Merge with Linux 2.1.55. More bugfixes and goodies from my private

CVS archive.

203 files changed, 23642 insertions, 4743 deletions

diff --git a/arch/alpha/config.in b/arch/alpha/config.in
index 6ac472f08..71c310aa1 100644
--- a/arch/alpha/config.in
+++ b/arch/alpha/config.in
@@ -94,9 +94,9 @@ bool 'System V IPC' CONFIG_SYSVIPC
 bool 'Sysctl support' CONFIG_SYSCTL
 tristate 'Kernel support for a.out (ECOFF) binaries' CONFIG_BINFMT_AOUT
 tristate 'Kernel support for ELF binaries' CONFIG_BINFMT_ELF
+tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
 if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
-  tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
-  tristate 'Kernel support for JAVA binaries' CONFIG_BINFMT_JAVA
+  tristate 'Kernel support for JAVA binaries (obsolete)' CONFIG_BINFMT_JAVA
 fi
 tristate 'Kernel support for Linux/Intel ELF binaries' CONFIG_BINFMT_EM86
 tristate 'Parallel port support' CONFIG_PARPORT
diff --git a/arch/alpha/defconfig b/arch/alpha/defconfig
index a36ec98a9..22adc7f62 100644
--- a/arch/alpha/defconfig
+++ b/arch/alpha/defconfig
@@ -43,7 +43,7 @@ CONFIG_SYSCTL=y
 CONFIG_BINFMT_AOUT=y
 CONFIG_BINFMT_ELF=y
 CONFIG_BINFMT_EM86=y
-# CONFIG_PNP_PARPORT is not set
+# CONFIG_PARPORT is not set
 
 #
 # Plug and Play support
@@ -67,7 +67,6 @@ CONFIG_BLK_DEV_FD=y
 # CONFIG_BLK_DEV_LOOP is not set
 # CONFIG_BLK_DEV_MD is not set
 # CONFIG_BLK_DEV_RAM is not set
-# CONFIG_BLK_DEV_INITRD is not set
 # CONFIG_BLK_DEV_XD is not set
 # CONFIG_BLK_DEV_EZ is not set
 # CONFIG_BLK_DEV_HD is not set
@@ -142,6 +141,7 @@ CONFIG_SCSI_CONSTANTS=y
 # CONFIG_SCSI_NCR53C406A is not set
 # CONFIG_SCSI_NCR53C7xx is not set
 # CONFIG_SCSI_NCR53C8XX is not set
+# CONFIG_SCSI_PPA is not set
 # CONFIG_SCSI_PAS16 is not set
 # CONFIG_SCSI_QLOGIC_FAS is not set
 CONFIG_SCSI_QLOGIC_ISP=y
@@ -150,6 +150,8 @@ CONFIG_SCSI_QLOGIC_ISP=y
 # CONFIG_SCSI_T128 is not set
 # CONFIG_SCSI_U14_34F is not set
 # CONFIG_SCSI_ULTRASTOR is not set
+# CONFIG_SCSI_MESH is not set
+# CONFIG_SCSI_MAC53C94 is not set
 
 #
 # Network device support
@@ -174,6 +176,7 @@ CONFIG_DE4X5=y
 # CONFIG_NET_POCKET is not set
 # CONFIG_FDDI is not set
 # CONFIG_DLCI is not set
+# CONFIG_PLIP is not set
 # CONFIG_PPP is not set
 # CONFIG_NET_RADIO is not set
 # CONFIG_SLIP is not set
@@ -215,6 +218,7 @@ CONFIG_ISO9660_FS=y
 # CONFIG_ROMFS_FS is not set
 # CONFIG_AUTOFS_FS is not set
 # CONFIG_UFS_FS is not set
+# CONFIG_MAC_PARTITION is not set
 
 #
 # Character devices
@@ -230,12 +234,15 @@ CONFIG_MOUSE=y
 # CONFIG_MS_BUSMOUSE is not set
 CONFIG_PSMOUSE=y
 # CONFIG_82C710_MOUSE is not set
+# CONFIG_PC110_PAD is not set
 # CONFIG_UMISC is not set
 # CONFIG_QIC02_TAPE is not set
 # CONFIG_FTAPE is not set
 # CONFIG_APM is not set
 # CONFIG_WATCHDOG is not set
 # CONFIG_RTC is not set
+# CONFIG_NVRAM is not set
+# CONFIG_JOYSTICK is not set
 
 #
 # Sound
diff --git a/arch/alpha/kernel/bios32.c b/arch/alpha/kernel/bios32.c
index 16d7474d1..385fc9494 100644
--- a/arch/alpha/kernel/bios32.c
+++ b/arch/alpha/kernel/bios32.c
@@ -186,6 +186,7 @@ static void layout_dev(struct pci_dev *dev)
 		pcibios_read_config_dword(bus->number, dev->devfn, reg, &base);
 		if (!base) {
 			/* this base-address register is unused */
+			dev->base_address[(reg - PCI_BASE_ADDRESS_0)>>2] = 0;
 			continue;
 		}
 
diff --git a/arch/alpha/kernel/entry.S b/arch/alpha/kernel/entry.S
index 8f896195d..e22313036 100644
--- a/arch/alpha/kernel/entry.S
+++ b/arch/alpha/kernel/entry.S
@@ -437,8 +437,8 @@ entUnaUser:
 	stq $13,32($30)
 	stq $14,40($30)
 	stq $15,48($30)
-	bis $31,$30,$19
 	lda $8,0x3fff
+	addq $30,56,$19
 	bic $30,$8,$8
 	jsr $26,do_entUnaUser
 	ldq $9,0($30)
diff --git a/arch/alpha/kernel/traps.c b/arch/alpha/kernel/traps.c
index c3838a8bb..12bee3fbc 100644
--- a/arch/alpha/kernel/traps.c
+++ b/arch/alpha/kernel/traps.c
@@ -491,8 +491,22 @@ static inline unsigned long s_reg_to_mem (unsigned long s_reg)
 			| 1L << 0x2c | 1L << 0x2d   /* stl stq */	\
 			| 1L << 0xd )		    /* stw */
 
+#define R(x)	((size_t) &((struct pt_regs *)0)->x)
+
+static int unauser_reg_offsets[32] = {
+	R(r0), R(r1), R(r2), R(r3), R(r4), R(r5), R(r6), R(r7), R(r8),
+	/* r9 ... r15 are stored in front of regs.  */
+	-56, -48, -40, -32, -24, -16, -8,
+	R(r16), R(r17), R(r18),
+	R(r19), R(r20), R(r21), R(r22), R(r23), R(r24), R(r25), R(r26),
+	R(r27), R(r28), R(gp),
+	0, 0
+};
+
+#undef R
+
 asmlinkage void do_entUnaUser(void * va, unsigned long opcode,
-			      unsigned long reg, unsigned long * frame)
+			      unsigned long reg, struct pt_regs *regs)
 {
 	extern void alpha_write_fp_reg (unsigned long reg, unsigned long val);
 	extern unsigned long alpha_read_fp_reg (unsigned long reg);
@@ -501,12 +515,10 @@ asmlinkage void do_entUnaUser(void * va, unsigned long opcode,
 	static long last_time = 0;
 
 	unsigned long tmp1, tmp2, tmp3, tmp4;
-	unsigned long *reg_addr, *pc_addr, fake_reg;
+	unsigned long fake_reg, *reg_addr = &fake_reg;
 	unsigned long uac_bits;
 	long error;
 
-	pc_addr = frame + 7 + 20 + 1;			/* pc in PAL frame */
-
 	/* Check the UAC bits to decide what the user wants us to do
 	   with the unaliged access.  */
 
@@ -519,7 +531,7 @@ asmlinkage void do_entUnaUser(void * va, unsigned long opcode,
 			lock_kernel();
 			printk("%s(%d): unaligned trap at %016lx: %p %lx %ld\n",
 			       current->comm, current->pid,
-			       *pc_addr - 4, va, opcode, reg);
+			       regs->pc - 4, va, opcode, reg);
 			unlock_kernel();
 		}
 		last_time = jiffies;
@@ -540,51 +552,19 @@ asmlinkage void do_entUnaUser(void * va, unsigned long opcode,
 
 	++unaligned[1].count;
 	unaligned[1].va = (unsigned long)va;
-	unaligned[1].pc = *pc_addr - 4;
+	unaligned[1].pc = regs->pc - 4;
 
-	reg_addr = frame;
 	if ((1L << opcode) & OP_INT_MASK) {
 		/* it's an integer load/store */
-		switch (reg) {
-		      case 0: case 1: case 2: case 3: case 4:
-		      case 5: case 6: case 7: case 8:
-			/* v0-t7 in SAVE_ALL frame */
-			reg_addr += 7 + reg;
-			break;
-
-		      case 9: case 10: case 11: case 12:
-		      case 13: case 14: case 15:
-			/* s0-s6 in entUna frame */
-			reg_addr += (reg - 9);
-			break;
-
-		      case 16: case 17: case 18:
-			/* a0-a2 in PAL frame */
-			reg_addr += 7 + 20 + 3 + (reg - 16);
-			break;
-
-		      case 19: case 20: case 21: case 22: case 23:
-		      case 24: case 25: case 26: case 27: case 28:
-			/* a3-at in SAVE_ALL frame */
-			reg_addr += 7 + 9 + (reg - 19);
-			break;
-
-		      case 29:
-			/* gp in PAL frame */
-			reg_addr += 7 + 20 + 2;
-			break;
-
-		      case 30:
+		if (reg < 30) {
+			reg_addr = (unsigned long *)
+			  ((char *)regs + unauser_reg_offsets[reg]);
+		} else if (reg == 30) {
 			/* usp in PAL regs */
 			fake_reg = rdusp();
-			reg_addr = &fake_reg;
-			break;
-
-		      case 31:
+		} else {
 			/* zero "register" */
 			fake_reg = 0;
-			reg_addr = &fake_reg;
-			break;
 		}
 	}
 
@@ -728,7 +708,6 @@ asmlinkage void do_entUnaUser(void * va, unsigned long opcode,
 
 	case 0x26: /* sts */
 		fake_reg = s_reg_to_mem(alpha_read_fp_reg(reg));
-		reg_addr = &fake_reg;
 		/* FALLTHRU */
 
 	case 0x2c: /* stl */
@@ -763,7 +742,6 @@ asmlinkage void do_entUnaUser(void * va, unsigned long opcode,
 
 	case 0x27: /* stt */
 		fake_reg = alpha_read_fp_reg(reg);
-		reg_addr = &fake_reg;
 		/* FALLTHRU */
 
 	case 0x2d: /* stq */
@@ -807,14 +785,14 @@ asmlinkage void do_entUnaUser(void * va, unsigned long opcode,
 	return;
 
 give_sigsegv:
-	*pc_addr -= 4;  /* make pc point to faulting insn */
+	regs->pc -= 4;  /* make pc point to faulting insn */
 	lock_kernel();
 	force_sig(SIGSEGV, current);
 	unlock_kernel();
 	return;
 
 give_sigbus:
-	*pc_addr -= 4;
+	regs->pc -= 4;
 	lock_kernel();
 	force_sig(SIGBUS, current);
 	unlock_kernel();
diff --git a/arch/i386/config.in b/arch/i386/config.in
index 887407c61..50d9de2a7 100644
--- a/arch/i386/config.in
+++ b/arch/i386/config.in
@@ -23,8 +23,10 @@ comment 'General setup'
 
 bool 'Kernel math emulation' CONFIG_MATH_EMULATION
 bool 'Networking support' CONFIG_NET
-bool 'PCI bios support' CONFIG_PCI
+bool 'PCI support' CONFIG_PCI
 if [ "$CONFIG_PCI" = "y" ]; then
+  bool '   PCI BIOS support' CONFIG_PCI_BIOS
+  bool '   PCI direct access support' CONFIG_PCI_DIRECT
   if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
     bool '   PCI bridge optimization (experimental)' CONFIG_PCI_OPTIMIZE
   fi
@@ -36,7 +38,7 @@ tristate 'Kernel support for a.out binaries' CONFIG_BINFMT_AOUT
 tristate 'Kernel support for ELF binaries' CONFIG_BINFMT_ELF
 tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
 if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
-  tristate 'Kernel support for JAVA binaries' CONFIG_BINFMT_JAVA
+  tristate 'Kernel support for JAVA binaries (obsolete)' CONFIG_BINFMT_JAVA
 fi
 
 choice 'Processor type' \
diff --git a/arch/i386/defconfig b/arch/i386/defconfig
index e3e50886c..93f300de7 100644
--- a/arch/i386/defconfig
+++ b/arch/i386/defconfig
@@ -20,6 +20,8 @@ CONFIG_MODULES=y
 # CONFIG_MATH_EMULATION is not set
 CONFIG_NET=y
 CONFIG_PCI=y
+CONFIG_PCI_BIOS=y
+# CONFIG_PCI_DIRECT is not set
 # CONFIG_MCA is not set
 CONFIG_SYSVIPC=y
 CONFIG_SYSCTL=y
@@ -139,7 +141,6 @@ CONFIG_SCSI_OMIT_FLASHPOINT=y
 # CONFIG_SCSI_NCR53C406A is not set
 # CONFIG_SCSI_NCR53C7xx is not set
 # CONFIG_SCSI_NCR53C8XX is not set
-# CONFIG_SCSI_PPA is not set
 # CONFIG_SCSI_PAS16 is not set
 # CONFIG_SCSI_QLOGIC_FAS is not set
 # CONFIG_SCSI_QLOGIC_ISP is not set
@@ -160,6 +161,7 @@ CONFIG_NET_ETHERNET=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_NET_ISA is not set
 CONFIG_NET_EISA=y
 # CONFIG_PCNET32 is not set
@@ -172,7 +174,6 @@ CONFIG_EEXPRESS_PRO100=y
 # CONFIG_NET_POCKET is not set
 # CONFIG_FDDI is not set
 # CONFIG_DLCI is not set
-# CONFIG_PLIP is not set
 # CONFIG_PPP is not set
 # CONFIG_NET_RADIO is not set
 # CONFIG_SLIP is not set
@@ -214,6 +215,7 @@ CONFIG_ISO9660_FS=y
 # CONFIG_ROMFS_FS is not set
 CONFIG_AUTOFS_FS=y
 # CONFIG_UFS_FS is not set
+# CONFIG_MAC_PARTITION is not set
 
 #
 # Character devices
diff --git a/arch/i386/kernel/bios32.c b/arch/i386/kernel/bios32.c
index 8f7a796fe..d1d7a85e0 100644
--- a/arch/i386/kernel/bios32.c
+++ b/arch/i386/kernel/bios32.c
@@ -1,7 +1,7 @@
 /*
  * bios32.c - BIOS32, PCI BIOS functions.
  *
- * $Id: bios32.c,v 1.1.1.1 1997/06/01 03:16:32 ralf Exp $
+ * $Id: bios32.c,v 1.2 1997/07/20 14:57:08 ralf Exp $
  *
  * Sponsored by
  *	iX Multiuser Multitasking Magazine
@@ -61,6 +61,9 @@
  * 
  * Jun 20, 1997 : Corrected problems in "conf1" type accesses.
  *      (paubert@iram.es)
+ *
+ * Aug 2,  1997 : Split to PCI BIOS handling and direct PCI access parts
+ *	and cleaned it up...     Martin Mares <mj@atrey.karlin.mff.cuni.cz>
  */
 
 #include <linux/config.h>
@@ -75,69 +78,12 @@
 #include <asm/system.h>
 #include <asm/io.h>
 
-#define PCIBIOS_PCI_FUNCTION_ID 	0xb1XX
-#define PCIBIOS_PCI_BIOS_PRESENT 	0xb101
-#define PCIBIOS_FIND_PCI_DEVICE		0xb102
-#define PCIBIOS_FIND_PCI_CLASS_CODE	0xb103
-#define PCIBIOS_GENERATE_SPECIAL_CYCLE	0xb106
-#define PCIBIOS_READ_CONFIG_BYTE	0xb108
-#define PCIBIOS_READ_CONFIG_WORD	0xb109
-#define PCIBIOS_READ_CONFIG_DWORD	0xb10a
-#define PCIBIOS_WRITE_CONFIG_BYTE	0xb10b
-#define PCIBIOS_WRITE_CONFIG_WORD	0xb10c
-#define PCIBIOS_WRITE_CONFIG_DWORD	0xb10d
-
-
-/* BIOS32 signature: "_32_" */
-#define BIOS32_SIGNATURE	(('_' << 0) + ('3' << 8) + ('2' << 16) + ('_' << 24))
-
-/* PCI signature: "PCI " */
-#define PCI_SIGNATURE		(('P' << 0) + ('C' << 8) + ('I' << 16) + (' ' << 24))
-
-/* PCI service signature: "$PCI" */
-#define PCI_SERVICE		(('$' << 0) + ('P' << 8) + ('C' << 16) + ('I' << 24))
-
-/*
- * This is the standard structure used to identify the entry point
- * to the BIOS32 Service Directory, as documented in
- * 	Standard BIOS 32-bit Service Directory Proposal
- * 	Revision 0.4 May 24, 1993
- * 	Phoenix Technologies Ltd.
- *	Norwood, MA
- * and the PCI BIOS specification.
- */
-
-union bios32 {
-	struct {
-		unsigned long signature;	/* _32_ */
-		unsigned long entry;		/* 32 bit physical address */
-		unsigned char revision;		/* Revision level, 0 */
-		unsigned char length;		/* Length in paragraphs should be 01 */
-		unsigned char checksum;		/* All bytes must add up to zero */
-		unsigned char reserved[5]; 	/* Must be zero */
-	} fields;
-	char chars[16];
-};
-
-#ifdef CONFIG_PCI
 /*
- * Physical address of the service directory.  I don't know if we're
- * allowed to have more than one of these or not, so just in case
- * we'll make pcibios_present() take a memory start parameter and store
- * the array there.
+ * Generic PCI access -- indirect calls according to detected HW.
  */
 
-static unsigned long bios32_entry = 0;
-static struct {
-	unsigned long address;
-	unsigned short segment;
-} bios32_indirect = { 0, KERNEL_CS };
-
-
-/*
- * function table for accessing PCI configuration space
- */
 struct pci_access {
+    int pci_present;
     int (*find_device)(unsigned short, unsigned short, unsigned short, unsigned char *, unsigned char *);
     int (*find_class)(unsigned int, unsigned short, unsigned char *, unsigned char *);
     int (*read_config_byte)(unsigned char, unsigned char, unsigned char, unsigned char *);
@@ -148,363 +94,137 @@ struct pci_access {
     int (*write_config_dword)(unsigned char, unsigned char, unsigned char, unsigned int);
 };
 
-/*
- * pointer to selected PCI access function table
- */
-static struct pci_access *access_pci = NULL;
-
-
-
-/*
- * Returns the entry point for the given service, NULL on error
- */
-
-static unsigned long bios32_service(unsigned long service)
+static int pci_stub(void)
 {
-	unsigned char return_code;	/* %al */
-	unsigned long address;		/* %ebx */
-	unsigned long length;		/* %ecx */
-	unsigned long entry;		/* %edx */
-	unsigned long flags;
-
-	save_flags(flags); cli();
-	__asm__("lcall (%%edi)"
-		: "=a" (return_code),
-		  "=b" (address),
-		  "=c" (length),
-		  "=d" (entry)
-		: "0" (service),
-		  "1" (0),
-		  "D" (&bios32_indirect));
-	restore_flags(flags);
-
-	switch (return_code) {
-		case 0:
-			return address + entry;
-		case 0x80:	/* Not present */
-			printk("bios32_service(0x%lx) : not present\n", service);
-			return 0;
-		default: /* Shouldn't happen */
-			printk("bios32_service(0x%lx) : returned 0x%x, mail drew@colorado.edu\n",
-				service, return_code);
-			return 0;
-	}
+	return PCIBIOS_FUNC_NOT_SUPPORTED;
 }
 
-static long pcibios_entry = 0;
-static struct {
-	unsigned long address;
-	unsigned short segment;
-} pci_indirect = { 0, KERNEL_CS };
+static struct pci_access pci_access_none = {
+	0,		   		/* No PCI present */
+	(void *) pci_stub,	        /* No functions implemented */
+	(void *) pci_stub,
+	(void *) pci_stub,
+	(void *) pci_stub,
+	(void *) pci_stub,
+	(void *) pci_stub,
+	(void *) pci_stub,
+	(void *) pci_stub
+};
 
+static struct pci_access *access_pci = &pci_access_none;
 
-__initfunc(static int check_pcibios(void))
+int pcibios_present(void)
 {
-	unsigned long signature;
-	unsigned char present_status;
-	unsigned char major_revision;
-	unsigned char minor_revision;
-	unsigned long flags;
-	int pack;
-
-	if ((pcibios_entry = bios32_service(PCI_SERVICE))) {
-		pci_indirect.address = pcibios_entry | PAGE_OFFSET;
-
-		save_flags(flags); cli();
-		__asm__("lcall (%%edi)\n\t"
-			"jc 1f\n\t"
-			"xor %%ah, %%ah\n"
-			"1:\tshl $8, %%eax\n\t"
-			"movw %%bx, %%ax"
-			: "=d" (signature),
-			  "=a" (pack)
-			: "1" (PCIBIOS_PCI_BIOS_PRESENT),
-			  "D" (&pci_indirect)
-			: "bx", "cx");
-		restore_flags(flags);
-
-		present_status = (pack >> 16) & 0xff;
-		major_revision = (pack >> 8) & 0xff;
-		minor_revision = pack & 0xff;
-		if (present_status || (signature != PCI_SIGNATURE)) {
-			printk ("pcibios_init : %s : BIOS32 Service Directory says PCI BIOS is present,\n"
-				"	but PCI_BIOS_PRESENT subfunction fails with present status of 0x%x\n"
-				"	and signature of 0x%08lx (%c%c%c%c).  mail drew@Colorado.EDU\n",
-				(signature == PCI_SIGNATURE) ?  "WARNING" : "ERROR",
-				present_status, signature,
-				(char) (signature >>  0), (char) (signature >>  8),
-				(char) (signature >> 16), (char) (signature >> 24));
-
-			if (signature != PCI_SIGNATURE)
-				pcibios_entry = 0;
-		}
-		if (pcibios_entry) {
-			printk ("pcibios_init : PCI BIOS revision %x.%02x entry at 0x%lx\n",
-				major_revision, minor_revision, pcibios_entry);
-			return 1;
-		}
-	}
-	return 0;
+	return access_pci->pci_present;
 }
 
-
-static int pci_bios_find_class (unsigned int class_code, unsigned short index,
+int pcibios_find_class (unsigned int class_code, unsigned short index,
 	unsigned char *bus, unsigned char *device_fn)
 {
-	unsigned long bx;
-	unsigned long ret;
-	unsigned long flags;
-
-	save_flags(flags); cli();
-	__asm__ ("lcall (%%edi)\n\t"
-		"jc 1f\n\t"
-		"xor %%ah, %%ah\n"
-		"1:"
-		: "=b" (bx),
-		  "=a" (ret)
-		: "1" (PCIBIOS_FIND_PCI_CLASS_CODE),
-		  "c" (class_code),
-		  "S" ((int) index),
-		  "D" (&pci_indirect));
-	restore_flags(flags);
-	*bus = (bx >> 8) & 0xff;
-	*device_fn = bx & 0xff;
-	return (int) (ret & 0xff00) >> 8;
+	return access_pci->find_class(class_code, index, bus, device_fn);
 }
 
-
-static int pci_bios_find_device (unsigned short vendor, unsigned short device_id,
+int pcibios_find_device (unsigned short vendor, unsigned short device_id,
 	unsigned short index, unsigned char *bus, unsigned char *device_fn)
 {
-	unsigned short bx;
-	unsigned short ret;
-	unsigned long flags;
-
-	save_flags(flags); cli();
-	__asm__("lcall (%%edi)\n\t"
-		"jc 1f\n\t"
-		"xor %%ah, %%ah\n"
-		"1:"
-		: "=b" (bx),
-		  "=a" (ret)
-		: "1" (PCIBIOS_FIND_PCI_DEVICE),
-		  "c" (device_id),
-		  "d" (vendor),
-		  "S" ((int) index),
-		  "D" (&pci_indirect));
-	restore_flags(flags);
-	*bus = (bx >> 8) & 0xff;
-	*device_fn = bx & 0xff;
-	return (int) (ret & 0xff00) >> 8;
+	return access_pci->find_device(vendor, device_id, index, bus, device_fn);
 }
 
-static int pci_bios_read_config_byte(unsigned char bus,
+int pcibios_read_config_byte (unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned char *value)
 {
-	unsigned long ret;
-	unsigned long bx = (bus << 8) | device_fn;
-	unsigned long flags;
-
-	save_flags(flags); cli();
-	__asm__("lcall (%%esi)\n\t"
-		"jc 1f\n\t"
-		"xor %%ah, %%ah\n"
-		"1:"
-		: "=c" (*value),
-		  "=a" (ret)
-		: "1" (PCIBIOS_READ_CONFIG_BYTE),
-		  "b" (bx),
-		  "D" ((long) where),
-		  "S" (&pci_indirect));
-	restore_flags(flags);
-	return (int) (ret & 0xff00) >> 8;
+	return access_pci->read_config_byte(bus, device_fn, where, value);
 }
 
-static int pci_bios_read_config_word (unsigned char bus,
+int pcibios_read_config_word (unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned short *value)
 {
-	unsigned long ret;
-	unsigned long bx = (bus << 8) | device_fn;
-	unsigned long flags;
-
-	save_flags(flags); cli();
-	__asm__("lcall (%%esi)\n\t"
-		"jc 1f\n\t"
-		"xor %%ah, %%ah\n"
-		"1:"
-		: "=c" (*value),
-		  "=a" (ret)
-		: "1" (PCIBIOS_READ_CONFIG_WORD),
-		  "b" (bx),
-		  "D" ((long) where),
-		  "S" (&pci_indirect));
-	restore_flags(flags);
-	return (int) (ret & 0xff00) >> 8;
+	return access_pci->read_config_word(bus, device_fn, where, value);
 }
 
-static int pci_bios_read_config_dword (unsigned char bus,
+int pcibios_read_config_dword (unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned int *value)
 {
-	unsigned long ret;
-	unsigned long bx = (bus << 8) | device_fn;
-	unsigned long flags;
-
-	save_flags(flags); cli();
-	__asm__("lcall (%%esi)\n\t"
-		"jc 1f\n\t"
-		"xor %%ah, %%ah\n"
-		"1:"
-		: "=c" (*value),
-		  "=a" (ret)
-		: "1" (PCIBIOS_READ_CONFIG_DWORD),
-		  "b" (bx),
-		  "D" ((long) where),
-		  "S" (&pci_indirect));
-	restore_flags(flags);
-	return (int) (ret & 0xff00) >> 8;
+	return access_pci->read_config_dword(bus, device_fn, where, value);
 }
 
-static int pci_bios_write_config_byte (unsigned char bus,
+int pcibios_write_config_byte (unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned char value)
 {
-	unsigned long ret;
-	unsigned long bx = (bus << 8) | device_fn;
-	unsigned long flags;
-
-	save_flags(flags); cli();
-	__asm__("lcall (%%esi)\n\t"
-		"jc 1f\n\t"
-		"xor %%ah, %%ah\n"
-		"1:"
-		: "=a" (ret)
-		: "0" (PCIBIOS_WRITE_CONFIG_BYTE),
-		  "c" (value),
-		  "b" (bx),
-		  "D" ((long) where),
-		  "S" (&pci_indirect));
-	restore_flags(flags);
-	return (int) (ret & 0xff00) >> 8;
+	return access_pci->write_config_byte(bus, device_fn, where, value);
 }
 
-static int pci_bios_write_config_word (unsigned char bus,
+int pcibios_write_config_word (unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned short value)
 {
-	unsigned long ret;
-	unsigned long bx = (bus << 8) | device_fn;
-	unsigned long flags;
-
-	save_flags(flags); cli();
-	__asm__("lcall (%%esi)\n\t"
-		"jc 1f\n\t"
-		"xor %%ah, %%ah\n"
-		"1:"
-		: "=a" (ret)
-		: "0" (PCIBIOS_WRITE_CONFIG_WORD),
-		  "c" (value),
-		  "b" (bx),
-		  "D" ((long) where),
-		  "S" (&pci_indirect));
-	restore_flags(flags);
-	return (int) (ret & 0xff00) >> 8;
+	return access_pci->write_config_word(bus, device_fn, where, value);
 }
 
-static int pci_bios_write_config_dword (unsigned char bus,
+int pcibios_write_config_dword (unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned int value)
 {
-	unsigned long ret;
-	unsigned long bx = (bus << 8) | device_fn;
-	unsigned long flags;
-
-	save_flags(flags); cli();
-	__asm__("lcall (%%esi)\n\t"
-		"jc 1f\n\t"
-		"xor %%ah, %%ah\n"
-		"1:"
-		: "=a" (ret)
-		: "0" (PCIBIOS_WRITE_CONFIG_DWORD),
-		  "c" (value),
-		  "b" (bx),
-		  "D" ((long) where),
-		  "S" (&pci_indirect));
-	restore_flags(flags);
-	return (int) (ret & 0xff00) >> 8;
+	return access_pci->write_config_dword(bus, device_fn, where, value);
 }
 
 /*
- * function table for BIOS32 access
+ * Direct access to PCI hardware...
  */
-static struct pci_access pci_bios_access = {
-      pci_bios_find_device,
-      pci_bios_find_class,
-      pci_bios_read_config_byte,
-      pci_bios_read_config_word,
-      pci_bios_read_config_dword,
-      pci_bios_write_config_byte,
-      pci_bios_write_config_word,
-      pci_bios_write_config_dword
-};
-
-
 
 /*
  * Given the vendor and device ids, find the n'th instance of that device
  * in the system.  
  */
+
+#ifdef CONFIG_PCI_DIRECT
+
 static int pci_direct_find_device (unsigned short vendor, unsigned short device_id,
 			   unsigned short index, unsigned char *bus,
 			   unsigned char *devfn)
 {
     unsigned int curr = 0;
     struct pci_dev *dev;
-    unsigned long flags;
 
