diff options
| author | Ralf Baechle <ralf@linux-mips.org> | 2001-01-31 22:22:27 +0000 |
|---|---|---|
| committer | Ralf Baechle <ralf@linux-mips.org> | 2001-01-31 22:22:27 +0000 |
| commit | 825423e4c4f18289df2393951cfd2a7a31fc0464 (patch) | |
| tree | 4ad80e981c3d9effa910d2247d118d254f9a5d09 | |
| parent | c4693dc4856ab907a5c02187a8d398861bebfc7e (diff) | |
Merge with Linux 2.4.1.
571 files changed, 48261 insertions, 12465 deletions
@@ -2363,10 +2363,11 @@ E: gt8134b@prism.gatech.edu D: Dosemu N: Hannu Savolainen -E: hannu@voxware.pp.fi -D: Kernel sound drivers -S: Hiekkalaiturintie 3 A 8 -S: 00980 Helsinki +E: hannu@opensound.com +D: Maintainer of the sound drivers until 2.1.x days. +D: Original compressed boot image support. +S: Valurink. 4A11 +S: 03600 Karkkila S: Finland N: Eric Schenk diff --git a/Documentation/Changes b/Documentation/Changes index 91c60eee1..0b2f66e6f 100644 --- a/Documentation/Changes +++ b/Documentation/Changes @@ -56,7 +56,7 @@ o modutils 2.4.0 # insmod -V o e2fsprogs 1.19 # tune2fs --version o pcmcia-cs 3.1.21 # cardmgr -V o PPP 2.4.0 # pppd --version -o isdn4k-utils 3.1beta7 # isdnctrl 2>&1|grep version +o isdn4k-utils 3.1pre1 # isdnctrl 2>&1|grep version Kernel compilation ================== @@ -335,9 +335,9 @@ o <ftp://ftp.isdn4linux.de/pub/isdn4linux/utils/testing/isdn4k-utils.v3.1beta7. Netfilter --------- -o <http://netfilter.filewatcher.org/iptables-1.1.1.tar.bz2> -o <http://www.samba.org/netfilter/iptables-1.1.1.tar.bz2> -o <http://netfilter.kernelnotes.org/iptables-1.1.1.tar.bz2> +o <http://netfilter.filewatcher.org/iptables-1.2.tar.bz2> +o <http://netfilter.samba.org/iptables-1.2.tar.bz2> +o <http://netfilter.kernelnotes.org/iptables-1.2.tar.bz2> Ip-route2 --------- diff --git a/Documentation/Configure.help b/Documentation/Configure.help index 05c0930ca..a6e9caa73 100644 --- a/Documentation/Configure.help +++ b/Documentation/Configure.help @@ -1450,15 +1450,6 @@ CONFIG_BLK_DEV_LVM want), say M here and read Documentation/modules.txt. The module will be called lvm-mod.o. -Logical Volume Manager /proc file system information -CONFIG_LVM_PROC_FS - If you say Y here, you are able to access overall Logical Volume - Manager, Volume Group, Logical and Physical Volume information in - /proc/lvm. - - To use this option, you have to check, that the "/proc file system - support" (CONFIG_PROC_FS) is enabled too. - Multiple devices driver support CONFIG_BLK_DEV_MD This driver lets you combine several hard disk partitions into one @@ -1565,20 +1556,6 @@ CONFIG_MD_RAID5 If unsure, say Y. -RAID Boot support -CONFIG_MD_BOOT - To boot with an initial raid volume (any type) you can select - autodetect, or answer Y here and appropriate options to the kernel - at boot time. - For lilo and loadlin options see the file Documentation/md.txt. - -RAID AutoDetect support -CONFIG_AUTODETECT_RAID - An alternative to "Raid Boot support" is autodetect support. - With this selected, any partitons of type 0xFD will be considered for - inclusion in a RAID array. Information in the RAID-superblock on - the partition will determine how it is included. - Support for Acer PICA 1 chipset CONFIG_ACER_PICA_61 This is a machine with a R4400 133/150 MHz CPU. To compile a Linux @@ -2086,6 +2063,72 @@ CONFIG_INET_ECN If in doubt, say N. +IP6 tables support (required for filtering/masq/NAT) +CONFIG_IP6_NF_IPTABLES + ip6tables is a general, extensible packet identification framework. + Currently only the packet filtering and packet mangling subsystem + for IPv6 use this, but connection tracking is going to follow. + Say 'Y' or 'M' here if you want to use either of those. + + If you want to compile it as a module, say M here and read + Documentation/modules.txt. If unsure, say `N'. + +IPv6 limit match support +CONFIG_IP6_NF_MATCH_LIMIT + limit matching allows you to control the rate at which a rule can be + matched: mainly useful in combination with the LOG target ("LOG + target support", below) and to avoid some Denial of Service attacks. + + If you want to compile it as a module, say M here and read + Documentation/modules.txt. If unsure, say `N'. + +MAC address match support +CONFIG_IP6_NF_MATCH_MAC + mac matching allows you to match packets based on the source + ethernet address of the packet. + + If you want to compile it as a module, say M here and read + Documentation/modules.txt. If unsure, say `N'. + +netfilter mark match support +CONFIG_IP6_NF_MATCH_MARK + Netfilter mark matching allows you to match packets based on the + `nfmark' value in the packet. This can be set by the MARK target + (see below). + + If you want to compile it as a module, say M here and read + Documentation/modules.txt. If unsure, say `N'. + +Packet filtering +CONFIG_IP6_NF_FILTER + Packet filtering defines a table `filter', which has a series of + rules for simple packet filtering at local input, forwarding and + local output. See the man page for iptables(8). + + If you want to compile it as a module, say M here and read + Documentation/modules.txt. If unsure, say `N'. + +Packet mangling +CONFIG_IP6_NF_MANGLE + This option adds a `mangle' table to iptables: see the man page for + iptables(8). This table is used for various packet alterations + which can effect how the packet is routed. + + If you want to compile it as a module, say M here and read + Documentation/modules.txt. If unsure, say `N'. + +MARK target support +CONFIG_IP6_NF_TARGET_MARK + This option adds a `MARK' target, which allows you to create rules + in the `mangle' table which alter the netfilter mark (nfmark) field + associated with the packet packet prior to routing. This can change + the routing method (see `IP: use netfilter MARK value as routing + key') and can also be used by other subsystems to change their + behavior. + + If you want to compile it as a module, say M here and read + Documentation/modules.txt. If unsure, say `N'. + SYN flood protection CONFIG_SYN_COOKIES Normal TCP/IP networking is open to an attack known as "SYN @@ -2642,13 +2685,14 @@ CONFIG_HOTPLUG One well known example of this is PCMCIA- or PC-cards, credit-card size devices such as network cards, modems or hard drives which are - plugged into slots found on all modern laptop computers. + plugged into slots found on all modern laptop computers. Another + example, used on modern desktops as well as laptops, is USB. - Another example, used on modern desktops as well as laptops, is USB. - Enable HOTPLUG with USB and KMOD, and your kernel will automatically - call out to a user mode "policy agent" to load modules and set up - software needed to use USB devices you plug in. Get agent software - (at http://www.linux-usb.org/policy.html) and install it. + Enable HOTPLUG and KMOD, and build a modular kernel. Get agent + software (at http://linux-hotplug.sourceforge.net) and install it. + Then your kernel will automatically call out to a user mode "policy + agent" (/sbin/hotplug) to load modules and set up software needed + to use devices as you hotplug them. PCMCIA/Cardbus support CONFIG_PCMCIA @@ -10762,6 +10806,46 @@ CONFIG_MINIX_FS called minix.o. Note that the file system of your root partition (the one containing the directory /) cannot be compiled as a module. +Reiserfs support +CONFIG_REISERFS_FS + + Stores not just filenames but the files themselves in a balanced + tree. Uses journaling. + + Balanced trees are more efficient than traditional + filesystem architectural foundations. + + You can use reiserfs in all cases where you use the ext2fs file + system, and you will gain in speed and disk space. It has fewer + worst case performance situations than other file systems + because balanced trees are hardier creatures than other algorithms + are (if that is not technical enough, read www.namesys.com....:-) ) + + It is more easily extended to have features currently found in + database and keyword search systems than block allocation based + filesystems are. The next version will be so extended, and will + support plugins consistent with our motto ``It takes more than a + license to make source code open.'' + + Read www.namesys.com to learn more about reiserfs. + + Sponsored by Threshold Networks, Emusic.com, and Bigstorage.com. + + If you like it, you can pay us to add new features to it that you + need, buy a support contract, or pay us to port it to another OS. + +Enable Reiserfs consistency checks +CONFIG_REISERFS_CHECK + If you set this to yes, then ReiserFS will perform every check it + can possibly imagine of its internal consistency throughout its + operation. It will also go substantially slower. More than once we + have forgotten that this was on, and then gone despondent over the + latest benchmarks.:-) Use of this option allows our team to go all + out in checking for consistency when debugging without fear of its + effect on end users. If you are on the verge of sending in a bug + report, say yes and you might get a useful error message. Almost + everyone should say no. + Second extended fs support CONFIG_EXT2_FS This is the de facto standard Linux file system (method to organize @@ -13242,7 +13326,7 @@ CONFIG_ACPI This support requires an ACPI compliant platform (hardware/firmware). If both ACPI and Advanced Power Management (APM) support are - configured, ACPI is used. + configured, whichever is loaded first shall be used. This code DOES NOT currently provide a complete OSPM implementation -- it has not yet reached APM's level of functionality. When fully @@ -14644,6 +14728,14 @@ CONFIG_ISDN_PPP_VJ This enables Van Jacobson header compression for synchronous PPP. Say Y if the other end of the connection supports it. +CONFIG_ISDN_PPP_BSDCOMP + Support for the BSD-Compress compression method for PPP, which uses + the LZW compression method to compress each PPP packet before it is + sent over the wire. The machine at the other end of the PPP link + (usually your ISP) has to support the BSD-Compress compression + method as well for this to be useful. Even if they don't support it, + it is safe to say Y here. + Support audio via ISDN CONFIG_ISDN_AUDIO If you say Y here, the modem-emulator will support a subset of the @@ -16980,6 +17072,20 @@ CONFIG_DISPLAY7SEG another UltraSPARC-IIi-cEngine boardset with a 7-segment display, you should say N to this option. +CP1XXX Hardware Watchdog support +CONFIG_WATCHDOG_CP1XXX + This is the driver for the hardware watchdog timers present on + Sun Microsystems CompactPCI models CP1400 and CP1500. + + This driver is also available as a module ( = code which can be + inserted in and removed from the running kernel whenever you want). + The module will be called cpwatchdog.o. If you want to compile it + as a module, say M here and read Documentation/modules.txt. + + If you do not have a CompactPCI model CP1400 or CP1500, or + another UltraSPARC-IIi-cEngine boardset with hardware watchdog, + you should say N to this option. + IA-64 system type CONFIG_IA64_GENERIC This selects the system type of your hardware. A "generic" kernel diff --git a/Documentation/cachetlb.txt b/Documentation/cachetlb.txt index f3ae78497..c47cd632d 100644 --- a/Documentation/cachetlb.txt +++ b/Documentation/cachetlb.txt @@ -167,7 +167,7 @@ Here are the routines, one by one: This interface flushes an entire user address space from the caches. That is, after running, there will be no cache - lines assosciated with 'mm'. + lines associated with 'mm'. This interface is used to handle whole address space page table operations such as what happens during @@ -209,7 +209,7 @@ require a whole different set of interfaces to handle properly. The biggest problem is that of virtual aliasing in the data cache of a processor. -Is your port subsceptible to virtual aliasing in it's D-cache? +Is your port susceptible to virtual aliasing in it's D-cache? Well, if your D-cache is virtually indexed, is larger in size than PAGE_SIZE, and does not prevent multiple cache lines for the same physical address from existing at once, you have this problem. @@ -221,6 +221,9 @@ size). This setting will force the SYSv IPC layer to only allow user processes to mmap shared memory at address which are a multiple of this value. +NOTE: This does not fix shared mmaps, check out the sparc64 port for +one way to solve this (in particular SPARC_FLAG_MMAPSHARED). + Next, you have two methods to solve the D-cache aliasing issue for all other cases. Please keep in mind that fact that, for a given page mapped into some user address space, there is always at least one more @@ -240,7 +243,7 @@ existing ports should move over to the new mechanism as well. The physical page 'page' is about to be place into the user address space of a process. If it is possible for stores done recently by the kernel into this physical - page, to not be visible to an arbitray mapping in userspace, + page, to not be visible to an arbitrary mapping in userspace, you must flush this page from the D-cache. If the D-cache is writeback in nature, the dirty data (if @@ -266,7 +269,7 @@ Here is the new interface: For example, a port may temporarily map 'from' and 'to' to kernel virtual addresses during the copy. The virtual address - for these two pages is choosen in such a way that the kernel + for these two pages is chosen in such a way that the kernel load/store instructions happen to virtual addresses which are of the same "color" as the user mapping of the page. Sparc64 for example, uses this technique. @@ -306,7 +309,7 @@ Here is the new interface: simply be defined as a nop on that architecture. There is a bit set aside in page->flags (PG_arch_1) as - "architecture private". The kernel guarentees that, + "architecture private". The kernel guarantees that, for pagecache pages, it will clear this bit when such a page first enters the pagecache. @@ -323,7 +326,14 @@ Here is the new interface: update_mmu_cache(), a check is made of this flag bit, and if set the flush is done and the flag bit is cleared. -XXX Not documented: flush_icache_page(), need to talk to Paul - Mackerras, David Mosberger-Tang, et al. - to see what the expected semantics of this - interface are. -DaveM + void flush_icache_range(unsigned long start, unsigned long end) + When the kernel stores into addresses that it will execute + out of (eg when loading modules), this function is called. + + If the icache does not snoop stores then this routine will need + to flush it. + + void flush_icache_page(struct vm_area_struct *vma, struct page *page) + All the functionality of flush_icache_page can be implemented in + flush_dcache_page and update_mmu_cache. In 2.5 the hope is to + remove this interface completely. diff --git a/Documentation/usb/hotplug.txt b/Documentation/usb/hotplug.txt index a526ffc67..0b95e275b 100644 --- a/Documentation/usb/hotplug.txt +++ b/Documentation/usb/hotplug.txt @@ -1,4 +1,4 @@ -USB HOTPLUGGING +LINUX HOTPLUGGING In hotpluggable busses like USB (and Cardbus PCI), end-users plug devices into the bus with power on. In most cases, users expect the devices to become @@ -8,7 +8,7 @@ immediately usable. That means the system must do many things, including: loading a kernel module; newer drivers can use modutils to publish their device (and class) support to user utilities. - - Bind a driver to that device. That's done using the USB + - Bind a driver to that device. Bus frameworks do that using a device driver's probe() routine. - Tell other subsystems to configure the new device. Print @@ -26,6 +26,10 @@ Those triggered actions must support a system's administrative policies; such programs are called "policy agents" here. Typically they involve shell scripts that dispatch to more familiar administration tools. +Because some of those actions rely on information about drivers (metadata) +that is currently available only when the drivers are dynamically linked, +you get the best hotplugging when you configure a highly modular system. + KERNEL HOTPLUG HELPER (/sbin/hotplug) @@ -40,9 +44,14 @@ some kernel event. That name is used as the first key for further event dispatch; any other argument and environment parameters are specified by the subsystem making that invocation. -A reference implementation of a /sbin/hotplug script is available at the -http://www.linux-usb.org website, which works USB for but also knows how to -delegate to any /etc/hotplug/$TYPE.agent policy agent present. +Hotplug software and other resources is available at: + + http://linux-hotplug.sourceforge.net + +Mailing list information is also available at that site. + + +-------------------------------------------------------------------------- USB POLICY AGENT diff --git a/Documentation/usb/usb-serial.txt b/Documentation/usb/usb-serial.txt index 2fc5ac8eb..e1371097b 100644 --- a/Documentation/usb/usb-serial.txt +++ b/Documentation/usb/usb-serial.txt @@ -223,7 +223,7 @@ Generic Serial driver To enable the generic driver to recognize your device, build the driver as a module and load it by the following invocation: - insmod usb-serial vendor=0x#### product=0x#### + insmod usbserial vendor=0x#### product=0x#### where the #### is replaced with the hex representation of your device's vendor id and product id. diff --git a/MAINTAINERS b/MAINTAINERS index 492d15a40..e30666fb7 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -196,6 +196,12 @@ L: linux-kernel@vger.kernel.org W: http://www.ocston.org/~tigran/patches/bfs S: Maintained +BLOCK LAYER +P: Jens Axboe +M: axboe@suse.de +L: linux-kernel@vger.kernel.org +S: Maintained + BTTV VIDEO4LINUX DRIVER P: Gerd Knorr M: kraxel@goldbach.in-berlin.de @@ -1,6 +1,6 @@ VERSION = 2 PATCHLEVEL = 4 -SUBLEVEL = 0 +SUBLEVEL = 1 EXTRAVERSION = KERNELRELEASE=$(VERSION).$(PATCHLEVEL).$(SUBLEVEL)$(EXTRAVERSION) @@ -110,8 +110,7 @@ export ROOT_DEV = CURRENT export SVGA_MODE = -DSVGA_MODE=NORMAL_VGA # -# if you want the RAM disk device, define this to be the -# size in blocks. +# if you want the RAM disk device, define this to be the size in blocks. # This is i386 specific. # @@ -458,9 +457,8 @@ export MODVERFILE depend dep: dep-files -# make checkconfig: Prune 'scripts' directory to avoid "false positives". checkconfig: - find * -name '*.[hcS]' -type f -print | grep -v scripts/ | sort | xargs $(PERL) -w scripts/checkconfig.pl + find * -name '*.[hcS]' -type f -print | sort | xargs $(PERL) -w scripts/checkconfig.pl checkhelp: find * -name [cC]onfig.in -print | sort | xargs $(PERL) -w scripts/checkhelp.pl diff --git a/arch/alpha/kernel/Makefile b/arch/alpha/kernel/Makefile index 85cc7ed48..abc04cca2 100644 --- a/arch/alpha/kernel/Makefile +++ b/arch/alpha/kernel/Makefile @@ -23,7 +23,7 @@ obj-y := entry.o traps.o process.o osf_sys.o irq.o irq_alpha.o \ # FIXME! # These should be made conditional on the stuff that needs them! # -obj-y += irq_i8259.o irq_srm.o irq_pyxis.o \ +obj-y += irq_i8259.o irq_srm.o \ es1888.o smc37c669.o smc37c93x.o ns87312.o ifdef CONFIG_VGA_HOSE @@ -43,7 +43,7 @@ obj-y += core_apecs.o core_cia.o core_irongate.o core_lca.o core_mcpcia.o \ sys_jensen.o sys_miata.o sys_mikasa.o sys_nautilus.o sys_titan.o \ sys_noritake.o sys_rawhide.o sys_ruffian.o sys_rx164.o \ sys_sable.o sys_sio.o sys_sx164.o sys_takara.o sys_rx164.o \ - sys_wildfire.o core_wildfire.o + sys_wildfire.o core_wildfire.o irq_pyxis.o else @@ -94,6 +94,10 @@ obj-$(CONFIG_ALPHA_SX164) += sys_sx164.o obj-$(CONFIG_ALPHA_TAKARA) += sys_takara.o obj-$(CONFIG_ALPHA_WILDFIRE) += sys_wildfire.o +ifneq ($(CONFIG_ALPHA_MIATA)$(CONFIG_ALPHA_RUFFIAN)$(CONFIG_ALPHA_SX164),) +obj-y += irq_pyxis.o +endif + endif # GENERIC all: kernel.o head.o diff --git a/arch/alpha/kernel/osf_sys.c b/arch/alpha/kernel/osf_sys.c index 4d574b824..cd28b07fc 100644 --- a/arch/alpha/kernel/osf_sys.c +++ b/arch/alpha/kernel/osf_sys.c @@ -904,7 +904,6 @@ extern int do_sys_settimeofday(struct timeval *tv, struct timezone *tz); extern int do_getitimer(int which, struct itimerval *value); extern int do_setitimer(int which, struct itimerval *, struct itimerval *); asmlinkage int sys_utimes(char *, struct timeval *); -extern int sys_wait4(pid_t, int *, int, struct rusage *); extern int do_adjtimex(struct timex *); struct timeval32 diff --git a/arch/alpha/kernel/signal.c b/arch/alpha/kernel/signal.c index 73df1e930..e3d75f7a1 100644 --- a/arch/alpha/kernel/signal.c +++ b/arch/alpha/kernel/signal.c @@ -30,7 +30,6 @@ #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -asmlinkage int sys_wait4(int, int *, int, struct rusage *); asmlinkage void ret_from_sys_call(void); asmlinkage int do_signal(sigset_t *, struct pt_regs *, struct switch_stack *, unsigned long, unsigned long); diff --git a/arch/alpha/kernel/sys_cabriolet.c b/arch/alpha/kernel/sys_cabriolet.c index 624c6732c..58a8f4658 100644 --- a/arch/alpha/kernel/sys_cabriolet.c +++ b/arch/alpha/kernel/sys_cabriolet.c @@ -42,7 +42,7 @@ static inline void cabriolet_update_irq_hw(unsigned int irq, unsigned long mask) { int ofs = (irq - 16) / 8; - outb(mask >> (16 + ofs*3), 0x804 + ofs); + outb(mask >> (16 + ofs * 8), 0x804 + ofs); } static inline void diff --git a/arch/arm/kernel/signal.c b/arch/arm/kernel/signal.c index 1af2abd84..1b8f88815 100644 --- a/arch/arm/kernel/signal.c +++ b/arch/arm/kernel/signal.c @@ -32,8 +32,6 @@ #define SWI_SYS_SIGRETURN (0xef000000|(__NR_sigreturn)) #define SWI_SYS_RT_SIGRETURN (0xef000000|(__NR_rt_sigreturn)) -asmlinkage int sys_wait4(pid_t pid, unsigned long * stat_addr, - int options, unsigned long *ru); asmlinkage int do_signal(sigset_t *oldset, struct pt_regs * regs, int syscall); int copy_siginfo_to_user(siginfo_t *to, siginfo_t *from) diff --git a/arch/i386/Makefile b/arch/i386/Makefile index 25b523026..f7941bcbe 100644 --- a/arch/i386/Makefile +++ b/arch/i386/Makefile @@ -50,7 +50,7 @@ ifdef CONFIG_M686 CFLAGS += -march=i686 endif -ifdef CONFIG_M686FXSR +ifdef CONFIG_MPENTIUMIII CFLAGS += -march=i686 endif diff --git a/arch/i386/boot/bootsect.S b/arch/i386/boot/bootsect.S index 26c67d5fd..5434a9df2 100644 --- a/arch/i386/boot/bootsect.S +++ b/arch/i386/boot/bootsect.S @@ -25,10 +25,9 @@ * loads pretty fast by getting whole tracks at a time whenever possible. */ -#include <linux/config.h> /* for CONFIG_ROOT_RDONLY */ #include <asm/boot.h> -SETUPSECS = 4 /* default nr of setup-sectors */ +SETUPSECTS = 4 /* default nr of setup-sectors */ BOOTSEG = 0x07C0 /* original address of boot-sector */ INITSEG = DEF_INITSEG /* we move boot here - out of the way */ SETUPSEG = DEF_SETUPSEG /* setup starts here */ @@ -46,8 +45,8 @@ SWAP_DEV = 0 /* SWAP_DEV is now written by "build" */ #define RAMDISK 0 #endif -#ifndef CONFIG_ROOT_RDONLY -#define CONFIG_ROOT_RDONLY 1 +#ifndef ROOT_RDONLY +#define ROOT_RDONLY 1 #endif .code16 @@ -394,11 +393,11 @@ disksizes: .byte 36, 18, 15, 9 msg1: .byte 13, 10 .ascii "Loading" -# XXX: This is a *very* snug fit. +# XXX: This is a fairly snug fit. .org 497 -setup_sects: .byte SETUPSECS -root_flags: .word CONFIG_ROOT_RDONLY +setup_sects: .byte SETUPSECTS +root_flags: .word ROOT_RDONLY syssize: .word SYSSIZE swap_dev: .word SWAP_DEV ram_size: .word RAMDISK diff --git a/arch/i386/config.in b/arch/i386/config.in index 4dc863995..2dec3d90b 100644 --- a/arch/i386/config.in +++ b/arch/i386/config.in @@ -33,7 +33,7 @@ choice 'Processor family' \ Pentium-Classic CONFIG_M586TSC \ Pentium-MMX CONFIG_M586MMX \ Pentium-Pro/Celeron/Pentium-II CONFIG_M686 \ - Pentium-III CONFIG_M686FXSR \ + Pentium-III CONFIG_MPENTIUMIII \ Pentium-4 CONFIG_MPENTIUM4 \ K6/K6-II/K6-III CONFIG_MK6 \ Athlon/K7 CONFIG_MK7 \ @@ -45,8 +45,6 @@ choice 'Processor family' \ # Define implied options from the CPU selection here # -unset CONFIG_X86_FXSR - if [ "$CONFIG_M386" = "y" ]; then define_bool CONFIG_X86_CMPXCHG n define_int CONFIG_X86_L1_CACHE_SHIFT 4 @@ -87,14 +85,12 @@ if [ "$CONFIG_M686" = "y" ]; then define_bool CONFIG_X86_PGE y define_bool CONFIG_X86_USE_PPRO_CHECKSUM y fi -if [ "$CONFIG_M686FXSR" = "y" ]; then +if [ "$CONFIG_MPENTIUMIII" = "y" ]; then define_int CONFIG_X86_L1_CACHE_SHIFT 5 define_bool CONFIG_X86_TSC y define_bool CONFIG_X86_GOOD_APIC y define_bool CONFIG_X86_PGE y define_bool CONFIG_X86_USE_PPRO_CHECKSUM y - define_bool CONFIG_X86_FXSR y - define_bool CONFIG_X86_XMM y fi if [ "$CONFIG_MPENTIUM4" = "y" ]; then define_int CONFIG_X86_L1_CACHE_SHIFT 7 @@ -102,8 +98,6 @@ if [ "$CONFIG_MPENTIUM4" = "y" ]; then define_bool CONFIG_X86_GOOD_APIC y define_bool CONFIG_X86_PGE y define_bool CONFIG_X86_USE_PPRO_CHECKSUM y - define_bool CONFIG_X86_FXSR y - define_bool CONFIG_X86_XMM y fi if [ "$CONFIG_MK6" = "y" ]; then define_int CONFIG_X86_L1_CACHE_SHIFT 5 @@ -158,9 +152,7 @@ if [ "$CONFIG_HIGHMEM64G" = "y" ]; then define_bool CONFIG_X86_PAE y fi -if [ "$CONFIG_X86_FXSR" != "y" ]; then - bool 'Math emulation' CONFIG_MATH_EMULATION -fi +bool 'Math emulation' CONFIG_MATH_EMULATION bool 'MTRR (Memory Type Range Register) support' CONFIG_MTRR bool 'Symmetric multi-processing support' CONFIG_SMP if [ "$CONFIG_SMP" != "y" ]; then diff --git a/arch/i386/defconfig b/arch/i386/defconfig index de38a46ef..05a2e3725 100644 --- a/arch/i386/defconfig +++ b/arch/i386/defconfig @@ -27,7 +27,7 @@ CONFIG_KMOD=y # CONFIG_M586TSC is not set # CONFIG_M586MMX is not set # CONFIG_M686 is not set -CONFIG_M686FXSR=y +CONFIG_MPENTIUMIII=y # CONFIG_MPENTIUM4 is not set # CONFIG_MK6 is not set # CONFIG_MK7 is not set @@ -45,8 +45,6 @@ CONFIG_X86_TSC=y CONFIG_X86_GOOD_APIC=y CONFIG_X86_PGE=y CONFIG_X86_USE_PPRO_CHECKSUM=y -CONFIG_X86_FXSR=y -CONFIG_X86_XMM=y # CONFIG_TOSHIBA is not set # CONFIG_MICROCODE is not set # CONFIG_X86_MSR is not set @@ -54,6 +52,7 @@ CONFIG_X86_XMM=y CONFIG_NOHIGHMEM=y # CONFIG_HIGHMEM4G is not set # CONFIG_HIGHMEM64G is not set +# CONFIG_MATH_EMULATION is not set # CONFIG_MTRR is not set CONFIG_SMP=y CONFIG_HAVE_DEC_LOCK=y @@ -134,7 +133,6 @@ CONFIG_BLK_DEV_FD=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -512,11 +510,18 @@ CONFIG_PSMOUSE=y # Ftape, the floppy tape device driver # # CONFIG_FTAPE is not set -# CONFIG_AGP is not set +CONFIG_AGP=y +CONFIG_AGP_INTEL=y +CONFIG_AGP_I810=y +CONFIG_AGP_VIA=y +CONFIG_AGP_AMD=y +CONFIG_AGP_SIS=y +CONFIG_AGP_ALI=y CONFIG_DRM=y CONFIG_DRM_TDFX=y # CONFIG_DRM_GAMMA is not set # CONFIG_DRM_R128 is not set +CONFIG_DRM_RADEON=y # CONFIG_DRM_I810 is not set # CONFIG_DRM_MGA is not set CONFIG_PCMCIA_SERIAL=y @@ -538,6 +543,8 @@ CONFIG_PCMCIA_SERIAL=y # CONFIG_QUOTA is not set # CONFIG_AUTOFS_FS is not set CONFIG_AUTOFS4_FS=y +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/i386/kernel/i387.c b/arch/i386/kernel/i387.c index 3031432d0..a3a526b4b 100644 --- a/arch/i386/kernel/i387.c +++ b/arch/i386/kernel/i387.c @@ -18,14 +18,6 @@ #include <asm/ptrace.h> #include <asm/uaccess.h> -#if defined(CONFIG_X86_FXSR) -#define HAVE_FXSR 1 -#elif defined(CONFIG_X86_RUNTIME_FXSR) -#define HAVE_FXSR (cpu_has_fxsr) -#else -#define HAVE_FXSR 0 -#endif - #ifdef CONFIG_MATH_EMULATION #define HAVE_HWFP (boot_cpu_data.hard_math) #else @@ -35,13 +27,13 @@ /* * The _current_ task is using the FPU for the first time * so initialize it and set the mxcsr to its default - * value at reset if we support FXSR and then + * value at reset if we support XMM instructions and then * remeber the current task has used the FPU. */ void init_fpu(void) { __asm__("fninit"); - if ( HAVE_FXSR ) + if ( cpu_has_xmm ) load_mxcsr(0x1f80); current->used_math = 1; @@ -51,9 +43,9 @@ void init_fpu(void) * FPU lazy state save handling. */ -void save_init_fpu( struct task_struct *tsk ) +static inline void __save_init_fpu( struct task_struct *tsk ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { asm volatile( "fxsave %0 ; fnclex" : "=m" (tsk->thread.i387.fxsave) ); } else { @@ -61,12 +53,28 @@ void save_init_fpu( struct task_struct *tsk ) : "=m" (tsk->thread.i387.fsave) ); } tsk->flags &= ~PF_USEDFPU; +} + +void save_init_fpu( struct task_struct *tsk ) +{ + __save_init_fpu(tsk); stts(); } +void kernel_fpu_begin(void) +{ + struct task_struct *tsk = current; + + if (tsk->flags & PF_USEDFPU) { + __save_init_fpu(tsk); + return; + } + clts(); +} + void restore_fpu( struct task_struct *tsk ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { asm volatile( "fxrstor %0" : : "m" (tsk->thread.i387.fxsave) ); } else { @@ -144,7 +152,7 @@ static inline unsigned long twd_fxsr_to_i387( struct i387_fxsave_struct *fxsave unsigned short get_fpu_cwd( struct task_struct *tsk ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { return tsk->thread.i387.fxsave.cwd; } else { return (unsigned short)tsk->thread.i387.fsave.cwd; @@ -153,7 +161,7 @@ unsigned short get_fpu_cwd( struct task_struct *tsk ) unsigned short get_fpu_swd( struct task_struct *tsk ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { return tsk->thread.i387.fxsave.swd; } else { return (unsigned short)tsk->thread.i387.fsave.swd; @@ -162,7 +170,7 @@ unsigned short get_fpu_swd( struct task_struct *tsk ) unsigned short get_fpu_twd( struct task_struct *tsk ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { return tsk->thread.i387.fxsave.twd; } else { return (unsigned short)tsk->thread.i387.fsave.twd; @@ -171,7 +179,7 @@ unsigned short get_fpu_twd( struct task_struct *tsk ) unsigned short get_fpu_mxcsr( struct task_struct *tsk ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { return tsk->thread.i387.fxsave.mxcsr; } else { return 0x1f80; @@ -180,7 +188,7 @@ unsigned short get_fpu_mxcsr( struct task_struct *tsk ) void set_fpu_cwd( struct task_struct *tsk, unsigned short cwd ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { tsk->thread.i387.fxsave.cwd = cwd; } else { tsk->thread.i387.fsave.cwd = ((long)cwd | 0xffff0000); @@ -189,7 +197,7 @@ void set_fpu_cwd( struct task_struct *tsk, unsigned short cwd ) void set_fpu_swd( struct task_struct *tsk, unsigned short swd ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { tsk->thread.i387.fxsave.swd = swd; } else { tsk->thread.i387.fsave.swd = ((long)swd | 0xffff0000); @@ -198,7 +206,7 @@ void set_fpu_swd( struct task_struct *tsk, unsigned short swd ) void set_fpu_twd( struct task_struct *tsk, unsigned short twd ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { tsk->thread.i387.fxsave.twd = twd_i387_to_fxsr(twd); } else { tsk->thread.i387.fsave.twd = ((long)twd | 0xffff0000); @@ -207,7 +215,7 @@ void set_fpu_twd( struct task_struct *tsk, unsigned short twd ) void set_fpu_mxcsr( struct task_struct *tsk, unsigned short mxcsr ) { - if ( HAVE_FXSR ) { + if ( cpu_has_xmm ) { tsk->thread.i387.fxsave.mxcsr = mxcsr; } } @@ -321,7 +329,7 @@ int save_i387( struct _fpstate *buf ) current->used_math = 0; if ( HAVE_HWFP ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { return save_i387_fxsave( buf ); } else { return save_i387_fsave( buf ); @@ -354,7 +362,7 @@ int restore_i387( struct _fpstate *buf ) int err; if ( HAVE_HWFP ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { err = restore_i387_fxsave( buf ); } else { err = restore_i387_fsave( buf ); @@ -387,7 +395,7 @@ static inline int get_fpregs_fxsave( struct user_i387_struct *buf, int get_fpregs( struct user_i387_struct *buf, struct task_struct *tsk ) { if ( HAVE_HWFP ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { return get_fpregs_fxsave( buf, tsk ); } else { return get_fpregs_fsave( buf, tsk ); @@ -415,7 +423,7 @@ static inline int set_fpregs_fxsave( struct task_struct *tsk, int set_fpregs( struct task_struct *tsk, struct user_i387_struct *buf ) { if ( HAVE_HWFP ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { return set_fpregs_fxsave( tsk, buf ); } else { return set_fpregs_fsave( tsk, buf ); @@ -428,9 +436,10 @@ int set_fpregs( struct task_struct *tsk, struct user_i387_struct *buf ) int get_fpxregs( struct user_fxsr_struct *buf, struct task_struct *tsk ) { - if ( HAVE_FXSR ) { - __copy_to_user( (void *)buf, &tsk->thread.i387.fxsave, - sizeof(struct user_fxsr_struct) ); + if ( cpu_has_fxsr ) { + if (__copy_to_user( (void *)buf, &tsk->thread.i387.fxsave, + sizeof(struct user_fxsr_struct) )) + return -EFAULT; return 0; } else { return -EIO; @@ -439,7 +448,7 @@ int get_fpxregs( struct user_fxsr_struct *buf, struct task_struct *tsk ) int set_fpxregs( struct task_struct *tsk, struct user_fxsr_struct *buf ) { - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { __copy_from_user( &tsk->thread.i387.fxsave, (void *)buf, sizeof(struct user_fxsr_struct) ); /* mxcsr bit 6 and 31-16 must be zero for security reasons */ @@ -485,7 +494,7 @@ int dump_fpu( struct pt_regs *regs, struct user_i387_struct *fpu ) fpvalid = tsk->used_math; if ( fpvalid ) { unlazy_fpu( tsk ); - if ( HAVE_FXSR ) { + if ( cpu_has_fxsr ) { copy_fpu_fxsave( tsk, fpu ); } else { copy_fpu_fsave( tsk, fpu ); @@ -500,7 +509,7 @@ int dump_extended_fpu( struct pt_regs *regs, struct user_fxsr_struct *fpu ) int fpvalid; struct task_struct *tsk = current; - fpvalid = tsk->used_math && HAVE_FXSR; + fpvalid = tsk->used_math && cpu_has_fxsr; if ( fpvalid ) { unlazy_fpu( tsk ); memcpy( fpu, &tsk->thread.i387.fxsave, diff --git a/arch/i386/kernel/io_apic.c b/arch/i386/kernel/io_apic.c index c5aece040..b17f499b4 100644 --- a/arch/i386/kernel/io_apic.c +++ b/arch/i386/kernel/io_apic.c @@ -253,7 +253,8 @@ static int __init find_isa_irq_pin(int irq, int type) * Find a specific PCI IRQ entry. * Not an __init, possibly needed by modules */ -static int __init pin_2_irq(int idx, int apic, int pin); +static int pin_2_irq(int idx, int apic, int pin); + int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pci_pin) { int apic, i, best_guess = -1; @@ -475,7 +476,7 @@ static inline int irq_trigger(int idx) return MPBIOS_trigger(idx); } -static int __init pin_2_irq(int idx, int apic, int pin) +static int pin_2_irq(int idx, int apic, int pin) { int irq, i; int bus = mp_irqs[idx].mpc_srcbus; diff --git a/arch/i386/kernel/pci-irq.c b/arch/i386/kernel/pci-irq.c index bd2bf22d3..8281e3f82 100644 --- a/arch/i386/kernel/pci-irq.c +++ b/arch/i386/kernel/pci-irq.c @@ -234,22 +234,107 @@ static int pirq_cyrix_set(struct pci_dev *router, struct pci_dev *dev, int pirq, return 1; } +/* + * PIRQ routing for SiS 85C503 router used in several SiS chipsets + * According to the SiS 5595 datasheet (preliminary V1.0, 12/24/1997) + * the related registers work as follows: + * + * general: one byte per re-routable IRQ, + * bit 7 IRQ mapping enabled (0) or disabled (1) + * bits [6:4] reserved + * bits [3:0] IRQ to map to + * allowed: 3-7, 9-12, 14-15 + * reserved: 0, 1, 2, 8, 13 + * + * individual registers in device config space: + * + * 0x41/0x42/0x43/0x44: PCI INT A/B/C/D - bits as in general case + * + * 0x61: IDEIRQ: bits as in general case - but: + * bits [6:5] must be written 01 + * bit 4 channel-select primary (0), secondary (1) + * + * 0x62: USBIRQ: bits as in general case - but: + * bit 4 OHCI function disabled (0), enabled (1) + * + * 0x6a: ACPI/SCI IRQ - bits as in general case + * + * 0x7e: Data Acq. Module IRQ - bits as in general case + * + * Apparently there are systems implementing PCI routing table using both + * link values 0x01-0x04 and 0x41-0x44 for PCI INTA..D, but register offsets + * like 0x62 as link values for USBIRQ e.g. So there is no simple + * "register = offset + pirq" relation. + * Currently we support PCI INTA..D and USBIRQ and try our best to handle + * both link mappings. + * IDE/ACPI/DAQ mapping is currently unsupported (left untouched as set by BIOS). + */ + static int pirq_sis_get(struct pci_dev *router, struct pci_dev *dev, int pirq) { u8 x; - int reg = 0x41 + (pirq - 'A') ; - - pci_read_config_byte(router, reg, &x); + int reg = pirq; + + switch(pirq) { + case 0x01: + case 0x02: + case 0x03: + case 0x04: + reg += 0x40; + case 0x41: + case 0x42: + case 0x43: + case 0x44: + case 0x62: + pci_read_config_byte(router, reg, &x); + if (reg != 0x62) + break; + if (!(x & 0x40)) + return 0; + break; + case 0x61: + case 0x6a: + case 0x7e: + printk("SiS pirq: advanced IDE/ACPI/DAQ mapping not yet implemented\n"); + return 0; + default: + printk("SiS router pirq escape (%d)\n", pirq); + return 0; + } return (x & 0x80) ? 0 : (x & 0x0f); } static int pirq_sis_set(struct pci_dev *router, struct pci_dev *dev, int pirq, int irq) { u8 x; - int reg = 0x41 + (pirq - 'A') ; - - pci_read_config_byte(router, reg, &x); - x = (pirq & 0x20) ? 0 : (irq & 0x0f); + int reg = pirq; + + switch(pirq) { + case 0x01: + case 0x02: + case 0x03: + case 0x04: + reg += 0x40; + case 0x41: + case 0x42: + case 0x43: + case 0x44: + case 0x62: + x = (irq&0x0f) ? (irq&0x0f) : 0x80; + if (reg != 0x62) + break; + /* always mark OHCI enabled, as nothing else knows about this */ + x |= 0x40; + break; + case 0x61: + case 0x6a: + case 0x7e: + printk("advanced SiS pirq mapping not yet implemented\n"); + return 0; + default: + printk("SiS router pirq escape (%d)\n", pirq); + return 0; + } pci_write_config_byte(router, reg, x); return 1; @@ -462,18 +547,9 @@ static int pcibios_lookup_irq(struct pci_dev *dev, int assign) irq = pirq & 0xf; DBG(" -> hardcoded IRQ %d\n", irq); msg = "Hardcoded"; - if (dev->irq && dev->irq != irq) { - printk("IRQ routing conflict in pirq table! Try 'pci=autoirq'\n"); - return 0; - } } else if (r->get && (irq = r->get(pirq_router_dev, dev, pirq))) { DBG(" -> got IRQ %d\n", irq); msg = "Found"; - /* We refuse to override the dev->irq information. Give a warning! */ - if (dev->irq && dev->irq != irq) { - printk("IRQ routing conflict in pirq table! Try 'pci=autoirq'\n"); - return 0; - } } else if (newirq && r->set && (dev->class >> 8) != PCI_CLASS_DISPLAY_VGA) { DBG(" -> assigning IRQ %d", newirq); if (r->set(pirq_router_dev, dev, pirq, newirq)) { @@ -504,6 +580,11 @@ static int pcibios_lookup_irq(struct pci_dev *dev, int assign) if (!info) continue; if (info->irq[pin].link == pirq) { + /* We refuse to override the dev->irq information. Give a warning! */ + if (dev2->irq && dev2->irq != irq) { + printk("IRQ routing conflict in pirq table for device %s\n", dev2->slot_name); + continue; + } dev2->irq = irq; pirq_penalty[irq]++; if (dev != dev2) diff --git a/arch/i386/kernel/setup.c b/arch/i386/kernel/setup.c index 09aa4a08c..2c3f61f7f 100644 --- a/arch/i386/kernel/setup.c +++ b/arch/i386/kernel/setup.c @@ -147,6 +147,7 @@ extern char _text, _etext, _edata, _end; extern unsigned long cpu_khz; static int disable_x86_serial_nr __initdata = 1; +static int disable_x86_fxsr __initdata = 0; /* * This is set up by the setup-routine at boot-time @@ -518,7 +519,7 @@ void __init setup_memory_region(void) e820.nr_map = 0; add_memory_region(0, LOWMEMSIZE(), E820_RAM); - add_memory_region(HIGH_MEMORY, (mem_size << 10) - HIGH_MEMORY, E820_RAM); + add_memory_region(HIGH_MEMORY, mem_size << 10, E820_RAM); } printk("BIOS-provided physical RAM map:\n"); print_memory_map(who); @@ -1400,9 +1401,6 @@ static void __init init_centaur(struct cpuinfo_x86 *c) wrmsr (0x1107, lo, hi); set_bit(X86_FEATURE_CX8, &c->x86_capability); - rdmsr (0x80000001, lo, hi); - if (hi & (1<<31)) - set_bit(X86_FEATURE_3DNOW, &c->x86_capability); get_model_name(c); display_cacheinfo(c); @@ -1796,6 +1794,13 @@ int __init x86_serial_nr_setup(char *s) } __setup("serialnumber", x86_serial_nr_setup); +int __init x86_fxsr_setup(char * s) +{ + disable_x86_fxsr = 1; + return 1; +} +__setup("nofxsr", x86_fxsr_setup); + /* Standard macro to see if a specific flag is changeable */ static inline int flag_is_changeable_p(u32 flag) @@ -1866,7 +1871,8 @@ static int __init id_and_try_enable_cpuid(struct cpuinfo_x86 *c) /* Detect Cyrix with disabled CPUID */ if ( c->x86 == 4 && test_cyrix_52div() ) { strcpy(c->x86_vendor_id, "CyrixInstead"); - } + c->x86_vendor = X86_VENDOR_CYRIX; + } else /* Detect NexGen with old hypercode */ if ( deep_magic_nexgen_probe() ) { @@ -2000,11 +2006,17 @@ void __init identify_cpu(struct cpuinfo_x86 *c) */ /* TSC disabled? */ -#ifdef CONFIG_TSC +#ifndef CONFIG_X86_TSC if ( tsc_disable ) clear_bit(X86_FEATURE_TSC, &c->x86_capability); #endif + /* FXSR disabled? */ + if (disable_x86_fxsr) { + clear_bit(X86_FEATURE_FXSR, &c->x86_capability); + clear_bit(X86_FEATURE_XMM, &c->x86_capability); + } + /* Disable the PN if appropriate */ squash_the_stupid_serial_number(c); diff --git a/arch/i386/kernel/signal.c b/arch/i386/kernel/signal.c index 2423a8b06..7df9c875c 100644 --- a/arch/i386/kernel/signal.c +++ b/arch/i386/kernel/signal.c @@ -26,8 +26,6 @@ #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); asmlinkage int FASTCALL(do_signal(struct pt_regs *regs, sigset_t *oldset)); int copy_siginfo_to_user(siginfo_t *to, siginfo_t *from) diff --git a/arch/i386/kernel/traps.c b/arch/i386/kernel/traps.c index 158d183b8..7c02813af 100644 --- a/arch/i386/kernel/traps.c +++ b/arch/i386/kernel/traps.c @@ -23,6 +23,7 @@ #include <linux/delay.h> #include <linux/spinlock.h> #include <linux/interrupt.h> +#include <linux/highmem.h> #ifdef CONFIG_MCA #include <linux/mca.h> diff --git a/arch/i386/lib/mmx.c b/arch/i386/lib/mmx.c index 4aca07e72..dfbf7ba8e 100644 --- a/arch/i386/lib/mmx.c +++ b/arch/i386/lib/mmx.c @@ -2,6 +2,8 @@ #include <linux/string.h> #include <linux/sched.h> +#include <asm/i387.h> + /* * MMX 3DNow! library helper functions * @@ -26,13 +28,7 @@ void *_mmx_memcpy(void *to, const void *from, size_t len) void *p=to; int i= len >> 6; /* len/64 */ - if (!(current->flags & PF_USEDFPU)) - clts(); - else - { - __asm__ __volatile__ ( " fnsave %0; fwait\n"::"m"(current->thread.i387)); - current->flags &= ~PF_USEDFPU; - } + kernel_fpu_begin(); __asm__ __volatile__ ( "1: prefetch (%0)\n" /* This set is 28 bytes */ @@ -88,20 +84,15 @@ void *_mmx_memcpy(void *to, const void *from, size_t len) * Now do the tail of the block */ __memcpy(to, from, len&63); - stts(); + kernel_fpu_end(); return p; } static void fast_clear_page(void *page) { int i; - if (!(current->flags & PF_USEDFPU)) - clts(); - else - { - __asm__ __volatile__ ( " fnsave %0; fwait\n"::"m"(current->thread.i387)); - current->flags &= ~PF_USEDFPU; - } + + kernel_fpu_begin(); __asm__ __volatile__ ( " pxor %%mm0, %%mm0\n" : : @@ -127,19 +118,14 @@ static void fast_clear_page(void *page) __asm__ __volatile__ ( " sfence \n" : : ); - stts(); + kernel_fpu_end(); } static void fast_copy_page(void *to, void *from) { int i; - if (!(current->flags & PF_USEDFPU)) - clts(); - else - { - __asm__ __volatile__ ( " fnsave %0; fwait\n"::"m"(current->thread.i387)); - current->flags &= ~PF_USEDFPU; - } + + kernel_fpu_begin(); /* maybe the prefetch stuff can go before the expensive fnsave... * but that is for later. -AV @@ -199,7 +185,7 @@ static void fast_copy_page(void *to, void *from) __asm__ __volatile__ ( " sfence \n" : : ); - stts(); + kernel_fpu_end(); } /* diff --git a/arch/i386/mm/init.c b/arch/i386/mm/init.c index b2e5a1bfc..6414b508c 100644 --- a/arch/i386/mm/init.c +++ b/arch/i386/mm/init.c @@ -317,7 +317,7 @@ static void __init pagetable_init (void) pgd_t *pgd, *pgd_base; int i, j, k; pmd_t *pmd; - pte_t *pte; + pte_t *pte, *pte_base; /* * This can be zero as well - no problem, in that case we exit @@ -366,11 +366,7 @@ static void __init pagetable_init (void) continue; } - pte = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE); - set_pmd(pmd, __pmd(_KERNPG_TABLE + __pa(pte))); - - if (pte != pte_offset(pmd, 0)) - BUG(); + pte_base = pte = (pte_t *) alloc_bootmem_low_pages(PAGE_SIZE); for (k = 0; k < PTRS_PER_PTE; pte++, k++) { vaddr = i*PGDIR_SIZE + j*PMD_SIZE + k*PAGE_SIZE; @@ -378,6 +374,10 @@ static void __init pagetable_init (void) break; *pte = mk_pte_phys(__pa(vaddr), PAGE_KERNEL); } + set_pmd(pmd, __pmd(_KERNPG_TABLE + __pa(pte_base))); + if (pte_base != pte_offset(pmd, 0)) + BUG(); + } } diff --git a/arch/ia64/ia32/sys_ia32.c b/arch/ia64/ia32/sys_ia32.c index f7b857b4c..25007aa43 100644 --- a/arch/ia64/ia32/sys_ia32.c +++ b/arch/ia64/ia32/sys_ia32.c @@ -1957,9 +1957,6 @@ put_rusage (struct rusage32 *ru, struct rusage *r) return err; } -extern asmlinkage long sys_wait4(pid_t pid,unsigned int * stat_addr, - int options, struct rusage * ru); - asmlinkage long sys32_wait4(__kernel_pid_t32 pid, unsigned int *stat_addr, int options, struct rusage32 *ru) diff --git a/arch/m68k/kernel/signal.c b/arch/m68k/kernel/signal.c index 9d754f9b8..e9c41780c 100644 --- a/arch/m68k/kernel/signal.c +++ b/arch/m68k/kernel/signal.c @@ -47,8 +47,6 @@ #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); asmlinkage int do_signal(sigset_t *oldset, struct pt_regs *regs); const int frame_extra_sizes[16] = { diff --git a/arch/mips/defconfig b/arch/mips/defconfig index e17cb4e09..924a848c7 100644 --- a/arch/mips/defconfig +++ b/arch/mips/defconfig @@ -115,7 +115,6 @@ CONFIG_SYSCTL=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -249,6 +248,8 @@ CONFIG_FONT_8x16=y # CONFIG_QUOTA is not set CONFIG_AUTOFS_FS=y CONFIG_AUTOFS4_FS=y +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips/defconfig-atlas b/arch/mips/defconfig-atlas index 56a5cbd26..73d74c356 100644 --- a/arch/mips/defconfig-atlas +++ b/arch/mips/defconfig-atlas @@ -111,7 +111,6 @@ CONFIG_BLK_DEV_RAM_SIZE=4096 # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -377,6 +376,8 @@ CONFIG_RTC=y # CONFIG_QUOTA is not set CONFIG_AUTOFS_FS=y # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips/defconfig-cp7000 b/arch/mips/defconfig-cp7000 index 1221ed1bf..60cb12ad0 100644 --- a/arch/mips/defconfig-cp7000 +++ b/arch/mips/defconfig-cp7000 @@ -113,7 +113,6 @@ CONFIG_SYSCTL=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -352,6 +351,8 @@ CONFIG_SERIAL_EXTENDED=y # CONFIG_QUOTA is not set # CONFIG_AUTOFS_FS is not set # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips/defconfig-ddb5476 b/arch/mips/defconfig-ddb5476 index 2f947f254..3c54bca48 100644 --- a/arch/mips/defconfig-ddb5476 +++ b/arch/mips/defconfig-ddb5476 @@ -115,7 +115,6 @@ CONFIG_SYSCTL=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -415,6 +414,8 @@ CONFIG_PSMOUSE=y # CONFIG_QUOTA is not set # CONFIG_AUTOFS_FS is not set # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips/defconfig-decstation b/arch/mips/defconfig-decstation index ac2e8d9e4..d633b27a2 100644 --- a/arch/mips/defconfig-decstation +++ b/arch/mips/defconfig-decstation @@ -109,7 +109,6 @@ CONFIG_TC=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -243,6 +242,8 @@ CONFIG_RTC=y # CONFIG_QUOTA is not set # CONFIG_AUTOFS_FS is not set # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips/defconfig-ev64120 b/arch/mips/defconfig-ev64120 index 1a24e68e5..78f51045a 100644 --- a/arch/mips/defconfig-ev64120 +++ b/arch/mips/defconfig-ev64120 @@ -117,7 +117,6 @@ CONFIG_SYSCTL=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -351,6 +350,8 @@ CONFIG_SERIAL=y # CONFIG_QUOTA is not set # CONFIG_AUTOFS_FS is not set # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips/defconfig-ev96100 b/arch/mips/defconfig-ev96100 index c04d274b0..48b2bcd61 100644 --- a/arch/mips/defconfig-ev96100 +++ b/arch/mips/defconfig-ev96100 @@ -113,7 +113,6 @@ CONFIG_NET=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -342,6 +341,8 @@ CONFIG_SERIAL_CONSOLE=y # CONFIG_QUOTA is not set # CONFIG_AUTOFS_FS is not set # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips/defconfig-ip22 b/arch/mips/defconfig-ip22 index e17cb4e09..924a848c7 100644 --- a/arch/mips/defconfig-ip22 +++ b/arch/mips/defconfig-ip22 @@ -115,7 +115,6 @@ CONFIG_SYSCTL=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -249,6 +248,8 @@ CONFIG_FONT_8x16=y # CONFIG_QUOTA is not set CONFIG_AUTOFS_FS=y CONFIG_AUTOFS4_FS=y +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips/defconfig-malta b/arch/mips/defconfig-malta index 43b031607..66125a703 100644 --- a/arch/mips/defconfig-malta +++ b/arch/mips/defconfig-malta @@ -110,7 +110,6 @@ CONFIG_BLK_DEV_RAM_SIZE=4096 # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -376,6 +375,8 @@ CONFIG_RTC=y # CONFIG_QUOTA is not set CONFIG_AUTOFS_FS=y # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips/defconfig-rm200 b/arch/mips/defconfig-rm200 index 0964d3957..a04708b93 100644 --- a/arch/mips/defconfig-rm200 +++ b/arch/mips/defconfig-rm200 @@ -119,7 +119,6 @@ CONFIG_BLK_DEV_FD=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -266,6 +265,8 @@ CONFIG_RTC=y # CONFIG_QUOTA is not set # CONFIG_AUTOFS_FS is not set # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips/kernel/irixsig.c b/arch/mips/kernel/irixsig.c index 022bb1f77..900e1004f 100644 --- a/arch/mips/kernel/irixsig.c +++ b/arch/mips/kernel/irixsig.c @@ -16,8 +16,6 @@ #include <asm/ptrace.h> #include <asm/uaccess.h> -asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); extern asmlinkage void syscall_trace(void); #undef DEBUG_SIG diff --git a/arch/mips/kernel/signal.c b/arch/mips/kernel/signal.c index a1e26db87..ff7c8d5a8 100644 --- a/arch/mips/kernel/signal.c +++ b/arch/mips/kernel/signal.c @@ -30,8 +30,6 @@ #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -extern asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); extern asmlinkage int do_signal(sigset_t *oldset, struct pt_regs *regs); extern asmlinkage int (*save_fp_context)(struct sigcontext *sc); diff --git a/arch/mips64/defconfig b/arch/mips64/defconfig index 1636cb6cb..8f8b780b2 100644 --- a/arch/mips64/defconfig +++ b/arch/mips64/defconfig @@ -91,7 +91,6 @@ CONFIG_PCI_NAMES=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -351,6 +350,8 @@ CONFIG_SERIAL_CONSOLE=y # CONFIG_QUOTA is not set CONFIG_AUTOFS_FS=y # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips64/defconfig-ip22 b/arch/mips64/defconfig-ip22 index d047a87c6..39518cb59 100644 --- a/arch/mips64/defconfig-ip22 +++ b/arch/mips64/defconfig-ip22 @@ -81,7 +81,6 @@ CONFIG_BINFMT_ELF=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -267,6 +266,8 @@ CONFIG_VT_CONSOLE=y # CONFIG_QUOTA is not set # CONFIG_AUTOFS_FS is not set # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips64/defconfig-ip27 b/arch/mips64/defconfig-ip27 index 1636cb6cb..8f8b780b2 100644 --- a/arch/mips64/defconfig-ip27 +++ b/arch/mips64/defconfig-ip27 @@ -91,7 +91,6 @@ CONFIG_PCI_NAMES=y # CONFIG_MD_RAID1 is not set # CONFIG_MD_RAID5 is not set # CONFIG_BLK_DEV_LVM is not set -# CONFIG_LVM_PROC_FS is not set # # Networking options @@ -351,6 +350,8 @@ CONFIG_SERIAL_CONSOLE=y # CONFIG_QUOTA is not set CONFIG_AUTOFS_FS=y # CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set diff --git a/arch/mips64/kernel/linux32.c b/arch/mips64/kernel/linux32.c index 2e64aa465..f1a6084e5 100644 --- a/arch/mips64/kernel/linux32.c +++ b/arch/mips64/kernel/linux32.c @@ -585,9 +585,6 @@ put_rusage (struct rusage32 *ru, struct rusage *r) return err; } -extern asmlinkage int sys_wait4(pid_t pid, unsigned int * stat_addr, - int options, struct rusage * ru); - asmlinkage int sys32_wait4(__kernel_pid_t32 pid, unsigned int * stat_addr, int options, struct rusage32 * ru) diff --git a/arch/mips64/kernel/signal.c b/arch/mips64/kernel/signal.c index 15b2fc60b..f30df53a3 100644 --- a/arch/mips64/kernel/signal.c +++ b/arch/mips64/kernel/signal.c @@ -31,8 +31,6 @@ #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -extern asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); extern asmlinkage int do_signal(sigset_t *oldset, struct pt_regs *regs); extern asmlinkage int save_fp_context(struct sigcontext *sc); extern asmlinkage int restore_fp_context(struct sigcontext *sc); diff --git a/arch/mips64/kernel/signal32.c b/arch/mips64/kernel/signal32.c index 548f271b5..6dad16160 100644 --- a/arch/mips64/kernel/signal32.c +++ b/arch/mips64/kernel/signal32.c @@ -30,8 +30,6 @@ #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -extern asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); extern asmlinkage int do_signal32(sigset_t *oldset, struct pt_regs *regs); extern asmlinkage int save_fp_context(struct sigcontext *sc); extern asmlinkage int restore_fp_context(struct sigcontext *sc); diff --git a/arch/ppc/8260_io/enet.c b/arch/ppc/8260_io/enet.c index e1ff3c092..77b0eb279 100644 --- a/arch/ppc/8260_io/enet.c +++ b/arch/ppc/8260_io/enet.c @@ -633,6 +633,9 @@ int __init scc_enet_init(void) /* Allocate some private information. */ cep = (struct scc_enet_private *)kmalloc(sizeof(*cep), GFP_KERNEL); + if (cep == NULL) + return -ENOMEM; + __clear_user(cep,sizeof(*cep)); spin_lock_init(&cep->lock); diff --git a/arch/ppc/8260_io/fcc_enet.c b/arch/ppc/8260_io/fcc_enet.c index da3991ae3..c8c31bd42 100644 --- a/arch/ppc/8260_io/fcc_enet.c +++ b/arch/ppc/8260_io/fcc_enet.c @@ -1082,6 +1082,9 @@ int __init fec_enet_init(void) */ cep = (struct fcc_enet_private *) kmalloc(sizeof(*cep), GFP_KERNEL); + if (cep == NULL) + return -ENOMEM; + __clear_user(cep,sizeof(*cep)); spin_lock_init(&cep->lock); cep->fip = fip; diff --git a/arch/ppc/8xx_io/enet.c b/arch/ppc/8xx_io/enet.c index b77cf32da..01eb2758b 100644 --- a/arch/ppc/8xx_io/enet.c +++ b/arch/ppc/8xx_io/enet.c @@ -652,7 +652,9 @@ int __init scc_enet_init(void) /* Allocate some private information. */ cep = (struct scc_enet_private *)kmalloc(sizeof(*cep), GFP_KERNEL); - /*memset(cep, 0, sizeof(*cep));*/ + if (cep == NULL) + return -ENOMEM; + __clear_user(cep,sizeof(*cep)); spin_lock_init(&cep->lock); diff --git a/arch/ppc/8xx_io/fec.c b/arch/ppc/8xx_io/fec.c index 659576e8c..6e5e9e686 100644 --- a/arch/ppc/8xx_io/fec.c +++ b/arch/ppc/8xx_io/fec.c @@ -1459,6 +1459,9 @@ int __init fec_enet_init(void) /* Allocate some private information. */ fep = (struct fec_enet_private *)kmalloc(sizeof(*fep), GFP_KERNEL); + if (fep == NULL) + return -ENOMEM; + __clear_user(fep,sizeof(*fep)); /* Create an Ethernet device instance. diff --git a/arch/ppc/Makefile b/arch/ppc/Makefile index d4172ed4a..488766003 100644 --- a/arch/ppc/Makefile +++ b/arch/ppc/Makefile @@ -145,9 +145,6 @@ endif clean_config: rm -f .config arch/ppc/defconfig -gemini_config: clean_config - cp -f arch/ppc/configs/gemini_defconfig arch/ppc/defconfig - pmac_config: clean_config cp -f arch/ppc/configs/pmac_defconfig arch/ppc/defconfig diff --git a/arch/ppc/boot/Makefile b/arch/ppc/boot/Makefile index 26961e8ea..32c2926a1 100644 --- a/arch/ppc/boot/Makefile +++ b/arch/ppc/boot/Makefile @@ -25,22 +25,12 @@ ZSZ = 0 IOFF = 0 ISZ = 0 -ifeq ($(CONFIG_ALL_PPC),y) -CONFIG_PREP=y -endif - ifeq ($(CONFIG_SMP),y) TFTPIMAGE=/tftpboot/zImage.prep.smp$(MSIZE) else TFTPIMAGE=/tftpboot/zImage.prep$(MSIZE) endif -ifeq ($(CONFIG_SMP),y) -TFTPSIMAGE=/tftpboot/sImage.smp -else -TFTPSIMAGE=/tftpboot/sImage -endif - ifeq ($(CONFIG_PPC64BRIDGE),y) MSIZE=.64 else @@ -81,8 +71,8 @@ zvmlinux.initrd: zvmlinux zvmlinux.initrd.tmp $@ rm zvmlinux.initrd.tmp -zImage: zvmlinux mkprep sImage -ifdef CONFIG_PREP +zImage: zvmlinux mkprep +ifdef CONFIG_ALL_PPC ./mkprep -pbp zvmlinux zImage endif ifdef CONFIG_APUS @@ -90,13 +80,8 @@ ifdef CONFIG_APUS gzip $(GZIP_FLAGS) vmapus endif -sImage: ../../../vmlinux -ifdef CONFIG_GEMINI - $(OBJCOPY) -I elf32-powerpc -O binary ../../../vmlinux sImage -endif - zImage.initrd: zvmlinux.initrd mkprep -ifdef CONFIG_PREP +ifdef CONFIG_ALL_PPC ./mkprep -pbp zvmlinux.initrd zImage.initrd endif @@ -128,18 +113,15 @@ mkprep : mkprep.c $(HOSTCC) -o mkprep mkprep.c znetboot : zImage -ifdef CONFIG_PREP +ifdef CONFIG_ALL_PPC cp zImage $(TFTPIMAGE) endif -ifdef CONFIG_GEMINI - cp sImage $(TFTPSIMAGE) -endif znetboot.initrd : zImage.initrd cp zImage.initrd $(TFTPIMAGE) clean: - rm -f vmlinux* zvmlinux* mkprep zImage* sImage* + rm -f vmlinux* zvmlinux* mkprep zImage* fastdep: $(TOPDIR)/scripts/mkdep *.[Sch] > .depend diff --git a/arch/ppc/boot/vreset.c b/arch/ppc/boot/vreset.c index c01361dc6..dfabd72e7 100644 --- a/arch/ppc/boot/vreset.c +++ b/arch/ppc/boot/vreset.c @@ -19,12 +19,16 @@ */ #include "iso_font.h" -#include <linux/delay.h> extern char *vidmem; extern int lines, cols; -/* estimate for delay */ -unsigned long loops_per_sec = 50000000;; + +static void mdelay(int ms) +{ + for (; ms > 0; --ms) + udelay(1000); +} + /* * VGA Register */ diff --git a/arch/ppc/chrpboot/Makefile b/arch/ppc/chrpboot/Makefile index f7ea3a3fa..9dbb7909d 100644 --- a/arch/ppc/chrpboot/Makefile +++ b/arch/ppc/chrpboot/Makefile @@ -23,11 +23,6 @@ LD_ARGS = -Ttext 0x00400000 OBJS = crt0.o start.o main.o misc.o ../coffboot/string.o ../coffboot/zlib.o image.o LIBS = $(TOPDIR)/lib/lib.a -ifeq ($(CONFIG_ALL_PPC),y) -# yes, we want to build chrp stuff -CONFIG_CHRP = y -endif - ifeq ($(CONFIG_SMP),y) TFTPIMAGE=/tftpboot/zImage.chrp.smp$(MSIZE) else @@ -37,10 +32,10 @@ endif all: $(TOPDIR)/zImage # -# Only build anything here if we're configured for CHRP +# Only build anything here if we're configured for ALL_PPC # -- cort # -ifeq ($(CONFIG_CHRP),y) +ifeq ($(CONFIG_ALL_PPC),y) znetboot: zImage cp zImage $(TFTPIMAGE) @@ -96,6 +91,7 @@ vmlinux.coff.initrd: clean: rm -f piggyback note addnote $(OBJS) zImage + rm -f zImage.rs6k zImage.initrd zImage.initrd.rs6k fastdep: $(TOPDIR)/scripts/mkdep *.[Sch] > .depend diff --git a/arch/ppc/coffboot/Makefile b/arch/ppc/coffboot/Makefile index 494026e42..2835e1f89 100644 --- a/arch/ppc/coffboot/Makefile +++ b/arch/ppc/coffboot/Makefile @@ -20,18 +20,13 @@ else MSIZE= endif -ifeq ($(CONFIG_ALL_PPC),y) -# yes, we want to build pmac stuff -CONFIG_PMAC = y -endif - ifeq ($(CONFIG_SMP),y) TFTPIMAGE=/tftpboot/zImage.pmac.smp$(MSIZE) else TFTPIMAGE=/tftpboot/zImage.pmac$(MSIZE) endif -ifeq ($(CONFIG_PMAC),y) +ifeq ($(CONFIG_ALL_PPC),y) chrpmain.o: chrpmain.c $(CC) $(CFLAGS) -DSYSMAP_OFFSET=0 -DSYSMAP_SIZE=0 -c chrpmain.c diff --git a/arch/ppc/coffboot/coffmain.c b/arch/ppc/coffboot/coffmain.c index b76ba5976..49e769f22 100644 --- a/arch/ppc/coffboot/coffmain.c +++ b/arch/ppc/coffboot/coffmain.c @@ -22,11 +22,18 @@ void gunzip(void *, int, unsigned char *, int *); #define get_32be(x) (*(unsigned *)(x)) #define RAM_START 0xc0000000 -#define PROG_START RAM_START #define RAM_END (RAM_START + 0x800000) /* only 8M mapped with BATs */ +#define PROG_START RAM_START +#define PROG_SIZE 0x00400000 + +#define SCRATCH_SIZE (128 << 10) + char *avail_ram; -char *end_avail; +char *begin_avail, *end_avail; +char *avail_high; +unsigned int heap_use; +unsigned int heap_max; extern char _start[], _end[]; extern char image_data[]; @@ -34,6 +41,7 @@ extern int image_len; extern char initrd_data[]; extern int initrd_len; +char heap[SCRATCH_SIZE]; boot(int a1, int a2, void *prom) { @@ -58,16 +66,18 @@ boot(int a1, int a2, void *prom) im = image_data; len = image_len; /* claim 3MB starting at 0 */ - claim(0, 3 << 20, 0); + claim(0, PROG_SIZE, 0); dst = (void *) RAM_START; if (im[0] == 0x1f && im[1] == 0x8b) { - /* claim 512kB for scratch space */ - avail_ram = claim(0, 512 << 10, 0x10) + RAM_START; - end_avail = avail_ram + (512 << 10); - printf("avail_ram = %x\n", avail_ram); + /* claim some memory for scratch space */ + begin_avail = avail_high = avail_ram = heap; + end_avail = heap + sizeof(heap); + printf("heap at 0x%x\n", avail_ram); printf("gunzipping (0x%x <- 0x%x:0x%0x)...", dst, im, im+len); - gunzip(dst, 3 << 20, im, &len); + gunzip(dst, PROG_SIZE, im, &len); printf("done %u bytes\n", len); + printf("%u bytes of heap consumed, max in use %u\n", + avail_high - begin_avail, heap_max); } else { memmove(dst, im, len); } @@ -78,9 +88,6 @@ boot(int a1, int a2, void *prom) sa = (unsigned long)PROG_START; printf("start address = 0x%x\n", sa); -#if 0 - pause(); -#endif (*(void (*)())sa)(a1, a2, prom); printf("returned?\n"); @@ -114,13 +121,33 @@ void make_bi_recs(unsigned long addr) rec = (struct bi_record *)((unsigned long)rec + rec->size); } +struct memchunk { + unsigned int size; + struct memchunk *next; +}; + +static struct memchunk *freechunks; + void *zalloc(void *x, unsigned items, unsigned size) { - void *p = avail_ram; + void *p; + struct memchunk **mpp, *mp; size *= items; size = (size + 7) & -8; + heap_use += size; + if (heap_use > heap_max) + heap_max = heap_use; + for (mpp = &freechunks; (mp = *mpp) != 0; mpp = &mp->next) { + if (mp->size == size) { + *mpp = mp->next; + return mp; + } + } + p = avail_ram; avail_ram += size; + if (avail_ram > avail_high) + avail_high = avail_ram; if (avail_ram > end_avail) { printf("oops... out of memory\n"); pause(); @@ -130,6 +157,17 @@ void *zalloc(void *x, unsigned items, unsigned size) void zfree(void *x, void *addr, unsigned nb) { + struct memchunk *mp = addr; + + nb = (nb + 7) & -8; + heap_use -= nb; + if (avail_ram == addr + nb) { + avail_ram = addr; + return; + } + mp->size = nb; + mp->next = freechunks; + freechunks = mp; } #define HEAD_CRC 2 diff --git a/arch/ppc/config.in b/arch/ppc/config.in index 34606f99a..3b4c2c141 100644 --- a/arch/ppc/config.in +++ b/arch/ppc/config.in @@ -24,21 +24,24 @@ mainmenu_option next_comment comment 'Platform support' define_bool CONFIG_PPC y choice 'Processor Type' \ - "6xx/7xx/7400 CONFIG_6xx \ + "6xx/7xx/74xx/8260 CONFIG_6xx \ 4xx CONFIG_4xx \ POWER3 CONFIG_POWER3 \ POWER4 CONFIG_POWER4 \ - 8260 CONFIG_8260 \ 8xx CONFIG_8xx" 6xx +if [ "$CONFIG_6xx" = "y" ]; then + bool 'MPC8260 CPM Support' CONFIG_8260 +fi + if [ "$CONFIG_POWER3" = "y" -o "$CONFIG_POWER4" = "y" ]; then define_bool CONFIG_PPC64BRIDGE y define_bool CONFIG_ALL_PPC y fi - + if [ "$CONFIG_8260" = "y" ]; then - define_bool CONFIG_6xx y define_bool CONFIG_SERIAL_CONSOLE y + bool 'Support for EST8260' CONFIG_EST8260 fi if [ "$CONFIG_4xx" = "y" ]; then @@ -59,25 +62,32 @@ if [ "$CONFIG_8xx" = "y" ]; then TQM860 CONFIG_TQM860 \ MBX CONFIG_MBX \ WinCept CONFIG_WINCEPT" RPX-Lite + + if [ "$CONFIG_TQM8xxL" = "y" ]; then + bool 'FPS850 Mainboard' CONFIG_FPS850 + fi fi -if [ "$CONFIG_6xx" = "y" ]; then + +if [ "$CONFIG_6xx" = "y" -a "$CONFIG_8260" = "n" ]; then choice 'Machine Type' \ "PowerMac/PReP/MTX/CHRP CONFIG_ALL_PPC \ - Gemini CONFIG_GEMINI \ - EST8260 CONFIG_EST8260 \ APUS CONFIG_APUS" PowerMac/PReP/MTX/CHRP fi +if [ "$CONFIG_PPC64BRIDGE" != "y" ]; then + bool 'Workarounds for PPC601 bugs' CONFIG_PPC601_SYNC_FIX +fi + if [ "$CONFIG_8xx" = "y" -o "$CONFIG_8260" = "y" ]; then define_bool CONFIG_ALL_PPC n fi -if [ "$CONFIG_TQM8xxL" = "y" ]; then - bool 'FPS850 Mainboard' CONFIG_FPS850 +bool 'Symmetric multi-processing support' CONFIG_SMP +if [ "$CONFIG_SMP" = "y" ]; then + bool ' Distribute interrupts on all CPUs by default' CONFIG_IRQ_ALL_CPUS fi -bool 'Symmetric multi-processing support' CONFIG_SMP -if [ "$CONFIG_6xx" = "y" ];then +if [ "$CONFIG_6xx" = "y" -a "$CONFIG_8260" = "n" ];then bool 'AltiVec Support' CONFIG_ALTIVEC fi @@ -106,14 +116,14 @@ define_bool CONFIG_MCA n if [ "$CONFIG_APUS" = "y" -o "$CONFIG_4xx" = "y" -o \ "$CONFIG_8260" = "y" ]; then - define_bool CONFIG_PCI n + define_bool CONFIG_PCI n else - if [ "$CONFIG_8xx" = "y" ]; then - bool 'QSpan PCI' CONFIG_PCI_QSPAN - define_bool CONFIG_PCI $CONFIG_PCI_QSPAN - else - define_bool CONFIG_PCI y - fi + if [ "$CONFIG_8xx" = "y" ]; then + bool 'QSpan PCI' CONFIG_PCI_QSPAN + define_bool CONFIG_PCI $CONFIG_PCI_QSPAN + else + define_bool CONFIG_PCI y + fi fi bool 'Networking support' CONFIG_NET @@ -123,40 +133,34 @@ bool 'BSD Process Accounting' CONFIG_BSD_PROCESS_ACCT # only elf supported, a.out is not -- Cort if [ "$CONFIG_PROC_FS" = "y" ]; then - define_bool CONFIG_KCORE_ELF y + define_bool CONFIG_KCORE_ELF y fi define_bool CONFIG_BINFMT_ELF y define_bool CONFIG_KERNEL_ELF y tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC source drivers/pci/Config.in -source drivers/zorro/Config.in bool 'Support for hot-pluggable devices' CONFIG_HOTPLUG if [ "$CONFIG_HOTPLUG" = "y" ]; then - source drivers/pcmcia/Config.in + source drivers/pcmcia/Config.in else - define_bool CONFIG_PCMCIA n + define_bool CONFIG_PCMCIA n fi source drivers/parport/Config.in if [ "$CONFIG_4xx" != "y" -a "$CONFIG_8xx" != "y" ]; then - bool 'Support for VGA Console' CONFIG_VGA_CONSOLE - bool 'Support for frame buffer devices' CONFIG_FB - if [ "$CONFIG_FB" = "y" ]; then - bool 'Backward compatibility mode for Xpmac' CONFIG_FB_COMPAT_XPMAC - fi - tristate 'Support for /dev/rtc' CONFIG_PPC_RTC - bool 'Support for Open Firmware device tree in /proc' CONFIG_PROC_DEVICETREE - bool 'Support for early boot text console (BootX only)' CONFIG_BOOTX_TEXT - bool 'Support for Motorola Hot Swap' CONFIG_MOTOROLA_HOTSWAP fi -if [ "$CONFIG_PREP" = "y" -o "$CONFIG_ALL_PPC" = "y" ]; then - bool 'PReP bootloader kernel arguments' CONFIG_CMDLINE_BOOL +if [ "$CONFIG_ALL_PPC" = "y" ]; then + bool 'Support for Open Firmware device tree in /proc' CONFIG_PROC_DEVICETREE + bool 'Support for RTAS (RunTime Abstraction Services) in /proc' CONFIG_PPC_RTAS + bool 'Support for early boot text console (BootX or OpenFirmware only)' CONFIG_BOOTX_TEXT + bool 'Support for PReP Residual Data' CONFIG_PREP_RESIDUAL + bool 'Default bootloader kernel arguments' CONFIG_CMDLINE_BOOL if [ "$CONFIG_CMDLINE_BOOL" = "y" ] ; then string 'Initial kernel command string' CONFIG_CMDLINE "console=ttyS0,9600 console=tty0 root=/dev/sda2" fi @@ -181,6 +185,7 @@ if [ "$CONFIG_APUS" = "y" ]; then fi bool 'Use power LED as a heartbeat' CONFIG_HEARTBEAT bool '/proc/hardware support' CONFIG_PROC_HARDWARE + source drivers/zorro/Config.in fi endmenu @@ -254,7 +259,13 @@ endmenu mainmenu_option next_comment comment 'Console drivers' -source drivers/video/Config.in +if [ "$CONFIG_4xx" != "y" -a "$CONFIG_8xx" != "y" ]; then + bool 'Support for VGA Console' CONFIG_VGA_CONSOLE +fi + source drivers/video/Config.in +if [ "$CONFIG_FB" = "y" -a "$CONFIG_ALL_PPC" = "y" ]; then + bool 'Backward compatibility mode for Xpmac' CONFIG_FB_COMPAT_XPMAC +fi endmenu source drivers/input/Config.in @@ -267,29 +278,26 @@ if [ "$CONFIG_ALL_PPC" = "y" ]; then bool 'Support for CUDA based PowerMacs' CONFIG_ADB_CUDA bool 'Support for PMU based PowerMacs' CONFIG_ADB_PMU if [ "$CONFIG_ADB_PMU" = "y" ]; then - bool ' Power management support for PowerBooks' CONFIG_PMAC_PBOOK - # made a separate option since backlight may end up beeing used - # on non-powerbook machines (but only on PMU based ones AFAIK) - bool ' Backlight control for LCD screens' CONFIG_PMAC_BACKLIGHT + bool ' Power management support for PowerBooks' CONFIG_PMAC_PBOOK + # made a separate option since backlight may end up beeing used + # on non-powerbook machines (but only on PMU based ones AFAIK) + bool ' Backlight control for LCD screens' CONFIG_PMAC_BACKLIGHT fi bool 'Support for PowerMac floppy' CONFIG_MAC_FLOPPY tristate 'Support for PowerMac serial ports' CONFIG_MAC_SERIAL if [ "$CONFIG_MAC_SERIAL" = "y" ]; then - bool ' Support for console on serial port' CONFIG_SERIAL_CONSOLE + bool ' Support for console on serial port' CONFIG_SERIAL_CONSOLE fi bool 'Apple Desktop Bus (ADB) support' CONFIG_ADB if [ "$CONFIG_ADB" = "y" ]; then - bool ' Include MacIO (CHRP) ADB driver' CONFIG_ADB_MACIO - fi -fi -if [ "$CONFIG_ADB" = "y" ]; then - dep_bool ' Use input layer for ADB devices' CONFIG_INPUT_ADBHID $CONFIG_INPUT - if [ "$CONFIG_INPUT_ADBHID" = "y" ]; then - define_bool CONFIG_MAC_HID y - bool ' Support for ADB raw keycodes' CONFIG_MAC_ADBKEYCODES - bool ' Support for mouse button 2+3 emulation' CONFIG_MAC_EMUMOUSEBTN - else - bool ' Support for ADB keyboard (old driver)' CONFIG_ADB_KEYBOARD + bool ' Include MacIO (CHRP) ADB driver' CONFIG_ADB_MACIO + dep_bool ' Use input layer for ADB devices' CONFIG_INPUT_ADBHID $CONFIG_INPUT + if [ "$CONFIG_INPUT_ADBHID" = "y" ]; then + bool ' Support for ADB raw keycodes' CONFIG_MAC_ADBKEYCODES + bool ' Support for mouse button 2+3 emulation' CONFIG_MAC_EMUMOUSEBTN + else + bool ' Support for ADB keyboard (old driver)' CONFIG_ADB_KEYBOARD + fi fi fi endmenu diff --git a/arch/ppc/configs/apus_defconfig b/arch/ppc/configs/apus_defconfig index ed930fff2..69e25a9b3 100644 --- a/arch/ppc/configs/apus_defconfig +++ b/arch/ppc/configs/apus_defconfig @@ -11,9 +11,7 @@ CONFIG_EXPERIMENTAL=y # # Loadable module support # -CONFIG_MODULES=y -# CONFIG_MODVERSIONS is not set -CONFIG_KMOD=y +# CONFIG_MODULES is not set # # Platform support @@ -23,14 +21,14 @@ CONFIG_6xx=y # CONFIG_4xx is not set # CONFIG_POWER3 is not set # CONFIG_POWER4 is not set -# CONFIG_8260 is not set # CONFIG_8xx is not set -CONFIG_ALL_PPC=y -# CONFIG_GEMINI is not set -# CONFIG_EST8260 is not set -# CONFIG_APUS is not set +# CONFIG_8260 is not set +# CONFIG_ALL_PPC is not set +CONFIG_APUS=y +CONFIG_PPC601_SYNC_FIX=y # CONFIG_SMP is not set # CONFIG_ALTIVEC is not set +CONFIG_MACH_SPECIFIC=y # # General setup @@ -38,8 +36,10 @@ CONFIG_ALL_PPC=y # CONFIG_HIGHMEM is not set # CONFIG_MOL is not set # CONFIG_ISA is not set +# CONFIG_EISA is not set # CONFIG_SBUS is not set -CONFIG_PCI=y +# CONFIG_MCA is not set +# CONFIG_PCI is not set CONFIG_NET=y CONFIG_SYSCTL=y CONFIG_SYSVIPC=y @@ -47,23 +47,35 @@ CONFIG_SYSVIPC=y CONFIG_KCORE_ELF=y CONFIG_BINFMT_ELF=y CONFIG_KERNEL_ELF=y -CONFIG_BINFMT_MISC=m -# CONFIG_PCI_NAMES is not set -# CONFIG_HOTPLUG is not set +# CONFIG_BINFMT_MISC is not set +CONFIG_HOTPLUG=y + +# +# PCMCIA/CardBus support +# # CONFIG_PCMCIA is not set # # Parallel port support # # CONFIG_PARPORT is not set -# CONFIG_VGA_CONSOLE is not set +CONFIG_PPC_RTC=y +CONFIG_FB_CONSOLE=y +CONFIG_AMIGA=y +CONFIG_ZORRO=y +CONFIG_AMIGAMOUSE=y +CONFIG_ABSTRACT_CONSOLE=y CONFIG_FB=y -CONFIG_FB_COMPAT_XPMAC=y -# CONFIG_PPC_RTC is not set -CONFIG_PROC_DEVICETREE=y -# CONFIG_BOOTX_TEXT is not set -# CONFIG_MOTOROLA_HOTSWAP is not set -# CONFIG_CMDLINE_BOOL is not set +CONFIG_AMIGA_BUILTIN_SERIAL=y +# CONFIG_M68K_PRINTER is not set +CONFIG_GVPIOEXT=y +# CONFIG_GVPIOEXT_LP is not set +CONFIG_GVPIOEXT_PLIP=y +CONFIG_MULTIFACE_III_TTY=y +# CONFIG_SERIAL_CONSOLE is not set +CONFIG_HEARTBEAT=y +CONFIG_PROC_HARDWARE=y +CONFIG_ZORRO_NAMES=y # # Memory Technology Devices (MTD) @@ -80,12 +92,14 @@ CONFIG_PROC_DEVICETREE=y # Block devices # # CONFIG_BLK_DEV_FD is not set +CONFIG_AMIGA_FLOPPY=y +CONFIG_AMIGA_Z2RAM=y # CONFIG_BLK_DEV_XD is not set # CONFIG_PARIDE is not set # CONFIG_BLK_CPQ_DA is not set # CONFIG_BLK_CPQ_CISS_DA is not set # CONFIG_BLK_DEV_DAC960 is not set -# CONFIG_BLK_DEV_LOOP is not set +CONFIG_BLK_DEV_LOOP=y # CONFIG_BLK_DEV_NBD is not set CONFIG_BLK_DEV_RAM=y CONFIG_BLK_DEV_RAM_SIZE=4096 @@ -106,21 +120,17 @@ CONFIG_BLK_DEV_INITRD=y # # Networking options # -CONFIG_PACKET=y -# CONFIG_PACKET_MMAP is not set -CONFIG_NETLINK=y -# CONFIG_RTNETLINK is not set -# CONFIG_NETLINK_DEV is not set +# CONFIG_PACKET is not set +# CONFIG_NETLINK is not set # CONFIG_NETFILTER is not set # CONFIG_FILTER is not set CONFIG_UNIX=y CONFIG_INET=y -CONFIG_IP_MULTICAST=y +# CONFIG_IP_MULTICAST is not set # CONFIG_IP_ADVANCED_ROUTER is not set # CONFIG_IP_PNP is not set # CONFIG_NET_IPIP is not set # CONFIG_NET_IPGRE is not set -# CONFIG_IP_MROUTE is not set # CONFIG_INET_ECN is not set # CONFIG_SYN_COOKIES is not set # CONFIG_IPV6 is not set @@ -131,7 +141,7 @@ CONFIG_IP_MULTICAST=y # # # CONFIG_IPX is not set -CONFIG_ATALK=m +# CONFIG_ATALK is not set # CONFIG_DECNET is not set # CONFIG_BRIDGE is not set # CONFIG_X25 is not set @@ -151,9 +161,48 @@ CONFIG_ATALK=m # # ATA/IDE/MFM/RLL support # -# CONFIG_IDE is not set -# CONFIG_BLK_DEV_IDE_MODES is not set +CONFIG_IDE=y + +# +# IDE, ATA and ATAPI Block devices +# +CONFIG_BLK_DEV_IDE=y + +# +# Please see Documentation/ide.txt for help/info on IDE drives +# +# CONFIG_BLK_DEV_HD_IDE is not set # CONFIG_BLK_DEV_HD is not set +CONFIG_BLK_DEV_IDEDISK=y +# CONFIG_IDEDISK_MULTI_MODE is not set +# CONFIG_BLK_DEV_IDEDISK_VENDOR is not set +# CONFIG_BLK_DEV_IDEDISK_FUJITSU is not set +# CONFIG_BLK_DEV_IDEDISK_IBM is not set +# CONFIG_BLK_DEV_IDEDISK_MAXTOR is not set +# CONFIG_BLK_DEV_IDEDISK_QUANTUM is not set +# CONFIG_BLK_DEV_IDEDISK_SEAGATE is not set +# CONFIG_BLK_DEV_IDEDISK_WD is not set +# CONFIG_BLK_DEV_COMMERIAL is not set +# CONFIG_BLK_DEV_TIVO is not set +# CONFIG_BLK_DEV_IDECS is not set +CONFIG_BLK_DEV_IDECD=y +# CONFIG_BLK_DEV_IDETAPE is not set +# CONFIG_BLK_DEV_IDEFLOPPY is not set +# CONFIG_BLK_DEV_IDESCSI is not set + +# +# IDE chipset support/bugfixes +# +# CONFIG_BLK_DEV_CMD640 is not set +# CONFIG_BLK_DEV_CMD640_ENHANCED is not set +# CONFIG_BLK_DEV_ISAPNP is not set +CONFIG_BLK_DEV_GAYLE=y +CONFIG_BLK_DEV_IDEDOUBLER=y +CONFIG_BLK_DEV_BUDDHA=y +# CONFIG_IDE_CHIPSETS is not set +# CONFIG_IDEDMA_AUTO is not set +# CONFIG_DMA_NONPCI is not set +# CONFIG_BLK_DEV_IDE_MODES is not set # # SCSI support @@ -165,7 +214,8 @@ CONFIG_SCSI=y # CONFIG_BLK_DEV_SD=y CONFIG_SD_EXTRA_DEVS=40 -CONFIG_CHR_DEV_ST=y +# CONFIG_CHR_DEV_ST is not set +# CONFIG_CHR_DEV_OSST is not set CONFIG_BLK_DEV_SR=y CONFIG_BLK_DEV_SR_VENDOR=y CONFIG_SR_EXTRA_DEVS=2 @@ -176,29 +226,23 @@ CONFIG_SR_EXTRA_DEVS=2 # # CONFIG_SCSI_DEBUG_QUEUES is not set # CONFIG_SCSI_MULTI_LUN is not set -CONFIG_SCSI_CONSTANTS=y +# CONFIG_SCSI_CONSTANTS is not set # CONFIG_SCSI_LOGGING is not set # # SCSI low-level drivers # -# CONFIG_BLK_DEV_3W_XXXX_RAID is not set # CONFIG_SCSI_7000FASST is not set # CONFIG_SCSI_ACARD is not set # CONFIG_SCSI_AHA152X is not set # CONFIG_SCSI_AHA1542 is not set # CONFIG_SCSI_AHA1740 is not set -CONFIG_SCSI_AIC7XXX=y -# CONFIG_AIC7XXX_TCQ_ON_BY_DEFAULT is not set -CONFIG_AIC7XXX_CMDS_PER_DEVICE=8 -CONFIG_AIC7XXX_PROC_STATS=y -CONFIG_AIC7XXX_RESET_DELAY=15 +# CONFIG_SCSI_AIC7XXX is not set # CONFIG_SCSI_ADVANSYS is not set # CONFIG_SCSI_IN2000 is not set # CONFIG_SCSI_AM53C974 is not set # CONFIG_SCSI_MEGARAID is not set # CONFIG_SCSI_BUSLOGIC is not set -# CONFIG_SCSI_CPQFCTS is not set # CONFIG_SCSI_DMX3191D is not set # CONFIG_SCSI_DTC3280 is not set # CONFIG_SCSI_EATA is not set @@ -211,30 +255,18 @@ CONFIG_AIC7XXX_RESET_DELAY=15 # CONFIG_SCSI_INIA100 is not set # CONFIG_SCSI_NCR53C406A is not set # CONFIG_SCSI_NCR53C7xx is not set -# CONFIG_SCSI_NCR53C8XX is not set -# CONFIG_SCSI_SYM53C8XX is not set # CONFIG_SCSI_PAS16 is not set # CONFIG_SCSI_PCI2000 is not set # CONFIG_SCSI_PCI2220I is not set # CONFIG_SCSI_PSI240I is not set # CONFIG_SCSI_QLOGIC_FAS is not set -# CONFIG_SCSI_QLOGIC_ISP is not set -# CONFIG_SCSI_QLOGIC_FC is not set -# CONFIG_SCSI_QLOGIC_1280 is not set # CONFIG_SCSI_SIM710 is not set # CONFIG_SCSI_SYM53C416 is not set -# CONFIG_SCSI_DC390T is not set # CONFIG_SCSI_T128 is not set # CONFIG_SCSI_U14_34F is not set # CONFIG_SCSI_DEBUG is not set -CONFIG_SCSI_MESH=y -CONFIG_SCSI_MESH_SYNC_RATE=5 -CONFIG_SCSI_MAC53C94=y - -# -# IEEE 1394 (FireWire) support -# -# CONFIG_IEEE1394 is not set +# CONFIG_SCSI_MESH is not set +# CONFIG_SCSI_MAC53C94 is not set # # Network device support @@ -245,34 +277,30 @@ CONFIG_NETDEVICES=y # ARCnet devices # # CONFIG_ARCNET is not set - -# -# Appletalk devices -# -# CONFIG_APPLETALK is not set # CONFIG_DUMMY is not set # CONFIG_BONDING is not set # CONFIG_EQUALIZER is not set # CONFIG_TUN is not set -# CONFIG_ETHERTAP is not set # CONFIG_NET_SB1000 is not set # # Ethernet (10 or 100Mbit) # CONFIG_NET_ETHERNET=y -CONFIG_MACE=y -CONFIG_BMAC=y +# CONFIG_MACE is not set +# CONFIG_BMAC is not set # CONFIG_GMAC is not set -# CONFIG_NCR885E is not set # CONFIG_OAKNET is not set +CONFIG_ARIADNE=y +# CONFIG_NE2K_ZORRO is not set +CONFIG_A2065=y +CONFIG_HYDRA=y # CONFIG_NET_VENDOR_3COM is not set # CONFIG_LANCE is not set # CONFIG_NET_VENDOR_SMC is not set # CONFIG_NET_VENDOR_RACAL is not set # CONFIG_AT1700 is not set # CONFIG_DEPCA is not set -# CONFIG_HP100 is not set # CONFIG_NET_ISA is not set # CONFIG_NET_PCI is not set # CONFIG_NET_POCKET is not set @@ -287,13 +315,16 @@ CONFIG_BMAC=y # CONFIG_FDDI is not set # CONFIG_HIPPI is not set CONFIG_PPP=y -# CONFIG_PPP_MULTILINK is not set -# CONFIG_PPP_ASYNC is not set -# CONFIG_PPP_SYNC_TTY is not set -# CONFIG_PPP_DEFLATE is not set -# CONFIG_PPP_BSDCOMP is not set +CONFIG_PPP_MULTILINK=y +CONFIG_PPP_ASYNC=y +CONFIG_PPP_SYNC_TTY=y +CONFIG_PPP_DEFLATE=y +CONFIG_PPP_BSDCOMP=y # CONFIG_PPPOE is not set -# CONFIG_SLIP is not set +CONFIG_SLIP=y +CONFIG_SLIP_COMPRESSED=y +CONFIG_SLIP_SMART=y +# CONFIG_SLIP_MODE_SLIP6 is not set # # Wireless LAN (non-hamradio) @@ -336,44 +367,47 @@ CONFIG_PPP=y # # Console drivers # +# CONFIG_VGA_CONSOLE is not set # # Frame-buffer support # CONFIG_FB=y CONFIG_DUMMY_CONSOLE=y -# CONFIG_FB_RIVA is not set -# CONFIG_FB_CLGEN is not set -# CONFIG_FB_PM2 is not set +CONFIG_FB_CLGEN=y +CONFIG_FB_PM2=y +CONFIG_FB_PM2_CVPPC=y # CONFIG_FB_CYBER2000 is not set -CONFIG_FB_OF=y -CONFIG_FB_CONTROL=y -CONFIG_FB_PLATINUM=y -CONFIG_FB_VALKYRIE=y -CONFIG_FB_CT65550=y -CONFIG_FB_IMSTT=y +CONFIG_FB_AMIGA=y +CONFIG_FB_AMIGA_OCS=y +CONFIG_FB_AMIGA_ECS=y +CONFIG_FB_AMIGA_AGA=y +CONFIG_FB_CYBER=y +CONFIG_FB_VIRGE=y +CONFIG_FB_RETINAZ3=y +# CONFIG_FB_FM2 is not set +# CONFIG_FB_OF is not set +# CONFIG_FB_CONTROL is not set +# CONFIG_FB_PLATINUM is not set +# CONFIG_FB_VALKYRIE is not set +# CONFIG_FB_CT65550 is not set +# CONFIG_FB_IMSTT is not set # CONFIG_FB_S3TRIO is not set # CONFIG_FB_VGA16 is not set -# CONFIG_FB_MATROX is not set -CONFIG_FB_ATY=y -# CONFIG_FB_ATY128 is not set -# CONFIG_FB_3DFX is not set -# CONFIG_FB_SIS is not set # CONFIG_FB_VIRTUAL is not set # CONFIG_FBCON_ADVANCED is not set +CONFIG_FBCON_MFB=y CONFIG_FBCON_CFB8=y CONFIG_FBCON_CFB16=y CONFIG_FBCON_CFB24=y CONFIG_FBCON_CFB32=y -# CONFIG_FBCON_FONTWIDTH8_ONLY is not set -CONFIG_FBCON_FONTS=y -# CONFIG_FONT_8x8 is not set +CONFIG_FBCON_AFB=y +CONFIG_FBCON_ILBM=y +CONFIG_FBCON_FONTWIDTH8_ONLY=y +# CONFIG_FBCON_FONTS is not set +CONFIG_FONT_8x8=y CONFIG_FONT_8x16=y -CONFIG_FONT_SUN8x16=y -CONFIG_FONT_SUN12x22=y -# CONFIG_FONT_6x11 is not set -# CONFIG_FONT_PEARL_8x8 is not set -# CONFIG_FONT_ACORN_8x8 is not set +CONFIG_FONT_PEARL_8x8=y # # Input core support @@ -383,23 +417,17 @@ CONFIG_FONT_SUN12x22=y # # Macintosh device drivers # -# CONFIG_ADB_CUDA is not set -# CONFIG_ADB_PMU is not set -CONFIG_MAC_FLOPPY=y -CONFIG_MAC_SERIAL=y -# CONFIG_SERIAL_CONSOLE is not set -# CONFIG_ADB is not set # # Character devices # CONFIG_VT=y CONFIG_VT_CONSOLE=y -CONFIG_SERIAL=m +CONFIG_SERIAL=y +# CONFIG_SERIAL_CONSOLE is not set # CONFIG_SERIAL_EXTENDED is not set # CONFIG_SERIAL_NONSTANDARD is not set -CONFIG_UNIX98_PTYS=y -CONFIG_UNIX98_PTY_COUNT=256 +# CONFIG_UNIX98_PTYS is not set # # I2C support @@ -427,7 +455,7 @@ CONFIG_UNIX98_PTY_COUNT=256 # # CONFIG_WATCHDOG is not set # CONFIG_INTEL_RNG is not set -CONFIG_NVRAM=y +# CONFIG_NVRAM is not set # CONFIG_RTC is not set # CONFIG_DTLK is not set # CONFIG_R3964 is not set @@ -449,24 +477,24 @@ CONFIG_NVRAM=y # File systems # # CONFIG_QUOTA is not set -CONFIG_AUTOFS_FS=y +# CONFIG_AUTOFS_FS is not set # CONFIG_AUTOFS4_FS is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set -# CONFIG_AFFS_FS is not set +CONFIG_AFFS_FS=y CONFIG_HFS_FS=y # CONFIG_BFS_FS is not set -CONFIG_FAT_FS=m -CONFIG_MSDOS_FS=m +CONFIG_FAT_FS=y +CONFIG_MSDOS_FS=y # CONFIG_UMSDOS_FS is not set -CONFIG_VFAT_FS=m +CONFIG_VFAT_FS=y # CONFIG_EFS_FS is not set # CONFIG_JFFS_FS is not set # CONFIG_CRAMFS is not set # CONFIG_RAMFS is not set CONFIG_ISO9660_FS=y # CONFIG_JOLIET is not set -# CONFIG_MINIX_FS is not set +CONFIG_MINIX_FS=y # CONFIG_NTFS_FS is not set # CONFIG_NTFS_RW is not set # CONFIG_HPFS_FS is not set @@ -474,7 +502,7 @@ CONFIG_PROC_FS=y # CONFIG_DEVFS_FS is not set # CONFIG_DEVFS_MOUNT is not set # CONFIG_DEVFS_DEBUG is not set -CONFIG_DEVPTS_FS=y +# CONFIG_DEVPTS_FS is not set # CONFIG_QNX4FS_FS is not set # CONFIG_QNX4FS_RW is not set # CONFIG_ROMFS_FS is not set @@ -493,7 +521,7 @@ CONFIG_EXT2_FS=y CONFIG_NFS_FS=y # CONFIG_NFS_V3 is not set # CONFIG_ROOT_NFS is not set -CONFIG_NFSD=y +# CONFIG_NFSD is not set # CONFIG_NFSD_V3 is not set CONFIG_SUNRPC=y CONFIG_LOCKD=y @@ -505,8 +533,6 @@ CONFIG_LOCKD=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -514,14 +540,15 @@ CONFIG_LOCKD=y # Partition Types # # CONFIG_PARTITION_ADVANCED is not set -CONFIG_MSDOS_PARTITION=y +CONFIG_AMIGA_PARTITION=y +# CONFIG_SMB_NLS is not set CONFIG_NLS=y # # Native Language Support # CONFIG_NLS_DEFAULT="iso8859-1" -CONFIG_NLS_CODEPAGE_437=y +# CONFIG_NLS_CODEPAGE_437 is not set # CONFIG_NLS_CODEPAGE_737 is not set # CONFIG_NLS_CODEPAGE_775 is not set # CONFIG_NLS_CODEPAGE_850 is not set @@ -541,7 +568,7 @@ CONFIG_NLS_CODEPAGE_437=y # CONFIG_NLS_CODEPAGE_936 is not set # CONFIG_NLS_CODEPAGE_949 is not set # CONFIG_NLS_CODEPAGE_950 is not set -# CONFIG_NLS_ISO8859_1 is not set +CONFIG_NLS_ISO8859_1=y # CONFIG_NLS_ISO8859_2 is not set # CONFIG_NLS_ISO8859_3 is not set # CONFIG_NLS_ISO8859_4 is not set @@ -558,23 +585,7 @@ CONFIG_NLS_CODEPAGE_437=y # # Sound # -CONFIG_SOUND=y -# CONFIG_DMASOUND_AWACS is not set -# CONFIG_SOUND_CMPCI is not set -# CONFIG_SOUND_EMU10K1 is not set -# CONFIG_SOUND_FUSION is not set -# CONFIG_SOUND_CS4281 is not set -# CONFIG_SOUND_ES1370 is not set -# CONFIG_SOUND_ES1371 is not set -# CONFIG_SOUND_ESSSOLO1 is not set -# CONFIG_SOUND_MAESTRO is not set -# CONFIG_SOUND_SONICVIBES is not set -# CONFIG_SOUND_TRIDENT is not set -# CONFIG_SOUND_MSNDCLAS is not set -# CONFIG_SOUND_MSNDPIN is not set -# CONFIG_SOUND_VIA82CXXX is not set -# CONFIG_SOUND_OSS is not set -# CONFIG_SOUND_TVMIXER is not set +# CONFIG_SOUND is not set # # USB support diff --git a/arch/ppc/configs/bseip_defconfig b/arch/ppc/configs/bseip_defconfig index f0ff611f4..82b7174bd 100644 --- a/arch/ppc/configs/bseip_defconfig +++ b/arch/ppc/configs/bseip_defconfig @@ -21,7 +21,6 @@ CONFIG_PPC=y # CONFIG_4xx is not set # CONFIG_POWER3 is not set # CONFIG_POWER4 is not set -# CONFIG_8260 is not set CONFIG_8xx=y CONFIG_SERIAL_CONSOLE=y # CONFIG_RPXLITE is not set @@ -32,6 +31,7 @@ CONFIG_BSEIP=y # CONFIG_TQM860 is not set # CONFIG_MBX is not set # CONFIG_WINCEPT is not set +CONFIG_PPC601_SYNC_FIX=y # CONFIG_ALL_PPC is not set # CONFIG_SMP is not set CONFIG_MACH_SPECIFIC=y @@ -43,7 +43,9 @@ CONFIG_MATH_EMULATION=y # CONFIG_HIGHMEM is not set # CONFIG_MOL is not set # CONFIG_ISA is not set +# CONFIG_EISA is not set # CONFIG_SBUS is not set +# CONFIG_MCA is not set # CONFIG_PCI_QSPAN is not set # CONFIG_PCI is not set CONFIG_NET=y @@ -182,7 +184,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_MACE is not set # CONFIG_BMAC is not set # CONFIG_GMAC is not set -# CONFIG_NCR885E is not set # CONFIG_OAKNET is not set # CONFIG_NET_VENDOR_3COM is not set # CONFIG_LANCE is not set @@ -190,7 +191,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_NET_VENDOR_RACAL is not set # CONFIG_AT1700 is not set # CONFIG_DEPCA is not set -# CONFIG_HP100 is not set # CONFIG_NET_ISA is not set # CONFIG_NET_PCI is not set # CONFIG_NET_POCKET is not set @@ -377,8 +377,6 @@ CONFIG_LOCKD=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -395,6 +393,7 @@ CONFIG_PARTITION_ADVANCED=y # CONFIG_SGI_PARTITION is not set # CONFIG_ULTRIX_PARTITION is not set # CONFIG_SUN_PARTITION is not set +# CONFIG_SMB_NLS is not set # CONFIG_NLS is not set # diff --git a/arch/ppc/configs/common_defconfig b/arch/ppc/configs/common_defconfig index 5574ecbdb..210887de7 100644 --- a/arch/ppc/configs/common_defconfig +++ b/arch/ppc/configs/common_defconfig @@ -23,12 +23,11 @@ CONFIG_6xx=y # CONFIG_4xx is not set # CONFIG_POWER3 is not set # CONFIG_POWER4 is not set -# CONFIG_8260 is not set # CONFIG_8xx is not set +# CONFIG_8260 is not set CONFIG_ALL_PPC=y -# CONFIG_GEMINI is not set -# CONFIG_EST8260 is not set # CONFIG_APUS is not set +CONFIG_PPC601_SYNC_FIX=y # CONFIG_SMP is not set CONFIG_ALTIVEC=y @@ -36,9 +35,11 @@ CONFIG_ALTIVEC=y # General setup # # CONFIG_HIGHMEM is not set -# CONFIG_MOL is not set +CONFIG_MOL=y # CONFIG_ISA is not set +# CONFIG_EISA is not set # CONFIG_SBUS is not set +# CONFIG_MCA is not set CONFIG_PCI=y CONFIG_NET=y CONFIG_SYSCTL=y @@ -49,21 +50,24 @@ CONFIG_BINFMT_ELF=y CONFIG_KERNEL_ELF=y CONFIG_BINFMT_MISC=m CONFIG_PCI_NAMES=y -# CONFIG_HOTPLUG is not set +CONFIG_HOTPLUG=y + +# +# PCMCIA/CardBus support +# # CONFIG_PCMCIA is not set # # Parallel port support # # CONFIG_PARPORT is not set -CONFIG_VGA_CONSOLE=y -CONFIG_FB=y -CONFIG_FB_COMPAT_XPMAC=y CONFIG_PPC_RTC=y CONFIG_PROC_DEVICETREE=y +CONFIG_PPC_RTAS=y CONFIG_BOOTX_TEXT=y -# CONFIG_MOTOROLA_HOTSWAP is not set -# CONFIG_CMDLINE_BOOL is not set +CONFIG_PREP_RESIDUAL=y +CONFIG_CMDLINE_BOOL=y +CONFIG_CMDLINE="console=ttyS0,9600 console=tty0 root=/dev/sda2" # # Memory Technology Devices (MTD) @@ -79,7 +83,7 @@ CONFIG_BOOTX_TEXT=y # # Block devices # -# CONFIG_BLK_DEV_FD is not set +CONFIG_BLK_DEV_FD=m # CONFIG_BLK_DEV_XD is not set # CONFIG_PARIDE is not set # CONFIG_BLK_CPQ_DA is not set @@ -111,7 +115,8 @@ CONFIG_PACKET=y CONFIG_NETLINK=y # CONFIG_RTNETLINK is not set # CONFIG_NETLINK_DEV is not set -# CONFIG_NETFILTER is not set +CONFIG_NETFILTER=y +# CONFIG_NETFILTER_DEBUG is not set # CONFIG_FILTER is not set CONFIG_UNIX=y CONFIG_INET=y @@ -123,6 +128,34 @@ CONFIG_IP_MULTICAST=y # CONFIG_IP_MROUTE is not set # CONFIG_INET_ECN is not set CONFIG_SYN_COOKIES=y + +# +# IP: Netfilter Configuration +# +CONFIG_IP_NF_CONNTRACK=m +CONFIG_IP_NF_FTP=m +# CONFIG_IP_NF_QUEUE is not set +CONFIG_IP_NF_IPTABLES=m +CONFIG_IP_NF_MATCH_LIMIT=m +CONFIG_IP_NF_MATCH_MAC=m +CONFIG_IP_NF_MATCH_MARK=m +CONFIG_IP_NF_MATCH_MULTIPORT=m +CONFIG_IP_NF_MATCH_TOS=m +CONFIG_IP_NF_MATCH_STATE=m +CONFIG_IP_NF_MATCH_UNCLEAN=m +CONFIG_IP_NF_MATCH_OWNER=m +CONFIG_IP_NF_FILTER=m +CONFIG_IP_NF_TARGET_REJECT=m +CONFIG_IP_NF_TARGET_MIRROR=m +CONFIG_IP_NF_NAT=m +CONFIG_IP_NF_NAT_NEEDED=y +CONFIG_IP_NF_TARGET_MASQUERADE=m +CONFIG_IP_NF_TARGET_REDIRECT=m +# CONFIG_IP_NF_MANGLE is not set +# CONFIG_IP_NF_TARGET_LOG is not set +CONFIG_IP_NF_COMPAT_IPCHAINS=m +CONFIG_IP_NF_NAT_NEEDED=y +# CONFIG_IP_NF_COMPAT_IPFWADM is not set # CONFIG_IPV6 is not set # CONFIG_KHTTPD is not set # CONFIG_ATM is not set @@ -239,6 +272,7 @@ CONFIG_SCSI=y CONFIG_BLK_DEV_SD=y CONFIG_SD_EXTRA_DEVS=40 CONFIG_CHR_DEV_ST=y +# CONFIG_CHR_DEV_OSST is not set CONFIG_BLK_DEV_SR=y CONFIG_BLK_DEV_SR_VENDOR=y CONFIG_SR_EXTRA_DEVS=2 @@ -344,7 +378,6 @@ CONFIG_NET_ETHERNET=y CONFIG_MACE=y CONFIG_BMAC=y CONFIG_GMAC=y -# CONFIG_NCR885E is not set # CONFIG_OAKNET is not set # CONFIG_NET_VENDOR_3COM is not set # CONFIG_LANCE is not set @@ -357,27 +390,28 @@ CONFIG_GMAC=y CONFIG_NET_PCI=y CONFIG_PCNET32=y # CONFIG_ADAPTEC_STARFIRE is not set -# CONFIG_AC3200 is not set # CONFIG_APRICOT is not set # CONFIG_CS89x0 is not set -CONFIG_DE4X5=y -# CONFIG_TULIP is not set +CONFIG_TULIP=y +CONFIG_DE4X5=m # CONFIG_DGRS is not set # CONFIG_DM9102 is not set # CONFIG_EEPRO100 is not set +# CONFIG_EEPRO100_PM is not set # CONFIG_LNE390 is not set # CONFIG_NATSEMI is not set # CONFIG_NE2K_PCI is not set # CONFIG_NE3210 is not set # CONFIG_ES3210 is not set -# CONFIG_RTL8129 is not set # CONFIG_8139TOO is not set +# CONFIG_RTL8129 is not set # CONFIG_SIS900 is not set # CONFIG_EPIC100 is not set # CONFIG_SUNDANCE is not set # CONFIG_TLAN is not set # CONFIG_VIA_RHINE is not set # CONFIG_WINBOND_840 is not set +# CONFIG_HAPPYMEAL is not set # CONFIG_NET_POCKET is not set # @@ -439,6 +473,7 @@ CONFIG_PPP_DEFLATE=y # # Console drivers # +CONFIG_VGA_CONSOLE=y # # Frame-buffer support @@ -460,7 +495,8 @@ CONFIG_FB_IMSTT=y CONFIG_FB_MATROX=y CONFIG_FB_MATROX_MILLENIUM=y CONFIG_FB_MATROX_MYSTIQUE=y -CONFIG_FB_MATROX_G100=y +# CONFIG_FB_MATROX_G100 is not set +# CONFIG_FB_MATROX_G450 is not set # CONFIG_FB_MATROX_MULTIHEAD is not set CONFIG_FB_ATY=y CONFIG_FB_ATY128=y @@ -481,6 +517,7 @@ CONFIG_FONT_SUN12x22=y # CONFIG_FONT_6x11 is not set # CONFIG_FONT_PEARL_8x8 is not set # CONFIG_FONT_ACORN_8x8 is not set +# CONFIG_FB_COMPAT_XPMAC is not set # # Input core support @@ -498,15 +535,13 @@ CONFIG_INPUT_EVDEV=y # CONFIG_ADB_CUDA=y CONFIG_ADB_PMU=y -CONFIG_PMAC_PBOOK=y -CONFIG_PMAC_BACKLIGHT=y -CONFIG_MAC_FLOPPY=y -CONFIG_MAC_SERIAL=y -# CONFIG_SERIAL_CONSOLE is not set +# CONFIG_PMAC_PBOOK is not set +# CONFIG_PMAC_BACKLIGHT is not set +# CONFIG_MAC_FLOPPY is not set +CONFIG_MAC_SERIAL=m CONFIG_ADB=y CONFIG_ADB_MACIO=y CONFIG_INPUT_ADBHID=y -CONFIG_MAC_HID=y CONFIG_MAC_ADBKEYCODES=y CONFIG_MAC_EMUMOUSEBTN=y @@ -575,17 +610,17 @@ CONFIG_NVRAM=y # File systems # # CONFIG_QUOTA is not set -CONFIG_AUTOFS_FS=y -CONFIG_AUTOFS4_FS=y +# CONFIG_AUTOFS_FS is not set +# CONFIG_AUTOFS4_FS is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set -# CONFIG_HFS_FS is not set +CONFIG_HFS_FS=m # CONFIG_BFS_FS is not set -CONFIG_FAT_FS=y -CONFIG_MSDOS_FS=y +CONFIG_FAT_FS=m +CONFIG_MSDOS_FS=m # CONFIG_UMSDOS_FS is not set -CONFIG_VFAT_FS=y +CONFIG_VFAT_FS=m # CONFIG_EFS_FS is not set # CONFIG_JFFS_FS is not set # CONFIG_CRAMFS is not set @@ -631,8 +666,6 @@ CONFIG_LOCKD=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -652,6 +685,7 @@ CONFIG_MSDOS_PARTITION=y # CONFIG_SGI_PARTITION is not set # CONFIG_ULTRIX_PARTITION is not set # CONFIG_SUN_PARTITION is not set +# CONFIG_SMB_NLS is not set CONFIG_NLS=y # @@ -678,7 +712,7 @@ CONFIG_NLS_DEFAULT="iso8859-1" # CONFIG_NLS_CODEPAGE_936 is not set # CONFIG_NLS_CODEPAGE_949 is not set # CONFIG_NLS_CODEPAGE_950 is not set -# CONFIG_NLS_ISO8859_1 is not set +CONFIG_NLS_ISO8859_1=m # CONFIG_NLS_ISO8859_2 is not set # CONFIG_NLS_ISO8859_3 is not set # CONFIG_NLS_ISO8859_4 is not set @@ -695,9 +729,9 @@ CONFIG_NLS_DEFAULT="iso8859-1" # # Sound # -CONFIG_SOUND=y -CONFIG_DMASOUND_AWACS=y -CONFIG_DMASOUND=y +CONFIG_SOUND=m +CONFIG_DMASOUND_AWACS=m +CONFIG_DMASOUND=m # CONFIG_SOUND_CMPCI is not set # CONFIG_SOUND_EMU10K1 is not set # CONFIG_SOUND_FUSION is not set @@ -711,43 +745,14 @@ CONFIG_DMASOUND=y # CONFIG_SOUND_MSNDCLAS is not set # CONFIG_SOUND_MSNDPIN is not set # CONFIG_SOUND_VIA82CXXX is not set -CONFIG_SOUND_OSS=y -# CONFIG_SOUND_TRACEINIT is not set -# CONFIG_SOUND_DMAP is not set -# CONFIG_SOUND_AD1816 is not set -# CONFIG_SOUND_SGALAXY is not set -# CONFIG_SOUND_ADLIB is not set -# CONFIG_SOUND_ACI_MIXER is not set -CONFIG_SOUND_CS4232=m -# CONFIG_SOUND_SSCAPE is not set -# CONFIG_SOUND_GUS is not set -# CONFIG_SOUND_ICH is not set -# CONFIG_SOUND_VMIDI is not set -# CONFIG_SOUND_TRIX is not set -# CONFIG_SOUND_MSS is not set -# CONFIG_SOUND_MPU401 is not set -# CONFIG_SOUND_NM256 is not set -# CONFIG_SOUND_MAD16 is not set -# CONFIG_SOUND_PAS is not set -# CONFIG_PAS_JOYSTICK is not set -# CONFIG_SOUND_PSS is not set -# CONFIG_SOUND_SB is not set -# CONFIG_SOUND_AWE32_SYNTH is not set -# CONFIG_SOUND_WAVEFRONT is not set -# CONFIG_SOUND_MAUI is not set -# CONFIG_SOUND_YM3812 is not set -# CONFIG_SOUND_OPL3SA1 is not set -# CONFIG_SOUND_OPL3SA2 is not set -# CONFIG_SOUND_YMPCI is not set -# CONFIG_SOUND_UART6850 is not set -# CONFIG_SOUND_AEDSP16 is not set +# CONFIG_SOUND_OSS is not set # CONFIG_SOUND_TVMIXER is not set # # USB support # CONFIG_USB=y -CONFIG_USB_DEBUG=y +# CONFIG_USB_DEBUG is not set # # Miscellaneous USB options @@ -763,33 +768,69 @@ CONFIG_USB_DEVICEFS=y CONFIG_USB_OHCI=y # -# USB Devices +# USB Device Class drivers # +# CONFIG_USB_AUDIO is not set +# CONFIG_USB_BLUETOOTH is not set +# CONFIG_USB_STORAGE is not set +CONFIG_USB_ACM=m # CONFIG_USB_PRINTER is not set + +# +# USB Human Interface Devices (HID) +# +CONFIG_USB_HID=y +# CONFIG_USB_WACOM is not set + +# +# USB Imaging devices +# +# CONFIG_USB_DC2XX is not set +# CONFIG_USB_MDC800 is not set # CONFIG_USB_SCANNER is not set # CONFIG_USB_MICROTEK is not set -# CONFIG_USB_AUDIO is not set -# CONFIG_USB_ACM is not set -# CONFIG_USB_SERIAL is not set + +# +# USB Multimedia devices +# # CONFIG_USB_IBMCAM is not set # CONFIG_USB_OV511 is not set -# CONFIG_USB_DC2XX is not set -# CONFIG_USB_MDC800 is not set -# CONFIG_USB_STORAGE is not set -# CONFIG_USB_USS720 is not set +# CONFIG_USB_DSBR is not set # CONFIG_USB_DABUSB is not set + +# +# USB Network adaptors +# # CONFIG_USB_PLUSB is not set # CONFIG_USB_PEGASUS is not set -# CONFIG_USB_RIO500 is not set -# CONFIG_USB_DSBR is not set -# CONFIG_USB_BLUETOOTH is not set # CONFIG_USB_NET1080 is not set # -# USB Human Interface Devices (HID) +# USB port drivers # -CONFIG_USB_HID=y -# CONFIG_USB_WACOM is not set +# CONFIG_USB_USS720 is not set + +# +# USB Serial Converter support +# +CONFIG_USB_SERIAL=m +# CONFIG_USB_SERIAL_DEBUG is not set +# CONFIG_USB_SERIAL_GENERIC is not set +# CONFIG_USB_SERIAL_BELKIN is not set +# CONFIG_USB_SERIAL_WHITEHEAT is not set +# CONFIG_USB_SERIAL_DIGI_ACCELEPORT is not set +# CONFIG_USB_SERIAL_EMPEG is not set +# CONFIG_USB_SERIAL_FTDI_SIO is not set +CONFIG_USB_SERIAL_VISOR=m +# CONFIG_USB_SERIAL_KEYSPAN_PDA is not set +# CONFIG_USB_SERIAL_KEYSPAN is not set +# CONFIG_USB_SERIAL_MCT_U232 is not set +# CONFIG_USB_SERIAL_OMNINET is not set + +# +# USB misc drivers +# +# CONFIG_USB_RIO500 is not set # # Kernel hacking diff --git a/arch/ppc/configs/est8260_defconfig b/arch/ppc/configs/est8260_defconfig index afb972c03..a4a44f0ab 100644 --- a/arch/ppc/configs/est8260_defconfig +++ b/arch/ppc/configs/est8260_defconfig @@ -17,21 +17,17 @@ CONFIG_EXPERIMENTAL=y # Platform support # CONFIG_PPC=y -# CONFIG_6xx is not set +CONFIG_6xx=y # CONFIG_4xx is not set # CONFIG_POWER3 is not set # CONFIG_POWER4 is not set -CONFIG_8260=y # CONFIG_8xx is not set -CONFIG_6xx=y +CONFIG_8260=y CONFIG_SERIAL_CONSOLE=y -# CONFIG_ALL_PPC is not set -# CONFIG_GEMINI is not set CONFIG_EST8260=y -# CONFIG_APUS is not set +CONFIG_PPC601_SYNC_FIX=y # CONFIG_ALL_PPC is not set # CONFIG_SMP is not set -# CONFIG_ALTIVEC is not set CONFIG_MACH_SPECIFIC=y # @@ -40,7 +36,9 @@ CONFIG_MACH_SPECIFIC=y # CONFIG_HIGHMEM is not set # CONFIG_MOL is not set # CONFIG_ISA is not set +# CONFIG_EISA is not set # CONFIG_SBUS is not set +# CONFIG_MCA is not set # CONFIG_PCI is not set CONFIG_NET=y CONFIG_SYSCTL=y @@ -57,12 +55,7 @@ CONFIG_KERNEL_ELF=y # Parallel port support # # CONFIG_PARPORT is not set -# CONFIG_VGA_CONSOLE is not set -# CONFIG_FB is not set # CONFIG_PPC_RTC is not set -# CONFIG_PROC_DEVICETREE is not set -# CONFIG_BOOTX_TEXT is not set -# CONFIG_MOTOROLA_HOTSWAP is not set # # Memory Technology Devices (MTD) @@ -184,7 +177,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_MACE is not set # CONFIG_BMAC is not set # CONFIG_GMAC is not set -# CONFIG_NCR885E is not set # CONFIG_OAKNET is not set # CONFIG_NET_VENDOR_3COM is not set # CONFIG_LANCE is not set @@ -192,7 +184,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_NET_VENDOR_RACAL is not set # CONFIG_AT1700 is not set # CONFIG_DEPCA is not set -# CONFIG_HP100 is not set # CONFIG_NET_ISA is not set # CONFIG_NET_PCI is not set # CONFIG_NET_POCKET is not set @@ -250,6 +241,7 @@ CONFIG_NET_ETHERNET=y # # Console drivers # +# CONFIG_VGA_CONSOLE is not set # # Frame-buffer support @@ -379,8 +371,6 @@ CONFIG_LOCKD=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -397,6 +387,7 @@ CONFIG_PARTITION_ADVANCED=y # CONFIG_SGI_PARTITION is not set # CONFIG_ULTRIX_PARTITION is not set # CONFIG_SUN_PARTITION is not set +# CONFIG_SMB_NLS is not set # CONFIG_NLS is not set # diff --git a/arch/ppc/configs/ibmchrp_defconfig b/arch/ppc/configs/ibmchrp_defconfig new file mode 100644 index 000000000..652b94209 --- /dev/null +++ b/arch/ppc/configs/ibmchrp_defconfig @@ -0,0 +1,636 @@ +# +# Automatically generated by make menuconfig: don't edit +# +# CONFIG_UID16 is not set + +# +# Code maturity level options +# +CONFIG_EXPERIMENTAL=y + +# +# Loadable module support +# +CONFIG_MODULES=y +# CONFIG_MODVERSIONS is not set +CONFIG_KMOD=y + +# +# Platform support +# +CONFIG_PPC=y +CONFIG_6xx=y +# CONFIG_4xx is not set +# CONFIG_POWER3 is not set +# CONFIG_POWER4 is not set +# CONFIG_8260 is not set +# CONFIG_8xx is not set +CONFIG_ALL_PPC=y +# CONFIG_GEMINI is not set +# CONFIG_APUS is not set +# CONFIG_PPC601_SYNC_FIX is not set +# CONFIG_SMP is not set +# CONFIG_ALTIVEC is not set + +# +# General setup +# +CONFIG_HIGHMEM=y +# CONFIG_MOL is not set +# CONFIG_ISA is not set +# CONFIG_EISA is not set +# CONFIG_SBUS is not set +# CONFIG_MCA is not set +CONFIG_PCI=y +CONFIG_NET=y +CONFIG_SYSCTL=y +CONFIG_SYSVIPC=y +# CONFIG_BSD_PROCESS_ACCT is not set +CONFIG_KCORE_ELF=y +CONFIG_BINFMT_ELF=y +CONFIG_KERNEL_ELF=y +CONFIG_BINFMT_MISC=y +CONFIG_PCI_NAMES=y +# CONFIG_HOTPLUG is not set +# CONFIG_PCMCIA is not set + +# +# Parallel port support +# +# CONFIG_PARPORT is not set +CONFIG_PPC_RTC=y +CONFIG_PROC_DEVICETREE=y +CONFIG_PPC_RTAS=y +# CONFIG_BOOTX_TEXT is not set +# CONFIG_PREP_RESIDUAL is not set +# CONFIG_CMDLINE_BOOL is not set + +# +# Memory Technology Devices (MTD) +# +# CONFIG_MTD is not set + +# +# Plug and Play configuration +# +# CONFIG_PNP is not set +# CONFIG_ISAPNP is not set + +# +# Block devices +# +CONFIG_BLK_DEV_FD=y +# CONFIG_BLK_DEV_XD is not set +# CONFIG_PARIDE is not set +# CONFIG_BLK_CPQ_DA is not set +# CONFIG_BLK_CPQ_CISS_DA is not set +# CONFIG_BLK_DEV_DAC960 is not set +CONFIG_BLK_DEV_LOOP=y +# CONFIG_BLK_DEV_NBD is not set +CONFIG_BLK_DEV_RAM=y +CONFIG_BLK_DEV_RAM_SIZE=4096 +CONFIG_BLK_DEV_INITRD=y + +# +# Multi-device support (RAID and LVM) +# +# CONFIG_MD is not set +# CONFIG_BLK_DEV_MD is not set +# CONFIG_MD_LINEAR is not set +# CONFIG_MD_RAID0 is not set +# CONFIG_MD_RAID1 is not set +# CONFIG_MD_RAID5 is not set +# CONFIG_BLK_DEV_LVM is not set +# CONFIG_LVM_PROC_FS is not set + +# +# Networking options +# +CONFIG_PACKET=y +# CONFIG_PACKET_MMAP is not set +CONFIG_NETLINK=y +# CONFIG_RTNETLINK is not set +# CONFIG_NETLINK_DEV is not set +CONFIG_NETFILTER=y +# CONFIG_NETFILTER_DEBUG is not set +# CONFIG_FILTER is not set +CONFIG_UNIX=y +CONFIG_INET=y +CONFIG_IP_MULTICAST=y +# CONFIG_IP_ADVANCED_ROUTER is not set +# CONFIG_IP_PNP is not set +# CONFIG_NET_IPIP is not set +# CONFIG_NET_IPGRE is not set +# CONFIG_IP_MROUTE is not set +# CONFIG_INET_ECN is not set +CONFIG_SYN_COOKIES=y + +# +# IP: Netfilter Configuration +# +CONFIG_IP_NF_CONNTRACK=m +CONFIG_IP_NF_FTP=m +# CONFIG_IP_NF_QUEUE is not set +CONFIG_IP_NF_IPTABLES=m +CONFIG_IP_NF_MATCH_LIMIT=m +CONFIG_IP_NF_MATCH_MAC=m +CONFIG_IP_NF_MATCH_MARK=m +CONFIG_IP_NF_MATCH_MULTIPORT=m +CONFIG_IP_NF_MATCH_TOS=m +CONFIG_IP_NF_MATCH_STATE=m +CONFIG_IP_NF_MATCH_UNCLEAN=m +CONFIG_IP_NF_MATCH_OWNER=m +CONFIG_IP_NF_FILTER=m +CONFIG_IP_NF_TARGET_REJECT=m +CONFIG_IP_NF_TARGET_MIRROR=m +CONFIG_IP_NF_NAT=m +CONFIG_IP_NF_NAT_NEEDED=y +CONFIG_IP_NF_TARGET_MASQUERADE=m +CONFIG_IP_NF_TARGET_REDIRECT=m +# CONFIG_IP_NF_MANGLE is not set +# CONFIG_IP_NF_TARGET_LOG is not set +CONFIG_IP_NF_COMPAT_IPCHAINS=m +CONFIG_IP_NF_NAT_NEEDED=y +# CONFIG_IP_NF_COMPAT_IPFWADM is not set +# CONFIG_IPV6 is not set +# CONFIG_KHTTPD is not set +# CONFIG_ATM is not set +# CONFIG_IPX is not set +# CONFIG_ATALK is not set +# CONFIG_DECNET is not set +# CONFIG_BRIDGE is not set +# CONFIG_X25 is not set +# CONFIG_LAPB is not set +# CONFIG_LLC is not set +# CONFIG_NET_DIVERT is not set +# CONFIG_ECONET is not set +# CONFIG_WAN_ROUTER is not set +# CONFIG_NET_FASTROUTE is not set +# CONFIG_NET_HW_FLOWCONTROL is not set + +# +# QoS and/or fair queueing +# +# CONFIG_NET_SCHED is not set + +# +# ATA/IDE/MFM/RLL support +# +# CONFIG_IDE is not set +# CONFIG_BLK_DEV_IDE_MODES is not set +# CONFIG_BLK_DEV_HD is not set + +# +# SCSI support +# +CONFIG_SCSI=y +CONFIG_BLK_DEV_SD=y +CONFIG_SD_EXTRA_DEVS=40 +CONFIG_CHR_DEV_ST=y +# CONFIG_CHR_DEV_OSST is not set +CONFIG_BLK_DEV_SR=y +CONFIG_BLK_DEV_SR_VENDOR=y +CONFIG_SR_EXTRA_DEVS=2 +CONFIG_CHR_DEV_SG=y +# CONFIG_SCSI_DEBUG_QUEUES is not set +# CONFIG_SCSI_MULTI_LUN is not set +CONFIG_SCSI_CONSTANTS=y +# CONFIG_SCSI_LOGGING is not set + +# +# SCSI low-level drivers +# +# CONFIG_BLK_DEV_3W_XXXX_RAID is not set +# CONFIG_SCSI_7000FASST is not set +# CONFIG_SCSI_ACARD is not set +# CONFIG_SCSI_AHA152X is not set +# CONFIG_SCSI_AHA1542 is not set +# CONFIG_SCSI_AHA1740 is not set +# CONFIG_SCSI_AIC7XXX is not set +# CONFIG_SCSI_ADVANSYS is not set +# CONFIG_SCSI_IN2000 is not set +# CONFIG_SCSI_AM53C974 is not set +# CONFIG_SCSI_MEGARAID is not set +# CONFIG_SCSI_BUSLOGIC is not set +# CONFIG_SCSI_CPQFCTS is not set +# CONFIG_SCSI_DMX3191D is not set +# CONFIG_SCSI_DTC3280 is not set +# CONFIG_SCSI_EATA is not set +# CONFIG_SCSI_EATA_DMA is not set +# CONFIG_SCSI_EATA_PIO is not set +# CONFIG_SCSI_FUTURE_DOMAIN is not set +# CONFIG_SCSI_GDTH is not set +# CONFIG_SCSI_GENERIC_NCR5380 is not set +# CONFIG_SCSI_INITIO is not set +# CONFIG_SCSI_INIA100 is not set +# CONFIG_SCSI_NCR53C406A is not set +# CONFIG_SCSI_NCR53C7xx is not set +# CONFIG_SCSI_NCR53C8XX is not set +CONFIG_SCSI_SYM53C8XX=y +CONFIG_SCSI_NCR53C8XX_DEFAULT_TAGS=8 +CONFIG_SCSI_NCR53C8XX_MAX_TAGS=32 +CONFIG_SCSI_NCR53C8XX_SYNC=20 +# CONFIG_SCSI_NCR53C8XX_PROFILE is not set +# CONFIG_SCSI_NCR53C8XX_IOMAPPED is not set +# CONFIG_SCSI_NCR53C8XX_PQS_PDS is not set +# CONFIG_SCSI_NCR53C8XX_SYMBIOS_COMPAT is not set +# CONFIG_SCSI_PAS16 is not set +# CONFIG_SCSI_PCI2000 is not set +# CONFIG_SCSI_PCI2220I is not set +# CONFIG_SCSI_PSI240I is not set +# CONFIG_SCSI_QLOGIC_FAS is not set +# CONFIG_SCSI_QLOGIC_ISP is not set +# CONFIG_SCSI_QLOGIC_FC is not set +# CONFIG_SCSI_QLOGIC_1280 is not set +# CONFIG_SCSI_SIM710 is not set +# CONFIG_SCSI_SYM53C416 is not set +# CONFIG_SCSI_DC390T is not set +# CONFIG_SCSI_T128 is not set +# CONFIG_SCSI_U14_34F is not set +# CONFIG_SCSI_DEBUG is not set +# CONFIG_SCSI_MESH is not set +# CONFIG_SCSI_MAC53C94 is not set + +# +# IEEE 1394 (FireWire) support +# +# CONFIG_IEEE1394 is not set + +# +# Network device support +# +CONFIG_NETDEVICES=y + +# +# ARCnet devices +# +# CONFIG_ARCNET is not set +# CONFIG_DUMMY is not set +# CONFIG_BONDING is not set +# CONFIG_EQUALIZER is not set +# CONFIG_TUN is not set +# CONFIG_ETHERTAP is not set +# CONFIG_NET_SB1000 is not set + +# +# Ethernet (10 or 100Mbit) +# +CONFIG_NET_ETHERNET=y +# CONFIG_MACE is not set +# CONFIG_BMAC is not set +# CONFIG_GMAC is not set +# CONFIG_NCR885E is not set +# CONFIG_OAKNET is not set +# CONFIG_NET_VENDOR_3COM is not set +# CONFIG_LANCE is not set +# CONFIG_NET_VENDOR_SMC is not set +# CONFIG_NET_VENDOR_RACAL is not set +# CONFIG_AT1700 is not set +# CONFIG_DEPCA is not set +# CONFIG_HP100 is not set +# CONFIG_NET_ISA is not set +CONFIG_NET_PCI=y +CONFIG_PCNET32=y +# CONFIG_ADAPTEC_STARFIRE is not set +# CONFIG_APRICOT is not set +# CONFIG_CS89x0 is not set +# CONFIG_TULIP is not set +CONFIG_DE4X5=y +# CONFIG_DGRS is not set +# CONFIG_DM9102 is not set +# CONFIG_EEPRO100 is not set +# CONFIG_EEPRO100_PM is not set +# CONFIG_LNE390 is not set +# CONFIG_NATSEMI is not set +# CONFIG_NE2K_PCI is not set +# CONFIG_NE3210 is not set +# CONFIG_ES3210 is not set +# CONFIG_8139TOO is not set +# CONFIG_RTL8129 is not set +# CONFIG_SIS900 is not set +# CONFIG_EPIC100 is not set +# CONFIG_SUNDANCE is not set +# CONFIG_TLAN is not set +# CONFIG_VIA_RHINE is not set +# CONFIG_WINBOND_840 is not set +# CONFIG_HAPPYMEAL is not set +# CONFIG_NET_POCKET is not set + +# +# Ethernet (1000 Mbit) +# +# CONFIG_ACENIC is not set +# CONFIG_HAMACHI is not set +# CONFIG_YELLOWFIN is not set +# CONFIG_SK98LIN is not set +# CONFIG_FDDI is not set +# CONFIG_HIPPI is not set +# CONFIG_PPP is not set +# CONFIG_SLIP is not set + +# +# Wireless LAN (non-hamradio) +# +# CONFIG_NET_RADIO is not set + +# +# Token Ring devices +# +CONFIG_TR=y +# CONFIG_IBMTR is not set +CONFIG_IBMOL=y +# CONFIG_IBMLS is not set +# CONFIG_TMS380TR is not set +# CONFIG_SMCTR is not set +# CONFIG_NET_FC is not set +# CONFIG_RCPCI is not set +# CONFIG_SHAPER is not set + +# +# Wan interfaces +# +# CONFIG_WAN is not set + +# +# Amateur Radio support +# +# CONFIG_HAMRADIO is not set + +# +# IrDA (infrared) support +# +# CONFIG_IRDA is not set + +# +# ISDN subsystem +# +# CONFIG_ISDN is not set + +# +# Old CD-ROM drivers (not SCSI, not IDE) +# +# CONFIG_CD_NO_IDESCSI is not set + +# +# Console drivers +# +CONFIG_VGA_CONSOLE=y + +# +# Frame-buffer support +# +CONFIG_FB=y +CONFIG_DUMMY_CONSOLE=y +# CONFIG_FB_RIVA is not set +# CONFIG_FB_CLGEN is not set +# CONFIG_FB_PM2 is not set +# CONFIG_FB_CYBER2000 is not set +CONFIG_FB_OF=y +# CONFIG_FB_CONTROL is not set +# CONFIG_FB_PLATINUM is not set +# CONFIG_FB_VALKYRIE is not set +# CONFIG_FB_CT65550 is not set +# CONFIG_FB_IMSTT is not set +# CONFIG_FB_S3TRIO is not set +# CONFIG_FB_VGA16 is not set +CONFIG_FB_MATROX=y +CONFIG_FB_MATROX_MILLENIUM=y +CONFIG_FB_MATROX_MYSTIQUE=y +CONFIG_FB_MATROX_G100=y +# CONFIG_FB_MATROX_G450 is not set +# CONFIG_FB_MATROX_MULTIHEAD is not set +# CONFIG_FB_ATY is not set +# CONFIG_FB_ATY128 is not set +CONFIG_FB_3DFX=y +# CONFIG_FB_SIS is not set +# CONFIG_FB_VIRTUAL is not set +# CONFIG_FBCON_ADVANCED is not set +CONFIG_FBCON_CFB8=y +CONFIG_FBCON_CFB16=y +CONFIG_FBCON_CFB24=y +CONFIG_FBCON_CFB32=y +# CONFIG_FBCON_FONTWIDTH8_ONLY is not set +CONFIG_FBCON_FONTS=y +# CONFIG_FONT_8x8 is not set +CONFIG_FONT_8x16=y +CONFIG_FONT_SUN8x16=y +CONFIG_FONT_SUN12x22=y +# CONFIG_FONT_6x11 is not set +# CONFIG_FONT_PEARL_8x8 is not set +# CONFIG_FONT_ACORN_8x8 is not set +# CONFIG_FB_COMPAT_XPMAC is not set + +# +# Input core support +# +CONFIG_INPUT=y +CONFIG_INPUT_KEYBDEV=y +CONFIG_INPUT_MOUSEDEV=y +CONFIG_INPUT_MOUSEDEV_SCREEN_X=1024 +CONFIG_INPUT_MOUSEDEV_SCREEN_Y=768 +# CONFIG_INPUT_JOYDEV is not set +CONFIG_INPUT_EVDEV=y + +# +# Macintosh device drivers +# +# CONFIG_ADB_CUDA is not set +# CONFIG_ADB_PMU is not set +# CONFIG_MAC_FLOPPY is not set +# CONFIG_MAC_SERIAL is not set +# CONFIG_ADB is not set + +# +# Character devices +# +CONFIG_VT=y +CONFIG_VT_CONSOLE=y +CONFIG_SERIAL=y +CONFIG_SERIAL_CONSOLE=y +# CONFIG_SERIAL_EXTENDED is not set +# CONFIG_SERIAL_NONSTANDARD is not set +CONFIG_UNIX98_PTYS=y +CONFIG_UNIX98_PTY_COUNT=256 + +# +# I2C support +# +# CONFIG_I2C is not set + +# +# Mice +# +CONFIG_BUSMOUSE=y +# CONFIG_ATIXL_BUSMOUSE is not set +# CONFIG_LOGIBUSMOUSE is not set +# CONFIG_MS_BUSMOUSE is not set +CONFIG_MOUSE=y +CONFIG_PSMOUSE=y +# CONFIG_82C710_MOUSE is not set +# CONFIG_PC110_PAD is not set + +# +# Joysticks +# +# CONFIG_JOYSTICK is not set +# CONFIG_QIC02_TAPE is not set + +# +# Watchdog Cards +# +# CONFIG_WATCHDOG is not set +# CONFIG_INTEL_RNG is not set +CONFIG_NVRAM=y +CONFIG_RTC=y +# CONFIG_DTLK is not set +# CONFIG_R3964 is not set +# CONFIG_APPLICOM is not set + +# +# Ftape, the floppy tape device driver +# +# CONFIG_FTAPE is not set +# CONFIG_AGP is not set +# CONFIG_DRM is not set + +# +# Multimedia devices +# +# CONFIG_VIDEO_DEV is not set + +# +# File systems +# +# CONFIG_QUOTA is not set +# CONFIG_AUTOFS_FS is not set +# CONFIG_AUTOFS4_FS is not set +# CONFIG_ADFS_FS is not set +# CONFIG_ADFS_FS_RW is not set +# CONFIG_AFFS_FS is not set +# CONFIG_HFS_FS is not set +# CONFIG_BFS_FS is not set +CONFIG_FAT_FS=m +CONFIG_MSDOS_FS=m +# CONFIG_UMSDOS_FS is not set +CONFIG_VFAT_FS=m +# CONFIG_EFS_FS is not set +# CONFIG_JFFS_FS is not set +# CONFIG_CRAMFS is not set +# CONFIG_RAMFS is not set +CONFIG_ISO9660_FS=y +# CONFIG_JOLIET is not set +# CONFIG_MINIX_FS is not set +# CONFIG_NTFS_FS is not set +# CONFIG_NTFS_RW is not set +# CONFIG_HPFS_FS is not set +CONFIG_PROC_FS=y +CONFIG_DEVFS_FS=y +# CONFIG_DEVFS_MOUNT is not set +# CONFIG_DEVFS_DEBUG is not set +CONFIG_DEVPTS_FS=y +# CONFIG_QNX4FS_FS is not set +# CONFIG_QNX4FS_RW is not set +# CONFIG_ROMFS_FS is not set +CONFIG_EXT2_FS=y +# CONFIG_SYSV_FS is not set +# CONFIG_SYSV_FS_WRITE is not set +# CONFIG_UDF_FS is not set +# CONFIG_UDF_RW is not set +# CONFIG_UFS_FS is not set +# CONFIG_UFS_FS_WRITE is not set + +# +# Network File Systems +# +# CONFIG_CODA_FS is not set +# CONFIG_NFS_FS is not set +# CONFIG_NFS_V3 is not set +# CONFIG_ROOT_NFS is not set +# CONFIG_NFSD is not set +# CONFIG_NFSD_V3 is not set +# CONFIG_SUNRPC is not set +# CONFIG_LOCKD is not set +# CONFIG_SMB_FS is not set +# CONFIG_NCP_FS is not set +# CONFIG_NCPFS_PACKET_SIGNING is not set +# CONFIG_NCPFS_IOCTL_LOCKING is not set +# CONFIG_NCPFS_STRONG is not set +# CONFIG_NCPFS_NFS_NS is not set +# CONFIG_NCPFS_OS2_NS is not set +# CONFIG_NCPFS_SMALLDOS is not set +# CONFIG_NCPFS_NLS is not set +# CONFIG_NCPFS_EXTRAS is not set + +# +# Partition Types +# +CONFIG_PARTITION_ADVANCED=y +# CONFIG_ACORN_PARTITION is not set +# CONFIG_OSF_PARTITION is not set +# CONFIG_AMIGA_PARTITION is not set +# CONFIG_ATARI_PARTITION is not set +CONFIG_MAC_PARTITION=y +CONFIG_MSDOS_PARTITION=y +# CONFIG_BSD_DISKLABEL is not set +# CONFIG_SOLARIS_X86_PARTITION is not set +# CONFIG_UNIXWARE_DISKLABEL is not set +# CONFIG_SGI_PARTITION is not set +# CONFIG_ULTRIX_PARTITION is not set +# CONFIG_SUN_PARTITION is not set +# CONFIG_SMB_NLS is not set +CONFIG_NLS=y + +# +# Native Language Support +# +CONFIG_NLS_DEFAULT="iso8859-1" +# CONFIG_NLS_CODEPAGE_437 is not set +# CONFIG_NLS_CODEPAGE_737 is not set +# CONFIG_NLS_CODEPAGE_775 is not set +# CONFIG_NLS_CODEPAGE_850 is not set +# CONFIG_NLS_CODEPAGE_852 is not set +# CONFIG_NLS_CODEPAGE_855 is not set +# CONFIG_NLS_CODEPAGE_857 is not set +# CONFIG_NLS_CODEPAGE_860 is not set +# CONFIG_NLS_CODEPAGE_861 is not set +# CONFIG_NLS_CODEPAGE_862 is not set +# CONFIG_NLS_CODEPAGE_863 is not set +# CONFIG_NLS_CODEPAGE_864 is not set +# CONFIG_NLS_CODEPAGE_865 is not set +# CONFIG_NLS_CODEPAGE_866 is not set +# CONFIG_NLS_CODEPAGE_869 is not set +# CONFIG_NLS_CODEPAGE_874 is not set +# CONFIG_NLS_CODEPAGE_932 is not set +# CONFIG_NLS_CODEPAGE_936 is not set +# CONFIG_NLS_CODEPAGE_949 is not set +# CONFIG_NLS_CODEPAGE_950 is not set +# CONFIG_NLS_ISO8859_1 is not set +# CONFIG_NLS_ISO8859_2 is not set +# CONFIG_NLS_ISO8859_3 is not set +# CONFIG_NLS_ISO8859_4 is not set +# CONFIG_NLS_ISO8859_5 is not set +# CONFIG_NLS_ISO8859_6 is not set +# CONFIG_NLS_ISO8859_7 is not set +# CONFIG_NLS_ISO8859_8 is not set +# CONFIG_NLS_ISO8859_9 is not set +# CONFIG_NLS_ISO8859_14 is not set +# CONFIG_NLS_ISO8859_15 is not set +# CONFIG_NLS_KOI8_R is not set +# CONFIG_NLS_UTF8 is not set + +# +# Sound +# +# CONFIG_SOUND is not set + +# +# USB support +# +# CONFIG_USB is not set + +# +# Kernel hacking +# +CONFIG_MAGIC_SYSRQ=y +# CONFIG_KGDB is not set +CONFIG_XMON=y diff --git a/arch/ppc/configs/mbx_defconfig b/arch/ppc/configs/mbx_defconfig index 6230f5877..81829327a 100644 --- a/arch/ppc/configs/mbx_defconfig +++ b/arch/ppc/configs/mbx_defconfig @@ -21,7 +21,6 @@ CONFIG_PPC=y # CONFIG_4xx is not set # CONFIG_POWER3 is not set # CONFIG_POWER4 is not set -# CONFIG_8260 is not set CONFIG_8xx=y CONFIG_SERIAL_CONSOLE=y # CONFIG_RPXLITE is not set @@ -32,6 +31,7 @@ CONFIG_SERIAL_CONSOLE=y # CONFIG_TQM860 is not set CONFIG_MBX=y # CONFIG_WINCEPT is not set +CONFIG_PPC601_SYNC_FIX=y # CONFIG_ALL_PPC is not set # CONFIG_SMP is not set CONFIG_MACH_SPECIFIC=y @@ -43,7 +43,9 @@ CONFIG_MATH_EMULATION=y # CONFIG_HIGHMEM is not set # CONFIG_MOL is not set # CONFIG_ISA is not set +# CONFIG_EISA is not set # CONFIG_SBUS is not set +# CONFIG_MCA is not set # CONFIG_PCI_QSPAN is not set # CONFIG_PCI is not set CONFIG_NET=y @@ -176,7 +178,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_MACE is not set # CONFIG_BMAC is not set # CONFIG_GMAC is not set -# CONFIG_NCR885E is not set # CONFIG_OAKNET is not set # CONFIG_NET_VENDOR_3COM is not set # CONFIG_LANCE is not set @@ -184,7 +185,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_NET_VENDOR_RACAL is not set # CONFIG_AT1700 is not set # CONFIG_DEPCA is not set -# CONFIG_HP100 is not set # CONFIG_NET_ISA is not set # CONFIG_NET_PCI is not set # CONFIG_NET_POCKET is not set @@ -370,8 +370,6 @@ CONFIG_LOCKD=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -388,6 +386,7 @@ CONFIG_PARTITION_ADVANCED=y # CONFIG_SGI_PARTITION is not set # CONFIG_ULTRIX_PARTITION is not set # CONFIG_SUN_PARTITION is not set +# CONFIG_SMB_NLS is not set # CONFIG_NLS is not set # diff --git a/arch/ppc/configs/oak_defconfig b/arch/ppc/configs/oak_defconfig index f33966d35..d8f39c36e 100644 --- a/arch/ppc/configs/oak_defconfig +++ b/arch/ppc/configs/oak_defconfig @@ -12,7 +12,7 @@ CONFIG_EXPERIMENTAL=y # Loadable module support # CONFIG_MODULES=y -# CONFIG_MODVERSIONS is not set +CONFIG_MODVERSIONS=y CONFIG_KMOD=y # @@ -23,10 +23,10 @@ CONFIG_PPC=y CONFIG_4xx=y # CONFIG_POWER3 is not set # CONFIG_POWER4 is not set -# CONFIG_8260 is not set # CONFIG_8xx is not set CONFIG_OAK=y # CONFIG_WALNUT is not set +CONFIG_PPC601_SYNC_FIX=y # CONFIG_SMP is not set CONFIG_MACH_SPECIFIC=y # CONFIG_MATH_EMULATION is not set @@ -37,7 +37,9 @@ CONFIG_MACH_SPECIFIC=y # CONFIG_HIGHMEM is not set # CONFIG_MOL is not set # CONFIG_ISA is not set +# CONFIG_EISA is not set # CONFIG_SBUS is not set +# CONFIG_MCA is not set # CONFIG_PCI is not set CONFIG_NET=y CONFIG_SYSCTL=y @@ -171,7 +173,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_MACE is not set # CONFIG_BMAC is not set # CONFIG_GMAC is not set -# CONFIG_NCR885E is not set CONFIG_OAKNET=y # CONFIG_NET_VENDOR_3COM is not set # CONFIG_LANCE is not set @@ -179,7 +180,6 @@ CONFIG_OAKNET=y # CONFIG_NET_VENDOR_RACAL is not set # CONFIG_AT1700 is not set # CONFIG_DEPCA is not set -# CONFIG_HP100 is not set # CONFIG_NET_ISA is not set # CONFIG_NET_PCI is not set # CONFIG_NET_POCKET is not set @@ -366,8 +366,6 @@ CONFIG_LOCKD=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -376,6 +374,7 @@ CONFIG_LOCKD=y # # CONFIG_PARTITION_ADVANCED is not set CONFIG_MSDOS_PARTITION=y +# CONFIG_SMB_NLS is not set # CONFIG_NLS is not set # diff --git a/arch/ppc/configs/power3_defconfig b/arch/ppc/configs/power3_defconfig new file mode 100644 index 000000000..0deb9b5e8 --- /dev/null +++ b/arch/ppc/configs/power3_defconfig @@ -0,0 +1,662 @@ +# +# Automatically generated by make menuconfig: don't edit +# +# CONFIG_UID16 is not set + +# +# Code maturity level options +# +CONFIG_EXPERIMENTAL=y + +# +# Loadable module support +# +CONFIG_MODULES=y +# CONFIG_MODVERSIONS is not set +CONFIG_KMOD=y + +# +# Platform support +# +CONFIG_PPC=y +# CONFIG_6xx is not set +# CONFIG_4xx is not set +CONFIG_POWER3=y +# CONFIG_POWER4 is not set +# CONFIG_8xx is not set +CONFIG_PPC64BRIDGE=y +CONFIG_ALL_PPC=y +CONFIG_SMP=y + +# +# General setup +# +CONFIG_HIGHMEM=y +# CONFIG_MOL is not set +# CONFIG_ISA is not set +# CONFIG_EISA is not set +# CONFIG_SBUS is not set +# CONFIG_MCA is not set +CONFIG_PCI=y +CONFIG_NET=y +CONFIG_SYSCTL=y +CONFIG_SYSVIPC=y +# CONFIG_BSD_PROCESS_ACCT is not set +CONFIG_KCORE_ELF=y +CONFIG_BINFMT_ELF=y +CONFIG_KERNEL_ELF=y +CONFIG_BINFMT_MISC=y +CONFIG_PCI_NAMES=y +# CONFIG_HOTPLUG is not set +# CONFIG_PCMCIA is not set + +# +# Parallel port support +# +CONFIG_PARPORT=m +CONFIG_PARPORT_PC=m +CONFIG_PARPORT_PC_FIFO=y +# CONFIG_PARPORT_PC_SUPERIO is not set +# CONFIG_PARPORT_AMIGA is not set +# CONFIG_PARPORT_MFC3 is not set +# CONFIG_PARPORT_ATARI is not set +# CONFIG_PARPORT_SUNBPP is not set +# CONFIG_PARPORT_OTHER is not set +# CONFIG_PARPORT_1284 is not set +# CONFIG_PPC_RTC is not set +CONFIG_PROC_DEVICETREE=y +CONFIG_PPC_RTAS=y +# CONFIG_BOOTX_TEXT is not set +# CONFIG_PREP_RESIDUAL is not set +# CONFIG_CMDLINE_BOOL is not set + +# +# Memory Technology Devices (MTD) +# +# CONFIG_MTD is not set + +# +# Plug and Play configuration +# +# CONFIG_PNP is not set +# CONFIG_ISAPNP is not set + +# +# Block devices +# +CONFIG_BLK_DEV_FD=y +# CONFIG_BLK_DEV_XD is not set +# CONFIG_PARIDE is not set +# CONFIG_BLK_CPQ_DA is not set +# CONFIG_BLK_CPQ_CISS_DA is not set +# CONFIG_BLK_DEV_DAC960 is not set +CONFIG_BLK_DEV_LOOP=y +# CONFIG_BLK_DEV_NBD is not set +CONFIG_BLK_DEV_RAM=y +CONFIG_BLK_DEV_RAM_SIZE=4096 +CONFIG_BLK_DEV_INITRD=y + +# +# Multi-device support (RAID and LVM) +# +CONFIG_MD=y +CONFIG_BLK_DEV_MD=y +CONFIG_MD_LINEAR=y +CONFIG_MD_RAID0=y +CONFIG_MD_RAID1=y +CONFIG_MD_RAID5=y +# CONFIG_MD_BOOT is not set +# CONFIG_AUTODETECT_RAID is not set +CONFIG_BLK_DEV_LVM=y +CONFIG_LVM_PROC_FS=y + +# +# Networking options +# +CONFIG_PACKET=y +# CONFIG_PACKET_MMAP is not set +CONFIG_NETLINK=y +# CONFIG_RTNETLINK is not set +# CONFIG_NETLINK_DEV is not set +# CONFIG_NETFILTER is not set +# CONFIG_FILTER is not set +CONFIG_UNIX=y +CONFIG_INET=y +CONFIG_IP_MULTICAST=y +# CONFIG_IP_ADVANCED_ROUTER is not set +# CONFIG_IP_PNP is not set +# CONFIG_NET_IPIP is not set +# CONFIG_NET_IPGRE is not set +# CONFIG_IP_MROUTE is not set +# CONFIG_INET_ECN is not set +CONFIG_SYN_COOKIES=y +# CONFIG_IPV6 is not set +# CONFIG_KHTTPD is not set +# CONFIG_ATM is not set +# CONFIG_IPX is not set +# CONFIG_ATALK is not set +# CONFIG_DECNET is not set +# CONFIG_BRIDGE is not set +# CONFIG_X25 is not set +# CONFIG_LAPB is not set +# CONFIG_LLC is not set +# CONFIG_NET_DIVERT is not set +# CONFIG_ECONET is not set +# CONFIG_WAN_ROUTER is not set +# CONFIG_NET_FASTROUTE is not set +# CONFIG_NET_HW_FLOWCONTROL is not set + +# +# QoS and/or fair queueing +# +# CONFIG_NET_SCHED is not set + +# +# ATA/IDE/MFM/RLL support +# +# CONFIG_IDE is not set +# CONFIG_BLK_DEV_IDE_MODES is not set +# CONFIG_BLK_DEV_HD is not set + +# +# SCSI support +# +CONFIG_SCSI=y +CONFIG_BLK_DEV_SD=y +CONFIG_SD_EXTRA_DEVS=40 +CONFIG_CHR_DEV_ST=y +# CONFIG_CHR_DEV_OSST is not set +CONFIG_BLK_DEV_SR=y +CONFIG_BLK_DEV_SR_VENDOR=y +CONFIG_SR_EXTRA_DEVS=2 +CONFIG_CHR_DEV_SG=y +# CONFIG_SCSI_DEBUG_QUEUES is not set +# CONFIG_SCSI_MULTI_LUN is not set +CONFIG_SCSI_CONSTANTS=y +CONFIG_SCSI_LOGGING=y + +# +# SCSI low-level drivers +# +# CONFIG_BLK_DEV_3W_XXXX_RAID is not set +# CONFIG_SCSI_7000FASST is not set +# CONFIG_SCSI_ACARD is not set +# CONFIG_SCSI_AHA152X is not set +# CONFIG_SCSI_AHA1542 is not set +# CONFIG_SCSI_AHA1740 is not set +# CONFIG_SCSI_AIC7XXX is not set +# CONFIG_SCSI_ADVANSYS is not set +# CONFIG_SCSI_IN2000 is not set +# CONFIG_SCSI_AM53C974 is not set +# CONFIG_SCSI_MEGARAID is not set +# CONFIG_SCSI_BUSLOGIC is not set +# CONFIG_SCSI_CPQFCTS is not set +# CONFIG_SCSI_DMX3191D is not set +# CONFIG_SCSI_DTC3280 is not set +# CONFIG_SCSI_EATA is not set +# CONFIG_SCSI_EATA_DMA is not set +# CONFIG_SCSI_EATA_PIO is not set +# CONFIG_SCSI_FUTURE_DOMAIN is not set +# CONFIG_SCSI_GDTH is not set +# CONFIG_SCSI_GENERIC_NCR5380 is not set +# CONFIG_SCSI_INITIO is not set +# CONFIG_SCSI_INIA100 is not set +# CONFIG_SCSI_PPA is not set +# CONFIG_SCSI_IMM is not set +# CONFIG_SCSI_NCR53C406A is not set +# CONFIG_SCSI_NCR53C7xx is not set +# CONFIG_SCSI_NCR53C8XX is not set +CONFIG_SCSI_SYM53C8XX=y +CONFIG_SCSI_NCR53C8XX_DEFAULT_TAGS=8 +CONFIG_SCSI_NCR53C8XX_MAX_TAGS=32 +CONFIG_SCSI_NCR53C8XX_SYNC=20 +# CONFIG_SCSI_NCR53C8XX_PROFILE is not set +# CONFIG_SCSI_NCR53C8XX_IOMAPPED is not set +# CONFIG_SCSI_NCR53C8XX_PQS_PDS is not set +# CONFIG_SCSI_NCR53C8XX_SYMBIOS_COMPAT is not set +# CONFIG_SCSI_PAS16 is not set +# CONFIG_SCSI_PCI2000 is not set +# CONFIG_SCSI_PCI2220I is not set +# CONFIG_SCSI_PSI240I is not set +# CONFIG_SCSI_QLOGIC_FAS is not set +# CONFIG_SCSI_QLOGIC_ISP is not set +# CONFIG_SCSI_QLOGIC_FC is not set +# CONFIG_SCSI_QLOGIC_1280 is not set +# CONFIG_SCSI_SIM710 is not set +# CONFIG_SCSI_SYM53C416 is not set +# CONFIG_SCSI_DC390T is not set +# CONFIG_SCSI_T128 is not set +# CONFIG_SCSI_U14_34F is not set +# CONFIG_SCSI_DEBUG is not set +# CONFIG_SCSI_MESH is not set +# CONFIG_SCSI_MAC53C94 is not set + +# +# IEEE 1394 (FireWire) support +# +# CONFIG_IEEE1394 is not set + +# +# Network device support +# +CONFIG_NETDEVICES=y + +# +# ARCnet devices +# +# CONFIG_ARCNET is not set +# CONFIG_DUMMY is not set +# CONFIG_BONDING is not set +# CONFIG_EQUALIZER is not set +# CONFIG_TUN is not set +# CONFIG_ETHERTAP is not set +# CONFIG_NET_SB1000 is not set + +# +# Ethernet (10 or 100Mbit) +# +CONFIG_NET_ETHERNET=y +# CONFIG_MACE is not set +# CONFIG_BMAC is not set +# CONFIG_GMAC is not set +# CONFIG_OAKNET is not set +# CONFIG_NET_VENDOR_3COM is not set +# CONFIG_LANCE is not set +# CONFIG_NET_VENDOR_SMC is not set +# CONFIG_NET_VENDOR_RACAL is not set +# CONFIG_AT1700 is not set +# CONFIG_DEPCA is not set +# CONFIG_HP100 is not set +# CONFIG_NET_ISA is not set +CONFIG_NET_PCI=y +CONFIG_PCNET32=y +# CONFIG_ADAPTEC_STARFIRE is not set +# CONFIG_APRICOT is not set +# CONFIG_CS89x0 is not set +# CONFIG_TULIP is not set +# CONFIG_DE4X5 is not set +# CONFIG_DGRS is not set +# CONFIG_DM9102 is not set +# CONFIG_EEPRO100 is not set +# CONFIG_EEPRO100_PM is not set +# CONFIG_LNE390 is not set +# CONFIG_NATSEMI is not set +# CONFIG_NE2K_PCI is not set +# CONFIG_NE3210 is not set +# CONFIG_ES3210 is not set +# CONFIG_8139TOO is not set +# CONFIG_RTL8129 is not set +# CONFIG_SIS900 is not set +# CONFIG_EPIC100 is not set +# CONFIG_SUNDANCE is not set +# CONFIG_TLAN is not set +# CONFIG_VIA_RHINE is not set +# CONFIG_WINBOND_840 is not set +# CONFIG_HAPPYMEAL is not set +# CONFIG_NET_POCKET is not set + +# +# Ethernet (1000 Mbit) +# +# CONFIG_ACENIC is not set +# CONFIG_HAMACHI is not set +# CONFIG_YELLOWFIN is not set +# CONFIG_SK98LIN is not set +# CONFIG_FDDI is not set +# CONFIG_HIPPI is not set +# CONFIG_PLIP is not set +# CONFIG_PPP is not set +# CONFIG_SLIP is not set + +# +# Wireless LAN (non-hamradio) +# +# CONFIG_NET_RADIO is not set + +# +# Token Ring devices +# +CONFIG_TR=y +# CONFIG_IBMTR is not set +CONFIG_IBMOL=y +# CONFIG_IBMLS is not set +# CONFIG_TMS380TR is not set +# CONFIG_SMCTR is not set +# CONFIG_NET_FC is not set +# CONFIG_RCPCI is not set +# CONFIG_SHAPER is not set + +# +# Wan interfaces +# +# CONFIG_WAN is not set + +# +# Amateur Radio support +# +# CONFIG_HAMRADIO is not set + +# +# IrDA (infrared) support +# +# CONFIG_IRDA is not set + +# +# ISDN subsystem +# +# CONFIG_ISDN is not set + +# +# Old CD-ROM drivers (not SCSI, not IDE) +# +# CONFIG_CD_NO_IDESCSI is not set + +# +# Console drivers +# +# CONFIG_VGA_CONSOLE is not set + +# +# Frame-buffer support +# +CONFIG_FB=y +CONFIG_DUMMY_CONSOLE=y +# CONFIG_FB_RIVA is not set +# CONFIG_FB_CLGEN is not set +# CONFIG_FB_PM2 is not set +# CONFIG_FB_CYBER2000 is not set +CONFIG_FB_OF=y +# CONFIG_FB_CONTROL is not set +# CONFIG_FB_PLATINUM is not set +# CONFIG_FB_VALKYRIE is not set +# CONFIG_FB_CT65550 is not set +# CONFIG_FB_IMSTT is not set +# CONFIG_FB_S3TRIO is not set +# CONFIG_FB_VGA16 is not set +CONFIG_FB_MATROX=y +CONFIG_FB_MATROX_MILLENIUM=y +CONFIG_FB_MATROX_MYSTIQUE=y +CONFIG_FB_MATROX_G100=y +CONFIG_FB_MATROX_I2C=y +# CONFIG_FB_MATROX_MAVEN is not set +# CONFIG_FB_MATROX_G450 is not set +CONFIG_FB_MATROX_MULTIHEAD=y +# CONFIG_FB_ATY is not set +# CONFIG_FB_ATY128 is not set +# CONFIG_FB_3DFX is not set +# CONFIG_FB_SIS is not set +# CONFIG_FB_VIRTUAL is not set +# CONFIG_FBCON_ADVANCED is not set +CONFIG_FBCON_CFB8=y +CONFIG_FBCON_CFB16=y +CONFIG_FBCON_CFB24=y +CONFIG_FBCON_CFB32=y +# CONFIG_FBCON_FONTWIDTH8_ONLY is not set +CONFIG_FBCON_FONTS=y +# CONFIG_FONT_8x8 is not set +CONFIG_FONT_8x16=y +CONFIG_FONT_SUN8x16=y +CONFIG_FONT_SUN12x22=y +# CONFIG_FONT_6x11 is not set +# CONFIG_FONT_PEARL_8x8 is not set +# CONFIG_FONT_ACORN_8x8 is not set +# CONFIG_FB_COMPAT_XPMAC is not set + +# +# Input core support +# +CONFIG_INPUT=y +CONFIG_INPUT_KEYBDEV=y +CONFIG_INPUT_MOUSEDEV=y +CONFIG_INPUT_MOUSEDEV_SCREEN_X=1024 +CONFIG_INPUT_MOUSEDEV_SCREEN_Y=768 +# CONFIG_INPUT_JOYDEV is not set +CONFIG_INPUT_EVDEV=y + +# +# Macintosh device drivers +# +# CONFIG_ADB_CUDA is not set +# CONFIG_ADB_PMU is not set +# CONFIG_MAC_FLOPPY is not set +# CONFIG_MAC_SERIAL is not set +# CONFIG_ADB is not set + +# +# Character devices +# +CONFIG_VT=y +CONFIG_VT_CONSOLE=y +CONFIG_SERIAL=y +CONFIG_SERIAL_CONSOLE=y +# CONFIG_SERIAL_EXTENDED is not set +# CONFIG_SERIAL_NONSTANDARD is not set +CONFIG_UNIX98_PTYS=y +CONFIG_UNIX98_PTY_COUNT=2048 +CONFIG_PRINTER=m +# CONFIG_LP_CONSOLE is not set +# CONFIG_PPDEV is not set + +# +# I2C support +# +CONFIG_I2C=y +CONFIG_I2C_ALGOBIT=y +# CONFIG_I2C_PHILIPSPAR is not set +# CONFIG_I2C_ELV is not set +# CONFIG_I2C_VELLEMAN is not set +CONFIG_I2C_ALGOPCF=y +# CONFIG_I2C_ELEKTOR is not set +CONFIG_I2C_CHARDEV=y + +# +# Mice +# +# CONFIG_BUSMOUSE is not set +CONFIG_MOUSE=y +CONFIG_PSMOUSE=y +# CONFIG_82C710_MOUSE is not set +# CONFIG_PC110_PAD is not set + +# +# Joysticks +# +# CONFIG_JOYSTICK is not set +# CONFIG_QIC02_TAPE is not set + +# +# Watchdog Cards +# +# CONFIG_WATCHDOG is not set +# CONFIG_INTEL_RNG is not set +CONFIG_NVRAM=y +# CONFIG_RTC is not set +# CONFIG_DTLK is not set +# CONFIG_R3964 is not set +# CONFIG_APPLICOM is not set + +# +# Ftape, the floppy tape device driver +# +# CONFIG_FTAPE is not set +# CONFIG_AGP is not set +# CONFIG_DRM is not set + +# +# Multimedia devices +# +# CONFIG_VIDEO_DEV is not set + +# +# File systems +# +# CONFIG_QUOTA is not set +# CONFIG_AUTOFS_FS is not set +# CONFIG_AUTOFS4_FS is not set +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set +# CONFIG_ADFS_FS is not set +# CONFIG_ADFS_FS_RW is not set +# CONFIG_AFFS_FS is not set +# CONFIG_HFS_FS is not set +# CONFIG_BFS_FS is not set +CONFIG_FAT_FS=y +CONFIG_MSDOS_FS=y +# CONFIG_UMSDOS_FS is not set +CONFIG_VFAT_FS=y +# CONFIG_EFS_FS is not set +# CONFIG_JFFS_FS is not set +# CONFIG_CRAMFS is not set +# CONFIG_RAMFS is not set +CONFIG_ISO9660_FS=y +CONFIG_JOLIET=y +# CONFIG_MINIX_FS is not set +# CONFIG_NTFS_FS is not set +# CONFIG_NTFS_RW is not set +# CONFIG_HPFS_FS is not set +CONFIG_PROC_FS=y +# CONFIG_DEVFS_FS is not set +# CONFIG_DEVFS_MOUNT is not set +# CONFIG_DEVFS_DEBUG is not set +CONFIG_DEVPTS_FS=y +# CONFIG_QNX4FS_FS is not set +# CONFIG_QNX4FS_RW is not set +# CONFIG_ROMFS_FS is not set +CONFIG_EXT2_FS=y +# CONFIG_SYSV_FS is not set +# CONFIG_SYSV_FS_WRITE is not set +# CONFIG_UDF_FS is not set +# CONFIG_UDF_RW is not set +# CONFIG_UFS_FS is not set +# CONFIG_UFS_FS_WRITE is not set + +# +# Network File Systems +# +# CONFIG_CODA_FS is not set +CONFIG_NFS_FS=y +# CONFIG_NFS_V3 is not set +# CONFIG_ROOT_NFS is not set +CONFIG_NFSD=y +# CONFIG_NFSD_V3 is not set +CONFIG_SUNRPC=y +CONFIG_LOCKD=y +# CONFIG_SMB_FS is not set +# CONFIG_NCP_FS is not set +# CONFIG_NCPFS_PACKET_SIGNING is not set +# CONFIG_NCPFS_IOCTL_LOCKING is not set +# CONFIG_NCPFS_STRONG is not set +# CONFIG_NCPFS_NFS_NS is not set +# CONFIG_NCPFS_OS2_NS is not set +# CONFIG_NCPFS_SMALLDOS is not set +# CONFIG_NCPFS_NLS is not set +# CONFIG_NCPFS_EXTRAS is not set + +# +# Partition Types +# +# CONFIG_PARTITION_ADVANCED is not set +CONFIG_MSDOS_PARTITION=y +# CONFIG_SMB_NLS is not set +CONFIG_NLS=y + +# +# Native Language Support +# +CONFIG_NLS_DEFAULT="iso8859-1" +CONFIG_NLS_CODEPAGE_437=y +# CONFIG_NLS_CODEPAGE_737 is not set +# CONFIG_NLS_CODEPAGE_775 is not set +# CONFIG_NLS_CODEPAGE_850 is not set +# CONFIG_NLS_CODEPAGE_852 is not set +# CONFIG_NLS_CODEPAGE_855 is not set +# CONFIG_NLS_CODEPAGE_857 is not set +# CONFIG_NLS_CODEPAGE_860 is not set +# CONFIG_NLS_CODEPAGE_861 is not set +# CONFIG_NLS_CODEPAGE_862 is not set +# CONFIG_NLS_CODEPAGE_863 is not set +# CONFIG_NLS_CODEPAGE_864 is not set +# CONFIG_NLS_CODEPAGE_865 is not set +# CONFIG_NLS_CODEPAGE_866 is not set +# CONFIG_NLS_CODEPAGE_869 is not set +# CONFIG_NLS_CODEPAGE_874 is not set +# CONFIG_NLS_CODEPAGE_932 is not set +# CONFIG_NLS_CODEPAGE_936 is not set +# CONFIG_NLS_CODEPAGE_949 is not set +# CONFIG_NLS_CODEPAGE_950 is not set +CONFIG_NLS_ISO8859_1=y +# CONFIG_NLS_ISO8859_2 is not set +# CONFIG_NLS_ISO8859_3 is not set +# CONFIG_NLS_ISO8859_4 is not set +# CONFIG_NLS_ISO8859_5 is not set +# CONFIG_NLS_ISO8859_6 is not set +# CONFIG_NLS_ISO8859_7 is not set +# CONFIG_NLS_ISO8859_8 is not set +# CONFIG_NLS_ISO8859_9 is not set +# CONFIG_NLS_ISO8859_14 is not set +# CONFIG_NLS_ISO8859_15 is not set +# CONFIG_NLS_KOI8_R is not set +# CONFIG_NLS_UTF8 is not set + +# +# Sound +# +CONFIG_SOUND=y +# CONFIG_DMASOUND_AWACS is not set +# CONFIG_SOUND_CMPCI is not set +# CONFIG_SOUND_EMU10K1 is not set +# CONFIG_SOUND_FUSION is not set +# CONFIG_SOUND_CS4281 is not set +# CONFIG_SOUND_ES1370 is not set +# CONFIG_SOUND_ES1371 is not set +# CONFIG_SOUND_ESSSOLO1 is not set +# CONFIG_SOUND_MAESTRO is not set +# CONFIG_SOUND_SONICVIBES is not set +# CONFIG_SOUND_TRIDENT is not set +# CONFIG_SOUND_MSNDCLAS is not set +# CONFIG_SOUND_MSNDPIN is not set +# CONFIG_SOUND_VIA82CXXX is not set +CONFIG_SOUND_OSS=y +CONFIG_SOUND_TRACEINIT=y +# CONFIG_SOUND_DMAP is not set +# CONFIG_SOUND_AD1816 is not set +# CONFIG_SOUND_SGALAXY is not set +# CONFIG_SOUND_ADLIB is not set +# CONFIG_SOUND_ACI_MIXER is not set +CONFIG_SOUND_CS4232=m +# CONFIG_SOUND_SSCAPE is not set +# CONFIG_SOUND_GUS is not set +# CONFIG_SOUND_ICH is not set +# CONFIG_SOUND_VMIDI is not set +# CONFIG_SOUND_TRIX is not set +CONFIG_SOUND_MSS=m +# CONFIG_SOUND_MPU401 is not set +# CONFIG_SOUND_NM256 is not set +# CONFIG_SOUND_MAD16 is not set +# CONFIG_SOUND_PAS is not set +# CONFIG_PAS_JOYSTICK is not set +# CONFIG_SOUND_PSS is not set +# CONFIG_SOUND_SB is not set +# CONFIG_SOUND_AWE32_SYNTH is not set +# CONFIG_SOUND_WAVEFRONT is not set +# CONFIG_SOUND_MAUI is not set +# CONFIG_SOUND_YM3812 is not set +# CONFIG_SOUND_OPL3SA1 is not set +# CONFIG_SOUND_OPL3SA2 is not set +# CONFIG_SOUND_YMPCI is not set +# CONFIG_SOUND_YMFPCI is not set +# CONFIG_SOUND_UART6850 is not set +# CONFIG_SOUND_AEDSP16 is not set +CONFIG_SOUND_TVMIXER=y + +# +# USB support +# +# CONFIG_USB is not set + +# +# Kernel hacking +# +CONFIG_MAGIC_SYSRQ=y +# CONFIG_KGDB is not set +CONFIG_XMON=y diff --git a/arch/ppc/configs/rpxcllf_defconfig b/arch/ppc/configs/rpxcllf_defconfig index 78ee82230..258d3e79f 100644 --- a/arch/ppc/configs/rpxcllf_defconfig +++ b/arch/ppc/configs/rpxcllf_defconfig @@ -21,7 +21,6 @@ CONFIG_PPC=y # CONFIG_4xx is not set # CONFIG_POWER3 is not set # CONFIG_POWER4 is not set -# CONFIG_8260 is not set CONFIG_8xx=y CONFIG_SERIAL_CONSOLE=y # CONFIG_RPXLITE is not set @@ -32,6 +31,7 @@ CONFIG_RPXCLASSIC=y # CONFIG_TQM860 is not set # CONFIG_MBX is not set # CONFIG_WINCEPT is not set +CONFIG_PPC601_SYNC_FIX=y # CONFIG_ALL_PPC is not set # CONFIG_SMP is not set CONFIG_MACH_SPECIFIC=y @@ -43,7 +43,9 @@ CONFIG_MATH_EMULATION=y # CONFIG_HIGHMEM is not set # CONFIG_MOL is not set # CONFIG_ISA is not set +# CONFIG_EISA is not set # CONFIG_SBUS is not set +# CONFIG_MCA is not set # CONFIG_PCI_QSPAN is not set # CONFIG_PCI is not set CONFIG_NET=y @@ -182,7 +184,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_MACE is not set # CONFIG_BMAC is not set # CONFIG_GMAC is not set -# CONFIG_NCR885E is not set # CONFIG_OAKNET is not set # CONFIG_NET_VENDOR_3COM is not set # CONFIG_LANCE is not set @@ -190,7 +191,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_NET_VENDOR_RACAL is not set # CONFIG_AT1700 is not set # CONFIG_DEPCA is not set -# CONFIG_HP100 is not set # CONFIG_NET_ISA is not set # CONFIG_NET_PCI is not set # CONFIG_NET_POCKET is not set @@ -377,8 +377,6 @@ CONFIG_LOCKD=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -395,6 +393,7 @@ CONFIG_PARTITION_ADVANCED=y # CONFIG_SGI_PARTITION is not set # CONFIG_ULTRIX_PARTITION is not set # CONFIG_SUN_PARTITION is not set +# CONFIG_SMB_NLS is not set # CONFIG_NLS is not set # diff --git a/arch/ppc/configs/rpxlite_defconfig b/arch/ppc/configs/rpxlite_defconfig index 741d9265f..df00f5ace 100644 --- a/arch/ppc/configs/rpxlite_defconfig +++ b/arch/ppc/configs/rpxlite_defconfig @@ -21,7 +21,6 @@ CONFIG_PPC=y # CONFIG_4xx is not set # CONFIG_POWER3 is not set # CONFIG_POWER4 is not set -# CONFIG_8260 is not set CONFIG_8xx=y CONFIG_SERIAL_CONSOLE=y CONFIG_RPXLITE=y @@ -32,6 +31,7 @@ CONFIG_RPXLITE=y # CONFIG_TQM860 is not set # CONFIG_MBX is not set # CONFIG_WINCEPT is not set +CONFIG_PPC601_SYNC_FIX=y # CONFIG_ALL_PPC is not set # CONFIG_SMP is not set CONFIG_MACH_SPECIFIC=y @@ -43,7 +43,9 @@ CONFIG_MATH_EMULATION=y # CONFIG_HIGHMEM is not set # CONFIG_MOL is not set # CONFIG_ISA is not set +# CONFIG_EISA is not set # CONFIG_SBUS is not set +# CONFIG_MCA is not set # CONFIG_PCI_QSPAN is not set # CONFIG_PCI is not set CONFIG_NET=y @@ -182,7 +184,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_MACE is not set # CONFIG_BMAC is not set # CONFIG_GMAC is not set -# CONFIG_NCR885E is not set # CONFIG_OAKNET is not set # CONFIG_NET_VENDOR_3COM is not set # CONFIG_LANCE is not set @@ -190,7 +191,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_NET_VENDOR_RACAL is not set # CONFIG_AT1700 is not set # CONFIG_DEPCA is not set -# CONFIG_HP100 is not set # CONFIG_NET_ISA is not set # CONFIG_NET_PCI is not set # CONFIG_NET_POCKET is not set @@ -377,8 +377,6 @@ CONFIG_LOCKD=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -395,6 +393,7 @@ CONFIG_PARTITION_ADVANCED=y # CONFIG_SGI_PARTITION is not set # CONFIG_ULTRIX_PARTITION is not set # CONFIG_SUN_PARTITION is not set +# CONFIG_SMB_NLS is not set # CONFIG_NLS is not set # diff --git a/arch/ppc/configs/walnut_defconfig b/arch/ppc/configs/walnut_defconfig index c3e95c6ac..eef439fc2 100644 --- a/arch/ppc/configs/walnut_defconfig +++ b/arch/ppc/configs/walnut_defconfig @@ -12,7 +12,7 @@ CONFIG_EXPERIMENTAL=y # Loadable module support # CONFIG_MODULES=y -# CONFIG_MODVERSIONS is not set +CONFIG_MODVERSIONS=y CONFIG_KMOD=y # @@ -23,10 +23,10 @@ CONFIG_PPC=y CONFIG_4xx=y # CONFIG_POWER3 is not set # CONFIG_POWER4 is not set -# CONFIG_8260 is not set # CONFIG_8xx is not set # CONFIG_OAK is not set CONFIG_WALNUT=y +CONFIG_PPC601_SYNC_FIX=y # CONFIG_SMP is not set CONFIG_MACH_SPECIFIC=y # CONFIG_MATH_EMULATION is not set @@ -37,7 +37,9 @@ CONFIG_MACH_SPECIFIC=y # CONFIG_HIGHMEM is not set # CONFIG_MOL is not set # CONFIG_ISA is not set +# CONFIG_EISA is not set # CONFIG_SBUS is not set +# CONFIG_MCA is not set # CONFIG_PCI is not set CONFIG_NET=y CONFIG_SYSCTL=y @@ -171,7 +173,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_MACE is not set # CONFIG_BMAC is not set # CONFIG_GMAC is not set -# CONFIG_NCR885E is not set # CONFIG_OAKNET is not set # CONFIG_NET_VENDOR_3COM is not set # CONFIG_LANCE is not set @@ -179,7 +180,6 @@ CONFIG_NET_ETHERNET=y # CONFIG_NET_VENDOR_RACAL is not set # CONFIG_AT1700 is not set # CONFIG_DEPCA is not set -# CONFIG_HP100 is not set # CONFIG_NET_ISA is not set # CONFIG_NET_PCI is not set # CONFIG_NET_POCKET is not set @@ -369,8 +369,6 @@ CONFIG_LOCKD=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -379,6 +377,7 @@ CONFIG_LOCKD=y # # CONFIG_PARTITION_ADVANCED is not set CONFIG_MSDOS_PARTITION=y +# CONFIG_SMB_NLS is not set # CONFIG_NLS is not set # diff --git a/arch/ppc/defconfig b/arch/ppc/defconfig index 5574ecbdb..210887de7 100644 --- a/arch/ppc/defconfig +++ b/arch/ppc/defconfig @@ -23,12 +23,11 @@ CONFIG_6xx=y # CONFIG_4xx is not set # CONFIG_POWER3 is not set # CONFIG_POWER4 is not set -# CONFIG_8260 is not set # CONFIG_8xx is not set +# CONFIG_8260 is not set CONFIG_ALL_PPC=y -# CONFIG_GEMINI is not set -# CONFIG_EST8260 is not set # CONFIG_APUS is not set +CONFIG_PPC601_SYNC_FIX=y # CONFIG_SMP is not set CONFIG_ALTIVEC=y @@ -36,9 +35,11 @@ CONFIG_ALTIVEC=y # General setup # # CONFIG_HIGHMEM is not set -# CONFIG_MOL is not set +CONFIG_MOL=y # CONFIG_ISA is not set +# CONFIG_EISA is not set # CONFIG_SBUS is not set +# CONFIG_MCA is not set CONFIG_PCI=y CONFIG_NET=y CONFIG_SYSCTL=y @@ -49,21 +50,24 @@ CONFIG_BINFMT_ELF=y CONFIG_KERNEL_ELF=y CONFIG_BINFMT_MISC=m CONFIG_PCI_NAMES=y -# CONFIG_HOTPLUG is not set +CONFIG_HOTPLUG=y + +# +# PCMCIA/CardBus support +# # CONFIG_PCMCIA is not set # # Parallel port support # # CONFIG_PARPORT is not set -CONFIG_VGA_CONSOLE=y -CONFIG_FB=y -CONFIG_FB_COMPAT_XPMAC=y CONFIG_PPC_RTC=y CONFIG_PROC_DEVICETREE=y +CONFIG_PPC_RTAS=y CONFIG_BOOTX_TEXT=y -# CONFIG_MOTOROLA_HOTSWAP is not set -# CONFIG_CMDLINE_BOOL is not set +CONFIG_PREP_RESIDUAL=y +CONFIG_CMDLINE_BOOL=y +CONFIG_CMDLINE="console=ttyS0,9600 console=tty0 root=/dev/sda2" # # Memory Technology Devices (MTD) @@ -79,7 +83,7 @@ CONFIG_BOOTX_TEXT=y # # Block devices # -# CONFIG_BLK_DEV_FD is not set +CONFIG_BLK_DEV_FD=m # CONFIG_BLK_DEV_XD is not set # CONFIG_PARIDE is not set # CONFIG_BLK_CPQ_DA is not set @@ -111,7 +115,8 @@ CONFIG_PACKET=y CONFIG_NETLINK=y # CONFIG_RTNETLINK is not set # CONFIG_NETLINK_DEV is not set -# CONFIG_NETFILTER is not set +CONFIG_NETFILTER=y +# CONFIG_NETFILTER_DEBUG is not set # CONFIG_FILTER is not set CONFIG_UNIX=y CONFIG_INET=y @@ -123,6 +128,34 @@ CONFIG_IP_MULTICAST=y # CONFIG_IP_MROUTE is not set # CONFIG_INET_ECN is not set CONFIG_SYN_COOKIES=y + +# +# IP: Netfilter Configuration +# +CONFIG_IP_NF_CONNTRACK=m +CONFIG_IP_NF_FTP=m +# CONFIG_IP_NF_QUEUE is not set +CONFIG_IP_NF_IPTABLES=m +CONFIG_IP_NF_MATCH_LIMIT=m +CONFIG_IP_NF_MATCH_MAC=m +CONFIG_IP_NF_MATCH_MARK=m +CONFIG_IP_NF_MATCH_MULTIPORT=m +CONFIG_IP_NF_MATCH_TOS=m +CONFIG_IP_NF_MATCH_STATE=m +CONFIG_IP_NF_MATCH_UNCLEAN=m +CONFIG_IP_NF_MATCH_OWNER=m +CONFIG_IP_NF_FILTER=m +CONFIG_IP_NF_TARGET_REJECT=m +CONFIG_IP_NF_TARGET_MIRROR=m +CONFIG_IP_NF_NAT=m +CONFIG_IP_NF_NAT_NEEDED=y +CONFIG_IP_NF_TARGET_MASQUERADE=m +CONFIG_IP_NF_TARGET_REDIRECT=m +# CONFIG_IP_NF_MANGLE is not set +# CONFIG_IP_NF_TARGET_LOG is not set +CONFIG_IP_NF_COMPAT_IPCHAINS=m +CONFIG_IP_NF_NAT_NEEDED=y +# CONFIG_IP_NF_COMPAT_IPFWADM is not set # CONFIG_IPV6 is not set # CONFIG_KHTTPD is not set # CONFIG_ATM is not set @@ -239,6 +272,7 @@ CONFIG_SCSI=y CONFIG_BLK_DEV_SD=y CONFIG_SD_EXTRA_DEVS=40 CONFIG_CHR_DEV_ST=y +# CONFIG_CHR_DEV_OSST is not set CONFIG_BLK_DEV_SR=y CONFIG_BLK_DEV_SR_VENDOR=y CONFIG_SR_EXTRA_DEVS=2 @@ -344,7 +378,6 @@ CONFIG_NET_ETHERNET=y CONFIG_MACE=y CONFIG_BMAC=y CONFIG_GMAC=y -# CONFIG_NCR885E is not set # CONFIG_OAKNET is not set # CONFIG_NET_VENDOR_3COM is not set # CONFIG_LANCE is not set @@ -357,27 +390,28 @@ CONFIG_GMAC=y CONFIG_NET_PCI=y CONFIG_PCNET32=y # CONFIG_ADAPTEC_STARFIRE is not set -# CONFIG_AC3200 is not set # CONFIG_APRICOT is not set # CONFIG_CS89x0 is not set -CONFIG_DE4X5=y -# CONFIG_TULIP is not set +CONFIG_TULIP=y +CONFIG_DE4X5=m # CONFIG_DGRS is not set # CONFIG_DM9102 is not set # CONFIG_EEPRO100 is not set +# CONFIG_EEPRO100_PM is not set # CONFIG_LNE390 is not set # CONFIG_NATSEMI is not set # CONFIG_NE2K_PCI is not set # CONFIG_NE3210 is not set # CONFIG_ES3210 is not set -# CONFIG_RTL8129 is not set # CONFIG_8139TOO is not set +# CONFIG_RTL8129 is not set # CONFIG_SIS900 is not set # CONFIG_EPIC100 is not set # CONFIG_SUNDANCE is not set # CONFIG_TLAN is not set # CONFIG_VIA_RHINE is not set # CONFIG_WINBOND_840 is not set +# CONFIG_HAPPYMEAL is not set # CONFIG_NET_POCKET is not set # @@ -439,6 +473,7 @@ CONFIG_PPP_DEFLATE=y # # Console drivers # +CONFIG_VGA_CONSOLE=y # # Frame-buffer support @@ -460,7 +495,8 @@ CONFIG_FB_IMSTT=y CONFIG_FB_MATROX=y CONFIG_FB_MATROX_MILLENIUM=y CONFIG_FB_MATROX_MYSTIQUE=y -CONFIG_FB_MATROX_G100=y +# CONFIG_FB_MATROX_G100 is not set +# CONFIG_FB_MATROX_G450 is not set # CONFIG_FB_MATROX_MULTIHEAD is not set CONFIG_FB_ATY=y CONFIG_FB_ATY128=y @@ -481,6 +517,7 @@ CONFIG_FONT_SUN12x22=y # CONFIG_FONT_6x11 is not set # CONFIG_FONT_PEARL_8x8 is not set # CONFIG_FONT_ACORN_8x8 is not set +# CONFIG_FB_COMPAT_XPMAC is not set # # Input core support @@ -498,15 +535,13 @@ CONFIG_INPUT_EVDEV=y # CONFIG_ADB_CUDA=y CONFIG_ADB_PMU=y -CONFIG_PMAC_PBOOK=y -CONFIG_PMAC_BACKLIGHT=y -CONFIG_MAC_FLOPPY=y -CONFIG_MAC_SERIAL=y -# CONFIG_SERIAL_CONSOLE is not set +# CONFIG_PMAC_PBOOK is not set +# CONFIG_PMAC_BACKLIGHT is not set +# CONFIG_MAC_FLOPPY is not set +CONFIG_MAC_SERIAL=m CONFIG_ADB=y CONFIG_ADB_MACIO=y CONFIG_INPUT_ADBHID=y -CONFIG_MAC_HID=y CONFIG_MAC_ADBKEYCODES=y CONFIG_MAC_EMUMOUSEBTN=y @@ -575,17 +610,17 @@ CONFIG_NVRAM=y # File systems # # CONFIG_QUOTA is not set -CONFIG_AUTOFS_FS=y -CONFIG_AUTOFS4_FS=y +# CONFIG_AUTOFS_FS is not set +# CONFIG_AUTOFS4_FS is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set # CONFIG_AFFS_FS is not set -# CONFIG_HFS_FS is not set +CONFIG_HFS_FS=m # CONFIG_BFS_FS is not set -CONFIG_FAT_FS=y -CONFIG_MSDOS_FS=y +CONFIG_FAT_FS=m +CONFIG_MSDOS_FS=m # CONFIG_UMSDOS_FS is not set -CONFIG_VFAT_FS=y +CONFIG_VFAT_FS=m # CONFIG_EFS_FS is not set # CONFIG_JFFS_FS is not set # CONFIG_CRAMFS is not set @@ -631,8 +666,6 @@ CONFIG_LOCKD=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -652,6 +685,7 @@ CONFIG_MSDOS_PARTITION=y # CONFIG_SGI_PARTITION is not set # CONFIG_ULTRIX_PARTITION is not set # CONFIG_SUN_PARTITION is not set +# CONFIG_SMB_NLS is not set CONFIG_NLS=y # @@ -678,7 +712,7 @@ CONFIG_NLS_DEFAULT="iso8859-1" # CONFIG_NLS_CODEPAGE_936 is not set # CONFIG_NLS_CODEPAGE_949 is not set # CONFIG_NLS_CODEPAGE_950 is not set -# CONFIG_NLS_ISO8859_1 is not set +CONFIG_NLS_ISO8859_1=m # CONFIG_NLS_ISO8859_2 is not set # CONFIG_NLS_ISO8859_3 is not set # CONFIG_NLS_ISO8859_4 is not set @@ -695,9 +729,9 @@ CONFIG_NLS_DEFAULT="iso8859-1" # # Sound # -CONFIG_SOUND=y -CONFIG_DMASOUND_AWACS=y -CONFIG_DMASOUND=y +CONFIG_SOUND=m +CONFIG_DMASOUND_AWACS=m +CONFIG_DMASOUND=m # CONFIG_SOUND_CMPCI is not set # CONFIG_SOUND_EMU10K1 is not set # CONFIG_SOUND_FUSION is not set @@ -711,43 +745,14 @@ CONFIG_DMASOUND=y # CONFIG_SOUND_MSNDCLAS is not set # CONFIG_SOUND_MSNDPIN is not set # CONFIG_SOUND_VIA82CXXX is not set -CONFIG_SOUND_OSS=y -# CONFIG_SOUND_TRACEINIT is not set -# CONFIG_SOUND_DMAP is not set -# CONFIG_SOUND_AD1816 is not set -# CONFIG_SOUND_SGALAXY is not set -# CONFIG_SOUND_ADLIB is not set -# CONFIG_SOUND_ACI_MIXER is not set -CONFIG_SOUND_CS4232=m -# CONFIG_SOUND_SSCAPE is not set -# CONFIG_SOUND_GUS is not set -# CONFIG_SOUND_ICH is not set -# CONFIG_SOUND_VMIDI is not set -# CONFIG_SOUND_TRIX is not set -# CONFIG_SOUND_MSS is not set -# CONFIG_SOUND_MPU401 is not set -# CONFIG_SOUND_NM256 is not set -# CONFIG_SOUND_MAD16 is not set -# CONFIG_SOUND_PAS is not set -# CONFIG_PAS_JOYSTICK is not set -# CONFIG_SOUND_PSS is not set -# CONFIG_SOUND_SB is not set -# CONFIG_SOUND_AWE32_SYNTH is not set -# CONFIG_SOUND_WAVEFRONT is not set -# CONFIG_SOUND_MAUI is not set -# CONFIG_SOUND_YM3812 is not set -# CONFIG_SOUND_OPL3SA1 is not set -# CONFIG_SOUND_OPL3SA2 is not set -# CONFIG_SOUND_YMPCI is not set -# CONFIG_SOUND_UART6850 is not set -# CONFIG_SOUND_AEDSP16 is not set +# CONFIG_SOUND_OSS is not set # CONFIG_SOUND_TVMIXER is not set # # USB support # CONFIG_USB=y -CONFIG_USB_DEBUG=y +# CONFIG_USB_DEBUG is not set # # Miscellaneous USB options @@ -763,33 +768,69 @@ CONFIG_USB_DEVICEFS=y CONFIG_USB_OHCI=y # -# USB Devices +# USB Device Class drivers # +# CONFIG_USB_AUDIO is not set +# CONFIG_USB_BLUETOOTH is not set +# CONFIG_USB_STORAGE is not set +CONFIG_USB_ACM=m # CONFIG_USB_PRINTER is not set + +# +# USB Human Interface Devices (HID) +# +CONFIG_USB_HID=y +# CONFIG_USB_WACOM is not set + +# +# USB Imaging devices +# +# CONFIG_USB_DC2XX is not set +# CONFIG_USB_MDC800 is not set # CONFIG_USB_SCANNER is not set # CONFIG_USB_MICROTEK is not set -# CONFIG_USB_AUDIO is not set -# CONFIG_USB_ACM is not set -# CONFIG_USB_SERIAL is not set + +# +# USB Multimedia devices +# # CONFIG_USB_IBMCAM is not set # CONFIG_USB_OV511 is not set -# CONFIG_USB_DC2XX is not set -# CONFIG_USB_MDC800 is not set -# CONFIG_USB_STORAGE is not set -# CONFIG_USB_USS720 is not set +# CONFIG_USB_DSBR is not set # CONFIG_USB_DABUSB is not set + +# +# USB Network adaptors +# # CONFIG_USB_PLUSB is not set # CONFIG_USB_PEGASUS is not set -# CONFIG_USB_RIO500 is not set -# CONFIG_USB_DSBR is not set -# CONFIG_USB_BLUETOOTH is not set # CONFIG_USB_NET1080 is not set # -# USB Human Interface Devices (HID) +# USB port drivers # -CONFIG_USB_HID=y -# CONFIG_USB_WACOM is not set +# CONFIG_USB_USS720 is not set + +# +# USB Serial Converter support +# +CONFIG_USB_SERIAL=m +# CONFIG_USB_SERIAL_DEBUG is not set +# CONFIG_USB_SERIAL_GENERIC is not set +# CONFIG_USB_SERIAL_BELKIN is not set +# CONFIG_USB_SERIAL_WHITEHEAT is not set +# CONFIG_USB_SERIAL_DIGI_ACCELEPORT is not set +# CONFIG_USB_SERIAL_EMPEG is not set +# CONFIG_USB_SERIAL_FTDI_SIO is not set +CONFIG_USB_SERIAL_VISOR=m +# CONFIG_USB_SERIAL_KEYSPAN_PDA is not set +# CONFIG_USB_SERIAL_KEYSPAN is not set +# CONFIG_USB_SERIAL_MCT_U232 is not set +# CONFIG_USB_SERIAL_OMNINET is not set + +# +# USB misc drivers +# +# CONFIG_USB_RIO500 is not set # # Kernel hacking diff --git a/arch/ppc/kernel/Makefile b/arch/ppc/kernel/Makefile index 4a9852612..545fce0cd 100644 --- a/arch/ppc/kernel/Makefile +++ b/arch/ppc/kernel/Makefile @@ -35,6 +35,7 @@ obj-$(CONFIG_PPC) := entry.o traps.o irq.o idle.o time.o misc.o \ process.o signal.o bitops.o ptrace.o \ ppc_htab.o semaphore.o syscalls.o \ align.o setup.o +obj-$(CONFIG_MODULES) += ppc_ksyms.o obj-$(CONFIG_POWER4) += xics.o obj-$(CONFIG_PCI) += pci.o pci-dma.o obj-$(CONFIG_KGDB) += ppc-stub.o @@ -62,11 +63,8 @@ obj-$(CONFIG_ALL_PPC) += pmac_pic.o pmac_setup.o pmac_time.o prom.o \ feature.o pmac_pci.o chrp_setup.o \ chrp_time.o chrp_pci.o open_pic.o \ indirect_pci.o i8259.o prep_pci.o \ - prep_time.o prep_nvram.o ppc_ksyms.o \ - prep_setup.o + prep_time.o prep_nvram.o prep_setup.o obj-$(CONFIG_PMAC_BACKLIGHT) += pmac_backlight.o -obj-$(CONFIG_GEMINI) += gemini_prom.o gemini_pci.o gemini_setup.o \ - open_pic.o obj-$(CONFIG_8260) += m8260_setup.o ppc8260_pic.o diff --git a/arch/ppc/kernel/apus_setup.c b/arch/ppc/kernel/apus_setup.c index 5979c9312..70a683d20 100644 --- a/arch/ppc/kernel/apus_setup.c +++ b/arch/ppc/kernel/apus_setup.c @@ -647,25 +647,6 @@ apus_ide_default_io_base(int index) return 0; } -int -apus_ide_check_region(ide_ioreg_t from, unsigned int extent) -{ - return 0; -} - -void -apus_ide_request_region(ide_ioreg_t from, - unsigned int extent, - const char *name) -{ -} - -void -apus_ide_release_region(ide_ioreg_t from, - unsigned int extent) -{ -} - void apus_ide_fix_driveid(struct hd_driveid *id) { @@ -1131,9 +1112,6 @@ void apus_init(unsigned long r3, unsigned long r4, unsigned long r5, ppc_ide_md.outsw = apus_ide_outsw; ppc_ide_md.default_irq = apus_ide_default_irq; ppc_ide_md.default_io_base = apus_ide_default_io_base; - ppc_ide_md.ide_check_region = apus_ide_check_region; - ppc_ide_md.ide_request_region = apus_ide_request_region; - ppc_ide_md.ide_release_region = apus_ide_release_region; ppc_ide_md.fix_driveid = apus_ide_fix_driveid; ppc_ide_md.ide_init_hwif = apus_ide_init_hwif_ports; diff --git a/arch/ppc/kernel/chrp_pci.c b/arch/ppc/kernel/chrp_pci.c index 43b678861..914ed02ff 100644 --- a/arch/ppc/kernel/chrp_pci.c +++ b/arch/ppc/kernel/chrp_pci.c @@ -8,8 +8,8 @@ #include <linux/delay.h> #include <linux/string.h> #include <linux/init.h> -#include <linux/openpic.h> #include <linux/ide.h> +#include <linux/bootmem.h> #include <asm/io.h> #include <asm/pgtable.h> @@ -19,16 +19,19 @@ #include <asm/gg2.h> #include <asm/machdep.h> #include <asm/init.h> +#include <asm/pci-bridge.h> +#include "open_pic.h" #include "pci.h" + #ifdef CONFIG_POWER4 -static unsigned long pci_address_offset(int, unsigned int); +extern unsigned long pci_address_offset(int, unsigned int); #endif /* CONFIG_POWER4 */ /* LongTrail */ -#define pci_config_addr(bus, dev, offset) \ -(GG2_PCI_CONFIG_BASE | ((bus)<<16) | ((dev)<<8) | (offset)) +#define pci_config_addr(dev, offset) \ +(GG2_PCI_CONFIG_BASE | ((dev->bus->number)<<16) | ((dev->devfn)<<8) | (offset)) volatile struct Hydra *Hydra = NULL; @@ -37,205 +40,127 @@ volatile struct Hydra *Hydra = NULL; * limit the bus number to 3 bits */ -int __chrp gg2_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char *val) -{ - if (bus > 7) { - *val = 0xff; - return PCIBIOS_DEVICE_NOT_FOUND; - } - *val = in_8((unsigned char *)pci_config_addr(bus, dev_fn, offset)); - return PCIBIOS_SUCCESSFUL; +#define cfg_read(val, addr, type, op) *val = op((type)(addr)) +#define cfg_write(val, addr, type, op) op((type *)(addr), (val)) + +#define cfg_read_bad(val, size) *val = bad_##size; +#define cfg_write_bad(val, size) + +#define bad_byte 0xff +#define bad_word 0xffff +#define bad_dword 0xffffffffU + +#define GG2_PCI_OP(rw, size, type, op) \ +int __chrp gg2_##rw##_config_##size(struct pci_dev *dev, int off, type val) \ +{ \ + if (dev->bus->number > 7) { \ + cfg_##rw##_bad(val, size) \ + return PCIBIOS_DEVICE_NOT_FOUND; \ + } \ + cfg_##rw(val, pci_config_addr(dev, off), type, op); \ + return PCIBIOS_SUCCESSFUL; \ } -int __chrp gg2_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short *val) -{ - if (bus > 7) { - *val = 0xffff; - return PCIBIOS_DEVICE_NOT_FOUND; - } - *val = in_le16((unsigned short *)pci_config_addr(bus, dev_fn, offset)); - return PCIBIOS_SUCCESSFUL; -} - - -int __chrp gg2_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int *val) -{ - if (bus > 7) { - *val = 0xffffffff; - return PCIBIOS_DEVICE_NOT_FOUND; - } - *val = in_le32((unsigned int *)pci_config_addr(bus, dev_fn, offset)); - return PCIBIOS_SUCCESSFUL; -} - -int __chrp gg2_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char val) -{ - if (bus > 7) - return PCIBIOS_DEVICE_NOT_FOUND; - out_8((unsigned char *)pci_config_addr(bus, dev_fn, offset), val); - return PCIBIOS_SUCCESSFUL; -} - -int __chrp gg2_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short val) -{ - if (bus > 7) - return PCIBIOS_DEVICE_NOT_FOUND; - out_le16((unsigned short *)pci_config_addr(bus, dev_fn, offset), val); - return PCIBIOS_SUCCESSFUL; -} +GG2_PCI_OP(read, byte, u8 *, in_8) +GG2_PCI_OP(read, word, u16 *, in_le16) +GG2_PCI_OP(read, dword, u32 *, in_le32) +GG2_PCI_OP(write, byte, u8, out_8) +GG2_PCI_OP(write, word, u16, out_le16) +GG2_PCI_OP(write, dword, u32, out_le32) -int __chrp gg2_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int val) +static struct pci_ops gg2_pci_ops = { - if (bus > 7) - return PCIBIOS_DEVICE_NOT_FOUND; - out_le32((unsigned int *)pci_config_addr(bus, dev_fn, offset), val); - return PCIBIOS_SUCCESSFUL; -} + gg2_read_config_byte, + gg2_read_config_word, + gg2_read_config_dword, + gg2_write_config_byte, + gg2_write_config_word, + gg2_write_config_dword +}; -#define python_config_address(bus) (unsigned *)((0xfef00000+0xf8000)-(bus*0x100000)) -#define python_config_data(bus) ((0xfef00000+0xf8010)-(bus*0x100000)) -#define PYTHON_CFA(b, d, o) (0x80 | ((b<<6) << 8) | ((d) << 16) \ +/* + * Access functions for PCI config space on IBM "python" host bridges. + */ +#define PYTHON_CFA(b, d, o) (0x80 | ((b) << 8) | ((d) << 16) \ | (((o) & ~3) << 24)) -unsigned int python_busnr = 0; - -int __chrp python_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char *val) -{ - if (bus > python_busnr) { - *val = 0xff; - return PCIBIOS_DEVICE_NOT_FOUND; - } - out_be32( python_config_address( bus ), PYTHON_CFA(bus,dev_fn,offset)); - *val = in_8((unsigned char *)python_config_data(bus) + (offset&3)); - return PCIBIOS_SUCCESSFUL; -} -int __chrp python_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short *val) -{ - if (bus > python_busnr) { - *val = 0xffff; - return PCIBIOS_DEVICE_NOT_FOUND; - } - out_be32( python_config_address( bus ), PYTHON_CFA(bus,dev_fn,offset)); - *val = in_le16((unsigned short *)(python_config_data(bus) + (offset&3))); - return PCIBIOS_SUCCESSFUL; +#define PYTHON_PCI_OP(rw, size, type, op, mask) \ +int __chrp \ +python_##rw##_config_##size(struct pci_dev *dev, int offset, type val) \ +{ \ + struct pci_controller *hose = dev->sysdata; \ + \ + out_be32(hose->cfg_addr, \ + PYTHON_CFA(dev->bus->number, dev->devfn, offset)); \ + cfg_##rw(val, hose->cfg_data + (offset & mask), type, op); \ + return PCIBIOS_SUCCESSFUL; \ } +PYTHON_PCI_OP(read, byte, u8 *, in_8, 3) +PYTHON_PCI_OP(read, word, u16 *, in_le16, 2) +PYTHON_PCI_OP(read, dword, u32 *, in_le32, 0) +PYTHON_PCI_OP(write, byte, u8, out_8, 3) +PYTHON_PCI_OP(write, word, u16, out_le16, 2) +PYTHON_PCI_OP(write, dword, u32, out_le32, 0) -int __chrp python_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int *val) -{ - if (bus > python_busnr) { - *val = 0xffffffff; - return PCIBIOS_DEVICE_NOT_FOUND; - } - out_be32( python_config_address( bus ), PYTHON_CFA(bus,dev_fn,offset)); - *val = in_le32((unsigned *)python_config_data(bus)); - return PCIBIOS_SUCCESSFUL; -} - -int __chrp python_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char val) +static struct pci_ops python_pci_ops = { - if (bus > python_busnr) - return PCIBIOS_DEVICE_NOT_FOUND; - out_be32( python_config_address( bus ), PYTHON_CFA(bus,dev_fn,offset)); - out_8((volatile unsigned char *)python_config_data(bus) + (offset&3), val); - return PCIBIOS_SUCCESSFUL; -} - -int __chrp python_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short val) -{ - if (bus > python_busnr) - return PCIBIOS_DEVICE_NOT_FOUND; - out_be32( python_config_address( bus ), PYTHON_CFA(bus,dev_fn,offset)); - out_le16((volatile unsigned short *)python_config_data(bus) + (offset&3), - val); - return PCIBIOS_SUCCESSFUL; -} - -int __chrp python_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int val) -{ - if (bus > python_busnr) - return PCIBIOS_DEVICE_NOT_FOUND; - out_be32( python_config_address( bus ), PYTHON_CFA(bus,dev_fn,offset)); - out_le32((unsigned *)python_config_data(bus) + (offset&3), val); - return PCIBIOS_SUCCESSFUL; -} - - -int __chrp rtas_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char *val) -{ - unsigned long addr = (offset&0xff) | ((dev_fn&0xff)<<8) | ((bus & 0xff)<<16); - unsigned long ret; - - if (call_rtas( "read-pci-config", 2, 2, &ret, addr, 1) != 0) - return PCIBIOS_DEVICE_NOT_FOUND; - *val = ret; - return PCIBIOS_SUCCESSFUL; -} - -int __chrp rtas_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short *val) -{ - unsigned long addr = (offset&0xff) | ((dev_fn&0xff)<<8) | ((bus & 0xff)<<16); - unsigned long ret; - - if (call_rtas("read-pci-config", 2, 2, &ret, addr, 2) != 0) - return PCIBIOS_DEVICE_NOT_FOUND; - *val = ret; - return PCIBIOS_SUCCESSFUL; -} - - -int __chrp rtas_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int *val) -{ - unsigned long addr = (offset&0xff) | ((dev_fn&0xff)<<8) | ((bus & 0xff)<<16); - unsigned long ret; + python_read_config_byte, + python_read_config_word, + python_read_config_dword, + python_write_config_byte, + python_write_config_word, + python_write_config_dword +}; - if (call_rtas("read-pci-config", 2, 2, &ret, addr, 4) != 0) - return PCIBIOS_DEVICE_NOT_FOUND; - *val = ret; - return PCIBIOS_SUCCESSFUL; +#ifdef CONFIG_POWER4 +/* + * Access functions for PCI config space using RTAS calls. + */ +#define RTAS_PCI_READ_OP(size, type, nbytes) \ +int __chrp \ +rtas_read_config_##size(struct pci_dev *dev, int offset, type val) \ +{ \ + unsigned long addr = (offset & 0xff) | ((dev->devfn & 0xff) << 8) \ + | ((dev->bus->number & 0xff) << 16); \ + unsigned long ret = ~0UL; \ + int rval; \ + \ + rval = call_rtas("read-pci-config", 2, 2, &ret, addr, nbytes); \ + *val = ret; \ + return rval? PCIBIOS_DEVICE_NOT_FOUND: PCIBIOS_SUCCESSFUL; \ } -int __chrp rtas_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char val) -{ - unsigned long addr = (offset&0xff) | ((dev_fn&0xff)<<8) | ((bus & 0xff)<<16); - if ( call_rtas( "write-pci-config", 3, 1, NULL, addr, 1, (ulong)val ) != 0 ) - return PCIBIOS_DEVICE_NOT_FOUND; - return PCIBIOS_SUCCESSFUL; +#define RTAS_PCI_WRITE_OP(size, type, nbytes) \ +int __chrp \ +rtas_write_config_##size(struct pci_dev *dev, int offset, type val) \ +{ \ + unsigned long addr = (offset & 0xff) | ((dev->devfn & 0xff) << 8) \ + | ((dev->bus->number & 0xff) << 16); \ + int rval; \ + \ + rval = call_rtas("write-pci-config", 3, 1, NULL, \ + addr, nbytes, (ulong)val); \ + return rval? PCIBIOS_DEVICE_NOT_FOUND: PCIBIOS_SUCCESSFUL; \ } -int __chrp rtas_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short val) -{ - unsigned long addr = (offset&0xff) | ((dev_fn&0xff)<<8) | ((bus & 0xff)<<16); - if ( call_rtas( "write-pci-config", 3, 1, NULL, addr, 2, (ulong)val ) != 0 ) - return PCIBIOS_DEVICE_NOT_FOUND; - return PCIBIOS_SUCCESSFUL; -} +RTAS_PCI_READ_OP(byte, u8 *, 1) +RTAS_PCI_READ_OP(word, u16 *, 2) +RTAS_PCI_READ_OP(dword, u32 *, 4) +RTAS_PCI_WRITE_OP(byte, u8, 1) +RTAS_PCI_WRITE_OP(word, u16, 2) +RTAS_PCI_WRITE_OP(dword, u32, 4) -int __chrp rtas_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int val) +static struct pci_ops rtas_pci_ops = { - unsigned long addr = (offset&0xff) | ((dev_fn&0xff)<<8) | ((bus & 0xff)<<16); - if ( call_rtas( "write-pci-config", 3, 1, NULL, addr, 4, (ulong)val ) != 0 ) - return PCIBIOS_DEVICE_NOT_FOUND; - return PCIBIOS_SUCCESSFUL; -} + rtas_read_config_byte, + rtas_read_config_word, + rtas_read_config_dword, + rtas_write_config_byte, + rtas_write_config_word, + rtas_write_config_dword +}; +#endif /* CONFIG_POWER4 */ /* * Temporary fixes for PCI devices. These should be replaced by OF query @@ -278,7 +203,7 @@ hydra_init(void) HYDRA_FC_MPIC_ENABLE | HYDRA_FC_SLOW_SCC_PCLK | HYDRA_FC_MPIC_IS_MASTER)); - OpenPIC = (volatile struct OpenPIC *)&Hydra->OpenPIC; + OpenPIC_Addr = &Hydra->OpenPIC; OpenPIC_InitSenses = hydra_openpic_initsenses; OpenPIC_NumInitSenses = sizeof(hydra_openpic_initsenses); return 1; @@ -316,54 +241,25 @@ void __init chrp_pcibios_fixup(void) { struct pci_dev *dev; - int *brp; struct device_node *np; - extern struct pci_ops generic_pci_ops; - -#ifndef CONFIG_POWER4 - np = find_devices("device-tree"); - if (np != 0) { - for (np = np->child; np != NULL; np = np->sibling) { - if (np->type == NULL || strcmp(np->type, "pci") != 0) - continue; - if ((brp = (int *) get_property(np, "bus-range", NULL)) == 0) - continue; - if (brp[0] != 0) /* bus 0 is already done */ - pci_scan_bus(brp[0], &generic_pci_ops, NULL); - } - } -#else - /* XXX kludge for now because we can't properly handle - physical addresses > 4GB. -- paulus */ - pci_scan_bus(0x1e, &generic_pci_ops, NULL); -#endif /* CONFIG_POWER4 */ /* PCI interrupts are controlled by the OpenPIC */ pci_for_each_dev(dev) { - np = find_pci_device_OFnode(dev->bus->number, dev->devfn); + np = pci_device_to_OF_node(dev); if ((np != 0) && (np->n_intrs > 0) && (np->intrs[0].line != 0)) dev->irq = np->intrs[0].line; - /* these need to be absolute addrs for OF and Matrox FB -- Cort */ - if ( dev->vendor == PCI_VENDOR_ID_MATROX ) - { - if ( dev->resource[0].start < isa_mem_base ) - dev->resource[0].start += isa_mem_base; - if ( dev->resource[1].start < isa_mem_base ) - dev->resource[1].start += isa_mem_base; - } + pci_write_config_byte(dev, PCI_INTERRUPT_LINE, dev->irq); + /* the F50 identifies the amd as a trident */ if ( (dev->vendor == PCI_VENDOR_ID_TRIDENT) && (dev->class>>8 == PCI_CLASS_NETWORK_ETHERNET) ) { dev->vendor = PCI_VENDOR_ID_AMD; - pcibios_write_config_word(dev->bus->number, - dev->devfn, PCI_VENDOR_ID, PCI_VENDOR_ID_AMD); + pci_write_config_word(dev, PCI_VENDOR_ID, + PCI_VENDOR_ID_AMD); } #ifdef CONFIG_POWER4 power4_fixup_dev(dev); -#else - if (dev->bus->number > 0 && python_busnr > 0) - dev->resource[0].start += dev->bus->number*0x01000000; #endif } } @@ -402,86 +298,213 @@ static void __init gg2_pcibios_fixup_bus(struct pci_bus *bus) bus->resource[1] = &gg2_resources.pci_mem; } -decl_config_access_method(grackle); -decl_config_access_method(indirect); -decl_config_access_method(rtas); +static void process_bridge_ranges(struct pci_controller *hose, + struct device_node *dev, int index) +{ + unsigned int *ranges; + int rlen = 0; + int memno = 0; + struct resource *res; + + hose->io_base_phys = 0; + ranges = (unsigned int *) get_property(dev, "ranges", &rlen); + while ((rlen -= 6 * sizeof(unsigned int)) >= 0) { + res = NULL; + switch (ranges[0] >> 24) { + case 1: /* I/O space */ + if (ranges[2] != 0) + break; + hose->io_base_phys = ranges[3]; + hose->io_base_virt = ioremap(ranges[3], ranges[5]); + if (index == 0) { + isa_io_base = (unsigned long) hose->io_base_virt; + printk("isa_io_base=%lx\n", isa_io_base); + } + res = &hose->io_resource; + res->flags = IORESOURCE_IO; + break; + case 2: /* memory space */ + if (index == 0 && ranges[1] == 0 && ranges[2] == 0){ + isa_mem_base = ranges[3]; + printk("isa_mem_base=%lx\n", isa_mem_base); + } + if (memno == 0) { + hose->pci_mem_offset = ranges[3] - ranges[2]; + printk("pci_mem_offset=%lx for this bridge\n", + hose->pci_mem_offset); + } + res = &hose->mem_resources[memno]; + res->flags = IORESOURCE_MEM; + ++memno; + break; + } + if (res != NULL) { + res->name = dev->full_name; + res->start = ranges[3]; + res->end = res->start + ranges[5] - 1; + res->parent = NULL; + res->sibling = NULL; + res->child = NULL; + } + ranges += 6; + } +} + +/* this is largely modeled and stolen after the pmac_pci code -- tgall + */ + +static void __init +ibm_add_bridges(struct device_node *dev) +{ + int *bus_range; + int len, index = 0; + struct pci_controller *hose; + volatile unsigned char *cfg; + unsigned int *dma; +#ifdef CONFIG_POWER3 + unsigned long *opprop = (unsigned long *) + get_property(find_path_device("/"), "platform-open-pic", NULL); +#endif + + for(; dev != NULL; dev = dev->next, ++index) { + if (dev->n_addrs < 1) { + printk(KERN_WARNING "Can't use %s: no address\n", + dev->full_name); + continue; + } + bus_range = (int *) get_property(dev, "bus-range", &len); + if (bus_range == NULL || len < 2 * sizeof(int)) { + printk(KERN_WARNING "Can't get bus-range for %s\n", + dev->full_name); + continue; + } + if (bus_range[1] == bus_range[0]) + printk(KERN_INFO "PCI bus %d", bus_range[0]); + else + printk(KERN_INFO "PCI buses %d..%d", + bus_range[0], bus_range[1]); + printk(" controlled by %s at %x\n", dev->type, + dev->addrs[0].address); + + hose = pcibios_alloc_controller(); + if (!hose) { + printk("Can't allocate PCI controller structure for %s\n", + dev->full_name); + continue; + } + hose->arch_data = dev; + hose->first_busno = bus_range[0]; + hose->last_busno = bus_range[1]; + hose->ops = &python_pci_ops; + + cfg = ioremap(dev->addrs[0].address + 0xf8000, 0x20); + hose->cfg_addr = (volatile unsigned int *) cfg; + hose->cfg_data = cfg + 0x10; + + process_bridge_ranges(hose, dev, index); + +#ifdef CONFIG_POWER3 + openpic_setup_ISU(index, opprop[index+1]); +#endif /* CONFIG_POWER3 */ + + /* check the first bridge for a property that we can + use to set pci_dram_offset */ + dma = (unsigned int *) + get_property(dev, "ibm,dma-ranges", &len); + if (index == 0 && dma != NULL && len >= 6 * sizeof(*dma)) { + pci_dram_offset = dma[2] - dma[3]; + printk("pci_dram_offset = %lx\n", pci_dram_offset); + } + } +} + +#ifdef CONFIG_POWER4 +void __init +power4_add_bridge(void) +{ + struct pci_controller* hose; + + hose = pcibios_alloc_controller(); + if (!hose) + return; + hose->first_busno = 0; + hose->last_busno = 0xff; + + hose->ops = &rtas_pci_ops; + pci_dram_offset = 0; +} +#endif /* CONFIG_POWER4 */ void __init -chrp_setup_pci_ptrs(void) +chrp_find_bridges(void) { struct device_node *py; + char *model, *name; + struct pci_controller* hose; ppc_md.pcibios_fixup = chrp_pcibios_fixup; + #ifdef CONFIG_POWER4 - set_config_access_method(rtas); - pci_dram_offset = 0; + power4_add_bridge(); #else /* CONFIG_POWER4 */ - if ( !strncmp("MOT", - get_property(find_path_device("/"), "model", NULL),3) ) - { - pci_dram_offset = 0; - isa_mem_base = 0xf7000000; - isa_io_base = 0xfe000000; - set_config_access_method(grackle); - } - else - { - if ((py = find_compatible_devices("pci", "IBM,python")) != 0 - || (py = find_compatible_devices("pci", "IBM,python3.0")) != 0) - { - char *name = get_property(find_path_device("/"), "name", NULL); - - /* find out how many pythons */ - while ( (py = py->next) ) python_busnr++; - set_config_access_method(python); - - /* - * We base these values on the machine type but should - * try to read them from the python controller itself. - * -- Cort - */ - if ( !strncmp("IBM,7025-F50", name, 12) ) - { - pci_dram_offset = 0x80000000; - isa_mem_base = 0xa0000000; - isa_io_base = 0x88000000; - } else if ( !strncmp("IBM,7043-260", name, 12) - || !strncmp("IBM,7044-270", name, 12)) - { - pci_dram_offset = 0x0; - isa_mem_base = 0xc0000000; - isa_io_base = 0xf8000000; - } - } - else - { - if ( !strncmp("IBM,7043-150", get_property(find_path_device("/"), "name", NULL),12) || - !strncmp("IBM,7046-155", get_property(find_path_device("/"), "name", NULL),12) || - !strncmp("IBM,7046-B50", get_property(find_path_device("/"), "name", NULL),12) ) - { - pci_dram_offset = 0; - isa_mem_base = 0x80000000; - isa_io_base = 0xfe000000; - pci_config_address = (unsigned int *)0xfec00000; - pci_config_data = (unsigned char *)0xfee00000; - set_config_access_method(indirect); - } - else - { - /* LongTrail */ - pci_dram_offset = 0; - isa_mem_base = 0xf7000000; - isa_io_base = 0xf8000000; - set_config_access_method(gg2); - ppc_md.pcibios_fixup = gg2_pcibios_fixup; - ppc_md.pcibios_fixup_bus = gg2_pcibios_fixup_bus; - } - } + model = get_property(find_path_device("/"), "model", NULL); + if (!strncmp("MOT", model, 3)) { + struct pci_controller* hose; + + hose = pcibios_alloc_controller(); + if (!hose) + return; + hose->first_busno = 0; + hose->last_busno = 0xff; + /* Check that please. This must be the root of the OF + * PCI tree (the root host bridge + */ + hose->arch_data = find_devices("pci"); + setup_grackle(hose, 0x20000); + return; } + + if ((py = find_compatible_devices("pci", "IBM,python"))) + { + /* XXX xmon_init_scc needs this set and the BAT + set up in MMU_init */ + ibm_add_bridges(find_devices("pci")); + return; + } + + + hose = pcibios_alloc_controller(); + if (!hose) + return; + hose->first_busno = 0; + hose->last_busno = 0xff; + /* Check that please. This must be the root of the OF + * PCI tree (the root host bridge + */ + hose->arch_data = find_devices("pci"); + name = get_property(find_path_device("/"), "name", NULL); + if (!strncmp("IBM,7043-150", name, 12) || + !strncmp("IBM,7046-155", name, 12) || + !strncmp("IBM,7046-B50", name, 12) ) { + setup_grackle(hose, 0x01000000); + isa_mem_base = 0x80000000; + return; + } + + /* LongTrail */ + hose->ops = &gg2_pci_ops; + pci_dram_offset = 0; + isa_mem_base = 0xf7000000; + hose->io_base_phys = (unsigned long) 0xf8000000; + hose->io_base_virt = ioremap(hose->io_base_phys, 0x10000); + isa_io_base = (unsigned long) hose->io_base_virt; + ppc_md.pcibios_fixup = gg2_pcibios_fixup; + ppc_md.pcibios_fixup_bus = gg2_pcibios_fixup_bus; #endif /* CONFIG_POWER4 */ } #ifdef CONFIG_PPC64BRIDGE +#ifdef CONFIG_POWER4 /* * Hack alert!!! * 64-bit machines like POWER3 and POWER4 have > 32 bit @@ -490,9 +513,7 @@ chrp_setup_pci_ptrs(void) * page table gives us into parts of the physical address * space above 4GB so we can access the I/O devices. */ - -#ifdef CONFIG_POWER4 -static unsigned long pci_address_offset(int busnr, unsigned int flags) +unsigned long pci_address_offset(int busnr, unsigned int flags) { unsigned long offset = 0; @@ -508,34 +529,5 @@ static unsigned long pci_address_offset(int busnr, unsigned int flags) } return offset; } - -unsigned long phys_to_bus(unsigned long pa) -{ - if (pa >= 0xf8000000) - pa -= 0x38000000; - else if (pa >= 0x80000000 && pa < 0xc0000000) - pa += 0x40000000; - return pa; -} - -unsigned long bus_to_phys(unsigned int ba, int busnr) -{ - return ba + pci_address_offset(busnr, IORESOURCE_MEM); -} - -#else /* CONFIG_POWER4 */ -/* - * For now assume I/O addresses are < 4GB and PCI bridges don't - * remap addresses on POWER3 machines. - */ -unsigned long phys_to_bus(unsigned long pa) -{ - return pa; -} - -unsigned long bus_to_phys(unsigned int ba, int busnr) -{ - return ba; -} #endif /* CONFIG_POWER4 */ #endif /* CONFIG_PPC64BRIDGE */ diff --git a/arch/ppc/kernel/chrp_setup.c b/arch/ppc/kernel/chrp_setup.c index ccc6621de..5682f3fad 100644 --- a/arch/ppc/kernel/chrp_setup.c +++ b/arch/ppc/kernel/chrp_setup.c @@ -30,7 +30,6 @@ #include <linux/ioport.h> #include <linux/console.h> #include <linux/pci.h> -#include <linux/openpic.h> #include <linux/version.h> #include <linux/adb.h> #include <linux/module.h> @@ -50,21 +49,19 @@ #include <asm/hydra.h> #include <asm/keyboard.h> #include <asm/init.h> - #include <asm/time.h> + #include "local_irq.h" #include "i8259.h" #include "open_pic.h" #include "xics.h" -extern volatile unsigned char *chrp_int_ack_special; - unsigned long chrp_get_rtc_time(void); int chrp_set_rtc_time(unsigned long nowtime); void chrp_calibrate_decr(void); long chrp_time_init(void); -void chrp_setup_pci_ptrs(void); +void chrp_find_bridges(void); void chrp_event_scan(void); void rtas_display_progress(char *, unsigned short); void rtas_indicator_progress(char *, unsigned short); @@ -92,7 +89,7 @@ kdev_t boot_dev; extern PTE *Hash, *Hash_end; extern unsigned long Hash_size, Hash_mask; extern int probingmem; -extern unsigned long loops_per_sec; +extern unsigned long loops_per_jiffy; extern int bootx_text_mapped; static int max_width; @@ -244,7 +241,7 @@ chrp_setup_arch(void) struct device_node *device; /* init to some ~sane value until calibrate_delay() runs */ - loops_per_sec = 50000000; + loops_per_jiffy = 50000000/HZ; #ifdef CONFIG_BLK_DEV_INITRD /* this is fine for chrp */ @@ -257,6 +254,9 @@ chrp_setup_arch(void) ROOT_DEV = to_kdev_t(0x0802); /* sda2 (sda1 is for the kernel) */ printk("Boot arguments: %s\n", cmd_line); + /* Lookup PCI host bridges */ + chrp_find_bridges(); + #ifndef CONFIG_PPC64BRIDGE /* PCI bridge config space access area - * appears to be not in devtree on longtrail. */ @@ -266,11 +266,12 @@ chrp_setup_arch(void) * -- Geert */ hydra_init(); /* Mac I/O */ + #endif /* CONFIG_PPC64BRIDGE */ #ifndef CONFIG_POWER4 /* Some IBM machines don't have the hydra -- Cort */ - if ( !OpenPIC ) + if ( !OpenPIC_Addr ) { unsigned long *opprop; @@ -279,7 +280,7 @@ chrp_setup_arch(void) if (opprop != 0) { printk("OpenPIC addrs: %lx %lx %lx\n", opprop[0], opprop[1], opprop[2]); - OpenPIC = ioremap(opprop[0], sizeof(struct OpenPIC)); + OpenPIC_Addr = ioremap(opprop[0], 0x40000); } } #endif @@ -292,23 +293,17 @@ chrp_setup_arch(void) conswitchp = &dummy_con; #endif -#ifndef CONFIG_PPC64BRIDGE - pmac_find_bridges(); -#endif /* CONFIG_PPC64BRIDGE */ - /* Get the event scan rate for the rtas so we know how * often it expects a heartbeat. -- Cort */ - if ( rtas_data ) - { + if ( rtas_data ) { struct property *p; device = find_devices("rtas"); for ( p = device->properties; p && strncmp(p->name, "rtas-event-scan-rate", 20); p = p->next ) /* nothing */ ; - if ( p && *(unsigned long *)p->value ) - { + if ( p && *(unsigned long *)p->value ) { ppc_md.heartbeat = chrp_event_scan; ppc_md.heartbeat_reset = (HZ/(*(unsigned long *)p->value)*30)-1; ppc_md.heartbeat_count = 1; @@ -365,79 +360,44 @@ chrp_irq_cannonicalize(u_int irq) } } -int __chrp chrp_get_irq( struct pt_regs *regs ) -{ - int irq; - - irq = openpic_irq( smp_processor_id() ); - if (irq == IRQ_8259_CASCADE) - { - /* - * This magic address generates a PCI IACK cycle. - */ - if ( chrp_int_ack_special ) - irq = *chrp_int_ack_special; - else - irq = i8259_irq( smp_processor_id() ); - openpic_eoi( smp_processor_id() ); - } - if (irq == OPENPIC_VEC_SPURIOUS) - /* - * Spurious interrupts should never be - * acknowledged - */ - irq = -1; - /* - * I would like to openpic_eoi here but there seem to be timing problems - * between the openpic ack and the openpic eoi. - * -- Cort - */ - return irq; -} - -void __chrp chrp_post_irq(struct pt_regs* regs, int irq) -{ - /* - * If it's an i8259 irq then we've already done the - * openpic irq. So we just check to make sure the controller - * is an openpic and if it is then eoi - * - * We do it this way since our irq_desc[irq].handler can change - * with RTL and no longer be open_pic -- Cort - */ - if ( irq >= open_pic_irq_offset) - openpic_eoi( smp_processor_id() ); -} - void __init chrp_init_IRQ(void) { struct device_node *np; int i; unsigned long *addrp; + unsigned char* chrp_int_ack_special = 0; + unsigned char init_senses[NR_IRQS - NUM_8259_INTERRUPTS]; + int nmi_irq = -1; +#if defined(CONFIG_VT) && defined(CONFIG_ADB_KEYBOARD) && defined(XMON) + struct device_node *kbd; +#endif if (!(np = find_devices("pci")) || !(addrp = (unsigned long *) get_property(np, "8259-interrupt-acknowledge", NULL))) printk("Cannot find pci to get ack address\n"); else - chrp_int_ack_special = (volatile unsigned char *) - ioremap(*addrp, 1); - open_pic_irq_offset = 16; - for ( i = 16 ; i < NR_IRQS ; i++ ) - irq_desc[i].handler = &open_pic; - openpic_init(1); - enable_irq(IRQ_8259_CASCADE); - for ( i = 0 ; i < 16 ; i++ ) + chrp_int_ack_special = (unsigned char *)ioremap(*addrp, 1); + /* hydra still sets OpenPIC_InitSenses to a static set of values */ + if (OpenPIC_InitSenses == NULL) { + prom_get_irq_senses(init_senses, NUM_8259_INTERRUPTS, NR_IRQS); + OpenPIC_InitSenses = init_senses; + OpenPIC_NumInitSenses = NR_IRQS - NUM_8259_INTERRUPTS; + } + openpic_init(1, NUM_8259_INTERRUPTS, chrp_int_ack_special, nmi_irq); + for ( i = 0 ; i < NUM_8259_INTERRUPTS ; i++ ) irq_desc[i].handler = &i8259_pic; i8259_init(); -#ifdef CONFIG_XMON - request_irq(openpic_to_irq(HYDRA_INT_ADB_NMI), - xmon_irq, 0, "NMI", 0); -#endif /* CONFIG_XMON */ -#ifdef CONFIG_SMP - request_irq(openpic_to_irq(OPENPIC_VEC_IPI), - openpic_ipi_action, 0, "IPI0", 0); -#endif /* CONFIG_SMP */ +#if defined(CONFIG_VT) && defined(CONFIG_ADB_KEYBOARD) && defined(XMON) + /* see if there is a keyboard in the device tree + with a parent of type "adb" */ + for (kbd = find_devices("keyboard"); kbd; kbd = kbd->next) + if (kbd->parent && kbd->parent->type + && strcmp(kbd->parent->type, "adb") == 0) + break; + if (kbd) + request_irq( HYDRA_INT_ADB_NMI, xmon_irq, 0, "XMON break", 0); +#endif } void __init @@ -556,12 +516,6 @@ chrp_ide_release_region(ide_ioreg_t from, } void __chrp -chrp_ide_fix_driveid(struct hd_driveid *id) -{ - ppc_generic_ide_fix_driveid(id); -} - -void __chrp chrp_ide_init_hwif_ports(hw_regs_t *hw, ide_ioreg_t data_port, ide_ioreg_t ctrl_port, int *irq) { ide_ioreg_t reg = data_port; @@ -586,7 +540,6 @@ void __init chrp_init(unsigned long r3, unsigned long r4, unsigned long r5, unsigned long r6, unsigned long r7) { - chrp_setup_pci_ptrs(); #ifdef CONFIG_BLK_DEV_INITRD /* take care of initrd if we have one */ if ( r6 ) @@ -596,10 +549,10 @@ void __init } #endif /* CONFIG_BLK_DEV_INITRD */ - /* pci_dram_offset/isa_io_base/isa_mem_base set by setup_pci_ptrs() */ ISA_DMA_THRESHOLD = ~0L; DMA_MODE_READ = 0x44; DMA_MODE_WRITE = 0x48; + isa_io_base = CHRP_ISA_IO_BASE; /* default value */ ppc_md.setup_arch = chrp_setup_arch; ppc_md.setup_residual = NULL; @@ -607,8 +560,8 @@ void __init ppc_md.irq_cannonicalize = chrp_irq_cannonicalize; #ifndef CONFIG_POWER4 ppc_md.init_IRQ = chrp_init_IRQ; - ppc_md.get_irq = chrp_get_irq; - ppc_md.post_irq = chrp_post_irq; + ppc_md.get_irq = openpic_get_irq; + ppc_md.post_irq = NULL; #else ppc_md.init_IRQ = xics_init_IRQ; ppc_md.get_irq = xics_get_irq; @@ -669,11 +622,12 @@ void __init ppc_ide_md.ide_check_region = chrp_ide_check_region; ppc_ide_md.ide_request_region = chrp_ide_request_region; ppc_ide_md.ide_release_region = chrp_ide_release_region; - ppc_ide_md.fix_driveid = chrp_ide_fix_driveid; + ppc_ide_md.fix_driveid = ppc_generic_ide_fix_driveid; ppc_ide_md.ide_init_hwif = chrp_ide_init_hwif_ports; ppc_ide_md.io_base = _IO_BASE; #endif + /* * Print the banner, then scroll down so boot progress * can be printed. -- Cort diff --git a/arch/ppc/kernel/entry.S b/arch/ppc/kernel/entry.S index 89922aaab..93bb8ca00 100644 --- a/arch/ppc/kernel/entry.S +++ b/arch/ppc/kernel/entry.S @@ -234,7 +234,6 @@ _GLOBAL(_switch) li r0,0x0ff0 stw r0,TRAP(r1) stw r1,KSP(r3) /* Set old stack pointer */ - sync tophys(r0,r4) CLR_TOP32(r0) mtspr SPRG3,r0 /* Update current THREAD phys addr */ @@ -251,7 +250,7 @@ _GLOBAL(_switch) #endif mtspr M_TWB,r9 /* Update MMU base address */ tlbia - SYNC + sync #endif /* CONFIG_8xx */ lwz r1,KSP(r4) /* Load new stack pointer */ /* save the old current 'last' for return value */ @@ -342,23 +341,23 @@ lost_irq_ret: do_bottom_half_ret: 2: lwz r3,_MSR(r1) /* Returning to user mode? */ andi. r3,r3,MSR_PR - beq+ restore /* if so, check need_resched and signals */ - .globl ret_to_user_hook -ret_to_user_hook: - nop + beq+ do_signal_ret /* if so, check need_resched and signals */ lwz r3,NEED_RESCHED(r2) cmpi 0,r3,0 /* check need_resched flag */ beq+ 7f bl schedule 7: lwz r5,SIGPENDING(r2) /* Check for pending unblocked signals */ cmpwi 0,r5,0 - beq+ restore + beq+ do_signal_ret li r3,0 addi r4,r1,STACK_FRAME_OVERHEAD MOL_HOOK_MMU(8,r8) bl do_signal .globl do_signal_ret do_signal_ret: + .globl ret_to_user_hook +ret_to_user_hook: + nop restore: lwz r3,_XER(r1) mtspr XER,r3 @@ -372,7 +371,7 @@ restore: */ mfmsr r0 /* Get current interrupt state */ rlwinm r0,r0,0,17,15 /* clear MSR_EE in r0 */ - sync /* Some chip revs have problems here... */ + SYNC /* Some chip revs have problems here... */ mtmsr r0 /* Update machine state */ /* if returning to user mode, set new sprg2 and save kernel SP */ @@ -468,7 +467,7 @@ enter_rtas: andc r0,r9,r0 li r10,MSR_IR|MSR_DR|MSR_FE0|MSR_FE1|MSR_FP andc r9,r0,r10 - sync /* disable interrupts so SRR0/1 */ + SYNC /* disable interrupts so SRR0/1 */ mtmsr r0 /* don't get trashed */ mtlr r6 CLR_TOP32(r7) diff --git a/arch/ppc/kernel/error_log.c b/arch/ppc/kernel/error_log.c new file mode 100644 index 000000000..5ee750dda --- /dev/null +++ b/arch/ppc/kernel/error_log.c @@ -0,0 +1,183 @@ +/* + * arch/ppc/kernel/error_log.c + * + * Copyright (c) 2000 Tilmann Bitterberg + * (tilmann@bitterberg.de) + * + * Error processing of errors found by rtas even-scan routine + * which is done with every heartbeat. (chrp_setup.c) + */ + +#include <linux/sched.h> + +#include <asm/prom.h> + +#include "error_log.h" + +/* ****************************************************************** */ +/* + * EVENT-SCAN + * The whole stuff below here doesn't take any action when it found + * an error, it just prints as much information as possible and + * then its up to the user to decide what to do. + * + * Returns 0 if no errors were found + * Returns 1 if there may be more errors + */ +int ppc_rtas_errorlog_scan(void) +{ +const char *_errlog_severity[] = { +#ifdef VERBOSE_ERRORS + "No Error\n\t\ +Should require no further information", + "Event\n\t\ +This is not really an error, it is an event. I use events\n\t\ +to communicate with RTAS back and forth.", + "Warning\n\t\ +Indicates a non-state-losing error, either fully recovered\n\t\ +by RTAS or not needing recovery. Ignore it.", + "Error sync\n\t\ +May only be fatal to a certain program or thread. Recovery\n\t\ +and continuation is possible, if I only had a handler for\n\t\ +this. Less serious", + "Error\n\t\ +Less serious, but still causing a loss of data and state.\n\t\ +I can't tell you exactly what to do, You have to decide\n\t\ +with help from the target and initiator field, what kind\n\t\ +of further actions may take place.", + "Fatal\n\t\ +Represent a permanent hardware failure and I believe this\n\t\ +affects my overall performance and behaviour. I would not\n\t\ +attempt to continue normal operation." +#else + "No Error", + "Event", + "Warning", + "Error sync", + "Error", + "Fatal" +#endif /* VERBOSE_ERRORS */ +}; + +#if 0 /* unused?? */ +const char *_errlog_disposition[] = { +#ifdef VERBOSE_ERRORS + "Fully recovered\n\t\ +There was an error, but it is fully recovered by RTAS.", + "Limited recovery\n\t\ +RTAS was able to recover the state of the machine, but some\n\t\ +feature of the machine has been disabled or lost (for example\n\t\ +error checking) or performance may suffer.", + "Not recovered\n\t\ +Whether RTAS did not try to recover anything or recovery failed:\n\t\ +HOUSTON, WE HAVE A PROBLEM!" +#else + "Fully recovered", + "Limited recovery", + "Not recovered" +#endif /* VERBOSE_ERRORS */ +}; +#endif + +const char *_errlog_extended[] = { +#ifdef VERBOSE_ERRORS + "Not present\n\t\ +Sad, the RTAS call didn't return an extended error log.", + "Present\n\t\ +The extended log is present and hopefully it contains a lot of\n\t\ +useful information, which leads to the solution of the problem." +#else + "Not present", + "Present" +#endif /* VERBOSE_ERRORS */ +}; + +const char *_errlog_initiator[] = { + "Unknown or not applicable", + "CPU", + "PCI", + "ISA", + "Memory", + "Power management" +}; + +const char *_errlog_target[] = { + "Unknown or not applicable", + "CPU", + "PCI", + "ISA", + "Memory", + "Power management" +}; + rtas_error_log error_log; + char logdata[1024]; + int error; +#if 0 /* unused?? */ + int retries = 0; /* if HW error, try 10 times */ +#endif + + error = call_rtas ("event-scan", 4, 1, (unsigned long *)&error_log, + INTERNAL_ERROR | EPOW_WARNING, + 0, __pa(logdata), 1024); + + if (error == 1) /* no errors found */ + return 0; + + if (error == -1) { + printk(KERN_ERR "Unable to get errors. Do you a favor and throw this box away\n"); + return 0; + } + if (error_log.version != 1) + printk(KERN_WARNING "Unknown version (%d), please implement me\n", + error_log.version); + + switch (error_log.disposition) { + case DISP_FULLY_RECOVERED: + /* there was an error, but everything is fine now */ + return 0; + case DISP_NOT_RECOVERED: + printk("We have a really serious Problem!\n"); + case DISP_LIMITED_RECOVERY: + printk("Error classification\n"); + printk("Severity : %s\n", + ppc_rtas_errorlog_check_severity (error_log)); + printk("Initiator : %s\n", + ppc_rtas_errorlog_check_initiator (error_log)); + printk("Target : %s\n", + ppc_rtas_errorlog_check_target (error_log)); + printk("Type : %s\n", + ppc_rtas_errorlog_check_type (error_log)); + printk("Ext. log : %s\n", + ppc_rtas_errorlog_check_extended (error_log)); + if (error_log.extended) + ppc_rtas_errorlog_disect_extended (logdata); + return 1; + default: + /* nothing */ + break; + } + return 0; +} +/* ****************************************************************** */ +const char * ppc_rtas_errorlog_check_type (rtas_error_log error_log) +{ + const char *_errlog_type[] = { + "unknown type", + "too many tries failed", + "TCE error", + "RTAS device failed", + "target timed out", + "parity error on data", /* 5 */ + "parity error on address", + "parity error on external cache", + "access to invalid address", + "uncorrectable ECC error", + "corrected ECC error" /* 10 */ + }; + if (error_log.type == TYPE_EPOW) + return "EPOW"; + if (error_log.type >= TYPE_PMGM_POWER_SW_ON) + return "PowerMGM Event (not handled right now)"; + return _errlog_type[error_log.type]; +} + diff --git a/arch/ppc/kernel/error_log.h b/arch/ppc/kernel/error_log.h new file mode 100644 index 000000000..579fc3e0d --- /dev/null +++ b/arch/ppc/kernel/error_log.h @@ -0,0 +1,95 @@ +#ifndef __ERROR_LOG_H__ +#define __ERROR_LOG_H__ + +#define VERBOSE_ERRORS 1 /* Maybe I enlarge the kernel too much */ +#undef VERBOSE_ERRORS + +/* Event classes */ +/* XXX: Endianess correct? NOW*/ +#define INTERNAL_ERROR 0x80000000 /* set bit 0 */ +#define EPOW_WARNING 0x40000000 /* set bit 1 */ +#define POWERMGM_EVENTS 0x20000000 /* set bit 2 */ + +/* event-scan returns */ +#define SEVERITY_FATAL 0x5 +#define SEVERITY_ERROR 0x4 +#define SEVERITY_ERROR_SYNC 0x3 +#define SEVERITY_WARNING 0x2 +#define SEVERITY_EVENT 0x1 +#define SEVERITY_NO_ERROR 0x0 +#define DISP_FULLY_RECOVERED 0x0 +#define DISP_LIMITED_RECOVERY 0x1 +#define DISP_NOT_RECOVERED 0x2 +#define PART_PRESENT 0x0 +#define PART_NOT_PRESENT 0x1 +#define INITIATOR_UNKNOWN 0x0 +#define INITIATOR_CPU 0x1 +#define INITIATOR_PCI 0x2 +#define INITIATOR_ISA 0x3 +#define INITIATOR_MEMORY 0x4 +#define INITIATOR_POWERMGM 0x5 +#define TARGET_UNKNOWN 0x0 +#define TARGET_CPU 0x1 +#define TARGET_PCI 0x2 +#define TARGET_ISA 0x3 +#define TARGET_MEMORY 0x4 +#define TARGET_POWERMGM 0x5 +#define TYPE_RETRY 0x01 +#define TYPE_TCE_ERR 0x02 +#define TYPE_INTERN_DEV_FAIL 0x03 +#define TYPE_TIMEOUT 0x04 +#define TYPE_DATA_PARITY 0x05 +#define TYPE_ADDR_PARITY 0x06 +#define TYPE_CACHE_PARITY 0x07 +#define TYPE_ADDR_INVALID 0x08 +#define TYPE_ECC_UNCORR 0x09 +#define TYPE_ECC_CORR 0x0a +#define TYPE_EPOW 0x40 +/* I don't add PowerMGM events right now, this is a different topic */ +#define TYPE_PMGM_POWER_SW_ON 0x60 +#define TYPE_PMGM_POWER_SW_OFF 0x61 +#define TYPE_PMGM_LID_OPEN 0x62 +#define TYPE_PMGM_LID_CLOSE 0x63 +#define TYPE_PMGM_SLEEP_BTN 0x64 +#define TYPE_PMGM_WAKE_BTN 0x65 +#define TYPE_PMGM_BATTERY_WARN 0x66 +#define TYPE_PMGM_BATTERY_CRIT 0x67 +#define TYPE_PMGM_SWITCH_TO_BAT 0x68 +#define TYPE_PMGM_SWITCH_TO_AC 0x69 +#define TYPE_PMGM_KBD_OR_MOUSE 0x6a +#define TYPE_PMGM_ENCLOS_OPEN 0x6b +#define TYPE_PMGM_ENCLOS_CLOSED 0x6c +#define TYPE_PMGM_RING_INDICATE 0x6d +#define TYPE_PMGM_LAN_ATTENTION 0x6e +#define TYPE_PMGM_TIME_ALARM 0x6f +#define TYPE_PMGM_CONFIG_CHANGE 0x70 +#define TYPE_PMGM_SERVICE_PROC 0x71 + +typedef struct _rtas_error_log { + unsigned long version:8; /* Architectural version */ + unsigned long severity:3; /* Severity level of error */ + unsigned long disposition:2; /* Degree of recovery */ + unsigned long extended:1; /* extended log present? */ + unsigned long /* reserved */ :2; /* Reserved for future use */ + unsigned long initiator:4; /* Initiator of event */ + unsigned long target:4; /* Target of failed operation */ + unsigned long type:8; /* General event or error*/ + unsigned long extended_log_length:32; /* length in bytes */ +} rtas_error_log; + +/* ****************************************************************** */ +#define ppc_rtas_errorlog_check_severity(x) \ + (_errlog_severity[x.severity]) +#define ppc_rtas_errorlog_check_target(x) \ + (_errlog_target[x.target]) +#define ppc_rtas_errorlog_check_initiator(x) \ + (_errlog_initiator[x.initiator]) +#define ppc_rtas_errorlog_check_extended(x) \ + (_errlog_extended[x.extended]) +#define ppc_rtas_errorlog_disect_extended(x) \ + do { /* implement me */ } while(0) +extern const char * ppc_rtas_errorlog_check_type (rtas_error_log error_log); +extern int ppc_rtas_errorlog_scan(void); + + +#endif /* __ERROR_LOG_H__ */ diff --git a/arch/ppc/kernel/feature.c b/arch/ppc/kernel/feature.c index 57599917a..f22f4b163 100644 --- a/arch/ppc/kernel/feature.c +++ b/arch/ppc/kernel/feature.c @@ -238,12 +238,15 @@ feature_add_controller(struct device_node *controller_device, fbit* bits); static struct feature_controller* feature_lookup_controller(struct device_node *device); +#ifdef CONFIG_PMAC_PBOOK static void heathrow_prepare_for_sleep(struct feature_controller* ctrler); static void heathrow_wakeup(struct feature_controller* ctrler); -static void keylargo_init(void); -static void uninorth_init(void); static void core99_prepare_for_sleep(struct feature_controller* ctrler); static void core99_wake_up(struct feature_controller* ctrler); +#endif /* CONFIG_PMAC_PBOOK */ + +static void keylargo_init(void); +static void uninorth_init(void); /* static variables */ static struct feature_controller controllers[MAX_FEATURE_CONTROLLERS]; @@ -255,6 +258,10 @@ static volatile u32* keylargo_base = NULL; static int uninorth_rev; static int keylargo_rev; +/* + * WARNING ! This function is called early in setup_arch, neither the IO base + * nor the udelay calibration have been done yet + */ void feature_init(void) { @@ -527,14 +534,31 @@ feature_set_usb_power(struct device_node* device, int power) void feature_set_firewire_power(struct device_node* device, int power) { + if (!uninorth_base) + return; + if (power) + UN_BIS(UNI_N_CLOCK_CNTL, UNI_N_CLOCK_CNTL_FW); + else + UN_BIC(UNI_N_CLOCK_CNTL, UNI_N_CLOCK_CNTL_FW); + udelay(20); +} + +#ifdef CONFIG_SMP +void +feature_core99_kick_cpu1(void) +{ + out_8((volatile u8 *)KL_FCR(KL_GPIO_KICK_CPU1), KL_GPIO_KICK_CPU1_UP); + udelay(1); + out_8((volatile u8 *)KL_FCR(KL_GPIO_KICK_CPU1), KL_GPIO_KICK_CPU1_DOWN); } +#endif /* CONFIG_SMP */ /* Initialize the Core99 UniNorth host bridge and memory controller */ static void uninorth_init(void) { - struct device_node* gmac; + struct device_node* gmac, *fw; unsigned long actrl; /* Set the arbitrer QAck delay according to what Apple does @@ -564,6 +588,11 @@ uninorth_init(void) } if (gmac) feature_set_gmac_power(gmac, 0); + + /* Kludge (enable FW before PCI probe) */ + fw = find_devices("firewire"); + if (fw && device_is_compatible(fw, "pci106b,18")) + feature_set_firewire_power(fw, 1); } /* Initialize the Core99 KeyLargo ASIC. Currently, we just make sure @@ -576,6 +605,43 @@ keylargo_init(void) } #ifdef CONFIG_PMAC_PBOOK + +static u32 save_fcr[5]; +static u32 save_mbcr; + +static void +heathrow_prepare_for_sleep(struct feature_controller* ctrler) +{ + save_mbcr = in_le32(FREG(ctrler, 0x34)); + save_fcr[0] = in_le32(FREG(ctrler, 0x38)); + save_fcr[1] = in_le32(FREG(ctrler, 0x3c)); + + out_le32(FREG(ctrler, 0x38), save_fcr[0] & ~HRW_IOBUS_ENABLE); +} + +static void +heathrow_wakeup(struct feature_controller* ctrler) +{ + out_le32(FREG(ctrler, 0x38), save_fcr[0]); + out_le32(FREG(ctrler, 0x3c), save_fcr[1]); + out_le32(FREG(ctrler, 0x34), save_mbcr); + mdelay(1); + out_le32(FREG(ctrler, 0x38), save_fcr[0] | HRW_IOBUS_ENABLE); + mdelay(1); +} + +static void +core99_prepare_for_sleep(struct feature_controller* ctrler) +{ + /* Not yet implemented */ +} + +static void +core99_wake_up(struct feature_controller* ctrler) +{ + /* Not yet implemented */ +} + void feature_prepare_for_sleep(void) { @@ -599,7 +665,6 @@ feature_prepare_for_sleep(void) } } - void feature_wake_up(void) { @@ -622,39 +687,4 @@ feature_wake_up(void) } } -static u32 save_fcr[5]; -static u32 save_mbcr; - -static void -heathrow_prepare_for_sleep(struct feature_controller* ctrler) -{ - save_mbcr = in_le32(FREG(ctrler, 0x34)); - save_fcr[0] = in_le32(FREG(ctrler, 0x38)); - save_fcr[1] = in_le32(FREG(ctrler, 0x3c)); - - out_le32(FREG(ctrler, 0x38), save_fcr[0] & ~HRW_IOBUS_ENABLE); -} - -static void -heathrow_wakeup(struct feature_controller* ctrler) -{ - out_le32(FREG(ctrler, 0x38), save_fcr[0]); - out_le32(FREG(ctrler, 0x3c), save_fcr[1]); - out_le32(FREG(ctrler, 0x34), save_mbcr); - mdelay(1); - out_le32(FREG(ctrler, 0x38), save_fcr[0] | HRW_IOBUS_ENABLE); - mdelay(1); -} - -static void -core99_prepare_for_sleep(struct feature_controller* ctrler) -{ - /* Not yet implemented */ -} - -static void -core99_wake_up(struct feature_controller* ctrler) -{ - /* Not yet implemented */ -} #endif /* CONFIG_PMAC_PBOOK */ diff --git a/arch/ppc/kernel/galaxy_pci.c b/arch/ppc/kernel/galaxy_pci.c index aeddd9a0e..36e7e14fb 100644 --- a/arch/ppc/kernel/galaxy_pci.c +++ b/arch/ppc/kernel/galaxy_pci.c @@ -36,67 +36,84 @@ /* Function Prototypes */ -decl_config_access_method(galaxy); - - void __init galaxy_pcibios_fixup(void) { } -void __init -galaxy_setup_pci_ptrs(void) +static int +galaxy_pcibios_read_config_byte(struct pci_controller* hose, + u8 bus, u8 dev, u8 offset, u8 *val) { - set_config_access_method(galaxy); - ppc_md.pcibios_fixup = galaxy_pcibios_fixup; + return (PCIBIOS_SUCCESSFUL); } -int -galaxy_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char *val) +static int +galaxy_pcibios_read_config_word(struct pci_controller* hose, + u8 bus, u8 dev, u8 offset, u16 *val) { return (PCIBIOS_SUCCESSFUL); } -int -galaxy_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short *val) +static int +galaxy_pcibios_read_config_dword(struct pci_controller* hose, + u8 bus, u8 dev, u8 offset, u32 *val) { return (PCIBIOS_SUCCESSFUL); } -int -galaxy_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int *val) +static int +galaxy_pcibios_write_config_byte(struct pci_controller* hose, + u8 bus, u8 dev, u8 offset, u8 val) { return (PCIBIOS_SUCCESSFUL); } -int -galaxy_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char val) +static int +galaxy_pcibios_write_config_word(struct pci_controller* hose, + u8 bus, u8 dev, u8 offset, u16 val) { return (PCIBIOS_SUCCESSFUL); } -int -galaxy_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short val) +static int +galaxy_pcibios_write_config_dword(struct pci_controller* hose, + u8 bus, u8 dev, u8 offset, u32 val) { return (PCIBIOS_SUCCESSFUL); } -int -galaxy_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int val) +static struct pci_controller_ops galaxy_pci_ops = { + galaxy_pcibios_read_config_byte, + galaxy_pcibios_read_config_word, + galaxy_pcibios_read_config_dword, + galaxy_pcibios_write_config_byte, + galaxy_pcibios_write_config_word, + galaxy_pcibios_write_config_dword +}; - return (PCIBIOS_SUCCESSFUL); +void __init +galaxy_find_bridges(void) +{ + struct pci_controller* hose; + + set_config_access_method(galaxy); + + ppc_md.pcibios_fixup = galaxy_pcibios_fixup; + hose = pcibios_alloc_controller(); + if (!hose) + return; + hose->ops = &galaxy_pci_ops; + /* Todo ... + hose->cfg_data = ioremap(PCICFGDATA, ...); + hose->cfg_addr = ioremap(PCICFGADDR, ...); + */ } diff --git a/arch/ppc/kernel/hashtable.S b/arch/ppc/kernel/hashtable.S index 06b7c6f1d..8ea3cdc7a 100644 --- a/arch/ppc/kernel/hashtable.S +++ b/arch/ppc/kernel/hashtable.S @@ -56,7 +56,6 @@ hash_page: #ifdef CONFIG_PPC64BRIDGE mfmsr r0 clrldi r0,r0,1 /* make sure it's in 32-bit mode */ - sync MTMSRD(r0) isync #endif @@ -112,23 +111,31 @@ hash_page: #endif tophys(r2,r5) rlwimi r2,r3,22,20,29 /* insert next 10 bits of address */ - lwz r6,0(r2) /* get linux-style pte */ ori r4,r4,1 /* set _PAGE_PRESENT bit in access */ + rlwinm r5,r4,5,24,24 /* _PAGE_RW access -> _PAGE_DIRTY */ + rlwimi r5,r4,7,22,22 /* _PAGE_RW -> _PAGE_HWWRITE */ + ori r5,r5,0x100 /* set _PAGE_ACCESSED */ +retry: + lwz r6,0(r2) /* get linux-style pte */ andc. r0,r4,r6 /* check access & ~permission */ #ifdef CONFIG_SMP bne- hash_page_out /* return if access not permitted */ #else bnelr- #endif + andc. r0,r5,r6 /* any bits not yet set? */ + beq 2f - ori r6,r6,0x100 /* set _PAGE_ACCESSED in pte */ - rlwinm r5,r4,5,24,24 /* _PAGE_RW access -> _PAGE_DIRTY */ - rlwimi r5,r4,7,22,22 /* _PAGE_RW -> _PAGE_HWWRITE */ - or r6,r6,r5 - stw r6,0(r2) /* update PTE (accessed/dirty bits) */ + /* Update the linux PTE atomically */ + lwarx r0,0,r2 /* refetch the pte and check */ + cmpw 0,r0,r6 /* that it hasn't been changed */ + bne- retry /* retry if it has */ + or r6,r6,r5 /* set accessed/dirty bits */ + stwcx. r6,0,r2 /* attempt to update PTE */ + bne- retry /* retry if someone got there first */ /* Convert linux-style PTE to low word of PPC-style PTE */ - rlwinm r4,r6,32-9,31,31 /* _PAGE_HWWRITE -> PP lsb */ +2: rlwinm r4,r6,32-9,31,31 /* _PAGE_HWWRITE -> PP lsb */ rlwimi r6,r6,32-1,31,31 /* _PAGE_USER -> PP (both bits now) */ ori r4,r4,0xe04 /* clear out reserved bits */ andc r6,r6,r4 /* PP=2 or 0, when _PAGE_HWWRITE */ @@ -166,12 +173,16 @@ hash_page_patch_A: rlwimi r4,r5,32-5,25-Hash_bits,24 /* (VSID & hash_mask) << 7 */ rlwinm r0,r3,32-5,25-Hash_bits,24 /* (PI & hash_mask) << 7 */ xor r4,r4,r0 /* make primary hash */ + li r2,8 /* PTEs/group */ +#ifndef CONFIG_SMP + /* We don't do this for SMP - another cpu could have put in + the appropriate PTE since we took the exception. -- paulus. */ /* See whether it was a PTE not found exception or a protection violation. */ andis. r0,r20,0x4000 - li r2,8 /* PTEs/group */ bne 10f /* no PTE: go look for an empty slot */ +#endif /* CONFIG_SMP */ tlbie r3 /* invalidate TLB entry */ /* Search the primary PTEG for a PTE whose 1st dword matches r5 */ @@ -263,7 +274,6 @@ found_empty: std r5,0(r3) found_slot: std r6,8(r3) - sync #else /* CONFIG_SMP */ /* @@ -311,12 +321,16 @@ hash_page_patch_A: rlwimi r4,r5,32-1,26-Hash_bits,25 /* (VSID & hash_mask) << 6 */ rlwinm r0,r3,32-6,26-Hash_bits,25 /* (PI & hash_mask) << 6 */ xor r4,r4,r0 /* make primary hash */ + li r2,8 /* PTEs/group */ +#ifndef CONFIG_SMP + /* We don't do this for SMP - another cpu could have put in + the appropriate PTE since we took the exception. -- paulus. */ /* See whether it was a PTE not found exception or a protection violation. */ andis. r0,r20,0x4000 - li r2,8 /* PTEs/group */ bne 10f /* no PTE: go look for an empty slot */ +#endif /* CONFIG_SMP */ tlbie r3 /* invalidate TLB entry */ /* Search the primary PTEG for a PTE whose 1st word matches r5 */ @@ -394,7 +408,6 @@ found_empty: stw r5,0(r3) found_slot: stw r6,4(r3) - sync #else /* CONFIG_SMP */ /* @@ -428,6 +441,8 @@ found_slot: #endif /* CONFIG_SMP */ #endif /* CONFIG_PPC64BRIDGE */ + sync /* make sure pte updates get to memory */ + /* * Update the hash table miss count. We only want misses here * that _are_ valid addresses and have a pte otherwise we don't @@ -517,7 +532,7 @@ _GLOBAL(flush_hash_segments) a hash table miss while we have the hash table locked, or we'll get a deadlock. -paulus */ mfmsr r10 - sync + SYNC rlwinm r0,r10,0,17,15 /* clear bit 16 (MSR_EE) */ mtmsr r0 SYNC @@ -616,7 +631,7 @@ _GLOBAL(flush_hash_page) a hash table miss while we have the hash table locked, or we'll get a deadlock. -paulus */ mfmsr r10 - sync + SYNC rlwinm r0,r10,0,17,15 /* clear bit 16 (MSR_EE) */ mtmsr r0 SYNC @@ -628,10 +643,14 @@ _GLOBAL(flush_hash_page) oris r8,r8,9 10: lwarx r7,0,r9 cmpi 0,r7,0 - bne- 10b + bne- 11f stwcx. r8,0,r9 - bne- 10b - eieio + beq+ 12f +11: lwz r7,0(r9) + cmpi 0,r7,0 + beq 10b + b 11b +12: eieio #endif #ifndef CONFIG_PPC64BRIDGE rlwinm r3,r3,11,1,20 /* put context into vsid */ diff --git a/arch/ppc/kernel/head.S b/arch/ppc/kernel/head.S index 71b5a96c5..f296d1606 100644 --- a/arch/ppc/kernel/head.S +++ b/arch/ppc/kernel/head.S @@ -149,6 +149,8 @@ __start: mr r28,r6 mr r27,r7 li r24,0 /* cpu # */ + /* N.B. prom_init clears the BSS even if it doesn't do + * anything else -- paulus. */ bl prom_init #ifdef CONFIG_APUS @@ -159,7 +161,6 @@ __start: bl fix_mem_constants #endif /* CONFIG_APUS */ -#ifndef CONFIG_GEMINI /* Switch MMU off, clear BATs and flush TLB. At this point, r3 contains * the physical address we are running at, returned by prom_init() */ @@ -167,7 +168,6 @@ __start: __after_mmu_off: bl clear_bats bl flush_tlbs -#endif #ifndef CONFIG_POWER4 /* POWER4 doesn't have BATs */ @@ -290,6 +290,7 @@ label: \ addi r3,r1,STACK_FRAME_OVERHEAD; \ li r20,MSR_KERNEL; \ bl transfer_to_handler; \ +i##n: \ .long hdlr; \ .long ret_from_except @@ -301,17 +302,13 @@ label: \ addi r3,r1,STACK_FRAME_OVERHEAD; \ li r20,MSR_KERNEL; \ bl transfer_to_handler; \ +i##n: \ .long hdlr; \ .long ret_from_except /* System reset */ -#ifdef CONFIG_SMP /* MVME/MTX and gemini start the secondary here */ -#ifdef CONFIG_GEMINI - . = 0x100 - b __secondary_start_gemini -#else /* CONFIG_GEMINI */ +#ifdef CONFIG_SMP /* MVME/MTX start the secondary here */ STD_EXCEPTION(0x100, Reset, __secondary_start_psurge) -#endif /* CONFIG_GEMINI */ #else STD_EXCEPTION(0x100, Reset, UnknownException) #endif @@ -344,6 +341,7 @@ DataAccess: li r20,MSR_KERNEL rlwimi r20,r23,0,16,16 /* copy EE bit from saved MSR */ bl transfer_to_handler +i0x300: .long do_page_fault .long ret_from_except @@ -384,6 +382,7 @@ InstructionAccess: li r20,MSR_KERNEL rlwimi r20,r23,0,16,16 /* copy EE bit from saved MSR */ bl transfer_to_handler +i0x400: .long do_page_fault .long ret_from_except @@ -429,6 +428,7 @@ Alignment: li r20,MSR_KERNEL rlwimi r20,r23,0,16,16 /* copy EE bit from saved MSR */ bl transfer_to_handler +i0x600: .long AlignmentException .long ret_from_except @@ -441,6 +441,7 @@ ProgramCheck: li r20,MSR_KERNEL rlwimi r20,r23,0,16,16 /* copy EE bit from saved MSR */ bl transfer_to_handler +i0x700: .long ProgramCheckException .long ret_from_except @@ -452,6 +453,7 @@ FPUnavailable: bne load_up_fpu /* if from user, just load it up */ li r20,MSR_KERNEL bl transfer_to_handler /* if from kernel, take a trap */ +i0x800: .long KernelFP .long ret_from_except @@ -575,7 +577,7 @@ InstructionAddressInvalid: mfmsr r0 /* Restore "normal" registers */ xoris r0,r0,MSR_TGPR>>16 mtcrf 0x80,r3 /* Restore CR0 */ - sync /* Some chip revs have problems here... */ + SYNC /* Some chip revs have problems here... */ mtmsr r0 b InstructionAccess @@ -646,7 +648,7 @@ DataAddressInvalid: mfmsr r0 /* Restore "normal" registers */ xoris r0,r0,MSR_TGPR>>16 mtcrf 0x80,r3 /* Restore CR0 */ - sync /* Some chip revs have problems here... */ + SYNC /* Some chip revs have problems here... */ mtmsr r0 b DataAccess @@ -843,7 +845,7 @@ load_up_fpu: #endif /* CONFIG_PPC64BRIDGE */ SYNC MTMSRD(r5) /* enable use of fpu now */ - SYNC + isync /* * For SMP, we don't do lazy FPU switching because it just gets too * horrendously complex, especially when a task switches from one CPU @@ -929,7 +931,7 @@ load_up_altivec: oris r5,r5,MSR_VEC@h SYNC mtmsr r5 /* enable use of AltiVec now */ - SYNC + isync /* * For SMP, we don't do lazy AltiVec switching because it just gets too * horrendously complex, especially when a task switches from one CPU @@ -1023,7 +1025,7 @@ giveup_altivec: oris r5,r5,MSR_VEC@h SYNC mtmsr r5 /* enable use of AltiVec now */ - SYNC + isync cmpi 0,r3,0 beqlr- /* if no previous owner, done */ addi r3,r3,THREAD /* want THREAD of task */ @@ -1064,7 +1066,7 @@ giveup_fpu: ori r5,r5,MSR_FP SYNC mtmsr r5 /* enable use of fpu now */ - SYNC + isync cmpi 0,r3,0 beqlr- /* if no previous owner, done */ addi r3,r3,THREAD /* want THREAD of task */ @@ -1163,6 +1165,7 @@ fix_mem_constants: icbi r0,r14 /* flush the icache line */ cmpw r12,r13 bne 1b + isync /* * Map the memory where the exception handlers will @@ -1208,9 +1211,9 @@ apus_interrupt_entry: mfmsr 20 xori r20,r20,MSR_DR - sync + SYNC mtmsr r20 - sync + isync lis r4,APUS_IPL_EMU@h @@ -1243,9 +1246,9 @@ apus_interrupt_entry: mfmsr r20 xori r20,r20,MSR_DR - sync + SYNC mtmsr r20 - sync + isync stw r3,(_CCR+4)(r21); @@ -1263,28 +1266,24 @@ apus_interrupt_entry: #endif /* CONFIG_APUS */ #ifdef CONFIG_SMP -#ifdef CONFIG_GEMINI - .globl __secondary_start_gemini -__secondary_start_gemini: - mfspr r4,HID0 - ori r4,r4,HID0_ICFI - li r3,0 - ori r3,r3,HID0_ICE - andc r4,r4,r3 - mtspr HID0,r4 - sync - bl prom_init - b __secondary_start -#endif /* CONFIG_GEMINI */ - .globl __secondary_start_psurge __secondary_start_psurge: li r24,1 /* cpu # */ + b __secondary_start_psurge99 + .globl __secondary_start_psurge2 +__secondary_start_psurge2: + li r24,2 /* cpu # */ + b __secondary_start_psurge99 + .globl __secondary_start_psurge3 +__secondary_start_psurge3: + li r24,3 /* cpu # */ + b __secondary_start_psurge99 +__secondary_start_psurge99: /* we come in here with IR=0 and DR=1, and DBAT 0 set to map the 0xf0000000 - 0xffffffff region */ mfmsr r0 rlwinm r0,r0,0,28,26 /* clear DR (0x10) */ - sync + SYNC mtmsr r0 isync @@ -1293,7 +1292,7 @@ __secondary_start: #ifdef CONFIG_PPC64BRIDGE mfmsr r0 clrldi r0,r0,1 /* make sure it's in 32-bit mode */ - sync + SYNC MTMSRD(r0) isync #else @@ -1445,21 +1444,6 @@ start_here: li r3,0 mtspr SPRG2,r3 /* 0 => r1 has kernel sp */ - /* Clear out the BSS */ - lis r11,_end@ha - addi r11,r11,_end@l - lis r8,__bss_start@ha - addi r8,r8,__bss_start@l - subf r11,r8,r11 - addi r11,r11,3 - rlwinm. r11,r11,30,2,31 - beq 2f - addi r8,r8,-4 - mtctr r11 - li r0,0 -3: stwu r0,4(r8) - bdnz 3b -2: /* stack */ addi r1,r2,TASK_UNION_SIZE li r0,0 @@ -1504,7 +1488,7 @@ start_here: RFI /* Load up the kernel context */ 2: - SYNC /* Force all PTE updates to finish */ + sync /* Force all PTE updates to finish */ tlbia /* Clear all TLB entries */ sync /* wait for tlbia/tlbie to finish */ #ifdef CONFIG_SMP @@ -1552,7 +1536,6 @@ _GLOBAL(set_context) * -- Cort */ clear_bats: -#if !defined(CONFIG_GEMINI) li r20,0 mfspr r9,PVR rlwinm r9,r9,16,16,31 /* r9 = 1 for 601, 4 for 604 */ @@ -1576,10 +1559,8 @@ clear_bats: mtspr IBAT2L,r20 mtspr IBAT3U,r20 mtspr IBAT3L,r20 -#endif /* !defined(CONFIG_GEMINI) */ blr -#ifndef CONFIG_GEMINI flush_tlbs: lis r20, 0x40 1: addic. r20, r20, -0x1000 @@ -1598,7 +1579,6 @@ mmu_off: mtspr SRR1,r3 sync RFI -#endif #ifndef CONFIG_POWER4 /* @@ -1745,3 +1725,12 @@ swapper_pg_dir: .globl cmd_line cmd_line: .space 512 + + .globl intercept_table +intercept_table: + .long 0, i0x100, i0x200, i0x300, i0x400, 0, i0x600, i0x700 + .long i0x800, 0, 0, 0, 0, i0xd00, 0, 0 + .long 0, 0, 0, i0x1300, 0, 0, 0, 0 + .long 0, 0, 0, 0, 0, 0, 0, 0 + .long 0, 0, 0, 0, 0, 0, 0, 0 + .long 0, 0, 0, 0, 0, 0, 0, 0 diff --git a/arch/ppc/kernel/i8259.c b/arch/ppc/kernel/i8259.c index bdb6ec844..7690d9091 100644 --- a/arch/ppc/kernel/i8259.c +++ b/arch/ppc/kernel/i8259.c @@ -10,12 +10,15 @@ unsigned char cached_8259[2] = { 0xff, 0xff }; #define cached_A1 (cached_8259[0]) #define cached_21 (cached_8259[1]) +spinlock_t i8259_lock = SPIN_LOCK_UNLOCKED; + int i8259_pic_irq_offset; int i8259_irq(int cpu) { int irq; + spin_lock/*_irqsave*/(&i8259_lock/*, flags*/); /* * Perform an interrupt acknowledge cycle on controller 1 */ @@ -40,14 +43,20 @@ int i8259_irq(int cpu) * interrupt */ outb(0x0b, 0x20); - if(~inb(0x20)&0x80) + if(~inb(0x20)&0x80) { + spin_unlock/*_irqrestore*/(&i8259_lock/*, flags*/); return -1; + } } + spin_unlock/*_irqrestore*/(&i8259_lock/*, flags*/); return irq; } static void i8259_mask_and_ack_irq(unsigned int irq_nr) { + unsigned long flags; + + spin_lock_irqsave(&i8259_lock, flags); if ( irq_nr >= i8259_pic_irq_offset ) irq_nr -= i8259_pic_irq_offset; @@ -63,6 +72,7 @@ static void i8259_mask_and_ack_irq(unsigned int irq_nr) outb(cached_21,0x21); outb(0x20,0x20); /* Non-specific EOI */ } + spin_unlock_irqrestore(&i8259_lock, flags); } static void i8259_set_irq_mask(int irq_nr) @@ -73,6 +83,9 @@ static void i8259_set_irq_mask(int irq_nr) static void i8259_mask_irq(unsigned int irq_nr) { + unsigned long flags; + + spin_lock_irqsave(&i8259_lock, flags); if ( irq_nr >= i8259_pic_irq_offset ) irq_nr -= i8259_pic_irq_offset; if ( irq_nr < 8 ) @@ -80,11 +93,14 @@ static void i8259_mask_irq(unsigned int irq_nr) else cached_A1 |= 1 << (irq_nr-8); i8259_set_irq_mask(irq_nr); + spin_unlock_irqrestore(&i8259_lock, flags); } static void i8259_unmask_irq(unsigned int irq_nr) { + unsigned long flags; + spin_lock_irqsave(&i8259_lock, flags); if ( irq_nr >= i8259_pic_irq_offset ) irq_nr -= i8259_pic_irq_offset; if ( irq_nr < 8 ) @@ -92,6 +108,13 @@ static void i8259_unmask_irq(unsigned int irq_nr) else cached_A1 &= ~(1 << (irq_nr-8)); i8259_set_irq_mask(irq_nr); + spin_unlock_irqrestore(&i8259_lock, flags); +} + +static void i8259_end_irq(unsigned int irq) +{ + if (!(irq_desc[irq].status & (IRQ_DISABLED|IRQ_INPROGRESS))) + i8259_unmask_irq(irq); } struct hw_interrupt_type i8259_pic = { @@ -101,11 +124,15 @@ struct hw_interrupt_type i8259_pic = { i8259_unmask_irq, i8259_mask_irq, i8259_mask_and_ack_irq, - 0 + i8259_end_irq, + NULL }; void __init i8259_init(void) { + unsigned long flags; + + spin_lock_irqsave(&i8259_lock, flags); /* init master interrupt controller */ outb(0x11, 0x20); /* Start init sequence */ outb(0x00, 0x21); /* Vector base */ @@ -120,7 +147,7 @@ void __init i8259_init(void) outb(0xFF, 0xA1); /* Mask all */ outb(cached_A1, 0xA1); outb(cached_21, 0x21); + spin_unlock_irqrestore(&i8259_lock, flags); request_irq( i8259_pic_irq_offset + 2, no_action, SA_INTERRUPT, "82c59 secondary cascade", NULL ); - enable_irq(i8259_pic_irq_offset + 2); /* Enable cascade interrupt */ } diff --git a/arch/ppc/kernel/idle.c b/arch/ppc/kernel/idle.c index a363a0e34..6ca616979 100644 --- a/arch/ppc/kernel/idle.c +++ b/arch/ppc/kernel/idle.c @@ -47,24 +47,35 @@ atomic_t zeropage_calls; /* # zero'd pages request that've been made */ int idled(void) { + int do_power_save = 0; + + /* only sleep on the 603-family/750 processors */ + switch (_get_PVR() >> 16) { + case 3: /* 603 */ + case 6: /* 603e */ + case 7: /* 603ev */ + case 8: /* 750 */ + case 12: /* 7400 */ + do_power_save = 1; + } + /* endless loop with no priority at all */ current->nice = 20; current->counter = -100; init_idle(); for (;;) { - __sti(); - - check_pgt_cache(); - - /*if ( !current->need_resched && zero_paged_on ) zero_paged();*/ - if ( !current->need_resched && htab_reclaim_on ) htab_reclaim(); - if ( !current->need_resched ) power_save(); + /*if ( !current->need_resched && zero_paged_on ) + zero_paged();*/ + if (!current->need_resched && htab_reclaim_on) + htab_reclaim(); + if (do_power_save && !current->need_resched) + power_save(); -#ifdef CONFIG_SMP - if (current->need_resched) -#endif + if (current->need_resched) { schedule(); + check_pgt_cache(); + } } return 0; } @@ -278,31 +289,31 @@ void zero_paged(void) void power_save(void) { - unsigned long msr, hid0; - - /* only sleep on the 603-family/750 processors */ - switch (_get_PVR() >> 16) { - case 3: /* 603 */ - case 6: /* 603e */ - case 7: /* 603ev */ - case 8: /* 750 */ - case 12: /* 7400 */ - save_flags(msr); - __cli(); - if (!current->need_resched) { - asm("mfspr %0,1008" : "=r" (hid0) :); - hid0 &= ~(HID0_NAP | HID0_SLEEP | HID0_DOZE); - hid0 |= (powersave_nap? HID0_NAP: HID0_DOZE) | HID0_DPM; - asm("mtspr 1008,%0" : : "r" (hid0)); + unsigned long hid0; + /* + * Disable interrupts to prevent a lost wakeup + * when going to sleep. This is necessary even with + * RTLinux since we are not guaranteed an interrupt + * didn't come in and is waiting for a __sti() before + * emulating one. This way, we really do hard disable. + * + * We assume that we're sti-ed when we come in here. We + * are in the idle loop so if we're cli-ed then it's a bug + * anyway. + * -- Cort + */ + _nmask_and_or_msr(MSR_EE, 0); + if (!current->need_resched) + { + asm("mfspr %0,1008" : "=r" (hid0) :); + hid0 &= ~(HID0_NAP | HID0_SLEEP | HID0_DOZE); + hid0 |= (powersave_nap? HID0_NAP: HID0_DOZE) | HID0_DPM; + asm("mtspr 1008,%0" : : "r" (hid0)); - /* set the POW bit in the MSR, and enable interrupts - * so we wake up sometime! */ - __sti(); /* this keeps rtl from getting confused -- Cort */ - _nmask_and_or_msr(0, MSR_POW | MSR_EE); - } - restore_flags(msr); - default: - return; + /* set the POW bit in the MSR, and enable interrupts + * so we wake up sometime! */ + _nmask_and_or_msr(0, MSR_POW | MSR_EE); } + _nmask_and_or_msr(0, MSR_EE); } diff --git a/arch/ppc/kernel/indirect_pci.c b/arch/ppc/kernel/indirect_pci.c index 552c552dc..3ba335d1f 100644 --- a/arch/ppc/kernel/indirect_pci.c +++ b/arch/ppc/kernel/indirect_pci.c @@ -9,113 +9,59 @@ * 2 of the License, or (at your option) any later version. */ +#include <linux/kernel.h> #include <linux/pci.h> -#include <asm/io.h> -#include <asm/system.h> - -unsigned int * pci_config_address; -unsigned char * pci_config_data; - -int indirect_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char *val) -{ - unsigned long flags; - - save_flags(flags); cli(); - - out_be32(pci_config_address, - ((offset&0xfc)<<24) | (dev_fn<<16) | (bus<<8) | 0x80); +#include <linux/delay.h> +#include <linux/string.h> +#include <linux/init.h> +#include <linux/bootmem.h> - *val= in_8(pci_config_data + (offset&3)); - - restore_flags(flags); - return PCIBIOS_SUCCESSFUL; +#include <asm/init.h> +#include <asm/io.h> +#include <asm/prom.h> +#include <asm/pci-bridge.h> +#include <asm/machdep.h> + +#include "pci.h" + +#define cfg_read(val, addr, type, op) *val = op((type)(addr)) +#define cfg_write(val, addr, type, op) op((type *)(addr), (val)) + +#define INDIRECT_PCI_OP(rw, size, type, op, mask) \ +static int \ +indirect_##rw##_config_##size(struct pci_dev *dev, int offset, type val) \ +{ \ + struct pci_controller *hose = dev->sysdata; \ + \ + out_be32(hose->cfg_addr, \ + ((offset & 0xfc) << 24) | (dev->devfn << 16) \ + | (dev->bus->number << 8) | 0x80); \ + cfg_##rw(val, hose->cfg_data + (offset & mask), type, op); \ + return PCIBIOS_SUCCESSFUL; \ } -int indirect_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short *val) -{ - unsigned long flags; - - if (offset&1) return PCIBIOS_BAD_REGISTER_NUMBER; - - save_flags(flags); cli(); - - out_be32(pci_config_address, - ((offset&0xfc)<<24) | (dev_fn<<16) | (bus<<8) | 0x80); +INDIRECT_PCI_OP(read, byte, u8 *, in_8, 3) +INDIRECT_PCI_OP(read, word, u16 *, in_le16, 2) +INDIRECT_PCI_OP(read, dword, u32 *, in_le32, 0) +INDIRECT_PCI_OP(write, byte, u8, out_8, 3) +INDIRECT_PCI_OP(write, word, u16, out_le16, 2) +INDIRECT_PCI_OP(write, dword, u32, out_le32, 0) - *val= in_le16((unsigned short *)(pci_config_data + (offset&3))); - - restore_flags(flags); - return PCIBIOS_SUCCESSFUL; -} - -int indirect_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int *val) +static struct pci_ops indirect_pci_ops = { - unsigned long flags; - - if (offset&3) return PCIBIOS_BAD_REGISTER_NUMBER; - - save_flags(flags); cli(); - - out_be32(pci_config_address, - ((offset&0xfc)<<24) | (dev_fn<<16) | (bus<<8) | 0x80); - - *val= in_le32((unsigned *)pci_config_data); - - restore_flags(flags); - return PCIBIOS_SUCCESSFUL; -} - -int indirect_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char val) + indirect_read_config_byte, + indirect_read_config_word, + indirect_read_config_dword, + indirect_write_config_byte, + indirect_write_config_word, + indirect_write_config_dword +}; + +void __init +setup_indirect_pci(struct pci_controller* hose, u32 cfg_addr, u32 cfg_data) { - unsigned long flags; - - save_flags(flags); cli(); - - out_be32(pci_config_address, - ((offset&0xfc)<<24) | (dev_fn<<16) | (bus<<8) | 0x80); - - out_8(pci_config_data + (offset&3), val); - - restore_flags(flags); - return PCIBIOS_SUCCESSFUL; -} - -int indirect_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short val) -{ - unsigned long flags; - - if (offset&1) return PCIBIOS_BAD_REGISTER_NUMBER; - - save_flags(flags); cli(); - - out_be32(pci_config_address, - ((offset&0xfc)<<24) | (dev_fn<<16) | (bus<<8) | 0x80); - - out_le16((unsigned short *)(pci_config_data + (offset&3)), val); - - restore_flags(flags); - return PCIBIOS_SUCCESSFUL; + hose->ops = &indirect_pci_ops; + hose->cfg_addr = (unsigned int *) ioremap(cfg_addr, 4); + hose->cfg_data = (unsigned char *) ioremap(cfg_data, 4); } -int indirect_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int val) -{ - unsigned long flags; - - if (offset&3) return PCIBIOS_BAD_REGISTER_NUMBER; - - save_flags(flags); cli(); - - out_be32(pci_config_address, - ((offset&0xfc)<<24) | (dev_fn<<16) | (bus<<8) | 0x80); - - out_le32((unsigned *)pci_config_data, val); - - restore_flags(flags); - return PCIBIOS_SUCCESSFUL; -} diff --git a/arch/ppc/kernel/irq.c b/arch/ppc/kernel/irq.c index a0caa4a4c..02ff4d964 100644 --- a/arch/ppc/kernel/irq.c +++ b/arch/ppc/kernel/irq.c @@ -41,11 +41,11 @@ #include <linux/config.h> #include <linux/init.h> #include <linux/malloc.h> -#include <linux/openpic.h> #include <linux/pci.h> #include <linux/delay.h> #include <linux/irq.h> #include <linux/proc_fs.h> +#include <linux/random.h> #include <asm/uaccess.h> #include <asm/bitops.h> @@ -64,19 +64,22 @@ #include "local_irq.h" -extern volatile unsigned long ipi_count; +atomic_t ipi_recv; +atomic_t ipi_sent; void enable_irq(unsigned int irq_nr); void disable_irq(unsigned int irq_nr); -volatile unsigned char *chrp_int_ack_special; +static void register_irq_proc (unsigned int irq); #define MAXCOUNT 10000000 -irq_desc_t irq_desc[NR_IRQS]; +irq_desc_t irq_desc[NR_IRQS] __cacheline_aligned = + { [0 ... NR_IRQS-1] = { 0, NULL, NULL, 0, SPIN_LOCK_UNLOCKED}}; + int ppc_spurious_interrupts = 0; struct irqaction *ppc_irq_action[NR_IRQS]; -unsigned int ppc_cached_irq_mask[NR_MASK_WORDS]; -unsigned int ppc_lost_interrupts[NR_MASK_WORDS]; +unsigned long ppc_cached_irq_mask[NR_MASK_WORDS]; +unsigned long ppc_lost_interrupts[NR_MASK_WORDS]; atomic_t ppc_n_lost_interrupts; /* nasty hack for shared irq's since we need to do kmalloc calls but @@ -115,52 +118,136 @@ void irq_kfree(void *ptr) kfree(ptr); } +int +setup_irq(unsigned int irq, struct irqaction * new) +{ + int shared = 0; + unsigned long flags; + struct irqaction *old, **p; + irq_desc_t *desc = irq_desc + irq; + + /* + * Some drivers like serial.c use request_irq() heavily, + * so we have to be careful not to interfere with a + * running system. + */ + if (new->flags & SA_SAMPLE_RANDOM) { + /* + * This function might sleep, we want to call it first, + * outside of the atomic block. + * Yes, this might clear the entropy pool if the wrong + * driver is attempted to be loaded, without actually + * installing a new handler, but is this really a problem, + * only the sysadmin is able to do this. + */ + rand_initialize_irq(irq); + } + + /* + * The following block of code has to be executed atomically + */ + spin_lock_irqsave(&desc->lock,flags); + p = &desc->action; + if ((old = *p) != NULL) { + /* Can't share interrupts unless both agree to */ + if (!(old->flags & new->flags & SA_SHIRQ)) { + spin_unlock_irqrestore(&desc->lock,flags); + return -EBUSY; + } + + /* add new interrupt at end of irq queue */ + do { + p = &old->next; + old = *p; + } while (old); + shared = 1; + } + + *p = new; + + if (!shared) { + desc->depth = 0; + desc->status &= ~(IRQ_DISABLED | IRQ_AUTODETECT | IRQ_WAITING); + unmask_irq(irq); + } + spin_unlock_irqrestore(&desc->lock,flags); + + register_irq_proc(irq); + return 0; +} + +/* This could be promoted to a real free_irq() ... */ +static int +do_free_irq(int irq, void* dev_id) +{ + irq_desc_t *desc; + struct irqaction **p; + unsigned long flags; + + desc = irq_desc + irq; + spin_lock_irqsave(&desc->lock,flags); + p = &desc->action; + for (;;) { + struct irqaction * action = *p; + if (action) { + struct irqaction **pp = p; + p = &action->next; + if (action->dev_id != dev_id) + continue; + + /* Found it - now remove it from the list of entries */ + *pp = action->next; + if (!desc->action) { + desc->status |= IRQ_DISABLED; + mask_irq(irq); + } + spin_unlock_irqrestore(&desc->lock,flags); + +#ifdef CONFIG_SMP + /* Wait to make sure it's not being used on another CPU */ + while (desc->status & IRQ_INPROGRESS) + barrier(); +#endif + irq_kfree(action); + return 0; + } + printk("Trying to free free IRQ%d\n",irq); + spin_unlock_irqrestore(&desc->lock,flags); + break; + } + return -ENOENT; +} + #if (defined(CONFIG_8xx) || defined(CONFIG_8260)) /* Name change so we can catch standard drivers that potentially mess up * the internal interrupt controller on 8xx and 8260. Just bear with me, * I don't like this either and I am searching a better solution. For * now, this is what I need. -- Dan */ -int request_8xxirq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *), +#define request_irq request_8xxirq #elif defined(CONFIG_APUS) -int request_sysirq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *), -#else -int request_irq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *), +#define request_irq request_sysirq +#define free_irq sys_free_irq #endif + +int request_irq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *), unsigned long irqflags, const char * devname, void *dev_id) { - struct irqaction *old, **p, *action; - unsigned long flags; + struct irqaction *action; + int retval; if (irq >= NR_IRQS) return -EINVAL; if (!handler) - { - /* Free */ - p = &irq_desc[irq].action; - while ((action = *p) != NULL && action->dev_id != dev_id) - p = &action->next; - if (action == NULL) - return -ENOENT; - - /* Found it - now free it */ - save_flags(flags); - cli(); - *p = action->next; - if (irq_desc[irq].action == NULL) - disable_irq(irq); - restore_flags(flags); - irq_kfree(action); - return 0; - } + /* We could implement really free_irq() instead of that... */ + return do_free_irq(irq, dev_id); action = (struct irqaction *) irq_kmalloc(sizeof(struct irqaction), GFP_KERNEL); - if (!action) + if (!action) { + printk(KERN_ERR "irq_kmalloc() failed for irq %d !\n", irq); return -ENOMEM; - - save_flags(flags); - cli(); + } action->handler = handler; action->flags = irqflags; @@ -168,57 +255,109 @@ int request_irq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *) action->name = devname; action->dev_id = dev_id; action->next = NULL; - enable_irq(irq); - p = &irq_desc[irq].action; - - if ((old = *p) != NULL) { - /* Can't share interrupts unless both agree to */ - if (!(old->flags & action->flags & SA_SHIRQ)) - return -EBUSY; - /* add new interrupt at end of irq queue */ - do { - p = &old->next; - old = *p; - } while (old); - } - *p = action; - - restore_flags(flags); + retval = setup_irq(irq, action); + if (retval) + kfree(action); + return 0; } -#ifdef CONFIG_APUS -void sys_free_irq(unsigned int irq, void *dev_id) -{ - sys_request_irq(irq, NULL, 0, NULL, dev_id); -} -#else void free_irq(unsigned int irq, void *dev_id) { -#if (defined(CONFIG_8xx) || defined(CONFIG_8260)) - request_8xxirq(irq, NULL, 0, NULL, dev_id); -#else request_irq(irq, NULL, 0, NULL, dev_id); -#endif } -#endif -/* XXX should implement irq disable depth like on intel */ -void disable_irq_nosync(unsigned int irq_nr) +/* + * Generic enable/disable code: this just calls + * down into the PIC-specific version for the actual + * hardware disable after having gotten the irq + * controller lock. + */ + +/** + * disable_irq_nosync - disable an irq without waiting + * @irq: Interrupt to disable + * + * Disable the selected interrupt line. Disables of an interrupt + * stack. Unlike disable_irq(), this function does not ensure existing + * instances of the IRQ handler have completed before returning. + * + * This function may be called from IRQ context. + */ + + void disable_irq_nosync(unsigned int irq) { - mask_irq(irq_nr); + irq_desc_t *desc = irq_desc + irq; + unsigned long flags; + + spin_lock_irqsave(&desc->lock, flags); + if (!desc->depth++) { + if (!(desc->status & IRQ_PER_CPU)) + desc->status |= IRQ_DISABLED; + mask_irq(irq); + } + spin_unlock_irqrestore(&desc->lock, flags); } -void disable_irq(unsigned int irq_nr) +/** + * disable_irq - disable an irq and wait for completion + * @irq: Interrupt to disable + * + * Disable the selected interrupt line. Disables of an interrupt + * stack. That is for two disables you need two enables. This + * function waits for any pending IRQ handlers for this interrupt + * to complete before returning. If you use this function while + * holding a resource the IRQ handler may need you will deadlock. + * + * This function may be called - with care - from IRQ context. + */ + +void disable_irq(unsigned int irq) { - mask_irq(irq_nr); - synchronize_irq(); + disable_irq_nosync(irq); + + if (!local_irq_count(smp_processor_id())) { + do { + barrier(); + } while (irq_desc[irq].status & IRQ_INPROGRESS); + } } -void enable_irq(unsigned int irq_nr) +/** + * enable_irq - enable interrupt handling on an irq + * @irq: Interrupt to enable + * + * Re-enables the processing of interrupts on this IRQ line + * providing no disable_irq calls are now in effect. + * + * This function may be called from IRQ context. + */ + +void enable_irq(unsigned int irq) { - unmask_irq(irq_nr); + irq_desc_t *desc = irq_desc + irq; + unsigned long flags; + + spin_lock_irqsave(&desc->lock, flags); + switch (desc->depth) { + case 1: { + unsigned int status = desc->status & ~IRQ_DISABLED; + desc->status = status; + if ((status & (IRQ_PENDING | IRQ_REPLAY)) == IRQ_PENDING) { + desc->status = status | IRQ_REPLAY; + hw_resend_irq(desc->handler,irq); + } + unmask_irq(irq); + /* fall-through */ + } + default: + desc->depth--; + break; + case 0: + printk("enable_irq(%u) unbalanced\n", irq); + } + spin_unlock_irqrestore(&desc->lock, flags); } int get_irq_list(char *buf) @@ -257,15 +396,41 @@ int get_irq_list(char *buf) } len += sprintf(buf+len, "\n"); } +#ifdef CONFIG_TAU + len += sprintf(buf+len, "TAU: "); + for (j = 0; j < smp_num_cpus; j++) + len += sprintf(buf+len, "%10u ", + tau_interrupts(j)); + len += sprintf(buf+len, "\n"); +#endif #ifdef CONFIG_SMP /* should this be per processor send/receive? */ - len += sprintf(buf+len, "IPI: %10lu\n", ipi_count); + len += sprintf(buf+len, "IPI (recv/sent): %10u/%u\n", + atomic_read(&ipi_recv), atomic_read(&ipi_sent)); #endif len += sprintf(buf+len, "BAD: %10u\n", ppc_spurious_interrupts); return len; #endif /* CONFIG_APUS */ } +static inline void +handle_irq_event(int irq, struct pt_regs *regs, struct irqaction *action) +{ + int status = 0; + + if (!(action->flags & SA_INTERRUPT)) + __sti(); + + do { + status |= action->flags; + action->handler(irq, action->dev_id, regs); + action = action->next; + } while (action); + if (status & SA_SAMPLE_RANDOM) + add_interrupt_randomness(irq); + __cli(); +} + /* * Eventually, this should take an array of interrupts and an array size * so it can dispatch multiple interrupts. @@ -275,33 +440,87 @@ void ppc_irq_dispatch_handler(struct pt_regs *regs, int irq) int status; struct irqaction *action; int cpu = smp_processor_id(); + irq_desc_t *desc = irq_desc + irq; - mask_and_ack_irq(irq); - status = 0; - action = irq_desc[irq].action; kstat.irqs[cpu][irq]++; - if (action && action->handler) { - if (!(action->flags & SA_INTERRUPT)) - __sti(); - do { - status |= action->flags; - action->handler(irq, action->dev_id, regs); - action = action->next; - } while ( action ); - __cli(); - if (irq_desc[irq].handler) { - if (irq_desc[irq].handler->end) - irq_desc[irq].handler->end(irq); - else if (irq_desc[irq].handler->enable) - irq_desc[irq].handler->enable(irq); + spin_lock(&desc->lock); + ack_irq(irq); + /* + REPLAY is when Linux resends an IRQ that was dropped earlier + WAITING is used by probe to mark irqs that are being tested + */ + status = desc->status & ~(IRQ_REPLAY | IRQ_WAITING); + if (!(status & IRQ_PER_CPU)) + status |= IRQ_PENDING; /* we _want_ to handle it */ + + /* + * If the IRQ is disabled for whatever reason, we cannot + * use the action we have. + */ + action = NULL; + if (!(status & (IRQ_DISABLED | IRQ_INPROGRESS))) { + action = desc->action; + if (!action || !action->handler) { + ppc_spurious_interrupts++; + printk(KERN_DEBUG "Unhandled interrupt %x, disabled\n", irq); + /* We can't call disable_irq here, it would deadlock */ + if (!desc->depth) + desc->depth = 1; + desc->status |= IRQ_DISABLED; + /* This is not a real spurrious interrupt, we + * have to eoi it, so we jump to out + */ + mask_irq(irq); + goto out; } - } else { - ppc_spurious_interrupts++; - printk(KERN_DEBUG "Unhandled interrupt %x, disabled\n", irq); - disable_irq(irq); + status &= ~IRQ_PENDING; /* we commit to handling */ + if (!(status & IRQ_PER_CPU)) + status |= IRQ_INPROGRESS; /* we are handling it */ + } + desc->status = status; + + /* + * If there is no IRQ handler or it was disabled, exit early. + Since we set PENDING, if another processor is handling + a different instance of this same irq, the other processor + will take care of it. + */ + if (!action) + goto out; + + + /* + * Edge triggered interrupts need to remember + * pending events. + * This applies to any hw interrupts that allow a second + * instance of the same irq to arrive while we are in do_IRQ + * or in the handler. But the code here only handles the _second_ + * instance of the irq, not the third or fourth. So it is mostly + * useful for irq hardware that does not mask cleanly in an + * SMP environment. + */ + for (;;) { + spin_unlock(&desc->lock); + handle_irq_event(irq, regs, action); + spin_lock(&desc->lock); + + if (!(desc->status & IRQ_PENDING)) + break; + desc->status &= ~IRQ_PENDING; + } + desc->status &= ~IRQ_INPROGRESS; +out: + /* + * The ->end() handler has to deal with interrupts which got + * disabled while the handler was running. + */ + if (irq_desc[irq].handler) { if (irq_desc[irq].handler->end) irq_desc[irq].handler->end(irq); + else if (irq_desc[irq].handler->enable) + irq_desc[irq].handler->enable(irq); } + spin_unlock(&desc->lock); } int do_IRQ(struct pt_regs *regs, int isfake) @@ -320,6 +539,7 @@ int do_IRQ(struct pt_regs *regs, int isfake) { printk(KERN_DEBUG "Bogus interrupt %d from PC = %lx\n", irq, regs->nip); + /* That's not SMP safe ... but who cares ? */ ppc_spurious_interrupts++; } goto out; @@ -362,7 +582,7 @@ void __init init_IRQ(void) #ifdef CONFIG_SMP unsigned char global_irq_holder = NO_PROC_ID; -unsigned volatile int global_irq_lock; +unsigned volatile long global_irq_lock; /* pendantic :long for set_bit--RR*/ atomic_t global_irq_count; atomic_t global_bh_count; @@ -634,7 +854,11 @@ static struct proc_dir_entry * root_irq_dir; static struct proc_dir_entry * irq_dir [NR_IRQS]; static struct proc_dir_entry * smp_affinity_entry [NR_IRQS]; +#ifdef CONFIG_IRQ_ALL_CPUS unsigned int irq_affinity [NR_IRQS] = { [0 ... NR_IRQS-1] = 0xffffffff}; +#else /* CONFIG_IRQ_ALL_CPUS */ +unsigned int irq_affinity [NR_IRQS] = { [0 ... NR_IRQS-1] = 0x00000000}; +#endif /* CONFIG_IRQ_ALL_CPUS */ #define HEX_DIGITS 8 @@ -694,6 +918,7 @@ static int irq_affinity_write_proc (struct file *file, const char *buffer, err = parse_hex_value(buffer, count, &new_value); +/* Why is this disabled ? --BenH */ #if 0/*CONFIG_SMP*/ /* * Do not allow disabling IRQs completely - it's a too easy diff --git a/arch/ppc/kernel/local_irq.h b/arch/ppc/kernel/local_irq.h index 5c616bbbd..319e3377c 100644 --- a/arch/ppc/kernel/local_irq.h +++ b/arch/ppc/kernel/local_irq.h @@ -15,8 +15,5 @@ void ppc_irq_dispatch_handler(struct pt_regs *regs, int irq); extern int ppc_spurious_interrupts; extern int ppc_second_irq; extern struct irqaction *ppc_irq_action[NR_IRQS]; -extern unsigned int ppc_cached_irq_mask[NR_MASK_WORDS]; -extern unsigned int ppc_lost_interrupts[NR_MASK_WORDS]; -extern atomic_t ppc_n_lost_interrupts; #endif /* _PPC_KERNEL_LOCAL_IRQ_H */ diff --git a/arch/ppc/kernel/m8260_setup.c b/arch/ppc/kernel/m8260_setup.c index 6e006a867..a55f5235d 100644 --- a/arch/ppc/kernel/m8260_setup.c +++ b/arch/ppc/kernel/m8260_setup.c @@ -62,7 +62,7 @@ extern void mackbd_leds(unsigned char leds); extern void mackbd_init_hw(void); #endif -extern unsigned long loops_per_sec; +extern unsigned long loops_per_jiffy; unsigned char __res[sizeof(bd_t)]; unsigned long empty_zero_page[1024]; @@ -286,10 +286,7 @@ m8260_init(unsigned long r3, unsigned long r4, unsigned long r5, ppc_ide_md.outsw = m8xx_ide_outsw; ppc_ide_md.default_irq = m8xx_ide_default_irq; ppc_ide_md.default_io_base = m8xx_ide_default_io_base; - ppc_ide_md.check_region = m8xx_ide_check_region; - ppc_ide_md.request_region = m8xx_ide_request_region; - ppc_ide_md.release_region = m8xx_ide_release_region; - ppc_ide_md.fix_driveid = m8xx_ide_fix_driveid; + ppc_ide_md.fix_driveid = ppc_generic_ide_fix_driveid; ppc_ide_md.ide_init_hwif = m8xx_ide_init_hwif_ports; ppc_ide_md.ide_request_irq = m8xx_ide_request_irq; @@ -297,13 +294,39 @@ m8260_init(unsigned long r3, unsigned long r4, unsigned long r5, #endif } -void +/* + * Copied from prom.c so I don't have include all of that crap. + * -- Dan + * + * prom_init() is called very early on, before the kernel text + * and data have been mapped to KERNELBASE. At this point the code + * is running at whatever address it has been loaded at, so + * references to extern and static variables must be relocated + * explicitly. The procedure reloc_offset() returns the address + * we're currently running at minus the address we were linked at. + * (Note that strings count as static variables.) + */ +extern unsigned long reloc_offset(void); +#define PTRRELOC(x) ((typeof(x))((unsigned long)(x) + offset)) + +__init +unsigned long prom_init(uint r3, uint r4, uint r5, uint r6) { - /* Nothing to do now, but we are called immediatedly upon - * kernel start up with MMU disabled, so if there is - * anything we need to do...... - */ + unsigned long offset = reloc_offset(); + unsigned long phys; + extern char __bss_start, _end; + + /* First zero the BSS -- use memset, some arches don't have + * caches on yet */ + memset_io(PTRRELOC(&__bss_start),0 , &_end - &__bss_start); + + /* Default */ + phys = offset + KERNELBASE; + + /* We are done. + */ + return phys; } /* Mainly for ksyms. diff --git a/arch/ppc/kernel/m8xx_setup.c b/arch/ppc/kernel/m8xx_setup.c index 7dc408a13..f9813fa75 100644 --- a/arch/ppc/kernel/m8xx_setup.c +++ b/arch/ppc/kernel/m8xx_setup.c @@ -63,8 +63,6 @@ extern void mackbd_leds(unsigned char leds); extern void mackbd_init_hw(void); #endif -extern unsigned long loops_per_sec; - unsigned char __res[sizeof(bd_t)]; unsigned long empty_zero_page[1024]; @@ -344,25 +342,6 @@ m8xx_ide_default_io_base(int index) } int -m8xx_ide_check_region(ide_ioreg_t from, unsigned int extent) -{ - return 0; -} - -void -m8xx_ide_request_region(ide_ioreg_t from, - unsigned int extent, - const char *name) -{ -} - -void -m8xx_ide_release_region(ide_ioreg_t from, - unsigned int extent) -{ -} - -int m8xx_ide_request_irq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *), unsigned long flags, @@ -376,12 +355,6 @@ m8xx_ide_request_irq(unsigned int irq, #endif } -void -m8xx_ide_fix_driveid(struct hd_driveid *id) -{ - ppc_generic_ide_fix_driveid(id); -} - /* We can use an external IDE controller or wire the IDE interface to * the internal PCMCIA controller. */ @@ -515,10 +488,7 @@ m8xx_init(unsigned long r3, unsigned long r4, unsigned long r5, ppc_ide_md.outsw = m8xx_ide_outsw; ppc_ide_md.default_irq = m8xx_ide_default_irq; ppc_ide_md.default_io_base = m8xx_ide_default_io_base; - ppc_ide_md.check_region = m8xx_ide_check_region; - ppc_ide_md.request_region = m8xx_ide_request_region; - ppc_ide_md.release_region = m8xx_ide_release_region; - ppc_ide_md.fix_driveid = m8xx_ide_fix_driveid; + ppc_ide_md.fix_driveid = ppc_generic_ide_fix_driveid; ppc_ide_md.ide_init_hwif = m8xx_ide_init_hwif_ports; ppc_ide_md.ide_request_irq = m8xx_ide_request_irq; diff --git a/arch/ppc/kernel/misc.S b/arch/ppc/kernel/misc.S index 94e1cd277..3f54003c7 100644 --- a/arch/ppc/kernel/misc.S +++ b/arch/ppc/kernel/misc.S @@ -37,6 +37,14 @@ .text + .align 5 +_GLOBAL(__delay) + cmpwi 0,r3,0 + mtctr r3 + beqlr +1: bdnz 1b + blr + /* * Returns (address we're running at) - (address we were linked at) * for use before the text and data are mapped to KERNELBASE. @@ -82,16 +90,16 @@ _GLOBAL(__no_use_restore_flags) lwz r7,ppc_n_lost_interrupts@l(r7) cmpi 0,r7,0 /* lost interrupts to process first? */ bne- do_lost_interrupts -1: sync +1: SYNC mtmsr r3 - isync + SYNC blr _GLOBAL(__no_use_cli) mfmsr r0 /* Get current interrupt state */ rlwinm r3,r0,16+1,32-1,31 /* Extract old value of 'EE' */ rlwinm r0,r0,0,17,15 /* clear MSR_EE in r0 */ - sync /* Some chip revs have problems here... */ + SYNC /* Some chip revs have problems here... */ mtmsr r0 /* Update machine state */ blr /* Done */ @@ -102,7 +110,7 @@ _GLOBAL(__no_use_sti) ori r3,r3,MSR_EE /* Turn on 'EE' bit */ cmpi 0,r4,0 /* lost interrupts to process first? */ bne- do_lost_interrupts - sync /* Some chip revs have problems here... */ + SYNC /* Some chip revs have problems here... */ mtmsr r3 /* Update machine state */ blr @@ -121,7 +129,7 @@ _GLOBAL(do_lost_interrupts) cmpi 0,r4,0 bne- 1b lwz r3,8(r1) - sync + SYNC mtmsr r3 lwz r0,20(r1) mtlr r0 @@ -137,8 +145,9 @@ _GLOBAL(do_lost_interrupts) mfmsr r0 /* Get current msr */ andc r0,r0,r3 /* And off the bits set in r3 (first parm) */ or r0,r0,r4 /* Or on the bits in r4 (second parm) */ - sync /* Some chip revs have problems here... */ + SYNC /* Some chip revs have problems here... */ mtmsr r0 /* Update machine state */ + isync blr /* Done */ @@ -148,7 +157,7 @@ _GLOBAL(do_lost_interrupts) _GLOBAL(_tlbia) #if defined(CONFIG_SMP) mfmsr r10 - sync + SYNC rlwinm r0,r10,0,17,15 /* clear bit 16 (MSR_EE) */ mtmsr r0 SYNC @@ -161,7 +170,6 @@ _GLOBAL(_tlbia) bne- 10b stwcx. r8,0,r9 bne- 10b - eieio #endif /* CONFIG_SMP */ sync tlbia @@ -182,7 +190,7 @@ _GLOBAL(_tlbia) _GLOBAL(_tlbie) #if defined(CONFIG_SMP) mfmsr r10 - sync + SYNC rlwinm r0,r10,0,17,15 /* clear bit 16 (MSR_EE) */ mtmsr r0 SYNC @@ -228,7 +236,7 @@ _GLOBAL(flush_instruction_cache) ori r3,r3,HID0_ICFI mtspr HID0,r3 #endif /* CONFIG_8xx */ - SYNC + isync blr /* @@ -259,7 +267,6 @@ _GLOBAL(flush_icache_range) 2: icbi 0,r6 addi r6,r6,CACHE_LINE_SIZE bdnz 2b - sync isync blr @@ -717,6 +724,8 @@ _GLOBAL(_get_SP) mr r3,r1 /* Close enough */ blr +#if 0 +/* isn't it just easier to use the mtspr/mfspr inline macros?? --Troy */ _GLOBAL(_get_THRM1) mfspr r3,THRM1 blr @@ -740,6 +749,7 @@ _GLOBAL(_set_THRM2) _GLOBAL(_set_THRM3) mtspr THRM3,r3 blr +#endif _GLOBAL(_get_PVR) mfspr r3,PVR @@ -755,6 +765,12 @@ _GLOBAL(_get_HID0) mfspr r3,HID0 blr +_GLOBAL(_set_HID0) + sync + mtspr HID0, r3 + SYNC /* Handle erratas in some cases */ + blr + _GLOBAL(_get_ICTC) mfspr r3,ICTC blr @@ -763,7 +779,6 @@ _GLOBAL(_set_ICTC) mtspr ICTC,r3 blr - /* L2CR functions Copyright © 1997-1998 by PowerLogix R & D, Inc. @@ -785,6 +800,17 @@ _GLOBAL(_set_ICTC) /* Thur, Dec. 12, 1998. - First public release, contributed by PowerLogix. + *********** + Sat, Aug. 7, 1999. + - Terry: Made sure code disabled interrupts before running. (Previously + it was assumed interrupts were already disabled). + - Terry: Updated for tentative G4 support. 4MB of memory is now flushed + instead of 2MB. (Prob. only 3 is necessary). + - Terry: Updated for workaround to HID0[DPM] processor bug + during global invalidates. + *********** + Thu, July 13, 2000. + - Terry: Added isync to correct for an errata. Author: Terry Greeniaus (tgree@phys.ualberta.ca) Please e-mail updates to this file to me, thanks! @@ -823,82 +849,94 @@ _GLOBAL(_set_ICTC) causes cache pushes from the L1 cache to go to the L2 cache instead of to main memory. */ - +/* + * Summary: this procedure ignores the L2I bit in the value passed in, + * flushes the cache if it was already enabled, always invalidates the + * cache, then enables the cache if the L2E bit is set in the value + * passed in. + * -- paulus. + */ _GLOBAL(_set_L2CR) - /* Make sure this is a 750 chip */ + /* Make sure this is a 750 or 7400 chip */ mfspr r4,PVR rlwinm r4,r4,16,16,31 - cmplwi r4,0x0008 - beq thisIs750 - cmplwi r4,0x000c - beq thisIs750 - li r3,-1 - blr - -thisIs750: - /* Get the current enable bit of the L2CR into r4 */ - mfspr r4,L2CR - mfmsr r7 - - /* See if we want to perform a global inval this time. */ - rlwinm r6,r3,0,10,10 /* r6 contains the new invalidate bit */ - rlwinm. r5,r3,0,0,0 /* r5 contains the new enable bit */ - rlwinm r3,r3,0,11,9 /* Turn off the invalidate bit */ - rlwimi r3,r4,0,0,0 /* Keep the enable bit the same as it was. */ - bne dontDisableCache /* Only disable the cache if L2CRApply - has the enable bit off */ - -disableCache: - /* Disable the cache. First, we turn off interrupts. - An interrupt while we are flushing the cache could bring - in data which may not get properly flushed. */ - rlwinm r4,r7,0,17,15 /* Turn off EE bit */ + cmpwi r4,0x0008 + cmpwi cr1,r4,0x000c + cror 2,2,4*cr1+2 + bne 99f + + /* Turn off interrupts and data relocation. */ + mfmsr r7 /* Save MSR in r7 */ + rlwinm r4,r7,0,17,15 + rlwinm r4,r4,0,28,26 /* Turn off DR bit */ sync mtmsr r4 - sync + isync + + /* Get the current enable bit of the L2CR into r4 */ + mfspr r4,L2CR -/* - Now, read the first 2MB of memory to put new data in the cache. - (Actually we only need the size of the L2 cache plus the size - of the L1 cache, but 2MB will cover everything just to be safe). -*/ - lis r4,0x0001 + /* Tweak some bits */ + rlwinm r5,r3,0,0,0 /* r5 contains the new enable bit */ + rlwinm r3,r3,0,11,9 /* Turn off the invalidate bit */ + rlwinm r3,r3,0,1,31 /* Turn off the enable bit */ + + /* Check to see if we need to flush */ + rlwinm. r4,r4,0,0,0 + beq 2f + + /* Flush the cache. First, read the first 4MB of memory (physical) to + * put new data in the cache. (Actually we only need + * the size of the L2 cache plus the size of the L1 cache, but 4MB will + * cover everything just to be safe). + */ + + /**** Might be a good idea to set L2DO here - to prevent instructions + from getting into the cache. But since we invalidate + the next time we enable the cache it doesn't really matter. + ****/ + + lis r4,0x0002 mtctr r4 - lis r4,KERNELBASE@h -1: lwzx r0,r0,r4 - addi r4,r4,0x0020 /* Go to start of next cache line */ + li r4,0 +1: + lwzx r0,r0,r4 + addi r4,r4,32 /* Go to start of next cache line */ bdnz 1b - /* Now, flush the first 2MB of memory */ - lis r4,0x0001 + /* Now, flush the first 4MB of memory */ + lis r4,0x0002 mtctr r4 - lis r4,KERNELBASE@h + li r4,0 sync -2: dcbf r0,r4 - addi r4,r4,0x0020 /* Go to start of next cache line */ - bdnz 2b - - /* Turn off the L2CR enable bit. */ - rlwinm r3,r3,0,1,31 - -dontDisableCache: - /* Set up the L2CR configuration bits */ +1: + dcbf r0,r4 + addi r4,r4,32 /* Go to start of next cache line */ + bdnz 1b + +2: + /* Set up the L2CR configuration bits (and switch L2 off) */ sync mtspr L2CR,r3 sync - /* Reenable interrupts if necessary. */ - mtmsr r7 + /* Before we perform the global invalidation, we must disable dynamic + * power management via HID0[DPM] to work around a processor bug where + * DPM can possibly interfere with the state machine in the processor + * that invalidates the L2 cache tags. + */ + mfspr r8,HID0 /* Save HID0 in r8 */ + rlwinm r4,r8,0,12,10 /* Turn off HID0[DPM] */ sync - - cmplwi r6,0 - beq noInval - + mtspr HID0,r4 /* Disable DPM */ + sync + /* Perform a global invalidation */ oris r3,r3,0x0020 sync mtspr L2CR,r3 sync + isync /* For errata */ /* Wait for the invalidation to complete */ 3: mfspr r3,L2CR @@ -910,27 +948,38 @@ dontDisableCache: mtspr L2CR,r3 sync -noInval: + /* Restore HID0[DPM] to whatever it was before */ + sync + mtspr 1008,r8 + sync + /* See if we need to enable the cache */ cmplwi r5,0 - beqlr + beq 4f /* Enable the cache */ oris r3,r3,0x8000 mtspr L2CR,r3 sync + + /* Restore MSR (restores EE and DR bits to original state) */ +4: SYNC + mtmsr r7 + isync + blr + +99: li r3,-1 blr _GLOBAL(_get_L2CR) /* Make sure this is a 750 chip */ mfspr r3,PVR - rlwinm r3,r3,16,16,31 - cmplwi r3,0x0008 - beq 1f - cmplwi r3,0x000c + srwi r3,r3,16 + cmpwi r3,0x0008 + cmpwi cr1,r3,0x000c li r3,0 + cror 2,2,4*cr1+2 bnelr -1: /* Return the L2CR contents */ mfspr r3,L2CR blr @@ -986,15 +1035,6 @@ _GLOBAL(cvt_df) blr #endif -_GLOBAL(__clear_msr_me) - mfmsr r0 /* Get current interrupt state */ - lis r3,0 - ori r3,r3,MSR_ME - andc r0,r0,r3 /* Clears bit in (r4) */ - sync /* Some chip revs have problems here */ - mtmsr r0 /* Update machine state */ - blr - /* * Create a kernel thread * kernel_thread(fn, arg, flags) @@ -1244,16 +1284,20 @@ _GLOBAL(sys_call_table) .long sys_getrlimit /* 190 */ .long sys_ni_syscall /* 191 */ /* Unused */ .long sys_ni_syscall /* 192 - reserved - mmap2 */ - .long sys_ni_syscall /* 193 - reserved - truncate64 */ - .long sys_ni_syscall /* 194 - reserved - ftruncate64 */ - .long sys_ni_syscall /* 195 - reserved - stat64 */ - .long sys_ni_syscall /* 196 - reserved - lstat64 */ - .long sys_ni_syscall /* 197 - reserved - fstat64 */ + .long sys_truncate64 /* 193 */ + .long sys_ftruncate64 /* 194 */ + .long sys_stat64 /* 195 */ + .long sys_lstat64 /* 196 */ + .long sys_fstat64 /* 197 */ .long sys_pciconfig_read /* 198 */ .long sys_pciconfig_write /* 199 */ .long sys_pciconfig_iobase /* 200 */ .long sys_ni_syscall /* 201 - reserved - MacOnLinux - new */ .long sys_getdents64 /* 202 */ + .long sys_pivot_root /* 203 */ + .long sys_fcntl64 /* 204 */ + .long sys_madvise /* 205 */ + .long sys_mincore /* 206 */ .rept NR_syscalls-(.-sys_call_table)/4 .long sys_ni_syscall .endr diff --git a/arch/ppc/kernel/open_pic.c b/arch/ppc/kernel/open_pic.c index 5ca365b1c..0cbd4f553 100644 --- a/arch/ppc/kernel/open_pic.c +++ b/arch/ppc/kernel/open_pic.c @@ -1,5 +1,5 @@ /* - * arch/ppc/kernel/openpic.c -- OpenPIC Interrupt Handling + * arch/ppc/kernel/open_pic.c -- OpenPIC Interrupt Handling * * Copyright (C) 1997 Geert Uytterhoeven * @@ -13,35 +13,38 @@ #include <linux/kernel.h> #include <linux/sched.h> #include <linux/init.h> -#include <linux/openpic.h> +#include <linux/irq.h> #include <asm/ptrace.h> #include <asm/signal.h> #include <asm/io.h> #include <asm/irq.h> #include <asm/prom.h> + #include "local_irq.h" +#include "open_pic.h" +#include "open_pic_defs.h" -volatile struct OpenPIC *OpenPIC = NULL; +void* OpenPIC_Addr; +static volatile struct OpenPIC *OpenPIC = NULL; u_int OpenPIC_NumInitSenses __initdata = 0; u_char *OpenPIC_InitSenses __initdata = NULL; -int open_pic_irq_offset; extern int use_of_interrupt_tree; -void chrp_mask_irq(unsigned int); -void chrp_unmask_irq(unsigned int); void find_ISUs(void); static u_int NumProcessors; static u_int NumSources; -OpenPIC_Source *ISU; -/* - * We should use this if we have > 1 ISU. - * We can just point each entry to the - * appropriate source regs but it wastes a lot of space - * so until we have >1 ISU I'll leave it unimplemented. - * -- Cort -OpenPIC_Source ISU[128]; -*/ +#ifdef CONFIG_POWER3 +static int NumISUs; +#endif +static int open_pic_irq_offset; +static volatile unsigned char* chrp_int_ack_special; + +OpenPIC_SourcePtr ISU[OPENPIC_MAX_ISU]; + +static void openpic_end_irq(unsigned int irq_nr); +static void openpic_ack_irq(unsigned int irq_nr); +static void openpic_set_affinity(unsigned int irq_nr, unsigned long cpumask); struct hw_interrupt_type open_pic = { " OpenPIC ", @@ -49,45 +52,69 @@ struct hw_interrupt_type open_pic = { NULL, openpic_enable_irq, openpic_disable_irq, - 0, + openpic_ack_irq, + openpic_end_irq, + openpic_set_affinity +}; + +#ifdef CONFIG_SMP +static void openpic_end_ipi(unsigned int irq_nr); +static void openpic_ack_ipi(unsigned int irq_nr); +static void openpic_enable_ipi(unsigned int irq_nr); +static void openpic_disable_ipi(unsigned int irq_nr); + +struct hw_interrupt_type open_pic_ipi = { + " OpenPIC ", + NULL, + NULL, + openpic_enable_ipi, + openpic_disable_ipi, + openpic_ack_ipi, + openpic_end_ipi, 0 }; +#endif /* CONFIG_SMP */ /* - * Accesses to the current processor's registers + * Accesses to the current processor's openpic registers */ -#ifndef __powerpc__ -#define THIS_CPU Private -#define CHECK_THIS_CPU do {} while (0) -#else +#ifdef CONFIG_SMP #define THIS_CPU Processor[cpu] +#define DECL_THIS_CPU int cpu = smp_hw_index[smp_processor_id()] #define CHECK_THIS_CPU check_arg_cpu(cpu) -#endif +#else +#define THIS_CPU Processor[0] +#define DECL_THIS_CPU +#define CHECK_THIS_CPU +#endif /* CONFIG_SMP */ #if 1 #define check_arg_ipi(ipi) \ if (ipi < 0 || ipi >= OPENPIC_NUM_IPI) \ - printk("openpic.c:%d: illegal ipi %d\n", __LINE__, ipi); + printk("open_pic.c:%d: illegal ipi %d\n", __LINE__, ipi); #define check_arg_timer(timer) \ if (timer < 0 || timer >= OPENPIC_NUM_TIMERS) \ - printk("openpic.c:%d: illegal timer %d\n", __LINE__, timer); + printk("open_pic.c:%d: illegal timer %d\n", __LINE__, timer); #define check_arg_vec(vec) \ if (vec < 0 || vec >= OPENPIC_NUM_VECTORS) \ - printk("openpic.c:%d: illegal vector %d\n", __LINE__, vec); + printk("open_pic.c:%d: illegal vector %d\n", __LINE__, vec); #define check_arg_pri(pri) \ if (pri < 0 || pri >= OPENPIC_NUM_PRI) \ - printk("openpic.c:%d: illegal priority %d\n", __LINE__, pri); + printk("open_pic.c:%d: illegal priority %d\n", __LINE__, pri); /* - * I changed this to return to keep us from from trying to use irq #'s - * that we're using for IPI's. - * -- Cort - */ + * Print out a backtrace if it's out of range, since if it's larger than NR_IRQ's + * data has probably been corrupted and we're going to panic or deadlock later + * anyway --Troy + */ +extern unsigned long* _get_SP(void); #define check_arg_irq(irq) \ - /*if (irq < 0 || irq >= (NumSources+open_pic_irq_offset)) \ - printk("openpic.c:%d: illegal irq %d\n", __LINE__, irq);*/ + if (irq < open_pic_irq_offset || irq >= (NumSources+open_pic_irq_offset)){ \ + printk("open_pic.c:%d: illegal irq %d\n", __LINE__, irq); \ + print_backtrace(_get_SP()); } #define check_arg_cpu(cpu) \ - if (cpu < 0 || cpu >= NumProcessors) \ - printk("openpic.c:%d: illegal cpu %d\n", __LINE__, cpu); + if (cpu < 0 || cpu >= NumProcessors){ \ + printk("open_pic.c:%d: illegal cpu %d\n", __LINE__, cpu); \ + print_backtrace(_get_SP()); } #else #define check_arg_ipi(ipi) do {} while (0) #define check_arg_timer(timer) do {} while (0) @@ -97,23 +124,10 @@ struct hw_interrupt_type open_pic = { #define check_arg_cpu(cpu) do {} while (0) #endif -#ifdef CONFIG_SMP -void openpic_ipi_action(int cpl, void *dev_id, struct pt_regs *regs) -{ - smp_message_recv(cpl-OPENPIC_VEC_IPI, regs); -} -#endif /* CONFIG_SMP */ - -#ifdef __i386__ -static inline u_int in_le32(volatile u_int *addr) -{ - return *addr; -} - -static inline void out_le32(volatile u_int *addr, u_int val) -{ - *addr = val; -} +#ifdef CONFIG_POWER3 + #define GET_ISU(source) ISU[(source) >> 4][(source) & 0xf] +#else + #define GET_ISU(source) ISU[0][(source)] #endif u_int openpic_read(volatile u_int *addr) @@ -156,22 +170,66 @@ static void openpic_safe_writefield(volatile u_int *addr, u_int mask, u_int field) { openpic_setfield(addr, OPENPIC_MASK); - /* wait until it's not in use */ - /* BenH: Is this code really enough ? I would rather check the result - * and eventually retry ... - */ while (openpic_read(addr) & OPENPIC_ACTIVITY); openpic_writefield(addr, mask | OPENPIC_MASK, field | OPENPIC_MASK); } -void __init openpic_init(int main_pic) +#ifdef CONFIG_SMP +/* yes this is right ... bug, feature, you decide! -- tgall */ +u_int openpic_read_IPI(volatile u_int* addr) +{ + u_int val = 0; +#ifdef CONFIG_POWER3 + val = in_be32(addr); +#else + val = in_le32(addr); +#endif + return val; +} + +/* because of the power3 be / le above, this is needed */ +inline void openpic_writefield_IPI(volatile u_int* addr, u_int mask, u_int field) +{ + u_int val = openpic_read_IPI(addr); + openpic_write(addr, (val & ~mask) | (field & mask)); +} + +static inline void openpic_clearfield_IPI(volatile u_int *addr, u_int mask) +{ + openpic_writefield_IPI(addr, mask, 0); +} + +static inline void openpic_setfield_IPI(volatile u_int *addr, u_int mask) +{ + openpic_writefield_IPI(addr, mask, mask); +} + +static void openpic_safe_writefield_IPI(volatile u_int *addr, u_int mask, u_int field) +{ + openpic_setfield_IPI(addr, OPENPIC_MASK); + + /* wait until it's not in use */ + /* BenH: Is this code really enough ? I would rather check the result + * and eventually retry ... + */ + while(openpic_read_IPI(addr) & OPENPIC_ACTIVITY); + + openpic_writefield_IPI(addr, mask | OPENPIC_MASK, field | OPENPIC_MASK); +} +#endif /* CONFIG_SMP */ + +void __init openpic_init(int main_pic, int offset, unsigned char* chrp_ack, + int programmer_switch_irq) { u_int t, i; u_int timerfreq; const char *version; - if (!OpenPIC) - panic("No OpenPIC found"); + if (!OpenPIC_Addr) { + printk("No OpenPIC found !\n"); + return; + } + OpenPIC = (volatile struct OpenPIC *)OpenPIC_Addr; if ( ppc_md.progress ) ppc_md.progress("openpic enter",0x122); @@ -194,179 +252,180 @@ void __init openpic_init(int main_pic) OPENPIC_FEATURE_LAST_PROCESSOR_SHIFT) + 1; NumSources = ((t & OPENPIC_FEATURE_LAST_SOURCE_MASK) >> OPENPIC_FEATURE_LAST_SOURCE_SHIFT) + 1; - if ( _machine != _MACH_Pmac ) - { - printk("OpenPIC Version %s (%d CPUs and %d IRQ sources) at %p\n", version, - NumProcessors, NumSources, OpenPIC); - timerfreq = openpic_read(&OpenPIC->Global.Timer_Frequency); - printk("OpenPIC timer frequency is "); - if (timerfreq) - printk("%d MHz\n", timerfreq>>20); - else - printk("not set\n"); + printk("OpenPIC Version %s (%d CPUs and %d IRQ sources) at %p\n", + version, NumProcessors, NumSources, OpenPIC); + timerfreq = openpic_read(&OpenPIC->Global.Timer_Frequency); + if (timerfreq) + printk("OpenPIC timer frequency is %d.%06d MHz\n", + timerfreq / 1000000, timerfreq % 1000000); + + if (!main_pic) + return; + + open_pic_irq_offset = offset; + chrp_int_ack_special = (volatile unsigned char*)chrp_ack; + + /* Initialize timer interrupts */ + if ( ppc_md.progress ) ppc_md.progress("openpic timer",0x3ba); + for (i = 0; i < OPENPIC_NUM_TIMERS; i++) { + /* Disabled, Priority 0 */ + openpic_inittimer(i, 0, OPENPIC_VEC_TIMER+i+offset); + /* No processor */ + openpic_maptimer(i, 0); } - - if ( main_pic ) - { - /* Initialize timer interrupts */ - if ( ppc_md.progress ) ppc_md.progress("openpic timer",0x3ba); - for (i = 0; i < OPENPIC_NUM_TIMERS; i++) { - /* Disabled, Priority 0 */ - openpic_inittimer(i, 0, OPENPIC_VEC_TIMER+i); - /* No processor */ - openpic_maptimer(i, 0); - } - - /* Initialize IPI interrupts */ - if ( ppc_md.progress ) ppc_md.progress("openpic ipi",0x3bb); - for (i = 0; i < OPENPIC_NUM_IPI; i++) { - /* Disabled, Priority 8 */ - openpic_initipi(i, 8, OPENPIC_VEC_IPI+i); - } - find_ISUs(); - if ( _machine != _MACH_Pmac ) - { - /* Initialize external interrupts */ - if ( ppc_md.progress ) ppc_md.progress("openpic ext",0x3bc); - /* SIOint (8259 cascade) is special */ - openpic_initirq(0, 8, open_pic_irq_offset, 1, 1); - openpic_mapirq(0, 1<<0); - for (i = 1; i < NumSources; i++) { - /* Enabled, Priority 8 */ - openpic_initirq(i, 8, open_pic_irq_offset+i, 0, - i < OpenPIC_NumInitSenses ? OpenPIC_InitSenses[i] : 1); - /* Processor 0 */ - openpic_mapirq(i, 1<<0); - } - } - else - { - /* Prevent any interrupt from occuring during initialisation. - * Hum... I believe this is not necessary, Apple does that in - * Darwin's PowerExpress code. - */ - openpic_set_priority(0, 0xf); - - /* First disable all interrupts and map them to CPU 0 */ - for (i = 0; i < NumSources; i++) { - openpic_disable_irq(i); - openpic_mapirq(i, 1<<0); - } - - /* If we use the device tree, then lookup all interrupts and - * initialize them according to sense infos found in the tree - */ - if (use_of_interrupt_tree) { - struct device_node* np = find_all_nodes(); - while(np) { - int j, pri; - pri = strcmp(np->name, "programmer-switch") ? 2 : 7; - for (j=0;j<np->n_intrs;j++) { - openpic_initirq(np->intrs[j].line, - pri, - np->intrs[j].line, - 0, - np->intrs[j].sense); - if (np->intrs[j].sense) - irq_desc[np->intrs[j].line].status = IRQ_LEVEL; - } - np = np->next; - } - } - } - - /* Initialize the spurious interrupt */ - if ( ppc_md.progress ) ppc_md.progress("openpic spurious",0x3bd); - openpic_set_spurious(OPENPIC_VEC_SPURIOUS); - if ( !(_machine & (_MACH_gemini|_MACH_Pmac)) ) - { - if (request_irq(IRQ_8259_CASCADE, no_action, SA_INTERRUPT, - "82c59 cascade", NULL)) - printk("Unable to get OpenPIC IRQ 0 for cascade\n"); - } - openpic_set_priority(0, 0); - openpic_disable_8259_pass_through(); + +#ifdef CONFIG_SMP + /* Initialize IPI interrupts */ + if ( ppc_md.progress ) ppc_md.progress("openpic ipi",0x3bb); + for (i = 0; i < OPENPIC_NUM_IPI; i++) { + /* Disabled, Priority 10..13 */ + openpic_initipi(i, 10+i, OPENPIC_VEC_IPI+i+offset); + /* IPIs are per-CPU */ + irq_desc[OPENPIC_VEC_IPI+i+offset].status |= IRQ_PER_CPU; + irq_desc[OPENPIC_VEC_IPI+i+offset].handler = &open_pic_ipi; + } +#endif + + find_ISUs(); + + /* Initialize external interrupts */ + if (ppc_md.progress) ppc_md.progress("openpic ext",0x3bc); + + openpic_set_priority(0xf); + + /* SIOint (8259 cascade) is special */ + if (offset) { + openpic_initirq(0, 8, offset, 1, 1); + openpic_mapirq(0, 1<<0); + } + + /* Init all external sources */ + for (i = 1; i < NumSources; i++) { + int pri, sense; + + /* the bootloader may have left it enabled (bad !) */ + openpic_disable_irq(i+offset); + + pri = (i == programmer_switch_irq)? 9: 8; + sense = (i < OpenPIC_NumInitSenses)? OpenPIC_InitSenses[i]: 1; + if (sense) + irq_desc[i+offset].status = IRQ_LEVEL; + + /* Enabled, Priority 8 or 9 */ + openpic_initirq(i, pri, i+offset, !sense, sense); + /* Processor 0 */ + openpic_mapirq(i, 1<<0); } - if ( ppc_md.progress ) ppc_md.progress("openpic exit",0x222); + + /* Init descriptors */ + for (i = offset; i < NumSources + offset; i++) + irq_desc[i].handler = &open_pic; + + /* Initialize the spurious interrupt */ + if (ppc_md.progress) ppc_md.progress("openpic spurious",0x3bd); + openpic_set_spurious(OPENPIC_VEC_SPURIOUS+offset); + + /* Initialize the cascade */ + if (offset) { + if (request_irq(offset, no_action, SA_INTERRUPT, + "82c59 cascade", NULL)) + printk("Unable to get OpenPIC IRQ 0 for cascade\n"); + } + openpic_set_priority(0); + openpic_disable_8259_pass_through(); + + if (ppc_md.progress) ppc_md.progress("openpic exit",0x222); +} + +#ifdef CONFIG_POWER3 +void openpic_setup_ISU(int isu_num, unsigned long addr) +{ + if (isu_num >= OPENPIC_MAX_ISU) + return; + ISU[isu_num] = (OpenPIC_SourcePtr) ioremap(addr, 0x400); + if (isu_num >= NumISUs) + NumISUs = isu_num + 1; } +#endif void find_ISUs(void) { -#ifdef CONFIG_PPC64BRIDGE - /* hardcode this for now since the IBM 260 is the only thing with - * a distributed openpic right now. -- Cort +#ifdef CONFIG_POWER3 + /* Use /interrupt-controller/reg and + * /interrupt-controller/interrupt-ranges from OF device tree + * the ISU array is setup in chrp_pci.c in ibm_add_bridges + * as a result + * -- tgall + */ + + /* basically each ISU is a bus, and this assumes that + * open_pic_isu_count interrupts per bus are possible + * ISU == Interrupt Source */ - ISU = (OpenPIC_Source *)0xfeff7c00; - NumSources = 0x10; + NumSources = NumISUs * 0x10; + #else /* for non-distributed OpenPIC implementations it's in the IDU -- Cort */ - ISU = (OpenPIC_Source *)OpenPIC->Source; + ISU[0] = (OpenPIC_Source *)OpenPIC->Source; #endif } -void openpic_reset(void) +static inline void openpic_reset(void) { openpic_setfield(&OpenPIC->Global.Global_Configuration0, OPENPIC_CONFIG_RESET); } -void openpic_enable_8259_pass_through(void) +static inline void openpic_enable_8259_pass_through(void) { openpic_clearfield(&OpenPIC->Global.Global_Configuration0, OPENPIC_CONFIG_8259_PASSTHROUGH_DISABLE); } -void openpic_disable_8259_pass_through(void) +static void openpic_disable_8259_pass_through(void) { openpic_setfield(&OpenPIC->Global.Global_Configuration0, OPENPIC_CONFIG_8259_PASSTHROUGH_DISABLE); } -#ifndef __i386__ /* * Find out the current interrupt */ -u_int openpic_irq(u_int cpu) +static u_int openpic_irq(void) { u_int vec; + DECL_THIS_CPU; - check_arg_cpu(cpu); + CHECK_THIS_CPU; vec = openpic_readfield(&OpenPIC->THIS_CPU.Interrupt_Acknowledge, OPENPIC_VECTOR_MASK); return vec; } -#endif -#ifndef __powerpc__ -void openpic_eoi(void) -#else -void openpic_eoi(u_int cpu) -#endif +static void openpic_eoi(void) { - check_arg_cpu(cpu); + DECL_THIS_CPU; + + CHECK_THIS_CPU; openpic_write(&OpenPIC->THIS_CPU.EOI, 0); /* Handle PCI write posting */ (void)openpic_read(&OpenPIC->THIS_CPU.EOI); } -#ifndef __powerpc__ -u_int openpic_get_priority(void) -#else -u_int openpic_get_priority(u_int cpu) -#endif +static inline u_int openpic_get_priority(void) { + DECL_THIS_CPU; + CHECK_THIS_CPU; return openpic_readfield(&OpenPIC->THIS_CPU.Current_Task_Priority, OPENPIC_CURRENT_TASK_PRIORITY_MASK); } -#ifndef __powerpc__ -void openpic_set_priority(u_int pri) -#else -void openpic_set_priority(u_int cpu, u_int pri) -#endif +static void openpic_set_priority(u_int pri) { + DECL_THIS_CPU; + CHECK_THIS_CPU; check_arg_pri(pri); openpic_writefield(&OpenPIC->THIS_CPU.Current_Task_Priority, @@ -376,24 +435,43 @@ void openpic_set_priority(u_int cpu, u_int pri) /* * Get/set the spurious vector */ -u_int openpic_get_spurious(void) +static inline u_int openpic_get_spurious(void) { return openpic_readfield(&OpenPIC->Global.Spurious_Vector, OPENPIC_VECTOR_MASK); } -void openpic_set_spurious(u_int vec) +static void openpic_set_spurious(u_int vec) { check_arg_vec(vec); openpic_writefield(&OpenPIC->Global.Spurious_Vector, OPENPIC_VECTOR_MASK, vec); } +#ifdef CONFIG_SMP +/* + * Convert a cpu mask from logical to physical cpu numbers. + */ +static inline u32 physmask(u32 cpumask) +{ + int i; + u32 mask = 0; + + for (i = 0; i < smp_num_cpus; ++i, cpumask >>= 1) + mask |= (cpumask & 1) << smp_hw_index[i]; + return mask; +} +#else +#define physmask(cpumask) (cpumask) +#endif + void openpic_init_processor(u_int cpumask) { - openpic_write(&OpenPIC->Global.Processor_Initialization, cpumask); + openpic_write(&OpenPIC->Global.Processor_Initialization, + physmask(cpumask)); } +#ifdef CONFIG_SMP /* * Initialize an interprocessor interrupt (and disable it) * @@ -401,35 +479,55 @@ void openpic_init_processor(u_int cpumask) * pri: interrupt source priority * vec: the vector it will produce */ -void openpic_initipi(u_int ipi, u_int pri, u_int vec) +static void __init openpic_initipi(u_int ipi, u_int pri, u_int vec) { - check_arg_timer(ipi); + check_arg_ipi(ipi); check_arg_pri(pri); check_arg_vec(vec); - openpic_safe_writefield(&OpenPIC->Global.IPI_Vector_Priority(ipi), + openpic_safe_writefield_IPI(&OpenPIC->Global.IPI_Vector_Priority(ipi), OPENPIC_PRIORITY_MASK | OPENPIC_VECTOR_MASK, (pri << OPENPIC_PRIORITY_SHIFT) | vec); } /* * Send an IPI to one or more CPUs + * + * Externally called, however, it takes an IPI number (0...OPENPIC_NUM_IPI) + * and not a system-wide interrupt number */ -#ifndef __powerpc__ void openpic_cause_IPI(u_int ipi, u_int cpumask) -#else -void openpic_cause_IPI(u_int cpu, u_int ipi, u_int cpumask) -#endif { + DECL_THIS_CPU; + CHECK_THIS_CPU; check_arg_ipi(ipi); - openpic_write(&OpenPIC->THIS_CPU.IPI_Dispatch(ipi), cpumask); + openpic_write(&OpenPIC->THIS_CPU.IPI_Dispatch(ipi), + physmask(cpumask)); } -void openpic_enable_IPI(u_int ipi) +void openpic_request_IPIs(void) { - check_arg_ipi(ipi); - openpic_clearfield(&OpenPIC->Global.IPI_Vector_Priority(ipi), - OPENPIC_MASK); + int i; + + /* + * Make sure this matches what is defined in smp.c for + * smp_message_{pass|recv}() or what shows up in + * /proc/interrupts will be wrong!!! --Troy */ + + if (OpenPIC == NULL) + return; + + request_irq(OPENPIC_VEC_IPI+open_pic_irq_offset, + openpic_ipi_action, 0, "IPI0 (call function)", 0); + request_irq(OPENPIC_VEC_IPI+open_pic_irq_offset+1, + openpic_ipi_action, 0, "IPI1 (reschedule)", 0); + request_irq(OPENPIC_VEC_IPI+open_pic_irq_offset+2, + openpic_ipi_action, 0, "IPI2 (invalidate tlb)", 0); + request_irq(OPENPIC_VEC_IPI+open_pic_irq_offset+3, + openpic_ipi_action, 0, "IPI3 (xmon break)", 0); + + for ( i = 0; i < OPENPIC_NUM_IPI ; i++ ) + openpic_enable_ipi(OPENPIC_VEC_IPI+open_pic_irq_offset+i); } /* @@ -437,21 +535,31 @@ void openpic_enable_IPI(u_int ipi) * * Get IPI's working and start taking interrupts. * -- Cort - */ -void do_openpic_setup_cpu(void) + */ +static spinlock_t openpic_setup_lock __initdata = SPIN_LOCK_UNLOCKED; + +void __init do_openpic_setup_cpu(void) { int i; - - for ( i = 0; i < OPENPIC_NUM_IPI ; i++ ) - openpic_enable_IPI(i); -#if 0 - /* let the openpic know we want intrs */ - for ( i = 0; i < NumSources ; i++ ) - openpic_mapirq(i, openpic_read(ISU[i].Destination) - | (1<<smp_processor_id()) ); -#endif - openpic_set_priority(smp_processor_id(), 0); + u32 msk = 1 << smp_hw_index[smp_processor_id()]; + + spin_lock(&openpic_setup_lock); + +#ifdef CONFIG_IRQ_ALL_CPUS + /* let the openpic know we want intrs. default affinity + * is 0xffffffff until changed via /proc + * That's how it's done on x86. If we want it differently, then + * we should make sure we also change the default values of irq_affinity + * in irq.c. + */ + for (i = 0; i < NumSources ; i++) + openpic_mapirq(i, openpic_read(&GET_ISU(i).Destination) | msk); +#endif /* CONFIG_IRQ_ALL_CPUS */ + openpic_set_priority(0); + + spin_unlock(&openpic_setup_lock); } +#endif /* CONFIG_SMP */ /* * Initialize a timer interrupt (and disable it) @@ -460,7 +568,7 @@ void do_openpic_setup_cpu(void) * pri: interrupt source priority * vec: the vector it will produce */ -void openpic_inittimer(u_int timer, u_int pri, u_int vec) +static void __init openpic_inittimer(u_int timer, u_int pri, u_int vec) { check_arg_timer(timer); check_arg_pri(pri); @@ -473,37 +581,99 @@ void openpic_inittimer(u_int timer, u_int pri, u_int vec) /* * Map a timer interrupt to one or more CPUs */ -void openpic_maptimer(u_int timer, u_int cpumask) +static void __init openpic_maptimer(u_int timer, u_int cpumask) { check_arg_timer(timer); - openpic_write(&OpenPIC->Global.Timer[timer].Destination, cpumask); + openpic_write(&OpenPIC->Global.Timer[timer].Destination, + physmask(cpumask)); } + +/* + * + * All functions below take an offset'ed irq argument + * + */ + + /* - * Enable/disable an interrupt source + * Enable/disable an external interrupt source + * + * Externally called, irq is an offseted system-wide interrupt number */ -void openpic_enable_irq(u_int irq) +static void openpic_enable_irq(u_int irq) { check_arg_irq(irq); - openpic_clearfield(&ISU[irq - open_pic_irq_offset].Vector_Priority, OPENPIC_MASK); + + /* + * Never want to disable a timer or ipi irq + * (only want to disable irqs within an ISU). + */ + if (((irq >= OPENPIC_VEC_IPI+open_pic_irq_offset) && + (irq < OPENPIC_VEC_IPI+open_pic_irq_offset+OPENPIC_NUM_IPI)) || + ((irq >= OPENPIC_VEC_TIMER+open_pic_irq_offset) && + (irq < OPENPIC_VEC_TIMER+open_pic_irq_offset+OPENPIC_NUM_TIMERS))) + { + /* silently ignore the enable of the timer or ipi irq. */ + return; + } + + + openpic_clearfield(&GET_ISU(irq - open_pic_irq_offset).Vector_Priority, OPENPIC_MASK); /* make sure mask gets to controller before we return to user */ do { mb(); /* sync is probably useless here */ - } while(openpic_readfield(&ISU[irq - open_pic_irq_offset].Vector_Priority, + } while(openpic_readfield(&GET_ISU(irq - open_pic_irq_offset).Vector_Priority, OPENPIC_MASK)); } -void openpic_disable_irq(u_int irq) +static void openpic_disable_irq(u_int irq) { + u32 vp; + check_arg_irq(irq); - openpic_setfield(&ISU[irq - open_pic_irq_offset].Vector_Priority, OPENPIC_MASK); + /* + * Never want to disable a timer or ipi irq + * (only want to disable irqs within an ISU). + */ + if (((irq >= OPENPIC_VEC_IPI+open_pic_irq_offset) && + (irq < OPENPIC_VEC_IPI+open_pic_irq_offset+OPENPIC_NUM_IPI)) || + ((irq >= OPENPIC_VEC_TIMER+open_pic_irq_offset) && + (irq < OPENPIC_VEC_TIMER+open_pic_irq_offset+OPENPIC_NUM_TIMERS))) + { + panic("openpic_disable_irq - disabling non-ISU irq"); + } + + + openpic_setfield(&GET_ISU(irq - open_pic_irq_offset).Vector_Priority, OPENPIC_MASK); /* make sure mask gets to controller before we return to user */ do { mb(); /* sync is probably useless here */ - } while(!openpic_readfield(&ISU[irq - open_pic_irq_offset].Vector_Priority, - OPENPIC_MASK)); + vp = openpic_readfield(&GET_ISU(irq - open_pic_irq_offset).Vector_Priority, + OPENPIC_MASK | OPENPIC_ACTIVITY); + } while((vp & OPENPIC_ACTIVITY) && !(vp & OPENPIC_MASK)); } +#ifdef CONFIG_SMP +/* + * Enable/disable an IPI interrupt source + * + * Externally called, irq is an offseted system-wide interrupt number + */ +void openpic_enable_ipi(u_int irq) +{ + irq -= (OPENPIC_VEC_IPI+open_pic_irq_offset); + check_arg_ipi(irq); + openpic_clearfield_IPI(&OpenPIC->Global.IPI_Vector_Priority(irq), OPENPIC_MASK); + +} +void openpic_disable_ipi(u_int irq) +{ + /* NEVER disable an IPI... that's just plain wrong! */ +} + +#endif + /* * Initialize an interrupt source (and disable it!) * @@ -513,12 +683,9 @@ void openpic_disable_irq(u_int irq) * pol: polarity (1 for positive, 0 for negative) * sense: 1 for level, 0 for edge */ -void openpic_initirq(u_int irq, u_int pri, u_int vec, int pol, int sense) +static void openpic_initirq(u_int irq, u_int pri, u_int vec, int pol, int sense) { - check_arg_irq(irq); - check_arg_pri(pri); - check_arg_vec(vec); - openpic_safe_writefield(&ISU[irq].Vector_Priority, + openpic_safe_writefield(&GET_ISU(irq).Vector_Priority, OPENPIC_PRIORITY_MASK | OPENPIC_VECTOR_MASK | OPENPIC_SENSE_MASK | OPENPIC_POLARITY_MASK, (pri << OPENPIC_PRIORITY_SHIFT) | vec | @@ -530,10 +697,9 @@ void openpic_initirq(u_int irq, u_int pri, u_int vec, int pol, int sense) /* * Map an interrupt source to one or more CPUs */ -void openpic_mapirq(u_int irq, u_int cpumask) +static void openpic_mapirq(u_int irq, u_int physmask) { - check_arg_irq(irq); - openpic_write(&ISU[irq].Destination, cpumask); + openpic_write(&GET_ISU(irq).Destination, physmask); } /* @@ -541,10 +707,92 @@ void openpic_mapirq(u_int irq, u_int cpumask) * * sense: 1 for level, 0 for edge */ -void openpic_set_sense(u_int irq, int sense) +static inline void openpic_set_sense(u_int irq, int sense) { - check_arg_irq(irq); - openpic_safe_writefield(&ISU[irq].Vector_Priority, + openpic_safe_writefield(&GET_ISU(irq).Vector_Priority, OPENPIC_SENSE_LEVEL, (sense ? OPENPIC_SENSE_LEVEL : 0)); } + +/* No spinlocks, should not be necessary with the OpenPIC + * (1 register = 1 interrupt and we have the desc lock). + */ +static void openpic_ack_irq(unsigned int irq_nr) +{ +#if 1 /* masking should be unnecessary, but I still get spurrious */ + openpic_disable_irq(irq_nr); +#endif + if ((irq_desc[irq_nr].status & IRQ_LEVEL) == 0) + openpic_eoi(); +} + +static void openpic_end_irq(unsigned int irq_nr) +{ + if ((irq_desc[irq_nr].status & IRQ_LEVEL) != 0) + openpic_eoi(); + +#if 1 /* masking should be unnecessary, but I still get spurrious */ + if (!(irq_desc[irq_nr].status & (IRQ_DISABLED|IRQ_INPROGRESS))) + openpic_enable_irq(irq_nr); +#endif +} + +static void openpic_set_affinity(unsigned int irq_nr, unsigned long cpumask) +{ + openpic_mapirq(irq_nr - open_pic_irq_offset, physmask(cpumask)); +} + +#ifdef CONFIG_SMP +static void openpic_ack_ipi(unsigned int irq_nr) +{ +} + +static void openpic_end_ipi(unsigned int irq_nr) +{ + /* IPIs are marked IRQ_PER_CPU. This has the side effect of + * preventing the IRQ_PENDING/IRQ_INPROGRESS logic from + * applying to them. We EOI them late to avoid re-entering. + * however, I'm wondering if we could simply let them have the + * SA_INTERRUPT flag and let them execute with all interrupts OFF. + * This would have the side effect of either running cross-CPU + * functions with interrupts off, or we can re-enable them explicitely + * with a __sti() in smp_call_function_interrupt(), since + * smp_call_function() is protected by a spinlock. + * Or maybe we shouldn't set the IRQ_PER_CPU flag on cross-CPU + * function calls IPI at all but that would make a special case. + */ + openpic_eoi(); +} + +static void openpic_ipi_action(int cpl, void *dev_id, struct pt_regs *regs) +{ + smp_message_recv(cpl-OPENPIC_VEC_IPI-open_pic_irq_offset, regs); +} + +#endif /* CONFIG_SMP */ + +/* This one may be merged with PReP and CHRP */ +int +openpic_get_irq(struct pt_regs *regs) +{ + extern int i8259_irq(int cpu); + + int irq = openpic_irq(); + + /* Management of the cascade should be moved out of here */ + if (open_pic_irq_offset && irq == open_pic_irq_offset) + { + /* + * This magic address generates a PCI IACK cycle. + */ + if ( chrp_int_ack_special ) + irq = *chrp_int_ack_special; + else + irq = i8259_irq( smp_processor_id() ); + openpic_eoi(); + } + if (irq == OPENPIC_VEC_SPURIOUS + open_pic_irq_offset) + irq = -1; + return irq; +} + diff --git a/arch/ppc/kernel/open_pic.h b/arch/ppc/kernel/open_pic.h index 3e51ffba3..cf924ba3a 100644 --- a/arch/ppc/kernel/open_pic.h +++ b/arch/ppc/kernel/open_pic.h @@ -1,11 +1,45 @@ +/* + * arch/ppc/kernel/open_pic.h -- OpenPIC Interrupt Handling + * + * Copyright (C) 1997 Geert Uytterhoeven + * + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file COPYING in the main directory of this archive + * for more details. + * + */ + #ifndef _PPC_KERNEL_OPEN_PIC_H #define _PPC_KERNEL_OPEN_PIC_H +#include <linux/config.h> + +#define OPENPIC_SIZE 0x40000 + +/* OpenPIC IRQ controller structure */ extern struct hw_interrupt_type open_pic; -void openpic_ipi_action(int cpl, void *dev_id, struct pt_regs *regs); -void openpic_enable_IPI(u_int ipi); -void do_openpic_setup_cpu(void); +/* OpenPIC IPI controller structure */ +#ifdef CONFIG_SMP +extern struct hw_interrupt_type open_pic_ipi; +#endif /* CONFIG_SMP */ + +extern u_int OpenPIC_NumInitSenses; +extern u_char *OpenPIC_InitSenses; +extern void* OpenPIC_Addr; + +/* Exported functions */ +extern void openpic_init(int, int, unsigned char *, int); +extern void openpic_request_IPIs(void); +extern void do_openpic_setup_cpu(void); +extern int openpic_get_irq(struct pt_regs *regs); +extern void openpic_init_processor(u_int cpumask); +extern void openpic_setup_ISU(int isu_num, unsigned long addr); +extern void openpic_cause_IPI(u_int ipi, u_int cpumask); -extern int open_pic_irq_offset; +extern inline int openpic_to_irq(int irq) +{ + return irq += NUM_8259_INTERRUPTS; +} +/*extern int open_pic_irq_offset;*/ #endif /* _PPC_KERNEL_OPEN_PIC_H */ diff --git a/arch/ppc/kernel/open_pic_defs.h b/arch/ppc/kernel/open_pic_defs.h new file mode 100644 index 000000000..fed4dd1cc --- /dev/null +++ b/arch/ppc/kernel/open_pic_defs.h @@ -0,0 +1,328 @@ +/* + * linux/openpic.h -- OpenPIC definitions + * + * Copyright (C) 1997 Geert Uytterhoeven + * + * This file is based on the following documentation: + * + * The Open Programmable Interrupt Controller (PIC) + * Register Interface Specification Revision 1.2 + * + * Issue Date: October 1995 + * + * Issued jointly by Advanced Micro Devices and Cyrix Corporation + * + * AMD is a registered trademark of Advanced Micro Devices, Inc. + * Copyright (C) 1995, Advanced Micro Devices, Inc. and Cyrix, Inc. + * All Rights Reserved. + * + * To receive a copy of this documentation, send an email to openpic@amd.com. + * + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file COPYING in the main directory of this archive + * for more details. + */ + +#ifndef _LINUX_OPENPIC_H +#define _LINUX_OPENPIC_H + +#ifdef __KERNEL__ + +#include <linux/config.h> + + /* + * OpenPIC supports up to 2048 interrupt sources and up to 32 processors + */ + +#define OPENPIC_MAX_SOURCES 2048 +#define OPENPIC_MAX_PROCESSORS 32 +#define OPENPIC_MAX_ISU 16 + +#define OPENPIC_NUM_TIMERS 4 +#define OPENPIC_NUM_IPI 4 +#define OPENPIC_NUM_PRI 16 +#define OPENPIC_NUM_VECTORS 256 + + + /* + * Non-offset'ed vector numbers + */ + +#define OPENPIC_VEC_TIMER 64 /* and up */ +#define OPENPIC_VEC_IPI 72 /* and up */ +#define OPENPIC_VEC_SPURIOUS 127 + + + /* + * OpenPIC Registers are 32 bits and aligned on 128 bit boundaries + */ + +typedef struct _OpenPIC_Reg { + u_int Reg; /* Little endian! */ + char Pad[0xc]; +} OpenPIC_Reg; + + + /* + * Per Processor Registers + */ + +typedef struct _OpenPIC_Processor { + /* + * Private Shadow Registers (for SLiC backwards compatibility) + */ + u_int IPI0_Dispatch_Shadow; /* Write Only */ + char Pad1[0x4]; + u_int IPI0_Vector_Priority_Shadow; /* Read/Write */ + char Pad2[0x34]; + /* + * Interprocessor Interrupt Command Ports + */ + OpenPIC_Reg _IPI_Dispatch[OPENPIC_NUM_IPI]; /* Write Only */ + /* + * Current Task Priority Register + */ + OpenPIC_Reg _Current_Task_Priority; /* Read/Write */ + char Pad3[0x10]; + /* + * Interrupt Acknowledge Register + */ + OpenPIC_Reg _Interrupt_Acknowledge; /* Read Only */ + /* + * End of Interrupt (EOI) Register + */ + OpenPIC_Reg _EOI; /* Read/Write */ + char Pad5[0xf40]; +} OpenPIC_Processor; + + + /* + * Timer Registers + */ + +typedef struct _OpenPIC_Timer { + OpenPIC_Reg _Current_Count; /* Read Only */ + OpenPIC_Reg _Base_Count; /* Read/Write */ + OpenPIC_Reg _Vector_Priority; /* Read/Write */ + OpenPIC_Reg _Destination; /* Read/Write */ +} OpenPIC_Timer; + + + /* + * Global Registers + */ + +typedef struct _OpenPIC_Global { + /* + * Feature Reporting Registers + */ + OpenPIC_Reg _Feature_Reporting0; /* Read Only */ + OpenPIC_Reg _Feature_Reporting1; /* Future Expansion */ + /* + * Global Configuration Registers + */ + OpenPIC_Reg _Global_Configuration0; /* Read/Write */ + OpenPIC_Reg _Global_Configuration1; /* Future Expansion */ + /* + * Vendor Specific Registers + */ + OpenPIC_Reg _Vendor_Specific[4]; + /* + * Vendor Identification Register + */ + OpenPIC_Reg _Vendor_Identification; /* Read Only */ + /* + * Processor Initialization Register + */ + OpenPIC_Reg _Processor_Initialization; /* Read/Write */ + /* + * IPI Vector/Priority Registers + */ + OpenPIC_Reg _IPI_Vector_Priority[OPENPIC_NUM_IPI]; /* Read/Write */ + /* + * Spurious Vector Register + */ + OpenPIC_Reg _Spurious_Vector; /* Read/Write */ + /* + * Global Timer Registers + */ + OpenPIC_Reg _Timer_Frequency; /* Read/Write */ + OpenPIC_Timer Timer[OPENPIC_NUM_TIMERS]; + char Pad1[0xee00]; +} OpenPIC_Global; + + + /* + * Interrupt Source Registers + */ + +typedef struct _OpenPIC_Source { + OpenPIC_Reg _Vector_Priority; /* Read/Write */ + OpenPIC_Reg _Destination; /* Read/Write */ +} OpenPIC_Source, *OpenPIC_SourcePtr; + + + /* + * OpenPIC Register Map + */ + +struct OpenPIC { + char Pad1[0x1000]; + /* + * Global Registers + */ + OpenPIC_Global Global; + /* + * Interrupt Source Configuration Registers + */ + OpenPIC_Source Source[OPENPIC_MAX_SOURCES]; + /* + * Per Processor Registers + */ + OpenPIC_Processor Processor[OPENPIC_MAX_PROCESSORS]; +}; + +extern volatile struct OpenPIC *OpenPIC; + + + /* + * Current Task Priority Register + */ + +#define OPENPIC_CURRENT_TASK_PRIORITY_MASK 0x0000000f + + /* + * Who Am I Register + */ + +#define OPENPIC_WHO_AM_I_ID_MASK 0x0000001f + + /* + * Feature Reporting Register 0 + */ + +#define OPENPIC_FEATURE_LAST_SOURCE_MASK 0x07ff0000 +#define OPENPIC_FEATURE_LAST_SOURCE_SHIFT 16 +#define OPENPIC_FEATURE_LAST_PROCESSOR_MASK 0x00001f00 +#define OPENPIC_FEATURE_LAST_PROCESSOR_SHIFT 8 +#define OPENPIC_FEATURE_VERSION_MASK 0x000000ff + + /* + * Global Configuration Register 0 + */ + +#define OPENPIC_CONFIG_RESET 0x80000000 +#define OPENPIC_CONFIG_8259_PASSTHROUGH_DISABLE 0x20000000 +#define OPENPIC_CONFIG_BASE_MASK 0x000fffff + + /* + * Vendor Identification Register + */ + +#define OPENPIC_VENDOR_ID_STEPPING_MASK 0x00ff0000 +#define OPENPIC_VENDOR_ID_STEPPING_SHIFT 16 +#define OPENPIC_VENDOR_ID_DEVICE_ID_MASK 0x0000ff00 +#define OPENPIC_VENDOR_ID_DEVICE_ID_SHIFT 8 +#define OPENPIC_VENDOR_ID_VENDOR_ID_MASK 0x000000ff + + /* + * Vector/Priority Registers + */ + +#define OPENPIC_MASK 0x80000000 +#define OPENPIC_ACTIVITY 0x40000000 /* Read Only */ +#define OPENPIC_PRIORITY_MASK 0x000f0000 +#define OPENPIC_PRIORITY_SHIFT 16 +#define OPENPIC_VECTOR_MASK 0x000000ff + + + /* + * Interrupt Source Registers + */ + +#define OPENPIC_POLARITY_POSITIVE 0x00800000 +#define OPENPIC_POLARITY_NEGATIVE 0x00000000 +#define OPENPIC_POLARITY_MASK 0x00800000 +#define OPENPIC_SENSE_LEVEL 0x00400000 +#define OPENPIC_SENSE_EDGE 0x00000000 +#define OPENPIC_SENSE_MASK 0x00400000 + + + /* + * Timer Registers + */ + +#define OPENPIC_COUNT_MASK 0x7fffffff +#define OPENPIC_TIMER_TOGGLE 0x80000000 +#define OPENPIC_TIMER_COUNT_INHIBIT 0x80000000 + + + /* + * Aliases to make life simpler + */ + +/* Per Processor Registers */ +#define IPI_Dispatch(i) _IPI_Dispatch[i].Reg +#define Current_Task_Priority _Current_Task_Priority.Reg +#define Interrupt_Acknowledge _Interrupt_Acknowledge.Reg +#define EOI _EOI.Reg + +/* Global Registers */ +#define Feature_Reporting0 _Feature_Reporting0.Reg +#define Feature_Reporting1 _Feature_Reporting1.Reg +#define Global_Configuration0 _Global_Configuration0.Reg +#define Global_Configuration1 _Global_Configuration1.Reg +#define Vendor_Specific(i) _Vendor_Specific[i].Reg +#define Vendor_Identification _Vendor_Identification.Reg +#define Processor_Initialization _Processor_Initialization.Reg +#define IPI_Vector_Priority(i) _IPI_Vector_Priority[i].Reg +#define Spurious_Vector _Spurious_Vector.Reg +#define Timer_Frequency _Timer_Frequency.Reg + +/* Timer Registers */ +#define Current_Count _Current_Count.Reg +#define Base_Count _Base_Count.Reg +#define Vector_Priority _Vector_Priority.Reg +#define Destination _Destination.Reg + +/* Interrupt Source Registers */ +#define Vector_Priority _Vector_Priority.Reg +#define Destination _Destination.Reg + + /* + * Local (static) OpenPIC Operations + */ + + +/* Global Operations */ +static void openpic_reset(void); +static void openpic_enable_8259_pass_through(void); +static void openpic_disable_8259_pass_through(void); +static u_int openpic_irq(void); +static void openpic_eoi(void); +static u_int openpic_get_priority(void); +static void openpic_set_priority(u_int pri); +static u_int openpic_get_spurious(void); +static void openpic_set_spurious(u_int vector); + +#ifdef CONFIG_SMP +/* Interprocessor Interrupts */ +static void openpic_initipi(u_int ipi, u_int pri, u_int vector); +static void openpic_ipi_action(int cpl, void *dev_id, struct pt_regs *regs); +#endif + +/* Timer Interrupts */ +static void openpic_inittimer(u_int timer, u_int pri, u_int vector); +static void openpic_maptimer(u_int timer, u_int cpumask); + +/* Interrupt Sources */ +static void openpic_enable_irq(u_int irq); +static void openpic_disable_irq(u_int irq); +static void openpic_initirq(u_int irq, u_int pri, u_int vector, int polarity, + int is_level); +static void openpic_mapirq(u_int irq, u_int cpumask); +static void openpic_set_sense(u_int irq, int sense); + +#endif /* __KERNEL__ */ + +#endif /* _LINUX_OPENPIC_H */ diff --git a/arch/ppc/kernel/pci.c b/arch/ppc/kernel/pci.c index 6d6f775f2..1e503fae4 100644 --- a/arch/ppc/kernel/pci.c +++ b/arch/ppc/kernel/pci.c @@ -3,15 +3,16 @@ * Common pmac/prep/chrp pci routines. -- Cort */ +#include <linux/config.h> #include <linux/kernel.h> #include <linux/pci.h> #include <linux/delay.h> #include <linux/string.h> #include <linux/init.h> -#include <linux/openpic.h> #include <linux/capability.h> #include <linux/sched.h> #include <linux/errno.h> +#include <linux/bootmem.h> #include <asm/processor.h> #include <asm/io.h> @@ -25,7 +26,7 @@ #include "pci.h" -#undef DEBUG +#define DEBUG #ifdef DEBUG #define DBG(x...) printk(x) @@ -37,54 +38,44 @@ unsigned long isa_io_base = 0; unsigned long isa_mem_base = 0; unsigned long pci_dram_offset = 0; -struct pci_fixup pcibios_fixups[] = { - { 0 } -}; +static u8* pci_to_OF_bus_map; -int generic_pcibios_read_byte(struct pci_dev *dev, int where, u8 *val) -{ - return ppc_md.pcibios_read_config_byte(dev->bus->number,dev->devfn,where,val); -} -int generic_pcibios_read_word(struct pci_dev *dev, int where, u16 *val) -{ - return ppc_md.pcibios_read_config_word(dev->bus->number,dev->devfn,where,val); -} -int generic_pcibios_read_dword(struct pci_dev *dev, int where, u32 *val) -{ - return ppc_md.pcibios_read_config_dword(dev->bus->number,dev->devfn,where,val); -} -int generic_pcibios_write_byte(struct pci_dev *dev, int where, u8 val) -{ - return ppc_md.pcibios_write_config_byte(dev->bus->number,dev->devfn,where,val); -} -int generic_pcibios_write_word(struct pci_dev *dev, int where, u16 val) -{ - return ppc_md.pcibios_write_config_word(dev->bus->number,dev->devfn,where,val); -} -int generic_pcibios_write_dword(struct pci_dev *dev, int where, u32 val) -{ - return ppc_md.pcibios_write_config_dword(dev->bus->number,dev->devfn,where,val); -} +static void pcibios_fixup_resources(struct pci_dev* dev); +#ifdef CONFIG_ALL_PPC +static void pcibios_fixup_cardbus(struct pci_dev* dev); +#endif -struct pci_ops generic_pci_ops = -{ - generic_pcibios_read_byte, - generic_pcibios_read_word, - generic_pcibios_read_dword, - generic_pcibios_write_byte, - generic_pcibios_write_word, - generic_pcibios_write_dword -}; +/* By default, we don't re-assign bus numbers. We do this only on + * some pmacs + */ +int pci_assign_all_busses; + +struct pci_controller* hose_head; +struct pci_controller** hose_tail = &hose_head; +static int pci_bus_count; +struct pci_fixup pcibios_fixups[] = { + { PCI_FIXUP_HEADER, PCI_ANY_ID, PCI_ANY_ID, pcibios_fixup_resources }, +#ifdef CONFIG_ALL_PPC + /* We should add per-machine fixup support in xxx_setup.c or xxx_pci.c */ + { PCI_FIXUP_FINAL, PCI_VENDOR_ID_TI, PCI_DEVICE_ID_TI_1211, pcibios_fixup_cardbus }, +#endif /* CONFIG_ALL_PPC */ + { 0 } +}; -void pcibios_update_resource(struct pci_dev *dev, struct resource *root, +void +pcibios_update_resource(struct pci_dev *dev, struct resource *root, struct resource *res, int resource) { u32 new, check; int reg; - - new = res->start | (res->flags & PCI_REGION_FLAG_MASK); + struct pci_controller* hose = dev->sysdata; + + new = res->start; + if (hose && res->flags & IORESOURCE_MEM) + new -= hose->pci_mem_offset; + new |= (res->flags & PCI_REGION_FLAG_MASK); if (resource < 6) { reg = PCI_BASE_ADDRESS_0 + 4*resource; } else if (resource == PCI_ROM_RESOURCE) { @@ -104,6 +95,62 @@ void pcibios_update_resource(struct pci_dev *dev, struct resource *root, } } +static void +pcibios_fixup_resources(struct pci_dev* dev) +{ + struct pci_controller* hose = + (struct pci_controller *)dev->sysdata; + int i; + if (!hose) { + printk("No hose for PCI dev %x.%x !\n", dev->bus->number, dev->devfn >> 3); + return; + } + for (i = 0; i < DEVICE_COUNT_RESOURCE; i++) { + struct resource *res = dev->resource + i; + if (!res->start) + continue; + if (res->flags & IORESOURCE_MEM) { + res->start += hose->pci_mem_offset; + res->end += hose->pci_mem_offset; +#ifdef DEBUG + printk("Fixup mem res, dev: %x.%x, res_start: %lx->%lx\n", + dev->bus->number, dev->devfn>>3, res->start-hose->pci_mem_offset, + res->start); +#endif + } + + if ((res->flags & IORESOURCE_IO) + && (unsigned long) hose->io_base_virt != isa_io_base) { + unsigned long offs = (unsigned long) hose->io_base_virt - isa_io_base; + res->start += offs; + res->end += offs; + printk("Fixup IO res, dev: %x.%x, res_start: %lx->%lx\n", + dev->bus->number, dev->devfn>>3, + res->start - offs, res->start); + } + } +} + +#ifdef CONFIG_ALL_PPC +static void +pcibios_fixup_cardbus(struct pci_dev* dev) +{ + /* + * Fix the interrupt routing on the TI1211 chip on the 1999 + * G3 powerbook, which doesn't get initialized properly by OF. + */ + if (dev->vendor == PCI_VENDOR_ID_TI + && dev->device == PCI_DEVICE_ID_TI_1211) { + u32 val; + /* 0x8c == TI122X_IRQMUX, 2 says to route the INTA + signal out the MFUNC0 pin */ + if (pci_read_config_dword(dev, 0x8c, &val) == 0 + && val == 0) + pci_write_config_dword(dev, 0x8c, 2); + } +} +#endif /* CONFIG_ALL_PPC */ + /* * We need to avoid collisions with `mirrored' VGA ports * and other strange ISA hardware, so we always want the @@ -172,7 +219,8 @@ pcibios_align_resource(void *data, struct resource *res, unsigned long size) * as well. */ -static void __init pcibios_allocate_bus_resources(struct list_head *bus_list) +static void __init +pcibios_allocate_bus_resources(struct list_head *bus_list) { struct list_head *ln; struct pci_bus *bus; @@ -197,7 +245,8 @@ static void __init pcibios_allocate_bus_resources(struct list_head *bus_list) } } -static void __init pcibios_allocate_resources(int pass) +static void __init +pcibios_allocate_resources(int pass) { struct pci_dev *dev; int idx, disabled; @@ -250,7 +299,8 @@ static void __init pcibios_allocate_resources(int pass) } } -static void __init pcibios_assign_resources(void) +static void __init +pcibios_assign_resources(void) { struct pci_dev *dev; int idx; @@ -278,7 +328,9 @@ static void __init pcibios_assign_resources(void) * the BIOS forgot to do so or because we have decided the old * address was unusable for some reason. */ - if (!r->start && r->end) + if (!r->start && r->end && + (!ppc_md.pcibios_enable_device_hook || + !ppc_md.pcibios_enable_device_hook(dev, 1))) pci_assign_resource(dev, idx); } @@ -293,7 +345,8 @@ static void __init pcibios_assign_resources(void) } -int pcibios_enable_resources(struct pci_dev *dev) +int +pcibios_enable_resources(struct pci_dev *dev) { u16 cmd, old_cmd; int idx; @@ -321,18 +374,267 @@ int pcibios_enable_resources(struct pci_dev *dev) return 0; } +struct pci_controller * __init +pcibios_alloc_controller(void) +{ + struct pci_controller *hose; + + hose = (struct pci_controller *)alloc_bootmem(sizeof(*hose)); + memset(hose, 0, sizeof(struct pci_controller)); + + *hose_tail = hose; + hose_tail = &hose->next; + + return hose; +} +static void +make_one_node_map(struct device_node* node, u8 pci_bus) +{ + int *bus_range; + int len; + + if (pci_bus >= pci_bus_count) + return; + bus_range = (int *) get_property(node, "bus-range", &len); + if (bus_range == NULL || len < 2 * sizeof(int)) { + printk(KERN_WARNING "Can't get bus-range for %s\n", + node->full_name); + return; + } + pci_to_OF_bus_map[pci_bus] = bus_range[0]; + + for (node=node->child; node != 0;node = node->sibling) { + struct pci_dev* dev; + unsigned int *class_code, *reg; + + class_code = (unsigned int *) get_property(node, "class-code", 0); + if (!class_code || ((*class_code >> 8) != PCI_CLASS_BRIDGE_PCI && + (*class_code >> 8) != PCI_CLASS_BRIDGE_CARDBUS)) + continue; + reg = (unsigned int *)get_property(node, "reg", 0); + if (!reg) + continue; + dev = pci_find_slot(pci_bus, ((reg[0] >> 8) & 0xff)); + if (!dev || !dev->subordinate) + continue; + make_one_node_map(node, dev->subordinate->number); + } +} + +void +pcibios_make_OF_bus_map(void) +{ + int i; + struct pci_controller* hose; + u8* of_prop_map; + + pci_to_OF_bus_map = (u8*)kmalloc(pci_bus_count, GFP_KERNEL); + if (!pci_to_OF_bus_map) { + printk(KERN_ERR "Can't allocate OF bus map !\n"); + return; + } + + /* We fill the bus map with invalid values, that helps + * debugging. + */ + for (i=0; i<pci_bus_count; i++) + pci_to_OF_bus_map[i] = 0xff; + + /* For each hose, we begin searching bridges */ + for(hose=hose_head; hose; hose=hose->next) { + struct device_node* node; + node = (struct device_node *)hose->arch_data; + if (!node) + continue; + make_one_node_map(node, hose->first_busno); + } + of_prop_map = get_property(find_path_device("/"), "pci-OF-bus-map", 0); + if (of_prop_map) + memcpy(of_prop_map, pci_to_OF_bus_map, pci_bus_count); +#ifdef DEBUG + printk("PCI->OF bus map:\n"); + for (i=0; i<pci_bus_count; i++) { + if (pci_to_OF_bus_map[i] == 0xff) + continue; + printk("%d -> %d\n", i, pci_to_OF_bus_map[i]); + } +#endif +} -void __init pcibios_init(void) +static struct device_node* +scan_OF_childs_for_device(struct device_node* node, u8 bus, u8 dev_fn) { + struct device_node* sub_node; + + for (; node != 0;node = node->sibling) { + unsigned int *class_code, *reg; + + reg = (unsigned int *) get_property(node, "reg", 0); + if (reg && ((reg[0] >> 8) & 0xff) == dev_fn + && ((reg[0] >> 16) & 0xff) == bus) + return node; + + /* For PCI<->PCI bridges or CardBus bridges, we go down */ + class_code = (unsigned int *) get_property(node, "class-code", 0); + if (!class_code || ((*class_code >> 8) != PCI_CLASS_BRIDGE_PCI && + (*class_code >> 8) != PCI_CLASS_BRIDGE_CARDBUS)) + continue; + sub_node = scan_OF_childs_for_device(node->child, bus, dev_fn); + if (sub_node) + return sub_node; + } + return NULL; +} + +/* + * Scans the OF tree for a device node matching a PCI device + */ +struct device_node* +pci_device_to_OF_node(struct pci_dev *dev) +{ + struct pci_controller *hose; + struct device_node *node; + int bus; + + if (!have_of) + return NULL; + + /* Lookup the hose */ + bus = dev->bus->number; + hose = pci_bus_to_hose(bus); + if (!hose) + return NULL; + + /* Check it has an OF node associated */ + node = (struct device_node *) hose->arch_data; + if (!node) + return NULL; + + /* Fixup bus number according to what OF think it is. */ + if (pci_to_OF_bus_map) + bus = pci_to_OF_bus_map[bus]; + if (bus == 0xff) + return NULL; + + /* Now, lookup childs of the hose */ + return scan_OF_childs_for_device(node->child, bus, dev->devfn); +} + +/* This routine is meant to be used early during boot, when the + * PCI bus numbers have not yet been assigned, and you need to + * issue PCI config cycles to an OF device. + * It could also be used to "fix" RTAS config cycles if you want + * to set pci_assign_all_busses to 1 and still use RTAS for PCI + * config cycles. + */ +struct pci_controller* +pci_find_hose_for_OF_device(struct device_node* node) +{ + if (!have_of) + return NULL; + while(node) { + struct pci_controller* hose; + for (hose=hose_head;hose;hose=hose->next) + if (hose->arch_data == node) + return hose; + node=node->parent; + } + return NULL; +} + +/* + * Returns the PCI device matching a given OF node + */ +int +pci_device_from_OF_node(struct device_node* node, u8* bus, u8* devfn) +{ + unsigned int *reg; + int i; + + if (!have_of) + return -ENODEV; + reg = (unsigned int *) get_property(node, "reg", 0); + if (!reg) + return -ENODEV; + *bus = (reg[0] >> 16) & 0xff; + for (i=0; pci_to_OF_bus_map && i<pci_bus_count; i++) + if (pci_to_OF_bus_map[i] == *bus) { + *bus = i; + break; + } + *devfn = ((reg[0] >> 8) & 0xff); + return 0; +} + +void __init +pcibios_init(void) +{ + struct pci_controller *hose; + struct pci_bus *bus; + int next_busno; + printk("PCI: Probing PCI hardware\n"); - pci_scan_bus(0, &generic_pci_ops, NULL); + + /* Scan all of the recorded PCI controllers. */ + for (next_busno = 0, hose = hose_head; hose; hose = hose->next) { + if (pci_assign_all_busses) + hose->first_busno = next_busno; + hose->last_busno = 0xff; + bus = pci_scan_bus(hose->first_busno, hose->ops, hose); + hose->bus = bus; + hose->last_busno = bus->subordinate; + if (pci_assign_all_busses || next_busno <= hose->last_busno) + next_busno = hose->last_busno+1; + } + pci_bus_count = next_busno; + + /* OpenFirmware based machines need a map of OF bus + * numbers vs. kernel bus numbers since we may have to + * remap them. + */ + if (pci_assign_all_busses && have_of) + pcibios_make_OF_bus_map(); + + /* Call machine dependant fixup */ if (ppc_md.pcibios_fixup) ppc_md.pcibios_fixup(); + + /* Allocate and assign resources */ pcibios_allocate_bus_resources(&pci_root_buses); pcibios_allocate_resources(0); pcibios_allocate_resources(1); pcibios_assign_resources(); + +#ifdef CONFIG_BLK_DEV_IDE + /* OF fails to initialize IDE controllers on macs + * (and maybe other machines) + * + * This late fixup is done here since I want it to happen after + * resource assignement, and there's no "late-init" arch hook + * + * Ideally, this should be moved to the IDE layer, but we need + * to check specifically with Andre Hedrick how to do it cleanly + * since the common IDE code seem to care about the fact that the + * BIOS may have disabled a controller. + * + * -- BenH + */ + if (_machine == _MACH_Pmac) { + struct pci_dev *dev; + pci_for_each_dev(dev) + { + if ((dev->class >> 16) == PCI_BASE_CLASS_STORAGE) + pci_enable_device(dev); + } + } +#endif /* CONFIG_BLK_DEV_IDE */ +} + +int __init +pcibios_assign_all_busses(void) +{ + return pci_assign_all_busses; } void __init @@ -344,9 +646,16 @@ pcibios_fixup_pbus_ranges(struct pci_bus * bus, struct pbus_set_ranges_data * ra ranges->mem_end -= bus->resource[1]->start; } +unsigned long resource_fixup(struct pci_dev * dev, struct resource * res, + unsigned long start, unsigned long size) +{ + return start; +} void __init pcibios_fixup_bus(struct pci_bus *bus) { + pci_read_bridge_bases(bus); + if ( ppc_md.pcibios_fixup_bus ) ppc_md.pcibios_fixup_bus(bus); } @@ -370,6 +679,10 @@ int pcibios_enable_device(struct pci_dev *dev) int idx; struct resource *r; + if (ppc_md.pcibios_enable_device_hook) + if (ppc_md.pcibios_enable_device_hook(dev, 0)) + return -EINVAL; + pci_read_config_word(dev, PCI_COMMAND, &cmd); old_cmd = cmd; for (idx=0; idx<6; idx++) { @@ -391,37 +704,99 @@ int pcibios_enable_device(struct pci_dev *dev) return 0; } -void * -pci_dev_io_base(unsigned char bus, unsigned char devfn, int physical) +struct pci_controller* +pci_bus_to_hose(int bus) { - if (!ppc_md.pci_dev_io_base) { - /* Please, someone fix this for non-pmac machines, we - * need either the virtual or physical PCI IO base - */ - return 0; - } - return ppc_md.pci_dev_io_base(bus, devfn, physical); + struct pci_controller* hose = hose_head; + + for (; hose; hose = hose->next) + if (bus >= hose->first_busno && bus <= hose->last_busno) + return hose; + return NULL; } -void * -pci_dev_mem_base(unsigned char bus, unsigned char devfn) +void* +pci_bus_io_base(unsigned int bus) { - /* Default memory base is 0 (1:1 mapping) */ - if (!ppc_md.pci_dev_mem_base) { - /* Please, someone fix this for non-pmac machines.*/ + struct pci_controller *hose; + + hose = pci_bus_to_hose(bus); + if (!hose) + return NULL; + return hose->io_base_virt; +} + +unsigned long +pci_bus_io_base_phys(unsigned int bus) +{ + struct pci_controller *hose; + + hose = pci_bus_to_hose(bus); + if (!hose) return 0; - } - return ppc_md.pci_dev_mem_base(bus, devfn); + return hose->io_base_phys; } -/* Returns the root-bridge number (Uni-N number) of a device */ -int -pci_dev_root_bridge(unsigned char bus, unsigned char devfn) +unsigned long +pci_bus_mem_base_phys(unsigned int bus) { - /* Defaults to 0 */ - if (!ppc_md.pci_dev_root_bridge) + struct pci_controller *hose; + + hose = pci_bus_to_hose(bus); + if (!hose) return 0; - return ppc_md.pci_dev_root_bridge(bus, devfn); + return hose->pci_mem_offset; +} + +#ifdef CONFIG_POWER4 +extern unsigned long pci_address_offset(int, unsigned int); +#endif /* CONFIG_POWER4 */ + +unsigned long +pci_resource_to_bus(struct pci_dev *pdev, struct resource *res) +{ + /* Hack alert again ! See comments in chrp_pci.c + */ +#ifdef CONFIG_POWER4 + unsigned long offset = pci_address_offset(pdev->bus->number, res->flags); + return res->start - offset; +#else /* CONFIG_POWER4 */ + struct pci_controller* hose = + (struct pci_controller *)pdev->sysdata; + if (hose && res->flags & IORESOURCE_MEM) + return res->start - hose->pci_mem_offset; + /* We may want to do something with IOs here... */ + return res->start; +#endif +} + +/* Obsolete functions. Should be removed once the symbios driver + * is fixed + */ +unsigned long +pci_phys_to_bus(unsigned long pa, int busnr) +{ +#ifdef CONFIG_POWER4 + return pa - pci_address_offset(busnr, IORESOURCE_MEM); +#else /* CONFIG_POWER4 */ + struct pci_controller* hose = pci_bus_to_hose(busnr); + if (!hose) + return pa; + return pa - hose->pci_mem_offset; +#endif +} + +unsigned long +pci_bus_to_phys(unsigned int ba, int busnr) +{ +#ifdef CONFIG_POWER4 + return ba + pci_address_offset(dev->bus->number, IORESOURCE_MEM); +#else /* CONFIG_POWER4 */ + struct pci_controller* hose = pci_bus_to_hose(busnr); + if (!hose) + return ba; + return ba + hose->pci_mem_offset; +#endif } /* Provide information on locations of various I/O regions in physical @@ -430,23 +805,93 @@ pci_dev_root_bridge(unsigned char bus, unsigned char devfn) * Note that the returned IO or memory base is a physical address */ -asmlinkage long +long sys_pciconfig_iobase(long which, unsigned long bus, unsigned long devfn) { + struct pci_controller* hose = pci_bus_to_hose(bus); long result = -EOPNOTSUPP; + + if (!hose) + return -ENODEV; switch (which) { case IOBASE_BRIDGE_NUMBER: - return (long)pci_dev_root_bridge(bus, devfn); + return (long)hose->first_busno; case IOBASE_MEMORY: - return (long)pci_dev_mem_base(bus, devfn); + return (long)hose->pci_mem_offset; case IOBASE_IO: - result = (long)pci_dev_io_base(bus, devfn, 1); - if (result == 0) - result = -EOPNOTSUPP; - break; + return (long)hose->io_base_phys; + case IOBASE_ISA_IO: + return (long)isa_io_base; + case IOBASE_ISA_MEM: + return (long)isa_mem_base; } return result; } +/* + * Null PCI config access functions, for the case when we can't + * find a hose. + */ +#define NULL_PCI_OP(rw, size, type) \ +static int \ +null_##rw##_config_##size(struct pci_dev *dev, int offset, type val) \ +{ \ + return PCIBIOS_DEVICE_NOT_FOUND; \ +} + +NULL_PCI_OP(read, byte, u8 *) +NULL_PCI_OP(read, word, u16 *) +NULL_PCI_OP(read, dword, u32 *) +NULL_PCI_OP(write, byte, u8) +NULL_PCI_OP(write, word, u16) +NULL_PCI_OP(write, dword, u32) + +static struct pci_ops null_pci_ops = +{ + null_read_config_byte, + null_read_config_word, + null_read_config_dword, + null_write_config_byte, + null_write_config_word, + null_write_config_dword +}; + +/* + * These functions are used early on before PCI scanning is done + * and all of the pci_dev and pci_bus structures have been created. + */ +static struct pci_dev * +fake_pci_dev(struct pci_controller *hose, int busnr, int devfn) +{ + static struct pci_dev dev; + static struct pci_bus bus; + + if (hose == 0) { + hose = pci_bus_to_hose(busnr); + if (hose == 0) + printk(KERN_ERR "Can't find hose for PCI bus %d!\n", busnr); + } + dev.bus = &bus; + dev.sysdata = hose; + dev.devfn = devfn; + bus.number = busnr; + bus.ops = hose? hose->ops: &null_pci_ops; + return &dev; +} + +#define EARLY_PCI_OP(rw, size, type) \ +int early_##rw##_config_##size(struct pci_controller *hose, int bus, \ + int devfn, int offset, type value) \ +{ \ + return pci_##rw##_config_##size(fake_pci_dev(hose, bus, devfn), \ + offset, value); \ +} + +EARLY_PCI_OP(read, byte, u8 *) +EARLY_PCI_OP(read, word, u16 *) +EARLY_PCI_OP(read, dword, u32 *) +EARLY_PCI_OP(write, byte, u8) +EARLY_PCI_OP(write, word, u16) +EARLY_PCI_OP(write, dword, u32) diff --git a/arch/ppc/kernel/pci.h b/arch/ppc/kernel/pci.h index d79eb0f4a..07a64d9f3 100644 --- a/arch/ppc/kernel/pci.h +++ b/arch/ppc/kernel/pci.h @@ -2,47 +2,23 @@ #ifndef __PPC_KERNEL_PCI_H__ #define __PPC_KERNEL_PCI_H__ +/* Configure those in your xxx_init() or xxx_setup_arch() function */ extern unsigned long isa_io_base; extern unsigned long isa_mem_base; extern unsigned long pci_dram_offset; -extern unsigned int *pci_config_address; -extern unsigned char *pci_config_data; +/* Set this to 1 if you want the kernel to re-assign all PCI + * bus numbers + */ +extern int pci_assign_all_busses; -void fix_intr(struct device_node *node, struct pci_dev *dev); -#if 0 -#define decl_config_access_method(name) \ -struct pci_ops name##_pci_ops = { \ - name##_pcibios_read_config_byte, \ - name##_pcibios_read_config_word, \ - name##_pcibios_read_config_dword, \ - name##_pcibios_write_config_byte, \ - name##_pcibios_write_config_word, \ - name##_pcibios_write_config_dword \ -} -#endif +extern struct pci_controller* pcibios_alloc_controller(void); +extern struct pci_controller* pci_find_hose_for_OF_device( + struct device_node* node); -#define decl_config_access_method(name) \ -extern int name##_pcibios_read_config_byte(unsigned char bus, \ - unsigned char dev_fn, unsigned char offset, unsigned char *val); \ -extern int name##_pcibios_read_config_word(unsigned char bus, \ - unsigned char dev_fn, unsigned char offset, unsigned short *val); \ -extern int name##_pcibios_read_config_dword(unsigned char bus, \ - unsigned char dev_fn, unsigned char offset, unsigned int *val); \ -extern int name##_pcibios_write_config_byte(unsigned char bus, \ - unsigned char dev_fn, unsigned char offset, unsigned char val); \ -extern int name##_pcibios_write_config_word(unsigned char bus, \ - unsigned char dev_fn, unsigned char offset, unsigned short val); \ -extern int name##_pcibios_write_config_dword(unsigned char bus, \ - unsigned char dev_fn, unsigned char offset, unsigned int val) - -#define set_config_access_method(name) \ - ppc_md.pcibios_read_config_byte = name##_pcibios_read_config_byte; \ - ppc_md.pcibios_read_config_word = name##_pcibios_read_config_word; \ - ppc_md.pcibios_read_config_dword = name##_pcibios_read_config_dword; \ - ppc_md.pcibios_write_config_byte = name##_pcibios_write_config_byte; \ - ppc_md.pcibios_write_config_word = name##_pcibios_write_config_word; \ - ppc_md.pcibios_write_config_dword = name##_pcibios_write_config_dword +extern void setup_indirect_pci(struct pci_controller* hose, + u32 cfg_addr, u32 cfg_data); +extern void setup_grackle(struct pci_controller *hose, unsigned io_space_size); #endif /* __PPC_KERNEL_PCI_H__ */ diff --git a/arch/ppc/kernel/pmac_pci.c b/arch/ppc/kernel/pmac_pci.c index 8f7b3d7c2..9e1fffb49 100644 --- a/arch/ppc/kernel/pmac_pci.c +++ b/arch/ppc/kernel/pmac_pci.c @@ -27,23 +27,14 @@ #include "pci.h" -struct bridge_data **bridges, *bridge_list; -static int max_bus; - -struct uninorth_data { - struct device_node* node; - volatile unsigned int* cfg_addr; - volatile unsigned int* cfg_data; - void* iobase; - unsigned long iobase_phys; -}; - -static struct uninorth_data uninorth_bridges[3]; -static int uninorth_count; -static int uninorth_default = -1; +#undef DEBUG static void add_bridges(struct device_node *dev); +/* XXX Could be per-controller, but I don't think we risk anything by + * assuming we won't have both UniNorth and Bandit */ +static int has_uninorth; + /* * Magic constants for enabling cache coherency in the bandit/PSX bridge. */ @@ -56,477 +47,201 @@ static void add_bridges(struct device_node *dev); #define BANDIT_MAGIC 0x50 #define BANDIT_COHERENT 0x40 -/* Obsolete, should be replaced by pmac_pci_dev_io_base() (below) */ -__pmac -void *pci_io_base(unsigned int bus) -{ - struct bridge_data *bp; - - if (bus > max_bus || (bp = bridges[bus]) == 0) - return 0; - return bp->io_base; -} - -__pmac -int pci_device_loc(struct device_node *dev, unsigned char *bus_ptr, - unsigned char *devfn_ptr) +static int __init +fixup_one_level_bus_range(struct device_node *node, int higher) { - unsigned int *reg; - int len; + for (; node != 0;node = node->sibling) { + int * bus_range; + unsigned int *class_code; + int len; - reg = (unsigned int *) get_property(dev, "reg", &len); - if (reg == 0 || len < 5 * sizeof(unsigned int)) { - /* doesn't look like a PCI device */ - *bus_ptr = 0xff; - *devfn_ptr = 0xff; - return -1; + /* For PCI<->PCI bridges or CardBus bridges, we go down */ + class_code = (unsigned int *) get_property(node, "class-code", 0); + if (!class_code || ((*class_code >> 8) != PCI_CLASS_BRIDGE_PCI && + (*class_code >> 8) != PCI_CLASS_BRIDGE_CARDBUS)) + continue; + bus_range = (int *) get_property(node, "bus-range", &len); + if (bus_range != NULL && len > 2 * sizeof(int)) { + if (bus_range[1] > higher) + higher = bus_range[1]; + } + higher = fixup_one_level_bus_range(node->child, higher); } - *bus_ptr = reg[0] >> 16; - *devfn_ptr = reg[0] >> 8; - return 0; + return higher; } -/* This routines figures out on which root bridge a given PCI device - * is attached. +/* This routine fixes the "bus-range" property of all bridges in the + * system since they tend to have their "last" member wrong on macs + * + * Note that the bus numbers manipulated here are OF bus numbers, they + * are not Linux bus numbers. */ -__pmac -int -pmac_pci_dev_root_bridge(unsigned char bus, unsigned char dev_fn) +static void __init +fixup_bus_range(struct device_node *bridge) { - struct device_node *node, *bridge_node; - int bridge = uninorth_default; - - if (uninorth_count == 0) - return 0; - if (bus == 0 && PCI_SLOT(dev_fn) < 11) - return 0; + int * bus_range; + int len; - /* We look for the OF device corresponding to this bus/devfn pair. If we - * don't find it, we default to the external PCI */ - bridge_node = NULL; - node = find_pci_device_OFnode(bus, dev_fn & 0xf8); - if (node) { - /* note: we don't stop on the first occurence since we need to go - * up to the root bridge */ - do { - if (node->type && !strcmp(node->type, "pci") - && device_is_compatible(node, "uni-north")) - bridge_node = node; - node=node->parent; - } while (node); - } - if (bridge_node) { - int i; - for (i=0;i<uninorth_count;i++) - if (uninorth_bridges[i].node == bridge_node) { - bridge = i; - break; - } - } - - if (bridge == -1) { - printk(KERN_WARNING "pmac_pci: no default bridge !\n"); - return 0; - } - - return bridge; -} - -__pmac -void * -pmac_pci_dev_io_base(unsigned char bus, unsigned char devfn, int physical) -{ - int bridge = -1; - if (uninorth_count != 0) - bridge = pmac_pci_dev_root_bridge(bus, devfn); - if (bridge == -1) { - struct bridge_data *bp; - - if (bus > max_bus || (bp = bridges[bus]) == 0) - return 0; - return physical ? (void *) bp->io_base_phys : bp->io_base; + /* Lookup the "bus-range" property for the hose */ + bus_range = (int *) get_property(bridge, "bus-range", &len); + if (bus_range == NULL || len < 2 * sizeof(int)) { + printk(KERN_WARNING "Can't get bus-range for %s\n", + bridge->full_name); + return; } - return physical ? (void *) uninorth_bridges[bridge].iobase_phys - : uninorth_bridges[bridge].iobase; -} - -__pmac -void * -pmac_pci_dev_mem_base(unsigned char bus, unsigned char devfn) -{ - return 0; + bus_range[1] = fixup_one_level_bus_range(bridge->child, bus_range[1]); } -/* This function only works for bus 0, uni-N uses a different mecanism for - * other busses (see below) +/* + * Apple MacRISC (UniNorth, Bandit) PCI controllers. + * + * The "Bandit" version is present in all early PCI PowerMacs, + * and up to the first ones using Grackle. Some machines may + * have 2 bandit controllers (2 PCI busses). + * + * The "UniNorth" version is present in all Core99 machines + * (iBook, G4, new IMacs, and all the recent Apple machines). + * It contains 3 controllers in one ASIC. */ -#define UNI_N_CFA0(devfn, off) \ + +#define MACRISC_CFA0(devfn, off) \ ((1 << (unsigned long)PCI_SLOT(dev_fn)) \ | (((unsigned long)PCI_FUNC(dev_fn)) << 8) \ | (((unsigned long)(off)) & 0xFCUL)) -/* This one is for type 1 config accesses */ -#define UNI_N_CFA1(bus, devfn, off) \ +#define MACRISC_CFA1(bus, devfn, off) \ ((((unsigned long)(bus)) << 16) \ |(((unsigned long)(devfn)) << 8) \ |(((unsigned long)(off)) & 0xFCUL) \ |1UL) -__pmac static -unsigned int -uni_north_access_data(unsigned char bus, unsigned char dev_fn, - unsigned char offset) +static unsigned int __pmac +macrisc_cfg_access(struct pci_controller* hose, u8 bus, u8 dev_fn, u8 offset) { - int bridge; unsigned int caddr; - - bridge = pmac_pci_dev_root_bridge(bus, dev_fn); - if (bus == 0) - caddr = UNI_N_CFA0(dev_fn, offset); - else - caddr = UNI_N_CFA1(bus, dev_fn, offset); - - if (bridge == -1) { - printk(KERN_WARNING "pmac_pci: no default bridge !\n"); - return 0; - } - - /* Uninorth will return garbage if we don't read back the value ! */ - out_le32(uninorth_bridges[bridge].cfg_addr, caddr); - (void)in_le32(uninorth_bridges[bridge].cfg_addr); - /* Yes, offset is & 7, not & 3 ! */ - return (unsigned int)(uninorth_bridges[bridge].cfg_data) + (offset & 0x07); -} - -__pmac -int uni_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char *val) -{ - unsigned int addr; - - *val = 0xff; - addr = uni_north_access_data(bus, dev_fn, offset); - if (!addr) - return PCIBIOS_DEVICE_NOT_FOUND; - *val = in_8((volatile unsigned char*)addr); - return PCIBIOS_SUCCESSFUL; -} - -__pmac -int uni_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short *val) -{ - unsigned int addr; - - *val = 0xffff; - addr = uni_north_access_data(bus, dev_fn, offset); - if (!addr) - return PCIBIOS_DEVICE_NOT_FOUND; - *val = in_le16((volatile unsigned short*)addr); - return PCIBIOS_SUCCESSFUL; -} - -__pmac -int uni_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int *val) -{ - unsigned int addr; - *val = 0xffff; - addr = uni_north_access_data(bus, dev_fn, offset); - if (!addr) - return PCIBIOS_DEVICE_NOT_FOUND; - *val = in_le32((volatile unsigned int*)addr); - return PCIBIOS_SUCCESSFUL; -} - -__pmac -int uni_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char val) -{ - unsigned int addr; - - addr = uni_north_access_data(bus, dev_fn, offset); - if (!addr) - return PCIBIOS_DEVICE_NOT_FOUND; - out_8((volatile unsigned char *)addr, val); - (void)in_8((volatile unsigned char *)addr); - return PCIBIOS_SUCCESSFUL; -} - -__pmac -int uni_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short val) -{ - unsigned int addr; - - addr = uni_north_access_data(bus, dev_fn, offset); - if (!addr) - return PCIBIOS_DEVICE_NOT_FOUND; - out_le16((volatile unsigned short *)addr, val); - (void)in_le16((volatile unsigned short *)addr); - return PCIBIOS_SUCCESSFUL; -} - -__pmac -int uni_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int val) -{ - unsigned int addr; - - addr = uni_north_access_data(bus, dev_fn, offset); - if (!addr) - return PCIBIOS_DEVICE_NOT_FOUND; - out_le32((volatile unsigned int *)addr, val); - (void)in_le32((volatile unsigned int *)addr); - return PCIBIOS_SUCCESSFUL; -} - -__pmac -int pmac_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char *val) -{ - struct bridge_data *bp; - - *val = 0xff; - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - if (bus == bp->bus_number) { - if (dev_fn < (11 << 3)) - return PCIBIOS_DEVICE_NOT_FOUND; - out_le32(bp->cfg_addr, - (1UL << (dev_fn >> 3)) + ((dev_fn & 7) << 8) - + (offset & ~3)); - } else { - /* Bus number once again taken into consideration. - * Change applied from 2.1.24. This makes devices located - * behind PCI-PCI bridges visible. - * -Ranjit Deshpande, 01/20/99 - */ - out_le32(bp->cfg_addr, (bus << 16) + (dev_fn << 8) + (offset & ~3) + 1); - } - udelay(2); - *val = in_8(bp->cfg_data + (offset & 3)); - return PCIBIOS_SUCCESSFUL; -} - -__pmac -int pmac_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short *val) -{ - struct bridge_data *bp; - - *val = 0xffff; - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - if ((offset & 1) != 0) - return PCIBIOS_BAD_REGISTER_NUMBER; - if (bus == bp->bus_number) { - if (dev_fn < (11 << 3)) - return PCIBIOS_DEVICE_NOT_FOUND; - out_le32(bp->cfg_addr, - (1UL << (dev_fn >> 3)) + ((dev_fn & 7) << 8) - + (offset & ~3)); - } else { - /* See pci_read_config_byte */ - out_le32(bp->cfg_addr, (bus << 16) + (dev_fn << 8) + (offset & ~3) + 1); - } - udelay(2); - *val = in_le16((volatile unsigned short *)(bp->cfg_data + (offset & 3))); - return PCIBIOS_SUCCESSFUL; -} - -__pmac -int pmac_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int *val) -{ - struct bridge_data *bp; - - *val = 0xffffffff; - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - if ((offset & 3) != 0) - return PCIBIOS_BAD_REGISTER_NUMBER; - if (bus == bp->bus_number) { - if (dev_fn < (11 << 3)) - return PCIBIOS_DEVICE_NOT_FOUND; - out_le32(bp->cfg_addr, - (1UL << (dev_fn >> 3)) + ((dev_fn & 7) << 8) - + offset); - } else { - /* See pci_read_config_byte */ - out_le32(bp->cfg_addr, (bus << 16) + (dev_fn << 8) + offset + 1); - } - udelay(2); - *val = in_le32((volatile unsigned int *)bp->cfg_data); - return PCIBIOS_SUCCESSFUL; -} - -__pmac -int pmac_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char val) -{ - struct bridge_data *bp; - - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - if (bus == bp->bus_number) { +#ifdef DEBUG +// printk("macrisc_config_access(hose: 0x%08lx, bus: 0x%x, devfb: 0x%x, offset: 0x%x)\n", +// hose, bus, dev_fn, offset); +#endif + if (bus == hose->first_busno) { if (dev_fn < (11 << 3)) - return PCIBIOS_DEVICE_NOT_FOUND; - out_le32(bp->cfg_addr, - (1UL << (dev_fn >> 3)) + ((dev_fn & 7) << 8) - + (offset & ~3)); - } else { - /* See pci_read_config_byte */ - out_le32(bp->cfg_addr, (bus << 16) + (dev_fn << 8) + (offset & ~3) + 1); - } - udelay(2); - out_8(bp->cfg_data + (offset & 3), val); - return PCIBIOS_SUCCESSFUL; -} - -__pmac -int pmac_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short val) -{ - struct bridge_data *bp; + return 0; + caddr = MACRISC_CFA0(dev_fn, offset); + } else + caddr = MACRISC_CFA1(bus, dev_fn, offset); + + /* Uninorth will return garbage if we don't read back the value ! */ + do { + out_le32(hose->cfg_addr, caddr); + } while(in_le32(hose->cfg_addr) != caddr); + + offset &= has_uninorth ? 