-    save_flags(flags);
     for (dev = pci_devices; dev; dev = dev->next) {
 	if (dev->vendor == vendor && dev->device == device_id) {
 	    if (curr == index) {
 		*devfn = dev->devfn;
 		*bus = dev->bus->number;
-		restore_flags(flags);
 		return PCIBIOS_SUCCESSFUL;
 	    }
 	    ++curr;
 	}
     }
-    restore_flags(flags);
     return PCIBIOS_DEVICE_NOT_FOUND;
 }
 
-
 /*
  * Given the class, find the n'th instance of that device
  * in the system.
  */
+
 static int pci_direct_find_class (unsigned int class_code, unsigned short index,
 			  unsigned char *bus, unsigned char *devfn)
 {
     unsigned int curr = 0;
     struct pci_dev *dev;
-    unsigned long flags;
 
-    save_flags(flags); cli();
     for (dev = pci_devices; dev; dev = dev->next) {
 	if (dev->class == class_code) {
 	    if (curr == index) {
 		*devfn = dev->devfn;
 		*bus = dev->bus->number;
-		restore_flags(flags);
 		return PCIBIOS_SUCCESSFUL;
 	    }
 	    ++curr;
 	}
     }
-    restore_flags(flags);
     return PCIBIOS_DEVICE_NOT_FOUND;
 }
 
 /*
  * Functions for accessing PCI configuration space with type 1 accesses
  */
+
 #define CONFIG_CMD(bus, device_fn, where)   (0x80000000 | (bus << 16) | (device_fn << 8) | (where & ~3))
 
 static int pci_conf1_read_config_byte(unsigned char bus, unsigned char device_fn,
@@ -585,10 +305,8 @@ static int pci_conf1_write_config_dword (unsigned char bus, unsigned char device
 
 #undef CONFIG_CMD
 
-/*
- * functiontable for type 1
- */
 static struct pci_access pci_direct_conf1 = {
+      1,
       pci_direct_find_device,
       pci_direct_find_class,
       pci_conf1_read_config_byte,
@@ -602,6 +320,7 @@ static struct pci_access pci_direct_conf1 = {
 /*
  * Functions for accessing PCI configuration space with type 2 accesses
  */
+
 #define IOADDR(devfn, where)   ((0xC000 | ((devfn & 0x78) << 5)) + where)
 #define FUNC(devfn)            (((devfn & 7) << 1) | 0xf0)
 
@@ -698,10 +417,8 @@ static int pci_conf2_write_config_dword (unsigned char bus, unsigned char device
 #undef IOADDR
 #undef FUNC
 
-/*
- * functiontable for type 2
- */
 static struct pci_access pci_direct_conf2 = {
+      1,
       pci_direct_find_device,
       pci_direct_find_class,
       pci_conf2_read_config_byte,
@@ -712,8 +429,7 @@ static struct pci_access pci_direct_conf2 = {
       pci_conf2_write_config_dword
 };
 
-
-__initfunc(static struct pci_access *check_direct_pci(void))
+__initfunc(static struct pci_access *pci_check_direct(void))
 {
     unsigned int tmp;
     unsigned long flags;
@@ -721,7 +437,7 @@ __initfunc(static struct pci_access *check_direct_pci(void))
     save_flags(flags); cli();
 
     /*
-     * check if configuration type 1 works
+     * Check if configuration type 1 works.
      */
     outb (0x01, 0xCFB);
     tmp = inl (0xCF8);
@@ -729,153 +445,385 @@ __initfunc(static struct pci_access *check_direct_pci(void))
     if (inl (0xCF8) == 0x80000000) {
 	outl (tmp, 0xCF8);
 	restore_flags(flags);
-	printk("pcibios_init: Using configuration type 1\n");
+	printk("PCI: Using configuration type 1\n");
 	return &pci_direct_conf1;
     }
     outl (tmp, 0xCF8);
 
     /*
-     * check if configuration type 2 works
+     * Check if configuration type 2 works.
      */
     outb (0x00, 0xCFB);
     outb (0x00, 0xCF8);
     outb (0x00, 0xCFA);
     if (inb (0xCF8) == 0x00 && inb (0xCFB) == 0x00) {
 	restore_flags(flags);
-	printk("pcibios_init: Using configuration type 2\n");
+	printk("PCI: Using configuration type 2\n");
 	return &pci_direct_conf2;
     }
     restore_flags(flags);
-    printk("pcibios_init: Not supported chipset for direct PCI access !\n");
+    printk("PCI: PCI hardware not found (i.e., not present or not supported).\n");
     return NULL;
 }
 
+#endif
 
 /*
- * access defined pcibios functions via
- * the function table
+ * BIOS32 and PCI BIOS handling.
  */
 
-int pcibios_present(void)
+#ifdef CONFIG_PCI_BIOS
+
+#define PCIBIOS_PCI_FUNCTION_ID 	0xb1XX
+#define PCIBIOS_PCI_BIOS_PRESENT 	0xb101
+#define PCIBIOS_FIND_PCI_DEVICE		0xb102
+#define PCIBIOS_FIND_PCI_CLASS_CODE	0xb103
+#define PCIBIOS_GENERATE_SPECIAL_CYCLE	0xb106
+#define PCIBIOS_READ_CONFIG_BYTE	0xb108
+#define PCIBIOS_READ_CONFIG_WORD	0xb109
+#define PCIBIOS_READ_CONFIG_DWORD	0xb10a
+#define PCIBIOS_WRITE_CONFIG_BYTE	0xb10b
+#define PCIBIOS_WRITE_CONFIG_WORD	0xb10c
+#define PCIBIOS_WRITE_CONFIG_DWORD	0xb10d
+
+/* BIOS32 signature: "_32_" */
+#define BIOS32_SIGNATURE	(('_' << 0) + ('3' << 8) + ('2' << 16) + ('_' << 24))
+
+/* PCI signature: "PCI " */
+#define PCI_SIGNATURE		(('P' << 0) + ('C' << 8) + ('I' << 16) + (' ' << 24))
+
+/* PCI service signature: "$PCI" */
+#define PCI_SERVICE		(('$' << 0) + ('P' << 8) + ('C' << 16) + ('I' << 24))
+
+/*
+ * This is the standard structure used to identify the entry point
+ * to the BIOS32 Service Directory, as documented in
+ * 	Standard BIOS 32-bit Service Directory Proposal
+ * 	Revision 0.4 May 24, 1993
+ * 	Phoenix Technologies Ltd.
+ *	Norwood, MA
+ * and the PCI BIOS specification.
+ */
+
+union bios32 {
+	struct {
+		unsigned long signature;	/* _32_ */
+		unsigned long entry;		/* 32 bit physical address */
+		unsigned char revision;		/* Revision level, 0 */
+		unsigned char length;		/* Length in paragraphs should be 01 */
+		unsigned char checksum;		/* All bytes must add up to zero */
+		unsigned char reserved[5]; 	/* Must be zero */
+	} fields;
+	char chars[16];
+};
+
+/*
+ * Physical address of the service directory.  I don't know if we're
+ * allowed to have more than one of these or not, so just in case
+ * we'll make pcibios_present() take a memory start parameter and store
+ * the array there.
+ */
+
+static unsigned long bios32_entry = 0;
+static struct {
+	unsigned long address;
+	unsigned short segment;
+} bios32_indirect = { 0, KERNEL_CS };
+
+/*
+ * Returns the entry point for the given service, NULL on error
+ */
+
+static unsigned long bios32_service(unsigned long service)
 {
-	return access_pci ? 1 : 0;
+	unsigned char return_code;	/* %al */
+	unsigned long address;		/* %ebx */
+	unsigned long length;		/* %ecx */
+	unsigned long entry;		/* %edx */
+	unsigned long flags;
+
+	save_flags(flags); cli();
+	__asm__("lcall (%%edi)"
+		: "=a" (return_code),
+		  "=b" (address),
+		  "=c" (length),
+		  "=d" (entry)
+		: "0" (service),
+		  "1" (0),
+		  "D" (&bios32_indirect));
+	restore_flags(flags);
+
+	switch (return_code) {
+		case 0:
+			return address + entry;
+		case 0x80:	/* Not present */
+			printk("bios32_service(0x%lx): not present\n", service);
+			return 0;
+		default: /* Shouldn't happen */
+			printk("bios32_service(0x%lx): returned 0x%x, mail drew@colorado.edu\n",
+				service, return_code);
+			return 0;
+	}
 }
 
-int pcibios_find_class (unsigned int class_code, unsigned short index,
+static long pcibios_entry = 0;
+static struct {
+	unsigned long address;
+	unsigned short segment;
+} pci_indirect = { 0, KERNEL_CS };
+
+__initfunc(static int check_pcibios(void))
+{
+	unsigned long signature;
+	unsigned char present_status;
+	unsigned char major_revision;
+	unsigned char minor_revision;
+	unsigned long flags;
+	int pack;
+
+	if ((pcibios_entry = bios32_service(PCI_SERVICE))) {
+		pci_indirect.address = pcibios_entry | PAGE_OFFSET;
+
+		save_flags(flags); cli();
+		__asm__("lcall (%%edi)\n\t"
+			"jc 1f\n\t"
+			"xor %%ah, %%ah\n"
+			"1:\tshl $8, %%eax\n\t"
+			"movw %%bx, %%ax"
+			: "=d" (signature),
+			  "=a" (pack)
+			: "1" (PCIBIOS_PCI_BIOS_PRESENT),
+			  "D" (&pci_indirect)
+			: "bx", "cx");
+		restore_flags(flags);
+
+		present_status = (pack >> 16) & 0xff;
+		major_revision = (pack >> 8) & 0xff;
+		minor_revision = pack & 0xff;
+		if (present_status || (signature != PCI_SIGNATURE)) {
+			printk ("PCI: %s: BIOS32 Service Directory says PCI BIOS is present,\n"
+				"	but PCI_BIOS_PRESENT subfunction fails with present status of 0x%x\n"
+				"	and signature of 0x%08lx (%c%c%c%c).  Mail drew@Colorado.EDU\n",
+				(signature == PCI_SIGNATURE) ?  "WARNING" : "ERROR",
+				present_status, signature,
+				(char) (signature >>  0), (char) (signature >>  8),
+				(char) (signature >> 16), (char) (signature >> 24));
+
+			if (signature != PCI_SIGNATURE)
+				pcibios_entry = 0;
+		}
+		if (pcibios_entry) {
+			printk ("PCI: PCI BIOS revision %x.%02x entry at 0x%lx\n",
+				major_revision, minor_revision, pcibios_entry);
+			return 1;
+		}
+	}
+	return 0;
+}
+
+static int pci_bios_find_class (unsigned int class_code, unsigned short index,
 	unsigned char *bus, unsigned char *device_fn)
 {
-   if (access_pci && access_pci->find_class)
-      return access_pci->find_class(class_code, index, bus, device_fn);
-    
-    return PCIBIOS_FUNC_NOT_SUPPORTED;
+	unsigned long bx;
+	unsigned long ret;
+	unsigned long flags;
+
+	save_flags(flags); cli();
+	__asm__ ("lcall (%%edi)\n\t"
+		"jc 1f\n\t"
+		"xor %%ah, %%ah\n"
+		"1:"
+		: "=b" (bx),
+		  "=a" (ret)
+		: "1" (PCIBIOS_FIND_PCI_CLASS_CODE),
+		  "c" (class_code),
+		  "S" ((int) index),
+		  "D" (&pci_indirect));
+	restore_flags(flags);
+	*bus = (bx >> 8) & 0xff;
+	*device_fn = bx & 0xff;
+	return (int) (ret & 0xff00) >> 8;
 }
 
-int pcibios_find_device (unsigned short vendor, unsigned short device_id,
+static int pci_bios_find_device (unsigned short vendor, unsigned short device_id,
 	unsigned short index, unsigned char *bus, unsigned char *device_fn)
 {
-    if (access_pci && access_pci->find_device)
-      return access_pci->find_device(vendor, device_id, index, bus, device_fn);
-    
-    return PCIBIOS_FUNC_NOT_SUPPORTED;
+	unsigned short bx;
+	unsigned short ret;
+	unsigned long flags;
+
+	save_flags(flags); cli();
+	__asm__("lcall (%%edi)\n\t"
+		"jc 1f\n\t"
+		"xor %%ah, %%ah\n"
+		"1:"
+		: "=b" (bx),
+		  "=a" (ret)
+		: "1" (PCIBIOS_FIND_PCI_DEVICE),
+		  "c" (device_id),
+		  "d" (vendor),
+		  "S" ((int) index),
+		  "D" (&pci_indirect));
+	restore_flags(flags);
+	*bus = (bx >> 8) & 0xff;
+	*device_fn = bx & 0xff;
+	return (int) (ret & 0xff00) >> 8;
 }
 
-int pcibios_read_config_byte (unsigned char bus,
+static int pci_bios_read_config_byte(unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned char *value)
 {
-    if (access_pci && access_pci->read_config_byte)
-      return access_pci->read_config_byte(bus, device_fn, where, value);
-    
-    return PCIBIOS_FUNC_NOT_SUPPORTED;
+	unsigned long ret;
+	unsigned long bx = (bus << 8) | device_fn;
+	unsigned long flags;
+
+	save_flags(flags); cli();
+	__asm__("lcall (%%esi)\n\t"
+		"jc 1f\n\t"
+		"xor %%ah, %%ah\n"
+		"1:"
+		: "=c" (*value),
+		  "=a" (ret)
+		: "1" (PCIBIOS_READ_CONFIG_BYTE),
+		  "b" (bx),
+		  "D" ((long) where),
+		  "S" (&pci_indirect));
+	restore_flags(flags);
+	return (int) (ret & 0xff00) >> 8;
 }
 
-int pcibios_read_config_word (unsigned char bus,
+static int pci_bios_read_config_word (unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned short *value)
 {
-    if (access_pci && access_pci->read_config_word)
-      return access_pci->read_config_word(bus, device_fn, where, value);
-    
-    return PCIBIOS_FUNC_NOT_SUPPORTED;
+	unsigned long ret;
+	unsigned long bx = (bus << 8) | device_fn;
+	unsigned long flags;
+
+	save_flags(flags); cli();
+	__asm__("lcall (%%esi)\n\t"
+		"jc 1f\n\t"
+		"xor %%ah, %%ah\n"
+		"1:"
+		: "=c" (*value),
+		  "=a" (ret)
+		: "1" (PCIBIOS_READ_CONFIG_WORD),
+		  "b" (bx),
+		  "D" ((long) where),
+		  "S" (&pci_indirect));
+	restore_flags(flags);
+	return (int) (ret & 0xff00) >> 8;
 }
 
-int pcibios_read_config_dword (unsigned char bus,
+static int pci_bios_read_config_dword (unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned int *value)
 {
-    if (access_pci && access_pci->read_config_dword)
-      return access_pci->read_config_dword(bus, device_fn, where, value);
-    
-    return PCIBIOS_FUNC_NOT_SUPPORTED;
+	unsigned long ret;
+	unsigned long bx = (bus << 8) | device_fn;
+	unsigned long flags;
+
+	save_flags(flags); cli();
+	__asm__("lcall (%%esi)\n\t"
+		"jc 1f\n\t"
+		"xor %%ah, %%ah\n"
+		"1:"
+		: "=c" (*value),
+		  "=a" (ret)
+		: "1" (PCIBIOS_READ_CONFIG_DWORD),
+		  "b" (bx),
+		  "D" ((long) where),
+		  "S" (&pci_indirect));
+	restore_flags(flags);
+	return (int) (ret & 0xff00) >> 8;
 }
 
-int pcibios_write_config_byte (unsigned char bus,
+static int pci_bios_write_config_byte (unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned char value)
 {
-    if (access_pci && access_pci->write_config_byte)
-      return access_pci->write_config_byte(bus, device_fn, where, value);
-    
-    return PCIBIOS_FUNC_NOT_SUPPORTED;
+	unsigned long ret;
+	unsigned long bx = (bus << 8) | device_fn;
+	unsigned long flags;
+
+	save_flags(flags); cli();
+	__asm__("lcall (%%esi)\n\t"
+		"jc 1f\n\t"
+		"xor %%ah, %%ah\n"
+		"1:"
+		: "=a" (ret)
+		: "0" (PCIBIOS_WRITE_CONFIG_BYTE),
+		  "c" (value),
+		  "b" (bx),
+		  "D" ((long) where),
+		  "S" (&pci_indirect));
+	restore_flags(flags);
+	return (int) (ret & 0xff00) >> 8;
 }
 
-int pcibios_write_config_word (unsigned char bus,
+static int pci_bios_write_config_word (unsigned char bus,
 	unsigned char device_fn, unsigned char where, unsigned short value)
 {
-    if (access_pci && access_pci->write_config_word)
-      return access_pci->write_config_word(bus, device_fn, where, value);
-    
-    return PCIBIOS_FUNC_NOT_SUPPORTED;    
-}
+	unsigned long ret;
+	unsigned long bx = (bus << 8) | device_fn;
+	unsigned long flags;
 
-int pcibios_write_config_dword (unsigned char bus,
-	unsigned char device_fn, unsigned char where, unsigned int value)
-{
-    if (access_pci && access_pci->write_config_dword)
-      return access_pci->write_config_dword(bus, device_fn, where, value);
-    
-    return PCIBIOS_FUNC_NOT_SUPPORTED;
+	save_flags(flags); cli();
+	__asm__("lcall (%%esi)\n\t"
+		"jc 1f\n\t"
+		"xor %%ah, %%ah\n"
+		"1:"
+		: "=a" (ret)
+		: "0" (PCIBIOS_WRITE_CONFIG_WORD),
+		  "c" (value),
+		  "b" (bx),
+		  "D" ((long) where),
+		  "S" (&pci_indirect));
+	restore_flags(flags);
+	return (int) (ret & 0xff00) >> 8;
 }
 
-const char *pcibios_strerror (int error)
+static int pci_bios_write_config_dword (unsigned char bus,
+	unsigned char device_fn, unsigned char where, unsigned int value)
 {
-	static char buf[80];
-
-	switch (error) {
-		case PCIBIOS_SUCCESSFUL:
-			return "SUCCESSFUL";
-
-		case PCIBIOS_FUNC_NOT_SUPPORTED:
-			return "FUNC_NOT_SUPPORTED";
-
-		case PCIBIOS_BAD_VENDOR_ID:
-			return "SUCCESSFUL";
-
-		case PCIBIOS_DEVICE_NOT_FOUND:
-			return "DEVICE_NOT_FOUND";
-
-		case PCIBIOS_BAD_REGISTER_NUMBER:
-			return "BAD_REGISTER_NUMBER";
-
-                case PCIBIOS_SET_FAILED:          
-			return "SET_FAILED";
-
-                case PCIBIOS_BUFFER_TOO_SMALL:    
-			return "BUFFER_TOO_SMALL";
+	unsigned long ret;
+	unsigned long bx = (bus << 8) | device_fn;
+	unsigned long flags;
 
-		default:
-			sprintf (buf, "UNKNOWN RETURN 0x%x", error);
-			return buf;
-	}
+	save_flags(flags); cli();
+	__asm__("lcall (%%esi)\n\t"
+		"jc 1f\n\t"
+		"xor %%ah, %%ah\n"
+		"1:"
+		: "=a" (ret)
+		: "0" (PCIBIOS_WRITE_CONFIG_DWORD),
+		  "c" (value),
+		  "b" (bx),
+		  "D" ((long) where),
+		  "S" (&pci_indirect));
+	restore_flags(flags);
+	return (int) (ret & 0xff00) >> 8;
 }
 
+/*
+ * Function table for BIOS32 access
+ */
 
-__initfunc(unsigned long pcibios_fixup(unsigned long mem_start, unsigned long mem_end))
-{
-    return mem_start;
-}
+static struct pci_access pci_bios_access = {
+      1,
+      pci_bios_find_device,
+      pci_bios_find_class,
+      pci_bios_read_config_byte,
+      pci_bios_read_config_word,
+      pci_bios_read_config_dword,
+      pci_bios_write_config_byte,
+      pci_bios_write_config_word,
+      pci_bios_write_config_dword
+};
 
-#endif
+/*
+ * Try to find PCI BIOS.
+ */
 
-__initfunc(unsigned long pcibios_init(unsigned long memory_start, unsigned long memory_end))
+__initfunc(static struct pci_access *pci_find_bios(void))
 {
-#ifdef CONFIG_PCI
 	union bios32 *check;
 	unsigned char sum;
 	int i, length;
@@ -884,7 +832,6 @@ __initfunc(unsigned long pcibios_init(unsigned long memory_start, unsigned long
 	 * Follow the standard procedure for locating the BIOS32 Service
 	 * directory by scanning the permissible address range from
 	 * 0xe0000 through 0xfffff for a valid BIOS32 structure.
-	 *
 	 */
 
 	for (check = (union bios32 *) __va(0xe0000);
@@ -901,24 +848,59 @@ __initfunc(unsigned long pcibios_init(unsigned long memory_start, unsigned long
 		if (sum != 0)
 			continue;
 		if (check->fields.revision != 0) {
-			printk("pcibios_init : unsupported revision %d at 0x%p, mail drew@colorado.edu\n",
+			printk("PCI: unsupported BIOS32 revision %d at 0x%p, mail drew@colorado.edu\n",
 				check->fields.revision, check);
 			continue;
 		}
-		printk ("pcibios_init : BIOS32 Service Directory structure at 0x%p\n", check);
-		if (!bios32_entry) {
-			if (check->fields.entry >= 0x100000) {
-				printk("pcibios_init: entry in high memory, trying direct PCI access\n");
- 				access_pci = check_direct_pci();
-			} else {
-				bios32_entry = check->fields.entry;
-				printk ("pcibios_init : BIOS32 Service Directory entry at 0x%lx\n", bios32_entry);
-				bios32_indirect.address = bios32_entry + PAGE_OFFSET;
-			}
+		printk ("PCI: BIOS32 Service Directory structure at 0x%p\n", check);
+		if (check->fields.entry >= 0x100000) {
+#ifdef CONFIG_PCI_DIRECT
+			printk("PCI: BIOS32 entry in high memory, trying direct PCI access.\n");
+			return pci_check_direct();
+#else
+			printk("PCI: BIOS32 entry in high memory, cannot use.\n");
+#endif
+		} else {
+			bios32_entry = check->fields.entry;
+			printk ("PCI: BIOS32 Service Directory entry at 0x%lx\n", bios32_entry);
+			bios32_indirect.address = bios32_entry + PAGE_OFFSET;
+			if (check_pcibios())
+				return &pci_bios_access;
 		}
+		break;	/* Hopefully more than one BIOS32 cannot happen... */
 	}
-	if (bios32_entry && check_pcibios())
- 		access_pci = &pci_bios_access;
+
+	return NULL;
+}
+
 #endif
+
+/*
+ * No fixup function used.
+ */
+
+__initfunc(unsigned long pcibios_fixup(unsigned long mem_start, unsigned long mem_end))
+{
+    return mem_start;
+}
+
+/*
+ * Initialization. Try all known PCI access methods.
+ */
+
+__initfunc(unsigned long pcibios_init(unsigned long memory_start, unsigned long memory_end))
+{
+	struct pci_access *a = NULL;
+
+#ifdef CONFIG_PCI_BIOS
+	a = pci_find_bios();
+#else
+#ifdef CONFIG_PCI_DIRECT
+	a = pci_check_direct();
+#endif
+#endif
+	if (a)
+		access_pci = a;
+
 	return memory_start;
 }
diff --git a/arch/i386/kernel/entry.S b/arch/i386/kernel/entry.S
index e7b9e0779..21bf04709 100644
--- a/arch/i386/kernel/entry.S
+++ b/arch/i386/kernel/entry.S
@@ -373,7 +373,7 @@ ENTRY(sys_call_table)
 	.long SYMBOL_NAME(sys_mknod)
 	.long SYMBOL_NAME(sys_chmod)		/* 15 */
 	.long SYMBOL_NAME(sys_chown)
-	.long SYMBOL_NAME(sys_break)
+	.long SYMBOL_NAME(sys_ni_syscall)				/* old break syscall holder */
 	.long SYMBOL_NAME(sys_stat)
 	.long SYMBOL_NAME(sys_lseek)
 	.long SYMBOL_NAME(sys_getpid)		/* 20 */
@@ -387,11 +387,11 @@ ENTRY(sys_call_table)
 	.long SYMBOL_NAME(sys_fstat)
 	.long SYMBOL_NAME(sys_pause)
 	.long SYMBOL_NAME(sys_utime)		/* 30 */
-	.long SYMBOL_NAME(sys_stty)
-	.long SYMBOL_NAME(sys_gtty)
+	.long SYMBOL_NAME(sys_ni_syscall)				/* old stty syscall holder */
+	.long SYMBOL_NAME(sys_ni_syscall)				/* old gtty syscall holder */
 	.long SYMBOL_NAME(sys_access)
 	.long SYMBOL_NAME(sys_nice)
-	.long SYMBOL_NAME(sys_ftime)		/* 35 */
+	.long SYMBOL_NAME(sys_ni_syscall)	/* 35 */		/* old ftime syscall holder */
 	.long SYMBOL_NAME(sys_sync)
 	.long SYMBOL_NAME(sys_kill)
 	.long SYMBOL_NAME(sys_rename)
@@ -400,7 +400,7 @@ ENTRY(sys_call_table)
 	.long SYMBOL_NAME(sys_dup)
 	.long SYMBOL_NAME(sys_pipe)
 	.long SYMBOL_NAME(sys_times)
-	.long SYMBOL_NAME(sys_prof)
+	.long SYMBOL_NAME(sys_ni_syscall)				/* old prof syscall holder */
 	.long SYMBOL_NAME(sys_brk)		/* 45 */
 	.long SYMBOL_NAME(sys_setgid)
 	.long SYMBOL_NAME(sys_getgid)
@@ -408,13 +408,13 @@ ENTRY(sys_call_table)
 	.long SYMBOL_NAME(sys_geteuid)
 	.long SYMBOL_NAME(sys_getegid)		/* 50 */
 	.long SYMBOL_NAME(sys_acct)
-	.long SYMBOL_NAME(sys_phys)
-	.long SYMBOL_NAME(sys_lock)
+	.long SYMBOL_NAME(sys_ni_syscall)				/* old phys syscall holder */
+	.long SYMBOL_NAME(sys_ni_syscall)				/* old lock syscall holder */
 	.long SYMBOL_NAME(sys_ioctl)
 	.long SYMBOL_NAME(sys_fcntl)		/* 55 */
-	.long SYMBOL_NAME(sys_mpx)
+	.long SYMBOL_NAME(sys_ni_syscall)				/* old mpx syscall holder */
 	.long SYMBOL_NAME(sys_setpgid)
-	.long SYMBOL_NAME(sys_ulimit)
+	.long SYMBOL_NAME(sys_ni_syscall)				/* old ulimit syscall holder */
 	.long SYMBOL_NAME(sys_olduname)
 	.long SYMBOL_NAME(sys_umask)		/* 60 */
 	.long SYMBOL_NAME(sys_chroot)
@@ -454,7 +454,7 @@ ENTRY(sys_call_table)
 	.long SYMBOL_NAME(sys_fchown)		/* 95 */
 	.long SYMBOL_NAME(sys_getpriority)
 	.long SYMBOL_NAME(sys_setpriority)
-	.long SYMBOL_NAME(sys_profil)
+	.long SYMBOL_NAME(sys_ni_syscall)				/* old profil syscall holder */
 	.long SYMBOL_NAME(sys_statfs)
 	.long SYMBOL_NAME(sys_fstatfs)		/* 100 */
 	.long SYMBOL_NAME(sys_ioperm)
diff --git a/arch/i386/kernel/head.S b/arch/i386/kernel/head.S
index bd4bf56cf..ef2cfa9a2 100644
--- a/arch/i386/kernel/head.S
+++ b/arch/i386/kernel/head.S
@@ -312,15 +312,16 @@ ignore_int:
 	pushl %eax
 	pushl %ecx
 	pushl %edx
+	push %es
 	push %ds
 	movl $(KERNEL_DS),%eax
 	mov %ax,%ds
 	mov %ax,%es
-	mov %ax,%fs
 	pushl $int_msg
 	call SYMBOL_NAME(printk)
 	popl %eax
 	pop %ds
+	pop %es
 	popl %edx
 	popl %ecx
 	popl %eax
diff --git a/arch/i386/kernel/i386_ksyms.c b/arch/i386/kernel/i386_ksyms.c
index a711d9a40..b740eb0ce 100644
--- a/arch/i386/kernel/i386_ksyms.c
+++ b/arch/i386/kernel/i386_ksyms.c
@@ -7,6 +7,7 @@
 #include <linux/sched.h>
 #include <linux/in6.h>
 #include <linux/interrupt.h>
+#include <linux/smp_lock.h>
 
 #include <asm/semaphore.h>
 #include <asm/processor.h>
@@ -49,6 +50,7 @@ EXPORT_SYMBOL_NOVERS(kernel_flag);
 EXPORT_SYMBOL_NOVERS(active_kernel_processor);
 EXPORT_SYMBOL(smp_invalidate_needed);
 EXPORT_SYMBOL_NOVERS(__lock_kernel);
+EXPORT_SYMBOL(lk_lockmsg);
 
 /* Global SMP irq stuff */
 EXPORT_SYMBOL(synchronize_irq);
diff --git a/arch/i386/kernel/irq.c b/arch/i386/kernel/irq.c
index 2e0f3e084..62d074508 100644
--- a/arch/i386/kernel/irq.c
+++ b/arch/i386/kernel/irq.c
@@ -42,8 +42,6 @@
 extern volatile unsigned long smp_local_timer_ticks[1+NR_CPUS];
 #endif
 
-#define CR0_NE 32
-
 unsigned int local_irq_count[NR_CPUS];
 #ifdef __SMP__
 atomic_t __intel_bh_counter;
@@ -55,6 +53,8 @@ int __intel_bh_counter;
 static unsigned int int_count[NR_CPUS][NR_IRQS] = {{0},};
 #endif
 
+atomic_t nmi_counter;
+
 /*
  * This contains the irq mask for both irq controllers
  */
@@ -199,7 +199,6 @@ static void no_action(int cpl, void *dev_id, struct pt_regs *regs) { }
  * be shot.
  */
  
-
 static void math_error_irq(int cpl, void *dev_id, struct pt_regs *regs)
 {
 	outb(0,0xF0);
@@ -224,28 +223,26 @@ static struct irqaction *irq_action[16] = {
 
 int get_irq_list(char *buf)
 {
-	int i, len = 0;
+	int i;
 	struct irqaction * action;
+	char *p = buf;
 
 	for (i = 0 ; i < NR_IRQS ; i++) {
 		action = irq_action[i];
 		if (!action) 
 			continue;
-		len += sprintf(buf+len, "%2d: %10u   %s",
+		p += sprintf(p, "%3d: %10u   %s",
 			i, kstat.interrupts[i], action->name);
 		for (action=action->next; action; action = action->next) {
-			len += sprintf(buf+len, ", %s", action->name);
+			p += sprintf(p, ", %s", action->name);
 		}
-		len += sprintf(buf+len, "\n");
+		*p++ = '\n';
 	}
-/*
- *	Linus - should you add NMI counts here ?????
- */
+	p += sprintf(p, "NMI: %10u\n", atomic_read(&nmi_counter));
 #ifdef __SMP_PROF__
-	len+=sprintf(buf+len, "IPI: %8lu received\n",
-		ipi_count);
+	p += sprintf(p, "IPI: %10lu\n", ipi_count);
 #endif		
-	return len;
+	return p - buf;
 }
 
 #ifdef __SMP_PROF__
diff --git a/arch/i386/kernel/smp.c b/arch/i386/kernel/smp.c
index d187128e1..eef030375 100644
--- a/arch/i386/kernel/smp.c
+++ b/arch/i386/kernel/smp.c
@@ -85,7 +85,7 @@ extern void update_one_process( struct task_struct *p,
  *	5AP.	Remote read is never used
  *	9AP.	XXX NEED TO CHECK WE HANDLE THIS XXX
  *	10AP.	XXX NEED TO CHECK WE HANDLE THIS XXX
- *	11AP.	Linux read the APIC between writes to avoid this, as per
+ *	11AP.	Linux reads the APIC between writes to avoid this, as per
  *		the documentation. Make sure you preserve this as it affects
  *		the C stepping chips too.
  *
@@ -887,10 +887,14 @@ __initfunc(void smp_boot_cpus(void))
 
 	/*
 	 *	Initialize the logical to physical cpu number mapping
+	 *	and the per-CPU profiling counter/multiplier
 	 */
 
-	for (i = 0; i < NR_CPUS; i++)
+	for (i = 0; i < NR_CPUS; i++) {
 		cpu_number_map[i] = -1;
+		prof_counter[i] = 1;
+		prof_multiplier[i] = 1;
+	}
 
 	/*
 	 *	Setup boot CPU information
@@ -917,15 +921,8 @@ __initfunc(void smp_boot_cpus(void))
 	 *	of here now!
 	 */
 
-	if (!smp_found_config) {
-		/*
-		 * For SMP-simulation on one CPU to work, we must initialize these 
-		 * values for the single CPU here:
-                 */
-	        prof_counter[0] = prof_multiplier[0] = 1;
-
+	if (!smp_found_config)
 		return;
-        }
 
 	/*
 	 *	Map the local APIC into kernel space
@@ -1625,7 +1622,6 @@ static unsigned int calibration_result;
 
 __initfunc(void setup_APIC_clock (void))
 {
-	int cpu = smp_processor_id();
 	unsigned long flags;
 
 	static volatile int calibration_lock;
@@ -1658,11 +1654,10 @@ __initfunc(void setup_APIC_clock (void))
 	}
 
 /*
- * Now set up the timer for real. Profiling multiplier is 1.
+ * Now set up the timer for real.
  */
-	setup_APIC_timer (calibration_result);
 
-	prof_counter[cpu] = prof_multiplier[cpu] = 1;
+	setup_APIC_timer (calibration_result);
 
 	/*
 	 * We ACK the APIC, just in case there is something pending.
diff --git a/arch/i386/kernel/sys_i386.c b/arch/i386/kernel/sys_i386.c
index b1eb2911d..f43dfe726 100644
--- a/arch/i386/kernel/sys_i386.c
+++ b/arch/i386/kernel/sys_i386.c
@@ -17,6 +17,7 @@
 #include <linux/stat.h>
 #include <linux/mman.h>
 #include <linux/file.h>
+#include <linux/utsname.h>
 
 #include <asm/uaccess.h>
 #include <asm/ipc.h>
@@ -203,3 +204,45 @@ out:
 	unlock_kernel();
 	return ret;
 }
+
+/*
+ * Old cruft
+ */
+asmlinkage int sys_uname(struct old_utsname * name)
+{
+	if (name && !copy_to_user(name, &system_utsname, sizeof (*name)))
+		return 0;
+	return -EFAULT;
+}
+
+asmlinkage int sys_olduname(struct oldold_utsname * name)
+{
+	int error;
+
+	if (!name)
+		return -EFAULT;
+	if (!access_ok(VERIFY_WRITE,name,sizeof(struct oldold_utsname)))
+		return -EFAULT;
+  
+	error = __copy_to_user(&name->sysname,&system_utsname.sysname,__OLD_UTS_LEN);
+	error -= __put_user(0,name->sysname+__OLD_UTS_LEN);
+	error -= __copy_to_user(&name->nodename,&system_utsname.nodename,__OLD_UTS_LEN);
+	error -= __put_user(0,name->nodename+__OLD_UTS_LEN);
+	error -= __copy_to_user(&name->release,&system_utsname.release,__OLD_UTS_LEN);
+	error -= __put_user(0,name->release+__OLD_UTS_LEN);
+	error -= __copy_to_user(&name->version,&system_utsname.version,__OLD_UTS_LEN);
+	error -= __put_user(0,name->version+__OLD_UTS_LEN);
+	error -= __copy_to_user(&name->machine,&system_utsname.machine,__OLD_UTS_LEN);
+	error = __put_user(0,name->machine+__OLD_UTS_LEN);
+	error = error ? -EFAULT : 0;
+
+	return error;
+}
+
+asmlinkage int sys_pause(void)
+{
+	current->state = TASK_INTERRUPTIBLE;
+	schedule();
+	return -ERESTARTNOHAND;
+}
+
diff --git a/arch/i386/kernel/traps.c b/arch/i386/kernel/traps.c
index b397fc76d..6fb18dc35 100644
--- a/arch/i386/kernel/traps.c
+++ b/arch/i386/kernel/traps.c
@@ -29,6 +29,7 @@
 #include <asm/uaccess.h>
 #include <asm/io.h>
 #include <asm/spinlock.h>
+#include <asm/atomic.h>
 
 asmlinkage int system_call(void);
 asmlinkage void lcall7(void);
@@ -271,7 +272,9 @@ static void unknown_nmi_error(unsigned char reason, struct pt_regs * regs)
 asmlinkage void do_nmi(struct pt_regs * regs, long error_code)
 {
 	unsigned char reason = inb(0x61);
+	extern atomic_t nmi_counter;
 
+	atomic_inc(&nmi_counter);
 	if (reason & 0x80)
 		mem_parity_error(reason, regs);
 	if (reason & 0x40)
@@ -348,7 +351,10 @@ asmlinkage void do_coprocessor_error(struct pt_regs * regs, long error_code)
 asmlinkage void do_spurious_interrupt_bug(struct pt_regs * regs,
 					  long error_code)
 {
+#if 0
+	/* No need to warn about this any longer. */
 	printk("Ignoring P6 Local APIC Spurious Interrupt Bug...\n");
+#endif
 }
 
 /*
diff --git a/arch/i386/lib/checksum.c b/arch/i386/lib/checksum.c
index 4a0f82440..2293a6c5e 100644
--- a/arch/i386/lib/checksum.c
+++ b/arch/i386/lib/checksum.c
@@ -111,13 +111,13 @@ unsigned int csum_partial(const unsigned char * buff, int len, unsigned int sum)
 #define SRC(y...)			\
 "	9999: "#y";			\n \
 	.section __ex_table, \"a\";	\n \
-	.long 9999b, src_access_fault	\n \
+	.long 9999b, 6001f		\n \
 	.previous"
 
 #define DST(y...)			\
 "	9999: "#y";			\n \
 	.section __ex_table, \"a\";	\n \
-	.long 9999b, dst_access_fault	\n \
+	.long 9999b, 6002f		\n \
 	.previous"
 
 unsigned int csum_partial_copy_generic (const char *src, char *dst,
@@ -203,28 +203,28 @@ unsigned int csum_partial_copy_generic (const char *src, char *dst,
 		adcl $0, %%eax
 	7:
 
-end_of_body:
+5000:
 
 # Exception handler:
 ################################################
 						#
 .section .fixup, \"a\"				#
 						#
-common_fixup:					#
+6000:						#
 						#
 	movl	%7, (%%ebx)			#
 						#
 # FIXME: do zeroing of rest of the buffer here. #
 						#
-	jmp	end_of_body			#
+	jmp	5000b				#
 						#
-src_access_fault:				#
+6001:						#
 	movl    %1, %%ebx			#
-	jmp	common_fixup			#
+	jmp	6000b				#
 						#
-dst_access_fault:				#
+6002:						#
 	movl    %2, %%ebx			#
-	jmp	common_fixup			#
+	jmp	6000b				#
 						#
 .previous					#
 						#
diff --git a/arch/i386/mm/fault.c b/arch/i386/mm/fault.c
index e4847c070..8181fd6c7 100644
--- a/arch/i386/mm/fault.c
+++ b/arch/i386/mm/fault.c
@@ -186,16 +186,15 @@ bad_area:
  *
  * First we check if it was the bootup rw-test, though..
  */
-	if (wp_works_ok < 0 && address == 0xc0000000 && (error_code & 1)) {
+	if (wp_works_ok < 0 && address == TASK_SIZE && (error_code & 1)) {
 		wp_works_ok = 1;
-		pg0[0] = pte_val(mk_pte(0, PAGE_SHARED));
+		pg0[0] = pte_val(mk_pte(TASK_SIZE, PAGE_SHARED));
 		flush_tlb();
 		goto out;
 	}
-	if (address < PAGE_SIZE) {
+	if (address < PAGE_SIZE)
 		printk(KERN_ALERT "Unable to handle kernel NULL pointer dereference");
-		pg0[0] = pte_val(mk_pte(0, PAGE_SHARED));
-	} else
+	else
 		printk(KERN_ALERT "Unable to handle kernel paging request");
 	printk(" at virtual address %08lx\n",address);
 	__asm__("movl %%cr3,%0" : "=r" (page));
diff --git a/arch/i386/mm/init.c b/arch/i386/mm/init.c
index 8f2852a48..6ed47e2ef 100644
--- a/arch/i386/mm/init.c
+++ b/arch/i386/mm/init.c
@@ -261,6 +261,43 @@ __initfunc(unsigned long paging_init(unsigned long start_mem, unsigned long end_
 	return free_area_init(start_mem, end_mem);
 }
 
+/*
+ * Test if the WP bit works in supervisor mode. It isn't supported on 386's
+ * and also on some strange 486's (NexGen etc.). All 586+'s are OK. The jumps
+ * before and after the test are here to work-around some nasty CPU bugs.
+ */
+
+__initfunc(void test_wp_bit(void))
+{
+	unsigned char tmp_reg;
+	unsigned long old = pg0[0];
+
+	printk("Checking if this processor honours the WP bit even in supervisor mode... ");
+	pg0[0] = pte_val(mk_pte(PAGE_OFFSET, PAGE_READONLY));
+	local_flush_tlb();
+	current->mm->mmap->vm_start += PAGE_SIZE;
+	__asm__ __volatile__(
+		"jmp 1f; 1:\n"
+		"movb %0,%1\n"
+		"movb %1,%0\n"
+		"jmp 1f; 1:\n"
+		:"=m" (*(char *) __va(0)),
+		 "=q" (tmp_reg)
+		:/* no inputs */
+		:"memory");
+	pg0[0] = old;
+	local_flush_tlb();
+	current->mm->mmap->vm_start -= PAGE_SIZE;
+	if (wp_works_ok < 0) {
+		wp_works_ok = 0;
+		printk("No.\n");
+#ifndef CONFIG_M386
+		panic("This kernel doesn't support CPU's with broken WP. Recompile it for a 386!");
+#endif
+	} else
+		printk("Ok.\n");
+}
+
 __initfunc(void mem_init(unsigned long start_mem, unsigned long end_mem))
 {
 	unsigned long start_low_mem = PAGE_SIZE;
@@ -339,30 +376,9 @@ __initfunc(void mem_init(unsigned long start_mem, unsigned long end_mem))
 		reservedpages << (PAGE_SHIFT-10),
 		datapages << (PAGE_SHIFT-10),
 		initpages << (PAGE_SHIFT-10));
-/* test if the WP bit is honoured in supervisor mode */
-	if (wp_works_ok < 0) {
-		unsigned char tmp_reg;
-		unsigned long old = pg0[0];
-		printk("Checking if this processor honours the WP bit even in supervisor mode... ");
-		pg0[0] = pte_val(mk_pte(PAGE_OFFSET, PAGE_READONLY));
-		local_flush_tlb();
-		current->mm->mmap->vm_start += PAGE_SIZE;
-		__asm__ __volatile__(
-			"movb %0,%1 ; movb %1,%0"
-			:"=m" (*(char *) __va(0)),
-			 "=q" (tmp_reg)
-			:/* no inputs */
-			:"memory");
-		pg0[0] = old;
-		local_flush_tlb();
-		current->mm->mmap->vm_start -= PAGE_SIZE;
-		if (wp_works_ok < 0) {
-			wp_works_ok = 0;
-			printk("No.\n");
-		} else
-			printk("Ok.\n");
-	}
-	return;
+
+	if (wp_works_ok < 0)
+		test_wp_bit();
 }
 
 void free_initmem(void)
diff --git a/arch/m68k/config.in b/arch/m68k/config.in
index 2390fb6fc..25d6ded04 100644
--- a/arch/m68k/config.in
+++ b/arch/m68k/config.in
@@ -52,9 +52,7 @@ bool 'System V IPC' CONFIG_SYSVIPC
 bool 'Sysctl support' CONFIG_SYSCTL
 tristate 'Kernel support for a.out binaries' CONFIG_BINFMT_AOUT
 tristate 'Kernel support for ELF binaries' CONFIG_BINFMT_ELF
-if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
-  tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
-fi
+tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
 
 if [ "$CONFIG_AMIGA" = "y" ]; then
   bool 'Amiga AutoConfig Identification' CONFIG_ZORRO
diff --git a/arch/m68k/kernel/Makefile b/arch/m68k/kernel/Makefile
index 67d87c1b9..a4fc10ab0 100644
--- a/arch/m68k/kernel/Makefile
+++ b/arch/m68k/kernel/Makefile
@@ -13,7 +13,7 @@
 all: kernel.o head.o
 O_TARGET := kernel.o
 O_OBJS := entry.o process.o traps.o ints.o signal.o ptrace.o \
-	  setup.o bios32.o sys_m68k.o time.o
+	  setup.o sys_m68k.o time.o
 ifdef CONFIG_VT
 O_OBJS += console.o
 endif
diff --git a/arch/m68k/kernel/process.c b/arch/m68k/kernel/process.c
index 27883a86e..f07a6a038 100644
--- a/arch/m68k/kernel/process.c
+++ b/arch/m68k/kernel/process.c
@@ -65,7 +65,7 @@ asmlinkage int sys_idle(void)
 	current->priority = -100;
 	current->counter = -100;
 	for (;;){
-		if (!need_resched)
+		if (!resched_needed())
 #if defined(CONFIG_ATARI) && !defined(CONFIG_AMIGA) && !defined(CONFIG_MAC)
 			/* block out HSYNC on the atari (falcon) */
 			__asm__("stop #0x2200" : : : "cc");
diff --git a/arch/m68k/kernel/ptrace.c b/arch/m68k/kernel/ptrace.c
index be4149cbb..6b660118a 100644
--- a/arch/m68k/kernel/ptrace.c
+++ b/arch/m68k/kernel/ptrace.c
@@ -526,7 +526,7 @@ asmlinkage void syscall_trace(void)
 		goto out;
 	current->exit_code = SIGTRAP;
 	current->state = TASK_STOPPED;
-	notify_parent(current);
+	notify_parent(current, SIGCHLD);
 	schedule();
 	/*
 	 * this isn't the same as continuing with a signal, but it will do
diff --git a/arch/m68k/kernel/signal.c b/arch/m68k/kernel/signal.c
index 52c13445d..ab887801c 100644
--- a/arch/m68k/kernel/signal.c
+++ b/arch/m68k/kernel/signal.c
@@ -441,7 +441,7 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs *regs)
 					regs->pc -= 2;
 				}
 			}
-			notify_parent(current);
+			notify_parent(current, SIGCHLD);
 			schedule();
 			if (!(signr = current->exit_code)) {
 			discard_frame:
@@ -489,7 +489,7 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs *regs)
 				current->exit_code = signr;
 				if (!(current->p_pptr->sig->action[SIGCHLD-1].sa_flags &
 				      SA_NOCLDSTOP))
-					notify_parent(current);
+					notify_parent(current, SIGCHLD);
 				schedule();
 				continue;
 
diff --git a/arch/mips/Makefile b/arch/mips/Makefile
index 516baec29..b1df65c86 100644
--- a/arch/mips/Makefile
+++ b/arch/mips/Makefile
@@ -13,21 +13,21 @@
 # Copyright (C) 1994, 1995, 1996 by Ralf Baechle
 # DECStation modifications by Paul M. Antoine, 1996
 #
-# $Id: Makefile,v 1.5 1997/07/01 08:54:40 ralf Exp $
+# $Id: Makefile,v 1.7 1997/08/30 04:51:27 ralf Exp $
 #
 
 #
 # Select the object file format to substitute into the linker script.
 #
 ifdef CONFIG_CPU_LITTLE_ENDIAN
-CROSS_COMPILE	= mipsel-linux-
+tool-prefix	= mipsel-linux-
 ifdef CONFIG_MIPS_ECOFF
 oformat		= ecoff-littlemips
 else
 oformat		= elf32-littlemips
 endif
 else
-CROSS_COMPILE	= mips-linux-
+tool-prefix	= mips-linux-
 ifdef CONFIG_MIPS_ECOFF
 oformat		= ecoff-bigmips
 else
@@ -35,6 +35,10 @@ oformat		= elf32-bigmips
 endif
 endif
 
+ifdef CONFIG_CROSSCOMPILE
+CROSS_COMPILE  = $(tool-prefix)
+endif
+
 LINKFLAGS	= -static -N
 MODFLAGS	+= -mlong-calls
 
diff --git a/arch/mips/config.in b/arch/mips/config.in
index 35903797a..9e820f29e 100644
--- a/arch/mips/config.in
+++ b/arch/mips/config.in
@@ -21,10 +21,6 @@ if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
 fi
 bool 'Support for Mips Magnum 4000' CONFIG_MIPS_MAGNUM_4000
 bool 'Support for Olivetti M700-10' CONFIG_OLIVETTI_M700
-if [ "$CONFIG_MIPS_MAGNUM_4000" = "y" -o \
-     "$CONFIG_OLIVETTI_M700" = "y" ]; then
-       define_bool CONFIG_VIDEO_G364 y
-fi
 if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
   bool 'Support for SGI workstations' CONFIG_SGI
   if [ "$CONFIG_SGI" = "y" ]; then
@@ -32,6 +28,21 @@ if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
   fi
 fi
 bool 'Support for SNI RM200 PCI' CONFIG_SNI_RM200_PCI
+
+#
+# Select some configuration options automatically for certain systems.
+#
+unset CONFIG_PCI
+unset CONFIG_MIPS_JAZZ
+unset CONFIG_VIDEO_G364
+
+if [ "$CONFIG_ALGOR_P4032" = "y" ]; then
+	define_bool CONFIG_PCI y
+fi
+if [ "$CONFIG_MIPS_MAGNUM_4000" = "y" -o \
+     "$CONFIG_OLIVETTI_M700" = "y" ]; then
+       define_bool CONFIG_VIDEO_G364 y
+fi
 if [ "$CONFIG_DESKSTATION_RPC44" = "y" -o \
      "$CONFIG_DESKSTATION_TYNE" = "y" ]; then
 	define_bool CONFIG_MIPS_ARC y
@@ -41,9 +52,7 @@ if [ "$CONFIG_ACER_PICA_61" = "y" -o \
      "$CONFIG_MIPS_MAGNUM_4000" = "y" ]; then
 	define_bool CONFIG_MIPS_JAZZ y
 fi
-unset CONFIG_PCI
-if [ "$CONFIG_SNI_RM200_PCI" = "y" -o \
-     "$CONFIG_ALGOR_P4032" = "y" ]; then
+if [ "$CONFIG_SNI_RM200_PCI" = "y" ]; then
 	define_bool CONFIG_PCI y
 fi
 endmenu
@@ -70,15 +79,19 @@ else
   define_bool CONFIG_ELF_KERNEL y
   bool 'Generate little endian code' CONFIG_CPU_LITTLE_ENDIAN
 fi
+
 if [ "$CONFIG_CPU_LITTLE_ENDIAN" = "n" ]; then
   define_bool CONFIG_BINFMT_IRIX y
   define_bool CONFIG_FORWARD_KEYBOARD y
 fi
-define_bool CONFIG_BINFMT_ELF y
 define_bool CONFIG_BINFMT_AOUT n
+define_bool CONFIG_BINFMT_ELF y
+tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
 if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
-  tristate 'Kernel support for JAVA binaries' CONFIG_BINFMT_JAVA
+  tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
+  tristate 'Kernel support for JAVA binaries (obsolete)' CONFIG_BINFMT_JAVA
 fi
+
 bool 'Networking support' CONFIG_NET
 bool 'System V IPC' CONFIG_SYSVIPC
 bool 'Sysctl support' CONFIG_SYSCTL
@@ -208,6 +221,10 @@ mainmenu_option next_comment
 comment 'Kernel hacking'
 
 #bool 'Debug kmalloc/kfree' CONFIG_DEBUG_MALLOC
+bool 'Are you using a crosscompiler' CONFIG_CROSSCOMPILE
+if [ "$CONFIG_MODULES" = "y" ]; then
+  bool ' Build fp execption handler module' CONFIG_MIPS_FPE_MODULE
+fi
 bool 'Remote GDB kernel debugging' CONFIG_REMOTE_DEBUG
 bool 'Kernel profiling support' CONFIG_PROFILE
 if [ "$CONFIG_PROFILE" = "y" ]; then
diff --git a/arch/mips/dec/decstation.S b/arch/mips/dec/decstation.S
index a5a4f9b6d..612fcea0b 100644
--- a/arch/mips/dec/decstation.S
+++ b/arch/mips/dec/decstation.S
@@ -14,6 +14,8 @@
  * 	  the end hasn't been modified to suit the DECStation's interrupts.
  *        (Paul, you need to fix this file to comply with NAPS.  Won't be
  *         too hard - Ralf)
+ *
+ * $Id:$
  */
 #include <asm/asm.h>
 #include <asm/mipsconfig.h>
@@ -223,12 +225,12 @@ loc_scsi:	PANIC("Unimplemented loc_scsi handler")
  *  comment - Ralf)
  * Andy, 6/16/95
  */
-loc_ethernet:	PANIC("Unimplemented loc_ethernet\n")
+loc_ethernet:	PANIC("Unimplemented loc_ethernet")
 
 /*
  * Keyboard interrupt, remapped to level 1
  */
-loc_keyboard:	PANIC("Unimplemented loc_keyboard\n")
+loc_keyboard:	PANIC("Unimplemented loc_keyboard")
 
 loc_mouse:	PANIC("Unimplemented loc_mouse handler")
 
@@ -291,7 +293,7 @@ ll_rtc:		PANIC("Unimplemented RTC interrupt handler")
  * Timer IRQ
  * We remap the timer irq to be more similar to a IBM compatible
  */
-ll_timer:	PANIC("Timer interrupt!\n");
+ll_timer:	PANIC("Timer interrupt!");
 /*
  * CPU count/compare IRQ (unused)
  */
diff --git a/arch/mips/dec/int-handler.S b/arch/mips/dec/int-handler.S
index 9191574d2..1706dc975 100644
--- a/arch/mips/dec/int-handler.S
+++ b/arch/mips/dec/int-handler.S
@@ -11,6 +11,8 @@
  *
  * FIXME: still plenty to do in this file, as much of the code hasn't been
  *	  modified to suit the DECStation's interrupts.
+ *
+ * $Id:$
  */
 #include <asm/asm.h>
 #include <asm/regdef.h>
@@ -114,12 +116,12 @@ loc_scsi:	PANIC("Unimplemented loc_scsi handler")
  * Only god knows why... Tell me if you find the reason!
  * Andy, 6/16/95
  */
-loc_ethernet:	PANIC("Unimplemented loc_ethernet\n")
+loc_ethernet:	PANIC("Unimplemented loc_ethernet")
 
 /*
  * Keyboard interrupt, remapped to level 1
  */
-loc_keyboard:	PANIC("Unimplemented loc_keyboard\n")
+loc_keyboard:	PANIC("Unimplemented loc_keyboard")
 
 loc_mouse:	PANIC("Unimplemented loc_mouse handler")
 
@@ -182,7 +184,7 @@ ll_rtc:		PANIC("Unimplemented RTC interrupt handler")
  * Timer IRQ
  * We remap the timer irq to be more similar to a IBM compatible
  */
-ll_timer:	PANIC("Timer interrupt!\n");
+ll_timer:	PANIC("Timer interrupt!");
 /*
  * CPU count/compare IRQ (unused)
  */
diff --git a/arch/mips/defconfig b/arch/mips/defconfig
index b1892eba6..ae3ba1f9d 100644
--- a/arch/mips/defconfig
+++ b/arch/mips/defconfig
@@ -33,12 +33,13 @@ CONFIG_CPU_R4X00=y
 #
 CONFIG_ELF_KERNEL=y
 CONFIG_CPU_LITTLE_ENDIAN=y
-CONFIG_BINFMT_ELF=y
 # CONFIG_BINFMT_AOUT is not set
+CONFIG_BINFMT_ELF=y
+# CONFIG_BINFMT_MISC is not set
 CONFIG_NET=y
 CONFIG_SYSVIPC=y
 CONFIG_SYSCTL=y
-# CONFIG_PARPORT is not set
+# CONFIG_PNP_PARPORT is not set
 
 #
 # Loadable module support
@@ -121,6 +122,7 @@ CONFIG_MIPS_JAZZ_SONIC=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_NET_ISA is not set
 CONFIG_NET_EISA=y
 CONFIG_PCNET32=y
@@ -133,7 +135,6 @@ CONFIG_PCNET32=y
 # CONFIG_NET_POCKET is not set
 # CONFIG_FDDI is not set
 # CONFIG_DLCI is not set
-# CONFIG_PLIP is not set
 # CONFIG_PPP is not set
 # CONFIG_NET_RADIO is not set
 # CONFIG_SLIP is not set
@@ -177,6 +178,7 @@ CONFIG_ISO9660_FS=y
 # CONFIG_ROMFS_FS is not set
 # CONFIG_AUTOFS_FS is not set
 # CONFIG_UFS_FS is not set
+# CONFIG_MAC_PARTITION is not set
 
 #
 # Character devices
@@ -192,6 +194,7 @@ CONFIG_SERIAL=y
 # CONFIG_APM is not set
 # CONFIG_WATCHDOG is not set
 # CONFIG_RTC is not set
+# CONFIG_NVRAM is not set
 # CONFIG_JOYSTICK is not set
 
 #
@@ -202,5 +205,6 @@ CONFIG_SERIAL=y
 #
 # Kernel hacking
 #
+# CONFIG_CROSSCOMPILE is not set
 # CONFIG_REMOTE_DEBUG is not set
 # CONFIG_PROFILE is not set
diff --git a/arch/mips/kernel/Makefile b/arch/mips/kernel/Makefile
index 21dd3610b..b3a550177 100644
--- a/arch/mips/kernel/Makefile
+++ b/arch/mips/kernel/Makefile
@@ -18,6 +18,10 @@ O_OBJS	:= branch.o process.o signal.o entry.o traps.o ptrace.o vm86.o \
 	   r2300_misc.o r2300_scall.o r2300_fpu.o r6000_fpu.o unaligned.o
 OX_OBJS := mips_ksyms.o
 
+ifdef CONFIG_MIPS_FPE_MODULE
+M_OBJS += fpe.o
+endif
+
 #
 # SGI's have very different interrupt/timer hardware.
 #
diff --git a/arch/mips/kernel/entry.S b/arch/mips/kernel/entry.S
index 00c321a35..c955b964e 100644
--- a/arch/mips/kernel/entry.S
+++ b/arch/mips/kernel/entry.S
@@ -6,6 +6,8 @@
  * for more details.
  *
  * Copyright (C) 1994, 1995 by Ralf Baechle
+ *
+ * $Id: entry.S,v 1.2 1997/08/11 04:21:31 ralf Exp $
  */
 
 /*
@@ -38,24 +40,17 @@
 		.set	noreorder
 		.set	mips3
 		.align 4
-/* XXX cli/sti ??? */
 handle_bottom_half:
-		mfc0	s3,CP0_STATUS		# Enable IRQs
-		ori	t0,s3, 0x1f
-		xori	t0,0x1e
-
 		jal	do_bottom_half
-		 mtc0	t0,CP0_STATUS
-
+		 nop
 		b	9f
-		 mtc0	s3,CP0_STATUS		# Restore old IRQ state
+		 nop
 