0x07 : 0x03; + return (unsigned int)(hose->cfg_data) + (unsigned int)offset; +} + +#define cfg_read(val, addr, type, op, op2) \ + *val = op((type)(addr)) +#define cfg_write(val, addr, type, op, op2) \ + op((type *)(addr), (val)); (void) op2((type *)(addr)) + +#define cfg_read_bad(val, size) *val = bad_##size; +#define cfg_write_bad(val, size) + +#define bad_byte 0xff +#define bad_word 0xffff +#define bad_dword 0xffffffffU + +#define MACRISC_PCI_OP(rw, size, type, op, op2) \ +static int __pmac \ +macrisc_##rw##_config_##size(struct pci_dev *dev, int off, type val) \ +{ \ + struct pci_controller *hose = dev->sysdata; \ + unsigned int addr; \ + \ + addr = macrisc_cfg_access(hose, dev->bus->number, dev->devfn, off); \ + if (!addr) { \ + cfg_##rw##_bad(val, size) \ + return PCIBIOS_DEVICE_NOT_FOUND; \ + } \ + cfg_##rw(val, addr, type, op, op2); \ + return PCIBIOS_SUCCESSFUL; \ +} + +MACRISC_PCI_OP(read, byte, u8 *, in_8, x) +MACRISC_PCI_OP(read, word, u16 *, in_le16, x) +MACRISC_PCI_OP(read, dword, u32 *, in_le32, x) +MACRISC_PCI_OP(write, byte, u8, out_8, in_8) +MACRISC_PCI_OP(write, word, u16, out_le16, in_le16) +MACRISC_PCI_OP(write, dword, u32, out_le32, in_le32) + +static struct pci_ops macrisc_pci_ops = +{ + macrisc_read_config_byte, + macrisc_read_config_word, + macrisc_read_config_dword, + macrisc_write_config_byte, + macrisc_write_config_word, + macrisc_write_config_dword +}; - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - if ((offset & 1) != 0) - return PCIBIOS_BAD_REGISTER_NUMBER; - if (bus == bp->bus_number) { - if (dev_fn < (11 << 3)) - return PCIBIOS_DEVICE_NOT_FOUND; - out_le32(bp->cfg_addr, - (1UL << (dev_fn >> 3)) + ((dev_fn & 7) << 8) - + (offset & ~3)); - } else { - /* See pci_read_config_byte */ - out_le32(bp->cfg_addr, (bus << 16) + (dev_fn << 8) + (offset & ~3) + 1); - } - udelay(2); - out_le16((volatile unsigned short *)(bp->cfg_data + (offset & 3)), val); - return PCIBIOS_SUCCESSFUL; -} -__pmac -int pmac_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int val) +/* + * Apple "Chaos" PCI controller. + * + * This controller is present on some first generation "PowerSurge" + * machines (8500, 8600, ...). It's a very weird beast and will die + * in flames if we try to probe the config space. + * The long-term solution is to provide a config space "emulation" + * based on what we find in OF device tree + */ + +static int chaos_config_read_byte(struct pci_dev *dev, int offset, u8 *val) { - struct bridge_data *bp; - - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - if ((offset & 3) != 0) - return PCIBIOS_BAD_REGISTER_NUMBER; - if (bus == bp->bus_number) { - if (dev_fn < (11 << 3)) - return PCIBIOS_DEVICE_NOT_FOUND; - out_le32(bp->cfg_addr, - (1UL << (dev_fn >> 3)) + ((dev_fn & 7) << 8) - + offset); - } else { - /* See pci_read_config_byte */ - out_le32(bp->cfg_addr, (bus << 16) + (dev_fn << 8) + (offset & ~3) + 1); - } - udelay(2); - out_le32((volatile unsigned int *)bp->cfg_data, val); - return PCIBIOS_SUCCESSFUL; + return PCIBIOS_DEVICE_NOT_FOUND; } -#define GRACKLE_CFA(b, d, o) (0x80 | ((b) << 8) | ((d) << 16) \ - | (((o) & ~3) << 24)) - -int grackle_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char *val) +static int chaos_config_read_word(struct pci_dev *dev, int offset, u16 *val) { - struct bridge_data *bp; - - *val = 0xff; - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - out_be32(bp->cfg_addr, GRACKLE_CFA(bus, dev_fn, offset)); - *val = in_8(bp->cfg_data + (offset & 3)); - return PCIBIOS_SUCCESSFUL; + return PCIBIOS_DEVICE_NOT_FOUND; } -int grackle_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short *val) +static int chaos_config_read_dword(struct pci_dev *dev, int offset, u32 *val) { - struct bridge_data *bp; - - *val = 0xffff; - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - if ((offset & 1) != 0) - return PCIBIOS_BAD_REGISTER_NUMBER; - out_be32(bp->cfg_addr, GRACKLE_CFA(bus, dev_fn, offset)); - *val = in_le16((volatile unsigned short *)(bp->cfg_data + (offset&3))); - return PCIBIOS_SUCCESSFUL; + return PCIBIOS_DEVICE_NOT_FOUND; } -int grackle_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int *val) +static int chaos_config_write_byte(struct pci_dev *dev, int offset, u8 val) { - struct bridge_data *bp; - - *val = 0xffffffff; - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - if ((offset & 3) != 0) - return PCIBIOS_BAD_REGISTER_NUMBER; - out_be32(bp->cfg_addr, GRACKLE_CFA(bus, dev_fn, offset)); - *val = in_le32((volatile unsigned int *)bp->cfg_data); - return PCIBIOS_SUCCESSFUL; + return PCIBIOS_DEVICE_NOT_FOUND; } -int grackle_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned char val) +static int chaos_config_write_word(struct pci_dev *dev, int offset, u16 val) { - struct bridge_data *bp; - - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - out_be32(bp->cfg_addr, GRACKLE_CFA(bus, dev_fn, offset)); - out_8(bp->cfg_data + (offset & 3), val); - return PCIBIOS_SUCCESSFUL; + return PCIBIOS_DEVICE_NOT_FOUND; } -int grackle_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short val) +static int chaos_config_write_dword(struct pci_dev *dev, int offset, u32 val) { - struct bridge_data *bp; - - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - if ((offset & 1) != 0) - return PCIBIOS_BAD_REGISTER_NUMBER; - out_be32(bp->cfg_addr, GRACKLE_CFA(bus, dev_fn, offset)); - out_le16((volatile unsigned short *)(bp->cfg_data + (offset&3)), val); - return PCIBIOS_SUCCESSFUL; + return PCIBIOS_DEVICE_NOT_FOUND; } -int grackle_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned int val) +static struct pci_ops chaos_pci_ops = { - struct bridge_data *bp; + chaos_config_read_byte, + chaos_config_read_word, + chaos_config_read_dword, + chaos_config_write_byte, + chaos_config_write_word, + chaos_config_write_dword +}; - if (bus > max_bus || (bp = bridges[bus]) == 0) - return PCIBIOS_DEVICE_NOT_FOUND; - if ((offset & 1) != 0) - return PCIBIOS_BAD_REGISTER_NUMBER; - out_be32(bp->cfg_addr, GRACKLE_CFA(bus, dev_fn, offset)); - out_le32((volatile unsigned int *)bp->cfg_data, val); - return PCIBIOS_SUCCESSFUL; -} /* * For a bandit bridge, turn on cache coherency if necessary. - * N.B. we can't use pcibios_*_config_* here because bridges[] - * is not initialized yet. + * N.B. we could clean this up using the hose ops directly. */ -static void __init init_bandit(struct bridge_data *bp) +static void __init init_bandit(struct pci_controller *bp) { unsigned int vendev, magic; int rev; @@ -543,8 +258,8 @@ static void __init init_bandit(struct bridge_data *bp) rev = in_8(bp->cfg_data); if (rev != BANDIT_REVID) printk(KERN_WARNING - "Unknown revision %d for bandit at %p\n", - rev, bp->io_base); + "Unknown revision %d for bandit at %08lx\n", + rev, bp->io_base_phys); } else if (vendev != (BANDIT_DEVID_2 << 16) + APPLE_VENDID) { printk(KERN_WARNING "bandit isn't? (%x)\n", vendev); return; @@ -555,8 +270,8 @@ static void __init init_bandit(struct bridge_data *bp) udelay(2); rev = in_8(bp->cfg_data); if (rev != BANDIT_REVID) - printk(KERN_WARNING "Unknown revision %d for bandit at %p\n", - rev, bp->io_base); + printk(KERN_WARNING "Unknown revision %d for bandit at %08lx\n", + rev, bp->io_base_phys); /* read the word at offset 0x50 */ out_le32(bp->cfg_addr, (1UL << BANDIT_DEVNUM) + BANDIT_MAGIC); @@ -567,16 +282,71 @@ static void __init init_bandit(struct bridge_data *bp) magic |= BANDIT_COHERENT; udelay(2); out_le32((volatile unsigned int *)bp->cfg_data, magic); - printk(KERN_INFO "Cache coherency enabled for bandit/PSX at %p\n", - bp->io_base); + printk(KERN_INFO "Cache coherency enabled for bandit/PSX at %08lx\n", + bp->io_base_phys); +} + + +/* + * Tweak the PCI-PCI bridge chip on the blue & white G3s. + */ +static void __init +init_p2pbridge(void) +{ + struct device_node *p2pbridge; + struct pci_controller* hose; + u8 bus, devfn; + u16 val; + + /* XXX it would be better here to identify the specific + PCI-PCI bridge chip we have. */ + if ((p2pbridge = find_devices("pci-bridge")) == 0 + || p2pbridge->parent == NULL + || strcmp(p2pbridge->parent->name, "pci") != 0) + return; + if (pci_device_from_OF_node(p2pbridge, &bus, &devfn) < 0) { +#ifdef DEBUG + printk("Can't find PCI infos for PCI<->PCI bridge\n"); +#endif + return; + } + /* Warning: At this point, we have not yet renumbered all busses. + * So we must use OF walking to find out hose + */ + hose = pci_find_hose_for_OF_device(p2pbridge); + if (!hose) { +#ifdef DEBUG + printk("Can't find hose for PCI<->PCI bridge\n"); +#endif + return; + } + if (early_read_config_word(hose, bus, devfn, + PCI_BRIDGE_CONTROL, &val) < 0) { + printk(KERN_ERR "init_p2pbridge: couldn't read bridge control\n"); + return; + } + val &= ~PCI_BRIDGE_CTL_MASTER_ABORT; + early_write_config_word(hose, bus, devfn, PCI_BRIDGE_CONTROL, val); } +void __init +pmac_find_bridges(void) +{ + add_bridges(find_devices("bandit")); + add_bridges(find_devices("chaos")); + add_bridges(find_devices("pci")); + init_p2pbridge(); +} + +#define GRACKLE_CFA(b, d, o) (0x80 | ((b) << 8) | ((d) << 16) \ + | (((o) & ~3) << 24)) + #define GRACKLE_PICR1_STG 0x00000040 #define GRACKLE_PICR1_LOOPSNOOP 0x00000010 /* N.B. this is called before bridges is initialized, so we can't use grackle_pcibios_{read,write}_config_dword. */ -static inline void grackle_set_stg(struct bridge_data *bp, int enable) +static inline void grackle_set_stg(struct pci_controller* bp, int enable) { unsigned int val; @@ -586,9 +356,10 @@ static inline void grackle_set_stg(struct bridge_data *bp, int enable) (val & ~GRACKLE_PICR1_STG); out_be32(bp->cfg_addr, GRACKLE_CFA(0, 0, 0xa8)); out_le32((volatile unsigned int *)bp->cfg_data, val); + (void)in_le32((volatile unsigned int *)bp->cfg_data); } -static inline void grackle_set_loop_snoop(struct bridge_data *bp, int enable) +static inline void grackle_set_loop_snoop(struct pci_controller *bp, int enable) { unsigned int val; @@ -598,25 +369,68 @@ static inline void grackle_set_loop_snoop(struct bridge_data *bp, int enable) (val & ~GRACKLE_PICR1_LOOPSNOOP); out_be32(bp->cfg_addr, GRACKLE_CFA(0, 0, 0xa8)); out_le32((volatile unsigned int *)bp->cfg_data, val); + (void)in_le32((volatile unsigned int *)bp->cfg_data); } +static void __init +setup_uninorth(struct pci_controller* hose, struct reg_property* addr) +{ + pci_assign_all_busses = 1; + has_uninorth = 1; + hose->ops = ¯isc_pci_ops; + hose->cfg_addr = ioremap(addr->address + 0x800000, 0x1000); + hose->cfg_data = ioremap(addr->address + 0xc00000, 0x1000); + hose->io_base_phys = addr->address; + /* is 0x10000 enough for io space ? */ + hose->io_base_virt = (void *)ioremap(addr->address, 0x10000); + + /* XXX This is the bridge with the PCI expansion bus. We route + * legacy IOs to it. + */ + if (addr->address == 0xf2000000) + isa_io_base = (unsigned long)hose->io_base_virt; +} -void __init pmac_find_bridges(void) +static void __init +setup_bandit(struct pci_controller* hose, struct reg_property* addr) { - int bus; - struct bridge_data *bridge; + hose->ops = ¯isc_pci_ops; + hose->cfg_addr = (volatile unsigned int *) + ioremap(addr->address + 0x800000, 0x1000); + hose->cfg_data = (volatile unsigned char *) + ioremap(addr->address + 0xc00000, 0x1000); + hose->io_base_phys = addr->address; + hose->io_base_virt = (void *) ioremap(addr->address, 0x10000); + init_bandit(hose); +} - bridge_list = 0; - max_bus = 0; - add_bridges(find_devices("bandit")); - add_bridges(find_devices("chaos")); - add_bridges(find_devices("pci")); - bridges = (struct bridge_data **) - alloc_bootmem((max_bus + 1) * sizeof(struct bridge_data *)); - memset(bridges, 0, (max_bus + 1) * sizeof(struct bridge_data *)); - for (bridge = bridge_list; bridge != NULL; bridge = bridge->next) - for (bus = bridge->bus_number; bus <= bridge->max_bus; ++bus) - bridges[bus] = bridge; +static void __init +setup_chaos(struct pci_controller* hose, struct reg_property* addr) +{ + /* assume a `chaos' bridge */ + hose->ops = &chaos_pci_ops; + hose->cfg_addr = (volatile unsigned int *) + ioremap(addr->address + 0x800000, 0x1000); + hose->cfg_data = (volatile unsigned char *) + ioremap(addr->address + 0xc00000, 0x1000); + hose->io_base_phys = addr->address; + hose->io_base_virt = (void *) ioremap(addr->address, 0x10000); +} + +void __init +setup_grackle(struct pci_controller *hose, unsigned io_space_size) +{ + setup_indirect_pci(hose, 0xfec00000, 0xfee00000); + hose->io_base_phys = 0xfe000000; + hose->io_base_virt = (void *) ioremap(0xfe000000, io_space_size); + pci_dram_offset = 0; + isa_mem_base = 0xfd000000; + isa_io_base = (unsigned long) hose->io_base_virt; + if (machine_is_compatible("AAPL,PowerBook1998")) + grackle_set_loop_snoop(hose, 1); +#if 0 /* Disabled for now, HW problems ??? */ + grackle_set_stg(hose, 1); +#endif } /* @@ -626,11 +440,12 @@ void __init pmac_find_bridges(void) */ static void __init add_bridges(struct device_node *dev) { - int *bus_range; int len; - struct bridge_data *bp; + struct pci_controller *hose; struct reg_property *addr; - + char* disp_name; + int *bus_range; + for (; dev != NULL; dev = dev->next) { addr = (struct reg_property *) get_property(dev, "reg", &len); if (addr == NULL || len < sizeof(*addr)) { @@ -640,108 +455,52 @@ static void __init add_bridges(struct device_node *dev) } bus_range = (int *) get_property(dev, "bus-range", &len); if (bus_range == NULL || len < 2 * sizeof(int)) { - printk(KERN_WARNING "Can't get bus-range for %s\n", - dev->full_name); - continue; - } - if (bus_range[1] == bus_range[0]) - printk(KERN_INFO "PCI bus %d", bus_range[0]); - else - printk(KERN_INFO "PCI buses %d..%d", bus_range[0], - bus_range[1]); - printk(" controlled by %s at %x\n", dev->name, addr->address); - if (device_is_compatible(dev, "uni-north")) { - int i = uninorth_count++; - uninorth_bridges[i].cfg_addr = ioremap(addr->address + 0x800000, 0x1000); - uninorth_bridges[i].cfg_data = ioremap(addr->address + 0xc00000, 0x1000); - uninorth_bridges[i].node = dev; - uninorth_bridges[i].iobase_phys = addr->address; - /* is 0x10000 enough for io space ? */ - uninorth_bridges[i].iobase = (void *)ioremap(addr->address, 0x10000); - /* XXX This is the bridge with the PCI expansion bus. This is also the - * address of the bus that will receive type 1 config accesses and io - * accesses. Appears to be correct for iMac DV and G4 Sawtooth too. - * That means that we cannot do io cycles on the AGP bus nor the internal - * ethernet/fw bus. Fortunately, they appear not to be needed on iMac DV - * and G4 neither. - */ - if (addr->address == 0xf2000000) - uninorth_default = i; - else - continue; + printk(KERN_WARNING "Can't get bus-range for %s, assume bus 0\n", + dev->full_name); } - bp = (struct bridge_data *) alloc_bootmem(sizeof(*bp)); + hose = pcibios_alloc_controller(); + if (!hose) + continue; + hose->arch_data = dev; + hose->first_busno = bus_range ? bus_range[0] : 0; + hose->last_busno = bus_range ? bus_range[1] : 0xff; + + disp_name = NULL; if (device_is_compatible(dev, "uni-north")) { - bp->cfg_addr = 0; - bp->cfg_data = 0; - bp->io_base = uninorth_bridges[uninorth_count-1].iobase; - bp->io_base_phys = uninorth_bridges[uninorth_count-1].iobase_phys; + setup_uninorth(hose, addr); + disp_name = "UniNorth"; } else if (strcmp(dev->name, "pci") == 0) { /* XXX assume this is a mpc106 (grackle) */ - bp->cfg_addr = (volatile unsigned int *) - ioremap(0xfec00000, 0x1000); - bp->cfg_data = (volatile unsigned char *) - ioremap(0xfee00000, 0x1000); - bp->io_base_phys = 0xfe000000; - bp->io_base = (void *) ioremap(0xfe000000, 0x20000); - if (machine_is_compatible("AAPL,PowerBook1998")) - grackle_set_loop_snoop(bp, 1); -#if 0 /* Disabled for now, HW problems ??? */ - grackle_set_stg(bp, 1); -#endif - } else { - /* a `bandit' or `chaos' bridge */ - bp->cfg_addr = (volatile unsigned int *) - ioremap(addr->address + 0x800000, 0x1000); - bp->cfg_data = (volatile unsigned char *) - ioremap(addr->address + 0xc00000, 0x1000); - bp->io_base_phys = addr->address; - bp->io_base = (void *) ioremap(addr->address, 0x10000); + setup_grackle(hose, 0x20000); + disp_name = "Grackle (MPC106)"; + } else if (strcmp(dev->name, "bandit") == 0) { + setup_bandit(hose, addr); + disp_name = "Bandit"; + } else if (strcmp(dev->name, "chaos") == 0) { + setup_chaos(hose, addr); + disp_name = "Chaos"; } + printk(KERN_INFO "Found %s PCI host bridge at 0x%08x. Firmware bus number: %d->%d\n", + disp_name, addr->address, hose->first_busno, hose->last_busno); +#ifdef DEBUG + printk(" ->Hose at 0x%08lx, cfg_addr=0x%08lx,cfg_data=0x%08lx\n", + hose, hose->cfg_addr, hose->cfg_data); +#endif + + /* Setup a default isa_io_base */ if (isa_io_base == 0) - isa_io_base = (unsigned long) bp->io_base; - bp->bus_number = bus_range[0]; - bp->max_bus = bus_range[1]; - bp->next = bridge_list; - bp->node = dev; - bridge_list = bp; - if (bp->max_bus > max_bus) - max_bus = bp->max_bus; - - if (strcmp(dev->name, "bandit") == 0) - init_bandit(bp); - } -} - -/* Recursively searches any node that is of type PCI-PCI bridge. Without - * this, the old code would miss children of P2P bridges and hence not - * fix IRQ's for cards located behind P2P bridges. - * - Ranjit Deshpande, 01/20/99 - */ -void __init -fix_intr(struct device_node *node, struct pci_dev *dev) -{ - unsigned int *reg, *class_code; + isa_io_base = (unsigned long)hose->io_base_virt; - for (; node != 0;node = node->sibling) { - class_code = (unsigned int *) get_property(node, "class-code", 0); - if(class_code && (*class_code >> 8) == PCI_CLASS_BRIDGE_PCI) - fix_intr(node->child, dev); - reg = (unsigned int *) get_property(node, "reg", 0); - if (reg == 0 || ((reg[0] >> 8) & 0xff) != dev->devfn) - continue; - /* this is the node, see if it has interrupts */ - if (node->n_intrs > 0) - dev->irq = node->intrs[0].line; - break; + /* Fixup "bus-range" OF property */ + fixup_bus_range(dev); } } -void __init -pmac_pcibios_fixup(void) -{ - struct pci_dev *dev; +static void +pcibios_fixup_OF_interrupts(void) +{ + struct pci_dev* dev; /* * FIXME: This is broken: We should not assign IRQ's to IRQless @@ -757,44 +516,43 @@ pmac_pcibios_fixup(void) * should find the device node and se if it has an * AAPL,interrupts property. */ - struct bridge_data *bp = bridges[dev->bus->number]; unsigned char pin; + struct device_node* node; - if (pci_read_config_byte(dev, PCI_INTERRUPT_PIN, &pin) || - !pin) + if (pci_read_config_byte(dev, PCI_INTERRUPT_PIN, &pin) || !pin) continue; /* No interrupt generated -> no fixup */ - /* We iterate all instances of uninorth for now */ - if (uninorth_count && dev->bus->number == 0) { - int i; - for (i=0;i<uninorth_count;i++) - fix_intr(uninorth_bridges[i].node->child, dev); - } else - fix_intr(bp->node->child, dev); + node = pci_device_to_OF_node(dev); + if (!node) { + printk("No OF node for device %x:%x\n", dev->bus->number, dev->devfn >> 3); + continue; + } + /* this is the node, see if it has interrupts */ + if (node->n_intrs > 0) + dev->irq = node->intrs[0].line; + pci_write_config_byte(dev, PCI_INTERRUPT_LINE, dev->irq); } } void __init -pmac_setup_pci_ptrs(void) +pmac_pcibios_fixup(void) { - struct device_node* np; + /* Fixup interrupts according to OF tree */ + pcibios_fixup_OF_interrupts(); +} - np = find_devices("pci"); - if (np != 0) - { - if (device_is_compatible(np, "uni-north")) - { - /* looks like an Core99 powermac */ - set_config_access_method(uni); - } else - { - /* looks like a G3 powermac */ - set_config_access_method(grackle); - } - } else - { - set_config_access_method(pmac); +/* We don't want to enable USB controllers absent from the OF tree + * (iBook second controller) + */ +int +pmac_pci_enable_device_hook(struct pci_dev *dev, int initial) +{ + if (dev->vendor == PCI_VENDOR_ID_APPLE + && dev->device == PCI_DEVICE_ID_APPLE_KL_USB) { + struct device_node* node; + node = pci_device_to_OF_node(dev); + if (!node) + return -EINVAL; } - - ppc_md.pcibios_fixup = pmac_pcibios_fixup; + return 0; } diff --git a/arch/ppc/kernel/pmac_pic.c b/arch/ppc/kernel/pmac_pic.c index efd767482..a9a1777ca 100644 --- a/arch/ppc/kernel/pmac_pic.c +++ b/arch/ppc/kernel/pmac_pic.c @@ -4,14 +4,15 @@ #include <linux/sched.h> #include <linux/signal.h> #include <linux/pci.h> -#include <linux/openpic.h> #include <asm/init.h> #include <asm/io.h> #include <asm/smp.h> #include <asm/prom.h> #include <asm/pci-bridge.h> + #include "pmac_pic.h" +#include "open_pic.h" /* pmac */struct pmac_irq_hw { unsigned int flag; @@ -30,16 +31,14 @@ static volatile struct pmac_irq_hw *pmac_irq_hw[4] = { static int max_irqs; static int max_real_irqs; -static int has_openpic = 0; +static int pmac_has_openpic; + +spinlock_t pmac_pic_lock = SPIN_LOCK_UNLOCKED; + #define GATWICK_IRQ_POOL_SIZE 10 static struct interrupt_info gatwick_int_pool[GATWICK_IRQ_POOL_SIZE]; -extern int pmac_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short *val); -extern int pmac_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn, - unsigned char offset, unsigned short val); - /* * Mark an irq as "lost". This is only used on the pmac * since it can lose interrupts (see pmac_set_irq_mask). @@ -51,48 +50,11 @@ void __pmac __no_use_set_lost(unsigned long irq_nr) atomic_inc(&ppc_n_lost_interrupts); } -static void pmac_openpic_mask_irq(unsigned int irq_nr) -{ - openpic_disable_irq(irq_nr); -} - -static void pmac_openpic_unmask_irq(unsigned int irq_nr) -{ - openpic_enable_irq(irq_nr); -} - -static void pmac_openpic_ack_irq(unsigned int irq_nr) -{ - if ((irq_desc[irq_nr].status & IRQ_LEVEL) == 0) - openpic_eoi(smp_processor_id()); - openpic_disable_irq(irq_nr); -} - -static void pmac_openpic_end_irq(unsigned int irq_nr) -{ - if ((irq_desc[irq_nr].status & IRQ_LEVEL) != 0) - openpic_eoi(smp_processor_id()); - openpic_enable_irq(irq_nr); -} - -struct hw_interrupt_type pmac_open_pic = { - " OpenPIC ", - NULL, - NULL, - pmac_openpic_unmask_irq, - pmac_openpic_mask_irq, - /* Theorically, the mask&ack should be NULL for OpenPIC. However, doing - * so shows tons of bogus interrupts coming in. - */ - pmac_openpic_ack_irq, - pmac_openpic_end_irq, - NULL -}; - static void __pmac pmac_mask_and_ack_irq(unsigned int irq_nr) { unsigned long bit = 1UL << (irq_nr & 0x1f); int i = irq_nr >> 5; + unsigned long flags; if ((unsigned)irq_nr >= max_irqs) return; @@ -100,6 +62,7 @@ static void __pmac pmac_mask_and_ack_irq(unsigned int irq_nr) clear_bit(irq_nr, ppc_cached_irq_mask); if (test_and_clear_bit(irq_nr, ppc_lost_interrupts)) atomic_dec(&ppc_n_lost_interrupts); + spin_lock_irqsave(&pmac_pic_lock, flags); out_le32(&pmac_irq_hw[i]->ack, bit); out_le32(&pmac_irq_hw[i]->enable, ppc_cached_irq_mask[i]); out_le32(&pmac_irq_hw[i]->ack, bit); @@ -108,16 +71,19 @@ static void __pmac pmac_mask_and_ack_irq(unsigned int irq_nr) interrupts */ mb(); } while(in_le32(&pmac_irq_hw[i]->flag) & bit); + spin_unlock_irqrestore(&pmac_pic_lock, flags); } static void __pmac pmac_set_irq_mask(unsigned int irq_nr) { unsigned long bit = 1UL << (irq_nr & 0x1f); int i = irq_nr >> 5; + unsigned long flags; if ((unsigned)irq_nr >= max_irqs) return; + spin_lock_irqsave(&pmac_pic_lock, flags); /* enable unmasked interrupts */ out_le32(&pmac_irq_hw[i]->enable, ppc_cached_irq_mask[i]); @@ -137,6 +103,7 @@ static void __pmac pmac_set_irq_mask(unsigned int irq_nr) && (ld_le32(&pmac_irq_hw[i]->level) & bit) && !(ld_le32(&pmac_irq_hw[i]->flag) & bit)) __set_lost((ulong)irq_nr); + spin_unlock_irqrestore(&pmac_pic_lock, flags); } static void __pmac pmac_mask_irq(unsigned int irq_nr) @@ -152,6 +119,15 @@ static void __pmac pmac_unmask_irq(unsigned int irq_nr) pmac_set_irq_mask(irq_nr); } +static void __pmac pmac_end_irq(unsigned int irq_nr) +{ + if (!(irq_desc[irq_nr].status & (IRQ_DISABLED|IRQ_INPROGRESS))) { + set_bit(irq_nr, ppc_cached_irq_mask); + pmac_set_irq_mask(irq_nr); + } +} + + struct hw_interrupt_type pmac_pic = { " PMAC-PIC ", NULL, @@ -159,7 +135,7 @@ struct hw_interrupt_type pmac_pic = { pmac_unmask_irq, pmac_mask_irq, pmac_mask_and_ack_irq, - pmac_unmask_irq, + pmac_end_irq, NULL }; @@ -170,7 +146,7 @@ struct hw_interrupt_type gatwick_pic = { pmac_unmask_irq, pmac_mask_irq, pmac_mask_and_ack_irq, - pmac_unmask_irq, + pmac_end_irq, NULL }; @@ -204,35 +180,22 @@ pmac_get_irq(struct pt_regs *regs) unsigned long bits = 0; #ifdef CONFIG_SMP - void pmac_smp_message_recv(struct pt_regs *); + void psurge_smp_message_recv(struct pt_regs *); - /* IPI's are a hack on the powersurge -- Cort */ - if ( smp_processor_id() != 0 ) - { - pmac_smp_message_recv(regs); + /* IPI's are a hack on the powersurge -- Cort */ + if ( smp_processor_id() != 0 ) { + psurge_smp_message_recv(regs); return -2; /* ignore, already handled */ } #endif /* CONFIG_SMP */ - - if (has_openpic) { - irq = openpic_irq(smp_processor_id()); - if (irq == OPENPIC_VEC_SPURIOUS) - /* We get those when doing polled ADB requests, - * using -2 is a temp hack to disable the printk - */ - irq = -2; /*-1; */ - } - else - { - for (irq = max_real_irqs; (irq -= 32) >= 0; ) { - int i = irq >> 5; - bits = ld_le32(&pmac_irq_hw[i]->flag) - | ppc_lost_interrupts[i]; - if (bits == 0) - continue; - irq += __ilog2(bits); - break; - } + for (irq = max_real_irqs; (irq -= 32) >= 0; ) { + int i = irq >> 5; + bits = ld_le32(&pmac_irq_hw[i]->flag) + | ppc_lost_interrupts[i]; + if (bits == 0) + continue; + irq += __ilog2(bits); + break; } return irq; @@ -336,11 +299,16 @@ static void __init enable_second_ohare(void) addr = (unsigned long) ioremap(irqctrler->addrs[0].address, 0x40); pmac_irq_hw[1] = (volatile struct pmac_irq_hw *)(addr + 0x20); max_irqs = 64; - if (pci_device_loc(irqctrler, &bus, &devfn) == 0) { - pmac_pcibios_read_config_word(bus, devfn, PCI_COMMAND, &cmd); - cmd |= PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER; - cmd &= ~PCI_COMMAND_IO; - pmac_pcibios_write_config_word(bus, devfn, PCI_COMMAND, cmd); + if (pci_device_from_OF_node(irqctrler, &bus, &devfn) == 0) { + struct pci_controller* hose = pci_find_hose_for_OF_device(irqctrler); + if (!hose) + printk(KERN_ERR "Can't find PCI hose for OHare2 !\n"); + else { + early_read_config_word(hose, bus, devfn, PCI_COMMAND, &cmd); + cmd |= PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER; + cmd &= ~PCI_COMMAND_IO; + early_write_config_word(hose, bus, devfn, PCI_COMMAND, cmd); + } } second_irq = irqctrler->intrs[0].line; @@ -378,20 +346,26 @@ pmac_pic_init(void) printk("PowerMac using OpenPIC irq controller\n"); if (irqctrler->n_addrs > 0) { + int nmi_irq = -1; + unsigned char senses[NR_IRQS]; #ifdef CONFIG_XMON struct device_node* pswitch; -#endif /* CONFIG_XMON */ - OpenPIC = (volatile struct OpenPIC *) - ioremap(irqctrler->addrs[0].address, - irqctrler->addrs[0].size); - for ( i = 0 ; i < NR_IRQS ; i++ ) - irq_desc[i].handler = &pmac_open_pic; - openpic_init(1); - has_openpic = 1; -#ifdef CONFIG_XMON + pswitch = find_devices("programmer-switch"); if (pswitch && pswitch->n_intrs) - request_irq(pswitch->intrs[0].line, xmon_irq, 0, + nmi_irq = pswitch->intrs[0].line; +#endif /* CONFIG_XMON */ + prom_get_irq_senses(senses, 0, NR_IRQS); + OpenPIC_InitSenses = senses; + OpenPIC_NumInitSenses = NR_IRQS; + ppc_md.get_irq = openpic_get_irq; + OpenPIC_Addr = ioremap(irqctrler->addrs[0].address, + irqctrler->addrs[0].size); + openpic_init(1, 0, 0, nmi_irq); + pmac_has_openpic = 1; +#ifdef CONFIG_XMON + if (nmi_irq >= 0) + request_irq(nmi_irq, xmon_irq, 0, "NMI - XMON", 0); #endif /* CONFIG_XMON */ return; diff --git a/arch/ppc/kernel/pmac_setup.c b/arch/ppc/kernel/pmac_setup.c index b5bf03abc..e7be1114e 100644 --- a/arch/ppc/kernel/pmac_setup.c +++ b/arch/ppc/kernel/pmac_setup.c @@ -47,6 +47,7 @@ #include <linux/cuda.h> #include <linux/pmu.h> +#include <asm/processor.h> #include <asm/init.h> #include <asm/prom.h> #include <asm/system.h> @@ -73,7 +74,8 @@ extern unsigned long pmac_get_rtc_time(void); extern int pmac_set_rtc_time(unsigned long nowtime); extern void pmac_read_rtc_time(void); extern void pmac_calibrate_decr(void); -extern void pmac_setup_pci_ptrs(void); +extern void pmac_pcibios_fixup(void); +extern void pmac_find_bridges(void); extern int mackbd_setkeycode(unsigned int scancode, unsigned int keycode); extern int mackbd_getkeycode(unsigned int scancode); @@ -99,9 +101,7 @@ extern char pckbd_unexpected_up(unsigned char keycode); extern int keyboard_sends_linux_keycodes; extern void pmac_nvram_update(void); -extern void *pmac_pci_dev_io_base(unsigned char bus, unsigned char devfn, int physical); -extern void *pmac_pci_dev_mem_base(unsigned char bus, unsigned char devfn); -extern int pmac_pci_dev_root_bridge(unsigned char bus, unsigned char devfn); +extern int pmac_pci_enable_device_hook(struct pci_dev *dev, int initial); unsigned char drive_info; @@ -119,13 +119,34 @@ extern int pmac_newworld; extern void zs_kgdb_hook(int tty_num); static void ohare_init(void); -static void init_p2pbridge(void); #ifdef CONFIG_BOOTX_TEXT void pmac_progress(char *s, unsigned short hex); #endif sys_ctrler_t sys_ctrler = SYS_CTRLER_UNKNOWN; +#ifdef CONFIG_SMP +volatile static long int core99_l2_cache; +void core99_init_l2(void) +{ + int cpu = smp_processor_id(); + + if ( (_get_PVR() >> 16) != 8 && (_get_PVR() >> 16) != 12 ) + return; + + if (cpu == 0){ + core99_l2_cache = _get_L2CR(); + printk("CPU0: L2CR is %lx\n", core99_l2_cache); + } else { + printk("CPU%d: L2CR was %lx\n", cpu, _get_L2CR()); + _set_L2CR(0); + _set_L2CR(core99_l2_cache); + printk("CPU%d: L2CR set to %lx\n", cpu, core99_l2_cache); + } +} +#endif /* CONFIG_SMP */ + + __pmac int pmac_get_cpuinfo(char *buffer) @@ -250,7 +271,7 @@ pmac_setup_arch(void) struct device_node *cpu; int *fp; - /* Set loops_per_sec to a half-way reasonable value, + /* Set loops_per_jiffy to a half-way reasonable value, for use until calibrate_delay gets called. */ cpu = find_type_devices("cpu"); if (cpu != 0) { @@ -263,13 +284,13 @@ pmac_setup_arch(void) case 10: /* mach V (604ev5) */ case 12: /* G4 */ case 20: /* 620 */ - loops_per_sec = *fp; + loops_per_jiffy = *fp / HZ; break; default: /* 601, 603, etc. */ - loops_per_sec = *fp / 2; + loops_per_jiffy = *fp / (2*HZ); } } else - loops_per_sec = 50000000; + loops_per_jiffy = 50000000 / HZ; } /* this area has the CPU identification register @@ -278,8 +299,8 @@ pmac_setup_arch(void) __ioremap(0xffc00000, 0x400000, pgprot_val(PAGE_READONLY)); ohare_init(); + /* Lookup PCI hosts */ pmac_find_bridges(); - init_p2pbridge(); /* Checks "l2cr-value" property in the registry */ if ( (_get_PVR() >> 16) == 8 || (_get_PVR() >> 16) == 12 ) { @@ -303,6 +324,11 @@ pmac_setup_arch(void) ppc_override_l2cr_value, (ppc_override_l2cr_value & 0x80000000) ? "enabled" : "disabled"); +#ifdef CONFIG_SMP + /* somewhat of a hack */ + core99_init_l2(); +#endif + #ifdef CONFIG_KGDB zs_kgdb_hook(0); #endif @@ -330,32 +356,6 @@ pmac_setup_arch(void) ROOT_DEV = to_kdev_t(DEFAULT_ROOT_DEVICE); } -/* - * Tweak the PCI-PCI bridge chip on the blue & white G3s. - */ -static void __init init_p2pbridge(void) -{ - struct device_node *p2pbridge; - unsigned char bus, devfn; - unsigned short val; - - /* XXX it would be better here to identify the specific - PCI-PCI bridge chip we have. */ - if ((p2pbridge = find_devices("pci-bridge")) == 0 - || p2pbridge->parent == NULL - || strcmp(p2pbridge->parent->name, "pci") != 0) - return; - if (pci_device_loc(p2pbridge, &bus, &devfn) < 0) - return; - if (ppc_md.pcibios_read_config_word(bus, devfn, PCI_BRIDGE_CONTROL, &val) < 0) { - printk(KERN_ERR "init_p2pbridge: couldn't read bridge control\n"); - return; - } - val &= ~PCI_BRIDGE_CTL_MASTER_ABORT; - ppc_md.pcibios_write_config_word(bus, devfn, PCI_BRIDGE_CONTROL, val); - ppc_md.pcibios_read_config_word(bus, devfn, PCI_BRIDGE_CONTROL, &val); -} - static void __init ohare_init(void) { /* @@ -448,7 +448,7 @@ kdev_t __init find_ide_boot(void) void __init find_boot_device(void) { -#ifdef CONFIG_SCSI +#if defined(CONFIG_SCSI) && defined(CONFIG_BLK_DEV_SD) if (boot_host != NULL) { boot_dev = sd_find_target(boot_host, boot_target); if (boot_dev != 0) @@ -493,7 +493,9 @@ pmac_restart(char *cmd) struct adb_request req; #endif /* CONFIG_ADB_CUDA */ +#ifdef CONFIG_NVRAM pmac_nvram_update(); +#endif switch (sys_ctrler) { #ifdef CONFIG_ADB_CUDA @@ -509,7 +511,7 @@ pmac_restart(char *cmd) pmu_restart(); break; #endif /* CONFIG_ADB_PMU */ - default: + default: ; } } @@ -520,7 +522,9 @@ pmac_power_off(void) struct adb_request req; #endif /* CONFIG_ADB_CUDA */ +#ifdef CONFIG_NVRAM pmac_nvram_update(); +#endif switch (sys_ctrler) { #ifdef CONFIG_ADB_CUDA @@ -536,7 +540,7 @@ pmac_power_off(void) pmu_shutdown(); break; #endif /* CONFIG_ADB_PMU */ - default: + default: ; } } @@ -566,17 +570,19 @@ pmac_ide_outsw(ide_ioreg_t port, void *buf, int ns) int pmac_ide_default_irq(ide_ioreg_t base) { - return 0; -} - #if defined(CONFIG_BLK_DEV_IDE) && defined(CONFIG_BLK_DEV_IDE_PMAC) -extern ide_ioreg_t pmac_ide_get_base(int index); + extern int pmac_ide_get_irq(ide_ioreg_t base); + return pmac_ide_get_irq(base); +#else + return 0; #endif +} ide_ioreg_t pmac_ide_default_io_base(int index) { #if defined(CONFIG_BLK_DEV_IDE) && defined(CONFIG_BLK_DEV_IDE_PMAC) + extern ide_ioreg_t pmac_ide_get_base(int index); return pmac_ide_get_base(index); #else return 0; @@ -586,7 +592,14 @@ pmac_ide_default_io_base(int index) int pmac_ide_check_region(ide_ioreg_t from, unsigned int extent) { - return 0; + /* + * We only do the check_region if `from' looks like a genuine + * I/O port number. If it actually refers to a memory-mapped + * register, it should be OK. + */ + if (from < ~_IO_BASE) + return 0; + return check_region(from, extent); } void @@ -594,24 +607,16 @@ pmac_ide_request_region(ide_ioreg_t from, unsigned int extent, const char *name) { + if (from < ~_IO_BASE) + request_region(from, extent, name); } void pmac_ide_release_region(ide_ioreg_t from, unsigned int extent) { -} - -/* Convert the shorts/longs in hd_driveid from little to big endian; - * chars are endian independant, of course, but strings need to be flipped. - * (Despite what it says in drivers/block/ide.h, they come up as little - * endian...) - * - * Changes to linux/hdreg.h may require changes here. */ -void -pmac_ide_fix_driveid(struct hd_driveid *id) -{ - ppc_generic_ide_fix_driveid(id); + if (from < ~_IO_BASE) + release_region(from, extent); } #if defined(CONFIG_BLK_DEV_IDE) && defined(CONFIG_BLK_DEV_IDE_PMAC) @@ -632,8 +637,6 @@ void __init pmac_init(unsigned long r3, unsigned long r4, unsigned long r5, unsigned long r6, unsigned long r7) { - pmac_setup_pci_ptrs(); - /* isa_io_base gets set in pmac_find_bridges */ isa_mem_base = PMAC_ISA_MEM_BASE; pci_dram_offset = PMAC_PCI_DRAM_OFFSET; @@ -646,8 +649,11 @@ pmac_init(unsigned long r3, unsigned long r4, unsigned long r5, ppc_md.get_cpuinfo = pmac_get_cpuinfo; ppc_md.irq_cannonicalize = NULL; ppc_md.init_IRQ = pmac_pic_init; - ppc_md.get_irq = pmac_get_irq; + ppc_md.get_irq = pmac_get_irq; /* Changed later on ... */ ppc_md.init = pmac_init2; + + ppc_md.pcibios_fixup = pmac_pcibios_fixup; + ppc_md.pcibios_enable_device_hook = pmac_pci_enable_device_hook; ppc_md.restart = pmac_restart; ppc_md.power_off = pmac_power_off; @@ -658,12 +664,8 @@ pmac_init(unsigned long r3, unsigned long r4, unsigned long r5, ppc_md.get_rtc_time = pmac_get_rtc_time; ppc_md.calibrate_decr = pmac_calibrate_decr; - ppc_md.pci_dev_io_base = pmac_pci_dev_io_base; - ppc_md.pci_dev_mem_base = pmac_pci_dev_mem_base; - ppc_md.pci_dev_root_bridge = pmac_pci_dev_root_bridge; - #ifdef CONFIG_VT -#ifdef CONFIG_INPUT_ADBHID +#ifdef CONFIG_INPUT ppc_md.kbd_init_hw = mac_hid_init_hw; ppc_md.kbd_translate = mac_hid_kbd_translate; ppc_md.kbd_unexpected_up = mac_hid_kbd_unexpected_up; @@ -682,35 +684,33 @@ pmac_init(unsigned long r3, unsigned long r4, unsigned long r5, } #endif /* CONFIG_MAGIC_SYSRQ */ #elif defined(CONFIG_ADB_KEYBOARD) - ppc_md.kbd_setkeycode = mackbd_setkeycode; - ppc_md.kbd_getkeycode = mackbd_getkeycode; - ppc_md.kbd_translate = mackbd_translate; - ppc_md.kbd_unexpected_up = mackbd_unexpected_up; - ppc_md.kbd_leds = mackbd_leds; - ppc_md.kbd_init_hw = mackbd_init_hw; + ppc_md.kbd_setkeycode = mackbd_setkeycode; + ppc_md.kbd_getkeycode = mackbd_getkeycode; + ppc_md.kbd_translate = mackbd_translate; + ppc_md.kbd_unexpected_up = mackbd_unexpected_up; + ppc_md.kbd_leds = mackbd_leds; + ppc_md.kbd_init_hw = mackbd_init_hw; #ifdef CONFIG_MAGIC_SYSRQ - ppc_md.ppc_kbd_sysrq_xlate = mackbd_sysrq_xlate; + ppc_md.ppc_kbd_sysrq_xlate = mackbd_sysrq_xlate; SYSRQ_KEY = 0x69; #endif /* CONFIG_MAGIC_SYSRQ */ #endif /* CONFIG_INPUT_ADBHID/CONFIG_ADB_KEYBOARD */ #endif /* CONFIG_VT */ #if defined(CONFIG_BLK_DEV_IDE) && defined(CONFIG_BLK_DEV_IDE_PMAC) - ppc_ide_md.insw = pmac_ide_insw; - ppc_ide_md.outsw = pmac_ide_outsw; - ppc_ide_md.default_irq = pmac_ide_default_irq; - ppc_ide_md.default_io_base = pmac_ide_default_io_base; - ppc_ide_md.ide_check_region = pmac_ide_check_region; - ppc_ide_md.ide_request_region = pmac_ide_request_region; - ppc_ide_md.ide_release_region = pmac_ide_release_region; - ppc_ide_md.fix_driveid = pmac_ide_fix_driveid; - ppc_ide_md.ide_init_hwif = pmac_ide_init_hwif_ports; - - ppc_ide_md.io_base = _IO_BASE; /* actually too early for this :-( */ -#endif + ppc_ide_md.insw = pmac_ide_insw; + ppc_ide_md.outsw = pmac_ide_outsw; + ppc_ide_md.default_irq = pmac_ide_default_irq; + ppc_ide_md.default_io_base = pmac_ide_default_io_base; + ppc_ide_md.ide_check_region = pmac_ide_check_region; + ppc_ide_md.ide_request_region = pmac_ide_request_region; + ppc_ide_md.ide_release_region = pmac_ide_release_region; + ppc_ide_md.fix_driveid = ppc_generic_ide_fix_driveid; + ppc_ide_md.ide_init_hwif = pmac_ide_init_hwif_ports; +#endif /* CONFIG_BLK_DEV_IDE && CONFIG_BLK_DEV_IDE_PMAC */ #ifdef CONFIG_BOOTX_TEXT ppc_md.progress = pmac_progress; -#endif +#endif /* CONFIG_BOOTX_TEXT */ if (ppc_md.progress) ppc_md.progress("pmac_init(): exit", 0); } @@ -727,5 +727,5 @@ pmac_progress(char *s, unsigned short hex) prom_drawstring(s); prom_drawchar('\n'); } -#endif CONFIG_BOOTX_TEXT +#endif /* CONFIG_BOOTX_TEXT */ diff --git a/arch/ppc/kernel/pmac_time.c b/arch/ppc/kernel/pmac_time.c index 00b6302a7..46d067e81 100644 --- a/arch/ppc/kernel/pmac_time.c +++ b/arch/ppc/kernel/pmac_time.c @@ -114,7 +114,7 @@ unsigned long pmac_get_rtc_time(void) + (req.reply[3] << 8) + req.reply[4]; return now - RTC_OFFSET; #endif /* CONFIG_ADB_PMU */ - default: + default: ; } return 0; } @@ -135,7 +135,7 @@ int pmac_set_rtc_time(unsigned long nowtime) return 0; while (!req.complete) cuda_poll(); -// if (req.reply_len != 7) + if ((req.reply_len != 3) && (req.reply_len != 7)) printk(KERN_ERR "pmac_set_rtc_time: got %d byte reply\n", req.reply_len); return 1; diff --git a/arch/ppc/kernel/ppc_asm.h b/arch/ppc/kernel/ppc_asm.h index 42b8c9c39..59c377c4d 100644 --- a/arch/ppc/kernel/ppc_asm.h +++ b/arch/ppc/kernel/ppc_asm.h @@ -50,7 +50,7 @@ */ #define LVX(r,a,b) .long (31<<26)+((r)<<21)+((a)<<16)+((b)<<11)+(103<<1) #define STVX(r,a,b) .long (31<<26)+((r)<<21)+((a)<<16)+((b)<<11)+(231<<1) -#define MFVSCR(r) .long (4<<26)+((r)<<21)+(1540<<1) +#define MFVSCR(r) .long (4<<26)+((r)<<21)+(770<<1) #define MTVSCR(r) .long (4<<26)+((r)<<11)+(802<<1) #define SAVE_VR(n,b,base) li b,THREAD_VR0+(16*(n)); STVX(n,b,base) @@ -66,9 +66,13 @@ #define REST_16VR(n,b,base) REST_8VR(n,b,base); REST_8VR(n+8,b,base) #define REST_32VR(n,b,base) REST_16VR(n,b,base); REST_16VR(n+16,b,base) +#ifdef CONFIG_PPC601_SYNC_FIX #define SYNC \ sync; \ isync +#else +#define SYNC +#endif /* * This instruction is not implemented on the PPC 603 or 601; however, on diff --git a/arch/ppc/kernel/ppc_htab.c b/arch/ppc/kernel/ppc_htab.c index 32f99ce0e..9768f2890 100644 --- a/arch/ppc/kernel/ppc_htab.c +++ b/arch/ppc/kernel/ppc_htab.c @@ -555,10 +555,7 @@ int proc_dol2crvec(ctl_table *table, int write, struct file *filp, break; buffer += len; left -= len; - _set_L2CR(0); _set_L2CR(val); - while ( _get_L2CR() & 0x1 ) - /* wait for invalidate to finish */; } else { p = buf; diff --git a/arch/ppc/kernel/ppc_ksyms.c b/arch/ppc/kernel/ppc_ksyms.c index 827e37540..6de2d94fe 100644 --- a/arch/ppc/kernel/ppc_ksyms.c +++ b/arch/ppc/kernel/ppc_ksyms.c @@ -12,6 +12,7 @@ #include <linux/console.h> #include <linux/irq.h> #include <linux/pci.h> +#include <linux/delay.h> #include <asm/page.h> #include <asm/semaphore.h> @@ -40,6 +41,7 @@ #include <asm/backlight.h> #ifdef CONFIG_SMP #include <asm/smplock.h> +#include <asm/smp.h> #endif /* CONFIG_SMP */ #include <asm/time.h> @@ -53,9 +55,10 @@ extern void MachineCheckException(struct pt_regs *regs); extern void AlignmentException(struct pt_regs *regs); extern void ProgramCheckException(struct pt_regs *regs); extern void SingleStepException(struct pt_regs *regs); -extern int sys_sigreturn(struct pt_regs *regs); extern void do_lost_interrupts(unsigned long); extern int do_signal(sigset_t *, struct pt_regs *); +extern int pmac_newworld; +extern int sys_sigreturn(struct pt_regs *regs); long long __ashrdi3(long long, int); long long __ashldi3(long long, int); @@ -98,10 +101,6 @@ EXPORT_SYMBOL(_prep_type); EXPORT_SYMBOL(ucSystemType); #endif #endif -#ifdef CONFIG_PCI -EXPORT_SYMBOL(pci_dev_io_base); -EXPORT_SYMBOL(pci_dev_mem_base); -#endif #if !__INLINE_BITOPS EXPORT_SYMBOL(set_bit); @@ -125,6 +124,7 @@ EXPORT_SYMBOL(strlen); EXPORT_SYMBOL(strnlen); EXPORT_SYMBOL(strcmp); EXPORT_SYMBOL(strncmp); +EXPORT_SYMBOL(strcasecmp); /* EXPORT_SYMBOL(csum_partial); already in net/netsyms.c */ EXPORT_SYMBOL(csum_partial_copy_generic); @@ -202,6 +202,10 @@ EXPORT_SYMBOL(_read_lock); EXPORT_SYMBOL(_read_unlock); EXPORT_SYMBOL(_write_lock); EXPORT_SYMBOL(_write_unlock); +EXPORT_SYMBOL(smp_call_function); +EXPORT_SYMBOL(smp_hw_index); +EXPORT_SYMBOL(smp_num_cpus); +EXPORT_SYMBOL(synchronize_irq); #endif #ifndef CONFIG_MACH_SPECIFIC @@ -234,12 +238,14 @@ EXPORT_SYMBOL(pmu_enable_irled); #ifdef CONFIG_PMAC_BACKLIGHT EXPORT_SYMBOL(get_backlight_level); EXPORT_SYMBOL(set_backlight_level); +EXPORT_SYMBOL(set_backlight_enable); +EXPORT_SYMBOL(register_backlight_controller); #endif /* CONFIG_PMAC_BACKLIGHT */ -#if defined(CONFIG_ALL_PPC) EXPORT_SYMBOL_NOVERS(sys_ctrler); #ifndef CONFIG_MACH_SPECIFIC EXPORT_SYMBOL_NOVERS(have_of); #endif /* CONFIG_MACH_SPECIFIC */ +#if defined(CONFIG_ALL_PPC) EXPORT_SYMBOL(find_devices); EXPORT_SYMBOL(find_type_devices); EXPORT_SYMBOL(find_compatible_devices); @@ -247,18 +253,29 @@ EXPORT_SYMBOL(find_path_device); EXPORT_SYMBOL(find_phandle); EXPORT_SYMBOL(device_is_compatible); EXPORT_SYMBOL(machine_is_compatible); -EXPORT_SYMBOL(find_pci_device_OFnode); EXPORT_SYMBOL(find_all_nodes); EXPORT_SYMBOL(get_property); -EXPORT_SYMBOL(pci_io_base); -EXPORT_SYMBOL(pci_device_loc); +EXPORT_SYMBOL(pci_bus_io_base); +EXPORT_SYMBOL(pci_bus_io_base_phys); +EXPORT_SYMBOL(pci_bus_mem_base_phys); +EXPORT_SYMBOL(pci_device_to_OF_node); +EXPORT_SYMBOL(pci_device_from_OF_node); +EXPORT_SYMBOL(pci_bus_to_hose); +EXPORT_SYMBOL(pci_resource_to_bus); +EXPORT_SYMBOL(pci_phys_to_bus); +EXPORT_SYMBOL(pci_bus_to_phys); +EXPORT_SYMBOL(pmac_newworld); EXPORT_SYMBOL(feature_set); EXPORT_SYMBOL(feature_clear); EXPORT_SYMBOL(feature_test); EXPORT_SYMBOL(feature_set_gmac_power); +EXPORT_SYMBOL(feature_set_gmac_phy_reset); EXPORT_SYMBOL(feature_set_usb_power); EXPORT_SYMBOL(feature_set_firewire_power); #endif /* defined(CONFIG_ALL_PPC) */ +#if defined(CONFIG_BOOTX_TEXT) +EXPORT_SYMBOL(bootx_update_display); +#endif #if defined(CONFIG_SCSI) && defined(CONFIG_ALL_PPC) EXPORT_SYMBOL(note_scsi_host); #endif @@ -286,6 +303,7 @@ EXPORT_SYMBOL(abs); EXPORT_SYMBOL(screen_info); #endif +EXPORT_SYMBOL(__delay); EXPORT_SYMBOL(int_control); EXPORT_SYMBOL(timer_interrupt_intercept); EXPORT_SYMBOL(timer_interrupt); @@ -300,6 +318,10 @@ EXPORT_SYMBOL(console_lock); #ifdef CONFIG_XMON EXPORT_SYMBOL(xmon); #endif +EXPORT_SYMBOL(__up); +EXPORT_SYMBOL(__down); +EXPORT_SYMBOL(__down_interruptible); +EXPORT_SYMBOL(__down_trylock); EXPORT_SYMBOL(down_read_failed); EXPORT_SYMBOL(down_write_failed); @@ -324,6 +346,12 @@ EXPORT_SYMBOL(do_softirq); EXPORT_SYMBOL(next_mmu_context); EXPORT_SYMBOL(set_context); EXPORT_SYMBOL(mmu_context_overflow); +#if !defined(CONFIG_8xx) && !defined(CONFIG_4xx) +extern long *intercept_table; +EXPORT_SYMBOL(intercept_table); +#endif +extern long *ret_from_intercept; +EXPORT_SYMBOL(ret_from_intercept); #ifdef CONFIG_MOL extern ulong mol_interface[]; diff --git a/arch/ppc/kernel/prep_nvram.c b/arch/ppc/kernel/prep_nvram.c index c3dcdea07..3d34a853f 100644 --- a/arch/ppc/kernel/prep_nvram.c +++ b/arch/ppc/kernel/prep_nvram.c @@ -16,19 +16,9 @@ #include <asm/machdep.h> #include <asm/prep_nvram.h> -/* - * Allow for a maximum of 32K of PReP NvRAM data - */ -#define MAX_PREP_NVRAM 0x8000 static char nvramData[MAX_PREP_NVRAM]; static NVRAM_MAP *nvram=(NVRAM_MAP *)&nvramData[0]; -#define PREP_NVRAM_AS0 0x74 -#define PREP_NVRAM_AS1 0x75 -#define PREP_NVRAM_DATA 0x77 - -unsigned char *rs_pcNvRAM; - unsigned char __prep prep_nvram_read_val(int addr) { outb(addr, PREP_NVRAM_AS0); @@ -44,20 +34,6 @@ void __prep prep_nvram_write_val(int addr, outb(val, PREP_NVRAM_DATA); } -/* - * Most Radstone boards have NvRAM memory mapped at offset 8M in ISA space - */ -unsigned char __prep rs_nvram_read_val(int addr) -{ - return rs_pcNvRAM[addr]; -} - -void __prep rs_nvram_write_val(int addr, - unsigned char val) -{ - rs_pcNvRAM[addr]=val; -} - void __init init_prep_nvram(void) { unsigned char *nvp; @@ -65,16 +41,6 @@ void __init init_prep_nvram(void) int nvramSize; /* - * I'm making the assumption that 32k will always cover the - * nvramsize. If this isn't the case please let me know and we can - * map the header, then get the size from the header, then map - * the whole size. -- Cort - */ - if ( _prep_type == _PREP_Radstone ) - rs_pcNvRAM = (unsigned char *)ioremap(_ISA_MEM_BASE+0x00800000, - 32<<10); - request_region(PREP_NVRAM_AS0, 0x8, "PReP NVRAM"); - /* * The following could fail if the NvRAM were corrupt but * we expect the boot firmware to have checked its checksum * before boot diff --git a/arch/ppc/kernel/prep_pci.c b/arch/ppc/kernel/prep_pci.c index fd14fc483..2f556d83d 100644 --- a/arch/ppc/kernel/prep_pci.c +++ b/arch/ppc/kernel/prep_pci.c @@ -7,11 +7,11 @@ * The motherboard routes/maps will disappear shortly. -- Cort */ +#include <linux/config.h> #include <linux/types.h> #include <linux/pci.h> #include <linux/kernel.h> #include <linux/init.h> -#include <linux/openpic.h> #include <asm/init.h> #include <asm/byteorder.h> @@ -25,6 +25,7 @@ #include <asm/machdep.h> #include "pci.h" +#include "open_pic.h" #define MAX_DEVNR 22 @@ -39,9 +40,6 @@ unsigned char *Motherboard_routes; /* Used for Motorola to store system config register */ static unsigned long *ProcInfo; -extern int chrp_get_irq(struct pt_regs *); -extern void chrp_post_irq(struct pt_regs* regs, int); - /* Tables for known hardware */ /* Motorola PowerStackII - Utah */ @@ -534,128 +532,45 @@ static char Nobis_pci_IRQ_routes[] __prepdata = { #define CFGPTR(dev) (0x80800000 | (1<<(dev>>3)) | ((dev&7)<<8) | offset) #define DEVNO(dev) (dev>>3) -__prep -int -prep_pcibios_read_config_dword (unsigned char bus, - unsigned char dev, unsigned char offset, unsigned int *val) -{ - unsigned long _val; - unsigned long *ptr; - - if ((bus != 0) || (DEVNO(dev) > MAX_DEVNR)) - { - *val = 0xFFFFFFFF; - return PCIBIOS_DEVICE_NOT_FOUND; - } else - { - ptr = (unsigned long *)CFGPTR(dev); - _val = le32_to_cpu(*ptr); - } - *val = _val; - return PCIBIOS_SUCCESSFUL; -} - -__prep -int -prep_pcibios_read_config_word (unsigned char bus, - unsigned char dev, unsigned char offset, unsigned short *val) -{ - unsigned short _val; - unsigned short *ptr; - - if ((bus != 0) || (DEVNO(dev) > MAX_DEVNR)) - { - *val = 0xFFFF; - return PCIBIOS_DEVICE_NOT_FOUND; - } else - { - ptr = (unsigned short *)CFGPTR(dev); - _val = le16_to_cpu(*ptr); - } - *val = _val; - return PCIBIOS_SUCCESSFUL; -} - -__prep -int -prep_pcibios_read_config_byte (unsigned char bus, - unsigned char dev, unsigned char offset, unsigned char *val) -{ - unsigned char _val; - unsigned char *ptr; - - if ((bus != 0) || (DEVNO(dev) > MAX_DEVNR)) - { - *val = 0xFF; - return PCIBIOS_DEVICE_NOT_FOUND; - } else - { - ptr = (unsigned char *)CFGPTR(dev); - _val = *ptr; - } - *val = _val; - return PCIBIOS_SUCCESSFUL; -} - -__prep -int -prep_pcibios_write_config_dword (unsigned char bus, - unsigned char dev, unsigned char offset, unsigned int val) -{ - unsigned long _val; - unsigned long *ptr; - - _val = le32_to_cpu(val); - if ((bus != 0) || (DEVNO(dev) > MAX_DEVNR)) - { - return PCIBIOS_DEVICE_NOT_FOUND; - } else - { - ptr = (unsigned long *)CFGPTR(dev); - *ptr = _val; - } - return PCIBIOS_SUCCESSFUL; +#define cfg_read(val, addr, type, op) *val = op((type)(addr)) +#define cfg_write(val, addr, type, op) op((type *)(addr), (val)) + +#define cfg_read_bad(val, size) *val = bad_##size; +#define cfg_write_bad(val, size) + +#define bad_byte 0xff +#define bad_word 0xffff +#define bad_dword 0xffffffffU + +#define PREP_PCI_OP(rw, size, type, op) \ +static int __prep \ +prep_##rw##_config_##size(struct pci_dev *dev, int offset, type val) \ +{ \ + if ((dev->bus->number != 0) || (DEVNO(dev->devfn) > MAX_DEVNR)) \ + { \ + cfg_##rw##_bad(val, size) \ + return PCIBIOS_DEVICE_NOT_FOUND; \ + } \ + cfg_##rw(val, CFGPTR(dev->devfn), type, op); \ + return PCIBIOS_SUCCESSFUL; \ } -__prep -int -prep_pcibios_write_config_word (unsigned char bus, - unsigned char dev, unsigned char offset, unsigned short val) -{ - unsigned short _val; - unsigned short *ptr; - - _val = le16_to_cpu(val); - if ((bus != 0) || (DEVNO(dev) > MAX_DEVNR)) - { - return PCIBIOS_DEVICE_NOT_FOUND; - } else - { - ptr = (unsigned short *)CFGPTR(dev); - *ptr = _val; - } - return PCIBIOS_SUCCESSFUL; -} +PREP_PCI_OP(read, byte, u8 *, in_8) +PREP_PCI_OP(read, word, u16 *, in_le16) +PREP_PCI_OP(read, dword, u32 *, in_le32) +PREP_PCI_OP(write, byte, u8, out_8) +PREP_PCI_OP(write, word, u16, out_le16) +PREP_PCI_OP(write, dword, u32, out_le32) -__prep -int -prep_pcibios_write_config_byte (unsigned char bus, - unsigned char dev, unsigned char offset, unsigned char val) +static struct pci_ops prep_pci_ops = { - unsigned char _val; - unsigned char *ptr; - - _val = val; - if ((bus != 0) || (DEVNO(dev) > MAX_DEVNR)) - { - return PCIBIOS_DEVICE_NOT_FOUND; - } else - { - ptr = (unsigned char *)CFGPTR(dev); - *ptr = _val; - } - return PCIBIOS_SUCCESSFUL; -} + prep_read_config_byte, + prep_read_config_word, + prep_read_config_dword, + prep_write_config_byte, + prep_write_config_word, + prep_write_config_dword +}; #define MOTOROLA_CPUTYPE_REG 0x800 #define MOTOROLA_BASETYPE_REG 0x803 @@ -685,7 +600,8 @@ static u_char mvme2600_openpic_initsenses[] __initdata = { #define MOT_HAWK_PRESENT 0x2 int prep_keybd_present = 1; -int MotMPIC = 0; +int MotMPIC; +int mot_multi; int __init raven_init(void) { @@ -695,18 +611,18 @@ int __init raven_init(void) /* Check to see if the Raven chip exists. */ if ( _prep_type != _PREP_Motorola) { - OpenPIC = NULL; + OpenPIC_Addr = NULL; return 0; } /* Check to see if this board is a type that might have a Raven. */ if ((inb(MOTOROLA_CPUTYPE_REG) & 0xF0) != 0xE0) { - OpenPIC = NULL; + OpenPIC_Addr = NULL; return 0; } /* Check the first PCI device to see if it is a Raven. */ - pcibios_read_config_dword(0, 0, PCI_VENDOR_ID, &devid); + early_read_config_dword(0, 0, 0, PCI_VENDOR_ID, &devid); switch (devid & 0xffff0000) { case MPIC_RAVEN_ID: @@ -716,33 +632,37 @@ int __init raven_init(void) MotMPIC = MOT_HAWK_PRESENT; break; default: - OpenPIC = NULL; + OpenPIC_Addr = NULL; return 0; } /* Read the memory base register. */ - pcibios_read_config_dword(0, 0, PCI_BASE_ADDRESS_1, &pci_membase); + early_read_config_dword(0, 0, 0, PCI_BASE_ADDRESS_1, &pci_membase); if (pci_membase == 0) { - OpenPIC = NULL; + OpenPIC_Addr = NULL; return 0; } /* Map the Raven MPIC registers to virtual memory. */ - OpenPIC = (struct OpenPIC *)ioremap(pci_membase+0xC0000000, 0x22000); + OpenPIC_Addr = ioremap(pci_membase+0xC0000000, 0x22000); OpenPIC_InitSenses = mvme2600_openpic_initsenses; OpenPIC_NumInitSenses = sizeof(mvme2600_openpic_initsenses); - ppc_md.get_irq = chrp_get_irq; - ppc_md.post_irq = chrp_post_irq; + ppc_md.get_irq = openpic_get_irq; /* If raven is present on Motorola store the system config register * for later use. */ ProcInfo = (unsigned long *)ioremap(0xfef80400, 4); + /* Indicate to system if this is a multiprocessor board */ + if (!(*ProcInfo & MOT_PROC2_BIT)) { + mot_multi = 1; + } + /* This is a hack. If this is a 2300 or 2400 mot board then there is * no keyboard controller and we have to indicate that. */ @@ -898,72 +818,8 @@ unsigned long __init prep_route_pci_interrupts(void) outb(pl_id|CAROLINA_IRQ_EDGE_MASK_HI, 0x04d1); pl_id=inb(0x04d1); /*printk("Hi mask now %#0x\n", pl_id);*/ - } else if ( _prep_type == _PREP_Radstone ) - { - unsigned char ucElcrM, ucElcrS; - - /* - * Set up edge/level - */ - switch(ucSystemType) - { - case RS_SYS_TYPE_PPC1: - { - if(ucBoardRevMaj<5) - { - ucElcrS=ELCRS_INT15_LVL; - } - else - { - ucElcrS=ELCRS_INT9_LVL | - ELCRS_INT11_LVL | - ELCRS_INT14_LVL | - ELCRS_INT15_LVL; - } - ucElcrM=ELCRM_INT5_LVL | ELCRM_INT7_LVL; - break; - } - - case RS_SYS_TYPE_PPC1a: - { - ucElcrS=ELCRS_INT9_LVL | - ELCRS_INT11_LVL | - ELCRS_INT14_LVL | - ELCRS_INT15_LVL; - ucElcrM=ELCRM_INT5_LVL; - break; - } - - case RS_SYS_TYPE_PPC2: - case RS_SYS_TYPE_PPC2a: - case RS_SYS_TYPE_PPC2ep: - case RS_SYS_TYPE_PPC4: - case RS_SYS_TYPE_PPC4a: - default: - { - ucElcrS=ELCRS_INT9_LVL | - ELCRS_INT10_LVL | - ELCRS_INT11_LVL | - ELCRS_INT14_LVL | - ELCRS_INT15_LVL; - ucElcrM=ELCRM_INT5_LVL | - ELCRM_INT7_LVL; - break; - } - } - - /* - * Write edge/level selection - */ - outb(ucElcrS, ISA8259_S_ELCR); - outb(ucElcrM, ISA8259_M_ELCR); - - /* - * Radstone boards have PCI interrupts all set up - * so leave well alone - */ - return 0; - } else + } + else { printk("No known machine pci routing!\n"); return -1; @@ -987,16 +843,10 @@ prep_pcibios_fixup(void) extern unsigned char *Motherboard_routes; unsigned char i; - if ( _prep_type == _PREP_Radstone ) - { - printk("Radstone boards require no PCI fixups\n"); - return; - } - prep_route_pci_interrupts(); printk("Setting PCI interrupts for a \"%s\"\n", Motherboard_map_name); - if (OpenPIC) { + if (OpenPIC_Addr) { /* PCI interrupts are controlled by the OpenPIC */ pci_for_each_dev(dev) { if (dev->bus->number == 0) { @@ -1018,7 +868,12 @@ prep_pcibios_fixup(void) for ( i = 0 ; i <= 5 ; i++ ) { - if ( dev->resource[i].start > 0x10000000 ) + /* + * Relocate PCI I/O resources if necessary so the + * standard 256MB BAT covers them. + */ + if ( (pci_resource_flags(dev, i) & IORESOURCE_IO) && + (dev->resource[i].start > 0x10000000) ) { printk("Relocating PCI address %lx -> %lx\n", dev->resource[i].start, @@ -1029,6 +884,8 @@ prep_pcibios_fixup(void) pci_write_config_dword(dev, PCI_BASE_ADDRESS_0+(i*0x4), dev->resource[i].start ); + dev->resource[i].end = + (dev->resource[i].end & 0x00FFFFFF) | 0x01000000; } } #if 0 @@ -1043,49 +900,50 @@ prep_pcibios_fixup(void) } } -decl_config_access_method(indirect); - void __init -prep_setup_pci_ptrs(void) +prep_find_bridges(void) { - PPC_DEVICE *hostbridge; - - printk("PReP architecture\n"); - if ( _prep_type == _PREP_Radstone ) - { - pci_config_address = (unsigned *)0x80000cf8; - pci_config_data = (char *)0x80000cfc; - set_config_access_method(indirect); - } - else - { - hostbridge = residual_find_device(PROCESSORDEVICE, NULL, - BridgeController, PCIBridge, -1, 0); - if (hostbridge && - hostbridge->DeviceId.Interface == PCIBridgeIndirect) { - PnP_TAG_PACKET * pkt; - set_config_access_method(indirect); - pkt = PnP_find_large_vendor_packet( + struct pci_controller* hose; + + hose = pcibios_alloc_controller(); + if (!hose) + return; + + hose->first_busno = 0; + hose->last_busno = 0xff; + hose->pci_mem_offset = PREP_ISA_MEM_BASE; + + printk("PReP architecture\n"); + { +#ifdef CONFIG_PREP_RESIDUAL + PPC_DEVICE *hostbridge; + + hostbridge = residual_find_device(PROCESSORDEVICE, NULL, + BridgeController, PCIBridge, -1, 0); + if (hostbridge && + hostbridge->DeviceId.Interface == PCIBridgeIndirect) { + PnP_TAG_PACKET * pkt; + pkt = PnP_find_large_vendor_packet( res->DevicePnPHeap+hostbridge->AllocatedOffset, 3, 0); - if(pkt) + if(pkt) { #define p pkt->L4_Pack.L4_Data.L4_PPCPack - pci_config_address= (unsigned *)ld_le32((unsigned *) p.PPCData); - pci_config_data= (unsigned char *)ld_le32((unsigned *) (p.PPCData+8)); - } + setup_indirect_pci(hose, + ld_le32((unsigned *) (p.PPCData)), + ld_le32((unsigned *) (p.PPCData+8))); + } else { - pci_config_address= (unsigned *) 0x80000cf8; - pci_config_data= (unsigned char *) 0x80000cfc; - } - } + setup_indirect_pci(hose, 0x80000cf8, 0x80000cfc); + } + } else +#endif /* CONFIG_PREP_RESIDUAL */ { - set_config_access_method(prep); - } - - } + hose->ops = &prep_pci_ops; + } + } ppc_md.pcibios_fixup = prep_pcibios_fixup; } diff --git a/arch/ppc/kernel/prep_setup.c b/arch/ppc/kernel/prep_setup.c index 6e30462aa..64ad519cb 100644 --- a/arch/ppc/kernel/prep_setup.c +++ b/arch/ppc/kernel/prep_setup.c @@ -32,7 +32,6 @@ #include <linux/console.h> #include <linux/timex.h> #include <linux/pci.h> -#include <linux/openpic.h> #include <linux/ide.h> #include <asm/init.h> @@ -48,8 +47,9 @@ #include <asm/prep_nvram.h> #include <asm/raven.h> #include <asm/keyboard.h> - +#include <asm/vga.h> #include <asm/time.h> + #include "local_irq.h" #include "i8259.h" #include "open_pic.h" @@ -84,7 +84,7 @@ extern void pckbd_leds(unsigned char leds); extern void pckbd_init_hw(void); extern unsigned char pckbd_sysrq_xlate[128]; -extern void prep_setup_pci_ptrs(void); +extern void prep_find_bridges(void); extern char saved_command_line[256]; int _prep_type; @@ -101,23 +101,23 @@ unsigned long empty_zero_page[1024]; extern PTE *Hash, *Hash_end; extern unsigned long Hash_size, Hash_mask; extern int probingmem; -extern unsigned long loops_per_sec; +extern unsigned long loops_per_jiffy; #ifdef CONFIG_BLK_DEV_RAM extern int rd_doload; /* 1 = load ramdisk, 0 = don't load */ extern int rd_prompt; /* 1 = prompt for ramdisk, 0 = don't prompt */ extern int rd_image_start; /* starting block # of image */ #endif -#ifdef CONFIG_VGA_CONSOLE -unsigned long vgacon_remap_base; -#endif int __prep prep_get_cpuinfo(char *buffer) { extern char *Motherboard_map_name; - int len, i; - + int len; +#ifdef CONFIG_PREP_RESIDUAL + int i; +#endif + #ifdef CONFIG_SMP #define CD(X) (cpu_data[n].X) #else @@ -190,7 +190,10 @@ prep_get_cpuinfo(char *buffer) } -no_l2: +no_l2: +#ifndef CONFIG_PREP_RESIDUAL + return len; +#else if ( res->ResidualLength == 0 ) return len; @@ -205,8 +208,8 @@ no_l2: res->Memories[i].SIMMSize); } len += sprintf(buffer+len,"\n"); - return len; +#endif } void __init @@ -214,11 +217,16 @@ prep_setup_arch(void) { extern char cmd_line[]; unsigned char reg; +#if 0 /* unused?? */ unsigned char ucMothMemType; unsigned char ucEquipPres1; +#endif /* init to some ~sane value until calibrate_delay() runs */ - loops_per_sec = 50000000; + loops_per_jiffy = 50000000; + + /* Lookup PCI host bridges */ + prep_find_bridges(); /* Set up floppy in PS/2 mode */ outb(0x09, SIO_CONFIG_RA); @@ -247,41 +255,6 @@ prep_setup_arch(void) *(unsigned char *)(0x8000081c) |= 3; ROOT_DEV = to_kdev_t(0x0802); /* sda2 */ break; - case _PREP_Radstone: - ROOT_DEV = to_kdev_t(0x0801); /* sda1 */ - - /* - * Determine system type - */ - ucMothMemType=inb(0x866); - ucEquipPres1=inb(0x80c); - - ucSystemType=((ucMothMemType&0x03)<<1) | - ((ucEquipPres1&0x80)>>7); - ucSystemType^=7; - - /* - * Determine board revision for use by - * rev. specific code - */ - ucBoardRev=inb(0x854); - ucBoardRevMaj=ucBoardRev>>5; - ucBoardRevMin=ucBoardRev&0x1f; - - /* - * Most Radstone boards have memory mapped NvRAM - */ - if((ucSystemType==RS_SYS_TYPE_PPC1) && (ucBoardRevMaj<5)) - { - ppc_md.nvram_read_val = prep_nvram_read_val; - ppc_md.nvram_write_val = prep_nvram_write_val; - } - else - { - ppc_md.nvram_read_val = rs_nvram_read_val; - ppc_md.nvram_write_val = rs_nvram_write_val; - } - break; } /* Read in NVRAM data */ @@ -341,12 +314,6 @@ prep_setup_arch(void) #endif /* CONFIG_SOUND_CS4232 */ /*print_residual_device_info();*/ - request_region(0x20,0x20,"pic1"); - request_region(0xa0,0x20,"pic2"); - request_region(0x00,0x20,"dma1"); - request_region(0x40,0x20,"timer"); - request_region(0x80,0x10,"dma page reg"); - request_region(0xc0,0x20,"dma2"); raven_init(); @@ -365,6 +332,7 @@ prep_setup_arch(void) */ void __init prep_res_calibrate_decr(void) { +#ifdef CONFIG_PREP_RESIDUAL unsigned long freq, divisor=4; freq = res->VitalProductData.ProcessorBusHz; @@ -372,6 +340,7 @@ void __init prep_res_calibrate_decr(void) (freq/divisor)/1000000, (freq/divisor)%1000000); tb_ticks_per_jiffy = freq / HZ / divisor; tb_to_us = mulhwu_scale_factor(freq/divisor, 1000000); +#endif } /* @@ -585,15 +554,16 @@ prep_setup_residual(char *buffer) { int len = 0; - /* PREP's without residual data will give incorrect values here */ len += sprintf(len+buffer, "clock\t\t: "); +#ifdef CONFIG_PREP_RESIDUAL if ( res->ResidualLength ) len += sprintf(len+buffer, "%ldMHz\n", (res->VitalProductData.ProcessorHz > 1024) ? res->VitalProductData.ProcessorHz>>20 : res->VitalProductData.ProcessorHz); else +#endif /* CONFIG_PREP_RESIDUAL */ len += sprintf(len+buffer, "???\n"); return len; @@ -640,19 +610,11 @@ prep_init_IRQ(void) { int i; - if (OpenPIC != NULL) { - for ( i = 16 ; i < 36 ; i++ ) - irq_desc[i].handler = &open_pic; - openpic_init(1); - } - - for ( i = 0 ; i < 16 ; i++ ) + if (OpenPIC_Addr != NULL) + openpic_init(1, NUM_8259_INTERRUPTS, 0, -1); + for ( i = 0 ; i < NUM_8259_INTERRUPTS ; i++ ) irq_desc[i].handler = &i8259_pic; - i8259_init(); -#ifdef CONFIG_SMP - request_irq(openpic_to_irq(OPENPIC_VEC_SPURIOUS), openpic_ipi_action, - 0, "IPI0", 0); -#endif /* CONFIG_SMP */ + i8259_init(); } #if defined(CONFIG_BLK_DEV_IDE) || defined(CONFIG_BLK_DEV_IDE_MODULE) @@ -708,19 +670,14 @@ prep_ide_request_region(ide_ioreg_t from, unsigned int extent, const char *name) { - request_region(from, extent, name); + request_region(from, extent, name); } void __prep prep_ide_release_region(ide_ioreg_t from, unsigned int extent) { - release_region(from, extent); -} - -void __prep -prep_ide_fix_driveid(struct hd_driveid *id) -{ + release_region(from, extent); } void __init @@ -743,17 +700,52 @@ prep_ide_init_hwif_ports (hw_regs_t *hw, ide_ioreg_t data_port, ide_ioreg_t ctrl } #endif +unsigned long *MotSave_SmpIar; +unsigned char *MotSave_CpusState[2]; + +void __init +prep_init2(void) +{ +#ifdef CONFIG_NVRAM + request_region(PREP_NVRAM_AS0, 0x8, "nvram"); +#endif + request_region(0x20,0x20,"pic1"); + request_region(0xa0,0x20,"pic2"); + request_region(0x00,0x20,"dma1"); + request_region(0x40,0x20,"timer"); + request_region(0x80,0x10,"dma page reg"); + request_region(0xc0,0x20,"dma2"); +} + void __init prep_init(unsigned long r3, unsigned long r4, unsigned long r5, unsigned long r6, unsigned long r7) { +#ifdef CONFIG_PREP_RESIDUAL + RESIDUAL *old_res = (RESIDUAL *)(r3 + KERNELBASE); + /* make a copy of residual data */ if ( r3 ) { memcpy((void *)res,(void *)(r3+KERNELBASE), sizeof(RESIDUAL)); + + /* These need to be saved for the Motorola Prep + * MVME4600 and Dual MTX boards. + */ + MotSave_SmpIar = &old_res->VitalProductData.SmpIar; + MotSave_CpusState[0] = &old_res->Cpus[0].CpuState; + MotSave_CpusState[1] = &old_res->Cpus[1].CpuState; } +#endif + /* Copy cmd_line parameters */ + if ( r6) + { + *(char *)(r7 + KERNELBASE) = 0; + strcpy(cmd_line, (char *)(r6 + KERNELBASE)); + } + isa_io_base = PREP_ISA_IO_BASE; isa_mem_base = PREP_ISA_MEM_BASE; pci_dram_offset = PREP_PCI_DRAM_OFFSET; @@ -762,29 +754,19 @@ prep_init(unsigned long r3, unsigned long r4, unsigned long r5, DMA_MODE_WRITE = 0x48; /* figure out what kind of prep workstation we are */ +#ifdef CONFIG_PREP_RESIDUAL if ( res->ResidualLength != 0 ) { if ( !strncmp(res->VitalProductData.PrintableModel,"IBM",3) ) _prep_type = _PREP_IBM; - else if (!strncmp(res->VitalProductData.PrintableModel, - "Radstone",8)) - { - extern char *Motherboard_map_name; - - _prep_type = _PREP_Radstone; - Motherboard_map_name= - res->VitalProductData.PrintableModel; - } - else - _prep_type = _PREP_Motorola; + _prep_type = _PREP_Motorola; } else /* assume motorola if no residual (netboot?) */ +#endif { _prep_type = _PREP_Motorola; } - prep_setup_pci_ptrs(); - ppc_md.setup_arch = prep_setup_arch; ppc_md.setup_residual = prep_setup_residual; ppc_md.get_cpuinfo = prep_get_cpuinfo; @@ -792,40 +774,14 @@ prep_init(unsigned long r3, unsigned long r4, unsigned long r5, ppc_md.init_IRQ = prep_init_IRQ; /* this gets changed later on if we have an OpenPIC -- Cort */ ppc_md.get_irq = prep_get_irq; - ppc_md.init = NULL; + ppc_md.init = prep_init2; ppc_md.restart = prep_restart; ppc_md.power_off = prep_power_off; ppc_md.halt = prep_halt; ppc_md.time_init = NULL; - if (_prep_type == _PREP_Radstone) { - /* - * We require a direct restart as port 92 does not work on - * all Radstone boards - */ - ppc_md.restart = prep_direct_restart; - /* - * The RTC device used varies according to board type - */ - if(((ucSystemType==RS_SYS_TYPE_PPC1) && (ucBoardRevMaj>=5)) || - (ucSystemType==RS_SYS_TYPE_PPC1a)) - { - ppc_md.set_rtc_time = mk48t59_set_rtc_time; - ppc_md.get_rtc_time = mk48t59_get_rtc_time; - ppc_md.time_init = mk48t59_init; - } - else - { - ppc_md.set_rtc_time = mc146818_set_rtc_time; - ppc_md.get_rtc_time = mc146818_get_rtc_time; - } - /* - * Determine the decrementer rate from the residual data - */ - ppc_md.calibrate_decr = prep_res_calibrate_decr; - } - else if (_prep_type == _PREP_IBM) { + if (_prep_type == _PREP_IBM) { ppc_md.set_rtc_time = mc146818_set_rtc_time; ppc_md.get_rtc_time = mc146818_get_rtc_time; ppc_md.calibrate_decr = prep_calibrate_decr; @@ -845,7 +801,7 @@ prep_init(unsigned long r3, unsigned long r4, unsigned long r5, ppc_ide_md.ide_check_region = prep_ide_check_region; ppc_ide_md.ide_request_region = prep_ide_request_region; ppc_ide_md.ide_release_region = prep_ide_release_region; - ppc_ide_md.fix_driveid = prep_ide_fix_driveid; + ppc_ide_md.fix_driveid = NULL; ppc_ide_md.ide_init_hwif = prep_ide_init_hwif_ports; #endif ppc_ide_md.io_base = _IO_BASE; diff --git a/arch/ppc/kernel/proc_rtas.c b/arch/ppc/kernel/proc_rtas.c new file mode 100644 index 000000000..2e8144204 --- /dev/null +++ b/arch/ppc/kernel/proc_rtas.c @@ -0,0 +1,784 @@ +/* + * arch/ppc/kernel/proc_rtas.c + * Copyright (C) 2000 Tilmann Bitterberg + * (tilmann@bitterberg.de) + * + * RTAS (Runtime Abstraction Services) stuff + * Intention is to provide a clean user interface + * to use the RTAS. + * + * TODO: + * Split off a header file and maybe move it to a different + * location. Write Documentation on what the /proc/rtas/ entries + * actually do. + */ + +#include <linux/errno.h> +#include <linux/sched.h> +#include <linux/proc_fs.h> +#include <linux/stat.h> +#include <linux/ctype.h> +#include <linux/time.h> +#include <linux/string.h> + +#include <asm/uaccess.h> +#include <asm/bitops.h> +#include <asm/processor.h> +#include <asm/io.h> +#include <asm/prom.h> +#include <asm/machdep.h> /* for ppc_md */ +#include <asm/time.h> + +/* Token for Sensors */ +#define KEY_SWITCH 0x0001 +#define ENCLOSURE_SWITCH 0x0002 +#define THERMAL_SENSOR 0x0003 +#define LID_STATUS 0x0004 +#define POWER_SOURCE 0x0005 +#define BATTERY_VOLTAGE 0x0006 +#define BATTERY_REMAINING 0x0007 +#define BATTERY_PERCENTAGE 0x0008 +#define EPOW_SENSOR 0x0009 +#define BATTERY_CYCLESTATE 0x000a +#define BATTERY_CHARGING 0x000b + +/* IBM specific sensors */ +#define IBM_SURVEILLANCE 0x2328 /* 9000 */ +#define IBM_FANRPM 0x2329 /* 9001 */ +#define IBM_VOLTAGE 0x232a /* 9002 */ +#define IBM_DRCONNECTOR 0x232b /* 9003 */ +#define IBM_POWERSUPPLY 0x232c /* 9004 */ +#define IBM_INTQUEUE 0x232d /* 9005 */ + +/* Status return values */ +#define SENSOR_CRITICAL_HIGH 13 +#define SENSOR_WARNING_HIGH 12 +#define SENSOR_NORMAL 11 +#define SENSOR_WARNING_LOW 10 +#define SENSOR_CRITICAL_LOW 9 +#define SENSOR_SUCCESS 0 +#define SENSOR_HW_ERROR -1 +#define SENSOR_BUSY -2 +#define SENSOR_NOT_EXIST -3 +#define SENSOR_DR_ENTITY -9000 + +/* Location Codes */ +#define LOC_SCSI_DEV_ADDR 'A' +#define LOC_SCSI_DEV_LOC 'B' +#define LOC_CPU 'C' +#define LOC_DISKETTE 'D' +#define LOC_ETHERNET 'E' +#define LOC_FAN 'F' +#define LOC_GRAPHICS 'G' +/* reserved / not used 'H' */ +#define LOC_IO_ADAPTER 'I' +/* reserved / not used 'J' */ +#define LOC_KEYBOARD 'K' +#define LOC_LCD 'L' +#define LOC_MEMORY 'M' +#define LOC_NV_MEMORY 'N' +#define LOC_MOUSE 'O' +#define LOC_PLANAR 'P' +#define LOC_OTHER_IO 'Q' +#define LOC_PARALLEL 'R' +#define LOC_SERIAL 'S' +#define LOC_DEAD_RING 'T' +#define LOC_RACKMOUNTED 'U' /* for _u_nit is rack mounted */ +#define LOC_VOLTAGE 'V' +#define LOC_SWITCH_ADAPTER 'W' +#define LOC_OTHER 'X' +#define LOC_FIRMWARE 'Y' +#define LOC_SCSI 'Z' + +/* Tokens for indicators */ +#define TONE_FREQUENCY 0x0001 /* 0 - 1000 (HZ)*/ +#define TONE_VOLUME 0x0002 /* 0 - 100 (%) */ +#define SYSTEM_POWER_STATE 0x0003 +#define WARNING_LIGHT 0x0004 +#define DISK_ACTIVITY_LIGHT 0x0005 +#define HEX_DISPLAY_UNIT 0x0006 +#define BATTERY_WARNING_TIME 0x0007 +#define CONDITION_CYCLE_REQUEST 0x0008 +#define SURVEILLANCE_INDICATOR 0x2328 /* 9000 */ +#define DR_ACTION 0x2329 /* 9001 */ +#define DR_INDICATOR 0x232a /* 9002 */ +/* 9003 - 9004: Vendor specific */ +#define GLOBAL_INTERRUPT_QUEUE 0x232d /* 9005 */ +/* 9006 - 9999: Vendor specific */ + +/* other */ +#define MAX_SENSORS 17 /* I only know of 17 sensors */ +#define MAX_LINELENGTH 256 +#define SENSOR_PREFIX "ibm,sensor-" +#define cel_to_fahr(x) ((x*9/5)+32) + + +/* Globals */ +static struct proc_dir_entry *proc_rtas; +static struct rtas_sensors sensors; +static struct device_node *rtas; +static unsigned long power_on_time = 0; /* Save the time the user set */ +static char progress_led[MAX_LINELENGTH]; + +static unsigned long rtas_tone_frequency = 1000; +static unsigned long rtas_tone_volume = 0; + +/* ****************STRUCTS******************************************* */ +struct individual_sensor { + unsigned int token; + unsigned int quant; +}; + +struct rtas_sensors { + struct individual_sensor sensor[MAX_SENSORS]; + unsigned int quant; +}; + +/* ****************************************************************** */ +/* Declarations */ +static int ppc_rtas_sensor_read(char * buf, char ** start, off_t off, + int count, int *eof, void *data); +static ssize_t ppc_rtas_clock_read(struct file * file, char * buf, + size_t count, loff_t *ppos); +static ssize_t ppc_rtas_clock_write(struct file * file, const char * buf, + size_t count, loff_t *ppos); +static ssize_t ppc_rtas_progress_read(struct file * file, char * buf, + size_t count, loff_t *ppos); +static ssize_t ppc_rtas_progress_write(struct file * file, const char * buf, + size_t count, loff_t *ppos); +static ssize_t ppc_rtas_poweron_read(struct file * file, char * buf, + size_t count, loff_t *ppos); +static ssize_t ppc_rtas_poweron_write(struct file * file, const char * buf, + size_t count, loff_t *ppos); + +static ssize_t ppc_rtas_tone_freq_write(struct file * file, const char * buf, + size_t count, loff_t *ppos); +static ssize_t ppc_rtas_tone_freq_read(struct file * file, char * buf, + size_t count, loff_t *ppos); +static ssize_t ppc_rtas_tone_volume_write(struct file * file, const char * buf, + size_t count, loff_t *ppos); +static ssize_t ppc_rtas_tone_volume_read(struct file * file, char * buf, + size_t count, loff_t *ppos); + +struct file_operations ppc_rtas_poweron_operations = { + read: ppc_rtas_poweron_read, + write: ppc_rtas_poweron_write +}; +struct file_operations ppc_rtas_progress_operations = { + read: ppc_rtas_progress_read, + write: ppc_rtas_progress_write +}; + +struct file_operations ppc_rtas_clock_operations = { + read: ppc_rtas_clock_read, + write: ppc_rtas_clock_write +}; + +struct file_operations ppc_rtas_tone_freq_operations = { + read: ppc_rtas_tone_freq_read, + write: ppc_rtas_tone_freq_write +}; +struct file_operations ppc_rtas_tone_volume_operations = { + read: ppc_rtas_tone_volume_read, + write: ppc_rtas_tone_volume_write +}; + +int ppc_rtas_find_all_sensors (void); +int ppc_rtas_process_sensor(struct individual_sensor s, int state, + int error, char * buf); +char * ppc_rtas_process_error(int error); +int get_location_code(struct individual_sensor s, char * buf); +int check_location_string (char *c, char * buf); +int check_location (char *c, int idx, char * buf); + +/* ****************************************************************** */ +/* MAIN */ +/* ****************************************************************** */ +void proc_rtas_init(void) +{ + struct proc_dir_entry *entry; + + rtas = find_devices("rtas"); + if ((rtas == 0) || (_machine != _MACH_chrp)) { + return; + } + + proc_rtas = proc_mkdir("rtas", 0); + if (proc_rtas == 0) + return; + + /* /proc/rtas entries */ + + entry = create_proc_entry("progress", S_IRUGO|S_IWUSR, proc_rtas); + if (entry) entry->proc_fops = &ppc_rtas_progress_operations; + + entry = create_proc_entry("clock", S_IRUGO|S_IWUSR, proc_rtas); + if (entry) entry->proc_fops = &ppc_rtas_clock_operations; + + entry = create_proc_entry("poweron", S_IWUSR|S_IRUGO, proc_rtas); + if (entry) entry->proc_fops = &ppc_rtas_poweron_operations; + + create_proc_read_entry("sensors", S_IRUGO, proc_rtas, + ppc_rtas_sensor_read, NULL); + + entry = create_proc_entry("frequency", S_IWUSR|S_IRUGO, proc_rtas); + if (entry) entry->proc_fops = &ppc_rtas_tone_freq_operations; + + entry = create_proc_entry("volume", S_IWUSR|S_IRUGO, proc_rtas); + if (entry) entry->proc_fops = &ppc_rtas_tone_volume_operations; +} + +/* ****************************************************************** */ +/* POWER-ON-TIME */ +/* ****************************************************************** */ +static ssize_t ppc_rtas_poweron_write(struct file * file, const char * buf, + size_t count, loff_t *ppos) +{ + struct rtc_time tm; + unsigned long nowtime; + char *dest; + int error; + + nowtime = simple_strtoul(buf, &dest, 10); + if (*dest != '\0' && *dest != '\n') { + printk("ppc_rtas_poweron_write: Invalid time\n"); + return count; + } + power_on_time = nowtime; /* save the time */ + + to_tm(nowtime, &tm); + + error = call_rtas("set-time-for-power-on", 7, 1, NULL, + tm.tm_year, tm.tm_mon, tm.tm_mday, + tm.tm_hour, tm.tm_min, tm.tm_sec, 0 /* nano */); + if (error != 0) + printk(KERN_WARNING "error: setting poweron time returned: %s\n", + ppc_rtas_process_error(error)); + return count; +} +/* ****************************************************************** */ +static ssize_t ppc_rtas_poweron_read(struct file * file, char * buf, + size_t count, loff_t *ppos) +{ + int n; + if (power_on_time == 0) + n = sprintf(buf, "Power on time not set\n"); + else + n = sprintf(buf, "%lu\n", power_on_time); + + if (*ppos >= strlen(buf)) + return 0; + if (n > strlen(buf) - *ppos) + n = strlen(buf) - *ppos; + if (n > count) + n = count; + *ppos += n; + return n; +} + +/* ****************************************************************** */ +/* PROGRESS */ +/* ****************************************************************** */ +static ssize_t ppc_rtas_progress_write(struct file * file, const char * buf, + size_t count, loff_t *ppos) +{ + unsigned long hex; + + strcpy(progress_led, buf); /* save the string */ + /* Lets see if the user passed hexdigits */ + hex = simple_strtoul(buf, NULL, 10); + + ppc_md.progress ((char *)buf, hex); + return count; + + /* clear the line */ /* ppc_md.progress(" ", 0xffff);*/ +} +/* ****************************************************************** */ +static ssize_t ppc_rtas_progress_read(struct file * file, char * buf, + size_t count, loff_t *ppos) +{ + int n = 0; + if (progress_led != NULL) + n = sprintf (buf, "%s\n", progress_led); + if (*ppos >= strlen(buf)) + return 0; + if (n > strlen(buf) - *ppos) + n = strlen(buf) - *ppos; + if (n > count) + n = count; + *ppos += n; + return n; +} + +/* ****************************************************************** */ +/* CLOCK */ +/* ****************************************************************** */ +static ssize_t ppc_rtas_clock_write(struct file * file, const char * buf, + size_t count, loff_t *ppos) +{ + struct rtc_time tm; + unsigned long nowtime; + char *dest; + int error; + + nowtime = simple_strtoul(buf, &dest, 10); + if (*dest != '\0' && *dest != '\n') { + printk("ppc_rtas_clock_write: Invalid time\n"); + return count; + } + + to_tm(nowtime, &tm); + error = call_rtas("set-time-of-day", 7, 1, NULL, + tm.tm_year, tm.tm_mon, tm.tm_mday, + tm.tm_hour, tm.tm_min, tm.tm_sec, 0); + if (error != 0) + printk(KERN_WARNING "error: setting the clock returned: %s\n", + ppc_rtas_process_error(error)); + return count; +} +/* ****************************************************************** */ +static ssize_t ppc_rtas_clock_read(struct file * file, char * buf, + size_t count, loff_t *ppos) +{ + unsigned int year, mon, day, hour, min, sec; + unsigned long *ret = kmalloc(4*8, GFP_KERNEL); + int n, error; + + error = call_rtas("get-time-of-day", 0, 8, ret); + + year = ret[0]; mon = ret[1]; day = ret[2]; + hour = ret[3]; min = ret[4]; sec = ret[5]; + + if (error != 0){ + printk(KERN_WARNING "error: reading the clock returned: %s\n", + ppc_rtas_process_error(error)); + n = sprintf (buf, "0"); + } else { + n = sprintf (buf, "%lu\n", mktime(year, mon, day, hour, min, sec)); + } + kfree(ret); + + if (*ppos >= strlen(buf)) + return 0; + if (n > strlen(buf) - *ppos) + n = strlen(buf) - *ppos; + if (n > count) + n = count; + *ppos += n; + return n; +} + +/* ****************************************************************** */ +/* SENSOR STUFF */ +/* ****************************************************************** */ +static int ppc_rtas_sensor_read(char * buf, char ** start, off_t off, + int count, int *eof, void *data) +{ + int i,j,n; + unsigned long ret; + int state, error; + char buffer[MAX_LINELENGTH*MAX_SENSORS]; /* May not be enough */ + + if (count < 0) + return -EINVAL; + + n = sprintf ( buffer , "RTAS (RunTime Abstraction Services) Sensor Information\n"); + n += sprintf ( buffer+n, "Sensor\t\tValue\t\tCondition\tLocation\n"); + n += sprintf ( buffer+n, "********************************************************\n"); + + if (ppc_rtas_find_all_sensors() != 0) { + n += sprintf ( buffer+n, "\nNo sensors are available\n"); + goto return_string; + } + + for (i=0; i<sensors.quant; i++) { + j = sensors.sensor[i].quant; + /* A sensor may have multiple instances */ + while (j >= 0) { + error = call_rtas("get-sensor-state", 2, 2, &ret, + sensors.sensor[i].token, sensors.sensor[i].quant-j); + state = (int) ret; + n += ppc_rtas_process_sensor(sensors.sensor[i], state, error, buffer+n ); + n += sprintf (buffer+n, "\n"); + j--; + } /* while */ + } /* for */ + +return_string: + if (off >= strlen(buffer)) { + *eof = 1; + return 0; + } + if (n > strlen(buffer) - off) + n = strlen(buffer) - off; + if (n > count) + n = count; + else + *eof = 1; + memcpy(buf, buffer + off, n); + *start = buf; + return n; +} + +/* ****************************************************************** */ + +int ppc_rtas_find_all_sensors (void) +{ + unsigned long *utmp; + int len, i, j; + + utmp = (unsigned long *) get_property(rtas, "rtas-sensors", &len); + if (utmp == NULL) { + printk (KERN_ERR "error: could not get rtas-sensors\n"); + return 1; + } + + sensors.quant = len / 8; /* int + int */ + + for (i=0, j=0; j<sensors.quant; i+=2, j++) { + sensors.sensor[j].token = utmp[i]; + sensors.sensor[j].quant = utmp[i+1]; + } + return 0; +} + +/* ****************************************************************** */ +/* + * Builds a string of what rtas returned + */ +char * ppc_rtas_process_error(int error) +{ + switch (error) { + case SENSOR_CRITICAL_HIGH: + return "(critical high)"; + case SENSOR_WARNING_HIGH: + return "(warning high)"; + case SENSOR_NORMAL: + return "(normal)"; + case SENSOR_WARNING_LOW: + return "(warning low)"; + case SENSOR_CRITICAL_LOW: + return "(critical low)"; + case SENSOR_SUCCESS: + return "(read ok)"; + case SENSOR_HW_ERROR: + return "(hardware error)"; + case SENSOR_BUSY: + return "(busy)"; + case SENSOR_NOT_EXIST: + return "(non existant)"; + case SENSOR_DR_ENTITY: + return "(dr entity removed)"; + default: + return "(UNKNOWN)"; + } +} + +/* ****************************************************************** */ +/* + * Builds a string out of what the sensor said + */ + +int ppc_rtas_process_sensor(struct individual_sensor s, int state, + int error, char * buf) +{ + /* Defined return vales */ + const char * key_switch[] = { "Off\t", "Normal\t", "Secure\t", "Mainenance" }; + const char * enclosure_switch[] = { "Closed", "Open" }; + const char * lid_status[] = { " ", "Open", "Closed" }; + const char * power_source[] = { "AC\t", "Battery", "AC & Battery" }; + const char * battery_remaining[] = { "Very Low", "Low", "Mid", "High" }; + const char * epow_sensor[] = { + "EPOW Reset", "Cooling warning", "Power warning", + "System shutdown", "System halt", "EPOW main enclosure", + "EPOW power off" }; + const char * battery_cyclestate[] = { "None", "In progress", "Requested" }; + const char * battery_charging[] = { "Charging", "Discharching", "No current flow" }; + const char * ibm_drconnector[] = { "Empty", "Present" }; + const char * ibm_intqueue[] = { "Disabled", "Enabled" }; + + int have_strings = 0; + int temperature = 0; + int unknown = 0; + int n = 0; + + /* What kind of sensor do we have here? */ + switch (s.token) { + case KEY_SWITCH: + n += sprintf(buf+n, "Key switch:\t"); + n += sprintf(buf+n, "%s\t", key_switch[state]); + have_strings = 1; + break; + case ENCLOSURE_SWITCH: + n += sprintf(buf+n, "Enclosure switch:\t"); + n += sprintf(buf+n, "%s\t", enclosure_switch[state]); + have_strings = 1; + break; + case THERMAL_SENSOR: + n += sprintf(buf+n, "Temp. (°C/°F):\t"); + temperature = 1; + break; + case LID_STATUS: + n += sprintf(buf+n, "Lid status:\t"); + n += sprintf(buf+n, "%s\t", lid_status[state]); + have_strings = 1; + break; + case POWER_SOURCE: + n += sprintf(buf+n, "Power source:\t"); + n += sprintf(buf+n, "%s\t", power_source[state]); + have_strings = 1; + break; + case BATTERY_VOLTAGE: + n += sprintf(buf+n, "Battery voltage:\t"); + break; + case BATTERY_REMAINING: + n += sprintf(buf+n, "Battery remaining:\t"); + n += sprintf(buf+n, "%s\t", battery_remaining[state]); + have_strings = 1; + break; + case BATTERY_PERCENTAGE: + n += sprintf(buf+n, "Battery percentage:\t"); + break; + case EPOW_SENSOR: + n += sprintf(buf+n, "EPOW Sensor:\t"); + n += sprintf(buf+n, "%s\t", epow_sensor[state]); + have_strings = 1; + break; + case BATTERY_CYCLESTATE: + n += sprintf(buf+n, "Battery cyclestate:\t"); + n += sprintf(buf+n, "%s\t", battery_cyclestate[state]); + have_strings = 1; + break; + case BATTERY_CHARGING: + n += sprintf(buf+n, "Battery Charging:\t"); + n += sprintf(buf+n, "%s\t", battery_charging[state]); + have_strings = 1; + break; + case IBM_SURVEILLANCE: + n += sprintf(buf+n, "Surveillance:\t"); + break; + case IBM_FANRPM: + n += sprintf(buf+n, "Fan (rpm):\t"); + break; + case IBM_VOLTAGE: + n += sprintf(buf+n, "Voltage (mv):\t"); + break; + case IBM_DRCONNECTOR: + n += sprintf(buf+n, "DR connector:\t"); + n += sprintf(buf+n, "%s\t", ibm_drconnector[state]); + have_strings = 1; + break; + case IBM_POWERSUPPLY: + n += sprintf(buf+n, "Powersupply:\t"); + break; + case IBM_INTQUEUE: + n += sprintf(buf+n, "Interrupt queue:\t"); + n += sprintf(buf+n, "%s\t", ibm_intqueue[state]); + have_strings = 1; + break; + default: + n += sprintf(buf+n, "Unkown sensor (type %d), ignoring it\n", + s.token); + unknown = 1; + have_strings = 1; + break; + } + if (have_strings == 0) { + if (temperature) { + n += sprintf(buf+n, "%4d /%4d\t", state, cel_to_fahr(state)); + } else + n += sprintf(buf+n, "%10d\t", state); + } + if (unknown == 0) { + n += sprintf ( buf+n, "%s\t", ppc_rtas_process_error(error)); + n += get_location_code(s, buf+n); + } + return n; +} + +/* ****************************************************************** */ + +int check_location (char *c, int idx, char * buf) +{ + int n = 0; + + switch (*(c+idx)) { + case LOC_PLANAR: + n += sprintf ( buf, "Planar #%c", *(c+idx+1)); + break; + case LOC_CPU: + n += sprintf ( buf, "CPU #%c", *(c+idx+1)); + break; + case LOC_FAN: + n += sprintf ( buf, "Fan #%c", *(c+idx+1)); + break; + case LOC_RACKMOUNTED: + n += sprintf ( buf, "Rack #%c", *(c+idx+1)); + break; + case LOC_VOLTAGE: + n += sprintf ( buf, "Voltage #%c", *(c+idx+1)); + break; + case LOC_LCD: + n += sprintf ( buf, "LCD #%c", *(c+idx+1)); + break; + case '.': + n += sprintf ( buf, "- %c", *(c+idx+1)); + default: + n += sprintf ( buf, "Unknown location"); + break; + } + return n; +} + + +/* ****************************************************************** */ +/* + * Format: + * ${LETTER}${NUMBER}[[-/]${LETTER}${NUMBER} [ ... ] ] + * the '.' may be an abbrevation + */ +int check_location_string (char *c, char *buf) +{ + int n=0,i=0; + + while (c[i]) { + if (isalpha(c[i]) || c[i] == '.') { + n += check_location(c, i, buf+n); + } + else if (c[i] == '/' || c[i] == '-') + n += sprintf(buf+n, " at "); + i++; + } + return n; +} + + +/* ****************************************************************** */ + +int get_location_code(struct individual_sensor s, char * buffer) +{ + char rstr[512], tmp[10], tmp2[10]; + int n=0, i=0, llen, len; + /* char *buf = kmalloc(MAX_LINELENGTH, GFP_KERNEL); */ + char *ret; + + static int pos = 0; /* remember position where buffer was */ + + /* construct the sensor number like 0003 */ + /* fill with zeros */ + n = sprintf(tmp, "%d", s.token); + len = strlen(tmp); + while (strlen(tmp) < 4) + n += sprintf (tmp+n, "0"); + + /* invert the string */ + while (tmp[i]) { + if (i<len) + tmp2[4-len+i] = tmp[i]; + else + tmp2[3-i] = tmp[i]; + i++; + } + tmp2[4] = '\0'; + + sprintf (rstr, SENSOR_PREFIX"%s", tmp2); + + ret = (char *) get_property(rtas, rstr, &llen); + + n=0; + if (ret[0] == '\0') + n += sprintf ( buffer+n, "--- ");/* does not have a location */ + else { + char t[50]; + ret += pos; + + n += check_location_string(ret, buffer + n); + n += sprintf ( buffer+n, " "); + /* see how many characters we have printed */ + sprintf ( t, "%s ", ret); + + pos += strlen(t); + if (pos >= llen) pos=0; + } + return n; +} +/* ****************************************************************** */ +/* INDICATORS - Tone Frequency */ +/* ****************************************************************** */ +static ssize_t ppc_rtas_tone_freq_write(struct file * file, const char * buf, + size_t count, loff_t *ppos) +{ + unsigned long freq; + char *dest; + int error; + freq = simple_strtoul(buf, &dest, 10); + if (*dest != '\0' && *dest != '\n') { + printk("ppc_rtas_tone_freq_write: Invalid tone freqency\n"); + return count; + } + if (freq < 0) freq = 0; + rtas_tone_frequency = freq; /* save it for later */ + error = call_rtas("set-indicator", 3, 1, NULL, + TONE_FREQUENCY, 0, freq); + if (error != 0) + printk(KERN_WARNING "error: setting tone frequency returned: %s\n", + ppc_rtas_process_error(error)); + return count; +} +/* ****************************************************************** */ +static ssize_t ppc_rtas_tone_freq_read(struct file * file, char * buf, + size_t count, loff_t *ppos) +{ + int n; + n = sprintf(buf, "%lu\n", rtas_tone_frequency); + + if (*ppos >= strlen(buf)) + return 0; + if (n > strlen(buf) - *ppos) + n = strlen(buf) - *ppos; + if (n > count) + n = count; + *ppos += n; + return n; +} +/* ****************************************************************** */ +/* INDICATORS - Tone Volume */ +/* ****************************************************************** */ +static ssize_t ppc_rtas_tone_volume_write(struct file * file, const char * buf, + size_t count, loff_t *ppos) +{ + unsigned long volume; + char *dest; + int error; + volume = simple_strtoul(buf, &dest, 10); + if (*dest != '\0' && *dest != '\n') { + printk("ppc_rtas_tone_volume_write: Invalid tone volume\n"); + return count; + } + if (volume < 0) volume = 0; + if (volume > 100) volume = 100; + + rtas_tone_volume = volume; /* save it for later */ + error = call_rtas("set-indicator", 3, 1, NULL, + TONE_VOLUME, 0, volume); + if (error != 0) + printk(KERN_WARNING "error: setting tone volume returned: %s\n", + ppc_rtas_process_error(error)); + return count; +} +/* ****************************************************************** */ +static ssize_t ppc_rtas_tone_volume_read(struct file * file, char * buf, + size_t count, loff_t *ppos) +{ + int n; + n = sprintf(buf, "%lu\n", rtas_tone_volume); + + if (*ppos >= strlen(buf)) + return 0; + if (n > strlen(buf) - *ppos) + n = strlen(buf) - *ppos; + if (n > count) + n = count; + *ppos += n; + return n; +} diff --git a/arch/ppc/kernel/process.c b/arch/ppc/kernel/process.c index a200b1c78..c4c76adec 100644 --- a/arch/ppc/kernel/process.c +++ b/arch/ppc/kernel/process.c @@ -222,16 +222,17 @@ _switch_to(struct task_struct *prev, struct task_struct *new, giveup_fpu(prev); #ifdef CONFIG_ALTIVEC /* - * If the previous thread 1) has some altivec regs it wants saved - * (has bits in vrsave set) and 2) used altivec in the last quantum + * If the previous thread used altivec in the last quantum * (thus changing altivec regs) then save them. + * We used to check the VRSAVE register but not all apps + * set it, so we don't rely on it now (and in fact we need + * to save & restore VSCR even if VRSAVE == 0). -- paulus * * On SMP we always save/restore altivec regs just to avoid the * complexity of changing processors. * -- Cort */ - if ( (prev->thread.regs && (prev->thread.regs->msr & MSR_VEC)) && - prev->thread.vrsave ) + if ((prev->thread.regs && (prev->thread.regs->msr & MSR_VEC))) giveup_altivec(prev); #endif /* CONFIG_ALTIVEC */ current_set[smp_processor_id()] = new; @@ -251,13 +252,15 @@ void show_regs(struct pt_regs * regs) { int i; - printk("NIP: %08lX XER: %08lX LR: %08lX REGS: %p TRAP: %04lx\n", - regs->nip, regs->xer, regs->link, regs,regs->trap); + printk("NIP: %08lX XER: %08lX LR: %08lX SP: %08lX REGS: %p TRAP: %04lx\n", + regs->nip, regs->xer, regs->link, regs->gpr[1], regs,regs->trap); printk("MSR: %08lx EE: %01x PR: %01x FP: %01x ME: %01x IR/DR: %01x%01x\n", regs->msr, regs->msr&MSR_EE ? 1 : 0, regs->msr&MSR_PR ? 1 : 0, regs->msr & MSR_FP ? 1 : 0,regs->msr&MSR_ME ? 1 : 0, regs->msr&MSR_IR ? 1 : 0, regs->msr&MSR_DR ? 1 : 0); + if (regs->trap == 0x300 || regs->trap == 0x600) + printk("DAR: %08lX, DSISR: %08lX\n", regs->dar, regs->dsisr); printk("TASK = %p[%d] '%s' ", current, current->pid, current->comm); printk("Last syscall: %ld ", current->thread.last_syscall); @@ -285,7 +288,7 @@ void show_regs(struct pt_regs * regs) printk("\n"); } } -out: +out: ; } void exit_thread(void) diff --git a/arch/ppc/kernel/prom.c b/arch/ppc/kernel/prom.c index 5494f2f52..f3b39c222 100644 --- a/arch/ppc/kernel/prom.c +++ b/arch/ppc/kernel/prom.c @@ -28,10 +28,11 @@ #include <asm/smp.h> #include <asm/bootx.h> #include <asm/system.h> -#include <asm/gemini.h> #include <asm/mmu.h> #include <asm/pgtable.h> #include <asm/bitops.h> +/* for openpic_to_irq */ +#include "open_pic.h" #ifdef CONFIG_FB #include <asm/linux_logo.h> @@ -95,26 +96,28 @@ static interpret_func interpret_root_props; #define FB_MAX 8 #endif char *prom_display_paths[FB_MAX] __initdata = { 0, }; -unsigned int prom_num_displays = 0; -char *of_stdout_device = 0; +unsigned int prom_num_displays __initdata = 0; +char *of_stdout_device __initdata = 0; +ihandle prom_disp_node __initdata = 0; -prom_entry prom = 0; -ihandle prom_chosen = 0, prom_stdout = 0, prom_disp_node = 0; +prom_entry prom __initdata = 0; +ihandle prom_chosen __initdata = 0; +ihandle prom_stdout __initdata = 0; extern char *klimit; -char *bootpath = 0; -char *bootdevice = 0; +char *bootpath; +char *bootdevice; -unsigned int rtas_data = 0; /* physical pointer */ -unsigned int rtas_entry = 0; /* physical pointer */ -unsigned int rtas_size = 0; -unsigned int old_rtas = 0; +unsigned int rtas_data; /* physical pointer */ +unsigned int rtas_entry; /* physical pointer */ +unsigned int rtas_size; +unsigned int old_rtas; /* Set for a newworld machine */ -int use_of_interrupt_tree = 0; -int pmac_newworld = 0; +int use_of_interrupt_tree; +int pmac_newworld; -static struct device_node *allnodes = 0; +static struct device_node *allnodes; #ifdef CONFIG_BOOTX_TEXT @@ -134,13 +137,12 @@ static void draw_byte_32(unsigned char *bits, unsigned long *base, int rb); static void draw_byte_16(unsigned char *bits, unsigned long *base, int rb); static void draw_byte_8(unsigned char *bits, unsigned long *base, int rb); -/* We want those in data, not BSS */ -static long g_loc_X = 0; -static long g_loc_Y = 0; -static long g_max_loc_X = 0; -static long g_max_loc_Y = 0; +static int g_loc_X; +static int g_loc_Y; +static int g_max_loc_X; +static int g_max_loc_Y; -unsigned long disp_BAT[2] = {0, 0}; +unsigned long disp_BAT[2] __initdata = {0, 0}; #define cmapsz (16*256) @@ -173,10 +175,10 @@ extern unsigned long reloc_offset(void); void phys_call_rtas(int, int, int, ...); extern char cmd_line[512]; /* XXX */ -boot_infos_t *boot_infos = 0; /* init it so it's in data segment not bss */ +boot_infos_t *boot_infos; #ifdef CONFIG_BOOTX_TEXT -boot_infos_t *disp_bi = 0; -boot_infos_t fake_bi = {0,}; +boot_infos_t *disp_bi; +boot_infos_t fake_bi; #endif unsigned long dev_tree_size; @@ -195,10 +197,6 @@ unsigned long dev_tree_size; * OF calls should be done within prom_init(), and prom_init() * and all routines called within it must be careful to relocate * references as necessary. - * - * Note that the bss is cleared *after* prom_init runs, so we have - * to make sure that any static or extern variables it accesses - * are put in the data segment. */ #define PTRRELOC(x) ((typeof(x))((unsigned long)(x) + offset)) #define PTRUNRELOC(x) ((typeof(x))((unsigned long)(x) - offset)) @@ -618,6 +616,11 @@ prom_init(int r3, int r4, prom_entry pp) char *p, *d; int prom_version = 0; unsigned long phys; + extern char __bss_start, _end; + + /* First zero the BSS -- use memset, some arches don't have + * caches on yet */ + memset_io(PTRRELOC(&__bss_start),0 , &_end - &__bss_start); /* Default */ phys = offset + KERNELBASE; @@ -948,34 +951,6 @@ check_display(unsigned long mem) if ((int) call_prom(RELOC("package-to-path"), 3, 1, node, path, 255) < 0) continue; - prom_print(RELOC("opening display ")); - prom_print(path); - ih = call_prom(RELOC("open"), 1, 1, path); - if (ih == 0 || ih == (ihandle) -1) { - prom_print(RELOC("... failed\n")); - continue; - } - prom_print(RELOC("... ok\n")); - - if (RELOC(prom_disp_node) == 0) - RELOC(prom_disp_node) = node; - - /* Setup a useable color table when the appropriate - * method is available. Should update this to set-colors */ - for (i = 0; i < 32; i++) - if (prom_set_color(ih, i, RELOC(default_colors)[i*3], - RELOC(default_colors)[i*3+1], - RELOC(default_colors)[i*3+2]) != 0) - break; - -#ifdef CONFIG_FB - for (i = 0; i < LINUX_LOGO_COLORS; i++) - if (prom_set_color(ih, i + 32, - RELOC(linux_logo_red)[i], - RELOC(linux_logo_green)[i], - RELOC(linux_logo_blue)[i]) != 0) - break; -#endif /* CONFIG_FB */ /* * If this display is the device that OF is using for stdout, @@ -990,9 +965,44 @@ check_display(unsigned long mem) = RELOC(prom_display_paths[i-1]); } RELOC(prom_display_paths[i]) = PTRUNRELOC(path); + if (i == 0) + RELOC(prom_disp_node) = node; if (RELOC(prom_num_displays) >= FB_MAX) break; } + + /* + * Open the first display and set its colormap. + */ + if (RELOC(prom_num_displays) > 0) { + path = PTRRELOC(RELOC(prom_display_paths[0])); + prom_print(RELOC("opening display ")); + prom_print(path); + ih = call_prom(RELOC("open"), 1, 1, path); + if (ih == 0 || ih == (ihandle) -1) { + prom_print(RELOC("... failed\n")); + } else { + prom_print(RELOC("... ok\n")); + + /* Setup a useable color table when the appropriate + * method is available. Should update this to set-colors */ + for (i = 0; i < 32; i++) + if (prom_set_color(ih, i, RELOC(default_colors)[i*3], + RELOC(default_colors)[i*3+1], + RELOC(default_colors)[i*3+2]) != 0) + break; + +#ifdef CONFIG_FB + for (i = 0; i < LINUX_LOGO_COLORS; i++) + if (prom_set_color(ih, i + 32, + RELOC(linux_logo_red)[i], + RELOC(linux_logo_green)[i], + RELOC(linux_logo_blue)[i]) != 0) + break; +#endif /* CONFIG_FB */ + } + } + return ALIGN(mem); } @@ -1277,7 +1287,7 @@ finish_node(struct device_node *np, unsigned long mem_start, if (!strcmp(np->name, "display")) np->name = get_property(np, "compatible", 0); - if (!strcmp(np->name, "device-tree")) + if (np->parent == NULL) ifunc = interpret_root_props; else if (np->type == 0) ifunc = NULL; @@ -1370,7 +1380,7 @@ finish_node_interrupts(struct device_node *np, unsigned long mem_start) np->intrs[i].line = *interrupts++; if (cvt_irq) np->intrs[i].line = openpic_to_irq(np->intrs[i].line); - np->intrs[i].sense = 0; + np->intrs[i].sense = 1; if (isize > 1) np->intrs[i].sense = *interrupts++; for (j=2; j<isize; j++) @@ -1540,7 +1550,7 @@ interpret_pci_props(struct device_node *np, unsigned long mem_start, for (i = 0; (ml -= cell_size) >= 0; ++i) { if (imp->addr.a_hi == devfn) { np->intrs[np->n_intrs].line = imp->intr; - np->intrs[np->n_intrs].sense = 0; /* FIXME */ + np->intrs[np->n_intrs].sense = 1; /* FIXME */ ++np->n_intrs; } imp = (struct pci_intr_map *)(((unsigned int)imp) @@ -1561,7 +1571,7 @@ interpret_pci_props(struct device_node *np, unsigned long mem_start, mem_start += np->n_intrs * sizeof(struct interrupt_info); for (i = 0; i < np->n_intrs; ++i) { np->intrs[i].line = *ip++; - np->intrs[i].sense = 0; + np->intrs[i].sense = 1; } } @@ -1614,7 +1624,7 @@ interpret_dbdma_props(struct device_node *np, unsigned long mem_start, mem_start += np->n_intrs * sizeof(struct interrupt_info); for (i = 0; i < np->n_intrs; ++i) { np->intrs[i].line = *ip++; - np->intrs[i].sense = 0; + np->intrs[i].sense = 1; } } @@ -1675,7 +1685,7 @@ interpret_macio_props(struct device_node *np, unsigned long mem_start, if (keylargo) np->intrs[i].sense = *ip++; else - np->intrs[i].sense = 0; + np->intrs[i].sense = 1; } } else { /* CHRP machines */ @@ -1771,7 +1781,7 @@ interpret_root_props(struct device_node *np, unsigned long mem_start, mem_start += np->n_intrs * sizeof(struct interrupt_info); for (i = 0; i < np->n_intrs; ++i) { np->intrs[i].line = *ip++; - np->intrs[i].sense = 0; + np->intrs[i].sense = 1; } } @@ -1779,6 +1789,30 @@ interpret_root_props(struct device_node *np, unsigned long mem_start, } /* + * Work out the sense (active-low level / active-high edge) + * of each interrupt from the device tree. + */ +void __init +prom_get_irq_senses(unsigned char *senses, int off, int max) +{ + struct device_node *np; + int i, j; + + /* default to level-triggered */ + memset(senses, 1, max - off); + if (!use_of_interrupt_tree) + return; + + for (np = allnodes; np != 0; np = np->allnext) { + for (j = 0; j < np->n_intrs; j++) { + i = np->intrs[j].line; + if (i >= off && i < max) + senses[i-off] = np->intrs[j].sense; + } + } +} + +/* * Construct and return a list of the device_nodes with a given name. */ __openfirmware @@ -1818,39 +1852,6 @@ find_type_devices(const char *type) return head; } -/* Finds a device node given its PCI bus number, device number - * and function number - */ -__openfirmware -struct device_node * -find_pci_device_OFnode(unsigned char bus, unsigned char dev_fn) -{ - struct device_node* np; - unsigned int *reg; - int l; - - for (np = allnodes; np != 0; np = np->allnext) { - int in_macio = 0; - struct device_node* parent = np->parent; - while(parent) { - char *pname = (char *)get_property(parent, "name", &l); - if (pname && strcmp(pname, "mac-io") == 0) { - in_macio = 1; - break; - } - parent = parent->parent; - } - if (in_macio) - continue; - reg = (unsigned int *) get_property(np, "reg", &l); - if (reg == 0 || l < sizeof(struct reg_property)) - continue; - if (((reg[0] >> 8) & 0xff) == dev_fn && ((reg[0] >> 16) & 0xff) == bus) - break; - } - return np; -} - /* * Returns all nodes linked together */ @@ -1983,6 +1984,21 @@ get_property(struct device_node *np, const char *name, int *lenp) return 0; } +/* + * Add a property to a node + */ +__openfirmware +void +prom_add_property(struct device_node* np, struct property* prop) +{ + struct property **next = &np->properties; + + prop->next = NULL; + while (*next) + next = &(*next)->next; + *next = prop; +} + #if 0 __openfirmware void diff --git a/arch/ppc/kernel/setup.c b/arch/ppc/kernel/setup.c index 6bafa57c1..58fb87e9c 100644 --- a/arch/ppc/kernel/setup.c +++ b/arch/ppc/kernel/setup.c @@ -12,6 +12,7 @@ #include <linux/delay.h> #include <linux/blk.h> #include <linux/ide.h> +#include <linux/bootmem.h> #include <asm/init.h> #include <asm/residual.h> @@ -71,12 +72,6 @@ extern void apus_init(unsigned long r3, unsigned long r6, unsigned long r7); -extern void gemini_init(unsigned long r3, - unsigned long r4, - unsigned long r5, - unsigned long r6, - unsigned long r7); - #ifdef CONFIG_XMON extern void xmon_map_scc(void); #endif @@ -106,6 +101,10 @@ int have_of = 0; unsigned long SYSRQ_KEY; #endif /* CONFIG_MAGIC_SYSRQ */ +#ifdef CONFIG_VGA_CONSOLE +unsigned long vgacon_remap_base; +#endif + struct machdep_calls ppc_md; /* @@ -377,9 +376,9 @@ int get_cpuinfo(char *buffer) len += sprintf(len+buffer, "revision\t: %hd.%hd\n", maj, min); len += sprintf(buffer+len, "bogomips\t: %lu.%02lu\n", - (CD(loops_per_sec)+2500)/500000, - (CD(loops_per_sec)+2500)/5000 % 100); - bogosum += CD(loops_per_sec); + (CD(loops_per_jiffy)+2500)/(500000/HZ), + (CD(loops_per_jiffy)+2500)/(5000/HZ) % 100); + bogosum += CD(loops_per_jiffy); } #ifdef CONFIG_SMP @@ -549,11 +548,6 @@ identify_machine(unsigned long r3, unsigned long r4, unsigned long r5, apus_init(r3, r4, r5, r6, r7); break; #endif -#ifdef CONFIG_GEMINI - case _MACH_gemini: - gemini_init(r3, r4, r5, r6, r7); - break; -#endif default: printk("Unknown machine type in identify_machine!\n"); } @@ -673,13 +667,14 @@ __setup("l2cr=", ppc_setup_l2cr); void __init ppc_init(void) { /* clear the progress line */ - if ( ppc_md.progress ) ppc_md.progress(" ", 0xffff); + if ( ppc_md.progress ) ppc_md.progress(" ", 0xffff); if (ppc_md.init != NULL) { ppc_md.init(); } } +/* Warning, IO base is not yet inited */ void __init setup_arch(char **cmdline_p) { extern int panic_timeout; @@ -688,7 +683,7 @@ void __init setup_arch(char **cmdline_p) extern void do_init_bootmem(void); /* so udelay does something sensible, assume <= 1000 bogomips */ - loops_per_sec = 500000000; + loops_per_jiffy = 500000000 / HZ; #ifdef CONFIG_ALL_PPC feature_init(); @@ -743,10 +738,34 @@ void __init setup_arch(char **cmdline_p) ppc_md.setup_arch(); if ( ppc_md.progress ) ppc_md.progress("arch: exit", 0x3eab); +#ifdef CONFIG_PCI + /* We create the "pci-OF-bus-map" property now so it appear in the + * /proc device tree + */ + if (have_of) { + struct property* of_prop; + + of_prop = (struct property*)alloc_bootmem(sizeof(struct property) + 256); + if (of_prop && find_path_device("/")) { + memset(of_prop, -1, sizeof(struct property) + 256); + of_prop->name = "pci-OF-bus-map"; + of_prop->length = 256; + of_prop->value = (unsigned char *)&of_prop[1]; + prom_add_property(find_path_device("/"), of_prop); + } + } +#endif /* CONFIG_PCI */ + paging_init(); sort_exception_table(); } +/* Convert the shorts/longs in hd_driveid from little to big endian; + * chars are endian independant, of course, but strings need to be flipped. + * (Despite what it says in drivers/block/ide.h, they come up as little + * endian...) + * + * Changes to linux/hdreg.h may require changes here. */ void ppc_generic_ide_fix_driveid(struct hd_driveid *id) { int i; diff --git a/arch/ppc/kernel/signal.c b/arch/ppc/kernel/signal.c index dd3d1ae1b..fe16fae25 100644 --- a/arch/ppc/kernel/signal.c +++ b/arch/ppc/kernel/signal.c @@ -53,8 +53,6 @@ #define MSR_USERCHANGE (MSR_FE0 | MSR_FE1) int do_signal(sigset_t *oldset, struct pt_regs *regs); -extern int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); int copy_siginfo_to_user(siginfo_t *to, siginfo_t *from) { diff --git a/arch/ppc/kernel/smp.c b/arch/ppc/kernel/smp.c index fb7f38444..7edf7209d 100644 --- a/arch/ppc/kernel/smp.c +++ b/arch/ppc/kernel/smp.c @@ -1,6 +1,4 @@ /* - * $Id: smp.c,v 1.68 1999/09/17 19:38:05 cort Exp $ - * * Smp support for ppc. * * Written by Cort Dougan (cort@cs.nmt.edu) borrowing a great @@ -8,8 +6,11 @@ * * Copyright (C) 1999 Cort Dougan <cort@cs.nmt.edu> * - * Support for PReP (Motorola MTX/MVME) SMP by Troy Benjegerdes - * (troy@microux.com, hozer@drgw.net) + * Support for PReP (Motorola MTX/MVME) and Macintosh G4 SMP + * by Troy Benjegerdes (hozer@drgw.net) + * + * Support for DayStar quad CPU cards + * Copyright (C) XLR8, Inc. 1994-2000 */ #include <linux/config.h> @@ -23,7 +24,6 @@ #define __KERNEL_SYSCALLS__ #include <linux/unistd.h> #include <linux/init.h> -#include <linux/openpic.h> #include <linux/spinlock.h> #include <asm/ptrace.h> @@ -37,47 +37,97 @@ #include <asm/io.h> #include <asm/prom.h> #include <asm/smp.h> -#include <asm/gemini.h> - +#include <asm/residual.h> +#include <asm/feature.h> #include <asm/time.h> + #include "open_pic.h" int smp_threads_ready; volatile int smp_commenced; int smp_num_cpus = 1; +int smp_tb_synchronized; struct cpuinfo_PPC cpu_data[NR_CPUS]; struct klock_info_struct klock_info = { KLOCK_CLEAR, 0 }; -volatile unsigned char active_kernel_processor = NO_PROC_ID; /* Processor holding kernel spinlock */ -volatile unsigned long ipi_count; +atomic_t ipi_recv; +atomic_t ipi_sent; spinlock_t kernel_flag = SPIN_LOCK_UNLOCKED; unsigned int prof_multiplier[NR_CPUS]; unsigned int prof_counter[NR_CPUS]; cycles_t cacheflush_time; +static int max_cpus __initdata = NR_CPUS; -/* this has to go in the data section because it is accessed from prom_init */ int smp_hw_index[NR_CPUS]; /* all cpu mappings are 1-1 -- Cort */ volatile unsigned long cpu_callin_map[NR_CPUS]; +#define TB_SYNC_PASSES 4 +volatile unsigned long __initdata tb_sync_flag = 0; +volatile unsigned long __initdata tb_offset = 0; + int start_secondary(void *); extern int cpu_idle(void *unused); -u_int openpic_read(volatile u_int *addr); void smp_call_function_interrupt(void); void smp_message_pass(int target, int msg, unsigned long data, int wait); +extern void __secondary_start_psurge(void); +extern void __secondary_start_psurge2(void); /* Temporary horrible hack */ +extern void __secondary_start_psurge3(void); /* Temporary horrible hack */ + +/* Addresses for powersurge registers */ +#define HAMMERHEAD_BASE 0xf8000000 +#define HHEAD_CONFIG 0x90 +#define HHEAD_SEC_INTR 0xc0 + /* register for interrupting the primary processor on the powersurge */ /* N.B. this is actually the ethernet ROM! */ -#define PSURGE_PRI_INTR 0xf3019000 -/* register for interrupting the secondary processor on the powersurge */ -#define PSURGE_SEC_INTR 0xf80000c0 +#define PSURGE_PRI_INTR 0xf3019000 + /* register for storing the start address for the secondary processor */ -#define PSURGE_START 0xf2800000 +/* N.B. this is the PCI config space address register for the 1st bridge */ +#define PSURGE_START 0xf2800000 + +/* Daystar/XLR8 4-CPU card */ +#define PSURGE_QUAD_REG_ADDR 0xf8800000 + +#define PSURGE_QUAD_IRQ_SET 0 +#define PSURGE_QUAD_IRQ_CLR 1 +#define PSURGE_QUAD_IRQ_PRIMARY 2 +#define PSURGE_QUAD_CKSTOP_CTL 3 +#define PSURGE_QUAD_PRIMARY_ARB 4 +#define PSURGE_QUAD_BOARD_ID 6 +#define PSURGE_QUAD_WHICH_CPU 7 +#define PSURGE_QUAD_CKSTOP_RDBK 8 +#define PSURGE_QUAD_RESET_CTL 11 + +#define PSURGE_QUAD_OUT(r, v) (out_8((u8 *)(quad_base+((r)<<2)+1), (v))) +#define PSURGE_QUAD_IN(r) (in_8((u8 *)(quad_base+((r)<<2)+1)) & 0x0f) +#define PSURGE_QUAD_BIS(r, v) (PSURGE_QUAD_OUT((r), PSURGE_QUAD_IN(r) | (v))) +#define PSURGE_QUAD_BIC(r, v) (PSURGE_QUAD_OUT((r), PSURGE_QUAD_IN(r) & ~(v))) + /* virtual addresses for the above */ -volatile u32 *psurge_pri_intr; -volatile u32 *psurge_sec_intr; -volatile u32 *psurge_start; +static volatile u8 *hhead_base; +static volatile u32 *quad_base; +static volatile u32 *psurge_pri_intr; +static volatile u8 *psurge_sec_intr; +static volatile u32 *psurge_start; + +/* what sort of powersurge board we have */ +static int psurge_type; + +/* values for psurge_type */ +#define PSURGE_DUAL 0 +#define PSURGE_QUAD_OKEE 1 +#define PSURGE_QUAD_COTTON 2 +#define PSURGE_QUAD_ICEGRASS 3 -/* Since OpenPIC has only 4 IPIs, we use slightly different message numbers. */ +/* l2 cache stuff for dual G4 macs */ +extern void core99_init_l2(void); + +/* Since OpenPIC has only 4 IPIs, we use slightly different message numbers. + * + * Make sure this matches openpic_request_IPIs in open_pic.c, or what shows up + * in /proc/interrupts will be wrong!!! --Troy */ #define PPC_MSG_CALL_FUNCTION 0 #define PPC_MSG_RESCHEDULE 1 #define PPC_MSG_INVALIDATE_TLB 2 @@ -85,10 +135,577 @@ volatile u32 *psurge_start; static inline void set_tb(unsigned int upper, unsigned int lower) { + mtspr(SPRN_TBWL, 0); mtspr(SPRN_TBWU, upper); mtspr(SPRN_TBWL, lower); } +/* + * Set and clear IPIs for powersurge. + */ +static inline void psurge_set_ipi(int cpu) +{ + if (cpu == 0) + in_be32(psurge_pri_intr); + else if (psurge_type == PSURGE_DUAL) + out_8(psurge_sec_intr, 0); + else + PSURGE_QUAD_OUT(PSURGE_QUAD_IRQ_SET, 1 << cpu); +} + +static inline void psurge_clr_ipi(int cpu) +{ + if (cpu > 0) { + if (psurge_type == PSURGE_DUAL) + out_8(psurge_sec_intr, ~0); + else + PSURGE_QUAD_OUT(PSURGE_QUAD_IRQ_CLR, 1 << cpu); + } +} + +/* + * On powersurge (old SMP powermac architecture) we don't have + * separate IPIs for separate messages like openpic does. Instead + * we have a bitmap for each processor, where a 1 bit means that + * the corresponding message is pending for that processor. + * Ideally each cpu's entry would be in a different cache line. + * -- paulus. + */ +static unsigned long psurge_smp_message[NR_CPUS]; + +void psurge_smp_message_recv(struct pt_regs *regs) +{ + int cpu = smp_processor_id(); + int msg; + + /* clear interrupt */ + psurge_clr_ipi(cpu); + + if (smp_num_cpus < 2) + return; + + /* make sure there is a message there */ + for (msg = 0; msg < 4; msg++) + if (test_and_clear_bit(msg, &psurge_smp_message[cpu])) + smp_message_recv(msg, regs); +} + +void +psurge_primary_intr(int irq, void *d, struct pt_regs *regs) +{ + psurge_smp_message_recv(regs); +} + +static void +smp_psurge_message_pass(int target, int msg, unsigned long data, int wait) +{ + int i; + + if (smp_num_cpus < 2) + return; + + for (i = 0; i < smp_num_cpus; i++) { + if (target == MSG_ALL + || (target == MSG_ALL_BUT_SELF && i != smp_processor_id()) + || target == i) { + set_bit(msg, &psurge_smp_message[i]); + psurge_set_ipi(i); + } + } +} + +/* + * Determine a quad card presence. We read the board ID register, we + * for the data bus to change to something else, and we read it again. + * It it's stable, then the register probably exist (ugh !) + */ +static int __init psurge_quad_probe(void) +{ + int type; + unsigned int i; + + type = PSURGE_QUAD_IN(PSURGE_QUAD_BOARD_ID); + if (type < PSURGE_QUAD_OKEE || type > PSURGE_QUAD_ICEGRASS + || type != PSURGE_QUAD_IN(PSURGE_QUAD_BOARD_ID)) + return PSURGE_DUAL; + + /* looks OK, try a slightly more rigorous test */ + /* bogus is not necessarily cacheline-aligned, + though I don't suppose that really matters. -- paulus */ + for (i = 0; i < 100; i++) { + volatile u32 bogus[8]; + bogus[(0+i)%8] = 0x00000000; + bogus[(1+i)%8] = 0x55555555; + bogus[(2+i)%8] = 0xFFFFFFFF; + bogus[(3+i)%8] = 0xAAAAAAAA; + bogus[(4+i)%8] = 0x33333333; + bogus[(5+i)%8] = 0xCCCCCCCC; + bogus[(6+i)%8] = 0xCCCCCCCC; + bogus[(7+i)%8] = 0x33333333; + wmb(); + asm volatile("dcbf 0,%0" : : "r" (bogus) : "memory"); + mb(); + if (type != PSURGE_QUAD_IN(PSURGE_QUAD_BOARD_ID)) + return PSURGE_DUAL; + } + return type; +} + +static void __init psurge_quad_init(void) +{ + int procbits; + + if (ppc_md.progress) ppc_md.progress("psurge_quad_init", 0x351); + procbits = ~PSURGE_QUAD_IN(PSURGE_QUAD_WHICH_CPU); + if (psurge_type == PSURGE_QUAD_ICEGRASS) + PSURGE_QUAD_BIS(PSURGE_QUAD_RESET_CTL, procbits); + else + PSURGE_QUAD_BIC(PSURGE_QUAD_CKSTOP_CTL, procbits); + mdelay(33); + out_8(psurge_sec_intr, ~0); + PSURGE_QUAD_OUT(PSURGE_QUAD_IRQ_CLR, procbits); + PSURGE_QUAD_BIS(PSURGE_QUAD_RESET_CTL, procbits); + if (psurge_type != PSURGE_QUAD_ICEGRASS) + PSURGE_QUAD_BIS(PSURGE_QUAD_CKSTOP_CTL, procbits); + PSURGE_QUAD_BIC(PSURGE_QUAD_PRIMARY_ARB, procbits); + mdelay(33); + PSURGE_QUAD_BIC(PSURGE_QUAD_RESET_CTL, procbits); + mdelay(33); + PSURGE_QUAD_BIS(PSURGE_QUAD_PRIMARY_ARB, procbits); + mdelay(33); +} + +static int __init smp_psurge_probe(void) +{ + int i, ncpus; + + /* We don't do SMP on the PPC601 -- paulus */ + if ((_get_PVR() >> 16) == 1) + return 1; + + /* + * The powersurge cpu board can be used in the generation + * of powermacs that have a socket for an upgradeable cpu card, + * including the 7500, 8500, 9500, 9600. + * The device tree doesn't tell you if you have 2 cpus because + * OF doesn't know anything about the 2nd processor. + * Instead we look for magic bits in magic registers, + * in the hammerhead memory controller in the case of the + * dual-cpu powersurge board. -- paulus. + */ + if (find_devices("hammerhead") == NULL) + return 1; + + hhead_base = ioremap(HAMMERHEAD_BASE, 0x800); + quad_base = ioremap(PSURGE_QUAD_REG_ADDR, 1024); + psurge_sec_intr = hhead_base + HHEAD_SEC_INTR; + + psurge_type = psurge_quad_probe(); + if (psurge_type != PSURGE_DUAL) { + psurge_quad_init(); + /* I believe we could "count" CPUs by counting 1 bits + * in procbits on a quad board. For now, we assume 4, + * non-present CPUs will just be seen as "stuck". + * (hope they are the higher-numbered ones -- paulus) + */ + ncpus = 4; + } else { + iounmap((void *) quad_base); + if ((in_8(hhead_base + HHEAD_CONFIG) & 0x02) == 0) { + /* not a dual-cpu card */ + iounmap((void *) hhead_base); + return 1; + } + ncpus = 2; + } + + psurge_start = ioremap(PSURGE_START, 4); + psurge_pri_intr = ioremap(PSURGE_PRI_INTR, 4); + + /* this is not actually strictly necessary -- paulus. */ + for (i = 1; i < ncpus; ++i) + smp_hw_index[i] = i; + + if (ppc_md.progress) ppc_md.progress("smp_psurge_probe - done", 0x352); + + return ncpus; +} + +static void __init smp_psurge_kick_cpu(int nr) +{ + void (*start)(void) = __secondary_start_psurge; + + if (ppc_md.progress) ppc_md.progress("smp_psurge_kick_cpu", 0x353); + + /* setup entry point of secondary processor */ + switch (nr) { + case 2: + start = __secondary_start_psurge2; + break; + case 3: + start = __secondary_start_psurge3; + break; + } + + out_be32(psurge_start, __pa(start)); + mb(); + + psurge_set_ipi(nr); + udelay(10); + psurge_clr_ipi(nr); + + if (ppc_md.progress) ppc_md.progress("smp_psurge_kick_cpu - done", 0x354); +} + +/* + * With the dual-cpu powersurge board, the decrementers and timebases + * of both cpus are frozen after the secondary cpu is started up, + * until we give the secondary cpu another interrupt. This routine + * uses this to get the timebases synchronized. + * -- paulus. + */ +static void __init psurge_dual_sync_tb(int cpu_nr) +{ + static volatile int sec_tb_reset = 0; + int t; + + set_dec(tb_ticks_per_jiffy); + set_tb(0, 0); + last_jiffy_stamp(cpu_nr) = 0; + + if (cpu_nr > 0) { + mb(); + sec_tb_reset = 1; + return; + } + + /* wait for the secondary to have reset its TB before proceeding */ + for (t = 10000000; t > 0 && !sec_tb_reset; --t) + ; + + /* now interrupt the secondary, starting both TBs */ + psurge_set_ipi(1); + + smp_tb_synchronized = 1; +} + +static void +smp_psurge_setup_cpu(int cpu_nr) +{ + + if (cpu_nr == 0) { + if (smp_num_cpus < 2) + return; + /* reset the entry point so if we get another intr we won't + * try to startup again */ + out_be32(psurge_start, 0x100); + if (request_irq(30, psurge_primary_intr, 0, "primary IPI", 0)) + printk(KERN_ERR "Couldn't get primary IPI interrupt"); + } + + if (psurge_type == PSURGE_DUAL) + psurge_dual_sync_tb(cpu_nr); +} + + +static void +smp_openpic_message_pass(int target, int msg, unsigned long data, int wait) +{ + /* make sure we're sending something that translates to an IPI */ + if ( msg > 0x3 ){ + printk("SMP %d: smp_message_pass: unknown msg %d\n", + smp_processor_id(), msg); + return; + } + switch ( target ) + { + case MSG_ALL: + openpic_cause_IPI(msg, 0xffffffff); + break; + case MSG_ALL_BUT_SELF: + openpic_cause_IPI(msg, + 0xffffffff & ~(1 << smp_hw_index[smp_processor_id()])); + + break; + default: + openpic_cause_IPI(msg, smp_hw_index[1<<target]); + break; + } +} + +static int +smp_core99_probe(void) +{ + struct device_node *cpus; + int *pp; + int i, ncpus = 1; + + if (ppc_md.progress) ppc_md.progress("smp_core99_probe", 0x345); +#if 0 /* Paulus method.. doesn't seem to work on earlier dual G4's??*/ + cpus = find_devices("cpus"); + if (cpus != 0) { + pp = (int *) get_property(cpus, "#cpus", NULL); + if (pp != NULL) + ncpus = *pp; + } +#else /* My original method -- Troy <hozer@drgw.net> */ + + cpus = find_type_devices("cpu"); + if (cpus){ + for ( ncpus = 1; cpus->next; cpus = cpus->next ){ + ncpus++; + } + } +#endif + printk("smp_core99_probe: OF reports %d cpus\n", ncpus); + if (ncpus > 1) { + openpic_request_IPIs(); + for (i = 1; i < ncpus; ++i) + smp_hw_index[i] = i; + } + + return ncpus; +} + +static void +smp_core99_kick_cpu(int nr) +{ + unsigned long save_int; + unsigned long flags; + volatile unsigned long *vector + = ((volatile unsigned long *)(KERNELBASE+0x500)); + + if (nr != 1) + return; + if (ppc_md.progress) ppc_md.progress("smp_core99_kick_cpu", 0x346); + + local_irq_save(flags); + local_irq_disable(); + + /* Save EE vector */ + save_int = *vector; + + /* Setup fake EE vector that does + * b __secondary_start_psurge - KERNELBASE + */ + *vector = 0x48000002 + + ((unsigned long)__secondary_start_psurge - KERNELBASE); + + /* flush data cache and inval instruction cache */ + flush_icache_range((unsigned long) vector, (unsigned long) vector + 4); + + /* Put some life in our friend */ + feature_core99_kick_cpu1(); + + /* FIXME: We wait a bit for the CPU to take the exception, I should + * instead wait for the entry code to set something for me. Well, + * ideally, all that crap will be done in prom.c and the CPU left + * in a RAM-based wait loop like CHRP. + */ + mdelay(1); + + /* Restore our exception vector */ + *vector = save_int; + flush_icache_range((unsigned long) vector, (unsigned long) vector + 4); + + local_irq_restore(flags); + if (ppc_md.progress) ppc_md.progress("smp_core99_probe done", 0x347); +} + +static void +smp_core99_setup_cpu(int cpu_nr) +{ + /* Setup openpic */ + do_openpic_setup_cpu(); + + /* Setup L2 */ + if (cpu_nr != 0) + core99_init_l2(); + else + if (ppc_md.progress) ppc_md.progress("core99_setup_cpu 0 done", 0x349); +} + +static int +smp_chrp_probe(void) +{ + extern unsigned long smp_chrp_cpu_nr; + + if (smp_chrp_cpu_nr > 1) + openpic_request_IPIs(); + + return smp_chrp_cpu_nr; +} + +static void +smp_chrp_kick_cpu(int nr) +{ + *(unsigned long *)KERNELBASE = nr; + asm volatile("dcbf 0,%0"::"r"(KERNELBASE):"memory"); +} + +static void +smp_chrp_setup_cpu(int cpu_nr) +{ + static atomic_t ready = ATOMIC_INIT(1); + static volatile int frozen = 0; + + if (cpu_nr == 0) { + /* wait for all the others */ + while (atomic_read(&ready) < smp_num_cpus) + barrier(); + atomic_set(&ready, 1); + /* freeze the timebase */ + call_rtas("freeze-time-base", 0, 1, NULL); + mb(); + frozen = 1; + /* XXX assumes this is not a 601 */ + set_tb(0, 0); + last_jiffy_stamp(0) = 0; + while (atomic_read(&ready) < smp_num_cpus) + barrier(); + /* thaw the timebase again */ + call_rtas("thaw-time-base", 0, 1, NULL); + mb(); + frozen = 0; + smp_tb_synchronized = 1; + } else { + atomic_inc(&ready); + while (!frozen) + barrier(); + set_tb(0, 0); + last_jiffy_stamp(0) = 0; + mb(); + atomic_inc(&ready); + while (frozen) + barrier(); + } + + if (OpenPIC_Addr) + do_openpic_setup_cpu(); +} + +#ifdef CONFIG_POWER4 +static void +smp_xics_message_pass(int target, int msg, unsigned long data, int wait) +{ + /* for now, only do reschedule messages + since we only have one IPI */ + if (msg != PPC_MSG_RESCHEDULE) + return; + for (i = 0; i < smp_num_cpus; ++i) { + if (target == MSG_ALL || target == i + || (target == MSG_ALL_BUT_SELF + && i != smp_processor_id())) + xics_cause_IPI(i); + } +} + +static int +smp_xics_probe(void) +{ + return smp_chrp_cpu_nr; +} + +static void +smp_xics_setup_cpu(int cpu_nr) +{ + if (cpu_nr > 0) + xics_setup_cpu(); +} +#endif /* CONFIG_POWER4 */ + +static int +smp_prep_probe(void) +{ + extern int mot_multi; + + if (mot_multi) { + openpic_request_IPIs(); + smp_hw_index[1] = 1; + return 2; + } + + return 1; +} + +static void +smp_prep_kick_cpu(int nr) +{ + extern unsigned long *MotSave_SmpIar; + extern unsigned char *MotSave_CpusState[2]; + + *MotSave_SmpIar = (unsigned long)__secondary_start_psurge - KERNELBASE; + *MotSave_CpusState[1] = CPU_GOOD; + printk("CPU1 reset, waiting\n"); +} + +static void +smp_prep_setup_cpu(int cpu_nr) +{ + if (OpenPIC_Addr) + do_openpic_setup_cpu(); +} + +static struct smp_ops_t { + void (*message_pass)(int target, int msg, unsigned long data, int wait); + int (*probe)(void); + void (*kick_cpu)(int nr); + void (*setup_cpu)(int nr); + +} *smp_ops; + +#define smp_message_pass(t,m,d,w) \ + do { if (smp_ops) \ + atomic_inc(&ipi_sent); \ + smp_ops->message_pass((t),(m),(d),(w)); \ + } while(0) + + +/* PowerSurge-style Macs */ +static struct smp_ops_t psurge_smp_ops = { + smp_psurge_message_pass, + smp_psurge_probe, + smp_psurge_kick_cpu, + smp_psurge_setup_cpu, +}; + +/* Core99 Macs (dual G4s) */ +static struct smp_ops_t core99_smp_ops = { + smp_openpic_message_pass, + smp_core99_probe, + smp_core99_kick_cpu, + smp_core99_setup_cpu, +}; + +/* CHRP with openpic */ +static struct smp_ops_t chrp_smp_ops = { + smp_openpic_message_pass, + smp_chrp_probe, + smp_chrp_kick_cpu, + smp_chrp_setup_cpu, +}; + +#ifdef CONFIG_POWER4 +/* CHRP with new XICS interrupt controller */ +static struct smp_ops_t xics_smp_ops = { + smp_xics_message_pass, + smp_xics_probe, + smp_chrp_kick_cpu, + smp_xics_setup_cpu, +}; +#endif /* CONFIG_POWER4 */ + +/* PReP (MTX) */ +static struct smp_ops_t prep_smp_ops = { + smp_openpic_message_pass, + smp_prep_probe, + smp_prep_kick_cpu, + smp_prep_setup_cpu, +}; + +/* + * Common functions + */ void smp_local_timer_interrupt(struct pt_regs * regs) { int cpu = smp_processor_id(); @@ -101,7 +718,7 @@ void smp_local_timer_interrupt(struct pt_regs * regs) void smp_message_recv(int msg, struct pt_regs *regs) { - ipi_count++; + atomic_inc(&ipi_recv); switch( msg ) { case PPC_MSG_CALL_FUNCTION: @@ -126,47 +743,6 @@ void smp_message_recv(int msg, struct pt_regs *regs) } /* - * As it is now, if we're sending two message at the same time - * we have race conditions on Pmac. The PowerSurge doesn't easily - * allow us to send IPI messages so we put the messages in - * smp_message[]. - * - * This is because don't have several IPI's on the PowerSurge even though - * we do on the chrp. It would be nice to use actual IPI's such as with - * openpic rather than this. - * -- Cort - */ -int pmac_smp_message[NR_CPUS]; -void pmac_smp_message_recv(struct pt_regs *regs) -{ - int cpu = smp_processor_id(); - int msg; - - /* clear interrupt */ - if (cpu == 1) - out_be32(psurge_sec_intr, ~0); - - if (smp_num_cpus < 2) - return; - - /* make sure there is a message there */ - msg = pmac_smp_message[cpu]; - if (msg == 0) - return; - - /* reset message */ - pmac_smp_message[cpu] = 0; - - smp_message_recv(msg - 1, regs); -} - -void -pmac_primary_intr(int irq, void *d, struct pt_regs *regs) -{ - pmac_smp_message_recv(regs); -} - -/* * 750's don't broadcast tlb invalidates so * we have to emulate that behavior. * -- Cort @@ -220,7 +796,7 @@ void smp_send_stop(void) */ static spinlock_t call_lock = SPIN_LOCK_UNLOCKED; -static volatile struct call_data_struct { +static struct call_data_struct { void (*func) (void *info); void *info; atomic_t started; @@ -317,87 +893,9 @@ void smp_call_function_interrupt(void) atomic_inc(&call_data->finished); } -void smp_message_pass(int target, int msg, unsigned long data, int wait) -{ - if ( !(_machine & (_MACH_Pmac|_MACH_chrp|_MACH_prep|_MACH_gemini)) ) - return; - - switch (_machine) { - case _MACH_Pmac: - /* - * IPI's on the Pmac are a hack but without reasonable - * IPI hardware SMP on Pmac is a hack. - * - * We assume here that the msg is not -1. If it is, - * the recipient won't know the message was destined - * for it. -- Cort - */ - if (smp_processor_id() == 0) { - /* primary cpu */ - if (target == 1 || target == MSG_ALL_BUT_SELF - || target == MSG_ALL) { - pmac_smp_message[1] = msg + 1; - /* interrupt secondary processor */ - out_be32(psurge_sec_intr, ~0); - out_be32(psurge_sec_intr, 0); - } - } else { - /* secondary cpu */ - if (target == 0 || target == MSG_ALL_BUT_SELF - || target == MSG_ALL) { - pmac_smp_message[0] = msg + 1; - /* interrupt primary processor */ - in_be32(psurge_pri_intr); - } - } - if (target == smp_processor_id() || target == MSG_ALL) { - /* sending a message to ourself */ - /* XXX maybe we shouldn't do this if ints are off */ - smp_message_recv(msg, NULL); - } - break; - case _MACH_chrp: - case _MACH_prep: - case _MACH_gemini: -#ifndef CONFIG_POWER4 - /* make sure we're sending something that translates to an IPI */ - if ( msg > 0x3 ) - break; - switch ( target ) - { - case MSG_ALL: - openpic_cause_IPI(smp_processor_id(), msg, 0xffffffff); - break; - case MSG_ALL_BUT_SELF: - openpic_cause_IPI(smp_processor_id(), msg, - 0xffffffff & ~(1 << smp_processor_id())); - break; - default: - openpic_cause_IPI(smp_processor_id(), msg, 1<<target); - break; - } -#else /* CONFIG_POWER4 */ - /* for now, only do reschedule messages - since we only have one IPI */ - if (msg != PPC_MSG_RESCHEDULE) - break; - for (i = 0; i < smp_num_cpus; ++i) { - if (target == MSG_ALL || target == i - || (target == MSG_ALL_BUT_SELF - && i != smp_processor_id())) - xics_cause_IPI(i); - } -#endif /* CONFIG_POWER4 */ - break; - } -} - void __init smp_boot_cpus(void) { extern struct task_struct *current_set[NR_CPUS]; - extern unsigned long smp_chrp_cpu_nr; - extern void __secondary_start_psurge(void); - extern void __secondary_start_chrp(void); int i, cpu_nr; struct task_struct *p; unsigned long a; @@ -411,7 +909,6 @@ void __init smp_boot_cpus(void) * cpu 0, the master -- Cort */ cpu_callin_map[0] = 1; - active_kernel_processor = 0; current->processor = 0; init_idle(); @@ -427,41 +924,40 @@ void __init smp_boot_cpus(void) */ cacheflush_time = 5 * 1024; - if ( !(_machine & (_MACH_Pmac|_MACH_chrp|_MACH_gemini)) ) - { - printk("SMP not supported on this machine.\n"); - return; - } - - switch ( _machine ) - { + /* To be later replaced by some arch-specific routine */ + switch(_machine) { case _MACH_Pmac: - /* assume powersurge board - 2 processors -- Cort */ - cpu_nr = 2; - psurge_pri_intr = ioremap(PSURGE_PRI_INTR, 4); - psurge_sec_intr = ioremap(PSURGE_SEC_INTR, 4); - psurge_start = ioremap(PSURGE_START, 4); + /* Check for Core99 */ + if (find_devices("uni-n")) + smp_ops = &core99_smp_ops; + else + smp_ops = &psurge_smp_ops; break; case _MACH_chrp: - if (OpenPIC) - for ( i = 0; i < 4 ; i++ ) - openpic_enable_IPI(i); - cpu_nr = smp_chrp_cpu_nr; +#ifndef CONFIG_POWER4 + smp_ops = &chrp_smp_ops; +#else + smp_ops = &xics_smp_ops; +#endif /* CONFIG_POWER4 */ break; - case _MACH_gemini: - for ( i = 0; i < 4 ; i++ ) - openpic_enable_IPI(i); - cpu_nr = (readb(GEMINI_CPUSTAT) & GEMINI_CPU_COUNT_MASK)>>2; - cpu_nr = (cpu_nr == 0) ? 4 : cpu_nr; + case _MACH_prep: + smp_ops = &prep_smp_ops; break; + default: + printk("SMP not supported on this machine.\n"); + return; } + + /* Probe arch for CPUs */ + cpu_nr = smp_ops->probe(); /* * only check for cpus we know exist. We keep the callin map * with cpus at the bottom -- Cort */ - for ( i = 1 ; i < cpu_nr; i++ ) - { + if (cpu_nr > max_cpus) + cpu_nr = max_cpus; + for (i = 1; i < cpu_nr; i++) { int c; struct pt_regs regs; @@ -487,25 +983,7 @@ void __init smp_boot_cpus(void) asm volatile("sync"); /* wake up cpus */ - switch ( _machine ) - { - case _MACH_Pmac: - /* setup entry point of secondary processor */ - out_be32(psurge_start, __pa(__secondary_start_psurge)); - /* interrupt secondary to begin executing code */ - out_be32(psurge_sec_intr, ~0); - udelay(1); - out_be32(psurge_sec_intr, 0); - break; - case _MACH_chrp: - *(unsigned long *)KERNELBASE = i; - asm volatile("dcbf 0,%0"::"r"(KERNELBASE):"memory"); - break; - case _MACH_gemini: - openpic_init_processor( 1<<i ); - openpic_init_processor( 0 ); - break; - } + smp_ops->kick_cpu(i); /* * wait to see if the cpu made a callin (is actually up). @@ -517,40 +995,108 @@ void __init smp_boot_cpus(void) if ( cpu_callin_map[i] ) { + char buf[32]; + sprintf(buf, "found cpu %d", i); + if (ppc_md.progress) ppc_md.progress(buf, 0x350+i); printk("Processor %d found.\n", i); smp_num_cpus++; } else { + char buf[32]; + sprintf(buf, "didn't find cpu %d", i); + if (ppc_md.progress) ppc_md.progress(buf, 0x360+i); printk("Processor %d is stuck.\n", i); } } - if (OpenPIC && (_machine & (_MACH_gemini|_MACH_chrp|_MACH_prep))) - do_openpic_setup_cpu(); + /* Setup CPU 0 last (important) */ + smp_ops->setup_cpu(0); +} - if ( _machine == _MACH_Pmac ) - { - /* reset the entry point so if we get another intr we won't - * try to startup again */ - out_be32(psurge_start, 0x100); - if (request_irq(30, pmac_primary_intr, 0, "primary IPI", 0)) - printk(KERN_ERR "Couldn't get primary IPI interrupt"); - /* - * The decrementers of both cpus are frozen at this point - * until we give the secondary cpu another interrupt. - * We set them both to decrementer_count and then send - * the interrupt. This should get the decrementers - * synchronized. - * -- paulus. - */ - set_dec(tb_ticks_per_jiffy); - if ((_get_PVR() >> 16) != 1) { - set_tb(0, 0); /* set timebase if not 601 */ - last_jiffy_stamp(0) = 0; +void __init smp_software_tb_sync(int cpu) +{ +#define PASSES 4 /* 4 passes.. */ + int pass; + int i, j; + + /* stop - start will be the number of timebase ticks it takes for cpu0 + * to send a message to all others and the first reponse to show up. + * + * ASSUMPTION: this time is similiar for all cpus + * ASSUMPTION: the time to send a one-way message is ping/2 + */ + register unsigned long start = 0; + register unsigned long stop = 0; + register unsigned long temp = 0; + + if (smp_num_cpus < 2) { + smp_tb_synchronized = 1; + return; + } + + /* This code need fixing on >2 CPUs --BenH/paulus */ + if (smp_num_cpus > 2) { + smp_tb_synchronized = 0; + return; + } + + set_tb(0, 0); + + /* multiple passes to get in l1 cache.. */ + for (pass = 2; pass < 2+PASSES; pass++){ + if (cpu == 0){ + mb(); + for (i = j = 1; i < smp_num_cpus; i++, j++){ + /* skip stuck cpus */ + while (!cpu_callin_map[j]) + ++j; + while (cpu_callin_map[j] != pass) + barrier(); + } + mb(); + tb_sync_flag = pass; + start = get_tbl(); /* start timing */ + while (tb_sync_flag) + mb(); + stop = get_tbl(); /* end timing */ + /* theoretically, the divisor should be 2, but + * I get better results on my dual mtx. someone + * please report results on other smp machines.. + */ + tb_offset = (stop-start)/4; + mb(); + tb_sync_flag = pass; + udelay(10); + mb(); + tb_sync_flag = 0; + mb(); + set_tb(0,0); + mb(); + } else { + cpu_callin_map[cpu] = pass; + mb(); + while (!tb_sync_flag) + mb(); /* wait for cpu0 */ + mb(); + tb_sync_flag = 0; /* send response for timing */ + mb(); + while (!tb_sync_flag) + mb(); + temp = tb_offset; /* make sure offset is loaded */ + while (tb_sync_flag) + mb(); + set_tb(0,temp); /* now, set the timebase */ + mb(); } - out_be32(psurge_sec_intr, ~0); - udelay(1); - out_be32(psurge_sec_intr, 0); } + if (cpu == 0) { + smp_tb_synchronized = 1; + printk("smp_software_tb_sync: %d passes, final offset: %ld\n", + PASSES, tb_offset); + } + /* so time.c doesn't get confused */ + set_dec(tb_ticks_per_jiffy); + last_jiffy_stamp(cpu) = 0; + cpu_callin_map[cpu] = 1; } void __init smp_commence(void) @@ -558,8 +1104,48 @@ void __init smp_commence(void) /* * Lets the callin's below out of their loop. */ + if (ppc_md.progress) ppc_md.progress("smp_commence", 0x370); wmb(); smp_commenced = 1; + /* if the smp_ops->setup_cpu function has not already synched the + * timebases with a nicer hardware-based method, do so now + * + * I am open to suggestions for improvements to this method + * -- Troy <hozer@drgw.net> + * + * NOTE: if you are debugging, set smp_tb_synchronized for now + * since if this code runs pretty early and needs all cpus that + * reported in in smp_callin_map to be working + * + * NOTE2: this code doesn't seem to work on > 2 cpus. -- paulus + */ + if (!smp_tb_synchronized) { + unsigned long flags; + __save_and_cli(flags); + smp_software_tb_sync(0); + __restore_flags(flags); + } +} + +void __init smp_callin(void) +{ + int cpu = current->processor; + + smp_store_cpu_info(cpu); + set_dec(tb_ticks_per_jiffy); + cpu_callin_map[cpu] = 1; + + smp_ops->setup_cpu(cpu); + + init_idle(); + + while(!smp_commenced) + barrier(); + /* see smp_commence for more info */ + if (!smp_tb_synchronized){ + smp_software_tb_sync(cpu); + } + __sti(); } /* intel needs this */ @@ -576,37 +1162,6 @@ int __init start_secondary(void *unused) return cpu_idle(NULL); } -void __init smp_callin(void) -{ - smp_store_cpu_info(current->processor); - set_dec(tb_ticks_per_jiffy); - if (_machine == _MACH_Pmac && (_get_PVR() >> 16) != 1) { - set_tb(0, 0); /* set timebase if not 601 */ - last_jiffy_stamp(current->processor) = 0; - } - init_idle(); - cpu_callin_map[current->processor] = 1; - -#ifndef CONFIG_POWER4 - /* - * Each processor has to do this and this is the best - * place to stick it for now. - * -- Cort - */ - if (OpenPIC && _machine & (_MACH_gemini|_MACH_chrp|_MACH_prep)) - do_openpic_setup_cpu(); -#else - xics_setup_cpu(); -#endif /* CONFIG_POWER4 */ -#ifdef CONFIG_GEMINI - if ( _machine == _MACH_gemini ) - gemini_init_l2(); -#endif - while(!smp_commenced) - barrier(); - __sti(); -} - void __init smp_setup(char *str, int *ints) { } @@ -621,6 +1176,14 @@ void __init smp_store_cpu_info(int id) struct cpuinfo_PPC *c = &cpu_data[id]; /* assume bogomips are same for everything */ - c->loops_per_sec = loops_per_sec; + c->loops_per_jiffy = loops_per_jiffy; c->pvr = _get_PVR(); } + +static int __init maxcpus(char *str) +{ + get_option(&str, &max_cpus); + return 1; +} + +__setup("maxcpus=", maxcpus); diff --git a/arch/ppc/kernel/time.c b/arch/ppc/kernel/time.c index f71c8cbbf..5b36cedcc 100644 --- a/arch/ppc/kernel/time.c +++ b/arch/ppc/kernel/time.c @@ -67,7 +67,12 @@ #include <asm/time.h> -void smp_local_timer_interrupt(struct pt_regs *); +#ifdef CONFIG_SMP +extern void smp_local_timer_interrupt(struct pt_regs *); +extern int smp_tb_synchronized; +#endif /* CONFIG_SMP */ + +extern int do_sys_settimeofday(struct timeval *tv, struct timezone *tz); /* keep track of when we need to update the rtc */ time_t last_rtc_update; @@ -97,6 +102,36 @@ static inline int tb_delta(unsigned *jiffy_stamp) { return delta; } +extern unsigned long prof_cpu_mask; +extern unsigned int * prof_buffer; +extern unsigned long prof_len; +extern unsigned long prof_shift; +extern char _stext; + +static inline void ppc_do_profile (unsigned long nip) +{ + if (!prof_buffer) + return; + + /* + * Only measure the CPUs specified by /proc/irq/prof_cpu_mask. + * (default is all CPUs.) + */ + if (!((1<<smp_processor_id()) & prof_cpu_mask)) + return; + + nip -= (unsigned long) &_stext; + nip >>= prof_shift; + /* + * Don't ignore out-of-bounds EIP values silently, + * put them into the last histogram slot, so if + * present, they will show up as a sharp peak. + */ + if (nip > prof_len-1) + nip = prof_len-1; + atomic_inc((atomic_t *)&prof_buffer[nip]); +} + /* * timer_interrupt - gets called when the decrementer overflows, * with interrupts disabled. @@ -110,6 +145,9 @@ int timer_interrupt(struct pt_regs * regs) hardirq_enter(cpu); + if (!user_mode(regs)) + ppc_do_profile(instruction_pointer(regs)); + do { jiffy_stamp += tb_ticks_per_jiffy; if (smp_processor_id()) continue; @@ -151,7 +189,7 @@ int timer_interrupt(struct pt_regs * regs) #ifdef CONFIG_SMP smp_local_timer_interrupt(regs); -#endif +#endif /* CONFIG_SMP */ if (ppc_md.heartbeat && !ppc_md.heartbeat_count--) ppc_md.heartbeat(); @@ -176,7 +214,7 @@ void do_gettimeofday(struct timeval *tv) /* As long as timebases are not in sync, gettimeofday can only * have jiffy resolution on SMP. */ - if (_machine != _MACH_Pmac) + if (!smp_tb_synchronized) delta = 0; #endif /* CONFIG_SMP */ lost_ticks = jiffies - wall_jiffies; diff --git a/arch/ppc/kernel/traps.c b/arch/ppc/kernel/traps.c index 3b7473dda..8bd9ebded 100644 --- a/arch/ppc/kernel/traps.c +++ b/arch/ppc/kernel/traps.c @@ -65,21 +65,37 @@ int (*debugger_dabr_match)(struct pt_regs *regs); void (*debugger_fault_handler)(struct pt_regs *regs); #endif #endif + /* * Trap & Exception support */ + +spinlock_t oops_lock = SPIN_LOCK_UNLOCKED; + +void die(const char * str, struct pt_regs * fp, long err) +{ + console_verbose(); + spin_lock_irq(&oops_lock); + printk("Oops: %s, sig: %ld\n", str, err); + show_regs(fp); + print_backtrace((unsigned long *)fp->gpr[1]); + spin_unlock_irq(&oops_lock); + /* do_exit() should take care of panic'ing from an interrupt + * context so we don't handle it here + */ + do_exit(err); +} + void _exception(int signr, struct pt_regs *regs) { if (!user_mode(regs)) { - show_regs(regs); #if defined(CONFIG_XMON) || defined(CONFIG_KGDB) debugger(regs); #endif - print_backtrace((unsigned long *)regs->gpr[1]); - panic("Exception in kernel pc %lx signal %d",regs->nip,signr); + die("Exception in kernel mode", regs, signr); } force_sig(signr, current); } @@ -98,7 +114,7 @@ MachineCheckException(struct pt_regs *regs) #if defined(CONFIG_8xx) && defined(CONFIG_PCI) /* the qspan pci read routines can cause machine checks -- Cort */ - bad_page_fault(regs, regs->dar); + bad_page_fault(regs, regs->dar, SIGBUS); return; #endif #if defined(CONFIG_XMON) || defined(CONFIG_KGDB) @@ -151,12 +167,10 @@ MachineCheckException(struct pt_regs *regs) default: printk("Unknown values in msr\n"); } - show_regs(regs); #if defined(CONFIG_XMON) || defined(CONFIG_KGDB) debugger(regs); #endif - print_backtrace((unsigned long *)regs->gpr[1]); - panic("machine check"); + die("machine check", regs, SIGBUS); } void @@ -217,13 +231,13 @@ emulate_instruction(struct pt_regs *regs) uint rd; uint retval; - retval = EFAULT; + retval = EINVAL; if (!user_mode(regs)) return retval; if (get_user(instword, (uint *)(regs->nip))) - return retval; + return EFAULT; /* Emulate the mfspr rD, PVR. */ @@ -337,12 +351,10 @@ SoftwareEmulation(struct pt_regs *regs) int errcode; if (!user_mode(regs)) { - show_regs(regs); #if defined(CONFIG_XMON) || defined(CONFIG_KGDB) debugger(regs); #endif - print_backtrace((unsigned long *)regs->gpr[1]); - panic("Kernel Mode Software FPU Emulation"); + die("Kernel Mode Software FPU Emulation", regs, SIGFPE); } #ifdef CONFIG_MATH_EMULATION diff --git a/arch/ppc/lib/Makefile b/arch/ppc/lib/Makefile index 1e973940c..b6f99174b 100644 --- a/arch/ppc/lib/Makefile +++ b/arch/ppc/lib/Makefile @@ -9,6 +9,6 @@ O_TARGET := lib.o obj-y := checksum.o string.o strcase.o -obj-$(CONFIG_SMP) := locks.o +obj-$(CONFIG_SMP) += locks.o include $(TOPDIR)/Rules.make diff --git a/arch/ppc/lib/locks.c b/arch/ppc/lib/locks.c index 60d8576da..a5b0a40a8 100644 --- a/arch/ppc/lib/locks.c +++ b/arch/ppc/lib/locks.c @@ -58,8 +58,9 @@ void _spin_unlock(spinlock_t *lp) { #ifdef DEBUG_LOCKS if ( !lp->lock ) - printk("_spin_unlock(%p): no lock cpu %d %s/%d\n", lp, - smp_processor_id(),current->comm,current->pid); + printk("_spin_unlock(%p): no lock cpu %d curr PC %p %s/%d\n", + lp, smp_processor_id(), __builtin_return_address(0), + current->comm, current->pid); if ( lp->owner_cpu != smp_processor_id() ) printk("_spin_unlock(%p): cpu %d trying clear of cpu %d pc %lx val %lx\n", lp, smp_processor_id(), (int)lp->owner_cpu, diff --git a/arch/ppc/mm/fault.c b/arch/ppc/mm/fault.c index b6da2cdfc..97bb6dbc1 100644 --- a/arch/ppc/mm/fault.c +++ b/arch/ppc/mm/fault.c @@ -48,7 +48,7 @@ unsigned long pte_errors = 0; /* updated by do_page_fault() */ unsigned int probingmem = 0; extern void die_if_kernel(char *, struct pt_regs *, long); -void bad_page_fault(struct pt_regs *, unsigned long); +void bad_page_fault(struct pt_regs *, unsigned long, int sig); void do_page_fault(struct pt_regs *, unsigned long, unsigned long); /* @@ -96,7 +96,7 @@ void do_page_fault(struct pt_regs *regs, unsigned long address, #endif /* CONFIG_XMON || CONFIG_KGDB */ if (in_interrupt() || mm == NULL) { - bad_page_fault(regs, address); + bad_page_fault(regs, address, SIGSEGV); return; } down(&mm->mmap_sem); @@ -182,7 +182,7 @@ bad_area: return; } - bad_page_fault(regs, address); + bad_page_fault(regs, address, SIGSEGV); return; /* @@ -194,7 +194,7 @@ out_of_memory: printk("VM: killing process %s\n", current->comm); if (user_mode(regs)) do_exit(SIGKILL); - bad_page_fault(regs, address); + bad_page_fault(regs, address, SIGKILL); return; do_sigbus: @@ -205,7 +205,7 @@ do_sigbus: info.si_addr = (void *)address; force_sig_info (SIGBUS, &info, current); if (!user_mode(regs)) - bad_page_fault(regs, address); + bad_page_fault(regs, address, SIGBUS); } /* @@ -214,8 +214,10 @@ do_sigbus: * in traps.c. */ void -bad_page_fault(struct pt_regs *regs, unsigned long address) +bad_page_fault(struct pt_regs *regs, unsigned long address, int sig) { + extern void die(const char *,struct pt_regs *,long); + unsigned long fixup; /* Are we prepared to handle this fault? */ @@ -225,14 +227,11 @@ bad_page_fault(struct pt_regs *regs, unsigned long address) } /* kernel has accessed a bad area */ - show_regs(regs); #if defined(CONFIG_XMON) || defined(CONFIG_KGDB) if (debugger_kernel_faults) debugger(regs); #endif - print_backtrace( (unsigned long *)regs->gpr[1] ); - panic("kernel access of bad area pc %lx lr %lx address %lX tsk %s/%d", - regs->nip,regs->link,address,current->comm,current->pid); + die("kernel access of bad area", regs, sig); } #ifdef CONFIG_8xx diff --git a/arch/ppc/mm/init.c b/arch/ppc/mm/init.c index c3d77a67e..839b618d0 100644 --- a/arch/ppc/mm/init.c +++ b/arch/ppc/mm/init.c @@ -34,7 +34,6 @@ #include <linux/vmalloc.h> #include <linux/init.h> #include <linux/delay.h> -#include <linux/openpic.h> #include <linux/bootmem.h> #include <linux/highmem.h> #ifdef CONFIG_BLK_DEV_INITRD @@ -62,7 +61,6 @@ #include <asm/machdep.h> #include <asm/setup.h> #include <asm/amigahw.h> -#include <asm/gemini.h> #include "mem_pieces.h" @@ -70,15 +68,17 @@ #include "4xx_tlb.h" #endif -#define MAX_LOW_MEM (640 << 20) +#define MAX_LOW_MEM (512 << 20) #define PGTOKB(pages) (((pages) * PAGE_SIZE) >> 10) int prom_trashed; atomic_t next_mmu_context; +rwlock_t context_overflow_lock __cacheline_aligned = RW_LOCK_UNLOCKED; unsigned long *end_of_DRAM; unsigned long total_memory; unsigned long total_lowmem; +unsigned long ram_phys_base; int mem_init_done; int init_bootmem_done; int boot_mapsize; @@ -114,7 +114,6 @@ static void *MMU_get_page(void); unsigned long prep_find_end_of_memory(void); unsigned long pmac_find_end_of_memory(void); unsigned long apus_find_end_of_memory(void); -unsigned long gemini_find_end_of_memory(void); extern unsigned long find_end_of_memory(void); #ifdef CONFIG_8xx unsigned long m8xx_find_end_of_memory(void); @@ -127,14 +126,13 @@ unsigned long m8260_find_end_of_memory(void); #endif /* CONFIG_8260 */ static void mapin_ram(void); void map_page(unsigned long va, unsigned long pa, int flags); -void set_phys_avail(struct mem_pieces *mp); +void set_phys_avail(unsigned long total_ram); extern void die_if_kernel(char *,struct pt_regs *,long); extern char _start[], _end[]; extern char _stext[], etext[]; extern struct task_struct *current_set[NR_CPUS]; -struct mem_pieces phys_mem; char *klimit = _end; struct mem_pieces phys_avail; @@ -199,6 +197,8 @@ int __map_without_bats; /* max amount of RAM to use */ unsigned long __max_memory; +/* max amount of low RAM to map in */ +unsigned long __max_low_memory = MAX_LOW_MEM; void __bad_pte(pmd_t *pmd) { @@ -399,8 +399,8 @@ __ioremap(unsigned long addr, unsigned long size, unsigned long flags) * If the address lies within the first 16 MB, assume it's in ISA * memory space */ - if (p < 16*1024*1024) - p += _ISA_MEM_BASE; + if ( p < 16*1024*1024 ) + p += _ISA_MEM_BASE; /* * Don't allow anybody to remap normal RAM that we're using. @@ -437,7 +437,11 @@ __ioremap(unsigned long addr, unsigned long size, unsigned long flags) return NULL; v = VMALLOC_VMADDR(area->addr); } else { +#ifndef CONFIG_HIGHMEM if (p >= ioremap_base) +#else + if (p >= ioremap_base && p < PKMAP_BASE) +#endif /* CONFIG_HIGHMEM */ v = p; else v = (ioremap_bot -= size); @@ -491,16 +495,13 @@ unsigned long iopa(unsigned long addr) void map_page(unsigned long va, unsigned long pa, int flags) { - pmd_t *pd, oldpd; + pmd_t *pd; pte_t *pg; /* Use upper 10 bits of VA to index the first level map */ pd = pmd_offset(pgd_offset_k(va), va); - oldpd = *pd; /* Use middle 10 bits of VA to index the second-level map */ pg = pte_alloc(pd, va); - if (pmd_none(oldpd) && mem_init_done) - set_pgdir(va, *(pgd_t *)pd); set_pte(pg, mk_pte_phys(pa & PAGE_MASK, __pgprot(flags))); if (mem_init_done) flush_hash_page(0, va); @@ -532,6 +533,8 @@ local_flush_tlb_all(void) * 0xd0000000 on 64-bit machines. */ flush_hash_segments(0xd, 0xffffff); #else + /* this could cause problems on SMP with nobats -- paulus */ + /* XXX no hash_table_lock? interesting -- paulus */ __clear_user(Hash, Hash_size); _tlbia(); #ifdef CONFIG_SMP @@ -548,6 +551,27 @@ local_flush_tlb_all(void) void local_flush_tlb_mm(struct mm_struct *mm) { + if (mm->context == 0) { + /* don't try to reassign a new context to the kernel */ + /* + * This could cause problems on SMP if we aren't using + * the BATs (e.g. on POWER4 or if the nobats option is used). + * The problem scenario is that one cpu is doing + * flush_hash_page or similar when another cpu clears + * out the HPTEs which map the flush_hash_page text + * and the hash table. hash_page will then deadlock. + * We need some way to have "protected" HPTEs or else + * do all hash-table manipulation with the MMU off. + * -- paulus. + */ +#ifdef CONFIG_PPC64BRIDGE + flush_hash_segments(0xd, 0xf); +#else + flush_hash_segments(0xc, 0xf); +#endif CONFIG_PPC64BRIDGE + _tlbia(); + return; + } mm->context = NO_CONTEXT; if (mm == current->mm) activate_mm(mm, mm); @@ -581,16 +605,18 @@ local_flush_tlb_range(struct mm_struct *mm, unsigned long start, unsigned long e { start &= PAGE_MASK; - if (end - start > 20 * PAGE_SIZE) - { - flush_tlb_mm(mm); - return; + if (mm->context != 0) { + if (end > TASK_SIZE) + end = TASK_SIZE; + if (end - start > 20 * PAGE_SIZE) { + flush_tlb_mm(mm); + return; + } } - for (; start < end && start < TASK_SIZE; start += PAGE_SIZE) - { + for (; start < end; start += PAGE_SIZE) flush_hash_page(mm->context, start); - } + #ifdef CONFIG_SMP smp_send_tlb_invalidate(0); #endif @@ -608,23 +634,36 @@ mmu_context_overflow(void) struct task_struct *tsk; printk(KERN_DEBUG "mmu_context_overflow\n"); - read_lock(&tasklist_lock); - for_each_task(tsk) { - if (tsk->mm) - tsk->mm->context = NO_CONTEXT; - } - read_unlock(&tasklist_lock); - flush_hash_segments(0x10, 0xffffff); + /* acquire the write lock for context overflow */ + write_lock (&context_overflow_lock); + /* recheck if overflow still exists */ + if (atomic_read(&next_mmu_context) == LAST_CONTEXT) { + read_lock(&tasklist_lock); + for_each_task(tsk) { + if (tsk->mm) + tsk->mm->context = NO_CONTEXT; + } + read_unlock(&tasklist_lock); + flush_hash_segments(0x10, 0xffffff); #ifdef CONFIG_SMP - smp_send_tlb_invalidate(0); + smp_send_tlb_invalidate(0); #endif - atomic_set(&next_mmu_context, 0); + atomic_set(&next_mmu_context, 0); + } + write_unlock (&context_overflow_lock); /* make sure current always has a context */ - current->mm->context = MUNGE_CONTEXT(atomic_inc_return(&next_mmu_context)); - /* The PGD is only a placeholder. It is only used on - * 8xx processors. - */ - set_context(current->mm->context, current->mm->pgd); + /* need to check to assure current task has an mm */ + /* - idle thread does not have an MM */ + if (current->mm) { + current->mm->context = MUNGE_CONTEXT(atomic_inc_return(&next_mmu_context)); + set_context(current->mm->context, current->mm->pgd); + } +} +#else /* CONFIG_8xx */ +void +mmu_context_overflow(void) +{ + atomic_set(&next_mmu_context, -1); } #endif /* CONFIG_8xx */ @@ -727,22 +766,20 @@ void __init setbat(int index, unsigned long virt, unsigned long phys, static void __init mapin_ram(void) { - int i; unsigned long v, p, s, f; #if !defined(CONFIG_4xx) && !defined(CONFIG_8xx) && !defined(CONFIG_POWER4) if (!__map_without_bats) { - unsigned long tot, mem_base, bl, done; + unsigned long tot, bl, done; unsigned long max_size = (256<<20); unsigned long align; /* Set up BAT2 and if necessary BAT3 to cover RAM. */ - mem_base = __pa(KERNELBASE); /* Make sure we don't map a block larger than the smallest alignment of the physical address. */ - /* alignment of mem_base */ - align = ~(mem_base-1) & mem_base; + /* alignment of ram_phys_base */ + align = ~(ram_phys_base-1) & ram_phys_base; /* set BAT block size to MIN(max_size, align) */ if (align && align < max_size) max_size = align; @@ -753,7 +790,7 @@ static void __init mapin_ram(void) break; } - setbat(2, KERNELBASE, mem_base, bl, RAM_PAGE); + setbat(2, KERNELBASE, ram_phys_base, bl, RAM_PAGE); done = (unsigned long)bat_addrs[2].limit - KERNELBASE + 1; if ((done < tot) && !bat_addrs[3].limit) { /* use BAT3 to cover a bit more */ @@ -761,41 +798,35 @@ static void __init mapin_ram(void) for (bl = 128<<10; bl < max_size; bl <<= 1) if (bl * 2 > tot) break; - setbat(3, KERNELBASE+done, mem_base+done, bl, + setbat(3, KERNELBASE+done, ram_phys_base+done, bl, RAM_PAGE); } } #endif /* !CONFIG_4xx && !CONFIG_8xx && !CONFIG_POWER4 */ - for (i = 0; i < phys_mem.n_regions; ++i) { - v = (ulong)__va(phys_mem.regions[i].address); - p = phys_mem.regions[i].address; - if (p >= total_lowmem) - break; - for (s = 0; s < phys_mem.regions[i].size; s += PAGE_SIZE) { - /* On the MPC8xx, we want the page shared so we - * don't get ASID compares on kernel space. - */ - f = _PAGE_PRESENT | _PAGE_ACCESSED | _PAGE_SHARED; + v = KERNELBASE; + p = ram_phys_base; + for (s = 0; s < total_lowmem; s += PAGE_SIZE) { + /* On the MPC8xx, we want the page shared so we + * don't get ASID compares on kernel space. + */ + f = _PAGE_PRESENT | _PAGE_ACCESSED | _PAGE_SHARED; #if defined(CONFIG_KGDB) || defined(CONFIG_XMON) - /* Allows stub to set breakpoints everywhere */ - f |= _PAGE_RW | _PAGE_DIRTY | _PAGE_HWWRITE; + /* Allows stub to set breakpoints everywhere */ + f |= _PAGE_RW | _PAGE_DIRTY | _PAGE_HWWRITE; #else - if ((char *) v < _stext || (char *) v >= etext) - f |= _PAGE_RW | _PAGE_DIRTY | _PAGE_HWWRITE; + if ((char *) v < _stext || (char *) v >= etext) + f |= _PAGE_RW | _PAGE_DIRTY | _PAGE_HWWRITE; #ifndef CONFIG_8xx - else - /* On the powerpc (not 8xx), no user access - forces R/W kernel access */ - f |= _PAGE_USER; + else + /* On the powerpc (not 8xx), no user access + forces R/W kernel access */ + f |= _PAGE_USER; #endif /* CONFIG_8xx */ #endif /* CONFIG_KGDB */ - map_page(v, p, f); - v += PAGE_SIZE; - p += PAGE_SIZE; - if (p >= total_lowmem) - break; - } + map_page(v, p, f); + v += PAGE_SIZE; + p += PAGE_SIZE; } } @@ -922,27 +953,64 @@ MMU_init(void) mtspr(SPRN_DCCR, 0x80000000); /* 128 MB of data space at 0x0. */ mtspr(SPRN_ICCR, 0x80000000); /* 128 MB of instr. space at 0x0. */ } -#else - /* How about ppc_md.md_find_end_of_memory instead of these - * ifdefs? -- Dan. - */ -#ifdef CONFIG_BOOTX_TEXT -extern boot_infos_t *disp_bi; + +#elif defined(CONFIG_8xx) +void __init MMU_init(void) +{ + if ( ppc_md.progress ) ppc_md.progress("MMU:enter", 0x111); + + total_memory = total_lowmem = m8xx_find_end_of_memory(); +#ifdef CONFIG_HIGHMEM + if (total_lowmem > MAX_LOW_MEM) { + total_lowmem = MAX_LOW_MEM; + mem_pieces_remove(&phys_avail, total_lowmem, + total_memory - total_lowmem, 0); + } +#endif /* CONFIG_HIGHMEM */ + end_of_DRAM = __va(total_lowmem); + set_phys_avail(total_lowmem); + + /* Map in all of RAM starting at KERNELBASE */ + mapin_ram(); + + /* Now map in some of the I/O space that is generically needed + * or shared with multiple devices. + * All of this fits into the same 4Mbyte region, so it only + * requires one page table page. + */ + ioremap(IMAP_ADDR, IMAP_SIZE); +#ifdef CONFIG_MBX + ioremap(NVRAM_ADDR, NVRAM_SIZE); + ioremap(MBX_CSR_ADDR, MBX_CSR_SIZE); + ioremap(PCI_CSR_ADDR, PCI_CSR_SIZE); + + /* Map some of the PCI/ISA I/O space to get the IDE interface. + */ + ioremap(PCI_ISA_IO_ADDR, 0x4000); + ioremap(PCI_IDE_ADDR, 0x4000); +#endif +#ifdef CONFIG_RPXLITE + ioremap(RPX_CSR_ADDR, RPX_CSR_SIZE); + ioremap(HIOX_CSR_ADDR, HIOX_CSR_SIZE); +#endif +#ifdef CONFIG_RPXCLASSIC + ioremap(PCI_CSR_ADDR, PCI_CSR_SIZE); + ioremap(RPX_CSR_ADDR, RPX_CSR_SIZE); #endif + if ( ppc_md.progress ) ppc_md.progress("MMU:exit", 0x211); +} + +#else /* not 4xx or 8xx */ void __init MMU_init(void) { if ( ppc_md.progress ) ppc_md.progress("MMU:enter", 0x111); -#ifndef CONFIG_8xx + if (have_of) total_memory = pmac_find_end_of_memory(); #ifdef CONFIG_APUS else if (_machine == _MACH_apus ) total_memory = apus_find_end_of_memory(); #endif -#ifdef CONFIG_GEMINI - else if ( _machine == _MACH_gemini ) - total_memory = gemini_find_end_of_memory(); -#endif /* CONFIG_GEMINI */ #if defined(CONFIG_8260) else total_memory = m8260_find_end_of_memory(); @@ -950,16 +1018,17 @@ void __init MMU_init(void) else /* prep */ total_memory = prep_find_end_of_memory(); #endif - + if (__max_memory && total_memory > __max_memory) + total_memory = __max_memory; total_lowmem = total_memory; -#ifdef CONFIG_HIGHMEM - if (total_lowmem > MAX_LOW_MEM) { - total_lowmem = MAX_LOW_MEM; - mem_pieces_remove(&phys_avail, total_lowmem, - total_memory - total_lowmem, 0); - } + if (total_lowmem > __max_low_memory) { + total_lowmem = __max_low_memory; +#ifndef CONFIG_HIGHMEM + total_memory = total_lowmem; #endif /* CONFIG_HIGHMEM */ + } end_of_DRAM = __va(total_lowmem); + set_phys_avail(total_lowmem); if ( ppc_md.progress ) ppc_md.progress("MMU:hash init", 0x300); hash_init(); @@ -991,12 +1060,15 @@ void __init MMU_init(void) ioremap_base = 0xf0000000; break; case _MACH_chrp: - setbat(0, 0xf8000000, 0xf8000000, 0x08000000, IO_PAGE); -#ifdef CONFIG_PPC64BRIDGE - setbat(1, 0x80000000, 0xc0000000, 0x10000000, IO_PAGE); -#else - setbat(1, 0x80000000, 0x80000000, 0x10000000, IO_PAGE); - setbat(3, 0x90000000, 0x90000000, 0x10000000, IO_PAGE); + /* + * The code below tends to get removed, please don't take it out. + * The F50 needs this mapping and it you take it out I'll track you + * down and slap your hands. If it causes problems please email me. + * -- Cort <cort@fsmlabs.com> + */ +#ifndef CONFIG_POWER3 + setbat(0, 0x80000000, 0x80000000, 0x10000000, IO_PAGE); + setbat(1, 0x90000000, 0x90000000, 0x10000000, IO_PAGE); #endif break; case _MACH_Pmac: @@ -1008,10 +1080,6 @@ void __init MMU_init(void) /* Map chip and ZorroII memory */ setbat(1, zTwoBase, 0x00000000, 0x01000000, IO_PAGE); break; - case _MACH_gemini: - setbat(0, 0xf0000000, 0xf0000000, 0x10000000, IO_PAGE); - setbat(1, 0x80000000, 0x80000000, 0x10000000, IO_PAGE); - break; case _MACH_8260: /* Map the IMMR, plus anything else we can cover * in that upper space according to the memory controller @@ -1025,46 +1093,7 @@ void __init MMU_init(void) } ioremap_bot = ioremap_base; #endif /* CONFIG_POWER4 */ -#else /* CONFIG_8xx */ - - total_memory = total_lowmem = m8xx_find_end_of_memory(); -#ifdef CONFIG_HIGHMEM - if (total_lowmem > MAX_LOW_MEM) { - total_lowmem = MAX_LOW_MEM; - mem_pieces_remove(&phys_avail, total_lowmem, - total_memory - total_lowmem, 0); - } -#endif /* CONFIG_HIGHMEM */ - end_of_DRAM = __va(total_lowmem); - - /* Map in all of RAM starting at KERNELBASE */ - mapin_ram(); - - /* Now map in some of the I/O space that is generically needed - * or shared with multiple devices. - * All of this fits into the same 4Mbyte region, so it only - * requires one page table page. - */ - ioremap(IMAP_ADDR, IMAP_SIZE); -#ifdef CONFIG_MBX - ioremap(NVRAM_ADDR, NVRAM_SIZE); - ioremap(MBX_CSR_ADDR, MBX_CSR_SIZE); - ioremap(PCI_CSR_ADDR, PCI_CSR_SIZE); - /* Map some of the PCI/ISA I/O space to get the IDE interface. - */ - ioremap(PCI_ISA_IO_ADDR, 0x4000); - ioremap(PCI_IDE_ADDR, 0x4000); -#endif -#ifdef CONFIG_RPXLITE - ioremap(RPX_CSR_ADDR, RPX_CSR_SIZE); - ioremap(HIOX_CSR_ADDR, HIOX_CSR_SIZE); -#endif -#ifdef CONFIG_RPXCLASSIC - ioremap(PCI_CSR_ADDR, PCI_CSR_SIZE); - ioremap(RPX_CSR_ADDR, RPX_CSR_SIZE); -#endif -#endif /* CONFIG_8xx */ if ( ppc_md.progress ) ppc_md.progress("MMU:exit", 0x211); #ifdef CONFIG_BOOTX_TEXT /* Must be done last, or ppc_md.progress will die */ @@ -1189,7 +1218,8 @@ void __init mem_init(void) #if defined(CONFIG_ALL_PPC) /* mark the RTAS pages as reserved */ if ( rtas_data ) - for (addr = rtas_data; addr < PAGE_ALIGN(rtas_data+rtas_size) ; + for (addr = (ulong)__va(rtas_data); + addr < PAGE_ALIGN((ulong)__va(rtas_data)+rtas_size) ; addr += PAGE_SIZE) SetPageReserved(virt_to_page(addr)); #endif /* defined(CONFIG_ALL_PPC) */ @@ -1249,7 +1279,7 @@ void __init mem_init(void) unsigned long __init pmac_find_end_of_memory(void) { unsigned long a, total; - unsigned long ram_limit = 0xe0000000 - KERNELBASE; + struct mem_pieces phys_mem; memory_node = find_devices("memory"); if (memory_node == NULL) { @@ -1260,8 +1290,7 @@ unsigned long __init pmac_find_end_of_memory(void) /* * Find out where physical memory is, and check that it * starts at 0 and is contiguous. It seems that RAM is - * always physically contiguous on Power Macintoshes, - * because MacOS can't cope if it isn't. + * always physically contiguous on Power Macintoshes. * * Supporting discontiguous physical memory isn't hard, * it just makes the virtual <-> physical mapping functions @@ -1274,23 +1303,14 @@ unsigned long __init pmac_find_end_of_memory(void) a = phys_mem.regions[0].address; if (a != 0) panic("RAM doesn't start at physical address 0"); - if (__max_memory == 0 || __max_memory > ram_limit) - __max_memory = ram_limit; - if (phys_mem.regions[0].size >= __max_memory) { - phys_mem.regions[0].size = __max_memory; - phys_mem.n_regions = 1; - } total = phys_mem.regions[0].size; - + if (phys_mem.n_regions > 1) { printk("RAM starting at 0x%x is not contiguous\n", phys_mem.regions[1].address); printk("Using RAM from 0 to 0x%lx\n", total-1); - phys_mem.n_regions = 1; } - set_phys_avail(&phys_mem); - return total; } #endif /* CONFIG_ALL_PPC */ @@ -1305,7 +1325,11 @@ unsigned long __init pmac_find_end_of_memory(void) unsigned long __init prep_find_end_of_memory(void) { unsigned long total; +#ifdef CONFIG_PREP_RESIDUAL total = res->TotalMemory; +#else + total = 0; +#endif if (total == 0 ) { @@ -1317,33 +1341,11 @@ unsigned long __init prep_find_end_of_memory(void) total = 0x02000000; printk("Ramsize default to be %ldM\n", total>>20); } - mem_pieces_append(&phys_mem, 0, total); - set_phys_avail(&phys_mem); return (total); } #endif /* defined(CONFIG_ALL_PPC) */ - -#if defined(CONFIG_GEMINI) -unsigned long __init gemini_find_end_of_memory(void) -{ - unsigned long total; - unsigned char reg; - - reg = readb(GEMINI_MEMCFG); - total = ((1<<((reg & 0x7) - 1)) * - (8<<((reg >> 3) & 0x7))); - total *= (1024*1024); - phys_mem.regions[0].address = 0; - phys_mem.regions[0].size = total; - phys_mem.n_regions = 1; - - set_phys_avail(&phys_mem); - return phys_mem.regions[0].size; -} -#endif /* defined(CONFIG_GEMINI) */ - #ifdef CONFIG_8260 /* * Same hack as 8xx. @@ -1355,12 +1357,7 @@ unsigned long __init m8260_find_end_of_memory(void) binfo = (bd_t *)__res; - phys_mem.regions[0].address = 0; - phys_mem.regions[0].size = binfo->bi_memsize; - phys_mem.n_regions = 1; - - set_phys_avail(&phys_mem); - return phys_mem.regions[0].size; + return binfo->bi_memsize; } #endif /* CONFIG_8260 */ @@ -1369,6 +1366,7 @@ unsigned long __init m8260_find_end_of_memory(void) unsigned long __init apus_find_end_of_memory(void) { int shadow = 0; + unsigned long total; /* The memory size reported by ADOS excludes the 512KB reserved for PPC exception registers and possibly 512KB @@ -1394,43 +1392,28 @@ unsigned long __init apus_find_end_of_memory(void) memory[0].size = ((size+0x001fffff) & 0xffe00000); } - /* Now register the memory block. */ - mem_pieces_append(&phys_mem, memory[0].addr, memory[0].size); - set_phys_avail(&phys_mem); + total = memory[0].size; /* Remove the memory chunks that are controlled by special Phase5 hardware. */ - { - unsigned long top = memory[0].addr + memory[0].size; - /* Remove the upper 512KB if it contains a shadow of - the ADOS ROM. FIXME: It might be possible to - disable this shadow HW. Check the booter - (ppc_boot.c) */ - if (shadow) - { - top -= HARDWARE_MAPPED_SIZE; - mem_pieces_remove(&phys_avail, top, - HARDWARE_MAPPED_SIZE, 0); - } - - /* Remove the upper 512KB where the PPC exception - vectors are mapped. */ - top -= HARDWARE_MAPPED_SIZE; -#if 0 - /* This would be neat, but it breaks on A3000 machines!? */ - mem_pieces_remove(&phys_avail, top, 16384, 0); -#else - mem_pieces_remove(&phys_avail, top, HARDWARE_MAPPED_SIZE, 0); -#endif + /* Remove the upper 512KB if it contains a shadow of + the ADOS ROM. FIXME: It might be possible to + disable this shadow HW. Check the booter + (ppc_boot.c) */ + if (shadow) + total -= HARDWARE_MAPPED_SIZE; - } + /* Remove the upper 512KB where the PPC exception + vectors are mapped. */ + total -= HARDWARE_MAPPED_SIZE; /* Linux/APUS only handles one block of memory -- the one on the PowerUP board. Other system memory is horrible slow in comparison. The user can use other memory for swapping using the z2ram device. */ - return memory[0].addr + memory[0].size; + ram_phys_base = memory[0].addr; + return total; } #endif /* CONFIG_APUS */ @@ -1440,12 +1423,11 @@ unsigned long __init apus_find_end_of_memory(void) static void __init hash_init(void) { int Hash_bits, mb, mb2; - unsigned int hmask, ramsize, h; + unsigned int hmask, h; extern unsigned int hash_page_patch_A[], hash_page_patch_B[], hash_page_patch_C[], hash_page[]; - ramsize = (ulong)end_of_DRAM - KERNELBASE; #ifdef CONFIG_PPC64BRIDGE /* The hash table has already been allocated and initialized in prom.c */ @@ -1464,7 +1446,7 @@ static void __init hash_init(void) * Allow 64k of hash table for every 16MB of memory, * up to a maximum of 2MB. */ - for (h = 64<<10; h < ramsize / 256 && h < (2<<20); h *= 2) + for (h = 64<<10; h < total_memory / 256 && h < (2<<20); h *= 2) ; Hash_size = h; Hash_mask = (h >> 6) - 1; @@ -1498,8 +1480,8 @@ static void __init hash_init(void) Hash = 0; #endif /* CONFIG_PPC64BRIDGE */ - printk("Total memory = %dMB; using %ldkB for hash table (at %p)\n", - ramsize >> 20, Hash_size >> 10, Hash); + printk("Total memory = %ldMB; using %ldkB for hash table (at %p)\n", + total_memory >> 20, Hash_size >> 10, Hash); if ( Hash_size ) { if ( ppc_md.progress ) ppc_md.progress("hash:patch", 0x345); @@ -1560,12 +1542,7 @@ unsigned long __init m8xx_find_end_of_memory(void) binfo = (bd_t *)__res; - phys_mem.regions[0].address = 0; - phys_mem.regions[0].size = binfo->bi_memsize; - phys_mem.n_regions = 1; - - set_phys_avail(&phys_mem); - return phys_mem.regions[0].address + phys_mem.regions[0].size; + return binfo->bi_memsize; } #endif /* !CONFIG_4xx && !CONFIG_8xx */ @@ -1582,20 +1559,16 @@ oak_find_end_of_memory(void) unsigned long *ret; bd_t *bip = (bd_t *)__res; - phys_mem.regions[0].address = 0; - phys_mem.regions[0].size = bip->bi_memsize; - phys_mem.n_regions = 1; - - set_phys_avail(&phys_mem); - return (phys_mem.regions[0].address + phys_mem.regions[0].size); + return bip->bi_memsize; } #endif /* - * Set phys_avail to phys_mem less the kernel text/data/bss. + * Set phys_avail to the amount of physical memory, + * less the kernel text/data/bss. */ void __init -set_phys_avail(struct mem_pieces *mp) +set_phys_avail(unsigned long total_memory) { unsigned long kstart, ksize; @@ -1604,7 +1577,9 @@ set_phys_avail(struct mem_pieces *mp) * physical memory. */ - phys_avail = *mp; + phys_avail.regions[0].address = 0; + phys_avail.regions[0].size = total_memory; + phys_avail.n_regions = 1; /* * Map out the kernel text/data/bss from the available physical diff --git a/arch/ppc/xmon/start.c b/arch/ppc/xmon/start.c index 499c5b6c8..611b1e069 100644 --- a/arch/ppc/xmon/start.c +++ b/arch/ppc/xmon/start.c @@ -14,6 +14,10 @@ #include <asm/bootx.h> #include <asm/feature.h> #include <asm/processor.h> +#include <asm/delay.h> +#ifdef CONFIG_SMP +#include <asm/bitops.h> +#endif static volatile unsigned char *sccc, *sccd; unsigned long TXRDY, RXRDY; @@ -22,10 +26,10 @@ extern void prom_drawchar(char); extern void prom_drawstring(const char *str); static int xmon_expect(const char *str, unsigned int timeout); -static int console = 0; -static int use_screen = 1; /* default */ -static int via_modem = 0; -static int xmon_use_sccb = 0; +static int console; +static int use_screen; +static int via_modem = 1; +static int xmon_use_sccb; static struct device_node *macio_node; #define TB_SPEED 25000000 @@ -53,28 +57,32 @@ xmon_map_scc(void) use_screen = 0; - if ( _machine == _MACH_Pmac ) - { + if (_machine == _MACH_Pmac) { struct device_node *np; unsigned long addr; #ifdef CONFIG_BOOTX_TEXT extern boot_infos_t *disp_bi; + /* see if there is a keyboard in the device tree + with a parent of type "adb" */ + for (np = find_devices("keyboard"); np; np = np->next) + if (np->parent && np->parent->type + && strcmp(np->parent->type, "adb") == 0) + break; + /* needs to be hacked if xmon_printk is to be used from within find_via_pmu() */ #ifdef CONFIG_ADB_PMU - if (!via_modem && disp_bi && find_via_pmu()) { - prom_drawstring("xmon uses screen and keyboard\n"); + if (np != NULL && disp_bi && find_via_pmu()) use_screen = 1; - } #endif #ifdef CONFIG_ADB_CUDA - if (!via_modem && disp_bi ) { - prom_drawstring("xmon uses screen and keyboard\n"); + if (np != NULL && disp_bi && find_via_cuda()) use_screen = 1; - } -#endif #endif + if (use_screen) + prom_drawstring("xmon uses screen and keyboard\n"); +#endif /* CONFIG_BOOTX_TEXT */ #ifdef CHRP_ESCC addr = 0xc1013020; @@ -93,15 +101,6 @@ xmon_map_scc(void) sccc = base + (addr & ~PAGE_MASK); sccd = sccc + 0x10; } - else if ( _machine & _MACH_gemini ) - { - /* should already be mapped by the kernel boot */ - sccc = (volatile unsigned char *) 0xffeffb0d; - sccd = (volatile unsigned char *) 0xffeffb08; - TXRDY = 0x20; - RXRDY = 1; - console = 1; - } else { /* should already be mapped by the kernel boot */ @@ -140,12 +139,22 @@ xmon_write(void *handle, void *ptr, int nb) char *p = ptr; int i, c, ct; +#ifdef CONFIG_SMP + static unsigned long xmon_write_lock; + int lock_wait = 1000000; + int locked; + + while ((locked = test_and_set_bit(0, &xmon_write_lock)) != 0) + if (--lock_wait == 0) + break; +#endif + #ifdef CONFIG_BOOTX_TEXT if (use_screen) { /* write it on the screen */ for (i = 0; i < nb; ++i) prom_drawchar(*p++); - return nb; + goto out; } #endif if (!scc_initialized) @@ -166,8 +175,15 @@ xmon_write(void *handle, void *ptr, int nb) } buf_access(); *sccd = c; + eieio(); } - return i; + + out: +#ifdef CONFIG_SMP + if (!locked) + clear_bit(0, &xmon_write_lock); +#endif + return nb; } int xmon_wants_key; diff --git a/arch/ppc/xmon/xmon.c b/arch/ppc/xmon/xmon.c index 49c3be834..65082d458 100644 --- a/arch/ppc/xmon/xmon.c +++ b/arch/ppc/xmon/xmon.c @@ -406,6 +406,7 @@ cmds(struct pt_regs *excp) break; case 'M': print_sysmap(); + break; case 'S': super_regs(); break; @@ -795,6 +796,8 @@ print_sysmap(void) extern char *sysmap; if ( sysmap ) printf("System.map: \n%s", sysmap); + else + printf("No System.map\n"); } void diff --git a/arch/s390/kernel/signal.c b/arch/s390/kernel/signal.c index a4df59c65..7809a4a2b 100644 --- a/arch/s390/kernel/signal.c +++ b/arch/s390/kernel/signal.c @@ -52,8 +52,6 @@ typedef struct struct ucontext uc; } rt_sigframe; -asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); asmlinkage int FASTCALL(do_signal(struct pt_regs *regs, sigset_t *oldset)); /* diff --git a/arch/sh/boot/compressed/head.S b/arch/sh/boot/compressed/head.S index 75d8b4ef9..0edf49085 100644 --- a/arch/sh/boot/compressed/head.S +++ b/arch/sh/boot/compressed/head.S @@ -43,7 +43,7 @@ bss_start_addr: end_addr: .long _end init_sr: - .long 0x40000000 /* Privileged mode, Bank=0, Block=0, I3-I0=0 */ + .long 0x400000F0 /* Privileged mode, Bank=0, Block=0, IMASK=0xF */ init_stack_addr: .long stack_start decompress_kernel_addr: diff --git a/arch/sh/kernel/entry.S b/arch/sh/kernel/entry.S index 37d29a6c4..f2756f72c 100644 --- a/arch/sh/kernel/entry.S +++ b/arch/sh/kernel/entry.S @@ -36,23 +36,23 @@ * NOTE: This code uses a convention that instructions in the delay slot * of a transfer-control instruction are indented by an extra space, thus: * - * jmp @$k0 ! control-transfer instruction - * ldc $k1, $ssr ! delay slot + * jmp @k0 ! control-transfer instruction + * ldc k1, ssr ! delay slot * * Stack layout in 'ret_from_syscall': * ptrace needs to have all regs on the stack. * if the order here is changed, it needs to be * updated in ptrace.c and ptrace.h * - * $r0 + * r0 * ... - * $r15 = stack pointer - * $spc - * $pr - * $ssr - * $gbr - * $mach - * $macl + * r15 = stack pointer + * spc + * pr + * ssr + * gbr + * mach + * macl * syscall # * */ @@ -88,16 +88,16 @@ MMU_TEA = 0xff00000c ! TLB Exception Address Register #endif /* Offsets to the stack */ -R0 = 0 /* Return value. New ABI also arg4 */ -R1 = 4 /* New ABI: arg5 */ -R2 = 8 /* New ABI: arg6 */ -R3 = 12 /* New ABI: syscall_nr */ -R4 = 16 /* New ABI: arg0 */ -R5 = 20 /* New ABI: arg1 */ -R6 = 24 /* New ABI: arg2 */ -R7 = 28 /* New ABI: arg3 */ -SP = (15*4) -SR = (16*4+8) +OFF_R0 = 0 /* Return value. New ABI also arg4 */ +OFF_R1 = 4 /* New ABI: arg5 */ +OFF_R2 = 8 /* New ABI: arg6 */ +OFF_R3 = 12 /* New ABI: syscall_nr */ +OFF_R4 = 16 /* New ABI: arg0 */ +OFF_R5 = 20 /* New ABI: arg1 */ +OFF_R6 = 24 /* New ABI: arg2 */ +OFF_R7 = 28 /* New ABI: arg3 */ +OFF_SP = (15*4) +OFF_SR = (16*4+8) SYSCALL_NR = (16*4+6*4) @@ -140,117 +140,139 @@ SYSCALL_NR = (16*4+6*4) ! #define STI() \ - mov.l __INV_IMASK, $r11; \ - stc $sr, $r10; \ - and $r11, $r10; \ - stc $k_g_imask, $r11; \ - or $r11, $r10; \ - ldc $r10, $sr + mov.l __INV_IMASK, r11; \ + stc sr, r10; \ + and r11, r10; \ + stc k_g_imask, r11; \ + or r11, r10; \ + ldc r10, sr .align 2 tlb_miss_load: bra call_dpf - mov #0, $r5 + mov #0, r5 .align 2 tlb_miss_store: bra call_dpf - mov #1, $r5 + mov #1, r5 .align 2 initial_page_write: bra call_dpf - mov #1, $r5 + mov #1, r5 .align 2 tlb_protection_violation_load: bra call_dpf - mov #0, $r5 + mov #0, r5 .align 2 tlb_protection_violation_store: bra call_dpf - mov #1, $r5 + mov #1, r5 call_dpf: - mov.l 1f, $r0 - mov $r5, $r8 - mov.l @$r0, $r6 - mov $r6, $r9 - mov.l 2f, $r0 - sts $pr, $r10 - jsr @$r0 - mov $r15, $r4 + mov.l 1f, r0 + mov r5, r8 + mov.l @r0, r6 + mov r6, r9 + mov.l 2f, r0 + sts pr, r10 + jsr @r0 + mov r15, r4 ! - tst #0xff, $r0 + tst #0xff, r0 bf/s 0f - lds $r10, $pr + lds r10, pr rts nop 0: STI() - mov.l 3f, $r0 - mov $r9, $r6 - mov $r8, $r5 - jmp @$r0 - mov $r15, $r4 + mov.l 3f, r0 + mov r9, r6 + mov r8, r5 + jmp @r0 + mov r15, r4 .align 2 1: .long MMU_TEA 2: .long SYMBOL_NAME(__do_page_fault) 3: .long SYMBOL_NAME(do_page_fault) + .align 2 +address_error_load: + bra call_dae + mov #0,r5 ! writeaccess = 0 + + .align 2 +address_error_store: + bra call_dae + mov #1,r5 ! writeaccess = 1 + +call_dae: + mov.l 1f, r0 + mov.l @r0, r6 ! address + mov.l 2f, r0 + jmp @r0 + mov r15, r4 ! regs + + .align 2 +1: .long MMU_TEA +2: .long SYMBOL_NAME(do_address_error) + #if defined(CONFIG_DEBUG_KERNEL_WITH_GDB_STUB) || defined(CONFIG_SH_STANDARD_BIOS) .align 2 /* Unwind the stack and jmp to the debug entry */ debug_kernel: - mov.l @$r15+, $r0 - mov.l @$r15+, $r1 - mov.l @$r15+, $r2 - mov.l @$r15+, $r3 - mov.l @$r15+, $r4 - mov.l @$r15+, $r5 - mov.l @$r15+, $r6 - mov.l @$r15+, $r7 - stc $sr, $r8 - mov.l 1f, $r9 ! BL =1, RB=1, IMASK=0x0F - or $r9, $r8 - ldc $r8, $sr ! here, change the register bank - mov.l @$r15+, $r8 - mov.l @$r15+, $r9 - mov.l @$r15+, $r10 - mov.l @$r15+, $r11 - mov.l @$r15+, $r12 - mov.l @$r15+, $r13 - mov.l @$r15+, $r14 - mov.l @$r15+, $k0 - ldc.l @$r15+, $spc - lds.l @$r15+, $pr - mov.l @$r15+, $k1 - ldc.l @$r15+, $gbr - lds.l @$r15+, $mach - lds.l @$r15+, $macl - mov $k0, $r15 + mov.l @r15+, r0 + mov.l @r15+, r1 + mov.l @r15+, r2 + mov.l @r15+, r3 + mov.l @r15+, r4 + mov.l @r15+, r5 + mov.l @r15+, r6 + mov.l @r15+, r7 + stc sr, r8 + mov.l 1f, r9 ! BL =1, RB=1, IMASK=0x0F + or r9, r8 + ldc r8, sr ! here, change the register bank + mov.l @r15+, r8 + mov.l @r15+, r9 + mov.l @r15+, r10 + mov.l @r15+, r11 + mov.l @r15+, r12 + mov.l @r15+, r13 + mov.l @r15+, r14 + mov.l @r15+, k0 + ldc.l @r15+, spc + lds.l @r15+, pr + mov.l @r15+, k1 + ldc.l @r15+, gbr + lds.l @r15+, mach + lds.l @r15+, macl + mov k0, r15 ! - mov.l 2f, $k0 - jmp @$k0 - ldc $k1, $ssr + mov.l 2f, k0 + mov.l @k0, k0 + jmp @k0 + ldc k1, ssr .align 2 1: .long 0x300000f0 -2: .long CONFIG_GDB_STUB_VBR + 0x100 +2: .long SYMBOL_NAME(gdb_vbr_vector) #endif .align 2 debug_trap: #if defined(CONFIG_DEBUG_KERNEL_WITH_GDB_STUB) || defined(CONFIG_SH_STANDARD_BIOS) - mov #SR, $r0 - mov.l @($r0,$r15), $r0 ! get status register - shll $r0 - shll $r0 ! kernel space? + mov #OFF_SR, r0 + mov.l @(r0,r15), r0 ! get status register + shll r0 + shll r0 ! kernel space? bt/s debug_kernel #endif - mov.l @$r15, $r0 - mov.l 1f, $r8 - jmp @$r8 + mov.l @r15, r0 + mov.l 1f, r8 + jmp @r8 nop .align 2 @@ -260,8 +282,8 @@ debug_trap: error: ! STI() - mov.l 1f, $r0 - jmp @$r0 + mov.l 1f, r0 + jmp @r0 nop .align 2 1: .long SYMBOL_NAME(do_exception_error) @@ -272,7 +294,7 @@ error: ! ENTRY(ret_from_fork) bra SYMBOL_NAME(ret_from_syscall) - add #4, $r15 ! pop down bogus r0 (see switch_to MACRO) + add #4, r15 ! pop down bogus r0 (see switch_to MACRO) /* * Old syscall interface: @@ -305,90 +327,90 @@ ENTRY(ret_from_fork) */ system_call: - mov.l __TRA, $r9 - mov.l @$r9, $r8 + mov.l __TRA, r9 + mov.l @r9, r8 ! ! Is the trap argument >= 0x20? (TRA will be >= 0x80) - mov #0x20, $r9 - extu.b $r9, $r9 - shll2 $r9 - cmp/hs $r9, $r8 + mov #0x20, r9 + extu.b r9, r9 + shll2 r9 + cmp/hs r9, r8 bt debug_trap ! - mov #SYSCALL_NR, $r14 - add $r15, $r14 + mov #SYSCALL_NR, r14 + add r15, r14 ! #ifdef COMPAT_OLD_SYSCALL_ABI - mov #0x40, $r9 - cmp/hs $r9, $r8 + mov #0x40, r9 + cmp/hs r9, r8 bf/s old_abi_system_call nop #endif ! New Syscall ABI - add #-0x40, $r8 - shlr2 $r8 - shll8 $r8 - shll8 $r8 ! $r8 = num_args<<16 - mov $r3, $r10 - or $r8, $r10 ! Encode syscall # and # of arguments - mov.l $r10, @$r14 ! set syscall_nr + add #-0x40, r8 + shlr2 r8 + shll8 r8 + shll8 r8 ! r8 = num_args<<16 + mov r3, r10 + or r8, r10 ! Encode syscall # and # of arguments + mov.l r10, @r14 ! set syscall_nr STI() ! - stc $k_current, $r11 - mov.l @(tsk_ptrace,$r11), $r10 ! Is current PTRACE_SYSCALL'd? - mov #PT_TRACESYS, $r11 - tst $r11, $r10 + stc k_current, r11 + mov.l @(tsk_ptrace,r11), r10 ! Is current PTRACE_SYSCALL'd? + mov #PT_TRACESYS, r11 + tst r11, r10 bt 5f ! Yes it is traced. - mov.l __syscall_trace, $r11 ! Call syscall_trace() which notifies - jsr @$r11 ! superior (will chomp $R[0-7]) + mov.l __syscall_trace, r11 ! Call syscall_trace() which notifies + jsr @r11 ! superior (will chomp R[0-7]) nop - ! Reload $R0-$R4 from kernel stack, where the + ! Reload R0-R4 from kernel stack, where the ! parent may have modified them using - ! ptrace(POKEUSR). (Note that $R0-$R2 are + ! ptrace(POKEUSR). (Note that R0-R2 are ! used by the system call handler directly ! from the kernel stack anyway, so don't need ! to be reloaded here.) This allows the parent ! to rewrite system calls and args on the fly. - mov.l @(R4,$r15), $r4 ! arg0 - mov.l @(R5,$r15), $r5 - mov.l @(R6,$r15), $r6 - mov.l @(R7,$r15), $r7 ! arg3 - mov.l @(R3,$r15), $r3 ! syscall_nr + mov.l @(OFF_R4,r15), r4 ! arg0 + mov.l @(OFF_R5,r15), r5 + mov.l @(OFF_R6,r15), r6 + mov.l @(OFF_R7,r15), r7 ! arg3 + mov.l @(OFF_R3,r15), r3 ! syscall_nr ! Arrange for syscall_trace() to be called ! again as the system call returns. - mov.l __syscall_ret_trace, $r10 + mov.l __syscall_ret_trace, r10 bra 6f - lds $r10, $pr + lds r10, pr ! No it isn't traced. ! Arrange for normal system call return. -5: mov.l __syscall_ret, $r10 - lds $r10, $pr +5: mov.l __syscall_ret, r10 + lds r10, pr ! Call the system call handler through the table. ! (both normal and ptrace'd) ! First check for bad syscall number -6: mov $r3, $r9 - mov.l __n_sys, $r10 - cmp/hs $r10, $r9 +6: mov r3, r9 + mov.l __n_sys, r10 + cmp/hs r10, r9 bf 2f ! Bad syscall number rts ! go to syscall_ret or syscall_ret_trace - mov #-ENOSYS, $r0 + mov #-ENOSYS, r0 ! Good syscall number -2: shll2 $r9 ! x4 - mov.l __sct, $r11 - add $r11, $r9 - mov.l @$r9, $r11 - jmp @$r11 ! jump to specific syscall handler +2: shll2 r9 ! x4 + mov.l __sct, r11 + add r11, r9 + mov.l @r9, r11 + jmp @r11 ! jump to specific syscall handler nop ! In case of trace syscall_ret_trace: - mov.l $r0, @(R0,$r15) ! save the return value - mov.l __syscall_trace, $r1 - mova SYMBOL_NAME(ret_from_syscall), $r0 - jmp @$r1 ! Call syscall_trace() which notifies superior - lds $r0, $pr ! Then return to ret_from_syscall() + mov.l r0, @(OFF_R0,r15) ! save the return value + mov.l __syscall_trace, r1 + mova SYMBOL_NAME(ret_from_syscall), r0 + jmp @r1 ! Call syscall_trace() which notifies superior + lds r0, pr ! Then return to ret_from_syscall() @@ -396,41 +418,41 @@ syscall_ret_trace: ! Handle old ABI system call. ! Note that ptrace(SYSCALL) is not supported for the old ABI. ! At this point: -! $r0, $r4-7 as per ABI -! $r8 = value of TRA register (= num_args<<2) -! $r14 = points to SYSCALL_NR in stack frame +! r0, r4-7 as per ABI +! r8 = value of TRA register (= num_args<<2) +! r14 = points to SYSCALL_NR in stack frame old_abi_system_call: - mov $r0, $r9 ! Save system call number in $r9 + mov r0, r9 ! Save system call number in r9 ! ! arrange for return which pops stack - mov.l __old_abi_syscall_ret, $r10 - lds $r10, $pr + mov.l __old_abi_syscall_ret, r10 + lds r10, pr ! Build the stack frame if TRA > 0 - mov $r8, $r10 - cmp/pl $r10 + mov r8, r10 + cmp/pl r10 bf 0f - mov.l @(SP,$r15), $r0 ! get original user stack -7: add #-4, $r10 -4: mov.l @($r0,$r10), $r1 ! May cause address error exception.. - mov.l $r1, @-$r15 - cmp/pl $r10 + mov.l @(OFF_SP,r15), r0 ! get original user stack +7: add #-4, r10 +4: mov.l @(r0,r10), r1 ! May cause address error exception.. + mov.l r1, @-r15 + cmp/pl r10 bt 7b 0: - mov.l $r9, @$r14 ! set syscall_nr + mov.l r9, @r14 ! set syscall_nr STI() ! Call the system call handler through the table. ! First check for bad syscall number - mov.l __n_sys, $r10 - cmp/hs $r10, $r9 + mov.l __n_sys, r10 + cmp/hs r10, r9 bf 2f ! Bad syscall number rts ! return to old_abi_syscall_ret - mov #-ENOSYS, $r0 + mov #-ENOSYS, r0 ! Good syscall number -2: shll2 $r9 ! x4 - mov.l __sct, $r11 - add $r11, $r9 - mov.l @$r9, $r11 - jmp @$r11 ! call specific syscall handler, +2: shll2 r9 ! x4 + mov.l __sct, r11 + add r11, r9 + mov.l @r9, r11 + jmp @r11 ! call specific syscall handler, nop .align 2 @@ -440,16 +462,16 @@ __old_abi_syscall_ret: ! This code gets called on address error exception when copying ! syscall arguments from user stack to kernel stack. It is ! supposed to return -EINVAL through old_abi_syscall_ret, but it - ! appears to have been broken for a long time in that the $r0 - ! return value will be saved into the kernel stack relative to $r15 - ! but the value of $r15 is not correct partway through the loop. - ! So the user prog is returned its old $r0 value, not -EINVAL. + ! appears to have been broken for a long time in that the r0 + ! return value will be saved into the kernel stack relative to r15 + ! but the value of r15 is not correct partway through the loop. + ! So the user prog is returned its old r0 value, not -EINVAL. ! Greg Banks 28 Aug 2000. .section .fixup,"ax" fixup_syscall_argerr: - ! First get $r15 back to + ! First get r15 back to rts - mov #-EINVAL, $r0 + mov #-EINVAL, r0 .previous .section __ex_table, "a" @@ -473,18 +495,18 @@ __syscall_ret: .align 2 reschedule: - mova SYMBOL_NAME(ret_from_syscall), $r0 - mov.l 1f, $r1 - jmp @$r1 - lds $r0, $pr + mova SYMBOL_NAME(ret_from_syscall), r0 + mov.l 1f, r1 + jmp @r1 + lds r0, pr .align 2 1: .long SYMBOL_NAME(schedule) ENTRY(ret_from_irq) - mov #SR, $r0 - mov.l @($r0,$r15), $r0 ! get status register - shll $r0 - shll $r0 ! kernel space? + mov #OFF_SR, r0 + mov.l @(r0,r15), r0 ! get status register + shll r0 + shll r0 ! kernel space? bt restore_all ! Yes, it's from kernel, go back soon ! STI() @@ -492,10 +514,10 @@ ENTRY(ret_from_irq) nop ENTRY(ret_from_exception) - mov #SR, $r0 - mov.l @($r0,$r15), $r0 ! get status register - shll $r0 - shll $r0 ! kernel space? + mov #OFF_SR, r0 + mov.l @(r0,r15), r0 ! get status register + shll r0 + shll r0 ! kernel space? bt restore_all ! Yes, it's from kernel, go back soon ! STI() @@ -508,38 +530,38 @@ __INV_IMASK: .align 2 #ifdef COMPAT_OLD_SYSCALL_ABI old_abi_syscall_ret: - add $r8, $r15 ! pop off the arguments + add r8, r15 ! pop off the arguments /* fall through */ #endif syscall_ret: - mov.l $r0, @(R0,$r15) ! save the return value + mov.l r0, @(OFF_R0,r15) ! save the return value /* fall through */ ENTRY(ret_from_syscall) - mov.l __irq_stat, $r0 ! softirq_active - mov.l @$r0, $r1 - mov.l @(4,$r0), $r2 ! softirq_mask - tst $r2, $r1 + mov.l __irq_stat, r0 ! softirq_active + mov.l @r0, r1 + mov.l @(4,r0), r2 ! softirq_mask + tst r2, r1 bt ret_with_reschedule handle_softirq: - mov.l __do_softirq, $r0 - jsr @$r0 + mov.l __do_softirq, r0 + jsr @r0 nop ret_with_reschedule: - stc $k_current, $r1 - mov.l @(need_resched,$r1), $r0 - tst #0xff, $r0 + stc k_current, r1 + mov.l @(need_resched,r1), r0 + tst #0xff, r0 bf reschedule - mov.l @(sigpending,$r1), $r0 - tst #0xff, $r0 + mov.l @(sigpending,r1), r0 + tst #0xff, r0 bt restore_all signal_return: - mov $r15, $r4 - mov #0, $r5 - mov.l __do_signal, $r1 - mova restore_all, $r0 - jmp @$r1 - lds $r0, $pr + mov r15, r4 + mov #0, r5 + mov.l __do_signal, r1 + mova restore_all, r0 + jmp @r1 + lds r0, pr .align 2 __do_signal: .long SYMBOL_NAME(do_signal) @@ -551,108 +573,108 @@ __do_softirq: .align 2 restore_all: #if defined(__SH4__) - mov.l __fpu_prepare_fd, $r0 - jsr @$r0 - stc $sr, $r4 + mov.l __fpu_prepare_fd, r0 + jsr @r0 + stc sr, r4 #endif ! - mov.l @$r15+, $r0 - mov.l @$r15+, $r1 - mov.l @$r15+, $r2 - mov.l @$r15+, $r3 - mov.l @$r15+, $r4 - mov.l @$r15+, $r5 - mov.l @$r15+, $r6 - mov.l @$r15+, $r7 + mov.l @r15+, r0 + mov.l @r15+, r1 + mov.l @r15+, r2 + mov.l @r15+, r3 + mov.l @r15+, r4 + mov.l @r15+, r5 + mov.l @r15+, r6 + mov.l @r15+, r7 ! - stc $sr, $r8 - mov.l __blrb_flags, $r9 ! BL =1, RB=1 - or $r9, $r8 - ldc $r8, $sr ! here, change the register bank + stc sr, r8 + mov.l __blrb_flags, r9 ! BL =1, RB=1 + or r9, r8 + ldc r8, sr ! here, change the register bank ! - mov.l @$r15+, $r8 - mov.l @$r15+, $r9 - mov.l @$r15+, $r10 - mov.l @$r15+, $r11 - mov.l @$r15+, $r12 - mov.l @$r15+, $r13 - mov.l @$r15+, $r14 - mov.l @$r15+, $k4 ! original stack pointer - ldc.l @$r15+, $spc - lds.l @$r15+, $pr - mov.l @$r15+, $k3 ! original SR - ldc.l @$r15+, $gbr - lds.l @$r15+, $mach - lds.l @$r15+, $macl - add #4, $r15 ! Skip syscall number + mov.l @r15+, r8 + mov.l @r15+, r9 + mov.l @r15+, r10 + mov.l @r15+, r11 + mov.l @r15+, r12 + mov.l @r15+, r13 + mov.l @r15+, r14 + mov.l @r15+, k4 ! original stack pointer + ldc.l @r15+, spc + lds.l @r15+, pr + mov.l @r15+, k3 ! original SR + ldc.l @r15+, gbr + lds.l @r15+, mach + lds.l @r15+, macl + add #4, r15 ! Skip syscall number ! ! Calculate new SR value - mov $k3, $k2 ! original SR value - mov.l 1f, $k1 - stc $sr, $k0 - and $k1, $k0 ! Get current FD-bit - mov.l 2f, $k1 - and $k1, $k2 ! Mask orignal SR value - or $k0, $k2 ! Inherit current FD-bit + mov k3, k2 ! original SR value + mov.l 1f, k1 + stc sr, k0 + and k1, k0 ! Get current FD-bit + mov.l 2f, k1 + and k1, k2 ! Mask orignal SR value + or k0, k2 ! Inherit current FD-bit ! - mov $k3, $k0 ! Calculate IMASK-bits - shlr2 $k0 - and #0x3c, $k0 - cmp/eq #0x3c, $k0 + mov k3, k0 ! Calculate IMASK-bits + shlr2 k0 + and #0x3c, k0 + cmp/eq #0x3c, k0 bt/s 7f - shll2 $k0 - mov $g_imask, $k0 + shll2 k0 + mov g_imask, k0 ! -7: or $k0, $k2 ! Set the IMASK-bits - ldc $k2, $ssr +7: or k0, k2 ! Set the IMASK-bits + ldc k2, ssr ! #if defined(__SH4__) - shll $k2 - shll $k2 + shll k2 + shll k2 bf 9f ! user mode /* Kernel to kernel transition */ - mov.l 1f, $k1 - tst $k1, $k3 + mov.l 1f, k1 + tst k1, k3 bf 9f ! it hadn't FPU ! Kernel to kernel and FPU was used ! There's the case we don't get FPU now - stc $sr, $k2 - tst $k1, $k2 + stc sr, k2 + tst k1, k2 bt 8f ! We need to grab FPU here - xor $k1, $k2 - ldc $k2, $sr ! Grab FPU - mov.l __init_task_flags, $k1 - mov.l @$k1, $k2 - mov.l __PF_USEDFPU, $k0 - or $k0, $k2 - mov.l $k2, @$k1 ! Set init_task.flags |= PF_USEDFPU + xor k1, k2 + ldc k2, sr ! Grab FPU + mov.l __init_task_flags, k1 + mov.l @k1, k2 + mov.l __PF_USEDFPU, k0 + or k0, k2 + mov.l k2, @k1 ! Set init_task.flags |= PF_USEDFPU ! ! Restoring FPU... ! -8: mov.l 3f, $k1 - lds $k1, $fpscr - fmov.s @$r15+, $fr0 - fmov.s @$r15+, $fr1 - fmov.s @$r15+, $fr2 - fmov.s @$r15+, $fr3 - fmov.s @$r15+, $fr4 - fmov.s @$r15+, $fr5 - fmov.s @$r15+, $fr6 - fmov.s @$r15+, $fr7 - fmov.s @$r15+, $fr8 - fmov.s @$r15+, $fr9 - fmov.s @$r15+, $fr10 - fmov.s @$r15+, $fr11 - fmov.s @$r15+, $fr12 - fmov.s @$r15+, $fr13 - fmov.s @$r15+, $fr14 - fmov.s @$r15+, $fr15 - lds.l @$r15+, $fpscr - lds.l @$r15+, $fpul +8: mov.l 3f, k1 + lds k1, fpscr + fmov.s @r15+, fr0 + fmov.s @r15+, fr1 + fmov.s @r15+, fr2 + fmov.s @r15+, fr3 + fmov.s @r15+, fr4 + fmov.s @r15+, fr5 + fmov.s @r15+, fr6 + fmov.s @r15+, fr7 + fmov.s @r15+, fr8 + fmov.s @r15+, fr9 + fmov.s @r15+, fr10 + fmov.s @r15+, fr11 + fmov.s @r15+, fr12 + fmov.s @r15+, fr13 + fmov.s @r15+, fr14 + fmov.s @r15+, fr15 + lds.l @r15+, fpscr + lds.l @r15+, fpul 9: #endif - mov $k4, $r15 + mov k4, r15 rte nop @@ -680,10 +702,10 @@ ENTRY(vbr_base) ! .balign 256,0,256 general_exception: - mov.l 1f, $k2 - mov.l 2f, $k3 + mov.l 1f, k2 + mov.l 2f, k3 bra handle_exception - mov.l @$k2, $k2 + mov.l @k2, k2 .align 2 2: .long SYMBOL_NAME(ret_from_exception) 1: .long EXPEVT @@ -691,17 +713,17 @@ general_exception: ! .balign 1024,0,1024 tlb_miss: - mov.l 1f, $k2 - mov.l 4f, $k3 + mov.l 1f, k2 + mov.l 4f, k3 bra handle_exception - mov.l @$k2, $k2 + mov.l @k2, k2 ! .balign 512,0,512 interrupt: - mov.l 2f, $k2 - mov.l 3f, $k3 + mov.l 2f, k2 + mov.l 3f, k3 bra handle_exception - mov.l @$k2, $k2 + mov.l @k2, k2 .align 2 1: .long EXPEVT @@ -715,102 +737,102 @@ handle_exception: ! Using k0, k1 for scratch registers (r0_bank1, r1_bank), ! save all registers onto stack. ! - stc $ssr, $k0 ! from kernel space? - shll $k0 ! Check MD bit (bit30) by shifting it into the T bit - shll $k0 + stc ssr, k0 ! from kernel space? + shll k0 ! Check MD bit (bit30) by shifting it into the T bit + shll k0 #if defined(__SH4__) bf/s 8f ! it's from user to kernel transition - mov $r15, $k0 ! save original stack to k0 + mov r15, k0 ! save original stack to k0 /* It's a kernel to kernel transition. */ /* Is the FPU disabled? */ - mov.l 2f, $k1 - stc $ssr, $k0 - tst $k1, $k0 - mov.l 4f, $k1 + mov.l 2f, k1 + stc ssr, k0 + tst k1, k0 + mov.l 4f, k1 bf/s 9f ! FPU is not enabled, no need to save it - mov $r15, $k0 ! save original stack to k0 + mov r15, k0 ! save original stack to k0 ! FPU is enabled, save it ! /* XXX: Need to save another bank of FPU if all FPU feature is used */ ! /* Currently it's not the case for GCC (only udivsi3_i4, divsi3_i4) */ - sts.l $fpul, @-$r15 - sts.l $fpscr, @-$r15 - mov.l 6f, $k1 - lds $k1, $fpscr - mov.l 3f, $k1 - fmov.s $fr15, @-$r15 - fmov.s $fr14, @-$r15 - fmov.s $fr13, @-$r15 - fmov.s $fr12, @-$r15 - fmov.s $fr11, @-$r15 - fmov.s $fr10, @-$r15 - fmov.s $fr9, @-$r15 - fmov.s $fr8, @-$r15 - fmov.s $fr7, @-$r15 - fmov.s $fr6, @-$r15 - fmov.s $fr5, @-$r15 - fmov.s $fr4, @-$r15 - fmov.s $fr3, @-$r15 - fmov.s $fr2, @-$r15 - fmov.s $fr1, @-$r15 + sts.l fpul, @-r15 + sts.l fpscr, @-r15 + mov.l 6f, k1 + lds k1, fpscr + mov.l 3f, k1 + fmov.s fr15, @-r15 + fmov.s fr14, @-r15 + fmov.s fr13, @-r15 + fmov.s fr12, @-r15 + fmov.s fr11, @-r15 + fmov.s fr10, @-r15 + fmov.s fr9, @-r15 + fmov.s fr8, @-r15 + fmov.s fr7, @-r15 + fmov.s fr6, @-r15 + fmov.s fr5, @-r15 + fmov.s fr4, @-r15 + fmov.s fr3, @-r15 + fmov.s fr2, @-r15 + fmov.s fr1, @-r15 bra 9f - fmov.s $fr0, @-$r15 + fmov.s fr0, @-r15 #else - mov.l 3f, $k1 + mov.l 3f, k1 bt/s 9f ! it's a kernel to kernel transition, and skip the FPU save. - mov $r15, $k0 ! save original stack to k0 anyway + mov r15, k0 ! save original stack to k0 anyway #endif 8: /* User space to kernel */ - mov #0x20, $k1 - shll8 $k1 ! $k1 <= 8192 == THREAD_SIZE - add $current, $k1 - mov $k1, $r15 ! change to kernel stack + mov #0x20, k1 + shll8 k1 ! k1 <= 8192 == THREAD_SIZE + add current, k1 + mov k1, r15 ! change to kernel stack ! - mov.l 4f, $k1 ! let kernel release FPU + mov.l 4f, k1 ! let kernel release FPU 9: ! Save the user registers on the stack. ! At this point, k1 should have been set to the new SR value - mov #-1, $k4 - mov.l $k4, @-$r15 ! syscall_nr (default: -1) + mov #-1, k4 + mov.l k4, @-r15 ! syscall_nr (default: -1) ! - sts.l $macl, @-$r15 - sts.l $mach, @-$r15 - stc.l $gbr, @-$r15 - stc.l $ssr, @-$r15 - sts.l $pr, @-$r15 - stc.l $spc, @-$r15 + sts.l macl, @-r15 + sts.l mach, @-r15 + stc.l gbr, @-r15 + stc.l ssr, @-r15 + sts.l pr, @-r15 + stc.l spc, @-r15 ! - lds $k3, $pr ! Set the return address to pr + lds k3, pr ! Set the return address to pr ! - mov.l $k0, @-$r15 ! save orignal stack - mov.l $r14, @-$r15 - mov.l $r13, @-$r15 - mov.l $r12, @-$r15 - mov.l $r11, @-$r15 - mov.l $r10, @-$r15 - mov.l $r9, @-$r15 - mov.l $r8, @-$r15 + mov.l k0, @-r15 ! save orignal stack + mov.l r14, @-r15 + mov.l r13, @-r15 + mov.l r12, @-r15 + mov.l r11, @-r15 + mov.l r10, @-r15 + mov.l r9, @-r15 + mov.l r8, @-r15 ! - stc $sr, $r8 ! Back to normal register bank, and - or $k1, $r8 ! Block all interrupts, may release FPU - mov.l 5f, $k1 - and $k1, $r8 ! ... - ldc $r8, $sr ! ...changed here. + stc sr, r8 ! Back to normal register bank, and + or k1, r8 ! Block all interrupts, may release FPU + mov.l 5f, k1 + and k1, r8 ! ... + ldc r8, sr ! ...changed here. ! - mov.l $r7, @-$r15 - mov.l $r6, @-$r15 - mov.l $r5, @-$r15 - mov.l $r4, @-$r15 - mov.l $r3, @-$r15 - mov.l $r2, @-$r15 - mov.l $r1, @-$r15 - mov.l $r0, @-$r15 + mov.l r7, @-r15 + mov.l r6, @-r15 + mov.l r5, @-r15 + mov.l r4, @-r15 + mov.l r3, @-r15 + mov.l r2, @-r15 + mov.l r1, @-r15 + mov.l r0, @-r15 ! Then, dispatch to the handler, according to the exception code. - stc $k_ex_code, $r8 - shlr2 $r8 - shlr $r8 - mov.l 1f, $r9 - add $r8, $r9 - mov.l @$r9, $r9 - jmp @$r9 + stc k_ex_code, r8 + shlr2 r8 + shlr r8 + mov.l 1f, r9 + add r8, r9 + mov.l @r9, r9 + jmp @r9 nop .align 2 1: .long SYMBOL_NAME(exception_handling_table) @@ -833,8 +855,8 @@ ENTRY(exception_handling_table) .long initial_page_write .long tlb_protection_violation_load .long tlb_protection_violation_store - .long error ! address_error_load (filled by trap_init) - .long error ! address_error_store (filled by trap_init) + .long address_error_load + .long address_error_store #if defined(__SH4__) .long SYMBOL_NAME(do_fpu_error) #else diff --git a/arch/sh/kernel/fpu.c b/arch/sh/kernel/fpu.c index cbeb60d31..9036b20c0 100644 --- a/arch/sh/kernel/fpu.c +++ b/arch/sh/kernel/fpu.c @@ -21,43 +21,43 @@ void save_fpu(struct task_struct *tsk) { - asm volatile("sts.l $fpul, @-%0\n\t" - "sts.l $fpscr, @-%0\n\t" - "lds %1, $fpscr\n\t" + asm volatile("sts.l fpul, @-%0\n\t" + "sts.l fpscr, @-%0\n\t" + "lds %1, fpscr\n\t" "frchg\n\t" - "fmov.s $fr15, @-%0\n\t" - "fmov.s $fr14, @-%0\n\t" - "fmov.s $fr13, @-%0\n\t" - "fmov.s $fr12, @-%0\n\t" - "fmov.s $fr11, @-%0\n\t" - "fmov.s $fr10, @-%0\n\t" - "fmov.s $fr9, @-%0\n\t" - "fmov.s $fr8, @-%0\n\t" - "fmov.s $fr7, @-%0\n\t" - "fmov.s $fr6, @-%0\n\t" - "fmov.s $fr5, @-%0\n\t" - "fmov.s $fr4, @-%0\n\t" - "fmov.s $fr3, @-%0\n\t" - "fmov.s $fr2, @-%0\n\t" - "fmov.s $fr1, @-%0\n\t" - "fmov.s $fr0, @-%0\n\t" + "fmov.s fr15, @-%0\n\t" + "fmov.s fr14, @-%0\n\t" + "fmov.s fr13, @-%0\n\t" + "fmov.s fr12, @-%0\n\t" + "fmov.s fr11, @-%0\n\t" + "fmov.s fr10, @-%0\n\t" + "fmov.s fr9, @-%0\n\t" + "fmov.s fr8, @-%0\n\t" + "fmov.s fr7, @-%0\n\t" + "fmov.s fr6, @-%0\n\t" + "fmov.s fr5, @-%0\n\t" + "fmov.s fr4, @-%0\n\t" + "fmov.s fr3, @-%0\n\t" + "fmov.s fr2, @-%0\n\t" + "fmov.s fr1, @-%0\n\t" + "fmov.s fr0, @-%0\n\t" "frchg\n\t" - "fmov.s $fr15, @-%0\n\t" - "fmov.s $fr14, @-%0\n\t" - "fmov.s $fr13, @-%0\n\t" - "fmov.s $fr12, @-%0\n\t" - "fmov.s $fr11, @-%0\n\t" - "fmov.s $fr10, @-%0\n\t" - "fmov.s $fr9, @-%0\n\t" - "fmov.s $fr8, @-%0\n\t" - "fmov.s $fr7, @-%0\n\t" - "fmov.s $fr6, @-%0\n\t" - "fmov.s $fr5, @-%0\n\t" - "fmov.s $fr4, @-%0\n\t" - "fmov.s $fr3, @-%0\n\t" - "fmov.s $fr2, @-%0\n\t" - "fmov.s $fr1, @-%0\n\t" - "fmov.s $fr0, @-%0" + "fmov.s fr15, @-%0\n\t" + "fmov.s fr14, @-%0\n\t" + "fmov.s fr13, @-%0\n\t" + "fmov.s fr12, @-%0\n\t" + "fmov.s fr11, @-%0\n\t" + "fmov.s fr10, @-%0\n\t" + "fmov.s fr9, @-%0\n\t" + "fmov.s fr8, @-%0\n\t" + "fmov.s fr7, @-%0\n\t" + "fmov.s fr6, @-%0\n\t" + "fmov.s fr5, @-%0\n\t" + "fmov.s fr4, @-%0\n\t" + "fmov.s fr3, @-%0\n\t" + "fmov.s fr2, @-%0\n\t" + "fmov.s fr1, @-%0\n\t" + "fmov.s fr0, @-%0" : /* no output */ : "r" ((char *)(&tsk->thread.fpu.hard.status)), "r" (FPSCR_INIT) @@ -70,43 +70,43 @@ save_fpu(struct task_struct *tsk) static void restore_fpu(struct task_struct *tsk) { - asm volatile("lds %1, $fpscr\n\t" - "fmov.s @%0+, $fr0\n\t" - "fmov.s @%0+, $fr1\n\t" - "fmov.s @%0+, $fr2\n\t" - "fmov.s @%0+, $fr3\n\t" - "fmov.s @%0+, $fr4\n\t" - "fmov.s @%0+, $fr5\n\t" - "fmov.s @%0+, $fr6\n\t" - "fmov.s @%0+, $fr7\n\t" - "fmov.s @%0+, $fr8\n\t" - "fmov.s @%0+, $fr9\n\t" - "fmov.s @%0+, $fr10\n\t" - "fmov.s @%0+, $fr11\n\t" - "fmov.s @%0+, $fr12\n\t" - "fmov.s @%0+, $fr13\n\t" - "fmov.s @%0+, $fr14\n\t" - "fmov.s @%0+, $fr15\n\t" + asm volatile("lds %1, fpscr\n\t" + "fmov.s @%0+, fr0\n\t" + "fmov.s @%0+, fr1\n\t" + "fmov.s @%0+, fr2\n\t" + "fmov.s @%0+, fr3\n\t" + "fmov.s @%0+, fr4\n\t" + "fmov.s @%0+, fr5\n\t" + "fmov.s @%0+, fr6\n\t" + "fmov.s @%0+, fr7\n\t" + "fmov.s @%0+, fr8\n\t" + "fmov.s @%0+, fr9\n\t" + "fmov.s @%0+, fr10\n\t" + "fmov.s @%0+, fr11\n\t" + "fmov.s @%0+, fr12\n\t" + "fmov.s @%0+, fr13\n\t" + "fmov.s @%0+, fr14\n\t" + "fmov.s @%0+, fr15\n\t" "frchg\n\t" - "fmov.s @%0+, $fr0\n\t" - "fmov.s @%0+, $fr1\n\t" - "fmov.s @%0+, $fr2\n\t" - "fmov.s @%0+, $fr3\n\t" - "fmov.s @%0+, $fr4\n\t" - "fmov.s @%0+, $fr5\n\t" - "fmov.s @%0+, $fr6\n\t" - "fmov.s @%0+, $fr7\n\t" - "fmov.s @%0+, $fr8\n\t" - "fmov.s @%0+, $fr9\n\t" - "fmov.s @%0+, $fr10\n\t" - "fmov.s @%0+, $fr11\n\t" - "fmov.s @%0+, $fr12\n\t" - "fmov.s @%0+, $fr13\n\t" - "fmov.s @%0+, $fr14\n\t" - "fmov.s @%0+, $fr15\n\t" + "fmov.s @%0+, fr0\n\t" + "fmov.s @%0+, fr1\n\t" + "fmov.s @%0+, fr2\n\t" + "fmov.s @%0+, fr3\n\t" + "fmov.s @%0+, fr4\n\t" + "fmov.s @%0+, fr5\n\t" + "fmov.s @%0+, fr6\n\t" + "fmov.s @%0+, fr7\n\t" + "fmov.s @%0+, fr8\n\t" + "fmov.s @%0+, fr9\n\t" + "fmov.s @%0+, fr10\n\t" + "fmov.s @%0+, fr11\n\t" + "fmov.s @%0+, fr12\n\t" + "fmov.s @%0+, fr13\n\t" + "fmov.s @%0+, fr14\n\t" + "fmov.s @%0+, fr15\n\t" "frchg\n\t" - "lds.l @%0+, $fpscr\n\t" - "lds.l @%0+, $fpul\n\t" + "lds.l @%0+, fpscr\n\t" + "lds.l @%0+, fpul\n\t" : /* no output */ : "r" (&tsk->thread.fpu), "r" (FPSCR_INIT) : "memory"); @@ -120,41 +120,41 @@ restore_fpu(struct task_struct *tsk) void fpu_init(void) { - asm volatile("lds %0, $fpul\n\t" - "lds %1, $fpscr\n\t" - "fsts $fpul, $fr0\n\t" - "fsts $fpul, $fr1\n\t" - "fsts $fpul, $fr2\n\t" - "fsts $fpul, $fr3\n\t" - "fsts $fpul, $fr4\n\t" - "fsts $fpul, $fr5\n\t" - "fsts $fpul, $fr6\n\t" - "fsts $fpul, $fr7\n\t" - "fsts $fpul, $fr8\n\t" - "fsts $fpul, $fr9\n\t" - "fsts $fpul, $fr10\n\t" - "fsts $fpul, $fr11\n\t" - "fsts $fpul, $fr12\n\t" - "fsts $fpul, $fr13\n\t" - "fsts $fpul, $fr14\n\t" - "fsts $fpul, $fr15\n\t" + asm volatile("lds %0, fpul\n\t" + "lds %1, fpscr\n\t" + "fsts fpul, fr0\n\t" + "fsts fpul, fr1\n\t" + "fsts fpul, fr2\n\t" + "fsts fpul, fr3\n\t" + "fsts fpul, fr4\n\t" + "fsts fpul, fr5\n\t" + "fsts fpul, fr6\n\t" + "fsts fpul, fr7\n\t" + "fsts fpul, fr8\n\t" + "fsts fpul, fr9\n\t" + "fsts fpul, fr10\n\t" + "fsts fpul, fr11\n\t" + "fsts fpul, fr12\n\t" + "fsts fpul, fr13\n\t" + "fsts fpul, fr14\n\t" + "fsts fpul, fr15\n\t" "frchg\n\t" - "fsts $fpul, $fr0\n\t" - "fsts $fpul, $fr1\n\t" - "fsts $fpul, $fr2\n\t" - "fsts $fpul, $fr3\n\t" - "fsts $fpul, $fr4\n\t" - "fsts $fpul, $fr5\n\t" - "fsts $fpul, $fr6\n\t" - "fsts $fpul, $fr7\n\t" - "fsts $fpul, $fr8\n\t" - "fsts $fpul, $fr9\n\t" - "fsts $fpul, $fr10\n\t" - "fsts $fpul, $fr11\n\t" - "fsts $fpul, $fr12\n\t" - "fsts $fpul, $fr13\n\t" - "fsts $fpul, $fr14\n\t" - "fsts $fpul, $fr15\n\t" + "fsts fpul, fr0\n\t" + "fsts fpul, fr1\n\t" + "fsts fpul, fr2\n\t" + "fsts fpul, fr3\n\t" + "fsts fpul, fr4\n\t" + "fsts fpul, fr5\n\t" + "fsts fpul, fr6\n\t" + "fsts fpul, fr7\n\t" + "fsts fpul, fr8\n\t" + "fsts fpul, fr9\n\t" + "fsts fpul, fr10\n\t" + "fsts fpul, fr11\n\t" + "fsts fpul, fr12\n\t" + "fsts fpul, fr13\n\t" + "fsts fpul, fr14\n\t" + "fsts fpul, fr15\n\t" "frchg" : /* no output */ : "r" (0), "r" (FPSCR_INIT)); @@ -192,9 +192,9 @@ do_fpu_state_restore(unsigned long r4, unsigned long r5, unsigned long r6, * * There's race condition in __cli: * - * (1) $SR --> register + * (1) SR --> register * (2) Set IMASK of register - * (3) $SR <-- register + * (3) SR <-- register * * Between (1) and (2), or (2) and (3) getting * interrupt, and interrupt handler (or diff --git a/arch/sh/kernel/head.S b/arch/sh/kernel/head.S index f1ac6fd17..73c13dc7d 100644 --- a/arch/sh/kernel/head.S +++ b/arch/sh/kernel/head.S @@ -37,40 +37,40 @@ ENTRY(empty_zero_page) */ ENTRY(_stext) ! Initialize Status Register - mov.l 1f, $r0 ! MD=1, RB=0, BL=0, IMASK=0xF - ldc $r0, $sr + mov.l 1f, r0 ! MD=1, RB=0, BL=0, IMASK=0xF + ldc r0, sr ! Initialize global interrupt mask - mov #0, $r0 - ldc $r0, $r6_bank + mov #0, r0 + ldc r0, r6_bank ! - mov.l 2f, $r0 - mov $r0, $r15 ! Set initial r15 (stack pointer) - mov #0x20, $r1 ! - shll8 $r1 ! $r1 = 8192 - sub $r1, $r0 ! - ldc $r0, $r7_bank ! ... and init_task + mov.l 2f, r0 + mov r0, r15 ! Set initial r15 (stack pointer) + mov #0x20, r1 ! + shll8 r1 ! r1 = 8192 + sub r1, r0 ! + ldc r0, r7_bank ! ... and init_task ! #if defined(__SH4__) ! Initialize fpu - mov.l 7f, $r0 - jsr @$r0 + mov.l 7f, r0 + jsr @r0 nop #endif ! Enable cache - mov.l 6f, $r0 - jsr @$r0 + mov.l 6f, r0 + jsr @r0 nop ! Clear BSS area - mov.l 3f, $r1 - add #4, $r1 - mov.l 4f, $r2 - mov #0, $r0 -9: cmp/hs $r2, $r1 + mov.l 3f, r1 + add #4, r1 + mov.l 4f, r2 + mov #0, r0 +9: cmp/hs r2, r1 bf/s 9b ! while (r1 < r2) - mov.l $r0,@-$r2 + mov.l r0,@-r2 ! Start kernel - mov.l 5f, $r0 - jmp @$r0 + mov.l 5f, r0 + jmp @r0 nop .balign 4 diff --git a/arch/sh/kernel/irq.c b/arch/sh/kernel/irq.c index 6451c4c9e..200148320 100644 --- a/arch/sh/kernel/irq.c +++ b/arch/sh/kernel/irq.c @@ -235,7 +235,7 @@ asmlinkage int do_IRQ(unsigned long r4, unsigned long r5, unsigned int status; /* Get IRQ number */ - asm volatile("stc $r2_bank, %0\n\t" + asm volatile("stc r2_bank, %0\n\t" "shlr2 %0\n\t" "shlr2 %0\n\t" "shlr %0\n\t" diff --git a/arch/sh/kernel/irq_imask.c b/arch/sh/kernel/irq_imask.c index 27d91b372..d0b3aea9f 100644 --- a/arch/sh/kernel/irq_imask.c +++ b/arch/sh/kernel/irq_imask.c @@ -59,16 +59,16 @@ void static inline set_interrupt_registers(int ip) { unsigned long __dummy; - asm volatile("ldc %2, $r6_bank\n\t" - "stc $sr, %0\n\t" + asm volatile("ldc %2, r6_bank\n\t" + "stc sr, %0\n\t" "and #0xf0, %0\n\t" "shlr2 %0\n\t" "cmp/eq #0x3c, %0\n\t" "bt/s 1f ! CLI-ed\n\t" - " stc $sr, %0\n\t" + " stc sr, %0\n\t" "and %1, %0\n\t" "or %2, %0\n\t" - "ldc %0, $sr\n" + "ldc %0, sr\n" "1:" : "=&z" (__dummy) : "r" (~0xf0), "r" (ip << 4) diff --git a/arch/sh/kernel/process.c b/arch/sh/kernel/process.c index c7511093b..1ce22f0fd 100644 --- a/arch/sh/kernel/process.c +++ b/arch/sh/kernel/process.c @@ -14,6 +14,8 @@ #define __KERNEL_SYSCALLS__ #include <stdarg.h> +#include <linux/config.h> + #include <linux/errno.h> #include <linux/sched.h> #include <linux/kernel.h> @@ -31,6 +33,7 @@ #include <linux/delay.h> #include <linux/reboot.h> #include <linux/init.h> +#include <linux/irq.h> #include <asm/uaccess.h> #include <asm/pgtable.h> @@ -39,8 +42,9 @@ #include <asm/processor.h> #include <asm/mmu_context.h> #include <asm/elf.h> - -#include <linux/irq.h> +#ifdef CONFIG_SH_STANDARD_BIOS +#include <asm/sh_bios.h> +#endif static int hlt_counter=0; @@ -79,11 +83,17 @@ void cpu_idle(void *unused) } void machine_restart(char * __unused) -{ /* Need to set MMU_TTB?? */ +{ +#ifdef CONFIG_SH_STANDARD_BIOS + sh_bios_shutdown(1); +#endif } void machine_halt(void) { +#ifdef CONFIG_SH_STANDARD_BIOS + sh_bios_shutdown(0); +#endif } void machine_power_off(void) @@ -93,7 +103,7 @@ void machine_power_off(void) void show_regs(struct pt_regs * regs) { printk("\n"); - printk("PC : %08lx SP : %08lx SR : %08lx TEA : %08lx\n", + printk("PC : %08lx SP : %08lx SR : %08lx TEA : %08x\n", regs->pc, regs->regs[15], regs->sr, ctrl_inl(MMU_TEA)); printk("R0 : %08lx R1 : %08lx R2 : %08lx R3 : %08lx\n", regs->regs[0],regs->regs[1], @@ -144,12 +154,12 @@ int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags) register unsigned long __sc9 __asm__ ("r9") = (long) fn; __asm__("trapa #0x12\n\t" /* Linux/SH system call */ - "tst #0xff, $r0\n\t" /* child or parent? */ + "tst #0xff, r0\n\t" /* child or parent? */ "bf 1f\n\t" /* parent - jump */ - "jsr @$r9\n\t" /* call fn */ - " mov $r8, $r4\n\t" /* push argument */ - "mov $r0, $r4\n\t" /* return value to arg of exit */ - "mov %1, $r3\n\t" /* exit */ + "jsr @r9\n\t" /* call fn */ + " mov r8, r4\n\t" /* push argument */ + "mov r0, r4\n\t" /* return value to arg of exit */ + "mov %1, r3\n\t" /* exit */ "trapa #0x11\n" "1:" : "=z" (__sc0) @@ -285,7 +295,7 @@ void __switch_to(struct task_struct *prev, struct task_struct *next) * Restore the kernel mode register * k7 (r7_bank1) */ - asm volatile("ldc %0, $r7_bank" + asm volatile("ldc %0, r7_bank" : /* no output */ :"r" (next)); } @@ -376,7 +386,7 @@ unsigned long get_wchan(struct task_struct *p) asmlinkage void print_syscall(int x) { unsigned long flags, sr; - asm("stc $sr, %0": "=r" (sr)); + asm("stc sr, %0": "=r" (sr)); save_and_cli(flags); printk("%c: %c %c, %c: SYSCALL\n", (x&63)+32, (current->flags&PF_USEDFPU)?'C':' ', diff --git a/arch/sh/kernel/setup.c b/arch/sh/kernel/setup.c index 8dad245df..b6fb1e9a8 100644 --- a/arch/sh/kernel/setup.c +++ b/arch/sh/kernel/setup.c @@ -503,8 +503,8 @@ int get_cpuinfo(char *buffer) "cache size\t: 8K-byte/16K-byte\n"); #endif p += sprintf(p, "bogomips\t: %lu.%02lu\n\n", - (loops_per_jiffy+2500)/(500000/HZ), - ((loops_per_jiffy+2500)/(5000/HZ)) % 100); + loops_per_jiffy/(500000/HZ), + (loops_per_jiffy/(5000/HZ)) % 100); p += sprintf(p, "Machine: %s\n", sh_mv.mv_name); #define PRINT_CLOCK(name, value) \ diff --git a/arch/sh/kernel/sh_bios.c b/arch/sh/kernel/sh_bios.c index 81a56b960..f72f71b98 100644 --- a/arch/sh/kernel/sh_bios.c +++ b/arch/sh/kernel/sh_bios.c @@ -1,4 +1,4 @@ -/* $Id: sh_bios.c,v 1.3 2000/09/30 03:43:30 gniibe Exp $ +/* $Id: sh_bios.c,v 1.5 2001/01/08 08:42:32 gniibe Exp $ * * linux/arch/sh/kernel/sh_bios.c * C interface for trapping into the standard LinuxSH BIOS. @@ -7,13 +7,12 @@ * */ -#include <linux/config.h> #include <asm/sh_bios.h> -#ifdef CONFIG_SH_STANDARD_BIOS - #define BIOS_CALL_CONSOLE_WRITE 0 -#define BIOS_CALL_READ_BLOCK 1 /* not implemented */ +#define BIOS_CALL_READ_BLOCK 1 +#define BIOS_CALL_ETH_NODE_ADDR 10 +#define BIOS_CALL_SHUTDOWN 11 #define BIOS_CALL_CHAR_OUT 0x1f /* TODO: hack */ #define BIOS_CALL_GDB_GET_MODE_PTR 0xfe #define BIOS_CALL_GDB_DETACH 0xff @@ -66,5 +65,12 @@ void sh_bios_gdb_detach(void) sh_bios_call(BIOS_CALL_GDB_DETACH, 0, 0, 0, 0); } -#endif +void sh_bios_get_node_addr (unsigned char *node_addr) +{ + sh_bios_call(BIOS_CALL_ETH_NODE_ADDR, 0, (long)node_addr, 0, 0); +} +void sh_bios_shutdown(unsigned int how) +{ + sh_bios_call(BIOS_CALL_SHUTDOWN, how, 0, 0, 0); +} diff --git a/arch/sh/kernel/sh_ksyms.c b/arch/sh/kernel/sh_ksyms.c index 2b0298699..2bdcd75ea 100644 --- a/arch/sh/kernel/sh_ksyms.c +++ b/arch/sh/kernel/sh_ksyms.c @@ -64,10 +64,11 @@ EXPORT_SYMBOL(get_vm_area); #define DECLARE_EXPORT(name) extern void name(void);EXPORT_SYMBOL_NOVERS(name) /* These symbols are generated by the compiler itself */ +DECLARE_EXPORT(__udivsi3); +DECLARE_EXPORT(__sdivsi3); + #ifdef __SH4__ -DECLARE_EXPORT(__udivsi3_i4); -DECLARE_EXPORT(__sdivsi3_i4); DECLARE_EXPORT(__movstr_i4_even); DECLARE_EXPORT(__movstr_i4_odd); DECLARE_EXPORT(__ashrdi3); diff --git a/arch/sh/kernel/signal.c b/arch/sh/kernel/signal.c index fe2f1b319..9095c5c79 100644 --- a/arch/sh/kernel/signal.c +++ b/arch/sh/kernel/signal.c @@ -29,8 +29,6 @@ #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); asmlinkage int do_signal(struct pt_regs *regs, sigset_t *oldset); int copy_siginfo_to_user(siginfo_t *to, siginfo_t *from) @@ -433,7 +431,7 @@ static void setup_frame(int sig, struct k_sigaction *ka, current->comm, current->pid, frame, regs->pc, regs->pr); #endif - flush_icache_range(regs->pr, regs->pr+4); + flush_cache_sigtramp(regs->pr); return; give_sigsegv: @@ -507,7 +505,7 @@ static void setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info, current->comm, current->pid, frame, regs->pc, regs->pr); #endif - flush_icache_range(regs->pr, regs->pr+4); + flush_cache_sigtramp(regs->pr); return; give_sigsegv: diff --git a/arch/sh/kernel/traps.c b/arch/sh/kernel/traps.c index 4aeaf8efa..f929fba2a 100644 --- a/arch/sh/kernel/traps.c +++ b/arch/sh/kernel/traps.c @@ -2,13 +2,16 @@ * * linux/arch/sh/traps.c * - * SuperH version: Copyright (C) 1999 Niibe Yutaka + * SuperH version: Copyright (C) 1999 Niibe Yutaka + * Copyright (C) 2000 Philipp Rumpf + * Copyright (C) 2000 David Howells */ /* * 'Traps.c' handles hardware traps and faults after we have saved some * state in 'entry.S'. */ +#include <linux/config.h> #include <linux/sched.h> #include <linux/kernel.h> #include <linux/string.h> @@ -35,7 +38,7 @@ asmlinkage void do_##name(unsigned long r4, unsigned long r5, \ { \ unsigned long error_code; \ \ - asm volatile("stc $r2_bank, %0": "=r" (error_code)); \ + asm volatile("stc r2_bank, %0": "=r" (error_code)); \ sti(); \ tsk->thread.error_code = error_code; \ tsk->thread.trap_no = trapnr; \ @@ -69,7 +72,16 @@ static inline void die_if_kernel(const char * str, struct pt_regs * regs, long e die(str, regs, err); } -static void die_if_no_fixup(const char * str, struct pt_regs * regs, long err) +static int handle_unaligned_notify_count = 10; + +/* + * try and fix up kernelspace address errors + * - userspace errors just cause EFAULT to be returned, resulting in SEGV + * - kernel/userspace interfaces cause a jump to an appropriate handler + * - other kernel errors are bad + * - return 0 if fixed-up, -EFAULT if non-fatal (to the kernel) fault + */ +static int die_if_no_fixup(const char * str, struct pt_regs * regs, long err) { if (!user_mode(regs)) { @@ -77,14 +89,407 @@ static void die_if_no_fixup(const char * str, struct pt_regs * regs, long err) fixup = search_exception_table(regs->pc); if (fixup) { regs->pc = fixup; - return; + return 0; } die(str, regs, err); } + return -EFAULT; +} + +/* + * handle an instruction that does an unaligned memory access by emulating the + * desired behaviour + * - note that PC _may not_ point to the faulting instruction + * (if that instruction is in a branch delay slot) + * - return 0 if emulation okay, -EFAULT on existential error + */ +static int handle_unaligned_ins(u16 instruction, struct pt_regs *regs) +{ + int ret, index, count; + unsigned long *rm, *rn; + unsigned char *src, *dst; + + index = (instruction>>8)&15; /* 0x0F00 */ + rn = ®s->regs[index]; + + index = (instruction>>4)&15; /* 0x00F0 */ + rm = ®s->regs[index]; + + count = 1<<(instruction&3); + + ret = -EFAULT; + switch (instruction>>12) { + case 0: /* mov.[bwl] to/from memory via r0+rn */ + if (instruction & 8) { + /* from memory */ + src = (unsigned char*) *rm; + src += regs->regs[0]; + dst = (unsigned char*) rn; + *(unsigned long*)dst = 0; + +#ifdef __LITTLE_ENDIAN__ + if (copy_from_user(dst, src, count)) + goto fetch_fault; + + if ((count == 2) && dst[1] & 0x80) { + dst[2] = 0xff; + dst[3] = 0xff; + } +#else + dst += 4-count; + + if (__copy_user(dst, src, count)) + goto fetch_fault; + + if ((count == 2) && dst[2] & 0x80) { + dst[0] = 0xff; + dst[1] = 0xff; + } +#endif + } else { + /* to memory */ + src = (unsigned char*) rm; +#if !defined(__LITTLE_ENDIAN__) + src += 4-count; +#endif + dst = (unsigned char*) *rn; + dst += regs->regs[0]; + + if (copy_to_user(dst, src, count)) + goto fetch_fault; + } + ret = 0; + break; + + case 1: /* mov.l Rm,@(disp,Rn) */ + src = (unsigned char*) rm; + dst = (unsigned char*) *rn; + dst += (instruction&0x000F)<<2; + + if (copy_to_user(dst,src,4)) + goto fetch_fault; + ret = 0; + break; + + case 2: /* mov.[bwl] to memory, possibly with pre-decrement */ + if (instruction & 4) + *rn -= count; + src = (unsigned char*) rm; + dst = (unsigned char*) *rn; +#if !defined(__LITTLE_ENDIAN__) + src += 4-count; +#endif + if (copy_to_user(dst, src, count)) + goto fetch_fault; + ret = 0; + break; + + case 5: /* mov.l @(disp,Rm),Rn */ + src = (unsigned char*) *rm; + src += (instruction&0x000F)<<2; + dst = (unsigned char*) rn; + *(unsigned long*)dst = 0; + + if (copy_from_user(dst,src,4)) + goto fetch_fault; + ret = 0; + break; + + case 6: /* mov.[bwl] from memory, possibly with post-increment */ + src = (unsigned char*) *rm; + if (instruction & 4) + *rm += count; + dst = (unsigned char*) rn; + *(unsigned long*)dst = 0; + +#ifdef __LITTLE_ENDIAN__ + if (copy_from_user(dst, src, count)) + goto fetch_fault; + + if ((count == 2) && dst[1] & 0x80) { + dst[2] = 0xff; + dst[3] = 0xff; + } +#else + dst += 4-count; + + if (copy_from_user(dst, src, count)) + goto fetch_fault; + + if ((count == 2) && dst[2] & 0x80) { + dst[0] = 0xff; + dst[1] = 0xff; + } +#endif + ret = 0; + break; + + case 8: + switch ((instruction&0xFF00)>>8) { + case 0x81: /* mov.w R0,@(disp,Rn) */ + src = (unsigned char*) ®s->regs[0]; +#if !defined(__LITTLE_ENDIAN__) + src += 2; +#endif + dst = (unsigned char*) *rm; /* called Rn in the spec */ + dst += (instruction&0x000F)<<1; + + if (copy_to_user(dst, src, 2)) + goto fetch_fault; + ret = 0; + break; + + case 0x85: /* mov.w @(disp,Rm),R0 */ + src = (unsigned char*) *rm; + src += (instruction&0x000F)<<1; + dst = (unsigned char*) ®s->regs[0]; + *(unsigned long*)dst = 0; + +#if !defined(__LITTLE_ENDIAN__) + dst += 2; +#endif + + if (copy_from_user(dst, src, 2)) + goto fetch_fault; + +#ifdef __LITTLE_ENDIAN__ + if (dst[1] & 0x80) { + dst[2] = 0xff; + dst[3] = 0xff; + } +#else + if (dst[2] & 0x80) { + dst[0] = 0xff; + dst[1] = 0xff; + } +#endif + ret = 0; + break; + } + break; + } + return ret; + + fetch_fault: + /* Argh. Address not only misaligned but also non-existent. + * Raise an EFAULT and see if it's trapped + */ + return die_if_no_fixup("Fault in unaligned fixup", regs, 0); +} + +/* + * emulate the instruction in the delay slot + * - fetches the instruction from PC+2 + */ +static inline int handle_unaligned_delayslot(struct pt_regs *regs) +{ + u16 instruction; + + if (copy_from_user(&instruction, (u16 *)(regs->pc+2), 2)) { + /* the instruction-fetch faulted */ + if (user_mode(regs)) + return -EFAULT; + + /* kernel */ + die("delay-slot-insn faulting in handle_unaligned_delayslot", regs, 0); + } + + return handle_unaligned_ins(instruction,regs); +} + +/* + * handle an instruction that does an unaligned memory access + * - have to be careful of branch delay-slot instructions that fault + * - if the branch would be taken PC points to the branch + * - if the branch would not be taken, PC points to delay-slot + * - return 0 if handled, -EFAULT if failed (may not return if in kernel) + */ +static int handle_unaligned_access(u16 instruction, struct pt_regs *regs) +{ + u_int rm; + int ret, index; + + index = (instruction>>8)&15; /* 0x0F00 */ + rm = regs->regs[index]; + + /* shout about the first ten userspace fixups */ + if (user_mode(regs) && handle_unaligned_notify_count>0) { + handle_unaligned_notify_count--; + + printk("Fixing up unaligned userspace access in \"%s\" pid=%d pc=0x%p ins=0x%04hx\n", + current->comm,current->pid,(u16*)regs->pc,instruction); + } + + ret = -EFAULT; + switch (instruction&0xF000) { + case 0x0000: + if (instruction==0x000B) { + /* rts */ + ret = handle_unaligned_delayslot(regs); + if (ret==0) + regs->pc = regs->pr; + } + else if ((instruction&0x00FF)==0x0023) { + /* braf @Rm */ + ret = handle_unaligned_delayslot(regs); + if (ret==0) + regs->pc += rm + 4; + } + else if ((instruction&0x00FF)==0x0003) { + /* bsrf @Rm */ + ret = handle_unaligned_delayslot(regs); + if (ret==0) { + regs->pr = regs->pc + 4; + regs->pc += rm + 4; + } + } + else { + /* mov.[bwl] to/from memory via r0+rn */ + goto simple; + } + break; + + case 0x1000: /* mov.l Rm,@(disp,Rn) */ + goto simple; + + case 0x2000: /* mov.[bwl] to memory, possibly with pre-decrement */ + goto simple; + + case 0x4000: + if ((instruction&0x00FF)==0x002B) { + /* jmp @Rm */ + ret = handle_unaligned_delayslot(regs); + if (ret==0) + regs->pc = rm; + } + else if ((instruction&0x00FF)==0x000B) { + /* jsr @Rm */ + ret = handle_unaligned_delayslot(regs); + if (ret==0) { + regs->pr = regs->pc + 4; + regs->pc = rm; + } + } + else { + /* mov.[bwl] to/from memory via r0+rn */ + goto simple; + } + break; + + case 0x5000: /* mov.l @(disp,Rm),Rn */ + goto simple; + + case 0x6000: /* mov.[bwl] from memory, possibly with post-increment */ + goto simple; + + case 0x8000: /* bf lab, bf/s lab, bt lab, bt/s lab */ + switch (instruction&0x0F00) { + case 0x0100: /* mov.w R0,@(disp,Rm) */ + goto simple; + case 0x0500: /* mov.w @(disp,Rm),R0 */ + goto simple; + case 0x0B00: /* bf lab - no delayslot*/ + break; + case 0x0F00: /* bf/s lab */ + ret = handle_unaligned_delayslot(regs); + if (ret==0) + regs->pc += (instruction&0x00FF)*2 + 4; + break; + case 0x0900: /* bt lab - no delayslot */ + break; + case 0x0D00: /* bt/s lab */ + ret = handle_unaligned_delayslot(regs); + if (ret==0) + regs->pc += (instruction&0x00FF)*2 + 4; + break; + } + break; + + case 0xA000: /* bra label */ + ret = handle_unaligned_delayslot(regs); + if (ret==0) + regs->pc += (instruction&0x0FFF)*2 + 4; + break; + + case 0xB000: /* bsr label */ + ret = handle_unaligned_delayslot(regs); + if (ret==0) { + regs->pr = regs->pc + 4; + regs->pc += (instruction&0x0FFF)*2 + 4; + } + break; + } + return ret; + + /* handle non-delay-slot instruction */ + simple: + ret = handle_unaligned_ins(instruction,regs); + if (ret==0) + regs->pc += 2; + return ret; +} + +/* + * Handle various address error exceptions + */ +asmlinkage void do_address_error(struct pt_regs *regs, + unsigned long writeaccess, + unsigned long address) +{ + unsigned long error_code; + mm_segment_t oldfs; + u16 instruction; + int tmp; + + asm volatile("stc r2_bank,%0": "=r" (error_code)); + + oldfs = get_fs(); + + if (user_mode(regs)) { + sti(); + current->thread.error_code = error_code; + current->thread.trap_no = (writeaccess) ? 8 : 7; + + /* bad PC is not something we can fix */ + if (regs->pc & 1) + goto uspace_segv; + + set_fs(USER_DS); + if (copy_from_user(&instruction, (u16 *)(regs->pc), 2)) { + /* Argh. Fault on the instruction itself. + This should never happen non-SMP + */ + set_fs(oldfs); + goto uspace_segv; + } + + tmp = handle_unaligned_access(instruction, regs); + set_fs(oldfs); + + if (tmp==0) + return; /* sorted */ + + uspace_segv: + printk(KERN_NOTICE "Killing process \"%s\" due to unaligned access\n", current->comm); + force_sig(SIGSEGV, current); + } else { + if (regs->pc & 1) + die("unaligned program counter", regs, error_code); + + set_fs(KERNEL_DS); + if (copy_from_user(&instruction, (u16 *)(regs->pc), 2)) { + /* Argh. Fault on the instruction itself. + This should never happen non-SMP + */ + set_fs(oldfs); + die("insn faulting in do_address_error", regs, 0); + } + + handle_unaligned_access(instruction, regs); + set_fs(oldfs); + } } -DO_ERROR( 7, SIGSEGV, "address error (load)", address_error_load, current) -DO_ERROR( 8, SIGSEGV, "address error (store)", address_error_store, current) DO_ERROR(12, SIGILL, "reserved instruction", reserved_inst, current) DO_ERROR(13, SIGILL, "illegal slot instruction", illegal_slot_inst, current) @@ -93,25 +498,42 @@ asmlinkage void do_exception_error(unsigned long r4, unsigned long r5, struct pt_regs regs) { long ex; - asm volatile("stc $r2_bank, %0" : "=r" (ex)); + asm volatile("stc r2_bank, %0" : "=r" (ex)); die_if_kernel("exception", ®s, ex); } +#if defined(CONFIG_DEBUG_KERNEL_WITH_GDB_STUB) || defined(CONFIG_SH_STANDARD_BIOS) +void *gdb_vbr_vector; +#endif + void __init trap_init(void) { extern void *vbr_base; extern void *exception_handling_table[14]; - exception_handling_table[7] = (void *)do_address_error_load; - exception_handling_table[8] = (void *)do_address_error_store; exception_handling_table[12] = (void *)do_reserved_inst; exception_handling_table[13] = (void *)do_illegal_slot_inst; +#if defined(CONFIG_DEBUG_KERNEL_WITH_GDB_STUB) || defined(CONFIG_SH_STANDARD_BIOS) + /* + * Read the old value of the VBR register to initialise + * the vector through which debug and BIOS traps are + * delegated by the Linux trap handler. + */ + { + register unsigned long vbr; + asm volatile("stc vbr, %0" : "=r" (vbr)); + gdb_vbr_vector = (void *)(vbr + 0x100); + printk("Setting GDB trap vector to 0x%08lx\n", + (unsigned long)gdb_vbr_vector); + } +#endif + /* NOTE: The VBR value should be at P1 (or P2, virtural "fixed" address space). It's definitely should not in physical address. */ - asm volatile("ldc %0, $vbr" + asm volatile("ldc %0, vbr" : /* no output */ : "r" (&vbr_base) : "memory"); @@ -123,8 +545,8 @@ void dump_stack(void) unsigned long *end; unsigned long *p; - asm("mov $r15, %0" : "=r" (start)); - asm("stc $r7_bank, %0" : "=r" (end)); + asm("mov r15, %0" : "=r" (start)); + asm("stc r7_bank, %0" : "=r" (end)); end += 8192/4; printk("%08lx:%08lx\n", (unsigned long)start, (unsigned long)end); diff --git a/arch/sh/lib/checksum.S b/arch/sh/lib/checksum.S index c7e96a394..9d8e0f476 100644 --- a/arch/sh/lib/checksum.S +++ b/arch/sh/lib/checksum.S @@ -49,99 +49,99 @@ ENTRY(csum_partial) * Fortunately, it is easy to convert 2-byte alignment to 4-byte * alignment for the unrolled loop. */ - mov $r5, $r1 - mov $r4, $r0 - tst #2, $r0 ! Check alignment. + mov r5, r1 + mov r4, r0 + tst #2, r0 ! Check alignment. bt 2f ! Jump if alignment is ok. ! - add #-2, $r5 ! Alignment uses up two bytes. - cmp/pz $r5 ! + add #-2, r5 ! Alignment uses up two bytes. + cmp/pz r5 ! bt/s 1f ! Jump if we had at least two bytes. clrt bra 6f - add #2, $r5 ! $r5 was < 2. Deal with it. + add #2, r5 ! r5 was < 2. Deal with it. 1: - mov.w @$r4+, $r0 - extu.w $r0, $r0 - addc $r0, $r6 + mov.w @r4+, r0 + extu.w r0, r0 + addc r0, r6 bf 2f - add #1, $r6 + add #1, r6 2: - mov #-5, $r0 - shld $r0, $r5 - tst $r5, $r5 + mov #-5, r0 + shld r0, r5 + tst r5, r5 bt/s 4f ! if it's =0, go to 4f clrt .align 2 3: - mov.l @$r4+, $r0 - mov.l @$r4+, $r2 - mov.l @$r4+, $r3 - addc $r0, $r6 - mov.l @$r4+, $r0 - addc $r2, $r6 - mov.l @$r4+, $r2 - addc $r3, $r6 - mov.l @$r4+, $r3 - addc $r0, $r6 - mov.l @$r4+, $r0 - addc $r2, $r6 - mov.l @$r4+, $r2 - addc $r3, $r6 - addc $r0, $r6 - addc $r2, $r6 - movt $r0 - dt $r5 + mov.l @r4+, r0 + mov.l @r4+, r2 + mov.l @r4+, r3 + addc r0, r6 + mov.l @r4+, r0 + addc r2, r6 + mov.l @r4+, r2 + addc r3, r6 + mov.l @r4+, r3 + addc r0, r6 + mov.l @r4+, r0 + addc r2, r6 + mov.l @r4+, r2 + addc r3, r6 + addc r0, r6 + addc r2, r6 + movt r0 + dt r5 bf/s 3b - cmp/eq #1, $r0 - ! here, we know $r5==0 - addc $r5, $r6 ! add carry to $r6 + cmp/eq #1, r0 + ! here, we know r5==0 + addc r5, r6 ! add carry to r6 4: - mov $r1, $r0 - and #0x1c, $r0 - tst $r0, $r0 + mov r1, r0 + and #0x1c, r0 + tst r0, r0 bt/s 6f - mov $r0, $r5 - shlr2 $r5 - mov #0, $r2 + mov r0, r5 + shlr2 r5 + mov #0, r2 5: - addc $r2, $r6 - mov.l @$r4+, $r2 - movt $r0 - dt $r5 + addc r2, r6 + mov.l @r4+, r2 + movt r0 + dt r5 bf/s 5b - cmp/eq #1, $r0 - addc $r2, $r6 - addc $r5, $r6 ! $r5==0 here, so it means add carry-bit + cmp/eq #1, r0 + addc r2, r6 + addc r5, r6 ! r5==0 here, so it means add carry-bit 6: - mov $r1, $r5 - mov #3, $r0 - and $r0, $r5 - tst $r5, $r5 + mov r1, r5 + mov #3, r0 + and r0, r5 + tst r5, r5 bt 9f ! if it's =0 go to 9f - mov #2, $r1 - cmp/hs $r1, $r5 + mov #2, r1 + cmp/hs r1, r5 bf 7f - mov.w @r4+, $r0 - extu.w $r0, $r0 - cmp/eq $r1, $r5 + mov.w @r4+, r0 + extu.w r0, r0 + cmp/eq r1, r5 bt/s 8f clrt - shll16 $r0 - addc $r0, $r6 + shll16 r0 + addc r0, r6 7: - mov.b @$r4+, $r0 - extu.b $r0, $r0 + mov.b @r4+, r0 + extu.b r0, r0 #ifndef __LITTLE_ENDIAN__ - shll8 $r0 + shll8 r0 #endif 8: - addc $r0, $r6 - mov #0, $r0 - addc $r0, $r6 + addc r0, r6 + mov #0, r0 + addc r0, r6 9: rts - mov $r6, $r0 + mov r6, r0 /* unsigned int csum_partial_copy_generic (const char *src, char *dst, int len, @@ -159,14 +159,14 @@ unsigned int csum_partial_copy_generic (const char *src, char *dst, int len, * them all but there's no guarantee. */ -#define SRC(x,y) \ - 9999: x,y; \ +#define SRC(...) \ + 9999: __VA_ARGS__ ; \ .section __ex_table, "a"; \ .long 9999b, 6001f ; \ .previous -#define DST(x,y) \ - 9999: x,y; \ +#define DST(...) \ + 9999: __VA_ARGS__ ; \ .section __ex_table, "a"; \ .long 9999b, 6002f ; \ .previous @@ -202,7 +202,7 @@ ENTRY(csum_partial_copy_generic) bt/s 1f clrt bra 4f - add #2,r6 ! $r6 was < 2. Deal with it. + add #2,r6 ! r6 was < 2. Deal with it. 3: ! Handle different src and dest alinments. ! This is not common, so simple byte by byte copy will do. @@ -211,7 +211,8 @@ ENTRY(csum_partial_copy_generic) tst r6, r6 bt 4f clrt -SRC(5: mov.b @r4+,r0 ) +5: +SRC( mov.b @r4+,r0 ) DST( mov.b r0,@r5 ) add #1, r5 SRC( mov.b @r4+,r1 ) @@ -244,7 +245,8 @@ DST( mov.b r1,@r5 ) ! src and dest equally aligned, but to a two byte boundary. ! Handle first two bytes as a special case .align 5 -SRC(1: mov.w @r4+,r0 ) +1: +SRC( mov.w @r4+,r0 ) DST( mov.w r0,@r5 ) add #2,r5 extu.w r0,r0 @@ -258,7 +260,8 @@ DST( mov.w r0,@r5 ) tst r6,r6 bt/s 2f clrt -SRC(1: mov.l @r4+,r0 ) +1: +SRC( mov.l @r4+,r0 ) SRC( mov.l @r4+,r1 ) addc r0,r7 DST( mov.l r0,@r5 ) @@ -276,7 +279,7 @@ DST( mov.l r0,@r5 ) DST( mov.l r1,@r5 ) add #4,r5 -SRC( mov.l @r4+,r0 ) +SRC( mov.l @r4+,r0 ) SRC( mov.l @r4+,r1 ) addc r0,r7 DST( mov.l r0,@r5 ) @@ -307,7 +310,8 @@ DST( mov.l r1,@r5 ) bf/s 4f clrt shlr2 r6 -SRC(3: mov.l @r4+,r0 ) +3: +SRC( mov.l @r4+,r0 ) addc r0,r7 DST( mov.l r0,@r5 ) add #4,r5 @@ -334,7 +338,8 @@ DST( mov.w r0,@r5 ) clrt shll16 r0 addc r0,r7 -SRC(5: mov.b @r4+,r0 ) +5: +SRC( mov.b @r4+,r0 ) DST( mov.b r0,@r5 ) extu.b r0,r0 #ifndef __LITTLE_ENDIAN__ diff --git a/arch/sh/lib/delay.c b/arch/sh/lib/delay.c index 790fd1808..e1fcc970f 100644 --- a/arch/sh/lib/delay.c +++ b/arch/sh/lib/delay.c @@ -21,7 +21,11 @@ void __delay(unsigned long loops) inline void __const_udelay(unsigned long xloops) { - xloops *= current_cpu_data.loops_per_jiffy; + __asm__("dmulu.l %0, %2\n\t" + "sts mach, %0" + : "=r" (xloops) + : "0" (xloops), "r" (current_cpu_data.loops_per_jiffy) + : "macl", "mach"); __delay(xloops * HZ); } diff --git a/arch/sh/mm/cache.c b/arch/sh/mm/cache.c index ab63535e4..2ea37e723 100644 --- a/arch/sh/mm/cache.c +++ b/arch/sh/mm/cache.c @@ -244,6 +244,22 @@ void flush_icache_range(unsigned long start, unsigned long end) } /* + * Write back the D-cache and purge the I-cache for signal trampoline. + */ +void flush_cache_sigtramp(unsigned long addr) +{ + unsigned long v, index; + + v = addr & ~(L1_CACHE_BYTES-1); + asm volatile("ocbwb %0" + : /* no output */ + : "m" (__m(v))); + + index = CACHE_IC_ADDRESS_ARRAY| (v&CACHE_IC_ENTRY_MASK); + ctrl_outl(0, index); /* Clear out Valid-bit */ +} + +/* * Invalidate the I-cache of the page (don't need to write back D-cache). * * Called from kernel/ptrace.c, mm/memory.c after flush_page_to_ram is called. diff --git a/arch/sparc/config.in b/arch/sparc/config.in index dbe1d1fb2..cb27cf77b 100644 --- a/arch/sparc/config.in +++ b/arch/sparc/config.in @@ -1,4 +1,4 @@ -# $Id: config.in,v 1.106 2000/11/17 04:27:52 davem Exp $ +# $Id: config.in,v 1.107 2001/01/06 00:46:44 davem Exp $ # For a description of the syntax of this configuration file, # see Documentation/kbuild/config-language.txt. # @@ -29,6 +29,9 @@ define_bool CONFIG_VT_CONSOLE y bool 'Symmetric multi-processing support (does not work on sun4/sun4c)' CONFIG_SMP +# Identify this as a Sparc32 build +define_bool CONFIG_SPARC32 y + # Global things across all Sun machines. define_bool CONFIG_ISA n define_bool CONFIG_EISA n diff --git a/arch/sparc/defconfig b/arch/sparc/defconfig index 19c9feba0..20c0782e7 100644 --- a/arch/sparc/defconfig +++ b/arch/sparc/defconfig @@ -22,7 +22,10 @@ CONFIG_KMOD=y CONFIG_VT=y CONFIG_VT_CONSOLE=y # CONFIG_SMP is not set +CONFIG_SPARC32=y # CONFIG_ISA is not set +# CONFIG_EISA is not set +# CONFIG_MCA is not set # CONFIG_PCMCIA is not set CONFIG_SBUS=y CONFIG_SBUSCHAR=y @@ -64,6 +67,7 @@ CONFIG_PROM_CONSOLE=y # CONFIG_FB=y CONFIG_DUMMY_CONSOLE=y +# CONFIG_FB_CYBER2000 is not set CONFIG_FB_SBUS=y CONFIG_FB_CGSIX=y CONFIG_FB_BWTWO=y @@ -111,10 +115,18 @@ CONFIG_SUN_AURORA=m CONFIG_BLK_DEV_FD=y CONFIG_BLK_DEV_LOOP=m CONFIG_BLK_DEV_NBD=m -CONFIG_BLK_DEV_MD=m -CONFIG_MD_LINEAR=m -CONFIG_MD_RAID0=m -CONFIG_MD_RAID1=m + +# +# Multi-device support (RAID and LVM) +# +# CONFIG_MD is not set +# CONFIG_BLK_DEV_MD is not set +# CONFIG_MD_LINEAR is not set +# CONFIG_MD_RAID0 is not set +# CONFIG_MD_RAID1 is not set +# CONFIG_MD_RAID5 is not set +# CONFIG_BLK_DEV_LVM is not set +# CONFIG_LVM_PROC_FS is not set CONFIG_BLK_DEV_RAM=y CONFIG_BLK_DEV_RAM_SIZE=4096 CONFIG_BLK_DEV_INITRD=y @@ -132,10 +144,8 @@ CONFIG_INET=y # CONFIG_IP_MULTICAST is not set # CONFIG_IP_ADVANCED_ROUTER is not set # CONFIG_IP_PNP is not set -# CONFIG_IP_ROUTER is not set # CONFIG_NET_IPIP is not set # CONFIG_NET_IPGRE is not set -# CONFIG_IP_ALIAS is not set # CONFIG_INET_ECN is not set # CONFIG_SYN_COOKIES is not set CONFIG_IPV6=m @@ -148,7 +158,6 @@ CONFIG_IPV6=m # CONFIG_IPX=m # CONFIG_IPX_INTERN is not set -# CONFIG_SPX is not set CONFIG_ATALK=m CONFIG_DECNET=m CONFIG_DECNET_SIOCGIFCONF=y @@ -157,6 +166,7 @@ CONFIG_DECNET_SIOCGIFCONF=y # CONFIG_X25 is not set # CONFIG_LAPB is not set # CONFIG_LLC is not set +# CONFIG_NET_DIVERT is not set # CONFIG_ECONET is not set # CONFIG_WAN_ROUTER is not set # CONFIG_NET_FASTROUTE is not set @@ -228,6 +238,7 @@ CONFIG_SCSI_FCAL=m CONFIG_NETDEVICES=y CONFIG_DUMMY=m CONFIG_BONDING=m +# CONFIG_TUN is not set CONFIG_PPP=m CONFIG_PPP_ASYNC=m CONFIG_PPP_SYNC_TTY=m @@ -302,6 +313,7 @@ CONFIG_NFSD=m CONFIG_SUNRPC=y CONFIG_LOCKD=y CONFIG_SMB_FS=m +# CONFIG_SMB_NLS_DEFAULT is not set # CONFIG_NCP_FS is not set # CONFIG_NCPFS_PACKET_SIGNING is not set # CONFIG_NCPFS_IOCTL_LOCKING is not set @@ -309,8 +321,6 @@ CONFIG_SMB_FS=m # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -320,6 +330,7 @@ CONFIG_SMB_FS=m # CONFIG_PARTITION_ADVANCED is not set CONFIG_MSDOS_PARTITION=y CONFIG_SUN_PARTITION=y +CONFIG_SMB_NLS=y CONFIG_NLS=y # diff --git a/arch/sparc/kernel/entry.S b/arch/sparc/kernel/entry.S index 70bbda891..47fcf6226 100644 --- a/arch/sparc/kernel/entry.S +++ b/arch/sparc/kernel/entry.S @@ -1,4 +1,4 @@ -/* $Id: entry.S,v 1.168 2001/01/01 01:46:15 davem Exp $ +/* $Id: entry.S,v 1.169 2001/01/25 21:47:20 davem Exp $ * arch/sparc/kernel/entry.S: Sparc trap low-level entry points. * * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu) @@ -999,6 +999,7 @@ C_LABEL(invalid_segment_patch1): ld [%l4 + 0x18], %l3 inc %l3 ! sun4c_kernel_ring.num_entries++ + st %l3, [%l4 + 0x18] b 4f ld [%l6 + 0x08], %l5 diff --git a/arch/sparc/kernel/pcic.c b/arch/sparc/kernel/pcic.c index c9ba85b27..9bc82a518 100644 --- a/arch/sparc/kernel/pcic.c +++ b/arch/sparc/kernel/pcic.c @@ -1,4 +1,4 @@ -/* $Id: pcic.c,v 1.20 2000/12/05 00:56:36 anton Exp $ +/* $Id: pcic.c,v 1.21 2001/01/18 00:23:00 davem Exp $ * pcic.c: Sparc/PCI controller support * * Copyright (C) 1998 V. Roganov and G. Raiko @@ -898,93 +898,6 @@ void pcic_nmi(unsigned int pend, struct pt_regs *regs) regs->npc += 4; } -/* - * XXX Gleb wrote me that he needs this for X server (only). - * Since we successfuly use XF86_FBDev, we do not need these anymore. - * - * Following code added to handle extra PCI-related system calls - */ -asmlinkage int sys_pciconfig_read(unsigned long bus, - unsigned long dfn, - unsigned long off, - unsigned long len, - unsigned char *buf) -{ - unsigned char ubyte; - unsigned short ushort; - unsigned int uint; - int err = 0; - - if(!suser()) - return -EPERM; - - switch(len) { - case 1: - pcibios_read_config_byte(bus, dfn, off, &ubyte); - put_user(ubyte, (unsigned char *)buf); - break; - case 2: - pcibios_read_config_word(bus, dfn, off, &ushort); - put_user(ushort, (unsigned short *)buf); - break; - case 4: - pcibios_read_config_dword(bus, dfn, off, &uint); - put_user(uint, (unsigned int *)buf); - break; - - default: - err = -EINVAL; - break; - }; - - return err; -} - -asmlinkage int sys_pciconfig_write(unsigned long bus, - unsigned long dfn, - unsigned long off, - unsigned long len, - unsigned char *buf) -{ - unsigned char ubyte; - unsigned short ushort; - unsigned int uint; - int err = 0; - - if(!suser()) - return -EPERM; - - switch(len) { - case 1: - err = get_user(ubyte, (unsigned char *)buf); - if(err) - break; - pcibios_write_config_byte(bus, dfn, off, ubyte); - break; - - case 2: - err = get_user(ushort, (unsigned short *)buf); - if(err) - break; - pcibios_write_config_byte(bus, dfn, off, ushort); - break; - - case 4: - err = get_user(uint, (unsigned int *)buf); - if(err) - break; - pcibios_write_config_byte(bus, dfn, off, uint); - break; - - default: - err = -EINVAL; - break; - - }; - - return err; -} - static inline unsigned long get_irqmask(int irq_nr) { return 1 << irq_nr; diff --git a/arch/sparc/kernel/semaphore.c b/arch/sparc/kernel/semaphore.c index c949717b4..259a0a71b 100644 --- a/arch/sparc/kernel/semaphore.c +++ b/arch/sparc/kernel/semaphore.c @@ -1,4 +1,4 @@ -/* $Id: semaphore.c,v 1.5 2000/12/29 10:35:05 anton Exp $ */ +/* $Id: semaphore.c,v 1.6 2001/01/05 04:40:07 davem Exp $ */ /* sparc32 semaphore implementation, based on i386 version */ diff --git a/arch/sparc/kernel/signal.c b/arch/sparc/kernel/signal.c index 6ac8c4b00..957ddc17c 100644 --- a/arch/sparc/kernel/signal.c +++ b/arch/sparc/kernel/signal.c @@ -1,4 +1,4 @@ -/* $Id: signal.c,v 1.107 2000/09/05 21:44:54 davem Exp $ +/* $Id: signal.c,v 1.108 2001/01/24 21:05:12 davem Exp $ * linux/arch/sparc/kernel/signal.c * * Copyright (C) 1991, 1992 Linus Torvalds @@ -28,9 +28,6 @@ #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); - extern void fpsave(unsigned long *fpregs, unsigned long *fsr, void *fpqueue, unsigned long *fpqdepth); extern void fpload(unsigned long *fpregs, unsigned long *fsr); diff --git a/arch/sparc/kernel/sparc_ksyms.c b/arch/sparc/kernel/sparc_ksyms.c index d9883654e..235d4b96e 100644 --- a/arch/sparc/kernel/sparc_ksyms.c +++ b/arch/sparc/kernel/sparc_ksyms.c @@ -1,4 +1,4 @@ -/* $Id: sparc_ksyms.c,v 1.105 2000/12/11 05:24:25 anton Exp $ +/* $Id: sparc_ksyms.c,v 1.106 2001/01/11 15:07:09 davem Exp $ * arch/sparc/kernel/ksyms.c: Sparc specific ksyms support. * * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu) @@ -154,6 +154,7 @@ EXPORT_SYMBOL(__cpu_logical_map); #endif EXPORT_SYMBOL(udelay); +EXPORT_SYMBOL(mostek_lock); EXPORT_SYMBOL(mstk48t02_regs); #if CONFIG_SUN_AUXIO EXPORT_SYMBOL(auxio_register); diff --git a/arch/sparc/kernel/sys_sunos.c b/arch/sparc/kernel/sys_sunos.c index ac252f55a..24ae3829d 100644 --- a/arch/sparc/kernel/sys_sunos.c +++ b/arch/sparc/kernel/sys_sunos.c @@ -1,4 +1,4 @@ -/* $Id: sys_sunos.c,v 1.130 2000/08/12 13:25:41 davem Exp $ +/* $Id: sys_sunos.c,v 1.131 2001/01/24 21:05:12 davem Exp $ * sys_sunos.c: SunOS specific syscall compatibility support. * * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu) @@ -834,7 +834,6 @@ asmlinkage int sunos_setpgrp(pid_t pid, pid_t pgid) } /* So stupid... */ -extern asmlinkage int sys_wait4(pid_t, unsigned int *, int, struct rusage *); asmlinkage int sunos_wait4(pid_t pid, unsigned int *stat_addr, int options, struct rusage *ru) { int ret; diff --git a/arch/sparc/kernel/time.c b/arch/sparc/kernel/time.c index 980981592..37113ef38 100644 --- a/arch/sparc/kernel/time.c +++ b/arch/sparc/kernel/time.c @@ -1,4 +1,4 @@ -/* $Id: time.c,v 1.57 2000/09/16 07:33:45 davem Exp $ +/* $Id: time.c,v 1.58 2001/01/11 15:07:09 davem Exp $ * linux/arch/sparc/kernel/time.c * * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu) @@ -44,6 +44,7 @@ extern rwlock_t xtime_lock; enum sparc_clock_type sp_clock_typ; +spinlock_t mostek_lock = SPIN_LOCK_UNLOCKED; unsigned long mstk48t02_regs = 0UL; static struct mostek48t08 *mstk48t08_regs = 0; static int set_rtc_mmss(unsigned long); @@ -158,12 +159,16 @@ static void __init kick_start_clock(void) prom_printf("CLOCK: Clock was stopped. Kick start "); + spin_lock_irq(&mostek_lock); + /* Turn on the kick start bit to start the oscillator. */ regs->creg |= MSTK_CREG_WRITE; regs->sec &= ~MSTK_STOP; regs->hour |= MSTK_KICK_START; regs->creg &= ~MSTK_CREG_WRITE; + spin_unlock_irq(&mostek_lock); + /* Delay to allow the clock oscillator to start. */ sec = MSTK_REG_SEC(regs); for (i = 0; i < 3; i++) { @@ -175,6 +180,8 @@ static void __init kick_start_clock(void) } prom_printf("\n"); + spin_lock_irq(&mostek_lock); + /* Turn off kick start and set a "valid" time and date. */ regs->creg |= MSTK_CREG_WRITE; regs->hour &= ~MSTK_KICK_START; @@ -187,12 +194,17 @@ static void __init kick_start_clock(void) MSTK_SET_REG_YEAR(regs,1996 - MSTK_YEAR_ZERO); regs->creg &= ~MSTK_CREG_WRITE; + spin_unlock_irq(&mostek_lock); + /* Ensure the kick start bit is off. If it isn't, turn it off. */ while (regs->hour & MSTK_KICK_START) { prom_printf("CLOCK: Kick start still on!\n"); + + spin_lock_irq(&mostek_lock); regs->creg |= MSTK_CREG_WRITE; regs->hour &= ~MSTK_KICK_START; regs->creg &= ~MSTK_CREG_WRITE; + spin_unlock_irq(&mostek_lock); } prom_printf("CLOCK: Kick start procedure successful.\n"); @@ -204,10 +216,12 @@ static __inline__ int has_low_battery(void) struct mostek48t02 *regs = (struct mostek48t02 *)mstk48t02_regs; unsigned char data1, data2; + spin_lock_irq(&mostek_lock); data1 = regs->eeprom[0]; /* Read some data. */ regs->eeprom[0] = ~data1; /* Write back the complement. */ data2 = regs->eeprom[0]; /* Read back the complement. */ regs->eeprom[0] = data1; /* Restore the original value. */ + spin_unlock_irq(&mostek_lock); return (data1 == data2); /* Was the write blocked? */ } @@ -376,6 +390,7 @@ void __init sbus_time_init(void) prom_printf("Something wrong, clock regs not mapped yet.\n"); prom_halt(); } + spin_lock_irq(&mostek_lock); mregs->creg |= MSTK_CREG_READ; sec = MSTK_REG_SEC(mregs); min = MSTK_REG_MIN(mregs); @@ -386,6 +401,7 @@ void __init sbus_time_init(void) xtime.tv_sec = mktime(year, mon, day, hour, min, sec); xtime.tv_usec = 0; mregs->creg &= ~MSTK_CREG_READ; + spin_unlock_irq(&mostek_lock); #ifdef CONFIG_SUN4 } else if(idprom->id_machtype == (SM_SUN4 | SM_4_260) ) { /* initialise the intersil on sun4 */ @@ -520,6 +536,7 @@ static int set_rtc_mmss(unsigned long nowtime) { int real_seconds, real_minutes, mostek_minutes; struct mostek48t02 *regs = (struct mostek48t02 *)mstk48t02_regs; + unsigned long flags; #ifdef CONFIG_SUN4 struct intersil *iregs = intersil_clock; int temp; @@ -557,6 +574,8 @@ static int set_rtc_mmss(unsigned long nowtime) } #endif } + + spin_lock_irqsave(&mostek_lock, flags); /* Read the current RTC minutes. */ regs->creg |= MSTK_CREG_READ; mostek_minutes = MSTK_REG_MIN(regs); @@ -579,8 +598,10 @@ static int set_rtc_mmss(unsigned long nowtime) MSTK_SET_REG_SEC(regs,real_seconds); MSTK_SET_REG_MIN(regs,real_minutes); regs->creg &= ~MSTK_CREG_WRITE; - } else + spin_unlock_irqrestore(&mostek_lock, flags); + return 0; + } else { + spin_unlock_irqrestore(&mostek_lock, flags); return -1; - - return 0; + } } diff --git a/arch/sparc64/config.in b/arch/sparc64/config.in index 19b05e28f..5c208915b 100644 --- a/arch/sparc64/config.in +++ b/arch/sparc64/config.in @@ -1,4 +1,4 @@ -# $Id: config.in,v 1.125 2000/10/10 01:05:53 davem Exp $ +# $Id: config.in,v 1.130 2001/01/18 04:47:44 davem Exp $ # For a description of the syntax of this configuration file, # see the Configure script. # @@ -26,6 +26,9 @@ define_bool CONFIG_VT_CONSOLE y bool 'Symmetric multi-processing support' CONFIG_SMP +# Identify this as a Sparc64 build +define_bool CONFIG_SPARC64 y + # Global things across all Sun machines. define_bool CONFIG_HAVE_DEC_LOCK y define_bool CONFIG_ISA n @@ -70,6 +73,7 @@ dep_tristate ' Parallel printer support' CONFIG_PRINTER $CONFIG_PARPORT if [ "$CONFIG_PCI" = "y" ]; then tristate 'SUNW, envctrl support' CONFIG_ENVCTRL tristate '7-Segment Display support' CONFIG_DISPLAY7SEG + tristate 'CP1XXX Hardware Watchdog support' CONFIG_WATCHDOG_CP1XXX fi endmenu diff --git a/arch/sparc64/defconfig b/arch/sparc64/defconfig index 7c2f852b3..d80f93cf6 100644 --- a/arch/sparc64/defconfig +++ b/arch/sparc64/defconfig @@ -20,8 +20,11 @@ CONFIG_KMOD=y CONFIG_VT=y CONFIG_VT_CONSOLE=y # CONFIG_SMP is not set +CONFIG_SPARC64=y CONFIG_HAVE_DEC_LOCK=y # CONFIG_ISA is not set +# CONFIG_EISA is not set +# CONFIG_MCA is not set # CONFIG_PCMCIA is not set CONFIG_SBUS=y CONFIG_SBUSCHAR=y @@ -66,6 +69,7 @@ CONFIG_PARPORT_1284=y CONFIG_PRINTER=m CONFIG_ENVCTRL=m CONFIG_DISPLAY7SEG=m +CONFIG_WATCHDOG_CP1XXX=m # # Console drivers @@ -122,9 +126,7 @@ CONFIG_SUN_AURORA=m # Linux/SPARC audio subsystem (EXPERIMENTAL) # CONFIG_SPARCAUDIO=y -# CONFIG_SPARCAUDIO_AMD7930 is not set CONFIG_SPARCAUDIO_CS4231=y -# CONFIG_SPARCAUDIO_DBRI is not set # CONFIG_SPARCAUDIO_DUMMY is not set # @@ -180,7 +182,6 @@ CONFIG_IPV6=m # CONFIG_IPX=m # CONFIG_IPX_INTERN is not set -# CONFIG_SPX is not set CONFIG_ATALK=m CONFIG_DECNET=m CONFIG_DECNET_SIOCGIFCONF=y @@ -415,6 +416,8 @@ CONFIG_DRM_FFB=m # CONFIG_QUOTA is not set CONFIG_AUTOFS_FS=m CONFIG_AUTOFS4_FS=m +# CONFIG_REISERFS_FS is not set +# CONFIG_REISERFS_CHECK is not set # CONFIG_ADFS_FS is not set # CONFIG_ADFS_FS_RW is not set CONFIG_AFFS_FS=m @@ -470,8 +473,6 @@ CONFIG_LOCKD_V4=y # CONFIG_NCPFS_NFS_NS is not set # CONFIG_NCPFS_OS2_NS is not set # CONFIG_NCPFS_SMALLDOS is not set -# CONFIG_NCPFS_MOUNT_SUBDIR is not set -# CONFIG_NCPFS_NDS_DOMAINS is not set # CONFIG_NCPFS_NLS is not set # CONFIG_NCPFS_EXTRAS is not set @@ -481,6 +482,7 @@ CONFIG_LOCKD_V4=y # CONFIG_PARTITION_ADVANCED is not set CONFIG_MSDOS_PARTITION=y CONFIG_SUN_PARTITION=y +# CONFIG_SMB_NLS is not set CONFIG_NLS=y # diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c index 81e11af70..67c1ec6d0 100644 --- a/arch/sparc64/kernel/ioctl32.c +++ b/arch/sparc64/kernel/ioctl32.c @@ -1,4 +1,4 @@ -/* $Id: ioctl32.c,v 1.104 2001/01/03 09:28:19 anton Exp $ +/* $Id: ioctl32.c,v 1.105 2001/01/18 04:47:44 davem Exp $ * ioctl32.c: Conversion between 32bit and 64bit native ioctls. * * Copyright (C) 1997-2000 Jakub Jelinek (jakub@redhat.com) @@ -73,6 +73,7 @@ #include <asm/audioio.h> #include <linux/ethtool.h> #include <asm/display7seg.h> +#include <asm/watchdog.h> #include <asm/module.h> #include <linux/soundcard.h> @@ -3600,6 +3601,16 @@ COMPATIBLE_IOCTL(DRM_IOCTL_FINISH) /* elevator */ COMPATIBLE_IOCTL(BLKELVGET) COMPATIBLE_IOCTL(BLKELVSET) +/* Big W */ +/* WIOC_GETSUPPORT not yet implemented -E */ +COMPATIBLE_IOCTL(WDIOC_GETSTATUS) +COMPATIBLE_IOCTL(WDIOC_GETBOOTSTATUS) +COMPATIBLE_IOCTL(WDIOC_GETTEMP) +COMPATIBLE_IOCTL(WDIOC_SETOPTIONS) +COMPATIBLE_IOCTL(WDIOC_KEEPALIVE) +COMPATIBLE_IOCTL(WIOCSTART) +COMPATIBLE_IOCTL(WIOCSTOP) +COMPATIBLE_IOCTL(WIOCGSTAT) /* And these ioctls need translation */ HANDLE_IOCTL(SIOCGIFNAME, dev_ifname32) HANDLE_IOCTL(SIOCGIFCONF, dev_ifconf) diff --git a/arch/sparc64/kernel/pci.c b/arch/sparc64/kernel/pci.c index 2b697dd87..009900165 100644 --- a/arch/sparc64/kernel/pci.c +++ b/arch/sparc64/kernel/pci.c @@ -1,4 +1,4 @@ -/* $Id: pci.c,v 1.20 2000/12/14 22:57:25 davem Exp $ +/* $Id: pci.c,v 1.21 2001/01/10 18:22:59 davem Exp $ * pci.c: UltraSparc PCI controller support. * * Copyright (C) 1997, 1998, 1999 David S. Miller (davem@redhat.com) @@ -78,6 +78,9 @@ volatile int pci_poke_faulted; /* Probe for all PCI controllers in the system. */ extern void sabre_init(int); extern void psycho_init(int); +#if 0 +extern void schizo_init(int); +#endif static struct { char *model_name; @@ -87,6 +90,10 @@ static struct { { "pci108e,a000", sabre_init }, { "SUNW,psycho", psycho_init }, { "pci108e,8000", psycho_init } +#if 0 + { "SUNW,schizo", schizo_init }, + { "pci108e,8001", schizo_init } +#endif }; #define PCI_NUM_CONTROLLER_TYPES (sizeof(pci_controller_table) / \ sizeof(pci_controller_table[0])) diff --git a/arch/sparc64/kernel/pci_iommu.c b/arch/sparc64/kernel/pci_iommu.c index d7267880a..08d5b8ee3 100644 --- a/arch/sparc64/kernel/pci_iommu.c +++ b/arch/sparc64/kernel/pci_iommu.c @@ -1,4 +1,4 @@ -/* $Id: pci_iommu.c,v 1.11 2000/03/10 02:42:15 davem Exp $ +/* $Id: pci_iommu.c,v 1.12 2001/01/11 16:26:45 davem Exp $ * pci_iommu.c: UltraSparc PCI controller IOM/STC support. * * Copyright (C) 1999 David S. Miller (davem@redhat.com) @@ -187,7 +187,7 @@ void *pci_alloc_consistent(struct pci_dev *pdev, size_t size, dma_addr_t *dma_ad memset((char *)first_page, 0, PAGE_SIZE << order); pcp = pdev->sysdata; - iommu = &pcp->pbm->parent->iommu; + iommu = pcp->pbm->iommu; spin_lock_irqsave(&iommu->lock, flags); iopte = alloc_consistent_cluster(iommu, size >> PAGE_SHIFT); @@ -241,7 +241,7 @@ void pci_free_consistent(struct pci_dev *pdev, size_t size, void *cpu, dma_addr_ npages = PAGE_ALIGN(size) >> PAGE_SHIFT; pcp = pdev->sysdata; - iommu = &pcp->pbm->parent->iommu; + iommu = pcp->pbm->iommu; iopte = iommu->page_table + ((dvma - iommu->page_table_map_base) >> PAGE_SHIFT); @@ -308,7 +308,7 @@ dma_addr_t pci_map_single(struct pci_dev *pdev, void *ptr, size_t sz, int direct unsigned long iopte_protection; pcp = pdev->sysdata; - iommu = &pcp->pbm->parent->iommu; + iommu = pcp->pbm->iommu; strbuf = &pcp->pbm->stc; if (direction == PCI_DMA_NONE) @@ -356,7 +356,7 @@ void pci_unmap_single(struct pci_dev *pdev, dma_addr_t bus_addr, size_t sz, int BUG(); pcp = pdev->sysdata; - iommu = &pcp->pbm->parent->iommu; + iommu = pcp->pbm->iommu; strbuf = &pcp->pbm->stc; npages = PAGE_ALIGN(bus_addr + sz) - (bus_addr & PAGE_MASK); @@ -504,7 +504,7 @@ int pci_map_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, int } pcp = pdev->sysdata; - iommu = &pcp->pbm->parent->iommu; + iommu = pcp->pbm->iommu; strbuf = &pcp->pbm->stc; if (direction == PCI_DMA_NONE) @@ -568,7 +568,7 @@ void pci_unmap_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, BUG(); pcp = pdev->sysdata; - iommu = &pcp->pbm->parent->iommu; + iommu = pcp->pbm->iommu; strbuf = &pcp->pbm->stc; bus_addr = sglist->dvma_address & PAGE_MASK; @@ -639,7 +639,7 @@ void pci_dma_sync_single(struct pci_dev *pdev, dma_addr_t bus_addr, size_t sz, i unsigned long flags, ctx, npages; pcp = pdev->sysdata; - iommu = &pcp->pbm->parent->iommu; + iommu = pcp->pbm->iommu; strbuf = &pcp->pbm->stc; if (!strbuf->strbuf_enabled) @@ -700,7 +700,7 @@ void pci_dma_sync_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelem unsigned long flags, ctx; pcp = pdev->sysdata; - iommu = &pcp->pbm->parent->iommu; + iommu = pcp->pbm->iommu; strbuf = &pcp->pbm->stc; if (!strbuf->strbuf_enabled) @@ -762,7 +762,7 @@ int pci_dma_supported(struct pci_dev *pdev, dma_addr_t device_mask) if (pdev == NULL) { dma_addr_mask = 0xffffffff; } else { - struct pci_iommu *iommu = &pcp->pbm->parent->iommu; + struct pci_iommu *iommu = pcp->pbm->iommu; dma_addr_mask = iommu->dma_addr_mask; } diff --git a/arch/sparc64/kernel/pci_psycho.c b/arch/sparc64/kernel/pci_psycho.c index aff2de594..fbd7832cd 100644 --- a/arch/sparc64/kernel/pci_psycho.c +++ b/arch/sparc64/kernel/pci_psycho.c @@ -1,4 +1,4 @@ -/* $Id: pci_psycho.c,v 1.17 2000/09/21 06:25:14 anton Exp $ +/* $Id: pci_psycho.c,v 1.18 2001/01/11 16:26:45 davem Exp $ * pci_psycho.c: PSYCHO/U2P specific PCI controller support. * * Copyright (C) 1997, 1998, 1999 David S. Miller (davem@caipfs.rutgers.edu) @@ -619,20 +619,21 @@ static void psycho_check_iommu_error(struct pci_controller_info *p, unsigned long afar, enum psycho_error_type type) { + struct pci_iommu *iommu = p->pbm_A.iommu; unsigned long iommu_tag[16]; unsigned long iommu_data[16]; unsigned long flags; u64 control; int i; - spin_lock_irqsave(&p->iommu.lock, flags); - control = psycho_read(p->iommu.iommu_control); + spin_lock_irqsave(&iommu->lock, flags); + control = psycho_read(iommu->iommu_control); if (control & PSYCHO_IOMMU_CTRL_XLTEERR) { char *type_string; /* Clear the error encountered bit. */ control &= ~PSYCHO_IOMMU_CTRL_XLTEERR; - psycho_write(p->iommu.iommu_control, control); + psycho_write(iommu->iommu_control, control); switch((control & PSYCHO_IOMMU_CTRL_XLTESTAT) >> 25UL) { case 0: @@ -662,7 +663,7 @@ static void psycho_check_iommu_error(struct pci_controller_info *p, * get as much diagnostic information to the * console as we can. */ - psycho_write(p->iommu.iommu_control, + psycho_write(iommu->iommu_control, control | PSYCHO_IOMMU_CTRL_DENAB); for (i = 0; i < 16; i++) { unsigned long base = p->controller_regs; @@ -678,7 +679,7 @@ static void psycho_check_iommu_error(struct pci_controller_info *p, } /* Leave diagnostic mode. */ - psycho_write(p->iommu.iommu_control, control); + psycho_write(iommu->iommu_control, control); for (i = 0; i < 16; i++) { unsigned long tag, data; @@ -717,7 +718,7 @@ static void psycho_check_iommu_error(struct pci_controller_info *p, } } __psycho_check_stc_error(p, afsr, afar, type); - spin_unlock_irqrestore(&p->iommu.lock, flags); + spin_unlock_irqrestore(&iommu->lock, flags); } /* Uncorrectable Errors. Cause of the error and the address are @@ -1255,24 +1256,25 @@ static void __init psycho_scan_bus(struct pci_controller_info *p) static void __init psycho_iommu_init(struct pci_controller_info *p) { + struct pci_iommu *iommu = p->pbm_A.iommu; unsigned long tsbbase, i; u64 control; /* Setup initial software IOMMU state. */ - spin_lock_init(&p->iommu.lock); - p->iommu.iommu_cur_ctx = 0; + spin_lock_init(&iommu->lock); + iommu->iommu_cur_ctx = 0; /* Register addresses. */ - p->iommu.iommu_control = p->controller_regs + PSYCHO_IOMMU_CONTROL; - p->iommu.iommu_tsbbase = p->controller_regs + PSYCHO_IOMMU_TSBBASE; - p->iommu.iommu_flush = p->controller_regs + PSYCHO_IOMMU_FLUSH; + iommu->iommu_control = p->controller_regs + PSYCHO_IOMMU_CONTROL; + iommu->iommu_tsbbase = p->controller_regs + PSYCHO_IOMMU_TSBBASE; + iommu->iommu_flush = p->controller_regs + PSYCHO_IOMMU_FLUSH; /* PSYCHO's IOMMU lacks ctx flushing. */ - p->iommu.iommu_ctxflush = 0; + iommu->iommu_ctxflush = 0; /* We use the main control register of PSYCHO as the write * completion register. */ - p->iommu.write_complete_reg = p->controller_regs + PSYCHO_CONTROL; + iommu->write_complete_reg = p->controller_regs + PSYCHO_CONTROL; /* * Invalidate TLB Entries. @@ -1298,19 +1300,19 @@ static void __init psycho_iommu_init(struct pci_controller_info *p) prom_printf("PSYCHO_IOMMU: Error, gfp(tsb) failed.\n"); prom_halt(); } - p->iommu.page_table = (iopte_t *)tsbbase; - p->iommu.page_table_sz_bits = 17; - p->iommu.page_table_map_base = 0xc0000000; - p->iommu.dma_addr_mask = 0xffffffff; + iommu->page_table = (iopte_t *)tsbbase; + iommu->page_table_sz_bits = 17; + iommu->page_table_map_base = 0xc0000000; + iommu->dma_addr_mask = 0xffffffff; memset((char *)tsbbase, 0, PAGE_SIZE << 7); /* We start with no consistent mappings. */ - p->iommu.lowest_consistent_map = - 1 << (p->iommu.page_table_sz_bits - PBM_LOGCLUSTERS); + iommu->lowest_consistent_map = + 1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS); for (i = 0; i < PBM_NCLUSTERS; i++) { - p->iommu.alloc_info[i].flush = 0; - p->iommu.alloc_info[i].next = 0; + iommu->alloc_info[i].flush = 0; + iommu->alloc_info[i].next = 0; } psycho_write(p->controller_regs + PSYCHO_IOMMU_TSBBASE, __pa(tsbbase)); @@ -1515,6 +1517,7 @@ void __init psycho_init(int node) { struct linux_prom64_registers pr_regs[3]; struct pci_controller_info *p; + struct pci_iommu *iommu; unsigned long flags; u32 upa_portid; int is_pbm_a, err; @@ -1538,6 +1541,13 @@ void __init psycho_init(int node) prom_halt(); } memset(p, 0, sizeof(*p)); + iommu = kmalloc(sizeof(struct pci_iommu), GFP_ATOMIC); + if (!iommu) { + prom_printf("PSYCHO: Fatal memory allocation error.\n"); + prom_halt(); + } + memset(iommu, 0, sizeof(*iommu)); + p->pbm_A.iommu = p->pbm_B.iommu = iommu; spin_lock_irqsave(&pci_controller_lock, flags); p->next = pci_controller_root; diff --git a/arch/sparc64/kernel/pci_sabre.c b/arch/sparc64/kernel/pci_sabre.c index f3a5adbec..ec74a3696 100644 --- a/arch/sparc64/kernel/pci_sabre.c +++ b/arch/sparc64/kernel/pci_sabre.c @@ -1,4 +1,4 @@ -/* $Id: pci_sabre.c,v 1.20 2000/06/26 19:40:27 davem Exp $ +/* $Id: pci_sabre.c,v 1.22 2001/01/16 13:03:48 anton Exp $ * pci_sabre.c: Sabre specific PCI controller support. * * Copyright (C) 1997, 1998, 1999 David S. Miller (davem@caipfs.rutgers.edu) @@ -672,14 +672,15 @@ static void sabre_check_iommu_error(struct pci_controller_info *p, unsigned long afsr, unsigned long afar) { + struct pci_iommu *iommu = p->pbm_A.iommu; unsigned long iommu_tag[16]; unsigned long iommu_data[16]; unsigned long flags; u64 control; int i; - spin_lock_irqsave(&p->iommu.lock, flags); - control = sabre_read(p->iommu.iommu_control); + spin_lock_irqsave(&iommu->lock, flags); + control = sabre_read(iommu->iommu_control); if (control & SABRE_IOMMUCTRL_ERR) { char *type_string; @@ -687,7 +688,7 @@ static void sabre_check_iommu_error(struct pci_controller_info *p, * NOTE: On Sabre this is write 1 to clear, * which is different from Psycho. */ - sabre_write(p->iommu.iommu_control, control); + sabre_write(iommu->iommu_control, control); switch((control & SABRE_IOMMUCTRL_ERRSTS) >> 25UL) { case 1: type_string = "Invalid Error"; @@ -706,7 +707,7 @@ static void sabre_check_iommu_error(struct pci_controller_info *p, * entries in the IOTLB. */ control &= ~(SABRE_IOMMUCTRL_ERRSTS | SABRE_IOMMUCTRL_ERR); - sabre_write(p->iommu.iommu_control, + sabre_write(iommu->iommu_control, (control | SABRE_IOMMUCTRL_DENAB)); for (i = 0; i < 16; i++) { unsigned long base = p->controller_regs; @@ -718,7 +719,7 @@ static void sabre_check_iommu_error(struct pci_controller_info *p, sabre_write(base + SABRE_IOMMU_TAG + (i * 8UL), 0); sabre_write(base + SABRE_IOMMU_DATA + (i * 8UL), 0); } - sabre_write(p->iommu.iommu_control, control); + sabre_write(iommu->iommu_control, control); for (i = 0; i < 16; i++) { unsigned long tag, data; @@ -752,7 +753,7 @@ static void sabre_check_iommu_error(struct pci_controller_info *p, ((data & SABRE_IOMMUDATA_PPN) << PAGE_SHIFT)); } } - spin_unlock_irqrestore(&p->iommu.lock, flags); + spin_unlock_irqrestore(&iommu->lock, flags); } static void sabre_ue_intr(int irq, void *dev_id, struct pt_regs *regs) @@ -1158,20 +1159,21 @@ static void __init sabre_iommu_init(struct pci_controller_info *p, int tsbsize, unsigned long dvma_offset, u32 dma_mask) { + struct pci_iommu *iommu = p->pbm_A.iommu; unsigned long tsbbase, i, order; u64 control; /* Setup initial software IOMMU state. */ - spin_lock_init(&p->iommu.lock); - p->iommu.iommu_cur_ctx = 0; + spin_lock_init(&iommu->lock); + iommu->iommu_cur_ctx = 0; /* Register addresses. */ - p->iommu.iommu_control = p->controller_regs + SABRE_IOMMU_CONTROL; - p->iommu.iommu_tsbbase = p->controller_regs + SABRE_IOMMU_TSBBASE; - p->iommu.iommu_flush = p->controller_regs + SABRE_IOMMU_FLUSH; - p->iommu.write_complete_reg = p->controller_regs + SABRE_WRSYNC; + iommu->iommu_control = p->controller_regs + SABRE_IOMMU_CONTROL; + iommu->iommu_tsbbase = p->controller_regs + SABRE_IOMMU_TSBBASE; + iommu->iommu_flush = p->controller_regs + SABRE_IOMMU_FLUSH; + iommu->write_complete_reg = p->controller_regs + SABRE_WRSYNC; /* Sabre's IOMMU lacks ctx flushing. */ - p->iommu.iommu_ctxflush = 0; + iommu->iommu_ctxflush = 0; /* Invalidate TLB Entries. */ control = sabre_read(p->controller_regs + SABRE_IOMMU_CONTROL); @@ -1192,9 +1194,9 @@ static void __init sabre_iommu_init(struct pci_controller_info *p, prom_printf("SABRE_IOMMU: Error, gfp(tsb) failed.\n"); prom_halt(); } - p->iommu.page_table = (iopte_t *)tsbbase; - p->iommu.page_table_map_base = dvma_offset; - p->iommu.dma_addr_mask = dma_mask; + iommu->page_table = (iopte_t *)tsbbase; + iommu->page_table_map_base = dvma_offset; + iommu->dma_addr_mask = dma_mask; memset((char *)tsbbase, 0, PAGE_SIZE << order); sabre_write(p->controller_regs + SABRE_IOMMU_TSBBASE, __pa(tsbbase)); @@ -1205,11 +1207,11 @@ static void __init sabre_iommu_init(struct pci_controller_info *p, switch(tsbsize) { case 64: control |= SABRE_IOMMU_TSBSZ_64K; - p->iommu.page_table_sz_bits = 16; + iommu->page_table_sz_bits = 16; break; case 128: control |= SABRE_IOMMU_TSBSZ_128K; - p->iommu.page_table_sz_bits = 17; + iommu->page_table_sz_bits = 17; break; default: prom_printf("iommu_init: Illegal TSB size %d\n", tsbsize); @@ -1219,12 +1221,12 @@ static void __init sabre_iommu_init(struct pci_controller_info *p, sabre_write(p->controller_regs + SABRE_IOMMU_CONTROL, control); /* We start with no consistent mappings. */ - p->iommu.lowest_consistent_map = - 1 << (p->iommu.page_table_sz_bits - PBM_LOGCLUSTERS); + iommu->lowest_consistent_map = + 1 << (iommu->page_table_sz_bits - PBM_LOGCLUSTERS); for (i = 0; i < PBM_NCLUSTERS; i++) { - p->iommu.alloc_info[i].flush = 0; - p->iommu.alloc_info[i].next = 0; + iommu->alloc_info[i].flush = 0; + iommu->alloc_info[i].next = 0; } } @@ -1368,6 +1370,7 @@ void __init sabre_init(int pnode) { struct linux_prom64_registers pr_regs[2]; struct pci_controller_info *p; + struct pci_iommu *iommu; unsigned long flags; int tsbsize, err; u32 busrange[2]; @@ -1380,10 +1383,17 @@ void __init sabre_init(int pnode) prom_printf("SABRE: Error, kmalloc(pci_controller_info) failed.\n"); prom_halt(); } + memset(p, 0, sizeof(*p)); - upa_portid = prom_getintdefault(pnode, "upa-portid", 0xff); + iommu = kmalloc(sizeof(*iommu), GFP_ATOMIC); + if (!iommu) { + prom_printf("SABRE: Error, kmalloc(pci_iommu) failed.\n"); + prom_halt(); + } + memset(iommu, 0, sizeof(*iommu)); + p->pbm_A.iommu = p->pbm_B.iommu = iommu; - memset(p, 0, sizeof(*p)); + upa_portid = prom_getintdefault(pnode, "upa-portid", 0xff); spin_lock_irqsave(&pci_controller_lock, flags); p->next = pci_controller_root; diff --git a/arch/sparc64/kernel/pci_schizo.c b/arch/sparc64/kernel/pci_schizo.c new file mode 100644 index 000000000..9299c2531 --- /dev/null +++ b/arch/sparc64/kernel/pci_schizo.c @@ -0,0 +1,169 @@ +/* $Id: pci_schizo.c,v 1.2 2001/01/12 02:43:30 davem Exp $ + * pci_schizo.c: SCHIZO specific PCI controller support. + * + * Copyright (C) 2001 David S. Miller (davem@redhat.com) + */ + +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/pci.h> +#include <linux/init.h> +#include <linux/malloc.h> + +#include <asm/pbm.h> +#include <asm/iommu.h> +#include <asm/irq.h> + +#include "pci_impl.h" + +static int schizo_read_byte(struct pci_dev *dev, int where, u8 *value) +{ + /* IMPLEMENT ME */ +} + +static int schizo_read_word(struct pci_dev *dev, int where, u16 *value) +{ + /* IMPLEMENT ME */ +} + +static int schizo_read_dword(struct pci_dev *dev, int where, u32 *value) +{ + /* IMPLEMENT ME */ +} + +static int schizo_write_byte(struct pci_dev *dev, int where, u8 value) +{ + /* IMPLEMENT ME */ +} + +static int schizo_write_word(struct pci_dev *dev, int where, u16 value) +{ + /* IMPLEMENT ME */ +} + +static int schizo_write_dword(struct pci_dev *dev, int where, u32 value) +{ + /* IMPLEMENT ME */ +} + +static struct pci_ops schizo_ops = { + schizo_read_byte, + schizo_read_word, + schizo_read_dword, + schizo_write_byte, + schizo_write_word, + schizo_write_dword +}; + +static void __init schizo_scan_bus(struct pci_controller_info *p) +{ + /* IMPLEMENT ME */ +} + +static unsigned int __init schizo_irq_build(struct pci_controller_info *p, + struct pci_dev *pdev, + unsigned int ino) +{ + /* IMPLEMENT ME */ +} + +static void __init schizo_base_address_update(struct pci_dev *pdev, int resource) +{ + /* IMPLEMENT ME */ +} + +static void __init schizo_resource_adjust(struct pci_dev *pdev, + struct resource *res, + struct resource *root) +{ + /* IMPLEMENT ME */ +} + +static void schizo_pbm_init(struct pci_controller_info *p, + int prom_node, int is_pbm_a) +{ + /* IMPLEMENT ME */ +} + +void __init schizo_init(int node) +{ + struct linux_prom64_registers pr_regs[3]; + struct pci_controller_info *p; + struct pci_iommu *iommu; + u32 portid; + int is_pbm_a, err; + + portid = prom_getintdefault(node, "portid", 0xff); + + spin_lock_irqsave(&pci_controller_lock, flags); + for(p = pci_controller_root; p; p = p->next) { + if (p->portid == portid) { + spin_unlock_irqrestore(&pci_controller_lock, flags); + is_pbm_a = (p->pbm_A.prom_node == 0); + schizo_pbm_init(p, node, is_pbm_a); + return; + } + } + spin_unlock_irqrestore(&pci_controller_lock, flags); + + p = kmalloc(sizeof(struct pci_controller_info), GFP_ATOMIC); + if (!p) { + prom_printf("SCHIZO: Fatal memory allocation error.\n"); + prom_halt(); + } + memset(p, 0, sizeof(*p)); + + iommu = kmalloc(sizeof(struct pci_iommu), GFP_ATOMIC); + if (!iommu) { + prom_printf("SCHIZO: Fatal memory allocation error.\n"); + prom_halt(); + } + memset(iommu, 0, sizeof(*iommu)); + p->pbm_A.iommu = iommu; + + iommu = kmalloc(sizeof(struct pci_iommu), GFP_ATOMIC); + if (!iommu) { + prom_printf("SCHIZO: Fatal memory allocation error.\n"); + prom_halt(); + } + memset(iommu, 0, sizeof(*iommu)); + p->pbm_B.iommu = iommu; + + spin_lock_irqsave(&pci_controller_lock, flags); + p->next = pci_controller_root; + pci_controller_root = p; + spin_unlock_irqrestore(&pci_controller_lock, flags); + + p->portid = portid; + p->index = pci_num_controllers++; + p->scan_bus = schizo_scan_bus; + p->irq_build = schizo_irq_build; + p->base_address_update = schizo_base_address_update; + p->resource_adjust = schizo_resource_adjust; + p->pci_ops = &schizo_ops; + +pbm_init: + /* Three OBP regs: + * 1) PBM controller regs + * 2) Schizo front-end controller regs (same for both PBMs) + * 3) Unknown... (0x7ffec000000 and 0x7ffee000000 on Excalibur) + */ + err = prom_getproperty(node, "reg", + (char *)&pr_regs[0], + sizeof(pr_regs)); + if (err == 0 || err == -1) { + prom_printf("SCHIZO: Fatal error, no reg property.\n"); + prom_halt(); + } + + /* XXX Read REG base, record in controller/pbm structures. */ + + /* XXX Report controller to console. */ + + /* XXX Setup pci_memspace_mask */ + + /* XXX Init core controller and IOMMU */ + + is_pbm_a = XXX; /* Figure out this test */ + schizo_pbm_init(p, node, is_pbm_a); +} diff --git a/arch/sparc64/kernel/signal.c b/arch/sparc64/kernel/signal.c index c2a7833fb..23d0774b4 100644 --- a/arch/sparc64/kernel/signal.c +++ b/arch/sparc64/kernel/signal.c @@ -1,4 +1,4 @@ -/* $Id: signal.c,v 1.54 2000/09/05 21:44:54 davem Exp $ +/* $Id: signal.c,v 1.55 2001/01/24 21:05:13 davem Exp $ * arch/sparc64/kernel/signal.c * * Copyright (C) 1991, 1992 Linus Torvalds @@ -31,9 +31,6 @@ #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); - asmlinkage int do_signal(sigset_t *oldset, struct pt_regs * regs, unsigned long orig_o0, int ret_from_syscall); diff --git a/arch/sparc64/kernel/signal32.c b/arch/sparc64/kernel/signal32.c index 6d06328dd..0886d9d39 100644 --- a/arch/sparc64/kernel/signal32.c +++ b/arch/sparc64/kernel/signal32.c @@ -1,4 +1,4 @@ -/* $Id: signal32.c,v 1.67 2000/09/05 21:44:54 davem Exp $ +/* $Id: signal32.c,v 1.68 2001/01/24 21:05:13 davem Exp $ * arch/sparc64/kernel/signal32.c * * Copyright (C) 1991, 1992 Linus Torvalds @@ -29,9 +29,6 @@ #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) -asmlinkage int sys_wait4(pid_t pid, unsigned long *stat_addr, - int options, unsigned long *ru); - asmlinkage int do_signal32(sigset_t *oldset, struct pt_regs *regs, unsigned long orig_o0, int ret_from_syscall); diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c index 25f6ecff3..76045d0d2 100644 --- a/arch/sparc64/kernel/smp.c +++ b/arch/sparc64/kernel/smp.c @@ -290,8 +290,8 @@ void __init smp_boot_cpus(void) } printk("Total of %d processors activated (%lu.%02lu BogoMIPS).\n", cpucount + 1, - (bogosum + 2500)/500000, - ((bogosum + 2500)/5000)%100); + bogosum/(500000/HZ), + (bogosum/(5000/HZ))%100); smp_activated = 1; smp_num_cpus = cpucount + 1; } diff --git a/arch/sparc64/kernel/sparc64_ksyms.c b/arch/sparc64/kernel/sparc64_ksyms.c index a0311626e..f35e38a5d 100644 --- a/arch/sparc64/kernel/sparc64_ksyms.c +++ b/arch/sparc64/kernel/sparc64_ksyms.c @@ -1,4 +1,4 @@ -/* $Id: sparc64_ksyms.c,v 1.99 2000/12/09 04:15:24 anton Exp $ +/* $Id: sparc64_ksyms.c,v 1.100 2001/01/11 15:07:09 davem Exp $ * arch/sparc64/kernel/sparc64_ksyms.c: Sparc64 specific ksyms support. * * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu) @@ -182,6 +182,7 @@ EXPORT_SYMBOL(__flushw_user); EXPORT_SYMBOL(flush_icache_range); EXPORT_SYMBOL(__flush_dcache_page); +EXPORT_SYMBOL(mostek_lock); EXPORT_SYMBOL(mstk48t02_regs); EXPORT_SYMBOL(request_fast_irq); #if CONFIG_SBUS diff --git a/arch/sparc64/kernel/sys_sparc32.c b/arch/sparc64/kernel/sys_sparc32.c index 775c986c8..184b4169d 100644 --- a/arch/sparc64/kernel/sys_sparc32.c +++ b/arch/sparc64/kernel/sys_sparc32.c @@ -1,4 +1,4 @@ -/* $Id: sys_sparc32.c,v 1.171 2000/12/13 16:34:55 davem Exp $ +/* $Id: sys_sparc32.c,v 1.172 2001/01/24 21:05:13 davem Exp $ * sys_sparc32.c: Conversion between 32bit and 64bit native syscalls. * * Copyright (C) 1997,1998 Jakub Jelinek (jj@sunsite.mff.cuni.cz) @@ -1794,9 +1794,6 @@ static int put_rusage (struct rusage32 *ru, struct rusage *r) return err; } -extern asmlinkage int sys_wait4(pid_t pid,unsigned int * stat_addr, - int options, struct rusage * ru); - asmlinkage int sys32_wait4(__kernel_pid_t32 pid, unsigned int *stat_addr, int options, struct rusage32 *ru) { if (!ru) diff --git a/arch/sparc64/kernel/sys_sunos32.c b/arch/sparc64/kernel/sys_sunos32.c index a9d143759..58e4704c4 100644 --- a/arch/sparc64/kernel/sys_sunos32.c +++ b/arch/sparc64/kernel/sys_sunos32.c @@ -1,4 +1,4 @@ -/* $Id: sys_sunos32.c,v 1.55 2000/11/18 02:10:59 davem Exp $ +/* $Id: sys_sunos32.c,v 1.56 2001/01/04 05:35:48 davem Exp $ * sys_sunos32.c: SunOS binary compatability layer on sparc64. * * Copyright (C) 1995, 1996, 1997 David S. Miller (davem@caip.rutgers.edu) @@ -1047,6 +1047,7 @@ asmlinkage int sunos_msgsys(int op, u32 arg1, u32 arg2, u32 arg3, u32 arg4) (current->thread.kregs->u_regs[UREG_FP] & 0xffffffffUL); if(get_user(arg5, &sp->xxargs[0])) { rval = -EFAULT; + kfree(kmbuf); break; } set_fs(KERNEL_DS); diff --git a/arch/sparc64/kernel/time.c b/arch/sparc64/kernel/time.c index 0264f9418..a3340f54a 100644 --- a/arch/sparc64/kernel/time.c +++ b/arch/sparc64/kernel/time.c @@ -1,4 +1,4 @@ -/* $Id: time.c,v 1.32 2000/09/22 23:02:13 davem Exp $ +/* $Id: time.c,v 1.33 2001/01/11 15:07:09 davem Exp $ * time.c: UltraSparc timer and TOD clock support. * * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu) @@ -34,7 +34,9 @@ extern rwlock_t xtime_lock; +spinlock_t mostek_lock = SPIN_LOCK_UNLOCKED; unsigned long mstk48t02_regs = 0UL; + static unsigned long mstk48t08_regs = 0UL; static unsigned long mstk48t59_regs = 0UL; @@ -187,6 +189,8 @@ static void __init kick_start_clock(void) prom_printf("CLOCK: Clock was stopped. Kick start "); + spin_lock_irq(&mostek_lock); + /* Turn on the kick start bit to start the oscillator. */ tmp = mostek_read(regs + MOSTEK_CREG); tmp |= MSTK_CREG_WRITE; @@ -201,6 +205,8 @@ static void __init kick_start_clock(void) tmp &= ~MSTK_CREG_WRITE; mostek_write(regs + MOSTEK_CREG, tmp); + spin_unlock_irq(&mostek_lock); + /* Delay to allow the clock oscillator to start. */ sec = MSTK_REG_SEC(regs); for (i = 0; i < 3; i++) { @@ -212,6 +218,8 @@ static void __init kick_start_clock(void) } prom_printf("\n"); + spin_lock_irq(&mostek_lock); + /* Turn off kick start and set a "valid" time and date. */ tmp = mostek_read(regs + MOSTEK_CREG); tmp |= MSTK_CREG_WRITE; @@ -230,9 +238,14 @@ static void __init kick_start_clock(void) tmp &= ~MSTK_CREG_WRITE; mostek_write(regs + MOSTEK_CREG, tmp); + spin_unlock_irq(&mostek_lock); + /* Ensure the kick start bit is off. If it isn't, turn it off. */ while (mostek_read(regs + MOSTEK_HOUR) & MSTK_KICK_START) { prom_printf("CLOCK: Kick start still on!\n"); + + spin_lock_irq(&mostek_lock); + tmp = mostek_read(regs + MOSTEK_CREG); tmp |= MSTK_CREG_WRITE; mostek_write(regs + MOSTEK_CREG, tmp); @@ -244,6 +257,8 @@ static void __init kick_start_clock(void) tmp = mostek_read(regs + MOSTEK_CREG); tmp &= ~MSTK_CREG_WRITE; mostek_write(regs + MOSTEK_CREG, tmp); + + spin_unlock_irq(&mostek_lock); } prom_printf("CLOCK: Kick start procedure successful.\n"); @@ -255,11 +270,15 @@ static int __init has_low_battery(void) unsigned long regs = mstk48t02_regs; u8 data1, data2; + spin_lock_irq(&mostek_lock); + data1 = mostek_read(regs + MOSTEK_EEPROM); /* Read some data. */ mostek_write(regs + MOSTEK_EEPROM, ~data1); /* Write back the complement. */ data2 = mostek_read(regs + MOSTEK_EEPROM); /* Read back the complement. */ mostek_write(regs + MOSTEK_EEPROM, data1); /* Restore original value. */ + spin_unlock_irq(&mostek_lock); + return (data1 == data2); /* Was the write blocked? */ } @@ -278,6 +297,8 @@ static void __init set_system_time(void) prom_halt(); } + spin_lock_irq(&mostek_lock); + tmp = mostek_read(mregs + MOSTEK_CREG); tmp |= MSTK_CREG_READ; mostek_write(mregs + MOSTEK_CREG, tmp); @@ -294,6 +315,8 @@ static void __init set_system_time(void) tmp = mostek_read(mregs + MOSTEK_CREG); tmp &= ~MSTK_CREG_READ; mostek_write(mregs + MOSTEK_CREG, tmp); + + spin_unlock_irq(&mostek_lock); } void __init clock_probe(void) @@ -512,6 +535,7 @@ static int set_rtc_mmss(unsigned long nowtime) { int real_seconds, real_minutes, mostek_minutes; unsigned long regs = mstk48t02_regs; + unsigned long flags; u8 tmp; /* @@ -521,6 +545,8 @@ static int set_rtc_mmss(unsigned long nowtime) if (!regs) return -1; + spin_lock_irqsave(&mostek_lock, flags); + /* Read the current RTC minutes. */ tmp = mostek_read(regs + MOSTEK_CREG); tmp |= MSTK_CREG_READ; @@ -555,8 +581,13 @@ static int set_rtc_mmss(unsigned long nowtime) tmp = mostek_read(regs + MOSTEK_CREG); tmp &= ~MSTK_CREG_WRITE; mostek_write(regs + MOSTEK_CREG, tmp); - } else - return -1; - return 0; + spin_unlock_irqrestore(&mostek_lock, flags); + + return 0; + } else { + spin_unlock_irqrestore(&mostek_lock, flags); + + return -1; + } } diff --git a/drivers/acpi/Makefile b/drivers/acpi/Makefile index 307cb08ad..e1e0745a4 100644 --- a/drivers/acpi/Makefile +++ b/drivers/acpi/Makefile @@ -4,7 +4,7 @@ O_TARGET := acpi.o -export-objs := ksyms.o +export-objs := acpi_ksyms.o export ACPI_CFLAGS ACPI_CFLAGS := -D_LINUX @@ -20,18 +20,26 @@ EXTRA_CFLAGS += -I./include EXTRA_CFLAGS += $(ACPI_CFLAGS) +# genksyms only reads $(CFLAGS), it should really read $(EXTRA_CFLAGS) as well. +# Without EXTRA_CFLAGS the gcc pass for genksyms fails, resulting in an empty +# include/linux/modules/acpi_ksyms.ver. Changing genkyms to use EXTRA_CFLAGS +# will hit everything, too risky in 2.4.0-prerelease. Bandaid by tweaking +# CFLAGS only for .ver targets. Review after 2.4.0 release. KAO + +$(MODINCL)/%.ver: CFLAGS := -I./include $(CFLAGS) + acpi-subdirs := common dispatcher events hardware \ interpreter namespace parser resources tables subdir-$(CONFIG_ACPI) += $(acpi-subdirs) obj-$(CONFIG_ACPI) := $(patsubst %,%.o,$(acpi-subdirs)) -obj-$(CONFIG_ACPI) += os.o ksyms.o +obj-$(CONFIG_ACPI) += os.o acpi_ksyms.o ifdef CONFIG_ACPI_KERNEL_CONFIG obj-$(CONFIG_ACPI) += acpiconf.o osconf.o else - obj-$(CONFIG_ACPI) += driver.o cmbatt.o cpu.o ec.o ksyms.o sys.o table.o power.o + obj-$(CONFIG_ACPI) += driver.o cmbatt.o cpu.o ec.o acpi_ksyms.o sys.o table.o power.o endif include $(TOPDIR)/Rules.make diff --git a/drivers/acpi/acpi_ksyms.c b/drivers/acpi/acpi_ksyms.c new file mode 100644 index 000000000..8e1774c9a --- /dev/null +++ b/drivers/acpi/acpi_ksyms.c @@ -0,0 +1,83 @@ +/* + * ksyms.c - ACPI exported symbols + * + * Copyright (C) 2000 Andrew Grover + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include <linux/module.h> +#include <linux/init.h> +#include <linux/kernel.h> +#include <linux/types.h> +#include <linux/acpi.h> +#include "acpi.h" +#include "acdebug.h" + +extern int acpi_in_debugger; + +#define _COMPONENT OS_DEPENDENT + MODULE_NAME ("symbols") + +#ifdef ENABLE_DEBUGGER +EXPORT_SYMBOL(acpi_in_debugger); +EXPORT_SYMBOL(acpi_db_user_commands); +#endif + +EXPORT_SYMBOL(acpi_os_free); +EXPORT_SYMBOL(acpi_os_breakpoint); +EXPORT_SYMBOL(acpi_os_printf); +EXPORT_SYMBOL(acpi_os_callocate); +EXPORT_SYMBOL(acpi_os_sleep); +EXPORT_SYMBOL(acpi_os_sleep_usec); +EXPORT_SYMBOL(acpi_os_in8); +EXPORT_SYMBOL(acpi_os_out8); +EXPORT_SYMBOL(acpi_os_queue_for_execution); + +EXPORT_SYMBOL(acpi_dbg_layer); +EXPORT_SYMBOL(acpi_dbg_level); +EXPORT_SYMBOL(function_exit); +EXPORT_SYMBOL(function_trace); +EXPORT_SYMBOL(function_status_exit); +EXPORT_SYMBOL(function_value_exit); +EXPORT_SYMBOL(debug_print_raw); +EXPORT_SYMBOL(debug_print_prefix); + +EXPORT_SYMBOL(acpi_cm_strncmp); +EXPORT_SYMBOL(acpi_cm_memcpy); +EXPORT_SYMBOL(acpi_cm_memset); + +EXPORT_SYMBOL(acpi_get_handle); +EXPORT_SYMBOL(acpi_get_parent); +EXPORT_SYMBOL(acpi_get_type); +EXPORT_SYMBOL(acpi_get_name); +EXPORT_SYMBOL(acpi_get_object_info); +EXPORT_SYMBOL(acpi_get_next_object); +EXPORT_SYMBOL(acpi_evaluate_object); + +EXPORT_SYMBOL(acpi_install_notify_handler); +EXPORT_SYMBOL(acpi_remove_notify_handler); +EXPORT_SYMBOL(acpi_install_gpe_handler); +EXPORT_SYMBOL(acpi_remove_gpe_handler); +EXPORT_SYMBOL(acpi_install_address_space_handler); +EXPORT_SYMBOL(acpi_remove_address_space_handler); + +EXPORT_SYMBOL(acpi_get_current_resources); +EXPORT_SYMBOL(acpi_get_possible_resources); +EXPORT_SYMBOL(acpi_set_current_resources); + +EXPORT_SYMBOL(acpi_enable_event); +EXPORT_SYMBOL(acpi_disable_event); +EXPORT_SYMBOL(acpi_clear_event); diff --git a/drivers/acpi/cmbatt.c b/drivers/acpi/cmbatt.c index c45aa810c..d5dc5977a 100644 --- a/drivers/acpi/cmbatt.c +++ b/drivers/acpi/cmbatt.c @@ -113,10 +113,10 @@ acpi_get_battery_status(ACPI_HANDLE handle, struct cmbatt_status *result) obj = (ACPI_OBJECT *) buf.pointer; objs = obj->package.elements; - result->state = objs[0].number.value; - result->present_rate = objs[1].number.value; - result->remaining_capacity = objs[2].number.value; - result->present_voltage = objs[3].number.value; + result->state = objs[0].integer.value; + result->present_rate = objs[1].integer.value; + result->remaining_capacity = objs[2].integer.value; + result->present_voltage = objs[3].integer.value; kfree(buf.pointer); @@ -153,15 +153,15 @@ acpi_get_battery_info(ACPI_HANDLE handle, struct cmbatt_info *result) obj = (ACPI_OBJECT *) buf.pointer; objs = obj->package.elements; - result->power_unit=objs[0].number.value; - result->design_capacity=objs[1].number.value; - result->last_full_capacity=objs[2].number.value; - result->battery_technology=objs[3].number.value; - result->design_voltage=objs[4].number.value; - result->design_capacity_warning=objs[5].number.value; - result->design_capacity_low=objs[6].number.value; - result->battery_capacity_granularity_1=objs[7].number.value; - result->battery_capacity_granularity_2=objs[8].number.value; + result->power_unit=objs[0].integer.value; + result->design_capacity=objs[1].integer.value; + result->last_full_capacity=objs[2].integer.value; + result->battery_technology=objs[3].integer.value; + result->design_voltage=objs[4].integer.value; + result->design_capacity_warning=objs[5].integer.value; + result->design_capacity_low=objs[6].integer.value; + result->battery_capacity_granularity_1=objs[7].integer.value; + result->battery_capacity_granularity_2=objs[8].integer.value; /* BUG: trailing NULL issue */ strncpy(result->model_number, objs[9].string.pointer, MAX_BATT_STRLEN-1); diff --git a/drivers/acpi/common/cmalloc.c b/drivers/acpi/common/cmalloc.c index b7a64e5b6..d8514698f 100644 --- a/drivers/acpi/common/cmalloc.c +++ b/drivers/acpi/common/cmalloc.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: cmalloc - local memory allocation routines - * $Revision: 79 $ + * $Revision: 84 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -57,8 +57,6 @@ _cm_allocate ( u32 line) { void *address = NULL; - DEBUG_ONLY_MEMBERS (\ - ACPI_STATUS status) /* Check for an inadvertent size of zero bytes */ @@ -107,8 +105,6 @@ _cm_callocate ( u32 line) { void *address = NULL; - DEBUG_ONLY_MEMBERS (\ - ACPI_STATUS status) /* Check for an inadvertent size of zero bytes */ diff --git a/drivers/acpi/common/cmclib.c b/drivers/acpi/common/cmclib.c index 5146b09c5..88e92d17f 100644 --- a/drivers/acpi/common/cmclib.c +++ b/drivers/acpi/common/cmclib.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: cmclib - Local implementation of C library functions - * $Revision: 28 $ + * $Revision: 32 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -347,7 +347,7 @@ acpi_cm_memcpy ( void * acpi_cm_memset ( void *dest, - u32 value, + NATIVE_UINT value, NATIVE_UINT count) { NATIVE_CHAR *new = (NATIVE_CHAR *) dest; @@ -522,6 +522,7 @@ static const u8 _acpi_ctype[257] = { #define IS_LOWER(c) (_acpi_ctype[(unsigned char)(c)] & (_ACPI_LO)) #define IS_DIGIT(c) (_acpi_ctype[(unsigned char)(c)] & (_ACPI_DI)) #define IS_SPACE(c) (_acpi_ctype[(unsigned char)(c)] & (_ACPI_SP)) +#define IS_XDIGIT(c) (_acpi_ctype[(unsigned char)(c)] & (_ACPI_XD)) /******************************************************************************* @@ -658,7 +659,7 @@ u32 acpi_cm_strtoul ( const NATIVE_CHAR *string, NATIVE_CHAR **terminator, - u32 base) + NATIVE_UINT base) { u32 converted = 0; u32 index; diff --git a/drivers/acpi/common/cmcopy.c b/drivers/acpi/common/cmcopy.c index 68b7bda01..da3851c17 100644 --- a/drivers/acpi/common/cmcopy.c +++ b/drivers/acpi/common/cmcopy.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: cmcopy - Internal to external object translation utilities - * $Revision: 61 $ + * $Revision: 62 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -118,9 +118,9 @@ acpi_cm_build_external_simple_object ( break; - case ACPI_TYPE_NUMBER: + case ACPI_TYPE_INTEGER: - external_obj->number.value= internal_obj->number.value; + external_obj->integer.value= internal_obj->integer.value; break; @@ -469,11 +469,11 @@ acpi_cm_build_internal_simple_object ( break; - case ACPI_TYPE_NUMBER: + case ACPI_TYPE_INTEGER: /* * Number is included in the object itself */ - internal_obj->number.value = external_obj->number.value; + internal_obj->integer.value = external_obj->integer.value; break; diff --git a/drivers/acpi/common/cmdebug.c b/drivers/acpi/common/cmdebug.c index a55372d5c..2d0022ac3 100644 --- a/drivers/acpi/common/cmdebug.c +++ b/drivers/acpi/common/cmdebug.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: cmdebug - Debug print routines - * $Revision: 61 $ + * $Revision: 64 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -281,7 +281,7 @@ function_value_exit ( u32 line_number, u32 component_id, NATIVE_CHAR *function_name, - NATIVE_UINT value) + ACPI_INTEGER value) { debug_print (module_name, line_number, component_id, TRACE_FUNCTIONS, diff --git a/drivers/acpi/common/cmdelete.c b/drivers/acpi/common/cmdelete.c index a6e74c12f..b516b691c 100644 --- a/drivers/acpi/common/cmdelete.c +++ b/drivers/acpi/common/cmdelete.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: cmdelete - object deletion and reference count utilities - * $Revision: 60 $ + * $Revision: 62 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/common/cmeval.c b/drivers/acpi/common/cmeval.c index 29a5cefa6..653e26a47 100644 --- a/drivers/acpi/common/cmeval.c +++ b/drivers/acpi/common/cmeval.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: cmeval - Object evaluation - * $Revision: 19 $ + * $Revision: 21 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -77,7 +77,7 @@ acpi_cm_evaluate_numeric_object ( /* Is the return object of the correct type? */ - if (obj_desc->common.type != ACPI_TYPE_NUMBER) { + if (obj_desc->common.type != ACPI_TYPE_INTEGER) { status = AE_TYPE; } else { @@ -85,7 +85,7 @@ acpi_cm_evaluate_numeric_object ( * Since the structure is a union, setting any field will set all * of the variables in the union */ - *address = obj_desc->number.value; + *address = obj_desc->integer.value; } /* On exit, we must delete the return object */ @@ -142,17 +142,17 @@ acpi_cm_execute_HID ( * a string */ - if ((obj_desc->common.type != ACPI_TYPE_NUMBER) && + if ((obj_desc->common.type != ACPI_TYPE_INTEGER) && (obj_desc->common.type != ACPI_TYPE_STRING)) { status = AE_TYPE; } else { - if (obj_desc->common.type == ACPI_TYPE_NUMBER) { + if (obj_desc->common.type == ACPI_TYPE_INTEGER) { /* Convert the Numeric HID to string */ - acpi_aml_eisa_id_to_string ((u32) obj_desc->number.value, hid->buffer); + acpi_aml_eisa_id_to_string ((u32) obj_desc->integer.value, hid->buffer); } else { @@ -217,17 +217,17 @@ acpi_cm_execute_UID ( * a string */ - if ((obj_desc->common.type != ACPI_TYPE_NUMBER) && + if ((obj_desc->common.type != ACPI_TYPE_INTEGER) && (obj_desc->common.type != ACPI_TYPE_STRING)) { status = AE_TYPE; } else { - if (obj_desc->common.type == ACPI_TYPE_NUMBER) { + if (obj_desc->common.type == ACPI_TYPE_INTEGER) { /* Convert the Numeric UID to string */ - acpi_aml_unsigned_integer_to_string (obj_desc->number.value, uid->buffer); + acpi_aml_unsigned_integer_to_string (obj_desc->integer.value, uid->buffer); } else { @@ -289,14 +289,14 @@ acpi_cm_execute_STA ( /* Is the return object of the correct type? */ - if (obj_desc->common.type != ACPI_TYPE_NUMBER) { + if (obj_desc->common.type != ACPI_TYPE_INTEGER) { status = AE_TYPE; } else { /* Extract the status flags */ - *flags = (u32) obj_desc->number.value; + *flags = (u32) obj_desc->integer.value; } /* On exit, we must delete the return object */ diff --git a/drivers/acpi/common/cmglobal.c b/drivers/acpi/common/cmglobal.c index 4b4460f46..a388907f3 100644 --- a/drivers/acpi/common/cmglobal.c +++ b/drivers/acpi/common/cmglobal.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: cmglobal - Global variables for the ACPI subsystem - * $Revision: 112 $ + * $Revision: 116 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -95,7 +95,7 @@ PREDEFINED_NAMES acpi_gbl_pre_defined_names[] = {"_SB_", INTERNAL_TYPE_DEF_ANY}, {"_SI_", INTERNAL_TYPE_DEF_ANY}, {"_TZ_", INTERNAL_TYPE_DEF_ANY}, - {"_REV", ACPI_TYPE_NUMBER, "2"}, + {"_REV", ACPI_TYPE_INTEGER, "2"}, {"_OS_", ACPI_TYPE_STRING, ACPI_OS_NAME}, {"_GL_", ACPI_TYPE_MUTEX, "0"}, @@ -152,6 +152,13 @@ u8 acpi_gbl_ns_properties[] = }; +/* Hex to ASCII conversion table */ + +NATIVE_CHAR acpi_gbl_hex_to_ascii[] = + {'0','1','2','3','4','5','6','7', + '8','9','A','B','C','D','E','F'}; + + /****************************************************************************** * * Table globals diff --git a/drivers/acpi/common/cminit.c b/drivers/acpi/common/cminit.c index e6cfb7655..0952720f1 100644 --- a/drivers/acpi/common/cminit.c +++ b/drivers/acpi/common/cminit.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: cminit - Common ACPI subsystem initialization - * $Revision: 91 $ + * $Revision: 93 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -35,6 +35,9 @@ MODULE_NAME ("cminit") +#define ACPI_OFFSET(d,o) ((u32) &(((d *)0)->o)) +#define ACPI_FADT_OFFSET(o) ACPI_OFFSET (FADT_DESCRIPTOR, o) + /******************************************************************************* * * FUNCTION: Acpi_cm_fadt_register_error @@ -53,12 +56,13 @@ static ACPI_STATUS acpi_cm_fadt_register_error ( NATIVE_CHAR *register_name, - u32 value) + u32 value, + u32 offset) { REPORT_ERROR ( - ("Invalid FADT register value, %s=%X (FADT=%p)\n", - register_name, value, acpi_gbl_FADT)); + ("Invalid FADT value %s=%lX at offset %lX FADT=%p\n", + register_name, value, offset, acpi_gbl_FADT)); return (AE_BAD_VALUE); @@ -91,39 +95,42 @@ acpi_cm_validate_fadt ( if (acpi_gbl_FADT->pm1_evt_len < 4) { status = acpi_cm_fadt_register_error ("PM1_EVT_LEN", - (u32) acpi_gbl_FADT->pm1_evt_len); + (u32) acpi_gbl_FADT->pm1_evt_len, + ACPI_FADT_OFFSET (pm1_evt_len)); } if (!acpi_gbl_FADT->pm1_cnt_len) { - status = acpi_cm_fadt_register_error ("PM1_CNT_LEN", - 0); + status = acpi_cm_fadt_register_error ("PM1_CNT_LEN", 0, + ACPI_FADT_OFFSET (pm1_cnt_len)); } if (!ACPI_VALID_ADDRESS (acpi_gbl_FADT->Xpm1a_evt_blk.address)) { - status = acpi_cm_fadt_register_error ("PM1a_EVT_BLK", - 0); + status = acpi_cm_fadt_register_error ("X_PM1a_EVT_BLK", 0, + ACPI_FADT_OFFSET (Xpm1a_evt_blk.address)); } if (!ACPI_VALID_ADDRESS (acpi_gbl_FADT->Xpm1a_cnt_blk.address)) { - status = acpi_cm_fadt_register_error ("PM1a_CNT_BLK", - 0); + status = acpi_cm_fadt_register_error ("X_PM1a_CNT_BLK", 0, + ACPI_FADT_OFFSET (Xpm1a_cnt_blk.address)); } if (!ACPI_VALID_ADDRESS (acpi_gbl_FADT->Xpm_tmr_blk.address)) { - status = acpi_cm_fadt_register_error ("PM_TMR_BLK", - 0); + status = acpi_cm_fadt_register_error ("X_PM_TMR_BLK", 0, + ACPI_FADT_OFFSET (Xpm_tmr_blk.address)); } if ((ACPI_VALID_ADDRESS (acpi_gbl_FADT->Xpm2_cnt_blk.address) && !acpi_gbl_FADT->pm2_cnt_len)) { status = acpi_cm_fadt_register_error ("PM2_CNT_LEN", - (u32) acpi_gbl_FADT->pm2_cnt_len); + (u32) acpi_gbl_FADT->pm2_cnt_len, + ACPI_FADT_OFFSET (pm2_cnt_len)); } if (acpi_gbl_FADT->pm_tm_len < 4) { status = acpi_cm_fadt_register_error ("PM_TM_LEN", - (u32) acpi_gbl_FADT->pm_tm_len); + (u32) acpi_gbl_FADT->pm_tm_len, + ACPI_FADT_OFFSET (pm_tm_len)); } /* length of GPE blocks must be a multiple of 2 */ @@ -132,15 +139,17 @@ acpi_cm_validate_fadt ( if (ACPI_VALID_ADDRESS (acpi_gbl_FADT->Xgpe0blk.address) && (acpi_gbl_FADT->gpe0blk_len & 1)) { - status = acpi_cm_fadt_register_error ("GPE0_BLK_LEN", - (u32) acpi_gbl_FADT->gpe0blk_len); + status = acpi_cm_fadt_register_error ("(x)GPE0_BLK_LEN", + (u32) acpi_gbl_FADT->gpe0blk_len, + ACPI_FADT_OFFSET (gpe0blk_len)); } if (ACPI_VALID_ADDRESS (acpi_gbl_FADT->Xgpe1_blk.address) && (acpi_gbl_FADT->gpe1_blk_len & 1)) { - status = acpi_cm_fadt_register_error ("GPE1_BLK_LEN", - (u32) acpi_gbl_FADT->gpe1_blk_len); + status = acpi_cm_fadt_register_error ("(x)GPE1_BLK_LEN", + (u32) acpi_gbl_FADT->gpe1_blk_len, + ACPI_FADT_OFFSET (gpe1_blk_len)); } return (status); diff --git a/drivers/acpi/common/cmobject.c b/drivers/acpi/common/cmobject.c index 95e70fb14..5f73abaaf 100644 --- a/drivers/acpi/common/cmobject.c +++ b/drivers/acpi/common/cmobject.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: cmobject - ACPI object create/delete/size/cache routines - * $Revision: 34 $ + * $Revision: 35 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -434,7 +434,7 @@ acpi_cm_get_simple_object_size ( break; - case ACPI_TYPE_NUMBER: + case ACPI_TYPE_INTEGER: case ACPI_TYPE_PROCESSOR: case ACPI_TYPE_POWER: diff --git a/drivers/acpi/common/cmutils.c b/drivers/acpi/common/cmutils.c index cb3d959c8..b0ee8b4d9 100644 --- a/drivers/acpi/common/cmutils.c +++ b/drivers/acpi/common/cmutils.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: cmutils - common utility procedures - * $Revision: 21 $ + * $Revision: 23 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -647,16 +647,16 @@ acpi_cm_resolve_package_references ( if (sub_object->common.type == INTERNAL_TYPE_REFERENCE) { if (sub_object->reference.op_code == AML_ZERO_OP) { - sub_object->common.type = ACPI_TYPE_NUMBER; - sub_object->number.value = 0; + sub_object->common.type = ACPI_TYPE_INTEGER; + sub_object->integer.value = 0; } else if (sub_object->reference.op_code == AML_ONE_OP) { - sub_object->common.type = ACPI_TYPE_NUMBER; - sub_object->number.value = 1; + sub_object->common.type = ACPI_TYPE_INTEGER; + sub_object->integer.value = 1; } else if (sub_object->reference.op_code == AML_ONES_OP) { - sub_object->common.type = ACPI_TYPE_NUMBER; - sub_object->number.value = ACPI_INTEGER_MAX; + sub_object->common.type = ACPI_TYPE_INTEGER; + sub_object->integer.value = ACPI_INTEGER_MAX; } } } diff --git a/drivers/acpi/common/cmxface.c b/drivers/acpi/common/cmxface.c index fc063850c..ce87b50d2 100644 --- a/drivers/acpi/common/cmxface.c +++ b/drivers/acpi/common/cmxface.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: cmxface - External interfaces for "global" ACPI functions - * $Revision: 55 $ + * $Revision: 62 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -304,16 +304,22 @@ acpi_get_system_info ( out_buffer->length = sizeof (ACPI_SYSTEM_INFO); info_ptr = (ACPI_SYSTEM_INFO *) out_buffer->pointer; - /* TBD [Future]: need a version number, or use the version string */ - info_ptr->acpi_ca_version = 0x1234; + info_ptr->acpi_ca_version = ACPI_CA_VERSION; /* System flags (ACPI capabilities) */ info_ptr->flags = acpi_gbl_system_flags; /* Timer resolution - 24 or 32 bits */ - - info_ptr->timer_resolution = acpi_hw_pmt_resolution (); + if (!acpi_gbl_FADT) { + info_ptr->timer_resolution = 0; + } + else if (acpi_gbl_FADT->tmr_val_ext == 0) { + info_ptr->timer_resolution = 24; + } + else { + info_ptr->timer_resolution = 32; + } /* Clear the reserved fields */ diff --git a/drivers/acpi/cpu.c b/drivers/acpi/cpu.c index 00ec59da5..3e4adcee2 100644 --- a/drivers/acpi/cpu.c +++ b/drivers/acpi/cpu.c @@ -39,6 +39,9 @@ static int acpi_c3_tested = 0; static int acpi_max_c_state = 1; static int acpi_pm_tmr_len; +#define MAX_C2_LATENCY 100 +#define MAX_C3_LATENCY 1000 + /* * Clear busmaster activity flag */ @@ -251,10 +254,7 @@ static ACPI_STATUS acpi_found_cpu(ACPI_HANDLE handle, u32 level, void *ctx, void **value) { ACPI_OBJECT obj; - ACPI_CX_STATE lat[4]; - ACPI_CPU_THROTTLING_STATE throttle[ACPI_MAX_THROTTLE]; ACPI_BUFFER buf; - int i, count; buf.length = sizeof(obj); buf.pointer = &obj; @@ -273,41 +273,28 @@ acpi_found_cpu(ACPI_HANDLE handle, u32 level, void *ctx, void **value) acpi_pblk = obj.processor.pblk_address; - buf.length = sizeof(lat); - buf.pointer = lat; - if (!ACPI_SUCCESS(acpi_get_processor_cx_info(handle, &buf))) - return AE_OK; + if (acpi_fadt.plvl2_lat + && acpi_fadt.plvl2_lat <= MAX_C2_LATENCY) { + acpi_c2_exit_latency + = ACPI_MICROSEC_TO_TMR_TICKS(acpi_fadt.plvl2_lat); + acpi_c2_enter_latency + = ACPI_MICROSEC_TO_TMR_TICKS(ACPI_TMR_HZ / 1000); + acpi_max_c_state = 2; - if (lat[2].latency < MAX_CX_STATE_LATENCY) { printk(KERN_INFO "ACPI: System firmware supports: C2"); - acpi_c2_exit_latency = lat[2].latency; - acpi_max_c_state = 2; - if (lat[3].latency < MAX_CX_STATE_LATENCY) { - printk(" C3"); - acpi_c3_exit_latency = lat[3].latency; + if (acpi_fadt.plvl3_lat + && acpi_fadt.plvl3_lat <= MAX_C3_LATENCY) { + acpi_c3_exit_latency + = ACPI_MICROSEC_TO_TMR_TICKS(acpi_fadt.plvl3_lat); + acpi_c3_enter_latency + = ACPI_MICROSEC_TO_TMR_TICKS(acpi_fadt.plvl3_lat * 5); acpi_max_c_state = 3; - } - printk("\n"); - } - memset(throttle, 0, sizeof(throttle)); - buf.length = sizeof(throttle); - buf.pointer = throttle; - - if (!ACPI_SUCCESS(acpi_get_processor_throttling_info(handle, &buf))) - return AE_OK; - - for (i = 0, count = 0; i < ACPI_MAX_THROTTLE; i++) { - if (throttle[i].percent_of_clock) - count++; - } - - /* 0% throttled really doesn't count */ - count--; + printk(" C3"); + } - if (count > 0) { - DEBUG_PRINT(ACPI_INFO, ("%d throttling states\n", count)); + printk("\n"); } return AE_OK; diff --git a/drivers/acpi/dispatcher/dsfield.c b/drivers/acpi/dispatcher/dsfield.c index 1ce75dbdb..897324547 100644 --- a/drivers/acpi/dispatcher/dsfield.c +++ b/drivers/acpi/dispatcher/dsfield.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: dsfield - Dispatcher field routines - * $Revision: 29 $ + * $Revision: 31 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/dispatcher/dsmethod.c b/drivers/acpi/dispatcher/dsmethod.c index d9d73ccea..4834f47ec 100644 --- a/drivers/acpi/dispatcher/dsmethod.c +++ b/drivers/acpi/dispatcher/dsmethod.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: dsmethod - Parser/Interpreter interface - control method parsing - * $Revision: 53 $ + * $Revision: 56 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/dispatcher/dsmthdat.c b/drivers/acpi/dispatcher/dsmthdat.c index e6913fabf..dfaa2a103 100644 --- a/drivers/acpi/dispatcher/dsmthdat.c +++ b/drivers/acpi/dispatcher/dsmthdat.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: dsmthdat - control method arguments and local variables - * $Revision: 36 $ + * $Revision: 39 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -40,7 +40,7 @@ * * FUNCTION: Acpi_ds_method_data_init * - * PARAMETERS: *Obj_desc + * PARAMETERS: Walk_state - Current walk state object * * RETURN: Status * @@ -97,7 +97,7 @@ acpi_ds_method_data_init ( * * FUNCTION: Acpi_ds_method_data_delete_all * - * PARAMETERS: None + * PARAMETERS: Walk_state - Current walk state object * * RETURN: Status * @@ -153,7 +153,9 @@ acpi_ds_method_data_delete_all ( * * FUNCTION: Acpi_ds_method_data_init_args * - * PARAMETERS: None + * PARAMETERS: *Params - Pointer to a parameter list for the method + * Max_param_count - The arg count for this method + * Walk_state - Current walk state object * * RETURN: Status * @@ -214,6 +216,7 @@ acpi_ds_method_data_init_args ( * Index - Which local_var or argument to get * Entry - Pointer to where a pointer to the stack * entry is returned. + * Walk_state - Current walk state object * * RETURN: Status * @@ -276,6 +279,7 @@ acpi_ds_method_data_get_entry ( * PARAMETERS: Type - Either MTH_TYPE_LOCAL or MTH_TYPE_ARG * Index - Which local_var or argument to get * Object - Object to be inserted into the stack entry + * Walk_state - Current walk state object * * RETURN: Status * @@ -320,6 +324,7 @@ acpi_ds_method_data_set_entry ( * PARAMETERS: Type - Either MTH_TYPE_LOCAL or MTH_TYPE_ARG * Index - Which local_var or argument whose type * to get + * Walk_state - Current walk state object * * RETURN: Data type of selected Arg or Local * Used only in Exec_monadic2()/Type_op. @@ -366,6 +371,7 @@ acpi_ds_method_data_get_type ( * PARAMETERS: Type - Either MTH_TYPE_LOCAL or MTH_TYPE_ARG * Index - Which local_var or argument whose type * to get + * Walk_state - Current walk state object * * RETURN: Get the Node associated with a local or arg. * @@ -418,7 +424,8 @@ acpi_ds_method_data_get_nte ( * * PARAMETERS: Type - Either MTH_TYPE_LOCAL or MTH_TYPE_ARG * Index - Which local_var or argument to get - * *Dest_desc - Descriptor into which selected Arg + * Walk_state - Current walk state object + * *Dest_desc - Ptr to Descriptor into which selected Arg * or Local value should be copied * * RETURN: Status @@ -474,10 +481,12 @@ acpi_ds_method_data_get_value ( switch (type) { case MTH_TYPE_ARG: + return (AE_AML_UNINITIALIZED_ARG); break; case MTH_TYPE_LOCAL: + return (AE_AML_UNINITIALIZED_LOCAL); break; } @@ -502,6 +511,7 @@ acpi_ds_method_data_get_value ( * * PARAMETERS: Type - Either MTH_TYPE_LOCAL or MTH_TYPE_ARG * Index - Which local_var or argument to delete + * Walk_state - Current walk state object * * RETURN: Status * @@ -548,7 +558,6 @@ acpi_ds_method_data_delete_value ( * Decrement the reference count by one to balance the * increment when the object was stored in the slot. */ - acpi_cm_remove_reference (object); } @@ -563,18 +572,14 @@ acpi_ds_method_data_delete_value ( * * PARAMETERS: Type - Either MTH_TYPE_LOCAL or MTH_TYPE_ARG * Index - Which local_var or argument to set - * *Src_desc - Value to be stored - * *Dest_desc - Descriptor into which *Src_desc - * can be copied, or NULL if one must - * be allocated for the purpose. If - * provided, this descriptor will be - * used for the new value. + * Src_desc - Value to be stored + * Walk_state - Current walk state * * RETURN: Status * * DESCRIPTION: Store a value in an Arg or Local. The Src_desc is installed * as the new value for the Arg or Local and the reference count - * is incremented. + * for Src_desc is incremented. * ******************************************************************************/ @@ -644,7 +649,6 @@ acpi_ds_method_data_set_value ( * Store this object into the Node * (do the indirect store) */ - status = acpi_ns_attach_object ((ACPI_NAMESPACE_NODE *) *entry, src_desc, src_desc->common.type); return (status); @@ -652,10 +656,18 @@ acpi_ds_method_data_set_value ( /* - * Otherwise, just delete the existing object - * before storing the new one + * Perform "Implicit conversion" of the new object to the type of the + * existing object */ + status = acpi_aml_convert_to_target_type ((*entry)->common.type, &src_desc, walk_state); + if (ACPI_FAILURE (status)) { + goto cleanup; + } + /* + * Delete the existing object + * before storing the new one + */ acpi_ds_method_data_delete_value (type, index, walk_state); } @@ -666,7 +678,6 @@ acpi_ds_method_data_set_value ( * Install the new object in the stack entry * (increments the object reference count by one) */ - status = acpi_ds_method_data_set_entry (type, index, src_desc, walk_state); if (ACPI_FAILURE (status)) { goto cleanup; diff --git a/drivers/acpi/dispatcher/dsobject.c b/drivers/acpi/dispatcher/dsobject.c index 042cc4a80..0a6c88b4d 100644 --- a/drivers/acpi/dispatcher/dsobject.c +++ b/drivers/acpi/dispatcher/dsobject.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: dsobject - Dispatcher object management routines - * $Revision: 53 $ + * $Revision: 56 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -241,14 +241,14 @@ acpi_ds_init_object_from_op ( /* We are expecting a number */ - if (arg_desc->common.type != ACPI_TYPE_NUMBER) { + if (arg_desc->common.type != ACPI_TYPE_INTEGER) { acpi_cm_remove_reference (arg_desc); return (AE_TYPE); } /* Get the value, delete the internal object */ - (*obj_desc)->buffer.length = (u32) arg_desc->number.value; + (*obj_desc)->buffer.length = (u32) arg_desc->integer.value; acpi_cm_remove_reference (arg_desc); /* Allocate the buffer */ @@ -304,8 +304,8 @@ acpi_ds_init_object_from_op ( status = acpi_ds_build_internal_object (walk_state, op, obj_desc); break; - case ACPI_TYPE_NUMBER: - (*obj_desc)->number.value = op->value.integer; + case ACPI_TYPE_INTEGER: + (*obj_desc)->integer.value = op->value.integer; break; @@ -411,18 +411,20 @@ acpi_ds_build_internal_simple_obj ( acpi_ns_externalize_name (ACPI_UINT32_MAX, op->value.string, &length, &name); if (name) { - REPORT_WARNING (("Reference %s AML %X not found\n", + REPORT_WARNING (("Reference %s at AML %X not found\n", name, op->aml_offset)); acpi_cm_free (name); } else { - REPORT_WARNING (("Reference %s AML %X not found\n", + REPORT_WARNING (("Reference %s at AML %X not found\n", op->value.string, op->aml_offset)); } *obj_desc_ptr = NULL; } - return (status); + else { + return (status); + } } } diff --git a/drivers/acpi/dispatcher/dsopcode.c b/drivers/acpi/dispatcher/dsopcode.c index 21e15bc26..ba06cf4db 100644 --- a/drivers/acpi/dispatcher/dsopcode.c +++ b/drivers/acpi/dispatcher/dsopcode.c @@ -2,12 +2,12 @@ * * Module Name: dsopcode - Dispatcher Op Region support and handling of * "control" opcodes - * $Revision: 28 $ + * $Revision: 30 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -355,7 +355,7 @@ acpi_ds_eval_field_unit_operands ( } - offset = (u32) off_desc->number.value; + offset = (u32) off_desc->integer.value; /* @@ -427,7 +427,7 @@ acpi_ds_eval_field_unit_operands ( /* Offset is in bits, count is in bits */ bit_offset = offset; - bit_count = (u16) cnt_desc->number.value; + bit_count = (u16) cnt_desc->integer.value; break; @@ -586,7 +586,7 @@ acpi_ds_eval_region_operands ( */ operand_desc = walk_state->operands[walk_state->num_operands - 1]; - obj_desc->region.length = (u32) operand_desc->number.value; + obj_desc->region.length = (u32) operand_desc->integer.value; acpi_cm_remove_reference (operand_desc); /* @@ -595,7 +595,7 @@ acpi_ds_eval_region_operands ( */ operand_desc = walk_state->operands[walk_state->num_operands - 2]; - obj_desc->region.address = (ACPI_PHYSICAL_ADDRESS) operand_desc->number.value; + obj_desc->region.address = (ACPI_PHYSICAL_ADDRESS) operand_desc->integer.value; acpi_cm_remove_reference (operand_desc); diff --git a/drivers/acpi/dispatcher/dsutils.c b/drivers/acpi/dispatcher/dsutils.c index 2efa43e51..d31464840 100644 --- a/drivers/acpi/dispatcher/dsutils.c +++ b/drivers/acpi/dispatcher/dsutils.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: dsutils - Dispatcher utilities - * $Revision: 50 $ + * $Revision: 52 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -566,7 +566,7 @@ acpi_ds_map_opcode_to_data_type ( case AML_WORD_OP: case AML_DWORD_OP: - data_type = ACPI_TYPE_NUMBER; + data_type = ACPI_TYPE_INTEGER; break; diff --git a/drivers/acpi/dispatcher/dswexec.c b/drivers/acpi/dispatcher/dswexec.c index 1f7e329ad..a2e1b73d3 100644 --- a/drivers/acpi/dispatcher/dswexec.c +++ b/drivers/acpi/dispatcher/dswexec.c @@ -2,12 +2,12 @@ * * Module Name: dswexec - Dispatcher method execution callbacks; * dispatch to interpreter. - * $Revision: 50 $ + * $Revision: 55 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -46,7 +46,7 @@ * * RETURN: Status * - * DESCRIPTION: + * DESCRIPTION: Get the result of a predicate evaluation * ****************************************************************************/ @@ -93,22 +93,22 @@ acpi_ds_get_predicate_value ( * be a number */ - if (obj_desc->common.type != ACPI_TYPE_NUMBER) { + if (obj_desc->common.type != ACPI_TYPE_INTEGER) { status = AE_AML_OPERAND_TYPE; goto cleanup; } - /* TBD: 64/32-bit */ + /* Truncate the predicate to 32-bits if necessary */ - obj_desc->number.value &= (UINT64) 0x00000000FFFFFFFF; + acpi_aml_truncate_for32bit_table (obj_desc, walk_state); /* * Save the result of the predicate evaluation on * the control stack */ - if (obj_desc->number.value) { + if (obj_desc->integer.value) { walk_state->control_state->common.value = TRUE; } @@ -330,6 +330,8 @@ acpi_ds_exec_end_op ( walk_state->num_operands = 0; walk_state->return_desc = NULL; + walk_state->op_info = op_info; + walk_state->opcode = opcode; /* Call debugger for single step support (DEBUG build only) */ diff --git a/drivers/acpi/dispatcher/dswload.c b/drivers/acpi/dispatcher/dswload.c index b3f1dc062..300d02b7f 100644 --- a/drivers/acpi/dispatcher/dswload.c +++ b/drivers/acpi/dispatcher/dswload.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: dswload - Dispatcher namespace load callbacks - * $Revision: 24 $ + * $Revision: 26 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/dispatcher/dswscope.c b/drivers/acpi/dispatcher/dswscope.c index 11b6a6cb9..332ab4ffd 100644 --- a/drivers/acpi/dispatcher/dswscope.c +++ b/drivers/acpi/dispatcher/dswscope.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: dswscope - Scope stack manipulation - * $Revision: 40 $ + * $Revision: 42 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/dispatcher/dswstate.c b/drivers/acpi/dispatcher/dswstate.c index a15a6f5f7..a436945df 100644 --- a/drivers/acpi/dispatcher/dswstate.c +++ b/drivers/acpi/dispatcher/dswstate.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: dswstate - Dispatcher parse tree walk management routines - * $Revision: 36 $ + * $Revision: 38 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -37,31 +37,6 @@ /******************************************************************************* * - * FUNCTION: Acpi_ds_result_stack_clear - * - * PARAMETERS: Walk_state - Current Walk state - * - * RETURN: Status - * - * DESCRIPTION: Reset this walk's result stack pointers to zero, thus setting - * the stack to zero. - * - ******************************************************************************/ - -ACPI_STATUS -xxx_acpi_ds_result_stack_clear ( - ACPI_WALK_STATE *walk_state) -{ -/* - Walk_state->Num_results = 0; - Walk_state->Current_result = 0; -*/ - return (AE_OK); -} - - -/******************************************************************************* - * * FUNCTION: Acpi_ds_result_insert * * PARAMETERS: Object - Object to push @@ -135,7 +110,7 @@ acpi_ds_result_remove ( /* Check for a valid result object */ if (!state->results.obj_desc [index]) { - return (AE_AML_NO_OPERAND); + return (AE_AML_NO_RETURN_VALUE); } /* Remove the object */ @@ -179,7 +154,7 @@ acpi_ds_result_pop ( if (!state->results.num_results) { - return (AE_STACK_UNDERFLOW); + return (AE_AML_NO_RETURN_VALUE); } /* Remove top element */ @@ -198,12 +173,12 @@ acpi_ds_result_pop ( } - return (AE_STACK_UNDERFLOW); + return (AE_AML_NO_RETURN_VALUE); } /******************************************************************************* * - * FUNCTION: Acpi_ds_result_pop + * FUNCTION: Acpi_ds_result_pop_from_bottom * * PARAMETERS: Object - Where to return the popped object * Walk_state - Current Walk state @@ -231,7 +206,7 @@ acpi_ds_result_pop_from_bottom ( if (!state->results.num_results) { - return (AE_STACK_UNDERFLOW); + return (AE_AML_NO_RETURN_VALUE); } /* Remove Bottom element */ @@ -250,7 +225,7 @@ acpi_ds_result_pop_from_bottom ( /* Check for a valid result object */ if (!*object) { - return (AE_AML_NO_OPERAND); + return (AE_AML_NO_RETURN_VALUE); } @@ -260,15 +235,14 @@ acpi_ds_result_pop_from_bottom ( /******************************************************************************* * - * FUNCTION: Acpi_ds_result_pop + * FUNCTION: Acpi_ds_result_push * * PARAMETERS: Object - Where to return the popped object * Walk_state - Current Walk state * * RETURN: Status * - * DESCRIPTION: Pop an object off the bottom of this walk's result stack. In - * other words, this is a FIFO. + * DESCRIPTION: Push an object onto the current result stack * ******************************************************************************/ @@ -282,7 +256,7 @@ acpi_ds_result_push ( state = walk_state->results; if (!state) { - return (AE_OK); + return (AE_AML_INTERNAL); } if (state->results.num_results == OBJ_NUM_OPERANDS) { @@ -719,6 +693,7 @@ acpi_ds_create_walk_state ( ACPI_WALK_LIST *walk_list) { ACPI_WALK_STATE *walk_state; + ACPI_STATUS status; acpi_cm_acquire_mutex (ACPI_MTX_CACHES); @@ -736,7 +711,7 @@ acpi_ds_create_walk_state ( acpi_gbl_walk_state_cache_depth--; acpi_cm_release_mutex (ACPI_MTX_CACHES); - } + } else { /* The cache is empty, create a new object */ @@ -762,6 +737,14 @@ acpi_ds_create_walk_state ( acpi_ds_method_data_init (walk_state); #endif + /* Create an initial result stack entry */ + + status = acpi_ds_result_stack_push (walk_state); + if (ACPI_FAILURE (status)) { + return (NULL); + } + + /* Put the new state at the head of the walk list */ acpi_ds_push_walk_state (walk_state, walk_list); diff --git a/drivers/acpi/driver.c b/drivers/acpi/driver.c index 222598120..241528fc6 100644 --- a/drivers/acpi/driver.c +++ b/drivers/acpi/driver.c @@ -55,7 +55,7 @@ struct acpi_run_entry static spinlock_t acpi_event_lock = SPIN_LOCK_UNLOCKED; static volatile u32 acpi_event_status = 0; -static volatile acpi_sstate_t acpi_event_state = ACPI_S0; +static volatile acpi_sstate_t acpi_event_state = ACPI_STATE_S0; static DECLARE_WAIT_QUEUE_HEAD(acpi_event_wait); static volatile int acpi_thread_pid = -1; @@ -243,7 +243,7 @@ acpi_do_event(ctl_table * ctl, /* * Enter system sleep state */ -static int +/*static int acpi_do_sleep(ctl_table * ctl, int write, struct file *file, @@ -257,13 +257,13 @@ acpi_do_sleep(ctl_table * ctl, } } else { - int status = acpi_enter_sx(ACPI_S1); + int status = acpi_enter_sx(ACPI_STATE_S1); if (status) return status; } file->f_pos += *len; return 0; -} +}*/ /* @@ -382,7 +382,8 @@ static struct ctl_table acpi_table[] = &acpi_c3_enter_latency, sizeof(acpi_c3_enter_latency), 0644, NULL, &acpi_do_ulong}, - {ACPI_SLEEP, "sleep", NULL, 0, 0600, NULL, &acpi_do_sleep}, +/* until it actually works */ +/* {ACPI_SLEEP, "sleep", NULL, 0, 0600, NULL, &acpi_do_sleep},*/ {ACPI_EVENT, "event", NULL, 0, 0400, NULL, &acpi_do_event}, @@ -415,7 +416,9 @@ static struct ctl_table acpi_dir_table[] = static int acpi_thread(void *context) { - ACPI_PHYSICAL_ADDRESS rsdp_phys; + ACPI_PHYSICAL_ADDRESS rsdp_phys; + ACPI_BUFFER buffer; + ACPI_SYSTEM_INFO sys_info; /* * initialize @@ -437,8 +440,6 @@ acpi_thread(void *context) rsdp_phys = efi.acpi; #endif - printk(KERN_ERR "ACPI: System description tables found\n"); - if (!ACPI_SUCCESS(acpi_find_and_load_tables(rsdp_phys))) return -ENODEV; @@ -448,6 +449,17 @@ acpi_thread(void *context) return -ENODEV; } + buffer.length = sizeof(sys_info); + buffer.pointer = &sys_info; + + if (!ACPI_SUCCESS (acpi_get_system_info(&buffer))) { + printk(KERN_ERR "ACPI: Could not get system info\n"); + acpi_terminate(); + return -ENODEV; + } + + printk(KERN_INFO "ACPI: Core Subsystem version [%x]\n", sys_info.acpi_ca_version); + if (!ACPI_SUCCESS(acpi_enable_subsystem(ACPI_FULL_INITIALIZATION))) { printk(KERN_ERR "ACPI: Subsystem enable failed\n"); acpi_terminate(); diff --git a/drivers/acpi/ec.c b/drivers/acpi/ec.c index 8f6f61e36..5e42683c6 100644 --- a/drivers/acpi/ec.c +++ b/drivers/acpi/ec.c @@ -270,25 +270,6 @@ ec_transaction ( return_ACPI_STATUS(status); } -static ACPI_STATUS -ec_space_setup ( - ACPI_HANDLE region_handle, - UINT32 function, - void *handler_context, - void **return_context) -{ - // TODO: What is this function for? - /* - * The ec object is in the handler context and is needed - * when calling the ec_space_handler. - */ - *return_context = handler_context; - - return AE_OK; -} - - - static void ec_query_handler ( @@ -419,8 +400,6 @@ ec_region_setup ( { FUNCTION_TRACE("acpi_ec_region_setup"); - printk("acpi_ec_region_setup\n"); - if (function == ACPI_REGION_DEACTIVATE) { if (*region_context) @@ -556,10 +535,10 @@ found_ec( buf.length = sizeof(obj); buf.pointer = &obj; if (!ACPI_SUCCESS(acpi_evaluate_object(handle, "_GPE", NULL, &buf)) - || obj.type != ACPI_TYPE_NUMBER) + || obj.type != ACPI_TYPE_INTEGER) return AE_OK; - ec_cxt->gpe_bit = obj.number.value; + ec_cxt->gpe_bit = obj.integer.value; /* determine if we need the Global Lock when accessing */ buf.length = sizeof(obj); @@ -568,12 +547,12 @@ found_ec( status = acpi_evaluate_object(handle, "_GLK", NULL, &buf); if (status == AE_NOT_FOUND) ec_cxt->need_global_lock = 0; - else if (!ACPI_SUCCESS(status) || obj.type != ACPI_TYPE_NUMBER) { + else if (!ACPI_SUCCESS(status) || obj.type != ACPI_TYPE_INTEGER) { DEBUG_PRINT(ACPI_ERROR, ("_GLK failed\n")); return AE_OK; } - ec_cxt->need_global_lock = obj.number.value; + ec_cxt->need_global_lock = obj.integer.value; printk(KERN_INFO "ACPI: found EC @ (0x%02x,0x%02x,gpe %d GL %d)\n", ec_cxt->data_port, ec_cxt->status_port, ec_cxt->gpe_bit, diff --git a/drivers/acpi/events/evevent.c b/drivers/acpi/events/evevent.c index d5ce143a8..afdd477db 100644 --- a/drivers/acpi/events/evevent.c +++ b/drivers/acpi/events/evevent.c @@ -2,12 +2,12 @@ * * Module Name: evevent - Fixed and General Purpose Acpi_event * handling and dispatch - * $Revision: 32 $ + * $Revision: 33 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/events/evmisc.c b/drivers/acpi/events/evmisc.c index a52f2dc3d..cae4a44b2 100644 --- a/drivers/acpi/events/evmisc.c +++ b/drivers/acpi/events/evmisc.c @@ -2,12 +2,12 @@ * * Module Name: evmisc - ACPI device notification handler dispatch * and ACPI Global Lock support - * $Revision: 20 $ + * $Revision: 22 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/events/evregion.c b/drivers/acpi/events/evregion.c index 53cae6392..071639a27 100644 --- a/drivers/acpi/events/evregion.c +++ b/drivers/acpi/events/evregion.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: evregion - ACPI Address_space (Op_region) handler dispatch - * $Revision: 93 $ + * $Revision: 94 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -122,8 +122,8 @@ acpi_ev_execute_reg_method ( u32 function) { ACPI_OPERAND_OBJECT *params[3]; - ACPI_OPERAND_OBJECT space_iD_obj; - ACPI_OPERAND_OBJECT function_obj; + ACPI_OPERAND_OBJECT space_id_desc; + ACPI_OPERAND_OBJECT function_desc; ACPI_STATUS status; @@ -141,24 +141,24 @@ acpi_ev_execute_reg_method ( * Passed as a parameter */ - acpi_cm_init_static_object (&space_iD_obj); - acpi_cm_init_static_object (&function_obj); + acpi_cm_init_static_object (&space_id_desc); + acpi_cm_init_static_object (&function_desc); /* * Method requires two parameters. */ - params [0] = &space_iD_obj; - params [1] = &function_obj; + params [0] = &space_id_desc; + params [1] = &function_desc; params [2] = NULL; /* * Set up the parameter objects */ - space_iD_obj.common.type = ACPI_TYPE_NUMBER; - space_iD_obj.number.value = region_obj->region.space_id; + space_id_desc.common.type = ACPI_TYPE_INTEGER; + space_id_desc.integer.value = region_obj->region.space_id; - function_obj.common.type = ACPI_TYPE_NUMBER; - function_obj.number.value = function; + function_desc.common.type = ACPI_TYPE_INTEGER; + function_desc.integer.value = function; /* * Execute the method, no return value diff --git a/drivers/acpi/events/evrgnini.c b/drivers/acpi/events/evrgnini.c index 92e5f198f..eb5e2033e 100644 --- a/drivers/acpi/events/evrgnini.c +++ b/drivers/acpi/events/evrgnini.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: evrgnini- ACPI Address_space (Op_region) init - * $Revision: 31 $ + * $Revision: 33 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/events/evsci.c b/drivers/acpi/events/evsci.c index 02320e93c..369d2f184 100644 --- a/drivers/acpi/events/evsci.c +++ b/drivers/acpi/events/evsci.c @@ -2,12 +2,12 @@ * * Module Name: evsci - System Control Interrupt configuration and * legacy to ACPI mode state transition functions - * $Revision: 67 $ + * $Revision: 69 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/events/evxface.c b/drivers/acpi/events/evxface.c index c3bbad0ff..f0e62934f 100644 --- a/drivers/acpi/events/evxface.c +++ b/drivers/acpi/events/evxface.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: evxface - External interfaces for ACPI events - * $Revision: 97 $ + * $Revision: 101 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -187,26 +187,25 @@ acpi_install_notify_handler ( return (AE_BAD_PARAMETER); } - /* Convert and validate the device handle */ - acpi_cm_acquire_mutex (ACPI_MTX_NAMESPACE); + /* Convert and validate the device handle */ + device_node = acpi_ns_convert_handle_to_entry (device); if (!device_node) { status = AE_BAD_PARAMETER; goto unlock_and_exit; } - /* - * Support for global notify handlers. These handlers are invoked for - * every notifiy of the type specifiec + * Root Object: + * ------------ + * Registering a notify handler on the root object indicates that the + * caller wishes to receive notifications for all objects. Note that + * only one <external> global handler can be regsitered (per notify type). */ - if (device == ACPI_ROOT_OBJECT) { - /* - * Make sure the handler is not already installed. - */ + /* Make sure the handler is not already installed */ if (((handler_type == ACPI_SYSTEM_NOTIFY) && acpi_gbl_sys_notify.handler) || @@ -222,94 +221,89 @@ acpi_install_notify_handler ( acpi_gbl_sys_notify.handler = handler; acpi_gbl_sys_notify.context = context; } - - else { + else /* ACPI_DEVICE_NOTIFY */ { acpi_gbl_drv_notify.node = device_node; acpi_gbl_drv_notify.handler = handler; acpi_gbl_drv_notify.context = context; } - /* Global notify handler installed */ - - goto unlock_and_exit; } - /* - * These are the ONLY objects that can receive ACPI notifications + * Other Objects: + * -------------- + * Caller will only receive notifications specific to the target object. + * Note that only certain object types can receive notifications. */ - - if ((device_node->type != ACPI_TYPE_DEVICE) && - (device_node->type != ACPI_TYPE_PROCESSOR) && - (device_node->type != ACPI_TYPE_POWER) && - (device_node->type != ACPI_TYPE_THERMAL)) - { - status = AE_BAD_PARAMETER; - goto unlock_and_exit; - } - - /* Check for an existing internal object */ - - obj_desc = acpi_ns_get_attached_object ((ACPI_HANDLE) device_node); - if (obj_desc) { + else { /* - * The object exists. - * Make sure the handler is not already installed. + * These are the ONLY objects that can receive ACPI notifications */ - - if (((handler_type == ACPI_SYSTEM_NOTIFY) && - obj_desc->device.sys_handler) || - ((handler_type == ACPI_DEVICE_NOTIFY) && - obj_desc->device.drv_handler)) + if ((device_node->type != ACPI_TYPE_DEVICE) && + (device_node->type != ACPI_TYPE_PROCESSOR) && + (device_node->type != ACPI_TYPE_POWER) && + (device_node->type != ACPI_TYPE_THERMAL)) { - status = AE_EXIST; + status = AE_BAD_PARAMETER; goto unlock_and_exit; } - } - else { - /* Create a new object */ + /* Check for an existing internal object */ - obj_desc = acpi_cm_create_internal_object (device_node->type); - if (!obj_desc) { - status = AE_NO_MEMORY; - goto unlock_and_exit; + obj_desc = acpi_ns_get_attached_object ((ACPI_HANDLE) device_node); + if (obj_desc) { + + /* Object exists - make sure there's no handler */ + + if (((handler_type == ACPI_SYSTEM_NOTIFY) && + obj_desc->device.sys_handler) || + ((handler_type == ACPI_DEVICE_NOTIFY) && + obj_desc->device.drv_handler)) + { + status = AE_EXIST; + goto unlock_and_exit; + } } - /* Attach new object to the Node */ + else { + /* Create a new object */ - status = acpi_ns_attach_object (device, obj_desc, (u8) device_node->type); + obj_desc = acpi_cm_create_internal_object (device_node->type); + if (!obj_desc) { + status = AE_NO_MEMORY; + goto unlock_and_exit; + } - if (ACPI_FAILURE (status)) { - goto unlock_and_exit; - } - } + /* Attach new object to the Node */ + status = acpi_ns_attach_object (device, obj_desc, (u8) device_node->type); - /* - * If we get here, we know that there is no handler installed - * so let's party - */ - notify_obj = acpi_cm_create_internal_object (INTERNAL_TYPE_NOTIFY); - if (!notify_obj) { - status = AE_NO_MEMORY; - goto unlock_and_exit; - } + if (ACPI_FAILURE (status)) { + goto unlock_and_exit; + } + } - notify_obj->notify_handler.node = device_node; - notify_obj->notify_handler.handler = handler; - notify_obj->notify_handler.context = context; + /* Install the handler */ + notify_obj = acpi_cm_create_internal_object (INTERNAL_TYPE_NOTIFY); + if (!notify_obj) { + status = AE_NO_MEMORY; + goto unlock_and_exit; + } - if (handler_type == ACPI_SYSTEM_NOTIFY) { - obj_desc->device.sys_handler = notify_obj; - } + notify_obj->notify_handler.node = device_node; + notify_obj->notify_handler.handler = handler; + notify_obj->notify_handler.context = context; - else { - obj_desc->device.drv_handler = notify_obj; - } + if (handler_type == ACPI_SYSTEM_NOTIFY) { + obj_desc->device.sys_handler = notify_obj; + } + else /* ACPI_DEVICE_NOTIFY */ { + obj_desc->device.drv_handler = notify_obj; + } + } unlock_and_exit: acpi_cm_release_mutex (ACPI_MTX_NAMESPACE); @@ -343,7 +337,6 @@ acpi_remove_notify_handler ( ACPI_NAMESPACE_NODE *device_node; ACPI_STATUS status = AE_OK; - /* Parameter validation */ if ((!handler) || @@ -363,63 +356,92 @@ acpi_remove_notify_handler ( } /* - * These are the ONLY objects that can receive ACPI notifications + * Root Object: + * ------------ */ + if (device == ACPI_ROOT_OBJECT) { - if ((device_node->type != ACPI_TYPE_DEVICE) && - (device_node->type != ACPI_TYPE_PROCESSOR) && - (device_node->type != ACPI_TYPE_POWER) && - (device_node->type != ACPI_TYPE_THERMAL)) - { - status = AE_BAD_PARAMETER; - goto unlock_and_exit; - } - - /* Check for an existing internal object */ + if (((handler_type == ACPI_SYSTEM_NOTIFY) && + !acpi_gbl_sys_notify.handler) || + ((handler_type == ACPI_DEVICE_NOTIFY) && + !acpi_gbl_drv_notify.handler)) + { + status = AE_NOT_EXIST; + goto unlock_and_exit; + } - obj_desc = acpi_ns_get_attached_object ((ACPI_HANDLE) device_node); - if (!obj_desc) { - status = AE_NOT_EXIST; - goto unlock_and_exit; + if (handler_type == ACPI_SYSTEM_NOTIFY) { + acpi_gbl_sys_notify.node = NULL; + acpi_gbl_sys_notify.handler = NULL; + acpi_gbl_sys_notify.context = NULL; + } + else { + acpi_gbl_drv_notify.node = NULL; + acpi_gbl_drv_notify.handler = NULL; + acpi_gbl_drv_notify.context = NULL; + } } /* - * The object exists. - * - * Make sure the handler is installed. + * Other Objects: + * -------------- */ - - if (handler_type == ACPI_SYSTEM_NOTIFY) { - notify_obj = obj_desc->device.sys_handler; - } else { - notify_obj = obj_desc->device.drv_handler; - } + /* + * These are the ONLY objects that can receive ACPI notifications + */ + if ((device_node->type != ACPI_TYPE_DEVICE) && + (device_node->type != ACPI_TYPE_PROCESSOR) && + (device_node->type != ACPI_TYPE_POWER) && + (device_node->type != ACPI_TYPE_THERMAL)) + { + status = AE_BAD_PARAMETER; + goto unlock_and_exit; + } - if ((!notify_obj) || - (notify_obj->notify_handler.handler != handler)) - { - status = AE_BAD_PARAMETER; - goto unlock_and_exit; - } + /* Check for an existing internal object */ - /* - * Now we can remove the handler - */ - if (handler_type == ACPI_SYSTEM_NOTIFY) { - obj_desc->device.sys_handler = NULL; - } - else { - obj_desc->device.drv_handler = NULL; + obj_desc = acpi_ns_get_attached_object ((ACPI_HANDLE) device_node); + if (!obj_desc) { + status = AE_NOT_EXIST; + goto unlock_and_exit; + } + + /* Object exists - make sure there's an existing handler */ + + if (handler_type == ACPI_SYSTEM_NOTIFY) { + notify_obj = obj_desc->device.sys_handler; + } + else { + notify_obj = obj_desc->device.drv_handler; + } + + if ((!notify_obj) || + (notify_obj->notify_handler.handler != handler)) + { + status = AE_BAD_PARAMETER; + goto unlock_and_exit; + } + + /* Remove the handler */ + + if (handler_type == ACPI_SYSTEM_NOTIFY) { + obj_desc->device.sys_handler = NULL; + } + else { + obj_desc->device.drv_handler = NULL; + } + + acpi_cm_remove_reference (notify_obj); } - acpi_cm_remove_reference (notify_obj); unlock_and_exit: acpi_cm_release_mutex (ACPI_MTX_NAMESPACE); return (status); } + /****************************************************************************** * * FUNCTION: Acpi_install_gpe_handler diff --git a/drivers/acpi/events/evxfevnt.c b/drivers/acpi/events/evxfevnt.c index 5b7652e52..9864301a3 100644 --- a/drivers/acpi/events/evxfevnt.c +++ b/drivers/acpi/events/evxfevnt.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: evxfevnt - External Interfaces, ACPI event disable/enable - * $Revision: 26 $ + * $Revision: 28 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/events/evxfregn.c b/drivers/acpi/events/evxfregn.c index 71116cfc8..e54bbab24 100644 --- a/drivers/acpi/events/evxfregn.c +++ b/drivers/acpi/events/evxfregn.c @@ -2,12 +2,12 @@ * * Module Name: evxfregn - External Interfaces, ACPI Operation Regions and * Address Spaces. - * $Revision: 26 $ + * $Revision: 27 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/hardware/hwacpi.c b/drivers/acpi/hardware/hwacpi.c index d2154a1a2..304d3dec6 100644 --- a/drivers/acpi/hardware/hwacpi.c +++ b/drivers/acpi/hardware/hwacpi.c @@ -1,12 +1,13 @@ + /****************************************************************************** * - * Module Name: hwacpi - ACPI hardware functions - mode and timer - * $Revision: 34 $ + * Module Name: hwacpi - ACPI Hardware Initialization/Mode Interface + * $Revision: 36 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -304,48 +305,3 @@ acpi_hw_get_mode_capabilities (void) } -/****************************************************************************** - * - * FUNCTION: Acpi_hw_pmt_ticks - * - * PARAMETERS: none - * - * RETURN: Current value of the ACPI PMT (timer) - * - * DESCRIPTION: Obtains current value of ACPI PMT - * - ******************************************************************************/ - -u32 -acpi_hw_pmt_ticks (void) -{ - u32 ticks; - - ticks = acpi_os_in32 ((ACPI_IO_ADDRESS) ACPI_GET_ADDRESS (acpi_gbl_FADT->Xpm_tmr_blk.address)); - - return (ticks); -} - - -/****************************************************************************** - * - * FUNCTION: Acpi_hw_pmt_resolution - * - * PARAMETERS: none - * - * RETURN: Number of bits of resolution in the PMT (either 24 or 32) - * - * DESCRIPTION: Obtains resolution of the ACPI PMT (either 24bit or 32bit) - * - ******************************************************************************/ - -u32 -acpi_hw_pmt_resolution (void) -{ - if (0 == acpi_gbl_FADT->tmr_val_ext) { - return (24); - } - - return (32); -} - diff --git a/drivers/acpi/hardware/hwgpe.c b/drivers/acpi/hardware/hwgpe.c index 2b413fac8..b7b777df6 100644 --- a/drivers/acpi/hardware/hwgpe.c +++ b/drivers/acpi/hardware/hwgpe.c @@ -1,12 +1,13 @@ + /****************************************************************************** * * Module Name: hwgpe - Low level GPE enable/disable/clear functions - * $Revision: 25 $ + * $Revision: 28 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/hardware/hwregs.c b/drivers/acpi/hardware/hwregs.c index 77b6a1c8c..fc96c51fb 100644 --- a/drivers/acpi/hardware/hwregs.c +++ b/drivers/acpi/hardware/hwregs.c @@ -3,12 +3,12 @@ * * Module Name: hwregs - Read/write access functions for the various ACPI * control and status registers. - * $Revision: 86 $ + * $Revision: 88 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -37,7 +37,7 @@ /* This matches the #defines in actypes.h. */ NATIVE_CHAR *sleep_state_table[] = {"\\_S0_","\\_S1_","\\_S2_","\\_S3_", - "\\_S4_","\\_S4_b","\\_S5_"}; + "\\_S4_","\\_S5_","\\_S4_b"}; /******************************************************************************* @@ -53,7 +53,7 @@ NATIVE_CHAR *sleep_state_table[] = {"\\_S0_","\\_S1_","\\_S2_"," * ******************************************************************************/ -static u32 +u32 acpi_hw_get_bit_shift ( u32 mask) { @@ -185,9 +185,9 @@ acpi_hw_obtain_sleep_type_register_data ( } else if (((obj_desc->package.elements[0])->common.type != - ACPI_TYPE_NUMBER) || + ACPI_TYPE_INTEGER) || ((obj_desc->package.elements[1])->common.type != - ACPI_TYPE_NUMBER)) + ACPI_TYPE_INTEGER)) { /* Must have two */ @@ -199,9 +199,9 @@ acpi_hw_obtain_sleep_type_register_data ( /* * Valid _Sx_ package size, type, and value */ - *slp_typ_a = (u8) (obj_desc->package.elements[0])->number.value; + *slp_typ_a = (u8) (obj_desc->package.elements[0])->integer.value; - *slp_typ_b = (u8) (obj_desc->package.elements[1])->number.value; + *slp_typ_b = (u8) (obj_desc->package.elements[1])->integer.value; } @@ -581,13 +581,8 @@ acpi_hw_register_read ( case PM1_CONTROL: /* 16-bit access */ - if (register_id != SLP_TYPE_B) { - value |= acpi_hw_low_level_read (16, &acpi_gbl_FADT->Xpm1a_cnt_blk, 0); - } - - if (register_id != SLP_TYPE_A) { - value |= acpi_hw_low_level_read (16, &acpi_gbl_FADT->Xpm1b_cnt_blk, 0); - } + value = acpi_hw_low_level_read (16, &acpi_gbl_FADT->Xpm1a_cnt_blk, 0); + value |= acpi_hw_low_level_read (16, &acpi_gbl_FADT->Xpm1b_cnt_blk, 0); break; @@ -696,30 +691,20 @@ acpi_hw_register_write ( case PM1_CONTROL: /* 16-bit access */ - /* - * If SLP_TYP_A or SLP_TYP_B, only write to one reg block. - * Otherwise, write to both. - */ - if (register_id == SLP_TYPE_A) { - acpi_hw_low_level_write (16, value, &acpi_gbl_FADT->Xpm1a_cnt_blk, 0); - } - else if (register_id == SLP_TYPE_B) { - acpi_hw_low_level_write (16, value, &acpi_gbl_FADT->Xpm1b_cnt_blk, 0); - } - else { - /* disable/re-enable interrupts if sleeping */ - if (register_id == SLP_EN) { - disable(); - } + acpi_hw_low_level_write (16, value, &acpi_gbl_FADT->Xpm1a_cnt_blk, 0); + acpi_hw_low_level_write (16, value, &acpi_gbl_FADT->Xpm1b_cnt_blk, 0); + break; - acpi_hw_low_level_write (16, value, &acpi_gbl_FADT->Xpm1a_cnt_blk, 0); - acpi_hw_low_level_write (16, value, &acpi_gbl_FADT->Xpm1b_cnt_blk, 0); - if (register_id == SLP_EN) { - enable(); - } - } + case PM1_a_CONTROL: /* 16-bit access */ + + acpi_hw_low_level_write (16, value, &acpi_gbl_FADT->Xpm1a_cnt_blk, 0); + break; + + + case PM1_b_CONTROL: /* 16-bit access */ + acpi_hw_low_level_write (16, value, &acpi_gbl_FADT->Xpm1b_cnt_blk, 0); break; diff --git a/drivers/acpi/hardware/hwsleep.c b/drivers/acpi/hardware/hwsleep.c new file mode 100644 index 000000000..892c721c6 --- /dev/null +++ b/drivers/acpi/hardware/hwsleep.c @@ -0,0 +1,189 @@ + +/****************************************************************************** + * + * Name: hwsleep.c - ACPI Hardware Sleep/Wake Interface + * $Revision: 5 $ + * + *****************************************************************************/ + +/* + * Copyright (C) 2000, 2001 R. Byron Moore + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include "acpi.h" +#include "acnamesp.h" +#include "achware.h" + +#define _COMPONENT HARDWARE + MODULE_NAME ("hwsleep") + + +/****************************************************************************** + * + * FUNCTION: Acpi_set_firmware_waking_vector + * + * PARAMETERS: Physical_address - Physical address of ACPI real mode + * entry point. + * + * RETURN: AE_OK or AE_ERROR + * + * DESCRIPTION: Access function for d_firmware_waking_vector field in FACS + * + ******************************************************************************/ + +ACPI_STATUS +acpi_set_firmware_waking_vector ( + ACPI_PHYSICAL_ADDRESS physical_address) +{ + + + /* Make sure that we have an FACS */ + + if (!acpi_gbl_FACS) { + return (AE_NO_ACPI_TABLES); + } + + /* Set the vector */ + + if (acpi_gbl_FACS->vector_width == 32) { + * (u32 *) acpi_gbl_FACS->firmware_waking_vector = (u32) physical_address; + } + else { + *acpi_gbl_FACS->firmware_waking_vector = physical_address; + } + + return (AE_OK); +} + + +/****************************************************************************** + * + * FUNCTION: Acpi_get_firmware_waking_vector + * + * PARAMETERS: *Physical_address - Output buffer where contents of + * the Firmware_waking_vector field of + * the FACS will be stored. + * + * RETURN: Status + * + * DESCRIPTION: Access function for d_firmware_waking_vector field in FACS + * + ******************************************************************************/ + +ACPI_STATUS +acpi_get_firmware_waking_vector ( + ACPI_PHYSICAL_ADDRESS *physical_address) +{ + + + if (!physical_address) { + return (AE_BAD_PARAMETER); + } + + /* Make sure that we have an FACS */ + + if (!acpi_gbl_FACS) { + return (AE_NO_ACPI_TABLES); + } + + /* Get the vector */ + + if (acpi_gbl_FACS->vector_width == 32) { + *physical_address = * (u32 *) acpi_gbl_FACS->firmware_waking_vector; + } + else { + *physical_address = *acpi_gbl_FACS->firmware_waking_vector; + } + + return (AE_OK); +} + +/****************************************************************************** + * + * FUNCTION: Acpi_enter_sleep_state + * + * PARAMETERS: Sleep_state - Which sleep state to enter + * + * RETURN: Status + * + * DESCRIPTION: Enter a system sleep state (see ACPI 2.0 spec p 231) + * + ******************************************************************************/ + +ACPI_STATUS +acpi_enter_sleep_state ( + u8 sleep_state) +{ + ACPI_STATUS status; + ACPI_OBJECT_LIST arg_list; + ACPI_OBJECT arg; + u8 type_a; + u8 type_b; + u16 PM1_acontrol; + u16 PM1_bcontrol; + + /* + * _PSW methods could be run here to enable wake-on keyboard, LAN, etc. + */ + + status = acpi_hw_obtain_sleep_type_register_data(sleep_state, &type_a, &type_b); + + if (!ACPI_SUCCESS(status)) { + return status; + } + + /* run the _PTS and _GTS methods */ + MEMSET(&arg_list, 0, sizeof(arg_list)); + arg_list.count = 1; + arg_list.pointer = &arg; + + MEMSET(&arg, 0, sizeof(arg)); + arg.type = ACPI_TYPE_INTEGER; + arg.integer.value = sleep_state; + + acpi_evaluate_object(NULL, "\\_PTS", &arg_list, NULL); + acpi_evaluate_object(NULL, "\\_GTS", &arg_list, NULL); + + /* clear wake status */ + acpi_hw_register_bit_access(ACPI_WRITE, ACPI_MTX_LOCK, WAK_STS, 1); + + PM1_acontrol = (u16) acpi_hw_register_read(ACPI_MTX_LOCK, PM1_CONTROL); + + /* mask off SLP_EN and SLP_TYP fields */ + PM1_acontrol &= 0xC3FF; + + /* mask in SLP_EN */ + PM1_acontrol |= (1 << acpi_hw_get_bit_shift (SLP_EN_MASK)); + + PM1_bcontrol = PM1_acontrol; + + /* mask in SLP_TYP */ + PM1_acontrol |= (type_a << acpi_hw_get_bit_shift (SLP_TYPE_X_MASK)); + PM1_bcontrol |= (type_b << acpi_hw_get_bit_shift (SLP_TYPE_X_MASK)); + + /* the old version was disabling interrupts. let's try it without + * and see how that works + */ + /*disable();*/ + + acpi_hw_register_write(ACPI_MTX_LOCK, PM1_a_CONTROL, PM1_acontrol); + acpi_hw_register_write(ACPI_MTX_LOCK, PM1_b_CONTROL, PM1_bcontrol); + + /*enable();*/ + + return (AE_OK); +} diff --git a/drivers/acpi/hardware/hwtimer.c b/drivers/acpi/hardware/hwtimer.c new file mode 100644 index 000000000..b7f529ccf --- /dev/null +++ b/drivers/acpi/hardware/hwtimer.c @@ -0,0 +1,196 @@ + +/****************************************************************************** + * + * Name: hwtimer.c - ACPI Power Management Timer Interface + * $Revision: 4 $ + * + *****************************************************************************/ + +/* + * Copyright (C) 2000, 2001 R. Byron Moore + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include "acpi.h" +#include "achware.h" + +#define _COMPONENT HARDWARE + MODULE_NAME ("hwtimer") + + +/****************************************************************************** + * + * FUNCTION: Acpi_get_timer_resolution + * + * PARAMETERS: none + * + * RETURN: Number of bits of resolution in the PM Timer (24 or 32). + * + * DESCRIPTION: Obtains resolution of the ACPI PM Timer. + * + ******************************************************************************/ + +ACPI_STATUS +acpi_get_timer_resolution ( + u32 *resolution) +{ + if (!resolution) { + return (AE_BAD_PARAMETER); + } + + if (0 == acpi_gbl_FADT->tmr_val_ext) { + *resolution = 24; + } + else { + *resolution = 32; + } + + return (AE_OK); +} + + +/****************************************************************************** + * + * FUNCTION: Acpi_get_timer + * + * PARAMETERS: none + * + * RETURN: Current value of the ACPI PM Timer (in ticks). + * + * DESCRIPTION: Obtains current value of ACPI PM Timer. + * + ******************************************************************************/ + +ACPI_STATUS +acpi_get_timer ( + u32 *ticks) +{ + if (!ticks) { + return (AE_BAD_PARAMETER); + } + + *ticks = acpi_os_in32 ((ACPI_IO_ADDRESS) ACPI_GET_ADDRESS (acpi_gbl_FADT->Xpm_tmr_blk.address)); + + return (AE_OK); +} + + +/****************************************************************************** + * + * FUNCTION: Acpi_get_timer_duration + * + * PARAMETERS: Start_ticks + * End_ticks + * Time_elapsed + * + * RETURN: Time_elapsed + * + * DESCRIPTION: Computes the time elapsed (in microseconds) between two + * PM Timer time stamps, taking into account the possibility of + * rollovers, the timer resolution, and timer frequency. + * + * The PM Timer's clock ticks at roughly 3.6 times per + * _microsecond_, and its clock continues through Cx state + * transitions (unlike many CPU timestamp counters) -- making it + * a versatile and accurate timer. + * + * Note that this function accomodates only a single timer + * rollover. Thus for 24-bit timers, this function should only + * be used for calculating durations less than ~4.6 seconds + * (~20 hours for 32-bit timers). + * + ******************************************************************************/ + +ACPI_STATUS +acpi_get_timer_duration ( + u32 start_ticks, + u32 end_ticks, + u32 *time_elapsed) +{ + u32 delta_ticks = 0; + u32 seconds = 0; + u32 milliseconds = 0; + u32 microseconds = 0; + u32 remainder = 0; + + if (!time_elapsed) { + return (AE_BAD_PARAMETER); + } + + /* + * Compute Tick Delta: + * ------------------- + * Handle timer rollovers on 24- versus 32-bit timers. + */ + if (start_ticks < end_ticks) { + delta_ticks = end_ticks - start_ticks; + } + else if (start_ticks > end_ticks) { + /* 24-bit Timer */ + if (0 == acpi_gbl_FADT->tmr_val_ext) { + delta_ticks = (0x00FFFFFF - start_ticks) + end_ticks; + } + /* 32-bit Timer */ + else { + delta_ticks = (0xFFFFFFFF - start_ticks) + end_ticks; + } + } + + /* + * Compute Duration: + * ----------------- + * Since certain compilers (gcc/Linux, argh!) don't support 64-bit + * divides in kernel-space we have to do some trickery to preserve + * accuracy while using 32-bit math. + * + * TODO: Change to use 64-bit math when supported. + * + * The process is as follows: + * 1. Compute the number of seconds by dividing Delta Ticks by + * the timer frequency. + * 2. Compute the number of milliseconds in the remainder from step #1 + * by multiplying by 1000 and then dividing by the timer frequency. + * 3. Compute the number of microseconds in the remainder from step #2 + * by multiplying by 1000 and then dividing by the timer frequency. + * 4. Add the results from steps 1, 2, and 3 to get the total duration. + * + * Example: The time elapsed for Delta_ticks = 0xFFFFFFFF should be + * 1199864031 microseconds. This is computed as follows: + * Step #1: Seconds = 1199; Remainder = 3092840 + * Step #2: Milliseconds = 864; Remainder = 113120 + * Step #3: Microseconds = 31; Remainder = <don't care!> + */ + + /* Step #1 */ + seconds = delta_ticks / PM_TIMER_FREQUENCY; + remainder = delta_ticks % PM_TIMER_FREQUENCY; + + /* Step #2 */ + milliseconds = (remainder * 1000) / PM_TIMER_FREQUENCY; + remainder = (remainder * 1000) % PM_TIMER_FREQUENCY; + + /* Step #3 */ + microseconds = (remainder * 1000) / PM_TIMER_FREQUENCY; + + /* Step #4 */ + *time_elapsed = seconds * 1000000; + *time_elapsed += milliseconds * 1000; + *time_elapsed += microseconds; + + return (AE_OK); +} + + diff --git a/drivers/acpi/include/accommon.h b/drivers/acpi/include/accommon.h index 37e13b228..61d54a220 100644 --- a/drivers/acpi/include/accommon.h +++ b/drivers/acpi/include/accommon.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: accommon.h -- prototypes for the common (subsystem-wide) procedures - * $Revision: 82 $ + * $Revision: 87 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -100,6 +100,8 @@ acpi_cm_allocate_owner_id ( * Cm_clib - Local implementations of C library functions */ +#ifndef ACPI_USE_SYSTEM_CLIBRARY + NATIVE_UINT acpi_cm_strlen ( const NATIVE_CHAR *string); @@ -141,7 +143,7 @@ u32 acpi_cm_strtoul ( const NATIVE_CHAR *string, NATIVE_CHAR **terminator, - u32 base); + NATIVE_UINT base); NATIVE_CHAR * acpi_cm_strstr ( @@ -161,7 +163,7 @@ acpi_cm_memcpy ( void * acpi_cm_memset ( void *dest, - u32 value, + NATIVE_UINT value, NATIVE_UINT count); u32 @@ -172,6 +174,7 @@ u32 acpi_cm_to_lower ( u32 c); +#endif /* ACPI_USE_SYSTEM_CLIBRARY */ /* * Cm_copy - Object construction and conversion interfaces @@ -297,7 +300,7 @@ function_value_exit ( u32 line_number, u32 component_id, NATIVE_CHAR *function_name, - NATIVE_UINT value); + ACPI_INTEGER value); void function_ptr_exit ( @@ -596,7 +599,7 @@ acpi_cm_init_static_object ( #define acpi_cm_callocate(a) _cm_callocate(a, _COMPONENT,_THIS_MODULE,__LINE__) #define acpi_cm_free(a) _cm_free(a,_COMPONENT,_THIS_MODULE,__LINE__) -#ifndef ACPI_DEBUG +#ifndef ACPI_DEBUG_TRACK_ALLOCATIONS #define acpi_cm_add_element_to_alloc_list(a,b,c,d,e,f) #define acpi_cm_delete_element_from_alloc_list(a,b,c,d) diff --git a/drivers/acpi/include/acconfig.h b/drivers/acpi/include/acconfig.h index 2b210339b..ea9be649d 100644 --- a/drivers/acpi/include/acconfig.h +++ b/drivers/acpi/include/acconfig.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acconfig.h - Global configuration constants - * $Revision: 48 $ + * $Revision: 53 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -53,7 +53,7 @@ /* Version string */ -#define ACPI_CA_VERSION __DATE__ +#define ACPI_CA_VERSION 0x20010125 /* Maximum objects in the various object caches */ @@ -122,6 +122,10 @@ #define MTH_NUM_ARGS 7 #define MTH_MAX_ARG 6 +/* Maximum length of resulting string when converting from a buffer */ + +#define ACPI_MAX_STRING_CONVERSION 200 + /* * Operand Stack (in WALK_STATE), Must be large enough to contain MTH_MAX_ARG */ diff --git a/drivers/acpi/include/acdebug.h b/drivers/acpi/include/acdebug.h index 2bc9e7165..f1fa7094e 100644 --- a/drivers/acpi/include/acdebug.h +++ b/drivers/acpi/include/acdebug.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acdebug.h - ACPI/AML debugger - * $Revision: 37 $ + * $Revision: 39 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/acdispat.h b/drivers/acpi/include/acdispat.h index 599e46c36..f5d52cc41 100644 --- a/drivers/acpi/include/acdispat.h +++ b/drivers/acpi/include/acdispat.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acdispat.h - dispatcher (parser to interpreter interface) - * $Revision: 33 $ + * $Revision: 35 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/acenv.h b/drivers/acpi/include/acenv.h index f867a348f..9e4e62339 100644 --- a/drivers/acpi/include/acenv.h +++ b/drivers/acpi/include/acenv.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acenv.h - Generation environment specific items - * $Revision: 65 $ + * $Revision: 70 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -28,7 +28,7 @@ /* - * Configuration for ACPI Utilities + * Configuration for ACPI tools and utilities */ #ifdef _ACPI_DUMP_APP @@ -55,6 +55,16 @@ #define ACPI_USE_SYSTEM_CLIBRARY #endif +/* + * Memory allocation tracking. Used only if + * 1) This is the debug version + * 2) This is NOT a 16-bit version of the code (not enough real-mode memory) + */ +#ifdef ACPI_DEBUG +#ifndef _IA16 +#define ACPI_DEBUG_TRACK_ALLOCATIONS +#endif +#endif /* * Environment configuration. The purpose of this file is to interface to the @@ -154,17 +164,17 @@ #define STRUPR(s) strupr((s)) #define STRLEN(s) strlen((s)) #define STRCPY(d,s) strcpy((d), (s)) -#define STRNCPY(d,s,n) strncpy((d), (s), (n)) -#define STRNCMP(d,s,n) strncmp((d), (s), (n)) +#define STRNCPY(d,s,n) strncpy((d), (s), (NATIVE_INT)(n)) +#define STRNCMP(d,s,n) strncmp((d), (s), (NATIVE_INT)(n)) #define STRCMP(d,s) strcmp((d), (s)) #define STRCAT(d,s) strcat((d), (s)) -#define STRNCAT(d,s,n) strncat((d), (s), (n)) -#define STRTOUL(d,s,n) strtoul((d), (s), (n)) -#define MEMCPY(d,s,n) memcpy((d), (s), (n)) -#define MEMSET(d,s,n) memset((d), (s), (n)) +#define STRNCAT(d,s,n) strncat((d), (s), (NATIVE_INT)(n)) +#define STRTOUL(d,s,n) strtoul((d), (s), (NATIVE_INT)(n)) +#define MEMCPY(d,s,n) memcpy((d), (s), (NATIVE_INT)(n)) +#define MEMSET(d,s,n) memset((d), (s), (NATIVE_INT)(n)) #define TOUPPER toupper #define TOLOWER tolower - +#define IS_XDIGIT isxdigit /****************************************************************************** * diff --git a/drivers/acpi/include/acevents.h b/drivers/acpi/include/acevents.h index 3e76370bf..b7d335451 100644 --- a/drivers/acpi/include/acevents.h +++ b/drivers/acpi/include/acevents.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acevents.h - Event subcomponent prototypes and defines - * $Revision: 62 $ + * $Revision: 63 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/acexcep.h b/drivers/acpi/include/acexcep.h index 1629a0934..8fb267585 100644 --- a/drivers/acpi/include/acexcep.h +++ b/drivers/acpi/include/acexcep.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acexcep.h - Exception codes returned by the ACPI subsystem - * $Revision: 37 $ + * $Revision: 41 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -116,8 +116,10 @@ #define AE_AML_NAME_NOT_FOUND (ACPI_STATUS) (0x0010 | AE_CODE_AML) #define AE_AML_INTERNAL (ACPI_STATUS) (0x0011 | AE_CODE_AML) #define AE_AML_INVALID_SPACE_ID (ACPI_STATUS) (0x0012 | AE_CODE_AML) +#define AE_AML_STRING_LIMIT (ACPI_STATUS) (0x0013 | AE_CODE_AML) +#define AE_AML_NO_RETURN_VALUE (ACPI_STATUS) (0x0014 | AE_CODE_AML) -#define AE_CODE_AML_MAX 0x0012 +#define AE_CODE_AML_MAX 0x0014 /* * Internal exceptions used for control @@ -202,6 +204,8 @@ static NATIVE_CHAR *acpi_gbl_exception_names_aml[] = "AE_AML_NAME_NOT_FOUND", "AE_AML_INTERNAL", "AE_AML_INVALID_SPACE_ID", + "AE_AML_STRING_LIMIT", + "AE_AML_NO_RETURN_VALUE", }; static NATIVE_CHAR *acpi_gbl_exception_names_ctrl[] = diff --git a/drivers/acpi/include/acgcc.h b/drivers/acpi/include/acgcc.h index 5992f493d..82b1e5139 100644 --- a/drivers/acpi/include/acgcc.h +++ b/drivers/acpi/include/acgcc.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acgcc.h - GCC specific defines, etc. - * $Revision: 2 $ + * $Revision: 4 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/acglobal.h b/drivers/acpi/include/acglobal.h index 248f72c9d..36444038c 100644 --- a/drivers/acpi/include/acglobal.h +++ b/drivers/acpi/include/acglobal.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acglobal.h - Declarations for global variables - * $Revision: 92 $ + * $Revision: 96 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -154,7 +154,6 @@ ACPI_EXTERN u8 acpi_gbl_global_lock_set; /* TBD: [Restr ACPI_EXTERN u8 acpi_gbl_step_to_next_call; ACPI_EXTERN u8 acpi_gbl_acpi_hardware_present; - ACPI_EXTERN ACPI_OBJECT_NOTIFY_HANDLER acpi_gbl_drv_notify; ACPI_EXTERN ACPI_OBJECT_NOTIFY_HANDLER acpi_gbl_sys_notify; @@ -162,7 +161,8 @@ ACPI_EXTERN ACPI_OBJECT_NOTIFY_HANDLER acpi_gbl_sys_notify; extern u8 acpi_gbl_shutdown; extern u32 acpi_gbl_system_flags; extern u32 acpi_gbl_startup_flags; -extern u8 acpi_gbl_decode_to8bit[]; +extern u8 acpi_gbl_decode_to8bit[8]; +extern NATIVE_CHAR acpi_gbl_hex_to_ascii[]; /***************************************************************************** diff --git a/drivers/acpi/include/achware.h b/drivers/acpi/include/achware.h index 1a206e8d2..efb97bd2f 100644 --- a/drivers/acpi/include/achware.h +++ b/drivers/acpi/include/achware.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: achware.h -- hardware specific interfaces - * $Revision: 48 $ + * $Revision: 53 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -27,6 +27,10 @@ #define __ACHWARE_H__ +/* PM Timer ticks per second (HZ) */ +#define PM_TIMER_FREQUENCY 3579545 + + /* Prototypes */ @@ -92,6 +96,10 @@ void acpi_hw_clear_acpi_status ( void); +u32 +acpi_hw_get_bit_shift ( + u32 mask); + /* GPE support */ @@ -121,89 +129,21 @@ acpi_hw_obtain_sleep_type_register_data ( u8 *slp_typ_b); -/* Cx State Prototypes */ - -ACPI_STATUS -acpi_hw_enter_c1( - ACPI_IO_ADDRESS pblk_address, - u32 *pm_timer_ticks); - -ACPI_STATUS -acpi_hw_enter_c2( - ACPI_IO_ADDRESS pblk_address, - u32 *pm_timer_ticks); - -ACPI_STATUS -acpi_hw_enter_c3( - ACPI_IO_ADDRESS pblk_address, - u32 *pm_timer_ticks); - -ACPI_STATUS -acpi_hw_enter_cx ( - ACPI_IO_ADDRESS pblk_address, - u32 *pm_timer_ticks); - -ACPI_STATUS -acpi_hw_set_cx ( - u32 cx_state); - -ACPI_STATUS -acpi_hw_get_cx_info ( - u32 cx_states[]); +/* ACPI Timer prototypes */ ACPI_STATUS -acpi_hw_get_cx_handler ( - u32 cx_state, - ACPI_C_STATE_HANDLER *handler); +acpi_get_timer_resolution ( + u32 *resolution); ACPI_STATUS -acpi_hw_set_cx_handler ( - u32 cx_state, - ACPI_C_STATE_HANDLER handler); - - -/* Throttling Prototypes */ - -void -acpi_hw_enable_throttling ( - ACPI_IO_ADDRESS pblk_address); - -void -acpi_hw_disable_throttling ( - ACPI_IO_ADDRESS pblk_address); - -u32 -acpi_hw_get_duty_cycle ( - u8 duty_offset, - ACPI_IO_ADDRESS pblk_address, - u32 num_throttle_states); - -void -acpi_hw_program_duty_cycle ( - u8 duty_offset, - u32 duty_cycle, - ACPI_IO_ADDRESS pblk_address, - u32 num_throttle_states); - -NATIVE_UINT -acpi_hw_local_pow ( - NATIVE_UINT x, - NATIVE_UINT y); - - -/* ACPI Timer prototypes */ - -u32 -acpi_hw_pmt_ticks ( - void); - -u32 -acpi_hw_pmt_resolution ( - void); +acpi_get_timer ( + u32 *ticks); ACPI_STATUS -acpi_get_timer ( - u32 *out_ticks); +acpi_get_timer_duration ( + u32 start_ticks, + u32 end_ticks, + u32 *time_elapsed); #endif /* __ACHWARE_H__ */ diff --git a/drivers/acpi/include/acinterp.h b/drivers/acpi/include/acinterp.h index c8c967492..6eb571e5f 100644 --- a/drivers/acpi/include/acinterp.h +++ b/drivers/acpi/include/acinterp.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acinterp.h - Interpreter subcomponent prototypes and defines - * $Revision: 86 $ + * $Revision: 91 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -70,9 +70,34 @@ acpi_aml_execute_method ( /* - * amfield - ACPI AML (p-code) execution - field manipulation + * amconvrt - object conversion */ +ACPI_STATUS +acpi_aml_convert_to_integer ( + ACPI_OPERAND_OBJECT **obj_desc, + ACPI_WALK_STATE *walk_state); + +ACPI_STATUS +acpi_aml_convert_to_buffer ( + ACPI_OPERAND_OBJECT **obj_desc, + ACPI_WALK_STATE *walk_state); + +ACPI_STATUS +acpi_aml_convert_to_string ( + ACPI_OPERAND_OBJECT **obj_desc, + ACPI_WALK_STATE *walk_state); + +ACPI_STATUS +acpi_aml_convert_to_target_type ( + OBJECT_TYPE_INTERNAL destination_type, + ACPI_OPERAND_OBJECT **obj_desc, + ACPI_WALK_STATE *walk_state); + + +/* + * amfield - ACPI AML (p-code) execution - field manipulation + */ ACPI_STATUS acpi_aml_read_field ( @@ -419,17 +444,75 @@ acpi_aml_exec_store ( ACPI_WALK_STATE *walk_state); ACPI_STATUS -acpi_aml_store_object_to_object ( +acpi_aml_store_object_to_index ( ACPI_OPERAND_OBJECT *val_desc, ACPI_OPERAND_OBJECT *dest_desc, ACPI_WALK_STATE *walk_state); ACPI_STATUS acpi_aml_store_object_to_node ( - ACPI_OPERAND_OBJECT *val_desc, + ACPI_OPERAND_OBJECT *source_desc, ACPI_NAMESPACE_NODE *node, ACPI_WALK_STATE *walk_state); +ACPI_STATUS +acpi_aml_store_object_to_object ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *dest_desc, + ACPI_WALK_STATE *walk_state); + + +/* + * + */ + +ACPI_STATUS +acpi_aml_resolve_object ( + ACPI_OPERAND_OBJECT **source_desc_ptr, + OBJECT_TYPE_INTERNAL target_type, + ACPI_WALK_STATE *walk_state); + +ACPI_STATUS +acpi_aml_store_object ( + ACPI_OPERAND_OBJECT *source_desc, + OBJECT_TYPE_INTERNAL target_type, + ACPI_OPERAND_OBJECT **target_desc_ptr, + ACPI_WALK_STATE *walk_state); + + +/* + * amcopy - object copy + */ + +ACPI_STATUS +acpi_aml_copy_buffer_to_buffer ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *target_desc); + +ACPI_STATUS +acpi_aml_copy_string_to_string ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *target_desc); + +ACPI_STATUS +acpi_aml_copy_integer_to_index_field ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *target_desc); + +ACPI_STATUS +acpi_aml_copy_integer_to_bank_field ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *target_desc); + +ACPI_STATUS +acpi_aml_copy_data_to_named_field ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_NAMESPACE_NODE *node); + +ACPI_STATUS +acpi_aml_copy_integer_to_field_unit ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *target_desc); /* * amutils - interpreter/scanner utilities diff --git a/drivers/acpi/include/aclinux.h b/drivers/acpi/include/aclinux.h index 673d5f96b..0cf0e2845 100644 --- a/drivers/acpi/include/aclinux.h +++ b/drivers/acpi/include/aclinux.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: aclinux.h - OS specific defines, etc. - * $Revision: 6 $ + * $Revision: 7 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/aclocal.h b/drivers/acpi/include/aclocal.h index a647026f1..2d931387e 100644 --- a/drivers/acpi/include/aclocal.h +++ b/drivers/acpi/include/aclocal.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: aclocal.h - Internal data types used across the ACPI subsystem - * $Revision: 95 $ + * $Revision: 100 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -74,7 +74,7 @@ typedef u32 ACPI_MUTEX_HANDLE; #define NUM_MTX MAX_MTX+1 -#ifdef ACPI_DEBUG +#if defined(ACPI_DEBUG) || defined(ENABLE_DEBUGGER) #ifdef DEFINE_ACPI_GLOBALS /* Names for the mutexes used in the subsystem */ @@ -476,7 +476,9 @@ typedef struct acpi_opcode_info u32 parse_args; /* Grammar/Parse time arguments */ u32 runtime_args; /* Interpret time arguments */ - DEBUG_ONLY_MEMBERS (NATIVE_CHAR *name) /* op name (debug only) */ +#ifdef _OPCODE_NAMES + NATIVE_CHAR *name; /* op name (debug only) */ +#endif } ACPI_OPCODE_INFO; @@ -591,6 +593,7 @@ typedef struct acpi_walk_state ACPI_GENERIC_STATE *scope_info; /* Stack of nested scopes */ ACPI_PARSE_STATE *parser_state; /* Current state of parser */ u8 *aml_last_while; + ACPI_OPCODE_INFO *op_info; /* Info on current opcode */ ACPI_PARSE_DOWNWARDS descending_callback; ACPI_PARSE_UPWARDS ascending_callback; @@ -605,6 +608,7 @@ typedef struct acpi_walk_state u32 parse_flags; u8 walk_type; u8 return_used; + u16 opcode; /* Current AML opcode */ u32 prev_arg_types; /* Debug support */ @@ -728,14 +732,16 @@ typedef struct acpi_get_devices_info #define PM1_STS 0x0100 #define PM1_EN 0x0200 #define PM1_CONTROL 0x0300 -#define PM2_CONTROL 0x0400 -#define PM_TIMER 0x0500 -#define PROCESSOR_BLOCK 0x0600 -#define GPE0_STS_BLOCK 0x0700 -#define GPE0_EN_BLOCK 0x0800 -#define GPE1_STS_BLOCK 0x0900 -#define GPE1_EN_BLOCK 0x0A00 -#define SMI_CMD_BLOCK 0x0B00 +#define PM1_a_CONTROL 0x0400 +#define PM1_b_CONTROL 0x0500 +#define PM2_CONTROL 0x0600 +#define PM_TIMER 0x0700 +#define PROCESSOR_BLOCK 0x0800 +#define GPE0_STS_BLOCK 0x0900 +#define GPE0_EN_BLOCK 0x0A00 +#define GPE1_STS_BLOCK 0x0B00 +#define GPE1_EN_BLOCK 0x0C00 +#define SMI_CMD_BLOCK 0x0D00 /* * Address space bitmasks for mmio or io spaces @@ -751,66 +757,66 @@ typedef struct acpi_get_devices_info /* * Control bit definitions */ -#define TMR_STS (PM1_STS | 0x01) -#define BM_STS (PM1_STS | 0x02) -#define GBL_STS (PM1_STS | 0x03) -#define PWRBTN_STS (PM1_STS | 0x04) -#define SLPBTN_STS (PM1_STS | 0x05) -#define RTC_STS (PM1_STS | 0x06) -#define WAK_STS (PM1_STS | 0x07) - -#define TMR_EN (PM1_EN | 0x01) +#define TMR_STS (PM1_STS | 0x01) +#define BM_STS (PM1_STS | 0x02) +#define GBL_STS (PM1_STS | 0x03) +#define PWRBTN_STS (PM1_STS | 0x04) +#define SLPBTN_STS (PM1_STS | 0x05) +#define RTC_STS (PM1_STS | 0x06) +#define WAK_STS (PM1_STS | 0x07) + +#define TMR_EN (PM1_EN | 0x01) /* no BM_EN */ -#define GBL_EN (PM1_EN | 0x03) -#define PWRBTN_EN (PM1_EN | 0x04) -#define SLPBTN_EN (PM1_EN | 0x05) -#define RTC_EN (PM1_EN | 0x06) -#define WAK_EN (PM1_EN | 0x07) +#define GBL_EN (PM1_EN | 0x03) +#define PWRBTN_EN (PM1_EN | 0x04) +#define SLPBTN_EN (PM1_EN | 0x05) +#define RTC_EN (PM1_EN | 0x06) +#define WAK_EN (PM1_EN | 0x07) -#define SCI_EN (PM1_CONTROL | 0x01) -#define BM_RLD (PM1_CONTROL | 0x02) -#define GBL_RLS (PM1_CONTROL | 0x03) -#define SLP_TYPE_A (PM1_CONTROL | 0x04) -#define SLP_TYPE_B (PM1_CONTROL | 0x05) -#define SLP_EN (PM1_CONTROL | 0x06) +#define SCI_EN (PM1_CONTROL | 0x01) +#define BM_RLD (PM1_CONTROL | 0x02) +#define GBL_RLS (PM1_CONTROL | 0x03) +#define SLP_TYPE_A (PM1_CONTROL | 0x04) +#define SLP_TYPE_B (PM1_CONTROL | 0x05) +#define SLP_EN (PM1_CONTROL | 0x06) -#define ARB_DIS (PM2_CONTROL | 0x01) +#define ARB_DIS (PM2_CONTROL | 0x01) -#define TMR_VAL (PM_TIMER | 0x01) +#define TMR_VAL (PM_TIMER | 0x01) -#define GPE0_STS (GPE0_STS_BLOCK | 0x01) -#define GPE0_EN (GPE0_EN_BLOCK | 0x01) +#define GPE0_STS (GPE0_STS_BLOCK | 0x01) +#define GPE0_EN (GPE0_EN_BLOCK | 0x01) -#define GPE1_STS (GPE1_STS_BLOCK | 0x01) -#define GPE1_EN (GPE1_EN_BLOCK | 0x01) +#define GPE1_STS (GPE1_STS_BLOCK | 0x01) +#define GPE1_EN (GPE1_EN_BLOCK | 0x01) -#define TMR_STS_MASK 0x0001 -#define BM_STS_MASK 0x0010 -#define GBL_STS_MASK 0x0020 -#define PWRBTN_STS_MASK 0x0100 -#define SLPBTN_STS_MASK 0x0200 -#define RTC_STS_MASK 0x0400 -#define WAK_STS_MASK 0x8000 +#define TMR_STS_MASK 0x0001 +#define BM_STS_MASK 0x0010 +#define GBL_STS_MASK 0x0020 +#define PWRBTN_STS_MASK 0x0100 +#define SLPBTN_STS_MASK 0x0200 +#define RTC_STS_MASK 0x0400 +#define WAK_STS_MASK 0x8000 -#define ALL_FIXED_STS_BITS (TMR_STS_MASK | BM_STS_MASK | GBL_STS_MASK \ - | PWRBTN_STS_MASK | SLPBTN_STS_MASK \ - | RTC_STS_MASK | WAK_STS_MASK) +#define ALL_FIXED_STS_BITS (TMR_STS_MASK | BM_STS_MASK | GBL_STS_MASK \ + | PWRBTN_STS_MASK | SLPBTN_STS_MASK \ + | RTC_STS_MASK | WAK_STS_MASK) -#define TMR_EN_MASK 0x0001 -#define GBL_EN_MASK 0x0020 -#define PWRBTN_EN_MASK 0x0100 -#define SLPBTN_EN_MASK 0x0200 -#define RTC_EN_MASK 0x0400 +#define TMR_EN_MASK 0x0001 +#define GBL_EN_MASK 0x0020 +#define PWRBTN_EN_MASK 0x0100 +#define SLPBTN_EN_MASK 0x0200 +#define RTC_EN_MASK 0x0400 -#define SCI_EN_MASK 0x0001 -#define BM_RLD_MASK 0x0002 -#define GBL_RLS_MASK 0x0004 -#define SLP_TYPE_X_MASK 0x1C00 -#define SLP_EN_MASK 0x2000 +#define SCI_EN_MASK 0x0001 +#define BM_RLD_MASK 0x0002 +#define GBL_RLS_MASK 0x0004 +#define SLP_TYPE_X_MASK 0x1C00 +#define SLP_EN_MASK 0x2000 -#define ARB_DIS_MASK 0x0001 -#define TMR_VAL_MASK 0xFFFFFFFF +#define ARB_DIS_MASK 0x0001 +#define TMR_VAL_MASK 0xFFFFFFFF #define GPE0_STS_MASK #define GPE0_EN_MASK @@ -819,8 +825,8 @@ typedef struct acpi_get_devices_info #define GPE1_EN_MASK -#define ACPI_READ 1 -#define ACPI_WRITE 2 +#define ACPI_READ 1 +#define ACPI_WRITE 2 /* Plug and play */ diff --git a/drivers/acpi/include/acmacros.h b/drivers/acpi/include/acmacros.h index 19cfa0591..7bed83ddd 100644 --- a/drivers/acpi/include/acmacros.h +++ b/drivers/acpi/include/acmacros.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acmacros.h - C macros for the entire subsystem. - * $Revision: 59 $ + * $Revision: 62 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -30,6 +30,14 @@ * Data manipulation macros */ +#ifndef LODWORD +#define LODWORD(l) ((u32)(UINT64)(l)) +#endif + +#ifndef HIDWORD +#define HIDWORD(l) ((u32)((((UINT64)(l)) >> 32) & 0xFFFFFFFF)) +#endif + #ifndef LOWORD #define LOWORD(l) ((u16)(NATIVE_UINT)(l)) #endif @@ -64,10 +72,18 @@ #ifdef _IA16 +/* + * For 16-bit addresses, we have to assume that the upper 32 bits + * are zero. + */ #define ACPI_GET_ADDRESS(a) ((a).lo) #define ACPI_STORE_ADDRESS(a,b) {(a).hi=0;(a).lo=(b);} -#define ACPI_VALID_ADDRESS(a) ((a).hi && (a).lo) +#define ACPI_VALID_ADDRESS(a) ((a).hi | (a).lo) + #else +/* + * Full 64-bit address on 32-bit and 64-bit platforms + */ #define ACPI_GET_ADDRESS(a) (a) #define ACPI_STORE_ADDRESS(a,b) ((a)=(b)) #define ACPI_VALID_ADDRESS(a) (a) @@ -335,7 +351,7 @@ */ #define return_VOID {function_exit(_THIS_MODULE,__LINE__,_COMPONENT,_proc_name);return;} #define return_ACPI_STATUS(s) {function_status_exit(_THIS_MODULE,__LINE__,_COMPONENT,_proc_name,s);return(s);} -#define return_VALUE(s) {function_value_exit(_THIS_MODULE,__LINE__,_COMPONENT,_proc_name,(NATIVE_UINT)s);return(s);} +#define return_VALUE(s) {function_value_exit(_THIS_MODULE,__LINE__,_COMPONENT,_proc_name,(ACPI_INTEGER)s);return(s);} #define return_PTR(s) {function_ptr_exit(_THIS_MODULE,__LINE__,_COMPONENT,_proc_name,(u8 *)s);return(s);} @@ -346,6 +362,8 @@ #define DEBUG_DEFINE(a) a; #define DEBUG_ONLY_MEMBERS(a) a; +#define _OPCODE_NAMES +#define _VERBOSE_STRUCTURES /* Stack and buffer dumping */ @@ -458,9 +476,8 @@ */ #ifdef _IA16 #undef DEBUG_ONLY_MEMBERS +#undef _VERBOSE_STRUCTURES #define DEBUG_ONLY_MEMBERS(a) -#undef OP_INFO_ENTRY -#define OP_INFO_ENTRY(flags,name,Pargs,Iargs) {flags,Pargs,Iargs} #endif diff --git a/drivers/acpi/include/acnamesp.h b/drivers/acpi/include/acnamesp.h index e010a8118..d6acb8444 100644 --- a/drivers/acpi/include/acnamesp.h +++ b/drivers/acpi/include/acnamesp.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acnamesp.h - Namespace subcomponent prototypes and defines - * $Revision: 100 $ + * $Revision: 101 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/acobject.h b/drivers/acpi/include/acobject.h index c801ff117..9394b470a 100644 --- a/drivers/acpi/include/acobject.h +++ b/drivers/acpi/include/acobject.h @@ -2,12 +2,12 @@ /****************************************************************************** * * Name: acobject.h - Definition of ACPI_OPERAND_OBJECT (Internal object only) - * $Revision: 75 $ + * $Revision: 78 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -113,10 +113,10 @@ typedef struct /* NUMBER - has value */ ACPI_INTEGER value; -} ACPI_OBJECT_NUMBER; +} ACPI_OBJECT_INTEGER; -typedef struct /* STRING - has length and pointer */ +typedef struct /* STRING - has length and pointer - Null terminated, ASCII characters only */ { ACPI_OBJECT_COMMON_HEADER @@ -126,13 +126,11 @@ typedef struct /* STRING - has length and pointer */ } ACPI_OBJECT_STRING; -typedef struct /* BUFFER - has length, sequence, and pointer */ +typedef struct /* BUFFER - has length and pointer - not null terminated */ { ACPI_OBJECT_COMMON_HEADER u32 length; - u32 sequence; /* Sequential count of buffers created */ - u8 *pointer; /* points to the buffer in allocated space */ } ACPI_OBJECT_BUFFER; @@ -398,7 +396,7 @@ typedef union acpi_operand_obj { ACPI_OBJECT_COMMON common; ACPI_OBJECT_CACHE_LIST cache; - ACPI_OBJECT_NUMBER number; + ACPI_OBJECT_INTEGER integer; ACPI_OBJECT_STRING string; ACPI_OBJECT_BUFFER buffer; ACPI_OBJECT_PACKAGE package; diff --git a/drivers/acpi/include/acoutput.h b/drivers/acpi/include/acoutput.h index 664a5f8a8..5c20943cb 100644 --- a/drivers/acpi/include/acoutput.h +++ b/drivers/acpi/include/acoutput.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acoutput.h -- debug output - * $Revision: 66 $ + * $Revision: 69 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -58,6 +58,7 @@ #define BATTERY 0x00040000 #define DEBUGGER 0x00100000 +#define COMPILER 0x00200000 #define ALL_COMPONENTS 0x001FFFFF #define COMPONENT_DEFAULT (ALL_COMPONENTS) diff --git a/drivers/acpi/include/acparser.h b/drivers/acpi/include/acparser.h index d657749da..9b4bfabea 100644 --- a/drivers/acpi/include/acparser.h +++ b/drivers/acpi/include/acparser.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: acparser.h - AML Parser subcomponent prototypes and defines - * $Revision: 47 $ + * $Revision: 49 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/acpi.h b/drivers/acpi/include/acpi.h index 4896c4a7f..344b01a77 100644 --- a/drivers/acpi/include/acpi.h +++ b/drivers/acpi/include/acpi.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acpi.h - Master include file, Publics and external data. - * $Revision: 48 $ + * $Revision: 50 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/acpiosxf.h b/drivers/acpi/include/acpiosxf.h index 2f9eb4c13..ee289f33e 100644 --- a/drivers/acpi/include/acpiosxf.h +++ b/drivers/acpi/include/acpiosxf.h @@ -9,7 +9,7 @@ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -121,6 +121,10 @@ acpi_os_unmap_memory ( void *logical_address, u32 length); +ACPI_STATUS +acpi_os_get_physical_address ( + void *logical_address, + ACPI_PHYSICAL_ADDRESS *physical_address); /* * Interrupt handlers diff --git a/drivers/acpi/include/acpixf.h b/drivers/acpi/include/acpixf.h index d70fa75a7..8d293818f 100644 --- a/drivers/acpi/include/acpixf.h +++ b/drivers/acpi/include/acpixf.h @@ -6,7 +6,7 @@ *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -285,6 +285,10 @@ acpi_get_firmware_waking_vector ( ACPI_PHYSICAL_ADDRESS *physical_address); ACPI_STATUS +acpi_enter_sleep_state ( + u8 sleep_state); + +ACPI_STATUS acpi_get_processor_throttling_info ( ACPI_HANDLE processor_handle, ACPI_BUFFER *user_buffer); diff --git a/drivers/acpi/include/acresrc.h b/drivers/acpi/include/acresrc.h index 3bb19490d..0852367d2 100644 --- a/drivers/acpi/include/acresrc.h +++ b/drivers/acpi/include/acresrc.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: acresrc.h - Resource Manager function prototypes - * $Revision: 20 $ + * $Revision: 22 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/actables.h b/drivers/acpi/include/actables.h index 4dd724517..be8b5e3b0 100644 --- a/drivers/acpi/include/actables.h +++ b/drivers/acpi/include/actables.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: actables.h - ACPI table management - * $Revision: 27 $ + * $Revision: 29 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/actbl.h b/drivers/acpi/include/actbl.h index 34631e820..7a41842d2 100644 --- a/drivers/acpi/include/actbl.h +++ b/drivers/acpi/include/actbl.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: actbl.h - Table data structures defined in ACPI specification - * $Revision: 43 $ + * $Revision: 45 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/actbl1.h b/drivers/acpi/include/actbl1.h index 019ba1680..5c68ca050 100644 --- a/drivers/acpi/include/actbl1.h +++ b/drivers/acpi/include/actbl1.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: actbl1.h - ACPI 1.0 tables - * $Revision: 15 $ + * $Revision: 17 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/actbl2.h b/drivers/acpi/include/actbl2.h index e3ccf2188..0a05efa1c 100644 --- a/drivers/acpi/include/actbl2.h +++ b/drivers/acpi/include/actbl2.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: actbl2.h - ACPI Specification Revision 2.0 Tables - * $Revision: 19 $ + * $Revision: 21 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -142,8 +142,8 @@ typedef struct u16 plvl3_lat; /* worst case HW latency to enter/exit C3 state */ u16 flush_size; /* number of flush strides that need to be read */ u16 flush_stride; /* Processor's memory cache line width, in bytes */ - u8 duty_offset; /* Processor’s duty cycle index in processor's P_CNT reg*/ - u8 duty_width; /* Processor’s duty cycle value bit width in P_CNT register.*/ + u8 duty_offset; /* Processor_’s duty cycle index in processor's P_CNT reg*/ + u8 duty_width; /* Processor_’s duty cycle value bit width in P_CNT register.*/ u8 day_alrm; /* index to day-of-month alarm in RTC CMOS RAM */ u8 mon_alrm; /* index to month-of-year alarm in RTC CMOS RAM */ u8 century; /* index to century in RTC CMOS RAM */ diff --git a/drivers/acpi/include/actbl71.h b/drivers/acpi/include/actbl71.h index 408ec402e..095806ccb 100644 --- a/drivers/acpi/include/actbl71.h +++ b/drivers/acpi/include/actbl71.h @@ -3,12 +3,12 @@ * Name: actbl71.h - IA-64 Extensions to the ACPI Spec Rev. 0.71 * This file includes tables specific to this * specification revision. - * $Revision: 7 $ + * $Revision: 9 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/include/actypes.h b/drivers/acpi/include/actypes.h index dfa28a9d9..826fc8901 100644 --- a/drivers/acpi/include/actypes.h +++ b/drivers/acpi/include/actypes.h @@ -1,12 +1,12 @@ /****************************************************************************** * * Name: actypes.h - Common data types for the entire ACPI subsystem - * $Revision: 159 $ + * $Revision: 163 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -97,7 +97,7 @@ typedef INT16 NATIVE_INT; typedef UINT32 ACPI_TBLPTR; typedef UINT32 ACPI_IO_ADDRESS; -typedef void *ACPI_PHYSICAL_ADDRESS; +typedef char *ACPI_PHYSICAL_ADDRESS; #define ALIGNED_ADDRESS_BOUNDARY 0x00000002 #define _HW_ALIGNMENT_SUPPORT @@ -249,8 +249,9 @@ typedef UINT64 ACPI_INTEGER; #define ACPI_STATE_S2 (u8) 2 #define ACPI_STATE_S3 (u8) 3 #define ACPI_STATE_S4 (u8) 4 -#define ACPI_STATE_S4_bIOS (u8) 5 -#define ACPI_STATE_S5 (u8) 6 +#define ACPI_STATE_S5 (u8) 5 +/* let's pretend S4_bIOS didn't exist for now. ASG */ +#define ACPI_STATE_S4_bIOS (u8) 6 #define ACPI_S_STATES_MAX ACPI_STATE_S5 @@ -284,8 +285,37 @@ typedef u32 ACPI_TABLE_TYPE; typedef u32 ACPI_OBJECT_TYPE; typedef u8 OBJECT_TYPE_INTERNAL; +#define ACPI_BTYPE_ANY 0x00000000 +#define ACPI_BTYPE_INTEGER 0x00000001 +#define ACPI_BTYPE_STRING 0x00000002 +#define ACPI_BTYPE_BUFFER 0x00000004 +#define ACPI_BTYPE_PACKAGE 0x00000008 +#define ACPI_BTYPE_FIELD_UNIT 0x00000010 +#define ACPI_BTYPE_DEVICE 0x00000020 +#define ACPI_BTYPE_EVENT 0x00000040 +#define ACPI_BTYPE_METHOD 0x00000080 +#define ACPI_BTYPE_MUTEX 0x00000100 +#define ACPI_BTYPE_REGION 0x00000200 +#define ACPI_BTYPE_POWER 0x00000400 +#define ACPI_BTYPE_PROCESSOR 0x00000800 +#define ACPI_BTYPE_THERMAL 0x00001000 +#define ACPI_BTYPE_BUFFER_FIELD 0x00002000 +#define ACPI_BTYPE_DDB_HANDLE 0x00004000 +#define ACPI_BTYPE_DEBUG_OBJECT 0x00008000 +#define ACPI_BTYPE_REFERENCE 0x00010000 +#define ACPI_BTYPE_RESOURCE 0x00020000 + +#define ACPI_BTYPE_COMPUTE_DATA (ACPI_BTYPE_INTEGER | ACPI_BTYPE_STRING | ACPI_BTYPE_BUFFER) + +#define ACPI_BTYPE_DATA (ACPI_BTYPE_COMPUTE_DATA | ACPI_BTYPE_PACKAGE) +#define ACPI_BTYPE_DATA_REFERENCE (ACPI_BTYPE_DATA | ACPI_BTYPE_REFERENCE | ACPI_BTYPE_DDB_HANDLE) +#define ACPI_BTYPE_DEVICE_OBJECTS (ACPI_BTYPE_DEVICE | ACPI_BTYPE_THERMAL | ACPI_BTYPE_PROCESSOR) +#define ACPI_BTYPE_OBJECTS_AND_REFS 0x00017FFF /* ARG or LOCAL */ +#define ACPI_BTYPE_ALL_OBJECTS 0x00007FFF + + #define ACPI_TYPE_ANY 0 /* 0x00 */ -#define ACPI_TYPE_NUMBER 1 /* 0x01 Byte/Word/Dword/Zero/One/Ones */ +#define ACPI_TYPE_INTEGER 1 /* 0x01 Byte/Word/Dword/Zero/One/Ones */ #define ACPI_TYPE_STRING 2 /* 0x02 */ #define ACPI_TYPE_BUFFER 3 /* 0x03 */ #define ACPI_TYPE_PACKAGE 4 /* 0x04 Byte_const, multiple Data_term/Constant/Super_name */ @@ -432,7 +462,7 @@ typedef union acpi_obj { ACPI_OBJECT_TYPE type; ACPI_INTEGER value; /* The actual number */ - } number; + } integer; struct { diff --git a/drivers/acpi/include/amlcode.h b/drivers/acpi/include/amlcode.h index 9a5cb2c52..707e7535d 100644 --- a/drivers/acpi/include/amlcode.h +++ b/drivers/acpi/include/amlcode.h @@ -3,12 +3,12 @@ * Name: amlcode.h - Definitions for AML, as included in "definition blocks" * Declarations and definitions contained herein are derived * directly from the ACPI specification. - * $Revision: 42 $ + * $Revision: 46 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -211,27 +211,42 @@ #define ARGP_TERMARG 0x0E #define ARGP_TERMLIST 0x0F #define ARGP_WORDDATA 0x10 +#define ARGP_QWORDDATA 0x11 +#define ARGP_SIMPLENAME 0x12 /* * Resolved argument types for the AML Interpreter * Each field in the Arg_types u32 is 5 bits, allowing for a maximum of 6 arguments. - * There can be up to 31 unique argument types + * There can be up to 31 unique argument types (0 is end-of-arg-list indicator) */ -#define ARGI_ANYTYPE 0x01 -#define ARGI_TARGETREF 0x02 -#define ARGI_REFERENCE 0x03 -#define ARGI_IF 0x04 -#define ARGI_NUMBER 0x05 -#define ARGI_STRING 0x06 -#define ARGI_BUFFER 0x07 -#define ARGI_PACKAGE 0x08 -#define ARGI_DATAOBJECT 0x09 /* Buffer, string, package or reference to a Node - Used only by Size_of operator*/ -#define ARGI_COMPLEXOBJ 0x0A /* Buffer or package */ -#define ARGI_MUTEX 0x0B -#define ARGI_EVENT 0x0C -#define ARGI_REGION 0x0D -#define ARGI_DDBHANDLE 0x0E +/* "Standard" ACPI types are 1-15 (0x0F) */ + +#define ARGI_INTEGER ACPI_TYPE_INTEGER /* 1 */ +#define ARGI_STRING ACPI_TYPE_STRING /* 2 */ +#define ARGI_BUFFER ACPI_TYPE_BUFFER /* 3 */ +#define ARGI_PACKAGE ACPI_TYPE_PACKAGE /* 4 */ +#define ARGI_EVENT ACPI_TYPE_EVENT +#define ARGI_MUTEX ACPI_TYPE_MUTEX +#define ARGI_REGION ACPI_TYPE_REGION +#define ARGI_DDBHANDLE ACPI_TYPE_DDB_HANDLE + +/* Custom types are 0x10 through 0x1F */ + +#define ARGI_IF 0x10 +#define ARGI_ANYOBJECT 0x11 +#define ARGI_ANYTYPE 0x12 +#define ARGI_COMPUTEDATA 0x13 /* Buffer, String, or Integer */ +#define ARGI_DATAOBJECT 0x14 /* Buffer, string, package or reference to a Node - Used only by Size_of operator*/ +#define ARGI_COMPLEXOBJ 0x15 /* Buffer or package */ +#define ARGI_INTEGER_REF 0x16 +#define ARGI_OBJECT_REF 0x17 +#define ARGI_DEVICE_REF 0x18 +#define ARGI_REFERENCE 0x19 +#define ARGI_TARGETREF 0x1A /* Target, subject to implicit conversion */ +#define ARGI_FIXED_TARGET 0x1B /* Target, no implicit conversion */ +#define ARGI_SIMPLE_TARGET 0x1C /* Name, Local, Arg -- no implicit conversion */ +#define ARGI_BUFFERSTRING 0x1D #define ARGI_INVALID_OPCODE 0xFFFFFFFF diff --git a/drivers/acpi/interpreter/amconfig.c b/drivers/acpi/interpreter/amconfig.c index 55e5b0510..497f35375 100644 --- a/drivers/acpi/interpreter/amconfig.c +++ b/drivers/acpi/interpreter/amconfig.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: amconfig - Namespace reconfiguration (Load/Unload opcodes) - * $Revision: 26 $ + * $Revision: 29 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -73,7 +73,8 @@ acpi_aml_exec_load_table ( table_header.length = 0; for (i = 0; i < sizeof (ACPI_TABLE_HEADER); i++) { status = acpi_ev_address_space_dispatch (rgn_desc, ADDRESS_SPACE_READ, - i, 8, (u32 *) ((u8 *) &table_header + i)); + (ACPI_PHYSICAL_ADDRESS) i, 8, + (u32 *) ((u8 *) &table_header + i)); if (ACPI_FAILURE (status)) { return (status); } @@ -96,7 +97,8 @@ acpi_aml_exec_load_table ( for (i = 0; i < table_header.length; i++) { status = acpi_ev_address_space_dispatch (rgn_desc, ADDRESS_SPACE_READ, - i, 8, (u32 *) (table_data_ptr + i)); + (ACPI_PHYSICAL_ADDRESS)i, 8, + (u32 *) (table_data_ptr + i)); if (ACPI_FAILURE (status)) { goto cleanup; } diff --git a/drivers/acpi/interpreter/amconvrt.c b/drivers/acpi/interpreter/amconvrt.c new file mode 100644 index 000000000..4c895c7ed --- /dev/null +++ b/drivers/acpi/interpreter/amconvrt.c @@ -0,0 +1,525 @@ +/****************************************************************************** + * + * Module Name: amconvrt - Object conversion routines + * $Revision: 3 $ + * + *****************************************************************************/ + +/* + * Copyright (C) 2000, 2001 R. Byron Moore + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + + +#include "acpi.h" +#include "acparser.h" +#include "acnamesp.h" +#include "acinterp.h" +#include "acevents.h" +#include "amlcode.h" +#include "acdispat.h" + + +#define _COMPONENT INTERPRETER + MODULE_NAME ("amconvrt") + + +/******************************************************************************* + * + * FUNCTION: Acpi_aml_convert_to_target_type + * + * PARAMETERS: *Obj_desc - Object to be converted. + * Walk_state - Current method state + * + * RETURN: Status + * + * DESCRIPTION: + * + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_convert_to_target_type ( + OBJECT_TYPE_INTERNAL destination_type, + ACPI_OPERAND_OBJECT **obj_desc, + ACPI_WALK_STATE *walk_state) +{ + ACPI_STATUS status = AE_OK; + + + /* + * If required by the target, + * perform implicit conversion on the source before we store it. + */ + + switch (GET_CURRENT_ARG_TYPE (walk_state->op_info->runtime_args)) + { + case ARGI_SIMPLE_TARGET: + case ARGI_FIXED_TARGET: + case ARGI_INTEGER_REF: /* Handles Increment, Decrement cases */ + + switch (destination_type) + { + case INTERNAL_TYPE_DEF_FIELD: + /* + * Named field can always handle conversions + */ + break; + + default: + /* No conversion allowed for these types */ + + if (destination_type != (*obj_desc)->common.type) { + status = AE_TYPE; + } + } + break; + + + case ARGI_TARGETREF: + + switch (destination_type) + { + case ACPI_TYPE_INTEGER: + case ACPI_TYPE_FIELD_UNIT: + case INTERNAL_TYPE_BANK_FIELD: + case INTERNAL_TYPE_INDEX_FIELD: + /* + * These types require an Integer operand. We can convert + * a Buffer or a String to an Integer if necessary. + */ + status = acpi_aml_convert_to_integer (obj_desc, walk_state); + break; + + + case ACPI_TYPE_STRING: + + /* + * The operand must be a String. We can convert an + * Integer or Buffer if necessary + */ + status = acpi_aml_convert_to_string (obj_desc, walk_state); + break; + + + case ACPI_TYPE_BUFFER: + + /* + * The operand must be a String. We can convert an + * Integer or Buffer if necessary + */ + status = acpi_aml_convert_to_buffer (obj_desc, walk_state); + break; + } + break; + + + case ARGI_REFERENCE: + /* + * Create_xxxx_field cases - we are storing the field object into the name + */ + break; + + + default: + status = AE_AML_INTERNAL; + } + + + /* + * Source-to-Target conversion semantics: + * + * If conversion to the target type cannot be performed, then simply + * overwrite the target with the new object and type. + */ + if (status == AE_TYPE) { + status = AE_OK; + } + + return (status); +} + + +/******************************************************************************* + * + * FUNCTION: Acpi_aml_convert_to_integer + * + * PARAMETERS: *Obj_desc - Object to be converted. Must be an + * Integer, Buffer, or String + * Walk_state - Current method state + * + * RETURN: Status + * + * DESCRIPTION: Convert an ACPI Object to an integer. + * + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_convert_to_integer ( + ACPI_OPERAND_OBJECT **obj_desc, + ACPI_WALK_STATE *walk_state) +{ + u32 i; + ACPI_OPERAND_OBJECT *ret_desc; + u32 count; + char *pointer; + ACPI_INTEGER result; + u32 integer_size = sizeof (ACPI_INTEGER); + + + switch ((*obj_desc)->common.type) + { + case ACPI_TYPE_INTEGER: + return (AE_OK); + + case ACPI_TYPE_STRING: + pointer = (*obj_desc)->string.pointer; + count = (*obj_desc)->string.length; + break; + + case ACPI_TYPE_BUFFER: + pointer = (char *) (*obj_desc)->buffer.pointer; + count = (*obj_desc)->buffer.length; + break; + + default: + return (AE_TYPE); + } + + /* + * Create a new integer + */ + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); + if (!ret_desc) { + return (AE_NO_MEMORY); + } + + + /* Handle both ACPI 1.0 and ACPI 2.0 Integer widths */ + + if (walk_state->method_node->flags & ANOBJ_DATA_WIDTH_32) { + /* + * We are running a method that exists in a 32-bit ACPI table. + * Truncate the value to 32 bits by zeroing out the upper 32-bit field + */ + integer_size = sizeof (u32); + } + + + /* + * Convert the buffer/string to an integer. Note that both buffers and + * strings are treated as raw data - we don't convert ascii to hex for + * strings. + * + * There are two terminating conditions for the loop: + * 1) The size of an integer has been reached, or + * 2) The end of the buffer or string has been reached + */ + result = 0; + + /* Transfer no more than an integer's worth of data */ + + if (count > integer_size) { + count = integer_size; + } + + /* + * String conversion is different than Buffer conversion + */ + switch ((*obj_desc)->common.type) + { + case ACPI_TYPE_STRING: + + /* TBD: Need to use 64-bit STRTOUL */ + + /* + * Convert string to an integer + * String must be hexadecimal as per the ACPI specification + */ + + result = STRTOUL (pointer, NULL, 16); + break; + + + case ACPI_TYPE_BUFFER: + + /* + * Buffer conversion - we simply grab enough raw data from the + * buffer to fill an integer + */ + for (i = 0; i < count; i++) { + /* + * Get next byte and shift it into the Result. + * Little endian is used, meaning that the first byte of the buffer + * is the LSB of the integer + */ + result |= (((ACPI_INTEGER) pointer[i]) << (i * 8)); + } + + break; + } + + /* Save the Result, delete original descriptor, store new descriptor */ + + ret_desc->integer.value = result; + + if (walk_state->opcode != AML_STORE_OP) { + acpi_cm_remove_reference (*obj_desc); + } + + *obj_desc = ret_desc; + + return (AE_OK); +} + + +/******************************************************************************* + * + * FUNCTION: Acpi_aml_convert_to_buffer + * + * PARAMETERS: *Obj_desc - Object to be converted. Must be an + * Integer, Buffer, or String + * Walk_state - Current method state + * + * RETURN: Status + * + * DESCRIPTION: Convert an ACPI Object to an Buffer + * + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_convert_to_buffer ( + ACPI_OPERAND_OBJECT **obj_desc, + ACPI_WALK_STATE *walk_state) +{ + ACPI_OPERAND_OBJECT *ret_desc; + u32 i; + u32 integer_size = sizeof (ACPI_INTEGER); + u8 *new_buf; + + + switch ((*obj_desc)->common.type) + { + case ACPI_TYPE_INTEGER: + + /* + * Create a new Buffer + */ + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_BUFFER); + if (!ret_desc) { + return (AE_NO_MEMORY); + } + + /* Handle both ACPI 1.0 and ACPI 2.0 Integer widths */ + + if (walk_state->method_node->flags & ANOBJ_DATA_WIDTH_32) { + /* + * We are running a method that exists in a 32-bit ACPI table. + * Truncate the value to 32 bits by zeroing out the upper + * 32-bit field + */ + integer_size = sizeof (u32); + } + + /* Need enough space for one integers */ + + ret_desc->buffer.length = integer_size; + new_buf = acpi_cm_callocate (integer_size); + if (!new_buf) { + REPORT_ERROR + (("Aml_exec_dyadic2_r/Concat_op: Buffer allocation failure\n")); + acpi_cm_remove_reference (ret_desc); + return (AE_NO_MEMORY); + } + + /* Copy the integer to the buffer */ + + for (i = 0; i < integer_size; i++) { + new_buf[i] = (u8) ((*obj_desc)->integer.value >> (i * 8)); + } + ret_desc->buffer.pointer = new_buf; + + /* Return the new buffer descriptor */ + + if (walk_state->opcode != AML_STORE_OP) { + acpi_cm_remove_reference (*obj_desc); + } + *obj_desc = ret_desc; + break; + + + case ACPI_TYPE_STRING: + break; + + + case ACPI_TYPE_BUFFER: + break; + + + default: + return (AE_TYPE); + break; + } + + return (AE_OK); +} + + +/******************************************************************************* + * + * FUNCTION: Acpi_aml_convert_to_string + * + * PARAMETERS: *Obj_desc - Object to be converted. Must be an + * Integer, Buffer, or String + * Walk_state - Current method state + * + * RETURN: Status + * + * DESCRIPTION: Convert an ACPI Object to a string + * + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_convert_to_string ( + ACPI_OPERAND_OBJECT **obj_desc, + ACPI_WALK_STATE *walk_state) +{ + ACPI_OPERAND_OBJECT *ret_desc; + u32 i; + u32 index; + u32 integer_size = sizeof (ACPI_INTEGER); + u8 *new_buf; + u8 *pointer; + + + switch ((*obj_desc)->common.type) + { + case ACPI_TYPE_INTEGER: + + /* + * Create a new String + */ + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_STRING); + if (!ret_desc) { + return (AE_NO_MEMORY); + } + + /* Handle both ACPI 1.0 and ACPI 2.0 Integer widths */ + + if (walk_state->method_node->flags & ANOBJ_DATA_WIDTH_32) { + /* + * We are running a method that exists in a 32-bit ACPI table. + * Truncate the value to 32 bits by zeroing out the upper + * 32-bit field + */ + integer_size = sizeof (u32); + } + + /* Need enough space for one ASCII integer plus null terminator */ + + ret_desc->string.length = (integer_size * 2) + 1; + new_buf = acpi_cm_callocate (ret_desc->string.length); + if (!new_buf) { + REPORT_ERROR + (("Aml_exec_dyadic2_r/Concat_op: Buffer allocation failure\n")); + acpi_cm_remove_reference (ret_desc); + return (AE_NO_MEMORY); + } + + /* Copy the integer to the buffer */ + + for (i = 0; i < (integer_size * 2); i++) { + new_buf[i] = acpi_gbl_hex_to_ascii [((*obj_desc)->integer.value >> (i * 4)) & 0xF]; + } + + /* Null terminate */ + + new_buf [i] = 0; + ret_desc->buffer.pointer = new_buf; + + /* Return the new buffer descriptor */ + + if (walk_state->opcode != AML_STORE_OP) { + acpi_cm_remove_reference (*obj_desc); + } + *obj_desc = ret_desc; + + return (AE_OK); + + + case ACPI_TYPE_BUFFER: + + if (((*obj_desc)->buffer.length * 3) > ACPI_MAX_STRING_CONVERSION) { + return (AE_AML_STRING_LIMIT); + } + + /* + * Create a new String + */ + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_STRING); + if (!ret_desc) { + return (AE_NO_MEMORY); + } + + /* Need enough space for one ASCII integer plus null terminator */ + + ret_desc->string.length = (*obj_desc)->buffer.length * 3; + new_buf = acpi_cm_callocate (ret_desc->string.length + 1); + if (!new_buf) { + REPORT_ERROR + (("Aml_exec_dyadic2_r/Concat_op: Buffer allocation failure\n")); + acpi_cm_remove_reference (ret_desc); + return (AE_NO_MEMORY); + } + + /* + * Convert each byte of the buffer to two ASCII characters plus a space. + */ + pointer = (*obj_desc)->buffer.pointer; + index = 0; + for (i = 0; i < (*obj_desc)->buffer.length; i++) { + new_buf[index + 0] = acpi_gbl_hex_to_ascii [pointer[i] & 0x0F]; + new_buf[index + 1] = acpi_gbl_hex_to_ascii [(pointer[i] >> 4) & 0x0F]; + new_buf[index + 2] = ' '; + index += 3; + } + + /* Null terminate */ + + new_buf [index] = 0; + ret_desc->buffer.pointer = new_buf; + + /* Return the new buffer descriptor */ + + if (walk_state->opcode != AML_STORE_OP) { + acpi_cm_remove_reference (*obj_desc); + } + *obj_desc = ret_desc; + break; + + + case ACPI_TYPE_STRING: + break; + + + default: + return (AE_TYPE); + break; + } + + return (AE_OK); +} + + diff --git a/drivers/acpi/interpreter/amcreate.c b/drivers/acpi/interpreter/amcreate.c index 02d7933e1..f19c422ce 100644 --- a/drivers/acpi/interpreter/amcreate.c +++ b/drivers/acpi/interpreter/amcreate.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: amcreate - Named object creation - * $Revision: 51 $ + * $Revision: 53 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -353,7 +353,7 @@ acpi_aml_exec_create_mutex ( goto cleanup; } - obj_desc->mutex.sync_level = (u8) sync_desc->number.value; + obj_desc->mutex.sync_level = (u8) sync_desc->integer.value; /* Obj_desc was on the stack top, and the name is below it */ diff --git a/drivers/acpi/interpreter/amdyadic.c b/drivers/acpi/interpreter/amdyadic.c index ba67b062a..a02173f29 100644 --- a/drivers/acpi/interpreter/amdyadic.c +++ b/drivers/acpi/interpreter/amdyadic.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: amdyadic - ACPI AML (p-code) execution for dyadic operators - * $Revision: 68 $ + * $Revision: 71 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -37,7 +37,174 @@ MODULE_NAME ("amdyadic") -/***************************************************************************** +/******************************************************************************* + * + * FUNCTION: Acpi_aml_do_concatenate + * + * PARAMETERS: *Obj_desc - Object to be converted. Must be an + * Integer, Buffer, or String + * + * RETURN: Status + * + * DESCRIPTION: Concatenate two objects OF THE SAME TYPE. + * + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_do_concatenate ( + ACPI_OPERAND_OBJECT *obj_desc, + ACPI_OPERAND_OBJECT *obj_desc2, + ACPI_OPERAND_OBJECT **actual_ret_desc, + ACPI_WALK_STATE *walk_state) +{ + ACPI_STATUS status; + u32 i; + ACPI_INTEGER this_integer; + ACPI_OPERAND_OBJECT *ret_desc; + NATIVE_CHAR *new_buf; + u32 integer_size = sizeof (ACPI_INTEGER); + + + /* + * There are three cases to handle: + * 1) Two Integers concatenated to produce a buffer + * 2) Two Strings concatenated to produce a string + * 3) Two Buffers concatenated to produce a buffer + */ + switch (obj_desc->common.type) + { + case ACPI_TYPE_INTEGER: + + /* Handle both ACPI 1.0 and ACPI 2.0 Integer widths */ + + if (walk_state->method_node->flags & ANOBJ_DATA_WIDTH_32) { + /* + * We are running a method that exists in a 32-bit ACPI table. + * Truncate the value to 32 bits by zeroing out the upper + * 32-bit field + */ + integer_size = sizeof (u32); + } + + /* Result of two integers is a buffer */ + + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_BUFFER); + if (!ret_desc) { + return (AE_NO_MEMORY); + } + + /* Need enough space for two integers */ + + ret_desc->buffer.length = integer_size * 2; + new_buf = acpi_cm_callocate (ret_desc->buffer.length); + if (!new_buf) { + REPORT_ERROR + (("Aml_exec_dyadic2_r/Concat_op: Buffer allocation failure\n")); + status = AE_NO_MEMORY; + goto cleanup; + } + + ret_desc->buffer.pointer = (u8 *) new_buf; + + /* Convert the first integer */ + + this_integer = obj_desc->integer.value; + for (i = 0; i < integer_size; i++) { + new_buf[i] = (u8) this_integer; + this_integer >>= 8; + } + + /* Convert the second integer */ + + this_integer = obj_desc2->integer.value; + for (; i < (integer_size * 2); i++) { + new_buf[i] = (u8) this_integer; + this_integer >>= 8; + } + + break; + + + case ACPI_TYPE_STRING: + + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_STRING); + if (!ret_desc) { + return (AE_NO_MEMORY); + } + + /* Operand1 is string */ + + new_buf = acpi_cm_allocate (obj_desc->string.length + + obj_desc2->string.length + 1); + if (!new_buf) { + REPORT_ERROR + (("Aml_exec_dyadic2_r/Concat_op: String allocation failure\n")); + status = AE_NO_MEMORY; + goto cleanup; + } + + STRCPY (new_buf, obj_desc->string.pointer); + STRCPY (new_buf + obj_desc->string.length, + obj_desc2->string.pointer); + + /* Point the return object to the new string */ + + ret_desc->string.pointer = new_buf; + ret_desc->string.length = obj_desc->string.length += + obj_desc2->string.length; + break; + + + case ACPI_TYPE_BUFFER: + + /* Operand1 is a buffer */ + + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_BUFFER); + if (!ret_desc) { + return (AE_NO_MEMORY); + } + + new_buf = acpi_cm_allocate (obj_desc->buffer.length + + obj_desc2->buffer.length); + if (!new_buf) { + REPORT_ERROR + (("Aml_exec_dyadic2_r/Concat_op: Buffer allocation failure\n")); + status = AE_NO_MEMORY; + goto cleanup; + } + + MEMCPY (new_buf, obj_desc->buffer.pointer, + obj_desc->buffer.length); + MEMCPY (new_buf + obj_desc->buffer.length, obj_desc2->buffer.pointer, + obj_desc2->buffer.length); + + /* + * Point the return object to the new buffer + */ + + ret_desc->buffer.pointer = (u8 *) new_buf; + ret_desc->buffer.length = obj_desc->buffer.length + + obj_desc2->buffer.length; + break; + + default: + status = AE_AML_INTERNAL; + ret_desc = NULL; + } + + + *actual_ret_desc = ret_desc; + return (AE_OK); + + +cleanup: + + acpi_cm_remove_reference (ret_desc); + return (status); +} + + +/******************************************************************************* * * FUNCTION: Acpi_aml_exec_dyadic1 * @@ -50,7 +217,7 @@ * * ALLOCATION: Deletes both operands * - ****************************************************************************/ + ******************************************************************************/ ACPI_STATUS acpi_aml_exec_dyadic1 ( @@ -106,7 +273,7 @@ acpi_aml_exec_dyadic1 ( /* Dispatch the notify to the appropriate handler */ - acpi_ev_notify_dispatch (node, (u32) val_desc->number.value); + acpi_ev_notify_dispatch (node, (u32) val_desc->integer.value); break; default: @@ -135,7 +302,7 @@ cleanup: } -/***************************************************************************** +/******************************************************************************* * * FUNCTION: Acpi_aml_exec_dyadic2_r * @@ -148,7 +315,7 @@ cleanup: * * ALLOCATION: Deletes one operand descriptor -- other remains on stack * - ****************************************************************************/ + ******************************************************************************/ ACPI_STATUS acpi_aml_exec_dyadic2_r ( @@ -164,7 +331,6 @@ acpi_aml_exec_dyadic2_r ( ACPI_OPERAND_OBJECT *ret_desc2 = NULL; ACPI_STATUS status = AE_OK; u32 num_operands = 3; - NATIVE_CHAR *new_buf; /* Resolve all operands */ @@ -201,7 +367,7 @@ acpi_aml_exec_dyadic2_r ( case AML_SHIFT_RIGHT_OP: case AML_SUBTRACT_OP: - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!ret_desc) { status = AE_NO_MEMORY; goto cleanup; @@ -222,8 +388,8 @@ acpi_aml_exec_dyadic2_r ( case AML_ADD_OP: - ret_desc->number.value = obj_desc->number.value + - obj_desc2->number.value; + ret_desc->integer.value = obj_desc->integer.value + + obj_desc2->integer.value; break; @@ -231,8 +397,8 @@ acpi_aml_exec_dyadic2_r ( case AML_BIT_AND_OP: - ret_desc->number.value = obj_desc->number.value & - obj_desc2->number.value; + ret_desc->integer.value = obj_desc->integer.value & + obj_desc2->integer.value; break; @@ -240,8 +406,8 @@ acpi_aml_exec_dyadic2_r ( case AML_BIT_NAND_OP: - ret_desc->number.value = ~(obj_desc->number.value & - obj_desc2->number.value); + ret_desc->integer.value = ~(obj_desc->integer.value & + obj_desc2->integer.value); break; @@ -249,8 +415,8 @@ acpi_aml_exec_dyadic2_r ( case AML_BIT_OR_OP: - ret_desc->number.value = obj_desc->number.value | - obj_desc2->number.value; + ret_desc->integer.value = obj_desc->integer.value | + obj_desc2->integer.value; break; @@ -258,8 +424,8 @@ acpi_aml_exec_dyadic2_r ( case AML_BIT_NOR_OP: - ret_desc->number.value = ~(obj_desc->number.value | - obj_desc2->number.value); + ret_desc->integer.value = ~(obj_desc->integer.value | + obj_desc2->integer.value); break; @@ -267,16 +433,16 @@ acpi_aml_exec_dyadic2_r ( case AML_BIT_XOR_OP: - ret_desc->number.value = obj_desc->number.value ^ - obj_desc2->number.value; + ret_desc->integer.value = obj_desc->integer.value ^ + obj_desc2->integer.value; break; - /* Def_divide := Divide_op Dividend Divisor Remainder Quotient */ + /* Def_divide := Divide_op Dividend Divisor Remainder Quotient */ case AML_DIVIDE_OP: - if (!obj_desc2->number.value) { + if (!obj_desc2->integer.value) { REPORT_ERROR (("Aml_exec_dyadic2_r/Divide_op: Divide by zero\n")); @@ -284,7 +450,7 @@ acpi_aml_exec_dyadic2_r ( goto cleanup; } - ret_desc2 = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + ret_desc2 = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!ret_desc2) { status = AE_NO_MEMORY; goto cleanup; @@ -292,13 +458,13 @@ acpi_aml_exec_dyadic2_r ( /* Remainder (modulo) */ - ret_desc->number.value = ACPI_MODULO (obj_desc->number.value, - obj_desc2->number.value); + ret_desc->integer.value = ACPI_MODULO (obj_desc->integer.value, + obj_desc2->integer.value); /* Result (what we used to call the quotient) */ - ret_desc2->number.value = ACPI_DIVIDE (obj_desc->number.value, - obj_desc2->number.value); + ret_desc2->integer.value = ACPI_DIVIDE (obj_desc->integer.value, + obj_desc2->integer.value); break; @@ -306,8 +472,8 @@ acpi_aml_exec_dyadic2_r ( case AML_MULTIPLY_OP: - ret_desc->number.value = obj_desc->number.value * - obj_desc2->number.value; + ret_desc->integer.value = obj_desc->integer.value * + obj_desc2->integer.value; break; @@ -315,8 +481,8 @@ acpi_aml_exec_dyadic2_r ( case AML_SHIFT_LEFT_OP: - ret_desc->number.value = obj_desc->number.value << - obj_desc2->number.value; + ret_desc->integer.value = obj_desc->integer.value << + obj_desc2->integer.value; break; @@ -324,8 +490,8 @@ acpi_aml_exec_dyadic2_r ( case AML_SHIFT_RIGHT_OP: - ret_desc->number.value = obj_desc->number.value >> - obj_desc2->number.value; + ret_desc->integer.value = obj_desc->integer.value >> + obj_desc2->integer.value; break; @@ -333,8 +499,8 @@ acpi_aml_exec_dyadic2_r ( case AML_SUBTRACT_OP: - ret_desc->number.value = obj_desc->number.value - - obj_desc2->number.value; + ret_desc->integer.value = obj_desc->integer.value - + obj_desc2->integer.value; break; @@ -342,79 +508,54 @@ acpi_aml_exec_dyadic2_r ( case AML_CONCAT_OP: - if (obj_desc2->common.type != obj_desc->common.type) { - status = AE_AML_OPERAND_TYPE; - goto cleanup; - } - /* Both operands are now known to be the same */ - - if (ACPI_TYPE_STRING == obj_desc->common.type) { - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_STRING); - if (!ret_desc) { - status = AE_NO_MEMORY; - goto cleanup; - } + /* + * Convert the second operand if necessary. The first operand + * determines the type of the second operand, (See the Data Types + * section of the ACPI specification.) Both object types are + * guaranteed to be either Integer/String/Buffer by the operand + * resolution mechanism above. + */ - /* Operand1 is string */ + switch (obj_desc->common.type) + { + case ACPI_TYPE_INTEGER: + status = acpi_aml_convert_to_integer (&obj_desc2, walk_state); + break; - new_buf = acpi_cm_allocate (obj_desc->string.length + - obj_desc2->string.length + 1); - if (!new_buf) { - REPORT_ERROR - (("Aml_exec_dyadic2_r/Concat_op: String allocation failure\n")); - status = AE_NO_MEMORY; - goto cleanup; - } + case ACPI_TYPE_STRING: + status = acpi_aml_convert_to_string (&obj_desc2, walk_state); + break; - STRCPY (new_buf, obj_desc->string.pointer); - STRCPY (new_buf + obj_desc->string.length, - obj_desc2->string.pointer); + case ACPI_TYPE_BUFFER: + status = acpi_aml_convert_to_buffer (&obj_desc2, walk_state); + break; - /* Point the return object to the new string */ - - ret_desc->string.pointer = new_buf; - ret_desc->string.length = obj_desc->string.length += - obj_desc2->string.length; + default: + status = AE_AML_INTERNAL; } - else { - /* Operand1 is not a string ==> must be a buffer */ - - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_BUFFER); - if (!ret_desc) { - status = AE_NO_MEMORY; - goto cleanup; - } - - new_buf = acpi_cm_allocate (obj_desc->buffer.length + - obj_desc2->buffer.length); - if (!new_buf) { - REPORT_ERROR - (("Aml_exec_dyadic2_r/Concat_op: Buffer allocation failure\n")); - status = AE_NO_MEMORY; - goto cleanup; - } - - MEMCPY (new_buf, obj_desc->buffer.pointer, - obj_desc->buffer.length); - MEMCPY (new_buf + obj_desc->buffer.length, obj_desc2->buffer.pointer, - obj_desc2->buffer.length); + if (ACPI_FAILURE (status)) { + goto cleanup; + } - /* - * Point the return object to the new buffer - */ - ret_desc->buffer.pointer = (u8 *) new_buf; - ret_desc->buffer.length = obj_desc->buffer.length + - obj_desc2->buffer.length; + /* + * Both operands are now known to be the same object type + * (Both are Integer, String, or Buffer), and we can now perform the + * concatenation. + */ + status = acpi_aml_do_concatenate (obj_desc, obj_desc2, &ret_desc, walk_state); + if (ACPI_FAILURE (status)) { + goto cleanup; } break; default: - REPORT_ERROR (("Acpi_aml_exec_dyadic2_r: Unknown dyadic opcode %X\n", opcode)); + REPORT_ERROR (("Acpi_aml_exec_dyadic2_r: Unknown dyadic opcode %X\n", + opcode)); status = AE_AML_BAD_OPCODE; goto cleanup; } @@ -474,7 +615,7 @@ cleanup: } -/***************************************************************************** +/******************************************************************************* * * FUNCTION: Acpi_aml_exec_dyadic2_s * @@ -486,7 +627,7 @@ cleanup: * * ALLOCATION: Deletes one operand descriptor -- other remains on stack * - ****************************************************************************/ + ******************************************************************************/ ACPI_STATUS acpi_aml_exec_dyadic2_s ( @@ -516,7 +657,7 @@ acpi_aml_exec_dyadic2_s ( /* Create the internal return object */ - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!ret_desc) { status = AE_NO_MEMORY; goto cleanup; @@ -524,7 +665,7 @@ acpi_aml_exec_dyadic2_s ( /* Default return value is FALSE, operation did not time out */ - ret_desc->number.value = 0; + ret_desc->integer.value = 0; /* Examine the opcode */ @@ -562,7 +703,7 @@ acpi_aml_exec_dyadic2_s ( */ if (status == AE_TIME) { - ret_desc->number.value = ACPI_INTEGER_MAX; /* TRUE, op timed out */ + ret_desc->integer.value = ACPI_INTEGER_MAX; /* TRUE, op timed out */ status = AE_OK; } @@ -591,7 +732,7 @@ cleanup: } -/***************************************************************************** +/******************************************************************************* * * FUNCTION: Acpi_aml_exec_dyadic2 * @@ -605,7 +746,7 @@ cleanup: * ALLOCATION: Deletes one operand descriptor -- other remains on stack * containing result value * - ****************************************************************************/ + ******************************************************************************/ ACPI_STATUS acpi_aml_exec_dyadic2 ( @@ -636,7 +777,7 @@ acpi_aml_exec_dyadic2 ( /* Create the internal return object */ - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!ret_desc) { status = AE_NO_MEMORY; goto cleanup; @@ -654,8 +795,8 @@ acpi_aml_exec_dyadic2 ( case AML_LAND_OP: - lboolean = (u8) (obj_desc->number.value && - obj_desc2->number.value); + lboolean = (u8) (obj_desc->integer.value && + obj_desc2->integer.value); break; @@ -663,8 +804,8 @@ acpi_aml_exec_dyadic2 ( case AML_LEQUAL_OP: - lboolean = (u8) (obj_desc->number.value == - obj_desc2->number.value); + lboolean = (u8) (obj_desc->integer.value == + obj_desc2->integer.value); break; @@ -672,8 +813,8 @@ acpi_aml_exec_dyadic2 ( case AML_LGREATER_OP: - lboolean = (u8) (obj_desc->number.value > - obj_desc2->number.value); + lboolean = (u8) (obj_desc->integer.value > + obj_desc2->integer.value); break; @@ -681,8 +822,8 @@ acpi_aml_exec_dyadic2 ( case AML_LLESS_OP: - lboolean = (u8) (obj_desc->number.value < - obj_desc2->number.value); + lboolean = (u8) (obj_desc->integer.value < + obj_desc2->integer.value); break; @@ -690,8 +831,8 @@ acpi_aml_exec_dyadic2 ( case AML_LOR_OP: - lboolean = (u8) (obj_desc->number.value || - obj_desc2->number.value); + lboolean = (u8) (obj_desc->integer.value || + obj_desc2->integer.value); break; @@ -707,10 +848,10 @@ acpi_aml_exec_dyadic2 ( /* Set return value to logical TRUE (all ones) or FALSE (zero) */ if (lboolean) { - ret_desc->number.value = ACPI_INTEGER_MAX; + ret_desc->integer.value = ACPI_INTEGER_MAX; } else { - ret_desc->number.value = 0; + ret_desc->integer.value = 0; } diff --git a/drivers/acpi/interpreter/amfield.c b/drivers/acpi/interpreter/amfield.c index 356be14c2..3ed442c95 100644 --- a/drivers/acpi/interpreter/amfield.c +++ b/drivers/acpi/interpreter/amfield.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: amfield - ACPI AML (p-code) execution - field manipulation - * $Revision: 74 $ + * $Revision: 77 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -174,7 +174,8 @@ acpi_aml_access_named_field ( u32 byte_field_length; - /* Basic data checking */ + /* Parameter validation */ + if ((!named_field) || (ACPI_READ == mode && !buffer)) { return (AE_AML_INTERNAL); } @@ -227,7 +228,7 @@ acpi_aml_access_named_field ( /* TBD: should these round down to a power of 2? */ - if (DIV_8(bit_granularity) > byte_field_length) { + if (DIV_8 (bit_granularity) > byte_field_length) { bit_granularity = MUL_8(byte_field_length); } diff --git a/drivers/acpi/interpreter/amfldio.c b/drivers/acpi/interpreter/amfldio.c index ce877c982..d16cd3d9b 100644 --- a/drivers/acpi/interpreter/amfldio.c +++ b/drivers/acpi/interpreter/amfldio.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: amfldio - Aml Field I/O - * $Revision: 32 $ + * $Revision: 37 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -132,7 +132,7 @@ acpi_aml_read_field ( u32 this_field_byte_offset; u32 this_field_datum_offset; u32 previous_raw_datum; - u32 this_raw_datum; + u32 this_raw_datum = 0; u32 valid_field_bits; u32 mask; u32 merged_datum = 0; @@ -203,32 +203,46 @@ acpi_aml_read_field ( while (this_field_datum_offset < datum_length) { /* - * Get the next raw datum, it contains bits of the current - * field datum + * If the field is aligned on a byte boundary, we don't want + * to perform a final read, since this would potentially read + * past the end of the region. + * + * TBD: [Investigate] It may make more sense to just split the aligned + * and non-aligned cases since the aligned case is so very simple, */ - - status = acpi_aml_read_field_data (obj_desc, - this_field_byte_offset + byte_granularity, - bit_granularity, &this_raw_datum); - if (ACPI_FAILURE (status)) { - goto cleanup; - } - - /* Before merging the data, make sure the unused bits are clear */ - - switch (byte_granularity) + if ((obj_desc->field.bit_offset != 0) || + ((obj_desc->field.bit_offset == 0) && + (this_field_datum_offset < (datum_length -1)))) { - case 1: - this_raw_datum &= 0x000000FF; - previous_raw_datum &= 0x000000FF; - break; - - case 2: - this_raw_datum &= 0x0000FFFF; - previous_raw_datum &= 0x0000FFFF; - break; + /* + * Get the next raw datum, it contains some or all bits + * of the current field datum + */ + + status = acpi_aml_read_field_data (obj_desc, + this_field_byte_offset + byte_granularity, + bit_granularity, &this_raw_datum); + if (ACPI_FAILURE (status)) { + goto cleanup; + } + + /* Before merging the data, make sure the unused bits are clear */ + + switch (byte_granularity) + { + case 1: + this_raw_datum &= 0x000000FF; + previous_raw_datum &= 0x000000FF; + break; + + case 2: + this_raw_datum &= 0x0000FFFF; + previous_raw_datum &= 0x0000FFFF; + break; + } } + /* * Put together bits of the two raw data to make a complete * field datum @@ -391,8 +405,6 @@ acpi_aml_write_field_data_with_update_rule ( merged_value = field_value; - /* Check if update rule needs to be applied (not if mask is all ones) */ - /* Decode the update rule */ @@ -401,13 +413,17 @@ acpi_aml_write_field_data_with_update_rule ( case UPDATE_PRESERVE: - /* - * Read the current contents of the byte/word/dword containing - * the field, and merge with the new field value. - */ - status = acpi_aml_read_field_data (obj_desc, this_field_byte_offset, - bit_granularity, ¤t_value); - merged_value |= (current_value & ~mask); + /* Check if update rule needs to be applied (not if mask is all ones) */ + + if (((1 << bit_granularity) -1) & ~mask) { + /* + * Read the current contents of the byte/word/dword containing + * the field, and merge with the new field value. + */ + status = acpi_aml_read_field_data (obj_desc, this_field_byte_offset, + bit_granularity, ¤t_value); + merged_value |= (current_value & ~mask); + } break; diff --git a/drivers/acpi/interpreter/ammisc.c b/drivers/acpi/interpreter/ammisc.c index 907169dfc..3c58daeab 100644 --- a/drivers/acpi/interpreter/ammisc.c +++ b/drivers/acpi/interpreter/ammisc.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: ammisc - ACPI AML (p-code) execution - specific opcodes - * $Revision: 71 $ + * $Revision: 73 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -172,7 +172,7 @@ acpi_aml_exec_index ( if (obj_desc->common.type == ACPI_TYPE_PACKAGE) { /* Object to be indexed is a Package */ - if (idx_desc->number.value >= obj_desc->package.count) { + if (idx_desc->integer.value >= obj_desc->package.count) { status = AE_AML_PACKAGE_LIMIT; goto cleanup; } @@ -195,7 +195,7 @@ acpi_aml_exec_index ( * we are after. */ - tmp_desc = obj_desc->package.elements[idx_desc->number.value]; + tmp_desc = obj_desc->package.elements[idx_desc->integer.value]; ret_desc->reference.op_code = AML_INDEX_OP; ret_desc->reference.target_type = tmp_desc->common.type; ret_desc->reference.object = tmp_desc; @@ -210,13 +210,13 @@ acpi_aml_exec_index ( */ ret_desc->reference.op_code = AML_INDEX_OP; ret_desc->reference.target_type = ACPI_TYPE_PACKAGE; - ret_desc->reference.where = &obj_desc->package.elements[idx_desc->number.value]; + ret_desc->reference.where = &obj_desc->package.elements[idx_desc->integer.value]; } else { /* Object to be indexed is a Buffer */ - if (idx_desc->number.value >= obj_desc->buffer.length) { + if (idx_desc->integer.value >= obj_desc->buffer.length) { status = AE_AML_BUFFER_LIMIT; goto cleanup; } @@ -224,7 +224,7 @@ acpi_aml_exec_index ( ret_desc->reference.op_code = AML_INDEX_OP; ret_desc->reference.target_type = ACPI_TYPE_BUFFER_FIELD; ret_desc->reference.object = obj_desc; - ret_desc->reference.offset = (u32) idx_desc->number.value; + ret_desc->reference.offset = (u32) idx_desc->integer.value; status = acpi_aml_exec_store (ret_desc, res_desc, walk_state); } @@ -314,20 +314,20 @@ acpi_aml_exec_match ( /* Validate match comparison sub-opcodes */ - if ((op1_desc->number.value > MAX_MATCH_OPERATOR) || - (op2_desc->number.value > MAX_MATCH_OPERATOR)) + if ((op1_desc->integer.value > MAX_MATCH_OPERATOR) || + (op2_desc->integer.value > MAX_MATCH_OPERATOR)) { status = AE_AML_OPERAND_VALUE; goto cleanup; } - index = (u32) start_desc->number.value; + index = (u32) start_desc->integer.value; if (index >= (u32) pkg_desc->package.count) { status = AE_AML_PACKAGE_LIMIT; goto cleanup; } - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!ret_desc) { status = AE_NO_MEMORY; goto cleanup; @@ -351,7 +351,7 @@ acpi_aml_exec_match ( * should we examine its value? */ if (!pkg_desc->package.elements[index] || - ACPI_TYPE_NUMBER != pkg_desc->package.elements[index]->common.type) + ACPI_TYPE_INTEGER != pkg_desc->package.elements[index]->common.type) { continue; } @@ -362,7 +362,7 @@ acpi_aml_exec_match ( * "continue" (proceed to next iteration of enclosing * "for" loop) signifies a non-match. */ - switch (op1_desc->number.value) + switch (op1_desc->integer.value) { case MATCH_MTR: /* always true */ @@ -372,8 +372,8 @@ acpi_aml_exec_match ( case MATCH_MEQ: /* true if equal */ - if (pkg_desc->package.elements[index]->number.value - != V1_desc->number.value) + if (pkg_desc->package.elements[index]->integer.value + != V1_desc->integer.value) { continue; } @@ -382,8 +382,8 @@ acpi_aml_exec_match ( case MATCH_MLE: /* true if less than or equal */ - if (pkg_desc->package.elements[index]->number.value - > V1_desc->number.value) + if (pkg_desc->package.elements[index]->integer.value + > V1_desc->integer.value) { continue; } @@ -392,8 +392,8 @@ acpi_aml_exec_match ( case MATCH_MLT: /* true if less than */ - if (pkg_desc->package.elements[index]->number.value - >= V1_desc->number.value) + if (pkg_desc->package.elements[index]->integer.value + >= V1_desc->integer.value) { continue; } @@ -402,8 +402,8 @@ acpi_aml_exec_match ( case MATCH_MGE: /* true if greater than or equal */ - if (pkg_desc->package.elements[index]->number.value - < V1_desc->number.value) + if (pkg_desc->package.elements[index]->integer.value + < V1_desc->integer.value) { continue; } @@ -412,8 +412,8 @@ acpi_aml_exec_match ( case MATCH_MGT: /* true if greater than */ - if (pkg_desc->package.elements[index]->number.value - <= V1_desc->number.value) + if (pkg_desc->package.elements[index]->integer.value + <= V1_desc->integer.value) { continue; } @@ -426,7 +426,7 @@ acpi_aml_exec_match ( } - switch(op2_desc->number.value) + switch(op2_desc->integer.value) { case MATCH_MTR: @@ -436,8 +436,8 @@ acpi_aml_exec_match ( case MATCH_MEQ: - if (pkg_desc->package.elements[index]->number.value - != V2_desc->number.value) + if (pkg_desc->package.elements[index]->integer.value + != V2_desc->integer.value) { continue; } @@ -446,8 +446,8 @@ acpi_aml_exec_match ( case MATCH_MLE: - if (pkg_desc->package.elements[index]->number.value - > V2_desc->number.value) + if (pkg_desc->package.elements[index]->integer.value + > V2_desc->integer.value) { continue; } @@ -456,8 +456,8 @@ acpi_aml_exec_match ( case MATCH_MLT: - if (pkg_desc->package.elements[index]->number.value - >= V2_desc->number.value) + if (pkg_desc->package.elements[index]->integer.value + >= V2_desc->integer.value) { continue; } @@ -466,8 +466,8 @@ acpi_aml_exec_match ( case MATCH_MGE: - if (pkg_desc->package.elements[index]->number.value - < V2_desc->number.value) + if (pkg_desc->package.elements[index]->integer.value + < V2_desc->integer.value) { continue; } @@ -476,8 +476,8 @@ acpi_aml_exec_match ( case MATCH_MGT: - if (pkg_desc->package.elements[index]->number.value - <= V2_desc->number.value) + if (pkg_desc->package.elements[index]->integer.value + <= V2_desc->integer.value) { continue; } @@ -497,7 +497,7 @@ acpi_aml_exec_match ( /* Match_value is the return value */ - ret_desc->number.value = match_value; + ret_desc->integer.value = match_value; cleanup: diff --git a/drivers/acpi/interpreter/ammonad.c b/drivers/acpi/interpreter/ammonad.c index df9671c06..ea8834a5c 100644 --- a/drivers/acpi/interpreter/ammonad.c +++ b/drivers/acpi/interpreter/ammonad.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: ammonad - ACPI AML (p-code) execution for monadic operators - * $Revision: 88 $ + * $Revision: 89 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -180,7 +180,7 @@ acpi_aml_exec_monadic1 ( case AML_SLEEP_OP: - acpi_aml_system_do_suspend ((u32) obj_desc->number.value); + acpi_aml_system_do_suspend ((u32) obj_desc->integer.value); break; @@ -188,7 +188,7 @@ acpi_aml_exec_monadic1 ( case AML_STALL_OP: - acpi_aml_system_do_stall ((u32) obj_desc->number.value); + acpi_aml_system_do_stall ((u32) obj_desc->integer.value); break; @@ -267,7 +267,7 @@ acpi_aml_exec_monadic2_r ( case AML_TO_BCD_OP: case AML_COND_REF_OF_OP: - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!ret_desc) { status = AE_NO_MEMORY; goto cleanup; @@ -283,7 +283,7 @@ acpi_aml_exec_monadic2_r ( case AML_BIT_NOT_OP: - ret_desc->number.value = ~obj_desc->number.value; + ret_desc->integer.value = ~obj_desc->integer.value; break; @@ -291,17 +291,17 @@ acpi_aml_exec_monadic2_r ( case AML_FIND_SET_LEFT_BIT_OP: - ret_desc->number.value = obj_desc->number.value; + ret_desc->integer.value = obj_desc->integer.value; /* * Acpi specification describes Integer type as a little * endian unsigned value, so this boundry condition is valid. */ - for (res_val = 0; ret_desc->number.value && res_val < ACPI_INTEGER_BIT_SIZE; ++res_val) { - ret_desc->number.value >>= 1; + for (res_val = 0; ret_desc->integer.value && res_val < ACPI_INTEGER_BIT_SIZE; ++res_val) { + ret_desc->integer.value >>= 1; } - ret_desc->number.value = res_val; + ret_desc->integer.value = res_val; break; @@ -309,19 +309,19 @@ acpi_aml_exec_monadic2_r ( case AML_FIND_SET_RIGHT_BIT_OP: - ret_desc->number.value = obj_desc->number.value; + ret_desc->integer.value = obj_desc->integer.value; /* * Acpi specification describes Integer type as a little * endian unsigned value, so this boundry condition is valid. */ - for (res_val = 0; ret_desc->number.value && res_val < ACPI_INTEGER_BIT_SIZE; ++res_val) { - ret_desc->number.value <<= 1; + for (res_val = 0; ret_desc->integer.value && res_val < ACPI_INTEGER_BIT_SIZE; ++res_val) { + ret_desc->integer.value <<= 1; } /* Since returns must be 1-based, subtract from 33 (65) */ - ret_desc->number.value = res_val == 0 ? 0 : (ACPI_INTEGER_BIT_SIZE + 1) - res_val; + ret_desc->integer.value = res_val == 0 ? 0 : (ACPI_INTEGER_BIT_SIZE + 1) - res_val; break; @@ -332,11 +332,11 @@ acpi_aml_exec_monadic2_r ( /* * The 64-bit ACPI integer can hold 16 4-bit BCD integers */ - ret_desc->number.value = 0; + ret_desc->integer.value = 0; for (i = 0; i < ACPI_MAX_BCD_DIGITS; i++) { /* Get one BCD digit */ - digit = (ACPI_INTEGER) ((obj_desc->number.value >> (i * 4)) & 0xF); + digit = (ACPI_INTEGER) ((obj_desc->integer.value >> (i * 4)) & 0xF); /* Check the range of the digit */ @@ -352,7 +352,7 @@ acpi_aml_exec_monadic2_r ( digit *= 10; } - ret_desc->number.value += digit; + ret_desc->integer.value += digit; } } break; @@ -363,16 +363,16 @@ acpi_aml_exec_monadic2_r ( case AML_TO_BCD_OP: - if (obj_desc->number.value > ACPI_MAX_BCD_VALUE) { + if (obj_desc->integer.value > ACPI_MAX_BCD_VALUE) { status = AE_AML_NUMERIC_OVERFLOW; goto cleanup; } - ret_desc->number.value = 0; + ret_desc->integer.value = 0; for (i = 0; i < ACPI_MAX_BCD_DIGITS; i++) { /* Divide by nth factor of 10 */ - digit = obj_desc->number.value; + digit = obj_desc->integer.value; for (j = 0; j < i; j++) { digit /= 10; } @@ -380,7 +380,7 @@ acpi_aml_exec_monadic2_r ( /* Create the BCD digit */ if (digit > 0) { - ret_desc->number.value += (ACPI_MODULO (digit, 10) << (i * 4)); + ret_desc->integer.value += (ACPI_MODULO (digit, 10) << (i * 4)); } } break; @@ -402,7 +402,7 @@ acpi_aml_exec_monadic2_r ( * return FALSE */ - ret_desc->number.value = 0; + ret_desc->integer.value = 0; /* * Must delete the result descriptor since there is no reference @@ -424,7 +424,7 @@ acpi_aml_exec_monadic2_r ( /* The object exists in the namespace, return TRUE */ - ret_desc->number.value = ACPI_INTEGER_MAX; + ret_desc->integer.value = ACPI_INTEGER_MAX; goto cleanup; break; @@ -579,13 +579,13 @@ acpi_aml_exec_monadic2 ( case AML_LNOT_OP: - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!ret_desc) { status = AE_NO_MEMORY; goto cleanup; } - ret_desc->number.value = !obj_desc->number.value; + ret_desc->integer.value = !obj_desc->integer.value; break; @@ -638,10 +638,10 @@ acpi_aml_exec_monadic2 ( /* Do the actual increment or decrement */ if (AML_INCREMENT_OP == opcode) { - ret_desc->number.value++; + ret_desc->integer.value++; } else { - ret_desc->number.value--; + ret_desc->integer.value--; } /* Store the result back in the original descriptor */ @@ -672,7 +672,7 @@ acpi_aml_exec_monadic2 ( /* Constants are of type Number */ - type = ACPI_TYPE_NUMBER; + type = ACPI_TYPE_INTEGER; break; @@ -733,13 +733,13 @@ acpi_aml_exec_monadic2 ( /* Allocate a descriptor to hold the type. */ - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!ret_desc) { status = AE_NO_MEMORY; goto cleanup; } - ret_desc->number.value = type; + ret_desc->integer.value = type; break; @@ -793,13 +793,13 @@ acpi_aml_exec_monadic2 ( * object to hold the value */ - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!ret_desc) { status = AE_NO_MEMORY; goto cleanup; } - ret_desc->number.value = value; + ret_desc->integer.value = value; break; @@ -910,14 +910,14 @@ acpi_aml_exec_monadic2 ( * sub-buffer of the main buffer, it is only a pointer to a * single element (byte) of the buffer! */ - ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + ret_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!ret_desc) { status = AE_NO_MEMORY; goto cleanup; } tmp_desc = obj_desc->reference.object; - ret_desc->number.value = + ret_desc->integer.value = tmp_desc->buffer.pointer[obj_desc->reference.offset]; /* TBD: [Investigate] (see below) Don't add an additional diff --git a/drivers/acpi/interpreter/amnames.c b/drivers/acpi/interpreter/amnames.c index ea4c26e08..07be1a37c 100644 --- a/drivers/acpi/interpreter/amnames.c +++ b/drivers/acpi/interpreter/amnames.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amnames - interpreter/scanner name load/execute - * $Revision: 71 $ + * $Revision: 73 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/interpreter/amprep.c b/drivers/acpi/interpreter/amprep.c index 266cb0105..2a56d4826 100644 --- a/drivers/acpi/interpreter/amprep.c +++ b/drivers/acpi/interpreter/amprep.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amprep - ACPI AML (p-code) execution - field prep utilities - * $Revision: 72 $ + * $Revision: 73 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/interpreter/amregion.c b/drivers/acpi/interpreter/amregion.c index dfe4fab85..1a2e7351d 100644 --- a/drivers/acpi/interpreter/amregion.c +++ b/drivers/acpi/interpreter/amregion.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amregion - ACPI default Op_region (address space) handlers - * $Revision: 41 $ + * $Revision: 44 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -99,8 +99,8 @@ acpi_aml_system_memory_space_handler ( */ if ((address < mem_info->mapped_physical_address) || - ((address + length) > - (mem_info->mapped_physical_address + mem_info->mapped_length))) + (((ACPI_INTEGER) address + length) > + ((ACPI_INTEGER) mem_info->mapped_physical_address + mem_info->mapped_length))) { /* * The request cannot be resolved by the current memory mapping; @@ -139,7 +139,7 @@ acpi_aml_system_memory_space_handler ( /* TBD: should these pointers go to 64-bit in all cases ? */ logical_addr_ptr = mem_info->mapped_logical_address + - (address - mem_info->mapped_physical_address); + ((ACPI_INTEGER) address - (ACPI_INTEGER) mem_info->mapped_physical_address); /* Perform the memory read or write */ diff --git a/drivers/acpi/interpreter/amresnte.c b/drivers/acpi/interpreter/amresnte.c index 20c6a0b11..0dc67b613 100644 --- a/drivers/acpi/interpreter/amresnte.c +++ b/drivers/acpi/interpreter/amresnte.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amresnte - AML Interpreter object resolution - * $Revision: 25 $ + * $Revision: 27 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -55,7 +55,7 @@ * can be either a pointer to an actual internal object or a pointer into the * AML stream itself. These types are currently: * - * ACPI_TYPE_NUMBER + * ACPI_TYPE_INTEGER * ACPI_TYPE_STRING * ACPI_TYPE_BUFFER * ACPI_TYPE_MUTEX @@ -208,14 +208,14 @@ acpi_aml_resolve_node_to_value ( break; - case ACPI_TYPE_NUMBER: + case ACPI_TYPE_INTEGER: /* * The Node has an attached internal object, make sure that it's a * number */ - if (ACPI_TYPE_NUMBER != val_desc->common.type) { + if (ACPI_TYPE_INTEGER != val_desc->common.type) { return (AE_AML_OPERAND_TYPE); } @@ -244,7 +244,7 @@ acpi_aml_resolve_node_to_value ( object_type = ACPI_TYPE_BUFFER; } else { - object_type = ACPI_TYPE_NUMBER; + object_type = ACPI_TYPE_INTEGER; } /* @@ -282,7 +282,7 @@ acpi_aml_resolve_node_to_value ( return (status); } - obj_desc->number.value = temp_val; + obj_desc->integer.value = temp_val; } @@ -330,12 +330,12 @@ acpi_aml_resolve_node_to_value ( /* Create an object for the result */ - obj_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + obj_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!obj_desc) { return (AE_NO_MEMORY); } - obj_desc->number.value = temp_val; + obj_desc->integer.value = temp_val; break; @@ -378,12 +378,12 @@ acpi_aml_resolve_node_to_value ( /* Create an object for the result */ - obj_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + obj_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!obj_desc) { return (AE_NO_MEMORY); } - obj_desc->number.value = temp_val; + obj_desc->integer.value = temp_val; break; @@ -477,12 +477,12 @@ acpi_aml_resolve_node_to_value ( /* Create object for result */ - obj_desc = acpi_cm_create_internal_object (ACPI_TYPE_NUMBER); + obj_desc = acpi_cm_create_internal_object (ACPI_TYPE_INTEGER); if (!obj_desc) { return (AE_NO_MEMORY); } - obj_desc->number.value = temp_val; + obj_desc->integer.value = temp_val; /* Truncate value if we are executing from a 32-bit ACPI table */ diff --git a/drivers/acpi/interpreter/amresolv.c b/drivers/acpi/interpreter/amresolv.c index 2fa59e9bb..c08a4731c 100644 --- a/drivers/acpi/interpreter/amresolv.c +++ b/drivers/acpi/interpreter/amresolv.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amresolv - AML Interpreter object resolution - * $Revision: 78 $ + * $Revision: 81 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -118,19 +118,19 @@ acpi_aml_get_field_unit_value ( mask = ACPI_UINT32_MAX; } - result_desc->number.type = (u8) ACPI_TYPE_NUMBER; + result_desc->integer.type = (u8) ACPI_TYPE_INTEGER; /* Get the 32 bit value at the location */ - MOVE_UNALIGNED32_TO_32 (&result_desc->number.value, location); + MOVE_UNALIGNED32_TO_32 (&result_desc->integer.value, location); /* * Shift the 32-bit word containing the field, and mask off the * resulting value */ - result_desc->number.value = - (result_desc->number.value >> field_desc->field_unit.bit_offset) & mask; + result_desc->integer.value = + (result_desc->integer.value >> field_desc->field_unit.bit_offset) & mask; /* Release global lock if we acquired it earlier */ @@ -150,7 +150,7 @@ acpi_aml_get_field_unit_value ( * * RETURN: Status * - * DESCRIPTION: Convert Reference entries on Obj_stack to Rvalues + * DESCRIPTION: Convert Reference objects to values * ******************************************************************************/ @@ -279,7 +279,7 @@ acpi_aml_resolve_object_to_value ( acpi_cm_remove_reference (stack_desc); *stack_ptr = obj_desc; - if (ACPI_TYPE_NUMBER == obj_desc->common.type) { + if (ACPI_TYPE_INTEGER == obj_desc->common.type) { /* Value is a Number */ } @@ -311,7 +311,7 @@ acpi_aml_resolve_object_to_value ( acpi_cm_remove_reference (stack_desc); *stack_ptr = obj_desc; - if (ACPI_TYPE_NUMBER == obj_desc->common.type) { + if (ACPI_TYPE_INTEGER == obj_desc->common.type) { /* Value is a Number */ } @@ -326,22 +326,22 @@ acpi_aml_resolve_object_to_value ( case AML_ZERO_OP: - stack_desc->common.type = (u8) ACPI_TYPE_NUMBER; - stack_desc->number.value = 0; + stack_desc->common.type = (u8) ACPI_TYPE_INTEGER; + stack_desc->integer.value = 0; break; case AML_ONE_OP: - stack_desc->common.type = (u8) ACPI_TYPE_NUMBER; - stack_desc->number.value = 1; + stack_desc->common.type = (u8) ACPI_TYPE_INTEGER; + stack_desc->integer.value = 1; break; case AML_ONES_OP: - stack_desc->common.type = (u8) ACPI_TYPE_NUMBER; - stack_desc->number.value = ACPI_INTEGER_MAX; + stack_desc->common.type = (u8) ACPI_TYPE_INTEGER; + stack_desc->integer.value = ACPI_INTEGER_MAX; /* Truncate value if we are executing from a 32-bit ACPI table */ diff --git a/drivers/acpi/interpreter/amresop.c b/drivers/acpi/interpreter/amresop.c index 83fda4455..a837fd66b 100644 --- a/drivers/acpi/interpreter/amresop.c +++ b/drivers/acpi/interpreter/amresop.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amresop - AML Interpreter operand/object resolution - * $Revision: 18 $ + * $Revision: 22 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -214,8 +214,13 @@ acpi_aml_resolve_operands ( switch (this_arg_type) { - case ARGI_REFERENCE: /* Reference */ - case ARGI_TARGETREF: + case ARGI_REFERENCE: /* References */ + case ARGI_INTEGER_REF: + case ARGI_OBJECT_REF: + case ARGI_DEVICE_REF: + case ARGI_TARGETREF: /* TBD: must implement implicit conversion rules before store */ + case ARGI_FIXED_TARGET: /* No implicit conversion before store to target */ + case ARGI_SIMPLE_TARGET: /* Name, Local, or Arg - no implicit conversion */ /* Need an operand of type INTERNAL_TYPE_REFERENCE */ @@ -283,20 +288,6 @@ acpi_aml_resolve_operands ( * For the simple cases, only one type of resolved object * is allowed */ - case ARGI_NUMBER: /* Number */ - - /* Need an operand of type ACPI_TYPE_NUMBER */ - - type_needed = ACPI_TYPE_NUMBER; - break; - - case ARGI_BUFFER: - - /* Need an operand of type ACPI_TYPE_BUFFER */ - - type_needed = ACPI_TYPE_BUFFER; - break; - case ARGI_MUTEX: /* Need an operand of type ACPI_TYPE_MUTEX */ @@ -344,11 +335,69 @@ acpi_aml_resolve_operands ( * The more complex cases allow multiple resolved object types */ + case ARGI_INTEGER: /* Number */ + + /* + * Need an operand of type ACPI_TYPE_INTEGER, + * But we can implicitly convert from a STRING or BUFFER + */ + status = acpi_aml_convert_to_integer (stack_ptr, walk_state); + if (ACPI_FAILURE (status)) { + if (status == AE_TYPE) { + return (AE_AML_OPERAND_TYPE); + } + + return (status); + } + + goto next_operand; + break; + + + case ARGI_BUFFER: + + /* + * Need an operand of type ACPI_TYPE_BUFFER, + * But we can implicitly convert from a STRING or INTEGER + */ + status = acpi_aml_convert_to_buffer (stack_ptr, walk_state); + if (ACPI_FAILURE (status)) { + if (status == AE_TYPE) { + return (AE_AML_OPERAND_TYPE); + } + + return (status); + } + + goto next_operand; + break; + + case ARGI_STRING: - /* Need an operand of type ACPI_TYPE_STRING or ACPI_TYPE_BUFFER */ + /* + * Need an operand of type ACPI_TYPE_STRING, + * But we can implicitly convert from a BUFFER or INTEGER + */ + status = acpi_aml_convert_to_string (stack_ptr, walk_state); + if (ACPI_FAILURE (status)) { + if (status == AE_TYPE) { + return (AE_AML_OPERAND_TYPE); + } + + return (status); + } + + goto next_operand; + break; + + + case ARGI_COMPUTEDATA: + + /* Need an operand of type INTEGER, STRING or BUFFER */ - if ((ACPI_TYPE_STRING != (*stack_ptr)->common.type) && + if ((ACPI_TYPE_INTEGER != (*stack_ptr)->common.type) && + (ACPI_TYPE_STRING != (*stack_ptr)->common.type) && (ACPI_TYPE_BUFFER != (*stack_ptr)->common.type)) { return (AE_AML_OPERAND_TYPE); diff --git a/drivers/acpi/interpreter/amstore.c b/drivers/acpi/interpreter/amstore.c index 8887e0997..9f350bb29 100644 --- a/drivers/acpi/interpreter/amstore.c +++ b/drivers/acpi/interpreter/amstore.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amstore - AML Interpreter object store support - * $Revision: 117 $ + * $Revision: 121 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -64,12 +64,7 @@ acpi_aml_exec_store ( ACPI_WALK_STATE *walk_state) { ACPI_STATUS status = AE_OK; - ACPI_OPERAND_OBJECT *delete_dest_desc = NULL; - ACPI_OPERAND_OBJECT *tmp_desc; - ACPI_NAMESPACE_NODE *node = NULL; - u8 value = 0; - u32 length; - u32 i; + ACPI_OPERAND_OBJECT *ref_desc = dest_desc; /* Validate parameters */ @@ -78,27 +73,23 @@ acpi_aml_exec_store ( return (AE_AML_NO_OPERAND); } - /* Examine the datatype of the Dest_desc */ + /* Dest_desc can be either a namespace node or an ACPI object */ if (VALID_DESCRIPTOR_TYPE (dest_desc, ACPI_DESC_TYPE_NAMED)) { - /* Dest is an ACPI_HANDLE, create a new object */ - - node = (ACPI_NAMESPACE_NODE *) dest_desc; - dest_desc = acpi_cm_create_internal_object (INTERNAL_TYPE_REFERENCE); - if (!dest_desc) { - /* Allocation failure */ - - return (AE_NO_MEMORY); - } + /* + * Dest is a namespace node, + * Storing an object into a Name "container" + */ + status = acpi_aml_store_object_to_node (val_desc, + (ACPI_NAMESPACE_NODE *) dest_desc, walk_state); - /* Build a new Reference wrapper around the handle */ + /* All done, that's it */ - dest_desc->reference.op_code = AML_NAME_OP; - dest_desc->reference.object = node; + return (status); } - /* Destination object must be of type Reference */ + /* Destination object must be an object of type Reference */ if (dest_desc->common.type != INTERNAL_TYPE_REFERENCE) { /* Destination is not an Reference */ @@ -106,36 +97,130 @@ acpi_aml_exec_store ( return (AE_AML_OPERAND_TYPE); } - /* Examine the Reference opcode */ - switch (dest_desc->reference.op_code) + /* + * Examine the Reference opcode. These cases are handled: + * + * 1) Store to Name (Change the object associated with a name) + * 2) Store to an indexed area of a Buffer or Package + * 3) Store to a Method Local or Arg + * 4) Store to the debug object + * 5) Store to a constant -- a noop + */ + + switch (ref_desc->reference.op_code) { case AML_NAME_OP: - /* - * Storing into a Name - */ - delete_dest_desc = dest_desc; - status = acpi_aml_store_object_to_node (val_desc, dest_desc->reference.object, - walk_state); + /* Storing an object into a Name "container" */ - break; /* Case Name_op */ + status = acpi_aml_store_object_to_node (val_desc, ref_desc->reference.object, + walk_state); + break; case AML_INDEX_OP: - delete_dest_desc = dest_desc; + /* Storing to an Index (pointer into a packager or buffer) */ + + status = acpi_aml_store_object_to_index (val_desc, ref_desc, walk_state); + break; + + + case AML_LOCAL_OP: + + status = acpi_ds_method_data_set_value (MTH_TYPE_LOCAL, + (ref_desc->reference.offset), val_desc, walk_state); + break; + + + case AML_ARG_OP: + + status = acpi_ds_method_data_set_value (MTH_TYPE_ARG, + (ref_desc->reference.offset), val_desc, walk_state); + break; + + + case AML_DEBUG_OP: /* - * Valid source value and destination reference pointer. + * Storing to the Debug object causes the value stored to be + * displayed and otherwise has no effect -- see ACPI Specification * - * ACPI Specification 1.0B section 15.2.3.4.2.13: - * Destination should point to either a buffer or a package + * TBD: print known object types "prettier". */ + break; + + + case AML_ZERO_OP: + case AML_ONE_OP: + case AML_ONES_OP: + /* - * Actually, storing to a package is not so simple. The source must be + * Storing to a constant is a no-op -- see ACPI Specification + * Delete the reference descriptor, however + */ + break; + + + default: + + /* TBD: [Restructure] use object dump routine !! */ + + status = AE_AML_INTERNAL; + break; + + } /* switch (Ref_desc->Reference.Op_code) */ + + + /* Always delete the reference descriptor object */ + + if (ref_desc) { + acpi_cm_remove_reference (ref_desc); + } + + return (status); +} + + +/******************************************************************************* + * + * FUNCTION: Acpi_aml_store_object_to_index + * + * PARAMETERS: *Val_desc - Value to be stored + * *Node - Named object to recieve the value + * + * RETURN: Status + * + * DESCRIPTION: Store the object to the named object. + * + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_store_object_to_index ( + ACPI_OPERAND_OBJECT *val_desc, + ACPI_OPERAND_OBJECT *dest_desc, + ACPI_WALK_STATE *walk_state) +{ + ACPI_STATUS status = AE_OK; + ACPI_OPERAND_OBJECT *obj_desc; + u32 length; + u32 i; + u8 value = 0; + + + /* + * Destination must be a reference pointer, and + * must point to either a buffer or a package + */ + + switch (dest_desc->reference.target_type) + { + case ACPI_TYPE_PACKAGE: + /* + * Storing to a package element is not simple. The source must be * evaluated and converted to the type of the destination and then the * source is copied into the destination - we can't just point to the * source object. @@ -145,8 +230,8 @@ acpi_aml_exec_store ( * The object at *(Dest_desc->Reference.Where) is the * element within the package that is to be modified. */ - tmp_desc = *(dest_desc->reference.where); - if (tmp_desc) { + obj_desc = *(dest_desc->reference.where); + if (obj_desc) { /* * If the Destination element is a package, we will delete * that object and construct a new one. @@ -155,41 +240,39 @@ acpi_aml_exec_store ( * to be packages? * && (Val_desc->Common.Type == ACPI_TYPE_PACKAGE) */ - if (tmp_desc->common.type == ACPI_TYPE_PACKAGE) { + if (obj_desc->common.type == ACPI_TYPE_PACKAGE) { /* * Take away the reference for being part of a package and * delete */ - acpi_cm_remove_reference (tmp_desc); - acpi_cm_remove_reference (tmp_desc); + acpi_cm_remove_reference (obj_desc); + acpi_cm_remove_reference (obj_desc); - tmp_desc = NULL; + obj_desc = NULL; } } - if (!tmp_desc) { + if (!obj_desc) { /* - * If the Tmp_desc is NULL, that means an uninitialized package - * has been used as a destination, therefore, we must create - * the destination element to match the type of the source - * element NOTE: Val_desc can be of any type. + * If the Obj_desc is NULL, it means that an uninitialized package + * element has been used as a destination (this is OK), therefore, + * we must create the destination element to match the type of the + * source element NOTE: Val_desc can be of any type. */ - tmp_desc = acpi_cm_create_internal_object (val_desc->common.type); - if (!tmp_desc) { - status = AE_NO_MEMORY; - goto cleanup; + obj_desc = acpi_cm_create_internal_object (val_desc->common.type); + if (!obj_desc) { + return (AE_NO_MEMORY); } /* * If the source is a package, copy the source to the new dest */ - if (ACPI_TYPE_PACKAGE == tmp_desc->common.type) { + if (ACPI_TYPE_PACKAGE == obj_desc->common.type) { status = acpi_aml_build_copy_internal_package_object ( - val_desc, tmp_desc, walk_state); + val_desc, obj_desc, walk_state); if (ACPI_FAILURE (status)) { - acpi_cm_remove_reference (tmp_desc); - tmp_desc = NULL; - goto cleanup; + acpi_cm_remove_reference (obj_desc); + return (status); } } @@ -199,38 +282,31 @@ acpi_aml_exec_store ( * part of the parent package */ - *(dest_desc->reference.where) = tmp_desc; - acpi_cm_add_reference (tmp_desc); + *(dest_desc->reference.where) = obj_desc; + acpi_cm_add_reference (obj_desc); } - if (ACPI_TYPE_PACKAGE != tmp_desc->common.type) { + if (ACPI_TYPE_PACKAGE != obj_desc->common.type) { /* * The destination element is not a package, so we need to * convert the contents of the source (Val_desc) and copy into - * the destination (Tmp_desc) + * the destination (Obj_desc) */ - status = acpi_aml_store_object_to_object (val_desc, tmp_desc, + status = acpi_aml_store_object_to_object (val_desc, obj_desc, walk_state); if (ACPI_FAILURE (status)) { /* * An error occurrered when copying the internal object * so delete the reference. */ - status = AE_AML_OPERAND_TYPE; + return (AE_AML_OPERAND_TYPE); } } - - break; } + break; - /* - * Check that the destination is a Buffer Field type - */ - if (dest_desc->reference.target_type != ACPI_TYPE_BUFFER_FIELD) { - status = AE_AML_OPERAND_TYPE; - break; - } + case ACPI_TYPE_BUFFER_FIELD: /* * Storing into a buffer at a location defined by an Index. * @@ -239,13 +315,11 @@ acpi_aml_exec_store ( */ /* - * Set the Tmp_desc to the destination object and type check. + * Set the Obj_desc to the destination object and type check. */ - tmp_desc = dest_desc->reference.object; - - if (tmp_desc->common.type != ACPI_TYPE_BUFFER) { - status = AE_AML_OPERAND_TYPE; - break; + obj_desc = dest_desc->reference.object; + if (obj_desc->common.type != ACPI_TYPE_BUFFER) { + return (AE_AML_OPERAND_TYPE); } /* @@ -256,15 +330,15 @@ acpi_aml_exec_store ( switch (val_desc->common.type) { /* - * If the type is Integer, the Length is 4. + * If the type is Integer, assign bytewise * This loop to assign each of the elements is somewhat - * backward because of the Big Endian-ness of IA-64 + * backward because of the Big Endian-ness of IA-64 */ - case ACPI_TYPE_NUMBER: - length = 4; + case ACPI_TYPE_INTEGER: + length = sizeof (ACPI_INTEGER); for (i = length; i != 0; i--) { - value = (u8)(val_desc->number.value >> (MUL_8 (i - 1))); - tmp_desc->buffer.pointer[dest_desc->reference.offset] = value; + value = (u8)(val_desc->integer.value >> (MUL_8 (i - 1))); + obj_desc->buffer.pointer[dest_desc->reference.offset] = value; } break; @@ -276,7 +350,7 @@ acpi_aml_exec_store ( length = val_desc->buffer.length; for (i = 0; i < length; i++) { value = *(val_desc->buffer.pointer + i); - tmp_desc->buffer.pointer[dest_desc->reference.offset] = value; + obj_desc->buffer.pointer[dest_desc->reference.offset] = value; } break; @@ -288,7 +362,7 @@ acpi_aml_exec_store ( length = val_desc->string.length; for (i = 0; i < length; i++) { value = *(val_desc->string.pointer + i); - tmp_desc->buffer.pointer[dest_desc->reference.offset] = value; + obj_desc->buffer.pointer[dest_desc->reference.offset] = value; } break; @@ -299,80 +373,207 @@ acpi_aml_exec_store ( status = AE_AML_OPERAND_TYPE; break; } + break; - /* - * If we had an error, break out of this case statement. - */ - if (ACPI_FAILURE (status)) { - break; - } - - /* - * Set the return pointer - */ - dest_desc = tmp_desc; + default: + status = AE_AML_OPERAND_TYPE; break; + } - case AML_ZERO_OP: - case AML_ONE_OP: - case AML_ONES_OP: - /* - * Storing to a constant is a no-op -- see ACPI Specification - * Delete the result descriptor. - */ + return (status); +} - delete_dest_desc = dest_desc; - break; + +/******************************************************************************* + * + * FUNCTION: Acpi_aml_store_object_to_node + * + * PARAMETERS: *Source_desc - Value to be stored + * *Node - Named object to recieve the value + * + * RETURN: Status + * + * DESCRIPTION: Store the object to the named object. + * + * The Assignment of an object to a named object is handled here + * The val passed in will replace the current value (if any) + * with the input value. + * + * When storing into an object the data is converted to the + * target object type then stored in the object. This means + * that the target object type (for an initialized target) will + * not be changed by a store operation. + * + * NOTE: the global lock is acquired early. This will result + * in the global lock being held a bit longer. Also, if the + * function fails during set up we may get the lock when we + * don't really need it. I don't think we care. + * + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_store_object_to_node ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_NAMESPACE_NODE *node, + ACPI_WALK_STATE *walk_state) +{ + ACPI_STATUS status = AE_OK; + ACPI_OPERAND_OBJECT *target_desc; + OBJECT_TYPE_INTERNAL target_type = ACPI_TYPE_ANY; - case AML_LOCAL_OP: + /* + * Assuming the parameters were already validated + */ + ACPI_ASSERT((node) && (source_desc)); - status = acpi_ds_method_data_set_value (MTH_TYPE_LOCAL, - (dest_desc->reference.offset), val_desc, walk_state); - delete_dest_desc = dest_desc; - break; + /* + * Get current type of the node, and object attached to Node + */ + target_type = acpi_ns_get_type (node); + target_desc = acpi_ns_get_attached_object (node); - case AML_ARG_OP: - status = acpi_ds_method_data_set_value (MTH_TYPE_ARG, - (dest_desc->reference.offset), val_desc, walk_state); - delete_dest_desc = dest_desc; + /* + * Resolve the source object to an actual value + * (If it is a reference object) + */ + status = acpi_aml_resolve_object (&source_desc, target_type, walk_state); + if (ACPI_FAILURE (status)) { + return (status); + } + + + /* + * Do the actual store operation + */ + switch (target_type) + { + case INTERNAL_TYPE_DEF_FIELD: + + /* Raw data copy for target types Integer/String/Buffer */ + + status = acpi_aml_copy_data_to_named_field (source_desc, node); break; - case AML_DEBUG_OP: + case ACPI_TYPE_INTEGER: + case ACPI_TYPE_STRING: + case ACPI_TYPE_BUFFER: + case INTERNAL_TYPE_BANK_FIELD: + case INTERNAL_TYPE_INDEX_FIELD: + case ACPI_TYPE_FIELD_UNIT: /* - * Storing to the Debug object causes the value stored to be - * displayed and otherwise has no effect -- see ACPI Specification + * These target types are all of type Integer/String/Buffer, and + * therefore support implicit conversion before the store. + * + * Copy and/or convert the source object to a new target object */ + status = acpi_aml_store_object (source_desc, target_type, &target_desc, walk_state); + if (ACPI_FAILURE (status)) { + return (status); + } - delete_dest_desc = dest_desc; + /* + * Store the new Target_desc as the new value of the Name, and set + * the Name's type to that of the value being stored in it. + * Source_desc reference count is incremented by Attach_object. + */ + status = acpi_ns_attach_object (node, target_desc, target_type); break; default: - /* TBD: [Restructure] use object dump routine !! */ + /* No conversions for all other types. Just attach the source object */ - delete_dest_desc = dest_desc; - status = AE_AML_INTERNAL; + status = acpi_ns_attach_object (node, source_desc, source_desc->common.type); - } /* switch(Dest_desc->Reference.Op_code) */ + break; + } -cleanup: + return (status); +} - /* Cleanup and exit*/ - if (delete_dest_desc) { - acpi_cm_remove_reference (delete_dest_desc); +/******************************************************************************* + * + * FUNCTION: Acpi_aml_store_object_to_object + * + * PARAMETERS: *Source_desc - Value to be stored + * *Dest_desc - Object to receive the value + * + * RETURN: Status + * + * DESCRIPTION: Store an object to another object. + * + * The Assignment of an object to another (not named) object + * is handled here. + * The val passed in will replace the current value (if any) + * with the input value. + * + * When storing into an object the data is converted to the + * target object type then stored in the object. This means + * that the target object type (for an initialized target) will + * not be changed by a store operation. + * + * This module allows destination types of Number, String, + * and Buffer. + * + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_store_object_to_object ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *dest_desc, + ACPI_WALK_STATE *walk_state) +{ + ACPI_STATUS status = AE_OK; + OBJECT_TYPE_INTERNAL destination_type = dest_desc->common.type; + + + /* + * Assuming the parameters are valid! + */ + ACPI_ASSERT((dest_desc) && (source_desc)); + + + /* + * From this interface, we only support Integers/Strings/Buffers + */ + switch (destination_type) + { + case ACPI_TYPE_INTEGER: + case ACPI_TYPE_STRING: + case ACPI_TYPE_BUFFER: + break; + + default: + return (AE_NOT_IMPLEMENTED); } + + /* + * Resolve the source object to an actual value + * (If it is a reference object) + */ + status = acpi_aml_resolve_object (&source_desc, destination_type, walk_state); + if (ACPI_FAILURE (status)) { + return (status); + } + + + /* + * Copy and/or convert the source object to the destination object + */ + status = acpi_aml_store_object (source_desc, destination_type, &dest_desc, walk_state); + + return (status); } - diff --git a/drivers/acpi/interpreter/amstoren.c b/drivers/acpi/interpreter/amstoren.c index 91b2095c5..eb62ed2f1 100644 --- a/drivers/acpi/interpreter/amstoren.c +++ b/drivers/acpi/interpreter/amstoren.c @@ -2,13 +2,13 @@ /****************************************************************************** * * Module Name: amstoren - AML Interpreter object store support, - * Store to Node (namespace object) - * $Revision: 24 $ + * Store to Node (namespace object) + * $Revision: 28 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -41,127 +41,72 @@ /******************************************************************************* * - * FUNCTION: Acpi_aml_store_object_to_node + * FUNCTION: Acpi_aml_resolve_object * - * PARAMETERS: *Val_desc - Value to be stored - * *Node - Named object to recieve the value + * PARAMETERS: Source_desc_ptr - Pointer to the source object + * Target_type - Current type of the target + * Walk_state - Current walk state * - * RETURN: Status - * - * DESCRIPTION: Store the object to the named object. - * - * The Assignment of an object to a named object is handled here - * The val passed in will replace the current value (if any) - * with the input value. + * RETURN: Status, resolved object in Source_desc_ptr. * - * When storing into an object the data is converted to the - * target object type then stored in the object. This means - * that the target object type (for an initialized target) will - * not be changed by a store operation. - * - * NOTE: the global lock is acquired early. This will result - * in the global lock being held a bit longer. Also, if the - * function fails during set up we may get the lock when we - * don't really need it. I don't think we care. + * DESCRIPTION: Resolve an object. If the object is a reference, dereference + * it and return the actual object in the Source_desc_ptr. * ******************************************************************************/ ACPI_STATUS -acpi_aml_store_object_to_node ( - ACPI_OPERAND_OBJECT *val_desc, - ACPI_NAMESPACE_NODE *node, +acpi_aml_resolve_object ( + ACPI_OPERAND_OBJECT **source_desc_ptr, + OBJECT_TYPE_INTERNAL target_type, ACPI_WALK_STATE *walk_state) { + ACPI_OPERAND_OBJECT *source_desc = *source_desc_ptr; ACPI_STATUS status = AE_OK; - u8 *buffer = NULL; - u32 length = 0; - u32 mask; - u32 new_value; - u8 locked = FALSE; - u8 *location=NULL; - ACPI_OPERAND_OBJECT *dest_desc; - OBJECT_TYPE_INTERNAL destination_type = ACPI_TYPE_ANY; /* - * Assuming the parameters are valid!!! + * Ensure we have a Source that can be stored in the target */ - ACPI_ASSERT((node) && (val_desc)); + switch (target_type) + { - destination_type = acpi_ns_get_type (node); + /* This case handles the "interchangeable" types Integer, String, and Buffer. */ /* - * First ensure we have a value that can be stored in the target + * These cases all require only Integers or values that + * can be converted to Integers (Strings or Buffers) */ - switch (destination_type) - { - /* Type of Name's existing value */ - - case INTERNAL_TYPE_ALIAS: - - /* - * Aliases are resolved by Acpi_aml_prep_operands - */ - - status = AE_AML_INTERNAL; - break; - - + case ACPI_TYPE_INTEGER: + case ACPI_TYPE_FIELD_UNIT: case INTERNAL_TYPE_BANK_FIELD: case INTERNAL_TYPE_INDEX_FIELD: - case ACPI_TYPE_FIELD_UNIT: - case ACPI_TYPE_NUMBER: - - /* - * These cases all require only number values or values that - * can be converted to numbers. - * - * If value is not a Number, try to resolve it to one. - */ - - if (val_desc->common.type != ACPI_TYPE_NUMBER) { - /* - * Initially not a number, convert - */ - status = acpi_aml_resolve_to_value (&val_desc, walk_state); - if (ACPI_SUCCESS (status) && - (val_desc->common.type != ACPI_TYPE_NUMBER)) - { - /* - * Conversion successful but still not a number - */ - status = AE_AML_OPERAND_TYPE; - } - } - - break; + /* + * Stores into a Field/Region or into a Buffer/String + * are all essentially the same. + */ case ACPI_TYPE_STRING: case ACPI_TYPE_BUFFER: case INTERNAL_TYPE_DEF_FIELD: /* - * Storing into a Field in a region or into a buffer or into - * a string all is essentially the same. - * - * If value is not a valid type, try to resolve it to one. + * If Source_desc is not a valid type, try to resolve it to one. */ - - if ((val_desc->common.type != ACPI_TYPE_NUMBER) && - (val_desc->common.type != ACPI_TYPE_BUFFER) && - (val_desc->common.type != ACPI_TYPE_STRING)) + if ((source_desc->common.type != ACPI_TYPE_INTEGER) && + (source_desc->common.type != ACPI_TYPE_BUFFER) && + (source_desc->common.type != ACPI_TYPE_STRING)) { /* - * Initially not a valid type, convert + * Initially not a valid type, convert */ - status = acpi_aml_resolve_to_value (&val_desc, walk_state); + status = acpi_aml_resolve_to_value (source_desc_ptr, walk_state); if (ACPI_SUCCESS (status) && - (val_desc->common.type != ACPI_TYPE_NUMBER) && - (val_desc->common.type != ACPI_TYPE_BUFFER) && - (val_desc->common.type != ACPI_TYPE_STRING)) + (source_desc->common.type != ACPI_TYPE_INTEGER) && + (source_desc->common.type != ACPI_TYPE_BUFFER) && + (source_desc->common.type != ACPI_TYPE_STRING)) { /* - * Conversion successful but still not a valid type + * Conversion successful but still not a valid type */ status = AE_AML_OPERAND_TYPE; } @@ -169,347 +114,133 @@ acpi_aml_store_object_to_node ( break; - case ACPI_TYPE_PACKAGE: + case INTERNAL_TYPE_ALIAS: /* - * TBD: [Unhandled] Not real sure what to do here + * Aliases are resolved by Acpi_aml_prep_operands */ - status = AE_NOT_IMPLEMENTED; + status = AE_AML_INTERNAL; break; + case ACPI_TYPE_PACKAGE: default: /* - * All other types than Alias and the various Fields come here. - * Store Val_desc as the new value of the Name, and set - * the Name's type to that of the value being stored in it. - * Val_desc reference count is incremented by Attach_object. + * All other types than Alias and the various Fields come here, + * including the untyped case - ACPI_TYPE_ANY. */ - - status = acpi_ns_attach_object (node, val_desc, val_desc->common.type); - - goto clean_up_and_bail_out; break; } - /* Exit now if failure above */ + return (status); +} + - if (ACPI_FAILURE (status)) { - goto clean_up_and_bail_out; - } +/******************************************************************************* + * + * FUNCTION: Acpi_aml_store_object + * + * PARAMETERS: Source_desc - Object to store + * Target_type - Current type of the target + * Target_desc_ptr - Pointer to the target + * Walk_state - Current walk state + * + * RETURN: Status + * + * DESCRIPTION: "Store" an object to another object. This may include + * converting the source type to the target type (implicit + * conversion), and a copy of the value of the source to + * the target. + * + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_store_object ( + ACPI_OPERAND_OBJECT *source_desc, + OBJECT_TYPE_INTERNAL target_type, + ACPI_OPERAND_OBJECT **target_desc_ptr, + ACPI_WALK_STATE *walk_state) +{ + ACPI_OPERAND_OBJECT *target_desc = *target_desc_ptr; + ACPI_STATUS status; - /* - * Get descriptor for object attached to Node - */ - dest_desc = acpi_ns_get_attached_object (node); - if (!dest_desc) { - /* - * There is no existing object attached to this Node - */ - status = AE_AML_INTERNAL; - goto clean_up_and_bail_out; - } /* - * Make sure the destination Object is the same as the Node + * Perform the "implicit conversion" of the source to the current type + * of the target - As per the ACPI specification. + * + * If no conversion performed, Source_desc is left alone, otherwise it + * is updated with a new object. */ - if (dest_desc->common.type != (u8) destination_type) { - status = AE_AML_INTERNAL; - goto clean_up_and_bail_out; + status = acpi_aml_convert_to_target_type (target_type, &source_desc, walk_state); + if (ACPI_FAILURE (status)) { + return (status); } + /* - * Acpi_everything is ready to execute now, We have - * a value we can handle, just perform the update + * We now have two objects of identical types, and we can perform a + * copy of the *value* of the source object. */ - - switch (destination_type) + switch (target_type) { - /* Type of Name's existing value */ - - case INTERNAL_TYPE_BANK_FIELD: + case ACPI_TYPE_ANY: + case INTERNAL_TYPE_DEF_ANY: /* - * Get the global lock if needed + * The target namespace node is uninitialized (has no target object), + * and will take on the type of the source object */ - locked = acpi_aml_acquire_global_lock (dest_desc->bank_field.lock_rule); - - /* - * Set Bank value to select proper Bank - * Perform the update (Set Bank Select) - */ - - status = acpi_aml_access_named_field (ACPI_WRITE, - dest_desc->bank_field.bank_select, - &dest_desc->bank_field.value, - sizeof (dest_desc->bank_field.value)); - if (ACPI_SUCCESS (status)) { - /* Set bank select successful, set data value */ - - status = acpi_aml_access_named_field (ACPI_WRITE, - dest_desc->bank_field.bank_select, - &val_desc->bank_field.value, - sizeof (val_desc->bank_field.value)); - } + *target_desc_ptr = source_desc; break; - case INTERNAL_TYPE_DEF_FIELD: - - /* - * Get the global lock if needed - */ - locked = acpi_aml_acquire_global_lock (val_desc->field.lock_rule); - - /* - * Perform the update - */ + case ACPI_TYPE_INTEGER: - switch (val_desc->common.type) - { - case ACPI_TYPE_NUMBER: - buffer = (u8 *) &val_desc->number.value; - length = sizeof (val_desc->number.value); - break; - - case ACPI_TYPE_BUFFER: - buffer = (u8 *) val_desc->buffer.pointer; - length = val_desc->buffer.length; - break; - - case ACPI_TYPE_STRING: - buffer = (u8 *) val_desc->string.pointer; - length = val_desc->string.length; - break; - } - - status = acpi_aml_access_named_field (ACPI_WRITE, - node, buffer, length); - - break; /* Global Lock released below */ + target_desc->integer.value = source_desc->integer.value; + /* Truncate value if we are executing from a 32-bit ACPI table */ - case ACPI_TYPE_STRING: - - /* - * Perform the update - */ - - switch (val_desc->common.type) - { - case ACPI_TYPE_NUMBER: - buffer = (u8 *) &val_desc->number.value; - length = sizeof (val_desc->number.value); - break; - - case ACPI_TYPE_BUFFER: - buffer = (u8 *) val_desc->buffer.pointer; - length = val_desc->buffer.length; - break; - - case ACPI_TYPE_STRING: - buffer = (u8 *) val_desc->string.pointer; - length = val_desc->string.length; - break; - } - - /* - * Setting a string value replaces the old string - */ - - if (length < dest_desc->string.length) { - /* - * Zero fill, not willing to do pointer arithmetic for - * archetecture independance. Just clear the whole thing - */ - MEMSET(dest_desc->string.pointer, 0, dest_desc->string.length); - MEMCPY(dest_desc->string.pointer, buffer, length); - } - else { - /* - * Free the current buffer, then allocate a buffer - * large enough to hold the value - */ - if ( dest_desc->string.pointer && - !acpi_tb_system_table_pointer (dest_desc->string.pointer)) - { - /* - * Only free if not a pointer into the DSDT - */ - - acpi_cm_free(dest_desc->string.pointer); - } + acpi_aml_truncate_for32bit_table (target_desc, walk_state); + break; - dest_desc->string.pointer = acpi_cm_allocate (length + 1); - dest_desc->string.length = length; - if (!dest_desc->string.pointer) { - status = AE_NO_MEMORY; - goto clean_up_and_bail_out; - } + case ACPI_TYPE_FIELD_UNIT: - MEMCPY(dest_desc->string.pointer, buffer, length); - } + status = acpi_aml_copy_integer_to_field_unit (source_desc, target_desc); break; - case ACPI_TYPE_BUFFER: - - /* - * Perform the update to the buffer - */ - - switch (val_desc->common.type) - { - case ACPI_TYPE_NUMBER: - buffer = (u8 *) &val_desc->number.value; - length = sizeof (val_desc->number.value); - break; - - case ACPI_TYPE_BUFFER: - buffer = (u8 *) val_desc->buffer.pointer; - length = val_desc->buffer.length; - break; - - case ACPI_TYPE_STRING: - buffer = (u8 *) val_desc->string.pointer; - length = val_desc->string.length; - break; - } + case INTERNAL_TYPE_BANK_FIELD: - /* - * Buffer is a static allocation, - * only place what will fit in the buffer. - */ - if (length <= dest_desc->buffer.length) { - /* - * Zero fill first, not willing to do pointer arithmetic for - * archetecture independence. Just clear the whole thing - */ - MEMSET(dest_desc->buffer.pointer, 0, dest_desc->buffer.length); - MEMCPY(dest_desc->buffer.pointer, buffer, length); - } - else { - /* - * truncate, copy only what will fit - */ - MEMCPY(dest_desc->buffer.pointer, buffer, dest_desc->buffer.length); - } + status = acpi_aml_copy_integer_to_bank_field (source_desc, target_desc); break; case INTERNAL_TYPE_INDEX_FIELD: - /* - * Get the global lock if needed - */ - locked = acpi_aml_acquire_global_lock (dest_desc->index_field.lock_rule); - - /* - * Set Index value to select proper Data register - * perform the update (Set index) - */ - - status = acpi_aml_access_named_field (ACPI_WRITE, - dest_desc->index_field.index, - &dest_desc->index_field.value, - sizeof (dest_desc->index_field.value)); - - if (ACPI_SUCCESS (status)) { - /* set index successful, next set Data value */ - - status = acpi_aml_access_named_field (ACPI_WRITE, - dest_desc->index_field.data, - &val_desc->number.value, - sizeof (val_desc->number.value)); - } + status = acpi_aml_copy_integer_to_index_field (source_desc, target_desc); break; - case ACPI_TYPE_FIELD_UNIT: - - - /* - * If the Field Buffer and Index have not been previously evaluated, - * evaluate them and save the results. - */ - if (!(dest_desc->common.flags & AOPOBJ_DATA_VALID)) { - status = acpi_ds_get_field_unit_arguments (dest_desc); - if (ACPI_FAILURE (status)) { - return (status); - } - } - - if ((!dest_desc->field_unit.container || - ACPI_TYPE_BUFFER != dest_desc->field_unit.container->common.type)) - { - status = AE_AML_INTERNAL; - goto clean_up_and_bail_out; - } - - /* - * Get the global lock if needed - */ - locked = acpi_aml_acquire_global_lock (dest_desc->field_unit.lock_rule); - - /* - * TBD: [Unhandled] REMOVE this limitation - * Make sure the operation is within the limits of our implementation - * this is not a Spec limitation!! - */ - if (dest_desc->field_unit.length + dest_desc->field_unit.bit_offset > 32) { - status = AE_NOT_IMPLEMENTED; - goto clean_up_and_bail_out; - } - - /* Field location is (base of buffer) + (byte offset) */ - - location = dest_desc->field_unit.container->buffer.pointer - + dest_desc->field_unit.offset; - - /* - * Construct Mask with 1 bits where the field is, - * 0 bits elsewhere - */ - mask = ((u32) 1 << dest_desc->field_unit.length) - ((u32)1 - << dest_desc->field_unit.bit_offset); - - /* Zero out the field in the buffer */ - - MOVE_UNALIGNED32_TO_32 (&new_value, location); - new_value &= ~mask; - - /* - * Shift and mask the new value into position, - * and or it into the buffer. - */ - new_value |= (val_desc->number.value << dest_desc->field_unit.bit_offset) & - mask; - - /* Store back the value */ - - MOVE_UNALIGNED32_TO_32 (location, &new_value); + case ACPI_TYPE_STRING: + status = acpi_aml_copy_string_to_string (source_desc, target_desc); break; - case ACPI_TYPE_NUMBER: - - - dest_desc->number.value = val_desc->number.value; - - /* Truncate value if we are executing from a 32-bit ACPI table */ + case ACPI_TYPE_BUFFER: - acpi_aml_truncate_for32bit_table (dest_desc, walk_state); + status = acpi_aml_copy_buffer_to_buffer (source_desc, target_desc); break; case ACPI_TYPE_PACKAGE: /* - * TBD: [Unhandled] Not real sure what to do here + * TBD: [Unhandled] Not real sure what to do here */ status = AE_NOT_IMPLEMENTED; break; @@ -518,24 +249,13 @@ acpi_aml_store_object_to_node ( default: /* - * All other types than Alias and the various Fields come here. - * Store Val_desc as the new value of the Name, and set - * the Name's type to that of the value being stored in it. - * Val_desc reference count is incremented by Attach_object. + * All other types come here. */ - status = AE_NOT_IMPLEMENTED; break; } -clean_up_and_bail_out: - - /* - * Release global lock if we acquired it earlier - */ - acpi_aml_release_global_lock (locked); - return (status); } diff --git a/drivers/acpi/interpreter/amstorob.c b/drivers/acpi/interpreter/amstorob.c index f3a098bd2..8118f08bd 100644 --- a/drivers/acpi/interpreter/amstorob.c +++ b/drivers/acpi/interpreter/amstorob.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amstorob - AML Interpreter object store support, store to object - * $Revision: 18 $ + * $Revision: 22 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -40,281 +40,383 @@ /******************************************************************************* * - * FUNCTION: Acpi_aml_store_object_to_object + * FUNCTION: Acpi_aml_copy_buffer_to_buffer * - * PARAMETERS: *Val_desc - Value to be stored - * *Dest_desc - Object to receive the value + * PARAMETERS: Source_desc - Source object to copy + * Target_desc - Destination object of the copy * * RETURN: Status * - * DESCRIPTION: Store an object to another object. + * DESCRIPTION: Copy a buffer object to another buffer object. * - * The Assignment of an object to another (not named) object - * is handled here. - * The val passed in will replace the current value (if any) - * with the input value. + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_copy_buffer_to_buffer ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *target_desc) +{ + u32 length; + u8 *buffer; + + /* + * We know that Source_desc is a buffer by now + */ + buffer = (u8 *) source_desc->buffer.pointer; + length = source_desc->buffer.length; + + /* + * Buffer is a static allocation, + * only place what will fit in the buffer. + */ + if (length <= target_desc->buffer.length) { + /* Clear existing buffer and copy in the new one */ + + MEMSET(target_desc->buffer.pointer, 0, target_desc->buffer.length); + MEMCPY(target_desc->buffer.pointer, buffer, length); + } + + else { + /* + * Truncate the source, copy only what will fit + */ + MEMCPY(target_desc->buffer.pointer, buffer, target_desc->buffer.length); + + } + + return (AE_OK); +} + + +/******************************************************************************* * - * When storing into an object the data is converted to the - * target object type then stored in the object. This means - * that the target object type (for an initialized target) will - * not be changed by a store operation. + * FUNCTION: Acpi_aml_copy_string_to_string + * + * PARAMETERS: Source_desc - Source object to copy + * Target_desc - Destination object of the copy + * + * RETURN: Status * - * This module allows destination types of Number, String, - * and Buffer. + * DESCRIPTION: Copy a String object to another String object * ******************************************************************************/ ACPI_STATUS -acpi_aml_store_object_to_object ( - ACPI_OPERAND_OBJECT *val_desc, - ACPI_OPERAND_OBJECT *dest_desc, - ACPI_WALK_STATE *walk_state) +acpi_aml_copy_string_to_string ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *target_desc) { - ACPI_STATUS status = AE_OK; - u8 *buffer = NULL; - u32 length = 0; - OBJECT_TYPE_INTERNAL destination_type = dest_desc->common.type; + u32 length; + u8 *buffer; /* - * Assuming the parameters are valid!!! + * We know that Source_desc is a string by now. */ - ACPI_ASSERT((dest_desc) && (val_desc)); + buffer = (u8 *) source_desc->string.pointer; + length = source_desc->string.length; /* - * First ensure we have a value that can be stored in the target + * Setting a string value replaces the old string */ - switch (destination_type) - { - /* Type of Name's existing value */ + if (length < target_desc->string.length) { + /* Clear old string and copy in the new one */ - case ACPI_TYPE_NUMBER: + MEMSET(target_desc->string.pointer, 0, target_desc->string.length); + MEMCPY(target_desc->string.pointer, buffer, length); + } + else { /* - * These cases all require only number values or values that - * can be converted to numbers. - * - * If value is not a Number, try to resolve it to one. + * Free the current buffer, then allocate a buffer + * large enough to hold the value */ - - if (val_desc->common.type != ACPI_TYPE_NUMBER) { + if (target_desc->string.pointer && + !acpi_tb_system_table_pointer (target_desc->string.pointer)) + { /* - * Initially not a number, convert + * Only free if not a pointer into the DSDT */ - status = acpi_aml_resolve_to_value (&val_desc, walk_state); - if (ACPI_SUCCESS (status) && - (val_desc->common.type != ACPI_TYPE_NUMBER)) - { - /* - * Conversion successful but still not a number - */ - status = AE_AML_OPERAND_TYPE; - } + acpi_cm_free(target_desc->string.pointer); } - break; + target_desc->string.pointer = acpi_cm_allocate (length + 1); + target_desc->string.length = length; - case ACPI_TYPE_STRING: - case ACPI_TYPE_BUFFER: + if (!target_desc->string.pointer) { + return (AE_NO_MEMORY); + } - /* - * Storing into a Field in a region or into a buffer or into - * a string all is essentially the same. - * - * If value is not a valid type, try to resolve it to one. - */ + MEMCPY(target_desc->string.pointer, buffer, length); + } - if ((val_desc->common.type != ACPI_TYPE_NUMBER) && - (val_desc->common.type != ACPI_TYPE_BUFFER) && - (val_desc->common.type != ACPI_TYPE_STRING)) - { - /* - * Initially not a valid type, convert - */ - status = acpi_aml_resolve_to_value (&val_desc, walk_state); - if (ACPI_SUCCESS (status) && - (val_desc->common.type != ACPI_TYPE_NUMBER) && - (val_desc->common.type != ACPI_TYPE_BUFFER) && - (val_desc->common.type != ACPI_TYPE_STRING)) - { - /* - * Conversion successful but still not a valid type - */ - status = AE_AML_OPERAND_TYPE; - } - } - break; + return (AE_OK); +} - default: +/******************************************************************************* + * + * FUNCTION: Acpi_aml_copy_integer_to_index_field + * + * PARAMETERS: Source_desc - Source object to copy + * Target_desc - Destination object of the copy + * + * RETURN: Status + * + * DESCRIPTION: Write an Integer to an Index Field + * + ******************************************************************************/ - /* - * TBD: [Unhandled] What other combinations must be implemented? - */ - status = AE_NOT_IMPLEMENTED; - break; - } +ACPI_STATUS +acpi_aml_copy_integer_to_index_field ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *target_desc) +{ + ACPI_STATUS status; + u8 locked; + + + /* + * Get the global lock if needed + */ + locked = acpi_aml_acquire_global_lock (target_desc->index_field.lock_rule); - /* Exit now if failure above */ + /* + * Set Index value to select proper Data register + * perform the update (Set index) + */ + status = acpi_aml_access_named_field (ACPI_WRITE, + target_desc->index_field.index, + &target_desc->index_field.value, + sizeof (target_desc->index_field.value)); + if (ACPI_SUCCESS (status)) { + /* Set_index was successful, next set Data value */ + + status = acpi_aml_access_named_field (ACPI_WRITE, + target_desc->index_field.data, + &source_desc->integer.value, + sizeof (source_desc->integer.value)); - if (ACPI_FAILURE (status)) { - goto clean_up_and_bail_out; } + + /* - * Acpi_everything is ready to execute now, We have - * a value we can handle, just perform the update + * Release global lock if we acquired it earlier */ + acpi_aml_release_global_lock (locked); - switch (destination_type) - { + return (status); +} - case ACPI_TYPE_STRING: - /* - * Perform the update - */ +/******************************************************************************* + * + * FUNCTION: Acpi_aml_copy_integer_to_bank_field + * + * PARAMETERS: Source_desc - Source object to copy + * Target_desc - Destination object of the copy + * + * RETURN: Status + * + * DESCRIPTION: Write an Integer to a Bank Field + * + ******************************************************************************/ - switch (val_desc->common.type) - { - case ACPI_TYPE_NUMBER: - buffer = (u8 *) &val_desc->number.value; - length = sizeof (val_desc->number.value); - break; - - case ACPI_TYPE_BUFFER: - buffer = (u8 *) val_desc->buffer.pointer; - length = val_desc->buffer.length; - break; - - case ACPI_TYPE_STRING: - buffer = (u8 *) val_desc->string.pointer; - length = val_desc->string.length; - break; - } +ACPI_STATUS +acpi_aml_copy_integer_to_bank_field ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *target_desc) +{ + ACPI_STATUS status; + u8 locked; - /* - * Setting a string value replaces the old string - */ - if (length < dest_desc->string.length) { - /* - * Zero fill, not willing to do pointer arithmetic for - * architecture independence. Just clear the whole thing - */ - MEMSET(dest_desc->string.pointer, 0, dest_desc->string.length); - MEMCPY(dest_desc->string.pointer, buffer, length); - } - else { - /* - * Free the current buffer, then allocate a buffer - * large enough to hold the value - */ - if ( dest_desc->string.pointer && - !acpi_tb_system_table_pointer (dest_desc->string.pointer)) - { - /* - * Only free if not a pointer into the DSDT - */ + /* + * Get the global lock if needed + */ + locked = acpi_aml_acquire_global_lock (target_desc->index_field.lock_rule); - acpi_cm_free(dest_desc->string.pointer); - } - dest_desc->string.pointer = acpi_cm_allocate (length + 1); - dest_desc->string.length = length; + /* + * Set Bank value to select proper Bank + * Perform the update (Set Bank Select) + */ - if (!dest_desc->string.pointer) { - status = AE_NO_MEMORY; - goto clean_up_and_bail_out; - } + status = acpi_aml_access_named_field (ACPI_WRITE, + target_desc->bank_field.bank_select, + &target_desc->bank_field.value, + sizeof (target_desc->bank_field.value)); + if (ACPI_SUCCESS (status)) { + /* Set bank select successful, set data value */ + + status = acpi_aml_access_named_field (ACPI_WRITE, + target_desc->bank_field.bank_select, + &source_desc->bank_field.value, + sizeof (source_desc->bank_field.value)); + } - MEMCPY(dest_desc->string.pointer, buffer, length); - } - break; - case ACPI_TYPE_BUFFER: + /* + * Release global lock if we acquired it earlier + */ + acpi_aml_release_global_lock (locked); - /* - * Perform the update to the buffer - */ + return (status); +} - switch (val_desc->common.type) - { - case ACPI_TYPE_NUMBER: - buffer = (u8 *) &val_desc->number.value; - length = sizeof (val_desc->number.value); - break; - - case ACPI_TYPE_BUFFER: - buffer = (u8 *) val_desc->buffer.pointer; - length = val_desc->buffer.length; - break; - - case ACPI_TYPE_STRING: - buffer = (u8 *) val_desc->string.pointer; - length = val_desc->string.length; - break; - } - /* - * If the buffer is uninitialized, - * memory needs to be allocated for the copy. - */ - if(0 == dest_desc->buffer.length) { - dest_desc->buffer.pointer = acpi_cm_callocate(length); - dest_desc->buffer.length = length; - - if (!dest_desc->buffer.pointer) { - status = AE_NO_MEMORY; - goto clean_up_and_bail_out; - } - } +/******************************************************************************* + * + * FUNCTION: Acpi_aml_copy_data_to_named_field + * + * PARAMETERS: Source_desc - Source object to copy + * Node - Destination Namespace node + * + * RETURN: Status + * + * DESCRIPTION: Copy raw data to a Named Field. No implicit conversion + * is performed on the source object + * + ******************************************************************************/ - /* - * Buffer is a static allocation, - * only place what will fit in the buffer. - */ - if (length <= dest_desc->buffer.length) { - /* - * Zero fill first, not willing to do pointer arithmetic for - * architecture independence. Just clear the whole thing - */ - MEMSET(dest_desc->buffer.pointer, 0, dest_desc->buffer.length); - MEMCPY(dest_desc->buffer.pointer, buffer, length); - } - else { - /* - * truncate, copy only what will fit - */ - MEMCPY(dest_desc->buffer.pointer, buffer, dest_desc->buffer.length); - } - break; +ACPI_STATUS +acpi_aml_copy_data_to_named_field ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_NAMESPACE_NODE *node) +{ + ACPI_STATUS status; + u8 locked; + u32 length; + u8 *buffer; - case ACPI_TYPE_NUMBER: - dest_desc->number.value = val_desc->number.value; + /* + * Named fields (Create_xxx_field) - We don't perform any conversions on the + * source operand, just use the raw data + */ + switch (source_desc->common.type) + { + case ACPI_TYPE_INTEGER: + buffer = (u8 *) &source_desc->integer.value; + length = sizeof (source_desc->integer.value); + break; - /* Truncate value if we are executing from a 32-bit ACPI table */ + case ACPI_TYPE_BUFFER: + buffer = (u8 *) source_desc->buffer.pointer; + length = source_desc->buffer.length; + break; - acpi_aml_truncate_for32bit_table (dest_desc, walk_state); + case ACPI_TYPE_STRING: + buffer = (u8 *) source_desc->string.pointer; + length = source_desc->string.length; break; default: + return (AE_TYPE); + } - /* - * All other types than Alias and the various Fields come here. - * Store Val_desc as the new value of the Name, and set - * the Name's type to that of the value being stored in it. - * Val_desc reference count is incremented by Attach_object. - */ + /* + * Get the global lock if needed before the update + * TBD: not needed! + */ + locked = acpi_aml_acquire_global_lock (source_desc->field.lock_rule); - status = AE_NOT_IMPLEMENTED; - break; - } + status = acpi_aml_access_named_field (ACPI_WRITE, + node, buffer, length); -clean_up_and_bail_out: + acpi_aml_release_global_lock (locked); return (status); } + +/******************************************************************************* + * + * FUNCTION: Acpi_aml_copy_integer_to_field_unit + * + * PARAMETERS: Source_desc - Source object to copy + * Target_desc - Destination object of the copy + * + * RETURN: Status + * + * DESCRIPTION: Write an Integer to a Field Unit. + * + ******************************************************************************/ + +ACPI_STATUS +acpi_aml_copy_integer_to_field_unit ( + ACPI_OPERAND_OBJECT *source_desc, + ACPI_OPERAND_OBJECT *target_desc) +{ + ACPI_STATUS status = AE_OK; + u8 *location = NULL; + u32 mask; + u32 new_value; + u8 locked = FALSE; + + + /* + * If the Field Buffer and Index have not been previously evaluated, + * evaluate them and save the results. + */ + if (!(target_desc->common.flags & AOPOBJ_DATA_VALID)) { + status = acpi_ds_get_field_unit_arguments (target_desc); + if (ACPI_FAILURE (status)) { + return (status); + } + } + + if ((!target_desc->field_unit.container || + ACPI_TYPE_BUFFER != target_desc->field_unit.container->common.type)) + { + return (AE_AML_INTERNAL); + } + + /* + * Get the global lock if needed + */ + locked = acpi_aml_acquire_global_lock (target_desc->field_unit.lock_rule); + + /* + * TBD: [Unhandled] REMOVE this limitation + * Make sure the operation is within the limits of our implementation + * this is not a Spec limitation!! + */ + if (target_desc->field_unit.length + target_desc->field_unit.bit_offset > 32) { + return (AE_NOT_IMPLEMENTED); + } + + /* Field location is (base of buffer) + (byte offset) */ + + location = target_desc->field_unit.container->buffer.pointer + + target_desc->field_unit.offset; + + /* + * Construct Mask with 1 bits where the field is, + * 0 bits elsewhere + */ + mask = ((u32) 1 << target_desc->field_unit.length) - ((u32)1 + << target_desc->field_unit.bit_offset); + + /* Zero out the field in the buffer */ + + MOVE_UNALIGNED32_TO_32 (&new_value, location); + new_value &= ~mask; + + /* + * Shift and mask the new value into position, + * and or it into the buffer. + */ + new_value |= (source_desc->integer.value << target_desc->field_unit.bit_offset) & + mask; + + /* Store back the value */ + + MOVE_UNALIGNED32_TO_32 (location, &new_value); + + return (AE_OK); +} + + diff --git a/drivers/acpi/interpreter/amsystem.c b/drivers/acpi/interpreter/amsystem.c index 9ad72c161..5e60538af 100644 --- a/drivers/acpi/interpreter/amsystem.c +++ b/drivers/acpi/interpreter/amsystem.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amsystem - Interface to OS services - * $Revision: 52 $ + * $Revision: 54 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -202,7 +202,7 @@ acpi_aml_system_acquire_mutex ( } status = acpi_aml_system_wait_semaphore (obj_desc->mutex.semaphore, - (u32) time_desc->number.value); + (u32) time_desc->integer.value); return (status); } @@ -299,7 +299,7 @@ acpi_aml_system_wait_event ( if (obj_desc) { status = acpi_aml_system_wait_semaphore (obj_desc->event.semaphore, - (u32) time_desc->number.value); + (u32) time_desc->integer.value); } diff --git a/drivers/acpi/interpreter/amutils.c b/drivers/acpi/interpreter/amutils.c index 4e1359888..e3456099c 100644 --- a/drivers/acpi/interpreter/amutils.c +++ b/drivers/acpi/interpreter/amutils.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amutils - interpreter/scanner utilities - * $Revision: 66 $ + * $Revision: 68 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -46,14 +46,11 @@ typedef struct internal_search_st /* Used to traverse nested packages when copying*/ +/* TBD: This must be removed! */ INTERNAL_PKG_SEARCH_INFO copy_level[MAX_PACKAGE_DEPTH]; -static NATIVE_CHAR hex[] = - {'0','1','2','3','4','5','6','7','8','9','A','B','C','D','E','F'}; - - /******************************************************************************* * * FUNCTION: Acpi_aml_enter_interpreter @@ -156,7 +153,7 @@ acpi_aml_truncate_for32bit_table ( */ if ((!obj_desc) || - (obj_desc->common.type != ACPI_TYPE_NUMBER) || + (obj_desc->common.type != ACPI_TYPE_INTEGER) || (!walk_state->method_node)) { return; @@ -167,7 +164,7 @@ acpi_aml_truncate_for32bit_table ( * We are running a method that exists in a 32-bit ACPI table. * Truncate the value to 32 bits by zeroing out the upper 32-bit field */ - obj_desc->number.value &= (ACPI_INTEGER) ACPI_UINT32_MAX; + obj_desc->integer.value &= (ACPI_INTEGER) ACPI_UINT32_MAX; } } @@ -343,10 +340,10 @@ acpi_aml_eisa_id_to_string ( out_string[0] = (char) ('@' + ((id >> 26) & 0x1f)); out_string[1] = (char) ('@' + ((id >> 21) & 0x1f)); out_string[2] = (char) ('@' + ((id >> 16) & 0x1f)); - out_string[3] = hex[(id >> 12) & 0xf]; - out_string[4] = hex[(id >> 8) & 0xf]; - out_string[5] = hex[(id >> 4) & 0xf]; - out_string[6] = hex[id & 0xf]; + out_string[3] = acpi_gbl_hex_to_ascii[(id >> 12) & 0xf]; + out_string[4] = acpi_gbl_hex_to_ascii[(id >> 8) & 0xf]; + out_string[5] = acpi_gbl_hex_to_ascii[(id >> 4) & 0xf]; + out_string[6] = acpi_gbl_hex_to_ascii[id & 0xf]; out_string[7] = 0; return (AE_OK); diff --git a/drivers/acpi/interpreter/amxface.c b/drivers/acpi/interpreter/amxface.c index fd589d39b..20cf9820e 100644 --- a/drivers/acpi/interpreter/amxface.c +++ b/drivers/acpi/interpreter/amxface.c @@ -2,12 +2,12 @@ /****************************************************************************** * * Module Name: amxface - External interpreter interfaces - * $Revision: 22 $ + * $Revision: 24 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/namespace/nsaccess.c b/drivers/acpi/namespace/nsaccess.c index 27c02e22a..c2e6a71e0 100644 --- a/drivers/acpi/namespace/nsaccess.c +++ b/drivers/acpi/namespace/nsaccess.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: nsaccess - Top-level functions for accessing ACPI namespace - * $Revision: 117 $ + * $Revision: 119 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -117,9 +117,9 @@ acpi_ns_root_initialize (void) switch (init_val->type) { - case ACPI_TYPE_NUMBER: + case ACPI_TYPE_INTEGER: - obj_desc->number.value = + obj_desc->integer.value = (ACPI_INTEGER) STRTOUL (init_val->val, NULL, 10); break; @@ -239,7 +239,7 @@ acpi_ns_lookup ( ACPI_NAMESPACE_NODE **return_node) { ACPI_STATUS status; - ACPI_NAMESPACE_NODE *prefix_node; + ACPI_NAMESPACE_NODE *prefix_node; ACPI_NAMESPACE_NODE *current_node = NULL; ACPI_NAMESPACE_NODE *scope_to_push = NULL; ACPI_NAMESPACE_NODE *this_node = NULL; @@ -248,8 +248,7 @@ acpi_ns_lookup ( u8 null_name_path = FALSE; OBJECT_TYPE_INTERNAL type_to_check_for; OBJECT_TYPE_INTERNAL this_search_type; - - DEBUG_ONLY_MEMBERS (u32 i) + u32 local_flags = flags & ~NS_ERROR_IF_FOUND; if (!return_node) { @@ -437,6 +436,7 @@ acpi_ns_lookup ( this_search_type = ACPI_TYPE_ANY; if (!num_segments) { this_search_type = type; + local_flags = flags; } /* Pluck one ACPI name from the front of the pathname */ @@ -447,7 +447,7 @@ acpi_ns_lookup ( status = acpi_ns_search_and_enter (simple_name, walk_state, current_node, interpreter_mode, - this_search_type, flags, + this_search_type, local_flags, &this_node); if (ACPI_FAILURE (status)) { diff --git a/drivers/acpi/namespace/nsalloc.c b/drivers/acpi/namespace/nsalloc.c index 9f1d5377e..5cac5d111 100644 --- a/drivers/acpi/namespace/nsalloc.c +++ b/drivers/acpi/namespace/nsalloc.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: nsalloc - Namespace allocation and deletion utilities - * $Revision: 43 $ + * $Revision: 45 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/namespace/nseval.c b/drivers/acpi/namespace/nseval.c index fbba7840c..6c040d22e 100644 --- a/drivers/acpi/namespace/nseval.c +++ b/drivers/acpi/namespace/nseval.c @@ -2,12 +2,12 @@ * * Module Name: nseval - Object evaluation interfaces -- includes control * method lookup and execution. - * $Revision: 79 $ + * $Revision: 81 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/namespace/nsinit.c b/drivers/acpi/namespace/nsinit.c index ddacd7b8c..76b535205 100644 --- a/drivers/acpi/namespace/nsinit.c +++ b/drivers/acpi/namespace/nsinit.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: nsinit - namespace initialization - * $Revision: 9 $ + * $Revision: 12 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -239,7 +239,9 @@ acpi_ns_init_one_device ( status = acpi_cm_execute_STA (node, &flags); if (ACPI_FAILURE (status)) { - return (status); + /* Ignore error and move on to next device */ + + return (AE_OK); } info->num_STA++; @@ -260,12 +262,15 @@ acpi_ns_init_one_device ( } else if (ACPI_FAILURE (status)) { - return (status); + /* Ignore error and move on to next device */ + } else { + /* Count of successfull INIs */ + info->num_INI++; } - return (status); + return (AE_OK); } diff --git a/drivers/acpi/namespace/nsload.c b/drivers/acpi/namespace/nsload.c index 28a26376d..f0eaf273b 100644 --- a/drivers/acpi/namespace/nsload.c +++ b/drivers/acpi/namespace/nsload.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: nsload - namespace loading/expanding/contracting procedures - * $Revision: 33 $ + * $Revision: 35 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/namespace/nsnames.c b/drivers/acpi/namespace/nsnames.c index faf8fe56a..c2fb49163 100644 --- a/drivers/acpi/namespace/nsnames.c +++ b/drivers/acpi/namespace/nsnames.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: nsnames - Name manipulation and search - * $Revision: 51 $ + * $Revision: 53 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/namespace/nsobject.c b/drivers/acpi/namespace/nsobject.c index 6d41b91ce..493cb1317 100644 --- a/drivers/acpi/namespace/nsobject.c +++ b/drivers/acpi/namespace/nsobject.c @@ -2,12 +2,12 @@ * * Module Name: nsobject - Utilities for objects attached to namespace * table entries - * $Revision: 47 $ + * $Revision: 49 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -205,7 +205,7 @@ acpi_ns_attach_object ( case AML_ZERO_OP: case AML_ONES_OP: case AML_ONE_OP: case AML_BYTE_OP: case AML_WORD_OP: case AML_DWORD_OP: - obj_type = ACPI_TYPE_NUMBER; + obj_type = ACPI_TYPE_INTEGER; break; diff --git a/drivers/acpi/namespace/nssearch.c b/drivers/acpi/namespace/nssearch.c index 001f57d9d..7a29e7542 100644 --- a/drivers/acpi/namespace/nssearch.c +++ b/drivers/acpi/namespace/nssearch.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: nssearch - Namespace search - * $Revision: 62 $ + * $Revision: 64 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -70,8 +70,6 @@ acpi_ns_search_node ( ACPI_NAMESPACE_NODE *next_node; - - /* * Search for name in this table, which is to say that we must search * for the name among the children of this object diff --git a/drivers/acpi/namespace/nsutils.c b/drivers/acpi/namespace/nsutils.c index 593064dbb..11457b33f 100644 --- a/drivers/acpi/namespace/nsutils.c +++ b/drivers/acpi/namespace/nsutils.c @@ -2,12 +2,12 @@ * * Module Name: nsutils - Utilities for accessing ACPI namespace, accessing * parents and siblings and Scope manipulation - * $Revision: 74 $ + * $Revision: 77 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/namespace/nswalk.c b/drivers/acpi/namespace/nswalk.c index 3682266b3..cf84be326 100644 --- a/drivers/acpi/namespace/nswalk.c +++ b/drivers/acpi/namespace/nswalk.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: nswalk - Functions for walking the APCI namespace - * $Revision: 17 $ + * $Revision: 19 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/namespace/nsxfname.c b/drivers/acpi/namespace/nsxfname.c index 2947f7f06..957481a6d 100644 --- a/drivers/acpi/namespace/nsxfname.c +++ b/drivers/acpi/namespace/nsxfname.c @@ -2,12 +2,12 @@ * * Module Name: nsxfname - Public interfaces to the ACPI subsystem * ACPI Namespace oriented interfaces - * $Revision: 73 $ + * $Revision: 75 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/namespace/nsxfobj.c b/drivers/acpi/namespace/nsxfobj.c index d4432ee8a..742c5da66 100644 --- a/drivers/acpi/namespace/nsxfobj.c +++ b/drivers/acpi/namespace/nsxfobj.c @@ -2,12 +2,12 @@ * * Module Name: nsxfobj - Public interfaces to the ACPI subsystem * ACPI Object oriented interfaces - * $Revision: 75 $ + * $Revision: 78 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -695,4 +695,3 @@ acpi_get_devices ( return (status); } - diff --git a/drivers/acpi/os.c b/drivers/acpi/os.c index 7bf86171b..172b6659b 100644 --- a/drivers/acpi/os.c +++ b/drivers/acpi/os.c @@ -24,8 +24,8 @@ #include <linux/mm.h> #include <linux/pci.h> #include <linux/acpi.h> +#include <linux/delay.h> #include <asm/io.h> -#include <asm/delay.h> #include "acpi.h" #include "driver.h" diff --git a/drivers/acpi/parser/psargs.c b/drivers/acpi/parser/psargs.c index 35d623668..da3ef9b30 100644 --- a/drivers/acpi/parser/psargs.c +++ b/drivers/acpi/parser/psargs.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: psargs - Parse AML opcode arguments - * $Revision: 42 $ + * $Revision: 43 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/parser/psopcode.c b/drivers/acpi/parser/psopcode.c index a7f061363..90a73f45b 100644 --- a/drivers/acpi/parser/psopcode.c +++ b/drivers/acpi/parser/psopcode.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: psopcode - Parser opcode information table - * $Revision: 24 $ + * $Revision: 27 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -44,7 +44,7 @@ #define _PFX 0x6D #define _UNKNOWN_OPCODE 0x02 /* An example unknown opcode */ -#define MAX_EXTENDED_OPCODE 0x87 +#define MAX_EXTENDED_OPCODE 0x88 #define NUM_EXTENDED_OPCODE MAX_EXTENDED_OPCODE + 1 #define MAX_INTERNAL_OPCODE #define NUM_INTERNAL_OPCODE MAX_INTERNAL_OPCODE + 1 @@ -83,9 +83,11 @@ #define ARGP_WORD_OP ARGP_LIST1 (ARGP_WORDDATA) #define ARGP_DWORD_OP ARGP_LIST1 (ARGP_DWORDDATA) #define ARGP_STRING_OP ARGP_LIST1 (ARGP_CHARLIST) +#define ARGP_QWORD_OP ARGP_LIST1 (ARGP_QWORDDATA) #define ARGP_SCOPE_OP ARGP_LIST3 (ARGP_PKGLENGTH, ARGP_NAME, ARGP_TERMLIST) #define ARGP_BUFFER_OP ARGP_LIST3 (ARGP_PKGLENGTH, ARGP_TERMARG, ARGP_BYTELIST) #define ARGP_PACKAGE_OP ARGP_LIST3 (ARGP_PKGLENGTH, ARGP_BYTEDATA, ARGP_DATAOBJLIST) +#define ARGP_VAR_PACKAGE_OP ARGP_LIST3 (ARGP_PKGLENGTH, ARGP_BYTEDATA, ARGP_DATAOBJLIST) #define ARGP_METHOD_OP ARGP_LIST4 (ARGP_PKGLENGTH, ARGP_NAME, ARGP_BYTEDATA, ARGP_TERMLIST) #define ARGP_LOCAL0 ARG_NONE #define ARGP_LOCAL1 ARG_NONE @@ -122,6 +124,8 @@ #define ARGP_FIND_SET_LEFT_BIT_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TARGET) #define ARGP_FIND_SET_RIGHT_BIT_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TARGET) #define ARGP_DEREF_OF_OP ARGP_LIST1 (ARGP_TERMARG) +#define ARGP_CONCAT_RES_OP ARGP_LIST3 (ARGP_TERMARG, ARGP_TERMARG, ARGP_TARGET) +#define ARGP_MOD_OP ARGP_LIST3 (ARGP_TERMARG, ARGP_TERMARG, ARGP_TARGET) #define ARGP_NOTIFY_OP ARGP_LIST2 (ARGP_SUPERNAME, ARGP_TERMARG) #define ARGP_SIZE_OF_OP ARGP_LIST1 (ARGP_SUPERNAME) #define ARGP_INDEX_OP ARGP_LIST3 (ARGP_TERMARG, ARGP_TERMARG, ARGP_TARGET) @@ -131,12 +135,21 @@ #define ARGP_BYTE_FIELD_OP ARGP_LIST3 (ARGP_TERMARG, ARGP_TERMARG, ARGP_NAME) #define ARGP_BIT_FIELD_OP ARGP_LIST3 (ARGP_TERMARG, ARGP_TERMARG, ARGP_NAME) #define ARGP_TYPE_OP ARGP_LIST1 (ARGP_SUPERNAME) +#define ARGP_QWORD_FIELD_OP ARGP_LIST3 (ARGP_TERMARG, ARGP_TERMARG, ARGP_NAME) #define ARGP_LAND_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TERMARG) #define ARGP_LOR_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TERMARG) #define ARGP_LNOT_OP ARGP_LIST1 (ARGP_TERMARG) #define ARGP_LEQUAL_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TERMARG) #define ARGP_LGREATER_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TERMARG) #define ARGP_LLESS_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TERMARG) +#define ARGP_TO_BUFFER_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TARGET) +#define ARGP_TO_DEC_STR_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TARGET) +#define ARGP_TO_HEX_STR_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TARGET) +#define ARGP_TO_INTEGER_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TARGET) +#define ARGP_TO_STRING_OP ARGP_LIST3 (ARGP_TERMARG, ARGP_TERMARG, ARGP_TARGET) +#define ARGP_COPY_OP ARGP_LIST2 (ARGP_SUPERNAME, ARGP_SIMPLENAME) +#define ARGP_MID_OP ARGP_LIST4 (ARGP_TERMARG, ARGP_TERMARG, ARGP_TERMARG, ARGP_TARGET) +#define ARGP_CONTINUE_OP ARG_NONE #define ARGP_IF_OP ARGP_LIST3 (ARGP_PKGLENGTH, ARGP_TERMARG, ARGP_TERMLIST) #define ARGP_ELSE_OP ARGP_LIST2 (ARGP_PKGLENGTH, ARGP_TERMLIST) #define ARGP_WHILE_OP ARGP_LIST3 (ARGP_PKGLENGTH, ARGP_TERMARG, ARGP_TERMLIST) @@ -149,6 +162,7 @@ #define ARGP_EVENT_OP ARGP_LIST1 (ARGP_NAME) #define ARGP_COND_REF_OF_OP ARGP_LIST2 (ARGP_SUPERNAME, ARGP_SUPERNAME) #define ARGP_CREATE_FIELD_OP ARGP_LIST4 (ARGP_TERMARG, ARGP_TERMARG, ARGP_TERMARG, ARGP_NAME) +#define ARGP_LOAD_TABLE_OP ARGP_LIST6 (ARGP_TERMARG, ARGP_TERMARG, ARGP_TERMARG, ARGP_TERMARG, ARGP_TERMARG, ARGP_TERMARG) #define ARGP_LOAD_OP ARGP_LIST2 (ARGP_NAMESTRING, ARGP_SUPERNAME) #define ARGP_STALL_OP ARGP_LIST1 (ARGP_TERMARG) #define ARGP_SLEEP_OP ARGP_LIST1 (ARGP_TERMARG) @@ -171,6 +185,7 @@ #define ARGP_THERMAL_ZONE_OP ARGP_LIST3 (ARGP_PKGLENGTH, ARGP_NAME, ARGP_OBJLIST) #define ARGP_INDEX_FIELD_OP ARGP_LIST5 (ARGP_PKGLENGTH, ARGP_NAMESTRING, ARGP_NAMESTRING,ARGP_BYTEDATA, ARGP_FIELDLIST) #define ARGP_BANK_FIELD_OP ARGP_LIST6 (ARGP_PKGLENGTH, ARGP_NAMESTRING, ARGP_NAMESTRING,ARGP_TERMARG, ARGP_BYTEDATA, ARGP_FIELDLIST) +#define ARGP_DATA_REGION_OP ARGP_LIST4 (ARGP_NAMESTRING, ARGP_TERMARG, ARGP_TERMARG, ARGP_TERMARG) #define ARGP_LNOTEQUAL_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TERMARG) #define ARGP_LLESSEQUAL_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TERMARG) #define ARGP_LGREATEREQUAL_OP ARGP_LIST2 (ARGP_TERMARG, ARGP_TERMARG) @@ -187,7 +202,7 @@ * All AML opcodes and the runtime arguments for each. Used by the AML interpreter Each list is compressed * into a 32-bit number and stored in the master opcode table at the end of this file. * - * (Used by Acpi_aml_prep_operands procedure) + * (Used by Acpi_aml_prep_operands procedure and the ASL Compiler) */ #define ARGI_ZERO_OP ARG_NONE @@ -198,9 +213,11 @@ #define ARGI_WORD_OP ARGI_INVALID_OPCODE #define ARGI_DWORD_OP ARGI_INVALID_OPCODE #define ARGI_STRING_OP ARGI_INVALID_OPCODE +#define ARGI_QWORD_OP ARGI_INVALID_OPCODE #define ARGI_SCOPE_OP ARGI_INVALID_OPCODE #define ARGI_BUFFER_OP ARGI_INVALID_OPCODE #define ARGI_PACKAGE_OP ARGI_INVALID_OPCODE +#define ARGI_VAR_PACKAGE_OP ARGI_INVALID_OPCODE #define ARGI_METHOD_OP ARGI_INVALID_OPCODE #define ARGI_LOCAL0 ARG_NONE #define ARGI_LOCAL1 ARG_NONE @@ -218,40 +235,51 @@ #define ARGI_ARG5 ARG_NONE #define ARGI_ARG6 ARG_NONE #define ARGI_STORE_OP ARGI_LIST2 (ARGI_ANYTYPE, ARGI_TARGETREF) -#define ARGI_REF_OF_OP ARGI_LIST1 (ARGI_REFERENCE) -#define ARGI_ADD_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_CONCAT_OP ARGI_LIST3 (ARGI_STRING, ARGI_STRING, ARGI_TARGETREF) -#define ARGI_SUBTRACT_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_INCREMENT_OP ARGI_LIST1 (ARGI_REFERENCE) -#define ARGI_DECREMENT_OP ARGI_LIST1 (ARGI_REFERENCE) -#define ARGI_MULTIPLY_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_DIVIDE_OP ARGI_LIST4 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF, ARGI_TARGETREF) -#define ARGI_SHIFT_LEFT_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_SHIFT_RIGHT_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_BIT_AND_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_BIT_NAND_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_BIT_OR_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_BIT_NOR_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_BIT_XOR_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_BIT_NOT_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_FIND_SET_LEFT_BIT_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_FIND_SET_RIGHT_BIT_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_TARGETREF) +#define ARGI_REF_OF_OP ARGI_LIST1 (ARGI_OBJECT_REF) +#define ARGI_ADD_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_CONCAT_OP ARGI_LIST3 (ARGI_COMPUTEDATA,ARGI_COMPUTEDATA, ARGI_TARGETREF) +#define ARGI_SUBTRACT_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_INCREMENT_OP ARGI_LIST1 (ARGI_INTEGER_REF) +#define ARGI_DECREMENT_OP ARGI_LIST1 (ARGI_INTEGER_REF) +#define ARGI_MULTIPLY_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_DIVIDE_OP ARGI_LIST4 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF, ARGI_TARGETREF) +#define ARGI_SHIFT_LEFT_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_SHIFT_RIGHT_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_BIT_AND_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_BIT_NAND_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_BIT_OR_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_BIT_NOR_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_BIT_XOR_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_BIT_NOT_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_FIND_SET_LEFT_BIT_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_FIND_SET_RIGHT_BIT_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_TARGETREF) #define ARGI_DEREF_OF_OP ARGI_LIST1 (ARGI_REFERENCE) -#define ARGI_NOTIFY_OP ARGI_LIST2 (ARGI_REFERENCE, ARGI_NUMBER) +#define ARGI_CONCAT_RES_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_BUFFER, ARGI_TARGETREF) +#define ARGI_MOD_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_NOTIFY_OP ARGI_LIST2 (ARGI_DEVICE_REF, ARGI_INTEGER) #define ARGI_SIZE_OF_OP ARGI_LIST1 (ARGI_DATAOBJECT) -#define ARGI_INDEX_OP ARGI_LIST3 (ARGI_COMPLEXOBJ, ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_MATCH_OP ARGI_LIST6 (ARGI_PACKAGE, ARGI_NUMBER, ARGI_NUMBER, ARGI_NUMBER, ARGI_NUMBER, ARGI_NUMBER) -#define ARGI_DWORD_FIELD_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_NUMBER, ARGI_REFERENCE) -#define ARGI_WORD_FIELD_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_NUMBER, ARGI_REFERENCE) -#define ARGI_BYTE_FIELD_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_NUMBER, ARGI_REFERENCE) -#define ARGI_BIT_FIELD_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_NUMBER, ARGI_REFERENCE) +#define ARGI_INDEX_OP ARGI_LIST3 (ARGI_COMPLEXOBJ, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_MATCH_OP ARGI_LIST6 (ARGI_PACKAGE, ARGI_INTEGER, ARGI_INTEGER, ARGI_INTEGER, ARGI_INTEGER, ARGI_INTEGER) +#define ARGI_DWORD_FIELD_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_INTEGER, ARGI_REFERENCE) +#define ARGI_WORD_FIELD_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_INTEGER, ARGI_REFERENCE) +#define ARGI_BYTE_FIELD_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_INTEGER, ARGI_REFERENCE) +#define ARGI_BIT_FIELD_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_INTEGER, ARGI_REFERENCE) #define ARGI_TYPE_OP ARGI_LIST1 (ARGI_ANYTYPE) -#define ARGI_LAND_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_NUMBER) -#define ARGI_LOR_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_NUMBER) -#define ARGI_LNOT_OP ARGI_LIST1 (ARGI_NUMBER) -#define ARGI_LEQUAL_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_NUMBER) -#define ARGI_LGREATER_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_NUMBER) -#define ARGI_LLESS_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_NUMBER) +#define ARGI_QWORD_FIELD_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_INTEGER, ARGI_REFERENCE) +#define ARGI_LAND_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_INTEGER) +#define ARGI_LOR_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_INTEGER) +#define ARGI_LNOT_OP ARGI_LIST1 (ARGI_INTEGER) +#define ARGI_LEQUAL_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_INTEGER) +#define ARGI_LGREATER_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_INTEGER) +#define ARGI_LLESS_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_INTEGER) +#define ARGI_TO_BUFFER_OP ARGI_LIST2 (ARGI_COMPUTEDATA,ARGI_FIXED_TARGET) +#define ARGI_TO_DEC_STR_OP ARGI_LIST2 (ARGI_COMPUTEDATA,ARGI_FIXED_TARGET) +#define ARGI_TO_HEX_STR_OP ARGI_LIST2 (ARGI_COMPUTEDATA,ARGI_FIXED_TARGET) +#define ARGI_TO_INTEGER_OP ARGI_LIST2 (ARGI_COMPUTEDATA,ARGI_FIXED_TARGET) +#define ARGI_TO_STRING_OP ARGI_LIST3 (ARGI_BUFFER, ARGI_INTEGER, ARGI_FIXED_TARGET) +#define ARGI_COPY_OP ARGI_LIST2 (ARGI_ANYTYPE, ARGI_SIMPLE_TARGET) +#define ARGI_MID_OP ARGI_LIST4 (ARGI_BUFFERSTRING,ARGI_INTEGER, ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_CONTINUE_OP ARGI_INVALID_OPCODE #define ARGI_IF_OP ARGI_INVALID_OPCODE #define ARGI_ELSE_OP ARGI_INVALID_OPCODE #define ARGI_WHILE_OP ARGI_INVALID_OPCODE @@ -262,23 +290,24 @@ #define ARGI_ONES_OP ARG_NONE #define ARGI_MUTEX_OP ARGI_INVALID_OPCODE #define ARGI_EVENT_OP ARGI_INVALID_OPCODE -#define ARGI_COND_REF_OF_OP ARGI_LIST2 (ARGI_REFERENCE, ARGI_TARGETREF) -#define ARGI_CREATE_FIELD_OP ARGI_LIST4 (ARGI_BUFFER, ARGI_NUMBER, ARGI_NUMBER, ARGI_REFERENCE) +#define ARGI_COND_REF_OF_OP ARGI_LIST2 (ARGI_OBJECT_REF, ARGI_TARGETREF) +#define ARGI_CREATE_FIELD_OP ARGI_LIST4 (ARGI_BUFFER, ARGI_INTEGER, ARGI_INTEGER, ARGI_REFERENCE) +#define ARGI_LOAD_TABLE_OP ARGI_LIST6 (ARGI_STRING, ARGI_STRING, ARGI_STRING, ARGI_STRING, ARGI_STRING, ARGI_TARGETREF) #define ARGI_LOAD_OP ARGI_LIST2 (ARGI_REGION, ARGI_TARGETREF) -#define ARGI_STALL_OP ARGI_LIST1 (ARGI_NUMBER) -#define ARGI_SLEEP_OP ARGI_LIST1 (ARGI_NUMBER) -#define ARGI_ACQUIRE_OP ARGI_LIST2 (ARGI_MUTEX, ARGI_NUMBER) +#define ARGI_STALL_OP ARGI_LIST1 (ARGI_INTEGER) +#define ARGI_SLEEP_OP ARGI_LIST1 (ARGI_INTEGER) +#define ARGI_ACQUIRE_OP ARGI_LIST2 (ARGI_MUTEX, ARGI_INTEGER) #define ARGI_SIGNAL_OP ARGI_LIST1 (ARGI_EVENT) -#define ARGI_WAIT_OP ARGI_LIST2 (ARGI_EVENT, ARGI_NUMBER) +#define ARGI_WAIT_OP ARGI_LIST2 (ARGI_EVENT, ARGI_INTEGER) #define ARGI_RESET_OP ARGI_LIST1 (ARGI_EVENT) #define ARGI_RELEASE_OP ARGI_LIST1 (ARGI_MUTEX) -#define ARGI_FROM_BCD_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_TARGETREF) -#define ARGI_TO_BCD_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_TARGETREF) +#define ARGI_FROM_BCD_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_TARGETREF) +#define ARGI_TO_BCD_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_FIXED_TARGET) #define ARGI_UNLOAD_OP ARGI_LIST1 (ARGI_DDBHANDLE) #define ARGI_REVISION_OP ARG_NONE #define ARGI_DEBUG_OP ARG_NONE -#define ARGI_FATAL_OP ARGI_LIST3 (ARGI_NUMBER, ARGI_NUMBER, ARGI_NUMBER) -#define ARGI_REGION_OP ARGI_LIST2 (ARGI_NUMBER, ARGI_NUMBER) +#define ARGI_FATAL_OP ARGI_LIST3 (ARGI_INTEGER, ARGI_INTEGER, ARGI_INTEGER) +#define ARGI_REGION_OP ARGI_LIST2 (ARGI_INTEGER, ARGI_INTEGER) #define ARGI_DEF_FIELD_OP ARGI_INVALID_OPCODE #define ARGI_DEVICE_OP ARGI_INVALID_OPCODE #define ARGI_PROCESSOR_OP ARGI_INVALID_OPCODE @@ -286,6 +315,7 @@ #define ARGI_THERMAL_ZONE_OP ARGI_INVALID_OPCODE #define ARGI_INDEX_FIELD_OP ARGI_INVALID_OPCODE #define ARGI_BANK_FIELD_OP ARGI_INVALID_OPCODE +#define ARGI_DATA_REGION_OP ARGI_LIST3 (ARGI_STRING, ARGI_STRING, ARGI_STRING) #define ARGI_LNOTEQUAL_OP ARGI_INVALID_OPCODE #define ARGI_LLESSEQUAL_OP ARGI_INVALID_OPCODE #define ARGI_LGREATEREQUAL_OP ARGI_INVALID_OPCODE @@ -307,8 +337,8 @@ static ACPI_OPCODE_INFO aml_op_info[] = { /* Index Opcode Type Class Has Arguments? Name Parser Args Interpreter Args */ -/* 00 */ /* AML_ZERO_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CONSTANT| AML_NO_ARGS, "Zero_op", ARGP_ZERO_OP, ARGI_ZERO_OP), -/* 01 */ /* AML_ONE_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CONSTANT| AML_NO_ARGS, "One_op", ARGP_ONE_OP, ARGI_ONE_OP), +/* 00 */ /* AML_ZERO_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CONSTANT| AML_NO_ARGS, "Zero", ARGP_ZERO_OP, ARGI_ZERO_OP), +/* 01 */ /* AML_ONE_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CONSTANT| AML_NO_ARGS, "One", ARGP_ONE_OP, ARGI_ONE_OP), /* 02 */ /* AML_ALIAS_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_NAMED_OBJECT| AML_HAS_ARGS, "Alias", ARGP_ALIAS_OP, ARGI_ALIAS_OP), /* 03 */ /* AML_NAME_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_NAMED_OBJECT| AML_HAS_ARGS, "Name", ARGP_NAME_OP, ARGI_NAME_OP), /* 04 */ /* AML_BYTE_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_LITERAL| AML_NO_ARGS, "Byte_const", ARGP_BYTE_OP, ARGI_BYTE_OP), @@ -337,7 +367,7 @@ static ACPI_OPCODE_INFO aml_op_info[] = /* 1_b */ /* AML_STORE_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2_r| AML_HAS_ARGS, "Store", ARGP_STORE_OP, ARGI_STORE_OP), /* 1_c */ /* AML_REF_OF_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2| AML_HAS_ARGS, "Ref_of", ARGP_REF_OF_OP, ARGI_REF_OF_OP), /* 1_d */ /* AML_ADD_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_DYADIC2_r| AML_HAS_ARGS, "Add", ARGP_ADD_OP, ARGI_ADD_OP), -/* 1_e */ /* AML_CONCAT_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_DYADIC2_r| AML_HAS_ARGS, "Concat", ARGP_CONCAT_OP, ARGI_CONCAT_OP), +/* 1_e */ /* AML_CONCAT_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_DYADIC2_r| AML_HAS_ARGS, "Concatenate", ARGP_CONCAT_OP, ARGI_CONCAT_OP), /* 1_f */ /* AML_SUBTRACT_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_DYADIC2_r| AML_HAS_ARGS, "Subtract", ARGP_SUBTRACT_OP, ARGI_SUBTRACT_OP), /* 20 */ /* AML_INCREMENT_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2| AML_HAS_ARGS, "Increment", ARGP_INCREMENT_OP, ARGI_INCREMENT_OP), /* 21 */ /* AML_DECREMENT_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2| AML_HAS_ARGS, "Decrement", ARGP_DECREMENT_OP, ARGI_DECREMENT_OP), @@ -376,7 +406,7 @@ static ACPI_OPCODE_INFO aml_op_info[] = /* 42 */ /* AML_RETURN_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CONTROL| AML_HAS_ARGS, "Return", ARGP_RETURN_OP, ARGI_RETURN_OP), /* 43 */ /* AML_BREAK_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CONTROL| AML_NO_ARGS, "Break", ARGP_BREAK_OP, ARGI_BREAK_OP), /* 44 */ /* AML_BREAK_POINT_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CONTROL| AML_NO_ARGS, "Break_point", ARGP_BREAK_POINT_OP, ARGI_BREAK_POINT_OP), -/* 45 */ /* AML_ONES_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CONSTANT| AML_NO_ARGS, "Ones_op", ARGP_ONES_OP, ARGI_ONES_OP), +/* 45 */ /* AML_ONES_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CONSTANT| AML_NO_ARGS, "Ones", ARGP_ONES_OP, ARGI_ONES_OP), /* Prefixed opcodes (Two-byte opcodes with a prefix op) */ @@ -402,7 +432,7 @@ static ACPI_OPCODE_INFO aml_op_info[] = /* 59 */ /* AML_DEF_FIELD_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_NAMED_OBJECT| AML_HAS_ARGS, "Field", ARGP_DEF_FIELD_OP, ARGI_DEF_FIELD_OP), /* 5_a */ /* AML_DEVICE_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_NAMED_OBJECT| AML_HAS_ARGS, "Device", ARGP_DEVICE_OP, ARGI_DEVICE_OP), /* 5_b */ /* AML_PROCESSOR_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_NAMED_OBJECT| AML_HAS_ARGS, "Processor", ARGP_PROCESSOR_OP, ARGI_PROCESSOR_OP), -/* 5_c */ /* AML_POWER_RES_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_NAMED_OBJECT| AML_HAS_ARGS, "Power_res", ARGP_POWER_RES_OP, ARGI_POWER_RES_OP), +/* 5_c */ /* AML_POWER_RES_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_NAMED_OBJECT| AML_HAS_ARGS, "Power_resource", ARGP_POWER_RES_OP, ARGI_POWER_RES_OP), /* 5_d */ /* AML_THERMAL_ZONE_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_NAMED_OBJECT| AML_HAS_ARGS, "Thermal_zone", ARGP_THERMAL_ZONE_OP, ARGI_THERMAL_ZONE_OP), /* 5_e */ /* AML_INDEX_FIELD_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_NAMED_OBJECT| AML_HAS_ARGS, "Index_field", ARGP_INDEX_FIELD_OP, ARGI_INDEX_FIELD_OP), /* 5_f */ /* AML_BANK_FIELD_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_NAMED_OBJECT| AML_HAS_ARGS, "Bank_field", ARGP_BANK_FIELD_OP, ARGI_BANK_FIELD_OP), @@ -423,6 +453,26 @@ static ACPI_OPCODE_INFO aml_op_info[] = /* 6_b */ /* UNKNOWN OPCODES */ OP_INFO_ENTRY (ACPI_OP_TYPE_UNKNOWN | OPTYPE_BOGUS| AML_HAS_ARGS, "UNKNOWN_OP!", ARG_NONE, ARG_NONE), /* 6_c */ /* ASCII CHARACTERS */ OP_INFO_ENTRY (ACPI_OP_TYPE_ASCII | OPTYPE_BOGUS| AML_HAS_ARGS, "ASCII_ONLY!", ARG_NONE, ARG_NONE), /* 6_d */ /* PREFIX CHARACTERS */ OP_INFO_ENTRY (ACPI_OP_TYPE_PREFIX | OPTYPE_BOGUS| AML_HAS_ARGS, "PREFIX_ONLY!", ARG_NONE, ARG_NONE), + + +/* ACPI 2.0 (new) opcodes */ + +/* 6_e */ /* AML_QWORD_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_LITERAL| AML_NO_ARGS, "Qword_const", ARGP_QWORD_OP, ARGI_QWORD_OP), +/* 6_f */ /* AML_VAR_PACKAGE_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_DATA_TERM| AML_HAS_ARGS, "Var_package", ARGP_VAR_PACKAGE_OP, ARGI_VAR_PACKAGE_OP), +/* 70 */ /* AML_CONCAT_RES_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_DYADIC2_r| AML_HAS_ARGS, "Concat_res", ARGP_CONCAT_RES_OP, ARGI_CONCAT_RES_OP), +/* 71 */ /* AML_MOD_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_DYADIC2_r| AML_HAS_ARGS, "Mod", ARGP_MOD_OP, ARGI_MOD_OP), +/* 72 */ /* AML_QWORD_FIELD_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CREATE_FIELD| AML_HAS_ARGS, "Create_qWord_field", ARGP_QWORD_FIELD_OP, ARGI_QWORD_FIELD_OP), +/* 73 */ /* AML_TO_BUFFER_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2_r| AML_HAS_ARGS, "To_buffer", ARGP_TO_BUFFER_OP, ARGI_TO_BUFFER_OP), +/* 74 */ /* AML_TO_DEC_STR_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2_r| AML_HAS_ARGS, "To_dec_string", ARGP_TO_DEC_STR_OP, ARGI_TO_DEC_STR_OP), +/* 75 */ /* AML_TO_HEX_STR_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2_r| AML_HAS_ARGS, "To_hex_string", ARGP_TO_HEX_STR_OP, ARGI_TO_HEX_STR_OP), +/* 76 */ /* AML_TO_INTEGER_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2_r| AML_HAS_ARGS, "To_integer", ARGP_TO_INTEGER_OP, ARGI_TO_INTEGER_OP), +/* 77 */ /* AML_TO_STRING_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2_r| AML_HAS_ARGS, "To_string", ARGP_TO_STRING_OP, ARGI_TO_STRING_OP), +/* 78 */ /* AML_COPY_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2_r| AML_HAS_ARGS, "Copy", ARGP_COPY_OP, ARGI_COPY_OP), +/* 79 */ /* AML_MID_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2_r| AML_HAS_ARGS, "Mid", ARGP_MID_OP, ARGI_MID_OP), +/* 7_a */ /* AML_CONTINUE_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_CONTROL| AML_NO_ARGS, "Continue", ARGP_CONTINUE_OP, ARGI_CONTINUE_OP), +/* 7_b */ /* AML_LOAD_TABLE_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2_r| AML_HAS_ARGS, "Load_table", ARGP_LOAD_TABLE_OP, ARGI_LOAD_TABLE_OP), +/* 7_c */ /* AML_DATA_REGION_OP */ OP_INFO_ENTRY (ACPI_OP_TYPE_OPCODE | OPTYPE_MONADIC2_r| AML_HAS_ARGS, "Data_op_region", ARGP_DATA_REGION_OP, ARGI_DATA_REGION_OP), + }; /* @@ -433,9 +483,10 @@ static ACPI_OPCODE_INFO aml_op_info[] = static u8 aml_short_op_info_index[256] = { /* 0 1 2 3 4 5 6 7 */ +/* 8 9 A B C D E F */ /* 0x00 */ 0x00, 0x01, _UNK, _UNK, _UNK, _UNK, 0x02, _UNK, -/* 0x08 */ 0x03, _UNK, 0x04, 0x05, 0x06, 0x07, _UNK, _UNK, -/* 0x10 */ 0x08, 0x09, 0x0a, _UNK, 0x0b, _UNK, _UNK, _UNK, +/* 0x08 */ 0x03, _UNK, 0x04, 0x05, 0x06, 0x07, 0x6E, _UNK, +/* 0x10 */ 0x08, 0x09, 0x0a, 0x6F, 0x0b, _UNK, _UNK, _UNK, /* 0x18 */ _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, /* 0x20 */ _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, /* 0x28 */ _UNK, _UNK, _UNK, _UNK, _UNK, 0x63, _PFX, _PFX, @@ -449,10 +500,10 @@ static u8 aml_short_op_info_index[256] = /* 0x68 */ 0x14, 0x15, 0x16, 0x17, 0x18, 0x19, 0x1a, _UNK, /* 0x70 */ 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20, 0x21, 0x22, /* 0x78 */ 0x23, 0x24, 0x25, 0x26, 0x27, 0x28, 0x29, 0x2a, -/* 0x80 */ 0x2b, 0x2c, 0x2d, 0x2e, _UNK, _UNK, 0x2f, 0x30, -/* 0x88 */ 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, _UNK, -/* 0x90 */ 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, _UNK, _UNK, -/* 0x98 */ _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, +/* 0x80 */ 0x2b, 0x2c, 0x2d, 0x2e, 0x70, 0x71, 0x2f, 0x30, +/* 0x88 */ 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, 0x72, +/* 0x90 */ 0x38, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x73, 0x74, +/* 0x98 */ 0x75, 0x76, _UNK, _UNK, 0x77, 0x78, 0x79, 0x7A, /* 0xA0 */ 0x3e, 0x3f, 0x40, 0x41, 0x42, 0x43, 0x60, 0x61, /* 0xA8 */ 0x62, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, /* 0xB0 */ _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, @@ -471,10 +522,11 @@ static u8 aml_short_op_info_index[256] = static u8 aml_long_op_info_index[NUM_EXTENDED_OPCODE] = { /* 0 1 2 3 4 5 6 7 */ +/* 8 9 A B C D E F */ /* 0x00 */ _UNK, 0x46, 0x47, _UNK, _UNK, _UNK, _UNK, _UNK, /* 0x08 */ _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, /* 0x10 */ _UNK, _UNK, 0x48, 0x49, _UNK, _UNK, _UNK, _UNK, -/* 0x18 */ _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, +/* 0x18 */ _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, 0x7B, /* 0x20 */ 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f, 0x50, 0x51, /* 0x28 */ 0x52, 0x53, 0x54, _UNK, _UNK, _UNK, _UNK, _UNK, /* 0x30 */ 0x55, 0x56, 0x57, _UNK, _UNK, _UNK, _UNK, _UNK, @@ -488,13 +540,10 @@ static u8 aml_long_op_info_index[NUM_EXTENDED_OPCODE] = /* 0x70 */ _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, /* 0x78 */ _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, _UNK, /* 0x80 */ 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, +/* 0x88 */ 0x7C, }; -/* 0 1 2 3 4 5 6 7 */ -/* 0x00 */ - - /******************************************************************************* * * FUNCTION: Acpi_ps_get_opcode_info diff --git a/drivers/acpi/parser/psparse.c b/drivers/acpi/parser/psparse.c index 7471efdcb..7373013da 100644 --- a/drivers/acpi/parser/psparse.c +++ b/drivers/acpi/parser/psparse.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: psparse - Parser top level AML parse routines - * $Revision: 71 $ + * $Revision: 74 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -537,6 +537,12 @@ acpi_ps_parse_loop ( */ status = acpi_ds_get_predicate_value (walk_state, NULL, TRUE); + if (ACPI_FAILURE (status) && + ((status & AE_CODE_MASK) != AE_CODE_CONTROL)) + { + return (status); + } + status = acpi_ps_next_parse_state (walk_state, op, status); } diff --git a/drivers/acpi/parser/psscope.c b/drivers/acpi/parser/psscope.c index 2e8926ad1..37b5d2265 100644 --- a/drivers/acpi/parser/psscope.c +++ b/drivers/acpi/parser/psscope.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: psscope - Parser scope stack management routines - * $Revision: 22 $ + * $Revision: 24 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/parser/pstree.c b/drivers/acpi/parser/pstree.c index a22bb646d..dc2e6d4d7 100644 --- a/drivers/acpi/parser/pstree.c +++ b/drivers/acpi/parser/pstree.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: pstree - Parser op tree manipulation/traversal/search - * $Revision: 25 $ + * $Revision: 27 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/parser/psutils.c b/drivers/acpi/parser/psutils.c index 3bac4a647..94518a920 100644 --- a/drivers/acpi/parser/psutils.c +++ b/drivers/acpi/parser/psutils.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: psutils - Parser miscellaneous utilities (Parser only) - * $Revision: 30 $ + * $Revision: 32 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/parser/pswalk.c b/drivers/acpi/parser/pswalk.c index 04a75917d..48c11eff3 100644 --- a/drivers/acpi/parser/pswalk.c +++ b/drivers/acpi/parser/pswalk.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: pswalk - Parser routines to walk parsed op tree(s) - * $Revision: 50 $ + * $Revision: 52 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/parser/psxface.c b/drivers/acpi/parser/psxface.c index a33ace051..328c755d9 100644 --- a/drivers/acpi/parser/psxface.c +++ b/drivers/acpi/parser/psxface.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: psxface - Parser external interfaces - * $Revision: 37 $ + * $Revision: 40 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -121,6 +121,12 @@ acpi_psx_execute ( return (AE_NO_MEMORY); } + + /* Init new op with the method name and pointer back to the NS node */ + + acpi_ps_set_name (op, method_node->name); + op->node = method_node; + /* * The walk of the parse tree is where we actually execute the method */ diff --git a/drivers/acpi/power.c b/drivers/acpi/power.c index 0422bf94d..d24392895 100644 --- a/drivers/acpi/power.c +++ b/drivers/acpi/power.c @@ -81,12 +81,12 @@ proc_read_ac_adapter_status(char *page, char **start, off_t off, buf.length = sizeof(obj); buf.pointer = &obj; if (!ACPI_SUCCESS(acpi_evaluate_object(ac_handle, "_PSR", NULL, &buf)) - || obj.type != ACPI_TYPE_NUMBER) { + || obj.type != ACPI_TYPE_INTEGER) { p += sprintf(p, "Could not read AC status\n"); goto end; } - if (obj.number.value) + if (obj.integer.value) p += sprintf(p, "on-line\n"); else p += sprintf(p, "off-line\n"); diff --git a/drivers/acpi/resources/rsaddr.c b/drivers/acpi/resources/rsaddr.c index e48666113..af91bc59f 100644 --- a/drivers/acpi/resources/rsaddr.c +++ b/drivers/acpi/resources/rsaddr.c @@ -4,12 +4,12 @@ * Acpi_rs_address16_stream * Acpi_rs_address32_resource * Acpi_rs_address32_stream - * $Revision: 12 $ + * $Revision: 14 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/resources/rscalc.c b/drivers/acpi/resources/rscalc.c index 2874aa3d0..86ff91be8 100644 --- a/drivers/acpi/resources/rscalc.c +++ b/drivers/acpi/resources/rscalc.c @@ -2,12 +2,12 @@ * * Module Name: rscalc - Acpi_rs_calculate_byte_stream_length * Acpi_rs_calculate_list_length - * $Revision: 16 $ + * $Revision: 18 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/resources/rscreate.c b/drivers/acpi/resources/rscreate.c index bc95686a7..f221a41dc 100644 --- a/drivers/acpi/resources/rscreate.c +++ b/drivers/acpi/resources/rscreate.c @@ -3,12 +3,12 @@ * Module Name: rscreate - Acpi_rs_create_resource_list * Acpi_rs_create_pci_routing_table * Acpi_rs_create_byte_stream - * $Revision: 22 $ + * $Revision: 24 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -223,9 +223,9 @@ acpi_rs_create_pci_routing_table ( /* * Dereference the Address */ - if (ACPI_TYPE_NUMBER == (*sub_object_list)->common.type) { + if (ACPI_TYPE_INTEGER == (*sub_object_list)->common.type) { user_prt->data.address = - (*sub_object_list)->number.value; + (*sub_object_list)->integer.value; } else { @@ -237,9 +237,9 @@ acpi_rs_create_pci_routing_table ( */ sub_object_list++; - if (ACPI_TYPE_NUMBER == (*sub_object_list)->common.type) { + if (ACPI_TYPE_INTEGER == (*sub_object_list)->common.type) { user_prt->data.pin = - (u32) (*sub_object_list)->number.value; + (u32) (*sub_object_list)->integer.value; } else { @@ -267,7 +267,7 @@ acpi_rs_create_pci_routing_table ( * is NULL, since the entire buffer was zeroed * out, we can leave this alone. */ - if (ACPI_TYPE_NUMBER == (*sub_object_list)->common.type) { + if (ACPI_TYPE_INTEGER == (*sub_object_list)->common.type) { /* * Add to the Length field the length of * the u32 NULL @@ -289,9 +289,9 @@ acpi_rs_create_pci_routing_table ( */ sub_object_list++; - if (ACPI_TYPE_NUMBER == (*sub_object_list)->common.type) { + if (ACPI_TYPE_INTEGER == (*sub_object_list)->common.type) { user_prt->data.source_index = - (u32) (*sub_object_list)->number.value; + (u32) (*sub_object_list)->integer.value; } else { diff --git a/drivers/acpi/resources/rsdump.c b/drivers/acpi/resources/rsdump.c index 073f4ddd4..03d2da6f2 100644 --- a/drivers/acpi/resources/rsdump.c +++ b/drivers/acpi/resources/rsdump.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: rsdump - Functions do dump out the resource structures. - * $Revision: 13 $ + * $Revision: 15 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/resources/rsio.c b/drivers/acpi/resources/rsio.c index 0d6c507c0..58fcf8cd8 100644 --- a/drivers/acpi/resources/rsio.c +++ b/drivers/acpi/resources/rsio.c @@ -6,12 +6,12 @@ * Acpi_rs_fixed_io_stream * Acpi_rs_dma_resource * Acpi_rs_dma_stream - * $Revision: 10 $ + * $Revision: 12 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/resources/rsirq.c b/drivers/acpi/resources/rsirq.c index 28a3a5045..eb47669ad 100644 --- a/drivers/acpi/resources/rsirq.c +++ b/drivers/acpi/resources/rsirq.c @@ -4,12 +4,12 @@ * Acpi_rs_irq_stream * Acpi_rs_extended_irq_resource * Acpi_rs_extended_irq_stream - * $Revision: 11 $ + * $Revision: 13 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/resources/rslist.c b/drivers/acpi/resources/rslist.c index 8e39ddded..de47563d0 100644 --- a/drivers/acpi/resources/rslist.c +++ b/drivers/acpi/resources/rslist.c @@ -2,12 +2,12 @@ * * Module Name: rslist - Acpi_rs_byte_stream_to_list * Acpi_list_to_byte_stream - * $Revision: 8 $ + * $Revision: 10 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/resources/rsmemory.c b/drivers/acpi/resources/rsmemory.c index 8aa4914f4..f9437e60c 100644 --- a/drivers/acpi/resources/rsmemory.c +++ b/drivers/acpi/resources/rsmemory.c @@ -6,12 +6,12 @@ * Acpi_rs_fixed_memory32_resource * Acpi_rs_memory32_range_stream * Acpi_rs_fixed_memory32_stream - * $Revision: 10 $ + * $Revision: 12 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/resources/rsmisc.c b/drivers/acpi/resources/rsmisc.c index 01bbb377e..be0a1a6a0 100644 --- a/drivers/acpi/resources/rsmisc.c +++ b/drivers/acpi/resources/rsmisc.c @@ -8,12 +8,12 @@ * Acpi_rs_end_dependent_functions_resource * Acpi_rs_start_dependent_functions_stream * Acpi_rs_end_dependent_functions_stream - * $Revision: 10 $ + * $Revision: 12 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/resources/rsutils.c b/drivers/acpi/resources/rsutils.c index 0a3f76668..2994bf0d3 100644 --- a/drivers/acpi/resources/rsutils.c +++ b/drivers/acpi/resources/rsutils.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: rsutils - Utilities for the resource manager - * $Revision: 12 $ + * $Revision: 14 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/resources/rsxface.c b/drivers/acpi/resources/rsxface.c index 8691f8200..5c126aaa5 100644 --- a/drivers/acpi/resources/rsxface.c +++ b/drivers/acpi/resources/rsxface.c @@ -1,12 +1,12 @@ /******************************************************************************* * * Module Name: rsxface - Public interfaces to the ACPI subsystem - * $Revision: 8 $ + * $Revision: 10 $ * ******************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/sys.c b/drivers/acpi/sys.c index 13648c255..2011174a1 100644 --- a/drivers/acpi/sys.c +++ b/drivers/acpi/sys.c @@ -27,18 +27,13 @@ #define _COMPONENT OS_DEPENDENT MODULE_NAME ("sys") -#define ACPI_SLP_TYP(typa, typb) (((int)(typa) << 8) | (int)(typb)) -#define ACPI_SLP_TYPA(value) ((value) >> 8) -#define ACPI_SLP_TYPB(value) ((value) & 0xff) - struct acpi_enter_sx_ctx { wait_queue_head_t wait; unsigned int state; }; -volatile acpi_sstate_t acpi_sleep_state = ACPI_S0; -static unsigned long acpi_slptyp[ACPI_S5 + 1] = {ACPI_INVALID,}; +volatile acpi_sstate_t acpi_sleep_state = ACPI_STATE_S0; /* * Enter system sleep state @@ -50,56 +45,32 @@ acpi_enter_sx_async(void *context) ACPI_OBJECT_LIST arg_list; ACPI_OBJECT arg; - /* - * _PSW methods could be run here to enable wake-on keyboard, LAN, etc. - */ - - // run the _PTS method - memset(&arg_list, 0, sizeof(arg_list)); - arg_list.count = 1; - arg_list.pointer = &arg; - - memset(&arg, 0, sizeof(arg)); - arg.type = ACPI_TYPE_NUMBER; - arg.number.value = ctx->state; + acpi_enter_sleep_state(ctx->state); - acpi_evaluate_object(NULL, "\\_PTS", &arg_list, NULL); - - // clear wake status by writing a 1 - acpi_hw_register_bit_access(ACPI_WRITE, ACPI_MTX_LOCK, WAK_STS, 1); - - acpi_sleep_state = ctx->state; - - // set ACPI_SLP_TYPA/b and ACPI_SLP_EN - acpi_hw_register_bit_access(ACPI_WRITE, ACPI_MTX_LOCK, SLP_TYPE_A, - ACPI_SLP_TYPA(acpi_slptyp[ctx->state])); - acpi_hw_register_bit_access(ACPI_WRITE, ACPI_MTX_LOCK, SLP_TYPE_B, - ACPI_SLP_TYPB(acpi_slptyp[ctx->state])); - acpi_hw_register_bit_access(ACPI_WRITE, ACPI_MTX_LOCK, SLP_EN, 1); - - if (ctx->state != ACPI_S1) { - /* we should have just shut off - what are we doing here? */ - printk(KERN_ERR "ACPI: S%d failed\n", ctx->state); + /* either we are in S1, or the transition failed, as the other states resume + from the waking vector */ + if (ctx->state != ACPI_STATE_S1) { + printk(KERN_ERR "Could not enter S%d\n", ctx->state); goto out; } - // wait until S1 is entered + /* wait until S1 is entered */ while (!(acpi_hw_register_bit_access(ACPI_READ, ACPI_MTX_LOCK, WAK_STS))) safe_halt(); - // run the _WAK method + /* run the _WAK method */ memset(&arg_list, 0, sizeof(arg_list)); arg_list.count = 1; arg_list.pointer = &arg; memset(&arg, 0, sizeof(arg)); - arg.type = ACPI_TYPE_NUMBER; - arg.number.value = ctx->state; + arg.type = ACPI_TYPE_INTEGER; + arg.integer.value = ctx->state; acpi_evaluate_object(NULL, "\\_WAK", &arg_list, NULL); out: - acpi_sleep_state = ACPI_S0; + acpi_sleep_state = ACPI_STATE_S0; if (waitqueue_active(&ctx->wait)) wake_up_interruptible(&ctx->wait); @@ -112,13 +83,9 @@ static void acpi_power_off(void) { struct acpi_enter_sx_ctx ctx; - - if ((STRNCMP(acpi_fadt.header.signature, ACPI_FADT_SIGNATURE, ACPI_SIG_LEN) != 0) - || acpi_slptyp[ACPI_S5] == ACPI_INVALID) - return; init_waitqueue_head(&ctx.wait); - ctx.state = ACPI_S5; + ctx.state = ACPI_STATE_S5; acpi_enter_sx_async(&ctx); } @@ -132,10 +99,6 @@ acpi_enter_sx(acpi_sstate_t state) DECLARE_WAITQUEUE(wait, current); int ret = 0; - if ((STRNCMP(acpi_fadt.header.signature, ACPI_FADT_SIGNATURE, ACPI_SIG_LEN) != 0) - || acpi_slptyp[state] == ACPI_INVALID) - return -EINVAL; - init_waitqueue_head(&ctx.wait); ctx.state = state; @@ -166,19 +129,14 @@ acpi_sys_init(void) printk(KERN_INFO "ACPI: System firmware supports:"); - for (sx = ACPI_S0; sx <= ACPI_S5; sx++) { - int ca_sx = (sx <= ACPI_S4) ? sx : (sx + 1); + for (sx = ACPI_STATE_S0; sx <= ACPI_STATE_S5; sx++) { if (ACPI_SUCCESS( - acpi_hw_obtain_sleep_type_register_data(ca_sx, + acpi_hw_obtain_sleep_type_register_data(sx, &type_a, &type_b))) { - acpi_slptyp[sx] = ACPI_SLP_TYP(type_a, type_b); printk(" S%d", sx); } - else { - acpi_slptyp[sx] = ACPI_INVALID; - } } printk("\n"); diff --git a/drivers/acpi/table.c b/drivers/acpi/table.c index 8379b4204..de0f99a27 100644 --- a/drivers/acpi/table.c +++ b/drivers/acpi/table.c @@ -47,21 +47,6 @@ acpi_fetch_fadt(void) return -ENODEV; } - if (acpi_fadt.plvl2_lat - && acpi_fadt.plvl2_lat <= ACPI_MAX_P_LVL2_LAT) { - acpi_c2_exit_latency - = ACPI_MICROSEC_TO_TMR_TICKS(acpi_fadt.plvl2_lat); - acpi_c2_enter_latency - = ACPI_MICROSEC_TO_TMR_TICKS(ACPI_TMR_HZ / 1000); - } - if (acpi_fadt.plvl3_lat - && acpi_fadt.plvl3_lat <= ACPI_MAX_P_LVL3_LAT) { - acpi_c3_exit_latency - = ACPI_MICROSEC_TO_TMR_TICKS(acpi_fadt.plvl3_lat); - acpi_c3_enter_latency - = ACPI_MICROSEC_TO_TMR_TICKS(acpi_fadt.plvl3_lat * 5); - } - return 0; } @@ -71,11 +56,7 @@ acpi_fetch_fadt(void) int acpi_find_and_load_tables(u64 rsdp) { - if (ACPI_SUCCESS(acpi_load_tables(rsdp))) - { - printk(KERN_INFO "ACPI: System description tables loaded\n"); - } - else { + if (!ACPI_SUCCESS(acpi_load_tables(rsdp))) { printk(KERN_INFO "ACPI: System description table load failed\n"); acpi_terminate(); return -1; diff --git a/drivers/acpi/tables/tbconvrt.c b/drivers/acpi/tables/tbconvrt.c index 624926d27..76d7fff15 100644 --- a/drivers/acpi/tables/tbconvrt.c +++ b/drivers/acpi/tables/tbconvrt.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: tbconvrt - ACPI Table conversion utilities - * $Revision: 15 $ + * $Revision: 18 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -47,7 +47,7 @@ a.register_bit_width = (u8) MUL_8 (b);\ a.register_bit_offset = 0;\ a.reserved = 0;\ - a.address = (UINT64) c;} + ACPI_STORE_ADDRESS (a.address,c);} /* ACPI V1.0 entries -- address space is always I/O */ @@ -126,8 +126,8 @@ acpi_tb_convert_to_xsdt ( new_table->table_offset_entry[i] = ((RSDT_DESCRIPTOR_REV071 *) table_info->pointer)->table_offset_entry[i]; #else - new_table->table_offset_entry[i] = - ((RSDT_DESCRIPTOR_REV1 *) table_info->pointer)->table_offset_entry[i]; + ACPI_STORE_ADDRESS (new_table->table_offset_entry[i], + ((RSDT_DESCRIPTOR_REV1 *) table_info->pointer)->table_offset_entry[i]); #endif } else { @@ -312,11 +312,11 @@ acpi_tb_convert_table_fadt (void) /* No 0.71 equivalence. Leave pre-zeroed. */ /* FADT2->Flush_stride = 0; */ - /* Processor’s duty cycle index in processor's P_CNT reg*/ + /* Processor's duty cycle index in processor's P_CNT reg*/ /* No 0.71 equivalence. Leave pre-zeroed. */ /* FADT2->Duty_offset = 0; */ - /* Processor’s duty cycle value bit width in P_CNT register.*/ + /* Processor's duty cycle value bit width in P_CNT register.*/ /* No 0.71 equivalence. Leave pre-zeroed. */ /* FADT2->Duty_width = 0; */ @@ -384,8 +384,8 @@ acpi_tb_convert_table_fadt (void) /* Convert table pointers to 64-bit fields */ - FADT2->Xfirmware_ctrl = (UINT64) FADT1->firmware_ctrl; - FADT2->Xdsdt = (UINT64) FADT1->dsdt; + ACPI_STORE_ADDRESS (FADT2->Xfirmware_ctrl, FADT1->firmware_ctrl); + ACPI_STORE_ADDRESS (FADT2->Xdsdt, FADT1->dsdt); /* System Interrupt Model isn't used in ACPI 2.0*/ /* FADT2->Reserved1 = 0; */ @@ -448,6 +448,7 @@ acpi_tb_convert_table_fadt (void) * Global FADT pointer will point to the common V2.0 FADT */ acpi_gbl_FADT = FADT2; + acpi_gbl_FADT->header.length = sizeof (FADT_DESCRIPTOR); /* Free the original table */ @@ -464,8 +465,6 @@ acpi_tb_convert_table_fadt (void) table_desc->length = sizeof (FADT_DESCRIPTOR_REV2); - /* Dump the FADT Header */ - /* Dump the entire FADT */ diff --git a/drivers/acpi/tables/tbget.c b/drivers/acpi/tables/tbget.c index 2cf8eede7..8e05e2c7d 100644 --- a/drivers/acpi/tables/tbget.c +++ b/drivers/acpi/tables/tbget.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: tbget - ACPI Table get* routines - * $Revision: 40 $ + * $Revision: 43 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -243,7 +243,7 @@ acpi_tb_get_all_tables ( /* Get the table via the XSDT */ status = acpi_tb_get_table ((ACPI_PHYSICAL_ADDRESS) - acpi_gbl_XSDT->table_offset_entry[index], + ACPI_GET_ADDRESS (acpi_gbl_XSDT->table_offset_entry[index]), table_ptr, &table_info); /* Ignore a table that failed verification */ @@ -326,7 +326,8 @@ acpi_tb_get_all_tables ( * Get the DSDT (We know that the FADT is valid now) */ - status = acpi_tb_get_table (acpi_gbl_FADT->Xdsdt, table_ptr, &table_info); + status = acpi_tb_get_table ((ACPI_PHYSICAL_ADDRESS) ACPI_GET_ADDRESS (acpi_gbl_FADT->Xdsdt), + table_ptr, &table_info); if (ACPI_FAILURE (status)) { return (status); } @@ -472,13 +473,14 @@ acpi_tb_get_table_rsdt ( /* 0.71 RSDP has 64bit Rsdt address field */ physical_address = ((RSDP_DESCRIPTOR_REV071 *)acpi_gbl_RSDP)->rsdt_physical_address; #else - physical_address = acpi_gbl_RSDP->rsdt_physical_address; + physical_address = (ACPI_PHYSICAL_ADDRESS) acpi_gbl_RSDP->rsdt_physical_address; #endif table_signature = RSDT_SIG; signature_length = sizeof (RSDT_SIG) -1; } else { - physical_address = (ACPI_PHYSICAL_ADDRESS) acpi_gbl_RSDP->xsdt_physical_address; + physical_address = (ACPI_PHYSICAL_ADDRESS) + ACPI_GET_ADDRESS (acpi_gbl_RSDP->xsdt_physical_address); table_signature = XSDT_SIG; signature_length = sizeof (XSDT_SIG) -1; } @@ -586,7 +588,7 @@ acpi_tb_get_table_facs ( else { /* Just map the physical memory to our address space */ - status = acpi_tb_map_acpi_table (acpi_gbl_FADT->Xfirmware_ctrl, + status = acpi_tb_map_acpi_table ((ACPI_PHYSICAL_ADDRESS) ACPI_GET_ADDRESS (acpi_gbl_FADT->Xfirmware_ctrl), &size, &table_ptr); if (ACPI_FAILURE(status)) { return (status); diff --git a/drivers/acpi/tables/tbinstal.c b/drivers/acpi/tables/tbinstal.c index b3926a0e3..e753917b7 100644 --- a/drivers/acpi/tables/tbinstal.c +++ b/drivers/acpi/tables/tbinstal.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: tbinstal - ACPI table installation and removal - * $Revision: 34 $ + * $Revision: 36 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/tables/tbutils.c b/drivers/acpi/tables/tbutils.c index baae6fb4c..d89b1508c 100644 --- a/drivers/acpi/tables/tbutils.c +++ b/drivers/acpi/tables/tbutils.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: tbutils - Table manipulation utilities - * $Revision: 31 $ + * $Revision: 33 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/tables/tbxface.c b/drivers/acpi/tables/tbxface.c index ee9eba62a..98c99873c 100644 --- a/drivers/acpi/tables/tbxface.c +++ b/drivers/acpi/tables/tbxface.c @@ -2,12 +2,12 @@ * * Module Name: tbxface - Public interfaces to the ACPI subsystem * ACPI table oriented interfaces - * $Revision: 32 $ + * $Revision: 34 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/acpi/tables/tbxfroot.c b/drivers/acpi/tables/tbxfroot.c index 4c31f1c37..606e8ad10 100644 --- a/drivers/acpi/tables/tbxfroot.c +++ b/drivers/acpi/tables/tbxfroot.c @@ -1,12 +1,12 @@ /****************************************************************************** * * Module Name: tbxfroot - Find the root ACPI table (RSDT) - * $Revision: 33 $ + * $Revision: 35 $ * *****************************************************************************/ /* - * Copyright (C) 2000 R. Byron Moore + * Copyright (C) 2000, 2001 R. Byron Moore * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by diff --git a/drivers/atm/Makefile b/drivers/atm/Makefile index 2157f1570..5b602f9ba 100644 --- a/drivers/atm/Makefile +++ b/drivers/atm/Makefile @@ -46,7 +46,7 @@ ifeq ($(CONFIG_ATM_FORE200E_SBA),y) endif endif -obj-$(CONFIG_ATM_FORE200E) += fore200e.o $(FORE200E_FW_OBJS) +obj-$(CONFIG_ATM_FORE200E) += fore_200e.o EXTRA_CFLAGS=-g diff --git a/drivers/block/DAC960.c b/drivers/block/DAC960.c index 5e8dc19f9..d0aac4c7d 100644 --- a/drivers/block/DAC960.c +++ b/drivers/block/DAC960.c @@ -1820,7 +1820,6 @@ static int DAC960_BackMergeFunction(RequestQueue_T *RequestQueue, Request->nr_segments < Controller->DriverScatterGatherLimit) { Request->nr_segments++; - RequestQueue->elevator.nr_segments++; return true; } return false; @@ -1844,7 +1843,6 @@ static int DAC960_FrontMergeFunction(RequestQueue_T *RequestQueue, Request->nr_segments < Controller->DriverScatterGatherLimit) { Request->nr_segments++; - RequestQueue->elevator.nr_segments++; return true; } return false; @@ -1864,17 +1862,12 @@ static int DAC960_MergeRequestsFunction(RequestQueue_T *RequestQueue, DAC960_Controller_T *Controller = (DAC960_Controller_T *) RequestQueue->queuedata; int TotalSegments = Request->nr_segments + NextRequest->nr_segments; - int SameSegment = 0; if (Request->bhtail->b_data + Request->bhtail->b_size == NextRequest->bh->b_data) - { TotalSegments--; - SameSegment = 1; - } if (TotalSegments > MaxSegments || TotalSegments > Controller->DriverScatterGatherLimit) return false; - RequestQueue->elevator.nr_segments -= SameSegment; Request->nr_segments = TotalSegments; return true; } @@ -2834,6 +2827,7 @@ static void DAC960_RequestFunction(RequestQueue_T *RequestQueue) static inline void DAC960_ProcessCompletedBuffer(BufferHeader_T *BufferHeader, boolean SuccessfulIO) { + blk_finished_io(BufferHeader->b_size >> 9); BufferHeader->b_end_io(BufferHeader, SuccessfulIO); } diff --git a/drivers/block/cciss.c b/drivers/block/cciss.c index 35afe2a7f..47937edcc 100644 --- a/drivers/block/cciss.c +++ b/drivers/block/cciss.c @@ -1086,6 +1086,7 @@ static inline void complete_buffers( struct buffer_head *bh, int status) { xbh = bh->b_reqnext; bh->b_reqnext = NULL; + blk_finished_io(bh->b_size >> 9); bh->b_end_io(bh, status); bh = xbh; } diff --git a/drivers/block/cpqarray.c b/drivers/block/cpqarray.c index afbf60ddc..7016cfff4 100644 --- a/drivers/block/cpqarray.c +++ b/drivers/block/cpqarray.c @@ -140,23 +140,7 @@ static int ida_release(struct inode *inode, struct file *filep); static int ida_ioctl(struct inode *inode, struct file *filep, unsigned int cmd, unsigned long arg); static int ida_ctlr_ioctl(int ctlr, int dsk, ida_ioctl_t *io); -static void do_ida_request(int i); -/* - * This is a hack. This driver eats a major number for each controller, and - * sets blkdev[xxx].request_fn to each one of these so the real request - * function knows what controller its working with. - */ -#define DO_IDA_REQUEST(x) { do_ida_request(x); } - -static void do_ida_request0(request_queue_t * q) DO_IDA_REQUEST(0); -static void do_ida_request1(request_queue_t * q) DO_IDA_REQUEST(1); -static void do_ida_request2(request_queue_t * q) DO_IDA_REQUEST(2); -static void do_ida_request3(request_queue_t * q) DO_IDA_REQUEST(3); -static void do_ida_request4(request_queue_t * q) DO_IDA_REQUEST(4); -static void do_ida_request5(request_queue_t * q) DO_IDA_REQUEST(5); -static void do_ida_request6(request_queue_t * q) DO_IDA_REQUEST(6); -static void do_ida_request7(request_queue_t * q) DO_IDA_REQUEST(7); - +static void do_ida_request(request_queue_t *q); static void start_io(ctlr_info_t *h); static inline void addQ(cmdlist_t **Qptr, cmdlist_t *c); @@ -362,6 +346,47 @@ void cleanup_module(void) } #endif /* MODULE */ +static inline int cpq_new_segment(request_queue_t *q, struct request *rq, + int max_segments) +{ + if (rq->nr_segments < SG_MAX) { + rq->nr_segments++; + return 1; + } + return 0; +} + +static int cpq_back_merge_fn(request_queue_t *q, struct request *rq, + struct buffer_head *bh, int max_segments) +{ + if (rq->bhtail->b_data + rq->bhtail->b_size == bh->b_data) + return 1; + return cpq_new_segment(q, rq, max_segments); +} + +static int cpq_front_merge_fn(request_queue_t *q, struct request *rq, + struct buffer_head *bh, int max_segments) +{ + if (bh->b_data + bh->b_size == rq->bh->b_data) + return 1; + return cpq_new_segment(q, rq, max_segments); +} + +static int cpq_merge_requests_fn(request_queue_t *q, struct request *rq, + struct request *nxt, int max_segments) +{ + int total_segments = rq->nr_segments + nxt->nr_segments; + + if (rq->bhtail->b_data + rq->bhtail->b_size == nxt->bh->b_data) + total_segments--; + + if (total_segments > SG_MAX) + return 0; + + rq->nr_segments = total_segments; + return 1; +} + /* * This is it. Find all the controllers and register them. I really hate * stealing all these major device numbers. @@ -369,12 +394,7 @@ void cleanup_module(void) */ int __init cpqarray_init(void) { - void (*request_fns[MAX_CTLR])(request_queue_t *) = { - do_ida_request0, do_ida_request1, - do_ida_request2, do_ida_request3, - do_ida_request4, do_ida_request5, - do_ida_request6, do_ida_request7, - }; + request_queue_t *q; int i,j; int num_cntlrs_reg = 0; @@ -495,16 +515,20 @@ int __init cpqarray_init(void) hba[i]->access.set_intr_mask(hba[i], FIFO_NOT_EMPTY); - ida_procinit(i); - blk_init_queue(BLK_DEFAULT_QUEUE(MAJOR_NR + i), - request_fns[i]); - blk_queue_headactive(BLK_DEFAULT_QUEUE(MAJOR_NR + i), 0); + q = BLK_DEFAULT_QUEUE(MAJOR_NR + i); + q->queuedata = hba[i]; + blk_init_queue(q, do_ida_request); + blk_queue_headactive(q, 0); blksize_size[MAJOR_NR+i] = ida_blocksizes + (i*256); hardsect_size[MAJOR_NR+i] = ida_hardsizes + (i*256); read_ahead[MAJOR_NR+i] = READ_AHEAD; + q->back_merge_fn = cpq_back_merge_fn; + q->front_merge_fn = cpq_front_merge_fn; + q->merge_requests_fn = cpq_merge_requests_fn; + ida_gendisk[i].major = MAJOR_NR + i; ida_gendisk[i].major_name = "ida"; ida_gendisk[i].minor_shift = NWD_SHIFT; @@ -872,37 +896,30 @@ static inline cmdlist_t *removeQ(cmdlist_t **Qptr, cmdlist_t *c) * are in here (either via the dummy do_ida_request functions or by being * called from the interrupt handler */ -static void do_ida_request(int ctlr) +static void do_ida_request(request_queue_t *q) { - ctlr_info_t *h = hba[ctlr]; + ctlr_info_t *h = q->queuedata; cmdlist_t *c; int seg, sect; char *lastdataend; - struct list_head * queue_head; + struct list_head * queue_head = &q->queue_head; struct buffer_head *bh; struct request *creq; - queue_head = &blk_dev[MAJOR_NR+ctlr].request_queue.queue_head; - - if (list_empty(queue_head)) - { + if (q->plugged || list_empty(queue_head)) { start_io(h); return; } creq = blkdev_entry_next_request(queue_head); - if (creq->rq_status == RQ_INACTIVE) - { - start_io(h); - return; - } - + if (creq->nr_segments > SG_MAX) + BUG(); - if (ctlr != MAJOR(creq->rq_dev)-MAJOR_NR || - ctlr > nr_ctlr || h == NULL) + if (h->ctlr != MAJOR(creq->rq_dev)-MAJOR_NR || h->ctlr > nr_ctlr) { printk(KERN_WARNING "doreq cmd for %d, %x at %p\n", - ctlr, creq->rq_dev, creq); + h->ctlr, creq->rq_dev, creq); + blkdev_dequeue_request(creq); complete_buffers(creq->bh, 0); start_io(h); return; @@ -916,12 +933,12 @@ static void do_ida_request(int ctlr) bh = creq->bh; - c->ctlr = ctlr; + c->ctlr = h->ctlr; c->hdr.unit = MINOR(creq->rq_dev) >> NWD_SHIFT; c->hdr.size = sizeof(rblk_t) >> 2; c->size += sizeof(rblk_t); - c->req.hdr.blk = ida[(ctlr<<CTLR_SHIFT) + MINOR(creq->rq_dev)].start_sect + creq->sector; + c->req.hdr.blk = ida[(h->ctlr<<CTLR_SHIFT) + MINOR(creq->rq_dev)].start_sect + creq->sector; c->bh = bh; DBGPX( if (bh == NULL) @@ -933,21 +950,16 @@ DBGPX( sect = 0; while(bh) { sect += bh->b_size/512; -DBGPX( - if (bh->b_size % 512) { - printk("Oh damn. %d+%d, size = %d\n", creq->sector, sect, bh->b_size); - panic("b_size %% 512 != 0"); - } -); if (bh->b_data == lastdataend) { c->req.sg[seg-1].size += bh->b_size; lastdataend += bh->b_size; } else { + if (seg == SG_MAX) + BUG(); c->req.sg[seg].size = bh->b_size; c->req.sg[seg].addr = (__u32)virt_to_bus(bh->b_data); lastdataend = bh->b_data + bh->b_size; - if (++seg == SG_MAX) - break; + seg++; } bh = bh->b_reqnext; } @@ -955,30 +967,24 @@ DBGPX( printk("Submitting %d sectors in %d segments\n", sect, seg); ); c->req.hdr.sg_cnt = seg; c->req.hdr.blk_cnt = sect; - creq->sector += sect; - creq->nr_sectors -= sect; - - /* Ready the next request: - * Fix up creq if we still have more buffers in the buffer chain, or - * mark this request as done and ready the next one. + /* + * Since we control our own merging, we know that this request + * is now fully setup and there's nothing left. */ - if (creq->nr_sectors) { -DBGPX( - if (bh==NULL) { - printk("sector=%d, nr_sectors=%d, sect=%d, seg=%d\n", - creq->sector, creq->nr_sectors, sect, seg); - panic("mother..."); - } -); - creq->bh = bh->b_reqnext; - bh->b_reqnext = NULL; -DBGPX( printk("More to do on same request %p\n", creq); ); - } else { -DBGPX( printk("Done with %p\n", creq); ); - blkdev_dequeue_request(creq); - end_that_request_last(creq); + if (creq->nr_sectors != sect) { + printk("ida: %ld != %d sectors\n", creq->nr_sectors, sect); + BUG(); } + blkdev_dequeue_request(creq); + + /* + * ehh, we can't really end the request here since it's not + * even started yet. for now it shouldn't hurt though + */ +DBGPX( printk("Done with %p\n", creq); ); + end_that_request_last(creq); + c->req.hdr.cmd = (creq->cmd == READ) ? IDA_READ : IDA_WRITE; c->type = CMD_RWREQ; @@ -1025,6 +1031,7 @@ static inline void complete_buffers(struct buffer_head *bh, int ok) xbh = bh->b_reqnext; bh->b_reqnext = NULL; + blk_finished_io(bh->b_size >> 9); bh->b_end_io(bh, ok); bh = xbh; @@ -1072,7 +1079,6 @@ static void do_ida_intr(int irq, void *dev_id, struct pt_regs *regs) unsigned long flags; __u32 a,a1; - istat = h->access.intr_pending(h); /* Is this interrupt for us? */ if (istat == 0) @@ -1116,7 +1122,7 @@ static void do_ida_intr(int irq, void *dev_id, struct pt_regs *regs) /* * See if we can queue up some more IO */ - do_ida_request(h->ctlr); + do_ida_request(BLK_DEFAULT_QUEUE(MAJOR_NR + h->ctlr)); spin_unlock_irqrestore(&io_request_lock, flags); } diff --git a/drivers/block/elevator.c b/drivers/block/elevator.c index 1200773c2..9917ad055 100644 --- a/drivers/block/elevator.c +++ b/drivers/block/elevator.c @@ -24,125 +24,115 @@ #include <linux/blkdev.h> #include <linux/elevator.h> #include <linux/blk.h> +#include <linux/module.h> #include <asm/uaccess.h> -/* - * Order ascending, but only allow a request to be skipped a certain - * number of times - */ -void elevator_linus(struct request *req, elevator_t *elevator, - struct list_head *real_head, - struct list_head *head, int orig_latency) -{ - struct list_head *entry = real_head; - struct request *tmp; - - req->elevator_sequence = orig_latency; - - while ((entry = entry->prev) != head) { - tmp = blkdev_entry_to_request(entry); - if (IN_ORDER(tmp, req)) - break; - if (!tmp->elevator_sequence) - break; - tmp->elevator_sequence--; - } - list_add(&req->queue, entry); -} - int elevator_linus_merge(request_queue_t *q, struct request **req, + struct list_head * head, struct buffer_head *bh, int rw, - int *max_sectors, int *max_segments) + int max_sectors, int max_segments) { - struct list_head *entry, *head = &q->queue_head; + struct list_head *entry = &q->queue_head; unsigned int count = bh->b_size >> 9, ret = ELEVATOR_NO_MERGE; - entry = head; - if (q->head_active && !q->plugged) - head = head->next; - while ((entry = entry->prev) != head) { - struct request *__rq = *req = blkdev_entry_to_request(entry); + struct request *__rq = blkdev_entry_to_request(entry); + + /* + * simply "aging" of requests in queue + */ + if (__rq->elevator_sequence-- <= 0) { + *req = __rq; + break; + } + if (__rq->sem) continue; if (__rq->cmd != rw) continue; - if (__rq->nr_sectors + count > *max_sectors) - continue; if (__rq->rq_dev != bh->b_rdev) continue; + if (__rq->nr_sectors + count > max_sectors) + continue; + if (__rq->elevator_sequence < count) + break; if (__rq->sector + __rq->nr_sectors == bh->b_rsector) { ret = ELEVATOR_BACK_MERGE; + *req = __rq; break; - } - if (!__rq->elevator_sequence) - break; - if (__rq->sector - count == bh->b_rsector) { - __rq->elevator_sequence--; + } else if (__rq->sector - count == bh->b_rsector) { ret = ELEVATOR_FRONT_MERGE; + __rq->elevator_sequence -= count; + *req = __rq; break; - } + } else if (!*req && BHRQ_IN_ORDER(bh, __rq)) + *req = __rq; } + return ret; +} + +void elevator_linus_merge_cleanup(request_queue_t *q, struct request *req, int count) +{ + struct list_head *entry = &req->queue, *head = &q->queue_head; + /* * second pass scan of requests that got passed over, if any */ - if (ret != ELEVATOR_NO_MERGE && *req) { - while ((entry = entry->next) != &q->queue_head) { - struct request *tmp = blkdev_entry_to_request(entry); - tmp->elevator_sequence--; - } + while ((entry = entry->next) != head) { + struct request *tmp = blkdev_entry_to_request(entry); + tmp->elevator_sequence -= count; } - - return ret; } -/* - * No request sorting, just add it to the back of the list - */ -void elevator_noop(struct request *req, elevator_t *elevator, - struct list_head *real_head, struct list_head *head, - int orig_latency) +void elevator_linus_merge_req(struct request *req, struct request *next) { - list_add_tail(&req->queue, real_head); + if (next->elevator_sequence < req->elevator_sequence) + req->elevator_sequence = next->elevator_sequence; } /* - * See if we can find a request that is buffer can be coalesced with. + * See if we can find a request that this buffer can be coalesced with. */ int elevator_noop_merge(request_queue_t *q, struct request **req, + struct list_head * head, struct buffer_head *bh, int rw, - int *max_sectors, int *max_segments) + int max_sectors, int max_segments) { - struct list_head *entry, *head = &q->queue_head; + struct list_head *entry; unsigned int count = bh->b_size >> 9; - if (q->head_active && !q->plugged) - head = head->next; + if (list_empty(&q->queue_head)) + return ELEVATOR_NO_MERGE; - entry = head; + entry = &q->queue_head; while ((entry = entry->prev) != head) { - struct request *__rq = *req = blkdev_entry_to_request(entry); - if (__rq->sem) - continue; + struct request *__rq = blkdev_entry_to_request(entry); + if (__rq->cmd != rw) continue; - if (__rq->nr_sectors + count > *max_sectors) - continue; if (__rq->rq_dev != bh->b_rdev) continue; - if (__rq->sector + __rq->nr_sectors == bh->b_rsector) + if (__rq->nr_sectors + count > max_sectors) + continue; + if (__rq->sem) + continue; + if (__rq->sector + __rq->nr_sectors == bh->b_rsector) { + *req = __rq; return ELEVATOR_BACK_MERGE; - if (__rq->sector - count == bh->b_rsector) + } else if (__rq->sector - count == bh->b_rsector) { + *req = __rq; return ELEVATOR_FRONT_MERGE; + } } + + *req = blkdev_entry_to_request(q->queue_head.prev); return ELEVATOR_NO_MERGE; } -/* - * The noop "elevator" does not do any accounting - */ -void elevator_noop_dequeue(struct request *req) {} +void elevator_noop_merge_cleanup(request_queue_t *q, struct request *req, int count) {} + +void elevator_noop_merge_req(struct request *req, struct request *next) {} int blkelvget_ioctl(elevator_t * elevator, blkelv_ioctl_arg_t * arg) { diff --git a/drivers/block/ll_rw_blk.c b/drivers/block/ll_rw_blk.c index ef71dddc7..836a89f1e 100644 --- a/drivers/block/ll_rw_blk.c +++ b/drivers/block/ll_rw_blk.c @@ -19,6 +19,7 @@ #include <linux/config.h> #include <linux/locks.h> #include <linux/mm.h> +#include <linux/swap.h> #include <linux/init.h> #include <linux/smp_lock.h> @@ -38,8 +39,6 @@ extern int mac_floppy_init(void); #endif -extern int lvm_init(void); - /* * For the allocated request tables */ @@ -118,6 +117,19 @@ int * max_readahead[MAX_BLKDEV]; */ int * max_sectors[MAX_BLKDEV]; +/* + * queued sectors for all devices, used to make sure we don't fill all + * of memory with locked buffers + */ +atomic_t queued_sectors; + +/* + * high and low watermark for above + */ +static int high_queued_sectors, low_queued_sectors; +static int batch_requests, queue_nr_requests; +static DECLARE_WAIT_QUEUE_HEAD(blk_buffers_wait); + static inline int get_max_sectors(kdev_t dev) { if (!max_sectors[MAJOR(dev)]) @@ -125,7 +137,7 @@ static inline int get_max_sectors(kdev_t dev) return max_sectors[MAJOR(dev)][MINOR(dev)]; } -static inline request_queue_t *__blk_get_queue(kdev_t dev) +inline request_queue_t *__blk_get_queue(kdev_t dev) { struct blk_dev_struct *bdev = blk_dev + MAJOR(dev); @@ -153,17 +165,14 @@ request_queue_t *blk_get_queue(kdev_t dev) static int __blk_cleanup_queue(struct list_head *head) { - struct list_head *entry; struct request *rq; int i = 0; if (list_empty(head)) return 0; - entry = head->next; do { - rq = list_entry(entry, struct request, table); - entry = entry->next; + rq = list_entry(head->next, struct request, table); list_del(&rq->table); kmem_cache_free(request_cachep, rq); i++; @@ -188,10 +197,12 @@ static int __blk_cleanup_queue(struct list_head *head) **/ void blk_cleanup_queue(request_queue_t * q) { - int count = QUEUE_NR_REQUESTS; + int count = queue_nr_requests; count -= __blk_cleanup_queue(&q->request_freelist[READ]); count -= __blk_cleanup_queue(&q->request_freelist[WRITE]); + count -= __blk_cleanup_queue(&q->pending_freelist[READ]); + count -= __blk_cleanup_queue(&q->pending_freelist[WRITE]); if (count) printk("blk_cleanup_queue: leaked requests (%d)\n", count); @@ -290,7 +301,6 @@ static inline int ll_new_segment(request_queue_t *q, struct request *req, int ma { if (req->nr_segments < max_segments) { req->nr_segments++; - q->elevator.nr_segments++; return 1; } return 0; @@ -316,18 +326,13 @@ static int ll_merge_requests_fn(request_queue_t *q, struct request *req, struct request *next, int max_segments) { int total_segments = req->nr_segments + next->nr_segments; - int same_segment; - same_segment = 0; - if (req->bhtail->b_data + req->bhtail->b_size == next->bh->b_data) { + if (req->bhtail->b_data + req->bhtail->b_size == next->bh->b_data) total_segments--; - same_segment = 1; - } if (total_segments > max_segments) return 0; - q->elevator.nr_segments -= same_segment; req->nr_segments = total_segments; return 1; } @@ -364,7 +369,7 @@ static inline void __generic_unplug_device(request_queue_t *q) } } -static void generic_unplug_device(void *data) +void generic_unplug_device(void *data) { request_queue_t *q = (request_queue_t *) data; unsigned long flags; @@ -379,19 +384,24 @@ static void blk_init_free_list(request_queue_t *q) struct request *rq; int i; + INIT_LIST_HEAD(&q->request_freelist[READ]); + INIT_LIST_HEAD(&q->request_freelist[WRITE]); + INIT_LIST_HEAD(&q->pending_freelist[READ]); + INIT_LIST_HEAD(&q->pending_freelist[WRITE]); + q->pending_free[READ] = q->pending_free[WRITE] = 0; + /* - * Divide requests in half between read and write. This used to - * be a 2/3 advantage for reads, but now reads can steal from - * the write free list. + * Divide requests in half between read and write */ - for (i = 0; i < QUEUE_NR_REQUESTS; i++) { + for (i = 0; i < queue_nr_requests; i++) { rq = kmem_cache_alloc(request_cachep, SLAB_KERNEL); + memset(rq, 0, sizeof(struct request)); rq->rq_status = RQ_INACTIVE; list_add(&rq->table, &q->request_freelist[i & 1]); } init_waitqueue_head(&q->wait_for_request); - spin_lock_init(&q->request_lock); + spin_lock_init(&q->queue_lock); } static int __make_request(request_queue_t * q, int rw, struct buffer_head * bh); @@ -426,14 +436,12 @@ static int __make_request(request_queue_t * q, int rw, struct buffer_head * bh); * blk_queue_headactive(). * * Note: - * blk_init_queue() must be paired with a blk_cleanup-queue() call + * blk_init_queue() must be paired with a blk_cleanup_queue() call * when the block device is deactivated (such as at module unload). **/ void blk_init_queue(request_queue_t * q, request_fn_proc * rfn) { INIT_LIST_HEAD(&q->queue_head); - INIT_LIST_HEAD(&q->request_freelist[READ]); - INIT_LIST_HEAD(&q->request_freelist[WRITE]); elevator_init(&q->elevator, ELEVATOR_LINUS); blk_init_free_list(q); q->request_fn = rfn; @@ -455,7 +463,6 @@ void blk_init_queue(request_queue_t * q, request_fn_proc * rfn) q->head_active = 1; } - #define blkdev_free_rq(list) list_entry((list)->next, struct request, table); /* * Get a free request. io_request_lock must be held and interrupts @@ -463,37 +470,16 @@ void blk_init_queue(request_queue_t * q, request_fn_proc * rfn) */ static inline struct request *get_request(request_queue_t *q, int rw) { - struct list_head *list = &q->request_freelist[rw]; - struct request *rq; - - /* - * Reads get preferential treatment and are allowed to steal - * from the write free list if necessary. - */ - if (!list_empty(list)) { - rq = blkdev_free_rq(list); - goto got_rq; - } + struct request *rq = NULL; - /* - * if the WRITE list is non-empty, we know that rw is READ - * and that the READ list is empty. allow reads to 'steal' - * from the WRITE list. - */ - if (!list_empty(&q->request_freelist[WRITE])) { - list = &q->request_freelist[WRITE]; - rq = blkdev_free_rq(list); - goto got_rq; + if (!list_empty(&q->request_freelist[rw])) { + rq = blkdev_free_rq(&q->request_freelist[rw]); + list_del(&rq->table); + rq->rq_status = RQ_ACTIVE; + rq->special = NULL; + rq->q = q; } - return NULL; - -got_rq: - list_del(&rq->table); - rq->free_list = list; - rq->rq_status = RQ_ACTIVE; - rq->special = NULL; - rq->q = q; return rq; } @@ -581,40 +567,42 @@ inline void drive_stat_acct (kdev_t dev, int rw, /* * add-request adds a request to the linked list. - * It disables interrupts (acquires the request spinlock) so that it can muck - * with the request-lists in peace. Thus it should be called with no spinlocks - * held. + * io_request_lock is held and interrupts disabled, as we muck with the + * request queue list. * * By this point, req->cmd is always either READ/WRITE, never READA, * which is important for drive_stat_acct() above. */ - static inline void add_request(request_queue_t * q, struct request * req, - struct list_head *head, int lat) + struct list_head *insert_here) { int major; drive_stat_acct(req->rq_dev, req->cmd, req->nr_sectors, 1); + if (!q->plugged && q->head_active && insert_here == &q->queue_head) { + spin_unlock_irq(&io_request_lock); + BUG(); + } + /* - * let selected elevator insert the request + * elevator indicated where it wants this request to be + * inserted at elevator_merge time */ - q->elevator.elevator_fn(req, &q->elevator, &q->queue_head, head, lat); - - /* - * FIXME(eric) I don't understand why there is a need for this - * special case code. It clearly doesn't fit any more with - * the new queueing architecture, and it got added in 2.3.10. - * I am leaving this in here until I hear back from the COMPAQ - * people. - */ + list_add(&req->queue, insert_here); + major = MAJOR(req->rq_dev); - if (major >= COMPAQ_SMART2_MAJOR+0 && major <= COMPAQ_SMART2_MAJOR+7) - (q->request_fn)(q); - if (major >= COMPAQ_CISS_MAJOR+0 && major <= COMPAQ_CISS_MAJOR+7) - (q->request_fn)(q); if (major >= DAC960_MAJOR+0 && major <= DAC960_MAJOR+7) - (q->request_fn)(q); + q->request_fn(q); +} + +void inline blk_refill_freelist(request_queue_t *q, int rw) +{ + if (q->pending_free[rw]) { + list_splice(&q->pending_freelist[rw], &q->request_freelist[rw]); + INIT_LIST_HEAD(&q->pending_freelist[rw]); + q->pending_free[rw] = 0; + } } /* @@ -622,15 +610,34 @@ static inline void add_request(request_queue_t * q, struct request * req, */ void inline blkdev_release_request(struct request *req) { + request_queue_t *q = req->q; + int rw = req->cmd; + req->rq_status = RQ_INACTIVE; + req->q = NULL; /* - * Request may not have originated from ll_rw_blk + * Request may not have originated from ll_rw_blk. if not, + * asumme it has free buffers and check waiters */ - if (req->free_list) { - list_add(&req->table, req->free_list); - req->free_list = NULL; - wake_up(&req->q->wait_for_request); + if (q) { + /* + * we've released enough buffers to start I/O again + */ + if (waitqueue_active(&blk_buffers_wait) + && atomic_read(&queued_sectors) < low_queued_sectors) + wake_up(&blk_buffers_wait); + + /* + * Add to pending free list and batch wakeups + */ + list_add(&req->table, &q->pending_freelist[rw]); + + if (++q->pending_free[rw] >= batch_requests) { + int wake_up = q->pending_free[rw]; + blk_refill_freelist(q, rw); + wake_up_nr(&q->wait_for_request, wake_up); + } } } @@ -658,9 +665,10 @@ static void attempt_merge(request_queue_t * q, * will have been updated to the appropriate number, * and we shouldn't do it here too. */ - if(!(q->merge_requests_fn)(q, req, next, max_segments)) + if (!q->merge_requests_fn(q, req, next, max_segments)) return; + q->elevator.elevator_merge_req_fn(req, next); req->bhtail->b_reqnext = next->bh; req->bhtail = next->bhtail; req->nr_sectors = req->hard_nr_sectors += next->hard_nr_sectors; @@ -699,7 +707,7 @@ static int __make_request(request_queue_t * q, int rw, int max_segments = MAX_SEGMENTS; struct request * req = NULL, *freereq = NULL; int rw_ahead, max_sectors, el_ret; - struct list_head *head; + struct list_head *head, *insert_here; int latency; elevator_t *elevator = &q->elevator; @@ -713,6 +721,7 @@ static int __make_request(request_queue_t * q, int rw, rw = READ; /* drop into READ */ case READ: case WRITE: + latency = elevator_request_latency(elevator, rw); break; default: BUG(); @@ -741,38 +750,33 @@ static int __make_request(request_queue_t * q, int rw, */ max_sectors = get_max_sectors(bh->b_rdev); - latency = elevator_request_latency(elevator, rw); - +again: + head = &q->queue_head; /* * Now we acquire the request spinlock, we have to be mega careful * not to schedule or do something nonatomic */ -again: spin_lock_irq(&io_request_lock); - /* - * skip first entry, for devices with active queue head - */ - head = &q->queue_head; - if (q->head_active && !q->plugged) - head = head->next; - + insert_here = head->prev; if (list_empty(head)) { q->plug_device_fn(q, bh->b_rdev); /* is atomic */ goto get_rq; - } + } else if (q->head_active && !q->plugged) + head = head->next; - el_ret = elevator->elevator_merge_fn(q, &req, bh, rw, - &max_sectors, &max_segments); + el_ret = elevator->elevator_merge_fn(q, &req, head, bh, rw, + max_sectors, max_segments); switch (el_ret) { case ELEVATOR_BACK_MERGE: if (!q->back_merge_fn(q, req, bh, max_segments)) break; + elevator->elevator_merge_cleanup_fn(q, req, count); req->bhtail->b_reqnext = bh; req->bhtail = bh; req->nr_sectors = req->hard_nr_sectors += count; - req->e = elevator; + blk_started_io(count); drive_stat_acct(req->rq_dev, req->cmd, count, 0); attempt_back_merge(q, req, max_sectors, max_segments); goto out; @@ -780,20 +784,29 @@ again: case ELEVATOR_FRONT_MERGE: if (!q->front_merge_fn(q, req, bh, max_segments)) break; + elevator->elevator_merge_cleanup_fn(q, req, count); bh->b_reqnext = req->bh; req->bh = bh; req->buffer = bh->b_data; req->current_nr_sectors = count; req->sector = req->hard_sector = sector; req->nr_sectors = req->hard_nr_sectors += count; - req->e = elevator; + blk_started_io(count); drive_stat_acct(req->rq_dev, req->cmd, count, 0); attempt_front_merge(q, head, req, max_sectors, max_segments); goto out; + /* * elevator says don't/can't merge. get new request */ case ELEVATOR_NO_MERGE: + /* + * use elevator hints as to where to insert the + * request. if no hints, just add it to the back + * of the queue + */ + if (req) + insert_here = &req->queue; break; default: @@ -802,10 +815,9 @@ again: } /* - * Grab a free request from the freelist. Read first try their - * own queue - if that is empty, we steal from the write list. - * Writes must block if the write list is empty, and read aheads - * are not crucial. + * Grab a free request from the freelist - if that is empty, check + * if we are doing read ahead and abort instead of blocking for + * a free slot. */ get_rq: if (freereq) { @@ -821,6 +833,7 @@ get_rq: } /* fill up the request-info, and add it to the queue */ + req->elevator_sequence = latency; req->cmd = rw; req->errors = 0; req->hard_sector = req->sector = sector; @@ -833,13 +846,13 @@ get_rq: req->bh = bh; req->bhtail = bh; req->rq_dev = bh->b_rdev; - req->e = elevator; - add_request(q, req, head, latency); + blk_started_io(count); + add_request(q, req, insert_here); out: - if (!q->plugged) - (q->request_fn)(q); if (freereq) blkdev_release_request(freereq); + if (!q->plugged) + q->request_fn(q); spin_unlock_irq(&io_request_lock); return 0; end_io: @@ -886,13 +899,13 @@ void generic_make_request (int rw, struct buffer_head * bh) int major = MAJOR(bh->b_rdev); request_queue_t *q; - if (!bh->b_end_io) BUG(); + if (!bh->b_end_io) + BUG(); + if (blk_size[major]) { unsigned long maxsector = (blk_size[major][MINOR(bh->b_rdev)] << 1) + 1; - unsigned int sector, count; - - count = bh->b_size >> 9; - sector = bh->b_rsector; + unsigned long sector = bh->b_rsector; + unsigned int count = bh->b_size >> 9; if (maxsector < count || maxsector - count < sector) { bh->b_state &= (1 << BH_Lock) | (1 << BH_Mapped); @@ -903,7 +916,7 @@ void generic_make_request (int rw, struct buffer_head * bh) when mounting a device. */ printk(KERN_INFO "attempt to access beyond end of device\n"); - printk(KERN_INFO "%s: rw=%d, want=%d, limit=%d\n", + printk(KERN_INFO "%s: rw=%d, want=%ld, limit=%d\n", kdevname(bh->b_rdev), rw, (sector + count)>>1, blk_size[major][MINOR(bh->b_rdev)]); @@ -930,15 +943,13 @@ void generic_make_request (int rw, struct buffer_head * bh) buffer_IO_error(bh); break; } - - } - while (q->make_request_fn(q, rw, bh)); + } while (q->make_request_fn(q, rw, bh)); } /** * submit_bh: submit a buffer_head to the block device later for I/O - * @rw: whether to %READ or %WRITE, or mayve to %READA (read ahead) + * @rw: whether to %READ or %WRITE, or maybe to %READA (read ahead) * @bh: The &struct buffer_head which describes the I/O * * submit_bh() is very similar in purpose to generic_make_request(), and @@ -961,7 +972,7 @@ void submit_bh(int rw, struct buffer_head * bh) * further remap this. */ bh->b_rdev = bh->b_dev; - bh->b_rsector = bh->b_blocknr * (bh->b_size>>9); + bh->b_rsector = bh->b_blocknr * (bh->b_size >> 9); generic_make_request(rw, bh); @@ -1021,6 +1032,9 @@ void ll_rw_block(int rw, int nr, struct buffer_head * bhs[]) int correct_size; int i; + if (!nr) + return; + major = MAJOR(bhs[0]->b_dev); /* Determine correct block size for this device. */ @@ -1033,9 +1047,8 @@ void ll_rw_block(int rw, int nr, struct buffer_head * bhs[]) /* Verify requested block sizes. */ for (i = 0; i < nr; i++) { - struct buffer_head *bh; - bh = bhs[i]; - if (bh->b_size != correct_size) { + struct buffer_head *bh = bhs[i]; + if (bh->b_size % correct_size) { printk(KERN_NOTICE "ll_rw_block: device %s: " "only %d-char blocks implemented (%u)\n", kdevname(bhs[0]->b_dev), @@ -1051,8 +1064,17 @@ void ll_rw_block(int rw, int nr, struct buffer_head * bhs[]) } for (i = 0; i < nr; i++) { - struct buffer_head *bh; - bh = bhs[i]; + struct buffer_head *bh = bhs[i]; + + /* + * don't lock any more buffers if we are above the high + * water mark. instead start I/O on the queued stuff. + */ + if (atomic_read(&queued_sectors) >= high_queued_sectors) { + run_task_queue(&tq_disk); + wait_event(blk_buffers_wait, + atomic_read(&queued_sectors) < low_queued_sectors); + } /* Only one thread can actually submit the I/O. */ if (test_and_set_bit(BH_Lock, &bh->b_state)) @@ -1096,12 +1118,25 @@ sorry: extern int stram_device_init (void); #endif -/* - * First step of what used to be end_request + +/** + * end_that_request_first - end I/O on one buffer. + * @req: the request being processed + * @uptodate: 0 for I/O error + * @name: the name printed for an I/O error * - * 0 means continue with end_that_request_last, - * 1 means we are done - */ + * Description: + * Ends I/O on the first buffer attached to @req, and sets it up + * for the next buffer_head (if any) in the cluster. + * + * Return: + * 0 - we are done with this request, call end_that_request_last() + * 1 - still buffers pending for this request + * + * Caveat: + * Drivers implementing their own end_request handling must call + * blk_finished_io() appropriately. + **/ int end_that_request_first (struct request *req, int uptodate, char *name) { @@ -1115,6 +1150,7 @@ int end_that_request_first (struct request *req, int uptodate, char *name) if ((bh = req->bh) != NULL) { nsect = bh->b_size >> 9; + blk_finished_io(nsect); req->bh = bh->b_reqnext; bh->b_reqnext = NULL; bh->b_end_io(bh, uptodate); @@ -1138,19 +1174,18 @@ int end_that_request_first (struct request *req, int uptodate, char *name) void end_that_request_last(struct request *req) { - if (req->e) { - printk("end_that_request_last called with non-dequeued req\n"); - BUG(); - } if (req->sem != NULL) up(req->sem); blkdev_release_request(req); } +#define MB(kb) ((kb) << 10) + int __init blk_dev_init(void) { struct blk_dev_struct *dev; + int total_ram; request_cachep = kmem_cache_create("blkdev_requests", sizeof(struct request), @@ -1165,6 +1200,44 @@ int __init blk_dev_init(void) memset(ro_bits,0,sizeof(ro_bits)); memset(max_readahead, 0, sizeof(max_readahead)); memset(max_sectors, 0, sizeof(max_sectors)); + + atomic_set(&queued_sectors, 0); + total_ram = nr_free_pages() << (PAGE_SHIFT - 10); + + /* + * Try to keep 128MB max hysteris. If not possible, + * use half of RAM + */ + high_queued_sectors = (total_ram * 2) / 3; + low_queued_sectors = high_queued_sectors / 3; + if (high_queued_sectors - low_queued_sectors > MB(128)) + low_queued_sectors = high_queued_sectors - MB(128); + + + /* + * make it sectors (512b) + */ + high_queued_sectors <<= 1; + low_queued_sectors <<= 1; + + /* + * Scale free request slots per queue too + */ + total_ram = (total_ram + MB(32) - 1) & ~(MB(32) - 1); + if ((queue_nr_requests = total_ram >> 9) > QUEUE_NR_REQUESTS) + queue_nr_requests = QUEUE_NR_REQUESTS; + + /* + * adjust batch frees according to queue length, with upper limit + */ + if ((batch_requests = queue_nr_requests >> 3) > 32) + batch_requests = 32; + + printk("block: queued sectors max/low %dkB/%dkB, %d slots per queue\n", + high_queued_sectors / 2, + low_queued_sectors / 2, + queue_nr_requests); + #ifdef CONFIG_AMIGA_Z2RAM z2_init(); #endif @@ -1268,9 +1341,6 @@ int __init blk_dev_init(void) #ifdef CONFIG_SUN_JSFLASH jsfd_init(); #endif -#ifdef CONFIG_BLK_DEV_LVM - lvm_init(); -#endif return 0; }; @@ -1279,9 +1349,12 @@ EXPORT_SYMBOL(end_that_request_first); EXPORT_SYMBOL(end_that_request_last); EXPORT_SYMBOL(blk_init_queue); EXPORT_SYMBOL(blk_get_queue); +EXPORT_SYMBOL(__blk_get_queue); EXPORT_SYMBOL(blk_cleanup_queue); EXPORT_SYMBOL(blk_queue_headactive); EXPORT_SYMBOL(blk_queue_pluggable); EXPORT_SYMBOL(blk_queue_make_request); EXPORT_SYMBOL(generic_make_request); EXPORT_SYMBOL(blkdev_release_request); +EXPORT_SYMBOL(generic_unplug_device); +EXPORT_SYMBOL(queued_sectors); diff --git a/drivers/block/paride/pd.c b/drivers/block/paride/pd.c index 9b1ff3a32..172acadbe 100644 --- a/drivers/block/paride/pd.c +++ b/drivers/block/paride/pd.c @@ -392,7 +392,6 @@ static inline int pd_new_segment(request_queue_t *q, struct request *req, int ma if (req->nr_segments < max_segments) { req->nr_segments++; - q->elevator.nr_segments++; return 1; } return 0; @@ -432,7 +431,6 @@ static int pd_merge_requests_fn(request_queue_t *q, struct request *req, if (total_segments > max_segments) return 0; - q->elevator.nr_segments -= same_segment; req->nr_segments = total_segments; return 1; } diff --git a/drivers/block/paride/pf.c b/drivers/block/paride/pf.c index 53f6ca3df..b71703f9b 100644 --- a/drivers/block/paride/pf.c +++ b/drivers/block/paride/pf.c @@ -346,7 +346,6 @@ static inline int pf_new_segment(request_queue_t *q, struct request *req, int ma if (req->nr_segments < max_segments) { req->nr_segments++; - q->elevator.nr_segments++; return 1; } return 0; @@ -386,7 +385,6 @@ static int pf_merge_requests_fn(request_queue_t *q, struct request *req, if (total_segments > max_segments) return 0; - q->elevator.nr_segments -= same_segment; req->nr_segments = total_segments; return 1; } diff --git a/drivers/cdrom/cdrom.c b/drivers/cdrom/cdrom.c index 1e555da55..cafe72afd 100644 --- a/drivers/cdrom/cdrom.c +++ b/drivers/cdrom/cdrom.c @@ -295,11 +295,11 @@ MODULE_PARM(check_media_type, "i"); /* These are used to simplify getting data in from and back to user land */ #define IOCTL_IN(arg, type, in) \ - if (copy_from_user(&in, (type *) arg, sizeof in)) \ + if (copy_from_user(&(in), (type *) (arg), sizeof (in))) \ return -EFAULT; #define IOCTL_OUT(arg, type, out) \ - if (copy_to_user((type *) arg, &out, sizeof out)) \ + if (copy_to_user((type *) (arg), &(out), sizeof (out))) \ return -EFAULT; /* The (cdo->capability & ~cdi->mask & CDC_XXX) construct was used in diff --git a/drivers/char/agp/agpgart_be.c b/drivers/char/agp/agpgart_be.c index 942142832..ccd53dacc 100644 --- a/drivers/char/agp/agpgart_be.c +++ b/drivers/char/agp/agpgart_be.c @@ -1384,9 +1384,11 @@ static void via_cleanup(void) aper_size_info_8 *previous_size; previous_size = A_SIZE_8(agp_bridge.previous_size); - pci_write_config_dword(agp_bridge.dev, VIA_ATTBASE, 0); pci_write_config_byte(agp_bridge.dev, VIA_APSIZE, previous_size->size_value); + /* Do not disable by writing 0 to VIA_ATTBASE, it screws things up + * during reinitialization. + */ } static void via_tlbflush(agp_memory * mem) @@ -2373,9 +2375,10 @@ static int __init agp_find_supported_device(void) if (i810_dev == NULL) { printk(KERN_ERR PFX "agpgart: Detected an " "Intel i815, but could not find the" - " secondary device.\n"); - agp_bridge.type = NOT_SUPPORTED; - return -ENODEV; + " secondary device.\n" + "Assuming user has added an external AGP" + " card\n"); + break; } printk(KERN_INFO PFX "agpgart: Detected an Intel i815 " "Chipset.\n"); diff --git a/drivers/char/drm/Config.in b/drivers/char/drm/Config.in index c69d37db5..6d7649eb0 100644 --- a/drivers/char/drm/Config.in +++ b/drivers/char/drm/Config.in @@ -9,7 +9,8 @@ bool 'Direct Rendering Manager (XFree86 DRI support)' CONFIG_DRM if [ "$CONFIG_DRM" != "n" ]; then tristate ' 3dfx Banshee/Voodoo3+' CONFIG_DRM_TDFX tristate ' 3dlabs GMX 2000' CONFIG_DRM_GAMMA - tristate ' ATI Rage 128' CONFIG_DRM_R128 + dep_tristate ' ATI Rage 128' CONFIG_DRM_R128 $CONFIG_AGP + dep_tristate ' ATI Radeon' CONFIG_DRM_RADEON $CONFIG_AGP dep_tristate ' Intel I810' CONFIG_DRM_I810 $CONFIG_AGP dep_tristate ' Matrox g200/g400' CONFIG_DRM_MGA $CONFIG_AGP fi diff --git a/drivers/char/drm/Makefile b/drivers/char/drm/Makefile index d25e33d47..41017340b 100644 --- a/drivers/char/drm/Makefile +++ b/drivers/char/drm/Makefile @@ -42,16 +42,18 @@ else endif endif -gamma-objs := gamma_drv.o gamma_dma.o -tdfx-objs := tdfx_drv.o tdfx_context.o -r128-objs := r128_drv.o r128_cce.o r128_context.o r128_bufs.o r128_state.o -ffb-objs := ffb_drv.o ffb_context.o -mga-objs := mga_drv.o mga_dma.o mga_context.o mga_bufs.o mga_state.o -i810-objs := i810_drv.o i810_dma.o i810_context.o i810_bufs.o +gamma-objs := gamma_drv.o gamma_dma.o +tdfx-objs := tdfx_drv.o tdfx_context.o +r128-objs := r128_drv.o r128_cce.o r128_context.o r128_bufs.o r128_state.o +ffb-objs := ffb_drv.o ffb_context.o +mga-objs := mga_drv.o mga_dma.o mga_context.o mga_bufs.o mga_state.o +i810-objs := i810_drv.o i810_dma.o i810_context.o i810_bufs.o +radeon-objs := radeon_drv.o radeon_cp.o radeon_context.o radeon_bufs.o radeon_state.o obj-$(CONFIG_DRM_GAMMA) += gamma.o obj-$(CONFIG_DRM_TDFX) += tdfx.o obj-$(CONFIG_DRM_R128) += r128.o +obj-$(CONFIG_DRM_RADEON)+= radeon.o obj-$(CONFIG_DRM_FFB) += ffb.o obj-$(CONFIG_DRM_MGA) += mga.o obj-$(CONFIG_DRM_I810) += i810.o @@ -96,5 +98,8 @@ i810.o: $(i810-objs) $(lib) r128.o: $(r128-objs) $(lib) $(LD) -r -o $@ $(r128-objs) $(lib) +radeon.o: $(radeon-objs) $(lib) + $(LD) -r -o $@ $(radeon-objs) $(lib) + ffb.o: $(ffb-objs) $(lib) $(LD) -r -o $@ $(ffb-objs) $(lib) diff --git a/drivers/char/drm/drm.h b/drivers/char/drm/drm.h index aeecca77b..101556540 100644 --- a/drivers/char/drm/drm.h +++ b/drivers/char/drm/drm.h @@ -83,6 +83,7 @@ typedef struct drm_clip_rect { #include "mga_drm.h" #include "i810_drm.h" #include "r128_drm.h" +#include "radeon_drm.h" #ifdef CONFIG_DRM_SIS #include "sis_drm.h" #endif @@ -298,100 +299,116 @@ typedef struct drm_agp_info { unsigned short id_device; } drm_agp_info_t; -#define DRM_IOCTL_BASE 'd' -#define DRM_IO(nr) _IO(DRM_IOCTL_BASE,nr) -#define DRM_IOR(nr,size) _IOR(DRM_IOCTL_BASE,nr,size) -#define DRM_IOW(nr,size) _IOW(DRM_IOCTL_BASE,nr,size) -#define DRM_IOWR(nr,size) _IOWR(DRM_IOCTL_BASE,nr,size) - - -#define DRM_IOCTL_VERSION DRM_IOWR(0x00, drm_version_t) -#define DRM_IOCTL_GET_UNIQUE DRM_IOWR(0x01, drm_unique_t) -#define DRM_IOCTL_GET_MAGIC DRM_IOR( 0x02, drm_auth_t) -#define DRM_IOCTL_IRQ_BUSID DRM_IOWR(0x03, drm_irq_busid_t) - -#define DRM_IOCTL_SET_UNIQUE DRM_IOW( 0x10, drm_unique_t) -#define DRM_IOCTL_AUTH_MAGIC DRM_IOW( 0x11, drm_auth_t) -#define DRM_IOCTL_BLOCK DRM_IOWR(0x12, drm_block_t) -#define DRM_IOCTL_UNBLOCK DRM_IOWR(0x13, drm_block_t) -#define DRM_IOCTL_CONTROL DRM_IOW( 0x14, drm_control_t) -#define DRM_IOCTL_ADD_MAP DRM_IOWR(0x15, drm_map_t) -#define DRM_IOCTL_ADD_BUFS DRM_IOWR(0x16, drm_buf_desc_t) -#define DRM_IOCTL_MARK_BUFS DRM_IOW( 0x17, drm_buf_desc_t) -#define DRM_IOCTL_INFO_BUFS DRM_IOWR(0x18, drm_buf_info_t) -#define DRM_IOCTL_MAP_BUFS DRM_IOWR(0x19, drm_buf_map_t) -#define DRM_IOCTL_FREE_BUFS DRM_IOW( 0x1a, drm_buf_free_t) - -#define DRM_IOCTL_ADD_CTX DRM_IOWR(0x20, drm_ctx_t) -#define DRM_IOCTL_RM_CTX DRM_IOWR(0x21, drm_ctx_t) -#define DRM_IOCTL_MOD_CTX DRM_IOW( 0x22, drm_ctx_t) -#define DRM_IOCTL_GET_CTX DRM_IOWR(0x23, drm_ctx_t) -#define DRM_IOCTL_SWITCH_CTX DRM_IOW( 0x24, drm_ctx_t) -#define DRM_IOCTL_NEW_CTX DRM_IOW( 0x25, drm_ctx_t) -#define DRM_IOCTL_RES_CTX DRM_IOWR(0x26, drm_ctx_res_t) -#define DRM_IOCTL_ADD_DRAW DRM_IOWR(0x27, drm_draw_t) -#define DRM_IOCTL_RM_DRAW DRM_IOWR(0x28, drm_draw_t) -#define DRM_IOCTL_DMA DRM_IOWR(0x29, drm_dma_t) -#define DRM_IOCTL_LOCK DRM_IOW( 0x2a, drm_lock_t) -#define DRM_IOCTL_UNLOCK DRM_IOW( 0x2b, drm_lock_t) -#define DRM_IOCTL_FINISH DRM_IOW( 0x2c, drm_lock_t) - -#define DRM_IOCTL_AGP_ACQUIRE DRM_IO( 0x30) -#define DRM_IOCTL_AGP_RELEASE DRM_IO( 0x31) -#define DRM_IOCTL_AGP_ENABLE DRM_IOW( 0x32, drm_agp_mode_t) -#define DRM_IOCTL_AGP_INFO DRM_IOR( 0x33, drm_agp_info_t) -#define DRM_IOCTL_AGP_ALLOC DRM_IOWR(0x34, drm_agp_buffer_t) -#define DRM_IOCTL_AGP_FREE DRM_IOW( 0x35, drm_agp_buffer_t) -#define DRM_IOCTL_AGP_BIND DRM_IOW( 0x36, drm_agp_binding_t) -#define DRM_IOCTL_AGP_UNBIND DRM_IOW( 0x37, drm_agp_binding_t) +#define DRM_IOCTL_BASE 'd' +#define DRM_IO(nr) _IO(DRM_IOCTL_BASE,nr) +#define DRM_IOR(nr,size) _IOR(DRM_IOCTL_BASE,nr,size) +#define DRM_IOW(nr,size) _IOW(DRM_IOCTL_BASE,nr,size) +#define DRM_IOWR(nr,size) _IOWR(DRM_IOCTL_BASE,nr,size) + + +#define DRM_IOCTL_VERSION DRM_IOWR(0x00, drm_version_t) +#define DRM_IOCTL_GET_UNIQUE DRM_IOWR(0x01, drm_unique_t) +#define DRM_IOCTL_GET_MAGIC DRM_IOR( 0x02, drm_auth_t) +#define DRM_IOCTL_IRQ_BUSID DRM_IOWR(0x03, drm_irq_busid_t) + +#define DRM_IOCTL_SET_UNIQUE DRM_IOW( 0x10, drm_unique_t) +#define DRM_IOCTL_AUTH_MAGIC DRM_IOW( 0x11, drm_auth_t) +#define DRM_IOCTL_BLOCK DRM_IOWR(0x12, drm_block_t) +#define DRM_IOCTL_UNBLOCK DRM_IOWR(0x13, drm_block_t) +#define DRM_IOCTL_CONTROL DRM_IOW( 0x14, drm_control_t) +#define DRM_IOCTL_ADD_MAP DRM_IOWR(0x15, drm_map_t) +#define DRM_IOCTL_ADD_BUFS DRM_IOWR(0x16, drm_buf_desc_t) +#define DRM_IOCTL_MARK_BUFS DRM_IOW( 0x17, drm_buf_desc_t) +#define DRM_IOCTL_INFO_BUFS DRM_IOWR(0x18, drm_buf_info_t) +#define DRM_IOCTL_MAP_BUFS DRM_IOWR(0x19, drm_buf_map_t) +#define DRM_IOCTL_FREE_BUFS DRM_IOW( 0x1a, drm_buf_free_t) + +#define DRM_IOCTL_ADD_CTX DRM_IOWR(0x20, drm_ctx_t) +#define DRM_IOCTL_RM_CTX DRM_IOWR(0x21, drm_ctx_t) +#define DRM_IOCTL_MOD_CTX DRM_IOW( 0x22, drm_ctx_t) +#define DRM_IOCTL_GET_CTX DRM_IOWR(0x23, drm_ctx_t) +#define DRM_IOCTL_SWITCH_CTX DRM_IOW( 0x24, drm_ctx_t) +#define DRM_IOCTL_NEW_CTX DRM_IOW( 0x25, drm_ctx_t) +#define DRM_IOCTL_RES_CTX DRM_IOWR(0x26, drm_ctx_res_t) +#define DRM_IOCTL_ADD_DRAW DRM_IOWR(0x27, drm_draw_t) +#define DRM_IOCTL_RM_DRAW DRM_IOWR(0x28, drm_draw_t) +#define DRM_IOCTL_DMA DRM_IOWR(0x29, drm_dma_t) +#define DRM_IOCTL_LOCK DRM_IOW( 0x2a, drm_lock_t) +#define DRM_IOCTL_UNLOCK DRM_IOW( 0x2b, drm_lock_t) +#define DRM_IOCTL_FINISH DRM_IOW( 0x2c, drm_lock_t) + +#define DRM_IOCTL_AGP_ACQUIRE DRM_IO( 0x30) +#define DRM_IOCTL_AGP_RELEASE DRM_IO( 0x31) +#define DRM_IOCTL_AGP_ENABLE DRM_IOW( 0x32, drm_agp_mode_t) +#define DRM_IOCTL_AGP_INFO DRM_IOR( 0x33, drm_agp_info_t) +#define DRM_IOCTL_AGP_ALLOC DRM_IOWR(0x34, drm_agp_buffer_t) +#define DRM_IOCTL_AGP_FREE DRM_IOW( 0x35, drm_agp_buffer_t) +#define DRM_IOCTL_AGP_BIND DRM_IOW( 0x36, drm_agp_binding_t) +#define DRM_IOCTL_AGP_UNBIND DRM_IOW( 0x37, drm_agp_binding_t) /* Mga specific ioctls */ -#define DRM_IOCTL_MGA_INIT DRM_IOW( 0x40, drm_mga_init_t) -#define DRM_IOCTL_MGA_SWAP DRM_IOW( 0x41, drm_mga_swap_t) -#define DRM_IOCTL_MGA_CLEAR DRM_IOW( 0x42, drm_mga_clear_t) -#define DRM_IOCTL_MGA_ILOAD DRM_IOW( 0x43, drm_mga_iload_t) -#define DRM_IOCTL_MGA_VERTEX DRM_IOW( 0x44, drm_mga_vertex_t) -#define DRM_IOCTL_MGA_FLUSH DRM_IOW( 0x45, drm_lock_t ) -#define DRM_IOCTL_MGA_INDICES DRM_IOW( 0x46, drm_mga_indices_t) -#define DRM_IOCTL_MGA_BLIT DRM_IOW( 0x47, drm_mga_blit_t) +#define DRM_IOCTL_MGA_INIT DRM_IOW( 0x40, drm_mga_init_t) +#define DRM_IOCTL_MGA_SWAP DRM_IOW( 0x41, drm_mga_swap_t) +#define DRM_IOCTL_MGA_CLEAR DRM_IOW( 0x42, drm_mga_clear_t) +#define DRM_IOCTL_MGA_ILOAD DRM_IOW( 0x43, drm_mga_iload_t) +#define DRM_IOCTL_MGA_VERTEX DRM_IOW( 0x44, drm_mga_vertex_t) +#define DRM_IOCTL_MGA_FLUSH DRM_IOW( 0x45, drm_lock_t ) +#define DRM_IOCTL_MGA_INDICES DRM_IOW( 0x46, drm_mga_indices_t) +#define DRM_IOCTL_MGA_BLIT DRM_IOW( 0x47, drm_mga_blit_t) /* I810 specific ioctls */ -#define DRM_IOCTL_I810_INIT DRM_IOW( 0x40, drm_i810_init_t) -#define DRM_IOCTL_I810_VERTEX DRM_IOW( 0x41, drm_i810_vertex_t) -#define DRM_IOCTL_I810_CLEAR DRM_IOW( 0x42, drm_i810_clear_t) -#define DRM_IOCTL_I810_FLUSH DRM_IO ( 0x43) -#define DRM_IOCTL_I810_GETAGE DRM_IO ( 0x44) -#define DRM_IOCTL_I810_GETBUF DRM_IOWR(0x45, drm_i810_dma_t) -#define DRM_IOCTL_I810_SWAP DRM_IO ( 0x46) -#define DRM_IOCTL_I810_COPY DRM_IOW( 0x47, drm_i810_copy_t) -#define DRM_IOCTL_I810_DOCOPY DRM_IO ( 0x48) +#define DRM_IOCTL_I810_INIT DRM_IOW( 0x40, drm_i810_init_t) +#define DRM_IOCTL_I810_VERTEX DRM_IOW( 0x41, drm_i810_vertex_t) +#define DRM_IOCTL_I810_CLEAR DRM_IOW( 0x42, drm_i810_clear_t) +#define DRM_IOCTL_I810_FLUSH DRM_IO( 0x43) +#define DRM_IOCTL_I810_GETAGE DRM_IO( 0x44) +#define DRM_IOCTL_I810_GETBUF DRM_IOWR(0x45, drm_i810_dma_t) +#define DRM_IOCTL_I810_SWAP DRM_IO( 0x46) +#define DRM_IOCTL_I810_COPY DRM_IOW( 0x47, drm_i810_copy_t) +#define DRM_IOCTL_I810_DOCOPY DRM_IO( 0x48) /* Rage 128 specific ioctls */ -#define DRM_IOCTL_R128_INIT DRM_IOW( 0x40, drm_r128_init_t) -#define DRM_IOCTL_R128_CCE_START DRM_IO( 0x41) -#define DRM_IOCTL_R128_CCE_STOP DRM_IOW( 0x42, drm_r128_cce_stop_t) -#define DRM_IOCTL_R128_CCE_RESET DRM_IO( 0x43) -#define DRM_IOCTL_R128_CCE_IDLE DRM_IO( 0x44) -#define DRM_IOCTL_R128_RESET DRM_IO( 0x46) -#define DRM_IOCTL_R128_SWAP DRM_IO( 0x47) -#define DRM_IOCTL_R128_CLEAR DRM_IOW( 0x48, drm_r128_clear_t) -#define DRM_IOCTL_R128_VERTEX DRM_IOW( 0x49, drm_r128_vertex_t) -#define DRM_IOCTL_R128_INDICES DRM_IOW( 0x4a, drm_r128_indices_t) -#define DRM_IOCTL_R128_BLIT DRM_IOW( 0x4b, drm_r128_blit_t) -#define DRM_IOCTL_R128_DEPTH DRM_IOW( 0x4c, drm_r128_depth_t) -#define DRM_IOCTL_R128_STIPPLE DRM_IOW( 0x4d, drm_r128_stipple_t) -#define DRM_IOCTL_R128_PACKET DRM_IOWR(0x4e, drm_r128_packet_t) +#define DRM_IOCTL_R128_INIT DRM_IOW( 0x40, drm_r128_init_t) +#define DRM_IOCTL_R128_CCE_START DRM_IO( 0x41) +#define DRM_IOCTL_R128_CCE_STOP DRM_IOW( 0x42, drm_r128_cce_stop_t) +#define DRM_IOCTL_R128_CCE_RESET DRM_IO( 0x43) +#define DRM_IOCTL_R128_CCE_IDLE DRM_IO( 0x44) +#define DRM_IOCTL_R128_RESET DRM_IO( 0x46) +#define DRM_IOCTL_R128_SWAP DRM_IO( 0x47) +#define DRM_IOCTL_R128_CLEAR DRM_IOW( 0x48, drm_r128_clear_t) +#define DRM_IOCTL_R128_VERTEX DRM_IOW( 0x49, drm_r128_vertex_t) +#define DRM_IOCTL_R128_INDICES DRM_IOW( 0x4a, drm_r128_indices_t) +#define DRM_IOCTL_R128_BLIT DRM_IOW( 0x4b, drm_r128_blit_t) +#define DRM_IOCTL_R128_DEPTH DRM_IOW( 0x4c, drm_r128_depth_t) +#define DRM_IOCTL_R128_STIPPLE DRM_IOW( 0x4d, drm_r128_stipple_t) +#define DRM_IOCTL_R128_PACKET DRM_IOWR(0x4e, drm_r128_packet_t) + +/* Radeon specific ioctls */ +#define DRM_IOCTL_RADEON_CP_INIT DRM_IOW( 0x40, drm_radeon_init_t) +#define DRM_IOCTL_RADEON_CP_START DRM_IO( 0x41) +#define DRM_IOCTL_RADEON_CP_STOP DRM_IOW( 0x42, drm_radeon_cp_stop_t) +#define DRM_IOCTL_RADEON_CP_RESET DRM_IO( 0x43) +#define DRM_IOCTL_RADEON_CP_IDLE DRM_IO( 0x44) +#define DRM_IOCTL_RADEON_RESET DRM_IO( 0x45) +#define DRM_IOCTL_RADEON_FULLSCREEN DRM_IOW( 0x46, drm_radeon_fullscreen_t) +#define DRM_IOCTL_RADEON_SWAP DRM_IO( 0x47) +#define DRM_IOCTL_RADEON_CLEAR DRM_IOW( 0x48, drm_radeon_clear_t) +#define DRM_IOCTL_RADEON_VERTEX DRM_IOW( 0x49, drm_radeon_vertex_t) +#define DRM_IOCTL_RADEON_INDICES DRM_IOW( 0x4a, drm_radeon_indices_t) +#define DRM_IOCTL_RADEON_BLIT DRM_IOW( 0x4b, drm_radeon_blit_t) +#define DRM_IOCTL_RADEON_STIPPLE DRM_IOW( 0x4c, drm_radeon_stipple_t) +#define DRM_IOCTL_RADEON_INDIRECT DRM_IOWR(0x4d, drm_radeon_indirect_t) #ifdef CONFIG_DRM_SIS /* SiS specific ioctls */ -#define SIS_IOCTL_FB_ALLOC DRM_IOWR( 0x44, drm_sis_mem_t) -#define SIS_IOCTL_FB_FREE DRM_IOW( 0x45, drm_sis_mem_t) -#define SIS_IOCTL_AGP_INIT DRM_IOWR( 0x53, drm_sis_agp_t) -#define SIS_IOCTL_AGP_ALLOC DRM_IOWR( 0x54, drm_sis_mem_t) -#define SIS_IOCTL_AGP_FREE DRM_IOW( 0x55, drm_sis_mem_t) -#define SIS_IOCTL_FLIP DRM_IOW( 0x48, drm_sis_flip_t) -#define SIS_IOCTL_FLIP_INIT DRM_IO( 0x49) -#define SIS_IOCTL_FLIP_FINAL DRM_IO( 0x50) +#define SIS_IOCTL_FB_ALLOC DRM_IOWR(0x44, drm_sis_mem_t) +#define SIS_IOCTL_FB_FREE DRM_IOW( 0x45, drm_sis_mem_t) +#define SIS_IOCTL_AGP_INIT DRM_IOWR(0x53, drm_sis_agp_t) +#define SIS_IOCTL_AGP_ALLOC DRM_IOWR(0x54, drm_sis_mem_t) +#define SIS_IOCTL_AGP_FREE DRM_IOW( 0x55, drm_sis_mem_t) +#define SIS_IOCTL_FLIP DRM_IOW( 0x48, drm_sis_flip_t) +#define SIS_IOCTL_FLIP_INIT DRM_IO( 0x49) +#define SIS_IOCTL_FLIP_FINAL DRM_IO( 0x50) #endif #endif diff --git a/drivers/char/drm/radeon_bufs.c b/drivers/char/drm/radeon_bufs.c new file mode 100644 index 000000000..9a3093eb1 --- /dev/null +++ b/drivers/char/drm/radeon_bufs.c @@ -0,0 +1,298 @@ +/* radeon_bufs.c -- IOCTLs to manage buffers -*- linux-c -*- + * + * Copyright 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Fremont, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: Kevin E. Martin <martin@valinux.com> + * Rickard E. (Rik) Faith <faith@valinux.com> + * Jeff Hartmann <jhartmann@valinux.com> + * + */ + +#define __NO_VERSION__ +#include <linux/config.h> +#include "drmP.h" +#include "radeon_drv.h" +#include "linux/un.h" + + +#if defined(CONFIG_AGP) || defined(CONFIG_AGP_MODULE) +int radeon_addbufs_agp(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_device_dma_t *dma = dev->dma; + drm_buf_desc_t request; + drm_buf_entry_t *entry; + drm_buf_t *buf; + unsigned long offset; + unsigned long agp_offset; + int count; + int order; + int size; + int alignment; + int page_order; + int total; + int byte_count; + int i; + + if (!dma) return -EINVAL; + + if (copy_from_user(&request, (drm_buf_desc_t *)arg, sizeof(request))) + return -EFAULT; + + count = request.count; + order = drm_order(request.size); + size = 1 << order; + + alignment = (request.flags & _DRM_PAGE_ALIGN) ? PAGE_ALIGN(size):size; + page_order = order - PAGE_SHIFT > 0 ? order - PAGE_SHIFT : 0; + total = PAGE_SIZE << page_order; + + byte_count = 0; + agp_offset = dev->agp->base + request.agp_start; + + DRM_DEBUG("count: %d\n", count); + DRM_DEBUG("order: %d\n", order); + DRM_DEBUG("size: %d\n", size); + DRM_DEBUG("agp_offset: %ld\n", agp_offset); + DRM_DEBUG("alignment: %d\n", alignment); + DRM_DEBUG("page_order: %d\n", page_order); + DRM_DEBUG("total: %d\n", total); + + if (order < DRM_MIN_ORDER || order > DRM_MAX_ORDER) return -EINVAL; + if (dev->queue_count) return -EBUSY; /* Not while in use */ + + spin_lock(&dev->count_lock); + if (dev->buf_use) { + spin_unlock(&dev->count_lock); + return -EBUSY; + } + atomic_inc(&dev->buf_alloc); + spin_unlock(&dev->count_lock); + + down(&dev->struct_sem); + entry = &dma->bufs[order]; + if (entry->buf_count) { + up(&dev->struct_sem); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; /* May only call once for each order */ + } + + entry->buflist = drm_alloc(count * sizeof(*entry->buflist), + DRM_MEM_BUFS); + if (!entry->buflist) { + up(&dev->struct_sem); + atomic_dec(&dev->buf_alloc); + return -ENOMEM; + } + memset(entry->buflist, 0, count * sizeof(*entry->buflist)); + + entry->buf_size = size; + entry->page_order = page_order; + offset = 0; + + for (offset = 0; + entry->buf_count < count; + offset += alignment, ++entry->buf_count) { + buf = &entry->buflist[entry->buf_count]; + buf->idx = dma->buf_count + entry->buf_count; + buf->total = alignment; + buf->order = order; + buf->used = 0; + buf->offset = (dma->byte_count + offset); + buf->address = (void *)(agp_offset + offset); + buf->next = NULL; + buf->waiting = 0; + buf->pending = 0; + init_waitqueue_head(&buf->dma_wait); + buf->pid = 0; + + buf->dev_priv_size = sizeof(drm_radeon_buf_priv_t); + buf->dev_private = drm_alloc(sizeof(drm_radeon_buf_priv_t), + DRM_MEM_BUFS); + memset(buf->dev_private, 0, buf->dev_priv_size); + +#if DRM_DMA_HISTOGRAM + buf->time_queued = 0; + buf->time_dispatched = 0; + buf->time_completed = 0; + buf->time_freed = 0; +#endif + + byte_count += PAGE_SIZE << page_order; + + DRM_DEBUG("buffer %d @ %p\n", + entry->buf_count, buf->address); + } + + DRM_DEBUG("byte_count: %d\n", byte_count); + + dma->buflist = drm_realloc(dma->buflist, + dma->buf_count * sizeof(*dma->buflist), + (dma->buf_count + entry->buf_count) + * sizeof(*dma->buflist), + DRM_MEM_BUFS); + for (i = dma->buf_count; i < dma->buf_count + entry->buf_count; i++) + dma->buflist[i] = &entry->buflist[i - dma->buf_count]; + + dma->buf_count += entry->buf_count; + dma->byte_count += byte_count; + + drm_freelist_create(&entry->freelist, entry->buf_count); + for (i = 0; i < entry->buf_count; i++) { + drm_freelist_put(dev, &entry->freelist, &entry->buflist[i]); + } + + up(&dev->struct_sem); + + request.count = entry->buf_count; + request.size = size; + + if (copy_to_user((drm_buf_desc_t *)arg, &request, sizeof(request))) + return -EFAULT; + + dma->flags = _DRM_DMA_USE_AGP; + + atomic_dec(&dev->buf_alloc); + return 0; +} +#endif + +int radeon_addbufs(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_buf_desc_t request; + + if (!dev_priv || dev_priv->is_pci) return -EINVAL; + + if (copy_from_user(&request, (drm_buf_desc_t *)arg, sizeof(request))) + return -EFAULT; + +#if defined(CONFIG_AGP) || defined(CONFIG_AGP_MODULE) + if (request.flags & _DRM_AGP_BUFFER) + return radeon_addbufs_agp(inode, filp, cmd, arg); + else +#endif + return -EINVAL; +} + +int radeon_mapbufs(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_device_dma_t *dma = dev->dma; + int retcode = 0; + const int zero = 0; + unsigned long virtual; + unsigned long address; + drm_buf_map_t request; + int i; + + if (!dma || !dev_priv || dev_priv->is_pci) return -EINVAL; + + DRM_DEBUG("\n"); + + spin_lock(&dev->count_lock); + if (atomic_read(&dev->buf_alloc)) { + spin_unlock(&dev->count_lock); + return -EBUSY; + } + ++dev->buf_use; /* Can't allocate more after this call */ + spin_unlock(&dev->count_lock); + + if (copy_from_user(&request, (drm_buf_map_t *)arg, sizeof(request))) + return -EFAULT; + + if (request.count >= dma->buf_count) { + if (dma->flags & _DRM_DMA_USE_AGP) { + drm_map_t *map; + + map = dev_priv->buffers; + if (!map) { + retcode = -EINVAL; + goto done; + } + + down(¤t->mm->mmap_sem); + virtual = do_mmap(filp, 0, map->size, + PROT_READ|PROT_WRITE, + MAP_SHARED, + (unsigned long)map->offset); + up(¤t->mm->mmap_sem); + } else { + down(¤t->mm->mmap_sem); + virtual = do_mmap(filp, 0, dma->byte_count, + PROT_READ|PROT_WRITE, MAP_SHARED, 0); + up(¤t->mm->mmap_sem); + } + if (virtual > -1024UL) { + /* Real error */ + retcode = (signed long)virtual; + goto done; + } + request.virtual = (void *)virtual; + + for (i = 0; i < dma->buf_count; i++) { + if (copy_to_user(&request.list[i].idx, + &dma->buflist[i]->idx, + sizeof(request.list[0].idx))) { + retcode = -EFAULT; + goto done; + } + if (copy_to_user(&request.list[i].total, + &dma->buflist[i]->total, + sizeof(request.list[0].total))) { + retcode = -EFAULT; + goto done; + } + if (copy_to_user(&request.list[i].used, + &zero, + sizeof(zero))) { + retcode = -EFAULT; + goto done; + } + address = virtual + dma->buflist[i]->offset; + if (copy_to_user(&request.list[i].address, + &address, + sizeof(address))) { + retcode = -EFAULT; + goto done; + } + } + } + done: + request.count = dma->buf_count; + DRM_DEBUG("%d buffers, retcode = %d\n", request.count, retcode); + + if (copy_to_user((drm_buf_map_t *)arg, &request, sizeof(request))) + return -EFAULT; + + return retcode; +} diff --git a/drivers/char/drm/radeon_context.c b/drivers/char/drm/radeon_context.c new file mode 100644 index 000000000..e428dc22c --- /dev/null +++ b/drivers/char/drm/radeon_context.c @@ -0,0 +1,215 @@ +/* radeon_context.c -- IOCTLs for Radeon contexts -*- linux-c -*- + * + * Copyright 1999, 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Fremont, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Author: Kevin E. Martin <martin@valinux.com> + * Rickard E. (Rik) Faith <faith@valinux.com> + * + */ + +#define __NO_VERSION__ +#include "drmP.h" +#include "radeon_drv.h" + +extern drm_ctx_t radeon_res_ctx; + +static int radeon_alloc_queue(drm_device_t *dev) +{ + return drm_ctxbitmap_next(dev); +} + +int radeon_context_switch(drm_device_t *dev, int old, int new) +{ + char buf[64]; + + atomic_inc(&dev->total_ctx); + + if (test_and_set_bit(0, &dev->context_flag)) { + DRM_ERROR("Reentering -- FIXME\n"); + return -EBUSY; + } + +#if DRM_DMA_HISTOGRAM + dev->ctx_start = get_cycles(); +#endif + + DRM_DEBUG("Context switch from %d to %d\n", old, new); + + if (new == dev->last_context) { + clear_bit(0, &dev->context_flag); + return 0; + } + + if (drm_flags & DRM_FLAG_NOCTX) { + radeon_context_switch_complete(dev, new); + } else { + sprintf(buf, "C %d %d\n", old, new); + drm_write_string(dev, buf); + } + + return 0; +} + +int radeon_context_switch_complete(drm_device_t *dev, int new) +{ + dev->last_context = new; /* PRE/POST: This is the _only_ writer. */ + dev->last_switch = jiffies; + + if (!_DRM_LOCK_IS_HELD(dev->lock.hw_lock->lock)) { + DRM_ERROR("Lock isn't held after context switch\n"); + } + + /* If a context switch is ever initiated + when the kernel holds the lock, release + that lock here. */ +#if DRM_DMA_HISTOGRAM + atomic_inc(&dev->histo.ctx[drm_histogram_slot(get_cycles() + - dev->ctx_start)]); + +#endif + clear_bit(0, &dev->context_flag); + wake_up(&dev->context_wait); + + return 0; +} + + +int radeon_resctx(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_ctx_res_t res; + drm_ctx_t ctx; + int i; + + DRM_DEBUG("%d\n", DRM_RESERVED_CONTEXTS); + if (copy_from_user(&res, (drm_ctx_res_t *)arg, sizeof(res))) + return -EFAULT; + if (res.count >= DRM_RESERVED_CONTEXTS) { + memset(&ctx, 0, sizeof(ctx)); + for (i = 0; i < DRM_RESERVED_CONTEXTS; i++) { + ctx.handle = i; + if (copy_to_user(&res.contexts[i], &i, sizeof(i))) + return -EFAULT; + } + } + res.count = DRM_RESERVED_CONTEXTS; + if (copy_to_user((drm_ctx_res_t *)arg, &res, sizeof(res))) + return -EFAULT; + return 0; +} + + +int radeon_addctx(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_ctx_t ctx; + + if (copy_from_user(&ctx, (drm_ctx_t *)arg, sizeof(ctx))) + return -EFAULT; + if ((ctx.handle = radeon_alloc_queue(dev)) == DRM_KERNEL_CONTEXT) { + /* Skip kernel's context and get a new one. */ + ctx.handle = radeon_alloc_queue(dev); + } + DRM_DEBUG("%d\n", ctx.handle); + if (ctx.handle == -1) { + DRM_DEBUG("Not enough free contexts.\n"); + /* Should this return -EBUSY instead? */ + return -ENOMEM; + } + + if (copy_to_user((drm_ctx_t *)arg, &ctx, sizeof(ctx))) + return -EFAULT; + return 0; +} + +int radeon_modctx(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_ctx_t ctx; + + if (copy_from_user(&ctx, (drm_ctx_t*)arg, sizeof(ctx))) + return -EFAULT; + if (ctx.flags==_DRM_CONTEXT_PRESERVED) + radeon_res_ctx.handle=ctx.handle; + return 0; +} + +int radeon_getctx(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_ctx_t ctx; + + if (copy_from_user(&ctx, (drm_ctx_t*)arg, sizeof(ctx))) + return -EFAULT; + /* This is 0, because we don't hanlde any context flags */ + ctx.flags = 0; + if (copy_to_user((drm_ctx_t*)arg, &ctx, sizeof(ctx))) + return -EFAULT; + return 0; +} + +int radeon_switchctx(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_ctx_t ctx; + + if (copy_from_user(&ctx, (drm_ctx_t *)arg, sizeof(ctx))) + return -EFAULT; + DRM_DEBUG("%d\n", ctx.handle); + return radeon_context_switch(dev, dev->last_context, ctx.handle); +} + +int radeon_newctx(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_ctx_t ctx; + + if (copy_from_user(&ctx, (drm_ctx_t *)arg, sizeof(ctx))) + return -EFAULT; + DRM_DEBUG("%d\n", ctx.handle); + radeon_context_switch_complete(dev, ctx.handle); + + return 0; +} + +int radeon_rmctx(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_ctx_t ctx; + + if (copy_from_user(&ctx, (drm_ctx_t *)arg, sizeof(ctx))) + return -EFAULT; + DRM_DEBUG("%d\n", ctx.handle); + drm_ctxbitmap_free(dev, ctx.handle); + + return 0; +} diff --git a/drivers/char/drm/radeon_cp.c b/drivers/char/drm/radeon_cp.c new file mode 100644 index 000000000..5d662bc08 --- /dev/null +++ b/drivers/char/drm/radeon_cp.c @@ -0,0 +1,1314 @@ +/* radeon_cp.c -- CP support for Radeon -*- linux-c -*- + * + * Copyright 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Fremont, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Kevin E. Martin <martin@valinux.com> + * Gareth Hughes <gareth@valinux.com> + * + */ + +#define __NO_VERSION__ +#include "drmP.h" +#include "radeon_drv.h" + +#include <linux/interrupt.h> /* For task queue support */ +#include <linux/delay.h> + +#define RADEON_FIFO_DEBUG 0 + + +/* CP microcode (from ATI) */ +static u32 radeon_cp_microcode[][2] = { + { 0x21007000, 0000000000 }, + { 0x20007000, 0000000000 }, + { 0x000000b4, 0x00000004 }, + { 0x000000b8, 0x00000004 }, + { 0x6f5b4d4c, 0000000000 }, + { 0x4c4c427f, 0000000000 }, + { 0x5b568a92, 0000000000 }, + { 0x4ca09c6d, 0000000000 }, + { 0xad4c4c4c, 0000000000 }, + { 0x4ce1af3d, 0000000000 }, + { 0xd8afafaf, 0000000000 }, + { 0xd64c4cdc, 0000000000 }, + { 0x4cd10d10, 0000000000 }, + { 0x000f0000, 0x00000016 }, + { 0x362f242d, 0000000000 }, + { 0x00000012, 0x00000004 }, + { 0x000f0000, 0x00000016 }, + { 0x362f282d, 0000000000 }, + { 0x000380e7, 0x00000002 }, + { 0x04002c97, 0x00000002 }, + { 0x000f0001, 0x00000016 }, + { 0x333a3730, 0000000000 }, + { 0x000077ef, 0x00000002 }, + { 0x00061000, 0x00000002 }, + { 0x00000021, 0x0000001a }, + { 0x00004000, 0x0000001e }, + { 0x00061000, 0x00000002 }, + { 0x00000021, 0x0000001a }, + { 0x00004000, 0x0000001e }, + { 0x00061000, 0x00000002 }, + { 0x00000021, 0x0000001a }, + { 0x00004000, 0x0000001e }, + { 0x00000017, 0x00000004 }, + { 0x0003802b, 0x00000002 }, + { 0x040067e0, 0x00000002 }, + { 0x00000017, 0x00000004 }, + { 0x000077e0, 0x00000002 }, + { 0x00065000, 0x00000002 }, + { 0x000037e1, 0x00000002 }, + { 0x040067e1, 0x00000006 }, + { 0x000077e0, 0x00000002 }, + { 0x000077e1, 0x00000002 }, + { 0x000077e1, 0x00000006 }, + { 0xffffffff, 0000000000 }, + { 0x10000000, 0000000000 }, + { 0x0003802b, 0x00000002 }, + { 0x040067e0, 0x00000006 }, + { 0x00007675, 0x00000002 }, + { 0x00007676, 0x00000002 }, + { 0x00007677, 0x00000002 }, + { 0x00007678, 0x00000006 }, + { 0x0003802c, 0x00000002 }, + { 0x04002676, 0x00000002 }, + { 0x00007677, 0x00000002 }, + { 0x00007678, 0x00000006 }, + { 0x0000002f, 0x00000018 }, + { 0x0000002f, 0x00000018 }, + { 0000000000, 0x00000006 }, + { 0x00000030, 0x00000018 }, + { 0x00000030, 0x00000018 }, + { 0000000000, 0x00000006 }, + { 0x01605000, 0x00000002 }, + { 0x00065000, 0x00000002 }, + { 0x00098000, 0x00000002 }, + { 0x00061000, 0x00000002 }, + { 0x64c0603e, 0x00000004 }, + { 0x000380e6, 0x00000002 }, + { 0x040025c5, 0x00000002 }, + { 0x00080000, 0x00000016 }, + { 0000000000, 0000000000 }, + { 0x0400251d, 0x00000002 }, + { 0x00007580, 0x00000002 }, + { 0x00067581, 0x00000002 }, + { 0x04002580, 0x00000002 }, + { 0x00067581, 0x00000002 }, + { 0x00000049, 0x00000004 }, + { 0x00005000, 0000000000 }, + { 0x000380e6, 0x00000002 }, + { 0x040025c5, 0x00000002 }, + { 0x00061000, 0x00000002 }, + { 0x0000750e, 0x00000002 }, + { 0x00019000, 0x00000002 }, + { 0x00011055, 0x00000014 }, + { 0x00000055, 0x00000012 }, + { 0x0400250f, 0x00000002 }, + { 0x0000504f, 0x00000004 }, + { 0x000380e6, 0x00000002 }, + { 0x040025c5, 0x00000002 }, + { 0x00007565, 0x00000002 }, + { 0x00007566, 0x00000002 }, + { 0x00000058, 0x00000004 }, + { 0x000380e6, 0x00000002 }, + { 0x040025c5, 0x00000002 }, + { 0x01e655b4, 0x00000002 }, + { 0x4401b0e4, 0x00000002 }, + { 0x01c110e4, 0x00000002 }, + { 0x26667066, 0x00000018 }, + { 0x040c2565, 0x00000002 }, + { 0x00000066, 0x00000018 }, + { 0x04002564, 0x00000002 }, + { 0x00007566, 0x00000002 }, + { 0x0000005d, 0x00000004 }, + { 0x00401069, 0x00000008 }, + { 0x00101000, 0x00000002 }, + { 0x000d80ff, 0x00000002 }, + { 0x0080006c, 0x00000008 }, + { 0x000f9000, 0x00000002 }, + { 0x000e00ff, 0x00000002 }, + { 0000000000, 0x00000006 }, + { 0x0000008f, 0x00000018 }, + { 0x0000005b, 0x00000004 }, + { 0x000380e6, 0x00000002 }, + { 0x040025c5, 0x00000002 }, + { 0x00007576, 0x00000002 }, + { 0x00065000, 0x00000002 }, + { 0x00009000, 0x00000002 }, + { 0x00041000, 0x00000002 }, + { 0x0c00350e, 0x00000002 }, + { 0x00049000, 0x00000002 }, + { 0x00051000, 0x00000002 }, + { 0x01e785f8, 0x00000002 }, + { 0x00200000, 0x00000002 }, + { 0x0060007e, 0x0000000c }, + { 0x00007563, 0x00000002 }, + { 0x006075f0, 0x00000021 }, + { 0x20007073, 0x00000004 }, + { 0x00005073, 0x00000004 }, + { 0x000380e6, 0x00000002 }, + { 0x040025c5, 0x00000002 }, + { 0x00007576, 0x00000002 }, + { 0x00007577, 0x00000002 }, + { 0x0000750e, 0x00000002 }, + { 0x0000750f, 0x00000002 }, + { 0x00a05000, 0x00000002 }, + { 0x00600083, 0x0000000c }, + { 0x006075f0, 0x00000021 }, + { 0x000075f8, 0x00000002 }, + { 0x00000083, 0x00000004 }, + { 0x000a750e, 0x00000002 }, + { 0x000380e6, 0x00000002 }, + { 0x040025c5, 0x00000002 }, + { 0x0020750f, 0x00000002 }, + { 0x00600086, 0x00000004 }, + { 0x00007570, 0x00000002 }, + { 0x00007571, 0x00000002 }, + { 0x00007572, 0x00000006 }, + { 0x000380e6, 0x00000002 }, + { 0x040025c5, 0x00000002 }, + { 0x00005000, 0x00000002 }, + { 0x00a05000, 0x00000002 }, + { 0x00007568, 0x00000002 }, + { 0x00061000, 0x00000002 }, + { 0x00000095, 0x0000000c }, + { 0x00058000, 0x00000002 }, + { 0x0c607562, 0x00000002 }, + { 0x00000097, 0x00000004 }, + { 0x000380e6, 0x00000002 }, + { 0x040025c5, 0x00000002 }, + { 0x00600096, 0x00000004 }, + { 0x400070e5, 0000000000 }, + { 0x000380e6, 0x00000002 }, + { 0x040025c5, 0x00000002 }, + { 0x000380e5, 0x00000002 }, + { 0x000000a8, 0x0000001c }, + { 0x000650aa, 0x00000018 }, + { 0x040025bb, 0x00000002 }, + { 0x000610ab, 0x00000018 }, + { 0x040075bc, 0000000000 }, + { 0x000075bb, 0x00000002 }, + { 0x000075bc, 0000000000 }, + { 0x00090000, 0x00000006 }, + { 0x00090000, 0x00000002 }, + { 0x000d8002, 0x00000006 }, + { 0x00007832, 0x00000002 }, + { 0x00005000, 0x00000002 }, + { 0x000380e7, 0x00000002 }, + { 0x04002c97, 0x00000002 }, + { 0x00007820, 0x00000002 }, + { 0x00007821, 0x00000002 }, + { 0x00007800, 0000000000 }, + { 0x01200000, 0x00000002 }, + { 0x20077000, 0x00000002 }, + { 0x01200000, 0x00000002 }, + { 0x20007000, 0x00000002 }, + { 0x00061000, 0x00000002 }, + { 0x0120751b, 0x00000002 }, + { 0x8040750a, 0x00000002 }, + { 0x8040750b, 0x00000002 }, + { 0x00110000, 0x00000002 }, + { 0x000380e5, 0x00000002 }, + { 0x000000c6, 0x0000001c }, + { 0x000610ab, 0x00000018 }, + { 0x844075bd, 0x00000002 }, + { 0x000610aa, 0x00000018 }, + { 0x840075bb, 0x00000002 }, + { 0x000610ab, 0x00000018 }, + { 0x844075bc, 0x00000002 }, + { 0x000000c9, 0x00000004 }, + { 0x804075bd, 0x00000002 }, + { 0x800075bb, 0x00000002 }, + { 0x804075bc, 0x00000002 }, + { 0x00108000, 0x00000002 }, + { 0x01400000, 0x00000002 }, + { 0x006000cd, 0x0000000c }, + { 0x20c07000, 0x00000020 }, + { 0x000000cf, 0x00000012 }, + { 0x00800000, 0x00000006 }, + { 0x0080751d, 0x00000006 }, + { 0000000000, 0000000000 }, + { 0x0000775c, 0x00000002 }, + { 0x00a05000, 0x00000002 }, + { 0x00661000, 0x00000002 }, + { 0x0460275d, 0x00000020 }, + { 0x00004000, 0000000000 }, + { 0x01e00830, 0x00000002 }, + { 0x21007000, 0000000000 }, + { 0x6464614d, 0000000000 }, + { 0x69687420, 0000000000 }, + { 0x00000073, 0000000000 }, + { 0000000000, 0000000000 }, + { 0x00005000, 0x00000002 }, + { 0x000380d0, 0x00000002 }, + { 0x040025e0, 0x00000002 }, + { 0x000075e1, 0000000000 }, + { 0x00000001, 0000000000 }, + { 0x000380e0, 0x00000002 }, + { 0x04002394, 0x00000002 }, + { 0x00005000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0x00000008, 0000000000 }, + { 0x00000004, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, +}; + + +#define DO_IOREMAP(_m) (_m)->handle = drm_ioremap((_m)->offset, (_m)->size) + +#define DO_IOREMAPFREE(_m) \ + do { \ + if ((_m)->handle && (_m)->size) \ + drm_ioremapfree((_m)->handle, (_m)->size); \ + } while (0) + +#define DO_FIND_MAP(_m, _o) \ + do { \ + int _i; \ + for (_i = 0; _i < dev->map_count; _i++) { \ + if (dev->maplist[_i]->offset == _o) { \ + _m = dev->maplist[_i]; \ + break; \ + } \ + } \ + } while (0) + + +int RADEON_READ_PLL(drm_device_t *dev, int addr) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + + RADEON_WRITE8(RADEON_CLOCK_CNTL_INDEX, addr & 0x1f); + return RADEON_READ(RADEON_CLOCK_CNTL_DATA); +} + +#if RADEON_FIFO_DEBUG +static void radeon_status( drm_radeon_private_t *dev_priv ) +{ + printk( "%s:\n", __FUNCTION__ ); + printk( "RBBM_STATUS = 0x%08x\n", + (unsigned int)RADEON_READ( RADEON_RBBM_STATUS ) ); + printk( "CP_RB_RTPR = 0x%08x\n", + (unsigned int)RADEON_READ( RADEON_CP_RB_RPTR ) ); + printk( "CP_RB_WTPR = 0x%08x\n", + (unsigned int)RADEON_READ( RADEON_CP_RB_WPTR ) ); +} +#endif + + +/* ================================================================ + * Engine, FIFO control + */ + +static int radeon_do_pixcache_flush( drm_radeon_private_t *dev_priv ) +{ + u32 tmp; + int i; + + tmp = RADEON_READ( RADEON_RB2D_DSTCACHE_CTLSTAT ); + tmp |= RADEON_RB2D_DC_FLUSH_ALL; + RADEON_WRITE( RADEON_RB2D_DSTCACHE_CTLSTAT, tmp ); + + for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) { + if ( !(RADEON_READ( RADEON_RB2D_DSTCACHE_CTLSTAT ) + & RADEON_RB2D_DC_BUSY) ) { + return 0; + } + udelay( 1 ); + } + +#if RADEON_FIFO_DEBUG + DRM_ERROR( "failed!\n" ); + radeon_status( dev_priv ); +#endif + return -EBUSY; +} + +static int radeon_do_wait_for_fifo( drm_radeon_private_t *dev_priv, + int entries ) +{ + int i; + + for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) { + int slots = ( RADEON_READ( RADEON_RBBM_STATUS ) + & RADEON_RBBM_FIFOCNT_MASK ); + if ( slots >= entries ) return 0; + udelay( 1 ); + } + +#if RADEON_FIFO_DEBUG + DRM_ERROR( "failed!\n" ); + radeon_status( dev_priv ); +#endif + return -EBUSY; +} + +static int radeon_do_wait_for_idle( drm_radeon_private_t *dev_priv ) +{ + int i, ret; + + ret = radeon_do_wait_for_fifo( dev_priv, 64 ); + if ( ret < 0 ) return ret; + + for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) { + if ( !(RADEON_READ( RADEON_RBBM_STATUS ) + & RADEON_RBBM_ACTIVE) ) { + radeon_do_pixcache_flush( dev_priv ); + return 0; + } + udelay( 1 ); + } + +#if RADEON_FIFO_DEBUG + DRM_ERROR( "failed!\n" ); + radeon_status( dev_priv ); +#endif + return -EBUSY; +} + + +/* ================================================================ + * CP control, initialization + */ + +/* Load the microcode for the CP */ +static void radeon_cp_load_microcode( drm_radeon_private_t *dev_priv ) +{ + int i; + + radeon_do_wait_for_idle( dev_priv ); + + RADEON_WRITE( RADEON_CP_ME_RAM_ADDR, 0 ); + for ( i = 0 ; i < 256 ; i++ ) { + RADEON_WRITE( RADEON_CP_ME_RAM_DATAH, + radeon_cp_microcode[i][1] ); + RADEON_WRITE( RADEON_CP_ME_RAM_DATAL, + radeon_cp_microcode[i][0] ); + } +} + +/* Flush any pending commands to the CP. This should only be used just + * prior to a wait for idle, as it informs the engine that the command + * stream is ending. + */ +static void radeon_do_cp_flush( drm_radeon_private_t *dev_priv ) +{ +#if 0 + u32 tmp; + + tmp = RADEON_READ( RADEON_CP_RB_WPTR ) | (1 << 31); + RADEON_WRITE( RADEON_CP_RB_WPTR, tmp ); +#endif +} + +/* Wait for the CP to go idle. + */ +int radeon_do_cp_idle( drm_radeon_private_t *dev_priv ) +{ + RING_LOCALS; + + BEGIN_RING( 6 ); + + RADEON_PURGE_CACHE(); + RADEON_PURGE_ZCACHE(); + RADEON_WAIT_UNTIL_IDLE(); + + ADVANCE_RING(); + + return radeon_do_wait_for_idle( dev_priv ); +} + +/* Start the Command Processor. + */ +static void radeon_do_cp_start( drm_radeon_private_t *dev_priv ) +{ + RING_LOCALS; + + radeon_do_wait_for_idle( dev_priv ); + + RADEON_WRITE( RADEON_CP_CSQ_CNTL, dev_priv->cp_mode ); + + dev_priv->cp_running = 1; + + BEGIN_RING( 6 ); + + RADEON_PURGE_CACHE(); + RADEON_PURGE_ZCACHE(); + RADEON_WAIT_UNTIL_IDLE(); + + ADVANCE_RING(); +} + +/* Reset the Command Processor. This will not flush any pending + * commands, so you must wait for the CP command stream to complete + * before calling this routine. + */ +static void radeon_do_cp_reset( drm_radeon_private_t *dev_priv ) +{ + u32 cur_read_ptr; + + cur_read_ptr = RADEON_READ( RADEON_CP_RB_RPTR ); + RADEON_WRITE( RADEON_CP_RB_WPTR, cur_read_ptr ); + *dev_priv->ring.head = cur_read_ptr; + dev_priv->ring.tail = cur_read_ptr; +} + +/* Stop the Command Processor. This will not flush any pending + * commands, so you must flush the command stream and wait for the CP + * to go idle before calling this routine. + */ +static void radeon_do_cp_stop( drm_radeon_private_t *dev_priv ) +{ + RADEON_WRITE( RADEON_CP_CSQ_CNTL, RADEON_CSQ_PRIDIS_INDDIS ); + + dev_priv->cp_running = 0; +} + +/* Reset the engine. This will stop the CP if it is running. + */ +static int radeon_do_engine_reset( drm_device_t *dev ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + u32 clock_cntl_index, mclk_cntl, rbbm_soft_reset; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + radeon_do_pixcache_flush( dev_priv ); + + clock_cntl_index = RADEON_READ( RADEON_CLOCK_CNTL_INDEX ); + mclk_cntl = RADEON_READ_PLL( dev, RADEON_MCLK_CNTL ); + + /* FIXME: remove magic number here and in radeon ddx driver!!! */ + RADEON_WRITE_PLL( RADEON_MCLK_CNTL, mclk_cntl | 0x003f00000 ); + + rbbm_soft_reset = RADEON_READ( RADEON_RBBM_SOFT_RESET ); + + RADEON_WRITE( RADEON_RBBM_SOFT_RESET, ( rbbm_soft_reset | + RADEON_SOFT_RESET_CP | + RADEON_SOFT_RESET_HI | + RADEON_SOFT_RESET_SE | + RADEON_SOFT_RESET_RE | + RADEON_SOFT_RESET_PP | + RADEON_SOFT_RESET_E2 | + RADEON_SOFT_RESET_RB | + RADEON_SOFT_RESET_HDP ) ); + RADEON_READ( RADEON_RBBM_SOFT_RESET ); + RADEON_WRITE( RADEON_RBBM_SOFT_RESET, ( rbbm_soft_reset & + ~( RADEON_SOFT_RESET_CP | + RADEON_SOFT_RESET_HI | + RADEON_SOFT_RESET_SE | + RADEON_SOFT_RESET_RE | + RADEON_SOFT_RESET_PP | + RADEON_SOFT_RESET_E2 | + RADEON_SOFT_RESET_RB | + RADEON_SOFT_RESET_HDP ) ) ); + RADEON_READ( RADEON_RBBM_SOFT_RESET ); + + + RADEON_WRITE_PLL( RADEON_MCLK_CNTL, mclk_cntl ); + RADEON_WRITE( RADEON_CLOCK_CNTL_INDEX, clock_cntl_index ); + RADEON_WRITE( RADEON_RBBM_SOFT_RESET, rbbm_soft_reset ); + + /* Reset the CP ring */ + radeon_do_cp_reset( dev_priv ); + + /* The CP is no longer running after an engine reset */ + dev_priv->cp_running = 0; + + /* Reset any pending vertex, indirect buffers */ + radeon_freelist_reset( dev ); + + return 0; +} + +static void radeon_cp_init_ring_buffer( drm_device_t *dev ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + u32 ring_start, cur_read_ptr; + u32 tmp; + + /* Initialize the memory controller */ + RADEON_WRITE( RADEON_MC_FB_LOCATION, + (dev_priv->agp_vm_start - 1) & 0xffff0000 ); + RADEON_WRITE( RADEON_MC_AGP_LOCATION, + (((dev_priv->agp_vm_start - 1 + + dev_priv->agp_size) & 0xffff0000) | + (dev_priv->agp_vm_start >> 16)) ); + + ring_start = (dev_priv->cp_ring->offset + - dev->agp->base + + dev_priv->agp_vm_start); + + RADEON_WRITE( RADEON_CP_RB_BASE, ring_start ); + + /* Set the write pointer delay */ + RADEON_WRITE( RADEON_CP_RB_WPTR_DELAY, 0 ); + + /* Initialize the ring buffer's read and write pointers */ + cur_read_ptr = RADEON_READ( RADEON_CP_RB_RPTR ); + RADEON_WRITE( RADEON_CP_RB_WPTR, cur_read_ptr ); + *dev_priv->ring.head = cur_read_ptr; + dev_priv->ring.tail = cur_read_ptr; + + RADEON_WRITE( RADEON_CP_RB_RPTR_ADDR, dev_priv->ring_rptr->offset ); + + /* Set ring buffer size */ + RADEON_WRITE( RADEON_CP_RB_CNTL, dev_priv->ring.size_l2qw ); + + radeon_do_wait_for_idle( dev_priv ); + + /* Turn off PCI GART */ + tmp = RADEON_READ( RADEON_AIC_CNTL ) & ~RADEON_PCIGART_TRANSLATE_EN; + RADEON_WRITE( RADEON_AIC_CNTL, tmp ); + + /* Turn on bus mastering */ + tmp = RADEON_READ( RADEON_BUS_CNTL ) & ~RADEON_BUS_MASTER_DIS; + RADEON_WRITE( RADEON_BUS_CNTL, tmp ); + + /* Sync everything up */ + RADEON_WRITE( RADEON_ISYNC_CNTL, + (RADEON_ISYNC_ANY2D_IDLE3D | + RADEON_ISYNC_ANY3D_IDLE2D | + RADEON_ISYNC_WAIT_IDLEGUI | + RADEON_ISYNC_CPSCRATCH_IDLEGUI) ); +} + +static int radeon_do_init_cp( drm_device_t *dev, drm_radeon_init_t *init ) +{ + drm_radeon_private_t *dev_priv; + int i; + + dev_priv = drm_alloc( sizeof(drm_radeon_private_t), DRM_MEM_DRIVER ); + if ( dev_priv == NULL ) + return -ENOMEM; + dev->dev_private = (void *)dev_priv; + + memset( dev_priv, 0, sizeof(drm_radeon_private_t) ); + + dev_priv->is_pci = init->is_pci; + + /* We don't support PCI cards until PCI GART is implemented. + * Fail here so we can remove all checks for PCI cards around + * the CP ring code. + */ + if ( dev_priv->is_pci ) { + drm_free( dev_priv, sizeof(*dev_priv), DRM_MEM_DRIVER ); + dev->dev_private = NULL; + return -EINVAL; + } + + dev_priv->usec_timeout = init->usec_timeout; + if ( dev_priv->usec_timeout < 1 || + dev_priv->usec_timeout > RADEON_MAX_USEC_TIMEOUT ) { + drm_free( dev_priv, sizeof(*dev_priv), DRM_MEM_DRIVER ); + dev->dev_private = NULL; + return -EINVAL; + } + + dev_priv->cp_mode = init->cp_mode; + + /* Simple idle check. + */ + atomic_set( &dev_priv->idle_count, 0 ); + + /* We don't support anything other than bus-mastering ring mode, + * but the ring can be in either AGP or PCI space for the ring + * read pointer. + */ + if ( ( init->cp_mode != RADEON_CSQ_PRIBM_INDDIS ) && + ( init->cp_mode != RADEON_CSQ_PRIBM_INDBM ) ) { + drm_free( dev_priv, sizeof(*dev_priv), DRM_MEM_DRIVER ); + dev->dev_private = NULL; + return -EINVAL; + } + + switch ( init->fb_bpp ) { + case 16: + dev_priv->color_fmt = RADEON_COLOR_FORMAT_RGB565; + break; + case 32: + default: + dev_priv->color_fmt = RADEON_COLOR_FORMAT_ARGB8888; + break; + } + dev_priv->front_offset = init->front_offset; + dev_priv->front_pitch = init->front_pitch; + dev_priv->back_offset = init->back_offset; + dev_priv->back_pitch = init->back_pitch; + + switch ( init->depth_bpp ) { + case 16: + dev_priv->depth_fmt = RADEON_DEPTH_FORMAT_16BIT_INT_Z; + break; + case 32: + default: + dev_priv->depth_fmt = RADEON_DEPTH_FORMAT_24BIT_INT_Z; + break; + } + dev_priv->depth_offset = init->depth_offset; + dev_priv->depth_pitch = init->depth_pitch; + + dev_priv->front_pitch_offset = (((dev_priv->front_pitch/64) << 22) | + (dev_priv->front_offset >> 10)); + dev_priv->back_pitch_offset = (((dev_priv->back_pitch/64) << 22) | + (dev_priv->back_offset >> 10)); + dev_priv->depth_pitch_offset = (((dev_priv->depth_pitch/64) << 22) | + (dev_priv->depth_offset >> 10)); + + /* Hardware state for depth clears. Remove this if/when we no + * longer clear the depth buffer with a 3D rectangle. Hard-code + * all values to prevent unwanted 3D state from slipping through + * and screwing with the clear operation. + */ + dev_priv->depth_clear.rb3d_cntl = (RADEON_PLANE_MASK_ENABLE | + RADEON_Z_ENABLE | + (dev_priv->color_fmt << 10) | + RADEON_ZBLOCK16); + + dev_priv->depth_clear.rb3d_zstencilcntl = (dev_priv->depth_fmt | + RADEON_Z_TEST_ALWAYS | + RADEON_STENCIL_TEST_ALWAYS | + RADEON_STENCIL_S_FAIL_KEEP | + RADEON_STENCIL_ZPASS_KEEP | + RADEON_STENCIL_ZFAIL_KEEP | + RADEON_Z_WRITE_ENABLE); + + dev_priv->depth_clear.se_cntl = (RADEON_FFACE_CULL_CW | + RADEON_BFACE_SOLID | + RADEON_FFACE_SOLID | + RADEON_FLAT_SHADE_VTX_LAST | + + RADEON_DIFFUSE_SHADE_FLAT | + RADEON_ALPHA_SHADE_FLAT | + RADEON_SPECULAR_SHADE_FLAT | + RADEON_FOG_SHADE_FLAT | + + RADEON_VTX_PIX_CENTER_OGL | + RADEON_ROUND_MODE_TRUNC | + RADEON_ROUND_PREC_8TH_PIX); + + /* FIXME: We want multiple shared areas, including one shared + * only by the X Server and kernel module. + */ + for ( i = 0 ; i < dev->map_count ; i++ ) { + if ( dev->maplist[i]->type == _DRM_SHM ) { + dev_priv->sarea = dev->maplist[i]; + break; + } + } + + DO_FIND_MAP( dev_priv->fb, init->fb_offset ); + DO_FIND_MAP( dev_priv->mmio, init->mmio_offset ); + DO_FIND_MAP( dev_priv->cp_ring, init->ring_offset ); + DO_FIND_MAP( dev_priv->ring_rptr, init->ring_rptr_offset ); + DO_FIND_MAP( dev_priv->buffers, init->buffers_offset ); + + if ( !dev_priv->is_pci ) { + DO_FIND_MAP( dev_priv->agp_textures, + init->agp_textures_offset ); + } + + dev_priv->sarea_priv = + (drm_radeon_sarea_t *)((u8 *)dev_priv->sarea->handle + + init->sarea_priv_offset); + + DO_IOREMAP( dev_priv->cp_ring ); + DO_IOREMAP( dev_priv->ring_rptr ); + DO_IOREMAP( dev_priv->buffers ); +#if 0 + if ( !dev_priv->is_pci ) { + DO_IOREMAP( dev_priv->agp_textures ); + } +#endif + + dev_priv->agp_size = init->agp_size; + dev_priv->agp_vm_start = RADEON_READ( RADEON_CONFIG_APER_SIZE ); + dev_priv->agp_buffers_offset = (dev_priv->buffers->offset + - dev->agp->base + + dev_priv->agp_vm_start); + + dev_priv->ring.head = ((__volatile__ u32 *) + dev_priv->ring_rptr->handle); + + dev_priv->ring.start = (u32 *)dev_priv->cp_ring->handle; + dev_priv->ring.end = ((u32 *)dev_priv->cp_ring->handle + + init->ring_size / sizeof(u32)); + dev_priv->ring.size = init->ring_size; + dev_priv->ring.size_l2qw = drm_order( init->ring_size / 8 ); + + dev_priv->ring.tail_mask = + (dev_priv->ring.size / sizeof(u32)) - 1; + +#if 0 + /* Initialize the scratch register pointer. This will cause + * the scratch register values to be written out to memory + * whenever they are updated. + * FIXME: This doesn't quite work yet, so we're disabling it + * for the release. + */ + RADEON_WRITE( RADEON_SCRATCH_ADDR, (dev_priv->ring_rptr->offset + + RADEON_SCRATCH_REG_OFFSET) ); + RADEON_WRITE( RADEON_SCRATCH_UMSK, 0x7 ); +#endif + + dev_priv->scratch = ((__volatile__ u32 *) + dev_priv->ring_rptr->handle + + (RADEON_SCRATCH_REG_OFFSET / sizeof(u32))); + + dev_priv->sarea_priv->last_frame = 0; + RADEON_WRITE( RADEON_LAST_FRAME_REG, + dev_priv->sarea_priv->last_frame ); + + dev_priv->sarea_priv->last_dispatch = 0; + RADEON_WRITE( RADEON_LAST_DISPATCH_REG, + dev_priv->sarea_priv->last_dispatch ); + + dev_priv->sarea_priv->last_clear = 0; + RADEON_WRITE( RADEON_LAST_CLEAR_REG, + dev_priv->sarea_priv->last_clear ); + + radeon_cp_load_microcode( dev_priv ); + radeon_cp_init_ring_buffer( dev ); + radeon_do_engine_reset( dev ); + +#if ROTATE_BUFS + dev_priv->last_buf = 0; +#endif + + return 0; +} + +static int radeon_do_cleanup_cp( drm_device_t *dev ) +{ + if ( dev->dev_private ) { + drm_radeon_private_t *dev_priv = dev->dev_private; + + DO_IOREMAPFREE( dev_priv->cp_ring ); + DO_IOREMAPFREE( dev_priv->ring_rptr ); + DO_IOREMAPFREE( dev_priv->buffers ); +#if 0 + if ( !dev_priv->is_pci ) { + DO_IOREMAPFREE( dev_priv->agp_textures ); + } +#endif + + drm_free( dev->dev_private, sizeof(drm_radeon_private_t), + DRM_MEM_DRIVER ); + dev->dev_private = NULL; + } + + return 0; +} + +int radeon_cp_init( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_init_t init; + + if ( copy_from_user( &init, (drm_radeon_init_t *)arg, sizeof(init) ) ) + return -EFAULT; + + switch ( init.func ) { + case RADEON_INIT_CP: + return radeon_do_init_cp( dev, &init ); + case RADEON_CLEANUP_CP: + return radeon_do_cleanup_cp( dev ); + } + + return -EINVAL; +} + +int radeon_cp_start( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + if ( dev_priv->cp_running ) { + DRM_DEBUG( "%s while CP running\n", __FUNCTION__ ); + return 0; + } + if ( dev_priv->cp_mode == RADEON_CSQ_PRIDIS_INDDIS ) { + DRM_DEBUG( "%s called with bogus CP mode (%d)\n", + __FUNCTION__, dev_priv->cp_mode ); + return 0; + } + + radeon_do_cp_start( dev_priv ); + + return 0; +} + +/* Stop the CP. The engine must have been idled before calling this + * routine. + */ +int radeon_cp_stop( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_cp_stop_t stop; + int ret; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + + if ( copy_from_user( &stop, (drm_radeon_init_t *)arg, sizeof(stop) ) ) + return -EFAULT; + + /* Flush any pending CP commands. This ensures any outstanding + * commands are exectuted by the engine before we turn it off. + */ + if ( stop.flush ) { + radeon_do_cp_flush( dev_priv ); + } + + /* If we fail to make the engine go idle, we return an error + * code so that the DRM ioctl wrapper can try again. + */ + if ( stop.idle ) { + ret = radeon_do_cp_idle( dev_priv ); + if ( ret < 0 ) return ret; + } + + /* Finally, we can turn off the CP. If the engine isn't idle, + * we will get some dropped triangles as they won't be fully + * rendered before the CP is shut down. + */ + radeon_do_cp_stop( dev_priv ); + + /* Reset the engine */ + radeon_do_engine_reset( dev ); + + return 0; +} + +/* Just reset the CP ring. Called as part of an X Server engine reset. + */ +int radeon_cp_reset( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + if ( !dev_priv ) { + DRM_DEBUG( "%s called before init done\n", __FUNCTION__ ); + return -EINVAL; + } + + radeon_do_cp_reset( dev_priv ); + + /* The CP is no longer running after an engine reset */ + dev_priv->cp_running = 0; + + return 0; +} + +int radeon_cp_idle( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + + return radeon_do_cp_idle( dev_priv ); +} + +int radeon_engine_reset( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + + return radeon_do_engine_reset( dev ); +} + + +/* ================================================================ + * Fullscreen mode + */ + +static int radeon_do_init_pageflip( drm_device_t *dev ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + dev_priv->crtc_offset = RADEON_READ( RADEON_CRTC_OFFSET ); + dev_priv->crtc_offset_cntl = RADEON_READ( RADEON_CRTC_OFFSET_CNTL ); + + RADEON_WRITE( RADEON_CRTC_OFFSET, dev_priv->front_offset ); + RADEON_WRITE( RADEON_CRTC_OFFSET_CNTL, + dev_priv->crtc_offset_cntl | + RADEON_CRTC_OFFSET_FLIP_CNTL ); + + dev_priv->page_flipping = 1; + dev_priv->current_page = 0; + + return 0; +} + +int radeon_do_cleanup_pageflip( drm_device_t *dev ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + RADEON_WRITE( RADEON_CRTC_OFFSET, dev_priv->crtc_offset ); + RADEON_WRITE( RADEON_CRTC_OFFSET_CNTL, dev_priv->crtc_offset_cntl ); + + dev_priv->page_flipping = 0; + dev_priv->current_page = 0; + + return 0; +} + +int radeon_fullscreen( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_fullscreen_t fs; + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + + if ( copy_from_user( &fs, (drm_radeon_fullscreen_t *)arg, + sizeof(fs) ) ) + return -EFAULT; + + switch ( fs.func ) { + case RADEON_INIT_FULLSCREEN: + return radeon_do_init_pageflip( dev ); + case RADEON_CLEANUP_FULLSCREEN: + return radeon_do_cleanup_pageflip( dev ); + } + + return -EINVAL; +} + + +/* ================================================================ + * Freelist management + */ +#define RADEON_BUFFER_USED 0xffffffff +#define RADEON_BUFFER_FREE 0 + +#if 0 +static int radeon_freelist_init( drm_device_t *dev ) +{ + drm_device_dma_t *dma = dev->dma; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_buf_t *buf; + drm_radeon_buf_priv_t *buf_priv; + drm_radeon_freelist_t *entry; + int i; + + dev_priv->head = drm_alloc( sizeof(drm_radeon_freelist_t), + DRM_MEM_DRIVER ); + if ( dev_priv->head == NULL ) + return -ENOMEM; + + memset( dev_priv->head, 0, sizeof(drm_radeon_freelist_t) ); + dev_priv->head->age = RADEON_BUFFER_USED; + + for ( i = 0 ; i < dma->buf_count ; i++ ) { + buf = dma->buflist[i]; + buf_priv = buf->dev_private; + + entry = drm_alloc( sizeof(drm_radeon_freelist_t), + DRM_MEM_DRIVER ); + if ( !entry ) return -ENOMEM; + + entry->age = RADEON_BUFFER_FREE; + entry->buf = buf; + entry->prev = dev_priv->head; + entry->next = dev_priv->head->next; + if ( !entry->next ) + dev_priv->tail = entry; + + buf_priv->discard = 0; + buf_priv->dispatched = 0; + buf_priv->list_entry = entry; + + dev_priv->head->next = entry; + + if ( dev_priv->head->next ) + dev_priv->head->next->prev = entry; + } + + return 0; + +} +#endif + +drm_buf_t *radeon_freelist_get( drm_device_t *dev ) +{ + drm_device_dma_t *dma = dev->dma; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_buf_priv_t *buf_priv; + drm_buf_t *buf; + int i, t; +#if ROTATE_BUFS + int start; +#endif + + /* FIXME: Optimize -- use freelist code */ + + for ( i = 0 ; i < dma->buf_count ; i++ ) { + buf = dma->buflist[i]; + buf_priv = buf->dev_private; + if ( buf->pid == 0 ) { + DRM_DEBUG( " ret buf=%d last=%d pid=0\n", + buf->idx, dev_priv->last_buf ); + return buf; + } + DRM_DEBUG( " skipping buf=%d pid=%d\n", + buf->idx, buf->pid ); + } + +#if ROTATE_BUFS + if ( ++dev_priv->last_buf >= dma->buf_count ) + dev_priv->last_buf = 0; + start = dev_priv->last_buf; +#endif + for ( t = 0 ; t < dev_priv->usec_timeout ; t++ ) { +#if 0 + /* FIXME: Disable this for now */ + u32 done_age = dev_priv->scratch[RADEON_LAST_DISPATCH]; +#else + u32 done_age = RADEON_READ( RADEON_LAST_DISPATCH_REG ); +#endif +#if ROTATE_BUFS + for ( i = start ; i < dma->buf_count ; i++ ) { +#else + for ( i = 0 ; i < dma->buf_count ; i++ ) { +#endif + buf = dma->buflist[i]; + buf_priv = buf->dev_private; + if ( buf->pending && buf_priv->age <= done_age ) { + /* The buffer has been processed, so it + * can now be used. + */ + buf->pending = 0; + DRM_DEBUG( " ret buf=%d last=%d age=%d done=%d\n", buf->idx, dev_priv->last_buf, buf_priv->age, done_age ); + return buf; + } + DRM_DEBUG( " skipping buf=%d age=%d done=%d\n", + buf->idx, buf_priv->age, + done_age ); +#if ROTATE_BUFS + start = 0; +#endif + } + udelay( 1 ); + } + + DRM_ERROR( "returning NULL!\n" ); + return NULL; +} + +void radeon_freelist_reset( drm_device_t *dev ) +{ + drm_device_dma_t *dma = dev->dma; +#if ROTATE_BUFS + drm_radeon_private_t *dev_priv = dev->dev_private; +#endif + int i; + +#if ROTATE_BUFS + dev_priv->last_buf = 0; +#endif + for ( i = 0 ; i < dma->buf_count ; i++ ) { + drm_buf_t *buf = dma->buflist[i]; + drm_radeon_buf_priv_t *buf_priv = buf->dev_private; + buf_priv->age = 0; + } +} + + +/* ================================================================ + * CP command submission + */ + +int radeon_wait_ring( drm_radeon_private_t *dev_priv, int n ) +{ + drm_radeon_ring_buffer_t *ring = &dev_priv->ring; + int i; + + for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) { + ring->space = *ring->head - ring->tail; + if ( ring->space <= 0 ) + ring->space += ring->size; + + if ( ring->space >= n ) + return 0; + + udelay( 1 ); + } + + /* FIXME: This return value is ignored in the BEGIN_RING macro! */ + DRM_ERROR( "failed!