-reschedule:
-		jal	schedule 
+reschedule:	jal	schedule 
 		 nop
 
 EXPORT(ret_from_sys_call)
-		 lw	t0,bh_mask
+		lw	t0,bh_mask
 		lw	t1,bh_active		# unused delay slot
 		and	t0,t1
 		bnez	t0,handle_bottom_half
diff --git a/arch/mips/kernel/fpe.c b/arch/mips/kernel/fpe.c
new file mode 100644
index 000000000..46ff43bb4
--- /dev/null
+++ b/arch/mips/kernel/fpe.c
@@ -0,0 +1,55 @@
+/*
+ * The real floating point exception handler.  While it doesn't really
+ * make sense to have this in a module, it makes debugging of this code
+ * in the kernel space a lot easier.  So far this handler in the released
+ * kernel source is just a dummy.
+ *
+ * Copyright (C) 1997 Ralf Baechle
+ *
+ * $Id:$
+ */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/signal.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+
+#include <asm/branch.h>
+#include <asm/ptrace.h>
+
+MODULE_AUTHOR("Ralf Baechle <ralf@gnu.ai.mit.edu>");
+MODULE_DESCRIPTION("Experimental floating point exception handler");
+MODULE_SUPPORTED_DEVICE("MIPS FPU");
+
+static void do_fpe(struct pt_regs *regs, unsigned int fcr31)
+{
+	lock_kernel();
+#ifdef CONF_DEBUG_EXCEPTIONS
+	show_regs(regs);
+#endif
+	printk("Caught floating exception at epc == %08lx, fcr31 == %08x\n",
+	       regs->cp0_epc, fcr31);
+	if (compute_return_epc(regs))
+		goto out;
+	force_sig(SIGFPE, current);
+out:
+	unlock_kernel();
+}
+
+/*
+ * For easier experimentation we never increment/decrement
+ * the module useable counter.
+ */
+int register_fpe(void (*handler)(struct pt_regs *regs, unsigned int fcr31));
+int unregister_fpe(void (*handler)(struct pt_regs *regs, unsigned int fcr31));
+
+int init_module(void)
+{
+	return register_fpe(do_fpe);
+}
+
+void cleanup_module(void)
+{
+	unregister_fpe(do_fpe);
+}
diff --git a/arch/mips/kernel/irq.c b/arch/mips/kernel/irq.c
index b2d112373..89ff7a3b0 100644
--- a/arch/mips/kernel/irq.c
+++ b/arch/mips/kernel/irq.c
@@ -1,17 +1,10 @@
 /*
- *	linux/arch/mips/kernel/irq.c
+ * Code to handle x86 style IRQs plus some generic interrupt stuff.
  *
- *	Copyright (C) 1992 Linus Torvalds
+ * Copyright (C) 1992 Linus Torvalds
+ * Copyright (C) 1994, 1995, 1996, 1997 Ralf Baechle
  *
- * This file contains the code used by various IRQ handling routines:
- * asking for different IRQ's should be done through these routines
- * instead of just grabbing them. Thus setups with different IRQ numbers
- * shouldn't result in any weird surprises, and installing new handlers
- * should be easier.
- *
- * Mips support by Ralf Baechle and Andreas Busse
- *
- * $Id: irq.c,v 1.2 1997/07/01 08:59:07 ralf Exp $
+ * $Id: irq.c,v 1.4 1997/09/11 20:35:50 ralf Exp $
  */
 #include <linux/config.h>
 #include <linux/errno.h>
@@ -30,13 +23,9 @@
 #include <asm/bootinfo.h>
 #include <asm/io.h>
 #include <asm/irq.h>
-#include <asm/jazz.h>
 #include <asm/mipsregs.h>
 #include <asm/system.h>
 #include <asm/vector.h>
-#ifdef CONFIG_SGI
-#include <asm/sgialib.h>
-#endif
 
 unsigned char cache_21 = 0xff;
 unsigned char cache_A1 = 0xff;
@@ -51,9 +40,6 @@ unsigned long spurious_count = 0;
 static inline void mask_irq(unsigned int irq_nr)
 {
 	unsigned char mask;
-    
-        if (irq_nr >= 16)
-	    return;
 
 	mask = 1 << (irq_nr & 7);
 	if (irq_nr < 8) {
@@ -69,9 +55,6 @@ static inline void unmask_irq(unsigned int irq_nr)
 {
 	unsigned char mask;
 
-        if (irq_nr >= 16)
-	    return;
-
 	mask = ~(1 << (irq_nr & 7));
 	if (irq_nr < 8) {
 		cache_21 &= mask;
@@ -139,13 +122,6 @@ int get_irq_list(char *buf)
 
 atomic_t __mips_bh_counter;
 
-#ifdef __SMP__
-#error Send superfluous SMP boxes to ralf@uni-koblenz.de
-#else
-#define irq_enter(cpu, irq)     (++local_irq_count[cpu])
-#define irq_exit(cpu, irq)      (--local_irq_count[cpu])
-#endif
-
 /*
  * do_IRQ handles IRQ's that have been installed without the
  * SA_INTERRUPT flag: it uses the full signal-handling return
@@ -255,7 +231,7 @@ int request_irq(unsigned int irq,
 	int retval;
 	struct irqaction * action;
 
-	if (irq > 31)
+	if (irq >= 32)
 		return -EINVAL;
 	if (!handler)
 		return -EINVAL;
diff --git a/arch/mips/kernel/mips_ksyms.c b/arch/mips/kernel/mips_ksyms.c
index d6657ac89..83712649c 100644
--- a/arch/mips/kernel/mips_ksyms.c
+++ b/arch/mips/kernel/mips_ksyms.c
@@ -7,7 +7,7 @@
  *
  * Copyright (C) 1996, 1997 by Ralf Baechle
  *
- * $Id: mips_ksyms.c,v 1.2 1997/07/26 19:31:40 ralf Exp $
+ * $Id: mips_ksyms.c,v 1.2 1997/07/29 03:58:54 ralf Exp $
  */
 #include <linux/config.h>
 #include <linux/module.h>
@@ -71,3 +71,19 @@ EXPORT_SYMBOL(port_base);
 #ifdef CONFIG_SGI
 EXPORT_SYMBOL(hpc3c0);
 #endif
+
+/*
+ * Kernel hacking ...
+ */
+#include <asm/branch.h>
+#include <linux/sched.h>
+
+int register_fpe(void (*handler)(struct pt_regs *regs, unsigned int fcr31));
+int unregister_fpe(void (*handler)(struct pt_regs *regs, unsigned int fcr31));
+
+#ifdef CONFIG_MIPS_FPE_MODULE
+EXPORT_SYMBOL(force_sig);
+EXPORT_SYMBOL(__compute_return_epc);
+EXPORT_SYMBOL(register_fpe);
+EXPORT_SYMBOL(unregister_fpe);
+#endif
diff --git a/arch/mips/kernel/pci.c b/arch/mips/kernel/pci.c
index 5e71233af..25ea56fdd 100644
--- a/arch/mips/kernel/pci.c
+++ b/arch/mips/kernel/pci.c
@@ -82,32 +82,6 @@ int pcibios_find_class (unsigned int class_code, unsigned short index,
 	return PCIBIOS_DEVICE_NOT_FOUND;
 }
 
-const char *pcibios_strerror (int error)
-{
-	static char buf[80];
-
-	switch (error) {
-	case PCIBIOS_SUCCESSFUL:
-		return "SUCCESSFUL";
-
-	case PCIBIOS_FUNC_NOT_SUPPORTED:
-		return "FUNC_NOT_SUPPORTED";
-
-	case PCIBIOS_BAD_VENDOR_ID:
-		return "SUCCESSFUL";
-
-	case PCIBIOS_DEVICE_NOT_FOUND:
-		return "DEVICE_NOT_FOUND";
-
-	case PCIBIOS_BAD_REGISTER_NUMBER:
-		return "BAD_REGISTER_NUMBER";
-
-	default:
-		sprintf (buf, "UNKNOWN RETURN 0x%x", error);
-		return buf;
-	}
-}
-
 /*
  * The functions below are machine specific and must be reimplented for
  * each PCI chipset configuration.  We just run the hook to the machine
diff --git a/arch/mips/kernel/ptrace.c b/arch/mips/kernel/ptrace.c
index 97a11abc4..b5a5a4428 100644
--- a/arch/mips/kernel/ptrace.c
+++ b/arch/mips/kernel/ptrace.c
@@ -59,12 +59,15 @@ repeat:
 		goto repeat;
 	}
 	page = pte_page(*pgtable);
-/* this is a hack for non-kernel-mapped video buffers and similar */
+	/* This is a hack for non-kernel-mapped video buffers and similar */
 	if (MAP_NR(page) >= MAP_NR(high_memory))
 		return 0;
 	page += addr & ~PAGE_MASK;
+	/* We can't use flush_page_to_ram() since we're running in
+	 * another context ...
+	 */
+	flush_cache_all();
 	retval = *(unsigned long *) page;
-	flush_page_to_ram(page);
 	return retval;
 }
 
@@ -117,14 +120,17 @@ repeat:
 		handle_mm_fault(tsk, vma, addr, 1);
 		goto repeat;
 	}
-/* this is a hack for non-kernel-mapped video buffers and similar */
-	flush_cache_page(vma, addr);
-	if (MAP_NR(page) < MAP_NR(high_memory)) {
-		*(unsigned long *) (page + (addr & ~PAGE_MASK)) = data;
+	/* This is a hack for non-kernel-mapped video buffers and similar */
+	if (MAP_NR(page) < MAP_NR(high_memory))
+		flush_cache_page(vma, addr);
+	*(unsigned long *) (page + (addr & ~PAGE_MASK)) = data;
+	if (MAP_NR(page) < MAP_NR(high_memory))
 		flush_page_to_ram(page);
-	}
-/* we're bypassing pagetables, so we have to set the dirty bit ourselves */
-/* this should also re-instate whatever read-only mode there was before */
+	/*
+	 * We're bypassing pagetables, so we have to set the dirty bit
+	 * ourselves this should also re-instate whatever read-only mode
+	 * there was before
+	 */
 	set_pte(pgtable, pte_mkdirty(mk_pte(page, vma->vm_page_prot)));
 	flush_tlb_page(vma, addr);
 }
@@ -369,7 +375,7 @@ asmlinkage int sys_ptrace(long request, long pid, long addr, long data)
 				case 5:
 					tmp = child->tss.fpu.hard.control;
 					break;
-				case 6:
+				case 6:	/* implementation / version register */
 					tmp = 0;
 					break;
 				default:
diff --git a/arch/mips/kernel/r4k_misc.S b/arch/mips/kernel/r4k_misc.S
index 0cbc2222f..090b3109f 100644
--- a/arch/mips/kernel/r4k_misc.S
+++ b/arch/mips/kernel/r4k_misc.S
@@ -1,10 +1,12 @@
-/* $Id: r4k_misc.S,v 1.8 1996/07/22 22:32:52 dm Exp $
+/*
  * r4k_misc.S: Misc. exception handling code for r4k.
  *
  * Copyright (C) 1994, 1995, 1996 by Ralf Baechle and Andreas Busse
  *
  * Multi-cpu abstraction and reworking:
  * Copyright (C) 1996 David S. Miller (dm@engr.sgi.com)
+ *
+ * $Id: r4k_misc.S,v 1.2 1997/06/09 06:20:52 ralf Exp $
  */
 #include <asm/asm.h>
 #include <asm/current.h>
@@ -21,7 +23,7 @@
 #include <asm/regdef.h>
 #include <asm/stackframe.h>
 
-#define NOTLB_OPTIMIZE /* If you are paranoid, define this. */
+#undef NOTLB_OPTIMIZE /* If you are paranoid, define this. */
 
 	/* ABUSE of CPP macros 101. */
 
@@ -108,10 +110,32 @@
 	.set	noreorder
 	.set	mips3
 
-	/* Note for many R4k variants tlb probes cannot be executed out
+/*
+ * From the IDT errata for the QED RM5230 (Nevada), processor revision 1.0:
+ * 2. A timing hazard exists for the TLBP instruction.
+ *
+ *      stalling_instruction
+ *      TLBP
+ *
+ * The JTLB is being read for the TLBP throughout the stall generated by the
+ * previous instruction. This is not really correct as the stalling instruction
+ * can modify the address used to access the JTLB.  The failure symptom is that
+ * the TLBP instruction will use an address created for the stalling instruction
+ * and not the address held in C0_ENHI and thus report the wrong results.
+ *
+ * The software work-around is to not allow the instruction preceding the TLBP
+ * to stall - make it an NOP or some other instruction guaranteed not to stall.
+ *
+ * Errata 2 will not be fixed.  This errata is also on the R5000.
+ *
+ * As if we MIPS hackers wouldn't know how to nop pipelines happy ...
+ */
+#define R5K_HAZARD nop
+
+	/*
+	 * Note for many R4k variants tlb probes cannot be executed out
 	 * of the instruction cache else you get bogus results.
 	 */
-
 	.align	5
 	NESTED(r4k_handle_tlbl, PT_SIZE, sp)
 	.set	noat
@@ -120,6 +144,7 @@ invalid_tlbl:
 #ifndef NOTLB_OPTIMIZE
 	/* Test present bit in entry. */
 	LOAD_PTE(k0, k1)
+	R5K_HAZARD
 	tlbp
 	PTE_PRESENT(k0, k1, nopage_tlbl)
 	PTE_MAKEVALID(k0, k1)
@@ -141,6 +166,7 @@ nopage_tlbl:
 	.set	noat
 #ifndef NOTLB_OPTIMIZE
 	LOAD_PTE(k0, k1)
+	R5K_HAZARD
 	tlbp				# find faulting entry
 	PTE_WRITABLE(k0, k1, nopage_tlbs)
 	PTE_MAKEWRITE(k0, k1)
@@ -162,6 +188,7 @@ nopage_tlbs:
 	.set	noat
 #ifndef NOTLB_OPTIMIZE
 	LOAD_PTE(k0, k1)
+	R5K_HAZARD
 	tlbp					# find faulting entry
 	andi	k0, k0, _PAGE_WRITE
 	beqz	k0, nowrite_mod
diff --git a/arch/mips/kernel/setup.c b/arch/mips/kernel/setup.c
index 1b00365a6..3ebf269f5 100644
--- a/arch/mips/kernel/setup.c
+++ b/arch/mips/kernel/setup.c
@@ -4,6 +4,8 @@
  *  Copyright (C) 1995  Linus Torvalds
  *  Copyright (C) 1995, 1996  Ralf Baechle
  *  Copyright (C) 1996  Stoned Elipot
+ *
+ * $Id:$
  */
 #include <linux/config.h>
 #include <linux/errno.h>
@@ -168,9 +170,10 @@ __initfunc(void setup_arch(char **cmdline_p,
 		break;
 #endif
 #if defined(CONFIG_MIPS_ARC) 
-/* Perhaps arch/mips/deskstation should be renommed arch/mips/arc.
- * For now CONFIG_MIPS_ARC means DeskStation. -Stoned.
- */
+	/*
+	 * Perhaps arch/mips/deskstation should be renamed to arch/mips/arc.
+	 * For now CONFIG_MIPS_ARC means DeskStation. -Stoned.
+	 */
 	case MACH_GROUP_ARC:
 		deskstation_setup();
 		break;
@@ -196,9 +199,6 @@ __initfunc(void setup_arch(char **cmdline_p,
 
 	atag = bi_TagFind(tag_drive_info);
 	memcpy(&drive_info, TAGVALPTR(atag), atag->size);
-#if 0
-	aux_device_present = AUX_DEVICE_INFO;
-#endif
 
 	memory_end = mips_memory_upper;
 	/*
diff --git a/arch/mips/kernel/signal.c b/arch/mips/kernel/signal.c
index 34ed87fb6..b2e76e8a7 100644
--- a/arch/mips/kernel/signal.c
+++ b/arch/mips/kernel/signal.c
@@ -4,7 +4,7 @@
  *  Copyright (C) 1991, 1992  Linus Torvalds
  *  Copyright (C) 1994, 1995, 1996  Ralf Baechle
  *
- * $Id: signal.c,v 1.3 1997/06/25 20:08:49 ralf Exp $
+ * $Id: signal.c,v 1.4 1997/08/06 19:15:07 miguel Exp $
  */
 #include <linux/config.h>
 #include <linux/sched.h>
@@ -42,7 +42,7 @@ asmlinkage int sys_sigsuspend(struct pt_regs *regs)
 	unsigned long mask;
 	sigset_t *uset, set;
 
-	uset = (sigset_t *)(long) regs->regs[4];
+	uset = (sigset_t *) regs->regs[4];
 	if (get_user(set, uset))
 		return -EFAULT;
 
@@ -392,3 +392,13 @@ asmlinkage unsigned long sys_signal(int signum, __sighandler_t handler)
 {
 	return -ENOSYS;
 }
+
+/*
+ * Compatibility syscall.  Can be replaced in libc.
+ */
+asmlinkage int sys_pause(void)
+{
+	current->state = TASK_INTERRUPTIBLE;
+	schedule();
+	return -ERESTARTNOHAND;
+}
diff --git a/arch/mips/kernel/syscall.c b/arch/mips/kernel/syscall.c
index f1b117a63..0a190e3ab 100644
--- a/arch/mips/kernel/syscall.c
+++ b/arch/mips/kernel/syscall.c
@@ -21,6 +21,7 @@
 #include <linux/smp_lock.h>
 #include <linux/mman.h>
 #include <linux/sched.h>
+#include <linux/utsname.h>
 #include <linux/unistd.h>
 #include <asm/branch.h>
 #include <asm/ptrace.h>
@@ -149,6 +150,43 @@ out:
 }
 
 /*
+ * Compacrapability ...
+ */
+asmlinkage int sys_uname(struct old_utsname * name)
+{
+	if (name && !copy_to_user(name, &system_utsname, sizeof (*name)))
+		return 0;
+	return -EFAULT;
+}
+
+/*
+ * Compacrapability ...
+ */
+asmlinkage int sys_olduname(struct oldold_utsname * name)
+{
+	int error;
+
+	if (!name)
+		return -EFAULT;
+	if (!access_ok(VERIFY_WRITE,name,sizeof(struct oldold_utsname)))
+		return -EFAULT;
+  
+	error = __copy_to_user(&name->sysname,&system_utsname.sysname,__OLD_UTS_LEN);
+	error -= __put_user(0,name->sysname+__OLD_UTS_LEN);
+	error -= __copy_to_user(&name->nodename,&system_utsname.nodename,__OLD_UTS_LEN);
+	error -= __put_user(0,name->nodename+__OLD_UTS_LEN);
+	error -= __copy_to_user(&name->release,&system_utsname.release,__OLD_UTS_LEN);
+	error -= __put_user(0,name->release+__OLD_UTS_LEN);
+	error -= __copy_to_user(&name->version,&system_utsname.version,__OLD_UTS_LEN);
+	error -= __put_user(0,name->version+__OLD_UTS_LEN);
+	error -= __copy_to_user(&name->machine,&system_utsname.machine,__OLD_UTS_LEN);
+	error = __put_user(0,name->machine+__OLD_UTS_LEN);
+	error = error ? -EFAULT : 0;
+
+	return error;
+}
+
+/*
  * Do the indirect syscall syscall.
  * Don't care about kernel locking; the actual syscall will do it.
  */
diff --git a/arch/mips/kernel/syscalls.h b/arch/mips/kernel/syscalls.h
index fd1b39bca..3232af01c 100644
--- a/arch/mips/kernel/syscalls.h
+++ b/arch/mips/kernel/syscalls.h
@@ -7,7 +7,7 @@
  *
  * Copyright (C) 1995, 1996, 1997 by Ralf Baechle
  *
- * $Id: syscalls.h,v 1.5 1997/07/20 14:57:27 ralf Exp $
+ * $Id: syscalls.h,v 1.6 1997/08/06 19:15:08 miguel Exp $
  */
 
 /*
@@ -35,7 +35,7 @@ SYS(sys_time, 1)
 SYS(sys_mknod, 3)
 SYS(sys_chmod, 2)				/* 4015 */
 SYS(sys_chown, 3)
-SYS(sys_break, 0)
+SYS(sys_ni_syscall, 0)
 SYS(sys_stat, 2)
 SYS(sys_lseek, 3)
 SYS(sys_getpid, 0)				/* 4020 */
@@ -49,11 +49,11 @@ SYS(sys_alarm, 1)
 SYS(sys_fstat, 2)
 SYS(sys_pause, 0)
 SYS(sys_utime, 2)				/* 4030 */
-SYS(sys_stty, 0)
-SYS(sys_gtty, 0)
+SYS(sys_ni_syscall, 0)
+SYS(sys_ni_syscall, 0)
 SYS(sys_access, 2)
 SYS(sys_nice, 1)
-SYS(sys_ftime, 0)				/* 4035 */
+SYS(sys_ni_syscall, 0)				/* 4035 */
 SYS(sys_sync, 0)
 SYS(sys_kill, 2)
 SYS(sys_rename, 2)
@@ -62,7 +62,7 @@ SYS(sys_rmdir, 1)				/* 4040 */
 SYS(sys_dup, 1)
 SYS(sys_pipe, 0)
 SYS(sys_times, 1)
-SYS(sys_prof, 0)
+SYS(sys_ni_syscall, 0)
 SYS(sys_brk, 1)					/* 4045 */
 SYS(sys_setgid, 1)
 SYS(sys_getgid, 0)
@@ -70,13 +70,13 @@ SYS(sys_signal, 2)
 SYS(sys_geteuid, 0)
 SYS(sys_getegid, 0)				/* 4050 */
 SYS(sys_acct, 0)
-SYS(sys_phys, 0)
-SYS(sys_lock, 0)
+SYS(sys_ni_syscall, 0)
+SYS(sys_ni_syscall, 0)
 SYS(sys_ioctl, 3)
 SYS(sys_fcntl, 3)				/* 4055 */
-SYS(sys_mpx, 2)
+SYS(sys_ni_syscall, 2)
 SYS(sys_setpgid, 2)
-SYS(sys_ulimit, 0)
+SYS(sys_ni_syscall, 0)
 SYS(sys_olduname, 1)
 SYS(sys_umask, 1)				/* 4060 */
 SYS(sys_chroot, 1)
@@ -116,7 +116,7 @@ SYS(sys_fchmod, 2)
 SYS(sys_fchown, 3)				/* 4095 */
 SYS(sys_getpriority, 2)
 SYS(sys_setpriority, 3)
-SYS(sys_profil, 0)
+SYS(sys_ni_syscall, 0)
 SYS(sys_statfs, 2)
 SYS(sys_fstatfs, 2)				/* 4100 */
 SYS(sys_ioperm, 3)
diff --git a/arch/mips/kernel/traps.c b/arch/mips/kernel/traps.c
index b7ceea080..c38fcf549 100644
--- a/arch/mips/kernel/traps.c
+++ b/arch/mips/kernel/traps.c
@@ -12,8 +12,6 @@
  * to mainly kill the offending process (probably by giving it a signal,
  * but possibly by killing it outright if necessary).
  *
- * FIXME: This is the place for a fpu emulator.
- *
  * Modified for R3000 by Paul M. Antoine, 1995, 1996
  */
 #include <linux/config.h>
@@ -228,8 +226,34 @@ out:
 	unlock_kernel();
 }
 
+#ifdef CONFIG_MIPS_FPE_MODULE
+static void (*fpe_handler)(struct pt_regs *regs, unsigned int fcr31);
+
+/*
+ * Register_fpe/unregister_fpe are for debugging purposes only.  To make
+ * this hack work a bit better there is no error checking.
+ */
+int register_fpe(void (*handler)(struct pt_regs *regs, unsigned int fcr31))
+{
+	fpe_handler = handler;
+	return 0;
+}
+
+int unregister_fpe(void (*handler)(struct pt_regs *regs, unsigned int fcr31))
+{
+	fpe_handler = NULL;
+	return 0;
+}
+#endif
+
 void do_fpe(struct pt_regs *regs, unsigned int fcr31)
 {
+#ifdef CONFIG_MIPS_FPE_MODULE
+	if (fpe_handler != NULL) {
+		fpe_handler(regs, fcr31);
+		return;
+	}
+#endif
 	lock_kernel();
 #ifdef CONF_DEBUG_EXCEPTIONS
 	show_regs(regs);
diff --git a/arch/mips/lib/copy_user.S b/arch/mips/lib/copy_user.S
index 30fa6d6e9..a7fdc74e5 100644
--- a/arch/mips/lib/copy_user.S
+++ b/arch/mips/lib/copy_user.S
@@ -1,13 +1,15 @@
 /*
- * arch/mips/mips1/memcpy.S
+ * arch/mips/lib/copy_user.S
  *
  * 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) 1996 by Ralf Baechle
+ * Copyright (c) 1996, 1997 by Ralf Baechle
  *
- * Less stupid memcpy/user_copy implementation for 32 bit MIPS CPUs.
+ * Less stupid user_copy implementation for 32 bit MIPS CPUs.
+ *
+ * $Id: copy_user.S,v 1.2 1997/08/11 04:26:12 ralf Exp $
  */
 #include <asm/asm.h>
 #include <asm/regdef.h>
@@ -37,8 +39,8 @@ not_even_the_same_alignment:
 		andi		v1,3
 		sltu		t0,v0,v1
 		MOVN(v1,v0,t0)
-		beqz		v1,3f			# -> finished
-		LONG_ADDU	v1,a0			# delay slot
+		beqz		v1,src_aligned
+		 LONG_ADDU	v1,a0
 1:		lb		$1,(a1)
 		EX(1b, fault)
 		LONG_ADDIU	a1,1
@@ -46,20 +48,22 @@ not_even_the_same_alignment:
 		EX(2b, fault)
 		LONG_ADDIU	a0,1
 		bne		a0,v1,1b
-		LONG_SUBU	v0,1			# delay slot
-3:
+		 LONG_SUBU	v0,1
+src_aligned:
+
 /*
  * Ok.  We've fixed the alignment of the copy src for this case.
  * Now let's copy in the usual BLOCK_SIZE byte blocks using unaligned
  * stores.
  * XXX Align the destination address.  This is better if the __copy_user
  *     encounters an access fault because we never have to deal with an
- *     only partially modified destination word.
+ *     only partially modified destination word.  This is required to
+ *     keep the semantics of the result of copy_*_user().
  */
 		ori		v1,v0,BLOCK_SIZE-1
 		xori		v1,BLOCK_SIZE-1
 		beqz		v1,copy_left_over
-		nop				# delay slot
+		 nop
 		LONG_SUBU	v0,v1
 		LONG_ADDU	v1,a0
 
@@ -81,16 +85,16 @@ not_even_the_same_alignment:
 		UEX(2b, fault_plus_12)
 		LONG_ADDIU	a0,BLOCK_SIZE
 		bne		a0,v1,1b
-		LONG_ADDIU	a1,BLOCK_SIZE	# delay slot
+		 LONG_ADDIU	a1,BLOCK_SIZE
 9:
 		b		copy_left_over	# < BLOCK_SIZE bytes left
-		nop				# delay slot
+		 nop
 
 /* ---------------------------------------------------------------------- */
 
 not_w_aligned:
 /*
- * Ok, src or destination are not 8-byte aligned.
+ * Ok, src or destination are not 4-byte aligned.
  * Try to fix that.  Do at least both addresses have the same alignment?
  */
 		xor		t0,a0,a1
@@ -107,8 +111,8 @@ not_w_aligned:
 		andi		v1,3
 		sltu		t0,v0,v1
 		MOVN(v1,v0,t0)
-		beqz		v1,align4		# -> finished
-		LONG_ADDU	v1,a0			# delay slot
+		beqz		v1,__copy_user
+		 LONG_ADDU	v1,a0
 1:		lb		$1,(a1)
 		EX(1b, fault)
 		LONG_ADDIU	a1,1
@@ -116,23 +120,23 @@ not_w_aligned:
 		EX(2b, fault)
 		LONG_ADDIU	a0,1
 		bne		a0,v1,1b
-		LONG_SUBU	v0,1			# delay slot
+		 LONG_SUBU	v0,1
 		b		align4
-		nop					# delay slot
+		 nop
 
 /* ---------------------------------------------------------------------- */
 
 LEAF(__copy_user)
 		or		t1,a0,a1
 		andi		t1,3
-		bnez		t1,not_w_aligned
-		move		v0,a2			# delay slot
+		bnez		t1,not_w_aligned	# not word alignment
+		 move		v0,a2
 
 align4:
 		ori		v1,v0,BLOCK_SIZE-1
 		xori		v1,BLOCK_SIZE-1
 		beqz		v1,copy_left_over
-		nop				# delay slot
+		 nop
 		LONG_SUBU	v0,v1
 		LONG_ADDU	v1,a0
 
@@ -154,7 +158,7 @@ align4:
 		EX(2b, fault_plus_12)
 		LONG_ADDIU	a0,BLOCK_SIZE
 		bne		a0,v1,1b
-		LONG_ADDIU	a1,BLOCK_SIZE	# delay slot
+		 LONG_ADDIU	a1,BLOCK_SIZE
 9:
 
 /*
@@ -162,7 +166,7 @@ align4:
  */
 copy_left_over:
 		beqz		v0,3f
-		nop					# delay slot
+		 nop
 1:		lb		$1,(a1)
 		EX(1b, fault)
 		LONG_ADDIU	a1,1
@@ -170,9 +174,11 @@ copy_left_over:
 		EX(2b, fault)
 		LONG_SUBU	v0,1
 		bnez		v0,1b
-		LONG_ADDIU	a0,1
-3:		jr		ra
-		nop				# delay slot
+		 LONG_ADDIU	a0,1
+3:
+
+done:		jr		ra
+		 nop
 
 		END(__copy_user)
 		.set		at
@@ -193,14 +199,3 @@ fault_plus_8:		LONG_ADDIU	v0,8
 			jr	ra
 fault_plus_12:		LONG_ADDIU	v0,12
 			jr	ra
-
-/* ---------------------------------------------------------------------- */
-
-/*
- * For now we use __copy_user for __memcpy, too.  This is effizient (one
- * instruction penatly) and smaller but adds unwanted error checking we don't
- * need.  This hopefully doesn't cover any bugs.  The memcpy() wrapper in
- * <asm/string.h> takes care of the return value in a way GCC can optimize.
- */
-		.globl	__memcpy
-__memcpy	=	__copy_user
diff --git a/arch/mips/mm/loadmmu.c b/arch/mips/mm/loadmmu.c
index 70597201a..c4b4258b7 100644
--- a/arch/mips/mm/loadmmu.c
+++ b/arch/mips/mm/loadmmu.c
@@ -1,7 +1,9 @@
-/* $Id: loadmmu.c,v 1.1.1.1 1997/06/01 03:16:38 ralf Exp $
+/*
  * loadmmu.c: Setup cpu/cache specific function ptrs at boot time.
  *
  * Copyright (C) 1996 David S. Miller (dm@engr.sgi.com)
+ *
+ * $Id: loadmmu.c,v 1.1.1.1 1997/06/01 03:16:38 ralf Exp $
  */
 
 #include <linux/kernel.h>
@@ -76,6 +78,7 @@ void loadmmu(void)
 	case CPU_R4700:
 	case CPU_R5000:
 	case CPU_R5000A:
+	case CPU_NEVADA:
 		printk("Loading R4000 MMU routines.\n");
 		ld_mmu_r4xx0();
 		break;
diff --git a/arch/mips/mm/r4xx0.c b/arch/mips/mm/r4xx0.c
index 2b32ca455..2586a961c 100644
--- a/arch/mips/mm/r4xx0.c
+++ b/arch/mips/mm/r4xx0.c
@@ -3,7 +3,7 @@
  *
  * Copyright (C) 1996 David S. Miller (dm@engr.sgi.com)
  *
- * $Id: r4xx0.c,v 1.5 1997/07/29 22:54:52 tsbogend Exp $
+ * $Id: r4xx0.c,v 1.6 1997/08/06 19:15:10 miguel Exp $
  */
 #include <linux/config.h>
 
@@ -39,7 +39,16 @@ static scache_size, sc_lsize;        /* Again, in bytes */
 #undef DEBUG_CACHE
 
 /*
- * Zero an entire page.
+ * On processors with QED R4600 style two set assosicative cache
+ * this is the bit which selects the way in the cache for the
+ * indexed cachops.
+ */
+#define waybit 0x2000
+
+/*
+ * Zero an entire page.  We have three flavours of the routine available.
+ * One for CPU with 16byte, with 32byte cachelines plus a special version
+ * with nops which handles the buggy R4600 v1.x.
  */
 
 static void r4k_clear_page_d16(unsigned long page)
@@ -1465,11 +1474,8 @@ static void r4k_flush_cache_page_d16i16(struct vm_area_struct *vma,
 		 */
 		page = (KSEG0 + (page & (dcache_size - 1)));
 		blast_dcache16_page_indexed(page);
-		blast_dcache16_page_indexed(page ^ 0x2000);
-		if(text) {
+		if(text)
 			blast_icache16_page_indexed(page);
-			blast_icache16_page_indexed(page ^ 0x2000);
-		}
 	}
 out:
 	restore_flags(flags);
@@ -1583,10 +1589,10 @@ static void r4k_flush_cache_page_d32i32_r4600(struct vm_area_struct *vma,
 		 */
 		page = (KSEG0 + (page & (dcache_size - 1)));
 		blast_dcache32_page_indexed(page);
-		blast_dcache32_page_indexed(page ^ 0x2000);
+		blast_dcache32_page_indexed(page ^ waybit);
 		if(text) {
 			blast_icache32_page_indexed(page);
-			blast_icache32_page_indexed(page ^ 0x2000);
+			blast_icache32_page_indexed(page ^ waybit);
 		}
 	}
 out:
@@ -1776,7 +1782,6 @@ static void r4k_flush_page_to_ram_d32i32_r4600(unsigned long page)
 		unsigned long flags;
 
 #ifdef DEBUG_CACHE
-		/* #if 1 */
 		printk("r4600_cram[%08lx]", page);
 #endif
 		/*
@@ -1786,7 +1791,6 @@ static void r4k_flush_page_to_ram_d32i32_r4600(unsigned long page)
 
 		save_and_cli(flags);
 		blast_dcache32_page(page);
-		blast_dcache32_page(page ^ 0x2000);
 #ifdef CONFIG_SGI
 		{
 			unsigned long tmp1, tmp2;
@@ -1824,6 +1828,7 @@ static void r4k_flush_page_to_ram_d32i32_r4600(unsigned long page)
 
 static void r4k_flush_cache_sigtramp(unsigned long addr)
 {
+	/* XXX  protect like uaccess.h loads/stores */
 	addr &= ~(dc_lsize - 1);
 	__asm__ __volatile__("nop;nop;nop;nop");
 	flush_dcache_line(addr);
@@ -2534,7 +2539,7 @@ void ld_mmu_r4xx0(void)
 	unsigned long cfg = read_32bit_cp0_register(CP0_CONFIG);
 	int sc_present = 0;
 
-	printk("CPU REVISION IS: %08x\n", read_32bit_cp0_register(CP0_PRID));
+	printk("CPU revision is: %08x\n", read_32bit_cp0_register(CP0_PRID));
 
 	probe_icache(cfg);
 	probe_dcache(cfg);
@@ -2546,7 +2551,6 @@ void ld_mmu_r4xx0(void)
 	case CPU_R4400PC:
 	case CPU_R4400SC:
 	case CPU_R4400MC:
-try_again:
 		probe_scache_kseg1 = (probe_func_t) (KSEG1ADDR(&probe_scache));
 		sc_present = probe_scache_kseg1(cfg);
 		break;
@@ -2554,38 +2558,40 @@ try_again:
 	case CPU_R4600:
 	case CPU_R4640:
 	case CPU_R4700:
-	case CPU_R5000:
+	case CPU_R5000:	/* XXX: We don't handle the true R5000 SCACHE */
+	case CPU_NEVADA:
 		probe_scache_kseg1 = (probe_func_t)
 			(KSEG1ADDR(&probe_scache_eeprom));
 		sc_present = probe_scache_eeprom(cfg);
 
-		/* Try using tags if eeprom give us bogus data. */
-		if(sc_present == -1)
-			goto try_again;
+		/* Try using tags if eeprom gives us bogus data. */
+		if(sc_present == -1) {
+			probe_scache_kseg1 =
+				(probe_func_t) (KSEG1ADDR(&probe_scache));
+			sc_present = probe_scache_kseg1(cfg);
+		}
 		break;
 	};
 
-	if(!sc_present) {
-		/* Lacks secondary cache. */
-		setup_noscache_funcs();
+	if(sc_present == 1
+	   && (mips_cputype == CPU_R4000SC
+               || mips_cputype == CPU_R4000MC
+               || mips_cputype == CPU_R4400SC
+               || mips_cputype == CPU_R4400MC)) {
+		/* Has a true secondary cache. */
+		setup_scache_funcs();
 	} else {
-		/* Has a secondary cache. */
-		if(mips_cputype != CPU_R4600 &&
-		   mips_cputype != CPU_R4640 &&
-		   mips_cputype != CPU_R4700 &&
-		   mips_cputype != CPU_R5000) {
-			setup_scache_funcs();
-		} else {
-			setup_noscache_funcs();
-			if((mips_cputype != CPU_R5000)) {
-				flush_cache_page =
-					r4k_flush_cache_page_d32i32_r4600;
-				flush_page_to_ram =
-					r4k_flush_page_to_ram_d32i32_r4600;
-			}
+		/* Lacks true secondary cache. */
+		setup_noscache_funcs();
+		if((mips_cputype != CPU_R5000)) { /* XXX */
+			flush_cache_page =
+				r4k_flush_cache_page_d32i32_r4600;
+			flush_page_to_ram =
+				r4k_flush_page_to_ram_d32i32_r4600;
 		}
 	}
 
+	/* XXX Handle true second level cache w/ split I/D */
 	flush_cache_sigtramp = r4k_flush_cache_sigtramp;
 
 	flush_tlb_all = r4k_flush_tlb_all;
diff --git a/arch/mips/sgi/kernel/indy_int.c b/arch/mips/sgi/kernel/indy_int.c
index 2d23372ae..343a71f9e 100644
--- a/arch/mips/sgi/kernel/indy_int.c
+++ b/arch/mips/sgi/kernel/indy_int.c
@@ -4,7 +4,7 @@
  *
  * Copyright (C) 1996 David S. Miller (dm@engr.sgi.com)
  *
- * $Id: indy_int.c,v 1.2 1997/07/01 09:00:58 ralf Exp $
+ * $Id: indy_int.c,v 1.3 1997/08/26 04:34:55 miguel Exp $
  */
 #include <linux/config.h>
 
@@ -260,13 +260,6 @@ int get_irq_list(char *buf)
 
 atomic_t __mips_bh_counter;
 
-#ifdef __SMP__
-#error Send superfluous SMP boxes to ralf@uni-koblenz.de
-#else
-#define irq_enter(cpu, irq)     (++local_irq_count[cpu])
-#define irq_exit(cpu, irq)      (--local_irq_count[cpu])
-#endif
-
 /*
  * do_IRQ handles IRQ's that have been installed without the
  * SA_INTERRUPT flag: it uses the full signal-handling return
@@ -434,7 +427,7 @@ void indy_local0_irqdispatch(struct pt_regs *regs)
 	struct irqaction *action;
 	unsigned char mask = ioc_icontrol->istat0;
 	unsigned char mask2 = 0;
-	int irq;
+	int irq, cpu = smp_processor_id();;
 
 	mask &= ioc_icontrol->imask0;
 	if(mask & ISTAT0_LIO2) {
@@ -446,13 +439,11 @@ void indy_local0_irqdispatch(struct pt_regs *regs)
 		irq = lc0msk_to_irqnr[mask];
 		action = local_irq_action[irq];
 	}
-#if 0
-	printk("local0_dispatch: got irq %d mask %2x mask2 %2x\n",
-	       irq, mask, mask2);
-	prom_getchar();
-#endif
+
+	irq_enter(cpu, irq);
 	kstat.interrupts[irq + 16]++;
 	action->handler(irq, action->dev_id, regs);
+	irq_exit(cpu, irq);
 }
 
 void indy_local1_irqdispatch(struct pt_regs *regs)
@@ -460,7 +451,7 @@ void indy_local1_irqdispatch(struct pt_regs *regs)
 	struct irqaction *action;
 	unsigned char mask = ioc_icontrol->istat1;
 	unsigned char mask2 = 0;
-	int irq;
+	int irq, cpu = smp_processor_id();;
 
 	mask &= ioc_icontrol->imask1;
 	if(mask & ISTAT1_LIO3) {
@@ -473,23 +464,24 @@ void indy_local1_irqdispatch(struct pt_regs *regs)
 		irq = lc1msk_to_irqnr[mask];
 		action = local_irq_action[irq];
 	}
-#if 0
-	printk("local1_dispatch: got irq %d mask %2x mask2 %2x\n",
-	       irq, mask, mask2);
-	prom_getchar();
-#endif
+	irq_enter(cpu, irq);
 	kstat.interrupts[irq + 24]++;
 	action->handler(irq, action->dev_id, regs);
+	irq_exit(cpu, irq);
 }
 
 void indy_buserror_irq(struct pt_regs *regs)
 {
-	kstat.interrupts[6]++;
+	int cpu = smp_processor_id();
+	int irq = 6;
+
+	irq_enter(cpu, irq);
+	kstat.interrupts[irq]++;
 	printk("Got a bus error IRQ, shouldn't happen yet\n");
 	show_regs(regs);
 	printk("Spinning...\n");
-	while(1)
-		;
+	while(1);
+	irq_exit(cpu, irq);
 }
 
 /* Misc. crap just to keep the kernel linking... */
diff --git a/arch/mips/sgi/kernel/indy_timer.c b/arch/mips/sgi/kernel/indy_timer.c
index 43c1c76c6..d0ba84929 100644
--- a/arch/mips/sgi/kernel/indy_timer.c
+++ b/arch/mips/sgi/kernel/indy_timer.c
@@ -3,7 +3,7 @@
  *
  * Copyright (C) 1996 David S. Miller (dm@engr.sgi.com)
  *
- * $Id: indy_timer.c,v 1.2 1997/06/28 23:27:29 ralf Exp $
+ * $Id: indy_timer.c,v 1.2 1997/07/01 09:00:59 ralf Exp $
  */
 
 #include <linux/errno.h>
@@ -101,10 +101,12 @@ static long last_rtc_update = 0;
 
 void indy_timer_interrupt(struct pt_regs *regs)
 {
+	int irq = 7;
+
 	/* Ack timer and compute new compare. */
 	r4k_cur = (read_32bit_cp0_register(CP0_COUNT) + r4k_offset);
 	ack_r4ktimer(r4k_cur);
-	kstat.interrupts[7]++;
+	kstat.interrupts[irq]++;
 	do_timer(regs);
 
 	/* We update the Dallas time of day approx. every 11 minutes,
@@ -272,10 +274,15 @@ void indy_timer_init(void)
 
 void indy_8254timer_irq(void)
 {
-	kstat.interrupts[4]++;
+	int cpu = smp_processor_id();
+	int irq = 4;
+
+	irq_enter(cpu, irq);
+	kstat.interrupts[irq]++;
 	printk("indy_8254timer_irq: Whoops, should not have gotten this IRQ\n");
 	prom_getchar();
 	prom_imode();
+	irq_exit(cpu, irq);
 }
 
 void do_gettimeofday(struct timeval *tv)
diff --git a/arch/ppc/Makefile b/arch/ppc/Makefile
index ac2719977..6c81701d9 100644
--- a/arch/ppc/Makefile
+++ b/arch/ppc/Makefile
@@ -1,6 +1,3 @@
-#
-# ppc/Makefile
-#
 # This file is included by the global makefile so that you can add your own
 # architecture-specific flags and dependencies. Remember to do have actions
 # for "archclean" and "archdep" for cleaning up and making dependencies for
@@ -12,29 +9,28 @@
 #
 # Copyright (C) 1994 by Linus Torvalds
 # Changes for PPC by Gary Thomas
-# Modified by Cort Dougan
+# Rewritten by Cort Dougan and Paul Mackerras
 #
 
+ifeq ($(CONFIG_PMAC),y)
+KERNELBASE	=0xc0000000
+else
+KERNELBASE	=0x90000000
+endif
+
 # PowerPC (cross) tools
-SUFFIX		= 
-AS		= as$(SUFFIX)
-ASFLAGS		= 
-LD		= ld$(SUFFIX)
-LINKFLAGS	= -T arch/ppc/ld.script -Ttext 0x90000000 
-HOSTCC		= gcc
-CC		= gcc$(SUFFIX)
+ifneq ($(shell uname -m),ppc)
+CROSS_COMPILE 	= ppc-linux-elf-
+else
+CHECKS		= checks
+endif
+
+ASFLAGS		=
+LINKFLAGS	= -T arch/ppc/vmlinux.lds -Ttext $(KERNELBASE) -Bstatic
 CFLAGSINC	= -D__KERNEL__ -I$(TOPDIR)/include -D__powerpc__
-CFLAGS		= $(CFLAGSINC) \
-		-Wstrict-prototypes -fomit-frame-pointer \
-		-fno-builtin \
-		-finhibit-size-directive \
-		-O2 -fsigned-char -pipe -ffixed-r2 -mstring -mmultiple -msoft-float
-# -fverbose-asm 
+CFLAGS		:= $(CFLAGS) -D__powerpc__ -fsigned-char -msoft-float -pipe \
+		-fno-builtin -ffixed-r2 -Wno-uninitialized -mmultiple -mstring
 CPP		= $(CC) -E $(CFLAGS)
-AR		= ar$(SUFFIX)
-RANLIB		= ranlib$(SUFFIX)
-STRIP		= strip$(SUFFIX)
-NM		= nm$(SUFFIX)
 
 ifdef CONFIG_601
 CFLAGS := $(CFLAGS) -mcpu=601 -DCPU=601
@@ -55,59 +51,35 @@ SUBDIRS := $(SUBDIRS) $(ARCH_SUBDIRS)
 ARCHIVES := arch/ppc/kernel/kernel.o arch/ppc/mm/mm.o arch/ppc/lib/lib.o $(ARCHIVES)
 CORE_FILES := arch/ppc/kernel/kernel.o arch/ppc/mm/mm.o arch/ppc/lib/lib.o $(CORE_FILES)
 
+ifdef CONFIG_XMON
+SUBDIRS += arch/ppc/xmon
+CORE_FILES += arch/ppc/xmon/x.o
+endif
+
+ifdef CONFIG_PMAC
+MAKEBOOT = $(MAKE) -C arch/$(ARCH)/coffboot
+else
+# PReP and CHRP systems
 MAKEBOOT = $(MAKE) -C arch/$(ARCH)/boot
+endif
 
 checks:
 	@$(MAKE) -C arch/$(ARCH)/kernel checks
 
-netboot: checks vmlinux
-	@$(MAKEBOOT) netboot
-
-znetboot: checks vmlinux
-	@$(MAKEBOOT) znetboot
+BOOT_TARGETS = netboot znetboot zImage floppy install \
+	vmlinux.coff znetboot.initrd zImage.initrd
 
-#rcpboot: checks vmlinux
-#	@$(MAKEBOOT) rcpboot
+$(BOOT_TARGETS): $(CHECKS) vmlinux
+	@$(MAKEBOOT) $@
 
-zImage: checks vmlinux
-	@$(MAKEBOOT) zImage
-
-floppy: checks vmlinux
-	@$(MAKEBOOT) floppy
-
-install: checks vmlinux
-	@$(MAKEBOOT) install
-
-vmlinux.coff : checks vmlinux
-	$(MAKE) -C arch/ppc/coffboot/ vmlinux.coff
-
-arch/ppc/kernel: dummy
-	$(MAKE) linuxsubdirs SUBDIRS=arch/ppc/kernel
-
-arch/ppc/mm: dummy
-	$(MAKE) linuxsubdirs SUBDIRS=arch/ppc/mm
-
-arch/ppc/lib: dummy
-	$(MAKE) linuxsubdirs SUBDIRS=arch/ppc/lib
-
-diffs:
-	arch/ppc/mkdiff
-
-tar:
-	arch/ppc/mktar
+tags:
+	etags */*.c include/{asm,linux}/*.h arch/ppc/kernel/*.{c,h}
 
 archclean:
-	rm -f arch/ppc/kernel/mk_defs arch/ppc/kernel/ppc_defs.h arch/ppc/kernel/checks TAGS
-	rm -f `find arch/ppc/ \( -name '*.[oas]' -o -name '*~' -o -name '#*#' \) -print`
-	rm -f `find include/asm-ppc/ \( -name '*.[oas]' -o -name '*~' -o -name '#*#' \) -print`
+	rm -f arch/ppc/kernel/mk_defs arch/ppc/kernel/ppc_defs.h
+	rm -f arch/ppc/kernel/checks
 	@$(MAKEBOOT) clean
 
 archdep:
-	$(MAKE) -C arch/ppc/boot fastdep
-	$(MAKE) -C arch/ppc/kernel fastdep
-	$(MAKE) -C arch/ppc/mm fastdep
-	$(MAKE) -C arch/ppc/lib fastdep
-
-tags :
-	etags arch/ppc/*/*.c arch/ppc/*/*.S include/asm/* */*.c 
+	$(MAKEBOOT) fastdep
 
diff --git a/arch/ppc/boot/Makefile b/arch/ppc/boot/Makefile
index 5dcac750e..a731ec2a0 100644
--- a/arch/ppc/boot/Makefile
+++ b/arch/ppc/boot/Makefile
@@ -22,58 +22,81 @@
 	$(CC) -D__ASSEMBLY__ -traditional -c -o $*.o $<
 
 
-ZLINKFLAGS = -T ../ld.script -Ttext 0x00800000
+ZLINKFLAGS = -T ../vmlinux.lds -Ttext 0x00800000
 GZIP_FLAGS = -v9
 
 SYSTEM = $(TOPDIR)/vmlinux
-
-OBJECTS = head.o inflate.o unzip.o misc.o vreset.o
-
+OBJECTS := head.o inflate.o unzip.o misc.o vreset.o kbd.o
 CFLAGS = -O2 -DSTDC_HEADERS -I$(TOPDIR)/include
+OBJCOPY = $(CROSS_COMPILE)objcopy
+OBJCOPY_ARGS = -O elf32-powerpc
+
 
 all:	$(TOPDIR)/zImage
 
 mkprep : mkprep.c
 	$(HOSTCC) $(CFLAGSINC) -o mkprep mkprep.c
 
+piggyback : piggyback.c
+	$(HOSTCC) $(CFLAGSINC) -o piggyback piggyback.c
+
 find_name : find_name.c
 	$(HOSTCC) $(CFLAGSINC) -o find_name find_name.c
 
-mk_type41: mk_type41.c
-	$(HOSTCC) $(CFLAGSINC) -o mk_type41 mk_type41.c
-
-piggyback: piggyback.c
-	$(HOSTCC) $(CFLAGS) -o piggyback piggyback.c
-
 floppy: $(TOPDIR)/vmlinux zImage 
 	dd if=$(TOPDIR)/zImage of=/dev/fd0H1440 bs=64b
 
-netboot : $(TOPDIR)/vmlinux mkprep
-	mkprep $(TOPDIR)/vmlinux $(TOPDIR)/netboot
+floppy.initrd: $(TOPDIR)/vmlinux zImage 
+	dd if=$(TOPDIR)/zImage.initrd of=/dev/fd0H1440 bs=64b
+
+znetboot : zImage mkprep
+	cp $(TOPDIR)/zImage /usr/local/tftpboot/vmlinux
 
-znetboot : zvmlinux mkprep
-	mkprep zvmlinux $(TOPDIR)/znetboot
-	cp $(TOPDIR)/znetboot /usr/local/tftpboot/vmlinux
+znetboot.initrd : zImage.initrd mkprep
+	cp $(TOPDIR)/zImage.initrd /usr/local/tftpboot/vmlinux
 
-rcpboot : znetboot
-	rcp $(TOPDIR)/znetboot charon:/usr/tftpboot/vmlinux
+#
+# This really needs to go away.  Perhaps a 
+# zImage.prep and zImage.chrp might be better.
+# Once we're able to get a lilo-ish program 
+# on prep systems this won't be a problem.
+#   -- Cort
+#
+ifdef CONFIG_CHRP
+zImage: zvmlinux
+	cp zvmlinux $(TOPDIR)/zImage
 
+zImage.initrd: zvmlinux.initrd
+	cp zvmlinux.initrd $(TOPDIR)/zImage.initrd
+
+zvmlinux: $(OBJECTS) $(SYSTEM) find_name vmlinux.gz piggyback
+	./piggyback < vmlinux.gz | $(AS) -o piggy.o
+	$(LD) $(ZLINKFLAGS) -o $@ $(OBJECTS) piggy.o
+	rm -f piggy.o
+else
 zImage: zvmlinux mkprep
 	mkprep -pbp zvmlinux $(TOPDIR)/zImage
 
-install: zImage
-	dd if=$(TOPDIR)/zImage of=/dev/sda4
-	ln -s /dev/sda4 $(INSTALL_PATH)/vmlinuz
-	cp $(TOPDIR)/System.map $(INSTALL_PATH)/	
+zImage.initrd: zvmlinux.initrd mkprep
+	mkprep -pbp zvmlinux.initrd $(TOPDIR)/zImage.initrd
 
-zvmlinux: $(OBJECTS) $(SYSTEM) mkprep  find_name
-	mkprep $(TOPDIR)/vmlinux -|gzip ${GZIP_FLAGS}|mkprep -asm - -|$(AS) -o piggy.o
-	$(LD) $(ZLINKFLAGS) -o zvmlinux $(OBJECTS) piggy.o
-	rm -f piggy.o
+zvmlinux: $(OBJECTS) $(SYSTEM) mkprep  find_name vmlinux.gz
+	$(LD) $(ZLINKFLAGS) -o zvmlinux.tmp $(OBJECTS)
+	$(OBJCOPY) $(OBJCOPY_ARGS) --add-section=image=vmlinux.gz \
+		zvmlinux.tmp $@
+	rm zvmlinux.tmp
+endif
+
+vmlinux.gz: $(TOPDIR)/vmlinux
+	dd bs=64k skip=1 if=$(TOPDIR)/vmlinux | gzip -vf9 - > vmlinux.gz
+
+zvmlinux.initrd: zvmlinux
+	$(OBJCOPY) $(OBJCOPY_ARGS) --add-section=initrd=ramdisk.image.gz \
+		zvmlinux $@
 
 clean:
-	rm -f piggyback zvmlinux mk_type41 mkprep mkboot find_name
-	rm -f $(TOPDIR)/{zImage,znetboot,netboot}
+	rm -f vmlinux* znetboot* zImage* zvmlinux* mkprep find_name
+	rm -f $(TOPDIR)/{zImage*,znetboot*,zvmlinux*,vmlinux*}
 
 fastdep:
 	$(TOPDIR)/scripts/mkdep *.[Sch] > .depend
diff --git a/arch/ppc/boot/find_name.c b/arch/ppc/boot/find_name.c
new file mode 100644
index 000000000..23646fb69
--- /dev/null
+++ b/arch/ppc/boot/find_name.c
@@ -0,0 +1,47 @@
+#include <stdio.h>
+#include <asm/page.h>
+#include <sys/mman.h>
+/*
+ * Finds a given address in the System.map and prints it out
+ * with its neighbors.  -- Cort
+ */
+
+void main(int argc, char **argv)
+{
+	unsigned long addr, cmp, i;
+	FILE *f;
+	char *ptr;
+	char s[256], last[256];
+	
+	if ( argc < 2 )
+	{
+		fprintf(stderr, "Usage: %s <address>\n", argv[0]);
+		exit(-1);
+	}
+
+	for ( i = 1 ; argv[i] ; i++ )
+	{
+		sscanf( argv[i], "%0x", &addr );
+		/* adjust if addr is relative to kernelbase */
+		if ( addr < PAGE_OFFSET )
+			addr += PAGE_OFFSET;
+		
+		if ( (f = fopen( "System.map", "r" )) == NULL )
+		{
+			perror("fopen()\n");
+			exit(-1);
+		}
+		
+		while ( !feof(f) )
+		{
+			fgets(s, 255 , f);
+			sscanf( s, "%0x", &cmp );
+			if ( addr < cmp )
+				break;
+			strcpy( last, s);
+		}
+		
+		printf( "%s", last);
+	}		
+	fclose(f);
+}
diff --git a/arch/ppc/boot/head.S b/arch/ppc/boot/head.S
index 6b25cb157..2cfda5cbe 100644
--- a/arch/ppc/boot/head.S
+++ b/arch/ppc/boot/head.S
@@ -13,17 +13,6 @@
 start:
 	bl	start_
 start_:
-/* TEMP - No residual data on BeBox (yet) */
-#if 0	
-#define IS_BE_BOX	0x42654278	/* 'BeBx' */
-	lis	r2,IS_BE_BOX>>16
-	ori	r2,r2,IS_BE_BOX&0xFFFF
-	cmp	0,r30,r2
-	bne	notBeBox
-	li	r3,0
-#endif	
-notBeBox:
-/* TEMP */
 	mr	r11,r3		/* Save pointer to residual data */
 	mfmsr	r3		/* Turn off interrupts */
 	li	r4,0
@@ -48,7 +37,9 @@ notBeBox:
 	mtlr	r21
 	mtctr	r22
 	bctr			/* Jump to code */
-/* Relocate code to final resting spot */
+/* 
+ * no matter where we're loaded, move ourselves to -Ttext address
+ */
 relocate:
 	mflr	r3		/* Compute code bias */
 	subi	r3,r3,4
@@ -98,13 +89,26 @@ start_ldr:
 	mr	r5,r6			/* Checksum */
 	mr	r6,r11			/* Residual data */
 	bl	decompress_kernel
-	/*mr	r29,r3*/			/* R3 = TotalMemory */
-	/*lis	r28,hold_residual@h
-	ori	r28,r28,hold_residual@l*/
+	
 	/* changed to use r3 (as firmware does) for kernel
 	   as ptr to residual -- Cort*/
-	li	r5,0x100		/* Kernel code starts here */
-	mtlr	r5
+	lis	r6,cmd_line@h
+	ori	r6,r6,cmd_line@l
+	subi	r7,r6,1
+00:	lbzu	r2,1(r7)
+	cmpi	0,r2,0
+	bne	00b
+
+	/* r4,r5 have initrd_start, size */
+	lis	r2,initrd_start@h
+	ori	r2,r2,initrd_start@l
+	lwz	r4,0(r2)
+	lis	r2,initrd_end@h
+	ori	r2,r2,initrd_end@l
+	lwz	r5,0(r2)
+		
+	li	r9,0x00c		/* Kernel code starts here */
+	mtlr	r9
 	blr	
 hang:
 	b	hang	
@@ -142,6 +146,45 @@ _put_HID0:
 	blr
 