\n" ); + return -EBUSY; +} + +void radeon_update_ring_snapshot( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_ring_buffer_t *ring = &dev_priv->ring; + + ring->space = *ring->head - ring->tail; + if ( ring->space == 0 ) + atomic_inc( &dev_priv->idle_count ); + if ( ring->space <= 0 ) + ring->space += ring->size; +} + +static int radeon_cp_get_buffers( drm_device_t *dev, drm_dma_t *d ) +{ + int i; + drm_buf_t *buf; + + for ( i = d->granted_count ; i < d->request_count ; i++ ) { + buf = radeon_freelist_get( dev ); + if ( !buf ) return -EAGAIN; + + buf->pid = current->pid; + + if ( copy_to_user( &d->request_indices[i], &buf->idx, + sizeof(buf->idx) ) ) + return -EFAULT; + if ( copy_to_user( &d->request_sizes[i], &buf->total, + sizeof(buf->total) ) ) + return -EFAULT; + + d->granted_count++; + } + return 0; +} + +int radeon_cp_buffers( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_device_dma_t *dma = dev->dma; + int ret = 0; + drm_dma_t d; + + if ( copy_from_user( &d, (drm_dma_t *) arg, sizeof(d) ) ) + return -EFAULT; + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + + /* Please don't send us buffers. + */ + if ( d.send_count != 0 ) { + DRM_ERROR( "Process %d trying to send %d buffers via drmDMA\n", + current->pid, d.send_count ); + return -EINVAL; + } + + /* We'll send you buffers. + */ + if ( d.request_count < 0 || d.request_count > dma->buf_count ) { + DRM_ERROR( "Process %d trying to get %d buffers (of %d max)\n", + current->pid, d.request_count, dma->buf_count ); + return -EINVAL; + } + + d.granted_count = 0; + + if ( d.request_count ) { + ret = radeon_cp_get_buffers( dev, &d ); + } + + if ( copy_to_user( (drm_dma_t *) arg, &d, sizeof(d) ) ) + return -EFAULT; + + return ret; +} diff --git a/drivers/char/drm/radeon_drm.h b/drivers/char/drm/radeon_drm.h new file mode 100644 index 000000000..c5f9f66d1 --- /dev/null +++ b/drivers/char/drm/radeon_drm.h @@ -0,0 +1,325 @@ +/* radeon_drm.h -- Public header for the radeon driver -*- linux-c -*- + * + * Copyright 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Fremont, California. + * All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Kevin E. Martin <martin@valinux.com> + * Gareth Hughes <gareth@valinux.com> + * + */ + +#ifndef __RADEON_DRM_H__ +#define __RADEON_DRM_H__ + +/* WARNING: If you change any of these defines, make sure to change the + * defines in the X server file (radeon_sarea.h) + */ +#ifndef __RADEON_SAREA_DEFINES__ +#define __RADEON_SAREA_DEFINES__ + +/* What needs to be changed for the current vertex buffer? + */ +#define RADEON_UPLOAD_CONTEXT 0x00000001 +#define RADEON_UPLOAD_VERTFMT 0x00000002 +#define RADEON_UPLOAD_LINE 0x00000004 +#define RADEON_UPLOAD_BUMPMAP 0x00000008 +#define RADEON_UPLOAD_MASKS 0x00000010 +#define RADEON_UPLOAD_VIEWPORT 0x00000020 +#define RADEON_UPLOAD_SETUP 0x00000040 +#define RADEON_UPLOAD_TCL 0x00000080 +#define RADEON_UPLOAD_MISC 0x00000100 +#define RADEON_UPLOAD_TEX0 0x00000200 +#define RADEON_UPLOAD_TEX1 0x00000400 +#define RADEON_UPLOAD_TEX2 0x00000800 +#define RADEON_UPLOAD_TEX0IMAGES 0x00001000 +#define RADEON_UPLOAD_TEX1IMAGES 0x00002000 +#define RADEON_UPLOAD_TEX2IMAGES 0x00004000 +#define RADEON_UPLOAD_CLIPRECTS 0x00008000 /* handled client-side */ +#define RADEON_REQUIRE_QUIESCENCE 0x00010000 +#define RADEON_UPLOAD_ALL 0x0001ffff + +#define RADEON_FRONT 0x1 +#define RADEON_BACK 0x2 +#define RADEON_DEPTH 0x4 + +/* Primitive types + */ +#define RADEON_POINTS 0x1 +#define RADEON_LINES 0x2 +#define RADEON_LINE_STRIP 0x3 +#define RADEON_TRIANGLES 0x4 +#define RADEON_TRIANGLE_FAN 0x5 +#define RADEON_TRIANGLE_STRIP 0x6 + +/* Vertex/indirect buffer size + */ +#define RADEON_BUFFER_SIZE 16384 + +/* Byte offsets for indirect buffer data + */ +#define RADEON_INDEX_PRIM_OFFSET 20 +#define RADEON_HOSTDATA_BLIT_OFFSET 32 + +#define RADEON_SCRATCH_REG_OFFSET 32 + +/* Keep these small for testing + */ +#define RADEON_NR_SAREA_CLIPRECTS 12 + +/* There are 2 heaps (local/AGP). Each region within a heap is a + * minimum of 64k, and there are at most 64 of them per heap. + */ +#define RADEON_LOCAL_TEX_HEAP 0 +#define RADEON_AGP_TEX_HEAP 1 +#define RADEON_NR_TEX_HEAPS 2 +#define RADEON_NR_TEX_REGIONS 64 +#define RADEON_LOG_TEX_GRANULARITY 16 + +#define RADEON_MAX_TEXTURE_LEVELS 11 +#define RADEON_MAX_TEXTURE_UNITS 3 + +#endif /* __RADEON_SAREA_DEFINES__ */ + +typedef struct { + unsigned int red; + unsigned int green; + unsigned int blue; + unsigned int alpha; +} radeon_color_regs_t; + +typedef struct { + /* Context state */ + unsigned int pp_misc; /* 0x1c14 */ + unsigned int pp_fog_color; + unsigned int re_solid_color; + unsigned int rb3d_blendcntl; + unsigned int rb3d_depthoffset; + unsigned int rb3d_depthpitch; + unsigned int rb3d_zstencilcntl; + + unsigned int pp_cntl; /* 0x1c38 */ + unsigned int rb3d_cntl; + unsigned int rb3d_coloroffset; + unsigned int re_width_height; + unsigned int rb3d_colorpitch; + unsigned int se_cntl; + + /* Vertex format state */ + unsigned int se_coord_fmt; /* 0x1c50 */ + + /* Line state */ + unsigned int re_line_pattern; /* 0x1cd0 */ + unsigned int re_line_state; + + unsigned int se_line_width; /* 0x1db8 */ + + /* Bumpmap state */ + unsigned int pp_lum_matrix; /* 0x1d00 */ + + unsigned int pp_rot_matrix_0; /* 0x1d58 */ + unsigned int pp_rot_matrix_1; + + /* Mask state */ + unsigned int rb3d_stencilrefmask; /* 0x1d7c */ + unsigned int rb3d_ropcntl; + unsigned int rb3d_planemask; + + /* Viewport state */ + unsigned int se_vport_xscale; /* 0x1d98 */ + unsigned int se_vport_xoffset; + unsigned int se_vport_yscale; + unsigned int se_vport_yoffset; + unsigned int se_vport_zscale; + unsigned int se_vport_zoffset; + + /* Setup state */ + unsigned int se_cntl_status; /* 0x2140 */ + +#ifdef TCL_ENABLE + /* TCL state */ + radeon_color_regs_t se_tcl_material_emmissive; /* 0x2210 */ + radeon_color_regs_t se_tcl_material_ambient; + radeon_color_regs_t se_tcl_material_diffuse; + radeon_color_regs_t se_tcl_material_specular; + unsigned int se_tcl_shininess; + unsigned int se_tcl_output_vtx_fmt; + unsigned int se_tcl_output_vtx_sel; + unsigned int se_tcl_matrix_select_0; + unsigned int se_tcl_matrix_select_1; + unsigned int se_tcl_ucp_vert_blend_ctl; + unsigned int se_tcl_texture_proc_ctl; + unsigned int se_tcl_light_model_ctl; + unsigned int se_tcl_per_light_ctl[4]; +#endif + + /* Misc state */ + unsigned int re_top_left; /* 0x26c0 */ + unsigned int re_misc; +} drm_radeon_context_regs_t; + +/* Setup registers for each texture unit + */ +typedef struct { + unsigned int pp_txfilter; + unsigned int pp_txformat; + unsigned int pp_txoffset; + unsigned int pp_txcblend; + unsigned int pp_txablend; + unsigned int pp_tfactor; + + unsigned int pp_border_color; + +#ifdef CUBIC_ENABLE + unsigned int pp_cubic_faces; + unsigned int pp_cubic_offset[5]; +#endif +} drm_radeon_texture_regs_t; + +typedef struct { + unsigned char next, prev; + unsigned char in_use; + int age; +} drm_radeon_tex_region_t; + +typedef struct { + /* The channel for communication of state information to the kernel + * on firing a vertex buffer. + */ + drm_radeon_context_regs_t context_state; + drm_radeon_texture_regs_t tex_state[RADEON_MAX_TEXTURE_UNITS]; + unsigned int dirty; + unsigned int vertsize; + unsigned int vc_format; + + /* The current cliprects, or a subset thereof. + */ + drm_clip_rect_t boxes[RADEON_NR_SAREA_CLIPRECTS]; + unsigned int nbox; + + /* Counters for client-side throttling of rendering clients. + */ + unsigned int last_frame; + unsigned int last_dispatch; + unsigned int last_clear; + + drm_radeon_tex_region_t tex_list[RADEON_NR_TEX_HEAPS][RADEON_NR_TEX_REGIONS+1]; + int tex_age[RADEON_NR_TEX_HEAPS]; + int ctx_owner; +} drm_radeon_sarea_t; + + +/* WARNING: If you change any of these defines, make sure to change the + * defines in the Xserver file (xf86drmRadeon.h) + */ +typedef struct drm_radeon_init { + enum { + RADEON_INIT_CP = 0x01, + RADEON_CLEANUP_CP = 0x02 + } func; + int sarea_priv_offset; + int is_pci; + int cp_mode; + int agp_size; + int ring_size; + int usec_timeout; + + unsigned int fb_bpp; + unsigned int front_offset, front_pitch; + unsigned int back_offset, back_pitch; + unsigned int depth_bpp; + unsigned int depth_offset, depth_pitch; + + unsigned int fb_offset; + unsigned int mmio_offset; + unsigned int ring_offset; + unsigned int ring_rptr_offset; + unsigned int buffers_offset; + unsigned int agp_textures_offset; +} drm_radeon_init_t; + +typedef struct drm_radeon_cp_stop { + int flush; + int idle; +} drm_radeon_cp_stop_t; + +typedef struct drm_radeon_fullscreen { + enum { + RADEON_INIT_FULLSCREEN = 0x01, + RADEON_CLEANUP_FULLSCREEN = 0x02 + } func; +} drm_radeon_fullscreen_t; + +#define CLEAR_X1 0 +#define CLEAR_Y1 1 +#define CLEAR_X2 2 +#define CLEAR_Y2 3 +#define CLEAR_DEPTH 4 + +typedef struct drm_radeon_clear { + unsigned int flags; + int x, y, w, h; + unsigned int clear_color; + unsigned int clear_depth; + union { + float f[5]; + unsigned int ui[5]; + } rect; +} drm_radeon_clear_t; + +typedef struct drm_radeon_vertex { + int prim; + int idx; /* Index of vertex buffer */ + int count; /* Number of vertices in buffer */ + int discard; /* Client finished with buffer? */ +} drm_radeon_vertex_t; + +typedef struct drm_radeon_indices { + int prim; + int idx; + int start; + int end; + int discard; /* Client finished with buffer? */ +} drm_radeon_indices_t; + +typedef struct drm_radeon_blit { + int idx; + int pitch; + int offset; + int format; + unsigned short x, y; + unsigned short width, height; +} drm_radeon_blit_t; + +typedef struct drm_radeon_stipple { + unsigned int *mask; +} drm_radeon_stipple_t; + +typedef struct drm_radeon_indirect { + int idx; + int start; + int end; + int discard; +} drm_radeon_indirect_t; + +#endif diff --git a/drivers/char/drm/radeon_drv.c b/drivers/char/drm/radeon_drv.c new file mode 100644 index 000000000..0113ed97c --- /dev/null +++ b/drivers/char/drm/radeon_drv.c @@ -0,0 +1,702 @@ +/* radeon_drv.c -- ATI Radeon driver -*- linux-c -*- + * + * Copyright 1999, 2000 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Fremont, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: Kevin E. Martin <martin@valinux.com> + * Rickard E. (Rik) Faith <faith@valinux.com> + * + */ + +#include <linux/config.h> +#include "drmP.h" +#include "radeon_drv.h" + +#define RADEON_NAME "radeon" +#define RADEON_DESC "ATI Radeon" +#define RADEON_DATE "20010105" +#define RADEON_MAJOR 1 +#define RADEON_MINOR 0 +#define RADEON_PATCHLEVEL 0 + +static drm_device_t radeon_device; +drm_ctx_t radeon_res_ctx; + +static struct file_operations radeon_fops = { +#if LINUX_VERSION_CODE >= 0x020400 + /* This started being used during 2.4.0-test */ + owner: THIS_MODULE, +#endif + open: radeon_open, + flush: drm_flush, + release: radeon_release, + ioctl: radeon_ioctl, + mmap: drm_mmap, + read: drm_read, + fasync: drm_fasync, + poll: drm_poll, +}; + +static struct miscdevice radeon_misc = { + minor: MISC_DYNAMIC_MINOR, + name: RADEON_NAME, + fops: &radeon_fops, +}; + +static drm_ioctl_desc_t radeon_ioctls[] = { + [DRM_IOCTL_NR(DRM_IOCTL_VERSION)] = { radeon_version, 0, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_GET_UNIQUE)] = { drm_getunique, 0, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_GET_MAGIC)] = { drm_getmagic, 0, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_IRQ_BUSID)] = { drm_irq_busid, 0, 1 }, + + [DRM_IOCTL_NR(DRM_IOCTL_SET_UNIQUE)] = { drm_setunique, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_BLOCK)] = { drm_block, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_UNBLOCK)] = { drm_unblock, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_AUTH_MAGIC)] = { drm_authmagic, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_ADD_MAP)] = { drm_addmap, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_ADD_BUFS)] = { radeon_addbufs, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_MARK_BUFS)] = { drm_markbufs, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_INFO_BUFS)] = { drm_infobufs, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_MAP_BUFS)] = { radeon_mapbufs, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_FREE_BUFS)] = { drm_freebufs, 1, 0 }, + + [DRM_IOCTL_NR(DRM_IOCTL_ADD_CTX)] = { radeon_addctx, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_RM_CTX)] = { radeon_rmctx, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_MOD_CTX)] = { radeon_modctx, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_GET_CTX)] = { radeon_getctx, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_SWITCH_CTX)] = { radeon_switchctx, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_NEW_CTX)] = { radeon_newctx, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_RES_CTX)] = { radeon_resctx, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_ADD_DRAW)] = { drm_adddraw, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_RM_DRAW)] = { drm_rmdraw, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_DMA)] = { radeon_cp_buffers, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_LOCK)] = { radeon_lock, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_UNLOCK)] = { radeon_unlock, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_FINISH)] = { drm_finish, 1, 0 }, + +#if defined(CONFIG_AGP) || defined(CONFIG_AGP_MODULE) + [DRM_IOCTL_NR(DRM_IOCTL_AGP_ACQUIRE)] = { drm_agp_acquire, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_RELEASE)] = { drm_agp_release, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_ENABLE)] = { drm_agp_enable, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_INFO)] = { drm_agp_info, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_ALLOC)] = { drm_agp_alloc, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_FREE)] = { drm_agp_free, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_BIND)] = { drm_agp_bind, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_AGP_UNBIND)] = { drm_agp_unbind, 1, 1 }, +#endif + + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_CP_INIT)] = { radeon_cp_init, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_CP_START)] = { radeon_cp_start, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_CP_STOP)] = { radeon_cp_stop, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_CP_RESET)] = { radeon_cp_reset, 1, 1 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_CP_IDLE)] = { radeon_cp_idle, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_RESET)] = { radeon_engine_reset, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_FULLSCREEN)] = { radeon_fullscreen, 1, 0 }, + + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_SWAP)] = { radeon_cp_swap, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_CLEAR)] = { radeon_cp_clear, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_VERTEX)] = { radeon_cp_vertex, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_INDICES)] = { radeon_cp_indices, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_BLIT)] = { radeon_cp_blit, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_STIPPLE)] = { radeon_cp_stipple, 1, 0 }, + [DRM_IOCTL_NR(DRM_IOCTL_RADEON_INDIRECT)]= { radeon_cp_indirect,1, 1 }, +}; +#define RADEON_IOCTL_COUNT DRM_ARRAY_SIZE(radeon_ioctls) + +#ifdef MODULE +static char *radeon = NULL; +#endif + +MODULE_AUTHOR("VA Linux Systems, Inc."); +MODULE_DESCRIPTION("radeon"); +MODULE_PARM(radeon, "s"); + +#ifndef MODULE +/* radeon_options is called by the kernel to parse command-line options + * passed via the boot-loader (e.g., LILO). It calls the insmod option + * routine, drm_parse_drm. + */ + +static int __init radeon_options(char *str) +{ + drm_parse_options(str); + return 1; +} + +__setup("radeon=", radeon_options); +#endif + +static int radeon_setup(drm_device_t *dev) +{ + int i; + + atomic_set(&dev->ioctl_count, 0); + atomic_set(&dev->vma_count, 0); + dev->buf_use = 0; + atomic_set(&dev->buf_alloc, 0); + + drm_dma_setup(dev); + + atomic_set(&dev->total_open, 0); + atomic_set(&dev->total_close, 0); + atomic_set(&dev->total_ioctl, 0); + atomic_set(&dev->total_irq, 0); + atomic_set(&dev->total_ctx, 0); + atomic_set(&dev->total_locks, 0); + atomic_set(&dev->total_unlocks, 0); + atomic_set(&dev->total_contends, 0); + atomic_set(&dev->total_sleeps, 0); + + for (i = 0; i < DRM_HASH_SIZE; i++) { + dev->magiclist[i].head = NULL; + dev->magiclist[i].tail = NULL; + } + dev->maplist = NULL; + dev->map_count = 0; + dev->vmalist = NULL; + dev->lock.hw_lock = NULL; + init_waitqueue_head(&dev->lock.lock_queue); + dev->queue_count = 0; + dev->queue_reserved = 0; + dev->queue_slots = 0; + dev->queuelist = NULL; + dev->irq = 0; + dev->context_flag = 0; + dev->interrupt_flag = 0; + dev->dma_flag = 0; + dev->last_context = 0; + dev->last_switch = 0; + dev->last_checked = 0; + init_timer(&dev->timer); + init_waitqueue_head(&dev->context_wait); + + dev->ctx_start = 0; + dev->lck_start = 0; + + dev->buf_rp = dev->buf; + dev->buf_wp = dev->buf; + dev->buf_end = dev->buf + DRM_BSZ; + dev->buf_async = NULL; + init_waitqueue_head(&dev->buf_readers); + init_waitqueue_head(&dev->buf_writers); + + radeon_res_ctx.handle = -1; + + DRM_DEBUG("\n"); + + /* The kernel's context could be created here, but is now created + in drm_dma_enqueue. This is more resource-efficient for + hardware that does not do DMA, but may mean that + drm_select_queue fails between the time the interrupt is + initialized and the time the queues are initialized. */ + + return 0; +} + + +static int radeon_takedown(drm_device_t *dev) +{ + int i; + drm_magic_entry_t *pt, *next; + drm_map_t *map; + drm_vma_entry_t *vma, *vma_next; + + DRM_DEBUG("\n"); + + down(&dev->struct_sem); + del_timer(&dev->timer); + + if (dev->devname) { + drm_free(dev->devname, strlen(dev->devname)+1, DRM_MEM_DRIVER); + dev->devname = NULL; + } + + if (dev->unique) { + drm_free(dev->unique, strlen(dev->unique)+1, DRM_MEM_DRIVER); + dev->unique = NULL; + dev->unique_len = 0; + } + /* Clear pid list */ + for (i = 0; i < DRM_HASH_SIZE; i++) { + for (pt = dev->magiclist[i].head; pt; pt = next) { + next = pt->next; + drm_free(pt, sizeof(*pt), DRM_MEM_MAGIC); + } + dev->magiclist[i].head = dev->magiclist[i].tail = NULL; + } + +#if defined(CONFIG_AGP) || defined(CONFIG_AGP_MODULE) + /* Clear AGP information */ + if (dev->agp) { + drm_agp_mem_t *entry; + drm_agp_mem_t *nexte; + + /* Remove AGP resources, but leave dev->agp + intact until radeon_cleanup is called. */ + for (entry = dev->agp->memory; entry; entry = nexte) { + nexte = entry->next; + if (entry->bound) drm_unbind_agp(entry->memory); + drm_free_agp(entry->memory, entry->pages); + drm_free(entry, sizeof(*entry), DRM_MEM_AGPLISTS); + } + dev->agp->memory = NULL; + + if (dev->agp->acquired) _drm_agp_release(); + + dev->agp->acquired = 0; + dev->agp->enabled = 0; + } +#endif + + /* Clear vma list (only built for debugging) */ + if (dev->vmalist) { + for (vma = dev->vmalist; vma; vma = vma_next) { + vma_next = vma->next; + drm_free(vma, sizeof(*vma), DRM_MEM_VMAS); + } + dev->vmalist = NULL; + } + + /* Clear map area and mtrr information */ + if (dev->maplist) { + for (i = 0; i < dev->map_count; i++) { + map = dev->maplist[i]; + switch (map->type) { + case _DRM_REGISTERS: + case _DRM_FRAME_BUFFER: +#ifdef CONFIG_MTRR + if (map->mtrr >= 0) { + int retcode; + retcode = mtrr_del(map->mtrr, + map->offset, + map->size); + DRM_DEBUG("mtrr_del = %d\n", retcode); + } +#endif + drm_ioremapfree(map->handle, map->size); + break; + case _DRM_SHM: + drm_free_pages((unsigned long)map->handle, + drm_order(map->size) + - PAGE_SHIFT, + DRM_MEM_SAREA); + break; + case _DRM_AGP: + /* Do nothing here, because this is all + handled in the AGP/GART driver. */ + break; + } + drm_free(map, sizeof(*map), DRM_MEM_MAPS); + } + drm_free(dev->maplist, + dev->map_count * sizeof(*dev->maplist), + DRM_MEM_MAPS); + dev->maplist = NULL; + dev->map_count = 0; + } + + drm_dma_takedown(dev); + + dev->queue_count = 0; + if (dev->lock.hw_lock) { + dev->lock.hw_lock = NULL; /* SHM removed */ + dev->lock.pid = 0; + wake_up_interruptible(&dev->lock.lock_queue); + } + up(&dev->struct_sem); + + return 0; +} + +/* radeon_init is called via init_module at module load time, or via + * linux/init/main.c (this is not currently supported). */ + +static int __init radeon_init(void) +{ + int retcode; + drm_device_t *dev = &radeon_device; + + DRM_DEBUG("\n"); + + memset((void *)dev, 0, sizeof(*dev)); + dev->count_lock = SPIN_LOCK_UNLOCKED; + sema_init(&dev->struct_sem, 1); + +#ifdef MODULE + drm_parse_options(radeon); +#endif + + if ((retcode = misc_register(&radeon_misc))) { + DRM_ERROR("Cannot register \"%s\"\n", RADEON_NAME); + return retcode; + } + dev->device = MKDEV(MISC_MAJOR, radeon_misc.minor); + dev->name = RADEON_NAME; + + drm_mem_init(); + drm_proc_init(dev); + +#if defined(CONFIG_AGP) || defined(CONFIG_AGP_MODULE) + dev->agp = drm_agp_init(); + if (dev->agp == NULL) { + DRM_ERROR("Cannot initialize agpgart module.\n"); + drm_proc_cleanup(); + misc_deregister(&radeon_misc); + radeon_takedown(dev); + return -ENOMEM; + } + +#ifdef CONFIG_MTRR + dev->agp->agp_mtrr = mtrr_add(dev->agp->agp_info.aper_base, + dev->agp->agp_info.aper_size*1024*1024, + MTRR_TYPE_WRCOMB, + 1); +#endif +#endif + + if((retcode = drm_ctxbitmap_init(dev))) { + DRM_ERROR("Cannot allocate memory for context bitmap.\n"); + drm_proc_cleanup(); + misc_deregister(&radeon_misc); + radeon_takedown(dev); + return retcode; + } + + DRM_INFO("Initialized %s %d.%d.%d %s on minor %d\n", + RADEON_NAME, + RADEON_MAJOR, + RADEON_MINOR, + RADEON_PATCHLEVEL, + RADEON_DATE, + radeon_misc.minor); + + return 0; +} + +/* radeon_cleanup is called via cleanup_module at module unload time. */ + +static void __exit radeon_cleanup(void) +{ + drm_device_t *dev = &radeon_device; + + DRM_DEBUG("\n"); + + drm_proc_cleanup(); + if (misc_deregister(&radeon_misc)) { + DRM_ERROR("Cannot unload module\n"); + } else { + DRM_INFO("Module unloaded\n"); + } + drm_ctxbitmap_cleanup(dev); + radeon_takedown(dev); +#if defined(CONFIG_AGP) || defined(CONFIG_AGP_MODULE) + if (dev->agp) { + drm_agp_uninit(); + drm_free(dev->agp, sizeof(*dev->agp), DRM_MEM_AGPLISTS); + dev->agp = NULL; + } +#endif +} + +module_init(radeon_init); +module_exit(radeon_cleanup); + + +int radeon_version(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_version_t version; + int len; + + if (copy_from_user(&version, + (drm_version_t *)arg, + sizeof(version))) + return -EFAULT; + +#define DRM_COPY(name,value) \ + len = strlen(value); \ + if (len > name##_len) len = name##_len; \ + name##_len = strlen(value); \ + if (len && name) { \ + if (copy_to_user(name, value, len)) \ + return -EFAULT; \ + } + + version.version_major = RADEON_MAJOR; + version.version_minor = RADEON_MINOR; + version.version_patchlevel = RADEON_PATCHLEVEL; + + DRM_COPY(version.name, RADEON_NAME); + DRM_COPY(version.date, RADEON_DATE); + DRM_COPY(version.desc, RADEON_DESC); + + if (copy_to_user((drm_version_t *)arg, + &version, + sizeof(version))) + return -EFAULT; + return 0; +} + +int radeon_open(struct inode *inode, struct file *filp) +{ + drm_device_t *dev = &radeon_device; + int retcode = 0; + + DRM_DEBUG("open_count = %d\n", dev->open_count); + if (!(retcode = drm_open_helper(inode, filp, dev))) { +#if LINUX_VERSION_CODE < 0x020333 + MOD_INC_USE_COUNT; /* Needed before Linux 2.3.51 */ +#endif + atomic_inc(&dev->total_open); + spin_lock(&dev->count_lock); + if (!dev->open_count++) { + spin_unlock(&dev->count_lock); + return radeon_setup(dev); + } + spin_unlock(&dev->count_lock); + } + + return retcode; +} + +int radeon_release(struct inode *inode, struct file *filp) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev; + int retcode = 0; + + lock_kernel(); + dev = priv->dev; + + DRM_DEBUG("open_count = %d\n", dev->open_count); + + /* Force the cleanup of page flipping when required */ + if ( dev->dev_private ) { + drm_radeon_private_t *dev_priv = dev->dev_private; + if ( dev_priv->page_flipping ) { + radeon_do_cleanup_pageflip( dev ); + } + } + + if (!(retcode = drm_release(inode, filp))) { +#if LINUX_VERSION_CODE < 0x020333 + MOD_DEC_USE_COUNT; /* Needed before Linux 2.3.51 */ +#endif + atomic_inc(&dev->total_close); + spin_lock(&dev->count_lock); + if (!--dev->open_count) { + if (atomic_read(&dev->ioctl_count) || dev->blocked) { + DRM_ERROR("Device busy: %d %d\n", + atomic_read(&dev->ioctl_count), + dev->blocked); + spin_unlock(&dev->count_lock); + unlock_kernel(); + return -EBUSY; + } + spin_unlock(&dev->count_lock); + unlock_kernel(); + return radeon_takedown(dev); + } + spin_unlock(&dev->count_lock); + } + + unlock_kernel(); + return retcode; +} + +/* radeon_ioctl is called whenever a process performs an ioctl on /dev/drm. */ + +int radeon_ioctl(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + int nr = DRM_IOCTL_NR(cmd); + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + int retcode = 0; + drm_ioctl_desc_t *ioctl; + drm_ioctl_t *func; + + atomic_inc(&dev->ioctl_count); + atomic_inc(&dev->total_ioctl); + ++priv->ioctl_count; + + DRM_DEBUG("pid = %d, cmd = 0x%02x, nr = 0x%02x, dev 0x%x, auth = %d\n", + current->pid, cmd, nr, dev->device, priv->authenticated); + + if (nr >= RADEON_IOCTL_COUNT) { + retcode = -EINVAL; + } else { + ioctl = &radeon_ioctls[nr]; + func = ioctl->func; + + if (!func) { + DRM_DEBUG("no function\n"); + retcode = -EINVAL; + } else if ((ioctl->root_only && !capable(CAP_SYS_ADMIN)) + || (ioctl->auth_needed && !priv->authenticated)) { + retcode = -EACCES; + } else { + retcode = (func)(inode, filp, cmd, arg); + } + } + + atomic_dec(&dev->ioctl_count); + return retcode; +} + +int radeon_lock(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + DECLARE_WAITQUEUE(entry, current); + int ret = 0; + drm_lock_t lock; +#if DRM_DMA_HISTOGRAM + cycles_t start; + + dev->lck_start = start = get_cycles(); +#endif + + if (copy_from_user(&lock, (drm_lock_t *)arg, sizeof(lock))) + return -EFAULT; + + if (lock.context == DRM_KERNEL_CONTEXT) { + DRM_ERROR("Process %d using kernel context %d\n", + current->pid, lock.context); + return -EINVAL; + } + + DRM_DEBUG("%d (pid %d) requests lock (0x%08x), flags = 0x%08x\n", + lock.context, current->pid, dev->lock.hw_lock->lock, + lock.flags); + + if (lock.context < 0 /* || lock.context >= dev->queue_count */) + return -EINVAL; + + if (!ret) { + add_wait_queue(&dev->lock.lock_queue, &entry); + for (;;) { + current->state = TASK_INTERRUPTIBLE; + if (!dev->lock.hw_lock) { + /* Device has been unregistered */ + ret = -EINTR; + break; + } + if (drm_lock_take(&dev->lock.hw_lock->lock, + lock.context)) { + dev->lock.pid = current->pid; + dev->lock.lock_time = jiffies; + atomic_inc(&dev->total_locks); + break; /* Got lock */ + } + + /* Contention */ + atomic_inc(&dev->total_sleeps); + schedule(); + if (signal_pending(current)) { + ret = -ERESTARTSYS; + break; + } + } + current->state = TASK_RUNNING; + remove_wait_queue(&dev->lock.lock_queue, &entry); + } + + if (!ret) { + sigemptyset(&dev->sigmask); + sigaddset(&dev->sigmask, SIGSTOP); + sigaddset(&dev->sigmask, SIGTSTP); + sigaddset(&dev->sigmask, SIGTTIN); + sigaddset(&dev->sigmask, SIGTTOU); + dev->sigdata.context = lock.context; + dev->sigdata.lock = dev->lock.hw_lock; + block_all_signals(drm_notifier, &dev->sigdata, &dev->sigmask); + if (lock.flags & _DRM_LOCK_READY) { + /* Wait for space in DMA/FIFO */ + } + if (lock.flags & _DRM_LOCK_QUIESCENT) { + /* Make hardware quiescent */ + DRM_DEBUG("not quiescent!\n"); +#if 0 + radeon_quiescent(dev); +#endif + } + } + +#if LINUX_VERSION_CODE < 0x020400 + if (lock.context != radeon_res_ctx.handle) { + current->counter = 5; + current->priority = DEF_PRIORITY/4; + } +#endif + DRM_DEBUG("%d %s\n", lock.context, ret ? "interrupted" : "has lock"); + +#if DRM_DMA_HISTOGRAM + atomic_inc(&dev->histo.lacq[drm_histogram_slot(get_cycles() - start)]); +#endif + + return ret; +} + + +int radeon_unlock(struct inode *inode, struct file *filp, unsigned int cmd, + unsigned long arg) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_lock_t lock; + + if (copy_from_user(&lock, (drm_lock_t *)arg, sizeof(lock))) + return -EFAULT; + + if (lock.context == DRM_KERNEL_CONTEXT) { + DRM_ERROR("Process %d using kernel context %d\n", + current->pid, lock.context); + return -EINVAL; + } + + DRM_DEBUG("%d frees lock (%d holds)\n", + lock.context, + _DRM_LOCKING_CONTEXT(dev->lock.hw_lock->lock)); + atomic_inc(&dev->total_unlocks); + if (_DRM_LOCK_IS_CONT(dev->lock.hw_lock->lock)) + atomic_inc(&dev->total_contends); + drm_lock_transfer(dev, &dev->lock.hw_lock->lock, DRM_KERNEL_CONTEXT); + /* FIXME: Try to send data to card here */ + if (!dev->context_flag) { + if (drm_lock_free(dev, &dev->lock.hw_lock->lock, + DRM_KERNEL_CONTEXT)) { + DRM_ERROR("\n"); + } + } + +#if LINUX_VERSION_CODE < 0x020400 + if (lock.context != radeon_res_ctx.handle) { + current->counter = 5; + current->priority = DEF_PRIORITY; + } +#endif + unblock_all_signals(); + return 0; +} diff --git a/drivers/char/drm/radeon_drv.h b/drivers/char/drm/radeon_drv.h new file mode 100644 index 000000000..06b541991 --- /dev/null +++ b/drivers/char/drm/radeon_drv.h @@ -0,0 +1,709 @@ +/* radeon_drv.h -- Private header for radeon driver -*- linux-c -*- + * + * Copyright 1999 Precision Insight, Inc., Cedar Park, Texas. + * Copyright 2000 VA Linux Systems, Inc., Fremont, California. + * All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Rickard E. (Rik) Faith <faith@valinux.com> + * Kevin E. Martin <martin@valinux.com> + * Gareth Hughes <gareth@valinux.com> + * + */ + +#ifndef __RADEON_DRV_H__ +#define __RADEON_DRV_H__ + +typedef struct drm_radeon_freelist { + unsigned int age; + drm_buf_t *buf; + struct drm_radeon_freelist *next; + struct drm_radeon_freelist *prev; +} drm_radeon_freelist_t; + +typedef struct drm_radeon_ring_buffer { + u32 *start; + u32 *end; + int size; + int size_l2qw; + + volatile u32 *head; + u32 tail; + u32 tail_mask; + int space; +} drm_radeon_ring_buffer_t; + +typedef struct drm_radeon_depth_clear_t { + u32 rb3d_cntl; + u32 rb3d_zstencilcntl; + u32 se_cntl; +} drm_radeon_depth_clear_t; + +typedef struct drm_radeon_private { + drm_radeon_ring_buffer_t ring; + drm_radeon_sarea_t *sarea_priv; + + int agp_size; + u32 agp_vm_start; + u32 agp_buffers_offset; + + int cp_mode; + int cp_running; + + drm_radeon_freelist_t *head; + drm_radeon_freelist_t *tail; +/* FIXME: ROTATE_BUFS is a hask to cycle through bufs until freelist + code is used. Note this hides a problem with the scratch register + (used to keep track of last buffer completed) being written to before + the last buffer has actually completed rendering. */ +#define ROTATE_BUFS 1 +#if ROTATE_BUFS + int last_buf; +#endif + volatile u32 *scratch; + + int usec_timeout; + int is_pci; + + atomic_t idle_count; + + int page_flipping; + int current_page; + u32 crtc_offset; + u32 crtc_offset_cntl; + + unsigned int color_fmt; + unsigned int front_offset; + unsigned int front_pitch; + unsigned int back_offset; + unsigned int back_pitch; + + unsigned int depth_fmt; + unsigned int depth_offset; + unsigned int depth_pitch; + + u32 front_pitch_offset; + u32 back_pitch_offset; + u32 depth_pitch_offset; + + drm_radeon_depth_clear_t depth_clear; + + drm_map_t *sarea; + drm_map_t *fb; + drm_map_t *mmio; + drm_map_t *cp_ring; + drm_map_t *ring_rptr; + drm_map_t *buffers; + drm_map_t *agp_textures; +} drm_radeon_private_t; + +typedef struct drm_radeon_buf_priv { + u32 age; + int prim; + int discard; + int dispatched; + drm_radeon_freelist_t *list_entry; +} drm_radeon_buf_priv_t; + + /* radeon_drv.c */ +extern int radeon_version( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_open( struct inode *inode, struct file *filp ); +extern int radeon_release( struct inode *inode, struct file *filp ); +extern int radeon_ioctl( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_lock( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_unlock( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); + + /* radeon_cp.c */ +extern int radeon_cp_init( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_start( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_stop( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_reset( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_idle( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_engine_reset( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_fullscreen( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_buffers( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); + +extern void radeon_freelist_reset( drm_device_t *dev ); +extern drm_buf_t *radeon_freelist_get( drm_device_t *dev ); + +extern int radeon_wait_ring( drm_radeon_private_t *dev_priv, int n ); +extern void radeon_update_ring_snapshot( drm_radeon_private_t *dev_priv ); + +extern int radeon_do_cp_idle( drm_radeon_private_t *dev_priv ); +extern int radeon_do_cleanup_pageflip( drm_device_t *dev ); + + /* radeon_state.c */ +extern int radeon_cp_clear( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_swap( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_vertex( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_indices( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_blit( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_stipple( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); +extern int radeon_cp_indirect( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ); + + /* radeon_bufs.c */ +extern int radeon_addbufs(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg); +extern int radeon_mapbufs(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg); + + /* radeon_context.c */ +extern int radeon_resctx(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg); +extern int radeon_addctx(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg); +extern int radeon_modctx(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg); +extern int radeon_getctx(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg); +extern int radeon_switchctx(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg); +extern int radeon_newctx(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg); +extern int radeon_rmctx(struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg); + +extern int radeon_context_switch(drm_device_t *dev, int old, int new); +extern int radeon_context_switch_complete(drm_device_t *dev, int new); + + +/* Register definitions, register access macros and drmAddMap constants + * for Radeon kernel driver. + */ + +#define RADEON_AUX_SCISSOR_CNTL 0x26f0 +# define RADEON_EXCLUSIVE_SCISSOR_0 (1 << 24) +# define RADEON_EXCLUSIVE_SCISSOR_1 (1 << 25) +# define RADEON_EXCLUSIVE_SCISSOR_2 (1 << 26) +# define RADEON_SCISSOR_0_ENABLE (1 << 28) +# define RADEON_SCISSOR_1_ENABLE (1 << 29) +# define RADEON_SCISSOR_2_ENABLE (1 << 30) + +#define RADEON_BUS_CNTL 0x0030 +# define RADEON_BUS_MASTER_DIS (1 << 6) + +#define RADEON_CLOCK_CNTL_DATA 0x000c +# define RADEON_PLL_WR_EN (1 << 7) +#define RADEON_CLOCK_CNTL_INDEX 0x0008 +#define RADEON_CONFIG_APER_SIZE 0x0108 +#define RADEON_CRTC_OFFSET 0x0224 +#define RADEON_CRTC_OFFSET_CNTL 0x0228 +# define RADEON_CRTC_TILE_EN (1 << 15) +# define RADEON_CRTC_OFFSET_FLIP_CNTL (1 << 16) + +#define RADEON_RB3D_COLORPITCH 0x1c48 +#define RADEON_RB3D_DEPTHCLEARVALUE 0x1c30 +#define RADEON_RB3D_DEPTHXY_OFFSET 0x1c60 + +#define RADEON_DP_GUI_MASTER_CNTL 0x146c +# define RADEON_GMC_SRC_PITCH_OFFSET_CNTL (1 << 0) +# define RADEON_GMC_DST_PITCH_OFFSET_CNTL (1 << 1) +# define RADEON_GMC_BRUSH_SOLID_COLOR (13 << 4) +# define RADEON_GMC_BRUSH_NONE (15 << 4) +# define RADEON_GMC_DST_16BPP (4 << 8) +# define RADEON_GMC_DST_24BPP (5 << 8) +# define RADEON_GMC_DST_32BPP (6 << 8) +# define RADEON_GMC_DST_DATATYPE_SHIFT 8 +# define RADEON_GMC_SRC_DATATYPE_COLOR (3 << 12) +# define RADEON_DP_SRC_SOURCE_MEMORY (2 << 24) +# define RADEON_DP_SRC_SOURCE_HOST_DATA (3 << 24) +# define RADEON_GMC_CLR_CMP_CNTL_DIS (1 << 28) +# define RADEON_GMC_WR_MSK_DIS (1 << 30) +# define RADEON_ROP3_S 0x00cc0000 +# define RADEON_ROP3_P 0x00f00000 +#define RADEON_DP_WRITE_MASK 0x16cc +#define RADEON_DST_PITCH_OFFSET 0x142c +#define RADEON_DST_PITCH_OFFSET_C 0x1c80 +# define RADEON_DST_TILE_LINEAR (0 << 30) +# define RADEON_DST_TILE_MACRO (1 << 30) +# define RADEON_DST_TILE_MICRO (2 << 30) +# define RADEON_DST_TILE_BOTH (3 << 30) + +#define RADEON_SCRATCH_REG0 0x15e0 +#define RADEON_SCRATCH_REG1 0x15e4 +#define RADEON_SCRATCH_REG2 0x15e8 +#define RADEON_SCRATCH_REG3 0x15ec +#define RADEON_SCRATCH_REG4 0x15f0 +#define RADEON_SCRATCH_REG5 0x15f4 +#define RADEON_SCRATCH_UMSK 0x0770 +#define RADEON_SCRATCH_ADDR 0x0774 + +#define RADEON_HOST_PATH_CNTL 0x0130 +# define RADEON_HDP_SOFT_RESET (1 << 26) +# define RADEON_HDP_WC_TIMEOUT_MASK (7 << 28) +# define RADEON_HDP_WC_TIMEOUT_28BCLK (7 << 28) + +#define RADEON_ISYNC_CNTL 0x1724 +# define RADEON_ISYNC_ANY2D_IDLE3D (1 << 0) +# define RADEON_ISYNC_ANY3D_IDLE2D (1 << 1) +# define RADEON_ISYNC_TRIG2D_IDLE3D (1 << 2) +# define RADEON_ISYNC_TRIG3D_IDLE2D (1 << 3) +# define RADEON_ISYNC_WAIT_IDLEGUI (1 << 4) +# define RADEON_ISYNC_CPSCRATCH_IDLEGUI (1 << 5) + +#define RADEON_MC_AGP_LOCATION 0x014c +#define RADEON_MC_FB_LOCATION 0x0148 +#define RADEON_MCLK_CNTL 0x0012 + +#define RADEON_PP_BORDER_COLOR_0 0x1d40 +#define RADEON_PP_BORDER_COLOR_1 0x1d44 +#define RADEON_PP_BORDER_COLOR_2 0x1d48 +#define RADEON_PP_CNTL 0x1c38 +# define RADEON_SCISSOR_ENABLE (1 << 1) +#define RADEON_PP_LUM_MATRIX 0x1d00 +#define RADEON_PP_MISC 0x1c14 +#define RADEON_PP_ROT_MATRIX_0 0x1d58 +#define RADEON_PP_TXFILTER_0 0x1c54 +#define RADEON_PP_TXFILTER_1 0x1c6c +#define RADEON_PP_TXFILTER_2 0x1c84 + +#define RADEON_RB2D_DSTCACHE_CTLSTAT 0x342c +# define RADEON_RB2D_DC_FLUSH (3 << 0) +# define RADEON_RB2D_DC_FREE (3 << 2) +# define RADEON_RB2D_DC_FLUSH_ALL 0xf +# define RADEON_RB2D_DC_BUSY (1 << 31) +#define RADEON_RB3D_CNTL 0x1c3c +# define RADEON_ALPHA_BLEND_ENABLE (1 << 0) +# define RADEON_PLANE_MASK_ENABLE (1 << 1) +# define RADEON_DITHER_ENABLE (1 << 2) +# define RADEON_ROUND_ENABLE (1 << 3) +# define RADEON_SCALE_DITHER_ENABLE (1 << 4) +# define RADEON_DITHER_INIT (1 << 5) +# define RADEON_ROP_ENABLE (1 << 6) +# define RADEON_STENCIL_ENABLE (1 << 7) +# define RADEON_Z_ENABLE (1 << 8) +# define RADEON_DEPTH_XZ_OFFEST_ENABLE (1 << 9) +# define RADEON_ZBLOCK8 (0 << 15) +# define RADEON_ZBLOCK16 (1 << 15) +#define RADEON_RB3D_DEPTHOFFSET 0x1c24 +#define RADEON_RB3D_PLANEMASK 0x1d84 +#define RADEON_RB3D_STENCILREFMASK 0x1d7c +#define RADEON_RB3D_ZCACHE_MODE 0x3250 +#define RADEON_RB3D_ZCACHE_CTLSTAT 0x3254 +# define RADEON_RB3D_ZC_FLUSH (1 << 0) +# define RADEON_RB3D_ZC_FREE (1 << 2) +# define RADEON_RB3D_ZC_FLUSH_ALL 0x5 +# define RADEON_RB3D_ZC_BUSY (1 << 31) +#define RADEON_RB3D_ZSTENCILCNTL 0x1c2c +# define RADEON_Z_TEST_MASK (7 << 4) +# define RADEON_Z_TEST_ALWAYS (7 << 4) +# define RADEON_STENCIL_TEST_ALWAYS (7 << 12) +# define RADEON_STENCIL_S_FAIL_KEEP (0 << 16) +# define RADEON_STENCIL_ZPASS_KEEP (0 << 20) +# define RADEON_STENCIL_ZFAIL_KEEP (0 << 20) +# define RADEON_Z_WRITE_ENABLE (1 << 30) +#define RADEON_RBBM_SOFT_RESET 0x00f0 +# define RADEON_SOFT_RESET_CP (1 << 0) +# define RADEON_SOFT_RESET_HI (1 << 1) +# define RADEON_SOFT_RESET_SE (1 << 2) +# define RADEON_SOFT_RESET_RE (1 << 3) +# define RADEON_SOFT_RESET_PP (1 << 4) +# define RADEON_SOFT_RESET_E2 (1 << 5) +# define RADEON_SOFT_RESET_RB (1 << 6) +# define RADEON_SOFT_RESET_HDP (1 << 7) +#define RADEON_RBBM_STATUS 0x0e40 +# define RADEON_RBBM_FIFOCNT_MASK 0x007f +# define RADEON_RBBM_ACTIVE (1 << 31) +#define RADEON_RE_LINE_PATTERN 0x1cd0 +#define RADEON_RE_MISC 0x26c4 +#define RADEON_RE_TOP_LEFT 0x26c0 +#define RADEON_RE_WIDTH_HEIGHT 0x1c44 +#define RADEON_RE_STIPPLE_ADDR 0x1cc8 +#define RADEON_RE_STIPPLE_DATA 0x1ccc + +#define RADEON_SCISSOR_TL_0 0x1cd8 +#define RADEON_SCISSOR_BR_0 0x1cdc +#define RADEON_SCISSOR_TL_1 0x1ce0 +#define RADEON_SCISSOR_BR_1 0x1ce4 +#define RADEON_SCISSOR_TL_2 0x1ce8 +#define RADEON_SCISSOR_BR_2 0x1cec +#define RADEON_SE_COORD_FMT 0x1c50 +#define RADEON_SE_CNTL 0x1c4c +# define RADEON_FFACE_CULL_CW (0 << 0) +# define RADEON_BFACE_SOLID (3 << 1) +# define RADEON_FFACE_SOLID (3 << 3) +# define RADEON_FLAT_SHADE_VTX_LAST (3 << 6) +# define RADEON_DIFFUSE_SHADE_FLAT (1 << 8) +# define RADEON_DIFFUSE_SHADE_GOURAUD (2 << 8) +# define RADEON_ALPHA_SHADE_FLAT (1 << 10) +# define RADEON_ALPHA_SHADE_GOURAUD (2 << 10) +# define RADEON_SPECULAR_SHADE_FLAT (1 << 12) +# define RADEON_SPECULAR_SHADE_GOURAUD (2 << 12) +# define RADEON_FOG_SHADE_FLAT (1 << 14) +# define RADEON_FOG_SHADE_GOURAUD (2 << 14) +# define RADEON_VPORT_XY_XFORM_ENABLE (1 << 24) +# define RADEON_VPORT_Z_XFORM_ENABLE (1 << 25) +# define RADEON_VTX_PIX_CENTER_OGL (1 << 27) +# define RADEON_ROUND_MODE_TRUNC (0 << 28) +# define RADEON_ROUND_PREC_8TH_PIX (1 << 30) +#define RADEON_SE_CNTL_STATUS 0x2140 +#define RADEON_SE_LINE_WIDTH 0x1db8 +#define RADEON_SE_VPORT_XSCALE 0x1d98 +#define RADEON_SURFACE_ACCESS_FLAGS 0x0bf8 +#define RADEON_SURFACE_ACCESS_CLR 0x0bfc +#define RADEON_SURFACE_CNTL 0x0b00 +# define RADEON_SURF_TRANSLATION_DIS (1 << 8) +# define RADEON_NONSURF_AP0_SWP_MASK (3 << 20) +# define RADEON_NONSURF_AP0_SWP_LITTLE (0 << 20) +# define RADEON_NONSURF_AP0_SWP_BIG16 (1 << 20) +# define RADEON_NONSURF_AP0_SWP_BIG32 (2 << 20) +# define RADEON_NONSURF_AP1_SWP_MASK (3 << 22) +# define RADEON_NONSURF_AP1_SWP_LITTLE (0 << 22) +# define RADEON_NONSURF_AP1_SWP_BIG16 (1 << 22) +# define RADEON_NONSURF_AP1_SWP_BIG32 (2 << 22) +#define RADEON_SURFACE0_INFO 0x0b0c +# define RADEON_SURF_PITCHSEL_MASK (0x1ff << 0) +# define RADEON_SURF_TILE_MODE_MASK (3 << 16) +# define RADEON_SURF_TILE_MODE_MACRO (0 << 16) +# define RADEON_SURF_TILE_MODE_MICRO (1 << 16) +# define RADEON_SURF_TILE_MODE_32BIT_Z (2 << 16) +# define RADEON_SURF_TILE_MODE_16BIT_Z (3 << 16) +#define RADEON_SURFACE0_LOWER_BOUND 0x0b04 +#define RADEON_SURFACE0_UPPER_BOUND 0x0b08 +#define RADEON_SURFACE1_INFO 0x0b1c +#define RADEON_SURFACE1_LOWER_BOUND 0x0b14 +#define RADEON_SURFACE1_UPPER_BOUND 0x0b18 +#define RADEON_SURFACE2_INFO 0x0b2c +#define RADEON_SURFACE2_LOWER_BOUND 0x0b24 +#define RADEON_SURFACE2_UPPER_BOUND 0x0b28 +#define RADEON_SURFACE3_INFO 0x0b3c +#define RADEON_SURFACE3_LOWER_BOUND 0x0b34 +#define RADEON_SURFACE3_UPPER_BOUND 0x0b38 +#define RADEON_SURFACE4_INFO 0x0b4c +#define RADEON_SURFACE4_LOWER_BOUND 0x0b44 +#define RADEON_SURFACE4_UPPER_BOUND 0x0b48 +#define RADEON_SURFACE5_INFO 0x0b5c +#define RADEON_SURFACE5_LOWER_BOUND 0x0b54 +#define RADEON_SURFACE5_UPPER_BOUND 0x0b58 +#define RADEON_SURFACE6_INFO 0x0b6c +#define RADEON_SURFACE6_LOWER_BOUND 0x0b64 +#define RADEON_SURFACE6_UPPER_BOUND 0x0b68 +#define RADEON_SURFACE7_INFO 0x0b7c +#define RADEON_SURFACE7_LOWER_BOUND 0x0b74 +#define RADEON_SURFACE7_UPPER_BOUND 0x0b78 +#define RADEON_SW_SEMAPHORE 0x013c + +#define RADEON_WAIT_UNTIL 0x1720 +# define RADEON_WAIT_CRTC_PFLIP (1 << 0) +# define RADEON_WAIT_2D_IDLECLEAN (1 << 16) +# define RADEON_WAIT_3D_IDLECLEAN (1 << 17) +# define RADEON_WAIT_HOST_IDLECLEAN (1 << 18) + +#define RADEON_RB3D_ZMASKOFFSET 0x1c34 +#define RADEON_RB3D_ZSTENCILCNTL 0x1c2c +# define RADEON_DEPTH_FORMAT_16BIT_INT_Z (0 << 0) +# define RADEON_DEPTH_FORMAT_24BIT_INT_Z (2 << 0) + + +/* CP registers */ +#define RADEON_CP_ME_RAM_ADDR 0x07d4 +#define RADEON_CP_ME_RAM_RADDR 0x07d8 +#define RADEON_CP_ME_RAM_DATAH 0x07dc +#define RADEON_CP_ME_RAM_DATAL 0x07e0 + +#define RADEON_CP_RB_BASE 0x0700 +#define RADEON_CP_RB_CNTL 0x0704 +#define RADEON_CP_RB_RPTR_ADDR 0x070c +#define RADEON_CP_RB_RPTR 0x0710 +#define RADEON_CP_RB_WPTR 0x0714 + +#define RADEON_CP_RB_WPTR_DELAY 0x0718 +# define RADEON_PRE_WRITE_TIMER_SHIFT 0 +# define RADEON_PRE_WRITE_LIMIT_SHIFT 23 + +#define RADEON_CP_IB_BASE 0x0738 + +#define RADEON_CP_CSQ_CNTL 0x0740 +# define RADEON_CSQ_CNT_PRIMARY_MASK (0xff << 0) +# define RADEON_CSQ_PRIDIS_INDDIS (0 << 28) +# define RADEON_CSQ_PRIPIO_INDDIS (1 << 28) +# define RADEON_CSQ_PRIBM_INDDIS (2 << 28) +# define RADEON_CSQ_PRIPIO_INDBM (3 << 28) +# define RADEON_CSQ_PRIBM_INDBM (4 << 28) +# define RADEON_CSQ_PRIPIO_INDPIO (15 << 28) + +#define RADEON_AIC_CNTL 0x01d0 +# define RADEON_PCIGART_TRANSLATE_EN (1 << 0) + +/* CP command packets */ +#define RADEON_CP_PACKET0 0x00000000 +# define RADEON_ONE_REG_WR (1 << 15) +#define RADEON_CP_PACKET1 0x40000000 +#define RADEON_CP_PACKET2 0x80000000 +#define RADEON_CP_PACKET3 0xC0000000 +# define RADEON_3D_RNDR_GEN_INDX_PRIM 0x00002300 +# define RADEON_WAIT_FOR_IDLE 0x00002600 +# define RADEON_3D_DRAW_IMMD 0x00002900 +# define RADEON_3D_CLEAR_ZMASK 0x00003200 +# define RADEON_CNTL_HOSTDATA_BLT 0x00009400 +# define RADEON_CNTL_PAINT_MULTI 0x00009A00 +# define RADEON_CNTL_BITBLT_MULTI 0x00009B00 + +#define RADEON_CP_PACKET_MASK 0xC0000000 +#define RADEON_CP_PACKET_COUNT_MASK 0x3fff0000 +#define RADEON_CP_PACKET0_REG_MASK 0x000007ff +#define RADEON_CP_PACKET1_REG0_MASK 0x000007ff +#define RADEON_CP_PACKET1_REG1_MASK 0x003ff800 + +#define RADEON_VTX_Z_PRESENT (1 << 31) + +#define RADEON_PRIM_TYPE_NONE (0 << 0) +#define RADEON_PRIM_TYPE_POINT (1 << 0) +#define RADEON_PRIM_TYPE_LINE (2 << 0) +#define RADEON_PRIM_TYPE_LINE_STRIP (3 << 0) +#define RADEON_PRIM_TYPE_TRI_LIST (4 << 0) +#define RADEON_PRIM_TYPE_TRI_FAN (5 << 0) +#define RADEON_PRIM_TYPE_TRI_STRIP (6 << 0) +#define RADEON_PRIM_TYPE_TRI_TYPE2 (7 << 0) +#define RADEON_PRIM_TYPE_RECT_LIST (8 << 0) +#define RADEON_PRIM_TYPE_3VRT_POINT_LIST (9 << 0) +#define RADEON_PRIM_TYPE_3VRT_LINE_LIST (10 << 0) +#define RADEON_PRIM_WALK_IND (1 << 4) +#define RADEON_PRIM_WALK_LIST (2 << 4) +#define RADEON_PRIM_WALK_RING (3 << 4) +#define RADEON_COLOR_ORDER_BGRA (0 << 6) +#define RADEON_COLOR_ORDER_RGBA (1 << 6) +#define RADEON_MAOS_ENABLE (1 << 7) +#define RADEON_VTX_FMT_R128_MODE (0 << 8) +#define RADEON_VTX_FMT_RADEON_MODE (1 << 8) +#define RADEON_NUM_VERTICES_SHIFT 16 + +#define RADEON_COLOR_FORMAT_CI8 2 +#define RADEON_COLOR_FORMAT_ARGB1555 3 +#define RADEON_COLOR_FORMAT_RGB565 4 +#define RADEON_COLOR_FORMAT_ARGB8888 6 +#define RADEON_COLOR_FORMAT_RGB332 7 +#define RADEON_COLOR_FORMAT_RGB8 9 +#define RADEON_COLOR_FORMAT_ARGB4444 15 + +#define RADEON_TXF_8BPP_I 0 +#define RADEON_TXF_16BPP_AI88 1 +#define RADEON_TXF_8BPP_RGB332 2 +#define RADEON_TXF_16BPP_ARGB1555 3 +#define RADEON_TXF_16BPP_RGB565 4 +#define RADEON_TXF_16BPP_ARGB4444 5 +#define RADEON_TXF_32BPP_ARGB8888 6 +#define RADEON_TXF_32BPP_RGBA8888 7 + +/* Constants */ +#define RADEON_MAX_USEC_TIMEOUT 100000 /* 100 ms */ + +#define RADEON_LAST_FRAME_REG RADEON_SCRATCH_REG0 +#define RADEON_LAST_DISPATCH_REG RADEON_SCRATCH_REG1 +#define RADEON_LAST_CLEAR_REG RADEON_SCRATCH_REG2 +#define RADEON_LAST_DISPATCH 1 + +#define RADEON_MAX_VB_AGE 0x7fffffff +#define RADEON_MAX_VB_VERTS (0xffff) + + +#define RADEON_BASE(reg) ((u32)(dev_priv->mmio->handle)) +#define RADEON_ADDR(reg) (RADEON_BASE(reg) + reg) + +#define RADEON_DEREF(reg) *(__volatile__ u32 *)RADEON_ADDR(reg) +#define RADEON_READ(reg) RADEON_DEREF(reg) +#define RADEON_WRITE(reg,val) do { RADEON_DEREF(reg) = val; } while (0) + +#define RADEON_DEREF8(reg) *(__volatile__ u8 *)RADEON_ADDR(reg) +#define RADEON_READ8(reg) RADEON_DEREF8(reg) +#define RADEON_WRITE8(reg,val) do { RADEON_DEREF8(reg) = val; } while (0) + +#define RADEON_WRITE_PLL(addr,val) \ +do { \ + RADEON_WRITE8(RADEON_CLOCK_CNTL_INDEX, \ + ((addr) & 0x1f) | RADEON_PLL_WR_EN); \ + RADEON_WRITE(RADEON_CLOCK_CNTL_DATA, (val)); \ +} while (0) + +extern int RADEON_READ_PLL(drm_device_t *dev, int addr); + + + +#define CP_PACKET0( reg, n ) \ + (RADEON_CP_PACKET0 | ((n) << 16) | ((reg) >> 2)) +#define CP_PACKET0_TABLE( reg, n ) \ + (RADEON_CP_PACKET0 | RADEON_ONE_REG_WR | ((n) << 16) | ((reg) >> 2)) +#define CP_PACKET1( reg0, reg1 ) \ + (RADEON_CP_PACKET1 | (((reg1) >> 2) << 15) | ((reg0) >> 2)) +#define CP_PACKET2() \ + (RADEON_CP_PACKET2) +#define CP_PACKET3( pkt, n ) \ + (RADEON_CP_PACKET3 | (pkt) | ((n) << 16)) + + +/* ================================================================ + * Engine control helper macros + */ + +#define RADEON_WAIT_UNTIL_2D_IDLE() \ +do { \ + OUT_RING( CP_PACKET0( RADEON_WAIT_UNTIL, 0 ) ); \ + OUT_RING( (RADEON_WAIT_2D_IDLECLEAN | \ + RADEON_WAIT_HOST_IDLECLEAN) ); \ +} while (0) + +#define RADEON_WAIT_UNTIL_3D_IDLE() \ +do { \ + OUT_RING( CP_PACKET0( RADEON_WAIT_UNTIL, 0 ) ); \ + OUT_RING( (RADEON_WAIT_3D_IDLECLEAN | \ + RADEON_WAIT_HOST_IDLECLEAN) ); \ +} while (0) + +#define RADEON_WAIT_UNTIL_IDLE() \ +do { \ + OUT_RING( CP_PACKET0( RADEON_WAIT_UNTIL, 0 ) ); \ + OUT_RING( (RADEON_WAIT_2D_IDLECLEAN | \ + RADEON_WAIT_3D_IDLECLEAN | \ + RADEON_WAIT_HOST_IDLECLEAN) ); \ +} while (0) + +#define RADEON_WAIT_UNTIL_PAGE_FLIPPED() \ +do { \ + OUT_RING( CP_PACKET0( RADEON_WAIT_UNTIL, 0 ) ); \ + OUT_RING( RADEON_WAIT_CRTC_PFLIP ); \ +} while (0) + +#define RADEON_FLUSH_CACHE() \ +do { \ + OUT_RING( CP_PACKET0( RADEON_RB2D_DSTCACHE_CTLSTAT, 0 ) ); \ + OUT_RING( RADEON_RB2D_DC_FLUSH ); \ +} while (0) + +#define RADEON_PURGE_CACHE() \ +do { \ + OUT_RING( CP_PACKET0( RADEON_RB2D_DSTCACHE_CTLSTAT, 0 ) ); \ + OUT_RING( RADEON_RB2D_DC_FLUSH_ALL ); \ +} while (0) + +#define RADEON_FLUSH_ZCACHE() \ +do { \ + OUT_RING( CP_PACKET0( RADEON_RB3D_ZCACHE_CTLSTAT, 0 ) ); \ + OUT_RING( RADEON_RB3D_ZC_FLUSH ); \ +} while (0) + +#define RADEON_PURGE_ZCACHE() \ +do { \ + OUT_RING( CP_PACKET0( RADEON_RB3D_ZCACHE_CTLSTAT, 0 ) ); \ + OUT_RING( RADEON_RB3D_ZC_FLUSH_ALL ); \ +} while (0) + + +/* ================================================================ + * Misc helper macros + */ + +#define VB_AGE_CHECK_WITH_RET( dev_priv ) \ +do { \ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; \ + if ( sarea_priv->last_dispatch >= RADEON_MAX_VB_AGE ) { \ + int __ret = radeon_do_cp_idle( dev_priv ); \ + if ( __ret < 0 ) return __ret; \ + sarea_priv->last_dispatch = 0; \ + radeon_freelist_reset( dev ); \ + } \ +} while (0) + +#define RADEON_DISPATCH_AGE( age ) \ +do { \ + OUT_RING( CP_PACKET0( RADEON_LAST_DISPATCH_REG, 0 ) ); \ + OUT_RING( age ); \ +} while (0) + +#define RADEON_FRAME_AGE( age ) \ +do { \ + OUT_RING( CP_PACKET0( RADEON_LAST_FRAME_REG, 0 ) ); \ + OUT_RING( age ); \ +} while (0) + +#define RADEON_CLEAR_AGE( age ) \ +do { \ + OUT_RING( CP_PACKET0( RADEON_LAST_CLEAR_REG, 0 ) ); \ + OUT_RING( age ); \ +} while (0) + + +/* ================================================================ + * Ring control + */ + +#define radeon_flush_write_combine() mb() + + +#define RADEON_VERBOSE 0 + +#define RING_LOCALS int write; unsigned int mask; volatile u32 *ring; + +#define BEGIN_RING( n ) do { \ + if ( RADEON_VERBOSE ) { \ + DRM_INFO( "BEGIN_RING( %d ) in %s\n", \ + n, __FUNCTION__ ); \ + } \ + if ( dev_priv->ring.space < (n) * sizeof(u32) ) { \ + radeon_wait_ring( dev_priv, (n) * sizeof(u32) ); \ + } \ + dev_priv->ring.space -= (n) * sizeof(u32); \ + ring = dev_priv->ring.start; \ + write = dev_priv->ring.tail; \ + mask = dev_priv->ring.tail_mask; \ +} while (0) + +#define ADVANCE_RING() do { \ + if ( RADEON_VERBOSE ) { \ + DRM_INFO( "ADVANCE_RING() tail=0x%06x wr=0x%06x\n", \ + write, dev_priv->ring.tail ); \ + } \ + radeon_flush_write_combine(); \ + dev_priv->ring.tail = write; \ + RADEON_WRITE( RADEON_CP_RB_WPTR, write ); \ +} while (0) + +#define OUT_RING( x ) do { \ + if ( RADEON_VERBOSE ) { \ + DRM_INFO( " OUT_RING( 0x%08x ) at 0x%x\n", \ + (unsigned int)(x), write ); \ + } \ + ring[write++] = (x); \ + write &= mask; \ +} while (0) + +#define RADEON_PERFORMANCE_BOXES 0 + +#endif /* __RADEON_DRV_H__ */ diff --git a/drivers/char/drm/radeon_state.c b/drivers/char/drm/radeon_state.c new file mode 100644 index 000000000..1a4da8005 --- /dev/null +++ b/drivers/char/drm/radeon_state.c @@ -0,0 +1,1447 @@ +/* radeon_state.c -- State support for Radeon -*- linux-c -*- + * + * Copyright 2000 VA Linux Systems, Inc., Fremont, California. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + * + * Authors: + * Kevin E. Martin <martin@valinux.com> + * Gareth Hughes <gareth@valinux.com> + * + */ + +#define __NO_VERSION__ +#include "drmP.h" +#include "radeon_drv.h" +#include "drm.h" +#include <linux/delay.h> + + +/* ================================================================ + * CP hardware state programming functions + */ + +static inline void radeon_emit_clip_rect( drm_radeon_private_t *dev_priv, + drm_clip_rect_t *box ) +{ + RING_LOCALS; + + DRM_DEBUG( " box: x1=%d y1=%d x2=%d y2=%d\n", + box->x1, box->y1, box->x2, box->y2 ); + + BEGIN_RING( 4 ); + + OUT_RING( CP_PACKET0( RADEON_RE_TOP_LEFT, 0 ) ); + OUT_RING( (box->y1 << 16) | box->x1 ); + + OUT_RING( CP_PACKET0( RADEON_RE_WIDTH_HEIGHT, 0 ) ); + OUT_RING( ((box->y2 - 1) << 16) | (box->x2 - 1) ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_context( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG( " %s\n", __FUNCTION__ ); + + BEGIN_RING( 14 ); + + OUT_RING( CP_PACKET0( RADEON_PP_MISC, 6 ) ); + OUT_RING( ctx->pp_misc ); + OUT_RING( ctx->pp_fog_color ); + OUT_RING( ctx->re_solid_color ); + OUT_RING( ctx->rb3d_blendcntl ); + OUT_RING( ctx->rb3d_depthoffset ); + OUT_RING( ctx->rb3d_depthpitch ); + OUT_RING( ctx->rb3d_zstencilcntl ); + + OUT_RING( CP_PACKET0( RADEON_PP_CNTL, 2 ) ); + OUT_RING( ctx->pp_cntl ); + OUT_RING( ctx->rb3d_cntl ); + OUT_RING( ctx->rb3d_coloroffset ); + + OUT_RING( CP_PACKET0( RADEON_RB3D_COLORPITCH, 0 ) ); + OUT_RING( ctx->rb3d_colorpitch ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_vertfmt( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG( " %s\n", __FUNCTION__ ); + + BEGIN_RING( 2 ); + + OUT_RING( CP_PACKET0( RADEON_SE_COORD_FMT, 0 ) ); + OUT_RING( ctx->se_coord_fmt ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_line( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG( " %s\n", __FUNCTION__ ); + + BEGIN_RING( 5 ); + + OUT_RING( CP_PACKET0( RADEON_RE_LINE_PATTERN, 1 ) ); + OUT_RING( ctx->re_line_pattern ); + OUT_RING( ctx->re_line_state ); + + OUT_RING( CP_PACKET0( RADEON_SE_LINE_WIDTH, 0 ) ); + OUT_RING( ctx->se_line_width ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_bumpmap( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG( " %s\n", __FUNCTION__ ); + + BEGIN_RING( 5 ); + + OUT_RING( CP_PACKET0( RADEON_PP_LUM_MATRIX, 0 ) ); + OUT_RING( ctx->pp_lum_matrix ); + + OUT_RING( CP_PACKET0( RADEON_PP_ROT_MATRIX_0, 1 ) ); + OUT_RING( ctx->pp_rot_matrix_0 ); + OUT_RING( ctx->pp_rot_matrix_1 ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_masks( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG( " %s\n", __FUNCTION__ ); + + BEGIN_RING( 4 ); + + OUT_RING( CP_PACKET0( RADEON_RB3D_STENCILREFMASK, 2 ) ); + OUT_RING( ctx->rb3d_stencilrefmask ); + OUT_RING( ctx->rb3d_ropcntl ); + OUT_RING( ctx->rb3d_planemask ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_viewport( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG( " %s\n", __FUNCTION__ ); + + BEGIN_RING( 7 ); + + OUT_RING( CP_PACKET0( RADEON_SE_VPORT_XSCALE, 5 ) ); + OUT_RING( ctx->se_vport_xscale ); + OUT_RING( ctx->se_vport_xoffset ); + OUT_RING( ctx->se_vport_yscale ); + OUT_RING( ctx->se_vport_yoffset ); + OUT_RING( ctx->se_vport_zscale ); + OUT_RING( ctx->se_vport_zoffset ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_setup( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG( " %s\n", __FUNCTION__ ); + + BEGIN_RING( 4 ); + + OUT_RING( CP_PACKET0( RADEON_SE_CNTL, 0 ) ); + OUT_RING( ctx->se_cntl ); + OUT_RING( CP_PACKET0( RADEON_SE_CNTL_STATUS, 0 ) ); + OUT_RING( ctx->se_cntl_status ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_tcl( drm_radeon_private_t *dev_priv ) +{ +#ifdef TCL_ENABLE + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG( " %s\n", __FUNCTION__ ); + + BEGIN_RING( 29 ); + + OUT_RING( CP_PACKET0( RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED, 27 ) ); + OUT_RING( ctx->se_tcl_material_emmissive.red ); + OUT_RING( ctx->se_tcl_material_emmissive.green ); + OUT_RING( ctx->se_tcl_material_emmissive.blue ); + OUT_RING( ctx->se_tcl_material_emmissive.alpha ); + OUT_RING( ctx->se_tcl_material_ambient.red ); + OUT_RING( ctx->se_tcl_material_ambient.green ); + OUT_RING( ctx->se_tcl_material_ambient.blue ); + OUT_RING( ctx->se_tcl_material_ambient.alpha ); + OUT_RING( ctx->se_tcl_material_diffuse.red ); + OUT_RING( ctx->se_tcl_material_diffuse.green ); + OUT_RING( ctx->se_tcl_material_diffuse.blue ); + OUT_RING( ctx->se_tcl_material_diffuse.alpha ); + OUT_RING( ctx->se_tcl_material_specular.red ); + OUT_RING( ctx->se_tcl_material_specular.green ); + OUT_RING( ctx->se_tcl_material_specular.blue ); + OUT_RING( ctx->se_tcl_material_specular.alpha ); + OUT_RING( ctx->se_tcl_shininess ); + OUT_RING( ctx->se_tcl_output_vtx_fmt ); + OUT_RING( ctx->se_tcl_output_vtx_sel ); + OUT_RING( ctx->se_tcl_matrix_select_0 ); + OUT_RING( ctx->se_tcl_matrix_select_1 ); + OUT_RING( ctx->se_tcl_ucp_vert_blend_ctl ); + OUT_RING( ctx->se_tcl_texture_proc_ctl ); + OUT_RING( ctx->se_tcl_light_model_ctl ); + for ( i = 0 ; i < 4 ; i++ ) { + OUT_RING( ctx->se_tcl_per_light_ctl[i] ); + } + + ADVANCE_RING(); +#else + DRM_ERROR( "TCL not enabled!\n" ); +#endif +} + +static inline void radeon_emit_misc( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_context_regs_t *ctx = &sarea_priv->context_state; + RING_LOCALS; + DRM_DEBUG( " %s\n", __FUNCTION__ ); + + BEGIN_RING( 2 ); + + OUT_RING( CP_PACKET0( RADEON_RE_MISC, 0 ) ); + OUT_RING( ctx->re_misc ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_tex0( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_texture_regs_t *tex = &sarea_priv->tex_state[0]; + RING_LOCALS; + DRM_DEBUG( " %s: offset=0x%x\n", __FUNCTION__, tex->pp_txoffset ); + + BEGIN_RING( 9 ); + + OUT_RING( CP_PACKET0( RADEON_PP_TXFILTER_0, 5 ) ); + OUT_RING( tex->pp_txfilter ); + OUT_RING( tex->pp_txformat ); + OUT_RING( tex->pp_txoffset ); + OUT_RING( tex->pp_txcblend ); + OUT_RING( tex->pp_txablend ); + OUT_RING( tex->pp_tfactor ); + + OUT_RING( CP_PACKET0( RADEON_PP_BORDER_COLOR_0, 0 ) ); + OUT_RING( tex->pp_border_color ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_tex1( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_texture_regs_t *tex = &sarea_priv->tex_state[1]; + RING_LOCALS; + DRM_DEBUG( " %s: offset=0x%x\n", __FUNCTION__, tex->pp_txoffset ); + + BEGIN_RING( 9 ); + + OUT_RING( CP_PACKET0( RADEON_PP_TXFILTER_1, 5 ) ); + OUT_RING( tex->pp_txfilter ); + OUT_RING( tex->pp_txformat ); + OUT_RING( tex->pp_txoffset ); + OUT_RING( tex->pp_txcblend ); + OUT_RING( tex->pp_txablend ); + OUT_RING( tex->pp_tfactor ); + + OUT_RING( CP_PACKET0( RADEON_PP_BORDER_COLOR_1, 0 ) ); + OUT_RING( tex->pp_border_color ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_tex2( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_texture_regs_t *tex = &sarea_priv->tex_state[2]; + RING_LOCALS; + DRM_DEBUG( " %s\n", __FUNCTION__ ); + + BEGIN_RING( 9 ); + + OUT_RING( CP_PACKET0( RADEON_PP_TXFILTER_2, 5 ) ); + OUT_RING( tex->pp_txfilter ); + OUT_RING( tex->pp_txformat ); + OUT_RING( tex->pp_txoffset ); + OUT_RING( tex->pp_txcblend ); + OUT_RING( tex->pp_txablend ); + OUT_RING( tex->pp_tfactor ); + + OUT_RING( CP_PACKET0( RADEON_PP_BORDER_COLOR_2, 0 ) ); + OUT_RING( tex->pp_border_color ); + + ADVANCE_RING(); +} + +static inline void radeon_emit_state( drm_radeon_private_t *dev_priv ) +{ + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + unsigned int dirty = sarea_priv->dirty; + + DRM_DEBUG( "%s: dirty=0x%08x\n", __FUNCTION__, dirty ); + + if ( dirty & RADEON_UPLOAD_CONTEXT ) { + radeon_emit_context( dev_priv ); + sarea_priv->dirty &= ~RADEON_UPLOAD_CONTEXT; + } + + if ( dirty & RADEON_UPLOAD_VERTFMT ) { + radeon_emit_vertfmt( dev_priv ); + sarea_priv->dirty &= ~RADEON_UPLOAD_VERTFMT; + } + + if ( dirty & RADEON_UPLOAD_LINE ) { + radeon_emit_line( dev_priv ); + sarea_priv->dirty &= ~RADEON_UPLOAD_LINE; + } + + if ( dirty & RADEON_UPLOAD_BUMPMAP ) { + radeon_emit_bumpmap( dev_priv ); + sarea_priv->dirty &= ~RADEON_UPLOAD_BUMPMAP; + } + + if ( dirty & RADEON_UPLOAD_MASKS ) { + radeon_emit_masks( dev_priv ); + sarea_priv->dirty &= ~RADEON_UPLOAD_MASKS; + } + + if ( dirty & RADEON_UPLOAD_VIEWPORT ) { + radeon_emit_viewport( dev_priv ); + sarea_priv->dirty &= ~RADEON_UPLOAD_VIEWPORT; + } + + if ( dirty & RADEON_UPLOAD_SETUP ) { + radeon_emit_setup( dev_priv ); + sarea_priv->dirty &= ~RADEON_UPLOAD_SETUP; + } + + if ( dirty & RADEON_UPLOAD_TCL ) { +#ifdef TCL_ENABLE + radeon_emit_tcl( dev_priv ); +#endif + sarea_priv->dirty &= ~RADEON_UPLOAD_TCL; + } + + if ( dirty & RADEON_UPLOAD_MISC ) { + radeon_emit_misc( dev_priv ); + sarea_priv->dirty &= ~RADEON_UPLOAD_MISC; + } + + if ( dirty & RADEON_UPLOAD_TEX0 ) { + radeon_emit_tex0( dev_priv ); + sarea_priv->dirty &= ~RADEON_UPLOAD_TEX0; + } + + if ( dirty & RADEON_UPLOAD_TEX1 ) { + radeon_emit_tex1( dev_priv ); + sarea_priv->dirty &= ~RADEON_UPLOAD_TEX1; + } + + if ( dirty & RADEON_UPLOAD_TEX2 ) { +#if 0 + radeon_emit_tex2( dev_priv ); +#endif + sarea_priv->dirty &= ~RADEON_UPLOAD_TEX2; + } + + sarea_priv->dirty &= ~(RADEON_UPLOAD_TEX0IMAGES | + RADEON_UPLOAD_TEX1IMAGES | + RADEON_UPLOAD_TEX2IMAGES | + RADEON_REQUIRE_QUIESCENCE); +} + + +#if RADEON_PERFORMANCE_BOXES +/* ================================================================ + * Performance monitoring functions + */ + +static void radeon_clear_box( drm_radeon_private_t *dev_priv, + int x, int y, int w, int h, + int r, int g, int b ) +{ + u32 pitch, offset; + u32 color; + RING_LOCALS; + + switch ( dev_priv->color_fmt ) { + case RADEON_COLOR_FORMAT_RGB565: + color = (((r & 0xf8) << 8) | + ((g & 0xfc) << 3) | + ((b & 0xf8) >> 3)); + break; + case RADEON_COLOR_FORMAT_ARGB8888: + default: + color = (((0xff) << 24) | (r << 16) | (g << 8) | b); + break; + } + + offset = dev_priv->back_offset; + pitch = dev_priv->back_pitch >> 3; + + BEGIN_RING( 6 ); + + OUT_RING( CP_PACKET3( RADEON_CNTL_PAINT_MULTI, 4 ) ); + OUT_RING( RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_SOLID_COLOR | + (dev_priv->color_fmt << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_P | + RADEON_GMC_CLR_CMP_CNTL_DIS ); + + OUT_RING( (pitch << 22) | (offset >> 5) ); + OUT_RING( color ); + + OUT_RING( (x << 16) | y ); + OUT_RING( (w << 16) | h ); + + ADVANCE_RING(); +} + +static void radeon_cp_performance_boxes( drm_radeon_private_t *dev_priv ) +{ + if ( atomic_read( &dev_priv->idle_count ) == 0 ) { + radeon_clear_box( dev_priv, 64, 4, 8, 8, 0, 255, 0 ); + } else { + atomic_set( &dev_priv->idle_count, 0 ); + } +} + +#endif + + +/* ================================================================ + * CP command dispatch functions + */ + +static void radeon_print_dirty( const char *msg, unsigned int flags ) +{ + DRM_DEBUG( "%s: (0x%x) %s%s%s%s%s%s%s%s%s%s%s%s%s%s\n", + msg, + flags, + (flags & RADEON_UPLOAD_CONTEXT) ? "context, " : "", + (flags & RADEON_UPLOAD_VERTFMT) ? "vertfmt, " : "", + (flags & RADEON_UPLOAD_LINE) ? "line, " : "", + (flags & RADEON_UPLOAD_BUMPMAP) ? "bumpmap, " : "", + (flags & RADEON_UPLOAD_MASKS) ? "masks, " : "", + (flags & RADEON_UPLOAD_VIEWPORT) ? "viewport, " : "", + (flags & RADEON_UPLOAD_SETUP) ? "setup, " : "", + (flags & RADEON_UPLOAD_TCL) ? "tcl, " : "", + (flags & RADEON_UPLOAD_MISC) ? "misc, " : "", + (flags & RADEON_UPLOAD_TEX0) ? "tex0, " : "", + (flags & RADEON_UPLOAD_TEX1) ? "tex1, " : "", + (flags & RADEON_UPLOAD_TEX2) ? "tex2, " : "", + (flags & RADEON_UPLOAD_CLIPRECTS) ? "cliprects, " : "", + (flags & RADEON_REQUIRE_QUIESCENCE) ? "quiescence, " : "" ); +} + +static void radeon_cp_dispatch_clear( drm_device_t *dev, + drm_radeon_clear_t *clear ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + int nbox = sarea_priv->nbox; + drm_clip_rect_t *pbox = sarea_priv->boxes; + unsigned int flags = clear->flags; + int i; + RING_LOCALS; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + radeon_update_ring_snapshot( dev_priv ); + + if ( dev_priv->page_flipping && dev_priv->current_page == 1 ) { + unsigned int tmp = flags; + + flags &= ~(RADEON_FRONT | RADEON_BACK); + if ( tmp & RADEON_FRONT ) flags |= RADEON_BACK; + if ( tmp & RADEON_BACK ) flags |= RADEON_FRONT; + } + + for ( i = 0 ; i < nbox ; i++ ) { + int x = pbox[i].x1; + int y = pbox[i].y1; + int w = pbox[i].x2 - x; + int h = pbox[i].y2 - y; + + DRM_DEBUG( "dispatch clear %d,%d-%d,%d flags 0x%x\n", + x, y, w, h, flags ); + + if ( flags & (RADEON_FRONT | RADEON_BACK) ) { + BEGIN_RING( 4 ); + + /* Ensure the 3D stream is idle before doing a + * 2D fill to clear the front or back buffer. + */ + RADEON_WAIT_UNTIL_3D_IDLE(); + + OUT_RING( CP_PACKET0( RADEON_DP_WRITE_MASK, 0 ) ); + OUT_RING( sarea_priv->context_state.rb3d_planemask ); + + ADVANCE_RING(); + + /* Make sure we restore the 3D state next time. + */ + dev_priv->sarea_priv->dirty |= (RADEON_UPLOAD_CONTEXT | + RADEON_UPLOAD_MASKS); + } + + if ( flags & RADEON_FRONT ) { + BEGIN_RING( 6 ); + + OUT_RING( CP_PACKET3( RADEON_CNTL_PAINT_MULTI, 4 ) ); + OUT_RING( RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_SOLID_COLOR | + (dev_priv->color_fmt << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_P | + RADEON_GMC_CLR_CMP_CNTL_DIS ); + + OUT_RING( dev_priv->front_pitch_offset ); + OUT_RING( clear->clear_color ); + + OUT_RING( (x << 16) | y ); + OUT_RING( (w << 16) | h ); + + ADVANCE_RING(); + } + + if ( flags & RADEON_BACK ) { + BEGIN_RING( 6 ); + + OUT_RING( CP_PACKET3( RADEON_CNTL_PAINT_MULTI, 4 ) ); + OUT_RING( RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_SOLID_COLOR | + (dev_priv->color_fmt << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_P | + RADEON_GMC_CLR_CMP_CNTL_DIS ); + + OUT_RING( dev_priv->back_pitch_offset ); + OUT_RING( clear->clear_color ); + + OUT_RING( (x << 16) | y ); + OUT_RING( (w << 16) | h ); + + ADVANCE_RING(); + + } + + if ( flags & RADEON_DEPTH ) { + drm_radeon_depth_clear_t *depth_clear = + &dev_priv->depth_clear; + + if ( sarea_priv->dirty & ~RADEON_UPLOAD_CLIPRECTS ) { + radeon_emit_state( dev_priv ); + } + + /* FIXME: Render a rectangle to clear the depth + * buffer. So much for those "fast Z clears"... + */ + BEGIN_RING( 23 ); + + RADEON_WAIT_UNTIL_2D_IDLE(); + + OUT_RING( CP_PACKET0( RADEON_PP_CNTL, 1 ) ); + OUT_RING( 0x00000000 ); + OUT_RING( depth_clear->rb3d_cntl ); + OUT_RING( CP_PACKET0( RADEON_RB3D_ZSTENCILCNTL, 0 ) ); + OUT_RING( depth_clear->rb3d_zstencilcntl ); + OUT_RING( CP_PACKET0( RADEON_RB3D_PLANEMASK, 0 ) ); + OUT_RING( 0x00000000 ); + OUT_RING( CP_PACKET0( RADEON_SE_CNTL, 0 ) ); + OUT_RING( depth_clear->se_cntl ); + + OUT_RING( CP_PACKET3( RADEON_3D_DRAW_IMMD, 10 ) ); + OUT_RING( RADEON_VTX_Z_PRESENT ); + OUT_RING( (RADEON_PRIM_TYPE_RECT_LIST | + RADEON_PRIM_WALK_RING | + RADEON_MAOS_ENABLE | + RADEON_VTX_FMT_RADEON_MODE | + (3 << RADEON_NUM_VERTICES_SHIFT)) ); + + OUT_RING( clear->rect.ui[CLEAR_X1] ); + OUT_RING( clear->rect.ui[CLEAR_Y1] ); + OUT_RING( clear->rect.ui[CLEAR_DEPTH] ); + + OUT_RING( clear->rect.ui[CLEAR_X1] ); + OUT_RING( clear->rect.ui[CLEAR_Y2] ); + OUT_RING( clear->rect.ui[CLEAR_DEPTH] ); + + OUT_RING( clear->rect.ui[CLEAR_X2] ); + OUT_RING( clear->rect.ui[CLEAR_Y2] ); + OUT_RING( clear->rect.ui[CLEAR_DEPTH] ); + + ADVANCE_RING(); + + /* Make sure we restore the 3D state next time. + */ + dev_priv->sarea_priv->dirty |= (RADEON_UPLOAD_CONTEXT | + RADEON_UPLOAD_SETUP | + RADEON_UPLOAD_MASKS); + } + } + + /* Increment the clear counter. The client-side 3D driver must + * wait on this value before performing the clear ioctl. We + * need this because the card's so damned fast... + */ + dev_priv->sarea_priv->last_clear++; + + BEGIN_RING( 4 ); + + RADEON_CLEAR_AGE( dev_priv->sarea_priv->last_clear ); + RADEON_WAIT_UNTIL_IDLE(); + + ADVANCE_RING(); +} + +static void radeon_cp_dispatch_swap( drm_device_t *dev ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + int nbox = sarea_priv->nbox; + drm_clip_rect_t *pbox = sarea_priv->boxes; + int i; + RING_LOCALS; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + radeon_update_ring_snapshot( dev_priv ); + +#if RADEON_PERFORMANCE_BOXES + /* Do some trivial performance monitoring... + */ + radeon_cp_performance_boxes( dev_priv ); +#endif + + /* Wait for the 3D stream to idle before dispatching the bitblt. + * This will prevent data corruption between the two streams. + */ + BEGIN_RING( 2 ); + + RADEON_WAIT_UNTIL_3D_IDLE(); + + ADVANCE_RING(); + + for ( i = 0 ; i < nbox ; i++ ) { + int x = pbox[i].x1; + int y = pbox[i].y1; + int w = pbox[i].x2 - x; + int h = pbox[i].y2 - y; + + DRM_DEBUG( "dispatch swap %d,%d-%d,%d\n", + x, y, w, h ); + + BEGIN_RING( 7 ); + + OUT_RING( CP_PACKET3( RADEON_CNTL_BITBLT_MULTI, 5 ) ); + OUT_RING( RADEON_GMC_SRC_PITCH_OFFSET_CNTL | + RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_NONE | + (dev_priv->color_fmt << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_S | + RADEON_DP_SRC_SOURCE_MEMORY | + RADEON_GMC_CLR_CMP_CNTL_DIS | + RADEON_GMC_WR_MSK_DIS ); + + OUT_RING( dev_priv->back_pitch_offset ); + OUT_RING( dev_priv->front_pitch_offset ); + + OUT_RING( (x << 16) | y ); + OUT_RING( (x << 16) | y ); + OUT_RING( (w << 16) | h ); + + ADVANCE_RING(); + } + + /* Increment the frame counter. The client-side 3D driver must + * throttle the framerate by waiting for this value before + * performing the swapbuffer ioctl. + */ + dev_priv->sarea_priv->last_frame++; + + BEGIN_RING( 4 ); + + RADEON_FRAME_AGE( dev_priv->sarea_priv->last_frame ); + RADEON_WAIT_UNTIL_2D_IDLE(); + + ADVANCE_RING(); +} + +static void radeon_cp_dispatch_flip( drm_device_t *dev ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + RING_LOCALS; + DRM_DEBUG( "%s: page=%d\n", __FUNCTION__, dev_priv->current_page ); + + radeon_update_ring_snapshot( dev_priv ); + +#if RADEON_PERFORMANCE_BOXES + /* Do some trivial performance monitoring... + */ + radeon_cp_performance_boxes( dev_priv ); +#endif + + BEGIN_RING( 6 ); + + RADEON_WAIT_UNTIL_3D_IDLE(); + RADEON_WAIT_UNTIL_PAGE_FLIPPED(); + + OUT_RING( CP_PACKET0( RADEON_CRTC_OFFSET, 0 ) ); + + if ( dev_priv->current_page == 0 ) { + OUT_RING( dev_priv->back_offset ); + dev_priv->current_page = 1; + } else { + OUT_RING( dev_priv->front_offset ); + dev_priv->current_page = 0; + } + + ADVANCE_RING(); + + /* Increment the frame counter. The client-side 3D driver must + * throttle the framerate by waiting for this value before + * performing the swapbuffer ioctl. + */ + dev_priv->sarea_priv->last_frame++; + + BEGIN_RING( 2 ); + + RADEON_FRAME_AGE( dev_priv->sarea_priv->last_frame ); + + ADVANCE_RING(); +} + +static void radeon_cp_dispatch_vertex( drm_device_t *dev, + drm_buf_t *buf ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_buf_priv_t *buf_priv = buf->dev_private; + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + int format = sarea_priv->vc_format; + int offset = dev_priv->agp_buffers_offset + buf->offset; + int size = buf->used; + int prim = buf_priv->prim; + int i = 0; + RING_LOCALS; + DRM_DEBUG( "%s: nbox=%d\n", __FUNCTION__, sarea_priv->nbox ); + + radeon_update_ring_snapshot( dev_priv ); + + if ( 0 ) + radeon_print_dirty( "dispatch_vertex", sarea_priv->dirty ); + + if ( buf->used ) { + buf_priv->dispatched = 1; + + if ( sarea_priv->dirty & ~RADEON_UPLOAD_CLIPRECTS ) { + radeon_emit_state( dev_priv ); + } + + do { + /* Emit the next set of up to three cliprects */ + if ( i < sarea_priv->nbox ) { + radeon_emit_clip_rect( dev_priv, + &sarea_priv->boxes[i] ); + } + + /* Emit the vertex buffer rendering commands */ + BEGIN_RING( 5 ); + + OUT_RING( CP_PACKET3( RADEON_3D_RNDR_GEN_INDX_PRIM, 3 ) ); + OUT_RING( offset ); + OUT_RING( size ); + OUT_RING( format ); + OUT_RING( prim | RADEON_PRIM_WALK_LIST | + RADEON_COLOR_ORDER_RGBA | + RADEON_VTX_FMT_RADEON_MODE | + (size << RADEON_NUM_VERTICES_SHIFT) ); + + ADVANCE_RING(); + + i++; + } while ( i < sarea_priv->nbox ); + } + + if ( buf_priv->discard ) { + buf_priv->age = dev_priv->sarea_priv->last_dispatch; + + /* Emit the vertex buffer age */ + BEGIN_RING( 2 ); + RADEON_DISPATCH_AGE( buf_priv->age ); + ADVANCE_RING(); + + buf->pending = 1; + buf->used = 0; + /* FIXME: Check dispatched field */ + buf_priv->dispatched = 0; + } + + dev_priv->sarea_priv->last_dispatch++; + + sarea_priv->dirty &= ~RADEON_UPLOAD_CLIPRECTS; + sarea_priv->nbox = 0; +} + + +static void radeon_cp_dispatch_indirect( drm_device_t *dev, + drm_buf_t *buf, + int start, int end ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_buf_priv_t *buf_priv = buf->dev_private; + RING_LOCALS; + DRM_DEBUG( "indirect: buf=%d s=0x%x e=0x%x\n", + buf->idx, start, end ); + + radeon_update_ring_snapshot( dev_priv ); + + if ( start != end ) { + int offset = (dev_priv->agp_buffers_offset + + buf->offset + start); + int dwords = (end - start + 3) / sizeof(u32); + + /* Indirect buffer data must be an even number of + * dwords, so if we've been given an odd number we must + * pad the data with a Type-2 CP packet. + */ + if ( dwords & 1 ) { + u32 *data = (u32 *) + ((char *)dev_priv->buffers->handle + + buf->offset + start); + data[dwords++] = RADEON_CP_PACKET2; + } + + buf_priv->dispatched = 1; + + /* Fire off the indirect buffer */ + BEGIN_RING( 3 ); + + OUT_RING( CP_PACKET0( RADEON_CP_IB_BASE, 1 ) ); + OUT_RING( offset ); + OUT_RING( dwords ); + + ADVANCE_RING(); + } + + if ( buf_priv->discard ) { + buf_priv->age = dev_priv->sarea_priv->last_dispatch; + + /* Emit the indirect buffer age */ + BEGIN_RING( 2 ); + RADEON_DISPATCH_AGE( buf_priv->age ); + ADVANCE_RING(); + + buf->pending = 1; + buf->used = 0; + /* FIXME: Check dispatched field */ + buf_priv->dispatched = 0; + } + + dev_priv->sarea_priv->last_dispatch++; +} + +static void radeon_cp_dispatch_indices( drm_device_t *dev, + drm_buf_t *buf, + int start, int end, + int count ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_buf_priv_t *buf_priv = buf->dev_private; + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + int format = sarea_priv->vc_format; + int offset = dev_priv->agp_buffers_offset; + int prim = buf_priv->prim; + u32 *data; + int dwords; + int i = 0; + RING_LOCALS; + DRM_DEBUG( "indices: s=%d e=%d c=%d\n", start, end, count ); + + radeon_update_ring_snapshot( dev_priv ); + + if ( 0 ) + radeon_print_dirty( "dispatch_indices", sarea_priv->dirty ); + + if ( start != end ) { + buf_priv->dispatched = 1; + + if ( sarea_priv->dirty & ~RADEON_UPLOAD_CLIPRECTS ) { + radeon_emit_state( dev_priv ); + } + + dwords = (end - start + 3) / sizeof(u32); + + data = (u32 *)((char *)dev_priv->buffers->handle + + buf->offset + start); + + data[0] = CP_PACKET3( RADEON_3D_RNDR_GEN_INDX_PRIM, dwords-2 ); + + data[1] = offset; + data[2] = RADEON_MAX_VB_VERTS; + data[3] = format; + data[4] = (prim | RADEON_PRIM_WALK_IND | + RADEON_COLOR_ORDER_RGBA | + RADEON_VTX_FMT_RADEON_MODE | + (count << RADEON_NUM_VERTICES_SHIFT) ); + + if ( count & 0x1 ) { + data[dwords-1] &= 0x0000ffff; + } + + do { + /* Emit the next set of up to three cliprects */ + if ( i < sarea_priv->nbox ) { + radeon_emit_clip_rect( dev_priv, + &sarea_priv->boxes[i] ); + } + + radeon_cp_dispatch_indirect( dev, buf, start, end ); + + i++; + } while ( i < sarea_priv->nbox ); + } + + if ( buf_priv->discard ) { + buf_priv->age = dev_priv->sarea_priv->last_dispatch; + + /* Emit the vertex buffer age */ + BEGIN_RING( 2 ); + RADEON_DISPATCH_AGE( buf_priv->age ); + ADVANCE_RING(); + + buf->pending = 1; + /* FIXME: Check dispatched field */ + buf_priv->dispatched = 0; + } + + dev_priv->sarea_priv->last_dispatch++; + + sarea_priv->dirty &= ~RADEON_UPLOAD_CLIPRECTS; + sarea_priv->nbox = 0; +} + +static int radeon_cp_dispatch_blit( drm_device_t *dev, + drm_radeon_blit_t *blit ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_device_dma_t *dma = dev->dma; + drm_buf_t *buf; + drm_radeon_buf_priv_t *buf_priv; + u32 format; + u32 *data; + int dword_shift, dwords; + RING_LOCALS; + DRM_DEBUG( "blit: ofs=0x%x p=%d f=%d x=%hd y=%hd w=%hd h=%hd\n", + blit->offset >> 10, blit->pitch, blit->format, + blit->x, blit->y, blit->width, blit->height ); + + radeon_update_ring_snapshot( dev_priv ); + + /* The compiler won't optimize away a division by a variable, + * even if the only legal values are powers of two. Thus, we'll + * use a shift instead. + */ + switch ( blit->format ) { + case RADEON_TXF_32BPP_ARGB8888: + case RADEON_TXF_32BPP_RGBA8888: + format = RADEON_COLOR_FORMAT_ARGB8888; + dword_shift = 0; + break; + case RADEON_TXF_16BPP_AI88: + case RADEON_TXF_16BPP_ARGB1555: + case RADEON_TXF_16BPP_RGB565: + case RADEON_TXF_16BPP_ARGB4444: + format = RADEON_COLOR_FORMAT_RGB565; + dword_shift = 1; + break; + case RADEON_TXF_8BPP_I: + case RADEON_TXF_8BPP_RGB332: + format = RADEON_COLOR_FORMAT_CI8; + dword_shift = 2; + break; + default: + DRM_ERROR( "invalid blit format %d\n", blit->format ); + return -EINVAL; + } + + /* Flush the pixel cache. This ensures no pixel data gets mixed + * up with the texture data from the host data blit, otherwise + * part of the texture image may be corrupted. + */ + BEGIN_RING( 4 ); + + RADEON_FLUSH_CACHE(); + RADEON_WAIT_UNTIL_IDLE(); + + ADVANCE_RING(); + + /* Dispatch the indirect buffer. + */ + buf = dma->buflist[blit->idx]; + buf_priv = buf->dev_private; + + if ( buf->pid != current->pid ) { + DRM_ERROR( "process %d using buffer owned by %d\n", + current->pid, buf->pid ); + return -EINVAL; + } + if ( buf->pending ) { + DRM_ERROR( "sending pending buffer %d\n", blit->idx ); + return -EINVAL; + } + + buf_priv->discard = 1; + + dwords = (blit->width * blit->height) >> dword_shift; + if ( !dwords ) dwords = 1; + + data = (u32 *)((char *)dev_priv->buffers->handle + buf->offset); + + data[0] = CP_PACKET3( RADEON_CNTL_HOSTDATA_BLT, dwords + 6 ); + data[1] = (RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_NONE | + (format << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_S | + RADEON_DP_SRC_SOURCE_HOST_DATA | + RADEON_GMC_CLR_CMP_CNTL_DIS | + RADEON_GMC_WR_MSK_DIS); + + data[2] = (blit->pitch << 22) | (blit->offset >> 10); + data[3] = 0xffffffff; + data[4] = 0xffffffff; + data[5] = (blit->y << 16) | blit->x; + data[6] = (blit->height << 16) | blit->width; + data[7] = dwords; + + buf->used = (dwords + 8) * sizeof(u32); + + radeon_cp_dispatch_indirect( dev, buf, 0, buf->used ); + + /* Flush the pixel cache after the blit completes. This ensures + * the texture data is written out to memory before rendering + * continues. + */ + BEGIN_RING( 4 ); + + RADEON_FLUSH_CACHE(); + RADEON_WAIT_UNTIL_2D_IDLE(); + + ADVANCE_RING(); + + return 0; +} + +static void radeon_cp_dispatch_stipple( drm_device_t *dev, u32 *stipple ) +{ + drm_radeon_private_t *dev_priv = dev->dev_private; + int i; + RING_LOCALS; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + radeon_update_ring_snapshot( dev_priv ); + + BEGIN_RING( 35 ); + + OUT_RING( CP_PACKET0( RADEON_RE_STIPPLE_ADDR, 0 ) ); + OUT_RING( 0x00000000 ); + + OUT_RING( CP_PACKET0_TABLE( RADEON_RE_STIPPLE_DATA, 31 ) ); + for ( i = 0 ; i < 32 ; i++ ) { + OUT_RING( stipple[i] ); + } + + ADVANCE_RING(); +} + + +/* ================================================================ + * IOCTL functions + */ + +int radeon_cp_clear( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + drm_radeon_clear_t clear; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + + if ( copy_from_user( &clear, (drm_radeon_clear_t *) arg, + sizeof(clear) ) ) + return -EFAULT; + + if ( sarea_priv->nbox > RADEON_NR_SAREA_CLIPRECTS ) + sarea_priv->nbox = RADEON_NR_SAREA_CLIPRECTS; + + radeon_cp_dispatch_clear( dev, &clear ); + + return 0; +} + +int radeon_cp_swap( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_sarea_t *sarea_priv = dev_priv->sarea_priv; + DRM_DEBUG( "%s\n", __FUNCTION__ ); + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + + if ( sarea_priv->nbox > RADEON_NR_SAREA_CLIPRECTS ) + sarea_priv->nbox = RADEON_NR_SAREA_CLIPRECTS; + + if ( !dev_priv->page_flipping ) { + radeon_cp_dispatch_swap( dev ); + dev_priv->sarea_priv->dirty |= (RADEON_UPLOAD_CONTEXT | + RADEON_UPLOAD_MASKS); + } else { + radeon_cp_dispatch_flip( dev ); + } + + return 0; +} + +int radeon_cp_vertex( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_device_dma_t *dma = dev->dma; + drm_buf_t *buf; + drm_radeon_buf_priv_t *buf_priv; + drm_radeon_vertex_t vertex; + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + if ( !dev_priv || dev_priv->is_pci ) { + DRM_ERROR( "%s called with a PCI card\n", __FUNCTION__ ); + return -EINVAL; + } + + if ( copy_from_user( &vertex, (drm_radeon_vertex_t *)arg, + sizeof(vertex) ) ) + return -EFAULT; + + DRM_DEBUG( "%s: pid=%d index=%d count=%d discard=%d\n", + __FUNCTION__, current->pid, + vertex.idx, vertex.count, vertex.discard ); + + if ( vertex.idx < 0 || vertex.idx >= dma->buf_count ) { + DRM_ERROR( "buffer index %d (of %d max)\n", + vertex.idx, dma->buf_count - 1 ); + return -EINVAL; + } + if ( vertex.prim < 0 || + vertex.prim > RADEON_PRIM_TYPE_3VRT_LINE_LIST ) { + DRM_ERROR( "buffer prim %d\n", vertex.prim ); + return -EINVAL; + } + + VB_AGE_CHECK_WITH_RET( dev_priv ); + + buf = dma->buflist[vertex.idx]; + buf_priv = buf->dev_private; + + if ( buf->pid != current->pid ) { + DRM_ERROR( "process %d using buffer owned by %d\n", + current->pid, buf->pid ); + return -EINVAL; + } + if ( buf->pending ) { + DRM_ERROR( "sending pending buffer %d\n", vertex.idx ); + return -EINVAL; + } + + buf->used = vertex.count; + buf_priv->prim = vertex.prim; + buf_priv->discard = vertex.discard; + + radeon_cp_dispatch_vertex( dev, buf ); + + return 0; +} + +int radeon_cp_indices( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_device_dma_t *dma = dev->dma; + drm_buf_t *buf; + drm_radeon_buf_priv_t *buf_priv; + drm_radeon_indices_t elts; + int count; + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + if ( !dev_priv || dev_priv->is_pci ) { + DRM_ERROR( "%s called with a PCI card\n", __FUNCTION__ ); + return -EINVAL; + } + + if ( copy_from_user( &elts, (drm_radeon_indices_t *)arg, + sizeof(elts) ) ) + return -EFAULT; + + DRM_DEBUG( "%s: pid=%d index=%d start=%d end=%d discard=%d\n", + __FUNCTION__, current->pid, + elts.idx, elts.start, elts.end, elts.discard ); + + if ( elts.idx < 0 || elts.idx >= dma->buf_count ) { + DRM_ERROR( "buffer index %d (of %d max)\n", + elts.idx, dma->buf_count - 1 ); + return -EINVAL; + } + if ( elts.prim < 0 || + elts.prim > RADEON_PRIM_TYPE_3VRT_LINE_LIST ) { + DRM_ERROR( "buffer prim %d\n", elts.prim ); + return -EINVAL; + } + + VB_AGE_CHECK_WITH_RET( dev_priv ); + + buf = dma->buflist[elts.idx]; + buf_priv = buf->dev_private; + + if ( buf->pid != current->pid ) { + DRM_ERROR( "process %d using buffer owned by %d\n", + current->pid, buf->pid ); + return -EINVAL; + } + if ( buf->pending ) { + DRM_ERROR( "sending pending buffer %d\n", elts.idx ); + return -EINVAL; + } + + count = (elts.end - elts.start) / sizeof(u16); + elts.start -= RADEON_INDEX_PRIM_OFFSET; + + if ( elts.start & 0x7 ) { + DRM_ERROR( "misaligned buffer 0x%x\n", elts.start ); + return -EINVAL; + } + if ( elts.start < buf->used ) { + DRM_ERROR( "no header 0x%x - 0x%x\n", elts.start, buf->used ); + return -EINVAL; + } + + buf->used = elts.end; + buf_priv->prim = elts.prim; + buf_priv->discard = elts.discard; + + radeon_cp_dispatch_indices( dev, buf, elts.start, elts.end, count ); + + return 0; +} + +int radeon_cp_blit( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_device_dma_t *dma = dev->dma; + drm_radeon_blit_t blit; + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + + if ( copy_from_user( &blit, (drm_radeon_blit_t *)arg, + sizeof(blit) ) ) + return -EFAULT; + + DRM_DEBUG( "%s: pid=%d index=%d\n", + __FUNCTION__, current->pid, blit.idx ); + + if ( blit.idx < 0 || blit.idx > dma->buf_count ) { + DRM_ERROR( "sending %d buffers (of %d max)\n", + blit.idx, dma->buf_count ); + return -EINVAL; + } + + VB_AGE_CHECK_WITH_RET( dev_priv ); + + return radeon_cp_dispatch_blit( dev, &blit ); +} + +int radeon_cp_stipple( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_stipple_t stipple; + u32 mask[32]; + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + + if ( copy_from_user( &stipple, (drm_radeon_stipple_t *)arg, + sizeof(stipple) ) ) + return -EFAULT; + + if ( copy_from_user( &mask, stipple.mask, + 32 * sizeof(u32) ) ) + return -EFAULT; + + radeon_cp_dispatch_stipple( dev, mask ); + + return 0; +} + +int radeon_cp_indirect( struct inode *inode, struct file *filp, + unsigned int cmd, unsigned long arg ) +{ + drm_file_t *priv = filp->private_data; + drm_device_t *dev = priv->dev; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_device_dma_t *dma = dev->dma; + drm_buf_t *buf; + drm_radeon_buf_priv_t *buf_priv; + drm_radeon_indirect_t indirect; + RING_LOCALS; + + if ( !_DRM_LOCK_IS_HELD( dev->lock.hw_lock->lock ) || + dev->lock.pid != current->pid ) { + DRM_ERROR( "%s called without lock held\n", __FUNCTION__ ); + return -EINVAL; + } + if ( !dev_priv || dev_priv->is_pci ) { + DRM_ERROR( "%s called with a PCI card\n", __FUNCTION__ ); + return -EINVAL; + } + + if ( copy_from_user( &indirect, (drm_radeon_indirect_t *)arg, + sizeof(indirect) ) ) + return -EFAULT; + + DRM_DEBUG( "indirect: idx=%d s=%d e=%d d=%d\n", + indirect.idx, indirect.start, + indirect.end, indirect.discard ); + + if ( indirect.idx < 0 || indirect.idx >= dma->buf_count ) { + DRM_ERROR( "buffer index %d (of %d max)\n", + indirect.idx, dma->buf_count - 1 ); + return -EINVAL; + } + + buf = dma->buflist[indirect.idx]; + buf_priv = buf->dev_private; + + if ( buf->pid != current->pid ) { + DRM_ERROR( "process %d using buffer owned by %d\n", + current->pid, buf->pid ); + return -EINVAL; + } + if ( buf->pending ) { + DRM_ERROR( "sending pending buffer %d\n", indirect.idx ); + return -EINVAL; + } + + if ( indirect.start < buf->used ) { + DRM_ERROR( "reusing indirect: start=0x%x actual=0x%x\n", + indirect.start, buf->used ); + return -EINVAL; + } + + VB_AGE_CHECK_WITH_RET( dev_priv ); + + buf->used = indirect.end; + buf_priv->discard = indirect.discard; + + /* Wait for the 3D stream to idle before the indirect buffer + * containing 2D acceleration commands is processed. + */ + BEGIN_RING( 2 ); + + RADEON_WAIT_UNTIL_3D_IDLE(); + + ADVANCE_RING(); + + /* Dispatch the indirect buffer full of commands from the + * X server. This is insecure and is thus only available to + * privileged clients. + */ + radeon_cp_dispatch_indirect( dev, buf, indirect.start, indirect.end ); + + return 0; +} diff --git a/drivers/char/misc.c b/drivers/char/misc.c index 4c141d668..e4edb9f07 100644 --- a/drivers/char/misc.c +++ b/drivers/char/misc.c @@ -70,7 +70,6 @@ extern int psaux_init(void); extern void gfx_register(void); #endif extern void streamable_init(void); -extern int rtc_sun_init(void); /* Combines MK48T02 and MK48T08 */ extern int rtc_DP8570A_init(void); extern int rtc_MK48T08_init(void); extern int ds1286_init(void); @@ -259,9 +258,6 @@ int __init misc_init(void) #ifdef CONFIG_BVME6000 rtc_DP8570A_init(); #endif -#if defined(CONFIG_SUN_MOSTEK_RTC) - rtc_sun_init(); -#endif #ifdef CONFIG_SGI_DS1286 ds1286_init(); #endif diff --git a/drivers/char/n_tty.c b/drivers/char/n_tty.c index 6f4ebe44c..c2ae319af 100644 --- a/drivers/char/n_tty.c +++ b/drivers/char/n_tty.c @@ -86,6 +86,15 @@ static inline void free_buf(unsigned char *buf) free_page((unsigned long) buf); } +static inline void put_tty_queue_nolock(unsigned char c, struct tty_struct *tty) +{ + if (tty->read_cnt < N_TTY_BUF_SIZE) { + tty->read_buf[tty->read_head] = c; + tty->read_head = (tty->read_head + 1) & (N_TTY_BUF_SIZE-1); + tty->read_cnt++; + } +} + static inline void put_tty_queue(unsigned char c, struct tty_struct *tty) { unsigned long flags; @@ -94,11 +103,7 @@ static inline void put_tty_queue(unsigned char c, struct tty_struct *tty) * Why didn't anyone see this one comming? --AJK */ spin_lock_irqsave(&tty->read_lock, flags); - if (tty->read_cnt < N_TTY_BUF_SIZE) { - tty->read_buf[tty->read_head] = c; - tty->read_head = (tty->read_head + 1) & (N_TTY_BUF_SIZE-1); - tty->read_cnt++; - } + put_tty_queue_nolock(c, tty); spin_unlock_irqrestore(&tty->read_lock, flags); } @@ -499,6 +504,8 @@ static inline void n_tty_receive_parity_error(struct tty_struct *tty, static inline void n_tty_receive_char(struct tty_struct *tty, unsigned char c) { + unsigned long flags; + if (tty->raw) { put_tty_queue(c, tty); return; @@ -651,10 +658,12 @@ send_signal: put_tty_queue(c, tty); handle_newline: + spin_lock_irqsave(&tty->read_lock, flags); set_bit(tty->read_head, &tty->read_flags); - put_tty_queue(c, tty); + put_tty_queue_nolock(c, tty); tty->canon_head = tty->read_head; tty->canon_data++; + spin_unlock_irqrestore(&tty->read_lock, flags); kill_fasync(&tty->fasync, SIGIO, POLL_IN); if (waitqueue_active(&tty->read_wait)) wake_up_interruptible(&tty->read_wait); @@ -1055,12 +1064,6 @@ do_it_again: tty->read_tail = ((tty->read_tail+1) & (N_TTY_BUF_SIZE-1)); tty->read_cnt--; - spin_unlock_irqrestore(&tty->read_lock, flags); - - if (!eol || (c != __DISABLED_CHAR)) { - put_user(c, b++); - nr--; - } if (eol) { /* this test should be redundant: * we shouldn't be reading data if @@ -1068,8 +1071,15 @@ do_it_again: */ if (--tty->canon_data < 0) tty->canon_data = 0; - break; } + spin_unlock_irqrestore(&tty->read_lock, flags); + + if (!eol || (c != __DISABLED_CHAR)) { + put_user(c, b++); + nr--; + } + if (eol) + break; } } else { int uncopied; diff --git a/drivers/i2o/i2o_block.c b/drivers/i2o/i2o_block.c index 8389039d6..857b32a7e 100644 --- a/drivers/i2o/i2o_block.c +++ b/drivers/i2o/i2o_block.c @@ -392,7 +392,6 @@ static inline int i2ob_new_segment(request_queue_t *q, struct request *req, if (req->nr_segments < max_segments) { req->nr_segments++; - q->elevator.nr_segments++; return 1; } return 0; @@ -421,22 +420,16 @@ static int i2ob_merge_requests(request_queue_t *q, { int max_segments = i2ob_dev[MINOR(req->rq_dev)].max_segments; int total_segments = req->nr_segments + next->nr_segments; - int same_segment; if (__max_segments < max_segments) max_segments = __max_segments; - same_segment = 0; if (req->bhtail->b_data + req->bhtail->b_size == next->bh->b_data) - { total_segments--; - same_segment = 1; - } if (total_segments > max_segments) return 0; - q->elevator.nr_segments -= same_segment; req->nr_segments = total_segments; return 1; } diff --git a/drivers/ide/hpt366.c b/drivers/ide/hpt366.c index 62e2437e5..38cde0b40 100644 --- a/drivers/ide/hpt366.c +++ b/drivers/ide/hpt366.c @@ -55,6 +55,15 @@ const char *bad_ata100_5[] = { }; const char *bad_ata66_4[] = { + "IBM-DTLA-307075", + "IBM-DTLA-307060", + "IBM-DTLA-307045", + "IBM-DTLA-307030", + "IBM-DTLA-307020", + "IBM-DTLA-307015", + "IBM-DTLA-305040", + "IBM-DTLA-305030", + "IBM-DTLA-305020", "WDC AC310200R", NULL }; diff --git a/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c index 3a6a9e8d9..433951e95 100644 --- a/drivers/ide/ide-cd.c +++ b/drivers/ide/ide-cd.c @@ -288,6 +288,9 @@ * 4.59 Aug 11, 2000 - Fix changer problem in cdrom_read_toc, we weren't * correctly sensing a disc change. * - Rearranged some code + * - Use extended sense on drives that support it for + * correctly reporting tray status -- from + * Michael D Johnson <johnsom@orst.edu> * *************************************************************************/ @@ -759,16 +762,13 @@ static ide_startstop_t cdrom_start_packet_command(ide_drive_t *drive, * changed 5 parameters to 3 for dvd-ram * struct packet_command *pc; now packet_command_t *pc; */ -#undef CLASSIC_PACKET_STRUCT static ide_startstop_t cdrom_transfer_packet_command (ide_drive_t *drive, struct packet_command *pc, ide_handler_t *handler) { -#ifdef CLASSIC_PACKET_STRUCT unsigned char *cmd_buf = pc->c; int cmd_len = sizeof(pc->c); unsigned int timeout = pc->timeout; -#endif ide_startstop_t startstop; if (CDROM_CONFIG_FLAGS (drive)->drq_interrupt) { @@ -786,22 +786,10 @@ static ide_startstop_t cdrom_transfer_packet_command (ide_drive_t *drive, } /* Arm the interrupt handler. */ -#ifdef CLASSIC_PACKET_STRUCT - /* Arm the interrupt handler. */ ide_set_handler (drive, handler, timeout, cdrom_timer_expiry); /* Send the command to the device. */ atapi_output_bytes (drive, cmd_buf, cmd_len); -#else /* !CLASSIC_PACKET_STRUCT */ - /* Arm the interrupt handler. */ -// ide_set_handler (drive, handler, (unsigned int) pc->timeout, cdrom_timer_expiry); - ide_set_handler (drive, handler, pc->timeout, cdrom_timer_expiry); - - /* Send the command to the device. */ -// atapi_output_bytes (drive, (void *)pc->c, (unsigned int) sizeof(pc->c)); - atapi_output_bytes (drive, pc->c, sizeof(pc->c)); -#endif /* CLASSIC_PACKET_STRUCT */ - return ide_started; } @@ -1884,9 +1872,6 @@ static int cdrom_read_toc(ide_drive_t *drive, struct request_sense *sense) If it is, just return. */ (void) cdrom_check_status(drive, sense); - if (CDROM_STATE_FLAGS(drive)->toc_valid) - return 0; - /* First read just the header, so we know how long the TOC is. */ stat = cdrom_read_tocentry(drive, 0, 1, 0, (char *) &toc->hdr, sizeof(struct atapi_toc_header), sense); @@ -2324,11 +2309,17 @@ int ide_cdrom_drive_status (struct cdrom_device_info *cdi, int slot_nr) sense.ascq == 0x04) return CDS_DISC_OK; + + /* + * If not using Mt Fuji extended media tray reports, + * just return TRAY_OPEN since ATAPI doesn't provide + * any other way to detect this... + */ if (sense.sense_key == NOT_READY) { - /* ATAPI doesn't have anything that can help - us decide whether the drive is really - emtpy or the tray is just open. irk. */ - return CDS_TRAY_OPEN; + if (sense.asc == 0x3a && (!sense.ascq||sense.ascq == 1)) + return CDS_NO_DISC; + else + return CDS_TRAY_OPEN; } return CDS_DRIVE_NOT_READY; @@ -2597,7 +2588,7 @@ int ide_cdrom_probe_capabilities (ide_drive_t *drive) if (CDROM_CONFIG_FLAGS (drive)->dvd_r|CDROM_CONFIG_FLAGS (drive)->dvd_ram) printk (" DVD%s%s", (CDROM_CONFIG_FLAGS (drive)->dvd_r)? "-R" : "", - (CDROM_CONFIG_FLAGS (drive)->dvd_ram)? "AM" : ""); + (CDROM_CONFIG_FLAGS (drive)->dvd_ram)? "-RAM" : ""); if (CDROM_CONFIG_FLAGS (drive)->cd_r|CDROM_CONFIG_FLAGS (drive)->cd_rw) printk (" CD%s%s", diff --git a/drivers/ide/ide-dma.c b/drivers/ide/ide-dma.c index 7caafab9b..7158d3461 100644 --- a/drivers/ide/ide-dma.c +++ b/drivers/ide/ide-dma.c @@ -226,6 +226,9 @@ static int ide_build_sglist (ide_hwif_t *hwif, struct request *rq) unsigned char *virt_addr = bh->b_data; unsigned int size = bh->b_size; + if (nents >= PRD_ENTRIES) + return 0; + while ((bh = bh->b_reqnext) != NULL) { if ((virt_addr + size) != (unsigned char *) bh->b_data) break; @@ -259,6 +262,9 @@ int ide_build_dmatable (ide_drive_t *drive, ide_dma_action_t func) HWIF(drive)->sg_nents = i = ide_build_sglist(HWIF(drive), HWGROUP(drive)->rq); + if (!i) + return 0; + sg = HWIF(drive)->sg_table; while (i && sg_dma_len(sg)) { u32 cur_addr; @@ -274,7 +280,7 @@ int ide_build_dmatable (ide_drive_t *drive, ide_dma_action_t func) */ while (cur_len) { - if (++count >= PRD_ENTRIES) { + if (count++ >= PRD_ENTRIES) { printk("%s: DMA table too small\n", drive->name); pci_unmap_sg(HWIF(drive)->pci_dev, HWIF(drive)->sg_table, diff --git a/drivers/ide/ide-probe.c b/drivers/ide/ide-probe.c index decbc9f8d..64d9121c8 100644 --- a/drivers/ide/ide-probe.c +++ b/drivers/ide/ide-probe.c @@ -134,7 +134,7 @@ static inline void do_identify (ide_drive_t *drive, byte cmd) break; } #endif - printk ("CDROM"); + printk ("CD/DVD-ROM"); break; case ide_tape: printk ("TAPE"); @@ -761,9 +761,10 @@ static void init_gendisk (ide_hwif_t *hwif) for (unit = 0; unit < minors; ++unit) { *bs++ = BLOCK_SIZE; #ifdef CONFIG_BLK_DEV_PDC4030 - *max_sect++ = ((hwif->chipset == ide_pdc4030) ? 127 : MAX_SECTORS); + *max_sect++ = ((hwif->chipset == ide_pdc4030) ? 127 : 256); #else - *max_sect++ = MAX_SECTORS; + /* IDE can do up to 128K per request. */ + *max_sect++ = 256; #endif *max_ra++ = MAX_READAHEAD; } diff --git a/drivers/ide/via82cxxx.c b/drivers/ide/via82cxxx.c index 3ea1e261e..356e7a4cc 100644 --- a/drivers/ide/via82cxxx.c +++ b/drivers/ide/via82cxxx.c @@ -602,7 +602,9 @@ void __init ide_init_via82cxxx(ide_hwif_t *hwif) #ifdef CONFIG_BLK_DEV_IDEDMA if (hwif->dma_base) { hwif->dmaproc = &via82cxxx_dmaproc; +#ifdef CONFIG_IDEDMA_AUTO hwif->autodma = 1; +#endif } #endif /* CONFIG_BLK_DEV_IDEDMA */ } diff --git a/drivers/isdn/hisax/Makefile b/drivers/isdn/hisax/Makefile index 66f6b5d4a..72d990134 100644 --- a/drivers/isdn/hisax/Makefile +++ b/drivers/isdn/hisax/Makefile @@ -34,8 +34,8 @@ hisax-objs-$(CONFIG_HISAX_DIEHLDIVA) += diva.o isac.o arcofi.o hscx.o hisax-objs-$(CONFIG_HISAX_ASUSCOM) += asuscom.o isac.o arcofi.o hscx.o hisax-objs-$(CONFIG_HISAX_TELEINT) += teleint.o isac.o arcofi.o hfc_2bs0.o hisax-objs-$(CONFIG_HISAX_SEDLBAUER) += sedlbauer.o isac.o arcofi.o hscx.o isar.o -hisax-objs-$(CONFIG_HISAX_SPORTSTER) += sportster.o isac.o arcofi.o hfc_2bs0.o -hisax-objs-$(CONFIG_HISAX_MIC) += mic.o isac.o arcofi.o hfc_2bs0.o +hisax-objs-$(CONFIG_HISAX_SPORTSTER) += sportster.o isac.o arcofi.o hscx.o +hisax-objs-$(CONFIG_HISAX_MIC) += mic.o isac.o arcofi.o hscx.o hisax-objs-$(CONFIG_HISAX_NETJET) += nj_s.o netjet.o isac.o arcofi.o hisax-objs-$(CONFIG_HISAX_NETJET_U) += nj_u.o netjet.o icc.o hisax-objs-$(CONFIG_HISAX_HFCS) += hfcscard.o hfc_2bds0.o diff --git a/drivers/isdn/hisax/config.c b/drivers/isdn/hisax/config.c index 0b60e89b3..6d0366cf2 100644 --- a/drivers/isdn/hisax/config.c +++ b/drivers/isdn/hisax/config.c @@ -1708,8 +1708,8 @@ int __devinit hisax_init_pcmcia(void *pcm_iob, int *busy_flag, struct IsdnCard * } static struct pci_device_id hisax_pci_tbl[] __initdata = { -#ifdef CONFIG_HISAX_FRTIZPCI - {PCI_VENDOR_ID_AVM, PCI_DEVICE_ID_AVM_FRITZ, PCI_ANY_ID, PCI_ANY_ID}, +#ifdef CONFIG_HISAX_FRITZPCI + {PCI_VENDOR_ID_AVM, PCI_DEVICE_ID_AVM_A1, PCI_ANY_ID, PCI_ANY_ID}, #endif #ifdef CONFIG_HISAX_DIEHLDIVA {PCI_VENDOR_ID_EICON, PCI_DEVICE_ID_EICON_DIVA20, PCI_ANY_ID, PCI_ANY_ID}, diff --git a/drivers/isdn/hisax/isdnl3.c b/drivers/isdn/hisax/isdnl3.c index b484635f1..1c2ab1042 100644 --- a/drivers/isdn/hisax/isdnl3.c +++ b/drivers/isdn/hisax/isdnl3.c @@ -566,7 +566,7 @@ l3_msg(struct PStack *st, int pr, void *arg) } else { struct sk_buff *skb = arg; - skb_queue_head(&st->l3.squeue, skb); + skb_queue_tail(&st->l3.squeue, skb); FsmEvent(&st->l3.l3m, EV_ESTABLISH_REQ, NULL); } break; diff --git a/drivers/isdn/hisax/md5sums.asc b/drivers/isdn/hisax/md5sums.asc index 4cc1f8b83..5ffd61db0 100644 --- a/drivers/isdn/hisax/md5sums.asc +++ b/drivers/isdn/hisax/md5sums.asc @@ -10,7 +10,7 @@ ca7bd9bac39203f3074f3f093948cc3c isac.c a2ad619fd404b3149099a2984de9d23c isdnl1.c d2a78e407f3d94876deac160c6f9aae6 isdnl2.c -a109841c2e75b11fc8ef2c8718e24c3e isdnl3.c +e7932ca7ae39c497c17f13a2e1434fcd isdnl3.c afb5f2f4ac296d6de45c856993b161e1 tei.c 00023e2a482cb86a26ea870577ade5d6 callc.c a1834e9b2ec068440cff2e899eff4710 cert.c @@ -25,9 +25,9 @@ a1834e9b2ec068440cff2e899eff4710 cert.c Version: 2.6.3i Charset: noconv -iQCVAwUBOlMTgDpxHvX/mS9tAQFSbgP/W9y6tnnWHTRLGqyr3EY1OHZiQXERkAAu -hp+Y8PIoX1GgAh4yZ7xhYwUsk6y0z5USdGuhC9ZHh+oZd57lPsJMnhkEZR5BVsYT -r7jHwelP527+QCLkVUCHIVIWUW0ANzeZBhDV2vefkFb+gWLiZsBhaHssbcKGsMNG -Ak4xS1ByqsM= -=lsIJ +iQCVAwUBOlxeLTpxHvX/mS9tAQH6RwP8DhyvqAnXFV6WIGi16iQ3vKikkPoqnDQs +GEn5uCW0dPYKlwthD2Grj/JbMYZhOmCFuDxF7ufJnjTSDe/D8XNe2wngxzAiwcIe +WjCrT8X95cuP3HZHscbFTEinVV0GAnoI0ZEgs5eBDhVHDqILLYMaTFBQaRH3jgXc +i5VH88jPfUM= +=qc+J -----END PGP SIGNATURE----- diff --git a/drivers/isdn/isdn_common.c b/drivers/isdn/isdn_common.c index 7a5074c79..f4c1d429d 100644 --- a/drivers/isdn/isdn_common.c +++ b/drivers/isdn/isdn_common.c @@ -1512,7 +1512,7 @@ isdn_ioctl(struct inode *inode, struct file *file, uint cmd, ulong arg) int i; if ((ret = verify_area(VERIFY_READ, (void *) arg, - (ISDN_MODEM_NUMREG + ISDN_MSNLEN) + (ISDN_MODEM_NUMREG + ISDN_MSNLEN + ISDN_LMSNLEN) * ISDN_MAX_CHANNELS))) return ret; @@ -1521,6 +1521,9 @@ isdn_ioctl(struct inode *inode, struct file *file, uint cmd, ulong arg) ISDN_MODEM_NUMREG)) return -EFAULT; p += ISDN_MODEM_NUMREG; + if (copy_from_user(dev->mdm.info[i].emu.plmsn, p, ISDN_LMSNLEN)) + return -EFAULT; + p += ISDN_LMSNLEN; if (copy_from_user(dev->mdm.info[i].emu.pmsn, p, ISDN_MSNLEN)) return -EFAULT; p += ISDN_MSNLEN; diff --git a/drivers/isdn/isdn_net.c b/drivers/isdn/isdn_net.c index 757c59ca9..672dd8c5c 100644 --- a/drivers/isdn/isdn_net.c +++ b/drivers/isdn/isdn_net.c @@ -2325,6 +2325,7 @@ isdn_net_new(char *name, struct net_device *master) memset(netdev, 0, sizeof(isdn_net_dev)); if (!(netdev->local = (isdn_net_local *) kmalloc(sizeof(isdn_net_local), GFP_KERNEL))) { printk(KERN_WARNING "isdn_net: Could not allocate device locals\n"); + kfree(netdev); return NULL; } memset(netdev->local, 0, sizeof(isdn_net_local)); diff --git a/drivers/isdn/isdn_ppp.c b/drivers/isdn/isdn_ppp.c index 3bebcdfcf..f1cbb8ab7 100644 --- a/drivers/isdn/isdn_ppp.c +++ b/drivers/isdn/isdn_ppp.c @@ -1131,9 +1131,9 @@ isdn_ppp_xmit(struct sk_buff *skb, struct net_device *netdev) proto = PPP_IPX; /* untested */ break; default: - dev_kfree_skb(skb); printk(KERN_ERR "isdn_ppp: skipped unsupported protocol: %#x.\n", skb->protocol); + dev_kfree_skb(skb); return 0; } @@ -2310,8 +2310,7 @@ static struct sk_buff *isdn_ppp_decompress(struct sk_buff *skb,struct ippp_struc rsparm.data = rsdata; rsparm.maxdlen = IPPP_RESET_MAXDATABYTES; - /* !!!HACK,HACK,HACK!!! 2048 is only assumed */ - skb_out = dev_alloc_skb(2048); + skb_out = dev_alloc_skb(is->mru + PPP_HDRLEN); len = ipc->decompress(stat, skb, skb_out, &rsparm); kfree_skb(skb); if (len <= 0) { @@ -2332,14 +2331,9 @@ static struct sk_buff *isdn_ppp_decompress(struct sk_buff *skb,struct ippp_struc kfree_skb(skb_out); return NULL; } - - if (isdn_ppp_skip_ac(ri, skb) < 0) { - kfree_skb(skb); - return NULL; - } - *proto = isdn_ppp_strip_proto(skb); + *proto = isdn_ppp_strip_proto(skb_out); if (*proto < 0) { - kfree_skb(skb); + kfree_skb(skb_out); return NULL; } return skb_out; diff --git a/drivers/isdn/isdn_v110.c b/drivers/isdn/isdn_v110.c index 08c54a866..5dd9f2c58 100644 --- a/drivers/isdn/isdn_v110.c +++ b/drivers/isdn/isdn_v110.c @@ -102,7 +102,7 @@ isdn_v110_open(unsigned char key, int hdrlen, int maxsize) int i; isdn_v110_stream *v; - if ((v = kmalloc(sizeof(isdn_v110_stream), GFP_KERNEL)) == NULL) + if ((v = kmalloc(sizeof(isdn_v110_stream), GFP_ATOMIC)) == NULL) return NULL; memset(v, 0, sizeof(isdn_v110_stream)); v->key = key; @@ -134,7 +134,7 @@ isdn_v110_open(unsigned char key, int hdrlen, int maxsize) v->b = 0; v->skbres = hdrlen; v->maxsize = maxsize - hdrlen; - if ((v->encodebuf = kmalloc(maxsize, GFP_KERNEL)) == NULL) { + if ((v->encodebuf = kmalloc(maxsize, GFP_ATOMIC)) == NULL) { kfree(v); return NULL; } diff --git a/drivers/md/Config.in b/drivers/md/Config.in index 565055a68..30438c77f 100644 --- a/drivers/md/Config.in +++ b/drivers/md/Config.in @@ -11,12 +11,7 @@ dep_tristate ' Linear (append) mode' CONFIG_MD_LINEAR $CONFIG_BLK_DEV_MD dep_tristate ' RAID-0 (striping) mode' CONFIG_MD_RAID0 $CONFIG_BLK_DEV_MD dep_tristate ' RAID-1 (mirroring) mode' CONFIG_MD_RAID1 $CONFIG_BLK_DEV_MD dep_tristate ' RAID-4/RAID-5 mode' CONFIG_MD_RAID5 $CONFIG_BLK_DEV_MD -if [ "$CONFIG_MD_LINEAR" = "y" -o "$CONFIG_MD_RAID0" = "y" -o "$CONFIG_MD_RAID1" = "y" -o "$CONFIG_MD_RAID5" = "y" ]; then - bool ' Boot support' CONFIG_MD_BOOT - bool ' Auto Detect support' CONFIG_AUTODETECT_RAID -fi dep_tristate ' Logical volume manager (LVM) support' CONFIG_BLK_DEV_LVM $CONFIG_MD -dep_mbool ' LVM information in proc filesystem' CONFIG_LVM_PROC_FS $CONFIG_BLK_DEV_LVM endmenu diff --git a/drivers/md/lvm-snap.c b/drivers/md/lvm-snap.c index 980694ee3..e28ffdbe9 100644 --- a/drivers/md/lvm-snap.c +++ b/drivers/md/lvm-snap.c @@ -21,6 +21,14 @@ * */ +/* + * Changelog + * + * 05/07/2000 - implemented persistent snapshot support + * 23/11/2000 - used cpu_to_le64 rather than my own macro + * + */ + #include <linux/kernel.h> #include <linux/vmalloc.h> #include <linux/blkdev.h> @@ -30,7 +38,9 @@ #include <linux/lvm.h> -static char *lvm_snap_version __attribute__ ((unused)) = "LVM 0.9 snapshot code (13/11/2000)\n"; +#include "lvm-snap.h" + +static char *lvm_snap_version __attribute__ ((unused)) = "LVM 0.9.1_beta2 snapshot code (18/01/2001)\n"; extern const char *const lvm_name; extern int lvm_blocksizes[]; @@ -214,10 +224,10 @@ void lvm_snapshot_fill_COW_page(vg_t * vg, lv_t * lv_snap) memset(lv_COW_table, 0, blksize_snap); for ( ; is < lv_snap->lv_remap_ptr; is++, id++) { /* store new COW_table entry */ - lv_COW_table[id].pv_org_number = LVM_TO_DISK64(lvm_pv_get_number(vg, lv_snap->lv_block_exception[is].rdev_org)); - lv_COW_table[id].pv_org_rsector = LVM_TO_DISK64(lv_snap->lv_block_exception[is].rsector_org); - lv_COW_table[id].pv_snap_number = LVM_TO_DISK64(lvm_pv_get_number(vg, lv_snap->lv_block_exception[is].rdev_new)); - lv_COW_table[id].pv_snap_rsector = LVM_TO_DISK64(lv_snap->lv_block_exception[is].rsector_new); + lv_COW_table[id].pv_org_number = cpu_to_le64(lvm_pv_get_number(vg, lv_snap->lv_block_exception[is].rdev_org)); + lv_COW_table[id].pv_org_rsector = cpu_to_le64(lv_snap->lv_block_exception[is].rsector_org); + lv_COW_table[id].pv_snap_number = cpu_to_le64(lvm_pv_get_number(vg, lv_snap->lv_block_exception[is].rdev_new)); + lv_COW_table[id].pv_snap_rsector = cpu_to_le64(lv_snap->lv_block_exception[is].rsector_new); } } @@ -227,8 +237,7 @@ void lvm_snapshot_fill_COW_page(vg_t * vg, lv_t * lv_snap) * */ -int lvm_write_COW_table_block(vg_t * vg, - lv_t * lv_snap) +int lvm_write_COW_table_block(vg_t * vg, lv_t * lv_snap) { int blksize_snap; int end_of_table; @@ -268,10 +277,10 @@ int lvm_write_COW_table_block(vg_t * vg, blocks[0] = (snap_pe_start + COW_table_sector_offset) >> (blksize_snap >> 10); /* store new COW_table entry */ - lv_COW_table[idx_COW_table].pv_org_number = LVM_TO_DISK64(lvm_pv_get_number(vg, lv_snap->lv_block_exception[idx].rdev_org)); - lv_COW_table[idx_COW_table].pv_org_rsector = LVM_TO_DISK64(lv_snap->lv_block_exception[idx].rsector_org); - lv_COW_table[idx_COW_table].pv_snap_number = LVM_TO_DISK64(lvm_pv_get_number(vg, snap_phys_dev)); - lv_COW_table[idx_COW_table].pv_snap_rsector = LVM_TO_DISK64(lv_snap->lv_block_exception[idx].rsector_new); + lv_COW_table[idx_COW_table].pv_org_number = cpu_to_le64(lvm_pv_get_number(vg, lv_snap->lv_block_exception[idx].rdev_org)); + lv_COW_table[idx_COW_table].pv_org_rsector = cpu_to_le64(lv_snap->lv_block_exception[idx].rsector_org); + lv_COW_table[idx_COW_table].pv_snap_number = cpu_to_le64(lvm_pv_get_number(vg, snap_phys_dev)); + lv_COW_table[idx_COW_table].pv_snap_rsector = cpu_to_le64(lv_snap->lv_block_exception[idx].rsector_new); length_tmp = iobuf->length; iobuf->length = blksize_snap; @@ -568,6 +577,7 @@ void lvm_snapshot_release(lv_t * lv) } if (lv->lv_iobuf) { + kiobuf_wait_for_io(lv->lv_iobuf); unmap_kiobuf(lv->lv_iobuf); free_kiovec(1, &lv->lv_iobuf); lv->lv_iobuf = NULL; diff --git a/drivers/md/lvm-snap.h b/drivers/md/lvm-snap.h new file mode 100644 index 000000000..23538a1b7 --- /dev/null +++ b/drivers/md/lvm-snap.h @@ -0,0 +1,47 @@ +/* + * kernel/lvm-snap.h + * + * Copyright (C) 2001 Sistina Software + * + * + * LVM driver is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2, or (at your option) + * any later version. + * + * LVM driver is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with GNU CC; see the file COPYING. If not, write to + * the Free Software Foundation, 59 Temple Place - Suite 330, + * Boston, MA 02111-1307, USA. + * + */ + +/* + * Changelog + * + * 05/01/2001:Joe Thornber - Factored this file out of lvm.c + * + */ + +#ifndef LVM_SNAP_H +#define LVM_SNAP_H + +/* external snapshot calls */ +extern inline int lvm_get_blksize(kdev_t); +extern int lvm_snapshot_alloc(lv_t *); +extern void lvm_snapshot_fill_COW_page(vg_t *, lv_t *); +extern int lvm_snapshot_COW(kdev_t, ulong, ulong, ulong, lv_t *); +extern int lvm_snapshot_remap_block(kdev_t *, ulong *, ulong, lv_t *); +extern void lvm_snapshot_release(lv_t *); +extern int lvm_write_COW_table_block(vg_t *, lv_t *); +extern inline void lvm_hash_link(lv_block_exception_t *, + kdev_t, ulong, lv_t *); +extern int lvm_snapshot_alloc_hash_table(lv_t *); +extern void lvm_drop_snapshot(lv_t *, const char *); + +#endif diff --git a/drivers/md/lvm.c b/drivers/md/lvm.c index ea276c57c..a4ca05e90 100644 --- a/drivers/md/lvm.c +++ b/drivers/md/lvm.c @@ -7,22 +7,23 @@ * April-May,July-August,November 1998 * January-March,May,July,September,October 1999 * January,February,July,September-November 2000 + * January 2001 * * * LVM driver is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2, or (at your option) * any later version. - * + * * LVM driver is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. - * + * * You should have received a copy of the GNU General Public License * along with GNU CC; see the file COPYING. If not, write to * the Free Software Foundation, 59 Temple Place - Suite 330, - * Boston, MA 02111-1307, USA. + * Boston, MA 02111-1307, USA. * */ @@ -138,13 +139,21 @@ * 01/11/2000 - added memory information on hash tables to * lvm_proc_get_global_info() * 02/11/2000 - implemented /proc/lvm/ hierarchy - * 07/12/2000 - make sure lvm_make_request_fn returns correct value - 0 or 1 - NeilBrown + * 22/11/2000 - changed lvm_do_create_proc_entry_of_pv () to work + * with devfs + * 26/11/2000 - corrected #ifdef locations for PROC_FS + * 28/11/2000 - fixed lvm_do_vg_extend() NULL pointer BUG + * - fixed lvm_do_create_proc_entry_of_pv() buffer tampering BUG + * 08/01/2001 - Removed conditional compiles related to PROC_FS, + * procfs is always supported now. (JT) + * 12/01/2001 - avoided flushing logical volume in case of shrinking + * because of unecessary overhead in case of heavy updates * */ -static char *lvm_version = "LVM version 0.9 by Heinz Mauelshagen (13/11/2000)\n"; -static char *lvm_short_version = "version 0.9 (13/11/2000)"; +static char *lvm_version = "LVM version 0.9.1_beta2 by Heinz Mauelshagen (18/01/2001)\n"; +static char *lvm_short_version = "version 0.9.1_beta2 (18/01/2001)"; #define MAJOR_NR LVM_BLK_MAJOR #define DEVICE_OFF(device) @@ -190,6 +199,8 @@ static char *lvm_short_version = "version 0.9 (13/11/2000)"; #include <linux/errno.h> #include <linux/lvm.h> +#include "lvm-snap.h" + #define LVM_CORRECT_READ_AHEAD( a) \ if ( a < LVM_MIN_READ_AHEAD || \ a > LVM_MAX_READ_AHEAD) a = LVM_MAX_READ_AHEAD; @@ -198,19 +209,28 @@ static char *lvm_short_version = "version 0.9 (13/11/2000)"; # define WRITEA WRITE #endif -/* - * External function prototypes - */ -#ifdef MODULE -int init_module(void); -void cleanup_module(void); +/* debug macros */ +#ifdef DEBUG_IOCTL +#define P_IOCTL(fmt, args...) printk(KERN_DEBUG "lvm ioctl: " fmt, ## args) #else -extern int lvm_init(void); +#define P_IOCTL(fmt, args...) #endif -static void lvm_dummy_device_request(request_queue_t *); -#define DEVICE_REQUEST lvm_dummy_device_request +#ifdef DEBUG_MAP +#define P_MAP(fmt, args...) printk(KERN_DEBUG "lvm map: " fmt, ## args) +#else +#define P_MAP(fmt, args...) +#endif + +#ifdef DEBUG_KFREE +#define P_KFREE(fmt, args...) printk(KERN_DEBUG "lvm kfree: " fmt, ## args) +#else +#define P_KFREE(fmt, args...) +#endif +/* + * External function prototypes + */ static int lvm_make_request_fn(request_queue_t*, int, struct buffer_head*); static int lvm_blk_ioctl(struct inode *, struct file *, uint, ulong); @@ -224,42 +244,29 @@ static int lvm_user_bmap(struct inode *, struct lv_bmap *); static int lvm_chr_ioctl(struct inode *, struct file *, uint, ulong); -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS int lvm_proc_read_vg_info(char *, char **, off_t, int, int *, void *); int lvm_proc_read_lv_info(char *, char **, off_t, int, int *, void *); int lvm_proc_read_pv_info(char *, char **, off_t, int, int *, void *); static int lvm_proc_get_global_info(char *, char **, off_t, int, int *, void *); + +void lvm_do_create_devfs_entry_of_vg ( vg_t *); + void lvm_do_create_proc_entry_of_vg ( vg_t *); -inline void lvm_do_remove_proc_entry_of_vg ( vg_t *); -inline void lvm_do_create_proc_entry_of_lv ( vg_t *, lv_t *); -inline void lvm_do_remove_proc_entry_of_lv ( vg_t *, lv_t *); -inline void lvm_do_create_proc_entry_of_pv ( vg_t *, pv_t *); -inline void lvm_do_remove_proc_entry_of_pv ( vg_t *, pv_t *); -#endif +void lvm_do_remove_proc_entry_of_vg ( vg_t *); +void lvm_do_create_proc_entry_of_lv ( vg_t *, lv_t *); +void lvm_do_remove_proc_entry_of_lv ( vg_t *, lv_t *); +void lvm_do_create_proc_entry_of_pv ( vg_t *, pv_t *); +void lvm_do_remove_proc_entry_of_pv ( vg_t *, pv_t *); -#ifdef LVM_HD_NAME -void lvm_hd_name(char *, int); -#endif /* End external function prototypes */ /* * Internal function prototypes */ +static void lvm_cleanup(void); static void lvm_init_vars(void); -/* external snapshot calls */ -extern inline int lvm_get_blksize(kdev_t); -extern int lvm_snapshot_alloc(lv_t *); -extern void lvm_snapshot_fill_COW_page(vg_t *, lv_t *); -extern int lvm_snapshot_COW(kdev_t, ulong, ulong, ulong, lv_t *); -extern int lvm_snapshot_remap_block(kdev_t *, ulong *, ulong, lv_t *); -extern void lvm_snapshot_release(lv_t *); -extern int lvm_write_COW_table_block(vg_t *, lv_t *); -extern inline void lvm_hash_link(lv_block_exception_t *, kdev_t, ulong, lv_t *); -extern int lvm_snapshot_alloc_hash_table(lv_t *); -extern void lvm_drop_snapshot(lv_t *, char *); - #ifdef LVM_HD_NAME extern void (*lvm_hd_name_ptr) (char *, int); #endif @@ -288,9 +295,9 @@ static int lvm_do_vg_reduce(vg_t *, void *); static int lvm_do_vg_rename(vg_t *, void *); static int lvm_do_vg_remove(int); static void lvm_geninit(struct gendisk *); -#ifdef LVM_GET_INODE -static struct inode *lvm_get_inode(int); -void lvm_clear_inode(struct inode *); +static char *lvm_show_uuid ( char *); +#ifdef LVM_HD_NAME +void lvm_hd_name(char *, int); #endif /* END Internal function prototypes */ @@ -298,12 +305,10 @@ void lvm_clear_inode(struct inode *); /* volume group descriptor area pointers */ static vg_t *vg[ABS_MAX_VG]; -#ifdef CONFIG_DEVFS_FS static devfs_handle_t lvm_devfs_handle; static devfs_handle_t vg_devfs_handle[MAX_VG]; static devfs_handle_t ch_devfs_handle[MAX_VG]; static devfs_handle_t lv_devfs_handle[MAX_LV]; -#endif static pv_t *pvp = NULL; static lv_t *lvp = NULL; @@ -340,18 +345,15 @@ static int loadtime = 0; static uint vg_count = 0; static long lvm_chr_open_count = 0; static ushort lvm_iop_version = LVM_DRIVER_IOP_VERSION; -static DECLARE_WAIT_QUEUE_HEAD(lvm_snapshot_wait); static DECLARE_WAIT_QUEUE_HEAD(lvm_wait); static DECLARE_WAIT_QUEUE_HEAD(lvm_map_wait); static spinlock_t lvm_lock = SPIN_LOCK_UNLOCKED; static spinlock_t lvm_snapshot_lock = SPIN_LOCK_UNLOCKED; -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS static struct proc_dir_entry *lvm_proc_dir = NULL; static struct proc_dir_entry *lvm_proc_vg_subdir = NULL; struct proc_dir_entry *pde = NULL; -#endif static struct file_operations lvm_chr_fops = { @@ -360,7 +362,7 @@ static struct file_operations lvm_chr_fops = ioctl: lvm_chr_ioctl, }; -#define BLOCK_DEVICE_OPERATIONS + /* block device operations structure needed for 2.3.38? and above */ static struct block_device_operations lvm_blk_dops = { @@ -391,22 +393,10 @@ static struct gendisk lvm_gendisk = NULL, /* pointer to next gendisk struct (internal) */ }; - -#ifdef MODULE -/* - * Module initialization... - */ -int init_module(void) -#else /* * Driver initialization... */ -#ifdef __initfunc -__initfunc(int lvm_init(void)) -#else -int __init lvm_init(void) -#endif -#endif /* #ifdef MODULE */ +int lvm_init(void) { struct gendisk *gendisk_ptr = NULL; @@ -414,11 +404,7 @@ int __init lvm_init(void) printk(KERN_ERR "%s -- register_chrdev failed\n", lvm_name); return -EIO; } -#ifdef BLOCK_DEVICE_OPERATIONS if (register_blkdev(MAJOR_NR, lvm_name, &lvm_blk_dops) < 0) -#else - if (register_blkdev(MAJOR_NR, lvm_name, &lvm_blk_fops) < 0) -#endif { printk("%s -- register_blkdev failed\n", lvm_name); if (unregister_chrdev(LVM_CHAR_MAJOR, lvm_name) < 0) @@ -426,21 +412,17 @@ int __init lvm_init(void) return -EIO; } -#ifdef CONFIG_DEVFS_FS lvm_devfs_handle = devfs_register( 0 , "lvm", 0, 0, LVM_CHAR_MAJOR, S_IFCHR | S_IRUSR | S_IWUSR | S_IRGRP, &lvm_chr_fops, NULL); -#endif -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS lvm_proc_dir = create_proc_entry (LVM_DIR, S_IFDIR, &proc_root); if (lvm_proc_dir != NULL) { lvm_proc_vg_subdir = create_proc_entry (LVM_VG_SUBDIR, S_IFDIR, lvm_proc_dir); pde = create_proc_entry(LVM_GLOBAL, S_IFREG, lvm_proc_dir); if ( pde != NULL) pde->read_proc = &lvm_proc_get_global_info; } -#endif lvm_init_vars(); lvm_geninit(&lvm_gendisk); @@ -464,9 +446,9 @@ int __init lvm_init(void) lvm_hd_name_ptr = lvm_hd_name; #endif - blk_init_queue(BLK_DEFAULT_QUEUE(MAJOR_NR), DEVICE_REQUEST); blk_queue_make_request(BLK_DEFAULT_QUEUE(MAJOR_NR), lvm_make_request_fn); + /* optional read root VGDA */ /* if ( *rootvg != 0) vg_read_with_pv_and_lv ( rootvg, &vg); @@ -483,20 +465,17 @@ int __init lvm_init(void) lvm_version, lvm_name); return 0; -} /* init_module() / lvm_init() */ +} /* lvm_init() */ -#ifdef MODULE /* - * Module cleanup... + * cleanup... */ -void cleanup_module(void) +static void lvm_cleanup(void) { struct gendisk *gendisk_ptr = NULL, *gendisk_ptr_prev = NULL; -#ifdef CONFIG_DEVFS_FS devfs_unregister (lvm_devfs_handle); -#endif if (unregister_chrdev(LVM_CHAR_MAJOR, lvm_name) < 0) { printk(KERN_ERR "%s -- unregister_chrdev failed\n", lvm_name); @@ -504,7 +483,7 @@ void cleanup_module(void) if (unregister_blkdev(MAJOR_NR, lvm_name) < 0) { printk(KERN_ERR "%s -- unregister_blkdev failed\n", lvm_name); } - blk_cleanup_queue(BLK_DEFAULT_QUEUE(MAJOR_NR)); + gendisk_ptr = gendisk_ptr_prev = gendisk_head; while (gendisk_ptr != NULL) { @@ -521,11 +500,9 @@ void cleanup_module(void) blksize_size[MAJOR_NR] = NULL; hardsect_size[MAJOR_NR] = NULL; -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS remove_proc_entry(LVM_GLOBAL, lvm_proc_dir); remove_proc_entry(LVM_VG_SUBDIR, lvm_proc_dir); remove_proc_entry(LVM_DIR, &proc_root); -#endif #ifdef LVM_HD_NAME /* reference from linux/drivers/block/genhd.c */ @@ -535,18 +512,13 @@ void cleanup_module(void) printk(KERN_INFO "%s -- Module successfully deactivated\n", lvm_name); return; -} /* void cleanup_module() */ -#endif /* #ifdef MODULE */ +} /* lvm_cleanup() */ /* * support function to initialize lvm variables */ -#ifdef __initfunc -__initfunc(void lvm_init_vars(void)) -#else void __init lvm_init_vars(void) -#endif { int v; @@ -626,13 +598,9 @@ static int lvm_chr_ioctl(struct inode *inode, struct file *file, /* otherwise cc will complain about unused variables */ (void) lvm_lock; - -#ifdef DEBUG_IOCTL - printk(KERN_DEBUG - "%s -- lvm_chr_ioctl: command: 0x%X MINOR: %d " - "VG#: %d mode: 0x%X\n", - lvm_name, command, minor, VG_CHR(minor), file->f_mode); -#endif + P_IOCTL("%s -- lvm_chr_ioctl: command: 0x%X MINOR: %d " + "VG#: %d mode: 0x%X\n", + lvm_name, command, minor, VG_CHR(minor), file->f_mode); #ifdef LVM_TOTAL_RESET if (lvm_reset_spindown > 0) return -EACCES; @@ -890,14 +858,12 @@ static int lvm_blk_open(struct inode *inode, struct file *file) if (lv_ptr->lv_status & LV_SPINDOWN) return -EPERM; /* Check inactive LV and open for read/write */ - if (file->f_mode & O_RDWR) { - if (!(lv_ptr->lv_status & LV_ACTIVE)) return -EPERM; - if (!(lv_ptr->lv_access & LV_WRITE)) return -EACCES; - } + if (!(lv_ptr->lv_status & LV_ACTIVE)) + return -EPERM; + if (!(lv_ptr->lv_access & LV_WRITE) && + (file->f_mode & FMODE_WRITE)) + return -EACCES; -#ifndef BLOCK_DEVICE_OPERATIONS - file->f_op = &lvm_blk_fops; -#endif /* be sure to increment VG counter */ if (lv_ptr->lv_open == 0) vg_ptr->lv_open++; @@ -930,24 +896,18 @@ static int lvm_blk_ioctl(struct inode *inode, struct file *file, void *arg = (void *) a; struct hd_geometry *hd = (struct hd_geometry *) a; -#ifdef DEBUG_IOCTL - printk(KERN_DEBUG - "%s -- lvm_blk_ioctl MINOR: %d command: 0x%X arg: %X " - "VG#: %dl LV#: %d\n", - lvm_name, minor, command, (ulong) arg, - VG_BLK(minor), LV_BLK(minor)); -#endif + P_IOCTL("%s -- lvm_blk_ioctl MINOR: %d command: 0x%X arg: %X " + "VG#: %dl LV#: %d\n", + lvm_name, minor, command, (ulong) arg, + VG_BLK(minor), LV_BLK(minor)); switch (command) { case BLKGETSIZE: /* return device size */ -#ifdef DEBUG_IOCTL - printk(KERN_DEBUG - "%s -- lvm_blk_ioctl -- BLKGETSIZE: %u\n", - lvm_name, lv_ptr->lv_size); -#endif + P_IOCTL("%s -- lvm_blk_ioctl -- BLKGETSIZE: %u\n", + lvm_name, lv_ptr->lv_size); if (put_user(lv_ptr->lv_size, (long *)arg)) - return -EFAULT; + return -EFAULT; break; @@ -955,10 +915,8 @@ static int lvm_blk_ioctl(struct inode *inode, struct file *file, /* flush buffer cache */ if (!capable(CAP_SYS_ADMIN)) return -EACCES; -#ifdef DEBUG_IOCTL - printk(KERN_DEBUG - "%s -- lvm_blk_ioctl -- BLKFLSBUF\n", lvm_name); -#endif + P_IOCTL("%s -- lvm_blk_ioctl -- BLKFLSBUF\n", lvm_name); + fsync_dev(inode->i_rdev); invalidate_buffers(inode->i_rdev); break; @@ -968,11 +926,9 @@ static int lvm_blk_ioctl(struct inode *inode, struct file *file, /* set read ahead for block device */ if (!capable(CAP_SYS_ADMIN)) return -EACCES; -#ifdef DEBUG_IOCTL - printk(KERN_DEBUG - "%s -- lvm_blk_ioctl -- BLKRASET: %d sectors for %02X:%02X\n", - lvm_name, (long) arg, MAJOR(inode->i_rdev), minor); -#endif + P_IOCTL("%s -- lvm_blk_ioctl -- BLKRASET: %d sectors for %02X:%02X\n", + lvm_name, (long) arg, MAJOR(inode->i_rdev), minor); + if ((long) arg < LVM_MIN_READ_AHEAD || (long) arg > LVM_MAX_READ_AHEAD) return -EINVAL; @@ -982,10 +938,7 @@ static int lvm_blk_ioctl(struct inode *inode, struct file *file, case BLKRAGET: /* get current read ahead setting */ -#ifdef DEBUG_IOCTL - printk(KERN_DEBUG - "%s -- lvm_blk_ioctl -- BLKRAGET\n", lvm_name); -#endif + P_IOCTL("%s -- lvm_blk_ioctl -- BLKRAGET\n", lvm_name); if (put_user(lv_ptr->lv_read_ahead, (long *)arg)) return -EFAULT; break; @@ -993,10 +946,7 @@ static int lvm_blk_ioctl(struct inode *inode, struct file *file, case HDIO_GETGEO: /* get disk geometry */ -#ifdef DEBUG_IOCTL - printk(KERN_DEBUG - "%s -- lvm_blk_ioctl -- HDIO_GETGEO\n", lvm_name); -#endif + P_IOCTL("%s -- lvm_blk_ioctl -- HDIO_GETGEO\n", lvm_name); if (hd == NULL) return -EINVAL; { @@ -1016,11 +966,8 @@ static int lvm_blk_ioctl(struct inode *inode, struct file *file, return -EFAULT; } -#ifdef DEBUG_IOCTL - printk(KERN_DEBUG - "%s -- lvm_blk_ioctl -- cylinders: %d\n", - lvm_name, lv_ptr->lv_size / heads / sectors); -#endif + P_IOCTL("%s -- lvm_blk_ioctl -- cylinders: %d\n", + lvm_name, lv_ptr->lv_size / heads / sectors); break; @@ -1127,22 +1074,22 @@ static int lvm_user_bmap(struct inode *inode, struct lv_bmap *user_result) struct buffer_head bh; unsigned long block; int err; - + if (get_user(block, &user_result->lv_block)) - return -EFAULT; - + return -EFAULT; + memset(&bh,0,sizeof bh); bh.b_rsector = block; bh.b_dev = bh.b_rdev = inode->i_dev; bh.b_size = lvm_get_blksize(bh.b_dev); if ((err=lvm_map(&bh, READ)) < 0) { - printk("lvm map failed: %d\n", err); - return -EINVAL; + printk("lvm map failed: %d\n", err); + return -EINVAL; } - - return put_user( kdev_t_to_nr(bh.b_rdev), &user_result->lv_dev) || + + return put_user(kdev_t_to_nr(bh.b_rdev), &user_result->lv_dev) || put_user(bh.b_rsector, &user_result->lv_block) ? -EFAULT : 0; -} +} /* @@ -1168,7 +1115,7 @@ int lvm_vg_info(vg_t *vg_ptr, char *buf) { vg_ptr->pe_total, vg_ptr->pe_allocated * vg_ptr->pe_size >> 1, vg_ptr->pe_allocated, - (vg_ptr->pe_total - vg_ptr->pe_allocated) * + (vg_ptr->pe_total - vg_ptr->pe_allocated) * vg_ptr->pe_size >> 1, vg_ptr->pe_total - vg_ptr->pe_allocated); return sz; @@ -1263,7 +1210,6 @@ int lvm_pv_info(pv_t *pv_ptr, char *buf) { } -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS /* * Support functions /proc-Filesystem */ @@ -1325,10 +1271,7 @@ static int lvm_proc_get_global_info(char *page, char **start, off_t pos, int cou lv_block_exception_t_bytes *= sizeof(lv_block_exception_t); if (buf != NULL) { -#ifdef DEBUG_KFREE - printk(KERN_DEBUG - "%s -- vfree %d\n", lvm_name, __LINE__); -#endif + P_KFREE("%s -- vfree %d\n", lvm_name, __LINE__); lock_kernel(); vfree(buf); unlock_kernel(); @@ -1452,7 +1395,6 @@ static int lvm_proc_get_global_info(char *page, char **start, off_t pos, int cou else return count; } /* lvm_proc_get_global_info() */ -#endif /* #if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS */ /* @@ -1478,7 +1420,7 @@ int lvm_proc_read_vg_info(char *page, char **start, off_t off, sz += sprintf ( page+sz, "PE size: %u\n", vg->pe_size / 2); sz += sprintf ( page+sz, "PE total: %u\n", vg->pe_total); sz += sprintf ( page+sz, "PE allocated: %u\n", vg->pe_allocated); - sz += sprintf ( page+sz, "uuid: %s\n", vg->vg_uuid); + sz += sprintf ( page+sz, "uuid: %s\n", lvm_show_uuid(vg->vg_uuid)); return sz; } @@ -1525,7 +1467,7 @@ int lvm_proc_read_pv_info(char *page, char **start, off_t off, sz += sprintf ( page+sz, "PE allocated: %u\n", pv->pe_allocated); sz += sprintf ( page+sz, "device: %02u:%02u\n", MAJOR(pv->pv_dev), MINOR(pv->pv_dev)); - sz += sprintf ( page+sz, "uuid: %s\n", pv->pv_uuid); + sz += sprintf ( page+sz, "uuid: %s\n", lvm_show_uuid(pv->pv_uuid)); return sz; @@ -1565,15 +1507,13 @@ static int lvm_map(struct buffer_head *bh, int rw) lvm_name, lv->lv_name); return -1; } -#ifdef DEBUG_MAP - printk(KERN_DEBUG - "%s - lvm_map minor:%d *rdev: %02d:%02d *rsector: %lu " - "size:%lu\n", - lvm_name, minor, - MAJOR(rdev_tmp), - MINOR(rdev_tmp), - rsector_tmp, size); -#endif + + P_MAP("%s - lvm_map minor:%d *rdev: %02d:%02d *rsector: %lu " + "size:%lu\n", + lvm_name, minor, + MAJOR(rdev_tmp), + MINOR(rdev_tmp), + rsector_tmp, size); if (rsector_tmp + size > lv->lv_size) { printk(KERN_ALERT @@ -1595,15 +1535,13 @@ lvm_second_remap: (rsector_tmp % vg_this->pe_size); rdev_tmp = lv->lv_current_pe[index].dev; -#ifdef DEBUG_MAP - printk(KERN_DEBUG - "lv_current_pe[%ld].pe: %ld rdev: %02d:%02d rsector:%ld\n", + P_MAP("lv_current_pe[%ld].pe: %ld rdev: %02d:%02d " + "rsector:%ld\n", index, lv->lv_current_pe[index].pe, MAJOR(rdev_tmp), MINOR(rdev_tmp), rsector_tmp); -#endif /* striped mapping */ } else { @@ -1624,9 +1562,7 @@ lvm_second_remap: rdev_tmp = lv->lv_current_pe[index].dev; } -#ifdef DEBUG_MAP - printk(KERN_DEBUG - "lv_current_pe[%ld].pe: %ld rdev: %02d:%02d rsector:%ld\n" + P_MAP("lv_current_pe[%ld].pe: %ld rdev: %02d:%02d rsector:%ld\n" "stripe_length: %ld stripe_index: %ld\n", index, lv->lv_current_pe[index].pe, @@ -1635,7 +1571,6 @@ lvm_second_remap: rsector_tmp, stripe_length, stripe_index); -#endif /* handle physical extents on the move */ if (pe_lock_req.lock == LOCK_PE) { @@ -1659,6 +1594,8 @@ lvm_second_remap: if (lv->lv_access & (LV_SNAPSHOT|LV_SNAPSHOT_ORG)) { /* original logical volume */ if (lv->lv_access & LV_SNAPSHOT_ORG) { + /* Serializes the access to the lv_snapshot_next list */ + down(&lv->lv_snapshot_sem); if (rw == WRITE || rw == WRITEA) { lv_t *lv_ptr; @@ -1669,7 +1606,8 @@ lvm_second_remap: lv_ptr = lv_ptr->lv_snapshot_next) { /* Check for inactive snapshot */ if (!(lv_ptr->lv_status & LV_ACTIVE)) continue; - down(&lv->lv_snapshot_org->lv_snapshot_sem); + /* Serializes the COW with the accesses to the snapshot device */ + down(&lv_ptr->lv_snapshot_sem); /* do we still have exception storage for this snapshot free? */ if (lv_ptr->lv_block_exception != NULL) { rdev_sav = rdev_tmp; @@ -1690,9 +1628,10 @@ lvm_second_remap: rdev_tmp = rdev_sav; rsector_tmp = rsector_sav; } - up(&lv->lv_snapshot_org->lv_snapshot_sem); + up(&lv_ptr->lv_snapshot_sem); } } + up(&lv->lv_snapshot_sem); } else { /* remap snapshot logical volume */ down(&lv->lv_snapshot_sem); @@ -1733,31 +1672,12 @@ void lvm_hd_name(char *buf, int minor) /* - * this one never should be called... - */ -static void lvm_dummy_device_request(request_queue_t * t) -{ - printk(KERN_EMERG - "%s -- oops, got lvm request for %02d:%02d [sector: %lu]\n", - lvm_name, - MAJOR(CURRENT->rq_dev), - MINOR(CURRENT->rq_dev), - CURRENT->sector); - return; -} - - -/* * make request function */ static int lvm_make_request_fn(request_queue_t *q, int rw, - struct buffer_head *bh) -{ - if (lvm_map(bh, rw)<0) - return 0; /* failure, buffer_IO_error has been called, don't recurse */ - else - return 1; /* all ok, mapping done, call lower level driver */ + struct buffer_head *bh) { + return (lvm_map(bh, rw) < 0) ? 0 : 1; } @@ -1774,10 +1694,8 @@ static int lvm_do_lock_lvm(void) lock_try_again: spin_lock(&lvm_lock); if (lock != 0 && lock != current->pid) { -#ifdef DEBUG_IOCTL - printk(KERN_INFO "lvm_do_lock_lvm: %s is locked by pid %d ...\n", - lvm_name, lock); -#endif + P_IOCTL("lvm_do_lock_lvm: %s is locked by pid %d ...\n", + lvm_name, lock); spin_unlock(&lvm_lock); interruptible_sleep_on(&lvm_wait); if (current->sigpending != 0) @@ -1966,6 +1884,8 @@ int lvm_do_vg_create(int minor, void *arg) } } + lvm_do_create_devfs_entry_of_vg ( vg_ptr); + /* Second path to correct snapshot logical volumes which are not in place during first path above */ for (l = 0; l < ls; l++) { @@ -1980,18 +1900,7 @@ int lvm_do_vg_create(int minor, void *arg) } } -#ifdef CONFIG_DEVFS_FS - vg_devfs_handle[vg_ptr->vg_number] = devfs_mk_dir(0, vg_ptr->vg_name, NULL); - ch_devfs_handle[vg_ptr->vg_number] = devfs_register( - vg_devfs_handle[vg_ptr->vg_number] , "group", - DEVFS_FL_DEFAULT, LVM_CHAR_MAJOR, vg_ptr->vg_number, - S_IFCHR | S_IRUSR | S_IWUSR | S_IRGRP, - &lvm_chr_fops, NULL); -#endif - -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS lvm_do_create_proc_entry_of_vg ( vg_ptr); -#endif vfree(snap_lv_ptr); @@ -2021,25 +1930,15 @@ static int lvm_do_vg_extend(vg_t *vg_ptr, void *arg) for (p = 0; p < vg_ptr->pv_max; p++) { if ( ( pv_ptr = vg_ptr->pv[p]) == NULL) { ret = lvm_do_pv_create(arg, vg_ptr, p); - lvm_do_create_proc_entry_of_pv ( vg_ptr, pv_ptr); if ( ret != 0) return ret; - - /* We don't need the PE list - in kernel space like LVs pe_t list */ - pv_ptr->pe = NULL; - vg_ptr->pv_cur++; - vg_ptr->pv_act++; - vg_ptr->pe_total += - pv_ptr->pe_total; -#ifdef LVM_GET_INODE - /* insert a dummy inode for fs_may_mount */ - pv_ptr->inode = lvm_get_inode(pv_ptr->pv_dev); -#endif + pv_ptr = vg_ptr->pv[p]; + vg_ptr->pe_total += pv_ptr->pe_total; + lvm_do_create_proc_entry_of_pv(vg_ptr, pv_ptr); return 0; } } } -return -EPERM; + return -EPERM; } /* lvm_do_vg_extend() */ @@ -2060,10 +1959,6 @@ static int lvm_do_vg_reduce(vg_t *vg_ptr, void *arg) { strcmp(pv_ptr->pv_name, pv_name) == 0) { if (pv_ptr->lv_cur > 0) return -EPERM; - vg_ptr->pe_total -= - pv_ptr->pe_total; - vg_ptr->pv_cur--; - vg_ptr->pv_act--; lvm_do_pv_remove(vg_ptr, p); /* Make PV pointer array contiguous */ for (; p < vg_ptr->pv_max - 1; p++) @@ -2091,9 +1986,7 @@ static int lvm_do_vg_rename(vg_t *vg_ptr, void *arg) if (copy_from_user(vg_name, arg, sizeof(vg_name)) != 0) return -EFAULT; -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS lvm_do_remove_proc_entry_of_vg ( vg_ptr); -#endif strncpy ( vg_ptr->vg_name, vg_name, sizeof ( vg_name)-1); for ( l = 0; l < vg_ptr->lv_max; l++) @@ -2115,9 +2008,7 @@ static int lvm_do_vg_rename(vg_t *vg_ptr, void *arg) strncpy(pv_ptr->vg_name, vg_name, NAME_LEN); } -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS lvm_do_create_proc_entry_of_vg ( vg_ptr); -#endif return 0; } /* lvm_do_vg_rename */ @@ -2166,27 +2057,17 @@ static int lvm_do_vg_remove(int minor) /* free PVs */ for (i = 0; i < vg_ptr->pv_max; i++) { if ((pv_ptr = vg_ptr->pv[i]) != NULL) { -#ifdef DEBUG_KFREE - printk(KERN_DEBUG - "%s -- kfree %d\n", lvm_name, __LINE__); -#endif + P_KFREE("%s -- kfree %d\n", lvm_name, __LINE__); lvm_do_pv_remove(vg_ptr, i); } } -#ifdef CONFIG_DEVFS_FS devfs_unregister (ch_devfs_handle[vg_ptr->vg_number]); devfs_unregister (vg_devfs_handle[vg_ptr->vg_number]); -#endif -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS lvm_do_remove_proc_entry_of_vg ( vg_ptr); -#endif - -#ifdef DEBUG_KFREE - printk(KERN_DEBUG "%s -- kfree %d\n", lvm_name, __LINE__); -#endif + P_KFREE("%s -- kfree %d\n", lvm_name, __LINE__); kfree(vg_ptr); vg[VG_CHR(minor)] = NULL; @@ -2222,11 +2103,6 @@ static int lvm_do_pv_create(pv_t *pvp, vg_t *vg_ptr, ulong p) { vg_ptr->pv_act++; vg_ptr->pv_cur++; -#ifdef LVM_GET_INODE - /* insert a dummy inode for fs_may_mount */ - pv_ptr->inode = lvm_get_inode(pv_ptr->pv_dev); -#endif - return 0; } /* lvm_do_pv_create() */ @@ -2237,11 +2113,8 @@ static int lvm_do_pv_create(pv_t *pvp, vg_t *vg_ptr, ulong p) { static int lvm_do_pv_remove(vg_t *vg_ptr, ulong p) { pv_t *pv_ptr = vg_ptr->pv[p]; -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS lvm_do_remove_proc_entry_of_pv ( vg_ptr, pv_ptr); -#endif - vg_ptr->pe_total -= - pv_ptr->pe_total; + vg_ptr->pe_total -= pv_ptr->pe_total; vg_ptr->pv_cur--; vg_ptr->pv_act--; #ifdef LVM_GET_INODE @@ -2320,11 +2193,9 @@ static int lvm_do_lv_create(int minor, char *lv_name, lv_t *lv) "%s -- LV_CREATE: vmalloc error LV_CURRENT_PE of %d Byte " "at line %d\n", lvm_name, size, __LINE__); -#ifdef DEBUG_KFREE - printk(KERN_DEBUG "%s -- kfree %d\n", lvm_name, __LINE__); -#endif + P_KFREE("%s -- kfree %d\n", lvm_name, __LINE__); kfree(lv_ptr); - vg[VG_CHR(minor)]->lv[l] = NULL; + vg_ptr->lv[l] = NULL; return -ENOMEM; } if (copy_from_user(lv_ptr->lv_current_pe, pep, size)) { @@ -2354,9 +2225,8 @@ static int lvm_do_lv_create(int minor, char *lv_name, lv_t *lv) "%s -- lvm_do_lv_create: vmalloc error LV_BLOCK_EXCEPTION " "of %d byte at line %d\n", lvm_name, size, __LINE__); -#ifdef DEBUG_KFREE - printk(KERN_DEBUG "%s -- kfree %d\n", lvm_name, __LINE__); -#endif + P_KFREE("%s -- kfree %d\n", lvm_name, + __LINE__); kfree(lv_ptr); vg_ptr->lv[l] = NULL; return -ENOMEM; @@ -2364,7 +2234,7 @@ static int lvm_do_lv_create(int minor, char *lv_name, lv_t *lv) if (copy_from_user(lv_ptr->lv_block_exception, lvbe, size)) { vfree(lv_ptr->lv_block_exception); kfree(lv_ptr); - vg[VG_CHR(minor)]->lv[l] = NULL; + vg_ptr->lv[l] = NULL; return -EFAULT; } /* point to the original logical volume */ @@ -2372,33 +2242,32 @@ static int lvm_do_lv_create(int minor, char *lv_name, lv_t *lv) lv_ptr->lv_snapshot_minor = 0; lv_ptr->lv_snapshot_org = lv_ptr; - lv_ptr->lv_snapshot_prev = NULL; - /* walk thrugh the snapshot list */ - while (lv_ptr->lv_snapshot_next != NULL) - lv_ptr = lv_ptr->lv_snapshot_next; - /* now lv_ptr points to the last existing snapshot in the chain */ - vg_ptr->lv[l]->lv_snapshot_prev = lv_ptr; /* our new one now back points to the previous last in the chain which can be the original logical volume */ lv_ptr = vg_ptr->lv[l]; /* now lv_ptr points to our new last snapshot logical volume */ - lv_ptr->lv_snapshot_org = lv_ptr->lv_snapshot_prev->lv_snapshot_org; - lv_ptr->lv_snapshot_next = NULL; lv_ptr->lv_current_pe = lv_ptr->lv_snapshot_org->lv_current_pe; + lv_ptr->lv_allocated_snapshot_le = lv_ptr->lv_allocated_le; lv_ptr->lv_allocated_le = lv_ptr->lv_snapshot_org->lv_allocated_le; lv_ptr->lv_current_le = lv_ptr->lv_snapshot_org->lv_current_le; lv_ptr->lv_size = lv_ptr->lv_snapshot_org->lv_size; lv_ptr->lv_stripes = lv_ptr->lv_snapshot_org->lv_stripes; lv_ptr->lv_stripesize = lv_ptr->lv_snapshot_org->lv_stripesize; + + /* Update the VG PE(s) used by snapshot reserve space. */ + vg_ptr->pe_allocated += lv_ptr->lv_allocated_snapshot_le; + if ((ret = lvm_snapshot_alloc(lv_ptr)) != 0) { vfree(lv_ptr->lv_block_exception); kfree(lv_ptr); - vg[VG_CHR(minor)]->lv[l] = NULL; + vg_ptr->lv[l] = NULL; return ret; } for ( e = 0; e < lv_ptr->lv_remap_ptr; e++) - lvm_hash_link (lv_ptr->lv_block_exception + e, lv_ptr->lv_block_exception[e].rdev_org, lv_ptr->lv_block_exception[e].rsector_org, lv_ptr); + lvm_hash_link (lv_ptr->lv_block_exception + e, + lv_ptr->lv_block_exception[e].rdev_org, + lv_ptr->lv_block_exception[e].rsector_org, lv_ptr); /* need to fill the COW exception table data into the page for disk i/o */ lvm_snapshot_fill_COW_page(vg_ptr, lv_ptr); @@ -2426,9 +2295,8 @@ static int lvm_do_lv_create(int minor, char *lv_name, lv_t *lv) vg_ptr->lv_cur++; lv_ptr->lv_status = lv_status_save; -#ifdef CONFIG_DEVFS_FS { - char *lv_tmp, *lv_buf = NULL; + char *lv_tmp, *lv_buf = lv->lv_name; strtok(lv->lv_name, "/"); /* /dev */ while((lv_tmp = strtok(NULL, "/")) != NULL) @@ -2440,24 +2308,29 @@ static int lvm_do_lv_create(int minor, char *lv_name, lv_t *lv) S_IFBLK | S_IRUSR | S_IWUSR | S_IRGRP, &lvm_blk_dops, NULL); } -#endif -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS lvm_do_create_proc_entry_of_lv ( vg_ptr, lv_ptr); -#endif /* optionally add our new snapshot LV */ if (lv_ptr->lv_access & LV_SNAPSHOT) { + lv_t *org = lv_ptr->lv_snapshot_org, *last; + /* sync the original logical volume */ - fsync_dev(lv_ptr->lv_snapshot_org->lv_dev); + fsync_dev(org->lv_dev); #ifdef LVM_VFS_ENHANCEMENT /* VFS function call to sync and lock the filesystem */ - fsync_dev_lockfs(lv_ptr->lv_snapshot_org->lv_dev); + fsync_dev_lockfs(org->lv_dev); #endif - lv_ptr->lv_snapshot_org->lv_access |= LV_SNAPSHOT_ORG; - lv_ptr->lv_access &= ~LV_SNAPSHOT_ORG; - /* put ourselve into the chain */ - lv_ptr->lv_snapshot_prev->lv_snapshot_next = lv_ptr; + + down(&org->lv_snapshot_sem); + org->lv_access |= LV_SNAPSHOT_ORG; + lv_ptr->lv_access &= ~LV_SNAPSHOT_ORG; /* this can only hide an userspace bug */ + + /* Link in the list of snapshot volumes */ + for (last = org; last->lv_snapshot_next; last = last->lv_snapshot_next); + lv_ptr->lv_snapshot_prev = last; + last->lv_snapshot_next = lv_ptr; + up(&org->lv_snapshot_sem); } /* activate the logical volume */ @@ -2513,6 +2386,31 @@ static int lvm_do_lv_remove(int minor, char *lv_name, int l) lv_ptr->lv_snapshot_next != NULL) return -EPERM; + if (lv_ptr->lv_access & LV_SNAPSHOT) { + /* + * Atomically make the the snapshot invisible + * to the original lv before playing with it. + */ + lv_t * org = lv_ptr->lv_snapshot_org; + down(&org->lv_snapshot_sem); + + /* remove this snapshot logical volume from the chain */ + lv_ptr->lv_snapshot_prev->lv_snapshot_next = lv_ptr->lv_snapshot_next; + if (lv_ptr->lv_snapshot_next != NULL) { + lv_ptr->lv_snapshot_next->lv_snapshot_prev = + lv_ptr->lv_snapshot_prev; + } + up(&org->lv_snapshot_sem); + + /* no more snapshots? */ + if (!org->lv_snapshot_next) + org->lv_access &= ~LV_SNAPSHOT_ORG; + lvm_snapshot_release(lv_ptr); + + /* Update the VG PE(s) used by snapshot reserve space. */ + vg_ptr->pe_allocated -= lv_ptr->lv_allocated_snapshot_le; + } + lv_ptr->lv_status |= LV_SPINDOWN; /* sync the buffers */ @@ -2532,7 +2430,8 @@ static int lvm_do_lv_remove(int minor, char *lv_name, int l) vg_lv_map[MINOR(lv_ptr->lv_dev)].vg_number = ABS_MAX_VG; vg_lv_map[MINOR(lv_ptr->lv_dev)].lv_number = -1; - /* correct the PE count in PVs if this is no snapshot logical volume */ + /* correct the PE count in PVs if this is not a snapshot + logical volume */ if (!(lv_ptr->lv_access & LV_SNAPSHOT)) { /* only if this is no snapshot logical volume because we share the lv_current_pe[] structs with the @@ -2546,31 +2445,13 @@ static int lvm_do_lv_remove(int minor, char *lv_name, int l) } } vfree(lv_ptr->lv_current_pe); - /* LV_SNAPSHOT */ - } else { - /* remove this snapshot logical volume from the chain */ - lv_ptr->lv_snapshot_prev->lv_snapshot_next = lv_ptr->lv_snapshot_next; - if (lv_ptr->lv_snapshot_next != NULL) { - lv_ptr->lv_snapshot_next->lv_snapshot_prev = - lv_ptr->lv_snapshot_prev; - } - /* no more snapshots? */ - if (lv_ptr->lv_snapshot_org->lv_snapshot_next == NULL) - lv_ptr->lv_snapshot_org->lv_access &= ~LV_SNAPSHOT_ORG; - lvm_snapshot_release(lv_ptr); } -#ifdef CONFIG_DEVFS_FS devfs_unregister(lv_devfs_handle[lv_ptr->lv_number]); -#endif -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS lvm_do_remove_proc_entry_of_lv ( vg_ptr, lv_ptr); -#endif -#ifdef DEBUG_KFREE - printk(KERN_DEBUG "%s -- kfree %d\n", lvm_name, __LINE__); -#endif + P_KFREE("%s -- kfree %d\n", lvm_name, __LINE__); kfree(lv_ptr); vg_ptr->lv[l] = NULL; vg_ptr->lv_cur--; @@ -2638,21 +2519,24 @@ static int lvm_do_lv_extend_reduce(int minor, char *lv_name, lv_t *lv) } memcpy(lvbe, lv_ptr->lv_block_exception, - (lv->lv_remap_end > lv_ptr->lv_remap_end ? lv_ptr->lv_remap_ptr : lv->lv_remap_end) * sizeof(lv_block_exception_t)); + (lv->lv_remap_end > lv_ptr->lv_remap_end ? + lv_ptr->lv_remap_ptr : lv->lv_remap_end) * sizeof(lv_block_exception_t)); lv_ptr->lv_block_exception = lvbe; lv_ptr->lv_remap_end = lv->lv_remap_end; if (lvm_snapshot_alloc_hash_table(lv_ptr) != 0) { - lvm_drop_snapshot(lv_ptr, "hash_alloc"); + lvm_drop_snapshot(lv_ptr, "no memory for hash table"); up(&lv_ptr->lv_snapshot_org->lv_snapshot_sem); vfree(lvbe_old); vfree(lvs_hash_table_old); - return 1; + return -ENOMEM; } for (e = 0; e < lv_ptr->lv_remap_ptr; e++) - lvm_hash_link (lv_ptr->lv_block_exception + e, lv_ptr->lv_block_exception[e].rdev_org, lv_ptr->lv_block_exception[e].rsector_org, lv_ptr); + lvm_hash_link (lv_ptr->lv_block_exception + e, + lv_ptr->lv_block_exception[e].rdev_org, + lv_ptr->lv_block_exception[e].rsector_org, lv_ptr); up(&lv_ptr->lv_snapshot_org->lv_snapshot_sem); @@ -2677,15 +2561,6 @@ static int lvm_do_lv_extend_reduce(int minor, char *lv_name, lv_t *lv) return -EFAULT; } -#ifdef DEBUG - printk(KERN_DEBUG - "%s -- fsync_dev and " - "invalidate_buffers for %s [%s] in %s\n", - lvm_name, lv_ptr->lv_name, - kdevname(lv_ptr->lv_dev), - vg_ptr->vg_name); -#endif - /* reduce allocation counters on PV(s) */ for (le = 0; le < lv_ptr->lv_allocated_le; le++) { vg_ptr->pe_allocated--; @@ -2714,9 +2589,6 @@ static int lvm_do_lv_extend_reduce(int minor, char *lv_name, lv_t *lv) /* save # of old allocated logical extents */ old_allocated_le = lv_ptr->lv_allocated_le; - /* in case of shrinking -> let's flush */ - if ( end > lv->lv_current_le) fsync_dev(lv_ptr->lv_dev); - /* copy preloaded LV */ memcpy((char *) lv_ptr, (char *) lv, sizeof(lv_t)); @@ -2914,15 +2786,11 @@ static int lvm_do_lv_rename(vg_t *vg_ptr, lv_req_t *lv_req, lv_t *lv) if ( (lv_ptr = vg_ptr->lv[l]) == NULL) continue; if (lv_ptr->lv_dev == lv->lv_dev) { -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS lvm_do_remove_proc_entry_of_lv ( vg_ptr, lv_ptr); -#endif strncpy(lv_ptr->lv_name, lv_req->lv_name, NAME_LEN); -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS lvm_do_create_proc_entry_of_lv ( vg_ptr, lv_ptr); -#endif break; } } @@ -3004,9 +2872,22 @@ static int lvm_do_pv_status(vg_t *vg_ptr, void *arg) /* + * create a devfs entry for a volume group + */ +void lvm_do_create_devfs_entry_of_vg ( vg_t *vg_ptr) { + vg_devfs_handle[vg_ptr->vg_number] = devfs_mk_dir(0, vg_ptr->vg_name, NULL); + ch_devfs_handle[vg_ptr->vg_number] = devfs_register( + vg_devfs_handle[vg_ptr->vg_number] , "group", + DEVFS_FL_DEFAULT, LVM_CHAR_MAJOR, vg_ptr->vg_number, + S_IFCHR | S_IRUSR | S_IWUSR | S_IRGRP, + &lvm_chr_fops, NULL); +} + + +/* * create a /proc entry for a logical volume */ -inline void lvm_do_create_proc_entry_of_lv ( vg_t *vg_ptr, lv_t *lv_ptr) { +void lvm_do_create_proc_entry_of_lv ( vg_t *vg_ptr, lv_t *lv_ptr) { char *basename; if ( vg_ptr->lv_subdir_pde != NULL) { @@ -3026,7 +2907,7 @@ inline void lvm_do_create_proc_entry_of_lv ( vg_t *vg_ptr, lv_t *lv_ptr) { /* * remove a /proc entry for a logical volume */ -inline void lvm_do_remove_proc_entry_of_lv ( vg_t *vg_ptr, lv_t *lv_ptr) { +void lvm_do_remove_proc_entry_of_lv ( vg_t *vg_ptr, lv_t *lv_ptr) { char *basename; if ( vg_ptr->lv_subdir_pde != NULL) { @@ -3041,13 +2922,17 @@ inline void lvm_do_remove_proc_entry_of_lv ( vg_t *vg_ptr, lv_t *lv_ptr) { /* * create a /proc entry for a physical volume */ -inline void lvm_do_create_proc_entry_of_pv ( vg_t *vg_ptr, pv_t *pv_ptr) { +void lvm_do_create_proc_entry_of_pv ( vg_t *vg_ptr, pv_t *pv_ptr) { + int offset = 0; char *basename; - - basename = strrchr(pv_ptr->pv_name, '/'); - if (basename == NULL) basename = pv_ptr->pv_name; - else basename++; - pde = create_proc_entry(basename, S_IFREG, vg_ptr->pv_subdir_pde); + char buffer[NAME_LEN]; + + basename = pv_ptr->pv_name; + if (strncmp(basename, "/dev/", 5) == 0) offset = 5; + strncpy(buffer, basename + offset, sizeof(buffer)); + basename = buffer; + while ( ( basename = strchr ( basename, '/')) != NULL) *basename = '_'; + pde = create_proc_entry(buffer, S_IFREG, vg_ptr->pv_subdir_pde); if ( pde != NULL) { pde->read_proc = lvm_proc_read_pv_info; pde->data = pv_ptr; @@ -3058,7 +2943,7 @@ inline void lvm_do_create_proc_entry_of_pv ( vg_t *vg_ptr, pv_t *pv_ptr) { /* * remove a /proc entry for a physical volume */ -inline void lvm_do_remove_proc_entry_of_pv ( vg_t *vg_ptr, pv_t *pv_ptr) { +void lvm_do_remove_proc_entry_of_pv ( vg_t *vg_ptr, pv_t *pv_ptr) { char *basename; basename = strrchr(pv_ptr->pv_name, '/'); @@ -3074,7 +2959,6 @@ inline void lvm_do_remove_proc_entry_of_pv ( vg_t *vg_ptr, pv_t *pv_ptr) { /* * create a /proc entry for a volume group */ -#if defined CONFIG_LVM_PROC_FS && defined CONFIG_PROC_FS void lvm_do_create_proc_entry_of_vg ( vg_t *vg_ptr) { int l, p; pv_t *pv_ptr; @@ -3090,24 +2974,25 @@ void lvm_do_create_proc_entry_of_vg ( vg_t *vg_ptr) { pde->read_proc = lvm_proc_read_vg_info; pde->data = vg_ptr; } - vg_ptr->lv_subdir_pde = - create_proc_entry(LVM_LV_SUBDIR, S_IFDIR, - vg_ptr->vg_dir_pde); - vg_ptr->pv_subdir_pde = - create_proc_entry(LVM_PV_SUBDIR, S_IFDIR, - vg_ptr->vg_dir_pde); - } - - if ( vg_ptr->pv_subdir_pde != NULL) { - for ( l = 0; l < vg_ptr->lv_max; l++) { - if ( ( lv_ptr = vg_ptr->lv[l]) == NULL) continue; - lvm_do_create_proc_entry_of_lv ( vg_ptr, lv_ptr); - } - for ( p = 0; p < vg_ptr->pv_max; p++) { - if ( ( pv_ptr = vg_ptr->pv[p]) == NULL) continue; - lvm_do_create_proc_entry_of_pv ( vg_ptr, pv_ptr); - } - } + pde = create_proc_entry(LVM_LV_SUBDIR, S_IFDIR, + vg_ptr->vg_dir_pde); + if ( pde != NULL) { + vg_ptr->lv_subdir_pde = pde; + for ( l = 0; l < vg_ptr->lv_max; l++) { + if ( ( lv_ptr = vg_ptr->lv[l]) == NULL) continue; + lvm_do_create_proc_entry_of_lv ( vg_ptr, lv_ptr); + } + } + pde = create_proc_entry(LVM_PV_SUBDIR, S_IFDIR, + vg_ptr->vg_dir_pde); + if ( pde != NULL) { + vg_ptr->pv_subdir_pde = pde; + for ( p = 0; p < vg_ptr->pv_max; p++) { + if ( ( pv_ptr = vg_ptr->pv[p]) == NULL) continue; + lvm_do_create_proc_entry_of_pv ( vg_ptr, pv_ptr); + } + } + } } /* @@ -3133,18 +3018,12 @@ void lvm_do_remove_proc_entry_of_vg ( vg_t *vg_ptr) { remove_proc_entry(vg_ptr->vg_name, lvm_proc_vg_subdir); } } -#endif /* * support function initialize gendisk variables */ -#ifdef __initfunc -__initfunc(void lvm_geninit(struct gendisk *lvm_gdisk)) -#else -void __init - lvm_geninit(struct gendisk *lvm_gdisk) -#endif +void __init lvm_geninit(struct gendisk *lvm_gdisk) { int i = 0; @@ -3166,39 +3045,30 @@ void __init } /* lvm_gen_init() */ -#ifdef LVM_GET_INODE /* - * support function to get an empty inode - * - * Gets an empty inode to be inserted into the inode hash, - * so that a physical volume can't be mounted. - * This is analog to drivers/block/md.c - * - * Is this the real thing? - * + * return a pointer to a '-' padded uuid */ -struct inode *lvm_get_inode(int dev) -{ - struct inode *inode_this = NULL; +static char *lvm_show_uuid ( char *uuidstr) { + int i, j; + static char uuid[NAME_LEN] = { 0, }; - /* Lock the device by inserting a dummy inode. */ - inode_this = get_empty_inode(); - inode_this->i_dev = dev; - insert_inode_hash(inode_this); - return inode_this; -} + memset ( uuid, 0, NAME_LEN); + i = 6; + memcpy ( uuid, uuidstr, i); + uuidstr += i; -/* - * support function to clear an inode - * - */ -void lvm_clear_inode(struct inode *inode) -{ -#ifdef I_FREEING - inode->i_state |= I_FREEING; -#endif - clear_inode(inode); - return; + for ( j = 0; j < 6; j++) { + uuid[i++] = '-'; + memcpy ( &uuid[i], uuidstr, 4); + uuidstr += 4; + i += 4; + } + + memcpy ( &uuid[i], uuidstr, 2 ); + + return uuid; } -#endif /* #ifdef LVM_GET_INODE */ + +module_init(lvm_init); +module_exit(lvm_cleanup); diff --git a/drivers/md/md.c b/drivers/md/md.c index 8542bc2b0..5d4bab6c9 100644 --- a/drivers/md/md.c +++ b/drivers/md/md.c @@ -2033,68 +2033,65 @@ abort: struct { int set; int noautodetect; +} raid_setup_args md__initdata; -} raid_setup_args md__initdata = { 0, 0 }; - -void md_setup_drive(void) md__init; +void md_setup_drive (void) md__init; /* * Searches all registered partitions for autorun RAID arrays * at boot time. */ -#ifdef CONFIG_AUTODETECT_RAID -static int detected_devices[128] md__initdata = { 0, }; -static int dev_cnt=0; +static int detected_devices[128] md__initdata; +static int dev_cnt; + void md_autodetect_dev(kdev_t dev) { if (dev_cnt >= 0 && dev_cnt < 127) detected_devices[dev_cnt++] = dev; } -#endif -int md__init md_run_setup(void) + +static void autostart_arrays (void) { -#ifdef CONFIG_AUTODETECT_RAID mdk_rdev_t *rdev; int i; - if (raid_setup_args.noautodetect) - printk(KERN_INFO "skipping autodetection of RAID arrays\n"); - else { - - printk(KERN_INFO "autodetecting RAID arrays\n"); + printk(KERN_INFO "autodetecting RAID arrays\n"); - for (i=0; i<dev_cnt; i++) { - kdev_t dev = detected_devices[i]; + for (i=0; i<dev_cnt; i++) { + kdev_t dev = detected_devices[i]; - if (md_import_device(dev,1)) { - printk(KERN_ALERT "could not import %s!\n", - partition_name(dev)); - continue; - } - /* - * Sanity checks: - */ - rdev = find_rdev_all(dev); - if (!rdev) { - MD_BUG(); - continue; - } - if (rdev->faulty) { - MD_BUG(); - continue; - } - md_list_add(&rdev->pending, &pending_raid_disks); + if (md_import_device(dev,1)) { + printk(KERN_ALERT "could not import %s!\n", + partition_name(dev)); + continue; } - - autorun_devices(-1); + /* + * Sanity checks: + */ + rdev = find_rdev_all(dev); + if (!rdev) { + MD_BUG(); + continue; + } + if (rdev->faulty) { + MD_BUG(); + continue; + } + md_list_add(&rdev->pending, &pending_raid_disks); } + autorun_devices(-1); +} + +int md__init md_run_setup(void) +{ + if (raid_setup_args.noautodetect) + printk(KERN_INFO "skipping autodetection of RAID arrays\n"); + else + autostart_arrays(); dev_cnt = -1; /* make sure further calls to md_autodetect_dev are ignored */ -#endif -#ifdef CONFIG_MD_BOOT md_setup_drive(); -#endif return 0; } @@ -2558,6 +2555,11 @@ static int md_ioctl (struct inode *inode, struct file *file, md_print_devices(); goto done_unlock; + case RAID_AUTORUN: + err = 0; + autostart_arrays(); + goto done; + case BLKGETSIZE: /* Return device size */ if (!arg) { err = -EINVAL; @@ -3639,14 +3641,12 @@ int md__init md_init (void) return (0); } -#ifdef CONFIG_MD_BOOT -#define MAX_MD_BOOT_DEVS 8 -struct { - unsigned long set; - int pers[MAX_MD_BOOT_DEVS]; - int chunk[MAX_MD_BOOT_DEVS]; - kdev_t devices[MAX_MD_BOOT_DEVS][MD_SB_DISKS]; -} md_setup_args md__initdata = { 0, }; +static struct { + char device_set [MAX_MD_DEVS]; + int pers[MAX_MD_DEVS]; + int chunk[MAX_MD_DEVS]; + kdev_t devices[MAX_MD_DEVS][MD_SB_DISKS]; +} md_setup_args md__initdata; /* * Parse the command-line parameters given our kernel, but do not @@ -3676,10 +3676,10 @@ static int md__init md_setup(char *str) printk("md: Too few arguments supplied to md=.\n"); return 0; } - if (minor >= MAX_MD_BOOT_DEVS) { + if (minor >= MAX_MD_DEVS) { printk ("md: Minor device number too high.\n"); return 0; - } else if (md_setup_args.set & (1 << minor)) { + } else if (md_setup_args.device_set[minor]) { printk ("md: Warning - md=%d,... has been specified twice;\n" " will discard the first definition.\n", minor); } @@ -3737,7 +3737,7 @@ static int md__init md_setup(char *str) printk ("md: Will configure md%d (%s) from %s, below.\n", minor, pername, devnames); md_setup_args.devices[minor][i] = (kdev_t) 0; - md_setup_args.set |= (1 << minor); + md_setup_args.device_set[minor] = 1; return 1; } @@ -3747,10 +3747,11 @@ void md__init md_setup_drive(void) kdev_t dev; mddev_t*mddev; - for (minor = 0; minor < MAX_MD_BOOT_DEVS; minor++) { + for (minor = 0; minor < MAX_MD_DEVS; minor++) { mdu_disk_info_t dinfo; - int err=0; - if (!(md_setup_args.set & (1 << minor))) + + int err = 0; + if (!md_setup_args.device_set[minor]) continue; printk("md: Loading md%d.\n", minor); if (mddev_map[minor].mddev) { @@ -3776,7 +3777,7 @@ void md__init md_setup_drive(void) ainfo.layout = 0; ainfo.chunk_size = md_setup_args.chunk[minor]; err = set_array_info(mddev, &ainfo); - for (i=0; !err && (dev = md_setup_args.devices[minor][i]); i++) { + for (i = 0; !err && (dev = md_setup_args.devices[minor][i]); i++) { dinfo.number = i; dinfo.raid_disk = i; dinfo.state = (1<<MD_DISK_ACTIVE)|(1<<MD_DISK_SYNC); @@ -3807,7 +3808,6 @@ void md__init md_setup_drive(void) } __setup("md=", md_setup); -#endif #ifdef MODULE int init_module (void) @@ -3859,9 +3859,7 @@ void cleanup_module (void) #endif __initcall(md_init); -#if defined(CONFIG_AUTODETECT_RAID) || defined(CONFIG_MD_BOOT) __initcall(md_run_setup); -#endif MD_EXPORT_SYMBOL(md_size); MD_EXPORT_SYMBOL(register_md_personality); diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c index 7981fe764..3ad3940a9 100644 --- a/drivers/md/raid5.c +++ b/drivers/md/raid5.c @@ -714,6 +714,11 @@ static void compute_parity(struct stripe_head *sh, int method) break; } spin_unlock_irq(&conf->device_lock); + if (count>1) { + xor_block(count, bh_ptr); + count = 1; + } + for (i = disks; i--;) if (chosen[i]) { struct buffer_head *bh = sh->bh_cache[i]; diff --git a/drivers/md/xor.c b/drivers/md/xor.c index b9b1cefe9..f0b76d466 100644 --- a/drivers/md/xor.c +++ b/drivers/md/xor.c @@ -57,8 +57,7 @@ xor_block(unsigned int count, struct buffer_head **bh_ptr) /* Set of all registered templates. */ static struct xor_block_template *template_list; -/* The -6*32 shift factor colors the cache. */ -#define BENCH_SIZE (PAGE_SIZE-6*32) +#define BENCH_SIZE (PAGE_SIZE) static void do_xor_speed(struct xor_block_template *tmpl, void *b1, void *b2) diff --git a/drivers/net/3c59x.c b/drivers/net/3c59x.c index ae4054d35..02c9204e9 100644 --- a/drivers/net/3c59x.c +++ b/drivers/net/3c59x.c @@ -118,6 +118,14 @@ LK1.1.11 13 Nov 2000 andrewm - Dump MOD_INC/DEC_USE_COUNT, use SET_MODULE_OWNER + LK1.1.12 1 Jan 2001 andrewm + - Call pci_enable_device before we request our IRQ (Tobias Ringstrom) + - Add 3c590 PCI latency timer hack to vortex_probe1 (from 0.99Ra) + - Added extended wait_for_completion for the 3c905CX. + - Look for an MII on PHY index 24 first (3c905CX oddity). + - Add HAS_NWAY to 3cSOHO100-TX (Brett Frankenberger) + - Don't free skbs we don't own on oom path in vortex_open(). + - See http://www.uow.edu.au/~andrewm/linux/#3c59x-2.3 for more details. - Also see Documentation/networking/vortex.txt */ @@ -203,7 +211,7 @@ static int rx_nocopy = 0, rx_copy = 0, queued_packet = 0, rx_csumhits; #include <linux/delay.h> static char version[] __devinitdata = -"3c59x.c:LK1.1.11 13 Nov 2000 Donald Becker and others. http://www.scyld.com/network/vortex.html " "$Revision: 1.102.2.46 $\n"; +"3c59x.c:LK1.1.12 06 Jan 2000 Donald Becker and others. http://www.scyld.com/network/vortex.html " "$Revision: 1.102.2.46 $\n"; MODULE_AUTHOR("Donald Becker <becker@scyld.com>"); MODULE_DESCRIPTION("3Com 3c59x/3c90x/3c575 series Vortex/Boomerang/Cyclone driver"); @@ -424,7 +432,7 @@ static struct vortex_chip_info { PCI_USES_IO|PCI_USES_MASTER, IS_CYCLONE, 128, }, {"3cSOHO100-TX Hurricane", - PCI_USES_IO|PCI_USES_MASTER, IS_CYCLONE, 128, }, + PCI_USES_IO|PCI_USES_MASTER, IS_CYCLONE|HAS_NWAY, 128, }, {"3c555 Laptop Hurricane", PCI_USES_IO|PCI_USES_MASTER, IS_CYCLONE|EEPROM_8BIT, 128, }, {"3c556 Laptop Tornado", @@ -843,10 +851,15 @@ static int __devinit vortex_init_one (struct pci_dev *pdev, { int rc; - rc = vortex_probe1 (pdev, pci_resource_start (pdev, 0), pdev->irq, - ent->driver_data, vortex_cards_found); - if (rc == 0) - vortex_cards_found++; + /* wake up and enable device */ + if (pci_enable_device (pdev)) { + rc = -EIO; + } else { + rc = vortex_probe1 (pdev, pci_resource_start (pdev, 0), pdev->irq, + ent->driver_data, vortex_cards_found); + if (rc == 0) + vortex_cards_found++; + } return rc; } @@ -863,7 +876,7 @@ static int __devinit vortex_probe1(struct pci_dev *pdev, struct vortex_private *vp; int option; unsigned int eeprom[0x40], checksum = 0; /* EEPROM contents */ - int i; + int i, step; struct net_device *dev; static int printed_version; int retval; @@ -889,7 +902,6 @@ static int __devinit vortex_probe1(struct pci_dev *pdev, vci->name, ioaddr); - /* private struct aligned and zeroed by init_etherdev */ vp = dev->priv; dev->base_addr = ioaddr; dev->irq = irq; @@ -908,19 +920,29 @@ static int __devinit vortex_probe1(struct pci_dev *pdev, if (pdev) { /* EISA resources already marked, so only PCI needs to do this here */ /* Ignore return value, because Cardbus drivers already allocate for us */ - if (request_region(ioaddr, vci->io_size, dev->name) != NULL) { + if (request_region(ioaddr, vci->io_size, dev->name) != NULL) vp->must_free_region = 1; - } - - /* wake up and enable device */ - if (pci_enable_device (pdev)) { - retval = -EIO; - goto free_region; - } /* enable bus-mastering if necessary */ if (vci->flags & PCI_USES_MASTER) pci_set_master (pdev); + + if (vci->drv_flags & IS_VORTEX) { + u8 pci_latency; + u8 new_latency = 248; + + /* Check the PCI latency value. On the 3c590 series the latency timer + must be set to the maximum value to avoid data corruption that occurs + when the timer expires during a transfer. This bug exists the Vortex + chip only. */ + pci_read_config_byte(pdev, PCI_LATENCY_TIMER, &pci_latency); + if (pci_latency < new_latency) { + printk(KERN_INFO "%s: Overriding PCI latency" + " timer (CFLT) setting of %d, new value is %d.\n", + dev->name, pci_latency, new_latency); + pci_write_config_byte(pdev, PCI_LATENCY_TIMER, new_latency); + } + } } spin_lock_init(&vp->lock); @@ -1025,6 +1047,13 @@ static int __devinit vortex_probe1(struct pci_dev *pdev, dev->irq); #endif + EL3WINDOW(4); + step = (inb(ioaddr + Wn4_NetDiag) & 0x1e) >> 1; + printk(KERN_INFO " product code '%c%c' rev %02x.%d date %02d-" + "%02d-%02d\n", eeprom[6]&0xff, eeprom[6]>>8, eeprom[0x14], + step, (eeprom[4]>>5) & 15, eeprom[4] & 31, eeprom[4]>>9); + + if (pdev && vci->drv_flags & HAS_CB_FNS) { unsigned long fn_st_addr; /* Cardbus function status space */ unsigned short n; @@ -1089,8 +1118,19 @@ static int __devinit vortex_probe1(struct pci_dev *pdev, mii_preamble_required++; mii_preamble_required++; mdio_read(dev, 24, 1); - for (phy = 1; phy <= 32 && phy_idx < sizeof(vp->phys); phy++) { - int mii_status, phyx = phy & 0x1f; + for (phy = 0; phy < 32 && phy_idx < 1; phy++) { + int mii_status, phyx; + + /* + * For the 3c905CX we look at index 24 first, because it bogusly + * reports an external PHY at all indices + */ + if (phy == 0) + phyx = 24; + else if (phy <= 24) + phyx = phy - 1; + else + phyx = phy; mii_status = mdio_read(dev, phyx, 1); if (mii_status && mii_status != 0xffff) { vp->phys[phy_idx++] = phyx; @@ -1135,12 +1175,13 @@ static int __devinit vortex_probe1(struct pci_dev *pdev, dev->set_multicast_list = set_rx_mode; dev->tx_timeout = vortex_tx_timeout; dev->watchdog_timeo = (watchdog * HZ) / 1000; - +// publish_netdev(dev); return 0; free_region: if (vp->must_free_region) release_region(ioaddr, vci->io_size); +// withdraw_netdev(dev); unregister_netdev(dev); kfree (dev); printk(KERN_ERR PFX "vortex_probe1 fails. Returns %d\n", retval); @@ -1150,13 +1191,23 @@ out: static void wait_for_completion(struct net_device *dev, int cmd) { - int i = 4000; + int i; outw(cmd, dev->base_addr + EL3_CMD); - while (--i > 0) { + for (i = 0; i < 2000; i++) { if (!(inw(dev->base_addr + EL3_STATUS) & CmdInProgress)) return; } + + /* OK, that didn't work. Do it the slow way. One second */ + for (i = 0; i < 100000; i++) { + if (!(inw(dev->base_addr + EL3_STATUS) & CmdInProgress)) { + printk(KERN_INFO "%s: command 0x%04x took %d usecs! Please tell andrewm@uow.edu.au\n", + dev->name, cmd, i * 10); + return; + } + udelay(10); + } printk(KERN_ERR "%s: command 0x%04x did not complete! Status=0x%x\n", dev->name, cmd, inw(dev->base_addr + EL3_STATUS)); } @@ -1331,6 +1382,7 @@ vortex_up(struct net_device *dev) set_rx_mode(dev); outw(StatsEnable, ioaddr + EL3_CMD); /* Turn on statistics. */ +// wait_for_completion(dev, SetTxStart|0x07ff); outw(RxEnable, ioaddr + EL3_CMD); /* Enable the receiver. */ outw(TxEnable, ioaddr + EL3_CMD); /* Enable transmitter. */ /* Allow status bits to be seen. */ @@ -1384,7 +1436,8 @@ vortex_open(struct net_device *dev) } if (i != RX_RING_SIZE) { int j; - for (j = 0; j < RX_RING_SIZE; j++) { + printk(KERN_EMERG "%s: no memory for rx ring\n", dev->name); + for (j = 0; j < i; j++) { if (vp->rx_skbuff[j]) { dev_kfree_skb(vp->rx_skbuff[j]); vp->rx_skbuff[j] = 0; @@ -1532,7 +1585,10 @@ static void vortex_tx_timeout(struct net_device *dev) printk(KERN_ERR "%s: transmit timed out, tx_status %2.2x status %4.4x.\n", dev->name, inb(ioaddr + TxStatus), inw(ioaddr + EL3_STATUS)); - + EL3WINDOW(4); + printk(KERN_ERR " diagnostics: net %04x media %04x dma %8.8x.\n", + inw(ioaddr + Wn4_NetDiag), inw(ioaddr + Wn4_Media), + inl(ioaddr + PktStatus)); /* Slight code bloat to be user friendly. */ if ((inb(ioaddr + TxStatus) & 0x88) == 0x88) printk(KERN_ERR "%s: Transmitter encountered 16 collisions --" @@ -1663,6 +1719,12 @@ vortex_error(struct net_device *dev, int status) dev->name, fifo_diag); /* Adapter failure requires Tx/Rx reset and reinit. */ if (vp->full_bus_master_tx) { + int bus_status = inl(ioaddr + PktStatus); + /* 0x80000000 PCI master abort. */ + /* 0x40000000 PCI target abort. */ + if (vortex_debug) + printk(KERN_ERR "%s: PCI bus error, bus status %8.8x\n", dev->name, bus_status); + /* In this case, blow the card away */ vortex_down(dev); wait_for_completion(dev, TotalReset | 0xff); diff --git a/drivers/net/Makefile b/drivers/net/Makefile index 290efb51c..5f9658fa5 100644 --- a/drivers/net/Makefile +++ b/drivers/net/Makefile @@ -26,7 +26,7 @@ ifeq ($(CONFIG_ISDN_PPP),y) obj-$(CONFIG_ISDN) += slhc.o endif -subdir-$(CONFIG_PCMCIA) += pcmcia +subdir-$(CONFIG_NET_PCMCIA) += pcmcia subdir-$(CONFIG_TULIP) += tulip subdir-$(CONFIG_IRDA) += irda subdir-$(CONFIG_TR) += tokenring diff --git a/drivers/net/depca.c b/drivers/net/depca.c index 448baa800..d5d3fa19f 100644 --- a/drivers/net/depca.c +++ b/drivers/net/depca.c @@ -1817,7 +1817,9 @@ EISA_signature(char *name, s32 eisa_id) ManCode[5]='\0'; for (i=0;(*signatures[i] != '\0') && (*name == '\0');i++) { - if (strstr(ManCode, signatures[i]) != NULL) { + const char * volatile lhs = ManCode; + const char * volatile rhs = signatures[i]; /* egcs-1.1.2 bug */ + if (strstr(lhs, rhs) != NULL) { strcpy(name,ManCode); status = 1; } diff --git a/drivers/net/dmfe.c b/drivers/net/dmfe.c index 481b28700..b160afd78 100644 --- a/drivers/net/dmfe.c +++ b/drivers/net/dmfe.c @@ -1596,10 +1596,10 @@ static int __init dmfe_init_module(void) break; } - rc = pci_register_driver(&dmfe_driver); + rc = pci_module_init(&dmfe_driver); if (rc < 0) return rc; - if (rc > 0) { + if (rc >= 0) { printk (KERN_INFO "Davicom DM91xx net driver loaded, version " DMFE_VERSION "\n"); return 0; diff --git a/drivers/net/eepro100.c b/drivers/net/eepro100.c index 298a317e2..2970fac9f 100644 --- a/drivers/net/eepro100.c +++ b/drivers/net/eepro100.c @@ -29,7 +29,7 @@ static const char *version = "eepro100.c:v1.09j-t 9/29/99 Donald Becker http://cesdis.gsfc.nasa.gov/linux/drivers/eepro100.html\n" -"eepro100.c: $Revision: 1.35 $ 2000/11/17 Modified by Andrey V. Savochkin <saw@saw.sw.com.sg> and others\n"; +"eepro100.c: $Revision: 1.36 $ 2000/11/17 Modified by Andrey V. Savochkin <saw@saw.sw.com.sg> and others\n"; /* A few user-configurable values that apply to all boards. First set is undocumented and spelled per Intel recommendations. */ @@ -698,6 +698,7 @@ static int speedo_found1(struct pci_dev *pdev, This takes less than 10usec and will easily finish before the next action. */ outl(PortReset, ioaddr + SCBPort); + inl(ioaddr + SCBPort); udelay(10); if (eeprom[3] & 0x0100) @@ -785,6 +786,7 @@ static int speedo_found1(struct pci_dev *pdev, #endif /* kernel_bloat */ outl(PortReset, ioaddr + SCBPort); + inl(ioaddr + SCBPort); udelay(10); /* Return the chip to its original power state. */ @@ -801,7 +803,7 @@ static int speedo_found1(struct pci_dev *pdev, sp->tx_ring = tx_ring_space; sp->tx_ring_dma = tx_ring_dma; sp->lstats = (struct speedo_stats *)(sp->tx_ring + TX_RING_SIZE); - sp->lstats_dma = cpu_to_le32(TX_RING_ELEM_DMA(sp, TX_RING_SIZE)); + sp->lstats_dma = TX_RING_ELEM_DMA(sp, TX_RING_SIZE); init_timer(&sp->timer); /* used in ioctl() */ sp->full_duplex = option >= 0 && (option & 0x10) ? 1 : 0; @@ -1002,7 +1004,9 @@ static void speedo_resume(struct net_device *dev) /* Set the segment registers to '0'. */ wait_for_cmd_done(ioaddr + SCBCmd); outl(0, ioaddr + SCBPointer); - inl(ioaddr + SCBPointer); /* XXX */ + /* impose a delay to avoid a bug */ + inl(ioaddr + SCBPointer); + udelay(10); outb(RxAddrLoad, ioaddr + SCBCmd); wait_for_cmd_done(ioaddr + SCBCmd); outb(CUCmdBase, ioaddr + SCBCmd); @@ -1010,7 +1014,6 @@ static void speedo_resume(struct net_device *dev) /* Load the statistics block and rx ring addresses. */ wait_for_cmd_done(ioaddr + SCBCmd); outl(sp->lstats_dma, ioaddr + SCBPointer); - inl(ioaddr + SCBPointer); /* XXX */ outb(CUStatsAddr, ioaddr + SCBCmd); sp->lstats->done_marker = 0; @@ -1045,7 +1048,7 @@ static void speedo_resume(struct net_device *dev) /* Start the chip's Tx process and unmask interrupts. */ wait_for_cmd_done(ioaddr + SCBCmd); - outl(cpu_to_le32(TX_RING_ELEM_DMA(sp, sp->dirty_tx % TX_RING_SIZE)), + outl(TX_RING_ELEM_DMA(sp, sp->dirty_tx % TX_RING_SIZE), ioaddr + SCBPointer); /* We are not ACK-ing FCP and ER in the interrupt handler yet so they should remain masked --Dragan */ @@ -1274,7 +1277,7 @@ static void speedo_tx_timeout(struct net_device *dev) /* Only the command unit has stopped. */ printk(KERN_WARNING "%s: Trying to restart the transmitter...\n", dev->name); - outl(cpu_to_le32(TX_RING_ELEM_DMA(sp, dirty_tx % TX_RING_SIZE])), + outl(TX_RING_ELEM_DMA(sp, dirty_tx % TX_RING_SIZE]), ioaddr + SCBPointer); outw(CUStart, ioaddr + SCBCmd); reset_mii(dev); diff --git a/drivers/net/hamradio/mkiss.c b/drivers/net/hamradio/mkiss.c index 7dab437e6..3ec09be5a 100644 --- a/drivers/net/hamradio/mkiss.c +++ b/drivers/net/hamradio/mkiss.c @@ -153,7 +153,7 @@ static int check_crc_flex(unsigned char *cp, int size) /* Find a free channel, and link in this `tty' line. */ static inline struct ax_disp *ax_alloc(void) { - ax25_ctrl_t *axp; + ax25_ctrl_t *axp=NULL; int i; for (i = 0; i < ax25_maxdev; i++) { diff --git a/drivers/net/hamradio/scc.c b/drivers/net/hamradio/scc.c index d65fddc41..cc31e9e30 100644 --- a/drivers/net/hamradio/scc.c +++ b/drivers/net/hamradio/scc.c @@ -136,7 +136,7 @@ /* ----------------------------------------------------------------------- */ -#undef SCC_LDELAY 1 /* slow it even a bit more down */ +#undef SCC_LDELAY /* slow it even a bit more down */ #undef SCC_DONT_CHECK /* don't look if the SCCs you specified are available */ #define SCC_MAXCHIPS 4 /* number of max. supported chips */ @@ -1776,7 +1776,7 @@ static int scc_net_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd) if (hwcfg.vector_latch) { if (!request_region(Vector_Latch, 1, "scc vector latch")) - printk(KERN_WARNING "z8530drv: warning, cannot reserve vector latch port 0x%x\n, disabled.", hwcfg.vector_latch); + printk(KERN_WARNING "z8530drv: warning, cannot reserve vector latch port 0x%x\n, disabled.", (unsigned int)hwcfg.vector_latch); else Vector_Latch = hwcfg.vector_latch; } diff --git a/drivers/net/myri_sbus.c b/drivers/net/myri_sbus.c index ba9c60250..56ca24689 100644 --- a/drivers/net/myri_sbus.c +++ b/drivers/net/myri_sbus.c @@ -892,6 +892,9 @@ static int __init myri_ether_init(struct net_device *dev, struct sbus_dev *sdev, DET(("myri_ether_init(%p,%p,%d):\n", dev, sdev, num)); dev = init_etherdev(0, sizeof(struct myri_eth)); + if (!dev) + return -ENOMEM; + if (version_printed++ == 0) printk(version); @@ -982,7 +985,7 @@ static int __init myri_ether_init(struct net_device *dev, struct sbus_dev *sdev, mp->reg_size, "MyriCOM Regs"); if (!mp->regs) { printk("MyriCOM: Cannot map MyriCOM registers.\n"); - return -ENODEV; + goto err; } mp->lanai = (unsigned short *) (mp->regs + (256 * 1024)); mp->lanai3 = (unsigned int *) mp->lanai; @@ -1059,7 +1062,7 @@ static int __init myri_ether_init(struct net_device *dev, struct sbus_dev *sdev, if (request_irq(dev->irq, &myri_interrupt, SA_SHIRQ, "MyriCOM Ethernet", (void *) dev)) { printk("MyriCOM: Cannot register interrupt handler.\n"); - return -ENODEV; + goto err; } DET(("ether_setup()\n")); @@ -1083,6 +1086,9 @@ static int __init myri_ether_init(struct net_device *dev, struct sbus_dev *sdev, root_myri_dev = mp; #endif return 0; +err: unregister_netdev(dev); + kfree(dev); + return -ENODEV; } static int __init myri_sbus_match(struct sbus_dev *sdev) diff --git a/drivers/net/pcnet32.c b/drivers/net/pcnet32.c index ccb92e8a5..fa8b951a7 100644 --- a/drivers/net/pcnet32.c +++ b/drivers/net/pcnet32.c @@ -695,6 +695,7 @@ pcnet32_probe1(unsigned long ioaddr, unsigned char irq_line, int shared, int car if (a == NULL) { printk(KERN_ERR "pcnet32: No access methods\n"); + pci_free_consistent(lp->pci_dev, sizeof(*lp), lp, lp->dma_addr); return -ENODEV; } lp->a = *a; @@ -741,6 +742,7 @@ pcnet32_probe1(unsigned long ioaddr, unsigned char irq_line, int shared, int car printk(", probed IRQ %d.\n", dev->irq); else { printk(", failed to detect IRQ line.\n"); + pci_free_consistent(lp->pci_dev, sizeof(*lp), lp, lp->dma_addr); return -ENODEV; } } diff --git a/drivers/net/ppp_async.c b/drivers/net/ppp_async.c index d6b2c47f3..c1234c5c8 100644 --- a/drivers/net/ppp_async.c +++ b/drivers/net/ppp_async.c @@ -33,13 +33,6 @@ #include <linux/init.h> #include <asm/uaccess.h> -#ifndef spin_trylock_bh -#define spin_trylock_bh(lock) ({ int __r; local_bh_disable(); \ - __r = spin_trylock(lock); \ - if (!__r) local_bh_enable(); \ - __r; }) -#endif - #define PPP_VERSION "2.4.1" #define OBUFSIZE 256 @@ -76,6 +69,7 @@ struct asyncppp { /* Bit numbers in xmit_flags */ #define XMIT_WAKEUP 0 #define XMIT_FULL 1 +#define XMIT_BUSY 2 /* State bits */ #define SC_TOSS 0x20000000 @@ -181,18 +175,14 @@ ppp_asynctty_close(struct tty_struct *tty) } /* - * Read does nothing. + * Read does nothing - no data is ever available this way. + * Pppd reads and writes packets via /dev/ppp instead. */ static ssize_t ppp_asynctty_read(struct tty_struct *tty, struct file *file, unsigned char *buf, size_t count) { - /* For now, do the same as the old 2.3.x code useta */ - struct asyncppp *ap = tty->disc_data; - - if (ap == 0) - return -ENXIO; - return ppp_channel_read(&ap->chan, file, buf, count); + return -EAGAIN; } /* @@ -203,12 +193,7 @@ static ssize_t ppp_asynctty_write(struct tty_struct *tty, struct file *file, const unsigned char *buf, size_t count) { - /* For now, do the same as the old 2.3.x code useta */ - struct asyncppp *ap = tty->disc_data; - - if (ap == 0) - return -ENXIO; - return ppp_channel_write(&ap->chan, buf, count); + return -EAGAIN; } static int @@ -259,25 +244,6 @@ ppp_asynctty_ioctl(struct tty_struct *tty, struct file *file, err = 0; break; -/* - * For now, do the same as the old 2.3 driver useta - */ - case PPPIOCGFLAGS: - case PPPIOCSFLAGS: - case PPPIOCGASYNCMAP: - case PPPIOCSASYNCMAP: - case PPPIOCGRASYNCMAP: - case PPPIOCSRASYNCMAP: - case PPPIOCGXASYNCMAP: - case PPPIOCSXASYNCMAP: - case PPPIOCGMRU: - case PPPIOCSMRU: - err = -EPERM; - if (!capable(CAP_NET_ADMIN)) - break; - err = ppp_async_ioctl(&ap->chan, cmd, arg); - break; - case PPPIOCATTACH: case PPPIOCDETACH: err = ppp_channel_ioctl(&ap->chan, cmd, arg); @@ -294,18 +260,7 @@ ppp_asynctty_ioctl(struct tty_struct *tty, struct file *file, static unsigned int ppp_asynctty_poll(struct tty_struct *tty, struct file *file, poll_table *wait) { - unsigned int mask; - struct asyncppp *ap = tty->disc_data; - - mask = POLLOUT | POLLWRNORM; -/* - * For now, do the same as the old 2.3 driver useta - */ - if (ap != 0) - mask |= ppp_channel_poll(&ap->chan, file, wait); - if (test_bit(TTY_OTHER_CLOSED, &tty->flags) || tty_hung_up_p(file)) - mask |= POLLHUP; - return mask; + return 0; } static int @@ -637,8 +592,18 @@ ppp_async_push(struct asyncppp *ap) int tty_stuffed = 0; set_bit(XMIT_WAKEUP, &ap->xmit_flags); - if (!spin_trylock_bh(&ap->xmit_lock)) + /* + * We can get called recursively here if the tty write + * function calls our wakeup function. This can happen + * for example on a pty with both the master and slave + * set to PPP line discipline. + * We use the XMIT_BUSY bit to detect this and get out, + * leaving the XMIT_WAKEUP bit set to tell the other + * instance that it may now be able to write more now. + */ + if (test_and_set_bit(XMIT_BUSY, &ap->xmit_flags)) return 0; + spin_lock_bh(&ap->xmit_lock); for (;;) { if (test_and_clear_bit(XMIT_WAKEUP, &ap->xmit_flags)) tty_stuffed = 0; @@ -653,7 +618,7 @@ ppp_async_push(struct asyncppp *ap) tty_stuffed = 1; continue; } - if (ap->optr == ap->olim && ap->tpkt != 0) { + if (ap->optr >= ap->olim && ap->tpkt != 0) { if (ppp_async_encode(ap)) { /* finished processing ap->tpkt */ clear_bit(XMIT_FULL, &ap->xmit_flags); @@ -661,17 +626,29 @@ ppp_async_push(struct asyncppp *ap) } continue; } - /* haven't made any progress */ - spin_unlock_bh(&ap->xmit_lock); + /* + * We haven't made any progress this time around. + * Clear XMIT_BUSY to let other callers in, but + * after doing so we have to check if anyone set + * XMIT_WAKEUP since we last checked it. If they + * did, we should try again to set XMIT_BUSY and go + * around again in case XMIT_BUSY was still set when + * the other caller tried. + */ + clear_bit(XMIT_BUSY, &ap->xmit_flags); + /* any more work to do? if not, exit the loop */ if (!(test_bit(XMIT_WAKEUP, &ap->xmit_flags) || (!tty_stuffed && ap->tpkt != 0))) break; - if (!spin_trylock_bh(&ap->xmit_lock)) + /* more work to do, see if we can do it now */ + if (test_and_set_bit(XMIT_BUSY, &ap->xmit_flags)) break; } + spin_unlock_bh(&ap->xmit_lock); return done; flush: + clear_bit(XMIT_BUSY, &ap->xmit_flags); if (ap->tpkt != 0) { kfree_skb(ap->tpkt); ap->tpkt = 0; diff --git a/drivers/net/pppoe.c b/drivers/net/pppoe.c index 8744f6186..7fffdf4f7 100644 --- a/drivers/net/pppoe.c +++ b/drivers/net/pppoe.c @@ -5,23 +5,24 @@ * PPPoE --- PPP over Ethernet (RFC 2516) * * - * Version: 0.6.4 + * Version: 0.6.5 * * 030700 : Fixed connect logic to allow for disconnect. - * 270700 : Fixed potential SMP problems; we must protect against - * simultaneous invocation of ppp_input + * 270700 : Fixed potential SMP problems; we must protect against + * simultaneous invocation of ppp_input * and ppp_unregister_channel. * 040800 : Respect reference count mechanisms on net-devices. * 200800 : fix kfree(skb) in pppoe_rcv (acme) * Module reference count is decremented in the right spot now, - * guards against sock_put not actually freeing the sk + * guards against sock_put not actually freeing the sk * in pppoe_release. * 051000 : Initialization cleanup. * 111100 : Fix recvmsg. + * 050101 : Fix PADT procesing. * * Author: Michal Ostrowski <mostrows@styx.uwaterloo.ca> * Contributors: - * Arnaldo Carvalho de Melo <acme@conectiva.com.br> + * Arnaldo Carvalho de Melo <acme@xconectiva.com.br> * * License: * This program is free software; you can redistribute it and/or @@ -110,7 +111,7 @@ static int hash_item(unsigned long sid, unsigned char *addr) hash ^= sid >> (i*PPPOE_HASH_BITS); return hash & ( PPPOE_HASH_SIZE - 1 ); -} +} static struct pppox_opt *item_hash_table[PPPOE_HASH_SIZE] = { 0, }; @@ -238,7 +239,7 @@ static int pppoe_device_event(struct notifier_block *this, struct net_device *dev = (struct net_device *) ptr; struct pppox_opt *po = NULL; int hash = 0; - + /* Only look at sockets that are using this specific device. */ switch (event) { case NETDEV_CHANGEMTU: @@ -255,13 +256,13 @@ static int pppoe_device_event(struct notifier_block *this, po = item_hash_table[hash]; ++hash; } - + while (po && hash < PPPOE_HASH_SIZE){ if(po->pppoe_dev == dev){ lock_sock(po->sk); if (po->sk->state & (PPPOX_CONNECTED|PPPOX_BOUND)){ pppox_unbind_sock(po->sk); - + dev_put(po->pppoe_dev); po->pppoe_dev = NULL; @@ -308,7 +309,7 @@ int pppoe_rcv_core(struct sock *sk, struct sk_buff *skb){ if (sk->state & PPPOX_BOUND) { skb_pull(skb, sizeof(struct pppoe_hdr)); - + ppp_input(&po->chan, skb); } else if( sk->state & PPPOX_RELAY ){ @@ -318,7 +319,7 @@ int pppoe_rcv_core(struct sock *sk, struct sk_buff *skb){ !( relay_po->sk->state & PPPOX_CONNECTED ) ){ goto abort; } - + skb_pull(skb, sizeof(struct pppoe_hdr)); if( !__pppoe_xmit( relay_po->sk , skb) ){ goto abort; @@ -369,7 +370,7 @@ static int pppoe_rcv(struct sk_buff *skb, }else{ ret = pppoe_rcv_core(sk, skb); } - + bh_unlock_sock(sk); sock_put(sk); return ret; @@ -412,13 +413,12 @@ static int pppoe_disc_rcv(struct sk_buff *skb, po = get_item((unsigned long) ph->sid, skb->mac.ethernet->h_source); if (!po) - goto abort_put; + goto abort; sk = po->sk; pppox_unbind_sock(sk); - abort_put: sock_put(sk); abort: kfree_skb(skb); @@ -466,9 +466,9 @@ static int pppoe_create(struct socket *sock) { int error = 0; struct sock *sk; - + MOD_INC_USE_COUNT; - + sk = sk_alloc(PF_PPPOX, GFP_KERNEL, 1); if (!sk) return -ENOMEM; @@ -528,7 +528,7 @@ int pppoe_release(struct socket *sock) po = sk->protinfo.pppox; if (po->pppoe_pa.sid) delete_item(po->pppoe_pa.sid, po->pppoe_pa.remote); - + if (po->pppoe_dev) dev_put(po->pppoe_dev); @@ -945,7 +945,7 @@ int pppoe_proc_info(char *buffer, char **start, off_t offset, int length) off_t begin = 0; int size; int i; - + len += sprintf(buffer, "Id Address Device\n"); pos = len; @@ -1025,9 +1025,10 @@ int __init pppoe_init(void) int err = register_pppox_proto(PX_PROTO_OE, &pppoe_proto); if (err == 0) { - printk(KERN_INFO "Registered PPPoE v0.6.4\n"); + printk(KERN_INFO "Registered PPPoE v0.6.5\n"); dev_add_pack(&pppoes_ptype); + dev_add_pack(&pppoed_ptype); register_netdevice_notifier(&pppoe_notifier); proc_net_create("pppoe", 0, pppoe_proc_info); } @@ -1038,6 +1039,7 @@ void __exit pppoe_exit(void) { unregister_pppox_proto(PX_PROTO_OE); dev_remove_pack(&pppoes_ptype); + dev_remove_pack(&pppoed_ptype); unregister_netdevice_notifier(&pppoe_notifier); proc_net_remove("pppoe"); } diff --git a/drivers/net/sunbmac.c b/drivers/net/sunbmac.c index 3036684f9..253548599 100644 --- a/drivers/net/sunbmac.c +++ b/drivers/net/sunbmac.c @@ -1,12 +1,9 @@ -/* $Id: sunbmac.c,v 1.21 2000/10/22 16:08:38 davem Exp $ +/* $Id: sunbmac.c,v 1.23 2001/01/20 03:36:40 davem Exp $ * sunbmac.c: Driver for Sparc BigMAC 100baseT ethernet adapters. * * Copyright (C) 1997, 1998, 1999 David S. Miller (davem@redhat.com) */ -static char *version = - "sunbmac.c:v1.9 11/Sep/99 David S. Miller (davem@redhat.com)\n"; - #include <linux/module.h> #include <linux/kernel.h> @@ -41,6 +38,9 @@ static char *version = #include "sunbmac.h" +static char version[] __initdata = + "sunbmac.c:v1.9 11/Sep/99 David S. Miller (davem@redhat.com)\n"; + #undef DEBUG_PROBE #undef DEBUG_TX #undef DEBUG_IRQ @@ -1051,10 +1051,10 @@ static void bigmac_set_multicast(struct net_device *dev) static int __init bigmac_ether_init(struct net_device *dev, struct sbus_dev *qec_sdev) { - static int version_printed = 0; - struct bigmac *bp = 0; + static int version_printed; + struct bigmac *bp; u8 bsizes, bsizes_more; - int i, res = ENOMEM; + int i; /* Get a new device struct for this interface. */ dev = init_etherdev(0, sizeof(struct bigmac)); @@ -1062,6 +1062,9 @@ static int __init bigmac_ether_init(struct net_device *dev, struct sbus_dev *qec if (version_printed++ == 0) printk(KERN_INFO "%s", version); + if (!dev) + return -ENOMEM; + /* Report what we have found to the user. */ printk(KERN_INFO "%s: BigMAC 100baseT Ethernet ", dev->name); dev->base_addr = (long) qec_sdev; @@ -1077,9 +1080,6 @@ static int __init bigmac_ether_init(struct net_device *dev, struct sbus_dev *qec spin_lock_init(&bp->lock); - /* All further failures we find return this. */ - res = ENODEV; - /* Verify the registers we expect, are actually there. */ if ((bp->bigmac_sdev->num_registers != 3) || (bp->qec_sdev->num_registers != 2)) { @@ -1205,28 +1205,25 @@ static int __init bigmac_ether_init(struct net_device *dev, struct sbus_dev *qec fail_and_cleanup: /* Something went wrong, undo whatever we did so far. */ - if (bp) { - /* Free register mappings if any. */ - if (bp->gregs) - sbus_iounmap(bp->gregs, GLOB_REG_SIZE); - if (bp->creg) - sbus_iounmap(bp->creg, CREG_REG_SIZE); - if (bp->bregs) - sbus_iounmap(bp->bregs, BMAC_REG_SIZE); - if (bp->tregs) - sbus_iounmap(bp->tregs, TCVR_REG_SIZE); - - if (bp->bmac_block) - sbus_free_consistent(bp->bigmac_sdev, - PAGE_SIZE, - bp->bmac_block, - bp->bblock_dvma); - - /* Free the BigMAC softc. */ - kfree(bp); - dev->priv = 0; - } - return res; /* Return error code. */ + /* Free register mappings if any. */ + if (bp->gregs) + sbus_iounmap(bp->gregs, GLOB_REG_SIZE); + if (bp->creg) + sbus_iounmap(bp->creg, CREG_REG_SIZE); + if (bp->bregs) + sbus_iounmap(bp->bregs, BMAC_REG_SIZE); + if (bp->tregs) + sbus_iounmap(bp->tregs, TCVR_REG_SIZE); + + if (bp->bmac_block) + sbus_free_consistent(bp->bigmac_sdev, + PAGE_SIZE, + bp->bmac_block, + bp->bblock_dvma); + + unregister_netdev(dev); + kfree(dev); + return -ENODEV; } /* QEC can be the parent of either QuadEthernet or diff --git a/drivers/net/tulip/ChangeLog b/drivers/net/tulip/ChangeLog index dfac94349..bb5b70f01 100644 --- a/drivers/net/tulip/ChangeLog +++ b/drivers/net/tulip/ChangeLog @@ -1,7 +1,37 @@ -2000-12-17 Alan Cox <alan@redhat.com> +2001-01-16 Jeff Garzik <jgarzik@mandrakesoft.com> - * merge support for the Davicom's quirks into the main tulip. Patch - by Tobias Ringstrom + * tulip_core.c: static vars no longer explicitly + initialized to zero. + * eeprom.c (tulip_read_eeprom): Make sure to delay between + EE_ENB and EE_ENB|EE_SHIFT_CLK. Merged from becker tulip.c. + +2001-01-05 Peter De Schrijver <p2@mind.be> + + * eeprom.c (tulip_parse_eeprom): Interpret a bit more of 21142 + extended format type 3 info blocks in a tulip SROM. + +2001-01-03 Matti Aarnio <matti.aarnio@zmailer.org> + + * media.c (tulip_select_media): Support media types 5 and 6 + +2001-??-?? ?? + + * tulip_core.c: Add comment about LanMedia needing + a different driver. + Enable workarounds for early PCI chipsets. + Add IA64 csr0 support, update HPPA csr0 support. + +2000-12-17 Alan Cox <alan@redhat.com> + + * eeprom.c, timer.c, tulip.h, tulip_core.c: Merge support + for the Davicom's quirks into the main tulip. + Patch by Tobias Ringstrom + +2000-11-08 Jim Studt <jim@federated.com> + + * eeprom.c (tulip_parse_eeprom): Check array bounds for + medianame[] and block_name[] arrays to avoid oops due + to bad values returned from hardware. 2000-11-02 Jeff Garzik <jgarzik@mandrakesoft.com> diff --git a/drivers/net/tulip/eeprom.c b/drivers/net/tulip/eeprom.c index 49e8bf42e..f42baefc2 100644 --- a/drivers/net/tulip/eeprom.c +++ b/drivers/net/tulip/eeprom.c @@ -207,8 +207,13 @@ subsequent_board: p += (p[0] & 0x3f) + 1; continue; } else if (p[1] & 1) { + int gpr_len, reset_len; + mtable->has_mii = 1; leaf->media = 11; + gpr_len=p[3]*2; + reset_len=p[4+gpr_len]*2; + new_advertise |= get_u16(&p[7+gpr_len+reset_len]); } else { mtable->has_nonmii = 1; leaf->media = p[2] & 0x0f; @@ -247,9 +252,9 @@ subsequent_board: } printk(KERN_INFO "%s: Index #%d - Media %s (#%d) described " "by a %s (%d) block.\n", - dev->name, i, medianame[leaf->media], leaf->media, - leaf->type >= ARRAY_SIZE(block_name) ? "UNKNOWN" : - block_name[leaf->type], leaf->type); + dev->name, i, medianame[leaf->media & 15], leaf->media, + leaf->type < ARRAY_SIZE(block_name) ? block_name[leaf->type] : "<unknown>", + leaf->type); } if (new_advertise) tp->to_advertise = new_advertise; @@ -278,6 +283,7 @@ int __devinit tulip_read_eeprom(long ioaddr, int location, int addr_len) retval = (retval << 1) | ((inl(ee_addr) & EE_DATA_READ) ? 1 : 0); } outl(EE_ENB, ee_addr); + eeprom_delay(); for (i = 16; i > 0; i--) { outl(EE_ENB | EE_SHIFT_CLK, ee_addr); diff --git a/drivers/net/tulip/media.c b/drivers/net/tulip/media.c index b8203f33a..bc8e7bbbc 100644 --- a/drivers/net/tulip/media.c +++ b/drivers/net/tulip/media.c @@ -148,7 +148,7 @@ void tulip_select_media(struct net_device *dev, int startup) long ioaddr = dev->base_addr; struct tulip_private *tp = (struct tulip_private *)dev->priv; struct mediatable *mtable = tp->mtable; - u32 new_csr6; + u32 new_csr6=0; int i; if (mtable) { @@ -265,7 +265,9 @@ void tulip_select_media(struct net_device *dev, int startup) } case 5: case 6: { u16 setup[5]; - u32 csr13val, csr14val, csr15dir, csr15val; + + new_csr6 = 0; /* FIXME */ + for (i = 0; i < 5; i++) setup[i] = get_u16(&p[i*2 + 1]); diff --git a/drivers/net/tulip/tulip_core.c b/drivers/net/tulip/tulip_core.c index 765cd6c7c..0517b90af 100644 --- a/drivers/net/tulip/tulip_core.c +++ b/drivers/net/tulip/tulip_core.c @@ -28,7 +28,7 @@ #include <asm/unaligned.h> static char version[] __devinitdata = - "Linux Tulip driver version 0.9.13 (January 2, 2001)\n"; + "Linux Tulip driver version 0.9.13a (January 20, 2001)\n"; /* A few user-configurable values. */ diff --git a/drivers/net/wan/lapbether.c b/drivers/net/wan/lapbether.c index eaba0cd34..3f23aec1c 100644 --- a/drivers/net/wan/lapbether.c +++ b/drivers/net/wan/lapbether.c @@ -45,25 +45,10 @@ #include <linux/lapb.h> #include <linux/init.h> -static char bcast_addr[6]={0xFF,0xFF,0xFF,0xFF,0xFF,0xFF}; - -static int lapbeth_rcv(struct sk_buff *, struct net_device *, struct packet_type *); -static int lapbeth_device_event(struct notifier_block *, unsigned long, void *); - -static struct packet_type lapbeth_packet_type = { - 0, /* ntohs(ETH_P_DEC),*/ - 0, /* copy */ - lapbeth_rcv, - NULL, - NULL, -}; - -static struct notifier_block lapbeth_dev_notifier = { - lapbeth_device_event, - 0 -}; - +static char bcast_addr[6] = {0xFF,0xFF,0xFF,0xFF,0xFF,0xFF}; +/* If this number is made larger, check that the temporary string buffer + * in lapbeth_new_device is large enough to store the probe device name.*/ #define MAXLAPBDEV 100 static struct lapbethdev { @@ -72,29 +57,14 @@ static struct lapbethdev { struct net_device *ethdev; /* link to ethernet device */ struct net_device axdev; /* lapbeth device (lapb#) */ struct net_device_stats stats; /* some statistics */ -} *lapbeth_devices = NULL; - +} *lapbeth_devices /* = NULL initially */; /* ------------------------------------------------------------------------ */ - -/* - * Get the ethernet device for a LAPB device - */ -#if 0 -static __inline__ struct net_device *lapbeth_get_ether_dev(struct net_device *dev) -{ - struct lapbethdev *lapbeth; - - lapbeth = (struct lapbethdev *)dev->priv; - - return (lapbeth != NULL) ? lapbeth->ethdev : NULL; -} -#endif /* * Get the LAPB device for the ethernet device */ -static __inline__ struct net_device *lapbeth_get_x25_dev(struct net_device *dev) +static inline struct net_device *lapbeth_get_x25_dev(struct net_device *dev) { struct lapbethdev *lapbeth; @@ -105,7 +75,7 @@ static __inline__ struct net_device *lapbeth_get_x25_dev(struct net_device *dev) return NULL; } -static __inline__ int dev_is_ethdev(struct net_device *dev) +static inline int dev_is_ethdev(struct net_device *dev) { return ( dev->type == ARPHRD_ETHER @@ -122,7 +92,7 @@ static int lapbeth_check_devices(struct net_device *dev) struct lapbethdev *lapbeth, *lapbeth_prev; int result = 0; unsigned long flags; - + save_flags(flags); cli(); @@ -134,7 +104,7 @@ static int lapbeth_check_devices(struct net_device *dev) lapbeth_prev->next = lapbeth->next; else lapbeth_devices = lapbeth->next; - + if (&lapbeth->axdev == dev) result = 1; @@ -145,16 +115,14 @@ static int lapbeth_check_devices(struct net_device *dev) lapbeth_prev = lapbeth; } - + restore_flags(flags); - + return result; } - /* ------------------------------------------------------------------------ */ - /* * Receive a LAPB frame via an ethernet interface. */ @@ -164,7 +132,7 @@ static int lapbeth_rcv(struct sk_buff *skb, struct net_device *dev, struct packe struct lapbethdev *lapbeth; skb->sk = NULL; /* Initially we don't know who it's for */ - + dev = lapbeth_get_x25_dev(dev); if (dev == NULL || !netif_running(dev)) { @@ -172,7 +140,7 @@ static int lapbeth_rcv(struct sk_buff *skb, struct net_device *dev, struct packe return 0; } - lapbeth = (struct lapbethdev *)dev->priv; + lapbeth = (struct lapbethdev *) dev->priv; lapbeth->stats.rx_packets++; @@ -191,7 +159,7 @@ static int lapbeth_rcv(struct sk_buff *skb, struct net_device *dev, struct packe static int lapbeth_data_indication(void *token, struct sk_buff *skb) { - struct lapbethdev *lapbeth = (struct lapbethdev *)token; + struct lapbethdev *lapbeth = (struct lapbethdev *) token; unsigned char *ptr; ptr = skb_push(skb, 1); @@ -206,14 +174,12 @@ static int lapbeth_data_indication(void *token, struct sk_buff *skb) } /* - * Send a LAPB frame via an ethernet interface + * Send a LAPB frame via an ethernet interface */ static int lapbeth_xmit(struct sk_buff *skb, struct net_device *dev) { - struct lapbethdev *lapbeth; + struct lapbethdev *lapbeth = (struct lapbethdev *) dev->priv; int err; - - lapbeth = (struct lapbethdev *)dev->priv; /* * Just to be *really* sure not to send anything if the interface @@ -253,10 +219,10 @@ static int lapbeth_xmit(struct sk_buff *skb, struct net_device *dev) return 0; } - + static void lapbeth_data_transmit(void *token, struct sk_buff *skb) { - struct lapbethdev *lapbeth = (struct lapbethdev *)token; + struct lapbethdev *lapbeth = (struct lapbethdev *) token; unsigned char *ptr; struct net_device *dev; int size; @@ -281,7 +247,7 @@ static void lapbeth_data_transmit(void *token, struct sk_buff *skb) static void lapbeth_connected(void *token, int reason) { - struct lapbethdev *lapbeth = (struct lapbethdev *)token; + struct lapbethdev *lapbeth = (struct lapbethdev *) token; struct sk_buff *skb; unsigned char *ptr; @@ -303,7 +269,7 @@ static void lapbeth_connected(void *token, int reason) static void lapbeth_disconnected(void *token, int reason) { - struct lapbethdev *lapbeth = (struct lapbethdev *)token; + struct lapbethdev *lapbeth = (struct lapbethdev *) token; struct sk_buff *skb; unsigned char *ptr; @@ -328,10 +294,7 @@ static void lapbeth_disconnected(void *token, int reason) */ static struct net_device_stats *lapbeth_get_stats(struct net_device *dev) { - struct lapbethdev *lapbeth; - - lapbeth = (struct lapbethdev *)dev->priv; - + struct lapbethdev *lapbeth = (struct lapbethdev *) dev->priv; return &lapbeth->stats; } @@ -340,18 +303,11 @@ static struct net_device_stats *lapbeth_get_stats(struct net_device *dev) */ static int lapbeth_set_mac_address(struct net_device *dev, void *addr) { - struct sockaddr *sa = (struct sockaddr *)addr; - + struct sockaddr *sa = (struct sockaddr *) addr; memcpy(dev->dev_addr, sa->sa_data, dev->addr_len); - return 0; } -static int lapbeth_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd) -{ - return -EINVAL; -} - /* * open/close a device */ @@ -363,8 +319,8 @@ static int lapbeth_open(struct net_device *dev) if (lapbeth_check_devices(dev)) return -ENODEV; /* oops, it's gone */ - - lapbeth = (struct lapbethdev *)dev->priv; + + lapbeth = (struct lapbethdev *) dev->priv; lapbeth_callbacks.connect_confirmation = lapbeth_connected; lapbeth_callbacks.connect_indication = lapbeth_connected; @@ -378,31 +334,20 @@ static int lapbeth_open(struct net_device *dev) return -ENODEV; } - MOD_INC_USE_COUNT; netif_start_queue(dev); - return 0; } static int lapbeth_close(struct net_device *dev) { - struct lapbethdev *lapbeth; + struct lapbethdev *lapbeth = (struct lapbethdev *) dev->priv; int err; netif_stop_queue(dev); - - lapbeth = (struct lapbethdev *)dev->priv; if ((err = lapb_unregister(lapbeth)) != LAPB_OK) printk(KERN_ERR "lapbeth: lapb_unregister error - %d\n", err); - MOD_DEC_USE_COUNT; - - return 0; -} - -static int lapbeth_dev_init(struct net_device *dev) -{ return 0; } @@ -414,14 +359,14 @@ static int lapbeth_dev_init(struct net_device *dev) static int lapbeth_new_device(struct net_device *dev) { int k; - unsigned char *buf; + unsigned char buf[14]; struct lapbethdev *lapbeth, *lapbeth2; - + if ((lapbeth = kmalloc(sizeof(struct lapbethdev), GFP_KERNEL)) == NULL) return -ENOMEM; - + memset(lapbeth, 0, sizeof(struct lapbethdev)); - + dev_hold(dev); lapbeth->ethdev = dev; @@ -429,7 +374,7 @@ static int lapbeth_new_device(struct net_device *dev) strncpy(lapbeth->ethname, dev->name, sizeof(lapbeth->ethname)-1); dev = &lapbeth->axdev; - buf = kmalloc(14, GFP_KERNEL); + SET_MODULE_OWNER(dev); for (k = 0; k < MAXLAPBDEV; k++) { struct net_device *odev; @@ -445,10 +390,9 @@ static int lapbeth_new_device(struct net_device *dev) kfree(lapbeth); return -ENODEV; } - + dev->priv = (void *)lapbeth; /* pointer back */ strcpy(dev->name, buf); - dev->init = lapbeth_dev_init; if (register_netdev(dev) != 0) { dev_put(dev); @@ -463,10 +407,6 @@ static int lapbeth_new_device(struct net_device *dev) dev->stop = lapbeth_close; dev->set_mac_address = lapbeth_set_mac_address; dev->get_stats = lapbeth_get_stats; - dev->do_ioctl = lapbeth_ioctl; - - dev->flags = 0; - dev->type = ARPHRD_X25; dev->hard_header_len = 3; dev->mtu = 1000; @@ -480,20 +420,19 @@ static int lapbeth_new_device(struct net_device *dev) for (lapbeth2 = lapbeth_devices; lapbeth2->next != NULL; lapbeth2 = lapbeth2->next); lapbeth2->next = lapbeth; } - + sti(); return 0; } - /* * Handle device status changes. */ -static int lapbeth_device_event(struct notifier_block *this,unsigned long event, void *ptr) +static int lapbeth_device_event(struct notifier_block *this, unsigned long event, void *ptr) { - struct net_device *dev = (struct net_device *)ptr; - + struct net_device *dev = (struct net_device *) ptr; + if (!dev_is_ethdev(dev)) return NOTIFY_DONE; @@ -518,23 +457,28 @@ static int lapbeth_device_event(struct notifier_block *this,unsigned long event, return NOTIFY_DONE; } - /* ------------------------------------------------------------------------ */ -/* - * Initialize driver. To be called from af_ax25 if not compiled as a - * module - */ -int lapbeth_init(void) +static struct packet_type lapbeth_packet_type = { + type: __constant_htons(ETH_P_DEC), + func: lapbeth_rcv, +}; + +static struct notifier_block lapbeth_dev_notifier = { + notifier_call: lapbeth_device_event, +}; + +static const char banner[] __initdata = KERN_INFO "LAPB Ethernet driver version 0.01\n"; + +static int __init lapbeth_init_driver(void) { struct net_device *dev; - lapbeth_packet_type.type = htons(ETH_P_DEC); dev_add_pack(&lapbeth_packet_type); register_netdevice_notifier(&lapbeth_dev_notifier); - printk(KERN_INFO "LAPB Ethernet driver version 0.01\n"); + printk(banner); read_lock_bh(&dev_base_lock); for (dev = dev_base; dev != NULL; dev = dev->next) { @@ -548,19 +492,9 @@ int lapbeth_init(void) return 0; } +module_init(lapbeth_init_driver); -#ifdef MODULE -EXPORT_NO_SYMBOLS; - -MODULE_AUTHOR("Jonathan Naylor <g4klx@g4klx.demon.co.uk>"); -MODULE_DESCRIPTION("The unofficial LAPB over Ethernet driver"); - -int init_module(void) -{ - return lapbeth_init(); -} - -void cleanup_module(void) +static void __exit lapbeth_cleanup_driver(void) { struct lapbethdev *lapbeth; @@ -571,4 +505,10 @@ void cleanup_module(void) for (lapbeth = lapbeth_devices; lapbeth != NULL; lapbeth = lapbeth->next) unregister_netdev(&lapbeth->axdev); } -#endif +module_exit(lapbeth_cleanup_driver); + +EXPORT_NO_SYMBOLS; + +MODULE_AUTHOR("Jonathan Naylor <g4klx@g4klx.demon.co.uk>"); +MODULE_DESCRIPTION("The unofficial LAPB over Ethernet driver"); + diff --git a/drivers/net/wan/lmc/lmc.h b/drivers/net/wan/lmc/lmc.h index 91b9e8f00..882e58c1b 100644 --- a/drivers/net/wan/lmc/lmc.h +++ b/drivers/net/wan/lmc/lmc.h @@ -29,4 +29,5 @@ extern lmc_media_t lmc_hssi_media; static void lmcEventLog( u_int32_t EventNum, u_int32_t arg2, u_int32_t arg3 ); #endif -#endif
\ No newline at end of file +#endif + diff --git a/drivers/net/wan/lmc/lmc_media.h b/drivers/net/wan/lmc/lmc_media.h index 7cc6c1650..ddcc00403 100644 --- a/drivers/net/wan/lmc/lmc_media.h +++ b/drivers/net/wan/lmc/lmc_media.h @@ -61,4 +61,5 @@ lmc_media_t lmc_t1_media = { }; -#endif
\ No newline at end of file +#endif + diff --git a/drivers/net/wan/lmc/lmc_prot.h b/drivers/net/wan/lmc/lmc_prot.h index 859ef0f00..f3b1df9e2 100644 --- a/drivers/net/wan/lmc/lmc_prot.h +++ b/drivers/net/wan/lmc/lmc_prot.h @@ -11,4 +11,5 @@ void lmc_proto_close(lmc_softc_t *sc const) unsigned short lmc_proto_type(lmc_softc_t *sc const, struct skbuff *skb) -#endif
\ No newline at end of file +#endif + diff --git a/drivers/net/wan/lmc/lmc_proto.h b/drivers/net/wan/lmc/lmc_proto.h index 6136dfad7..080a55773 100644 --- a/drivers/net/wan/lmc/lmc_proto.h +++ b/drivers/net/wan/lmc/lmc_proto.h @@ -12,4 +12,5 @@ unsigned short lmc_proto_type(lmc_softc_t *sc, struct sk_buff *skb); void lmc_proto_netif(lmc_softc_t *sc, struct sk_buff *skb); int lmc_skb_rawpackets(char *buf, char **start, off_t offset, int len, int unused); -#endif
\ No newline at end of file +#endif + diff --git a/drivers/net/wan/sdla.c b/drivers/net/wan/sdla.c index ef489b1a6..1eb66b44e 100644 --- a/drivers/net/wan/sdla.c +++ b/drivers/net/wan/sdla.c @@ -744,7 +744,7 @@ static void sdla_receive(struct net_device *dev) struct buf_entry *pbuf; unsigned long flags; - int i, received, success, addr, buf_base, buf_top; + int i=0, received, success, addr, buf_base, buf_top; short dlci, len, len2, split; flp = dev->priv; diff --git a/drivers/pnp/isapnp_proc.c b/drivers/pnp/isapnp_proc.c index ee4c2aee3..9f3cc0fc3 100644 --- a/drivers/pnp/isapnp_proc.c +++ b/drivers/pnp/isapnp_proc.c @@ -783,7 +783,10 @@ static int isapnp_set_card(char *line) unsigned int id; char index[16], value[32]; - isapnp_info_card = NULL; + if (isapnp_info_card) { + isapnp_cfg_end(); + isapnp_info_card = NULL; + } line = isapnp_get_str(index, line, sizeof(index)); isapnp_get_str(value, line, sizeof(value)); idx = idx1 = simple_strtoul(index, NULL, 0); @@ -853,10 +856,7 @@ static int isapnp_set_device(char *line) static int isapnp_autoconfigure(void) { - if (isapnp_info_device == NULL) { - printk("isapnp: device is not set\n"); - return 0; - } + isapnp_cfg_end(); if (isapnp_info_device->active) isapnp_info_device->deactivate(isapnp_info_device); if (isapnp_info_device->prepare(isapnp_info_device) < 0) { @@ -867,6 +867,13 @@ static int isapnp_autoconfigure(void) printk("isapnp: cannot activate device"); return 0; } + if (isapnp_cfg_begin(isapnp_info_card->number, -1)<0) { + printk("isapnp: configuration start sequence for card %d failed\n", isapnp_info_card->number); + isapnp_info_card = NULL; + isapnp_info_device = NULL; + return 1; + } + isapnp_device(isapnp_info_device->devfn); return 0; } diff --git a/drivers/s390/block/dasd.c b/drivers/s390/block/dasd.c index 55e27c7d8..a48213aea 100644 --- a/drivers/s390/block/dasd.c +++ b/drivers/s390/block/dasd.c @@ -952,7 +952,6 @@ do_dasd_request (request_queue_t *queue) go = 1; while (go && !list_empty(&queue->queue_head)) { req = blkdev_entry_next_request(&queue->queue_head); - req = blkdev_entry_next_request(&queue->queue_head); di = DEVICE_NR (req->rq_dev); dasd_debug ((unsigned long) req); /* req */ dasd_debug (0xc4d90000 + /* DR## */ diff --git a/drivers/sbus/audio/Config.in b/drivers/sbus/audio/Config.in index a9419f78c..40a431f81 100644 --- a/drivers/sbus/audio/Config.in +++ b/drivers/sbus/audio/Config.in @@ -8,9 +8,11 @@ if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then comment 'Linux/SPARC audio subsystem (EXPERIMENTAL)' tristate 'Audio support (EXPERIMENTAL)' CONFIG_SPARCAUDIO - dep_tristate ' AMD7930 Lowlevel Driver' CONFIG_SPARCAUDIO_AMD7930 $CONFIG_SPARCAUDIO + if [ "$CONFIG_SPARC64" != "y" ]; then + dep_tristate ' AMD7930 Lowlevel Driver' CONFIG_SPARCAUDIO_AMD7930 $CONFIG_SPARCAUDIO + dep_tristate ' DBRI Lowlevel Driver' CONFIG_SPARCAUDIO_DBRI $CONFIG_SPARCAUDIO + fi dep_tristate ' CS4231 Lowlevel Driver' CONFIG_SPARCAUDIO_CS4231 $CONFIG_SPARCAUDIO - dep_tristate ' DBRI Lowlevel Driver' CONFIG_SPARCAUDIO_DBRI $CONFIG_SPARCAUDIO dep_tristate ' Dummy Lowlevel Driver' CONFIG_SPARCAUDIO_DUMMY $CONFIG_SPARCAUDIO endmenu fi diff --git a/drivers/sbus/audio/amd7930.c b/drivers/sbus/audio/amd7930.c index f3f593bce..1ddfb5017 100644 --- a/drivers/sbus/audio/amd7930.c +++ b/drivers/sbus/audio/amd7930.c @@ -1,4 +1,4 @@ -/* $Id: amd7930.c,v 1.24 2000/01/22 05:10:27 anton Exp $ +/* $Id: amd7930.c,v 1.25 2001/01/08 04:19:16 davem Exp $ * drivers/sbus/audio/amd7930.c * * Copyright (C) 1996,1997 Thomas K. Dyas (tdyas@eden.rutgers.edu) @@ -107,7 +107,7 @@ static __u8 mulaw2bilinear(__u8 data); static __u8 linear2mulaw(__u16 data); static __u16 mulaw2linear(__u8 data); -#if defined (AMD79C30_ISDN) || defined (LINUX_VERSION_CODE) && LINUX_VERSION_CODE > 0x200ff +#if defined (AMD79C30_ISDN) && defined (LINUX_VERSION_CODE) && LINUX_VERSION_CODE > 0x200ff #include "../../isdn/hisax/hisax.h" #include "../../isdn/hisax/isdnl1.h" #include "../../isdn/hisax/foreign.h" @@ -1131,7 +1131,7 @@ static int amd7930_ioctl(struct inode * inode, struct file * file, * */ -#if defined (AMD79C30_ISDN) || defined (LINUX_VERSION_CODE) && LINUX_VERSION_CODE > 0x200ff +#if defined (AMD79C30_ISDN) && defined (LINUX_VERSION_CODE) && LINUX_VERSION_CODE > 0x200ff static int amd7930_get_irqnum(int dev) { struct amd7930_info *info; diff --git a/drivers/sbus/audio/dbri.c b/drivers/sbus/audio/dbri.c index ad0d879b1..665bb4b31 100644 --- a/drivers/sbus/audio/dbri.c +++ b/drivers/sbus/audio/dbri.c @@ -1,4 +1,4 @@ -/* $Id: dbri.c,v 1.22 2000/10/27 07:01:38 uzi Exp $ +/* $Id: dbri.c,v 1.23 2001/01/08 04:19:16 davem Exp $ * drivers/sbus/audio/dbri.c * * Copyright (C) 1997 Rudolf Koenig (rfkoenig@immd4.informatik.uni-erlangen.de) @@ -61,7 +61,7 @@ #include <asm/audioio.h> #include "dbri.h" -#if defined(DBRI_ISDN) || defined (LINUX_VERSION_CODE) && LINUX_VERSION_CODE > 0x200ff +#if defined(DBRI_ISDN) && defined (LINUX_VERSION_CODE) && LINUX_VERSION_CODE > 0x200ff #include "../../isdn/hisax/hisax.h" #include "../../isdn/hisax/isdnl1.h" #include "../../isdn/hisax/foreign.h" @@ -2227,7 +2227,7 @@ void dbri_brecv(int dev, unsigned int chan, recv_on_pipe(dbri, 8+chan, buffer, size, callback, callback_arg); } -#if defined(DBRI_ISDN) || defined (LINUX_VERSION_CODE) && LINUX_VERSION_CODE > 0x200ff +#if defined(DBRI_ISDN) && defined (LINUX_VERSION_CODE) && LINUX_VERSION_CODE > 0x200ff struct foreign_interface dbri_foreign_interface = { dbri_get_irqnum, dbri_get_liu_state, diff --git a/drivers/sbus/char/Makefile b/drivers/sbus/char/Makefile index b3a736db5..437f2d28c 100644 --- a/drivers/sbus/char/Makefile +++ b/drivers/sbus/char/Makefile @@ -19,6 +19,7 @@ obj-$(CONFIG_PCI) += su.o pcikbd.o obj-$(CONFIG_SAB82532) += sab82532.o obj-$(CONFIG_ENVCTRL) += envctrl.o obj-$(CONFIG_DISPLAY7SEG) += display7seg.o +obj-$(CONFIG_WATCHDOG_CP1XXX) += cpwatchdog.o obj-$(CONFIG_OBP_FLASH) += flash.o obj-$(CONFIG_SUN_OPENPROMIO) += openprom.o obj-$(CONFIG_SUN_MOSTEK_RTC) += rtc.o diff --git a/drivers/sbus/char/bpp.c b/drivers/sbus/char/bpp.c index c96a141a5..2be051503 100644 --- a/drivers/sbus/char/bpp.c +++ b/drivers/sbus/char/bpp.c @@ -17,6 +17,7 @@ #include <linux/errno.h> #include <linux/sched.h> #include <linux/smp_lock.h> +#include <linux/spinlock.h> #include <linux/timer.h> #include <linux/ioport.h> #include <linux/major.h> @@ -432,6 +433,7 @@ static int terminate(unsigned minor) return 0; } +static spinlock_t bpp_open_lock = SPIN_LOCK_UNLOCKED; /* * Allow only one process to open the device at a time. @@ -439,13 +441,25 @@ static int terminate(unsigned minor) static int bpp_open(struct inode *inode, struct file *f) { unsigned minor = MINOR(inode->i_rdev); - if (minor >= BPP_NO) return -ENODEV; - if (! instances[minor].present) return -ENODEV; - if (instances[minor].opened) return -EBUSY; - - instances[minor].opened = 1; + int ret; + + spin_lock(&bpp_open_lock); + ret = 0; + if (minor >= BPP_NO) { + ret = -ENODEV; + } else { + if (! instances[minor].present) { + ret = -ENODEV; + } else { + if (instances[minor].opened) + ret = -EBUSY; + else + instances[minor].opened = 1; + } + } + spin_unlock(&bpp_open_lock); - return 0; + return ret; } /* @@ -458,12 +472,14 @@ static int bpp_release(struct inode *inode, struct file *f) { unsigned minor = MINOR(inode->i_rdev); - lock_kernel(); + spin_lock(&bpp_open_lock); instances[minor].opened = 0; if (instances[minor].mode != COMPATIBILITY) - terminate(minor); - unlock_kernel(); + terminate(minor); + + spin_unlock(&bpp_open_lock); + return 0; } diff --git a/drivers/sbus/char/cpwatchdog.c b/drivers/sbus/char/cpwatchdog.c new file mode 100644 index 000000000..c319b4d33 --- /dev/null +++ b/drivers/sbus/char/cpwatchdog.c @@ -0,0 +1,838 @@ +/* cpwatchdog.c - driver implementation for hardware watchdog + * timers found on Sun Microsystems CP1400 and CP1500 boards. + * + * This device supports both the generic Linux watchdog + * interface and Solaris-compatible ioctls as best it is + * able. + * + * NOTE: CP1400 systems appear to have a defective intr_mask + * register on the PLD, preventing the disabling of + * timer interrupts. We use a timer to periodically + * reset 'stopped' watchdogs on affected platforms. + * + * TODO: DevFS support (/dev/watchdogs/0 ... /dev/watchdogs/2) + * + * Copyright (c) 2000 Eric Brower (ebrower@usa.net) + */ + +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/version.h> +#include <linux/fs.h> +#include <linux/errno.h> +#include <linux/major.h> +#include <linux/init.h> +#include <linux/miscdevice.h> +#include <linux/sched.h> +#include <linux/interrupt.h> +#include <linux/ioport.h> +#include <linux/timer.h> +#include <asm/irq.h> +#include <asm/ebus.h> +#include <asm/oplib.h> +#include <asm/uaccess.h> + +#include <asm/watchdog.h> + +#define WD_OBPNAME "watchdog" +#define WD_BADMODEL "SUNW,501-5336" +#define WD_BTIMEOUT (jiffies + (HZ * 1000)) +#define WD_BLIMIT 0xFFFF + +#define WD0_DEVNAME "watchdog0" +#define WD1_DEVNAME "watchdog1" +#define WD2_DEVNAME "watchdog2" + +#define WD0_MINOR 212 +#define WD1_MINOR 213 +#define WD2_MINOR 214 + + +/* Internal driver definitions + */ +#define WD0_ID 0 /* Watchdog0 */ +#define WD1_ID 1 /* Watchdog1 */ +#define WD2_ID 2 /* Watchdog2 */ +#define WD_NUMDEVS 3 /* Device contains 3 timers */ + +#define WD_INTR_OFF 0 /* Interrupt disable value */ +#define WD_INTR_ON 1 /* Interrupt enable value */ + +#define WD_STAT_INIT 0x01 /* Watchdog timer is initialized */ +#define WD_STAT_BSTOP 0x02 /* Watchdog timer is brokenstopped */ +#define WD_STAT_SVCD 0x04 /* Watchdog interrupt occurred */ + +/* Register value definitions + */ +#define WD0_INTR_MASK 0x01 /* Watchdog device interrupt masks */ +#define WD1_INTR_MASK 0x02 +#define WD2_INTR_MASK 0x04 + +#define WD_S_RUNNING 0x01 /* Watchdog device status running */ +#define WD_S_EXPIRED 0x02 /* Watchdog device status expired */ + +/* Sun uses Altera PLD EPF8820ATC144-4 + * providing three hardware watchdogs: + * + * 1) RIC - sends an interrupt when triggered + * 2) XIR - asserts XIR_B_RESET when triggered, resets CPU + * 3) POR - asserts POR_B_RESET when triggered, resets CPU, backplane, board + * + *** Timer register block definition (struct wd_timer_regblk) + * + * dcntr and limit registers (halfword access): + * ------------------- + * | 15 | ...| 1 | 0 | + * ------------------- + * |- counter val -| + * ------------------- + * dcntr - Current 16-bit downcounter value. + * When downcounter reaches '0' watchdog expires. + * Reading this register resets downcounter with 'limit' value. + * limit - 16-bit countdown value in 1/10th second increments. + * Writing this register begins countdown with input value. + * Reading from this register does not affect counter. + * NOTES: After watchdog reset, dcntr and limit contain '1' + * + * status register (byte access): + * --------------------------- + * | 7 | ... | 2 | 1 | 0 | + * --------------+------------ + * |- UNUSED -| EXP | RUN | + * --------------------------- + * status- Bit 0 - Watchdog is running + * Bit 1 - Watchdog has expired + * + *** PLD register block definition (struct wd_pld_regblk) + * + * intr_mask register (byte access): + * --------------------------------- + * | 7 | ... | 3 | 2 | 1 | 0 | + * +-------------+------------------ + * |- UNUSED -| WD3 | WD2 | WD1 | + * --------------------------------- + * WD3 - 1 == Interrupt disabled for watchdog 3 + * WD2 - 1 == Interrupt disabled for watchdog 2 + * WD1 - 1 == Interrupt disabled for watchdog 1 + * + * pld_status register (byte access): + * UNKNOWN, MAGICAL MYSTERY REGISTER + * + */ +struct wd_timer_regblk { + volatile __u16 dcntr; /* down counter - hw */ + volatile __u16 dcntr_pad; + volatile __u16 limit; /* limit register - hw */ + volatile __u16 limit_pad; + volatile __u8 status; /* status register - b */ + volatile __u8 status_pad; + volatile __u16 status_pad2; + volatile __u32 pad32; /* yet more padding */ +}; + +struct wd_pld_regblk { + volatile __u8 intr_mask; /* interrupt mask - b */ + volatile __u8 intr_mask_pad; + volatile __u16 intr_mask_pad2; + volatile __u8 status; /* device status - b */ + volatile __u8 status_pad; + volatile __u16 status_pad2; +}; + +struct wd_regblk { + volatile struct wd_timer_regblk wd0_regs; + volatile struct wd_timer_regblk wd1_regs; + volatile struct wd_timer_regblk wd2_regs; + volatile struct wd_pld_regblk pld_regs; +}; + +/* Individual timer structure + */ +struct wd_timer { + __u16 timeout; + __u8 intr_mask; + unsigned char runstatus; + volatile struct wd_timer_regblk* regs; +}; + +/* Device structure + */ +struct wd_device { + int irq; + spinlock_t lock; + unsigned char isbaddoggie; /* defective PLD */ + unsigned char opt_enable; + unsigned char opt_reboot; + unsigned short opt_timeout; + unsigned char initialized; + struct wd_timer watchdog[WD_NUMDEVS]; + volatile struct wd_regblk* regs; +}; + +static struct wd_device wd_dev = { + 0, SPIN_LOCK_UNLOCKED, 0, 0, 0, 0, +}; + +struct timer_list wd_timer; + +static int wd0_timeout = 0; +static int wd1_timeout = 0; +static int wd2_timeout = 0; + +#ifdef MODULE +EXPORT_NO_SYMBOLS; + +MODULE_PARM (wd0_timeout, "i"); +MODULE_PARM_DESC(wd0_timeout, "Default watchdog0 timeout in 1/10secs"); +MODULE_PARM (wd1_timeout, "i"); +MODULE_PARM_DESC(wd1_timeout, "Default watchdog1 timeout in 1/10secs"); +MODULE_PARM (wd2_timeout, "i"); +MODULE_PARM_DESC(wd2_timeout, "Default watchdog2 timeout in 1/10secs"); + +MODULE_AUTHOR + ("Eric Brower <ebrower@usa.net>"); +MODULE_DESCRIPTION + ("Hardware watchdog driver for Sun Microsystems CP1400/1500"); +MODULE_SUPPORTED_DEVICE + ("watchdog"); +#endif /* ifdef MODULE */ + +/* Forward declarations of internal methods + */ +void wd_dumpregs(void); +void wd_interrupt(int irq, void *dev_id, struct pt_regs *regs); +void wd_toggleintr(struct wd_timer* pTimer, int enable); +void wd_pingtimer(struct wd_timer* pTimer); +void wd_starttimer(struct wd_timer* pTimer); +void wd_resetbrokentimer(struct wd_timer* pTimer); +void wd_stoptimer(struct wd_timer* pTimer); +void wd_brokentimer(unsigned long data); +int wd_getstatus(struct wd_timer* pTimer); + +/* PLD expects words to be written in LSB format, + * so we must flip all words prior to writing them to regs + */ +inline unsigned short flip_word(unsigned short word) +{ + return ((word & 0xff) << 8) | ((word >> 8) & 0xff); +} + +#define wd_writew(val, addr) (writew(flip_word(val), addr)) +#define wd_readw(addr) (flip_word(readw(addr))) +#define wd_writeb(val, addr) (writeb(val, addr)) +#define wd_readb(addr) (readb(addr)) + + +/* CP1400s seem to have broken PLD implementations-- + * the interrupt_mask register cannot be written, so + * no timer interrupts can be masked within the PLD. + */ +static inline int wd_isbroken(void) +{ + /* we could test this by read/write/read/restore + * on the interrupt mask register only if OBP + * 'watchdog-enable?' == FALSE, but it seems + * ubiquitous on CP1400s + */ + char val[32]; + prom_getproperty(prom_root_node, "model", val, sizeof(val)); + return((!strcmp(val, WD_BADMODEL)) ? 1 : 0); +} + +/* Retrieve watchdog-enable? option from OBP + * Returns 0 if false, 1 if true + */ +static inline int wd_opt_enable(void) +{ + int opt_node; + + opt_node = prom_getchild(prom_root_node); + opt_node = prom_searchsiblings(opt_node, "options"); + return((-1 == prom_getint(opt_node, "watchdog-enable?")) ? 0 : 1); +} + +/* Retrieve watchdog-reboot? option from OBP + * Returns 0 if false, 1 if true + */ +static inline int wd_opt_reboot(void) +{ + int opt_node; + + opt_node = prom_getchild(prom_root_node); + opt_node = prom_searchsiblings(opt_node, "options"); + return((-1 == prom_getint(opt_node, "watchdog-reboot?")) ? 0 : 1); +} + +/* Retrieve watchdog-timeout option from OBP + * Returns OBP value, or 0 if not located + */ +static inline int wd_opt_timeout(void) +{ + int opt_node; + char value[32]; + char *p = value; + + opt_node = prom_getchild(prom_root_node); + opt_node = prom_searchsiblings(opt_node, "options"); + opt_node = prom_getproperty(opt_node, + "watchdog-timeout", + value, + sizeof(value)); + if(-1 != opt_node) { + /* atoi implementation */ + for(opt_node = 0; /* nop */; p++) { + if(*p >= '0' && *p <= '9') { + opt_node = (10*opt_node)+(*p-'0'); + } + else { + break; + } + } + } + return((-1 == opt_node) ? (0) : (opt_node)); +} + +static int wd_open(struct inode *inode, struct file *f) +{ + switch(MINOR(inode->i_rdev)) + { + case WD0_MINOR: + f->private_data = &wd_dev.watchdog[WD0_ID]; + break; + case WD1_MINOR: + f->private_data = &wd_dev.watchdog[WD1_ID]; + break; + case WD2_MINOR: + f->private_data = &wd_dev.watchdog[WD2_ID]; + break; + default: + return(-ENODEV); + } + + /* Register IRQ on first open of device */ + if(0 == wd_dev.initialized) + { + if (request_irq(wd_dev.irq, + &wd_interrupt, + SA_SHIRQ, + WD_OBPNAME, + (void *)wd_dev.regs)) { + printk("%s: Cannot register IRQ %s\n", + WD_OBPNAME, __irq_itoa(wd_dev.irq)); + return(-EBUSY); + } + wd_dev.initialized = 1; + } + + MOD_INC_USE_COUNT; + return(0); +} + +static int wd_release(struct inode *inode, struct file *file) +{ + MOD_DEC_USE_COUNT; + return 0; +} + +static int wd_ioctl(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg) +{ + int setopt = 0; + struct wd_timer* pTimer = (struct wd_timer*)file->private_data; + struct watchdog_info info = { + 0, + 0, + "Altera EPF8820ATC144-4" + }; + + if(NULL == pTimer) { + return(-EINVAL); + } + + switch(cmd) + { + /* Generic Linux IOCTLs */ + case WDIOC_GETSUPPORT: + if(copy_to_user((struct watchdog_info *)arg, + (struct watchdog_info *)&info, + sizeof(struct watchdog_info *))) { + return(-EFAULT); + } + break; + case WDIOC_KEEPALIVE: + wd_pingtimer(pTimer); + break; + case WDIOC_SETOPTIONS: + if(copy_from_user(&setopt, (void*) arg, sizeof(unsigned int))) { + return -EFAULT; + } + if(setopt & WDIOS_DISABLECARD) { + if(wd_dev.opt_enable) { + printk( + "%s: cannot disable watchdog in ENABLED mode\n", + WD_OBPNAME); + return(-EINVAL); + } + wd_stoptimer(pTimer); + } + else if(setopt & WDIOS_ENABLECARD) { + wd_starttimer(pTimer); + } + else { + return(-EINVAL); + } + break; + /* Solaris-compatible IOCTLs */ + case WIOCGSTAT: + setopt = wd_getstatus(pTimer); + if(copy_to_user((void*)arg, &setopt, sizeof(unsigned int))) { + return(-EFAULT); + } + break; + case WIOCSTART: + wd_starttimer(pTimer); + break; + case WIOCSTOP: + if(wd_dev.opt_enable) { + printk("%s: cannot disable watchdog in ENABLED mode\n", + WD_OBPNAME); + return(-EINVAL); + } + wd_stoptimer(pTimer); + break; + default: + return(-EINVAL); + } + return(0); +} + +static ssize_t wd_write( struct file *file, + const char *buf, + size_t count, + loff_t *ppos) +{ + struct wd_timer* pTimer = (struct wd_timer*)file->private_data; + + if(NULL == pTimer) { + return(-EINVAL); + } + + wd_pingtimer(pTimer); + return(count); +} + +static ssize_t wd_read(struct file * file, char * buffer, + size_t count, loff_t *ppos) +{ +#ifdef WD_DEBUG + wd_dumpregs(); + return(0); +#else + return(-EINVAL); +#endif /* ifdef WD_DEBUG */ +} + +void wd_interrupt(int irq, void *dev_id, struct pt_regs *regs) +{ + /* Only WD0 will interrupt-- others are NMI and we won't + * see them here.... + */ + spin_lock_irq(&wd_dev.lock); + if((unsigned long)wd_dev.regs == (unsigned long)dev_id) + { + wd_stoptimer(&wd_dev.watchdog[WD0_ID]); + wd_dev.watchdog[WD0_ID].runstatus |= WD_STAT_SVCD; + } + spin_unlock_irq(&wd_dev.lock); + return; +} + +static struct file_operations wd_fops = { + owner: THIS_MODULE, + ioctl: wd_ioctl, + open: wd_open, + write: wd_write, + read: wd_read, + release: wd_release, +}; + +static struct miscdevice wd0_miscdev = { WD0_MINOR, WD0_DEVNAME, &wd_fops }; +static struct miscdevice wd1_miscdev = { WD1_MINOR, WD1_DEVNAME, &wd_fops }; +static struct miscdevice wd2_miscdev = { WD2_MINOR, WD2_DEVNAME, &wd_fops }; + +void wd_dumpregs(void) +{ + /* Reading from downcounters initiates watchdog countdown-- + * Example is included below for illustration purposes. + */ + int i; + printk("%s: dumping register values\n", WD_OBPNAME); + for(i = WD0_ID; i < WD_NUMDEVS; ++i) { + /* printk("\t%s%i: dcntr at 0x%lx: 0x%x\n", + * WD_OBPNAME, + * i, + * (unsigned long)(&wd_dev.watchdog[i].regs->dcntr), + * readw(&wd_dev.watchdog[i].regs->dcntr)); + */ + printk("\t%s%i: limit at 0x%lx: 0x%x\n", + WD_OBPNAME, + i, + (unsigned long)(&wd_dev.watchdog[i].regs->limit), + readw(&wd_dev.watchdog[i].regs->limit)); + printk("\t%s%i: status at 0x%lx: 0x%x\n", + WD_OBPNAME, + i, + (unsigned long)(&wd_dev.watchdog[i].regs->status), + readb(&wd_dev.watchdog[i].regs->status)); + printk("\t%s%i: driver status: 0x%x\n", + WD_OBPNAME, + i, + wd_getstatus(&wd_dev.watchdog[i])); + } + printk("\tintr_mask at 0x%lx: 0x%x\n", + (unsigned long)(&wd_dev.regs->pld_regs.intr_mask), + readb(&wd_dev.regs->pld_regs.intr_mask)); + printk("\tpld_status at 0x%lx: 0x%x\n", + (unsigned long)(&wd_dev.regs->pld_regs.status), + readb(&wd_dev.regs->pld_regs.status)); +} + +/* Enable or disable watchdog interrupts + * Because of the CP1400 defect this should only be + * called during initialzation or by wd_[start|stop]timer() + * + * pTimer - pointer to timer device, or NULL to indicate all timers + * enable - non-zero to enable interrupts, zero to disable + */ +void wd_toggleintr(struct wd_timer* pTimer, int enable) +{ + unsigned char curregs = wd_readb(&wd_dev.regs->pld_regs.intr_mask); + unsigned char setregs = + (NULL == pTimer) ? + (WD0_INTR_MASK | WD1_INTR_MASK | WD2_INTR_MASK) : + (pTimer->intr_mask); + + (WD_INTR_ON == enable) ? + (curregs &= ~setregs): + (curregs |= setregs); + + wd_writeb(curregs, &wd_dev.regs->pld_regs.intr_mask); + return; +} + +/* Reset countdown timer with 'limit' value and continue countdown. + * This will not start a stopped timer. + * + * pTimer - pointer to timer device + */ +void wd_pingtimer(struct wd_timer* pTimer) +{ + if(wd_readb(&pTimer->regs->status) & WD_S_RUNNING) { + wd_readb(&pTimer->regs->dcntr); + } +} + +/* Stop a running watchdog timer-- the timer actually keeps + * running, but the interrupt is masked so that no action is + * taken upon expiration. + * + * pTimer - pointer to timer device + */ +void wd_stoptimer(struct wd_timer* pTimer) +{ + if(wd_readb(&pTimer->regs->status) & WD_S_RUNNING) { + wd_toggleintr(pTimer, WD_INTR_OFF); + + if(wd_dev.isbaddoggie) { + pTimer->runstatus |= WD_STAT_BSTOP; + wd_brokentimer((unsigned long)&wd_dev); + } + } +} + +/* Start a watchdog timer with the specified limit value + * If the watchdog is running, it will be restarted with + * the provided limit value. + * + * This function will enable interrupts on the specified + * watchdog. + * + * pTimer - pointer to timer device + * limit - limit (countdown) value in 1/10th seconds + */ +void wd_starttimer(struct wd_timer* pTimer) +{ + if(wd_dev.isbaddoggie) { + pTimer->runstatus &= ~WD_STAT_BSTOP; + } + pTimer->runstatus &= ~WD_STAT_SVCD; + + wd_writew(pTimer->timeout, &pTimer->regs->limit); + wd_toggleintr(pTimer, WD_INTR_ON); +} + +/* Restarts timer with maximum limit value and + * does not unset 'brokenstop' value. + */ +void wd_resetbrokentimer(struct wd_timer* pTimer) +{ + wd_toggleintr(pTimer, WD_INTR_ON); + wd_writew(WD_BLIMIT, &pTimer->regs->limit); +} + +/* Timer device initialization helper. + * Returns 0 on success, other on failure + */ +int wd_inittimer(int whichdog) +{ + struct miscdevice *whichmisc; + volatile struct wd_timer_regblk *whichregs; + char whichident[8]; + int whichmask; + __u16 whichlimit; + + switch(whichdog) + { + case WD0_ID: + whichmisc = &wd0_miscdev; + strcpy(whichident, "RIC"); + whichregs = &wd_dev.regs->wd0_regs; + whichmask = WD0_INTR_MASK; + whichlimit= (0 == wd0_timeout) ? + (wd_dev.opt_timeout): + (wd0_timeout); + break; + case WD1_ID: + whichmisc = &wd1_miscdev; + strcpy(whichident, "XIR"); + whichregs = &wd_dev.regs->wd1_regs; + whichmask = WD1_INTR_MASK; + whichlimit= (0 == wd1_timeout) ? + (wd_dev.opt_timeout): + (wd1_timeout); + break; + case WD2_ID: + whichmisc = &wd2_miscdev; + strcpy(whichident, "POR"); + whichregs = &wd_dev.regs->wd2_regs; + whichmask = WD2_INTR_MASK; + whichlimit= (0 == wd2_timeout) ? + (wd_dev.opt_timeout): + (wd2_timeout); + break; + default: + printk("%s: %s: invalid watchdog id: %i\n", + WD_OBPNAME, __FUNCTION__, whichdog); + return(1); + } + if(0 != misc_register(whichmisc)) + { + return(1); + } + wd_dev.watchdog[whichdog].regs = whichregs; + wd_dev.watchdog[whichdog].timeout = whichlimit; + wd_dev.watchdog[whichdog].intr_mask = whichmask; + wd_dev.watchdog[whichdog].runstatus &= ~WD_STAT_BSTOP; + wd_dev.watchdog[whichdog].runstatus |= WD_STAT_INIT; + + printk("%s%i: %s hardware watchdog [%01i.%i sec] %s\n", + WD_OBPNAME, + whichdog, + whichident, + wd_dev.watchdog[whichdog].timeout / 10, + wd_dev.watchdog[whichdog].timeout % 10, + (0 != wd_dev.opt_enable) ? "in ENABLED mode" : ""); + return(0); +} + +/* Timer method called to reset stopped watchdogs-- + * because of the PLD bug on CP1400, we cannot mask + * interrupts within the PLD so me must continually + * reset the timers ad infinitum. + */ +void wd_brokentimer(unsigned long data) +{ + struct wd_device* pDev = (struct wd_device*)data; + int id, tripped = 0; + + /* kill a running timer instance, in case we + * were called directly instead of by kernel timer + */ + if(timer_pending(&wd_timer)) { + del_timer(&wd_timer); + } + + for(id = WD0_ID; id < WD_NUMDEVS; ++id) { + if(pDev->watchdog[id].runstatus & WD_STAT_BSTOP) { + ++tripped; + wd_resetbrokentimer(&pDev->watchdog[id]); + } + } + + if(tripped) { + /* there is at least one timer brokenstopped-- reschedule */ + wd_timer.expires = WD_BTIMEOUT; + add_timer(&wd_timer); + } +} + +int wd_getstatus(struct wd_timer* pTimer) +{ + unsigned char stat = wd_readb(&pTimer->regs->status); + unsigned char intr = wd_readb(&wd_dev.regs->pld_regs.intr_mask); + unsigned char ret = WD_STOPPED; + + /* determine STOPPED */ + if(0 == stat ) { + return(ret); + } + /* determine EXPIRED vs FREERUN vs RUNNING */ + else if(WD_S_EXPIRED & stat) { + ret = WD_EXPIRED; + } + else if(WD_S_RUNNING & stat) { + if(intr & pTimer->intr_mask) { + ret = WD_FREERUN; + } + else { + /* Fudge WD_EXPIRED status for defective CP1400-- + * IF timer is running + * AND brokenstop is set + * AND an interrupt has been serviced + * we are WD_EXPIRED. + * + * IF timer is running + * AND brokenstop is set + * AND no interrupt has been serviced + * we are WD_FREERUN. + */ + if(wd_dev.isbaddoggie && (pTimer->runstatus & WD_STAT_BSTOP)) { + if(pTimer->runstatus & WD_STAT_SVCD) { + ret = WD_EXPIRED; + } + else { + /* we could as well pretend we are expired */ + ret = WD_FREERUN; + } + } + else { + ret = WD_RUNNING; + } + } + } + + /* determine SERVICED */ + if(pTimer->runstatus & WD_STAT_SVCD) { + ret |= WD_SERVICED; + } + + return(ret); +} + +static int __init wd_init(void) +{ + int id; + struct linux_ebus *ebus = NULL; + struct linux_ebus_device *edev = NULL; + + for_each_ebus(ebus) { + for_each_ebusdev(edev, ebus) { + if (!strcmp(edev->prom_name, WD_OBPNAME)) + goto ebus_done; + } + } + +ebus_done: + if(!edev) { + printk("%s: unable to locate device\n", WD_OBPNAME); + return -ENODEV; + } + + wd_dev.regs = + ioremap(edev->resource[0].start, sizeof(struct wd_regblk)); + + if(NULL == wd_dev.regs) { + printk("%s: unable to map registers\n", WD_OBPNAME); + return(-ENODEV); + } + + /* initialize device structure from OBP parameters */ + wd_dev.irq = edev->irqs[0]; + wd_dev.opt_enable = wd_opt_enable(); + wd_dev.opt_reboot = wd_opt_reboot(); + wd_dev.opt_timeout = wd_opt_timeout(); + wd_dev.isbaddoggie = wd_isbroken(); + + /* disable all interrupts unless watchdog-enabled? == true */ + if(! wd_dev.opt_enable) { + wd_toggleintr(NULL, WD_INTR_OFF); + } + + /* register miscellaneous devices */ + for(id = WD0_ID; id < WD_NUMDEVS; ++id) { + if(0 != wd_inittimer(id)) { + printk("%s%i: unable to initialize\n", WD_OBPNAME, id); + } + } + + /* warn about possible defective PLD */ + if(wd_dev.isbaddoggie) { + init_timer(&wd_timer); + wd_timer.function = wd_brokentimer; + wd_timer.data = (unsigned long)&wd_dev; + wd_timer.expires = WD_BTIMEOUT; + + printk("%s: PLD defect workaround enabled for model %s\n", + WD_OBPNAME, WD_BADMODEL); + } + return(0); +} + +static void __exit wd_cleanup(void) +{ + int id; + + /* if 'watchdog-enable?' == TRUE, timers are not stopped + * when module is unloaded. All brokenstopped timers will + * also now eventually trip. + */ + for(id = WD0_ID; id < WD_NUMDEVS; ++id) { + if(WD_S_RUNNING == wd_readb(&wd_dev.watchdog[id].regs->status)) { + if(wd_dev.opt_enable) { + printk(KERN_WARNING "%s%i: timer not stopped at release\n", + WD_OBPNAME, id); + } + else { + wd_stoptimer(&wd_dev.watchdog[id]); + if(wd_dev.watchdog[id].runstatus & WD_STAT_BSTOP) { + wd_resetbrokentimer(&wd_dev.watchdog[id]); + printk(KERN_WARNING + "%s%i: defect workaround disabled at release, "\ + "timer expires in ~%01i sec\n", + WD_OBPNAME, id, + wd_readw(&wd_dev.watchdog[id].regs->limit) / 10); + } + } + } + } + + if(wd_dev.isbaddoggie && timer_pending(&wd_timer)) { + del_timer(&wd_timer); + } + if(0 != (wd_dev.watchdog[WD0_ID].runstatus & WD_STAT_INIT)) { + misc_deregister(&wd0_miscdev); + } + if(0 != (wd_dev.watchdog[WD1_ID].runstatus & WD_STAT_INIT)) { + misc_deregister(&wd1_miscdev); + } + if(0 != (wd_dev.watchdog[WD2_ID].runstatus & WD_STAT_INIT)) { + misc_deregister(&wd2_miscdev); + } + if(0 != wd_dev.initialized) { + free_irq(wd_dev.irq, (void *)wd_dev.regs); + } + iounmap(wd_dev.regs); +} + +module_init(wd_init); +module_exit(wd_cleanup); diff --git a/drivers/sbus/char/flash.c b/drivers/sbus/char/flash.c index 71809b59d..c6c927712 100644 --- a/drivers/sbus/char/flash.c +++ b/drivers/sbus/char/flash.c @@ -1,4 +1,4 @@ -/* $Id: flash.c,v 1.20 2000/11/08 04:57:49 davem Exp $ +/* $Id: flash.c,v 1.21 2001/01/11 15:29:36 davem Exp $ * flash.c: Allow mmap access to the OBP Flash, for OBP updates. * * Copyright (C) 1997 Eddie C. Dost (ecd@skynet.be) @@ -14,6 +14,7 @@ #include <linux/poll.h> #include <linux/init.h> #include <linux/smp_lock.h> +#include <linux/spinlock.h> #include <asm/system.h> #include <asm/uaccess.h> @@ -22,6 +23,7 @@ #include <asm/sbus.h> #include <asm/ebus.h> +static spinlock_t flash_lock = SPIN_LOCK_UNLOCKED; static struct { unsigned long read_base; /* Physical read address */ unsigned long write_base; /* Physical write address */ @@ -38,14 +40,14 @@ flash_mmap(struct file *file, struct vm_area_struct *vma) unsigned long addr; unsigned long size; - lock_kernel(); + spin_lock(&flash_lock); if (flash.read_base == flash.write_base) { addr = flash.read_base; size = flash.read_size; } else { if ((vma->vm_flags & VM_READ) && (vma->vm_flags & VM_WRITE)) { - unlock_kernel(); + spin_unlock(&flash_lock); return -EINVAL; } if (vma->vm_flags & VM_READ) { @@ -55,11 +57,11 @@ flash_mmap(struct file *file, struct vm_area_struct *vma) addr = flash.write_base; size = flash.write_size; } else { - unlock_kernel(); + spin_unlock(&flash_lock); return -ENXIO; } } - unlock_kernel(); + spin_unlock(&flash_lock); if ((vma->vm_pgoff << PAGE_SHIFT) > size) return -ENXIO; @@ -127,9 +129,10 @@ flash_open(struct inode *inode, struct file *file) static int flash_release(struct inode *inode, struct file *file) { - lock_kernel(); + spin_lock(&flash_lock); flash.busy = 0; - unlock_kernel(); + spin_unlock(&flash_lock); + return 0; } diff --git a/drivers/sbus/char/jsflash.c b/drivers/sbus/char/jsflash.c index f87850b2c..8faec19dd 100644 --- a/drivers/sbus/char/jsflash.c +++ b/drivers/sbus/char/jsflash.c @@ -505,9 +505,7 @@ static int jsfd_open(struct inode *inode, struct file *file) static int jsf_release(struct inode *inode, struct file *file) { - lock_kernel(); jsf0.busy = 0; - unlock_kernel(); return 0; } diff --git a/drivers/sbus/char/pcikbd.c b/drivers/sbus/char/pcikbd.c index 6ddccf114..ea37949fa 100644 --- a/drivers/sbus/char/pcikbd.c +++ b/drivers/sbus/char/pcikbd.c @@ -1,4 +1,4 @@ -/* $Id: pcikbd.c,v 1.49 2000/07/13 08:06:40 davem Exp $ +/* $Id: pcikbd.c,v 1.50 2001/01/11 15:29:36 davem Exp $ * pcikbd.c: Ultra/AX PC keyboard support. * * Copyright (C) 1997 Eddie C. Dost (ecd@skynet.be) @@ -746,13 +746,13 @@ static int aux_release(struct inode * inode, struct file * file) { unsigned long flags; - lock_kernel(); aux_fasync(-1, file, 0); - if (--aux_count) - goto out; spin_lock_irqsave(&pcikbd_lock, flags); + if (--aux_count) + goto out; + /* Disable controller ints */ aux_write_cmd(AUX_INTS_OFF); poll_aux_status(); @@ -761,9 +761,8 @@ static int aux_release(struct inode * inode, struct file * file) pcimouse_outb(KBD_CCMD_MOUSE_DISABLE, pcimouse_iobase + KBD_CNTL_REG); poll_aux_status(); - spin_unlock_irqrestore(&pcikbd_lock, flags); out: - unlock_kernel(); + spin_unlock_irqrestore(&pcikbd_lock, flags); return 0; } @@ -780,11 +779,13 @@ static int aux_open(struct inode * inode, struct file * file) if (!aux_present) return -ENODEV; - if (aux_count++) - return 0; - spin_lock_irqsave(&pcikbd_lock, flags); + if (aux_count++) { + spin_unlock_irqrestore(&pcikbd_lock, flags); + return 0; + } + if (!poll_aux_status()) { aux_count--; spin_unlock_irqrestore(&pcikbd_lock, flags); diff --git a/drivers/sbus/char/rtc.c b/drivers/sbus/char/rtc.c index d8454cf25..6105e4229 100644 --- a/drivers/sbus/char/rtc.c +++ b/drivers/sbus/char/rtc.c @@ -1,4 +1,4 @@ -/* $Id: rtc.c,v 1.23 2000/08/29 07:01:55 davem Exp $ +/* $Id: rtc.c,v 1.24 2001/01/11 15:07:09 davem Exp $ * * Linux/SPARC Real Time Clock Driver * Copyright (C) 1996 Thomas K. Dyas (tdyas@eden.rutgers.edu) @@ -31,11 +31,9 @@ static int rtc_busy = 0; void get_rtc_time(struct rtc_time *t) { unsigned long regs = mstk48t02_regs; - unsigned long flags; u8 tmp; - save_flags(flags); - cli(); + spin_lock_irq(&mostek_lock); tmp = mostek_read(regs + MOSTEK_CREG); tmp |= MSTK_CREG_READ; @@ -52,18 +50,18 @@ void get_rtc_time(struct rtc_time *t) tmp = mostek_read(regs + MOSTEK_CREG); tmp &= ~MSTK_CREG_READ; mostek_write(regs + MOSTEK_CREG, tmp); - restore_flags(flags); + + spin_unlock_irq(&mostek_lock); } /* Set the current date and time inthe real time clock. */ void set_rtc_time(struct rtc_time *t) { unsigned long regs = mstk48t02_regs; - unsigned long flags; u8 tmp; - save_flags(flags); - cli(); + spin_lock_irq(&mostek_lock); + tmp = mostek_read(regs + MOSTEK_CREG); tmp |= MSTK_CREG_WRITE; mostek_write(regs + MOSTEK_CREG, tmp); @@ -79,7 +77,8 @@ void set_rtc_time(struct rtc_time *t) tmp = mostek_read(regs + MOSTEK_CREG); tmp &= ~MSTK_CREG_WRITE; mostek_write(regs + MOSTEK_CREG, tmp); - restore_flags(flags); + + spin_unlock_irq(&mostek_lock); } static long long rtc_lseek(struct file *file, long long offset, int origin) @@ -121,20 +120,24 @@ static int rtc_ioctl(struct inode *inode, struct file *file, unsigned int cmd, static int rtc_open(struct inode *inode, struct file *file) { + int ret; + + spin_lock_irq(&mostek_lock); + if (rtc_busy) { + ret = -EBUSY; + } else { + rtc_busy = 1; + ret = 0; + } + spin_unlock_irq(&mostek_lock); - if (rtc_busy) - return -EBUSY; - - rtc_busy = 1; - - return 0; + return ret; } static int rtc_release(struct inode *inode, struct file *file) { - lock_kernel(); rtc_busy = 0; - unlock_kernel(); + return 0; } @@ -150,11 +153,7 @@ static struct miscdevice rtc_dev = { RTC_MINOR, "rtc", &rtc_fops }; EXPORT_NO_SYMBOLS; -#ifdef MODULE -int init_module(void) -#else -int __init rtc_sun_init(void) -#endif +static int __init rtc_sun_init(void) { int error; @@ -173,9 +172,10 @@ int __init rtc_sun_init(void) return 0; } -#ifdef MODULE -void cleanup_module(void) +static void __exit rtc_sun_cleanup(void) { misc_deregister(&rtc_dev); } -#endif + +module_init(rtc_sun_init); +module_exit(rtc_sun_cleanup); diff --git a/drivers/sbus/char/sunkbd.c b/drivers/sbus/char/sunkbd.c index f34c12250..467200699 100644 --- a/drivers/sbus/char/sunkbd.c +++ b/drivers/sbus/char/sunkbd.c @@ -1521,15 +1521,17 @@ kbd_ioctl (struct inode *i, struct file *f, unsigned int cmd, unsigned long arg) static int kbd_open (struct inode *i, struct file *f) { + spin_lock_irq(&kbd_queue_lock); kbd_active++; if (kbd_opened) - return 0; + goto out; kbd_opened = fg_console + 1; - spin_lock_irq(&kbd_queue_lock); kbd_head = kbd_tail = 0; + + out: spin_unlock_irq(&kbd_queue_lock); return 0; @@ -1538,7 +1540,7 @@ kbd_open (struct inode *i, struct file *f) static int kbd_close (struct inode *i, struct file *f) { - lock_kernel(); + spin_lock_irq(&kbd_queue_lock); if (!--kbd_active) { if (kbd_redirected) kbd_table [kbd_redirected-1].kbdmode = VC_XLATE; @@ -1546,7 +1548,8 @@ kbd_close (struct inode *i, struct file *f) kbd_opened = 0; kbd_fasync (-1, f, 0); } - unlock_kernel(); + spin_unlock_irq(&kbd_queue_lock); + return 0; } diff --git a/drivers/sbus/char/sunmouse.c b/drivers/sbus/char/sunmouse.c index fd92bc434..8bb45cbde 100644 --- a/drivers/sbus/char/sunmouse.c +++ b/drivers/sbus/char/sunmouse.c @@ -391,11 +391,14 @@ sun_mouse_inbyte(unsigned char byte, int is_break) static int sun_mouse_open(struct inode * inode, struct file * file) { + spin_lock_irq(&sunmouse.lock); if (sunmouse.active++) - return 0; + goto out; sunmouse.delta_x = sunmouse.delta_y = 0; sunmouse.button_state = 0x80; sunmouse.vuid_mode = VUID_NATIVE; +out: + spin_unlock_irq(&sunmouse.lock); return 0; } @@ -412,10 +415,12 @@ static int sun_mouse_fasync (int fd, struct file *filp, int on) static int sun_mouse_close(struct inode *inode, struct file *file) { - lock_kernel(); sun_mouse_fasync (-1, file, 0); + + spin_lock_irq(&sunmouse.lock); sunmouse.active--; - unlock_kernel(); + spin_unlock_irq(&sunmouse.lock); + return 0; } diff --git a/drivers/sbus/char/vfc_dev.c b/drivers/sbus/char/vfc_dev.c index 73158f351..a4c0f347e 100644 --- a/drivers/sbus/char/vfc_dev.c +++ b/drivers/sbus/char/vfc_dev.c @@ -22,6 +22,7 @@ #include <linux/sched.h> #include <linux/fs.h> #include <linux/smp_lock.h> +#include <linux/spinlock.h> #include <asm/openprom.h> #include <asm/oplib.h> @@ -181,17 +182,26 @@ struct vfc_dev *vfc_get_dev_ptr(int instance) return vfc_dev_lst[instance]; } +static spinlock_t vfc_dev_lock = SPIN_LOCK_UNLOCKED; + static int vfc_open(struct inode *inode, struct file *file) { struct vfc_dev *dev; + spin_lock(&vfc_dev_lock); dev = vfc_get_dev_ptr(MINOR(inode->i_rdev)); - if (dev == NULL) + if (dev == NULL) { + spin_unlock(&vfc_dev_lock); return -ENODEV; - if (dev->busy) + } + if (dev->busy) { + spin_unlock(&vfc_dev_lock); return -EBUSY; + } dev->busy = 1; + spin_unlock(&vfc_dev_lock); + vfc_lock_device(dev); vfc_csr_init(dev); @@ -209,14 +219,14 @@ static int vfc_release(struct inode *inode,struct file *file) { struct vfc_dev *dev; - lock_kernel(); + spin_lock(&vfc_dev_lock); dev = vfc_get_dev_ptr(MINOR(inode->i_rdev)); if (!dev || !dev->busy) { - unlock_kernel(); + spin_unlock(&vfc_dev_lock); return -EINVAL; } dev->busy = 0; - unlock_kernel(); + spin_unlock(&vfc_dev_lock); return 0; } @@ -611,12 +621,10 @@ static int vfc_mmap(struct inode *inode, struct file *file, unsigned int map_size, ret, map_offset; struct vfc_dev *dev; - lock_kernel(); dev = vfc_get_dev_ptr(MINOR(inode->i_rdev)); - if(dev == NULL) { - unlock_kernel(); + if(dev == NULL) return -ENODEV; - } + map_size = vma->vm_end - vma->vm_start; if(map_size > sizeof(struct vfc_regs)) map_size = sizeof(struct vfc_regs); @@ -626,7 +634,7 @@ static int vfc_mmap(struct inode *inode, struct file *file, map_offset = (unsigned int) (long)dev->phys_regs; ret = io_remap_page_range(vma->vm_start, map_offset, map_size, vma->vm_page_prot, dev->which_io); - unlock_kernel(); + if(ret) return -EAGAIN; diff --git a/drivers/sbus/sbus.c b/drivers/sbus/sbus.c index 942a6c59b..032776c70 100644 --- a/drivers/sbus/sbus.c +++ b/drivers/sbus/sbus.c @@ -1,4 +1,4 @@ -/* $Id: sbus.c,v 1.91 2000/11/08 05:04:06 davem Exp $ +/* $Id: sbus.c,v 1.92 2001/01/25 17:15:59 davem Exp $ * sbus.c: SBus support routines. * * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu) @@ -238,10 +238,14 @@ static void __init __apply_ranges_to_regs(struct linux_prom_ranges *ranges, break; } if (rngnum == num_ranges) { - prom_printf("sbus_apply_ranges: Cannot find matching " - "range nregs[%d] nranges[%d].\n", - num_regs, num_ranges); - prom_halt(); + /* We used to flag this as an error. Actually + * some devices do not report the regs as we expect. + * For example, see SUNW,pln device. In that case + * the reg property is in a format internal to that + * node, ie. it is not in the SBUS register space + * per se. -DaveM + */ + return; } regs[regnum].which_io = ranges[rngnum].ot_parent_space; regs[regnum].phys_addr += ranges[rngnum].ot_parent_base; diff --git a/drivers/scsi/constants.c b/drivers/scsi/constants.c index 5a74bc2df..73bd8029e 100644 --- a/drivers/scsi/constants.c +++ b/drivers/scsi/constants.c @@ -776,7 +776,7 @@ void print_sense_internal(const char * devclass, printk("%s%s: sns = %2x %2x\n", devclass, kdevname(dev), sense_buffer[0], sense_buffer[2]); - printk("Non-extended sense class %d code 0x%0x ", sense_class, code); + printk("Non-extended sense class %d code 0x%0x\n", sense_class, code); s = 4; } diff --git a/drivers/scsi/ibmmca.c b/drivers/scsi/ibmmca.c index f966bf5a7..0e5ee6935 100644 --- a/drivers/scsi/ibmmca.c +++ b/drivers/scsi/ibmmca.c @@ -16,6 +16,7 @@ #if LINUX_VERSION_CODE < KERNEL_VERSION(2,4,0) #error "This driver works only with kernel 2.4.0 or higher!" #endif +#include <linux/module.h> #include <linux/kernel.h> #include <linux/types.h> #include <linux/ctype.h> @@ -38,7 +39,7 @@ #include <linux/config.h> /* current version of this driver-source: */ -#define IBMMCA_SCSI_DRIVER_VERSION "4.0a" +#define IBMMCA_SCSI_DRIVER_VERSION "4.0b" #define IBMLOCK spin_lock_irqsave(&io_request_lock, flags); #define IBMUNLOCK spin_unlock_irqrestore(&io_request_lock, flags); @@ -443,7 +444,6 @@ static int scsi_id[IM_MAX_HOSTS] = { 7, 7, 7, 7, 7, 7, 7, 7 }; (that is kernel version 2.1.x) */ #if defined(MODULE) static char *boot_options = NULL; -#include <linux/module.h> MODULE_PARM(boot_options, "s"); MODULE_PARM(io_port, "1-" __MODULE_STRING(IM_MAX_HOSTS) "i"); MODULE_PARM(scsi_id, "1-" __MODULE_STRING(IM_MAX_HOSTS) "i"); @@ -2563,9 +2563,6 @@ static int option_setup(char *str) __setup("ibmmcascsi=", option_setup); -#ifdef MODULE -/* Eventually this will go into an include file, but this will be later */ -Scsi_Host_Template driver_template = IBMMCA; +static Scsi_Host_Template driver_template = IBMMCA; #include "scsi_module.c" -#endif diff --git a/drivers/scsi/megaraid.c b/drivers/scsi/megaraid.c index 264ae2fe0..106e39b8a 100644 --- a/drivers/scsi/megaraid.c +++ b/drivers/scsi/megaraid.c @@ -149,7 +149,6 @@ #include <linux/version.h> #ifdef MODULE -#include <linux/modversions.h> #include <linux/module.h> char kernel_version[] = UTS_RELEASE; diff --git a/drivers/scsi/ppa.c b/drivers/scsi/ppa.c index d82eed0b7..fc0af0cbf 100644 --- a/drivers/scsi/ppa.c +++ b/drivers/scsi/ppa.c @@ -222,8 +222,8 @@ int ppa_detect(Scsi_Host_Template * host) printk(" supported by the imm (ZIP Plus) driver. If the\n"); printk(" cable is marked with \"AutoDetect\", this is what has\n"); printk(" happened.\n"); - return 0; spin_lock_irq(&io_request_lock); + return 0; } try_again = 1; goto retry_entry; diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c index ace7c7e2c..dcde48437 100644 --- a/drivers/scsi/scsi_lib.c +++ b/drivers/scsi/scsi_lib.c @@ -50,6 +50,50 @@ * This entire source file deals with the new queueing code. */ +/* + * Function: __scsi_insert_special() + * + * Purpose: worker for scsi_insert_special_*() + * + * Arguments: q - request queue where request should be inserted + * rq - request to be inserted + * data - private data + * at_head - insert request at head or tail of queue + * + * Lock status: Assumed that io_request_lock is not held upon entry. + * + * Returns: Nothing + */ +static void __scsi_insert_special(request_queue_t *q, struct request *rq, + void *data, int at_head) +{ + unsigned long flags; + + ASSERT_LOCK(&io_request_lock, 0); + + rq->cmd = SPECIAL; + rq->special = data; + rq->q = NULL; + rq->nr_segments = 0; + rq->elevator_sequence = 0; + + /* + * We have the option of inserting the head or the tail of the queue. + * Typically we use the tail for new ioctls and so forth. We use the + * head of the queue for things like a QUEUE_FULL message from a + * device, or a host that is unable to accept a particular command. + */ + spin_lock_irqsave(&io_request_lock, flags); + + if (at_head) + list_add(&rq->queue, &q->queue_head); + else + list_add_tail(&rq->queue, &q->queue_head); + + q->request_fn(q); + spin_unlock_irqrestore(&io_request_lock, flags); +} + /* * Function: scsi_insert_special_cmd() @@ -73,52 +117,9 @@ */ int scsi_insert_special_cmd(Scsi_Cmnd * SCpnt, int at_head) { - unsigned long flags; - request_queue_t *q; - - ASSERT_LOCK(&io_request_lock, 0); - - /* - * The SCpnt already contains a request structure - we will doctor the - * thing up with the appropriate values and use that in the actual - * request queue. - */ - q = &SCpnt->device->request_queue; - SCpnt->request.cmd = SPECIAL; - SCpnt->request.special = (void *) SCpnt; - SCpnt->request.q = NULL; - SCpnt->request.free_list = NULL; - SCpnt->request.nr_segments = 0; - - /* - * We have the option of inserting the head or the tail of the queue. - * Typically we use the tail for new ioctls and so forth. We use the - * head of the queue for things like a QUEUE_FULL message from a - * device, or a host that is unable to accept a particular command. - */ - spin_lock_irqsave(&io_request_lock, flags); - - if (at_head) { - list_add(&SCpnt->request.queue, &q->queue_head); - } else { - /* - * FIXME(eric) - we always insert at the tail of the - * list. Otherwise ioctl commands would always take - * precedence over normal I/O. An ioctl on a busy - * disk might be delayed indefinitely because the - * request might not float high enough in the queue - * to be scheduled. - */ - list_add_tail(&SCpnt->request.queue, &q->queue_head); - } + request_queue_t *q = &SCpnt->device->request_queue; - /* - * Now hit the requeue function for the queue. If the host is - * already busy, so be it - we have nothing special to do. If - * the host can queue it, then send it off. - */ - q->request_fn(q); - spin_unlock_irqrestore(&io_request_lock, flags); + __scsi_insert_special(q, &SCpnt->request, SCpnt, at_head); return 0; } @@ -144,51 +145,9 @@ int scsi_insert_special_cmd(Scsi_Cmnd * SCpnt, int at_head) */ int scsi_insert_special_req(Scsi_Request * SRpnt, int at_head) { - unsigned long flags; - request_queue_t *q; - - ASSERT_LOCK(&io_request_lock, 0); - - /* - * The SCpnt already contains a request structure - we will doctor the - * thing up with the appropriate values and use that in the actual - * request queue. - */ - q = &SRpnt->sr_device->request_queue; - SRpnt->sr_request.cmd = SPECIAL; - SRpnt->sr_request.special = (void *) SRpnt; - SRpnt->sr_request.q = NULL; - SRpnt->sr_request.nr_segments = 0; - - /* - * We have the option of inserting the head or the tail of the queue. - * Typically we use the tail for new ioctls and so forth. We use the - * head of the queue for things like a QUEUE_FULL message from a - * device, or a host that is unable to accept a particular command. - */ - spin_lock_irqsave(&io_request_lock, flags); + request_queue_t *q = &SRpnt->sr_device->request_queue; - if (at_head) { - list_add(&SRpnt->sr_request.queue, &q->queue_head); - } else { - /* - * FIXME(eric) - we always insert at the tail of the - * list. Otherwise ioctl commands would always take - * precedence over normal I/O. An ioctl on a busy - * disk might be delayed indefinitely because the - * request might not float high enough in the queue - * to be scheduled. - */ - list_add_tail(&SRpnt->sr_request.queue, &q->queue_head); - } - - /* - * Now hit the requeue function for the queue. If the host is - * already busy, so be it - we have nothing special to do. If - * the host can queue it, then send it off. - */ - q->request_fn(q); - spin_unlock_irqrestore(&io_request_lock, flags); + __scsi_insert_special(q, &SRpnt->sr_request, SRpnt, at_head); return 0; } @@ -403,6 +362,7 @@ static Scsi_Cmnd *__scsi_end_request(Scsi_Cmnd * SCpnt, struct request *req; struct buffer_head *bh; Scsi_Device * SDpnt; + int nsect; ASSERT_LOCK(&io_request_lock, 0); @@ -414,11 +374,13 @@ static Scsi_Cmnd *__scsi_end_request(Scsi_Cmnd * SCpnt, } do { if ((bh = req->bh) != NULL) { + nsect = bh->b_size >> 9; + blk_finished_io(nsect); req->bh = bh->b_reqnext; - req->nr_sectors -= bh->b_size >> 9; - req->sector += bh->b_size >> 9; + req->nr_sectors -= nsect; + req->sector += nsect; bh->b_reqnext = NULL; - sectors -= bh->b_size >> 9; + sectors -= nsect; bh->b_end_io(bh, uptodate); if ((bh = req->bh) != NULL) { req->current_nr_sectors = bh->b_size >> 9; @@ -863,17 +825,6 @@ void scsi_request_fn(request_queue_t * q) SHpnt = SDpnt->host; /* - * If the host for this device is in error recovery mode, don't - * do anything at all here. When the host leaves error recovery - * mode, it will automatically restart things and start queueing - * commands again. Same goes if the queue is actually plugged, - * if the device itself is blocked, or if the host is fully - * occupied. - */ - if (SHpnt->in_recovery || q->plugged) - return; - - /* * To start with, we keep looping until the queue is empty, or until * the host is no longer able to accept any more requests. */ @@ -896,10 +847,11 @@ void scsi_request_fn(request_queue_t * q) || (SHpnt->host_blocked) || (SHpnt->host_self_blocked)) { /* - * If we are unable to process any commands at all for this - * device, then we consider it to be starved. What this means - * is that there are no outstanding commands for this device - * and hence we need a little help getting it started again + * If we are unable to process any commands at all for + * this device, then we consider it to be starved. + * What this means is that there are no outstanding + * commands for this device and hence we need a + * little help getting it started again * once the host isn't quite so busy. */ if (SDpnt->device_busy == 0) { @@ -1000,8 +952,8 @@ void scsi_request_fn(request_queue_t * q) } /* * If so, we are ready to do something. Bump the count - * while the queue is locked and then break out of the loop. - * Otherwise loop around and try another request. + * while the queue is locked and then break out of the + * loop. Otherwise loop around and try another request. */ if (!SCpnt) { break; @@ -1029,8 +981,9 @@ void scsi_request_fn(request_queue_t * q) memcpy(&SCpnt->request, req, sizeof(struct request)); /* - * We have copied the data out of the request block - it is now in - * a field in SCpnt. Release the request block. + * We have copied the data out of the request block - + * it is now in a field in SCpnt. Release the request + * block. */ blkdev_release_request(req); } @@ -1047,12 +1000,14 @@ void scsi_request_fn(request_queue_t * q) /* * This will do a couple of things: * 1) Fill in the actual SCSI command. - * 2) Fill in any other upper-level specific fields (timeout). + * 2) Fill in any other upper-level specific fields + * (timeout). * - * If this returns 0, it means that the request failed (reading - * past end of disk, reading offline device, etc). This won't - * actually talk to the device, but some kinds of consistency - * checking may cause the request to be rejected immediately. + * If this returns 0, it means that the request failed + * (reading past end of disk, reading offline device, + * etc). This won't actually talk to the device, but + * some kinds of consistency checking may cause the + * request to be rejected immediately. */ if (STpnt == NULL) { STpnt = scsi_get_request_dev(req); @@ -1103,8 +1058,8 @@ void scsi_request_fn(request_queue_t * q) scsi_dispatch_cmd(SCpnt); /* - * Now we need to grab the lock again. We are about to mess with - * the request queue and try to find another command. + * Now we need to grab the lock again. We are about to mess + * with the request queue and try to find another command. */ spin_lock_irq(&io_request_lock); } diff --git a/drivers/scsi/scsi_merge.c b/drivers/scsi/scsi_merge.c index 7483eb970..4ebadd868 100644 --- a/drivers/scsi/scsi_merge.c +++ b/drivers/scsi/scsi_merge.c @@ -324,7 +324,6 @@ static inline int scsi_new_mergeable(request_queue_t * q, req->nr_segments >= SHpnt->sg_tablesize) return 0; req->nr_segments++; - q->elevator.nr_segments++; return 1; } @@ -341,11 +340,8 @@ static inline int scsi_new_segment(request_queue_t * q, if (req->nr_hw_segments >= SHpnt->sg_tablesize || req->nr_segments >= SHpnt->sg_tablesize) return 0; - if (req->nr_segments >= max_segments) - return 0; req->nr_hw_segments++; req->nr_segments++; - q->elevator.nr_segments++; return 1; } #else @@ -361,7 +357,6 @@ static inline int scsi_new_segment(request_queue_t * q, * counter. */ req->nr_segments++; - q->elevator.nr_segments++; return 1; } else { return 0; @@ -417,8 +412,10 @@ __inline static int __scsi_back_merge_fn(request_queue_t * q, SDpnt = (Scsi_Device *) q->queuedata; SHpnt = SDpnt->host; +#ifdef DMA_CHUNK_SIZE if (max_segments > 64) max_segments = 64; +#endif if (use_clustering) { /* @@ -471,8 +468,10 @@ __inline static int __scsi_front_merge_fn(request_queue_t * q, SDpnt = (Scsi_Device *) q->queuedata; SHpnt = SDpnt->host; +#ifdef DMA_CHUNK_SIZE if (max_segments > 64) max_segments = 64; +#endif if (use_clustering) { /* @@ -601,10 +600,10 @@ __inline static int __scsi_merge_requests_fn(request_queue_t * q, SDpnt = (Scsi_Device *) q->queuedata; SHpnt = SDpnt->host; +#ifdef DMA_CHUNK_SIZE if (max_segments > 64) max_segments = 64; -#ifdef DMA_CHUNK_SIZE /* If it would not fit into prepared memory space for sg chain, * then don't allow the merge. */ @@ -664,7 +663,6 @@ __inline static int __scsi_merge_requests_fn(request_queue_t * q, * This one is OK. Let it go. */ req->nr_segments += next->nr_segments - 1; - q->elevator.nr_segments--; #ifdef DMA_CHUNK_SIZE req->nr_hw_segments += next->nr_hw_segments - 1; #endif diff --git a/drivers/scsi/sg.c b/drivers/scsi/sg.c index 04e4a6a28..c064dfde4 100644 --- a/drivers/scsi/sg.c +++ b/drivers/scsi/sg.c @@ -694,6 +694,7 @@ static int sg_common_write(Sg_fd * sfp, Sg_request * srp, (void *)SRpnt->sr_buffer, hp->dxfer_len, sg_cmd_done_bh, timeout, SG_DEFAULT_RETRIES); /* dxfer_len overwrites SRpnt->sr_bufflen, hence need for b_malloc_len */ + generic_unplug_device(&SRpnt->sr_device->request_queue); return 0; } diff --git a/drivers/scsi/sr.c b/drivers/scsi/sr.c index 71ec104f4..7a2cde23b 100644 --- a/drivers/scsi/sr.c +++ b/drivers/scsi/sr.c @@ -671,12 +671,14 @@ void get_capabilities(int i) cmd[3] = cmd[5] = 0; rc = sr_do_ioctl(i, cmd, buffer, 128, 1, SCSI_DATA_READ, NULL); - if (-EINVAL == rc) { - /* failed, drive has'nt this mode page */ + if (rc) { + /* failed, drive doesn't have capabilities mode page */ scsi_CDs[i].cdi.speed = 1; - /* disable speed select, drive probably can't do this either */ - scsi_CDs[i].cdi.mask |= CDC_SELECT_SPEED; + scsi_CDs[i].cdi.mask |= (CDC_CD_R | CDC_CD_RW | CDC_DVD_R | + CDC_DVD | CDC_DVD_RAM | + CDC_SELECT_DISC | CDC_SELECT_SPEED); scsi_free(buffer, 512); + printk("sr%i: scsi-1 drive\n", i); return; } n = buffer[3] + 4; diff --git a/drivers/sound/Config.in b/drivers/sound/Config.in index 9fd35efcb..10235dd2e 100644 --- a/drivers/sound/Config.in +++ b/drivers/sound/Config.in @@ -142,9 +142,9 @@ if [ "$CONFIG_SOUND_OSS" = "y" -o "$CONFIG_SOUND_OSS" = "m" ]; then dep_tristate ' Yamaha FM synthesizer (YM3812/OPL-3) support' CONFIG_SOUND_YM3812 $CONFIG_SOUND_OSS dep_tristate ' Yamaha OPL3-SA1 audio controller' CONFIG_SOUND_OPL3SA1 $CONFIG_SOUND_OSS dep_tristate ' Yamaha OPL3-SA2, SA3, and SAx based PnP cards' CONFIG_SOUND_OPL3SA2 $CONFIG_SOUND_OSS - dep_tristate ' Yamaha YMF7xx PCI audio (legacy mode)' CONFIG_SOUND_YMPCI $CONFIG_SOUND_OSS $CONFIG_PCI - if [ "$CONFIG_SOUND_YMPCI" = "n" ]; then - dep_tristate ' Yamaha YMF7xx PCI audio (native mode) (EXPERIMENTAL)' CONFIG_SOUND_YMFPCI $CONFIG_SOUND_OSS $CONFIG_PCI $CONFIG_EXPERIMENTAL + dep_tristate ' Yamaha YMF7xx PCI audio (native mode)' CONFIG_SOUND_YMFPCI $CONFIG_SOUND_OSS $CONFIG_PCI + if [ "$CONFIG_SOUND_YMFPCI" != "n" ]; then + bool ' Yamaha PCI legacy ports support' CONFIG_SOUND_YMFPCI_LEGACY fi dep_tristate ' 6850 UART support' CONFIG_SOUND_UART6850 $CONFIG_SOUND_OSS diff --git a/drivers/sound/Makefile b/drivers/sound/Makefile index e7ee317e6..824f08ec8 100644 --- a/drivers/sound/Makefile +++ b/drivers/sound/Makefile @@ -47,8 +47,10 @@ obj-$(CONFIG_SOUND_ACI_MIXER) += aci.o obj-$(CONFIG_SOUND_AWE32_SYNTH) += awe_wave.o obj-$(CONFIG_SOUND_VIA82CXXX) += via82cxxx_audio.o ac97_codec.o -obj-$(CONFIG_SOUND_YMPCI) += ymf_sb.o sb_lib.o uart401.o obj-$(CONFIG_SOUND_YMFPCI) += ymfpci.o ac97_codec.o +ifeq ($(CONFIG_SOUND_YMFPCI_LEGACY),y) + obj-$(CONFIG_SOUND_YMFPCI) += opl3.o uart401.o +endif obj-$(CONFIG_SOUND_MSNDCLAS) += msnd.o msnd_classic.o obj-$(CONFIG_SOUND_MSNDPIN) += msnd.o msnd_pinnacle.o obj-$(CONFIG_SOUND_VWSND) += vwsnd.o diff --git a/drivers/sound/emu10k1/audio.c b/drivers/sound/emu10k1/audio.c index 9623dcb2e..2280ca236 100644 --- a/drivers/sound/emu10k1/audio.c +++ b/drivers/sound/emu10k1/audio.c @@ -375,8 +375,10 @@ static int emu10k1_audio_ioctl(struct inode *inode, struct file *file, unsigned format = wiinst->format; format.samplingrate = val; - if (emu10k1_wavein_setformat(wave_dev, &format) < 0) + if (emu10k1_wavein_setformat(wave_dev, &format) < 0) { + spin_unlock_irqrestore(&wiinst->lock, flags); return -EINVAL; + } val = wiinst->format.samplingrate; @@ -393,8 +395,10 @@ static int emu10k1_audio_ioctl(struct inode *inode, struct file *file, unsigned format = woinst->format; format.samplingrate = val; - if (emu10k1_waveout_setformat(wave_dev, &format) < 0) + if (emu10k1_waveout_setformat(wave_dev, &format) < 0) { + spin_unlock_irqrestore(&woinst->lock, flags); return -EINVAL; + } val = woinst->format.samplingrate; @@ -430,8 +434,10 @@ static int emu10k1_audio_ioctl(struct inode *inode, struct file *file, unsigned format = wiinst->format; format.channels = val ? 2 : 1; - if (emu10k1_wavein_setformat(wave_dev, &format) < 0) + if (emu10k1_wavein_setformat(wave_dev, &format) < 0) { + spin_unlock_irqrestore(&wiinst->lock, flags); return -EINVAL; + } val = wiinst->format.channels - 1; @@ -447,8 +453,10 @@ static int emu10k1_audio_ioctl(struct inode *inode, struct file *file, unsigned format = woinst->format; format.channels = val ? 2 : 1; - if (emu10k1_waveout_setformat(wave_dev, &format) < 0) + if (emu10k1_waveout_setformat(wave_dev, &format) < 0) { + spin_unlock_irqrestore(&woinst->lock, flags); return -EINVAL; + } val = woinst->format.channels - 1; @@ -478,8 +486,10 @@ static int emu10k1_audio_ioctl(struct inode *inode, struct file *file, unsigned format = wiinst->format; format.channels = val; - if (emu10k1_wavein_setformat(wave_dev, &format) < 0) + if (emu10k1_wavein_setformat(wave_dev, &format) < 0) { + spin_unlock_irqrestore(&wiinst->lock, flags); return -EINVAL; + } val = wiinst->format.channels; @@ -495,8 +505,10 @@ static int emu10k1_audio_ioctl(struct inode *inode, struct file *file, unsigned format = woinst->format; format.channels = val; - if (emu10k1_waveout_setformat(wave_dev, &format) < 0) + if (emu10k1_waveout_setformat(wave_dev, &format) < 0) { + spin_unlock_irqrestore(&woinst->lock, flags); return -EINVAL; + } val = woinst->format.channels; @@ -542,8 +554,10 @@ static int emu10k1_audio_ioctl(struct inode *inode, struct file *file, unsigned format = wiinst->format; format.bitsperchannel = val; - if (emu10k1_wavein_setformat(wave_dev, &format) < 0) + if (emu10k1_wavein_setformat(wave_dev, &format) < 0) { + spin_unlock_irqrestore(&wiinst->lock, flags); return -EINVAL; + } val = wiinst->format.bitsperchannel; @@ -559,8 +573,10 @@ static int emu10k1_audio_ioctl(struct inode *inode, struct file *file, unsigned format = woinst->format; format.bitsperchannel = val; - if (emu10k1_waveout_setformat(wave_dev, &format) < 0) + if (emu10k1_waveout_setformat(wave_dev, &format) < 0) { + spin_unlock_irqrestore(&woinst->lock, flags); return -EINVAL; + } val = woinst->format.bitsperchannel; @@ -968,6 +984,7 @@ static int emu10k1_audio_mmap(struct file *file, struct vm_area_struct *vma) for (i = 0; i < woinst->buffer.pages; i++) { if (remap_page_range(vma->vm_start + (i * PAGE_SIZE), virt_to_phys(woinst->buffer.addr[i]), PAGE_SIZE, vma->vm_page_prot)) { spin_unlock_irqrestore(&woinst->lock, flags); + unlock_kernel(); return -EAGAIN; } } diff --git a/drivers/sound/trix.c b/drivers/sound/trix.c index 73c9dcf54..03b4b1999 100644 --- a/drivers/sound/trix.c +++ b/drivers/sound/trix.c @@ -17,7 +17,6 @@ * Arnaldo C. de Melo Got rid of attach_uart401 */ -#include <linux/config.h> #include <linux/init.h> #include <linux/module.h> diff --git a/drivers/sound/via82cxxx_audio.c b/drivers/sound/via82cxxx_audio.c index f609d1e2b..fbe02bffd 100644 --- a/drivers/sound/via82cxxx_audio.c +++ b/drivers/sound/via82cxxx_audio.c @@ -15,7 +15,7 @@ */ -#define VIA_VERSION "1.1.14" +#define VIA_VERSION "1.1.14a" #include <linux/config.h> @@ -76,8 +76,16 @@ #define VIA_COUNTER_LIMIT 100000 /* size of DMA buffers */ -#define VIA_DMA_BUFFERS 16 -#define VIA_DMA_BUF_SIZE PAGE_SIZE +#define VIA_MAX_BUFFER_DMA_PAGES 32 + +/* buffering default values in ms */ +#define VIA_DEFAULT_FRAG_TIME 20 +#define VIA_DEFAULT_BUFFER_TIME 500 + +#define VIA_MAX_FRAG_SIZE PAGE_SIZE +#define VIA_MIN_FRAG_SIZE 64 + +#define VIA_MIN_FRAG_NUMBER 2 #ifndef AC97_PCM_LR_ADC_RATE # define AC97_PCM_LR_ADC_RATE AC97_PCM_LR_DAC_RATE @@ -102,7 +110,6 @@ #define VIA_BASE0_PCM_OUT_CHAN_STATUS 0x00 #define VIA_BASE0_PCM_OUT_CHAN_CTRL 0x01 #define VIA_BASE0_PCM_OUT_CHAN_TYPE 0x02 -#define VIA_BASE0_PCM_OUT_BLOCK_COUNT 0x0C #define VIA_BASE0_PCM_IN_CHAN 0x10 /* input PCM from user */ #define VIA_BASE0_PCM_IN_CHAN_STATUS 0x10 @@ -114,6 +121,7 @@ #define VIA_PCM_CONTROL 0x01 #define VIA_PCM_TYPE 0x02 #define VIA_PCM_TABLE_ADDR 0x04 +#define VIA_PCM_BLOCK_COUNT 0x0C /* XXX unused DMA channel for FM PCM data */ #define VIA_BASE0_FM_OUT_CHAN 0x20 @@ -223,14 +231,14 @@ enum via_channel_states { }; -struct via_sgd_data { +struct via_buffer_pgtbl { dma_addr_t handle; void *cpuaddr; }; struct via_channel { - atomic_t n_bufs; + atomic_t n_frags; atomic_t hw_ptr; wait_queue_head_t wait; @@ -246,11 +254,14 @@ struct via_channel { u8 pcm_fmt; /* VIA_PCM_FMT_xxx */ unsigned rate; /* sample rate */ + unsigned int frag_size; + unsigned int frag_number; volatile struct via_sgd_table *sgtable; dma_addr_t sgt_handle; - struct via_sgd_data sgbuf [VIA_DMA_BUFFERS]; + unsigned int page_number; + struct via_buffer_pgtbl pgtbl[VIA_MAX_BUFFER_DMA_PAGES]; long iobase; @@ -301,17 +312,16 @@ static unsigned int via_dsp_poll(struct file *file, struct poll_table_struct *wa static int via_dsp_ioctl (struct inode *inode, struct file *file, unsigned int cmd, unsigned long arg); static int via_dsp_open (struct inode *inode, struct file *file); static int via_dsp_release(struct inode *inode, struct file *file); -#ifdef VIA_SUPPORT_MMAP static int via_dsp_mmap(struct file *file, struct vm_area_struct *vma); -#endif static u16 via_ac97_read_reg (struct ac97_codec *codec, u8 reg); static void via_ac97_write_reg (struct ac97_codec *codec, u8 reg, u16 value); static u8 via_ac97_wait_idle (struct via_info *card); static void via_chan_free (struct via_info *card, struct via_channel *chan); -static void via_chan_clear (struct via_channel *chan); +static void via_chan_clear (struct via_info *card, struct via_channel *chan); static void via_chan_pcm_fmt (struct via_channel *chan, int reset); +static void via_chan_buffer_free (struct via_info *card, struct via_channel *chan); #ifdef VIA_PROC_FS static int via_init_proc (void); @@ -569,25 +579,53 @@ static void via_chan_init_defaults (struct via_info *card, struct via_channel *c chan->pcm_fmt = VIA_PCM_FMT_MASK; chan->is_enabled = 1; - if (chan->is_record) - atomic_set (&chan->n_bufs, 0); - else - atomic_set (&chan->n_bufs, VIA_DMA_BUFFERS); + chan->frag_number = 0; + chan->frag_size = 0; + atomic_set(&chan->n_frags, 0); atomic_set (&chan->hw_ptr, 0); } +/** + * via_chan_init - Initialize PCM channel + * @card: Private audio chip info + * @chan: Channel to be initialized + * + * Performs some of the preparations necessary to begin + * using a PCM channel. + * + * Currently the preparations consist in + * setting the + * PCM channel to a known state. + */ + + +static void via_chan_init (struct via_info *card, struct via_channel *chan) +{ + + DPRINTK ("ENTER\n"); + + /* bzero channel structure, and init members to defaults */ + via_chan_init_defaults (card, chan); + + /* stop any existing channel output */ + via_chan_clear (card, chan); + via_chan_status_clear (chan->iobase); + via_chan_pcm_fmt (chan, 1); + + DPRINTK ("EXIT\n"); +} /** - * via_chan_init - Initialize PCM channel + * via_chan_buffer_init - Initialize PCM channel buffer * @card: Private audio chip info * @chan: Channel to be initialized * - * Performs all the preparations necessary to begin + * Performs some of the preparations necessary to begin * using a PCM channel. * * Currently the preparations include allocating the - * scatter-gather DMA table and buffers, setting the - * PCM channel to a known state, and passing the + * scatter-gather DMA table and buffers, + * and passing the * address of the DMA table to the hardware. * * Note that special care is taken when passing the @@ -596,18 +634,21 @@ static void via_chan_init_defaults (struct via_info *card, struct via_channel *c * always "take" the address. */ -static int via_chan_init (struct via_info *card, struct via_channel *chan) +static int via_chan_buffer_init (struct via_info *card, struct via_channel *chan) { + int page, offset; int i; DPRINTK ("ENTER\n"); - /* bzero channel structure, and init members to defaults */ - via_chan_init_defaults (card, chan); + if (chan->sgtable != NULL) { + DPRINTK ("EXIT\n"); + return 0; + } /* alloc DMA-able memory for scatter-gather table */ chan->sgtable = pci_alloc_consistent (card->pdev, - (sizeof (struct via_sgd_table) * VIA_DMA_BUFFERS), + (sizeof (struct via_sgd_table) * chan->frag_number), &chan->sgt_handle); if (!chan->sgtable) { printk (KERN_ERR PFX "DMA table alloc fail, aborting\n"); @@ -616,45 +657,54 @@ static int via_chan_init (struct via_info *card, struct via_channel *chan) } memset ((void*)chan->sgtable, 0, - (sizeof (struct via_sgd_table) * VIA_DMA_BUFFERS)); + (sizeof (struct via_sgd_table) * chan->frag_number)); /* alloc DMA-able memory for scatter-gather buffers */ - for (i = 0; i < VIA_DMA_BUFFERS; i++) { - chan->sgbuf[i].cpuaddr = - pci_alloc_consistent (card->pdev, VIA_DMA_BUF_SIZE, - &chan->sgbuf[i].handle); - if (!chan->sgbuf[i].cpuaddr) - goto err_out_nomem; + chan->page_number = (chan->frag_number * chan->frag_size) / PAGE_SIZE + + (((chan->frag_number * chan->frag_size) % PAGE_SIZE) ? 1 : 0); + + for (i = 0; i < chan->page_number; i++) { + chan->pgtbl[i].cpuaddr = pci_alloc_consistent (card->pdev, PAGE_SIZE, + &chan->pgtbl[i].handle); - if (i < (VIA_DMA_BUFFERS - 1)) - chan->sgtable[i].count = cpu_to_le32 (VIA_DMA_BUF_SIZE | VIA_FLAG); - else - chan->sgtable[i].count = cpu_to_le32 (VIA_DMA_BUF_SIZE | VIA_EOL); - chan->sgtable[i].addr = cpu_to_le32 (chan->sgbuf[i].handle); + if (!chan->pgtbl[i].cpuaddr) { + chan->page_number = i; + goto err_out_nomem; + } #ifndef VIA_NDEBUG - memset (chan->sgbuf[i].cpuaddr, 0xBC, VIA_DMA_BUF_SIZE); + memset (chan->pgtbl[i].cpuaddr, 0xBC, chan->frag_size); #endif #if 1 - DPRINTK ("dmabuf #%d (h=%lx, 32(h)=%lx, v2p=%lx, a=%p)\n", - i, (long)chan->sgbuf[i].handle, - (long)chan->sgtable[i].addr, - virt_to_phys(chan->sgbuf[i].cpuaddr), - chan->sgbuf[i].cpuaddr); + DPRINTK ("dmabuf_pg #%d (h=%lx, v2p=%lx, a=%p)\n", + i, (long)chan->pgtbl[i].handle, + virt_to_phys(chan->pgtbl[i].cpuaddr), + chan->pgtbl[i].cpuaddr); #endif - - assert ((VIA_DMA_BUF_SIZE % PAGE_SIZE) == 0); } - /* stop any existing channel output */ - via_chan_clear (chan); - via_chan_status_clear (chan->iobase); - via_chan_pcm_fmt (chan, 1); + for (i = 0; i < chan->frag_number; i++) { + + page = i / (PAGE_SIZE / chan->frag_size); + offset = (i % (PAGE_SIZE / chan->frag_size)) * chan->frag_size; + + chan->sgtable[i].count = cpu_to_le32 (chan->frag_size | VIA_FLAG); + chan->sgtable[i].addr = cpu_to_le32 (chan->pgtbl[page].handle + offset); + +#if 1 + DPRINTK ("dmabuf #%d (32(h)=%lx)\n", + i, + (long)chan->sgtable[i].addr); +#endif + } + + /* overwrite the last buffer information */ + chan->sgtable[chan->frag_number - 1].count = cpu_to_le32 (chan->frag_size | VIA_EOL); /* set location of DMA-able scatter-gather info table */ - DPRINTK("outl (0x%X, 0x%04lX)\n", + DPRINTK ("outl (0x%X, 0x%04lX)\n", cpu_to_le32 (chan->sgt_handle), chan->iobase + VIA_PCM_TABLE_ADDR); @@ -664,7 +714,7 @@ static int via_chan_init (struct via_info *card, struct via_channel *chan) udelay (20); via_ac97_wait_idle (card); - DPRINTK("inl (0x%lX) = %x\n", + DPRINTK ("inl (0x%lX) = %x\n", chan->iobase + VIA_PCM_TABLE_ADDR, inl(chan->iobase + VIA_PCM_TABLE_ADDR)); @@ -673,7 +723,7 @@ static int via_chan_init (struct via_info *card, struct via_channel *chan) err_out_nomem: printk (KERN_ERR PFX "DMA buffer alloc fail, aborting\n"); - via_chan_free (card, chan); + via_chan_buffer_free (card, chan); DPRINTK ("EXIT\n"); return -ENOMEM; } @@ -695,8 +745,6 @@ err_out_nomem: static void via_chan_free (struct via_info *card, struct via_channel *chan) { - int i; - DPRINTK ("ENTER\n"); synchronize_irq(); @@ -710,23 +758,33 @@ static void via_chan_free (struct via_info *card, struct via_channel *chan) spin_unlock_irq (&card->lock); + DPRINTK ("EXIT\n"); +} + +static void via_chan_buffer_free (struct via_info *card, struct via_channel *chan) +{ + int i; + + DPRINTK ("ENTER\n"); + /* zero location of DMA-able scatter-gather info table */ via_ac97_wait_idle(card); outl (0, chan->iobase + VIA_PCM_TABLE_ADDR); - for (i = 0; i < VIA_DMA_BUFFERS; i++) - if (chan->sgbuf[i].cpuaddr) { - assert ((VIA_DMA_BUF_SIZE % PAGE_SIZE) == 0); - pci_free_consistent (card->pdev, VIA_DMA_BUF_SIZE, - chan->sgbuf[i].cpuaddr, - chan->sgbuf[i].handle); - chan->sgbuf[i].cpuaddr = NULL; - chan->sgbuf[i].handle = 0; + for (i = 0; i < chan->page_number; i++) + if (chan->pgtbl[i].cpuaddr) { + pci_free_consistent (card->pdev, PAGE_SIZE, + chan->pgtbl[i].cpuaddr, + chan->pgtbl[i].handle); + chan->pgtbl[i].cpuaddr = NULL; + chan->pgtbl[i].handle = 0; } + chan->page_number = 0; + if (chan->sgtable) { pci_free_consistent (card->pdev, - (sizeof (struct via_sgd_table) * VIA_DMA_BUFFERS), + (sizeof (struct via_sgd_table) * chan->frag_number), (void*)chan->sgtable, chan->sgt_handle); chan->sgtable = NULL; } @@ -771,11 +829,11 @@ static void via_chan_pcm_fmt (struct via_channel *chan, int reset) if (!chan->is_record) chan->pcm_fmt |= VIA_CHAN_TYPE_INT_SELECT; - outb (chan->pcm_fmt, chan->iobase + 2); + outb (chan->pcm_fmt, chan->iobase + VIA_PCM_TYPE); DPRINTK ("EXIT, pcm_fmt = 0x%02X, reg = 0x%02X\n", chan->pcm_fmt, - inb (chan->iobase + 2)); + inb (chan->iobase + VIA_PCM_TYPE)); } @@ -787,10 +845,11 @@ static void via_chan_pcm_fmt (struct via_channel *chan, int reset) * all software pointers which track DMA operation. */ -static void via_chan_clear (struct via_channel *chan) +static void via_chan_clear (struct via_info *card, struct via_channel *chan) { DPRINTK ("ENTER\n"); via_chan_stop (chan->iobase); + via_chan_buffer_free(card, chan); chan->is_active = 0; chan->is_mapped = 0; chan->is_enabled = 1; @@ -798,10 +857,6 @@ static void via_chan_clear (struct via_channel *chan) chan->sw_ptr = 0; chan->n_irqs = 0; atomic_set (&chan->hw_ptr, 0); - if (chan->is_record) - atomic_set (&chan->n_bufs, 0); - else - atomic_set (&chan->n_bufs, VIA_DMA_BUFFERS); DPRINTK ("EXIT\n"); } @@ -826,7 +881,7 @@ static int via_chan_set_speed (struct via_info *card, { DPRINTK ("ENTER, requested rate = %d\n", val); - via_chan_clear (chan); + via_chan_clear (card, chan); val = via_set_rate (&card->ac97, chan, val); @@ -858,7 +913,7 @@ static int via_chan_set_fmt (struct via_info *card, val == AFMT_S16_LE ? "AFMT_S16_LE" : "unknown"); - via_chan_clear (chan); + via_chan_clear (card, chan); assert (val != AFMT_QUERY); /* this case is handled elsewhere */ @@ -907,7 +962,7 @@ static int via_chan_set_stereo (struct via_info *card, { DPRINTK ("ENTER, channels = %d\n", val); - via_chan_clear (chan); + via_chan_clear (card, chan); switch (val) { @@ -934,6 +989,78 @@ static int via_chan_set_stereo (struct via_info *card, return val; } +static int via_chan_set_buffering (struct via_info *card, + struct via_channel *chan, int val) +{ + int shift; + + DPRINTK ("ENTER\n"); + + /* in both cases the buffer cannot be changed */ + if (chan->is_active || chan->is_mapped) { + DPRINTK ("EXIT\n"); + return -EINVAL; + } + + /* called outside SETFRAGMENT */ + /* set defaults or do nothing */ + if (val < 0) { + + if (chan->frag_size && chan->frag_number) + goto out; + + DPRINTK ("\n"); + + chan->frag_size = (VIA_DEFAULT_FRAG_TIME * chan->rate * + ((chan->pcm_fmt & VIA_PCM_FMT_STEREO) ? 2 : 1) * + ((chan->pcm_fmt & VIA_PCM_FMT_16BIT) ? 2 : 1)) / 1000 - 1; + + shift = 0; + while (chan->frag_size) { + chan->frag_size >>= 1; + shift++; + } + chan->frag_size = 1 << shift; + + chan->frag_number = (VIA_DEFAULT_BUFFER_TIME / VIA_DEFAULT_FRAG_TIME); + + DPRINTK ("setting default values %d %d\n", chan->frag_size, chan->frag_number); + } else { + chan->frag_size = 1 << (val & 0xFFFF); + chan->frag_number = (val >> 16) & 0xFFFF; + + DPRINTK ("using user values %d %d\n", chan->frag_size, chan->frag_number); + } + + /* quake3 wants frag_number to be a power of two */ + shift = 0; + while (chan->frag_number) { + chan->frag_number >>= 1; + shift++; + } + chan->frag_number = 1 << shift; + + if (chan->frag_size > VIA_MAX_FRAG_SIZE) + chan->frag_size = VIA_MAX_FRAG_SIZE; + else if (chan->frag_size < VIA_MIN_FRAG_SIZE) + chan->frag_size = VIA_MIN_FRAG_SIZE; + + if (chan->frag_number < VIA_MIN_FRAG_NUMBER) + chan->frag_number = VIA_MIN_FRAG_NUMBER; + + if ((chan->frag_number * chan->frag_size) / PAGE_SIZE > VIA_MAX_BUFFER_DMA_PAGES) + chan->frag_number = (VIA_MAX_BUFFER_DMA_PAGES * PAGE_SIZE) / chan->frag_size; + +out: + if (chan->is_record) + atomic_set (&chan->n_frags, 0); + else + atomic_set (&chan->n_frags, chan->frag_number); + + DPRINTK ("EXIT\n"); + + return 0; +} #ifdef VIA_CHAN_DUMP_BUFS /** @@ -948,7 +1075,7 @@ static void via_chan_dump_bufs (struct via_channel *chan) { int i; - for (i = 0; i < VIA_DMA_BUFFERS; i++) { + for (i = 0; i < chan->frag_number; i++) { DPRINTK ("#%02d: addr=%x, count=%u, flag=%d, eol=%d\n", i, chan->sgtable[i].addr, chan->sgtable[i].count & 0x00FFFFFF, @@ -975,15 +1102,15 @@ static void via_chan_flush_frag (struct via_channel *chan) assert (chan->slop_len > 0); - if (chan->sw_ptr == (VIA_DMA_BUFFERS - 1)) + if (chan->sw_ptr == (chan->frag_number - 1)) chan->sw_ptr = 0; else chan->sw_ptr++; chan->slop_len = 0; - assert (atomic_read (&chan->n_bufs) > 0); - atomic_dec (&chan->n_bufs); + assert (atomic_read (&chan->n_frags) > 0); + atomic_dec (&chan->n_frags); DPRINTK ("EXIT\n"); } @@ -1003,7 +1130,7 @@ static inline void via_chan_maybe_start (struct via_channel *chan) if (!chan->is_active && chan->is_enabled) { chan->is_active = 1; sg_begin (chan); - DPRINTK("starting channel %s\n", chan->name); + DPRINTK ("starting channel %s\n", chan->name); } } @@ -1213,7 +1340,7 @@ static loff_t via_llseek(struct file *file, loff_t offset, int origin) { DPRINTK ("ENTER\n"); - DPRINTK("EXIT, returning -ESPIPE\n"); + DPRINTK ("EXIT, returning -ESPIPE\n"); return -ESPIPE; } @@ -1245,7 +1372,7 @@ static int __init via_ac97_reset (struct via_info *card) pci_read_config_byte (card->pdev, 0x43, &r43); pci_read_config_byte (card->pdev, 0x44, &r44); pci_read_config_byte (card->pdev, 0x48, &r48); - DPRINTK("PCI config: %02X %02X %02X %02X %02X %02X\n", + DPRINTK ("PCI config: %02X %02X %02X %02X %02X %02X\n", r40,r41,r42,r43,r44,r48); spin_lock_irq (&card->lock); @@ -1334,7 +1461,7 @@ static int __init via_ac97_init (struct via_info *card) card->ac97.dev_mixer = register_sound_mixer (&via_mixer_fops, -1); if (card->ac97.dev_mixer < 0) { printk (KERN_ERR PFX "unable to register AC97 mixer, aborting\n"); - DPRINTK("EXIT, returning -EIO\n"); + DPRINTK ("EXIT, returning -EIO\n"); return -EIO; } @@ -1359,21 +1486,21 @@ static int __init via_ac97_init (struct via_info *card) err_out: unregister_sound_mixer (card->ac97.dev_mixer); - DPRINTK("EXIT, returning %d\n", rc); + DPRINTK ("EXIT, returning %d\n", rc); return rc; } static void via_ac97_cleanup (struct via_info *card) { - DPRINTK("ENTER\n"); + DPRINTK ("ENTER\n"); assert (card != NULL); assert (card->ac97.dev_mixer >= 0); unregister_sound_mixer (card->ac97.dev_mixer); - DPRINTK("EXIT\n"); + DPRINTK ("EXIT\n"); } @@ -1414,24 +1541,24 @@ static void via_intr_channel (struct via_channel *chan) /* sanity check: make sure our h/w ptr doesn't have a weird value */ assert (n >= 0); - assert (n < VIA_DMA_BUFFERS); + assert (n < chan->frag_number); /* reset SGD data structure in memory to reflect a full buffer, * and advance the h/w ptr, wrapping around to zero if needed */ - if (n == (VIA_DMA_BUFFERS - 1)) { - chan->sgtable[n].count = (VIA_DMA_BUF_SIZE | VIA_EOL); + if (n == (chan->frag_number - 1)) { + chan->sgtable[n].count = (chan->frag_size | VIA_EOL); atomic_set (&chan->hw_ptr, 0); } else { - chan->sgtable[n].count = (VIA_DMA_BUF_SIZE | VIA_FLAG); + chan->sgtable[n].count = (chan->frag_size | VIA_FLAG); atomic_inc (&chan->hw_ptr); } /* accounting crap for SNDCTL_DSP_GETxPTR */ chan->n_irqs++; - chan->bytes += VIA_DMA_BUF_SIZE; + chan->bytes += chan->frag_size; if (chan->bytes < 0) /* handle overflow of 31-bit value */ - chan->bytes = VIA_DMA_BUF_SIZE; + chan->bytes = chan->frag_size; /* wake up anyone listening to see when interrupts occur */ if (waitqueue_active (&chan->wait)) @@ -1445,25 +1572,25 @@ static void via_intr_channel (struct via_channel *chan) if (chan->is_mapped) return; - /* If we are recording, then n_bufs represents the number - * of buffers waiting to be handled by userspace. - * If we are playback, then n_bufs represents the number - * of buffers remaining to be filled by userspace. - * We increment here. If we reach max buffers (VIA_DMA_BUFFERS), + /* If we are recording, then n_frags represents the number + * of fragments waiting to be handled by userspace. + * If we are playback, then n_frags represents the number + * of fragments remaining to be filled by userspace. + * We increment here. If we reach max number of fragments, * this indicates an underrun/overrun. For this case under OSS, * we stop the record/playback process. */ - if (atomic_read (&chan->n_bufs) < VIA_DMA_BUFFERS) - atomic_inc (&chan->n_bufs); - assert (atomic_read (&chan->n_bufs) <= VIA_DMA_BUFFERS); + if (atomic_read (&chan->n_frags) < chan->frag_number) + atomic_inc (&chan->n_frags); + assert (atomic_read (&chan->n_frags) <= chan->frag_number); - if (atomic_read (&chan->n_bufs) == VIA_DMA_BUFFERS) { + if (atomic_read (&chan->n_frags) == chan->frag_number) { chan->is_active = 0; via_chan_stop (chan->iobase); } - DPRINTK ("%s intr, channel n_bufs == %d\n", chan->name, - atomic_read (&chan->n_bufs)); + DPRINTK ("%s intr, channel n_frags == %d\n", chan->name, + atomic_read (&chan->n_frags)); } @@ -1618,9 +1745,7 @@ static struct file_operations via_dsp_fops = { poll: via_dsp_poll, llseek: via_llseek, ioctl: via_dsp_ioctl, -#ifdef VIA_SUPPORT_MMAP mmap: via_dsp_mmap, -#endif }; @@ -1668,7 +1793,6 @@ static void via_dsp_cleanup (struct via_info *card) } -#ifdef VIA_SUPPORT_MMAP static struct page * via_mm_nopage (struct vm_area_struct * vma, unsigned long address, int write_access) { @@ -1685,8 +1809,6 @@ static struct page * via_mm_nopage (struct vm_area_struct * vma, address, write_access); - assert (VIA_DMA_BUF_SIZE == PAGE_SIZE); - if (address > vma->vm_end) { DPRINTK ("EXIT, returning NOPAGE_SIGBUS\n"); return NOPAGE_SIGBUS; /* Disallow mremap */ @@ -1702,7 +1824,7 @@ static struct page * via_mm_nopage (struct vm_area_struct * vma, #ifndef VIA_NDEBUG { - unsigned long max_bufs = VIA_DMA_BUFFERS; + unsigned long max_bufs = chan->frag_number; if (rd && wr) max_bufs *= 2; /* via_dsp_mmap() should ensure this */ assert (pgoff < max_bufs); @@ -1711,17 +1833,17 @@ static struct page * via_mm_nopage (struct vm_area_struct * vma, /* if full-duplex (read+write) and we have two sets of bufs, * then the playback buffers come first, sez soundcard.c */ - if (pgoff >= VIA_DMA_BUFFERS) { - pgoff -= VIA_DMA_BUFFERS; + if (pgoff >= chan->page_number) { + pgoff -= chan->page_number; chan = &card->ch_in; } else if (!wr) chan = &card->ch_in; - assert ((((unsigned long)chan->sgbuf[pgoff].cpuaddr) % PAGE_SIZE) == 0); + assert ((((unsigned long)chan->pgtbl[pgoff].cpuaddr) % PAGE_SIZE) == 0); - dmapage = virt_to_page (chan->sgbuf[pgoff].cpuaddr); + dmapage = virt_to_page (chan->pgtbl[pgoff].cpuaddr); DPRINTK ("EXIT, returning page %p for cpuaddr %lXh\n", - dmapage, (unsigned long) chan->sgbuf[pgoff].cpuaddr); + dmapage, (unsigned long) chan->pgtbl[pgoff].cpuaddr); get_page (dmapage); return dmapage; } @@ -1761,16 +1883,18 @@ static int via_dsp_mmap(struct file *file, struct vm_area_struct *vma) vma->vm_end - vma->vm_start, vma->vm_pgoff); - assert (VIA_DMA_BUF_SIZE == PAGE_SIZE); - max_size = 0; - if (file->f_mode & FMODE_READ) { + if (vma->vm_flags & VM_READ) { rd = 1; - max_size += (VIA_DMA_BUFFERS * VIA_DMA_BUF_SIZE); + via_chan_set_buffering(card, &card->ch_in, -1); + via_chan_buffer_init (card, &card->ch_in); + max_size += card->ch_in.page_number << PAGE_SHIFT; } - if (file->f_mode & FMODE_WRITE) { + if (vma->vm_flags & VM_WRITE) { wr = 1; - max_size += (VIA_DMA_BUFFERS * VIA_DMA_BUF_SIZE); + via_chan_set_buffering(card, &card->ch_out, -1); + via_chan_buffer_init (card, &card->ch_out); + max_size += card->ch_out.page_number << PAGE_SHIFT; } start = vma->vm_start; @@ -1802,10 +1926,9 @@ static int via_dsp_mmap(struct file *file, struct vm_area_struct *vma) rc = 0; out: - DPRINTK("EXIT, returning %d\n", rc); + DPRINTK ("EXIT, returning %d\n", rc); return rc; } -#endif /* VIA_SUPPORT_MMAP */ static ssize_t via_dsp_do_read (struct via_info *card, @@ -1831,13 +1954,13 @@ handle_one_block: */ n = chan->sw_ptr; - /* n_bufs represents the number of buffers waiting + /* n_frags represents the number of fragments waiting * to be copied to userland. sleep until at least * one buffer has been read from the audio hardware. */ - tmp = atomic_read (&chan->n_bufs); + tmp = atomic_read (&chan->n_frags); assert (tmp >= 0); - assert (tmp <= VIA_DMA_BUFFERS); + assert (tmp <= chan->frag_number); while (tmp == 0) { if (nonblock || !chan->is_active) return -EAGAIN; @@ -1848,18 +1971,18 @@ handle_one_block: if (signal_pending (current)) return -ERESTARTSYS; - tmp = atomic_read (&chan->n_bufs); + tmp = atomic_read (&chan->n_frags); } /* Now that we have a buffer we can read from, send * as much as sample data possible to userspace. */ - while ((count > 0) && (chan->slop_len < VIA_DMA_BUF_SIZE)) { - size_t slop_left = VIA_DMA_BUF_SIZE - chan->slop_len; + while ((count > 0) && (chan->slop_len < chan->frag_size)) { + size_t slop_left = chan->frag_size - chan->slop_len; size = (count < slop_left) ? count : slop_left; if (copy_to_user (userbuf, - chan->sgbuf[n].cpuaddr + chan->slop_len, + chan->pgtbl[n / (PAGE_SIZE / chan->frag_size)].cpuaddr + n % (PAGE_SIZE / chan->frag_size) + chan->slop_len, size)) return -EFAULT; @@ -1871,7 +1994,7 @@ handle_one_block: /* If we didn't copy the buffer completely to userspace, * stop now. */ - if (chan->slop_len < VIA_DMA_BUF_SIZE) + if (chan->slop_len < chan->frag_size) goto out; /* @@ -1882,20 +2005,20 @@ handle_one_block: /* advance channel software pointer to point to * the next buffer from which we will copy */ - if (chan->sw_ptr == (VIA_DMA_BUFFERS - 1)) + if (chan->sw_ptr == (chan->frag_number - 1)) chan->sw_ptr = 0; else chan->sw_ptr++; /* mark one less buffer waiting to be processed */ - assert (atomic_read (&chan->n_bufs) > 0); - atomic_dec (&chan->n_bufs); + assert (atomic_read (&chan->n_frags) > 0); + atomic_dec (&chan->n_frags); /* we are at a block boundary, there is no fragment data */ chan->slop_len = 0; - DPRINTK("Flushed block %u, sw_ptr now %u, n_bufs now %d\n", - n, chan->sw_ptr, atomic_read (&chan->n_bufs)); + DPRINTK ("Flushed block %u, sw_ptr now %u, n_frags now %d\n", + n, chan->sw_ptr, atomic_read (&chan->n_frags)); DPRINTK ("regs==%02X %02X %02X %08X %08X %08X %08X\n", inb (card->baseaddr + 0x00), @@ -1941,12 +2064,18 @@ static ssize_t via_dsp_read(struct file *file, char *buffer, size_t count, loff_ goto out_up; } + via_chan_set_buffering(card, &card->ch_in, -1); + rc = via_chan_buffer_init (card, &card->ch_in); + + if (rc) + goto out_up; + rc = via_dsp_do_read (card, buffer, count, nonblock); out_up: up (&card->syscall_sem); out: - DPRINTK("EXIT, returning %ld\n",(long) rc); + DPRINTK ("EXIT, returning %ld\n",(long) rc); return rc; } @@ -1966,40 +2095,40 @@ handle_one_block: if (current->need_resched) schedule (); - /* grab current channel software pointer. In the case of - * playback, this is pointing to the next buffer that + /* grab current channel fragment pointer. In the case of + * playback, this is pointing to the next fragment that * should receive data from userland. */ n = chan->sw_ptr; - /* n_bufs represents the number of buffers remaining + /* n_frags represents the number of fragments remaining * to be filled by userspace. Sleep until - * at least one buffer is available for our use. + * at least one fragment is available for our use. */ - tmp = atomic_read (&chan->n_bufs); + tmp = atomic_read (&chan->n_frags); assert (tmp >= 0); - assert (tmp <= VIA_DMA_BUFFERS); + assert (tmp <= chan->frag_number); while (tmp == 0) { if (nonblock || !chan->is_enabled) return -EAGAIN; - DPRINTK ("Sleeping on block %d, tmp==%d, ir==%d\n", n, tmp, chan->is_record); + DPRINTK ("Sleeping on page %d, tmp==%d, ir==%d\n", n, tmp, chan->is_record); interruptible_sleep_on (&chan->wait); if (signal_pending (current)) return -ERESTARTSYS; - tmp = atomic_read (&chan->n_bufs); + tmp = atomic_read (&chan->n_frags); } - /* Now that we have a buffer we can write to, fill it up + /* Now that we have at least one fragment we can write to, fill the buffer * as much as possible with data from userspace. */ - while ((count > 0) && (chan->slop_len < VIA_DMA_BUF_SIZE)) { - size_t slop_left = VIA_DMA_BUF_SIZE - chan->slop_len; + while ((count > 0) && (chan->slop_len < chan->frag_size)) { + size_t slop_left = chan->frag_size - chan->slop_len; size = (count < slop_left) ? count : slop_left; - if (copy_from_user (chan->sgbuf[n].cpuaddr + chan->slop_len, + if (copy_from_user (chan->pgtbl[n / (PAGE_SIZE / chan->frag_size)].cpuaddr + (n % (PAGE_SIZE / chan->frag_size)) * chan->frag_size + chan->slop_len, userbuf, size)) return -EFAULT; @@ -2009,36 +2138,36 @@ handle_one_block: } /* If we didn't fill up the buffer with data, stop now. - * Put a 'stop' marker in the DMA table too, to tell the - * audio hardware to stop if it gets here. - */ - if (chan->slop_len < VIA_DMA_BUF_SIZE) { + * Put a 'stop' marker in the DMA table too, to tell the + * audio hardware to stop if it gets here. + */ + if (chan->slop_len < chan->frag_size) { sgtable[n].count = cpu_to_le32 (chan->slop_len | VIA_EOL | VIA_STOP); goto out; } /* - * If we get to this point, we have filled a buffer with - * audio data, flush the buffer to audio hardware. - */ + * If we get to this point, we have filled a buffer with + * audio data, flush the buffer to audio hardware. + */ /* Record the true size for the audio hardware to notice */ - if (n == (VIA_DMA_BUFFERS - 1)) - sgtable[n].count = cpu_to_le32 (VIA_DMA_BUF_SIZE | VIA_EOL); - else - sgtable[n].count = cpu_to_le32 (VIA_DMA_BUF_SIZE | VIA_FLAG); + if (n == (chan->frag_number - 1)) + sgtable[n].count = cpu_to_le32 (chan->frag_size | VIA_EOL); + else + sgtable[n].count = cpu_to_le32 (chan->frag_size | VIA_FLAG); /* advance channel software pointer to point to * the next buffer we will fill with data */ - if (chan->sw_ptr == (VIA_DMA_BUFFERS - 1)) + if (chan->sw_ptr == (chan->frag_number - 1)) chan->sw_ptr = 0; else chan->sw_ptr++; /* mark one less buffer as being available for userspace consumption */ - assert (atomic_read (&chan->n_bufs) > 0); - atomic_dec (&chan->n_bufs); + assert (atomic_read (&chan->n_frags) > 0); + atomic_dec (&chan->n_frags); /* we are at a block boundary, there is no fragment data */ chan->slop_len = 0; @@ -2046,8 +2175,8 @@ handle_one_block: /* if SGD has not yet been started, start it */ via_chan_maybe_start (chan); - DPRINTK("Flushed block %u, sw_ptr now %u, n_bufs now %d\n", - n, chan->sw_ptr, atomic_read (&chan->n_bufs)); + DPRINTK ("Flushed block %u, sw_ptr now %u, n_frags now %d\n", + n, chan->sw_ptr, atomic_read (&chan->n_frags)); DPRINTK ("regs==%02X %02X %02X %08X %08X %08X %08X\n", inb (card->baseaddr + 0x00), @@ -2093,12 +2222,18 @@ static ssize_t via_dsp_write(struct file *file, const char *buffer, size_t count goto out_up; } + via_chan_set_buffering(card, &card->ch_out, -1); + rc = via_chan_buffer_init (card, &card->ch_out); + + if (rc) + goto out_up; + rc = via_dsp_do_write (card, buffer, count, nonblock); out_up: up (&card->syscall_sem); out: - DPRINTK("EXIT, returning %ld\n",(long) rc); + DPRINTK ("EXIT, returning %ld\n",(long) rc); return rc; } @@ -2117,23 +2252,27 @@ static unsigned int via_dsp_poll(struct file *file, struct poll_table_struct *wa rd = (file->f_mode & FMODE_READ); wr = (file->f_mode & FMODE_WRITE); - if (wr && (atomic_read (&card->ch_out.n_bufs) == 0)) { + if (wr && (atomic_read (&card->ch_out.n_frags) == 0)) { assert (card->ch_out.is_active); poll_wait(file, &card->ch_out.wait, wait); } if (rd) { /* XXX is it ok, spec-wise, to start DMA here? */ + if (!card->ch_in.is_active) { + via_chan_set_buffering(card, &card->ch_in, -1); + via_chan_buffer_init(card, &card->ch_in); + } via_chan_maybe_start (&card->ch_in); - if (atomic_read (&card->ch_in.n_bufs) == 0) + if (atomic_read (&card->ch_in.n_frags) == 0) poll_wait(file, &card->ch_in.wait, wait); } - if (wr && (atomic_read (&card->ch_out.n_bufs) > 0)) + if (wr && ((atomic_read (&card->ch_out.n_frags) > 0) || !card->ch_out.is_active)) mask |= POLLOUT | POLLWRNORM; - if (rd && (atomic_read (&card->ch_in.n_bufs) > 0)) + if (rd && (atomic_read (&card->ch_in.n_frags) > 0)) mask |= POLLIN | POLLRDNORM; - DPRINTK("EXIT, returning %u\n", mask); + DPRINTK ("EXIT, returning %u\n", mask); return mask; } @@ -2158,12 +2297,12 @@ static int via_dsp_drain_playback (struct via_info *card, if (chan->slop_len > 0) via_chan_flush_frag (chan); - if (atomic_read (&chan->n_bufs) == VIA_DMA_BUFFERS) + if (atomic_read (&chan->n_frags) == chan->frag_number) goto out; via_chan_maybe_start (chan); - while (atomic_read (&chan->n_bufs) < VIA_DMA_BUFFERS) { + while (atomic_read (&chan->n_frags) < chan->frag_number) { if (nonblock) { DPRINTK ("EXIT, returning -EAGAIN\n"); return -EAGAIN; @@ -2178,7 +2317,7 @@ static int via_dsp_drain_playback (struct via_info *card, pci_read_config_byte (card->pdev, 0x43, &r43); pci_read_config_byte (card->pdev, 0x44, &r44); pci_read_config_byte (card->pdev, 0x48, &r48); - DPRINTK("PCI config: %02X %02X %02X %02X %02X %02X\n", + DPRINTK ("PCI config: %02X %02X %02X %02X %02X %02X\n", r40,r41,r42,r43,r44,r48); DPRINTK ("regs==%02X %02X %02X %08X %08X %08X %08X\n", @@ -2195,7 +2334,7 @@ static int via_dsp_drain_playback (struct via_info *card, printk (KERN_ERR "sleeping but not active\n"); #endif - DPRINTK ("sleeping, nbufs=%d\n", atomic_read (&chan->n_bufs)); + DPRINTK ("sleeping, nbufs=%d\n", atomic_read (&chan->n_frags)); interruptible_sleep_on (&chan->wait); if (signal_pending (current)) { @@ -2213,7 +2352,7 @@ static int via_dsp_drain_playback (struct via_info *card, pci_read_config_byte (card->pdev, 0x43, &r43); pci_read_config_byte (card->pdev, 0x44, &r44); pci_read_config_byte (card->pdev, 0x48, &r48); - DPRINTK("PCI config: %02X %02X %02X %02X %02X %02X\n", + DPRINTK ("PCI config: %02X %02X %02X %02X %02X %02X\n", r40,r41,r42,r43,r44,r48); DPRINTK ("regs==%02X %02X %02X %08X %08X %08X %08X\n", @@ -2225,7 +2364,7 @@ static int via_dsp_drain_playback (struct via_info *card, inl (card->baseaddr + 0x80), inl (card->baseaddr + 0x84)); - DPRINTK ("final nbufs=%d\n", atomic_read (&chan->n_bufs)); + DPRINTK ("final nbufs=%d\n", atomic_read (&chan->n_frags)); } #endif @@ -2252,21 +2391,23 @@ static int via_dsp_ioctl_space (struct via_info *card, { audio_buf_info info; - info.fragstotal = VIA_DMA_BUFFERS; - info.fragsize = VIA_DMA_BUF_SIZE; + via_chan_set_buffering(card, chan, -1); + + info.fragstotal = chan->frag_number; + info.fragsize = chan->frag_size; /* number of full fragments we can read/write without blocking */ - info.fragments = atomic_read (&chan->n_bufs); + info.fragments = atomic_read (&chan->n_frags); - if ((chan->slop_len > 0) && (info.fragments > 0)) + if ((chan->slop_len % chan->frag_size > 0) && (info.fragments > 0)) info.fragments--; /* number of bytes that can be read or written immediately * without blocking. */ - info.bytes = (info.fragments * VIA_DMA_BUF_SIZE); - if (chan->slop_len > 0) - info.bytes += VIA_DMA_BUF_SIZE - chan->slop_len; + info.bytes = (info.fragments * chan->frag_size); + if (chan->slop_len % chan->frag_size > 0) + info.bytes += chan->frag_size - (chan->slop_len % chan->frag_size); DPRINTK ("EXIT, returning fragstotal=%d, fragsize=%d, fragments=%d, bytes=%d\n", info.fragstotal, @@ -2305,8 +2446,8 @@ static int via_dsp_ioctl_ptr (struct via_info *card, if (chan->is_active) { unsigned long extra; - info.ptr = atomic_read (&chan->hw_ptr) * VIA_DMA_BUF_SIZE; - extra = VIA_DMA_BUF_SIZE - inl (chan->iobase + VIA_BASE0_PCM_OUT_BLOCK_COUNT); + info.ptr = atomic_read (&chan->hw_ptr) * chan->frag_size; + extra = chan->frag_size - inl (chan->iobase + VIA_PCM_BLOCK_COUNT); info.ptr += extra; info.bytes += extra; } else { @@ -2386,13 +2527,13 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, /* OSS API version. XXX unverified */ case OSS_GETVERSION: - DPRINTK("ioctl OSS_GETVERSION, EXIT, returning SOUND_VERSION\n"); + DPRINTK ("ioctl OSS_GETVERSION, EXIT, returning SOUND_VERSION\n"); rc = put_user (SOUND_VERSION, (int *)arg); break; /* list of supported PCM data formats */ case SNDCTL_DSP_GETFMTS: - DPRINTK("DSP_GETFMTS, EXIT, returning AFMT U8|S16_LE\n"); + DPRINTK ("DSP_GETFMTS, EXIT, returning AFMT U8|S16_LE\n"); rc = put_user (AFMT_U8 | AFMT_S16_LE, (int *)arg); break; @@ -2402,20 +2543,19 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, rc = -EFAULT; break; } - DPRINTK("DSP_SETFMT, val==%d\n", val); + DPRINTK ("DSP_SETFMT, val==%d\n", val); if (val != AFMT_QUERY) { rc = 0; - if (rc == 0 && rd) + if (rd) rc = via_chan_set_fmt (card, &card->ch_in, val); - if (rc == 0 && wr) + + if (rc >= 0 && wr) rc = via_chan_set_fmt (card, &card->ch_out, val); - if (rc <= 0) { - if (rc == 0) - rc = -EINVAL; + if (rc < 0) break; - } + val = rc; } else { if ((rd && (card->ch_in.pcm_fmt & VIA_PCM_FMT_16BIT)) || @@ -2424,7 +2564,7 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, else val = AFMT_U8; } - DPRINTK("SETFMT EXIT, returning %d\n", val); + DPRINTK ("SETFMT EXIT, returning %d\n", val); rc = put_user (val, (int *)arg); break; @@ -2434,18 +2574,19 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, rc = -EFAULT; break; } - DPRINTK("DSP_CHANNELS, val==%d\n", val); + DPRINTK ("DSP_CHANNELS, val==%d\n", val); if (val != 0) { rc = 0; - if (rc == 0 && rd) + + if (rd) rc = via_chan_set_stereo (card, &card->ch_in, val); - if (rc == 0 && wr) + + if (rc >= 0 && wr) rc = via_chan_set_stereo (card, &card->ch_out, val); - if (rc <= 0) { - if (rc == 0) - rc = -EINVAL; + + if (rc < 0) break; - } + val = rc; } else { if ((rd && (card->ch_in.pcm_fmt & VIA_PCM_FMT_STEREO)) || @@ -2454,7 +2595,7 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, else val = 1; } - DPRINTK("CHANNELS EXIT, returning %d\n", val); + DPRINTK ("CHANNELS EXIT, returning %d\n", val); rc = put_user (val, (int *)arg); break; @@ -2464,21 +2605,21 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, rc = -EFAULT; break; } - DPRINTK("DSP_STEREO, val==%d\n", val); + DPRINTK ("DSP_STEREO, val==%d\n", val); rc = 0; - if (rc == 0 && rd) + if (rd) rc = via_chan_set_stereo (card, &card->ch_in, val ? 2 : 1); - if (rc == 0 && wr) + if (rc >= 0 && wr) rc = via_chan_set_stereo (card, &card->ch_out, val ? 2 : 1); - if (rc <= 0) { - if (rc == 0) - rc = -EINVAL; + if (rc < 0) break; - } - DPRINTK("STEREO EXIT, returning %d\n", val); - rc = 0; + + val = rc - 1; + + DPRINTK ("STEREO EXIT, returning %d\n", val); + rc = put_user(val, (int *) arg); break; /* query or set sampling rate */ @@ -2487,7 +2628,7 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, rc = -EFAULT; break; } - DPRINTK("DSP_SPEED, val==%d\n", val); + DPRINTK ("DSP_SPEED, val==%d\n", val); if (val < 0) { rc = -EINVAL; break; @@ -2495,16 +2636,14 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, if (val > 0) { rc = 0; - if (rc == 0 && rd) + if (rd) rc = via_chan_set_speed (card, &card->ch_in, val); - if (rc == 0 && wr) + if (rc >= 0 && wr) rc = via_chan_set_speed (card, &card->ch_out, val); - if (rc <= 0) { - if (rc == 0) - rc = -EINVAL; + if (rc < 0) break; - } + val = rc; } else { if (rd) @@ -2514,7 +2653,7 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, else val = 0; } - DPRINTK("SPEED EXIT, returning %d\n", val); + DPRINTK ("SPEED EXIT, returning %d\n", val); rc = put_user (val, (int *)arg); break; @@ -2522,7 +2661,7 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, case SNDCTL_DSP_SYNC: DPRINTK ("DSP_SYNC\n"); if (wr) { - DPRINTK("SYNC EXIT (after calling via_dsp_drain_playback)\n"); + DPRINTK ("SYNC EXIT (after calling via_dsp_drain_playback)\n"); rc = via_dsp_drain_playback (card, &card->ch_out, nonblock); } break; @@ -2531,12 +2670,19 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, case SNDCTL_DSP_RESET: DPRINTK ("DSP_RESET\n"); if (rd) { - via_chan_clear (&card->ch_in); + via_chan_clear (card, &card->ch_in); via_chan_pcm_fmt (&card->ch_in, 1); + card->ch_in.frag_number = 0; + card->ch_in.frag_size = 0; + atomic_set(&card->ch_in.n_frags, 0); } + if (wr) { - via_chan_clear (&card->ch_out); + via_chan_clear (card, &card->ch_out); via_chan_pcm_fmt (&card->ch_out, 1); + card->ch_out.frag_number = 0; + card->ch_out.frag_size = 0; + atomic_set(&card->ch_out.n_frags, 0); } rc = 0; @@ -2544,40 +2690,47 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, /* obtain bitmask of device capabilities, such as mmap, full duplex, etc. */ case SNDCTL_DSP_GETCAPS: - DPRINTK("DSP_GETCAPS\n"); + DPRINTK ("DSP_GETCAPS\n"); rc = put_user(VIA_DSP_CAP, (int *)arg); break; - /* obtain bitmask of device capabilities, such as mmap, full duplex, etc. */ + /* obtain buffer fragment size */ case SNDCTL_DSP_GETBLKSIZE: - DPRINTK("DSP_GETBLKSIZE\n"); - rc = put_user(VIA_DMA_BUF_SIZE, (int *)arg); + DPRINTK ("DSP_GETBLKSIZE\n"); + + if (rd) { + via_chan_set_buffering(card, &card->ch_in, -1); + rc = put_user(card->ch_in.frag_size, (int *)arg); + } else if (wr) { + via_chan_set_buffering(card, &card->ch_out, -1); + rc = put_user(card->ch_out.frag_size, (int *)arg); + } break; /* obtain information about input buffering */ case SNDCTL_DSP_GETISPACE: - DPRINTK("DSP_GETISPACE\n"); + DPRINTK ("DSP_GETISPACE\n"); if (rd) rc = via_dsp_ioctl_space (card, &card->ch_in, (void*) arg); break; /* obtain information about output buffering */ case SNDCTL_DSP_GETOSPACE: - DPRINTK("DSP_GETOSPACE\n"); + DPRINTK ("DSP_GETOSPACE\n"); if (wr) rc = via_dsp_ioctl_space (card, &card->ch_out, (void*) arg); break; /* obtain information about input hardware pointer */ case SNDCTL_DSP_GETIPTR: - DPRINTK("DSP_GETIPTR\n"); + DPRINTK ("DSP_GETIPTR\n"); if (rd) rc = via_dsp_ioctl_ptr (card, &card->ch_in, (void*) arg); break; /* obtain information about output hardware pointer */ case SNDCTL_DSP_GETOPTR: - DPRINTK("DSP_GETOPTR\n"); + DPRINTK ("DSP_GETOPTR\n"); if (wr) rc = via_dsp_ioctl_ptr (card, &card->ch_out, (void*) arg); break; @@ -2585,25 +2738,29 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, /* return number of bytes remaining to be played by DMA engine */ case SNDCTL_DSP_GETODELAY: { - DPRINTK("DSP_GETODELAY\n"); + DPRINTK ("DSP_GETODELAY\n"); chan = &card->ch_out; if (!wr) break; - val = VIA_DMA_BUFFERS - atomic_read (&chan->n_bufs); + if (chan->is_active) { - if (val > 0) { - val *= VIA_DMA_BUF_SIZE; - val -= VIA_DMA_BUF_SIZE - - inl (chan->iobase + VIA_BASE0_PCM_OUT_BLOCK_COUNT); - } - val += chan->slop_len; + val = chan->frag_number - atomic_read (&chan->n_frags); - assert (val <= (VIA_DMA_BUF_SIZE * VIA_DMA_BUFFERS)); + if (val > 0) { + val *= chan->frag_size; + val -= chan->frag_size - + inl (chan->iobase + VIA_PCM_BLOCK_COUNT); + } + val += chan->slop_len % chan->frag_size; + } else + val = 0; - DPRINTK("GETODELAY EXIT, val = %d bytes\n", val); + assert (val <= (chan->frag_size * chan->frag_number)); + + DPRINTK ("GETODELAY EXIT, val = %d bytes\n", val); rc = put_user (val, (int *)arg); break; } @@ -2617,7 +2774,7 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, rc = -EFAULT; break; } - DPRINTK("DSP_SETTRIGGER, rd=%d, wr=%d, act=%d/%d, en=%d/%d\n", + DPRINTK ("DSP_SETTRIGGER, rd=%d, wr=%d, act=%d/%d, en=%d/%d\n", rd, wr, card->ch_in.is_active, card->ch_out.is_active, card->ch_in.is_enabled, card->ch_out.is_enabled); @@ -2625,6 +2782,7 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, if (rd) rc = via_dsp_ioctl_trigger (&card->ch_in, val); + if (!rc && wr) rc = via_dsp_ioctl_trigger (&card->ch_out, val); @@ -2634,7 +2792,7 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, * with O_RDWR, this is mainly a no-op that always returns success. */ case SNDCTL_DSP_SETDUPLEX: - DPRINTK("DSP_SETDUPLEX\n"); + DPRINTK ("DSP_SETDUPLEX\n"); if (!rd || !wr) break; rc = 0; @@ -2646,7 +2804,13 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, rc = -EFAULT; break; } - DPRINTK("DSP_SETFRAGMENT, val==%d\n", val); + DPRINTK ("DSP_SETFRAGMENT, val==%d\n", val); + + if (rd) + rc = via_chan_set_buffering(card, &card->ch_in, val); + + if (wr) + rc = via_chan_set_buffering(card, &card->ch_out, val); DPRINTK ("SNDCTL_DSP_SETFRAGMENT (fragshift==0x%04X (%d), maxfrags==0x%04X (%d))\n", val & 0xFFFF, @@ -2654,13 +2818,12 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, (val >> 16) & 0xFFFF, (val >> 16) & 0xFFFF); - /* just to shut up some programs */ rc = 0; break; /* inform device of an upcoming pause in input (or output). */ case SNDCTL_DSP_POST: - DPRINTK("DSP_POST\n"); + DPRINTK ("DSP_POST\n"); if (wr) { if (card->ch_out.slop_len > 0) via_chan_flush_frag (&card->ch_out); @@ -2678,15 +2841,14 @@ static int via_dsp_ioctl (struct inode *inode, struct file *file, } up (&card->syscall_sem); - DPRINTK("EXIT, returning %d\n", rc); + DPRINTK ("EXIT, returning %d\n", rc); return rc; } static int via_dsp_open (struct inode *inode, struct file *file) { - int rc, minor = MINOR(inode->i_rdev); - int got_read_chan = 0; + int minor = MINOR(inode->i_rdev); struct via_info *card; struct pci_dev *pdev; struct via_channel *chan; @@ -2733,17 +2895,13 @@ match: } file->private_data = card; - DPRINTK("file->f_mode == 0x%x\n", file->f_mode); + DPRINTK ("file->f_mode == 0x%x\n", file->f_mode); /* handle input from analog source */ if (file->f_mode & FMODE_READ) { chan = &card->ch_in; - rc = via_chan_init (card, chan); - if (rc) - goto err_out; - - got_read_chan = 1; + via_chan_init (card, chan); /* why is this forced to 16-bit stereo in all drivers? */ chan->pcm_fmt = VIA_PCM_FMT_16BIT | VIA_PCM_FMT_STEREO; @@ -2756,30 +2914,28 @@ match: if (file->f_mode & FMODE_WRITE) { chan = &card->ch_out; - rc = via_chan_init (card, chan); - if (rc) - goto err_out_read_chan; + via_chan_init (card, chan); - if ((minor & 0xf) == SND_DEV_DSP16) { - chan->pcm_fmt |= VIA_PCM_FMT_16BIT; - via_set_rate (&card->ac97, chan, 44100); + if (file->f_mode & FMODE_READ) { + /* if in duplex mode make the recording and playback channels + have the same settings */ + chan->pcm_fmt = VIA_PCM_FMT_16BIT | VIA_PCM_FMT_STEREO; + via_chan_pcm_fmt (chan, 0); + via_set_rate (&card->ac97, chan, 44100); } else { - via_set_rate (&card->ac97, chan, 8000); + if ((minor & 0xf) == SND_DEV_DSP16) { + chan->pcm_fmt = VIA_PCM_FMT_16BIT; + via_chan_pcm_fmt (chan, 0); + via_set_rate (&card->ac97, chan, 44100); + } else { + via_chan_pcm_fmt (chan, 0); + via_set_rate (&card->ac97, chan, 8000); + } } - - via_chan_pcm_fmt (chan, 0); } DPRINTK ("EXIT, returning 0\n"); return 0; - -err_out_read_chan: - if (got_read_chan) - via_chan_free (card, &card->ch_in); -err_out: - up (&card->open_sem); - DPRINTK("ERROR EXIT, returning %d\n", rc); - return rc; } @@ -2807,15 +2963,18 @@ static int via_dsp_release(struct inode *inode, struct file *file) printk (KERN_DEBUG "via_audio: ignoring drain playback error %d\n", rc); via_chan_free (card, &card->ch_out); + via_chan_buffer_free(card, &card->ch_out); } - if (file->f_mode & FMODE_READ) + if (file->f_mode & FMODE_READ) { via_chan_free (card, &card->ch_in); + via_chan_buffer_free (card, &card->ch_in); + } up (&card->syscall_sem); up (&card->open_sem); - DPRINTK("EXIT, returning 0\n"); + DPRINTK ("EXIT, returning 0\n"); return 0; } @@ -2934,9 +3093,9 @@ static int __init via_init_one (struct pci_dev *pdev, const struct pci_device_id tmp &= 0xF0; tmp |= pdev->irq; pci_write_config_byte (pdev, 0x3C, tmp); - DPRINTK("new 0x3c==0x%02x\n", tmp); + DPRINTK ("new 0x3c==0x%02x\n", tmp); } else { - DPRINTK("IRQ reg 0x3c==0x%02x, irq==%d\n", + DPRINTK ("IRQ reg 0x3c==0x%02x, irq==%d\n", tmp, tmp & 0x0F); } @@ -3036,12 +3195,12 @@ static int __init init_via82cxxx_audio(void) static void __exit cleanup_via82cxxx_audio(void) { - DPRINTK("ENTER\n"); + DPRINTK ("ENTER\n"); pci_unregister_driver (&via_driver); via_cleanup_proc (); - DPRINTK("EXIT\n"); + DPRINTK ("EXIT\n"); } @@ -3133,7 +3292,7 @@ static int via_info_read_proc (char *page, char **start, off_t off, ); - DPRINTK("EXIT, returning %d\n", len); + DPRINTK ("EXIT, returning %d\n", len); return len; #undef YN diff --git a/drivers/sound/ymfpci.c b/drivers/sound/ymfpci.c index a8cdc58f2..9670951f8 100644 --- a/drivers/sound/ymfpci.c +++ b/drivers/sound/ymfpci.c @@ -24,19 +24,28 @@ * * TODO: * - Use P44Slot for 44.1 playback. - * - Capture and duplex * - 96KHz playback for DVD - use pitch of 2.0. * - uLaw for Sun apps. + * : Alan says firmly "no software format conversion in kernel". * - Retain DMA buffer on close, do not wait the end of frame. * - Cleanup - * ? merge ymf_pcm and state - * ? pcm interrupt no pointer * ? underused structure members * - Remove remaining P3 tags (debug messages). * - Resolve XXX tagged questions. * - Cannot play 5133Hz. + * - 2001/01/07 Consider if we can remove voice_lock, like so: + * : Allocate/deallocate voices in open/close under semafore. + * : We access voices in interrupt, that only for pcms that open. + * voice_lock around playback_prepare closes interrupts for insane duration. + * - Revisit the way voice_alloc is done - too confusing, overcomplicated. + * Should support various channel types, however. + * - Remove prog_dmabuf from read/write, leave it in open. + * - 2001/01/07 Replace the OPL3 part of CONFIG_SOUND_YMFPCI_LEGACY code with + * native synthesizer through a playback slot. + * - Use new 2.3.x cache coherent PCI DMA routines instead of virt_to_bus. */ +#include <linux/config.h> #include <linux/module.h> #include <linux/init.h> #include <linux/ioport.h> @@ -50,19 +59,19 @@ #include <asm/dma.h> #include <asm/uaccess.h> +#ifdef CONFIG_SOUND_YMFPCI_LEGACY +# include "sound_config.h" +# include "mpu401.h" +#endif #include "ymfpci.h" -#define snd_magic_cast(t, p, err) ((t *)(p)) - -/* Channels, such as play and record. I do only play a.t.m. XXX */ -#define NR_HW_CH 1 - -static int ymf_playback_trigger(ymfpci_t *codec, ymfpci_pcm_t *ypcm, int cmd); -static int ymfpci_voice_alloc(ymfpci_t *codec, ymfpci_voice_type_t type, - int pair, ymfpci_voice_t **rvoice); -static int ymfpci_voice_free(ymfpci_t *codec, ymfpci_voice_t *pvoice); -static int ymf_playback_prepare(ymfpci_t *codec, struct ymf_state *state); -static int ymf_state_alloc(ymfpci_t *unit, int nvirt); +static int ymf_playback_trigger(ymfpci_t *codec, struct ymf_pcm *ypcm, int cmd); +static void ymf_capture_trigger(ymfpci_t *unit, struct ymf_pcm *ypcm, int cmd); +static void ymfpci_voice_free(ymfpci_t *codec, ymfpci_voice_t *pvoice); +static int ymf_capture_alloc(struct ymf_unit *unit, int *pbank); +static int ymf_playback_prepare(struct ymf_state *state); +static int ymf_capture_prepare(struct ymf_state *state); +static struct ymf_state *ymf_state_alloc(ymfpci_t *unit); static LIST_HEAD(ymf_devs); @@ -87,7 +96,7 @@ MODULE_DEVICE_TABLE(pci, ymf_id_tbl); /* * Mindlessly copied from cs46xx XXX */ -extern __inline__ unsigned ld2(unsigned int x) +static inline unsigned ld2(unsigned int x) { unsigned r = 0; @@ -281,18 +290,13 @@ static void ymf_pcm_update_shift(struct ymf_pcm_format *f) f->shift++; } -/* - * Whole OSS-style DMA machinery is taken from cs46xx. - */ - /* Are you sure 32K is not too much? See if mpg123 skips on loaded systems. */ #define DMABUF_DEFAULTORDER (15-PAGE_SHIFT) #define DMABUF_MINORDER 1 /* allocate DMA buffer, playback and recording buffer should be allocated seperately */ -static int alloc_dmabuf(struct ymf_state *state) +static int alloc_dmabuf(struct ymf_dmabuf *dmabuf) { - struct ymf_dmabuf *dmabuf = &state->dmabuf; void *rawbuf = NULL; int order; struct page * map, * mapend; @@ -323,9 +327,8 @@ static int alloc_dmabuf(struct ymf_state *state) } /* free DMA buffer */ -static void dealloc_dmabuf(struct ymf_state *state) +static void dealloc_dmabuf(struct ymf_dmabuf *dmabuf) { - struct ymf_dmabuf *dmabuf = &state->dmabuf; struct page *map, *mapend; if (dmabuf->rawbuf) { @@ -339,9 +342,9 @@ static void dealloc_dmabuf(struct ymf_state *state) dmabuf->mapped = dmabuf->ready = 0; } -static int prog_dmabuf(struct ymf_state *state, unsigned rec) +static int prog_dmabuf(struct ymf_state *state, int rec) { - struct ymf_dmabuf *dmabuf = &state->dmabuf; + struct ymf_dmabuf *dmabuf; int w_16; unsigned bytepersec; unsigned bufsize; @@ -350,6 +353,7 @@ static int prog_dmabuf(struct ymf_state *state, unsigned rec) int ret; w_16 = ymf_pcm_format_width(state->format.format) == 16; + dmabuf = rec ? &state->rpcm.dmabuf : &state->wpcm.dmabuf; spin_lock_irqsave(&state->unit->reg_lock, flags); dmabuf->hwptr = dmabuf->swptr = 0; @@ -359,7 +363,7 @@ static int prog_dmabuf(struct ymf_state *state, unsigned rec) /* allocate DMA buffer if not allocated yet */ if (!dmabuf->rawbuf) - if ((ret = alloc_dmabuf(state))) + if ((ret = alloc_dmabuf(dmabuf))) return ret; bytepersec = state->format.rate << state->format.shift; @@ -383,7 +387,6 @@ static int prog_dmabuf(struct ymf_state *state, unsigned rec) dmabuf->numfrag = bufsize >> dmabuf->fragshift; } dmabuf->fragsize = 1 << dmabuf->fragshift; - dmabuf->fragsamples = dmabuf->fragsize >> state->format.shift; dmabuf->dmasize = dmabuf->numfrag << dmabuf->fragshift; /* @@ -414,15 +417,20 @@ static int prog_dmabuf(struct ymf_state *state, unsigned rec) * Now set up the ring */ - spin_lock_irqsave(&state->unit->reg_lock, flags); + /* XXX ret = rec? cap_pre(): pbk_pre(); */ + spin_lock_irqsave(&state->unit->voice_lock, flags); if (rec) { - /* ymf_rec_setup(state); */ + if ((ret = ymf_capture_prepare(state)) != 0) { + spin_unlock_irqrestore(&state->unit->voice_lock, flags); + return ret; + } } else { - if ((ret = ymf_playback_prepare(state->unit, state)) != 0) { + if ((ret = ymf_playback_prepare(state)) != 0) { + spin_unlock_irqrestore(&state->unit->voice_lock, flags); return ret; } } - spin_unlock_irqrestore(&state->unit->reg_lock, flags); + spin_unlock_irqrestore(&state->unit->voice_lock, flags); /* set the ready flag for the dma buffer (this comment is not stupid) */ dmabuf->ready = 1; @@ -439,9 +447,14 @@ static int prog_dmabuf(struct ymf_state *state, unsigned rec) static void ymf_start_dac(struct ymf_state *state) { - ymf_playback_trigger(state->unit, &state->ypcm, 1); + ymf_playback_trigger(state->unit, &state->wpcm, 1); } +// static void ymf_start_adc(struct ymf_state *state) +// { +// ymf_capture_trigger(state->unit, &state->rpcm, 1); +// } + /* * Wait until output is drained. * This does not kill the hardware for the sake of ioctls. @@ -449,14 +462,14 @@ static void ymf_start_dac(struct ymf_state *state) static void ymf_wait_dac(struct ymf_state *state) { struct ymf_unit *unit = state->unit; - ymfpci_pcm_t *ypcm = &state->ypcm; + struct ymf_pcm *ypcm = &state->wpcm; DECLARE_WAITQUEUE(waita, current); unsigned long flags; - add_wait_queue(&state->dmabuf.wait, &waita); + add_wait_queue(&ypcm->dmabuf.wait, &waita); spin_lock_irqsave(&unit->reg_lock, flags); - if (state->dmabuf.count != 0 && !state->ypcm.running) { + if (ypcm->dmabuf.count != 0 && !ypcm->running) { ymf_playback_trigger(unit, ypcm, 1); } @@ -479,7 +492,7 @@ static void ymf_wait_dac(struct ymf_state *state) spin_unlock_irqrestore(&unit->reg_lock, flags); set_current_state(TASK_RUNNING); - remove_wait_queue(&state->dmabuf.wait, &waita); + remove_wait_queue(&ypcm->dmabuf.wait, &waita); /* * This function may take up to 4 seconds to reach this point @@ -487,6 +500,17 @@ static void ymf_wait_dac(struct ymf_state *state) */ } +/* Can just stop, without wait. Or can we? */ +static void ymf_stop_adc(struct ymf_state *state) +{ + struct ymf_unit *unit = state->unit; + unsigned long flags; + + spin_lock_irqsave(&unit->reg_lock, flags); + ymf_capture_trigger(unit, &state->rpcm, 0); + spin_unlock_irqrestore(&unit->reg_lock, flags); +} + /* * Hardware start management */ @@ -523,12 +547,11 @@ static void ymfpci_hw_stop(ymfpci_t *codec) * Playback voice management */ -static int voice_alloc(ymfpci_t *codec, ymfpci_voice_type_t type, int pair, ymfpci_voice_t **rvoice) +static int voice_alloc(ymfpci_t *codec, ymfpci_voice_type_t type, int pair, ymfpci_voice_t *rvoice[]) { ymfpci_voice_t *voice, *voice2; int idx; - - *rvoice = NULL; + for (idx = 0; idx < 64; idx += pair ? 2 : 1) { voice = &codec->voices[idx]; voice2 = pair ? &codec->voices[idx+1] : NULL; @@ -551,52 +574,29 @@ static int voice_alloc(ymfpci_t *codec, ymfpci_voice_type_t type, int pair, ymfp break; } ymfpci_hw_start(codec); - if (voice2) + rvoice[0] = voice; + if (voice2) { ymfpci_hw_start(codec); - *rvoice = voice; + rvoice[1] = voice2; + } return 0; } - return -ENOMEM; + return -EBUSY; /* Your audio channel is open by someone else. */ } -static int ymfpci_voice_alloc(ymfpci_t *codec, ymfpci_voice_type_t type, - int pair, ymfpci_voice_t **rvoice) +static void ymfpci_voice_free(ymfpci_t *unit, ymfpci_voice_t *pvoice) { - unsigned long flags; - int result; - - spin_lock_irqsave(&codec->voice_lock, flags); - for (;;) { - result = voice_alloc(codec, type, pair, rvoice); - if (result == 0 || type != YMFPCI_PCM) - break; - /* TODO: synth/midi voice deallocation */ - break; - } - spin_unlock_irqrestore(&codec->voice_lock, flags); - return result; -} - -static int ymfpci_voice_free(ymfpci_t *codec, ymfpci_voice_t *pvoice) -{ - unsigned long flags; - - ymfpci_hw_stop(codec); - spin_lock_irqsave(&codec->voice_lock, flags); + ymfpci_hw_stop(unit); pvoice->use = pvoice->pcm = pvoice->synth = pvoice->midi = 0; pvoice->ypcm = NULL; - pvoice->interrupt = NULL; - spin_unlock_irqrestore(&codec->voice_lock, flags); - return 0; } /* - * PCM part */ static void ymf_pcm_interrupt(ymfpci_t *codec, ymfpci_voice_t *voice) { - ymfpci_pcm_t *ypcm; + struct ymf_pcm *ypcm; int redzone; int pos, delta, swptr; int played, distance; @@ -611,7 +611,7 @@ static void ymf_pcm_interrupt(ymfpci_t *codec, ymfpci_voice_t *voice) ypcm->running = 0; // lock it return; } - dmabuf = &state->dmabuf; + dmabuf = &ypcm->dmabuf; spin_lock(&codec->reg_lock); if (ypcm->running) { /* P3 */ /** printk("ymfpci: %d, intr bank %d count %d start 0x%x:%x\n", @@ -627,10 +627,9 @@ static void ymf_pcm_interrupt(ymfpci_t *codec, ymfpci_voice_t *voice) pos = voice->bank[codec->active_bank].start; pos <<= state->format.shift; if (pos < 0 || pos >= dmabuf->dmasize) { /* ucode bug */ - printk(KERN_ERR - "ymfpci%d: %d: runaway: hwptr %d dmasize %d\n", + printk(KERN_ERR "ymfpci%d: runaway voice %d: hwptr %d=>%d dmasize %d\n", codec->dev_audio, voice->number, - dmabuf->hwptr, dmabuf->dmasize); + dmabuf->hwptr, pos, dmabuf->dmasize); pos = 0; } if (pos < dmabuf->hwptr) { @@ -706,36 +705,66 @@ static void ymf_pcm_interrupt(ymfpci_t *codec, ymfpci_voice_t *voice) spin_unlock(&codec->reg_lock); } -#if HAVE_RECORD -static void ymfpci_pcm_capture_interrupt(snd_pcm_subchn_t *substream) +static void ymf_cap_interrupt(ymfpci_t *unit, struct ymf_capture *cap) { - snd_pcm_runtime_t *runtime = substream->runtime; - ymfpci_pcm_t *ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, ); - ymfpci_t *codec = ypcm->codec; - u32 pos, delta; - - spin_lock(&codec->reg_lock); + struct ymf_pcm *ypcm; + int redzone; + struct ymf_state *state; + struct ymf_dmabuf *dmabuf; + int pos, delta; + int cnt; + + if ((ypcm = cap->ypcm) == NULL) { + return; + } + if ((state = ypcm->state) == NULL) { + ypcm->running = 0; // lock it + return; + } + dmabuf = &ypcm->dmabuf; + spin_lock(&unit->reg_lock); if (ypcm->running) { - pos = codec->bank_capture[ypcm->capture_bank_number][codec->active_bank]->start << ypcm->shift_offset; - if (pos < ypcm->last_pos) // <-- dmabuf->hwptr - delta = pos + (ypcm->buffer_size - ypcm->last_pos); - else - delta = pos - ypcm->last_pos; - ypcm->frag_pos += delta; - ypcm->last_pos = pos; - while (ypcm->frag_pos >= ypcm->frag_size) { - ypcm->frag_pos -= ypcm->frag_size; - // printk("done - active_bank = 0x%x, start = 0x%x\n", codec->active_bank, voice->bank[codec->active_bank].start); - spin_unlock(&codec->reg_lock); - snd_pcm_transfer_done(substream); - spin_lock(&codec->reg_lock); + redzone = ymf_calc_lend(state->format.rate); + redzone <<= (state->format.shift + 1); + + pos = cap->bank[unit->active_bank].start; + // pos <<= state->format.shift; + if (pos < 0 || pos >= dmabuf->dmasize) { /* ucode bug */ + printk(KERN_ERR "ymfpci%d: runaway capture %d: hwptr %d=>%d dmasize %d\n", + unit->dev_audio, ypcm->capture_bank_number, + dmabuf->hwptr, pos, dmabuf->dmasize); + pos = 0; + } + if (pos < dmabuf->hwptr) { + delta = dmabuf->dmasize - dmabuf->hwptr; + delta += pos; + } else { + delta = pos - dmabuf->hwptr; + } + dmabuf->hwptr = pos; + + cnt = dmabuf->count; + cnt += delta; + if (cnt + redzone > dmabuf->dmasize) { + /* Overflow - bump swptr */ + dmabuf->count = dmabuf->dmasize - redzone; + dmabuf->swptr = dmabuf->hwptr + redzone; + if (dmabuf->swptr >= dmabuf->dmasize) { + dmabuf->swptr -= dmabuf->dmasize; + } + } else { + dmabuf->count = cnt; + } + + dmabuf->total_bytes += delta; + if (dmabuf->count) { /* && is_sleeping XXX */ + wake_up(&dmabuf->wait); } } - spin_unlock(&codec->reg_lock); + spin_unlock(&unit->reg_lock); } -#endif -static int ymf_playback_trigger(ymfpci_t *codec, ymfpci_pcm_t *ypcm, int cmd) +static int ymf_playback_trigger(ymfpci_t *codec, struct ymf_pcm *ypcm, int cmd) { if (ypcm->voices[0] == NULL) { @@ -755,40 +784,29 @@ static int ymf_playback_trigger(ymfpci_t *codec, ymfpci_pcm_t *ypcm, int cmd) return 0; } -#if HAVE_RECORD -static int ymfpci_capture_trigger(void *private_data, - snd_pcm_subchn_t * substream, - int cmd) +static void ymf_capture_trigger(ymfpci_t *codec, struct ymf_pcm *ypcm, int cmd) { - unsigned long flags; - ymfpci_t *codec = snd_magic_cast(ymfpci_t, private_data, -ENXIO); - ymfpci_pcm_t *ypcm = snd_magic_cast(ymfpci_pcm_t, substream->runtime->private_data, -ENXIO); - int result = 0; u32 tmp; - spin_lock_irqsave(&codec->reg_lock, flags); - if (cmd == SND_PCM_TRIGGER_GO) { + if (cmd != 0) { tmp = ymfpci_readl(codec, YDSXGR_MAPOFREC) | (1 << ypcm->capture_bank_number); ymfpci_writel(codec, YDSXGR_MAPOFREC, tmp); ypcm->running = 1; - } else if (cmd == SND_PCM_TRIGGER_STOP) { + } else { tmp = ymfpci_readl(codec, YDSXGR_MAPOFREC) & ~(1 << ypcm->capture_bank_number); ymfpci_writel(codec, YDSXGR_MAPOFREC, tmp); ypcm->running = 0; - } else { - result = -EINVAL; } - spin_unlock_irqrestore(&codec->reg_lock, flags); - return result; } -#endif -static int ymfpci_pcm_voice_alloc(ymfpci_pcm_t *ypcm, int voices) +static int ymfpci_pcm_voice_alloc(struct ymf_pcm *ypcm, int voices) { + struct ymf_unit *unit; int err; + unit = ypcm->state->unit; if (ypcm->voices[1] != NULL && voices < 2) { - ymfpci_voice_free(ypcm->codec, ypcm->voices[1]); + ymfpci_voice_free(unit, ypcm->voices[1]); ypcm->voices[1] = NULL; } if (voices == 1 && ypcm->voices[0] != NULL) @@ -797,18 +815,17 @@ static int ymfpci_pcm_voice_alloc(ymfpci_pcm_t *ypcm, int voices) return 0; /* already allocated */ if (voices > 1) { if (ypcm->voices[0] != NULL && ypcm->voices[1] == NULL) { - ymfpci_voice_free(ypcm->codec, ypcm->voices[0]); + ymfpci_voice_free(unit, ypcm->voices[0]); ypcm->voices[0] = NULL; } - } - err = ymfpci_voice_alloc(ypcm->codec, YMFPCI_PCM, voices > 1, &ypcm->voices[0]); - if (err < 0) - return err; - ypcm->voices[0]->ypcm = ypcm; - ypcm->voices[0]->interrupt = ymf_pcm_interrupt; - if (voices > 1) { - ypcm->voices[1] = &ypcm->codec->voices[ypcm->voices[0]->number + 1]; + if ((err = voice_alloc(unit, YMFPCI_PCM, 1, ypcm->voices)) < 0) + return err; + ypcm->voices[0]->ypcm = ypcm; ypcm->voices[1]->ypcm = ypcm; + } else { + if ((err = voice_alloc(unit, YMFPCI_PCM, 0, ypcm->voices)) < 0) + return err; + ypcm->voices[0]->ypcm = ypcm; } return 0; } @@ -901,17 +918,32 @@ static void ymf_pcm_init_voice(ymfpci_voice_t *voice, int stereo, } /* - * XXX Use new cache coherent PCI DMA routines instead of virt_to_bus. + * XXX Capture channel allocation is entirely fake at the moment. + * We use only one channel and mark it busy as required. */ -static int ymf_playback_prepare(ymfpci_t *codec, struct ymf_state *state) +static int ymf_capture_alloc(struct ymf_unit *unit, int *pbank) { - ymfpci_pcm_t *ypcm = &state->ypcm; + struct ymf_capture *cap; + int cbank; + + cbank = 1; /* Only ADC slot is used for now. */ + cap = &unit->capture[cbank]; + if (cap->use) + return -EBUSY; + cap->use = 1; + *pbank = cbank; + return 0; +} + +static int ymf_playback_prepare(struct ymf_state *state) +{ + struct ymf_pcm *ypcm = &state->wpcm; int err, nvoice; if ((err = ymfpci_pcm_voice_alloc(ypcm, state->format.voices)) < 0) { - /* Cannot be unless we leak voices in ymf_release! */ - printk(KERN_ERR "ymfpci%d: cannot allocate voice!