 /*
+ * Delay for a number of microseconds
+ * -- Use the BUS timer (assumes 66MHz)
+ */
+	.globl	udelay
+udelay:		
+	mfspr	r4,PVR
+	srwi	r4,r4,16
+	cmpi	0,r4,1		/* 601 ? */
+	bne	.udelay_not_601
+00:	li	r0,86	/* Instructions / microsecond? */
+	mtctr	r0
+10:	addi	r0,r0,0 /* NOP */
+	bdnz	10b
+	subic.	r3,r3,1
+	bne	00b
+	blr
+
+.udelay_not_601:		
+	mulli	r4,r3,1000	/* nanoseconds */
+	addi	r4,r4,59
+	li	r5,60
+	divw	r4,r4,r5	/* BUS ticks */
+1:	mftbu	r5
+	mftb	r6
+	mftbu	r7
+	cmp	0,r5,r7
+	bne	1b		/* Get [synced] base time */
+	addc	r9,r6,r4	/* Compute end time */
+	addze	r8,r5
+2:	mftbu	r5
+	cmp	0,r5,r8
+	blt	2b
+	bgt	3f
+	mftb	r6
+	cmp	0,r6,r9
+	blt	2b
+3:	blr		
+	
+/*
  * This space [buffer] is used to forceably flush the data cache when
  * running in copyback mode.  This is necessary IFF the data cache could
  * contain instructions for which the instruction cache has stale data.
diff --git a/arch/ppc/boot/kbd.c b/arch/ppc/boot/kbd.c
new file mode 100644
index 000000000..6c0065b00
--- /dev/null
+++ b/arch/ppc/boot/kbd.c
@@ -0,0 +1,166 @@
+/* Keyboard handler */
+
+#include <../drivers/char/defkeymap.c> /* yeah I know it's bad */
+
+#define	L		0x0001	/* locking function */
+#define	SHF		0x0002	/* keyboard shift */
+#define	ALT		0x0004	/* alternate shift -- alternate chars */
+#define	NUM		0x0008	/* numeric shift  cursors vs. numeric */
+#define	CTL		0x0010	/* control shift  -- allows ctl function */
+#define	CPS		0x0020	/* caps shift -- swaps case of letter */
+#define	ASCII		0x0040	/* ascii code for this key */
+#define	STP		0x0080	/* stop output */
+#define	FUNC		0x0100	/* function key */
+#define	SCROLL		0x0200	/* scroll lock key */
+
+unsigned char shfts, ctls, alts, caps, num, stp;
+
+#define	KBDATAP		0x60	/* kbd data port */
+#define	KBSTATUSPORT	0x61	/* kbd status */
+#define	KBSTATP		0x64	/* kbd status port */
+#define	KBINRDY		0x01
+#define	KBOUTRDY	0x02
+
+#define _x__ 0x00  /* Unknown / unmapped */
+
+const unsigned short action[] = {
+	0,     ASCII, ASCII, ASCII, ASCII, ASCII, ASCII, ASCII,	/* scan  0- 7 */
+	ASCII, ASCII, ASCII, ASCII, ASCII, ASCII, ASCII, ASCII,	/* scan  8-15 */
+	ASCII, ASCII, ASCII, ASCII, ASCII, ASCII, ASCII, ASCII,	/* scan 16-23 */
+	ASCII, ASCII, ASCII, ASCII, ASCII,   CTL, ASCII, ASCII,	/* scan 24-31 */
+	ASCII, ASCII, ASCII, ASCII, ASCII, ASCII, ASCII, ASCII,	/* scan 32-39 */
+	ASCII, ASCII,   SHF, ASCII, ASCII, ASCII, ASCII, ASCII,	/* scan 40-47 */
+	ASCII, ASCII, ASCII, ASCII, ASCII, ASCII,   SHF, ASCII,	/* scan 48-55 */
+	  ALT, ASCII,   CPS,  FUNC,  FUNC,  FUNC,  FUNC,  FUNC,	/* scan 56-63 */
+	 FUNC,  FUNC,  FUNC,  FUNC,  FUNC,   NUM,SCROLL, ASCII,	/* scan 64-71 */
+	ASCII, ASCII, ASCII, ASCII, ASCII, ASCII, ASCII, ASCII,	/* scan 72-79 */
+	ASCII, ASCII, ASCII, ASCII,     0,     0,     0,     0,	/* scan 80-87 */
+	0,0,0,0,0,0,0,0,	/* scan 88-95 */
+	0,0,0,0,0,0,0,0,	/* scan 96-103 */
+	0,0,0,0,0,0,0,0,	/* scan 104-111 */
+	0,0,0,0,0,0,0,0,	/* scan 112-119 */
+	0,0,0,0,0,0,0,0,	/* scan 120-127 */
+};
+
+static int
+kbd(noblock)
+	int noblock;
+{
+	unsigned char dt, brk, act;
+	int first = 1;	
+loop:
+	if (noblock) {
+		if ((inb(KBSTATP) & KBINRDY) == 0)
+			return (-1);
+	} else while((inb(KBSTATP) & KBINRDY) == 0) ;
+
+	dt = inb(KBDATAP);
+
+	brk = dt & 0x80;	/* brk == 1 on key release */
+	dt = dt & 0x7f;		/* keycode */
+
+	act = action[dt];
+	if (/*act&SHF*/ dt == 54)
+		shfts = brk ? 0 : 1;
+	if (/*act&ALT*/ dt == 48)
+		alts = brk ? 0 : 1;
+	if (/*act&NUM*/ dt == 69)
+		if (act&L) {
+			/* NUM lock */
+			if(!brk)
+				num = !num;
+		} else
+			num = brk ? 0 : 1;
+	if (/*act&CTL*/ dt == 29)
+		ctls = brk ? 0 : 1;
+	if (/*act&CPS*/ dt == 58)
+		if (act&L) {
+			/* CAPS lock */
+			if(!brk)
+				caps = !caps;
+		} else
+			caps = brk ? 0 : 1;
+	if (0/*act&STP*/)
+		if (act&L) {
+			if(!brk)
+				stp = !stp;
+		} else
+			stp = brk ? 0 : 1;
+
+	if ((act&ASCII) && !brk) {
+		unsigned char chr;
+		if (shfts)
+			chr = shift_map[dt];
+		else if (ctls)
+			chr = ctrl_map[dt];
+		else
+			chr = plain_map[dt];
+		if (alts)
+			chr |= 0x80;
+
+		if (caps && (chr >= 'a' && chr <= 'z'))
+			chr -= 'a' - 'A' ;
+		if ( chr == 0x01 ) chr = '\n'; /* hack */
+#define CTRL(s) (s & 0x1F)			
+		if ((chr == '\r') || (chr == '\n') || (chr == CTRL('A')) || (chr == CTRL('S')))
+		{
+			/* Wait for key up */
+			while (1)
+			{
+				while((inb(KBSTATP) & KBINRDY) == 0) ;
+				dt = inb(KBDATAP);
+				if (dt & 0x80) /* key up */ break;
+			}
+		}
+		return (chr);
+	}
+	if (first && brk) return (0);  /* Ignore initial 'key up' codes */
+	goto loop;
+}
+
+static
+scankbd(void) {
+	return (kbd(1) != -1);
+}
+
+static 
+kbdreset(void)
+{
+	unsigned char c;
+	int i;
+
+	/* Send self-test */
+	while (inb(KBSTATP) & KBOUTRDY) ;
+	outb(KBSTATP,0xAA);
+	while ((inb(KBSTATP) & KBINRDY) == 0) ;	/* wait input ready */
+	if ((c = inb(KBDATAP)) != 0x55)
+	{
+		puts("Keyboard self test failed - result:");
+		puthex(c);
+		puts("\n");
+	}
+	/* Enable interrupts and keyboard controller */
+	while (inb(KBSTATP) & KBOUTRDY) ;
+	outb(KBSTATP,0x60);	
+	while (inb(KBSTATP) & KBOUTRDY) ;
+	outb(KBDATAP,0x45);
+	for (i = 0;  i < 10000;  i++) udelay(1);
+	while (inb(KBSTATP) & KBOUTRDY) ;
+	outb(KBSTATP,0xAE);
+}
+
+static int kbd_reset = 0;
+
+CRT_getc(void)
+{
+	int c;
+	if (!kbd_reset) {kbdreset(); kbd_reset++; }
+	while ((c = kbd(0)) == 0) ;
+	return(c);
+}
+
+CRT_tstc(void)
+{
+	if (!kbd_reset) {kbdreset(); kbd_reset++; }
+	return ((inb(KBSTATP) & KBINRDY) != 0);
+}
diff --git a/arch/ppc/boot/misc.c b/arch/ppc/boot/misc.c
index a983ea106..4d2a723f5 100644
--- a/arch/ppc/boot/misc.c
+++ b/arch/ppc/boot/misc.c
@@ -8,13 +8,23 @@
  * puts by Nick Holloway 1993
  *
  * Adapted for PowerPC by Gary Thomas
+ * Updated and modified by Cort Dougan (cort@cs.nmt.edu)
  */
 
 #include "gzip.h"
 #include "lzw.h"
 #include "asm/residual.h"
+#include <elf.h>
 
 RESIDUAL hold_residual;
+unsigned long initrd_start = 0, initrd_end = 0;
+char *zimage_start;
+int zimage_size;
+extern char input_data[];
+extern int input_len;
+void cksum_text(void);
+void verify_data(unsigned long load_addr);
+
 void dump_buf(unsigned char *p, int s);
 #define EOF -1
 
@@ -26,8 +36,7 @@ unsigned outcnt;
 unsigned insize;
 unsigned inptr;
 
-extern char input_data[];
-extern int input_len;
+char cmd_line[256];
 
 int input_ptr;
 
@@ -56,6 +65,9 @@ int lines, cols;
 int orig_x, orig_y;
 
 void puts(const char *);
+void putc(const char c);
+void puthex(unsigned long val);
+void _bcopy(char *src, char *dst, int len);
 
 void *malloc(int size)
 {
@@ -77,7 +89,7 @@ void *malloc(int size)
 	 */
           
 	if (free_mem_ptr < (long)&end) {
-		if (free_mem_ptr > (long)&input_data[input_ptr])
+		if (free_mem_ptr > (long)&zimage_start[input_ptr])
 			error("\nOut of memory\n");
 
 		return p;
@@ -87,7 +99,7 @@ void *malloc(int size)
 #endif	
 	return p;
 	puts("large kernel, low 1M tight...");
-	free_mem_ptr = (long)input_data;
+	free_mem_ptr = (long)zimage_start;
 	}
 }
 
@@ -115,6 +127,53 @@ static void scroll()
 		vidmem[i] = ' ';
 }
 
+tstc(void)
+{
+	return (CRT_tstc() );
+}
+
+getc(void)
+{
+	while (1) {
+		if (CRT_tstc()) return (CRT_getc());
+	}
+}
+
+void 
+putc(const char c)
+{
+	int x,y;
+
+	x = orig_x;
+	y = orig_y;
+
+	if ( c == '\n' ) {
+		x = 0;
+		if ( ++y >= lines ) {
+			scroll();
+			y--;
+		}
+	} else if (c == '\b') {
+		if (x > 0) {
+			x--;
+		}
+	} else {
+		vidmem [ ( x + cols * y ) * 2 ] = c; 
+		if ( ++x >= cols ) {
+			x = 0;
+			if ( ++y >= lines ) {
+				scroll();
+				y--;
+			}
+		}
+	}
+
+	cursor(x, y);
+
+	orig_x = x;
+	orig_y = y;
+}
+
 void puts(const char *s)
 {
 	int x,y;
@@ -229,10 +288,10 @@ puts("*");
     insize = 0;
     do {
 	len = INBUFSIZ-insize;
-	if (len > (input_len-input_ptr+1)) len=input_len-input_ptr+1;
+	if (len > (zimage_size-input_ptr+1)) len=zimage_size-input_ptr+1;
         if (len == 0 || len == EOF) break;
 
-        for (i=0;i<len;i++) inbuf[insize+i] = input_data[input_ptr+i];
+        for (i=0;i<len;i++) inbuf[insize+i] = zimage_start[input_ptr+i];
 	insize += len;
 	input_ptr += len;
     } while (insize < INBUFSIZ);
@@ -313,7 +372,15 @@ void error(char *x)
 unsigned long
 decompress_kernel(unsigned long load_addr, int num_words, unsigned long cksum, RESIDUAL *residual)
 {
-  unsigned long TotalMemory;
+  int timer;
+  char *cp, ch;
+  Elf32_Ehdr *eh;
+  Elf32_Shdr *sh, *strtab_shdr;
+  char *strtab;
+  unsigned long i;
+  extern unsigned long start(void);
+
+  
   output_data = (char *)0x0;	/* Points to 0 */
   lines = 25;
   cols = 80;
@@ -339,33 +406,135 @@ decompress_kernel(unsigned long load_addr, int num_words, unsigned long cksum, R
   clear_bufs();
   makecrc();
 
-  puts("Loaded at ");  puthex(load_addr);  puts(", ");  puthex(num_words);  puts(" words");
-  puts(", cksum = ");  puthex(cksum);  puts("\n");
+  puts("Cksum: ");  puthex(cksum);  puts("\n");
+  puts("Loaded at: ");  puthex(load_addr); puts(" ");  puthex(num_words+load_addr);
+  puts("\n");
+  puts("Boot code relocated to: ");  puthex((unsigned long)start); puts(" ");
+  puthex((unsigned long)(num_words+start));
+  puts("\n");
   if (residual) {
-    _bcopy(residual, &hold_residual, sizeof(hold_residual));
-    puts("Residual data at ");  puthex(residual);  puts("\n");
-    show_residual_data(residual);
-    TotalMemory = residual->TotalMemory;
-  } else {
-    TotalMemory = 0x01000000;
+    _bcopy((char *)residual, (char *)&hold_residual, sizeof(hold_residual));
+    puts("Residual data at: ");  puthex((unsigned long)residual); puts(" ");
+    puthex((unsigned long)((unsigned long)(residual->ResidualLength) + residual));  puts("\n");
+    puts("Residual data relocated to: ");  puthex((unsigned long)&hold_residual); puts("\n");
   }
 
-  puts("Uncompressing Linux...");
+  /*
+   * Take care of initrd if we have one
+   */
+  /*
+   * the _actual_ load addr is 64k (elf hdr size) lower but the
+   * firmware gives us the starting exec point not the load ptr
+   * -- Cort
+   */
+  eh = (Elf32_Ehdr *)(load_addr - 65536);
+  sh = (Elf32_Shdr *)((unsigned long)eh + eh->e_shoff );
+  /*puts("Entry point: "); puthex(eh->e_entry); puts("\n");*/
+  
+  /* find string table */
+  for ( i = 0 ; i <= eh->e_shnum ; i++,sh++)
+  {
+	  /*puts("Section: "); puthex(i);
+	  puts(" type: "); puthex(sh->sh_type);
+	  puts(" offset: "); puthex(sh->sh_offset);
+	  puts("\n");*/
+
+	  if ( sh->sh_type == SHT_STRTAB )
+	  {
+		  strtab_shdr = sh;
+		  strtab = (char *)(sh->sh_offset + (unsigned long)eh);
+		  /*puts("Found strtab at: "); puthex((unsigned long)strtab);
+		  puts("\n");*/
+		  break;
+	  }
+  }  
 
-  method = get_method(0);
+  /* find the initrd and image sections */
+  if ( strtab_shdr )
+  {
+	  sh = (Elf32_Shdr *)((unsigned long)eh + eh->e_shoff );
+	  for ( i = 0 ; i <= eh->e_shnum ; i++,sh++)
+	  {
+		  if ( !memcmp("initrd", (char *)(strtab+sh->sh_name), 6 ) )
+		  {
+			  initrd_start = (unsigned long)eh + sh->sh_offset;
+			  initrd_end =  initrd_start + sh->sh_size;
+			  puts("Found initrd at: "); puthex(initrd_start);
+			  puts(" "); puthex(initrd_end);
+			  puts("\n");
+		  }
+		  if ( !memcmp("image", (char *)(strtab+sh->sh_name), 5 ) )
+		  {
+			  zimage_start = (char *)((unsigned long)eh + sh->sh_offset);
+			  zimage_size =  sh->sh_size;
+			  puts("Found zimage at: "); puthex((unsigned long)zimage_start);
+			  puts(" "); puthex((unsigned long)(zimage_size+zimage_start));
+			  puts("\n");
+		  }
+	  }
+  }
+  else
+  {
+	  puts("Couldn't find string table for boot image!\n");
+  }
+  
+  /* relocate initrd */
+  if ( initrd_start )
+  {
+	  memcpy ((void *)0x00D00000,(void *)initrd_start,
+		  initrd_end - initrd_start );
+	  initrd_end = 0x00D00000 + initrd_end - initrd_start;
+	  initrd_start = 0x00D00000;
+	  puts("Moved initrd to: "); puthex(initrd_start);
+	  puts(" "); puthex(initrd_end); puts("\n");
+  }
 
-  work(0, 0);
 
+  CRT_tstc();  /* Forces keyboard to be initialized */
+  puts("\nLinux/PPC load: ");
+  timer = 0;
+  cp = cmd_line;
+  while (timer++ < 5*1000) {
+	  if (tstc()) {
+		  while ((ch = getc()) != '\n' && ch != '\r') {
+			  if (ch == '\b') {
+				  if (cp != cmd_line) {
+					  cp--;
+					  puts("\b \b");
+				  }
+			  } else {
+				  *cp++ = ch;
+				  putc(ch);
+			  }
+		  }
+		  break;  /* Exit 'timer' loop */
+	  }
+	  udelay(1000);  /* 1 msec */
+  }
+  *cp = 0;
+  puts("\n");
+
+  /* mappings on early boot can only handle 16M */
+  if ( (int)(&cmd_line[0]) > (16<<20))
+	  puts("cmd_line > 16M\n");
+  if ( (int)&hold_residual > (16<<20))
+	  puts("hold_residual > 16M\n");
+  if ( initrd_start > (16<<20))
+	  puts("initrd_start > 16M\n");
+       
+  
+  puts("Uncompressing Linux...");
+  method = get_method(0);
+  work(0, 0);
   puts("done.\n");
+  
   puts("Now booting the kernel\n");
-  /*return (TotalMemory);*/              /* Later this can be a pointer to saved residual data */
-  return &hold_residual;
+  return (unsigned long)&hold_residual;
 }
 
 show_residual_data(RESIDUAL *res)
 {
   puts("Residual data: ");  puthex(res->ResidualLength);  puts(" bytes\n");
-  puts("Total memory: ");  puthex(res->TotalMemory);  puts("\n");
 #if 0
   puts("Residual structure = ");  puthex(sizeof(*res));  puts(" bytes\n");
   dump_buf(&hold_residual, 32);
@@ -407,14 +576,14 @@ cksum_data()
 {
   unsigned int *ptr, len, cksum, cnt;
   cksum = cnt = 0;
-  ptr = input_data;
+  ptr = (unsigned int *)zimage_start;
   puts("Checksums: ");
-  for (len = 0;  len < input_len;  len += 4) {
+  for (len = 0;  len < zimage_size;  len += 4) {
     cksum ^= *ptr++;
     if (len && ((len & 0x0FFF) == 0)) {
       if (cnt == 0) {
 	puts("\n  [");
-	puthex(ptr-1);
+	puthex((unsigned long)ptr-1);
 	puts("] ");
       }
       puthex(cksum);
@@ -429,7 +598,7 @@ cksum_data()
   puts("Data cksum = ");  puthex(cksum);  puts("\n");
 }
 
-cksum_text()
+void cksum_text(void)
 {
   extern int start, etext;
   unsigned int *ptr, len, text_len, cksum, cnt;
@@ -442,7 +611,7 @@ cksum_text()
     if (len && ((len & 0x0FFF) == 0)) {
       if (cnt == 0) {
 	puts("\n  [");
-	puthex(ptr-1);
+	puthex((unsigned long)ptr-1);
 	puts("] ");
       }
       puthex(cksum);
@@ -457,24 +626,24 @@ cksum_text()
   puts("TEXT cksum = ");  puthex(cksum);  puts("\n");
 }
 
-verify_data(unsigned long load_addr)
+void verify_data(unsigned long load_addr)
 {
   extern int start, etext;
   unsigned int *orig_ptr, *copy_ptr, len, errors;
   errors = 0;
-  copy_ptr = input_data;
-  orig_ptr = (unsigned int *)(load_addr + ((unsigned int)input_data - (unsigned int)&start));
-  for (len = 0;  len < input_len;  len += 4) {
+  copy_ptr = (unsigned int *)zimage_start;
+  orig_ptr = (unsigned int *)(load_addr + ((unsigned int)zimage_start - (unsigned int)&start));
+  for (len = 0;  len < zimage_size;  len += 4) {
     if (*copy_ptr++ != *orig_ptr++) {
       errors++;
     }    
   }
-  copy_ptr = input_data;
-  orig_ptr = (unsigned int *)(load_addr + ((unsigned int)input_data - (unsigned int)&start));
-  for (len = 0;  len < input_len;  len += 4) {
+  copy_ptr = (unsigned int *)zimage_start;
+  orig_ptr = (unsigned int *)(load_addr + ((unsigned int)zimage_start - (unsigned int)&start));
+  for (len = 0;  len < zimage_size;  len += 4) {
     if (*copy_ptr++ != *orig_ptr++) {
-      dump_buf(copy_ptr-1, 128);
-      dump_buf(orig_ptr-1, 128);
+      dump_buf((unsigned char *) (copy_ptr-1), 128);
+      dump_buf((unsigned char *) (orig_ptr-1), 128);
       puts("Total errors = ");  puthex(errors*4);  puts("\n");
       while (1) ;
     }    
@@ -486,9 +655,9 @@ test_data(unsigned long load_addr)
   extern int start, etext;
   unsigned int *orig_ptr, *copy_ptr, len, errors;
   errors = 0;
-  copy_ptr = input_data;
-  orig_ptr = (unsigned int *)(load_addr + ((unsigned int)input_data - (unsigned int)&start));
-  for (len = 0;  len < input_len;  len += 4) {
+  copy_ptr = (unsigned int *)zimage_start;
+  orig_ptr = (unsigned int *)(load_addr + ((unsigned int)zimage_start - (unsigned int)&start));
+  for (len = 0;  len < zimage_size;  len += 4) {
     if (*copy_ptr++ != *orig_ptr++) {
       errors++;
     }    
@@ -532,7 +701,7 @@ void dump_buf(unsigned char *p, int s)
    }
    while (s > 0)
    {
-      puthex(p);  puts(": ");
+      puthex((unsigned long)p);  puts(": ");
       for (i = 0;  i < 16;  i++)
       {
          if (i < s)
diff --git a/arch/ppc/boot/mkprep.c b/arch/ppc/boot/mkprep.c
index 417334670..06b08df98 100644
--- a/arch/ppc/boot/mkprep.c
+++ b/arch/ppc/boot/mkprep.c
@@ -131,17 +131,18 @@ int main(int argc, char *argv[])
   argptr++;
 
   /* skip elf header in input file */
-  lseek(in_fd, elfhdr_size, SEEK_SET);
+  if ( !prep )
+	  lseek(in_fd, elfhdr_size, SEEK_SET);
   
   /* write prep partition if necessary */
   if ( prep )
-    write_prep_partition( in_fd, out_fd );
+	  write_prep_partition( in_fd, out_fd );
 
   /* write input image to bootimage */
   if ( asmoutput )
-    write_asm_data( in_fd, out_fd );
+	  write_asm_data( in_fd, out_fd );
   else
-    copy_image(in_fd, out_fd);
+	  copy_image(in_fd, out_fd);
   
   return 0;
 }
@@ -161,28 +162,11 @@ void write_prep_partition(int in, int out)
   }
   
   bzero( block, sizeof block );
-
  
-  /* set entry point and boot image size */
-  *entry = cpu_to_le32(0x400);
-  /* need use size - elfheader? */
+  /* set entry point and boot image size skipping over elf header */
+  *entry = cpu_to_le32(0x400+65536);
   *length = cpu_to_le32(info.st_size+0x400);
 
-  /*
-   * Writes the "boot record", which contains the partition table, to the
-   * diskette, followed by the dummy PC boot block and load image descriptor
-   * block.  It returns the number of bytes it has written to the load
-   * image.
-   *
-   * The boot record is the first block of the diskette and identifies the
-   * "PReP" partition.  The "PReP" partition contains the "load image" starting
-   * at offset zero within the partition.  The first block of the load image is
-   * a dummy PC boot block.  The second block is the "load image descriptor"
-   * which contains the size of the load image and the entry point into the
-   * image.  The actual boot image starts at offset 1024 bytes (third sector)
-   * in the partition.
-   */
-  
   /* sets magic number for msdos partition (used by linux) */
   block[510] = 0x55;
   block[511] = 0xAA;
@@ -220,7 +204,7 @@ void write_prep_partition(int in, int out)
   /* This has to be 0 on the PowerStack? */   
   pe->beginning_sector  = cpu_to_le32(0);
 #endif    
-/*pe->number_of_sectors = cpu_to_le32(2*18*80-1);*/
+  pe->number_of_sectors = cpu_to_le32(2*18*80-1);
 
   write( out, block, sizeof(block) );
   write( out, entry, sizeof(*entry) );
diff --git a/arch/ppc/coffboot/.cvsignore b/arch/ppc/coffboot/.cvsignore
new file mode 100644
index 000000000..45968cd8d
--- /dev/null
+++ b/arch/ppc/coffboot/.cvsignore
@@ -0,0 +1,7 @@
+.depend
+vmlinux.coff.initrd
+coffboot
+elfextract
+hack-coff
+vmlinux.coff
+zImage
diff --git a/arch/ppc/coffboot/Makefile b/arch/ppc/coffboot/Makefile
new file mode 100644
index 000000000..e145411c6
--- /dev/null
+++ b/arch/ppc/coffboot/Makefile
@@ -0,0 +1,50 @@
+# Makefile for making XCOFF bootable images for booting on PowerMacs
+# using Open Firmware.
+#
+# Paul Mackerras	January 1997
+
+# PowerPC (cross) tools
+ifneq ($(shell uname -m),ppc)
+CROSS_COMPILE 	=powerpc-eabi-
+endif
+
+HOSTCC	= gcc
+HOSTCFLAGS = -O -I$(TOPDIR)/include
+
+CC	= $(CROSS_COMPILE)gcc
+LD	= $(CROSS_COMPILE)ld
+CFLAGS	= -O -fno-builtin -I$(TOPDIR)/include
+OBJCOPY = $(CROSS_COMPILE)objcopy
+OBJCOPY_ARGS = -O aixcoff-rs6000 -R .stab -R .stabstr -R .comment \
+	--add-section=image=zImage
+LD_ARGS = -e _start -T ld.script -Ttext 500000 -Tdata 510000 -Bstatic
+GZIP = gzip -9
+
+OBJS = crt0.o start.o main.o misc.o string.o zlib.o
+LIBS = $(TOPDIR)/lib/lib.a
+
+vmlinux.coff: coffboot hack-coff zImage
+	$(OBJCOPY) $(OBJCOPY_ARGS) coffboot $@
+	./hack-coff $@
+
+vmlinux.coff.initrd: coffboot hack-coff zImage ramdisk.image.gz
+	$(OBJCOPY) $(OBJCOPY_ARGS) --add-section=initrd=ramdisk.image.gz \
+		coffboot $@
+	./hack-coff $@
+
+coffboot: $(OBJS) ld.script
+	$(LD) -o coffboot $(LD_ARGS) $(OBJS) $(LIBS)
+
+zImage: $(TOPDIR)/vmlinux elfextract
+	./elfextract $(TOPDIR)/vmlinux | $(GZIP) >zImage
+
+hack-coff: hack-coff.c
+	$(HOSTCC) $(HOSTCFLAGS) -o hack-coff hack-coff.c
+
+elfextract: elfextract.c
+	$(HOSTCC) $(HOSTCFLAGS) -o elfextract elfextract.c
+
+clean:
+	rm -f elfextract hack-coff coffboot zImage vmlinux.coff
+
+fastdep:
diff --git a/arch/ppc/coffboot/crt0.S b/arch/ppc/coffboot/crt0.S
new file mode 100644
index 000000000..43ae4e0ec
--- /dev/null
+++ b/arch/ppc/coffboot/crt0.S
@@ -0,0 +1,24 @@
+/*
+ * Copyright (C) Paul Mackerras 1997.
+ *
+ * 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.
+ */
+	.text
+	.globl	_start
+_start:
+	.long	__start,0,0
+
+	.globl	__start
+__start:
+	lis	9,_start@h
+	lis	8,_etext@ha
+	addi	8,8,_etext@l
+1:	dcbf	0,9
+	icbi	0,9
+	addi	9,9,0x20
+	cmplwi	0,9,8
+	blt	1b
+	b	start
diff --git a/arch/ppc/coffboot/elfextract.c b/arch/ppc/coffboot/elfextract.c
new file mode 100644
index 000000000..1b50ca475
--- /dev/null
+++ b/arch/ppc/coffboot/elfextract.c
@@ -0,0 +1,98 @@
+/*
+ * Extract the loadable program segment from an elf file.
+ *
+ * Copyright 1996 Paul Mackerras.
+ *
+ * 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.
+ */
+#include <stdio.h>
+#include <linux/elf.h>
+
+FILE *fi, *fo;
+char *ni, *no;
+char buf[65536];
+
+void
+rd(void *buf, int len)
+{
+    int nr;
+
+    nr = fread(buf, 1, len, fi);
+    if (nr == len)
+	return;
+    if (ferror(fi))
+	fprintf(stderr, "%s: read error\n", ni);
+    else
+	fprintf(stderr, "%s: short file\n", ni);
+    exit(1);
+}
+
+main(int ac, char **av)
+{
+    unsigned nb, len;
+    Elf32_Ehdr eh;
+    Elf32_Phdr ph;
+
+    if (ac > 3 || ac > 1 && av[1][0] == '-') {
+	fprintf(stderr, "Usage: %s [elf-file [image-file]]\n", av[0]);
+	exit(0);
+    }
+
+    fi = stdin;
+    ni = "(stdin)";
+    fo = stdout;
+    no = "(stdout)";
+
+    if (ac > 1) {
+	ni = av[1];
+	fi = fopen(ni, "rb");
+	if (fi == NULL) {
+	    perror(ni);
+	    exit(1);
+	}
+    }
+
+    rd(&eh, sizeof(eh));
+    if (eh.e_ident[EI_MAG0] != ELFMAG0
+	|| eh.e_ident[EI_MAG1] != ELFMAG1
+	|| eh.e_ident[EI_MAG2] != ELFMAG2
+	|| eh.e_ident[EI_MAG3] != ELFMAG3) {
+	fprintf(stderr, "%s: not an ELF file\n", ni);
+	exit(1);
+    }
+
+    fseek(fi, eh.e_phoff + (eh.e_phnum - 1) * sizeof(ph), 0);
+    rd(&ph, sizeof(ph));
+    if (ph.p_type != PT_LOAD) {
+	fprintf(stderr, "%s: doesn't have a loadable segment\n", ni);
+	exit(1);
+    }
+
+    if (ac > 2) {
+	no = av[2];
+	fo = fopen(no, "wb");
+	if (fo == NULL) {
+	    perror(no);
+	    exit(1);
+	}
+    }
+
+    fseek(fi, ph.p_offset, 0);
+    for (len = ph.p_filesz; len != 0; len -= nb) {
+	nb = len;
+	if (nb > sizeof(buf))
+	    nb = sizeof(buf);
+	rd(buf, nb);
+	if (fwrite(buf, 1, nb, fo) != nb) {
+	    fprintf(stderr, "%s: write error\n", no);
+	    exit(1);
+	}
+    }
+
+    fclose(fo);
+    fclose(fi);
+    exit(0);
+}
diff --git a/arch/ppc/coffboot/hack-coff.c b/arch/ppc/coffboot/hack-coff.c
new file mode 100644
index 000000000..3dfa8d800
--- /dev/null
+++ b/arch/ppc/coffboot/hack-coff.c
@@ -0,0 +1,79 @@
+/*
+ * hack-coff.c - hack the header of an xcoff file to fill in
+ * a few fields needed by the Open Firmware xcoff loader on
+ * Power Macs but not initialized by objcopy.
+ *
+ * Copyright (C) Paul Mackerras 1997.
+ *
+ * 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.
+ */
+#include <stdio.h>
+#include "rs6000.h"
+
+#define AOUT_MAGIC	0x010b
+
+#define get_16be(x)	((((unsigned char *)(x))[0] << 8) \
+			 + ((unsigned char *)(x))[1])
+#define put_16be(x, v)	(((unsigned char *)(x))[0] = (v) >> 8, \
+			 ((unsigned char *)(x))[1] = (v) & 0xff)
+#define get_32be(x)	((((unsigned char *)(x))[0] << 24) \
+			 + (((unsigned char *)(x))[1] << 16) \
+			 + (((unsigned char *)(x))[2] << 8) \
+			 + ((unsigned char *)(x))[3])
+
+main(int ac, char **av)
+{
+    int fd;
+    int i, nsect;
+    int aoutsz;
+    struct external_filehdr fhdr;
+    AOUTHDR aout;
+    struct external_scnhdr shdr;
+
+    if (ac != 2) {
+	fprintf(stderr, "Usage: hack-coff coff-file\n");
+	exit(1);
+    }
+    if ((fd = open(av[1], 2)) == -1) {
+	perror(av[2]);
+	exit(1);
+    }
+    if (read(fd, &fhdr, sizeof(fhdr)) != sizeof(fhdr))
+	goto readerr;
+    i = get_16be(fhdr.f_magic);
+    if (i != U802TOCMAGIC && i != U802WRMAGIC && i != U802ROMAGIC) {
+	fprintf(stderr, "%s: not an xcoff file\n", av[1]);
+	exit(1);
+    }
+    aoutsz = get_16be(fhdr.f_opthdr);
+    if (read(fd, &aout, aoutsz) != aoutsz)
+	goto readerr;
+    nsect = get_16be(fhdr.f_nscns);
+    for (i = 0; i < nsect; ++i) {
+	if (read(fd, &shdr, sizeof(shdr)) != sizeof(shdr))
+	    goto readerr;
+	if (strcmp(shdr.s_name, ".text") == 0) {
+	    put_16be(aout.o_snentry, i+1);
+	    put_16be(aout.o_sntext, i+1);
+	} else if (strcmp(shdr.s_name, ".data") == 0) {
+	    put_16be(aout.o_sndata, i+1);
+	} else if (strcmp(shdr.s_name, ".bss") == 0) {
+	    put_16be(aout.o_snbss, i+1);
+	}
+    }
+    put_16be(aout.magic, AOUT_MAGIC);
+    if (lseek(fd, (long) sizeof(struct external_filehdr), 0) == -1
+	|| write(fd, &aout, aoutsz) != aoutsz) {
+	fprintf(stderr, "%s: write error\n", av[1]);
+	exit(1);
+    }
+    close(fd);
+    exit(0);
+
+readerr:
+    fprintf(stderr, "%s: read error or file too short\n", av[1]);
+    exit(1);
+}
diff --git a/arch/ppc/coffboot/ld.script b/arch/ppc/coffboot/ld.script
new file mode 100644
index 000000000..2469ed65d
--- /dev/null
+++ b/arch/ppc/coffboot/ld.script
@@ -0,0 +1,68 @@
+OUTPUT_ARCH(powerpc)
+SEARCH_DIR(/lib); SEARCH_DIR(/usr/lib); SEARCH_DIR(/usr/local/lib); SEARCH_DIR(/usr/local/powerpc-any-elf/lib);
+/* Do we need any of these for elf?
+   __DYNAMIC = 0;    */
+SECTIONS
+{
+  /* Read-only sections, merged into text segment: */
+  . = + SIZEOF_HEADERS;
+  .interp : { *(.interp) }
+  .hash          : { *(.hash)		}
+  .dynsym        : { *(.dynsym)		}
+  .dynstr        : { *(.dynstr)		}
+  .rel.text      : { *(.rel.text)		}
+  .rela.text     : { *(.rela.text) 	}
+  .rel.data      : { *(.rel.data)		}
+  .rela.data     : { *(.rela.data) 	}
+  .rel.rodata    : { *(.rel.rodata) 	}
+  .rela.rodata   : { *(.rela.rodata) 	}
+  .rel.got       : { *(.rel.got)		}
+  .rela.got      : { *(.rela.got)		}
+  .rel.ctors     : { *(.rel.ctors)	}
+  .rela.ctors    : { *(.rela.ctors)	}
+  .rel.dtors     : { *(.rel.dtors)	}
+  .rela.dtors    : { *(.rela.dtors)	}
+  .rel.bss       : { *(.rel.bss)		}
+  .rela.bss      : { *(.rela.bss)		}
+  .rel.plt       : { *(.rel.plt)		}
+  .rela.plt      : { *(.rela.plt)		}
+  .init          : { *(.init)	} =0
+  .plt : { *(.plt) }
+  .text      :
+  {
+    *(.text)
+    *(.rodata)
+    *(.rodata1)
+    *(.got1)
+  }
+  .fini      : { *(.fini)    } =0
+  .ctors     : { *(.ctors)   }
+  .dtors     : { *(.dtors)   }
+  _etext = .;
+  PROVIDE (etext = .);
+  /* Read-write section, merged into data segment: */
+  . = (. + 0x0FFF) & 0xFFFFF000;
+  .data    :
+  {
+    *(.data)
+    *(.data1)
+    *(.sdata)
+    *(.sdata2)
+    *(.got.plt) *(.got)
+    *(.dynamic)
+    CONSTRUCTORS
+  }
+  _edata  =  .;
+  PROVIDE (edata = .);
+  __bss_start = .;
+  .bss       :
+  {
+   *(.sbss) *(.scommon)
+   *(.dynbss)
+   *(.bss)
+   *(COMMON)
+  }
+  _end = . ;
+  PROVIDE (end = .);
+}
+
diff --git a/arch/ppc/coffboot/main.c b/arch/ppc/coffboot/main.c
new file mode 100644
index 000000000..ab0ee6c4a
--- /dev/null
+++ b/arch/ppc/coffboot/main.c
@@ -0,0 +1,185 @@
+/*
+ * Copyright (C) Paul Mackerras 1997.
+ *
+ * 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.
+ */
+#include "nonstdio.h"
+#include "rs6000.h"
+#include "zlib.h"
+
+extern void *finddevice(const char *);
+extern int getprop(void *, const char *, void *, int);
+void gunzip(void *, int, unsigned char *, int *);
+
+#define get_16be(x)	(*(unsigned short *)(x))
+#define get_32be(x)	(*(unsigned *)(x))
+
+#define RAM_START	0xc0000000
+#define RAM_END		0xc0800000	/* only 8M mapped with BATs */
+
+#define RAM_FREE	0xc0540000	/* after image of coffboot */
+
+char *avail_ram;
+char *end_avail;
+
+coffboot(int a1, int a2, void *prom)
+{
+    void *options;
+    unsigned loadbase;
+    struct external_filehdr *eh;
+    struct external_scnhdr *sp;
+    struct external_scnhdr *isect, *rsect;
+    int ns, oh, i;
+    unsigned sa, len;
+    void *dst;
+    unsigned char *im;
+    unsigned initrd_start, initrd_size;
+
+    printf("coffboot starting\n");
+    options = finddevice("/options");
+    if (options == (void *) -1)
+	exit();
+    if (getprop(options, "load-base", &loadbase, sizeof(loadbase))
+	!= sizeof(loadbase)) {
+	printf("error getting load-base\n");
+	exit();
+    }
+    setup_bats();
+
+    eh = (struct external_filehdr *) loadbase;
+    ns = get_16be(eh->f_nscns);
+    oh = get_16be(eh->f_opthdr);
+
+    sp = (struct external_scnhdr *) (loadbase + sizeof(struct external_filehdr) + oh);
+    isect = rsect = NULL;
+    for (i = 0; i < ns; ++i, ++sp) {
+	if (strcmp(sp->s_name, "image") == 0)
+	    isect = sp;
+	else if (strcmp(sp->s_name, "initrd") == 0)
+	    rsect = sp;
+    }
+    if (isect == NULL) {
+	printf("image section not found\n");
+	exit();
+    }
+
+    if (rsect != NULL && (initrd_size = get_32be(rsect->s_size)) != 0) {
+	initrd_start = (RAM_END - initrd_size) & ~0xFFF;
+	a1 = initrd_start;
+	a2 = initrd_size;
+	printf("initial ramdisk at %x (%u bytes)\n",
+	       initrd_start, initrd_size);
+	memcpy((char *) initrd_start,
+	       (char *) (loadbase + get_32be(rsect->s_scnptr)),
+	       initrd_size);
+	end_avail = (char *) initrd_start;
+    } else {
+	end_avail = (char *) RAM_END;
+    }
+	
+    im = (unsigned char *)(loadbase + get_32be(isect->s_scnptr));
+    len = get_32be(isect->s_size);
+    dst = (void *) RAM_START;
+
+    if (im[0] == 0x1f && im[1] == 0x8b) {
+	void *cp = (void *) RAM_FREE;
+	avail_ram = (void *) (RAM_FREE + ((len + 7) & -8));
+	memcpy(cp, im, len);
+	printf("gunzipping... ");
+	gunzip(dst, 0x400000, cp, &len);
+	printf("done\n");
+
+    } else {
+	memmove(dst, im, len);
+    }
+
+    flush_cache(dst, len);
+
+    sa = *(unsigned *)dst + RAM_START;
+    printf("start address = 0x%x\n", sa);
+
+#if 0
+    pause();
+#endif
+    (*(void (*)())sa)(a1, a2, prom);
+
+    printf("returned?\n");
+
+    pause();
+}
+
+void *zalloc(void *x, unsigned items, unsigned size)
+{
+    void *p = avail_ram;
+
+    size *= items;
+    size = (size + 7) & -8;
+    avail_ram += size;
+    if (avail_ram > end_avail) {
+	printf("oops... out of memory\n");
+	pause();
+    }
+    return p;
+}
+
+void zfree(void *x, void *addr, unsigned nb)
+{
+}
+
+#define HEAD_CRC	2
+#define EXTRA_FIELD	4
+#define ORIG_NAME	8
+#define COMMENT		0x10
+#define RESERVED	0xe0
+
+#define DEFLATED	8
+
+void gunzip(void *dst, int dstlen, unsigned char *src, int *lenp)
+{
+    z_stream s;
+    int r, i, flags;
+
+    /* skip header */
+    i = 10;
+    flags = src[3];
+    if (src[2] != DEFLATED || (flags & RESERVED) != 0) {
+	printf("bad gzipped data\n");
+	exit();
+    }
+    if ((flags & EXTRA_FIELD) != 0)
+	i = 12 + src[10] + (src[11] << 8);
+    if ((flags & ORIG_NAME) != 0)
+	while (src[i++] != 0)
+	    ;
+    if ((flags & COMMENT) != 0)
+	while (src[i++] != 0)
+	    ;
+    if ((flags & HEAD_CRC) != 0)
+	i += 2;
+    if (i >= *lenp) {
+	printf("gunzip: ran out of data in header\n");
+	exit();
+    }
+
+    s.zalloc = zalloc;
+    s.zfree = zfree;
+    r = inflateInit2(&s, -MAX_WBITS);
+    if (r != Z_OK) {
+	printf("inflateInit2 returned %d\n", r);
+	exit();
+    }
+    s.next_in = src + i;
+    s.avail_in = *lenp - i;
+    s.next_out = dst;
+    s.avail_out = dstlen;
+    r = inflate(&s, Z_FINISH);
+    if (r != Z_OK && r != Z_STREAM_END) {
+	printf("inflate returned %d\n", r);
+	exit();
+    }
+    *lenp = s.next_out - (unsigned char *) dst;
+    inflateEnd(&s);
+}
diff --git a/arch/ppc/coffboot/misc.S b/arch/ppc/coffboot/misc.S
new file mode 100644
index 000000000..ae08ee2fa
--- /dev/null
+++ b/arch/ppc/coffboot/misc.S
@@ -0,0 +1,50 @@
+/*
+ * Copyright (C) Paul Mackerras 1997.
+ *
+ * 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.
+ */
+	.text
+
+/*
+ * Use the BAT0 registers to map the 1st 8MB of RAM to 0xc0000000.
+ */
+	.globl	setup_bats
+setup_bats:
+	mfpvr	3
+	rlwinm	3,3,16,16,31		/* r3 = 1 for 601, 4 for 604 */
+	cmpi	0,3,1
+	lis	4,0xc000
+	bne	4f
+	ori	4,4,4			/* set up BAT registers for 601 */
+	li	5,0x7f
+	b	5f
+4:	ori	4,4,0xff		/* set up BAT registers for 604 */
+	li	5,2
+	mtdbatu	0,4
+	mtdbatl	0,5
+5:	mtibatu	0,4
+	mtibatl	0,5
+	isync
+	blr
+
+/*
+ * Flush the dcache and invalidate the icache for a range of addresses.
+ *
+ * flush_cache(addr, len)
+ */
+	.global	flush_cache
+flush_cache:
+	addi	4,4,0x1f	/* len = (len + 0x1f) / 0x20 */
+	rlwinm.	4,4,27,5,31
+	mtctr	4
+	beqlr
+1:	dcbf	0,3
+	icbi	0,3
+	addi	3,3,0x20
+	bdnz	1b
+	sync
+	isync
+	blr
diff --git a/arch/ppc/coffboot/nonstdio.h b/arch/ppc/coffboot/nonstdio.h
new file mode 100644
index 000000000..664b8384a
--- /dev/null
+++ b/arch/ppc/coffboot/nonstdio.h
@@ -0,0 +1,18 @@
+/*
+ * Copyright (C) Paul Mackerras 1997.
+ *
+ * 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.
+ */
+typedef int	FILE;
+extern FILE *stdin, *stdout;
+#define NULL	((void *)0)
+#define EOF	(-1)
+#define fopen(n, m)	NULL
+#define fflush(f)	0
+#define fclose(f)	0
+extern char *fgets();
+
+#define perror(s)	printf("%s: no files!\n", (s))
diff --git a/arch/ppc/coffboot/rs6000.h b/arch/ppc/coffboot/rs6000.h
new file mode 100644
index 000000000..0def1d95d
--- /dev/null
+++ b/arch/ppc/coffboot/rs6000.h
@@ -0,0 +1,243 @@
+/* IBM RS/6000 "XCOFF" file definitions for BFD.
+   Copyright (C) 1990, 1991 Free Software Foundation, Inc.
+   FIXME: Can someone provide a transliteration of this name into ASCII?
+   Using the following chars caused a compiler warning on HIUX (so I replaced
+   them with octal escapes), and isn't useful without an understanding of what
+   character set it is.
+   Written by Mimi Ph\373\364ng-Th\345o V\365 of IBM
+   and John Gilmore of Cygnus Support.  */
+
+/********************** FILE HEADER **********************/
+
+struct external_filehdr {
+	char f_magic[2];	/* magic number			*/
+	char f_nscns[2];	/* number of sections		*/
+	char f_timdat[4];	/* time & date stamp		*/
+	char f_symptr[4];	/* file pointer to symtab	*/
+	char f_nsyms[4];	/* number of symtab entries	*/
+	char f_opthdr[2];	/* sizeof(optional hdr)		*/
+	char f_flags[2];	/* flags			*/
+};
+
+        /* IBM RS/6000 */
+#define U802WRMAGIC     0730    /* writeable text segments **chh**      */
+#define U802ROMAGIC     0735    /* readonly sharable text segments      */
+#define U802TOCMAGIC    0737    /* readonly text segments and TOC       */
+
+#define BADMAG(x)	\
+	((x).f_magic != U802ROMAGIC && (x).f_magic != U802WRMAGIC && \
+	 (x).f_magic != U802TOCMAGIC)
+
+#define	FILHDR	struct external_filehdr
+#define	FILHSZ	20
+
+
+/********************** AOUT "OPTIONAL HEADER" **********************/
+
+
+typedef struct 
+{
+  unsigned char	magic[2];	/* type of file			*/
+  unsigned char	vstamp[2];	/* version stamp		*/
+  unsigned char	tsize[4];	/* text size in bytes, padded to FW bdry */
+  unsigned char	dsize[4];	/* initialized data "  "	*/
+  unsigned char	bsize[4];	/* uninitialized data "   "	*/
+  unsigned char	entry[4];	/* entry pt.			*/
+  unsigned char	text_start[4];	/* base of text used for this file */
+  unsigned char	data_start[4];	/* base of data used for this file */
+  unsigned char	o_toc[4];	/* address of TOC */
+  unsigned char	o_snentry[2];	/* section number of entry point */
+  unsigned char	o_sntext[2];	/* section number of .text section */
+  unsigned char	o_sndata[2];	/* section number of .data section */
+  unsigned char	o_sntoc[2];	/* section number of TOC */
+  unsigned char	o_snloader[2];	/* section number of .loader section */
+  unsigned char	o_snbss[2];	/* section number of .bss section */
+  unsigned char	o_algntext[2];	/* .text alignment */
+  unsigned char	o_algndata[2];	/* .data alignment */
+  unsigned char	o_modtype[2];	/* module type (??) */
+  unsigned char o_cputype[2];	/* cpu type */
+  unsigned char	o_maxstack[4];	/* max stack size (??) */
+  unsigned char o_maxdata[4];	/* max data size (??) */
+  unsigned char	o_resv2[12];	/* reserved */
+}
+AOUTHDR;
+
+#define AOUTSZ 72
+#define SMALL_AOUTSZ (28)
+#define AOUTHDRSZ 72
+
+#define	RS6K_AOUTHDR_OMAGIC	0x0107	/* old: text & data writeable */
+#define	RS6K_AOUTHDR_NMAGIC	0x0108	/* new: text r/o, data r/w */
+#define	RS6K_AOUTHDR_ZMAGIC	0x010B	/* paged: text r/o, both page-aligned */
+
+
+/********************** SECTION HEADER **********************/
+
+
+struct external_scnhdr {
+	char		s_name[8];	/* section name			*/
+	char		s_paddr[4];	/* physical address, aliased s_nlib */
+	char		s_vaddr[4];	/* virtual address		*/
+	char		s_size[4];	/* section size			*/
+	char		s_scnptr[4];	/* file ptr to raw data for section */
+	char		s_relptr[4];	/* file ptr to relocation	*/
+	char		s_lnnoptr[4];	/* file ptr to line numbers	*/
+	char		s_nreloc[2];	/* number of relocation entries	*/
+	char		s_nlnno[2];	/* number of line number entries*/
+	char		s_flags[4];	/* flags			*/
+};
+
+/*
+ * names of "special" sections
+ */
+#define _TEXT	".text"
+#define _DATA	".data"
+#define _BSS	".bss"
+#define _PAD	".pad"
+#define _LOADER	".loader"
+
+#define	SCNHDR	struct external_scnhdr
+#define	SCNHSZ	40
+
+/* XCOFF uses a special .loader section with type STYP_LOADER.  */
+#define STYP_LOADER 0x1000
+
+/* XCOFF uses a special .debug section with type STYP_DEBUG.  */
+#define STYP_DEBUG 0x2000
+
+/* XCOFF handles line number or relocation overflow by creating
+   another section header with STYP_OVRFLO set.  */
+#define STYP_OVRFLO 0x8000
+
+/********************** LINE NUMBERS **********************/
+
+/* 1 line number entry for every "breakpointable" source line in a section.
+ * Line numbers are grouped on a per function basis; first entry in a function
+ * grouping will have l_lnno = 0 and in place of physical address will be the
+ * symbol table index of the function name.
+ */
+struct external_lineno {
+	union {
+		char l_symndx[4];	/* function name symbol index, iff l_lnno == 0*/
+		char l_paddr[4];	/* (physical) address of line number	*/
+	} l_addr;
+	char l_lnno[2];	/* line number		*/
+};
+
+
+#define	LINENO	struct external_lineno
+#define	LINESZ	6
+
+
+/********************** SYMBOLS **********************/
+
+#define E_SYMNMLEN	8	/* # characters in a symbol name	*/
+#define E_FILNMLEN	14	/* # characters in a file name		*/
+#define E_DIMNUM	4	/* # array dimensions in auxiliary entry */
+
+struct external_syment 
+{
+  union {
+    char e_name[E_SYMNMLEN];
+    struct {
+      char e_zeroes[4];
+      char e_offset[4];
+    } e;
+  } e;
+  char e_value[4];
+  char e_scnum[2];
+  char e_type[2];
+  char e_sclass[1];
+  char e_numaux[1];
+};
+
+
+
+#define N_BTMASK	(017)
+#define N_TMASK		(060)
+#define N_BTSHFT	(4)
+#define N_TSHIFT	(2)
+  
+
+union external_auxent {
+	struct {
+		char x_tagndx[4];	/* str, un, or enum tag indx */
+		union {
+			struct {
+			    char  x_lnno[2]; /* declaration line number */
+			    char  x_size[2]; /* str/union/array size */
+			} x_lnsz;
+			char x_fsize[4];	/* size of function */
+		} x_misc;
+		union {
+			struct {		/* if ISFCN, tag, or .bb */
+			    char x_lnnoptr[4];	/* ptr to fcn line # */
+			    char x_endndx[4];	/* entry ndx past block end */
+			} x_fcn;
+			struct {		/* if ISARY, up to 4 dimen. */
+			    char x_dimen[E_DIMNUM][2];
+			} x_ary;
+		} x_fcnary;
+		char x_tvndx[2];		/* tv index */
+	} x_sym;
+
+	union {
+		char x_fname[E_FILNMLEN];
+		struct {
+			char x_zeroes[4];
+			char x_offset[4];
+		} x_n;
+	} x_file;
+
+	struct {
+		char x_scnlen[4];			/* section length */
+		char x_nreloc[2];	/* # relocation entries */
+		char x_nlinno[2];	/* # line numbers */
+	} x_scn;
+
+        struct {
+		char x_tvfill[4];	/* tv fill value */
+		char x_tvlen[2];	/* length of .tv */
+		char x_tvran[2][2];	/* tv range */
+	} x_tv;		/* info about .tv section (in auxent of symbol .tv)) */
+
+	struct {
+		unsigned char x_scnlen[4];
+		unsigned char x_parmhash[4];
+		unsigned char x_snhash[2];
+		unsigned char x_smtyp[1];
+		unsigned char x_smclas[1];
+		unsigned char x_stab[4];
+		unsigned char x_snstab[2];
+	} x_csect;
+
+};
+
+#define	SYMENT	struct external_syment
+#define	SYMESZ	18	
+#define	AUXENT	union external_auxent
+#define	AUXESZ	18
+#define DBXMASK 0x80		/* for dbx storage mask */
+#define SYMNAME_IN_DEBUG(symptr) ((symptr)->n_sclass & DBXMASK)
+
+
+
+/********************** RELOCATION DIRECTIVES **********************/
+
+
+struct external_reloc {
+  char r_vaddr[4];
+  char r_symndx[4];
+  char r_size[1];
+  char r_type[1];
+};
+
+
+#define RELOC struct external_reloc
+#define RELSZ 10
+
+#define DEFAULT_DATA_SECTION_ALIGNMENT 4
+#define DEFAULT_BSS_SECTION_ALIGNMENT 4
+#define DEFAULT_TEXT_SECTION_ALIGNMENT 4
+/* For new sections we havn't heard of before */
+#define DEFAULT_SECTION_ALIGNMENT 4
diff --git a/arch/ppc/coffboot/start.c b/arch/ppc/coffboot/start.c
new file mode 100644
index 000000000..363e65962
--- /dev/null
+++ b/arch/ppc/coffboot/start.c
@@ -0,0 +1,280 @@
+/*
+ * Copyright (C) Paul Mackerras 1997.
+ *
+ * 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.
+ */
+#include <stdarg.h>
+
+int (*prom)();
+
+void *chosen_handle;
+void *stdin;
+void *stdout;
+void *stderr;
+
+void exit(void);
+void *finddevice(const char *name);
+int getprop(void *phandle, const char *name, void *buf, int buflen);
+
+void
+start(int a1, int a2, void *promptr)
+{
+    prom = (int (*)()) promptr;
+    chosen_handle = finddevice("/chosen");
+    if (chosen_handle == (void *) -1)
+	exit();
+    if (getprop(chosen_handle, "stdout", &stdout, sizeof(stdout)) != 4)
+	exit();
+    stderr = stdout;
+    if (getprop(chosen_handle, "stdin", &stdin, sizeof(stdin)) != 4)
+	exit();
+
+    coffboot(a1, a2, promptr);
+    for (;;)
+	exit();
+}
+
+int
+write(void *handle, void *ptr, int nb)
+{
+    struct prom_args {
+	char *service;
+	int nargs;
+	int nret;
+	void *ihandle;
+	void *addr;
+	int len;
+	int actual;
+    } args;
+
+    args.service = "write";
+    args.nargs = 3;
+    args.nret = 1;
+    args.ihandle = handle;
+    args.addr = ptr;
+    args.len = nb;
+    args.actual = -1;
+    (*prom)(&args);
+    return args.actual;
+}
+
+int
+read(void *handle, void *ptr, int nb)
+{
+    struct prom_args {
+	char *service;
+	int nargs;
+	int nret;
+	void *ihandle;
+	void *addr;
+	int len;
+	int actual;
+    } args;
+
+    args.service = "read";
+    args.nargs = 3;
+    args.nret = 1;
+    args.ihandle = handle;
+    args.addr = ptr;
+    args.len = nb;
+    args.actual = -1;
+    (*prom)(&args);
+    return args.actual;
+}
+
+void
+exit()
+{
+    struct prom_args {
+	char *service;
+    } args;
+
+    for (;;) {
+	args.service = "exit";
+	(*prom)(&args);
+    }
+}
+
+void
+pause()
+{
+    struct prom_args {
+	char *service;
+    } args;
+
+    args.service = "enter";
+    (*prom)(&args);
+}
+
+void *
+finddevice(const char *name)
+{
+    struct prom_args {
+	char *service;
+	int nargs;
+	int nret;
+	const char *devspec;
+	void *phandle;
+    } args;
+
+    args.service = "finddevice";
+    args.nargs = 1;
+    args.nret = 1;
+    args.devspec = name;
+    args.phandle = (void *) -1;
+    (*prom)(&args);
+    return args.phandle;
+}
+
+int
+getprop(void *phandle, const char *name, void *buf, int buflen)
+{
+    struct prom_args {
+	char *service;
+	int nargs;
+	int nret;
+	void *phandle;
+	const char *name;
+	void *buf;
+	int buflen;
+	int size;
+    } args;
+
+    args.service = "getprop";
+    args.nargs = 4;
+    args.nret = 1;
+    args.phandle = phandle;
+    args.name = name;
+    args.buf = buf;
+    args.buflen = buflen;
+    args.size = -1;
+    (*prom)(&args);
+    return args.size;
+}
+
+int
+putc(int c, void *f)
+{
+    char ch = c;
+
+    if (c == '\n')
+	putc('\r', f);
+    return write(f, &ch, 1) == 1? c: -1;
+}
+
+int
+putchar(int c)
+{
+    return putc(c, stdout);
+}
+
+int
+fputs(char *str, void *f)
+{
+    int n = strlen(str);
+
+    return write(f, str, n) == n? 0: -1;
+}
+
+int
+readchar()
+{
+    char ch;
+
+    for (;;) {
+	switch (read(stdin, &ch, 1)) {
+	case 1:
+	    return ch;
+	case -1:
+	    printk("read(stdin) returned -1\r\n");
+	    return -1;
+	}
+    }
+}
+
+static char line[256];
+static char *lineptr;
+static int lineleft;
+
+int
+getchar()
+{
+    int c;
+
+    if (lineleft == 0) {
+	lineptr = line;
+	for (;;) {
+	    c = readchar();
+	    if (c == -1 || c == 4)
+		break;
+	    if (c == '\r' || c == '\n') {
+		*lineptr++ = '\n';
+		putchar('\n');
+		break;
+	    }
+	    switch (c) {
+	    case 0177:
+	    case '\b':
+		if (lineptr > line) {
+		    putchar('\b');
+		    putchar(' ');
+		    putchar('\b');
+		    --lineptr;
+		}
+		break;
+	    case 'U' & 0x1F:
+		while (lineptr > line) {
+		    putchar('\b');
+		    putchar(' ');
+		    putchar('\b');
+		    --lineptr;