\n", - codec->dev_audio); + /* Somebody started 32 mpg123's in parallel? */ + /* P3 */ printk("ymfpci%d: cannot allocate voice\n", + state->unit->dev_audio); return err; } @@ -919,101 +951,76 @@ static int ymf_playback_prepare(ymfpci_t *codec, struct ymf_state *state) ymf_pcm_init_voice(ypcm->voices[nvoice], state->format.voices == 2, state->format.rate, ymf_pcm_format_width(state->format.format) == 16, - virt_to_bus(state->dmabuf.rawbuf), state->dmabuf.dmasize, + virt_to_bus(ypcm->dmabuf.rawbuf), ypcm->dmabuf.dmasize, ypcm->spdif); } return 0; } -#if 0 /* old */ -static int ymfpci_capture_prepare(void *private_data, - snd_pcm_subchn_t * substream) +static int ymf_capture_prepare(struct ymf_state *state) { - ymfpci_t *codec = snd_magic_cast(ymfpci_t, private_data, -ENXIO); - snd_pcm_runtime_t *runtime = substream->runtime; - ymfpci_pcm_t *ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, -ENXIO); + ymfpci_t *unit = state->unit; + struct ymf_pcm *ypcm = &state->rpcm; ymfpci_capture_bank_t * bank; - int nbank; + /* XXX This is confusing, gotta rename one of them banks... */ + int nbank; /* flip-flop bank */ + int cbank; /* input [super-]bank */ + struct ymf_capture *cap; u32 rate, format; - ypcm->frag_size = snd_pcm_lib_transfer_fragment(substream); - ypcm->buffer_size = snd_pcm_lib_transfer_size(substream); - ypcm->frag_pos = 0; - ypcm->last_pos = 0; - ypcm->shift_offset = 0; - rate = ((48000 * 4096) / runtime->format.rate) - 1; + if (ypcm->capture_bank_number == -1) { + if (ymf_capture_alloc(unit, &cbank) != 0) + return -EBUSY; + + ypcm->capture_bank_number = cbank; + + cap = &unit->capture[cbank]; + cap->bank = unit->bank_capture[cbank][0]; + cap->ypcm = ypcm; + ymfpci_hw_start(unit); + } + + // ypcm->frag_size = snd_pcm_lib_transfer_fragment(substream); + // frag_size is replaced with nonfragged byte-aligned rolling buffer + rate = ((48000 * 4096) / state->format.rate) - 1; format = 0; - if (runtime->format.voices == 2) + if (state->format.voices == 2) format |= 2; - if (snd_pcm_format_width(runtime->format.format) == 8) + if (ymf_pcm_format_width(state->format.format) == 8) format |= 1; switch (ypcm->capture_bank_number) { case 0: - ymfpci_writel(codec, YDSXGR_RECFORMAT, format); - ymfpci_writel(codec, YDSXGR_RECSLOTSR, rate); + ymfpci_writel(unit, YDSXGR_RECFORMAT, format); + ymfpci_writel(unit, YDSXGR_RECSLOTSR, rate); break; case 1: - ymfpci_writel(codec, YDSXGR_ADCFORMAT, format); - ymfpci_writel(codec, YDSXGR_ADCSLOTSR, rate); + ymfpci_writel(unit, YDSXGR_ADCFORMAT, format); + ymfpci_writel(unit, YDSXGR_ADCSLOTSR, rate); break; } for (nbank = 0; nbank < 2; nbank++) { - bank = codec->bank_capture[ypcm->capture_bank_number][nbank]; - bank->base = virt_to_bus(runtime->dma_area->buf); - bank->loop_end = ypcm->buffer_size; + bank = unit->bank_capture[ypcm->capture_bank_number][nbank]; + bank->base = virt_to_bus(ypcm->dmabuf.rawbuf); + // bank->loop_end = ypcm->dmabuf.dmasize >> state->format.shift; + bank->loop_end = ypcm->dmabuf.dmasize; bank->start = 0; bank->num_of_loops = 0; } - if (runtime->digital.dig_valid) - /*runtime->digital.type == SND_PCM_DIG_AES_IEC958*/ - ymfpci_writew(codec, YDSXGR_SPDIFOUTSTATUS, runtime->digital.dig_status[0] | - (runtime->digital.dig_status[1] << 8)); +#if 0 /* s/pdif */ + if (state->digital.dig_valid) + /*state->digital.type == SND_PCM_DIG_AES_IEC958*/ + ymfpci_writew(codec, YDSXGR_SPDIFOUTSTATUS, + state->digital.dig_status[0] | (state->digital.dig_status[1] << 8)); +#endif return 0; } -static unsigned int ymfpci_playback_pointer(void *private_data, - snd_pcm_subchn_t * substream) -{ - ymfpci_t *codec = snd_magic_cast(ymfpci_t, private_data, -ENXIO); - snd_pcm_runtime_t *runtime = substream->runtime; - ymfpci_pcm_t *ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, -ENXIO); - ymfpci_voice_t *voice = ypcm->voices[0]; - unsigned long flags; - unsigned int result; - - spin_lock_irqsave(&codec->reg_lock, flags); - if (ypcm->running && voice) - result = voice->bank[codec->active_bank].start << ypcm->shift_offset; - else - result = 0; - spin_unlock_irqrestore(&codec->reg_lock, flags); - return result; -} - -static unsigned int ymfpci_capture_pointer(void *private_data, - snd_pcm_subchn_t * substream) -{ - ymfpci_t *codec = snd_magic_cast(ymfpci_t, private_data, -ENXIO); - snd_pcm_runtime_t *runtime = substream->runtime; - ymfpci_pcm_t *ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, -ENXIO); - unsigned long flags; - unsigned int result; - - spin_lock_irqsave(&codec->reg_lock, flags); - if (ypcm->running) - result = codec->bank_capture[ypcm->capture_bank_number][codec->active_bank]->start << ypcm->shift_offset; - else - result = 0; - spin_unlock_irqrestore(&codec->reg_lock, flags); - return result; -} -#endif /* old */ - void ymf_interrupt(int irq, void *dev_id, struct pt_regs *regs) { ymfpci_t *codec = dev_id; u32 status, nvoice, mode; - ymfpci_voice_t *voice; + struct ymf_voice *voice; + struct ymf_capture *cap; status = ymfpci_readl(codec, YDSXGR_STATUS); if (status & 0x80000000) { @@ -1026,8 +1033,13 @@ void ymf_interrupt(int irq, void *dev_id, struct pt_regs *regs) spin_lock(&codec->voice_lock); for (nvoice = 0; nvoice < 64; nvoice++) { voice = &codec->voices[nvoice]; - if (voice->interrupt) - voice->interrupt(codec, voice); + if (voice->use) + ymf_pcm_interrupt(codec, voice); + } + for (nvoice = 0; nvoice < 5; nvoice++) { + cap = &codec->capture[nvoice]; + if (cap->use) + ymf_cap_interrupt(codec, cap); } spin_unlock(&codec->voice_lock); } @@ -1039,22 +1051,32 @@ void ymf_interrupt(int irq, void *dev_id, struct pt_regs *regs) } } -static void ymf_pcm_free_substream(ymfpci_pcm_t *ypcm) +static void ymf_pcm_free_substream(struct ymf_pcm *ypcm) { - ymfpci_t *codec; + unsigned long flags; + struct ymf_unit *unit; + + unit = ypcm->state->unit; - if (ypcm) { - codec = ypcm->codec; + if (ypcm->type == PLAYBACK_VOICE) { + spin_lock_irqsave(&unit->voice_lock, flags); if (ypcm->voices[1]) - ymfpci_voice_free(codec, ypcm->voices[1]); + ymfpci_voice_free(unit, ypcm->voices[1]); if (ypcm->voices[0]) - ymfpci_voice_free(codec, ypcm->voices[0]); + ymfpci_voice_free(unit, ypcm->voices[0]); + spin_unlock_irqrestore(&unit->voice_lock, flags); + } else { + if (ypcm->capture_bank_number != -1) { + unit->capture[ypcm->capture_bank_number].use = 0; + ypcm->capture_bank_number = -1; + ymfpci_hw_stop(unit); + } } } -static int ymf_state_alloc(ymfpci_t *unit, int nvirt) +static struct ymf_state *ymf_state_alloc(ymfpci_t *unit) { - ymfpci_pcm_t *ypcm; + struct ymf_pcm *ypcm; struct ymf_state *state; if ((state = kmalloc(sizeof(struct ymf_state), GFP_KERNEL)) == NULL) { @@ -1062,61 +1084,31 @@ static int ymf_state_alloc(ymfpci_t *unit, int nvirt) } memset(state, 0, sizeof(struct ymf_state)); - init_waitqueue_head(&state->dmabuf.wait); - - ypcm = &state->ypcm; + ypcm = &state->wpcm; ypcm->state = state; - ypcm->codec = unit; ypcm->type = PLAYBACK_VOICE; + ypcm->capture_bank_number = -1; + init_waitqueue_head(&ypcm->dmabuf.wait); + + ypcm = &state->rpcm; + ypcm->state = state; + ypcm->type = CAPTURE_AC97; + ypcm->capture_bank_number = -1; + init_waitqueue_head(&ypcm->dmabuf.wait); state->unit = unit; - state->virt = nvirt; state->format.format = AFMT_U8; state->format.rate = 8000; state->format.voices = 1; ymf_pcm_update_shift(&state->format); - unit->states[nvirt] = state; - return 0; + return state; out0: - return -ENOMEM; + return NULL; } -#if HAVE_RECORD - -static int ymfpci_capture_open(void *private_data, - snd_pcm_subchn_t * substream, - u32 capture_bank_number) -{ - ymfpci_t *codec = snd_magic_cast(ymfpci_t, private_data, -ENXIO); - snd_pcm_runtime_t *runtime = substream->runtime; - ymfpci_pcm_t *ypcm; - int err; - - if ((err = snd_pcm_dma_alloc(substream, !capture_bank_number ? codec->dma2ptr : codec->dma3ptr, "YMFPCI - ADC")) < 0) - return err; - ypcm = snd_magic_kcalloc(ymfpci_pcm_t, 0, GFP_KERNEL); - if (ypcm == NULL) { - snd_pcm_dma_free(substream); - return -ENOMEM; - } - ypcm->codec = codec; - ypcm->type = capture_bank_number + CAPTURE_REC; - ypcm->substream = substream; - ypcm->capture_bank_number = capture_bank_number; - codec->capture_substream[capture_bank_number] = substream; - runtime->hw = &ymfpci_capture; - snd_pcm_set_mixer(substream, codec->mixer->device, codec->ac97->me_capture); - runtime->private_data = ypcm; - runtime->private_free = ymfpci_pcm_free_substream; - ymfpci_hw_start(codec); - return 0; -} - -#endif /* old */ - /* AES/IEC958 channel status bits */ #define SND_PCM_AES0_PROFESSIONAL (1<<0) /* 0 = consumer, 1 = professional */ #define SND_PCM_AES0_NONAUDIO (1<<1) /* 0 = audio, 1 = non-audio */ @@ -1202,24 +1194,6 @@ static int ymfpci_capture_open(void *private_data, #define SND_PCM_AES3_CON_CLOCK_50PPM (1<<4) /* 50 ppm */ #define SND_PCM_AES3_CON_CLOCK_VARIABLE (2<<4) /* variable pitch */ -#if HAVE_RECORD /* old */ - -static int ymfpci_capture_close(void *private_data, - snd_pcm_subchn_t * substream) -{ - ymfpci_t *codec = snd_magic_cast(ymfpci_t, private_data, -ENXIO); - snd_pcm_runtime_t *runtime = substream->runtime; - ymfpci_pcm_t *ypcm = snd_magic_cast(ymfpci_pcm_t, runtime->private_data, -ENXIO); - - if (ypcm != NULL) { - codec->capture_substream[ypcm->capture_bank_number] = NULL; - ymfpci_hw_stop(codec); - } - snd_pcm_dma_free(substream); - return 0; -} -#endif - /* * User interface */ @@ -1229,21 +1203,21 @@ static loff_t ymf_llseek(struct file *file, loff_t offset, int origin) return -ESPIPE; } -/* in this loop, dmabuf.count signifies the amount of data that is waiting to be copied to - the user's buffer. it is filled by the dma machine and drained by this loop. */ -static ssize_t ymf_read(struct file *file, char *buffer, size_t count, loff_t *ppos) +/* + * in this loop, dmabuf.count signifies the amount of data that is + * waiting to be copied to the user's buffer. it is filled by the dma + * machine and drained by this loop. + */ +static ssize_t +ymf_read(struct file *file, char *buffer, size_t count, loff_t *ppos) { -#if HAVE_RECORD - struct cs_state *state = (struct cs_state *)file->private_data; - struct dmabuf *dmabuf = &state->dmabuf; + struct ymf_state *state = (struct ymf_state *)file->private_data; + struct ymf_dmabuf *dmabuf = &state->rpcm.dmabuf; + DECLARE_WAITQUEUE(waita, current); ssize_t ret; unsigned long flags; - unsigned swptr; - int cnt; - -#ifdef DEBUG - printk("cs461x: cs_read called, count = %d\n", count); -#endif + unsigned int swptr; + int cnt; /* This many to go in this revolution */ if (ppos != &file->f_pos) return -ESPIPE; @@ -1255,19 +1229,14 @@ static ssize_t ymf_read(struct file *file, char *buffer, size_t count, loff_t *p return -EFAULT; ret = 0; + add_wait_queue(&dmabuf->wait, &waita); while (count > 0) { - spin_lock_irqsave(&state->card->lock, flags); - if (dmabuf->count > (signed) dmabuf->dmasize) { - /* buffer overrun, we are recovering from sleep_on_timeout, - resync hwptr and swptr, make process flush the buffer */ - dmabuf->count = dmabuf->dmasize; - dmabuf->swptr = dmabuf->hwptr; - } + spin_lock_irqsave(&state->unit->reg_lock, flags); swptr = dmabuf->swptr; cnt = dmabuf->dmasize - swptr; if (dmabuf->count < cnt) cnt = dmabuf->count; - spin_unlock_irqrestore(&state->card->lock, flags); + spin_unlock_irqrestore(&state->unit->reg_lock, flags); if (cnt > count) cnt = count; @@ -1275,14 +1244,18 @@ static ssize_t ymf_read(struct file *file, char *buffer, size_t count, loff_t *p unsigned long tmo; /* buffer is empty, start the dma machine and wait for data to be recorded */ - start_adc(state); + spin_lock_irqsave(&state->unit->reg_lock, flags); + if (!state->rpcm.running) { + ymf_capture_trigger(state->unit, &state->rpcm, 1); + } + spin_unlock_irqrestore(&state->unit->reg_lock, flags); if (file->f_flags & O_NONBLOCK) { if (!ret) ret = -EAGAIN; - return ret; + break; } /* This isnt strictly right for the 810 but it'll do */ - tmo = (dmabuf->dmasize * HZ) / (dmabuf->rate * 2); - tmo >>= sample_shift[dmabuf->fmt]; + tmo = (dmabuf->dmasize * HZ) / (state->format.rate * 2); + tmo >>= state->format.shift; /* There are two situations when sleep_on_timeout returns, one is when the interrupt is serviced correctly and the process is waked up by ISR ON TIME. Another is when timeout is expired, which means that @@ -1290,50 +1263,56 @@ static ssize_t ymf_read(struct file *file, char *buffer, size_t count, loff_t *p is TOO LATE for the process to be scheduled to run (scheduler latency) which results in a (potential) buffer overrun. And worse, there is NOTHING we can do to prevent it. */ - if (!interruptible_sleep_on_timeout(&dmabuf->wait, tmo)) { -#ifdef DEBUG - printk(KERN_ERR "cs461x: recording schedule timeout, " - "dmasz %u fragsz %u count %i hwptr %u swptr %u\n", - dmabuf->dmasize, dmabuf->fragsize, dmabuf->count, - dmabuf->hwptr, dmabuf->swptr); -#endif - /* a buffer overrun, we delay the recovery untill next time the - while loop begin and we REALLY have space to record */ + set_current_state(TASK_INTERRUPTIBLE); + tmo = schedule_timeout(tmo); + spin_lock_irqsave(&state->unit->reg_lock, flags); + if (tmo == 0 && dmabuf->count == 0) { + printk(KERN_ERR "ymfpci%d: recording schedule timeout, " + "dmasz %u fragsz %u count %i hwptr %u swptr %u\n", + state->unit->dev_audio, + dmabuf->dmasize, dmabuf->fragsize, dmabuf->count, + dmabuf->hwptr, dmabuf->swptr); } + spin_unlock_irqrestore(&state->unit->reg_lock, flags); if (signal_pending(current)) { ret = ret ? ret : -ERESTARTSYS; - return ret; + break; } continue; } if (copy_to_user(buffer, dmabuf->rawbuf + swptr, cnt)) { if (!ret) ret = -EFAULT; - return ret; + break; } swptr = (swptr + cnt) % dmabuf->dmasize; - spin_lock_irqsave(&state->card->lock, flags); + spin_lock_irqsave(&state->unit->reg_lock, flags); dmabuf->swptr = swptr; dmabuf->count -= cnt; - spin_unlock_irqrestore(&state->card->lock, flags); + // spin_unlock_irqrestore(&state->unit->reg_lock, flags); count -= cnt; buffer += cnt; ret += cnt; - start_adc(state); + // spin_lock_irqsave(&state->unit->reg_lock, flags); + if (!state->rpcm.running) { + ymf_capture_trigger(state->unit, &state->rpcm, 1); + } + spin_unlock_irqrestore(&state->unit->reg_lock, flags); } + set_current_state(TASK_RUNNING); + remove_wait_queue(&dmabuf->wait, &waita); + return ret; -#else - return -EINVAL; -#endif } -static ssize_t ymf_write(struct file *file, const char *buffer, size_t count, loff_t *ppos) +static ssize_t +ymf_write(struct file *file, const char *buffer, size_t count, loff_t *ppos) { struct ymf_state *state = (struct ymf_state *)file->private_data; - struct ymf_dmabuf *dmabuf = &state->dmabuf; + struct ymf_dmabuf *dmabuf = &state->wpcm.dmabuf; DECLARE_WAITQUEUE(waita, current); ssize_t ret; unsigned long flags; @@ -1375,7 +1354,7 @@ static ssize_t ymf_write(struct file *file, const char *buffer, size_t count, lo } if (dmabuf->count == 0) { swptr = dmabuf->hwptr; - if (state->ypcm.running) { + if (state->wpcm.running) { /* * Add uncertainty reserve. */ @@ -1410,8 +1389,8 @@ static ssize_t ymf_write(struct file *file, const char *buffer, size_t count, lo * wait for data to be played */ spin_lock_irqsave(&state->unit->reg_lock, flags); - if (!state->ypcm.running) { - ymf_playback_trigger(state->unit, &state->ypcm, 1); + if (!state->wpcm.running) { + ymf_playback_trigger(state->unit, &state->wpcm, 1); } spin_unlock_irqrestore(&state->unit->reg_lock, flags); if (file->f_flags & O_NONBLOCK) { @@ -1448,8 +1427,8 @@ static ssize_t ymf_write(struct file *file, const char *buffer, size_t count, lo */ delay = state->format.rate / 20; /* 50ms */ delay <<= state->format.shift; - if (dmabuf->count >= delay && !state->ypcm.running) { - ymf_playback_trigger(state->unit, &state->ypcm, 1); + if (dmabuf->count >= delay && !state->wpcm.running) { + ymf_playback_trigger(state->unit, &state->wpcm, 1); } spin_unlock_irqrestore(&state->unit->reg_lock, flags); @@ -1469,19 +1448,23 @@ static ssize_t ymf_write(struct file *file, const char *buffer, size_t count, lo static unsigned int ymf_poll(struct file *file, struct poll_table_struct *wait) { struct ymf_state *state = (struct ymf_state *)file->private_data; - struct ymf_dmabuf *dmabuf = &state->dmabuf; + struct ymf_dmabuf *dmabuf; unsigned long flags; unsigned int mask = 0; - if (file->f_mode & (FMODE_WRITE | FMODE_READ)) - poll_wait(file, &dmabuf->wait, wait); + if (file->f_mode & FMODE_WRITE) + poll_wait(file, &state->wpcm.dmabuf.wait, wait); + // if (file->f_mode & FMODE_READ) + // poll_wait(file, &dmabuf->wait, wait); spin_lock_irqsave(&state->unit->reg_lock, flags); if (file->f_mode & FMODE_READ) { + dmabuf = &state->rpcm.dmabuf; if (dmabuf->count >= (signed)dmabuf->fragsize) mask |= POLLIN | POLLRDNORM; } if (file->f_mode & FMODE_WRITE) { + dmabuf = &state->wpcm.dmabuf; if (dmabuf->mapped) { if (dmabuf->count >= (signed)dmabuf->fragsize) mask |= POLLOUT | POLLWRNORM; @@ -1498,10 +1481,9 @@ static unsigned int ymf_poll(struct file *file, struct poll_table_struct *wait) static int ymf_mmap(struct file *file, struct vm_area_struct *vma) { struct ymf_state *state = (struct ymf_state *)file->private_data; - struct ymf_dmabuf *dmabuf = &state->dmabuf; + struct ymf_dmabuf *dmabuf = &state->wpcm.dmabuf; int ret; unsigned long size; - if (vma->vm_flags & VM_WRITE) { if ((ret = prog_dmabuf(state, 0)) != 0) @@ -1529,7 +1511,7 @@ static int ymf_ioctl(struct inode *inode, struct file *file, unsigned int cmd, unsigned long arg) { struct ymf_state *state = (struct ymf_state *)file->private_data; - struct ymf_dmabuf *dmabuf = &state->dmabuf; + struct ymf_dmabuf *dmabuf; unsigned long flags; audio_buf_info abinfo; count_info cinfo; @@ -1542,28 +1524,30 @@ static int ymf_ioctl(struct inode *inode, struct file *file, case SNDCTL_DSP_RESET: if (file->f_mode & FMODE_WRITE) { ymf_wait_dac(state); + dmabuf = &state->wpcm.dmabuf; spin_lock_irqsave(&state->unit->reg_lock, flags); dmabuf->ready = 0; - dmabuf->swptr = dmabuf->hwptr = 0; + dmabuf->swptr = dmabuf->hwptr; dmabuf->count = dmabuf->total_bytes = 0; spin_unlock_irqrestore(&state->unit->reg_lock, flags); } -#if HAVE_RECORD if (file->f_mode & FMODE_READ) { - stop_adc(state); - synchronize_irq(); + ymf_stop_adc(state); + dmabuf = &state->rpcm.dmabuf; + spin_lock_irqsave(&state->unit->reg_lock, flags); dmabuf->ready = 0; - dmabuf->swptr = dmabuf->hwptr = 0; + dmabuf->swptr = dmabuf->hwptr; dmabuf->count = dmabuf->total_bytes = 0; + spin_unlock_irqrestore(&state->unit->reg_lock, flags); } -#endif return 0; case SNDCTL_DSP_SYNC: if (file->f_mode & FMODE_WRITE) { + dmabuf = &state->wpcm.dmabuf; if (file->f_flags & O_NONBLOCK) { spin_lock_irqsave(&state->unit->reg_lock, flags); - if (dmabuf->count != 0 && !state->ypcm.running) { + if (dmabuf->count != 0 && !state->wpcm.running) { ymf_start_dac(state); } spin_unlock_irqrestore(&state->unit->reg_lock, flags); @@ -1571,6 +1555,7 @@ static int ymf_ioctl(struct inode *inode, struct file *file, ymf_wait_dac(state); } } + /* XXX What does this do for reading? dmabuf->count=0; ? */ return 0; case SNDCTL_DSP_SPEED: /* set smaple rate */ @@ -1579,17 +1564,22 @@ static int ymf_ioctl(struct inode *inode, struct file *file, if (val >= 8000 && val <= 48000) { if (file->f_mode & FMODE_WRITE) { ymf_wait_dac(state); + dmabuf = &state->wpcm.dmabuf; + spin_lock_irqsave(&state->unit->reg_lock, flags); + dmabuf->ready = 0; + state->format.rate = val; + ymf_pcm_update_shift(&state->format); + spin_unlock_irqrestore(&state->unit->reg_lock, flags); } -#if HAVE_RECORD if (file->f_mode & FMODE_READ) { - stop_adc(state); + ymf_stop_adc(state); + dmabuf = &state->rpcm.dmabuf; + spin_lock_irqsave(&state->unit->reg_lock, flags); + dmabuf->ready = 0; + state->format.rate = val; + ymf_pcm_update_shift(&state->format); + spin_unlock_irqrestore(&state->unit->reg_lock, flags); } -#endif - spin_lock_irqsave(&state->unit->reg_lock, flags); - dmabuf->ready = 0; - state->format.rate = val; - ymf_pcm_update_shift(&state->format); - spin_unlock_irqrestore(&state->unit->reg_lock, flags); } return put_user(state->format.rate, (int *)arg); @@ -1597,41 +1587,41 @@ static int ymf_ioctl(struct inode *inode, struct file *file, * OSS manual does not mention SNDCTL_DSP_STEREO at all. * All channels are mono and if you want stereo, you * play into two channels with SNDCTL_DSP_CHANNELS. - * However, mpg123 uses it. I wonder, why Michael Hipp uses it. + * However, mpg123 calls it. I wonder, why Michael Hipp used it. */ case SNDCTL_DSP_STEREO: /* set stereo or mono channel */ if (get_user(val, (int *)arg)) return -EFAULT; if (file->f_mode & FMODE_WRITE) { ymf_wait_dac(state); + dmabuf = &state->wpcm.dmabuf; spin_lock_irqsave(&state->unit->reg_lock, flags); dmabuf->ready = 0; state->format.voices = val ? 2 : 1; ymf_pcm_update_shift(&state->format); spin_unlock_irqrestore(&state->unit->reg_lock, flags); } -#if HAVE_RECORD if (file->f_mode & FMODE_READ) { - /* stop_adc(state); */ + ymf_stop_adc(state); + dmabuf = &state->rpcm.dmabuf; + spin_lock_irqsave(&state->unit->reg_lock, flags); dmabuf->ready = 0; - if(val) - dmabuf->fmt |= CS_FMT_STEREO; - else - dmabuf->fmt &= ~CS_FMT_STEREO; + state->format.voices = val ? 2 : 1; + ymf_pcm_update_shift(&state->format); + spin_unlock_irqrestore(&state->unit->reg_lock, flags); } -#endif return 0; case SNDCTL_DSP_GETBLKSIZE: if (file->f_mode & FMODE_WRITE) { if ((val = prog_dmabuf(state, 0))) return val; - return put_user(dmabuf->fragsize, (int *)arg); + return put_user(state->wpcm.dmabuf.fragsize, (int *)arg); } if (file->f_mode & FMODE_READ) { if ((val = prog_dmabuf(state, 1))) return val; - return put_user(dmabuf->fragsize, (int *)arg); + return put_user(state->rpcm.dmabuf.fragsize, (int *)arg); } return -EINVAL; @@ -1644,17 +1634,22 @@ static int ymf_ioctl(struct inode *inode, struct file *file, if (val == AFMT_S16_LE || val == AFMT_U8) { if (file->f_mode & FMODE_WRITE) { ymf_wait_dac(state); + dmabuf = &state->wpcm.dmabuf; + spin_lock_irqsave(&state->unit->reg_lock, flags); + dmabuf->ready = 0; + state->format.format = val; + ymf_pcm_update_shift(&state->format); + spin_unlock_irqrestore(&state->unit->reg_lock, flags); } -#if HAVE_RECORD if (file->f_mode & FMODE_READ) { - stop_adc(state); + ymf_stop_adc(state); + dmabuf = &state->rpcm.dmabuf; + spin_lock_irqsave(&state->unit->reg_lock, flags); + dmabuf->ready = 0; + state->format.format = val; + ymf_pcm_update_shift(&state->format); + spin_unlock_irqrestore(&state->unit->reg_lock, flags); } -#endif - spin_lock_irqsave(&state->unit->reg_lock, flags); - dmabuf->ready = 0; - state->format.format = val; - ymf_pcm_update_shift(&state->format); - spin_unlock_irqrestore(&state->unit->reg_lock, flags); } return put_user(state->format.format, (int *)arg); @@ -1667,20 +1662,24 @@ static int ymf_ioctl(struct inode *inode, struct file *file, ymf_wait_dac(state); if (val == 1 || val == 2) { spin_lock_irqsave(&state->unit->reg_lock, flags); + dmabuf = &state->wpcm.dmabuf; dmabuf->ready = 0; state->format.voices = val; ymf_pcm_update_shift(&state->format); spin_unlock_irqrestore(&state->unit->reg_lock, flags); } } -#if HAVE_RECORD if (file->f_mode & FMODE_READ) { - spin_lock_irqsave(&state->unit->reg_lock, flags); - stop_adc(state); - dmabuf->ready = 0; - spin_unlock_irqrestore(&state->unit->reg_lock, flags); + ymf_stop_adc(state); + if (val == 1 || val == 2) { + spin_lock_irqsave(&state->unit->reg_lock, flags); + dmabuf = &state->rpcm.dmabuf; + dmabuf->ready = 0; + state->format.voices = val; + ymf_pcm_update_shift(&state->format); + spin_unlock_irqrestore(&state->unit->reg_lock, flags); + } } -#endif } return put_user(state->format.voices, (int *)arg); @@ -1696,15 +1695,16 @@ static int ymf_ioctl(struct inode *inode, struct file *file, * The paragraph above is a clumsy way to say "flush ioctl". * This ioctl is used by mpg123. */ - /* P3 */ /* printk("ymfpci: ioctl SNDCTL_DSP_POST\n"); */ spin_lock_irqsave(&state->unit->reg_lock, flags); - if (dmabuf->count != 0 && !state->ypcm.running) { + if (state->wpcm.dmabuf.count != 0 && !state->wpcm.running) { ymf_start_dac(state); } spin_unlock_irqrestore(&state->unit->reg_lock, flags); return 0; +#if 0 /* XXX Was dummy implementation anyways. Make sense of this. */ case SNDCTL_DSP_SUBDIVIDE: + dmabuf = &state->wpcm.dmabuf; if (dmabuf->subdivision) return -EINVAL; if (get_user(val, (int *)arg)) @@ -1713,6 +1713,7 @@ static int ymf_ioctl(struct inode *inode, struct file *file, return -EINVAL; dmabuf->subdivision = val; return 0; +#endif case SNDCTL_DSP_SETFRAGMENT: if (get_user(val, (int *)arg)) @@ -1720,6 +1721,7 @@ static int ymf_ioctl(struct inode *inode, struct file *file, /* P3: these frags are for Doom. Amasingly, it sets [2,2**11]. */ /* P3 */ // printk("ymfpci: ioctl SNDCTL_DSP_SETFRAGMENT 0x%x\n", val); + dmabuf = &state->wpcm.dmabuf; dmabuf->ossfragshift = val & 0xffff; dmabuf->ossmaxfrags = (val >> 16) & 0xffff; switch (dmabuf->ossmaxfrags) { @@ -1736,10 +1738,10 @@ static int ymf_ioctl(struct inode *inode, struct file *file, case SNDCTL_DSP_GETOSPACE: if (!(file->f_mode & FMODE_WRITE)) return -EINVAL; + dmabuf = &state->wpcm.dmabuf; if (!dmabuf->ready && (val = prog_dmabuf(state, 0)) != 0) return val; spin_lock_irqsave(&state->unit->reg_lock, flags); - /* cs_update_ptr(state); */ /* XXX Always up to date? */ abinfo.fragsize = dmabuf->fragsize; abinfo.bytes = dmabuf->dmasize - dmabuf->count; abinfo.fragstotal = dmabuf->numfrag; @@ -1747,21 +1749,19 @@ static int ymf_ioctl(struct inode *inode, struct file *file, spin_unlock_irqrestore(&state->unit->reg_lock, flags); return copy_to_user((void *)arg, &abinfo, sizeof(abinfo)) ? -EFAULT : 0; -#if HAVE_RECORD case SNDCTL_DSP_GETISPACE: if (!(file->f_mode & FMODE_READ)) return -EINVAL; + dmabuf = &state->rpcm.dmabuf; if (!dmabuf->ready && (val = prog_dmabuf(state, 1)) != 0) return val; - spin_lock_irqsave(&state->card->lock, flags); - cs_update_ptr(state); + spin_lock_irqsave(&state->unit->reg_lock, flags); abinfo.fragsize = dmabuf->fragsize; abinfo.bytes = dmabuf->count; abinfo.fragstotal = dmabuf->numfrag; abinfo.fragments = abinfo.bytes >> dmabuf->fragshift; - spin_unlock_irqrestore(&state->card->lock, flags); + spin_unlock_irqrestore(&state->unit->reg_lock, flags); return copy_to_user((void *)arg, &abinfo, sizeof(abinfo)) ? -EFAULT : 0; -#endif case SNDCTL_DSP_NONBLOCK: file->f_flags |= O_NONBLOCK; @@ -1772,9 +1772,10 @@ static int ymf_ioctl(struct inode *inode, struct file *file, (int *)arg); */ return put_user(0, (int *)arg); -#if 0 /* old */ +#if 0 /* not implememnted, yet? */ case SNDCTL_DSP_GETTRIGGER: val = 0; + dmabuf = &state->wpcm.dmabuf; if (file->f_mode & FMODE_READ && dmabuf->enable) val |= PCM_ENABLE_INPUT; if (file->f_mode & FMODE_WRITE && dmabuf->enable) @@ -1785,6 +1786,7 @@ static int ymf_ioctl(struct inode *inode, struct file *file, if (get_user(val, (int *)arg)) return -EFAULT; if (file->f_mode & FMODE_READ) { + dmabuf = &state->rpcm.dmabuf; if (val & PCM_ENABLE_INPUT) { if (!dmabuf->ready && (ret = prog_dmabuf(state, 1))) return ret; @@ -1793,6 +1795,7 @@ static int ymf_ioctl(struct inode *inode, struct file *file, stop_adc(state); } if (file->f_mode & FMODE_WRITE) { + dmabuf = &state->wpcm.dmabuf; if (val & PCM_ENABLE_OUTPUT) { if (!dmabuf->ready && (ret = prog_dmabuf(state, 0))) return ret; @@ -1801,48 +1804,48 @@ static int ymf_ioctl(struct inode *inode, struct file *file, stop_dac(state); } return 0; - #endif -#if HAVE_RECORD case SNDCTL_DSP_GETIPTR: if (!(file->f_mode & FMODE_READ)) return -EINVAL; + dmabuf = &state->rpcm.dmabuf; spin_lock_irqsave(&state->unit->reg_lock, flags); - cs_update_ptr(state); cinfo.bytes = dmabuf->total_bytes; cinfo.blocks = dmabuf->count >> dmabuf->fragshift; cinfo.ptr = dmabuf->hwptr; + /* XXX fishy - breaks invariant count=hwptr-swptr */ if (dmabuf->mapped) dmabuf->count &= dmabuf->fragsize-1; spin_unlock_irqrestore(&state->unit->reg_lock, flags); - return copy_to_user((void *)arg, &cinfo, sizeof(cinfo)); -#endif + return copy_to_user((void *)arg, &cinfo, sizeof(cinfo)) ? -EFAULT : 0; case SNDCTL_DSP_GETOPTR: if (!(file->f_mode & FMODE_WRITE)) return -EINVAL; + dmabuf = &state->wpcm.dmabuf; spin_lock_irqsave(&state->unit->reg_lock, flags); - /* cs_update_ptr(state); */ /* Always up to date */ cinfo.bytes = dmabuf->total_bytes; cinfo.blocks = dmabuf->count >> dmabuf->fragshift; cinfo.ptr = dmabuf->hwptr; + /* XXX fishy - breaks invariant count=swptr-hwptr */ if (dmabuf->mapped) dmabuf->count &= dmabuf->fragsize-1; spin_unlock_irqrestore(&state->unit->reg_lock, flags); - return copy_to_user((void *)arg, &cinfo, sizeof(cinfo)); + return copy_to_user((void *)arg, &cinfo, sizeof(cinfo)) ? -EFAULT : 0; case SNDCTL_DSP_SETDUPLEX: /* XXX TODO */ return -EINVAL; -#if 0 /* old */ +#if 0 /* XXX implement when an app found that uses it. */ case SNDCTL_DSP_GETODELAY: if (!(file->f_mode & FMODE_WRITE)) return -EINVAL; spin_lock_irqsave(&state->unit->reg_lock, flags); cs_update_ptr(state); + dmabuf = &state->wpcm.dmabuf; val = dmabuf->count; - spin_unlock_irqrestore(&state->card->lock, flags); + spin_unlock_irqrestore(&state->unit->reg_lock, flags); return put_user(val, (int *)arg); #endif @@ -1850,7 +1853,7 @@ static int ymf_ioctl(struct inode *inode, struct file *file, return put_user(state->format.rate, (int *)arg); case SOUND_PCM_READ_CHANNELS: - return put_user(state->format.voices, (int *)arg); + return put_user(state->format.voices, (int *)arg); case SOUND_PCM_READ_BITS: return put_user(AFMT_S16_LE, (int *)arg); @@ -1873,28 +1876,25 @@ static int ymf_ioctl(struct inode *inode, struct file *file, return -ENOTTY; } +/* + * open(2) + * We use upper part of the minor to distinguish between soundcards. + * Channels are opened with a clone open. + */ static int ymf_open(struct inode *inode, struct file *file) { struct list_head *list; ymfpci_t *unit; int minor; struct ymf_state *state; - int nvirt; int err; - /* - * This is how we do it currently: only one channel is used - * in every board, so that we could use several boards in one machine. - * We waste 63 out of 64 playback slots, but so what. - * OSS model is constructed for devices with single playback channel. - */ minor = MINOR(inode->i_rdev); if ((minor & 0x0F) == 3) { /* /dev/dspN */ ; } else { return -ENXIO; } - nvirt = 0; /* Such is the partitioning of minor */ for (list = ymf_devs.next; list != &ymf_devs; list = list->next) { unit = list_entry(list, ymfpci_t, ymf_devs); @@ -1905,34 +1905,36 @@ static int ymf_open(struct inode *inode, struct file *file) return -ENODEV; down(&unit->open_sem); - if (unit->states[nvirt] != NULL) { - up(&unit->open_sem); - return -EBUSY; - } - if ((err = ymf_state_alloc(unit, nvirt)) != 0) { + if ((state = ymf_state_alloc(unit)) == NULL) { up(&unit->open_sem); - return err; + return -ENOMEM; } - state = unit->states[nvirt]; + list_add_tail(&state->chain, &unit->states); file->private_data = state; /* - * XXX This ymf_playback_prepare is totally unneeded here. - * The question is if we want to allow write to fail if - * prog_dmabuf fails... Say, no memory in DMA zone? + * ymf_read and ymf_write that we borrowed from cs46xx + * allocate buffers with prog_dmabuf(). We call prog_dmabuf + * here so that in case of DMA memory exhaustion open + * fails rather than write. + * + * XXX prog_dmabuf allocates voice. Should allocate explicitly, above. */ - if ((err = ymf_playback_prepare(unit, state)) != 0) { - /* XXX This recovery is ugly as hell. */ - - ymf_pcm_free_substream(&state->ypcm); - - unit->states[state->virt] = NULL; - kfree(state); - - up(&unit->open_sem); - return err; + if (file->f_mode & FMODE_WRITE) { + if (!state->wpcm.dmabuf.ready) { + if ((err = prog_dmabuf(state, 0)) != 0) { + goto out_nodma; + } + } + } + if (file->f_mode & FMODE_READ) { + if (!state->rpcm.dmabuf.ready) { + if ((err = prog_dmabuf(state, 1)) != 0) { + goto out_nodma; + } + } } #if 0 /* test if interrupts work */ @@ -1941,10 +1943,26 @@ static int ymf_open(struct inode *inode, struct file *file) (YDSXGR_TIMERCTRL_TEN|YDSXGR_TIMERCTRL_TIEN)); #endif up(&unit->open_sem); - /* XXX Is it correct to have MOD_INC_USE_COUNT outside of sem.? */ MOD_INC_USE_COUNT; return 0; + +out_nodma: + /* + * XXX Broken custom: "goto out_xxx" in other place is + * a nestable exception, but here it is not nestable due to semaphore. + * XXX Doubtful technique of self-describing objects.... + */ + dealloc_dmabuf(&state->wpcm.dmabuf); + dealloc_dmabuf(&state->rpcm.dmabuf); + ymf_pcm_free_substream(&state->wpcm); + ymf_pcm_free_substream(&state->rpcm); + + list_del(&state->chain); + kfree(state); + + up(&unit->open_sem); + return err; } static int ymf_release(struct inode *inode, struct file *file) @@ -1956,12 +1974,6 @@ static int ymf_release(struct inode *inode, struct file *file) ymfpci_writeb(codec, YDSXGR_TIMERCTRL, 0); #endif - if (state != codec->states[state->virt]) { - printk(KERN_ERR "ymfpci%d.%d: state mismatch\n", - state->unit->dev_audio, state->virt); - return -EIO; - } - down(&codec->open_sem); /* @@ -1969,10 +1981,14 @@ static int ymf_release(struct inode *inode, struct file *file) * Deallocate when unloading the driver and we can wait. */ ymf_wait_dac(state); - dealloc_dmabuf(state); - ymf_pcm_free_substream(&state->ypcm); - - codec->states[state->virt] = NULL; + ymf_stop_adc(state); /* fortunately, it's immediate */ + dealloc_dmabuf(&state->wpcm.dmabuf); + dealloc_dmabuf(&state->rpcm.dmabuf); + ymf_pcm_free_substream(&state->wpcm); + ymf_pcm_free_substream(&state->rpcm); + + list_del(&state->chain); + file->private_data = NULL; /* Can you tell I programmed Solaris */ kfree(state); up(&codec->open_sem); @@ -2045,6 +2061,83 @@ static /*const*/ struct file_operations ymf_mixer_fops = { * initialization routines */ +#ifdef CONFIG_SOUND_YMFPCI_LEGACY + +static int ymfpci_setup_legacy(ymfpci_t *unit, struct pci_dev *pcidev) +{ + int v; + int mpuio = -1, oplio = -1; + + switch (unit->iomidi) { + case 0x330: + mpuio = 0; + break; + case 0x300: + mpuio = 1; + break; + case 0x332: + mpuio = 2; + break; + case 0x334: + mpuio = 3; + break; + default: ; + } + + switch (unit->iosynth) { + case 0x388: + oplio = 0; + break; + case 0x398: + oplio = 1; + break; + case 0x3a0: + oplio = 2; + break; + case 0x3a8: + oplio = 3; + break; + default: ; + } + + if (mpuio >= 0 || oplio >= 0) { + v = 0x003e; + pci_write_config_word(pcidev, PCIR_LEGCTRL, v); + + switch (pcidev->device) { + case PCI_DEVICE_ID_YAMAHA_724: + case PCI_DEVICE_ID_YAMAHA_740: + case PCI_DEVICE_ID_YAMAHA_724F: + case PCI_DEVICE_ID_YAMAHA_740C: + v = 0x8800; + if (mpuio >= 0) { v |= mpuio<<4; } + if (oplio >= 0) { v |= oplio; } + pci_write_config_word(pcidev, PCIR_ELEGCTRL, v); + break; + + case PCI_DEVICE_ID_YAMAHA_744: + case PCI_DEVICE_ID_YAMAHA_754: + v = 0x8800; + pci_write_config_word(pcidev, PCIR_ELEGCTRL, v); + if (oplio >= 0) { + pci_write_config_word(pcidev, PCIR_OPLADR, unit->iosynth); + } + if (mpuio >= 0) { + pci_write_config_word(pcidev, PCIR_MPUADR, unit->iomidi); + } + break; + + default: + printk(KERN_ERR "ymfpci: Unknown device ID: 0x%x\n", + pcidev->device); + return -EINVAL; + } + } + + return 0; +} +#endif /* CONFIG_SOUND_YMFPCI_LEGACY */ + static void ymfpci_aclink_reset(struct pci_dev * pci) { u8 cmd; @@ -2149,6 +2242,12 @@ static int ymfpci_memalloc(ymfpci_t *codec) ptr += 0x00ff; (long)ptr &= ~0x00ff; + /* + * Hardware requires only ptr[playback_ctrl_size] zeroed, + * but in our judgement it is a wrong kind of savings, so clear it all. + */ + memset(ptr, 0, size); + codec->bank_base_playback = ptr; codec->ctrl_playback = (u32 *)ptr; codec->ctrl_playback[0] = YDSXG_PLAYBACK_VOICES; @@ -2213,7 +2312,7 @@ static void ymfpci_memfree(ymfpci_t *codec) kfree(codec->work_ptr); } -static int ymf_ac97_init(ymfpci_t *card, int num_ac97) +static int ymf_ac97_init(ymfpci_t *unit, int num_ac97) { struct ac97_codec *codec; u16 eid; @@ -2224,7 +2323,7 @@ static int ymf_ac97_init(ymfpci_t *card, int num_ac97) /* initialize some basic codec information, other fields will be filled in ac97_probe_codec */ - codec->private_data = card; + codec->private_data = unit; codec->id = num_ac97; codec->codec_read = ymfpci_codec_read; @@ -2241,14 +2340,14 @@ static int ymf_ac97_init(ymfpci_t *card, int num_ac97) goto out_kfree; } - card->ac97_features = eid; + unit->ac97_features = eid; if ((codec->dev_mixer = register_sound_mixer(&ymf_mixer_fops, -1)) < 0) { printk(KERN_ERR "ymfpci: couldn't register mixer!\n"); goto out_kfree; } - card->ac97_codec[num_ac97] = codec; + unit->ac97_codec[num_ac97] = codec; return 0; out_kfree: @@ -2256,6 +2355,19 @@ static int ymf_ac97_init(ymfpci_t *card, int num_ac97) return -ENODEV; } +#ifdef CONFIG_SOUND_YMFPCI_LEGACY +# ifdef MODULE +static int mpu_io = 0; +static int synth_io = 0; +MODULE_PARM(mpu_io, "i"); +MODULE_PARM(synth_io, "i"); +# else +static int mpu_io = 0x330; +static int synth_io = 0x388; +# endif +static int assigned; +#endif /* CONFIG_SOUND_YMFPCI_LEGACY */ + static int __devinit ymf_probe_one(struct pci_dev *pcidev, const struct pci_device_id *ent) { u16 ctrl; @@ -2277,10 +2389,15 @@ static int __devinit ymf_probe_one(struct pci_dev *pcidev, const struct pci_devi spin_lock_init(&codec->reg_lock); spin_lock_init(&codec->voice_lock); init_MUTEX(&codec->open_sem); + INIT_LIST_HEAD(&codec->states); codec->pci = pcidev; pci_read_config_byte(pcidev, PCI_REVISION_ID, &codec->rev); codec->reg_area_virt = ioremap(pci_resource_start(pcidev, 0), 0x8000); + if (codec->reg_area_virt == NULL) { + printk(KERN_ERR "ymfpci: unable to map registers\n"); + goto out_free; + } printk(KERN_INFO "ymfpci: %s at 0x%lx IRQ %d\n", (char *)ent->driver_data, pci_resource_start(pcidev, 0), pcidev->irq); @@ -2289,6 +2406,16 @@ static int __devinit ymf_probe_one(struct pci_dev *pcidev, const struct pci_devi if (ymfpci_codec_ready(codec, 0, 1) < 0) goto out_unmap; +#ifdef CONFIG_SOUND_YMFPCI_LEGACY + if (assigned == 0) { + codec->iomidi = mpu_io; + codec->iosynth = synth_io; + if (ymfpci_setup_legacy(codec, pcidev) < 0) + goto out_unmap; + assigned = 1; + } +#endif + ymfpci_download_image(codec); udelay(100); /* seems we need some delay after downloading image.. */ @@ -2296,11 +2423,11 @@ static int __devinit ymf_probe_one(struct pci_dev *pcidev, const struct pci_devi if (ymfpci_memalloc(codec) < 0) goto out_disable_dsp; - /* ymfpci_proc_init(card, codec); */ + /* ymfpci_proc_init(unit, codec); */ if (request_irq(pcidev->irq, ymf_interrupt, SA_SHIRQ, "ymfpci", codec) != 0) { - printk(KERN_ERR "ymfpci%d: unable to request IRQ %d\n", - codec->dev_audio, pcidev->irq); + printk(KERN_ERR "ymfpci: unable to request IRQ %d\n", + pcidev->irq); goto out_memfree; } @@ -2317,6 +2444,23 @@ static int __devinit ymf_probe_one(struct pci_dev *pcidev, const struct pci_devi if ((err = ymf_ac97_init(codec, 0)) != 0) goto out_unregister_sound_dsp; +#ifdef CONFIG_SOUND_YMFPCI_LEGACY + codec->opl3_data.name = "ymfpci"; + codec->mpu_data.name = "ymfpci"; + + codec->opl3_data.io_base = codec->iosynth; + codec->opl3_data.irq = -1; + + codec->mpu_data.io_base = codec->iomidi; + codec->mpu_data.irq = -1; /* XXX Make it ours. */ + + if (codec->iomidi) { + if (!probe_uart401(&codec->mpu_data, THIS_MODULE)) { + codec->iomidi = 0; /* XXX kludge */ + } + } +#endif /* CONFIG_SOUND_YMFPCI_LEGACY */ + /* put it into driver list */ list_add_tail(&codec->ymf_devs, &ymf_devs); pci_set_drvdata(pcidev, codec); @@ -2336,6 +2480,7 @@ static int __devinit ymf_probe_one(struct pci_dev *pcidev, const struct pci_devi ymfpci_writel(codec, YDSXGR_STATUS, ~0); out_unmap: iounmap(codec->reg_area_virt); + out_free: kfree(codec); return -ENODEV; } @@ -2358,6 +2503,11 @@ static void __devexit ymf_remove_one(struct pci_dev *pcidev) ctrl = ymfpci_readw(codec, YDSXGR_GLOBALCTRL); ymfpci_writew(codec, YDSXGR_GLOBALCTRL, ctrl & ~0x0007); iounmap(codec->reg_area_virt); +#ifdef CONFIG_SOUND_YMFPCI_LEGACY + if (codec->iomidi) { + unload_uart401(&codec->mpu_data); + } +#endif /* CONFIG_SOUND_YMFPCI_LEGACY */ kfree(codec); } diff --git a/drivers/sound/ymfpci.h b/drivers/sound/ymfpci.h index 6e1a8d5f3..bd2f4c2a3 100644 --- a/drivers/sound/ymfpci.h +++ b/drivers/sound/ymfpci.h @@ -21,6 +21,7 @@ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. * */ +#include <linux/config.h> /* * Direct registers @@ -131,7 +132,12 @@ #define YDSXG_AC97READCMD 0x8000 #define YDSXG_AC97WRITECMD 0x0000 +#define PCIR_LEGCTRL 0x40 +#define PCIR_ELEGCTRL 0x42 #define PCIR_DSXGCTRL 0x48 +#define PCIR_OPLADR 0x60 +#define PCIR_SBADR 0x62 +#define PCIR_MPUADR 0x64 #define YDSXG_DSPLENGTH 0x0080 #define YDSXG_CTRLLENGTH 0x3000 @@ -185,8 +191,8 @@ typedef struct stru_ymfpci_playback_bank { } ymfpci_playback_bank_t; typedef struct stru_ymfpci_capture_bank { - u32 base; /* 32-bit address */ - u32 loop_end; /* 32-bit offset */ + u32 base; /* 32-bit address (aligned at 4) */ + u32 loop_end; /* size in BYTES (aligned at 4) */ u32 start; /* 32-bit offset */ u32 num_of_loops; /* counter */ } ymfpci_capture_bank_t; @@ -198,8 +204,7 @@ typedef struct stru_ymfpci_effect_bank { u32 temp; } ymfpci_effect_bank_t; -typedef struct stru_ymfpci_voice ymfpci_voice_t; -typedef struct ymf_pcm ymfpci_pcm_t; +typedef struct ymf_voice ymfpci_voice_t; /* * Throughout the code Yaroslav names YMF unit pointer "codec" * even though it does not correspond to any codec. Must be historic. @@ -214,52 +219,35 @@ typedef enum { YMFPCI_MIDI } ymfpci_voice_type_t; -struct stru_ymfpci_voice { - ymfpci_t *codec; +struct ymf_voice { + // ymfpci_t *codec; int number; - int use: 1, - pcm: 1, - synth: 1, - midi: 1; + char use, pcm, synth, midi; // bool ymfpci_playback_bank_t *bank; - void (*interrupt)(ymfpci_t *codec, ymfpci_voice_t *voice); - ymfpci_pcm_t *ypcm; + struct ymf_pcm *ypcm; }; -typedef enum { - PLAYBACK_VOICE, - CAPTURE_REC, - CAPTURE_AC97, - EFFECT_DRY_LEFT, - EFFECT_DRY_RIGHT, - EFFECT_EFF1, - EFFECT_EFF2, - EFFECT_EFF3 -} ymfpci_pcm_type_t; - -struct ymf_pcm { - ymfpci_t *codec; - ymfpci_pcm_type_t type; - struct ymf_state *state; - ymfpci_voice_t *voices[2]; /* playback only */ - int running; // + - int spdif; +struct ymf_capture { + // struct ymf_unit *unit; + int use; + ymfpci_capture_bank_t *bank; + struct ymf_pcm *ypcm; }; struct ymf_unit { u8 rev; /* PCI revision */ void *reg_area_virt; - void *work_ptr; // + + void *work_ptr; unsigned int bank_size_playback; unsigned int bank_size_capture; unsigned int bank_size_effect; unsigned int work_size; - void *bank_base_playback; // + - void *bank_base_capture; // + - void *bank_base_effect; // + - void *work_base; // + + void *bank_base_playback; + void *bank_base_capture; + void *bank_base_effect; + void *work_base; u32 *ctrl_playback; ymfpci_playback_bank_t *bank_playback[YDSXG_PLAYBACK_VOICES][2]; @@ -269,13 +257,20 @@ struct ymf_unit { int start_count; u32 active_bank; - ymfpci_voice_t voices[64]; + struct ymf_voice voices[64]; + struct ymf_capture capture[5]; struct ac97_codec *ac97_codec[NR_AC97]; u16 ac97_features; struct pci_dev *pci; +#ifdef CONFIG_SOUND_YMFPCI_LEGACY + /* legacy hardware resources */ + unsigned int iosynth, iomidi; + struct address_info opl3_data, mpu_data; +#endif + spinlock_t reg_lock; spinlock_t voice_lock; @@ -284,14 +279,11 @@ struct ymf_unit { struct semaphore open_sem; struct list_head ymf_devs; - struct ymf_state *states[1]; // * - /* ypcm may be the same thing as state, but not for record, effects. */ + struct list_head states; /* List of states for this unit */ + /* For the moment we do not traverse list of states so it is + * entirely useless. Will be used (PM) or killed. XXX */ }; -/* - * "Software" or virtual channel, an instance of opened /dev/dsp. - */ - struct ymf_dmabuf { /* OSS buffer management stuff */ @@ -312,7 +304,6 @@ struct ymf_dmabuf { /* redundant, but makes calculations easier */ unsigned fragsize; unsigned dmasize; /* Total rawbuf[] size */ - unsigned fragsamples; /* OSS stuff */ unsigned mapped:1; @@ -329,15 +320,40 @@ struct ymf_pcm_format { int shift; /* redundant, computed from the above */ }; -struct ymf_state { - struct ymf_unit *unit; /* backpointer */ +typedef enum { + PLAYBACK_VOICE, + CAPTURE_REC, + CAPTURE_AC97, + EFFECT_DRY_LEFT, + EFFECT_DRY_RIGHT, + EFFECT_EFF1, + EFFECT_EFF2, + EFFECT_EFF3 +} ymfpci_pcm_type_t; - /* virtual channel number */ - int virt; // * unused a.t.m. +/* This is variant record, but we hate unions. Little waste on pointers []. */ +struct ymf_pcm { + ymfpci_pcm_type_t type; + struct ymf_state *state; + + ymfpci_voice_t *voices[2]; + int capture_bank_number; + + struct ymf_dmabuf dmabuf; + int running; + int spdif; +}; - struct ymf_pcm ypcm; // * - struct ymf_dmabuf dmabuf; // * - struct ymf_pcm_format format; // * +/* + * "Software" or virtual channel, an instance of opened /dev/dsp. + * It may have two physical channels (pcms) for duplex operations. + */ + +struct ymf_state { + struct list_head chain; + struct ymf_unit *unit; /* backpointer */ + struct ymf_pcm rpcm, wpcm; + struct ymf_pcm_format format; }; #endif /* __YMFPCI_H */ diff --git a/drivers/usb/rio500.c b/drivers/usb/rio500.c index 6d7b9043d..29b926056 100644 --- a/drivers/usb/rio500.c +++ b/drivers/usb/rio500.c @@ -247,6 +247,7 @@ ioctl_rio(struct inode *inode, struct file *file, unsigned int cmd, return 0; err_out: + up(&(rio->lock)); return retval; } @@ -389,7 +390,7 @@ read_rio(struct file *file, char *buffer, size_t count, loff_t * ppos) result, partial, this_read); return -EIO; } else { - unlock_kernel(); + up(&(rio->lock)); return (0); } diff --git a/drivers/usb/serial/Config.in b/drivers/usb/serial/Config.in index 6d949bc0b..0032f6e15 100644 --- a/drivers/usb/serial/Config.in +++ b/drivers/usb/serial/Config.in @@ -22,6 +22,7 @@ if [ "$CONFIG_USB_SERIAL" != "n" ]; then bool ' USB Keyspan USA-19 Firmware' CONFIG_USB_SERIAL_KEYSPAN_USA19 bool ' USB Keyspan USA-18X Firmware' CONFIG_USB_SERIAL_KEYSPAN_USA18X bool ' USB Keyspan USA-19W Firmware' CONFIG_USB_SERIAL_KEYSPAN_USA19W + bool ' USB Keyspan USA-49W Firmware' CONFIG_USB_SERIAL_KEYSPAN_USA49W fi dep_tristate ' USB MCT Single Port Serial Driver (EXPERIMENTAL)' CONFIG_USB_SERIAL_MCT_U232 $CONFIG_USB_SERIAL $CONFIG_EXPERIMENTAL dep_tristate ' USB ZyXEL omni.net LCD Plus Driver (EXPERIMENTAL)' CONFIG_USB_SERIAL_OMNINET $CONFIG_USB_SERIAL $CONFIG_EXPERIMENTAL diff --git a/drivers/usb/serial/usbserial.c b/drivers/usb/serial/usbserial.c index 1a07a03d5..3a3451e49 100644 --- a/drivers/usb/serial/usbserial.c +++ b/drivers/usb/serial/usbserial.c @@ -14,8 +14,14 @@ * based on a driver by Brad Keryan) * * See Documentation/usb/usb-serial.txt for more information on using this driver - * - * (12/29/2000) gkh + * + * (12/12/2000) gkh + * Removed MOD_INC and MOD_DEC from poll and disconnect functions, and + * moved them to the serial_open and serial_close functions. + * Also fixed bug with there not being a MOD_DEC for the generic driver + * (thanks to Gary Brubaker for finding this.) + * + * (11/29/2000) gkh * Small NULL pointer initialization cleanup which saves a bit of disk image * * (11/01/2000) Adam J. Richter @@ -471,6 +477,8 @@ static int serial_open (struct tty_struct *tty, struct file * filp) return -ENODEV; } + MOD_INC_USE_COUNT; + /* set up our port structure making the tty driver remember our port object, and us it */ portNumber = MINOR(tty->device) - serial->minor; port = &serial->port[portNumber]; @@ -508,6 +516,8 @@ static void serial_close(struct tty_struct *tty, struct file * filp) } else { generic_close(port, filp); } + + MOD_DEC_USE_COUNT; } @@ -721,12 +731,13 @@ static int generic_open (struct usb_serial_port *port, struct file *filp) if (port_paranoia_check (port, __FUNCTION__)) return -ENODEV; + MOD_INC_USE_COUNT; + dbg(__FUNCTION__ " - port %d", port->number); spin_lock_irqsave (&port->port_lock, flags); ++port->open_count; - MOD_INC_USE_COUNT; if (!port->active) { port->active = 1; @@ -776,6 +787,7 @@ static void generic_close (struct usb_serial_port *port, struct file * filp) } spin_unlock_irqrestore (&port->port_lock, flags); + MOD_DEC_USE_COUNT; } @@ -1069,7 +1081,6 @@ static void * usb_serial_probe(struct usb_device *dev, unsigned int ifnum, } /* found all that we need */ - MOD_INC_USE_COUNT; info("%s converter detected", type->name); #ifdef CONFIG_USB_SERIAL_GENERIC @@ -1077,7 +1088,6 @@ static void * usb_serial_probe(struct usb_device *dev, unsigned int ifnum, num_ports = num_bulk_out; if (num_ports == 0) { err("Generic device with no bulk out, not allowed."); - MOD_DEC_USE_COUNT; return NULL; } } else @@ -1087,7 +1097,6 @@ static void * usb_serial_probe(struct usb_device *dev, unsigned int ifnum, serial = get_free_serial (num_ports, &minor); if (serial == NULL) { err("No more free serial devices"); - MOD_DEC_USE_COUNT; return NULL; } @@ -1233,7 +1242,6 @@ probe_error: /* free up any memory that we allocated */ kfree (serial); - MOD_DEC_USE_COUNT; return NULL; } @@ -1300,7 +1308,6 @@ static void usb_serial_disconnect(struct usb_device *dev, void *ptr) info("device disconnected"); } - MOD_DEC_USE_COUNT; } diff --git a/drivers/usb/serial/visor.c b/drivers/usb/serial/visor.c index e363916b9..d8a60a2a7 100644 --- a/drivers/usb/serial/visor.c +++ b/drivers/usb/serial/visor.c @@ -10,6 +10,10 @@ * (at your option) any later version. * * See Documentation/usb/usb-serial.txt for more information on using this driver + * + * (12/12/2000) gkh + * Moved MOD_DEC to end of visor_close to be nicer, as the final write + * message can sleep. * * (11/12/2000) gkh * Fixed bug with data being dropped on the floor by forcing tty->low_latency @@ -214,7 +218,6 @@ static void visor_close (struct usb_serial_port *port, struct file * filp) spin_lock_irqsave (&port->port_lock, flags); --port->open_count; - MOD_DEC_USE_COUNT; if (port->open_count <= 0) { transfer_buffer = kmalloc (0x12, GFP_KERNEL); @@ -237,6 +240,8 @@ static void visor_close (struct usb_serial_port *port, struct file * filp) /* Uncomment the following line if you want to see some statistics in your syslog */ /* info ("Bytes In = %d Bytes Out = %d", bytes_in, bytes_out); */ + + MOD_DEC_USE_COUNT; } diff --git a/drivers/usb/storage/debug.h b/drivers/usb/storage/debug.h index 5123f85f8..c5477dcaf 100644 --- a/drivers/usb/storage/debug.h +++ b/drivers/usb/storage/debug.h @@ -1,7 +1,7 @@ /* Driver for USB Mass Storage compliant devices * Debugging Functions Header File * - * $Id: debug.h,v 1.5 2000/09/04 02:12:47 groovyjava Exp $ + * $Id: debug.h,v 1.6 2001/01/12 23:51:04 mdharm Exp $ * * Current development and maintenance by: * (c) 1999, 2000 Matthew Dharm (mdharm-usb@one-eyed-alien.net) diff --git a/drivers/usb/storage/scsiglue.c b/drivers/usb/storage/scsiglue.c index 241f7d5f6..5c50b578c 100644 --- a/drivers/usb/storage/scsiglue.c +++ b/drivers/usb/storage/scsiglue.c @@ -248,7 +248,7 @@ static int bus_reset( Scsi_Cmnd *srb ) for (i = 0; i < us->pusb_dev->actconfig->bNumInterfaces; i++) { struct usb_interface *intf = &us->pusb_dev->actconfig->interface[i]; - const struct usb_device_id *id; + struct usb_device_id *id; /* if this is an unclaimed interface, skip it */ if (!intf->driver) { diff --git a/drivers/usb/storage/unusual_devs.h b/drivers/usb/storage/unusual_devs.h new file mode 100644 index 000000000..d29307433 --- /dev/null +++ b/drivers/usb/storage/unusual_devs.h @@ -0,0 +1,232 @@ +/* Driver for USB Mass Storage compliant devices + * Ununsual Devices File + * + * $Id: unusual_devs.h,v 1.1 2000/12/05 05:38:31 mdharm Exp $ + * + * Current development and maintenance by: + * (c) 2000 Matthew Dharm (mdharm-usb@one-eyed-alien.net) + * + * Initial work by: + * (c) 2000 Adam J. Richter (adam@yggdrasil.com), Yggdrasil Computing, Inc. + * + * Please see http://www.one-eyed-alien.net/~mdharm/linux-usb for more + * information about this driver. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2, or (at your option) any + * later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public License along + * with this program; if not, write to the Free Software Foundation, Inc., + * 675 Mass Ave, Cambridge, MA 02139, USA. + */ + +/* IMPORTANT NOTE: This file must be included in another file which does + * the following thing for it to work: + * The macro UNUSUAL_DEV() must be defined before this file is included + */ +#include <linux/config.h> + +/* If you edit this file, please try to keep it sorted first by VendorID, + * then by ProductID. + */ + +UNUSUAL_DEV( 0x03ee, 0x0000, 0x0000, 0x0245, + "Mitsumi", + "CD-R/RW Drive", + US_SC_8020, US_PR_CBI, NULL, 0), + +UNUSUAL_DEV( 0x03f0, 0x0107, 0x0200, 0x0200, + "HP", + "CD-Writer+", + US_SC_8070, US_PR_CB, NULL, 0), + +#ifdef CONFIG_USB_STORAGE_HP8200e +UNUSUAL_DEV( 0x03f0, 0x0207, 0x0001, 0x0001, + "HP", + "CD-Writer+ 8200e", + US_SC_8070, US_PR_SCM_ATAPI, init_8200e, 0), +#endif + +UNUSUAL_DEV( 0x04e6, 0x0001, 0x0200, 0x0200, + "Matshita", + "LS-120", + US_SC_8020, US_PR_CB, NULL, 0), + +UNUSUAL_DEV( 0x04e6, 0x0002, 0x0100, 0x0100, + "Shuttle", + "eUSCSI Bridge", + US_SC_SCSI, US_PR_BULK, usb_stor_euscsi_init, + US_FL_SCM_MULT_TARG ), + +#ifdef CONFIG_USB_STORAGE_SDDR09 +UNUSUAL_DEV( 0x04e6, 0x0003, 0x0000, 0x9999, + "Sandisk", + "ImageMate SDDR09", + US_SC_SCSI, US_PR_EUSB_SDDR09, NULL, + US_FL_SINGLE_LUN | US_FL_START_STOP ), +#endif + +#ifdef CONFIG_USB_STORAGE_DPCM +UNUSUAL_DEV( 0x0436, 0x0005, 0x0100, 0x0100, + "Microtech", + "CameraMate (DPCM_USB)", + US_SC_SCSI, US_PR_DPCM_USB, NULL, + US_FL_START_STOP ), +#endif + +UNUSUAL_DEV( 0x04e6, 0x0006, 0x0100, 0x0200, + "Shuttle", + "eUSB MMC Adapter", + US_SC_SCSI, US_PR_CB, NULL, + US_FL_SINGLE_LUN), + +UNUSUAL_DEV( 0x04e6, 0x0007, 0x0100, 0x0200, + "Sony", + "Hifd", + US_SC_SCSI, US_PR_CB, NULL, + US_FL_SINGLE_LUN), + +UNUSUAL_DEV( 0x04e6, 0x0009, 0x0200, 0x0200, + "Shuttle", + "eUSB ATA/ATAPI Adapter", + US_SC_8020, US_PR_CB, NULL, 0), + +UNUSUAL_DEV( 0x04e6, 0x000a, 0x0200, 0x0200, + "Shuttle", + "eUSB CompactFlash Adapter", + US_SC_8020, US_PR_CB, NULL, 0), + +UNUSUAL_DEV( 0x04e6, 0x000B, 0x0100, 0x0100, + "Shuttle", + "eUSCSI Bridge", + US_SC_SCSI, US_PR_BULK, usb_stor_euscsi_init, + US_FL_SCM_MULT_TARG ), + +UNUSUAL_DEV( 0x04e6, 0x000C, 0x0100, 0x0100, + "Shuttle", + "eUSCSI Bridge", + US_SC_SCSI, US_PR_BULK, usb_stor_euscsi_init, + US_FL_SCM_MULT_TARG ), + +UNUSUAL_DEV( 0x04e6, 0x0101, 0x0200, 0x0200, + "Shuttle", + "CD-RW Device", + US_SC_8020, US_PR_CB, NULL, 0), + +UNUSUAL_DEV( 0x054c, 0x0010, 0x0106, 0x0210, + "Sony", + "DSC-S30/S70/505V/F505", + US_SC_SCSI, US_PR_CB, NULL, + US_FL_SINGLE_LUN | US_FL_START_STOP | US_FL_MODE_XLATE ), + +UNUSUAL_DEV( 0x054c, 0x002d, 0x0100, 0x0100, + "Sony", + "Memorystick MSAC-US1", + US_SC_UFI, US_PR_CB, NULL, + US_FL_SINGLE_LUN | US_FL_START_STOP ), + +UNUSUAL_DEV( 0x057b, 0x0000, 0x0000, 0x0299, + "Y-E Data", + "Flashbuster-U", + US_SC_UFI, US_PR_CB, NULL, + US_FL_SINGLE_LUN), + +UNUSUAL_DEV( 0x057b, 0x0000, 0x0300, 0x9999, + "Y-E Data", + "Flashbuster-U", + US_SC_UFI, US_PR_CBI, NULL, + US_FL_SINGLE_LUN), + +UNUSUAL_DEV( 0x059f, 0xa601, 0x0200, 0x0200, + "LaCie", + "USB Hard Disk", + US_SC_RBC, US_PR_CB, NULL, 0 ), + +UNUSUAL_DEV( 0x05ab, 0x0031, 0x0100, 0x0100, + "In-System", + "USB/IDE Bridge (ATAPI ONLY!)", + US_SC_8070, US_PR_BULK, NULL, 0 ), + +UNUSUAL_DEV( 0x0644, 0x0000, 0x0100, 0x0100, + "TEAC", + "Floppy Drive", + US_SC_UFI, US_PR_CB, NULL, 0 ), + +#ifdef CONFIG_USB_STORAGE_SDDR09 +UNUSUAL_DEV( 0x066b, 0x0105, 0x0100, 0x0100, + "Olympus", + "Camedia MAUSB-2", + US_SC_SCSI, US_PR_EUSB_SDDR09, NULL, + US_FL_SINGLE_LUN | US_FL_START_STOP ), +#endif + +UNUSUAL_DEV( 0x0693, 0x0002, 0x0100, 0x0100, + "Hagiwara", + "FlashGate SmartMedia", + US_SC_SCSI, US_PR_BULK, NULL, 0 ), + +UNUSUAL_DEV( 0x0693, 0x0005, 0x0100, 0x0100, + "Hagiwara", + "Flashgate", + US_SC_SCSI, US_PR_BULK, NULL, 0 ), + +UNUSUAL_DEV( 0x0781, 0x0001, 0x0200, 0x0200, + "Sandisk", + "ImageMate SDDR-05a", + US_SC_SCSI, US_PR_CB, NULL, + US_FL_SINGLE_LUN | US_FL_START_STOP), + +UNUSUAL_DEV( 0x0781, 0x0100, 0x0100, 0x0100, + "Sandisk", + "ImageMate SDDR-12", + US_SC_SCSI, US_PR_CB, NULL, + US_FL_SINGLE_LUN ), + +#ifdef CONFIG_USB_STORAGE_SDDR09 +UNUSUAL_DEV( 0x0781, 0x0200, 0x0100, 0x0100, + "Sandisk", + "ImageMate SDDR-09", + US_SC_SCSI, US_PR_EUSB_SDDR09, NULL, + US_FL_SINGLE_LUN | US_FL_START_STOP ), +#endif + +UNUSUAL_DEV( 0x0781, 0x0002, 0x0009, 0x0009, + "Sandisk", + "ImageMate SDDR-31", + US_SC_SCSI, US_PR_BULK, NULL, + US_FL_IGNORE_SER), + +UNUSUAL_DEV( 0x07af, 0x0004, 0x0100, 0x0100, + "Microtech", + "USB-SCSI-DB25", + US_SC_SCSI, US_PR_BULK, usb_stor_euscsi_init, + US_FL_SCM_MULT_TARG ), + +#ifdef CONFIG_USB_STORAGE_FREECOM +UNUSUAL_DEV( 0x07ab, 0xfc01, 0x0000, 0x9999, + "Freecom", + "USB-IDE", + US_SC_QIC, US_PR_FREECOM, freecom_init, 0), +#endif + +UNUSUAL_DEV( 0x07af, 0x0005, 0x0100, 0x0100, + "Microtech", + "USB-SCSI-HD50", + US_SC_SCSI, US_PR_BULK, usb_stor_euscsi_init, + US_FL_SCM_MULT_TARG ), + +#ifdef CONFIG_USB_STORAGE_DPCM +UNUSUAL_DEV( 0x07af, 0x0006, 0x0100, 0x0100, + "Microtech", + "CameraMate (DPCM_USB)", + US_SC_SCSI, US_PR_DPCM_USB, NULL, + US_FL_START_STOP ), +#endif + diff --git a/drivers/usb/storage/usb.c b/drivers/usb/storage/usb.c index 6c254c701..1e0f40f50 100644 --- a/drivers/usb/storage/usb.c +++ b/drivers/usb/storage/usb.c @@ -1,6 +1,6 @@ /* Driver for USB Mass Storage compliant devices * - * $Id: usb.c,v 1.57 2000/11/21 02:56:41 mdharm Exp $ + * $Id: usb.c,v 1.61 2001/01/13 00:10:59 mdharm Exp $ * * Current development and maintenance by: * (c) 1999, 2000 Matthew Dharm (mdharm-usb@one-eyed-alien.net) @@ -11,6 +11,9 @@ * Initial work by: * (c) 1999 Michael Gee (michael@linuxspecific.com) * + * usb_device_id support by Adam J. Richter (adam@yggdrasil.com): + * (c) 2000 Yggdrasil Computing, Inc. + * * This driver is based on the 'USB Mass Storage Class' document. This * describes in detail the protocol used to communicate with such * devices. Clearly, the designers had SCSI and ATAPI commands in @@ -95,11 +98,133 @@ struct semaphore us_list_semaphore; static void * storage_probe(struct usb_device *dev, unsigned int ifnum, const struct usb_device_id *id); + static void storage_disconnect(struct usb_device *dev, void *ptr); + +/* The entries in this table, except for final ones here + * (USB_MASS_STORAGE_CLASS and the empty entry), correspond, + * line for line with the entries of us_unsuaul_dev_list[]. + * For now, we duplicate idVendor and idProduct in us_unsual_dev_list, + * just to avoid alignment bugs. + */ + +#define UNUSUAL_DEV(id_vendor, id_product, bcdDeviceMin, bcdDeviceMax, \ + vendorName, productName,useProtocol, useTransport, \ + initFunction, flags) \ +{ USB_DEVICE_VER(id_vendor, id_product, bcdDeviceMin,bcdDeviceMax) } + +static struct usb_device_id storage_usb_ids [] = { + +# include "unusual_devs.h" +#undef UNUSUAL_DEV + /* Control/Bulk transport for all SubClass values */ + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_RBC, US_PR_CB) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_8020, US_PR_CB) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_QIC, US_PR_CB) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_UFI, US_PR_CB) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_8070, US_PR_CB) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_SCSI, US_PR_CB) }, + + /* Control/Bulk/Interrupt transport for all SubClass values */ + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_RBC, US_PR_CBI) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_8020, US_PR_CBI) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_QIC, US_PR_CBI) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_UFI, US_PR_CBI) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_8070, US_PR_CBI) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_SCSI, US_PR_CBI) }, + + /* Bulk-only transport for all SubClass values */ + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_RBC, US_PR_BULK) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_8020, US_PR_BULK) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_QIC, US_PR_BULK) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_UFI, US_PR_BULK) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_8070, US_PR_BULK) }, + { USB_INTERFACE_INFO(USB_CLASS_MASS_STORAGE, US_SC_SCSI, US_PR_BULK) }, + + /* Terminating entry */ + { } +}; + +MODULE_DEVICE_TABLE (usb, storage_usb_ids); + +/* This is the list of devices we recognize, along with their flag data */ + +/* The vendor name should be kept at eight characters or less, and + * the product name should be kept at 16 characters or less. If a device + * has the US_FL_DUMMY_INQUIRY flag, then the vendor and product names + * normally generated by a device thorugh the INQUIRY response will be + * taken from this list, and this is the reason for the above size + * restriction. However, if the flag is not present, then you + * are free to use as many characters as you like. + */ + +#undef UNUSUAL_DEV +#define UNUSUAL_DEV(idVendor, idProduct, bcdDeviceMin, bcdDeviceMax, \ + vendor_name, product_name, use_protocol, use_transport, \ + init_function, Flags) \ +{ \ + vendorName: vendor_name, \ + productName: product_name, \ + useProtocol: use_protocol, \ + useTransport: use_transport, \ + initFunction : init_function, \ + flags: Flags, \ +} + +static struct us_unusual_dev us_unusual_dev_list[] = { +# include "unusual_devs.h" +# undef UNUSUAL_DEV + /* Control/Bulk transport for all SubClass values */ + { useProtocol: US_SC_RBC, + useTransport: US_PR_CB}, + { useProtocol: US_SC_8020, + useTransport: US_PR_CB}, + { useProtocol: US_SC_QIC, + useTransport: US_PR_CB}, + { useProtocol: US_SC_UFI, + useTransport: US_PR_CB}, + { useProtocol: US_SC_8070, + useTransport: US_PR_CB}, + { useProtocol: US_SC_SCSI, + useTransport: US_PR_CB}, + + /* Control/Bulk/Interrupt transport for all SubClass values */ + { useProtocol: US_SC_RBC, + useTransport: US_PR_CBI}, + { useProtocol: US_SC_8020, + useTransport: US_PR_CBI}, + { useProtocol: US_SC_QIC, + useTransport: US_PR_CBI}, + { useProtocol: US_SC_UFI, + useTransport: US_PR_CBI}, + { useProtocol: US_SC_8070, + useTransport: US_PR_CBI}, + { useProtocol: US_SC_SCSI, + useTransport: US_PR_CBI}, + + /* Bulk-only transport for all SubClass values */ + { useProtocol: US_SC_RBC, + useTransport: US_PR_BULK}, + { useProtocol: US_SC_8020, + useTransport: US_PR_BULK}, + { useProtocol: US_SC_QIC, + useTransport: US_PR_BULK}, + { useProtocol: US_SC_UFI, + useTransport: US_PR_BULK}, + { useProtocol: US_SC_8070, + useTransport: US_PR_BULK}, + { useProtocol: US_SC_SCSI, + useTransport: US_PR_BULK}, + + /* Terminating entry */ + { 0 } +}; + struct usb_driver usb_storage_driver = { name: "usb-storage", probe: storage_probe, disconnect: storage_disconnect, + id_table: storage_usb_ids, }; /* @@ -325,246 +450,6 @@ static int usb_stor_control_thread(void * __us) return 0; } -/* This is the list of devices we recognize, along with their flag data */ - -/* The vendor name should be kept at eight characters or less, and - * the product name should be kept at 16 characters or less. If a device - * has the US_FL_DUMMY_INQUIRY flag, then the vendor and product names - * normally generated by a device thorugh the INQUIRY response will be - * taken from this list, and this is the reason for the above size - * restriction. However, if the flag is not present, then you - * are free to use as many characters as you like. - */ -static struct us_unusual_dev us_unusual_dev_list[] = { - - { 0x03ee, 0x0000, 0x0000, 0x0245, - "Mitsumi", - "CD-R/RW Drive", - US_SC_8020, US_PR_CBI, NULL, 0}, - - { 0x03f0, 0x0107, 0x0200, 0x0200, - "HP", - "CD-Writer+", - US_SC_8070, US_PR_CB, NULL, 0}, - -#ifdef CONFIG_USB_STORAGE_HP8200e - { 0x03f0, 0x0207, 0x0001, 0x0001, - "HP", - "CD-Writer+ 8200e", - US_SC_8070, US_PR_SCM_ATAPI, init_8200e, 0}, -#endif - - { 0x04e6, 0x0001, 0x0200, 0x0200, - "Matshita", - "LS-120", - US_SC_8020, US_PR_CB, NULL, 0}, - - { 0x04e6, 0x0002, 0x0100, 0x0100, - "Shuttle", - "eUSCSI Bridge", - US_SC_SCSI, US_PR_BULK, usb_stor_euscsi_init, - US_FL_SCM_MULT_TARG }, - -#ifdef CONFIG_USB_STORAGE_SDDR09 - { 0x04e6, 0x0003, 0x0000, 0x9999, - "Sandisk", - "ImageMate SDDR09", - US_SC_SCSI, US_PR_EUSB_SDDR09, NULL, - US_FL_SINGLE_LUN | US_FL_START_STOP }, -#endif - -#ifdef CONFIG_USB_STORAGE_DPCM - { 0x0436, 0x0005, 0x0100, 0x0100, - "Microtech", - "CameraMate (DPCM_USB)", - US_SC_SCSI, US_PR_DPCM_USB, NULL, - US_FL_START_STOP }, -#endif - - { 0x04e6, 0x0006, 0x0100, 0x0200, - "Shuttle", - "eUSB MMC Adapter", - US_SC_SCSI, US_PR_CB, NULL, - US_FL_SINGLE_LUN}, - - { 0x04e6, 0x0007, 0x0100, 0x0200, - "Sony", - "Hifd", - US_SC_SCSI, US_PR_CB, NULL, - US_FL_SINGLE_LUN}, - - { 0x04e6, 0x0009, 0x0200, 0x0200, - "Shuttle", - "eUSB ATA/ATAPI Adapter", - US_SC_8020, US_PR_CB, NULL, 0}, - - { 0x04e6, 0x000a, 0x0200, 0x0200, - "Shuttle", - "eUSB CompactFlash Adapter", - US_SC_8020, US_PR_CB, NULL, 0}, - - { 0x04e6, 0x000B, 0x0100, 0x0100, - "Shuttle", - "eUSCSI Bridge", - US_SC_SCSI, US_PR_BULK, usb_stor_euscsi_init, - US_FL_SCM_MULT_TARG }, - - { 0x04e6, 0x000C, 0x0100, 0x0100, - "Shuttle", - "eUSCSI Bridge", - US_SC_SCSI, US_PR_BULK, usb_stor_euscsi_init, - US_FL_SCM_MULT_TARG }, - - { 0x04e6, 0x0101, 0x0200, 0x0200, - "Shuttle", - "CD-RW Device", - US_SC_8020, US_PR_CB, NULL, 0}, - - { 0x054c, 0x0010, 0x0106, 0x0210, - "Sony", - "DSC-S30/S70/505V/F505", - US_SC_SCSI, US_PR_CB, NULL, - US_FL_SINGLE_LUN | US_FL_START_STOP | US_FL_MODE_XLATE }, - - { 0x054c, 0x002d, 0x0100, 0x0100, - "Sony", - "Memorystick MSAC-US1", - US_SC_UFI, US_PR_CB, NULL, - US_FL_SINGLE_LUN | US_FL_START_STOP }, - - { 0x057b, 0x0000, 0x0000, 0x0299, - "Y-E Data", - "Flashbuster-U", - US_SC_UFI, US_PR_CB, NULL, - US_FL_SINGLE_LUN}, - - { 0x057b, 0x0000, 0x0300, 0x9999, - "Y-E Data", - "Flashbuster-U", - US_SC_UFI, US_PR_CBI, NULL, - US_FL_SINGLE_LUN}, - - { 0x059f, 0xa601, 0x0200, 0x0200, - "LaCie", - "USB Hard Disk", - US_SC_RBC, US_PR_CB, NULL, 0 }, - - { 0x05ab, 0x0031, 0x0100, 0x0100, - "In-System", - "USB/IDE Bridge (ATAPI ONLY!)", - US_SC_8070, US_PR_BULK, NULL, 0 }, - - { 0x0644, 0x0000, 0x0100, 0x0100, - "TEAC", - "Floppy Drive", - US_SC_UFI, US_PR_CB, NULL, 0 }, - -#ifdef CONFIG_USB_STORAGE_SDDR09 - { 0x066b, 0x0105, 0x0100, 0x0100, - "Olympus", - "Camedia MAUSB-2", - US_SC_SCSI, US_PR_EUSB_SDDR09, NULL, - US_FL_SINGLE_LUN | US_FL_START_STOP }, -#endif - - { 0x0693, 0x0002, 0x0100, 0x0100, - "Hagiwara", - "FlashGate SmartMedia", - US_SC_SCSI, US_PR_BULK, NULL, 0 }, - - { 0x0693, 0x0005, 0x0100, 0x0100, - "Hagiwara", - "Flashgate", - US_SC_SCSI, US_PR_BULK, NULL, 0 }, - - { 0x0781, 0x0001, 0x0200, 0x0200, - "Sandisk", - "ImageMate SDDR-05a", - US_SC_SCSI, US_PR_CB, NULL, - US_FL_SINGLE_LUN | US_FL_START_STOP}, - - { 0x0781, 0x0100, 0x0100, 0x0100, - "Sandisk", - "ImageMate SDDR-12", - US_SC_SCSI, US_PR_CB, NULL, - US_FL_SINGLE_LUN }, - -#ifdef CONFIG_USB_STORAGE_SDDR09 - { 0x0781, 0x0200, 0x0100, 0x0208, - "Sandisk", - "ImageMate SDDR-09", - US_SC_SCSI, US_PR_EUSB_SDDR09, NULL, - US_FL_SINGLE_LUN | US_FL_START_STOP }, -#endif - - { 0x0781, 0x0002, 0x0009, 0x0009, - "Sandisk", - "ImageMate SDDR-31", - US_SC_SCSI, US_PR_BULK, NULL, - US_FL_IGNORE_SER}, - - { 0x07af, 0x0004, 0x0100, 0x0100, - "Microtech", - "USB-SCSI-DB25", - US_SC_SCSI, US_PR_BULK, usb_stor_euscsi_init, - US_FL_SCM_MULT_TARG }, - -#ifdef CONFIG_USB_STORAGE_FREECOM - { 0x07ab, 0xfc01, 0x0000, 0x9999, - "Freecom", - "USB-IDE", - US_SC_QIC, US_PR_FREECOM, freecom_init, 0}, -#endif - - { 0x07af, 0x0005, 0x0100, 0x0100, - "Microtech", - "USB-SCSI-HD50", - US_SC_SCSI, US_PR_BULK, usb_stor_euscsi_init, - US_FL_SCM_MULT_TARG }, - -#ifdef CONFIG_USB_STORAGE_DPCM - { 0x07af, 0x0006, 0x0100, 0x0100, - "Microtech", - "CameraMate (DPCM_USB)", - US_SC_SCSI, US_PR_DPCM_USB, NULL, - US_FL_START_STOP }, -#endif - { 0 } -}; - -/* Search our ususual device list, based on vendor/product combinations - * to see if we can support this device. Returns a pointer to a structure - * defining how we should support this device, or NULL if it's not in the - * list - */ -static struct us_unusual_dev* us_find_dev(u16 idVendor, u16 idProduct, - u16 bcdDevice) -{ - struct us_unusual_dev* ptr; - - US_DEBUGP("Searching unusual device list for (0x%x, 0x%x, 0x%x)...\n", - idVendor, idProduct, bcdDevice); - - ptr = us_unusual_dev_list; - while ((ptr->idVendor != 0x0000) && - !((ptr->idVendor == idVendor) && - (ptr->idProduct == idProduct) && - (ptr->bcdDeviceMin <= bcdDevice) && - (ptr->bcdDeviceMax >= bcdDevice))) - ptr++; - - /* if the search ended because we hit the end record, we failed */ - if (ptr->idVendor == 0x0000) { - US_DEBUGP("-- did not find a matching device\n"); - return NULL; - } - - /* otherwise, we found one! */ - US_DEBUGP("-- found matching device: %s %s\n", ptr->vendorName, - ptr->productName); - return ptr; -} - /* Set up the IRQ pipe and handler * Note that this function assumes that all the data in the us_data * strucuture is current. This includes the ep_int field, which gives us @@ -620,6 +505,7 @@ static void * storage_probe(struct usb_device *dev, unsigned int ifnum, const struct usb_device_id *id) { int i; + const int id_index = id - storage_usb_ids; char mf[USB_STOR_STRING_LEN]; /* manufacturer */ char prod[USB_STOR_STRING_LEN]; /* product */ char serial[USB_STOR_STRING_LEN]; /* serial number */ @@ -640,47 +526,48 @@ static void * storage_probe(struct usb_device *dev, unsigned int ifnum, u8 subclass = 0; u8 protocol = 0; - /* the altsettting 0 on the interface we're probing */ - struct usb_interface_descriptor *altsetting = - &(dev->actconfig->interface[ifnum].altsetting[0]); + /* the altsettting on the interface we're probing that matched our + * usb_match_id table + */ + struct usb_interface *intf = dev->actconfig->interface; + struct usb_interface_descriptor *altsetting = + intf[ifnum].altsetting + intf[ifnum].act_altsetting; + US_DEBUGP("act_altsettting is %d\n", intf[ifnum].act_altsetting); /* clear the temporary strings */ memset(mf, 0, sizeof(mf)); memset(prod, 0, sizeof(prod)); memset(serial, 0, sizeof(serial)); - /* search for this device in our unusual device list */ - unusual_dev = us_find_dev(dev->descriptor.idVendor, - dev->descriptor.idProduct, - dev->descriptor.bcdDevice); - /* * Can we support this device, either because we know about it * from our unusual device list, or because it advertises that it's * compliant to the specification? + * + * id_index is calculated in the declaration to be the index number + * of the match from the usb_device_id table, so we can find the + * corresponding entry in the private table. */ - if (!unusual_dev && - !(dev->descriptor.bDeviceClass == 0 && - altsetting->bInterfaceClass == USB_CLASS_MASS_STORAGE && - altsetting->bInterfaceSubClass >= US_SC_MIN && - altsetting->bInterfaceSubClass <= US_SC_MAX)) { - /* if it's not a mass storage, we go no further */ + US_DEBUGP("id_index calculated to be: %d\n", id_index); + US_DEBUGP("Array length appears to be: %d\n", sizeof(us_unusual_dev_list) / sizeof(us_unusual_dev_list[0])); + if (id_index < + sizeof(us_unusual_dev_list) / sizeof(us_unusual_dev_list[0])) { + unusual_dev = &us_unusual_dev_list[id_index]; + if (unusual_dev->vendorName) + US_DEBUGP("Vendor: %s\n", unusual_dev->vendorName); + if (unusual_dev->productName) + US_DEBUGP("Product: %s\n", unusual_dev->productName); + } else + /* no, we can't support it */ return NULL; - } /* At this point, we know we've got a live one */ US_DEBUGP("USB Mass Storage device detected\n"); /* Determine subclass and protocol, or copy from the interface */ - if (unusual_dev) { - subclass = unusual_dev->useProtocol; - protocol = unusual_dev->useTransport; - flags = unusual_dev->flags; - } else { - subclass = altsetting->bInterfaceSubClass; - protocol = altsetting->bInterfaceProtocol; - flags = 0; - } + subclass = unusual_dev->useProtocol; + protocol = unusual_dev->useTransport; + flags = unusual_dev->flags; /* * Find the endpoints we need @@ -728,7 +615,7 @@ static void * storage_probe(struct usb_device *dev, unsigned int ifnum, /* Do some basic sanity checks, and bail if we find a problem */ if (!ep_in || !ep_out || (protocol == US_PR_CBI && !ep_int)) { - US_DEBUGP("Sanity check failed. Rejecting device.\n"); + US_DEBUGP("Endpoint sanity check failed! Rejecting dev.\n"); return NULL; } @@ -861,7 +748,7 @@ static void * storage_probe(struct usb_device *dev, unsigned int ifnum, strncpy(ss->product, prod, USB_STOR_STRING_LEN); strncpy(ss->serial, serial, USB_STOR_STRING_LEN); if (strlen(ss->vendor) == 0) { - if (unusual_dev) + if (unusual_dev->vendorName) strncpy(ss->vendor, unusual_dev->vendorName, USB_STOR_STRING_LEN); else @@ -869,7 +756,7 @@ static void * storage_probe(struct usb_device *dev, unsigned int ifnum, USB_STOR_STRING_LEN); } if (strlen(ss->product) == 0) { - if (unusual_dev) + if (unusual_dev->productName) strncpy(ss->product, unusual_dev->productName, USB_STOR_STRING_LEN); else @@ -1118,6 +1005,8 @@ static void storage_disconnect(struct usb_device *dev, void *ptr) int __init usb_stor_init(void) { + printk(KERN_INFO "Initializing USB Mass Storage driver...\n"); + /* initialize internal global data elements */ us_list = NULL; init_MUTEX(&us_list_semaphore); diff --git a/drivers/usb/storage/usb.h b/drivers/usb/storage/usb.h index e0dac568d..9c8076c8d 100644 --- a/drivers/usb/storage/usb.h +++ b/drivers/usb/storage/usb.h @@ -1,7 +1,7 @@ /* Driver for USB Mass Storage compliant devices * Main Header File * - * $Id: usb.h,v 1.11 2000/11/13 22:38:55 mdharm Exp $ + * $Id: usb.h,v 1.12 2000/12/05 03:33:49 mdharm Exp $ * * Current development and maintenance by: * (c) 1999, 2000 Matthew Dharm (mdharm-usb@one-eyed-alien.net) @@ -84,13 +84,6 @@ struct us_data; */ struct us_unusual_dev { - /* we search the list based on these parameters */ - __u16 idVendor; - __u16 idProduct; - __u16 bcdDeviceMin; - __u16 bcdDeviceMax; - - /* the list specifies these parameters */ const char* vendorName; const char* productName; __u8 useProtocol; diff --git a/drivers/video/sbusfb.c b/drivers/video/sbusfb.c index f9fa0d213..4d5b66386 100644 --- a/drivers/video/sbusfb.c +++ b/drivers/video/sbusfb.c @@ -678,7 +678,7 @@ static int sbusfb_ioctl(struct inode *inode, struct file *file, u_int cmd, int end, count, index; struct fbcmap *cmap; - if (!fb->loadcmap) + if (!fb->loadcmap || !fb->color_map) return -EINVAL; i = verify_area (VERIFY_READ, (void *) arg, sizeof (struct fbcmap)); if (i) return i; @@ -1110,6 +1110,8 @@ sizechange: } if (!p) { + if (fb->color_map) + kfree(fb->color_map); kfree(fb); return; } @@ -1147,6 +1149,8 @@ sizechange: sbusfb_set_var(var, -1, &fb->info); if (register_framebuffer(&fb->info) < 0) { + if (fb->color_map) + kfree(fb->color_map); kfree(fb); return; } diff --git a/drivers/video/vfb.c b/drivers/video/vfb.c index 8833cbeac..b494e25c6 100644 --- a/drivers/video/vfb.c +++ b/drivers/video/vfb.c @@ -461,7 +461,7 @@ static u_long get_line_length(int xres_virtual, int bpp) { u_long length; - length = (xres_virtual+bpp-1)/bpp; + length = xres_virtual*bpp; length = (length+31)&-32; length >>= 3; return(length); diff --git a/fs/Config.in b/fs/Config.in index 53380d067..19430baa6 100644 --- a/fs/Config.in +++ b/fs/Config.in @@ -8,6 +8,8 @@ bool 'Quota support' CONFIG_QUOTA tristate 'Kernel automounter support' CONFIG_AUTOFS_FS tristate 'Kernel automounter version 4 support (also supports v3)' CONFIG_AUTOFS4_FS +dep_tristate 'Reiserfs support' CONFIG_REISERFS_FS $CONFIG_EXPERIMENTAL +dep_mbool ' Have reiserfs do extra internal checking' CONFIG_REISERFS_CHECK $CONFIG_REISERFS_FS $CONFIG_EXPERIMENTAL dep_tristate 'ADFS file system support' CONFIG_ADFS_FS $CONFIG_EXPERIMENTAL dep_mbool ' ADFS write support (DANGEROUS)' CONFIG_ADFS_FS_RW $CONFIG_ADFS_FS $CONFIG_EXPERIMENTAL diff --git a/fs/Makefile b/fs/Makefile index d63fd61bd..738e4418c 100644 --- a/fs/Makefile +++ b/fs/Makefile @@ -58,6 +58,7 @@ subdir-$(CONFIG_UDF_FS) += udf subdir-$(CONFIG_AUTOFS_FS) += autofs subdir-$(CONFIG_AUTOFS4_FS) += autofs4 subdir-$(CONFIG_ADFS_FS) += adfs +subdir-$(CONFIG_REISERFS_FS) += reiserfs subdir-$(CONFIG_DEVPTS_FS) += devpts subdir-$(CONFIG_SUN_OPENPROMFS) += openpromfs diff --git a/fs/buffer.c b/fs/buffer.c index 0096524bf..bea5f09e5 100644 --- a/fs/buffer.c +++ b/fs/buffer.c @@ -834,6 +834,10 @@ still_busy: return; } +void set_buffer_async_io(struct buffer_head *bh) { + bh->b_end_io = end_buffer_io_async ; +} + /* * Synchronise all the inode's dirty buffers to the disk. * @@ -1151,7 +1155,7 @@ void __bforget(struct buffer_head * buf) /* grab the lru lock here to block bdflush. */ spin_lock(&lru_list_lock); write_lock(&hash_table_lock); - if (!atomic_dec_and_test(&buf->b_count) || buffer_locked(buf)) + if (!atomic_dec_and_test(&buf->b_count) || buffer_locked(buf) || buffer_protected(buf)) goto in_use; __hash_unlink(buf); remove_inode_queue(buf); @@ -2411,6 +2415,7 @@ busy_buffer_page: loop = 1; goto cleaned_buffers_try_again; } + wakeup_bdflush(0); } return 0; } diff --git a/fs/coda/cnode.c b/fs/coda/cnode.c index c8dc3dd8f..12de14e7b 100644 --- a/fs/coda/cnode.c +++ b/fs/coda/cnode.c @@ -14,17 +14,21 @@ extern int coda_debug; extern int coda_print_entry; +static ViceFid NullFID = { 0, 0, 0 }; + inline int coda_fideq(ViceFid *fid1, ViceFid *fid2) { - if (fid1->Vnode != fid2->Vnode) - return 0; - if (fid1->Volume != fid2->Volume) - return 0; - if (fid1->Unique != fid2->Unique) - return 0; + if (fid1->Vnode != fid2->Vnode) return 0; + if (fid1->Volume != fid2->Volume) return 0; + if (fid1->Unique != fid2->Unique) return 0; return 1; } +static int coda_inocmp(struct inode *inode, unsigned long ino, void *opaque) +{ + return (coda_fideq((ViceFid *)opaque, &(ITOC(inode)->c_fid))); +} + static struct inode_operations coda_symlink_inode_operations = { readlink: page_readlink, follow_link: page_follow_link, @@ -62,7 +66,8 @@ struct inode * coda_iget(struct super_block * sb, ViceFid * fid, struct coda_inode_info *cii; ino_t ino = attr->va_fileid; - inode = iget(sb, ino); + inode = iget4(sb, ino, coda_inocmp, fid); + if ( !inode ) { CDEBUG(D_CNODE, "coda_iget: no inode\n"); return ERR_PTR(-ENOMEM); @@ -70,25 +75,16 @@ struct inode * coda_iget(struct super_block * sb, ViceFid * fid, /* check if the inode is already initialized */ cii = ITOC(inode); - if (cii->c_fid.Volume != 0 || cii->c_fid.Vnode != 0 || cii->c_fid.Unique != 0) { - /* see if it is the right one (might have an inode collision) */ - if ( !coda_fideq(fid, &cii->c_fid) ) { - printk("coda_iget: initialized inode old %s new %s!\n", - coda_f2s(&cii->c_fid), coda_f2s2(fid)); - iput(inode); - return ERR_PTR(-ENOENT); - } - /* we will still replace the attributes, type might have changed */ - goto out; + if (coda_fideq(&cii->c_fid, &NullFID)) { + /* new, empty inode found... initializing */ + cii->c_fid = *fid; + cii->c_vnode = inode; } - /* new, empty inode found... initializing */ - - /* Initialize the Coda inode info structure */ - cii->c_fid = *fid; - cii->c_vnode = inode; + /* we shouldnt see inode collisions anymore */ + if ( !coda_fideq(fid, &cii->c_fid) ) BUG(); -out: + /* always replace the attributes, type might have changed */ coda_fill_inode(inode, attr); return inode; } @@ -107,7 +103,6 @@ int coda_cnode_make(struct inode **inode, ViceFid *fid, struct super_block *sb) ENTRY; /* We get inode numbers from Venus -- see venus source */ - error = venus_getattr(sb, fid, &attr); if ( error ) { CDEBUG(D_CNODE, @@ -183,7 +178,7 @@ struct inode *coda_fid_to_inode(ViceFid *fid, struct super_block *sb) if ( coda_fideq(&cii->c_fid, fid) ) { inode = cii->c_vnode; CDEBUG(D_INODE, "volume root, found %ld\n", inode->i_ino); - iget(sb, inode->i_ino); + iget4(sb, inode->i_ino, coda_inocmp, fid); return inode; } } @@ -192,7 +187,7 @@ struct inode *coda_fid_to_inode(ViceFid *fid, struct super_block *sb) /* fid is not weird: ino should be computable */ nr = coda_f2i(fid); - inode = iget(sb, nr); + inode = iget4(sb, nr, coda_inocmp, fid); if ( !inode ) { printk("coda_fid_to_inode: null from iget, sb %p, nr %ld.\n", sb, (long)nr); @@ -202,18 +197,11 @@ struct inode *coda_fid_to_inode(ViceFid *fid, struct super_block *sb) /* check if this inode is linked to a cnode */ cii = ITOC(inode); - /* make sure this is the one we want */ - if ( coda_fideq(fid, &cii->c_fid) ) { - CDEBUG(D_INODE, "found %ld\n", inode->i_ino); - return inode; - } - -#if 0 - printk("coda_fid2inode: bad cnode (ino %ld, fid %s)", nr, coda_f2s(fid)); -#endif - iput(inode); - return NULL; + /* we shouldn't have inode collisions anymore */ + if ( !coda_fideq(fid, &cii->c_fid) ) BUG(); + CDEBUG(D_INODE, "found %ld\n", inode->i_ino); + return inode; } /* the CONTROL inode is made without asking attributes from Venus */ @@ -388,10 +388,8 @@ static int exec_mmap(void) old_mm = current->mm; if (old_mm && atomic_read(&old_mm->mm_users) == 1) { - flush_cache_mm(old_mm); mm_release(); exit_mmap(old_mm); - flush_tlb_mm(old_mm); return 0; } @@ -407,6 +405,7 @@ static int exec_mmap(void) /* Add it to the list of mm's */ spin_lock(&mmlist_lock); list_add(&mm->mmlist, &init_mm.mmlist); + mmlist_nr++; spin_unlock(&mmlist_lock); task_lock(current); @@ -938,12 +937,14 @@ int do_coredump(long signr, struct pt_regs * regs) #else corename[4] = '\0'; #endif - file = filp_open(corename, O_CREAT | 2 | O_TRUNC | O_NOFOLLOW, 0600); + file = filp_open(corename, O_CREAT | 2 | O_NOFOLLOW, 0600); if (IS_ERR(file)) goto fail; inode = file->f_dentry->d_inode; if (inode->i_nlink > 1) goto close_fail; /* multiple links - don't dump */ + if (d_unhashed(file->f_dentry)) + goto close_fail; if (!S_ISREG(inode->i_mode)) goto close_fail; @@ -951,6 +952,8 @@ int do_coredump(long signr, struct pt_regs * regs) goto close_fail; if (!file->f_op->write) goto close_fail; + if (do_truncate(file->f_dentry, 0) != 0) + goto close_fail; if (!binfmt->core_dump(signr, regs, file)) goto close_fail; unlock_kernel(); diff --git a/fs/fat/cache.c b/fs/fat/cache.c index 0cc0ee242..35874bf50 100644 --- a/fs/fat/cache.c +++ b/fs/fat/cache.c @@ -24,6 +24,7 @@ #endif static struct fat_cache *fat_cache,cache[FAT_CACHE]; +static spinlock_t fat_cache_lock = SPIN_LOCK_UNLOCKED; /* Returns the this'th FAT entry, -1 if it is an end-of-file entry. If new_value is != -1, that FAT entry is replaced by it. */ @@ -133,13 +134,16 @@ int default_fat_access(struct super_block *sb,int nr,int new_value) return next; } - void fat_cache_init(void) { static int initialized = 0; int count; - if (initialized) return; + spin_lock(&fat_cache_lock); + if (initialized) { + spin_unlock(&fat_cache_lock); + return; + } fat_cache = &cache[0]; for (count = 0; count < FAT_CACHE; count++) { cache[count].device = 0; @@ -147,6 +151,7 @@ void fat_cache_init(void) &cache[count+1]; } initialized = 1; + spin_unlock(&fat_cache_lock); } @@ -157,6 +162,7 @@ void fat_cache_lookup(struct inode *inode,int cluster,int *f_clu,int *d_clu) if (!first) return; + spin_lock(&fat_cache_lock); for (walk = fat_cache; walk; walk = walk->next) if (inode->i_dev == walk->device && walk->start_cluster == first @@ -166,8 +172,12 @@ void fat_cache_lookup(struct inode *inode,int cluster,int *f_clu,int *d_clu) #ifdef DEBUG printk("cache hit: %d (%d)\n",walk->file_cluster,*d_clu); #endif - if ((*f_clu = walk->file_cluster) == cluster) return; + if ((*f_clu = walk->file_cluster) == cluster) { + spin_unlock(&fat_cache_lock); + return; + } } + spin_unlock(&fat_cache_lock); #ifdef DEBUG printk("cache miss\n"); #endif @@ -197,6 +207,7 @@ void fat_cache_add(struct inode *inode,int f_clu,int d_clu) int first = MSDOS_I(inode)->i_start; last = NULL; + spin_lock(&fat_cache_lock); for (walk = fat_cache; walk->next; walk = (last = walk)->next) if (inode->i_dev == walk->device && walk->start_cluster == first @@ -204,17 +215,22 @@ void fat_cache_add(struct inode *inode,int f_clu,int d_clu) if (walk->disk_cluster != d_clu) { printk("FAT cache corruption inode=%ld\n", inode->i_ino); + spin_unlock(&fat_cache_lock); fat_cache_inval_inode(inode); return; } /* update LRU */ - if (last == NULL) return; + if (last == NULL) { + spin_unlock(&fat_cache_lock); + return; + } last->next = walk->next; walk->next = fat_cache; fat_cache = walk; #ifdef DEBUG list_cache(); #endif + spin_unlock(&fat_cache_lock); return; } walk->device = inode->i_dev; @@ -224,6 +240,7 @@ list_cache(); last->next = NULL; walk->next = fat_cache; fat_cache = walk; + spin_unlock(&fat_cache_lock); #ifdef DEBUG list_cache(); #endif @@ -238,10 +255,12 @@ void fat_cache_inval_inode(struct inode *inode) struct fat_cache *walk; int first = MSDOS_I(inode)->i_start; + spin_lock(&fat_cache_lock); for (walk = fat_cache; walk; walk = walk->next) if (walk->device == inode->i_dev && walk->start_cluster == first) walk->device = 0; + spin_unlock(&fat_cache_lock); } @@ -249,9 +268,11 @@ void fat_cache_inval_dev(kdev_t device) { struct fat_cache *walk; + spin_lock(&fat_cache_lock); for (walk = fat_cache; walk; walk = walk->next) if (walk->device == device) walk->device = 0; + spin_unlock(&fat_cache_lock); } diff --git a/fs/hpfs/inode.c b/fs/hpfs/inode.c index 06f5d0783..aa09b0dd9 100644 --- a/fs/hpfs/inode.c +++ b/fs/hpfs/inode.c @@ -299,6 +299,7 @@ int hpfs_notify_change(struct dentry *dentry, struct iattr *attr) { struct inode *inode = dentry->d_inode; int error; + if ((attr->ia_valid & ATTR_SIZE) && attr->ia_size > inode->i_size) return -EINVAL; if (inode->i_sb->s_hpfs_root == inode->i_ino) return -EINVAL; if ((error = inode_change_ok(inode, attr))) return error; inode_setattr(inode, attr); diff --git a/fs/inode.c b/fs/inode.c index 1d2b23b70..d51bbddbe 100644 --- a/fs/inode.c +++ b/fs/inode.c @@ -136,6 +136,16 @@ void __mark_inode_dirty(struct inode *inode, int flags) struct super_block * sb = inode->i_sb; if (sb) { + /* Don't do this for I_DIRTY_PAGES - that doesn't actually dirty the inode itself */ + if (flags & (I_DIRTY | I_DIRTY_SYNC)) { + if (sb->s_op && sb->s_op->dirty_inode) + sb->s_op->dirty_inode(inode); + } + + /* avoid the locking if we can */ + if ((inode->i_state & flags) == flags) + return; + spin_lock(&inode_lock); if ((inode->i_state & flags) != flags) { inode->i_state |= flags; @@ -676,7 +686,17 @@ static struct inode * get_new_inode(struct super_block *sb, unsigned long ino, s spin_unlock(&inode_lock); clean_inode(inode); - sb->s_op->read_inode(inode); + + /* reiserfs specific hack right here. We don't + ** want this to last, and are looking for VFS changes + ** that will allow us to get rid of it. + ** -- mason@suse.com + */ + if (sb->s_op->read_inode2) { + sb->s_op->read_inode2(inode, opaque) ; + } else { + sb->s_op->read_inode(inode); + } /* * This is special! We do not need the spinlock diff --git a/fs/ncpfs/dir.c b/fs/ncpfs/dir.c index a187b1199..c16d93614 100644 --- a/fs/ncpfs/dir.c +++ b/fs/ncpfs/dir.c @@ -871,6 +871,12 @@ int ncp_create_new(struct inode *dir, struct dentry *dentry, int mode, goto out; error = -EACCES; + + if (S_ISREG(mode) && + (server->m.flags & NCP_MOUNT_EXTRAS) && + (mode & S_IXUGO)) + attributes |= aSYSTEM; + result = ncp_open_create_file_or_subdir(server, dir, __name, OC_MODE_CREATE | OC_MODE_OPEN | OC_MODE_REPLACE, attributes, AR_READ | AR_WRITE, &finfo); diff --git a/fs/ncpfs/sock.c b/fs/ncpfs/sock.c index 171f0cb51..470ea5e96 100644 --- a/fs/ncpfs/sock.c +++ b/fs/ncpfs/sock.c @@ -456,7 +456,10 @@ static int ncp_do_request(struct ncp_server *server, int size, spin_lock_irqsave(¤t->sigmask_lock, flags); old_set = current->blocked; - mask = sigmask(SIGKILL) | sigmask(SIGSTOP); + if (current->flags & PF_EXITING) + mask = 0; + else + mask = sigmask(SIGKILL); if (server->m.flags & NCP_MOUNT_INTR) { /* FIXME: This doesn't seem right at all. So, like, we can't handle SIGINT and get whatever to stop? diff --git a/fs/nfs/flushd.c b/fs/nfs/flushd.c index 700504ded..502b490b8 100644 --- a/fs/nfs/flushd.c +++ b/fs/nfs/flushd.c @@ -34,7 +34,7 @@ #include <linux/sunrpc/clnt.h> #include <linux/sunrpc/sched.h> -#include <linux/spinlock.h> +#include <linux/smp_lock.h> #include <linux/nfs.h> #include <linux/nfs_fs.h> @@ -53,11 +53,6 @@ static struct rpc_wait_queue flushd_queue = RPC_INIT_WAITQ("nfs_flushd"); /* - * Spinlock - */ -spinlock_t nfs_flushd_lock = SPIN_LOCK_UNLOCKED; - -/* * Local function declarations. */ static void nfs_flushd(struct rpc_task *); @@ -68,19 +63,20 @@ int nfs_reqlist_init(struct nfs_server *server) { struct nfs_reqlist *cache; struct rpc_task *task; - int status = 0; + int status; dprintk("NFS: writecache_init\n"); - spin_lock(&nfs_flushd_lock); - cache = server->rw_requests; - if (cache->task) + lock_kernel(); + status = -ENOMEM; + /* Create the RPC task */ + if (!(task = rpc_new_task(server->client, NULL, RPC_TASK_ASYNC))) goto out_unlock; - /* Create the RPC task */ - status = -ENOMEM; - task = rpc_new_task(server->client, NULL, RPC_TASK_ASYNC); - if (!task) + cache = server->rw_requests; + + status = 0; + if (cache->task) goto out_unlock; task->tk_calldata = server; @@ -94,11 +90,13 @@ int nfs_reqlist_init(struct nfs_server *server) task->tk_action = nfs_flushd; task->tk_exit = nfs_flushd_exit; - spin_unlock(&nfs_flushd_lock); rpc_execute(task); + unlock_kernel(); return 0; out_unlock: - spin_unlock(&nfs_flushd_lock); + if (task) + rpc_release_task(task); + unlock_kernel(); return status; } @@ -106,23 +104,24 @@ void nfs_reqlist_exit(struct nfs_server *server) { struct nfs_reqlist *cache; + lock_kernel(); cache = server->rw_requests; if (!cache) - return; + goto out; dprintk("NFS: reqlist_exit (ptr %p rpc %p)\n", cache, cache->task); + while (cache->task || cache->inodes) { - spin_lock(&nfs_flushd_lock); if (!cache->task) { - spin_unlock(&nfs_flushd_lock); nfs_reqlist_init(server); } else { cache->task->tk_status = -ENOMEM; rpc_wake_up_task(cache->task); - spin_unlock(&nfs_flushd_lock); } interruptible_sleep_on_timeout(&cache->request_wait, 1 * HZ); } + out: + unlock_kernel(); } int nfs_reqlist_alloc(struct nfs_server *server) @@ -161,7 +160,6 @@ static void inode_append_flushd(struct inode *inode) struct nfs_reqlist *cache = NFS_REQUESTLIST(inode); struct inode **q; - spin_lock(&nfs_flushd_lock); if (NFS_FLAGS(inode) & NFS_INO_FLUSH) goto out; inode->u.nfs_i.hash_next = NULL; @@ -177,7 +175,6 @@ static void inode_append_flushd(struct inode *inode) NFS_FLAGS(inode) |= NFS_INO_FLUSH; atomic_inc(&inode->i_count); out: - spin_unlock(&nfs_flushd_lock); } void inode_remove_flushd(struct inode *inode) @@ -185,7 +182,7 @@ void inode_remove_flushd(struct inode *inode) struct nfs_reqlist *cache = NFS_REQUESTLIST(inode); struct inode **q; - spin_lock(&nfs_flushd_lock); + lock_kernel(); if (!(NFS_FLAGS(inode) & NFS_INO_FLUSH)) goto out; @@ -198,7 +195,7 @@ void inode_remove_flushd(struct inode *inode) iput(inode); } out: - spin_unlock(&nfs_flushd_lock); + unlock_kernel(); } void inode_schedule_scan(struct inode *inode, unsigned long time) @@ -207,6 +204,7 @@ void inode_schedule_scan(struct inode *inode, unsigned long time) struct rpc_task *task; unsigned long mintimeout; + lock_kernel(); if (time_after(NFS_NEXTSCAN(inode), time)) NFS_NEXTSCAN(inode) = time; mintimeout = jiffies + 1 * HZ; @@ -214,16 +212,14 @@ void inode_schedule_scan(struct inode *inode, unsigned long time) mintimeout = NFS_NEXTSCAN(inode); inode_append_flushd(inode); - spin_lock(&nfs_flushd_lock); task = cache->task; if (!task) { - spin_unlock(&nfs_flushd_lock); nfs_reqlist_init(NFS_SERVER(inode)); } else { if (time_after(cache->runat, mintimeout)) rpc_wake_up_task(task); - spin_unlock(&nfs_flushd_lock); } + unlock_kernel(); } @@ -240,10 +236,8 @@ nfs_flushd(struct rpc_task *task) server = (struct nfs_server *) task->tk_calldata; cache = server->rw_requests; - spin_lock(&nfs_flushd_lock); next = cache->inodes; cache->inodes = NULL; - spin_unlock(&nfs_flushd_lock); while ((inode = next) != NULL) { next = next->u.nfs_i.hash_next; @@ -280,13 +274,11 @@ nfs_flushd(struct rpc_task *task) task->tk_timeout = delay; cache->runat = jiffies + task->tk_timeout; - spin_lock(&nfs_flushd_lock); if (!atomic_read(&cache->nr_requests) && !cache->inodes) { cache->task = NULL; task->tk_action = NULL; } else rpc_sleep_on(&flushd_queue, task, NULL, NULL); - spin_unlock(&nfs_flushd_lock); } static void @@ -297,10 +289,8 @@ nfs_flushd_exit(struct rpc_task *task) server = (struct nfs_server *) task->tk_calldata; cache = server->rw_requests; - spin_lock(&nfs_flushd_lock); if (cache->task == task) cache->task = NULL; - spin_unlock(&nfs_flushd_lock); wake_up(&cache->request_wait); } diff --git a/fs/partitions/msdos.c b/fs/partitions/msdos.c index 69f6f0ce9..0d72631c8 100644 --- a/fs/partitions/msdos.c +++ b/fs/partitions/msdos.c @@ -36,7 +36,7 @@ #include "check.h" #include "msdos.h" -#if CONFIG_BLK_DEV_MD && CONFIG_AUTODETECT_RAID +#if CONFIG_BLK_DEV_MD extern void md_autodetect_dev(kdev_t dev); #endif @@ -136,7 +136,7 @@ static void extended_partition(struct gendisk *hd, kdev_t dev) add_gd_partition(hd, current_minor, this_sector+START_SECT(p)*sector_size, NR_SECTS(p)*sector_size); -#if CONFIG_BLK_DEV_MD && CONFIG_AUTODETECT_RAID +#if CONFIG_BLK_DEV_MD if (SYS_IND(p) == LINUX_RAID_PARTITION) { md_autodetect_dev(MKDEV(hd->major,current_minor)); } @@ -448,7 +448,7 @@ check_table: continue; add_gd_partition(hd, minor, first_sector+START_SECT(p)*sector_size, NR_SECTS(p)*sector_size); -#if CONFIG_BLK_DEV_MD && CONFIG_AUTODETECT_RAID +#if CONFIG_BLK_DEV_MD if (SYS_IND(p) == LINUX_RAID_PARTITION) { md_autodetect_dev(MKDEV(hd->major,minor)); } diff --git a/fs/proc/kcore.c b/fs/proc/kcore.c index 00c16769e..f95e9f624 100644 --- a/fs/proc/kcore.c +++ b/fs/proc/kcore.c @@ -17,6 +17,7 @@ #include <linux/elf.h> #include <linux/elfcore.h> #include <linux/vmalloc.h> +#include <linux/highmem.h> #include <asm/uaccess.h> #include <asm/io.h> diff --git a/fs/ramfs/inode.c b/fs/ramfs/inode.c index 898ab8178..69ea5fb23 100644 --- a/fs/ramfs/inode.c +++ b/fs/ramfs/inode.c @@ -81,6 +81,7 @@ static int ramfs_readpage(struct file *file, struct page * page) static int ramfs_writepage(struct page *page) { SetPageDirty(page); + UnlockPage(page); return 0; } diff --git a/fs/reiserfs/.cvsignore b/fs/reiserfs/.cvsignore new file mode 100644 index 000000000..857dd22e9 --- /dev/null +++ b/fs/reiserfs/.cvsignore @@ -0,0 +1,2 @@ +.depend +.*.flags diff --git a/fs/reiserfs/Makefile b/fs/reiserfs/Makefile new file mode 100644 index 000000000..fda0b17ec --- /dev/null +++ b/fs/reiserfs/Makefile @@ -0,0 +1,20 @@ +# +# Makefile for the linux reiser-filesystem routines. +# +# Note! Dependencies are done automagically by 'make dep', which also +# removes any old dependencies. DON'T put your own dependencies here +# unless it's something special (ie not a .c file). +# +# Note 2! The CFLAGS definitions are now in the main makefile... + +O_TARGET := reiserfs.o +obj-y := bitmap.o do_balan.o namei.o inode.o file.o dir.o fix_node.o super.o prints.o objectid.o \ +lbalance.o ibalance.o stree.o hashes.o buffer2.o tail_conversion.o journal.o resize.o tail_conversion.o version.o item_ops.o ioctl.o + +obj-m := $(O_TARGET) + +include $(TOPDIR)/Rules.make + +TAGS: + etags *.c + diff --git a/fs/reiserfs/README b/fs/reiserfs/README new file mode 100644 index 000000000..37b23959c --- /dev/null +++ b/fs/reiserfs/README @@ -0,0 +1,157 @@ +[LICENSING] + +ReiserFS is hereby licensed under the GNU General +Public License version 2. + +Source code files that contain the phrase "licensing governed by +reiserfs/README" are "governed files" throughout this file. Governed +files are licensed under the GPL. The portions of them owned by Hans +Reiser, or authorized to be licensed by him, have been in the past, +and likely will be in the future, licensed to other parties under +other licenses. If you add your code to governed files, and don't +want it to be owned by Hans Reiser, put your copyright label on that +code so the poor blight and his customers can keep things straight. +All portions of governed files not labeled otherwise are owned by Hans +Reiser, and by adding your code to it, widely distributing it to +others or sending us a patch, and leaving the sentence in stating that +licensing is governed by the statement in this file, you accept this. +It will be a kindness if you identify whether Hans Reiser is allowed +to license code labeled as owned by you on your behalf other than +under the GPL, because he wants to know if it is okay to do so and put +a check in the mail to you (for non-trivial improvements) when he +makes his next sale. He makes no guarantees as to the amount if any, +though he feels motivated to motivate contributors, and you can surely +discuss this with him before or after contributing. You have the +right to decline to allow him to license your code contribution other +than under the GPL. + +Further licensing options are available for commercial and/or other +interests directly from Hans Reiser: hans@reiser.to. If you interpret +the GPL as not allowing those additional licensing options, you read +it wrongly, and Richard Stallman agrees with me, when carefully read +you can see that those restrictions on additional terms do not apply +to the owner of the copyright, and my interpretation of this shall +govern for this license. + +Finally, nothing in this license shall be interpreted to allow you to +fail to fairly credit me, or to remove my credits, without my +permission, unless you are an end user not redistributing to others. +If you have doubts about how to properly do that, or about what is +fair, ask. (Last I spoke with him Richard was contemplating how best +to address the fair crediting issue in the next GPL version.) + +[END LICENSING] + +Reiserfs is a file system based on balanced tree algorithms, which is +described at http://devlinux.com/namesys. + +Stop reading here. Go there, then return. + +Send bug reports to yura@namesys.botik.ru. + +mkreiserfs and other utilities are in reiserfs/utils, or wherever your +Linux provider put them. There is some disagreement about how useful +it is for users to get their fsck and mkreiserfs out of sync with the +version of reiserfs that is in their kernel, with many important +distributors wanting them out of sync.:-) Please try to remember to +recompile and reinstall fsck and mkreiserfs with every update of +reiserfs, this is a common source of confusion. Note that some of the +utilities cannot be compiled without accessing the balancing code +which is in the kernel code, and relocating the utilities may require +you to specify where that code can be found. + +Yes, if you update your reiserfs kernel module you do have to +recompile your kernel, most of the time. The errors you get will be +quite cryptic if your forget to do so. + +Real users, as opposed to folks who want to hack and then understand +what went wrong, will want REISERFS_CHECK off. + +Hideous Commercial Pitch: Spread your development costs across other OS +vendors. Select from the best in the world, not the best in your +building, by buying from third party OS component suppliers. Leverage +the software component development power of the internet. Be the most +aggressive in taking advantage of the commercial possibilities of +decentralized internet development, and add value through your branded +integration that you sell as an operating system. Let your competitors +be the ones to compete against the entire internet by themselves. Be +hip, get with the new economic trend, before your competitors do. Send +email to hans@reiser.to. + +To understand the code, after reading the website, start reading the +code by reading reiserfs_fs.h first. + +Hans Reiser was the project initiator, primary architect, source of all +funding for the first 5.5 years, and one of the programmers. He owns +the copyright. + +Vladimir Saveljev was one of the programmers, and he worked long hours +writing the cleanest code. He always made the effort to be the best he +could be, and to make his code the best that it could be. What resulted +was quite remarkable. I don't think that money can ever motivate someone +to work the way he did, he is one of the most selfless men I know. + +Yura helps with benchmarking, coding hashes, and block pre-allocation +code. + +Anatoly Pinchuk is a former member of our team who worked closely with +Vladimir throughout the project's development. He wrote a quite +substantial portion of the total code. He realized that there was a +space problem with packing tails of files for files larger than a node +that start on a node aligned boundary (there are reasons to want to node +align files), and he invented and implemented indirect items and +unformatted nodes as the solution. + +Konstantin Shvachko, with the help of the Russian version of a VC, +tried to put me in a position where I was forced into giving control +of the project to him. (Fortunately, as the person paying the money +for all salaries from my dayjob I owned all copyrights, and you can't +really force takeovers of sole proprietorships.) This was something +curious, because he never really understood the value of our project, +why we should do what we do, or why innovation was possible in +general, but he was sure that he ought to be controlling it. Every +innovation had to be forced past him while he was with us. He added +two years to the time required to complete reiserfs, and was a net +loss for me. Mikhail Gilula was a brilliant innovator who also left +in a destructive way that erased the value of his contributions, and +that he was shown much generosity just makes it more painful. + +Grigory Zaigralin was an extremely effective system administrator for +our group. + +Igor Krasheninnikov was wonderful at hardware procurement, repair, and +network installation. + +Jeremy Fitzhardinge wrote the teahash.c code, and he gives credit to a +textbook he got the algorithm from in the code. Note that his analysis +of how we could use the hashing code in making 32 bit NFS cookies work +was probably more important than the actual algorithm. Colin Plumb also +contributed to it. + +Chris Mason dived right into our code, and in just a few months produced +the journaling code that dramatically increased the value of ReiserFS. +He is just an amazing programmer. + +Igor Zagorovsky is writing much of the new item handler and extent code +for our next major release. + +Alexander Zarochentcev (sometimes known as zam, or sasha), wrote the +resizer, and is hard at work on implementing allocate on flush. SGI +implemented allocate on flush before us for XFS, and generously took +the time to convince me we should do it also. They are great people, +and a great company. + +Yuri Shevchuk and Nikita Danilov are doing squid cache optimization. + +Vitaly Fertman is doing fsck. + +SuSE, IntegratedLinux.com, Ecila, MP3.com, bigstorage.com, and the +Alpha PC Company made it possible for me to not have a day job +anymore, and to dramatically increase our staffing. Ecila funded +hypertext feature development, MP3.com funded journaling, SuSE funded +core development, IntegratedLinux.com funded squid web cache +appliances, bigstorage.com funded HSM, and the alpha PC company funded +the alpha port. Many of these tasks were helped by sponsors other +than the ones just named. SuSE has helped in much more than just +funding.... + diff --git a/fs/reiserfs/bitmap.c b/fs/reiserfs/bitmap.c new file mode 100644 index 000000000..ea83fd549 --- /dev/null +++ b/fs/reiserfs/bitmap.c @@ -0,0 +1,679 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/sched.h> +#include <linux/reiserfs_fs.h> +#include <linux/locks.h> +#include <asm/bitops.h> + +#else + +#include "nokernel.h" + +#endif + + +#ifdef CONFIG_REISERFS_CHECK + +/* this is a safety check to make sure +** blocks are reused properly. used for debugging only. +** +** this checks, that block can be reused, and it has correct state +** (free or busy) +*/ +int is_reusable (struct super_block * s, unsigned long block, int bit_value) +{ + int i, j; + + if (block == 0 || block >= SB_BLOCK_COUNT (s)) { + reiserfs_warning ("vs-4010: is_reusable: block number is out of range %lu (%u)\n", + block, SB_BLOCK_COUNT (s)); + return 0; + } + + /* it can't be one of the bitmap blocks */ + for (i = 0; i < SB_BMAP_NR (s); i ++) + if (block == SB_AP_BITMAP (s)[i]->b_blocknr) { + reiserfs_warning ("vs: 4020: is_reusable: " + "bitmap block %lu(%u) can't be freed or reused\n", + block, SB_BMAP_NR (s)); + return 0; + } + + i = block / (s->s_blocksize << 3); + if (i >= SB_BMAP_NR (s)) { + reiserfs_warning ("vs-4030: is_reusable: there is no so many bitmap blocks: " + "block=%lu, bitmap_nr=%d\n", block, i); + return 0; + } + + j = block % (s->s_blocksize << 3); + if ((bit_value == 0 && + reiserfs_test_le_bit(j, SB_AP_BITMAP(s)[i]->b_data)) || + (bit_value == 1 && + reiserfs_test_le_bit(j, SB_AP_BITMAP (s)[i]->b_data) == 0)) { + reiserfs_warning ("vs-4040: is_reusable: corresponding bit of block %lu does not " + "match required value (i==%d, j==%d) test_bit==%d\n", + block, i, j, reiserfs_test_le_bit (j, SB_AP_BITMAP (s)[i]->b_data)); + return 0; + } + + if (bit_value == 0 && block == SB_ROOT_BLOCK (s)) { + reiserfs_warning ("vs-4050: is_reusable: this is root block (%u), " + "it must be busy", SB_ROOT_BLOCK (s)); + return 0; + } + + return 1; +} + + + + +#endif /* CONFIG_REISERFS_CHECK */ + +#if 0 +/*%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%*/ +int is_used (struct super_block * s, unsigned long block) +{ + int i, j; + + i = block / (s->s_blocksize << 3); + j = block % (s->s_blocksize << 3); + if (reiserfs_test_le_bit(j, SB_AP_BITMAP (s)[i]->b_data)) + return 1; + return 0; + +} +/*%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%*/ +#endif + + +/* get address of corresponding bit (bitmap block number and offset in it) */ +static inline void get_bit_address (struct super_block * s, unsigned long block, int * bmap_nr, int * offset) +{ + /* It is in the bitmap block number equal to the block number divided by the number of + bits in a block. */ + *bmap_nr = block / (s->s_blocksize << 3); + /* Within that bitmap block it is located at bit offset *offset. */ + *offset = block % (s->s_blocksize << 3); + return; +} + + +/* There would be a modest performance benefit if we write a version + to free a list of blocks at once. -Hans */ + /* I wonder if it would be less modest + now that we use journaling. -Hans */ +void reiserfs_free_block (struct reiserfs_transaction_handle *th, unsigned long block) +{ + struct super_block * s = th->t_super; + struct reiserfs_super_block * rs; + struct buffer_head * sbh; + struct buffer_head ** apbh; + int nr, offset; + +#ifdef CONFIG_REISERFS_CHECK + if (!s) + reiserfs_panic (s, "vs-4060: reiserfs_free_block: trying to free block on nonexistent device"); + + if (is_reusable (s, block, 1) == 0) + reiserfs_panic (s, "vs-4070: reiserfs_free_block: can not free such block"); +#endif + + rs = SB_DISK_SUPER_BLOCK (s); + sbh = SB_BUFFER_WITH_SB (s); + apbh = SB_AP_BITMAP (s); + + get_bit_address (s, block, &nr, &offset); + + /* mark it before we clear it, just in case */ + journal_mark_freed(th, s, block) ; + + reiserfs_prepare_for_journal(s, apbh[nr], 1 ) ; + + /* clear bit for the given block in bit map */ + if (!reiserfs_test_and_clear_le_bit (offset, apbh[nr]->b_data)) { + reiserfs_warning ("vs-4080: reiserfs_free_block: " + "free_block (%04x:%lu)[dev:blocknr]: bit already cleared\n", + s->s_dev, block); + } + journal_mark_dirty (th, s, apbh[nr]); + + reiserfs_prepare_for_journal(s, sbh, 1) ; + /* update super block */ + rs->s_free_blocks = cpu_to_le32 (le32_to_cpu (rs->s_free_blocks) + 1); + + journal_mark_dirty (th, s, sbh); + s->s_dirt = 1; +} + + + +/* beginning from offset-th bit in bmap_nr-th bitmap block, + find_forward finds the closest zero bit. It returns 1 and zero + bit address (bitmap, offset) if zero bit found or 0 if there is no + zero bit in the forward direction */ +/* The function is NOT SCHEDULE-SAFE! */ +static int find_forward (struct super_block * s, int * bmap_nr, int * offset, int for_unformatted) +{ + int i, j; + struct buffer_head * bh; + unsigned long block_to_try = 0; + unsigned long next_block_to_try = 0 ; + + for (i = *bmap_nr; i < SB_BMAP_NR (s); i ++, *offset = 0) { + /* get corresponding bitmap block */ + bh = SB_AP_BITMAP (s)[i]; + if (buffer_locked (bh)) { + __wait_on_buffer (bh); + } +retry: + j = reiserfs_find_next_zero_le_bit ((unsigned long *)bh->b_data, + s->s_blocksize << 3, *offset); + + /* wow, this really needs to be redone. We can't allocate a block if + ** it is in the journal somehow. reiserfs_in_journal makes a suggestion + ** for a good block if the one you ask for is in the journal. Note, + ** reiserfs_in_journal might reject the block it suggests. The big + ** gain from the suggestion is when a big file has been deleted, and + ** many blocks show free in the real bitmap, but are all not free + ** in the journal list bitmaps. + ** + ** this whole system sucks. The bitmaps should reflect exactly what + ** can and can't be allocated, and the journal should update them as + ** it goes. TODO. + */ + if (j < (s->s_blocksize << 3)) { + block_to_try = (i * (s->s_blocksize << 3)) + j; + + /* the block is not in the journal, we can proceed */ + if (!(reiserfs_in_journal(s, s->s_dev, block_to_try, s->s_blocksize, for_unformatted, &next_block_to_try))) { + *bmap_nr = i; + *offset = j; + return 1; + } + /* the block is in the journal */ + else if ((j+1) < (s->s_blocksize << 3)) { /* try again */ + /* reiserfs_in_journal suggested a new block to try */ + if (next_block_to_try > 0) { + int new_i ; + get_bit_address (s, next_block_to_try, &new_i, offset); + + /* block is not in this bitmap. reset i and continue + ** we only reset i if new_i is in a later bitmap. + */ + if (new_i > i) { + i = (new_i - 1 ); /* i gets incremented by the for loop */ + continue ; + } + } else { + /* no suggestion was made, just try the next block */ + *offset = j+1 ; + } + goto retry ; + } + } + } + /* zero bit not found */ + return 0; +} + +/* return 0 if no free blocks, else return 1 */ +/* The function is NOT SCHEDULE-SAFE! +** because the bitmap block we want to change could be locked, and on its +** way to the disk when we want to read it, and because of the +** flush_async_commits. Per bitmap block locks won't help much, and +** really aren't needed, as we retry later on if we try to set the bit +** and it is already set. +*/ +static int find_zero_bit_in_bitmap (struct super_block * s, + unsigned long search_start, + int * bmap_nr, int * offset, + int for_unformatted) +{ + int retry_count = 0 ; + /* get bit location (bitmap number and bit offset) of search_start block */ + get_bit_address (s, search_start, bmap_nr, offset); + + /* note that we search forward in the bitmap, benchmarks have shown that it is better to allocate in increasing + sequence, which is probably due to the disk spinning in the forward direction.. */ + if (find_forward (s, bmap_nr, offset, for_unformatted) == 0) { + /* there wasn't a free block with number greater than our + starting point, so we are going to go to the beginning of the disk */ + +retry: + search_start = 0; /* caller will reset search_start for itself also. */ + get_bit_address (s, search_start, bmap_nr, offset); + if (find_forward (s, bmap_nr,offset,for_unformatted) == 0) { + if (for_unformatted) { /* why only unformatted nodes? -Hans */ + if (retry_count == 0) { + /* we've got a chance that flushing async commits will free up + ** some space. Sync then retry + */ + flush_async_commits(s) ; + retry_count++ ; + goto retry ; + } else if (retry_count > 0) { + /* nothing more we can do. Make the others wait, flush + ** all log blocks to disk, and flush to their home locations. + ** this will free up any blocks held by the journal + */ + SB_JOURNAL(s)->j_must_wait = 1 ; + } + } + return 0; + } + } + return 1; +} + +/* get amount_needed free block numbers from scanning the bitmap of + free/used blocks. + + Optimize layout by trying to find them starting from search_start + and moving in increasing blocknr direction. (This was found to be + faster than using a bi-directional elevator_direction, in part + because of disk spin direction, in part because by the time one + reaches the end of the disk the beginning of the disk is the least + congested). + + search_start is the block number of the left + semantic neighbor of the node we create. + + return CARRY_ON if everything is ok + return NO_DISK_SPACE if out of disk space + return NO_MORE_UNUSED_CONTIGUOUS_BLOCKS if the block we found is not contiguous to the last one + + return block numbers found, in the array free_blocknrs. assumes + that any non-zero entries already present in the array are valid. + This feature is perhaps convenient coding when one might not have + used all blocknrs from the last time one called this function, or + perhaps it is an archaism from the days of schedule tracking, one + of us ought to reread the code that calls this, and analyze whether + it is still the right way to code it. + + spare space is used only when priority is set to 1. reiserfsck has + its own reiserfs_new_blocknrs, which can use reserved space + + exactly what reserved space? the SPARE_SPACE? if so, please comment reiserfs.h. + + Give example of who uses spare space, and say that it is a deadlock + avoidance mechanism. -Hans */ + +/* This function is NOT SCHEDULE-SAFE! */ + +static int do_reiserfs_new_blocknrs (struct reiserfs_transaction_handle *th, + unsigned long * free_blocknrs, + unsigned long search_start, + int amount_needed, int priority, + int for_unformatted, + int for_prealloc) +{ + struct super_block * s = th->t_super; + int i, j; + unsigned long * block_list_start = free_blocknrs; + int init_amount_needed = amount_needed; + unsigned long new_block = 0 ; + + if (SB_FREE_BLOCKS (s) < SPARE_SPACE && !priority) + /* we can answer NO_DISK_SPACE being asked for new block with + priority 0 */ + return NO_DISK_SPACE; + +#ifdef CONFIG_REISERFS_CHECK + if (!s) + reiserfs_panic (s, "vs-4090: reiserfs_new_blocknrs: trying to get new block from nonexistent device"); + + if (search_start == MAX_B_NUM) + reiserfs_panic (s, "vs-4100: reiserfs_new_blocknrs: we are optimizing location based on " + "the bogus location of a temp buffer (%lu).", search_start); + + if (amount_needed < 1 || amount_needed > 2) + reiserfs_panic (s, "vs-4110: reiserfs_new_blocknrs: amount_needed parameter incorrect (%d)", amount_needed); +#endif /* CONFIG_REISERFS_CHECK */ + + /* We continue the while loop if another process snatches our found + * free block from us after we find it but before we successfully + * mark it as in use, or if we need to use sync to free up some + * blocks on the preserve list. */ + + while (amount_needed--) { + /* skip over any blocknrs already gotten last time. */ + if (*(free_blocknrs) != 0) { +#ifdef CONFIG_REISERFS_CHECK + if (is_reusable (s, *free_blocknrs, 1) == 0) + reiserfs_panic(s, "vs-4120: reiserfs_new_blocknrs: bad blocknr on free_blocknrs list"); +#endif /* CONFIG_REISERFS_CHECK */ + free_blocknrs++; + continue; + } + /* look for zero bits in bitmap */ + if (find_zero_bit_in_bitmap(s,search_start, &i, &j,for_unformatted) == 0) { + if (find_zero_bit_in_bitmap(s,search_start,&i,&j, for_unformatted) == 0) { + /* recode without the goto and without + the if. It will require a + duplicate for. This is worth the + code clarity. Your way was + admirable, and just a bit too + clever in saving instructions.:-) + I'd say create a new function, but + that would slow things also, yes? + -Hans */ +free_and_return: + for ( ; block_list_start != free_blocknrs; block_list_start++) { + reiserfs_free_block (th, *block_list_start); + *block_list_start = 0; + } + if (for_prealloc) + return NO_MORE_UNUSED_CONTIGUOUS_BLOCKS; + else + return NO_DISK_SPACE; + } + } + + /* i and j now contain the results of the search. i = bitmap block + number containing free block, j = offset in this block. we + compute the blocknr which is our result, store it in + free_blocknrs, and increment the pointer so that on the next + loop we will insert into the next location in the array. Also + in preparation for the next loop, search_start is changed so + that the next search will not rescan the same range but will + start where this search finished. Note that while it is + possible that schedule has occurred and blocks have been freed + in that range, it is perhaps more important that the blocks + returned be near each other than that they be near their other + neighbors, and it also simplifies and speeds the code this way. */ + + /* journal: we need to make sure the block we are giving out is not + ** a log block, horrible things would happen there. + */ + new_block = (i * (s->s_blocksize << 3)) + j; + if (for_prealloc && (new_block - 1) != search_start) { + /* preallocated blocks must be contiguous, bail if we didnt find one. + ** this is not a bug. We want to do the check here, before the + ** bitmap block is prepared, and before we set the bit and log the + ** bitmap. + ** + ** If we do the check after this function returns, we have to + ** call reiserfs_free_block for new_block, which would be pure + ** overhead. + ** + ** for_prealloc should only be set if the caller can deal with the + ** NO_MORE_UNUSED_CONTIGUOUS_BLOCKS return value. This can be + ** returned before the disk is actually full + */ + goto free_and_return ; + } + search_start = new_block ; + if (search_start >= reiserfs_get_journal_block(s) && + search_start < (reiserfs_get_journal_block(s) + JOURNAL_BLOCK_COUNT)) { + reiserfs_warning("vs-4130: reiserfs_new_blocknrs: trying to allocate log block %lu\n", + search_start) ; + search_start++ ; + amount_needed++ ; + continue ; + } + + + reiserfs_prepare_for_journal(s, SB_AP_BITMAP(s)[i], 1) ; + +#ifdef CONFIG_REISERFS_CHECK + if (buffer_locked (SB_AP_BITMAP (s)[i]) || is_reusable (s, search_start, 0) == 0) + reiserfs_panic (s, "vs-4140: reiserfs_new_blocknrs: bitmap block is locked or bad block number found"); +#endif + + /* if this bit was already set, we've scheduled, and someone else + ** has allocated it. loop around and try again + */ + if (reiserfs_test_and_set_le_bit (j, SB_AP_BITMAP (s)[i]->b_data)) { + reiserfs_warning("vs-4150: reiserfs_new_blocknrs, block not free"); + reiserfs_restore_prepared_buffer(s, SB_AP_BITMAP(s)[i]) ; + amount_needed++ ; + continue ; + } + journal_mark_dirty (th, s, SB_AP_BITMAP (s)[i]); + *free_blocknrs = search_start ; + free_blocknrs ++; + } + + reiserfs_prepare_for_journal(s, SB_BUFFER_WITH_SB(s), 1) ; + /* update free block count in super block */ + s->u.reiserfs_sb.s_rs->s_free_blocks = cpu_to_le32 (SB_FREE_BLOCKS (s) - init_amount_needed); + journal_mark_dirty (th, s, SB_BUFFER_WITH_SB (s)); + s->s_dirt = 1; + + return CARRY_ON; +} + +// this is called only by get_empty_nodes with for_preserve_list==0 +int reiserfs_new_blocknrs (struct reiserfs_transaction_handle *th, unsigned long * free_blocknrs, + unsigned long search_start, int amount_needed) { + return do_reiserfs_new_blocknrs(th, free_blocknrs, search_start, amount_needed, 0/*for_preserve_list-priority*/, 0/*for_formatted*/, 0/*for_prealloc */) ; +} + + +// called by get_new_buffer and by reiserfs_get_block with amount_needed == 1 and for_preserve_list == 0 +int reiserfs_new_unf_blocknrs(struct reiserfs_transaction_handle *th, unsigned long * free_blocknrs, + unsigned long search_start) { + return do_reiserfs_new_blocknrs(th, free_blocknrs, search_start, + 1/*amount_needed*/, + 0/*for_preserve_list-priority*/, + 1/*for formatted*/, + 0/*for prealloc */) ; +} + +#ifdef REISERFS_PREALLOCATE + +/* +** We pre-allocate 8 blocks. Pre-allocation is used for files > 16 KB only. +** This lowers fragmentation on large files by grabbing a contiguous set of +** blocks at once. It also limits the number of times the bitmap block is +** logged by making X number of allocation changes in a single transaction. +** +** We are using a border to divide the disk into two parts. The first part +** is used for tree blocks, which have a very high turnover rate (they +** are constantly allocated then freed) +** +** The second part of the disk is for the unformatted nodes of larger files. +** Putting them away from the tree blocks lowers fragmentation, and makes +** it easier to group files together. There are a number of different +** allocation schemes being tried right now, each is documented below. +** +** A great deal of the allocator's speed comes because reiserfs_get_block +** sends us the block number of the last unformatted node in the file. Once +** a given block is allocated past the border, we don't collide with the +** blocks near the search_start again. +** +*/ +int reiserfs_new_unf_blocknrs2 (struct reiserfs_transaction_handle *th, + struct inode * p_s_inode, + unsigned long * free_blocknrs, + unsigned long search_start) +{ + int ret=0, blks_gotten=0; + unsigned long border = 0; + unsigned long bstart = 0; + unsigned long hash_in, hash_out; + int allocated[PREALLOCATION_SIZE]; + int blks; + + if (!reiserfs_no_border(th->t_super)) { + /* we default to having the border at the 10% mark of the disk. This + ** is an arbitrary decision and it needs tuning. It also needs a limit + ** to prevent it from taking too much space on huge drives. + */ + bstart = (SB_BLOCK_COUNT(th->t_super) / 10); + } + if (!reiserfs_no_unhashed_relocation(th->t_super)) { + /* this is a very simple first attempt at preventing too much grouping + ** around the border value. Since k_dir_id is never larger than the + ** highest allocated oid, it is far from perfect, and files will tend + ** to be grouped towards the start of the border + */ + border = (INODE_PKEY(p_s_inode)->k_dir_id) % (SB_BLOCK_COUNT(th->t_super) - bstart - 1) ; + } else { + /* why would we want to delcare a local variable to this if statement + ** name border????? -chris + ** unsigned long border = 0; + */ + if (!reiserfs_hashed_relocation(th->t_super)) { + hash_in = (INODE_PKEY(p_s_inode))->k_dir_id; + /* I wonder if the CPU cost of the + hash will obscure the layout + effect? Of course, whether that + effect is good or bad we don't + know.... :-) */ + + hash_out = keyed_hash(((char *) (&hash_in)), 4); + border = hash_out % (SB_BLOCK_COUNT(th->t_super) - bstart - 1) ; + } + } + border += bstart ; + allocated[0] = 0 ; /* important. Allows a check later on to see if at + * least one block was allocated. This prevents false + * no disk space returns + */ + + if ( (p_s_inode->i_size < 4 * 4096) || + !(S_ISREG(p_s_inode->i_mode)) ) + { + if ( search_start < border + || ( + /* allow us to test whether it is a + good idea to prevent files from + getting too far away from their + packing locality by some unexpected + means. This might be poor code for + directories whose files total + larger than 1/10th of the disk, and + it might be good code for + suffering from old insertions when the disk + was almost full. */ + /* changed from !reiserfs_test3(th->t_super), which doesn't + ** seem like a good idea. Think about adding blocks to + ** a large file. If you've allocated 10% of the disk + ** in contiguous blocks, you start over at the border value + ** for every new allocation. This throws away all the + ** information sent in about the last block that was allocated + ** in the file. Not a good general case at all. + ** -chris + */ + reiserfs_test4(th->t_super) && + (search_start > border + (SB_BLOCK_COUNT(th->t_super) / 10)) + ) + ) + search_start=border; + + ret = do_reiserfs_new_blocknrs(th, free_blocknrs, search_start, + 1/*amount_needed*/, + 0/*use reserved blocks for root */, + 1/*for_formatted*/, + 0/*for prealloc */) ; + return ret; + } + + /* take a block off the prealloc list and return it -Hans */ + if (p_s_inode->u.reiserfs_i.i_prealloc_count > 0) { + p_s_inode->u.reiserfs_i.i_prealloc_count--; + *free_blocknrs = p_s_inode->u.reiserfs_i.i_prealloc_block++; + return ret; + } + + /* else get a new preallocation for the file */ + reiserfs_discard_prealloc (th, p_s_inode); + /* this uses the last preallocated block as the search_start. discard + ** prealloc does not zero out this number. + */ + if (search_start <= p_s_inode->u.reiserfs_i.i_prealloc_block) { + search_start = p_s_inode->u.reiserfs_i.i_prealloc_block; + } + + /* doing the compare again forces search_start to be >= the border, + ** even if the file already had prealloction done. This seems extra, + ** and should probably be removed + */ + if ( search_start < border ) search_start=border; + + *free_blocknrs = 0; + blks = PREALLOCATION_SIZE-1; + for (blks_gotten=0; blks_gotten<PREALLOCATION_SIZE; blks_gotten++) { + ret = do_reiserfs_new_blocknrs(th, free_blocknrs, search_start, + 1/*amount_needed*/, + 0/*for root reserved*/, + 1/*for_formatted*/, + (blks_gotten > 0)/*must_be_contiguous*/) ; + /* if we didn't find a block this time, adjust blks to reflect + ** the actual number of blocks allocated + */ + if (ret != CARRY_ON) { + blks = blks_gotten > 0 ? (blks_gotten - 1) : 0 ; + break ; + } + allocated[blks_gotten]= *free_blocknrs; +#ifdef CONFIG_REISERFS_CHECK + if ( (blks_gotten>0) && (allocated[blks_gotten] - allocated[blks_gotten-1]) != 1 ) { + /* this should be caught by new_blocknrs now, checking code */ + reiserfs_warning("yura-1, reiserfs_new_unf_blocknrs2: pre-allocated not contiguous set of blocks!\n") ; + reiserfs_free_block(th, allocated[blks_gotten]); + blks = blks_gotten-1; + break; + } +#endif + if (blks_gotten==0) { + p_s_inode->u.reiserfs_i.i_prealloc_block = *free_blocknrs; + } + search_start = *free_blocknrs; + *free_blocknrs = 0; + } + p_s_inode->u.reiserfs_i.i_prealloc_count = blks; + *free_blocknrs = p_s_inode->u.reiserfs_i.i_prealloc_block; + p_s_inode->u.reiserfs_i.i_prealloc_block++; + + /* we did actually manage to get 1 block */ + if (ret != CARRY_ON && allocated[0] > 0) { + return CARRY_ON ; + } + /* NO_MORE_UNUSED_CONTIGUOUS_BLOCKS should only mean something to + ** the preallocation code. The rest of the filesystem asks for a block + ** and should either get it, or know the disk is full. The code + ** above should never allow ret == NO_MORE_UNUSED_CONTIGUOUS_BLOCK, + ** as it doesn't send for_prealloc = 1 to do_reiserfs_new_blocknrs + ** unless it has already successfully allocated at least one block. + ** Just in case, we translate into a return value the rest of the + ** filesystem can understand. + ** + ** It is an error to change this without making the + ** rest of the filesystem understand NO_MORE_UNUSED_CONTIGUOUS_BLOCKS + ** If you consider it a bug to return NO_DISK_SPACE here, fix the rest + ** of the fs first. + */ + if (ret == NO_MORE_UNUSED_CONTIGUOUS_BLOCKS) { +#ifdef CONFIG_REISERFS_CHECK + reiserfs_warning("reiser-2015: this shouldn't happen, may cause false out of disk space error"); +#endif + return NO_DISK_SPACE; + } + return ret; +} + +// +// a portion of this function, was derived from minix or ext2's +// analog. You should be able to tell which portion by looking at the +// ext2 code and comparing. + +void reiserfs_discard_prealloc (struct reiserfs_transaction_handle *th, + struct inode * inode) +{ + if (inode->u.reiserfs_i.i_prealloc_count > 0) { + while (inode->u.reiserfs_i.i_prealloc_count--) { + reiserfs_free_block(th,inode->u.reiserfs_i.i_prealloc_block); + inode->u.reiserfs_i.i_prealloc_block++; + } + } + inode->u.reiserfs_i.i_prealloc_count = 0; +} +#endif diff --git a/fs/reiserfs/buffer2.c b/fs/reiserfs/buffer2.c new file mode 100644 index 000000000..c05854ce5 --- /dev/null +++ b/fs/reiserfs/buffer2.c @@ -0,0 +1,358 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + + +/* + * Contains code from + * + * linux/include/linux/lock.h and linux/fs/buffer.c /linux/fs/minix/fsync.c + * + * Copyright (C) 1991, 1992 Linus Torvalds + */ +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/sched.h> +#include <linux/locks.h> +#include <linux/reiserfs_fs.h> +#include <linux/smp_lock.h> + +#else + +#include "nokernel.h" + +#endif + + +/* + * wait_buffer_until_released + * reiserfs_bread + * reiserfs_getblk + * get_new_buffer + */ + + + +/* when we allocate a new block (get_new_buffer, get_empty_nodes) and + get buffer for it, it is possible that it is held by someone else + or even by this process. In this function we wait until all other + holders release buffer. To make sure, that current process does not + hold we did free all buffers in tree balance structure + (get_empty_nodes and get_nodes_for_preserving) or in path structure + only (get_new_buffer) just before calling this */ +void wait_buffer_until_released (struct buffer_head * bh) +{ + int repeat_counter = 0; + + while (atomic_read (&(bh->b_count)) > 1) { + + if ( !(++repeat_counter % 30000000) ) { + reiserfs_warning ("vs-3050: wait_buffer_until_released: nobody releases buffer (%b). Still waiting (%d) %cJDIRTY %cJWAIT\n", + bh, repeat_counter, buffer_journaled(bh) ? ' ' : '!', + buffer_journal_dirty(bh) ? ' ' : '!'); + } + run_task_queue(&tq_disk); + current->policy |= SCHED_YIELD; + /*current->counter = 0;*/ + schedule(); + } + if (repeat_counter > 30000000) { + reiserfs_warning("vs-3051: done waiting, ignore vs-3050 messages for (%b)\n", bh) ; + } +} + +/* + * reiserfs_bread() reads a specified block and returns the buffer that contains + * it. It returns NULL if the block was unreadable. + */ +/* It first tries to find the block in cache, and if it cannot do so + then it creates a new buffer and schedules I/O to read the + block. */ +/* The function is NOT SCHEDULE-SAFE! */ + +struct buffer_head * reiserfs_bread (kdev_t n_dev, int n_block, int n_size) +{ + return bread (n_dev, n_block, n_size); +} + +/* This function looks for a buffer which contains a given block. If + the block is in cache it returns it, otherwise it returns a new + buffer which is not uptodate. This is called by reiserfs_bread and + other functions. Note that get_new_buffer ought to be called this + and this ought to be called get_new_buffer, since this doesn't + actually get the block off of the disk. */ +/* The function is NOT SCHEDULE-SAFE! */ + +struct buffer_head * reiserfs_getblk (kdev_t n_dev, int n_block, int n_size) +{ + return getblk (n_dev, n_block, n_size); +} + +#ifdef NEW_GET_NEW_BUFFER + +/* returns one buffer with a blocknr near blocknr. */ +static int get_new_buffer_near_blocknr( + struct super_block * p_s_sb, + int blocknr, + struct buffer_head ** pp_s_new_bh, + struct path * p_s_path + ) { + unsigned long n_new_blocknumber = 0; + int n_ret_value, + n_repeat = CARRY_ON; + +#ifdef CONFIG_REISERFS_CHECK + int repeat_counter = 0; + + if (!blocknr) + printk ("blocknr passed to get_new_buffer_near_blocknr was 0"); +#endif + + + if ( (n_ret_value = reiserfs_new_blocknrs (p_s_sb, &n_new_blocknumber, + blocknr, 1)) == NO_DISK_SPACE ) + return NO_DISK_SPACE; + + *pp_s_new_bh = reiserfs_getblk(p_s_sb->s_dev, n_new_blocknumber, p_s_sb->s_blocksize); + if ( buffer_uptodate(*pp_s_new_bh) ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( buffer_dirty(*pp_s_new_bh) || (*pp_s_new_bh)->b_dev == NODEV ) { + reiserfs_panic(p_s_sb, "PAP-14080: get_new_buffer: invalid uptodate buffer %b for the new block", *pp_s_new_bh); + } +#endif + + /* Free path buffers to prevent deadlock. */ + /* It is possible that this process has the buffer, which this function is getting, already in + its path, and is responsible for double incrementing the value of b_count. If we recalculate + the path after schedule we can avoid risking an endless loop. This problematic situation is + possible in a multiple processing environment. Suppose process 1 has acquired a path P; then + process 2 balanced and remove block A from the tree. Process 1 continues and runs + get_new_buffer, that returns buffer with block A. If node A was on the path P, then it will + have b_count == 2. If we now will simply wait in while ( (*pp_s_new_bh)->b_count > 1 ) we get + into an endless loop, as nobody will release this buffer and the current process holds buffer + twice. That is why we do decrement_counters_in_path(p_s_path) before waiting until b_count + becomes 1. (it there were other processes holding node A, then eventually we will get a + moment, when all of them released a buffer). */ + if ( atomic_read (&((*pp_s_new_bh)->b_count)) > 1 ) { + decrement_counters_in_path(p_s_path); + n_ret_value |= SCHEDULE_OCCURRED; + } + + while ( atomic_read (&((*pp_s_new_bh)->b_count)) > 1 ) { + +#ifdef REISERFS_INFO + printk("get_new_buffer() calls schedule to decrement b_count\n"); +#endif + +#ifdef CONFIG_REISERFS_CHECK + if ( ! (++repeat_counter % 10000) ) + printk("get_new_buffer(%u): counter(%d) too big", current->pid, repeat_counter); +#endif + + current->counter = 0; + schedule(); + } + +#ifdef CONFIG_REISERFS_CHECK + if ( buffer_dirty(*pp_s_new_bh) || (*pp_s_new_bh)->b_dev == NODEV ) { + print_buffer_head(*pp_s_new_bh,"get_new_buffer"); + reiserfs_panic(p_s_sb, "PAP-14090: get_new_buffer: invalid uptodate buffer %b for the new block(case 2)", *pp_s_new_bh); + } +#endif + + } + else { + ; + +#ifdef CONFIG_REISERFS_CHECK + if (atomic_read (&((*pp_s_new_bh)->b_count)) != 1) { + reiserfs_panic(p_s_sb,"PAP-14100: get_new_buffer: not uptodate buffer %b for the new block has b_count more than one", + *pp_s_new_bh); + } +#endif + + } + return (n_ret_value | n_repeat); +} + + +/* returns the block number of the last unformatted node, assumes p_s_key_to_search.k_offset is a byte in the tail of + the file, Useful for when you want to append to a file, and convert a direct item into an unformatted node near the + last unformatted node of the file. Putting the unformatted node near the direct item is potentially very bad to do. + If there is no unformatted node in the file, then we return the block number of the direct item. */ +/* The function is NOT SCHEDULE-SAFE! */ +inline int get_last_unformatted_node_blocknr_of_file( struct key * p_s_key_to_search, struct super_block * p_s_sb, + struct buffer_head * p_s_bh + struct path * p_unf_search_path, struct inode * p_s_inode) + +{ + struct key unf_key_to_search; + struct item_head * p_s_ih; + int n_pos_in_item; + struct buffer_head * p_indirect_item_bh; + + copy_key(&unf_key_to_search,p_s_key_to_search); + unf_key_to_search.k_uniqueness = TYPE_INDIRECT; + unf_key_to_search.k_offset = p_s_inode->u.reiserfs_i.i_first_direct_byte - 1; + + /* p_s_key_to_search->k_offset - MAX_ITEM_LEN(p_s_sb->s_blocksize); */ + if (search_for_position_by_key (p_s_sb, &unf_key_to_search, p_unf_search_path, &n_pos_in_item) == POSITION_FOUND) + { + p_s_ih = B_N_PITEM_HEAD(p_indirect_item_bh = PATH_PLAST_BUFFER(p_unf_search_path), PATH_LAST_POSITION(p_unf_search_path)); + return (B_I_POS_UNFM_POINTER(p_indirect_item_bh, p_s_ih, n_pos_in_item)); + } + /* else */ + printk("reiser-1800: search for unformatted node failed, p_s_key_to_search->k_offset = %u, unf_key_to_search.k_offset = %u, MAX_ITEM_LEN(p_s_sb->s_blocksize) = %ld, debug this\n", p_s_key_to_search->k_offset, unf_key_to_search.k_offset, MAX_ITEM_LEN(p_s_sb->s_blocksize) ); + print_buffer_head(PATH_PLAST_BUFFER(p_unf_search_path), "the buffer holding the item before the key we failed to find"); + print_block_head(PATH_PLAST_BUFFER(p_unf_search_path), "the block head"); + return 0; /* keeps the compiler quiet */ +} + + + /* hasn't been out of disk space tested */ +/* The function is NOT SCHEDULE-SAFE! */ +static int get_buffer_near_last_unf ( struct super_block * p_s_sb, struct key * p_s_key_to_search, + struct inode * p_s_inode, struct buffer_head * p_s_bh, + struct buffer_head ** pp_s_un_bh, struct path * p_s_search_path) +{ + int unf_blocknr = 0, /* blocknr from which we start search for a free block for an unformatted node, if 0 + then we didn't find an unformatted node though we might have found a file hole */ + n_repeat = CARRY_ON; + struct key unf_key_to_search; + struct path unf_search_path; + + copy_key(&unf_key_to_search,p_s_key_to_search); + unf_key_to_search.k_uniqueness = TYPE_INDIRECT; + + if ( + (p_s_inode->u.reiserfs_i.i_first_direct_byte > 4095) /* i_first_direct_byte gets used for all sorts of + crap other than what the name indicates, thus + testing to see if it is 0 is not enough */ + && (p_s_inode->u.reiserfs_i.i_first_direct_byte < MAX_KEY_OFFSET) /* if there is no direct item then + i_first_direct_byte = MAX_KEY_OFFSET */ + ) + { + /* actually, we don't want the last unformatted node, we want the last unformatted node + which is before the current file offset */ + unf_key_to_search.k_offset = ((p_s_inode->u.reiserfs_i.i_first_direct_byte -1) < unf_key_to_search.k_offset) ? p_s_inode->u.reiserfs_i.i_first_direct_byte -1 : unf_key_to_search.k_offset; + + while (unf_key_to_search.k_offset > -1) + { + /* This is our poorly documented way of initializing paths. -Hans */ + init_path (&unf_search_path); + /* get the blocknr from which we start the search for a free block. */ + unf_blocknr = get_last_unformatted_node_blocknr_of_file( p_s_key_to_search, /* assumes this points to the file tail */ + p_s_sb, /* lets us figure out the block size */ + p_s_bh, /* if there is no unformatted node in the file, + then it returns p_s_bh->b_blocknr */ + &unf_search_path, + p_s_inode + ); +/* printk("in while loop: unf_blocknr = %d, *pp_s_un_bh = %p\n", unf_blocknr, *pp_s_un_bh); */ + if (unf_blocknr) + break; + else /* release the path and search again, this could be really slow for huge + holes.....better to spend the coding time adding compression though.... -Hans */ + { + /* Vladimir, is it a problem that I don't brelse these buffers ?-Hans */ + decrement_counters_in_path(&unf_search_path); + unf_key_to_search.k_offset -= 4096; + } + } + if (unf_blocknr) { + n_repeat |= get_new_buffer_near_blocknr(p_s_sb, unf_blocknr, pp_s_un_bh, p_s_search_path); + } + else { /* all unformatted nodes are holes */ + n_repeat |= get_new_buffer_near_blocknr(p_s_sb, p_s_bh->b_blocknr, pp_s_un_bh, p_s_search_path); + } + } + else { /* file has no unformatted nodes */ + n_repeat |= get_new_buffer_near_blocknr(p_s_sb, p_s_bh->b_blocknr, pp_s_un_bh, p_s_search_path); +/* printk("in else: unf_blocknr = %d, *pp_s_un_bh = %p\n", unf_blocknr, *pp_s_un_bh); */ +/* print_path (0, p_s_search_path); */ + } + + return n_repeat; +} + +#endif /* NEW_GET_NEW_BUFFER */ + + +#ifdef OLD_GET_NEW_BUFFER + +/* The function is NOT SCHEDULE-SAFE! */ +int get_new_buffer( + struct reiserfs_transaction_handle *th, + struct buffer_head * p_s_bh, + struct buffer_head ** pp_s_new_bh, + struct path * p_s_path + ) { + unsigned long n_new_blocknumber = 0; + int n_repeat; + struct super_block * p_s_sb = th->t_super; + + if ( (n_repeat = reiserfs_new_unf_blocknrs (th, &n_new_blocknumber, p_s_bh->b_blocknr)) == NO_DISK_SPACE ) + return NO_DISK_SPACE; + + *pp_s_new_bh = reiserfs_getblk(p_s_sb->s_dev, n_new_blocknumber, p_s_sb->s_blocksize); + if (atomic_read (&(*pp_s_new_bh)->b_count) > 1) { + /* Free path buffers to prevent deadlock which can occur in the + situation like : this process holds p_s_path; Block + (*pp_s_new_bh)->b_blocknr is on the path p_s_path, but it is + not necessary, that *pp_s_new_bh is in the tree; process 2 + could remove it from the tree and freed block + (*pp_s_new_bh)->b_blocknr. Reiserfs_new_blocknrs in above + returns block (*pp_s_new_bh)->b_blocknr. Reiserfs_getblk gets + buffer for it, and it has b_count > 1. If we now will simply + wait in while ( (*pp_s_new_bh)->b_count > 1 ) we get into an + endless loop, as nobody will release this buffer and the + current process holds buffer twice. That is why we do + decrement_counters_in_path(p_s_path) before waiting until + b_count becomes 1. (it there were other processes holding node + pp_s_new_bh, then eventually we will get a moment, when all of + them released a buffer). */ + decrement_counters_in_path(p_s_path); + wait_buffer_until_released (*pp_s_new_bh); + n_repeat |= SCHEDULE_OCCURRED; + } + +#ifdef CONFIG_REISERFS_CHECK + if ( atomic_read (&((*pp_s_new_bh)->b_count)) != 1 || buffer_dirty (*pp_s_new_bh)) { + reiserfs_panic(p_s_sb,"PAP-14100: get_new_buffer: not free or dirty buffer %b for the new block", + *pp_s_new_bh); + } +#endif + + return n_repeat; +} + +#endif /* OLD_GET_NEW_BUFFER */ + + +#ifdef GET_MANY_BLOCKNRS + /* code not yet functional */ +get_next_blocknr ( + unsigned long * p_blocknr_array, /* we get a whole bunch of blocknrs all at once for + the write. This is better than getting them one at + a time. */ + unsigned long ** p_blocknr_index, /* pointer to current offset into the array. */ + unsigned long blocknr_array_length +) +{ + unsigned long return_value; + + if (*p_blocknr_index < p_blocknr_array + blocknr_array_length) { + return_value = **p_blocknr_index; + **p_blocknr_index = 0; + *p_blocknr_index++; + return (return_value); + } + else + { + kfree (p_blocknr_array); + } +} +#endif /* GET_MANY_BLOCKNRS */ + diff --git a/fs/reiserfs/dir.c b/fs/reiserfs/dir.c new file mode 100644 index 000000000..48ca45c0c --- /dev/null +++ b/fs/reiserfs/dir.c @@ -0,0 +1,249 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/string.h> +#include <linux/errno.h> +#include <linux/fs.h> +#include <linux/reiserfs_fs.h> +#include <linux/stat.h> +#include <linux/smp_lock.h> +#include <asm/uaccess.h> + +#else + +#include "nokernel.h" + +#endif + +extern struct key MIN_KEY; + +static int reiserfs_readdir (struct file *, void *, filldir_t); +int reiserfs_dir_fsync(struct file *filp, struct dentry *dentry, int datasync) ; + +struct file_operations reiserfs_dir_operations = { + read: generic_read_dir, + readdir: reiserfs_readdir, + fsync: reiserfs_dir_fsync, +}; + +/* + * directories can handle most operations... + */ +struct inode_operations reiserfs_dir_inode_operations = { + //&reiserfs_dir_operations, /* default_file_ops */ + create: reiserfs_create, + lookup: reiserfs_lookup, + link: reiserfs_link, + unlink: reiserfs_unlink, + symlink: reiserfs_symlink, + mkdir: reiserfs_mkdir, + rmdir: reiserfs_rmdir, + mknod: reiserfs_mknod, + rename: reiserfs_rename, +}; + +int reiserfs_dir_fsync(struct file *filp, struct dentry *dentry, int datasync) { + int ret = 0 ; + int windex ; + struct reiserfs_transaction_handle th ; + + journal_begin(&th, dentry->d_inode->i_sb, 1) ; + windex = push_journal_writer("dir_fsync") ; + reiserfs_prepare_for_journal(th.t_super, SB_BUFFER_WITH_SB(th.t_super), 1) ; + journal_mark_dirty(&th, dentry->d_inode->i_sb, SB_BUFFER_WITH_SB (dentry->d_inode->i_sb)) ; + pop_journal_writer(windex) ; + journal_end_sync(&th, dentry->d_inode->i_sb, 1) ; + + return ret ; +} + + +#define store_ih(where,what) copy_item_head (where, what) + +// +static int reiserfs_readdir (struct file * filp, void * dirent, filldir_t filldir) +{ + struct inode *inode = filp->f_dentry->d_inode; + struct cpu_key pos_key; /* key of current position in the directory (key of directory entry) */ + INITIALIZE_PATH (path_to_entry); + struct buffer_head * bh; + int item_num, entry_num; + struct key * rkey; + struct item_head * ih, tmp_ih; + int search_res; + char * local_buf; + loff_t next_pos; + char small_buf[32] ; /* avoid kmalloc if we can */ + struct reiserfs_dir_entry de; + + + reiserfs_check_lock_depth("readdir") ; + + /* form key for search the next directory entry using f_pos field of + file structure */ + make_cpu_key (&pos_key, inode, (filp->f_pos) ? (filp->f_pos) : DOT_OFFSET, + TYPE_DIRENTRY, 3); + next_pos = cpu_key_k_offset (&pos_key); + + /* reiserfs_warning ("reiserfs_readdir 1: f_pos = %Ld\n", filp->f_pos);*/ + + while (1) { + research: + /* search the directory item, containing entry with specified key */ + search_res = search_by_entry_key (inode->i_sb, &pos_key, &path_to_entry, &de); + if (search_res == IO_ERROR) { + // FIXME: we could just skip part of directory which could + // not be read + return -EIO; + } + entry_num = de.de_entry_num; + bh = de.de_bh; + item_num = de.de_item_num; + ih = de.de_ih; + store_ih (&tmp_ih, ih); + +#ifdef CONFIG_REISERFS_CHECK + /* we must have found item, that is item of this directory, */ + if (COMP_SHORT_KEYS (&(ih->ih_key), &pos_key)) + reiserfs_panic (inode->i_sb, "vs-9000: reiserfs_readdir: " + "found item %h does not match to dir we readdir %k", + ih, &pos_key); + + if (item_num > B_NR_ITEMS (bh) - 1) + reiserfs_panic (inode->i_sb, "vs-9005: reiserfs_readdir: " + "item_num == %d, item amount == %d", + item_num, B_NR_ITEMS (bh)); + + /* and entry must be not more than number of entries in the item */ + if (I_ENTRY_COUNT (ih) < entry_num) + reiserfs_panic (inode->i_sb, "vs-9010: reiserfs_readdir: " + "entry number is too big %d (%d)", + entry_num, I_ENTRY_COUNT (ih)); +#endif /* CONFIG_REISERFS_CHECK */ + + if (search_res == POSITION_FOUND || entry_num < I_ENTRY_COUNT (ih)) { + /* go through all entries in the directory item beginning from the entry, that has been found */ + struct reiserfs_de_head * deh = B_I_DEH (bh, ih) + entry_num; + + for (; entry_num < I_ENTRY_COUNT (ih); entry_num ++, deh ++) { + int d_reclen; + char * d_name; + off_t d_off; + ino_t d_ino; + + if (!de_visible (deh)) + /* it is hidden entry */ + continue; + d_reclen = entry_length (bh, ih, entry_num); + d_name = B_I_DEH_ENTRY_FILE_NAME (bh, ih, deh); + if (!d_name[d_reclen - 1]) + d_reclen = strlen (d_name); + + if (d_reclen > REISERFS_MAX_NAME_LEN(inode->i_sb->s_blocksize)){ + /* too big to send back to VFS */ + continue ; + } + d_off = deh_offset (deh); + filp->f_pos = d_off ; + d_ino = deh_objectid (deh); + if (d_reclen <= 32) { + local_buf = small_buf ; + } else { + local_buf = kmalloc(d_reclen, GFP_BUFFER) ; + if (!local_buf) { + pathrelse (&path_to_entry); + return -ENOMEM ; + } + if (item_moved (&tmp_ih, &path_to_entry)) { + kfree(local_buf) ; + goto research; + } + } + // Note, that we copy name to user space via temporary + // buffer (local_buf) because filldir will block if + // user space buffer is swapped out. At that time + // entry can move to somewhere else + memcpy (local_buf, d_name, d_reclen); + if (filldir (dirent, d_name, d_reclen, d_off, d_ino, + DT_UNKNOWN) < 0) { + if (local_buf != small_buf) { + kfree(local_buf) ; + } + goto end; + } + if (local_buf != small_buf) { + kfree(local_buf) ; + } + + // next entry should be looked for with such offset + next_pos = deh_offset (deh) + 1; + + if (item_moved (&tmp_ih, &path_to_entry)) { + reiserfs_warning ("vs-9020: reiserfs_readdir " + "things are moving under hands. Researching..\n"); + goto research; + } + } /* for */ + } + + if (item_num != B_NR_ITEMS (bh) - 1) + // end of directory has been reached + goto end; + + /* item we went through is last item of node. Using right + delimiting key check is it directory end */ + rkey = get_rkey (&path_to_entry, inode->i_sb); + if (! comp_le_keys (rkey, &MIN_KEY)) { +#ifdef CONFIG_REISERFS_CHECK + reiserfs_warning ("vs-9025: reiserfs_readdir:" + "get_rkey failed. Researching..\n"); +#endif + /* set pos_key to key, that is the smallest and greater + that key of the last entry in the item */ + set_cpu_key_k_offset (&pos_key, next_pos); + continue; + } + + if ( COMP_SHORT_KEYS (rkey, &pos_key)) { + // end of directory has been reached + goto end; + } + + /* directory continues in the right neighboring block */ + set_cpu_key_k_offset (&pos_key, le_key_k_offset (ITEM_VERSION_1, rkey)); + + } /* while */ + + + end: + // FIXME: ext2_readdir does not reset f_pos + filp->f_pos = next_pos; + pathrelse (&path_to_entry); + reiserfs_check_path(&path_to_entry) ; + return 0; +} + + + + + + + + + + + + + + + + + + + + + diff --git a/fs/reiserfs/do_balan.c b/fs/reiserfs/do_balan.c new file mode 100644 index 000000000..852135254 --- /dev/null +++ b/fs/reiserfs/do_balan.c @@ -0,0 +1,2043 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +/* Now we have all buffers that must be used in balancing of the tree */ +/* Further calculations can not cause schedule(), and thus the buffer */ +/* tree will be stable until the balancing will be finished */ +/* balance the tree according to the analysis made before, */ +/* and using buffers obtained after all above. */ + + +/** + ** balance_leaf_when_delete + ** balance_leaf + ** do_balance + ** + **/ + +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <asm/uaccess.h> +#include <linux/sched.h> +#include <linux/reiserfs_fs.h> + +#else + +#include "nokernel.h" + +#endif + + +#ifdef CONFIG_REISERFS_CHECK + +struct tree_balance * cur_tb = NULL; /* detects whether more than one + copy of tb exists as a means + of checking whether schedule + is interrupting do_balance */ +#endif + + +inline void do_balance_mark_leaf_dirty (struct tree_balance * tb, + struct buffer_head * bh, int flag) +{ + if (reiserfs_dont_log(tb->tb_sb)) { + if (!test_and_set_bit(BH_Dirty, &bh->b_state)) { + __mark_buffer_dirty(bh) ; + tb->need_balance_dirty = 1; + } + } else { + int windex = push_journal_writer("do_balance") ; + journal_mark_dirty(tb->transaction_handle, tb->transaction_handle->t_super, bh) ; + pop_journal_writer(windex) ; + } +} + +#define do_balance_mark_internal_dirty do_balance_mark_leaf_dirty +#define do_balance_mark_sb_dirty do_balance_mark_leaf_dirty + + +/* summary: + if deleting something ( tb->insert_size[0] < 0 ) + return(balance_leaf_when_delete()); (flag d handled here) + else + if lnum is larger than 0 we put items into the left node + if rnum is larger than 0 we put items into the right node + if snum1 is larger than 0 we put items into the new node s1 + if snum2 is larger than 0 we put items into the new node s2 +Note that all *num* count new items being created. + +It would be easier to read balance_leaf() if each of these summary +lines was a separate procedure rather than being inlined. I think +that there are many passages here and in balance_leaf_when_delete() in +which two calls to one procedure can replace two passages, and it +might save cache space and improve software maintenance costs to do so. + +Vladimir made the perceptive comment that we should offload most of +the decision making in this function into fix_nodes/check_balance, and +then create some sort of structure in tb that says what actions should +be performed by do_balance. + +-Hans */ + + + +/* Balance leaf node in case of delete or cut: insert_size[0] < 0 + * + * lnum, rnum can have values >= -1 + * -1 means that the neighbor must be joined with S + * 0 means that nothing should be done with the neighbor + * >0 means to shift entirely or partly the specified number of items to the neighbor + */ +static int balance_leaf_when_delete (struct tree_balance * tb, int flag) +{ + struct buffer_head * tbS0 = PATH_PLAST_BUFFER (tb->tb_path); + int item_pos = PATH_LAST_POSITION (tb->tb_path); + int pos_in_item = tb->tb_path->pos_in_item; + struct buffer_info bi; + int n; + struct item_head * ih; + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->FR[0] && B_LEVEL (tb->FR[0]) != DISK_LEAF_NODE_LEVEL + 1) + reiserfs_panic (tb->tb_sb, + "vs- 12000: balance_leaf_when_delete:level: wrong FR %z\n", tb->FR[0]); + if ( tb->blknum[0] > 1 ) + reiserfs_panic (tb->tb_sb, + "PAP-12005: balance_leaf_when_delete: " + "tb->blknum == %d, can not be > 1", tb->blknum[0]); + + if ( ! tb->blknum[0] && ! PATH_H_PPARENT(tb->tb_path, 0)) + reiserfs_panic (tb->tb_sb, "PAP-12010: balance_leaf_when_delete: tree can not be empty"); +#endif + + ih = B_N_PITEM_HEAD (tbS0, item_pos); + + /* Delete or truncate the item */ + + switch (flag) { + case M_DELETE: /* delete item in S[0] */ + +#ifdef CONFIG_REISERFS_CHECK + if (le16_to_cpu (ih->ih_item_len) + IH_SIZE != -tb->insert_size [0]) + reiserfs_panic (tb->tb_sb, "vs-12013: balance_leaf_when_delete: " + "mode Delete, insert size %d, ih to be deleted %h", ih); + +#if 0 /* rigth delim key not supported */ + if ( ! item_pos && (! tb->L[0] || COMP_KEYS(B_PRIGHT_DELIM_KEY(tb->L[0]), B_N_PKEY(tbS0, 0))) ) { + print_cur_tb ("12015"); + reiserfs_panic (tb->tb_sb, "PAP-12015: balance_leaf_when_delete: L0's rkey does not match to 1st key of S0: " + "rkey in L %k, first key in S0 %k, rkey in CFL %k", + tb->L[0] ? B_PRIGHT_DELIM_KEY(tb->L[0]) : 0, + B_N_PKEY(tbS0, 0), + tb->CFL[0] ? B_N_PDELIM_KEY(tb->CFL[0],tb->lkey[0]) : 0); + } +#endif + +#endif + + bi.tb = tb; + bi.bi_bh = tbS0; + bi.bi_parent = PATH_H_PPARENT (tb->tb_path, 0); + bi.bi_position = PATH_H_POSITION (tb->tb_path, 1); + leaf_delete_items (&bi, 0, item_pos, 1, -1); + + if ( ! item_pos && tb->CFL[0] ) { + if ( B_NR_ITEMS(tbS0) ) { + replace_key(tb, tb->CFL[0],tb->lkey[0],tbS0,0); +#if 0 /* right delim key support */ + copy_key(B_PRIGHT_DELIM_KEY(tb->L[0]), B_N_PKEY(tbS0, 0)); + reiserfs_mark_buffer_dirty (tb->L[0], 0); +#endif + } + else { + if ( ! PATH_H_POSITION (tb->tb_path, 1) ) + replace_key(tb, tb->CFL[0],tb->lkey[0],PATH_H_PPARENT(tb->tb_path, 0),0); +#if 0 /* right delim key support */ + copy_key(B_PRIGHT_DELIM_KEY(tb->L[0]), B_PRIGHT_DELIM_KEY(tbS0)); + reiserfs_mark_buffer_dirty (tb->L[0], 0); +#endif + } + } + +#ifdef CONFIG_REISERFS_CHECK +#if 0 + if (! item_pos && (!tb->CFL[0] || !tb->L[0])) +#endif + if (! item_pos && !tb->CFL[0]) + reiserfs_panic (tb->tb_sb, "PAP-12020: balance_leaf_when_delete: tb->CFL[0]==%p, tb->L[0]==%p", tb->CFL[0], tb->L[0]); +#endif + + break; + + case M_CUT: { /* cut item in S[0] */ + bi.tb = tb; + bi.bi_bh = tbS0; + bi.bi_parent = PATH_H_PPARENT (tb->tb_path, 0); + bi.bi_position = PATH_H_POSITION (tb->tb_path, 1); + if (is_direntry_le_ih (ih)) { + +#ifdef CONFIG_REISERFS_CHECK +#if 0 /* right delim key support */ + if ( ! item_pos && ! pos_in_item && (! tb->L[0] || COMP_KEYS(B_PRIGHT_DELIM_KEY(tb->L[0]), + B_N_PKEY(tbS0, 0))) ) + reiserfs_panic(tb->tb_sb, "PAP-12025: balance_leaf_when_delete: illegal right delimiting key"); +#endif +#endif + + /* UFS unlink semantics are such that you can only delete one directory entry at a time. */ + /* when we cut a directory tb->insert_size[0] means number of entries to be cut (always 1) */ + tb->insert_size[0] = -1; + leaf_cut_from_buffer (&bi, item_pos, pos_in_item, -tb->insert_size[0]); + +#ifdef CONFIG_REISERFS_CHECK + if (! item_pos && ! pos_in_item && ! tb->CFL[0]) + reiserfs_panic (tb->tb_sb, "PAP-12030: balance_leaf_when_delete: can not change delimiting key. CFL[0]=%p", tb->CFL[0]); +#endif /* CONFIG_REISERFS_CHECK */ + + if ( ! item_pos && ! pos_in_item && tb->CFL[0] ) { + replace_key(tb, tb->CFL[0],tb->lkey[0],tbS0,0); +#if 0/* right delim key support */ + copy_key(B_PRIGHT_DELIM_KEY(tb->L[0]), B_N_PKEY(tbS0, 0)); + reiserfs_mark_buffer_dirty (tb->L[0], 0); +#endif + } + } else { + leaf_cut_from_buffer (&bi, item_pos, pos_in_item, -tb->insert_size[0]); + +#ifdef CONFIG_REISERFS_CHECK + if (! ih->ih_item_len) + reiserfs_panic (tb->tb_sb, "PAP-12035: balance_leaf_when_delete: cut must leave non-zero dynamic length of item"); +#endif /* CONFIG_REISERFS_CHECK */ + } + break; + } + + default: + print_cur_tb ("12040"); + reiserfs_panic (tb->tb_sb, "PAP-12040: balance_leaf_when_delete: unexpectable mode: %s(%d)", + (flag == M_PASTE) ? "PASTE" : ((flag == M_INSERT) ? "INSERT" : "UNKNOWN"), flag); + } + + /* the rule is that no shifting occurs unless by shifting a node can be freed */ + n = B_NR_ITEMS(tbS0); + if ( tb->lnum[0] ) /* L[0] takes part in balancing */ + { + if ( tb->lnum[0] == -1 ) /* L[0] must be joined with S[0] */ + { + if ( tb->rnum[0] == -1 ) /* R[0] must be also joined with S[0] */ + { + if ( tb->FR[0] == PATH_H_PPARENT(tb->tb_path, 0) ) + { + /* all contents of all the 3 buffers will be in L[0] */ + if ( PATH_H_POSITION (tb->tb_path, 1) == 0 && 1 < B_NR_ITEMS(tb->FR[0]) ) + replace_key(tb, tb->CFL[0],tb->lkey[0],tb->FR[0],1); + + /* update right_delimiting_key field */ +#if 0 + copy_key (B_PRIGHT_DELIM_KEY (tb->L[0]), B_PRIGHT_DELIM_KEY (tb->R[0])); +#endif + leaf_move_items (LEAF_FROM_S_TO_L, tb, n, -1, 0); + leaf_move_items (LEAF_FROM_R_TO_L, tb, B_NR_ITEMS(tb->R[0]), -1, 0); + +#if 0/*preserve list*/ + preserve_invalidate(tb, tbS0, tb->L[0]); + preserve_invalidate(tb, tb->R[0], tb->L[0]); +#endif + reiserfs_invalidate_buffer (tb, tbS0); + reiserfs_invalidate_buffer (tb, tb->R[0]); + + return 0; + } + /* all contents of all the 3 buffers will be in R[0] */ + leaf_move_items (LEAF_FROM_S_TO_R, tb, n, -1, 0); + leaf_move_items (LEAF_FROM_L_TO_R, tb, B_NR_ITEMS(tb->L[0]), -1, 0); + + /* right_delimiting_key is correct in R[0] */ + replace_key(tb, tb->CFR[0],tb->rkey[0],tb->R[0],0); + +#if 0 + /* mark tb->R[0] as suspected recipient */ + preserve_invalidate(tb,tbS0, tb->R[0]); + preserve_invalidate(tb,tb->L[0], tb->R[0]); +#endif + reiserfs_invalidate_buffer (tb, tbS0); + reiserfs_invalidate_buffer (tb, tb->L[0]); + + return -1; + } + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->rnum[0] != 0 ) + reiserfs_panic (tb->tb_sb, "PAP-12045: balance_leaf_when_delete: " + "rnum must be 0 (%d)", tb->rnum[0]); +#endif /* CONFIG_REISERFS_CHECK */ + + /* all contents of L[0] and S[0] will be in L[0] */ + leaf_shift_left(tb, n, -1); + +#if 0/*preserve list*/ + preserve_invalidate(tb, tbS0, tb->L[0]); /* preserved, shifting */ +#endif + reiserfs_invalidate_buffer (tb, tbS0); + + return 0; + } + /* a part of contents of S[0] will be in L[0] and the rest part of S[0] will be in R[0] */ + +#ifdef CONFIG_REISERFS_CHECK + if (( tb->lnum[0] + tb->rnum[0] < n ) || ( tb->lnum[0] + tb->rnum[0] > n+1 )) + reiserfs_panic (tb->tb_sb, "PAP-12050: balance_leaf_when_delete: " + "rnum(%d) and lnum(%d) and item number in S[0] are not consistent", + tb->rnum[0], tb->lnum[0], n); + + if (( tb->lnum[0] + tb->rnum[0] == n ) && (tb->lbytes != -1 || tb->rbytes != -1)) + reiserfs_panic (tb->tb_sb, "PAP-12055: balance_leaf_when_delete: " + "bad rbytes (%d)/lbytes (%d) parameters when items are not split", + tb->rbytes, tb->lbytes); + if (( tb->lnum[0] + tb->rnum[0] == n + 1 ) && (tb->lbytes < 1 || tb->rbytes != -1)) + reiserfs_panic (tb->tb_sb, "PAP-12060: balance_leaf_when_delete: " + "bad rbytes (%d)/lbytes (%d) parameters when items are split", + tb->rbytes, tb->lbytes); +#endif + + leaf_shift_left (tb, tb->lnum[0], tb->lbytes); + leaf_shift_right(tb, tb->rnum[0], tb->rbytes); + +#if 0/*preserve list*/ + preserve_invalidate (tb, tbS0, tb->L[0]); + mark_suspected_recipient (tb->tb_sb, tb->R[0]); +#endif + reiserfs_invalidate_buffer (tb, tbS0); + + return 0; + } + + if ( tb->rnum[0] == -1 ) { + /* all contents of R[0] and S[0] will be in R[0] */ + leaf_shift_right(tb, n, -1); +#if 0/*preserve list*/ + preserve_invalidate(tb, tbS0, tb->R[0]); +#endif + reiserfs_invalidate_buffer (tb, tbS0); + return 0; + } + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->rnum[0] ) + reiserfs_panic (tb->tb_sb, "PAP-12065: balance_leaf_when_delete: " + "bad rnum parameter must be 0 (%d)", tb->rnum[0]); +#endif + + return 0; +} + + +static int balance_leaf (struct tree_balance * tb, + struct item_head * ih, /* item header of inserted item (this is on little endian) */ + const char * body, /* body of inserted item or bytes to paste */ + int flag, /* i - insert, d - delete, c - cut, p - paste + (see comment to do_balance) */ + struct item_head * insert_key, /* in our processing of one level we sometimes determine what + must be inserted into the next higher level. This insertion + consists of a key or two keys and their corresponding + pointers */ + struct buffer_head ** insert_ptr /* inserted node-ptrs for the next level */ + ) +{ + struct buffer_head * tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#if 0/*preserve list*/ + struct buffer_head * tbF0 = PATH_H_PPARENT (tb->tb_path, 0); + int S0_b_item_order = PATH_H_B_ITEM_ORDER (tb->tb_path, 0); +#endif + int item_pos = PATH_LAST_POSITION (tb->tb_path); /* index into the array of item headers in S[0] + of the affected item */ + struct buffer_info bi; + struct buffer_head *S_new[2]; /* new nodes allocated to hold what could not fit into S */ + int snum[2]; /* number of items that will be placed + into S_new (includes partially shifted + items) */ + int sbytes[2]; /* if an item is partially shifted into S_new then + if it is a directory item + it is the number of entries from the item that are shifted into S_new + else + it is the number of bytes from the item that are shifted into S_new + */ + int n, i; + int ret_val; + int pos_in_item; + int zeros_num; + +#if 0 + if (tb->insert_size [0] % 4) { + reiserfs_panic (tb->tb_sb, "balance_leaf: wrong insert_size %d", + tb->insert_size [0]); + } +#endif + /* Make balance in case insert_size[0] < 0 */ + if ( tb->insert_size[0] < 0 ) + return balance_leaf_when_delete (tb, flag); + + zeros_num = 0; + if (flag == M_INSERT && body == 0) + zeros_num = le16_to_cpu (ih->ih_item_len); + + pos_in_item = tb->tb_path->pos_in_item; + /* for indirect item pos_in_item is measured in unformatted node + pointers. Recalculate to bytes */ + if (flag != M_INSERT && is_indirect_le_ih (B_N_PITEM_HEAD (tbS0, item_pos))) + pos_in_item *= UNFM_P_SIZE; + + if ( tb->lnum[0] > 0 ) { + /* Shift lnum[0] items from S[0] to the left neighbor L[0] */ + if ( item_pos < tb->lnum[0] ) { + /* new item or it part falls to L[0], shift it too */ + n = B_NR_ITEMS(tb->L[0]); + + switch (flag) { + case M_INSERT: /* insert item into L[0] */ + + if ( item_pos == tb->lnum[0] - 1 && tb->lbytes != -1 ) { + /* part of new item falls into L[0] */ + int new_item_len; + int version; + +#ifdef CONFIG_REISERFS_CHECK + if (!is_direct_le_ih (ih)) + reiserfs_panic (tb->tb_sb, "PAP-12075: balance_leaf: " + "only direct inserted item can be broken. %h", ih); +#endif + ret_val = leaf_shift_left (tb, tb->lnum[0]-1, -1); + /* when reading the if conditions preceding the subsequent preserve_shifted + lines understand that their goal is to determine if all that we are + shifting is the new data being added */ +#if 0/*preserve list*/ + if (tb->lnum[0] - 1 > 0) { + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->L[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); + } +#endif + + /* Calculate item length to insert to S[0] */ + new_item_len = le16_to_cpu (ih->ih_item_len) - tb->lbytes; + /* Calculate and check item length to insert to L[0] */ + ih->ih_item_len -= new_item_len; + +#ifdef CONFIG_REISERFS_CHECK + if ( (int)(ih->ih_item_len) <= 0 ) + reiserfs_panic(tb->tb_sb, "PAP-12080: balance_leaf: " + "there is nothing to insert into L[0]: ih_item_len=%d", + (int)ih->ih_item_len); +#endif + + /* Insert new item into L[0] */ + bi.tb = tb; + bi.bi_bh = tb->L[0]; + bi.bi_parent = tb->FL[0]; + bi.bi_position = get_left_neighbor_position (tb, 0); + leaf_insert_into_buf (&bi, n + item_pos - ret_val, ih, body, + zeros_num > ih->ih_item_len ? ih->ih_item_len : zeros_num); + + version = ih_version (ih); + + /* Calculate key component, item length and body to insert into S[0] */ + set_le_key_k_offset (ih_version (ih), &(ih->ih_key), + le_key_k_offset (ih_version (ih), &(ih->ih_key)) + tb->lbytes); + ih->ih_item_len = cpu_to_le16 (new_item_len); + if ( tb->lbytes > zeros_num ) { + body += (tb->lbytes - zeros_num); + zeros_num = 0; + } + else + zeros_num -= tb->lbytes; + +#ifdef CONFIG_REISERFS_CHECK + if ( (int)(ih->ih_item_len) <= 0 ) + reiserfs_panic(tb->tb_sb, "PAP-12085: balance_leaf: " + "there is nothing to insert into S[0]: ih_item_len=%d", + (int)ih->ih_item_len); +#endif + } else { + /* new item in whole falls into L[0] */ + /* Shift lnum[0]-1 items to L[0] */ + ret_val = leaf_shift_left(tb, tb->lnum[0]-1, tb->lbytes); +#if 0/*preserve list*/ + if (tb->lnum[0] > 1) { + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->L[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); + } +#endif + /* Insert new item into L[0] */ + bi.tb = tb; + bi.bi_bh = tb->L[0]; + bi.bi_parent = tb->FL[0]; + bi.bi_position = get_left_neighbor_position (tb, 0); + leaf_insert_into_buf (&bi, n + item_pos - ret_val, ih, body, zeros_num); +#if 0/*preserve list*/ + if (tb->preserve_mode == PRESERVE_INDIRECT_TO_DIRECT){ + mark_suspected_recipient (tb->tb_sb, bi.bi_bh); + } +#endif + tb->insert_size[0] = 0; + zeros_num = 0; + } + break; + + case M_PASTE: /* append item in L[0] */ + + if ( item_pos == tb->lnum[0] - 1 && tb->lbytes != -1 ) { + /* we must shift the part of the appended item */ + if ( is_direntry_le_ih (B_N_PITEM_HEAD (tbS0, item_pos))) { + +#ifdef CONFIG_REISERFS_CHECK + if ( zeros_num ) + reiserfs_panic(tb->tb_sb, "PAP-12090: balance_leaf: illegal parameter in case of a directory"); +#endif + + /* directory item */ + if ( tb->lbytes > pos_in_item ) { + /* new directory entry falls into L[0] */ + struct item_head * pasted; + int l_pos_in_item = pos_in_item; + + /* Shift lnum[0] - 1 items in whole. Shift lbytes - 1 entries from given directory item */ + ret_val = leaf_shift_left(tb, tb->lnum[0], tb->lbytes - 1); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->L[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + if ( ret_val && ! item_pos ) { + pasted = B_N_PITEM_HEAD(tb->L[0],B_NR_ITEMS(tb->L[0])-1); + l_pos_in_item += I_ENTRY_COUNT(pasted) - (tb->lbytes-1); + } + + /* Append given directory entry to directory item */ + bi.tb = tb; + bi.bi_bh = tb->L[0]; + bi.bi_parent = tb->FL[0]; + bi.bi_position = get_left_neighbor_position (tb, 0); + leaf_paste_in_buffer (&bi, n + item_pos - ret_val, l_pos_in_item, + tb->insert_size[0], body, zeros_num); + + /* previous string prepared space for pasting new entry, following string pastes this entry */ + + /* when we have merge directory item, pos_in_item has been changed too */ + + /* paste new directory entry. 1 is entry number */ + leaf_paste_entries (bi.bi_bh, n + item_pos - ret_val, l_pos_in_item, 1, + (struct reiserfs_de_head *)body, + body + DEH_SIZE, tb->insert_size[0] + ); + tb->insert_size[0] = 0; + } else { + /* new directory item doesn't fall into L[0] */ + /* Shift lnum[0]-1 items in whole. Shift lbytes directory entries from directory item number lnum[0] */ + leaf_shift_left (tb, tb->lnum[0], tb->lbytes); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->L[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + } + /* Calculate new position to append in item body */ + pos_in_item -= tb->lbytes; + } + else { + /* regular object */ + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->lbytes <= 0 ) + reiserfs_panic(tb->tb_sb, "PAP-12095: balance_leaf: " + "there is nothing to shift to L[0]. lbytes=%d", + tb->lbytes); + if ( pos_in_item != B_N_PITEM_HEAD(tbS0, item_pos)->ih_item_len ) + reiserfs_panic(tb->tb_sb, "PAP-12100: balance_leaf: " + "incorrect position to paste: item_len=%d, pos_in_item=%d", + B_N_PITEM_HEAD(tbS0,item_pos)->ih_item_len, pos_in_item); +#endif + + if ( tb->lbytes >= pos_in_item ) { + /* appended item will be in L[0] in whole */ + int l_n; + + /* this bytes number must be appended to the last item of L[h] */ + l_n = tb->lbytes - pos_in_item; + + /* Calculate new insert_size[0] */ + tb->insert_size[0] -= l_n; + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->insert_size[0] <= 0 ) + reiserfs_panic(tb->tb_sb, "PAP-12105: balance_leaf: " + "there is nothing to paste into L[0]. insert_size=%d", + tb->insert_size[0]); +#endif + + ret_val = leaf_shift_left(tb,tb->lnum[0], + B_N_PITEM_HEAD(tbS0,item_pos)->ih_item_len); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->L[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + /* Append to body of item in L[0] */ + bi.tb = tb; + bi.bi_bh = tb->L[0]; + bi.bi_parent = tb->FL[0]; + bi.bi_position = get_left_neighbor_position (tb, 0); + leaf_paste_in_buffer( + &bi,n + item_pos - ret_val, + B_N_PITEM_HEAD(tb->L[0],n+item_pos-ret_val)->ih_item_len, + l_n,body, zeros_num > l_n ? l_n : zeros_num + ); + +#ifdef CONFIG_REISERFS_CHECK + if (l_n && is_indirect_le_ih(B_N_PITEM_HEAD(tb->L[0], + n + item_pos - ret_val))) + reiserfs_panic(tb->tb_sb, "PAP-12110: balance_leaf: " + "pasting more than 1 unformatted node pointer into indirect item"); +#endif + + /* 0-th item in S0 can be only of DIRECT type when l_n != 0*/ + { + int version; + + version = le16_to_cpu (B_N_PITEM_HEAD (tbS0, 0)->ih_version); + set_le_key_k_offset (version, B_N_PKEY (tbS0, 0), + le_key_k_offset (version, B_N_PKEY (tbS0, 0)) + l_n); + set_le_key_k_offset (version, B_N_PDELIM_KEY(tb->CFL[0],tb->lkey[0]), + le_key_k_offset (version, B_N_PDELIM_KEY(tb->CFL[0],tb->lkey[0])) + l_n); + } +#if 0 + set_le_key_k_offset (B_PRIGHT_DELIM_KEY(tb->L[0]), le_key_k_offset (B_PRIGHT_DELIM_KEY(tb->L[0])) + l_n); +#endif + /* k_offset (B_N_PKEY (tbS0, 0)) += l_n; + k_offset (B_N_PDELIM_KEY(tb->CFL[0],tb->lkey[0])) += l_n; + k_offset (B_PRIGHT_DELIM_KEY(tb->L[0])) += l_n;*/ + +#ifdef NO_CONFIG_REISERFS_CHECK /* journal victim */ + if (!buffer_dirty (tbS0) || !buffer_dirty (tb->CFL[0]) || !buffer_dirty (tb->L[0])) + reiserfs_panic(tb->tb_sb, "PAP-12115: balance_leaf: L, CLF and S must be dirty already"); +#endif + + /* Calculate new body, position in item and insert_size[0] */ + if ( l_n > zeros_num ) { + body += (l_n - zeros_num); + zeros_num = 0; + } + else + zeros_num -= l_n; + pos_in_item = 0; + +#ifdef CONFIG_REISERFS_CHECK + if (comp_short_le_keys (B_N_PKEY(tbS0,0), + B_N_PKEY(tb->L[0],B_NR_ITEMS(tb->L[0])-1)) || + !op_is_left_mergeable (B_N_PKEY (tbS0, 0), tbS0->b_size) || + !op_is_left_mergeable(B_N_PDELIM_KEY(tb->CFL[0],tb->lkey[0]), tbS0->b_size)) + reiserfs_panic (tb->tb_sb, "PAP-12120: balance_leaf: " + "item must be merge-able with left neighboring item"); +#endif + + } + else /* only part of the appended item will be in L[0] */ + { + /* Calculate position in item for append in S[0] */ + pos_in_item -= tb->lbytes; + +#ifdef CONFIG_REISERFS_CHECK + if ( pos_in_item <= 0 ) + reiserfs_panic(tb->tb_sb, "PAP-12125: balance_leaf: " + "no place for paste. pos_in_item=%d", pos_in_item); +#endif + + /* Shift lnum[0] - 1 items in whole. Shift lbytes - 1 byte from item number lnum[0] */ + leaf_shift_left(tb,tb->lnum[0],tb->lbytes); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->L[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + } + } + } + else /* appended item will be in L[0] in whole */ + { + struct item_head * pasted; + +#ifdef REISERFS_FSCK + if ( ! item_pos && is_left_mergeable (tb->tb_sb, tb->tb_path) == 1 ) +#else + if ( ! item_pos && op_is_left_mergeable (B_N_PKEY (tbS0, 0), tbS0->b_size) ) +#endif + { /* if we paste into first item of S[0] and it is left mergable */ + /* then increment pos_in_item by the size of the last item in L[0] */ + pasted = B_N_PITEM_HEAD(tb->L[0],n-1); + if ( is_direntry_le_ih (pasted) ) + pos_in_item += le16_to_cpu (pasted->u.ih_entry_count); + else + pos_in_item += le16_to_cpu (pasted->ih_item_len); + } + + /* Shift lnum[0] - 1 items in whole. Shift lbytes - 1 byte from item number lnum[0] */ + ret_val = leaf_shift_left(tb,tb->lnum[0],tb->lbytes); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->L[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + /* Append to body of item in L[0] */ + bi.tb = tb; + bi.bi_bh = tb->L[0]; + bi.bi_parent = tb->FL[0]; + bi.bi_position = get_left_neighbor_position (tb, 0); + leaf_paste_in_buffer (&bi, n + item_pos - ret_val, pos_in_item, tb->insert_size[0], + body, zeros_num); + + /* if appended item is directory, paste entry */ + pasted = B_N_PITEM_HEAD (tb->L[0], n + item_pos - ret_val); + if (is_direntry_le_ih (pasted)) + leaf_paste_entries ( + bi.bi_bh, n + item_pos - ret_val, pos_in_item, 1, + (struct reiserfs_de_head *)body, body + DEH_SIZE, tb->insert_size[0] + ); + /* if appended item is indirect item, put unformatted node into un list */ + if (is_indirect_le_ih (pasted)) + set_ih_free_space (pasted, ((struct unfm_nodeinfo*)body)->unfm_freespace); + tb->insert_size[0] = 0; + zeros_num = 0; + } + break; + default: /* cases d and t */ + reiserfs_panic (tb->tb_sb, "PAP-12130: balance_leaf: lnum > 0: unexpectable mode: %s(%d)", + (flag == M_DELETE) ? "DELETE" : ((flag == M_CUT) ? "CUT" : "UNKNOWN"), flag); + } + } else { + /* new item doesn't fall into L[0] */ + leaf_shift_left(tb,tb->lnum[0],tb->lbytes); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->L[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + } + } /* tb->lnum[0] > 0 */ + + /* Calculate new item position */ + item_pos -= ( tb->lnum[0] - (( tb->lbytes != -1 ) ? 1 : 0)); + + if ( tb->rnum[0] > 0 ) { + /* shift rnum[0] items from S[0] to the right neighbor R[0] */ + n = B_NR_ITEMS(tbS0); + switch ( flag ) { + + case M_INSERT: /* insert item */ + if ( n - tb->rnum[0] < item_pos ) + { /* new item or its part falls to R[0] */ + if ( item_pos == n - tb->rnum[0] + 1 && tb->rbytes != -1 ) + { /* part of new item falls into R[0] */ + int old_key_comp, old_len, r_zeros_number; + const char * r_body; + int version; + loff_t offset; + +#ifdef CONFIG_REISERFS_CHECK + if ( !is_direct_le_ih (ih) ) + reiserfs_panic(tb->tb_sb, "PAP-12135: balance_leaf: " + "only direct item can be split. (%h)", ih); +#endif + + leaf_shift_right(tb,tb->rnum[0]-1,-1); +#if 0/*preserve list*/ + if (tb->rnum[0]>1) { + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->R[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); + } +#endif + + version = le16_to_cpu (ih->ih_version); + /* Remember key component and item length */ + old_key_comp = le_key_k_offset (version, &(ih->ih_key)); + old_len = le16_to_cpu (ih->ih_item_len); + + /* Calculate key component and item length to insert into R[0] */ + offset = le_key_k_offset (version, &(ih->ih_key)) + (old_len - tb->rbytes); + set_le_key_k_offset (version, &(ih->ih_key), offset); + ih->ih_item_len = cpu_to_le16 (tb->rbytes); + /* Insert part of the item into R[0] */ + bi.tb = tb; + bi.bi_bh = tb->R[0]; + bi.bi_parent = tb->FR[0]; + bi.bi_position = get_right_neighbor_position (tb, 0); + if ( offset - old_key_comp > zeros_num ) { + r_zeros_number = 0; + r_body = body + offset - old_key_comp - zeros_num; + } + else { + r_body = body; + r_zeros_number = zeros_num - (offset - old_key_comp); + zeros_num -= r_zeros_number; + } + + leaf_insert_into_buf (&bi, 0, ih, r_body, r_zeros_number); + + /* Replace right delimiting key by first key in R[0] */ + replace_key(tb, tb->CFR[0],tb->rkey[0],tb->R[0],0); + + /* Calculate key component and item length to insert into S[0] */ + set_le_key_k_offset (version, &(ih->ih_key), old_key_comp); + ih->ih_item_len = cpu_to_le16 (old_len - tb->rbytes); + + tb->insert_size[0] -= tb->rbytes; + + } + else /* whole new item falls into R[0] */ + { + /* Shift rnum[0]-1 items to R[0] */ + ret_val = leaf_shift_right(tb,tb->rnum[0]-1,tb->rbytes); +#if 0/*preserve list*/ + if (tb->rnum[0]>1) { + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->R[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); + } +#endif + /* Insert new item into R[0] */ + bi.tb = tb; + bi.bi_bh = tb->R[0]; + bi.bi_parent = tb->FR[0]; + bi.bi_position = get_right_neighbor_position (tb, 0); + leaf_insert_into_buf (&bi, item_pos - n + tb->rnum[0] - 1, ih, body, zeros_num); +#if 0/*preserve list*/ + if (tb->preserve_mode == PRESERVE_INDIRECT_TO_DIRECT){ + mark_suspected_recipient (tb->tb_sb, bi.bi_bh); + } +#endif + + /* If we insert new item in the begin of R[0] change the right delimiting key */ + if ( item_pos - n + tb->rnum[0] - 1 == 0 ) { + replace_key(tb, tb->CFR[0],tb->rkey[0],tb->R[0],0); + +#if 0 + /* update right delimiting key */ + copy_key(B_PRIGHT_DELIM_KEY(tbS0), &(ih->ih_key)); + reiserfs_mark_buffer_dirty (tbS0, 0); +#endif + } + zeros_num = tb->insert_size[0] = 0; + } + } + else /* new item or part of it doesn't fall into R[0] */ + { + leaf_shift_right(tb,tb->rnum[0],tb->rbytes); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->R[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + } + break; + + case M_PASTE: /* append item */ + + if ( n - tb->rnum[0] <= item_pos ) /* pasted item or part of it falls to R[0] */ + { + if ( item_pos == n - tb->rnum[0] && tb->rbytes != -1 ) + { /* we must shift the part of the appended item */ + if ( is_direntry_le_ih (B_N_PITEM_HEAD(tbS0, item_pos))) + { /* we append to directory item */ + int entry_count; + +#ifdef CONFIG_REISERFS_CHECK + if ( zeros_num ) + reiserfs_panic(tb->tb_sb, "PAP-12145: balance_leaf: illegal parametr in case of a directory"); +#endif + + entry_count = I_ENTRY_COUNT(B_N_PITEM_HEAD(tbS0, item_pos)); + if ( entry_count - tb->rbytes < pos_in_item ) + /* new directory entry falls into R[0] */ + { + int paste_entry_position; + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->rbytes - 1 >= entry_count || ! tb->insert_size[0] ) + reiserfs_panic(tb->tb_sb, "PAP-12150: balance_leaf: " + "no enough of entries to shift to R[0]: rbytes=%d, entry_count=%d", + tb->rbytes, entry_count); +#endif + + /* Shift rnum[0]-1 items in whole. Shift rbytes-1 directory entries from directory item number rnum[0] */ + leaf_shift_right(tb,tb->rnum[0],tb->rbytes - 1); +#if 0/*preserve list*/ + /* if we are shifting more than just the new entry */ + if (tb->rbytes > 1 || tb->rnum[0] > 1) { + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->R[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); + } +#endif + /* Paste given directory entry to directory item */ + paste_entry_position = pos_in_item - entry_count + tb->rbytes - 1; + bi.tb = tb; + bi.bi_bh = tb->R[0]; + bi.bi_parent = tb->FR[0]; + bi.bi_position = get_right_neighbor_position (tb, 0); + leaf_paste_in_buffer (&bi, 0, paste_entry_position, + tb->insert_size[0],body,zeros_num); + /* paste entry */ + leaf_paste_entries ( + bi.bi_bh, 0, paste_entry_position, 1, (struct reiserfs_de_head *)body, + body + DEH_SIZE, tb->insert_size[0] + ); + + if ( paste_entry_position == 0 ) { + /* change delimiting keys */ + replace_key(tb, tb->CFR[0],tb->rkey[0],tb->R[0],0); +#if 0 + copy_key(B_PRIGHT_DELIM_KEY(tbS0), B_N_PKEY(tb->R[0], 0)); + reiserfs_mark_buffer_dirty (tbS0, 0); +#endif + } + + tb->insert_size[0] = 0; + pos_in_item++; + } + else /* new directory entry doesn't fall into R[0] */ + { + leaf_shift_right(tb,tb->rnum[0],tb->rbytes); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->R[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + } + } + else /* regular object */ + { + int n_shift, n_rem, r_zeros_number; + const char * r_body; + + /* Calculate number of bytes which must be shifted from appended item */ + if ( (n_shift = tb->rbytes - tb->insert_size[0]) < 0 ) + n_shift = 0; + +#ifdef CONFIG_REISERFS_CHECK + if (pos_in_item != B_N_PITEM_HEAD (tbS0, item_pos)->ih_item_len) + reiserfs_panic(tb->tb_sb,"PAP-12155: balance_leaf: invalid position to paste. ih_item_len=%d, pos_in_item=%d", + pos_in_item, B_N_PITEM_HEAD(tbS0,item_pos)->ih_item_len); +#endif + + leaf_shift_right(tb,tb->rnum[0],n_shift); +#if 0/*preserve list*/ + /* if we are shifting an old part from the appended item or more than the appended item is going into R */ + if (n_shift || tb->rnum[0] > 1) { + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->R[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); + } +#endif + /* Calculate number of bytes which must remain in body after appending to R[0] */ + if ( (n_rem = tb->insert_size[0] - tb->rbytes) < 0 ) + n_rem = 0; + + { + int version; + + version = ih_version (B_N_PITEM_HEAD (tb->R[0],0)); + set_le_key_k_offset (version, B_N_PKEY(tb->R[0],0), + le_key_k_offset (version, B_N_PKEY(tb->R[0],0)) + n_rem); + set_le_key_k_offset (version, B_N_PDELIM_KEY(tb->CFR[0],tb->rkey[0]), + le_key_k_offset (version, B_N_PDELIM_KEY(tb->CFR[0],tb->rkey[0])) + n_rem); + } +/* k_offset (B_N_PKEY(tb->R[0],0)) += n_rem; + k_offset (B_N_PDELIM_KEY(tb->CFR[0],tb->rkey[0])) += n_rem;*/ + do_balance_mark_internal_dirty (tb, tb->CFR[0], 0); + +#if 0 + set_le_key_k_offset (B_PRIGHT_DELIM_KEY(tbS0), le_key_k_offset (B_PRIGHT_DELIM_KEY(tbS0)) + n_rem); +/* k_offset (B_PRIGHT_DELIM_KEY(tbS0)) += n_rem;*/ + reiserfs_mark_buffer_dirty (tbS0, 0); +#endif + /* Append part of body into R[0] */ + bi.tb = tb; + bi.bi_bh = tb->R[0]; + bi.bi_parent = tb->FR[0]; + bi.bi_position = get_right_neighbor_position (tb, 0); + if ( n_rem > zeros_num ) { + r_zeros_number = 0; + r_body = body + n_rem - zeros_num; + } + else { + r_body = body; + r_zeros_number = zeros_num - n_rem; + zeros_num -= r_zeros_number; + } + + leaf_paste_in_buffer(&bi, 0, n_shift, tb->insert_size[0] - n_rem, r_body, r_zeros_number); + + if (is_indirect_le_ih (B_N_PITEM_HEAD(tb->R[0],0))) { + +#ifdef CONFIG_REISERFS_CHECK + if (n_rem) + reiserfs_panic(tb->tb_sb, "PAP-12160: balance_leaf: paste more than one unformatted node pointer"); +#endif + + set_ih_free_space (B_N_PITEM_HEAD(tb->R[0],0), ((struct unfm_nodeinfo*)body)->unfm_freespace); + } + + tb->insert_size[0] = n_rem; + if ( ! n_rem ) + pos_in_item ++; + } + } + else /* pasted item in whole falls into R[0] */ + { + struct item_head * pasted; + + ret_val = leaf_shift_right(tb,tb->rnum[0],tb->rbytes); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->R[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + /* append item in R[0] */ + if ( pos_in_item >= 0 ) { + bi.tb = tb; + bi.bi_bh = tb->R[0]; + bi.bi_parent = tb->FR[0]; + bi.bi_position = get_right_neighbor_position (tb, 0); + leaf_paste_in_buffer(&bi,item_pos - n + tb->rnum[0], pos_in_item, + tb->insert_size[0],body, zeros_num); + } + + /* paste new entry, if item is directory item */ + pasted = B_N_PITEM_HEAD(tb->R[0], item_pos - n + tb->rnum[0]); + if (is_direntry_le_ih (pasted) && pos_in_item >= 0 ) { + leaf_paste_entries ( + bi.bi_bh, item_pos - n + tb->rnum[0], pos_in_item, 1, + (struct reiserfs_de_head *)body, body + DEH_SIZE, tb->insert_size[0] + ); + if ( ! pos_in_item ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( item_pos - n + tb->rnum[0] ) + reiserfs_panic (tb->tb_sb, "PAP-12165: balance_leaf: " + "directory item must be first item of node when pasting is in 0th position"); +#endif + + /* update delimiting keys */ + replace_key(tb, tb->CFR[0],tb->rkey[0],tb->R[0],0); +#if 0 + copy_key(B_PRIGHT_DELIM_KEY(tbS0),B_N_PKEY(tb->R[0], 0)); + reiserfs_mark_buffer_dirty (tbS0, 0); +#endif + } + } + + if (is_indirect_le_ih (pasted)) + set_ih_free_space (pasted, ((struct unfm_nodeinfo*)body)->unfm_freespace); + zeros_num = tb->insert_size[0] = 0; + } + } + else /* new item doesn't fall into R[0] */ + { + leaf_shift_right(tb,tb->rnum[0],tb->rbytes); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, tb->R[0]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + } + break; + default: /* cases d and t */ + reiserfs_panic (tb->tb_sb, "PAP-12175: balance_leaf: rnum > 0: unexpectable mode: %s(%d)", + (flag == M_DELETE) ? "DELETE" : ((flag == M_CUT) ? "CUT" : "UNKNOWN"), flag); + } + + } /* tb->rnum[0] > 0 */ + + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->blknum[0] > 3 ) + reiserfs_panic (tb->tb_sb, "PAP-12180: balance_leaf: blknum can not be %d. It must be <= 3", tb->blknum[0]); + + if ( tb->blknum[0] < 0 ) + reiserfs_panic (tb->tb_sb, "PAP-12185: balance_leaf: blknum can not be %d. It must be >= 0", tb->blknum[0]); +#endif + + /* if while adding to a node we discover that it is possible to split + it in two, and merge the left part into the left neighbor and the + right part into the right neighbor, eliminating the node */ + if ( tb->blknum[0] == 0 ) { /* node S[0] is empty now */ + +#ifdef CONFIG_REISERFS_CHECK + if ( ! tb->lnum[0] || ! tb->rnum[0] ) + reiserfs_panic(tb->tb_sb, "PAP-12190: balance_leaf: lnum and rnum must not be zero"); +#if 0 + if (COMP_KEYS (B_N_PKEY(tb->R[0], 0), B_PRIGHT_DELIM_KEY(tbS0))) + reiserfs_panic (tb->tb_sb, "vs-12192: balance_leaf: S[0] is being removed from the tree, it has incorrect right delimiting key"); +#endif +#endif + +#if 0 + /* if insertion was done before 0-th position in R[0], right + delimiting key of the tb->L[0]'s and left delimiting key are + not set correctly */ + if (tb->L[0]) { + copy_key(B_PRIGHT_DELIM_KEY(tb->L[0]), B_PRIGHT_DELIM_KEY(tbS0)); + reiserfs_mark_buffer_dirty (tb->L[0], 0); + } + + if (tb->CFL[0]) { + copy_key (B_N_PDELIM_KEY (tb->CFL[0], tb->lkey[0]), B_PRIGHT_DELIM_KEY(tbS0)); + reiserfs_mark_buffer_dirty (tb->CFL[0], 0); + } +#endif + + /* if insertion was done before 0-th position in R[0], right + delimiting key of the tb->L[0]'s and left delimiting key are + not set correctly */ + if (tb->CFL[0]) { + if (!tb->CFR[0]) + reiserfs_panic (tb->tb_sb, "vs-12195: balance_leaf: CFR not initialized"); + copy_key (B_N_PDELIM_KEY (tb->CFL[0], tb->lkey[0]), B_N_PDELIM_KEY (tb->CFR[0], tb->rkey[0])); + do_balance_mark_internal_dirty (tb, tb->CFL[0], 0); + } + + reiserfs_invalidate_buffer(tb,tbS0); + return 0; + } + + + /* Fill new nodes that appear in place of S[0] */ + + /* I am told that this copying is because we need an array to enable + the looping code. -Hans */ + snum[0] = tb->s1num, + snum[1] = tb->s2num; + sbytes[0] = tb->s1bytes; + sbytes[1] = tb->s2bytes; + for( i = tb->blknum[0] - 2; i >= 0; i-- ) { + +#ifdef CONFIG_REISERFS_CHECK + if (!snum[i]) + reiserfs_panic(tb->tb_sb,"PAP-12200: balance_leaf: snum[%d] == %d. Must be > 0", i, snum[i]); +#endif /* CONFIG_REISERFS_CHECK */ + + /* here we shift from S to S_new nodes */ + + S_new[i] = get_FEB(tb); + + /* initialized block type and tree level */ + B_BLK_HEAD(S_new[i])->blk_level = cpu_to_le16 (DISK_LEAF_NODE_LEVEL); + + + n = B_NR_ITEMS(tbS0); + + switch (flag) { + case M_INSERT: /* insert item */ + + if ( n - snum[i] < item_pos ) + { /* new item or it's part falls to first new node S_new[i]*/ + if ( item_pos == n - snum[i] + 1 && sbytes[i] != -1 ) + { /* part of new item falls into S_new[i] */ + int old_key_comp, old_len, r_zeros_number; + const char * r_body; + int version; + +#ifdef CONFIG_REISERFS_CHECK + if ( !is_direct_le_ih(ih) ) + /* The items which can be inserted are: + Stat_data item, direct item, indirect item and directory item which consist of only two entries "." and "..". + These items must not be broken except for a direct one. */ + reiserfs_panic(tb->tb_sb, "PAP-12205: balance_leaf: " + "non-direct item can not be broken when inserting"); +#endif + + /* Move snum[i]-1 items from S[0] to S_new[i] */ + leaf_move_items (LEAF_FROM_S_TO_SNEW, tb, snum[i] - 1, -1, S_new[i]); +#if 0/*preserve list*/ + if (snum[i] > 1 ) { + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, S_new[i]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); + } +#endif + /* Remember key component and item length */ + version = ih_version (ih); + old_key_comp = le_key_k_offset (version, &(ih->ih_key)); + old_len = le16_to_cpu (ih->ih_item_len); + + /* Calculate key component and item length to insert into S_new[i] */ + set_le_key_k_offset (version, &(ih->ih_key), + le_key_k_offset (version, &(ih->ih_key)) + (old_len - sbytes[i])); + + ih->ih_item_len = cpu_to_le16 (sbytes[i]); + + /* Insert part of the item into S_new[i] before 0-th item */ + bi.tb = tb; + bi.bi_bh = S_new[i]; + bi.bi_parent = 0; + bi.bi_position = 0; + + if ( le_key_k_offset (version, &(ih->ih_key)) - old_key_comp > zeros_num ) { + r_zeros_number = 0; + r_body = body + (le_key_k_offset (version, &(ih->ih_key)) - old_key_comp) - zeros_num; + } + else { + r_body = body; + r_zeros_number = zeros_num - (le_key_k_offset (version, &(ih->ih_key)) - old_key_comp); + zeros_num -= r_zeros_number; + } + + leaf_insert_into_buf (&bi, 0, ih, r_body, r_zeros_number); + + /* Calculate key component and item length to insert into S[i] */ + set_le_key_k_offset (version, &(ih->ih_key), old_key_comp); + ih->ih_item_len = cpu_to_le16 (old_len - sbytes[i]); + tb->insert_size[0] -= sbytes[i]; + } + else /* whole new item falls into S_new[i] */ + { + /* Shift snum[0] - 1 items to S_new[i] (sbytes[i] of split item) */ + leaf_move_items (LEAF_FROM_S_TO_SNEW, tb, snum[i] - 1, sbytes[i], S_new[i]); + + /* Insert new item into S_new[i] */ + bi.tb = tb; + bi.bi_bh = S_new[i]; + bi.bi_parent = 0; + bi.bi_position = 0; + leaf_insert_into_buf (&bi, item_pos - n + snum[i] - 1, ih, body, zeros_num); +#if 0/*preserve list*/ + if (tb->preserve_mode == PRESERVE_INDIRECT_TO_DIRECT){ + mark_suspected_recipient (tb->tb_sb, bi.bi_bh); + } +#endif + + zeros_num = tb->insert_size[0] = 0; + } + } + + else /* new item or it part don't falls into S_new[i] */ + { + leaf_move_items (LEAF_FROM_S_TO_SNEW, tb, snum[i], sbytes[i], S_new[i]); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, S_new[i]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + } + break; + + case M_PASTE: /* append item */ + + if ( n - snum[i] <= item_pos ) /* pasted item or part if it falls to S_new[i] */ + { + if ( item_pos == n - snum[i] && sbytes[i] != -1 ) + { /* we must shift part of the appended item */ + struct item_head * aux_ih; + +#ifdef CONFIG_REISERFS_CHECK + if ( ih ) + reiserfs_panic (tb->tb_sb, "PAP-12210: balance_leaf: ih must be 0"); +#endif /* CONFIG_REISERFS_CHECK */ + + if ( is_direntry_le_ih (aux_ih = B_N_PITEM_HEAD(tbS0,item_pos))) { + /* we append to directory item */ + + int entry_count; + + entry_count = le16_to_cpu (aux_ih->u.ih_entry_count); + + if ( entry_count - sbytes[i] < pos_in_item && pos_in_item <= entry_count ) { + /* new directory entry falls into S_new[i] */ + +#ifdef CONFIG_REISERFS_CHECK + if ( ! tb->insert_size[0] ) + reiserfs_panic (tb->tb_sb, "PAP-12215: balance_leaif: insert_size is already 0"); + if ( sbytes[i] - 1 >= entry_count ) + reiserfs_panic (tb->tb_sb, "PAP-12220: balance_leaf: " + "there are no so much entries (%d), only %d", + sbytes[i] - 1, entry_count); +#endif + + /* Shift snum[i]-1 items in whole. Shift sbytes[i] directory entries from directory item number snum[i] */ + leaf_move_items (LEAF_FROM_S_TO_SNEW, tb, snum[i], sbytes[i]-1, S_new[i]); +#if 0/*preserve list*/ + /* if more than the affected item is shifted, or if more than + one entry (from the affected item) is shifted */ + if (snum[i] > 1 || sbytes[i] > 1) { + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, S_new[i]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); + } +#endif + /* Paste given directory entry to directory item */ + bi.tb = tb; + bi.bi_bh = S_new[i]; + bi.bi_parent = 0; + bi.bi_position = 0; + leaf_paste_in_buffer (&bi, 0, pos_in_item - entry_count + sbytes[i] - 1, + tb->insert_size[0], body,zeros_num); + /* paste new directory entry */ + leaf_paste_entries ( + bi.bi_bh, 0, pos_in_item - entry_count + sbytes[i] - 1, + 1, (struct reiserfs_de_head *)body, body + DEH_SIZE, + tb->insert_size[0] + ); + tb->insert_size[0] = 0; + pos_in_item++; + } else { /* new directory entry doesn't fall into S_new[i] */ + leaf_move_items (LEAF_FROM_S_TO_SNEW, tb, snum[i], sbytes[i], S_new[i]); + } + } + else /* regular object */ + { + int n_shift, n_rem, r_zeros_number; + const char * r_body; + +#ifdef CONFIG_REISERFS_CHECK + if ( pos_in_item != B_N_PITEM_HEAD(tbS0,item_pos)->ih_item_len || + tb->insert_size[0] <= 0 ) + reiserfs_panic (tb->tb_sb, "PAP-12225: balance_leaf: item too short or insert_size <= 0"); +#endif + + /* Calculate number of bytes which must be shifted from appended item */ + n_shift = sbytes[i] - tb->insert_size[0]; + if ( n_shift < 0 ) + n_shift = 0; + leaf_move_items (LEAF_FROM_S_TO_SNEW, tb, snum[i], n_shift, S_new[i]); + + /* Calculate number of bytes which must remain in body after append to S_new[i] */ + n_rem = tb->insert_size[0] - sbytes[i]; + if ( n_rem < 0 ) + n_rem = 0; + /* Append part of body into S_new[0] */ + bi.tb = tb; + bi.bi_bh = S_new[i]; + bi.bi_parent = 0; + bi.bi_position = 0; + + if ( n_rem > zeros_num ) { + r_zeros_number = 0; + r_body = body + n_rem - zeros_num; + } + else { + r_body = body; + r_zeros_number = zeros_num - n_rem; + zeros_num -= r_zeros_number; + } + + leaf_paste_in_buffer(&bi, 0, n_shift, tb->insert_size[0]-n_rem, r_body,r_zeros_number); + { + struct item_head * tmp; + + tmp = B_N_PITEM_HEAD(S_new[i],0); + if (is_indirect_le_ih (tmp)) { + if (n_rem) + reiserfs_panic (tb->tb_sb, "PAP-12230: balance_leaf: invalid action with indirect item"); + set_ih_free_space (tmp, ((struct unfm_nodeinfo*)body)->unfm_freespace); + } + set_le_key_k_offset (ih_version (tmp), &tmp->ih_key, + le_key_k_offset (ih_version (tmp), &tmp->ih_key) + n_rem); + } + + tb->insert_size[0] = n_rem; + if ( ! n_rem ) + pos_in_item++; + } + } + else + /* item falls wholly into S_new[i] */ + { + int ret_val; + struct item_head * pasted; + +#ifdef CONFIG_REISERFS_CHECK + struct item_head * ih = B_N_PITEM_HEAD(tbS0,item_pos); + + if ( ! is_direntry_le_ih(ih) && (pos_in_item != ih->ih_item_len || + tb->insert_size[0] <= 0) ) + reiserfs_panic (tb->tb_sb, "PAP-12235: balance_leaf: pos_in_item must be equal to ih_item_len"); +#endif /* CONFIG_REISERFS_CHECK */ + + ret_val = leaf_move_items (LEAF_FROM_S_TO_SNEW, tb, snum[i], sbytes[i], S_new[i]); +#if 0/*preserve list*/ + /* we must preserve that which we are pasting onto the end of and shifting */ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, S_new[i]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + +#ifdef CONFIG_REISERFS_CHECK + if ( ret_val ) + reiserfs_panic (tb->tb_sb, "PAP-12240: balance_leaf: " + "unexpected value returned by leaf_move_items (%d)", + ret_val); +#endif /* CONFIG_REISERFS_CHECK */ + + /* paste into item */ + bi.tb = tb; + bi.bi_bh = S_new[i]; + bi.bi_parent = 0; + bi.bi_position = 0; + leaf_paste_in_buffer(&bi, item_pos - n + snum[i], pos_in_item, tb->insert_size[0], body, zeros_num); + + pasted = B_N_PITEM_HEAD(S_new[i], item_pos - n + snum[i]); + if (is_direntry_le_ih (pasted)) + { + leaf_paste_entries ( + bi.bi_bh, item_pos - n + snum[i], pos_in_item, 1, + (struct reiserfs_de_head *)body, body + DEH_SIZE, tb->insert_size[0] + ); + } + + /* if we paste to indirect item update ih_free_space */ + if (is_indirect_le_ih (pasted)) + set_ih_free_space (pasted, ((struct unfm_nodeinfo*)body)->unfm_freespace); + zeros_num = tb->insert_size[0] = 0; + } + } + + else /* pasted item doesn't fall into S_new[i] */ + { + leaf_move_items (LEAF_FROM_S_TO_SNEW, tb, snum[i], sbytes[i], S_new[i]); +#if 0/*preserve list*/ + preserve_shifted(tb, &(PATH_PLAST_BUFFER (tb->tb_path)), tbF0, S0_b_item_order, S_new[i]); + tbS0 = PATH_PLAST_BUFFER (tb->tb_path); +#endif + } + break; + default: /* cases d and t */ + reiserfs_panic (tb->tb_sb, "PAP-12245: balance_leaf: blknum > 2: unexpectable mode: %s(%d)", + (flag == M_DELETE) ? "DELETE" : ((flag == M_CUT) ? "CUT" : "UNKNOWN"), flag); + } + + memcpy (insert_key + i,B_N_PKEY(S_new[i],0),KEY_SIZE); + insert_ptr[i] = S_new[i]; + +#ifdef CONFIG_REISERFS_CHECK + if (atomic_read (&(S_new[i]->b_count)) != 1) { + if (atomic_read(&(S_new[i]->b_count)) != 2 || + !(buffer_journaled(S_new[i]) || buffer_journal_dirty(S_new[i]))) { + reiserfs_panic (tb->tb_sb, "PAP-12247: balance_leaf: S_new[%d] : (%b)\n", i, S_new[i]); + } + } +#endif + +#if 0 + /* update right_delimiting_key fields */ + copy_key (B_PRIGHT_DELIM_KEY (S_new[i]), B_PRIGHT_DELIM_KEY (tbS0)); + copy_key (B_PRIGHT_DELIM_KEY (tbS0), B_N_PKEY (S_new[i], 0)); + reiserfs_mark_buffer_dirty (tbS0, 0); +#endif + + } + + /* if the affected item was not wholly shifted then we perform all necessary operations on that part or whole of the + affected item which remains in S */ + if ( 0 <= item_pos && item_pos < tb->s0num ) + { /* if we must insert or append into buffer S[0] */ + + switch (flag) + { + case M_INSERT: /* insert item into S[0] */ + bi.tb = tb; + bi.bi_bh = tbS0; + bi.bi_parent = PATH_H_PPARENT (tb->tb_path, 0); + bi.bi_position = PATH_H_POSITION (tb->tb_path, 1); + leaf_insert_into_buf (&bi, item_pos, ih, body, zeros_num); +#if 0/*preserve list*/ + if (tb->preserve_mode == PRESERVE_INDIRECT_TO_DIRECT){ + mark_suspected_recipient (tb->tb_sb, bi.bi_bh); + } +#endif + + /* If we insert the first key change the delimiting key */ + if( item_pos == 0 ) { + if (tb->CFL[0]) /* can be 0 in reiserfsck */ + replace_key(tb, tb->CFL[0], tb->lkey[0],tbS0,0); + +#if 0 /* right delim key support */ +#ifdef CONFIG_REISERFS_CHECK + if ( ! tb->CFL[0] || ! tb->L[0] || (B_NR_ITEMS (tbS0) > 1 && + COMP_KEYS(B_PRIGHT_DELIM_KEY(tb->L[0]), B_N_PKEY(tbS0, 1))) ) + reiserfs_panic(tb->tb_sb, "PAP-12250: balance_leaf: invalid right delimiting key"); + if (!buffer_dirty (tb->L[0]) && !(buffer_journaled(tb->L[0]) || + buffer_journal_dirty(tb->L[0]))) + reiserfs_panic (tb->tb_sb, "PAP-12255: balance_leaf: tb->L[0] must be dirty"); +#endif + if (tb->L[0]) /* can be 0 in reiserfsck */ + copy_key (B_PRIGHT_DELIM_KEY (tb->L[0]), &(ih->ih_key)); +#endif /* right delim key support */ + } + break; + + case M_PASTE: { /* append item in S[0] */ + struct item_head * pasted; + + pasted = B_N_PITEM_HEAD (tbS0, item_pos); + /* when directory, may be new entry already pasted */ + if (is_direntry_le_ih (pasted)) { + if ( pos_in_item >= 0 && pos_in_item <= le16_to_cpu (pasted->u.ih_entry_count) ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( ! tb->insert_size[0] ) + reiserfs_panic (tb->tb_sb, "PAP-12260: balance_leaf: insert_size is 0 already"); +#endif /* CONFIG_REISERFS_CHECK */ + + /* prepare space */ + bi.tb = tb; + bi.bi_bh = tbS0; + bi.bi_parent = PATH_H_PPARENT (tb->tb_path, 0); + bi.bi_position = PATH_H_POSITION (tb->tb_path, 1); + leaf_paste_in_buffer(&bi, item_pos, pos_in_item, tb->insert_size[0], body, zeros_num); + + +#ifdef CONFIG_REISERFS_CHECK +#if 0 + if ( ! item_pos && ! pos_in_item && (! tb->L[0] || COMP_KEYS(B_PRIGHT_DELIM_KEY(tb->L[0]), + B_N_PKEY(tbS0, 0))) ) + reiserfs_panic(tb->tb_sb, "PAP-12265: balance_leaf: invalid right delimiting key"); +#endif +#endif + + /* paste entry */ + leaf_paste_entries ( + bi.bi_bh, item_pos, pos_in_item, 1, (struct reiserfs_de_head *)body, + body + DEH_SIZE, tb->insert_size[0] + ); + if ( ! item_pos && ! pos_in_item ) { + +#ifdef CONFIG_REISERFS_CHECK + if (!tb->CFL[0] || !tb->L[0]) + reiserfs_panic (tb->tb_sb, "PAP-12270: balance_leaf: CFL[0]/L[0] must be specified"); +#endif /* CONFIG_REISERFS_CHECK */ + + if (tb->CFL[0]) { + replace_key(tb, tb->CFL[0], tb->lkey[0],tbS0,0); + +#if 0 + /* update right delimiting key */ + copy_key (B_PRIGHT_DELIM_KEY (tb->L[0]), B_N_PKEY(tbS0, 0)); + /* probably not needed as something has been shifted to tb->L[0] already */ + reiserfs_mark_buffer_dirty (tb->L[0], 0); +#endif + } + } + tb->insert_size[0] = 0; + } + } else { /* regular object */ + if ( pos_in_item == pasted->ih_item_len ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->insert_size[0] <= 0 ) + reiserfs_panic (tb->tb_sb, + "PAP-12275: balance_leaf: insert size must not be %d", tb->insert_size[0]); +#endif /* CONFIG_REISERFS_CHECK */ + bi.tb = tb; + bi.bi_bh = tbS0; + bi.bi_parent = PATH_H_PPARENT (tb->tb_path, 0); + bi.bi_position = PATH_H_POSITION (tb->tb_path, 1); + leaf_paste_in_buffer (&bi, item_pos, pos_in_item, tb->insert_size[0], body, zeros_num); + + if (is_indirect_le_ih (pasted)) { + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->insert_size[0] != UNFM_P_SIZE ) + reiserfs_panic (tb->tb_sb, + "PAP-12280: balance_leaf: insert_size for indirect item must be %d, not %d", + UNFM_P_SIZE, tb->insert_size[0]); +#endif /* CONFIG_REISERFS_CHECK */ + + set_ih_free_space (pasted, ((struct unfm_nodeinfo*)body)->unfm_freespace); + } + tb->insert_size[0] = 0; + } + +#ifdef CONFIG_REISERFS_CHECK + else { + if ( tb->insert_size[0] ) { + print_cur_tb ("12285"); + reiserfs_panic (tb->tb_sb, "PAP-12285: balance_leaf: insert_size must be 0 (%d)", tb->insert_size[0]); + } + } +#endif /* CONFIG_REISERFS_CHECK */ + + } + } /* case M_PASTE: */ + } + } + +#ifdef CONFIG_REISERFS_CHECK + if ( flag == M_PASTE && tb->insert_size[0] ) { + print_cur_tb ("12290"); + reiserfs_panic (tb->tb_sb, "PAP-12290: balance_leaf: insert_size is still not 0 (%d)", tb->insert_size[0]); + } +#endif /* CONFIG_REISERFS_CHECK */ + + return 0; +} /* Leaf level of the tree is balanced (end of balance_leaf) */ + + + +/* Make empty node */ +void make_empty_node (struct buffer_info * bi) +{ + struct block_head * blkh; + +#ifdef CONFIG_REISERFS_CHECK + if (bi->bi_bh == NULL) + reiserfs_panic (0, "PAP-12295: make_empty_node: pointer to the buffer is NULL"); +#endif + + (blkh = B_BLK_HEAD(bi->bi_bh))->blk_nr_item = cpu_to_le16 (0); + blkh->blk_free_space = cpu_to_le16 (MAX_CHILD_SIZE(bi->bi_bh)); + + if (bi->bi_parent) + B_N_CHILD (bi->bi_parent, bi->bi_position)->dc_size = 0; +} + + +/* Get first empty buffer */ +struct buffer_head * get_FEB (struct tree_balance * tb) +{ + int i; + struct buffer_head * first_b; + struct buffer_info bi; + + for (i = 0; i < MAX_FEB_SIZE; i ++) + if (tb->FEB[i] != 0) + break; + + if (i == MAX_FEB_SIZE) + reiserfs_panic(tb->tb_sb, "vs-12300: get_FEB: FEB list is empty"); + + bi.tb = tb; + bi.bi_bh = first_b = tb->FEB[i]; + bi.bi_parent = 0; + bi.bi_position = 0; + make_empty_node (&bi); + set_bit(BH_Uptodate, &first_b->b_state); + tb->FEB[i] = 0; + tb->used[i] = first_b; + +#ifdef REISERFS_FSCK + mark_block_formatted (first_b->b_blocknr); +#endif + + return(first_b); +} + + +/* This is now used because reiserfs_free_block has to be able to +** schedule. +*/ +static void store_thrown (struct tree_balance * tb, struct buffer_head * bh) +{ + int i; + + if (buffer_dirty (bh)) + printk ("store_thrown deals with dirty buffer\n"); + for (i = 0; i < sizeof (tb->thrown)/sizeof (tb->thrown[0]); i ++) + if (!tb->thrown[i]) { + tb->thrown[i] = bh; + atomic_inc(&bh->b_count) ; /* decremented in free_thrown */ + return; + } + reiserfs_warning ("store_thrown: too many thrown buffers\n"); +} + +static void free_thrown(struct tree_balance *tb) { + int i ; + unsigned long blocknr ; + for (i = 0; i < sizeof (tb->thrown)/sizeof (tb->thrown[0]); i++) { + if (tb->thrown[i]) { + blocknr = tb->thrown[i]->b_blocknr ; + if (buffer_dirty (tb->thrown[i])) + printk ("free_thrown deals with dirty buffer %ld\n", blocknr); + brelse(tb->thrown[i]) ; /* incremented in store_thrown */ + reiserfs_free_block (tb->transaction_handle, blocknr); + } + } +} + +void reiserfs_invalidate_buffer (struct tree_balance * tb, struct buffer_head * bh) +{ + B_BLK_HEAD (bh)->blk_level = cpu_to_le16 (FREE_LEVEL)/*0*/; + B_BLK_HEAD (bh)->blk_nr_item = cpu_to_le16 (0); + mark_buffer_clean (bh); + /* reiserfs_free_block is no longer schedule safe + reiserfs_free_block (tb->transaction_handle, tb->tb_sb, bh->b_blocknr); + */ + + store_thrown (tb, bh); +#if 0 +#ifdef REISERFS_FSCK + { + struct buffer_head * to_be_forgotten; + + to_be_forgotten = find_buffer (bh->b_dev, bh->b_blocknr, bh->b_size); + if (to_be_forgotten) { + to_be_forgotten->b_count ++; + bforget (to_be_forgotten); + } + unmark_block_formatted (bh->b_blocknr); + } +#endif +#endif +} + +/* Replace n_dest'th key in buffer dest by n_src'th key of buffer src.*/ +void replace_key (struct tree_balance * tb, struct buffer_head * dest, int n_dest, + struct buffer_head * src, int n_src) +{ + +#ifdef CONFIG_REISERFS_CHECK + if (dest == NULL || src == NULL) + reiserfs_panic (0, "vs-12305: replace_key: sourse or destination buffer is 0 (src=%p, dest=%p)", src, dest); + + if ( ! B_IS_KEYS_LEVEL (dest) ) + reiserfs_panic (0, "vs-12310: replace_key: invalid level (%z) for destination buffer. dest must be leaf", + dest); + + if (n_dest < 0 || n_src < 0) + reiserfs_panic (0, "vs-12315: replace_key: src(%d) or dest(%d) key number less than 0", n_src, n_dest); + + if (n_dest >= B_NR_ITEMS(dest) || n_src >= B_NR_ITEMS(src)) + reiserfs_panic (0, "vs-12320: replace_key: src(%d(%d)) or dest(%d(%d)) key number is too big", + n_src, B_NR_ITEMS(src), n_dest, B_NR_ITEMS(dest)); +#endif /* CONFIG_REISERFS_CHECK */ + + if (B_IS_ITEMS_LEVEL (src)) + /* source buffer contains leaf node */ + memcpy (B_N_PDELIM_KEY(dest,n_dest), B_N_PITEM_HEAD(src,n_src), KEY_SIZE); + else + memcpy (B_N_PDELIM_KEY(dest,n_dest), B_N_PDELIM_KEY(src,n_src), KEY_SIZE); + + do_balance_mark_internal_dirty (tb, dest, 0); +} + + +int get_left_neighbor_position ( + struct tree_balance * tb, + int h + ) +{ + int Sh_position = PATH_H_POSITION (tb->tb_path, h + 1); + +#ifdef CONFIG_REISERFS_CHECK + if (PATH_H_PPARENT (tb->tb_path, h) == 0 || tb->FL[h] == 0) + reiserfs_panic (tb->tb_sb, "vs-12325: get_left_neighbor_position: FL[%d](%p) or F[%d](%p) does not exist", + h, tb->FL[h], h, PATH_H_PPARENT (tb->tb_path, h)); +#endif + + if (Sh_position == 0) + return B_NR_ITEMS (tb->FL[h]); + else + return Sh_position - 1; +} + + +int get_right_neighbor_position (struct tree_balance * tb, int h) +{ + int Sh_position = PATH_H_POSITION (tb->tb_path, h + 1); + +#ifdef CONFIG_REISERFS_CHECK + if (PATH_H_PPARENT (tb->tb_path, h) == 0 || tb->FR[h] == 0) + reiserfs_panic (tb->tb_sb, "vs-12330: get_right_neighbor_position: F[%d](%p) or FR[%d](%p) does not exist", + h, PATH_H_PPARENT (tb->tb_path, h), h, tb->FR[h]); +#endif + + if (Sh_position == B_NR_ITEMS (PATH_H_PPARENT (tb->tb_path, h))) + return 0; + else + return Sh_position + 1; +} + + +#ifdef CONFIG_REISERFS_CHECK + +int is_reusable (struct super_block * s, unsigned long block, int bit_value); +static void check_internal_node (struct super_block * s, struct buffer_head * bh, char * mes) +{ + struct disk_child * dc; + int i; + + if (!bh) + reiserfs_panic (s, "PAP-12336: check_internal_node: bh == 0"); + + if (!bh || !B_IS_IN_TREE (bh)) + return; + + if (!buffer_dirty (bh) && + !(buffer_journaled(bh) || buffer_journal_dirty(bh))) { + reiserfs_panic (s, "PAP-12337: check_internal_node: buffer (%b) must be dirty", bh); + } + + dc = B_N_CHILD (bh, 0); + + for (i = 0; i <= B_NR_ITEMS (bh); i ++, dc ++) { + if (!is_reusable (s, dc->dc_block_number, 1) ) { + print_cur_tb (mes); + reiserfs_panic (s, "PAP-12338: check_internal_node: invalid child pointer %y in %b", dc, bh); + } + } +} + + +static int locked_or_not_in_tree (struct buffer_head * bh, char * which) +{ + if ( buffer_locked (bh) || !B_IS_IN_TREE (bh) ) { + reiserfs_warning ("vs-12339: locked_or_not_in_tree: %s (%b)\n", which, bh); + return 1; + } + return 0; +} + + +static int check_before_balancing (struct tree_balance * tb) +{ + int retval = 0; + + if ( cur_tb ) { + reiserfs_panic (tb->tb_sb, "vs-12335: check_before_balancing: " + "suspect that schedule occurred based on cur_tb not being null at this point in code. " + "do_balance cannot properly handle schedule occuring while it runs."); + } + + /* double check that buffers that we will modify are unlocked. (fix_nodes should already have + prepped all of these for us). */ + if ( tb->lnum[0] ) { + retval |= locked_or_not_in_tree (tb->L[0], "L[0]"); + retval |= locked_or_not_in_tree (tb->FL[0], "FL[0]"); + retval |= locked_or_not_in_tree (tb->CFL[0], "CFL[0]"); + check_leaf (tb->L[0]); + } + if ( tb->rnum[0] ) { + retval |= locked_or_not_in_tree (tb->R[0], "R[0]"); + retval |= locked_or_not_in_tree (tb->FR[0], "FR[0]"); + retval |= locked_or_not_in_tree (tb->CFR[0], "CFR[0]"); + check_leaf (tb->R[0]); + } + retval |= locked_or_not_in_tree (PATH_PLAST_BUFFER (tb->tb_path), "S[0]"); + check_leaf (PATH_PLAST_BUFFER (tb->tb_path)); + + return retval; +} + + +void check_after_balance_leaf (struct tree_balance * tb) +{ + if (tb->lnum[0]) { + if (B_FREE_SPACE (tb->L[0]) != + MAX_CHILD_SIZE (tb->L[0]) - B_N_CHILD (tb->FL[0], get_left_neighbor_position (tb, 0))->dc_size) { + print_cur_tb ("12221"); + reiserfs_panic (tb->tb_sb, "PAP-12355: check_after_balance_leaf: shift to left was incorrect"); + } + } + if (tb->rnum[0]) { + if (B_FREE_SPACE (tb->R[0]) != + MAX_CHILD_SIZE (tb->R[0]) - B_N_CHILD (tb->FR[0], get_right_neighbor_position (tb, 0))->dc_size) { + print_cur_tb ("12222"); + reiserfs_panic (tb->tb_sb, "PAP-12360: check_after_balance_leaf: shift to right was incorrect"); + } + } + if (PATH_H_PBUFFER(tb->tb_path,1) && (B_FREE_SPACE (PATH_H_PBUFFER(tb->tb_path,0)) != + (MAX_CHILD_SIZE (PATH_H_PBUFFER(tb->tb_path,0)) - + B_N_CHILD (PATH_H_PBUFFER(tb->tb_path,1), + PATH_H_POSITION (tb->tb_path, 1))->dc_size))) { + print_cur_tb ("12223"); + reiserfs_panic (tb->tb_sb, "PAP-12365: check_after_balance_leaf: S is incorrect"); + } +} + + +void check_leaf_level (struct tree_balance * tb) +{ + check_leaf (tb->L[0]); + check_leaf (tb->R[0]); + check_leaf (PATH_PLAST_BUFFER (tb->tb_path)); +} + +void check_internal_levels (struct tree_balance * tb) +{ + int h; + + /* check all internal nodes */ + for (h = 1; tb->insert_size[h]; h ++) { + check_internal_node (tb->tb_sb, PATH_H_PBUFFER (tb->tb_path, h), "BAD BUFFER ON PATH"); + if (tb->lnum[h]) + check_internal_node (tb->tb_sb, tb->L[h], "BAD L"); + if (tb->rnum[h]) + check_internal_node (tb->tb_sb, tb->R[h], "BAD R"); + } + +} + +#endif + + + + + + +/* Now we have all of the buffers that must be used in balancing of + the tree. We rely on the assumption that schedule() will not occur + while do_balance works. ( Only interrupt handlers are acceptable.) + We balance the tree according to the analysis made before this, + using buffers already obtained. For SMP support it will someday be + necessary to add ordered locking of tb. */ + +/* Some interesting rules of balancing: + + we delete a maximum of two nodes per level per balancing: we never + delete R, when we delete two of three nodes L, S, R then we move + them into R. + + we only delete L if we are deleting two nodes, if we delete only + one node we delete S + + if we shift leaves then we shift as much as we can: this is a + deliberate policy of extremism in node packing which results in + higher average utilization after repeated random balance operations + at the cost of more memory copies and more balancing as a result of + small insertions to full nodes. + + if we shift internal nodes we try to evenly balance the node + utilization, with consequent less balancing at the cost of lower + utilization. + + one could argue that the policy for directories in leaves should be + that of internal nodes, but we will wait until another day to + evaluate this.... It would be nice to someday measure and prove + these assumptions as to what is optimal.... + +*/ + +static inline void do_balance_starts (struct tree_balance *tb) +{ + /* use print_cur_tb() to see initial state of struct + tree_balance */ + + /* store_print_tb (tb); */ + +#ifdef CONFIG_REISERFS_CHECK + + /* do not delete, just comment it out */ +/* print_tb(flag, PATH_LAST_POSITION(tb->tb_path), tb->tb_path->pos_in_item, tb, + "check");*/ + + if (check_before_balancing (tb)) + reiserfs_panic (tb->tb_sb, "PAP-12340: do_balance: locked buffers in TB"); + +#ifndef __KERNEL__ + if ( atomic_read(&(PATH_PLAST_BUFFER(tb->tb_path)->b_count)) > 1 || (tb->L[0] && atomic_read(&(tb->L[0]->b_count)) > 1) || + (tb->R[0] && atomic_read(&(tb->R[0]->b_count)) > 1) ) { + print_cur_tb ("first three parameters are invalid"); + reiserfs_panic (tb->tb_sb, "PAP-12345: do_balance: counter too big"); + } +#endif /* !__KERNEL__ */ + cur_tb = tb; + +#endif /* CONFIG_REISERFS_CHECK */ +} + + +static inline void do_balance_completed (struct tree_balance * tb) +{ + +#ifdef CONFIG_REISERFS_CHECK + check_leaf_level (tb); + check_internal_levels (tb); + cur_tb = NULL; +#endif + + /* reiserfs_free_block is no longer schedule safe. So, we need to + ** put the buffers we want freed on the thrown list during do_balance, + ** and then free them now + */ + + tb->tb_sb->u.reiserfs_sb.s_do_balance ++; + + + /* release all nodes hold to perform the balancing */ + unfix_nodes(tb); + + free_thrown(tb) ; +} + + + + + +void do_balance (struct tree_balance * tb, /* tree_balance structure */ + struct item_head * ih, /* item header of inserted item */ + const char * body, /* body of inserted item or bytes to paste */ + int flag) /* i - insert, d - delete + c - cut, p - paste + + Cut means delete part of an item + (includes removing an entry from a + directory). + + Delete means delete whole item. + + Insert means add a new item into the + tree. + + Paste means to append to the end of an + existing file or to insert a directory + entry. */ +{ + int child_pos, /* position of a child node in its parent */ + h; /* level of the tree being processed */ + struct item_head insert_key[2]; /* in our processing of one level + we sometimes determine what + must be inserted into the next + higher level. This insertion + consists of a key or two keys + and their corresponding + pointers */ + struct buffer_head *insert_ptr[2]; /* inserted node-ptrs for the next + level */ + + tb->tb_mode = flag; + tb->need_balance_dirty = 0; + + if (FILESYSTEM_CHANGED_TB(tb)) { + reiserfs_panic(tb->tb_sb, "clm-6000: do_balance, fs generation has changed\n") ; + } + /* if we have no real work to do */ + if ( ! tb->insert_size[0] ) { + reiserfs_warning ("PAP-12350: do_balance: insert_size == 0, mode == %c", + flag); + unfix_nodes(tb); + return; + } + + atomic_inc (&(fs_generation (tb->tb_sb))); + do_balance_starts (tb); + +#ifdef REISERFS_FSCK + if (flag == M_INTERNAL) { + insert_ptr[0] = (struct buffer_head *)body; + /* we must prepare insert_key */ + + if (PATH_H_B_ITEM_ORDER (tb->tb_path, 0)/*LAST_POSITION (tb->tb_path)*//*item_pos*/ == -1) { + /* get delimiting key from buffer in tree */ + copy_key (&insert_key[0].ih_key, B_N_PKEY (PATH_PLAST_BUFFER (tb->tb_path), 0)); + /*insert_ptr[0]->b_item_order = 0;*/ + } else { + /* get delimiting key from new buffer */ + copy_key (&insert_key[0].ih_key, B_N_PKEY((struct buffer_head *)body,0)); + /*insert_ptr[0]->b_item_order = item_pos;*/ + } + + /* and insert_ptr instead of balance_leaf */ + child_pos = PATH_H_B_ITEM_ORDER (tb->tb_path, 0)/*item_pos*/; + } else +#endif + + /* balance leaf returns 0 except if combining L R and S into + one node. see balance_internal() for explanation of this + line of code.*/ + child_pos = PATH_H_B_ITEM_ORDER (tb->tb_path, 0) + + balance_leaf (tb, ih, body, flag, insert_key, insert_ptr); + +#ifdef CONFIG_REISERFS_CHECK + check_after_balance_leaf (tb); +#endif + + /* Balance internal level of the tree. */ + for ( h = 1; h < MAX_HEIGHT && tb->insert_size[h]; h++ ) + child_pos = balance_internal (tb, h, child_pos, insert_key, insert_ptr); + + + do_balance_completed (tb); + +} diff --git a/fs/reiserfs/file.c b/fs/reiserfs/file.c new file mode 100644 index 000000000..c7a62c0e7 --- /dev/null +++ b/fs/reiserfs/file.c @@ -0,0 +1,124 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + + +#ifdef __KERNEL__ + +#include <linux/sched.h> +#include <linux/reiserfs_fs.h> +#include <linux/smp_lock.h> + +#else + +#include "nokernel.h" + +#endif + +/* +** We pack the tails of files on file close, not at the time they are written. +** This implies an unnecessary copy of the tail and an unnecessary indirect item +** insertion/balancing, for files that are written in one write. +** It avoids unnecessary tail packings (balances) for files that are written in +** multiple writes and are small enough to have tails. +** +** file_release is called by the VFS layer when the file is closed. If +** this is the last open file descriptor, and the file +** small enough to have a tail, and the tail is currently in an +** unformatted node, the tail is converted back into a direct item. +** +** We use reiserfs_truncate_file to pack the tail, since it already has +** all the conditions coded. +*/ +static int reiserfs_file_release (struct inode * inode, struct file * filp) +{ + + struct reiserfs_transaction_handle th ; + int windex ; + + if (!S_ISREG (inode->i_mode)) + BUG (); + + /* fast out for when nothing needs to be done */ + if ((atomic_read(&inode->i_count) > 1 || + !inode->u.reiserfs_i.i_pack_on_close || + !tail_has_to_be_packed(inode)) && + inode->u.reiserfs_i.i_prealloc_count <= 0) { + return 0; + } + + lock_kernel() ; + down (&inode->i_sem); + journal_begin(&th, inode->i_sb, JOURNAL_PER_BALANCE_CNT * 3) ; + +#ifdef REISERFS_PREALLOCATE + reiserfs_discard_prealloc (&th, inode); +#endif + journal_end(&th, inode->i_sb, JOURNAL_PER_BALANCE_CNT * 3) ; + + if (atomic_read(&inode->i_count) <= 1 && + inode->u.reiserfs_i.i_pack_on_close && + tail_has_to_be_packed (inode)) { + /* if regular file is released by last holder and it has been + appended (we append by unformatted node only) or its direct + item(s) had to be converted, then it may have to be + indirect2direct converted */ + windex = push_journal_writer("file_release") ; + reiserfs_truncate_file(inode, 0) ; + pop_journal_writer(windex) ; + } + up (&inode->i_sem); + unlock_kernel() ; + return 0; +} + +static void reiserfs_vfs_truncate_file(struct inode *inode) { + reiserfs_truncate_file(inode, 1) ; +} + +/* Sync a reiserfs file. */ +static int reiserfs_sync_file( + struct file * p_s_filp, + struct dentry * p_s_dentry, + int datasync + ) { + struct inode * p_s_inode = p_s_dentry->d_inode; + struct reiserfs_transaction_handle th ; + int n_err = 0; + int windex ; + int jbegin_count = 1 ; + + lock_kernel() ; + + if (!S_ISREG(p_s_inode->i_mode)) + BUG (); + + n_err = fsync_inode_buffers(p_s_inode) ; + /* commit the current transaction to flush any metadata + ** changes. sys_fsync takes care of flushing the dirty pages for us + */ + journal_begin(&th, p_s_inode->i_sb, jbegin_count) ; + windex = push_journal_writer("sync_file") ; + reiserfs_update_sd(&th, p_s_inode); + pop_journal_writer(windex) ; + journal_end_sync(&th, p_s_inode->i_sb,jbegin_count) ; + unlock_kernel() ; + return ( n_err < 0 ) ? -EIO : 0; +} + + +struct file_operations reiserfs_file_operations = { + read: generic_file_read, + write: generic_file_write, + ioctl: reiserfs_ioctl, + mmap: generic_file_mmap, + release: reiserfs_file_release, + fsync: reiserfs_sync_file, +}; + + +struct inode_operations reiserfs_file_inode_operations = { + truncate: reiserfs_vfs_truncate_file, +}; + + diff --git a/fs/reiserfs/fix_node.c b/fs/reiserfs/fix_node.c new file mode 100644 index 000000000..daa30b588 --- /dev/null +++ b/fs/reiserfs/fix_node.c @@ -0,0 +1,2908 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +/** + ** old_item_num + ** old_entry_num + ** set_entry_sizes + ** create_virtual_node + ** check_left + ** check_right + ** directory_part_size + ** get_num_ver + ** set_parameters + ** is_leaf_removable + ** are_leaves_removable + ** get_empty_nodes + ** get_lfree + ** get_rfree + ** is_left_neighbor_in_cache + ** decrement_key + ** get_far_parent + ** get_parents + ** can_node_be_removed + ** ip_check_balance + ** dc_check_balance_internal + ** dc_check_balance_leaf + ** dc_check_balance + ** check_balance + ** get_direct_parent + ** get_neighbors + ** fix_nodes + ** + ** + **/ + + +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/sched.h> +#include <linux/string.h> +#include <linux/locks.h> +#include <linux/reiserfs_fs.h> + +#else + +#include "nokernel.h" + +#endif + + + +/* To make any changes in the tree we find a node, that contains item + to be changed/deleted or position in the node we insert a new item + to. We call this node S. To do balancing we need to decide what we + will shift to left/right neighbor, or to a new node, where new item + will be etc. To make this analysis simpler we build virtual + node. Virtual node is an array of items, that will replace items of + node S. (For instance if we are going to delete an item, virtual + node does not contain it). Virtual node keeps information about + item sizes and types, mergeability of first and last items, sizes + of all entries in directory item. We use this array of items when + calculating what we can shift to neighbors and how many nodes we + have to have if we do not any shiftings, if we shift to left/right + neighbor or to both. */ + + +/* taking item number in virtual node, returns number of item, that it has in source buffer */ +static inline int old_item_num (int new_num, int affected_item_num, int mode) +{ + if (mode == M_PASTE || mode == M_CUT || new_num < affected_item_num) + return new_num; + + if (mode == M_INSERT) { + +#ifdef CONFIG_REISERFS_CHECK + if (new_num == 0) + reiserfs_panic (0,"vs-8005: old_item_num: for INSERT mode and item number of inserted item"); +#endif + + return new_num - 1; + } + +#ifdef CONFIG_REISERFS_CHECK + if (mode != M_DELETE) + reiserfs_panic (0, "vs-8010: old_item_num: mode must be M_DELETE (mode = \'%c\'", mode); +#endif + + /* delete mode */ + return new_num + 1; +} + +static void create_virtual_node (struct tree_balance * tb, int h) +{ + struct item_head * ih; + struct virtual_node * vn = tb->tb_vn; + int new_num; + struct buffer_head * Sh; /* this comes from tb->S[h] */ + + Sh = PATH_H_PBUFFER (tb->tb_path, h); + + /* size of changed node */ + vn->vn_size = MAX_CHILD_SIZE (Sh) - B_FREE_SPACE (Sh) + tb->insert_size[h]; + + /* for internal nodes array if virtual items is not created */ + if (h) { + vn->vn_nr_item = (vn->vn_size - DC_SIZE) / (DC_SIZE + KEY_SIZE); + return; + } + + /* number of items in virtual node */ + vn->vn_nr_item = B_NR_ITEMS (Sh) + ((vn->vn_mode == M_INSERT)? 1 : 0) - ((vn->vn_mode == M_DELETE)? 1 : 0); + + /* first virtual item */ + vn->vn_vi = (struct virtual_item *)(tb->tb_vn + 1); + memset (vn->vn_vi, 0, vn->vn_nr_item * sizeof (struct virtual_item)); + vn->vn_free_ptr += vn->vn_nr_item * sizeof (struct virtual_item); + + + /* first item in the node */ + ih = B_N_PITEM_HEAD (Sh, 0); + + /* define the mergeability for 0-th item (if it is not being deleted) */ +#ifdef REISERFS_FSCK + if (is_left_mergeable (tb->tb_sb, tb->tb_path) == 1 && (vn->vn_mode != M_DELETE || vn->vn_affected_item_num)) +#else + if (op_is_left_mergeable (&(ih->ih_key), Sh->b_size) && (vn->vn_mode != M_DELETE || vn->vn_affected_item_num)) +#endif + vn->vn_vi[0].vi_type |= VI_TYPE_LEFT_MERGEABLE; + + /* go through all items those remain in the virtual node (except for the new (inserted) one) */ + for (new_num = 0; new_num < vn->vn_nr_item; new_num ++) { + int j; + struct virtual_item * vi = vn->vn_vi + new_num; + int is_affected = ((new_num != vn->vn_affected_item_num) ? 0 : 1); + + + if (is_affected && vn->vn_mode == M_INSERT) + continue; + + /* get item number in source node */ + j = old_item_num (new_num, vn->vn_affected_item_num, vn->vn_mode); + + vi->vi_item_len += ih[j].ih_item_len + IH_SIZE; + vi->vi_ih = ih + j; + vi->vi_item = B_I_PITEM (Sh, ih + j); + vi->vi_uarea = vn->vn_free_ptr; + + // FIXME: there is no check, that item operation did not + // consume too much memory + vn->vn_free_ptr += op_create_vi (vn, vi, is_affected, tb->insert_size [0]); + if (tb->vn_buf + tb->vn_buf_size < vn->vn_free_ptr) + reiserfs_panic (tb->tb_sb, "vs-8030: create_virtual_node: " + "virtual node space consumed"); + + if (!is_affected) + /* this is not being changed */ + continue; + + if (vn->vn_mode == M_PASTE || vn->vn_mode == M_CUT) { + vn->vn_vi[new_num].vi_item_len += tb->insert_size[0]; + vi->vi_new_data = vn->vn_data; // pointer to data which is going to be pasted + } + } + + + /* virtual inserted item is not defined yet */ + if (vn->vn_mode == M_INSERT) { + struct virtual_item * vi = vn->vn_vi + vn->vn_affected_item_num; + +#ifdef CONFIG_REISERFS_CHECK + if (vn->vn_ins_ih == 0) + reiserfs_panic (0, "vs-8040: create_virtual_node: item header of inserted item is not specified"); +#endif + + vi->vi_item_len = tb->insert_size[0]; + vi->vi_ih = vn->vn_ins_ih; + vi->vi_item = vn->vn_data; + vi->vi_uarea = vn->vn_free_ptr; + + op_create_vi (vn, vi, 0/*not pasted or cut*/, tb->insert_size [0]); +#if 0 + switch (type/*le_key_k_type (ih_version (vn->vn_ins_ih), &(vn->vn_ins_ih->ih_key))*/) { + case TYPE_STAT_DATA: + vn->vn_vi[vn->vn_affected_item_num].vi_type |= VI_TYPE_STAT_DATA; + break; + case TYPE_DIRECT: + vn->vn_vi[vn->vn_affected_item_num].vi_type |= VI_TYPE_DIRECT; + break; + case TYPE_INDIRECT: + vn->vn_vi[vn->vn_affected_item_num].vi_type |= VI_TYPE_INDIRECT; + break; + default: + /* inseted item is directory (it must be item with "." and "..") */ + vn->vn_vi[vn->vn_affected_item_num].vi_type |= + (VI_TYPE_DIRECTORY | VI_TYPE_FIRST_DIRECTORY_ITEM | VI_TYPE_INSERTED_DIRECTORY_ITEM); + + /* this directory item can not be split, so do not set sizes of entries */ + break; + } +#endif + } + + /* set right merge flag we take right delimiting key and check whether it is a mergeable item */ + if (tb->CFR[0]) { + struct key * key; + + key = B_N_PDELIM_KEY (tb->CFR[0], tb->rkey[0]); +#ifdef REISERFS_FSCK + if (is_right_mergeable (tb->tb_sb, tb->tb_path) == 1 && (vn->vn_mode != M_DELETE || + vn->vn_affected_item_num != B_NR_ITEMS (Sh) - 1)) +#else + if (op_is_left_mergeable (key, Sh->b_size) && (vn->vn_mode != M_DELETE || + vn->vn_affected_item_num != B_NR_ITEMS (Sh) - 1)) +#endif + vn->vn_vi[vn->vn_nr_item-1].vi_type |= VI_TYPE_RIGHT_MERGEABLE; + +#ifdef CONFIG_REISERFS_CHECK + if (op_is_left_mergeable (key, Sh->b_size) && + !(vn->vn_mode != M_DELETE || vn->vn_affected_item_num != B_NR_ITEMS (Sh) - 1) ) { + /* we delete last item and it could be merged with right neighbor's first item */ + if (!(B_NR_ITEMS (Sh) == 1 && is_direntry_le_ih (B_N_PITEM_HEAD (Sh, 0)) && + I_ENTRY_COUNT (B_N_PITEM_HEAD (Sh, 0)) == 1)) { + /* node contains more than 1 item, or item is not directory item, or this item contains more than 1 entry */ + print_block (Sh, 0, -1, -1); + reiserfs_panic (tb->tb_sb, "vs-8045: create_virtual_node: rdkey %k, affected item==%d (mode==%c) Must be %c", + key, vn->vn_affected_item_num, vn->vn_mode, M_DELETE); + } else + /* we can delete directory item, that has only one directory entry in it */ + ; + } +#endif + + } +} + + +/* using virtual node check, how many items can be shifted to left + neighbor */ +static void check_left (struct tree_balance * tb, int h, int cur_free) +{ + int i; + struct virtual_node * vn = tb->tb_vn; + struct virtual_item * vi; + int d_size, ih_size; + +#ifdef CONFIG_REISERFS_CHECK + if (cur_free < 0) + reiserfs_panic (0, "vs-8050: check_left: cur_free (%d) < 0", cur_free); +#endif + + /* internal level */ + if (h > 0) { + tb->lnum[h] = cur_free / (DC_SIZE + KEY_SIZE); + return; + } + + /* leaf level */ + + if (!cur_free || !vn->vn_nr_item) { + /* no free space or nothing to move */ + tb->lnum[h] = 0; + tb->lbytes = -1; + return; + } + +#ifdef CONFIG_REISERFS_CHECK + if (!PATH_H_PPARENT (tb->tb_path, 0)) + reiserfs_panic (0, "vs-8055: check_left: parent does not exist or invalid"); +#endif + + vi = vn->vn_vi; + if ((unsigned int)cur_free >= (vn->vn_size - ((vi->vi_type & VI_TYPE_LEFT_MERGEABLE) ? IH_SIZE : 0))) { + /* all contents of S[0] fits into L[0] */ + +#ifdef CONFIG_REISERFS_CHECK + if (vn->vn_mode == M_INSERT || vn->vn_mode == M_PASTE) + reiserfs_panic (0, "vs-8055: check_left: invalid mode or balance condition failed"); +#endif + + tb->lnum[0] = vn->vn_nr_item; + tb->lbytes = -1; + return; + } + + + d_size = 0, ih_size = IH_SIZE; + + /* first item may be merge with last item in left neighbor */ + if (vi->vi_type & VI_TYPE_LEFT_MERGEABLE) + d_size = -((int)IH_SIZE), ih_size = 0; + + tb->lnum[0] = 0; + for (i = 0; i < vn->vn_nr_item; i ++, ih_size = IH_SIZE, d_size = 0, vi ++) { + d_size += vi->vi_item_len; + if (cur_free >= d_size) { + /* the item can be shifted entirely */ + cur_free -= d_size; + tb->lnum[0] ++; + continue; + } + + /* the item cannot be shifted entirely, try to split it */ + /* check whether L[0] can hold ih and at least one byte of the item body */ + if (cur_free <= ih_size) { + /* cannot shift even a part of the current item */ + tb->lbytes = -1; + return; + } + cur_free -= ih_size; + + tb->lbytes = op_check_left (vi, cur_free, 0, 0); + if (tb->lbytes != -1) + /* count partially shifted item */ + tb->lnum[0] ++; + + break; + } + + return; +} + + +/* using virtual node check, how many items can be shifted to right + neighbor */ +static void check_right (struct tree_balance * tb, int h, int cur_free) +{ + int i; + struct virtual_node * vn = tb->tb_vn; + struct virtual_item * vi; + int d_size, ih_size; + +#ifdef CONFIG_REISERFS_CHECK + if (cur_free < 0) + reiserfs_panic (tb->tb_sb, "vs-8070: check_right: cur_free < 0"); +#endif + + /* internal level */ + if (h > 0) { + tb->rnum[h] = cur_free / (DC_SIZE + KEY_SIZE); + return; + } + + /* leaf level */ + + if (!cur_free || !vn->vn_nr_item) { + /* no free space */ + tb->rnum[h] = 0; + tb->rbytes = -1; + return; + } + +#ifdef CONFIG_REISERFS_CHECK + if (!PATH_H_PPARENT (tb->tb_path, 0)) + reiserfs_panic (tb->tb_sb, "vs-8075: check_right: parent does not exist or invalid"); +#endif + + vi = vn->vn_vi + vn->vn_nr_item - 1; + if ((unsigned int)cur_free >= (vn->vn_size - ((vi->vi_type & VI_TYPE_RIGHT_MERGEABLE) ? IH_SIZE : 0))) { + /* all contents of S[0] fits into R[0] */ + +#ifdef CONFIG_REISERFS_CHECK + if (vn->vn_mode == M_INSERT || vn->vn_mode == M_PASTE) + reiserfs_panic (tb->tb_sb, "vs-8080: check_right: invalid mode or balance condition failed"); +#endif + + tb->rnum[h] = vn->vn_nr_item; + tb->rbytes = -1; + return; + } + + d_size = 0, ih_size = IH_SIZE; + + /* last item may be merge with first item in right neighbor */ + if (vi->vi_type & VI_TYPE_RIGHT_MERGEABLE) + d_size = -(int)IH_SIZE, ih_size = 0; + + tb->rnum[0] = 0; + for (i = vn->vn_nr_item - 1; i >= 0; i --, d_size = 0, ih_size = IH_SIZE, vi --) { + d_size += vi->vi_item_len; + if (cur_free >= d_size) { + /* the item can be shifted entirely */ + cur_free -= d_size; + tb->rnum[0] ++; + continue; + } + + /* check whether R[0] can hold ih and at least one byte of the item body */ + if ( cur_free <= ih_size ) { /* cannot shift even a part of the current item */ + tb->rbytes = -1; + return; + } + + /* R[0] can hold the header of the item and at least one byte of its body */ + cur_free -= ih_size; /* cur_free is still > 0 */ + + tb->rbytes = op_check_right (vi, cur_free); + if (tb->rbytes != -1) + /* count partially shifted item */ + tb->rnum[0] ++; + + break; + } + + return; +} + + +/* + * from - number of items, which are shifted to left neighbor entirely + * to - number of item, which are shifted to right neighbor entirely + * from_bytes - number of bytes of boundary item (or directory entries) which are shifted to left neighbor + * to_bytes - number of bytes of boundary item (or directory entries) which are shifted to right neighbor */ +static int get_num_ver (int mode, struct tree_balance * tb, int h, + int from, int from_bytes, + int to, int to_bytes, + short * snum012, int flow + ) +{ + int i; + int cur_free; + // int bytes; + int units; + struct virtual_node * vn = tb->tb_vn; + // struct virtual_item * vi; + + int total_node_size, max_node_size, current_item_size; + int needed_nodes; + int start_item, /* position of item we start filling node from */ + end_item, /* position of item we finish filling node by */ + start_bytes,/* number of first bytes (entries for directory) of start_item-th item + we do not include into node that is being filled */ + end_bytes; /* number of last bytes (entries for directory) of end_item-th item + we do node include into node that is being filled */ + int split_item_positions[2]; /* these are positions in virtual item of + items, that are split between S[0] and + S1new and S1new and S2new */ + + split_item_positions[0] = -1; + split_item_positions[1] = -1; + +#ifdef CONFIG_REISERFS_CHECK + /* We only create additional nodes if we are in insert or paste mode + or we are in replace mode at the internal level. If h is 0 and + the mode is M_REPLACE then in fix_nodes we change the mode to + paste or insert before we get here in the code. */ + if ( tb->insert_size[h] < 0 || (mode != M_INSERT && mode != M_PASTE)) + reiserfs_panic (0, "vs-8100: get_num_ver: insert_size < 0 in overflow"); +#endif + + max_node_size = MAX_CHILD_SIZE (PATH_H_PBUFFER (tb->tb_path, h)); + + /* snum012 [0-2] - number of items, that lay + to S[0], first new node and second new node */ + snum012[3] = -1; /* s1bytes */ + snum012[4] = -1; /* s2bytes */ + + /* internal level */ + if (h > 0) { + i = ((to - from) * (KEY_SIZE + DC_SIZE) + DC_SIZE); + if (i == max_node_size) + return 1; + return (i / max_node_size + 1); + } + + /* leaf level */ + needed_nodes = 1; + total_node_size = 0; + cur_free = max_node_size; + + // start from 'from'-th item + start_item = from; + // skip its first 'start_bytes' units + start_bytes = ((from_bytes != -1) ? from_bytes : 0); + + // last included item is the 'end_item'-th one + end_item = vn->vn_nr_item - to - 1; + // do not count last 'end_bytes' units of 'end_item'-th item + end_bytes = (to_bytes != -1) ? to_bytes : 0; + + /* go through all item begining from the start_item-th item and ending by + the end_item-th item. Do not count first 'start_bytes' units of + 'start_item'-th item and last 'end_bytes' of 'end_item'-th item */ + + for (i = start_item; i <= end_item; i ++) { + struct virtual_item * vi = vn->vn_vi + i; + int skip_from_end = ((i == end_item) ? end_bytes : 0); + +#ifdef CONFIG_REISERFS_CHECK + if (needed_nodes > 3) { + reiserfs_panic (tb->tb_sb, "vs-8105: get_num_ver: too many nodes are needed"); + } +#endif + + /* get size of current item */ + current_item_size = vi->vi_item_len; + + /* do not take in calculation head part (from_bytes) of from-th item */ + current_item_size -= op_part_size (vi, 0/*from start*/, start_bytes); + + /* do not take in calculation tail part of last item */ + current_item_size -= op_part_size (vi, 1/*from end*/, skip_from_end); + + /* if item fits into current node entierly */ + if (total_node_size + current_item_size <= max_node_size) { + snum012[needed_nodes - 1] ++; + total_node_size += current_item_size; + start_bytes = 0; + continue; + } + + if (current_item_size > max_node_size) { + /* virtual item length is longer, than max size of item in + a node. It is impossible for direct item */ +#ifdef CONFIG_REISERFS_CHECK + if (is_direct_le_ih (vi->vi_ih)) + reiserfs_panic (tb->tb_sb, "vs-8110: get_num_ver: " + "direct item length is %d. It can not be longer than %d", + current_item_size, max_node_size); +#endif + /* we will try to split it */ + flow = 1; + } + + if (!flow) { + /* as we do not split items, take new node and continue */ + needed_nodes ++; i --; total_node_size = 0; + continue; + } + + // calculate number of item units which fit into node being + // filled + { + int free_space; + + free_space = max_node_size - total_node_size - IH_SIZE; + units = op_check_left (vi, free_space, start_bytes, skip_from_end); + if (units == -1) { + /* nothing fits into current node, take new node and continue */ + needed_nodes ++, i--, total_node_size = 0; + continue; + } + } + + /* something fits into the current node */ + //if (snum012[3] != -1 || needed_nodes != 1) + // reiserfs_panic (tb->tb_sb, "vs-8115: get_num_ver: too many nodes required"); + //snum012[needed_nodes - 1 + 3] = op_unit_num (vi) - start_bytes - units; + start_bytes += units; + snum012[needed_nodes - 1 + 3] = units; + + if (needed_nodes > 2) + reiserfs_warning ("vs-8111: get_num_ver: split_item_position is out of boundary\n"); + snum012[needed_nodes - 1] ++; + split_item_positions[needed_nodes - 1] = i; + needed_nodes ++; + /* continue from the same item with start_bytes != -1 */ + start_item = i; + i --; + total_node_size = 0; + } + + // sum012[4] (if it is not -1) contains number of units of which + // are to be in S1new, snum012[3] - to be in S0. They are supposed + // to be S1bytes and S2bytes correspondingly, so recalculate + if (snum012[4] > 0) { + int split_item_num; + int bytes_to_r, bytes_to_l; + int bytes_to_S1new; + + split_item_num = split_item_positions[1]; + bytes_to_l = ((from == split_item_num && from_bytes != -1) ? from_bytes : 0); + bytes_to_r = ((end_item == split_item_num && end_bytes != -1) ? end_bytes : 0); + bytes_to_S1new = ((split_item_positions[0] == split_item_positions[1]) ? snum012[3] : 0); + + // s2bytes + snum012[4] = op_unit_num (&vn->vn_vi[split_item_num]) - snum012[4] - bytes_to_r - bytes_to_l - bytes_to_S1new; + + if (vn->vn_vi[split_item_num].vi_index != TYPE_DIRENTRY) + reiserfs_warning ("vs-8115: get_num_ver: not directory item\n"); + } + + /* now we know S2bytes, calculate S1bytes */ + if (snum012[3] > 0) { + int split_item_num; + int bytes_to_r, bytes_to_l; + int bytes_to_S2new; + + split_item_num = split_item_positions[0]; + bytes_to_l = ((from == split_item_num && from_bytes != -1) ? from_bytes : 0); + bytes_to_r = ((end_item == split_item_num && end_bytes != -1) ? end_bytes : 0); + bytes_to_S2new = ((split_item_positions[0] == split_item_positions[1] && snum012[4] != -1) ? snum012[4] : 0); + + // s1bytes + snum012[3] = op_unit_num (&vn->vn_vi[split_item_num]) - snum012[3] - bytes_to_r - bytes_to_l - bytes_to_S2new; + } + + return needed_nodes; +} + + +#ifdef CONFIG_REISERFS_CHECK +extern struct tree_balance * cur_tb; +#endif + + +/* Set parameters for balancing. + * Performs write of results of analysis of balancing into structure tb, + * where it will later be used by the functions that actually do the balancing. + * Parameters: + * tb tree_balance structure; + * h current level of the node; + * lnum number of items from S[h] that must be shifted to L[h]; + * rnum number of items from S[h] that must be shifted to R[h]; + * blk_num number of blocks that S[h] will be splitted into; + * s012 number of items that fall into splitted nodes. + * lbytes number of bytes which flow to the left neighbor from the item that is not + * not shifted entirely + * rbytes number of bytes which flow to the right neighbor from the item that is not + * not shifted entirely + * s1bytes number of bytes which flow to the first new node when S[0] splits (this number is contained in s012 array) + */ + +static void set_parameters (struct tree_balance * tb, int h, int lnum, + int rnum, int blk_num, short * s012, int lb, int rb) +{ + + tb->lnum[h] = lnum; + tb->rnum[h] = rnum; + tb->blknum[h] = blk_num; + + if (h == 0) + { /* only for leaf level */ + if (s012 != NULL) + { + tb->s0num = * s012 ++, + tb->s1num = * s012 ++, + tb->s2num = * s012 ++; + tb->s1bytes = * s012 ++; + tb->s2bytes = * s012; + } + tb->lbytes = lb; + tb->rbytes = rb; + } +} + + + +/* check, does node disappear if we shift tb->lnum[0] items to left + neighbor and tb->rnum[0] to the right one. */ +static int is_leaf_removable (struct tree_balance * tb) +{ + struct virtual_node * vn = tb->tb_vn; + int to_left, to_right; + int size; + int remain_items; + + /* number of items, that will be shifted to left (right) neighbor + entirely */ + to_left = tb->lnum[0] - ((tb->lbytes != -1) ? 1 : 0); + to_right = tb->rnum[0] - ((tb->rbytes != -1) ? 1 : 0); + remain_items = vn->vn_nr_item; + + /* how many items remain in S[0] after shiftings to neighbors */ + remain_items -= (to_left + to_right); + + if (remain_items < 1) { + /* all content of node can be shifted to neighbors */ + set_parameters (tb, 0, to_left, vn->vn_nr_item - to_left, 0, NULL, -1, -1); + return 1; + } + + if (remain_items > 1 || tb->lbytes == -1 || tb->rbytes == -1) + /* S[0] is not removable */ + return 0; + + /* check, whether we can divide 1 remaining item between neighbors */ + + /* get size of remaining item (in item units) */ + size = op_unit_num (&(vn->vn_vi[to_left])); + + if (tb->lbytes + tb->rbytes >= size) { + set_parameters (tb, 0, to_left + 1, to_right + 1, 0, NULL, tb->lbytes, -1); + return 1; + } + + return 0; +} + + +/* check whether L, S, R can be joined in one node */ +static int are_leaves_removable (struct tree_balance * tb, int lfree, int rfree) +{ + struct virtual_node * vn = tb->tb_vn; + int ih_size; + struct buffer_head *S0; + + S0 = PATH_H_PBUFFER (tb->tb_path, 0); + + ih_size = 0; + if (vn->vn_nr_item) { + if (vn->vn_vi[0].vi_type & VI_TYPE_LEFT_MERGEABLE) + ih_size += IH_SIZE; + + if (vn->vn_vi[vn->vn_nr_item-1].vi_type & VI_TYPE_RIGHT_MERGEABLE) + ih_size += IH_SIZE; + } else { + /* there was only one item and it will be deleted */ + struct item_head * ih; + +#ifdef CONFIG_REISERFS_CHECK + if (B_NR_ITEMS (S0) != 1) + reiserfs_panic (0, "vs-8125: are_leaves_removable: item number must be 1: it is %d", B_NR_ITEMS(S0)); +#endif + + ih = B_N_PITEM_HEAD (S0, 0); + if (tb->CFR[0] && !comp_short_le_keys (&(ih->ih_key), B_N_PDELIM_KEY (tb->CFR[0], tb->rkey[0]))) + if (is_direntry_le_ih (ih)) { +#ifndef REISERFS_FSCK + + /* Directory must be in correct state here: that is + somewhere at the left side should exist first directory + item. But the item being deleted can not be that first + one because its right neighbor is item of the same + directory. (But first item always gets deleted in last + turn). So, neighbors of deleted item can be merged, so + we can save ih_size */ + ih_size = IH_SIZE; + +#ifdef CONFIG_REISERFS_CHECK + /* we might check that left neighbor exists and is of the + same directory */ + if (le_key_k_offset (ih_version (ih), &(ih->ih_key)) == DOT_OFFSET) + reiserfs_panic (tb->tb_sb, "vs-8130: are_leaves_removable: " + "first directory item can not be removed until directory is not empty"); +#endif + + +#else /* REISERFS_FSCK */ + + /* we can delete any directory item in fsck (if it is unreachable) */ + if (ih->ih_key.k_offset != DOT_OFFSET) { + /* must get left neighbor here to make sure, that left + neighbor is of the same directory */ + struct buffer_head * left; + + left = get_left_neighbor (tb->tb_sb, tb->tb_path); + if (left) { + struct item_head * last; + + if (B_NR_ITEMS (left) == 0) + reiserfs_panic (tb->tb_sb, "vs-8135: are_leaves_removable: " + "empty node in the tree"); + last = B_N_PITEM_HEAD (left, B_NR_ITEMS (left) - 1); + if (!comp_short_keys (&last->ih_key, &ih->ih_key)) + ih_size = IH_SIZE; + brelse (left); + } + } +#endif + } + + } + + if (MAX_CHILD_SIZE (S0) + vn->vn_size <= rfree + lfree + ih_size) { + set_parameters (tb, 0, -1, -1, -1, NULL, -1, -1); + return 1; + } + return 0; + +} + + + +/* when we do not split item, lnum and rnum are numbers of entire items */ +#define SET_PAR_SHIFT_LEFT \ +if (h)\ +{\ + int to_l;\ + \ + to_l = (MAX_NR_KEY(Sh)+1 - lpar + vn->vn_nr_item + 1) / 2 -\ + (MAX_NR_KEY(Sh) + 1 - lpar);\ + \ + set_parameters (tb, h, to_l, 0, lnver, NULL, -1, -1);\ +}\ +else \ +{\ + if (lset==LEFT_SHIFT_FLOW)\ + set_parameters (tb, h, lpar, 0, lnver, snum012+lset,\ + tb->lbytes, -1);\ + else\ + set_parameters (tb, h, lpar - (tb->lbytes!=-1), 0, lnver, snum012+lset,\ + -1, -1);\ +} + + +#define SET_PAR_SHIFT_RIGHT \ +if (h)\ +{\ + int to_r;\ + \ + to_r = (MAX_NR_KEY(Sh)+1 - rpar + vn->vn_nr_item + 1) / 2 - (MAX_NR_KEY(Sh) + 1 - rpar);\ + \ + set_parameters (tb, h, 0, to_r, rnver, NULL, -1, -1);\ +}\ +else \ +{\ + if (rset==RIGHT_SHIFT_FLOW)\ + set_parameters (tb, h, 0, rpar, rnver, snum012+rset,\ + -1, tb->rbytes);\ + else\ + set_parameters (tb, h, 0, rpar - (tb->rbytes!=-1), rnver, snum012+rset,\ + -1, -1);\ +} + + +void free_buffers_in_tb ( + struct tree_balance * p_s_tb + ) { + int n_counter; + + decrement_counters_in_path(p_s_tb->tb_path); + + for ( n_counter = 0; n_counter < MAX_HEIGHT; n_counter++ ) { + decrement_bcount(p_s_tb->L[n_counter]); + p_s_tb->L[n_counter] = NULL; + decrement_bcount(p_s_tb->R[n_counter]); + p_s_tb->R[n_counter] = NULL; + decrement_bcount(p_s_tb->FL[n_counter]); + p_s_tb->FL[n_counter] = NULL; + decrement_bcount(p_s_tb->FR[n_counter]); + p_s_tb->FR[n_counter] = NULL; + decrement_bcount(p_s_tb->CFL[n_counter]); + p_s_tb->CFL[n_counter] = NULL; + decrement_bcount(p_s_tb->CFR[n_counter]); + p_s_tb->CFR[n_counter] = NULL; + } +} + + +/* Get new buffers for storing new nodes that are created while balancing. + * Returns: SCHEDULE_OCCURED - schedule occured while the function worked; + * CARRY_ON - schedule didn't occur while the function worked; + * NO_DISK_SPACE - no disk space. + */ +/* The function is NOT SCHEDULE-SAFE! */ +static int get_empty_nodes( + struct tree_balance * p_s_tb, + int n_h + ) { + struct buffer_head * p_s_new_bh, + * p_s_Sh = PATH_H_PBUFFER (p_s_tb->tb_path, n_h); + unsigned long * p_n_blocknr, + a_n_blocknrs[MAX_AMOUNT_NEEDED] = {0, }; + int n_counter, + n_number_of_freeblk, + n_amount_needed,/* number of needed empty blocks */ + n_retval = CARRY_ON; + struct super_block * p_s_sb = p_s_tb->tb_sb; + + +#ifdef REISERFS_FSCK + if (n_h == 0 && p_s_tb->insert_size[n_h] == 0x7fff) + return CARRY_ON; +#endif + + /* number_of_freeblk is the number of empty blocks which have been + acquired for use by the balancing algorithm minus the number of + empty blocks used in the previous levels of the analysis, + number_of_freeblk = tb->cur_blknum can be non-zero if a schedule occurs + after empty blocks are acquired, and the balancing analysis is + then restarted, amount_needed is the number needed by this level + (n_h) of the balancing analysis. + + Note that for systems with many processes writing, it would be + more layout optimal to calculate the total number needed by all + levels and then to run reiserfs_new_blocks to get all of them at once. */ + + /* Initiate number_of_freeblk to the amount acquired prior to the restart of + the analysis or 0 if not restarted, then subtract the amount needed + by all of the levels of the tree below n_h. */ + /* blknum includes S[n_h], so we subtract 1 in this calculation */ + for ( n_counter = 0, n_number_of_freeblk = p_s_tb->cur_blknum; n_counter < n_h; n_counter++ ) + n_number_of_freeblk -= ( p_s_tb->blknum[n_counter] ) ? (p_s_tb->blknum[n_counter] - 1) : 0; + + /* Allocate missing empty blocks. */ + /* if p_s_Sh == 0 then we are getting a new root */ + n_amount_needed = ( p_s_Sh ) ? (p_s_tb->blknum[n_h] - 1) : 1; + /* Amount_needed = the amount that we need more than the amount that we have. */ + if ( n_amount_needed > n_number_of_freeblk ) + n_amount_needed -= n_number_of_freeblk; + else /* If we have enough already then there is nothing to do. */ + return CARRY_ON; + + if ( reiserfs_new_blocknrs (p_s_tb->transaction_handle, a_n_blocknrs, + PATH_PLAST_BUFFER(p_s_tb->tb_path)->b_blocknr, n_amount_needed) == NO_DISK_SPACE ) + return NO_DISK_SPACE; + + /* for each blocknumber we just got, get a buffer and stick it on FEB */ + for ( p_n_blocknr = a_n_blocknrs, n_counter = 0; n_counter < n_amount_needed; + p_n_blocknr++, n_counter++ ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( ! *p_n_blocknr ) + reiserfs_panic(p_s_sb, "PAP-8135: get_empty_nodes: reiserfs_new_blocknrs failed when got new blocks"); +#endif + + p_s_new_bh = reiserfs_getblk(p_s_sb->s_dev, *p_n_blocknr, p_s_sb->s_blocksize); + if (atomic_read (&(p_s_new_bh->b_count)) > 1) { +/*&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&*/ +/* + reiserfs_warning ("waiting for buffer %b, iput inode pid = %d, this pid %d, mode %c, %h\n", + p_s_new_bh, put_inode_pid, current->pid, p_s_tb->tb_vn->vn_mode, p_s_tb->tb_vn->vn_ins_ih); + print_tb (0, 0, 0, p_s_tb, "tb"); +*/ +/*&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&*/ + if (atomic_read(&(p_s_new_bh->b_count)) > 2 || + !(buffer_journaled(p_s_new_bh) || buffer_journal_dirty(p_s_new_bh))) { + n_retval = REPEAT_SEARCH ; + free_buffers_in_tb (p_s_tb); + wait_buffer_until_released (p_s_new_bh); + } + } +#ifdef CONFIG_REISERFS_CHECK + if (atomic_read (&(p_s_new_bh->b_count)) != 1 || buffer_dirty (p_s_new_bh)) { + if (atomic_read(&(p_s_new_bh->b_count)) > 2 || + !(buffer_journaled(p_s_new_bh) || buffer_journal_dirty(p_s_new_bh))) { + reiserfs_panic(p_s_sb,"PAP-8140: get_empty_nodes: not free or dirty buffer %b for the new block", + p_s_new_bh); + } + } +#endif + + /* Put empty buffers into the array. */ + if (p_s_tb->FEB[p_s_tb->cur_blknum]) + BUG(); + + p_s_tb->FEB[p_s_tb->cur_blknum++] = p_s_new_bh; + } + + if ( n_retval == CARRY_ON && FILESYSTEM_CHANGED_TB (p_s_tb) ) + n_retval = REPEAT_SEARCH ; + + return n_retval; +} + + +/* Get free space of the left neighbor, which is stored in the parent + * node of the left neighbor. */ +static int get_lfree (struct tree_balance * tb, int h) +{ + struct buffer_head * l, * f; + int order; + + if ((f = PATH_H_PPARENT (tb->tb_path, h)) == 0 || (l = tb->FL[h]) == 0) + return 0; + + if (f == l) + order = PATH_H_B_ITEM_ORDER (tb->tb_path, h) - 1; + else { + order = B_NR_ITEMS (l); + f = l; + } + + return (MAX_CHILD_SIZE(f) - le16_to_cpu (B_N_CHILD(f,order)->dc_size)); +} + + +/* Get free space of the right neighbor, + * which is stored in the parent node of the right neighbor. + */ +static int get_rfree (struct tree_balance * tb, int h) +{ + struct buffer_head * r, * f; + int order; + + if ((f = PATH_H_PPARENT (tb->tb_path, h)) == 0 || (r = tb->FR[h]) == 0) + return 0; + + if (f == r) + order = PATH_H_B_ITEM_ORDER (tb->tb_path, h) + 1; + else { + order = 0; + f = r; + } + + return (MAX_CHILD_SIZE(f) - B_N_CHILD(f,order)->dc_size); + +} + + +/* Check whether left neighbor is in memory. */ +static int is_left_neighbor_in_cache( + struct tree_balance * p_s_tb, + int n_h + ) { + struct buffer_head * p_s_father, * left; + struct super_block * p_s_sb = p_s_tb->tb_sb; + unsigned long n_left_neighbor_blocknr; + int n_left_neighbor_position; + + if ( ! p_s_tb->FL[n_h] ) /* Father of the left neighbor does not exist. */ + return 0; + + /* Calculate father of the node to be balanced. */ + p_s_father = PATH_H_PBUFFER(p_s_tb->tb_path, n_h + 1); + +#ifdef CONFIG_REISERFS_CHECK + if ( ! p_s_father || ! B_IS_IN_TREE (p_s_father) || ! B_IS_IN_TREE (p_s_tb->FL[n_h]) || + ! buffer_uptodate (p_s_father) || ! buffer_uptodate (p_s_tb->FL[n_h]) ) { + reiserfs_panic (p_s_sb, "vs-8165: is_left_neighbor_in_cache: F[h] (%b) or FL[h] (%b) is invalid", + p_s_father, p_s_tb->FL[n_h]); + } +#endif + + + /* Get position of the pointer to the left neighbor into the left father. */ + n_left_neighbor_position = ( p_s_father == p_s_tb->FL[n_h] ) ? + p_s_tb->lkey[n_h] : B_NR_ITEMS (p_s_tb->FL[n_h]); + /* Get left neighbor block number. */ + n_left_neighbor_blocknr = B_N_CHILD_NUM(p_s_tb->FL[n_h], n_left_neighbor_position); + /* Look for the left neighbor in the cache. */ + if ( (left = get_hash_table(p_s_sb->s_dev, n_left_neighbor_blocknr, p_s_sb->s_blocksize)) ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( buffer_uptodate (left) && ! B_IS_IN_TREE(left) ) { + reiserfs_panic(p_s_sb, "vs-8170: is_left_neighbor_in_cache: left neighbor (%b %z) is not in the tree", + left, left); + } +#endif + atomic_dec (&(left->b_count)); + return 1; + } + + return 0; +} + + +#define LEFT_PARENTS 'l' +#define RIGHT_PARENTS 'r' + + +static void decrement_key (struct cpu_key * p_s_key) +{ + // call item specific function for this key + item_ops[cpu_key_k_type (p_s_key)]->decrement_key (p_s_key); + + +#if 0 /* this works wrong when key is key of second part of tail: it + sets key to be of indirect type. It looks like it makes no + harm but it is unclear */ + + unsigned long * p_n_key_field = (unsigned long *)p_s_key + REISERFS_FULL_KEY_LEN - 1; + int n_counter; + + for( n_counter = 0; n_counter < REISERFS_FULL_KEY_LEN; n_counter++, p_n_key_field-- ) { + if ( *p_n_key_field ) { + (*p_n_key_field)--; + break; + } + } +#ifdef CONFIG_REISERFS_CHECK + if ( n_counter == REISERFS_FULL_KEY_LEN ) + reiserfs_panic(NULL, "PAP-8175: decrement_key: zero key"); +#endif + +#endif /*0*/ + +} + + + + +/* Calculate far left/right parent of the left/right neighbor of the current node, that + * is calculate the left/right (FL[h]/FR[h]) neighbor of the parent F[h]. + * Calculate left/right common parent of the current node and L[h]/R[h]. + * Calculate left/right delimiting key position. + * Returns: PATH_INCORRECT - path in the tree is not correct; + SCHEDULE_OCCURRED - schedule occured while the function worked; + * CARRY_ON - schedule didn't occur while the function worked; + */ +static int get_far_parent (struct tree_balance * p_s_tb, + int n_h, + struct buffer_head ** pp_s_father, + struct buffer_head ** pp_s_com_father, + char c_lr_par) +{ + struct buffer_head * p_s_parent; + INITIALIZE_PATH (s_path_to_neighbor_father); + struct path * p_s_path = p_s_tb->tb_path; + struct cpu_key s_lr_father_key; + int n_counter, + n_position = MAX_INT, + n_first_last_position = 0, + n_path_offset = PATH_H_PATH_OFFSET(p_s_path, n_h); + + /* Starting from F[n_h] go upwards in the tree, and look for the common + ancestor of F[n_h], and its neighbor l/r, that should be obtained. */ + + n_counter = n_path_offset; + +#ifdef CONFIG_REISERFS_CHECK + if ( n_counter < FIRST_PATH_ELEMENT_OFFSET ) + reiserfs_panic(p_s_tb->tb_sb, "PAP-8180: get_far_parent: invalid path length"); +#endif + + + for ( ; n_counter > FIRST_PATH_ELEMENT_OFFSET; n_counter-- ) { + /* Check whether parent of the current buffer in the path is really parent in the tree. */ + if ( ! B_IS_IN_TREE(p_s_parent = PATH_OFFSET_PBUFFER(p_s_path, n_counter - 1)) ) + return REPEAT_SEARCH; + /* Check whether position in the parent is correct. */ + if ( (n_position = PATH_OFFSET_POSITION(p_s_path, n_counter - 1)) > B_NR_ITEMS(p_s_parent) ) + return REPEAT_SEARCH; + /* Check whether parent at the path really points to the child. */ + if ( B_N_CHILD_NUM(p_s_parent, n_position) != + PATH_OFFSET_PBUFFER(p_s_path, n_counter)->b_blocknr ) + return REPEAT_SEARCH; + /* Return delimiting key if position in the parent is not equal to first/last one. */ + if ( c_lr_par == RIGHT_PARENTS ) + n_first_last_position = B_NR_ITEMS (p_s_parent); + if ( n_position != n_first_last_position ) { + *pp_s_com_father = p_s_parent; + atomic_inc (&((*pp_s_com_father)->b_count)); + /*(*pp_s_com_father = p_s_parent)->b_count++;*/ + break; + } + } + + /* if we are in the root of the tree, then there is no common father */ + if ( n_counter == FIRST_PATH_ELEMENT_OFFSET ) { + /* Check whether first buffer in the path is the root of the tree. */ + if ( PATH_OFFSET_PBUFFER(p_s_tb->tb_path, FIRST_PATH_ELEMENT_OFFSET)->b_blocknr == + SB_ROOT_BLOCK (p_s_tb->tb_sb) ) { + *pp_s_father = *pp_s_com_father = NULL; + return CARRY_ON; + } + return REPEAT_SEARCH; + } + +#ifdef CONFIG_REISERFS_CHECK + if ( B_LEVEL (*pp_s_com_father) <= DISK_LEAF_NODE_LEVEL ) { + reiserfs_panic(p_s_tb->tb_sb, "PAP-8185: get_far_parent: (%b %z) level too small", *pp_s_com_father, *pp_s_com_father); + } +#endif + + /* Check whether the common parent is locked. */ + + if ( buffer_locked (*pp_s_com_father) ) { + __wait_on_buffer(*pp_s_com_father); + if ( FILESYSTEM_CHANGED_TB (p_s_tb) ) { + decrement_bcount(*pp_s_com_father); + return REPEAT_SEARCH; + } + } + + /* So, we got common parent of the current node and its left/right neighbor. + Now we are geting the parent of the left/right neighbor. */ + + /* Form key to get parent of the left/right neighbor. */ + le_key2cpu_key (&s_lr_father_key, B_N_PDELIM_KEY(*pp_s_com_father, ( c_lr_par == LEFT_PARENTS ) ? + (p_s_tb->lkey[n_h - 1] = n_position - 1) : (p_s_tb->rkey[n_h - 1] = n_position))); + + + if ( c_lr_par == LEFT_PARENTS ) + decrement_key(&s_lr_father_key); + + if (search_by_key(p_s_tb->tb_sb, &s_lr_father_key, &s_path_to_neighbor_father, n_h + 1) == IO_ERROR) + // path is released + return IO_ERROR; + + if ( FILESYSTEM_CHANGED_TB (p_s_tb) ) { + decrement_counters_in_path(&s_path_to_neighbor_father); + decrement_bcount(*pp_s_com_father); + return REPEAT_SEARCH; + } + + *pp_s_father = PATH_PLAST_BUFFER(&s_path_to_neighbor_father); + +#ifdef CONFIG_REISERFS_CHECK + if ( B_LEVEL (*pp_s_father) != n_h + 1 ) { + reiserfs_panic(p_s_tb->tb_sb, "PAP-8190: get_far_parent: (%b %z) level too small", *pp_s_father, *pp_s_father); + } + + if ( s_path_to_neighbor_father.path_length < FIRST_PATH_ELEMENT_OFFSET ) + reiserfs_panic(0, "PAP-8192: get_far_parent: path length is too small"); + +#endif + + s_path_to_neighbor_father.path_length--; + decrement_counters_in_path(&s_path_to_neighbor_father); + return CARRY_ON; +} + + +/* Get parents of neighbors of node in the path(S[n_path_offset]) and common parents of + * S[n_path_offset] and L[n_path_offset]/R[n_path_offset]: F[n_path_offset], FL[n_path_offset], + * FR[n_path_offset], CFL[n_path_offset], CFR[n_path_offset]. + * Calculate numbers of left and right delimiting keys position: lkey[n_path_offset], rkey[n_path_offset]. + * Returns: SCHEDULE_OCCURRED - schedule occured while the function worked; + * CARRY_ON - schedule didn't occur while the function worked; + */ +static int get_parents (struct tree_balance * p_s_tb, int n_h) +{ + struct path * p_s_path = p_s_tb->tb_path; + int n_position, + n_ret_value, + n_path_offset = PATH_H_PATH_OFFSET(p_s_tb->tb_path, n_h); + struct buffer_head * p_s_curf, + * p_s_curcf; + + /* Current node is the root of the tree or will be root of the tree */ + if ( n_path_offset <= FIRST_PATH_ELEMENT_OFFSET ) { + /* The root can not have parents. + Release nodes which previously were obtained as parents of the current node neighbors. */ + decrement_bcount(p_s_tb->FL[n_h]); + decrement_bcount(p_s_tb->CFL[n_h]); + decrement_bcount(p_s_tb->FR[n_h]); + decrement_bcount(p_s_tb->CFR[n_h]); + p_s_tb->FL[n_h] = p_s_tb->CFL[n_h] = p_s_tb->FR[n_h] = p_s_tb->CFR[n_h] = NULL; + return CARRY_ON; + } + + /* Get parent FL[n_path_offset] of L[n_path_offset]. */ + if ( (n_position = PATH_OFFSET_POSITION(p_s_path, n_path_offset - 1)) ) { + /* Current node is not the first child of its parent. */ + /*(p_s_curf = p_s_curcf = PATH_OFFSET_PBUFFER(p_s_path, n_path_offset - 1))->b_count += 2;*/ + p_s_curf = p_s_curcf = PATH_OFFSET_PBUFFER(p_s_path, n_path_offset - 1); + atomic_inc (&(p_s_curf->b_count)); + atomic_inc (&(p_s_curf->b_count)); + p_s_tb->lkey[n_h] = n_position - 1; + } + else { + /* Calculate current parent of L[n_path_offset], which is the left neighbor of the current node. + Calculate current common parent of L[n_path_offset] and the current node. Note that + CFL[n_path_offset] not equal FL[n_path_offset] and CFL[n_path_offset] not equal F[n_path_offset]. + Calculate lkey[n_path_offset]. */ + if ( (n_ret_value = get_far_parent(p_s_tb, n_h + 1, &p_s_curf, + &p_s_curcf, LEFT_PARENTS)) != CARRY_ON ) + return n_ret_value; + } + + decrement_bcount(p_s_tb->FL[n_h]); + p_s_tb->FL[n_h] = p_s_curf; /* New initialization of FL[n_h]. */ + decrement_bcount(p_s_tb->CFL[n_h]); + p_s_tb->CFL[n_h] = p_s_curcf; /* New initialization of CFL[n_h]. */ + +#ifdef CONFIG_REISERFS_CHECK + if ((p_s_curf && !B_IS_IN_TREE (p_s_curf)) || (p_s_curcf && !B_IS_IN_TREE (p_s_curcf))) { + reiserfs_panic (p_s_tb->tb_sb, "PAP-8195: get_parents: FL (%b) or CFL (%b) is invalid", p_s_curf, p_s_curcf); + } +#endif + +/* Get parent FR[n_h] of R[n_h]. */ + +/* Current node is the last child of F[n_h]. FR[n_h] != F[n_h]. */ + if ( n_position == B_NR_ITEMS (PATH_H_PBUFFER(p_s_path, n_h + 1)) ) { +/* Calculate current parent of R[n_h], which is the right neighbor of F[n_h]. + Calculate current common parent of R[n_h] and current node. Note that CFR[n_h] + not equal FR[n_path_offset] and CFR[n_h] not equal F[n_h]. */ + if ( (n_ret_value = get_far_parent(p_s_tb, n_h + 1, &p_s_curf, &p_s_curcf, RIGHT_PARENTS)) != CARRY_ON ) + return n_ret_value; + } + else { +/* Current node is not the last child of its parent F[n_h]. */ + /*(p_s_curf = p_s_curcf = PATH_OFFSET_PBUFFER(p_s_path, n_path_offset - 1))->b_count += 2;*/ + p_s_curf = p_s_curcf = PATH_OFFSET_PBUFFER(p_s_path, n_path_offset - 1); + atomic_inc (&(p_s_curf->b_count)); + atomic_inc (&(p_s_curf->b_count)); + p_s_tb->rkey[n_h] = n_position; + } + + decrement_bcount(p_s_tb->FR[n_h]); + p_s_tb->FR[n_h] = p_s_curf; /* New initialization of FR[n_path_offset]. */ + + decrement_bcount(p_s_tb->CFR[n_h]); + p_s_tb->CFR[n_h] = p_s_curcf; /* New initialization of CFR[n_path_offset]. */ + +#ifdef CONFIG_REISERFS_CHECK +#if 0 + if (n_h == 0 && p_s_tb->CFR[n_h] && COMP_KEYS (B_PRIGHT_DELIM_KEY (PATH_H_PBUFFER(p_s_path, n_h)), + B_N_PDELIM_KEY (p_s_tb->CFR[n_h], p_s_tb->rkey[n_h]))) { + reiserfs_panic (p_s_tb->tb_sb, "PAP-8200: get_parents: rdkey in S0 %k, rdkey in CFR0 %k do not match", + B_PRIGHT_DELIM_KEY (PATH_H_PBUFFER(p_s_path, n_h)), B_N_PDELIM_KEY (p_s_tb->CFR[n_h], p_s_tb->rkey[n_h])); + } +#endif + if ((p_s_curf && !B_IS_IN_TREE (p_s_curf)) || (p_s_curcf && !B_IS_IN_TREE (p_s_curcf))) { + reiserfs_panic (p_s_tb->tb_sb, "PAP-8205: get_parents: FR (%b) or CFR (%b) is invalid", p_s_curf, p_s_curcf); + } +#endif + + return CARRY_ON; +} + + +/* it is possible to remove node as result of shiftings to + neighbors even when we insert or paste item. */ +static inline int can_node_be_removed (int mode, int lfree, int sfree, int rfree, struct tree_balance * tb, int h) +{ + struct buffer_head * Sh = PATH_H_PBUFFER (tb->tb_path, h); + int levbytes = tb->insert_size[h]; + struct item_head * ih; + struct key * r_key = NULL; + + ih = B_N_PITEM_HEAD (Sh, 0); + if ( tb->CFR[h] ) + r_key = B_N_PDELIM_KEY(tb->CFR[h],tb->rkey[h]); + + if ( + lfree + rfree + sfree < MAX_CHILD_SIZE(Sh) + levbytes + /* shifting may merge items which might save space */ +#ifdef REISERFS_FSCK + - (( ! h && is_left_mergeable (tb->tb_sb, tb->tb_path) == 1 ) ? IH_SIZE : 0) + - (( ! h && r_ih && is_right_mergeable (tb->tb_sb, tb->tb_path) == 1 ) ? IH_SIZE : 0) +#else + - (( ! h && op_is_left_mergeable (&(ih->ih_key), Sh->b_size) ) ? IH_SIZE : 0) + - (( ! h && r_key && op_is_left_mergeable (r_key, Sh->b_size) ) ? IH_SIZE : 0) +#endif + + (( h ) ? KEY_SIZE : 0)) + { + /* node can not be removed */ + if (sfree >= levbytes ) { /* new item fits into node S[h] without any shifting */ + if ( ! h ) + tb->s0num = B_NR_ITEMS(Sh) + ((mode == M_INSERT ) ? 1 : 0); + set_parameters (tb, h, 0, 0, 1, NULL, -1, -1); + return NO_BALANCING_NEEDED; + } + } + return !NO_BALANCING_NEEDED; +} + + + +/* Check whether current node S[h] is balanced when increasing its size by + * Inserting or Pasting. + * Calculate parameters for balancing for current level h. + * Parameters: + * tb tree_balance structure; + * h current level of the node; + * inum item number in S[h]; + * mode i - insert, p - paste; + * Returns: 1 - schedule occured; + * 0 - balancing for higher levels needed; + * -1 - no balancing for higher levels needed; + * -2 - no disk space. + */ +/* ip means Inserting or Pasting */ +static int ip_check_balance (struct tree_balance * tb, int h) +{ + struct virtual_node * vn = tb->tb_vn; + int levbytes, /* Number of bytes that must be inserted into (value + is negative if bytes are deleted) buffer which + contains node being balanced. The mnemonic is + that the attempted change in node space used level + is levbytes bytes. */ + n_ret_value; + + int lfree, sfree, rfree /* free space in L, S and R */; + + /* nver is short for number of vertixes, and lnver is the number if + we shift to the left, rnver is the number if we shift to the + right, and lrnver is the number if we shift in both directions. + The goal is to minimize first the number of vertixes, and second, + the number of vertixes whose contents are changed by shifting, + and third the number of uncached vertixes whose contents are + changed by shifting and must be read from disk. */ + int nver, lnver, rnver, lrnver; + + /* used at leaf level only, S0 = S[0] is the node being balanced, + sInum [ I = 0,1,2 ] is the number of items that will + remain in node SI after balancing. S1 and S2 are new + nodes that might be created. */ + + /* we perform 8 calls to get_num_ver(). For each call we calculate five parameters. + where 4th parameter is s1bytes and 5th - s2bytes + */ + short snum012[40] = {0,}; /* s0num, s1num, s2num for 8 cases + 0,1 - do not shift and do not shift but bottle + 2 - shift only whole item to left + 3 - shift to left and bottle as much as possible + 4,5 - shift to right (whole items and as much as possible + 6,7 - shift to both directions (whole items and as much as possible) + */ + + /* Sh is the node whose balance is currently being checked */ + struct buffer_head * Sh; + +#ifdef REISERFS_FSCK + /* special mode for insert pointer to the most low internal node */ + if (h == 0 && vn->vn_mode == M_INTERNAL) { + /* blk_num == 2 is to get pointer inserted to the next level */ + set_parameters (tb, h, 0, 0, 2, NULL, -1, -1); + return 0; + } +#endif + + Sh = PATH_H_PBUFFER (tb->tb_path, h); + levbytes = tb->insert_size[h]; + + /* Calculate balance parameters for creating new root. */ + if ( ! Sh ) { + if ( ! h ) + reiserfs_panic (tb->tb_sb, "vs-8210: ip_check_balance: S[0] can not be 0"); + switch ( n_ret_value = get_empty_nodes (tb, h) ) { + case CARRY_ON: + set_parameters (tb, h, 0, 0, 1, NULL, -1, -1); + return NO_BALANCING_NEEDED; /* no balancing for higher levels needed */ + + case NO_DISK_SPACE: + case REPEAT_SEARCH: + return n_ret_value; + default: + reiserfs_panic(tb->tb_sb, "vs-8215: ip_check_balance: incorrect return value of get_empty_nodes"); + } + } + + if ( (n_ret_value = get_parents (tb, h)) != CARRY_ON ) /* get parents of S[h] neighbors. */ + return n_ret_value; + + sfree = B_FREE_SPACE (Sh); + + /* get free space of neighbors */ + rfree = get_rfree (tb, h); + lfree = get_lfree (tb, h); + + if (can_node_be_removed (vn->vn_mode, lfree, sfree, rfree, tb, h) == NO_BALANCING_NEEDED) + /* and new item fits into node S[h] without any shifting */ + return NO_BALANCING_NEEDED; + + create_virtual_node (tb, h); + + /* + determine maximal number of items we can shift to the left neighbor (in tb structure) + and the maximal number of bytes that can flow to the left neighbor + from the left most liquid item that cannot be shifted from S[0] entirely (returned value) + */ + check_left (tb, h, lfree); + + /* + determine maximal number of items we can shift to the right neighbor (in tb structure) + and the maximal number of bytes that can flow to the right neighbor + from the right most liquid item that cannot be shifted from S[0] entirely (returned value) + */ + check_right (tb, h, rfree); + + + /* all contents of internal node S[h] can be moved into its + neighbors, S[h] will be removed after balancing */ + if (h && (tb->rnum[h] + tb->lnum[h] >= vn->vn_nr_item + 1)) { + int to_r; + + /* Since we are working on internal nodes, and our internal + nodes have fixed size entries, then we can balance by the + number of items rather than the space they consume. In this + routine we set the left node equal to the right node, + allowing a difference of less than or equal to 1 child + pointer. */ + to_r = ((MAX_NR_KEY(Sh)<<1)+2-tb->lnum[h]-tb->rnum[h]+vn->vn_nr_item+1)/2 - + (MAX_NR_KEY(Sh) + 1 - tb->rnum[h]); + set_parameters (tb, h, vn->vn_nr_item + 1 - to_r, to_r, 0, NULL, -1, -1); + return CARRY_ON; + } + +#ifdef CONFIG_REISERFS_CHECK + /* this checks balance condition, that any two neighboring nodes can not fit in one node */ + if ( h && ( tb->lnum[h] >= vn->vn_nr_item + 1 || tb->rnum[h] >= vn->vn_nr_item + 1) ) + reiserfs_panic (tb->tb_sb, "vs-8220: ip_check_balance: tree is not balanced on internal level"); + + if ( ! h && ((tb->lnum[h] >= vn->vn_nr_item && (tb->lbytes == -1)) || + (tb->rnum[h] >= vn->vn_nr_item && (tb->rbytes == -1)) )) + reiserfs_panic(tb->tb_sb, "vs-8225: ip_check_balance: tree is not balanced on leaf level"); +#endif + + /* all contents of S[0] can be moved into its neighbors + S[0] will be removed after balancing. */ + if (!h && is_leaf_removable (tb)) + return CARRY_ON; + + + /* why do we perform this check here rather than earlier?? + Answer: we can win 1 node in some cases above. Moreover we + checked it above, when we checked, that S[0] is not removable + in principle */ + if (sfree >= levbytes) { /* new item fits into node S[h] without any shifting */ + if ( ! h ) + tb->s0num = vn->vn_nr_item; + set_parameters (tb, h, 0, 0, 1, NULL, -1, -1); + return NO_BALANCING_NEEDED; + } + + + { + int lpar, rpar, nset, lset, rset, lrset; + /* + * regular overflowing of the node + */ + + /* get_num_ver works in 2 modes (FLOW & NO_FLOW) + lpar, rpar - number of items we can shift to left/right neighbor (including splitting item) + nset, lset, rset, lrset - shows, whether flowing items give better packing + */ +#define FLOW 1 +#define NO_FLOW 0 /* do not any splitting */ + + /* we choose one the following */ +#define NOTHING_SHIFT_NO_FLOW 0 +#define NOTHING_SHIFT_FLOW 5 +#define LEFT_SHIFT_NO_FLOW 10 +#define LEFT_SHIFT_FLOW 15 +#define RIGHT_SHIFT_NO_FLOW 20 +#define RIGHT_SHIFT_FLOW 25 +#define LR_SHIFT_NO_FLOW 30 +#define LR_SHIFT_FLOW 35 + + + lpar = tb->lnum[h]; + rpar = tb->rnum[h]; + + + /* calculate number of blocks S[h] must be split into when + nothing is shifted to the neighbors, + as well as number of items in each part of the split node (s012 numbers), + and number of bytes (s1bytes) of the shared drop which flow to S1 if any */ + nset = NOTHING_SHIFT_NO_FLOW; + nver = get_num_ver (vn->vn_mode, tb, h, + 0, -1, h?vn->vn_nr_item:0, -1, + snum012, NO_FLOW); + + if (!h) + { + int nver1; + + /* note, that in this case we try to bottle between S[0] and S1 (S1 - the first new node) */ + nver1 = get_num_ver (vn->vn_mode, tb, h, + 0, -1, 0, -1, + snum012 + NOTHING_SHIFT_FLOW, FLOW); + if (nver > nver1) + nset = NOTHING_SHIFT_FLOW, nver = nver1; + } + + + /* calculate number of blocks S[h] must be split into when + l_shift_num first items and l_shift_bytes of the right most + liquid item to be shifted are shifted to the left neighbor, + as well as number of items in each part of the splitted node (s012 numbers), + and number of bytes (s1bytes) of the shared drop which flow to S1 if any + */ + lset = LEFT_SHIFT_NO_FLOW; + lnver = get_num_ver (vn->vn_mode, tb, h, + lpar - (( h || tb->lbytes == -1 ) ? 0 : 1), -1, h ? vn->vn_nr_item:0, -1, + snum012 + LEFT_SHIFT_NO_FLOW, NO_FLOW); + if (!h) + { + int lnver1; + + lnver1 = get_num_ver (vn->vn_mode, tb, h, + lpar - ((tb->lbytes != -1) ? 1 : 0), tb->lbytes, 0, -1, + snum012 + LEFT_SHIFT_FLOW, FLOW); + if (lnver > lnver1) + lset = LEFT_SHIFT_FLOW, lnver = lnver1; + } + + + /* calculate number of blocks S[h] must be split into when + r_shift_num first items and r_shift_bytes of the left most + liquid item to be shifted are shifted to the right neighbor, + as well as number of items in each part of the splitted node (s012 numbers), + and number of bytes (s1bytes) of the shared drop which flow to S1 if any + */ + rset = RIGHT_SHIFT_NO_FLOW; + rnver = get_num_ver (vn->vn_mode, tb, h, + 0, -1, h ? (vn->vn_nr_item-rpar) : (rpar - (( tb->rbytes != -1 ) ? 1 : 0)), -1, + snum012 + RIGHT_SHIFT_NO_FLOW, NO_FLOW); + if (!h) + { + int rnver1; + + rnver1 = get_num_ver (vn->vn_mode, tb, h, + 0, -1, (rpar - ((tb->rbytes != -1) ? 1 : 0)), tb->rbytes, + snum012 + RIGHT_SHIFT_FLOW, FLOW); + + if (rnver > rnver1) + rset = RIGHT_SHIFT_FLOW, rnver = rnver1; + } + + + /* calculate number of blocks S[h] must be split into when + items are shifted in both directions, + as well as number of items in each part of the splitted node (s012 numbers), + and number of bytes (s1bytes) of the shared drop which flow to S1 if any + */ + lrset = LR_SHIFT_NO_FLOW; + lrnver = get_num_ver (vn->vn_mode, tb, h, + lpar - ((h || tb->lbytes == -1) ? 0 : 1), -1, h ? (vn->vn_nr_item-rpar):(rpar - ((tb->rbytes != -1) ? 1 : 0)), -1, + snum012 + LR_SHIFT_NO_FLOW, NO_FLOW); + if (!h) + { + int lrnver1; + + lrnver1 = get_num_ver (vn->vn_mode, tb, h, + lpar - ((tb->lbytes != -1) ? 1 : 0), tb->lbytes, (rpar - ((tb->rbytes != -1) ? 1 : 0)), tb->rbytes, + snum012 + LR_SHIFT_FLOW, FLOW); + if (lrnver > lrnver1) + lrset = LR_SHIFT_FLOW, lrnver = lrnver1; + } + + + + /* Our general shifting strategy is: + 1) to minimized number of new nodes; + 2) to minimized number of neighbors involved in shifting; + 3) to minimized number of disk reads; */ + + /* we can win TWO or ONE nodes by shifting in both directions */ + if (lrnver < lnver && lrnver < rnver) + { +#ifdef CONFIG_REISERFS_CHECK + if (h && (tb->lnum[h] != 1 || tb->rnum[h] != 1 || lrnver != 1 || rnver != 2 || lnver != 2 || h != 1)) + reiserfs_panic (0, "vs-8230: check_balance: bad h"); +#endif + if (lrset == LR_SHIFT_FLOW) + set_parameters (tb, h, tb->lnum[h], tb->rnum[h], lrnver, snum012 + lrset, + tb->lbytes, tb->rbytes); + else + set_parameters (tb, h, tb->lnum[h] - ((tb->lbytes == -1) ? 0 : 1), + tb->rnum[h] - ((tb->rbytes == -1) ? 0 : 1), lrnver, snum012 + lrset, -1, -1); + + return CARRY_ON; + } + + /* if shifting doesn't lead to better packing then don't shift */ + if (nver == lrnver) + { + set_parameters (tb, h, 0, 0, nver, snum012 + nset, -1, -1); + return CARRY_ON; + } + + + /* now we know that for better packing shifting in only one + direction either to the left or to the right is required */ + + /* if shifting to the left is better than shifting to the right */ + if (lnver < rnver) + { + SET_PAR_SHIFT_LEFT; + return CARRY_ON; + } + + /* if shifting to the right is better than shifting to the left */ + if (lnver > rnver) + { + SET_PAR_SHIFT_RIGHT; + return CARRY_ON; + } + + + /* now shifting in either direction gives the same number + of nodes and we can make use of the cached neighbors */ + if (is_left_neighbor_in_cache (tb,h)) + { + SET_PAR_SHIFT_LEFT; + return CARRY_ON; + } + + /* shift to the right independently on whether the right neighbor in cache or not */ + SET_PAR_SHIFT_RIGHT; + return CARRY_ON; + } +} + + +/* Check whether current node S[h] is balanced when Decreasing its size by + * Deleting or Cutting for INTERNAL node of S+tree. + * Calculate parameters for balancing for current level h. + * Parameters: + * tb tree_balance structure; + * h current level of the node; + * inum item number in S[h]; + * mode i - insert, p - paste; + * Returns: 1 - schedule occured; + * 0 - balancing for higher levels needed; + * -1 - no balancing for higher levels needed; + * -2 - no disk space. + * + * Note: Items of internal nodes have fixed size, so the balance condition for + * the internal part of S+tree is as for the B-trees. + */ +static int dc_check_balance_internal (struct tree_balance * tb, int h) +{ + struct virtual_node * vn = tb->tb_vn; + + /* Sh is the node whose balance is currently being checked, + and Fh is its father. */ + struct buffer_head * Sh, * Fh; + int maxsize, + n_ret_value; + int lfree, rfree /* free space in L and R */; + + Sh = PATH_H_PBUFFER (tb->tb_path, h); + Fh = PATH_H_PPARENT (tb->tb_path, h); + + maxsize = MAX_CHILD_SIZE(Sh); + +/* using tb->insert_size[h], which is negative in this case, create_virtual_node calculates: */ +/* new_nr_item = number of items node would have if operation is */ +/* performed without balancing (new_nr_item); */ + create_virtual_node (tb, h); + + if ( ! Fh ) + { /* S[h] is the root. */ + if ( vn->vn_nr_item > 0 ) + { + set_parameters (tb, h, 0, 0, 1, NULL, -1, -1); + return NO_BALANCING_NEEDED; /* no balancing for higher levels needed */ + } + /* new_nr_item == 0. + * Current root will be deleted resulting in + * decrementing the tree height. */ + set_parameters (tb, h, 0, 0, 0, NULL, -1, -1); + return CARRY_ON; + } + + if ( (n_ret_value = get_parents(tb,h)) != CARRY_ON ) + return n_ret_value; + + + /* get free space of neighbors */ + rfree = get_rfree (tb, h); + lfree = get_lfree (tb, h); + + /* determine maximal number of items we can fit into neighbors */ + check_left (tb, h, lfree); + check_right (tb, h, rfree); + + + if ( vn->vn_nr_item >= MIN_NR_KEY(Sh) ) + { /* Balance condition for the internal node is valid. + * In this case we balance only if it leads to better packing. */ + if ( vn->vn_nr_item == MIN_NR_KEY(Sh) ) + { /* Here we join S[h] with one of its neighbors, + * which is impossible with greater values of new_nr_item. */ + if ( tb->lnum[h] >= vn->vn_nr_item + 1 ) + { + /* All contents of S[h] can be moved to L[h]. */ + int n; + int order_L; + + order_L = ((n=PATH_H_B_ITEM_ORDER(tb->tb_path, h))==0) ? B_NR_ITEMS(tb->FL[h]) : n - 1; + n = B_N_CHILD(tb->FL[h],order_L)->dc_size / (DC_SIZE + KEY_SIZE); + set_parameters (tb, h, -n-1, 0, 0, NULL, -1, -1); + return CARRY_ON; + } + + if ( tb->rnum[h] >= vn->vn_nr_item + 1 ) + { + /* All contents of S[h] can be moved to R[h]. */ + int n; + int order_R; + + order_R = ((n=PATH_H_B_ITEM_ORDER(tb->tb_path, h))==B_NR_ITEMS(Fh)) ? 0 : n + 1; + n = B_N_CHILD(tb->FR[h],order_R)->dc_size / (DC_SIZE + KEY_SIZE); + set_parameters (tb, h, 0, -n-1, 0, NULL, -1, -1); + return CARRY_ON; + } + } + + if (tb->rnum[h] + tb->lnum[h] >= vn->vn_nr_item + 1) + { + /* All contents of S[h] can be moved to the neighbors (L[h] & R[h]). */ + int to_r; + + to_r = ((MAX_NR_KEY(Sh)<<1)+2-tb->lnum[h]-tb->rnum[h]+vn->vn_nr_item+1)/2 - + (MAX_NR_KEY(Sh) + 1 - tb->rnum[h]); + set_parameters (tb, h, vn->vn_nr_item + 1 - to_r, to_r, 0, NULL, -1, -1); + return CARRY_ON; + } + + /* Balancing does not lead to better packing. */ + set_parameters (tb, h, 0, 0, 1, NULL, -1, -1); + return NO_BALANCING_NEEDED; + } + + /* Current node contain insufficient number of items. Balancing is required. */ + /* Check whether we can merge S[h] with left neighbor. */ + if (tb->lnum[h] >= vn->vn_nr_item + 1) + if (is_left_neighbor_in_cache (tb,h) || tb->rnum[h] < vn->vn_nr_item + 1 || !tb->FR[h]) + { + int n; + int order_L; + + order_L = ((n=PATH_H_B_ITEM_ORDER(tb->tb_path, h))==0) ? B_NR_ITEMS(tb->FL[h]) : n - 1; + n = B_N_CHILD(tb->FL[h],order_L)->dc_size / (DC_SIZE + KEY_SIZE); + set_parameters (tb, h, -n-1, 0, 0, NULL, -1, -1); + return CARRY_ON; + } + + /* Check whether we can merge S[h] with right neighbor. */ + if (tb->rnum[h] >= vn->vn_nr_item + 1) + { + int n; + int order_R; + + order_R = ((n=PATH_H_B_ITEM_ORDER(tb->tb_path, h))==B_NR_ITEMS(Fh)) ? 0 : (n + 1); + n = B_N_CHILD(tb->FR[h],order_R)->dc_size / (DC_SIZE + KEY_SIZE); + set_parameters (tb, h, 0, -n-1, 0, NULL, -1, -1); + return CARRY_ON; + } + + /* All contents of S[h] can be moved to the neighbors (L[h] & R[h]). */ + if (tb->rnum[h] + tb->lnum[h] >= vn->vn_nr_item + 1) + { + int to_r; + + to_r = ((MAX_NR_KEY(Sh)<<1)+2-tb->lnum[h]-tb->rnum[h]+vn->vn_nr_item+1)/2 - + (MAX_NR_KEY(Sh) + 1 - tb->rnum[h]); + set_parameters (tb, h, vn->vn_nr_item + 1 - to_r, to_r, 0, NULL, -1, -1); + return CARRY_ON; + } + + /* For internal nodes try to borrow item from a neighbor */ +#ifdef CONFIG_REISERFS_CHECK + if (!tb->FL[h] && !tb->FR[h]) + reiserfs_panic (0, "vs-8235: dc_check_balance_internal: trying to borrow for root"); +#endif + + /* Borrow one or two items from caching neighbor */ + if (is_left_neighbor_in_cache (tb,h) || !tb->FR[h]) + { + int from_l; + + from_l = (MAX_NR_KEY(Sh) + 1 - tb->lnum[h] + vn->vn_nr_item + 1) / 2 - (vn->vn_nr_item + 1); + set_parameters (tb, h, -from_l, 0, 1, NULL, -1, -1); + return CARRY_ON; + } + + set_parameters (tb, h, 0, -((MAX_NR_KEY(Sh)+1-tb->rnum[h]+vn->vn_nr_item+1)/2-(vn->vn_nr_item+1)), 1, + NULL, -1, -1); + return CARRY_ON; +} + + +/* Check whether current node S[h] is balanced when Decreasing its size by + * Deleting or Truncating for LEAF node of S+tree. + * Calculate parameters for balancing for current level h. + * Parameters: + * tb tree_balance structure; + * h current level of the node; + * inum item number in S[h]; + * mode i - insert, p - paste; + * Returns: 1 - schedule occured; + * 0 - balancing for higher levels needed; + * -1 - no balancing for higher levels needed; + * -2 - no disk space. + */ +static int dc_check_balance_leaf (struct tree_balance * tb, int h) +{ + struct virtual_node * vn = tb->tb_vn; + + /* Number of bytes that must be deleted from + (value is negative if bytes are deleted) buffer which + contains node being balanced. The mnemonic is that the + attempted change in node space used level is levbytes bytes. */ + int levbytes; + /* the maximal item size */ + int maxsize, + n_ret_value; + /* S0 is the node whose balance is currently being checked, + and F0 is its father. */ + struct buffer_head * S0, * F0; + int lfree, rfree /* free space in L and R */; + + S0 = PATH_H_PBUFFER (tb->tb_path, 0); + F0 = PATH_H_PPARENT (tb->tb_path, 0); + + levbytes = tb->insert_size[h]; + + maxsize = MAX_CHILD_SIZE(S0); /* maximal possible size of an item */ + + if ( ! F0 ) + { /* S[0] is the root now. */ + +#ifdef CONFIG_REISERFS_CHECK + if ( -levbytes >= maxsize - B_FREE_SPACE (S0) ) + reiserfs_panic (tb->tb_sb, "vs-8240: dc_check_balance_leaf: attempt to create empty buffer tree"); +#endif + + set_parameters (tb, h, 0, 0, 1, NULL, -1, -1); + return NO_BALANCING_NEEDED; + } + + if ( (n_ret_value = get_parents(tb,h)) != CARRY_ON ) + return n_ret_value; + + /* get free space of neighbors */ + rfree = get_rfree (tb, h); + lfree = get_lfree (tb, h); + + create_virtual_node (tb, h); + + /* if 3 leaves can be merge to one, set parameters and return */ + if (are_leaves_removable (tb, lfree, rfree)) + return CARRY_ON; + + /* determine maximal number of items we can shift to the left/right neighbor + and the maximal number of bytes that can flow to the left/right neighbor + from the left/right most liquid item that cannot be shifted from S[0] entirely + */ + check_left (tb, h, lfree); + check_right (tb, h, rfree); + + /* check whether we can merge S with left neighbor. */ + if (tb->lnum[0] >= vn->vn_nr_item && tb->lbytes == -1) + if (is_left_neighbor_in_cache (tb,h) || + ((tb->rnum[0] - ((tb->rbytes == -1) ? 0 : 1)) < vn->vn_nr_item) || /* S can not be merged with R */ + !tb->FR[h]) { + +#ifdef CONFIG_REISERFS_CHECK + if (!tb->FL[h]) + reiserfs_panic (0, "vs-8245: dc_check_balance_leaf: FL[h] must exist"); +#endif + + /* set parameter to merge S[0] with its left neighbor */ + set_parameters (tb, h, -1, 0, 0, NULL, -1, -1); + return CARRY_ON; + } + + /* check whether we can merge S[0] with right neighbor. */ + if (tb->rnum[0] >= vn->vn_nr_item && tb->rbytes == -1) { + set_parameters (tb, h, 0, -1, 0, NULL, -1, -1); + return CARRY_ON; + } + + /* All contents of S[0] can be moved to the neighbors (L[0] & R[0]). Set parameters and return */ + if (is_leaf_removable (tb)) + return CARRY_ON; + + /* Balancing is not required. */ + tb->s0num = vn->vn_nr_item; + set_parameters (tb, h, 0, 0, 1, NULL, -1, -1); + return NO_BALANCING_NEEDED; +} + + + +/* Check whether current node S[h] is balanced when Decreasing its size by + * Deleting or Cutting. + * Calculate parameters for balancing for current level h. + * Parameters: + * tb tree_balance structure; + * h current level of the node; + * inum item number in S[h]; + * mode d - delete, c - cut. + * Returns: 1 - schedule occured; + * 0 - balancing for higher levels needed; + * -1 - no balancing for higher levels needed; + * -2 - no disk space. + */ +static int dc_check_balance (struct tree_balance * tb, int h) +{ + +#ifdef CONFIG_REISERFS_CHECK + if ( ! (PATH_H_PBUFFER (tb->tb_path, h)) ) + reiserfs_panic(tb->tb_sb, "vs-8250: dc_check_balance: S is not initialized"); +#endif + + if ( h ) + return dc_check_balance_internal (tb, h); + else + return dc_check_balance_leaf (tb, h); +} + + + +/* Check whether current node S[h] is balanced. + * Calculate parameters for balancing for current level h. + * Parameters: + * + * tb tree_balance structure: + * + * tb is a large structure that must be read about in the header file + * at the same time as this procedure if the reader is to successfully + * understand this procedure + * + * h current level of the node; + * inum item number in S[h]; + * mode i - insert, p - paste, d - delete, c - cut. + * Returns: 1 - schedule occured; + * 0 - balancing for higher levels needed; + * -1 - no balancing for higher levels needed; + * -2 - no disk space. + */ +static int check_balance (int mode, + struct tree_balance * tb, + int h, + int inum, + int pos_in_item, + struct item_head * ins_ih, + const void * data + ) +{ + struct virtual_node * vn; + + vn = tb->tb_vn = (struct virtual_node *)(tb->vn_buf); + vn->vn_free_ptr = (char *)(tb->tb_vn + 1); + vn->vn_mode = mode; + vn->vn_affected_item_num = inum; + vn->vn_pos_in_item = pos_in_item; + vn->vn_ins_ih = ins_ih; + vn->vn_data = data; + +#ifdef CONFIG_REISERFS_CHECK + if (mode == M_INSERT && !vn->vn_ins_ih) + reiserfs_panic (0, "vs-8255: check_balance: ins_ih can not be 0 in insert mode"); +#endif + + if ( tb->insert_size[h] > 0 ) + /* Calculate balance parameters when size of node is increasing. */ + return ip_check_balance (tb, h); + + /* Calculate balance parameters when size of node is decreasing. */ + return dc_check_balance (tb, h); +} + + + +/* Check whether parent at the path is the really parent of the current node.*/ +static int get_direct_parent( + struct tree_balance * p_s_tb, + int n_h + ) { + struct buffer_head * p_s_bh; + struct path * p_s_path = p_s_tb->tb_path; + int n_position, + n_path_offset = PATH_H_PATH_OFFSET(p_s_tb->tb_path, n_h); + + /* We are in the root or in the new root. */ + if ( n_path_offset <= FIRST_PATH_ELEMENT_OFFSET ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( n_path_offset < FIRST_PATH_ELEMENT_OFFSET - 1 ) + reiserfs_panic(p_s_tb->tb_sb, "PAP-8260: get_direct_parent: illegal offset in the path"); +#endif + + if ( PATH_OFFSET_PBUFFER(p_s_path, FIRST_PATH_ELEMENT_OFFSET)->b_blocknr == + SB_ROOT_BLOCK (p_s_tb->tb_sb) ) { + /* Root is not changed. */ + PATH_OFFSET_PBUFFER(p_s_path, n_path_offset - 1) = NULL; + PATH_OFFSET_POSITION(p_s_path, n_path_offset - 1) = 0; + return CARRY_ON; + } + return REPEAT_SEARCH; /* Root is changed and we must recalculate the path. */ + } + + if ( ! B_IS_IN_TREE(p_s_bh = PATH_OFFSET_PBUFFER(p_s_path, n_path_offset - 1)) ) + return REPEAT_SEARCH; /* Parent in the path is not in the tree. */ + + if ( (n_position = PATH_OFFSET_POSITION(p_s_path, n_path_offset - 1)) > B_NR_ITEMS(p_s_bh) ) + return REPEAT_SEARCH; + + if ( B_N_CHILD_NUM(p_s_bh, n_position) != PATH_OFFSET_PBUFFER(p_s_path, n_path_offset)->b_blocknr ) + /* Parent in the path is not parent of the current node in the tree. */ + return REPEAT_SEARCH; + + if ( buffer_locked(p_s_bh) ) { + __wait_on_buffer(p_s_bh); + if ( FILESYSTEM_CHANGED_TB (p_s_tb) ) + return REPEAT_SEARCH; + } + + return CARRY_ON; /* Parent in the path is unlocked and really parent of the current node. */ +} + + +/* Using lnum[n_h] and rnum[n_h] we should determine what neighbors + * of S[n_h] we + * need in order to balance S[n_h], and get them if necessary. + * Returns: SCHEDULE_OCCURRED - schedule occured while the function worked; + * CARRY_ON - schedule didn't occur while the function worked; + */ +static int get_neighbors( + struct tree_balance * p_s_tb, + int n_h + ) { + int n_child_position, + n_path_offset = PATH_H_PATH_OFFSET(p_s_tb->tb_path, n_h + 1); + unsigned long n_son_number; + struct super_block * p_s_sb = p_s_tb->tb_sb; + struct buffer_head * p_s_bh; + + + if ( p_s_tb->lnum[n_h] ) { + /* We need left neighbor to balance S[n_h]. */ + p_s_bh = PATH_OFFSET_PBUFFER(p_s_tb->tb_path, n_path_offset); + +#ifdef CONFIG_REISERFS_CHECK + if ( p_s_bh == p_s_tb->FL[n_h] && ! PATH_OFFSET_POSITION(p_s_tb->tb_path, n_path_offset) ) + reiserfs_panic (p_s_tb->tb_sb, "PAP-8270: get_neighbors: invalid position in the parent"); +#endif + + n_child_position = ( p_s_bh == p_s_tb->FL[n_h] ) ? p_s_tb->lkey[n_h] : B_NR_ITEMS (p_s_tb->FL[n_h]); + n_son_number = B_N_CHILD_NUM(p_s_tb->FL[n_h], n_child_position); + p_s_bh = reiserfs_bread(p_s_sb->s_dev, n_son_number, p_s_sb->s_blocksize); + if (!p_s_bh) + return IO_ERROR; + if ( FILESYSTEM_CHANGED_TB (p_s_tb) ) { + decrement_bcount(p_s_bh); + return REPEAT_SEARCH; + } + +#ifdef CONFIG_REISERFS_CHECK + if ( ! B_IS_IN_TREE(p_s_tb->FL[n_h]) || n_child_position > B_NR_ITEMS(p_s_tb->FL[n_h]) || + B_N_CHILD_NUM(p_s_tb->FL[n_h], n_child_position) != p_s_bh->b_blocknr ) + reiserfs_panic (p_s_tb->tb_sb, "PAP-8275: get_neighbors: invalid parent"); + if ( ! B_IS_IN_TREE(p_s_bh) ) + reiserfs_panic (p_s_tb->tb_sb, "PAP-8280: get_neighbors: invalid child"); + + if (! n_h && B_FREE_SPACE (p_s_bh) != MAX_CHILD_SIZE (p_s_bh) - B_N_CHILD (p_s_tb->FL[0],n_child_position)->dc_size) + reiserfs_panic (p_s_tb->tb_sb, "PAP-8290: get_neighbors: invalid child size of left neighbor"); +#endif + + decrement_bcount(p_s_tb->L[n_h]); + p_s_tb->L[n_h] = p_s_bh; + } + + + if ( p_s_tb->rnum[n_h] ) { /* We need right neighbor to balance S[n_path_offset]. */ + p_s_bh = PATH_OFFSET_PBUFFER(p_s_tb->tb_path, n_path_offset); + +#ifdef CONFIG_REISERFS_CHECK + if ( p_s_bh == p_s_tb->FR[n_h] && PATH_OFFSET_POSITION(p_s_tb->tb_path, n_path_offset) >= B_NR_ITEMS(p_s_bh) ) + reiserfs_panic (p_s_tb->tb_sb, "PAP-8295: get_neighbors: invalid position in the parent"); +#endif + + n_child_position = ( p_s_bh == p_s_tb->FR[n_h] ) ? p_s_tb->rkey[n_h] + 1 : 0; + n_son_number = B_N_CHILD_NUM(p_s_tb->FR[n_h], n_child_position); + p_s_bh = reiserfs_bread(p_s_sb->s_dev, n_son_number, p_s_sb->s_blocksize); + if (!p_s_bh) + return IO_ERROR; + if ( FILESYSTEM_CHANGED_TB (p_s_tb) ) { + decrement_bcount(p_s_bh); + return REPEAT_SEARCH; + } + decrement_bcount(p_s_tb->R[n_h]); + p_s_tb->R[n_h] = p_s_bh; + +#ifdef CONFIG_REISERFS_CHECK + if (! n_h && B_FREE_SPACE (p_s_bh) != MAX_CHILD_SIZE (p_s_bh) - B_N_CHILD (p_s_tb->FR[0],n_child_position)->dc_size) { + reiserfs_panic (p_s_tb->tb_sb, "PAP-8300: get_neighbors: invalid child size of right neighbor (%d != %d - %d)", + B_FREE_SPACE (p_s_bh), MAX_CHILD_SIZE (p_s_bh), B_N_CHILD (p_s_tb->FR[0],n_child_position)->dc_size); + } +#endif + + } + return CARRY_ON; +} + + +void * reiserfs_kmalloc (size_t size, int flags, struct super_block * s) +{ + void * vp; + static size_t malloced; + + + vp = kmalloc (size, flags); + if (vp) { + s->u.reiserfs_sb.s_kmallocs += size; + if (s->u.reiserfs_sb.s_kmallocs > malloced + 200000) { + reiserfs_warning ("vs-8301: reiserfs_kmalloc: allocated memory %d\n", s->u.reiserfs_sb.s_kmallocs); + malloced = s->u.reiserfs_sb.s_kmallocs; + } + } +/*printk ("malloc : size %d, allocated %d\n", size, s->u.reiserfs_sb.s_kmallocs);*/ + return vp; +} + +void reiserfs_kfree (const void * vp, size_t size, struct super_block * s) +{ + kfree (vp); + + s->u.reiserfs_sb.s_kmallocs -= size; + if (s->u.reiserfs_sb.s_kmallocs < 0) + reiserfs_warning ("vs-8302: reiserfs_kfree: allocated memory %d\n", s->u.reiserfs_sb.s_kmallocs); + +} + + +static int get_virtual_node_size (struct super_block * sb, struct buffer_head * bh) +{ + // int size = sizeof (struct virtual_item); /* for new item in case of insert */ + // int i, nr_items; + // struct item_head * ih; + + // this is enough for _ALL_ currently possible cases. In 4 k block + // one may put < 170 empty items. Each virtual item eats 12 + // byte. The biggest direntry item may have < 256 entries. Each + // entry would eat 2 byte of virtual node space + return sb->s_blocksize; + +#if 0 + size = sizeof (struct virtual_node) + sizeof (struct virtual_item); + ih = B_N_PITEM_HEAD (bh, 0); + nr_items = B_NR_ITEMS (bh); + for (i = 0; i < nr_items; i ++, ih ++) { + /* each item occupies some space in virtual node */ + size += sizeof (struct virtual_item); + if (is_direntry_le_ih (ih)) + /* each entry and new one occupeis 2 byte in the virtual node */ + size += (le16_to_cpu (ih->u.ih_entry_count) + 1) * sizeof (__u16); + } + + /* 1 bit for each bitmap block to note whether bitmap block was + dirtied in the operation */ + /* size += (SB_BMAP_NR (sb) * 2 / 8 + 4);*/ + return size; +#endif +} + + + +/* maybe we should fail balancing we are going to perform when kmalloc + fails several times. But now it will loop until kmalloc gets + required memory */ +static int get_mem_for_virtual_node (struct tree_balance * tb) +{ + int check_fs = 0; + int size; + char * buf; + + size = get_virtual_node_size (tb->tb_sb, PATH_PLAST_BUFFER (tb->tb_path)); + + if (size > tb->vn_buf_size) { + /* we have to allocate more memory for virtual node */ + if (tb->vn_buf) { + /* free memory allocated before */ + reiserfs_kfree (tb->vn_buf, tb->vn_buf_size, tb->tb_sb); + /* this is not needed if kfree is atomic */ + check_fs = 1; + } + + /* virtual node requires now more memory */ + tb->vn_buf_size = size; + + /* get memory for virtual item */ + buf = reiserfs_kmalloc(size, GFP_ATOMIC, tb->tb_sb); + if ( ! buf ) { + /* getting memory with GFP_KERNEL priority may involve + balancing now (due to indirect_to_direct conversion on + dcache shrinking). So, release path and collected + resourses here */ + free_buffers_in_tb (tb); + buf = reiserfs_kmalloc(size, GFP_BUFFER, tb->tb_sb); + if ( !buf ) { +#ifdef CONFIG_REISERFS_CHECK + reiserfs_warning ("vs-8345: get_mem_for_virtual_node: " + "kmalloc failed. reiserfs kmalloced %d bytes\n", + tb->tb_sb->u.reiserfs_sb.s_kmallocs); +#endif + tb->vn_buf_size = 0; + } + tb->vn_buf = buf; + schedule() ; + return REPEAT_SEARCH; + } + + tb->vn_buf = buf; + } + + if ( check_fs && FILESYSTEM_CHANGED_TB (tb) ) + return REPEAT_SEARCH; + + return CARRY_ON; +} + + +#ifdef CONFIG_REISERFS_CHECK +static void tb_buffer_sanity_check (struct super_block * p_s_sb, + struct buffer_head * p_s_bh, + const char *descr, int level) { + if (p_s_bh) { + if (atomic_read (&(p_s_bh->b_count)) <= 0) { + + reiserfs_panic (p_s_sb, "tb_buffer_sanity_check(): negative or zero reference counter for buffer %s[%d] (%b)\n", descr, level, p_s_bh); + } + + if ( ! buffer_uptodate (p_s_bh) ) { + reiserfs_panic (p_s_sb, "tb_buffer_sanity_check(): buffer is not up to date %s[%d] (%b)\n", descr, level, p_s_bh); + } + + if ( ! B_IS_IN_TREE (p_s_bh) ) { + reiserfs_panic (p_s_sb, "tb_buffer_sanity_check(): buffer is not in tree %s[%d] (%b)\n", descr, level, p_s_bh); + } + + if (p_s_bh->b_dev != p_s_sb->s_dev || + p_s_bh->b_size != p_s_sb->s_blocksize || + p_s_bh->b_blocknr > SB_BLOCK_COUNT(p_s_sb)) { + reiserfs_panic (p_s_sb, "tb_buffer_sanity_check(): check failed for buffer %s[%d] (%b)\n", descr, level, p_s_bh); + } + } +} +#endif + +static void clear_all_dirty_bits(struct super_block *s, + struct buffer_head *bh) { + reiserfs_prepare_for_journal(s, bh, 0) ; +} + +static int wait_tb_buffers_until_unlocked (struct tree_balance * p_s_tb) +{ + struct buffer_head * locked; +#ifdef CONFIG_REISERFS_CHECK + int repeat_counter = 0; +#endif + int i; + + do { + + locked = NULL; + + for ( i = p_s_tb->tb_path->path_length; !locked && i > ILLEGAL_PATH_ELEMENT_OFFSET; i-- ) { + if ( PATH_OFFSET_PBUFFER (p_s_tb->tb_path, i) ) { + /* if I understand correctly, we can only be sure the last buffer + ** in the path is in the tree --clm + */ +#ifdef CONFIG_REISERFS_CHECK + if (PATH_PLAST_BUFFER(p_s_tb->tb_path) == + PATH_OFFSET_PBUFFER(p_s_tb->tb_path, i)) { + tb_buffer_sanity_check (p_s_tb->tb_sb, + PATH_OFFSET_PBUFFER (p_s_tb->tb_path, i), + "S", + p_s_tb->tb_path->path_length - i); + } +#endif + clear_all_dirty_bits(p_s_tb->tb_sb, + PATH_OFFSET_PBUFFER (p_s_tb->tb_path, i)) ; + + if ( buffer_locked (PATH_OFFSET_PBUFFER (p_s_tb->tb_path, i)) ) + locked = PATH_OFFSET_PBUFFER (p_s_tb->tb_path, i); + } + } + + for ( i = 0; !locked && i < MAX_HEIGHT && p_s_tb->insert_size[i]; i++ ) { + + if (p_s_tb->lnum[i] ) { + + if ( p_s_tb->L[i] ) { +#ifdef CONFIG_REISERFS_CHECK + tb_buffer_sanity_check (p_s_tb->tb_sb, p_s_tb->L[i], "L", i); +#endif + clear_all_dirty_bits(p_s_tb->tb_sb, p_s_tb->L[i]) ; + if ( buffer_locked (p_s_tb->L[i]) ) + locked = p_s_tb->L[i]; + } + + if ( !locked && p_s_tb->FL[i] ) { +#ifdef CONFIG_REISERFS_CHECK + tb_buffer_sanity_check (p_s_tb->tb_sb, p_s_tb->FL[i], "FL", i); +#endif + clear_all_dirty_bits(p_s_tb->tb_sb, p_s_tb->FL[i]) ; + if ( buffer_locked (p_s_tb->FL[i]) ) + locked = p_s_tb->FL[i]; + } + + if ( !locked && p_s_tb->CFL[i] ) { +#ifdef CONFIG_REISERFS_CHECK + tb_buffer_sanity_check (p_s_tb->tb_sb, p_s_tb->CFL[i], "CFL", i); +#endif + clear_all_dirty_bits(p_s_tb->tb_sb, p_s_tb->CFL[i]) ; + if ( buffer_locked (p_s_tb->CFL[i]) ) + locked = p_s_tb->CFL[i]; + } + + } + + if ( !locked && (p_s_tb->rnum[i]) ) { + + if ( p_s_tb->R[i] ) { +#ifdef CONFIG_REISERFS_CHECK + tb_buffer_sanity_check (p_s_tb->tb_sb, p_s_tb->R[i], "R", i); +#endif + clear_all_dirty_bits(p_s_tb->tb_sb, p_s_tb->R[i]) ; + if ( buffer_locked (p_s_tb->R[i]) ) + locked = p_s_tb->R[i]; + } + + + if ( !locked && p_s_tb->FR[i] ) { +#ifdef CONFIG_REISERFS_CHECK + tb_buffer_sanity_check (p_s_tb->tb_sb, p_s_tb->FR[i], "FR", i); +#endif + clear_all_dirty_bits(p_s_tb->tb_sb, p_s_tb->FR[i]) ; + if ( buffer_locked (p_s_tb->FR[i]) ) + locked = p_s_tb->FR[i]; + } + + if ( !locked && p_s_tb->CFR[i] ) { +#ifdef CONFIG_REISERFS_CHECK + tb_buffer_sanity_check (p_s_tb->tb_sb, p_s_tb->CFR[i], "CFR", i); +#endif + clear_all_dirty_bits(p_s_tb->tb_sb, p_s_tb->CFR[i]) ; + if ( buffer_locked (p_s_tb->CFR[i]) ) + locked = p_s_tb->CFR[i]; + } + } + } + /* as far as I can tell, this is not required. The FEB list seems + ** to be full of newly allocated nodes, which will never be locked, + ** dirty, or anything else. + ** To be safe, I'm putting in the checks and waits in. For the moment, + ** they are needed to keep the code in journal.c from complaining + ** about the buffer. That code is inside CONFIG_REISERFS_CHECK as well. + ** --clm + */ + for ( i = 0; !locked && i < MAX_FEB_SIZE; i++ ) { + if ( p_s_tb->FEB[i] ) { + clear_all_dirty_bits(p_s_tb->tb_sb, p_s_tb->FEB[i]) ; + if (buffer_locked(p_s_tb->FEB[i])) { + locked = p_s_tb->FEB[i] ; + } + } + } + + if (locked) { +#ifdef CONFIG_REISERFS_CHECK + repeat_counter++; + if ( (repeat_counter % 10000) == 0) { + reiserfs_warning ("wait_tb_buffers_until_released(): too many iterations waiting for buffer to unlock (%b)\n", locked); + + /* Don't loop forever. Try to recover from possible error. */ + + return ( FILESYSTEM_CHANGED_TB (p_s_tb) ) ? REPEAT_SEARCH : CARRY_ON; + } +#endif + __wait_on_buffer (locked); + if ( FILESYSTEM_CHANGED_TB (p_s_tb) ) { + return REPEAT_SEARCH; + } + } + + } while (locked); + + return CARRY_ON; +} + + +/* Prepare for balancing, that is + * get all necessary parents, and neighbors; + * analyze what and where should be moved; + * get sufficient number of new nodes; + * Balancing will start only after all resources will be collected at a time. + * + * When ported to SMP kernels, only at the last moment after all needed nodes + * are collected in cache, will the resources be locked using the usual + * textbook ordered lock acquisition algorithms. Note that ensuring that + * this code neither write locks what it does not need to write lock nor locks out of order + * will be a pain in the butt that could have been avoided. Grumble grumble. -Hans + * + * fix is meant in the sense of render unchanging + * + * Latency might be improved by first gathering a list of what buffers are needed + * and then getting as many of them in parallel as possible? -Hans + * + * Parameters: + * op_mode i - insert, d - delete, c - cut (truncate), p - paste (append) + * tb tree_balance structure; + * inum item number in S[h]; + * pos_in_item - comment this if you can + * ins_ih & ins_sd are used when inserting + * Returns: 1 - schedule occurred while the function worked; + * 0 - schedule didn't occur while the function worked; + * -1 - if no_disk_space + */ + + +int fix_nodes (int n_op_mode, + struct tree_balance * p_s_tb, + struct item_head * p_s_ins_ih, // item head of item being inserted + const void * data // inserted item or data to be pasted + ) { + int n_ret_value, + n_h, + n_item_num = PATH_LAST_POSITION(p_s_tb->tb_path); + int n_pos_in_item; + + /* we set wait_tb_buffers_run when we have to restore any dirty bits cleared + ** during wait_tb_buffers_run + */ + int wait_tb_buffers_run = 0 ; + int windex ; + struct buffer_head * p_s_tbS0 = PATH_PLAST_BUFFER(p_s_tb->tb_path); + + n_pos_in_item = p_s_tb->tb_path->pos_in_item; + + + p_s_tb->fs_gen = get_generation (p_s_tb->tb_sb); + + /* we prepare and log the super here so it will already be in the + ** transaction when do_balance needs to change it. + ** This way do_balance won't have to schedule when trying to prepare + ** the super for logging + */ + reiserfs_prepare_for_journal(p_s_tb->tb_sb, + SB_BUFFER_WITH_SB(p_s_tb->tb_sb), 1) ; + journal_mark_dirty(p_s_tb->transaction_handle, p_s_tb->tb_sb, + SB_BUFFER_WITH_SB(p_s_tb->tb_sb)) ; + if ( FILESYSTEM_CHANGED_TB (p_s_tb) ) + return REPEAT_SEARCH; + + /* if it possible in indirect_to_direct conversion */ + if (buffer_locked (p_s_tbS0)) { + __wait_on_buffer (p_s_tbS0); + if ( FILESYSTEM_CHANGED_TB (p_s_tb) ) + return REPEAT_SEARCH; + } + +#ifndef __KERNEL__ + if ( atomic_read (&(p_s_tbS0->b_count)) > 1 || + (p_s_tb->L[0] && atomic_read (&(p_s_tb->L[0]->b_count)) > 1) || + (p_s_tb->R[0] && atomic_read (&(p_s_tb->R[0]->b_count)) > 1) ) { + printk ("mode=%c, insert_size=%d\n", n_op_mode, p_s_tb->insert_size[0]); + print_cur_tb ("first three parameters are invalid"); + reiserfs_panic (p_s_tb->tb_sb, "PAP-8310: fix_nodes: all buffers must be hold once in one thread processing"); + } +#endif + +#ifdef CONFIG_REISERFS_CHECK + if ( cur_tb ) { + print_cur_tb ("fix_nodes"); + reiserfs_panic(p_s_tb->tb_sb,"PAP-8305: fix_nodes: there is pending do_balance"); + } + + if (!buffer_uptodate (p_s_tbS0) || !B_IS_IN_TREE (p_s_tbS0)) { + reiserfs_panic (p_s_tb->tb_sb, "PAP-8320: fix_nodes: S[0] (%b %z) is not uptodate " + "at the beginning of fix_nodes or not in tree (mode %c)", p_s_tbS0, p_s_tbS0, n_op_mode); + } + + // FIXME: new items have to be of 8 byte multiples. Including new + // directory items those look like old ones + /* + if (p_s_tb->insert_size[0] % 8) + reiserfs_panic (p_s_tb->tb_sb, "vs-: fix_nodes: incorrect insert_size %d, " + "mode %c", + p_s_tb->insert_size[0], n_op_mode); + */ + + /* Check parameters. */ + switch (n_op_mode) { +#ifdef REISERFS_FSCK + case M_INTERNAL: + break; + case M_INSERT: + if ( n_item_num < 0 || n_item_num > B_NR_ITEMS(p_s_tbS0) ) + reiserfs_panic(p_s_tb->tb_sb,"PAP-8325: fix_nodes: Incorrect item number %d (in S0 - %d) in case of insert", + n_item_num, B_NR_ITEMS(p_s_tbS0)); +#else + case M_INSERT: + if ( n_item_num <= 0 || n_item_num > B_NR_ITEMS(p_s_tbS0) ) + reiserfs_panic(p_s_tb->tb_sb,"PAP-8330: fix_nodes: Incorrect item number %d (in S0 - %d) in case of insert", + n_item_num, B_NR_ITEMS(p_s_tbS0)); +#endif + break; + case M_PASTE: + case M_DELETE: + case M_CUT: + if ( n_item_num < 0 || n_item_num >= B_NR_ITEMS(p_s_tbS0) ) { + print_block (p_s_tbS0, 0, -1, -1); + printk("mode = %c insert_size = %d\n", n_op_mode, p_s_tb->insert_size[0]); + reiserfs_panic(p_s_tb->tb_sb,"PAP-8335: fix_nodes: Incorrect item number(%d)", n_item_num); + } + break; + default: + reiserfs_panic(p_s_tb->tb_sb,"PAP-8340: fix_nodes: Incorrect mode of operation"); + } +#endif + + if (get_mem_for_virtual_node (p_s_tb) == REPEAT_SEARCH) + // FIXME: maybe -ENOMEM when tb->vn_buf == 0? Now just repeat + return REPEAT_SEARCH; + + + /* Starting from the leaf level; for all levels n_h of the tree. */ + for ( n_h = 0; n_h < MAX_HEIGHT && p_s_tb->insert_size[n_h]; n_h++ ) { + if ( (n_ret_value = get_direct_parent(p_s_tb, n_h)) != CARRY_ON ) { + goto repeat; + return n_ret_value; + } + + if ( (n_ret_value = check_balance (n_op_mode, p_s_tb, n_h, n_item_num, + n_pos_in_item, p_s_ins_ih, data)) != CARRY_ON ) { + if ( n_ret_value == NO_BALANCING_NEEDED ) { + /* No balancing for higher levels needed. */ + if ( (n_ret_value = get_neighbors(p_s_tb, n_h)) != CARRY_ON ) { + goto repeat; + return n_ret_value; + } + if ( n_h != MAX_HEIGHT - 1 ) + p_s_tb->insert_size[n_h + 1] = 0; + /* ok, analysis and resource gathering are complete */ + break; + } + goto repeat; + return n_ret_value; + } + + if ( (n_ret_value = get_neighbors(p_s_tb, n_h)) != CARRY_ON ) { + goto repeat; + return n_ret_value; + } + + if ( (n_ret_value = get_empty_nodes(p_s_tb, n_h)) != CARRY_ON ) { + goto repeat; + return n_ret_value; /* No disk space, or schedule occurred and + analysis may be invalid and needs to be redone. */ + } + + if ( ! PATH_H_PBUFFER(p_s_tb->tb_path, n_h) ) { + /* We have a positive insert size but no nodes exist on this + level, this means that we are creating a new root. */ + +#ifdef CONFIG_REISERFS_CHECK + if ( p_s_tb->blknum[n_h] != 1 ) + reiserfs_panic(p_s_tb->tb_sb,"PAP-8350: fix_nodes: creating new empty root"); +#endif /* CONFIG_REISERFS_CHECK */ + + if ( n_h < MAX_HEIGHT - 1 ) + p_s_tb->insert_size[n_h + 1] = 0; + } + else + if ( ! PATH_H_PBUFFER(p_s_tb->tb_path, n_h + 1) ) { + if ( p_s_tb->blknum[n_h] > 1 ) { + /* The tree needs to be grown, so this node S[n_h] + which is the root node is split into two nodes, + and a new node (S[n_h+1]) will be created to + become the root node. */ + +#ifdef CONFIG_REISERFS_CHECK + if ( n_h == MAX_HEIGHT - 1 ) + reiserfs_panic(p_s_tb->tb_sb, "PAP-8355: fix_nodes: attempt to create too high of a tree"); +#endif /* CONFIG_REISERFS_CHECK */ + + p_s_tb->insert_size[n_h + 1] = (DC_SIZE + KEY_SIZE) * (p_s_tb->blknum[n_h] - 1) + DC_SIZE; + } + else + if ( n_h < MAX_HEIGHT - 1 ) + p_s_tb->insert_size[n_h + 1] = 0; + } + else + p_s_tb->insert_size[n_h + 1] = (DC_SIZE + KEY_SIZE) * (p_s_tb->blknum[n_h] - 1); + } + + + windex = push_journal_writer("fix_nodes") ; + if ((n_ret_value = wait_tb_buffers_until_unlocked (p_s_tb)) == CARRY_ON) { + pop_journal_writer(windex) ; + if (FILESYSTEM_CHANGED_TB(p_s_tb)) { + wait_tb_buffers_run = 1 ; + n_ret_value = REPEAT_SEARCH ; + goto repeat; + } else { + return CARRY_ON; + } + } else { + wait_tb_buffers_run = 1 ; + pop_journal_writer(windex) ; + goto repeat; + } + + repeat: + // fix_nodes was unable to perform its calculation due to + // filesystem got changed under us, lack of free disk space or i/o + // failure. If the first is the case - the search will be + // repeated. For now - free all resources acquired so far except + // for the new allocated nodes + { + int i; + + /* Release path buffers. */ + if (wait_tb_buffers_run) { + pathrelse_and_restore(p_s_tb->tb_sb, p_s_tb->tb_path) ; + } else { + pathrelse (p_s_tb->tb_path); + } + /* brelse all resources collected for balancing */ + for ( i = 0; i < MAX_HEIGHT; i++ ) { + if (wait_tb_buffers_run) { + reiserfs_restore_prepared_buffer(p_s_tb->tb_sb, p_s_tb->L[i]); + reiserfs_restore_prepared_buffer(p_s_tb->tb_sb, p_s_tb->R[i]); + reiserfs_restore_prepared_buffer(p_s_tb->tb_sb, p_s_tb->FL[i]); + reiserfs_restore_prepared_buffer(p_s_tb->tb_sb, p_s_tb->FR[i]); + reiserfs_restore_prepared_buffer(p_s_tb->tb_sb, p_s_tb->CFL[i]); + reiserfs_restore_prepared_buffer(p_s_tb->tb_sb, p_s_tb->CFR[i]); + } + + brelse (p_s_tb->L[i]);p_s_tb->L[i] = 0; + brelse (p_s_tb->R[i]);p_s_tb->R[i] = 0; + brelse (p_s_tb->FL[i]);p_s_tb->FL[i] = 0; + brelse (p_s_tb->FR[i]);p_s_tb->FR[i] = 0; + brelse (p_s_tb->CFL[i]);p_s_tb->CFL[i] = 0; + brelse (p_s_tb->CFR[i]);p_s_tb->CFR[i] = 0; + } + + if (wait_tb_buffers_run) { + for ( i = 0; i < MAX_FEB_SIZE; i++ ) { + if ( p_s_tb->FEB[i] ) { + reiserfs_restore_prepared_buffer(p_s_tb->tb_sb, + p_s_tb->FEB[i]) ; + } + } + } + return n_ret_value; + } + +} + + +/* Anatoly will probably forgive me renaming p_s_tb to tb. I just + wanted to make lines shorter */ +void unfix_nodes (struct tree_balance * tb) +{ + int i; + +#ifdef CONFIG_REISERFS_CHECK + if ( ! tb->vn_buf ) + reiserfs_panic (tb->tb_sb, + "PAP-16050: unfix_nodes: pointer to the virtual node is NULL"); +#endif + + /* Release path buffers. */ + pathrelse_and_restore (tb->tb_sb, tb->tb_path); + + /* brelse all resources collected for balancing */ + for ( i = 0; i < MAX_HEIGHT; i++ ) { + reiserfs_restore_prepared_buffer (tb->tb_sb, tb->L[i]); + reiserfs_restore_prepared_buffer (tb->tb_sb, tb->R[i]); + reiserfs_restore_prepared_buffer (tb->tb_sb, tb->FL[i]); + reiserfs_restore_prepared_buffer (tb->tb_sb, tb->FR[i]); + reiserfs_restore_prepared_buffer (tb->tb_sb, tb->CFL[i]); + reiserfs_restore_prepared_buffer (tb->tb_sb, tb->CFR[i]); + + brelse (tb->L[i]); + brelse (tb->R[i]); + brelse (tb->FL[i]); + brelse (tb->FR[i]); + brelse (tb->CFL[i]); + brelse (tb->CFR[i]); + } + + /* deal with list of allocated (used and unused) nodes */ + for ( i = 0; i < MAX_FEB_SIZE; i++ ) { + if ( tb->FEB[i] ) { + unsigned long blocknr = tb->FEB[i]->b_blocknr ; + /* de-allocated block which was not used by balancing and + bforget about buffer for it */ + brelse (tb->FEB[i]); + reiserfs_free_block (tb->transaction_handle, blocknr); + } + if (tb->used[i]) { + /* release used as new nodes including a new root */ + brelse (tb->used[i]); + } + } + +#if 0 /* shouldn't this be in CONFIG_REISERFS_CHECK??? */ + /* make sure, that all we have released got really freed */ + for (i = 0; i < sizeof (tb->thrown) / sizeof (tb->thrown[0]); i ++) + if (tb->thrown[i]) { + if (atomic_read (&(tb->thrown[i]->b_count))) { + /* the log will have the count at one and the buffers marked */ + if (atomic_read(&(tb->thrown[i]->b_count)) > 1 || + !(buffer_journaled(tb->thrown[i]) || + buffer_journal_dirty(tb->thrown[i]))) { + foo_print (tb->thrown[i], tb->tb_sb); + printk ("unfix_nodes: Waiting...(block %lu, count %d)\n", + tb->thrown[i]->b_blocknr, + atomic_read (&(tb->thrown[i]->b_count))); + wait_buffer_until_released (tb->thrown[i]); + printk ("unfix_nodes: Done (block %lu, count %d)\n", + tb->thrown[i]->b_blocknr, + atomic_read (&(tb->thrown[i]->b_count))); + } + } + } +#endif /* 0 */ + reiserfs_kfree (tb->vn_buf, tb->vn_buf_size, tb->tb_sb); + +} + + + +#ifndef REISERFS_FSCK + +// is_left_mergeable is now one of the item methods + +#else + +// this works only in fsck + +int are_items_mergeable (struct item_head * left, struct item_head * right, int bsize) +{ + if (comp_keys (&left->ih_key, &right->ih_key) != -1) { + reiserfs_panic (0, "vs-16070: are_items_mergeable: left %k, right %k", &(left->ih_key), &(right->ih_key)); + } + + if (comp_short_keys (&left->ih_key, &right->ih_key)) + return 0; + + if (I_IS_DIRECTORY_ITEM (left)) { + return 1; + } + + if ((I_IS_DIRECT_ITEM (left) && I_IS_DIRECT_ITEM (right)) || + (I_IS_INDIRECT_ITEM (left) && I_IS_INDIRECT_ITEM (right))) + return (left->ih_key.k_offset + I_BYTES_NUMBER (left, bsize) == right->ih_key.k_offset) ? 1 : 0; + + return 0; +} + +/* get left neighbor of the leaf node */ +static struct buffer_head * get_left_neighbor (struct super_block * s, struct path * path) +{ + struct key key; + INITIALIZE_PATH (path_to_left_neighbor); + struct buffer_head * bh; + + copy_key (&key, B_N_PKEY (PATH_PLAST_BUFFER (path), 0)); + decrement_key (&key); + +/* init_path (&path_to_left_neighbor);*/ + search_by_key (s, &key, &path_to_left_neighbor, DISK_LEAF_NODE_LEVEL, READ_BLOCKS); + // FIXME: fsck is to handle I/O failures somehow as well + if (PATH_LAST_POSITION (&path_to_left_neighbor) == 0) { + pathrelse (&path_to_left_neighbor); + return 0; + } + bh = PATH_PLAST_BUFFER (&path_to_left_neighbor); + bh->b_count ++; + pathrelse (&path_to_left_neighbor); + return bh; +} + +extern struct key MIN_KEY; +static struct buffer_head * get_right_neighbor (struct super_block * s, struct path * path) +{ + struct key key; + struct key * rkey; + INITIALIZE_PATH (path_to_right_neighbor); + struct buffer_head * bh; + + rkey = get_rkey (path, s); + if (comp_keys (rkey, &MIN_KEY) == 0) + reiserfs_panic (s, "vs-16080: get_right_neighbor: get_rkey returned min key (path has changed)"); + copy_key (&key, rkey); + + + /*init_path (&path_to_right_neighbor);*/ + search_by_key (s, &key, &path_to_right_neighbor, DISK_LEAF_NODE_LEVEL, READ_BLOCKS); + if (PATH_PLAST_BUFFER (&path_to_right_neighbor) == PATH_PLAST_BUFFER (path)) { + pathrelse (&path_to_right_neighbor); + return 0; + } + bh = PATH_PLAST_BUFFER (&path_to_right_neighbor); + bh->b_count ++; + pathrelse (&path_to_right_neighbor); + return bh; +} + + +int is_left_mergeable (struct super_block * s, struct path * path) +{ + struct item_head * right; + struct buffer_head * bh; + int retval; + + right = B_N_PITEM_HEAD (PATH_PLAST_BUFFER (path), 0); + + bh = get_left_neighbor (s, path); + if (bh == 0) { + return 0; + } + retval = are_items_mergeable (B_N_PITEM_HEAD (bh, B_NR_ITEMS (bh) - 1), right, bh->b_size); + brelse (bh); + return retval; +} + + +int is_right_mergeable (struct super_block * s, struct path * path) +{ + struct item_head * left; + struct buffer_head * bh; + int retval; + + left = B_N_PITEM_HEAD (PATH_PLAST_BUFFER (path), B_NR_ITEMS (PATH_PLAST_BUFFER (path)) - 1); + + bh = get_right_neighbor (s, path); + if (bh == 0) { + return 0; + } + retval = are_items_mergeable (left, B_N_PITEM_HEAD (bh, 0), bh->b_size); + brelse (bh); + return retval; +} + +#endif /* REISERFS_FSCK */ + + + + + diff --git a/fs/reiserfs/hashes.c b/fs/reiserfs/hashes.c new file mode 100644 index 000000000..b707aec4b --- /dev/null +++ b/fs/reiserfs/hashes.c @@ -0,0 +1,226 @@ + +/* + * Keyed 32-bit hash function using TEA in a Davis-Meyer function + * H0 = Key + * Hi = E Mi(Hi-1) + Hi-1 + * + * (see Applied Cryptography, 2nd edition, p448). + * + * Jeremy Fitzhardinge <jeremy@zip.com.au> 1998 + * + * Jeremy has agreed to the contents of reiserfs/README. -Hans + * Yura's function is added (04/07/2000) + */ + +// +// keyed_hash +// yura_hash +// r5_hash +// + +#include <asm/types.h> + + + +#define DELTA 0x9E3779B9 +#define FULLROUNDS 10 /* 32 is overkill, 16 is strong crypto */ +#define PARTROUNDS 6 /* 6 gets complete mixing */ + +#ifndef __KERNEL__ +typedef __u32 u32; +#endif + +/* a, b, c, d - data; h0, h1 - accumulated hash */ +#define TEACORE(rounds) \ + do { \ + u32 sum = 0; \ + int n = rounds; \ + u32 b0, b1; \ + \ + b0 = h0; \ + b1 = h1; \ + \ + do \ + { \ + sum += DELTA; \ + b0 += ((b1 << 4)+a) ^ (b1+sum) ^ ((b1 >> 5)+b); \ + b1 += ((b0 << 4)+c) ^ (b0+sum) ^ ((b0 >> 5)+d); \ + } while(--n); \ + \ + h0 += b0; \ + h1 += b1; \ + } while(0) + + +u32 keyed_hash(const char *msg, int len) +{ + u32 k[] = { 0x9464a485, 0x542e1a94, 0x3e846bff, 0xb75bcfc3}; + + u32 h0 = k[0], h1 = k[1]; + u32 a, b, c, d; + u32 pad; + int i; + + + // assert(len >= 0 && len < 256); + + pad = (u32)len | ((u32)len << 8); + pad |= pad << 16; + + while(len >= 16) + { + a = (u32)msg[ 0] | + (u32)msg[ 1] << 8 | + (u32)msg[ 2] << 16| + (u32)msg[ 3] << 24; + b = (u32)msg[ 4] | + (u32)msg[ 5] << 8 | + (u32)msg[ 6] << 16| + (u32)msg[ 7] << 24; + c = (u32)msg[ 8] | + (u32)msg[ 9] << 8 | + (u32)msg[10] << 16| + (u32)msg[11] << 24; + d = (u32)msg[12] | + (u32)msg[13] << 8 | + (u32)msg[14] << 16| + (u32)msg[15] << 24; + + TEACORE(PARTROUNDS); + + len -= 16; + msg += 16; + } + + if (len >= 12) + { + //assert(len < 16); + if (len >= 16) + *(int *)0 = 0; + + a = (u32)msg[ 0] | + (u32)msg[ 1] << 8 | + (u32)msg[ 2] << 16| + (u32)msg[ 3] << 24; + b = (u32)msg[ 4] | + (u32)msg[ 5] << 8 | + (u32)msg[ 6] << 16| + (u32)msg[ 7] << 24; + c = (u32)msg[ 8] | + (u32)msg[ 9] << 8 | + (u32)msg[10] << 16| + (u32)msg[11] << 24; + + d = pad; + for(i = 12; i < len; i++) + { + d <<= 8; + d |= msg[i]; + } + } + else if (len >= 8) + { + //assert(len < 12); + if (len >= 12) + *(int *)0 = 0; + a = (u32)msg[ 0] | + (u32)msg[ 1] << 8 | + (u32)msg[ 2] << 16| + (u32)msg[ 3] << 24; + b = (u32)msg[ 4] | + (u32)msg[ 5] << 8 | + (u32)msg[ 6] << 16| + (u32)msg[ 7] << 24; + + c = d = pad; + for(i = 8; i < len; i++) + { + c <<= 8; + c |= msg[i]; + } + } + else if (len >= 4) + { + //assert(len < 8); + if (len >= 8) + *(int *)0 = 0; + a = (u32)msg[ 0] | + (u32)msg[ 1] << 8 | + (u32)msg[ 2] << 16| + (u32)msg[ 3] << 24; + + b = c = d = pad; + for(i = 4; i < len; i++) + { + b <<= 8; + b |= msg[i]; + } + } + else + { + //assert(len < 4); + if (len >= 4) + *(int *)0 = 0; + a = b = c = d = pad; + for(i = 0; i < len; i++) + { + a <<= 8; + a |= msg[i]; + } + } + + TEACORE(FULLROUNDS); + +/* return 0;*/ + return h0^h1; +} + +/* What follows in this file is copyright 2000 by Hans Reiser, and the + * licensing of what follows is governed by reiserfs/README */ + +u32 yura_hash (const char *msg, int len) +{ + int j, pow; + u32 a, c; + int i; + + for (pow=1,i=1; i < len; i++) pow = pow * 10; + + if (len == 1) + a = msg[0]-48; + else + a = (msg[0] - 48) * pow; + + for (i=1; i < len; i++) { + c = msg[i] - 48; + for (pow=1,j=i; j < len-1; j++) pow = pow * 10; + a = a + c * pow; + } + + for (; i < 40; i++) { + c = '0' - 48; + for (pow=1,j=i; j < len-1; j++) pow = pow * 10; + a = a + c * pow; + } + + for (; i < 256; i++) { + c = i; + for (pow=1,j=i; j < len-1; j++) pow = pow * 10; + a = a + c * pow; + } + + a = a << 7; + return a; +} + +u32 r5_hash (const char *msg, int len) +{ + u32 a=0; + while(*msg) { + a += *msg << 4; + a += *msg >> 4; + a *= 11; + msg++; + } + return a; +} diff --git a/fs/reiserfs/ibalance.c b/fs/reiserfs/ibalance.c new file mode 100644 index 000000000..6007d5158 --- /dev/null +++ b/fs/reiserfs/ibalance.c @@ -0,0 +1,1140 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <asm/uaccess.h> +#include <linux/string.h> +#include <linux/sched.h> +#include <linux/reiserfs_fs.h> + +#else + +#include "nokernel.h" + +#endif + + +/* this is one and only function that is used outside (do_balance.c) */ +int balance_internal ( + struct tree_balance * , + int, + int, + struct item_head * , + struct buffer_head ** + ); + +/* modes of internal_shift_left, internal_shift_right and internal_insert_childs */ +#define INTERNAL_SHIFT_FROM_S_TO_L 0 +#define INTERNAL_SHIFT_FROM_R_TO_S 1 +#define INTERNAL_SHIFT_FROM_L_TO_S 2 +#define INTERNAL_SHIFT_FROM_S_TO_R 3 +#define INTERNAL_INSERT_TO_S 4 +#define INTERNAL_INSERT_TO_L 5 +#define INTERNAL_INSERT_TO_R 6 + +static void internal_define_dest_src_infos ( + int shift_mode, + struct tree_balance * tb, + int h, + struct buffer_info * dest_bi, + struct buffer_info * src_bi, + int * d_key, + struct buffer_head ** cf + ) +{ +#ifdef CONFIG_REISERFS_CHECK + memset (dest_bi, 0, sizeof (struct buffer_info)); + memset (src_bi, 0, sizeof (struct buffer_info)); +#endif + /* define dest, src, dest parent, dest position */ + switch (shift_mode) { + case INTERNAL_SHIFT_FROM_S_TO_L: /* used in internal_shift_left */ + src_bi->tb = tb; + src_bi->bi_bh = PATH_H_PBUFFER (tb->tb_path, h); + src_bi->bi_parent = PATH_H_PPARENT (tb->tb_path, h); + src_bi->bi_position = PATH_H_POSITION (tb->tb_path, h + 1); + dest_bi->tb = tb; + dest_bi->bi_bh = tb->L[h]; + dest_bi->bi_parent = tb->FL[h]; + dest_bi->bi_position = get_left_neighbor_position (tb, h); + *d_key = tb->lkey[h]; + *cf = tb->CFL[h]; + break; + case INTERNAL_SHIFT_FROM_L_TO_S: + src_bi->tb = tb; + src_bi->bi_bh = tb->L[h]; + src_bi->bi_parent = tb->FL[h]; + src_bi->bi_position = get_left_neighbor_position (tb, h); + dest_bi->tb = tb; + dest_bi->bi_bh = PATH_H_PBUFFER (tb->tb_path, h); + dest_bi->bi_parent = PATH_H_PPARENT (tb->tb_path, h); + dest_bi->bi_position = PATH_H_POSITION (tb->tb_path, h + 1); /* dest position is analog of dest->b_item_order */ + *d_key = tb->lkey[h]; + *cf = tb->CFL[h]; + break; + + case INTERNAL_SHIFT_FROM_R_TO_S: /* used in internal_shift_left */ + src_bi->tb = tb; + src_bi->bi_bh = tb->R[h]; + src_bi->bi_parent = tb->FR[h]; + src_bi->bi_position = get_right_neighbor_position (tb, h); + dest_bi->tb = tb; + dest_bi->bi_bh = PATH_H_PBUFFER (tb->tb_path, h); + dest_bi->bi_parent = PATH_H_PPARENT (tb->tb_path, h); + dest_bi->bi_position = PATH_H_POSITION (tb->tb_path, h + 1); + *d_key = tb->rkey[h]; + *cf = tb->CFR[h]; + break; + + case INTERNAL_SHIFT_FROM_S_TO_R: + src_bi->tb = tb; + src_bi->bi_bh = PATH_H_PBUFFER (tb->tb_path, h); + src_bi->bi_parent = PATH_H_PPARENT (tb->tb_path, h); + src_bi->bi_position = PATH_H_POSITION (tb->tb_path, h + 1); + dest_bi->tb = tb; + dest_bi->bi_bh = tb->R[h]; + dest_bi->bi_parent = tb->FR[h]; + dest_bi->bi_position = get_right_neighbor_position (tb, h); + *d_key = tb->rkey[h]; + *cf = tb->CFR[h]; + break; + + case INTERNAL_INSERT_TO_L: + dest_bi->tb = tb; + dest_bi->bi_bh = tb->L[h]; + dest_bi->bi_parent = tb->FL[h]; + dest_bi->bi_position = get_left_neighbor_position (tb, h); + break; + + case INTERNAL_INSERT_TO_S: + dest_bi->tb = tb; + dest_bi->bi_bh = PATH_H_PBUFFER (tb->tb_path, h); + dest_bi->bi_parent = PATH_H_PPARENT (tb->tb_path, h); + dest_bi->bi_position = PATH_H_POSITION (tb->tb_path, h + 1); + break; + + case INTERNAL_INSERT_TO_R: + dest_bi->tb = tb; + dest_bi->bi_bh = tb->R[h]; + dest_bi->bi_parent = tb->FR[h]; + dest_bi->bi_position = get_right_neighbor_position (tb, h); + break; + + default: + reiserfs_panic (tb->tb_sb, "internal_define_dest_src_infos", "shift type is unknown (%d)", shift_mode); + } +} + + + +/* Insert count node pointers into buffer cur before position to + 1. + * Insert count items into buffer cur before position to. + * Items and node pointers are specified by inserted and bh respectively. + */ +static void internal_insert_childs (struct buffer_info * cur_bi, + int to, int count, + struct item_head * inserted, + struct buffer_head ** bh + ) +{ + struct buffer_head * cur = cur_bi->bi_bh; + struct block_head * blkh; + int nr; + struct key * ih; + struct disk_child new_dc[2]; + struct disk_child * dc; + int i; + + if (count <= 0) + return; + + nr = le16_to_cpu ((blkh = B_BLK_HEAD(cur))->blk_nr_item); + +#ifdef CONFIG_REISERFS_CHECK + if (count > 2) + reiserfs_panic (0, "internal_insert_childs", "too many children (%d) are to be inserted", count); + if (B_FREE_SPACE (cur) < count * (KEY_SIZE + DC_SIZE)) + reiserfs_panic (0, "internal_insert_childs", "no enough free space (%d), needed %d bytes", + B_FREE_SPACE (cur), count * (KEY_SIZE + DC_SIZE)); +#endif /* CONFIG_REISERFS_CHECK */ + + /* prepare space for count disk_child */ + dc = B_N_CHILD(cur,to+1); + + memmove (dc + count, dc, (nr+1-(to+1)) * DC_SIZE); + + /* copy to_be_insert disk children */ + for (i = 0; i < count; i ++) { + new_dc[i].dc_size = + cpu_to_le16 (MAX_CHILD_SIZE(bh[i]) - B_FREE_SPACE (bh[i])); + new_dc[i].dc_block_number = cpu_to_le32 (bh[i]->b_blocknr); + } + memcpy (dc, new_dc, DC_SIZE * count); + + + /* prepare space for count items */ + ih = B_N_PDELIM_KEY (cur, ((to == -1) ? 0 : to)); + + memmove (ih + count, ih, (nr - to) * KEY_SIZE + (nr + 1 + count) * DC_SIZE); + + /* copy item headers (keys) */ + memcpy (ih, inserted, KEY_SIZE); + if ( count > 1 ) + memcpy (ih + 1, inserted + 1, KEY_SIZE); + + /* sizes, item number */ + blkh->blk_nr_item = cpu_to_le16 (le16_to_cpu (blkh->blk_nr_item) + count); + blkh->blk_free_space = cpu_to_le16 (le16_to_cpu (blkh->blk_free_space) - count * (DC_SIZE + KEY_SIZE)); + + do_balance_mark_internal_dirty (cur_bi->tb, cur,0); + + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + check_internal (cur); + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + + if (cur_bi->bi_parent) { + B_N_CHILD (cur_bi->bi_parent,cur_bi->bi_position)->dc_size += count * (DC_SIZE + KEY_SIZE); + do_balance_mark_internal_dirty(cur_bi->tb, cur_bi->bi_parent, 0); + + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + check_internal (cur_bi->bi_parent); + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + } + +} + + +/* Delete del_num items and node pointers from buffer cur starting from * + * the first_i'th item and first_p'th pointers respectively. */ +static void internal_delete_pointers_items ( + struct buffer_info * cur_bi, + int first_p, + int first_i, + int del_num + ) +{ + struct buffer_head * cur = cur_bi->bi_bh; + int nr; + struct block_head * blkh; + struct key * key; + struct disk_child * dc; + +#ifdef CONFIG_REISERFS_CHECK + if (cur == NULL) + reiserfs_panic (0, "internal_delete_pointers_items1: buffer is 0"); + + if (del_num < 0) + reiserfs_panic (0, "internal_delete_pointers_items2", + "negative number of items (%d) can not be deleted", del_num); + + if (first_p < 0 || first_p + del_num > B_NR_ITEMS (cur) + 1 || first_i < 0) + reiserfs_panic (0, "internal_delete_pointers_items3", + "first pointer order (%d) < 0 or " + "no so many pointers (%d), only (%d) or " + "first key order %d < 0", first_p, + first_p + del_num, B_NR_ITEMS (cur) + 1, first_i); +#endif /* CONFIG_REISERFS_CHECK */ + if ( del_num == 0 ) + return; + + nr = le16_to_cpu ((blkh = B_BLK_HEAD(cur))->blk_nr_item); + + if ( first_p == 0 && del_num == nr + 1 ) { +#ifdef CONFIG_REISERFS_CHECK + if ( first_i != 0 ) + reiserfs_panic (0, "internal_delete_pointers_items5", + "first deleted key must have order 0, not %d", first_i); +#endif /* CONFIG_REISERFS_CHECK */ + make_empty_node (cur_bi); + return; + } + +#ifdef CONFIG_REISERFS_CHECK + if (first_i + del_num > B_NR_ITEMS (cur)) { + printk("first_i = %d del_num = %d\n",first_i,del_num); + reiserfs_panic (0, "internal_delete_pointers_items4: :" + "no so many keys (%d) in the node (%b)(%z)", first_i + del_num, cur, cur); + } +#endif /* CONFIG_REISERFS_CHECK */ + + + /* deleting */ + dc = B_N_CHILD (cur, first_p); + + memmove (dc, dc + del_num, (nr + 1 - first_p - del_num) * DC_SIZE); + key = B_N_PDELIM_KEY (cur, first_i); + memmove (key, key + del_num, (nr - first_i - del_num) * KEY_SIZE + (nr + 1 - del_num) * DC_SIZE); + + + /* sizes, item number */ + blkh->blk_nr_item = cpu_to_le16 (le16_to_cpu (blkh->blk_nr_item) - del_num); + blkh->blk_free_space = cpu_to_le16 (le16_to_cpu (blkh->blk_free_space) + del_num * (KEY_SIZE + DC_SIZE)); + + do_balance_mark_internal_dirty (cur_bi->tb, cur, 0); + /*&&&&&&&&&&&&&&&&&&&&&&&*/ + check_internal (cur); + /*&&&&&&&&&&&&&&&&&&&&&&&*/ + + if (cur_bi->bi_parent) { + B_N_CHILD (cur_bi->bi_parent, cur_bi->bi_position)->dc_size -= del_num * (KEY_SIZE + DC_SIZE); + do_balance_mark_internal_dirty (cur_bi->tb, cur_bi->bi_parent,0); + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + check_internal (cur_bi->bi_parent); + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + } +} + + +/* delete n node pointers and items starting from given position */ +static void internal_delete_childs (struct buffer_info * cur_bi, + int from, int n) +{ + int i_from; + + i_from = (from == 0) ? from : from - 1; + + /* delete n pointers starting from `from' position in CUR; + delete n keys starting from 'i_from' position in CUR; + */ + internal_delete_pointers_items (cur_bi, from, i_from, n); +} + + +/* copy cpy_num node pointers and cpy_num - 1 items from buffer src to buffer dest +* last_first == FIRST_TO_LAST means, that we copy first items from src to tail of dest + * last_first == LAST_TO_FIRST means, that we copy last items from src to head of dest + */ +static void internal_copy_pointers_items ( + struct buffer_info * dest_bi, + struct buffer_head * src, + int last_first, int cpy_num + ) +{ + /* ATTENTION! Number of node pointers in DEST is equal to number of items in DEST * + * as delimiting key have already inserted to buffer dest.*/ + struct buffer_head * dest = dest_bi->bi_bh; + int nr_dest, nr_src; + int dest_order, src_order; + struct block_head * blkh; + struct key * key; + struct disk_child * dc; + + nr_src = B_NR_ITEMS (src); + +#ifdef CONFIG_REISERFS_CHECK + if ( dest == NULL || src == NULL ) + reiserfs_panic (0, "internal_copy_pointers_items", "src (%p) or dest (%p) buffer is 0", src, dest); + + if (last_first != FIRST_TO_LAST && last_first != LAST_TO_FIRST) + reiserfs_panic (0, "internal_copy_pointers_items", + "invalid last_first parameter (%d)", last_first); + + if ( nr_src < cpy_num - 1 ) + reiserfs_panic (0, "internal_copy_pointers_items", "no so many items (%d) in src (%d)", cpy_num, nr_src); + + if ( cpy_num < 0 ) + reiserfs_panic (0, "internal_copy_pointers_items", "cpy_num less than 0 (%d)", cpy_num); + + if (cpy_num - 1 + B_NR_ITEMS(dest) > (int)MAX_NR_KEY(dest)) + reiserfs_panic (0, "internal_copy_pointers_items", + "cpy_num (%d) + item number in dest (%d) can not be more than MAX_NR_KEY(%d)", + cpy_num, B_NR_ITEMS(dest), MAX_NR_KEY(dest)); +#endif + + if ( cpy_num == 0 ) + return; + + /* coping */ + nr_dest = le16_to_cpu ((blkh = B_BLK_HEAD(dest))->blk_nr_item); + + /*dest_order = (last_first == LAST_TO_FIRST) ? 0 : nr_dest;*/ + /*src_order = (last_first == LAST_TO_FIRST) ? (nr_src - cpy_num + 1) : 0;*/ + (last_first == LAST_TO_FIRST) ? (dest_order = 0, src_order = nr_src - cpy_num + 1) : + (dest_order = nr_dest, src_order = 0); + + /* prepare space for cpy_num pointers */ + dc = B_N_CHILD (dest, dest_order); + + memmove (dc + cpy_num, dc, (nr_dest - dest_order) * DC_SIZE); + + /* insert pointers */ + memcpy (dc, B_N_CHILD (src, src_order), DC_SIZE * cpy_num); + + + /* prepare space for cpy_num - 1 item headers */ + key = B_N_PDELIM_KEY(dest, dest_order); + memmove (key + cpy_num - 1, key, + KEY_SIZE * (nr_dest - dest_order) + DC_SIZE * (nr_dest + cpy_num)); + + + /* insert headers */ + memcpy (key, B_N_PDELIM_KEY (src, src_order), KEY_SIZE * (cpy_num - 1)); + + /* sizes, item number */ + blkh->blk_nr_item = cpu_to_le16 (le16_to_cpu (blkh->blk_nr_item) + (cpy_num - 1)); + blkh->blk_free_space = cpu_to_le16 (le16_to_cpu (blkh->blk_free_space) - (KEY_SIZE * (cpy_num - 1) + DC_SIZE * cpy_num)); + + do_balance_mark_internal_dirty (dest_bi->tb, dest, 0); + + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + check_internal (dest); + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + + if (dest_bi->bi_parent) { + B_N_CHILD(dest_bi->bi_parent,dest_bi->bi_position)->dc_size += + KEY_SIZE * (cpy_num - 1) + DC_SIZE * cpy_num; + + do_balance_mark_internal_dirty (dest_bi->tb, dest_bi->bi_parent,0); + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + check_internal (dest_bi->bi_parent); + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + } + +} + + +/* Copy cpy_num node pointers and cpy_num - 1 items from buffer src to buffer dest. + * Delete cpy_num - del_par items and node pointers from buffer src. + * last_first == FIRST_TO_LAST means, that we copy/delete first items from src. + * last_first == LAST_TO_FIRST means, that we copy/delete last items from src. + */ +static void internal_move_pointers_items (struct buffer_info * dest_bi, + struct buffer_info * src_bi, + int last_first, int cpy_num, int del_par) +{ + int first_pointer; + int first_item; + + internal_copy_pointers_items (dest_bi, src_bi->bi_bh, last_first, cpy_num); + + if (last_first == FIRST_TO_LAST) { /* shift_left occurs */ + first_pointer = 0; + first_item = 0; + /* delete cpy_num - del_par pointers and keys starting for pointers with first_pointer, + for key - with first_item */ + internal_delete_pointers_items (src_bi, first_pointer, first_item, cpy_num - del_par); + } else { /* shift_right occurs */ + int i, j; + + i = ( cpy_num - del_par == ( j = B_NR_ITEMS(src_bi->bi_bh)) + 1 ) ? 0 : j - cpy_num + del_par; + + internal_delete_pointers_items (src_bi, j + 1 - cpy_num + del_par, i, cpy_num - del_par); + } +} + +/* Insert n_src'th key of buffer src before n_dest'th key of buffer dest. */ +static void internal_insert_key (struct buffer_info * dest_bi, + int dest_position_before, /* insert key before key with n_dest number */ + struct buffer_head * src, + int src_position) +{ + struct buffer_head * dest = dest_bi->bi_bh; + int nr; + struct block_head * blkh; + struct key * key; + +#ifdef CONFIG_REISERFS_CHECK + if (dest == NULL || src == NULL) + reiserfs_panic (0, "internal_insert_key", "sourse(%p) or dest(%p) buffer is 0", src, dest); + + if (dest_position_before < 0 || src_position < 0) + reiserfs_panic (0, "internal_insert_key", "source(%d) or dest(%d) key number less than 0", + src_position, dest_position_before); + + if (dest_position_before > B_NR_ITEMS (dest) || src_position >= B_NR_ITEMS(src)) + reiserfs_panic (0, "internal_insert_key", + "invalid position in dest (%d (key number %d)) or in src (%d (key number %d))", + dest_position_before, B_NR_ITEMS (dest), src_position, B_NR_ITEMS(src)); + + if (B_FREE_SPACE (dest) < KEY_SIZE) + reiserfs_panic (0, "internal_insert_key", + "no enough free space (%d) in dest buffer", B_FREE_SPACE (dest)); +#endif + + nr = le16_to_cpu ((blkh=B_BLK_HEAD(dest))->blk_nr_item); + + /* prepare space for inserting key */ + key = B_N_PDELIM_KEY (dest, dest_position_before); + memmove (key + 1, key, (nr - dest_position_before) * KEY_SIZE + (nr + 1) * DC_SIZE); + + /* insert key */ + memcpy (key, B_N_PDELIM_KEY(src, src_position), KEY_SIZE); + + /* Change dirt, free space, item number fields. */ + blkh->blk_nr_item = cpu_to_le16 (le16_to_cpu (blkh->blk_nr_item) + 1); + blkh->blk_free_space = cpu_to_le16 (le16_to_cpu (blkh->blk_free_space) - KEY_SIZE); + + do_balance_mark_internal_dirty (dest_bi->tb, dest, 0); + + if (dest_bi->bi_parent) { + B_N_CHILD(dest_bi->bi_parent,dest_bi->bi_position)->dc_size += KEY_SIZE; + do_balance_mark_internal_dirty (dest_bi->tb, dest_bi->bi_parent,0); + } +} + + + +/* Insert d_key'th (delimiting) key from buffer cfl to tail of dest. + * Copy pointer_amount node pointers and pointer_amount - 1 items from buffer src to buffer dest. + * Replace d_key'th key in buffer cfl. + * Delete pointer_amount items and node pointers from buffer src. + */ +/* this can be invoked both to shift from S to L and from R to S */ +static void internal_shift_left ( + int mode, /* INTERNAL_FROM_S_TO_L | INTERNAL_FROM_R_TO_S */ + struct tree_balance * tb, + int h, + int pointer_amount + ) +{ + struct buffer_info dest_bi, src_bi; + struct buffer_head * cf; + int d_key_position; + + internal_define_dest_src_infos (mode, tb, h, &dest_bi, &src_bi, &d_key_position, &cf); + + /*printk("pointer_amount = %d\n",pointer_amount);*/ + + if (pointer_amount) { + /* insert delimiting key from common father of dest and src to node dest into position B_NR_ITEM(dest) */ + internal_insert_key (&dest_bi, B_NR_ITEMS(dest_bi.bi_bh), cf, d_key_position); + + if (B_NR_ITEMS(src_bi.bi_bh) == pointer_amount - 1) { + if (src_bi.bi_position/*src->b_item_order*/ == 0) + replace_key (tb, cf, d_key_position, src_bi.bi_parent/*src->b_parent*/, 0); + } else + replace_key (tb, cf, d_key_position, src_bi.bi_bh, pointer_amount - 1); + } + /* last parameter is del_parameter */ + internal_move_pointers_items (&dest_bi, &src_bi, FIRST_TO_LAST, pointer_amount, 0); + +} + +/* Insert delimiting key to L[h]. + * Copy n node pointers and n - 1 items from buffer S[h] to L[h]. + * Delete n - 1 items and node pointers from buffer S[h]. + */ +/* it always shifts from S[h] to L[h] */ +static void internal_shift1_left ( + struct tree_balance * tb, + int h, + int pointer_amount + ) +{ + struct buffer_info dest_bi, src_bi; + struct buffer_head * cf; + int d_key_position; + + internal_define_dest_src_infos (INTERNAL_SHIFT_FROM_S_TO_L, tb, h, &dest_bi, &src_bi, &d_key_position, &cf); + + if ( pointer_amount > 0 ) /* insert lkey[h]-th key from CFL[h] to left neighbor L[h] */ + internal_insert_key (&dest_bi, B_NR_ITEMS(dest_bi.bi_bh), cf, d_key_position); + /* internal_insert_key (tb->L[h], B_NR_ITEM(tb->L[h]), tb->CFL[h], tb->lkey[h]);*/ + + /* last parameter is del_parameter */ + internal_move_pointers_items (&dest_bi, &src_bi, FIRST_TO_LAST, pointer_amount, 1); + /* internal_move_pointers_items (tb->L[h], tb->S[h], FIRST_TO_LAST, pointer_amount, 1);*/ +} + + +/* Insert d_key'th (delimiting) key from buffer cfr to head of dest. + * Copy n node pointers and n - 1 items from buffer src to buffer dest. + * Replace d_key'th key in buffer cfr. + * Delete n items and node pointers from buffer src. + */ +static void internal_shift_right ( + int mode, /* INTERNAL_FROM_S_TO_R | INTERNAL_FROM_L_TO_S */ + struct tree_balance * tb, + int h, + int pointer_amount + ) +{ + struct buffer_info dest_bi, src_bi; + struct buffer_head * cf; + int d_key_position; + int nr; + + + internal_define_dest_src_infos (mode, tb, h, &dest_bi, &src_bi, &d_key_position, &cf); + + nr = B_NR_ITEMS (src_bi.bi_bh); + + if (pointer_amount > 0) { + /* insert delimiting key from common father of dest and src to dest node into position 0 */ + internal_insert_key (&dest_bi, 0, cf, d_key_position); + if (nr == pointer_amount - 1) { +#ifdef CONFIG_REISERFS_CHECK + if ( src_bi.bi_bh != PATH_H_PBUFFER (tb->tb_path, h)/*tb->S[h]*/ || dest_bi.bi_bh != tb->R[h]) + reiserfs_panic (tb->tb_sb, "internal_shift_right", "src (%p) must be == tb->S[h](%p) when it disappears", + src_bi.bi_bh, PATH_H_PBUFFER (tb->tb_path, h)); +#endif + /* when S[h] disappers replace left delemiting key as well */ + if (tb->CFL[h]) + replace_key (tb, cf, d_key_position, tb->CFL[h], tb->lkey[h]); + } else + replace_key (tb, cf, d_key_position, src_bi.bi_bh, nr - pointer_amount); + } + + /* last parameter is del_parameter */ + internal_move_pointers_items (&dest_bi, &src_bi, LAST_TO_FIRST, pointer_amount, 0); +} + +/* Insert delimiting key to R[h]. + * Copy n node pointers and n - 1 items from buffer S[h] to R[h]. + * Delete n - 1 items and node pointers from buffer S[h]. + */ +/* it always shift from S[h] to R[h] */ +static void internal_shift1_right ( + struct tree_balance * tb, + int h, + int pointer_amount + ) +{ + struct buffer_info dest_bi, src_bi; + struct buffer_head * cf; + int d_key_position; + + internal_define_dest_src_infos (INTERNAL_SHIFT_FROM_S_TO_R, tb, h, &dest_bi, &src_bi, &d_key_position, &cf); + + if (pointer_amount > 0) /* insert rkey from CFR[h] to right neighbor R[h] */ + internal_insert_key (&dest_bi, 0, cf, d_key_position); + /* internal_insert_key (tb->R[h], 0, tb->CFR[h], tb->rkey[h]);*/ + + /* last parameter is del_parameter */ + internal_move_pointers_items (&dest_bi, &src_bi, LAST_TO_FIRST, pointer_amount, 1); + /* internal_move_pointers_items (tb->R[h], tb->S[h], LAST_TO_FIRST, pointer_amount, 1);*/ +} + + +/* Delete insert_num node pointers together with their left items + * and balance current node.*/ +static void balance_internal_when_delete (struct tree_balance * tb, + int h, int child_pos) +{ + int insert_num; + int n; + struct buffer_head * tbSh = PATH_H_PBUFFER (tb->tb_path, h); + struct buffer_info bi; + + insert_num = tb->insert_size[h] / ((int)(DC_SIZE + KEY_SIZE)); + + /* delete child-node-pointer(s) together with their left item(s) */ + bi.tb = tb; + bi.bi_bh = tbSh; + bi.bi_parent = PATH_H_PPARENT (tb->tb_path, h); + bi.bi_position = PATH_H_POSITION (tb->tb_path, h + 1); + + internal_delete_childs (&bi, child_pos, -insert_num); + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->blknum[h] > 1 ) + reiserfs_panic (tb->tb_sb, "balance_internal_when_delete", "tb->blknum[%d]=%d when insert_size < 0", + h, tb->blknum[h]); +#endif /* CONFIG_REISERFS_CHECK */ + + n = B_NR_ITEMS(tbSh); + + if ( tb->lnum[h] == 0 && tb->rnum[h] == 0 ) { + if ( tb->blknum[h] == 0 ) { + /* node S[h] (root of the tree) is empty now */ + struct buffer_head *new_root; + +#ifdef CONFIG_REISERFS_CHECK + if (n || B_FREE_SPACE (tbSh) != MAX_CHILD_SIZE(tbSh) - DC_SIZE) + reiserfs_panic (tb->tb_sb, "balance_internal_when_delete", "buffer must have only 0 keys (%d)", + n); + + if (bi.bi_parent) + reiserfs_panic (tb->tb_sb, "balance_internal_when_delete", "root has parent (%p)", bi.bi_parent); +#endif /* CONFIG_REISERFS_CHECK */ + + /* choose a new root */ + if ( ! tb->L[h-1] || ! B_NR_ITEMS(tb->L[h-1]) ) + new_root = tb->R[h-1]; + else + new_root = tb->L[h-1]; + /* switch super block's tree root block number to the new value */ + tb->tb_sb->u.reiserfs_sb.s_rs->s_root_block = cpu_to_le32 (new_root->b_blocknr); + //tb->tb_sb->u.reiserfs_sb.s_rs->s_tree_height --; + tb->tb_sb->u.reiserfs_sb.s_rs->s_tree_height = cpu_to_le16 (SB_TREE_HEIGHT (tb->tb_sb) - 1); + + do_balance_mark_sb_dirty (tb, tb->tb_sb->u.reiserfs_sb.s_sbh, 1); + /*&&&&&&&&&&&&&&&&&&&&&&*/ + if (h > 1) + /* use check_internal if new root is an internal node */ + check_internal (new_root); + /*&&&&&&&&&&&&&&&&&&&&&&*/ + tb->tb_sb->s_dirt = 1; + + /* do what is needed for buffer thrown from tree */ + reiserfs_invalidate_buffer(tb, tbSh); + return; + } + return; + } + + if ( tb->L[h] && tb->lnum[h] == -B_NR_ITEMS(tb->L[h]) - 1 ) { /* join S[h] with L[h] */ + +#ifdef CONFIG_REISERFS_CHECK + if ( tb->rnum[h] != 0 ) + reiserfs_panic (tb->tb_sb, "balance_internal_when_delete", "invalid tb->rnum[%d]==%d when joining S[h] with L[h]", + h, tb->rnum[h]); +#endif /* CONFIG_REISERFS_CHECK */ + + internal_shift_left (INTERNAL_SHIFT_FROM_S_TO_L, tb, h, n + 1); + reiserfs_invalidate_buffer(tb, tbSh); + + return; + } + + if ( tb->R[h] && tb->rnum[h] == -B_NR_ITEMS(tb->R[h]) - 1 ) { /* join S[h] with R[h] */ +#ifdef CONFIG_REISERFS_CHECK + if ( tb->lnum[h] != 0 ) + reiserfs_panic (tb->tb_sb, "balance_internal_when_delete", "invalid tb->lnum[%d]==%d when joining S[h] with R[h]", + h, tb->lnum[h]); +#endif /* CONFIG_REISERFS_CHECK */ + + internal_shift_right (INTERNAL_SHIFT_FROM_S_TO_R, tb, h, n + 1); + + reiserfs_invalidate_buffer(tb,tbSh); + return; + } + + if ( tb->lnum[h] < 0 ) { /* borrow from left neighbor L[h] */ +#ifdef CONFIG_REISERFS_CHECK + if ( tb->rnum[h] != 0 ) + reiserfs_panic (tb->tb_sb, "balance_internal_when_delete", "invalid tb->rnum[%d]==%d when borrow from L[h]", + h, tb->rnum[h]); +#endif /* CONFIG_REISERFS_CHECK */ + /*internal_shift_right (tb, h, tb->L[h], tb->CFL[h], tb->lkey[h], tb->S[h], -tb->lnum[h]);*/ + internal_shift_right (INTERNAL_SHIFT_FROM_L_TO_S, tb, h, -tb->lnum[h]); + return; + } + + if ( tb->rnum[h] < 0 ) { /* borrow from right neighbor R[h] */ +#ifdef CONFIG_REISERFS_CHECK + if ( tb->lnum[h] != 0 ) + reiserfs_panic (tb->tb_sb, "balance_internal_when_delete", "invalid tb->lnum[%d]==%d when borrow from R[h]", + h, tb->lnum[h]); +#endif /* CONFIG_REISERFS_CHECK */ + internal_shift_left (INTERNAL_SHIFT_FROM_R_TO_S, tb, h, -tb->rnum[h]);/*tb->S[h], tb->CFR[h], tb->rkey[h], tb->R[h], -tb->rnum[h]);*/ + return; + } + + if ( tb->lnum[h] > 0 ) { /* split S[h] into two parts and put them into neighbors */ +#ifdef CONFIG_REISERFS_CHECK + if ( tb->rnum[h] == 0 || tb->lnum[h] + tb->rnum[h] != n + 1 ) + reiserfs_panic (tb->tb_sb, "balance_internal_when_delete", + "invalid tb->lnum[%d]==%d or tb->rnum[%d]==%d when S[h](item number == %d) is split between them", + h, tb->lnum[h], h, tb->rnum[h], n); +#endif /* CONFIG_REISERFS_CHECK */ + + internal_shift_left (INTERNAL_SHIFT_FROM_S_TO_L, tb, h, tb->lnum[h]);/*tb->L[h], tb->CFL[h], tb->lkey[h], tb->S[h], tb->lnum[h]);*/ + internal_shift_right (INTERNAL_SHIFT_FROM_S_TO_R, tb, h, tb->rnum[h]); + + reiserfs_invalidate_buffer (tb, tbSh); + + return; + } + reiserfs_panic (tb->tb_sb, "balance_internal_when_delete", "unexpected tb->lnum[%d]==%d or tb->rnum[%d]==%d", + h, tb->lnum[h], h, tb->rnum[h]); +} + + +/* Replace delimiting key of buffers L[h] and S[h] by the given key.*/ +void replace_lkey ( + struct tree_balance * tb, + int h, + struct item_head * key + ) +{ +#ifdef CONFIG_REISERFS_CHECK + if (tb->L[h] == NULL || tb->CFL[h] == NULL) + reiserfs_panic (tb->tb_sb, "replace_lkey: 12255: " + "L[h](%p) and CFL[h](%p) must exist in replace_lkey", tb->L[h], tb->CFL[h]); +#endif + + if (B_NR_ITEMS(PATH_H_PBUFFER(tb->tb_path, h)) == 0) + return; + + memcpy (B_N_PDELIM_KEY(tb->CFL[h],tb->lkey[h]), key, KEY_SIZE); + + do_balance_mark_internal_dirty (tb, tb->CFL[h],0); +} + + +/* Replace delimiting key of buffers S[h] and R[h] by the given key.*/ +void replace_rkey ( + struct tree_balance * tb, + int h, + struct item_head * key + ) +{ +#ifdef CONFIG_REISERFS_CHECK + if (tb->R[h] == NULL || tb->CFR[h] == NULL) + reiserfs_panic (tb->tb_sb, "replace_rkey: 12260: " + "R[h](%p) and CFR[h](%p) must exist in replace_rkey", tb->R[h], tb->CFR[h]); + + if (B_NR_ITEMS(tb->R[h]) == 0) + reiserfs_panic (tb->tb_sb, "replace_rkey: 12265: " + "R[h] can not be empty if it exists (item number=%d)", B_NR_ITEMS(tb->R[h])); +#endif + + memcpy (B_N_PDELIM_KEY(tb->CFR[h],tb->rkey[h]), key, KEY_SIZE); + + do_balance_mark_internal_dirty (tb, tb->CFR[h], 0); +} + + +int balance_internal (struct tree_balance * tb, /* tree_balance structure */ + int h, /* level of the tree */ + int child_pos, + struct item_head * insert_key, /* key for insertion on higher level */ + struct buffer_head ** insert_ptr /* node for insertion on higher level*/ + ) + /* if inserting/pasting + { + child_pos is the position of the node-pointer in S[h] that * + pointed to S[h-1] before balancing of the h-1 level; * + this means that new pointers and items must be inserted AFTER * + child_pos + } + else + { + it is the position of the leftmost pointer that must be deleted (together with + its corresponding key to the left of the pointer) + as a result of the previous level's balancing. + } +*/ +{ + struct buffer_head * tbSh = PATH_H_PBUFFER (tb->tb_path, h); + struct buffer_info bi; + int order; /* we return this: it is 0 if there is no S[h], else it is tb->S[h]->b_item_order */ + int insert_num, n, k; + struct buffer_head * S_new; + struct item_head new_insert_key; + struct buffer_head * new_insert_ptr = NULL; + struct item_head * new_insert_key_addr = insert_key; + +#ifdef CONFIG_REISERFS_CHECK + if ( h < 1 ) + reiserfs_panic (tb->tb_sb, "balance_internal", "h (%d) can not be < 1 on internal level", h); +#endif /* CONFIG_REISERFS_CHECK */ + + order = ( tbSh ) ? PATH_H_POSITION (tb->tb_path, h + 1)/*tb->S[h]->b_item_order*/ : 0; + + /* Using insert_size[h] calculate the number insert_num of items + that must be inserted to or deleted from S[h]. */ + insert_num = tb->insert_size[h]/((int)(KEY_SIZE + DC_SIZE)); + + /* Check whether insert_num is proper **/ +#ifdef CONFIG_REISERFS_CHECK + if ( insert_num < -2 || insert_num > 2 ) + reiserfs_panic (tb->tb_sb, "balance_internal", + "incorrect number of items inserted to the internal node (%d)", insert_num); + + if ( h > 1 && (insert_num > 1 || insert_num < -1) ) + reiserfs_panic (tb->tb_sb, "balance_internal", + "incorrect number of items (%d) inserted to the internal node on a level (h=%d) higher than last internal level", + insert_num, h); +#endif /* CONFIG_REISERFS_CHECK */ + + /* Make balance in case insert_num < 0 */ + if ( insert_num < 0 ) { + balance_internal_when_delete (tb, h, child_pos); + return order; + } + + k = 0; + if ( tb->lnum[h] > 0 ) { + /* shift lnum[h] items from S[h] to the left neighbor L[h]. + check how many of new items fall into L[h] or CFL[h] after + shifting */ + n = B_NR_ITEMS (tb->L[h]); /* number of items in L[h] */ + if ( tb->lnum[h] <= child_pos ) { + /* new items don't fall into L[h] or CFL[h] */ + internal_shift_left (INTERNAL_SHIFT_FROM_S_TO_L, tb, h, tb->lnum[h]); + /*internal_shift_left (tb->L[h],tb->CFL[h],tb->lkey[h],tbSh,tb->lnum[h]);*/ + child_pos -= tb->lnum[h]; + } else if ( tb->lnum[h] > child_pos + insert_num ) { + /* all new items fall into L[h] */ + internal_shift_left (INTERNAL_SHIFT_FROM_S_TO_L, tb, h, tb->lnum[h] - insert_num); + /* internal_shift_left(tb->L[h],tb->CFL[h],tb->lkey[h],tbSh, + tb->lnum[h]-insert_num); + */ + /* insert insert_num keys and node-pointers into L[h] */ + bi.tb = tb; + bi.bi_bh = tb->L[h]; + bi.bi_parent = tb->FL[h]; + bi.bi_position = get_left_neighbor_position (tb, h); + internal_insert_childs (&bi,/*tb->L[h], tb->S[h-1]->b_next*/ n + child_pos + 1, + insert_num,insert_key,insert_ptr); + + insert_num = 0; + } else { + struct disk_child * dc; + + /* some items fall into L[h] or CFL[h], but some don't fall */ + internal_shift1_left(tb,h,child_pos+1); + /* calculate number of new items that fall into L[h] */ + k = tb->lnum[h] - child_pos - 1; + bi.tb = tb; + bi.bi_bh = tb->L[h]; + bi.bi_parent = tb->FL[h]; + bi.bi_position = get_left_neighbor_position (tb, h); + internal_insert_childs (&bi,/*tb->L[h], tb->S[h-1]->b_next,*/ n + child_pos + 1,k, + insert_key,insert_ptr); + + replace_lkey(tb,h,insert_key + k); + + /* replace the first node-ptr in S[h] by node-ptr to insert_ptr[k] */ + dc = B_N_CHILD(tbSh, 0); + dc->dc_size = cpu_to_le16 (MAX_CHILD_SIZE(insert_ptr[k]) - B_FREE_SPACE (insert_ptr[k])); + dc->dc_block_number = cpu_to_le32 (insert_ptr[k]->b_blocknr); + + do_balance_mark_internal_dirty (tb, tbSh, 0); + + k++; + insert_key += k; + insert_ptr += k; + insert_num -= k; + child_pos = 0; + } + } /* tb->lnum[h] > 0 */ + + if ( tb->rnum[h] > 0 ) { + /*shift rnum[h] items from S[h] to the right neighbor R[h]*/ + /* check how many of new items fall into R or CFR after shifting */ + n = B_NR_ITEMS (tbSh); /* number of items in S[h] */ + if ( n - tb->rnum[h] >= child_pos ) + /* new items fall into S[h] */ + /*internal_shift_right(tb,h,tbSh,tb->CFR[h],tb->rkey[h],tb->R[h],tb->rnum[h]);*/ + internal_shift_right (INTERNAL_SHIFT_FROM_S_TO_R, tb, h, tb->rnum[h]); + else + if ( n + insert_num - tb->rnum[h] < child_pos ) + { + /* all new items fall into R[h] */ + /*internal_shift_right(tb,h,tbSh,tb->CFR[h],tb->rkey[h],tb->R[h], + tb->rnum[h] - insert_num);*/ + internal_shift_right (INTERNAL_SHIFT_FROM_S_TO_R, tb, h, tb->rnum[h] - insert_num); + + /* insert insert_num keys and node-pointers into R[h] */ + bi.tb = tb; + bi.bi_bh = tb->R[h]; + bi.bi_parent = tb->FR[h]; + bi.bi_position = get_right_neighbor_position (tb, h); + internal_insert_childs (&bi, /*tb->R[h],tb->S[h-1]->b_next*/ child_pos - n - insert_num + tb->rnum[h] - 1, + insert_num,insert_key,insert_ptr); + insert_num = 0; + } + else + { + struct disk_child * dc; + + /* one of the items falls into CFR[h] */ + internal_shift1_right(tb,h,n - child_pos + 1); + /* calculate number of new items that fall into R[h] */ + k = tb->rnum[h] - n + child_pos - 1; + bi.tb = tb; + bi.bi_bh = tb->R[h]; + bi.bi_parent = tb->FR[h]; + bi.bi_position = get_right_neighbor_position (tb, h); + internal_insert_childs (&bi, /*tb->R[h], tb->R[h]->b_child,*/ 0, k, insert_key + 1, insert_ptr + 1); + + replace_rkey(tb,h,insert_key + insert_num - k - 1); + + /* replace the first node-ptr in R[h] by node-ptr insert_ptr[insert_num-k-1]*/ + dc = B_N_CHILD(tb->R[h], 0); + dc->dc_size = + cpu_to_le16 (MAX_CHILD_SIZE(insert_ptr[insert_num-k-1]) - + B_FREE_SPACE (insert_ptr[insert_num-k-1])); + dc->dc_block_number = cpu_to_le32 (insert_ptr[insert_num-k-1]->b_blocknr); + + do_balance_mark_internal_dirty (tb, tb->R[h],0); + + insert_num -= (k + 1); + } + } + + /** Fill new node that appears instead of S[h] **/ +#ifdef CONFIG_REISERFS_CHECK + if ( tb->blknum[h] > 2 ) + reiserfs_panic(0, "balance_internal", "blknum can not be > 2 for internal level"); + if ( tb->blknum[h] < 0 ) + reiserfs_panic(0, "balance_internal", "blknum can not be < 0"); +#endif /* CONFIG_REISERFS_CHECK */ + + if ( ! tb->blknum[h] ) + { /* node S[h] is empty now */ +#ifdef CONFIG_REISERFS_CHECK + if ( ! tbSh ) + reiserfs_panic(0,"balance_internal", "S[h] is equal NULL"); +#endif /* CONFIG_REISERFS_CHECK */ + + /* do what is needed for buffer thrown from tree */ + reiserfs_invalidate_buffer(tb,tbSh); + return order; + } + + if ( ! tbSh ) { + /* create new root */ + struct disk_child * dc; + struct buffer_head * tbSh_1 = PATH_H_PBUFFER (tb->tb_path, h - 1); + + + if ( tb->blknum[h] != 1 ) + reiserfs_panic(0, "balance_internal", "One new node required for creating the new root"); + /* S[h] = empty buffer from the list FEB. */ + tbSh = get_FEB (tb); + B_BLK_HEAD(tbSh)->blk_level = cpu_to_le16 (h + 1); + + /* Put the unique node-pointer to S[h] that points to S[h-1]. */ + + dc = B_N_CHILD(tbSh, 0); + dc->dc_block_number = cpu_to_le32 (tbSh_1->b_blocknr); + dc->dc_size = cpu_to_le16 (MAX_CHILD_SIZE (tbSh_1) - B_FREE_SPACE (tbSh_1)); + + tb->insert_size[h] -= DC_SIZE; + B_BLK_HEAD(tbSh)->blk_free_space = cpu_to_le16 (B_FREE_SPACE (tbSh) - DC_SIZE); + + do_balance_mark_internal_dirty (tb, tbSh, 0); + + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + check_internal (tbSh); + /*&&&&&&&&&&&&&&&&&&&&&&&&*/ + + /* put new root into path structure */ + PATH_OFFSET_PBUFFER(tb->tb_path, ILLEGAL_PATH_ELEMENT_OFFSET) = tbSh; + + /* Change root in structure super block. */ + tb->tb_sb->u.reiserfs_sb.s_rs->s_root_block = cpu_to_le32 (tbSh->b_blocknr); + tb->tb_sb->u.reiserfs_sb.s_rs->s_tree_height = cpu_to_le16 (SB_TREE_HEIGHT (tb->tb_sb) + 1); + do_balance_mark_sb_dirty (tb, tb->tb_sb->u.reiserfs_sb.s_sbh, 1); + tb->tb_sb->s_dirt = 1; + } + + if ( tb->blknum[h] == 2 ) { + int snum; + struct buffer_info dest_bi, src_bi; + + + /* S_new = free buffer from list FEB */ + S_new = get_FEB(tb); + + B_BLK_HEAD(S_new)->blk_level = cpu_to_le16 (h + 1); + + dest_bi.tb = tb; + dest_bi.bi_bh = S_new; + dest_bi.bi_parent = 0; + dest_bi.bi_position = 0; + src_bi.tb = tb; + src_bi.bi_bh = tbSh; + src_bi.bi_parent = PATH_H_PPARENT (tb->tb_path, h); + src_bi.bi_position = PATH_H_POSITION (tb->tb_path, h + 1); + + n = B_NR_ITEMS (tbSh); /* number of items in S[h] */ + snum = (insert_num + n + 1)/2; + if ( n - snum >= child_pos ) { + /* new items don't fall into S_new */ + /* store the delimiting key for the next level */ + /* new_insert_key = (n - snum)'th key in S[h] */ + memcpy (&new_insert_key,B_N_PDELIM_KEY(tbSh,n - snum), + KEY_SIZE); + /* last parameter is del_par */ + internal_move_pointers_items (&dest_bi, &src_bi, LAST_TO_FIRST, snum, 0); + /* internal_move_pointers_items(S_new, tbSh, LAST_TO_FIRST, snum, 0);*/ + } else if ( n + insert_num - snum < child_pos ) { + /* all new items fall into S_new */ + /* store the delimiting key for the next level */ + /* new_insert_key = (n + insert_item - snum)'th key in S[h] */ + memcpy(&new_insert_key,B_N_PDELIM_KEY(tbSh,n + insert_num - snum), + KEY_SIZE); + /* last parameter is del_par */ + internal_move_pointers_items (&dest_bi, &src_bi, LAST_TO_FIRST, snum - insert_num, 0); + /* internal_move_pointers_items(S_new,tbSh,1,snum - insert_num,0);*/ + + /* insert insert_num keys and node-pointers into S_new */ + internal_insert_childs (&dest_bi, /*S_new,tb->S[h-1]->b_next,*/child_pos - n - insert_num + snum - 1, + insert_num,insert_key,insert_ptr); + + insert_num = 0; + } else { + struct disk_child * dc; + + /* some items fall into S_new, but some don't fall */ + /* last parameter is del_par */ + internal_move_pointers_items (&dest_bi, &src_bi, LAST_TO_FIRST, n - child_pos + 1, 1); + /* internal_move_pointers_items(S_new,tbSh,1,n - child_pos + 1,1);*/ + /* calculate number of new items that fall into S_new */ + k = snum - n + child_pos - 1; + + internal_insert_childs (&dest_bi, /*S_new,*/ 0, k, insert_key + 1, insert_ptr+1); + + /* new_insert_key = insert_key[insert_num - k - 1] */ + memcpy(&new_insert_key,insert_key + insert_num - k - 1, + KEY_SIZE); + /* replace first node-ptr in S_new by node-ptr to insert_ptr[insert_num-k-1] */ + + dc = B_N_CHILD(S_new,0); + dc->dc_size = cpu_to_le16 (MAX_CHILD_SIZE(insert_ptr[insert_num-k-1]) - + B_FREE_SPACE(insert_ptr[insert_num-k-1])); + dc->dc_block_number = cpu_to_le32 (insert_ptr[insert_num-k-1]->b_blocknr); + + do_balance_mark_internal_dirty (tb, S_new,0); + + insert_num -= (k + 1); + } + /* new_insert_ptr = node_pointer to S_new */ + new_insert_ptr = S_new; + +#ifdef CONFIG_REISERFS_CHECK + if ( buffer_locked(S_new) || atomic_read (&(S_new->b_count)) != 1) + if (buffer_locked(S_new) || atomic_read(&(S_new->b_count)) > 2 || + !(buffer_journaled(S_new) || buffer_journal_dirty(S_new))) { + reiserfs_panic (tb->tb_sb, "cm-00001: balance_internal: bad S_new (%b)", S_new); + } +#endif /* CONFIG_REISERFS_CHECK */ + + // S_new is released in unfix_nodes + } + + n = B_NR_ITEMS (tbSh); /*number of items in S[h] */ + +#ifdef REISERFS_FSCK + if ( -1 <= child_pos && child_pos <= n && insert_num > 0 ) { +#else + if ( 0 <= child_pos && child_pos <= n && insert_num > 0 ) { +#endif + bi.tb = tb; + bi.bi_bh = tbSh; + bi.bi_parent = PATH_H_PPARENT (tb->tb_path, h); + bi.bi_position = PATH_H_POSITION (tb->tb_path, h + 1); +#ifdef REISERFS_FSCK + if (child_pos == -1) { + /* this is a little different from original do_balance: + here we insert the minimal keys in the tree, that has never happened when file system works */ + if (tb->CFL[h-1] || insert_num != 1 || h != 1) + die ("balance_internal: invalid child_pos"); +/* insert_child (tb->S[h], tb->S[h-1], child_pos, insert_num, B_N_ITEM_HEAD(tb->S[0],0), insert_ptr);*/ + internal_insert_childs (&bi, child_pos, insert_num, B_N_PITEM_HEAD (PATH_PLAST_BUFFER (tb->tb_path), 0), insert_ptr); + } else +#endif + internal_insert_childs ( + &bi,/*tbSh,*/ + /* ( tb->S[h-1]->b_parent == tb->S[h] ) ? tb->S[h-1]->b_next : tb->S[h]->b_child->b_next,*/ + child_pos,insert_num,insert_key,insert_ptr + ); + } + + + memcpy (new_insert_key_addr,&new_insert_key,KEY_SIZE); + insert_ptr[0] = new_insert_ptr; + + return order; + } + + + diff --git a/fs/reiserfs/inode.c b/fs/reiserfs/inode.c new file mode 100644 index 000000000..12ab13894 --- /dev/null +++ b/fs/reiserfs/inode.c @@ -0,0 +1,1881 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/sched.h> +#include <linux/reiserfs_fs.h> +#include <linux/locks.h> +#include <linux/smp_lock.h> +#include <asm/uaccess.h> + +#else + +#include "nokernel.h" + +#endif + +/* args for the create parameter of reiserfs_get_block */ +#define GET_BLOCK_NO_CREATE 0 /* don't create new blocks or convert tails */ +#define GET_BLOCK_CREATE 1 /* add anything you need to find block */ +#define GET_BLOCK_NO_HOLE 2 /* return -ENOENT for file holes */ +#define GET_BLOCK_READ_DIRECT 4 /* read the tail if indirect item not found */ + +// +// initially this function was derived from minix or ext2's analog and +// evolved as the prototype did +// +void reiserfs_delete_inode (struct inode * inode) +{ + int jbegin_count = JOURNAL_PER_BALANCE_CNT * 2; + int windex ; + struct reiserfs_transaction_handle th ; + + + lock_kernel() ; + + /* The = 0 happens when we abort creating a new inode for some reason like lack of space.. */ + if (INODE_PKEY(inode)->k_objectid != 0) { /* also handles bad_inode case */ + down (&inode->i_sem); + + journal_begin(&th, inode->i_sb, jbegin_count) ; + windex = push_journal_writer("delete_inode") ; + + reiserfs_delete_object (&th, inode); + reiserfs_remove_page_from_flush_list(&th, inode) ; + pop_journal_writer(windex) ; + reiserfs_release_objectid (&th, inode->i_ino); + + journal_end(&th, inode->i_sb, jbegin_count) ; + + up (&inode->i_sem); + } else { + /* no object items are in the tree */ + ; + } + clear_inode (inode); /* note this must go after the journal_end to prevent deadlock */ + unlock_kernel() ; +} + +static void _make_cpu_key (struct cpu_key * key, int version, __u32 dirid, __u32 objectid, + loff_t offset, int type, int length) +{ + key->version = version; + + key->on_disk_key.k_dir_id = dirid; + key->on_disk_key.k_objectid = objectid; + set_cpu_key_k_offset (key, offset); + set_cpu_key_k_type (key, type); + key->key_length = length; +} + + +/* take base of inode_key (it comes from inode always) (dirid, objectid) and version from an inode, set + offset and type of key */ +void make_cpu_key (struct cpu_key * key, const struct inode * inode, loff_t offset, + int type, int length) +{ + _make_cpu_key (key, inode_items_version (inode), le32_to_cpu (INODE_PKEY (inode)->k_dir_id), + le32_to_cpu (INODE_PKEY (inode)->k_objectid), + offset, type, length); +} + + +// +// when key is 0, do not set version and short key +// +inline void make_le_item_head (struct item_head * ih, struct cpu_key * key, int version, + loff_t offset, int type, int length, int entry_count/*or ih_free_space*/) +{ + if (key) { + ih->ih_key.k_dir_id = cpu_to_le32 (key->on_disk_key.k_dir_id); + ih->ih_key.k_objectid = cpu_to_le32 (key->on_disk_key.k_objectid); + } + ih->ih_version = cpu_to_le16 (version); + set_le_ih_k_offset (ih, offset); + set_le_ih_k_type (ih, type); + ih->ih_item_len = cpu_to_le16 (length); + /* set_ih_free_space (ih, 0);*/ + // for directory items it is entry count, for directs and stat + // datas - 0xffff, for indirects - 0 + ih->u.ih_entry_count = cpu_to_le16 (entry_count); +} + + +// +// FIXME: we might cache recently accessed indirect item (or at least +// first 15 pointers just like ext2 does + +// Ugh. Not too eager for that.... +// I cut the code until such time as I see a convincing argument (benchmark). +// I don't want a bloated inode struct..., and I don't like code complexity.... + +/* cutting the code is fine, since it really isn't in use yet and is easy +** to add back in. But, Vladimir has a really good idea here. Think +** about what happens for reading a file. For each page, +** The VFS layer calls reiserfs_readpage, who searches the tree to find +** an indirect item. This indirect item has X number of pointers, where +** X is a big number if we've done the block allocation right. But, +** we only use one or two of these pointers during each call to readpage, +** needlessly researching again later on. +** +** The size of the cache could be dynamic based on the size of the file. +** +** I'd also like to see us cache the location the stat data item, since +** we are needlessly researching for that frequently. +** +** --chris +*/ + +/* people who call journal_begin with a page locked must call this +** BEFORE calling journal_begin +*/ +static int prevent_flush_page_lock(struct page *page, + struct inode *inode) { + struct reiserfs_page_list *pl ; + struct super_block *s = inode->i_sb ; + /* we don't care if the inode has a stale pointer from an old + ** transaction + */ + if(!page || inode->u.reiserfs_i.i_conversion_trans_id != SB_JOURNAL(s)->j_trans_id) { + return 0 ; + } + pl = inode->u.reiserfs_i.i_converted_page ; + if (pl && pl->page == page) { + pl->do_not_lock = 1 ; + } + /* this last part is really important. The address space operations have + ** the page locked before they call the journal functions. So it is possible + ** for one process to be waiting in flush_pages_before_commit for a + ** page, then for the process with the page locked to call journal_begin. + ** + ** We'll deadlock because the process flushing pages will never notice + ** the process with the page locked has called prevent_flush_page_lock. + ** So, we wake up the page waiters, even though the page is still locked. + ** The process waiting in flush_pages_before_commit must check the + ** pl->do_not_lock flag, and stop trying to lock the page. + */ + wake_up(&page->wait) ; + return 0 ; + +} +/* people who call journal_end with a page locked must call this +** AFTER calling journal_end +*/ +static int allow_flush_page_lock(struct page *page, + struct inode *inode) { + + struct reiserfs_page_list *pl ; + struct super_block *s = inode->i_sb ; + /* we don't care if the inode has a stale pointer from an old + ** transaction + */ + if(!page || inode->u.reiserfs_i.i_conversion_trans_id != SB_JOURNAL(s)->j_trans_id) { + return 0 ; + } + pl = inode->u.reiserfs_i.i_converted_page ; + if (pl && pl->page == page) { + pl->do_not_lock = 0 ; + } + return 0 ; + +} + +/* If this page has a file tail in it, and +** it was read in by get_block_create_0, the page data is valid, +** but tail is still sitting in a direct item, and we can't write to +** it. So, look through this page, and check all the mapped buffers +** to make sure they have valid block numbers. Any that don't need +** to be unmapped, so that block_prepare_write will correctly call +** reiserfs_get_block to convert the tail into an unformatted node +*/ +static inline void fix_tail_page_for_writing(struct page *page) { + struct buffer_head *head, *next, *bh ; + + if (page && page->buffers) { + head = page->buffers ; + bh = head ; + do { + next = bh->b_this_page ; + if (buffer_mapped(bh) && bh->b_blocknr == 0) { + reiserfs_unmap_buffer(bh) ; + } + bh = next ; + } while (bh != head) ; + } +} + + + + +/* we need to allocate a block for new unformatted node. Try to figure out + what point in bitmap reiserfs_new_blocknrs should start from. */ +static b_blocknr_t find_tag (struct buffer_head * bh, struct item_head * ih, + __u32 * item, int pos_in_item) +{ + if (!is_indirect_le_ih (ih)) + /* something more complicated could be here */ + return bh->b_blocknr; + + /* for indirect item: go to left and look for the first non-hole entry in + the indirect item */ + if (pos_in_item == I_UNFM_NUM (ih)) + pos_in_item --; + while (pos_in_item >= 0) { + if (item [pos_in_item]) + return item [pos_in_item]; + pos_in_item --; + } + return bh->b_blocknr; +} + + +/* reiserfs_get_block does not need to allocate a block only if it has been + done already or non-hole position has been found in the indirect item */ +static inline int allocation_needed (int retval, b_blocknr_t allocated, + struct item_head * ih, + __u32 * item, int pos_in_item) +{ + if (allocated) + return 0; + if (retval == POSITION_FOUND && is_indirect_le_ih (ih) && item[pos_in_item]) + return 0; + return 1; +} + +static inline int indirect_item_found (int retval, struct item_head * ih) +{ + return (retval == POSITION_FOUND) && is_indirect_le_ih (ih); +} + + +static inline void set_block_dev_mapped (struct buffer_head * bh, + b_blocknr_t block, struct inode * inode) +{ + bh->b_dev = inode->i_dev; + bh->b_blocknr = block; + bh->b_state |= (1UL << BH_Mapped); +} + + +// +// files which were created in the earlier version can not be longer, +// than 2 gb +// +int file_capable (struct inode * inode, long block) +{ + if (inode_items_version (inode) != ITEM_VERSION_1 || // it is new file. + block < (1 << (31 - inode->i_sb->s_blocksize_bits))) // old file, but 'block' is inside of 2gb + return 1; + + return 0; +} + +/*static*/ void restart_transaction(struct reiserfs_transaction_handle *th, + struct inode *inode, struct path *path) { + struct super_block *s = th->t_super ; + int len = th->t_blocks_allocated ; + + pathrelse(path) ; + reiserfs_update_sd(th, inode) ; + journal_end(th, s, len) ; + journal_begin(th, s, len) ; +} + +// it is called by get_block when create == 0. Returns block number +// for 'block'-th logical block of file. When it hits direct item it +// returns 0 (being called from bmap) or read direct item into piece +// of page (bh_result) + +// Please improve the english/clarity in the comment above, as it is +// hard to understand. + +static int _get_block_create_0 (struct inode * inode, long block, + struct buffer_head * bh_result, + int args) +{ + INITIALIZE_PATH (path); + struct cpu_key key; + struct buffer_head * bh; + struct item_head * ih, tmp_ih; + int fs_gen ; + int blocknr; + char * p = NULL; + int chars; + int ret ; + unsigned long offset ; + + // prepare the key to look for the 'block'-th block of file + make_cpu_key (&key, inode, + (loff_t)block * inode->i_sb->s_blocksize + 1, TYPE_ANY, 3); + +research: + if (search_for_position_by_key (inode->i_sb, &key, &path) != POSITION_FOUND) { + pathrelse (&path); + if (p) + kunmap(bh_result->b_page) ; + if ((args & GET_BLOCK_NO_HOLE)) { + return -ENOENT ; + } + return 0 ; + } + + // + bh = get_bh (&path); + ih = get_ih (&path); + if (is_indirect_le_ih (ih)) { + __u32 * ind_item = (__u32 *)B_I_PITEM (bh, ih); + + /* FIXME: here we could cache indirect item or part of it in + the inode to avoid search_by_key in case of subsequent + access to file */ + blocknr = le32_to_cpu (ind_item [path.pos_in_item]); + ret = 0 ; + if (blocknr) { + bh_result->b_dev = inode->i_dev; + bh_result->b_blocknr = blocknr; + bh_result->b_state |= (1UL << BH_Mapped); + } else if ((args & GET_BLOCK_NO_HOLE)) { + ret = -ENOENT ; + } + pathrelse (&path); + if (p) + kunmap(bh_result->b_page) ; + return ret ; + } + + // requested data are in direct item(s) + if (!(args & GET_BLOCK_READ_DIRECT)) { + // we are called by bmap. FIXME: we can not map block of file + // when it is stored in direct item(s) + pathrelse (&path); + if (p) + kunmap(bh_result->b_page) ; + return -ENOENT; + } + + // read file tail into part of page + offset = (cpu_key_k_offset(&key) - 1) & (PAGE_CACHE_SIZE - 1) ; + fs_gen = get_generation(inode->i_sb) ; + copy_item_head (&tmp_ih, ih); + + /* we only want to kmap if we are reading the tail into the page. + ** this is not the common case, so we don't kmap until we are + ** sure we need to. But, this means the item might move if + ** kmap schedules + */ + p = (char *)kmap(bh_result->b_page) ; + if (fs_changed (fs_gen, inode->i_sb) && item_moved (&tmp_ih, &path)) { + goto research; + } + p += offset ; + memset (p, 0, inode->i_sb->s_blocksize); + do { + if (!is_direct_le_ih (ih)) { + BUG (); + } + chars = le16_to_cpu (ih->ih_item_len) - path.pos_in_item; + memcpy (p, B_I_PITEM (bh, ih) + path.pos_in_item, chars); + p += chars; + + if (PATH_LAST_POSITION (&path) != (B_NR_ITEMS (bh) - 1)) + // we done, if read direct item is not the last item of + // node FIXME: we could try to check right delimiting key + // to see whether direct item continues in the right + // neighbor or rely on i_size + break; + + // update key to look for the next piece + set_cpu_key_k_offset (&key, cpu_key_k_offset (&key) + chars); + if (search_for_position_by_key (inode->i_sb, &key, &path) != POSITION_FOUND) + // we read something from tail, even if now we got IO_ERROR + break; + bh = get_bh (&path); + ih = get_ih (&path); + } while (1); + + pathrelse (&path); + + // FIXME: b_blocknr == 0 here. but b_data contains correct data + // from tail. ll_rw_block will skip uptodate buffers + bh_result->b_blocknr = 0 ; + bh_result->b_dev = inode->i_dev; + mark_buffer_uptodate (bh_result, 1); + bh_result->b_state |= (1UL << BH_Mapped); + kunmap(bh_result->b_page) ; + + return 0; +} + + +// this is called to create file map. So, _get_block_create_0 will not +// read direct item +int reiserfs_bmap (struct inode * inode, long block, + struct buffer_head * bh_result, int create) +{ + if (!file_capable (inode, block)) + return -EFBIG; + + lock_kernel() ; + /* do not read the direct item */ + _get_block_create_0 (inode, block, bh_result, 0) ; + unlock_kernel() ; + return 0; +} + +/* special version of get_block that is only used by grab_tail_page right +** now. It is sent to block_prepare_write, and when you try to get a +** block past the end of the file (or a block from a hole) it returns +** -ENOENT instead of a valid buffer. block_prepare_write expects to +** be able to do i/o on the buffers returned, unless an error value +** is also returned. +** +** So, this allows block_prepare_write to be used for reading a single block +** in a page. Where it does not produce a valid page for holes, or past the +** end of the file. This turns out to be exactly what we need for reading +** tails for conversion. +** +** The point of the wrapper is forcing a certain value for create, even +** though the VFS layer is calling this function with create==1. If you +** don't want to send create == GET_BLOCK_NO_HOLE to reiserfs_get_block, +** don't use this function. +*/ +static int reiserfs_get_block_create_0 (struct inode * inode, long block, + struct buffer_head * bh_result, int create) { + return reiserfs_get_block(inode, block, bh_result, GET_BLOCK_NO_HOLE) ; +} + +/* +** helper function for when reiserfs_get_block is called for a hole +** but the file tail is still in a direct item +** bh_result is the buffer head for the hole +** tail_offset is the offset of the start of the tail in the file +** +** This calls prepare_write, which will start a new transaction +** you should not be in a transaction, or have any paths held when you +** call this. +*/ +static int convert_tail_for_hole(struct inode *inode, + struct buffer_head *bh_result, + loff_t tail_offset) { + unsigned long index ; + unsigned long tail_end ; + unsigned long tail_start ; + struct page * tail_page ; + struct page * hole_page = bh_result->b_page ; + int retval = 0 ; + + if ((tail_offset & (bh_result->b_size - 1)) != 1) + return -EIO ; + + /* always try to read until the end of the block */ + tail_start = tail_offset & (PAGE_CACHE_SIZE - 1) ; + tail_end = (tail_start | (bh_result->b_size - 1)) + 1 ; + + index = tail_offset >> PAGE_CACHE_SHIFT ; + if (index != hole_page->index) { + tail_page = grab_cache_page(inode->i_mapping, index) ; + retval = PTR_ERR(tail_page) ; + if (IS_ERR(tail_page)) { + goto out ; + } + } else { + tail_page = hole_page ; + } + + /* we don't have to make sure the conversion did not happen while + ** we were locking the page because anyone that could convert + ** must first take i_sem. + ** + ** We must fix the tail page for writing because it might have buffers + ** that are mapped, but have a block number of 0. This indicates tail + ** data that has been read directly into the page, and block_prepare_write + ** won't trigger a get_block in this case. + */ + fix_tail_page_for_writing(tail_page) ; + retval = block_prepare_write(tail_page, tail_start, tail_end, + reiserfs_get_block) ; + if (retval) + goto unlock ; + + /* tail conversion might change the data in the page */ + flush_dcache_page(tail_page) ; + + retval = generic_commit_write(NULL, tail_page, tail_start, tail_end) ; + +unlock: + if (tail_page != hole_page) { + UnlockPage(tail_page) ; + page_cache_release(tail_page) ; + } +out: + return retval ; +} + +// +// initially this function was derived from ext2's analog and evolved +// as the prototype did. You'll need to look at the ext2 version to +// determine which parts are derivative, if any, understanding that +// there are only so many ways to code to a given interface. +// +int reiserfs_get_block (struct inode * inode, long block, + struct buffer_head * bh_result, int create) +{ + int repeat, retval; + unsigned long tag; + b_blocknr_t allocated_block_nr = 0;// b_blocknr_t is unsigned long + INITIALIZE_PATH(path); + int pos_in_item; + struct cpu_key key; + struct buffer_head * bh, * unbh = 0; + struct item_head * ih, tmp_ih; + __u32 * item; + int done; + int fs_gen; + int windex ; + struct reiserfs_transaction_handle th ; + int jbegin_count = JOURNAL_PER_BALANCE_CNT * 3 ; + int version; + int transaction_started = 0 ; + loff_t new_offset = (block << inode->i_sb->s_blocksize_bits) + 1 ; + + /* bad.... */ + lock_kernel() ; + th.t_trans_id = 0 ; + version = inode_items_version (inode); + + if (!file_capable (inode, block)) { + unlock_kernel() ; + return -EFBIG; + } + + /* if !create, we aren't changing the FS, so we don't need to + ** log anything, so we don't need to start a transaction + */ + if (!(create & GET_BLOCK_CREATE)) { + int ret ; + /* find number of block-th logical block of the file */ + ret = _get_block_create_0 (inode, block, bh_result, + create | GET_BLOCK_READ_DIRECT) ; + unlock_kernel() ; + return ret; + } + + if (block < 0) { + unlock_kernel(); + return -EIO; + } + + prevent_flush_page_lock(bh_result->b_page, inode) ; + inode->u.reiserfs_i.i_pack_on_close = 1 ; + + windex = push_journal_writer("reiserfs_get_block") ; + + /* set the key of the first byte in the 'block'-th block of file */ + make_cpu_key (&key, inode, + (loff_t)block * inode->i_sb->s_blocksize + 1, // k_offset + TYPE_ANY, 3/*key length*/); + if ((new_offset + inode->i_sb->s_blocksize) >= inode->i_size) { + journal_begin(&th, inode->i_sb, jbegin_count) ; + transaction_started = 1 ; + } + research: + + retval = search_for_position_by_key (inode->i_sb, &key, &path); + if (retval == IO_ERROR) { + retval = -EIO; + goto failure; + } + + bh = get_bh (&path); + ih = get_ih (&path); + item = get_item (&path); + pos_in_item = path.pos_in_item; + + fs_gen = get_generation (inode->i_sb); + copy_item_head (&tmp_ih, ih); + + if (allocation_needed (retval, allocated_block_nr, ih, item, pos_in_item)) { + /* we have to allocate block for the unformatted node */ + tag = find_tag (bh, ih, item, pos_in_item); + if (!transaction_started) { + pathrelse(&path) ; + journal_begin(&th, inode->i_sb, jbegin_count) ; + transaction_started = 1 ; + goto research ; + } + +#ifdef REISERFS_PREALLOCATE + repeat = reiserfs_new_unf_blocknrs2 (&th, inode, &allocated_block_nr, tag); +#else + repeat = reiserfs_new_unf_blocknrs (&th, &allocated_block_nr, tag); +#endif + + if (repeat == NO_DISK_SPACE) { + /* restart the transaction to give the journal a chance to free + ** some blocks. releases the path, so we have to go back to + ** research if we succeed on the second try + */ + restart_transaction(&th, inode, &path) ; +#ifdef REISERFS_PREALLOCATE + repeat = reiserfs_new_unf_blocknrs2 (&th, inode, &allocated_block_nr, tag); +#else + repeat = reiserfs_new_unf_blocknrs (&th, &allocated_block_nr, tag); +#endif + + if (repeat != NO_DISK_SPACE) { + goto research ; + } + retval = -ENOSPC; + goto failure; + } + + if (fs_changed (fs_gen, inode->i_sb) && item_moved (&tmp_ih, &path)) { + goto research; + } + } + + if (indirect_item_found (retval, ih)) { + /* 'block'-th block is in the file already (there is + corresponding cell in some indirect item). But it may be + zero unformatted node pointer (hole) */ + if (!item[pos_in_item]) { + /* use allocated block to plug the hole */ + reiserfs_prepare_for_journal(inode->i_sb, bh, 1) ; + if (fs_changed (fs_gen, inode->i_sb) && item_moved (&tmp_ih, &path)) { + reiserfs_restore_prepared_buffer(inode->i_sb, bh) ; + goto research; + } + bh_result->b_state |= (1UL << BH_New); + item[pos_in_item] = cpu_to_le32 (allocated_block_nr); + journal_mark_dirty (&th, inode->i_sb, bh); + inode->i_blocks += (inode->i_sb->s_blocksize / 512) ; + reiserfs_update_sd(&th, inode) ; + } + set_block_dev_mapped(bh_result, le32_to_cpu (item[pos_in_item]), inode); + pathrelse (&path); +#ifdef REISERFS_CHECK + pop_journal_writer(windex) ; +#endif /* REISERFS_CHECK */ + if (transaction_started) + journal_end(&th, inode->i_sb, jbegin_count) ; + + allow_flush_page_lock(bh_result->b_page, inode) ; + unlock_kernel() ; + + /* the item was found, so new blocks were not added to the file + ** there is no need to make sure the inode is updated with this + ** transaction + */ + return 0; + } + + if (!transaction_started) { + /* if we don't pathrelse, we could vs-3050 on the buffer if + ** someone is waiting for it (they can't finish until the buffer + ** is released, we can start a new transaction until they finish) + */ + pathrelse(&path) ; + journal_begin(&th, inode->i_sb, jbegin_count) ; + transaction_started = 1 ; + goto research; + } + + /* desired position is not found or is in the direct item. We have + to append file with holes up to 'block'-th block converting + direct items to indirect one if necessary */ + done = 0; + do { + if (is_statdata_le_ih (ih)) { + __u32 unp = 0; + struct cpu_key tmp_key; + + /* indirect item has to be inserted */ + make_le_item_head (&tmp_ih, &key, version, 1, TYPE_INDIRECT, + UNFM_P_SIZE, 0/* free_space */); + + if (cpu_key_k_offset (&key) == 1) { + /* we are going to add 'block'-th block to the file. Use + allocated block for that */ + unp = cpu_to_le32 (allocated_block_nr); + set_block_dev_mapped (bh_result, allocated_block_nr, inode); + bh_result->b_state |= (1UL << BH_New); + done = 1; + } + tmp_key = key; // ;) + set_cpu_key_k_offset (&tmp_key, 1); + PATH_LAST_POSITION(&path) ++; + + retval = reiserfs_insert_item (&th, &path, &tmp_key, &tmp_ih, (char *)&unp); + if (retval) { + reiserfs_free_block (&th, allocated_block_nr); + +#ifdef REISERFS_PREALLOCATE + reiserfs_discard_prealloc (&th, inode); +#endif + goto failure; // retval == -ENOSPC or -EIO or -EEXIST + } + if (unp) + inode->i_blocks += inode->i_sb->s_blocksize / 512; + //mark_tail_converted (inode); + } else if (is_direct_le_ih (ih)) { + /* direct item has to be converted */ + loff_t tail_offset; + + tail_offset = ((le_ih_k_offset (ih) - 1) & ~(inode->i_sb->s_blocksize - 1)) + 1; + if (tail_offset == cpu_key_k_offset (&key)) { + /* direct item we just found fits into block we have + to map. Convert it into unformatted node: use + bh_result for the conversion */ + set_block_dev_mapped (bh_result, allocated_block_nr, inode); + unbh = bh_result; + done = 1; + } else { + /* we have to padd file tail stored in direct item(s) + up to block size and convert it to unformatted + node. FIXME: this should also get into page cache */ + + pathrelse(&path) ; + journal_end(&th, inode->i_sb, jbegin_count) ; + transaction_started = 0 ; + + retval = convert_tail_for_hole(inode, bh_result, tail_offset) ; + if (retval) { + printk("clm-6004: convert tail failed inode %lu, error %d\n", inode->i_ino, retval) ; + if (allocated_block_nr) + reiserfs_free_block (&th, allocated_block_nr); + goto failure ; + } + goto research ; + } + retval = direct2indirect (&th, inode, &path, unbh, tail_offset); + /* it is important the mark_buffer_uptodate is done after + ** the direct2indirect. The buffer might contain valid + ** data newer than the data on disk (read by readpage, changed, + ** and then sent here by writepage). direct2indirect needs + ** to know if unbh was already up to date, so it can decide + ** if the data in unbh needs to be replaced with data from + ** the disk + */ + mark_buffer_uptodate (unbh, 1); + if (retval) { + reiserfs_free_block (&th, allocated_block_nr); + +#ifdef REISERFS_PREALLOCATE + reiserfs_discard_prealloc (&th, inode); +#endif + goto failure; + } + /* we've converted the tail, so we must + ** flush unbh before the transaction commits + */ + reiserfs_add_page_to_flush_list(&th, inode, unbh) ; + + //inode->i_blocks += inode->i_sb->s_blocksize / 512; + //mark_tail_converted (inode); + } else { + /* append indirect item with holes if needed, when appending + pointer to 'block'-th block use block, which is already + allocated */ + struct cpu_key tmp_key; + struct unfm_nodeinfo un = {0, 0}; + +#ifdef CONFIG_REISERFS_CHECK + if (pos_in_item != le16_to_cpu (ih->ih_item_len) / UNFM_P_SIZE) + reiserfs_panic (inode->i_sb, "vs-: reiserfs_get_block: " + "invalid position for append"); +#endif + /* indirect item has to be appended, set up key of that position */ + make_cpu_key (&tmp_key, inode, + le_key_k_offset (version, &(ih->ih_key)) + op_bytes_number (ih, inode->i_sb->s_blocksize), + //pos_in_item * inode->i_sb->s_blocksize, + TYPE_INDIRECT, 3);// key type is unimportant + + if (cpu_key_k_offset (&tmp_key) == cpu_key_k_offset (&key)) { + /* we are going to add target block to the file. Use allocated + block for that */ + un.unfm_nodenum = cpu_to_le32 (allocated_block_nr); + set_block_dev_mapped (bh_result, allocated_block_nr, inode); + bh_result->b_state |= (1UL << BH_New); + done = 1; + } else { + /* paste hole to the indirect item */ + } + retval = reiserfs_paste_into_item (&th, &path, &tmp_key, (char *)&un, UNFM_P_SIZE); + if (retval) { + reiserfs_free_block (&th, allocated_block_nr); + +#ifdef REISERFS_PREALLOCATE + reiserfs_discard_prealloc (&th, inode); +#endif + goto failure; + } + if (un.unfm_nodenum) + inode->i_blocks += inode->i_sb->s_blocksize / 512; + //mark_tail_converted (inode); + } + + if (done == 1) + break; + + /* this loop could log more blocks than we had originally asked + ** for. So, we have to allow the transaction to end if it is + ** too big or too full. Update the inode so things are + ** consistent if we crash before the function returns + ** + ** release the path so that anybody waiting on the path before + ** ending their transaction will be able to continue. + */ + if (journal_transaction_should_end(&th, th.t_blocks_allocated)) { + restart_transaction(&th, inode, &path) ; + } + /* inserting indirect pointers for a hole can take a + ** long time. reschedule if needed + */ + if (current->need_resched) + schedule() ; + + retval = search_for_position_by_key (inode->i_sb, &key, &path); + if (retval == IO_ERROR) { + retval = -EIO; + goto failure; + } + if (retval == POSITION_FOUND) { + reiserfs_warning ("vs-: reiserfs_get_block: " + "%k should not be found", &key); + retval = -EEXIST; + pathrelse(&path) ; + goto failure; + } + bh = get_bh (&path); + ih = get_ih (&path); + item = get_item (&path); + pos_in_item = path.pos_in_item; + } while (1); + + + retval = 0; + reiserfs_check_path(&path) ; + + failure: + if (transaction_started) { + reiserfs_update_sd(&th, inode) ; + journal_end(&th, inode->i_sb, jbegin_count) ; + } + pop_journal_writer(windex) ; + allow_flush_page_lock(bh_result->b_page, inode) ; + unlock_kernel() ; + reiserfs_check_path(&path) ; + return retval; +} + + +// +// BAD: new directories have stat data of new type and all other items +// of old type. Version stored in the inode says about body items, so +// in update_stat_data we can not rely on inode, but have to check +// item version directly +// + +// called by read_inode +static void init_inode (struct inode * inode, struct path * path) +{ + struct buffer_head * bh; + struct item_head * ih; + __u32 rdev; + //int version = ITEM_VERSION_1; + + bh = PATH_PLAST_BUFFER (path); + ih = PATH_PITEM_HEAD (path); + + + copy_key (INODE_PKEY (inode), &(ih->ih_key)); + inode->i_generation = INODE_PKEY (inode)->k_dir_id; + inode->i_blksize = PAGE_SIZE; + + if (stat_data_v1 (ih)) { + struct stat_data_v1 * sd = (struct stat_data_v1 *)B_I_PITEM (bh, ih); + unsigned long blocks; + + inode_items_version (inode) = ITEM_VERSION_1; + inode->i_mode = le16_to_cpu (sd->sd_mode); + inode->i_nlink = le16_to_cpu (sd->sd_nlink); + inode->i_uid = le16_to_cpu (sd->sd_uid); + inode->i_gid = le16_to_cpu (sd->sd_gid); + inode->i_size = le32_to_cpu (sd->sd_size); + inode->i_atime = le32_to_cpu (sd->sd_atime); + inode->i_mtime = le32_to_cpu (sd->sd_mtime); + inode->i_ctime = le32_to_cpu (sd->sd_ctime); + + inode->i_blocks = le32_to_cpu (sd->u.sd_blocks); + blocks = (inode->i_size + 511) >> 9; + blocks = _ROUND_UP (blocks, inode->i_blksize >> 9); + if (inode->i_blocks > blocks) { + // there was a bug in <=3.5.23 when i_blocks could take negative + // values. Starting from 3.5.17 this value could even be stored in + // stat data. For such files we set i_blocks based on file + // size. Just 2 notes: this can be wrong for sparce files. On-disk value will be + // only updated if file's inode will ever change + inode->i_blocks = blocks; + } + + rdev = le32_to_cpu (sd->u.sd_rdev); + inode->u.reiserfs_i.i_first_direct_byte = le32_to_cpu (sd->sd_first_direct_byte); + } else { + // new stat data found, but object may have old items + // (directories and symlinks) + struct stat_data * sd = (struct stat_data *)B_I_PITEM (bh, ih); + + /* both old and new directories have old keys */ + //version = (S_ISDIR (sd->sd_mode) ? ITEM_VERSION_1 : ITEM_VERSION_2); + if (S_ISDIR (sd->sd_mode) || S_ISLNK (sd->sd_mode)) + inode_items_version (inode) = ITEM_VERSION_1; + else + inode_items_version (inode) = ITEM_VERSION_2; + inode->i_mode = le16_to_cpu (sd->sd_mode); + inode->i_nlink = le32_to_cpu (sd->sd_nlink); + inode->i_uid = le32_to_cpu (sd->sd_uid); + inode->i_size = le64_to_cpu (sd->sd_size); + inode->i_gid = le32_to_cpu (sd->sd_gid); + inode->i_mtime = le32_to_cpu (sd->sd_mtime); + inode->i_atime = le32_to_cpu (sd->sd_atime); + inode->i_ctime = le32_to_cpu (sd->sd_ctime); + inode->i_blocks = le32_to_cpu (sd->sd_blocks); + rdev = le32_to_cpu (sd->u.sd_rdev); + } + + /* nopack = 0, by default */ + inode->u.reiserfs_i.nopack = 0; + + pathrelse (path); + if (S_ISREG (inode->i_mode)) { + inode->i_op = &reiserfs_file_inode_operations; + inode->i_fop = &reiserfs_file_operations; + inode->i_mapping->a_ops = &reiserfs_address_space_operations ; + } else if (S_ISDIR (inode->i_mode)) { + inode->i_op = &reiserfs_dir_inode_operations; + inode->i_fop = &reiserfs_dir_operations; + } else if (S_ISLNK (inode->i_mode)) { + inode->i_op = &page_symlink_inode_operations; + inode->i_mapping->a_ops = &reiserfs_address_space_operations; + } else { + inode->i_blocks = 0; + init_special_inode(inode, inode->i_mode, rdev) ; + } +} + + +// update new stat data with inode fields +static void inode2sd (void * sd, struct inode * inode) +{ + struct stat_data * sd_v2 = (struct stat_data *)sd; + + sd_v2->sd_mode = cpu_to_le16 (inode->i_mode); + sd_v2->sd_nlink = cpu_to_le16 (inode->i_nlink); + sd_v2->sd_uid = cpu_to_le32 (inode->i_uid); + sd_v2->sd_size = cpu_to_le64 (inode->i_size); + sd_v2->sd_gid = cpu_to_le32 (inode->i_gid); + sd_v2->sd_mtime = cpu_to_le32 (inode->i_mtime); + sd_v2->sd_atime = cpu_to_le32 (inode->i_atime); + sd_v2->sd_ctime = cpu_to_le32 (inode->i_ctime); + sd_v2->sd_blocks = cpu_to_le32 (inode->i_blocks); + if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) + sd_v2->u.sd_rdev = cpu_to_le32 (inode->i_rdev); +} + + +// used to copy inode's fields to old stat data +static void inode2sd_v1 (void * sd, struct inode * inode) +{ + struct stat_data_v1 * sd_v1 = (struct stat_data_v1 *)sd; + + sd_v1->sd_mode = cpu_to_le16 (inode->i_mode); + sd_v1->sd_uid = cpu_to_le16 (inode->i_uid); + sd_v1->sd_gid = cpu_to_le16 (inode->i_gid); + sd_v1->sd_nlink = cpu_to_le16 (inode->i_nlink); + sd_v1->sd_size = cpu_to_le32 (inode->i_size); + sd_v1->sd_atime = cpu_to_le32 (inode->i_atime); + sd_v1->sd_ctime = cpu_to_le32 (inode->i_ctime); + sd_v1->sd_mtime = cpu_to_le32 (inode->i_mtime); + if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) + sd_v1->u.sd_rdev = cpu_to_le32 (inode->i_rdev); + else + sd_v1->u.sd_blocks = cpu_to_le32 (inode->i_blocks); + + // Sigh. i_first_direct_byte is back + sd_v1->sd_first_direct_byte = cpu_to_le32 (inode->u.reiserfs_i.i_first_direct_byte); +} + + +/* NOTE, you must prepare the buffer head before sending it here, +** and then log it after the call +*/ +static void update_stat_data (struct path * path, struct inode * inode) +{ + struct buffer_head * bh; + struct item_head * ih; + + bh = PATH_PLAST_BUFFER (path); + ih = PATH_PITEM_HEAD (path); + + if (!is_statdata_le_ih (ih)) + reiserfs_panic (inode->i_sb, "vs-13065: update_stat_data: key %k, found item %h", + INODE_PKEY (inode), ih); + + if (stat_data_v1 (ih)) { + // path points to old stat data + inode2sd_v1 (B_I_PITEM (bh, ih), inode); + } else { + inode2sd (B_I_PITEM (bh, ih), inode); + } + + return; +} + + +void reiserfs_update_sd (struct reiserfs_transaction_handle *th, + struct inode * inode) +{ + struct cpu_key key; + INITIALIZE_PATH(path); + struct buffer_head *bh ; + int fs_gen ; + struct item_head *ih, tmp_ih ; + int retval; + + make_cpu_key (&key, inode, SD_OFFSET, TYPE_STAT_DATA, 3);//key type is unimportant + + for(;;) { + int pos; + /* look for the object's stat data */ + retval = search_item (inode->i_sb, &key, &path); + if (retval == IO_ERROR) { + reiserfs_warning ("vs-13050: reiserfs_update_sd: " + "i/o failure occurred trying to update %K stat data", + &key); + return; + } + if (retval == ITEM_NOT_FOUND) { + pos = PATH_LAST_POSITION (&path); + pathrelse(&path) ; + if (inode->i_nlink == 0) { + /*printk ("vs-13050: reiserfs_update_sd: i_nlink == 0, stat data not found\n");*/ + return; + } + reiserfs_warning ("vs-13060: reiserfs_update_sd: " + "stat data of object %k (nlink == %d) not found (pos %d)\n", + INODE_PKEY (inode), inode->i_nlink, pos); + reiserfs_check_path(&path) ; + return; + } + + /* sigh, prepare_for_journal might schedule. When it schedules the + ** FS might change. We have to detect that, and loop back to the + ** search if the stat data item has moved + */ + bh = get_bh(&path) ; + ih = get_ih(&path) ; + copy_item_head (&tmp_ih, ih); + fs_gen = get_generation (inode->i_sb); + reiserfs_prepare_for_journal(inode->i_sb, bh, 1) ; + if (fs_changed (fs_gen, inode->i_sb) && item_moved(&tmp_ih, &path)) { + reiserfs_restore_prepared_buffer(inode->i_sb, bh) ; + continue ; /* Stat_data item has been moved after scheduling. */ + } + break; + } + update_stat_data (&path, inode); + journal_mark_dirty(th, th->t_super, bh) ; + pathrelse (&path); + return; +} + +void reiserfs_read_inode(struct inode *inode) { + make_bad_inode(inode) ; +} + + +// +// initially this function was derived from minix or ext2's analog and +// evolved as the prototype did +// + +/* looks for stat data in the tree, and fills up the fields of in-core + inode stat data fields */ +void reiserfs_read_inode2 (struct inode * inode, void *p) +{ + INITIALIZE_PATH (path_to_sd); + struct cpu_key key; + struct reiserfs_iget4_args *args = (struct reiserfs_iget4_args *)p ; + unsigned long dirino; + int retval; + + if (!p) { + make_bad_inode(inode) ; + return; + } + + dirino = args->objectid ; + + /* set version 1, version 2 could be used too, because stat data + key is the same in both versions */ + key.version = ITEM_VERSION_1; + key.on_disk_key.k_dir_id = dirino; + key.on_disk_key.k_objectid = inode->i_ino; + key.on_disk_key.u.k_offset_v1.k_offset = SD_OFFSET; + key.on_disk_key.u.k_offset_v1.k_uniqueness = SD_UNIQUENESS; + + /* look for the object's stat data */ + retval = search_item (inode->i_sb, &key, &path_to_sd); + if (retval == IO_ERROR) { + reiserfs_warning ("vs-13070: reiserfs_read_inode2: " + "i/o failure occurred trying to find stat data of %K\n", + &key); + make_bad_inode(inode) ; + return; + } + if (retval != ITEM_FOUND) { + reiserfs_warning ("vs-13042: reiserfs_read_inode2: %K not found\n", &key); + pathrelse (&path_to_sd); + make_bad_inode(inode) ; + return; + } + + init_inode (inode, &path_to_sd); + reiserfs_check_path(&path_to_sd) ; /* init inode should be relsing */ + +} + + +struct inode * reiserfs_iget (struct super_block * s, struct cpu_key * key) +{ + struct inode * inode; + struct reiserfs_iget4_args args ; + + args.objectid = key->on_disk_key.k_dir_id ; + inode = iget4 (s, key->on_disk_key.k_objectid, 0, (void *)(&args)); + if (!inode) + return inode ; + + // if (comp_short_keys (INODE_PKEY (inode), key)) { + if (is_bad_inode (inode)) { + reiserfs_warning ("vs-13048: reiserfs_iget: " + "bad_inode. Stat data of (%lu %lu) not found\n", + key->on_disk_key.k_dir_id, key->on_disk_key.k_objectid); + iput (inode); + inode = 0; + } + return inode; +} + + +// +// initially this function was derived from minix or ext2's analog and +// evolved as the prototype did +// +/* looks for stat data, then copies fields to it, marks the buffer + containing stat data as dirty */ +/* reiserfs inodes are never really dirty, since the dirty inode call +** always logs them. This call allows the VFS inode marking routines +** to properly mark inodes for datasync and such, but only actually +** does something when called for a synchronous update. +*/ +void reiserfs_write_inode (struct inode * inode, int do_sync) { + struct reiserfs_transaction_handle th ; + int jbegin_count = 1 ; + + if (inode->i_sb->s_flags & MS_RDONLY) { + reiserfs_warning("clm-6005: writing inode %lu on readonly FS\n", + inode->i_ino) ; + return ; + } + if (do_sync) { + lock_kernel() ; + journal_begin(&th, inode->i_sb, jbegin_count) ; + reiserfs_update_sd (&th, inode); + journal_end_sync(&th, inode->i_sb, jbegin_count) ; + unlock_kernel() ; + } +} + +void reiserfs_dirty_inode (struct inode * inode) { + struct reiserfs_transaction_handle th ; + + if (inode->i_sb->s_flags & MS_RDONLY) { + reiserfs_warning("clm-6006: writing inode %lu on readonly FS\n", + inode->i_ino) ; + return ; + } + lock_kernel() ; + journal_begin(&th, inode->i_sb, 1) ; + reiserfs_update_sd (&th, inode); + journal_end(&th, inode->i_sb, 1) ; + unlock_kernel() ; +} + + +/* FIXME: no need any more. right? */ +int reiserfs_sync_inode (struct reiserfs_transaction_handle *th, struct inode * inode) +{ + int err = 0; + + reiserfs_update_sd (th, inode); + return err; +} + + +/* stat data of new object is inserted already, this inserts the item + containing "." and ".." entries */ +static int reiserfs_new_directory (struct reiserfs_transaction_handle *th, + struct item_head * ih, struct path * path, const struct inode * dir) +{ + struct super_block * sb = th->t_super; + char empty_dir [EMPTY_DIR_SIZE]; + char * body = empty_dir; + struct cpu_key key; + int retval; + + _make_cpu_key (&key, ITEM_VERSION_1, le32_to_cpu (ih->ih_key.k_dir_id), + le32_to_cpu (ih->ih_key.k_objectid), DOT_OFFSET, TYPE_DIRENTRY, 3/*key length*/); + + /* compose item head for new item. Directories consist of items of + old type (ITEM_VERSION_1). Do not set key (second arg is 0), it + is done by reiserfs_new_inode */ + if (old_format_only (sb)) { + make_le_item_head (ih, 0, ITEM_VERSION_1, DOT_OFFSET, TYPE_DIRENTRY, EMPTY_DIR_SIZE_V1, 2); + + make_empty_dir_item_v1 (body, ih->ih_key.k_dir_id, ih->ih_key.k_objectid, + le32_to_cpu (INODE_PKEY (dir)->k_dir_id), + le32_to_cpu (INODE_PKEY (dir)->k_objectid)); + } else { + make_le_item_head (ih, 0, ITEM_VERSION_1, DOT_OFFSET, TYPE_DIRENTRY, EMPTY_DIR_SIZE, 2); + + make_empty_dir_item (body, ih->ih_key.k_dir_id, ih->ih_key.k_objectid, + le32_to_cpu (INODE_PKEY (dir)->k_dir_id), + le32_to_cpu (INODE_PKEY (dir)->k_objectid)); + } + + /* look for place in the tree for new item */ + retval = search_item (sb, &key, path); + if (retval == IO_ERROR) { + reiserfs_warning ("vs-13080: reiserfs_new_directory: " + "i/o failure occured creating new directory\n"); + return -EIO; + } + if (retval == ITEM_FOUND) { + pathrelse (path); + reiserfs_warning ("vs-13070: reiserfs_new_directory: " + "object with this key exists (%k)", &(ih->ih_key)); + return -EEXIST; + } + + /* insert item, that is empty directory item */ + return reiserfs_insert_item (th, path, &key, ih, body); +} + + +/* stat data of object has been inserted, this inserts the item + containing the body of symlink */ +static int reiserfs_new_symlink (struct reiserfs_transaction_handle *th, + struct item_head * ih, + struct path * path, const char * symname, int item_len) +{ + struct super_block * sb = th->t_super; + struct cpu_key key; + int retval; + + _make_cpu_key (&key, ITEM_VERSION_1, + le32_to_cpu (ih->ih_key.k_dir_id), + le32_to_cpu (ih->ih_key.k_objectid), + 1, TYPE_DIRECT, 3/*key length*/); + + make_le_item_head (ih, 0, ITEM_VERSION_1, 1, TYPE_DIRECT, item_len, 0/*free_space*/); + + /* look for place in the tree for new item */ + retval = search_item (sb, &key, path); + if (retval == IO_ERROR) { + reiserfs_warning ("vs-13080: reiserfs_new_symlinik: " + "i/o failure occured creating new symlink\n"); + return -EIO; + } + if (retval == ITEM_FOUND) { + pathrelse (path); + reiserfs_warning ("vs-13080: reiserfs_new_symlink: " + "object with this key exists (%k)", &(ih->ih_key)); + return -EEXIST; + } + + /* insert item, that is body of symlink */ + return reiserfs_insert_item (th, path, &key, ih, symname); +} + + +/* inserts the stat data into the tree, and then calls + reiserfs_new_directory (to insert ".", ".." item if new object is + directory) or reiserfs_new_symlink (to insert symlink body if new + object is symlink) or nothing (if new object is regular file) */ +struct inode * reiserfs_new_inode (struct reiserfs_transaction_handle *th, + const struct inode * dir, int mode, + const char * symname, + int i_size, /* 0 for regular, EMTRY_DIR_SIZE for dirs, + strlen (symname) for symlinks)*/ + struct dentry *dentry, struct inode *inode, int * err) +{ + struct super_block * sb; + INITIALIZE_PATH (path_to_key); + struct cpu_key key; + struct item_head ih; + struct stat_data sd; + int retval; + + if (!dir || !dir->i_nlink) { + *err = -EPERM; + iput(inode) ; + return NULL; + } + + sb = dir->i_sb; + inode->i_sb = sb; + inode->i_flags = 0;//inode->i_sb->s_flags; + + /* item head of new item */ + ih.ih_key.k_dir_id = INODE_PKEY (dir)->k_objectid; + ih.ih_key.k_objectid = cpu_to_le32 (reiserfs_get_unused_objectid (th)); + if (!ih.ih_key.k_objectid) { + iput(inode) ; + *err = -ENOMEM; + return NULL; + } + if (old_format_only (sb)) + make_le_item_head (&ih, 0, ITEM_VERSION_1, SD_OFFSET, TYPE_STAT_DATA, SD_V1_SIZE, MAX_US_INT); + else + make_le_item_head (&ih, 0, ITEM_VERSION_2, SD_OFFSET, TYPE_STAT_DATA, SD_SIZE, MAX_US_INT); + + + /* key to search for correct place for new stat data */ + _make_cpu_key (&key, ITEM_VERSION_2, le32_to_cpu (ih.ih_key.k_dir_id), + le32_to_cpu (ih.ih_key.k_objectid), SD_OFFSET, TYPE_STAT_DATA, 3/*key length*/); + + /* find proper place for inserting of stat data */ + retval = search_item (sb, &key, &path_to_key); + if (retval == IO_ERROR) { + iput (inode); + *err = -EIO; + return NULL; + } + if (retval == ITEM_FOUND) { + pathrelse (&path_to_key); + iput (inode); + *err = -EEXIST; + return NULL; + } + + /* fill stat data */ + inode->i_mode = mode; + inode->i_nlink = (S_ISDIR (mode) ? 2 : 1); + inode->i_uid = current->fsuid; + if (dir->i_mode & S_ISGID) { + inode->i_gid = dir->i_gid; + if (S_ISDIR(mode)) + inode->i_mode |= S_ISGID; + } else + inode->i_gid = current->fsgid; + + inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME; + inode->i_size = i_size; + inode->i_blocks = (inode->i_size + 511) >> 9; + inode->u.reiserfs_i.i_first_direct_byte = S_ISLNK(mode) ? 1 : + U32_MAX/*NO_BYTES_IN_DIRECT_ITEM*/; + + if (old_format_only (sb)) + inode2sd_v1 (&sd, inode); + else + inode2sd (&sd, inode); + + // these do not go to on-disk stat data + inode->i_ino = le32_to_cpu (ih.ih_key.k_objectid); + inode->i_blksize = PAGE_SIZE; + inode->i_dev = sb->s_dev; + + // store in in-core inode the key of stat data and version all + // object items will have (directory items will have old offset + // format, other new objects will consist of new items) + memcpy (INODE_PKEY (inode), &(ih.ih_key), KEY_SIZE); + if (old_format_only (sb) || S_ISDIR(mode) || S_ISLNK(mode)) + inode_items_version (inode) = ITEM_VERSION_1; + else + inode_items_version (inode) = ITEM_VERSION_2; + + /* insert the stat data into the tree */ + retval = reiserfs_insert_item (th, &path_to_key, &key, &ih, (char *)(&sd)); + if (retval) { + iput (inode); + *err = retval; + reiserfs_check_path(&path_to_key) ; + return NULL; + } + + if (S_ISDIR(mode)) { + /* insert item with "." and ".." */ + retval = reiserfs_new_directory (th, &ih, &path_to_key, dir); + } + + if (S_ISLNK(mode)) { + /* insert body of symlink */ + if (!old_format_only (sb)) + i_size = ROUND_UP(i_size); + retval = reiserfs_new_symlink (th, &ih, &path_to_key, symname, i_size); + } + if (retval) { + inode->i_nlink = 0; + iput (inode); + *err = retval; + reiserfs_check_path(&path_to_key) ; + return NULL; + } + + /* not a perfect generation count, as object ids can be reused, but this + ** is as good as reiserfs can do right now + */ + inode->i_generation = INODE_PKEY (inode)->k_dir_id; + insert_inode_hash (inode); + // we do not mark inode dirty: on disk content matches to the + // in-core one + reiserfs_check_path(&path_to_key) ; + + return inode; +} + +/* +** finds the tail page in the page cache, +** reads the last block in. +** +** On success, page_result is set to a locked, pinned page, and bh_result +** is set to an up to date buffer for the last block in the file. returns 0. +** +** tail conversion is not done, so bh_result might not be valid for writing +** check buffer_mapped(bh_result) and bh_result->b_blocknr != 0 before +** trying to write the block. +** +** on failure, nonzero is returned, page_result and bh_result are untouched. +*/ +static int grab_tail_page(struct inode *p_s_inode, + struct page **page_result, + struct buffer_head **bh_result) { + + /* we want the page with the last byte in the file, + ** not the page that will hold the next byte for appending + */ + unsigned long index = (p_s_inode->i_size-1) >> PAGE_CACHE_SHIFT ; + unsigned long pos = 0 ; + unsigned long start = 0 ; + unsigned long blocksize = p_s_inode->i_sb->s_blocksize ; + unsigned long offset = (p_s_inode->i_size) & (PAGE_CACHE_SIZE - 1) ; + struct buffer_head *bh ; + struct buffer_head *head ; + struct page * page ; + int error ; + + /* we know that we are only called with inode->i_size > 0. + ** we also know that a file tail can never be as big as a block + ** If i_size % blocksize == 0, our file is currently block aligned + ** and it won't need converting or zeroing after a truncate. + */ + if ((offset & (blocksize - 1)) == 0) { + return -ENOENT ; + } + page = grab_cache_page(p_s_inode->i_mapping, index) ; + error = PTR_ERR(page) ; + if (IS_ERR(page)) { + goto out ; + } + /* start within the page of the last block in the file */ + start = (offset / blocksize) * blocksize ; + + error = block_prepare_write(page, start, offset, + reiserfs_get_block_create_0) ; + if (error) + goto unlock ; + + kunmap(page) ; /* mapped by block_prepare_write */ + + head = page->buffers ; + bh = head; + do { + if (pos >= start) { + break ; + } + bh = bh->b_this_page ; + pos += blocksize ; + } while(bh != head) ; + + if (!buffer_uptodate(bh)) { + /* note, this should never happen, prepare_write should + ** be taking care of this for us. If the buffer isn't up to date, + ** I've screwed up the code to find the buffer, or the code to + ** call prepare_write + */ + reiserfs_warning("clm-6000: error reading block %lu on dev %s\n", + bh->b_blocknr, kdevname(bh->b_dev)) ; + error = -EIO ; + goto unlock ; + } + *bh_result = bh ; + *page_result = page ; + +out: + return error ; + +unlock: + UnlockPage(page) ; + page_cache_release(page) ; + return error ; +} + +/* +** vfs version of truncate file. Must NOT be called with +** a transaction already started. +** +** some code taken from block_truncate_page +*/ +void reiserfs_truncate_file(struct inode *p_s_inode, int update_timestamps) { + struct reiserfs_transaction_handle th ; + int windex ; + + /* we want the offset for the first byte after the end of the file */ + unsigned long offset = p_s_inode->i_size & (PAGE_CACHE_SIZE - 1) ; + unsigned blocksize = p_s_inode->i_sb->s_blocksize ; + unsigned length ; + struct page *page = NULL ; + int error ; + struct buffer_head *bh = NULL ; + + if (p_s_inode->i_size > 0) { + if ((error = grab_tail_page(p_s_inode, &page, &bh))) { + // -ENOENT means we truncated past the end of the file, + // and get_block_create_0 could not find a block to read in, + // which is ok. + if (error != -ENOENT) + reiserfs_warning("clm-6001: grab_tail_page failed %d\n", error); + page = NULL ; + bh = NULL ; + } + } + + /* so, if page != NULL, we have a buffer head for the offset at + ** the end of the file. if the bh is mapped, and bh->b_blocknr != 0, + ** then we have an unformatted node. Otherwise, we have a direct item, + ** and no zeroing is required. We zero after the truncate, because the + ** truncate might pack the item anyway (it will unmap bh if it packs). + */ + prevent_flush_page_lock(page, p_s_inode) ; + journal_begin(&th, p_s_inode->i_sb, JOURNAL_PER_BALANCE_CNT * 2 ) ; + windex = push_journal_writer("reiserfs_vfs_truncate_file") ; + reiserfs_do_truncate (&th, p_s_inode, page, update_timestamps) ; + pop_journal_writer(windex) ; + journal_end(&th, p_s_inode->i_sb, JOURNAL_PER_BALANCE_CNT * 2 ) ; + allow_flush_page_lock(page, p_s_inode) ; + + if (page && buffer_mapped(bh) && bh->b_blocknr != 0) { + length = offset & (blocksize - 1) ; + /* if we are not on a block boundary */ + if (length) { + length = blocksize - length ; + memset((char *)kmap(page) + offset, 0, length) ; + flush_dcache_page(page) ; + kunmap(page) ; + mark_buffer_dirty(bh) ; + } + } + + if (page) { + UnlockPage(page) ; + page_cache_release(page) ; + } + return ; +} + +static int map_block_for_writepage(struct inode *inode, + struct buffer_head *bh_result, + unsigned long block) { + struct reiserfs_transaction_handle th ; + int fs_gen ; + struct item_head tmp_ih ; + struct item_head *ih ; + struct buffer_head *bh ; + __u32 *item ; + struct cpu_key key ; + INITIALIZE_PATH(path) ; + int pos_in_item ; + int jbegin_count = JOURNAL_PER_BALANCE_CNT ; + loff_t byte_offset = (block << inode->i_sb->s_blocksize_bits) + 1 ; + int retval ; + int use_get_block = 0 ; + int bytes_copied = 0 ; + int copy_size ; + +start_over: + lock_kernel() ; + prevent_flush_page_lock(bh_result->b_page, inode) ; + journal_begin(&th, inode->i_sb, jbegin_count) ; + + make_cpu_key(&key, inode, byte_offset, TYPE_ANY, 3) ; + +research: + retval = search_for_position_by_key(inode->i_sb, &key, &path) ; + if (retval != POSITION_FOUND) { + use_get_block = 1; + goto out ; + } + + bh = get_bh(&path) ; + ih = get_ih(&path) ; + item = get_item(&path) ; + pos_in_item = path.pos_in_item ; + + /* we've found an unformatted node */ + if (indirect_item_found(retval, ih)) { + if (bytes_copied > 0) { + reiserfs_warning("clm-6002: bytes_copied %d\n", bytes_copied) ; + } + if (!item[pos_in_item]) { + /* crap, we are writing to a hole */ + use_get_block = 1; + goto out ; + } + set_block_dev_mapped(bh_result, le32_to_cpu(item[pos_in_item]), inode); + } else if (is_direct_le_ih(ih)) { + char *p ; + p = page_address(bh_result->b_page) ; + p += (byte_offset -1) & (PAGE_CACHE_SIZE - 1) ; + copy_size = le16_to_cpu(ih->ih_item_len) - pos_in_item ; + + fs_gen = get_generation(inode->i_sb) ; + copy_item_head(&tmp_ih, ih) ; + reiserfs_prepare_for_journal(inode->i_sb, bh, 1) ; + if (fs_changed (fs_gen, inode->i_sb) && item_moved (&tmp_ih, &path)) { + reiserfs_restore_prepared_buffer(inode->i_sb, bh) ; + goto research; + } + + memcpy( B_I_PITEM(bh, ih) + pos_in_item, p + bytes_copied, copy_size) ; + + journal_mark_dirty(&th, inode->i_sb, bh) ; + bytes_copied += copy_size ; + set_block_dev_mapped(bh_result, 0, inode); + + /* are there still bytes left? */ + if (bytes_copied < bh_result->b_size && + (byte_offset + bytes_copied) < inode->i_size) { + set_cpu_key_k_offset(&key, cpu_key_k_offset(&key) + copy_size) ; + goto research ; + } + } else { + reiserfs_warning("clm-6003: bad item inode %lu, device %s\n", inode->i_ino, kdevname(inode->i_sb->s_dev)) ; + retval = -EIO ; + goto out ; + } + retval = 0 ; + +out: + pathrelse(&path) ; + journal_end(&th, inode->i_sb, jbegin_count) ; + allow_flush_page_lock(bh_result->b_page, inode) ; + unlock_kernel() ; + + /* this is where we fill in holes in the file. */ + if (use_get_block) { + kmap(bh_result->b_page) ; + retval = reiserfs_get_block(inode, block, bh_result, 1) ; + kunmap(bh_result->b_page) ; + if (!retval) { + if (!buffer_mapped(bh_result) || bh_result->b_blocknr == 0) { + /* get_block failed to find a mapped unformatted node. */ + use_get_block = 0 ; + goto start_over ; + } + } + } + return retval ; +} + +/* helper func to get a buffer head ready for writepage to send to +** ll_rw_block +*/ +static inline void submit_bh_for_writepage(struct buffer_head **bhp, int nr) { + struct buffer_head *bh ; + int i; + for(i = 0 ; i < nr ; i++) { + bh = bhp[i] ; + lock_buffer(bh) ; + atomic_inc(&bh->b_count) ; /* async end_io handler decs this */ + set_buffer_async_io(bh) ; + /* submit_bh doesn't care if the buffer is dirty, but nobody + ** later on in the call chain will be cleaning it. So, we + ** clean the buffer here, it still gets written either way. + */ + clear_bit(BH_Dirty, &bh->b_state) ; + set_bit(BH_Uptodate, &bh->b_state) ; + submit_bh(WRITE, bh) ; + } +} + +static int reiserfs_write_full_page(struct page *page) { + struct inode *inode = page->mapping->host ; + unsigned long end_index = inode->i_size >> PAGE_CACHE_SHIFT ; + unsigned last_offset = PAGE_CACHE_SIZE; + int error = 0; + unsigned long block ; + unsigned cur_offset = 0 ; + struct buffer_head *head, *bh ; + int partial = 0 ; + struct buffer_head *arr[PAGE_CACHE_SIZE/512] ; + int nr = 0 ; + + if (!page->buffers) { + block_prepare_write(page, 0, 0, NULL) ; + kunmap(page) ; + } + /* last page in the file, zero out any contents past the + ** last byte in the file + */ + if (page->index >= end_index) { + last_offset = inode->i_size & (PAGE_CACHE_SIZE - 1) ; + /* no file contents in this page */ + if (page->index >= end_index + 1 || !last_offset) { + error = -EIO ; + goto fail ; + } + memset((char *)kmap(page)+last_offset, 0, PAGE_CACHE_SIZE-last_offset) ; + flush_dcache_page(page) ; + kunmap(page) ; + } + head = page->buffers ; + bh = head ; + block = page->index << (PAGE_CACHE_SHIFT - inode->i_sb->s_blocksize_bits) ; + do { + /* if this offset in the page is outside the file */ + if (cur_offset >= last_offset) { + if (!buffer_uptodate(bh)) + partial = 1 ; + } else { + /* fast path, buffer mapped to an unformatted node */ + if (buffer_mapped(bh) && bh->b_blocknr != 0) { + arr[nr++] = bh ; + } else { + /* buffer not mapped yet, or points to a direct item. + ** search and dirty or log + */ + if ((error = map_block_for_writepage(inode, bh, block))) { + goto fail ; + } + /* map_block_for_writepage either found an unformatted node + ** and mapped it for us, or it found a direct item + ** and logged the changes. + */ + if (buffer_mapped(bh) && bh->b_blocknr != 0) { + arr[nr++] = bh ; + } + } + } + bh = bh->b_this_page ; + cur_offset += bh->b_size ; + block++ ; + } while(bh != head) ; + + /* if this page only had a direct item, it is very possible for + ** nr == 0 without there being any kind of error. + */ + if (nr) { + submit_bh_for_writepage(arr, nr) ; + } else { + UnlockPage(page) ; + } + if (!partial) + SetPageUptodate(page) ; + + return 0 ; + +fail: + if (nr) { + submit_bh_for_writepage(arr, nr) ; + } else { + UnlockPage(page) ; + } + ClearPageUptodate(page) ; + return error ; +} + +// +// this is exactly what 2.3.99-pre9's ext2_readpage is +// +static int reiserfs_readpage (struct file *f, struct page * page) +{ + return block_read_full_page (page, reiserfs_get_block); +} + + +// +// modified from ext2_writepage is +// +static int reiserfs_writepage (struct page * page) +{ + struct inode *inode = page->mapping->host ; + reiserfs_wait_on_write_block(inode->i_sb) ; + return reiserfs_write_full_page(page) ; +} + + +// +// from ext2_prepare_write, but modified +// +int reiserfs_prepare_write(struct file *f, struct page *page, unsigned from, unsigned to) { + struct inode *inode = page->mapping->host ; + reiserfs_wait_on_write_block(inode->i_sb) ; + fix_tail_page_for_writing(page) ; + return block_prepare_write(page, from, to, reiserfs_get_block) ; +} + + +// +// this is exactly what 2.3.99-pre9's ext2_bmap is +// +static int reiserfs_aop_bmap(struct address_space *as, long block) { + return generic_block_bmap(as, block, reiserfs_bmap) ; +} + + +static int reiserfs_commit_write(struct file *f, struct page *page, + unsigned from, unsigned to) { + struct inode *inode = page->mapping->host ; + int ret ; + struct reiserfs_transaction_handle th ; + + reiserfs_wait_on_write_block(inode->i_sb) ; + lock_kernel(); + prevent_flush_page_lock(page, inode) ; + ret = generic_commit_write(f, page, from, to) ; + /* we test for O_SYNC here so we can commit the transaction + ** for any packed tails the file might have had + */ + if (f->f_flags & O_SYNC) { + journal_begin(&th, inode->i_sb, 1) ; + reiserfs_prepare_for_journal(inode->i_sb, + SB_BUFFER_WITH_SB(inode->i_sb), 1) ; + journal_mark_dirty(&th, inode->i_sb, SB_BUFFER_WITH_SB(inode->i_sb)) ; + journal_end_sync(&th, inode->i_sb, 1) ; + } + allow_flush_page_lock(page, inode) ; + unlock_kernel(); + return ret ; +} + +struct address_space_operations reiserfs_address_space_operations = { + writepage: reiserfs_writepage, + readpage: reiserfs_readpage, + sync_page: block_sync_page, + prepare_write: reiserfs_prepare_write, + commit_write: reiserfs_commit_write, + bmap: reiserfs_aop_bmap +} ; diff --git a/fs/reiserfs/ioctl.c b/fs/reiserfs/ioctl.c new file mode 100644 index 000000000..72054ce3a --- /dev/null +++ b/fs/reiserfs/ioctl.c @@ -0,0 +1,101 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +#ifdef __KERNEL__ + +#include <linux/fs.h> +#include <linux/reiserfs_fs.h> +#include <linux/sched.h> +#include <asm/uaccess.h> +#include <linux/smp_lock.h> +#include <linux/locks.h> + +#else + +#include "nokernel.h" + +#endif + +/* +** reiserfs_ioctl - handler for ioctl for inode +** supported commands: +** 1) REISERFS_IOC_UNPACK - try to unpack tail from direct item into indirect +** and prevent packing file (argument arg has to be non-zero) +** 2) That's all for a while ... +*/ +int reiserfs_ioctl (struct inode * inode, struct file * filp, unsigned int cmd, + unsigned long arg) +{ + switch (cmd) { + case REISERFS_IOC_UNPACK: + if (arg) + return reiserfs_unpack (inode, filp); + + default: + return -ENOTTY; + } +} + +/* +** reiserfs_unpack +** Function try to convert tail from direct item into indirect. +** It set up nopack attribute in the inode.u.reiserfs_i.nopack +*/ +int reiserfs_unpack (struct inode * inode, struct file * filp) +{ + int retval = 0; + int index ; + struct page *page ; + unsigned long write_from ; + unsigned long blocksize = inode->i_sb->s_blocksize ; + + if (inode->i_size == 0) { + return -EINVAL ; + } + /* ioctl already done */ + if (inode->u.reiserfs_i.nopack) { + return 0 ; + } + lock_kernel(); + + /* we need to make sure nobody is changing the file size beneath + ** us + */ + down(&inode->i_sem) ; + + write_from = inode->i_size & (blocksize - 1) ; + /* if we are on a block boundary, we are already unpacked. */ + if ( write_from == 0) { + inode->u.reiserfs_i.nopack = 1; + goto out ; + } + + /* we unpack by finding the page with the tail, and calling + ** reiserfs_prepare_write on that page. This will force a + ** reiserfs_get_block to unpack the tail for us. + */ + index = inode->i_size >> PAGE_CACHE_SHIFT ; + page = grab_cache_page(inode->i_mapping, index) ; + retval = PTR_ERR(page) ; + if (IS_ERR(page)) { + goto out ; + } + retval = reiserfs_prepare_write(NULL, page, write_from, blocksize) ; + if (retval) + goto out_unlock ; + + /* conversion can change page contents, must flush */ + flush_dcache_page(page) ; + inode->u.reiserfs_i.nopack = 1; + kunmap(page) ; /* mapped by prepare_write */ + +out_unlock: + UnlockPage(page) ; + page_cache_release(page) ; + +out: + up(&inode->i_sem) ; + unlock_kernel(); + return retval; +} diff --git a/fs/reiserfs/item_ops.c b/fs/reiserfs/item_ops.c new file mode 100644 index 000000000..b3f962d05 --- /dev/null +++ b/fs/reiserfs/item_ops.c @@ -0,0 +1,718 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +#ifdef __KERNEL__ + +#include <linux/sched.h> +#include <linux/reiserfs_fs.h> + +#else + +#include "nokernel.h" + +#endif + + +// this contains item handlers for old item types: sd, direct, +// indirect, directory + +/* and where are the comments? how about saying where we can find an + explanation of each item handler method? -Hans */ + +////////////////////////////////////////////////////////////////////////////// +// stat data functions +// +static int sd_bytes_number (struct item_head * ih, int block_size) +{ + return 0; +} + +static void sd_decrement_key (struct cpu_key * key) +{ + key->on_disk_key.k_objectid --; + set_cpu_key_k_type (key, TYPE_ANY); + set_cpu_key_k_offset(key, (loff_t)(-1)); +} + +static int sd_is_left_mergeable (struct key * key, unsigned long bsize) +{ + return 0; +} + + + +static char * print_time (time_t t) +{ + static char timebuf[256]; + +#ifndef __KERNEL__ +// struct tm *loctime; +// loctime = localtime (&t); + sprintf (timebuf, "%s", asctime (localtime (&t))); + timebuf[strlen (timebuf) - 1] = 0; +#else + sprintf (timebuf, "%ld", t); +#endif + return timebuf; +} + + +static void sd_print_item (struct item_head * ih, char * item) +{ + printk ("\tmode | size | nlinks | first direct | mtime\n"); + if (stat_data_v1 (ih)) { + struct stat_data_v1 * sd = (struct stat_data_v1 *)item; + + printk ("\t0%-6o | %6u | %2u | %d | %s\n", sd->sd_mode, sd->sd_size, + sd->sd_nlink, sd->sd_first_direct_byte, print_time (sd->sd_mtime)); + } else { + struct stat_data * sd = (struct stat_data *)item; + + printk ("\t0%-6o | %6Lu | %2u | %d | %s\n", sd->sd_mode, (unsigned long long)(sd->sd_size), + sd->sd_nlink, sd->u.sd_rdev, print_time (sd->sd_mtime)); + } +} + +static void sd_check_item (struct item_head * ih, char * item) +{ + // FIXME: type something here! +} + + +static int sd_create_vi (struct virtual_node * vn, + struct virtual_item * vi, + int is_affected, + int insert_size) +{ + vi->vi_index = TYPE_STAT_DATA; + //vi->vi_type |= VI_TYPE_STAT_DATA;// not needed? + return 0; +} + + +static int sd_check_left (struct virtual_item * vi, int free, + int start_skip, int end_skip) +{ + if (start_skip || end_skip) + BUG (); + return -1; +} + + +static int sd_check_right (struct virtual_item * vi, int free) +{ + return -1; +} + +static int sd_part_size (struct virtual_item * vi, int first, int count) +{ + if (count) + BUG (); + return 0; +} + +static int sd_unit_num (struct virtual_item * vi) +{ + return vi->vi_item_len - IH_SIZE; +} + + +static void sd_print_vi (struct virtual_item * vi) +{ + reiserfs_warning ("STATDATA, index %d, type 0x%x, %h\n", + vi->vi_index, vi->vi_type, vi->vi_ih); +} + +struct item_operations stat_data_ops = { + sd_bytes_number, + sd_decrement_key, + sd_is_left_mergeable, + sd_print_item, + sd_check_item, + + sd_create_vi, + sd_check_left, + sd_check_right, + sd_part_size, + sd_unit_num, + sd_print_vi +}; + + + +////////////////////////////////////////////////////////////////////////////// +// direct item functions +// +static int direct_bytes_number (struct item_head * ih, int block_size) +{ + return le16_to_cpu (ih->ih_item_len); +} + + +// FIXME: this should probably switch to indirect as well +static void direct_decrement_key (struct cpu_key * key) +{ + cpu_key_k_offset_dec (key); + if (cpu_key_k_offset (key) == 0) + set_cpu_key_k_type (key, TYPE_STAT_DATA); +} + + +static int direct_is_left_mergeable (struct key * key, unsigned long bsize) +{ + int version = le_key_version (key); + return ((le_key_k_offset (version, key) & (bsize - 1)) != 1); +} + + +static void direct_print_item (struct item_head * ih, char * item) +{ + int j = 0; + +// return; + printk ("\""); + while (j < ih->ih_item_len) + printk ("%c", item[j++]); + printk ("\"\n"); +} + + +static void direct_check_item (struct item_head * ih, char * item) +{ + // FIXME: type something here! +} + + +static int direct_create_vi (struct virtual_node * vn, + struct virtual_item * vi, + int is_affected, + int insert_size) +{ + vi->vi_index = TYPE_DIRECT; + //vi->vi_type |= VI_TYPE_DIRECT; + return 0; +} + +static int direct_check_left (struct virtual_item * vi, int free, + int start_skip, int end_skip) +{ + int bytes; + + bytes = free - free % 8; + return bytes ?: -1; +} + + +static int direct_check_right (struct virtual_item * vi, int free) +{ + return direct_check_left (vi, free, 0, 0); +} + +static int direct_part_size (struct virtual_item * vi, int first, int count) +{ + return count; +} + + +static int direct_unit_num (struct virtual_item * vi) +{ + return vi->vi_item_len - IH_SIZE; +} + + +static void direct_print_vi (struct virtual_item * vi) +{ + reiserfs_warning ("DIRECT, index %d, type 0x%x, %h\n", + vi->vi_index, vi->vi_type, vi->vi_ih); +} + +struct item_operations direct_ops = { + direct_bytes_number, + direct_decrement_key, + direct_is_left_mergeable, + direct_print_item, + direct_check_item, + + direct_create_vi, + direct_check_left, + direct_check_right, + direct_part_size, + direct_unit_num, + direct_print_vi +}; + + + +////////////////////////////////////////////////////////////////////////////// +// indirect item functions +// + +static int indirect_bytes_number (struct item_head * ih, int block_size) +{ + return le16_to_cpu (ih->ih_item_len) / UNFM_P_SIZE * block_size; //- get_ih_free_space (ih); +} + + +// decrease offset, if it becomes 0, change type to stat data +static void indirect_decrement_key (struct cpu_key * key) +{ + cpu_key_k_offset_dec (key); + if (cpu_key_k_offset (key) == 0) + set_cpu_key_k_type (key, TYPE_STAT_DATA); +} + + +// if it is not first item of the body, then it is mergeable +static int indirect_is_left_mergeable (struct key * key, unsigned long bsize) +{ + int version = le_key_version (key); + return (le_key_k_offset (version, key) != 1); +} + + +// printing of indirect item +static void start_new_sequence (__u32 * start, int * len, __u32 new) +{ + *start = new; + *len = 1; +} + + +static int sequence_finished (__u32 start, int * len, __u32 new) +{ + if (start == INT_MAX) + return 1; + + if (start == 0 && new == 0) { + (*len) ++; + return 0; + } + if (start != 0 && (start + *len) == new) { + (*len) ++; + return 0; + } + return 1; +} + +static void print_sequence (__u32 start, int len) +{ + if (start == INT_MAX) + return; + + if (len == 1) + printk (" %d", start); + else + printk (" %d(%d)", start, len); +} + + +static void indirect_print_item (struct item_head * ih, char * item) +{ + int j; + __u32 * unp, prev = INT_MAX; + int num; + + unp = (__u32 *)item; + + if (ih->ih_item_len % UNFM_P_SIZE) + printk ("indirect_print_item: invalid item len"); + + printk ("%d pointers\n[ ", (int)I_UNFM_NUM (ih)); + for (j = 0; j < I_UNFM_NUM (ih); j ++) { + if (sequence_finished (prev, &num, unp[j])) { + print_sequence (prev, num); + start_new_sequence (&prev, &num, unp[j]); + } + } + print_sequence (prev, num); + printk ("]\n"); +} + +static void indirect_check_item (struct item_head * ih, char * item) +{ + // FIXME: type something here! +} + + +static int indirect_create_vi (struct virtual_node * vn, + struct virtual_item * vi, + int is_affected, + int insert_size) +{ + vi->vi_index = TYPE_INDIRECT; + //vi->vi_type |= VI_TYPE_INDIRECT; + return 0; +} + +static int indirect_check_left (struct virtual_item * vi, int free, + int start_skip, int end_skip) +{ + int bytes; + + bytes = free - free % UNFM_P_SIZE; + return bytes ?: -1; +} + + +static int indirect_check_right (struct virtual_item * vi, int free) +{ + return indirect_check_left (vi, free, 0, 0); +} + + + +// return size in bytes of 'units' units. If first == 0 - calculate from the head (left), otherwise - from tail (right) +static int indirect_part_size (struct virtual_item * vi, int first, int units) +{ + // unit of indirect item is byte (yet) + return units; +} + +static int indirect_unit_num (struct virtual_item * vi) +{ + // unit of indirect item is byte (yet) + return vi->vi_item_len - IH_SIZE; +} + +static void indirect_print_vi (struct virtual_item * vi) +{ + reiserfs_warning ("INDIRECT, index %d, type 0x%x, %h\n", + vi->vi_index, vi->vi_type, vi->vi_ih); +} + +struct item_operations indirect_ops = { + indirect_bytes_number, + indirect_decrement_key, + indirect_is_left_mergeable, + indirect_print_item, + indirect_check_item, + + indirect_create_vi, + indirect_check_left, + indirect_check_right, + indirect_part_size, + indirect_unit_num, + indirect_print_vi +}; + + +////////////////////////////////////////////////////////////////////////////// +// direntry functions +// + + +static int direntry_bytes_number (struct item_head * ih, int block_size) +{ + reiserfs_warning ("vs-16090: direntry_bytes_number: " + "bytes number is asked for direntry"); + return 0; +} + +static void direntry_decrement_key (struct cpu_key * key) +{ + cpu_key_k_offset_dec (key); + if (cpu_key_k_offset (key) == 0) + set_cpu_key_k_type (key, TYPE_STAT_DATA); +} + + +static int direntry_is_left_mergeable (struct key * key, unsigned long bsize) +{ + if (le32_to_cpu (key->u.k_offset_v1.k_offset) == DOT_OFFSET) + return 0; + return 1; + +} + + +static void direntry_print_item (struct item_head * ih, char * item) +{ + int i; + int namelen; + struct reiserfs_de_head * deh; + char * name; + static char namebuf [80]; + + + printk ("\n # %-15s%-30s%-15s%-15s%-15s\n", "Name", "Key of pointed object", "Hash", "Gen number", "Status"); + + deh = (struct reiserfs_de_head *)item; + + for (i = 0; i < I_ENTRY_COUNT (ih); i ++, deh ++) { + namelen = (i ? ((deh - 1)->deh_location) : ih->ih_item_len) - deh->deh_location; + name = item + deh->deh_location; + if (name[namelen-1] == 0) + namelen = strlen (name); + namebuf[0] = '"'; + if (namelen > sizeof (namebuf) - 3) { + strncpy (namebuf + 1, name, sizeof (namebuf) - 3); + namebuf[sizeof (namebuf) - 2] = '"'; + namebuf[sizeof (namebuf) - 1] = 0; + } else { + memcpy (namebuf + 1, name, namelen); + namebuf[namelen + 1] = '"'; + namebuf[namelen + 2] = 0; + } + + printk ("%d: %-15s%-15d%-15d%-15Ld%-15Ld(%s)\n", + i, namebuf, + deh->deh_dir_id, deh->deh_objectid, + GET_HASH_VALUE (deh_offset (deh)), GET_GENERATION_NUMBER ((deh_offset (deh))), + (de_hidden (deh)) ? "HIDDEN" : "VISIBLE"); + } +} + + +static void direntry_check_item (struct item_head * ih, char * item) +{ + int i; + struct reiserfs_de_head * deh; + + // FIXME: type something here! + deh = (struct reiserfs_de_head *)item; + for (i = 0; i < I_ENTRY_COUNT (ih); i ++, deh ++) { + ; + } +} + + + +#define DIRENTRY_VI_FIRST_DIRENTRY_ITEM 1 + +struct direntry_uarea { + int flags; + short entry_count; + short entry_sizes[1]; +}; + + +/* + * function returns old entry number in directory item in real node + * using new entry number in virtual item in virtual node */ +static inline int old_entry_num (int is_affected, int virtual_entry_num, int pos_in_item, int mode) +{ + if ( mode == M_INSERT || mode == M_DELETE) + return virtual_entry_num; + + if (!is_affected) + /* cut or paste is applied to another item */ + return virtual_entry_num; + + if (virtual_entry_num < pos_in_item) + return virtual_entry_num; + + if (mode == M_CUT) + return virtual_entry_num + 1; + +#ifdef CONFIG_REISERFS_CHECK + if (mode != M_PASTE || virtual_entry_num == 0) + reiserfs_panic (0, "vs-8015: old_entry_num: mode must be M_PASTE (mode = \'%c\'", mode); +#endif + + return virtual_entry_num - 1; +} + + + + +/* Create an array of sizes of directory entries for virtual + item. Return space used by an item. FIXME: no control over + consuming of space used by this item handler */ +static int direntry_create_vi (struct virtual_node * vn, + struct virtual_item * vi, + int is_affected, + int insert_size) +{ + struct direntry_uarea * dir_u = vi->vi_uarea; + int i, j; + int size = sizeof (struct direntry_uarea); + struct reiserfs_de_head * deh; + + vi->vi_index = TYPE_DIRENTRY; + + if (!(vi->vi_ih) || !vi->vi_item) + BUG (); + + + dir_u->flags = 0; + if (le_ih_k_offset (vi->vi_ih) == DOT_OFFSET) + dir_u->flags |= DIRENTRY_VI_FIRST_DIRENTRY_ITEM; + + deh = (struct reiserfs_de_head *)(vi->vi_item); + + + /* virtual directory item have this amount of entry after */ + dir_u->entry_count = ih_entry_count (vi->vi_ih) + + ((is_affected) ? ((vn->vn_mode == M_CUT) ? -1 : + (vn->vn_mode == M_PASTE ? 1 : 0)) : 0); + + for (i = 0; i < dir_u->entry_count; i ++) { + j = old_entry_num (is_affected, i, vn->vn_pos_in_item, vn->vn_mode); + dir_u->entry_sizes[i] = (j ? le16_to_cpu (deh[j - 1].deh_location) : le16_to_cpu (vi->vi_ih->ih_item_len)) - + le16_to_cpu (deh[j].deh_location) + DEH_SIZE; + } + + size += (dir_u->entry_count * sizeof (short)); + + /* set size of pasted entry */ + if (is_affected && vn->vn_mode == M_PASTE) + dir_u->entry_sizes[vn->vn_pos_in_item] = insert_size; + + +#ifdef CONFIG_REISERFS_CHECK + /* compare total size of entries with item length */ + { + int k, l; + + l = 0; + for (k = 0; k < dir_u->entry_count; k ++) + l += dir_u->entry_sizes[k]; + + if (l + IH_SIZE != vi->vi_item_len + + ((is_affected && (vn->vn_mode == M_PASTE || vn->vn_mode == M_CUT)) ? insert_size : 0) ) { + reiserfs_panic (0, "vs-8025: set_entry_sizes: (mode==%c, insert_size==%d), invalid length of directory item", + vn->vn_mode, insert_size); + } + } +#endif + + return size; + + +} + + +// +// return number of entries which may fit into specified amount of +// free space, or -1 if free space is not enough even for 1 entry +// +static int direntry_check_left (struct virtual_item * vi, int free, + int start_skip, int end_skip) +{ + int i; + int entries = 0; + struct direntry_uarea * dir_u = vi->vi_uarea; + + for (i = start_skip; i < dir_u->entry_count - end_skip; i ++) { + if (dir_u->entry_sizes[i] > free) + /* i-th entry doesn't fit into the remaining free space */ + break; + + free -= dir_u->entry_sizes[i]; + entries ++; + } + + if (entries == dir_u->entry_count) { + printk ("free spze %d, entry_count %d\n", free, dir_u->entry_count); + BUG (); + } + + /* "." and ".." can not be separated from each other */ + if (start_skip == 0 && (dir_u->flags & DIRENTRY_VI_FIRST_DIRENTRY_ITEM) && entries < 2) + entries = 0; + + return entries ?: -1; +} + + +static int direntry_check_right (struct virtual_item * vi, int free) +{ + int i; + int entries = 0; + struct direntry_uarea * dir_u = vi->vi_uarea; + + for (i = dir_u->entry_count - 1; i >= 0; i --) { + if (dir_u->entry_sizes[i] > free) + /* i-th entry doesn't fit into the remaining free space */ + break; + + free -= dir_u->entry_sizes[i]; + entries ++; + } + if (entries == dir_u->entry_count) + BUG (); + + /* "." and ".." can not be separated from each other */ + if ((dir_u->flags & DIRENTRY_VI_FIRST_DIRENTRY_ITEM) && entries > dir_u->entry_count - 2) + entries = dir_u->entry_count - 2; + + return entries ?: -1; +} + + +/* sum of entry sizes between from-th and to-th entries including both edges */ +static int direntry_part_size (struct virtual_item * vi, int first, int count) +{ + int i, retval; + int from, to; + struct direntry_uarea * dir_u = vi->vi_uarea; + + retval = 0; + if (first == 0) + from = 0; + else + from = dir_u->entry_count - count; + to = from + count - 1; + + for (i = from; i <= to; i ++) + retval += dir_u->entry_sizes[i]; + + return retval; +} + +static int direntry_unit_num (struct virtual_item * vi) +{ + struct direntry_uarea * dir_u = vi->vi_uarea; + + return dir_u->entry_count; +} + + + +static void direntry_print_vi (struct virtual_item * vi) +{ + int i; + struct direntry_uarea * dir_u = vi->vi_uarea; + + reiserfs_warning ("DIRENTRY, index %d, type 0x%x, %h, flags 0x%x\n", + vi->vi_index, vi->vi_type, vi->vi_ih, dir_u->flags); + printk ("%d entries: ", dir_u->entry_count); + for (i = 0; i < dir_u->entry_count; i ++) + printk ("%d ", dir_u->entry_sizes[i]); + printk ("\n"); +} + +struct item_operations direntry_ops = { + direntry_bytes_number, + direntry_decrement_key, + direntry_is_left_mergeable, + direntry_print_item, + direntry_check_item, + + direntry_create_vi, + direntry_check_left, + direntry_check_right, + direntry_part_size, + direntry_unit_num, + direntry_print_vi +}; + + +////////////////////////////////////////////////////////////////////////////// +// +// +#if ! (TYPE_STAT_DATA == 0 && TYPE_INDIRECT == 1 && TYPE_DIRECT == 2 && TYPE_DIRENTRY == 3) + do not compile +#endif + +struct item_operations * item_ops [4] = { + &stat_data_ops, + &indirect_ops, + &direct_ops, + &direntry_ops +}; + + + + diff --git a/fs/reiserfs/journal.c b/fs/reiserfs/journal.c new file mode 100644 index 000000000..01a597156 --- /dev/null +++ b/fs/reiserfs/journal.c @@ -0,0 +1,3215 @@ +/* +** Write ahead logging implementation copyright Chris Mason 2000 +** +** The background commits make this code very interelated, and +** overly complex. I need to rethink things a bit....The major players: +** +** journal_begin -- call with the number of blocks you expect to log. +** If the current transaction is too +** old, it will block until the current transaction is +** finished, and then start a new one. +** Usually, your transaction will get joined in with +** previous ones for speed. +** +** journal_join -- same as journal_begin, but won't block on the current +** transaction regardless of age. Don't ever call +** this. Ever. There are only two places it should be +** called from, and they are both inside this file. +** +** journal_mark_dirty -- adds blocks into this transaction. clears any flags +** that might make them get sent to disk +** and then marks them BH_JDirty. Puts the buffer head +** into the current transaction hash. +** +** journal_end -- if the current transaction is batchable, it does nothing +** otherwise, it could do an async/synchronous commit, or +** a full flush of all log and real blocks in the +** transaction. +** +** flush_old_commits -- if the current transaction is too old, it is ended and +** commit blocks are sent to disk. Forces commit blocks +** to disk for all backgrounded commits that have been +** around too long. +** -- Note, if you call this as an immediate flush from +** from within kupdate, it will ignore the immediate flag +** +** The commit thread -- a writer process for async commits. It allows a +** a process to request a log flush on a task queue. +** the commit will happen once the commit thread wakes up. +** The benefit here is the writer (with whatever +** related locks it has) doesn't have to wait for the +** log blocks to hit disk if it doesn't want to. +*/ + +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <asm/uaccess.h> +#include <asm/system.h> + +#include <linux/sched.h> +#include <asm/semaphore.h> + +#include <linux/vmalloc.h> +#include <linux/reiserfs_fs.h> + +#include <linux/kernel.h> +#include <linux/errno.h> +#include <linux/fcntl.h> +#include <linux/locks.h> +#include <linux/stat.h> +#include <linux/string.h> +#include <linux/smp_lock.h> + +#else + +#include "nokernel.h" + +#endif + + +/* the number of mounted filesystems. This is used to decide when to +** start and kill the commit thread +*/ +static int reiserfs_mounted_fs_count = 0 ; + +/* wake this up when you add something to the commit thread task queue */ +DECLARE_WAIT_QUEUE_HEAD(reiserfs_commit_thread_wait) ; + +/* wait on this if you need to be sure you task queue entries have been run */ +static DECLARE_WAIT_QUEUE_HEAD(reiserfs_commit_thread_done) ; +DECLARE_TASK_QUEUE(reiserfs_commit_thread_tq) ; + +#define JOURNAL_TRANS_HALF 1018 /* must be correct to keep the desc and commit structs at 4k */ + +/* cnode stat bits. Move these into reiserfs_fs.h */ + +#define BLOCK_FREED 2 /* this block was freed, and can't be written. */ +#define BLOCK_FREED_HOLDER 3 /* this block was freed during this transaction, and can't be written */ + +#define BLOCK_NEEDS_FLUSH 4 /* used in flush_journal_list */ + +/* flags for do_journal_end */ +#define FLUSH_ALL 1 /* flush commit and real blocks */ +#define COMMIT_NOW 2 /* end and commit this transaction */ +#define WAIT 4 /* wait for the log blocks to hit the disk*/ + +/* state bits for the journal */ +#define WRITERS_BLOCKED 1 /* set when new writers not allowed */ + +static int do_journal_end(struct reiserfs_transaction_handle *,struct super_block *,unsigned long nblocks,int flags) ; +static int flush_journal_list(struct super_block *s, struct reiserfs_journal_list *jl, int flushall) ; +static int flush_commit_list(struct super_block *s, struct reiserfs_journal_list *jl, int flushall) ; +static int can_dirty(struct reiserfs_journal_cnode *cn) ; + +static void init_journal_hash(struct super_block *p_s_sb) { + memset(SB_JOURNAL(p_s_sb)->j_hash_table, 0, JOURNAL_HASH_SIZE * sizeof(struct reiserfs_journal_cnode *)) ; +} + +/* +** clears BH_Dirty and sticks the buffer on the clean list. Called because I can't allow refile_buffer to +** make schedule happen after I've freed a block. Look at remove_from_transaction and journal_mark_freed for +** more details. +*/ +static int reiserfs_clean_and_file_buffer(struct buffer_head *bh) { + if (bh) { + clear_bit(BH_Dirty, &bh->b_state) ; +#if 0 + if (bh->b_list != BUF_CLEAN) { + reiserfs_file_buffer(bh, BUF_CLEAN) ; + } +#endif + } + return 0 ; +} + +static struct reiserfs_bitmap_node * +allocate_bitmap_node(struct super_block *p_s_sb) { + struct reiserfs_bitmap_node *bn ; + static int id = 0 ; + + bn = kmalloc(sizeof(struct reiserfs_bitmap_node), GFP_BUFFER) ; + if (!bn) { + return NULL ; + } + bn->data = kmalloc(p_s_sb->s_blocksize, GFP_BUFFER) ; + if (!bn->data) { + kfree(bn) ; + return NULL ; + } + bn->id = id++ ; + memset(bn->data, 0, p_s_sb->s_blocksize) ; + INIT_LIST_HEAD(&bn->list) ; + return bn ; +} + +static struct reiserfs_bitmap_node * +get_bitmap_node(struct super_block *p_s_sb) { + struct reiserfs_bitmap_node *bn = NULL; + struct list_head *entry = SB_JOURNAL(p_s_sb)->j_bitmap_nodes.next ; + + SB_JOURNAL(p_s_sb)->j_used_bitmap_nodes++ ; +repeat: + + if(entry != &SB_JOURNAL(p_s_sb)->j_bitmap_nodes) { + bn = list_entry(entry, struct reiserfs_bitmap_node, list) ; + list_del(entry) ; + memset(bn->data, 0, p_s_sb->s_blocksize) ; + SB_JOURNAL(p_s_sb)->j_free_bitmap_nodes-- ; + return bn ; + } + bn = allocate_bitmap_node(p_s_sb) ; + if (!bn) { + current->policy = SCHED_YIELD ; + schedule() ; + goto repeat ; + } + return bn ; +} +static inline void free_bitmap_node(struct super_block *p_s_sb, + struct reiserfs_bitmap_node *bn) { + SB_JOURNAL(p_s_sb)->j_used_bitmap_nodes-- ; + if (SB_JOURNAL(p_s_sb)->j_free_bitmap_nodes > REISERFS_MAX_BITMAP_NODES) { + kfree(bn->data) ; + kfree(bn) ; + } else { + list_add(&bn->list, &SB_JOURNAL(p_s_sb)->j_bitmap_nodes) ; + SB_JOURNAL(p_s_sb)->j_free_bitmap_nodes++ ; + } +} + +static void allocate_bitmap_nodes(struct super_block *p_s_sb) { + int i ; + struct reiserfs_bitmap_node *bn = NULL ; + for (i = 0 ; i < REISERFS_MIN_BITMAP_NODES ; i++) { + bn = allocate_bitmap_node(p_s_sb) ; + if (bn) { + list_add(&bn->list, &SB_JOURNAL(p_s_sb)->j_bitmap_nodes) ; + SB_JOURNAL(p_s_sb)->j_free_bitmap_nodes++ ; + } else { + break ; // this is ok, we'll try again when more are needed + } + } +} + +static int set_bit_in_list_bitmap(struct super_block *p_s_sb, int block, + struct reiserfs_list_bitmap *jb) { + int bmap_nr = block / (p_s_sb->s_blocksize << 3) ; + int bit_nr = block % (p_s_sb->s_blocksize << 3) ; + + if (!jb->bitmaps[bmap_nr]) { + jb->bitmaps[bmap_nr] = get_bitmap_node(p_s_sb) ; + } + set_bit(bit_nr, jb->bitmaps[bmap_nr]->data) ; + return 0 ; +} + +static void cleanup_bitmap_list(struct super_block *p_s_sb, + struct reiserfs_list_bitmap *jb) { + int i; + for (i = 0 ; i < SB_BMAP_NR(p_s_sb) ; i++) { + if (jb->bitmaps[i]) { + free_bitmap_node(p_s_sb, jb->bitmaps[i]) ; + jb->bitmaps[i] = NULL ; + } + } +} + +/* +** only call this on FS unmount. +*/ +static int free_list_bitmaps(struct super_block *p_s_sb, + struct reiserfs_list_bitmap *jb_array) { + int i ; + struct reiserfs_list_bitmap *jb ; + for (i = 0 ; i < JOURNAL_NUM_BITMAPS ; i++) { + jb = jb_array + i ; + jb->journal_list = NULL ; + cleanup_bitmap_list(p_s_sb, jb) ; + vfree(jb->bitmaps) ; + jb->bitmaps = NULL ; + } + return 0; +} + +static int free_bitmap_nodes(struct super_block *p_s_sb) { + struct list_head *next = SB_JOURNAL(p_s_sb)->j_bitmap_nodes.next ; + struct reiserfs_bitmap_node *bn ; + + while(next != &SB_JOURNAL(p_s_sb)->j_bitmap_nodes) { + bn = list_entry(next, struct reiserfs_bitmap_node, list) ; + list_del(next) ; + kfree(bn->data) ; + kfree(bn) ; + next = SB_JOURNAL(p_s_sb)->j_bitmap_nodes.next ; + SB_JOURNAL(p_s_sb)->j_free_bitmap_nodes-- ; + } + + return 0 ; +} + +/* +** get memory for JOURNAL_NUM_BITMAPS worth of bitmaps. +** jb_array is the array to be filled in. +*/ +int reiserfs_allocate_list_bitmaps(struct super_block *p_s_sb, + struct reiserfs_list_bitmap *jb_array, + int bmap_nr) { + int i ; + int failed = 0 ; + struct reiserfs_list_bitmap *jb ; + int mem = bmap_nr * sizeof(struct reiserfs_bitmap_node *) ; + + for (i = 0 ; i < JOURNAL_NUM_BITMAPS ; i++) { + jb = jb_array + i ; + jb->journal_list = NULL ; + jb->bitmaps = vmalloc( mem ) ; + if (!jb->bitmaps) { + reiserfs_warning("clm-2000, unable to allocate bitmaps for journal lists\n") ; + failed = 1; + break ; + } + memset(jb->bitmaps, 0, mem) ; + } + if (failed) { + free_list_bitmaps(p_s_sb, jb_array) ; + return -1 ; + } + return 0 ; +} + +/* +** find an available list bitmap. If you can't find one, flush a commit list +** and try again +*/ +static struct reiserfs_list_bitmap * +get_list_bitmap(struct super_block *p_s_sb, struct reiserfs_journal_list *jl) { + int i,j ; + struct reiserfs_list_bitmap *jb = NULL ; + + for (j = 0 ; j < (JOURNAL_NUM_BITMAPS * 3) ; j++) { + i = SB_JOURNAL(p_s_sb)->j_list_bitmap_index ; + SB_JOURNAL(p_s_sb)->j_list_bitmap_index = (i + 1) % JOURNAL_NUM_BITMAPS ; + jb = SB_JOURNAL(p_s_sb)->j_list_bitmap + i ; + if (SB_JOURNAL(p_s_sb)->j_list_bitmap[i].journal_list) { + flush_commit_list(p_s_sb, SB_JOURNAL(p_s_sb)->j_list_bitmap[i].journal_list, 1) ; + if (!SB_JOURNAL(p_s_sb)->j_list_bitmap[i].journal_list) { + break ; + } + } else { + break ; + } + } + if (jb->journal_list) { /* double check to make sure if flushed correctly */ + return NULL ; + } + jb->journal_list = jl ; + return jb ; +} + +/* +** allocates a new chunk of X nodes, and links them all together as a list. +** Uses the cnode->next and cnode->prev pointers +** returns NULL on failure +*/ +static struct reiserfs_journal_cnode *allocate_cnodes(int num_cnodes) { + struct reiserfs_journal_cnode *head ; + int i ; + if (num_cnodes <= 0) { + return NULL ; + } + head = vmalloc(num_cnodes * sizeof(struct reiserfs_journal_cnode)) ; + if (!head) { + return NULL ; + } + memset(head, 0, num_cnodes * sizeof(struct reiserfs_journal_cnode)) ; + head[0].prev = NULL ; + head[0].next = head + 1 ; + for (i = 1 ; i < num_cnodes; i++) { + head[i].prev = head + (i - 1) ; + head[i].next = head + (i + 1) ; /* if last one, overwrite it after the if */ + } + head[num_cnodes -1].next = NULL ; + return head ; +} + +/* +** pulls a cnode off the free list, or returns NULL on failure +*/ +static struct reiserfs_journal_cnode *get_cnode(struct super_block *p_s_sb) { + struct reiserfs_journal_cnode *cn ; + + reiserfs_check_lock_depth("get_cnode") ; + + if (SB_JOURNAL(p_s_sb)->j_cnode_free <= 0) { + return NULL ; + } + SB_JOURNAL(p_s_sb)->j_cnode_used++ ; + SB_JOURNAL(p_s_sb)->j_cnode_free-- ; + cn = SB_JOURNAL(p_s_sb)->j_cnode_free_list ; + if (!cn) { + return cn ; + } + if (cn->next) { + cn->next->prev = NULL ; + } + SB_JOURNAL(p_s_sb)->j_cnode_free_list = cn->next ; + memset(cn, 0, sizeof(struct reiserfs_journal_cnode)) ; + return cn ; +} + +/* +** returns a cnode to the free list +*/ +static void free_cnode(struct super_block *p_s_sb, struct reiserfs_journal_cnode *cn) { + + reiserfs_check_lock_depth("free_cnode") ; + + SB_JOURNAL(p_s_sb)->j_cnode_used-- ; + SB_JOURNAL(p_s_sb)->j_cnode_free++ ; + /* memset(cn, 0, sizeof(struct reiserfs_journal_cnode)) ; */ + cn->next = SB_JOURNAL(p_s_sb)->j_cnode_free_list ; + if (SB_JOURNAL(p_s_sb)->j_cnode_free_list) { + SB_JOURNAL(p_s_sb)->j_cnode_free_list->prev = cn ; + } + cn->prev = NULL ; /* not needed with the memset, but I might kill the memset, and forget to do this */ + SB_JOURNAL(p_s_sb)->j_cnode_free_list = cn ; +} + +static int clear_prepared_bits(struct buffer_head *bh) { + clear_bit(BH_JPrepared, &bh->b_state) ; + return 0 ; +} + +/* buffer is in current transaction */ +inline int buffer_journaled(struct buffer_head *bh) { + if (bh) + return test_bit(BH_JDirty, &bh->b_state) ; + else + return 0 ; +} + +/* disk block was taken off free list before being in a finished transation, or written to disk +** journal_new blocks can be reused immediately, for any purpose +*/ +inline int buffer_journal_new(struct buffer_head *bh) { + if (bh) + return test_bit(BH_JNew, &bh->b_state) ; + else + return 0 ; +} + +inline int mark_buffer_journal_new(struct buffer_head *bh) { + if (bh) { + set_bit(BH_JNew, &bh->b_state) ; + } + return 0 ; +} + +inline int mark_buffer_not_journaled(struct buffer_head *bh) { + if (bh) + clear_bit(BH_JDirty, &bh->b_state) ; + return 0 ; +} + +/* utility function to force a BUG if it is called without the big +** kernel lock held. caller is the string printed just before calling BUG() +*/ +void reiserfs_check_lock_depth(char *caller) { +#ifdef CONFIG_SMP + if (current->lock_depth < 0) { + printk("%s called without kernel lock held\n", caller) ; + show_reiserfs_locks() ; + BUG() ; + } +#else + ; +#endif +} + +/* return a cnode with same dev, block number and size in table, or null if not found */ +static inline struct reiserfs_journal_cnode *get_journal_hash_dev(struct reiserfs_journal_cnode **table, + kdev_t dev,long bl,int size) { + struct reiserfs_journal_cnode *cn ; + cn = journal_hash(table, dev, bl) ; + while(cn) { + if ((cn->blocknr == bl) && (cn->dev == dev)) + return cn ; + cn = cn->hnext ; + } + return (struct reiserfs_journal_cnode *)0 ; +} + +/* returns a cnode with same size, block number and dev as bh in the current transaction hash. NULL if not found */ +static inline struct reiserfs_journal_cnode *get_journal_hash(struct super_block *p_s_sb, struct buffer_head *bh) { + struct reiserfs_journal_cnode *cn ; + if (bh) { + cn = get_journal_hash_dev(SB_JOURNAL(p_s_sb)->j_hash_table, bh->b_dev, bh->b_blocknr, bh->b_size) ; + } + else { + return (struct reiserfs_journal_cnode *)0 ; + } + return cn ; +} + +/* once upon a time, the journal would deadlock. a lot. Now, when +** CONFIG_REISERFS_CHECK is defined, anytime someone enters a +** transaction, it pushes itself into this ugly static list, and pops +** itself off before calling journal_end. I made a SysRq key to dump +** the list, and tell me what the writers are when I'm deadlocked. */ + + /* are you depending on the compiler + to optimize this function away + everywhere it is called? It is not + obvious how this works, but I + suppose debugging code need not be + clear. -Hans */ +static char *journal_writers[512] ; +int push_journal_writer(char *s) { +#ifdef CONFIG_REISERFS_CHECK + int i ; + for (i = 0 ; i < 512 ; i++) { + if (!journal_writers[i]) { + journal_writers[i] = s ; + return i ; + } + } + return -1 ; +#else + return 0 ; +#endif +} +int pop_journal_writer(int index) { +#ifdef CONFIG_REISERFS_CHECK + if (index >= 0) { + journal_writers[index] = NULL ; + } +#endif + return 0 ; +} + +int dump_journal_writers(void) { + int i ; + for (i = 0 ; i < 512 ; i++) { + if (journal_writers[i]) { + printk("%d: %s\n", i, journal_writers[i]) ; + } + } + return 0 ; +} + +/* +** this actually means 'can this block be reallocated yet?'. If you set search_all, a block can only be allocated +** if it is not in the current transaction, was not freed by the current transaction, and has no chance of ever +** being overwritten by a replay after crashing. +** +** If you don't set search_all, a block can only be allocated if it is not in the current transaction. Since deleting +** a block removes it from the current transaction, this case should never happen. If you don't set search_all, make +** sure you never write the block without logging it. +** +** next_zero_bit is a suggestion about the next block to try for find_forward. +** when bl is rejected because it is set in a journal list bitmap, we search +** for the next zero bit in the bitmap that rejected bl. Then, we return that +** through next_zero_bit for find_forward to try. +** +** Just because we return something in next_zero_bit does not mean we won't +** reject it on the next call to reiserfs_in_journal +** +*/ +int reiserfs_in_journal(struct super_block *p_s_sb, kdev_t dev, + unsigned long bl, int size, int search_all, + unsigned long *next_zero_bit) { + struct reiserfs_journal_cnode *cn ; + struct reiserfs_list_bitmap *jb ; + int i ; + int bmap_nr = bl / (p_s_sb->s_blocksize << 3) ; + int bit_nr = bl % (p_s_sb->s_blocksize << 3) ; + int tmp_bit ; + + *next_zero_bit = 0 ; /* always start this at zero. */ + + /* we aren't logging all blocks are safe for reuse */ + if (reiserfs_dont_log(p_s_sb)) { + return 0 ; + } + + /* If we aren't doing a search_all, this is a metablock, and it will be logged before use. + ** if we crash before the transaction that freed it commits, this transaction won't + ** have committed either, and the block will never be written + */ + if (search_all) { + for (i = 0 ; i < JOURNAL_NUM_BITMAPS ; i++) { + jb = SB_JOURNAL(p_s_sb)->j_list_bitmap + i ; + if (jb->journal_list && jb->bitmaps[bmap_nr] && + test_bit(bit_nr, jb->bitmaps[bmap_nr]->data)) { + tmp_bit = find_next_zero_bit((unsigned long *) + (jb->bitmaps[bmap_nr]->data), + p_s_sb->s_blocksize << 3, bit_nr+1) ; + *next_zero_bit = bmap_nr * (p_s_sb->s_blocksize << 3) + tmp_bit ; + return 1 ; + } + } + } + + /* is it in any old transactions? */ + if (search_all && (cn = get_journal_hash_dev(SB_JOURNAL(p_s_sb)->j_list_hash_table, dev,bl,size))) { + return 1; + } + + /* is it in the current transaction. This should never happen */ + if ((cn = get_journal_hash_dev(SB_JOURNAL(p_s_sb)->j_hash_table, dev,bl,size))) { + return 1; + } + + /* safe for reuse */ + return 0 ; +} + +/* insert cn into table +*/ +inline void insert_journal_hash(struct reiserfs_journal_cnode **table, struct reiserfs_journal_cnode *cn) { + struct reiserfs_journal_cnode *cn_orig ; + + cn_orig = journal_hash(table, cn->dev, cn->blocknr) ; + cn->hnext = cn_orig ; + cn->hprev = NULL ; + if (cn_orig) { + cn_orig->hprev = cn ; + } + journal_hash(table, cn->dev, cn->blocknr) = cn ; +} + +/* lock the current transaction */ +inline static void lock_journal(struct super_block *p_s_sb) { + while(atomic_read(&(SB_JOURNAL(p_s_sb)->j_wlock)) > 0) { + sleep_on(&(SB_JOURNAL(p_s_sb)->j_wait)) ; + } + atomic_set(&(SB_JOURNAL(p_s_sb)->j_wlock), 1) ; +} + +/* unlock the current transaction */ +inline static void unlock_journal(struct super_block *p_s_sb) { + atomic_dec(&(SB_JOURNAL(p_s_sb)->j_wlock)) ; + wake_up(&(SB_JOURNAL(p_s_sb)->j_wait)) ; +} + +/* +** this used to be much more involved, and I'm keeping it just in case things get ugly again. +** it gets called by flush_commit_list, and cleans up any data stored about blocks freed during a +** transaction. +*/ +static void cleanup_freed_for_journal_list(struct super_block *p_s_sb, struct reiserfs_journal_list *jl) { + + struct reiserfs_list_bitmap *jb = jl->j_list_bitmap ; + if (jb) { + cleanup_bitmap_list(p_s_sb, jb) ; + } + jl->j_list_bitmap->journal_list = NULL ; + jl->j_list_bitmap = NULL ; +} + +/* +** if this journal list still has commit blocks unflushed, send them to disk. +** +** log areas must be flushed in order (transaction 2 can't commit before transaction 1) +** Before the commit block can by written, every other log block must be safely on disk +** +*/ +static int flush_commit_list(struct super_block *s, struct reiserfs_journal_list *jl, int flushall) { + int i, count ; + int index = 0 ; + int bn ; + int retry_count = 0 ; + int orig_commit_left = 0 ; + struct buffer_head *tbh = NULL ; + struct reiserfs_journal_list *other_jl ; + + reiserfs_check_lock_depth("flush_commit_list") ; + + if (atomic_read(&jl->j_older_commits_done)) { + return 0 ; + } + + /* before we can put our commit blocks on disk, we have to make sure everyone older than + ** us is on disk too + */ + if (jl->j_len <= 0) { + return 0 ; + } + if (flushall) { + /* we _must_ make sure the transactions are committed in order. Start with the + ** index after this one, wrap all the way around + */ + index = (jl - SB_JOURNAL_LIST(s)) + 1 ; + for (i = 0 ; i < JOURNAL_LIST_COUNT ; i++) { + other_jl = SB_JOURNAL_LIST(s) + ( (index + i) % JOURNAL_LIST_COUNT) ; + if (other_jl && other_jl != jl && other_jl->j_len > 0 && other_jl->j_trans_id > 0 && + other_jl->j_trans_id <= jl->j_trans_id && (atomic_read(&(jl->j_older_commits_done)) == 0)) { + flush_commit_list(s, other_jl, 0) ; + } + } + } + + count = 0 ; + /* don't flush the commit list for the current transactoin */ + if (jl == ((SB_JOURNAL_LIST(s) + SB_JOURNAL_LIST_INDEX(s)))) { + return 0 ; + } + + /* make sure nobody is trying to flush this one at the same time */ + if (atomic_read(&(jl->j_commit_flushing))) { + sleep_on(&(jl->j_commit_wait)) ; + if (flushall) { + atomic_set(&(jl->j_older_commits_done), 1) ; + } + return 0 ; + } + + /* this commit is done, exit */ + if (atomic_read(&(jl->j_commit_left)) <= 0) { + if (flushall) { + atomic_set(&(jl->j_older_commits_done), 1) ; + } + return 0 ; + } + /* keeps others from flushing while we are flushing */ + atomic_set(&(jl->j_commit_flushing), 1) ; + + + if (jl->j_len > JOURNAL_TRANS_MAX) { + reiserfs_panic(s, "journal-512: flush_commit_list: length is %lu, list number %d\n", jl->j_len, jl - SB_JOURNAL_LIST(s)) ; + return 0 ; + } + + orig_commit_left = atomic_read(&(jl->j_commit_left)) ; + + /* start by checking all the commit blocks in this transaction. + ** Add anyone not on disk into tbh. Stop checking once commit_left <= 1, because that means we + ** only have the commit block left + */ +retry: + count = 0 ; + for (i = 0 ; atomic_read(&(jl->j_commit_left)) > 1 && i < (jl->j_len + 1) ; i++) { /* everything but commit_bh */ + bn = reiserfs_get_journal_block(s) + (jl->j_start+i) % JOURNAL_BLOCK_COUNT; + tbh = get_hash_table(s->s_dev, bn, s->s_blocksize) ; + +/* kill this sanity check */ +if (count > (orig_commit_left + 2)) { +reiserfs_panic(s, "journal-539: flush_commit_list: BAD count(%d) > orig_commit_left(%d)!\n", count, orig_commit_left) ; +} + if (tbh) { + if (buffer_locked(tbh)) { /* wait on it, redo it just to make sure */ + wait_on_buffer(tbh) ; + if (!buffer_uptodate(tbh)) { + reiserfs_panic(s, "journal-584, buffer write failed\n") ; + } + } + if (buffer_dirty(tbh)) { + printk("journal-569: flush_commit_list, block already dirty!\n") ; + } else { + mark_buffer_dirty(tbh) ; + } + ll_rw_block(WRITE, 1, &tbh) ; + count++ ; + atomic_dec(&(tbh->b_count)) ; /* once for our get_hash */ + } + } + + /* wait on everyone in tbh before writing commit block*/ + if (count > 0) { + for (i = 0 ; atomic_read(&(jl->j_commit_left)) > 1 && + i < (jl->j_len + 1) ; i++) { /* everything but commit_bh */ + bn = reiserfs_get_journal_block(s) + (jl->j_start + i) % JOURNAL_BLOCK_COUNT ; + tbh = get_hash_table(s->s_dev, bn, s->s_blocksize) ; + + wait_on_buffer(tbh) ; + if (!buffer_uptodate(tbh)) { + reiserfs_panic(s, "journal-601, buffer write failed\n") ; + } + atomic_dec(&(tbh->b_count)) ; /* once for our get_hash */ + bforget(tbh) ; /* once due to original getblk in do_journal_end */ + atomic_dec(&(jl->j_commit_left)) ; + } + } + + if (atomic_read(&(jl->j_commit_left)) != 1) { /* just the commit_bh left, flush it without calling getblk for everyone */ + if (retry_count < 2) { + printk("journal-582: flush_commit_list, not all log blocks on disk yet, trying again\n") ; + retry_count++ ; + goto retry; + } + reiserfs_panic(s, "journal-563: flush_commit_list: BAD, j_commit_left is %lu, should be 1\n", + atomic_read(&(jl->j_commit_left))); + } + + mark_buffer_dirty(jl->j_commit_bh) ; + ll_rw_block(WRITE, 1, &(jl->j_commit_bh)) ; + wait_on_buffer(jl->j_commit_bh) ; + if (!buffer_uptodate(jl->j_commit_bh)) { + reiserfs_panic(s, "journal-615: buffer write failed\n") ; + } + atomic_dec(&(jl->j_commit_left)) ; + bforget(jl->j_commit_bh) ; + + /* now, every commit block is on the disk. It is safe to allow blocks freed during this transaction to be reallocated */ + cleanup_freed_for_journal_list(s, jl) ; + + if (flushall) { + atomic_set(&(jl->j_older_commits_done), 1) ; + } + atomic_set(&(jl->j_commit_flushing), 0) ; + wake_up(&(jl->j_commit_wait)) ; + return 0 ; +} + +/* +** flush_journal_list frequently needs to find a newer transaction for a given block. This does that, or +** returns NULL if it can't find anything +*/ +static struct reiserfs_journal_list *find_newer_jl_for_cn(struct reiserfs_journal_cnode *cn) { + kdev_t dev = cn->dev; + unsigned long blocknr = cn->blocknr ; + + cn = cn->hprev ; + while(cn) { + if (cn->dev == dev && cn->blocknr == blocknr && cn->jlist) { + return cn->jlist ; + } + cn = cn->hprev ; + } + return NULL ; +} + + +/* +** once all the real blocks have been flushed, it is safe to remove them from the +** journal list for this transaction. Aside from freeing the cnode, this also allows the +** block to be reallocated for data blocks if it had been deleted. +*/ +static void remove_all_from_journal_list(struct super_block *p_s_sb, struct reiserfs_journal_list *jl, int debug) { + struct buffer_head fake_bh ; + struct reiserfs_journal_cnode *cn, *last ; + cn = jl->j_realblock ; + + /* which is better, to lock once around the whole loop, or + ** to lock for each call to remove_from_journal_list? + */ + while(cn) { + if (cn->blocknr != 0) { + if (debug) { + printk("block %lu, bh is %d, state %d\n", cn->blocknr, cn->bh ? 1: 0, + cn->state) ; + } + fake_bh.b_blocknr = cn->blocknr ; + fake_bh.b_dev = cn->dev ; + cn->state = 0 ; + remove_from_journal_list(p_s_sb, jl, &fake_bh, 1) ; + } + last = cn ; + cn = cn->next ; + free_cnode(p_s_sb, last) ; + } + jl->j_realblock = NULL ; +} + +/* +** if this timestamp is greater than the timestamp we wrote last to the header block, write it to the header block. +** once this is done, I can safely say the log area for this transaction won't ever be replayed, and I can start +** releasing blocks in this transaction for reuse as data blocks. +** called by flush_journal_list, before it calls remove_all_from_journal_list +** +*/ +static int update_journal_header_block(struct super_block *p_s_sb, unsigned long offset, unsigned long trans_id) { + struct reiserfs_journal_header *jh ; + if (trans_id >= SB_JOURNAL(p_s_sb)->j_last_flush_trans_id) { + if (buffer_locked((SB_JOURNAL(p_s_sb)->j_header_bh))) { + wait_on_buffer((SB_JOURNAL(p_s_sb)->j_header_bh)) ; + if (!buffer_uptodate(SB_JOURNAL(p_s_sb)->j_header_bh)) { + reiserfs_panic(p_s_sb, "journal-699: buffer write failed\n") ; + } + } + SB_JOURNAL(p_s_sb)->j_last_flush_trans_id = trans_id ; + SB_JOURNAL(p_s_sb)->j_first_unflushed_offset = offset ; + jh = (struct reiserfs_journal_header *)(SB_JOURNAL(p_s_sb)->j_header_bh->b_data) ; + jh->j_last_flush_trans_id = cpu_to_le32(trans_id) ; + jh->j_first_unflushed_offset = cpu_to_le32(offset) ; + jh->j_mount_id = cpu_to_le32(SB_JOURNAL(p_s_sb)->j_mount_id) ; + set_bit(BH_Dirty, &(SB_JOURNAL(p_s_sb)->j_header_bh->b_state)) ; + ll_rw_block(WRITE, 1, &(SB_JOURNAL(p_s_sb)->j_header_bh)) ; + wait_on_buffer((SB_JOURNAL(p_s_sb)->j_header_bh)) ; + if (!buffer_uptodate(SB_JOURNAL(p_s_sb)->j_header_bh)) { + reiserfs_panic(p_s_sb, "journal-712: buffer write failed\n") ; + } + } + return 0 ; +} + +/* +** flush any and all journal lists older than you are +** can only be called from flush_journal_list +*/ +static int flush_older_journal_lists(struct super_block *p_s_sb, struct reiserfs_journal_list *jl, unsigned long trans_id) { + int i, index ; + struct reiserfs_journal_list *other_jl ; + + index = jl - SB_JOURNAL_LIST(p_s_sb) ; + for (i = 0 ; i < JOURNAL_LIST_COUNT ; i++) { + other_jl = SB_JOURNAL_LIST(p_s_sb) + ((index + i) % JOURNAL_LIST_COUNT) ; + if (other_jl && other_jl->j_len > 0 && + other_jl->j_trans_id > 0 && + other_jl->j_trans_id < trans_id && + other_jl != jl) { + /* do not flush all */ + flush_journal_list(p_s_sb, other_jl, 0) ; + } + } + return 0 ; +} + +static void reiserfs_end_buffer_io_sync(struct buffer_head *bh, int uptodate) { + if (buffer_journaled(bh)) { + reiserfs_warning("clm-2084: pinned buffer %u:%s sent to disk\n", + bh->b_blocknr, kdevname(bh->b_dev)) ; + } + mark_buffer_uptodate(bh, uptodate) ; + unlock_buffer(bh) ; +} +static void submit_logged_buffer(struct buffer_head *bh) { + lock_buffer(bh) ; + bh->b_end_io = reiserfs_end_buffer_io_sync ; + mark_buffer_notjournal_new(bh) ; + clear_bit(BH_Dirty, &bh->b_state) ; + submit_bh(WRITE, bh) ; +} + +/* flush a journal list, both commit and real blocks +** +** always set flushall to 1, unless you are calling from inside +** flush_journal_list +** +** IMPORTANT. This can only be called while there are no journal writers, +** and the journal is locked. That means it can only be called from +** do_journal_end, or by journal_release +*/ +static int flush_journal_list(struct super_block *s, + struct reiserfs_journal_list *jl, int flushall) { + struct reiserfs_journal_list *pjl ; + struct reiserfs_journal_cnode *cn, *last ; + int count ; + int was_jwait = 0 ; + int was_dirty = 0 ; + struct buffer_head *saved_bh ; + unsigned long j_len_saved = jl->j_len ; + + if (j_len_saved <= 0) { + return 0 ; + } + + if (atomic_read(&SB_JOURNAL(s)->j_wcount) != 0) { + reiserfs_warning("clm-2048: flush_journal_list called with wcount %d\n", + atomic_read(&SB_JOURNAL(s)->j_wcount)) ; + } + /* if someone is getting the commit list, we must wait for them */ + while (atomic_read(&(jl->j_commit_flushing))) { + sleep_on(&(jl->j_commit_wait)) ; + } + /* if someone is flushing this list, we must wait for them */ + while (atomic_read(&(jl->j_flushing))) { + sleep_on(&(jl->j_flush_wait)) ; + } + + /* this list is now ours, we can change anything we want */ + atomic_set(&(jl->j_flushing), 1) ; + + count = 0 ; + if (j_len_saved > JOURNAL_TRANS_MAX) { + reiserfs_panic(s, "journal-715: flush_journal_list, length is %lu, list number %d\n", j_len_saved, jl - SB_JOURNAL_LIST(s)) ; + atomic_dec(&(jl->j_flushing)) ; + return 0 ; + } + + /* if all the work is already done, get out of here */ + if (atomic_read(&(jl->j_nonzerolen)) <= 0 && + atomic_read(&(jl->j_commit_left)) <= 0) { + goto flush_older_and_return ; + } + + /* start by putting the commit list on disk. This will also flush + ** the commit lists of any olders transactions + */ + flush_commit_list(s, jl, 1) ; + + /* are we done now? */ + if (atomic_read(&(jl->j_nonzerolen)) <= 0 && + atomic_read(&(jl->j_commit_left)) <= 0) { + goto flush_older_and_return ; + } + + /* loop through each cnode, see if we need to write it, + ** or wait on a more recent transaction, or just ignore it + */ + if (atomic_read(&(SB_JOURNAL(s)->j_wcount)) != 0) { + reiserfs_panic(s, "journal-844: panic journal list is flushing, wcount is not 0\n") ; + } + cn = jl->j_realblock ; + while(cn) { + was_jwait = 0 ; + was_dirty = 0 ; + saved_bh = NULL ; + /* blocknr of 0 is no longer in the hash, ignore it */ + if (cn->blocknr == 0) { + goto free_cnode ; + } + pjl = find_newer_jl_for_cn(cn) ; + /* the order is important here. We check pjl to make sure we + ** don't clear BH_JDirty_wait if we aren't the one writing this + ** block to disk + */ + if (!pjl && cn->bh) { + saved_bh = cn->bh ; + + /* we do this to make sure nobody releases the buffer while + ** we are working with it + */ + atomic_inc(&(saved_bh->b_count)) ; + + if (buffer_journal_dirty(saved_bh)) { + was_jwait = 1 ; + mark_buffer_notjournal_dirty(saved_bh) ; + /* brelse the inc from journal_mark_dirty */ + atomic_dec(&(saved_bh->b_count)) ; + } + if (can_dirty(cn)) { + was_dirty = 1 ; + } + } + + /* if someone has this block in a newer transaction, just make + ** sure they are commited, and don't try writing it to disk + */ + if (pjl) { + flush_commit_list(s, pjl, 1) ; + goto free_cnode ; + } + + /* bh == NULL when the block got to disk on its own, OR, + ** the block got freed in a future transaction + */ + if (saved_bh == NULL) { + goto free_cnode ; + } + + /* this should never happen. kupdate_one_transaction has this list + ** locked while it works, so we should never see a buffer here that + ** is not marked JDirty_wait + */ + if ((!was_jwait) && !buffer_locked(saved_bh)) { +printk("journal-813: BAD! buffer %lu %cdirty %cjwait, not in a newer tranasction\n", saved_bh->b_blocknr, + was_dirty ? ' ' : '!', was_jwait ? ' ' : '!') ; + } + /* kupdate_one_transaction waits on the buffers it is writing, so we + ** should never see locked buffers here + */ + if (buffer_locked(saved_bh)) { + printk("clm-2083: locked buffer %lu in flush_journal_list\n", + saved_bh->b_blocknr) ; + wait_on_buffer(saved_bh) ; + if (!buffer_uptodate(saved_bh)) { + reiserfs_panic(s, "journal-923: buffer write failed\n") ; + } + } + if (was_dirty) { + /* we inc again because saved_bh gets decremented at free_cnode */ + atomic_inc(&(saved_bh->b_count)) ; + set_bit(BLOCK_NEEDS_FLUSH, &cn->state) ; + submit_logged_buffer(saved_bh) ; + count++ ; + } else { + printk("clm-2082: Unable to flush buffer %lu in flush_journal_list\n", + saved_bh->b_blocknr) ; + } +free_cnode: + last = cn ; + cn = cn->next ; + if (saved_bh) { + /* we incremented this to keep others from taking the buffer head away */ + atomic_dec(&(saved_bh->b_count)); + if (atomic_read(&(saved_bh->b_count)) < 0) { + printk("journal-945: saved_bh->b_count < 0") ; + } + } + } + if (count > 0) { + cn = jl->j_realblock ; + while(cn) { + if (test_bit(BLOCK_NEEDS_FLUSH, &cn->state)) { + if (!cn->bh) { + reiserfs_panic(s, "journal-1011: cn->bh is NULL\n") ; + } + wait_on_buffer(cn->bh) ; + if (!cn->bh) { + reiserfs_panic(s, "journal-1012: cn->bh is NULL\n") ; + } + if (!buffer_uptodate(cn->bh)) { + reiserfs_panic(s, "journal-949: buffer write failed\n") ; + } + refile_buffer(cn->bh) ; + brelse(cn->bh) ; + } + cn = cn->next ; + } + } + +flush_older_and_return: + /* before we can update the journal header block, we _must_ flush all + ** real blocks from all older transactions to disk. This is because + ** once the header block is updated, this transaction will not be + ** replayed after a crash + */ + if (flushall) { + flush_older_journal_lists(s, jl, jl->j_trans_id) ; + } + + /* before we can remove everything from the hash tables for this + ** transaction, we must make sure it can never be replayed + ** + ** since we are only called from do_journal_end, we know for sure there + ** are no allocations going on while we are flushing journal lists. So, + ** we only need to update the journal header block for the last list + ** being flushed + */ + if (flushall) { + update_journal_header_block(s, (jl->j_start + jl->j_len + 2) % JOURNAL_BLOCK_COUNT, jl->j_trans_id) ; + } + remove_all_from_journal_list(s, jl, 0) ; + jl->j_len = 0 ; + atomic_set(&(jl->j_nonzerolen), 0) ; + jl->j_start = 0 ; + jl->j_realblock = NULL ; + jl->j_commit_bh = NULL ; + jl->j_trans_id = 0 ; + atomic_dec(&(jl->j_flushing)) ; + wake_up(&(jl->j_flush_wait)) ; + return 0 ; +} + + +static int kupdate_one_transaction(struct super_block *s, + struct reiserfs_journal_list *jl) +{ + struct reiserfs_journal_list *pjl ; /* previous list for this cn */ + struct reiserfs_journal_cnode *cn, *walk_cn ; + unsigned long blocknr ; + int run = 0 ; + int orig_trans_id = jl->j_trans_id ; + struct buffer_head *saved_bh ; + int ret = 0 ; + + /* if someone is getting the commit list, we must wait for them */ + while (atomic_read(&(jl->j_commit_flushing))) { + sleep_on(&(jl->j_commit_wait)) ; + } + /* if someone is flushing this list, we must wait for them */ + while (atomic_read(&(jl->j_flushing))) { + sleep_on(&(jl->j_flush_wait)) ; + } + /* was it flushed while we slept? */ + if (jl->j_len <= 0 || jl->j_trans_id != orig_trans_id) { + return 0 ; + } + + /* this list is now ours, we can change anything we want */ + atomic_set(&(jl->j_flushing), 1) ; + +loop_start: + cn = jl->j_realblock ; + while(cn) { + saved_bh = NULL ; + /* if the blocknr == 0, this has been cleared from the hash, + ** skip it + */ + if (cn->blocknr == 0) { + goto next ; + } + /* look for a more recent transaction that logged this + ** buffer. Only the most recent transaction with a buffer in + ** it is allowed to send that buffer to disk + */ + pjl = find_newer_jl_for_cn(cn) ; + if (run == 0 && !pjl && cn->bh && buffer_journal_dirty(cn->bh) && + can_dirty(cn)) + { + if (!test_bit(BH_JPrepared, &cn->bh->b_state)) { + set_bit(BLOCK_NEEDS_FLUSH, &cn->state) ; + submit_logged_buffer(cn->bh) ; + } else { + /* someone else is using this buffer. We can't + ** send it to disk right now because they might + ** be changing/logging it. + */ + ret = 1 ; + } + } else if (test_bit(BLOCK_NEEDS_FLUSH, &cn->state)) { + clear_bit(BLOCK_NEEDS_FLUSH, &cn->state) ; + if (!pjl && cn->bh) { + wait_on_buffer(cn->bh) ; + } + /* check again, someone could have logged while we scheduled */ + pjl = find_newer_jl_for_cn(cn) ; + + /* before the JDirty_wait bit is set, the + ** buffer is added to the hash list. So, if we are + ** run in the middle of a do_journal_end, we will notice + ** if this buffer was logged and added from the latest + ** transaction. In this case, we don't want to decrement + ** b_count + */ + if (!pjl && cn->bh && buffer_journal_dirty(cn->bh)) { + blocknr = cn->blocknr ; + walk_cn = cn ; + saved_bh= cn->bh ; + /* update all older transactions to show this block + ** was flushed + */ + mark_buffer_notjournal_dirty(cn->bh) ; + while(walk_cn) { + if (walk_cn->bh && walk_cn->blocknr == blocknr && + walk_cn->dev == cn->dev) { + if (walk_cn->jlist) { + atomic_dec(&(walk_cn->jlist->j_nonzerolen)) ; + } + walk_cn->bh = NULL ; + } + walk_cn = walk_cn->hnext ; + } + if (atomic_read(&saved_bh->b_count) < 1) { + reiserfs_warning("clm-2081: bad count on %lu\n", + saved_bh->b_blocknr) ; + } + brelse(saved_bh) ; + } + } + /* + ** if the more recent transaction is committed to the log, + ** this buffer can be considered flushed. Decrement our + ** counters to reflect one less buffer that needs writing. + ** + ** note, this relies on all of the above code being + ** schedule free once pjl comes back non-null. + */ + if (pjl && cn->bh && atomic_read(&pjl->j_commit_left) == 0) { + atomic_dec(&cn->jlist->j_nonzerolen) ; + cn->bh = NULL ; + } +next: + cn = cn->next ; + } + /* the first run through the loop sends all the dirty buffers to + ** ll_rw_block. + ** the second run through the loop does all the accounting + */ + if (run++ == 0) { + goto loop_start ; + } + + atomic_set(&(jl->j_flushing), 0) ; + wake_up(&(jl->j_flush_wait)) ; + return ret ; +} +/* since we never give dirty buffers to bdflush/kupdate, we have to +** flush them ourselves. This runs through the journal lists, finds +** old metadata in need of flushing and sends it to disk. +** this does not end transactions, commit anything, or free +** cnodes. +** +** returns the highest transaction id that was flushed last time +*/ +static unsigned long reiserfs_journal_kupdate(struct super_block *s) { + struct reiserfs_journal_list *jl ; + int i ; + int start ; + time_t age ; + int ret = 0 ; + + start = SB_JOURNAL_LIST_INDEX(s) ; + + /* safety check to prevent flush attempts during a mount */ + if (start < 0) { + return 0 ; + } + i = (start + 1) % JOURNAL_LIST_COUNT ; + while(i != start) { + jl = SB_JOURNAL_LIST(s) + i ; + age = CURRENT_TIME - jl->j_timestamp ; + if (jl->j_len > 0 && // age >= (JOURNAL_MAX_COMMIT_AGE * 2) && + atomic_read(&(jl->j_nonzerolen)) > 0 && + atomic_read(&(jl->j_commit_left)) == 0) { + + if (jl->j_trans_id == SB_JOURNAL(s)->j_trans_id) { + break ; + } + /* if ret was already 1, we want to preserve that */ + ret |= kupdate_one_transaction(s, jl) ; + } + if (atomic_read(&(jl->j_nonzerolen)) > 0) { + ret |= 1 ; + } + i = (i + 1) % JOURNAL_LIST_COUNT ; + } + return ret ; +} + +/* +** removes any nodes in table with name block and dev as bh. +** only touchs the hnext and hprev pointers. +*/ +void remove_journal_hash(struct reiserfs_journal_cnode **table, struct reiserfs_journal_list *jl,struct buffer_head *bh, + int remove_freed){ + struct reiserfs_journal_cnode *cur ; + struct reiserfs_journal_cnode **head ; + + if (!bh) + return ; + + head= &(journal_hash(table, bh->b_dev, bh->b_blocknr)) ; + if (!head) { + return ; + } + cur = *head ; + while(cur) { + if (cur->blocknr == bh->b_blocknr && cur->dev == bh->b_dev && (jl == NULL || jl == cur->jlist) && + (!test_bit(BLOCK_FREED, &cur->state) || remove_freed)) { + if (cur->hnext) { + cur->hnext->hprev = cur->hprev ; + } + if (cur->hprev) { + cur->hprev->hnext = cur->hnext ; + } else { + *head = cur->hnext ; + } + cur->blocknr = 0 ; + cur->dev = 0 ; + cur->state = 0 ; + if (cur->bh && cur->jlist) /* anybody who clears the cur->bh will also dec the nonzerolen */ + atomic_dec(&(cur->jlist->j_nonzerolen)) ; + cur->bh = NULL ; + cur->jlist = NULL ; + } + cur = cur->hnext ; + } +} + +static void free_journal_ram(struct super_block *p_s_sb) { + vfree(SB_JOURNAL(p_s_sb)->j_cnode_free_orig) ; + free_list_bitmaps(p_s_sb, SB_JOURNAL(p_s_sb)->j_list_bitmap) ; + free_bitmap_nodes(p_s_sb) ; /* must be after free_list_bitmaps */ + if (SB_JOURNAL(p_s_sb)->j_header_bh) { + brelse(SB_JOURNAL(p_s_sb)->j_header_bh) ; + } + vfree(SB_JOURNAL(p_s_sb)) ; +} + +/* +** call on unmount. Only set error to 1 if you haven't made your way out +** of read_super() yet. Any other caller must keep error at 0. +*/ +static int do_journal_release(struct reiserfs_transaction_handle *th, struct super_block *p_s_sb, int error) { + struct reiserfs_transaction_handle myth ; + + /* we only want to flush out transactions if we were called with error == 0 + */ + if (!error && !(p_s_sb->s_flags & MS_RDONLY)) { + /* end the current trans */ + do_journal_end(th, p_s_sb,10, FLUSH_ALL) ; + + /* make sure something gets logged to force our way into the flush code */ + journal_join(&myth, p_s_sb, 1) ; + reiserfs_prepare_for_journal(p_s_sb, SB_BUFFER_WITH_SB(p_s_sb), 1) ; + journal_mark_dirty(&myth, p_s_sb, SB_BUFFER_WITH_SB(p_s_sb)) ; + do_journal_end(&myth, p_s_sb,1, FLUSH_ALL) ; + } + + /* we decrement before we wake up, because the commit thread dies off + ** when it has been woken up and the count is <= 0 + */ + reiserfs_mounted_fs_count-- ; + wake_up(&reiserfs_commit_thread_wait) ; + sleep_on(&reiserfs_commit_thread_done) ; + + free_journal_ram(p_s_sb) ; + + return 0 ; +} + +/* +** call on unmount. flush all journal trans, release all alloc'd ram +*/ +int journal_release(struct reiserfs_transaction_handle *th, struct super_block *p_s_sb) { + return do_journal_release(th, p_s_sb, 0) ; +} +/* +** only call from an error condition inside reiserfs_read_super! +*/ +int journal_release_error(struct reiserfs_transaction_handle *th, struct super_block *p_s_sb) { + return do_journal_release(th, p_s_sb, 1) ; +} + +/* compares description block with commit block. returns 1 if they differ, 0 if they are the same */ +static int journal_compare_desc_commit(struct super_block *p_s_sb, struct reiserfs_journal_desc *desc, + struct reiserfs_journal_commit *commit) { + if (le32_to_cpu(commit->j_trans_id) != le32_to_cpu(desc->j_trans_id) || + le32_to_cpu(commit->j_len) != le32_to_cpu(desc->j_len) || + le32_to_cpu(commit->j_len) > JOURNAL_TRANS_MAX || + le32_to_cpu(commit->j_len) <= 0 + ) { + return 1 ; + } + return 0 ; +} +/* returns 0 if it did not find a description block +** returns -1 if it found a corrupt commit block +** returns 1 if both desc and commit were valid +*/ +static int journal_transaction_is_valid(struct super_block *p_s_sb, struct buffer_head *d_bh, unsigned long *oldest_invalid_trans_id, unsigned long *newest_mount_id) { + struct reiserfs_journal_desc *desc ; + struct reiserfs_journal_commit *commit ; + struct buffer_head *c_bh ; + unsigned long offset ; + + desc = (struct reiserfs_journal_desc *)d_bh->b_data ; + if (le32_to_cpu(desc->j_len) > 0 && !memcmp(desc->j_magic, JOURNAL_DESC_MAGIC, 8)) { + if (oldest_invalid_trans_id && *oldest_invalid_trans_id && le32_to_cpu(desc->j_trans_id) > *oldest_invalid_trans_id) { + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-986: transaction " + "is valid returning because trans_id %d is greater than " + "oldest_invalid %lu\n", le32_to_cpu(desc->j_trans_id), + *oldest_invalid_trans_id); + return 0 ; + } + if (newest_mount_id && *newest_mount_id > le32_to_cpu(desc->j_mount_id)) { + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1087: transaction " + "is valid returning because mount_id %d is less than " + "newest_mount_id %lu\n", desc->j_mount_id, + *newest_mount_id) ; + return -1 ; + } + offset = d_bh->b_blocknr - reiserfs_get_journal_block(p_s_sb) ; + + /* ok, we have a journal description block, lets see if the transaction was valid */ + c_bh = bread(p_s_sb->s_dev, reiserfs_get_journal_block(p_s_sb) + ((offset + le32_to_cpu(desc->j_len) + 1) % JOURNAL_BLOCK_COUNT), + p_s_sb->s_blocksize) ; + if (!c_bh) + return 0 ; + commit = (struct reiserfs_journal_commit *)c_bh->b_data ; + if (journal_compare_desc_commit(p_s_sb, desc, commit)) { + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, + "journal_transaction_is_valid, commit offset %ld had bad " + "time %d or length %d\n", + c_bh->b_blocknr - reiserfs_get_journal_block(p_s_sb), + le32_to_cpu(commit->j_trans_id), + le32_to_cpu(commit->j_len)); + brelse(c_bh) ; + if (oldest_invalid_trans_id) + *oldest_invalid_trans_id = le32_to_cpu(desc->j_trans_id) ; + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1004: " + "transaction_is_valid setting oldest invalid trans_id " + "to %d\n", le32_to_cpu(desc->j_trans_id)) ; + return -1; + } + brelse(c_bh) ; + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1006: found valid " + "transaction start offset %lu, len %d id %d\n", + d_bh->b_blocknr - reiserfs_get_journal_block(p_s_sb), + le32_to_cpu(desc->j_len), le32_to_cpu(desc->j_trans_id)) ; + return 1 ; + } else { + return 0 ; + } +} + +static void brelse_array(struct buffer_head **heads, int num) { + int i ; + for (i = 0 ; i < num ; i++) { + brelse(heads[i]) ; + } +} + +/* +** given the start, and values for the oldest acceptable transactions, +** this either reads in a replays a transaction, or returns because the transaction +** is invalid, or too old. +*/ +static int journal_read_transaction(struct super_block *p_s_sb, unsigned long cur_dblock, unsigned long oldest_start, + unsigned long oldest_trans_id, unsigned long newest_mount_id) { + struct reiserfs_journal_desc *desc ; + struct reiserfs_journal_commit *commit ; + unsigned long trans_id = 0 ; + struct buffer_head *c_bh ; + struct buffer_head *d_bh ; + struct buffer_head **log_blocks = NULL ; + struct buffer_head **real_blocks = NULL ; + unsigned long trans_offset ; + int i; + + d_bh = bread(p_s_sb->s_dev, cur_dblock, p_s_sb->s_blocksize) ; + if (!d_bh) + return 1 ; + desc = (struct reiserfs_journal_desc *)d_bh->b_data ; + trans_offset = d_bh->b_blocknr - reiserfs_get_journal_block(p_s_sb) ; + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1037: " + "journal_read_transaction, offset %lu, len %d mount_id %d\n", + d_bh->b_blocknr - reiserfs_get_journal_block(p_s_sb), + le32_to_cpu(desc->j_len), le32_to_cpu(desc->j_mount_id)) ; + if (le32_to_cpu(desc->j_trans_id) < oldest_trans_id) { + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1039: " + "journal_read_trans skipping because %lu is too old\n", + cur_dblock - reiserfs_get_journal_block(p_s_sb)) ; + brelse(d_bh) ; + return 1 ; + } + if (le32_to_cpu(desc->j_mount_id) != newest_mount_id) { + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1146: " + "journal_read_trans skipping because %d is != " + "newest_mount_id %lu\n", le32_to_cpu(desc->j_mount_id), + newest_mount_id) ; + brelse(d_bh) ; + return 1 ; + } + c_bh = bread(p_s_sb->s_dev, reiserfs_get_journal_block(p_s_sb) + ((trans_offset + le32_to_cpu(desc->j_len) + 1) % JOURNAL_BLOCK_COUNT), + p_s_sb->s_blocksize) ; + if (!c_bh) { + brelse(d_bh) ; + return 1 ; + } + commit = (struct reiserfs_journal_commit *)c_bh->b_data ; + if (journal_compare_desc_commit(p_s_sb, desc, commit)) { + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal_read_transaction, " + "commit offset %ld had bad time %d or length %d\n", + c_bh->b_blocknr - reiserfs_get_journal_block(p_s_sb), + le32_to_cpu(commit->j_trans_id), le32_to_cpu(commit->j_len)); + brelse(c_bh) ; + brelse(d_bh) ; + return 1; + } + trans_id = le32_to_cpu(desc->j_trans_id) ; + /* now we know we've got a good transaction, and it was inside the valid time ranges */ + log_blocks = kmalloc(le32_to_cpu(desc->j_len) * sizeof(struct buffer_head *), GFP_BUFFER) ; + real_blocks = kmalloc(le32_to_cpu(desc->j_len) * sizeof(struct buffer_head *), GFP_BUFFER) ; + if (!log_blocks || !real_blocks) { + brelse(c_bh) ; + brelse(d_bh) ; + kfree(log_blocks) ; + kfree(real_blocks) ; + reiserfs_warning("journal-1169: kmalloc failed, unable to mount FS\n") ; + return -1 ; + } + /* get all the buffer heads */ + for(i = 0 ; i < le32_to_cpu(desc->j_len) ; i++) { + log_blocks[i] = getblk(p_s_sb->s_dev, reiserfs_get_journal_block(p_s_sb) + (trans_offset + 1 + i) % JOURNAL_BLOCK_COUNT, p_s_sb->s_blocksize); + if (i < JOURNAL_TRANS_HALF) { + real_blocks[i] = getblk(p_s_sb->s_dev, le32_to_cpu(desc->j_realblock[i]), p_s_sb->s_blocksize) ; + } else { + real_blocks[i] = getblk(p_s_sb->s_dev, le32_to_cpu(commit->j_realblock[i - JOURNAL_TRANS_HALF]), p_s_sb->s_blocksize) ; + } + if (real_blocks[i]->b_blocknr >= reiserfs_get_journal_block(p_s_sb) && + real_blocks[i]->b_blocknr < (reiserfs_get_journal_block(p_s_sb)+JOURNAL_BLOCK_COUNT)) { + reiserfs_warning("journal-1204: REPLAY FAILURE fsck required! Trying to replay onto a log block\n") ; + brelse_array(log_blocks, i) ; + brelse_array(real_blocks, i) ; + brelse(c_bh) ; + brelse(d_bh) ; + kfree(log_blocks) ; + kfree(real_blocks) ; + return -1 ; + } + } + /* read in the log blocks, memcpy to the corresponding real block */ + ll_rw_block(READ, le32_to_cpu(desc->j_len), log_blocks) ; + for (i = 0 ; i < le32_to_cpu(desc->j_len) ; i++) { + wait_on_buffer(log_blocks[i]) ; + if (!buffer_uptodate(log_blocks[i])) { + reiserfs_warning("journal-1212: REPLAY FAILURE fsck required! buffer write failed\n") ; + brelse_array(log_blocks + i, le32_to_cpu(desc->j_len) - i) ; + brelse_array(real_blocks, le32_to_cpu(desc->j_len)) ; + brelse(c_bh) ; + brelse(d_bh) ; + kfree(log_blocks) ; + kfree(real_blocks) ; + return -1 ; + } + memcpy(real_blocks[i]->b_data, log_blocks[i]->b_data, real_blocks[i]->b_size) ; + mark_buffer_uptodate(real_blocks[i], 1) ; + brelse(log_blocks[i]) ; + } + /* flush out the real blocks */ + for (i = 0 ; i < le32_to_cpu(desc->j_len) ; i++) { + set_bit(BH_Dirty, &(real_blocks[i]->b_state)) ; + ll_rw_block(WRITE, 1, real_blocks + i) ; + } + for (i = 0 ; i < le32_to_cpu(desc->j_len) ; i++) { + wait_on_buffer(real_blocks[i]) ; + if (!buffer_uptodate(real_blocks[i])) { + reiserfs_warning("journal-1226: REPLAY FAILURE, fsck required! buffer write failed\n") ; + brelse_array(real_blocks + i, le32_to_cpu(desc->j_len) - i) ; + brelse(c_bh) ; + brelse(d_bh) ; + kfree(log_blocks) ; + kfree(real_blocks) ; + return -1 ; + } + brelse(real_blocks[i]) ; + } + cur_dblock = reiserfs_get_journal_block(p_s_sb) + ((trans_offset + le32_to_cpu(desc->j_len) + 2) % JOURNAL_BLOCK_COUNT) ; + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1095: setting journal " + "start to offset %ld\n", + cur_dblock - reiserfs_get_journal_block(p_s_sb)) ; + + /* init starting values for the first transaction, in case this is the last transaction to be replayed. */ + SB_JOURNAL(p_s_sb)->j_start = cur_dblock - reiserfs_get_journal_block(p_s_sb) ; + SB_JOURNAL(p_s_sb)->j_last_flush_trans_id = trans_id ; + SB_JOURNAL(p_s_sb)->j_trans_id = trans_id + 1; + brelse(c_bh) ; + brelse(d_bh) ; + kfree(log_blocks) ; + kfree(real_blocks) ; + return 0 ; +} + +/* +** read and replay the log +** on a clean unmount, the journal header's next unflushed pointer will be to an invalid +** transaction. This tests that before finding all the transactions in the log, whic makes normal mount times fast. +** +** After a crash, this starts with the next unflushed transaction, and replays until it finds one too old, or invalid. +** +** On exit, it sets things up so the first transaction will work correctly. +*/ +static int journal_read(struct super_block *p_s_sb) { + struct reiserfs_journal_desc *desc ; + unsigned long last_flush_trans_id = 0 ; + unsigned long oldest_trans_id = 0; + unsigned long oldest_invalid_trans_id = 0 ; + time_t start ; + unsigned long last_flush_start = 0; + unsigned long oldest_start = 0; + unsigned long cur_dblock = 0 ; + unsigned long newest_mount_id = 9 ; + struct buffer_head *d_bh ; + struct reiserfs_journal_header *jh ; + int valid_journal_header = 0 ; + int replay_count = 0 ; + int continue_replay = 1 ; + int ret ; + + cur_dblock = reiserfs_get_journal_block(p_s_sb) ; + printk("reiserfs: checking transaction log (device %s) ...\n", + kdevname(p_s_sb->s_dev)) ; + start = CURRENT_TIME ; + + /* step 1, read in the journal header block. Check the transaction it says + ** is the first unflushed, and if that transaction is not valid, + ** replay is done + */ + SB_JOURNAL(p_s_sb)->j_header_bh = bread(p_s_sb->s_dev, + reiserfs_get_journal_block(p_s_sb) + + JOURNAL_BLOCK_COUNT, + p_s_sb->s_blocksize) ; + if (!SB_JOURNAL(p_s_sb)->j_header_bh) { + return 1 ; + } + jh = (struct reiserfs_journal_header *)(SB_JOURNAL(p_s_sb)->j_header_bh->b_data) ; + if (le32_to_cpu(jh->j_first_unflushed_offset) >= 0 && + le32_to_cpu(jh->j_first_unflushed_offset) < JOURNAL_BLOCK_COUNT && + le32_to_cpu(jh->j_last_flush_trans_id) > 0) { + last_flush_start = reiserfs_get_journal_block(p_s_sb) + + le32_to_cpu(jh->j_first_unflushed_offset) ; + last_flush_trans_id = le32_to_cpu(jh->j_last_flush_trans_id) ; + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1153: found in " + "header: first_unflushed_offset %d, last_flushed_trans_id " + "%lu\n", le32_to_cpu(jh->j_first_unflushed_offset), + last_flush_trans_id) ; + valid_journal_header = 1 ; + + /* now, we try to read the first unflushed offset. If it is not valid, + ** there is nothing more we can do, and it makes no sense to read + ** through the whole log. + */ + d_bh = bread(p_s_sb->s_dev, reiserfs_get_journal_block(p_s_sb) + le32_to_cpu(jh->j_first_unflushed_offset), p_s_sb->s_blocksize) ; + ret = journal_transaction_is_valid(p_s_sb, d_bh, NULL, NULL) ; + if (!ret) { + continue_replay = 0 ; + } + brelse(d_bh) ; + } + + if (continue_replay && is_read_only(p_s_sb->s_dev)) { + printk("clm-2076: device is readonly, unable to replay log\n") ; + brelse(SB_JOURNAL(p_s_sb)->j_header_bh) ; + SB_JOURNAL(p_s_sb)->j_header_bh = NULL ; + return -1 ; + } + if (continue_replay && (p_s_sb->s_flags & MS_RDONLY)) { + printk("Warning, log replay starting on readonly filesystem\n") ; + } + + /* ok, there are transactions that need to be replayed. start with the first log block, find + ** all the valid transactions, and pick out the oldest. + */ + while(continue_replay && cur_dblock < (reiserfs_get_journal_block(p_s_sb) + JOURNAL_BLOCK_COUNT)) { + d_bh = bread(p_s_sb->s_dev, cur_dblock, p_s_sb->s_blocksize) ; + ret = journal_transaction_is_valid(p_s_sb, d_bh, &oldest_invalid_trans_id, &newest_mount_id) ; + if (ret == 1) { + desc = (struct reiserfs_journal_desc *)d_bh->b_data ; + if (oldest_start == 0) { /* init all oldest_ values */ + oldest_trans_id = le32_to_cpu(desc->j_trans_id) ; + oldest_start = d_bh->b_blocknr ; + newest_mount_id = le32_to_cpu(desc->j_mount_id) ; + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1179: Setting " + "oldest_start to offset %lu, trans_id %lu\n", + oldest_start - reiserfs_get_journal_block(p_s_sb), + oldest_trans_id) ; + } else if (oldest_trans_id > le32_to_cpu(desc->j_trans_id)) { + /* one we just read was older */ + oldest_trans_id = le32_to_cpu(desc->j_trans_id) ; + oldest_start = d_bh->b_blocknr ; + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1180: Resetting " + "oldest_start to offset %lu, trans_id %lu\n", + oldest_start - reiserfs_get_journal_block(p_s_sb), + oldest_trans_id) ; + } + if (newest_mount_id < le32_to_cpu(desc->j_mount_id)) { + newest_mount_id = le32_to_cpu(desc->j_mount_id) ; + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1299: Setting " + "newest_mount_id to %d\n", le32_to_cpu(desc->j_mount_id)); + } + cur_dblock += le32_to_cpu(desc->j_len) + 2 ; + } + else { + cur_dblock++ ; + } + brelse(d_bh) ; + } + /* step three, starting at the oldest transaction, replay */ + if (last_flush_start > 0) { + oldest_start = last_flush_start ; + oldest_trans_id = last_flush_trans_id ; + } + cur_dblock = oldest_start ; + if (oldest_trans_id) { + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1206: Starting replay " + "from offset %lu, trans_id %lu\n", + cur_dblock - reiserfs_get_journal_block(p_s_sb), + oldest_trans_id) ; + + } + replay_count = 0 ; + while(continue_replay && oldest_trans_id > 0) { + ret = journal_read_transaction(p_s_sb, cur_dblock, oldest_start, oldest_trans_id, newest_mount_id) ; + if (ret < 0) { + return ret ; + } else if (ret != 0) { + break ; + } + cur_dblock = reiserfs_get_journal_block(p_s_sb) + SB_JOURNAL(p_s_sb)->j_start ; + replay_count++ ; + } + + if (oldest_trans_id == 0) { + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1225: No valid " + "transactions found\n") ; + } + /* j_start does not get set correctly if we don't replay any transactions. + ** if we had a valid journal_header, set j_start to the first unflushed transaction value, + ** copy the trans_id from the header + */ + if (valid_journal_header && replay_count == 0) { + SB_JOURNAL(p_s_sb)->j_start = le32_to_cpu(jh->j_first_unflushed_offset) ; + SB_JOURNAL(p_s_sb)->j_trans_id = le32_to_cpu(jh->j_last_flush_trans_id) + 1; + SB_JOURNAL(p_s_sb)->j_last_flush_trans_id = le32_to_cpu(jh->j_last_flush_trans_id) ; + SB_JOURNAL(p_s_sb)->j_mount_id = le32_to_cpu(jh->j_mount_id) + 1; + } else { + SB_JOURNAL(p_s_sb)->j_mount_id = newest_mount_id + 1 ; + } + reiserfs_debug(p_s_sb, REISERFS_DEBUG_CODE, "journal-1299: Setting " + "newest_mount_id to %lu\n", SB_JOURNAL(p_s_sb)->j_mount_id) ; + SB_JOURNAL(p_s_sb)->j_first_unflushed_offset = SB_JOURNAL(p_s_sb)->j_start ; + if (replay_count > 0) { + printk("reiserfs: replayed %d transactions in %lu seconds\n", replay_count, + CURRENT_TIME - start) ; + } + if (!is_read_only(p_s_sb->s_dev)) { + update_journal_header_block(p_s_sb, SB_JOURNAL(p_s_sb)->j_start, + SB_JOURNAL(p_s_sb)->j_last_flush_trans_id) ; + } + return 0 ; +} + + +struct reiserfs_journal_commit_task { + struct super_block *p_s_sb ; + int jindex ; + int wake_on_finish ; /* if this is one, we wake the task_done queue, if it + ** is zero, we free the whole struct on finish + */ + struct reiserfs_journal_commit_task *self ; + struct wait_queue *task_done ; + struct tq_struct task ; +} ; + +static void reiserfs_journal_commit_task_func(struct reiserfs_journal_commit_task *ct) { + + struct reiserfs_journal_list *jl ; + jl = SB_JOURNAL_LIST(ct->p_s_sb) + ct->jindex ; + + flush_commit_list(ct->p_s_sb, SB_JOURNAL_LIST(ct->p_s_sb) + ct->jindex, 1) ; + if (jl->j_len > 0 && atomic_read(&(jl->j_nonzerolen)) > 0 && + atomic_read(&(jl->j_commit_left)) == 0) { + kupdate_one_transaction(ct->p_s_sb, jl) ; + } + kfree(ct->self) ; +} + +static void setup_commit_task_arg(struct reiserfs_journal_commit_task *ct, + struct super_block *p_s_sb, + int jindex) { + if (!ct) { + reiserfs_panic(NULL, "journal-1360: setup_commit_task_arg called with NULL struct\n") ; + } + ct->p_s_sb = p_s_sb ; + ct->jindex = jindex ; + ct->task_done = NULL ; + INIT_LIST_HEAD(&ct->task.list) ; + ct->task.sync = 0 ; + ct->task.routine = (void *)(void *)reiserfs_journal_commit_task_func ; + ct->self = ct ; + ct->task.data = (void *)ct ; +} + +static void commit_flush_async(struct super_block *p_s_sb, int jindex) { + struct reiserfs_journal_commit_task *ct ; + /* using GFP_BUFFER, GFP_KERNEL could try to flush inodes, which will try + ** to start/join a transaction, which will deadlock + */ + ct = kmalloc(sizeof(struct reiserfs_journal_commit_task), GFP_BUFFER) ; + if (ct) { + setup_commit_task_arg(ct, p_s_sb, jindex) ; + queue_task(&(ct->task), &reiserfs_commit_thread_tq); + wake_up(&reiserfs_commit_thread_wait) ; + } else { +#ifdef CONFIG_REISERFS_CHECK + reiserfs_warning("journal-1540: kmalloc failed, doing sync commit\n") ; +#endif + flush_commit_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + jindex, 1) ; + } +} + +/* +** this is the commit thread. It is started with kernel_thread on +** FS mount, and journal_release() waits for it to exit. +** +** It could do a periodic commit, but there is a lot code for that +** elsewhere right now, and I only wanted to implement this little +** piece for starters. +** +** All we do here is sleep on the j_commit_thread_wait wait queue, and +** then run the per filesystem commit task queue when we wakeup. +*/ +static int reiserfs_journal_commit_thread(void *nullp) { + exit_files(current); + exit_mm(current); + + spin_lock_irq(¤t->sigmask_lock); + sigfillset(¤t->blocked); + recalc_sigpending(current); + spin_unlock_irq(¤t->sigmask_lock); + + current->session = 1; + current->pgrp = 1; + sprintf(current->comm, "kreiserfsd") ; + lock_kernel() ; + while(1) { + + while(TQ_ACTIVE(reiserfs_commit_thread_tq)) { + run_task_queue(&reiserfs_commit_thread_tq) ; + } + + /* if there aren't any more filesystems left, break */ + if (reiserfs_mounted_fs_count <= 0) { + run_task_queue(&reiserfs_commit_thread_tq) ; + break ; + } + wake_up(&reiserfs_commit_thread_done) ; + interruptible_sleep_on_timeout(&reiserfs_commit_thread_wait, 5) ; + } + unlock_kernel() ; + wake_up(&reiserfs_commit_thread_done) ; + return 0 ; +} + +static void journal_list_init(struct super_block *p_s_sb) { + int i ; + for (i = 0 ; i < JOURNAL_LIST_COUNT ; i++) { + init_waitqueue_head(&(SB_JOURNAL_LIST(p_s_sb)[i].j_commit_wait)) ; + init_waitqueue_head(&(SB_JOURNAL_LIST(p_s_sb)[i].j_flush_wait)) ; + } +} + +/* +** must be called once on fs mount. calls journal_read for you +*/ +int journal_init(struct super_block *p_s_sb) { + int num_cnodes = JOURNAL_BLOCK_COUNT * 2 ; + + if (sizeof(struct reiserfs_journal_commit) != 4096 || + sizeof(struct reiserfs_journal_desc) != 4096 + ) { + printk("journal-1249: commit or desc struct not 4096 %Zd %Zd\n", sizeof(struct reiserfs_journal_commit), + sizeof(struct reiserfs_journal_desc)) ; + return 1 ; + } + /* sanity check to make sure they don't overflow the journal */ + if (JOURNAL_BLOCK_COUNT > reiserfs_get_journal_orig_size(p_s_sb)) { + printk("journal-1393: current JOURNAL_BLOCK_COUNT (%d) is too big. This FS was created with a journal size of %lu blocks\n", + JOURNAL_BLOCK_COUNT, reiserfs_get_journal_orig_size(p_s_sb)) ; + return 1 ; + } + SB_JOURNAL(p_s_sb) = vmalloc(sizeof (struct reiserfs_journal)) ; + + if (!SB_JOURNAL(p_s_sb)) { + printk("journal-1256: unable to get memory for journal structure\n") ; + return 1 ; + } + memset(SB_JOURNAL(p_s_sb), 0, sizeof(struct reiserfs_journal)) ; + + SB_JOURNAL(p_s_sb)->j_list_bitmap_index = 0 ; + SB_JOURNAL_LIST_INDEX(p_s_sb) = -10000 ; /* make sure flush_old_commits does not try to flush a list while replay is on */ + + /* clear out the journal list array */ + memset(SB_JOURNAL_LIST(p_s_sb), 0, sizeof(struct reiserfs_journal_list) * JOURNAL_LIST_COUNT) ; + journal_list_init(p_s_sb) ; + + memset(SB_JOURNAL(p_s_sb)->j_list_hash_table, 0, JOURNAL_HASH_SIZE * sizeof(struct reiserfs_journal_cnode *)) ; + memset(journal_writers, 0, sizeof(char *) * 512) ; /* debug code */ + + INIT_LIST_HEAD(&SB_JOURNAL(p_s_sb)->j_bitmap_nodes) ; + reiserfs_allocate_list_bitmaps(p_s_sb, SB_JOURNAL(p_s_sb)->j_list_bitmap, + SB_BMAP_NR(p_s_sb)) ; + allocate_bitmap_nodes(p_s_sb) ; + + SB_JOURNAL(p_s_sb)->j_start = 0 ; + SB_JOURNAL(p_s_sb)->j_len = 0 ; + SB_JOURNAL(p_s_sb)->j_len_alloc = 0 ; + atomic_set(&(SB_JOURNAL(p_s_sb)->j_wcount), 0) ; + SB_JOURNAL(p_s_sb)->j_bcount = 0 ; + SB_JOURNAL(p_s_sb)->j_trans_start_time = 0 ; + SB_JOURNAL(p_s_sb)->j_last = NULL ; + SB_JOURNAL(p_s_sb)->j_first = NULL ; + init_waitqueue_head(&(SB_JOURNAL(p_s_sb)->j_join_wait)) ; + init_waitqueue_head(&(SB_JOURNAL(p_s_sb)->j_wait)) ; + + SB_JOURNAL(p_s_sb)->j_trans_id = 10 ; + SB_JOURNAL(p_s_sb)->j_mount_id = 10 ; + SB_JOURNAL(p_s_sb)->j_state = 0 ; + atomic_set(&(SB_JOURNAL(p_s_sb)->j_jlock), 0) ; + atomic_set(&(SB_JOURNAL(p_s_sb)->j_wlock), 0) ; + SB_JOURNAL(p_s_sb)->j_cnode_free_list = allocate_cnodes(num_cnodes) ; + SB_JOURNAL(p_s_sb)->j_cnode_free_orig = SB_JOURNAL(p_s_sb)->j_cnode_free_list ; + SB_JOURNAL(p_s_sb)->j_cnode_free = SB_JOURNAL(p_s_sb)->j_cnode_free_list ? num_cnodes : 0 ; + SB_JOURNAL(p_s_sb)->j_cnode_used = 0 ; + SB_JOURNAL(p_s_sb)->j_must_wait = 0 ; + init_journal_hash(p_s_sb) ; + SB_JOURNAL_LIST(p_s_sb)[0].j_list_bitmap = get_list_bitmap(p_s_sb, SB_JOURNAL_LIST(p_s_sb)) ; + if (!(SB_JOURNAL_LIST(p_s_sb)[0].j_list_bitmap)) { + reiserfs_warning("journal-2005, get_list_bitmap failed for journal list 0\n") ; + return 1 ; + } + if (journal_read(p_s_sb) < 0) { + reiserfs_warning("Replay Failure, unable to mount\n") ; + free_journal_ram(p_s_sb) ; + return 1 ; + } + SB_JOURNAL_LIST_INDEX(p_s_sb) = 0 ; /* once the read is done, we can set this where it belongs */ + + if (reiserfs_dont_log (p_s_sb)) + return 0; + + reiserfs_mounted_fs_count++ ; + if (reiserfs_mounted_fs_count <= 1) { + kernel_thread((void *)(void *)reiserfs_journal_commit_thread, NULL, + CLONE_FS | CLONE_FILES | CLONE_VM) ; + } + return 0 ; +} + +/* +** test for a polite end of the current transaction. Used by file_write, and should +** be used by delete to make sure they don't write more than can fit inside a single +** transaction +*/ +int journal_transaction_should_end(struct reiserfs_transaction_handle *th, int new_alloc) { + time_t now = CURRENT_TIME ; + if (reiserfs_dont_log(th->t_super)) + return 0 ; + if ( SB_JOURNAL(th->t_super)->j_must_wait > 0 || + (SB_JOURNAL(th->t_super)->j_len_alloc + new_alloc) >= JOURNAL_MAX_BATCH || + atomic_read(&(SB_JOURNAL(th->t_super)->j_jlock)) || + (now - SB_JOURNAL(th->t_super)->j_trans_start_time) > JOURNAL_MAX_TRANS_AGE || + SB_JOURNAL(th->t_super)->j_cnode_free < (JOURNAL_TRANS_MAX * 3)) { + return 1 ; + } + return 0 ; +} + +/* this must be called inside a transaction, and requires the +** kernel_lock to be held +*/ +void reiserfs_block_writes(struct reiserfs_transaction_handle *th) { + struct super_block *s = th->t_super ; + SB_JOURNAL(s)->j_must_wait = 1 ; + set_bit(WRITERS_BLOCKED, &SB_JOURNAL(s)->j_state) ; + return ; +} + +/* this must be called without a transaction started, and does not +** require BKL +*/ +void reiserfs_allow_writes(struct super_block *s) { + clear_bit(WRITERS_BLOCKED, &SB_JOURNAL(s)->j_state) ; + wake_up(&SB_JOURNAL(s)->j_join_wait) ; +} + +/* this must be called without a transaction started, and does not +** require BKL +*/ +void reiserfs_wait_on_write_block(struct super_block *s) { + wait_event(SB_JOURNAL(s)->j_join_wait, + !test_bit(WRITERS_BLOCKED, &SB_JOURNAL(s)->j_state)) ; +} + +/* join == true if you must join an existing transaction. +** join == false if you can deal with waiting for others to finish +** +** this will block until the transaction is joinable. send the number of blocks you +** expect to use in nblocks. +*/ +static int do_journal_begin_r(struct reiserfs_transaction_handle *th, struct super_block * p_s_sb,unsigned long nblocks,int join) { + time_t now = CURRENT_TIME ; + int old_trans_id ; + + reiserfs_check_lock_depth("journal_begin") ; +#ifdef CONFIG_REISERFS_CHECK + if (p_s_sb->s_flags & MS_RDONLY) { + printk("clm-2078: calling journal_begin on readonly FS\n") ; + BUG() ; + } +#endif + + if (reiserfs_dont_log(p_s_sb)) { + th->t_super = p_s_sb ; /* others will check this for the don't log flag */ + return 0 ; + } + +relock: + lock_journal(p_s_sb) ; + + if (test_bit(WRITERS_BLOCKED, &SB_JOURNAL(p_s_sb)->j_state)) { + unlock_journal(p_s_sb) ; + reiserfs_wait_on_write_block(p_s_sb) ; + goto relock ; + } + + /* if there is no room in the journal OR + ** if this transaction is too old, and we weren't called joinable, wait for it to finish before beginning + ** we don't sleep if there aren't other writers + */ + + + if ( (!join && SB_JOURNAL(p_s_sb)->j_must_wait > 0) || + ( !join && (SB_JOURNAL(p_s_sb)->j_len_alloc + nblocks + 2) >= JOURNAL_MAX_BATCH) || + (!join && atomic_read(&(SB_JOURNAL(p_s_sb)->j_wcount)) > 0 && SB_JOURNAL(p_s_sb)->j_trans_start_time > 0 && + (now - SB_JOURNAL(p_s_sb)->j_trans_start_time) > JOURNAL_MAX_TRANS_AGE) || + (!join && atomic_read(&(SB_JOURNAL(p_s_sb)->j_jlock)) ) || + (!join && SB_JOURNAL(p_s_sb)->j_cnode_free < (JOURNAL_TRANS_MAX * 3))) { + + unlock_journal(p_s_sb) ; /* allow others to finish this transaction */ + + /* if writer count is 0, we can just force this transaction to end, and start + ** a new one afterwards. + */ + if (atomic_read(&(SB_JOURNAL(p_s_sb)->j_wcount)) <= 0) { + struct reiserfs_transaction_handle myth ; + journal_join(&myth, p_s_sb, 1) ; + reiserfs_prepare_for_journal(p_s_sb, SB_BUFFER_WITH_SB(p_s_sb), 1) ; + journal_mark_dirty(&myth, p_s_sb, SB_BUFFER_WITH_SB(p_s_sb)) ; + do_journal_end(&myth, p_s_sb,1,COMMIT_NOW) ; + } else { + /* but if the writer count isn't zero, we have to wait for the current writers to finish. + ** They won't batch on transaction end once we set j_jlock + */ + atomic_set(&(SB_JOURNAL(p_s_sb)->j_jlock), 1) ; + old_trans_id = SB_JOURNAL(p_s_sb)->j_trans_id ; + while(atomic_read(&(SB_JOURNAL(p_s_sb)->j_jlock)) && + SB_JOURNAL(p_s_sb)->j_trans_id == old_trans_id) { + sleep_on(&(SB_JOURNAL(p_s_sb)->j_join_wait)) ; + } + } + lock_journal(p_s_sb) ; /* relock to continue */ + } + + if (SB_JOURNAL(p_s_sb)->j_trans_start_time == 0) { /* we are the first writer, set trans_id */ + SB_JOURNAL(p_s_sb)->j_trans_start_time = now ; + } + atomic_inc(&(SB_JOURNAL(p_s_sb)->j_wcount)) ; + SB_JOURNAL(p_s_sb)->j_len_alloc += nblocks ; + th->t_blocks_logged = 0 ; + th->t_blocks_allocated = nblocks ; + th->t_super = p_s_sb ; + th->t_trans_id = SB_JOURNAL(p_s_sb)->j_trans_id ; + th->t_caller = "Unknown" ; + unlock_journal(p_s_sb) ; + p_s_sb->s_dirt = 1; + return 0 ; +} + + +int journal_join(struct reiserfs_transaction_handle *th, struct super_block *p_s_sb, unsigned long nblocks) { + return do_journal_begin_r(th, p_s_sb, nblocks, 1) ; +} + +int journal_begin(struct reiserfs_transaction_handle *th, struct super_block * p_s_sb, unsigned long nblocks) { + return do_journal_begin_r(th, p_s_sb, nblocks, 0) ; +} + +/* not used at all */ +int journal_prepare(struct super_block * p_s_sb, struct buffer_head *bh) { + return 0 ; +} + +/* +** puts bh into the current transaction. If it was already there, reorders removes the +** old pointers from the hash, and puts new ones in (to make sure replay happen in the right order). +** +** if it was dirty, cleans and files onto the clean list. I can't let it be dirty again until the +** transaction is committed. +** +** if j_len, is bigger than j_len_alloc, it pushes j_len_alloc to 10 + j_len. +*/ +int journal_mark_dirty(struct reiserfs_transaction_handle *th, struct super_block *p_s_sb, struct buffer_head *bh) { + struct reiserfs_journal_cnode *cn = NULL; + int count_already_incd = 0 ; + int prepared = 0 ; + + if (reiserfs_dont_log(th->t_super)) { + mark_buffer_dirty(bh) ; + return 0 ; + } + + if (th->t_trans_id != SB_JOURNAL(p_s_sb)->j_trans_id) { + reiserfs_panic(th->t_super, "journal-1577: handle trans id %d != current trans id %d\n", + th->t_trans_id, SB_JOURNAL(p_s_sb)->j_trans_id); + } + p_s_sb->s_dirt = 1 ; + + prepared = test_and_clear_bit(BH_JPrepared, &bh->b_state) ; + /* already in this transaction, we are done */ + if (buffer_journaled(bh)) { + return 0 ; + } + + /* this must be turned into a panic instead of a warning. We can't allow + ** a dirty or journal_dirty or locked buffer to be logged, as some changes + ** could get to disk too early. NOT GOOD. + */ + if (!prepared || buffer_locked(bh)) { + printk("journal-1777: buffer %lu bad state %cPREPARED %cLOCKED %cDIRTY %cJDIRTY_WAIT\n", bh->b_blocknr, prepared ? ' ' : '!', + buffer_locked(bh) ? ' ' : '!', + buffer_dirty(bh) ? ' ' : '!', + buffer_journal_dirty(bh) ? ' ' : '!') ; + show_reiserfs_locks() ; + } + count_already_incd = clear_prepared_bits(bh) ; + + if (atomic_read(&(SB_JOURNAL(p_s_sb)->j_wcount)) <= 0) { + printk("journal-1409: journal_mark_dirty returning because j_wcount was %d\n", atomic_read(&(SB_JOURNAL(p_s_sb)->j_wcount))) ; + return 1 ; + } + /* this error means I've screwed up, and we've overflowed the transaction. + ** Nothing can be done here, except make the FS readonly or panic. + */ + if (SB_JOURNAL(p_s_sb)->j_len >= JOURNAL_TRANS_MAX) { + reiserfs_panic(th->t_super, "journal-1413: journal_mark_dirty: j_len (%lu) is too big\n", SB_JOURNAL(p_s_sb)->j_len) ; + } + + if (buffer_journal_dirty(bh)) { + count_already_incd = 1 ; + mark_buffer_notjournal_dirty(bh) ; + } + + if (buffer_dirty(bh)) { + clear_bit(BH_Dirty, &bh->b_state) ; + } + + if (buffer_journaled(bh)) { /* must double check after getting lock */ + goto done ; + } + + if (SB_JOURNAL(p_s_sb)->j_len > SB_JOURNAL(p_s_sb)->j_len_alloc) { + SB_JOURNAL(p_s_sb)->j_len_alloc = SB_JOURNAL(p_s_sb)->j_len + JOURNAL_PER_BALANCE_CNT ; + } + + set_bit(BH_JDirty, &bh->b_state) ; + + /* now put this guy on the end */ + if (!cn) { + cn = get_cnode(p_s_sb) ; + if (!cn) { + reiserfs_panic(p_s_sb, "get_cnode failed!\n"); + } + + if (th->t_blocks_logged == th->t_blocks_allocated) { + th->t_blocks_allocated += JOURNAL_PER_BALANCE_CNT ; + SB_JOURNAL(p_s_sb)->j_len_alloc += JOURNAL_PER_BALANCE_CNT ; + } + th->t_blocks_logged++ ; + SB_JOURNAL(p_s_sb)->j_len++ ; + + cn->bh = bh ; + cn->blocknr = bh->b_blocknr ; + cn->dev = bh->b_dev ; + cn->jlist = NULL ; + insert_journal_hash(SB_JOURNAL(p_s_sb)->j_hash_table, cn) ; + if (!count_already_incd) { + atomic_inc(&(bh->b_count)) ; + } + } + cn->next = NULL ; + cn->prev = SB_JOURNAL(p_s_sb)->j_last ; + cn->bh = bh ; + if (SB_JOURNAL(p_s_sb)->j_last) { + SB_JOURNAL(p_s_sb)->j_last->next = cn ; + SB_JOURNAL(p_s_sb)->j_last = cn ; + } else { + SB_JOURNAL(p_s_sb)->j_first = cn ; + SB_JOURNAL(p_s_sb)->j_last = cn ; + } +done: + return 0 ; +} + +/* +** if buffer already in current transaction, do a journal_mark_dirty +** otherwise, just mark it dirty and move on. Used for writes to meta blocks +** that don't need journaling +*/ +int journal_mark_dirty_nolog(struct reiserfs_transaction_handle *th, struct super_block *p_s_sb, struct buffer_head *bh) { + if (reiserfs_dont_log(th->t_super) || buffer_journaled(bh) || + buffer_journal_dirty(bh)) { + return journal_mark_dirty(th, p_s_sb, bh) ; + } + if (get_journal_hash_dev(SB_JOURNAL(p_s_sb)->j_list_hash_table, bh->b_dev,bh->b_blocknr,bh->b_size)) { + return journal_mark_dirty(th, p_s_sb, bh) ; + } + mark_buffer_dirty(bh) ; + return 0 ; +} + +int journal_end(struct reiserfs_transaction_handle *th, struct super_block *p_s_sb, unsigned long nblocks) { + return do_journal_end(th, p_s_sb, nblocks, 0) ; +} + +/* removes from the current transaction, relsing and descrementing any counters. +** also files the removed buffer directly onto the clean list +** +** called by journal_mark_freed when a block has been deleted +** +** returns 1 if it cleaned and relsed the buffer. 0 otherwise +*/ +int remove_from_transaction(struct super_block *p_s_sb, unsigned long blocknr, int already_cleaned) { + struct buffer_head *bh ; + struct reiserfs_journal_cnode *cn ; + int ret = 0; + + cn = get_journal_hash_dev(SB_JOURNAL(p_s_sb)->j_hash_table, p_s_sb->s_dev, blocknr, p_s_sb->s_blocksize) ; + if (!cn || !cn->bh) { + return ret ; + } + bh = cn->bh ; + if (cn->prev) { + cn->prev->next = cn->next ; + } + if (cn->next) { + cn->next->prev = cn->prev ; + } + if (cn == SB_JOURNAL(p_s_sb)->j_first) { + SB_JOURNAL(p_s_sb)->j_first = cn->next ; + } + if (cn == SB_JOURNAL(p_s_sb)->j_last) { + SB_JOURNAL(p_s_sb)->j_last = cn->prev ; + } + remove_journal_hash(SB_JOURNAL(p_s_sb)->j_hash_table, NULL, bh, 0) ; + mark_buffer_not_journaled(bh) ; /* don't log this one */ + + if (!already_cleaned) { + mark_buffer_notjournal_dirty(bh) ; + atomic_dec(&(bh->b_count)) ; + if (atomic_read(&(bh->b_count)) < 0) { + printk("journal-1752: remove from trans, b_count < 0\n") ; + } + if (!buffer_locked(bh)) reiserfs_clean_and_file_buffer(bh) ; + ret = 1 ; + } + SB_JOURNAL(p_s_sb)->j_len-- ; + SB_JOURNAL(p_s_sb)->j_len_alloc-- ; + free_cnode(p_s_sb, cn) ; + return ret ; +} + +/* removes from a specific journal list hash */ +int remove_from_journal_list(struct super_block *s, struct reiserfs_journal_list *jl, struct buffer_head *bh, int remove_freed) { + remove_journal_hash(SB_JOURNAL(s)->j_list_hash_table, jl, bh, remove_freed) ; + return 0 ; +} + +/* +** for any cnode in a journal list, it can only be dirtied of all the +** transactions that include it are commited to disk. +** this checks through each transaction, and returns 1 if you are allowed to dirty, +** and 0 if you aren't +** +** it is called by dirty_journal_list, which is called after flush_commit_list has gotten all the log +** blocks for a given transaction on disk +** +*/ +static int can_dirty(struct reiserfs_journal_cnode *cn) { + kdev_t dev = cn->dev ; + unsigned long blocknr = cn->blocknr ; + struct reiserfs_journal_cnode *cur = cn->hprev ; + int can_dirty = 1 ; + + /* first test hprev. These are all newer than cn, so any node here + ** with the name block number and dev means this node can't be sent + ** to disk right now. + */ + while(cur && can_dirty) { + if (cur->jlist && cur->bh && cur->blocknr && cur->dev == dev && + cur->blocknr == blocknr) { + can_dirty = 0 ; + } + cur = cur->hprev ; + } + /* then test hnext. These are all older than cn. As long as they + ** are committed to the log, it is safe to write cn to disk + */ + cur = cn->hnext ; + while(cur && can_dirty) { + if (cur->jlist && cur->jlist->j_len > 0 && + atomic_read(&(cur->jlist->j_commit_left)) > 0 && cur->bh && + cur->blocknr && cur->dev == dev && cur->blocknr == blocknr) { + can_dirty = 0 ; + } + cur = cur->hnext ; + } + return can_dirty ; +} + +/* syncs the commit blocks, but does not force the real buffers to disk +** will wait until the current transaction is done/commited before returning +*/ +int journal_end_sync(struct reiserfs_transaction_handle *th, struct super_block *p_s_sb, unsigned long nblocks) { + return do_journal_end(th, p_s_sb, nblocks, COMMIT_NOW | WAIT) ; +} + +#ifdef __KERNEL__ +int show_reiserfs_locks(void) { + + dump_journal_writers() ; +#if 0 /* debugging code for when we are compiled static don't delete */ + p_s_sb = sb_entry(super_blocks.next); + while (p_s_sb != sb_entry(&super_blocks)) { + if (reiserfs_is_super(p_s_sb)) { +printk("journal lock is %d, join lock is %d, writers %d must wait is %d\n", + atomic_read(&(SB_JOURNAL(p_s_sb)->j_wlock)), + atomic_read(&(SB_JOURNAL(p_s_sb)->j_jlock)), + atomic_read(&(SB_JOURNAL(p_s_sb)->j_wcount)), + SB_JOURNAL(p_s_sb)->j_must_wait) ; + printk("used cnodes %d, free cnodes %d\n", SB_JOURNAL(p_s_sb)->j_cnode_used, SB_JOURNAL(p_s_sb)->j_cnode_free) ; + } + p_s_sb = sb_entry(p_s_sb->s_list.next); + } +#endif + return 0 ; +} +#endif + +/* +** used to get memory back from async commits that are floating around +** and to reclaim any blocks deleted but unusable because their commits +** haven't hit disk yet. called from bitmap.c +** +** if it starts flushing things, it ors SCHEDULE_OCCURRED into repeat. +** note, this is just if schedule has a chance of occuring. I need to +** change flush_commit_lists to have a repeat parameter too. +** +*/ +void flush_async_commits(struct super_block *p_s_sb) { + int i ; + + for (i = 0 ; i < JOURNAL_LIST_COUNT ; i++) { + if (i != SB_JOURNAL_LIST_INDEX(p_s_sb)) { + flush_commit_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + i, 1) ; + } + } +} + +/* +** flushes any old transactions to disk +** ends the current transaction if it is too old +** +** also calls flush_journal_list with old_only == 1, which allows me to reclaim +** memory and such from the journal lists whose real blocks are all on disk. +** +** called by sync_dev_journal from buffer.c +*/ +int flush_old_commits(struct super_block *p_s_sb, int immediate) { + int i ; + int count = 0; + int start ; + time_t now ; + int keep_dirty = 0 ; + struct reiserfs_transaction_handle th ; + + start = SB_JOURNAL_LIST_INDEX(p_s_sb) ; + now = CURRENT_TIME ; + + /* safety check so we don't flush while we are replaying the log during mount */ + if (SB_JOURNAL_LIST_INDEX(p_s_sb) < 0) { + return 0 ; + } + if (!strcmp(current->comm, "kupdate")) { + immediate = 0 ; + keep_dirty = 1 ; + } + /* starting with oldest, loop until we get to the start */ + i = (SB_JOURNAL_LIST_INDEX(p_s_sb) + 1) % JOURNAL_LIST_COUNT ; + while(i != start) { + if (SB_JOURNAL_LIST(p_s_sb)[i].j_len > 0 && ((now - SB_JOURNAL_LIST(p_s_sb)[i].j_timestamp) > JOURNAL_MAX_COMMIT_AGE || + immediate)) { + /* we have to check again to be sure the current transaction did not change */ + if (i != SB_JOURNAL_LIST_INDEX(p_s_sb)) { + flush_commit_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + i, 1) ; + } + } + i = (i + 1) % JOURNAL_LIST_COUNT ; + count++ ; + } + /* now, check the current transaction. If there are no writers, and it is too old, finish it, and + ** force the commit blocks to disk + */ + if (!immediate && atomic_read(&(SB_JOURNAL(p_s_sb)->j_wcount)) <= 0 && + SB_JOURNAL(p_s_sb)->j_trans_start_time > 0 && + SB_JOURNAL(p_s_sb)->j_len > 0 && + (now - SB_JOURNAL(p_s_sb)->j_trans_start_time) > JOURNAL_MAX_TRANS_AGE) { + journal_join(&th, p_s_sb, 1) ; + reiserfs_prepare_for_journal(p_s_sb, SB_BUFFER_WITH_SB(p_s_sb), 1) ; + journal_mark_dirty(&th, p_s_sb, SB_BUFFER_WITH_SB(p_s_sb)) ; + do_journal_end(&th, p_s_sb,1, COMMIT_NOW) ; + keep_dirty = 0 ; + } else if (immediate) { /* belongs above, but I wanted this to be very explicit as a special case. If they say to + flush, we must be sure old transactions hit the disk too. */ + journal_join(&th, p_s_sb, 1) ; + reiserfs_prepare_for_journal(p_s_sb, SB_BUFFER_WITH_SB(p_s_sb), 1) ; + journal_mark_dirty(&th, p_s_sb, SB_BUFFER_WITH_SB(p_s_sb)) ; + do_journal_end(&th, p_s_sb,1, COMMIT_NOW | WAIT) ; + } + keep_dirty |= reiserfs_journal_kupdate(p_s_sb) ; + return keep_dirty ; +} + +/* +** returns 0 if do_journal_end should return right away, returns 1 if do_journal_end should finish the commit +** +** if the current transaction is too old, but still has writers, this will wait on j_join_wait until all +** the writers are done. By the time it wakes up, the transaction it was called has already ended, so it just +** flushes the commit list and returns 0. +** +** Won't batch when flush or commit_now is set. Also won't batch when others are waiting on j_join_wait. +** +** Note, we can't allow the journal_end to proceed while there are still writers in the log. +*/ +static int check_journal_end(struct reiserfs_transaction_handle *th, struct super_block * p_s_sb, + unsigned long nblocks, int flags) { + + time_t now ; + int flush = flags & FLUSH_ALL ; + int commit_now = flags & COMMIT_NOW ; + int wait_on_commit = flags & WAIT ; + + if (th->t_trans_id != SB_JOURNAL(p_s_sb)->j_trans_id) { + reiserfs_panic(th->t_super, "journal-1577: handle trans id %d != current trans id %d\n", + th->t_trans_id, SB_JOURNAL(p_s_sb)->j_trans_id); + } + + SB_JOURNAL(p_s_sb)->j_len_alloc -= (th->t_blocks_allocated - th->t_blocks_logged) ; + if (atomic_read(&(SB_JOURNAL(p_s_sb)->j_wcount)) > 0) { /* <= 0 is allowed. unmounting might not call begin */ + atomic_dec(&(SB_JOURNAL(p_s_sb)->j_wcount)) ; + } + + /* BUG, deal with case where j_len is 0, but people previously freed blocks need to be released + ** will be dealt with by next transaction that actually writes something, but should be taken + ** care of in this trans + */ + if (SB_JOURNAL(p_s_sb)->j_len == 0) { + int wcount = atomic_read(&(SB_JOURNAL(p_s_sb)->j_wcount)) ; + unlock_journal(p_s_sb) ; + if (atomic_read(&(SB_JOURNAL(p_s_sb)->j_jlock)) > 0 && wcount <= 0) { + atomic_dec(&(SB_JOURNAL(p_s_sb)->j_jlock)) ; + wake_up(&(SB_JOURNAL(p_s_sb)->j_join_wait)) ; + } + return 0 ; + } + /* if wcount > 0, and we are called to with flush or commit_now, + ** we wait on j_join_wait. We will wake up when the last writer has + ** finished the transaction, and started it on its way to the disk. + ** Then, we flush the commit or journal list, and just return 0 + ** because the rest of journal end was already done for this transaction. + */ + if (atomic_read(&(SB_JOURNAL(p_s_sb)->j_wcount)) > 0) { + if (flush || commit_now) { + int orig_jindex = SB_JOURNAL_LIST_INDEX(p_s_sb) ; + atomic_set(&(SB_JOURNAL(p_s_sb)->j_jlock), 1) ; + if (flush) { + SB_JOURNAL(p_s_sb)->j_next_full_flush = 1 ; + } + unlock_journal(p_s_sb) ; + /* sleep while the current transaction is still j_jlocked */ + while(atomic_read(&(SB_JOURNAL(p_s_sb)->j_jlock)) && + SB_JOURNAL(p_s_sb)->j_trans_id == th->t_trans_id) { + sleep_on(&(SB_JOURNAL(p_s_sb)->j_join_wait)) ; + } + if (commit_now) { + if (wait_on_commit) { + flush_commit_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + orig_jindex, 1) ; + } else { + commit_flush_async(p_s_sb, orig_jindex) ; + } + } + return 0 ; + } + unlock_journal(p_s_sb) ; + return 0 ; + } + + /* deal with old transactions where we are the last writers */ + now = CURRENT_TIME ; + if ((now - SB_JOURNAL(p_s_sb)->j_trans_start_time) > JOURNAL_MAX_TRANS_AGE) { + commit_now = 1 ; + SB_JOURNAL(p_s_sb)->j_next_async_flush = 1 ; + } + /* don't batch when someone is waiting on j_join_wait */ + /* don't batch when syncing the commit or flushing the whole trans */ + if (!(SB_JOURNAL(p_s_sb)->j_must_wait > 0) && !(atomic_read(&(SB_JOURNAL(p_s_sb)->j_jlock))) && !flush && !commit_now && + (SB_JOURNAL(p_s_sb)->j_len < JOURNAL_MAX_BATCH) && + SB_JOURNAL(p_s_sb)->j_len_alloc < JOURNAL_MAX_BATCH && SB_JOURNAL(p_s_sb)->j_cnode_free > (JOURNAL_TRANS_MAX * 3)) { + SB_JOURNAL(p_s_sb)->j_bcount++ ; + unlock_journal(p_s_sb) ; + return 0 ; + } + + if (SB_JOURNAL(p_s_sb)->j_start > JOURNAL_BLOCK_COUNT) { + reiserfs_panic(p_s_sb, "journal-003: journal_end: j_start (%d) is too high\n", SB_JOURNAL(p_s_sb)->j_start) ; + } + return 1 ; +} + +/* +** Does all the work that makes deleting blocks safe. +** when deleting a block mark BH_JNew, just remove it from the current transaction, clean it's buffer_head and move on. +** +** otherwise: +** set a bit for the block in the journal bitmap. That will prevent it from being allocated for unformatted nodes +** before this transaction has finished. +** +** mark any cnodes for this block as BLOCK_FREED, and clear their bh pointers. That will prevent any old transactions with +** this block from trying to flush to the real location. Since we aren't removing the cnode from the journal_list_hash, +** the block can't be reallocated yet. +** +** Then remove it from the current transaction, decrementing any counters and filing it on the clean list. +*/ +int journal_mark_freed(struct reiserfs_transaction_handle *th, struct super_block *p_s_sb, unsigned long blocknr) { + struct reiserfs_journal_cnode *cn = NULL ; + struct buffer_head *bh = NULL ; + struct reiserfs_list_bitmap *jb = NULL ; + int cleaned = 0 ; + + if (reiserfs_dont_log(th->t_super)) { + bh = get_hash_table(p_s_sb->s_dev, blocknr, p_s_sb->s_blocksize) ; + if (bh && buffer_dirty (bh)) { + printk ("journal_mark_freed(dont_log): dirty buffer on hash list: %lx %ld\n", bh->b_state, blocknr); + BUG (); + } + brelse (bh); + return 0 ; + } + bh = get_hash_table(p_s_sb->s_dev, blocknr, p_s_sb->s_blocksize) ; + /* if it is journal new, we just remove it from this transaction */ + if (bh && buffer_journal_new(bh)) { + clear_prepared_bits(bh) ; + cleaned = remove_from_transaction(p_s_sb, blocknr, cleaned) ; + } else { + /* set the bit for this block in the journal bitmap for this transaction */ + jb = SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_list_bitmap ; + if (!jb) { + reiserfs_panic(p_s_sb, "journal-1702: journal_mark_freed, journal_list_bitmap is NULL\n") ; + } + set_bit_in_list_bitmap(p_s_sb, blocknr, jb) ; + + /* Note, the entire while loop is not allowed to schedule. */ + + if (bh) { + clear_prepared_bits(bh) ; + } + cleaned = remove_from_transaction(p_s_sb, blocknr, cleaned) ; + + /* find all older transactions with this block, make sure they don't try to write it out */ + cn = get_journal_hash_dev(SB_JOURNAL(p_s_sb)->j_list_hash_table, p_s_sb->s_dev, blocknr, p_s_sb->s_blocksize) ; + while (cn) { + if (p_s_sb->s_dev == cn->dev && blocknr == cn->blocknr) { + set_bit(BLOCK_FREED, &cn->state) ; + if (cn->bh) { + if (!cleaned) { + /* remove_from_transaction will brelse the buffer if it was + ** in the current trans + */ + mark_buffer_notjournal_dirty(cn->bh) ; + if (!buffer_locked(cn->bh)) { + reiserfs_clean_and_file_buffer(cn->bh) ; + } + cleaned = 1 ; + atomic_dec(&(cn->bh->b_count)) ; + if (atomic_read(&(cn->bh->b_count)) < 0) { + printk("journal-2138: cn->bh->b_count < 0\n") ; + } + } + if (cn->jlist) { /* since we are clearing the bh, we MUST dec nonzerolen */ + atomic_dec(&(cn->jlist->j_nonzerolen)) ; + } + cn->bh = NULL ; + } + } + cn = cn->hnext ; + } + } + + if (bh) { + atomic_dec(&(bh->b_count)) ; /* get_hash incs this */ + if (atomic_read(&(bh->b_count)) < 0) { + printk("journal-2165: bh->b_count < 0\n") ; + } + } + return 0 ; +} + +void reiserfs_restore_prepared_buffer(struct super_block *p_s_sb, + struct buffer_head *bh) { + if (reiserfs_dont_log (p_s_sb)) + return; + + if (!bh) { + return ; + } + clear_bit(BH_JPrepared, &bh->b_state) ; +} + +extern struct tree_balance *cur_tb ; +/* +** before we can change a metadata block, we have to make sure it won't +** be written to disk while we are altering it. So, we must: +** clean it +** wait on it. +** +*/ +void reiserfs_prepare_for_journal(struct super_block *p_s_sb, + struct buffer_head *bh, int wait) { + int retry_count = 0 ; + + if (reiserfs_dont_log (p_s_sb)) + return; + + while(!test_bit(BH_JPrepared, &bh->b_state) || + (wait && buffer_locked(bh))) { + if (buffer_journaled(bh)) { + set_bit(BH_JPrepared, &bh->b_state) ; + return ; + } + set_bit(BH_JPrepared, &bh->b_state) ; + if (wait) { +#ifdef CONFIG_REISERFS_CHECK + if (buffer_locked(bh) && cur_tb != NULL) { + printk("reiserfs_prepare_for_journal, waiting while do_balance was running\n") ; + BUG() ; + } +#endif + wait_on_buffer(bh) ; + } + retry_count++ ; + } +} + +/* + * Wait for a page to get unlocked. + * + * This must be called with the caller "holding" the page, + * ie with increased "page->count" so that the page won't + * go away during the wait.. + */ +static void ___reiserfs_wait_on_page(struct reiserfs_page_list *pl) +{ + struct task_struct *tsk = current; + struct page *page = pl->page ; + DECLARE_WAITQUEUE(wait, tsk); + + add_wait_queue(&page->wait, &wait); + do { + block_sync_page(page); + set_task_state(tsk, TASK_UNINTERRUPTIBLE); + if (!PageLocked(page) || pl->do_not_lock) + break; + schedule(); + } while (PageLocked(page)); + tsk->state = TASK_RUNNING; + remove_wait_queue(&page->wait, &wait); +} + +/* + * Get an exclusive lock on the page.. + * but, every time you get woken up, check the page to make sure + * someone hasn't called a journal_begin with it locked. + * + * the page should always be locked when this returns + * + * returns 0 if you've got the page locked + * returns 1 if it returns because someone else has called journal_begin + * with the page locked + * this is only useful to the code that flushes pages before a + * commit. Do not export this hack. Ever. + */ +static int reiserfs_try_lock_page(struct reiserfs_page_list *pl) +{ + struct page *page = pl->page ; + while (TryLockPage(page)) { + if (pl->do_not_lock) { + /* the page is locked, but we cannot have it */ + return 1 ; + } + ___reiserfs_wait_on_page(pl); + } + /* we have the page locked */ + return 0 ; +} + + +/* +** This can only be called from do_journal_end. +** it runs through the list things that need flushing before the +** transaction can commit, and writes each of them to disk +** +*/ + +static void flush_pages_before_commit(struct reiserfs_transaction_handle *th, + struct super_block *p_s_sb) { + struct reiserfs_page_list *pl = SB_JOURNAL(p_s_sb)->j_flush_pages ; + struct reiserfs_page_list *pl_tmp ; + struct buffer_head *bh, *head ; + int count = 0 ; + + /* first write each dirty unlocked buffer in the list */ + + while(pl) { + /* ugly. journal_end can be called from get_block, which has a + ** page locked. So, we have to check to see if pl->page is the page + ** currently locked by the calling function, and if so, skip the + ** lock + */ + if (reiserfs_try_lock_page(pl)) { + goto setup_next ; + } + if (!PageLocked(pl->page)) { + BUG() ; + } + if (pl->page->buffers) { + head = pl->page->buffers ; + bh = head ; + do { + if (bh->b_blocknr == pl->blocknr && buffer_dirty(bh) && + !buffer_locked(bh) && buffer_uptodate(bh) ) { + ll_rw_block(WRITE, 1, &bh) ; + } + bh = bh->b_this_page ; + } while (bh != head) ; + } + if (!pl->do_not_lock) { + UnlockPage(pl->page) ; + } +setup_next: + pl = pl->next ; + } + + /* now wait on them */ + + pl = SB_JOURNAL(p_s_sb)->j_flush_pages ; + while(pl) { + if (reiserfs_try_lock_page(pl)) { + goto remove_page ; + } + if (!PageLocked(pl->page)) { + BUG() ; + } + if (pl->page->buffers) { + head = pl->page->buffers ; + bh = head ; + do { + if (bh->b_blocknr == pl->blocknr) { + count++ ; + wait_on_buffer(bh) ; + if (!buffer_uptodate(bh)) { + reiserfs_panic(p_s_sb, "journal-2443: flush_pages_before_commit, error writing block %lu\n", bh->b_blocknr) ; + } + } + bh = bh->b_this_page ; + } while (bh != head) ; + } + if (!pl->do_not_lock) { + UnlockPage(pl->page) ; + } +remove_page: + /* we've waited on the I/O, we can remove the page from the + ** list, and free our pointer struct to it. + */ + if (pl->prev) { + pl->prev->next = pl->next ; + } + if (pl->next) { + pl->next->prev = pl->prev ; + } + put_page(pl->page) ; + pl_tmp = pl ; + pl = pl->next ; + reiserfs_kfree(pl_tmp, sizeof(struct reiserfs_page_list), p_s_sb) ; + } + SB_JOURNAL(p_s_sb)->j_flush_pages = NULL ; +} + +/* +** called when a indirect item is converted back into a tail. +** +** The reiserfs part of the inode stores enough information to find +** our page_list struct in the flush list. We remove it from the list +** and free the struct. +** +** Note, it is possible for this to happen: +** +** reiserfs_add_page_to_flush_list(inode) +** transaction ends, list is flushed +** reiserfs_remove_page_from_flush_list(inode) +** +** This would be bad because the page_list pointer in the inode is not +** updated when the list is flushed, so we can't know if the pointer is +** valid. So, in the inode, we also store the transaction id when the +** page was added. If we are trying to remove something from an old +** transaction, we just clear out the pointer in the inode and return. +** +** Normal case is to use the reiserfs_page_list pointer in the inode to +** find and remove the page from the flush list. +*/ +int reiserfs_remove_page_from_flush_list(struct reiserfs_transaction_handle *th, + struct inode *inode) { + struct reiserfs_page_list *pl ; + + /* was this conversion done in a previous transaction? If so, return */ + if (inode->u.reiserfs_i.i_conversion_trans_id < th->t_trans_id) { + inode->u.reiserfs_i.i_converted_page = NULL ; + inode->u.reiserfs_i.i_conversion_trans_id = 0 ; + return 0 ; + } + + /* remove the page_list struct from the list, release our hold on the + ** page, and free the page_list struct + */ + pl = inode->u.reiserfs_i.i_converted_page ; + if (pl) { + if (pl->next) { + pl->next->prev = pl->prev ; + } + if (pl->prev) { + pl->prev->next = pl->next ; + } + if (SB_JOURNAL(inode->i_sb)->j_flush_pages == pl) { + SB_JOURNAL(inode->i_sb)->j_flush_pages = pl->next ; + } + put_page(pl->page) ; + reiserfs_kfree(pl, sizeof(struct reiserfs_page_list), inode->i_sb) ; + inode->u.reiserfs_i.i_converted_page = NULL ; + inode->u.reiserfs_i.i_conversion_trans_id = 0 ; + } + return 0 ; +} + +/* +** Called after a direct to indirect transaction. The unformatted node +** must be flushed to disk before the transaction commits, otherwise, we +** risk losing the data from the direct item. This adds the page +** containing the unformatted node to a list of pages that need flushing. +** +** it calls get_page(page), so the page won't disappear until we've +** flushed or removed it from our list. +** +** pointers to the reiserfs_page_list struct are stored in the inode, +** so this page can be quickly removed from the list after the tail is +** converted back into a direct item. +** +** If we fail to find the memory for the reiserfs_page_list struct, we +** just sync the page now. Not good, but safe. +** +** since this must be called with the page locked, we always set +** the do_not_lock field in the page_list struct we allocate +** +*/ +int reiserfs_add_page_to_flush_list(struct reiserfs_transaction_handle *th, + struct inode *inode, + struct buffer_head *bh) { + struct reiserfs_page_list *new_pl ; + +/* debugging use ONLY. Do not define this on data you care about. */ +#ifdef REISERFS_NO_FLUSH_AFTER_CONVERT + return 0 ; +#endif + + get_page(bh->b_page) ; + new_pl = reiserfs_kmalloc(sizeof(struct reiserfs_page_list), GFP_BUFFER, + inode->i_sb) ; + if (!new_pl) { + put_page(bh->b_page) ; + reiserfs_warning("journal-2480: forced to flush page, out of memory\n") ; + ll_rw_block(WRITE, 1, &bh) ; + wait_on_buffer(bh) ; + if (!buffer_uptodate(bh)) { + reiserfs_panic(inode->i_sb, "journal-2484: error writing buffer %lu to disk\n", bh->b_blocknr) ; + } + inode->u.reiserfs_i.i_converted_page = NULL ; + return 0 ; + } + + new_pl->page = bh->b_page ; + new_pl->do_not_lock = 1 ; + new_pl->blocknr = bh->b_blocknr ; + new_pl->next = SB_JOURNAL(inode->i_sb)->j_flush_pages; + if (new_pl->next) { + new_pl->next->prev = new_pl ; + } + new_pl->prev = NULL ; + SB_JOURNAL(inode->i_sb)->j_flush_pages = new_pl ; + + /* if we have numbers from an old transaction, zero the converted + ** page, it has already been flushed and freed + */ + if (inode->u.reiserfs_i.i_conversion_trans_id && + inode->u.reiserfs_i.i_conversion_trans_id < th->t_trans_id) { + inode->u.reiserfs_i.i_converted_page = NULL ; + } + if (inode->u.reiserfs_i.i_converted_page) { + reiserfs_panic(inode->i_sb, "journal-2501: inode already had a converted page\n") ; + } + inode->u.reiserfs_i.i_converted_page = new_pl ; + inode->u.reiserfs_i.i_conversion_trans_id = th->t_trans_id ; + return 0 ; +} + +/* +** long and ugly. If flush, will not return until all commit +** blocks and all real buffers in the trans are on disk. +** If no_async, won't return until all commit blocks are on disk. +** +** keep reading, there are comments as you go along +*/ +static int do_journal_end(struct reiserfs_transaction_handle *th, struct super_block * p_s_sb, unsigned long nblocks, + int flags) { + struct reiserfs_journal_cnode *cn, *next, *jl_cn; + struct reiserfs_journal_cnode *last_cn = NULL; + struct reiserfs_journal_desc *desc ; + struct reiserfs_journal_commit *commit ; + struct buffer_head *c_bh ; /* commit bh */ + struct buffer_head *d_bh ; /* desc bh */ + int cur_write_start = 0 ; /* start index of current log write */ + int cur_blocks_left = 0 ; /* number of journal blocks left to write */ + int old_start ; + int i ; + int jindex ; + int orig_jindex ; + int flush = flags & FLUSH_ALL ; + int commit_now = flags & COMMIT_NOW ; + int wait_on_commit = flags & WAIT ; + struct reiserfs_super_block *rs ; + + if (reiserfs_dont_log(th->t_super)) { + return 0 ; + } + + lock_journal(p_s_sb) ; + if (SB_JOURNAL(p_s_sb)->j_next_full_flush) { + flags |= FLUSH_ALL ; + flush = 1 ; + } + if (SB_JOURNAL(p_s_sb)->j_next_async_flush) { + flags |= COMMIT_NOW ; + commit_now = 1 ; + } + + /* check_journal_end locks the journal, and unlocks if it does not return 1 + ** it tells us if we should continue with the journal_end, or just return + */ + if (!check_journal_end(th, p_s_sb, nblocks, flags)) { + return 0 ; + } + + /* check_journal_end might set these, check again */ + if (SB_JOURNAL(p_s_sb)->j_next_full_flush) { + flush = 1 ; + } + if (SB_JOURNAL(p_s_sb)->j_next_async_flush) { + commit_now = 1 ; + } + /* + ** j must wait means we have to flush the log blocks, and the real blocks for + ** this transaction + */ + if (SB_JOURNAL(p_s_sb)->j_must_wait > 0) { + flush = 1 ; + } + + rs = SB_DISK_SUPER_BLOCK(p_s_sb) ; + /* setup description block */ + d_bh = getblk(p_s_sb->s_dev, reiserfs_get_journal_block(p_s_sb) + SB_JOURNAL(p_s_sb)->j_start, p_s_sb->s_blocksize) ; + mark_buffer_uptodate(d_bh, 1) ; + desc = (struct reiserfs_journal_desc *)(d_bh)->b_data ; + memset(desc, 0, sizeof(struct reiserfs_journal_desc)) ; + memcpy(desc->j_magic, JOURNAL_DESC_MAGIC, 8) ; + desc->j_trans_id = cpu_to_le32(SB_JOURNAL(p_s_sb)->j_trans_id) ; + + /* setup commit block. Don't write (keep it clean too) this one until after everyone else is written */ + c_bh = getblk(p_s_sb->s_dev, reiserfs_get_journal_block(p_s_sb) + + ((SB_JOURNAL(p_s_sb)->j_start + SB_JOURNAL(p_s_sb)->j_len + 1) % JOURNAL_BLOCK_COUNT), + p_s_sb->s_blocksize) ; + commit = (struct reiserfs_journal_commit *)c_bh->b_data ; + memset(commit, 0, sizeof(struct reiserfs_journal_commit)) ; + commit->j_trans_id = cpu_to_le32(SB_JOURNAL(p_s_sb)->j_trans_id) ; + mark_buffer_uptodate(c_bh, 1) ; + + /* init this journal list */ + atomic_set(&(SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_older_commits_done), 0) ; + SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_trans_id = SB_JOURNAL(p_s_sb)->j_trans_id ; + SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_timestamp = SB_JOURNAL(p_s_sb)->j_trans_start_time ; + SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_commit_bh = c_bh ; + SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_start = SB_JOURNAL(p_s_sb)->j_start ; + SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_len = SB_JOURNAL(p_s_sb)->j_len ; + atomic_set(&(SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_nonzerolen), SB_JOURNAL(p_s_sb)->j_len) ; + atomic_set(&(SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_commit_left), SB_JOURNAL(p_s_sb)->j_len + 2); + SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_realblock = NULL ; + atomic_set(&(SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_commit_flushing), 1) ; + atomic_set(&(SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_flushing), 1) ; + + /* which is faster, locking/unlocking at the start and end of the for + ** or locking once per iteration around the insert_journal_hash? + ** eitherway, we are write locking insert_journal_hash. The ENTIRE FOR + ** LOOP MUST not cause schedule to occur. + */ + + /* for each real block, add it to the journal list hash, + ** copy into real block index array in the commit or desc block + */ + for (i = 0, cn = SB_JOURNAL(p_s_sb)->j_first ; cn ; cn = cn->next, i++) { + if (test_bit(BH_JDirty, &cn->bh->b_state) ) { + jl_cn = get_cnode(p_s_sb) ; + if (!jl_cn) { + reiserfs_panic(p_s_sb, "journal-1676, get_cnode returned NULL\n") ; + } + if (i == 0) { + SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_realblock = jl_cn ; + } + jl_cn->prev = last_cn ; + jl_cn->next = NULL ; + if (last_cn) { + last_cn->next = jl_cn ; + } + last_cn = jl_cn ; + if (cn->bh->b_blocknr >= reiserfs_get_journal_block(p_s_sb) && + cn->bh->b_blocknr < (reiserfs_get_journal_block(p_s_sb) + JOURNAL_BLOCK_COUNT)) { + reiserfs_panic(p_s_sb, "journal-2332: Trying to log block %lu, which is a log block\n", cn->bh->b_blocknr) ; + } + jl_cn->blocknr = cn->bh->b_blocknr ; + jl_cn->state = 0 ; + jl_cn->dev = cn->bh->b_dev ; + jl_cn->bh = cn->bh ; + jl_cn->jlist = SB_JOURNAL_LIST(p_s_sb) + SB_JOURNAL_LIST_INDEX(p_s_sb) ; + insert_journal_hash(SB_JOURNAL(p_s_sb)->j_list_hash_table, jl_cn) ; + if (i < JOURNAL_TRANS_HALF) { + desc->j_realblock[i] = cpu_to_le32(cn->bh->b_blocknr) ; + } else { + commit->j_realblock[i - JOURNAL_TRANS_HALF] = cpu_to_le32(cn->bh->b_blocknr) ; + } + } else { + i-- ; + } + } + + desc->j_len = cpu_to_le32(SB_JOURNAL(p_s_sb)->j_len) ; + desc->j_mount_id = cpu_to_le32(SB_JOURNAL(p_s_sb)->j_mount_id) ; + desc->j_trans_id = cpu_to_le32(SB_JOURNAL(p_s_sb)->j_trans_id) ; + commit->j_len = cpu_to_le32(SB_JOURNAL(p_s_sb)->j_len) ; + + /* special check in case all buffers in the journal were marked for not logging */ + if (SB_JOURNAL(p_s_sb)->j_len == 0) { + brelse(d_bh) ; + brelse(c_bh) ; + unlock_journal(p_s_sb) ; +printk("journal-2020: do_journal_end: BAD desc->j_len is ZERO\n") ; + atomic_set(&(SB_JOURNAL(p_s_sb)->j_jlock), 0) ; + wake_up(&(SB_JOURNAL(p_s_sb)->j_join_wait)) ; + return 0 ; + } + + /* first data block is j_start + 1, so add one to cur_write_start wherever you use it */ + cur_write_start = SB_JOURNAL(p_s_sb)->j_start ; + cur_blocks_left = SB_JOURNAL(p_s_sb)->j_len ; + cn = SB_JOURNAL(p_s_sb)->j_first ; + jindex = 1 ; /* start at one so we don't get the desc again */ + while(cur_blocks_left > 0) { + /* copy all the real blocks into log area. dirty log blocks */ + if (test_bit(BH_JDirty, &cn->bh->b_state)) { + struct buffer_head *tmp_bh ; + tmp_bh = getblk(p_s_sb->s_dev, reiserfs_get_journal_block(p_s_sb) + + ((cur_write_start + jindex) % JOURNAL_BLOCK_COUNT), + p_s_sb->s_blocksize) ; + mark_buffer_uptodate(tmp_bh, 1) ; + memcpy(tmp_bh->b_data, cn->bh->b_data, cn->bh->b_size) ; + jindex++ ; + } else { + /* JDirty cleared sometime during transaction. don't log this one */ + printk("journal-2048: do_journal_end: BAD, buffer in journal hash, but not JDirty!\n") ; + } + cn = cn->next ; + cur_blocks_left-- ; + } + + /* we are done with both the c_bh and d_bh, but + ** c_bh must be written after all other commit blocks, + ** so we dirty/relse c_bh in flush_commit_list, with commit_left <= 1. + */ + + /* now loop through and mark all buffers from this transaction as JDirty_wait + ** clear the JDirty bit, clear BH_JNew too. + ** if they weren't JDirty, they weren't logged, just relse them and move on + */ + cn = SB_JOURNAL(p_s_sb)->j_first ; + while(cn) { + clear_bit(BH_JNew, &(cn->bh->b_state)) ; + if (test_bit(BH_JDirty, &(cn->bh->b_state))) { + set_bit(BH_JDirty_wait, &(cn->bh->b_state)) ; + clear_bit(BH_JDirty, &(cn->bh->b_state)) ; + } else { + brelse(cn->bh) ; + } + next = cn->next ; + free_cnode(p_s_sb, cn) ; + cn = next ; + } + + /* unlock the journal list for committing and flushing */ + atomic_set(&(SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_commit_flushing), 0) ; + atomic_set(&(SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_flushing), 0) ; + + orig_jindex = SB_JOURNAL_LIST_INDEX(p_s_sb) ; + jindex = (SB_JOURNAL_LIST_INDEX(p_s_sb) + 1) % JOURNAL_LIST_COUNT ; + SB_JOURNAL_LIST_INDEX(p_s_sb) = jindex ; + + /* make sure to flush any data converted from direct items to + ** indirect items before allowing the commit blocks to reach the + ** disk + */ + flush_pages_before_commit(th, p_s_sb) ; + + /* honor the flush and async wishes from the caller */ + if (flush) { + + flush_commit_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + orig_jindex, 1) ; + flush_journal_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + orig_jindex , 1) ; + } else if (commit_now) { + if (wait_on_commit) { + flush_commit_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + orig_jindex, 1) ; + } else { + commit_flush_async(p_s_sb, orig_jindex) ; + } + } + + /* reset journal values for the next transaction */ + old_start = SB_JOURNAL(p_s_sb)->j_start ; + SB_JOURNAL(p_s_sb)->j_start = (SB_JOURNAL(p_s_sb)->j_start + SB_JOURNAL(p_s_sb)->j_len + 2) % JOURNAL_BLOCK_COUNT; + atomic_set(&(SB_JOURNAL(p_s_sb)->j_wcount), 0) ; + SB_JOURNAL(p_s_sb)->j_bcount = 0 ; + SB_JOURNAL(p_s_sb)->j_last = NULL ; + SB_JOURNAL(p_s_sb)->j_first = NULL ; + SB_JOURNAL(p_s_sb)->j_len = 0 ; + SB_JOURNAL(p_s_sb)->j_trans_start_time = 0 ; + SB_JOURNAL(p_s_sb)->j_trans_id++ ; + SB_JOURNAL(p_s_sb)->j_must_wait = 0 ; + SB_JOURNAL(p_s_sb)->j_len_alloc = 0 ; + SB_JOURNAL(p_s_sb)->j_next_full_flush = 0 ; + SB_JOURNAL(p_s_sb)->j_next_async_flush = 0 ; + init_journal_hash(p_s_sb) ; + + /* if the next transaction has any chance of wrapping, flush + ** transactions that might get overwritten. If any journal lists are very + ** old flush them as well. + */ + for (i = 0 ; i < JOURNAL_LIST_COUNT ; i++) { + jindex = i ; + if (SB_JOURNAL_LIST(p_s_sb)[jindex].j_len > 0 && SB_JOURNAL(p_s_sb)->j_start <= SB_JOURNAL_LIST(p_s_sb)[jindex].j_start) { + if ((SB_JOURNAL(p_s_sb)->j_start + JOURNAL_TRANS_MAX + 1) >= SB_JOURNAL_LIST(p_s_sb)[jindex].j_start) { + flush_journal_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + jindex, 1) ; + } + } else if (SB_JOURNAL_LIST(p_s_sb)[jindex].j_len > 0 && + (SB_JOURNAL(p_s_sb)->j_start + JOURNAL_TRANS_MAX + 1) > JOURNAL_BLOCK_COUNT) { + if (((SB_JOURNAL(p_s_sb)->j_start + JOURNAL_TRANS_MAX + 1) % JOURNAL_BLOCK_COUNT) >= + SB_JOURNAL_LIST(p_s_sb)[jindex].j_start) { + flush_journal_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + jindex, 1 ) ; + } + } + /* this check should always be run, to send old lists to disk */ + if (SB_JOURNAL_LIST(p_s_sb)[jindex].j_len > 0 && + SB_JOURNAL_LIST(p_s_sb)[jindex].j_timestamp < + (CURRENT_TIME - (JOURNAL_MAX_TRANS_AGE * 4))) { + flush_journal_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + jindex, 1 ) ; + } + } + + /* if the next journal_list is still in use, flush it */ + if (SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_len != 0) { + flush_journal_list(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + SB_JOURNAL_LIST_INDEX(p_s_sb), 1) ; + } + + /* we don't want anyone flushing the new transaction's list */ + atomic_set(&(SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_commit_flushing), 1) ; + atomic_set(&(SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_flushing), 1) ; + SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_list_bitmap = get_list_bitmap(p_s_sb, SB_JOURNAL_LIST(p_s_sb) + + SB_JOURNAL_LIST_INDEX(p_s_sb)) ; + + if (!(SB_JOURNAL_LIST(p_s_sb)[SB_JOURNAL_LIST_INDEX(p_s_sb)].j_list_bitmap)) { + reiserfs_panic(p_s_sb, "journal-1996: do_journal_end, could not get a list bitmap\n") ; + } + unlock_journal(p_s_sb) ; + atomic_set(&(SB_JOURNAL(p_s_sb)->j_jlock), 0) ; + /* wake up any body waiting to join. */ + wake_up(&(SB_JOURNAL(p_s_sb)->j_join_wait)) ; + return 0 ; +} + + + diff --git a/fs/reiserfs/lbalance.c b/fs/reiserfs/lbalance.c new file mode 100644 index 000000000..c18f3bdd0 --- /dev/null +++ b/fs/reiserfs/lbalance.c @@ -0,0 +1,1326 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <asm/uaccess.h> +#include <linux/string.h> +#include <linux/sched.h> +#include <linux/reiserfs_fs.h> + +#else + +#include "nokernel.h" + +#endif + +/* these are used in do_balance.c */ + +/* leaf_move_items + leaf_shift_left + leaf_shift_right + leaf_delete_items + leaf_insert_into_buf + leaf_paste_in_buffer + leaf_cut_from_buffer + leaf_paste_entries + */ + + +/* copy copy_count entries from source directory item to dest buffer (creating new item if needed) */ +static void leaf_copy_dir_entries (struct buffer_info * dest_bi, struct buffer_head * source, + int last_first, int item_num, int from, int copy_count) +{ + struct buffer_head * dest = dest_bi->bi_bh; + int item_num_in_dest; /* either the number of target item, + or if we must create a new item, + the number of the item we will + create it next to */ + struct item_head * ih; + struct reiserfs_de_head * deh; + int copy_records_len; /* length of all records in item to be copied */ + char * records; + + ih = B_N_PITEM_HEAD (source, item_num); + +#ifdef CONFIG_REISERFS_CHECK + if (!is_direntry_le_ih (ih)) + reiserfs_panic(0, "vs-10000: leaf_copy_dir_entries: item must be directory item"); +#endif + + /* length of all record to be copied and first byte of the last of them */ + deh = B_I_DEH (source, ih); + if (copy_count) { + copy_records_len = (from ? deh[from - 1].deh_location : ih->ih_item_len) - + deh[from + copy_count - 1].deh_location; + records = source->b_data + ih->ih_item_location + deh[from + copy_count - 1].deh_location; + } else { + copy_records_len = 0; + records = 0; + } + + /* when copy last to first, dest buffer can contain 0 items */ + item_num_in_dest = (last_first == LAST_TO_FIRST) ? (( B_NR_ITEMS(dest) ) ? 0 : -1) : (B_NR_ITEMS(dest) - 1); + + /* if there are no items in dest or the first/last item in dest is not item of the same directory */ + if ( (item_num_in_dest == - 1) || +#ifdef REISERFS_FSCK + (last_first == FIRST_TO_LAST && are_items_mergeable (B_N_PITEM_HEAD (dest, item_num_in_dest), ih, dest->b_size) == 0) || + (last_first == LAST_TO_FIRST && are_items_mergeable (ih, B_N_PITEM_HEAD (dest, item_num_in_dest), dest->b_size) == 0)) { +#else + (last_first == FIRST_TO_LAST && le_key_k_offset (ih_version (ih), &(ih->ih_key)) == DOT_OFFSET) || + (last_first == LAST_TO_FIRST && comp_short_le_keys/*COMP_SHORT_KEYS*/ (&ih->ih_key, B_N_PKEY (dest, item_num_in_dest)))) { +#endif + /* create new item in dest */ + struct item_head new_ih; + + /* form item header */ + memcpy (&new_ih.ih_key, &ih->ih_key, KEY_SIZE); + new_ih.ih_version = cpu_to_le16 (ITEM_VERSION_1); + /* calculate item len */ + new_ih.ih_item_len = cpu_to_le16 (DEH_SIZE * copy_count + copy_records_len); + I_ENTRY_COUNT(&new_ih) = 0; + + if (last_first == LAST_TO_FIRST) { + /* form key by the following way */ + if (from < I_ENTRY_COUNT(ih)) { + set_le_ih_k_offset (&new_ih, cpu_to_le32 (le32_to_cpu (deh[from].deh_offset))); + /*memcpy (&new_ih.ih_key.k_offset, &deh[from].deh_offset, SHORT_KEY_SIZE);*/ + } else { + /* no entries will be copied to this item in this function */ + set_le_ih_k_offset (&new_ih, cpu_to_le32 (U32_MAX)); + /* this item is not yet valid, but we want I_IS_DIRECTORY_ITEM to return 1 for it, so we -1 */ + } + set_le_key_k_type (ITEM_VERSION_1, &(new_ih.ih_key), TYPE_DIRENTRY); + } + + /* insert item into dest buffer */ + leaf_insert_into_buf (dest_bi, (last_first == LAST_TO_FIRST) ? 0 : B_NR_ITEMS(dest), &new_ih, NULL, 0); + } else { + /* prepare space for entries */ + leaf_paste_in_buffer (dest_bi, (last_first==FIRST_TO_LAST) ? (B_NR_ITEMS(dest) - 1) : 0, MAX_US_INT, + DEH_SIZE * copy_count + copy_records_len, records, 0 + ); + } + + item_num_in_dest = (last_first == FIRST_TO_LAST) ? (B_NR_ITEMS(dest)-1) : 0; + + leaf_paste_entries (dest_bi->bi_bh, item_num_in_dest, + (last_first == FIRST_TO_LAST) ? I_ENTRY_COUNT(B_N_PITEM_HEAD (dest, item_num_in_dest)) : 0, + copy_count, deh + from, records, + DEH_SIZE * copy_count + copy_records_len + ); +} + + +/* Copy the first (if last_first == FIRST_TO_LAST) or last (last_first == LAST_TO_FIRST) item or + part of it or nothing (see the return 0 below) from SOURCE to the end + (if last_first) or beginning (!last_first) of the DEST */ +/* returns 1 if anything was copied, else 0 */ +static int leaf_copy_boundary_item (struct buffer_info * dest_bi, struct buffer_head * src, int last_first, + int bytes_or_entries) +{ + struct buffer_head * dest = dest_bi->bi_bh; + int dest_nr_item, src_nr_item; /* number of items in the source and destination buffers */ + struct item_head * ih; + struct item_head * dih; + + dest_nr_item = B_NR_ITEMS(dest); + + if ( last_first == FIRST_TO_LAST ) { + /* if ( DEST is empty or first item of SOURCE and last item of DEST are the items of different objects + or of different types ) then there is no need to treat this item differently from the other items + that we copy, so we return */ + ih = B_N_PITEM_HEAD (src, 0); + dih = B_N_PITEM_HEAD (dest, dest_nr_item - 1); +#ifdef REISERFS_FSCK + if (!dest_nr_item || (are_items_mergeable (dih, ih, src->b_size) == 0)) +#else + if (!dest_nr_item || (!op_is_left_mergeable (&(ih->ih_key), src->b_size))) +#endif + /* there is nothing to merge */ + return 0; + +#ifdef CONFIG_REISERFS_CHECK + if ( ! ih->ih_item_len ) + reiserfs_panic (0, "vs-10010: leaf_copy_boundary_item: item can not have empty dynamic length"); +#endif + + if ( is_direntry_le_ih (ih) ) { + if ( bytes_or_entries == -1 ) + /* copy all entries to dest */ + bytes_or_entries = le16_to_cpu (ih->u.ih_entry_count); + leaf_copy_dir_entries (dest_bi, src, FIRST_TO_LAST, 0, 0, bytes_or_entries); + return 1; + } + + /* copy part of the body of the first item of SOURCE to the end of the body of the last item of the DEST + part defined by 'bytes_or_entries'; if bytes_or_entries == -1 copy whole body; don't create new item header + */ + if ( bytes_or_entries == -1 ) + bytes_or_entries = le16_to_cpu (ih->ih_item_len); + +#ifdef CONFIG_REISERFS_CHECK + else { + if (bytes_or_entries == le16_to_cpu (ih->ih_item_len) && is_indirect_le_ih(ih)) + if (get_ih_free_space (ih)) + reiserfs_panic (0, "vs-10020: leaf_copy_boundary_item: " + "last unformatted node must be filled entirely (%h)", + ih); + } +#endif + + /* merge first item (or its part) of src buffer with the last + item of dest buffer. Both are of the same file */ + leaf_paste_in_buffer (dest_bi, + dest_nr_item - 1, dih->ih_item_len, bytes_or_entries, B_I_PITEM(src,ih), 0 + ); + + if (is_indirect_le_ih (dih)) { +#ifdef CONFIG_REISERFS_CHECK + if (get_ih_free_space (dih)) + reiserfs_panic (0, "vs-10030: leaf_copy_boundary_item: " + "merge to left: last unformatted node of non-last indirect item %h must have zerto free space", + ih); +#endif + if (bytes_or_entries == le16_to_cpu (ih->ih_item_len)) + set_ih_free_space (dih, get_ih_free_space (ih)); + } + + return 1; + } + + + /* copy boundary item to right (last_first == LAST_TO_FIRST) */ + + /* ( DEST is empty or last item of SOURCE and first item of DEST + are the items of different object or of different types ) + */ + src_nr_item = B_NR_ITEMS (src); + ih = B_N_PITEM_HEAD (src, src_nr_item - 1); + dih = B_N_PITEM_HEAD (dest, 0); + +#ifdef REISERFS_FSCK + if (!dest_nr_item || are_items_mergeable (ih, dih, src->b_size) == 0) +#else + if (!dest_nr_item || !op_is_left_mergeable (&(dih->ih_key), src->b_size)) +#endif + return 0; + + if ( is_direntry_le_ih (ih)) { + if ( bytes_or_entries == -1 ) + /* bytes_or_entries = entries number in last item body of SOURCE */ + bytes_or_entries = le16_to_cpu (ih->u.ih_entry_count); + + leaf_copy_dir_entries (dest_bi, src, LAST_TO_FIRST, src_nr_item - 1, le16_to_cpu (ih->u.ih_entry_count) - bytes_or_entries, bytes_or_entries); + return 1; + } + + /* copy part of the body of the last item of SOURCE to the begin of the body of the first item of the DEST; + part defined by 'bytes_or_entries'; if byte_or_entriess == -1 copy whole body; change first item key of the DEST; + don't create new item header + */ + +#ifdef CONFIG_REISERFS_CHECK + if (is_indirect_le_ih(ih) && get_ih_free_space (ih)) + reiserfs_panic (0, "vs-10040: leaf_copy_boundary_item: " + "merge to right: last unformatted node of non-last indirect item must be filled entirely (%h)", + ih); +#endif + + if ( bytes_or_entries == -1 ) { + /* bytes_or_entries = length of last item body of SOURCE */ + bytes_or_entries = ih->ih_item_len; + +#ifdef CONFIG_REISERFS_CHECK + if (le_ih_k_offset (dih) != le_ih_k_offset (ih) + op_bytes_number (ih, src->b_size)) + reiserfs_panic (0, "vs-10050: leaf_copy_boundary_item: items %h and %h do not match", ih, dih); +#endif + + /* change first item key of the DEST */ + set_le_ih_k_offset (dih, le_ih_k_offset (ih)); + + /* item becomes non-mergeable */ + /* or mergeable if left item was */ + set_le_ih_k_type (dih, le_ih_k_type (ih)); + } else { + /* merge to right only part of item */ +#ifdef CONFIG_REISERFS_CHECK + if ( le16_to_cpu (ih->ih_item_len) <= bytes_or_entries ) + reiserfs_panic (0, "vs-10060: leaf_copy_boundary_item: no so much bytes %lu (needed %lu)", + ih->ih_item_len, bytes_or_entries); +#endif + + /* change first item key of the DEST */ + if ( is_direct_le_ih (dih) ) { +#ifdef CONFIG_REISERFS_CHECK + if (le_ih_k_offset (dih) <= (unsigned long)bytes_or_entries) + reiserfs_panic (0, "vs-10070: leaf_copy_boundary_item: dih %h, bytes_or_entries(%d)", + dih, bytes_or_entries); +#endif + set_le_ih_k_offset (dih, le_ih_k_offset (dih) - bytes_or_entries); + } else { +#ifdef CONFIG_REISERFS_CHECK + if (le_ih_k_offset (dih) <= (bytes_or_entries / UNFM_P_SIZE) * dest->b_size ) + reiserfs_panic (0, "vs-10080: leaf_copy_boundary_item: dih %h, bytes_or_entries(%d)", + dih, (bytes_or_entries/UNFM_P_SIZE)*dest->b_size); +#endif + set_le_ih_k_offset (dih, le_ih_k_offset (dih) - ((bytes_or_entries / UNFM_P_SIZE) * dest->b_size)); + } + } + + leaf_paste_in_buffer (dest_bi, 0, 0, bytes_or_entries, B_I_PITEM(src,ih) + ih->ih_item_len - bytes_or_entries, 0); + return 1; +} + + +/* copy cpy_mun items from buffer src to buffer dest + * last_first == FIRST_TO_LAST means, that we copy cpy_num items beginning from first-th item in src to tail of dest + * last_first == LAST_TO_FIRST means, that we copy cpy_num items beginning from first-th item in src to head of dest + */ +static void leaf_copy_items_entirely (struct buffer_info * dest_bi, struct buffer_head * src, int last_first, + int first, int cpy_num) +{ + struct buffer_head * dest; + int nr; + int dest_before; + int last_loc, last_inserted_loc, location; + int i, j; + struct block_head * blkh; + struct item_head * ih; + +#ifdef CONFIG_REISERFS_CHECK + if (last_first != LAST_TO_FIRST && last_first != FIRST_TO_LAST) + reiserfs_panic (0, "vs-10090: leaf_copy_items_entirely: bad last_first parameter %d", last_first); + + if (B_NR_ITEMS (src) - first < cpy_num) + reiserfs_panic (0, "vs-10100: leaf_copy_items_entirely: too few items in source %d, required %d from %d", + B_NR_ITEMS(src), cpy_num, first); + + if (cpy_num < 0) + reiserfs_panic (0, "vs-10110: leaf_copy_items_entirely: can not copy negative amount of items"); + + if ( ! dest_bi ) + reiserfs_panic (0, "vs-10120: leaf_copy_items_entirely: can not copy negative amount of items"); +#endif + + dest = dest_bi->bi_bh; + +#ifdef CONFIG_REISERFS_CHECK + if ( ! dest ) + reiserfs_panic (0, "vs-10130: leaf_copy_items_entirely: can not copy negative amount of items"); +#endif + + if (cpy_num == 0) + return; + + nr = le16_to_cpu ((blkh = B_BLK_HEAD(dest))->blk_nr_item); + + /* we will insert items before 0-th or nr-th item in dest buffer. It depends of last_first parameter */ + dest_before = (last_first == LAST_TO_FIRST) ? 0 : nr; + + /* location of head of first new item */ + ih = B_N_PITEM_HEAD (dest, dest_before); + +#ifdef CONFIG_REISERFS_CHECK + if (le16_to_cpu (blkh->blk_free_space) < cpy_num * IH_SIZE) { + reiserfs_panic (0, "vs-10140: leaf_copy_items_entirely: " + "not enough free space for headers %d (needed %d)", + B_FREE_SPACE (dest), cpy_num * IH_SIZE); + } +#endif + + /* prepare space for headers */ + memmove (ih + cpy_num, ih, (nr-dest_before) * IH_SIZE); + + /* copy item headers */ + memcpy (ih, B_N_PITEM_HEAD (src, first), cpy_num * IH_SIZE); + + blkh->blk_free_space = cpu_to_le16 (le16_to_cpu (blkh->blk_free_space) - IH_SIZE * cpy_num); + + /* location of unmovable item */ + j = location = (dest_before == 0) ? dest->b_size : (ih-1)->ih_item_location; + for (i = dest_before; i < nr + cpy_num; i ++) + ih[i-dest_before].ih_item_location = + (location -= ih[i-dest_before].ih_item_len); + + /* prepare space for items */ + last_loc = ih[nr+cpy_num-1-dest_before].ih_item_location; + last_inserted_loc = ih[cpy_num-1].ih_item_location; + + /* check free space */ +#ifdef CONFIG_REISERFS_CHECK + if (le16_to_cpu (blkh->blk_free_space) < j - last_inserted_loc) { + reiserfs_panic (0, "vs-10150: leaf_copy_items_entirely: not enough free space for items %d (needed %d)", + le16_to_cpu (blkh->blk_free_space), j - last_inserted_loc); + } +#endif + + memmove (dest->b_data + last_loc, + dest->b_data + last_loc + j - last_inserted_loc, + last_inserted_loc - last_loc); + + /* copy items */ + memcpy (dest->b_data + last_inserted_loc, B_N_PITEM(src,(first + cpy_num - 1)), + j - last_inserted_loc); + + /* sizes, item number */ + blkh->blk_nr_item = cpu_to_le16 (le16_to_cpu (blkh->blk_nr_item) + cpy_num); + blkh->blk_free_space = cpu_to_le16 (le16_to_cpu (blkh->blk_free_space) - (j - last_inserted_loc)); + + do_balance_mark_leaf_dirty (dest_bi->tb, dest, 0); + + if (dest_bi->bi_parent) { +#ifdef CONFIG_REISERFS_CHECK + if (B_N_CHILD (dest_bi->bi_parent, dest_bi->bi_position)->dc_block_number != dest->b_blocknr) { + reiserfs_panic (0, "vs-10160: leaf_copy_items_entirely: " + "block number in bh does not match to field in disk_child structure %lu and %lu", + dest->b_blocknr, B_N_CHILD (dest_bi->bi_parent, dest_bi->bi_position)->dc_block_number); + } +#endif + B_N_CHILD (dest_bi->bi_parent, dest_bi->bi_position)->dc_size += + j - last_inserted_loc + IH_SIZE * cpy_num; + + do_balance_mark_internal_dirty (dest_bi->tb, dest_bi->bi_parent, 0); + } +} + + +/* This function splits the (liquid) item into two items (useful when + shifting part of an item into another node.) */ +static void leaf_item_bottle (struct buffer_info * dest_bi, struct buffer_head * src, int last_first, + int item_num, int cpy_bytes) +{ + struct buffer_head * dest = dest_bi->bi_bh; + struct item_head * ih; + +#ifdef CONFIG_REISERFS_CHECK + if ( cpy_bytes == -1 ) + reiserfs_panic (0, "vs-10170: leaf_item_bottle: bytes == - 1 means: do not split item"); +#endif + + if ( last_first == FIRST_TO_LAST ) { + /* if ( if item in position item_num in buffer SOURCE is directory item ) */ + if (is_direntry_le_ih (ih = B_N_PITEM_HEAD(src,item_num))) + leaf_copy_dir_entries (dest_bi, src, FIRST_TO_LAST, item_num, 0, cpy_bytes); + else { + struct item_head n_ih; + + /* copy part of the body of the item number 'item_num' of SOURCE to the end of the DEST + part defined by 'cpy_bytes'; create new item header; change old item_header (????); + n_ih = new item_header; + */ + memcpy (&n_ih, ih, IH_SIZE); + n_ih.ih_item_len = cpu_to_le16 (cpy_bytes); + if (is_indirect_le_ih (ih)) { +#ifdef CONFIG_REISERFS_CHECK + if (cpy_bytes == le16_to_cpu (ih->ih_item_len) && get_ih_free_space (ih)) + reiserfs_panic (0, "vs-10180: leaf_item_bottle: " + "when whole indirect item is bottle to left neighbor, it must have free_space==0 (not %lu)", + get_ih_free_space (ih)); +#endif + set_ih_free_space (&n_ih, 0); + } + +#ifdef CONFIG_REISERFS_CHECK + if (op_is_left_mergeable (&(ih->ih_key), src->b_size)) + reiserfs_panic (0, "vs-10190: leaf_item_bottle: bad mergeability of item %h", ih); +#endif + n_ih.ih_version = ih->ih_version;; + leaf_insert_into_buf (dest_bi, B_NR_ITEMS(dest), &n_ih, B_N_PITEM (src, item_num), 0); + } + } else { + /* if ( if item in position item_num in buffer SOURCE is directory item ) */ + if (is_direntry_le_ih(ih = B_N_PITEM_HEAD (src, item_num))) + leaf_copy_dir_entries (dest_bi, src, LAST_TO_FIRST, item_num, I_ENTRY_COUNT(ih) - cpy_bytes, cpy_bytes); + else { + struct item_head n_ih; + + /* copy part of the body of the item number 'item_num' of SOURCE to the begin of the DEST + part defined by 'cpy_bytes'; create new item header; + n_ih = new item_header; + */ + memcpy (&n_ih, ih, SHORT_KEY_SIZE); + n_ih.ih_version = cpu_to_le16 (ih_version (ih)); + if (is_direct_le_ih (ih)) { + set_le_ih_k_offset (&n_ih, le_ih_k_offset (ih) + le16_to_cpu (ih->ih_item_len) - cpy_bytes); + set_le_ih_k_type (&n_ih, TYPE_DIRECT); + set_ih_free_space (&n_ih, MAX_US_INT); + } else { + /* indirect item */ +#ifdef CONFIG_REISERFS_CHECK + if (!cpy_bytes && get_ih_free_space (ih)) + reiserfs_panic (0, "vs-10200: leaf_item_bottle: ih->ih_free_space must be 0 when indirect item will be appended"); +#endif + set_le_ih_k_offset (&n_ih, le_ih_k_offset (ih) + (le16_to_cpu (ih->ih_item_len) - cpy_bytes) / UNFM_P_SIZE * dest->b_size); + set_le_ih_k_type (&n_ih, TYPE_INDIRECT); + set_ih_free_space (&n_ih, get_ih_free_space (ih)); + } + + /* set item length */ + n_ih.ih_item_len = cpu_to_le16 (cpy_bytes); + n_ih.ih_version = cpu_to_le16 (le16_to_cpu (ih->ih_version)); + leaf_insert_into_buf (dest_bi, 0, &n_ih, B_N_PITEM(src,item_num) + le16_to_cpu (ih->ih_item_len) - cpy_bytes, 0); + } + } +} + + +/* If cpy_bytes equals minus one than copy cpy_num whole items from SOURCE to DEST. + If cpy_bytes not equal to minus one than copy cpy_num-1 whole items from SOURCE to DEST. + From last item copy cpy_num bytes for regular item and cpy_num directory entries for + directory item. */ +static int leaf_copy_items (struct buffer_info * dest_bi, struct buffer_head * src, int last_first, int cpy_num, + int cpy_bytes) +{ + struct buffer_head * dest; + int pos, i, src_nr_item, bytes; + + dest = dest_bi->bi_bh; +#ifdef CONFIG_REISERFS_CHECK + if (!dest || !src) + reiserfs_panic (0, "vs-10210: leaf_copy_items: !dest || !src"); + + if ( last_first != FIRST_TO_LAST && last_first != LAST_TO_FIRST ) + reiserfs_panic (0, "vs-10220: leaf_copy_items: last_first != FIRST_TO_LAST && last_first != LAST_TO_FIRST"); + + if ( B_NR_ITEMS(src) < cpy_num ) + reiserfs_panic (0, "vs-10230: leaf_copy_items: No enough items: %d, required %d", B_NR_ITEMS(src), cpy_num); + + if ( cpy_num < 0 ) + reiserfs_panic (0, "vs-10240: leaf_copy_items: cpy_num < 0 (%d)", cpy_num); +#endif + + if ( cpy_num == 0 ) + return 0; + + if ( last_first == FIRST_TO_LAST ) { + /* copy items to left */ + pos = 0; + if ( cpy_num == 1 ) + bytes = cpy_bytes; + else + bytes = -1; + + /* copy the first item or it part or nothing to the end of the DEST (i = leaf_copy_boundary_item(DEST,SOURCE,0,bytes)) */ + i = leaf_copy_boundary_item (dest_bi, src, FIRST_TO_LAST, bytes); + cpy_num -= i; + if ( cpy_num == 0 ) + return i; + pos += i; + if ( cpy_bytes == -1 ) + /* copy first cpy_num items starting from position 'pos' of SOURCE to end of DEST */ + leaf_copy_items_entirely (dest_bi, src, FIRST_TO_LAST, pos, cpy_num); + else { + /* copy first cpy_num-1 items starting from position 'pos-1' of the SOURCE to the end of the DEST */ + leaf_copy_items_entirely (dest_bi, src, FIRST_TO_LAST, pos, cpy_num-1); + + /* copy part of the item which number is cpy_num+pos-1 to the end of the DEST */ + leaf_item_bottle (dest_bi, src, FIRST_TO_LAST, cpy_num+pos-1, cpy_bytes); + } + } else { + /* copy items to right */ + src_nr_item = B_NR_ITEMS (src); + if ( cpy_num == 1 ) + bytes = cpy_bytes; + else + bytes = -1; + + /* copy the last item or it part or nothing to the begin of the DEST (i = leaf_copy_boundary_item(DEST,SOURCE,1,bytes)); */ + i = leaf_copy_boundary_item (dest_bi, src, LAST_TO_FIRST, bytes); + + cpy_num -= i; + if ( cpy_num == 0 ) + return i; + + pos = src_nr_item - cpy_num - i; + if ( cpy_bytes == -1 ) { + /* starting from position 'pos' copy last cpy_num items of SOURCE to begin of DEST */ + leaf_copy_items_entirely (dest_bi, src, LAST_TO_FIRST, pos, cpy_num); + } else { + /* copy last cpy_num-1 items starting from position 'pos+1' of the SOURCE to the begin of the DEST; */ + leaf_copy_items_entirely (dest_bi, src, LAST_TO_FIRST, pos+1, cpy_num-1); + + /* copy part of the item which number is pos to the begin of the DEST */ + leaf_item_bottle (dest_bi, src, LAST_TO_FIRST, pos, cpy_bytes); + } + } + return i; +} + + +/* there are types of coping: from S[0] to L[0], from S[0] to R[0], + from R[0] to L[0]. for each of these we have to define parent and + positions of destination and source buffers */ +static void leaf_define_dest_src_infos (int shift_mode, struct tree_balance * tb, struct buffer_info * dest_bi, + struct buffer_info * src_bi, int * first_last, + struct buffer_head * Snew) +{ +#ifdef CONFIG_REISERFS_CHECK + memset (dest_bi, 0, sizeof (struct buffer_info)); + memset (src_bi, 0, sizeof (struct buffer_info)); +#endif + + /* define dest, src, dest parent, dest position */ + switch (shift_mode) { + case LEAF_FROM_S_TO_L: /* it is used in leaf_shift_left */ + src_bi->tb = tb; + src_bi->bi_bh = PATH_PLAST_BUFFER (tb->tb_path); + src_bi->bi_parent = PATH_H_PPARENT (tb->tb_path, 0); + src_bi->bi_position = PATH_H_B_ITEM_ORDER (tb->tb_path, 0); /* src->b_item_order */ + dest_bi->tb = tb; + dest_bi->bi_bh = tb->L[0]; + dest_bi->bi_parent = tb->FL[0]; + dest_bi->bi_position = get_left_neighbor_position (tb, 0); + *first_last = FIRST_TO_LAST; + break; + + case LEAF_FROM_S_TO_R: /* it is used in leaf_shift_right */ + src_bi->tb = tb; + src_bi->bi_bh = PATH_PLAST_BUFFER (tb->tb_path); + src_bi->bi_parent = PATH_H_PPARENT (tb->tb_path, 0); + src_bi->bi_position = PATH_H_B_ITEM_ORDER (tb->tb_path, 0); + dest_bi->tb = tb; + dest_bi->bi_bh = tb->R[0]; + dest_bi->bi_parent = tb->FR[0]; + dest_bi->bi_position = get_right_neighbor_position (tb, 0); + *first_last = LAST_TO_FIRST; + break; + + case LEAF_FROM_R_TO_L: /* it is used in balance_leaf_when_delete */ + src_bi->tb = tb; + src_bi->bi_bh = tb->R[0]; + src_bi->bi_parent = tb->FR[0]; + src_bi->bi_position = get_right_neighbor_position (tb, 0); + dest_bi->tb = tb; + dest_bi->bi_bh = tb->L[0]; + dest_bi->bi_parent = tb->FL[0]; + dest_bi->bi_position = get_left_neighbor_position (tb, 0); + *first_last = FIRST_TO_LAST; + break; + + case LEAF_FROM_L_TO_R: /* it is used in balance_leaf_when_delete */ + src_bi->tb = tb; + src_bi->bi_bh = tb->L[0]; + src_bi->bi_parent = tb->FL[0]; + src_bi->bi_position = get_left_neighbor_position (tb, 0); + dest_bi->tb = tb; + dest_bi->bi_bh = tb->R[0]; + dest_bi->bi_parent = tb->FR[0]; + dest_bi->bi_position = get_right_neighbor_position (tb, 0); + *first_last = LAST_TO_FIRST; + break; + + case LEAF_FROM_S_TO_SNEW: + src_bi->tb = tb; + src_bi->bi_bh = PATH_PLAST_BUFFER (tb->tb_path); + src_bi->bi_parent = PATH_H_PPARENT (tb->tb_path, 0); + src_bi->bi_position = PATH_H_B_ITEM_ORDER (tb->tb_path, 0); + dest_bi->tb = tb; + dest_bi->bi_bh = Snew; + dest_bi->bi_parent = 0; + dest_bi->bi_position = 0; + *first_last = LAST_TO_FIRST; + break; + + default: + reiserfs_panic (0, "vs-10250: leaf_define_dest_src_infos: shift type is unknown (%d)", shift_mode); + } +#ifdef CONFIG_REISERFS_CHECK + if (src_bi->bi_bh == 0 || dest_bi->bi_bh == 0) { + reiserfs_panic (0, "vs-10260: leaf_define_dest_src_etc: mode==%d, source (%p) or dest (%p) buffer is initialized incorrectly", + shift_mode, src_bi->bi_bh, dest_bi->bi_bh); + } +#endif +} + + + + +/* copy mov_num items and mov_bytes of the (mov_num-1)th item to + neighbor. Delete them from source */ +int leaf_move_items (int shift_mode, struct tree_balance * tb, int mov_num, int mov_bytes, struct buffer_head * Snew) +{ + int ret_value; + struct buffer_info dest_bi, src_bi; + int first_last; + + leaf_define_dest_src_infos (shift_mode, tb, &dest_bi, &src_bi, &first_last, Snew); + + ret_value = leaf_copy_items (&dest_bi, src_bi.bi_bh, first_last, mov_num, mov_bytes); + + leaf_delete_items (&src_bi, first_last, (first_last == FIRST_TO_LAST) ? 0 : (B_NR_ITEMS(src_bi.bi_bh) - mov_num), mov_num, mov_bytes); + + + return ret_value; +} + + +/* Shift shift_num items (and shift_bytes of last shifted item if shift_bytes != -1) + from S[0] to L[0] and replace the delimiting key */ +int leaf_shift_left (struct tree_balance * tb, int shift_num, int shift_bytes) +{ + struct buffer_head * S0 = PATH_PLAST_BUFFER (tb->tb_path); + int i; + + /* move shift_num (and shift_bytes bytes) items from S[0] to left neighbor L[0] */ + i = leaf_move_items (LEAF_FROM_S_TO_L, tb, shift_num, shift_bytes, 0); + + if ( shift_num ) { + if (B_NR_ITEMS (S0) == 0) { /* number of items in S[0] == 0 */ + +#ifdef CONFIG_REISERFS_CHECK + if ( shift_bytes != -1 ) + reiserfs_panic (tb->tb_sb, "vs-10270: leaf_shift_left: S0 is empty now, but shift_bytes != -1 (%d)", shift_bytes); + + if (tb->tb_mode == M_PASTE || tb->tb_mode == M_INSERT) { + print_cur_tb ("vs-10275"); + reiserfs_panic (tb->tb_sb, "vs-10275: leaf_shift_left: balance condition corrupted (%c)", tb->tb_mode); + } +#endif + + if (PATH_H_POSITION (tb->tb_path, 1) == 0) + replace_key (tb, tb->CFL[0], tb->lkey[0], PATH_H_PPARENT (tb->tb_path, 0), 0); + +#if 0 + /* change right_delimiting_key field in L0's block header */ + copy_key (B_PRIGHT_DELIM_KEY(tb->L[0]), B_PRIGHT_DELIM_KEY (S0)); +#endif + } else { + /* replace lkey in CFL[0] by 0-th key from S[0]; */ + replace_key (tb, tb->CFL[0], tb->lkey[0], S0, 0); + +#if 0 + /* change right_delimiting_key field in L0's block header */ + copy_key (B_PRIGHT_DELIM_KEY(tb->L[0]), B_N_PKEY (S0, 0)); +#endif +#ifdef CONFIG_REISERFS_CHECK + if (shift_bytes != -1 && !(is_direntry_le_ih (B_N_PITEM_HEAD (S0, 0)) + && !I_ENTRY_COUNT (B_N_PITEM_HEAD (S0, 0)))) { + if (!op_is_left_mergeable (B_N_PKEY (S0, 0), S0->b_size)) { + reiserfs_panic (tb->tb_sb, "vs-10280: leaf_shift_left: item must be mergeable"); + } + } +#endif + } + } + + return i; +} + + + + + +/* CLEANING STOPPED HERE */ + + + + +/* Shift shift_num (shift_bytes) items from S[0] to the right neighbor, and replace the delimiting key */ +int leaf_shift_right( + struct tree_balance * tb, + int shift_num, + int shift_bytes + ) +{ + // struct buffer_head * S0 = PATH_PLAST_BUFFER (tb->tb_path); + int ret_value; + + /* move shift_num (and shift_bytes) items from S[0] to right neighbor R[0] */ + ret_value = leaf_move_items (LEAF_FROM_S_TO_R, tb, shift_num, shift_bytes, 0); + + /* replace rkey in CFR[0] by the 0-th key from R[0] */ + if (shift_num) { + replace_key (tb, tb->CFR[0], tb->rkey[0], tb->R[0], 0); + +#if 0 + /* change right_delimiting_key field in S0's block header */ + copy_key (B_PRIGHT_DELIM_KEY(S0), B_N_PKEY (tb->R[0], 0)); +#endif + } + + return ret_value; +} + + + +static void leaf_delete_items_entirely (struct buffer_info * bi, + int first, int del_num); +/* If del_bytes == -1, starting from position 'first' delete del_num items in whole in buffer CUR. + If not. + If last_first == 0. Starting from position 'first' delete del_num-1 items in whole. Delete part of body of + the first item. Part defined by del_bytes. Don't delete first item header + If last_first == 1. Starting from position 'first+1' delete del_num-1 items in whole. Delete part of body of + the last item . Part defined by del_bytes. Don't delete last item header. +*/ +void leaf_delete_items (struct buffer_info * cur_bi, int last_first, + int first, int del_num, int del_bytes) +{ + struct buffer_head * bh; + int item_amount = B_NR_ITEMS (bh = cur_bi->bi_bh); + +#ifdef CONFIG_REISERFS_CHECK + if ( !bh ) + reiserfs_panic (0, "leaf_delete_items: 10155: bh is not defined"); + + if ( del_num < 0 ) + reiserfs_panic (0, "leaf_delete_items: 10160: del_num can not be < 0. del_num==%d", del_num); + + if ( first < 0 || first + del_num > item_amount ) + reiserfs_panic (0, "leaf_delete_items: 10165: invalid number of first item to be deleted (%d) or " + "no so much items (%d) to delete (only %d)", first, first + del_num, item_amount); +#endif + + if ( del_num == 0 ) + return; + + if ( first == 0 && del_num == item_amount && del_bytes == -1 ) { + make_empty_node (cur_bi); + do_balance_mark_leaf_dirty (cur_bi->tb, bh, 0); + return; + } + + if ( del_bytes == -1 ) + /* delete del_num items beginning from item in position first */ + leaf_delete_items_entirely (cur_bi, first, del_num); + else { + if ( last_first == FIRST_TO_LAST ) { + /* delete del_num-1 items beginning from item in position first */ + leaf_delete_items_entirely (cur_bi, first, del_num-1); + + /* delete the part of the first item of the bh + do not delete item header + */ + leaf_cut_from_buffer (cur_bi, 0, 0, del_bytes); + } else { + struct item_head * ih; + int len; + + /* delete del_num-1 items beginning from item in position first+1 */ + leaf_delete_items_entirely (cur_bi, first+1, del_num-1); + + if (is_direntry_le_ih (ih = B_N_PITEM_HEAD(bh, B_NR_ITEMS(bh)-1))) /* the last item is directory */ + /* len = numbers of directory entries in this item */ + len = le16_to_cpu (ih->u.ih_entry_count); + else + /* len = body len of item */ + len = le16_to_cpu (ih->ih_item_len); + + /* delete the part of the last item of the bh + do not delete item header + */ + leaf_cut_from_buffer (cur_bi, B_NR_ITEMS(bh)-1, len - del_bytes, del_bytes); + } + } +} + + +/* insert item into the leaf node in position before */ +void leaf_insert_into_buf (struct buffer_info * bi, int before, + struct item_head * inserted_item_ih, + const char * inserted_item_body, + int zeros_number) +{ + struct buffer_head * bh = bi->bi_bh; + int nr; + struct block_head * blkh; + struct item_head * ih; + int i; + int last_loc, unmoved_loc; + char * to; + + + nr = le16_to_cpu ((blkh = B_BLK_HEAD (bh))->blk_nr_item); + +#ifdef CONFIG_REISERFS_CHECK + /* check free space */ + if (le16_to_cpu (blkh->blk_free_space) < le16_to_cpu (inserted_item_ih->ih_item_len) + IH_SIZE) + reiserfs_panic (0, "leaf_insert_into_buf: 10170: " + "not enough free space in block %z, new item %h", + bh, inserted_item_ih); + if (zeros_number > inserted_item_ih->ih_item_len) + reiserfs_panic (0, "vs-10172: leaf_insert_into_buf: " + "zero number == %d, item length == %d", zeros_number, inserted_item_ih->ih_item_len); +#endif /* CONFIG_REISERFS_CHECK */ + + + /* get item new item must be inserted before */ + ih = B_N_PITEM_HEAD (bh, before); + + /* prepare space for the body of new item */ + last_loc = nr ? ih[nr - before - 1].ih_item_location : bh->b_size; + unmoved_loc = before ? (ih-1)->ih_item_location : bh->b_size; + + memmove (bh->b_data + last_loc - inserted_item_ih->ih_item_len, + bh->b_data + last_loc, unmoved_loc - last_loc); + + to = bh->b_data + unmoved_loc - inserted_item_ih->ih_item_len; + memset (to, 0, zeros_number); + to += zeros_number; + + /* copy body to prepared space */ + if (inserted_item_body) + memmove (to, inserted_item_body, inserted_item_ih->ih_item_len - zeros_number); + else + memset(to, '\0', inserted_item_ih->ih_item_len - zeros_number); + + /* insert item header */ + memmove (ih + 1, ih, IH_SIZE * (nr - before)); + memmove (ih, inserted_item_ih, IH_SIZE); + + /* change locations */ + for (i = before; i < nr + 1; i ++) + ih[i-before].ih_item_location = + (unmoved_loc -= ih[i-before].ih_item_len); + + /* sizes, free space, item number */ + blkh->blk_nr_item = cpu_to_le16 (le16_to_cpu (blkh->blk_nr_item) + 1); + blkh->blk_free_space = cpu_to_le16 (le16_to_cpu (blkh->blk_free_space) - + (IH_SIZE + inserted_item_ih->ih_item_len)); + + do_balance_mark_leaf_dirty (bi->tb, bh, 1); + + if (bi->bi_parent) { + B_N_CHILD (bi->bi_parent, bi->bi_position)->dc_size += (IH_SIZE + inserted_item_ih->ih_item_len); + do_balance_mark_internal_dirty (bi->tb, bi->bi_parent, 0); + } +} + + +/* paste paste_size bytes to affected_item_num-th item. + When item is a directory, this only prepare space for new entries */ +void leaf_paste_in_buffer (struct buffer_info * bi, int affected_item_num, + int pos_in_item, int paste_size, + const char * body, + int zeros_number) +{ + struct buffer_head * bh = bi->bi_bh; + int nr; + struct block_head * blkh; + struct item_head * ih; + int i; + int last_loc, unmoved_loc; + + + nr = le16_to_cpu ((blkh = B_BLK_HEAD(bh))->blk_nr_item); + +#ifdef CONFIG_REISERFS_CHECK + /* check free space */ + if (le16_to_cpu (blkh->blk_free_space) < paste_size) + reiserfs_panic (0, "leaf_paste_in_buffer: 10175: not enough free space: needed %d, available %d", + paste_size, le16_to_cpu (blkh->blk_free_space)); + if (zeros_number > paste_size) { + print_cur_tb ("10177"); + reiserfs_panic (0, "vs-10177: leaf_paste_in_buffer: zero number == %d, paste_size == %d", + zeros_number, paste_size); + } +#endif /* CONFIG_REISERFS_CHECK */ + + + /* item to be appended */ + ih = B_N_PITEM_HEAD(bh, affected_item_num); + + last_loc = ih[nr - affected_item_num - 1].ih_item_location; + unmoved_loc = affected_item_num ? (ih-1)->ih_item_location : bh->b_size; + + /* prepare space */ + memmove (bh->b_data + last_loc - paste_size, bh->b_data + last_loc, + unmoved_loc - last_loc); + + + /* change locations */ + for (i = affected_item_num; i < nr; i ++) + ih[i-affected_item_num].ih_item_location -= paste_size; + + if ( body ) { + if (!is_direntry_le_ih (ih)) { + if (!pos_in_item) { + /* shift data to right */ + memmove (bh->b_data + ih->ih_item_location + paste_size, + bh->b_data + ih->ih_item_location, ih->ih_item_len); + /* paste data in the head of item */ + memset (bh->b_data + ih->ih_item_location, 0, zeros_number); + memcpy (bh->b_data + ih->ih_item_location + zeros_number, body, paste_size - zeros_number); + } else { + memset (bh->b_data + unmoved_loc - paste_size, 0, zeros_number); + memcpy (bh->b_data + unmoved_loc - paste_size + zeros_number, body, paste_size - zeros_number); + } + } + } + else + memset(bh->b_data + unmoved_loc - paste_size,'\0',paste_size); + + ih->ih_item_len += paste_size; + + /* change free space */ + blkh->blk_free_space = cpu_to_le16 (le16_to_cpu (blkh->blk_free_space) - paste_size); + + do_balance_mark_leaf_dirty (bi->tb, bh, 0); + + if (bi->bi_parent) { + B_N_CHILD (bi->bi_parent, bi->bi_position)->dc_size += paste_size; + do_balance_mark_internal_dirty (bi->tb, bi->bi_parent, 0); + } +} + + +/* cuts DEL_COUNT entries beginning from FROM-th entry. Directory item + does not have free space, so it moves DEHs and remaining records as + necessary. Return value is size of removed part of directory item + in bytes. */ +static int leaf_cut_entries ( + struct buffer_head * bh, + struct item_head * ih, + int from, + int del_count + ) +{ + char * item; + struct reiserfs_de_head * deh; + int prev_record_offset; /* offset of record, that is (from-1)th */ + char * prev_record; /* */ + int cut_records_len; /* length of all removed records */ + int i; + + +#ifdef CONFIG_REISERFS_CHECK + /* make sure, that item is directory and there are enough entries to + remove */ + if (!is_direntry_le_ih (ih)) + reiserfs_panic (0, "leaf_cut_entries: 10180: item is not directory item"); + + if (I_ENTRY_COUNT(ih) < from + del_count) + reiserfs_panic (0, "leaf_cut_entries: 10185: item contains not enough entries: entry_cout = %d, from = %d, to delete = %d", + I_ENTRY_COUNT(ih), from, del_count); +#endif + + if (del_count == 0) + return 0; + + /* first byte of item */ + item = bh->b_data + ih->ih_item_location; + + /* entry head array */ + deh = B_I_DEH (bh, ih); + + /* first byte of remaining entries, those are BEFORE cut entries + (prev_record) and length of all removed records (cut_records_len) */ + prev_record_offset = (from ? deh[from - 1].deh_location : ih->ih_item_len); + cut_records_len = prev_record_offset/*from_record*/ - deh[from + del_count - 1].deh_location; + prev_record = item + prev_record_offset; + + + /* adjust locations of remaining entries */ + for (i = I_ENTRY_COUNT(ih) - 1; i > from + del_count - 1; i --) + deh[i].deh_location -= (DEH_SIZE * del_count); + + for (i = 0; i < from; i ++) + deh[i].deh_location -= DEH_SIZE * del_count + cut_records_len; + + I_ENTRY_COUNT(ih) -= del_count; + + /* shift entry head array and entries those are AFTER removed entries */ + memmove ((char *)(deh + from), + deh + from + del_count, + prev_record - cut_records_len - (char *)(deh + from + del_count)); + + /* shift records, those are BEFORE removed entries */ + memmove (prev_record - cut_records_len - DEH_SIZE * del_count, + prev_record, item + ih->ih_item_len - prev_record); + + return DEH_SIZE * del_count + cut_records_len; +} + + +/* when cut item is part of regular file + pos_in_item - first byte that must be cut + cut_size - number of bytes to be cut beginning from pos_in_item + + when cut item is part of directory + pos_in_item - number of first deleted entry + cut_size - count of deleted entries + */ +void leaf_cut_from_buffer (struct buffer_info * bi, int cut_item_num, + int pos_in_item, int cut_size) +{ + int nr; + struct buffer_head * bh = bi->bi_bh; + struct block_head * blkh; + struct item_head * ih; + int last_loc, unmoved_loc; + int i; + + nr = le16_to_cpu ((blkh = B_BLK_HEAD (bh))->blk_nr_item); + + /* item head of truncated item */ + ih = B_N_PITEM_HEAD (bh, cut_item_num); + + if (is_direntry_le_ih (ih)) { + /* first cut entry ()*/ + cut_size = leaf_cut_entries (bh, ih, pos_in_item, cut_size); + if (pos_in_item == 0) { + /* change key */ +#ifdef CONFIG_REISERFS_CHECK + if (cut_item_num) + reiserfs_panic (0, "leaf_cut_from_buffer: 10190: " + "when 0-th enrty of item is cut, that item must be first in the node, not %d-th", cut_item_num); +#endif + /* change item key by key of first entry in the item */ + set_le_ih_k_offset (ih, le32_to_cpu (B_I_DEH (bh, ih)->deh_offset)); + /*memcpy (&ih->ih_key.k_offset, &(B_I_DEH (bh, ih)->deh_offset), SHORT_KEY_SIZE);*/ + } + } else { + /* item is direct or indirect */ +#ifdef CONFIG_REISERFS_CHECK + if (is_statdata_le_ih (ih)) + reiserfs_panic (0, "leaf_cut_from_buffer: 10195: item is stat data"); + + if (pos_in_item && pos_in_item + cut_size != le16_to_cpu (ih->ih_item_len) ) + reiserfs_panic (0, "cut_from_buf: 10200: invalid offset (%lu) or trunc_size (%lu) or ih_item_len (%lu)", + pos_in_item, cut_size, le16_to_cpu (ih->ih_item_len)); +#endif + + /* shift item body to left if cut is from the head of item */ + if (pos_in_item == 0) { + memmove (bh->b_data + le16_to_cpu (ih->ih_item_location), bh->b_data + le16_to_cpu (ih->ih_item_location) + cut_size, + le16_to_cpu (ih->ih_item_len) - cut_size); + + /* change key of item */ + if (is_direct_le_ih (ih)) + set_le_ih_k_offset (ih, le_ih_k_offset (ih) + cut_size); + else { + set_le_ih_k_offset (ih, le_ih_k_offset (ih) + (cut_size / UNFM_P_SIZE) * bh->b_size); +#ifdef CONFIG_REISERFS_CHECK + if ( le16_to_cpu (ih->ih_item_len) == cut_size && get_ih_free_space (ih) ) + reiserfs_panic (0, "leaf_cut_from_buf: 10205: invalid ih_free_space (%h)", ih); +#endif + } + } + } + + + /* location of the last item */ + last_loc = le16_to_cpu (ih[nr - cut_item_num - 1].ih_item_location); + + /* location of the item, which is remaining at the same place */ + unmoved_loc = cut_item_num ? le16_to_cpu ((ih-1)->ih_item_location) : bh->b_size; + + + /* shift */ + memmove (bh->b_data + last_loc + cut_size, bh->b_data + last_loc, + unmoved_loc - last_loc - cut_size); + + /* change item length */ +/* ih->ih_item_len -= cut_size;*/ + ih->ih_item_len = cpu_to_le16 (le16_to_cpu (ih->ih_item_len) - cut_size); + + if (is_indirect_le_ih (ih)) { + if (pos_in_item) + set_ih_free_space (ih, 0); + } + + /* change locations */ + for (i = cut_item_num; i < nr; i ++) +/* ih[i-cut_item_num].ih_item_location += cut_size;*/ + ih[i-cut_item_num].ih_item_location = + cpu_to_le16 (le16_to_cpu (ih[i-cut_item_num].ih_item_location) + cut_size); + + /* size, free space */ + blkh->blk_free_space = cpu_to_le16 (le16_to_cpu (blkh->blk_free_space) + cut_size); + + do_balance_mark_leaf_dirty (bi->tb, bh, 0); + + if (bi->bi_parent) { + B_N_CHILD (bi->bi_parent, bi->bi_position)->dc_size -= cut_size; + do_balance_mark_internal_dirty (bi->tb, bi->bi_parent, 0); + } +} + + +/* delete del_num items from buffer starting from the first'th item */ +static void leaf_delete_items_entirely (struct buffer_info * bi, + int first, int del_num) +{ + struct buffer_head * bh = bi->bi_bh; + int nr; + int i, j; + int last_loc, last_removed_loc; + struct block_head * blkh; + struct item_head * ih; + +#ifdef CONFIG_REISERFS_CHECK + if (bh == NULL) + reiserfs_panic (0, "leaf_delete_items_entirely: 10210: buffer is 0"); + + if (del_num < 0) + reiserfs_panic (0, "leaf_delete_items_entirely: 10215: del_num less than 0 (%d)", del_num); +#endif /* CONFIG_REISERFS_CHECK */ + + if (del_num == 0) + return; + + nr = le16_to_cpu ((blkh = B_BLK_HEAD(bh))->blk_nr_item); + +#ifdef CONFIG_REISERFS_CHECK + if (first < 0 || first + del_num > nr) + reiserfs_panic (0, "leaf_delete_items_entirely: 10220: first=%d, number=%d, there is %d items", first, del_num, nr); +#endif /* CONFIG_REISERFS_CHECK */ + + if (first == 0 && del_num == nr) { + /* this does not work */ + make_empty_node (bi); + + do_balance_mark_leaf_dirty (bi->tb, bh, 0); + return; + } + + ih = B_N_PITEM_HEAD (bh, first); + + /* location of unmovable item */ + j = (first == 0) ? bh->b_size : (ih-1)->ih_item_location; + + /* delete items */ + last_loc = ih[nr-1-first].ih_item_location; + last_removed_loc = ih[del_num-1].ih_item_location; + + memmove (bh->b_data + last_loc + j - last_removed_loc, + bh->b_data + last_loc, last_removed_loc - last_loc); + + /* delete item headers */ + memmove (ih, ih + del_num, (nr - first - del_num) * IH_SIZE); + + /* change item location */ + for (i = first; i < nr - del_num; i ++) + ih[i-first].ih_item_location += j - last_removed_loc; + + /* sizes, item number */ + blkh->blk_nr_item = cpu_to_le16 (le16_to_cpu (blkh->blk_nr_item) - del_num); + blkh->blk_free_space = cpu_to_le16 (le16_to_cpu (blkh->blk_free_space) + (j - last_removed_loc + IH_SIZE * del_num)); + + do_balance_mark_leaf_dirty (bi->tb, bh, 0); + + if (bi->bi_parent) { + B_N_CHILD (bi->bi_parent, bi->bi_position)->dc_size -= j - last_removed_loc + IH_SIZE * del_num; + do_balance_mark_internal_dirty (bi->tb, bi->bi_parent, 0); + } +} + + + + + +/* paste new_entry_count entries (new_dehs, records) into position before to item_num-th item */ +void leaf_paste_entries ( + struct buffer_head * bh, + int item_num, + int before, + int new_entry_count, + struct reiserfs_de_head * new_dehs, + const char * records, + int paste_size + ) +{ + struct item_head * ih; + char * item; + struct reiserfs_de_head * deh; + char * insert_point; + int i, old_entry_num; + + if (new_entry_count == 0) + return; + + ih = B_N_PITEM_HEAD(bh, item_num); + +#ifdef CONFIG_REISERFS_CHECK + /* make sure, that item is directory, and there are enough records in it */ + if (!is_direntry_le_ih (ih)) + reiserfs_panic (0, "leaf_paste_entries: 10225: item is not directory item"); + + if (I_ENTRY_COUNT (ih) < before) + reiserfs_panic (0, "leaf_paste_entries: 10230: there are no entry we paste entries before. entry_count = %d, before = %d", + I_ENTRY_COUNT (ih), before); +#endif + + + /* first byte of dest item */ + item = bh->b_data + ih->ih_item_location; + + /* entry head array */ + deh = B_I_DEH (bh, ih); + + /* new records will be pasted at this point */ + insert_point = item + (before ? deh[before - 1].deh_location : (ih->ih_item_len - paste_size)); + + /* adjust locations of records that will be AFTER new records */ + for (i = I_ENTRY_COUNT(ih) - 1; i >= before; i --) + deh[i].deh_location += DEH_SIZE * new_entry_count; + + /* adjust locations of records that will be BEFORE new records */ + for (i = 0; i < before; i ++) + deh[i].deh_location += paste_size; + + old_entry_num = I_ENTRY_COUNT(ih); + I_ENTRY_COUNT(ih) += new_entry_count; + + /* prepare space for pasted records */ + memmove (insert_point + paste_size, insert_point, item + (ih->ih_item_len - paste_size) - insert_point); + + /* copy new records */ + memcpy (insert_point + DEH_SIZE * new_entry_count, records, + paste_size - DEH_SIZE * new_entry_count); + + /* prepare space for new entry heads */ + deh += before; + memmove ((char *)(deh + new_entry_count), deh, insert_point - (char *)deh); + + /* copy new entry heads */ + deh = (struct reiserfs_de_head *)((char *)deh); + memcpy (deh, new_dehs, DEH_SIZE * new_entry_count); + + /* set locations of new records */ + for (i = 0; i < new_entry_count; i ++) + deh[i].deh_location += + (- new_dehs[new_entry_count - 1].deh_location + insert_point + DEH_SIZE * new_entry_count - item); + + + /* change item key if neccessary (when we paste before 0-th entry */ + if (!before) + { +#ifdef CONFIG_REISERFS_CHECK +/* + if ( old_entry_num && COMP_SHORT_KEYS ((unsigned long *)&ih->ih_key.k_offset, + &(new_dehs->deh_offset)) <= 0) + reiserfs_panic (0, "leaf_paste_entries: 10235: new key must be less, that old key"); +*/ +#endif + set_le_ih_k_offset (ih, le32_to_cpu (new_dehs->deh_offset)); +/* memcpy (&ih->ih_key.k_offset, + &new_dehs->deh_offset, SHORT_KEY_SIZE);*/ + } + +#ifdef CONFIG_REISERFS_CHECK + { + int prev, next; + /* check record locations */ + deh = B_I_DEH (bh, ih); + for (i = 0; i < I_ENTRY_COUNT(ih); i ++) { + next = (i < I_ENTRY_COUNT(ih) - 1) ? deh[i + 1].deh_location : 0; + prev = (i != 0) ? deh[i - 1].deh_location : 0; + + if (prev && prev <= deh[i].deh_location) + reiserfs_warning ("vs-10240: leaf_paste_entries: directory item (%h) corrupted (prev %a, cur(%d) %a)\n", + ih, deh + i - 1, i, deh + i); + if (next && next >= deh[i].deh_location) + reiserfs_warning ("vs-10250: leaf_paste_entries: directory item (%h) corrupted (cur(%d) %a, next %a)\n", + ih, i, deh + i, deh + i + 1); + } + } +#endif + +} + + + diff --git a/fs/reiserfs/namei.c b/fs/reiserfs/namei.c new file mode 100644 index 000000000..ffb8920bd --- /dev/null +++ b/fs/reiserfs/namei.c @@ -0,0 +1,1221 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/sched.h> +#include <linux/reiserfs_fs.h> +#include <linux/smp_lock.h> + +#else + +#include "nokernel.h" + +#endif + + /* there should be an overview right + here, as there should be in every + conceptual grouping of code. This + should be combined with dir.c and + called dir.c (naming will become + too large to be called one file in + a few years), stop senselessly + imitating the incoherent + structuring of code used by other + filesystems. */ + +#define INC_DIR_INODE_NLINK(i) if (i->i_nlink != 1) { i->i_nlink++; if (i->i_nlink >= REISERFS_LINK_MAX) i->i_nlink=1; } +#define DEC_DIR_INODE_NLINK(i) if (i->i_nlink != 1) i->i_nlink--; + +// directory item contains array of entry headers. This performs +// binary search through that array +static int bin_search_in_dir_item (struct reiserfs_dir_entry * de, loff_t off) +{ + struct item_head * ih = de->de_ih; + struct reiserfs_de_head * deh = de->de_deh; + int rbound, lbound, j; + + lbound = 0; + rbound = I_ENTRY_COUNT (ih) - 1; + + for (j = (rbound + lbound) / 2; lbound <= rbound; j = (rbound + lbound) / 2) { + if (off < deh_offset (deh + j)) { + rbound = j - 1; + continue; + } + if (off > deh_offset (deh + j)) { + lbound = j + 1; + continue; + } + // this is not name found, but matched third key component + de->de_entry_num = j; + return NAME_FOUND; + } + + de->de_entry_num = lbound; + return NAME_NOT_FOUND; +} + + +// comment? maybe something like set de to point to what the path points to? +static inline void set_de_item_location (struct reiserfs_dir_entry * de, struct path * path) +{ + de->de_bh = get_bh (path); + de->de_ih = get_ih (path); + de->de_deh = B_I_DEH (de->de_bh, de->de_ih); + de->de_item_num = PATH_LAST_POSITION (path); +} + + +// de_bh, de_ih, de_deh (points to first element of array), de_item_num is set +inline void set_de_name_and_namelen (struct reiserfs_dir_entry * de) +{ + struct reiserfs_de_head * deh = de->de_deh + de->de_entry_num; + + if (de->de_entry_num >= ih_entry_count (de->de_ih)) + BUG (); + + de->de_entrylen = entry_length (de->de_bh, de->de_ih, de->de_entry_num); + de->de_namelen = de->de_entrylen - (de_with_sd (deh) ? SD_SIZE : 0); + de->de_name = B_I_PITEM (de->de_bh, de->de_ih) + le16_to_cpu (deh->deh_location); + if (de->de_name[de->de_namelen - 1] == 0) + de->de_namelen = strlen (de->de_name); +} + + +// what entry points to +static inline void set_de_object_key (struct reiserfs_dir_entry * de) +{ + if (de->de_entry_num >= ih_entry_count (de->de_ih)) + BUG (); + de->de_dir_id = le32_to_cpu (de->de_deh[de->de_entry_num].deh_dir_id); + de->de_objectid = le32_to_cpu (de->de_deh[de->de_entry_num].deh_objectid); +} + + +static inline void store_de_entry_key (struct reiserfs_dir_entry * de) +{ + struct reiserfs_de_head * deh = de->de_deh + de->de_entry_num; + + if (de->de_entry_num >= ih_entry_count (de->de_ih)) + BUG (); + + /* store key of the found entry */ + de->de_entry_key.version = ITEM_VERSION_1; + de->de_entry_key.on_disk_key.k_dir_id = le32_to_cpu (de->de_ih->ih_key.k_dir_id); + de->de_entry_key.on_disk_key.k_objectid = le32_to_cpu (de->de_ih->ih_key.k_objectid); + set_cpu_key_k_offset (&(de->de_entry_key), deh_offset (deh)); + set_cpu_key_k_type (&(de->de_entry_key), TYPE_DIRENTRY); +} + + +/* We assign a key to each directory item, and place multiple entries +in a single directory item. A directory item has a key equal to the +key of the first directory entry in it. + +This function first calls search_by_key, then, if item whose first +entry matches is not found it looks for the entry inside directory +item found by search_by_key. Fills the path to the entry, and to the +entry position in the item + +*/ + +/* The function is NOT SCHEDULE-SAFE! */ +int search_by_entry_key (struct super_block * sb, struct cpu_key * key, + struct path * path, struct reiserfs_dir_entry * de) +{ + int retval; + + retval = search_item (sb, key, path); + switch (retval) { + case ITEM_NOT_FOUND: + if (!PATH_LAST_POSITION (path)) { + reiserfs_warning ("vs-7000: search_by_entry_key: search_by_key returned item position == 0"); + pathrelse(path) ; + return IO_ERROR ; + } + PATH_LAST_POSITION (path) --; + + case ITEM_FOUND: + break; + + case IO_ERROR: + return retval; + + default: + pathrelse (path); + reiserfs_warning ("vs-7002: search_by_entry_key: no path to here"); + return IO_ERROR; + } + + set_de_item_location (de, path); + +#ifdef CONFIG_REISERFS_CHECK + if (!is_direntry_le_ih (de->de_ih) || + COMP_SHORT_KEYS (&(de->de_ih->ih_key), key)) { + print_block (de->de_bh, 0, -1, -1); + reiserfs_panic (sb, "vs-7005: search_by_entry_key: found item %h is not directory item or " + "does not belong to the same directory as key %k", de->de_ih, key); + } +#endif /* CONFIG_REISERFS_CHECK */ + + /* binary search in directory item by third componen t of the + key. sets de->de_entry_num of de */ + retval = bin_search_in_dir_item (de, cpu_key_k_offset (key)); + path->pos_in_item = de->de_entry_num; + if (retval != NAME_NOT_FOUND) { + // ugly, but rename needs de_bh, de_deh, de_name, de_namelen, de_objectid set + set_de_name_and_namelen (de); + set_de_object_key (de); + } + return retval; +} + + + +/* Keyed 32-bit hash function using TEA in a Davis-Meyer function */ + +/* The third component is hashed, and you can choose from more than + one hash function. Per directory hashes are not yet implemented + but are thought about. This function should be moved to hashes.c + Jedi, please do so. -Hans */ + +static __u32 get_third_component (struct super_block * s, + const char * name, int len) +{ + __u32 res; + + if (!len || (len == 1 && name[0] == '.')) + return DOT_OFFSET; + if (len == 2 && name[0] == '.' && name[1] == '.') + return DOT_DOT_OFFSET; + + res = s->u.reiserfs_sb.s_hash_function (name, len); + + // take bits from 7-th to 30-th including both bounds + res = GET_HASH_VALUE(res); + if (res == 0) + // needed to have no names before "." and ".." those have hash + // value == 0 and generation conters 1 and 2 accordingly + res = 128; + return res + MAX_GENERATION_NUMBER; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +static int reiserfs_match (struct reiserfs_dir_entry * de, + const char * name, int namelen) +{ + int retval = NAME_NOT_FOUND; + + if ((namelen == de->de_namelen) && + !memcmp(de->de_name, name, de->de_namelen)) + retval = (de_visible (de->de_deh + de->de_entry_num) ? NAME_FOUND : NAME_FOUND_INVISIBLE); + + return retval; +} + + +/* de's de_bh, de_ih, de_deh, de_item_num, de_entry_num are set already */ + + /* used when hash collisions exist */ + + +static int linear_search_in_dir_item (struct cpu_key * key, struct reiserfs_dir_entry * de, + const char * name, int namelen) +{ + struct reiserfs_de_head * deh = de->de_deh; + int retval; + int i; + + i = de->de_entry_num; + + if (i == I_ENTRY_COUNT (de->de_ih) || + GET_HASH_VALUE (deh_offset (deh + i)) != GET_HASH_VALUE (cpu_key_k_offset (key))) { + i --; + } + +#ifdef CONFIG_REISERFS_CHECK + if (de->de_deh != B_I_DEH (de->de_bh, de->de_ih)) + reiserfs_panic (0, "vs-7010: linear_search_in_dir_item: array of entry headers not found"); +#endif /* CONFIG_REISERFS_CHECK */ + + deh += i; + + for (; i >= 0; i --, deh --) { + if (GET_HASH_VALUE (deh_offset (deh)) != + GET_HASH_VALUE (cpu_key_k_offset (key))) { + // hash value does not match, no need to check whole name + return NAME_NOT_FOUND; + } + + /* mark, that this generation number is used */ + if (de->de_gen_number_bit_string) + set_bit (GET_GENERATION_NUMBER (deh_offset (deh)), de->de_gen_number_bit_string); + + // calculate pointer to name and namelen + de->de_entry_num = i; + set_de_name_and_namelen (de); + + if ((retval = reiserfs_match (de, name, namelen)) != NAME_NOT_FOUND) { + // de's de_name, de_namelen, de_recordlen are set. Fill the rest: + + // key of pointed object + set_de_object_key (de); + + store_de_entry_key (de); + + // retval can be NAME_FOUND or NAME_FOUND_INVISIBLE + return retval; + } + } + + if (GET_GENERATION_NUMBER (le_ih_k_offset (de->de_ih)) == 0) + /* we have reached left most entry in the node. In common we + have to go to the left neighbor, but if generation counter + is 0 already, we know for sure, that there is no name with + the same hash value */ + // FIXME: this work correctly only because hash value can not + // be 0. Btw, in case of Yura's hash it is probably possible, + // so, this is a bug + return NAME_NOT_FOUND; + +#ifdef CONFIG_REISERFS_CHECK + if (de->de_item_num) + reiserfs_panic (0, "vs-7015: linear_search_in_dir_item: " + "two diritems of the same directory in one node?"); +#endif /* CONFIG_REISERFS_CHECK */ + + return GOTO_PREVIOUS_ITEM; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +// may return NAME_FOUND, NAME_FOUND_INVISIBLE, NAME_NOT_FOUND +// FIXME: should add something like IOERROR +static int reiserfs_find_entry (struct inode * dir, const char * name, int namelen, + struct path * path_to_entry, struct reiserfs_dir_entry * de) +{ + struct cpu_key key_to_search; + int retval; + + + if (namelen > REISERFS_MAX_NAME_LEN (dir->i_sb->s_blocksize)) + return NAME_NOT_FOUND; + + /* we will search for this key in the tree */ + make_cpu_key (&key_to_search, dir, + get_third_component (dir->i_sb, name, namelen), TYPE_DIRENTRY, 3); + + while (1) { + retval = search_by_entry_key (dir->i_sb, &key_to_search, path_to_entry, de); + if (retval == IO_ERROR) + // FIXME: still has to be dealt with + + /* I want you to conform to our error + printing standard. How many times + do I have to ask? -Hans */ + + BUG (); + + /* compare names for all entries having given hash value */ + retval = linear_search_in_dir_item (&key_to_search, de, name, namelen); + if (retval != GOTO_PREVIOUS_ITEM) { + /* there is no need to scan directory anymore. Given entry found or does not exist */ + path_to_entry->pos_in_item = de->de_entry_num; + return retval; + } + + /* there is left neighboring item of this directory and given entry can be there */ + set_cpu_key_k_offset (&key_to_search, le_ih_k_offset (de->de_ih) - 1); + pathrelse (path_to_entry); + + } /* while (1) */ +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +struct dentry * reiserfs_lookup (struct inode * dir, struct dentry * dentry) +{ + int retval; + struct inode * inode = 0; + struct reiserfs_dir_entry de; + INITIALIZE_PATH (path_to_entry); + + reiserfs_check_lock_depth("lookup") ; + + if (dentry->d_name.len > REISERFS_MAX_NAME_LEN (dir->i_sb->s_blocksize)) + return ERR_PTR(-ENAMETOOLONG); + + de.de_gen_number_bit_string = 0; + retval = reiserfs_find_entry (dir, dentry->d_name.name, dentry->d_name.len, &path_to_entry, &de); + pathrelse (&path_to_entry); + if (retval == NAME_FOUND) { + inode = reiserfs_iget (dir->i_sb, (struct cpu_key *)&(de.de_dir_id)); + if (!inode) { + return ERR_PTR(-EACCES); + } + } + + d_add(dentry, inode); + return NULL; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// + +/* add entry to the directory (entry can be hidden). + +insert definition of when hidden directories are used here -Hans + + Does not mark dir inode dirty, do it after successesfull call to it */ + +static int reiserfs_add_entry (struct reiserfs_transaction_handle *th, struct inode * dir, + const char * name, int namelen, struct inode * inode, + int visible) +{ + struct cpu_key entry_key; + struct reiserfs_de_head * deh; + INITIALIZE_PATH (path); + struct reiserfs_dir_entry de; + int bit_string [MAX_GENERATION_NUMBER / (sizeof(int) * 8) + 1]; + int gen_number; + char small_buf[32+DEH_SIZE] ; /* 48 bytes now and we avoid kmalloc + if we create file with short name */ + char * buffer; + int buflen, paste_size; + int retval; + + + /* cannot allow items to be added into a busy deleted directory */ + if (!namelen) + return -EINVAL; + + if (namelen > REISERFS_MAX_NAME_LEN (dir->i_sb->s_blocksize)) + return -ENAMETOOLONG; + + /* each entry has unique key. compose it */ + make_cpu_key (&entry_key, dir, + get_third_component (dir->i_sb, name, namelen), TYPE_DIRENTRY, 3); + + /* get memory for composing the entry */ + buflen = DEH_SIZE + ROUND_UP (namelen); + if (buflen > sizeof (small_buf)) { + buffer = reiserfs_kmalloc (buflen, GFP_BUFFER, dir->i_sb); + if (buffer == 0) + return -ENOMEM; + } else + buffer = small_buf; + + paste_size = (old_format_only (dir->i_sb)) ? (DEH_SIZE + namelen) : buflen; + + /* fill buffer : directory entry head, name[, dir objectid | , stat data | ,stat data, dir objectid ] */ + deh = (struct reiserfs_de_head *)buffer; + deh->deh_location = 0; + deh->deh_offset = cpu_to_le32 (cpu_key_k_offset (&entry_key)); + deh->deh_state = 0; + /* put key (ino analog) to de */ + deh->deh_dir_id = INODE_PKEY (inode)->k_dir_id; + deh->deh_objectid = INODE_PKEY (inode)->k_objectid; + + /* copy name */ + memcpy ((char *)(deh + 1), name, namelen); + /* padd by 0s to the 4 byte boundary */ + padd_item ((char *)(deh + 1), ROUND_UP (namelen), namelen); + + /* entry is ready to be pasted into tree, set 'visibility' and 'stat data in entry' attributes */ + mark_de_without_sd (deh); + visible ? mark_de_visible (deh) : mark_de_hidden (deh); + + /* find the proper place for the new entry */ + memset (bit_string, 0, sizeof (bit_string)); + de.de_gen_number_bit_string = (char *)bit_string; + if (reiserfs_find_entry (dir, name, namelen, &path, &de) == NAME_FOUND) { + if (buffer != small_buf) + reiserfs_kfree (buffer, buflen, dir->i_sb); + pathrelse (&path); + return -EEXIST; + } + + if (find_first_nonzero_bit (bit_string, MAX_GENERATION_NUMBER + 1) < MAX_GENERATION_NUMBER + 1) { + /* there are few names with given hash value */ + gen_number = find_first_zero_bit (bit_string, MAX_GENERATION_NUMBER + 1); + if (gen_number > MAX_GENERATION_NUMBER) { + /* there is no free generation number */ + reiserfs_warning ("reiserfs_add_entry: Congratulations! we have got hash function screwed up\n"); + if (buffer != small_buf) + reiserfs_kfree (buffer, buflen, dir->i_sb); + pathrelse (&path); + return -EHASHCOLLISION;//EBADSLT + } + /* adjust offset of directory enrty */ + deh->deh_offset = cpu_to_le32 (SET_GENERATION_NUMBER (deh_offset (deh), gen_number)); + set_cpu_key_k_offset (&entry_key, le32_to_cpu (deh->deh_offset)); + + /* find place for new entry */ + if (search_by_entry_key (dir->i_sb, &entry_key, &path, &de) == NAME_FOUND) { + reiserfs_warning ("vs-7032: reiserfs_add_entry: " + "entry with this key (%k) already exists", &entry_key); + if (buffer != small_buf) + reiserfs_kfree (buffer, buflen, dir->i_sb); + pathrelse (&path); + return -EHASHCOLLISION; + } + } else { + deh->deh_offset = cpu_to_le32 (SET_GENERATION_NUMBER (le32_to_cpu (deh->deh_offset), 0)); + set_cpu_key_k_offset (&entry_key, le32_to_cpu (deh->deh_offset)); + } + + /* perform the insertion of the entry that we have prepared */ + retval = reiserfs_paste_into_item (th, &path, &entry_key, buffer, paste_size); + if (buffer != small_buf) + reiserfs_kfree (buffer, buflen, dir->i_sb); + if (retval) { + reiserfs_check_path(&path) ; + return retval; + } + + dir->i_size += paste_size; + dir->i_blocks = ((dir->i_size + 511) >> 9); + dir->i_mtime = dir->i_ctime = CURRENT_TIME; + if (!S_ISDIR (inode->i_mode) && visible) + // reiserfs_mkdir or reiserfs_rename will do that by itself + reiserfs_update_sd (th, dir); + + reiserfs_check_path(&path) ; + return 0; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +int reiserfs_create (struct inode * dir, struct dentry *dentry, int mode) +{ + int retval; + struct inode * inode; + int windex ; + int jbegin_count = JOURNAL_PER_BALANCE_CNT * 2 ; + struct reiserfs_transaction_handle th ; + + + inode = get_empty_inode() ; + if (!inode) { + return -ENOMEM ; + } + journal_begin(&th, dir->i_sb, jbegin_count) ; + th.t_caller = "create" ; + windex = push_journal_writer("reiserfs_create") ; + inode = reiserfs_new_inode (&th, dir, mode, 0, 0/*i_size*/, dentry, inode, &retval); + if (!inode) { + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return retval; + } + + inode->i_op = &reiserfs_file_inode_operations; + inode->i_fop = &reiserfs_file_operations; + inode->i_mapping->a_ops = &reiserfs_address_space_operations ; + + retval = reiserfs_add_entry (&th, dir, dentry->d_name.name, dentry->d_name.len, + inode, 1/*visible*/); + if (retval) { + inode->i_nlink--; + reiserfs_update_sd (&th, inode); + pop_journal_writer(windex) ; + // FIXME: should we put iput here and have stat data deleted + // in the same transactioin + journal_end(&th, dir->i_sb, jbegin_count) ; + iput (inode); + return retval; + } + + d_instantiate(dentry, inode); + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return 0; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +int reiserfs_mknod (struct inode * dir, struct dentry *dentry, int mode, int rdev) +{ + int retval; + struct inode * inode; + int windex ; + struct reiserfs_transaction_handle th ; + int jbegin_count = JOURNAL_PER_BALANCE_CNT * 3; + + inode = get_empty_inode() ; + if (!inode) { + return -ENOMEM ; + } + journal_begin(&th, dir->i_sb, jbegin_count) ; + windex = push_journal_writer("reiserfs_mknod") ; + + inode = reiserfs_new_inode (&th, dir, mode, 0, 0/*i_size*/, dentry, inode, &retval); + if (!inode) { + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return retval; + } + + init_special_inode(inode, mode, rdev) ; + + //FIXME: needed for block and char devices only + reiserfs_update_sd (&th, inode); + + retval = reiserfs_add_entry (&th, dir, dentry->d_name.name, dentry->d_name.len, + inode, 1/*visible*/); + if (retval) { + inode->i_nlink--; + reiserfs_update_sd (&th, inode); + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + iput (inode); + return retval; + } + + d_instantiate(dentry, inode); + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return 0; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +int reiserfs_mkdir (struct inode * dir, struct dentry *dentry, int mode) +{ + int retval; + struct inode * inode; + int windex ; + struct reiserfs_transaction_handle th ; + int jbegin_count = JOURNAL_PER_BALANCE_CNT * 3; + + inode = get_empty_inode() ; + if (!inode) { + return -ENOMEM ; + } + journal_begin(&th, dir->i_sb, jbegin_count) ; + windex = push_journal_writer("reiserfs_mkdir") ; + + /* inc the link count now, so another writer doesn't overflow it while + ** we sleep later on. + */ + INC_DIR_INODE_NLINK(dir) + + mode = S_IFDIR | mode; + inode = reiserfs_new_inode (&th, dir, mode, 0/*symlink*/, + old_format_only (dir->i_sb) ? EMPTY_DIR_SIZE_V1 : EMPTY_DIR_SIZE, + dentry, inode, &retval); + if (!inode) { + pop_journal_writer(windex) ; + dir->i_nlink-- ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return retval; + } + + inode->i_op = &reiserfs_dir_inode_operations; + inode->i_fop = &reiserfs_dir_operations; + + // note, _this_ add_entry will not update dir's stat data + retval = reiserfs_add_entry (&th, dir, dentry->d_name.name, dentry->d_name.len, + inode, 1/*visible*/); + if (retval) { + inode->i_nlink = 0; + DEC_DIR_INODE_NLINK(dir); + reiserfs_update_sd (&th, inode); + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + iput (inode); + return retval; + } + + // the above add_entry did not update dir's stat data + reiserfs_update_sd (&th, dir); + + d_instantiate(dentry, inode); + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return 0; +} + +static inline int reiserfs_empty_dir(struct inode *inode) { + /* we can cheat because an old format dir cannot have + ** EMPTY_DIR_SIZE, and a new format dir cannot have + ** EMPTY_DIR_SIZE_V1. So, if the inode is either size, + ** regardless of disk format version, the directory is empty. + */ + if (inode->i_size != EMPTY_DIR_SIZE && + inode->i_size != EMPTY_DIR_SIZE_V1) { + return 0 ; + } + return 1 ; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +int reiserfs_rmdir (struct inode * dir, struct dentry *dentry) +{ + int retval; + struct inode * inode; + int windex ; + struct reiserfs_transaction_handle th ; + int jbegin_count = JOURNAL_PER_BALANCE_CNT * 3; + INITIALIZE_PATH (path); + struct reiserfs_dir_entry de; + + + journal_begin(&th, dir->i_sb, jbegin_count) ; + windex = push_journal_writer("reiserfs_rmdir") ; + + de.de_gen_number_bit_string = 0; + if (reiserfs_find_entry (dir, dentry->d_name.name, dentry->d_name.len, &path, &de) == NAME_NOT_FOUND) { + retval = -ENOENT; + goto end_rmdir; + } + inode = dentry->d_inode; + + if (de.de_objectid != inode->i_ino) { + // FIXME: compare key of an object and a key found in the + // entry + retval = -EIO; + goto end_rmdir; + } + if (!reiserfs_empty_dir(inode)) { + retval = -ENOTEMPTY; + goto end_rmdir; + } + + /* cut entry from dir directory */ + retval = reiserfs_cut_from_item (&th, &path, &(de.de_entry_key), dir, + NULL, /* page */ + 0/*new file size - not used here*/); + if (retval < 0) + goto end_rmdir; + + if ( inode->i_nlink != 2 && inode->i_nlink != 1 ) + printk ("reiserfs_rmdir: empty directory has nlink != 2 (%d)\n", inode->i_nlink); + + inode->i_nlink = 0; + inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME; + reiserfs_update_sd (&th, inode); + + DEC_DIR_INODE_NLINK(dir) + dir->i_size -= (DEH_SIZE + de.de_entrylen); + dir->i_blocks = ((dir->i_size + 511) >> 9); + reiserfs_update_sd (&th, dir); + + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + reiserfs_check_path(&path) ; + return 0; + + end_rmdir: + /* we must release path, because we did not call + reiserfs_cut_from_item, or reiserfs_cut_from_item does not + release path if operation was not complete */ + pathrelse (&path); + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return retval; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +int reiserfs_unlink (struct inode * dir, struct dentry *dentry) +{ + int retval; + struct inode * inode; + struct reiserfs_dir_entry de; + INITIALIZE_PATH (path); + int windex ; + struct reiserfs_transaction_handle th ; + int jbegin_count = JOURNAL_PER_BALANCE_CNT * 3; + + journal_begin(&th, dir->i_sb, jbegin_count) ; + windex = push_journal_writer("reiserfs_unlink") ; + + de.de_gen_number_bit_string = 0; + if (reiserfs_find_entry (dir, dentry->d_name.name, dentry->d_name.len, &path, &de) == NAME_NOT_FOUND) { + retval = -ENOENT; + goto end_unlink; + } + inode = dentry->d_inode; + + if (de.de_objectid != inode->i_ino) { + // FIXME: compare key of an object and a key found in the + // entry + retval = -EIO; + goto end_unlink; + } + + if (!inode->i_nlink) { + printk("reiserfs_unlink: deleting nonexistent file (%s:%lu), %d\n", + kdevname(inode->i_dev), inode->i_ino, inode->i_nlink); + inode->i_nlink = 1; + } + + retval = reiserfs_cut_from_item (&th, &path, &(de.de_entry_key), dir, NULL, 0); + if (retval < 0) + goto end_unlink; + + inode->i_nlink--; + inode->i_ctime = CURRENT_TIME; + reiserfs_update_sd (&th, inode); + + dir->i_size -= (de.de_entrylen + DEH_SIZE); + dir->i_blocks = ((dir->i_size + 511) >> 9); + dir->i_ctime = dir->i_mtime = CURRENT_TIME; + reiserfs_update_sd (&th, dir); + + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + reiserfs_check_path(&path) ; + return 0; + + end_unlink: + pathrelse (&path); + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + reiserfs_check_path(&path) ; + return retval; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +int reiserfs_symlink (struct inode * dir, struct dentry * dentry, const char * symname) +{ + int retval; + struct inode * inode; + char * name; + int item_len; + int windex ; + struct reiserfs_transaction_handle th ; + int jbegin_count = JOURNAL_PER_BALANCE_CNT * 3; + + + inode = get_empty_inode() ; + if (!inode) { + return -ENOMEM ; + } + + item_len = ROUND_UP (strlen (symname)); + if (item_len > MAX_ITEM_LEN (dir->i_sb->s_blocksize)) { + iput(inode) ; + return -ENAMETOOLONG; + } + + name = kmalloc (item_len, GFP_BUFFER); + if (!name) { + iput(inode) ; + return -ENOMEM; + } + memcpy (name, symname, strlen (symname)); + padd_item (name, item_len, strlen (symname)); + + journal_begin(&th, dir->i_sb, jbegin_count) ; + windex = push_journal_writer("reiserfs_symlink") ; + + inode = reiserfs_new_inode (&th, dir, S_IFLNK | S_IRWXUGO, name, strlen (symname), dentry, + inode, &retval); + kfree (name); + if (inode == 0) { /* reiserfs_new_inode iputs for us */ + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return retval; + } + + inode->i_op = &page_symlink_inode_operations; + inode->i_mapping->a_ops = &reiserfs_address_space_operations; + + // must be sure this inode is written with this transaction + // + //reiserfs_update_sd (&th, inode, READ_BLOCKS); + + retval = reiserfs_add_entry (&th, dir, dentry->d_name.name, dentry->d_name.len, + inode, 1/*visible*/); + if (retval) { + inode->i_nlink--; + reiserfs_update_sd (&th, inode); + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + iput (inode); + return retval; + } + + d_instantiate(dentry, inode); + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return 0; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +int reiserfs_link (struct dentry * old_dentry, struct inode * dir, struct dentry * dentry) +{ + int retval; + struct inode *inode = old_dentry->d_inode; + int windex ; + struct reiserfs_transaction_handle th ; + int jbegin_count = JOURNAL_PER_BALANCE_CNT * 3; + + + if (S_ISDIR(inode->i_mode)) + return -EPERM; + + if (inode->i_nlink >= REISERFS_LINK_MAX) { + //FIXME: sd_nlink is 32 bit for new files + return -EMLINK; + } + + journal_begin(&th, dir->i_sb, jbegin_count) ; + windex = push_journal_writer("reiserfs_link") ; + + /* create new entry */ + retval = reiserfs_add_entry (&th, dir, dentry->d_name.name, dentry->d_name.len, + inode, 1/*visible*/); + if (retval) { + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return retval; + } + + inode->i_nlink++; + inode->i_ctime = CURRENT_TIME; + reiserfs_update_sd (&th, inode); + + atomic_inc(&inode->i_count) ; + d_instantiate(dentry, inode); + pop_journal_writer(windex) ; + journal_end(&th, dir->i_sb, jbegin_count) ; + return 0; +} + + +// de contains information pointing to an entry which +static int de_still_valid (const char * name, int len, struct reiserfs_dir_entry * de) +{ + struct reiserfs_dir_entry tmp = *de; + + // recalculate pointer to name and name length + set_de_name_and_namelen (&tmp); + // FIXME: could check more + if (tmp.de_namelen != len || memcmp (name, de->de_name, len)) + return 0; + return 1; +} + + +static int entry_points_to_object (const char * name, int len, struct reiserfs_dir_entry * de, struct inode * inode) +{ + if (!de_still_valid (name, len, de)) + return 0; + + if (inode) { + if (!de_visible (de->de_deh + de->de_entry_num)) + reiserfs_panic (0, "vs-7042: entry_points_to_object: entry must be visible"); + return (de->de_objectid == inode->i_ino) ? 1 : 0; + } + + /* this must be added hidden entry */ + if (de_visible (de->de_deh + de->de_entry_num)) + reiserfs_panic (0, "vs-7043: entry_points_to_object: entry must be visible"); + + return 1; +} + + +/* sets key of objectid the entry has to point to */ +static void set_ino_in_dir_entry (struct reiserfs_dir_entry * de, struct key * key) +{ + de->de_deh[de->de_entry_num].deh_dir_id = key->k_dir_id; + de->de_deh[de->de_entry_num].deh_objectid = key->k_objectid; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// + +/* + * process, that is going to call fix_nodes/do_balance must hold only + * one path. If it holds 2 or more, it can get into endless waiting in + * get_empty_nodes or its clones + */ +int reiserfs_rename (struct inode * old_dir, struct dentry *old_dentry, + struct inode * new_dir, struct dentry *new_dentry) +{ + int retval; + INITIALIZE_PATH (old_entry_path); + INITIALIZE_PATH (new_entry_path); + INITIALIZE_PATH (dot_dot_entry_path); + struct item_head new_entry_ih, old_entry_ih ; + struct reiserfs_dir_entry old_de, new_de, dot_dot_de; + struct inode * old_inode, * new_inode; + int windex ; + struct reiserfs_transaction_handle th ; + int jbegin_count = JOURNAL_PER_BALANCE_CNT * 3; + + + old_inode = old_dentry->d_inode; + new_inode = new_dentry->d_inode; + + // make sure, that oldname still exists and points to an object we + // are going to rename + old_de.de_gen_number_bit_string = 0; + retval = reiserfs_find_entry (old_dir, old_dentry->d_name.name, old_dentry->d_name.len, + &old_entry_path, &old_de); + pathrelse (&old_entry_path); + if (retval != NAME_FOUND || old_de.de_objectid != old_inode->i_ino) { + // FIXME: IO error is possible here + return -ENOENT; + } + + if (S_ISDIR(old_inode->i_mode)) { + // make sure, that directory being renamed has correct ".." + // and that its new parent directory has not too many links + // already + + if (new_inode) { + if (!reiserfs_empty_dir(new_inode)) { + return -ENOTEMPTY; + } + } + + /* directory is renamed, its parent directory will be changed, + ** so find ".." entry + */ + dot_dot_de.de_gen_number_bit_string = 0; + retval = reiserfs_find_entry (old_inode, "..", 2, &dot_dot_entry_path, &dot_dot_de); + pathrelse (&dot_dot_entry_path); + if (retval != NAME_FOUND) + return -EIO; + + /* inode number of .. must equal old_dir->i_ino */ + if (dot_dot_de.de_objectid != old_dir->i_ino) + return -EIO; + } + + journal_begin(&th, old_dir->i_sb, jbegin_count) ; + windex = push_journal_writer("reiserfs_rename") ; + + /* add new entry (or find the existing one) */ + retval = reiserfs_add_entry (&th, new_dir, new_dentry->d_name.name, new_dentry->d_name.len, + old_inode, 0); + if (retval == -EEXIST) { + // FIXME: is it possible, that new_inode == 0 here? If yes, it + // is not clear how does ext2 handle that + if (!new_inode) { + printk ("reiserfs_rename: new entry is found, new inode == 0\n"); + BUG (); + } + } else if (retval) { + pop_journal_writer(windex) ; + journal_end(&th, old_dir->i_sb, jbegin_count) ; + return retval; + } + + + while (1) { + // look for old name using corresponding entry key (found by reiserfs_find_entry) + if (search_by_entry_key (new_dir->i_sb, &old_de.de_entry_key, &old_entry_path, &old_de) != NAME_FOUND) + BUG (); + + copy_item_head(&old_entry_ih, get_ih(&old_entry_path)) ; + + // look for new name by reiserfs_find_entry + new_de.de_gen_number_bit_string = 0; + retval = reiserfs_find_entry (new_dir, new_dentry->d_name.name, new_dentry->d_name.len, + &new_entry_path, &new_de); + if (retval != NAME_FOUND_INVISIBLE && retval != NAME_FOUND) + BUG (); + + copy_item_head(&new_entry_ih, get_ih(&new_entry_path)) ; + + reiserfs_prepare_for_journal(old_inode->i_sb, new_de.de_bh, 1) ; + + if (S_ISDIR(old_inode->i_mode)) { + if (search_by_entry_key (new_dir->i_sb, &dot_dot_de.de_entry_key, &dot_dot_entry_path, &dot_dot_de) != NAME_FOUND) + BUG (); + // node containing ".." gets into transaction + reiserfs_prepare_for_journal(old_inode->i_sb, dot_dot_de.de_bh, 1) ; + } + /* we should check seals here, not do + this stuff, yes? Then, having + gathered everything into RAM we + should lock the buffers, yes? -Hans */ + /* probably. our rename needs to hold more + ** than one path at once. The seals would + ** have to be written to deal with multi-path + ** issues -chris + */ + /* sanity checking before doing the rename - avoid races many + ** of the above checks could have scheduled. We have to be + ** sure our items haven't been shifted by another process. + */ + if (!entry_points_to_object(new_dentry->d_name.name, + new_dentry->d_name.len, + &new_de, new_inode) || + item_moved(&new_entry_ih, &new_entry_path) || + item_moved(&old_entry_ih, &old_entry_path) || + !entry_points_to_object (old_dentry->d_name.name, + old_dentry->d_name.len, + &old_de, old_inode)) { + reiserfs_restore_prepared_buffer (old_inode->i_sb, new_de.de_bh); + if (S_ISDIR(old_inode->i_mode)) + reiserfs_restore_prepared_buffer (old_inode->i_sb, dot_dot_de.de_bh); +#if 0 + // FIXME: do we need this? shouldn't we simply continue? + run_task_queue(&tq_disk); + current->policy |= SCHED_YIELD; + /*current->counter = 0;*/ + schedule(); +#endif + continue; + } + +#ifdef CONFIG_REISERFS_CHECK + if (S_ISDIR(old_inode->i_mode) && + (!entry_points_to_object ("..", 2, &dot_dot_de, old_dir) || + !reiserfs_buffer_prepared(dot_dot_de.de_bh))) { + // this should be not changed + BUG (); + } +#endif + + break; + } + + /* ok, all the changes can be done in one fell swoop when we + have claimed all the buffers needed.*/ + + mark_de_visible (new_de.de_deh + new_de.de_entry_num); + set_ino_in_dir_entry (&new_de, INODE_PKEY (old_inode)); + journal_mark_dirty (&th, old_dir->i_sb, new_de.de_bh); + + mark_de_hidden (old_de.de_deh + old_de.de_entry_num); + old_dir->i_ctime = old_dir->i_mtime = CURRENT_TIME; + new_dir->i_ctime = new_dir->i_mtime = CURRENT_TIME; + + if (new_inode) { + // adjust link number of the victim + if (S_ISDIR(new_inode->i_mode)) { + DEC_DIR_INODE_NLINK(new_inode) + } else { + new_inode->i_nlink--; + } + new_inode->i_ctime = CURRENT_TIME; + } + + if (S_ISDIR(old_inode->i_mode)) { + //if (dot_dot_de.de_bh) { + // adjust ".." of renamed directory + set_ino_in_dir_entry (&dot_dot_de, INODE_PKEY (new_dir)); + journal_mark_dirty (&th, new_dir->i_sb, dot_dot_de.de_bh); + + DEC_DIR_INODE_NLINK(old_dir) + if (new_inode) { + if (S_ISDIR(new_inode->i_mode)) { + DEC_DIR_INODE_NLINK(new_inode) + } else { + new_inode->i_nlink--; + } + } else { + INC_DIR_INODE_NLINK(new_dir) + } + } + + // looks like in 2.3.99pre3 brelse is atomic. so we can use pathrelse + pathrelse (&new_entry_path); + pathrelse (&dot_dot_entry_path); + + // FIXME: this reiserfs_cut_from_item's return value may screw up + // anybody, but it will panic if will not be able to find the + // entry. This needs one more clean up + if (reiserfs_cut_from_item (&th, &old_entry_path, &(old_de.de_entry_key), old_dir, NULL, 0) < 0) + reiserfs_warning ("vs-: reiserfs_rename: coudl not cut old name. Fsck later?\n"); + + old_dir->i_size -= DEH_SIZE + old_de.de_entrylen; + old_dir->i_blocks = ((old_dir->i_size + 511) >> 9); + + reiserfs_update_sd (&th, old_dir); + reiserfs_update_sd (&th, new_dir); + if (new_inode) + reiserfs_update_sd (&th, new_inode); + + pop_journal_writer(windex) ; + journal_end(&th, old_dir->i_sb, jbegin_count) ; + return 0; +} + diff --git a/fs/reiserfs/objectid.c b/fs/reiserfs/objectid.c new file mode 100644 index 000000000..0bcbd4559 --- /dev/null +++ b/fs/reiserfs/objectid.c @@ -0,0 +1,211 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/string.h> +#include <linux/locks.h> +#include <linux/sched.h> +#include <linux/reiserfs_fs.h> + +#else + +#include "nokernel.h" + +#endif + + +// find where objectid map starts +#define objectid_map(s,rs) (old_format_only (s) ? \ + (__u32 *)((struct reiserfs_super_block_v1 *)rs + 1) :\ + (__u32 *)(rs + 1)) + + +#ifdef CONFIG_REISERFS_CHECK + +static void check_objectid_map (struct super_block * s, __u32 * map) +{ + if (le32_to_cpu (map[0]) != 1) + reiserfs_panic (s, "vs-15010: check_objectid_map: map corrupted"); + + // FIXME: add something else here +} + +#endif + + +/* When we allocate objectids we allocate the first unused objectid. + Each sequence of objectids in use (the odd sequences) is followed + by a sequence of objectids not in use (the even sequences). We + only need to record the last objectid in each of these sequences + (both the odd and even sequences) in order to fully define the + boundaries of the sequences. A consequence of allocating the first + objectid not in use is that under most conditions this scheme is + extremely compact. The exception is immediately after a sequence + of operations which deletes a large number of objects of + non-sequential objectids, and even then it will become compact + again as soon as more objects are created. Note that many + interesting optimizations of layout could result from complicating + objectid assignment, but we have deferred making them for now. */ + + +/* get unique object identifier */ +__u32 reiserfs_get_unused_objectid (struct reiserfs_transaction_handle *th) +{ + struct super_block * s = th->t_super; + struct reiserfs_super_block * rs = SB_DISK_SUPER_BLOCK (s); + __u32 * map = objectid_map (s, rs); + __u32 unused_objectid; + + +#ifdef CONFIG_REISERFS_CHECK + check_objectid_map (s, map); +#endif + + reiserfs_prepare_for_journal(s, SB_BUFFER_WITH_SB(s), 1) ; + /* comment needed -Hans */ + unused_objectid = le32_to_cpu (map[1]); + if (unused_objectid == U32_MAX) { + printk ("REISERFS: get_objectid: no more object ids\n"); + reiserfs_restore_prepared_buffer(s, SB_BUFFER_WITH_SB(s)) ; + return 0; + } + + /* This incrementation allocates the first unused objectid. That + is to say, the first entry on the objectid map is the first + unused objectid, and by incrementing it we use it. See below + where we check to see if we eliminated a sequence of unused + objectids.... */ + map[1] = cpu_to_le32 (unused_objectid + 1); + + /* Now we check to see if we eliminated the last remaining member of + the first even sequence (and can eliminate the sequence by + eliminating its last objectid from oids), and can collapse the + first two odd sequences into one sequence. If so, then the net + result is to eliminate a pair of objectids from oids. We do this + by shifting the entire map to the left. */ + if (le16_to_cpu (rs->s_oid_cursize) > 2 && map[1] == map[2]) { + memmove (map + 1, map + 3, (le16_to_cpu (rs->s_oid_cursize) - 3) * sizeof(__u32)); + //rs->s_oid_cursize -= 2; + rs->s_oid_cursize = cpu_to_le16 (le16_to_cpu (rs->s_oid_cursize) - 2); + } + + journal_mark_dirty(th, s, SB_BUFFER_WITH_SB (s)); + s->s_dirt = 1; + return unused_objectid; +} + + +/* makes object identifier unused */ +void reiserfs_release_objectid (struct reiserfs_transaction_handle *th, + __u32 objectid_to_release) +{ + struct super_block * s = th->t_super; + struct reiserfs_super_block * rs = SB_DISK_SUPER_BLOCK (s); + __u32 * map = objectid_map (s, rs); + int i = 0; + + //return; +#ifdef CONFIG_REISERFS_CHECK + check_objectid_map (s, map); +#endif + + reiserfs_prepare_for_journal(s, SB_BUFFER_WITH_SB(s), 1) ; + journal_mark_dirty(th, s, SB_BUFFER_WITH_SB (s)); + s->s_dirt = 1; + + + /* start at the beginning of the objectid map (i = 0) and go to + the end of it (i = disk_sb->s_oid_cursize). Linear search is + what we use, though it is possible that binary search would be + more efficient after performing lots of deletions (which is + when oids is large.) We only check even i's. */ + while (i < le16_to_cpu (rs->s_oid_cursize)) { + if (objectid_to_release == le32_to_cpu (map[i])) { + /* This incrementation unallocates the objectid. */ + //map[i]++; + map[i] = cpu_to_le32 (le32_to_cpu (map[i]) + 1); + + /* Did we unallocate the last member of an odd sequence, and can shrink oids? */ + if (map[i] == map[i+1]) { + /* shrink objectid map */ + memmove (map + i, map + i + 2, + (le16_to_cpu (rs->s_oid_cursize) - i - 2) * sizeof (__u32)); + //disk_sb->s_oid_cursize -= 2; + rs->s_oid_cursize = cpu_to_le16 (le16_to_cpu (rs->s_oid_cursize) - 2); + +#ifdef CONFIG_REISERFS_CHECK + if (le16_to_cpu (rs->s_oid_cursize) < 2 || + le16_to_cpu (rs->s_oid_cursize) > le16_to_cpu (rs->s_oid_maxsize)) + reiserfs_panic (s, "vs-15005: reiserfs_release_objectid: " + "objectid map corrupted cur_size == %d (max == %d)", + le16_to_cpu (rs->s_oid_cursize), le16_to_cpu (rs->s_oid_maxsize)); +#endif + } + return; + } + + if (objectid_to_release > le32_to_cpu (map[i]) && + objectid_to_release < le32_to_cpu (map[i + 1])) { + /* size of objectid map is not changed */ + if (objectid_to_release + 1 == le32_to_cpu (map[i + 1])) { + //objectid_map[i+1]--; + map[i + 1] = cpu_to_le32 (le32_to_cpu (map[i + 1]) - 1); + return; + } + + if (rs->s_oid_cursize == rs->s_oid_maxsize) + /* objectid map must be expanded, but there is no space */ + return; + + /* expand the objectid map*/ + memmove (map + i + 3, map + i + 1, + (le16_to_cpu (rs->s_oid_cursize) - i - 1) * sizeof(__u32)); + map[i + 1] = cpu_to_le32 (objectid_to_release); + map[i + 2] = cpu_to_le32 (objectid_to_release + 1); + rs->s_oid_cursize = cpu_to_le16 (le16_to_cpu (rs->s_oid_cursize) + 2); + return; + } + i += 2; + } + + reiserfs_warning ("vs-15010: reiserfs_release_objectid: tried to free free object id (%lu)", + objectid_to_release); +} + + +int reiserfs_convert_objectid_map_v1(struct super_block *s) { + struct reiserfs_super_block *disk_sb = SB_DISK_SUPER_BLOCK (s); + int cur_size = le16_to_cpu(disk_sb->s_oid_cursize) ; + int new_size = (s->s_blocksize - SB_SIZE) / sizeof(__u32) / 2 * 2 ; + int old_max = le16_to_cpu(disk_sb->s_oid_maxsize) ; + struct reiserfs_super_block_v1 *disk_sb_v1 ; + __u32 *objectid_map, *new_objectid_map ; + int i ; + + disk_sb_v1=(struct reiserfs_super_block_v1 *)(SB_BUFFER_WITH_SB(s)->b_data); + objectid_map = (__u32 *)(disk_sb_v1 + 1) ; + new_objectid_map = (__u32 *)(disk_sb + 1) ; + + if (cur_size > new_size) { + /* mark everyone used that was listed as free at the end of the objectid + ** map + */ + objectid_map[new_size - 1] = objectid_map[cur_size - 1] ; + disk_sb->s_oid_cursize = cpu_to_le16(new_size) ; + } + /* move the smaller objectid map past the end of the new super */ + for (i = new_size - 1 ; i >= 0 ; i--) { + objectid_map[i + (old_max - new_size)] = objectid_map[i] ; + } + + + /* set the max size so we don't overflow later */ + disk_sb->s_oid_maxsize = cpu_to_le16(new_size) ; + + /* finally, zero out the unused chunk of the new super */ + memset(disk_sb->s_unused, 0, sizeof(disk_sb->s_unused)) ; + return 0 ; +} + diff --git a/fs/reiserfs/prints.c b/fs/reiserfs/prints.c new file mode 100644 index 000000000..40cabc500 --- /dev/null +++ b/fs/reiserfs/prints.c @@ -0,0 +1,881 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/sched.h> +#include <linux/fs.h> +#include <linux/reiserfs_fs.h> +#include <linux/string.h> + +#else + +#include "nokernel.h" +#include <limits.h> + +#endif + +#include <stdarg.h> + +static char error_buf[1024]; +static char fmt_buf[1024]; +static char off_buf[80]; + + +static char * cpu_offset (struct cpu_key * key) +{ + if (cpu_key_k_type(key) == TYPE_DIRENTRY) + sprintf (off_buf, "%Lu(%Lu)", + (unsigned long long)GET_HASH_VALUE (cpu_key_k_offset (key)), + (unsigned long long)GET_GENERATION_NUMBER (cpu_key_k_offset (key))); + else + sprintf (off_buf, "0x%Lx", (unsigned long long)cpu_key_k_offset (key)); + return off_buf; +} + + +static char * le_offset (struct key * key) +{ + int version; + + version = le_key_version (key); + if (le_key_k_type (version, key) == TYPE_DIRENTRY) + sprintf (off_buf, "%Lu(%Lu)", + (unsigned long long)GET_HASH_VALUE (le_key_k_offset (version, key)), + (unsigned long long)GET_GENERATION_NUMBER (le_key_k_offset (version, key))); + else + sprintf (off_buf, "0x%Lx", (unsigned long long)le_key_k_offset (version, key)); + return off_buf; +} + + +static char * cpu_type (struct cpu_key * key) +{ + if (cpu_key_k_type (key) == TYPE_STAT_DATA) + return "SD"; + if (cpu_key_k_type (key) == TYPE_DIRENTRY) + return "DIR"; + if (cpu_key_k_type (key) == TYPE_DIRECT) + return "DIRECT"; + if (cpu_key_k_type (key) == TYPE_INDIRECT) + return "IND"; + return "UNKNOWN"; +} + + +static char * le_type (struct key * key) +{ + int version; + + version = le_key_version (key); + + if (le_key_k_type (version, key) == TYPE_STAT_DATA) + return "SD"; + if (le_key_k_type (version, key) == TYPE_DIRENTRY) + return "DIR"; + if (le_key_k_type (version, key) == TYPE_DIRECT) + return "DIRECT"; + if (le_key_k_type (version, key) == TYPE_INDIRECT) + return "IND"; + return "UNKNOWN"; +} + + +/* %k */ +static void sprintf_le_key (char * buf, struct key * key) +{ + if (key) + sprintf (buf, "[%d %d %s %s]", le32_to_cpu (key->k_dir_id), + le32_to_cpu (key->k_objectid), le_offset (key), le_type (key)); + else + sprintf (buf, "[NULL]"); +} + + +/* %K */ +static void sprintf_cpu_key (char * buf, struct cpu_key * key) +{ + if (key) + sprintf (buf, "[%d %d %s %s]", key->on_disk_key.k_dir_id, + key->on_disk_key.k_objectid, cpu_offset (key), cpu_type (key)); + else + sprintf (buf, "[NULL]"); +} + + +static void sprintf_item_head (char * buf, struct item_head * ih) +{ + if (ih) { + sprintf (buf, "%s", (ih_version (ih) == ITEM_VERSION_2) ? "*NEW* " : "*OLD*"); + sprintf_le_key (buf + strlen (buf), &(ih->ih_key)); + sprintf (buf + strlen (buf), ", item_len %d, item_location %d, " + "free_space(entry_count) %d", + ih->ih_item_len, ih->ih_item_location, ih_free_space (ih)); + } else + sprintf (buf, "[NULL]"); +} + + +static void sprintf_direntry (char * buf, struct reiserfs_dir_entry * de) +{ + char name[20]; + + memcpy (name, de->de_name, de->de_namelen > 19 ? 19 : de->de_namelen); + name [de->de_namelen > 19 ? 19 : de->de_namelen] = 0; + sprintf (buf, "\"%s\"==>[%d %d]", name, de->de_dir_id, de->de_objectid); +} + + +static void sprintf_block_head (char * buf, struct buffer_head * bh) +{ + sprintf (buf, "level=%d, nr_items=%d, free_space=%d rdkey ", + B_LEVEL (bh), B_NR_ITEMS (bh), B_FREE_SPACE (bh)); +#if 0 + if (B_LEVEL (bh) == DISK_LEAF_NODE_LEVEL) + sprintf_le_key (buf + strlen (buf), B_PRIGHT_DELIM_KEY (bh)); +#endif +} + + +static void sprintf_buffer_head (char * buf, struct buffer_head * bh) +{ + sprintf (buf, "dev %s, size %d, blocknr %ld, count %d, list %d, state 0x%lx, page %p, (%s, %s, %s)", + kdevname (bh->b_dev), bh->b_size, bh->b_blocknr, atomic_read (&(bh->b_count)), bh->b_list, + bh->b_state, bh->b_page, + buffer_uptodate (bh) ? "UPTODATE" : "!UPTODATE", + buffer_dirty (bh) ? "DIRTY" : "CLEAN", + buffer_locked (bh) ? "LOCKED" : "UNLOCKED"); +} + + +static void sprintf_disk_child (char * buf, struct disk_child * dc) +{ + sprintf (buf, "[dc_number=%d, dc_size=%u]", dc->dc_block_number, dc->dc_size); +} + + +static char * is_there_reiserfs_struct (char * fmt, int * what, int * skip) +{ + char * k = fmt; + + *skip = 0; + + while (1) { + k = strstr (k, "%"); + if (!k) + break; + if (k && (k[1] == 'k' || k[1] == 'K' || k[1] == 'h' || k[1] == 't' || + k[1] == 'z' || k[1] == 'b' || k[1] == 'y')) { + *what = k[1]; + break; + } + (*skip) ++; + k ++; + } + return k; +} + + +/* debugging reiserfs we used to print out a lot of different + variables, like keys, item headers, buffer heads etc. Values of + most fields matter. So it took a long time just to write + appropriative printk. With this reiserfs_warning you can use format + specification for complex structures like you used to do with + printfs for integers, doubles and pointers. For instance, to print + out key structure you have to write just: + reiserfs_warning ("bad key %k", key); + instead of + printk ("bad key %lu %lu %lu %lu", key->k_dir_id, key->k_objectid, + key->k_offset, key->k_uniqueness); +*/ + +#define do_reiserfs_warning \ +{\ + char * fmt1 = fmt_buf;\ + va_list args;\ + int i, j;\ + char * k;\ + char * p = error_buf;\ + int what, skip;\ +\ + strcpy (fmt1, fmt);\ + va_start(args, fmt);\ +\ + while (1) {\ + k = is_there_reiserfs_struct (fmt1, &what, &skip);\ + if (k != 0) {\ + *k = 0;\ + p += vsprintf (p, fmt1, args);\ +\ + for (i = 0; i < skip; i ++)\ + j = va_arg (args, int);\ +\ + switch (what) {\ + case 'k':\ + sprintf_le_key (p, va_arg(args, struct key *));\ + break;\ + case 'K':\ + sprintf_cpu_key (p, va_arg(args, struct cpu_key *));\ + break;\ + case 'h':\ + sprintf_item_head (p, va_arg(args, struct item_head *));\ + break;\ + case 't':\ + sprintf_direntry (p, va_arg(args, struct reiserfs_dir_entry *));\ + break;\ + case 'y':\ + sprintf_disk_child (p, va_arg(args, struct disk_child *));\ + break;\ + case 'z':\ + sprintf_block_head (p, va_arg(args, struct buffer_head *));\ + break;\ + case 'b':\ + sprintf_buffer_head (p, va_arg(args, struct buffer_head *));\ + break;\ + }\ + p += strlen (p);\ + fmt1 = k + 2;\ + } else {\ + i = vsprintf (p, fmt1, args);\ + break;\ + }\ + }\ +\ + va_end(args);\ +} + + +/* in addition to usual conversion specifiers this accepts reiserfs + specific conversion specifiers: + %k to print little endian key, + %K to print cpu key, + %h to print item_head, + %t to print directory entry + %z to print block head (arg must be struct buffer_head * + %b to print buffer_head +*/ +void reiserfs_warning (const char * fmt, ...) +{ + do_reiserfs_warning; + /* console_print (error_buf); */ + printk ("%s", error_buf); +} + +void reiserfs_debug (struct super_block *s, int level, const char * fmt, ...) +{ +#ifdef CONFIG_REISERFS_CHECK + do_reiserfs_warning; + printk ("%s", error_buf); +#else + ; +#endif +} + +/* The format: + + maintainer-errorid: [function-name:] message + + where errorid is unique to the maintainer and function-name is + optional, is recommended, so that anyone can easily find the bug + with a simple grep for the short to type string + maintainer-errorid. Don't bother with reusing errorids, there are + lots of numbers out there. + + Example: + + reiserfs_panic( + p_sb, "reiser-29: reiserfs_new_blocknrs: " + "one of search_start or rn(%d) is equal to MAX_B_NUM," + "which means that we are optimizing location based on the bogus location of a temp buffer (%p).", + rn, bh + ); + + Regular panic()s sometimes clear the screen before the message can + be read, thus the need for the while loop. + + Numbering scheme for panic used by Vladimir and Anatoly( Hans completely ignores this scheme, and considers it + pointless complexity): + + panics in reiserfs_fs.h have numbers from 1000 to 1999 + super.c 2000 to 2999 + preserve.c 3000 to 3999 + bitmap.c 4000 to 4999 + stree.c 5000 to 5999 + prints.c 6000 to 6999 + namei.c 7000 to 7999 + fix_nodes.c 8000 to 8999 + dir.c 9000 to 9999 + lbalance.c 10000 to 10999 + ibalance.c 11000 to 11999 not ready + do_balan.c 12000 to 12999 + inode.c 13000 to 13999 + file.c 14000 to 14999 + objectid.c 15000 - 15999 + buffer.c 16000 - 16999 + symlink.c 17000 - 17999 + + . */ + + +#ifdef CONFIG_REISERFS_CHECK +extern struct tree_balance * cur_tb; +#endif + +void reiserfs_panic (struct super_block * sb, const char * fmt, ...) +{ +#ifdef __KERNEL__ + show_reiserfs_locks() ; +#endif + do_reiserfs_warning; + printk ("%s", error_buf); + BUG (); + // console_print (error_buf); + // for (;;); + +#ifdef __KERNEL__ + + /* comment before release */ + //for (;;); + +#if 0 /* this is not needed, the state is ignored */ + if (sb && !(sb->s_flags & MS_RDONLY)) { + sb->u.reiserfs_sb.s_mount_state |= REISERFS_ERROR_FS; + sb->u.reiserfs_sb.s_rs->s_state = REISERFS_ERROR_FS; + + mark_buffer_dirty(sb->u.reiserfs_sb.s_sbh) ; + sb->s_dirt = 1; + } +#endif + + /* this is to prevent panic from syncing this filesystem */ + if (sb && sb->s_lock) + sb->s_lock=0; + if (sb) + sb->s_flags |= MS_RDONLY; + + panic ("REISERFS: panic (device %s): %s\n", + sb ? kdevname(sb->s_dev) : "sb == 0", error_buf); +#else + exit (0); +#endif +} + + +void print_virtual_node (struct virtual_node * vn) +{ + int i; + struct virtual_item * vi; + + printk ("VIRTUAL NODE CONTAINS %d items, has size %d,%s,%s, ITEM_POS=%d POS_IN_ITEM=%d MODE=\'%c\'\n", + vn->vn_nr_item, vn->vn_size, + (vn->vn_vi[0].vi_type & VI_TYPE_LEFT_MERGEABLE )? "left mergeable" : "", + (vn->vn_vi[vn->vn_nr_item - 1].vi_type & VI_TYPE_RIGHT_MERGEABLE) ? "right mergeable" : "", + vn->vn_affected_item_num, vn->vn_pos_in_item, vn->vn_mode); + + vi = vn->vn_vi; + for (i = 0; i < vn->vn_nr_item; i ++, vi ++) + op_print_vi (vi); + +} + + +void print_path (struct tree_balance * tb, struct path * path) +{ + int h = 0; + struct buffer_head * bh; + + if (tb) { + while (tb->insert_size[h]) { + bh = PATH_H_PBUFFER (path, h); + printk ("block %lu (level=%d), position %d\n", bh ? bh->b_blocknr : 0, + bh ? B_LEVEL (bh) : 0, PATH_H_POSITION (path, h)); + h ++; + } + } else { + int offset = path->path_length; + struct buffer_head * bh; + printk ("Offset Bh (b_blocknr, b_count) Position Nr_item\n"); + while ( offset > ILLEGAL_PATH_ELEMENT_OFFSET ) { + bh = PATH_OFFSET_PBUFFER (path, offset); + printk ("%6d %10p (%9lu, %7d) %8d %7d\n", offset, + bh, bh ? bh->b_blocknr : 0, bh ? atomic_read (&(bh->b_count)) : 0, + PATH_OFFSET_POSITION (path, offset), bh ? B_NR_ITEMS (bh) : -1); + + offset --; + } + } + +} + + +/* this prints internal nodes (4 keys/items in line) (dc_number, + dc_size)[k_dirid, k_objectid, k_offset, k_uniqueness](dc_number, + dc_size)...*/ +static int print_internal (struct buffer_head * bh, int first, int last) +{ + struct key * key; + struct disk_child * dc; + int i; + int from, to; + + if (!B_IS_KEYS_LEVEL (bh)) + return 1; + + check_internal (bh); + + if (first == -1) { + from = 0; + to = B_NR_ITEMS (bh); + } else { + from = first; + to = last < B_NR_ITEMS (bh) ? last : B_NR_ITEMS (bh); + } + + reiserfs_warning ("INTERNAL NODE (%ld) contains %z\n", bh->b_blocknr, bh); + + dc = B_N_CHILD (bh, from); + reiserfs_warning ("PTR %d: %y ", from, dc); + + for (i = from, key = B_N_PDELIM_KEY (bh, from), dc ++; i < to; i ++, key ++, dc ++) { + reiserfs_warning ("KEY %d: %k PTR %d: %y ", i, key, i + 1, dc); + if (i && i % 4 == 0) + printk ("\n"); + } + printk ("\n"); + return 0; +} + + + + + +static int print_leaf (struct buffer_head * bh, int print_mode, int first, int last) +{ + struct block_head * blkh; + struct item_head * ih; + int i; + int from, to; + + if (!B_IS_ITEMS_LEVEL (bh)) + return 1; + + check_leaf (bh); + + blkh = B_BLK_HEAD (bh); + ih = B_N_PITEM_HEAD (bh,0); + + printk ("\n===================================================================\n"); + reiserfs_warning ("LEAF NODE (%ld) contains %z\n", bh->b_blocknr, bh); + + if (!(print_mode & PRINT_LEAF_ITEMS)) { + reiserfs_warning ("FIRST ITEM_KEY: %k, LAST ITEM KEY: %k\n", + &(ih->ih_key), &((ih + le16_to_cpu (blkh->blk_nr_item) - 1)->ih_key)); + return 0; + } + + if (first < 0 || first > le16_to_cpu (blkh->blk_nr_item) - 1) + from = 0; + else + from = first; + + if (last < 0 || last > le16_to_cpu (blkh->blk_nr_item)) + to = le16_to_cpu (blkh->blk_nr_item); + else + to = last; + + ih += from; + printk ("-------------------------------------------------------------------------------\n"); + printk ("|##| type | key | ilen | free_space | version | loc |\n"); + for (i = from; i < to; i++, ih ++) { + printk ("-------------------------------------------------------------------------------\n"); + reiserfs_warning ("|%2d| %h |\n", i, ih); + if (print_mode & PRINT_LEAF_ITEMS) + op_print_item (ih, B_I_PITEM (bh, ih)); + } + + printk ("===================================================================\n"); + + return 0; +} + +static char * reiserfs_version (char * buf) +{ + __u16 * pversion; + + pversion = (__u16 *)(buf) + 36; + if (*pversion == 0) + return "0"; + if (*pversion == 2) + return "2"; + return "Unknown"; +} + + +/* return 1 if this is not super block */ +static int print_super_block (struct buffer_head * bh) +{ + struct reiserfs_super_block * rs = (struct reiserfs_super_block *)(bh->b_data); + int skipped, data_blocks; + + + if (strncmp (rs->s_magic, REISERFS_SUPER_MAGIC_STRING, strlen ( REISERFS_SUPER_MAGIC_STRING)) && + strncmp (rs->s_magic, REISER2FS_SUPER_MAGIC_STRING, strlen ( REISER2FS_SUPER_MAGIC_STRING))) + return 1; + + printk ("%s\'s super block in block %ld\n======================\n", kdevname (bh->b_dev), bh->b_blocknr); + printk ("Reiserfs version %s\n", reiserfs_version (bh->b_data)); + printk ("Block count %u\n", le32_to_cpu (rs->s_block_count)); + printk ("Blocksize %d\n", le16_to_cpu (rs->s_blocksize)); + printk ("Free blocks %u\n", le32_to_cpu (rs->s_free_blocks)); + skipped = bh->b_blocknr; // FIXME: this would be confusing if + // someone stores reiserfs super block in some data block ;) + data_blocks = le32_to_cpu (rs->s_block_count) - skipped - 1 - + le16_to_cpu (rs->s_bmap_nr) - (le32_to_cpu (rs->s_orig_journal_size) + 1) - + le32_to_cpu (rs->s_free_blocks); + printk ("Busy blocks (skipped %d, bitmaps - %d, journal blocks - %d\n" + "1 super blocks, %d data blocks\n", + skipped, le16_to_cpu (rs->s_bmap_nr), + (le32_to_cpu (rs->s_orig_journal_size) + 1), data_blocks); + printk ("Root block %u\n", le32_to_cpu (rs->s_root_block)); + printk ("Journal block (first) %d\n", le32_to_cpu (rs->s_journal_block)); + printk ("Journal dev %d\n", le32_to_cpu (rs->s_journal_dev)); + printk ("Journal orig size %d\n", le32_to_cpu (rs->s_orig_journal_size)); + printk ("Filesystem state %s\n", + (le16_to_cpu (rs->s_state) == REISERFS_VALID_FS) ? "VALID" : "ERROR"); + printk ("Hash function \"%s\"\n", le16_to_cpu (rs->s_hash_function_code) == TEA_HASH ? "tea" : + ((le16_to_cpu (rs->s_hash_function_code) == YURA_HASH) ? "rupasov" : "unknown")); + +#if 0 + __u32 s_journal_trans_max ; /* max number of blocks in a transaction. */ + __u32 s_journal_block_count ; /* total size of the journal. can change over time */ + __u32 s_journal_max_batch ; /* max number of blocks to batch into a trans */ + __u32 s_journal_max_commit_age ; /* in seconds, how old can an async commit be */ + __u32 s_journal_max_trans_age ; /* in seconds, how old can a transaction be */ +#endif + printk ("Tree height %d\n", rs->s_tree_height); + return 0; +} + + +static int print_desc_block (struct buffer_head * bh) +{ + struct reiserfs_journal_desc * desc; + + desc = (struct reiserfs_journal_desc *)(bh->b_data); + if (memcmp(desc->j_magic, JOURNAL_DESC_MAGIC, 8)) + return 1; + + printk ("Desc block %lu (j_trans_id %d, j_mount_id %d, j_len %d)", + bh->b_blocknr, desc->j_trans_id, desc->j_mount_id, desc->j_len); + + return 0; +} + + +void print_block (struct buffer_head * bh, ...)//int print_mode, int first, int last) +{ + va_list args; + int mode, first, last; + + va_start (args, bh); + + if ( ! bh ) { + printk("print_block: buffer is NULL\n"); + return; + } + + mode = va_arg (args, int); + first = va_arg (args, int); + last = va_arg (args, int); + if (print_leaf (bh, mode, first, last)) + if (print_internal (bh, first, last)) + if (print_super_block (bh)) + if (print_desc_block (bh)) + printk ("Block %ld contains unformatted data\n", bh->b_blocknr); +} + + + +char print_tb_buf[2048]; + +/* this stores initial state of tree balance in the print_tb_buf */ +void store_print_tb (struct tree_balance * tb) +{ + int h = 0; + int i; + struct buffer_head * tbSh, * tbFh; + + if (!tb) + return; + + sprintf (print_tb_buf, "\n" + "BALANCING %d\n" + "MODE=%c, ITEM_POS=%d POS_IN_ITEM=%d\n" + "=====================================================================\n" + "* h * S * L * R * F * FL * FR * CFL * CFR *\n", + tb->tb_sb->u.reiserfs_sb.s_do_balance, + tb->tb_mode, PATH_LAST_POSITION (tb->tb_path), tb->tb_path->pos_in_item); + + for (h = 0; h < sizeof(tb->insert_size) / sizeof (tb->insert_size[0]); h ++) { + if (PATH_H_PATH_OFFSET (tb->tb_path, h) <= tb->tb_path->path_length && + PATH_H_PATH_OFFSET (tb->tb_path, h) > ILLEGAL_PATH_ELEMENT_OFFSET) { + tbSh = PATH_H_PBUFFER (tb->tb_path, h); + tbFh = PATH_H_PPARENT (tb->tb_path, h); + } else { + tbSh = 0; + tbFh = 0; + } + sprintf (print_tb_buf + strlen (print_tb_buf), + "* %d * %3ld(%2d) * %3ld(%2d) * %3ld(%2d) * %5ld * %5ld * %5ld * %5ld * %5ld *\n", + h, + (tbSh) ? (tbSh->b_blocknr):(-1), + (tbSh) ? atomic_read (&(tbSh->b_count)) : -1, + (tb->L[h]) ? (tb->L[h]->b_blocknr):(-1), + (tb->L[h]) ? atomic_read (&(tb->L[h]->b_count)) : -1, + (tb->R[h]) ? (tb->R[h]->b_blocknr):(-1), + (tb->R[h]) ? atomic_read (&(tb->R[h]->b_count)) : -1, + (tbFh) ? (tbFh->b_blocknr):(-1), + (tb->FL[h]) ? (tb->FL[h]->b_blocknr):(-1), + (tb->FR[h]) ? (tb->FR[h]->b_blocknr):(-1), + (tb->CFL[h]) ? (tb->CFL[h]->b_blocknr):(-1), + (tb->CFR[h]) ? (tb->CFR[h]->b_blocknr):(-1)); + } + + sprintf (print_tb_buf + strlen (print_tb_buf), + "=====================================================================\n" + "* h * size * ln * lb * rn * rb * blkn * s0 * s1 * s1b * s2 * s2b * curb * lk * rk *\n" + "* 0 * %4d * %2d * %2d * %2d * %2d * %4d * %2d * %2d * %3d * %2d * %3d * %4d * %2d * %2d *\n", + tb->insert_size[0], tb->lnum[0], tb->lbytes, tb->rnum[0],tb->rbytes, tb->blknum[0], + tb->s0num, tb->s1num,tb->s1bytes, tb->s2num, tb->s2bytes, tb->cur_blknum, tb->lkey[0], tb->rkey[0]); + + /* this prints balance parameters for non-leaf levels */ + h = 0; + do { + h++; + sprintf (print_tb_buf + strlen (print_tb_buf), + "* %d * %4d * %2d * * %2d * * %2d *\n", + h, tb->insert_size[h], tb->lnum[h], tb->rnum[h], tb->blknum[h]); + } while (tb->insert_size[h]); + + sprintf (print_tb_buf + strlen (print_tb_buf), + "=====================================================================\n" + "FEB list: "); + + /* print FEB list (list of buffers in form (bh (b_blocknr, b_count), that will be used for new nodes) */ + h = 0; + for (i = 0; i < sizeof (tb->FEB) / sizeof (tb->FEB[0]); i ++) + sprintf (print_tb_buf + strlen (print_tb_buf), + "%p (%lu %d)%s", tb->FEB[i], tb->FEB[i] ? tb->FEB[i]->b_blocknr : 0, + tb->FEB[i] ? atomic_read (&(tb->FEB[i]->b_count)) : 0, + (i == sizeof (tb->FEB) / sizeof (tb->FEB[0]) - 1) ? "\n" : ", "); + + sprintf (print_tb_buf + strlen (print_tb_buf), + "======================== the end ====================================\n"); +} + +void print_cur_tb (char * mes) +{ + printk ("%s\n%s", mes, print_tb_buf); +} + + +#ifndef __KERNEL__ + +void print_bmap_block (int i, char * data, int size, int silent) +{ + int j, k; + int bits = size * 8; + int zeros = 0, ones = 0; + + + if (test_bit (0, data)) { + /* first block addressed by this bitmap block is used */ + ones ++; + if (!silent) + printf ("Busy (%d-", i * bits); + for (j = 1; j < bits; j ++) { + while (test_bit (j, data)) { + ones ++; + if (j == bits - 1) { + if (!silent) + printf ("%d)\n", j + i * bits); + goto end; + } + j++; + } + if (!silent) + printf ("%d) Free(%d-", j - 1 + i * bits, j + i * bits); + + while (!test_bit (j, data)) { + zeros ++; + if (j == bits - 1) { + if (!silent) + printf ("%d)\n", j + i * bits); + goto end; + } + j++; + } + if (!silent) + printf ("%d) Busy(%d-", j - 1 + i * bits, j + i * bits); + + j --; + end: + } + } else { + /* first block addressed by this bitmap is free */ + zeros ++; + if (!silent) + printf ("Free (%d-", i * bits); + for (j = 1; j < bits; j ++) { + k = 0; + while (!test_bit (j, data)) { + k ++; + if (j == bits - 1) { + if (!silent) + printf ("%d)\n", j + i * bits); + zeros += k; + goto end2; + } + j++; + } + zeros += k; + if (!silent) + printf ("%d) Busy(%d-", j - 1 + i * bits, j + i * bits); + + k = 0; + while (test_bit (j, data)) { + ones ++; + if (j == bits - 1) { + if (!silent) + printf ("%d)\n", j + i * bits); + ones += k; + goto end2; + } + j++; + } + ones += k; + if (!silent) + printf ("%d) Busy(%d-", j - 1 + i * bits, j + i * bits); + + j --; + end2: + } + } + + printf ("used %d, free %d\n", ones, zeros); +} + + +/* if silent == 1, do not print details */ +void print_bmap (struct super_block * s, int silent) +{ + int bmapnr = SB_BMAP_NR (s); + int i; + + printf ("Bitmap blocks are:\n"); + for (i = 0; i < bmapnr; i ++) { + printf ("#%d: block %lu: ", i, SB_AP_BITMAP(s)[i]->b_blocknr); + print_bmap_block (i, SB_AP_BITMAP(s)[i]->b_data, s->s_blocksize, silent); + } + +} + + + + +void print_objectid_map (struct super_block * s) +{ + int i; + struct reiserfs_super_block * rs; + unsigned long * omap; + + rs = SB_DISK_SUPER_BLOCK (s); + omap = (unsigned long *)(rs + 1); + printk ("Map of objectids\n"); + + for (i = 0; i < rs->s_oid_cursize; i ++) { + if (i % 2 == 0) + printk ("busy(%lu-%lu) ", omap[i], omap[i+1] - 1); + else + printk ("free(%lu-%lu) ", + omap[i], ((i+1) == rs->s_oid_cursize) ? -1 : omap[i+1] - 1); + } + printk ("\n"); + + printk ("Object id array has size %d (max %d):", rs->s_oid_cursize, + rs->s_oid_maxsize); + + for (i = 0; i < rs->s_oid_cursize; i ++) + printk ("%lu ", omap[i]); + printk ("\n"); + +} + +#endif /* #ifndef __KERNEL__ */ + + +static void check_leaf_block_head (struct buffer_head * bh) +{ + struct block_head * blkh; + + blkh = B_BLK_HEAD (bh); + if (le16_to_cpu (blkh->blk_nr_item) > (bh->b_size - BLKH_SIZE) / IH_SIZE) + reiserfs_panic (0, "vs-6010: check_leaf_block_head: invalid item number %z", bh); + if (le16_to_cpu (blkh->blk_free_space) > + bh->b_size - BLKH_SIZE - IH_SIZE * le16_to_cpu (blkh->blk_nr_item)) + reiserfs_panic (0, "vs-6020: check_leaf_block_head: invalid free space %z", bh); + +} + +static void check_internal_block_head (struct buffer_head * bh) +{ + struct block_head * blkh; + + blkh = B_BLK_HEAD (bh); + if (!(B_LEVEL (bh) > DISK_LEAF_NODE_LEVEL && B_LEVEL (bh) <= MAX_HEIGHT)) + reiserfs_panic (0, "vs-6025: check_internal_block_head: invalid level %z", bh); + + if (B_NR_ITEMS (bh) > (bh->b_size - BLKH_SIZE) / IH_SIZE) + reiserfs_panic (0, "vs-6030: check_internal_block_head: invalid item number %z", bh); + + if (B_FREE_SPACE (bh) != + bh->b_size - BLKH_SIZE - KEY_SIZE * B_NR_ITEMS (bh) - DC_SIZE * (B_NR_ITEMS (bh) + 1)) + reiserfs_panic (0, "vs-6040: check_internal_block_head: invalid free space %z", bh); + +} + + +void check_leaf (struct buffer_head * bh) +{ + int i; + struct item_head * ih; + + if (!bh) + return; + check_leaf_block_head (bh); + for (i = 0, ih = B_N_PITEM_HEAD (bh, 0); i < B_NR_ITEMS (bh); i ++, ih ++) + op_check_item (ih, B_I_PITEM (bh, ih)); +} + + +void check_internal (struct buffer_head * bh) +{ + if (!bh) + return; + check_internal_block_head (bh); +} + + +void print_statistics (struct super_block * s) +{ + + /* + printk ("reiserfs_put_super: session statistics: balances %d, fix_nodes %d, preserve list freeings %d, \ +bmap with search %d, without %d, dir2ind %d, ind2dir %d\n", + s->u.reiserfs_sb.s_do_balance, s->u.reiserfs_sb.s_fix_nodes, s->u.reiserfs_sb.s_preserve_list_freeings, + s->u.reiserfs_sb.s_bmaps, s->u.reiserfs_sb.s_bmaps_without_search, + s->u.reiserfs_sb.s_direct2indirect, s->u.reiserfs_sb.s_indirect2direct); + */ + +} diff --git a/fs/reiserfs/resize.c b/fs/reiserfs/resize.c new file mode 100644 index 000000000..87ebd6ee6 --- /dev/null +++ b/fs/reiserfs/resize.c @@ -0,0 +1,168 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +/* + * Written by Alexander Zarochentcev. + * + * The kernel part of the (on-line) reiserfs resizer. + */ + +#ifdef __KERNEL__ + +#include <linux/kernel.h> +#include <linux/vmalloc.h> +#include <linux/locks.h> +#include <linux/string.h> +#include <linux/errno.h> +#include <linux/reiserfs_fs.h> +#include <linux/reiserfs_fs_sb.h> + +#else + +#include "nokernel.h" + +#endif + +int reiserfs_resize (struct super_block * s, unsigned long block_count_new) +{ + struct reiserfs_super_block * sb; + struct buffer_head ** bitmap, * bh; + struct reiserfs_transaction_handle th; + unsigned int bmap_nr_new, bmap_nr; + unsigned int block_r_new, block_r; + + struct reiserfs_list_bitmap * jb; + struct reiserfs_list_bitmap jbitmap[JOURNAL_NUM_BITMAPS]; + + unsigned long int block_count, free_blocks; + int i; + int copy_size ; + + sb = SB_DISK_SUPER_BLOCK(s); + + if (SB_BLOCK_COUNT(s) >= block_count_new) { + printk("can\'t shrink filesystem on-line\n"); + return -EINVAL; + } + + /* check the device size */ + bh = bread(s->s_dev, block_count_new - 1, s->s_blocksize); + if (!bh) { + printk("reiserfs_resize: can\'t read last block\n"); + return -EINVAL; + } + bforget(bh); + + /* old disk layout detection; those partitions can be mounted, but + * cannot be resized */ + if (SB_BUFFER_WITH_SB(s)->b_blocknr * SB_BUFFER_WITH_SB(s)->b_size + != REISERFS_DISK_OFFSET_IN_BYTES ) { + printk("reiserfs_resize: unable to resize a reiserfs without distributed bitmap (fs version < 3.5.12)\n"); + return -ENOTSUPP; + } + + /* count used bits in last bitmap block */ + block_r = SB_BLOCK_COUNT(s) - + (SB_BMAP_NR(s) - 1) * s->s_blocksize * 8; + + /* count bitmap blocks in new fs */ + bmap_nr_new = block_count_new / ( s->s_blocksize * 8 ); + block_r_new = block_count_new - bmap_nr_new * s->s_blocksize * 8; + if (block_r_new) + bmap_nr_new++; + else + block_r_new = s->s_blocksize * 8; + + /* save old values */ + block_count = SB_BLOCK_COUNT(s); + bmap_nr = SB_BMAP_NR(s); + + /* resizing of reiserfs bitmaps (journal and real), if needed */ + if (bmap_nr_new > bmap_nr) { + /* reallocate journal bitmaps */ + if (reiserfs_allocate_list_bitmaps(s, jbitmap, bmap_nr_new) < 0) { + printk("reiserfs_resize: unable to allocate memory for journal bitmaps\n"); + unlock_super(s) ; + return -ENOMEM ; + } + /* the new journal bitmaps are zero filled, now we copy in the bitmap + ** node pointers from the old journal bitmap structs, and then + ** transfer the new data structures into the journal struct. + ** + ** using the copy_size var below allows this code to work for + ** both shrinking and expanding the FS. + */ + copy_size = bmap_nr_new < bmap_nr ? bmap_nr_new : bmap_nr ; + copy_size = copy_size * sizeof(struct reiserfs_list_bitmap_node *) ; + for (i = 0 ; i < JOURNAL_NUM_BITMAPS ; i++) { + struct reiserfs_bitmap_node **node_tmp ; + jb = SB_JOURNAL(s)->j_list_bitmap + i ; + memcpy(jbitmap[i].bitmaps, jb->bitmaps, copy_size) ; + + /* just in case vfree schedules on us, copy the new + ** pointer into the journal struct before freeing the + ** old one + */ + node_tmp = jb->bitmaps ; + jb->bitmaps = jbitmap[i].bitmaps ; + vfree(node_tmp) ; + } + + /* allocate additional bitmap blocks, reallocate array of bitmap + * block pointers */ + bitmap = reiserfs_kmalloc(sizeof(struct buffer_head *) * bmap_nr_new, GFP_KERNEL, s); + if (!bitmap) { + printk("reiserfs_resize: unable to allocate memory.\n"); + return -ENOMEM; + } + for (i = 0; i < bmap_nr; i++) + bitmap[i] = SB_AP_BITMAP(s)[i]; + for (i = bmap_nr; i < bmap_nr_new; i++) { + bitmap[i] = reiserfs_getblk(s->s_dev, i * s->s_blocksize * 8, s->s_blocksize); + memset(bitmap[i]->b_data, 0, sb->s_blocksize); + reiserfs_test_and_set_le_bit(0, bitmap[i]->b_data); + + mark_buffer_dirty(bitmap[i]) ; + mark_buffer_uptodate(bitmap[i], 1); + ll_rw_block(WRITE, 1, bitmap + i); + wait_on_buffer(bitmap[i]); + } + /* free old bitmap blocks array */ + reiserfs_kfree(SB_AP_BITMAP(s), + sizeof(struct buffer_head *) * bmap_nr, s); + SB_AP_BITMAP(s) = bitmap; + } + + /* begin transaction */ + journal_begin(&th, s, 10); + + /* correct last bitmap blocks in old and new disk layout */ + reiserfs_prepare_for_journal(s, SB_AP_BITMAP(s)[bmap_nr - 1], 1); + for (i = block_r; i < s->s_blocksize * 8; i++) + reiserfs_test_and_clear_le_bit(i, + SB_AP_BITMAP(s)[bmap_nr - 1]->b_data); + journal_mark_dirty(&th, s, SB_AP_BITMAP(s)[bmap_nr - 1]); + + reiserfs_prepare_for_journal(s, SB_AP_BITMAP(s)[bmap_nr_new - 1], 1); + for (i = block_r_new; i < s->s_blocksize * 8; i++) + reiserfs_test_and_set_le_bit(i, + SB_AP_BITMAP(s)[bmap_nr_new - 1]->b_data); + journal_mark_dirty(&th, s, SB_AP_BITMAP(s)[bmap_nr_new - 1]); + + /* update super */ + reiserfs_prepare_for_journal(s, SB_BUFFER_WITH_SB(s), 1) ; + free_blocks = SB_FREE_BLOCKS(s); + PUT_SB_FREE_BLOCKS(s, free_blocks + (block_count_new - block_count - (bmap_nr_new - bmap_nr))); + PUT_SB_BLOCK_COUNT(s, block_count_new); + PUT_SB_BMAP_NR(s, bmap_nr_new); + s->s_dirt = 1; + + journal_mark_dirty(&th, s, SB_BUFFER_WITH_SB(s)); + + SB_JOURNAL(s)->j_must_wait = 1; + journal_end(&th, s, 10); + + return 0; +} + diff --git a/fs/reiserfs/stree.c b/fs/reiserfs/stree.c new file mode 100644 index 000000000..f4b24b590 --- /dev/null +++ b/fs/reiserfs/stree.c @@ -0,0 +1,2078 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +/* + * Written by Anatoly P. Pinchuk pap@namesys.botik.ru + * Programm System Institute + * Pereslavl-Zalessky Russia + */ + +/* + * This file contains functions dealing with S+tree + * + * B_IS_IN_TREE + * copy_short_key + * copy_item_head + * comp_short_keys + * comp_keys + * comp_cpu_keys + * comp_short_le_keys + * comp_short_cpu_keys + * cpu_key2cpu_key + * le_key2cpu_key + * comp_le_keys + * bin_search + * get_lkey + * get_rkey + * key_in_buffer + * decrement_bcount + * decrement_counters_in_path + * reiserfs_check_path + * pathrelse_and_restore + * pathrelse + * search_by_key_reada + * search_by_key + * search_for_position_by_key + * comp_items + * prepare_for_direct_item + * prepare_for_direntry_item + * prepare_for_delete_or_cut + * calc_deleted_bytes_number + * init_tb_struct + * padd_item + * reiserfs_delete_item + * reiserfs_delete_solid_item + * reiserfs_delete_object + * maybe_indirect_to_direct + * indirect_to_direct_roll_back + * reiserfs_cut_from_item + * truncate_directory + * reiserfs_do_truncate + * reiserfs_paste_into_item + * reiserfs_insert_item + */ +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/sched.h> +#include <linux/string.h> +#include <linux/locks.h> +#include <linux/pagemap.h> +#include <linux/reiserfs_fs.h> +#include <linux/smp_lock.h> + +#else + +#include "nokernel.h" + +#endif + + + +/* Does the buffer contain a disk block which is in the tree. */ +inline int B_IS_IN_TREE (struct buffer_head * p_s_bh) +{ + +#ifdef CONFIG_REISERFS_CHECK + + if ( B_LEVEL (p_s_bh) > MAX_HEIGHT ) { + reiserfs_panic(0, "PAP-1010: B_IS_IN_TREE: block (%b) has too big level (%z)", + p_s_bh, p_s_bh); + } +#endif + + return ( B_LEVEL (p_s_bh) != FREE_LEVEL ); +} + + + + +inline void copy_short_key (void * to, void * from) +{ + memcpy (to, from, SHORT_KEY_SIZE); +} + +// +// to gets item head in le form +// +inline void copy_item_head(void * p_v_to, void * p_v_from) +{ + memcpy (p_v_to, p_v_from, IH_SIZE); +} + + +/* k1 is pointer to on-disk structure which is stored in little-endian + form. k2 is pointer to cpu variable. For key of items of the same + object this returns 0. + Returns: -1 if key1 < key2 + 0 if key1 == key2 + 1 if key1 > key2 */ +inline int comp_short_keys (struct key * le_key, struct cpu_key * cpu_key) +{ + __u32 * p_s_le_u32, * p_s_cpu_u32; + int n_key_length = REISERFS_SHORT_KEY_LEN; + + p_s_le_u32 = (__u32 *)le_key; + p_s_cpu_u32 = (__u32 *)cpu_key; + for( ; n_key_length--; ++p_s_le_u32, ++p_s_cpu_u32 ) { + if ( le32_to_cpu (*p_s_le_u32) < *p_s_cpu_u32 ) + return -1; + if ( le32_to_cpu (*p_s_le_u32) > *p_s_cpu_u32 ) + return 1; + } + + return 0; +} + + +/* k1 is pointer to on-disk structure which is stored in little-endian + form. k2 is pointer to cpu variable. + Compare keys using all 4 key fields. + Returns: -1 if key1 < key2 0 + if key1 = key2 1 if key1 > key2 */ +inline int comp_keys (struct key * le_key, struct cpu_key * cpu_key) +{ + int retval; + + retval = comp_short_keys (le_key, cpu_key); + if (retval) + return retval; + if (le_key_k_offset (cpu_key->version, le_key) < cpu_key_k_offset (cpu_key)) + return -1; + if (le_key_k_offset (cpu_key->version, le_key) > cpu_key_k_offset (cpu_key)) + return 1; + + if (cpu_key->key_length == 3) + return 0; + + /* this part is needed only when tail conversion is in progress */ + if (le_key_k_type (cpu_key->version, le_key) < cpu_key_k_type (cpu_key)) + return -1; + + if (le_key_k_type (cpu_key->version, le_key) > cpu_key_k_type (cpu_key)) + return 1; + + return 0; +} + + +// +// FIXME: not used yet +// +inline int comp_cpu_keys (struct cpu_key * key1, struct cpu_key * key2) +{ + if (key1->on_disk_key.k_dir_id < key2->on_disk_key.k_dir_id) + return -1; + if (key1->on_disk_key.k_dir_id > key2->on_disk_key.k_dir_id) + return 1; + + if (key1->on_disk_key.k_objectid < key2->on_disk_key.k_objectid) + return -1; + if (key1->on_disk_key.k_objectid > key2->on_disk_key.k_objectid) + return 1; + + if (cpu_key_k_offset (key1) < cpu_key_k_offset (key2)) + return -1; + if (cpu_key_k_offset (key1) > cpu_key_k_offset (key2)) + return 1; + + reiserfs_warning ("comp_cpu_keys: type are compared for %k and %k\n", + key1, key2); + + if (cpu_key_k_type (key1) < cpu_key_k_type (key2)) + return -1; + if (cpu_key_k_type (key1) > cpu_key_k_type (key2)) + return 1; + return 0; +} + +inline int comp_short_le_keys (struct key * key1, struct key * key2) +{ + __u32 * p_s_1_u32, * p_s_2_u32; + int n_key_length = REISERFS_SHORT_KEY_LEN; + + p_s_1_u32 = (__u32 *)key1; + p_s_2_u32 = (__u32 *)key2; + for( ; n_key_length--; ++p_s_1_u32, ++p_s_2_u32 ) { + if ( le32_to_cpu (*p_s_1_u32) < le32_to_cpu (*p_s_2_u32) ) + return -1; + if ( le32_to_cpu (*p_s_1_u32) > le32_to_cpu (*p_s_2_u32) ) + return 1; + } + return 0; +} + +inline int comp_short_cpu_keys (struct cpu_key * key1, + struct cpu_key * key2) +{ + __u32 * p_s_1_u32, * p_s_2_u32; + int n_key_length = REISERFS_SHORT_KEY_LEN; + + p_s_1_u32 = (__u32 *)key1; + p_s_2_u32 = (__u32 *)key2; + + for( ; n_key_length--; ++p_s_1_u32, ++p_s_2_u32 ) { + if ( *p_s_1_u32 < *p_s_2_u32 ) + return -1; + if ( *p_s_1_u32 > *p_s_2_u32 ) + return 1; + } + return 0; +} + + + +inline void cpu_key2cpu_key (struct cpu_key * to, struct cpu_key * from) +{ + memcpy (to, from, sizeof (struct cpu_key)); +} + + +inline void le_key2cpu_key (struct cpu_key * to, struct key * from) +{ + to->on_disk_key.k_dir_id = le32_to_cpu (from->k_dir_id); + to->on_disk_key.k_objectid = le32_to_cpu (from->k_objectid); + + // find out version of the key + to->version = le_key_version (from); + if (to->version == ITEM_VERSION_1) { + to->on_disk_key.u.k_offset_v1.k_offset = le32_to_cpu (from->u.k_offset_v1.k_offset); + to->on_disk_key.u.k_offset_v1.k_uniqueness = le32_to_cpu (from->u.k_offset_v1.k_uniqueness); + } else { + to->on_disk_key.u.k_offset_v2.k_offset = le64_to_cpu (from->u.k_offset_v2.k_offset); + to->on_disk_key.u.k_offset_v2.k_type = le16_to_cpu (from->u.k_offset_v2.k_type); + } +} + + + +// this does not say which one is bigger, it only returns 1 if keys +// are not equal, 0 otherwise +inline int comp_le_keys (struct key * k1, struct key * k2) +{ + return memcmp (k1, k2, sizeof (struct key)); +} + +/************************************************************************** + * Binary search toolkit function * + * Search for an item in the array by the item key * + * Returns: 1 if found, 0 if not found; * + * *p_n_pos = number of the searched element if found, else the * + * number of the first element that is larger than p_v_key. * + **************************************************************************/ +/* For those not familiar with binary search: n_lbound is the leftmost item that it + could be, n_rbound the rightmost item that it could be. We examine the item + halfway between n_lbound and n_rbound, and that tells us either that we can increase + n_lbound, or decrease n_rbound, or that we have found it, or if n_lbound <= n_rbound that + there are no possible items, and we have not found it. With each examination we + cut the number of possible items it could be by one more than half rounded down, + or we find it. */ +inline int bin_search ( + void * p_v_key, /* Key to search for. */ + void * p_v_base, /* First item in the array. */ + int p_n_num, /* Number of items in the array. */ + int p_n_width, /* Item size in the array. + searched. Lest the reader be + confused, note that this is crafted + as a general function, and when it + is applied specifically to the array + of item headers in a node, p_n_width + is actually the item header size not + the item size. */ + int * p_n_pos /* Number of the searched for element. */ + ) { + int n_rbound, n_lbound, n_j; + + for ( n_j = ((n_rbound = p_n_num - 1) + (n_lbound = 0))/2; n_lbound <= n_rbound; n_j = (n_rbound + n_lbound)/2 ) + switch( COMP_KEYS((struct key *)((char * )p_v_base + n_j * p_n_width), (struct cpu_key *)p_v_key) ) { + case -1: n_lbound = n_j + 1; continue; + case 1: n_rbound = n_j - 1; continue; + case 0: *p_n_pos = n_j; return ITEM_FOUND; /* Key found in the array. */ + } + + /* bin_search did not find given key, it returns position of key, + that is minimal and greater than the given one. */ + *p_n_pos = n_lbound; + return ITEM_NOT_FOUND; +} + +#ifdef CONFIG_REISERFS_CHECK +extern struct tree_balance * cur_tb; +#endif + + + +/* Minimal possible key. It is never in the tree. */ +struct key MIN_KEY = {0, 0, {{0, 0},}}; + +/* Maximal possible key. It is never in the tree. */ +struct key MAX_KEY = {0xffffffff, 0xffffffff, {{0xffffffff, 0xffffffff},}}; + + +/* Get delimiting key of the buffer by looking for it in the buffers in the path, starting from the bottom + of the path, and going upwards. We must check the path's validity at each step. If the key is not in + the path, there is no delimiting key in the tree (buffer is first or last buffer in tree), and in this + case we return a special key, either MIN_KEY or MAX_KEY. */ +inline struct key * get_lkey ( + struct path * p_s_chk_path, + struct super_block * p_s_sb + ) { + int n_position, n_path_offset = p_s_chk_path->path_length; + struct buffer_head * p_s_parent; + +#ifdef CONFIG_REISERFS_CHECK + if ( n_path_offset < FIRST_PATH_ELEMENT_OFFSET ) + reiserfs_panic(p_s_sb,"PAP-5010: get_lkey: illegal offset in the path"); +#endif + + /* While not higher in path than first element. */ + while ( n_path_offset-- > FIRST_PATH_ELEMENT_OFFSET ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( ! buffer_uptodate(PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)) ) + reiserfs_panic(p_s_sb, "PAP-5020: get_lkey: parent is not uptodate"); +#endif + + /* Parent at the path is not in the tree now. */ + if ( ! B_IS_IN_TREE(p_s_parent = PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)) ) + return &MAX_KEY; + /* Check whether position in the parent is correct. */ + if ( (n_position = PATH_OFFSET_POSITION(p_s_chk_path, n_path_offset)) > B_NR_ITEMS(p_s_parent) ) + return &MAX_KEY; + /* Check whether parent at the path really points to the child. */ + if ( B_N_CHILD_NUM(p_s_parent, n_position) != + PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset + 1)->b_blocknr ) + return &MAX_KEY; + /* Return delimiting key if position in the parent is not equal to zero. */ + if ( n_position ) + return B_N_PDELIM_KEY(p_s_parent, n_position - 1); + } + /* Return MIN_KEY if we are in the root of the buffer tree. */ + if ( PATH_OFFSET_PBUFFER(p_s_chk_path, FIRST_PATH_ELEMENT_OFFSET)->b_blocknr == + SB_ROOT_BLOCK (p_s_sb) ) + return &MIN_KEY; + return &MAX_KEY; +} + + +/* Get delimiting key of the buffer at the path and its right neighbor. */ +inline struct key * get_rkey ( + struct path * p_s_chk_path, + struct super_block * p_s_sb + ) { + int n_position, + n_path_offset = p_s_chk_path->path_length; + struct buffer_head * p_s_parent; + +#ifdef CONFIG_REISERFS_CHECK + if ( n_path_offset < FIRST_PATH_ELEMENT_OFFSET ) + reiserfs_panic(p_s_sb,"PAP-5030: get_rkey: illegal offset in the path"); +#endif + + while ( n_path_offset-- > FIRST_PATH_ELEMENT_OFFSET ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( ! buffer_uptodate(PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)) ) + reiserfs_panic(p_s_sb, "PAP-5040: get_rkey: parent is not uptodate"); +#endif + + /* Parent at the path is not in the tree now. */ + if ( ! B_IS_IN_TREE(p_s_parent = PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset)) ) + return &MIN_KEY; + /* Check whether position in the parrent is correct. */ + if ( (n_position = PATH_OFFSET_POSITION(p_s_chk_path, n_path_offset)) > B_NR_ITEMS(p_s_parent) ) + return &MIN_KEY; + /* Check whether parent at the path really points to the child. */ + if ( B_N_CHILD_NUM(p_s_parent, n_position) != + PATH_OFFSET_PBUFFER(p_s_chk_path, n_path_offset + 1)->b_blocknr ) + return &MIN_KEY; + /* Return delimiting key if position in the parent is not the last one. */ + if ( n_position != B_NR_ITEMS(p_s_parent) ) + return B_N_PDELIM_KEY(p_s_parent, n_position); + } + /* Return MAX_KEY if we are in the root of the buffer tree. */ + if ( PATH_OFFSET_PBUFFER(p_s_chk_path, FIRST_PATH_ELEMENT_OFFSET)->b_blocknr == + SB_ROOT_BLOCK (p_s_sb) ) + return &MAX_KEY; + return &MIN_KEY; +} + + +/* Check whether a key is contained in the tree rooted from a buffer at a path. */ +/* This works by looking at the left and right delimiting keys for the buffer in the last path_element in + the path. These delimiting keys are stored at least one level above that buffer in the tree. If the + buffer is the first or last node in the tree order then one of the delimiting keys may be absent, and in + this case get_lkey and get_rkey return a special key which is MIN_KEY or MAX_KEY. */ +static inline int key_in_buffer ( + struct path * p_s_chk_path, /* Path which should be checked. */ + struct cpu_key * p_s_key, /* Key which should be checked. */ + struct super_block * p_s_sb /* Super block pointer. */ + ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( ! p_s_key || p_s_chk_path->path_length < FIRST_PATH_ELEMENT_OFFSET || + p_s_chk_path->path_length > MAX_HEIGHT ) + reiserfs_panic(p_s_sb, "PAP-5050: key_in_buffer: pointer to the key(%p) is NULL or illegal path length(%d)", + p_s_key, p_s_chk_path->path_length); + + if ( PATH_PLAST_BUFFER(p_s_chk_path)->b_dev == NODEV ) + reiserfs_panic(p_s_sb, "PAP-5060: key_in_buffer: device must not be NODEV"); +#endif + + if ( COMP_KEYS(get_lkey(p_s_chk_path, p_s_sb), p_s_key) == 1 ) + /* left delimiting key is bigger, that the key we look for */ + return 0; + // if ( COMP_KEYS(p_s_key, get_rkey(p_s_chk_path, p_s_sb)) != -1 ) + if ( COMP_KEYS(get_rkey(p_s_chk_path, p_s_sb), p_s_key) != 1 ) + /* p_s_key must be less than right delimitiing key */ + return 0; + return 1; +} + + +inline void decrement_bcount( + struct buffer_head * p_s_bh + ) { + if ( p_s_bh ) { + if ( atomic_read (&(p_s_bh->b_count)) ) { + atomic_dec (&(p_s_bh->b_count)); + return; + } + reiserfs_panic(NULL, "PAP-5070: decrement_bcount: trying to free free buffer %b", p_s_bh); + } +} + + +/* Decrement b_count field of the all buffers in the path. */ +void decrement_counters_in_path ( + struct path * p_s_search_path + ) { + int n_path_offset = p_s_search_path->path_length; + +#ifdef CONFIG_REISERFS_CHECK + if ( n_path_offset < ILLEGAL_PATH_ELEMENT_OFFSET || + n_path_offset > EXTENDED_MAX_HEIGHT - 1 ) + reiserfs_panic(NULL, "PAP-5080: decrement_counters_in_path: illegal path offset of %d", n_path_offset); +#endif + + while ( n_path_offset > ILLEGAL_PATH_ELEMENT_OFFSET ) { + struct buffer_head * bh; + + bh = PATH_OFFSET_PBUFFER(p_s_search_path, n_path_offset--); + decrement_bcount (bh); + } + p_s_search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET; +} + + +int reiserfs_check_path(struct path *p) { +#ifdef CONFIG_REISERFS_CHECK + if (p->path_length != ILLEGAL_PATH_ELEMENT_OFFSET) { + reiserfs_warning("check_path, path not properly relsed\n") ; + BUG() ; + } +#endif + return 0 ; +} + + +/* Release all buffers in the path. Restore dirty bits clean +** when preparing the buffer for the log +** +** only called from fix_nodes() +*/ +void pathrelse_and_restore ( + struct super_block *s, + struct path * p_s_search_path + ) { + int n_path_offset = p_s_search_path->path_length; + +#ifdef CONFIG_REISERFS_CHECK + if ( n_path_offset < ILLEGAL_PATH_ELEMENT_OFFSET ) + reiserfs_panic(NULL, "clm-4000: pathrelse: illegal path offset"); +#endif + + while ( n_path_offset > ILLEGAL_PATH_ELEMENT_OFFSET ) { + reiserfs_restore_prepared_buffer(s, PATH_OFFSET_PBUFFER(p_s_search_path, + n_path_offset)); + brelse(PATH_OFFSET_PBUFFER(p_s_search_path, n_path_offset--)); + } + p_s_search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET; +} + +/* Release all buffers in the path. */ +void pathrelse ( + struct path * p_s_search_path + ) { + int n_path_offset = p_s_search_path->path_length; + +#ifdef CONFIG_REISERFS_CHECK + if ( n_path_offset < ILLEGAL_PATH_ELEMENT_OFFSET ) + reiserfs_panic(NULL, "PAP-5090: pathrelse: illegal path offset"); +#endif + + while ( n_path_offset > ILLEGAL_PATH_ELEMENT_OFFSET ) + brelse(PATH_OFFSET_PBUFFER(p_s_search_path, n_path_offset--)); + + p_s_search_path->path_length = ILLEGAL_PATH_ELEMENT_OFFSET; +} + + + +static int is_leaf (char * buf, int blocksize, struct buffer_head * bh) +{ + struct block_head * blkh; + struct item_head * ih; + int used_space; + int prev_location; + int i; + int nr; + + blkh = (struct block_head *)buf; + if (le16_to_cpu (blkh->blk_level) != DISK_LEAF_NODE_LEVEL) { + printk ("is_leaf: this should be caught earlier\n"); + return 0; + } + + nr = le16_to_cpu (blkh->blk_nr_item); + if (nr < 1 || nr > ((blocksize - BLKH_SIZE) / (IH_SIZE + MIN_ITEM_LEN))) { + /* item number is too big or too small */ + reiserfs_warning ("is_leaf: nr_item seems wrong: %z\n", bh); + return 0; + } + ih = (struct item_head *)(buf + BLKH_SIZE) + nr - 1; + used_space = BLKH_SIZE + IH_SIZE * nr + (blocksize - ih_location (ih)); + if (used_space != blocksize - le16_to_cpu (blkh->blk_free_space)) { + /* free space does not match to calculated amount of use space */ + reiserfs_warning ("is_leaf: free space seems wrong: %z\n", bh); + return 0; + } + + // FIXME: it is_leaf will hit performance too much - we may have + // return 1 here + + /* check tables of item heads */ + ih = (struct item_head *)(buf + BLKH_SIZE); + prev_location = blocksize; + for (i = 0; i < nr; i ++, ih ++) { + if (ih_location (ih) >= blocksize || ih_location (ih) < IH_SIZE * nr) { + reiserfs_warning ("is_leaf: item location seems wrong: %h\n", ih); + return 0; + } + if (ih_item_len (ih) < 1 || ih_item_len (ih) > MAX_ITEM_LEN (blocksize)) { + reiserfs_warning ("is_leaf: item length seems wrong: %h\n", ih); + return 0; + } + if (prev_location - ih_location (ih) != ih_item_len (ih)) { + reiserfs_warning ("is_leaf: item location seems wrong (second one): %h\n", ih); + return 0; + } + prev_location = ih_location (ih); + } + + // one may imagine much more checks + return 1; +} + + +/* returns 1 if buf looks like an internal node, 0 otherwise */ +static int is_internal (char * buf, int blocksize, struct buffer_head * bh) +{ + struct block_head * blkh; + int nr; + int used_space; + + blkh = (struct block_head *)buf; + if (le16_to_cpu (blkh->blk_level) <= DISK_LEAF_NODE_LEVEL || + le16_to_cpu (blkh->blk_level) > MAX_HEIGHT) { + /* this level is not possible for internal nodes */ + printk ("is_internal: this should be caught earlier\n"); + return 0; + } + + nr = le16_to_cpu (blkh->blk_nr_item); + if (nr > (blocksize - BLKH_SIZE - DC_SIZE) / (KEY_SIZE + DC_SIZE)) { + /* for internal which is not root we might check min number of keys */ + reiserfs_warning ("is_internal: number of key seems wrong: %z\n", bh); + return 0; + } + + used_space = BLKH_SIZE + KEY_SIZE * nr + DC_SIZE * (nr + 1); + if (used_space != blocksize - le16_to_cpu (blkh->blk_free_space)) { + reiserfs_warning ("is_internal: free space seems wrong: %z\n", bh); + return 0; + } + + // one may imagine much more checks + return 1; +} + + +// make sure that bh contains formatted node of reiserfs tree of +// 'level'-th level +static int is_tree_node (struct buffer_head * bh, int level) +{ + if (B_LEVEL (bh) != level) { + printk ("is_tree_node: node level %d does not match to the expected one %d\n", + B_LEVEL (bh), level); + return 0; + } + if (level == DISK_LEAF_NODE_LEVEL) + return is_leaf (bh->b_data, bh->b_size, bh); + + return is_internal (bh->b_data, bh->b_size, bh); +} + + + +#ifdef SEARCH_BY_KEY_READA + +/* The function is NOT SCHEDULE-SAFE! */ +static void search_by_key_reada (struct super_block * s, int blocknr) +{ + struct buffer_head * bh; + + if (blocknr == 0) + return; + + bh = reiserfs_getblk (s->s_dev, blocknr, s->s_blocksize); + + if (!buffer_uptodate (bh)) { + ll_rw_block (READA, 1, &bh); + } + bh->b_count --; +} + +#endif + +/************************************************************************** + * Algorithm SearchByKey * + * look for item in the Disk S+Tree by its key * + * Input: p_s_sb - super block * + * p_s_key - pointer to the key to search * + * Output: ITEM_FOUND, ITEM_NOT_FOUND or IO_ERROR * + * p_s_search_path - path from the root to the needed leaf * + **************************************************************************/ + +/* This function fills up the path from the root to the leaf as it + descends the tree looking for the key. It uses reiserfs_bread to + try to find buffers in the cache given their block number. If it + does not find them in the cache it reads them from disk. For each + node search_by_key finds using reiserfs_bread it then uses + bin_search to look through that node. bin_search will find the + position of the block_number of the next node if it is looking + through an internal node. If it is looking through a leaf node + bin_search will find the position of the item which has key either + equal to given key, or which is the maximal key less than the given + key. search_by_key returns a path that must be checked for the + correctness of the top of the path but need not be checked for the + correctness of the bottom of the path */ +/* The function is NOT SCHEDULE-SAFE! */ +int search_by_key (struct super_block * p_s_sb, + struct cpu_key * p_s_key, /* Key to search. */ + struct path * p_s_search_path, /* This structure was + allocated and initialized + by the calling + function. It is filled up + by this function. */ + int n_stop_level /* How far down the tree to search. To + stop at leaf level - set to + DISK_LEAF_NODE_LEVEL */ + ) { + kdev_t n_dev = p_s_sb->s_dev; + int n_block_number = SB_ROOT_BLOCK (p_s_sb), + expected_level = SB_TREE_HEIGHT (p_s_sb), + n_block_size = p_s_sb->s_blocksize; + struct buffer_head * p_s_bh; + struct path_element * p_s_last_element; + int n_node_level, n_retval; + int right_neighbor_of_leaf_node; + int fs_gen; + +#ifdef CONFIG_REISERFS_CHECK + int n_repeat_counter = 0; +#endif + + /* As we add each node to a path we increase its count. This means that + we must be careful to release all nodes in a path before we either + discard the path struct or re-use the path struct, as we do here. */ + + decrement_counters_in_path(p_s_search_path); + + right_neighbor_of_leaf_node = 0; + + /* With each iteration of this loop we search through the items in the + current node, and calculate the next current node(next path element) + for the next iteration of this loop.. */ + while ( 1 ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( !(++n_repeat_counter % 50000) ) + reiserfs_warning ("PAP-5100: search_by_key: %s:" + "there were %d iterations of while loop " + "looking for key %K\n", + current->comm, n_repeat_counter, p_s_key); +#endif + + /* prep path to have another element added to it. */ + p_s_last_element = PATH_OFFSET_PELEMENT(p_s_search_path, ++p_s_search_path->path_length); + fs_gen = get_generation (p_s_sb); + expected_level --; + +#ifdef SEARCH_BY_KEY_READA + /* schedule read of right neighbor */ + search_by_key_reada (p_s_sb, right_neighbor_of_leaf_node); +#endif + + /* Read the next tree node, and set the last element in the path to + have a pointer to it. */ + if ( ! (p_s_bh = p_s_last_element->pe_buffer = + reiserfs_bread(n_dev, n_block_number, n_block_size)) ) { + p_s_search_path->path_length --; + pathrelse(p_s_search_path); + return IO_ERROR; + } + + /* It is possible that schedule occured. We must check whether the key + to search is still in the tree rooted from the current buffer. If + not then repeat search from the root. */ + if ( fs_changed (fs_gen, p_s_sb) && + (!B_IS_IN_TREE (p_s_bh) || !key_in_buffer(p_s_search_path, p_s_key, p_s_sb)) ) { + decrement_counters_in_path(p_s_search_path); + + /* Get the root block number so that we can repeat the search + starting from the root. */ + n_block_number = SB_ROOT_BLOCK (p_s_sb); + expected_level = SB_TREE_HEIGHT (p_s_sb); + right_neighbor_of_leaf_node = 0; + + /* repeat search from the root */ + continue; + } + +#ifdef CONFIG_REISERFS_CHECK + + if ( ! key_in_buffer(p_s_search_path, p_s_key, p_s_sb) ) + reiserfs_panic(p_s_sb, "PAP-5130: search_by_key: key is not in the buffer"); + if ( cur_tb ) { + print_cur_tb ("5140"); + reiserfs_panic(p_s_sb, "PAP-5140: search_by_key: schedule occurred in do_balance!"); + } + +#endif + + // make sure, that the node contents look like a node of + // certain level + if (!is_tree_node (p_s_bh, expected_level)) { + reiserfs_warning ("vs-5150: search_by_key: " + "invalid format found in block %d. Fsck?\n", p_s_bh->b_blocknr); + pathrelse (p_s_search_path); + return IO_ERROR; + } + + /* ok, we have acquired next formatted node in the tree */ + n_node_level = B_LEVEL (p_s_bh); + +#ifdef CONFIG_REISERFS_CHECK + + if (n_node_level < n_stop_level) + reiserfs_panic (p_s_sb, "vs-5152: search_by_key: tree level is less than stop level (%d)", + n_node_level, n_stop_level); + +#endif + + n_retval = bin_search (p_s_key, B_N_PITEM_HEAD(p_s_bh, 0), B_NR_ITEMS(p_s_bh), + ( n_node_level == DISK_LEAF_NODE_LEVEL ) ? IH_SIZE : KEY_SIZE, &(p_s_last_element->pe_position)); + if (n_node_level == n_stop_level) { + return n_retval; + } + + /* we are not in the stop level */ + if (n_retval == ITEM_FOUND) + /* item has been found, so we choose the pointer which is to the right of the found one */ + p_s_last_element->pe_position++; + + /* if item was not found we choose the position which is to + the left of the found item. This requires no code, + bin_search did it already.*/ + + /* So we have chosen a position in the current node which is + an internal node. Now we calculate child block number by + position in the node. */ + n_block_number = B_N_CHILD_NUM(p_s_bh, p_s_last_element->pe_position); + +#ifdef SEARCH_BY_KEY_READA + /* if we are going to read leaf node, then calculate its right neighbor if possible */ + if (n_node_level == DISK_LEAF_NODE_LEVEL + 1 && p_s_last_element->pe_position < B_NR_ITEMS (p_s_bh)) + right_neighbor_of_leaf_node = B_N_CHILD_NUM(p_s_bh, p_s_last_element->pe_position + 1); +#endif + } +} + + +/* Form the path to an item and position in this item which contains + file byte defined by p_s_key. If there is no such item + corresponding to the key, we point the path to the item with + maximal key less than p_s_key, and *p_n_pos_in_item is set to one + past the last entry/byte in the item. If searching for entry in a + directory item, and it is not found, *p_n_pos_in_item is set to one + entry more than the entry with maximal key which is less than the + sought key. + + Note that if there is no entry in this same node which is one more, + then we point to an imaginary entry. for direct items, the + position is in units of bytes, for indirect items the position is + in units of blocknr entries, for directory items the position is in + units of directory entries. */ + +/* The function is NOT SCHEDULE-SAFE! */ +int search_for_position_by_key (struct super_block * p_s_sb, /* Pointer to the super block. */ + struct cpu_key * p_cpu_key, /* Key to search (cpu variable) */ + struct path * p_s_search_path /* Filled up by this function. */ + ) { + struct item_head * p_le_ih; /* pointer to on-disk structure */ + int n_blk_size; + loff_t item_offset, offset; + struct reiserfs_dir_entry de; + int retval; + + /* If searching for directory entry. */ + if ( is_direntry_cpu_key (p_cpu_key) ) + return search_by_entry_key (p_s_sb, p_cpu_key, p_s_search_path, &de); + + /* If not searching for directory entry. */ + + /* If item is found. */ + retval = search_item (p_s_sb, p_cpu_key, p_s_search_path); + if (retval == IO_ERROR) + return retval; + if ( retval == ITEM_FOUND ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( ! B_N_PITEM_HEAD(PATH_PLAST_BUFFER(p_s_search_path), + PATH_LAST_POSITION(p_s_search_path))->ih_item_len ) + reiserfs_panic(p_s_sb, "PAP-5165: search_for_position_by_key: item length equals zero"); +#endif + + pos_in_item(p_s_search_path) = 0; + return POSITION_FOUND; + } + +#ifdef CONFIG_REISERFS_CHECK + if ( ! PATH_LAST_POSITION(p_s_search_path) ) + reiserfs_panic(p_s_sb, "PAP-5170: search_for_position_by_key: position equals zero"); +#endif + + /* Item is not found. Set path to the previous item. */ + p_le_ih = B_N_PITEM_HEAD(PATH_PLAST_BUFFER(p_s_search_path), --PATH_LAST_POSITION(p_s_search_path)); + n_blk_size = p_s_sb->s_blocksize; + + if (comp_short_keys (&(p_le_ih->ih_key), p_cpu_key)) { + return FILE_NOT_FOUND; + } + +#if 0 +/*#ifdef CONFIG_REISERFS_CHECK*/ + + /* we expect to find stat data or item of the same type */ + if ( ! is_statdata_le_ih(p_le_ih) && ((is_indirect_cpu_key(p_cpu_key) && ! is_indirect_le_ih(p_le_ih)) || + (is_direct_cpu_key(p_cpu_key) && ! is_direct_le_ih(p_le_ih))) ) { + print_block (PATH_PLAST_BUFFER(p_s_search_path), PRINT_LEAF_ITEMS, + PATH_LAST_POSITION (p_s_search_path) - 2, + PATH_LAST_POSITION (p_s_search_path) + 2); + reiserfs_panic(p_s_sb, "PAP-5190: search_for_position_by_key: " + "found item %h type does not match to the expected one %k", + p_le_ih, p_cpu_key); + } +/*#endif*/ +#endif + + // FIXME: quite ugly this far + + item_offset = le_ih_k_offset (p_le_ih); + offset = cpu_key_k_offset (p_cpu_key); + + /* Needed byte is contained in the item pointed to by the path.*/ + if (item_offset <= offset && + item_offset + op_bytes_number (p_le_ih, n_blk_size) > offset) { + pos_in_item (p_s_search_path) = offset - item_offset; + if ( is_indirect_le_ih(p_le_ih) ) { + pos_in_item (p_s_search_path) /= n_blk_size; + } + return POSITION_FOUND; + } + + /* Needed byte is not contained in the item pointed to by the + path. Set pos_in_item out of the item. */ + if ( is_indirect_le_ih (p_le_ih) ) + pos_in_item (p_s_search_path) = le16_to_cpu (p_le_ih->ih_item_len) / UNFM_P_SIZE; + else + pos_in_item (p_s_search_path) = le16_to_cpu (p_le_ih->ih_item_len); + + return POSITION_NOT_FOUND; +} + + +/* Compare given item and item pointed to by the path. */ +int comp_items (struct item_head * stored_ih, struct path * p_s_path) +{ + struct buffer_head * p_s_bh; + struct item_head * ih; + + /* Last buffer at the path is not in the tree. */ + if ( ! B_IS_IN_TREE(p_s_bh = PATH_PLAST_BUFFER(p_s_path)) ) + return 1; + + /* Last path position is invalid. */ + if ( PATH_LAST_POSITION(p_s_path) >= B_NR_ITEMS(p_s_bh) ) + return 1; + + /* we need only to know, whether it is the same item */ + ih = get_ih (p_s_path); + return memcmp (stored_ih, ih, IH_SIZE); + +#if 0 + /* Get item at the path. */ + p_s_path_item = PATH_PITEM_HEAD(p_s_path); + /* Compare keys. */ + if ( COMP_KEYS(&(p_s_path_item->ih_key), &(p_cpu_ih->ih_key)) ) + return 1; + + /* Compare other items fields. */ + if ( le16_to_cpu (p_s_path_item->u.ih_entry_count) != p_cpu_ih->u.ih_entry_count || + le16_to_cpu (p_s_path_item->ih_item_len) != p_cpu_ih->ih_item_len || + le16_to_cpu ( p_s_path_item->ih_item_location) != p_cpu_ih->ih_item_location ) + return 1; + + /* Items are equal. */ + return 0; +#endif +} + + +/* unformatted nodes are not logged anymore, ever. This is safe +** now +*/ +#define held_by_others(bh) (atomic_read(&(bh)->b_count) > 1) + +// block can not be forgotten as it is in I/O or held by someone +#define block_in_use(bh) (buffer_locked(bh) || (held_by_others(bh))) + + + +// prepare for delete or cut of direct item +static inline int prepare_for_direct_item (struct path * path, + struct item_head * le_ih, + struct inode * inode, + loff_t new_file_length, + int * cut_size) +{ + loff_t round_len; + + + if ( new_file_length == max_reiserfs_offset (inode) ) { + /* item has to be deleted */ + *cut_size = -(IH_SIZE + le16_to_cpu (le_ih->ih_item_len)); + return M_DELETE; + } + + // new file gets truncated + if (inode_items_version (inode) == ITEM_VERSION_2) { + // + round_len = ROUND_UP (new_file_length); + /* this was n_new_file_length < le_ih ... */ + if ( round_len < le_ih_k_offset (le_ih) ) { + *cut_size = -(IH_SIZE + le16_to_cpu (le_ih->ih_item_len)); + return M_DELETE; /* Delete this item. */ + } + /* Calculate first position and size for cutting from item. */ + pos_in_item (path) = round_len - (le_ih_k_offset (le_ih) - 1); + *cut_size = -(le16_to_cpu (le_ih->ih_item_len) - pos_in_item(path)); + + return M_CUT; /* Cut from this item. */ + } + + + // old file: items may have any length + + if ( new_file_length < le_ih_k_offset (le_ih) ) { + *cut_size = -(IH_SIZE + le16_to_cpu (le_ih->ih_item_len)); + return M_DELETE; /* Delete this item. */ + } + /* Calculate first position and size for cutting from item. */ + *cut_size = -(le16_to_cpu (le_ih->ih_item_len) - + (pos_in_item (path) = new_file_length + 1 - le_ih_k_offset (le_ih))); + return M_CUT; /* Cut from this item. */ +} + + +static inline int prepare_for_direntry_item (struct path * path, + struct item_head * le_ih, + struct inode * inode, + loff_t new_file_length, + int * cut_size) +{ + if (le_ih_k_offset (le_ih) == DOT_OFFSET && + new_file_length == max_reiserfs_offset (inode)) { +#ifdef CONFIG_REISERFS_CHECK + if (ih_entry_count (le_ih) != 2) + reiserfs_panic(inode->i_sb,"PAP-5220: prepare_for_delete_or_cut: " + "incorrect empty directory item (%h)", le_ih); +#endif + *cut_size = -(IH_SIZE + le16_to_cpu (le_ih->ih_item_len)); + return M_DELETE; /* Delete the directory item containing "." and ".." entry. */ + } + + if ( ih_entry_count (le_ih) == 1 ) { + /* Delete the directory item such as there is one record only + in this item*/ + *cut_size = -(IH_SIZE + le16_to_cpu (le_ih->ih_item_len)); + return M_DELETE; + } + + /* Cut one record from the directory item. */ + *cut_size = -(DEH_SIZE + entry_length (get_bh (path), le_ih, pos_in_item (path))); + return M_CUT; +} + + +/* If the path points to a directory or direct item, calculate mode and the size cut, for balance. + If the path points to an indirect item, remove some number of its unformatted nodes. + In case of file truncate calculate whether this item must be deleted/truncated or last + unformatted node of this item will be converted to a direct item. + This function returns a determination of what balance mode the calling function should employ. */ +static char prepare_for_delete_or_cut( + struct reiserfs_transaction_handle *th, + struct inode * inode, + struct path * p_s_path, + struct cpu_key * p_s_item_key, + int * p_n_removed, /* Number of unformatted nodes which were removed + from end of the file. */ + int * p_n_cut_size, + unsigned long long n_new_file_length /* MAX_KEY_OFFSET in case of delete. */ + ) { + struct super_block * p_s_sb = inode->i_sb; + struct item_head * p_le_ih = PATH_PITEM_HEAD(p_s_path); + struct buffer_head * p_s_bh = PATH_PLAST_BUFFER(p_s_path); + +#ifdef CONFIG_REISERFS_CHECK + int n_repeat_counter = 0; +#endif + + /* Stat_data item. */ + if ( is_statdata_le_ih (p_le_ih) ) { + +#ifdef CONFIG_REISERFS_CHECK + if ( n_new_file_length != max_reiserfs_offset (inode) ) + reiserfs_panic(p_s_sb, "PAP-5210: prepare_for_delete_or_cut: mode must be M_DELETE"); +#endif + + *p_n_cut_size = -(IH_SIZE + le16_to_cpu (p_le_ih->ih_item_len)); + return M_DELETE; + } + + + /* Directory item. */ + if ( is_direntry_le_ih (p_le_ih) ) + return prepare_for_direntry_item (p_s_path, p_le_ih, inode, n_new_file_length, p_n_cut_size); + + /* Direct item. */ + if ( is_direct_le_ih (p_le_ih) ) + return prepare_for_direct_item (p_s_path, p_le_ih, inode, n_new_file_length, p_n_cut_size); + + + /* Case of an indirect item. */ + { + int n_unfm_number, /* Number of the item unformatted nodes. */ + n_counter, + n_retry, /* Set to one if there is unformatted node buffer in use. */ + n_blk_size; + __u32 * p_n_unfm_pointer; /* Pointer to the unformatted node number. */ + __u32 tmp; + struct item_head s_ih; /* Item header. */ + char c_mode; /* Returned mode of the balance. */ + struct buffer_head * p_s_un_bh; + int need_research; + + + n_blk_size = p_s_sb->s_blocksize; + + /* Search for the needed object indirect item until there are no unformatted nodes to be removed. */ + do { + need_research = 0; + p_s_bh = PATH_PLAST_BUFFER(p_s_path); + /* Copy indirect item header to a temp variable. */ + copy_item_head(&s_ih, PATH_PITEM_HEAD(p_s_path)); + /* Calculate number of unformatted nodes in this item. */ + n_unfm_number = I_UNFM_NUM(&s_ih); + +#ifdef CONFIG_REISERFS_CHECK + if ( ! is_indirect_le_ih(&s_ih) || ! n_unfm_number || + pos_in_item (p_s_path) + 1 != n_unfm_number ) { + printk("n_unfm_number = %d *p_n_pos_in_item = %d\n",n_unfm_number, pos_in_item (p_s_path)); + reiserfs_panic(p_s_sb, "PAP-5240: prepare_for_delete_or_cut: illegal item %h", &s_ih); + } +#endif + + /* Calculate balance mode and position in the item to remove unformatted nodes. */ + if ( n_new_file_length == max_reiserfs_offset (inode) ) {/* Case of delete. */ + pos_in_item (p_s_path) = 0; + *p_n_cut_size = -(IH_SIZE + le16_to_cpu (s_ih.ih_item_len)); + c_mode = M_DELETE; + } + else { /* Case of truncate. */ + if ( n_new_file_length < le_ih_k_offset (&s_ih) ) { + pos_in_item (p_s_path) = 0; + *p_n_cut_size = -(IH_SIZE + le16_to_cpu (s_ih.ih_item_len)); + c_mode = M_DELETE; /* Delete this item. */ + } + else { + /* indirect item must be truncated starting from *p_n_pos_in_item-th position */ + pos_in_item (p_s_path) = (n_new_file_length + n_blk_size - le_ih_k_offset (&s_ih) ) >> p_s_sb->s_blocksize_bits; + +#ifdef CONFIG_REISERFS_CHECK + if ( pos_in_item (p_s_path) > n_unfm_number ) + reiserfs_panic(p_s_sb, "PAP-5250: prepare_for_delete_or_cut: illegal position in the item"); +#endif + + /* Either convert last unformatted node of indirect item to direct item or increase + its free space. */ + if ( pos_in_item (p_s_path) == n_unfm_number ) { + *p_n_cut_size = 0; /* Nothing to cut. */ + return M_CONVERT; /* Maybe convert last unformatted node to the direct item. */ + } + /* Calculate size to cut. */ + *p_n_cut_size = -(s_ih.ih_item_len - pos_in_item (p_s_path) * UNFM_P_SIZE); + + c_mode = M_CUT; /* Cut from this indirect item. */ + } + } + +#ifdef CONFIG_REISERFS_CHECK + if ( n_unfm_number <= pos_in_item (p_s_path) ) + reiserfs_panic(p_s_sb, "PAP-5260: prepare_for_delete_or_cut: illegal position in the indirect item"); +#endif + + /* pointers to be cut */ + n_unfm_number -= pos_in_item (p_s_path); + /* Set pointer to the last unformatted node pointer that is to be cut. */ + p_n_unfm_pointer = (__u32 *)B_I_PITEM(p_s_bh, &s_ih) + I_UNFM_NUM(&s_ih) - 1 - *p_n_removed; + + + /* We go through the unformatted nodes pointers of the indirect + item and look for the unformatted nodes in the cache. If we + found some of them we free it, zero corresponding indirect item + entry and log buffer containing that indirect item. For this we + need to prepare last path element for logging. If some + unformatted node has b_count > 1 we must not free this + unformatted node since it is in use. */ + reiserfs_prepare_for_journal(p_s_sb, p_s_bh, 1); + // note: path could be changed, first line in for loop takes care + // of it + + for ( n_retry = 0, n_counter = *p_n_removed; + n_counter < n_unfm_number; n_counter++, p_n_unfm_pointer-- ) { + + if (item_moved (&s_ih, p_s_path)) { + need_research = 1 ; + break; + } +#ifdef CONFIG_REISERFS_CHECK + if (p_n_unfm_pointer < (__u32 *)B_I_PITEM(p_s_bh, &s_ih) || + p_n_unfm_pointer > (__u32 *)B_I_PITEM(p_s_bh, &s_ih) + I_UNFM_NUM(&s_ih) - 1) + reiserfs_panic (p_s_sb, "vs-5265: prepare_for_delete_or_cut: pointer out of range"); +#endif + + if ( ! *p_n_unfm_pointer ) { /* Hole, nothing to remove. */ + if ( ! n_retry ) + (*p_n_removed)++; + continue; + } + /* Search for the buffer in cache. */ + p_s_un_bh = get_hash_table(p_s_sb->s_dev, *p_n_unfm_pointer, n_blk_size); + + if (p_s_un_bh && buffer_locked(p_s_un_bh)) { + __wait_on_buffer(p_s_un_bh) ; + if ( item_moved (&s_ih, p_s_path) ) { + need_research = 1; + brelse(p_s_un_bh) ; + break ; + } + } + if ( p_s_un_bh && block_in_use (p_s_un_bh)) { + /* Block is locked or held more than by one holder and by + journal. */ + +#ifndef __KERNEL__ + reiserfs_panic(p_s_sb, "PAP-5270: prepare_for_delete_or_cut: b_count != 1"); +#endif + +#ifdef CONFIG_REISERFS_CHECK + if (n_repeat_counter && (n_repeat_counter % 100000) == 0) { + printk("prepare_for_delete, waiting on buffer %lu, b_count %d, %s%cJDIRTY %cJDIRTY_WAIT\n", + p_s_un_bh->b_blocknr, atomic_read (&p_s_un_bh->b_count), + buffer_locked (p_s_un_bh) ? "locked, " : "", + buffer_journaled(p_s_un_bh) ? ' ' : '!', + buffer_journal_dirty(p_s_un_bh) ? ' ' : '!') ; + + } +#endif + n_retry = 1; + brelse (p_s_un_bh); + continue; + } + + if ( ! n_retry ) + (*p_n_removed)++; + +#ifdef CONFIG_REISERFS_CHECK + if ( p_s_un_bh && (*p_n_unfm_pointer != p_s_un_bh->b_blocknr )) + // note: minix_truncate allows that. As truncate is + // protected by down (inode->i_sem), two truncates can not + // co-exist + reiserfs_panic(p_s_sb, "PAP-5280: prepare_for_delete_or_cut: blocks numbers are different"); +#endif + + tmp = *p_n_unfm_pointer; + *p_n_unfm_pointer = 0; + journal_mark_dirty (th, p_s_sb, p_s_bh); + bforget (p_s_un_bh); + inode->i_blocks -= p_s_sb->s_blocksize / 512; + reiserfs_free_block(th, tmp); + if ( item_moved (&s_ih, p_s_path) ) { + need_research = 1; + break ; +#if 0 + reiserfs_prepare_for_journal(p_s_sb, + PATH_PLAST_BUFFER(p_s_path), + 1) ; + if ( comp_items(&s_ih, p_s_path) ) { + reiserfs_restore_prepared_buffer(p_s_sb, + PATH_PLAST_BUFFER(p_s_path)) ; + brelse(p_s_un_bh); + break; + } + *p_n_unfm_pointer = 0; + journal_mark_dirty (th,p_s_sb,PATH_PLAST_BUFFER(p_s_path)); + + reiserfs_free_block(th, p_s_sb, block_addr); + if (p_s_un_bh) { + mark_buffer_clean (p_s_un_bh); + brelse (p_s_un_bh); + } + if ( comp_items(&s_ih, p_s_path) ) { + break ; + } +#endif + } + + } + + /* a trick. If the buffer has been logged, this + ** will do nothing. If we've broken the loop without + ** logging it, it will restore the buffer + ** + */ + reiserfs_restore_prepared_buffer(p_s_sb, p_s_bh); + + if ( n_retry ) { + /* There is block in use. Wait, they should release it soon */ + +#ifdef CONFIG_REISERFS_CHECK + if ( *p_n_removed >= n_unfm_number ) + reiserfs_panic(p_s_sb, "PAP-5290: prepare_for_delete_or_cut: illegal case"); + if ( !(++n_repeat_counter % 500000) ) { + reiserfs_warning("PAP-5300: prepare_for_delete_or_cut: (pid %u): " + "could not delete item %k in (%d) iterations. New file length %Lu. (inode %Ld), Still trying\n", + current->pid, p_s_item_key, n_repeat_counter, n_new_file_length, inode->i_size); + if (n_repeat_counter == 5000000) { + print_block (PATH_PLAST_BUFFER(p_s_path), 3, + PATH_LAST_POSITION (p_s_path) - 2, PATH_LAST_POSITION (p_s_path) + 2); + reiserfs_panic(p_s_sb, "PAP-5305: prepare_for_delete_or_cut: key %k, new_file_length %Ld", + p_s_item_key, n_new_file_length); + } + } +#endif + +#ifdef __KERNEL__ + run_task_queue(&tq_disk); + current->policy |= SCHED_YIELD; + schedule(); +#endif + } + /* This loop can be optimized. */ + } while ( (*p_n_removed < n_unfm_number || need_research) && + search_for_position_by_key(p_s_sb, p_s_item_key, p_s_path) == POSITION_FOUND ); + +#ifdef CONFIG_REISERFS_CHECK + if ( *p_n_removed < n_unfm_number ) + reiserfs_panic(p_s_sb, "PAP-5310: prepare_for_delete_or_cut: indirect item is not found"); + + if (item_moved (&s_ih, p_s_path) ) { + printk("prepare_for_delete_or_cut: after while, comp failed, retry\n") ; + BUG (); + } +#endif + + if (c_mode == M_CUT) + pos_in_item (p_s_path) *= UNFM_P_SIZE; + return c_mode; + } +} + + +/* Calculate bytes number which will be deleted or cutted in the balance. */ +int calc_deleted_bytes_number( + struct tree_balance * p_s_tb, + char c_mode + ) { + int n_del_size; + struct item_head * p_le_ih = PATH_PITEM_HEAD(p_s_tb->tb_path); + + if ( is_statdata_le_ih (p_le_ih) ) + return 0; + + if ( is_direntry_le_ih (p_le_ih) ) { + // return EMPTY_DIR_SIZE; /* We delete emty directoris only. */ + // we can't use EMPTY_DIR_SIZE, as old format dirs have a different + // empty size. ick. FIXME, is this right? + // + return le16_to_cpu(p_le_ih->ih_item_len) ; + } + n_del_size = ( c_mode == M_DELETE ) ? le16_to_cpu (p_le_ih->ih_item_len) : -p_s_tb->insert_size[0]; + + if ( is_indirect_le_ih (p_le_ih) ) + n_del_size = (n_del_size/UNFM_P_SIZE)* + (PATH_PLAST_BUFFER(p_s_tb->tb_path)->b_size);// - get_ih_free_space (p_le_ih); + return n_del_size; +} + +static void init_tb_struct( + struct reiserfs_transaction_handle *th, + struct tree_balance * p_s_tb, + struct super_block * p_s_sb, + struct path * p_s_path, + int n_size + ) { + memset (p_s_tb,'\0',sizeof(struct tree_balance)); + p_s_tb->transaction_handle = th ; + p_s_tb->tb_sb = p_s_sb; + p_s_tb->tb_path = p_s_path; + PATH_OFFSET_PBUFFER(p_s_path, ILLEGAL_PATH_ELEMENT_OFFSET) = NULL; + PATH_OFFSET_POSITION(p_s_path, ILLEGAL_PATH_ELEMENT_OFFSET) = 0; + p_s_tb->insert_size[0] = n_size; +} + + + +void padd_item (char * item, int total_length, int length) +{ + int i; + + for (i = total_length; i > length; ) + item [--i] = 0; +} + + +/* Delete object item. */ +int reiserfs_delete_item (struct reiserfs_transaction_handle *th, + struct path * p_s_path, /* Path to the deleted item. */ + struct cpu_key * p_s_item_key, /* Key to search for the deleted item. */ + struct inode * p_s_inode,/* inode is here just to update i_blocks */ + struct buffer_head * p_s_un_bh) /* NULL or unformatted node pointer. */ +{ + struct super_block * p_s_sb = p_s_inode->i_sb; + struct tree_balance s_del_balance; + struct item_head s_ih; + int n_ret_value, + n_del_size, + n_removed; + +#ifdef CONFIG_REISERFS_CHECK + char c_mode; + int n_iter = 0; +#endif + + init_tb_struct(th, &s_del_balance, p_s_sb, p_s_path, 0/*size is unknown*/); + + while ( 1 ) { + n_removed = 0; + +#ifdef CONFIG_REISERFS_CHECK + n_iter++; + c_mode = +#endif + prepare_for_delete_or_cut(th, p_s_inode, p_s_path, p_s_item_key, &n_removed, &n_del_size, max_reiserfs_offset (p_s_inode)); + +#ifdef CONFIG_REISERFS_CHECK + if ( c_mode != M_DELETE ) + reiserfs_panic(p_s_sb, "PAP-5320: reiserfs_delete_item: mode must be M_DELETE"); +#endif + + copy_item_head(&s_ih, PATH_PITEM_HEAD(p_s_path)); + s_del_balance.insert_size[0] = n_del_size; + + n_ret_value = fix_nodes(M_DELETE, &s_del_balance, NULL, 0); + if ( n_ret_value != REPEAT_SEARCH ) + break; + + // file system changed, repeat search + n_ret_value = search_for_position_by_key(p_s_sb, p_s_item_key, p_s_path); + if (n_ret_value == IO_ERROR) + break; + if (n_ret_value == FILE_NOT_FOUND) { + reiserfs_warning ("vs-5340: reiserfs_delete_item: " + "no items of the file %K found\n", p_s_item_key); + break; + } + } /* while (1) */ + + if ( n_ret_value != CARRY_ON ) { + unfix_nodes(&s_del_balance); + return 0; + } + + // reiserfs_delete_item returns item length when success + n_ret_value = calc_deleted_bytes_number(&s_del_balance, M_DELETE); + + if ( p_s_un_bh ) { + int off; + int block_off ; + char *data ; + + /* We are in direct2indirect conversion, so move tail contents + to the unformatted node */ + /* note, we do the copy before preparing the buffer because we + ** don't care about the contents of the unformatted node yet. + ** the only thing we really care about is the direct item's data + ** is in the unformatted node. + ** + ** Otherwise, we would have to call reiserfs_prepare_for_journal on + ** the unformatted node, which might schedule, meaning we'd have to + ** loop all the way back up to the start of the while loop. + ** + ** The unformatted node is prepared and logged after the do_balance. + ** + ** p_s_un_bh is from the page cache (all unformatted nodes are + ** from the page cache) and might be a highmem page. So, we + ** can't use p_s_un_bh->b_data. But, the page has already been + ** kmapped, so we can use page_address() + ** -clm + */ + + data = page_address(p_s_un_bh->b_page) ; + off = ((le_ih_k_offset (&s_ih) - 1) & (PAGE_CACHE_SIZE - 1)); + block_off = off & (p_s_un_bh->b_size - 1) ; + memcpy(data + off, + B_I_PITEM(PATH_PLAST_BUFFER(p_s_path), &s_ih), n_ret_value); + + /* clear out the rest of the block past the end of the file. */ + if (block_off + n_ret_value < p_s_un_bh->b_size) { + memset(data + off + n_ret_value, 0, + p_s_un_bh->b_size - block_off - n_ret_value) ; + } + } + + /* Perform balancing after all resources have been collected at once. */ + do_balance(&s_del_balance, NULL, NULL, M_DELETE); + + /* see comment above for why this is after the do_balance */ + if (p_s_un_bh) { + mark_buffer_dirty(p_s_un_bh) ; + } + + /* Return deleted body length */ + return n_ret_value; +} + + +/* Summary Of Mechanisms For Handling Collisions Between Processes: + + deletion of the body of the object is performed by iput(), with the + result that if multiple processes are operating on a file, the + deletion of the body of the file is deferred until the last process + that has an open inode performs its iput(). + + writes and truncates are protected from collisions by use of + semaphores. + + creates, linking, and mknod are protected from collisions with other + processes by making the reiserfs_add_entry() the last step in the + creation, and then rolling back all changes if there was a collision. + - Hans +*/ + + +/* this deletes item which never gets split */ +static void reiserfs_delete_solid_item (struct reiserfs_transaction_handle *th, + struct key * key) +{ + struct tree_balance tb; + INITIALIZE_PATH (path); + int item_len; + int tb_init = 0 ; + struct cpu_key cpu_key; + int retval; + + le_key2cpu_key (&cpu_key, key); + + while (1) { + retval = search_item (th->t_super, &cpu_key, &path); + if (retval == IO_ERROR) { + reiserfs_warning ("vs-: reiserfs_delete_solid_item: " + "i/o failure occured trying to delete %K\n", &cpu_key); + break; + } + if (retval != ITEM_FOUND) { + pathrelse (&path); + reiserfs_warning ("vs-: reiserfs_delete_solid_item: %k not found", + key); + break; + } + if (!tb_init) { + tb_init = 1 ; + item_len = le16_to_cpu (PATH_PITEM_HEAD (&path)->ih_item_len); + init_tb_struct (th, &tb, th->t_super, &path, - (IH_SIZE + item_len)); + } + + retval = fix_nodes (M_DELETE, &tb, NULL, 0); + if (retval == REPEAT_SEARCH) + continue; + + if (retval == CARRY_ON) { + do_balance (&tb, 0, 0, M_DELETE); + break; + } + + // IO_ERROR, NO_DISK_SPACE, etc + reiserfs_warning ("vs-: reiserfs_delete_solid_item: " + "could not delete %K due to fix_nodes failure\n", &cpu_key); + unfix_nodes (&tb); + break; + } + + reiserfs_check_path(&path) ; +} + + +void reiserfs_delete_object (struct reiserfs_transaction_handle *th, struct inode * inode) +{ + inode->i_size = 0; + + /* for directory this deletes item containing "." and ".." */ + reiserfs_do_truncate (th, inode, NULL, 0/*no timestamp updates*/); + + /* delete stat data */ + /* this debug code needs to go away. Trying to find a truncate race + ** -- clm -- 4/1/2000 + */ +#if 0 + if (inode->i_nlink != 0) { + reiserfs_warning("clm-4001: deleting inode with link count==%d\n", inode->i_nlink) ; + } +#endif + reiserfs_delete_solid_item (th, INODE_PKEY (inode)); +} + + +static int maybe_indirect_to_direct (struct reiserfs_transaction_handle *th, + struct inode * p_s_inode, + struct page *page, + struct path * p_s_path, + struct cpu_key * p_s_item_key, + loff_t n_new_file_size, + char * p_c_mode + ) { + struct super_block * p_s_sb = p_s_inode->i_sb; + int n_block_size = p_s_sb->s_blocksize; + int cut_bytes; + + if (n_new_file_size != p_s_inode->i_size) + BUG (); + + /* the page being sent in could be NULL if there was an i/o error + ** reading in the last block. The user will hit problems trying to + ** read the file, but for now we just skip the indirect2direct + */ + if (atomic_read(&p_s_inode->i_count) > 1 || + !tail_has_to_be_packed (p_s_inode) || + !page || p_s_inode->u.reiserfs_i.nopack) { + // leave tail in an unformatted node + *p_c_mode = M_SKIP_BALANCING; + cut_bytes = n_block_size - (n_new_file_size & (n_block_size - 1)); + pathrelse(p_s_path); + return cut_bytes; + } + /* Permorm the conversion to a direct_item. */ + /*return indirect_to_direct (p_s_inode, p_s_path, p_s_item_key, n_new_file_size, p_c_mode);*/ + return indirect2direct (th, p_s_inode, page, p_s_path, p_s_item_key, n_new_file_size, p_c_mode); +} + + +/* we did indirect_to_direct conversion. And we have inserted direct + item successesfully, but there were no disk space to cut unfm + pointer being converted. Therefore we have to delete inserted + direct item(s) */ +static void indirect_to_direct_roll_back (struct reiserfs_transaction_handle *th, struct inode * inode, struct path * path) +{ + struct cpu_key tail_key; + int tail_len; + int removed; + + make_cpu_key (&tail_key, inode, inode->i_size + 1, TYPE_DIRECT, 4);// !!!! + tail_key.key_length = 4; + + tail_len = (cpu_key_k_offset (&tail_key) & (inode->i_sb->s_blocksize - 1)) - 1; + while (tail_len) { + /* look for the last byte of the tail */ + if (search_for_position_by_key (inode->i_sb, &tail_key, path) == POSITION_NOT_FOUND) + reiserfs_panic (inode->i_sb, "vs-5615: indirect_to_direct_roll_back: found invalid item"); +#ifdef CONFIG_REISERFS_CHECK + if (path->pos_in_item != PATH_PITEM_HEAD (path)->ih_item_len - 1) + reiserfs_panic (inode->i_sb, "vs-5616: indirect_to_direct_roll_back: appended bytes found"); +#endif + PATH_LAST_POSITION (path) --; + + removed = reiserfs_delete_item (th, path, &tail_key, inode, 0/*unbh not needed*/); +#ifdef CONFIG_REISERFS_CHECK + if (removed <= 0 || removed > tail_len) + reiserfs_panic (inode->i_sb, "vs-5617: indirect_to_direct_roll_back: " + "there was tail %d bytes, removed item length %d bytes", + tail_len, removed); +#endif + tail_len -= removed; + set_cpu_key_k_offset (&tail_key, cpu_key_k_offset (&tail_key) - removed); + } + printk ("indirect_to_direct_roll_back: indirect_to_direct conversion has been rolled back due to lack of disk space\n"); + //mark_file_without_tail (inode); + mark_inode_dirty (inode); +} + + +/* (Truncate or cut entry) or delete object item. Returns < 0 on failure */ +int reiserfs_cut_from_item (struct reiserfs_transaction_handle *th, + struct path * p_s_path, + struct cpu_key * p_s_item_key, + struct inode * p_s_inode, + struct page *page, + loff_t n_new_file_size) +{ + struct super_block * p_s_sb = p_s_inode->i_sb; + /* Every function which is going to call do_balance must first + create a tree_balance structure. Then it must fill up this + structure by using the init_tb_struct and fix_nodes functions. + After that we can make tree balancing. */ + struct tree_balance s_cut_balance; + int n_cut_size = 0, /* Amount to be cut. */ + n_ret_value = CARRY_ON, + n_removed = 0, /* Number of the removed unformatted nodes. */ + n_is_inode_locked = 0; + char c_mode; /* Mode of the balance. */ + int retval2 = -1; + + + init_tb_struct(th, &s_cut_balance, p_s_inode->i_sb, p_s_path, n_cut_size); + + + /* Repeat this loop until we either cut the item without needing + to balance, or we fix_nodes without schedule occuring */ + while ( 1 ) { + /* Determine the balance mode, position of the first byte to + be cut, and size to be cut. In case of the indirect item + free unformatted nodes which are pointed to by the cut + pointers. */ + + c_mode = prepare_for_delete_or_cut(th, p_s_inode, p_s_path, p_s_item_key, &n_removed, + &n_cut_size, n_new_file_size); + if ( c_mode == M_CONVERT ) { + /* convert last unformatted node to direct item or leave + tail in the unformatted node */ +#ifdef CONFIG_REISERFS_CHECK + if ( n_ret_value != CARRY_ON ) + reiserfs_panic (p_s_sb, "PAP-5570: reiserfs_cut_from_item: can not convert twice"); +#endif + + n_ret_value = maybe_indirect_to_direct (th, p_s_inode, page, p_s_path, p_s_item_key, + n_new_file_size, &c_mode); + if ( c_mode == M_SKIP_BALANCING ) + /* tail has been left in the unformatted node */ + return n_ret_value; + + n_is_inode_locked = 1; + + /* removing of last unformatted node will change value we + have to return to truncate. Save it */ + retval2 = n_ret_value; + /*retval2 = p_s_sb->s_blocksize - (n_new_file_size & (p_s_sb->s_blocksize - 1));*/ + + /* So, we have performed the first part of the conversion: + inserting the new direct item. Now we are removing the + last unformatted node pointer. Set key to search for + it. */ + set_cpu_key_k_type (p_s_item_key, TYPE_INDIRECT); + p_s_item_key->key_length = 4; + n_new_file_size -= (n_new_file_size & (p_s_sb->s_blocksize - 1)); + set_cpu_key_k_offset (p_s_item_key, n_new_file_size + 1); + if ( search_for_position_by_key(p_s_sb, p_s_item_key, p_s_path) == POSITION_NOT_FOUND ){ + print_block (PATH_PLAST_BUFFER (p_s_path), 3, PATH_LAST_POSITION (p_s_path) - 1, PATH_LAST_POSITION (p_s_path) + 1); + reiserfs_panic(p_s_sb, "PAP-5580: reiserfs_cut_from_item: item to convert does not exist (%k)", p_s_item_key); + } + continue; + } + if (n_cut_size == 0) { + pathrelse (p_s_path); + return 0; + } + + s_cut_balance.insert_size[0] = n_cut_size; + + n_ret_value = fix_nodes(c_mode, &s_cut_balance, NULL, 0); + if ( n_ret_value != REPEAT_SEARCH ) + break; + + n_ret_value = search_for_position_by_key(p_s_sb, p_s_item_key, p_s_path); + if (n_ret_value == POSITION_FOUND) + continue; + + reiserfs_warning ("PAP-5610: reiserfs_cut_from_item: item %K not found\n", p_s_item_key); + pathrelse (p_s_path); + return (n_ret_value == IO_ERROR) ? -EIO : -ENOENT; + } /* while */ + + // check fix_nodes results (IO_ERROR or NO_DISK_SPACE) + if ( n_ret_value != CARRY_ON ) { + if ( n_is_inode_locked ) { + // FIXME: this seems to be not needed: we are always able + // to cut item + indirect_to_direct_roll_back (th, p_s_inode, p_s_path); + } + if (n_ret_value == NO_DISK_SPACE) + reiserfs_warning (""); + unfix_nodes (&s_cut_balance); + return -EIO; + } + + /* go ahead and perform balancing */ + +#ifdef CONFIG_REISERFS_CHECK + if ( c_mode == M_PASTE || c_mode == M_INSERT ) + reiserfs_panic (p_s_sb, "PAP-5640: reiserfs_cut_from_item: illegal mode"); +#endif + + /* Calculate number of bytes that need to be cut from the item. */ + if (retval2 == -1) + n_ret_value = calc_deleted_bytes_number(&s_cut_balance, c_mode); + else + n_ret_value = retval2; + + if ( c_mode == M_DELETE ) { + struct item_head * p_le_ih = PATH_PITEM_HEAD (s_cut_balance.tb_path); + + if ( is_direct_le_ih (p_le_ih) && (le_ih_k_offset (p_le_ih) & (p_s_sb->s_blocksize - 1)) == 1 ) { + /* we delete first part of tail which was stored in direct + item(s) */ + // FIXME: this is to keep 3.5 happy + p_s_inode->u.reiserfs_i.i_first_direct_byte = U32_MAX; + p_s_inode->i_blocks -= p_s_sb->s_blocksize / 512; + } + } + +#ifdef CONFIG_REISERFS_CHECK + if (n_is_inode_locked) { + struct item_head * le_ih = PATH_PITEM_HEAD (s_cut_balance.tb_path); + /* we are going to complete indirect2direct conversion. Make + sure, that we exactly remove last unformatted node pointer + of the item */ + if (!is_indirect_le_ih (le_ih)) + reiserfs_panic (p_s_sb, "vs-5652: reiserfs_cut_from_item: " + "item must be indirect %h", le_ih); + + if (c_mode == M_DELETE && le16_to_cpu (le_ih->ih_item_len) != UNFM_P_SIZE) + reiserfs_panic (p_s_sb, "vs-5653: reiserfs_cut_from_item: " + "completing indirect2direct conversion indirect item %h" + "being deleted must be of 4 byte long", le_ih); + + if (c_mode == M_CUT && s_cut_balance.insert_size[0] != -UNFM_P_SIZE) { + reiserfs_panic (p_s_sb, "vs-5654: reiserfs_cut_from_item: " + "can not complete indirect2direct conversion of %h (CUT, insert_size==%d)", + le_ih, s_cut_balance.insert_size[0]); + } + /* it would be useful to make sure, that right neighboring + item is direct item of this file */ + } +#endif + + do_balance(&s_cut_balance, NULL, NULL, c_mode); + if ( n_is_inode_locked ) { + /* we've converted from indirect to direct, we must remove + ** ourselves from the list of pages that need flushing before + ** this transaction can commit + */ + reiserfs_remove_page_from_flush_list(th, p_s_inode) ; + p_s_inode->u.reiserfs_i.i_pack_on_close = 0 ; + } + return n_ret_value; +} + + +static void truncate_directory (struct reiserfs_transaction_handle *th, struct inode * inode) +{ + if (inode->i_nlink) + reiserfs_warning ("vs-5655: truncate_directory: link count != 0"); + + set_le_key_k_offset (ITEM_VERSION_1, INODE_PKEY (inode), DOT_OFFSET); + set_le_key_k_type (ITEM_VERSION_1, INODE_PKEY (inode), TYPE_DIRENTRY); + reiserfs_delete_solid_item (th, INODE_PKEY (inode)); + + set_le_key_k_offset (ITEM_VERSION_1, INODE_PKEY (inode), SD_OFFSET); + set_le_key_k_type (ITEM_VERSION_1, INODE_PKEY (inode), TYPE_STAT_DATA); +} + + + + +/* Truncate file to the new size. Note, this must be called with a transaction + already started */ +void reiserfs_do_truncate (struct reiserfs_transaction_handle *th, + struct inode * p_s_inode, /* ->i_size contains new + size */ + struct page *page, /* up to date for last block */ + int update_timestamps /* when it is called by + file_release to convert + the tail - no timestamps + should be updated */ + ) { + INITIALIZE_PATH (s_search_path); /* Path to the current object item. */ + struct item_head * p_le_ih; /* Pointer to an item header. */ + struct cpu_key s_item_key; /* Key to search for a previous file item. */ + loff_t n_file_size, /* Old file size. */ + n_new_file_size;/* New file size. */ + int n_deleted; /* Number of deleted or truncated bytes. */ + int retval; + + if ( ! (S_ISREG(p_s_inode->i_mode) || S_ISDIR(p_s_inode->i_mode) || S_ISLNK(p_s_inode->i_mode)) ) + return; + + if (S_ISDIR(p_s_inode->i_mode)) { + // deletion of directory - no need to update timestamps + truncate_directory (th, p_s_inode); + return; + } + + /* Get new file size. */ + n_new_file_size = p_s_inode->i_size; + + // FIXME: note, that key type is unimportant here + make_cpu_key (&s_item_key, p_s_inode, max_reiserfs_offset (p_s_inode), TYPE_DIRECT, 3); + + retval = search_for_position_by_key(p_s_inode->i_sb, &s_item_key, &s_search_path); + if (retval == IO_ERROR) { + reiserfs_warning ("vs-5657: reiserfs_do_truncate: " + "i/o failure occured trying to truncate %K\n", &s_item_key); + return; + } + if (retval == POSITION_FOUND || retval == FILE_NOT_FOUND) { + reiserfs_warning ("PAP-5660: reiserfs_do_truncate: " + "wrong result %d of search for %K\n", retval, &s_item_key); + return; + } + + s_search_path.pos_in_item --; + + /* Get real file size (total length of all file items) */ + p_le_ih = PATH_PITEM_HEAD(&s_search_path); + if ( is_statdata_le_ih (p_le_ih) ) + n_file_size = 0; + else { + loff_t offset = le_ih_k_offset (p_le_ih); + int bytes = op_bytes_number (p_le_ih,p_s_inode->i_sb->s_blocksize); + + /* this may mismatch with real file size: if last direct item + had no padding zeros and last unformatted node had no free + space, this file would have this file size */ + n_file_size = offset + bytes - 1; + } + + if ( n_file_size == 0 || n_file_size < n_new_file_size ) { + pathrelse(&s_search_path); + return; + } + /* Update key to search for the last file item. */ + set_cpu_key_k_offset (&s_item_key, n_file_size); + + do { + /* Cut or delete file item. */ + n_deleted = reiserfs_cut_from_item(th, &s_search_path, &s_item_key, p_s_inode, page, n_new_file_size); + if (n_deleted < 0) { + reiserfs_warning ("vs-5665: reiserfs_truncate_file: cut_from_item failed"); + reiserfs_check_path(&s_search_path) ; + return; + } + +#ifdef CONFIG_REISERFS_CHECK + if ( n_deleted > n_file_size ){ + reiserfs_panic (p_s_inode->i_sb, "PAP-5670: reiserfs_truncate_file: " + "reiserfs_truncate_file returns too big number: deleted %d, file_size %lu, item_key %k", + n_deleted, n_file_size, &s_item_key); + } +#endif + + /* Change key to search the last file item. */ + n_file_size -= n_deleted; + + set_cpu_key_k_offset (&s_item_key, n_file_size); + + /* While there are bytes to truncate and previous file item is presented in the tree. */ + + /* + ** This loop could take a really long time, and could log + ** many more blocks than a transaction can hold. So, we do a polite + ** journal end here, and if the transaction needs ending, we make + ** sure the file is consistent before ending the current trans + ** and starting a new one + */ + if (journal_transaction_should_end(th, th->t_blocks_allocated)) { + int orig_len_alloc = th->t_blocks_allocated ; + decrement_counters_in_path(&s_search_path) ; + + if (update_timestamps) { + p_s_inode->i_mtime = p_s_inode->i_ctime = CURRENT_TIME; + // FIXME: sd gets wrong size here + } + reiserfs_update_sd(th, p_s_inode) ; + + journal_end(th, p_s_inode->i_sb, orig_len_alloc) ; + journal_begin(th, p_s_inode->i_sb, orig_len_alloc) ; + } + } while ( n_file_size > ROUND_UP (n_new_file_size) && + search_for_position_by_key(p_s_inode->i_sb, &s_item_key, &s_search_path) == POSITION_FOUND ) ; + +#ifdef CONFIG_REISERFS_CHECK + if ( n_file_size > ROUND_UP (n_new_file_size) ) + reiserfs_panic (p_s_inode->i_sb, "PAP-5680: reiserfs_truncate_file: " + "truncate did not finish: new_file_size %Ld, current %Ld, oid %d\n", + n_new_file_size, n_file_size, s_item_key.on_disk_key.k_objectid); +#endif + + if (update_timestamps) { + // this is truncate, not file closing + p_s_inode->i_mtime = p_s_inode->i_ctime = CURRENT_TIME; + } + reiserfs_update_sd (th, p_s_inode); + + pathrelse(&s_search_path) ; +} + + +#ifdef CONFIG_REISERFS_CHECK +// this makes sure, that we __append__, not overwrite or add holes +static void check_research_for_paste (struct path * path, struct cpu_key * p_s_key) +{ + struct item_head * found_ih = get_ih (path); + + if (is_direct_le_ih (found_ih)) { + if (le_ih_k_offset (found_ih) + op_bytes_number (found_ih, get_bh (path)->b_size) != + cpu_key_k_offset (p_s_key) || + op_bytes_number (found_ih, get_bh (path)->b_size) != pos_in_item (path)) + reiserfs_panic (0, "PAP-5720: check_research_for_paste: " + "found direct item %h or position (%d) does not match to key %K", + found_ih, pos_in_item (path), p_s_key); + } + if (is_indirect_le_ih (found_ih)) { + if (le_ih_k_offset (found_ih) + op_bytes_number (found_ih, get_bh (path)->b_size) != cpu_key_k_offset (p_s_key) || + I_UNFM_NUM (found_ih) != pos_in_item (path) || + get_ih_free_space (found_ih) != 0) + reiserfs_panic (0, "PAP-5730: check_research_for_paste: " + "found indirect item (%h) or position (%d) does not match to key (%K)", + found_ih, pos_in_item (path), p_s_key); + } +} +#endif /* config reiserfs check */ + + +/* Paste bytes to the existing item. Returns bytes number pasted into the item. */ +int reiserfs_paste_into_item (struct reiserfs_transaction_handle *th, + struct path * p_s_search_path, /* Path to the pasted item. */ + struct cpu_key * p_s_key, /* Key to search for the needed item.*/ + const char * p_c_body, /* Pointer to the bytes to paste. */ + int n_pasted_size) /* Size of pasted bytes. */ +{ + struct tree_balance s_paste_balance; + int retval; + + init_tb_struct(th, &s_paste_balance, th->t_super, p_s_search_path, n_pasted_size); + + while ( (retval = fix_nodes(M_PASTE, &s_paste_balance, NULL, p_c_body)) == REPEAT_SEARCH ) { + /* file system changed while we were in the fix_nodes */ + retval = search_for_position_by_key (th->t_super, p_s_key, p_s_search_path); + if (retval == IO_ERROR) + return -EIO; + if (retval == POSITION_FOUND) { + reiserfs_warning ("PAP-5710: reiserfs_paste_into_item: entry or pasted byte (%K) exists", p_s_key); + pathrelse (p_s_search_path); + return -EEXIST; + } + +#ifdef CONFIG_REISERFS_CHECK + check_research_for_paste (p_s_search_path, p_s_key); +#endif + } + + /* Perform balancing after all resources are collected by fix_nodes, and + accessing them will not risk triggering schedule. */ + if ( retval == CARRY_ON ) { + do_balance(&s_paste_balance, NULL/*ih*/, p_c_body, M_PASTE); + return 0; + } + + unfix_nodes(&s_paste_balance); + return (retval == NO_DISK_SPACE) ? -ENOSPC : -EIO; +} + + +/* Insert new item into the buffer at the path. */ +int reiserfs_insert_item(struct reiserfs_transaction_handle *th, + struct path * p_s_path, /* Path to the inserteded item. */ + struct cpu_key * key, + struct item_head * p_s_ih, /* Pointer to the item header to insert.*/ + const char * p_c_body) /* Pointer to the bytes to insert. */ +{ + struct tree_balance s_ins_balance; + int retval; + + init_tb_struct(th, &s_ins_balance, th->t_super, p_s_path, IH_SIZE + p_s_ih->ih_item_len); + + /* + if (p_c_body == 0) + n_zeros_num = p_s_ih->ih_item_len; + */ + // le_key2cpu_key (&key, &(p_s_ih->ih_key)); + + while ( (retval = fix_nodes(M_INSERT, &s_ins_balance, p_s_ih, p_c_body)) == REPEAT_SEARCH) { + /* file system changed while we were in the fix_nodes */ + retval = search_item (th->t_super, key, p_s_path); + if (retval == IO_ERROR) + return -EIO; + + if (retval == ITEM_FOUND) { + reiserfs_warning ("PAP-5760: reiserfs_insert_item: " + "key %K already exists in the tree\n", key); + pathrelse (p_s_path); + return -EEXIST; + } + } + + /* make balancing after all resources will be collected at a time */ + if ( retval == CARRY_ON ) { + do_balance (&s_ins_balance, p_s_ih, p_c_body, M_INSERT); + return 0; + } + + unfix_nodes(&s_ins_balance); + return (retval == NO_DISK_SPACE) ? -ENOSPC : -EIO; +} + + + + diff --git a/fs/reiserfs/super.c b/fs/reiserfs/super.c new file mode 100644 index 000000000..9838a1dcd --- /dev/null +++ b/fs/reiserfs/super.c @@ -0,0 +1,879 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/module.h> +#include <linux/sched.h> +#include <asm/uaccess.h> +#include <linux/reiserfs_fs.h> +#include <linux/smp_lock.h> +#include <linux/locks.h> +#include <linux/init.h> + +#else + +#include "nokernel.h" +#include <stdlib.h> // for simple_strtoul + +#endif + +#define SUPPORT_OLD_FORMAT + +#define REISERFS_OLD_BLOCKSIZE 4096 +#define REISERFS_SUPER_MAGIC_STRING_OFFSET_NJ 20 + + +#if 0 +// this one is not used currently +inline void reiserfs_mark_buffer_dirty (struct buffer_head * bh, int flag) +{ + mark_buffer_dirty (bh, flag); +} +#endif + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +void reiserfs_write_super (struct super_block * s) +{ + + int dirty = 0 ; + lock_kernel() ; + if (!(s->s_flags & MS_RDONLY)) { + dirty = flush_old_commits(s, 1) ; + } + s->s_dirt = dirty; + unlock_kernel() ; +} + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +void reiserfs_write_super_lockfs (struct super_block * s) +{ + + int dirty = 0 ; + struct reiserfs_transaction_handle th ; + lock_kernel() ; + if (!(s->s_flags & MS_RDONLY)) { + journal_begin(&th, s, 1) ; + journal_mark_dirty(&th, s, SB_BUFFER_WITH_SB (s)); + reiserfs_block_writes(&th) ; + journal_end(&th, s, 1) ; + } + s->s_dirt = dirty; + unlock_kernel() ; +} + +void reiserfs_unlockfs(struct super_block *s) { + reiserfs_allow_writes(s) ; +} + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +/* there should be no suspected recipients already. True and cautious + bitmaps should not differ. We only have to free preserve list and + write both bitmaps */ +void reiserfs_put_super (struct super_block * s) +{ + int i; + struct reiserfs_transaction_handle th ; + + /* change file system state to current state if it was mounted with read-write permissions */ + if (!(s->s_flags & MS_RDONLY)) { + journal_begin(&th, s, 10) ; + reiserfs_prepare_for_journal(s, SB_BUFFER_WITH_SB(s), 1) ; + s->u.reiserfs_sb.s_rs->s_state = le16_to_cpu (s->u.reiserfs_sb.s_mount_state); + journal_mark_dirty(&th, s, SB_BUFFER_WITH_SB (s)); + } + + /* note, journal_release checks for readonly mount, and can decide not + ** to do a journal_end + */ + journal_release(&th, s) ; + + for (i = 0; i < SB_BMAP_NR (s); i ++) + brelse (SB_AP_BITMAP (s)[i]); + + reiserfs_kfree (SB_AP_BITMAP (s), sizeof (struct buffer_head *) * SB_BMAP_NR (s), s); + + brelse (SB_BUFFER_WITH_SB (s)); + + print_statistics (s); + + if (s->u.reiserfs_sb.s_kmallocs != 0) { + reiserfs_warning ("vs-2004: reiserfs_put_super: aloocated memory left %d\n", + s->u.reiserfs_sb.s_kmallocs); + } + + return; +} + +struct super_operations reiserfs_sops = +{ + read_inode: reiserfs_read_inode, + read_inode2: reiserfs_read_inode2, + write_inode: reiserfs_write_inode, + dirty_inode: reiserfs_dirty_inode, + delete_inode: reiserfs_delete_inode, + put_super: reiserfs_put_super, + write_super: reiserfs_write_super, + write_super_lockfs: reiserfs_write_super_lockfs, + unlockfs: reiserfs_unlockfs, + statfs: reiserfs_statfs, + remount_fs: reiserfs_remount, + +}; + +/* this was (ext2)parse_options */ +static int parse_options (char * options, unsigned long * mount_options, unsigned long * blocks) +{ + char * this_char; + char * value; + + *blocks = 0; + if (!options) + /* use default configuration: create tails, journaling on, no + conversion to newest format */ + return 1; + for (this_char = strtok (options, ","); this_char != NULL; this_char = strtok (NULL, ",")) { + if ((value = strchr (this_char, '=')) != NULL) + *value++ = 0; + if (!strcmp (this_char, "notail")) { + set_bit (NOTAIL, mount_options); + } else if (!strcmp (this_char, "conv")) { + // if this is set, we update super block such that + // the partition will not be mounable by 3.5.x anymore + set_bit (REISERFS_CONVERT, mount_options); + } else if (!strcmp (this_char, "noborder")) { + /* this is used for benchmarking + experimental variations, it is not + intended for users to use, only for + developers who want to casually + hack in something to test */ + set_bit (REISERFS_NO_BORDER, mount_options); + } else if (!strcmp (this_char, "no_unhashed_relocation")) { + set_bit (REISERFS_NO_UNHASHED_RELOCATION, mount_options); + } else if (!strcmp (this_char, "hashed_relocation")) { + set_bit (REISERFS_HASHED_RELOCATION, mount_options); + } else if (!strcmp (this_char, "test4")) { + set_bit (REISERFS_TEST4, mount_options); + } else if (!strcmp (this_char, "nolog")) { + reiserfs_warning("reiserfs: nolog mount option not supported yet\n"); + } else if (!strcmp (this_char, "replayonly")) { + set_bit (REPLAYONLY, mount_options); + } else if (!strcmp (this_char, "resize")) { + if (value && *value){ + *blocks = simple_strtoul (value, &value, 0); + } else { + printk("reiserfs: resize option requires a value\n"); + return 0; + } + } else if (!strcmp (this_char, "hash")) { + if (value && *value) { + /* if they specify any hash option, we force detection + ** to make sure they aren't using the wrong hash + */ + if (!strcmp(value, "rupasov")) { + set_bit (FORCE_RUPASOV_HASH, mount_options); + set_bit (FORCE_HASH_DETECT, mount_options); + } else if (!strcmp(value, "tea")) { + set_bit (FORCE_TEA_HASH, mount_options); + set_bit (FORCE_HASH_DETECT, mount_options); + } else if (!strcmp(value, "r5")) { + set_bit (FORCE_R5_HASH, mount_options); + set_bit (FORCE_HASH_DETECT, mount_options); + } else if (!strcmp(value, "detect")) { + set_bit (FORCE_HASH_DETECT, mount_options); + } else { + printk("reiserfs: invalid hash function specified\n") ; + return 0 ; + } + } else { + printk("reiserfs: hash option requires a value\n"); + return 0 ; + } + } else { + printk ("reiserfs: Unrecognized mount option %s\n", this_char); + return 0; + } + } + return 1; +} + + +int reiserfs_is_super(struct super_block *s) { + return (s->s_dev != 0 && s->s_op == &reiserfs_sops) ; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +int reiserfs_remount (struct super_block * s, int * flags, char * data) +{ + struct reiserfs_super_block * rs; + struct reiserfs_transaction_handle th ; + unsigned long blocks; + unsigned long mount_options; + + rs = SB_DISK_SUPER_BLOCK (s); + + if (!parse_options(data, &mount_options, &blocks)) + return 0; + + if(blocks) { + int rc = reiserfs_resize(s, blocks); + if (rc != 0) + return rc; + } + + if ((unsigned long)(*flags & MS_RDONLY) == (s->s_flags & MS_RDONLY)) { + /* there is nothing to do to remount read-only fs as read-only fs */ + return 0; + } + + if (*flags & MS_RDONLY) { + /* try to remount file system with read-only permissions */ + if (le16_to_cpu (rs->s_state) == REISERFS_VALID_FS || s->u.reiserfs_sb.s_mount_state != REISERFS_VALID_FS) { + return 0; + } + + journal_begin(&th, s, 10) ; + /* Mounting a rw partition read-only. */ + reiserfs_prepare_for_journal(s, SB_BUFFER_WITH_SB(s), 1) ; + rs->s_state = cpu_to_le16 (s->u.reiserfs_sb.s_mount_state); + journal_mark_dirty(&th, s, SB_BUFFER_WITH_SB (s)); + s->s_dirt = 0; + } else { + s->u.reiserfs_sb.s_mount_state = le16_to_cpu(rs->s_state) ; + s->s_flags &= ~MS_RDONLY ; /* now it is safe to call journal_begin */ + journal_begin(&th, s, 10) ; + + /* Mount a partition which is read-only, read-write */ + reiserfs_prepare_for_journal(s, SB_BUFFER_WITH_SB(s), 1) ; + s->u.reiserfs_sb.s_mount_state = le16_to_cpu (rs->s_state); + s->s_flags &= ~MS_RDONLY; + rs->s_state = cpu_to_le16 (REISERFS_ERROR_FS); + /* mark_buffer_dirty (SB_BUFFER_WITH_SB (s), 1); */ + journal_mark_dirty(&th, s, SB_BUFFER_WITH_SB (s)); + s->s_dirt = 0; + s->u.reiserfs_sb.s_mount_state = REISERFS_VALID_FS ; + } + /* this will force a full flush of all journal lists */ + SB_JOURNAL(s)->j_must_wait = 1 ; + journal_end(&th, s, 10) ; + return 0; +} + + +static int read_bitmaps (struct super_block * s) +{ + int i, bmp, dl ; + struct reiserfs_super_block * rs = SB_DISK_SUPER_BLOCK(s); + + SB_AP_BITMAP (s) = reiserfs_kmalloc (sizeof (struct buffer_head *) * le16_to_cpu (rs->s_bmap_nr), GFP_BUFFER, s); + if (SB_AP_BITMAP (s) == 0) + return 1; + memset (SB_AP_BITMAP (s), 0, sizeof (struct buffer_head *) * le16_to_cpu (rs->s_bmap_nr)); + + /* reiserfs leaves the first 64k unused so that any partition + labeling scheme currently used will have enough space. Then we + need one block for the super. -Hans */ + bmp = (REISERFS_DISK_OFFSET_IN_BYTES / s->s_blocksize) + 1; /* first of bitmap blocks */ + SB_AP_BITMAP (s)[0] = reiserfs_bread (s->s_dev, bmp, s->s_blocksize); + if(!SB_AP_BITMAP(s)[0]) + return 1; + for (i = 1, bmp = dl = rs->s_blocksize * 8; i < le16_to_cpu (rs->s_bmap_nr); i ++) { + SB_AP_BITMAP (s)[i] = reiserfs_bread (s->s_dev, bmp, s->s_blocksize); + if (!SB_AP_BITMAP (s)[i]) + return 1; + bmp += dl; + } + + return 0; +} + +static int read_old_bitmaps (struct super_block * s) +{ + int i ; + struct reiserfs_super_block * rs = SB_DISK_SUPER_BLOCK(s); + int bmp1 = (REISERFS_OLD_DISK_OFFSET_IN_BYTES / s->s_blocksize) + 1; /* first of bitmap blocks */ + + /* read true bitmap */ + SB_AP_BITMAP (s) = reiserfs_kmalloc (sizeof (struct buffer_head *) * le16_to_cpu (rs->s_bmap_nr), GFP_BUFFER, s); + if (SB_AP_BITMAP (s) == 0) + return 1; + + memset (SB_AP_BITMAP (s), 0, sizeof (struct buffer_head *) * le16_to_cpu (rs->s_bmap_nr)); + + for (i = 0; i < le16_to_cpu (rs->s_bmap_nr); i ++) { + SB_AP_BITMAP (s)[i] = reiserfs_bread (s->s_dev, bmp1 + i, s->s_blocksize); + if (!SB_AP_BITMAP (s)[i]) + return 1; + } + + return 0; +} + +void check_bitmap (struct super_block * s) +{ + int i = 0; + int free = 0; + char * buf; + + while (i < SB_BLOCK_COUNT (s)) { + buf = SB_AP_BITMAP (s)[i / (s->s_blocksize * 8)]->b_data; + if (!reiserfs_test_le_bit (i % (s->s_blocksize * 8), buf)) + free ++; + i ++; + } + + if (free != SB_FREE_BLOCKS (s)) + reiserfs_warning ("vs-4000: check_bitmap: %d free blocks, must be %d\n", + free, SB_FREE_BLOCKS (s)); +} + +#ifdef SUPPORT_OLD_FORMAT + +/* support old disk layout */ +static int read_old_super_block (struct super_block * s, int size) +{ + struct buffer_head * bh; + struct reiserfs_super_block * rs; + + printk("read_old_super_block: try to find super block in old location\n"); + /* there are only 4k-sized blocks in v3.5.10 */ + if (size != REISERFS_OLD_BLOCKSIZE) + set_blocksize(s->s_dev, REISERFS_OLD_BLOCKSIZE); + bh = bread (s->s_dev, + REISERFS_OLD_DISK_OFFSET_IN_BYTES / REISERFS_OLD_BLOCKSIZE, + REISERFS_OLD_BLOCKSIZE); + if (!bh) { + printk("read_old_super_block: unable to read superblock on dev %s\n", kdevname(s->s_dev)); + return 1; + } + + rs = (struct reiserfs_super_block *)bh->b_data; + if (strncmp (rs->s_magic, REISERFS_SUPER_MAGIC_STRING, strlen ( REISERFS_SUPER_MAGIC_STRING))) { + /* pre-journaling version check */ + if(!strncmp((char*)rs + REISERFS_SUPER_MAGIC_STRING_OFFSET_NJ, + REISERFS_SUPER_MAGIC_STRING, strlen(REISERFS_SUPER_MAGIC_STRING))) { + printk("read_old_super_blockr: a pre-journaling reiserfs filesystem isn't suitable there.\n"); + brelse(bh); + return 1; + } + + brelse (bh); + printk ("read_old_super_block: can't find a reiserfs filesystem on dev %s.\n", kdevname(s->s_dev)); + return 1; + } + + if(REISERFS_OLD_BLOCKSIZE != le16_to_cpu (rs->s_blocksize)) { + printk("read_old_super_block: blocksize mismatch, super block corrupted\n"); + brelse(bh); + return 1; + } + + s->s_blocksize = REISERFS_OLD_BLOCKSIZE; + s->s_blocksize_bits = 0; + while ((1 << s->s_blocksize_bits) != s->s_blocksize) + s->s_blocksize_bits ++; + + SB_BUFFER_WITH_SB (s) = bh; + SB_DISK_SUPER_BLOCK (s) = rs; + s->s_op = &reiserfs_sops; + return 0; +} +#endif + +// +// FIXME: mounting old filesystems we _must_ change magic string to +// make then unmountable by reiserfs of 3.5.x +// +static int read_super_block (struct super_block * s, int size) +{ + struct buffer_head * bh; + struct reiserfs_super_block * rs; + + bh = bread (s->s_dev, REISERFS_DISK_OFFSET_IN_BYTES / size, size); + if (!bh) { + printk("read_super_block: unable to read superblock on dev %s\n", kdevname(s->s_dev)); + return 1; + } + + rs = (struct reiserfs_super_block *)bh->b_data; + if (!is_reiserfs_magic_string (rs)) { + printk ("read_super_block: can't find a reiserfs filesystem on dev %s\n", + kdevname(s->s_dev)); + brelse (bh); + return 1; + } + + // + // ok, reiserfs signature (old or new) found in 64-th 1k block of + // the device + // + +#ifndef SUPPORT_OLD_FORMAT + // with SUPPORT_OLD_FORMAT undefined - detect old format by + // checking super block version + if (le16_to_cpu (rs->s_version) != REISERFS_VERSION_2) { + brelse (bh); + printk ("read_super_block: unsupported version (%d) of reiserfs found on dev %s\n", + le16_to_cpu (rs->s_version), kdevname(s->s_dev)); + return 1; + } +#endif + + s->s_blocksize = le16_to_cpu (rs->s_blocksize); + s->s_blocksize_bits = 0; + while ((1 << s->s_blocksize_bits) != s->s_blocksize) + s->s_blocksize_bits ++; + + brelse (bh); + + if (s->s_blocksize != size) + set_blocksize (s->s_dev, s->s_blocksize); + bh = reiserfs_bread (s->s_dev, REISERFS_DISK_OFFSET_IN_BYTES / s->s_blocksize, s->s_blocksize); + if (!bh) { + printk("read_super_block: unable to read superblock on dev %s\n", kdevname(s->s_dev)); + return 1; + } + + rs = (struct reiserfs_super_block *)bh->b_data; + if (!is_reiserfs_magic_string (rs) || + le16_to_cpu (rs->s_blocksize) != s->s_blocksize) { + brelse (bh); + printk ("read_super_block: can't find a reiserfs filesystem on dev %s.\n", kdevname(s->s_dev)); + return 1; + } + /* must check to be sure we haven't pulled an old format super out + ** of the old format's log. This is a kludge of a check, but it + ** will work. If block we've just read in is inside the + ** journal for that super, it can't be valid. + */ + if (bh->b_blocknr >= le32_to_cpu(rs->s_journal_block) && + bh->b_blocknr < (le32_to_cpu(rs->s_journal_block) + JOURNAL_BLOCK_COUNT)) { + brelse(bh) ; + printk("super-459: read_super_block: super found at block %lu is within its own log. " + "It must not be of this format type.\n", bh->b_blocknr) ; + return 1 ; + } + SB_BUFFER_WITH_SB (s) = bh; + SB_DISK_SUPER_BLOCK (s) = rs; + s->s_op = &reiserfs_sops; + return 0; +} + +/* after journal replay, reread all bitmap and super blocks */ +static int reread_meta_blocks(struct super_block *s) { + int i ; + ll_rw_block(READ, 1, &(SB_BUFFER_WITH_SB(s))) ; + wait_on_buffer(SB_BUFFER_WITH_SB(s)) ; + if (!buffer_uptodate(SB_BUFFER_WITH_SB(s))) { + printk("reread_meta_blocks, error reading the super\n") ; + return 1 ; + } + + for (i = 0; i < SB_BMAP_NR(s) ; i++) { + ll_rw_block(READ, 1, &(SB_AP_BITMAP(s)[i])) ; + wait_on_buffer(SB_AP_BITMAP(s)[i]) ; + if (!buffer_uptodate(SB_AP_BITMAP(s)[i])) { + printk("reread_meta_blocks, error reading bitmap block number %d at %ld\n", i, SB_AP_BITMAP(s)[i]->b_blocknr) ; + return 1 ; + } + } + return 0 ; + +} + + +///////////////////////////////////////////////////// +// hash detection stuff + + +// if root directory is empty - we set default - Yura's - hash and +// warn about it +// FIXME: we look for only one name in a directory. If tea and yura +// bith have the same value - we ask user to send report to the +// mailing list +__u32 find_hash_out (struct super_block * s) +{ + int retval; + struct inode * inode; + struct cpu_key key; + INITIALIZE_PATH (path); + struct reiserfs_dir_entry de; + __u32 hash = DEFAULT_HASH; + + inode = s->s_root->d_inode; + + while (1) { + make_cpu_key (&key, inode, ~0, TYPE_DIRENTRY, 3); + retval = search_by_entry_key (s, &key, &path, &de); + if (retval == IO_ERROR) { + pathrelse (&path); + return UNSET_HASH ; + } + if (retval == NAME_NOT_FOUND) + de.de_entry_num --; + set_de_name_and_namelen (&de); + if (le32_to_cpu (de.de_deh[de.de_entry_num].deh_offset) == DOT_DOT_OFFSET) { + /* allow override in this case */ + if (reiserfs_rupasov_hash(s)) { + hash = YURA_HASH ; + } + reiserfs_warning("reiserfs: FS seems to be empty, autodetect " + "is using the default hash\n"); + break; + } + if (GET_HASH_VALUE(yura_hash (de.de_name, de.de_namelen)) == + GET_HASH_VALUE(keyed_hash (de.de_name, de.de_namelen))) { + reiserfs_warning ("reiserfs: Could not detect hash function " + "please mount with -o hash={tea,rupasov,r5}\n") ; + hash = UNSET_HASH ; + break; + } + if (GET_HASH_VALUE(le32_to_cpu(de.de_deh[de.de_entry_num].deh_offset))== + GET_HASH_VALUE (yura_hash (de.de_name, de.de_namelen))) + hash = YURA_HASH; + else + hash = TEA_HASH; + break; + } + + pathrelse (&path); + return hash; +} + +// finds out which hash names are sorted with +static int what_hash (struct super_block * s) +{ + __u32 code; + + code = le32_to_cpu (s->u.reiserfs_sb.s_rs->s_hash_function_code); + + /* reiserfs_hash_detect() == true if any of the hash mount options + ** were used. We must check them to make sure the user isn't + ** using a bad hash value + */ + if (code == UNSET_HASH || reiserfs_hash_detect(s)) + code = find_hash_out (s); + + if (code != UNSET_HASH && reiserfs_hash_detect(s)) { + /* detection has found the hash, and we must check against the + ** mount options + */ + if (reiserfs_rupasov_hash(s) && code != YURA_HASH) { + printk("REISERFS: Error, tea hash detected, " + "unable to force rupasov hash\n") ; + code = UNSET_HASH ; + } else if (reiserfs_tea_hash(s) && code != TEA_HASH) { + printk("REISERFS: Error, rupasov hash detected, " + "unable to force tea hash\n") ; + code = UNSET_HASH ; + } else if (reiserfs_r5_hash(s) && code != R5_HASH) { + printk("REISERFS: Error, r5 hash detected, " + "unable to force r5 hash\n") ; + code = UNSET_HASH ; + } + } else { + /* find_hash_out was not called or could not determine the hash */ + if (reiserfs_rupasov_hash(s)) { + code = YURA_HASH ; + } else if (reiserfs_tea_hash(s)) { + code = TEA_HASH ; + } else if (reiserfs_r5_hash(s)) { + code = R5_HASH ; + } + } + + /* if we are mounted RW, and we have a new valid hash code, update + ** the super + */ + if (code != UNSET_HASH && + !(s->s_flags & MS_RDONLY) && + code != le32_to_cpu (s->u.reiserfs_sb.s_rs->s_hash_function_code)) { + s->u.reiserfs_sb.s_rs->s_hash_function_code = cpu_to_le32(code) ; + } + return code; +} + +// return pointer to appropriate function +static hashf_t hash_function (struct super_block * s) +{ + switch (what_hash (s)) { + case TEA_HASH: + reiserfs_warning ("Using tea hash to sort names\n"); + return keyed_hash; + case YURA_HASH: + reiserfs_warning ("Using rupasov hash to sort names\n"); + return yura_hash; + case R5_HASH: + reiserfs_warning ("Using r5 hash to sort names\n"); + return r5_hash; + } + return NULL; +} + +// this is used to set up correct value for old partitions +int function2code (hashf_t func) +{ + if (func == keyed_hash) + return TEA_HASH; + if (func == yura_hash) + return YURA_HASH; + if (func == r5_hash) + return R5_HASH; + + BUG() ; // should never happen + + return 0; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +struct super_block * reiserfs_read_super (struct super_block * s, void * data, int silent) +{ + int size; + struct inode *root_inode; + kdev_t dev = s->s_dev; + int j; + extern int *blksize_size[]; + struct reiserfs_transaction_handle th ; + int old_format = 0; + unsigned long blocks; + int jinit_done = 0 ; + struct reiserfs_iget4_args args ; + + + memset (&s->u.reiserfs_sb, 0, sizeof (struct reiserfs_sb_info)); + + if (parse_options ((char *) data, &(s->u.reiserfs_sb.s_mount_opt), &blocks) == 0) { + return NULL; + } + + if (blocks) { + printk("reserfs: resize option for remount only\n"); + return NULL; + } + + if (blksize_size[MAJOR(dev)] && blksize_size[MAJOR(dev)][MINOR(dev)] != 0) { + /* as blocksize is set for partition we use it */ + size = blksize_size[MAJOR(dev)][MINOR(dev)]; + } else { + size = BLOCK_SIZE; + set_blocksize (s->s_dev, BLOCK_SIZE); + } + + /* read block (64-th 1k block), which can contain reiserfs super block */ + if (read_super_block (s, size)) { +#ifdef SUPPORT_OLD_FORMAT + // try old format (undistributed bitmap, super block in 8-th 1k block of a device) + if(read_old_super_block(s,size)) + goto error; + else + old_format = 1; +#endif + goto error ; + } + + s->u.reiserfs_sb.s_mount_state = le16_to_cpu (SB_DISK_SUPER_BLOCK (s)->s_state); /* journal victim */ + s->u.reiserfs_sb.s_mount_state = REISERFS_VALID_FS ; + + if (old_format ? read_old_bitmaps(s) : read_bitmaps(s)) { + printk ("reiserfs_read_super: unable to read bitmap\n"); + goto error; + } +#ifdef CONFIG_REISERFS_CHECK + printk("reiserfs:warning: CONFIG_REISERFS_CHECK is set ON\n"); + printk("reiserfs:warning: - it is slow mode for debugging.\n"); +#endif + + // set_device_ro(s->s_dev, 1) ; + if (journal_init(s)) { + printk("reiserfs_read_super: unable to initialize journal space\n") ; + goto error ; + } else { + jinit_done = 1 ; /* once this is set, journal_release must be called + ** if we error out of the mount + */ + } + if (reread_meta_blocks(s)) { + printk("reiserfs_read_super: unable to reread meta blocks after journal init\n") ; + goto error ; + } + + if (replay_only (s)) + goto error; + + if (is_read_only(s->s_dev) && !(s->s_flags & MS_RDONLY)) { + printk("clm-7000: Detected readonly device, marking FS readonly\n") ; + s->s_flags |= MS_RDONLY ; + } + args.objectid = REISERFS_ROOT_PARENT_OBJECTID ; + root_inode = iget4 (s, REISERFS_ROOT_OBJECTID, 0, (void *)(&args)); + if (!root_inode) { + printk ("reiserfs_read_super: get root inode failed\n"); + goto error; + } + + s->s_root = d_alloc_root(root_inode); + if (!s->s_root) { + iput(root_inode); + goto error; + } + + // define and initialize hash function + s->u.reiserfs_sb.s_hash_function = hash_function (s); + if (s->u.reiserfs_sb.s_hash_function == NULL) { + dput(s->s_root) ; + s->s_root = NULL ; + goto error ; + } + + if (!(s->s_flags & MS_RDONLY)) { + struct reiserfs_super_block * rs = SB_DISK_SUPER_BLOCK (s); + + journal_begin(&th, s, 1) ; + reiserfs_prepare_for_journal(s, SB_BUFFER_WITH_SB(s), 1) ; + + rs->s_state = cpu_to_le16 (REISERFS_ERROR_FS); + + if (strncmp (rs->s_magic, REISER2FS_SUPER_MAGIC_STRING, + strlen ( REISER2FS_SUPER_MAGIC_STRING))) { + if (le16_to_cpu(rs->s_version) != 0) + BUG (); + // filesystem created under 3.5.x found + if (!old_format_only (s)) { + reiserfs_warning("reiserfs: converting 3.5.x filesystem to the new format\n") ; + // after this 3.5.x will not be able to mount this partition + memcpy (rs->s_magic, REISER2FS_SUPER_MAGIC_STRING, + sizeof (REISER2FS_SUPER_MAGIC_STRING)); + + reiserfs_convert_objectid_map_v1(s) ; + } else { + reiserfs_warning("reiserfs: using 3.5.x disk format\n") ; + } + } else { + // new format found + set_bit (REISERFS_CONVERT, &(s->u.reiserfs_sb.s_mount_opt)); + } + + // mark hash in super block: it could be unset. overwrite should be ok + rs->s_hash_function_code = cpu_to_le32 (function2code (s->u.reiserfs_sb.s_hash_function)); + + journal_mark_dirty(&th, s, SB_BUFFER_WITH_SB (s)); + journal_end(&th, s, 1) ; + s->s_dirt = 0; + } else { + struct reiserfs_super_block * rs = SB_DISK_SUPER_BLOCK (s); + if (strncmp (rs->s_magic, REISER2FS_SUPER_MAGIC_STRING, + strlen ( REISER2FS_SUPER_MAGIC_STRING))) { + reiserfs_warning("reiserfs: using 3.5.x disk format\n") ; + } + } + + init_waitqueue_head (&(s->u.reiserfs_sb.s_wait)); + + printk("%s\n", reiserfs_get_version_string()) ; + return s; + + error: + if (jinit_done) { /* kill the commit thread, free journal ram */ + journal_release_error(NULL, s) ; + } + if (SB_DISK_SUPER_BLOCK (s)) { + for (j = 0; j < SB_BMAP_NR (s); j ++) { + if (SB_AP_BITMAP (s)) + brelse (SB_AP_BITMAP (s)[j]); + } + if (SB_AP_BITMAP (s)) + reiserfs_kfree (SB_AP_BITMAP (s), sizeof (struct buffer_head *) * SB_BMAP_NR (s), s); + } + if (SB_BUFFER_WITH_SB (s)) + brelse(SB_BUFFER_WITH_SB (s)); + + return NULL; +} + + +// +// a portion of this function, particularly the VFS interface portion, +// was derived from minix or ext2's analog and evolved as the +// prototype did. You should be able to tell which portion by looking +// at the ext2 code and comparing. It's subfunctions contain no code +// used as a template unless they are so labeled. +// +int reiserfs_statfs (struct super_block * s, struct statfs * buf) +{ + struct reiserfs_super_block * rs = SB_DISK_SUPER_BLOCK (s); + + /* changed to accomodate gcc folks.*/ + buf->f_type = REISERFS_SUPER_MAGIC; + buf->f_bsize = le32_to_cpu (s->s_blocksize); + buf->f_blocks = le32_to_cpu (rs->s_block_count) - le16_to_cpu (rs->s_bmap_nr) - 1; + buf->f_bfree = le32_to_cpu (rs->s_free_blocks); + buf->f_bavail = buf->f_bfree; + buf->f_files = -1; + buf->f_ffree = -1; + buf->f_namelen = (REISERFS_MAX_NAME_LEN (s->s_blocksize)); + return 0; +} + +#ifdef __KERNEL__ + +static DECLARE_FSTYPE_DEV(reiserfs_fs_type,"reiserfs",reiserfs_read_super); + +// +// this is exactly what 2.3.99-pre9's init_ext2_fs is +// +static int __init init_reiserfs_fs (void) +{ + return register_filesystem(&reiserfs_fs_type); +} + +EXPORT_NO_SYMBOLS; + +// +// this is exactly what 2.3.99-pre9's init_ext2_fs is +// +static void __exit exit_reiserfs_fs(void) +{ + unregister_filesystem(&reiserfs_fs_type); +} + +module_init(init_reiserfs_fs) ; +module_exit(exit_reiserfs_fs) ; + +#endif + + + diff --git a/fs/reiserfs/tail_conversion.c b/fs/reiserfs/tail_conversion.c new file mode 100644 index 000000000..25da132b6 --- /dev/null +++ b/fs/reiserfs/tail_conversion.c @@ -0,0 +1,297 @@ +/* + * Copyright 1999 Hans Reiser, see reiserfs/README for licensing and copyright details + */ + +#ifdef __KERNEL__ + +#include <linux/config.h> +#include <linux/sched.h> +#include <linux/pagemap.h> +#include <linux/reiserfs_fs.h> +#include <linux/locks.h> + +#else + +#include "nokernel.h" + +#endif + + +/* access to tail : when one is going to read tail it must make sure, that is not running. + direct2indirect and indirect2direct can not run concurrently */ + + +/* Converts direct items to an unformatted node. Panics if file has no + tail. -ENOSPC if no disk space for conversion */ +/* path points to first direct item of the file regarless of how many of + them are there */ +int direct2indirect (struct reiserfs_transaction_handle *th, struct inode * inode, + struct path * path, struct buffer_head * unbh, + loff_t tail_offset) +{ + struct super_block * sb = inode->i_sb; + struct buffer_head *up_to_date_bh ; + struct item_head * p_le_ih = PATH_PITEM_HEAD (path); + struct cpu_key end_key; /* Key to search for the last byte of the + converted item. */ + struct item_head ind_ih; /* new indirect item to be inserted or + key of unfm pointer to be pasted */ + int n_blk_size, + n_retval; /* returned value for reiserfs_insert_item and clones */ + struct unfm_nodeinfo unfm_ptr; /* Handle on an unformatted node + that will be inserted in the + tree. */ + + + sb->u.reiserfs_sb.s_direct2indirect ++; + + n_blk_size = sb->s_blocksize; + + /* and key to search for append or insert pointer to the new + unformatted node. */ + copy_item_head (&ind_ih, p_le_ih); + set_le_ih_k_offset (&ind_ih, tail_offset); + set_le_ih_k_type (&ind_ih, TYPE_INDIRECT); + + /* Set the key to search for the place for new unfm pointer */ + make_cpu_key (&end_key, inode, tail_offset, TYPE_INDIRECT, 4); + + // FIXME: we could avoid this + if ( search_for_position_by_key (sb, &end_key, path) == POSITION_FOUND ) + reiserfs_panic (sb, "PAP-14030: direct2indirect: " + "pasted or inserted byte exists in the tree"); + + p_le_ih = PATH_PITEM_HEAD (path); + + unfm_ptr.unfm_nodenum = cpu_to_le32 (unbh->b_blocknr); + unfm_ptr.unfm_freespace = 0; // ??? + + if ( is_statdata_le_ih (p_le_ih) ) { + /* Insert new indirect item. */ + set_ih_free_space (&ind_ih, 0); /* delete at nearest future */ + ind_ih.ih_item_len = cpu_to_le16 (UNFM_P_SIZE); + PATH_LAST_POSITION (path)++; + n_retval = reiserfs_insert_item (th, path, &end_key, &ind_ih, + (char *)&unfm_ptr); + } else { + /* Paste into last indirect item of an object. */ + n_retval = reiserfs_paste_into_item(th, path, &end_key, + (char *)&unfm_ptr, UNFM_P_SIZE); + } + if ( n_retval ) { + return n_retval; + } + + // note: from here there are two keys which have matching first + // three key components. They only differ by the fourth one. + + + /* Set the key to search for the direct items of the file */ + make_cpu_key (&end_key, inode, max_reiserfs_offset (inode), TYPE_DIRECT, 4); + + /* Move bytes from the direct items to the new unformatted node + and delete them. */ + while (1) { + int item_len, first_direct; + + /* end_key.k_offset is set so, that we will always have found + last item of the file */ + if ( search_for_position_by_key (sb, &end_key, path) == POSITION_FOUND ) + reiserfs_panic (sb, "PAP-14050: direct2indirect: " + "direct item (%k) not found", &end_key); + p_le_ih = PATH_PITEM_HEAD (path); +#ifdef CONFIG_REISERFS_CHECK + if (!is_direct_le_ih (p_le_ih)) + reiserfs_panic (sb, "vs-14055: direct2indirect: " + "direct item expected, found %h", p_le_ih); +#endif + if ((le_ih_k_offset (p_le_ih) & (n_blk_size - 1)) == 1) + first_direct = 1; + else + first_direct = 0; + item_len = le16_to_cpu (p_le_ih->ih_item_len); + + /* we only send the unbh pointer if the buffer is not up to date. + ** this avoids overwriting good data from writepage() with old data + ** from the disk or buffer cache + */ + if (buffer_uptodate(unbh) || Page_Uptodate(unbh->b_page)) { + up_to_date_bh = NULL ; + } else { + up_to_date_bh = unbh ; + } + n_retval = reiserfs_delete_item (th, path, &end_key, inode, + up_to_date_bh) ; + + if (first_direct && item_len == n_retval) + // done: file does not have direct items anymore + break; + + } + + inode->u.reiserfs_i.i_first_direct_byte = U32_MAX; + + return 0; +} + + +/* stolen from fs/buffer.c */ +void reiserfs_unmap_buffer(struct buffer_head *bh) { + if (buffer_mapped(bh)) { + if (buffer_journaled(bh) || buffer_journal_dirty(bh)) { + BUG() ; + } + mark_buffer_clean(bh) ; + wait_on_buffer(bh) ; + // clear_bit(BH_Uptodate, &bh->b_state) ; + clear_bit(BH_Mapped, &bh->b_state) ; + clear_bit(BH_Req, &bh->b_state) ; + clear_bit(BH_New, &bh->b_state) ; + } +} + +static void +unmap_buffers(struct page *page, loff_t pos) { + struct buffer_head *bh ; + struct buffer_head *head ; + struct buffer_head *next ; + unsigned long tail_index ; + unsigned long cur_index ; + + if (page) { + if (page->buffers) { + tail_index = pos & (PAGE_CACHE_SIZE - 1) ; + cur_index = 0 ; + head = page->buffers ; + bh = head ; + do { + next = bh->b_this_page ; + + /* we want to unmap the buffers that contain the tail, and + ** all the buffers after it (since the tail must be at the + ** end of the file). We don't want to unmap file data + ** before the tail, since it might be dirty and waiting to + ** reach disk + */ + cur_index += bh->b_size ; + if (cur_index > tail_index) { + reiserfs_unmap_buffer(bh) ; + } + bh = next ; + } while (bh != head) ; + } + } +} + +/* this first locks inode (neither reads nor sync are permitted), + reads tail through page cache, insert direct item. When direct item + inserted successfully inode is left locked. Return value is always + what we expect from it (number of cut bytes). But when tail remains + in the unformatted node, we set mode to SKIP_BALANCING and unlock + inode */ +int indirect2direct (struct reiserfs_transaction_handle *th, + struct inode * p_s_inode, + struct page *page, + struct path * p_s_path, /* path to the indirect item. */ + struct cpu_key * p_s_item_key, /* Key to look for unformatted node pointer to be cut. */ + loff_t n_new_file_size, /* New file size. */ + char * p_c_mode) +{ + struct super_block * p_s_sb = p_s_inode->i_sb; + struct item_head s_ih; + unsigned long n_block_size = p_s_sb->s_blocksize; + char * tail; + int tail_len, round_tail_len; + loff_t pos, pos1; /* position of first byte of the tail */ + struct cpu_key key; + + p_s_sb->u.reiserfs_sb.s_indirect2direct ++; + + *p_c_mode = M_SKIP_BALANCING; + + /* store item head path points to. */ + copy_item_head (&s_ih, PATH_PITEM_HEAD(p_s_path)); + + tail_len = (n_new_file_size & (n_block_size - 1)); + if (!old_format_only (p_s_sb)) + round_tail_len = ROUND_UP (tail_len); + else + round_tail_len = tail_len; + + pos = le_ih_k_offset (&s_ih) - 1 + (le16_to_cpu (s_ih.ih_item_len) / UNFM_P_SIZE - 1) * p_s_sb->s_blocksize; + pos1 = pos; + + // we are protected by i_sem. The tail can not disapper, not + // append can be done either + // we are in truncate or packing tail in file_release + + tail = (char *)kmap(page) ; /* this can schedule */ + + if (path_changed (&s_ih, p_s_path)) { + /* re-search indirect item */ + if ( search_for_position_by_key (p_s_sb, p_s_item_key, p_s_path) == POSITION_NOT_FOUND ) + reiserfs_panic(p_s_sb, "PAP-5520: indirect2direct: " + "item to be converted %k does not exist", p_s_item_key); + copy_item_head(&s_ih, PATH_PITEM_HEAD(p_s_path)); +#ifdef CONFIG_REISERFS_CHECK + pos = le_ih_k_offset (&s_ih) - 1 + + (le16_to_cpu (s_ih.ih_item_len) / UNFM_P_SIZE - 1) * p_s_sb->s_blocksize; + if (pos != pos1) + reiserfs_panic (p_s_sb, "vs-5530: indirect2direct: " + "tail position changed while we were reading it"); +#endif + } + + + /* Set direct item header to insert. */ + make_le_item_head (&s_ih, 0, inode_items_version (p_s_inode), pos1 + 1, + TYPE_DIRECT, round_tail_len, 0xffff/*ih_free_space*/); + + /* we want a pointer to the first byte of the tail in the page. + ** the page was locked and this part of the page was up to date when + ** indirect2direct was called, so we know the bytes are still valid + */ + tail = tail + (pos & (PAGE_CACHE_SIZE - 1)) ; + + PATH_LAST_POSITION(p_s_path)++; + + key = *p_s_item_key; + set_cpu_key_k_type (&key, TYPE_DIRECT); + key.key_length = 4; + /* Insert tail as new direct item in the tree */ + if ( reiserfs_insert_item(th, p_s_path, &key, &s_ih, + tail ? tail : NULL) < 0 ) { + /* No disk memory. So we can not convert last unformatted node + to the direct item. In this case we used to adjust + indirect items's ih_free_space. Now ih_free_space is not + used, it would be ideal to write zeros to corresponding + unformatted node. For now i_size is considered as guard for + going out of file size */ + kunmap(page) ; + return n_block_size - round_tail_len; + } + kunmap(page) ; + + /* this will invalidate all the buffers in the page after + ** pos1 + */ + unmap_buffers(page, pos1) ; + + // note: we have now the same as in above direct2indirect + // conversion: there are two keys which have matching first three + // key components. They only differ by the fouhth one. + + /* We have inserted new direct item and must remove last + unformatted node. */ + p_s_inode->i_blocks += (p_s_sb->s_blocksize / 512); + *p_c_mode = M_CUT; + + /* we store position of first direct item in the in-core inode */ + //mark_file_with_tail (p_s_inode, pos1 + 1); + p_s_inode->u.reiserfs_i.i_first_direct_byte = pos1 + 1; + + return n_block_size - round_tail_len; +} + + + diff --git a/fs/reiserfs/version.c b/fs/reiserfs/version.c new file mode 100644 index 000000000..361884e80 --- /dev/null +++ b/fs/reiserfs/version.c @@ -0,0 +1,7 @@ +/* + * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README + */ + +char *reiserfs_get_version_string(void) { + return "ReiserFS version 3.6.25" ; +} diff --git a/fs/super.c b/fs/super.c index 1c0b48c11..6262c7675 100644 --- a/fs/super.c +++ b/fs/super.c @@ -54,6 +54,7 @@ extern void wait_for_keypress(void); extern int root_mountflags; static int do_remount_sb(struct super_block *sb, int flags, char * data); +static int do_remount(const char *dir, int flags, char * data); /* this is initialized in init/main.c */ kdev_t ROOT_DEV; @@ -1025,15 +1026,12 @@ static int do_umount(struct vfsmount *mnt, int umount_root, int flags) * call reboot(9). Then init(8) could umount root and exec /reboot. */ if (mnt == current->fs->rootmnt && !umount_root) { - int retval = 0; /* * Special case for "unmounting" root ... * we just try to remount it readonly. */ mntput(mnt); - if (!(sb->s_flags & MS_RDONLY)) - retval = do_remount_sb(sb, MS_RDONLY, 0); - return retval; + return do_remount("/", MS_RDONLY, NULL); } spin_lock(&dcache_lock); diff --git a/fs/udf/file.c b/fs/udf/file.c index 9ab57d370..02aa4f025 100644 --- a/fs/udf/file.c +++ b/fs/udf/file.c @@ -86,6 +86,7 @@ static int udf_adinicb_writepage(struct page *page) brelse(bh); SetPageUptodate(page); kunmap(page); + UnlockPage(page); return 0; } diff --git a/fs/udf/inode.c b/fs/udf/inode.c index 3783b43fd..8c818666c 100644 --- a/fs/udf/inode.c +++ b/fs/udf/inode.c @@ -203,7 +203,6 @@ void udf_expand_file_adinicb(struct inode * inode, int newsize, int * err) udf_release_data(bh); inode->i_data.a_ops->writepage(page); - UnlockPage(page); page_cache_release(page); mark_inode_dirty(inode); diff --git a/include/asm-alpha/errno.h b/include/asm-alpha/errno.h index 724f03b77..fd5b8fafc 100644 --- a/include/asm-alpha/errno.h +++ b/include/asm-alpha/errno.h @@ -139,4 +139,6 @@ #define ENOMEDIUM 129 /* No medium found */ #define EMEDIUMTYPE 130 /* Wrong medium type */ +#define EHASHCOLLISION 131 /* Number of hash collisons exceeds maximum generation counter value. */ + #endif diff --git a/include/asm-alpha/unistd.h b/include/asm-alpha/unistd.h index 4a5b98607..265c1b2dc 100644 --- a/include/asm-alpha/unistd.h +++ b/include/asm-alpha/unistd.h @@ -572,7 +572,6 @@ static inline long sync(void) return sys_sync(); } -extern long sys_wait4(int, int *, int, struct rusage *); static inline pid_t waitpid(int pid, int * wait_stat, int flags) { return sys_wait4(pid, wait_stat, flags, NULL); diff --git a/include/asm-i386/bugs.h b/include/asm-i386/bugs.h index 4e77e5d8a..257f1f879 100644 --- a/include/asm-i386/bugs.h +++ b/include/asm-i386/bugs.h @@ -76,26 +76,23 @@ static void __init check_fpu(void) } /* Enable FXSR and company _before_ testing for FP problems. */ -#if defined(CONFIG_X86_FXSR) || defined(CONFIG_X86_RUNTIME_FXSR) /* * Verify that the FXSAVE/FXRSTOR data will be 16-byte aligned. */ - if (offsetof(struct task_struct, thread.i387.fxsave) & 15) - panic("Kernel compiled for PII/PIII+ with FXSR, data not 16-byte aligned!"); - + if (offsetof(struct task_struct, thread.i387.fxsave) & 15) { + extern void __buggy_fxsr_alignment(void); + __buggy_fxsr_alignment(); + } if (cpu_has_fxsr) { printk(KERN_INFO "Enabling fast FPU save and restore... "); set_in_cr4(X86_CR4_OSFXSR); printk("done.\n"); } -#endif -#ifdef CONFIG_X86_XMM if (cpu_has_xmm) { printk(KERN_INFO "Enabling unmasked SIMD FPU exception support... "); set_in_cr4(X86_CR4_OSXMMEXCPT); printk("done.\n"); } -#endif /* Test for the divl bug.. */ __asm__("fninit\n\t" @@ -203,14 +200,6 @@ static void __init check_config(void) && (boot_cpu_data.x86_mask < 6 || boot_cpu_data.x86_mask == 11)) panic("Kernel compiled for PMMX+, assumes a local APIC without the read-before-write bug!"); #endif - -/* - * If we configured ourselves for FXSR, we'd better have it. - */ -#ifdef CONFIG_X86_FXSR - if (!cpu_has_fxsr) - panic("Kernel compiled for PII/PIII+, requires FXSR feature!"); -#endif } static void __init check_bugs(void) diff --git a/include/asm-i386/errno.h b/include/asm-i386/errno.h index 7cf599f4d..d22c4472b 100644 --- a/include/asm-i386/errno.h +++ b/include/asm-i386/errno.h @@ -128,5 +128,6 @@ #define ENOMEDIUM 123 /* No medium found */ #define EMEDIUMTYPE 124 /* Wrong medium type */ +#define EHASHCOLLISION 125 /* Number of hash collisons exceeds maximum generation counter value. */ #endif diff --git a/include/asm-i386/i387.h b/include/asm-i386/i387.h index 04ba635e5..1cf8dc2ab 100644 --- a/include/asm-i386/i387.h +++ b/include/asm-i386/i387.h @@ -23,6 +23,10 @@ extern void init_fpu(void); extern void save_init_fpu( struct task_struct *tsk ); extern void restore_fpu( struct task_struct *tsk ); +extern void kernel_fpu_begin(void); +#define kernel_fpu_end() stts() + + #define unlazy_fpu( tsk ) do { \ if ( tsk->flags & PF_USEDFPU ) \ save_init_fpu( tsk ); \ @@ -50,10 +54,8 @@ extern void set_fpu_twd( struct task_struct *tsk, unsigned short twd ); extern void set_fpu_mxcsr( struct task_struct *tsk, unsigned short mxcsr ); #define load_mxcsr( val ) do { \ - if ( cpu_has_xmm ) { \ - unsigned long __mxcsr = ((unsigned long)(val) & 0xffff); \ - asm volatile( "ldmxcsr %0" : : "m" (__mxcsr) ); \ - } \ + unsigned long __mxcsr = ((unsigned long)(val) & 0xffbf); \ + asm volatile( "ldmxcsr %0" : : "m" (__mxcsr) ); \ } while (0) /* diff --git a/include/asm-i386/pgtable.h b/include/asm-i386/pgtable.h index bf32a7449..a25f3bcfd 100644 --- a/include/asm-i386/pgtable.h +++ b/include/asm-i386/pgtable.h @@ -140,7 +140,11 @@ extern unsigned long empty_zero_page[1024]; #define VMALLOC_START (((unsigned long) high_memory + 2*VMALLOC_OFFSET-1) & \ ~(VMALLOC_OFFSET-1)) #define VMALLOC_VMADDR(x) ((unsigned long)(x)) -#define VMALLOC_END (FIXADDR_START) +#if CONFIG_HIGHMEM +# define VMALLOC_END (PKMAP_BASE-2*PAGE_SIZE) +#else +# define VMALLOC_END (FIXADDR_START-2*PAGE_SIZE) +#endif /* * The 4MB page is guessing.. Detailed in the infamous "Chapter H" diff --git a/include/asm-i386/system.h b/include/asm-i386/system.h index d3b01ab8b..52e24682e 100644 --- a/include/asm-i386/system.h +++ b/include/asm-i386/system.h @@ -267,15 +267,8 @@ static inline unsigned long __cmpxchg(volatile void *ptr, unsigned long old, * I expect future Intel CPU's to have a weaker ordering, * but I'd also expect them to finally get their act together * and add some real memory barriers if so. - * - * The Pentium III does add a real memory barrier with the - * sfence instruction, so we use that where appropriate. */ -#ifndef CONFIG_X86_XMM #define mb() __asm__ __volatile__ ("lock; addl $0,0(%%esp)": : :"memory") -#else -#define mb() __asm__ __volatile__ ("sfence": : :"memory") -#endif #define rmb() mb() #define wmb() __asm__ __volatile__ ("": : :"memory") diff --git a/include/asm-mips/errno.h b/include/asm-mips/errno.h index 0763d0e99..736774592 100644 --- a/include/asm-mips/errno.h +++ b/include/asm-mips/errno.h @@ -1,11 +1,9 @@ /* - * include/asm-mips/errno.h - * * This file is subject to the terms and conditions of the GNU General Public * License. See the file "COPYING" in the main directory of this archive * for more details. * - * Copyright (C) 1995, 1999 by Ralf Baechle + * Copyright (C) 1995, 1999, 2001 by Ralf Baechle */ #ifndef _ASM_ERRNO_H #define _ASM_ERRNO_H @@ -143,6 +141,7 @@ */ #define ENOMEDIUM 159 /* No medium found */ #define EMEDIUMTYPE 160 /* Wrong medium type */ +#define EHASHCOLLISION 161 /* Number of hash collisons exceeds maximum generation counter value. */ #define EDQUOT 1133 /* Quota exceeded */ diff --git a/include/asm-mips64/errno.h b/include/asm-mips64/errno.h index b2f507f4f..7864a4039 100644 --- a/include/asm-mips64/errno.h +++ b/include/asm-mips64/errno.h @@ -1,10 +1,10 @@ -/* $Id: errno.h,v 1.1 1999/08/18 23:37:51 ralf Exp $ - * +/* * This file is subject to the terms and conditions of the GNU General Public * License. See the file "COPYING" in the main directory of this archive * for more details. * - * Copyright (C) 1995 by Ralf Baechle + * Copyright (C) 1995, 2001 by Ralf Baechle + * Copyright (C) 2001 Silicon Graphics, Inc. */ #ifndef _ASM_ERRNO_H #define _ASM_ERRNO_H @@ -142,6 +142,7 @@ */ #define ENOMEDIUM 159 /* No medium found */ #define EMEDIUMTYPE 160 /* Wrong medium type */ +#define EHASHCOLLISION 125 /* Number of hash collisons exceeds maximum generation counter value. */ #define EDQUOT 1133 /* Quota exceeded */ diff --git a/include/asm-ppc/delay.h b/include/asm-ppc/delay.h index 2116a2f2c..8cc1fde24 100644 --- a/include/asm-ppc/delay.h +++ b/include/asm-ppc/delay.h @@ -2,6 +2,8 @@ #ifndef _PPC_DELAY_H #define _PPC_DELAY_H +#include <asm/param.h> + /* * Copyright 1996, Paul Mackerras. * @@ -11,25 +13,38 @@ * 2 of the License, or (at your option) any later version. */ -extern unsigned long loops_per_sec; +extern unsigned long loops_per_jiffy; -extern __inline__ void __delay(unsigned int loops) -{ - if (loops != 0) - __asm__ __volatile__("mtctr %0; 1: bdnz 1b" : : - "r" (loops) : "ctr"); -} +/* maximum permitted argument to udelay */ +#define __MAX_UDELAY 1000000 -extern __inline__ void udelay(unsigned long usecs) +extern void __delay(unsigned int loops); + +/* N.B. the `secs' parameter here is a fixed-point number with + the binary point to the left of the most-significant bit. */ +extern __inline__ void __const_udelay(unsigned int secs) { - unsigned long loops; + unsigned int loops; - /* compute (usecs * 2^32 / 10^6) * loops_per_sec / 2^32 */ - usecs *= 0x10c6; /* 2^32 / 10^6 */ __asm__("mulhwu %0,%1,%2" : "=r" (loops) : - "r" (usecs), "r" (loops_per_sec)); - __delay(loops); + "r" (secs), "r" (loops_per_jiffy)); + __delay(loops * HZ); } +/* + * note that 4294 == 2^32 / 10^6, multiplying by 4294 converts from + * microseconds to a 32-bit fixed-point number of seconds. + */ +extern __inline__ void __udelay(unsigned int usecs) +{ + __const_udelay(usecs * 4294); +} + +extern void __bad_udelay(void); /* deliberately undefined */ + +#define udelay(n) (__builtin_constant_p(n)? \ + ((n) > __MAX_UDELAY? __bad_udelay(): __const_udelay((n) * 4294u)) : \ + __udelay(n)) + #endif /* defined(_PPC_DELAY_H) */ #endif /* __KERNEL__ */ diff --git a/include/asm-ppc/dma.h b/include/asm-ppc/dma.h index 735d91d1f..56f56cd0a 100644 --- a/include/asm-ppc/dma.h +++ b/include/asm-ppc/dma.h @@ -102,13 +102,13 @@ extern unsigned long ISA_DMA_THRESHOLD; /* used in nasty hack for sound - see prep_setup_arch() -- Cort */ extern long ppc_cs4232_dma, ppc_cs4232_dma2; #if defined(CONFIG_CS4232) -#if defined(CONFIG_PREP) || defined(CONFIG_ALL_PPC) +#if defined(CONFIG_ALL_PPC) #define SND_DMA1 ppc_cs4232_dma #define SND_DMA2 ppc_cs4232_dma2 -#else /* !CONFIG_PREP && !CONFIG_ALL_PPC */ +#else /* !CONFIG_ALL_PPC */ #define SND_DMA1 -1 #define SND_DMA2 -1 -#endif /* !CONFIG_PREP */ +#endif /* CONFIG_ALL_PPC */ #elif defined(CONFIG_MSS) #define SND_DMA1 CONFIG_MSS_DMA #define SND_DMA2 CONFIG_MSS_DMA2 @@ -201,40 +201,8 @@ static __inline__ void release_dma_lock(unsigned long flags) /* enable/disable a specific DMA channel */ static __inline__ void enable_dma(unsigned int dmanr) { - /* - * The Radstone PPC2 and PPC2a boards have inverted DREQ - * lines (active low) so each command needs to be logically - * ORed with 0x40 - */ unsigned char ucDmaCmd=0x00; -#if defined(CONFIG_PREP) || defined(CONFIG_ALL_PPC) - if(_prep_type==_PREP_Radstone) - { - switch(ucSystemType) - { - case RS_SYS_TYPE_PPC2: - case RS_SYS_TYPE_PPC2a: - case RS_SYS_TYPE_PPC2ep: - { - /* - * DREQ lines are active low - */ - ucDmaCmd=0x40; - break; - } - - default: - { - /* - * DREQ lines are active high - */ - break; - } - } - } -#endif /* CONFIG_PREP || CONFIG_ALL_PPC */ - if (dmanr != 4) { dma_outb(0, DMA2_MASK_REG); /* This may not be enabled */ diff --git a/include/asm-ppc/elf.h b/include/asm-ppc/elf.h index 6a0e2e874..85124797a 100644 --- a/include/asm-ppc/elf.h +++ b/include/asm-ppc/elf.h @@ -70,5 +70,24 @@ typedef elf_vrreg_t elf_vrregset_t[ELF_NVRREG]; #define SET_PERSONALITY(ex, ibcs2) set_personality((ibcs2)?PER_SVR4:PER_LINUX) +/* + * We need to put in some extra aux table entries to tell glibc what + * the cache block size is, so it can use the dcbz instruction safely. + */ +#define AT_DCACHEBSIZE 17 +#define AT_ICACHEBSIZE 18 +#define AT_UCACHEBSIZE 19 + +extern int dcache_bsize; +extern int icache_bsize; +extern int ucache_bsize; + +#define DLINFO_EXTRA_ITEMS 3 +#define EXTRA_DLINFO do { \ + NEW_AUX_ENT(0, AT_DCACHEBSIZE, dcache_bsize); \ + NEW_AUX_ENT(1, AT_ICACHEBSIZE, icache_bsize); \ + NEW_AUX_ENT(2, AT_UCACHEBSIZE, ucache_bsize); \ +} while (0) + #endif /* __KERNEL__ */ #endif diff --git a/include/asm-ppc/feature.h b/include/asm-ppc/feature.h index 7a33ea8d3..9e9f831d5 100644 --- a/include/asm-ppc/feature.h +++ b/include/asm-ppc/feature.h @@ -86,6 +86,8 @@ extern void feature_set_usb_power(struct device_node* device, int power); extern void feature_set_firewire_power(struct device_node* device, int power); +extern void feature_core99_kick_cpu1(void); + /* * Sleep related functions. At term, they should be high-priority notifiers */ diff --git a/include/asm-ppc/hardirq.h b/include/asm-ppc/hardirq.h index 8a270c8c1..66aa6abeb 100644 --- a/include/asm-ppc/hardirq.h +++ b/include/asm-ppc/hardirq.h @@ -47,7 +47,7 @@ typedef struct { #include <asm/atomic.h> extern unsigned char global_irq_holder; -extern unsigned volatile int global_irq_lock; +extern unsigned volatile long global_irq_lock; extern atomic_t global_irq_count; static inline void release_irqlock(int cpu) @@ -66,8 +66,8 @@ static inline void hardirq_enter(int cpu) ++local_irq_count(cpu); atomic_inc(&global_irq_count); while (test_bit(0,&global_irq_lock)) { - if (smp_processor_id() == global_irq_holder) { - printk("uh oh, interrupt while we hold global irq lock!\n"); + if (cpu == global_irq_holder) { + printk("uh oh, interrupt while we hold global irq lock! (CPU %d)\n", cpu); #ifdef CONFIG_XMON xmon(0); #endif diff --git a/include/asm-ppc/hw_irq.h b/include/asm-ppc/hw_irq.h index 7d4790171..54e4d7111 100644 --- a/include/asm-ppc/hw_irq.h +++ b/include/asm-ppc/hw_irq.h @@ -28,17 +28,23 @@ extern void __no_use_set_lost(unsigned long); #define __cli() int_control.int_cli() #define __sti() int_control.int_sti() -#define __save_flags(flags) int_control.int_save_flags(&flags) -#define __restore_flags(flags) int_control.int_restore_flags(flags) +#define __save_flags(flags) int_control.int_save_flags((unsigned long *)&flags) +#define __restore_flags(flags) int_control.int_restore_flags((unsigned long)flags) #define __save_and_cli(flags) ({__save_flags(flags);__cli();}) -#define __set_lost(irq) ({ if ((ulong)int_control.int_set_lost) int_control.int_set_lost(irq); }) +#define __set_lost(irq) ({ if ((unsigned long)int_control.int_set_lost) int_control.int_set_lost(irq); }) extern void do_lost_interrupts(unsigned long); -extern atomic_t ppc_n_lost_interrupts; #define mask_irq(irq) ({if (irq_desc[irq].handler && irq_desc[irq].handler->disable) irq_desc[irq].handler->disable(irq);}) #define unmask_irq(irq) ({if (irq_desc[irq].handler && irq_desc[irq].handler->enable) irq_desc[irq].handler->enable(irq);}) -#define mask_and_ack_irq(irq) ({if (irq_desc[irq].handler && irq_desc[irq].handler->ack) irq_desc[irq].handler->ack(irq);}) +#define ack_irq(irq) ({if (irq_desc[irq].handler && irq_desc[irq].handler->ack) irq_desc[irq].handler->ack(irq);}) + +/* Should we handle this via lost interrupts and IPIs or should we don't care like + * we do now ? --BenH. + */ +struct hw_interrupt_type; +static inline void hw_resend_irq(struct hw_interrupt_type *h, unsigned int i) {} + #endif /* _PPC_HW_IRQ_H */ #endif /* __KERNEL__ */ diff --git a/include/asm-ppc/ide.h b/include/asm-ppc/ide.h index 179bdec67..a9e3baf0d 100644 --- a/include/asm-ppc/ide.h +++ b/include/asm-ppc/ide.h @@ -37,6 +37,8 @@ extern ide_ioreg_t chrp_idedma_regbase; /* one for both channels */ extern unsigned int chrp_ide_irq; extern void chrp_ide_probe(void); +extern void ppc_generic_ide_fix_driveid(struct hd_driveid *id); + struct ide_machdep_calls { void (*insw)(ide_ioreg_t port, void *buf, int ns); void (*outsw)(ide_ioreg_t port, void *buf, int ns); @@ -90,10 +92,9 @@ static __inline__ int ide_default_irq(ide_ioreg_t base) static __inline__ ide_ioreg_t ide_default_io_base(int index) { - if ( ppc_ide_md.default_io_base ) + if (ppc_ide_md.default_io_base) return ppc_ide_md.default_io_base(index); - else - return -1; + return 0; } static __inline__ void ide_init_hwif_ports(hw_regs_t *hw, @@ -124,10 +125,9 @@ static __inline__ void ide_init_default_hwifs(void) static __inline__ int ide_check_region (ide_ioreg_t from, unsigned int extent) { - if ( ppc_ide_md.ide_check_region ) + if (ppc_ide_md.ide_check_region) return ppc_ide_md.ide_check_region(from, extent); - else - return -1; + return 0; } static __inline__ void ide_request_region (ide_ioreg_t from, unsigned int extent, const char *name) @@ -148,19 +148,6 @@ static __inline__ void ide_fix_driveid (struct hd_driveid *id) ppc_ide_md.fix_driveid(id); } -#if 0 /* inb/outb from io.h is OK now -- paulus */ -#undef inb -#define inb(port) in_8((unsigned char *)((port) + ppc_ide_md.io_base)) -#undef inb_p -#define inb_p(port) inb(port) - -#undef outb -#define outb(val, port) \ - out_8((unsigned char *)((port) + ppc_ide_md.io_base), (val) ) -#undef outb_p -#define outb_p(val, port) outb(val, port) -#endif - typedef union { unsigned all : 8; /* all of the bits together */ struct { diff --git a/include/asm-ppc/io.h b/include/asm-ppc/io.h index 03f9db09a..602363511 100644 --- a/include/asm-ppc/io.h +++ b/include/asm-ppc/io.h @@ -222,20 +222,6 @@ extern inline void * bus_to_virt(unsigned long address) } /* - * The PCI bus bridge can translate addresses issued by the processor(s) - * into a different address on the PCI bus. On 32-bit cpus, we assume - * this mapping is 1-1, but on 64-bit systems it often isn't. - */ -#ifndef CONFIG_PPC64BRIDGE -#define phys_to_bus(x) (x) -#define bus_to_phys(x) (x) - -#else -extern unsigned long phys_to_bus(unsigned long pa); -extern unsigned long bus_to_phys(unsigned int ba, int busnr); -#endif /* CONFIG_PPC64BRIDGE */ - -/* * Change virtual addresses to physical addresses and vv, for * addresses in the area where the kernel has the RAM mapped. */ @@ -364,6 +350,13 @@ out: return retval; } +/* Make some pcmcia drivers happy */ +static inline int isa_check_signature(unsigned long io_addr, + const unsigned char *signature, int length) +{ + return 0; +} + /* Nothing to do */ #define dma_cache_inv(_start,_size) do { } while (0) diff --git a/include/asm-ppc/ioctls.h b/include/asm-ppc/ioctls.h index 0f879c163..e700d29fd 100644 --- a/include/asm-ppc/ioctls.h +++ b/include/asm-ppc/ioctls.h @@ -9,6 +9,7 @@ #define FIONBIO _IOW('f', 126, int) #define FIONREAD _IOR('f', 127, int) #define TIOCINQ FIONREAD +#define FIOQSIZE _IOR('f', 128, loff_t) #define TIOCGETP _IOR('t', 8, struct sgttyb) #define TIOCSETP _IOW('t', 9, struct sgttyb) diff --git a/include/asm-ppc/irq.h b/include/asm-ppc/irq.h index c3ae15763..139dfea0c 100644 --- a/include/asm-ppc/irq.h +++ b/include/asm-ppc/irq.h @@ -4,6 +4,7 @@ #include <linux/config.h> #include <asm/machdep.h> /* ppc_md */ +#include <asm/atomic.h> extern void disable_irq(unsigned int); extern void disable_irq_nosync(unsigned int); @@ -163,9 +164,6 @@ extern irq_node_t *new_irq_node(void); #ifndef CONFIG_8260 #define NUM_8259_INTERRUPTS 16 -#define IRQ_8259_CASCADE 16 -#define openpic_to_irq(n) ((n)+NUM_8259_INTERRUPTS) -#define irq_to_openpic(n) ((n)-NUM_8259_INTERRUPTS) #else /* CONFIG_8260 */ @@ -214,7 +212,10 @@ static __inline__ int irq_cannonicalize(int irq) #endif #define NR_MASK_WORDS ((NR_IRQS + 31) / 32) -extern unsigned int ppc_lost_interrupts[NR_MASK_WORDS]; +/* pendatic: these are long because they are used with set_bit --RR */ +extern unsigned long ppc_cached_irq_mask[NR_MASK_WORDS]; +extern unsigned long ppc_lost_interrupts[NR_MASK_WORDS]; +extern atomic_t ppc_n_lost_interrupts; #endif /* _ASM_IRQ_H */ #endif /* __KERNEL__ */ diff --git a/include/asm-ppc/keylargo.h b/include/asm-ppc/keylargo.h index 5408262a1..02d3d58b2 100644 --- a/include/asm-ppc/keylargo.h +++ b/include/asm-ppc/keylargo.h @@ -20,10 +20,19 @@ #define KEYLARGO_GPIO_CNT 17 /* Specific GPIO regs */ + #define KL_GPIO_ETH_PHY_RESET (KEYLARGO_GPIO_0+0x10) #define KL_GPIO_ETH_PHY_RESET_ASSERT 0x04 #define KL_GPIO_ETH_PHY_RESET_RELEASE 0x05 #define KL_GPIO_ETH_PHY_RESET_TRISTATE 0x00 + +#define KL_GPIO_KICK_CPU1 (KEYLARGO_GPIO_0+0x0a) +#define KL_GPIO_KICK_CPU1_UP 0x04 +#define KL_GPIO_KICK_CPU1_DOWN 0x38 + +#define KL_GPIO_PMU_MESSAGE_IRQ (KEYLARGO_GPIO_EXTINT_0+0x09) +#define KL_GPIO_PMU_MESSAGE_BIT 0x02 + /* * Bits in feature control register */ diff --git a/include/asm-ppc/linux_logo.h b/include/asm-ppc/linux_logo.h index c35e6db5b..d3aa2d267 100644 --- a/include/asm-ppc/linux_logo.h +++ b/include/asm-ppc/linux_logo.h @@ -34,9 +34,6 @@ extern unsigned char linux_logo_green[]; extern unsigned char linux_logo_blue[]; extern unsigned char linux_logo[]; extern unsigned char linux_logo_bw[]; -extern unsigned char linux_logo16_red[]; -extern unsigned char linux_logo16_green[]; -extern unsigned char linux_logo16_blue[]; extern unsigned char linux_logo16[]; #endif diff --git a/include/asm-ppc/machdep.h b/include/asm-ppc/machdep.h index 67111965f..ca254d0e7 100644 --- a/include/asm-ppc/machdep.h +++ b/include/asm-ppc/machdep.h @@ -10,6 +10,7 @@ struct pt_regs; struct pci_bus; +struct pci_dev; struct machdep_calls { void (*setup_arch)(void); @@ -45,7 +46,7 @@ struct machdep_calls { unsigned char (*nvram_read_val)(int addr); void (*nvram_write_val)(int addr, unsigned char val); -/* Tons of keyboard stuff. */ + /* Tons of keyboard stuff. */ int (*kbd_setkeycode)(unsigned int scancode, unsigned int keycode); int (*kbd_getkeycode)(unsigned int scancode); @@ -59,25 +60,25 @@ struct machdep_calls { unsigned char *ppc_kbd_sysrq_xlate; #endif - /* PCI interfaces */ - int (*pcibios_read_config_byte)(unsigned char bus, - unsigned char dev_fn, unsigned char offset, unsigned char *val); - int (*pcibios_read_config_word)(unsigned char bus, - unsigned char dev_fn, unsigned char offset, unsigned short *val); - int (*pcibios_read_config_dword)(unsigned char bus, - unsigned char dev_fn, unsigned char offset, unsigned int *val); - int (*pcibios_write_config_byte)(unsigned char bus, - unsigned char dev_fn, unsigned char offset, unsigned char val); - int (*pcibios_write_config_word)(unsigned char bus, - unsigned char dev_fn, unsigned char offset, unsigned short val); - int (*pcibios_write_config_dword)(unsigned char bus, - unsigned char dev_fn, unsigned char offset, unsigned int val); + /* + * optional PCI "hooks" + */ + + /* Called after scanning the bus, before allocating + * resources + */ void (*pcibios_fixup)(void); - void (*pcibios_fixup_bus)(struct pci_bus *); - void* (*pci_dev_io_base)(unsigned char bus, unsigned char devfn, int physical); - void* (*pci_dev_mem_base)(unsigned char bus, unsigned char devfn); - int (*pci_dev_root_bridge)(unsigned char bus, unsigned char devfn); + /* Called for each PCI bus in the system + * when it's probed + */ + void (*pcibios_fixup_bus)(struct pci_bus *); + + /* Called when pci_enable_device() is called (initial=0) or + * when a device with no assigned resource is found (initial=1). + * Returns 0 to allow assignement/enabling of the device + */ + int (*pcibios_enable_device_hook)(struct pci_dev *, int initial); /* this is for modules, since _machine can be a define -- Cort */ int ppc_machine; diff --git a/include/asm-ppc/mman.h b/include/asm-ppc/mman.h index 64abf0c58..1c0dbe205 100644 --- a/include/asm-ppc/mman.h +++ b/include/asm-ppc/mman.h @@ -13,6 +13,7 @@ #define MAP_ANONYMOUS 0x20 /* don't use a file */ #define MAP_RENAME MAP_ANONYMOUS /* In SunOS terminology */ #define MAP_NORESERVE 0x40 /* don't reserve swap pages */ +#define MAP_LOCKED 0x80 #define MAP_GROWSDOWN 0x0100 /* stack-like segment */ #define MAP_DENYWRITE 0x0800 /* ETXTBSY */ diff --git a/include/asm-ppc/mmu.h b/include/asm-ppc/mmu.h index 3efe39d00..976ea6cc9 100644 --- a/include/asm-ppc/mmu.h +++ b/include/asm-ppc/mmu.h @@ -6,12 +6,13 @@ #ifndef _PPC_MMU_H_ #define _PPC_MMU_H_ -/* Default "unsigned long" context */ -typedef unsigned long mm_context_t; - #include <linux/config.h> #ifndef __ASSEMBLY__ + +/* Default "unsigned long" context */ +typedef unsigned long mm_context_t; + /* Hardware Page Table Entry */ typedef struct _PTE { #ifdef CONFIG_PPC64BRIDGE diff --git a/include/asm-ppc/mmu_context.h b/include/asm-ppc/mmu_context.h index fed474d14..f8ef4e604 100644 --- a/include/asm-ppc/mmu_context.h +++ b/include/asm-ppc/mmu_context.h @@ -56,13 +56,6 @@ extern void mmu_context_overflow(void); */ extern void set_context(int context, void *pgd); -#ifdef CONFIG_8xx -extern inline void mmu_context_overflow(void) -{ - atomic_set(&next_mmu_context, -1); -} -#endif - /* * Get a new mmu context for task tsk if necessary. */ diff --git a/include/asm-ppc/parport.h b/include/asm-ppc/parport.h new file mode 100644 index 000000000..11f96d3de --- /dev/null +++ b/include/asm-ppc/parport.h @@ -0,0 +1,18 @@ +/* + * parport.h: platform-specific PC-style parport initialisation + * + * Copyright (C) 1999, 2000 Tim Waugh <tim@cyberelk.demon.co.uk> + * + * This file should only be included by drivers/parport/parport_pc.c. + */ + +#ifndef _ASM_PPC_PARPORT_H +#define _ASM_PPC_PARPORT_H + +static int __devinit parport_pc_find_isa_ports (int autoirq, int autodma); +static int __devinit parport_pc_find_nonpci_ports (int autoirq, int autodma) +{ + return parport_pc_find_isa_ports (autoirq, autodma); +} + +#endif /* !(_ASM_PPC_PARPORT_H) */ diff --git a/include/asm-ppc/pci-bridge.h b/include/asm-ppc/pci-bridge.h index 9e5385e51..a6c8d8919 100644 --- a/include/asm-ppc/pci-bridge.h +++ b/include/asm-ppc/pci-bridge.h @@ -2,49 +2,70 @@ #ifndef _ASM_PCI_BRIDGE_H #define _ASM_PCI_BRIDGE_H -void pmac_find_bridges(void); +struct device_node; +struct pci_controller; /* * pci_io_base returns the memory address at which you can access * the I/O space for PCI bus number `bus' (or NULL on error). - * - * NOTE: This doesn't handle the new Uni-N chip which requires - * per-device io_base. */ -void *pci_io_base(unsigned int bus); - -/* This version handles the new Uni-N host bridge, the iobase is now - * a per-device thing. I also added the memory base so PReP can - * be fixed to return 0xc0000000 (I didn't actually implement it) - * - * pci_dev_io_base() returns either a virtual (ioremap'ed) address or - * a physical address. In-kernel clients will use logical while the - * sys_pciconfig_iobase syscall returns a physical one to userland. +extern void *pci_bus_io_base(unsigned int bus); +extern unsigned long pci_bus_io_base_phys(unsigned int bus); +extern unsigned long pci_bus_mem_base_phys(unsigned int bus); + +/* + * PCI <-> OF matching functions */ -void *pci_dev_io_base(unsigned char bus, unsigned char devfn, int physical); -void *pci_dev_mem_base(unsigned char bus, unsigned char devfn); +extern int pci_device_from_OF_node(struct device_node *node, + u8* bus, u8* devfn); +extern struct device_node* pci_device_to_OF_node(struct pci_dev *); -/* Returns the root-bridge number (Uni-N number) of a device */ -int pci_dev_root_bridge(unsigned char bus, unsigned char devfn); +/* Get the PCI host controller for a bus */ +extern struct pci_controller* pci_bus_to_hose(int bus); + +/* Get the PCI host controller for an OF device */ +extern struct pci_controller* +pci_find_hose_for_OF_device(struct device_node* node); /* - * pci_device_loc returns the bus number and device/function number - * for a device on a PCI bus, given its device_node struct. - * It returns 0 if OK, -1 on error. + * Structure of a PCI controller (host bridge) */ -int pci_device_loc(struct device_node *dev, unsigned char *bus_ptr, - unsigned char *devfn_ptr); +struct pci_controller { + struct pci_controller *next; + struct pci_bus *bus; + void *arch_data; + + int first_busno; + int last_busno; + + void *io_base_virt; + unsigned long io_base_phys; + + /* Some machines (PReP) have a non 1:1 mapping of + * the PCI memory space in the CPU bus space + */ + unsigned long pci_mem_offset; -struct bridge_data { + struct pci_ops *ops; volatile unsigned int *cfg_addr; volatile unsigned char *cfg_data; - void *io_base; /* virtual */ - unsigned long io_base_phys; - int bus_number; - int max_bus; - struct bridge_data *next; - struct device_node *node; + + /* Currently, we limit ourselves to 1 IO range and 3 mem + * ranges since the common pci_bus structure can't handle more + */ + struct resource io_resource; + struct resource mem_resources[3]; + int mem_resource_count; }; +/* These are used for config access before all the PCI probing + has been done. */ +int early_read_config_byte(struct pci_controller *hose, int bus, int dev_fn, int where, u8 *val); +int early_read_config_word(struct pci_controller *hose, int bus, int dev_fn, int where, u16 *val); +int early_read_config_dword(struct pci_controller *hose, int bus, int dev_fn, int where, u32 *val); +int early_write_config_byte(struct pci_controller *hose, int bus, int dev_fn, int where, u8 val); +int early_write_config_word(struct pci_controller *hose, int bus, int dev_fn, int where, u16 val); +int early_write_config_dword(struct pci_controller *hose, int bus, int dev_fn, int where, u32 val); + #endif #endif /* __KERNEL__ */ diff --git a/include/asm-ppc/pci.h b/include/asm-ppc/pci.h index 1a661f050..a1cfa7d31 100644 --- a/include/asm-ppc/pci.h +++ b/include/asm-ppc/pci.h @@ -6,13 +6,11 @@ #define IOBASE_BRIDGE_NUMBER 0 #define IOBASE_MEMORY 1 #define IOBASE_IO 2 +#define IOBASE_ISA_IO 3 +#define IOBASE_ISA_MEM 4 -/* Can be used to override the logic in pci_scan_bus for skipping - * already-configured bus numbers - to be used for buggy BIOSes - * or architectures with incomplete PCI setup by the loader. - */ -#define pcibios_assign_all_busses() 0 +extern int pcibios_assign_all_busses(void); #define PCIBIOS_MIN_IO 0x1000 #define PCIBIOS_MIN_MEM 0x10000000 @@ -27,6 +25,18 @@ extern inline void pcibios_penalize_isa_irq(int irq) /* We don't do dynamic PCI IRQ allocation */ } +extern unsigned long pci_resource_to_bus(struct pci_dev *pdev, struct resource *res); + +/* + * The PCI bus bridge can translate addresses issued by the processor(s) + * into a different address on the PCI bus. On 32-bit cpus, we assume + * this mapping is 1-1, but on 64-bit systems it often isn't. + * + * Obsolete ! Drivers should now use pci_resource_to_bus + */ +extern unsigned long pci_phys_to_bus(unsigned long pa, int busnr); +extern unsigned long pci_bus_to_phys(unsigned int ba, int busnr); + /* Dynamic DMA Mapping stuff * ++ajoshi */ diff --git a/include/asm-ppc/pgtable.h b/include/asm-ppc/pgtable.h index 0f4579cd7..353f5c9bf 100644 --- a/include/asm-ppc/pgtable.h +++ b/include/asm-ppc/pgtable.h @@ -17,22 +17,22 @@ extern void local_flush_tlb_mm(struct mm_struct *mm); extern void local_flush_tlb_page(struct vm_area_struct *vma, unsigned long vmaddr); extern void local_flush_tlb_range(struct mm_struct *mm, unsigned long start, unsigned long end); -extern inline void flush_hash_page(unsigned context, unsigned long va) +static inline void flush_hash_page(unsigned context, unsigned long va) { } #elif defined(CONFIG_8xx) #define __tlbia() asm volatile ("tlbia" : : ) -extern inline void local_flush_tlb_all(void) +static inline void local_flush_tlb_all(void) { __tlbia(); } -extern inline void local_flush_tlb_mm(struct mm_struct *mm) +static inline void local_flush_tlb_mm(struct mm_struct *mm) { __tlbia(); } -extern inline void local_flush_tlb_page(struct vm_area_struct *vma, +static inline void local_flush_tlb_page(struct vm_area_struct *vma, unsigned long vmaddr) { __tlbia(); } -extern inline void local_flush_tlb_range(struct mm_struct *mm, +static inline void local_flush_tlb_range(struct mm_struct *mm, unsigned long start, unsigned long end) { __tlbia(); } -extern inline void flush_hash_page(unsigned context, unsigned long va) +static inline void flush_hash_page(unsigned context, unsigned long va) { } #else struct mm_struct; @@ -49,7 +49,7 @@ extern void local_flush_tlb_range(struct mm_struct *mm, unsigned long start, #define flush_tlb_page local_flush_tlb_page #define flush_tlb_range local_flush_tlb_range -extern inline void flush_tlb_pgtables(struct mm_struct *mm, +static inline void flush_tlb_pgtables(struct mm_struct *mm, unsigned long start, unsigned long end) { /* PPC has hw page tables. */ @@ -323,9 +323,9 @@ extern pte_t * __bad_pagetable(void); * setup: the pgd is never bad, and a pmd always exists (as it's folded * into the pgd entry) */ -extern inline int pgd_none(pgd_t pgd) { return 0; } -extern inline int pgd_bad(pgd_t pgd) { return 0; } -extern inline int pgd_present(pgd_t pgd) { return 1; } +static inline int pgd_none(pgd_t pgd) { return 0; } +static inline int pgd_bad(pgd_t pgd) { return 0; } +static inline int pgd_present(pgd_t pgd) { return 1; } #define pgd_clear(xp) do { } while (0) #define pgd_page(pgd) \ @@ -335,45 +335,45 @@ extern inline int pgd_present(pgd_t pgd) { return 1; } * The following only work if pte_present() is true. * Undefined behaviour if not.. */ -extern inline int pte_read(pte_t pte) { return pte_val(pte) & _PAGE_USER; } -extern inline int pte_write(pte_t pte) { return pte_val(pte) & _PAGE_RW; } -extern inline int pte_exec(pte_t pte) { return pte_val(pte) & _PAGE_USER; } -extern inline int pte_dirty(pte_t pte) { return pte_val(pte) & _PAGE_DIRTY; } -extern inline int pte_young(pte_t pte) { return pte_val(pte) & _PAGE_ACCESSED; } +static inline int pte_read(pte_t pte) { return pte_val(pte) & _PAGE_USER; } +static inline int pte_write(pte_t pte) { return pte_val(pte) & _PAGE_RW; } +static inline int pte_exec(pte_t pte) { return pte_val(pte) & _PAGE_USER; } +static inline int pte_dirty(pte_t pte) { return pte_val(pte) & _PAGE_DIRTY; } +static inline int pte_young(pte_t pte) { return pte_val(pte) & _PAGE_ACCESSED; } -extern inline void pte_uncache(pte_t pte) { pte_val(pte) |= _PAGE_NO_CACHE; } -extern inline void pte_cache(pte_t pte) { pte_val(pte) &= ~_PAGE_NO_CACHE; } +static inline void pte_uncache(pte_t pte) { pte_val(pte) |= _PAGE_NO_CACHE; } +static inline void pte_cache(pte_t pte) { pte_val(pte) &= ~_PAGE_NO_CACHE; } -extern inline pte_t pte_rdprotect(pte_t pte) { +static inline pte_t pte_rdprotect(pte_t pte) { pte_val(pte) &= ~_PAGE_USER; return pte; } -extern inline pte_t pte_exprotect(pte_t pte) { +static inline pte_t pte_exprotect(pte_t pte) { pte_val(pte) &= ~_PAGE_USER; return pte; } -extern inline pte_t pte_wrprotect(pte_t pte) { +static inline pte_t pte_wrprotect(pte_t pte) { pte_val(pte) &= ~(_PAGE_RW | _PAGE_HWWRITE); return pte; } -extern inline pte_t pte_mkclean(pte_t pte) { +static inline pte_t pte_mkclean(pte_t pte) { pte_val(pte) &= ~(_PAGE_DIRTY | _PAGE_HWWRITE); return pte; } -extern inline pte_t pte_mkold(pte_t pte) { +static inline pte_t pte_mkold(pte_t pte) { pte_val(pte) &= ~_PAGE_ACCESSED; return pte; } -extern inline pte_t pte_mkread(pte_t pte) { +static inline pte_t pte_mkread(pte_t pte) { pte_val(pte) |= _PAGE_USER; return pte; } -extern inline pte_t pte_mkexec(pte_t pte) { +static inline pte_t pte_mkexec(pte_t pte) { pte_val(pte) |= _PAGE_USER; return pte; } -extern inline pte_t pte_mkwrite(pte_t pte) +static inline pte_t pte_mkwrite(pte_t pte) { pte_val(pte) |= _PAGE_RW; if (pte_val(pte) & _PAGE_DIRTY) pte_val(pte) |= _PAGE_HWWRITE; return pte; } -extern inline pte_t pte_mkdirty(pte_t pte) +static inline pte_t pte_mkdirty(pte_t pte) { pte_val(pte) |= _PAGE_DIRTY; if (pte_val(pte) & _PAGE_RW) pte_val(pte) |= _PAGE_HWWRITE; return pte; } -extern inline pte_t pte_mkyoung(pte_t pte) { +static inline pte_t pte_mkyoung(pte_t pte) { pte_val(pte) |= _PAGE_ACCESSED; return pte; } /* Certain architectures need to do special things when pte's @@ -387,7 +387,7 @@ extern inline pte_t pte_mkyoung(pte_t pte) { * and a page entry and page directory to the page they refer to. */ -extern inline pte_t mk_pte_phys(unsigned long physpage, pgprot_t pgprot) +static inline pte_t mk_pte_phys(unsigned long physpage, pgprot_t pgprot) { pte_t pte; pte_val(pte) = physpage | pgprot_val(pgprot); @@ -401,12 +401,73 @@ extern inline pte_t mk_pte_phys(unsigned long physpage, pgprot_t pgprot) pte; \ }) -extern inline pte_t pte_modify(pte_t pte, pgprot_t newprot) +static inline pte_t pte_modify(pte_t pte, pgprot_t newprot) { pte_val(pte) = (pte_val(pte) & _PAGE_CHG_MASK) | pgprot_val(newprot); return pte; } +/* + * Atomic PTE updates. + * + * pte_update clears and sets bit atomically, and returns + * the old pte value. + */ +static inline unsigned long pte_update(pte_t *p, unsigned long clr, + unsigned long set) +{ + unsigned long old, tmp; + + __asm__ __volatile__("\ +1: lwarx %0,0,%3 + andc %1,%0,%4 + or %1,%1,%5 + stwcx. %1,0,%3 + bne- 1b" + : "=&r" (old), "=&r" (tmp), "=m" (*p) + : "r" (p), "r" (clr), "r" (set), "m" (*p) + : "cc" ); + return old; +} + +static inline int ptep_test_and_clear_young(pte_t *ptep) +{ + return (pte_update(ptep, _PAGE_ACCESSED, 0) & _PAGE_ACCESSED) != 0; +} + +static inline int ptep_test_and_clear_dirty(pte_t *ptep) +{ + return (pte_update(ptep, _PAGE_DIRTY | _PAGE_HWWRITE, 0) + & _PAGE_DIRTY) != 0; +} + +static inline pte_t ptep_get_and_clear(pte_t *ptep) +{ + return __pte(pte_update(ptep, ~0UL, 0)); +} + +static inline void ptep_set_wrprotect(pte_t *ptep) +{ + pte_update(ptep, _PAGE_RW | _PAGE_HWWRITE, 0); +} + +static inline void ptep_mkdirty(pte_t *ptep) +{ + /* + * N.B. this doesn't set the _PAGE_HWWRITE bit in the case + * where _PAGE_RW is set and _PAGE_DIRTY was clear. This + * doesn't matter; all it will mean is that if the next call + * to hash_page for this page is for a read, it will put a + * readonly HPTE into the hash table rather than a R/W HPTE. + * A call to hash_page for a write to this page will set + * _PAGE_HWWRITE and put a R/W HPTE into the hash table. + * -- paulus. + */ + pte_update(ptep, 0, _PAGE_DIRTY); +} + +#define pte_same(A,B) (pte_val(A) == pte_val(B)) + #define pmd_page(pmd) (pmd_val(pmd)) /* to find an entry in a kernel page-table-directory */ @@ -417,13 +478,13 @@ extern inline pte_t pte_modify(pte_t pte, pgprot_t newprot) #define pgd_offset(mm, address) ((mm)->pgd + pgd_index(address)) /* Find an entry in the second-level page table.. */ -extern inline pmd_t * pmd_offset(pgd_t * dir, unsigned long address) +static inline pmd_t * pmd_offset(pgd_t * dir, unsigned long address) { return (pmd_t *) dir; } /* Find an entry in the third-level page table.. */ -extern inline pte_t * pte_offset(pmd_t * dir, unsigned long address) +static inline pte_t * pte_offset(pmd_t * dir, unsigned long address) { return (pte_t *) pmd_page(*dir) + ((address >> PAGE_SHIFT) & (PTRS_PER_PTE - 1)); } @@ -486,8 +547,6 @@ extern void kernel_set_cachemode (unsigned long address, unsigned long size, #define io_remap_page_range remap_page_range -#include <asm-generic/pgtable.h> - -#endif __ASSEMBLY__ +#endif /* __ASSEMBLY__ */ #endif /* _PPC_PGTABLE_H */ #endif /* __KERNEL__ */ diff --git a/include/asm-ppc/prep_nvram.h b/include/asm-ppc/prep_nvram.h index 82ec21508..85694899c 100644 --- a/include/asm-ppc/prep_nvram.h +++ b/include/asm-ppc/prep_nvram.h @@ -21,6 +21,11 @@ #ifndef _PPC_PREP_NVRAM_H #define _PPC_PREP_NVRAM_H +#define MAX_PREP_NVRAM 0x8000 +#define PREP_NVRAM_AS0 0x74 +#define PREP_NVRAM_AS1 0x75 +#define PREP_NVRAM_DATA 0x77 + #define NVSIZE 4096 /* size of NVRAM */ #define OSAREASIZE 512 /* size of OSArea space */ #define CONFSIZE 1024 /* guess at size of Configuration space */ diff --git a/include/asm-ppc/processor.h b/include/asm-ppc/processor.h index 5908d7f32..cd87f248e 100644 --- a/include/asm-ppc/processor.h +++ b/include/asm-ppc/processor.h @@ -203,6 +203,12 @@ #define SPRN_IMISS 0x3D4 /* Instruction TLB Miss Register */ #define SPRN_IMMR 0x27E /* Internal Memory Map Register */ #define SPRN_L2CR 0x3F9 /* Level 2 Cache Control Regsiter */ +#define L2CR_PIPE_LATEWR (0x01800000) /* late-write SRAM */ +#define L2CR_L2CTL (0x00100000) /* RAM control */ +#define L2CR_INST_DISABLE (0x00400000) /* disable for insn's */ +#define L2CR_L2I (0x00200000) /* global invalidate */ +#define L2CR_L2E (0x80000000) /* enable */ +#define L2CR_L2WT (0x00080000) /* write-through */ #define SPRN_LR 0x008 /* Link Register */ #define SPRN_MMCR0 0x3B8 /* Monitor Mode Control Register 0 */ #define SPRN_MMCR1 0x3BC /* Monitor Mode Control Register 1 */ @@ -233,14 +239,14 @@ #define SPRN_SRR1 0x01B /* Save/Restore Register 1 */ #define SPRN_SRR2 0x3DE /* Save/Restore Register 2 */ #define SPRN_SRR3 0x3DF /* Save/Restore Register 3 */ -#define SPRN_TBHI 0x3DC /* Time Base High */ -#define SPRN_TBHU 0x3CC /* Time Base High User-mode */ -#define SPRN_TBLO 0x3DD /* Time Base Low */ -#define SPRN_TBLU 0x3CD /* Time Base Low User-mode */ -#define SPRN_TBRL 0x10D /* Time Base Read Lower Register */ -#define SPRN_TBRU 0x10C /* Time Base Read Upper Register */ -#define SPRN_TBWL 0x11D /* Time Base Write Lower Register */ -#define SPRN_TBWU 0x11C /* Time Base Write Upper Register */ +#define SPRN_TBRL 0x10C /* Time Base Read Lower Register (user, R/O) */ +#define SPRN_TBRU 0x10D /* Time Base Read Upper Register (user, R/O) */ +#define SPRN_TBWL 0x11C /* Time Base Lower Register (supervisor, R/W) */ +#define SPRN_TBWU 0x11D /* Time Base Upper Register (supervisor, R/W) */ +#define SPRN_TBHI 0x3DC /* Time Base High (4xx) */ +#define SPRN_TBHU 0x3CC /* Time Base High User-mode (4xx) */ +#define SPRN_TBLO 0x3DD /* Time Base Low (4xx) */ +#define SPRN_TBLU 0x3CD /* Time Base Low User-mode (4xx) */ #define SPRN_TCR 0x3DA /* Timer Control Register */ #define TCR_WP(x) (((x)&0x3)<<30) /* WDT Period */ #define WP_2_17 0 /* 2^17 clocks */ @@ -262,15 +268,17 @@ #define TCR_FIE 0x00800000 /* FIT Interrupt Enable */ #define TCR_ARE 0x00400000 /* Auto Reload Enable */ #define SPRN_THRM1 0x3FC /* Thermal Management Register 1 */ -#define THRM1_TIN (1<<0) -#define THRM1_TIV (1<<1) -#define THRM1_THRES (0x7f<<2) -#define THRM1_TID (1<<29) -#define THRM1_TIE (1<<30) -#define THRM1_V (1<<31) +/* these bits were defined in inverted endian sense originally, ugh, confusing */ +#define THRM1_TIN (1 << 31) +#define THRM1_TIV (1 << 30) +#define THRM1_THRES(x) ((x&0x7f)<<23) +#define THRM3_SITV(x) ((x&0x3fff)<<1) +#define THRM1_TID (1<<2) +#define THRM1_TIE (1<<1) +#define THRM1_V (1<<0) #define SPRN_THRM2 0x3FD /* Thermal Management Register 2 */ #define SPRN_THRM3 0x3FE /* Thermal Management Register 3 */ -#define THRM3_E (1<<31) +#define THRM3_E (1<<0) #define SPRN_TSR 0x3D8 /* Timer Status Register */ #define TSR_ENW 0x80000000 /* Enable Next Watchdog */ #define TSR_WIS 0x40000000 /* WDT Interrupt Status */ @@ -500,8 +508,8 @@ #define _MACH_fads 0x00000020 /* Motorola FADS board */ #define _MACH_rpxlite 0x00000040 /* RPCG RPX-Lite 8xx board */ #define _MACH_bseip 0x00000080 /* Bright Star Engineering ip-Engine */ -#define _MACH_yk 0x00000100 /* Motorola Yellowknife */ -#define _MACH_gemini 0x00000200 /* Synergy Microsystems gemini board */ +#define _MACH_unused0 0x00000100 /* Now free to be used */ +#define _MACH_unused1 0x00000200 /* Now free to be used */ #define _MACH_classic 0x00000400 /* RPCG RPX-Classic 8xx board */ #define _MACH_oak 0x00000800 /* IBM "Oak" 403 eval. board */ #define _MACH_walnut 0x00001000 /* IBM "Walnut" 405GP eval. board */ @@ -509,24 +517,11 @@ #define _MACH_tqm860 0x00004000 /* TQM860/L */ #define _MACH_tqm8xxL 0x00008000 /* TQM8xxL */ - /* see residual.h for these */ #define _PREP_Motorola 0x01 /* motorola prep */ #define _PREP_Firm 0x02 /* firmworks prep */ #define _PREP_IBM 0x00 /* ibm prep */ #define _PREP_Bull 0x03 /* bull prep */ -#define _PREP_Radstone 0x04 /* Radstone Technology PLC prep */ - -/* - * Radstone board types - */ -#define RS_SYS_TYPE_PPC1 0 -#define RS_SYS_TYPE_PPC2 1 -#define RS_SYS_TYPE_PPC1a 2 -#define RS_SYS_TYPE_PPC2a 3 -#define RS_SYS_TYPE_PPC4 4 -#define RS_SYS_TYPE_PPC4a 5 -#define RS_SYS_TYPE_PPC2ep 6 /* these are arbitrary */ #define _CHRP_Motorola 0x04 /* motorola chrp, the cobra */ @@ -715,9 +710,6 @@ void _nmask_and_or_msr(unsigned long nmask, unsigned long or_val); #elif defined(CONFIG_APUS) #define _machine _MACH_apus #define have_of 0 -#elif defined(CONFIG_GEMINI) -#define _machine _MACH_gemini -#define have_of 0 #elif defined(CONFIG_8260) #define _machine _MACH_8260 #define have_of 0 diff --git a/include/asm-ppc/prom.h b/include/asm-ppc/prom.h index 0c57690ea..b80c460b3 100644 --- a/include/asm-ppc/prom.h +++ b/include/asm-ppc/prom.h @@ -80,14 +80,15 @@ extern struct device_node *find_type_devices(const char *type); extern struct device_node *find_path_device(const char *path); extern struct device_node *find_compatible_devices(const char *type, const char *compat); -extern struct device_node *find_pci_device_OFnode(unsigned char bus, - unsigned char dev_fn); extern struct device_node *find_phandle(phandle); extern struct device_node *find_all_nodes(void); extern int device_is_compatible(struct device_node *device, const char *); extern int machine_is_compatible(const char *compat); extern unsigned char *get_property(struct device_node *node, const char *name, int *lenp); +extern void prom_add_property(struct device_node* np, struct property* prop); +extern void prom_get_irq_senses(unsigned char *, int, int); + extern void print_properties(struct device_node *node); extern int call_rtas(const char *service, int nargs, int nret, unsigned long *outputs, ...); @@ -96,7 +97,8 @@ extern void prom_drawhex(unsigned long v); extern void prom_drawchar(char c); extern void map_bootx_text(void); - +extern void bootx_update_display(unsigned long phys, int width, int height, + int depth, int pitch); #endif /* _PPC_PROM_H */ #endif /* __KERNEL__ */ diff --git a/include/asm-ppc/raven.h b/include/asm-ppc/raven.h index ee873ff82..e912088b0 100644 --- a/include/asm-ppc/raven.h +++ b/include/asm-ppc/raven.h @@ -31,5 +31,5 @@ extern struct hw_interrupt_type raven_pic; extern int raven_init(void); -#endif _ASMPPC_RAVEN_H +#endif /* _ASMPPC_RAVEN_H */ #endif /* __KERNEL__ */ diff --git a/include/asm-ppc/segment.h b/include/asm-ppc/segment.h index 0eef1e5e7..0f2f7428d 100644 --- a/include/asm-ppc/segment.h +++ b/include/asm-ppc/segment.h @@ -1,7 +1 @@ -#ifndef __PPC_SEGMENT_H -#define __PPC_SEGMENT_H - -/* Only here because we have some old header files that expect it.. */ - -#endif #include <asm/uaccess.h> diff --git a/include/asm-ppc/serial.h b/include/asm-ppc/serial.h index 60c1a14ff..721476591 100644 --- a/include/asm-ppc/serial.h +++ b/include/asm-ppc/serial.h @@ -5,10 +5,6 @@ #ifdef __KERNEL__ #include <linux/config.h> -#ifdef CONFIG_GEMINI -#include <asm/gemini_serial.h> -#else - /* * This assumes you have a 1.8432 MHz clock for your UART. * @@ -127,5 +123,4 @@ HUB6_SERIAL_PORT_DFNS \ MCA_SERIAL_PORT_DFNS -#endif /* CONFIG_GEMINI */ #endif /* __KERNEL__ */ diff --git a/include/asm-ppc/smp.h b/include/asm-ppc/smp.h index 4851e13fd..6dfc778b9 100644 --- a/include/asm-ppc/smp.h +++ b/include/asm-ppc/smp.h @@ -15,7 +15,7 @@ #ifndef __ASSEMBLY__ struct cpuinfo_PPC { - unsigned long loops_per_sec; + unsigned long loops_per_jiffy; unsigned long pvr; unsigned long *pgd_cache; unsigned long *pte_cache; diff --git a/include/asm-ppc/smplock.h b/include/asm-ppc/smplock.h index 5fdd5733b..4b7ba58e9 100644 --- a/include/asm-ppc/smplock.h +++ b/include/asm-ppc/smplock.h @@ -39,13 +39,13 @@ do { \ * so we only need to worry about other * CPU's. */ -extern __inline__ void lock_kernel(void) +static __inline__ void lock_kernel(void) { if (!++current->lock_depth) spin_lock(&kernel_flag); } -extern __inline__ void unlock_kernel(void) +static __inline__ void unlock_kernel(void) { if (--current->lock_depth < 0) spin_unlock(&kernel_flag); diff --git a/include/asm-ppc/termios.h b/include/asm-ppc/termios.h index 2a9b8b025..22f53ce8b 100644 --- a/include/asm-ppc/termios.h +++ b/include/asm-ppc/termios.h @@ -43,6 +43,7 @@ struct ltchars { #define FIONBIO _IOW('f', 126, int) #define FIONREAD _IOR('f', 127, int) #define TIOCINQ FIONREAD +#define FIOQSIZE _IOR('f', 128, loff_t) #define TIOCGETP _IOR('t', 8, struct sgttyb) #define TIOCSETP _IOW('t', 9, struct sgttyb) diff --git a/include/asm-ppc/unistd.h b/include/asm-ppc/unistd.h index 5c432792b..c0a69ef72 100644 --- a/include/asm-ppc/unistd.h +++ b/include/asm-ppc/unistd.h @@ -206,6 +206,10 @@ #define __NR_pciconfig_iobase 200 #define __NR_multiplexer 201 #define __NR_getdents64 202 +#define __NR_pivot_root 203 +#define __NR_fcntl64 204 +#define __NR_madvise 205 +#define __NR_mincore 206 #define __NR(n) #n diff --git a/include/asm-sh/current.h b/include/asm-sh/current.h index 355a2cd14..c690c233b 100644 --- a/include/asm-sh/current.h +++ b/include/asm-sh/current.h @@ -12,7 +12,7 @@ static __inline__ struct task_struct * get_current(void) { struct task_struct *current; - __asm__("stc $r7_bank, %0" + __asm__("stc r7_bank, %0" :"=r" (current)); return current; diff --git a/include/asm-sh/pgtable.h b/include/asm-sh/pgtable.h index b1a6f9461..a448c6f00 100644 --- a/include/asm-sh/pgtable.h +++ b/include/asm-sh/pgtable.h @@ -39,6 +39,7 @@ extern void paging_init(void); #define flush_dcache_page(page) do { } while (0) #define flush_icache_range(start, end) do { } while (0) #define flush_icache_page(vma,pg) do { } while (0) +#define flush_cache_sigtramp(vaddr) do { } while (0) #elif defined(__SH4__) /* * Caches are broken on SH-4, so we need them. @@ -52,6 +53,7 @@ extern void flush_page_to_ram(struct page *page); extern void flush_dcache_page(struct page *pg); extern void flush_icache_range(unsigned long start, unsigned long end); extern void flush_icache_page(struct vm_area_struct *vma, struct page *pg); +extern void flush_cache_sigtramp(unsigned long addr); #endif /* @@ -125,11 +127,7 @@ extern unsigned long empty_zero_page[1024]; /* Hardware flags: SZ=1 (4k-byte) */ #define _PAGE_FLAGS_HARD 0x00000010 -#if defined(__sh3__) -#define _PAGE_SHARED _PAGE_HW_SHARED -#elif defined(__SH4__) #define _PAGE_SHARED _PAGE_U0_SHARED -#endif #define _PAGE_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER | _PAGE_ACCESSED | _PAGE_DIRTY) #define _KERNPG_TABLE (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY) diff --git a/include/asm-sh/sh_bios.h b/include/asm-sh/sh_bios.h index cdb73bdea..0ca261956 100644 --- a/include/asm-sh/sh_bios.h +++ b/include/asm-sh/sh_bios.h @@ -13,5 +13,7 @@ extern void sh_bios_char_out(char ch); extern int sh_bios_in_gdb_mode(void); extern void sh_bios_gdb_detach(void); +extern void sh_bios_get_node_addr(unsigned char *node_addr); +extern void sh_bios_shutdown(unsigned int how); #endif /* __ASM_SH_BIOS_H */ diff --git a/include/asm-sh/system.h b/include/asm-sh/system.h index 86786c730..a8f720582 100644 --- a/include/asm-sh/system.h +++ b/include/asm-sh/system.h @@ -6,6 +6,7 @@ */ #include <linux/config.h> +#include <linux/kernel.h> /* * switch_to() should switch tasks to task nr n, first @@ -28,37 +29,37 @@ typedef struct { register unsigned long *__ts6 __asm__ ("r6") = &next->thread.sp; \ register unsigned long __ts7 __asm__ ("r7") = next->thread.pc; \ __asm__ __volatile__ (".balign 4\n\t" \ - "stc.l $gbr, @-$r15\n\t" \ - "sts.l $pr, @-$r15\n\t" \ - "mov.l $r8, @-$r15\n\t" \ - "mov.l $r9, @-$r15\n\t" \ - "mov.l $r10, @-$r15\n\t" \ - "mov.l $r11, @-$r15\n\t" \ - "mov.l $r12, @-$r15\n\t" \ - "mov.l $r13, @-$r15\n\t" \ - "mov.l $r14, @-$r15\n\t" \ - "mov.l $r15, @$r1 ! save SP\n\t" \ - "mov.l @$r6, $r15 ! change to new stack\n\t" \ - "mov.l %0, @-$r15 ! push R0 onto new stack\n\t" \ + "stc.l gbr, @-r15\n\t" \ + "sts.l pr, @-r15\n\t" \ + "mov.l r8, @-r15\n\t" \ + "mov.l r9, @-r15\n\t" \ + "mov.l r10, @-r15\n\t" \ + "mov.l r11, @-r15\n\t" \ + "mov.l r12, @-r15\n\t" \ + "mov.l r13, @-r15\n\t" \ + "mov.l r14, @-r15\n\t" \ + "mov.l r15, @r1 ! save SP\n\t" \ + "mov.l @r6, r15 ! change to new stack\n\t" \ + "mov.l %0, @-r15 ! push R0 onto new stack\n\t" \ "mova 1f, %0\n\t" \ - "mov.l %0, @$r2 ! save PC\n\t" \ + "mov.l %0, @r2 ! save PC\n\t" \ "mov.l 2f, %0\n\t" \ "jmp @%0 ! call __switch_to\n\t" \ - " lds $r7, $pr ! with return to new PC\n\t" \ + " lds r7, pr ! with return to new PC\n\t" \ ".balign 4\n" \ "2:\n\t" \ ".long " "__switch_to\n" \ "1:\n\t" \ - "mov.l @$r15+, %0 ! pop R0 from new stack\n\t" \ - "mov.l @$r15+, $r14\n\t" \ - "mov.l @$r15+, $r13\n\t" \ - "mov.l @$r15+, $r12\n\t" \ - "mov.l @$r15+, $r11\n\t" \ - "mov.l @$r15+, $r10\n\t" \ - "mov.l @$r15+, $r9\n\t" \ - "mov.l @$r15+, $r8\n\t" \ - "lds.l @$r15+, $pr\n\t" \ - "ldc.l @$r15+, $gbr\n\t" \ + "mov.l @r15+, %0 ! pop R0 from new stack\n\t" \ + "mov.l @r15+, r14\n\t" \ + "mov.l @r15+, r13\n\t" \ + "mov.l @r15+, r12\n\t" \ + "mov.l @r15+, r11\n\t" \ + "mov.l @r15+, r10\n\t" \ + "mov.l @r15+, r9\n\t" \ + "mov.l @r15+, r8\n\t" \ + "lds.l @r15+, pr\n\t" \ + "ldc.l @r15+, gbr\n\t" \ :"=&z" (__last) \ :"0" (prev), \ "r" (__ts1), "r" (__ts2), \ @@ -107,11 +108,11 @@ static __inline__ void __sti(void) { unsigned long __dummy0, __dummy1; - __asm__ __volatile__("stc $sr, %0\n\t" + __asm__ __volatile__("stc sr, %0\n\t" "and %1, %0\n\t" - "stc $r6_bank, %1\n\t" + "stc r6_bank, %1\n\t" "or %1, %0\n\t" - "ldc %0, $sr" + "ldc %0, sr" : "=&r" (__dummy0), "=r" (__dummy1) : "1" (~0x000000f0) : "memory"); @@ -120,9 +121,9 @@ static __inline__ void __sti(void) static __inline__ void __cli(void) { unsigned long __dummy; - __asm__ __volatile__("stc $sr, %0\n\t" + __asm__ __volatile__("stc sr, %0\n\t" "or #0xf0, %0\n\t" - "ldc %0, $sr" + "ldc %0, sr" : "=&z" (__dummy) : /* no inputs */ : "memory"); @@ -131,7 +132,7 @@ static __inline__ void __cli(void) #define __save_flags(x) \ x = (__extension__ ({ unsigned long __sr; \ __asm__ __volatile__( \ - "stc $sr, %0" \ + "stc sr, %0" \ : "=&r" (__sr) \ : /* no inputs */ \ : "memory"); \ @@ -140,10 +141,10 @@ x = (__extension__ ({ unsigned long __sr; \ #define __save_and_cli(x) \ x = (__extension__ ({ unsigned long __dummy,__sr; \ __asm__ __volatile__( \ - "stc $sr, %1\n\t" \ + "stc sr, %1\n\t" \ "mov %1, %0\n\t" \ "or #0xf0, %0\n\t" \ - "ldc %0, $sr" \ + "ldc %0, sr" \ : "=&z" (__dummy), "=&r" (__sr) \ : /* no inputs */ \ : "memory"); (__sr & 0x000000f0); })) diff --git a/include/asm-sparc/mostek.h b/include/asm-sparc/mostek.h index c9a10cd86..be60a9afa 100644 --- a/include/asm-sparc/mostek.h +++ b/include/asm-sparc/mostek.h @@ -1,4 +1,4 @@ -/* $Id: mostek.h,v 1.12 1999/08/31 18:51:41 davem Exp $ +/* $Id: mostek.h,v 1.13 2001/01/11 15:07:09 davem Exp $ * mostek.h: Describes the various Mostek time of day clock registers. * * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu) @@ -64,6 +64,7 @@ struct mostek48t02 { volatile unsigned char year; /* Year (0-99) */ }; +extern spinlock_t mostek_lock; extern unsigned long mstk48t02_regs; /* Control register values. */ diff --git a/include/asm-sparc64/mostek.h b/include/asm-sparc64/mostek.h index e153a36cf..b000c1586 100644 --- a/include/asm-sparc64/mostek.h +++ b/include/asm-sparc64/mostek.h @@ -1,4 +1,4 @@ -/* $Id: mostek.h,v 1.3 1999/08/30 10:14:50 davem Exp $ +/* $Id: mostek.h,v 1.4 2001/01/11 15:07:09 davem Exp $ * mostek.h: Describes the various Mostek time of day clock registers. * * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu) @@ -66,6 +66,7 @@ static __inline__ void mostek_write(unsigned long addr, u8 val) #define MOSTEK_MONTH 0x07feUL #define MOSTEK_YEAR 0x07ffUL +extern spinlock_t mostek_lock; extern unsigned long mstk48t02_regs; /* Control register values. */ diff --git a/include/asm-sparc64/pbm.h b/include/asm-sparc64/pbm.h index c17e9bf23..082626725 100644 --- a/include/asm-sparc64/pbm.h +++ b/include/asm-sparc64/pbm.h @@ -1,4 +1,4 @@ -/* $Id: pbm.h,v 1.22 2000/03/25 05:18:30 davem Exp $ +/* $Id: pbm.h,v 1.23 2001/01/11 16:26:45 davem Exp $ * pbm.h: UltraSparc PCI controller software state. * * Copyright (C) 1997, 1998, 1999 David S. Miller (davem@redhat.com) @@ -18,11 +18,12 @@ /* The abstraction used here is that there are PCI controllers, * each with one (Sabre) or two (PSYCHO/SCHIZO) PCI bus modules - * underneath. Each PCI controller has a single IOMMU shared - * by the PCI bus modules underneath, and if a streaming buffer + * underneath. Each PCI bus module uses an IOMMU (shared by both + * PBMs of a controller, or per-PBM), and if a streaming buffer * is present, each PCI bus module has it's own. (ie. the IOMMU - * is shared between PBMs, the STC is not) Furthermore, each - * PCI bus module controls it's own autonomous PCI bus. + * might be shared between PBMs, the STC is never shared) + * Furthermore, each PCI bus module controls it's own autonomous + * PCI bus. */ #define PBM_LOGCLUSTERS 3 @@ -150,6 +151,9 @@ struct pci_pbm_info { /* This PBM's streaming buffer. */ struct pci_strbuf stc; + /* IOMMU state, potentially shared by both PBM segments. */ + struct pci_iommu *iommu; + /* Now things for the actual PCI bus probes. */ unsigned int pci_first_busno; unsigned int pci_last_busno; @@ -189,9 +193,6 @@ struct pci_controller_info { unsigned int pci_first_busno; unsigned int pci_last_busno; - /* IOMMU state shared by both PBM segments. */ - struct pci_iommu iommu; - void *starfire_cookie; }; diff --git a/include/asm-sparc64/watchdog.h b/include/asm-sparc64/watchdog.h new file mode 100644 index 000000000..4d5b03b60 --- /dev/null +++ b/include/asm-sparc64/watchdog.h @@ -0,0 +1,31 @@ +/* $Id: watchdog.h,v 1.1 2001/01/18 04:47:44 davem Exp $ + * + * watchdog - Driver interface for the hardware watchdog timers + * present on Sun Microsystems boardsets + * + * Copyright (c) 2000 Eric Brower <ebrower@usa.net> + * + */ + +#ifndef _SPARC64_WATCHDOG_H +#define _SPARC64_WATCHDOG_H + +#include <linux/watchdog.h> + +/* Solaris compatibility ioctls-- + * Ref. <linux/watchdog.h> for standard linux watchdog ioctls + */ +#define WIOCSTART _IO (WATCHDOG_IOCTL_BASE, 10) /* Start Timer */ +#define WIOCSTOP _IO (WATCHDOG_IOCTL_BASE, 11) /* Stop Timer */ +#define WIOCGSTAT _IOR(WATCHDOG_IOCTL_BASE, 12, int)/* Get Timer Status */ + +/* Status flags from WIOCGSTAT ioctl + */ +#define WD_FREERUN 0x01 /* timer is running, interrupts disabled */ +#define WD_EXPIRED 0x02 /* timer has expired */ +#define WD_RUNNING 0x04 /* timer is running, interrupts enabled */ +#define WD_STOPPED 0x08 /* timer has not been started */ +#define WD_SERVICED 0x10 /* timer interrupt was serviced */ + +#endif /* ifndef _SPARC64_WATCHDOG_H */ + diff --git a/include/linux/acpi.h b/include/linux/acpi.h index ff1dcaf45..4fce3f847 100644 --- a/include/linux/acpi.h +++ b/include/linux/acpi.h @@ -28,21 +28,6 @@ #include <linux/wait.h> #endif /* __KERNEL__ */ -u64 acpi_get_rsdp_ptr(void); - -/* - * System sleep states - */ -enum -{ - ACPI_S0, /* working */ - ACPI_S1, /* sleep */ - ACPI_S2, /* sleep */ - ACPI_S3, /* sleep */ - ACPI_S4, /* non-volatile sleep */ - ACPI_S5, /* soft-off */ -}; - typedef int acpi_sstate_t; /* diff --git a/include/linux/blk.h b/include/linux/blk.h index 11db342a3..3f9c707c7 100644 --- a/include/linux/blk.h +++ b/include/linux/blk.h @@ -87,10 +87,6 @@ void initrd_init(void); static inline void blkdev_dequeue_request(struct request * req) { - if (req->e) { - req->e->dequeue_fn(req); - req->e = NULL; - } list_del(&req->queue); } @@ -322,7 +318,7 @@ static void floppy_off(unsigned int nr); #define DEVICE_NAME "ida" #define TIMEOUT_VALUE (25*HZ) -#define DEVICE_REQUEST do_ida_request0 +#define DEVICE_REQUEST do_ida_request #define DEVICE_NR(device) (MINOR(device) >> 4) #endif /* MAJOR_NR == whatever */ diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h index 851bf3c53..01cd38a83 100644 --- a/include/linux/blkdev.h +++ b/include/linux/blkdev.h @@ -23,8 +23,6 @@ struct request { int elevator_sequence; struct list_head table; - struct list_head *free_list; - volatile int rq_status; /* should split this into a few status bits */ #define RQ_INACTIVE (-1) #define RQ_ACTIVE 1 @@ -47,7 +45,6 @@ struct request { struct buffer_head * bh; struct buffer_head * bhtail; request_queue_t *q; - elevator_t *e; }; #include <linux/elevator.h> @@ -67,9 +64,10 @@ typedef void (plug_device_fn) (request_queue_t *q, kdev_t device); typedef void (unplug_device_fn) (void *q); /* - * Default nr free requests per queue + * Default nr free requests per queue, ll_rw_blk will scale it down + * according to available RAM at init time */ -#define QUEUE_NR_REQUESTS 256 +#define QUEUE_NR_REQUESTS 8192 struct request_queue { @@ -77,6 +75,8 @@ struct request_queue * the queue request freelist, one for reads and one for writes */ struct list_head request_freelist[2]; + struct list_head pending_freelist[2]; + int pending_free[2]; /* * Together with queue_head for cacheline sharing @@ -116,7 +116,7 @@ struct request_queue * Is meant to protect the queue in the future instead of * io_request_lock */ - spinlock_t request_lock; + spinlock_t queue_lock; /* * Tasks wait here for free request @@ -152,6 +152,7 @@ extern void grok_partitions(struct gendisk *dev, int drive, unsigned minors, lon extern void register_disk(struct gendisk *dev, kdev_t first, unsigned minors, struct block_device_operations *ops, long size); extern void generic_make_request(int rw, struct buffer_head * bh); extern request_queue_t *blk_get_queue(kdev_t dev); +extern inline request_queue_t *__blk_get_queue(kdev_t dev); extern void blkdev_release_request(struct request *); /* @@ -162,6 +163,7 @@ extern void blk_cleanup_queue(request_queue_t *); extern void blk_queue_headactive(request_queue_t *, int); extern void blk_queue_pluggable(request_queue_t *, plug_device_fn *); extern void blk_queue_make_request(request_queue_t *, make_request_fn *); +extern void generic_unplug_device(void *); extern int * blk_size[MAX_BLKDEV]; @@ -175,9 +177,10 @@ extern int * max_sectors[MAX_BLKDEV]; extern int * max_segments[MAX_BLKDEV]; -#define MAX_SECTORS 254 +extern atomic_t queued_sectors; -#define MAX_SEGMENTS MAX_SECTORS +#define MAX_SEGMENTS 128 +#define MAX_SECTORS (MAX_SEGMENTS*8) #define PageAlignSize(size) (((size) + PAGE_SIZE -1) & PAGE_MASK) @@ -203,5 +206,14 @@ static inline int get_hardsect_size(kdev_t dev) return 512; } +#define blk_finished_io(nsects) \ + atomic_sub(nsects, &queued_sectors); \ + if (atomic_read(&queued_sectors) < 0) { \ + printk("block: queued_sectors < 0\n"); \ + atomic_set(&queued_sectors, 0); \ + } + +#define blk_started_io(nsects) \ + atomic_add(nsects, &queued_sectors); #endif diff --git a/include/linux/dn.h b/include/linux/dn.h index c7448158b..782cae49e 100644 --- a/include/linux/dn.h +++ b/include/linux/dn.h @@ -45,7 +45,12 @@ #define DSO_LINKINFO 7 /* Set/Get link information */ #define DSO_STREAM 8 /* Set socket type to stream */ #define DSO_SEQPACKET 9 /* Set socket type to sequenced packet */ -#define DSO_MAX 10 /* Maximum option number */ +#define DSO_MAXWINDOW 11 /* Maximum window size allowed */ +#define DSO_NODELAY 12 /* Turn off nagle */ +#define DSO_CORK 13 /* Wait for more data! */ +#define DSO_SERVICES 14 /* NSP Services field */ +#define DSO_INFO 15 /* NSP Info field */ +#define DSO_MAX 15 /* Maximum option number */ /* LINK States */ diff --git a/include/linux/elevator.h b/include/linux/elevator.h index 16fa9cae6..9e9b27dfa 100644 --- a/include/linux/elevator.h +++ b/include/linux/elevator.h @@ -7,34 +7,32 @@ typedef void (elevator_fn) (struct request *, elevator_t *, struct list_head *, struct list_head *, int); -typedef int (elevator_merge_fn) (request_queue_t *, struct request **, - struct buffer_head *, int, int *, int *); +typedef int (elevator_merge_fn) (request_queue_t *, struct request **, struct list_head *, + struct buffer_head *, int, int, int); -typedef void (elevator_dequeue_fn) (struct request *); +typedef void (elevator_merge_cleanup_fn) (request_queue_t *, struct request *, int); + +typedef void (elevator_merge_req_fn) (struct request *, struct request *); struct elevator_s { - int sequence; - int read_latency; int write_latency; - int max_bomb_segments; - unsigned int nr_segments; - int read_pendings; - - elevator_fn * elevator_fn; elevator_merge_fn *elevator_merge_fn; - elevator_dequeue_fn *dequeue_fn; + elevator_merge_cleanup_fn *elevator_merge_cleanup_fn; + elevator_merge_req_fn *elevator_merge_req_fn; unsigned int queue_ID; }; -void elevator_noop(struct request *, elevator_t *, struct list_head *, struct list_head *, int); -int elevator_noop_merge(request_queue_t *, struct request **, struct buffer_head *, int, int *, int *); -void elevator_noop_dequeue(struct request *); -void elevator_linus(struct request *, elevator_t *, struct list_head *, struct list_head *, int); -int elevator_linus_merge(request_queue_t *, struct request **, struct buffer_head *, int, int *, int *); +int elevator_noop_merge(request_queue_t *, struct request **, struct list_head *, struct buffer_head *, int, int, int); +void elevator_noop_merge_cleanup(request_queue_t *, struct request *, int); +void elevator_noop_merge_req(struct request *, struct request *); + +int elevator_linus_merge(request_queue_t *, struct request **, struct list_head *, struct buffer_head *, int, int, int); +void elevator_linus_merge_cleanup(request_queue_t *, struct request *, int); +void elevator_linus_merge_req(struct request *, struct request *); typedef struct blkelv_ioctl_arg_s { int queue_ID; @@ -69,6 +67,10 @@ extern void elevator_init(elevator_t *, elevator_t); (s1)->sector < (s2)->sector)) || \ (s1)->rq_dev < (s2)->rq_dev) +#define BHRQ_IN_ORDER(bh, rq) \ + (((bh)->b_rdev == (rq)->rq_dev && \ + (bh)->b_rsector < (rq)->sector)) + static inline int elevator_request_latency(elevator_t * elevator, int rw) { int latency; @@ -80,36 +82,24 @@ static inline int elevator_request_latency(elevator_t * elevator, int rw) return latency; } -#define ELEVATOR_NOOP \ -((elevator_t) { \ - 0, /* sequence */ \ - \ - 0, /* read_latency */ \ - 0, /* write_latency */ \ - 0, /* max_bomb_segments */ \ - \ - 0, /* nr_segments */ \ - 0, /* read_pendings */ \ - \ - elevator_noop, /* elevator_fn */ \ - elevator_noop_merge, /* elevator_merge_fn */ \ - elevator_noop_dequeue, /* dequeue_fn */ \ +#define ELEVATOR_NOOP \ +((elevator_t) { \ + 0, /* read_latency */ \ + 0, /* write_latency */ \ + \ + elevator_noop_merge, /* elevator_merge_fn */ \ + elevator_noop_merge_cleanup, /* elevator_merge_cleanup_fn */ \ + elevator_noop_merge_req, /* elevator_merge_req_fn */ \ }) -#define ELEVATOR_LINUS \ -((elevator_t) { \ - 0, /* not used */ \ - \ - 1000000, /* read passovers */ \ - 2000000, /* write passovers */ \ - 0, /* max_bomb_segments */ \ - \ - 0, /* not used */ \ - 0, /* not used */ \ - \ - elevator_linus, /* elevator_fn */ \ - elevator_linus_merge, /* elevator_merge_fn */ \ - elevator_noop_dequeue, /* dequeue_fn */ \ +#define ELEVATOR_LINUS \ +((elevator_t) { \ + 8192, /* read passovers */ \ + 16384, /* write passovers */ \ + \ + elevator_linus_merge, /* elevator_merge_fn */ \ + elevator_linus_merge_cleanup, /* elevator_merge_cleanup_fn */ \ + elevator_linus_merge_req, /* elevator_merge_req_fn */ \ }) #endif diff --git a/include/linux/fs.h b/include/linux/fs.h index 8032db992..686aef31a 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -288,6 +288,7 @@ extern void set_bh_page(struct buffer_head *bh, struct page *page, unsigned long #include <linux/hfs_fs_i.h> #include <linux/adfs_fs_i.h> #include <linux/qnx4_fs_i.h> +#include <linux/reiserfs_fs_i.h> #include <linux/bfs_fs_i.h> #include <linux/udf_fs_i.h> #include <linux/ncp_fs_i.h> @@ -450,6 +451,7 @@ struct inode { struct hfs_inode_info hfs_i; struct adfs_inode_info adfs_i; struct qnx4_inode_info qnx4_i; + struct reiserfs_inode_info reiserfs_i; struct bfs_inode_info bfs_i; struct udf_inode_info udf_i; struct ncp_inode_info ncpfs_i; @@ -460,35 +462,6 @@ struct inode { } u; }; -/* Inode state bits.. */ -#define I_DIRTY_SYNC 1 /* Not dirty enough for O_DATASYNC */ -#define I_DIRTY_DATASYNC 2 /* Data-related inode changes pending */ -#define I_DIRTY_PAGES 4 /* Data-related inode changes pending */ -#define I_LOCK 8 -#define I_FREEING 16 -#define I_CLEAR 32 - -#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES) - -extern void __mark_inode_dirty(struct inode *, int); -static inline void mark_inode_dirty(struct inode *inode) -{ - if ((inode->i_state & I_DIRTY) != I_DIRTY) - __mark_inode_dirty(inode, I_DIRTY); -} - -static inline void mark_inode_dirty_sync(struct inode *inode) -{ - if (!(inode->i_state & I_DIRTY_SYNC)) - __mark_inode_dirty(inode, I_DIRTY_SYNC); -} - -static inline void mark_inode_dirty_pages(struct inode *inode) -{ - if (inode && !(inode->i_state & I_DIRTY_PAGES)) - __mark_inode_dirty(inode, I_DIRTY_PAGES); -} - struct fown_struct { int pid; /* pid or -pgrp where SIGIO should be sent */ uid_t uid, euid; /* uid/euid of process setting the owner */ @@ -654,6 +627,7 @@ struct quota_mount_options #include <linux/hfs_fs_sb.h> #include <linux/adfs_fs_sb.h> #include <linux/qnx4_fs_sb.h> +#include <linux/reiserfs_fs_sb.h> #include <linux/bfs_fs_sb.h> #include <linux/udf_fs_sb.h> #include <linux/ncp_fs_sb.h> @@ -702,6 +676,7 @@ struct super_block { struct hfs_sb_info hfs_sb; struct adfs_sb_info adfs_sb; struct qnx4_sb_info qnx4_sb; + struct reiserfs_sb_info reiserfs_sb; struct bfs_sb_info bfs_sb; struct udf_sb_info udf_sb; struct ncp_sb_info ncpfs_sb; @@ -815,17 +790,54 @@ struct inode_operations { */ struct super_operations { void (*read_inode) (struct inode *); + + /* reiserfs kludge. reiserfs needs 64 bits of information to + ** find an inode. We are using the read_inode2 call to get + ** that information. We don't like this, and are waiting on some + ** VFS changes for the real solution. + ** iget4 calls read_inode2, iff it is defined + */ + void (*read_inode2) (struct inode *, void *) ; + void (*dirty_inode) (struct inode *); void (*write_inode) (struct inode *, int); void (*put_inode) (struct inode *); void (*delete_inode) (struct inode *); void (*put_super) (struct super_block *); void (*write_super) (struct super_block *); + void (*write_super_lockfs) (struct super_block *); + void (*unlockfs) (struct super_block *); int (*statfs) (struct super_block *, struct statfs *); int (*remount_fs) (struct super_block *, int *, char *); void (*clear_inode) (struct inode *); void (*umount_begin) (struct super_block *); }; +/* Inode state bits.. */ +#define I_DIRTY_SYNC 1 /* Not dirty enough for O_DATASYNC */ +#define I_DIRTY_DATASYNC 2 /* Data-related inode changes pending */ +#define I_DIRTY_PAGES 4 /* Data-related inode changes pending */ +#define I_LOCK 8 +#define I_FREEING 16 +#define I_CLEAR 32 + +#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES) + +extern void __mark_inode_dirty(struct inode *, int); +static inline void mark_inode_dirty(struct inode *inode) +{ + __mark_inode_dirty(inode, I_DIRTY); +} + +static inline void mark_inode_dirty_sync(struct inode *inode) +{ + __mark_inode_dirty(inode, I_DIRTY_SYNC); +} + +static inline void mark_inode_dirty_pages(struct inode *inode) +{ + __mark_inode_dirty(inode, I_DIRTY_PAGES); +} + struct dquot_operations { void (*initialize) (struct inode *, short); void (*drop) (struct inode *); @@ -988,6 +1000,9 @@ extern int fs_may_remount_ro(struct super_block *); extern int try_to_free_buffers(struct page *, int); extern void refile_buffer(struct buffer_head * buf); +/* reiserfs_writepage needs this */ +extern void set_buffer_async_io(struct buffer_head *bh) ; + #define BUF_CLEAN 0 #define BUF_LOCKED 1 /* Buffers scheduled for write */ #define BUF_DIRTY 2 /* Dirty buffers, not yet scheduled for write */ diff --git a/include/linux/lvm.h b/include/linux/lvm.h index 7b4ff94f9..4e95eb73d 100644 --- a/include/linux/lvm.h +++ b/include/linux/lvm.h @@ -9,6 +9,7 @@ * May-July 1998 * January-March,July,September,October,Dezember 1999 * January,February,July,November 2000 + * January 2001 * * lvm is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -57,6 +58,8 @@ * 26/06/2000 - implemented snapshot persistency and resizing support * 02/11/2000 - added hash table size member to lv structure * 12/11/2000 - removed unneeded timestamp definitions + * 24/12/2000 - removed LVM_TO_{CORE,DISK}*, use cpu_{from, to}_le* + * instead - Christoph Hellwig * */ @@ -64,11 +67,10 @@ #ifndef _LVM_H_INCLUDE #define _LVM_H_INCLUDE -#define _LVM_KERNEL_H_VERSION "LVM 0.9 (13/11/2000)" +#define _LVM_KERNEL_H_VERSION "LVM 0.9.1_beta2 (18/01/2001)" #include <linux/config.h> #include <linux/version.h> -#include <endian.h> /* * preprocessor definitions @@ -77,8 +79,7 @@ #define LVM_TOTAL_RESET #ifdef __KERNEL__ -#define LVM_GET_INODE -#undef LVM_HD_NAME /* display nice names in /proc/partitions */ +#undef LVM_HD_NAME /* display nice names in /proc/partitions */ /* lots of debugging output (see driver source) #define DEBUG_LVM_GET_INFO @@ -109,6 +110,7 @@ #ifdef __KERNEL__ #include <linux/spinlock.h> + #include <asm/semaphore.h> #endif /* #ifdef __KERNEL__ */ @@ -216,15 +218,13 @@ * * 1K volume group structure ~200 byte * - * 5K time stamp structure ~ - * * 6K namelist of physical volumes 128 byte each * - * 6k + n * 128byte n logical volume structures ~300 byte each + * 6k + n * ~300byte n logical volume structures ~300 byte each * - * + m * 328byte m physical extent alloc. structs 4 byte each + * + m * 4byte m physical extent alloc. structs 4 byte each * - * End of disk - first physical extent typical 4 megabyte + * End of disk - first physical extent typically 4 megabyte * PE total * * PE size * @@ -292,7 +292,7 @@ #define LVM_MAX_PE_SIZE ( 16L * 1024L * 1024L / SECTOR_SIZE * 1024) /* 16GB in sectors */ #define LVM_DEFAULT_PE_SIZE ( 4096L * 1024 / SECTOR_SIZE) /* 4 MB in sectors */ #define LVM_DEFAULT_STRIPE_SIZE 16L /* 16 KB */ -#define LVM_MIN_STRIPE_SIZE ( PAGE_SIZE>>9) /* PAGESIZE in sectors */ +#define LVM_MIN_STRIPE_SIZE ( PAGE_SIZE/SECTOR_SIZE) /* PAGESIZE in sectors */ #define LVM_MAX_STRIPE_SIZE ( 512L * 1024 / SECTOR_SIZE) /* 512 KB in sectors */ #define LVM_MAX_STRIPES 128 /* max # of stripes */ #define LVM_MAX_SIZE ( 1024LU * 1024 / SECTOR_SIZE * 1024 * 1024) /* 1TB[sectors] */ @@ -326,51 +326,6 @@ COW_table_entries_per_PE - COW_table_chunks_per_PE;}) -/* to disk and to core data conversion macros */ -#if __BYTE_ORDER == __BIG_ENDIAN - -#define LVM_TO_CORE16(x) ( \ - ((uint16_t)((((uint16_t)(x) & 0x00FFU) << 8) | \ - (((uint16_t)(x) & 0xFF00U) >> 8)))) - -#define LVM_TO_DISK16(x) LVM_TO_CORE16(x) - -#define LVM_TO_CORE32(x) ( \ - ((uint32_t)((((uint32_t)(x) & 0x000000FFU) << 24) | \ - (((uint32_t)(x) & 0x0000FF00U) << 8))) \ - (((uint32_t)(x) & 0x00FF0000U) >> 8))) \ - (((uint32_t)(x) & 0xFF000000U) >> 24)))) - -#define LVM_TO_DISK32(x) LVM_TO_CORE32(x) - -#define LVM_TO_CORE64(x) \ - ((uint64_t)((((uint64_t)(x) & 0x00000000000000FFULL) << 56) | \ - (((uint64_t)(x) & 0x000000000000FF00ULL) << 40) | \ - (((uint64_t)(x) & 0x0000000000FF0000ULL) << 24) | \ - (((uint64_t)(x) & 0x00000000FF000000ULL) << 8) | \ - (((uint64_t)(x) & 0x000000FF00000000ULL) >> 8) | \ - (((uint64_t)(x) & 0x0000FF0000000000ULL) >> 24) | \ - (((uint64_t)(x) & 0x00FF000000000000ULL) >> 40) | \ - (((uint64_t)(x) & 0xFF00000000000000ULL) >> 56))) - -#define LVM_TO_DISK64(x) LVM_TO_CORE64(x) - -#elif __BYTE_ORDER == __LITTLE_ENDIAN - -#define LVM_TO_CORE16(x) x -#define LVM_TO_DISK16(x) x -#define LVM_TO_CORE32(x) x -#define LVM_TO_DISK32(x) x -#define LVM_TO_CORE64(x) x -#define LVM_TO_DISK64(x) x - -#else - -#error "__BYTE_ORDER must be defined as __LITTLE_ENDIAN or __BIG_ENDIAN" - -#endif /* #if __BYTE_ORDER == __BIG_ENDIAN */ - - /* * ioctls */ @@ -687,6 +642,8 @@ typedef struct lv_v4 { wait_queue_head_t lv_snapshot_wait; int lv_snapshot_use_rate; void *vg; + + uint lv_allocated_snapshot_le; #else char dummy[200]; #endif diff --git a/include/linux/mm.h b/include/linux/mm.h index 245201241..713b01316 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -386,6 +386,7 @@ extern void clear_page_tables(struct mm_struct *, unsigned long, int); struct page * shmem_nopage(struct vm_area_struct * vma, unsigned long address, int no_share); struct file *shmem_file_setup(char * name, loff_t size); +extern void shmem_lock(struct file * file, int lock); extern int shmem_zero_setup(struct vm_area_struct *); extern void zap_page_range(struct mm_struct *mm, unsigned long address, unsigned long size); @@ -464,6 +465,7 @@ extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int); #else #define __GFP_HIGHMEM 0x0 /* noop */ #endif +#define __GFP_VM 0x20 #define GFP_BUFFER (__GFP_HIGH | __GFP_WAIT) diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 1ee4dd616..4c77c2081 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -74,7 +74,7 @@ struct nfs_readargs { struct nfs_readres { struct nfs_fattr * fattr; - unsigned int count; + __u32 count; int eof; }; @@ -84,7 +84,7 @@ struct nfs_readres { #define NFS_WRITE_MAXIOV 8 struct nfs_writeargs { struct nfs_fh * fh; - __u32 offset; + __u64 offset; __u32 count; enum nfs3_stable_how stable; unsigned int nriov; diff --git a/include/linux/raid/md_u.h b/include/linux/raid/md_u.h index c96b0e404..22a154380 100644 --- a/include/linux/raid/md_u.h +++ b/include/linux/raid/md_u.h @@ -22,6 +22,7 @@ #define GET_ARRAY_INFO _IOR (MD_MAJOR, 0x11, mdu_array_info_t) #define GET_DISK_INFO _IOR (MD_MAJOR, 0x12, mdu_disk_info_t) #define PRINT_RAID_DEBUG _IO (MD_MAJOR, 0x13) +#define RAID_AUTORUN _IO (MD_MAJOR, 0x14) /* configuration */ #define CLEAR_ARRAY _IO (MD_MAJOR, 0x20) diff --git a/include/linux/reiserfs_fs.h b/include/linux/reiserfs_fs.h new file mode 100644 index 000000000..277281d6d --- /dev/null +++ b/include/linux/reiserfs_fs.h @@ -0,0 +1,2074 @@ +/* + * Copyright 1996, 1997, 1998 Hans Reiser, see reiserfs/README for licensing and copyright details + */ + + /* this file has an amazingly stupid + name, yura please fix it to be + reiserfs.h, and merge all the rest + of our .h files that are in this + directory into it. */ + + +#ifndef _LINUX_REISER_FS_H +#define _LINUX_REISER_FS_H + + +#include <linux/types.h> +#ifdef __KERNEL__ +#include <linux/malloc.h> +#include <linux/tqueue.h> +#endif + +/* + * include/linux/reiser_fs.h + * + * Reiser File System constants and structures + * + */ + +/* in reading the #defines, it may help to understand that they employ + the following abbreviations: + + B = Buffer + I = Item header + H = Height within the tree (should be changed to LEV) + N = Number of the item in the node + STAT = stat data + DEH = Directory Entry Header + EC = Entry Count + E = Entry number + UL = Unsigned Long + BLKH = BLocK Header + UNFM = UNForMatted node + DC = Disk Child + P = Path + + These #defines are named by concatenating these abbreviations, + where first comes the arguments, and last comes the return value, + of the macro. + +*/ + + /* Vladimir, what is the story with + new_get_new_buffer nowadays? I + want a complete explanation written + here. */ + +/* NEW_GET_NEW_BUFFER will try to allocate new blocks better */ +/*#define NEW_GET_NEW_BUFFER*/ +#define OLD_GET_NEW_BUFFER + + /* Vladimir, what about this one too? */ +/* if this is undefined, all inode changes get into stat data immediately, if it can be found in RAM */ +#define DIRTY_LATER + +/* enable journalling */ +#define ENABLE_JOURNAL + +#ifdef __KERNEL__ + +/* #define REISERFS_CHECK */ + +#define REISERFS_PREALLOCATE +#endif +#define PREALLOCATION_SIZE 8 + +/* if this is undefined, all inode changes get into stat data + immediately, if it can be found in RAM */ +#define DIRTY_LATER + + +/*#define READ_LOCK_REISERFS*/ + + +/* n must be power of 2 */ +#define _ROUND_UP(x,n) (((x)+(n)-1u) & ~((n)-1u)) + +// to be ok for alpha and others we have to align structures to 8 byte +// boundary. +// FIXME: do not change 4 by anything else: there is code which relies on that + /* what 4? -Hans */ +#define ROUND_UP(x) _ROUND_UP(x,8LL) + +/* debug levels. Right now, CONFIG_REISERFS_CHECK means print all debug +** messages. +*/ +#define REISERFS_DEBUG_CODE 5 /* extra messages to help find/debug errors */ + +/* + * Disk Data Structures + */ + +/***************************************************************************/ +/* SUPER BLOCK */ +/***************************************************************************/ + +/* + * Structure of super block on disk, a version of which in RAM is often accessed as s->u.reiserfs_sb.s_rs + * the version in RAM is part of a larger structure containing fields never written to disk. + */ + + /* used by gcc */ +#define REISERFS_SUPER_MAGIC 0x52654973 + /* used by file system utilities that + look at the superblock, etc. */ +#define REISERFS_SUPER_MAGIC_STRING "ReIsErFs" +#define REISER2FS_SUPER_MAGIC_STRING "ReIsEr2Fs" + +extern inline int is_reiserfs_magic_string (struct reiserfs_super_block * rs) +{ + return (!strncmp (rs->s_magic, REISERFS_SUPER_MAGIC_STRING, + strlen ( REISERFS_SUPER_MAGIC_STRING)) || + !strncmp (rs->s_magic, REISER2FS_SUPER_MAGIC_STRING, + strlen ( REISER2FS_SUPER_MAGIC_STRING))); +} + + /* ReiserFS leaves the first 64k unused, + so that partition labels have enough + space. If someone wants to write a + fancy bootloader that needs more than + 64k, let us know, and this will be + increased in size. This number must + be larger than than the largest block + size on any platform, or code will + break. -Hans */ +#define REISERFS_DISK_OFFSET_IN_BYTES (64 * 1024) +#define REISERFS_FIRST_BLOCK unused_define + +/* the spot for the super in versions 3.5 - 3.5.10 (inclusive) */ +#define REISERFS_OLD_DISK_OFFSET_IN_BYTES (8 * 1024) + + +// reiserfs internal error code (used by search_by_key adn fix_nodes)) +#define CARRY_ON 0 +#define REPEAT_SEARCH -1 +#define IO_ERROR -2 +#define NO_DISK_SPACE -3 +#define NO_BALANCING_NEEDED (-4) +#define NO_MORE_UNUSED_CONTIGUOUS_BLOCKS (-5) + +//#define SCHEDULE_OCCURRED 1 +//#define PATH_INCORRECT 2 + +//#define NO_DISK_SPACE (-1) + + + +typedef unsigned long b_blocknr_t; +typedef __u32 unp_t; + + /* who is responsible for this + completely uncommented struct? */ +struct unfm_nodeinfo { + /* This is what? */ + unp_t unfm_nodenum; + /* now this I know what it is, and + most of the people on our project + know what it is, but I bet nobody + new I hire will have a clue. */ + unsigned short unfm_freespace; +}; + + +/* when reiserfs_file_write is called with a byte count >= MIN_PACK_ON_CLOSE, +** it sets the inode to pack on close, and when extending the file, will only +** use unformatted nodes. +** +** This is a big speed up for the journal, which is badly hurt by direct->indirect +** conversions (they must be logged). +*/ +#define MIN_PACK_ON_CLOSE 512 + +/* the defines below say, that if file size is >= + DIRECT_TAIL_SUPPRESSION_SIZE * blocksize, then if tail is longer + than MAX_BYTES_SUPPRESS_DIRECT_TAIL, it will be stored in + unformatted node */ +#define DIRECT_TAIL_SUPPRESSION_SIZE 1024 +#define MAX_BYTES_SUPPRESS_DIRECT_TAIL 1024 + +#if 0 + +// +#define mark_file_with_tail(inode,offset) \ +{\ +inode->u.reiserfs_i.i_has_tail = 1;\ +} + +#define mark_file_without_tail(inode) \ +{\ +inode->u.reiserfs_i.i_has_tail = 0;\ +} + +#endif + +// this says about version of all items (but stat data) the object +// consists of +#define inode_items_version(inode) ((inode)->u.reiserfs_i.i_version) + + +/* We store tail in unformatted node if it is too big to fit into a + formatted node or if DIRECT_TAIL_SUPPRESSION_SIZE, + MAX_BYTES_SUPPRESS_DIRECT_TAIL and file size say that. */ +/* #define STORE_TAIL_IN_UNFM(n_file_size,n_tail_size,n_block_size) \ */ +/* ( ((n_tail_size) > MAX_DIRECT_ITEM_LEN(n_block_size)) || \ */ +/* ( ( (n_file_size) >= (n_block_size) * DIRECT_TAIL_SUPPRESSION_SIZE ) && \ */ +/* ( (n_tail_size) >= MAX_BYTES_SUPPRESS_DIRECT_TAIL ) ) ) */ + + /* This is an aggressive tail suppression policy, I am hoping it + improves our benchmarks. The principle behind it is that + percentage space saving is what matters, not absolute space + saving. This is non-intuitive, but it helps to understand it if + you consider that the cost to access 4 blocks is not much more + than the cost to access 1 block, if you have to do a seek and + rotate. A tail risks a non-linear disk access that is + significant as a percentage of total time cost for a 4 block file + and saves an amount of space that is less significant as a + percentage of space, or so goes the hypothesis. -Hans */ +#define STORE_TAIL_IN_UNFM(n_file_size,n_tail_size,n_block_size) \ +(\ + (!(n_tail_size)) || \ + (((n_tail_size) > MAX_DIRECT_ITEM_LEN(n_block_size)) || \ + ( (n_file_size) >= (n_block_size) * 4 ) || \ + ( ( (n_file_size) >= (n_block_size) * 3 ) && \ + ( (n_tail_size) >= (MAX_DIRECT_ITEM_LEN(n_block_size))/4) ) || \ + ( ( (n_file_size) >= (n_block_size) * 2 ) && \ + ( (n_tail_size) >= (MAX_DIRECT_ITEM_LEN(n_block_size))/2) ) || \ + ( ( (n_file_size) >= (n_block_size) ) && \ + ( (n_tail_size) >= (MAX_DIRECT_ITEM_LEN(n_block_size) * 3)/4) ) ) \ +) + + +/* + * values for s_state field + */ +#define REISERFS_VALID_FS 1 +#define REISERFS_ERROR_FS 2 + + + +/***************************************************************************/ +/* KEY & ITEM HEAD */ +/***************************************************************************/ + +// +// we do support for old format of reiserfs: the problem is to +// distinuquish keys with 32 bit offset and keys with 60 bit ones. On +// leaf level we use ih_version of struct item_head (was +// ih_reserved). For all old items it is set to 0 +// (ITEM_VERSION_1). For new items it is ITEM_VERSION_2. On internal +// levels we have to know version of item key belongs to. +// +#define ITEM_VERSION_1 0 +#define ITEM_VERSION_2 1 + + +/* loff_t - long long */ + + +// +// directories use this key as well as old files +// +struct offset_v1 { + __u32 k_offset; + __u32 k_uniqueness; +} __attribute__ ((__packed__)); + +struct offset_v2 { + __u64 k_offset:60; + __u64 k_type: 4; +} __attribute__ ((__packed__)); + + + +/* Key of an item determines its location in the S+tree, and + is composed of 4 components */ +struct key { + __u32 k_dir_id; /* packing locality: by default parent + directory object id */ + __u32 k_objectid; /* object identifier */ + union { + struct offset_v1 k_offset_v1; + struct offset_v2 k_offset_v2; + } __attribute__ ((__packed__)) u; +} __attribute__ ((__packed__)); + + +struct cpu_key { + struct key on_disk_key; + int version; + int key_length; /* 3 in all cases but direct2indirect and + indirect2direct conversion */ +}; + + + + + + + + /* Our function for comparing keys can compare keys of different + lengths. It takes as a parameter the length of the keys it is to + compare. These defines are used in determining what is to be + passed to it as that parameter. */ +#define REISERFS_FULL_KEY_LEN 4 + +#define REISERFS_SHORT_KEY_LEN 2 + +/* The result of the key compare */ +#define FIRST_GREATER 1 +#define SECOND_GREATER -1 +#define KEYS_IDENTICAL 0 +#define KEY_FOUND 1 +#define KEY_NOT_FOUND 0 + + +#define KEY_SIZE (sizeof(struct key)) +#define SHORT_KEY_SIZE (sizeof (__u32) + sizeof (__u32)) + +/* return values for search_by_key and clones */ +#define ITEM_FOUND 1 +#define ITEM_NOT_FOUND 0 +#define ENTRY_FOUND 1 +#define ENTRY_NOT_FOUND 0 +#define DIRECTORY_NOT_FOUND -1 +#define REGULAR_FILE_FOUND -2 +#define DIRECTORY_FOUND -3 +#define BYTE_FOUND 1 +#define BYTE_NOT_FOUND 0 +#define FILE_NOT_FOUND -1 + +#define POSITION_FOUND 1 +#define POSITION_NOT_FOUND 0 + +// return values for reiserfs_find_entry and search_by_entry_key +#define NAME_FOUND 1 +#define NAME_NOT_FOUND 0 +#define GOTO_PREVIOUS_ITEM 2 +#define NAME_FOUND_INVISIBLE 3 + + + +/* Everything in the filesystem is stored as a set of items. The + item head contains the key of the item, its free space (for + indirect items) and specifies the location of the item itself + within the block. */ + +struct item_head +{ + struct key ih_key; /* Everything in the tree is found by searching for it based on its key.*/ + + /* This is bloat, this should be part + of the item not the item + header. -Hans */ + union { + __u16 ih_free_space_reserved; /* The free space in the last unformatted node of an indirect item if this + is an indirect item. This equals 0xFFFF iff this is a direct item or + stat data item. Note that the key, not this field, is used to determine + the item type, and thus which field this union contains. */ + __u16 ih_entry_count; /* Iff this is a directory item, this field equals the number of directory + entries in the directory item. */ + } __attribute__ ((__packed__)) u; + __u16 ih_item_len; /* total size of the item body */ + __u16 ih_item_location; /* an offset to the item body within the block */ + /* I thought we were going to use this + for having lots of item types? Why + don't you use this for item type + not item version. That is how you + talked me into this field a year + ago, remember? I am still not + convinced it needs to be 16 bits + (for at least many years), but at + least I can sympathize with that + hope. Change the name from version + to type, and tell people not to use + FFFF in case 16 bits is someday too + small and needs to be extended:-). */ + __u16 ih_version; /* 0 for all old items, 2 for new + ones. Highest bit is set by fsck + temporary, cleaned after all done */ +} __attribute__ ((__packed__)); +/* size of item header */ +#define IH_SIZE (sizeof(struct item_head)) + +#define ih_free_space(ih) le16_to_cpu((ih)->u.ih_free_space_reserved) +#define ih_version(ih) le16_to_cpu((ih)->ih_version) +#define ih_entry_count(ih) le16_to_cpu((ih)->u.ih_entry_count) +#define ih_location(ih) le16_to_cpu((ih)->ih_item_location) +#define ih_item_len(ih) le16_to_cpu((ih)->ih_item_len) + +#define put_ih_free_space(ih, val) do { (ih)->u.ih_free_space_reserved = cpu_to_le16(val); } while(0) +#define put_ih_version(ih, val) do { (ih)->ih_version = cpu_to_le16(val); } while (0) +#define put_ih_entry_count(ih, val) do { (ih)->u.ih_entry_count = cpu_to_le16(val); } while (0) +#define put_ih_location(ih, val) do { (ih)->ih_item_location = cpu_to_le16(val); } while (0) +#define put_ih_item_len(ih, val) do { (ih)->ih_item_len = cpu_to_le16(val); } while (0) + + +// FIXME: now would that work for other than i386 archs +#define unreachable_item(ih) (ih->ih_version & (1 << 15)) + +#define get_ih_free_space(ih) (ih_version (ih) == ITEM_VERSION_2 ? 0 : ih_free_space (ih)) +#define set_ih_free_space(ih,val) put_ih_free_space((ih), ((ih_version(ih) == ITEM_VERSION_2) ? 0 : (val))) + + +// +// there are 5 item types currently +// +#define TYPE_STAT_DATA 0 +#define TYPE_INDIRECT 1 +#define TYPE_DIRECT 2 +#define TYPE_DIRENTRY 3 +#define TYPE_ANY 15 // FIXME: comment is required + +// +// in old version uniqueness field shows key type +// +#define V1_SD_UNIQUENESS 0 +#define V1_INDIRECT_UNIQUENESS 0xfffffffe +#define V1_DIRECT_UNIQUENESS 0xffffffff +#define V1_DIRENTRY_UNIQUENESS 500 +#define V1_ANY_UNIQUENESS 555 // FIXME: comment is required + +// +// here are conversion routines +// +extern inline int uniqueness2type (__u32 uniqueness) +{ + switch (uniqueness) { + case V1_SD_UNIQUENESS: return TYPE_STAT_DATA; + case V1_INDIRECT_UNIQUENESS: return TYPE_INDIRECT; + case V1_DIRECT_UNIQUENESS: return TYPE_DIRECT; + case V1_DIRENTRY_UNIQUENESS: return TYPE_DIRENTRY; + } +/* + if (uniqueness != V1_ANY_UNIQUENESS) { + printk ("uniqueness %d\n", uniqueness); + BUG (); + } +*/ + return TYPE_ANY; +} + +extern inline __u32 type2uniqueness (int type) +{ + switch (type) { + case TYPE_STAT_DATA: return V1_SD_UNIQUENESS; + case TYPE_INDIRECT: return V1_INDIRECT_UNIQUENESS; + case TYPE_DIRECT: return V1_DIRECT_UNIQUENESS; + case TYPE_DIRENTRY: return V1_DIRENTRY_UNIQUENESS; + } + /* + if (type != TYPE_ANY) + BUG (); + */ + return V1_ANY_UNIQUENESS; +} + + +// +// key is pointer to on disk key which is stored in le, result is cpu, +// there is no way to get version of object from key, so, provide +// version to these defines +// +extern inline loff_t le_key_k_offset (int version, struct key * key) +{ + return (version == ITEM_VERSION_1) ? key->u.k_offset_v1.k_offset : + le64_to_cpu (key->u.k_offset_v2.k_offset); +} +extern inline loff_t le_ih_k_offset (struct item_head * ih) +{ + return le_key_k_offset (ih_version (ih), &(ih->ih_key)); +} + + +extern inline loff_t le_key_k_type (int version, struct key * key) +{ + return (version == ITEM_VERSION_1) ? uniqueness2type (key->u.k_offset_v1.k_uniqueness) : + le16_to_cpu (key->u.k_offset_v2.k_type); +} +extern inline loff_t le_ih_k_type (struct item_head * ih) +{ + return le_key_k_type (ih_version (ih), &(ih->ih_key)); +} + + +extern inline void set_le_key_k_offset (int version, struct key * key, loff_t offset) +{ + (version == ITEM_VERSION_1) ? (key->u.k_offset_v1.k_offset = offset) : + (key->u.k_offset_v2.k_offset = cpu_to_le64 (offset)); +} +extern inline void set_le_ih_k_offset (struct item_head * ih, loff_t offset) +{ + set_le_key_k_offset (ih_version (ih), &(ih->ih_key), offset); +} + + + +extern inline void set_le_key_k_type (int version, struct key * key, int type) +{ + (version == ITEM_VERSION_1) ? (key->u.k_offset_v1.k_uniqueness = type2uniqueness (type)) : + (key->u.k_offset_v2.k_type = cpu_to_le16 (type)); +} +extern inline void set_le_ih_k_type (struct item_head * ih, int type) +{ + set_le_key_k_type (ih_version (ih), &(ih->ih_key), type); +} + + +#define is_direntry_le_key(version,key) (le_key_k_type (version, key) == TYPE_DIRENTRY) +#define is_direct_le_key(version,key) (le_key_k_type (version, key) == TYPE_DIRECT) +#define is_indirect_le_key(version,key) (le_key_k_type (version, key) == TYPE_INDIRECT) +#define is_statdata_le_key(version,key) (le_key_k_type (version, key) == TYPE_STAT_DATA) + +// +// item header has version. +// +#define is_direntry_le_ih(ih) is_direntry_le_key (ih_version (ih), &((ih)->ih_key)) +#define is_direct_le_ih(ih) is_direct_le_key (ih_version (ih), &((ih)->ih_key)) +#define is_indirect_le_ih(ih) is_indirect_le_key (ih_version(ih), &((ih)->ih_key)) +#define is_statdata_le_ih(ih) is_statdata_le_key (ih_version (ih), &((ih)->ih_key)) + + + +// +// key is pointer to cpu key, result is cpu +// +extern inline loff_t cpu_key_k_offset (struct cpu_key * key) +{ + return (key->version == ITEM_VERSION_1) ? key->on_disk_key.u.k_offset_v1.k_offset : + key->on_disk_key.u.k_offset_v2.k_offset; +} + +extern inline loff_t cpu_key_k_type (struct cpu_key * key) +{ + return (key->version == ITEM_VERSION_1) ? uniqueness2type (key->on_disk_key.u.k_offset_v1.k_uniqueness) : + key->on_disk_key.u.k_offset_v2.k_type; +} + +extern inline void set_cpu_key_k_offset (struct cpu_key * key, loff_t offset) +{ + (key->version == ITEM_VERSION_1) ? (key->on_disk_key.u.k_offset_v1.k_offset = offset) : + (key->on_disk_key.u.k_offset_v2.k_offset = offset); +} + + +extern inline void set_cpu_key_k_type (struct cpu_key * key, int type) +{ + (key->version == ITEM_VERSION_1) ? (key->on_disk_key.u.k_offset_v1.k_uniqueness = type2uniqueness (type)) : + (key->on_disk_key.u.k_offset_v2.k_type = type); +} + +extern inline void cpu_key_k_offset_dec (struct cpu_key * key) +{ + if (key->version == ITEM_VERSION_1) + key->on_disk_key.u.k_offset_v1.k_offset --; + else + key->on_disk_key.u.k_offset_v2.k_offset --; +} + + +#define is_direntry_cpu_key(key) (cpu_key_k_type (key) == TYPE_DIRENTRY) +#define is_direct_cpu_key(key) (cpu_key_k_type (key) == TYPE_DIRECT) +#define is_indirect_cpu_key(key) (cpu_key_k_type (key) == TYPE_INDIRECT) +#define is_statdata_cpu_key(key) (cpu_key_k_type (key) == TYPE_STAT_DATA) + + +/* are these used ? */ +#define is_direntry_cpu_ih(ih) (is_direntry_cpu_key (&((ih)->ih_key))) +#define is_direct_cpu_ih(ih) (is_direct_cpu_key (&((ih)->ih_key))) +#define is_indirect_cpu_ih(ih) (is_indirect_cpu_key (&((ih)->ih_key))) +#define is_statdata_cpu_ih(ih) (is_statdata_cpu_key (&((ih)->ih_key))) + + + + + +#define I_K_KEY_IN_ITEM(p_s_ih, p_s_key, n_blocksize) \ + ( ! COMP_SHORT_KEYS(p_s_ih, p_s_key) && \ + I_OFF_BYTE_IN_ITEM(p_s_ih, k_offset (p_s_key), n_blocksize) ) + +/* maximal length of item */ +#define MAX_ITEM_LEN(block_size) (block_size - BLKH_SIZE - IH_SIZE) +#define MIN_ITEM_LEN 1 + + +/* object identifier for root dir */ +#define REISERFS_ROOT_OBJECTID 2 +#define REISERFS_ROOT_PARENT_OBJECTID 1 +extern struct key root_key; + + + + +/* + * Picture represents a leaf of the S+tree + * ______________________________________________________ + * | | Array of | | | + * |Block | Object-Item | F r e e | Objects- | + * | head | Headers | S p a c e | Items | + * |______|_______________|___________________|___________| + */ + +/* Header of a disk block. More precisely, header of a formatted leaf + or internal node, and not the header of an unformatted node. */ +struct block_head { + __u16 blk_level; /* Level of a block in the tree. */ + __u16 blk_nr_item; /* Number of keys/items in a block. */ + __u16 blk_free_space; /* Block free space in bytes. */ + __u16 blk_reserved; + /* dump this in v4/planA */ + struct key blk_right_delim_key; /* kept only for compatibility */ +}; + +#define BLKH_SIZE (sizeof(struct block_head)) + +/* + * values for blk_level field of the struct block_head + */ + +#define FREE_LEVEL 0 /* when node gets removed from the tree its + blk_level is set to FREE_LEVEL. It is then + used to see whether the node is still in the + tree */ + +#define DISK_LEAF_NODE_LEVEL 1 /* Leaf node level.*/ + +/* Given the buffer head of a formatted node, resolve to the block head of that node. */ +#define B_BLK_HEAD(p_s_bh) ((struct block_head *)((p_s_bh)->b_data)) +/* Number of items that are in buffer. */ +#define B_NR_ITEMS(p_s_bh) (le16_to_cpu ( B_BLK_HEAD(p_s_bh)->blk_nr_item )) +#define B_LEVEL(bh) (le16_to_cpu ( B_BLK_HEAD(bh)->blk_level )) +#define B_FREE_SPACE(bh) (le16_to_cpu ( B_BLK_HEAD(bh)->blk_free_space )) + +#define PUT_B_NR_ITEMS(p_s_bh) do { B_BLK_HEAD(p_s_bh)->blk_nr_item = cpu_to_le16(val); } while (0) +#define PUT_B_LEVEL(bh, val) do { B_BLK_HEAD(bh)->blk_level = cpu_to_le16(val); } while (0) +#define PUT_B_FREE_SPACE(bh) do { B_BLK_HEAD(bh)->blk_free_space = cpu_to_le16(val); } while (0) + +/* Get right delimiting key. */ +#define B_PRIGHT_DELIM_KEY(p_s_bh) ( &(B_BLK_HEAD(p_s_bh)->blk_right_delim_key) ) + +/* Does the buffer contain a disk leaf. */ +#define B_IS_ITEMS_LEVEL(p_s_bh) ( B_BLK_HEAD(p_s_bh)->blk_level == DISK_LEAF_NODE_LEVEL ) + +/* Does the buffer contain a disk internal node */ +#define B_IS_KEYS_LEVEL(p_s_bh) ( B_BLK_HEAD(p_s_bh)->blk_level > DISK_LEAF_NODE_LEVEL &&\ + B_BLK_HEAD(p_s_bh)->blk_level <= MAX_HEIGHT ) + + + + +/***************************************************************************/ +/* STAT DATA */ +/***************************************************************************/ + + +// +// old stat data is 32 bytes long. We are going to distinguish new one by +// different size +// +struct stat_data_v1 +{ + __u16 sd_mode; /* file type, permissions */ + __u16 sd_nlink; /* number of hard links */ + __u16 sd_uid; /* owner */ + __u16 sd_gid; /* group */ + __u32 sd_size; /* file size */ + __u32 sd_atime; /* time of last access */ + __u32 sd_mtime; /* time file was last modified */ + __u32 sd_ctime; /* time inode (stat data) was last changed (except changes to sd_atime and sd_mtime) */ + union { + __u32 sd_rdev; + __u32 sd_blocks; /* number of blocks file uses */ + } __attribute__ ((__packed__)) u; + __u32 sd_first_direct_byte; /* first byte of file which is stored + in a direct item: except that if it + equals 1 it is a symlink and if it + equals ~(__u32)0 there is no + direct item. The existence of this + field really grates on me. Let's + replace it with a macro based on + sd_size and our tail suppression + policy. Someday. -Hans */ +} __attribute__ ((__packed__)); + +#define SD_V1_SIZE (sizeof(struct stat_data_v1)) + + +/* Stat Data on disk (reiserfs version of UFS disk inode minus the + address blocks) */ +struct stat_data { + __u16 sd_mode; /* file type, permissions */ + __u16 sd_reserved; + __u32 sd_nlink; /* number of hard links */ + __u64 sd_size; /* file size */ + __u32 sd_uid; /* owner */ + __u32 sd_gid; /* group */ + __u32 sd_atime; /* time of last access */ + __u32 sd_mtime; /* time file was last modified */ + __u32 sd_ctime; /* time inode (stat data) was last changed (except changes to sd_atime and sd_mtime) */ + __u32 sd_blocks; + union { + __u32 sd_rdev; + //__u32 sd_first_direct_byte; + /* first byte of file which is stored in a + direct item: except that if it equals 1 + it is a symlink and if it equals + ~(__u32)0 there is no direct item. The + existence of this field really grates + on me. Let's replace it with a macro + based on sd_size and our tail + suppression policy? */ + } __attribute__ ((__packed__)) u; +} __attribute__ ((__packed__)); +// +// this is 40 bytes long +// +#define SD_SIZE (sizeof(struct stat_data)) + +#define stat_data_v1(ih) (ih_version (ih) == ITEM_VERSION_1) + + +/***************************************************************************/ +/* DIRECTORY STRUCTURE */ +/***************************************************************************/ +/* + Picture represents the structure of directory items + ________________________________________________ + | Array of | | | | | | + | directory |N-1| N-2 | .... | 1st |0th| + | entry headers | | | | | | + |_______________|___|_____|________|_______|___| + <---- directory entries ------> + + First directory item has k_offset component 1. We store "." and ".." + in one item, always, we never split "." and ".." into differing + items. This makes, among other things, the code for removing + directories simpler. */ +#define SD_OFFSET 0 +#define SD_UNIQUENESS 0 +#define DOT_OFFSET 1 +#define DOT_DOT_OFFSET 2 +#define DIRENTRY_UNIQUENESS 500 + +/* */ +#define FIRST_ITEM_OFFSET 1 + +/* + Q: How to get key of object pointed to by entry from entry? + + A: Each directory entry has its header. This header has deh_dir_id and deh_objectid fields, those are key + of object, entry points to */ + +/* NOT IMPLEMENTED: + Directory will someday contain stat data of object */ + + + +struct reiserfs_de_head +{ + __u32 deh_offset; /* third component of the directory entry key */ + __u32 deh_dir_id; /* objectid of the parent directory of the object, that is referenced + by directory entry */ + __u32 deh_objectid; /* objectid of the object, that is referenced by directory entry */ + __u16 deh_location; /* offset of name in the whole item */ + __u16 deh_state; /* whether 1) entry contains stat data (for future), and 2) whether + entry is hidden (unlinked) */ +} __attribute__ ((__packed__)); +#define DEH_SIZE sizeof(struct reiserfs_de_head) + +/* empty directory contains two entries "." and ".." and their headers */ +#define EMPTY_DIR_SIZE \ +(DEH_SIZE * 2 + ROUND_UP (strlen (".")) + ROUND_UP (strlen (".."))) + +/* old format directories have this size when empty */ +#define EMPTY_DIR_SIZE_V1 (DEH_SIZE * 2 + 3) + +#define DEH_Statdata 0 /* not used now */ +#define DEH_Visible 2 + +/* bitops which deals with unaligned addrs; + needed for alpha port. --zam */ +#ifdef __alpha__ +# define ADDR_UNALIGNED_BITS (5) +#endif + +#ifdef ADDR_UNALIGNED_BITS + +# define aligned_address(addr) ((void *)((long)(addr) & ~((1UL << ADDR_UNALIGNED_BITS) - 1))) +# define unaligned_offset(addr) (((int)((long)(addr) & ((1 << ADDR_UNALIGNED_BITS) - 1))) << 3) + +# define set_bit_unaligned(nr, addr) set_bit((nr) + unaligned_offset(addr), aligned_address(addr)) +# define clear_bit_unaligned(nr, addr) clear_bit((nr) + unaligned_offset(addr), aligned_address(addr)) +# define test_bit_unaligned(nr, addr) test_bit((nr) + unaligned_offset(addr), aligned_address(addr)) + +#else + +# define set_bit_unaligned(nr, addr) set_bit(nr, addr) +# define clear_bit_unaligned(nr, addr) clear_bit(nr, addr) +# define test_bit_unaligned(nr, addr) test_bit(nr, addr) + +#endif + +#define deh_dir_id(deh) (__le32_to_cpu ((deh)->deh_dir_id)) +#define deh_objectid(deh) (__le32_to_cpu ((deh)->deh_objectid)) +#define deh_offset(deh) (__le32_to_cpu ((deh)->deh_offset)) + + +#define mark_de_with_sd(deh) set_bit_unaligned (DEH_Statdata, &((deh)->deh_state)) +#define mark_de_without_sd(deh) clear_bit_unaligned (DEH_Statdata, &((deh)->deh_state)) +#define mark_de_visible(deh) set_bit_unaligned (DEH_Visible, &((deh)->deh_state)) +#define mark_de_hidden(deh) clear_bit_unaligned (DEH_Visible, &((deh)->deh_state)) + +#define de_with_sd(deh) test_bit_unaligned (DEH_Statdata, &((deh)->deh_state)) +#define de_visible(deh) test_bit_unaligned (DEH_Visible, &((deh)->deh_state)) +#define de_hidden(deh) !test_bit_unaligned (DEH_Visible, &((deh)->deh_state)) + +/* compose directory item containing "." and ".." entries (entries are + not aligned to 4 byte boundary) */ +extern inline void make_empty_dir_item_v1 (char * body, __u32 dirid, __u32 objid, + __u32 par_dirid, __u32 par_objid) +{ + struct reiserfs_de_head * deh; + + memset (body, 0, EMPTY_DIR_SIZE_V1); + deh = (struct reiserfs_de_head *)body; + + /* direntry header of "." */ + deh[0].deh_offset = cpu_to_le32 (DOT_OFFSET); + deh[0].deh_dir_id = cpu_to_le32 (dirid); + deh[0].deh_objectid = cpu_to_le32 (objid); + deh[0].deh_location = cpu_to_le16 (EMPTY_DIR_SIZE_V1 - strlen (".")); + deh[0].deh_state = 0; + mark_de_visible(&(deh[0])); + + /* direntry header of ".." */ + deh[1].deh_offset = cpu_to_le32 (DOT_DOT_OFFSET); + /* key of ".." for the root directory */ + deh[1].deh_dir_id = cpu_to_le32 (par_dirid); + deh[1].deh_objectid = cpu_to_le32 (par_objid); + deh[1].deh_location = cpu_to_le16 (le16_to_cpu (deh[0].deh_location) - strlen ("..")); + deh[1].deh_state = 0; + mark_de_visible(&(deh[1])); + + /* copy ".." and "." */ + memcpy (body + deh[0].deh_location, ".", 1); + memcpy (body + deh[1].deh_location, "..", 2); +} + +/* compose directory item containing "." and ".." entries */ +extern inline void make_empty_dir_item (char * body, __u32 dirid, __u32 objid, + __u32 par_dirid, __u32 par_objid) +{ + struct reiserfs_de_head * deh; + + memset (body, 0, EMPTY_DIR_SIZE); + deh = (struct reiserfs_de_head *)body; + + /* direntry header of "." */ + deh[0].deh_offset = cpu_to_le32 (DOT_OFFSET); + deh[0].deh_dir_id = cpu_to_le32 (dirid); + deh[0].deh_objectid = cpu_to_le32 (objid); + deh[0].deh_location = cpu_to_le16 (EMPTY_DIR_SIZE - ROUND_UP (strlen ("."))); + deh[0].deh_state = 0; + mark_de_visible(&(deh[0])); + + /* direntry header of ".." */ + deh[1].deh_offset = cpu_to_le32 (DOT_DOT_OFFSET); + /* key of ".." for the root directory */ + deh[1].deh_dir_id = cpu_to_le32 (par_dirid); + deh[1].deh_objectid = cpu_to_le32 (par_objid); + deh[1].deh_location = cpu_to_le16 (le16_to_cpu (deh[0].deh_location) - ROUND_UP (strlen (".."))); + deh[1].deh_state = 0; + mark_de_visible(&(deh[1])); + + /* copy ".." and "." */ + memcpy (body + deh[0].deh_location, ".", 1); + memcpy (body + deh[1].deh_location, "..", 2); +} + + +/* array of the entry headers */ + /* get item body */ +#define B_I_PITEM(bh,ih) ( (bh)->b_data + (ih)->ih_item_location ) +#define B_I_DEH(bh,ih) ((struct reiserfs_de_head *)(B_I_PITEM(bh,ih))) + +/* length of the directory entry in directory item. This define + calculates length of i-th directory entry using directory entry + locations from dir entry head. When it calculates length of 0-th + directory entry, it uses length of whole item in place of entry + location of the non-existent following entry in the calculation. + See picture above.*/ +/* +#define I_DEH_N_ENTRY_LENGTH(ih,deh,i) \ +((i) ? (((deh)-1)->deh_location - (deh)->deh_location) : ((ih)->ih_item_len) - (deh)->deh_location) +*/ +extern inline int entry_length (struct buffer_head * bh, struct item_head * ih, + int pos_in_item) +{ + struct reiserfs_de_head * deh; + + deh = B_I_DEH (bh, ih) + pos_in_item; + if (pos_in_item) + return (le16_to_cpu ((deh - 1)->deh_location) - le16_to_cpu (deh->deh_location)); + return (le16_to_cpu (ih->ih_item_len) - le16_to_cpu (deh->deh_location)); +} + + + +/* number of entries in the directory item, depends on ENTRY_COUNT being at the start of directory dynamic data. */ +#define I_ENTRY_COUNT(ih) ((ih)->u.ih_entry_count) + + +/* name by bh, ih and entry_num */ +#define B_I_E_NAME(bh,ih,entry_num) ((char *)(bh->b_data + ih->ih_item_location + (B_I_DEH(bh,ih)+(entry_num))->deh_location)) + +// two entries per block (at least) +//#define REISERFS_MAX_NAME_LEN(block_size) +//((block_size - BLKH_SIZE - IH_SIZE - DEH_SIZE * 2) / 2) + +// two entries per block (at least) +#define REISERFS_MAX_NAME_LEN(block_size) 255 + + + + +/* this structure is used for operations on directory entries. It is + not a disk structure. */ +/* When reiserfs_find_entry or search_by_entry_key find directory + entry, they return filled reiserfs_dir_entry structure */ +struct reiserfs_dir_entry +{ + struct buffer_head * de_bh; + int de_item_num; + struct item_head * de_ih; + int de_entry_num; + struct reiserfs_de_head * de_deh; + int de_entrylen; + int de_namelen; + char * de_name; + char * de_gen_number_bit_string; + + __u32 de_dir_id; + __u32 de_objectid; + + struct cpu_key de_entry_key; +}; + +/* these defines are useful when a particular member of a reiserfs_dir_entry is needed */ + +/* pointer to file name, stored in entry */ +#define B_I_DEH_ENTRY_FILE_NAME(bh,ih,deh) (B_I_PITEM (bh, ih) + (deh)->deh_location) + +/* length of name */ +#define I_DEH_N_ENTRY_FILE_NAME_LENGTH(ih,deh,entry_num) \ +(I_DEH_N_ENTRY_LENGTH (ih, deh, entry_num) - (de_with_sd (deh) ? SD_SIZE : 0)) + + + +/* hash value occupies bits from 7 up to 30 */ +#define GET_HASH_VALUE(offset) ((offset) & 0x7fffff80LL) +/* generation number occupies 7 bits starting from 0 up to 6 */ +#define GET_GENERATION_NUMBER(offset) ((offset) & 0x7fLL) +#define MAX_GENERATION_NUMBER 127 + +#define SET_GENERATION_NUMBER(offset,gen_number) (GET_HASH_VALUE(offset)|(gen_number)) + + +/* + * Picture represents an internal node of the reiserfs tree + * ______________________________________________________ + * | | Array of | Array of | Free | + * |block | keys | pointers | space | + * | head | N | N+1 | | + * |______|_______________|___________________|___________| + */ + +/***************************************************************************/ +/* DISK CHILD */ +/***************************************************************************/ +/* Disk child pointer: The pointer from an internal node of the tree + to a node that is on disk. */ +struct disk_child { + __u32 dc_block_number; /* Disk child's block number. */ + __u16 dc_size; /* Disk child's used space. */ + __u16 dc_reserved; +}; + +#define DC_SIZE (sizeof(struct disk_child)) + +/* Get disk child by buffer header and position in the tree node. */ +#define B_N_CHILD(p_s_bh,n_pos) ((struct disk_child *)\ +((p_s_bh)->b_data+BLKH_SIZE+B_NR_ITEMS(p_s_bh)*KEY_SIZE+DC_SIZE*(n_pos))) + +/* Get disk child number by buffer header and position in the tree node. */ +#define B_N_CHILD_NUM(p_s_bh,n_pos) (le32_to_cpu (B_N_CHILD(p_s_bh,n_pos)->dc_block_number)) +#define PUT_B_N_CHILD_NUM(p_s_bh,n_pos, val) do { B_N_CHILD(p_s_bh,n_pos)->dc_block_number = cpu_to_le32(val); } while (0) + + /* maximal value of field child_size in structure disk_child */ + /* child size is the combined size of all items and their headers */ +#define MAX_CHILD_SIZE(bh) ((int)( (bh)->b_size - BLKH_SIZE )) + +/* amount of used space in buffer (not including block head) */ +#define B_CHILD_SIZE(cur) (MAX_CHILD_SIZE(cur)-(B_FREE_SPACE(cur))) + +/* max and min number of keys in internal node */ +#define MAX_NR_KEY(bh) ( (MAX_CHILD_SIZE(bh)-DC_SIZE)/(KEY_SIZE+DC_SIZE) ) +#define MIN_NR_KEY(bh) (MAX_NR_KEY(bh)/2) + +/***************************************************************************/ +/* PATH STRUCTURES AND DEFINES */ +/***************************************************************************/ + + +/* Search_by_key fills up the path from the root to the leaf as it descends the tree looking for the + key. It uses reiserfs_bread to try to find buffers in the cache given their block number. If it + does not find them in the cache it reads them from disk. For each node search_by_key finds using + reiserfs_bread it then uses bin_search to look through that node. bin_search will find the + position of the block_number of the next node if it is looking through an internal node. If it + is looking through a leaf node bin_search will find the position of the item which has key either + equal to given key, or which is the maximal key less than the given key. */ + +struct path_element { + struct buffer_head * pe_buffer; /* Pointer to the buffer at the path in the tree. */ + int pe_position; /* Position in the tree node which is placed in the */ + /* buffer above. */ +}; + +#define MAX_HEIGHT 5 /* maximal height of a tree. don't change this without changing JOURNAL_PER_BALANCE_CNT */ +#define EXTENDED_MAX_HEIGHT 7 /* Must be equals MAX_HEIGHT + FIRST_PATH_ELEMENT_OFFSET */ +#define FIRST_PATH_ELEMENT_OFFSET 2 /* Must be equal to at least 2. */ + +#define ILLEGAL_PATH_ELEMENT_OFFSET 1 /* Must be equal to FIRST_PATH_ELEMENT_OFFSET - 1 */ +#define MAX_FEB_SIZE 6 /* this MUST be MAX_HEIGHT + 1. See about FEB below */ + + + +/* We need to keep track of who the ancestors of nodes are. When we + perform a search we record which nodes were visited while + descending the tree looking for the node we searched for. This list + of nodes is called the path. This information is used while + performing balancing. Note that this path information may become + invalid, and this means we must check it when using it to see if it + is still valid. You'll need to read search_by_key and the comments + in it, especially about decrement_counters_in_path(), to understand + this structure. + +Paths make the code so much harder to work with and debug.... An +enormous number of bugs are due to them, and trying to write or modify +code that uses them just makes my head hurt. They are based on an +excessive effort to avoid disturbing the precious VFS code.:-( The +gods only know how we are going to SMP the code that uses them. +znodes are the way! */ + + +struct path { + int path_length; /* Length of the array above. */ + struct path_element path_elements[EXTENDED_MAX_HEIGHT]; /* Array of the path elements. */ + int pos_in_item; +}; + +#define pos_in_item(path) ((path)->pos_in_item) + +#define INITIALIZE_PATH(var) \ +struct path var = {ILLEGAL_PATH_ELEMENT_OFFSET, } + +/* Get path element by path and path position. */ +#define PATH_OFFSET_PELEMENT(p_s_path,n_offset) ((p_s_path)->path_elements +(n_offset)) + +/* Get buffer header at the path by path and path position. */ +#define PATH_OFFSET_PBUFFER(p_s_path,n_offset) (PATH_OFFSET_PELEMENT(p_s_path,n_offset)->pe_buffer) + +/* Get position in the element at the path by path and path position. */ +#define PATH_OFFSET_POSITION(p_s_path,n_offset) (PATH_OFFSET_PELEMENT(p_s_path,n_offset)->pe_position) + + +#define PATH_PLAST_BUFFER(p_s_path) (PATH_OFFSET_PBUFFER((p_s_path), (p_s_path)->path_length)) + /* you know, to the person who didn't + write this the macro name does not + at first suggest what it does. + Maybe POSITION_FROM_PATH_END? Or + maybe we should just focus on + dumping paths... -Hans */ +#define PATH_LAST_POSITION(p_s_path) (PATH_OFFSET_POSITION((p_s_path), (p_s_path)->path_length)) + + +#define PATH_PITEM_HEAD(p_s_path) B_N_PITEM_HEAD(PATH_PLAST_BUFFER(p_s_path),PATH_LAST_POSITION(p_s_path)) + +/* in do_balance leaf has h == 0 in contrast with path structure, + where root has level == 0. That is why we need these defines */ +#define PATH_H_PBUFFER(p_s_path, h) PATH_OFFSET_PBUFFER (p_s_path, p_s_path->path_length - (h)) /* tb->S[h] */ +#define PATH_H_PPARENT(path, h) PATH_H_PBUFFER (path, (h) + 1) /* tb->F[h] or tb->S[0]->b_parent */ +#define PATH_H_POSITION(path, h) PATH_OFFSET_POSITION (path, path->path_length - (h)) +#define PATH_H_B_ITEM_ORDER(path, h) PATH_H_POSITION(path, h + 1) /* tb->S[h]->b_item_order */ + +#define PATH_H_PATH_OFFSET(p_s_path, n_h) ((p_s_path)->path_length - (n_h)) + +#define get_bh(path) PATH_PLAST_BUFFER(path) +#define get_ih(path) PATH_PITEM_HEAD(path) +#define get_item_pos(path) PATH_LAST_POSITION(path) +#define get_item(path) ((void *)B_N_PITEM(PATH_PLAST_BUFFER(path), PATH_LAST_POSITION (path))) +#define item_moved(ih,path) comp_items(ih, path) +#define path_changed(ih,path) comp_items (ih, path) + + +/***************************************************************************/ +/* MISC */ +/***************************************************************************/ + +/* Size of pointer to the unformatted node. */ +#define UNFM_P_SIZE (sizeof(unp_t)) + +// in in-core inode key is stored on le form +#define INODE_PKEY(inode) ((struct key *)((inode)->u.reiserfs_i.i_key)) +//#define mark_tail_converted(inode) (atomic_set(&((inode)->u.reiserfs_i.i_converted),1)) +//#define unmark_tail_converted(inode) (atomic_set(&((inode)->u.reiserfs_i.i_converted), 0)) +//#define is_tail_converted(inode) (atomic_read(&((inode)->u.reiserfs_i.i_converted))) + + + +#define MAX_UL_INT 0xffffffff +#define MAX_INT 0x7ffffff +#define MAX_US_INT 0xffff + +///#define TOO_LONG_LENGTH (~0ULL) + +// reiserfs version 2 has max offset 60 bits. Version 1 - 32 bit offset +#define U32_MAX (~(__u32)0) +extern inline loff_t max_reiserfs_offset (struct inode * inode) +{ + if (inode_items_version (inode) == ITEM_VERSION_1) + return (loff_t)U32_MAX; + + return (loff_t)((~(__u64)0) >> 4); +} + + +/*#define MAX_KEY_UNIQUENESS MAX_UL_INT*/ +#define MAX_KEY_OBJECTID MAX_UL_INT + + +#define MAX_B_NUM MAX_UL_INT +#define MAX_FC_NUM MAX_US_INT + + +/* the purpose is to detect overflow of an unsigned short */ +#define REISERFS_LINK_MAX (MAX_US_INT - 1000) + + +/* The following defines are used in reiserfs_insert_item and reiserfs_append_item */ +#define REISERFS_KERNEL_MEM 0 /* reiserfs kernel memory mode */ +#define REISERFS_USER_MEM 1 /* reiserfs user memory mode */ + +#define fs_generation(s) ((s)->u.reiserfs_sb.s_generation_counter) +#define get_generation(s) atomic_read (&fs_generation(s)) +#define FILESYSTEM_CHANGED_TB(tb) (get_generation((tb)->tb_sb) != (tb)->fs_gen) +#define fs_changed(gen,s) (gen != get_generation (s)) + + +/***************************************************************************/ +/* FIXATE NODES */ +/***************************************************************************/ + +//#define VI_TYPE_STAT_DATA 1 +//#define VI_TYPE_DIRECT 2 +//#define VI_TYPE_INDIRECT 4 +//#define VI_TYPE_DIRECTORY 8 +//#define VI_TYPE_FIRST_DIRECTORY_ITEM 16 +//#define VI_TYPE_INSERTED_DIRECTORY_ITEM 32 + +#define VI_TYPE_LEFT_MERGEABLE 1 +#define VI_TYPE_RIGHT_MERGEABLE 2 + +/* To make any changes in the tree we always first find node, that + contains item to be changed/deleted or place to insert a new + item. We call this node S. To do balancing we need to decide what + we will shift to left/right neighbor, or to a new node, where new + item will be etc. To make this analysis simpler we build virtual + node. Virtual node is an array of items, that will replace items of + node S. (For instance if we are going to delete an item, virtual + node does not contain it). Virtual node keeps information about + item sizes and types, mergeability of first and last items, sizes + of all entries in directory item. We use this array of items when + calculating what we can shift to neighbors and how many nodes we + have to have if we do not any shiftings, if we shift to left/right + neighbor or to both. */ +struct virtual_item +{ + int vi_index; // index in the array of item operations + unsigned short vi_type; // left/right mergeability + unsigned short vi_item_len; /* length of item that it will have after balancing */ + struct item_head * vi_ih; + const char * vi_item; // body of item (old or new) + const void * vi_new_data; // 0 always but paste mode + void * vi_uarea; // item specific area +}; + + +struct virtual_node +{ + char * vn_free_ptr; /* this is a pointer to the free space in the buffer */ + unsigned short vn_nr_item; /* number of items in virtual node */ + short vn_size; /* size of node , that node would have if it has unlimited size and no balancing is performed */ + short vn_mode; /* mode of balancing (paste, insert, delete, cut) */ + short vn_affected_item_num; + short vn_pos_in_item; + struct item_head * vn_ins_ih; /* item header of inserted item, 0 for other modes */ + const void * vn_data; + struct virtual_item * vn_vi; /* array of items (including a new one, excluding item to be deleted) */ +}; + + +/***************************************************************************/ +/* TREE BALANCE */ +/***************************************************************************/ + +/* This temporary structure is used in tree balance algorithms, and + constructed as we go to the extent that its various parts are + needed. It contains arrays of nodes that can potentially be + involved in the balancing of node S, and parameters that define how + each of the nodes must be balanced. Note that in these algorithms + for balancing the worst case is to need to balance the current node + S and the left and right neighbors and all of their parents plus + create a new node. We implement S1 balancing for the leaf nodes + and S0 balancing for the internal nodes (S1 and S0 are defined in + our papers.)*/ + +#define MAX_FREE_BLOCK 7 /* size of the array of buffers to free at end of do_balance */ + +/* maximum number of FEB blocknrs on a single level */ +#define MAX_AMOUNT_NEEDED 2 + +/* someday somebody will prefix every field in this struct with tb_ */ +struct tree_balance +{ + int tb_mode; + int need_balance_dirty; + struct super_block * tb_sb; + struct reiserfs_transaction_handle *transaction_handle ; + struct path * tb_path; + struct buffer_head * L[MAX_HEIGHT]; /* array of left neighbors of nodes in the path */ + struct buffer_head * R[MAX_HEIGHT]; /* array of right neighbors of nodes in the path*/ + struct buffer_head * FL[MAX_HEIGHT]; /* array of fathers of the left neighbors */ + struct buffer_head * FR[MAX_HEIGHT]; /* array of fathers of the right neighbors */ + struct buffer_head * CFL[MAX_HEIGHT]; /* array of common parents of center node and its left neighbor */ + struct buffer_head * CFR[MAX_HEIGHT]; /* array of common parents of center node and its right neighbor */ + + struct buffer_head * FEB[MAX_FEB_SIZE]; /* array of empty buffers. Number of buffers in array equals + cur_blknum. */ + struct buffer_head * used[MAX_FEB_SIZE]; + struct buffer_head * thrown[MAX_FEB_SIZE]; + int lnum[MAX_HEIGHT]; /* array of number of items which must be + shifted to the left in order to balance the + current node; for leaves includes item that + will be partially shifted; for internal + nodes, it is the number of child pointers + rather than items. It includes the new item + being created. The code sometimes subtracts + one to get the number of wholly shifted + items for other purposes. */ + int rnum[MAX_HEIGHT]; /* substitute right for left in comment above */ + int lkey[MAX_HEIGHT]; /* array indexed by height h mapping the key delimiting L[h] and + S[h] to its item number within the node CFL[h] */ + int rkey[MAX_HEIGHT]; /* substitute r for l in comment above */ + int insert_size[MAX_HEIGHT]; /* the number of bytes by we are trying to add or remove from + S[h]. A negative value means removing. */ + int blknum[MAX_HEIGHT]; /* number of nodes that will replace node S[h] after + balancing on the level h of the tree. If 0 then S is + being deleted, if 1 then S is remaining and no new nodes + are being created, if 2 or 3 then 1 or 2 new nodes is + being created */ + + /* fields that are used only for balancing leaves of the tree */ + int cur_blknum; /* number of empty blocks having been already allocated */ + int s0num; /* number of items that fall into left most node when S[0] splits */ + int s1num; /* number of items that fall into first new node when S[0] splits */ + int s2num; /* number of items that fall into second new node when S[0] splits */ + int lbytes; /* number of bytes which can flow to the left neighbor from the left */ + /* most liquid item that cannot be shifted from S[0] entirely */ + /* if -1 then nothing will be partially shifted */ + int rbytes; /* number of bytes which will flow to the right neighbor from the right */ + /* most liquid item that cannot be shifted from S[0] entirely */ + /* if -1 then nothing will be partially shifted */ + int s1bytes; /* number of bytes which flow to the first new node when S[0] splits */ + /* note: if S[0] splits into 3 nodes, then items do not need to be cut */ + int s2bytes; + struct buffer_head * buf_to_free[MAX_FREE_BLOCK]; /* buffers which are to be freed after do_balance finishes by unfix_nodes */ + char * vn_buf; /* kmalloced memory. Used to create + virtual node and keep map of + dirtied bitmap blocks */ + int vn_buf_size; /* size of the vn_buf */ + struct virtual_node * tb_vn; /* VN starts after bitmap of bitmap blocks */ + + int fs_gen; /* saved value of `reiserfs_generation' counter + see FILESYSTEM_CHANGED() macro in reiserfs_fs.h */ +} ; + + +#if 0 + /* when balancing we potentially affect a 3 node wide column of nodes + in the tree (the top of the column may be tapered). C is the nodes + at the center of this column, and L and R are the nodes to the + left and right. */ + struct seal * L_path_seals[MAX_HEIGHT]; + struct seal * C_path_seals[MAX_HEIGHT]; + struct seal * R_path_seals[MAX_HEIGHT]; + char L_path_lock_types[MAX_HEIGHT]; /* 'r', 'w', or 'n' for read, write, or none */ + char C_path_lock_types[MAX_HEIGHT]; + char R_path_lock_types[MAX_HEIGHT]; + + + struct seal_list_elem * C_seal[MAX_HEIGHT]; /* array of seals on nodes in the path */ + struct seal_list_elem * L_seal[MAX_HEIGHT]; /* array of seals on left neighbors of nodes in the path */ + struct seal_list_elem * R_seal[MAX_HEIGHT]; /* array of seals on right neighbors of nodes in the path*/ + struct seal_list_elem * FL_seal[MAX_HEIGHT]; /* array of seals on fathers of the left neighbors */ + struct seal_list_elem * FR_seal[MAX_HEIGHT]; /* array of seals on fathers of the right neighbors */ + struct seal_list_elem * CFL_seal[MAX_HEIGHT]; /* array of seals on common parents of center node and its left neighbor */ + struct seal_list_elem * CFR_seal[MAX_HEIGHT]; /* array of seals on common parents of center node and its right neighbor */ + + struct char C_desired_lock_type[MAX_HEIGHT]; /* 'r', 'w', or 'n' for read, write, or none */ + struct char L_desired_lock_type[MAX_HEIGHT]; + struct char R_desired_lock_type[MAX_HEIGHT]; + struct char FL_desired_lock_type[MAX_HEIGHT]; + struct char FR_desired_lock_type[MAX_HEIGHT]; + struct char CFL_desired_lock_type[MAX_HEIGHT]; + struct char CFR_desired_lock_type[MAX_HEIGHT]; +#endif + + + + + +/* These are modes of balancing */ + +/* When inserting an item. */ +#define M_INSERT 'i' +/* When inserting into (directories only) or appending onto an already + existant item. */ +#define M_PASTE 'p' +/* When deleting an item. */ +#define M_DELETE 'd' +/* When truncating an item or removing an entry from a (directory) item. */ +#define M_CUT 'c' + +/* used when balancing on leaf level skipped (in reiserfsck) */ +#define M_INTERNAL 'n' + +/* When further balancing is not needed, then do_balance does not need + to be called. */ +#define M_SKIP_BALANCING 's' +#define M_CONVERT 'v' + +/* modes of leaf_move_items */ +#define LEAF_FROM_S_TO_L 0 +#define LEAF_FROM_S_TO_R 1 +#define LEAF_FROM_R_TO_L 2 +#define LEAF_FROM_L_TO_R 3 +#define LEAF_FROM_S_TO_SNEW 4 + +#define FIRST_TO_LAST 0 +#define LAST_TO_FIRST 1 + +/* used in do_balance for passing parent of node information that has + been gotten from tb struct */ +struct buffer_info { + struct tree_balance * tb; + struct buffer_head * bi_bh; + struct buffer_head * bi_parent; + int bi_position; +}; + + +/* there are 4 types of items: stat data, directory item, indirect, direct. ++-------------------+------------+--------------+------------+ +| | k_offset | k_uniqueness | mergeable? | ++-------------------+------------+--------------+------------+ +| stat data | 0 | 0 | no | ++-------------------+------------+--------------+------------+ +| 1st directory item| DOT_OFFSET |DIRENTRY_UNIQUENESS| no | +| non 1st directory | hash value | | yes | +| item | | | | ++-------------------+------------+--------------+------------+ +| indirect item | offset + 1 |TYPE_INDIRECT | if this is not the first indirect item of the object ++-------------------+------------+--------------+------------+ +| direct item | offset + 1 |TYPE_DIRECT | if not this is not the first direct item of the object ++-------------------+------------+--------------+------------+ +*/ + +struct item_operations { + int (*bytes_number) (struct item_head * ih, int block_size); + void (*decrement_key) (struct cpu_key *); + int (*is_left_mergeable) (struct key * ih, unsigned long bsize); + void (*print_item) (struct item_head *, char * item); + void (*check_item) (struct item_head *, char * item); + + int (*create_vi) (struct virtual_node * vn, struct virtual_item * vi, + int is_affected, int insert_size); + int (*check_left) (struct virtual_item * vi, int free, + int start_skip, int end_skip); + int (*check_right) (struct virtual_item * vi, int free); + int (*part_size) (struct virtual_item * vi, int from, int to); + int (*unit_num) (struct virtual_item * vi); + void (*print_vi) (struct virtual_item * vi); +}; + + +extern struct item_operations stat_data_ops, indirect_ops, direct_ops, + direntry_ops; +extern struct item_operations * item_ops [4]; + +#define op_bytes_number(ih,bsize) item_ops[le_ih_k_type (ih)]->bytes_number (ih, bsize) +#define op_is_left_mergeable(key,bsize) item_ops[le_key_k_type (le_key_version (key), key)]->is_left_mergeable (key, bsize) +#define op_print_item(ih,item) item_ops[le_ih_k_type (ih)]->print_item (ih, item) +#define op_check_item(ih,item) item_ops[le_ih_k_type (ih)]->check_item (ih, item) +#define op_create_vi(vn,vi,is_affected,insert_size) item_ops[le_ih_k_type ((vi)->vi_ih)]->create_vi (vn,vi,is_affected,insert_size) +#define op_check_left(vi,free,start_skip,end_skip) item_ops[(vi)->vi_index]->check_left (vi, free, start_skip, end_skip) +#define op_check_right(vi,free) item_ops[(vi)->vi_index]->check_right (vi, free) +#define op_part_size(vi,from,to) item_ops[(vi)->vi_index]->part_size (vi, from, to) +#define op_unit_num(vi) item_ops[(vi)->vi_index]->unit_num (vi) +#define op_print_vi(vi) item_ops[(vi)->vi_index]->print_vi (vi) + + + + + +#define COMP_KEYS comp_keys +#define COMP_SHORT_KEYS comp_short_keys +#define keys_of_same_object comp_short_keys + +/*#define COMP_KEYS(p_s_key1, p_s_key2) comp_keys((unsigned long *)(p_s_key1), (unsigned long *)(p_s_key2)) +#define COMP_SHORT_KEYS(p_s_key1, p_s_key2) comp_short_keys((unsigned long *)(p_s_key1), (unsigned long *)(p_s_key2))*/ + + +/* number of blocks pointed to by the indirect item */ +#define I_UNFM_NUM(p_s_ih) ( (p_s_ih)->ih_item_len / UNFM_P_SIZE ) + +/* the used space within the unformatted node corresponding to pos within the item pointed to by ih */ +#define I_POS_UNFM_SIZE(ih,pos,size) (((pos) == I_UNFM_NUM(ih) - 1 ) ? (size) - (ih)->u.ih_free_space : (size)) + +/* number of bytes contained by the direct item or the unformatted nodes the indirect item points to */ + + +/* get the item header */ +#define B_N_PITEM_HEAD(bh,item_num) ( (struct item_head * )((bh)->b_data + BLKH_SIZE) + (item_num) ) + +/* get key */ +#define B_N_PDELIM_KEY(bh,item_num) ( (struct key * )((bh)->b_data + BLKH_SIZE) + (item_num) ) + +/* get the key */ +#define B_N_PKEY(bh,item_num) ( &(B_N_PITEM_HEAD(bh,item_num)->ih_key) ) + +/* get item body */ +#define B_N_PITEM(bh,item_num) ( (bh)->b_data + B_N_PITEM_HEAD((bh),(item_num))->ih_item_location) + +/* get the stat data by the buffer header and the item order */ +#define B_N_STAT_DATA(bh,nr) \ +( (struct stat_data *)((bh)->b_data+B_N_PITEM_HEAD((bh),(nr))->ih_item_location ) ) + + /* following defines use reiserfs buffer header and item header */ + +/* get stat-data */ +#define B_I_STAT_DATA(bh, ih) ( (struct stat_data * )((bh)->b_data + (ih)->ih_item_location) ) + +// this is 3976 for size==4096 +#define MAX_DIRECT_ITEM_LEN(size) ((size) - BLKH_SIZE - 2*IH_SIZE - SD_SIZE - UNFM_P_SIZE) + +/* indirect items consist of entries which contain blocknrs, pos + indicates which entry, and B_I_POS_UNFM_POINTER resolves to the + blocknr contained by the entry pos points to */ +#define B_I_POS_UNFM_POINTER(bh,ih,pos) (*(((unp_t *)B_I_PITEM(bh,ih)) + (pos))) +#define PUT_B_I_POS_UNFM_POINTER(bh,ih,pos, val) do {*(((unp_t *)B_I_PITEM(bh,ih)) + (pos)) = cpu_to_le32(val); } while (0) + +/* Reiserfs buffer cache statistics. */ +#ifdef REISERFS_CACHE_STAT + struct reiserfs_cache_stat + { + int nr_reiserfs_ll_r_block; /* Number of block reads. */ + int nr_reiserfs_ll_w_block; /* Number of block writes. */ + int nr_reiserfs_schedule; /* Number of locked buffers waits. */ + unsigned long nr_reiserfs_bread; /* Number of calls to reiserfs_bread function */ + unsigned long nr_returns; /* Number of breads of buffers that were hoped to contain a key but did not after bread completed + (usually due to object shifting while bread was executing.) + In the code this manifests as the number + of times that the repeat variable is nonzero in search_by_key.*/ + unsigned long nr_fixed; /* number of calls of fix_nodes function */ + unsigned long nr_failed; /* number of calls of fix_nodes in which schedule occurred while the function worked */ + unsigned long nr_find1; /* How many times we access a child buffer using its direct pointer from an internal node.*/ + unsigned long nr_find2; /* Number of times there is neither a direct pointer to + nor any entry in the child list pointing to the buffer. */ + unsigned long nr_find3; /* When parent is locked (meaning that there are no direct pointers) + or parent is leaf and buffer to be found is an unformatted node. */ + } cache_stat; +#endif + +struct reiserfs_iget4_args { + __u32 objectid ; +} ; + +/***************************************************************************/ +/* FUNCTION DECLARATIONS */ +/***************************************************************************/ + +/*#ifdef __KERNEL__*/ + +/* journal.c see journal.c for all the comments here */ + +#define JOURNAL_TRANS_HALF 1018 /* must be correct to keep the desc and commit structs at 4k */ + + +/* first block written in a commit. */ +struct reiserfs_journal_desc { + __u32 j_trans_id ; /* id of commit */ + __u32 j_len ; /* length of commit. len +1 is the commit block */ + __u32 j_mount_id ; /* mount id of this trans*/ + __u32 j_realblock[JOURNAL_TRANS_HALF] ; /* real locations for each block */ + char j_magic[12] ; +} ; + +/* last block written in a commit */ +struct reiserfs_journal_commit { + __u32 j_trans_id ; /* must match j_trans_id from the desc block */ + __u32 j_len ; /* ditto */ + __u32 j_realblock[JOURNAL_TRANS_HALF] ; /* real locations for each block */ + char j_digest[16] ; /* md5 sum of all the blocks involved, including desc and commit. not used, kill it */ +} ; + +/* this header block gets written whenever a transaction is considered fully flushed, and is more recent than the +** last fully flushed transaction. fully flushed means all the log blocks and all the real blocks are on disk, +** and this transaction does not need to be replayed. +*/ +struct reiserfs_journal_header { + __u32 j_last_flush_trans_id ; /* id of last fully flushed transaction */ + __u32 j_first_unflushed_offset ; /* offset in the log of where to start replay after a crash */ + __u32 j_mount_id ; +} ; + +/* these are used to keep flush pages that contain converted direct items. +** if the page is not flushed before the transaction that converted it +** is committed, we risk losing data +** +** note, while a page is in this list, its counter is incremented. +*/ +struct reiserfs_page_list { + struct reiserfs_page_list *next ; + struct reiserfs_page_list *prev ; + struct page *page ; + unsigned long blocknr ; /* block number holding converted data */ + + /* if a transaction writer has the page locked the flush_page_list + ** function doesn't need to (and can't) get the lock while flushing + ** the page. do_not_lock needs to be set by anyone who calls journal_end + ** with a page lock held. They have to look in the inode and see + ** if the inode has the page they have locked in the flush list. + ** + ** this sucks. + */ + int do_not_lock ; +} ; + +extern task_queue reiserfs_commit_thread_tq ; +extern wait_queue_head_t reiserfs_commit_thread_wait ; + +/* biggest tunable defines are right here */ +#define JOURNAL_BLOCK_COUNT 8192 /* number of blocks in the journal */ +#define JOURNAL_MAX_BATCH 900 /* max blocks to batch into one transaction, don't make this any bigger than 900 */ +#define JOURNAL_MAX_COMMIT_AGE 30 +#define JOURNAL_MAX_TRANS_AGE 30 +#define JOURNAL_PER_BALANCE_CNT 12 /* must be >= (5 + 2 * (MAX_HEIGHT-2) + 1) */ + +/* both of these can be as low as 1, or as high as you want. The min is the +** number of 4k bitmap nodes preallocated on mount. New nodes are allocated +** as needed, and released when transactions are committed. On release, if +** the current number of nodes is > max, the node is freed, otherwise, +** it is put on a free list for faster use later. +*/ +#define REISERFS_MIN_BITMAP_NODES 10 +#define REISERFS_MAX_BITMAP_NODES 100 + +#define JBH_HASH_SHIFT 13 /* these are based on journal hash size of 8192 */ +#define JBH_HASH_MASK 8191 + +/* After several hours of tedious analysis, the following hash + * function won. Do not mess with it... -DaveM + */ +#define _jhashfn(dev,block) \ + ((((dev)<<(JBH_HASH_SHIFT - 6)) ^ ((dev)<<(JBH_HASH_SHIFT - 9))) ^ \ + (((block)<<(JBH_HASH_SHIFT - 6)) ^ ((block) >> 13) ^ ((block) << (JBH_HASH_SHIFT - 12)))) +#define journal_hash(t,dev,block) ((t)[_jhashfn((dev),(block)) & JBH_HASH_MASK]) + +/* finds n'th buffer with 0 being the start of this commit. Needs to go away, j_ap_blocks has changed +** since I created this. One chunk of code in journal.c needs changing before deleting it +*/ +#define JOURNAL_BUFFER(j,n) ((j)->j_ap_blocks[((j)->j_start + (n)) % JOURNAL_BLOCK_COUNT]) + +void reiserfs_wait_on_write_block(struct super_block *s) ; +void reiserfs_block_writes(struct reiserfs_transaction_handle *th) ; +void reiserfs_allow_writes(struct super_block *s) ; +void reiserfs_check_lock_depth(char *caller) ; +void reiserfs_prepare_for_journal(struct super_block *, struct buffer_head *bh, int wait) ; +void reiserfs_restore_prepared_buffer(struct super_block *, struct buffer_head *bh) ; +int journal_init(struct super_block *) ; +int journal_release(struct reiserfs_transaction_handle*, struct super_block *) ; +int journal_release_error(struct reiserfs_transaction_handle*, struct super_block *) ; +int journal_end(struct reiserfs_transaction_handle *, struct super_block *, unsigned long) ; +int journal_end_sync(struct reiserfs_transaction_handle *, struct super_block *, unsigned long) ; +int journal_mark_dirty_nolog(struct reiserfs_transaction_handle *, struct super_block *, struct buffer_head *bh) ; +int journal_mark_freed(struct reiserfs_transaction_handle *, struct super_block *, unsigned long blocknr) ; +int push_journal_writer(char *w) ; +int pop_journal_writer(int windex) ; +int journal_lock_dobalance(struct super_block *p_s_sb) ; +int journal_unlock_dobalance(struct super_block *p_s_sb) ; +int journal_transaction_should_end(struct reiserfs_transaction_handle *, int) ; +int reiserfs_in_journal(struct super_block *p_s_sb, kdev_t dev, unsigned long bl, int size, int searchall, unsigned long *next) ; +int journal_begin(struct reiserfs_transaction_handle *, struct super_block *p_s_sb, unsigned long) ; +int journal_join(struct reiserfs_transaction_handle *, struct super_block *p_s_sb, unsigned long) ; +struct super_block *reiserfs_get_super(kdev_t dev) ; +void flush_async_commits(struct super_block *p_s_sb) ; + +int remove_from_transaction(struct super_block *p_s_sb, unsigned long blocknr, int already_cleaned) ; +int remove_from_journal_list(struct super_block *s, struct reiserfs_journal_list *jl, struct buffer_head *bh, int remove_freed) ; + +int buffer_journaled(struct buffer_head *bh) ; +int mark_buffer_journal_new(struct buffer_head *bh) ; +int reiserfs_sync_all_buffers(kdev_t dev, int wait) ; +int reiserfs_sync_buffers(kdev_t dev, int wait) ; +int reiserfs_add_page_to_flush_list(struct reiserfs_transaction_handle *, + struct inode *, struct buffer_head *) ; +int reiserfs_remove_page_from_flush_list(struct reiserfs_transaction_handle *, + struct inode *) ; + +int reiserfs_allocate_list_bitmaps(struct super_block *s, struct reiserfs_list_bitmap *, int) ; + + /* why is this kerplunked right here? */ +static inline int reiserfs_buffer_prepared(struct buffer_head *bh) { + if (bh && test_bit(BH_JPrepared, &bh->b_state)) + return 1 ; + else + return 0 ; +} + +/* buffer was journaled, waiting to get to disk */ +static inline int buffer_journal_dirty(struct buffer_head *bh) { + if (bh) + return test_bit(BH_JDirty_wait, &bh->b_state) ; + else + return 0 ; +} +static inline int mark_buffer_notjournal_dirty(struct buffer_head *bh) { + if (bh) + clear_bit(BH_JDirty_wait, &bh->b_state) ; + return 0 ; +} +static inline int mark_buffer_notjournal_new(struct buffer_head *bh) { + if (bh) { + clear_bit(BH_JNew, &bh->b_state) ; + } + return 0 ; +} + +/* objectid.c */ +__u32 reiserfs_get_unused_objectid (struct reiserfs_transaction_handle *th); +void reiserfs_release_objectid (struct reiserfs_transaction_handle *th, __u32 objectid_to_release); +int reiserfs_convert_objectid_map_v1(struct super_block *) ; + +/* stree.c */ +int B_IS_IN_TREE(struct buffer_head *); +extern inline void copy_key (void * to, void * from); +extern inline void copy_short_key (void * to, void * from); +extern inline void copy_item_head(void * p_v_to, void * p_v_from); + +// first key is in cpu form, second - le +extern inline int comp_keys (struct key * le_key, struct cpu_key * cpu_key); +extern inline int comp_short_keys (struct key * le_key, struct cpu_key * cpu_key); +extern inline void le_key2cpu_key (struct cpu_key * to, struct key * from); + +// both are cpu keys +extern inline int comp_cpu_keys (struct cpu_key *, struct cpu_key *); +extern inline int comp_short_cpu_keys (struct cpu_key *, struct cpu_key *); +extern inline void cpu_key2cpu_key (struct cpu_key *, struct cpu_key *); + +// both are in le form +extern inline int comp_le_keys (struct key *, struct key *); +extern inline int comp_short_le_keys (struct key *, struct key *); + +// +// get key version from on disk key - kludge +// +extern inline int le_key_version (struct key * key) +{ + int type; + + type = le16_to_cpu (key->u.k_offset_v2.k_type); + if (type != TYPE_DIRECT && type != TYPE_INDIRECT && type != TYPE_DIRENTRY) + return ITEM_VERSION_1; + + return ITEM_VERSION_2; + +} + + +extern inline void copy_key (void * to, void * from) +{ + memcpy (to, from, KEY_SIZE); +} + + +int comp_items (struct item_head * p_s_ih, struct path * p_s_path); +struct key * get_rkey (struct path * p_s_chk_path, struct super_block * p_s_sb); +inline int bin_search (void * p_v_key, void * p_v_base, int p_n_num, int p_n_width, int * p_n_pos); +int search_by_key (struct super_block *, struct cpu_key *, struct path *, int); +#define search_item(s,key,path) search_by_key (s, key, path, DISK_LEAF_NODE_LEVEL) +int search_for_position_by_key (struct super_block * p_s_sb, struct cpu_key * p_s_cpu_key, struct path * p_s_search_path); +extern inline void decrement_bcount (struct buffer_head * p_s_bh); +void decrement_counters_in_path (struct path * p_s_search_path); +void pathrelse (struct path * p_s_search_path); +int reiserfs_check_path(struct path *p) ; +void pathrelse_and_restore (struct super_block *s, struct path * p_s_search_path); + +int reiserfs_insert_item (struct reiserfs_transaction_handle *th, + struct path * path, + struct cpu_key * key, + struct item_head * ih, const char * body); + +int reiserfs_paste_into_item (struct reiserfs_transaction_handle *th, + struct path * path, + struct cpu_key * key, + const char * body, int paste_size); + +int reiserfs_cut_from_item (struct reiserfs_transaction_handle *th, + struct path * path, + struct cpu_key * key, + struct inode * inode, + struct page *page, + loff_t new_file_size); + +int reiserfs_delete_item (struct reiserfs_transaction_handle *th, + struct path * path, + struct cpu_key * key, + struct inode * inode, + struct buffer_head * p_s_un_bh); + + +void reiserfs_delete_object (struct reiserfs_transaction_handle *th, struct inode * p_s_inode); +void reiserfs_do_truncate (struct reiserfs_transaction_handle *th, + struct inode * p_s_inode, struct page *, + int update_timestamps); +// +//void lock_inode_to_convert (struct inode * p_s_inode); +//void unlock_inode_after_convert (struct inode * p_s_inode); +//void increment_i_read_sync_counter (struct inode * p_s_inode); +//void decrement_i_read_sync_counter (struct inode * p_s_inode); + + +#define block_size(inode) ((inode)->i_sb->s_blocksize) +#define file_size(inode) ((inode)->i_size) +#define tail_size(inode) (file_size (inode) & (block_size (inode) - 1)) + +#define tail_has_to_be_packed(inode) (!dont_have_tails ((inode)->i_sb) &&\ +!STORE_TAIL_IN_UNFM(file_size (inode), tail_size(inode), block_size (inode))) + +/* +int get_buffer_by_range (struct super_block * p_s_sb, struct key * p_s_range_begin, struct key * p_s_range_end, + struct buffer_head ** pp_s_buf, unsigned long * p_n_objectid); +int get_buffers_from_range (struct super_block * p_s_sb, struct key * p_s_range_start, struct key * p_s_range_end, + struct buffer_head ** p_s_range_buffers, + int n_max_nr_buffers_to_return); +*/ + +#ifndef REISERFS_FSCK + +//inline int is_left_mergeable (struct item_head * ih, unsigned long bsize); + +#else + +int is_left_mergeable (struct super_block * s, struct path * path); +int is_right_mergeable (struct super_block * s, struct path * path); +int are_items_mergeable (struct item_head * left, struct item_head * right, int bsize); + +#endif +void padd_item (char * item, int total_length, int length); + + +/* inode.c */ + +int reiserfs_prepare_write(struct file *, struct page *, unsigned, unsigned) ; +void reiserfs_truncate_file(struct inode *, int update_timestamps) ; +void make_cpu_key (struct cpu_key * cpu_key, const struct inode * inode, loff_t offset, + int type, int key_length); +void make_le_item_head (struct item_head * ih, struct cpu_key * key, int version, + loff_t offset, int type, int length, int entry_count); +/*void store_key (struct key * key); +void forget_key (struct key * key);*/ +int reiserfs_get_block (struct inode * inode, long block, + struct buffer_head * bh_result, int create); +struct inode * reiserfs_iget (struct super_block * s, struct cpu_key * key); +void reiserfs_read_inode (struct inode * inode) ; +void reiserfs_read_inode2(struct inode * inode, void *p) ; +void reiserfs_delete_inode (struct inode * inode); +extern int reiserfs_notify_change(struct dentry * dentry, struct iattr * attr); +void reiserfs_write_inode (struct inode * inode, int) ; + +/* we don't mark inodes dirty, we just log them */ +void reiserfs_dirty_inode (struct inode * inode) ; + +struct inode * reiserfs_new_inode (struct reiserfs_transaction_handle *th, const struct inode * dir, int mode, + const char * symname, int item_len, + struct dentry *dentry, struct inode *inode, int * err); +int reiserfs_sync_inode (struct reiserfs_transaction_handle *th, struct inode * inode); +void reiserfs_update_sd (struct reiserfs_transaction_handle *th, struct inode * inode); +int reiserfs_inode_setattr(struct dentry *, struct iattr * attr); + +/* namei.c */ +inline void set_de_name_and_namelen (struct reiserfs_dir_entry * de); +int search_by_entry_key (struct super_block * sb, struct cpu_key * key, struct path * path, + struct reiserfs_dir_entry * de); +struct dentry * reiserfs_lookup (struct inode * dir, struct dentry *dentry); +int reiserfs_create (struct inode * dir, struct dentry *dentry, int mode); +int reiserfs_mknod (struct inode * dir_inode, struct dentry *dentry, int mode, int rdev); +int reiserfs_mkdir (struct inode * dir, struct dentry *dentry, int mode); +int reiserfs_rmdir (struct inode * dir, struct dentry *dentry); +int reiserfs_unlink (struct inode * dir, struct dentry *dentry); +int reiserfs_symlink (struct inode * dir, struct dentry *dentry, const char * symname); +int reiserfs_link (struct dentry * old_dentry, struct inode * dir, struct dentry *dentry); +int reiserfs_rename (struct inode * old_dir, struct dentry *old_dentry, struct inode * new_dir, struct dentry *new_dentry); + +/* super.c */ +inline void reiserfs_mark_buffer_dirty (struct buffer_head * bh, int flag); +inline void reiserfs_mark_buffer_clean (struct buffer_head * bh); +void reiserfs_panic (struct super_block * s, const char * fmt, ...); +void reiserfs_write_super (struct super_block * s); +void reiserfs_put_super (struct super_block * s); +int reiserfs_remount (struct super_block * s, int * flags, char * data); +/*int read_super_block (struct super_block * s, int size); +int read_bitmaps (struct super_block * s); +int read_old_bitmaps (struct super_block * s); +int read_old_super_block (struct super_block * s, int size);*/ +struct super_block * reiserfs_read_super (struct super_block * s, void * data, int silent); +int reiserfs_statfs (struct super_block * s, struct statfs * buf); + +/* dir.c */ +extern struct inode_operations reiserfs_dir_inode_operations; +extern struct file_operations reiserfs_dir_operations; + +/* tail_conversion.c */ +int direct2indirect (struct reiserfs_transaction_handle *, struct inode *, struct path *, struct buffer_head *, loff_t); +int indirect2direct (struct reiserfs_transaction_handle *, struct inode *, struct page *, struct path *, struct cpu_key *, loff_t, char *); +void reiserfs_unmap_buffer(struct buffer_head *) ; + + +/* file.c */ +extern struct inode_operations reiserfs_file_inode_operations; +extern struct file_operations reiserfs_file_operations; +extern struct address_space_operations reiserfs_address_space_operations ; +int get_new_buffer (struct reiserfs_transaction_handle *th, struct buffer_head *, + struct buffer_head **, struct path *); + + +/* buffer2.c */ +struct buffer_head * reiserfs_getblk (kdev_t n_dev, int n_block, int n_size); +void wait_buffer_until_released (struct buffer_head * bh); +struct buffer_head * reiserfs_bread (kdev_t n_dev, int n_block, int n_size); + + +/* fix_nodes.c */ +void * reiserfs_kmalloc (size_t size, int flags, struct super_block * s); +void reiserfs_kfree (const void * vp, size_t size, struct super_block * s); +int fix_nodes (int n_op_mode, struct tree_balance * p_s_tb, struct item_head * p_s_ins_ih, const void *); +void unfix_nodes (struct tree_balance *); +void free_buffers_in_tb (struct tree_balance * p_s_tb); + + +/* prints.c */ +void reiserfs_panic (struct super_block * s, const char * fmt, ...); +void reiserfs_warning (const char * fmt, ...); +void reiserfs_debug (struct super_block *s, int level, const char * fmt, ...); +void print_virtual_node (struct virtual_node * vn); +void print_indirect_item (struct buffer_head * bh, int item_num); +void store_print_tb (struct tree_balance * tb); +void print_cur_tb (char * mes); +void print_de (struct reiserfs_dir_entry * de); +void print_bi (struct buffer_info * bi, char * mes); +#define PRINT_LEAF_ITEMS 1 /* print all items */ +#define PRINT_DIRECTORY_ITEMS 2 /* print directory items */ +#define PRINT_DIRECT_ITEMS 4 /* print contents of direct items */ +void print_block (struct buffer_head * bh, ...); +void print_path (struct tree_balance * tb, struct path * path); +void print_bmap (struct super_block * s, int silent); +void print_bmap_block (int i, char * data, int size, int silent); +/*void print_super_block (struct super_block * s, char * mes);*/ +void print_objectid_map (struct super_block * s); +void print_block_head (struct buffer_head * bh, char * mes); +void check_leaf (struct buffer_head * bh); +void check_internal (struct buffer_head * bh); +void print_statistics (struct super_block * s); + +/* lbalance.c */ +int leaf_move_items (int shift_mode, struct tree_balance * tb, int mov_num, int mov_bytes, struct buffer_head * Snew); +int leaf_shift_left (struct tree_balance * tb, int shift_num, int shift_bytes); +int leaf_shift_right (struct tree_balance * tb, int shift_num, int shift_bytes); +void leaf_delete_items (struct buffer_info * cur_bi, int last_first, int first, int del_num, int del_bytes); +void leaf_insert_into_buf (struct buffer_info * bi, int before, + struct item_head * inserted_item_ih, const char * inserted_item_body, int zeros_number); +void leaf_paste_in_buffer (struct buffer_info * bi, int pasted_item_num, + int pos_in_item, int paste_size, const char * body, int zeros_number); +void leaf_cut_from_buffer (struct buffer_info * bi, int cut_item_num, int pos_in_item, + int cut_size); +void leaf_paste_entries (struct buffer_head * bh, int item_num, int before, + int new_entry_count, struct reiserfs_de_head * new_dehs, const char * records, int paste_size); +/* ibalance.c */ +int balance_internal (struct tree_balance * , int, int, struct item_head * , + struct buffer_head **); + +/* do_balance.c */ +inline void do_balance_mark_leaf_dirty (struct tree_balance * tb, + struct buffer_head * bh, int flag); +#define do_balance_mark_internal_dirty do_balance_mark_leaf_dirty +#define do_balance_mark_sb_dirty do_balance_mark_leaf_dirty + +void do_balance (struct tree_balance * tb, struct item_head * ih, + const char * body, int flag); +void reiserfs_invalidate_buffer (struct tree_balance * tb, struct buffer_head * bh); + +int get_left_neighbor_position (struct tree_balance * tb, int h); +int get_right_neighbor_position (struct tree_balance * tb, int h); +void replace_key (struct tree_balance * tb, struct buffer_head *, int, struct buffer_head *, int); +void replace_lkey (struct tree_balance *, int, struct item_head *); +void replace_rkey (struct tree_balance *, int, struct item_head *); +void make_empty_node (struct buffer_info *); +struct buffer_head * get_FEB (struct tree_balance *); + +/* bitmap.c */ +int is_reusable (struct super_block * s, unsigned long block, int bit_value); +void reiserfs_free_block (struct reiserfs_transaction_handle *th, unsigned long); +int reiserfs_new_blocknrs (struct reiserfs_transaction_handle *th, + unsigned long * pblocknrs, unsigned long start_from, int amount_needed); +int reiserfs_new_unf_blocknrs (struct reiserfs_transaction_handle *th, + unsigned long * pblocknr, unsigned long start_from); +#ifdef REISERFS_PREALLOCATE +int reiserfs_new_unf_blocknrs2 (struct reiserfs_transaction_handle *th, + struct inode * inode, + unsigned long * pblocknr, + unsigned long start_from); + +void reiserfs_discard_prealloc (struct reiserfs_transaction_handle *th, + struct inode * inode); +#endif + +/* hashes.c */ +__u32 keyed_hash (const char *msg, int len); +__u32 yura_hash (const char *msg, int len); +__u32 r5_hash (const char *msg, int len); + +/* version.c */ +char *reiserfs_get_version_string(void) ; + +/* the ext2 bit routines adjust for big or little endian as +** appropriate for the arch, so in our laziness we use them rather +** than using the bit routines they call more directly. These +** routines must be used when changing on disk bitmaps. */ +#define reiserfs_test_and_set_le_bit ext2_set_bit +#define reiserfs_test_and_clear_le_bit ext2_clear_bit +#define reiserfs_test_le_bit ext2_test_bit +#define reiserfs_find_next_zero_le_bit ext2_find_next_zero_bit + + +// +// this was totally copied from from linux's +// find_first_zero_bit and changed a bit +// + +#ifdef __i386__ + +extern __inline__ int +find_first_nonzero_bit(void * addr, unsigned size) { + int res; + int __d0; + void *__d1; + + + if (!size) { + return (0); + } + __asm__ __volatile__ ( + "cld\n\t" + "xorl %%eax,%%eax\n\t" + "repe; scasl\n\t" + "je 1f\n\t" + "movl -4(%%edi),%%eax\n\t" + "subl $4, %%edi\n\t" + "bsfl %%eax,%%eax\n\t" + "1:\tsubl %%edx,%%edi\n\t" + "shll $3,%%edi\n\t" + "addl %%edi,%%eax" + :"=a" (res), + "=c"(__d0), "=D"(__d1) + :"1" ((size + 31) >> 5), "d" (addr), "2" (addr)); + return (res); +} + +#else /* __i386__ */ + +extern __inline__ int find_next_nonzero_bit(void * addr, unsigned size, unsigned offset) +{ + unsigned int * p = ((unsigned int *) addr) + (offset >> 5); + unsigned int result = offset & ~31UL; + unsigned int tmp; + + if (offset >= size) + return size; + size -= result; + offset &= 31UL; + if (offset) { + tmp = *p++; + /* set to zero first offset bits */ + tmp &= ~(~0UL >> (32-offset)); + if (size < 32) + goto found_first; + if (tmp != 0U) + goto found_middle; + size -= 32; + result += 32; + } + while (size >= 32) { + if ((tmp = *p++) != 0U) + goto found_middle; + result += 32; + size -= 32; + } + if (!size) + return result; + tmp = *p; +found_first: +found_middle: + return result + ffs(tmp); +} + +#define find_first_nonzero_bit(addr,size) find_next_nonzero_bit((addr), (size), 0) + +#endif /* 0 */ + +/* sometimes reiserfs_truncate may require to allocate few new blocks + to perform indirect2direct conversion. People probably used to + think, that truncate should work without problems on a filesystem + without free disk space. They may complain that they can not + truncate due to lack of free disk space. This spare space allows us + to not worry about it. 500 is probably too much, but it should be + absolutely safe */ +#define SPARE_SPACE 500 + +extern inline unsigned long reiserfs_get_journal_block(struct super_block *s) { + return le32_to_cpu(SB_DISK_SUPER_BLOCK(s)->s_journal_block) ; +} +extern inline unsigned long reiserfs_get_journal_orig_size(struct super_block *s) { + return le32_to_cpu(SB_DISK_SUPER_BLOCK(s)->s_orig_journal_size) ; +} + +/* prototypes from ioctl.c */ +int reiserfs_ioctl (struct inode * inode, struct file * filp, + unsigned int cmd, unsigned long arg); +int reiserfs_unpack (struct inode * inode, struct file * filp); + +/* ioctl's command */ +#define REISERFS_IOC_UNPACK _IOW(0xCD,1,long) + +#endif /* _LINUX_REISER_FS_H */ + + diff --git a/include/linux/reiserfs_fs_i.h b/include/linux/reiserfs_fs_i.h new file mode 100644 index 000000000..3a60b8667 --- /dev/null +++ b/include/linux/reiserfs_fs_i.h @@ -0,0 +1,63 @@ +#ifndef _REISER_FS_I +#define _REISER_FS_I + +/* these are used to keep track of the pages that need +** flushing before the current transaction can commit +*/ +struct reiserfs_page_list ; + +struct reiserfs_inode_info { + __u32 i_key [4];/* key is still 4 32 bit integers */ + + /* this comment will be totally + cryptic to readers not familiar + with 3.5/3.6 format conversion, and + it does not consider that that 3.6 + might not be the last version */ + int i_version; // this says whether file is old or new + + int i_pack_on_close ; // file might need tail packing on close + + __u32 i_first_direct_byte; // offset of first byte stored in direct item. + + /* pointer to the page that must be flushed before + ** the current transaction can commit. + ** + ** this pointer is only used when the tail is converted back into + ** a direct item, or the file is deleted + */ + struct reiserfs_page_list *i_converted_page ; + + /* we save the id of the transaction when we did the direct->indirect + ** conversion. That allows us to flush the buffers to disk + ** without having to update this inode to zero out the converted + ** page variable + */ + int i_conversion_trans_id ; + + /* My guess is this contains the first + unused block of a sequence of + blocks plus the length of the + sequence, which I think is always + at least two at the time of the + preallocation. I really prefer + allocate on flush conceptually..... + + You know, it really annoys me when + code is this badly commented that I + have to guess what it does. + Neither I nor anyone else has time + for guessing what your + datastructures mean. -Hans */ + //For preallocation + int i_prealloc_block; + int i_prealloc_count; + + /* I regret that you think the below + is a comment you should make.... -Hans */ + //nopack-attribute + int nopack; +}; + + +#endif diff --git a/include/linux/reiserfs_fs_sb.h b/include/linux/reiserfs_fs_sb.h new file mode 100644 index 000000000..a6921dad0 --- /dev/null +++ b/include/linux/reiserfs_fs_sb.h @@ -0,0 +1,398 @@ +/* Copyright 1996-2000 Hans Reiser, see reiserfs/README for licensing + * and copyright details */ + +#ifndef _LINUX_REISER_FS_SB +#define _LINUX_REISER_FS_SB + +#ifdef __KERNEL__ +#include <linux/tqueue.h> +#endif + +// +// super block's field values +// +/*#define REISERFS_VERSION 0 undistributed bitmap */ +/*#define REISERFS_VERSION 1 distributed bitmap and resizer*/ +#define REISERFS_VERSION_2 2 /* distributed bitmap, resizer, 64-bit, etc*/ +#define UNSET_HASH 0 // read_super will guess about, what hash names + // in directories were sorted with +#define TEA_HASH 1 +#define YURA_HASH 2 +#define R5_HASH 3 +#define DEFAULT_HASH R5_HASH + +/* this is the on disk super block */ + +struct reiserfs_super_block +{ + __u32 s_block_count; + __u32 s_free_blocks; /* free blocks count */ + __u32 s_root_block; /* root block number */ + __u32 s_journal_block; /* journal block number */ + __u32 s_journal_dev; /* journal device number */ + + /* Since journal size is currently a #define in a header file, if + ** someone creates a disk with a 16MB journal and moves it to a + ** system with 32MB journal default, they will overflow their journal + ** when they mount the disk. s_orig_journal_size, plus some checks + ** while mounting (inside journal_init) prevent that from happening + */ + + /* great comment Chris. Thanks. -Hans */ + + __u32 s_orig_journal_size; + __u32 s_journal_trans_max ; /* max number of blocks in a transaction. */ + __u32 s_journal_block_count ; /* total size of the journal. can change over time */ + __u32 s_journal_max_batch ; /* max number of blocks to batch into a trans */ + __u32 s_journal_max_commit_age ; /* in seconds, how old can an async commit be */ + __u32 s_journal_max_trans_age ; /* in seconds, how old can a transaction be */ + __u16 s_blocksize; /* block size */ + __u16 s_oid_maxsize; /* max size of object id array, see get_objectid() commentary */ + __u16 s_oid_cursize; /* current size of object id array */ + __u16 s_state; /* valid or error */ + char s_magic[12]; /* reiserfs magic string indicates that file system is reiserfs */ + __u32 s_hash_function_code; /* indicate, what hash function is being use to sort names in a directory*/ + __u16 s_tree_height; /* height of disk tree */ + __u16 s_bmap_nr; /* amount of bitmap blocks needed to address each block of file system */ + __u16 s_version; /* I'd prefer it if this was a string, + something like "3.6.4", and maybe + 16 bytes long mostly unused. We + don't need to save bytes in the + superblock. -Hans */ + __u16 s_reserved; + char s_unused[128] ; /* zero filled by mkreiserfs */ +} __attribute__ ((__packed__)); + +#define SB_SIZE (sizeof(struct reiserfs_super_block)) + +/* this is the super from 3.5.X, where X >= 10 */ +struct reiserfs_super_block_v1 +{ + __u32 s_block_count; /* blocks count */ + __u32 s_free_blocks; /* free blocks count */ + __u32 s_root_block; /* root block number */ + __u32 s_journal_block; /* journal block number */ + __u32 s_journal_dev; /* journal device number */ + __u32 s_orig_journal_size; /* size of the journal on FS creation. used to make sure they don't overflow it */ + __u32 s_journal_trans_max ; /* max number of blocks in a transaction. */ + __u32 s_journal_block_count ; /* total size of the journal. can change over time */ + __u32 s_journal_max_batch ; /* max number of blocks to batch into a trans */ + __u32 s_journal_max_commit_age ; /* in seconds, how old can an async commit be */ + __u32 s_journal_max_trans_age ; /* in seconds, how old can a transaction be */ + __u16 s_blocksize; /* block size */ + __u16 s_oid_maxsize; /* max size of object id array, see get_objectid() commentary */ + __u16 s_oid_cursize; /* current size of object id array */ + __u16 s_state; /* valid or error */ + char s_magic[16]; /* reiserfs magic string indicates that file system is reiserfs */ + __u16 s_tree_height; /* height of disk tree */ + __u16 s_bmap_nr; /* amount of bitmap blocks needed to address each block of file system */ + __u32 s_reserved; +} __attribute__ ((__packed__)); + +#define SB_SIZE_V1 (sizeof(struct reiserfs_super_block_v1)) + +/* LOGGING -- */ + +/* These all interelate for performance. +** +** If the journal block count is smaller than n transactions, you lose speed. +** I don't know what n is yet, I'm guessing 8-16. +** +** typical transaction size depends on the application, how often fsync is +** called, and how many metadata blocks you dirty in a 30 second period. +** The more small files (<16k) you use, the larger your transactions will +** be. +** +** If your journal fills faster than dirty buffers get flushed to disk, it must flush them before allowing the journal +** to wrap, which slows things down. If you need high speed meta data updates, the journal should be big enough +** to prevent wrapping before dirty meta blocks get to disk. +** +** If the batch max is smaller than the transaction max, you'll waste space at the end of the journal +** because journal_end sets the next transaction to start at 0 if the next transaction has any chance of wrapping. +** +** The large the batch max age, the better the speed, and the more meta data changes you'll lose after a crash. +** +*/ + +/* don't mess with these for a while */ + /* we have a node size define somewhere in reiserfs_fs.h. -Hans */ +#define JOURNAL_BLOCK_SIZE 4096 /* BUG gotta get rid of this */ +#define JOURNAL_MAX_CNODE 1500 /* max cnodes to allocate. */ +#define JOURNAL_TRANS_MAX 1024 /* biggest possible single transaction, don't change for now (8/3/99) */ +#define JOURNAL_HASH_SIZE 8192 +#define JOURNAL_NUM_BITMAPS 5 /* number of copies of the bitmaps to have floating. Must be >= 2 */ +#define JOURNAL_LIST_COUNT 64 + +/* these are bh_state bit flag offset numbers, for use in the buffer head */ + +#define BH_JDirty 16 /* journal data needs to be written before buffer can be marked dirty */ +#define BH_JDirty_wait 18 /* commit is done, buffer marked dirty */ +#define BH_JNew 19 /* buffer allocated during this transaction, no need to write if freed during this trans too */ + +/* ugly. metadata blocks must be prepared before they can be logged. +** prepared means unlocked and cleaned. If the block is prepared, but not +** logged for some reason, any bits cleared while preparing it must be +** set again. +*/ +#define BH_JPrepared 20 /* block has been prepared for the log */ +#define BH_JRestore_dirty 22 /* restore the dirty bit later */ + +/* One of these for every block in every transaction +** Each one is in two hash tables. First, a hash of the current transaction, and after journal_end, a +** hash of all the in memory transactions. +** next and prev are used by the current transaction (journal_hash). +** hnext and hprev are used by journal_list_hash. If a block is in more than one transaction, the journal_list_hash +** links it in multiple times. This allows flush_journal_list to remove just the cnode belonging +** to a given transaction. +*/ +struct reiserfs_journal_cnode { + struct buffer_head *bh ; /* real buffer head */ + kdev_t dev ; /* dev of real buffer head */ + unsigned long blocknr ; /* block number of real buffer head, == 0 when buffer on disk */ + int state ; + struct reiserfs_journal_list *jlist ; /* journal list this cnode lives in */ + struct reiserfs_journal_cnode *next ; /* next in transaction list */ + struct reiserfs_journal_cnode *prev ; /* prev in transaction list */ + struct reiserfs_journal_cnode *hprev ; /* prev in hash list */ + struct reiserfs_journal_cnode *hnext ; /* next in hash list */ +}; + +struct reiserfs_bitmap_node { + int id ; + char *data ; + struct list_head list ; +} ; + +struct reiserfs_list_bitmap { + struct reiserfs_journal_list *journal_list ; + struct reiserfs_bitmap_node **bitmaps ; +} ; + +/* +** transaction handle which is passed around for all journal calls +*/ +struct reiserfs_transaction_handle { + /* ifdef it. -Hans */ + char *t_caller ; /* debugging use */ + int t_blocks_logged ; /* number of blocks this writer has logged */ + int t_blocks_allocated ; /* number of blocks this writer allocated */ + unsigned long t_trans_id ; /* sanity check, equals the current trans id */ + struct super_block *t_super ; /* super for this FS when journal_begin was + called. saves calls to reiserfs_get_super */ + +} ; + +/* +** one of these for each transaction. The most important part here is the j_realblock. +** this list of cnodes is used to hash all the blocks in all the commits, to mark all the +** real buffer heads dirty once all the commits hit the disk, +** and to make sure every real block in a transaction is on disk before allowing the log area +** to be overwritten */ +struct reiserfs_journal_list { + unsigned long j_start ; + unsigned long j_len ; + atomic_t j_nonzerolen ; + atomic_t j_commit_left ; + atomic_t j_flushing ; + atomic_t j_commit_flushing ; + atomic_t j_older_commits_done ; /* all commits older than this on disk*/ + unsigned long j_trans_id ; + time_t j_timestamp ; + struct reiserfs_list_bitmap *j_list_bitmap ; + struct buffer_head *j_commit_bh ; /* commit buffer head */ + struct reiserfs_journal_cnode *j_realblock ; + struct reiserfs_journal_cnode *j_freedlist ; /* list of buffers that were freed during this trans. free each of these on flush */ + wait_queue_head_t j_commit_wait ; /* wait for all the commit blocks to be flushed */ + wait_queue_head_t j_flush_wait ; /* wait for all the real blocks to be flushed */ +} ; + +struct reiserfs_page_list ; /* defined in reiserfs_fs.h */ + +struct reiserfs_journal { + struct buffer_head ** j_ap_blocks ; /* journal blocks on disk */ + struct reiserfs_journal_cnode *j_last ; /* newest journal block */ + struct reiserfs_journal_cnode *j_first ; /* oldest journal block. start here for traverse */ + + int j_state ; + unsigned long j_trans_id ; + unsigned long j_mount_id ; + unsigned long j_start ; /* start of current waiting commit (index into j_ap_blocks) */ + unsigned long j_len ; /* lenght of current waiting commit */ + unsigned long j_len_alloc ; /* number of buffers requested by journal_begin() */ + atomic_t j_wcount ; /* count of writers for current commit */ + unsigned long j_bcount ; /* batch count. allows turning X transactions into 1 */ + unsigned long j_first_unflushed_offset ; /* first unflushed transactions offset */ + unsigned long j_last_flush_trans_id ; /* last fully flushed journal timestamp */ + struct buffer_head *j_header_bh ; + + /* j_flush_pages must be flushed before the current transaction can + ** commit + */ + struct reiserfs_page_list *j_flush_pages ; + time_t j_trans_start_time ; /* time this transaction started */ + wait_queue_head_t j_wait ; /* wait journal_end to finish I/O */ + atomic_t j_wlock ; /* lock for j_wait */ + wait_queue_head_t j_join_wait ; /* wait for current transaction to finish before starting new one */ + atomic_t j_jlock ; /* lock for j_join_wait */ + int j_journal_list_index ; /* journal list number of the current trans */ + int j_list_bitmap_index ; /* number of next list bitmap to use */ + int j_must_wait ; /* no more journal begins allowed. MUST sleep on j_join_wait */ + int j_next_full_flush ; /* next journal_end will flush all journal list */ + int j_next_async_flush ; /* next journal_end will flush all async commits */ + + int j_cnode_used ; /* number of cnodes on the used list */ + int j_cnode_free ; /* number of cnodes on the free list */ + + struct reiserfs_journal_cnode *j_cnode_free_list ; + struct reiserfs_journal_cnode *j_cnode_free_orig ; /* orig pointer returned from vmalloc */ + + int j_free_bitmap_nodes ; + int j_used_bitmap_nodes ; + struct list_head j_bitmap_nodes ; + struct reiserfs_list_bitmap j_list_bitmap[JOURNAL_NUM_BITMAPS] ; /* array of bitmaps to record the deleted blocks */ + struct reiserfs_journal_list j_journal_list[JOURNAL_LIST_COUNT] ; /* array of all the journal lists */ + struct reiserfs_journal_cnode *j_hash_table[JOURNAL_HASH_SIZE] ; /* hash table for real buffer heads in current trans */ + struct reiserfs_journal_cnode *j_list_hash_table[JOURNAL_HASH_SIZE] ; /* hash table for all the real buffer heads in all + the transactions */ +}; + +#define JOURNAL_DESC_MAGIC "ReIsErLB" /* ick. magic string to find desc blocks in the journal */ + + +typedef __u32 (*hashf_t) (const char *, int); + +/* reiserfs union of in-core super block data */ +struct reiserfs_sb_info +{ + struct buffer_head * s_sbh; /* Buffer containing the super block */ + /* both the comment and the choice of + name are unclear for s_rs -Hans */ + struct reiserfs_super_block * s_rs; /* Pointer to the super block in the buffer */ + struct buffer_head ** s_ap_bitmap; /* array of buffers, holding block bitmap */ + struct reiserfs_journal *s_journal ; /* pointer to journal information */ + unsigned short s_mount_state; /* reiserfs state (valid, invalid) */ + + /* Comment? -Hans */ + void (*end_io_handler)(struct buffer_head *, int); + hashf_t s_hash_function; /* pointer to function which is used + to sort names in directory. Set on + mount */ + unsigned long s_mount_opt; /* reiserfs's mount options are set + here (currently - NOTAIL, NOLOG, + REPLAYONLY) */ + + /* Comment? -Hans */ + wait_queue_head_t s_wait; + /* To be obsoleted soon by per buffer seals.. -Hans */ + atomic_t s_generation_counter; // increased by one every time the + // tree gets re-balanced + + /* session statistics */ + int s_kmallocs; + int s_disk_reads; + int s_disk_writes; + int s_fix_nodes; + int s_do_balance; + int s_unneeded_left_neighbor; + int s_good_search_by_key_reada; + int s_bmaps; + int s_bmaps_without_search; + int s_direct2indirect; + int s_indirect2direct; +}; + + +#define NOTAIL 0 /* -o notail: no tails will be created in a session */ +#define REPLAYONLY 3 /* replay journal and return 0. Use by fsck */ +#define REISERFS_NOLOG 4 /* -o nolog: turn journalling off */ +#define REISERFS_CONVERT 5 /* -o conv: causes conversion of old + format super block to the new + format. If not specified - old + partition will be dealt with in a + manner of 3.5.x */ + +/* -o hash={tea, rupasov, r5, detect} is meant for properly mounting +** reiserfs disks from 3.5.19 or earlier. 99% of the time, this option +** is not required. If the normal autodection code can't determine which +** hash to use (because both hases had the same value for a file) +** use this option to force a specific hash. It won't allow you to override +** the existing hash on the FS, so if you have a tea hash disk, and mount +** with -o hash=rupasov, the mount will fail. +*/ +#define FORCE_TEA_HASH 6 /* try to force tea hash on mount */ +#define FORCE_RUPASOV_HASH 7 /* try to force rupasov hash on mount */ +#define FORCE_R5_HASH 8 /* try to force rupasov hash on mount */ +#define FORCE_HASH_DETECT 9 /* try to detect hash function on mount */ + + +/* used for testing experimental features, makes benchmarking new + features with and without more convenient, should never be used by + users in any code shipped to users (ideally) */ + +#define REISERFS_NO_BORDER 11 +#define REISERFS_NO_UNHASHED_RELOCATION 12 +#define REISERFS_HASHED_RELOCATION 13 +#define REISERFS_TEST4 14 + +#define REISERFS_TEST1 11 +#define REISERFS_TEST2 12 +#define REISERFS_TEST3 13 +#define REISERFS_TEST4 14 + +#define reiserfs_r5_hash(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << FORCE_R5_HASH)) +#define reiserfs_rupasov_hash(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << FORCE_RUPASOV_HASH)) +#define reiserfs_tea_hash(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << FORCE_TEA_HASH)) +#define reiserfs_hash_detect(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << FORCE_HASH_DETECT)) +#define reiserfs_no_border(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << REISERFS_NO_BORDER)) +#define reiserfs_no_unhashed_relocation(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << REISERFS_NO_UNHASHED_RELOCATION)) +#define reiserfs_hashed_relocation(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << REISERFS_HASHED_RELOCATION)) +#define reiserfs_test4(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << REISERFS_TEST4)) + +#define dont_have_tails(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << NOTAIL)) +#define replay_only(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << REPLAYONLY)) +#define reiserfs_dont_log(s) ((s)->u.reiserfs_sb.s_mount_opt & (1 << REISERFS_NOLOG)) +#define old_format_only(s) ((SB_VERSION(s) != REISERFS_VERSION_2) && !((s)->u.reiserfs_sb.s_mount_opt & (1 << REISERFS_CONVERT))) + + +void reiserfs_file_buffer (struct buffer_head * bh, int list); +int reiserfs_is_super(struct super_block *s) ; +int journal_mark_dirty(struct reiserfs_transaction_handle *, struct super_block *, struct buffer_head *bh) ; +int flush_old_commits(struct super_block *s, int) ; +int show_reiserfs_locks(void) ; +int reiserfs_resize(struct super_block *, unsigned long) ; + +#define CARRY_ON 0 +#define SCHEDULE_OCCURRED 1 + + +#define SB_BUFFER_WITH_SB(s) ((s)->u.reiserfs_sb.s_sbh) +#define SB_JOURNAL(s) ((s)->u.reiserfs_sb.s_journal) +#define SB_JOURNAL_LIST(s) (SB_JOURNAL(s)->j_journal_list) +#define SB_JOURNAL_LIST_INDEX(s) (SB_JOURNAL(s)->j_journal_list_index) +#define SB_JOURNAL_LEN_FREE(s) (SB_JOURNAL(s)->j_journal_len_free) +#define SB_AP_BITMAP(s) ((s)->u.reiserfs_sb.s_ap_bitmap) + + +// on-disk super block fields converted to cpu form +#define SB_DISK_SUPER_BLOCK(s) ((s)->u.reiserfs_sb.s_rs) +#define SB_BLOCK_COUNT(s) le32_to_cpu ((SB_DISK_SUPER_BLOCK(s)->s_block_count)) +#define SB_FREE_BLOCKS(s) le32_to_cpu ((SB_DISK_SUPER_BLOCK(s)->s_free_blocks)) +#define SB_REISERFS_MAGIC(s) (SB_DISK_SUPER_BLOCK(s)->s_magic) +#define SB_ROOT_BLOCK(s) le32_to_cpu ((SB_DISK_SUPER_BLOCK(s)->s_root_block)) +#define SB_TREE_HEIGHT(s) le16_to_cpu ((SB_DISK_SUPER_BLOCK(s)->s_tree_height)) +#define SB_REISERFS_STATE(s) le16_to_cpu ((SB_DISK_SUPER_BLOCK(s)->s_state)) +#define SB_VERSION(s) le16_to_cpu ((SB_DISK_SUPER_BLOCK(s)->s_version)) +#define SB_BMAP_NR(s) le16_to_cpu ((SB_DISK_SUPER_BLOCK(s)->s_bmap_nr)) + +#define PUT_SB_BLOCK_COUNT(s, val) do { SB_DISK_SUPER_BLOCK(s)->s_block_count = cpu_to_le32(val); } while (0) +#define PUT_SB_FREE_BLOCKS(s, val) do { SB_DISK_SUPER_BLOCK(s)->s_free_blocks = cpu_to_le32(val); } while (0) +#define PUT_SB_ROOT_BLOCK(s, val) do { SB_DISK_SUPER_BLOCK(s)->s_root_block = cpu_to_le32(val); } while (0) +#define PUT_SB_TREE_HEIGHT(s, val) do { SB_DISK_SUPER_BLOCK(s)->s_tree_height = cpu_to_le16(val); } while (0) +#define PUT_SB_REISERFS_STATE(s, val) do { SB_DISK_SUPER_BLOCK(s)->s_state = cpu_to_le16(val); } while (0) +#define PUT_SB_VERSION(s, val) do { SB_DISK_SUPER_BLOCK(s)->s_version = cpu_to_le16(val); } while (0) +#define PUT_SB_BMAP_NR(s, val) do { SB_DISK_SUPER_BLOCK(s)->s_bmap_nr = cpu_to_le16 (val); } while (0) + +#endif /* _LINUX_REISER_FS_SB */ + + + diff --git a/include/linux/rtc.h b/include/linux/rtc.h index fba9111da..df958b66a 100644 --- a/include/linux/rtc.h +++ b/include/linux/rtc.h @@ -8,7 +8,7 @@ * Copyright (C) 1999 Hewlett-Packard Co. * Copyright (C) 1999 Stephane Eranian <eranian@hpl.hp.com> */ -#ifndef _LINUX_RTC_H +#ifndef _LINUX_RTC_H_ #define _LINUX_RTC_H_ /* diff --git a/include/linux/sched.h b/include/linux/sched.h index 25196fc53..a382b6736 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -219,13 +219,14 @@ struct mm_struct { unsigned long rss, total_vm, locked_vm; unsigned long def_flags; unsigned long cpu_vm_mask; - unsigned long swap_cnt; /* number of pages to swap on next pass */ unsigned long swap_address; /* Architecture-specific MM context */ mm_context_t context; }; +extern int mmlist_nr; + #define INIT_MM(name) \ { \ mmap: &init_mmap, \ @@ -542,8 +543,8 @@ extern unsigned long prof_shift; #define CURRENT_TIME (xtime.tv_sec) -extern void FASTCALL(__wake_up(wait_queue_head_t *q, unsigned int mode, unsigned int wq_mode)); -extern void FASTCALL(__wake_up_sync(wait_queue_head_t *q, unsigned int mode, unsigned int wq_mode)); +extern void FASTCALL(__wake_up(wait_queue_head_t *q, unsigned int mode, int nr)); +extern void FASTCALL(__wake_up_sync(wait_queue_head_t *q, unsigned int mode, int nr)); extern void FASTCALL(sleep_on(wait_queue_head_t *q)); extern long FASTCALL(sleep_on_timeout(wait_queue_head_t *q, signed long timeout)); @@ -552,12 +553,17 @@ extern long FASTCALL(interruptible_sleep_on_timeout(wait_queue_head_t *q, signed long timeout)); extern void FASTCALL(wake_up_process(struct task_struct * tsk)); -#define wake_up(x) __wake_up((x),TASK_UNINTERRUPTIBLE | TASK_INTERRUPTIBLE,WQ_FLAG_EXCLUSIVE) -#define wake_up_all(x) __wake_up((x),TASK_UNINTERRUPTIBLE | TASK_INTERRUPTIBLE,0) -#define wake_up_sync(x) __wake_up_sync((x),TASK_UNINTERRUPTIBLE | TASK_INTERRUPTIBLE,WQ_FLAG_EXCLUSIVE) -#define wake_up_interruptible(x) __wake_up((x),TASK_INTERRUPTIBLE,WQ_FLAG_EXCLUSIVE) -#define wake_up_interruptible_all(x) __wake_up((x),TASK_INTERRUPTIBLE,0) -#define wake_up_interruptible_sync(x) __wake_up_sync((x),TASK_INTERRUPTIBLE,WQ_FLAG_EXCLUSIVE) +#define wake_up(x) __wake_up((x),TASK_UNINTERRUPTIBLE | TASK_INTERRUPTIBLE, 1) +#define wake_up_nr(x, nr) __wake_up((x),TASK_UNINTERRUPTIBLE | TASK_INTERRUPTIBLE, nr) +#define wake_up_all(x) __wake_up((x),TASK_UNINTERRUPTIBLE | TASK_INTERRUPTIBLE, 0) +#define wake_up_sync(x) __wake_up_sync((x),TASK_UNINTERRUPTIBLE | TASK_INTERRUPTIBLE, 1) +#define wake_up_sync_nr(x, nr) __wake_up_sync((x),TASK_UNINTERRUPTIBLE | TASK_INTERRUPTIBLE, nr) +#define wake_up_interruptible(x) __wake_up((x),TASK_INTERRUPTIBLE, 1) +#define wake_up_interruptible_nr(x, nr) __wake_up((x),TASK_INTERRUPTIBLE, nr) +#define wake_up_interruptible_all(x) __wake_up((x),TASK_INTERRUPTIBLE, 0) +#define wake_up_interruptible_sync(x) __wake_up_sync((x),TASK_INTERRUPTIBLE, 1) +#define wake_up_interruptible_sync_nr(x) __wake_up_sync((x),TASK_INTERRUPTIBLE, nr) +asmlinkage long sys_wait4(pid_t pid,unsigned int * stat_addr, int options, struct rusage * ru); extern int in_group_p(gid_t); extern int in_egroup_p(gid_t); diff --git a/include/linux/swap.h b/include/linux/swap.h index e001de887..afb1d96b3 100644 --- a/include/linux/swap.h +++ b/include/linux/swap.h @@ -107,7 +107,7 @@ extern wait_queue_head_t kreclaimd_wait; extern int page_launder(int, int); extern int free_shortage(void); extern int inactive_shortage(void); -extern void wakeup_kswapd(int); +extern void wakeup_kswapd(void); extern int try_to_free_pages(unsigned int gfp_mask); /* linux/mm/page_io.c */ diff --git a/include/linux/vt_kern.h b/include/linux/vt_kern.h index 80feae4c7..f24e4de6b 100644 --- a/include/linux/vt_kern.h +++ b/include/linux/vt_kern.h @@ -30,7 +30,7 @@ extern struct vt_struct { wait_queue_head_t paste_wait; } *vt_cons[MAX_NR_CONSOLES]; -void (*kd_mksound)(unsigned int hz, unsigned int ticks); +extern void (*kd_mksound)(unsigned int hz, unsigned int ticks); /* console.c */ diff --git a/include/net/dn.h b/include/net/dn.h index 52c6ba44d..88a0bab3e 100644 --- a/include/net/dn.h +++ b/include/net/dn.h @@ -9,6 +9,8 @@ typedef unsigned short dn_address; #define dn_ntohs(x) le16_to_cpu((unsigned short)(x)) #define dn_htons(x) cpu_to_le16((unsigned short)(x)) +#define DN_SK(sk) (&sk->protinfo.dn) + struct dn_scp /* Session Control Port */ { unsigned char state; @@ -44,9 +46,22 @@ struct dn_scp /* Session Control Port */ #define DN_SEND 2 #define DN_DONTSEND 1 #define DN_NOCHANGE 0 + unsigned short flowrem_dat; + unsigned short flowrem_oth; + unsigned short flowloc_dat; + unsigned short flowloc_oth; + unsigned char services_rem; + unsigned char services_loc; + unsigned char info_rem; + unsigned char info_loc; + + unsigned short segsize_rem; + unsigned short segsize_loc; + + unsigned char nonagle; + unsigned char multi_ireq; unsigned char accept_mode; - unsigned short mss; - unsigned long seg_size; /* Running total of current segment */ + unsigned long seg_total; /* Running total of current segment */ struct optdata_dn conndata_in; struct optdata_dn conndata_out; @@ -80,7 +95,8 @@ struct dn_scp /* Session Control Port */ * multipliers. */ #define NSP_MIN_WINDOW 1 -#define NSP_MAX_WINDOW 512 +#define NSP_MAX_WINDOW (0x07fe) + unsigned long max_window; unsigned long snd_window; #define NSP_INITIAL_SRTT (HZ) unsigned long nsp_srtt; @@ -116,6 +132,7 @@ struct dn_scp /* Session Control Port */ struct timer_list delack_timer; int delack_pending; void (*delack_fxn)(struct sock *sk); + }; /* @@ -128,7 +145,7 @@ struct dn_scp /* Session Control Port */ * segsize : Size of segment * segnum : Number, for data, otherdata and linkservice * xmit_count : Number of times we've transmitted this skb - * stamp : Time stamp of first transmission, used in RTT calculations + * stamp : Time stamp of most recent transmission, used in RTT calculations * iif: Input interface number * * As a general policy, this structure keeps all addresses in network @@ -136,6 +153,7 @@ struct dn_scp /* Session Control Port */ * and src_port are in network order. All else is in host order. * */ +#define DN_SKB_CB(skb) ((struct dn_skb_cb *)(skb)->cb) struct dn_skb_cb { unsigned short dst; unsigned short src; diff --git a/include/net/dn_nsp.h b/include/net/dn_nsp.h index 8b628d2e5..3948c30a3 100644 --- a/include/net/dn_nsp.h +++ b/include/net/dn_nsp.h @@ -24,12 +24,12 @@ extern void dn_nsp_send_disc(struct sock *sk, unsigned char type, unsigned short reason, int gfp); extern void dn_nsp_return_disc(struct sk_buff *skb, unsigned char type, unsigned short reason); -extern void dn_nsp_send_lnk(struct sock *sk, unsigned short flags); +extern void dn_nsp_send_link(struct sock *sk, unsigned char lsflags, char fcval); extern void dn_nsp_send_conninit(struct sock *sk, unsigned char flags); extern void dn_nsp_output(struct sock *sk); extern int dn_nsp_check_xmit_queue(struct sock *sk, struct sk_buff *skb, struct sk_buff_head *q, unsigned short acknum); -extern void dn_nsp_queue_xmit(struct sock *sk, struct sk_buff *skb, int oob); +extern void dn_nsp_queue_xmit(struct sock *sk, struct sk_buff *skb, int gfp, int oob); extern unsigned long dn_nsp_persist(struct sock *sk); extern int dn_nsp_xmit_timeout(struct sock *sk); @@ -120,6 +120,7 @@ struct nsp_conn_init_msg #define NSP_FC_NONE 0x00 /* Flow Control None */ #define NSP_FC_SRC 0x04 /* Seg Req. Count */ #define NSP_FC_SCMC 0x08 /* Sess. Control Mess */ +#define NSP_FC_MASK 0x0c /* FC type mask */ unsigned char info __attribute__((packed)); unsigned short segsize __attribute__((packed)); }; @@ -178,13 +179,13 @@ static __inline__ int before_or_equal(unsigned short seq1, unsigned short seq2) static __inline__ void seq_add(unsigned short *seq, unsigned short off) { - *seq += off; - *seq &= 0x0fff; + (*seq) += off; + (*seq) &= 0x0fff; } static __inline__ int seq_next(unsigned short seq1, unsigned short seq2) { - return (((seq2&0x0fff) - (seq1&0x0fff)) == 1); + return equal(seq1 + 1, seq2); } /* diff --git a/include/net/ipx.h b/include/net/ipx.h index bc023a662..ba788670f 100644 --- a/include/net/ipx.h +++ b/include/net/ipx.h @@ -73,6 +73,14 @@ typedef struct ipx_route { struct ipx_route *ir_next; } ipx_route; +#ifdef __KERNEL__ +struct ipx_cb { + u8 ipx_tctrl; + u32 ipx_dest_net; + u32 ipx_source_net; + int last_hop_index; +}; +#endif #define IPX_MIN_EPHEMERAL_SOCKET 0x4000 #define IPX_MAX_EPHEMERAL_SOCKET 0x7fff diff --git a/include/net/x25.h b/include/net/x25.h index 257618d44..2e1e88426 100644 --- a/include/net/x25.h +++ b/include/net/x25.h @@ -217,6 +217,7 @@ extern void x25_start_t23timer(struct sock *); extern void x25_stop_heartbeat(struct sock *); extern void x25_stop_timer(struct sock *); extern unsigned long x25_display_timer(struct sock *); +extern void x25_check_rbuf(struct sock *); /* sysctl_net_x25.c */ extern void x25_register_sysctl(void); @@ -121,6 +121,7 @@ static void shm_destroy (struct shmid_kernel *shp) { shm_tot -= (shp->shm_segsz + PAGE_SIZE - 1) >> PAGE_SHIFT; shm_rmid (shp->id); + shmem_lock(shp->shm_file, 0); fput (shp->shm_file); kfree (shp); } @@ -467,10 +468,10 @@ asmlinkage long sys_shmctl (int shmid, int cmd, struct shmid_ds *buf) if(err) goto out_unlock; if(cmd==SHM_LOCK) { - shp->shm_file->f_dentry->d_inode->u.shmem_i.locked = 1; + shmem_lock(shp->shm_file, 1); shp->shm_flags |= SHM_LOCKED; } else { - shp->shm_file->f_dentry->d_inode->u.shmem_i.locked = 0; + shmem_lock(shp->shm_file, 0); shp->shm_flags &= ~SHM_LOCKED; } shm_unlock(shmid); diff --git a/kernel/context.c b/kernel/context.c index 864a70131..6bfaaca92 100644 --- a/kernel/context.c +++ b/kernel/context.c @@ -148,7 +148,7 @@ void flush_scheduled_tasks(void) int start_context_thread(void) { - kernel_thread(context_thread, NULL, CLONE_FS | CLONE_FILES | CLONE_SIGHAND); + kernel_thread(context_thread, NULL, CLONE_FS | CLONE_FILES); return 0; } diff --git a/kernel/fork.c b/kernel/fork.c index 99c1f2317..e578a9644 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -134,7 +134,6 @@ static inline int dup_mmap(struct mm_struct * mm) mm->mmap_cache = NULL; mm->map_count = 0; mm->cpu_vm_mask = 0; - mm->swap_cnt = 0; mm->swap_address = 0; pprev = &mm->mmap; for (mpnt = current->mm->mmap ; mpnt ; mpnt = mpnt->vm_next) { @@ -193,6 +192,7 @@ fail_nomem: } spinlock_t mmlist_lock __cacheline_aligned = SPIN_LOCK_UNLOCKED; +int mmlist_nr; #define allocate_mm() (kmem_cache_alloc(mm_cachep, SLAB_KERNEL)) #define free_mm(mm) (kmem_cache_free(mm_cachep, (mm))) @@ -246,6 +246,7 @@ void mmput(struct mm_struct *mm) { if (atomic_dec_and_lock(&mm->mm_users, &mmlist_lock)) { list_del(&mm->mmlist); + mmlist_nr--; spin_unlock(&mmlist_lock); exit_mmap(mm); mmdrop(mm); @@ -326,6 +327,7 @@ static int copy_mm(unsigned long clone_flags, struct task_struct * tsk) */ spin_lock(&mmlist_lock); list_add(&mm->mmlist, &oldmm->mmlist); + mmlist_nr++; spin_unlock(&mmlist_lock); if (retval) @@ -445,7 +447,7 @@ static int copy_files(unsigned long clone_flags, struct task_struct * tsk) if (size > __FD_SETSIZE) { newf->max_fdset = 0; write_lock(&newf->file_lock); - error = expand_fdset(newf, size); + error = expand_fdset(newf, size-1); write_unlock(&newf->file_lock); if (error) goto out_release; @@ -464,7 +466,7 @@ static int copy_files(unsigned long clone_flags, struct task_struct * tsk) read_unlock(&oldf->file_lock); newf->max_fds = 0; write_lock(&newf->file_lock); - error = expand_fd_array(newf, open_files); + error = expand_fd_array(newf, open_files-1); write_unlock(&newf->file_lock); if (error) goto out_release; diff --git a/kernel/ksyms.c b/kernel/ksyms.c index 8afe07cca..f36261500 100644 --- a/kernel/ksyms.c +++ b/kernel/ksyms.c @@ -159,6 +159,7 @@ EXPORT_SYMBOL(d_alloc); EXPORT_SYMBOL(d_lookup); EXPORT_SYMBOL(__d_path); EXPORT_SYMBOL(mark_buffer_dirty); +EXPORT_SYMBOL(set_buffer_async_io); /* for reiserfs_writepage */ EXPORT_SYMBOL(__mark_buffer_dirty); EXPORT_SYMBOL(__mark_inode_dirty); EXPORT_SYMBOL(get_empty_filp); @@ -394,6 +395,7 @@ EXPORT_SYMBOL(unmap_kiobuf); EXPORT_SYMBOL(lock_kiovec); EXPORT_SYMBOL(unlock_kiovec); EXPORT_SYMBOL(brw_kiovec); +EXPORT_SYMBOL(kiobuf_wait_for_io); /* dma handling */ EXPORT_SYMBOL(request_dma); diff --git a/kernel/sched.c b/kernel/sched.c index bc2dcfa70..ec1f463d0 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -690,19 +690,15 @@ scheduling_in_interrupt: } static inline void __wake_up_common (wait_queue_head_t *q, unsigned int mode, - unsigned int wq_mode, const int sync) + int nr_exclusive, const int sync) { struct list_head *tmp, *head; - struct task_struct *p, *best_exclusive; + struct task_struct *p; unsigned long flags; - int best_cpu, irq; if (!q) goto out; - best_cpu = smp_processor_id(); - irq = in_interrupt(); - best_exclusive = NULL; wq_write_lock_irqsave(&q->lock, flags); #if WAITQUEUE_DEBUG @@ -730,47 +726,27 @@ static inline void __wake_up_common (wait_queue_head_t *q, unsigned int mode, #if WAITQUEUE_DEBUG curr->__waker = (long)__builtin_return_address(0); #endif - /* - * If waking up from an interrupt context then - * prefer processes which are affine to this - * CPU. - */ - if (irq && (curr->flags & wq_mode & WQ_FLAG_EXCLUSIVE)) { - if (!best_exclusive) - best_exclusive = p; - if (p->processor == best_cpu) { - best_exclusive = p; - break; - } - } else { - if (sync) - wake_up_process_synchronous(p); - else - wake_up_process(p); - if (curr->flags & wq_mode & WQ_FLAG_EXCLUSIVE) - break; - } + if (sync) + wake_up_process_synchronous(p); + else + wake_up_process(p); + if ((curr->flags & WQ_FLAG_EXCLUSIVE) && !--nr_exclusive) + break; } } - if (best_exclusive) { - if (sync) - wake_up_process_synchronous(best_exclusive); - else - wake_up_process(best_exclusive); - } wq_write_unlock_irqrestore(&q->lock, flags); out: return; } -void __wake_up(wait_queue_head_t *q, unsigned int mode, unsigned int wq_mode) +void __wake_up(wait_queue_head_t *q, unsigned int mode, int nr) { - __wake_up_common(q, mode, wq_mode, 0); + __wake_up_common(q, mode, nr, 0); } -void __wake_up_sync(wait_queue_head_t *q, unsigned int mode, unsigned int wq_mode) +void __wake_up_sync(wait_queue_head_t *q, unsigned int mode, int nr) { - __wake_up_common(q, mode, wq_mode, 1); + __wake_up_common(q, mode, nr, 1); } #define SLEEP_ON_VAR \ diff --git a/mm/filemap.c b/mm/filemap.c index dedd7911e..4c89ad3e9 100644 --- a/mm/filemap.c +++ b/mm/filemap.c @@ -143,7 +143,8 @@ void __set_page_dirty(struct page *page) list_add(&page->list, &mapping->dirty_pages); spin_unlock(&pagecache_lock); - mark_inode_dirty_pages(mapping->host); + if (mapping->host) + mark_inode_dirty_pages(mapping->host); } /** @@ -306,7 +307,7 @@ inside: */ age_page_up(page); if (inactive_shortage() > inactive_target / 2 && free_shortage()) - wakeup_kswapd(0); + wakeup_kswapd(); not_found: return page; } @@ -974,10 +975,6 @@ static void generic_file_readahead(int reada_ok, * accessed sequentially. */ if (ahead) { - if (reada_ok == 2) { - run_task_queue(&tq_disk); - } - filp->f_ralen += ahead; filp->f_rawin += filp->f_ralen; filp->f_raend = raend + ahead + 1; @@ -1835,7 +1832,8 @@ static long madvise_fixup_start(struct vm_area_struct * vma, n->vm_end = end; setup_read_behavior(n, behavior); n->vm_raend = 0; - get_file(n->vm_file); + if (n->vm_file) + get_file(n->vm_file); if (n->vm_ops && n->vm_ops->open) n->vm_ops->open(n); lock_vma_mappings(vma); @@ -1861,7 +1859,8 @@ static long madvise_fixup_end(struct vm_area_struct * vma, n->vm_pgoff += (n->vm_start - vma->vm_start) >> PAGE_SHIFT; setup_read_behavior(n, behavior); n->vm_raend = 0; - get_file(n->vm_file); + if (n->vm_file) + get_file(n->vm_file); if (n->vm_ops && n->vm_ops->open) n->vm_ops->open(n); lock_vma_mappings(vma); @@ -1893,7 +1892,8 @@ static long madvise_fixup_middle(struct vm_area_struct * vma, right->vm_pgoff += (right->vm_start - left->vm_start) >> PAGE_SHIFT; left->vm_raend = 0; right->vm_raend = 0; - atomic_add(2, &vma->vm_file->f_count); + if (vma->vm_file) + atomic_add(2, &vma->vm_file->f_count); if (vma->vm_ops && vma->vm_ops->open) { vma->vm_ops->open(left); diff --git a/mm/memory.c b/mm/memory.c index 6f1f318a3..7fc8de5eb 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -207,7 +207,8 @@ skip_copy_pte_range: address = (address + PMD_SIZE) & PMD_MASK; src_pte = pte_offset(src_pmd, address); dst_pte = pte_offset(dst_pmd, address); - + + spin_lock(&src->page_table_lock); do { pte_t pte = *src_pte; struct page *ptepage; @@ -240,10 +241,11 @@ skip_copy_pte_range: address = (address + PMD_SIZE) & PMD_MASK; cont_copy_pte_range: set_pte(dst_pte, pte); cont_copy_pte_range_noset: address += PAGE_SIZE; if (address >= end) - goto out; + goto out_unlock; src_pte++; dst_pte++; } while ((unsigned long)src_pte & PTE_TABLE_MASK); + spin_unlock(&src->page_table_lock); cont_copy_pmd_range: src_pmd++; dst_pmd++; @@ -252,6 +254,10 @@ cont_copy_pmd_range: src_pmd++; out: return 0; +out_unlock: + spin_unlock(&src->page_table_lock); + return 0; + nomem: return -ENOMEM; } @@ -939,7 +945,6 @@ void vmtruncate(struct inode * inode, loff_t offset) if (inode->i_size < offset) goto do_expand; inode->i_size = offset; - truncate_inode_pages(mapping, offset); spin_lock(&mapping->i_shared_lock); if (!mapping->i_mmap && !mapping->i_mmap_shared) goto out_unlock; @@ -954,8 +959,7 @@ void vmtruncate(struct inode * inode, loff_t offset) out_unlock: spin_unlock(&mapping->i_shared_lock); - /* this should go into ->truncate */ - inode->i_size = offset; + truncate_inode_pages(mapping, offset); if (inode->i_op && inode->i_op->truncate) inode->i_op->truncate(inode); return; @@ -883,6 +883,8 @@ void exit_mmap(struct mm_struct * mm) mm->rss = 0; mm->total_vm = 0; mm->locked_vm = 0; + + flush_cache_mm(mm); while (mpnt) { struct vm_area_struct * next = mpnt->vm_next; unsigned long start = mpnt->vm_start; @@ -895,13 +897,13 @@ void exit_mmap(struct mm_struct * mm) } mm->map_count--; remove_shared_vm_struct(mpnt); - flush_cache_range(mm, start, end); zap_page_range(mm, start, size); if (mpnt->vm_file) fput(mpnt->vm_file); kmem_cache_free(vm_area_cachep, mpnt); mpnt = next; } + flush_tlb_mm(mm); /* This is just debugging */ if (mm->map_count) diff --git a/mm/page_alloc.c b/mm/page_alloc.c index b67aa4913..09ac27284 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -16,6 +16,7 @@ #include <linux/interrupt.h> #include <linux/pagemap.h> #include <linux/bootmem.h> +#include <linux/slab.h> int nr_swap_pages; int nr_active_pages; @@ -303,7 +304,7 @@ struct page * __alloc_pages(zonelist_t *zonelist, unsigned long order) * an inactive page shortage, wake up kswapd. */ if (inactive_shortage() > inactive_target / 2 && free_shortage()) - wakeup_kswapd(0); + wakeup_kswapd(); /* * If we are about to get low on free pages and cleaning * the inactive_dirty pages would fix the situation, @@ -379,7 +380,7 @@ try_again: * - if we don't have __GFP_IO set, kswapd may be * able to free some memory we can't free ourselves */ - wakeup_kswapd(0); + wakeup_kswapd(); if (gfp_mask & __GFP_WAIT) { __set_current_state(TASK_RUNNING); current->policy |= SCHED_YIELD; @@ -404,7 +405,7 @@ try_again: * - we're doing a higher-order allocation * --> move pages to the free list until we succeed * - we're /really/ tight on memory - * --> wait on the kswapd waitqueue until memory is freed + * --> try to free pages ourselves with page_launder */ if (!(current->flags & PF_MEMALLOC)) { /* @@ -443,36 +444,20 @@ try_again: /* * When we arrive here, we are really tight on memory. * - * We wake up kswapd and sleep until kswapd wakes us - * up again. After that we loop back to the start. - * - * We have to do this because something else might eat - * the memory kswapd frees for us and we need to be - * reliable. Note that we don't loop back for higher - * order allocations since it is possible that kswapd - * simply cannot free a large enough contiguous area - * of memory *ever*. + * We try to free pages ourselves by: + * - shrinking the i/d caches. + * - reclaiming unused memory from the slab caches. + * - swapping/syncing pages to disk (done by page_launder) + * - moving clean pages from the inactive dirty list to + * the inactive clean list. (done by page_launder) */ - if ((gfp_mask & (__GFP_WAIT|__GFP_IO)) == (__GFP_WAIT|__GFP_IO)) { - wakeup_kswapd(1); + if (gfp_mask & __GFP_WAIT) { memory_pressure++; - if (!order) - goto try_again; - /* - * If __GFP_IO isn't set, we can't wait on kswapd because - * kswapd just might need some IO locks /we/ are holding ... - * - * SUBTLE: The scheduling point above makes sure that - * kswapd does get the chance to free memory we can't - * free ourselves... - */ - } else if (gfp_mask & __GFP_WAIT) { try_to_free_pages(gfp_mask); - memory_pressure++; + wakeup_bdflush(0); if (!order) goto try_again; } - } /* @@ -554,14 +539,8 @@ void __free_pages(struct page *page, unsigned long order) void free_pages(unsigned long addr, unsigned long order) { - struct page *fpage; - -#ifdef CONFIG_DISCONTIGMEM - if (addr == 0) return; -#endif - fpage = virt_to_page(addr); - if (VALID_PAGE(fpage)) - __free_pages(fpage, order); + if (addr != 0) + __free_pages(virt_to_page(addr), order); } /* diff --git a/mm/shmem.c b/mm/shmem.c index a81a74659..00426ca27 100644 --- a/mm/shmem.c +++ b/mm/shmem.c @@ -117,11 +117,43 @@ shmem_truncate_part (swp_entry_t * dir, unsigned long size, return 0; } +/* + * shmem_recalc_inode - recalculate the size of an inode + * + * @inode: inode to recalc + * + * We have to calculate the free blocks since the mm can drop pages + * behind our back + * + * But we know that normally + * inodes->i_blocks == inode->i_mapping->nrpages + info->swapped + * + * So the mm freed + * inodes->i_blocks - (inode->i_mapping->nrpages + info->swapped) + * + * It has to be called with the spinlock held. + */ + +static void shmem_recalc_inode(struct inode * inode) +{ + unsigned long freed; + + freed = inode->i_blocks - + (inode->i_mapping->nrpages + inode->u.shmem_i.swapped); + if (freed){ + struct shmem_sb_info * info = &inode->i_sb->u.shmem_sb; + inode->i_blocks -= freed; + spin_lock (&info->stat_lock); + info->free_blocks += freed; + spin_unlock (&info->stat_lock); + } +} + static void shmem_truncate (struct inode * inode) { int clear_base; unsigned long start; - unsigned long mmfreed, freed = 0; + unsigned long freed = 0; swp_entry_t **base, **ptr; struct shmem_inode_info * info = &inode->u.shmem_i; @@ -154,26 +186,9 @@ static void shmem_truncate (struct inode * inode) info->i_indirect = 0; out: - - /* - * We have to calculate the free blocks since we do not know - * how many pages the mm discarded - * - * But we know that normally - * inodes->i_blocks == inode->i_mapping->nrpages + info->swapped - * - * So the mm freed - * inodes->i_blocks - (inode->i_mapping->nrpages + info->swapped) - */ - - mmfreed = inode->i_blocks - (inode->i_mapping->nrpages + info->swapped); info->swapped -= freed; - inode->i_blocks -= freed + mmfreed; + shmem_recalc_inode(inode); spin_unlock (&info->lock); - - spin_lock (&inode->i_sb->u.shmem_sb.stat_lock); - inode->i_sb->u.shmem_sb.free_blocks += freed + mmfreed; - spin_unlock (&inode->i_sb->u.shmem_sb.stat_lock); } static void shmem_delete_inode(struct inode * inode) @@ -201,13 +216,15 @@ static int shmem_writepage(struct page * page) swp_entry_t *entry, swap; info = &page->mapping->host->u.shmem_i; - if (info->locked) - return 1; swap = __get_swap_page(2); - if (!swap.val) - return 1; + if (!swap.val) { + set_page_dirty(page); + UnlockPage(page); + return -ENOMEM; + } spin_lock(&info->lock); + shmem_recalc_inode(page->mapping->host); entry = shmem_swp_entry (info, page->index); if (!entry) /* this had been allocted on page allocation */ BUG(); @@ -269,6 +286,9 @@ struct page * shmem_nopage(struct vm_area_struct * vma, unsigned long address, i entry = shmem_swp_entry (info, idx); if (!entry) goto oom; + spin_lock (&info->lock); + shmem_recalc_inode(inode); + spin_unlock (&info->lock); if (entry->val) { unsigned long flags; @@ -310,6 +330,8 @@ struct page * shmem_nopage(struct vm_area_struct * vma, unsigned long address, i } /* We have the page */ SetPageUptodate (page); + if (info->locked) + page_cache_get(page); cached_page: UnlockPage (page); @@ -374,8 +396,7 @@ struct inode *shmem_get_inode(struct super_block *sb, int mode, int dev) inode->i_fop = &shmem_dir_operations; break; case S_IFLNK: - inode->i_op = &page_symlink_inode_operations; - break; + BUG(); } spin_lock (&shmem_ilock); list_add (&inode->u.shmem_i.list, &shmem_inodes); @@ -401,6 +422,32 @@ static int shmem_statfs(struct super_block *sb, struct statfs *buf) return 0; } +void shmem_lock(struct file * file, int lock) +{ + struct inode * inode = file->f_dentry->d_inode; + struct shmem_inode_info * info = &inode->u.shmem_i; + struct page * page; + unsigned long idx, size; + + if (info->locked == lock) + return; + down(&inode->i_sem); + info->locked = lock; + size = (inode->i_size + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; + for (idx = 0; idx < size; idx++) { + page = find_lock_page(inode->i_mapping, idx); + if (!page) + continue; + if (!lock) { + /* release the extra count and our reference */ + page_cache_release(page); + page_cache_release(page); + } + UnlockPage(page); + } + up(&inode->i_sem); +} + /* * Lookup the data. This is trivial - if the dentry didn't already * exist, we know it is negative. @@ -528,19 +575,6 @@ static int shmem_rename(struct inode * old_dir, struct dentry *old_dentry, struc return error; } -static int shmem_symlink(struct inode * dir, struct dentry *dentry, const char * symname) -{ - int error; - - error = shmem_mknod(dir, dentry, S_IFLNK | S_IRWXUGO, 0); - if (!error) { - int l = strlen(symname)+1; - struct inode *inode = dentry->d_inode; - error = block_symlink(inode, symname, l); - } - return error; -} - static int shmem_mmap(struct file * file, struct vm_area_struct * vma) { struct vm_operations_struct * ops; @@ -677,7 +711,6 @@ static struct inode_operations shmem_dir_inode_operations = { lookup: shmem_lookup, link: shmem_link, unlink: shmem_unlink, - symlink: shmem_symlink, mkdir: shmem_mkdir, rmdir: shmem_rmdir, mknod: shmem_mknod, @@ -1702,7 +1702,7 @@ static void enable_all_cpucaches (void) * kmem_cache_reap - Reclaim memory from caches. * @gfp_mask: the type of memory required. * - * Called from try_to_free_page(). + * Called from do_try_to_free_pages() and __alloc_pages() */ void kmem_cache_reap (int gfp_mask) { @@ -30,8 +30,7 @@ * start background swapping if we fall below freepages.high free * pages, and we begin intensive swapping below freepages.low. * - * Actual initialization is done in mm/page_alloc.c or - * arch/sparc(64)/mm/init.c. + * Actual initialization is done in mm/page_alloc.c */ freepages_t freepages = { 0, /* freepages.min */ diff --git a/mm/vmalloc.c b/mm/vmalloc.c index 62ce5f1ff..93edab662 100644 --- a/mm/vmalloc.c +++ b/mm/vmalloc.c @@ -9,6 +9,7 @@ #include <linux/malloc.h> #include <linux/vmalloc.h> #include <linux/spinlock.h> +#include <linux/highmem.h> #include <linux/smp_lock.h> #include <asm/uaccess.h> diff --git a/mm/vmscan.c b/mm/vmscan.c index afa5261c1..f41c53328 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -35,45 +35,21 @@ * using a process that no longer actually exists (it might * have died while we slept). */ -static int try_to_swap_out(struct mm_struct * mm, struct vm_area_struct* vma, unsigned long address, pte_t * page_table, int gfp_mask) +static void try_to_swap_out(struct mm_struct * mm, struct vm_area_struct* vma, unsigned long address, pte_t * page_table, struct page *page) { pte_t pte; swp_entry_t entry; - struct page * page; - int onlist; - - pte = *page_table; - if (!pte_present(pte)) - goto out_failed; - page = pte_page(pte); - if ((!VALID_PAGE(page)) || PageReserved(page)) - goto out_failed; - - if (!mm->swap_cnt) - return 1; - - mm->swap_cnt--; - onlist = PageActive(page); /* Don't look at this pte if it's been accessed recently. */ if (ptep_test_and_clear_young(page_table)) { - age_page_up(page); - goto out_failed; + page->age += PAGE_AGE_ADV; + if (page->age > PAGE_AGE_MAX) + page->age = PAGE_AGE_MAX; + return; } - if (!onlist) - /* The page is still mapped, so it can't be freeable... */ - age_page_down_ageonly(page); - - /* - * If the page is in active use by us, or if the page - * is in active use by others, don't unmap it or - * (worse) start unneeded IO. - */ - if (page->age > 0) - goto out_failed; if (TryLockPage(page)) - goto out_failed; + return; /* From this point on, the odds are that we're going to * nuke this pte, so read and clear the pte. This hook @@ -87,9 +63,6 @@ static int try_to_swap_out(struct mm_struct * mm, struct vm_area_struct* vma, un * Is the page already in the swap cache? If so, then * we can just drop our reference to it without doing * any IO - it's already up-to-date on disk. - * - * Return 0, as we didn't actually free any real - * memory, and we should just continue our scan. */ if (PageSwapCache(page)) { entry.val = page->index; @@ -99,12 +72,12 @@ set_swap_pte: swap_duplicate(entry); set_pte(page_table, swp_entry_to_pte(entry)); drop_pte: - UnlockPage(page); mm->rss--; - deactivate_page(page); + if (!page->age) + deactivate_page(page); + UnlockPage(page); page_cache_release(page); -out_failed: - return 0; + return; } /* @@ -153,34 +126,20 @@ out_failed: out_unlock_restore: set_pte(page_table, pte); UnlockPage(page); - return 0; + return; } -/* - * A new implementation of swap_out(). We do not swap complete processes, - * but only a small number of blocks, before we continue with the next - * process. The number of blocks actually swapped is determined on the - * number of page faults, that this process actually had in the last time, - * so we won't swap heavily used processes all the time ... - * - * Note: the priority argument is a hint on much CPU to waste with the - * swap block search, not a hint, of how much blocks to swap with - * each process. - * - * (C) 1993 Kai Petzke, wpp@marie.physik.tu-berlin.de - */ - -static inline int swap_out_pmd(struct mm_struct * mm, struct vm_area_struct * vma, pmd_t *dir, unsigned long address, unsigned long end, int gfp_mask) +static int swap_out_pmd(struct mm_struct * mm, struct vm_area_struct * vma, pmd_t *dir, unsigned long address, unsigned long end, int count) { pte_t * pte; unsigned long pmd_end; if (pmd_none(*dir)) - return 0; + return count; if (pmd_bad(*dir)) { pmd_ERROR(*dir); pmd_clear(dir); - return 0; + return count; } pte = pte_offset(dir, address); @@ -190,28 +149,33 @@ static inline int swap_out_pmd(struct mm_struct * mm, struct vm_area_struct * vm end = pmd_end; do { - int result; - mm->swap_address = address + PAGE_SIZE; - result = try_to_swap_out(mm, vma, address, pte, gfp_mask); - if (result) - return result; + if (pte_present(*pte)) { + struct page *page = pte_page(*pte); + + if (VALID_PAGE(page) && !PageReserved(page)) { + try_to_swap_out(mm, vma, address, pte, page); + if (!--count) + break; + } + } address += PAGE_SIZE; pte++; } while (address && (address < end)); - return 0; + mm->swap_address = address + PAGE_SIZE; + return count; } -static inline int swap_out_pgd(struct mm_struct * mm, struct vm_area_struct * vma, pgd_t *dir, unsigned long address, unsigned long end, int gfp_mask) +static inline int swap_out_pgd(struct mm_struct * mm, struct vm_area_struct * vma, pgd_t *dir, unsigned long address, unsigned long end, int count) { pmd_t * pmd; unsigned long pgd_end; if (pgd_none(*dir)) - return 0; + return count; if (pgd_bad(*dir)) { pgd_ERROR(*dir); pgd_clear(dir); - return 0; + return count; } pmd = pmd_offset(dir, address); @@ -221,23 +185,23 @@ static inline int swap_out_pgd(struct mm_struct * mm, struct vm_area_struct * vm end = pgd_end; do { - int result = swap_out_pmd(mm, vma, pmd, address, end, gfp_mask); - if (result) - return result; + count = swap_out_pmd(mm, vma, pmd, address, end, count); + if (!count) + break; address = (address + PMD_SIZE) & PMD_MASK; pmd++; } while (address && (address < end)); - return 0; + return count; } -static int swap_out_vma(struct mm_struct * mm, struct vm_area_struct * vma, unsigned long address, int gfp_mask) +static int swap_out_vma(struct mm_struct * mm, struct vm_area_struct * vma, unsigned long address, int count) { pgd_t *pgdir; unsigned long end; /* Don't swap out areas which are locked down */ if (vma->vm_flags & (VM_LOCKED|VM_RESERVED)) - return 0; + return count; pgdir = pgd_offset(mm, address); @@ -245,18 +209,17 @@ static int swap_out_vma(struct mm_struct * mm, struct vm_area_struct * vma, unsi if (address >= end) BUG(); do { - int result = swap_out_pgd(mm, vma, pgdir, address, end, gfp_mask); - if (result) - return result; + count = swap_out_pgd(mm, vma, pgdir, address, end, count); + if (!count) + break; address = (address + PGDIR_SIZE) & PGDIR_MASK; pgdir++; } while (address && (address < end)); - return 0; + return count; } -static int swap_out_mm(struct mm_struct * mm, int gfp_mask) +static int swap_out_mm(struct mm_struct * mm, int count) { - int result = 0; unsigned long address; struct vm_area_struct* vma; @@ -276,8 +239,8 @@ static int swap_out_mm(struct mm_struct * mm, int gfp_mask) address = vma->vm_start; for (;;) { - result = swap_out_vma(mm, vma, address, gfp_mask); - if (result) + count = swap_out_vma(mm, vma, address, count); + if (!count) goto out_unlock; vma = vma->vm_next; if (!vma) @@ -287,94 +250,63 @@ static int swap_out_mm(struct mm_struct * mm, int gfp_mask) } /* Reset to 0 when we reach the end of address space */ mm->swap_address = 0; - mm->swap_cnt = 0; out_unlock: spin_unlock(&mm->page_table_lock); - return result; + return !count; } /* - * Select the task with maximal swap_cnt and try to swap out a page. * N.B. This function returns only 0 or 1. Return values != 1 from * the lower level routines result in continued processing. */ #define SWAP_SHIFT 5 #define SWAP_MIN 8 +static inline int swap_amount(struct mm_struct *mm) +{ + int nr = mm->rss >> SWAP_SHIFT; + return nr < SWAP_MIN ? SWAP_MIN : nr; +} + static int swap_out(unsigned int priority, int gfp_mask) { int counter; - int __ret = 0; - - /* - * We make one or two passes through the task list, indexed by - * assign = {0, 1}: - * Pass 1: select the swappable task with maximal RSS that has - * not yet been swapped out. - * Pass 2: re-assign rss swap_cnt values, then select as above. - * - * With this approach, there's no need to remember the last task - * swapped out. If the swap-out fails, we clear swap_cnt so the - * task won't be selected again until all others have been tried. - * - * Think of swap_cnt as a "shadow rss" - it tells us which process - * we want to page out (always try largest first). - */ - counter = (nr_threads << SWAP_SHIFT) >> priority; - if (counter < 1) - counter = 1; + int retval = 0; + struct mm_struct *mm = current->mm; - for (; counter >= 0; counter--) { + /* Always start by trying to penalize the process that is allocating memory */ + if (mm) + retval = swap_out_mm(mm, swap_amount(mm)); + + /* Then, look at the other mm's */ + counter = mmlist_nr >> priority; + do { struct list_head *p; - unsigned long max_cnt = 0; - struct mm_struct *best = NULL; - int assign = 0; - int found_task = 0; - select: + spin_lock(&mmlist_lock); p = init_mm.mmlist.next; - for (; p != &init_mm.mmlist; p = p->next) { - struct mm_struct *mm = list_entry(p, struct mm_struct, mmlist); - if (mm->rss <= 0) - continue; - found_task++; - /* Refresh swap_cnt? */ - if (assign == 1) { - mm->swap_cnt = (mm->rss >> SWAP_SHIFT); - if (mm->swap_cnt < SWAP_MIN) - mm->swap_cnt = SWAP_MIN; - } - if (mm->swap_cnt > max_cnt) { - max_cnt = mm->swap_cnt; - best = mm; - } - } + if (p == &init_mm.mmlist) + goto empty; + + /* Move it to the back of the queue.. */ + list_del(p); + list_add_tail(p, &init_mm.mmlist); + mm = list_entry(p, struct mm_struct, mmlist); - /* Make sure it doesn't disappear */ - if (best) - atomic_inc(&best->mm_users); + /* Make sure the mm doesn't disappear when we drop the lock.. */ + atomic_inc(&mm->mm_users); spin_unlock(&mmlist_lock); - /* - * We have dropped the tasklist_lock, but we - * know that "mm" still exists: we are running - * with the big kernel lock, and exit_mm() - * cannot race with us. - */ - if (!best) { - if (!assign && found_task > 0) { - assign = 1; - goto select; - } - break; - } else { - __ret = swap_out_mm(best, gfp_mask); - mmput(best); - break; - } - } - return __ret; + /* Walk about 6% of the address space each time */ + retval |= swap_out_mm(mm, swap_amount(mm)); + mmput(mm); + } while (--counter >= 0); + return retval; + +empty: + spin_unlock(&mmlist_lock); + return 0; } @@ -540,7 +472,6 @@ dirty_page_rescan: */ if (PageDirty(page)) { int (*writepage)(struct page *) = page->mapping->a_ops->writepage; - int result; if (!writepage) goto page_active; @@ -558,16 +489,12 @@ dirty_page_rescan: page_cache_get(page); spin_unlock(&pagemap_lru_lock); - result = writepage(page); + writepage(page); page_cache_release(page); /* And re-start the thing.. */ spin_lock(&pagemap_lru_lock); - if (result != 1) - continue; - /* writepage refused to do anything */ - set_page_dirty(page); - goto page_active; + continue; } /* @@ -808,6 +735,9 @@ int free_shortage(void) int inactive_shortage(void) { int shortage = 0; + pg_data_t *pgdat = pgdat_list; + + /* Is the inactive dirty list too small? */ shortage += freepages.high; shortage += inactive_target; @@ -818,7 +748,27 @@ int inactive_shortage(void) if (shortage > 0) return shortage; - return 0; + /* If not, do we have enough per-zone pages on the inactive list? */ + + shortage = 0; + + do { + int i; + for(i = 0; i < MAX_NR_ZONES; i++) { + int zone_shortage; + zone_t *zone = pgdat->node_zones+ i; + + zone_shortage = zone->pages_high; + zone_shortage -= zone->inactive_dirty_pages; + zone_shortage -= zone->inactive_clean_pages; + zone_shortage -= zone->free_pages; + if (zone_shortage > 0) + shortage += zone_shortage; + } + pgdat = pgdat->node_next; + } while (pgdat); + + return shortage; } /* @@ -833,72 +783,35 @@ int inactive_shortage(void) * really care about latency. In that case we don't try * to free too many pages. */ +#define DEF_PRIORITY (6) static int refill_inactive(unsigned int gfp_mask, int user) { - int priority, count, start_count, made_progress; + int count, start_count, maxtry; count = inactive_shortage() + free_shortage(); if (user) count = (1 << page_cluster); start_count = count; - /* Always trim SLAB caches when memory gets low. */ - kmem_cache_reap(gfp_mask); - - priority = 6; + maxtry = 6; do { - made_progress = 0; - if (current->need_resched) { __set_current_state(TASK_RUNNING); schedule(); } - while (refill_inactive_scan(priority, 1)) { - made_progress = 1; - if (--count <= 0) - goto done; - } - - /* - * don't be too light against the d/i cache since - * refill_inactive() almost never fail when there's - * really plenty of memory free. - */ - shrink_dcache_memory(priority, gfp_mask); - shrink_icache_memory(priority, gfp_mask); - - /* - * Then, try to page stuff out.. - */ - while (swap_out(priority, gfp_mask)) { - made_progress = 1; + while (refill_inactive_scan(DEF_PRIORITY, 1)) { if (--count <= 0) goto done; } - /* - * If we either have enough free memory, or if - * page_launder() will be able to make enough - * free memory, then stop. - */ - if (!inactive_shortage() || !free_shortage()) - goto done; + /* If refill_inactive_scan failed, try to page stuff out.. */ + swap_out(DEF_PRIORITY, gfp_mask); - /* - * Only switch to a lower "priority" if we - * didn't make any useful progress in the - * last loop. - */ - if (!made_progress) - priority--; - } while (priority >= 0); - - /* Always end on a refill_inactive.., may sleep... */ - while (refill_inactive_scan(0, 1)) { - if (--count <= 0) - goto done; - } + if (--maxtry <= 0) + return 0; + + } while (inactive_shortage()); done: return (count < start_count); @@ -922,20 +835,29 @@ static int do_try_to_free_pages(unsigned int gfp_mask, int user) /* * If needed, we move pages from the active list - * to the inactive list. We also "eat" pages from - * the inode and dentry cache whenever we do this. + * to the inactive list. */ - if (free_shortage() || inactive_shortage()) { - shrink_dcache_memory(6, gfp_mask); - shrink_icache_memory(6, gfp_mask); + if (inactive_shortage()) ret += refill_inactive(gfp_mask, user); + + /* + * Delete pages from the inode and dentry caches and + * reclaim unused slab cache if memory is low. + */ + if (free_shortage()) { + shrink_dcache_memory(DEF_PRIORITY, gfp_mask); + shrink_icache_memory(DEF_PRIORITY, gfp_mask); } else { /* - * Reclaim unused slab cache memory. + * Illogical, but true. At least for now. + * + * If we're _not_ under shortage any more, we + * reap the caches. Why? Because a noticeable + * part of the caches are the buffer-heads, + * which we'll want to keep if under shortage. */ kmem_cache_reap(gfp_mask); - ret = 1; - } + } return ret; } @@ -988,13 +910,8 @@ int kswapd(void *unused) static int recalc = 0; /* If needed, try to free some memory. */ - if (inactive_shortage() || free_shortage()) { - int wait = 0; - /* Do we need to do some synchronous flushing? */ - if (waitqueue_active(&kswapd_done)) - wait = 1; - do_try_to_free_pages(GFP_KSWAPD, wait); - } + if (inactive_shortage() || free_shortage()) + do_try_to_free_pages(GFP_KSWAPD, 0); /* * Do some (very minimal) background scanning. This @@ -1002,7 +919,7 @@ int kswapd(void *unused) * every minute. This clears old referenced bits * and moves unused pages to the inactive list. */ - refill_inactive_scan(6, 0); + refill_inactive_scan(DEF_PRIORITY, 0); /* Once a second, recalculate some VM stats. */ if (time_after(jiffies, recalc + HZ)) { @@ -1010,11 +927,6 @@ int kswapd(void *unused) recalculate_vm_stats(); } - /* - * Wake up everybody waiting for free memory - * and unplug the disk queue. - */ - wake_up_all(&kswapd_done); run_task_queue(&tq_disk); /* @@ -1045,33 +957,10 @@ int kswapd(void *unused) } } -void wakeup_kswapd(int block) +void wakeup_kswapd(void) { - DECLARE_WAITQUEUE(wait, current); - - if (current == kswapd_task) - return; - - if (!block) { - if (waitqueue_active(&kswapd_wait)) - wake_up(&kswapd_wait); - return; - } - - /* - * Kswapd could wake us up before we get a chance - * to sleep, so we have to be very careful here to - * prevent SMP races... - */ - __set_current_state(TASK_UNINTERRUPTIBLE); - add_wait_queue(&kswapd_done, &wait); - - if (waitqueue_active(&kswapd_wait)) - wake_up(&kswapd_wait); - schedule(); - - remove_wait_queue(&kswapd_done, &wait); - __set_current_state(TASK_RUNNING); + if (current != kswapd_task) + wake_up_process(kswapd_task); } /* @@ -1096,7 +985,7 @@ DECLARE_WAIT_QUEUE_HEAD(kreclaimd_wait); /* * Kreclaimd will move pages from the inactive_clean list to the * free list, in order to keep atomic allocations possible under - * all circumstances. Even when kswapd is blocked on IO. + * all circumstances. */ int kreclaimd(void *unused) { diff --git a/net/appletalk/aarp.c b/net/appletalk/aarp.c index c67e3fdd0..ce5fc6d0d 100644 --- a/net/appletalk/aarp.c +++ b/net/appletalk/aarp.c @@ -25,6 +25,7 @@ * Jaume Grau - flush caches on AARP_PROBE * Rob Newberry - Added proxy AARP and AARP proc fs, * moved probing from DDP module. + * Arnaldo C. Melo - don't mangle rx packets * */ @@ -58,16 +59,12 @@ #include <linux/init.h> #include <linux/proc_fs.h> - int sysctl_aarp_expiry_time = AARP_EXPIRY_TIME; int sysctl_aarp_tick_time = AARP_TICK_TIME; int sysctl_aarp_retransmit_limit = AARP_RETRANSMIT_LIMIT; int sysctl_aarp_resolve_time = AARP_RESOLVE_TIME; -/* - * Lists of aarp entries - */ - +/* Lists of aarp entries */ struct aarp_entry { /* These first two are only used for unresolved entries */ unsigned long last_sent; /* Last time we xmitted the aarp request */ @@ -81,22 +78,16 @@ struct aarp_entry { struct aarp_entry *next; /* Next entry in chain */ }; -/* - * Hashed list of resolved, unresolved and proxy entries - */ - +/* Hashed list of resolved, unresolved and proxy entries */ static struct aarp_entry *resolved[AARP_HASH_SIZE]; static struct aarp_entry *unresolved[AARP_HASH_SIZE]; static struct aarp_entry *proxies[AARP_HASH_SIZE]; -static int unresolved_count = 0; +static int unresolved_count; /* One lock protects it all. */ static spinlock_t aarp_lock = SPIN_LOCK_UNLOCKED; -/* - * Used to walk the list and purge/kick entries. - */ - +/* Used to walk the list and purge/kick entries. */ static struct timer_list aarp_timer; /* @@ -108,7 +99,7 @@ static void __aarp_expire(struct aarp_entry *a) { struct sk_buff *skb; - while ((skb=skb_dequeue(&a->packet_queue)) != NULL) + while ((skb = skb_dequeue(&a->packet_queue)) != NULL) kfree_skb(skb); kfree(a); @@ -125,33 +116,29 @@ static void __aarp_send_query(struct aarp_entry *a) static char aarp_eth_multicast[ETH_ALEN] = { 0x09, 0x00, 0x07, 0xFF, 0xFF, 0xFF }; struct net_device *dev = a->dev; - int len = dev->hard_header_len + sizeof(struct elapaarp) + aarp_dl->header_length; + int len = dev->hard_header_len + sizeof(struct elapaarp) + + aarp_dl->header_length; struct sk_buff *skb = alloc_skb(len, GFP_ATOMIC); - struct elapaarp *eah; struct at_addr *sat = atalk_find_dev_addr(dev); + struct elapaarp *eah; - if (skb == NULL) + if (!skb) return; - if (sat == NULL) { + if (!sat) { kfree_skb(skb); return; } - /* - * Set up the buffer. - */ - + /* Set up the buffer */ skb_reserve(skb, dev->hard_header_len + aarp_dl->header_length); - eah = (struct elapaarp *)skb_put(skb, sizeof(struct elapaarp)); + eah = (struct elapaarp *)skb_put(skb, + sizeof(struct elapaarp)); skb->protocol = htons(ETH_P_ATALK); skb->nh.raw = skb->h.raw = (void *) eah; skb->dev = dev; - /* - * Set up the ARP. - */ - + /* Set up the ARP */ eah->hw_type = htons(AARP_HW_TYPE_ETHERNET); eah->pa_type = htons(ETH_P_ATALK); eah->hw_len = ETH_ALEN; @@ -170,52 +157,37 @@ static void __aarp_send_query(struct aarp_entry *a) eah->pa_dst_net = a->target_addr.s_net; eah->pa_dst_node= a->target_addr.s_node; - /* - * Add ELAP headers and set target to the AARP multicast. - */ - + /* Add ELAP headers and set target to the AARP multicast */ aarp_dl->datalink_header(aarp_dl, skb, aarp_eth_multicast); - /* - * Send it. - */ - + /* Send it */ dev_queue_xmit(skb); - - /* - * Update the sending count - */ - + /* Update the sending count */ a->xmit_count++; } -/* This runs under aarp_lock and in softint context, so only - * atomic memory allocations can be used. - */ +/* This runs under aarp_lock and in softint context, so only atomic memory + * allocations can be used. */ static void aarp_send_reply(struct net_device *dev, struct at_addr *us, struct at_addr *them, unsigned char *sha) { - int len = dev->hard_header_len + sizeof(struct elapaarp) + aarp_dl->header_length; + int len = dev->hard_header_len + sizeof(struct elapaarp) + + aarp_dl->header_length; struct sk_buff *skb = alloc_skb(len, GFP_ATOMIC); struct elapaarp *eah; - if (skb == NULL) + if (!skb) return; - /* - * Set up the buffer. - */ - + /* Set up the buffer */ skb_reserve(skb, dev->hard_header_len + aarp_dl->header_length); - eah = (struct elapaarp *)skb_put(skb, sizeof(struct elapaarp)); + eah = (struct elapaarp *)skb_put(skb, + sizeof(struct elapaarp)); skb->protocol = htons(ETH_P_ATALK); skb->nh.raw = skb->h.raw = (void *) eah; skb->dev = dev; - /* - * Set up the ARP. - */ - + /* Set up the ARP */ eah->hw_type = htons(AARP_HW_TYPE_ETHERNET); eah->pa_type = htons(ETH_P_ATALK); eah->hw_len = ETH_ALEN; @@ -228,7 +200,7 @@ static void aarp_send_reply(struct net_device *dev, struct at_addr *us, eah->pa_src_net = us->s_net; eah->pa_src_node= us->s_node; - if (sha == NULL) + if (!sha) memset(eah->hw_dst, '\0', ETH_ALEN); else memcpy(eah->hw_dst, sha, ETH_ALEN); @@ -237,47 +209,38 @@ static void aarp_send_reply(struct net_device *dev, struct at_addr *us, eah->pa_dst_net = them->s_net; eah->pa_dst_node= them->s_node; - /* - * Add ELAP headers and set target to the AARP multicast. - */ - + /* Add ELAP headers and set target to the AARP multicast */ aarp_dl->datalink_header(aarp_dl, skb, sha); - - /* - * Send it. - */ + /* Send it */ dev_queue_xmit(skb); } /* - * Send probe frames. Called from aarp_probe_network and aarp_proxy_probe_network. + * Send probe frames. Called from aarp_probe_network and + * aarp_proxy_probe_network. */ void aarp_send_probe(struct net_device *dev, struct at_addr *us) { - int len = dev->hard_header_len + sizeof(struct elapaarp) + aarp_dl->header_length; + int len = dev->hard_header_len + sizeof(struct elapaarp) + + aarp_dl->header_length; struct sk_buff *skb = alloc_skb(len, GFP_ATOMIC); - struct elapaarp *eah; static char aarp_eth_multicast[ETH_ALEN] = { 0x09, 0x00, 0x07, 0xFF, 0xFF, 0xFF }; + struct elapaarp *eah; - if (skb == NULL) + if (!skb) return; - /* - * Set up the buffer. - */ - + /* Set up the buffer */ skb_reserve(skb, dev->hard_header_len + aarp_dl->header_length); - eah = (struct elapaarp *)skb_put(skb, sizeof(struct elapaarp)); + eah = (struct elapaarp *)skb_put(skb, + sizeof(struct elapaarp)); skb->protocol = htons(ETH_P_ATALK); skb->nh.raw = skb->h.raw = (void *) eah; skb->dev = dev; - /* - * Set up the ARP. - */ - + /* Set up the ARP */ eah->hw_type = htons(AARP_HW_TYPE_ETHERNET); eah->pa_type = htons(ETH_P_ATALK); eah->hw_len = ETH_ALEN; @@ -296,15 +259,9 @@ void aarp_send_probe(struct net_device *dev, struct at_addr *us) eah->pa_dst_net = us->s_net; eah->pa_dst_node= us->s_node; - /* - * Add ELAP headers and set target to the AARP multicast. - */ - + /* Add ELAP headers and set target to the AARP multicast */ aarp_dl->datalink_header(aarp_dl, skb, aarp_eth_multicast); - - /* - * Send it. - */ + /* Send it */ dev_queue_xmit(skb); } @@ -318,16 +275,14 @@ static void __aarp_expire_timer(struct aarp_entry **n) { struct aarp_entry *t; - while ((*n) != NULL) { + while (*n) /* Expired ? */ - if(time_after(jiffies, (*n)->expires_at)) { + if (time_after(jiffies, (*n)->expires_at)) { t = *n; *n = (*n)->next; __aarp_expire(t); - } else { + } else n = &((*n)->next); - } - } } /* @@ -340,10 +295,8 @@ static void __aarp_kick(struct aarp_entry **n) { struct aarp_entry *t; - while ((*n) != NULL) { - /* Expired - if this will be the 11th transmit, we delete - * instead. - */ + while (*n) + /* Expired: if this will be the 11th tx, we delete instead. */ if ((*n)->xmit_count >= sysctl_aarp_retransmit_limit) { t = *n; *n = (*n)->next; @@ -352,7 +305,6 @@ static void __aarp_kick(struct aarp_entry **n) __aarp_send_query(*n); n = &((*n)->next); } - } } /* @@ -366,21 +318,16 @@ static void __aarp_expire_device(struct aarp_entry **n, struct net_device *dev) { struct aarp_entry *t; - while ((*n) != NULL) { + while (*n) if ((*n)->dev == dev) { t = *n; *n = (*n)->next; __aarp_expire(t); - } else { + } else n = &((*n)->next); - } - } } -/* - * Handle the timer event - */ - +/* Handle the timer event */ static void aarp_expire_timeout(unsigned long unused) { int ct; @@ -395,17 +342,14 @@ static void aarp_expire_timeout(unsigned long unused) } spin_unlock_bh(&aarp_lock); - mod_timer(&aarp_timer, jiffies + - (unresolved_count ? sysctl_aarp_tick_time: + (unresolved_count ? sysctl_aarp_tick_time : sysctl_aarp_expiry_time)); } -/* - * Network device notifier chain handler. - */ - -static int aarp_device_event(struct notifier_block *this, unsigned long event, void *ptr) +/* Network device notifier chain handler. */ +static int aarp_device_event(struct notifier_block *this, unsigned long event, + void *ptr) { int ct; @@ -432,11 +376,8 @@ static struct aarp_entry *aarp_alloc(void) { struct aarp_entry *a = kmalloc(sizeof(struct aarp_entry), GFP_ATOMIC); - if (a == NULL) - return NULL; - - skb_queue_head_init(&a->packet_queue); - + if (a) + skb_queue_head_init(&a->packet_queue); return a; } @@ -464,10 +405,8 @@ static struct aarp_entry *__aarp_find_entry(struct aarp_entry *list, /* Called from the DDP code, and thus must be exported. */ void aarp_proxy_remove(struct net_device *dev, struct at_addr *sa) { + int hash = sa->s_node % (AARP_HASH_SIZE - 1); struct aarp_entry *a; - int hash; - - hash = sa->s_node % (AARP_HASH_SIZE-1); spin_lock_bh(&aarp_lock); @@ -479,23 +418,15 @@ void aarp_proxy_remove(struct net_device *dev, struct at_addr *sa) } /* This must run under aarp_lock. */ -static struct at_addr *__aarp_proxy_find(struct net_device *dev, struct at_addr *sa) +static struct at_addr *__aarp_proxy_find(struct net_device *dev, + struct at_addr *sa) { - struct at_addr *retval; - struct aarp_entry *a; - int hash; + int hash = sa->s_node % (AARP_HASH_SIZE - 1); + struct aarp_entry *a = __aarp_find_entry(proxies[hash], dev, sa); - hash = sa->s_node % (AARP_HASH_SIZE-1); - - retval = NULL; - a = __aarp_find_entry(proxies[hash], dev, sa); - if (a != NULL) - retval = sa; - - return retval; + return a ? sa : NULL; } - /* * Probe a Phase 1 device or a device that requires its Net:Node to * be set via an ioctl. @@ -506,13 +437,13 @@ void aarp_send_probe_phase1(struct atalk_iface *iface) struct sockaddr_at *sa = (struct sockaddr_at *)&atreq.ifr_addr; sa->sat_addr.s_node = iface->address.s_node; - sa->sat_addr.s_net = ntohs(iface->address.s_net); + sa->sat_addr.s_net = ntohs(iface->address.s_net); /* We pass the Net:Node to the drivers/cards by a Device ioctl. */ if (!(iface->dev->do_ioctl(iface->dev, &atreq, SIOCSIFADDR))) { (void)iface->dev->do_ioctl(iface->dev, &atreq, SIOCGIFADDR); - if ((iface->address.s_net != htons(sa->sat_addr.s_net)) || - (iface->address.s_node != sa->sat_addr.s_node)) + if (iface->address.s_net != htons(sa->sat_addr.s_net) || + iface->address.s_node != sa->sat_addr.s_node) iface->status |= ATIF_PROBE_FAIL; iface->address.s_net = htons(sa->sat_addr.s_net); @@ -523,17 +454,16 @@ void aarp_send_probe_phase1(struct atalk_iface *iface) void aarp_probe_network(struct atalk_iface *atif) { - if(atif->dev->type == ARPHRD_LOCALTLK || atif->dev->type == ARPHRD_PPP) { + if (atif->dev->type == ARPHRD_LOCALTLK || + atif->dev->type == ARPHRD_PPP) aarp_send_probe_phase1(atif); - } else { + else { unsigned int count; for (count = 0; count < AARP_RETRANSMIT_LIMIT; count++) { aarp_send_probe(atif->dev, &atif->address); - /* - * Defer 1/10th - */ + /* Defer 1/10th */ current->state = TASK_INTERRUPTIBLE; schedule_timeout(HZ/10); @@ -545,9 +475,9 @@ void aarp_probe_network(struct atalk_iface *atif) int aarp_proxy_probe_network(struct atalk_iface *atif, struct at_addr *sa) { - struct aarp_entry *entry; + int hash, retval = 1; + struct aarp_entry *entry; unsigned int count; - int hash, retval; /* * we don't currently support LocalTalk or PPP for proxy AARP; @@ -564,7 +494,7 @@ int aarp_proxy_probe_network(struct atalk_iface *atif, struct at_addr *sa) * we need this one to hang around even if it's in use */ entry = aarp_alloc(); - if (entry == NULL) + if (!entry) return -ENOMEM; entry->expires_at = -1; @@ -582,56 +512,38 @@ int aarp_proxy_probe_network(struct atalk_iface *atif, struct at_addr *sa) for (count = 0; count < AARP_RETRANSMIT_LIMIT; count++) { aarp_send_probe(atif->dev, sa); - /* - * Defer 1/10th - */ + /* Defer 1/10th */ current->state = TASK_INTERRUPTIBLE; - spin_unlock_bh(&aarp_lock); - schedule_timeout(HZ/10); - spin_lock_bh(&aarp_lock); if (entry->status & ATIF_PROBE_FAIL) break; } - retval = 1; - if (entry->status & ATIF_PROBE_FAIL) { - /* free the entry */ - entry->expires_at = jiffies - 1; - - /* return network full */ - retval = -EADDRINUSE; - } else { - /* clear the probing flag */ + entry->expires_at = jiffies - 1; /* free the entry */ + retval = -EADDRINUSE; /* return network full */ + } else /* clear the probing flag */ entry->status &= ~ATIF_PROBE; - } spin_unlock_bh(&aarp_lock); - return retval; } - -/* - * Send a DDP frame - */ -int aarp_send_ddp(struct net_device *dev,struct sk_buff *skb, struct at_addr *sa, void *hwaddr) +/* Send a DDP frame */ +int aarp_send_ddp(struct net_device *dev,struct sk_buff *skb, + struct at_addr *sa, void *hwaddr) { - static char ddp_eth_multicast[ETH_ALEN] = { 0x09, 0x00, 0x07, 0xFF, 0xFF, 0xFF }; + static char ddp_eth_multicast[ETH_ALEN] = + { 0x09, 0x00, 0x07, 0xFF, 0xFF, 0xFF }; int hash; struct aarp_entry *a; skb->nh.raw = skb->data; - /* - * Check for LocalTalk first - */ - - + /* Check for LocalTalk first */ if (dev->type == ARPHRD_LOCALTLK) { struct at_addr *at = atalk_find_dev_addr(dev); struct ddpehdr *ddp = (struct ddpehdr *)skb->data; @@ -644,8 +556,8 @@ int aarp_send_ddp(struct net_device *dev,struct sk_buff *skb, struct at_addr *sa * (zero matches anything) */ - if( ( ddp->deh_snet==0 || at->s_net==ddp->deh_snet) && - ( ddp->deh_dnet==0 || at->s_net==ddp->deh_dnet) ) { + if ((!ddp->deh_snet || at->s_net == ddp->deh_snet) && + (!ddp->deh_dnet || at->s_net == ddp->deh_dnet)) { skb_pull(skb, sizeof(struct ddpehdr) - 4); /* @@ -665,104 +577,58 @@ int aarp_send_ddp(struct net_device *dev,struct sk_buff *skb, struct at_addr *sa skb->data[0] = sa->s_node; skb->data[1] = at->s_node; skb->data[2] = ft; - - if (skb->sk) - skb->priority = skb->sk->priority; skb->dev = dev; - dev_queue_xmit(skb); - return 1; + goto sendit; } - /* - * On a PPP link we neither compress nor aarp. - */ + /* On a PPP link we neither compress nor aarp. */ if (dev->type == ARPHRD_PPP) { skb->protocol = htons(ETH_P_PPPTALK); - if (skb->sk) - skb->priority = skb->sk->priority; skb->dev = dev; - dev_queue_xmit(skb); - return 1; + goto sendit; } - /* - * Non ELAP we cannot do. - */ - + /* Non ELAP we cannot do. */ if (dev->type != ARPHRD_ETHER) return -1; skb->dev = dev; skb->protocol = htons(ETH_P_ATALK); - hash = sa->s_node % (AARP_HASH_SIZE - 1); - /* - * Do we have a resolved entry ? - */ - + /* Do we have a resolved entry? */ if (sa->s_node == ATADDR_BCAST) { ddp_dl->datalink_header(ddp_dl, skb, ddp_eth_multicast); - - if (skb->sk) - skb->priority = skb->sk->priority; - dev_queue_xmit(skb); - return 1; + goto sendit; } spin_lock_bh(&aarp_lock); - a = __aarp_find_entry(resolved[hash], dev, sa); - if (a != NULL) { - /* - * Return 1 and fill in the address - */ - + if (a) { /* Return 1 and fill in the address */ a->expires_at = jiffies + (sysctl_aarp_expiry_time * 10); ddp_dl->datalink_header(ddp_dl, skb, a->hwaddr); - if(skb->sk) - skb->priority = skb->sk->priority; - dev_queue_xmit(skb); - spin_unlock_bh(&aarp_lock); - return 1; + goto sendit; } - /* - * Do we have an unresolved entry: This is the less common path - */ - + /* Do we have an unresolved entry: This is the less common path */ a = __aarp_find_entry(unresolved[hash], dev, sa); - if (a != NULL) { - /* - * Queue onto the unresolved queue - */ - + if (a) { /* Queue onto the unresolved queue */ skb_queue_tail(&a->packet_queue, skb); - spin_unlock_bh(&aarp_lock); return 0; } - /* - * Allocate a new entry - */ - + /* Allocate a new entry */ a = aarp_alloc(); - if (a == NULL) { - /* - * Whoops slipped... good job it's an unreliable - * protocol 8) - */ + if (!a) { + /* Whoops slipped... good job it's an unreliable protocol 8) */ spin_unlock_bh(&aarp_lock); return -1; } - /* - * Set up the queue - */ - + /* Set up the queue */ skb_queue_tail(&a->packet_queue, skb); a->expires_at = jiffies + sysctl_aarp_resolve_time; a->dev = dev; @@ -772,10 +638,7 @@ int aarp_send_ddp(struct net_device *dev,struct sk_buff *skb, struct at_addr *sa unresolved[hash] = a; unresolved_count++; - /* - * Send an initial request for the address - */ - + /* Send an initial request for the address */ __aarp_send_query(a); /* @@ -786,18 +649,16 @@ int aarp_send_ddp(struct net_device *dev,struct sk_buff *skb, struct at_addr *sa if (unresolved_count == 1) mod_timer(&aarp_timer, jiffies + sysctl_aarp_tick_time); - - /* - * Now finally, it is safe to drop the lock. - */ - + /* Now finally, it is safe to drop the lock. */ spin_unlock_bh(&aarp_lock); - /* - * Tell the ddp layer we have taken over for this frame. - */ - + /* Tell the ddp layer we have taken over for this frame. */ return 0; + +sendit: if (skb->sk) + skb->priority = skb->sk->priority; + dev_queue_xmit(skb); + return 1; } /* @@ -806,170 +667,112 @@ int aarp_send_ddp(struct net_device *dev,struct sk_buff *skb, struct at_addr *sa * * Must run under aarp_lock. */ -static void __aarp_resolved(struct aarp_entry **list, struct aarp_entry *a, int hash) +static void __aarp_resolved(struct aarp_entry **list, struct aarp_entry *a, + int hash) { struct sk_buff *skb; - while (*list != NULL) { + while (*list) if (*list == a) { unresolved_count--; - *list = a->next; - /* - * Move into the resolved list - */ - + /* Move into the resolved list */ a->next = resolved[hash]; resolved[hash] = a; - /* - * Kick frames off - */ - + /* Kick frames off */ while ((skb = skb_dequeue(&a->packet_queue)) != NULL) { - a->expires_at = jiffies + (sysctl_aarp_expiry_time*10); + a->expires_at = jiffies + + sysctl_aarp_expiry_time * 10; ddp_dl->datalink_header(ddp_dl, skb, a->hwaddr); if (skb->sk) skb->priority = skb->sk->priority; dev_queue_xmit(skb); } - } else { + } else list = &((*list)->next); - } - } } /* * This is called by the SNAP driver whenever we see an AARP SNAP * frame. We currently only support Ethernet. */ -static int aarp_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt) +static int aarp_rcv(struct sk_buff *skb, struct net_device *dev, + struct packet_type *pt) { - struct elapaarp *ea=(struct elapaarp *)skb->h.raw; + struct elapaarp *ea = (struct elapaarp *)skb->h.raw; + int hash, ret = 0; + __u16 function; struct aarp_entry *a; struct at_addr sa, *ma, da; - int hash; struct atalk_iface *ifa; - /* - * We only do Ethernet SNAP AARP. - */ - - if (dev->type != ARPHRD_ETHER) { - kfree_skb(skb); - return 0; - } - - /* - * Frame size ok ? - */ - - if (!skb_pull(skb, sizeof(*ea))) { - kfree_skb(skb); - return 0; - } - - ea->function = ntohs(ea->function); + /* We only do Ethernet SNAP AARP. */ + if (dev->type != ARPHRD_ETHER) + goto out0; - /* - * Sanity check fields. - */ + /* Frame size ok? */ + if (!skb_pull(skb, sizeof(*ea))) + goto out0; - if (ea->function < AARP_REQUEST || - ea->function > AARP_PROBE || - ea->hw_len != ETH_ALEN || - ea->pa_len != AARP_PA_ALEN || - ea->pa_src_zero != 0 || - ea->pa_dst_zero != 0) { - kfree_skb(skb); - return 0; - } + function = ntohs(ea->function); - /* - * Looks good. - */ + /* Sanity check fields. */ + if (function < AARP_REQUEST || function > AARP_PROBE || + ea->hw_len != ETH_ALEN || ea->pa_len != AARP_PA_ALEN || + ea->pa_src_zero || ea->pa_dst_zero) + goto out0; + /* Looks good. */ hash = ea->pa_src_node % (AARP_HASH_SIZE - 1); - /* - * Build an address. - */ - + /* Build an address. */ sa.s_node = ea->pa_src_node; sa.s_net = ea->pa_src_net; - /* - * Process the packet. - * Check for replies of me. - */ - + /* Process the packet. Check for replies of me. */ ifa = atalk_find_dev(dev); - if (ifa == NULL) { - kfree_skb(skb); - return 1; - } - - if (ifa->status & ATIF_PROBE) { - if (ifa->address.s_node == ea->pa_dst_node && - ifa->address.s_net == ea->pa_dst_net) { - /* - * Fail the probe (in use) - */ - - ifa->status |= ATIF_PROBE_FAIL; - kfree_skb(skb); - return 1; - } + if (!ifa) + goto out1; + + if (ifa->status & ATIF_PROBE && + ifa->address.s_node == ea->pa_dst_node && + ifa->address.s_net == ea->pa_dst_net) { + ifa->status |= ATIF_PROBE_FAIL; /* Fail the probe (in use) */ + goto out1; } - /* - * Check for replies of proxy AARP entries - */ - + /* Check for replies of proxy AARP entries */ da.s_node = ea->pa_dst_node; da.s_net = ea->pa_dst_net; spin_lock_bh(&aarp_lock); - a = __aarp_find_entry(proxies[hash], dev, &da); - if (a != NULL) { - if (a->status & ATIF_PROBE) { - a->status |= ATIF_PROBE_FAIL; - - spin_unlock_bh(&aarp_lock); - - /* - * we do not respond to probe or request packets for - * this address while we are probing this address - */ - kfree_skb(skb); - - return 1; - } + if (a && a->status & ATIF_PROBE) { + a->status |= ATIF_PROBE_FAIL; + /* + * we do not respond to probe or request packets for + * this address while we are probing this address + */ + goto unlock; } - switch (ea->function) { + switch (function) { case AARP_REPLY: - if (unresolved_count == 0) /* Speed up */ + if (!unresolved_count) /* Speed up */ break; - /* - * Find the entry. - */ - - if ((a = __aarp_find_entry(unresolved[hash],dev,&sa)) == NULL || - (dev != a->dev)) + /* Find the entry. */ + a = __aarp_find_entry(unresolved[hash],dev,&sa); + if (!a || dev != a->dev) break; - /* - * We can fill one in - this is good. - */ - + /* We can fill one in - this is good. */ memcpy(a->hwaddr,ea->hw_src,ETH_ALEN); __aarp_resolved(&unresolved[hash],a,hash); - if (unresolved_count == 0) + if (!unresolved_count) mod_timer(&aarp_timer, jiffies + sysctl_aarp_expiry_time); break; @@ -977,49 +780,45 @@ static int aarp_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_t case AARP_REQUEST: case AARP_PROBE: /* - * If it is my address set ma to my address and reply. We can treat probe and - * request the same. Probe simply means we shouldn't cache the querying host, - * as in a probe they are proposing an address not using one. + * If it is my address set ma to my address and + * reply. We can treat probe and request the + * same. Probe simply means we shouldn't cache + * the querying host, as in a probe they are + * proposing an address not using one. * - * Support for proxy-AARP added. We check if the address is one - * of our proxies before we toss the packet out. + * Support for proxy-AARP added. We check if the + * address is one of our proxies before we toss + * the packet out. */ sa.s_node = ea->pa_dst_node; sa.s_net = ea->pa_dst_net; - /* - * See if we have a matching proxy. - */ + /* See if we have a matching proxy. */ ma = __aarp_proxy_find(dev, &sa); - if (!ma) { + if (!ma) ma = &ifa->address; - } else { - /* - * We need to make a copy of the entry. - */ + else { /* We need to make a copy of the entry. */ da.s_node = sa.s_node; da.s_net = da.s_net; ma = &da; } - if (ea->function == AARP_PROBE) { + if (function == AARP_PROBE) { /* A probe implies someone trying to get an * address. So as a precaution flush any - * entries we have for this address. - */ + * entries we have for this address. */ struct aarp_entry *a = __aarp_find_entry( - resolved[sa.s_node%(AARP_HASH_SIZE-1)], - skb->dev, - &sa); + resolved[sa.s_node%(AARP_HASH_SIZE-1)], + skb->dev, &sa); /* Make it expire next tick - that avoids us - * getting into a probe/flush/learn/probe/flush/learn - * cycle during probing of a slow to respond host addr. - */ - if (a != NULL) - { + * getting into a probe/flush/learn/probe/ + * flush/learn cycle during probing of a slow + * to respond host addr. */ + if (a) { a->expires_at = jiffies - 1; - mod_timer(&aarp_timer, jiffies + sysctl_aarp_tick_time); + mod_timer(&aarp_timer, jiffies + + sysctl_aarp_tick_time); } } @@ -1032,18 +831,16 @@ static int aarp_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_t sa.s_node = ea->pa_src_node; sa.s_net = ea->pa_src_net; - /* - * aarp_my_address has found the address to use for us. - */ - + /* aarp_my_address has found the address to use for us. + */ aarp_send_reply(dev, ma, &sa, ea->hw_src); break; - }; - - spin_unlock_bh(&aarp_lock); + } - kfree_skb(skb); - return 1; +unlock: spin_unlock_bh(&aarp_lock); +out1: ret = 1; +out0: kfree_skb(skb); + return ret; } static struct notifier_block aarp_notifier = { @@ -1054,10 +851,10 @@ static struct notifier_block aarp_notifier = { static char aarp_snap_id[] = { 0x00, 0x00, 0x00, 0x80, 0xF3 }; - void __init aarp_proto_init(void) { - if ((aarp_dl = register_snap_client(aarp_snap_id, aarp_rcv)) == NULL) + aarp_dl = register_snap_client(aarp_snap_id, aarp_rcv); + if (!aarp_dl) printk(KERN_CRIT "Unable to register AARP with SNAP.\n"); init_timer(&aarp_timer); aarp_timer.function = aarp_expire_timeout; @@ -1067,9 +864,7 @@ void __init aarp_proto_init(void) register_netdevice_notifier(&aarp_notifier); } -/* - * Remove the AARP entries associated with a device. - */ +/* Remove the AARP entries associated with a device. */ void aarp_device_down(struct net_device *dev) { int ct; @@ -1085,14 +880,12 @@ void aarp_device_down(struct net_device *dev) spin_unlock_bh(&aarp_lock); } -/* - * Called from proc fs - */ +/* Called from proc fs */ static int aarp_get_info(char *buffer, char **start, off_t offset, int length) { /* we should dump all our AARP entries */ - struct aarp_entry *entry; - int len, ct; + struct aarp_entry *entry; + int len, ct; len = sprintf(buffer, "%-10.10s ""%-10.10s""%-18.18s""%12.12s""%12.12s"" xmit_count status\n", @@ -1171,25 +964,19 @@ static int aarp_get_info(char *buffer, char **start, off_t offset, int length) } spin_unlock_bh(&aarp_lock); - return len; } #ifdef MODULE -/* - * General module cleanup. Called from cleanup_module() in ddp.c. - */ +/* General module cleanup. Called from cleanup_module() in ddp.c. */ void aarp_cleanup_module(void) { del_timer(&aarp_timer); unregister_netdevice_notifier(&aarp_notifier); unregister_snap_client(aarp_snap_id); } - #endif /* MODULE */ - #ifdef CONFIG_PROC_FS - void aarp_register_proc_fs(void) { proc_net_create("aarp", 0, aarp_get_info); @@ -1199,7 +986,5 @@ void aarp_unregister_proc_fs(void) { proc_net_remove("aarp"); } - #endif - #endif /* CONFIG_ATALK || CONFIG_ATALK_MODULE */ diff --git a/net/appletalk/ddp.c b/net/appletalk/ddp.c index e7784dabd..0e514faae 100644 --- a/net/appletalk/ddp.c +++ b/net/appletalk/ddp.c @@ -29,15 +29,17 @@ * driver file. (ipddp.c & ipddp.h) * Jay Schulist : Made work as module with * AppleTalk drivers, cleaned it. - * Rob Newberry : Added proxy AARP and AARP proc fs, - * moved probing to AARP module. + * Rob Newberry : Added proxy AARP and AARP + * procfs, moved probing to AARP + * module. * Adrian Sun/ * Michael Zuelsdorff : fix for net.0 packets. don't * allow illegal ether/tokentalk * port assignment. we lose a * valid localtalk port as a * result. - * + * Arnaldo C. de Melo : Cleanup, in preparation for + * shared skb support 8) * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public License @@ -84,19 +86,19 @@ #include <linux/stat.h> #include <linux/init.h> - #ifdef CONFIG_PROC_FS extern void aarp_register_proc_fs(void); extern void aarp_unregister_proc_fs(void); #endif +extern void aarp_cleanup_module(void); + extern void aarp_probe_network(struct atalk_iface *atif); -extern int aarp_proxy_probe_network(struct atalk_iface *atif, struct at_addr *sa); +extern int aarp_proxy_probe_network(struct atalk_iface *atif, + struct at_addr *sa); extern void aarp_proxy_remove(struct net_device *dev, struct at_addr *sa); - #undef APPLETALK_DEBUG - #ifdef APPLETALK_DEBUG #define DPRINT(x) print(x) #else @@ -119,13 +121,14 @@ static struct proto_ops atalk_dgram_ops; * * \**************************************************************************/ -static struct sock *atalk_sockets = NULL; +static struct sock *atalk_sockets; static spinlock_t atalk_sockets_lock = SPIN_LOCK_UNLOCKED; extern inline void atalk_insert_socket(struct sock *sk) { spin_lock_bh(&atalk_sockets_lock); - if ((sk->next = atalk_sockets) != NULL) + sk->next = atalk_sockets; + if (sk->next) atalk_sockets->pprev = &sk->next; atalk_sockets = sk; sk->pprev = &atalk_sockets; @@ -135,7 +138,7 @@ extern inline void atalk_insert_socket(struct sock *sk) extern inline void atalk_remove_socket(struct sock *sk) { spin_lock_bh(&atalk_sockets_lock); - if (sk->pprev != NULL) { + if (sk->pprev) { if (sk->next) sk->next->pprev = sk->pprev; *sk->pprev = sk->next; @@ -144,12 +147,13 @@ extern inline void atalk_remove_socket(struct sock *sk) spin_unlock_bh(&atalk_sockets_lock); } -static struct sock *atalk_search_socket(struct sockaddr_at *to, struct atalk_iface *atif) +static struct sock *atalk_search_socket(struct sockaddr_at *to, + struct atalk_iface *atif) { struct sock *s; spin_lock_bh(&atalk_sockets_lock); - for (s = atalk_sockets; s != NULL; s = s->next) { + for (s = atalk_sockets; s; s = s->next) { if (to->sat_port != s->protinfo.af_at.src_port) continue; @@ -174,7 +178,6 @@ static struct sock *atalk_search_socket(struct sockaddr_at *to, struct atalk_ifa } } spin_unlock_bh(&atalk_sockets_lock); - return s; } @@ -185,29 +188,28 @@ static struct sock *atalk_search_socket(struct sockaddr_at *to, struct atalk_ifa * * This entire operation must execute atomically. */ -static struct sock *atalk_find_or_insert_socket(struct sock *sk, struct sockaddr_at *sat) +static struct sock *atalk_find_or_insert_socket(struct sock *sk, + struct sockaddr_at *sat) { struct sock *s; spin_lock_bh(&atalk_sockets_lock); - - for (s = atalk_sockets; s != NULL; s = s->next) { + for (s = atalk_sockets; s; s = s->next) if (s->protinfo.af_at.src_net == sat->sat_addr.s_net && s->protinfo.af_at.src_node == sat->sat_addr.s_node && s->protinfo.af_at.src_port == sat->sat_port) break; - } if (!s) { /* Wheee, it's free, assign and insert. */ - if ((sk->next = atalk_sockets) != NULL) + sk->next = atalk_sockets; + if (sk->next) atalk_sockets->pprev = &sk->next; atalk_sockets = sk; sk->pprev = &atalk_sockets; } spin_unlock_bh(&atalk_sockets_lock); - return s; } @@ -215,9 +217,8 @@ static void atalk_destroy_timer(unsigned long data) { struct sock *sk = (struct sock *) data; - if (atomic_read(&sk->wmem_alloc) == 0 && - atomic_read(&sk->rmem_alloc) == 0 && - sk->dead) { + if (!atomic_read(&sk->wmem_alloc) && + !atomic_read(&sk->rmem_alloc) && sk->dead) { sock_put(sk); MOD_DEC_USE_COUNT; } else { @@ -228,16 +229,11 @@ static void atalk_destroy_timer(unsigned long data) extern inline void atalk_destroy_socket(struct sock *sk) { - struct sk_buff *skb; - atalk_remove_socket(sk); + skb_queue_purge(&sk->receive_queue); - while ((skb = skb_dequeue(&sk->receive_queue)) != NULL) - kfree_skb(skb); - - if (atomic_read(&sk->wmem_alloc) == 0 && - atomic_read(&sk->rmem_alloc) == 0 && - sk->dead) { + if (!atomic_read(&sk->wmem_alloc) && + !atomic_read(&sk->rmem_alloc) && sk->dead) { sock_put(sk); MOD_DEC_USE_COUNT; } else { @@ -249,37 +245,31 @@ extern inline void atalk_destroy_socket(struct sock *sk) } } -/* - * Called from proc fs - */ +/* Called from proc fs */ static int atalk_get_info(char *buffer, char **start, off_t offset, int length) { - struct sock *s; - int len = 0; off_t pos = 0; off_t begin = 0; - - /* - * Output the AppleTalk data for the /proc filesystem. - */ - - len += sprintf(buffer,"Type local_addr remote_addr tx_queue rx_queue st uid\n"); + int len = sprintf(buffer, "Type local_addr remote_addr tx_queue " + "rx_queue st uid\n"); + struct sock *s; + /* Output the AppleTalk data for the /proc filesystem */ spin_lock_bh(&atalk_sockets_lock); - for (s = atalk_sockets; s != NULL; s = s->next) { - len += sprintf(buffer+len,"%02X ", s->type); - len += sprintf(buffer+len,"%04X:%02X:%02X ", + for (s = atalk_sockets; s; s = s->next) { + len += sprintf(buffer + len,"%02X ", s->type); + len += sprintf(buffer + len,"%04X:%02X:%02X ", ntohs(s->protinfo.af_at.src_net), s->protinfo.af_at.src_node, s->protinfo.af_at.src_port); - len += sprintf(buffer+len,"%04X:%02X:%02X ", + len += sprintf(buffer + len,"%04X:%02X:%02X ", ntohs(s->protinfo.af_at.dest_net), s->protinfo.af_at.dest_node, s->protinfo.af_at.dest_port); - len += sprintf(buffer+len,"%08X:%08X ", + len += sprintf(buffer + len,"%08X:%08X ", atomic_read(&s->wmem_alloc), atomic_read(&s->rmem_alloc)); - len += sprintf(buffer+len,"%02X %d\n", s->state, + len += sprintf(buffer + len,"%02X %d\n", s->state, SOCK_INODE(s->socket)->i_uid); /* Are we still dumping unwanted data then discard the record */ @@ -295,8 +285,8 @@ static int atalk_get_info(char *buffer, char **start, off_t offset, int length) spin_unlock_bh(&atalk_sockets_lock); /* The data in question runs from begin to begin+len */ - *start = buffer + (offset - begin); /* Start of wanted data */ - len -= (offset - begin); /* Remove unwanted header data from length */ + *start = buffer + offset - begin; /* Start of wanted data */ + len -= offset - begin; /* Remove unwanted header data from length */ if (len > length) len = length; /* Remove unwanted tail data from length */ @@ -310,18 +300,16 @@ static int atalk_get_info(char *buffer, char **start, off_t offset, int length) \**************************************************************************/ /* Anti-deadlock ordering is router_lock --> iface_lock -DaveM */ -static struct atalk_route *atalk_router_list = NULL; +static struct atalk_route *atalk_router_list; static rwlock_t atalk_router_lock = RW_LOCK_UNLOCKED; -static struct atalk_iface *atalk_iface_list = NULL; +static struct atalk_iface *atalk_iface_list; static spinlock_t atalk_iface_lock = SPIN_LOCK_UNLOCKED; -static struct atalk_route atrtr_default; /* For probing devices or in a routerless network */ - -/* - * AppleTalk interface control - */ +/* For probing devices or in a routerless network */ +static struct atalk_route atrtr_default; +/* AppleTalk interface control */ /* * Drop a device. Doesn't drop any of its routes - that is the caller's * problem. Called when we down the interface or delete the address. @@ -344,12 +332,12 @@ static void atif_drop_device(struct net_device *dev) spin_unlock_bh(&atalk_iface_lock); } -static struct atalk_iface *atif_add_device(struct net_device *dev, struct at_addr *sa) +static struct atalk_iface *atif_add_device(struct net_device *dev, + struct at_addr *sa) { - struct atalk_iface *iface = (struct atalk_iface *) - kmalloc(sizeof(*iface), GFP_KERNEL); + struct atalk_iface *iface = kmalloc(sizeof(*iface), GFP_KERNEL); - if (iface == NULL) + if (!iface) return NULL; iface->dev = dev; @@ -363,51 +351,37 @@ static struct atalk_iface *atif_add_device(struct net_device *dev, struct at_add spin_unlock_bh(&atalk_iface_lock); MOD_INC_USE_COUNT; - return iface; } - -/* - * Perform phase 2 AARP probing on our tentative address. - */ +/* Perform phase 2 AARP probing on our tentative address */ static int atif_probe_device(struct atalk_iface *atif) { - int netrange = ntohs(atif->nets.nr_lastnet) - ntohs(atif->nets.nr_firstnet) + 1; + int netrange = ntohs(atif->nets.nr_lastnet) - + ntohs(atif->nets.nr_firstnet) + 1; int probe_net = ntohs(atif->address.s_net); int probe_node = atif->address.s_node; int netct, nodect; - /* - * Offset the network we start probing with. - */ - + /* Offset the network we start probing with */ if (probe_net == ATADDR_ANYNET) { - if (!netrange) - probe_net = ntohs(atif->nets.nr_firstnet); - else - probe_net = ntohs(atif->nets.nr_firstnet) + (jiffies % netrange); + probe_net = ntohs(atif->nets.nr_firstnet); + if (netrange) + probe_net += jiffies % netrange; } - if (probe_node == ATADDR_ANYNODE) probe_node = jiffies & 0xFF; - /* - * Scan the networks. - */ + /* Scan the networks */ atif->status |= ATIF_PROBE; for (netct = 0; netct <= netrange; netct++) { - /* - * Sweep the available nodes from a given start. - */ - + /* Sweep the available nodes from a given start */ atif->address.s_net = htons(probe_net); for (nodect = 0; nodect < 256; nodect++) { atif->address.s_node = ((nodect+probe_node) & 0xFF); - if (atif->address.s_node > 0 && atif->address.s_node < 254) { - /* - * Probe a proposed address. - */ + if (atif->address.s_node > 0 && + atif->address.s_node < 254) { + /* Probe a proposed address */ aarp_probe_network(atif); if (!(atif->status & ATIF_PROBE_FAIL)) { @@ -427,58 +401,45 @@ static int atif_probe_device(struct atalk_iface *atif) } -/* - * Perform AARP probing for a proxy address - */ -static int atif_proxy_probe_device(struct atalk_iface *atif, struct at_addr* proxy_addr) +/* Perform AARP probing for a proxy address */ +static int atif_proxy_probe_device(struct atalk_iface *atif, + struct at_addr* proxy_addr) { - int netrange = ntohs(atif->nets.nr_lastnet) - ntohs(atif->nets.nr_firstnet) + 1; - int probe_net = ntohs(atif->address.s_net); /* we probe the interface's network */ + int netrange = ntohs(atif->nets.nr_lastnet) - + ntohs(atif->nets.nr_firstnet) + 1; + /* we probe the interface's network */ + int probe_net = ntohs(atif->address.s_net); int probe_node = ATADDR_ANYNODE; /* we'll take anything */ int netct, nodect; - /* - * Offset the network we start probing with. - */ - + /* Offset the network we start probing with */ if (probe_net == ATADDR_ANYNET) { - if (!netrange) - probe_net = ntohs(atif->nets.nr_firstnet); - else - probe_net = ntohs(atif->nets.nr_firstnet) + (jiffies % netrange); + probe_net = ntohs(atif->nets.nr_firstnet); + if (netrange) + probe_net += jiffies % netrange; } if (probe_node == ATADDR_ANYNODE) probe_node = jiffies & 0xFF; - /* - * Scan the networks. - */ - + /* Scan the networks */ for (netct = 0; netct <= netrange; netct++) { - /* - * Sweep the available nodes from a given start. - */ - + /* Sweep the available nodes from a given start */ proxy_addr->s_net = htons(probe_net); for (nodect = 0; nodect < 256; nodect++) { proxy_addr->s_node = ((nodect + probe_node) & 0xFF); - if ((proxy_addr->s_node > 0) && (proxy_addr->s_node < 254)) { - /* - * Tell AARP to probe a proposed address. - */ - int probe_result = aarp_proxy_probe_network(atif, - proxy_addr); - - if (probe_result == 0) - return 0; - - if (probe_result != -EADDRINUSE) - return probe_result; + if (proxy_addr->s_node > 0 && + proxy_addr->s_node < 254) { + /* Tell AARP to probe a proposed address */ + int ret = aarp_proxy_probe_network(atif, + proxy_addr); + + if (ret != -EADDRINUSE) + return ret; } } probe_net++; - if(probe_net > ntohs(atif->nets.nr_lastnet)) + if (probe_net > ntohs(atif->nets.nr_lastnet)) probe_net = ntohs(atif->nets.nr_firstnet); } @@ -489,25 +450,21 @@ static int atif_proxy_probe_device(struct atalk_iface *atif, struct at_addr* pro struct at_addr *atalk_find_dev_addr(struct net_device *dev) { struct atalk_iface *iface = dev->atalk_ptr; - - if(iface) - return &iface->address; - - return NULL; + return iface ? &iface->address : NULL; } static struct at_addr *atalk_find_primary(void) { + struct atalk_iface *fiface = NULL; struct at_addr *retval; struct atalk_iface *iface; - struct atalk_iface *fiface = NULL; /* * Return a point-to-point interface only if * there is no non-ptp interface available. */ spin_lock_bh(&atalk_iface_lock); - for (iface = atalk_iface_list; iface != NULL; iface = iface->next) { + for (iface = atalk_iface_list; iface; iface = iface->next) { if (!fiface && !(iface->dev->flags & IFF_LOOPBACK)) fiface = iface; if (!(iface->dev->flags & (IFF_LOOPBACK | IFF_POINTOPOINT))) { @@ -516,16 +473,13 @@ static struct at_addr *atalk_find_primary(void) } } - if (fiface) { + if (fiface) retval = &fiface->address; - } else if (atalk_iface_list != NULL) { + else if (atalk_iface_list) retval = &atalk_iface_list->address; - } else { + else retval = NULL; - } -out: - spin_unlock_bh(&atalk_iface_lock); - +out: spin_unlock_bh(&atalk_iface_lock); return retval; } @@ -535,9 +489,9 @@ out: */ static struct atalk_iface *atalk_find_anynet(int node, struct net_device *dev) { - struct atalk_iface *iface=dev->atalk_ptr; + struct atalk_iface *iface = dev->atalk_ptr; - if (iface==NULL || (iface->status & ATIF_PROBE)) + if (!iface || iface->status & ATIF_PROBE) return NULL; if (node == ATADDR_BCAST || @@ -548,15 +502,13 @@ static struct atalk_iface *atalk_find_anynet(int node, struct net_device *dev) return NULL; } -/* - * Find a match for a specific network:node pair - */ +/* Find a match for a specific network:node pair */ static struct atalk_iface *atalk_find_interface(int net, int node) { struct atalk_iface *iface; spin_lock_bh(&atalk_iface_lock); - for (iface = atalk_iface_list; iface != NULL; iface = iface->next) { + for (iface = atalk_iface_list; iface; iface = iface->next) { if ((node == ATADDR_BCAST || node == ATADDR_ANYNODE || iface->address.s_node == node) && @@ -565,13 +517,12 @@ static struct atalk_iface *atalk_find_interface(int net, int node) break; /* XXXX.0 -- net.0 returns the iface associated with net */ - if ((node == ATADDR_ANYNODE) && (net != ATADDR_ANYNET) && - (ntohs(iface->nets.nr_firstnet) <= ntohs(net)) && - (ntohs(net) <= ntohs(iface->nets.nr_lastnet))) + if (node == ATADDR_ANYNODE && net != ATADDR_ANYNET && + ntohs(iface->nets.nr_firstnet) <= ntohs(net) && + ntohs(net) <= ntohs(iface->nets.nr_lastnet)) break; } spin_unlock_bh(&atalk_iface_lock); - return iface; } @@ -588,11 +539,11 @@ static struct atalk_route *atrtr_find(struct at_addr *target) * host route, because some host routes might overlap * network routes */ - struct atalk_route *r; struct atalk_route *net_route = NULL; + struct atalk_route *r; read_lock_bh(&atalk_router_lock); - for (r = atalk_router_list; r != NULL; r = r->next) { + for (r = atalk_router_list; r; r = r->next) { if (!(r->flags & RTF_UP)) continue; @@ -604,13 +555,12 @@ static struct atalk_route *atrtr_find(struct at_addr *target) */ if (r->target.s_node == target->s_node) goto out; - } else { + } else /* * this route will work if there isn't a * direct host route, so cache it */ net_route = r; - } } } @@ -618,19 +568,13 @@ static struct atalk_route *atrtr_find(struct at_addr *target) * if we found a network route but not a direct host * route, then return it */ - if (net_route != NULL) { + if (net_route) r = net_route; - } else if (atrtr_default.dev) { + else if (atrtr_default.dev) r = &atrtr_default; - } else { - /* - * No route can be found. - */ + else /* No route can be found */ r = NULL; - } - -out: - read_unlock_bh(&atalk_router_lock); +out: read_unlock_bh(&atalk_router_lock); return r; } @@ -642,16 +586,10 @@ out: struct net_device *atrtr_get_dev(struct at_addr *sa) { struct atalk_route *atr = atrtr_find(sa); - - if (atr == NULL) - return NULL; - else - return atr->dev; + return atr ? atr->dev : NULL; } -/* - * Set up a default router. - */ +/* Set up a default router */ static void atrtr_set_default(struct net_device *dev) { atrtr_default.dev = dev; @@ -678,20 +616,16 @@ static int atrtr_create(struct rtentry *r, struct net_device *devhint) * operations. */ - /* - * Validate the request - */ + /* Validate the request */ if (ta->sat_family != AF_APPLETALK) return -EINVAL; - if (devhint == NULL && ga->sat_family != AF_APPLETALK) + if (!devhint && ga->sat_family != AF_APPLETALK) return -EINVAL; - /* - * Now walk the routing table and make our decisions. - */ + /* Now walk the routing table and make our decisions */ write_lock_bh(&atalk_router_lock); - for (rt = atalk_router_list; rt != NULL; rt = rt->next) { + for (rt = atalk_router_list; rt; rt = rt->next) { if (r->rt_flags != rt->flags) continue; @@ -703,65 +637,59 @@ static int atrtr_create(struct rtentry *r, struct net_device *devhint) } } - if(devhint == NULL) { + if (!devhint) { riface = NULL; spin_lock_bh(&atalk_iface_lock); for (iface = atalk_iface_list; iface; iface = iface->next) { - if (riface == NULL && - ntohs(ga->sat_addr.s_net) >= ntohs(iface->nets.nr_firstnet) && - ntohs(ga->sat_addr.s_net) <= ntohs(iface->nets.nr_lastnet)) + if (!riface && + ntohs(ga->sat_addr.s_net) >= + ntohs(iface->nets.nr_firstnet) && + ntohs(ga->sat_addr.s_net) <= + ntohs(iface->nets.nr_lastnet)) riface = iface; - if (ga->sat_addr.s_net == iface->address.s_net && + if (ga->sat_addr.s_net == iface->address.s_net && ga->sat_addr.s_node == iface->address.s_node) riface = iface; } spin_unlock_bh(&atalk_iface_lock); retval = -ENETUNREACH; - if (riface == NULL) + if (!riface) goto out; devhint = riface->dev; } - if (rt == NULL) { - rt = (struct atalk_route *) - kmalloc(sizeof(struct atalk_route), GFP_ATOMIC); + if (!rt) { + rt = kmalloc(sizeof(struct atalk_route), GFP_ATOMIC); retval = -ENOBUFS; - if (rt == NULL) + if (!rt) goto out; rt->next = atalk_router_list; atalk_router_list = rt; } - /* - * Fill in the routing entry. - */ + /* Fill in the routing entry */ rt->target = ta->sat_addr; rt->dev = devhint; rt->flags = r->rt_flags; rt->gateway = ga->sat_addr; retval = 0; - -out: - write_unlock_bh(&atalk_router_lock); - +out: write_unlock_bh(&atalk_router_lock); return retval; } -/* - * Delete a route. Find it and discard it. - */ +/* Delete a route. Find it and discard it */ static int atrtr_delete(struct at_addr * addr) { struct atalk_route **r = &atalk_router_list; - struct atalk_route *tmp; int retval = 0; + struct atalk_route *tmp; write_lock_bh(&atalk_router_lock); while ((tmp = *r) != NULL) { @@ -775,8 +703,7 @@ static int atrtr_delete(struct at_addr * addr) r = &tmp->next; } retval = -ENOENT; -out: - write_unlock_bh(&atalk_router_lock); +out: write_unlock_bh(&atalk_router_lock); return retval; } @@ -794,9 +721,8 @@ void atrtr_device_down(struct net_device *dev) if (tmp->dev == dev) { *r = tmp->next; kfree(tmp); - } else { + } else r = &tmp->next; - } } write_unlock_bh(&atalk_router_lock); @@ -804,9 +730,7 @@ void atrtr_device_down(struct net_device *dev) atrtr_set_default(NULL); } -/* - * Actually down the interface. - */ +/* Actually down the interface */ static inline void atalk_dev_down(struct net_device *dev) { atrtr_device_down(dev); /* Remove all routes for the device */ @@ -818,27 +742,22 @@ static inline void atalk_dev_down(struct net_device *dev) * A device event has occurred. Watch for devices going down and * delete our use of them (iface and route). */ -static int ddp_device_event(struct notifier_block *this, unsigned long event, void *ptr) +static int ddp_device_event(struct notifier_block *this, unsigned long event, + void *ptr) { - if (event == NETDEV_DOWN) { + if (event == NETDEV_DOWN) /* Discard any use of this */ atalk_dev_down((struct net_device *) ptr); - } return NOTIFY_DONE; } -/* - * ioctl calls. Shouldn't even need touching. - */ - -/* - * Device configuration ioctl calls. - */ +/* ioctl calls. Shouldn't even need touching */ +/* Device configuration ioctl calls */ int atif_ioctl(int cmd, void *arg) { - struct ifreq atreq; static char aarp_mcast[6] = {0x09, 0x00, 0x00, 0xFF, 0xFF, 0xFF}; + struct ifreq atreq; struct netrange *nr; struct sockaddr_at *sa; struct net_device *dev; @@ -846,12 +765,13 @@ int atif_ioctl(int cmd, void *arg) int ct; int limit; struct rtentry rtdef; - int add_route; + int add_route; if (copy_from_user(&atreq, arg, sizeof(atreq))) return -EFAULT; - if ((dev = __dev_get_by_name(atreq.ifr_name)) == NULL) + dev = __dev_get_by_name(atreq.ifr_name); + if (!dev) return -ENODEV; sa = (struct sockaddr_at*) &atreq.ifr_addr; @@ -870,16 +790,19 @@ int atif_ioctl(int cmd, void *arg) return -EPROTONOSUPPORT; nr = (struct netrange *) &sa->sat_zero[0]; - add_route = 1; /* - * if this is a point-to-point iface, and we already have an - * iface for this AppleTalk address, then we should not add a route + * if this is a point-to-point iface, and we already + * have an iface for this AppleTalk address, then we + * should not add a route */ if ((dev->flags & IFF_POINTOPOINT) && - atalk_find_interface(sa->sat_addr.s_net, sa->sat_addr.s_node)) { - printk(KERN_DEBUG "AppleTalk: point-to-point interface added with existing address\n"); + atalk_find_interface(sa->sat_addr.s_net, + sa->sat_addr.s_node)) { + printk(KERN_DEBUG "AppleTalk: point-to-point " + "interface added with " + "existing address\n"); add_route = 0; } @@ -889,13 +812,11 @@ int atif_ioctl(int cmd, void *arg) */ if (dev->type == ARPHRD_ETHER && nr->nr_phase != 2) return -EPROTONOSUPPORT; - if (sa->sat_addr.s_node == ATADDR_BCAST || + if (sa->sat_addr.s_node == ATADDR_BCAST || sa->sat_addr.s_node == 254) return -EINVAL; if (atif) { - /* - * Already setting address. - */ + /* Already setting address */ if (atif->status & ATIF_PROBE) return -EBUSY; @@ -904,7 +825,7 @@ int atif_ioctl(int cmd, void *arg) atrtr_device_down(dev); /* Flush old routes */ } else { atif = atif_add_device(dev, &sa->sat_addr); - if (atif == NULL) + if (!atif) return -ENOMEM; } atif->nets = *nr; @@ -921,10 +842,7 @@ int atif_ioctl(int cmd, void *arg) return -EADDRINUSE; } - /* - * Hey it worked - add the direct routes. - */ - + /* Hey it worked - add the direct routes */ sa = (struct sockaddr_at *) &rtdef.rt_gateway; sa->sat_family = AF_APPLETALK; sa->sat_addr.s_net = atif->address.s_net; @@ -933,13 +851,11 @@ int atif_ioctl(int cmd, void *arg) rtdef.rt_flags = RTF_UP; sa->sat_family = AF_APPLETALK; sa->sat_addr.s_node = ATADDR_ANYNODE; - if ((dev->flags & IFF_LOOPBACK) || - (dev->flags & IFF_POINTOPOINT)) + if (dev->flags & IFF_LOOPBACK || + dev->flags & IFF_POINTOPOINT) rtdef.rt_flags |= RTF_HOST; - /* - * Routerless initial state. - */ + /* Routerless initial state */ if (nr->nr_firstnet == htons(0) && nr->nr_lastnet == htons(0xFFFE)) { sa->sat_addr.s_net = atif->address.s_net; @@ -948,37 +864,35 @@ int atif_ioctl(int cmd, void *arg) } else { limit = ntohs(nr->nr_lastnet); if (limit - ntohs(nr->nr_firstnet) > 4096) { - printk(KERN_WARNING "Too many routes/iface.\n"); + printk(KERN_WARNING "Too many routes/" + "iface.\n"); return -EINVAL; } - if (add_route) { - for(ct = ntohs(nr->nr_firstnet);ct <= limit; ct++) { + if (add_route) + for (ct = ntohs(nr->nr_firstnet); + ct <= limit; ct++) { sa->sat_addr.s_net = htons(ct); atrtr_create(&rtdef, dev); } - } } dev_mc_add(dev, aarp_mcast, 6, 1); return 0; case SIOCGIFADDR: - if (atif == NULL) + if (!atif) return -EADDRNOTAVAIL; - ((struct sockaddr_at *)(&atreq.ifr_addr))->sat_family = - AF_APPLETALK; - ((struct sockaddr_at *)(&atreq.ifr_addr))->sat_addr = - atif->address; + + sa->sat_family = AF_APPLETALK; + sa->sat_addr = atif->address; break; case SIOCGIFBRDADDR: - if (atif == NULL) + if (!atif) return -EADDRNOTAVAIL; - ((struct sockaddr_at *)(&atreq.ifr_addr))->sat_family = - AF_APPLETALK; - ((struct sockaddr_at *)(&atreq.ifr_addr))->sat_addr.s_net = - atif->address.s_net; - ((struct sockaddr_at *)(&atreq.ifr_addr))->sat_addr.s_node = - ATADDR_BCAST; + + sa->sat_family = AF_APPLETALK; + sa->sat_addr.s_net = atif->address.s_net; + sa->sat_addr.s_node = ATADDR_BCAST; break; case SIOCATALKDIFADDR: @@ -995,7 +909,7 @@ int atif_ioctl(int cmd, void *arg) return -EPERM; if (sa->sat_family != AF_APPLETALK) return -EINVAL; - if (atif == NULL) + if (!atif) return -EADDRNOTAVAIL; /* @@ -1007,9 +921,10 @@ int atif_ioctl(int cmd, void *arg) /* * atif points to the current interface on this network; - * we aren't concerned about its current status (at least for now), - * but it has all the settings about the network we're going - * to probe. consequently, it must exist. + * we aren't concerned about its current status (at + * least for now), but it has all the settings about + * the network we're going to probe. Consequently, it + * must exist. */ if (!atif) return -EADDRNOTAVAIL; @@ -1034,10 +949,10 @@ int atif_ioctl(int cmd, void *arg) return -EADDRINUSE; /* - * We now have an address on the local network, and the AARP - * code will defend it for us until we take it down. - * We don't set up any routes right now, because ATCP will - * install them manually via SIOCADDRT. + * We now have an address on the local network, and + * the AARP code will defend it for us until we take it + * down. We don't set up any routes right now, because + * ATCP will install them manually via SIOCADDRT. */ break; @@ -1046,30 +961,22 @@ int atif_ioctl(int cmd, void *arg) return -EPERM; if (sa->sat_family != AF_APPLETALK) return -EINVAL; - if (atif == NULL) + if (!atif) return -EADDRNOTAVAIL; - /* - * give to aarp module to remove proxy entry - */ + /* give to aarp module to remove proxy entry */ aarp_proxy_remove(atif->dev, &(sa->sat_addr)); - return 0; - }; - - if (copy_to_user(arg, &atreq, sizeof(atreq))) - return -EFAULT; + } - return 0; + return copy_to_user(arg, &atreq, sizeof(atreq)) ? -EFAULT : 0; } -/* - * Routing ioctl() calls - */ +/* Routing ioctl() calls */ static int atrtr_ioctl(unsigned int cmd, void *arg) { - struct rtentry rt; struct net_device *dev = NULL; + struct rtentry rt; if (copy_from_user(&rt, arg, sizeof(rt))) return -EFAULT; @@ -1078,37 +985,38 @@ static int atrtr_ioctl(unsigned int cmd, void *arg) case SIOCDELRT: if (rt.rt_dst.sa_family != AF_APPLETALK) return -EINVAL; - return atrtr_delete(&((struct sockaddr_at *)&rt.rt_dst)->sat_addr); + return atrtr_delete(&((struct sockaddr_at *) + &rt.rt_dst)->sat_addr); case SIOCADDRT: - /* FIX ME: the name of the device is still in user space, isn't it? */ - if (rt.rt_dev != NULL) { - if ((dev = __dev_get_by_name(rt.rt_dev)) == NULL) + /* FIXME: the name of the device is still in user + * space, isn't it? */ + if (rt.rt_dev) { + dev = __dev_get_by_name(rt.rt_dev); + if (!dev) return -ENODEV; } return atrtr_create(&rt, dev); - - default: - return -EINVAL; - }; + } + return -EINVAL; } /* Called from proc fs - just make it print the ifaces neatly */ - -static int atalk_if_get_info(char *buffer, char **start, off_t offset, int length) +static int atalk_if_get_info(char *buffer, char **start, off_t offset, + int length) { - struct atalk_iface *iface; - int len = 0; off_t pos = 0; off_t begin = 0; - - len += sprintf(buffer,"Interface Address Networks Status\n"); + struct atalk_iface *iface; + int len = sprintf(buffer, "Interface Address " + "Networks Status\n"); spin_lock_bh(&atalk_iface_lock); - for (iface = atalk_iface_list; iface != NULL; iface = iface->next) { + for (iface = atalk_iface_list; iface; iface = iface->next) { len += sprintf(buffer+len,"%-16s %04X:%02X %04X-%04X %d\n", iface->dev->name, ntohs(iface->address.s_net), - iface->address.s_node, ntohs(iface->nets.nr_firstnet), + iface->address.s_node, + ntohs(iface->nets.nr_firstnet), ntohs(iface->nets.nr_lastnet), iface->status); pos = begin + len; if (pos < offset) { @@ -1124,33 +1032,32 @@ static int atalk_if_get_info(char *buffer, char **start, off_t offset, int lengt len -= (offset - begin); if (len > length) len = length; - - return (len); + return len; } /* Called from proc fs - just make it print the routes neatly */ - -static int atalk_rt_get_info(char *buffer, char **start, off_t offset, int length) +static int atalk_rt_get_info(char *buffer, char **start, off_t offset, + int length) { - struct atalk_route *rt; - int len = 0; off_t pos = 0; off_t begin = 0; + int len = sprintf(buffer, "Target Router Flags Dev\n"); + struct atalk_route *rt; - len += sprintf(buffer,"Target Router Flags Dev\n"); if (atrtr_default.dev) { rt = &atrtr_default; - len += sprintf(buffer+len,"Default %04X:%02X %-4d %s\n", + len += sprintf(buffer + len,"Default %04X:%02X %-4d %s\n", ntohs(rt->gateway.s_net), rt->gateway.s_node, rt->flags, rt->dev->name); } read_lock_bh(&atalk_router_lock); - for (rt = atalk_router_list; rt != NULL; rt = rt->next) { - len += sprintf(buffer+len,"%04X:%02X %04X:%02X %-4d %s\n", + for (rt = atalk_router_list; rt; rt = rt->next) { + len += sprintf(buffer + len, + "%04X:%02X %04X:%02X %-4d %s\n", ntohs(rt->target.s_net), rt->target.s_node, - ntohs(rt->gateway.s_net), rt->gateway.s_node, rt->flags, - rt->dev->name); + ntohs(rt->gateway.s_net), rt->gateway.s_node, + rt->flags, rt->dev->name); pos = begin + len; if (pos < offset) { len = 0; @@ -1165,7 +1072,6 @@ static int atalk_rt_get_info(char *buffer, char **start, off_t offset, int lengt len -= (offset - begin); if (len > length) len = length; - return len; } @@ -1198,11 +1104,8 @@ unsigned short atalk_checksum(struct ddpehdr *ddp, int len) } data++; } - - if (sum) - return htons((unsigned short) sum); - - return 0xFFFF; /* Use 0xFFFF for 0. 0 itself means none */ + /* Use 0xFFFF for 0. 0 itself means none */ + return sum ? htons((unsigned short) sum) : 0xFFFF; } /* @@ -1211,10 +1114,9 @@ unsigned short atalk_checksum(struct ddpehdr *ddp, int len) */ static int atalk_create(struct socket *sock, int protocol) { - struct sock *sk; + struct sock *sk = sk_alloc(PF_APPLETALK, GFP_KERNEL, 1); - sk = sk_alloc(PF_APPLETALK, GFP_KERNEL, 1); - if (sk == NULL) + if (!sk) return -ENOMEM; switch (sock->type) { @@ -1230,36 +1132,32 @@ static int atalk_create(struct socket *sock, int protocol) case SOCK_STREAM: /* - * TO DO: if you want to implement ADSP, here's the place to start + * TODO: if you want to implement ADSP, here's the + * place to start */ /* sock->ops = &atalk_stream_ops; break; */ default: - sk_free((void *) sk); + sk_free(sk); return -ESOCKTNOSUPPORT; - }; + } MOD_INC_USE_COUNT; - sock_init_data(sock, sk); - sk->destruct = NULL; /* Checksums on by default */ sk->zapped = 1; - return 0; } -/* - * Free a socket. No work needed - */ +/* Free a socket. No work needed */ static int atalk_release(struct socket *sock) { - struct sock *sk=sock->sk; + struct sock *sk = sock->sk; - if (sk == NULL) + if (!sk) return 0; if (!sk->dead) @@ -1268,7 +1166,6 @@ static int atalk_release(struct socket *sock) sk->dead = 1; sock->sk = NULL; atalk_destroy_socket(sk); - return 0; } @@ -1289,21 +1186,21 @@ static int atalk_pick_and_bind_port(struct sock *sk, struct sockaddr_at *sat) for (sat->sat_port = ATPORT_RESERVED; sat->sat_port < ATPORT_LAST; sat->sat_port++) { - for (s = atalk_sockets; s != NULL; s = s->next) { + for (s = atalk_sockets; s; s = s->next) { if (s->protinfo.af_at.src_net == sat->sat_addr.s_net && - s->protinfo.af_at.src_node == sat->sat_addr.s_node && + s->protinfo.af_at.src_node == + sat->sat_addr.s_node && s->protinfo.af_at.src_port == sat->sat_port) goto try_next_port; } /* Wheee, it's free, assign and insert. */ - if ((sk->next = atalk_sockets) != NULL) + sk->next = atalk_sockets; + if (sk->next) atalk_sockets->pprev = &sk->next; atalk_sockets = sk; sk->pprev = &atalk_sockets; - sk->protinfo.af_at.src_port = sat->sat_port; - retval = 0; goto out; @@ -1312,61 +1209,53 @@ static int atalk_pick_and_bind_port(struct sock *sk, struct sockaddr_at *sat) } retval = -EBUSY; -out: - spin_unlock_bh(&atalk_sockets_lock); - +out: spin_unlock_bh(&atalk_sockets_lock); return retval; } static int atalk_autobind(struct sock *sk) { - struct at_addr *ap = atalk_find_primary(); struct sockaddr_at sat; int n; + struct at_addr *ap = atalk_find_primary(); - if (ap == NULL || ap->s_net == htons(ATADDR_ANYNET)) + if (!ap || ap->s_net == htons(ATADDR_ANYNET)) return -EADDRNOTAVAIL; sk->protinfo.af_at.src_net = sat.sat_addr.s_net = ap->s_net; sk->protinfo.af_at.src_node = sat.sat_addr.s_node = ap->s_node; - if ((n = atalk_pick_and_bind_port(sk, &sat)) < 0) + n = atalk_pick_and_bind_port(sk, &sat); + if (n < 0) return n; sk->zapped = 0; - return 0; } -/* - * Set the address 'our end' of the connection. - */ +/* Set the address 'our end' of the connection */ static int atalk_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) { - struct sock *sk; struct sockaddr_at *addr = (struct sockaddr_at *)uaddr; + struct sock *sk = sock->sk; - sk = sock->sk; - - if(sk->zapped == 0) - return -EINVAL; - - if(addr_len != sizeof(struct sockaddr_at)) + if (!sk->zapped || addr_len != sizeof(struct sockaddr_at)) return -EINVAL; - if(addr->sat_family != AF_APPLETALK) + if (addr->sat_family != AF_APPLETALK) return -EAFNOSUPPORT; - if(addr->sat_addr.s_net == htons(ATADDR_ANYNET)) { + if (addr->sat_addr.s_net == htons(ATADDR_ANYNET)) { struct at_addr *ap = atalk_find_primary(); - if(ap == NULL) + if (!ap) return -EADDRNOTAVAIL; sk->protinfo.af_at.src_net = addr->sat_addr.s_net = ap->s_net; sk->protinfo.af_at.src_node = addr->sat_addr.s_node= ap->s_node; } else { - if (atalk_find_interface(addr->sat_addr.s_net, addr->sat_addr.s_node) == NULL) + if (!atalk_find_interface(addr->sat_addr.s_net, + addr->sat_addr.s_node)) return -EADDRNOTAVAIL; sk->protinfo.af_at.src_net = addr->sat_addr.s_net; @@ -1381,18 +1270,15 @@ static int atalk_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) } else { sk->protinfo.af_at.src_port = addr->sat_port; - if (atalk_find_or_insert_socket(sk, addr) != NULL) + if (atalk_find_or_insert_socket(sk, addr)) return -EADDRINUSE; } sk->zapped = 0; - return 0; } -/* - * Set the address we talk to. - */ +/* Set the address we talk to */ static int atalk_connect(struct socket *sock, struct sockaddr *uaddr, int addr_len, int flags) { @@ -1412,19 +1298,20 @@ static int atalk_connect(struct socket *sock, struct sockaddr *uaddr, if (addr->sat_addr.s_node == ATADDR_BCAST && !sk->broadcast) { #if 1 - printk(KERN_WARNING "%s is broken and did not set SO_BROADCAST. It will break when 2.2 is released.\n", + printk(KERN_WARNING "%s is broken and did not set " + "SO_BROADCAST. It will break when 2.2 is " + "released.\n", current->comm); #else return -EACCES; #endif } - if (sk->zapped) { + if (sk->zapped) if (atalk_autobind(sk) < 0) return -EBUSY; - } - if (atrtr_get_dev(&addr->sat_addr) == NULL) + if (!atrtr_get_dev(&addr->sat_addr)) return -ENETUNREACH; sk->protinfo.af_at.dest_port = addr->sat_port; @@ -1433,7 +1320,6 @@ static int atalk_connect(struct socket *sock, struct sockaddr *uaddr, sock->state = SS_CONNECTED; sk->state = TCP_ESTABLISHED; - return 0; } @@ -1446,13 +1332,11 @@ static int atalk_getname(struct socket *sock, struct sockaddr *uaddr, int *uaddr_len, int peer) { struct sockaddr_at sat; - struct sock *sk; + struct sock *sk = sock->sk; - sk = sock->sk; - if (sk->zapped) { + if (sk->zapped) if (atalk_autobind(sk) < 0) return -ENOBUFS; - } *uaddr_len = sizeof(struct sockaddr_at); @@ -1471,7 +1355,6 @@ static int atalk_getname(struct socket *sock, struct sockaddr *uaddr, sat.sat_family = AF_APPLETALK; memcpy(uaddr, &sat, sizeof(sat)); - return 0; } @@ -1482,20 +1365,19 @@ static int atalk_getname(struct socket *sock, struct sockaddr *uaddr, * extracted. PPP should probably pass frames marked as for this layer. * [ie ARPHRD_ETHERTALK] */ -static int atalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt) +static int atalk_rcv(struct sk_buff *skb, struct net_device *dev, + struct packet_type *pt) { - struct sock *sock; struct ddpehdr *ddp = (void *) skb->h.raw; + struct sock *sock; struct atalk_iface *atif; struct sockaddr_at tosat; int origlen; struct ddpebits ddphv; /* Size check */ - if (skb->len < sizeof(*ddp)) { - kfree_skb(skb); - return 0; - } + if (skb->len < sizeof(*ddp)) + goto freeit; /* * Fix up the length field [Ok this is horrible but otherwise @@ -1506,15 +1388,10 @@ static int atalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ * happens to be safe BUT.. (Its safe as user space will not * run until we put it back) */ - *((__u16 *)&ddphv) = ntohs(*((__u16 *)ddp)); - /* - * Trim buffer in case of stray trailing data - */ - + /* Trim buffer in case of stray trailing data */ origlen = skb->len; - skb_trim(skb, min(skb->len, ddphv.deh_len)); /* @@ -1522,32 +1399,26 @@ static int atalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ * (Otherwise we'll detonate most spectacularly * in the middle of recvmsg()). */ - if (skb->len < sizeof(*ddp)) { - kfree_skb(skb); - return 0; - } + if (skb->len < sizeof(*ddp)) + goto freeit; /* * Any checksums. Note we don't do htons() on this == is assumed to be * valid for net byte orders all over the networking code... */ - if (ddp->deh_sum && atalk_checksum(ddp, ddphv.deh_len) != ddp->deh_sum) { + if (ddp->deh_sum && + atalk_checksum(ddp, ddphv.deh_len) != ddp->deh_sum) /* Not a valid AppleTalk frame - dustbin time */ - kfree_skb(skb); - return 0; - } + goto freeit; /* Check the packet is aimed at us */ - - if (ddp->deh_dnet == 0) /* Net 0 is 'this network' */ + if (!ddp->deh_dnet) /* Net 0 is 'this network' */ atif = atalk_find_anynet(ddp->deh_dnode, dev); else atif = atalk_find_interface(ddp->deh_dnet, ddp->deh_dnode); - /* - * Not ours, so we route the packet via the correct AppleTalk interface. - */ - if (atif == NULL) { + /* Not ours, so we route the packet via the correct AppleTalk iface */ + if (!atif) { struct atalk_route *rt; struct at_addr ta; @@ -1555,16 +1426,16 @@ static int atalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ * Don't route multicast, etc., packets, or packets * sent to "this network" */ - if (skb->pkt_type != PACKET_HOST || ddp->deh_dnet == 0) { - /* - * FIX ME: - * Can it ever happen that a packet is from a PPP iface and needs to be broadcast onto the default network? - */ + if (skb->pkt_type != PACKET_HOST || !ddp->deh_dnet) { + /* FIXME: + * Can it ever happen that a packet is from a PPP + * iface and needs to be broadcast onto the default + * network? */ if (dev->type == ARPHRD_PPP) - printk(KERN_DEBUG "AppleTalk: didn't forward broadcast packet received from PPP iface\n"); - - kfree_skb(skb); - return 0; + printk(KERN_DEBUG "AppleTalk: didn't forward " + "broadcast packet received " + "from PPP iface\n"); + goto freeit; } ta.s_net = ddp->deh_dnet; @@ -1572,10 +1443,8 @@ static int atalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ /* Route the packet */ rt = atrtr_find(&ta); - if (rt == NULL || ddphv.deh_hops == DDP_MAXHOPS) { - kfree_skb(skb); - return 0; - } + if (!rt || ddphv.deh_hops == DDP_MAXHOPS) + goto freeit; ddphv.deh_hops++; /* @@ -1606,40 +1475,33 @@ static int atalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ * * Note. ddp-> becomes invalid at the realloc. */ - if (skb_headroom(skb) < 22) - { - struct sk_buff *newskb; + if (skb_headroom(skb) < 22) { /* 22 bytes - 12 ether, 2 len, 3 802.2 5 snap */ - newskb = skb_realloc_headroom(skb, 32); + struct sk_buff *nskb = skb_realloc_headroom(skb, 32); kfree_skb(skb); - if (!newskb) - return 0; - skb = newskb; - } - else + if (!nskb) + goto out; + skb = nskb; + } else skb = skb_unshare(skb, GFP_ATOMIC); /* * If the buffer didn't vanish into the lack of * space bitbucket we can send it. */ - if (skb) { - if (aarp_send_ddp(rt->dev, skb, &ta, NULL) == -1) - kfree_skb(skb); - } - - return 0; + if (skb && aarp_send_ddp(rt->dev, skb, &ta, NULL) == -1) + goto freeit; + goto out; } #if defined(CONFIG_IPDDP) || defined(CONFIG_IPDDP_MODULE) - /* - * Check if IP-over-DDP - */ + /* Check if IP-over-DDP */ if (skb->data[12] == 22) { - struct net_device *dev; + struct net_device *dev = __dev_get_by_name("ipddp0"); + struct net_device_stats *stats; /* This needs to be able to handle ipddp"N" devices */ - if ((dev = __dev_get_by_name("ipddp0")) == NULL) + if (!dev) return -ENODEV; skb->protocol = htons(ETH_P_IP); @@ -1647,14 +1509,13 @@ static int atalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ skb->dev = dev; skb->h.raw = skb->data; - ((struct net_device_stats *)dev->priv)->rx_packets++; - ((struct net_device_stats *)dev->priv)->rx_bytes += skb->len + 13; + stats = dev->priv; + stats->rx_packets++; + stats->rx_bytes += skb->len + 13; netif_rx(skb); /* Send the SKB up to a higher place. */ - - return 0; + goto out; } #endif - /* * Which socket - atalk_search_socket() looks for a *full match* * of the <net,node,port> tuple. @@ -1664,23 +1525,17 @@ static int atalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ tosat.sat_port = ddp->deh_dport; sock = atalk_search_socket(&tosat, atif); - - if (sock == NULL) { - /* But not one of our sockets */ - kfree_skb(skb); - return 0; - } + if (!sock) /* But not one of our sockets */ + goto freeit; - /* - * Queue packet (standard) - */ - + /* Queue packet (standard) */ skb->sk = sock; if (sock_queue_rcv_skb(sock, skb) < 0) - kfree_skb(skb); - - return 0; + goto freeit; + goto out; +freeit: kfree_skb(skb); +out: return 0; } /* @@ -1688,21 +1543,18 @@ static int atalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ * Caller must provide enough headroom on the packet to pull the short * header and append a long one. */ -static int ltalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt) +static int ltalk_rcv(struct sk_buff *skb, struct net_device *dev, + struct packet_type *pt) { struct ddpehdr *ddp; struct at_addr *ap; - /* - * Expand any short form frames. - */ + /* Expand any short form frames */ if (skb->mac.raw[2] == 1) { - /* - * Find our address. - */ + /* Find our address */ ap = atalk_find_dev_addr(dev); - if (ap == NULL || skb->len < sizeof(struct ddpshdr)) { + if (!ap || skb->len < sizeof(struct ddpshdr)) { kfree_skb(skb); return 0; } @@ -1715,9 +1567,7 @@ static int ltalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ skb_push(skb, sizeof(*ddp) - 4); ddp = (struct ddpehdr *)skb->data; - /* - * Now fill in the long header. - */ + /* Now fill in the long header */ /* * These two first. The mac overlays the new source/dest @@ -1737,7 +1587,6 @@ static int ltalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ ddp->deh_len = skb->len; ddp->deh_hops = DDP_MAXHOPS; /* Non routable, so force a drop if we slip up later */ - /* Mend the byte order */ *((__u16 *)ddp) = htons(*((__u16 *)ddp)); } @@ -1746,19 +1595,20 @@ static int ltalk_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_ return atalk_rcv(skb, dev, pt); } -static int atalk_sendmsg(struct socket *sock, struct msghdr *msg, int len, struct scm_cookie *scm) +static int atalk_sendmsg(struct socket *sock, struct msghdr *msg, int len, + struct scm_cookie *scm) { struct sock *sk = sock->sk; struct sockaddr_at *usat = (struct sockaddr_at *)msg->msg_name; + int flags = msg->msg_flags; + int loopback = 0; struct sockaddr_at local_satalk, gsat; struct sk_buff *skb; struct net_device *dev; struct ddpehdr *ddp; int size; struct atalk_route *rt; - int loopback = 0; int err; - int flags = msg->msg_flags; if (flags & ~MSG_DONTWAIT) return -EINVAL; @@ -1767,19 +1617,18 @@ static int atalk_sendmsg(struct socket *sock, struct msghdr *msg, int len, struc return -EMSGSIZE; if (usat) { - if(sk->zapped) { + if (sk->zapped) if (atalk_autobind(sk) < 0) return -EBUSY; - } - if (msg->msg_namelen < sizeof(*usat)) - return -EINVAL; - if (usat->sat_family != AF_APPLETALK) + if (msg->msg_namelen < sizeof(*usat) || + usat->sat_family != AF_APPLETALK) return -EINVAL; /* netatalk doesn't implement this check */ if (usat->sat_addr.s_node == ATADDR_BCAST && !sk->broadcast) { - printk(KERN_INFO "SO_BROADCAST: Fix your netatalk as it will break before 2.2\n"); + printk(KERN_INFO "SO_BROADCAST: Fix your netatalk as " + "it will break before 2.2\n"); #if 0 return -EPERM; #endif @@ -1795,15 +1644,14 @@ static int atalk_sendmsg(struct socket *sock, struct msghdr *msg, int len, struc } /* Build a packet */ - SOCK_DEBUG(sk, "SK %p: Got address.\n", sk); /* For headers */ size = sizeof(struct ddpehdr) + len + ddp_dl->header_length; - if (usat->sat_addr.s_net != 0 || usat->sat_addr.s_node == ATADDR_ANYNODE) { + if (usat->sat_addr.s_net || usat->sat_addr.s_node == ATADDR_ANYNODE) { rt = atrtr_find(&usat->sat_addr); - if (rt == NULL) + if (!rt) return -ENETUNREACH; dev = rt->dev; @@ -1814,24 +1662,23 @@ static int atalk_sendmsg(struct socket *sock, struct msghdr *msg, int len, struc at_hint.s_net = sk->protinfo.af_at.src_net; rt = atrtr_find(&at_hint); - if (rt == NULL) + if (!rt) return -ENETUNREACH; dev = rt->dev; } - SOCK_DEBUG(sk, "SK %p: Size needed %d, device %s\n", sk, size, dev->name); + SOCK_DEBUG(sk, "SK %p: Size needed %d, device %s\n", + sk, size, dev->name); size += dev->hard_header_len; - skb = sock_alloc_send_skb(sk, size, 0, (flags & MSG_DONTWAIT), &err); - if (skb == NULL) + if (!skb) return err; skb->sk = sk; skb_reserve(skb, ddp_dl->header_length); skb_reserve(skb, dev->hard_header_len); - skb->dev = dev; SOCK_DEBUG(sk, "SK %p: Begin build.\n", sk); @@ -1871,28 +1718,30 @@ static int atalk_sendmsg(struct socket *sock, struct msghdr *msg, int len, struc * Loopback broadcast packets to non gateway targets (ie routes * to group we are in) */ - if (ddp->deh_dnode == ATADDR_BCAST) { - if ((!(rt->flags&RTF_GATEWAY)) && (!(dev->flags&IFF_LOOPBACK))) { - struct sk_buff *skb2 = skb_copy(skb, GFP_KERNEL); - if (skb2) { - loopback = 1; - SOCK_DEBUG(sk, "SK %p: send out(copy).\n", sk); - if (aarp_send_ddp(dev, skb2, &usat->sat_addr, NULL) == -1) - kfree_skb(skb2); + if (ddp->deh_dnode == ATADDR_BCAST && + !(rt->flags & RTF_GATEWAY) && !(dev->flags & IFF_LOOPBACK)) { + struct sk_buff *skb2 = skb_copy(skb, GFP_KERNEL); + + if (skb2) { + loopback = 1; + SOCK_DEBUG(sk, "SK %p: send out(copy).\n", sk); + if (aarp_send_ddp(dev, skb2, + &usat->sat_addr, NULL) == -1) + kfree_skb(skb2); /* else queued/sent above in the aarp queue */ - } } } - if ((dev->flags & IFF_LOOPBACK) || loopback) { + if (dev->flags & IFF_LOOPBACK || loopback) { SOCK_DEBUG(sk, "SK %p: Loop back.\n", sk); /* loop back */ skb_orphan(skb); ddp_dl->datalink_header(ddp_dl, skb, dev->dev_addr); skb->mac.raw = skb->data; - skb->h.raw = skb->data + ddp_dl->header_length + dev->hard_header_len; - skb_pull(skb,dev->hard_header_len); - skb_pull(skb,ddp_dl->header_length); + skb->h.raw = skb->data + ddp_dl->header_length + + dev->hard_header_len; + skb_pull(skb, dev->hard_header_len); + skb_pull(skb, ddp_dl->header_length); atalk_rcv(skb, dev, NULL); } else { SOCK_DEBUG(sk, "SK %p: send out.\n", sk); @@ -1915,15 +1764,15 @@ static int atalk_recvmsg(struct socket *sock, struct msghdr *msg, int size, { struct sock *sk = sock->sk; struct sockaddr_at *sat = (struct sockaddr_at *)msg->msg_name; - struct ddpehdr *ddp = NULL; - struct ddpebits ddphv; + struct ddpehdr *ddp = NULL; int copied = 0; - struct sk_buff *skb; int err = 0; + struct ddpebits ddphv; + struct sk_buff *skb; - skb = skb_recv_datagram(sk, (flags & ~MSG_DONTWAIT), - (flags & MSG_DONTWAIT), &err); - if (skb == NULL) + skb = skb_recv_datagram(sk, flags & ~MSG_DONTWAIT, + flags & MSG_DONTWAIT, &err); + if (!skb) return err; ddp = (struct ddpehdr *)(skb->h.raw); @@ -1943,7 +1792,8 @@ static int atalk_recvmsg(struct socket *sock, struct msghdr *msg, int size, copied = size; msg->msg_flags |= MSG_TRUNC; } - err = skb_copy_datagram_iovec(skb, sizeof(*ddp), msg->msg_iov, copied); + err = skb_copy_datagram_iovec(skb, sizeof(*ddp), + msg->msg_iov, copied); } if (!err) { @@ -1957,7 +1807,6 @@ static int atalk_recvmsg(struct socket *sock, struct msghdr *msg, int size, } skb_free_datagram(sk, skb); /* Free the datagram. */ - return err ? err : copied; } @@ -1967,50 +1816,40 @@ static int atalk_recvmsg(struct socket *sock, struct msghdr *msg, int size, */ static int atalk_ioctl(struct socket *sock,unsigned int cmd, unsigned long arg) { - long amount=0; - struct sock *sk=sock->sk; + long amount = 0; + struct sock *sk = sock->sk; - switch(cmd) - { - /* - * Protocol layer - */ + switch (cmd) { + /* Protocol layer */ case TIOCOUTQ: amount = sk->sndbuf - atomic_read(&sk->wmem_alloc); - if(amount < 0) + if (amount < 0) amount = 0; break; - case TIOCINQ: { - struct sk_buff *skb; - /* These two are safe on a single CPU system as only user tasks fiddle here */ - if((skb = skb_peek(&sk->receive_queue)) != NULL) + /* These two are safe on a single CPU system as only + * user tasks fiddle here */ + struct sk_buff *skb = skb_peek(&sk->receive_queue); + + if (skb) amount = skb->len-sizeof(struct ddpehdr); break; } - case SIOCGSTAMP: - if(sk) - { - if(sk->stamp.tv_sec == 0) - return -ENOENT; - return (copy_to_user((void *)arg,&sk->stamp,sizeof(struct timeval)) ? -EFAULT : 0); - } - return (-EINVAL); - - /* - * Routing - */ + if (!sk) + return -EINVAL; + if (!sk->stamp.tv_sec) + return -ENOENT; + return copy_to_user((void *)arg, &sk->stamp, + sizeof(struct timeval)) ? -EFAULT : 0; + /* Routing */ case SIOCADDRT: case SIOCDELRT: - if(!capable(CAP_NET_ADMIN)) + if (!capable(CAP_NET_ADMIN)) return -EPERM; - return (atrtr_ioctl(cmd,(void *)arg)); - - /* - * Interface - */ + return atrtr_ioctl(cmd, (void *)arg); + /* Interface */ case SIOCGIFADDR: case SIOCSIFADDR: case SIOCGIFBRDADDR: @@ -2018,11 +1857,8 @@ static int atalk_ioctl(struct socket *sock,unsigned int cmd, unsigned long arg) case SIOCDIFADDR: case SIOCSARP: /* proxy AARP */ case SIOCDARP: /* proxy AARP */ - return (atif_ioctl(cmd,(void *)arg)); - - /* - * Physical layer ioctl calls - */ + return atif_ioctl(cmd, (void *)arg); + /* Physical layer ioctl calls */ case SIOCSIFLINK: case SIOCGIFHWADDR: case SIOCSIFHWADDR: @@ -2035,8 +1871,7 @@ static int atalk_ioctl(struct socket *sock,unsigned int cmd, unsigned long arg) case SIOCGIFCOUNT: case SIOCGIFINDEX: case SIOCGIFNAME: - return ((dev_ioctl(cmd,(void *) arg))); - + return dev_ioctl(cmd,(void *) arg); case SIOCSIFMETRIC: case SIOCSIFBRDADDR: case SIOCGIFNETMASK: @@ -2045,16 +1880,15 @@ static int atalk_ioctl(struct socket *sock,unsigned int cmd, unsigned long arg) case SIOCSIFMEM: case SIOCGIFDSTADDR: case SIOCSIFDSTADDR: - return (-EINVAL); - + return -EINVAL; default: - return (-EINVAL); + return -EINVAL; } - return (put_user(amount, (int *)arg)); + return put_user(amount, (int *)arg); } -static struct net_proto_family atalk_family_ops= +static struct net_proto_family atalk_family_ops = { PF_APPLETALK, atalk_create @@ -2111,19 +1945,17 @@ struct packet_type ppptalk_packet_type= static char ddp_snap_id[] = {0x08, 0x00, 0x07, 0x80, 0x9B}; -/* - * Export symbols for use by drivers when AppleTalk is a module. - */ +/* Export symbols for use by drivers when AppleTalk is a module */ EXPORT_SYMBOL(aarp_send_ddp); EXPORT_SYMBOL(atrtr_get_dev); EXPORT_SYMBOL(atalk_find_dev_addr); /* Called by proto.c on kernel start up */ - static int __init atalk_init(void) { (void) sock_register(&atalk_family_ops); - if((ddp_dl = register_snap_client(ddp_snap_id, atalk_rcv)) == NULL) + ddp_dl = register_snap_client(ddp_snap_id, atalk_rcv); + if (!ddp_dl) printk(KERN_CRIT "Unable to register DDP with SNAP.\n"); ltalk_packet_type.type = htons(ETH_P_LOCALTALK); @@ -2135,19 +1967,16 @@ static int __init atalk_init(void) register_netdevice_notifier(&ddp_notifier); aarp_proto_init(); -#ifdef CONFIG_PROC_FS proc_net_create("appletalk", 0, atalk_get_info); proc_net_create("atalk_route", 0, atalk_rt_get_info); proc_net_create("atalk_iface", 0, atalk_if_get_info); - +#ifdef CONFIG_PROC_FS aarp_register_proc_fs(); #endif /* CONFIG_PROC_FS */ - #ifdef CONFIG_SYSCTL atalk_register_sysctl(); #endif /* CONFIG_SYSCTL */ - - printk(KERN_INFO "NET4: AppleTalk 0.18 for Linux NET4.0\n"); + printk(KERN_INFO "NET4: AppleTalk 0.18a for Linux NET4.0\n"); return 0; } module_init(atalk_init); @@ -2165,32 +1994,24 @@ module_init(atalk_init); * Ergo, before the AppleTalk module can be removed, all AppleTalk * sockets be closed from user space. */ - static void __exit atalk_exit(void) { #ifdef CONFIG_SYSCTL atalk_unregister_sysctl(); #endif /* CONFIG_SYSCTL */ - -#ifdef CONFIG_PROC_FS proc_net_remove("appletalk"); proc_net_remove("atalk_route"); proc_net_remove("atalk_iface"); - +#ifdef CONFIG_PROC_FS aarp_unregister_proc_fs(); #endif /* CONFIG_PROC_FS */ - aarp_cleanup_module(); /* General aarp clean-up. */ - unregister_netdevice_notifier(&ddp_notifier); dev_remove_pack(<alk_packet_type); dev_remove_pack(&ppptalk_packet_type); unregister_snap_client(ddp_snap_id); sock_unregister(PF_APPLETALK); - - return; } module_exit(atalk_exit); #endif /* MODULE */ - #endif /* CONFIG_ATALK || CONFIG_ATALK_MODULE */ diff --git a/net/atm/lec.c b/net/atm/lec.c index 51be9f077..ec0dc2671 100644 --- a/net/atm/lec.c +++ b/net/atm/lec.c @@ -51,6 +51,11 @@ static unsigned char bridge_ula_lec[] = {0x01, 0x80, 0xc2, 0x00, 0x00}; #define DPRINTK(format,args...) #endif +struct net_bridge_fdb_entry *(*br_fdb_get_hook)(struct net_bridge *br, + unsigned char *addr); +void (*br_fdb_put_hook)(struct net_bridge_fdb_entry *ent); + + #define DUMP_PACKETS 0 /* 0 = None, * 1 = 30 first bytes * 2 = Whole packet @@ -853,8 +858,11 @@ static void __exit lane_module_cleanup(void) if (dev_lec[i] != NULL) { priv = (struct lec_priv *)dev_lec[i]->priv; #if defined(CONFIG_TR) - unregister_trdev(dev_lec[i]); + if (priv->is_trdev) + unregister_trdev(dev_lec[i]); + else #endif + unregister_netdev(dev_lec[i]); kfree(dev_lec[i]); dev_lec[i] = NULL; } diff --git a/net/atm/lec.h b/net/atm/lec.h index f40a37fa3..5dd0494e9 100644 --- a/net/atm/lec.h +++ b/net/atm/lec.h @@ -16,9 +16,9 @@ #if defined (CONFIG_BRIDGE) || defined(CONFIG_BRIDGE_MODULE) #include <linux/if_bridge.h> -struct net_bridge_fdb_entry *(*br_fdb_get_hook)(struct net_bridge *br, +extern struct net_bridge_fdb_entry *(*br_fdb_get_hook)(struct net_bridge *br, unsigned char *addr); -void (*br_fdb_put_hook)(struct net_bridge_fdb_entry *ent); +extern void (*br_fdb_put_hook)(struct net_bridge_fdb_entry *ent); #endif /* defined(CONFIG_BRIDGE) || defined(CONFIG_BRIDGE_MODULE) */ #define LEC_HEADER_LEN 16 diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h index 940d69be0..8fcfe3fe4 100644 --- a/net/bridge/br_private.h +++ b/net/bridge/br_private.h @@ -4,7 +4,7 @@ * Authors: * Lennert Buytenhek <buytenh@gnu.org> * - * $Id: br_private.h,v 1.3 2000/05/05 02:17:17 davem Exp $ + * $Id: br_private.h,v 1.4 2001/01/19 04:51:48 davem Exp $ * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public License @@ -112,8 +112,8 @@ struct net_bridge int gc_interval; }; -struct notifier_block br_device_notifier; -unsigned char bridge_ula[6]; +extern struct notifier_block br_device_notifier; +extern unsigned char bridge_ula[6]; /* br.c */ void br_dec_use_count(void); diff --git a/net/core/netfilter.c b/net/core/netfilter.c index f4bb62818..b12e1d5ec 100644 --- a/net/core/netfilter.c +++ b/net/core/netfilter.c @@ -442,7 +442,6 @@ static void nf_queue(struct sk_buff *skb, } } -/* We have BR_NETPROTO_LOCK here */ int nf_hook_slow(int pf, unsigned int hook, struct sk_buff *skb, struct net_device *indev, struct net_device *outdev, @@ -452,6 +451,9 @@ int nf_hook_slow(int pf, unsigned int hook, struct sk_buff *skb, unsigned int verdict; int ret = 0; + /* We may already have this, but read-locks nest anyway */ + br_read_lock_bh(BR_NETPROTO_LOCK); + #ifdef CONFIG_NETFILTER_DEBUG if (skb->nf_debug & (1 << hook)) { printk("nf_hook: hook %i already set.\n", hook); @@ -479,6 +481,7 @@ int nf_hook_slow(int pf, unsigned int hook, struct sk_buff *skb, break; } + br_read_unlock_bh(BR_NETPROTO_LOCK); return ret; } diff --git a/net/decnet/Makefile b/net/decnet/Makefile index ac7a1a462..11f7c8b08 100644 --- a/net/decnet/Makefile +++ b/net/decnet/Makefile @@ -11,5 +11,3 @@ obj-y += sysctl_net_decnet.o include $(TOPDIR)/Rules.make -tar: - tar -cvf /dev/f1 . diff --git a/net/decnet/TODO b/net/decnet/TODO index 1607d6d1b..c8ea8178d 100644 --- a/net/decnet/TODO +++ b/net/decnet/TODO @@ -43,15 +43,15 @@ Steve's quick list of things that need finishing off: o Hello messages should be generated for each primary address on each interface. - o Add more information into /proc/net/decnet and finalise the format to - allow DECnet support in netstat. - - o Make sure that returned connect messages are generated when they should - be, and that the correct error messages are sent too. - o Add the routing message grabbing netfilter module [written, tested, awaiting merge] o Add perfect socket hashing - an idea suggested by Paul Koning [part written, awaiting debugging and merge] + o Add session control message flow control + + o Add NSP message flow control + + o DECnet sendpages() function + diff --git a/net/decnet/af_decnet.c b/net/decnet/af_decnet.c index bc51b636d..b930878c2 100644 --- a/net/decnet/af_decnet.c +++ b/net/decnet/af_decnet.c @@ -35,6 +35,7 @@ * Arnaldo C. Melo: use capable, not suser * Steve Whitehouse: Removed unused code. Fix to use sk->allocation * when required. + * Patrick Caulfield: /proc/net/decnet now has object name/number */ @@ -128,8 +129,6 @@ Version 0.0.6 2.1.110 07-aug-98 Eduardo Marcelo Serrat #include <net/dn_fib.h> #include <net/dn_neigh.h> -#define MAX(a,b) ((a)>(b)?(a):(b)) - static void dn_keepalive(struct sock *sk); /* @@ -141,15 +140,15 @@ unsigned char decnet_ether_address[ETH_ALEN] = { 0xAA, 0x00, 0x04, 0x00, 0x00, 0 static struct proto_ops dn_proto_ops; rwlock_t dn_hash_lock = RW_LOCK_UNLOCKED; -static struct sock *dn_sklist = NULL; -static struct sock *dn_wild_sk = NULL; +static struct sock *dn_sklist; +static struct sock *dn_wild_sk; static int __dn_setsockopt(struct socket *sock, int level, int optname, char *optval, int optlen, int flags); static int __dn_getsockopt(struct socket *sock, int level, int optname, char *optval, int *optlen, int flags); static struct sock **dn_find_list(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); if (scp->addr.sdn_flags & SDF_WILD) return dn_wild_sk ? NULL : &dn_wild_sk; @@ -159,7 +158,7 @@ static struct sock **dn_find_list(struct sock *sk) static unsigned short port_alloc(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); static unsigned short port = 0x2000; if (port == 0) @@ -177,12 +176,17 @@ static unsigned short port = 0x2000; */ static int dn_hash_sock(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); struct sock **skp; int rv = -EUSERS; - write_lock_bh(&dn_hash_lock); + if (sk->next) + BUG(); + if (sk->pprev) + BUG(); + write_lock_bh(&dn_hash_lock); + if (!scp->addrloc && !port_alloc(sk)) goto out; @@ -327,7 +331,7 @@ struct sock *dn_sklist_find_listener(struct sockaddr_dn *addr) read_lock(&dn_hash_lock); for(sk = dn_sklist; sk != NULL; sk = sk->next) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); if (sk->state != TCP_LISTEN) continue; if (scp->addr.sdn_objnum) { @@ -355,13 +359,13 @@ struct sock *dn_sklist_find_listener(struct sockaddr_dn *addr) struct sock *dn_find_by_skb(struct sk_buff *skb) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); struct sock *sk; struct dn_scp *scp; read_lock(&dn_hash_lock); for(sk = dn_sklist; sk != NULL; sk = sk->next) { - scp = &sk->protinfo.dn; + scp = DN_SK(sk); if (cb->src != dn_saddr2dn(&scp->peer)) continue; if (cb->dst_port != scp->addrloc) @@ -383,7 +387,7 @@ struct sock *dn_find_by_skb(struct sk_buff *skb) static void dn_destruct(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); skb_queue_purge(&scp->data_xmit_queue); skb_queue_purge(&scp->other_xmit_queue); @@ -394,25 +398,26 @@ static void dn_destruct(struct sock *sk) MOD_DEC_USE_COUNT; } -struct sock *dn_alloc_sock(struct socket *sock, int flags) +struct sock *dn_alloc_sock(struct socket *sock, int gfp) { struct sock *sk; struct dn_scp *scp; - if ((sk = sk_alloc(PF_DECnet, flags, 1)) == NULL) + if ((sk = sk_alloc(PF_DECnet, gfp, 1)) == NULL) goto no_sock; if (sock) { sock->ops = &dn_proto_ops; } sock_init_data(sock,sk); - scp = &sk->protinfo.dn; + scp = DN_SK(sk); sk->backlog_rcv = dn_nsp_backlog_rcv; sk->destruct = dn_destruct; sk->no_check = 1; sk->family = PF_DECnet; sk->protocol = 0; + sk->allocation = gfp; /* Initialization of DECnet Session Control Port */ scp->state = DN_O; /* Open */ @@ -424,13 +429,25 @@ struct sock *dn_alloc_sock(struct socket *sock, int flags) scp->ackrcv_oth = 0; /* Last oth data ack rec*/ scp->flowrem_sw = DN_SEND; scp->flowloc_sw = DN_SEND; + scp->flowrem_dat = 0; + scp->flowrem_oth = 1; + scp->flowloc_dat = 0; + scp->flowloc_oth = 1; + scp->services_rem = 0; + scp->services_loc = 1 | NSP_FC_NONE; + scp->info_rem = 0; + scp->info_loc = 0x03; /* NSP version 4.1 */ + scp->segsize_rem = 230; /* Default: Updated by remote segsize */ + scp->segsize_loc = 1450; /* Best guess for ethernet */ + scp->nonagle = 0; + scp->multi_ireq = 1; scp->accept_mode = ACC_IMMED; scp->addr.sdn_family = AF_DECnet; scp->peer.sdn_family = AF_DECnet; scp->accessdata.acc_accl = 5; memcpy(scp->accessdata.acc_acc, "LINUX", 5); - scp->mss = 1460; + scp->max_window = NSP_MAX_WINDOW; scp->snd_window = NSP_MIN_WINDOW; scp->nsp_srtt = NSP_INITIAL_SRTT; scp->nsp_rttvar = NSP_INITIAL_RTTVAR; @@ -464,7 +481,7 @@ no_sock: */ static void dn_keepalive(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); /* * By checking the other_data transmit queue is empty @@ -472,7 +489,7 @@ static void dn_keepalive(struct sock *sk) * many of these keepalive frames. */ if (skb_queue_len(&scp->other_xmit_queue) == 0) - dn_nsp_send_lnk(sk, DN_NOCHANGE); + dn_nsp_send_link(sk, DN_NOCHANGE, 0); } @@ -485,7 +502,7 @@ static void dn_keepalive(struct sock *sk) */ int dn_destroy_timer(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); scp->persist = dn_nsp_persist(sk); @@ -527,7 +544,7 @@ int dn_destroy_timer(struct sock *sk) static void dn_destroy_sock(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); scp->nsp_rxtshift = 0; /* reset back off */ @@ -674,7 +691,7 @@ dn_release(struct socket *sock) static int dn_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) { struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); struct sockaddr_dn *saddr = (struct sockaddr_dn *)uaddr; struct net_device *dev; int rv; @@ -722,11 +739,8 @@ static int dn_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) memcpy(&scp->addr, saddr, addr_len); sk->zapped = 0; - if ((rv = dn_hash_sock(sk)) == 0) - goto out; - - sk->zapped = 1; -out: + if ((rv = dn_hash_sock(sk)) != 0) + sk->zapped = 1; return rv; } @@ -735,7 +749,7 @@ out: static int dn_auto_bind(struct socket *sock) { struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); sk->zapped = 0; @@ -769,7 +783,7 @@ static int dn_connect(struct socket *sock, struct sockaddr *uaddr, int addr_len, { struct sockaddr_dn *addr = (struct sockaddr_dn *)uaddr; struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); int err = -EISCONN; lock_sock(sk); @@ -788,7 +802,7 @@ static int dn_connect(struct socket *sock, struct sockaddr *uaddr, int addr_len, } err = -EINVAL; - if (sk->protinfo.dn.state != DN_O) + if (DN_SK(sk)->state != DN_O) goto out; if (addr_len != sizeof(struct sockaddr_dn)) @@ -812,7 +826,7 @@ static int dn_connect(struct socket *sock, struct sockaddr *uaddr, int addr_len, sk->state = TCP_SYN_SENT; sock->state = SS_CONNECTING; - sk->protinfo.dn.state = DN_CI; + DN_SK(sk)->state = DN_CI; dn_nsp_send_conninit(sk, NSP_CI); @@ -853,7 +867,7 @@ out: return err; } -static int dn_access_copy(struct sk_buff *skb, struct accessdata_dn *acc) +static void dn_access_copy(struct sk_buff *skb, struct accessdata_dn *acc) { unsigned char *ptr = skb->data; @@ -870,10 +884,9 @@ static int dn_access_copy(struct sk_buff *skb, struct accessdata_dn *acc) skb_pull(skb, acc->acc_accl + acc->acc_passl + acc->acc_userl + 3); - return 0; } -static int dn_user_copy(struct sk_buff *skb, struct optdata_dn *opt) +static void dn_user_copy(struct sk_buff *skb, struct optdata_dn *opt) { unsigned char *ptr = skb->data; @@ -882,7 +895,6 @@ static int dn_user_copy(struct sk_buff *skb, struct optdata_dn *opt) memcpy(opt->opt_data, ptr, opt->opt_optl); skb_pull(skb, opt->opt_optl + 1); - return 0; } @@ -910,7 +922,7 @@ static int dn_wait_accept(struct socket *sock, int flags) return -ERESTARTSYS; /* But of course you don't! */ } - if ((sk->protinfo.dn.state != DN_RUN) && (sk->protinfo.dn.state != DN_DRC)) { + if ((DN_SK(sk)->state != DN_RUN) && (DN_SK(sk)->state != DN_DRC)) { sock->state = SS_UNCONNECTED; return sock_error(sk); } @@ -937,7 +949,7 @@ static int dn_accept(struct socket *sock, struct socket *newsock, int flags) return -EINVAL; } - if (sk->protinfo.dn.state != DN_O) { + if (DN_SK(sk)->state != DN_O) { release_sock(sk); return -EINVAL; } @@ -967,7 +979,7 @@ static int dn_accept(struct socket *sock, struct socket *newsock, int flags) } } while (skb == NULL); - cb = (struct dn_skb_cb *)skb->cb; + cb = DN_SKB_CB(skb); if ((newsk = dn_alloc_sock(newsock, sk->allocation)) == NULL) { release_sock(sk); @@ -980,50 +992,52 @@ static int dn_accept(struct socket *sock, struct socket *newsock, int flags) dst_release(xchg(&newsk->dst_cache, skb->dst)); skb->dst = NULL; - newsk->protinfo.dn.state = DN_CR; - newsk->protinfo.dn.addrrem = cb->src_port; - newsk->protinfo.dn.mss = cb->segsize; - newsk->protinfo.dn.accept_mode = sk->protinfo.dn.accept_mode; + DN_SK(newsk)->state = DN_CR; + DN_SK(newsk)->addrrem = cb->src_port; + DN_SK(newsk)->services_rem = cb->services; + DN_SK(newsk)->info_rem = cb->info; + DN_SK(newsk)->segsize_rem = cb->segsize; + DN_SK(newsk)->accept_mode = DN_SK(sk)->accept_mode; - if (newsk->protinfo.dn.mss < 230) - newsk->protinfo.dn.mss = 230; + if (DN_SK(newsk)->segsize_rem < 230) + DN_SK(newsk)->segsize_rem = 230; newsk->state = TCP_LISTEN; newsk->zapped = 0; - memcpy(&newsk->protinfo.dn.addr, &sk->protinfo.dn.addr, sizeof(struct sockaddr_dn)); + memcpy(&(DN_SK(newsk)->addr), &(DN_SK(sk)->addr), sizeof(struct sockaddr_dn)); /* * If we are listening on a wild socket, we don't want * the newly created socket on the wrong hash queue. */ - newsk->protinfo.dn.addr.sdn_flags &= ~SDF_WILD; + DN_SK(newsk)->addr.sdn_flags &= ~SDF_WILD; - skb_pull(skb, dn_username2sockaddr(skb->data, skb->len, &newsk->protinfo.dn.addr, &type)); - skb_pull(skb, dn_username2sockaddr(skb->data, skb->len, &newsk->protinfo.dn.peer, &type)); - *(dn_address *)newsk->protinfo.dn.peer.sdn_add.a_addr = cb->src; - *(dn_address *)newsk->protinfo.dn.addr.sdn_add.a_addr = cb->dst; + skb_pull(skb, dn_username2sockaddr(skb->data, skb->len, &(DN_SK(newsk)->addr), &type)); + skb_pull(skb, dn_username2sockaddr(skb->data, skb->len, &(DN_SK(newsk)->peer), &type)); + *(dn_address *)(DN_SK(newsk)->peer.sdn_add.a_addr) = cb->src; + *(dn_address *)(DN_SK(newsk)->addr.sdn_add.a_addr) = cb->dst; menuver = *skb->data; skb_pull(skb, 1); if (menuver & DN_MENUVER_ACC) - dn_access_copy(skb, &newsk->protinfo.dn.accessdata); + dn_access_copy(skb, &(DN_SK(newsk)->accessdata)); if (menuver & DN_MENUVER_USR) - dn_user_copy(skb, &newsk->protinfo.dn.conndata_in); + dn_user_copy(skb, &(DN_SK(newsk)->conndata_in)); if (menuver & DN_MENUVER_PRX) - newsk->protinfo.dn.peer.sdn_flags |= SDF_PROXY; + DN_SK(newsk)->peer.sdn_flags |= SDF_PROXY; if (menuver & DN_MENUVER_UIC) - newsk->protinfo.dn.peer.sdn_flags |= SDF_UICPROXY; + DN_SK(newsk)->peer.sdn_flags |= SDF_UICPROXY; kfree_skb(skb); - memcpy(&newsk->protinfo.dn.conndata_out, &sk->protinfo.dn.conndata_out, + memcpy(&(DN_SK(newsk)->conndata_out), &(DN_SK(sk)->conndata_out), sizeof(struct optdata_dn)); - memcpy(&newsk->protinfo.dn.discdata_out, &sk->protinfo.dn.discdata_out, + memcpy(&(DN_SK(newsk)->discdata_out), &(DN_SK(sk)->discdata_out), sizeof(struct optdata_dn)); lock_sock(newsk); @@ -1031,9 +1045,13 @@ static int dn_accept(struct socket *sock, struct socket *newsock, int flags) dn_send_conn_ack(newsk); - if (newsk->protinfo.dn.accept_mode == ACC_IMMED) { - newsk->protinfo.dn.state = DN_CC; - dn_send_conn_conf(newsk, newsk->allocation); + /* + * Here we use sk->allocation since although the conn conf is + * for the newsk, the context is the old socket. + */ + if (DN_SK(newsk)->accept_mode == ACC_IMMED) { + DN_SK(newsk)->state = DN_CC; + dn_send_conn_conf(newsk, sk->allocation); err = dn_wait_accept(newsock, flags); } @@ -1046,7 +1064,7 @@ static int dn_getname(struct socket *sock, struct sockaddr *uaddr,int *uaddr_len { struct sockaddr_dn *sa = (struct sockaddr_dn *)uaddr; struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); *uaddr_len = sizeof(struct sockaddr_dn); @@ -1070,7 +1088,7 @@ static int dn_getname(struct socket *sock, struct sockaddr *uaddr,int *uaddr_len static unsigned int dn_poll(struct file *file, struct socket *sock, poll_table *wait) { struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); int mask = datagram_poll(file, sock, wait); if (skb_queue_len(&scp->other_receive_queue)) @@ -1082,100 +1100,32 @@ static unsigned int dn_poll(struct file *file, struct socket *sock, poll_table static int dn_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg) { struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); int err = -EOPNOTSUPP; unsigned long amount = 0; struct sk_buff *skb; + int val; -#if 0 - struct dn_naddr dnaddr; -#endif switch(cmd) { case SIOCGIFADDR: case SIOCSIFADDR: return dn_dev_ioctl(cmd, (void *)arg); + case SIOCATMARK: + lock_sock(sk); + val = (skb_queue_len(&scp->other_receive_queue) != 0); + if (scp->state != DN_RUN) + val = -ENOTCONN; + release_sock(sk); + return val; + #ifdef CONFIG_DECNET_ROUTER case SIOCADDRT: case SIOCDELRT: return dn_fib_ioctl(sock, cmd, arg); #endif /* CONFIG_DECNET_ROUTER */ -#if 0 - case SIOCSIFADDR: - if (!capable(CAP_NET_ADMIN)) return -EPERM; - - if ((err = copy_from_user(devname, ioarg->devname, 5)) != 0) - break; - if ((err = copy_from_user(addr, ioarg->exec_addr, 6)) != 0) - break; - if ((dev = dev_get(devname)) == NULL) { - err = -ENODEV; - break; - } - if (dev->dn_ptr == NULL) { - err = -ENODEV; - break; - } - - dn_dev_devices_off(); - - decnet_default_device = dev; - memcpy(decnet_ether_address, addr, ETH_ALEN); - decnet_address = dn_htons(dn_eth2dn(decnet_ether_address)); - - dn_dev_devices_on(); - - break; - - case SIOCGIFADDR: - if (decnet_default_device) - strcpy(devname, decnet_default_device->name); - else - memset(devname, 0, 6); - - if ((err = copy_to_user(ioarg->devname, devname, 5)) != 0) - break; - - if ((err = copy_to_user(ioarg->exec_addr, decnet_ether_address, 6)) != 0) - break; - - break; -#endif - -#if 0 - case SIOCSNETADDR: - if (!capable(CAP_NET_ADMIN)) { - err = -EPERM; - break; - } - - if ((err = copy_from_user(&dnaddr, (void *)arg, sizeof(struct dn_naddr))) != 0) - break; - - if (dnaddr.a_len != ETH_ALEN) { - err = -EINVAL; - break; - } - - dn_dev_devices_off(); - - memcpy(decnet_ether_address, dnaddr.a_addr, ETH_ALEN); - decnet_address = dn_htons(dn_eth2dn(decnet_ether_address)); - - dn_dev_devices_on(); - break; - - case SIOCGNETADDR: - dnaddr.a_len = ETH_ALEN; - memcpy(dnaddr.a_addr, decnet_ether_address, ETH_ALEN); - - if ((err = copy_to_user((void *)arg, &dnaddr, sizeof(struct dn_naddr))) != 0) - break; - - break; -#endif case OSIOCSNETADDR: if (!capable(CAP_NET_ADMIN)) { err = -EPERM; @@ -1237,7 +1187,7 @@ static int dn_listen(struct socket *sock, int backlog) if (sk->zapped) goto out; - if ((sk->protinfo.dn.state != DN_O) || (sk->state == TCP_LISTEN)) + if ((DN_SK(sk)->state != DN_O) || (sk->state == TCP_LISTEN)) goto out; sk->max_ack_backlog = backlog; @@ -1255,7 +1205,7 @@ out: static int dn_shutdown(struct socket *sock, int how) { struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); int err = -ENOTCONN; lock_sock(sk); @@ -1300,14 +1250,27 @@ static int dn_setsockopt(struct socket *sock, int level, int optname, char *optv static int __dn_setsockopt(struct socket *sock, int level,int optname, char *optval, int optlen, int flags) { struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; - struct optdata_dn opt; - struct accessdata_dn acc; + struct dn_scp *scp = DN_SK(sk); + union { + struct optdata_dn opt; + struct accessdata_dn acc; + int mode; + unsigned long win; + int val; + unsigned char services; + unsigned char info; + } u; int err; if (optlen && !optval) return -EINVAL; + if (optlen > sizeof(u)) + return -EINVAL; + + if (copy_from_user(&u, optval, optlen)) + return -EFAULT; + switch(optname) { case DSO_CONDATA: if (sock->state == SS_CONNECTED) @@ -1318,29 +1281,23 @@ static int __dn_setsockopt(struct socket *sock, int level,int optname, char *opt if (optlen != sizeof(struct optdata_dn)) return -EINVAL; - if (copy_from_user(&opt, optval, optlen)) - return -EFAULT; - - if (opt.opt_optl > 16) + if (u.opt.opt_optl > 16) return -EINVAL; - memcpy(&scp->conndata_out, &opt, sizeof(struct optdata_dn)); + memcpy(&scp->conndata_out, &u.opt, optlen); break; case DSO_DISDATA: - if (sock->state != SS_CONNECTED && sk->protinfo.dn.accept_mode == ACC_IMMED) + if (sock->state != SS_CONNECTED && scp->accept_mode == ACC_IMMED) return -ENOTCONN; if (optlen != sizeof(struct optdata_dn)) return -EINVAL; - if (copy_from_user(&opt, optval, sizeof(struct optdata_dn))) - return -EFAULT; - - if (opt.opt_optl > 16) + if (u.opt.opt_optl > 16) return -EINVAL; - memcpy(&scp->discdata_out, &opt, sizeof(struct optdata_dn)); + memcpy(&scp->discdata_out, &u.opt, optlen); break; case DSO_CONACCESS: @@ -1352,15 +1309,12 @@ static int __dn_setsockopt(struct socket *sock, int level,int optname, char *opt if (optlen != sizeof(struct accessdata_dn)) return -EINVAL; - if (copy_from_user(&acc, optval, sizeof(struct accessdata_dn))) - return -EFAULT; - - if ((acc.acc_accl > DN_MAXACCL) || - (acc.acc_passl > DN_MAXACCL) || - (acc.acc_userl > DN_MAXACCL)) + if ((u.acc.acc_accl > DN_MAXACCL) || + (u.acc.acc_passl > DN_MAXACCL) || + (u.acc.acc_userl > DN_MAXACCL)) return -EINVAL; - memcpy(&scp->accessdata, &acc, sizeof(struct accessdata_dn)); + memcpy(&scp->accessdata, &u.acc, optlen); break; case DSO_ACCEPTMODE: @@ -1372,16 +1326,10 @@ static int __dn_setsockopt(struct socket *sock, int level,int optname, char *opt if (optlen != sizeof(int)) return -EINVAL; - { - int mode; - - if (get_user(mode, optval)) - return -EFAULT; - if ((mode != ACC_IMMED) && (mode != ACC_DEFER)) - return -EINVAL; + if ((u.mode != ACC_IMMED) && (u.mode != ACC_DEFER)) + return -EINVAL; - scp->accept_mode = (unsigned char)mode; - } + scp->accept_mode = (unsigned char)u.mode; break; case DSO_CONACCEPT: @@ -1411,8 +1359,55 @@ static int __dn_setsockopt(struct socket *sock, int level,int optname, char *opt case DSO_LINKINFO: case DSO_STREAM: case DSO_SEQPACKET: - return -ENOPROTOOPT; + + case DSO_MAXWINDOW: + if (optlen != sizeof(unsigned long)) + return -EINVAL; + if (u.win > NSP_MAX_WINDOW) + u.win = NSP_MAX_WINDOW; + if (u.win == 0) + return -EINVAL; + scp->max_window = u.win; + if (scp->snd_window > u.win) + scp->snd_window = u.win; + break; + + case DSO_NODELAY: + if (optlen != sizeof(int)) + return -EINVAL; + if (scp->nonagle == 2) + return -EINVAL; + scp->nonagle = (u.val == 0) ? 0 : 1; + /* if (scp->nonagle == 1) { Push pending frames } */ + break; + + case DSO_CORK: + if (optlen != sizeof(int)) + return -EINVAL; + if (scp->nonagle == 1) + return -EINVAL; + scp->nonagle = (u.val == 0) ? 0 : 2; + /* if (scp->nonagle == 0) { Push pending frames } */ + break; + + case DSO_SERVICES: + if (optlen != sizeof(unsigned char)) + return -EINVAL; + if ((u.services & ~NSP_FC_MASK) != 0x01) + return -EINVAL; + if ((u.services & NSP_FC_MASK) == NSP_FC_MASK) + return -EINVAL; + scp->services_loc = u.services; + break; + + case DSO_INFO: + if (optlen != sizeof(unsigned char)) + return -EINVAL; + if (u.info & 0xfc) + return -EINVAL; + scp->info_loc = u.info; + break; } return 0; @@ -1433,44 +1428,40 @@ static int dn_getsockopt(struct socket *sock, int level, int optname, char *optv static int __dn_getsockopt(struct socket *sock, int level,int optname, char *optval,int *optlen, int flags) { struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); struct linkinfo_dn link; - int mode = scp->accept_mode; + int r_len = *optlen; + void *r_data = NULL; + int val; switch(optname) { case DSO_CONDATA: - if (*optlen != sizeof(struct optdata_dn)) - return -EINVAL; - - if (copy_to_user(optval, &scp->conndata_in, sizeof(struct optdata_dn))) - return -EFAULT; + if (r_len > sizeof(struct optdata_dn)) + r_len = sizeof(struct optdata_dn); + r_data = &scp->conndata_in; break; case DSO_DISDATA: - if (*optlen != sizeof(struct optdata_dn)) - return -EINVAL; - - if (copy_to_user(optval, &scp->discdata_in, sizeof(struct optdata_dn))) - return -EFAULT; - + if (r_len > sizeof(struct optdata_dn)) + r_len = sizeof(struct optdata_dn); + r_data = &scp->discdata_in; break; case DSO_CONACCESS: - if (*optlen != sizeof(struct accessdata_dn)) - return -EINVAL; - - if (copy_to_user(optval, &scp->accessdata, sizeof(struct accessdata_dn))) - return -EFAULT; + if (r_len > sizeof(struct accessdata_dn)) + r_len = sizeof(struct accessdata_dn); + r_data = &scp->accessdata; break; case DSO_ACCEPTMODE: - if (put_user(mode, optval)) - return -EFAULT; + if (r_len > sizeof(unsigned char)) + r_len = sizeof(unsigned char); + r_data = &scp->accept_mode; break; case DSO_LINKINFO: - if (*optlen != sizeof(struct linkinfo_dn)) - return -EINVAL; + if (r_len > sizeof(struct linkinfo_dn)) + r_len = sizeof(struct linkinfo_dn); switch(sock->state) { case SS_CONNECTING: @@ -1486,10 +1477,8 @@ static int __dn_getsockopt(struct socket *sock, int level,int optname, char *opt link.idn_linkstate = LL_INACTIVE; } - link.idn_segsize = scp->mss; - - if (copy_to_user(optval, &link, sizeof(struct linkinfo_dn))) - return -EFAULT; + link.idn_segsize = scp->segsize_rem; + r_data = &link; break; default: @@ -1508,6 +1497,45 @@ static int __dn_getsockopt(struct socket *sock, int level,int optname, char *opt case DSO_CONACCEPT: case DSO_CONREJECT: return -ENOPROTOOPT; + + case DSO_MAXWINDOW: + if (r_len > sizeof(unsigned long)) + r_len = sizeof(unsigned long); + r_data = &scp->max_window; + break; + + case DSO_NODELAY: + if (r_len > sizeof(int)) + r_len = sizeof(int); + val = (scp->nonagle == 1); + r_data = &val; + break; + + case DSO_CORK: + if (r_len > sizeof(int)) + r_len = sizeof(int); + val = (scp->nonagle == 2); + r_data = &val; + break; + + case DSO_SERVICES: + if (r_len > sizeof(unsigned char)) + r_len = sizeof(unsigned char); + r_data = &scp->services_rem; + break; + + case DSO_INFO: + if (r_len > sizeof(unsigned char)) + r_len = sizeof(unsigned char); + r_data = &scp->info_rem; + break; + } + + if (r_data) { + if (copy_to_user(optval, r_data, r_len)) + return -EFAULT; + if (put_user(r_len, optlen)) + return -EFAULT; } return 0; @@ -1520,7 +1548,7 @@ static int __dn_getsockopt(struct socket *sock, int level,int optname, char *opt */ static int dn_wait_run(struct sock *sk, int flags) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); int err = 0; switch(scp->state) { @@ -1572,7 +1600,7 @@ static int dn_data_ready(struct sock *sk, struct sk_buff_head *q, int flags, int return skb_queue_len(q) ? 1 : 0; while(skb != (struct sk_buff *)q) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); len += skb->len; if (cb->nsp_flags & 0x40) { @@ -1599,7 +1627,7 @@ static int dn_recvmsg(struct socket *sock, struct msghdr *msg, int size, int flags, struct scm_cookie *scm) { struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); struct sk_buff_head *queue = &sk->receive_queue; int target = size > 1 ? 1 : 0; int copied = 0; @@ -1681,7 +1709,7 @@ static int dn_recvmsg(struct socket *sock, struct msghdr *msg, int size, for(skb = queue->next; skb != (struct sk_buff *)queue; skb = nskb) { int chunk = skb->len; - cb = (struct dn_skb_cb *)skb->cb; + cb = DN_SKB_CB(skb); if ((chunk + copied) > size) chunk = size - copied; @@ -1693,7 +1721,7 @@ static int dn_recvmsg(struct socket *sock, struct msghdr *msg, int size, copied += chunk; if (!(flags & MSG_PEEK)) - skb->len -= chunk; + skb_pull(skb, chunk); eor = cb->nsp_flags & 0x40; nskb = skb->next; @@ -1707,7 +1735,7 @@ static int dn_recvmsg(struct socket *sock, struct msghdr *msg, int size, */ if ((scp->flowloc_sw == DN_DONTSEND) && !dn_congested(sk)) { scp->flowloc_sw = DN_SEND; - dn_nsp_send_lnk(sk, DN_SEND); + dn_nsp_send_link(sk, DN_SEND, 0); } } @@ -1727,6 +1755,7 @@ static int dn_recvmsg(struct socket *sock, struct msghdr *msg, int size, rv = copied; + if (eor && (sk->type == SOCK_SEQPACKET)) msg->msg_flags |= MSG_EOR; @@ -1745,16 +1774,31 @@ out: } +static inline int dn_queue_too_long(struct dn_scp *scp, struct sk_buff_head *queue, int flags) +{ + unsigned char fctype = scp->services_rem & NSP_FC_MASK; + if (skb_queue_len(queue) >= scp->snd_window) + return 1; + if (fctype != NSP_FC_NONE) { + if (flags & MSG_OOB) { + if (scp->flowrem_oth == 0) + return 1; + } else { + if (scp->flowrem_dat == 0) + return 1; + } + } + return 0; +} + static int dn_sendmsg(struct socket *sock, struct msghdr *msg, int size, struct scm_cookie *scm) { struct sock *sk = sock->sk; - struct dn_scp *scp = &sk->protinfo.dn; - int mss = scp->mss; - int mtu = 230 - 11; /* maximum value thats always safe */ + struct dn_scp *scp = DN_SK(sk); + int mss; struct sk_buff_head *queue = &scp->data_xmit_queue; int flags = msg->msg_flags; - unsigned short numseg = 0; int err = 0; int sent = 0; int addr_len = msg->msg_namelen; @@ -1765,6 +1809,7 @@ static int dn_sendmsg(struct socket *sock, struct msghdr *msg, int size, unsigned char *ptr; unsigned short ack; int len; + unsigned char fctype; if (flags & ~(MSG_TRYHARD|MSG_OOB|MSG_DONTWAIT|MSG_EOR)) return -EOPNOTSUPP; @@ -1801,16 +1846,19 @@ static int dn_sendmsg(struct socket *sock, struct msghdr *msg, int size, if ((flags & MSG_TRYHARD) && sk->dst_cache) dst_negative_advice(&sk->dst_cache); + mss = scp->segsize_rem; + fctype = scp->services_rem & NSP_FC_MASK; + if (sk->dst_cache && sk->dst_cache->neighbour) { struct dn_neigh *dn = (struct dn_neigh *)sk->dst_cache->neighbour; - if (dn->blksize > 230) - mtu = dn->blksize - 11; + if (dn->blksize < (mss + 11)) + mss = dn->blksize - 11; } /* * The only difference between SEQPACKET & STREAM sockets under DECnet - * AFAIK is that SEQPACKET sockets set the MSG_EOR flag for the last - * session control message segment. + * is that SEQPACKET sockets set the MSG_EOR flag for the last + * session control message segment. */ if (flags & MSG_OOB) { @@ -1822,9 +1870,6 @@ static int dn_sendmsg(struct socket *sock, struct msghdr *msg, int size, } } - if (mss < mtu) - mtu = mss; - scp->persist_fxn = dn_nsp_xmit_timeout; while(sent < size) { @@ -1842,14 +1887,14 @@ static int dn_sendmsg(struct socket *sock, struct msghdr *msg, int size, */ len = size - sent; - if (len > mtu) - len = mtu; + if (len > mss) + len = mss; /* * Wait for queue size to go down below the window * size. */ - if (skb_queue_len(queue) >= scp->snd_window) { + if (dn_queue_too_long(scp, queue, flags)) { if (flags & MSG_DONTWAIT) { err = -EWOULDBLOCK; goto out; @@ -1857,7 +1902,7 @@ static int dn_sendmsg(struct socket *sock, struct msghdr *msg, int size, SOCK_SLEEP_PRE(sk) - if (skb_queue_len(queue) >= scp->snd_window) + if (dn_queue_too_long(scp, queue, flags)) schedule(); SOCK_SLEEP_POST(sk) @@ -1876,7 +1921,7 @@ static int dn_sendmsg(struct socket *sock, struct msghdr *msg, int size, if (!skb) continue; - cb = (struct dn_skb_cb *)skb->cb; + cb = DN_SKB_CB(skb); ptr = skb_put(skb, 9); @@ -1886,26 +1931,34 @@ static int dn_sendmsg(struct socket *sock, struct msghdr *msg, int size, } if (flags & MSG_OOB) { - cb->segnum = scp->numoth++; - scp->numoth &= 0x0fff; + cb->segnum = scp->numoth; + seq_add(&scp->numoth, 1); msgflg = 0x30; - ack = scp->ackxmt_oth | 0x8000; + ack = (scp->numoth_rcv & 0x0FFF) | 0x8000; + scp->ackxmt_oth = scp->numoth_rcv; + if (fctype != NSP_FC_NONE) + scp->flowrem_oth--; } else { - cb->segnum = scp->numdat++; - scp->numdat &= 0x0fff; + cb->segnum = scp->numdat; + seq_add(&scp->numdat, 1); msgflg = 0x00; if (sock->type == SOCK_STREAM) msgflg = 0x60; - if (scp->seg_size == 0) + if (scp->seg_total == 0) msgflg |= 0x20; - scp->seg_size += len; + scp->seg_total += len; if (((sent + len) == size) && (flags & MSG_EOR)) { msgflg |= 0x40; - scp->seg_size = 0; + scp->seg_total = 0; + if (fctype == NSP_FC_SCMC) + scp->flowrem_dat--; } - ack = scp->ackxmt_dat | 0x8000; + ack = (scp->numdat_rcv & 0x0FFF) | 0x8000; + scp->ackxmt_dat = scp->numdat_rcv; + if (fctype == NSP_FC_SRC) + scp->flowrem_dat--; } *ptr++ = msgflg; @@ -1918,8 +1971,7 @@ static int dn_sendmsg(struct socket *sock, struct msghdr *msg, int size, *(__u16 *)ptr = dn_htons(cb->segnum); sent += len; - dn_nsp_queue_xmit(sk, skb, flags & MSG_OOB); - numseg++; + dn_nsp_queue_xmit(sk, skb, sk->allocation, flags & MSG_OOB); skb = NULL; scp->persist = dn_nsp_persist(sk); @@ -1955,21 +2007,38 @@ static int dn_device_event(struct notifier_block *this, unsigned long event, } static struct notifier_block dn_dev_notifier = { - dn_device_event, - 0 + notifier_call: dn_device_event, }; extern int dn_route_rcv(struct sk_buff *, struct net_device *, struct packet_type *); -static struct packet_type dn_dix_packet_type = -{ - __constant_htons(ETH_P_DNA_RT), - NULL, /* All devices */ - dn_route_rcv, - (void*)1, - NULL, +static struct packet_type dn_dix_packet_type = { + type: __constant_htons(ETH_P_DNA_RT), + dev: NULL, /* All devices */ + func: dn_route_rcv, + data: (void*)1, }; +#define IS_NOT_PRINTABLE(x) ((x) < 32 || (x) > 126) + +static void dn_printable_object(struct sockaddr_dn *dn, unsigned char *buf) +{ + int i; + + switch (dn_ntohs(dn->sdn_objnamel)) { + case 0: + sprintf(buf, "%d", dn->sdn_objnum); + break; + default: + for (i = 0; i < dn_ntohs(dn->sdn_objnamel); i++) { + buf[i] = dn->sdn_objname[i]; + if (IS_NOT_PRINTABLE(buf[i])) + buf[i] = '.'; + } + buf[i] = 0; + } +} + static int dn_get_info(char *buffer, char **start, off_t offset, int length) { struct sock *sk; @@ -1979,15 +2048,20 @@ static int dn_get_info(char *buffer, char **start, off_t offset, int length) off_t begin = 0; char buf1[DN_ASCBUF_LEN]; char buf2[DN_ASCBUF_LEN]; + char local_object[DN_MAXOBJL+3]; + char remote_object[DN_MAXOBJL+3]; - len += sprintf(buffer + len, "Local Remote\n"); + len += sprintf(buffer + len, "Local Remote\n"); read_lock(&dn_hash_lock); for(sk = dn_sklist; sk != NULL; sk = sk->next) { - scp = &sk->protinfo.dn; + scp = DN_SK(sk); + + dn_printable_object(&scp->addr, local_object); + dn_printable_object(&scp->peer, remote_object); len += sprintf(buffer + len, - "%6s/%04X %04d:%04d %04d:%04d %01d %6s/%04X %04d:%04d %04d:%04d %01d %4s %s\n", + "%6s/%04X %04d:%04d %04d:%04d %01d %-16s %6s/%04X %04d:%04d %04d:%04d %01d %-16s %4s %s\n", dn_addr2asc(dn_ntohs(dn_saddr2dn(&scp->addr)), buf1), scp->addrloc, scp->numdat, @@ -1995,6 +2069,7 @@ static int dn_get_info(char *buffer, char **start, off_t offset, int length) scp->ackxmt_dat, scp->ackxmt_oth, scp->flowloc_sw, + local_object, dn_addr2asc(dn_ntohs(dn_saddr2dn(&scp->peer)), buf2), scp->addrrem, scp->numdat_rcv, @@ -2002,6 +2077,7 @@ static int dn_get_info(char *buffer, char **start, off_t offset, int length) scp->ackrcv_dat, scp->ackrcv_oth, scp->flowrem_sw, + remote_object, dn_state2asc(scp->state), ((scp->accept_mode == ACC_IMMED) ? "IMMED" : "DEFER")); @@ -2026,8 +2102,8 @@ static int dn_get_info(char *buffer, char **start, off_t offset, int length) static struct net_proto_family dn_family_ops = { - AF_DECnet, - dn_create + family: AF_DECnet, + create: dn_create, }; static struct proto_ops dn_proto_ops = { @@ -2067,6 +2143,7 @@ MODULE_PARM(addr, "2i"); MODULE_PARM_DESC(addr, "The DECnet address of this machine: area,node"); #endif +static char banner[] __initdata = KERN_INFO "NET4: DECnet for Linux: V.2.4.0-test12s (C) 1995-2000 Linux DECnet Project Team\n"; static int __init decnet_init(void) { @@ -2085,7 +2162,7 @@ static int __init decnet_init(void) dn_dn2eth(decnet_ether_address, dn_ntohs(decnet_address)); #endif - printk(KERN_INFO "NET4: DECnet for Linux: V.2.4.0-test10s (C) 1995-2000 Linux DECnet Project Team\n"); + printk(banner); sock_register(&dn_family_ops); dev_add_pack(&dn_dix_packet_type); diff --git a/net/decnet/dn_dev.c b/net/decnet/dn_dev.c index d8f91ac38..056eaa043 100644 --- a/net/decnet/dn_dev.c +++ b/net/decnet/dn_dev.c @@ -52,7 +52,7 @@ static unsigned char dn_eco_version[3] = {0x02,0x00,0x00}; extern struct neigh_table dn_neigh_table; -struct net_device *decnet_default_device = NULL; +struct net_device *decnet_default_device; static struct dn_dev *dn_dev_create(struct net_device *dev, int *err); static void dn_dev_delete(struct net_device *dev); @@ -1286,9 +1286,7 @@ void __exit dn_dev_cleanup(void) } #endif /* CONFIG_SYSCTL */ -#ifdef CONFIG_PROC_FS proc_net_remove("decnet_dev"); -#endif /* CONFIG_PROC_FS */ dn_dev_devices_off(); } diff --git a/net/decnet/dn_fib.c b/net/decnet/dn_fib.c index aff2dc05d..f82e26406 100644 --- a/net/decnet/dn_fib.c +++ b/net/decnet/dn_fib.c @@ -55,7 +55,7 @@ extern int dn_cache_dump(struct sk_buff *skb, struct netlink_callback *cb); #endif /* CONFIG_RTNETLINK */ -static struct dn_fib_info *dn_fib_info_list = NULL; +static struct dn_fib_info *dn_fib_info_list; static rwlock_t dn_fib_info_lock = RW_LOCK_UNLOCKED; int dn_fib_info_cnt; @@ -641,15 +641,11 @@ static int decnet_rt_get_info(char *buffer, char **start, off_t offset, int leng return 0; } - #endif /* CONFIG_PROC_FS */ - void __exit dn_fib_cleanup(void) { -#ifdef CONFIG_PROC_FS proc_net_remove("decnet_route"); -#endif dn_fib_table_cleanup(); dn_fib_rules_cleanup(); diff --git a/net/decnet/dn_neigh.c b/net/decnet/dn_neigh.c index 00c27cdec..f25f37275 100644 --- a/net/decnet/dn_neigh.c +++ b/net/decnet/dn_neigh.c @@ -18,6 +18,8 @@ * forwarding now stands a good chance of * working. * Steve Whitehouse : Fixed neighbour states (for now anyway). + * Steve Whitehouse : Made error_report functions dummies. This + * is not the right place to return skbs. * */ @@ -52,81 +54,66 @@ static int dn_phase3_output(struct sk_buff *); * For talking to broadcast devices: Ethernet & PPP */ static struct neigh_ops dn_long_ops = { - AF_DECnet, - NULL, - NULL, - dn_long_error_report, - dn_long_output, - dn_long_output, - dev_queue_xmit, - dev_queue_xmit + family: AF_DECnet, + error_report: dn_long_error_report, + output: dn_long_output, + connected_output: dn_long_output, + hh_output: dev_queue_xmit, + queue_xmit: dev_queue_xmit, }; /* * For talking to pointopoint and multidrop devices: DDCMP and X.25 */ static struct neigh_ops dn_short_ops = { - AF_DECnet, - NULL, - NULL, - dn_short_error_report, - dn_short_output, - dn_short_output, - dev_queue_xmit, - dev_queue_xmit + family: AF_DECnet, + error_report: dn_short_error_report, + output: dn_short_output, + connected_output: dn_short_output, + hh_output: dev_queue_xmit, + queue_xmit: dev_queue_xmit, }; /* * For talking to DECnet phase III nodes */ static struct neigh_ops dn_phase3_ops = { - AF_DECnet, - NULL, - NULL, - dn_short_error_report, /* Can use short version here */ - dn_phase3_output, - dn_phase3_output, - dev_queue_xmit, - dev_queue_xmit + family: AF_DECnet, + error_report: dn_short_error_report, /* Can use short version here */ + output: dn_phase3_output, + connected_output: dn_phase3_output, + hh_output: dev_queue_xmit, + queue_xmit: dev_queue_xmit }; struct neigh_table dn_neigh_table = { - NULL, - PF_DECnet, - sizeof(struct dn_neigh), - sizeof(dn_address), - dn_neigh_hash, - dn_neigh_construct, - NULL, /* pconstructor */ - NULL, /* pdestructor */ - NULL, /* proxyredo */ - "dn_neigh_cache", - { - NULL, - NULL, - &dn_neigh_table, - 0, - NULL, - NULL, - 30 * HZ, /* base_reachable_time */ - 1 * HZ, /* retrans_time */ - 60 * HZ, /* gc_staletime */ - 30 * HZ, /* reachable_time */ - 5 * HZ, /* delay_probe_time */ - 3, /* queue_len */ - 0, /* ucast_probes */ - 0, /* app_probes */ - 0, /* mcast_probes */ - 0, /* anycast_delay */ - 0, /* proxy_delay */ - 0, /* proxy_qlen */ - 1 * HZ, /* locktime */ + family: PF_DECnet, + entry_size: sizeof(struct dn_neigh), + key_len: sizeof(dn_address), + hash: dn_neigh_hash, + constructor: dn_neigh_construct, + id: "dn_neigh_cache", + parms: { + tbl: &dn_neigh_table, + entries: 0, + base_reachable_time: 30 * HZ, + retrans_time: 1 * HZ, + gc_staletime: 60 * HZ, + reachable_time: 30 * HZ, + delay_probe_time: 5 * HZ, + queue_len: 3, + ucast_probes: 0, + app_probes: 0, + mcast_probes: 0, + anycast_delay: 0, + proxy_delay: 0, + proxy_qlen: 0, + locktime: 1 * HZ, }, - 30 * HZ, /* gc_interval */ - 128, /* gc_thresh1 */ - 512, /* gc_thresh2 */ - 1024, /* gc_thresh3 */ - + gc_interval: 30 * HZ, + gc_thresh1: 128, + gc_thresh2: 512, + gc_thresh3: 1024, }; static u32 dn_neigh_hash(const void *pkey, const struct net_device *dev) @@ -180,66 +167,15 @@ static int dn_neigh_construct(struct neighbour *neigh) static void dn_long_error_report(struct neighbour *neigh, struct sk_buff *skb) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; - unsigned char *ptr; - printk(KERN_DEBUG "dn_long_error_report: called\n"); - - if (!(cb->rt_flags & DN_RT_F_RQR)) { - kfree_skb(skb); - return; - } - - skb_push(skb, skb->data - skb->nh.raw); - ptr = skb->data; - - *(unsigned short *)ptr = dn_htons(skb->len - 2); - ptr += 2; - - if (*ptr & DN_RT_F_PF) { - char padlen = (*ptr & ~DN_RT_F_PF); - ptr += padlen; - } - - *ptr++ |= (cb->rt_flags & ~DN_RT_F_RQR) | DN_RT_F_RTS; - - ptr += 2; - dn_dn2eth(ptr, dn_ntohs(cb->src)); - ptr += 8; - dn_dn2eth(ptr, dn_ntohs(cb->dst)); - ptr += 6; - *ptr = 0; - - skb->dst->neighbour->ops->queue_xmit(skb); + kfree_skb(skb); } static void dn_short_error_report(struct neighbour *neigh, struct sk_buff *skb) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; - unsigned char *ptr; - printk(KERN_DEBUG "dn_short_error_report: called\n"); - - if (!(cb->rt_flags & DN_RT_F_RQR)) { - kfree_skb(skb); - return; - } - - skb_push(skb, skb->data - skb->nh.raw); - ptr = skb->data; - - *(unsigned short *)ptr = dn_htons(skb->len - 2); - ptr += 2; - *ptr++ = (cb->rt_flags & ~DN_RT_F_RQR) | DN_RT_F_RTS; - - *(dn_address *)ptr = cb->src; - ptr += 2; - *(dn_address *)ptr = cb->dst; - ptr += 2; - *ptr = 0; - - skb->dst->neighbour->ops->queue_xmit(skb); + kfree_skb(skb); } static int dn_neigh_output_packet(struct sk_buff *skb) @@ -266,7 +202,7 @@ static int dn_long_output(struct sk_buff *skb) int headroom = dev->hard_header_len + sizeof(struct dn_long_packet) + 3; unsigned char *data; struct dn_long_packet *lp; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); if (skb_headroom(skb) < headroom) { @@ -312,7 +248,7 @@ static int dn_short_output(struct sk_buff *skb) int headroom = dev->hard_header_len + sizeof(struct dn_short_packet) + 2; struct dn_short_packet *sp; unsigned char *data; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); if (skb_headroom(skb) < headroom) { @@ -355,7 +291,7 @@ static int dn_phase3_output(struct sk_buff *skb) int headroom = dev->hard_header_len + sizeof(struct dn_short_packet) + 2; struct dn_short_packet *sp; unsigned char *data; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); if (skb_headroom(skb) < headroom) { struct sk_buff *skb2 = skb_realloc_headroom(skb, headroom); @@ -659,8 +595,6 @@ void __init dn_neigh_init(void) void __exit dn_neigh_cleanup(void) { -#ifdef CONFIG_PROC_FS proc_net_remove("decnet_neigh"); -#endif /* CONFIG_PROC_FS */ neigh_table_clear(&dn_neigh_table); } diff --git a/net/decnet/dn_nsp_in.c b/net/decnet/dn_nsp_in.c index 361729458..11a4e8237 100644 --- a/net/decnet/dn_nsp_in.c +++ b/net/decnet/dn_nsp_in.c @@ -25,6 +25,9 @@ * Steve Whitehouse: * Patrick Caulfield: Checking conninits for correctness & sending of error * responses. + * Steve Whitehouse: Added backlog congestion level return codes. + * Patrick Caulfield: + * Steve Whitehouse: Added flow control support (outbound) */ /****************************************************************************** @@ -79,7 +82,7 @@ static void dn_log_martian(struct sk_buff *skb, const char *msg) { if (decnet_log_martians && net_ratelimit()) { char *devname = skb->dev ? skb->dev->name : "???"; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); printk(KERN_INFO "DECnet: Martian packet (%s) dev=%s src=0x%04hx dst=0x%04hx srcport=0x%04hx dstport=0x%04hx\n", msg, devname, cb->src, cb->dst, cb->src_port, cb->dst_port); } } @@ -91,7 +94,7 @@ static void dn_log_martian(struct sk_buff *skb, const char *msg) */ static void dn_ack(struct sock *sk, struct sk_buff *skb, unsigned short ack) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); unsigned short type = ((ack >> 12) & 0x0003); int wakeup = 0; @@ -212,7 +215,7 @@ static struct { */ static struct sock *dn_find_listener(struct sk_buff *skb, unsigned short *reason) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); struct nsp_conn_init_msg *msg = (struct nsp_conn_init_msg *)skb->data; struct sockaddr_dn dstaddr; struct sockaddr_dn srcaddr; @@ -331,33 +334,26 @@ static void dn_nsp_conn_init(struct sock *sk, struct sk_buff *skb) static void dn_nsp_conn_conf(struct sock *sk, struct sk_buff *skb) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_skb_cb *cb = DN_SKB_CB(skb); + struct dn_scp *scp = DN_SK(sk); + unsigned char *ptr; - if (skb->len < 3) + if (skb->len < 4) goto out; - cb->services = *skb->data; - cb->info = *(skb->data+1); - skb_pull(skb, 2); - cb->segsize = dn_ntohs(*(__u16 *)skb->data); - skb_pull(skb, 2); - - /* - * FIXME: Check out services and info fields to check that - * we can talk to this kind of node. - */ + ptr = skb->data; + cb->services = *ptr++; + cb->info = *ptr++; + cb->segsize = dn_ntohs(*(__u16 *)ptr); if ((scp->state == DN_CI) || (scp->state == DN_CD)) { scp->persist = 0; scp->addrrem = cb->src_port; sk->state = TCP_ESTABLISHED; scp->state = DN_RUN; - - if (scp->mss > cb->segsize) - scp->mss = cb->segsize; - if (scp->mss < 230) - scp->mss = 230; + scp->services_rem = cb->services; + scp->info_rem = cb->info; + scp->segsize_rem = cb->segsize; if (skb->len > 0) { unsigned char dlen = *skb->data; @@ -366,7 +362,7 @@ static void dn_nsp_conn_conf(struct sock *sk, struct sk_buff *skb) memcpy(scp->conndata_in.opt_data, skb->data + 1, dlen); } } - dn_nsp_send_lnk(sk, DN_NOCHANGE); + dn_nsp_send_link(sk, DN_NOCHANGE, 0); if (!sk->dead) sk->state_change(sk); } @@ -377,7 +373,7 @@ out: static void dn_nsp_conn_ack(struct sock *sk, struct sk_buff *skb) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); if (scp->state == DN_CI) { scp->state = DN_CD; @@ -389,8 +385,8 @@ static void dn_nsp_conn_ack(struct sock *sk, struct sk_buff *skb) static void dn_nsp_disc_init(struct sock *sk, struct sk_buff *skb) { - struct dn_scp *scp = &sk->protinfo.dn; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_scp *scp = DN_SK(sk); + struct dn_skb_cb *cb = DN_SKB_CB(skb); unsigned short reason; if (skb->len < 2) @@ -448,7 +444,7 @@ out: */ static void dn_nsp_disc_conf(struct sock *sk, struct sk_buff *skb) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); unsigned short reason; if (skb->len != 2) @@ -492,38 +488,65 @@ out: static void dn_nsp_linkservice(struct sock *sk, struct sk_buff *skb) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_scp *scp = DN_SK(sk); unsigned short segnum; unsigned char lsflags; char fcval; + int wake_up = 0; + char *ptr = skb->data; + unsigned char fctype = scp->services_rem & NSP_FC_MASK; if (skb->len != 4) goto out; - cb->segnum = segnum = dn_ntohs(*(__u16 *)skb->data); - skb_pull(skb, 2); - lsflags = *(unsigned char *)skb->data; - skb_pull(skb, 1); - fcval = *(char *)skb->data; + segnum = dn_ntohs(*(__u16 *)ptr); + ptr += 2; + lsflags = *(unsigned char *)ptr++; + fcval = *ptr; - if (lsflags & 0xf0) + /* + * Here we ignore erronous packets which should really + * should cause a connection abort. It is not critical + * for now though. + */ + if (lsflags & 0xf8) goto out; - if (((sk->protinfo.dn.numoth_rcv + 1) & 0x0FFF) == (segnum & 0x0FFF)) { - sk->protinfo.dn.numoth_rcv += 1; - switch(lsflags & 0x03) { - case 0x00: - break; - case 0x01: - sk->protinfo.dn.flowrem_sw = DN_DONTSEND; - break; - case 0x02: - sk->protinfo.dn.flowrem_sw = DN_SEND; + if (seq_next(scp->numoth_rcv, segnum)) { + seq_add(&scp->numoth_rcv, 1); + switch(lsflags & 0x04) { /* FCVAL INT */ + case 0x00: /* Normal Request */ + switch(lsflags & 0x03) { /* FCVAL MOD */ + case 0x00: /* Request count */ + if (fcval < 0) { + unsigned char p_fcval = -fcval; + if ((scp->flowrem_dat > p_fcval) && + (fctype == NSP_FC_SCMC)) { + scp->flowrem_dat -= p_fcval; + } + } else if (fcval > 0) { + scp->flowrem_dat += fcval; + wake_up = 1; + } + break; + case 0x01: /* Stop outgoing data */ + scp->flowrem_sw = DN_DONTSEND; + break; + case 0x02: /* Ok to start again */ + scp->flowrem_sw = DN_SEND; dn_nsp_output(sk); - if (!sk->dead) - sk->state_change(sk); + wake_up = 1; + } + break; + case 0x04: /* Interrupt Request */ + if (fcval > 0) { + scp->flowrem_oth += fcval; + wake_up = 1; + } + break; } - + if (wake_up && !sk->dead) + sk->state_change(sk); } dn_nsp_send_oth_ack(sk); @@ -582,9 +605,9 @@ static __inline__ int dn_queue_skb(struct sock *sk, struct sk_buff *skb, int sig static void dn_nsp_otherdata(struct sock *sk, struct sk_buff *skb) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); unsigned short segnum; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); int queued = 0; if (skb->len < 2) @@ -593,10 +616,10 @@ static void dn_nsp_otherdata(struct sock *sk, struct sk_buff *skb) cb->segnum = segnum = dn_ntohs(*(__u16 *)skb->data); skb_pull(skb, 2); - if (((sk->protinfo.dn.numoth_rcv + 1) & 0x0fff) == (segnum & 0x0fff)) { + if (seq_next(scp->numoth_rcv, segnum)) { if (dn_queue_skb(sk, skb, SIGURG, &scp->other_receive_queue) == 0) { - sk->protinfo.dn.numoth_rcv++; + seq_add(&scp->numoth_rcv, 1); scp->other_report = 0; queued = 1; } @@ -612,8 +635,8 @@ static void dn_nsp_data(struct sock *sk, struct sk_buff *skb) { int queued = 0; unsigned short segnum; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_skb_cb *cb = DN_SKB_CB(skb); + struct dn_scp *scp = DN_SK(sk); if (skb->len < 2) goto out; @@ -621,17 +644,15 @@ static void dn_nsp_data(struct sock *sk, struct sk_buff *skb) cb->segnum = segnum = dn_ntohs(*(__u16 *)skb->data); skb_pull(skb, 2); - if (((sk->protinfo.dn.numdat_rcv + 1) & 0x0FFF) == - (segnum & 0x0FFF)) { - + if (seq_next(scp->numdat_rcv, segnum)) { if (dn_queue_skb(sk, skb, SIGIO, &sk->receive_queue) == 0) { - sk->protinfo.dn.numdat_rcv++; + seq_add(&scp->numdat_rcv, 1); queued = 1; } if ((scp->flowloc_sw == DN_SEND) && dn_congested(sk)) { scp->flowloc_sw = DN_DONTSEND; - dn_nsp_send_lnk(sk, DN_DONTSEND); + dn_nsp_send_link(sk, DN_DONTSEND, 0); } } @@ -648,7 +669,7 @@ out: */ static void dn_returned_conn_init(struct sock *sk, struct sk_buff *skb) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); if (scp->state == DN_CI) { scp->state = DN_NC; @@ -660,28 +681,37 @@ static void dn_returned_conn_init(struct sock *sk, struct sk_buff *skb) kfree_skb(skb); } -static void dn_nsp_no_socket(struct sk_buff *skb, unsigned short reason) +static int dn_nsp_no_socket(struct sk_buff *skb, unsigned short reason) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); + int ret = NET_RX_DROP; + + /* Must not reply to returned packets */ + if (cb->rt_flags & DN_RT_F_RTS) + goto out; if ((reason != NSP_REASON_OK) && ((cb->nsp_flags & 0x0c) == 0x08)) { switch(cb->nsp_flags & 0x70) { case 0x10: case 0x60: /* (Retransmitted) Connect Init */ dn_nsp_return_disc(skb, NSP_DISCINIT, reason); + ret = NET_RX_SUCCESS; break; case 0x20: /* Connect Confirm */ dn_nsp_return_disc(skb, NSP_DISCCONF, reason); + ret = NET_RX_SUCCESS; break; } } +out: kfree_skb(skb); + return ret; } static int dn_nsp_rx_packet(struct sk_buff *skb) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); struct sock *sk = NULL; unsigned char *ptr = (unsigned char *)skb->data; unsigned short reason = NSP_REASON_NL; @@ -754,14 +784,19 @@ static int dn_nsp_rx_packet(struct sk_buff *skb) sk = dn_find_by_skb(skb); got_it: if (sk != NULL) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); int ret; /* Reset backoff */ scp->nsp_rxtshift = 0; bh_lock_sock(sk); - ret = 0; + ret = NET_RX_SUCCESS; + if (decnet_debug_level & 8) + printk(KERN_DEBUG "NSP: 0x%02x 0x%02x 0x%04x 0x%04x %d\n", + (int)cb->rt_flags, (int)cb->nsp_flags, + (int)cb->src_port, (int)cb->dst_port, + (int)sk->lock.users); if (sk->lock.users == 0) ret = dn_nsp_backlog_rcv(sk, skb); else @@ -772,12 +807,11 @@ got_it: return ret; } - dn_nsp_no_socket(skb, reason); - return 1; + return dn_nsp_no_socket(skb, reason); free_out: kfree_skb(skb); - return 0; + return NET_RX_DROP; } int dn_nsp_rx(struct sk_buff *skb) @@ -792,12 +826,12 @@ int dn_nsp_rx(struct sk_buff *skb) */ int dn_nsp_backlog_rcv(struct sock *sk, struct sk_buff *skb) { - struct dn_scp *scp = &sk->protinfo.dn; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_scp *scp = DN_SK(sk); + struct dn_skb_cb *cb = DN_SKB_CB(skb); if (cb->rt_flags & DN_RT_F_RTS) { dn_returned_conn_init(sk, skb); - return 0; + return NET_RX_SUCCESS; } /* @@ -875,6 +909,6 @@ free_out: } } - return 0; + return NET_RX_SUCCESS; } diff --git a/net/decnet/dn_nsp_out.c b/net/decnet/dn_nsp_out.c index 6965cbf42..5e8482ce0 100644 --- a/net/decnet/dn_nsp_out.c +++ b/net/decnet/dn_nsp_out.c @@ -20,6 +20,7 @@ * Steve Whitehouse: New output state machine * Paul Koning: Connect Confirm message fix. * Eduardo Serrat: Fix to stop dn_nsp_do_disc() sending malformed packets. + * Steve Whitehouse: dn_nsp_output() and friends needed a spring clean */ /****************************************************************************** @@ -165,7 +166,7 @@ struct sk_buff *dn_alloc_send_skb(struct sock *sk, int *size, int noblock, int * */ unsigned long dn_nsp_persist(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); unsigned long t = ((scp->nsp_srtt >> 2) + scp->nsp_rttvar) >> 1; @@ -188,7 +189,7 @@ unsigned long dn_nsp_persist(struct sock *sk) */ static void dn_nsp_rtt(struct sock *sk, long rtt) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); long srtt = (long)scp->nsp_srtt; long rttvar = (long)scp->nsp_rttvar; long delta; @@ -223,65 +224,64 @@ static void dn_nsp_rtt(struct sock *sk, long rtt) /* printk(KERN_DEBUG "srtt=%lu rttvar=%lu\n", scp->nsp_srtt, scp->nsp_rttvar); */ } -/* - * Walk the queues, otherdata/linkservice first. Send as many - * frames as the window allows, increment send counts on all - * skbs which are sent. Reduce the window if we are retransmitting - * frames. +/** + * dn_nsp_clone_and_send - Send a data packet by cloning it + * @skb: The packet to clone and transmit + * @gfp: memory allocation flag + * + * Clone a queued data or other data packet and transmit it. + * + * Returns: The number of times the packet has been sent previously */ -void dn_nsp_output(struct sock *sk) +static inline unsigned dn_nsp_clone_and_send(struct sk_buff *skb, int gfp) { - struct dn_scp *scp = &sk->protinfo.dn; - unsigned long win = scp->snd_window; - struct sk_buff *skb, *skb2, *list; - struct dn_skb_cb *cb; - int reduce_win = 0; + struct dn_skb_cb *cb = DN_SKB_CB(skb); + struct sk_buff *skb2; + int ret = 0; - /* printk(KERN_DEBUG "dn_nsp_output: ping\n"); */ + if ((skb2 = skb_clone(skb, gfp)) != NULL) { + ret = cb->xmit_count; + cb->xmit_count++; + cb->stamp = jiffies; + skb2->sk = skb->sk; + dn_nsp_send(skb2); + } + + return ret; +} + +/** + * dn_nsp_output - Try and send something from socket queues + * @sk: The socket whose queues are to be investigated + * @gfp: The memory allocation flags + * + * Try and send the packet on the end of the data and other data queues. + * Other data gets priority over data, and if we retransmit a packet we + * reduce the window by dividing it in two. + * + */ +void dn_nsp_output(struct sock *sk) +{ + struct dn_scp *scp = DN_SK(sk); + struct sk_buff *skb; + unsigned reduce_win = 0; /* * First we check for otherdata/linkservice messages */ - skb = scp->other_xmit_queue.next; - list = (struct sk_buff *)&scp->other_xmit_queue; - while(win && (skb != list)) { - if ((skb2 = skb_clone(skb, GFP_ATOMIC)) != NULL) { - cb = (struct dn_skb_cb *)skb; - if (cb->xmit_count > 0) - reduce_win = 1; - else - cb->stamp = jiffies; - cb->xmit_count++; - skb2->sk = sk; - dn_nsp_send(skb2); - } - skb = skb->next; - win--; - } + if ((skb = skb_peek(&scp->other_xmit_queue)) != NULL) + reduce_win = dn_nsp_clone_and_send(skb, GFP_ATOMIC); /* * If we may not send any data, we don't. - * Should this apply to otherdata as well ? - SJW + * If we are still trying to get some other data down the + * channel, we don't try and send any data. */ - if (scp->flowrem_sw != DN_SEND) + if (reduce_win || (scp->flowrem_sw != DN_SEND)) goto recalc_window; - skb = scp->data_xmit_queue.next; - list = (struct sk_buff *)&scp->data_xmit_queue; - while(win && (skb != list)) { - if ((skb2 = skb_clone(skb, GFP_ATOMIC)) != NULL) { - cb = (struct dn_skb_cb *)skb; - if (cb->xmit_count > 0) - reduce_win = 1; - else - cb->stamp = jiffies; - cb->xmit_count++; - skb2->sk = sk; - dn_nsp_send(skb2); - } - skb = skb->next; - win--; - } + if ((skb = skb_peek(&scp->data_xmit_queue)) != NULL) + reduce_win = dn_nsp_clone_and_send(skb, GFP_ATOMIC); /* * If we've sent any frame more than once, we cut the @@ -290,7 +290,6 @@ void dn_nsp_output(struct sock *sk) */ recalc_window: if (reduce_win) { - /* printk(KERN_DEBUG "Window reduction %ld\n", scp->snd_window); */ scp->snd_window >>= 1; if (scp->snd_window < NSP_MIN_WINDOW) scp->snd_window = NSP_MIN_WINDOW; @@ -299,7 +298,7 @@ recalc_window: int dn_nsp_xmit_timeout(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); dn_nsp_output(sk); @@ -309,14 +308,60 @@ int dn_nsp_xmit_timeout(struct sock *sk) return 0; } -void dn_nsp_queue_xmit(struct sock *sk, struct sk_buff *skb, int oth) +static inline unsigned char *dn_mk_common_header(struct dn_scp *scp, struct sk_buff *skb, unsigned char msgflag, int len) +{ + unsigned char *ptr = skb_push(skb, len); + + if (len < 5) + BUG(); + + *ptr++ = msgflag; + *((unsigned short *)ptr) = scp->addrrem; + ptr += 2; + *((unsigned short *)ptr) = scp->addrloc; + ptr += 2; + return ptr; +} + +static unsigned short *dn_mk_ack_header(struct sock *sk, struct sk_buff *skb, unsigned char msgflag, int hlen, int other) +{ + struct dn_scp *scp = DN_SK(sk); + unsigned short acknum = scp->numdat_rcv & 0x0FFF; + unsigned short ackcrs = scp->numoth_rcv & 0x0FFF; + unsigned short *ptr; + + if (hlen < 9) + BUG(); + + scp->ackxmt_dat = acknum; + scp->ackxmt_oth = ackcrs; + acknum |= 0x8000; + ackcrs |= 0x8000; + + /* If this is an "other data/ack" message, swap acknum and ackcrs */ + if (other) { + unsigned short tmp = acknum; + acknum = ackcrs; + ackcrs = tmp; + } + + /* Set "cross subchannel" bit in ackcrs */ + ackcrs |= 0x2000; + + ptr = (unsigned short *)dn_mk_common_header(scp, skb, msgflag, hlen); + + *ptr++ = dn_htons(acknum); + *ptr++ = dn_htons(ackcrs); + + return ptr; +} + +void dn_nsp_queue_xmit(struct sock *sk, struct sk_buff *skb, int gfp, int oth) { - struct dn_scp *scp = &sk->protinfo.dn; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_scp *scp = DN_SK(sk); + struct dn_skb_cb *cb = DN_SKB_CB(skb); unsigned long t = ((scp->nsp_srtt >> 2) + scp->nsp_rttvar) >> 1; - struct sk_buff *skb2; - if (t < HZ) t = HZ; /* * Slow start: If we have been idle for more than * one RTT, then reset window to min size. @@ -336,20 +381,17 @@ void dn_nsp_queue_xmit(struct sock *sk, struct sk_buff *skb, int oth) if (scp->flowrem_sw != DN_SEND) return; - if ((skb2 = skb_clone(skb, GFP_ATOMIC)) != NULL) { - cb->stamp = jiffies; - cb->xmit_count++; - skb2->sk = sk; - dn_nsp_send(skb2); - } + dn_nsp_clone_and_send(skb, gfp); } + int dn_nsp_check_xmit_queue(struct sock *sk, struct sk_buff *skb, struct sk_buff_head *q, unsigned short acknum) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_skb_cb *cb = DN_SKB_CB(skb); + struct dn_scp *scp = DN_SK(sk); struct sk_buff *skb2, *list, *ack = NULL; int wakeup = 0; + int try_retrans = 0; unsigned long reftime = cb->stamp; unsigned long pkttime; unsigned short xmit_count; @@ -358,7 +400,7 @@ int dn_nsp_check_xmit_queue(struct sock *sk, struct sk_buff *skb, struct sk_buff skb2 = q->next; list = (struct sk_buff *)q; while(list != skb2) { - struct dn_skb_cb *cb2 = (struct dn_skb_cb *)skb2->cb; + struct dn_skb_cb *cb2 = DN_SKB_CB(skb2); if (before_or_equal(cb2->segnum, acknum)) ack = skb2; @@ -372,27 +414,50 @@ int dn_nsp_check_xmit_queue(struct sock *sk, struct sk_buff *skb, struct sk_buff /* printk(KERN_DEBUG "check_xmit_queue: %04x, %d\n", acknum, cb2->xmit_count); */ + /* Does _last_ packet acked have xmit_count > 1 */ + try_retrans = 0; + /* Remember to wake up the sending process */ wakeup = 1; + /* Keep various statistics */ pkttime = cb2->stamp; xmit_count = cb2->xmit_count; segnum = cb2->segnum; + /* Remove and drop ack'ed packet */ skb_unlink(ack); kfree_skb(ack); ack = NULL; + + /* + * We don't expect to see acknowledgements for packets we + * haven't sent yet. + */ + if (xmit_count == 0) + BUG(); + /* + * If the packet has only been sent once, we can use it + * to calculate the RTT and also open the window a little + * further. + */ if (xmit_count == 1) { if (equal(segnum, acknum)) dn_nsp_rtt(sk, (long)(pkttime - reftime)); - if (scp->snd_window < NSP_MAX_WINDOW) + if (scp->snd_window < scp->max_window) scp->snd_window++; } + + /* + * Packet has been sent more than once. If this is the last + * packet to be acknowledged then we want to send the next + * packet in the send queue again (assumes the remote host does + * go-back-N error control). + */ + if (xmit_count > 1) + try_retrans = 1; } -#if 0 /* Turned off due to possible interference in socket shutdown */ - if ((skb_queue_len(&scp->data_xmit_queue) == 0) && - (skb_queue_len(&scp->other_xmit_queue) == 0)) - scp->persist = 0; -#endif + if (try_retrans) + dn_nsp_output(sk); return wakeup; } @@ -400,47 +465,31 @@ int dn_nsp_check_xmit_queue(struct sock *sk, struct sk_buff *skb, struct sk_buff void dn_nsp_send_data_ack(struct sock *sk) { struct sk_buff *skb = NULL; - struct nsp_data_ack_msg *msg; - if ((skb = dn_alloc_skb(sk, 200, GFP_ATOMIC)) == NULL) + if ((skb = dn_alloc_skb(sk, 9, GFP_ATOMIC)) == NULL) return; - - msg = (struct nsp_data_ack_msg *)skb_put(skb,sizeof(*msg)); - - msg->msgflg = 0x04; /* data ack message */ - msg->dstaddr = sk->protinfo.dn.addrrem; - msg->srcaddr = sk->protinfo.dn.addrloc; - msg->acknum = dn_htons((sk->protinfo.dn.numdat_rcv & 0x0FFF) | 0x8000); - - sk->protinfo.dn.ackxmt_dat = sk->protinfo.dn.numdat_rcv; + skb_reserve(skb, 9); + dn_mk_ack_header(sk, skb, 0x04, 9, 0); dn_nsp_send(skb); } void dn_nsp_send_oth_ack(struct sock *sk) { struct sk_buff *skb = NULL; - struct nsp_data_ack_msg *msg; - if ((skb = dn_alloc_skb(sk, 200, GFP_ATOMIC)) == NULL) + if ((skb = dn_alloc_skb(sk, 9, GFP_ATOMIC)) == NULL) return; - - msg = (struct nsp_data_ack_msg *)skb_put(skb,sizeof(*msg)); - - msg->msgflg = 0x14; /* oth ack message */ - msg->dstaddr = sk->protinfo.dn.addrrem; - msg->srcaddr = sk->protinfo.dn.addrloc; - msg->acknum = dn_htons((sk->protinfo.dn.numoth_rcv & 0x0FFF) | 0x8000); - - sk->protinfo.dn.ackxmt_oth = sk->protinfo.dn.numoth_rcv; + skb_reserve(skb, 9); + dn_mk_ack_header(sk, skb, 0x14, 9, 1); dn_nsp_send(skb); } void dn_send_conn_ack (struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); struct sk_buff *skb = NULL; struct nsp_conn_ack_msg *msg; @@ -456,7 +505,7 @@ void dn_send_conn_ack (struct sock *sk) void dn_nsp_delayed_ack(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); if (scp->ackxmt_oth != scp->numoth_rcv) dn_nsp_send_oth_ack(sk); @@ -467,7 +516,7 @@ void dn_nsp_delayed_ack(struct sock *sk) static int dn_nsp_retrans_conn_conf(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); if (scp->state == DN_CC) dn_send_conn_conf(sk, GFP_ATOMIC); @@ -477,7 +526,7 @@ static int dn_nsp_retrans_conn_conf(struct sock *sk) void dn_send_conn_conf(struct sock *sk, int gfp) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); struct sk_buff *skb = NULL; struct nsp_conn_init_msg *msg; unsigned char len = scp->conndata_out.opt_optl; @@ -489,9 +538,9 @@ void dn_send_conn_conf(struct sock *sk, int gfp) msg->msgflg = 0x28; msg->dstaddr = scp->addrrem; msg->srcaddr = scp->addrloc; - msg->services = 0x01; - msg->info = 0x03; - msg->segsize = dn_htons(0x05B3); + msg->services = scp->services_loc; + msg->info = scp->info_loc; + msg->segsize = dn_htons(scp->segsize_loc); *skb_put(skb,1) = len; @@ -551,7 +600,7 @@ static __inline__ void dn_nsp_do_disc(struct sock *sk, unsigned char msgflg, void dn_nsp_send_disc(struct sock *sk, unsigned char msgflg, unsigned short reason, int gfp) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); int ddl = 0; if (msgflg == NSP_DISCINIT) @@ -568,7 +617,7 @@ void dn_nsp_send_disc(struct sock *sk, unsigned char msgflg, void dn_nsp_return_disc(struct sk_buff *skb, unsigned char msgflg, unsigned short reason) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); int ddl = 0; int gfp = GFP_ATOMIC; @@ -577,38 +626,35 @@ void dn_nsp_return_disc(struct sk_buff *skb, unsigned char msgflg, } -void dn_nsp_send_lnk(struct sock *sk, unsigned short flgs) +void dn_nsp_send_link(struct sock *sk, unsigned char lsflags, char fcval) { - struct dn_scp *scp = &sk->protinfo.dn; - struct sk_buff *skb = NULL; - struct nsp_data_seg_msg *msg; - struct nsp_data_opt_msg *msg1; - struct dn_skb_cb *cb; + struct dn_scp *scp = DN_SK(sk); + struct sk_buff *skb; + unsigned short *segnum; + unsigned char *ptr; + int gfp = GFP_ATOMIC; - if ((skb = dn_alloc_skb(sk, 80, GFP_ATOMIC)) == NULL) + if ((skb = dn_alloc_skb(sk, 13, gfp)) == NULL) return; - cb = (struct dn_skb_cb *)skb->cb; - msg = (struct nsp_data_seg_msg *)skb_put(skb, sizeof(*msg)); - msg->msgflg = 0x10; /* Link svc message */ - msg->dstaddr = scp->addrrem; - msg->srcaddr = scp->addrloc; + skb_reserve(skb, 13); + segnum = dn_mk_ack_header(sk, skb, 0x10, 13, 1); + *segnum = dn_htons(scp->numoth); + DN_SKB_CB(skb)->segnum = scp->numoth; + seq_add(&scp->numoth, 1); + ptr = (unsigned char *)(segnum + 1); + *ptr++ = lsflags; + *ptr = fcval; - msg1 = (struct nsp_data_opt_msg *)skb_put(skb, sizeof(*msg1)); - msg1->acknum = dn_htons((scp->ackxmt_oth & 0x0FFF) | 0x8000); - msg1->segnum = dn_htons(cb->segnum = (scp->numoth++ & 0x0FFF)); - msg1->lsflgs = flgs; - - dn_nsp_queue_xmit(sk, skb, 1); + dn_nsp_queue_xmit(sk, skb, gfp, 1); scp->persist = dn_nsp_persist(sk); scp->persist_fxn = dn_nsp_xmit_timeout; - } static int dn_nsp_retrans_conninit(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); if (scp->state == DN_CI) dn_nsp_send_conninit(sk, NSP_RCI); @@ -618,7 +664,7 @@ static int dn_nsp_retrans_conninit(struct sock *sk) void dn_nsp_send_conninit(struct sock *sk, unsigned char msgflg) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); struct sk_buff *skb = NULL; struct nsp_conn_init_msg *msg; unsigned char aux; @@ -629,16 +675,16 @@ void dn_nsp_send_conninit(struct sock *sk, unsigned char msgflg) if ((skb = dn_alloc_skb(sk, 200, (msgflg == NSP_CI) ? sk->allocation : GFP_ATOMIC)) == NULL) return; - cb = (struct dn_skb_cb *)skb->cb; + cb = DN_SKB_CB(skb); msg = (struct nsp_conn_init_msg *)skb_put(skb,sizeof(*msg)); msg->msgflg = msgflg; msg->dstaddr = 0x0000; /* Remote Node will assign it*/ - msg->srcaddr = sk->protinfo.dn.addrloc; - msg->services = 1 | NSP_FC_NONE; /* Requested flow control */ - msg->info = 0x03; /* Version Number */ - msg->segsize = dn_htons(1459); /* Max segment size */ + msg->srcaddr = scp->addrloc; + msg->services = scp->services_loc; /* Requested flow control */ + msg->info = scp->info_loc; /* Version Number */ + msg->segsize = dn_htons(scp->segsize_loc); /* Max segment size */ if (scp->peer.sdn_objnum) type = 0; @@ -674,8 +720,8 @@ void dn_nsp_send_conninit(struct sock *sk, unsigned char msgflg) if (aux > 0) memcpy(skb_put(skb,aux), scp->conndata_out.opt_data, aux); - sk->protinfo.dn.persist = dn_nsp_persist(sk); - sk->protinfo.dn.persist_fxn = dn_nsp_retrans_conninit; + scp->persist = dn_nsp_persist(sk); + scp->persist_fxn = dn_nsp_retrans_conninit; cb->rt_flags = DN_RT_F_RQR; diff --git a/net/decnet/dn_route.c b/net/decnet/dn_route.c index 70646fc11..b40c601b6 100644 --- a/net/decnet/dn_route.c +++ b/net/decnet/dn_route.c @@ -33,6 +33,9 @@ * Steve Whitehouse : Real SMP at last :-) Also new netfilter * stuff. Look out raw sockets your days * are numbered! + * Steve Whitehouse : Added return-to-sender functions. Added + * backlog congestion level return codes. + * */ /****************************************************************************** @@ -109,17 +112,16 @@ static struct timer_list dn_rt_flush_timer = { function: dn_run_flush }; int decnet_dst_gc_interval = 2; static struct dst_ops dn_dst_ops = { - PF_DECnet, - __constant_htons(ETH_P_DNA_RT), - 128, - dn_dst_gc, - dn_dst_check, - dn_dst_reroute, - NULL, - dn_dst_negative_advice, - dn_dst_link_failure, - sizeof(struct dn_route), - ATOMIC_INIT(0) + family: PF_DECnet, + protocol: __constant_htons(ETH_P_DNA_RT), + gc_thresh: 128, + gc: dn_dst_gc, + check: dn_dst_check, + reroute: dn_dst_reroute, + negative_advice: dn_dst_negative_advice, + link_failure: dn_dst_link_failure, + entry_size: sizeof(struct dn_route), + entries: ATOMIC_INIT(0), }; static __inline__ unsigned dn_hash(unsigned short src, unsigned short dst) @@ -294,21 +296,131 @@ void dn_rt_cache_flush(int delay) spin_unlock_bh(&dn_rt_flush_lock); } +/** + * dn_return_short - Return a short packet to its sender + * @skb: The packet to return + * + */ +static int dn_return_short(struct sk_buff *skb) +{ + struct dn_skb_cb *cb; + unsigned char *ptr; + dn_address *src; + dn_address *dst; + dn_address tmp; + + /* Add back headers */ + skb_push(skb, skb->data - skb->nh.raw); + + if ((skb = skb_unshare(skb, GFP_ATOMIC)) == NULL) + return NET_RX_DROP; + + cb = DN_SKB_CB(skb); + /* Skip packet length and point to flags */ + ptr = skb->data + 2; + *ptr++ = (cb->rt_flags & ~DN_RT_F_RQR) | DN_RT_F_RTS; + + dst = (dn_address *)ptr; + ptr += 2; + src = (dn_address *)ptr; + ptr += 2; + *ptr = 0; /* Zero hop count */ + + /* Swap source and destination */ + tmp = *src; + *src = *dst; + *dst = tmp; + + skb->pkt_type = PACKET_OUTGOING; + dn_rt_finish_output(skb, NULL); + return NET_RX_SUCCESS; +} + +/** + * dn_return_long - Return a long packet to its sender + * @skb: The long format packet to return + * + */ +static int dn_return_long(struct sk_buff *skb) +{ + struct dn_skb_cb *cb; + unsigned char *ptr; + unsigned char *src_addr, *dst_addr; + unsigned char tmp[ETH_ALEN]; + + /* Add back all headers */ + skb_push(skb, skb->data - skb->nh.raw); + + if ((skb = skb_unshare(skb, GFP_ATOMIC)) == NULL) + return NET_RX_DROP; + cb = DN_SKB_CB(skb); + /* Ignore packet length and point to flags */ + ptr = skb->data + 2; + + /* Skip padding */ + if (*ptr & DN_RT_F_PF) { + char padlen = (*ptr & ~DN_RT_F_PF); + ptr += padlen; + } + + *ptr++ = (cb->rt_flags & ~DN_RT_F_RQR) | DN_RT_F_RTS; + ptr += 2; + dst_addr = ptr; + ptr += 8; + src_addr = ptr; + ptr += 6; + *ptr = 0; /* Zero hop count */ + + /* Swap source and destination */ + memcpy(tmp, src_addr, ETH_ALEN); + memcpy(src_addr, dst_addr, ETH_ALEN); + memcpy(dst_addr, tmp, ETH_ALEN); + + skb->pkt_type = PACKET_OUTGOING; + dn_rt_finish_output(skb, tmp); + return NET_RX_SUCCESS; +} + +/** + * dn_route_rx_packet - Try and find a route for an incoming packet + * @skb: The packet to find a route for + * + * Returns: result of input function if route is found, error code otherwise + */ static int dn_route_rx_packet(struct sk_buff *skb) { + struct dn_skb_cb *cb = DN_SKB_CB(skb); int err; if ((err = dn_route_input(skb)) == 0) return skb->dst->input(skb); + if (decnet_debug_level & 4) { + char *devname = skb->dev ? skb->dev->name : "???"; + struct dn_skb_cb *cb = DN_SKB_CB(skb); + printk(KERN_DEBUG + "DECnet: dn_route_rx_packet: rt_flags=0x%02x dev=%s len=%d src=0x%04hx dst=0x%04hx err=%d type=%d\n", + (int)cb->rt_flags, devname, skb->len, cb->src, cb->dst, + err, skb->pkt_type); + } + + if ((skb->pkt_type == PACKET_HOST) && (cb->rt_flags & DN_RT_F_RQR)) { + switch(cb->rt_flags & DN_RT_PKT_MSK) { + case DN_RT_PKT_SHORT: + return dn_return_short(skb); + case DN_RT_PKT_LONG: + return dn_return_long(skb); + } + } + kfree_skb(skb); - return err; + return NET_RX_DROP; } static int dn_route_rx_long(struct sk_buff *skb) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); unsigned char *ptr = skb->data; if (skb->len < 21) /* 20 for long header, 1 for shortest nsp */ @@ -339,14 +451,14 @@ static int dn_route_rx_long(struct sk_buff *skb) drop_it: kfree_skb(skb); - return 0; + return NET_RX_DROP; } static int dn_route_rx_short(struct sk_buff *skb) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); unsigned char *ptr = skb->data; if (skb->len < 6) /* 5 for short header + 1 for shortest nsp */ @@ -365,29 +477,33 @@ static int dn_route_rx_short(struct sk_buff *skb) drop_it: kfree_skb(skb); - return 0; + return NET_RX_DROP; } static int dn_route_discard(struct sk_buff *skb) { + /* + * I know we drop the packet here, but thats considered success in + * this case + */ kfree_skb(skb); - return 0; + return NET_RX_SUCCESS; } static int dn_route_ptp_hello(struct sk_buff *skb) { dn_dev_hello(skb); dn_neigh_pointopoint_hello(skb); - return 0; + return NET_RX_SUCCESS; } int dn_route_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt) { struct dn_skb_cb *cb; unsigned char flags = 0; - int padlen = 0; __u16 len = dn_ntohs(*(__u16 *)skb->data); struct dn_dev *dn = (struct dn_dev *)dev->dn_ptr; + unsigned char padlen = 0; if (dn == NULL) goto dump_it; @@ -404,7 +520,7 @@ int dn_route_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type flags = *skb->data; - cb = (struct dn_skb_cb *)skb->cb; + cb = DN_SKB_CB(skb); cb->stamp = jiffies; cb->iif = dev->ifindex; @@ -448,20 +564,16 @@ int dn_route_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type switch(flags & DN_RT_CNTL_MSK) { case DN_RT_PKT_HELO: - NF_HOOK(PF_DECnet, NF_DN_HELLO, skb, skb->dev, NULL, dn_route_ptp_hello); - goto out; + return NF_HOOK(PF_DECnet, NF_DN_HELLO, skb, skb->dev, NULL, dn_route_ptp_hello); case DN_RT_PKT_L1RT: case DN_RT_PKT_L2RT: - NF_HOOK(PF_DECnet, NF_DN_ROUTE, skb, skb->dev, NULL, dn_route_discard); - goto out; + return NF_HOOK(PF_DECnet, NF_DN_ROUTE, skb, skb->dev, NULL, dn_route_discard); case DN_RT_PKT_ERTH: - NF_HOOK(PF_DECnet, NF_DN_HELLO, skb, skb->dev, NULL, dn_neigh_router_hello); - goto out; + return NF_HOOK(PF_DECnet, NF_DN_HELLO, skb, skb->dev, NULL, dn_neigh_router_hello); case DN_RT_PKT_EEDH: - NF_HOOK(PF_DECnet, NF_DN_HELLO, skb, skb->dev, NULL, dn_neigh_endnode_hello); - goto out; + return NF_HOOK(PF_DECnet, NF_DN_HELLO, skb, skb->dev, NULL, dn_neigh_endnode_hello); } } else { if (dn->parms.state != DN_DEV_S_RU) @@ -480,7 +592,7 @@ int dn_route_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type dump_it: kfree_skb(skb); out: - return 0; + return NET_RX_DROP; } static int dn_output(struct sk_buff *skb) @@ -488,7 +600,7 @@ static int dn_output(struct sk_buff *skb) struct dst_entry *dst = skb->dst; struct dn_route *rt = (struct dn_route *)dst; struct net_device *dev = dst->dev; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); struct neighbour *neigh; int err = -EINVAL; @@ -524,7 +636,7 @@ error: #ifdef CONFIG_DECNET_ROUTER static int dn_forward(struct sk_buff *skb) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); struct dst_entry *dst = skb->dst; struct net_device *dev = skb->dev; struct neighbour *neigh; @@ -536,7 +648,7 @@ static int dn_forward(struct sk_buff *skb) /* * Hop count exceeded. */ - err = 0; + err = NET_RX_DROP; if (++cb->hops > 30) goto drop; @@ -573,7 +685,7 @@ drop: static int dn_blackhole(struct sk_buff *skb) { kfree_skb(skb); - return 0; + return NET_RX_DROP; } /* @@ -583,7 +695,7 @@ static int dn_blackhole(struct sk_buff *skb) static int dn_rt_bug(struct sk_buff *skb) { if (net_ratelimit()) { - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); printk(KERN_DEBUG "dn_rt_bug: skb from:%04x to:%04x\n", cb->src, cb->dst); @@ -591,7 +703,7 @@ static int dn_rt_bug(struct sk_buff *skb) kfree_skb(skb); - return -EINVAL; + return NET_RX_BAD; } static int dn_route_output_slow(struct dst_entry **pprt, dn_address dst, dn_address src, int flags) @@ -732,7 +844,7 @@ int dn_route_output(struct dst_entry **pprt, dn_address dst, dn_address src, int static int dn_route_input_slow(struct sk_buff *skb) { struct dn_route *rt = NULL; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); struct net_device *dev = skb->dev; struct dn_dev *dn_db; struct neighbour *neigh = NULL; @@ -880,7 +992,7 @@ add_entry: int dn_route_input(struct sk_buff *skb) { struct dn_route *rt; - struct dn_skb_cb *cb = (struct dn_skb_cb *)skb->cb; + struct dn_skb_cb *cb = DN_SKB_CB(skb); unsigned hash = dn_hash(cb->src, cb->dst); if (skb->dst) @@ -964,7 +1076,7 @@ int dn_cache_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, void *arg) if (skb == NULL) return -ENOBUFS; skb->mac.raw = skb->data; - cb = (struct dn_skb_cb *)skb->cb; + cb = DN_SKB_CB(skb); if (rta[RTA_SRC-1]) memcpy(&src, RTA_DATA(rta[RTA_SRC-1]), 2); @@ -1185,8 +1297,6 @@ void __exit dn_route_cleanup(void) del_timer(&dn_route_timer); dn_run_flush(0); -#ifdef CONFIG_PROC_FS proc_net_remove("decnet_cache"); -#endif /* CONFIG_PROC_FS */ } diff --git a/net/decnet/dn_rules.c b/net/decnet/dn_rules.c index 133591f0b..32adfecba 100644 --- a/net/decnet/dn_rules.c +++ b/net/decnet/dn_rules.c @@ -57,7 +57,12 @@ struct dn_fib_rule int r_dead; }; -static struct dn_fib_rule default_rule = { NULL, ATOMIC_INIT(2), 0x7fff, DN_DEFAULT_TABLE, RTN_UNICAST }; +static struct dn_fib_rule default_rule = { + r_clntref: ATOMIC_INIT(2), + r_preference: 0x7fff, + r_table: DN_DEFAULT_TABLE, + r_action: RTN_UNICAST +}; static struct dn_fib_rule *dn_fib_rules = &default_rule; static rwlock_t dn_fib_rules_lock = RW_LOCK_UNLOCKED; @@ -291,9 +296,7 @@ static int dn_fib_rules_event(struct notifier_block *this, unsigned long event, static struct notifier_block dn_fib_rules_notifier = { - dn_fib_rules_event, - NULL, - 0 + notifier_call: dn_fib_rules_event, }; #ifdef CONFIG_RTNETLINK diff --git a/net/decnet/dn_table.c b/net/decnet/dn_table.c index ce7a3ac06..bc21acb4b 100644 --- a/net/decnet/dn_table.c +++ b/net/decnet/dn_table.c @@ -78,7 +78,7 @@ static rwlock_t dn_fib_tables_lock = RW_LOCK_UNLOCKED; static struct dn_fib_table *dn_fib_tables[DN_NUM_TABLES + 1]; static kmem_cache_t *dn_hash_kmem; -static int dn_fib_hash_zombies = 0; +static int dn_fib_hash_zombies; static __inline__ dn_fib_idx_t dn_hash(dn_fib_key_t key, struct dn_zone *dz) { diff --git a/net/decnet/dn_timer.c b/net/decnet/dn_timer.c index bbba58b02..41a4aa602 100644 --- a/net/decnet/dn_timer.c +++ b/net/decnet/dn_timer.c @@ -52,7 +52,7 @@ void dn_stop_slow_timer(struct sock *sk) static void dn_slow_timer(unsigned long arg) { struct sock *sk = (struct sock *)arg; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); sock_hold(sk); bh_lock_sock(sk); @@ -112,7 +112,7 @@ out: static void dn_fast_timer(unsigned long arg) { struct sock *sk = (struct sock *)arg; - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); bh_lock_sock(sk); if (sk->lock.users != 0) { @@ -131,7 +131,7 @@ out: void dn_start_fast_timer(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); if (!scp->delack_pending) { scp->delack_pending = 1; @@ -145,7 +145,7 @@ void dn_start_fast_timer(struct sock *sk) void dn_stop_fast_timer(struct sock *sk) { - struct dn_scp *scp = &sk->protinfo.dn; + struct dn_scp *scp = DN_SK(sk); if (scp->delack_pending) { scp->delack_pending = 0; diff --git a/net/ipv4/igmp.c b/net/ipv4/igmp.c index 1d8002bdd..81fae9233 100644 --- a/net/ipv4/igmp.c +++ b/net/ipv4/igmp.c @@ -504,8 +504,8 @@ void ip_mc_inc_group(struct in_device *in_dev, u32 addr) im->timer.function=&igmp_timer_expire; im->unsolicit_count = IGMP_Unsolicited_Report_Count; im->reporter = 0; - im->loaded = 0; #endif + im->loaded = 0; write_lock_bh(&in_dev->lock); im->next=in_dev->mc_list; in_dev->mc_list=im; diff --git a/net/ipv4/netfilter/Config.in b/net/ipv4/netfilter/Config.in index 406d2ea3d..5887658fb 100644 --- a/net/ipv4/netfilter/Config.in +++ b/net/ipv4/netfilter/Config.in @@ -37,11 +37,20 @@ if [ "$CONFIG_IP_NF_IPTABLES" != "n" ]; then fi if [ "$CONFIG_IP_NF_CONNTRACK" != "n" ]; then - dep_tristate ' Full NAT' CONFIG_IP_NF_NAT $CONFIG_IP_NF_IPTABLES + dep_tristate ' Full NAT' CONFIG_IP_NF_NAT $CONFIG_IP_NF_IPTABLES $CONFIG_IP_NF_CONNTRACK if [ "$CONFIG_IP_NF_NAT" != "n" ]; then define_bool CONFIG_IP_NF_NAT_NEEDED y dep_tristate ' MASQUERADE target support' CONFIG_IP_NF_TARGET_MASQUERADE $CONFIG_IP_NF_NAT dep_tristate ' REDIRECT target support' CONFIG_IP_NF_TARGET_REDIRECT $CONFIG_IP_NF_NAT + # If they want FTP, set to $CONFIG_IP_NF_NAT (m or y), + # or $CONFIG_IP_NF_FTP (m or y), whichever is weaker. Argh. + if [ "$CONFIG_IP_NF_FTP" = "m" ]; then + define_tristate CONFIG_IP_NF_NAT_FTP m + else + if [ "$CONFIG_IP_NF_FTP" = "y" ]; then + define_tristate CONFIG_IP_NF_NAT_FTP $CONFIG_IP_NF_NAT + fi + fi fi fi diff --git a/net/ipv4/netfilter/Makefile b/net/ipv4/netfilter/Makefile index 995860767..c40caa75e 100644 --- a/net/ipv4/netfilter/Makefile +++ b/net/ipv4/netfilter/Makefile @@ -35,7 +35,7 @@ obj-$(CONFIG_IP_NF_CONNTRACK) += ip_conntrack.o obj-$(CONFIG_IP_NF_FTP) += ip_conntrack_ftp.o # NAT helpers -obj-$(CONFIG_IP_NF_FTP) += ip_nat_ftp.o +obj-$(CONFIG_IP_NF_NAT_FTP) += ip_nat_ftp.o # generic IP tables obj-$(CONFIG_IP_NF_IPTABLES) += ip_tables.o diff --git a/net/ipv4/netfilter/ip_conntrack_core.c b/net/ipv4/netfilter/ip_conntrack_core.c index 2e4dd82ee..bc7e64c8b 100644 --- a/net/ipv4/netfilter/ip_conntrack_core.c +++ b/net/ipv4/netfilter/ip_conntrack_core.c @@ -882,10 +882,15 @@ ip_ct_gather_frags(struct sk_buff *skb) #ifdef CONFIG_NETFILTER_DEBUG unsigned int olddebug = skb->nf_debug; #endif - if (sk) sock_hold(sk); + if (sk) { + sock_hold(sk); + skb_orphan(skb); + } + local_bh_disable(); skb = ip_defrag(skb); - local_bh_enable(); + local_bh_enable(); + if (!skb) { if (sk) sock_put(sk); return skb; diff --git a/net/ipv4/netfilter/ip_nat_core.c b/net/ipv4/netfilter/ip_nat_core.c index 9ba62dc84..cc19e1f0b 100644 --- a/net/ipv4/netfilter/ip_nat_core.c +++ b/net/ipv4/netfilter/ip_nat_core.c @@ -438,8 +438,27 @@ get_unique_tuple(struct ip_conntrack_tuple *tuple, conntrack)); ret = 1; goto clear_fulls; + } else if (HOOK2MANIP(hooknum) == IP_NAT_MANIP_DST) { + /* Try implicit source NAT; protocol + may be able to play with ports to + make it unique. */ + struct ip_nat_range r + = { IP_NAT_RANGE_MAP_IPS, + tuple->src.ip, tuple->src.ip, + { 0 }, { 0 } }; + DEBUGP("Trying implicit mapping\n"); + if (proto->unique_tuple(tuple, &r, + IP_NAT_MANIP_SRC, + conntrack)) { + /* Must be unique. */ + IP_NF_ASSERT(!ip_nat_used_tuple + (tuple, conntrack)); + ret = 1; + goto clear_fulls; + } } - DEBUGP("Protocol can't get unique tuple.\n"); + DEBUGP("Protocol can't get unique tuple %u.\n", + hooknum); } /* Eliminate that from range, and try again. */ diff --git a/net/ipv4/netfilter/ipt_MASQUERADE.c b/net/ipv4/netfilter/ipt_MASQUERADE.c index 99164a7a0..f2a19702d 100644 --- a/net/ipv4/netfilter/ipt_MASQUERADE.c +++ b/net/ipv4/netfilter/ipt_MASQUERADE.c @@ -1,5 +1,6 @@ /* Masquerade. Simple mapping which alters range to a local IP address (depending on route). */ +#include <linux/config.h> #include <linux/types.h> #include <linux/ip.h> #include <linux/timer.h> @@ -68,6 +69,7 @@ masquerade_target(struct sk_buff **pskb, struct ip_nat_multi_range newrange; u_int32_t newsrc; struct rtable *rt; + struct rt_key key; IP_NF_ASSERT(hooknum == NF_IP_POST_ROUTING); @@ -82,10 +84,14 @@ masquerade_target(struct sk_buff **pskb, mr = targinfo; - if (ip_route_output(&rt, (*pskb)->nh.iph->daddr, - 0, - RT_TOS((*pskb)->nh.iph->tos)|RTO_CONN, - out->ifindex) != 0) { + key.dst = (*pskb)->nh.iph->daddr; + key.src = 0; /* Unknown: that's what we're trying to establish */ + key.tos = RT_TOS((*pskb)->nh.iph->tos)|RTO_CONN; + key.oif = out->ifindex; +#ifdef CONFIG_IP_ROUTE_FWMARK + key.fwmark = (*pskb)->nfmark; +#endif + if (ip_route_output_key(&rt, &key) != 0) { /* Shouldn't happen */ printk("MASQUERADE: No route: Rusty's brain broke!\n"); return NF_DROP; diff --git a/net/ipv4/netfilter/ipt_REJECT.c b/net/ipv4/netfilter/ipt_REJECT.c index 9c1088e76..cc5ffbc4a 100644 --- a/net/ipv4/netfilter/ipt_REJECT.c +++ b/net/ipv4/netfilter/ipt_REJECT.c @@ -167,27 +167,9 @@ static unsigned int reject(struct sk_buff **pskb, case IPT_ICMP_HOST_PROHIBITED: icmp_send(*pskb, ICMP_DEST_UNREACH, ICMP_HOST_ANO, 0); break; - case IPT_ICMP_ECHOREPLY: { - struct icmphdr *icmph = (struct icmphdr *) - ((u_int32_t *)(*pskb)->nh.iph + (*pskb)->nh.iph->ihl); - unsigned int datalen = (*pskb)->len - (*pskb)->nh.iph->ihl * 4; - - /* Not non-head frags, or truncated */ - if (((ntohs((*pskb)->nh.iph->frag_off) & IP_OFFSET) == 0) - && datalen >= 4) { - /* Usually I don't like cut & pasting code, - but dammit, my party is starting in 45 - mins! --RR */ - struct icmp_bxm icmp_param; - - icmp_param.icmph=*icmph; - icmp_param.icmph.type=ICMP_ECHOREPLY; - icmp_param.data_ptr=(icmph+1); - icmp_param.data_len=datalen; - icmp_reply(&icmp_param, *pskb); - } - } - break; + case IPT_ICMP_ECHOREPLY: + printk("REJECT: ECHOREPLY no longer supported.\n"); + break; case IPT_TCP_RESET: send_reset(*pskb, hooknum == NF_IP_LOCAL_IN); break; diff --git a/net/ipv4/netfilter/iptable_mangle.c b/net/ipv4/netfilter/iptable_mangle.c index c52ada64e..60d4698fb 100644 --- a/net/ipv4/netfilter/iptable_mangle.c +++ b/net/ipv4/netfilter/iptable_mangle.c @@ -53,7 +53,7 @@ static struct sizeof(struct ipt_entry), sizeof(struct ipt_standard), 0, { 0, 0 }, { } }, - { { { { sizeof(struct ipt_standard_target), "" } }, { } }, + { { { { IPT_ALIGN(sizeof(struct ipt_standard_target)), "" } }, { } }, -NF_ACCEPT - 1 } }, /* LOCAL_OUT */ { { { { 0 }, { 0 }, { 0 }, { 0 }, "", "", { 0 }, { 0 }, 0, 0, 0 }, @@ -61,7 +61,7 @@ static struct sizeof(struct ipt_entry), sizeof(struct ipt_standard), 0, { 0, 0 }, { } }, - { { { { sizeof(struct ipt_standard_target), "" } }, { } }, + { { { { IPT_ALIGN(sizeof(struct ipt_standard_target)), "" } }, { } }, -NF_ACCEPT - 1 } } }, /* ERROR */ @@ -70,7 +70,7 @@ static struct sizeof(struct ipt_entry), sizeof(struct ipt_error), 0, { 0, 0 }, { } }, - { { { { sizeof(struct ipt_error_target), IPT_ERROR_TARGET } }, + { { { { IPT_ALIGN(sizeof(struct ipt_error_target)), IPT_ERROR_TARGET } }, { } }, "ERROR" } diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index b370fcdf9..6660e0f72 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -954,7 +954,7 @@ int tcp_sendmsg(struct sock *sk, struct msghdr *msg, int size) */ skb = sk->write_queue.prev; if (tp->send_head && - (mss_now - skb->len) > 0) { + (mss_now > skb->len)) { copy = skb->len; if (skb_tailroom(skb) > 0) { int last_byte_was_odd = (copy % 4); diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 4e3eab087..a3f83272b 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -1705,7 +1705,7 @@ static __inline__ void tcp_ack_packets_out(struct sock *sk, struct tcp_opt *tp) if ((__s32)when < (__s32)tp->rttvar) when = tp->rttvar; - tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, when); + tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, min(when, TCP_RTO_MAX)); } } diff --git a/net/ipv6/netfilter/Makefile b/net/ipv6/netfilter/Makefile index aec5db879..fcc9bc058 100644 --- a/net/ipv6/netfilter/Makefile +++ b/net/ipv6/netfilter/Makefile @@ -19,6 +19,7 @@ obj-$(CONFIG_IP6_NF_MATCH_MARK) += ip6t_mark.o obj-$(CONFIG_IP6_NF_MATCH_MAC) += ip6t_mac.o obj-$(CONFIG_IP6_NF_MATCH_MULTIPORT) += ip6t_multiport.o obj-$(CONFIG_IP6_NF_FILTER) += ip6table_filter.o +obj-$(CONFIG_IP6_NF_MANGLE) += ip6table_mangle.o obj-$(CONFIG_IP6_NF_TARGET_MARK) += ip6t_MARK.o include $(TOPDIR)/Rules.make diff --git a/net/ipv6/netfilter/ip6_tables.c b/net/ipv6/netfilter/ip6_tables.c index 57430f29f..659bb3a1e 100644 --- a/net/ipv6/netfilter/ip6_tables.c +++ b/net/ipv6/netfilter/ip6_tables.c @@ -11,7 +11,7 @@ #include <linux/module.h> #include <linux/tcp.h> #include <linux/udp.h> -#include <linux/icmp.h> +#include <linux/icmpv6.h> #include <net/ip.h> #include <asm/uaccess.h> #include <asm/semaphore.h> @@ -1642,7 +1642,7 @@ udp_checkentry(const char *tablename, /* Returns 1 if the type and code is matched by the range, 0 otherwise */ static inline int -icmp_type_code_match(u_int8_t test_type, u_int8_t min_code, u_int8_t max_code, +icmp6_type_code_match(u_int8_t test_type, u_int8_t min_code, u_int8_t max_code, u_int8_t type, u_int8_t code, int invert) { @@ -1651,7 +1651,7 @@ icmp_type_code_match(u_int8_t test_type, u_int8_t min_code, u_int8_t max_code, } static int -icmp_match(const struct sk_buff *skb, +icmp6_match(const struct sk_buff *skb, const struct net_device *in, const struct net_device *out, const void *matchinfo, @@ -1660,7 +1660,7 @@ icmp_match(const struct sk_buff *skb, u_int16_t datalen, int *hotdrop) { - const struct icmphdr *icmp = hdr; + const struct icmp6hdr *icmp = hdr; const struct ip6t_icmp *icmpinfo = matchinfo; if (offset == 0 && datalen < 2) { @@ -1673,16 +1673,16 @@ icmp_match(const struct sk_buff *skb, /* Must not be a fragment. */ return !offset - && icmp_type_code_match(icmpinfo->type, + && icmp6_type_code_match(icmpinfo->type, icmpinfo->code[0], icmpinfo->code[1], - icmp->type, icmp->code, + icmp->icmp6_type, icmp->icmp6_code, !!(icmpinfo->invflags&IP6T_ICMP_INV)); } /* Called when user tries to insert an entry of this type. */ static int -icmp_checkentry(const char *tablename, +icmp6_checkentry(const char *tablename, const struct ip6t_ip6 *ipv6, void *matchinfo, unsigned int matchsize, @@ -1691,7 +1691,7 @@ icmp_checkentry(const char *tablename, const struct ip6t_icmp *icmpinfo = matchinfo; /* Must specify proto == ICMP, and no unknown invflags */ - return ipv6->proto == IPPROTO_ICMP + return ipv6->proto == IPPROTO_ICMPV6 && !(ipv6->invflags & IP6T_INV_PROTO) && matchsize == IP6T_ALIGN(sizeof(struct ip6t_icmp)) && !(icmpinfo->invflags & ~IP6T_ICMP_INV); @@ -1711,8 +1711,8 @@ static struct ip6t_match tcp_matchstruct = { { NULL, NULL }, "tcp", &tcp_match, &tcp_checkentry, NULL }; static struct ip6t_match udp_matchstruct = { { NULL, NULL }, "udp", &udp_match, &udp_checkentry, NULL }; -static struct ip6t_match icmp_matchstruct -= { { NULL, NULL }, "icmp", &icmp_match, &icmp_checkentry, NULL }; +static struct ip6t_match icmp6_matchstruct += { { NULL, NULL }, "icmp6", &icmp6_match, &icmp6_checkentry, NULL }; #ifdef CONFIG_PROC_FS static inline int print_name(const struct ip6t_table *t, @@ -1761,7 +1761,7 @@ static int __init init(void) list_append(&ip6t_target, &ip6t_error_target); list_append(&ip6t_match, &tcp_matchstruct); list_append(&ip6t_match, &udp_matchstruct); - list_append(&ip6t_match, &icmp_matchstruct); + list_append(&ip6t_match, &icmp6_matchstruct); up(&ip6t_mutex); /* Register setsockopt */ diff --git a/net/ipv6/netfilter/ip6t_MARK.c b/net/ipv6/netfilter/ip6t_MARK.c index dd8bb3226..08df336e8 100644 --- a/net/ipv6/netfilter/ip6t_MARK.c +++ b/net/ipv6/netfilter/ip6t_MARK.c @@ -4,8 +4,8 @@ #include <linux/ip.h> #include <net/checksum.h> -#include <linux/netfilter_ipv4/ip_tables.h> -#include <linux/netfilter_ipv4/ipt_MARK.h> +#include <linux/netfilter_ipv6/ip6_tables.h> +#include <linux/netfilter_ipv6/ip6t_MARK.h> static unsigned int target(struct sk_buff **pskb, @@ -15,26 +15,26 @@ target(struct sk_buff **pskb, const void *targinfo, void *userinfo) { - const struct ipt_mark_target_info *markinfo = targinfo; + const struct ip6t_mark_target_info *markinfo = targinfo; if((*pskb)->nfmark != markinfo->mark) { (*pskb)->nfmark = markinfo->mark; (*pskb)->nfcache |= NFC_ALTERED; } - return IPT_CONTINUE; + return IP6T_CONTINUE; } static int checkentry(const char *tablename, - const struct ipt_entry *e, + const struct ip6t_entry *e, void *targinfo, unsigned int targinfosize, unsigned int hook_mask) { - if (targinfosize != IPT_ALIGN(sizeof(struct ipt_mark_target_info))) { + if (targinfosize != IP6T_ALIGN(sizeof(struct ip6t_mark_target_info))) { printk(KERN_WARNING "MARK: targinfosize %u != %Zu\n", targinfosize, - IPT_ALIGN(sizeof(struct ipt_mark_target_info))); + IP6T_ALIGN(sizeof(struct ip6t_mark_target_info))); return 0; } @@ -46,12 +46,13 @@ checkentry(const char *tablename, return 1; } -static struct ipt_target ipt_mark_reg +static struct ip6t_target ip6t_mark_reg = { { NULL, NULL }, "MARK", target, checkentry, NULL, THIS_MODULE }; static int __init init(void) { - if (ipt_register_target(&ipt_mark_reg)) + printk(KERN_DEBUG "registreing ipv6 mark target\n"); + if (ip6t_register_target(&ip6t_mark_reg)) return -EINVAL; return 0; @@ -59,7 +60,7 @@ static int __init init(void) static void __exit fini(void) { - ipt_unregister_target(&ipt_mark_reg); + ip6t_unregister_target(&ip6t_mark_reg); } module_init(init); diff --git a/net/ipv6/netfilter/ip6t_mark.c b/net/ipv6/netfilter/ip6t_mark.c index babe202a4..9a78b1ca2 100644 --- a/net/ipv6/netfilter/ip6t_mark.c +++ b/net/ipv6/netfilter/ip6t_mark.c @@ -2,7 +2,7 @@ #include <linux/module.h> #include <linux/skbuff.h> -#include <linux/netfilter_ipv4/ipt_mark.h> +#include <linux/netfilter_ipv6/ip6t_mark.h> #include <linux/netfilter_ipv6/ip6_tables.h> static int @@ -15,7 +15,7 @@ match(const struct sk_buff *skb, u_int16_t datalen, int *hotdrop) { - const struct ipt_mark_info *info = matchinfo; + const struct ip6t_mark_info *info = matchinfo; return ((skb->nfmark & info->mask) == info->mark) ^ info->invert; } @@ -27,7 +27,7 @@ checkentry(const char *tablename, unsigned int matchsize, unsigned int hook_mask) { - if (matchsize != IP6T_ALIGN(sizeof(struct ipt_mark_info))) + if (matchsize != IP6T_ALIGN(sizeof(struct ip6t_mark_info))) return 0; return 1; diff --git a/net/ipv6/netfilter/ip6table_mangle.c b/net/ipv6/netfilter/ip6table_mangle.c new file mode 100644 index 000000000..612c292c4 --- /dev/null +++ b/net/ipv6/netfilter/ip6table_mangle.c @@ -0,0 +1,189 @@ +/* + * IPv6 packet mangling table, a port of the IPv4 mangle table to IPv6 + * + * Copyright (C) 2000 by Harald Welte <laforge@gnumonks.org> + */ +#include <linux/module.h> +#include <linux/netfilter_ipv6/ip6_tables.h> + +#define MANGLE_VALID_HOOKS ((1 << NF_IP6_PRE_ROUTING) | (1 << NF_IP6_LOCAL_OUT)) + +#if 1 +#define DEBUGP(x, args...) printk(KERN_DEBUG x, ## args) +#else +#define DEBUGP(x, args...) +#endif + +/* Standard entry. */ +struct ip6t_standard +{ + struct ip6t_entry entry; + struct ip6t_standard_target target; +}; + +struct ip6t_error_target +{ + struct ip6t_entry_target target; + char errorname[IP6T_FUNCTION_MAXNAMELEN]; +}; + +struct ip6t_error +{ + struct ip6t_entry entry; + struct ip6t_error_target target; +}; + +static struct +{ + struct ip6t_replace repl; + struct ip6t_standard entries[2]; + struct ip6t_error term; +} initial_table __initdata += { { "mangle", MANGLE_VALID_HOOKS, 3, + sizeof(struct ip6t_standard) * 2 + sizeof(struct ip6t_error), + { [NF_IP6_PRE_ROUTING] 0, + [NF_IP6_LOCAL_OUT] sizeof(struct ip6t_standard) }, + { [NF_IP6_PRE_ROUTING] 0, + [NF_IP6_LOCAL_OUT] sizeof(struct ip6t_standard) }, + 0, NULL, { } }, + { + /* PRE_ROUTING */ + { { { { { { 0 } } }, { { { 0 } } }, { { { 0 } } }, { { { 0 } } }, "", "", { 0 }, { 0 }, 0, 0, 0 }, + 0, + sizeof(struct ip6t_entry), + sizeof(struct ip6t_standard), + 0, { 0, 0 }, { } }, + { { { { IP6T_ALIGN(sizeof(struct ip6t_standard_target)), "" } }, { } }, + -NF_ACCEPT - 1 } }, + /* LOCAL_OUT */ + { { { { { { 0 } } }, { { { 0 } } }, { { { 0 } } }, { { { 0 } } }, "", "", { 0 }, { 0 }, 0, 0, 0 }, + 0, + sizeof(struct ip6t_entry), + sizeof(struct ip6t_standard), + 0, { 0, 0 }, { } }, + { { { { IP6T_ALIGN(sizeof(struct ip6t_standard_target)), "" } }, { } }, + -NF_ACCEPT - 1 } } + }, + /* ERROR */ + { { { { { { 0 } } }, { { { 0 } } }, { { { 0 } } }, { { { 0 } } }, "", "", { 0 }, { 0 }, 0, 0, 0 }, + 0, + sizeof(struct ip6t_entry), + sizeof(struct ip6t_error), + 0, { 0, 0 }, { } }, + { { { { IP6T_ALIGN(sizeof(struct ip6t_error_target)), IP6T_ERROR_TARGET } }, + { } }, + "ERROR" + } + } +}; + +static struct ip6t_table packet_mangler += { { NULL, NULL }, "mangle", &initial_table.repl, + MANGLE_VALID_HOOKS, RW_LOCK_UNLOCKED, NULL }; + +/* The work comes in here from netfilter.c. */ +static unsigned int +ip6t_hook(unsigned int hook, + struct sk_buff **pskb, + const struct net_device *in, + const struct net_device *out, + int (*okfn)(struct sk_buff *)) +{ + return ip6t_do_table(pskb, hook, in, out, &packet_mangler, NULL); +} + +static unsigned int +ip6t_local_out_hook(unsigned int hook, + struct sk_buff **pskb, + const struct net_device *in, + const struct net_device *out, + int (*okfn)(struct sk_buff *)) +{ + + unsigned long nfmark; + unsigned int ret; + struct in6_addr saddr, daddr; + u_int8_t hop_limit; + u_int32_t flowlabel; + +#if 0 + /* root is playing with raw sockets. */ + if ((*pskb)->len < sizeof(struct iphdr) + || (*pskb)->nh.iph->ihl * 4 < sizeof(struct iphdr)) { + if (net_ratelimit()) + printk("ip6t_hook: happy cracking.\n"); + return NF_ACCEPT; + } +#endif + + /* save source/dest address, nfmark, hoplimit, flowlabel, priority, */ + memcpy(&saddr, &(*pskb)->nh.ipv6h->saddr, sizeof(saddr)); + memcpy(&daddr, &(*pskb)->nh.ipv6h->daddr, sizeof(daddr)); + nfmark = (*pskb)->nfmark; + hop_limit = (*pskb)->nh.ipv6h->hop_limit; + + /* flowlabel and prio (includes version, which shouldn't change either */ + flowlabel = (u_int32_t) (*pskb)->nh.ipv6h; + + ret = ip6t_do_table(pskb, hook, in, out, &packet_mangler, NULL); + + if (ret != NF_DROP && ret != NF_STOLEN + && (memcmp(&(*pskb)->nh.ipv6h->saddr, &saddr, sizeof(saddr)) + || memcmp(&(*pskb)->nh.ipv6h->daddr, &daddr, sizeof(daddr)) + || (*pskb)->nfmark != nfmark + || (*pskb)->nh.ipv6h->hop_limit != hop_limit)) { + + /* something which could affect routing has changed */ + + DEBUGP("ip6table_mangle: we'd need to re-route a packet\n"); + } + + return ret; +} + +static struct nf_hook_ops ip6t_ops[] += { { { NULL, NULL }, ip6t_hook, PF_INET6, NF_IP6_PRE_ROUTING, NF_IP6_PRI_MANGLE }, + { { NULL, NULL }, ip6t_local_out_hook, PF_INET6, NF_IP6_LOCAL_OUT, + NF_IP6_PRI_MANGLE } +}; + +static int __init init(void) +{ + int ret; + + /* Register table */ + ret = ip6t_register_table(&packet_mangler); + if (ret < 0) + return ret; + + /* Register hooks */ + ret = nf_register_hook(&ip6t_ops[0]); + if (ret < 0) + goto cleanup_table; + + ret = nf_register_hook(&ip6t_ops[1]); + if (ret < 0) + goto cleanup_hook0; + + return ret; + + cleanup_hook0: + nf_unregister_hook(&ip6t_ops[0]); + cleanup_table: + ip6t_unregister_table(&packet_mangler); + + return ret; +} + +static void __exit fini(void) +{ + unsigned int i; + + for (i = 0; i < sizeof(ip6t_ops)/sizeof(struct nf_hook_ops); i++) + nf_unregister_hook(&ip6t_ops[i]); + + ip6t_unregister_table(&packet_mangler); +} + +module_init(init); +module_exit(fini); diff --git a/net/ipx/af_ipx.c b/net/ipx/af_ipx.c index 26cc63348..fb73651bf 100644 --- a/net/ipx/af_ipx.c +++ b/net/ipx/af_ipx.c @@ -61,6 +61,9 @@ * suggestions and guidance. * Arnaldo Carvalho de Melo <acme@conectiva.com.br>, * November, 2000 + * Revision 043: Shared SKBs, don't mangle packets, some cleanups + * Arnaldo Carvalho de Melo <acme@conectiva.com.br>, + * December, 2000 * * Protect the module by a MOD_INC_USE_COUNT/MOD_DEC_USE_COUNT * pair. Also, now usage count is managed this way @@ -140,28 +143,23 @@ static ipx_interface *ipx_internal_net; atomic_t ipx_sock_nr; #endif -static int ipxcfg_set_auto_create(char val) +static void ipxcfg_set_auto_create(char val) { - if(ipxcfg_auto_create_interfaces != val) - { - if(val) + if (ipxcfg_auto_create_interfaces != val) { + if (val) MOD_INC_USE_COUNT; else MOD_DEC_USE_COUNT; ipxcfg_auto_create_interfaces = val; } - - return (0); } -static int ipxcfg_set_auto_select(char val) +static void ipxcfg_set_auto_select(char val) { ipxcfg_auto_select_primary = val; - if(val && (ipx_primary_net == NULL)) + if (val && !ipx_primary_net) ipx_primary_net = ipx_interfaces; - - return (0); } static int ipxcfg_get_config_data(ipx_config_data *arg) @@ -171,7 +169,7 @@ static int ipxcfg_get_config_data(ipx_config_data *arg) vals.ipxcfg_auto_create_interfaces = ipxcfg_auto_create_interfaces; vals.ipxcfg_auto_select_primary = ipxcfg_auto_select_primary; - return (copy_to_user(arg, &vals, sizeof(vals)) ? -EFAULT : 0); + return copy_to_user(arg, &vals, sizeof(vals)) ? -EFAULT : 0; } /**************************************************************************\ @@ -213,22 +211,19 @@ void ipx_remove_socket(struct sock *sk) /* Determine interface with which socket is associated */ intrfc = sk->protinfo.af_ipx.intrfc; - if(intrfc == NULL) + if (!intrfc) return; ipxitf_hold(intrfc); spin_lock_bh(&intrfc->if_sklist_lock); s = intrfc->if_sklist; - if(s == sk) - { + if (s == sk) { intrfc->if_sklist = s->next; goto out; } - while(s && s->next) - { - if(s->next == sk) - { + while (s && s->next) { + if (s->next == sk) { s->next = sk->next; goto out; } @@ -264,7 +259,7 @@ static ipx_route * ipxrtr_lookup(__u32); static void ipxitf_clear_primary_net(void) { - if(ipxcfg_auto_select_primary && (ipx_interfaces != NULL)) + if (ipxcfg_auto_select_primary && ipx_interfaces) ipx_primary_net = ipx_interfaces; else ipx_primary_net = NULL; @@ -273,14 +268,14 @@ static void ipxitf_clear_primary_net(void) static ipx_interface *__ipxitf_find_using_phys(struct net_device *dev, unsigned short datalink) { - ipx_interface *i; + ipx_interface *i; - for(i = ipx_interfaces; - i && ((i->if_dev != dev) || (i->if_dlink_type != datalink)); + for (i = ipx_interfaces; + i && (i->if_dev != dev || i->if_dlink_type != datalink); i = i->if_next) ; - return (i); + return i; } static ipx_interface *ipxitf_find_using_phys(struct net_device *dev, @@ -298,12 +293,12 @@ static ipx_interface *ipxitf_find_using_phys(struct net_device *dev, static ipx_interface *ipxitf_find_using_net(__u32 net) { - ipx_interface *i; + ipx_interface *i; spin_lock_bh(&ipx_interfaces_lock); - if(net) - for(i = ipx_interfaces; i && (i->if_netnum != net); - i = i->if_next) + if (net) + for (i = ipx_interfaces; i && i->if_netnum != net; + i = i->if_next) ; else i = ipx_primary_net; @@ -311,7 +306,7 @@ static ipx_interface *ipxitf_find_using_net(__u32 net) ipxitf_hold(i); spin_unlock_bh(&ipx_interfaces_lock); - return (i); + return i; } /* Sockets are bound to a particular IPX interface. */ @@ -324,11 +319,10 @@ static void ipxitf_insert_socket(ipx_interface *intrfc, struct sock *sk) spin_lock_bh(&intrfc->if_sklist_lock); sk->protinfo.af_ipx.intrfc = intrfc; sk->next = NULL; - if(intrfc->if_sklist == NULL) + if (!intrfc->if_sklist) intrfc->if_sklist = sk; - else - { - for (s = intrfc->if_sklist; s->next != NULL; s = s->next) + else { + for (s = intrfc->if_sklist; s->next; s = s->next) ; s->next = sk; } @@ -337,19 +331,21 @@ static void ipxitf_insert_socket(ipx_interface *intrfc, struct sock *sk) } /* caller must hold intrfc->if_sklist_lock */ -static struct sock *__ipxitf_find_socket(ipx_interface *intrfc, unsigned short port) +static struct sock *__ipxitf_find_socket(ipx_interface *intrfc, + unsigned short port) { struct sock *s; - for(s = intrfc->if_sklist; - (s != NULL) && (s->protinfo.af_ipx.port != port); - s = s->next) + for (s = intrfc->if_sklist; + s && s->protinfo.af_ipx.port != port; + s = s->next) ; return s; } /* caller must hold a reference to intrfc */ -static struct sock *ipxitf_find_socket(ipx_interface *intrfc, unsigned short port) +static struct sock *ipxitf_find_socket(ipx_interface *intrfc, + unsigned short port) { struct sock *s; @@ -359,11 +355,10 @@ static struct sock *ipxitf_find_socket(ipx_interface *intrfc, unsigned short por sock_hold(s); spin_unlock_bh(&intrfc->if_sklist_lock); - return (s); + return s; } #ifdef CONFIG_IPX_INTERN - static struct sock *ipxitf_find_internal_socket(ipx_interface *intrfc, unsigned char *node, unsigned short port) { @@ -373,19 +368,16 @@ static struct sock *ipxitf_find_internal_socket(ipx_interface *intrfc, spin_lock_bh(&intrfc->if_sklist_lock); s = intrfc->if_sklist; - while(s != NULL) - { - if((s->protinfo.af_ipx.port == port) - && (memcmp(node, s->protinfo.af_ipx.node, IPX_NODE_LEN) == 0)) - { + while (s) { + if (s->protinfo.af_ipx.port == port && + !memcmp(node, s->protinfo.af_ipx.node, IPX_NODE_LEN)) break; - } s = s->next; } spin_unlock_bh(&intrfc->if_sklist_lock); ipxitf_put(intrfc); - return (s); + return s; } #endif @@ -401,8 +393,7 @@ static void __ipxitf_down(ipx_interface *intrfc) spin_lock_bh(&intrfc->if_sklist_lock); /* error sockets */ - for(s = intrfc->if_sklist; s != NULL; ) - { + for (s = intrfc->if_sklist; s; ) { s->err = ENOLINK; s->error_report(s); s->protinfo.af_ipx.intrfc = NULL; @@ -416,30 +407,27 @@ static void __ipxitf_down(ipx_interface *intrfc) spin_unlock_bh(&intrfc->if_sklist_lock); /* remove this interface from list */ - if(intrfc == ipx_interfaces) + if (intrfc == ipx_interfaces) ipx_interfaces = intrfc->if_next; - else - { - for(i = ipx_interfaces; - (i != NULL) && (i->if_next != intrfc); - i = i->if_next) + else { + for (i = ipx_interfaces; + i && i->if_next != intrfc; + i = i->if_next) ; - if((i != NULL) && (i->if_next == intrfc)) + if (i && i->if_next == intrfc) i->if_next = intrfc->if_next; } /* remove this interface from *special* networks */ - if(intrfc == ipx_primary_net) + if (intrfc == ipx_primary_net) ipxitf_clear_primary_net(); - if(intrfc == ipx_internal_net) + if (intrfc == ipx_internal_net) ipx_internal_net = NULL; if (intrfc->if_dev) dev_put(intrfc->if_dev); kfree(intrfc); MOD_DEC_USE_COUNT; - - return; } static void ipxitf_down(ipx_interface *intrfc) @@ -449,36 +437,31 @@ static void ipxitf_down(ipx_interface *intrfc) spin_unlock_bh(&ipx_interfaces_lock); } -static int ipxitf_device_event(struct notifier_block *notifier, unsigned long event, void *ptr) +static int ipxitf_device_event(struct notifier_block *notifier, + unsigned long event, void *ptr) { struct net_device *dev = ptr; ipx_interface *i, *tmp; - if(event != NETDEV_DOWN) + if (event != NETDEV_DOWN) return NOTIFY_DONE; spin_lock_bh(&ipx_interfaces_lock); - for(i = ipx_interfaces; i != NULL;) - { + for (i = ipx_interfaces; i;) { tmp = i->if_next; - if(i->if_dev == dev) + if (i->if_dev == dev) __ipxitf_put(i); i = tmp; } spin_unlock_bh(&ipx_interfaces_lock); - - return (NOTIFY_DONE); + return NOTIFY_DONE; } -static int ipxitf_def_skb_handler(struct sock *sock, struct sk_buff *skb) +static void ipxitf_def_skb_handler(struct sock *sock, struct sk_buff *skb) { - int retval; - - if((retval = sock_queue_rcv_skb(sock, skb)) < 0) + if (sock_queue_rcv_skb(sock, skb) < 0) kfree_skb(skb); - - return (retval); } /* @@ -500,57 +483,50 @@ static int ipxitf_demux_socket(ipx_interface *intrfc, struct sk_buff *skb, int c spin_lock_bh(&intrfc->if_sklist_lock); s = intrfc->if_sklist; - while(s != NULL) - { - if((s->protinfo.af_ipx.port == ipx->ipx_dest.sock) - && (is_broadcast - || (memcmp(ipx->ipx_dest.node, s->protinfo.af_ipx.node, - IPX_NODE_LEN) == 0))) - { + while (s) { + if (s->protinfo.af_ipx.port == ipx->ipx_dest.sock && + (is_broadcast || !memcmp(ipx->ipx_dest.node, + s->protinfo.af_ipx.node, + IPX_NODE_LEN))) { /* We found a socket to which to send */ struct sk_buff *skb1; - if(copy != 0) - { + if (copy) { skb1 = skb_clone(skb, GFP_ATOMIC); ret = -ENOMEM; - if (skb1 == NULL) + if (!skb1) goto out; - } - else - { + } else { skb1 = skb; copy = 1; /* skb may only be used once */ } ipxitf_def_skb_handler(s, skb1); /* On an external interface, one socket can listen */ - if(intrfc != ipx_internal_net) + if (intrfc != ipx_internal_net) break; } s = s->next; } /* skb was solely for us, and we did not make a copy, so free it. */ - if(copy == 0) + if (!copy) kfree_skb(skb); ret = 0; out: spin_unlock_bh(&intrfc->if_sklist_lock); return ret; } - #else - -static int ipxitf_demux_socket(ipx_interface *intrfc, struct sk_buff *skb, int copy) +static int ipxitf_demux_socket(ipx_interface *intrfc, struct sk_buff *skb, + int copy) { struct ipxhdr *ipx = skb->nh.ipxh; struct sock *sock1 = NULL, *sock2 = NULL; struct sk_buff *skb1 = NULL, *skb2 = NULL; int ret; - if (intrfc == ipx_primary_net && ntohs(ipx->ipx_dest.sock) == 0x451) - { + if (intrfc == ipx_primary_net && ntohs(ipx->ipx_dest.sock) == 0x451) { /* * The packet's target is a NCP connection handler. We want to * hand it to the correct socket directly within the kernel, @@ -562,48 +538,34 @@ static int ipxitf_demux_socket(ipx_interface *intrfc, struct sk_buff *skb, int c * VERY fast as well. */ int connection = 0; - - if (*((char*)(ipx+1)) == 0x22 && *((char*)(ipx+1)+1) == 0x22) - { - /* - * The packet is a NCP request - */ - connection = ( ((int) *((char*)(ipx+1)+5)) << 8 ) - | (int) *((char*)(ipx+1)+3); - } - else if (*((char*)(ipx+1))== 0x77 && *((char*)(ipx+1)+1) == 0x77) - { - /* - * The packet is a BURST packet - */ - connection = ( ((int) *((char*)(ipx+1)+9)) << 8 ) - | (int) *((char*)(ipx+1)+8); - } - - if (connection) - { - /* - * Now we have to look for a special NCP connection handling - * socket. Only these sockets have ipx_ncp_conn != 0, set - * by SIOCIPXNCPCONN. - */ + u8 *ncphdr = (u8 *)(ipx + 1); + + if (*ncphdr == 0x22 && *(ncphdr + 1) == 0x22) + /* The packet is a NCP request */ + connection = (((int) *(ncphdr + 5)) << 8) | + (int) *(ncphdr+3); + else if (*ncphdr == 0x77 && *(ncphdr + 1) == 0x77) + /* The packet is a BURST packet */ + connection = (((int) *(ncphdr+9)) << 8) | + (int) *(ncphdr+8); + + if (connection) { + /* Now we have to look for a special NCP connection + * handling socket. Only these sockets have + * ipx_ncp_conn != 0, set by SIOCIPXNCPCONN. */ spin_lock_bh(&intrfc->if_sklist_lock); - for (sock1=intrfc->if_sklist; - (sock1 != NULL) && - (sock1->protinfo.af_ipx.ipx_ncp_conn != connection); - sock1=sock1->next); + for (sock1 = intrfc->if_sklist; + sock1 && + sock1->protinfo.af_ipx.ipx_ncp_conn != connection; + sock1 = sock1->next); if (sock1) sock_hold(sock1); spin_unlock_bh(&intrfc->if_sklist_lock); } } - if (sock1 == NULL) - { - /* No special socket found, forward the packet the - * normal way. - */ + if (!sock1) + /* No special socket found, forward the packet the normal way */ sock1 = ipxitf_find_socket(intrfc, ipx->ipx_dest.sock); - } /* * We need to check if there is a primary net and if @@ -613,10 +575,8 @@ static int ipxitf_demux_socket(ipx_interface *intrfc, struct sk_buff *skb, int c * 0x456(Diagnostic). */ - if(ipx_primary_net && (intrfc != ipx_primary_net)) - { - switch(ntohs(ipx->ipx_dest.sock)) - { + if (ipx_primary_net && intrfc != ipx_primary_net) { + switch (ntohs(ipx->ipx_dest.sock)) { case 0x452: case 0x453: case 0x456: @@ -637,11 +597,10 @@ static int ipxitf_demux_socket(ipx_interface *intrfc, struct sk_buff *skb, int c /* * If there is nothing to do return. The kfree will cancel any charging. */ - if(sock1 == NULL && sock2 == NULL) - { - if(!copy) + if (!sock1 && !sock2) { + if (!copy) kfree_skb(skb); - return (0); + return 0; } /* @@ -652,30 +611,30 @@ static int ipxitf_demux_socket(ipx_interface *intrfc, struct sk_buff *skb, int c * copies, we do as much as is possible. */ - if(copy) + if (copy) skb1 = skb_clone(skb, GFP_ATOMIC); else skb1 = skb; ret = -ENOMEM; - if(skb1 == NULL) + if (!skb1) goto out; /* Do we need 2 SKBs? */ - if(sock1 && sock2) + if (sock1 && sock2) skb2 = skb_clone(skb1, GFP_ATOMIC); else skb2 = skb1; - if(sock1) - (void) ipxitf_def_skb_handler(sock1, skb1); + if (sock1) + ipxitf_def_skb_handler(sock1, skb1); ret = -ENOMEM; - if(skb2 == NULL) + if (!skb2) goto out; - if(sock2) - (void) ipxitf_def_skb_handler(sock2, skb2); + if (sock2) + ipxitf_def_skb_handler(sock2, skb2); ret = 0; out: if (sock1) @@ -686,7 +645,8 @@ out: if (sock1) } #endif /* CONFIG_IPX_INTERN */ -static struct sk_buff *ipxitf_adjust_skbuff(ipx_interface *intrfc, struct sk_buff *skb) +static struct sk_buff *ipxitf_adjust_skbuff(ipx_interface *intrfc, + struct sk_buff *skb) { struct sk_buff *skb2; int in_offset = skb->h.raw - skb->head; @@ -694,21 +654,21 @@ static struct sk_buff *ipxitf_adjust_skbuff(ipx_interface *intrfc, struct sk_buf int len; /* Hopefully, most cases */ - if(in_offset >= out_offset) - return (skb); + if (in_offset >= out_offset) + return skb; /* Need new SKB */ len = skb->len + out_offset; skb2 = alloc_skb(len, GFP_ATOMIC); - if(skb2 != NULL) - { + if (skb2) { skb_reserve(skb2, out_offset); skb2->nh.raw = skb2->h.raw = skb_put(skb2,skb->len); memcpy(skb2->h.raw, skb->h.raw, skb->len); + memcpy(skb2->cb, skb->cb, sizeof(skb->cb)); } kfree_skb(skb); - return (skb2); + return skb2; } /* caller must hold a reference to intrfc */ @@ -716,6 +676,7 @@ static struct sk_buff *ipxitf_adjust_skbuff(ipx_interface *intrfc, struct sk_buf static int ipxitf_send(ipx_interface *intrfc, struct sk_buff *skb, char *node) { struct ipxhdr *ipx = skb->nh.ipxh; + struct ipx_cb *cb = (struct ipx_cb *) skb->cb; struct net_device *dev = intrfc->if_dev; struct datalink_proto *dl = intrfc->if_dlink; char dest_node[IPX_NODE_LEN]; @@ -727,7 +688,7 @@ static int ipxitf_send(ipx_interface *intrfc, struct sk_buff *skb, char *node) * packet to avoid unnecessary copies. */ - if((dl == NULL) || (dev == NULL) || (dev->flags & IFF_LOOPBACK)) + if (!dl || !dev || dev->flags & IFF_LOOPBACK) send_to_wire = 0; /* No non looped */ /* @@ -737,30 +698,27 @@ static int ipxitf_send(ipx_interface *intrfc, struct sk_buff *skb, char *node) * up clones. */ - if(ipx->ipx_dest.net == intrfc->if_netnum) - { + if (cb->ipx_dest_net == intrfc->if_netnum) { /* * To our own node, loop and free the original. * The internal net will receive on all node address. */ - if((intrfc == ipx_internal_net) - || memcmp(intrfc->if_node, node, IPX_NODE_LEN) == 0) - { + if (intrfc == ipx_internal_net || + !memcmp(intrfc->if_node, node, IPX_NODE_LEN)) { /* Don't charge sender */ skb_orphan(skb); /* Will charge receiver */ - return (ipxitf_demux_socket(intrfc, skb, 0)); + return ipxitf_demux_socket(intrfc, skb, 0); } /* Broadcast, loop and possibly keep to send on. */ - if(memcmp(ipx_broadcast_node, node, IPX_NODE_LEN) == 0) - { - if(!send_to_wire) + if (!memcmp(ipx_broadcast_node, node, IPX_NODE_LEN)) { + if (!send_to_wire) skb_orphan(skb); ipxitf_demux_socket(intrfc, skb, send_to_wire); - if(!send_to_wire) - return (0); + if (!send_to_wire) + return 0; } } @@ -769,36 +727,45 @@ static int ipxitf_send(ipx_interface *intrfc, struct sk_buff *skb, char *node) * We are still charging the sender. Which is right - the driver * free will handle this fairly. */ - if(ipx->ipx_source.net != intrfc->if_netnum) - { + if (cb->ipx_source_net != intrfc->if_netnum) { /* * Unshare the buffer before modifying the count in * case its a flood or tcpdump */ skb = skb_unshare(skb, GFP_ATOMIC); - if(!skb) - return (0); - if(++(ipx->ipx_tctrl) > ipxcfg_max_hops) + if (!skb) + return 0; + if (++(cb->ipx_tctrl) > ipxcfg_max_hops) send_to_wire = 0; } - if(!send_to_wire) - { + if (!send_to_wire) { kfree_skb(skb); - return (0); + return 0; } /* Determine the appropriate hardware address */ addr_len = dev->addr_len; - if(memcmp(ipx_broadcast_node, node, IPX_NODE_LEN) == 0) + if (!memcmp(ipx_broadcast_node, node, IPX_NODE_LEN)) memcpy(dest_node, dev->broadcast, addr_len); else memcpy(dest_node, &(node[IPX_NODE_LEN-addr_len]), addr_len); /* Make any compensation for differing physical/data link size */ skb = ipxitf_adjust_skbuff(intrfc, skb); - if(skb == NULL) - return (0); + if (!skb) + return 0; + + ipx->ipx_tctrl = cb->ipx_tctrl; + ipx->ipx_dest.net = cb->ipx_dest_net; + ipx->ipx_source.net = cb->ipx_source_net; + /* see if we need to include the netnum in the route list */ + if (cb->last_hop_index >= 0) { + u32 *last_hop = (u32 *)(((u8 *) skb->data) + + sizeof(struct ipxhdr) + cb->last_hop_index * + sizeof(u32)); + *last_hop = intrfc->if_netnum; + } /* set up data link and physical headers */ skb->dev = dev; @@ -807,15 +774,14 @@ static int ipxitf_send(ipx_interface *intrfc, struct sk_buff *skb, char *node) /* Send it out */ dev_queue_xmit(skb); - - return (0); + return 0; } static int ipxrtr_add_route(__u32, ipx_interface *, unsigned char *); static int ipxitf_add_local_route(ipx_interface *intrfc) { - return (ipxrtr_add_route(intrfc->if_netnum, intrfc, NULL)); + return ipxrtr_add_route(intrfc->if_netnum, intrfc, NULL); } static const char * ipx_frame_name(unsigned short); @@ -824,29 +790,26 @@ static const char * ipx_device_name(ipx_interface *); static int ipxitf_rcv(ipx_interface *intrfc, struct sk_buff *skb) { struct ipxhdr *ipx = skb->nh.ipxh; - ipx_interface *i; + struct ipx_cb *cb = (struct ipx_cb *) skb->cb; int ret = 0; ipxitf_hold(intrfc); /* See if we should update our network number */ - if(!intrfc->if_netnum /* net number of intrfc not known yet (== 0) */ - && (ipx->ipx_source.net == ipx->ipx_dest.net) /* intra packet */ - && ipx->ipx_source.net) /* source net number of packet != 0 */ - { + if (!intrfc->if_netnum && /* net number of intrfc not known yet */ + cb->ipx_source_net == cb->ipx_dest_net && /* intra packet */ + cb->ipx_source_net) { + ipx_interface *i = ipxitf_find_using_net(cb->ipx_source_net); /* NB: NetWare servers lie about their hop count so we * dropped the test based on it. This is the best way * to determine this is a 0 hop count packet. */ - if((i=ipxitf_find_using_net(ipx->ipx_source.net)) == NULL) - { - intrfc->if_netnum = ipx->ipx_source.net; - (void) ipxitf_add_local_route(intrfc); - } - else - { + if (!i) { + intrfc->if_netnum = cb->ipx_source_net; + ipxitf_add_local_route(intrfc); + } else { printk(KERN_WARNING "IPX: Network number collision %lx\n %s %s and %s %s\n", - (long unsigned int) htonl(ipx->ipx_source.net), + (long unsigned int) htonl(cb->ipx_source_net), ipx_device_name(i), ipx_frame_name(i->if_dlink_type), ipx_device_name(intrfc), @@ -854,75 +817,63 @@ static int ipxitf_rcv(ipx_interface *intrfc, struct sk_buff *skb) ipxitf_put(i); } } + + cb->last_hop_index = -1; - if(ipx->ipx_type == IPX_TYPE_PPROP - && ipx->ipx_tctrl < 8 - && skb->pkt_type != PACKET_OTHERHOST - /* header + 8 network numbers */ - && ntohs(ipx->ipx_pktsize) >= sizeof(struct ipxhdr) + 8 * 4) - { + if (ipx->ipx_type == IPX_TYPE_PPROP && cb->ipx_tctrl < 8 && + skb->pkt_type != PACKET_OTHERHOST && + /* header + 8 network numbers */ + ntohs(ipx->ipx_pktsize) >= sizeof(struct ipxhdr) + 8 * 4) { int i; ipx_interface *ifcs; struct sk_buff *skb2; - __u32 *l; - char *c; - - c = (char *) skb->data; - c += sizeof(struct ipxhdr); - l = (__u32 *) c; - - i = 0; + char *c = ((char *) skb->data) + sizeof(struct ipxhdr); + u32 *l = (u32 *) c; /* Dump packet if already seen this net */ - for( ; i < ipx->ipx_tctrl; i++) - if(*l++ == intrfc->if_netnum) + for (i = 0; i < cb->ipx_tctrl; i++) + if (*l++ == intrfc->if_netnum) break; - if(i == ipx->ipx_tctrl) - { + if (i == cb->ipx_tctrl) { /* < 8 hops && input itfc not in list */ - *l = intrfc->if_netnum; /* insert recvd netnum into list */ - ipx->ipx_tctrl++; + /* insert recvd netnum into list */ + cb->last_hop_index = i; + cb->ipx_tctrl++; /* xmit on all other interfaces... */ spin_lock_bh(&ipx_interfaces_lock); - for(ifcs = ipx_interfaces; ifcs != NULL; ifcs = ifcs->if_next) - { + for (ifcs = ipx_interfaces; ifcs; + ifcs = ifcs->if_next) { /* Except unconfigured interfaces */ - if(ifcs->if_netnum == 0) + if (!ifcs->if_netnum) continue; /* That aren't in the list */ l = (__u32 *) c; - for(i = 0; i <= ipx->ipx_tctrl; i++) - if(ifcs->if_netnum == *l++) + for (i = 0; i <= cb->ipx_tctrl; i++) + if (ifcs->if_netnum == *l++) break; - if(i - 1 == ipx->ipx_tctrl) - { - ipx->ipx_dest.net = ifcs->if_netnum; + if (i - 1 == cb->ipx_tctrl) { + cb->ipx_dest_net = ifcs->if_netnum; skb2=skb_clone(skb, GFP_ATOMIC); if (skb2) ipxrtr_route_skb(skb2); } } spin_unlock_bh(&ipx_interfaces_lock); - - /* Reset network number in packet */ - ipx->ipx_dest.net = intrfc->if_netnum; } } - if(!ipx->ipx_dest.net) - ipx->ipx_dest.net = intrfc->if_netnum; - if(!ipx->ipx_source.net) - ipx->ipx_source.net = intrfc->if_netnum; + if (!cb->ipx_dest_net) + cb->ipx_dest_net = intrfc->if_netnum; + if (!cb->ipx_source_net) + cb->ipx_source_net = intrfc->if_netnum; - if(intrfc->if_netnum != ipx->ipx_dest.net) - { + if (intrfc->if_netnum != cb->ipx_dest_net) { /* We only route point-to-point packets. */ - if(skb->pkt_type == PACKET_HOST) - { + if (skb->pkt_type == PACKET_HOST) { skb=skb_unshare(skb, GFP_ATOMIC); - if(skb) + if (skb) ret = ipxrtr_route_skb(skb); goto out_intrfc; } @@ -931,9 +882,8 @@ static int ipxitf_rcv(ipx_interface *intrfc, struct sk_buff *skb) } /* see if we should keep it */ - if((memcmp(ipx_broadcast_node, ipx->ipx_dest.node, IPX_NODE_LEN) == 0) - || (memcmp(intrfc->if_node, ipx->ipx_dest.node, IPX_NODE_LEN) == 0)) - { + if (!memcmp(ipx_broadcast_node, ipx->ipx_dest.node, IPX_NODE_LEN) || + !memcmp(intrfc->if_node, ipx->ipx_dest.node, IPX_NODE_LEN)) { ret = ipxitf_demux_socket(intrfc, skb, 0); goto out_intrfc; } @@ -952,20 +902,17 @@ static void ipxitf_insert(ipx_interface *intrfc) intrfc->if_next = NULL; spin_lock_bh(&ipx_interfaces_lock); - if(ipx_interfaces == NULL) + if (!ipx_interfaces) ipx_interfaces = intrfc; - else - { - for(i = ipx_interfaces; i->if_next != NULL; i = i->if_next) + else { + for (i = ipx_interfaces; i->if_next; i = i->if_next) ; i->if_next = intrfc; } spin_unlock_bh(&ipx_interfaces_lock); - if(ipxcfg_auto_select_primary && (ipx_primary_net == NULL)) + if (ipxcfg_auto_select_primary && !ipx_primary_net) ipx_primary_net = intrfc; - - return; } static int ipxitf_create_internal(ipx_interface_definition *idef) @@ -974,21 +921,21 @@ static int ipxitf_create_internal(ipx_interface_definition *idef) int ret; /* Only one primary network allowed */ - if(ipx_primary_net != NULL) - return (-EEXIST); + if (ipx_primary_net) + return -EEXIST; /* Must have a valid network number */ - if(!idef->ipx_network) - return (-EADDRNOTAVAIL); + if (!idef->ipx_network) + return -EADDRNOTAVAIL; intrfc = ipxitf_find_using_net(idef->ipx_network); - if(intrfc != NULL) { + if (intrfc) { ipxitf_put(intrfc); - return (-EADDRINUSE); + return -EADDRINUSE; } - intrfc = (ipx_interface *)kmalloc(sizeof(ipx_interface),GFP_ATOMIC); - if(intrfc == NULL) - return (-EAGAIN); + intrfc = kmalloc(sizeof(ipx_interface),GFP_ATOMIC); + if (!intrfc) + return -EAGAIN; intrfc->if_dev = NULL; intrfc->if_netnum = idef->ipx_network; intrfc->if_dlink_type = 0; @@ -998,8 +945,7 @@ static int ipxitf_create_internal(ipx_interface_definition *idef) intrfc->if_ipx_offset = 0; intrfc->if_sknum = IPX_MIN_EPHEMERAL_SOCKET; memcpy((char *)&(intrfc->if_node), idef->ipx_node, IPX_NODE_LEN); - ipx_internal_net = intrfc; - ipx_primary_net = intrfc; + ipx_internal_net = ipx_primary_net = intrfc; spin_lock_init(&intrfc->if_sklist_lock); atomic_set(&intrfc->refcnt, 1); MOD_INC_USE_COUNT; @@ -1013,22 +959,21 @@ static int ipxitf_create_internal(ipx_interface_definition *idef) static int ipx_map_frame_type(unsigned char type) { - switch(type) - { + switch (type) { case IPX_FRAME_ETHERII: - return (htons(ETH_P_IPX)); + return htons(ETH_P_IPX); case IPX_FRAME_8022: - return (htons(ETH_P_802_2)); + return htons(ETH_P_802_2); case IPX_FRAME_SNAP: - return (htons(ETH_P_SNAP)); + return htons(ETH_P_SNAP); case IPX_FRAME_8023: - return (htons(ETH_P_802_3)); + return htons(ETH_P_802_3); } - return (0); + return 0; } static int ipxitf_create(ipx_interface_definition *idef) @@ -1039,29 +984,29 @@ static int ipxitf_create(ipx_interface_definition *idef) ipx_interface *intrfc; int err; - if(idef->ipx_special == IPX_INTERNAL) - return (ipxitf_create_internal(idef)); + if (idef->ipx_special == IPX_INTERNAL) + return ipxitf_create_internal(idef); - if((idef->ipx_special == IPX_PRIMARY) && (ipx_primary_net != NULL)) - return (-EEXIST); + if (idef->ipx_special == IPX_PRIMARY && ipx_primary_net) + return -EEXIST; intrfc = ipxitf_find_using_net(idef->ipx_network); - if(idef->ipx_network && intrfc != NULL) { + if (idef->ipx_network && intrfc) { ipxitf_put(intrfc); - return (-EADDRINUSE); + return -EADDRINUSE; } if (intrfc) ipxitf_put(intrfc); dev = dev_get_by_name(idef->ipx_device); - if(dev == NULL) - return (-ENODEV); + if (!dev) + return -ENODEV; - switch(idef->ipx_dlink_type) - { + switch (idef->ipx_dlink_type) { case IPX_FRAME_TR_8022: - printk("IPX frame type 802.2TR is obsolete. Use 802.2 instead.\n"); + printk(KERN_WARNING "IPX frame type 802.2TR is " + "obsolete Use 802.2 instead.\n"); /* fall through */ case IPX_FRAME_8022: @@ -1070,14 +1015,14 @@ static int ipxitf_create(ipx_interface_definition *idef) break; case IPX_FRAME_ETHERII: - if (dev->type != ARPHRD_IEEE802) - { + if (dev->type != ARPHRD_IEEE802) { dlink_type = htons(ETH_P_IPX); datalink = pEII_datalink; break; - } - else - printk("IPX frame type EtherII over token-ring is obsolete. Use SNAP instead.\n"); + } else + printk(KERN_WARNING "IPX frame type EtherII " + "over token-ring is obsolete. Use SNAP " + "instead.\n"); /* fall through */ case IPX_FRAME_SNAP: @@ -1096,24 +1041,24 @@ static int ipxitf_create(ipx_interface_definition *idef) } err = -ENETDOWN; - if(!(dev->flags & IFF_UP)) + if (!(dev->flags & IFF_UP)) goto out_dev; /* Check addresses are suitable */ err = -EINVAL; - if(dev->addr_len > IPX_NODE_LEN) + if (dev->addr_len > IPX_NODE_LEN) goto out_dev; err = -EPROTONOSUPPORT; - if(datalink == NULL) + if (!datalink) goto out_dev; - if((intrfc = ipxitf_find_using_phys(dev, dlink_type)) == NULL) - { + intrfc = ipxitf_find_using_phys(dev, dlink_type); + if (!intrfc) { /* Ok now create */ - intrfc = (ipx_interface *)kmalloc(sizeof(ipx_interface),GFP_ATOMIC); + intrfc = kmalloc(sizeof(ipx_interface), GFP_ATOMIC); err = -EAGAIN; - if(intrfc == NULL) + if (!intrfc) goto out_dev; intrfc->if_dev = dev; intrfc->if_netnum = idef->ipx_network; @@ -1122,17 +1067,16 @@ static int ipxitf_create(ipx_interface_definition *idef) intrfc->if_sklist = NULL; intrfc->if_sknum = IPX_MIN_EPHEMERAL_SOCKET; /* Setup primary if necessary */ - if((idef->ipx_special == IPX_PRIMARY)) + if ((idef->ipx_special == IPX_PRIMARY)) ipx_primary_net = intrfc; intrfc->if_internal = 0; intrfc->if_ipx_offset = dev->hard_header_len + datalink->header_length; - if(memcmp(idef->ipx_node, "\000\000\000\000\000\000", IPX_NODE_LEN) == 0) - { + if (!memcmp(idef->ipx_node, "\000\000\000\000\000\000", + IPX_NODE_LEN)) { memset(intrfc->if_node, 0, IPX_NODE_LEN); memcpy((char *)&(intrfc->if_node[IPX_NODE_LEN-dev->addr_len]), dev->dev_addr, dev->addr_len); - } - else + } else memcpy(intrfc->if_node, idef->ipx_node, IPX_NODE_LEN); spin_lock_init(&intrfc->if_sklist_lock); atomic_set(&intrfc->refcnt, 1); @@ -1144,7 +1088,7 @@ static int ipxitf_create(ipx_interface_definition *idef) /* If the network number is known, add a route */ err = 0; - if(!intrfc->if_netnum) + if (!intrfc->if_netnum) goto out_intrfc; err = ipxitf_add_local_route(intrfc); @@ -1164,10 +1108,8 @@ static int ipxitf_delete(ipx_interface_definition *idef) int ret = 0; spin_lock_bh(&ipx_interfaces_lock); - if(idef->ipx_special == IPX_INTERNAL) - { - if(ipx_internal_net != NULL) - { + if (idef->ipx_special == IPX_INTERNAL) { + if (ipx_internal_net) { __ipxitf_put(ipx_internal_net); goto out; } @@ -1176,19 +1118,19 @@ static int ipxitf_delete(ipx_interface_definition *idef) } dlink_type = ipx_map_frame_type(idef->ipx_dlink_type); - if(dlink_type == 0) { + if (!dlink_type) { ret = -EPROTONOSUPPORT; goto out; } dev = __dev_get_by_name(idef->ipx_device); - if(dev == NULL) { + if (!dev) { ret = -ENODEV; goto out; } intrfc = __ipxitf_find_using_phys(dev, dlink_type); - if(intrfc != NULL) + if (intrfc) __ipxitf_put(intrfc); else ret = -EINVAL; @@ -1198,13 +1140,12 @@ out: spin_unlock_bh(&ipx_interfaces_lock); } static ipx_interface *ipxitf_auto_create(struct net_device *dev, - unsigned short dlink_type) + unsigned short dlink_type) { struct datalink_proto *datalink = NULL; ipx_interface *intrfc; - switch(htons(dlink_type)) - { + switch (htons(dlink_type)) { case ETH_P_IPX: datalink = pEII_datalink; break; @@ -1222,19 +1163,18 @@ static ipx_interface *ipxitf_auto_create(struct net_device *dev, break; default: - return (NULL); + return NULL; } - if(dev == NULL) - return (NULL); + if (!dev) + return NULL; /* Check addresses are suitable */ - if(dev->addr_len>IPX_NODE_LEN) - return (NULL); + if (dev->addr_len > IPX_NODE_LEN) + return NULL; - intrfc = (ipx_interface *)kmalloc(sizeof(ipx_interface),GFP_ATOMIC); - if(intrfc != NULL) - { + intrfc = kmalloc(sizeof(ipx_interface), GFP_ATOMIC); + if (intrfc) { intrfc->if_dev = dev; intrfc->if_netnum = 0; intrfc->if_dlink_type = dlink_type; @@ -1253,27 +1193,25 @@ static ipx_interface *ipxitf_auto_create(struct net_device *dev, ipxitf_insert(intrfc); } - return (intrfc); + return intrfc; } static int ipxitf_ioctl(unsigned int cmd, void *arg) { struct ifreq ifr; - int err, val; + int err = 0, val; - switch(cmd) - { - case SIOCSIFADDR: - { + switch (cmd) { + case SIOCSIFADDR: { struct sockaddr_ipx *sipx; ipx_interface_definition f; - if(copy_from_user(&ifr, arg, sizeof(ifr))) - return (-EFAULT); + if (copy_from_user(&ifr, arg, sizeof(ifr))) + return -EFAULT; sipx = (struct sockaddr_ipx *)&ifr.ifr_addr; - if(sipx->sipx_family != AF_IPX) - return (-EINVAL); + if (sipx->sipx_family != AF_IPX) + return -EINVAL; f.ipx_network = sipx->sipx_network; memcpy(f.ipx_device,ifr.ifr_name,sizeof(f.ipx_device)); @@ -1281,10 +1219,10 @@ static int ipxitf_ioctl(unsigned int cmd, void *arg) f.ipx_dlink_type = sipx->sipx_type; f.ipx_special = sipx->sipx_special; - if(sipx->sipx_action == IPX_DLTITF) - return (ipxitf_delete(&f)); + if (sipx->sipx_action == IPX_DLTITF) + return ipxitf_delete(&f); else - return (ipxitf_create(&f)); + return ipxitf_create(&f); } case SIOCGIFADDR: @@ -1293,50 +1231,46 @@ static int ipxitf_ioctl(unsigned int cmd, void *arg) ipx_interface *ipxif; struct net_device *dev; - if(copy_from_user(&ifr, arg, sizeof(ifr))) - return (-EFAULT); + if (copy_from_user(&ifr, arg, sizeof(ifr))) + return -EFAULT; sipx = (struct sockaddr_ipx *)&ifr.ifr_addr; dev = __dev_get_by_name(ifr.ifr_name); - if(!dev) - return (-ENODEV); + if (!dev) + return -ENODEV; ipxif = ipxitf_find_using_phys(dev, ipx_map_frame_type(sipx->sipx_type)); - if(ipxif == NULL) - return (-EADDRNOTAVAIL); + if (!ipxif) + return -EADDRNOTAVAIL; sipx->sipx_family = AF_IPX; sipx->sipx_network = ipxif->if_netnum; - memcpy(sipx->sipx_node, ipxif->if_node, sizeof(sipx->sipx_node)); - err = -EFAULT; - if(!copy_to_user(arg, &ifr, sizeof(ifr))) - err = 0; + memcpy(sipx->sipx_node, ipxif->if_node, + sizeof(sipx->sipx_node)); + if (copy_to_user(arg, &ifr, sizeof(ifr))) + err = -EFAULT; ipxitf_put(ipxif); - return (err); + return err; } case SIOCAIPXITFCRT: - { - err = get_user(val, (unsigned char *) arg); - if(err) - return (err); - - return (ipxcfg_set_auto_create(val)); - } + if (get_user(val, (unsigned char *) arg)) + return -EFAULT; + ipxcfg_set_auto_create(val); + break; case SIOCAIPXPRISLT: - { - err = get_user(val, (unsigned char *) arg); - if(err) - return (err); - - return (ipxcfg_set_auto_select(val)); - } + if (get_user(val, (unsigned char *) arg)) + return -EFAULT; + ipxcfg_set_auto_select(val); + break; default: - return (-EINVAL); + return -EINVAL; } + + return 0; } /**************************************************************************\ @@ -1350,11 +1284,11 @@ static ipx_route *ipxrtr_lookup(__u32 net) ipx_route *r; read_lock_bh(&ipx_routes_lock); - for(r = ipx_routes; (r != NULL) && (r->ir_net != net); r = r->ir_next) + for (r = ipx_routes; r && r->ir_net != net; r = r->ir_next) ; read_unlock_bh(&ipx_routes_lock); - return (r); + return r; } /* caller must hold a reference to intrfc */ @@ -1365,34 +1299,30 @@ static int ipxrtr_add_route(__u32 network, ipx_interface *intrfc, unsigned char /* Get a route structure; either existing or create */ rt = ipxrtr_lookup(network); - if(rt == NULL) - { - rt = (ipx_route *)kmalloc(sizeof(ipx_route),GFP_ATOMIC); - if(rt == NULL) - return (-EAGAIN); + if (!rt) { + rt = kmalloc(sizeof(ipx_route),GFP_ATOMIC); + if (!rt) + return -EAGAIN; write_lock_bh(&ipx_routes_lock); rt->ir_next = ipx_routes; ipx_routes = rt; write_unlock_bh(&ipx_routes_lock); } - else if(intrfc == ipx_internal_net) - return (-EEXIST); + else if (intrfc == ipx_internal_net) + return -EEXIST; rt->ir_net = network; rt->ir_intrfc = intrfc; - if(node == NULL) - { + if (!node) { memset(rt->ir_router_node, '\0', IPX_NODE_LEN); rt->ir_routed = 0; - } - else - { + } else { memcpy(rt->ir_router_node, node, IPX_NODE_LEN); rt->ir_routed = 1; } - return (0); + return 0; } static void ipxrtr_del_routes(ipx_interface *intrfc) @@ -1400,14 +1330,11 @@ static void ipxrtr_del_routes(ipx_interface *intrfc) ipx_route **r, *tmp; write_lock_bh(&ipx_routes_lock); - for(r = &ipx_routes; (tmp = *r) != NULL;) - { - if(tmp->ir_intrfc == intrfc) - { + for (r = &ipx_routes; (tmp = *r) != NULL;) { + if (tmp->ir_intrfc == intrfc) { *r = tmp->ir_next; kfree(tmp); - } - else + } else r = &(tmp->ir_next); } write_unlock_bh(&ipx_routes_lock); @@ -1420,8 +1347,8 @@ static int ipxrtr_create(ipx_route_definition *rd) /* Find the appropriate interface */ intrfc = ipxitf_find_using_net(rd->ipx_router_network); - if(intrfc == NULL) - return (-ENETUNREACH); + if (!intrfc) + return -ENETUNREACH; ret = ipxrtr_add_route(rd->ipx_network, intrfc, rd->ipx_router_node); ipxitf_put(intrfc); return ret; @@ -1434,13 +1361,11 @@ static int ipxrtr_delete(long net) int err; write_lock_bh(&ipx_routes_lock); - for(r = &ipx_routes; (tmp = *r) != NULL;) - { - if(tmp->ir_net == net) - { + for (r = &ipx_routes; (tmp = *r) != NULL;) { + if (tmp->ir_net == net) { /* Directly connected; can't lose route */ err = -EPERM; - if(!(tmp->ir_routed)) + if (!tmp->ir_routed) goto out; *r = tmp->ir_next; @@ -1461,45 +1386,38 @@ out: write_unlock_bh(&ipx_routes_lock); */ /* Note: We assume ipx_tctrl==0 and htons(length)==ipx_pktsize */ +/* This functions should *not* mess with packet contents */ -static __u16 ipx_set_checksum(struct ipxhdr *packet,int length) +static __u16 ipx_cksum(struct ipxhdr *packet,int length) { /* * NOTE: sum is a net byte order quantity, which optimizes the * loop. This only works on big and little endian machines. (I * don't know of a machine that isn't.) */ - - __u32 sum = 0; - - /* Pointer to second word - We skip the checksum field */ - __u16 *p = (__u16 *)&packet->ipx_pktsize; - - /* Number of complete words */ - __u32 i = length >> 1; - char hops = packet->ipx_tctrl; - - /* Hop count excluded from checksum calc */ - packet->ipx_tctrl = 0; - - /* Loop through all complete words except the checksum field */ - while(--i) + /* start at ipx_dest - We skip the checksum field and start with + * ipx_type before the loop, not considering ipx_tctrl in the calc */ + __u16 *p = (__u16 *)&packet->ipx_dest; + __u32 i = (length >> 1) - 1; /* Number of complete words */ + __u32 sum = packet->ipx_type << sizeof(packet->ipx_tctrl); + + /* Loop through all complete words except the checksum field, + * ipx_type (accounted above) and ipx_tctrl (not used in the cksum) */ + while (--i) sum += *p++; /* Add on the last part word if it exists */ - if(packet->ipx_pktsize & htons(1)) + if (packet->ipx_pktsize & htons(1)) sum += ntohs(0xff00) & *p; - packet->ipx_tctrl = hops; - /* Do final fixup */ sum = (sum & 0xffff) + (sum >> 16); /* It's a pity there's no concept of carry in C */ - if(sum >= 0x10000) + if (sum >= 0x10000) sum++; - return (~sum); + return ~sum; } /* @@ -1510,80 +1428,76 @@ static int ipxrtr_route_packet(struct sock *sk, struct sockaddr_ipx *usipx, stru struct sk_buff *skb; ipx_interface *intrfc; struct ipxhdr *ipx; + struct ipx_cb *cb; int size; int ipx_offset; ipx_route *rt = NULL; int err; /* Find the appropriate interface on which to send packet */ - if(!usipx->sipx_network && (ipx_primary_net != NULL)) - { + if (!usipx->sipx_network && ipx_primary_net) { usipx->sipx_network = ipx_primary_net->if_netnum; intrfc = ipx_primary_net; - } - else - { + } else { rt = ipxrtr_lookup(usipx->sipx_network); - if(rt == NULL) - return (-ENETUNREACH); + if (!rt) + return -ENETUNREACH; intrfc = rt->ir_intrfc; } ipxitf_hold(intrfc); ipx_offset = intrfc->if_ipx_offset; - size = sizeof(struct ipxhdr) + len; - size += ipx_offset; + size = sizeof(struct ipxhdr) + len + ipx_offset; skb = sock_alloc_send_skb(sk, size, 0, noblock, &err); - if(skb == NULL) + if (!skb) goto out; skb_reserve(skb,ipx_offset); skb->sk = sk; + cb = (struct ipx_cb *) skb->cb; /* Fill in IPX header */ ipx = (struct ipxhdr *)skb_put(skb, sizeof(struct ipxhdr)); ipx->ipx_pktsize= htons(len + sizeof(struct ipxhdr)); - ipx->ipx_tctrl = 0; + cb->ipx_tctrl = 0; ipx->ipx_type = usipx->sipx_type; skb->h.raw = (void *)skb->nh.ipxh = ipx; - ipx->ipx_source.net = sk->protinfo.af_ipx.intrfc->if_netnum; + cb->last_hop_index = -1; #ifdef CONFIG_IPX_INTERN + cb->ipx_source_net = sk->protinfo.af_ipx.intrfc->if_netnum; memcpy(ipx->ipx_source.node, sk->protinfo.af_ipx.node, IPX_NODE_LEN); #else - if((err = ntohs(sk->protinfo.af_ipx.port)) == 0x453 || err == 0x452) - { + err = ntohs(sk->protinfo.af_ipx.port); + if (err == 0x453 || err == 0x452) { /* RIP/SAP special handling for mars_nwe */ - ipx->ipx_source.net = intrfc->if_netnum; + cb->ipx_source_net = intrfc->if_netnum; memcpy(ipx->ipx_source.node, intrfc->if_node, IPX_NODE_LEN); - } - else - { - ipx->ipx_source.net = sk->protinfo.af_ipx.intrfc->if_netnum; + } else { + cb->ipx_source_net = sk->protinfo.af_ipx.intrfc->if_netnum; memcpy(ipx->ipx_source.node, sk->protinfo.af_ipx.intrfc->if_node, IPX_NODE_LEN); } #endif /* CONFIG_IPX_INTERN */ ipx->ipx_source.sock = sk->protinfo.af_ipx.port; - ipx->ipx_dest.net = usipx->sipx_network; + cb->ipx_dest_net = usipx->sipx_network; memcpy(ipx->ipx_dest.node,usipx->sipx_node,IPX_NODE_LEN); ipx->ipx_dest.sock = usipx->sipx_port; err = memcpy_fromiovec(skb_put(skb,len),iov,len); - if(err) - { + if (err) { kfree_skb(skb); goto out; } /* Apply checksum. Not allowed on 802.3 links. */ - if(sk->no_check || intrfc->if_dlink_type == IPX_FRAME_8023) + if (sk->no_check || intrfc->if_dlink_type == IPX_FRAME_8023) ipx->ipx_checksum=0xFFFF; else - ipx->ipx_checksum = ipx_set_checksum(ipx, len + sizeof(struct ipxhdr)); + ipx->ipx_checksum = ipx_cksum(ipx, len + sizeof(struct ipxhdr)); err = ipxitf_send(intrfc, skb, (rt && rt->ir_routed) ? rt->ir_router_node : ipx->ipx_dest.node); @@ -1594,21 +1508,20 @@ out: ipxitf_put(intrfc); int ipxrtr_route_skb(struct sk_buff *skb) { struct ipxhdr *ipx = skb->nh.ipxh; - ipx_route *r; + struct ipx_cb *cb = (struct ipx_cb *) skb->cb; + ipx_route *r = ipxrtr_lookup(cb->ipx_dest_net); - r = ipxrtr_lookup(ipx->ipx_dest.net); - if(r == NULL) /* no known route */ - { + if (!r) { /* no known route */ kfree_skb(skb); - return (0); + return 0; } ipxitf_hold(r->ir_intrfc); - (void)ipxitf_send(r->ir_intrfc, skb, (r->ir_routed) ? + ipxitf_send(r->ir_intrfc, skb, (r->ir_routed) ? r->ir_router_node : ipx->ipx_dest.node); ipxitf_put(r->ir_intrfc); - return (0); + return 0; } /* @@ -1621,23 +1534,22 @@ static int ipxrtr_ioctl(unsigned int cmd, void *arg) int err; err = copy_from_user(&rt,arg,sizeof(rt)); - if(err) - return (-EFAULT); + if (err) + return -EFAULT; sg = (struct sockaddr_ipx *)&rt.rt_gateway; st = (struct sockaddr_ipx *)&rt.rt_dst; - if(!(rt.rt_flags & RTF_GATEWAY)) - return (-EINVAL); /* Direct routes are fixed */ - if(sg->sipx_family != AF_IPX) - return (-EINVAL); - if(st->sipx_family != AF_IPX) - return (-EINVAL); + if (!(rt.rt_flags & RTF_GATEWAY)) + return -EINVAL; /* Direct routes are fixed */ + if (sg->sipx_family != AF_IPX) + return -EINVAL; + if (st->sipx_family != AF_IPX) + return -EINVAL; - switch(cmd) - { + switch (cmd) { case SIOCDELRT: - return (ipxrtr_delete(st->sipx_network)); + return ipxrtr_delete(st->sipx_network); case SIOCADDRT: { @@ -1645,42 +1557,40 @@ static int ipxrtr_ioctl(unsigned int cmd, void *arg) f.ipx_network=st->sipx_network; f.ipx_router_network=sg->sipx_network; memcpy(f.ipx_router_node, sg->sipx_node, IPX_NODE_LEN); - return (ipxrtr_create(&f)); + return ipxrtr_create(&f); } - - default: - return (-EINVAL); } + + return -EINVAL; } static const char *ipx_frame_name(unsigned short frame) { - switch(ntohs(frame)) - { + switch (ntohs(frame)) { case ETH_P_IPX: - return ("EtherII"); + return "EtherII"; case ETH_P_802_2: - return ("802.2"); + return "802.2"; case ETH_P_SNAP: - return ("SNAP"); + return "SNAP"; case ETH_P_802_3: - return ("802.3"); + return "802.3"; case ETH_P_TR_802_2: - return ("802.2TR"); + return "802.2TR"; default: - return ("None"); + return "None"; } } static const char *ipx_device_name(ipx_interface *intrfc) { - return (intrfc->if_internal ? "Internal" : - (intrfc->if_dev ? intrfc->if_dev->name : "Unknown")); + return intrfc->if_internal ? "Internal" : + intrfc->if_dev ? intrfc->if_dev->name : "Unknown"; } /* Called from proc fs */ @@ -1700,13 +1610,12 @@ static int ipx_interface_get_info(char *buffer, char **start, off_t offset, #endif strcat(buffer+len++, "\n"); spin_lock_bh(&ipx_interfaces_lock); - for(i = ipx_interfaces; i != NULL; i = i->if_next) - { + for (i = ipx_interfaces; i; i = i->if_next) { len += sprintf(buffer+len, "%08lX ", (long unsigned int)ntohl(i->if_netnum)); len += sprintf(buffer+len,"%02X%02X%02X%02X%02X%02X ", i->if_node[0], i->if_node[1], i->if_node[2], i->if_node[3], i->if_node[4], i->if_node[5]); - len += sprintf(buffer+len, "%-9s", (i == ipx_primary_net) ? + len += sprintf(buffer+len, "%-9s", i == ipx_primary_net ? "Yes" : "No"); len += sprintf(buffer+len, "%-11s", ipx_device_name(i)); len += sprintf(buffer+len, "%-9s", @@ -1718,12 +1627,11 @@ static int ipx_interface_get_info(char *buffer, char **start, off_t offset, /* Are we still dumping unwanted data then discard the record */ pos = begin + len; - if(pos < offset) - { + if (pos < offset) { len = 0; /* Keep dumping into the buffer start */ begin = pos; } - if(pos > offset + length) /* We have dumped enough */ + if (pos > offset + length) /* We have dumped enough */ break; } spin_unlock_bh(&ipx_interfaces_lock); @@ -1731,10 +1639,10 @@ static int ipx_interface_get_info(char *buffer, char **start, off_t offset, /* The data in question runs from begin to begin+len */ *start = buffer + (offset - begin); /* Start of wanted data */ len -= (offset - begin); /* Remove unwanted header data from length */ - if(len > length) + if (len > length) len = length; /* Remove unwanted tail data from length */ - return (len); + return len; } static int ipx_get_info(char *buffer, char **start, off_t offset, int length) @@ -1755,12 +1663,10 @@ static int ipx_get_info(char *buffer, char **start, off_t offset, int length) "State", "Uid"); spin_lock_bh(&ipx_interfaces_lock); - for(i = ipx_interfaces; i != NULL; i = i->if_next) - { + for (i = ipx_interfaces; i; i = i->if_next) { ipxitf_hold(i); spin_lock_bh(&i->if_sklist_lock); - for(s = i->if_sklist; s != NULL; s = s->next) - { + for (s = i->if_sklist; s; s = s->next) { #ifdef CONFIG_IPX_INTERN len += sprintf(buffer+len, "%08lX:%02X%02X%02X%02X%02X%02X:%04X ", @@ -1778,10 +1684,9 @@ static int ipx_get_info(char *buffer, char **start, off_t offset, int length) htons(s->protinfo.af_ipx.port)); #endif /* CONFIG_IPX_INTERN */ - if(s->state != TCP_ESTABLISHED) + if (s->state != TCP_ESTABLISHED) len += sprintf(buffer+len, "%-28s", "Not_Connected"); - else - { + else { len += sprintf(buffer+len, "%08lX:%02X%02X%02X%02X%02X%02X:%04X ", (unsigned long) htonl(s->protinfo.af_ipx.dest_addr.net), @@ -1801,13 +1706,12 @@ static int ipx_get_info(char *buffer, char **start, off_t offset, int length) s->state, SOCK_INODE(s->socket)->i_uid); pos = begin + len; - if(pos < offset) - { + if (pos < offset) { len = 0; begin = pos; } - if(pos > offset + length) /* We have dumped enough */ + if (pos > offset + length) /* We have dumped enough */ break; } spin_unlock_bh(&i->if_sklist_lock); @@ -1818,10 +1722,10 @@ static int ipx_get_info(char *buffer, char **start, off_t offset, int length) /* The data in question runs from begin to begin+len */ *start = buffer + (offset-begin); len -= (offset - begin); - if(len > length) + if (len > length) len = length; - return (len); + return len; } static int ipx_rt_get_info(char *buffer, char **start, off_t offset, int length) @@ -1833,41 +1737,36 @@ static int ipx_rt_get_info(char *buffer, char **start, off_t offset, int length) len += sprintf(buffer,"%-11s%-13s%s\n", "Network", "Router_Net", "Router_Node"); read_lock_bh(&ipx_routes_lock); - for(rt = ipx_routes; rt != NULL; rt = rt->ir_next) - { + for (rt = ipx_routes; rt; rt = rt->ir_next) { len += sprintf(buffer+len,"%08lX ", (long unsigned int) ntohl(rt->ir_net)); - if(rt->ir_routed) - { + if (rt->ir_routed) { len += sprintf(buffer+len,"%08lX %02X%02X%02X%02X%02X%02X\n", (long unsigned int) ntohl(rt->ir_intrfc->if_netnum), rt->ir_router_node[0], rt->ir_router_node[1], rt->ir_router_node[2], rt->ir_router_node[3], rt->ir_router_node[4], rt->ir_router_node[5]); - } - else - { + } else { len += sprintf(buffer+len, "%-13s%s\n", "Directly", "Connected"); } pos = begin + len; - if(pos < offset) - { - len = 0; + if (pos < offset) { + len = 0; begin = pos; } - if(pos > offset + length) + if (pos > offset + length) break; } read_unlock_bh(&ipx_routes_lock); *start = buffer + (offset - begin); len -= (offset - begin); - if(len > length) + if (len > length) len = length; - return (len); + return len; } /**************************************************************************\ @@ -1877,89 +1776,80 @@ static int ipx_rt_get_info(char *buffer, char **start, off_t offset, int length) * * \**************************************************************************/ -static int ipx_setsockopt(struct socket *sock, int level, int optname, char *optval, int optlen) +static int ipx_setsockopt(struct socket *sock, int level, int optname, + char *optval, int optlen) { - struct sock *sk; - int err, opt; - - sk = sock->sk; + struct sock *sk = sock->sk; + int opt; - if(optlen != sizeof(int)) - return (-EINVAL); + if (optlen != sizeof(int)) + return -EINVAL; - err = get_user(opt, (unsigned int *)optval); - if(err) - return (err); + if (get_user(opt, (unsigned int *)optval)) + return -EFAULT; - switch(level) - { + switch (level) { case SOL_IPX: - switch(optname) - { + switch (optname) { case IPX_TYPE: sk->protinfo.af_ipx.type = opt; - return (0); + return 0; default: - return (-ENOPROTOOPT); + return -ENOPROTOOPT; } break; default: - return (-ENOPROTOOPT); + return -ENOPROTOOPT; } } static int ipx_getsockopt(struct socket *sock, int level, int optname, char *optval, int *optlen) { - struct sock *sk; - int val=0; + struct sock *sk = sock->sk; + int val = 0; int len; - sk = sock->sk; - - switch(level) - { + switch (level) { case SOL_IPX: - switch(optname) - { + switch (optname) { case IPX_TYPE: val = sk->protinfo.af_ipx.type; break; default: - return (-ENOPROTOOPT); + return -ENOPROTOOPT; } break; default: - return (-ENOPROTOOPT); + return -ENOPROTOOPT; } - if(get_user(len, optlen)) - return (-EFAULT); + if (get_user(len, optlen)) + return -EFAULT; len = min(len, sizeof(int)); - if(put_user(len, optlen)) - return (-EFAULT); + if (put_user(len, optlen)) + return -EFAULT; - if(copy_to_user(optval, &val, len)) - return (-EFAULT); + if (copy_to_user(optval, &val, len)) + return -EFAULT; - return (0); + return 0; } static int ipx_create(struct socket *sock, int protocol) { struct sock *sk; - switch(sock->type) - { + switch (sock->type) { case SOCK_DGRAM: sk = sk_alloc(PF_IPX, GFP_KERNEL, 1); - if(sk == NULL) - return (-ENOMEM); + if (!sk) + return -ENOMEM; sock->ops = &ipx_dgram_ops; break; @@ -1968,12 +1858,12 @@ static int ipx_create(struct socket *sock, int protocol) * From this point on SPX sockets are handled * by af_spx.c and the methods replaced. */ - if(spx_family_ops) - return (spx_family_ops->create(sock,protocol)); + if (spx_family_ops) + return spx_family_ops->create(sock,protocol); /* Fall through if SPX is not loaded */ case SOCK_STREAM: /* Allow higher levels to piggyback */ default: - return (-ESOCKTNOSUPPORT); + return -ESOCKTNOSUPPORT; } #ifdef IPX_REFCNT_DEBUG atomic_inc(&ipx_sock_nr); @@ -1985,31 +1875,30 @@ static int ipx_create(struct socket *sock, int protocol) sk->no_check = 1; /* Checksum off by default */ MOD_INC_USE_COUNT; - - return (0); + return 0; } static int ipx_release(struct socket *sock) { struct sock *sk = sock->sk; - if(sk == NULL) - return (0); + if (!sk) + return 0; - if(!sk->dead) + if (!sk->dead) sk->state_change(sk); sk->dead = 1; sock->sk = NULL; ipx_destroy_socket(sk); - if(sock->type == SOCK_DGRAM) + if (sock->type == SOCK_DGRAM) MOD_DEC_USE_COUNT; - return (0); + return 0; } -/* caller must hold a referente to intrfc */ +/* caller must hold a reference to intrfc */ static unsigned short ipx_first_free_socketnum(ipx_interface *intrfc) { @@ -2017,91 +1906,80 @@ static unsigned short ipx_first_free_socketnum(ipx_interface *intrfc) spin_lock_bh(&intrfc->if_sklist_lock); - if(socketNum < IPX_MIN_EPHEMERAL_SOCKET) + if (socketNum < IPX_MIN_EPHEMERAL_SOCKET) socketNum = IPX_MIN_EPHEMERAL_SOCKET; - while(__ipxitf_find_socket(intrfc, ntohs(socketNum)) != NULL) - { - if(socketNum > IPX_MAX_EPHEMERAL_SOCKET) + while (__ipxitf_find_socket(intrfc, ntohs(socketNum))) + if (socketNum > IPX_MAX_EPHEMERAL_SOCKET) socketNum = IPX_MIN_EPHEMERAL_SOCKET; else socketNum++; - } spin_unlock_bh(&intrfc->if_sklist_lock); intrfc->if_sknum = socketNum; - return (ntohs(socketNum)); + return ntohs(socketNum); } static int ipx_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) { - struct sock *sk; + struct sock *sk = sock->sk; ipx_interface *intrfc; struct sockaddr_ipx *addr = (struct sockaddr_ipx *)uaddr; int ret; - sk = sock->sk; - - if(sk->zapped == 0) - return (-EINVAL); + if (!sk->zapped) + return -EINVAL; - if(addr_len != sizeof(struct sockaddr_ipx)) - return (-EINVAL); + if (addr_len != sizeof(struct sockaddr_ipx)) + return -EINVAL; intrfc = ipxitf_find_using_net(addr->sipx_network); - if(intrfc == NULL) - return (-EADDRNOTAVAIL); + if (!intrfc) + return -EADDRNOTAVAIL; - if(addr->sipx_port == 0) - { + if (!addr->sipx_port) { addr->sipx_port = ipx_first_free_socketnum(intrfc); ret = -EINVAL; - if(addr->sipx_port == 0) + if (!addr->sipx_port) goto out; } /* protect IPX system stuff like routing/sap */ ret = -EACCES; - if(ntohs(addr->sipx_port) < IPX_MIN_EPHEMERAL_SOCKET && !capable(CAP_NET_ADMIN)) + if (ntohs(addr->sipx_port) < IPX_MIN_EPHEMERAL_SOCKET && + !capable(CAP_NET_ADMIN)) goto out; sk->protinfo.af_ipx.port = addr->sipx_port; #ifdef CONFIG_IPX_INTERN - if(intrfc == ipx_internal_net) - { + if (intrfc == ipx_internal_net) { /* The source address is to be set explicitly if the * socket is to be bound on the internal network. If a * node number 0 was specified, the default is used. */ ret = -EINVAL; - if(memcmp(addr->sipx_node,ipx_broadcast_node,IPX_NODE_LEN) == 0) + if (!memcmp(addr->sipx_node,ipx_broadcast_node,IPX_NODE_LEN)) goto out; - if(memcmp(addr->sipx_node, ipx_this_node, IPX_NODE_LEN) == 0) - { + if (!memcmp(addr->sipx_node, ipx_this_node, IPX_NODE_LEN)) memcpy(sk->protinfo.af_ipx.node, intrfc->if_node, IPX_NODE_LEN); - } else - { - memcpy(sk->protinfo.af_ipx.node, addr->sipx_node, IPX_NODE_LEN); - } + memcpy(sk->protinfo.af_ipx.node, addr->sipx_node, + IPX_NODE_LEN); ret = -EADDRINUSE; - if(ipxitf_find_internal_socket(intrfc, - sk->protinfo.af_ipx.node, - sk->protinfo.af_ipx.port) != NULL) - { + if (ipxitf_find_internal_socket(intrfc, + sk->protinfo.af_ipx.node, + sk->protinfo.af_ipx.port)) { SOCK_DEBUG(sk, "IPX: bind failed because port %X in use.\n", ntohs((int)addr->sipx_port)); goto out; } - } - else - { + } else { /* Source addresses are easy. It must be our * network:node pair for an interface routed to IPX * with the ipx routing ioctl() @@ -2111,8 +1989,7 @@ static int ipx_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) IPX_NODE_LEN); ret = -EADDRINUSE; - if(ipxitf_find_socket(intrfc, addr->sipx_port) != NULL) - { + if (ipxitf_find_socket(intrfc, addr->sipx_port)) { SOCK_DEBUG(sk, "IPX: bind failed because port %X in use.\n", ntohs((int)addr->sipx_port)); @@ -2126,8 +2003,7 @@ static int ipx_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len) an interface routed to IPX with the ipx routing ioctl() */ ret = -EADDRINUSE; - if(ipxitf_find_socket(intrfc, addr->sipx_port) != NULL) - { + if (ipxitf_find_socket(intrfc, addr->sipx_port)) { SOCK_DEBUG(sk, "IPX: bind failed because port %X in use.\n", ntohs((int)addr->sipx_port)); goto out; @@ -2153,13 +2029,12 @@ static int ipx_connect(struct socket *sock, struct sockaddr *uaddr, sk->state = TCP_CLOSE; sock->state = SS_UNCONNECTED; - if(addr_len != sizeof(*addr)) - return (-EINVAL); + if (addr_len != sizeof(*addr)) + return -EINVAL; addr = (struct sockaddr_ipx *)uaddr; /* put the autobinding in */ - if(sk->protinfo.af_ipx.port == 0) - { + if (!sk->protinfo.af_ipx.port) { struct sockaddr_ipx uaddr; int ret; @@ -2167,7 +2042,7 @@ static int ipx_connect(struct socket *sock, struct sockaddr *uaddr, uaddr.sipx_network = 0; #ifdef CONFIG_IPX_INTERN - if(sk->protinfo.af_ipx.intrfc) + if (sk->protinfo.af_ipx.intrfc) memcpy(uaddr.sipx_node, sk->protinfo.af_ipx.intrfc->if_node,IPX_NODE_LEN); else return -ENETDOWN; /* Someone zonked the iface */ @@ -2175,13 +2050,15 @@ static int ipx_connect(struct socket *sock, struct sockaddr *uaddr, ret = ipx_bind(sock, (struct sockaddr *)&uaddr, sizeof(struct sockaddr_ipx)); - if(ret != 0) - return (ret); + if (ret) + return ret; } - /* We can either connect to primary network or somewhere we can route to */ - if( !(addr->sipx_network == 0 && ipx_primary_net != NULL) && ipxrtr_lookup(addr->sipx_network) == NULL) - return (-ENETUNREACH); + /* We can either connect to primary network or somewhere + * we can route to */ + if (!(!addr->sipx_network && ipx_primary_net) && + !ipxrtr_lookup(addr->sipx_network)) + return -ENETUNREACH; sk->protinfo.af_ipx.dest_addr.net = addr->sipx_network; sk->protinfo.af_ipx.dest_addr.sock = addr->sipx_port; @@ -2189,41 +2066,34 @@ static int ipx_connect(struct socket *sock, struct sockaddr *uaddr, addr->sipx_node,IPX_NODE_LEN); sk->protinfo.af_ipx.type = addr->sipx_type; - if(sock->type == SOCK_DGRAM ) - { + if (sock->type == SOCK_DGRAM) { sock->state = SS_CONNECTED; sk->state = TCP_ESTABLISHED; } - return (0); + return 0; } static int ipx_getname(struct socket *sock, struct sockaddr *uaddr, - int *uaddr_len, int peer) + int *uaddr_len, int peer) { ipx_address *addr; struct sockaddr_ipx sipx; - struct sock *sk; - - sk = sock->sk; + struct sock *sk = sock->sk; *uaddr_len = sizeof(struct sockaddr_ipx); - if(peer) - { - if(sk->state != TCP_ESTABLISHED) - return (-ENOTCONN); + if (peer) { + if (sk->state != TCP_ESTABLISHED) + return -ENOTCONN; addr = &sk->protinfo.af_ipx.dest_addr; sipx.sipx_network = addr->net; memcpy(sipx.sipx_node,addr->node,IPX_NODE_LEN); sipx.sipx_port = addr->sock; - } - else - { - if(sk->protinfo.af_ipx.intrfc != NULL) - { + } else { + if (sk->protinfo.af_ipx.intrfc) { sipx.sipx_network=sk->protinfo.af_ipx.intrfc->if_netnum; #ifdef CONFIG_IPX_INTERN memcpy(sipx.sipx_node, sk->protinfo.af_ipx.node, IPX_NODE_LEN); @@ -2231,9 +2101,7 @@ static int ipx_getname(struct socket *sock, struct sockaddr *uaddr, memcpy(sipx.sipx_node, sk->protinfo.af_ipx.intrfc->if_node, IPX_NODE_LEN); #endif /* CONFIG_IPX_INTERN */ - } - else - { + } else { sipx.sipx_network = 0; memset(sipx.sipx_node, '\0', IPX_NODE_LEN); } @@ -2242,10 +2110,10 @@ static int ipx_getname(struct socket *sock, struct sockaddr *uaddr, } sipx.sipx_family = AF_IPX; - sipx.sipx_type = sk->protinfo.af_ipx.type; + sipx.sipx_type = sk->protinfo.af_ipx.type; memcpy(uaddr,&sipx,sizeof(sipx)); - return (0); + return 0; } int ipx_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt) @@ -2253,97 +2121,96 @@ int ipx_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt) /* NULL here for pt means the packet was looped back */ ipx_interface *intrfc; struct ipxhdr *ipx; + struct ipx_cb *cb; + u16 ipx_pktsize; int ret; - - ipx = skb->nh.ipxh; - - /* Too small? */ - if(ntohs(ipx->ipx_pktsize) < sizeof(struct ipxhdr)) - goto drop; - - /* Invalid header */ - if(ntohs(ipx->ipx_pktsize) > skb->len) - goto drop; /* Not ours */ if (skb->pkt_type == PACKET_OTHERHOST) goto drop; + + if ((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL) + goto out; + + ipx = skb->nh.ipxh; + ipx_pktsize = ntohs(ipx->ipx_pktsize); + + /* Too small or invalid header? */ + if (ipx_pktsize < sizeof(struct ipxhdr) || ipx_pktsize > skb->len) + goto drop; - if(ipx->ipx_checksum != IPX_NO_CHECKSUM) - { - if(ipx_set_checksum(ipx, ntohs(ipx->ipx_pktsize)) != ipx->ipx_checksum) - goto drop; - } + if (ipx->ipx_checksum != IPX_NO_CHECKSUM && + ipx->ipx_checksum != ipx_cksum(ipx, ipx_pktsize)) + goto drop; + + cb = (struct ipx_cb *) skb->cb; + cb->ipx_tctrl = ipx->ipx_tctrl; + cb->ipx_dest_net = ipx->ipx_dest.net; + cb->ipx_source_net = ipx->ipx_source.net; /* Determine what local ipx endpoint this is */ intrfc = ipxitf_find_using_phys(dev, pt->type); - if(intrfc == NULL) - { - if(ipxcfg_auto_create_interfaces - && ntohl(ipx->ipx_dest.net) != 0L) - { + if (!intrfc) { + if (ipxcfg_auto_create_interfaces && + ntohl(cb->ipx_dest_net)) { intrfc = ipxitf_auto_create(dev, pt->type); ipxitf_hold(intrfc); } - if(intrfc == NULL) /* Not one of ours */ + if (!intrfc) /* Not one of ours */ goto drop; } ret = ipxitf_rcv(intrfc, skb); ipxitf_put(intrfc); return ret; -drop: - kfree_skb(skb); - return (0); +drop: kfree_skb(skb); +out: return 0; } static int ipx_sendmsg(struct socket *sock, struct msghdr *msg, int len, struct scm_cookie *scm) { struct sock *sk = sock->sk; - struct sockaddr_ipx *usipx=(struct sockaddr_ipx *)msg->msg_name; + struct sockaddr_ipx *usipx = (struct sockaddr_ipx *)msg->msg_name; struct sockaddr_ipx local_sipx; int retval; int flags = msg->msg_flags; /* Socket gets bound below anyway */ -/* if(sk->zapped) - return (-EIO); */ /* Socket not bound */ - if(flags & ~MSG_DONTWAIT) - return (-EINVAL); - - if(usipx) - { - if(sk->protinfo.af_ipx.port == 0) - { +/* if (sk->zapped) + return -EIO; */ /* Socket not bound */ + if (flags & ~MSG_DONTWAIT) + return -EINVAL; + + if (usipx) { + if (!sk->protinfo.af_ipx.port) { struct sockaddr_ipx uaddr; int ret; uaddr.sipx_port = 0; uaddr.sipx_network = 0L; #ifdef CONFIG_IPX_INTERN - if(sk->protinfo.af_ipx.intrfc) - memcpy(uaddr.sipx_node, sk->protinfo.af_ipx.intrfc - ->if_node,IPX_NODE_LEN); + if (sk->protinfo.af_ipx.intrfc) + memcpy(uaddr.sipx_node, + sk->protinfo.af_ipx.intrfc->if_node, + IPX_NODE_LEN); else return -ENETDOWN; /* Someone zonked the iface */ #endif ret = ipx_bind(sock, (struct sockaddr *)&uaddr, sizeof(struct sockaddr_ipx)); - if(ret != 0) - return (ret); + if (ret) + return ret; } - if(msg->msg_namelen < sizeof(*usipx)) - return (-EINVAL); - if(usipx->sipx_family != AF_IPX) - return (-EINVAL); - } - else - { - if(sk->state != TCP_ESTABLISHED) - return (-ENOTCONN); + if (msg->msg_namelen < sizeof(*usipx)) + return -EINVAL; + if (usipx->sipx_family != AF_IPX) + return -EINVAL; + } else { + if (sk->state != TCP_ESTABLISHED) + return -ENOTCONN; usipx=&local_sipx; usipx->sipx_family = AF_IPX; @@ -2353,11 +2220,12 @@ static int ipx_sendmsg(struct socket *sock, struct msghdr *msg, int len, memcpy(usipx->sipx_node,sk->protinfo.af_ipx.dest_addr.node,IPX_NODE_LEN); } - retval = ipxrtr_route_packet(sk, usipx, msg->msg_iov, len, flags&MSG_DONTWAIT); - if(retval < 0) - return (retval); + retval = ipxrtr_route_packet(sk, usipx, msg->msg_iov, len, + flags & MSG_DONTWAIT); + if (retval < 0) + return retval; - return (len); + return len; } @@ -2371,8 +2239,7 @@ static int ipx_recvmsg(struct socket *sock, struct msghdr *msg, int size, int copied, err; /* put the autobinding in */ - if(sk->protinfo.af_ipx.port == 0) - { + if (!sk->protinfo.af_ipx.port) { struct sockaddr_ipx uaddr; int ret; @@ -2380,7 +2247,7 @@ static int ipx_recvmsg(struct socket *sock, struct msghdr *msg, int size, uaddr.sipx_network = 0; #ifdef CONFIG_IPX_INTERN - if(sk->protinfo.af_ipx.intrfc) + if (sk->protinfo.af_ipx.intrfc) memcpy(uaddr.sipx_node, sk->protinfo.af_ipx.intrfc->if_node,IPX_NODE_LEN); else return -ENETDOWN; /* Someone zonked the iface */ @@ -2388,47 +2255,45 @@ static int ipx_recvmsg(struct socket *sock, struct msghdr *msg, int size, ret = ipx_bind(sock, (struct sockaddr *)&uaddr, sizeof(struct sockaddr_ipx)); - if(ret != 0) - return (ret); + if (ret) + return ret; } - if(sk->zapped) - return (-ENOTCONN); + if (sk->zapped) + return -ENOTCONN; skb = skb_recv_datagram(sk,flags&~MSG_DONTWAIT,flags&MSG_DONTWAIT,&err); - if(!skb) + if (!skb) goto out; ipx = skb->nh.ipxh; copied = ntohs(ipx->ipx_pktsize) - sizeof(struct ipxhdr); - if(copied > size) - { + if (copied > size) { copied=size; msg->msg_flags |= MSG_TRUNC; } err = skb_copy_datagram_iovec(skb, sizeof(struct ipxhdr), msg->msg_iov, copied); - if(err) + if (err) goto out_free; sk->stamp = skb->stamp; msg->msg_namelen = sizeof(*sipx); - if(sipx) - { + if (sipx) { + struct ipx_cb *cb = (struct ipx_cb *) skb->cb; sipx->sipx_family = AF_IPX; sipx->sipx_port = ipx->ipx_source.sock; memcpy(sipx->sipx_node,ipx->ipx_source.node,IPX_NODE_LEN); - sipx->sipx_network = ipx->ipx_source.net; + sipx->sipx_network = cb->ipx_source_net; sipx->sipx_type = ipx->ipx_type; } err = copied; out_free: skb_free_datagram(sk, skb); -out: - return (err); +out: return err; } @@ -2437,40 +2302,39 @@ static int ipx_ioctl(struct socket *sock,unsigned int cmd, unsigned long arg) long amount = 0; struct sock *sk = sock->sk; - switch(cmd) - { + switch (cmd) { case TIOCOUTQ: amount = sk->sndbuf - atomic_read(&sk->wmem_alloc); - if(amount < 0) + if (amount < 0) amount = 0; - return (put_user(amount, (int *)arg)); + return put_user(amount, (int *)arg); case TIOCINQ: { - struct sk_buff *skb; + struct sk_buff *skb = skb_peek(&sk->receive_queue); /* These two are safe on a single CPU system as only user tasks fiddle here */ - if((skb = skb_peek(&sk->receive_queue)) != NULL) + if (skb) amount = skb->len - sizeof(struct ipxhdr); - return (put_user(amount, (int *)arg)); + return put_user(amount, (int *)arg); } case SIOCADDRT: case SIOCDELRT: - if(!capable(CAP_NET_ADMIN)) - return (-EPERM); - return (ipxrtr_ioctl(cmd,(void *)arg)); + if (!capable(CAP_NET_ADMIN)) + return -EPERM; + return ipxrtr_ioctl(cmd,(void *)arg); case SIOCSIFADDR: case SIOCAIPXITFCRT: case SIOCAIPXPRISLT: - if(!capable(CAP_NET_ADMIN)) - return (-EPERM); + if (!capable(CAP_NET_ADMIN)) + return -EPERM; case SIOCGIFADDR: - return (ipxitf_ioctl(cmd,(void *)arg)); + return ipxitf_ioctl(cmd,(void *)arg); case SIOCIPXCFGDATA: - return (ipxcfg_get_config_data((void *)arg)); + return ipxcfg_get_config_data((void *)arg); case SIOCIPXNCPCONN: { @@ -2479,24 +2343,24 @@ static int ipx_ioctl(struct socket *sock,unsigned int cmd, unsigned long arg) * handed to us in arg. */ if (!capable(CAP_NET_ADMIN)) - return(-EPERM); - return get_user(sk->protinfo.af_ipx.ipx_ncp_conn, (const unsigned short *)(arg)); + return -EPERM; + return get_user(sk->protinfo.af_ipx.ipx_ncp_conn, + (const unsigned short *)(arg)); } case SIOCGSTAMP: { int ret = -EINVAL; - if(sk) - { - if(sk->stamp.tv_sec == 0) - return (-ENOENT); + if (sk) { + if (!sk->stamp.tv_sec) + return -ENOENT; ret = -EFAULT; - if(!copy_to_user((void *)arg, &sk->stamp, + if (!copy_to_user((void *)arg, &sk->stamp, sizeof(struct timeval))) ret = 0; } - return (ret); + return ret; } case SIOCGIFDSTADDR: @@ -2505,14 +2369,14 @@ static int ipx_ioctl(struct socket *sock,unsigned int cmd, unsigned long arg) case SIOCSIFBRDADDR: case SIOCGIFNETMASK: case SIOCSIFNETMASK: - return (-EINVAL); + return -EINVAL; default: - return (dev_ioctl(cmd,(void *) arg)); + return dev_ioctl(cmd,(void *) arg); } /*NOT REACHED*/ - return (0); + return 0; } /* @@ -2521,19 +2385,19 @@ static int ipx_ioctl(struct socket *sock,unsigned int cmd, unsigned long arg) int ipx_register_spx(struct proto_ops **p, struct net_proto_family *spx) { - if(spx_family_ops!=NULL) + if (spx_family_ops) return -EBUSY; cli(); MOD_INC_USE_COUNT; - *p=&ipx_dgram_ops; - spx_family_ops=spx; + *p = &ipx_dgram_ops; + spx_family_ops = spx; sti(); return 0; } int ipx_unregister_spx(void) { - spx_family_ops=NULL; + spx_family_ops = NULL; MOD_DEC_USE_COUNT; return 0; } @@ -2576,7 +2440,7 @@ static struct packet_type ipx_8023_packet_type = __constant_htons(ETH_P_802_3), NULL, /* All devices */ ipx_rcv, - NULL, + (void *) 1, /* yap, I understand shared skbs :-) */ NULL, }; @@ -2585,7 +2449,7 @@ static struct packet_type ipx_dix_packet_type = __constant_htons(ETH_P_IPX), NULL, /* All devices */ ipx_rcv, - NULL, + (void *) 1, /* yap, I understand shared skbs :-) */ NULL, }; @@ -2604,12 +2468,9 @@ extern void destroy_8023_client(struct datalink_proto *); static unsigned char ipx_8022_type = 0xE0; static unsigned char ipx_snap_id[5] = { 0x0, 0x0, 0x0, 0x81, 0x37 }; - - - static int __init ipx_init(void) { - (void) sock_register(&ipx_family_ops); + sock_register(&ipx_family_ops); pEII_datalink = make_EII_client(); dev_add_pack(&ipx_dix_packet_type); @@ -2617,35 +2478,34 @@ static int __init ipx_init(void) p8023_datalink = make_8023_client(); dev_add_pack(&ipx_8023_packet_type); - if((p8022_datalink = register_8022_client(ipx_8022_type,ipx_rcv)) == NULL) + p8022_datalink = register_8022_client(ipx_8022_type,ipx_rcv); + if (!p8022_datalink) printk(KERN_CRIT "IPX: Unable to register with 802.2\n"); - if((pSNAP_datalink = register_snap_client(ipx_snap_id,ipx_rcv)) == NULL) + pSNAP_datalink = register_snap_client(ipx_snap_id,ipx_rcv); + if (!pSNAP_datalink) printk(KERN_CRIT "IPX: Unable to register with SNAP\n"); register_netdevice_notifier(&ipx_dev_notifier); - #ifdef CONFIG_PROC_FS proc_net_create("ipx", 0, ipx_get_info); proc_net_create("ipx_interface", 0, ipx_interface_get_info); proc_net_create("ipx_route", 0, ipx_rt_get_info); #endif - - printk(KERN_INFO "NET4: Linux IPX 0.42v4 for NET4.0\n"); + printk(KERN_INFO "NET4: Linux IPX 0.43 for NET4.0\n"); printk(KERN_INFO "IPX Portions Copyright (c) 1995 Caldera, Inc.\n"); printk(KERN_INFO "IPX Portions Copyright (c) 2000 Conectiva, Inc.\n"); return 0; } + module_init(ipx_init); /* Higher layers need this info to prep tx pkts */ int ipx_if_offset(unsigned long ipx_net_number) { - ipx_route *rt = NULL; - - rt = ipxrtr_lookup(ipx_net_number); + ipx_route *rt = ipxrtr_lookup(ipx_net_number); - return (rt ? rt->ir_intrfc->if_ipx_offset : -ENETUNREACH); + return rt ? rt->ir_intrfc->if_ipx_offset : -ENETUNREACH; } /* Export symbols for higher layers */ @@ -2683,24 +2543,22 @@ static void ipx_proto_finito(void) unregister_netdevice_notifier(&ipx_dev_notifier); unregister_snap_client(ipx_snap_id); - pSNAP_datalink = NULL; + pSNAP_datalink = NULL; unregister_8022_client(ipx_8022_type); - p8022_datalink = NULL; + p8022_datalink = NULL; dev_remove_pack(&ipx_8023_packet_type); destroy_8023_client(p8023_datalink); - p8023_datalink = NULL; + p8023_datalink = NULL; dev_remove_pack(&ipx_dix_packet_type); destroy_EII_client(pEII_datalink); - pEII_datalink = NULL; - - (void) sock_unregister(ipx_family_ops.family); + pEII_datalink = NULL; - return; + sock_unregister(ipx_family_ops.family); } + module_exit(ipx_proto_finito); #endif /* MODULE */ - #endif /* CONFIG_IPX || CONFIG_IPX_MODULE */ diff --git a/net/irda/af_irda.c b/net/irda/af_irda.c index e5504ee96..250b22ae9 100644 --- a/net/irda/af_irda.c +++ b/net/irda/af_irda.c @@ -2409,6 +2409,7 @@ int __init irda_proto_init(void) #endif return 0; } +module_init(irda_proto_init); /* * Function irda_proto_cleanup (void) @@ -2429,11 +2430,9 @@ void irda_proto_cleanup(void) return; } -module_init(irda_proto_init); module_exit(irda_proto_cleanup); MODULE_AUTHOR("Dag Brattli <dagb@cs.uit.no>"); MODULE_DESCRIPTION("The Linux IrDA Protocol Subsystem"); MODULE_PARM(irda_debug, "1l"); #endif /* MODULE */ - diff --git a/net/sched/cls_u32.c b/net/sched/cls_u32.c index 550519e0b..43ed8e26b 100644 --- a/net/sched/cls_u32.c +++ b/net/sched/cls_u32.c @@ -252,7 +252,7 @@ static u32 gen_new_htid(struct tc_u_common *tp_c) do { if (++tp_c->hgenerator == 0x7FF) tp_c->hgenerator = 1; - } while (i>0 && u32_lookup_ht(tp_c, (tp_c->hgenerator|0x800)<<20)); + } while (--i>0 && u32_lookup_ht(tp_c, (tp_c->hgenerator|0x800)<<20)); return i > 0 ? (tp_c->hgenerator|0x800)<<20 : 0; } diff --git a/net/sched/sch_dsmark.c b/net/sched/sch_dsmark.c index 76f2fc394..a66fd2b45 100644 --- a/net/sched/sch_dsmark.c +++ b/net/sched/sch_dsmark.c @@ -39,16 +39,22 @@ * x:y y>0 y+1 use entry [y] * ... ... ... * x:indices-1 indices use entry [indices-1] + * ... ... ... + * x:y y+1 use entry [y & (indices-1)] + * ... ... ... + * 0xffff 0x10000 use entry [indices-1] */ +#define NO_DEFAULT_INDEX (1 << 16) + struct dsmark_qdisc_data { struct Qdisc *q; struct tcf_proto *filter_list; __u8 *mask; /* "owns" the array */ __u8 *value; __u16 indices; - __u16 default_index; + __u32 default_index; /* index range is 0...0xffff */ int set_tc_index; }; @@ -217,7 +223,7 @@ static int dsmark_enqueue(struct sk_buff *skb,struct Qdisc *sch) case TC_POLICE_UNSPEC: /* fall through */ default: - if (p->default_index) + if (p->default_index != NO_DEFAULT_INDEX) skb->tc_index = p->default_index; break; }; @@ -325,14 +331,12 @@ int dsmark_init(struct Qdisc *sch,struct rtattr *opt) if (tmp & 1) return -EINVAL; } - p->default_index = 0; + p->default_index = NO_DEFAULT_INDEX; if (tb[TCA_DSMARK_DEFAULT_INDEX-1]) { if (RTA_PAYLOAD(tb[TCA_DSMARK_DEFAULT_INDEX-1]) < sizeof(__u16)) return -EINVAL; p->default_index = *(__u16 *) RTA_DATA(tb[TCA_DSMARK_DEFAULT_INDEX-1]); - if (!p->default_index || p->default_index >= p->indices) - return -EINVAL; } p->set_tc_index = !!tb[TCA_DSMARK_SET_TC_INDEX-1]; p->mask = kmalloc(p->indices*2,GFP_KERNEL); @@ -411,9 +415,11 @@ static int dsmark_dump(struct Qdisc *sch, struct sk_buff *skb) rta = (struct rtattr *) b; RTA_PUT(skb,TCA_OPTIONS,0,NULL); RTA_PUT(skb,TCA_DSMARK_INDICES,sizeof(__u16),&p->indices); - if (p->default_index) - RTA_PUT(skb,TCA_DSMARK_DEFAULT_INDEX, sizeof(__u16), - &p->default_index); + if (p->default_index != NO_DEFAULT_INDEX) { + __u16 tmp = p->default_index; + + RTA_PUT(skb,TCA_DSMARK_DEFAULT_INDEX, sizeof(__u16), &tmp); + } if (p->set_tc_index) RTA_PUT(skb, TCA_DSMARK_SET_TC_INDEX, 0, NULL); rta->rta_len = skb->tail-b; diff --git a/net/sched/sch_gred.c b/net/sched/sch_gred.c index 4983e898e..3a48c3ebf 100644 --- a/net/sched/sch_gred.c +++ b/net/sched/sch_gred.c @@ -110,12 +110,9 @@ gred_enqueue(struct sk_buff *skb, struct Qdisc* sch) unsigned long qave=0; int i=0; - if (!t->initd) { - DPRINTK("NO GRED Queues setup yet! Enqueued anyway\n"); - if (q->backlog <= q->limit) { - __skb_queue_tail(&sch->q, skb); - return NET_XMIT_DROP; /* @@@@ */ - } + if (!t->initd && skb_queue_len(&sch->q) <= sch->dev->tx_queue_len) { + D2PRINTK("NO GRED Queues setup yet! Enqueued anyway\n"); + goto do_enqueue; } @@ -179,11 +176,12 @@ gred_enqueue(struct sk_buff *skb, struct Qdisc* sch) q->qcount = -1; enqueue: if (q->backlog <= q->limit) { + q->backlog += skb->len; +do_enqueue: __skb_queue_tail(&sch->q, skb); sch->stats.backlog += skb->len; sch->stats.bytes += skb->len; sch->stats.packets++; - q->backlog += skb->len; return 0; } else { q->pdrop++; diff --git a/net/sunrpc/auth.c b/net/sunrpc/auth.c index a295ecaf7..a6af993c8 100644 --- a/net/sunrpc/auth.c +++ b/net/sunrpc/auth.c @@ -267,22 +267,26 @@ rpcauth_holdcred(struct rpc_task *task) dprintk("RPC: %4d holding %s cred %p\n", task->tk_pid, task->tk_auth->au_ops->au_name, task->tk_msg.rpc_cred); if (task->tk_msg.rpc_cred) { + spin_lock(&rpc_credcache_lock); task->tk_msg.rpc_cred->cr_count++; task->tk_msg.rpc_cred->cr_expire = jiffies + task->tk_auth->au_expire; + spin_unlock(&rpc_credcache_lock); } } void rpcauth_releasecred(struct rpc_auth *auth, struct rpc_cred *cred) { + spin_lock(&rpc_credcache_lock); if (cred != NULL && cred->cr_count > 0) { - cred->cr_count--; - if (cred->cr_flags & RPCAUTH_CRED_DEAD) { + if (!--cred->cr_count && (cred->cr_flags & RPCAUTH_CRED_DEAD)) { + spin_unlock(&rpc_credcache_lock); rpcauth_remove_credcache(auth, cred); - if (!cred->cr_count) - rpcauth_crdestroy(auth, cred); + rpcauth_crdestroy(auth, cred); + return; } } + spin_unlock(&rpc_credcache_lock); } void @@ -335,13 +339,19 @@ rpcauth_invalcred(struct rpc_task *task) { dprintk("RPC: %4d invalidating %s cred %p\n", task->tk_pid, task->tk_auth->au_ops->au_name, task->tk_msg.rpc_cred); + spin_lock(&rpc_credcache_lock); if (task->tk_msg.rpc_cred) task->tk_msg.rpc_cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE; + spin_unlock(&rpc_credcache_lock); } int rpcauth_uptodatecred(struct rpc_task *task) { - return !(task->tk_msg.rpc_cred) || + int retval; + spin_lock(&rpc_credcache_lock); + retval = !(task->tk_msg.rpc_cred) || (task->tk_msg.rpc_cred->cr_flags & RPCAUTH_CRED_UPTODATE); + spin_unlock(&rpc_credcache_lock); + return retval; } diff --git a/net/sunrpc/sunrpc_syms.c b/net/sunrpc/sunrpc_syms.c index e53759e3e..1c5fd3740 100644 --- a/net/sunrpc/sunrpc_syms.c +++ b/net/sunrpc/sunrpc_syms.c @@ -36,6 +36,7 @@ EXPORT_SYMBOL(rpciod_down); EXPORT_SYMBOL(rpciod_up); EXPORT_SYMBOL(rpc_new_task); EXPORT_SYMBOL(rpc_wake_up_status); +EXPORT_SYMBOL(rpc_release_task); /* RPC client functions */ EXPORT_SYMBOL(rpc_create_client); diff --git a/net/sysctl_net.c b/net/sysctl_net.c index bda8d54d8..61f948390 100644 --- a/net/sysctl_net.c +++ b/net/sysctl_net.c @@ -26,10 +26,6 @@ extern ctl_table ipx_table[]; extern ctl_table core_table[]; -#ifdef CONFIG_UNIX -extern ctl_table unix_table[]; -#endif - #ifdef CONFIG_NET extern ctl_table ether_table[], e802_table[]; #endif @@ -48,9 +44,6 @@ extern ctl_table econet_table[]; ctl_table net_table[] = { {NET_CORE, "core", NULL, 0, 0555, core_table}, -#ifdef CONFIG_UNIX - {NET_UNIX, "unix", NULL, 0, 0555, unix_table}, -#endif #ifdef CONFIG_NET {NET_802, "802", NULL, 0, 0555, e802_table}, {NET_ETHER, "ethernet", NULL, 0, 0555, ether_table}, diff --git a/net/unix/Makefile b/net/unix/Makefile index 88a262671..9a840afde 100644 --- a/net/unix/Makefile +++ b/net/unix/Makefile @@ -16,5 +16,3 @@ obj-$(CONFIG_SYSCTL) += sysctl_net_unix.o include $(TOPDIR)/Rules.make -tar: - tar -cvf /dev/f1 . diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c index e48b8549a..121e4d9de 100644 --- a/net/unix/af_unix.c +++ b/net/unix/af_unix.c @@ -1,5 +1,5 @@ /* - * NET3: Implementation of BSD Unix domain sockets. + * NET4: Implementation of BSD Unix domain sockets. * * Authors: Alan Cox, <alan.cox@linux.org> * @@ -8,7 +8,7 @@ * as published by the Free Software Foundation; either version * 2 of the License, or (at your option) any later version. * - * Version: $Id: af_unix.c,v 1.108 2000/11/10 04:02:04 davem Exp $ + * Version: $Id: af_unix.c,v 1.109 2001/01/06 00:42:23 davem Exp $ * * Fixes: * Linus Torvalds : Assorted bug cures. @@ -124,13 +124,12 @@ static atomic_t unix_nr_socks = ATOMIC_INIT(0); #define UNIX_ABSTRACT(sk) ((sk)->protinfo.af_unix.addr->hash!=UNIX_HASH_SIZE) /* - SMP locking strategy. - * hash table is protceted with rwlock unix_table_lock - * each socket state is protected by separate rwlock. - + * SMP locking strategy: + * hash table is protected with rwlock unix_table_lock + * each socket state is protected by separate rwlock. */ -extern __inline__ unsigned unix_hash_fold(unsigned hash) +static inline unsigned unix_hash_fold(unsigned hash) { hash ^= hash>>16; hash ^= hash>>8; @@ -139,17 +138,17 @@ extern __inline__ unsigned unix_hash_fold(unsigned hash) #define unix_peer(sk) ((sk)->pair) -extern __inline__ int unix_our_peer(unix_socket *sk, unix_socket *osk) +static inline int unix_our_peer(unix_socket *sk, unix_socket *osk) { return unix_peer(osk) == sk; } -extern __inline__ int unix_may_send(unix_socket *sk, unix_socket *osk) +static inline int unix_may_send(unix_socket *sk, unix_socket *osk) { return (unix_peer(osk) == NULL || unix_our_peer(sk, osk)); } -static __inline__ unix_socket * unix_peer_get(unix_socket *s) +static inline unix_socket * unix_peer_get(unix_socket *s) { unix_socket *peer; @@ -161,7 +160,7 @@ static __inline__ unix_socket * unix_peer_get(unix_socket *s) return peer; } -extern __inline__ void unix_release_addr(struct unix_address *addr) +extern inline void unix_release_addr(struct unix_address *addr) { if (atomic_dec_and_test(&addr->refcnt)) kfree(addr); @@ -231,14 +230,14 @@ static void __unix_insert_socket(unix_socket **list, unix_socket *sk) sock_hold(sk); } -static __inline__ void unix_remove_socket(unix_socket *sk) +static inline void unix_remove_socket(unix_socket *sk) { write_lock(&unix_table_lock); __unix_remove_socket(sk); write_unlock(&unix_table_lock); } -static __inline__ void unix_insert_socket(unix_socket **list, unix_socket *sk) +static inline void unix_insert_socket(unix_socket **list, unix_socket *sk) { write_lock(&unix_table_lock); __unix_insert_socket(list, sk); @@ -258,7 +257,7 @@ static unix_socket *__unix_find_socket_byname(struct sockaddr_un *sunname, return NULL; } -static __inline__ unix_socket * +static inline unix_socket * unix_find_socket_byname(struct sockaddr_un *sunname, int len, int type, unsigned hash) { @@ -291,7 +290,7 @@ static unix_socket *unix_find_socket_byinode(struct inode *i) return s; } -static __inline__ int unix_writable(struct sock *sk) +static inline int unix_writable(struct sock *sk) { return ((atomic_read(&sk->wmem_alloc)<<2) <= sk->sndbuf); } @@ -1823,7 +1822,7 @@ struct proto_ops unix_stream_ops = { struct proto_ops unix_dgram_ops = { family: PF_UNIX, - + release: unix_release, bind: unix_bind, connect: unix_dgram_connect, @@ -1842,20 +1841,25 @@ struct proto_ops unix_dgram_ops = { }; struct net_proto_family unix_family_ops = { - PF_UNIX, - unix_create + family: PF_UNIX, + create: unix_create }; #ifdef CONFIG_SYSCTL extern void unix_sysctl_register(void); extern void unix_sysctl_unregister(void); +#else +static inline unix_sysctl_register() {}; +static inline unix_sysctl_unregister() {}; #endif +static const char banner[] __initdata = KERN_INFO "NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.\n"; + static int __init af_unix_init(void) { struct sk_buff *dummy_skb; - - printk(KERN_INFO "NET4: Unix domain sockets 1.0/SMP for Linux NET4.0.\n"); + + printk(banner); if (sizeof(struct unix_skb_parms) > sizeof(dummy_skb->cb)) { printk(KERN_CRIT "unix_proto_init: panic\n"); @@ -1865,23 +1869,15 @@ static int __init af_unix_init(void) #ifdef CONFIG_PROC_FS create_proc_read_entry("net/unix", 0, 0, unix_read_proc, NULL); #endif - -#ifdef CONFIG_SYSCTL unix_sysctl_register(); -#endif - return 0; } static void __exit af_unix_exit(void) { sock_unregister(PF_UNIX); -#ifdef CONFIG_SYSCTL unix_sysctl_unregister(); -#endif -#ifdef CONFIG_PROC_FS remove_proc_entry("net/unix", 0); -#endif } module_init(af_unix_init); diff --git a/net/unix/sysctl_net_unix.c b/net/unix/sysctl_net_unix.c index 885119da3..eceea5c95 100644 --- a/net/unix/sysctl_net_unix.c +++ b/net/unix/sysctl_net_unix.c @@ -1,10 +1,8 @@ /* - * NET3: Sysctl interface to net af_unix subsystem. + * NET4: Sysctl interface to net af_unix subsystem. * * Authors: Mike Shaver. * - * Added /proc/sys/net/unix directory entry (empty =) ). - * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public License * as published by the Free Software Foundation; either version @@ -13,9 +11,6 @@ #include <linux/mm.h> #include <linux/sysctl.h> -#include <linux/config.h> - -#ifdef CONFIG_SYSCTL extern int sysctl_unix_max_dgram_qlen; @@ -26,20 +21,18 @@ ctl_table unix_table[] = { {0} }; -static struct ctl_table_header * unix_sysctl_header; -static struct ctl_table unix_root_table[]; -static struct ctl_table unix_net_table[]; - -ctl_table unix_root_table[] = { - {CTL_NET, "net", NULL, 0, 0555, unix_net_table}, +static ctl_table unix_net_table[] = { + {NET_UNIX, "unix", NULL, 0, 0555, unix_table}, {0} }; -ctl_table unix_net_table[] = { - {NET_UNIX, "unix", NULL, 0, 0555, unix_table}, +static ctl_table unix_root_table[] = { + {CTL_NET, "net", NULL, 0, 0555, unix_net_table}, {0} }; +static struct ctl_table_header * unix_sysctl_header; + void unix_sysctl_register(void) { unix_sysctl_header = register_sysctl_table(unix_root_table, 0); @@ -50,4 +43,3 @@ void unix_sysctl_unregister(void) unregister_sysctl_table(unix_sysctl_header); } -#endif /* CONFIG_SYSCTL */ diff --git a/net/x25/Makefile b/net/x25/Makefile index 0890eff84..d0082933f 100644 --- a/net/x25/Makefile +++ b/net/x25/Makefile @@ -17,5 +17,3 @@ obj-$(CONFIG_SYSCTL) += sysctl_net_x25.o include $(TOPDIR)/Rules.make -tar: - tar -cvf /dev/f1 . diff --git a/net/x25/af_x25.c b/net/x25/af_x25.c index c8f8c3e97..a938f3bc1 100644 --- a/net/x25/af_x25.c +++ b/net/x25/af_x25.c @@ -1084,6 +1084,9 @@ static int x25_recvmsg(struct socket *sock, struct msghdr *msg, int size, int fl msg->msg_namelen = sizeof(struct sockaddr_x25); skb_free_datagram(sk, skb); + lock_sock(sk); + x25_check_rbuf(sk); + release_sock(sk); return copied; } @@ -1258,8 +1261,8 @@ static int x25_get_info(char *buffer, char **start, off_t offset, int length) } struct net_proto_family x25_family_ops = { - AF_X25, - x25_create + family: AF_X25, + create: x25_create, }; static struct proto_ops SOCKOPS_WRAPPED(x25_proto_ops) = { @@ -1286,18 +1289,13 @@ static struct proto_ops SOCKOPS_WRAPPED(x25_proto_ops) = { SOCKOPS_WRAP(x25_proto, AF_X25); -static struct packet_type x25_packet_type = -{ - 0, /* MUTTER ntohs(ETH_P_X25),*/ - 0, /* copy */ - x25_lapb_receive_frame, - NULL, - NULL, +static struct packet_type x25_packet_type = { + type: __constant_htons(ETH_P_X25), + func: x25_lapb_receive_frame, }; struct notifier_block x25_dev_notifier = { - x25_device_event, - 0 + notifier_call: x25_device_event, }; void x25_kill_by_neigh(struct x25_neigh *neigh) @@ -1317,7 +1315,6 @@ static int __init x25_init(void) #endif /* MODULE */ sock_register(&x25_family_ops); - x25_packet_type.type = htons(ETH_P_X25); dev_add_pack(&x25_packet_type); register_netdevice_notifier(&x25_dev_notifier); diff --git a/net/x25/x25_dev.c b/net/x25/x25_dev.c index fbc781dce..1abb28358 100644 --- a/net/x25/x25_dev.c +++ b/net/x25/x25_dev.c @@ -18,7 +18,6 @@ */ #include <linux/config.h> -#if defined(CONFIG_X25) || defined(CONFIG_X25_MODULE) #include <linux/errno.h> #include <linux/types.h> #include <linux/socket.h> @@ -249,8 +248,3 @@ void x25_send_frame(struct sk_buff *skb, struct x25_neigh *neigh) dev_queue_xmit(skb); } - -#endif - - - diff --git a/net/x25/x25_facilities.c b/net/x25/x25_facilities.c index ad41d1cde..adf757e56 100644 --- a/net/x25/x25_facilities.c +++ b/net/x25/x25_facilities.c @@ -18,8 +18,6 @@ * negotiation. */ -#include <linux/config.h> -#if defined(CONFIG_X25) || defined(CONFIG_X25_MODULE) #include <linux/errno.h> #include <linux/types.h> #include <linux/socket.h> @@ -232,7 +230,3 @@ void x25_limit_facilities(struct x25_facilities *facilities, } } } - -#endif - - diff --git a/net/x25/x25_in.c b/net/x25/x25_in.c index bcb5f1cf4..73a567e89 100644 --- a/net/x25/x25_in.c +++ b/net/x25/x25_in.c @@ -22,8 +22,6 @@ * i-frames. */ -#include <linux/config.h> -#if defined(CONFIG_X25) || defined(CONFIG_X25_MODULE) #include <linux/errno.h> #include <linux/types.h> #include <linux/socket.h> @@ -369,5 +367,3 @@ int x25_backlog_rcv(struct sock *sk, struct sk_buff *skb) return 0; } - -#endif diff --git a/net/x25/x25_link.c b/net/x25/x25_link.c index d6b878371..841499325 100644 --- a/net/x25/x25_link.c +++ b/net/x25/x25_link.c @@ -20,8 +20,6 @@ * 2000-09-04 Henner Eisen dev_hold() / dev_put() for x25_neigh. */ -#include <linux/config.h> -#if defined(CONFIG_X25) || defined(CONFIG_X25_MODULE) #include <linux/errno.h> #include <linux/types.h> #include <linux/socket.h> @@ -42,9 +40,10 @@ #include <linux/fcntl.h> #include <linux/mm.h> #include <linux/interrupt.h> +#include <linux/init.h> #include <net/x25.h> -static struct x25_neigh *x25_neigh_list = NULL; +static struct x25_neigh *x25_neigh_list /* = NULL initially */; static void x25_t20timer_expiry(unsigned long); @@ -422,12 +421,11 @@ int x25_subscr_ioctl(unsigned int cmd, void *arg) return 0; } -#ifdef MODULE /* * Release all memory associated with X.25 neighbour structures. */ -void x25_link_free(void) +void __exit x25_link_free(void) { struct x25_neigh *neigh, *x25_neigh = x25_neigh_list; @@ -438,7 +436,3 @@ void x25_link_free(void) x25_remove_neigh(neigh); } } - -#endif - -#endif diff --git a/net/x25/x25_out.c b/net/x25/x25_out.c index b3ce30477..2521078a6 100644 --- a/net/x25/x25_out.c +++ b/net/x25/x25_out.c @@ -21,8 +21,6 @@ * needed cleaned seq-number fields. */ -#include <linux/config.h> -#if defined(CONFIG_X25) || defined(CONFIG_X25_MODULE) #include <linux/errno.h> #include <linux/types.h> #include <linux/socket.h> @@ -229,5 +227,3 @@ void x25_enquiry_response(struct sock *sk) x25_stop_timer(sk); } - -#endif diff --git a/net/x25/x25_route.c b/net/x25/x25_route.c index 4cb51300b..a4651699e 100644 --- a/net/x25/x25_route.c +++ b/net/x25/x25_route.c @@ -17,7 +17,6 @@ */ #include <linux/config.h> -#if defined(CONFIG_X25) || defined(CONFIG_X25_MODULE) #include <linux/errno.h> #include <linux/types.h> #include <linux/socket.h> @@ -42,9 +41,10 @@ #include <linux/mm.h> #include <linux/interrupt.h> #include <linux/notifier.h> +#include <linux/init.h> #include <net/x25.h> -static struct x25_route *x25_route_list = NULL; +static struct x25_route *x25_route_list /* = NULL initially */; /* * Add a new route. @@ -255,12 +255,10 @@ int x25_routes_get_info(char *buffer, char **start, off_t offset, int length) return len; } -#ifdef MODULE - /* * Release all memory associated with X.25 routing structures. */ -void x25_route_free(void) +void __exit x25_route_free(void) { struct x25_route *route, *x25_route = x25_route_list; @@ -271,7 +269,3 @@ void x25_route_free(void) x25_remove_route(route); } } - -#endif - -#endif diff --git a/net/x25/x25_subr.c b/net/x25/x25_subr.c index 25a700af9..635c872e5 100644 --- a/net/x25/x25_subr.c +++ b/net/x25/x25_subr.c @@ -19,8 +19,6 @@ * negotiation. */ -#include <linux/config.h> -#if defined(CONFIG_X25) || defined(CONFIG_X25_MODULE) #include <linux/errno.h> #include <linux/types.h> #include <linux/socket.h> @@ -351,4 +349,18 @@ void x25_disconnect(struct sock *sk, int reason, unsigned char cause, unsigned c sk->dead = 1; } -#endif +/* + * Clear an own-rx-busy condition and tell the peer about this, provided + * that there is a significant amount of free receive buffer space available. + */ +void x25_check_rbuf(struct sock *sk) +{ + if (atomic_read(&sk->rmem_alloc) < (sk->rcvbuf / 2) && + (sk->protinfo.x25->condition & X25_COND_OWN_RX_BUSY)) { + sk->protinfo.x25->condition &= ~X25_COND_OWN_RX_BUSY; + sk->protinfo.x25->condition &= ~X25_COND_ACK_PENDING; + sk->protinfo.x25->vl = sk->protinfo.x25->vr; + x25_write_internal(sk, X25_RR); + x25_stop_timer(sk); + } +} diff --git a/net/x25/x25_timer.c b/net/x25/x25_timer.c index 20672419e..cf9b42606 100644 --- a/net/x25/x25_timer.c +++ b/net/x25/x25_timer.c @@ -18,8 +18,6 @@ * Centralised disconnection processing. */ -#include <linux/config.h> -#if defined(CONFIG_X25) || defined(CONFIG_X25_MODULE) #include <linux/errno.h> #include <linux/types.h> #include <linux/socket.h> @@ -121,6 +119,11 @@ static void x25_heartbeat_expiry(unsigned long param) { struct sock *sk = (struct sock *)param; + bh_lock_sock(sk); + if (sk->lock.users) { /* can currently only occur in state 3 */ + goto restart_heartbeat; + } + switch (sk->protinfo.x25->state) { case X25_STATE_0: @@ -128,7 +131,7 @@ static void x25_heartbeat_expiry(unsigned long param) is accepted() it isn't 'dead' so doesn't get removed. */ if (sk->destroy || (sk->state == TCP_LISTEN && sk->dead)) { x25_destroy_socket(sk); - return; + goto unlock; } break; @@ -136,29 +139,21 @@ static void x25_heartbeat_expiry(unsigned long param) /* * Check for the state of the receive buffer. */ - if (atomic_read(&sk->rmem_alloc) < (sk->rcvbuf / 2) && - (sk->protinfo.x25->condition & X25_COND_OWN_RX_BUSY)) { - sk->protinfo.x25->condition &= ~X25_COND_OWN_RX_BUSY; - sk->protinfo.x25->condition &= ~X25_COND_ACK_PENDING; - sk->protinfo.x25->vl = sk->protinfo.x25->vr; - x25_write_internal(sk, X25_RR); - x25_stop_timer(sk); - break; - } + x25_check_rbuf(sk); break; } - + restart_heartbeat: x25_start_heartbeat(sk); + unlock: + bh_unlock_sock(sk); } /* * Timer has expired, it may have been T2, T21, T22, or T23. We can tell * by the state machine state. */ -static void x25_timer_expiry(unsigned long param) +static inline void x25_do_timer_expiry(struct sock * sk) { - struct sock *sk = (struct sock *)param; - switch (sk->protinfo.x25->state) { case X25_STATE_3: /* T2 */ @@ -181,4 +176,17 @@ static void x25_timer_expiry(unsigned long param) } } -#endif +static void x25_timer_expiry(unsigned long param) +{ + struct sock *sk = (struct sock *)param; + + bh_lock_sock(sk); + if (sk->lock.users) { /* can currently only occur in state 3 */ + if (sk->protinfo.x25->state == X25_STATE_3) { + x25_start_t2timer(sk); + } + } else { + x25_do_timer_expiry(sk); + } + bh_unlock_sock(sk); +} diff --git a/scripts/checkconfig.pl b/scripts/checkconfig.pl index fc352e436..8868471fd 100644 --- a/scripts/checkconfig.pl +++ b/scripts/checkconfig.pl @@ -14,6 +14,7 @@ foreach $file (@ARGV) # Initialize variables. my $fInComment = 0; + my $fInString = 0; my $fUseConfig = 0; my $iLinuxConfig = 0; my %configList = (); @@ -24,6 +25,10 @@ foreach $file (@ARGV) $fInComment && (s+^.*?\*/+ +o ? ($fInComment = 0) : next); m+/\*+o && (s+/\*.*?\*/+ +go, (s+/\*.*$+ +o && ($fInComment = 1))); + # Strip strings. + $fInString && (s+^.*?"+ +o ? ($fInString = 0) : next); + m+"+o && (s+".*?"+ +go, (s+".*$+ +o && ($fInString = 1))); + # Pick up definitions. if ( m/^\s*#/o ) { |