+		}
+		break;
+	    default:
+		if (lineptr >= &line[sizeof(line) - 1])
+		    putchar('\a');
+		else {
+		    putchar(c);
+		    *lineptr++ = c;
+		}
+	    }
+	}
+	lineleft = lineptr - line;
+	lineptr = line;
+    }
+    if (lineleft == 0)
+	return -1;
+    --lineleft;
+    return *lineptr++;
+}
+
+extern int vsprintf(char *buf, const char *fmt, va_list args);
+static char sprint_buf[1024];
+
+void
+printk(char *fmt, ...)
+{
+	va_list args;
+	int n;
+
+	va_start(args, fmt);
+	n = vsprintf(sprint_buf, fmt, args);
+	va_end(args);
+	write(stdout, sprint_buf, n);
+}
+
+int
+printf(char *fmt, ...)
+{
+	va_list args;
+	int n;
+
+	va_start(args, fmt);
+	n = vsprintf(sprint_buf, fmt, args);
+	va_end(args);
+	write(stdout, sprint_buf, n);
+	return n;
+}
diff --git a/arch/ppc/coffboot/string.S b/arch/ppc/coffboot/string.S
new file mode 100644
index 000000000..ba83591b7
--- /dev/null
+++ b/arch/ppc/coffboot/string.S
@@ -0,0 +1,206 @@
+/*
+ * String handling functions for PowerPC.
+ *
+ * Copyright (C) 1996 Paul Mackerras.
+ *
+ * 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.
+ */
+#define r0	0
+#define r3	3
+#define r4	4
+#define r5	5
+#define r6	6
+#define r7	7
+#define r8	8
+
+	.globl	strcpy
+strcpy:
+	addi	r5,r3,-1
+	addi	r4,r4,-1
+1:	lbzu	r0,1(r4)
+	cmpwi	0,r0,0
+	stbu	r0,1(r5)
+	bne	1b
+	blr
+
+	.globl	strncpy
+strncpy:
+	cmpwi	0,r5,0
+	beqlr
+	mtctr	r5
+	addi	r6,r3,-1
+	addi	r4,r4,-1
+1:	lbzu	r0,1(r4)
+	cmpwi	0,r0,0
+	stbu	r0,1(r6)
+	bdnzf	2,1b		/* dec ctr, branch if ctr != 0 && !cr0.eq */
+	blr
+
+	.globl	strcat
+strcat:
+	addi	r5,r3,-1
+	addi	r4,r4,-1
+1:	lbzu	r0,1(r5)
+	cmpwi	0,r0,0
+	bne	1b
+	addi	r5,r5,-1
+1:	lbzu	r0,1(r4)
+	cmpwi	0,r0,0
+	stbu	r0,1(r5)
+	bne	1b
+	blr
+
+	.globl	strcmp
+strcmp:
+	addi	r5,r3,-1
+	addi	r4,r4,-1
+1:	lbzu	r3,1(r5)
+	cmpwi	1,r3,0
+	lbzu	r0,1(r4)
+	subf.	r3,r0,r3
+	beqlr	1
+	beq	1b
+	blr
+
+	.globl	strlen
+strlen:
+	addi	r4,r3,-1
+1:	lbzu	r0,1(r4)
+	cmpwi	0,r0,0
+	bne	1b
+	subf	r3,r3,r4
+	blr
+
+	.globl	memset
+memset:
+	rlwimi	r4,r4,8,16,23
+	rlwimi	r4,r4,16,0,15
+	addi	r6,r3,-4
+	cmplwi	0,r5,4
+	blt	7f
+	stwu	r4,4(r6)
+	beqlr
+	andi.	r0,r6,3
+	add	r5,r0,r5
+	subf	r6,r0,r6
+	rlwinm	r0,r5,32-2,2,31
+	mtctr	r0
+	bdz	6f
+1:	stwu	r4,4(r6)
+	bdnz	1b
+6:	andi.	r5,r5,3
+7:	cmpwi	0,r5,0
+	beqlr
+	mtctr	r5
+	addi	r6,r6,3
+8:	stbu	r4,1(r6)
+	bdnz	8b
+	blr
+
+	.globl	bcopy
+bcopy:
+	mr	r6,r3
+	mr	r3,r4
+	mr	r4,r6
+	b	memcpy
+
+	.globl	memmove
+memmove:
+	cmplw	0,r3,r4
+	bgt	backwards_memcpy
+	/* fall through */
+
+	.globl	memcpy
+memcpy:
+	rlwinm.	r7,r5,32-3,3,31		/* r0 = r5 >> 3 */
+	addi	r6,r3,-4
+	addi	r4,r4,-4
+	beq	2f			/* if less than 8 bytes to do */
+	andi.	r0,r6,3			/* get dest word aligned */
+	mtctr	r7
+	bne	5f
+1:	lwz	r7,4(r4)
+	lwzu	r8,8(r4)
+	stw	r7,4(r6)
+	stwu	r8,8(r6)
+	bdnz	1b
+	andi.	r5,r5,7
+2:	cmplwi	0,r5,4
+	blt	3f
+	lwzu	r0,4(r4)
+	addi	r5,r5,-4
+	stwu	r0,4(r6)
+3:	cmpwi	0,r5,0
+	beqlr
+	mtctr	r5
+	addi	r4,r4,3
+	addi	r6,r6,3
+4:	lbzu	r0,1(r4)
+	stbu	r0,1(r6)
+	bdnz	4b
+	blr
+5:	subfic	r0,r0,4
+	mtctr	r0
+6:	lbz	r7,4(r4)
+	addi	r4,r4,1
+	stb	r7,4(r6)
+	addi	r6,r6,1
+	bdnz	6b
+	subf	r5,r0,r5
+	rlwinm.	r7,r5,32-3,3,31
+	beq	2b
+	mtctr	r7
+	b	1b
+
+	.globl	backwards_memcpy
+backwards_memcpy:
+	rlwinm.	r7,r5,32-3,3,31		/* r0 = r5 >> 3 */
+	add	r6,r3,r5
+	add	r4,r4,r5
+	beq	2f
+	andi.	r0,r6,3
+	mtctr	r7
+	bne	5f
+1:	lwz	r7,-4(r4)
+	lwzu	r8,-8(r4)
+	stw	r7,-4(r6)
+	stwu	r8,-8(r6)
+	bdnz	1b
+	andi.	r5,r5,7
+2:	cmplwi	0,r5,4
+	blt	3f
+	lwzu	r0,-4(r4)
+	subi	r5,r5,4
+	stwu	r0,-4(r6)
+3:	cmpwi	0,r5,0
+	beqlr
+	mtctr	r5
+4:	lbzu	r0,-1(r4)
+	stbu	r0,-1(r6)
+	bdnz	4b
+	blr
+5:	mtctr	r0
+6:	lbzu	r7,-1(r4)
+	stbu	r7,-1(r6)
+	bdnz	6b
+	subf	r5,r0,r5
+	rlwinm.	r7,r5,32-3,3,31
+	beq	2b
+	mtctr	r7
+	b	1b
+
+	.globl	memcmp
+memcmp:
+	cmpwi	0,r5,0
+	blelr
+	mtctr	r5
+	addi	r6,r3,-1
+	addi	r4,r4,-1
+1:	lbzu	r3,1(r6)
+	lbzu	r0,1(r4)
+	subf.	r3,r0,r3
+	bdnzt	2,1b
+	blr
diff --git a/arch/ppc/coffboot/zlib.c b/arch/ppc/coffboot/zlib.c
new file mode 100644
index 000000000..12d07df41
--- /dev/null
+++ b/arch/ppc/coffboot/zlib.c
@@ -0,0 +1,2143 @@
+/*
+ * This file is derived from various .h and .c files from the zlib-0.95
+ * distribution by Jean-loup Gailly and Mark Adler, with some additions
+ * by Paul Mackerras to aid in implementing Deflate compression and
+ * decompression for PPP packets.  See zlib.h for conditions of
+ * distribution and use.
+ *
+ * Changes that have been made include:
+ * - changed functions not used outside this file to "local"
+ * - added minCompression parameter to deflateInit2
+ * - added Z_PACKET_FLUSH (see zlib.h for details)
+ * - added inflateIncomp
+ *
+ * $Id: zlib.c,v 1.1 1997/08/30 04:51:48 ralf Exp $
+ */
+
+/*+++++*/
+/* zutil.h -- internal interface and configuration of the compression library
+ * Copyright (C) 1995 Jean-loup Gailly.
+ * For conditions of distribution and use, see copyright notice in zlib.h
+ */
+
+/* WARNING: this file should *not* be used by applications. It is
+   part of the implementation of the compression library and is
+   subject to change. Applications should only use zlib.h.
+ */
+
+/* From: zutil.h,v 1.9 1995/05/03 17:27:12 jloup Exp */
+
+#define _Z_UTIL_H
+
+#include "zlib.h"
+
+#ifndef local
+#  define local static
+#endif
+/* compile with -Dlocal if your debugger can't find static symbols */
+
+#define FAR
+
+typedef unsigned char  uch;
+typedef uch FAR uchf;
+typedef unsigned short ush;
+typedef ush FAR ushf;
+typedef unsigned long  ulg;
+
+extern char *z_errmsg[]; /* indexed by 1-zlib_error */
+
+#define ERR_RETURN(strm,err) return (strm->msg=z_errmsg[1-err], err)
+/* To be used only when the state is known to be valid */
+
+#ifndef NULL
+#define NULL	((void *) 0)
+#endif
+
+        /* common constants */
+
+#define DEFLATED   8
+
+#ifndef DEF_WBITS
+#  define DEF_WBITS MAX_WBITS
+#endif
+/* default windowBits for decompression. MAX_WBITS is for compression only */
+
+#if MAX_MEM_LEVEL >= 8
+#  define DEF_MEM_LEVEL 8
+#else
+#  define DEF_MEM_LEVEL  MAX_MEM_LEVEL
+#endif
+/* default memLevel */
+
+#define STORED_BLOCK 0
+#define STATIC_TREES 1
+#define DYN_TREES    2
+/* The three kinds of block type */
+
+#define MIN_MATCH  3
+#define MAX_MATCH  258
+/* The minimum and maximum match lengths */
+
+         /* functions */
+
+#include <string.h>
+#define zmemcpy memcpy
+#define zmemzero(dest, len)	memset(dest, 0, len)
+
+/* Diagnostic functions */
+#ifdef DEBUG_ZLIB
+#  include <stdio.h>
+#  ifndef verbose
+#    define verbose 0
+#  endif
+#  define Assert(cond,msg) {if(!(cond)) z_error(msg);}
+#  define Trace(x) fprintf x
+#  define Tracev(x) {if (verbose) fprintf x ;}
+#  define Tracevv(x) {if (verbose>1) fprintf x ;}
+#  define Tracec(c,x) {if (verbose && (c)) fprintf x ;}
+#  define Tracecv(c,x) {if (verbose>1 && (c)) fprintf x ;}
+#else
+#  define Assert(cond,msg)
+#  define Trace(x)
+#  define Tracev(x)
+#  define Tracevv(x)
+#  define Tracec(c,x)
+#  define Tracecv(c,x)
+#endif
+
+
+typedef uLong (*check_func) OF((uLong check, Bytef *buf, uInt len));
+
+/* voidpf zcalloc OF((voidpf opaque, unsigned items, unsigned size)); */
+/* void   zcfree  OF((voidpf opaque, voidpf ptr)); */
+
+#define ZALLOC(strm, items, size) \
+           (*((strm)->zalloc))((strm)->opaque, (items), (size))
+#define ZFREE(strm, addr, size)	\
+	   (*((strm)->zfree))((strm)->opaque, (voidpf)(addr), (size))
+#define TRY_FREE(s, p, n) {if (p) ZFREE(s, p, n);}
+
+/* deflate.h -- internal compression state
+ * Copyright (C) 1995 Jean-loup Gailly
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* WARNING: this file should *not* be used by applications. It is
+   part of the implementation of the compression library and is
+   subject to change. Applications should only use zlib.h.
+ */
+
+/*+++++*/
+/* infblock.h -- header to use infblock.c
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* WARNING: this file should *not* be used by applications. It is
+   part of the implementation of the compression library and is
+   subject to change. Applications should only use zlib.h.
+ */
+
+struct inflate_blocks_state;
+typedef struct inflate_blocks_state FAR inflate_blocks_statef;
+
+local inflate_blocks_statef * inflate_blocks_new OF((
+    z_stream *z,
+    check_func c,               /* check function */
+    uInt w));                   /* window size */
+
+local int inflate_blocks OF((
+    inflate_blocks_statef *,
+    z_stream *,
+    int));                      /* initial return code */
+
+local void inflate_blocks_reset OF((
+    inflate_blocks_statef *,
+    z_stream *,
+    uLongf *));                  /* check value on output */
+
+local int inflate_blocks_free OF((
+    inflate_blocks_statef *,
+    z_stream *,
+    uLongf *));                  /* check value on output */
+
+local int inflate_addhistory OF((
+    inflate_blocks_statef *,
+    z_stream *));
+
+local int inflate_packet_flush OF((
+    inflate_blocks_statef *));
+
+/*+++++*/
+/* inftrees.h -- header to use inftrees.c
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* WARNING: this file should *not* be used by applications. It is
+   part of the implementation of the compression library and is
+   subject to change. Applications should only use zlib.h.
+ */
+
+/* Huffman code lookup table entry--this entry is four bytes for machines
+   that have 16-bit pointers (e.g. PC's in the small or medium model). */
+
+typedef struct inflate_huft_s FAR inflate_huft;
+
+struct inflate_huft_s {
+  union {
+    struct {
+      Byte Exop;        /* number of extra bits or operation */
+      Byte Bits;        /* number of bits in this code or subcode */
+    } what;
+    uInt Nalloc;	/* number of these allocated here */
+    Bytef *pad;         /* pad structure to a power of 2 (4 bytes for */
+  } word;               /*  16-bit, 8 bytes for 32-bit machines) */
+  union {
+    uInt Base;          /* literal, length base, or distance base */
+    inflate_huft *Next; /* pointer to next level of table */
+  } more;
+};
+
+#ifdef DEBUG_ZLIB
+  local uInt inflate_hufts;
+#endif
+
+local int inflate_trees_bits OF((
+    uIntf *,                    /* 19 code lengths */
+    uIntf *,                    /* bits tree desired/actual depth */
+    inflate_huft * FAR *,       /* bits tree result */
+    z_stream *));               /* for zalloc, zfree functions */
+
+local int inflate_trees_dynamic OF((
+    uInt,                       /* number of literal/length codes */
+    uInt,                       /* number of distance codes */
+    uIntf *,                    /* that many (total) code lengths */
+    uIntf *,                    /* literal desired/actual bit depth */
+    uIntf *,                    /* distance desired/actual bit depth */
+    inflate_huft * FAR *,       /* literal/length tree result */
+    inflate_huft * FAR *,       /* distance tree result */
+    z_stream *));               /* for zalloc, zfree functions */
+
+local int inflate_trees_fixed OF((
+    uIntf *,                    /* literal desired/actual bit depth */
+    uIntf *,                    /* distance desired/actual bit depth */
+    inflate_huft * FAR *,       /* literal/length tree result */
+    inflate_huft * FAR *));     /* distance tree result */
+
+local int inflate_trees_free OF((
+    inflate_huft *,             /* tables to free */
+    z_stream *));               /* for zfree function */
+
+
+/*+++++*/
+/* infcodes.h -- header to use infcodes.c
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* WARNING: this file should *not* be used by applications. It is
+   part of the implementation of the compression library and is
+   subject to change. Applications should only use zlib.h.
+ */
+
+struct inflate_codes_state;
+typedef struct inflate_codes_state FAR inflate_codes_statef;
+
+local inflate_codes_statef *inflate_codes_new OF((
+    uInt, uInt,
+    inflate_huft *, inflate_huft *,
+    z_stream *));
+
+local int inflate_codes OF((
+    inflate_blocks_statef *,
+    z_stream *,
+    int));
+
+local void inflate_codes_free OF((
+    inflate_codes_statef *,
+    z_stream *));
+
+
+/*+++++*/
+/* inflate.c -- zlib interface to inflate modules
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* inflate private state */
+struct internal_state {
+
+  /* mode */
+  enum {
+      METHOD,   /* waiting for method byte */
+      FLAG,     /* waiting for flag byte */
+      BLOCKS,   /* decompressing blocks */
+      CHECK4,   /* four check bytes to go */
+      CHECK3,   /* three check bytes to go */
+      CHECK2,   /* two check bytes to go */
+      CHECK1,   /* one check byte to go */
+      DONE,     /* finished check, done */
+      BAD}      /* got an error--stay here */
+    mode;               /* current inflate mode */
+
+  /* mode dependent information */
+  union {
+    uInt method;        /* if FLAGS, method byte */
+    struct {
+      uLong was;                /* computed check value */
+      uLong need;               /* stream check value */
+    } check;            /* if CHECK, check values to compare */
+    uInt marker;        /* if BAD, inflateSync's marker bytes count */
+  } sub;        /* submode */
+
+  /* mode independent information */
+  int  nowrap;          /* flag for no wrapper */
+  uInt wbits;           /* log2(window size)  (8..15, defaults to 15) */
+  inflate_blocks_statef 
+    *blocks;            /* current inflate_blocks state */
+
+};
+
+
+int inflateReset(z)
+z_stream *z;
+{
+  uLong c;
+
+  if (z == Z_NULL || z->state == Z_NULL)
+    return Z_STREAM_ERROR;
+  z->total_in = z->total_out = 0;
+  z->msg = Z_NULL;
+  z->state->mode = z->state->nowrap ? BLOCKS : METHOD;
+  inflate_blocks_reset(z->state->blocks, z, &c);
+  Trace((stderr, "inflate: reset\n"));
+  return Z_OK;
+}
+
+
+int inflateEnd(z)
+z_stream *z;
+{
+  uLong c;
+
+  if (z == Z_NULL || z->state == Z_NULL || z->zfree == Z_NULL)
+    return Z_STREAM_ERROR;
+  if (z->state->blocks != Z_NULL)
+    inflate_blocks_free(z->state->blocks, z, &c);
+  ZFREE(z, z->state, sizeof(struct internal_state));
+  z->state = Z_NULL;
+  Trace((stderr, "inflate: end\n"));
+  return Z_OK;
+}
+
+
+int inflateInit2(z, w)
+z_stream *z;
+int w;
+{
+  /* initialize state */
+  if (z == Z_NULL)
+    return Z_STREAM_ERROR;
+/*  if (z->zalloc == Z_NULL) z->zalloc = zcalloc; */
+/*  if (z->zfree == Z_NULL) z->zfree = zcfree; */
+  if ((z->state = (struct internal_state FAR *)
+       ZALLOC(z,1,sizeof(struct internal_state))) == Z_NULL)
+    return Z_MEM_ERROR;
+  z->state->blocks = Z_NULL;
+
+  /* handle undocumented nowrap option (no zlib header or check) */
+  z->state->nowrap = 0;
+  if (w < 0)
+  {
+    w = - w;
+    z->state->nowrap = 1;
+  }
+
+  /* set window size */
+  if (w < 8 || w > 15)
+  {
+    inflateEnd(z);
+    return Z_STREAM_ERROR;
+  }
+  z->state->wbits = (uInt)w;
+
+  /* create inflate_blocks state */
+  if ((z->state->blocks =
+       inflate_blocks_new(z, z->state->nowrap ? Z_NULL : adler32, 1 << w))
+      == Z_NULL)
+  {
+    inflateEnd(z);
+    return Z_MEM_ERROR;
+  }
+  Trace((stderr, "inflate: allocated\n"));
+
+  /* reset state */
+  inflateReset(z);
+  return Z_OK;
+}
+
+
+int inflateInit(z)
+z_stream *z;
+{
+  return inflateInit2(z, DEF_WBITS);
+}
+
+
+#define NEEDBYTE {if(z->avail_in==0)goto empty;r=Z_OK;}
+#define NEXTBYTE (z->avail_in--,z->total_in++,*z->next_in++)
+
+int inflate(z, f)
+z_stream *z;
+int f;
+{
+  int r;
+  uInt b;
+
+  if (z == Z_NULL || z->next_in == Z_NULL)
+    return Z_STREAM_ERROR;
+  r = Z_BUF_ERROR;
+  while (1) switch (z->state->mode)
+  {
+    case METHOD:
+      NEEDBYTE
+      if (((z->state->sub.method = NEXTBYTE) & 0xf) != DEFLATED)
+      {
+        z->state->mode = BAD;
+        z->msg = "unknown compression method";
+        z->state->sub.marker = 5;       /* can't try inflateSync */
+        break;
+      }
+      if ((z->state->sub.method >> 4) + 8 > z->state->wbits)
+      {
+        z->state->mode = BAD;
+        z->msg = "invalid window size";
+        z->state->sub.marker = 5;       /* can't try inflateSync */
+        break;
+      }
+      z->state->mode = FLAG;
+    case FLAG:
+      NEEDBYTE
+      if ((b = NEXTBYTE) & 0x20)
+      {
+        z->state->mode = BAD;
+        z->msg = "invalid reserved bit";
+        z->state->sub.marker = 5;       /* can't try inflateSync */
+        break;
+      }
+      if (((z->state->sub.method << 8) + b) % 31)
+      {
+        z->state->mode = BAD;
+        z->msg = "incorrect header check";
+        z->state->sub.marker = 5;       /* can't try inflateSync */
+        break;
+      }
+      Trace((stderr, "inflate: zlib header ok\n"));
+      z->state->mode = BLOCKS;
+    case BLOCKS:
+      r = inflate_blocks(z->state->blocks, z, r);
+      if (f == Z_PACKET_FLUSH && z->avail_in == 0 && z->avail_out != 0)
+	  r = inflate_packet_flush(z->state->blocks);
+      if (r == Z_DATA_ERROR)
+      {
+        z->state->mode = BAD;
+        z->state->sub.marker = 0;       /* can try inflateSync */
+        break;
+      }
+      if (r != Z_STREAM_END)
+        return r;
+      r = Z_OK;
+      inflate_blocks_reset(z->state->blocks, z, &z->state->sub.check.was);
+      if (z->state->nowrap)
+      {
+        z->state->mode = DONE;
+        break;
+      }
+      z->state->mode = CHECK4;
+    case CHECK4:
+      NEEDBYTE
+      z->state->sub.check.need = (uLong)NEXTBYTE << 24;
+      z->state->mode = CHECK3;
+    case CHECK3:
+      NEEDBYTE
+      z->state->sub.check.need += (uLong)NEXTBYTE << 16;
+      z->state->mode = CHECK2;
+    case CHECK2:
+      NEEDBYTE
+      z->state->sub.check.need += (uLong)NEXTBYTE << 8;
+      z->state->mode = CHECK1;
+    case CHECK1:
+      NEEDBYTE
+      z->state->sub.check.need += (uLong)NEXTBYTE;
+
+      if (z->state->sub.check.was != z->state->sub.check.need)
+      {
+        z->state->mode = BAD;
+        z->msg = "incorrect data check";
+        z->state->sub.marker = 5;       /* can't try inflateSync */
+        break;
+      }
+      Trace((stderr, "inflate: zlib check ok\n"));
+      z->state->mode = DONE;
+    case DONE:
+      return Z_STREAM_END;
+    case BAD:
+      return Z_DATA_ERROR;
+    default:
+      return Z_STREAM_ERROR;
+  }
+
+ empty:
+  if (f != Z_PACKET_FLUSH)
+    return r;
+  z->state->mode = BAD;
+  z->state->sub.marker = 0;       /* can try inflateSync */
+  return Z_DATA_ERROR;
+}
+
+/*
+ * This subroutine adds the data at next_in/avail_in to the output history
+ * without performing any output.  The output buffer must be "caught up";
+ * i.e. no pending output (hence s->read equals s->write), and the state must
+ * be BLOCKS (i.e. we should be willing to see the start of a series of
+ * BLOCKS).  On exit, the output will also be caught up, and the checksum
+ * will have been updated if need be.
+ */
+
+int inflateIncomp(z)
+z_stream *z;
+{
+    if (z->state->mode != BLOCKS)
+	return Z_DATA_ERROR;
+    return inflate_addhistory(z->state->blocks, z);
+}
+
+
+int inflateSync(z)
+z_stream *z;
+{
+  uInt n;       /* number of bytes to look at */
+  Bytef *p;     /* pointer to bytes */
+  uInt m;       /* number of marker bytes found in a row */
+  uLong r, w;   /* temporaries to save total_in and total_out */
+
+  /* set up */
+  if (z == Z_NULL || z->state == Z_NULL)
+    return Z_STREAM_ERROR;
+  if (z->state->mode != BAD)
+  {
+    z->state->mode = BAD;
+    z->state->sub.marker = 0;
+  }
+  if ((n = z->avail_in) == 0)
+    return Z_BUF_ERROR;
+  p = z->next_in;
+  m = z->state->sub.marker;
+
+  /* search */
+  while (n && m < 4)
+  {
+    if (*p == (Byte)(m < 2 ? 0 : 0xff))
+      m++;
+    else if (*p)
+      m = 0;
+    else
+      m = 4 - m;
+    p++, n--;
+  }
+
+  /* restore */
+  z->total_in += p - z->next_in;
+  z->next_in = p;
+  z->avail_in = n;
+  z->state->sub.marker = m;
+
+  /* return no joy or set up to restart on a new block */
+  if (m != 4)
+    return Z_DATA_ERROR;
+  r = z->total_in;  w = z->total_out;
+  inflateReset(z);
+  z->total_in = r;  z->total_out = w;
+  z->state->mode = BLOCKS;
+  return Z_OK;
+}
+
+#undef NEEDBYTE
+#undef NEXTBYTE
+
+/*+++++*/
+/* infutil.h -- types and macros common to blocks and codes
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* WARNING: this file should *not* be used by applications. It is
+   part of the implementation of the compression library and is
+   subject to change. Applications should only use zlib.h.
+ */
+
+/* inflate blocks semi-private state */
+struct inflate_blocks_state {
+
+  /* mode */
+  enum {
+      TYPE,     /* get type bits (3, including end bit) */
+      LENS,     /* get lengths for stored */
+      STORED,   /* processing stored block */
+      TABLE,    /* get table lengths */
+      BTREE,    /* get bit lengths tree for a dynamic block */
+      DTREE,    /* get length, distance trees for a dynamic block */
+      CODES,    /* processing fixed or dynamic block */
+      DRY,      /* output remaining window bytes */
+      DONEB,     /* finished last block, done */
+      BADB}      /* got a data error--stuck here */
+    mode;               /* current inflate_block mode */
+
+  /* mode dependent information */
+  union {
+    uInt left;          /* if STORED, bytes left to copy */
+    struct {
+      uInt table;               /* table lengths (14 bits) */
+      uInt index;               /* index into blens (or border) */
+      uIntf *blens;             /* bit lengths of codes */
+      uInt bb;                  /* bit length tree depth */
+      inflate_huft *tb;         /* bit length decoding tree */
+      int nblens;		/* # elements allocated at blens */
+    } trees;            /* if DTREE, decoding info for trees */
+    struct {
+      inflate_huft *tl, *td;    /* trees to free */
+      inflate_codes_statef 
+         *codes;
+    } decode;           /* if CODES, current state */
+  } sub;                /* submode */
+  uInt last;            /* true if this block is the last block */
+
+  /* mode independent information */
+  uInt bitk;            /* bits in bit buffer */
+  uLong bitb;           /* bit buffer */
+  Bytef *window;        /* sliding window */
+  Bytef *end;           /* one byte after sliding window */
+  Bytef *read;          /* window read pointer */
+  Bytef *write;         /* window write pointer */
+  check_func checkfn;   /* check function */
+  uLong check;          /* check on output */
+
+};
+
+
+/* defines for inflate input/output */
+/*   update pointers and return */
+#define UPDBITS {s->bitb=b;s->bitk=k;}
+#define UPDIN {z->avail_in=n;z->total_in+=p-z->next_in;z->next_in=p;}
+#define UPDOUT {s->write=q;}
+#define UPDATE {UPDBITS UPDIN UPDOUT}
+#define LEAVE {UPDATE return inflate_flush(s,z,r);}
+/*   get bytes and bits */
+#define LOADIN {p=z->next_in;n=z->avail_in;b=s->bitb;k=s->bitk;}
+#define NEEDBYTE {if(n)r=Z_OK;else LEAVE}
+#define NEXTBYTE (n--,*p++)
+#define NEEDBITS(j) {while(k<(j)){NEEDBYTE;b|=((uLong)NEXTBYTE)<<k;k+=8;}}
+#define DUMPBITS(j) {b>>=(j);k-=(j);}
+/*   output bytes */
+#define WAVAIL (q<s->read?s->read-q-1:s->end-q)
+#define LOADOUT {q=s->write;m=WAVAIL;}
+#define WRAP {if(q==s->end&&s->read!=s->window){q=s->window;m=WAVAIL;}}
+#define FLUSH {UPDOUT r=inflate_flush(s,z,r); LOADOUT}
+#define NEEDOUT {if(m==0){WRAP if(m==0){FLUSH WRAP if(m==0) LEAVE}}r=Z_OK;}
+#define OUTBYTE(a) {*q++=(Byte)(a);m--;}
+/*   load local pointers */
+#define LOAD {LOADIN LOADOUT}
+
+/* And'ing with mask[n] masks the lower n bits */
+local uInt inflate_mask[] = {
+    0x0000,
+    0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff,
+    0x01ff, 0x03ff, 0x07ff, 0x0fff, 0x1fff, 0x3fff, 0x7fff, 0xffff
+};
+
+/* copy as much as possible from the sliding window to the output area */
+local int inflate_flush OF((
+    inflate_blocks_statef *,
+    z_stream *,
+    int));
+
+/*+++++*/
+/* inffast.h -- header to use inffast.c
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* WARNING: this file should *not* be used by applications. It is
+   part of the implementation of the compression library and is
+   subject to change. Applications should only use zlib.h.
+ */
+
+local int inflate_fast OF((
+    uInt,
+    uInt,
+    inflate_huft *,
+    inflate_huft *,
+    inflate_blocks_statef *,
+    z_stream *));
+
+
+/*+++++*/
+/* infblock.c -- interpret and process block types to last block
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* Table for deflate from PKZIP's appnote.txt. */
+local uInt border[] = { /* Order of the bit length code lengths */
+        16, 17, 18, 0, 8, 7, 9, 6, 10, 5, 11, 4, 12, 3, 13, 2, 14, 1, 15};
+
+/*
+   Notes beyond the 1.93a appnote.txt:
+
+   1. Distance pointers never point before the beginning of the output
+      stream.
+   2. Distance pointers can point back across blocks, up to 32k away.
+   3. There is an implied maximum of 7 bits for the bit length table and
+      15 bits for the actual data.
+   4. If only one code exists, then it is encoded using one bit.  (Zero
+      would be more efficient, but perhaps a little confusing.)  If two
+      codes exist, they are coded using one bit each (0 and 1).
+   5. There is no way of sending zero distance codes--a dummy must be
+      sent if there are none.  (History: a pre 2.0 version of PKZIP would
+      store blocks with no distance codes, but this was discovered to be
+      too harsh a criterion.)  Valid only for 1.93a.  2.04c does allow
+      zero distance codes, which is sent as one code of zero bits in
+      length.
+   6. There are up to 286 literal/length codes.  Code 256 represents the
+      end-of-block.  Note however that the static length tree defines
+      288 codes just to fill out the Huffman codes.  Codes 286 and 287
+      cannot be used though, since there is no length base or extra bits
+      defined for them.  Similarily, there are up to 30 distance codes.
+      However, static trees define 32 codes (all 5 bits) to fill out the
+      Huffman codes, but the last two had better not show up in the data.
+   7. Unzip can check dynamic Huffman blocks for complete code sets.
+      The exception is that a single code would not be complete (see #4).
+   8. The five bits following the block type is really the number of
+      literal codes sent minus 257.
+   9. Length codes 8,16,16 are interpreted as 13 length codes of 8 bits
+      (1+6+6).  Therefore, to output three times the length, you output
+      three codes (1+1+1), whereas to output four times the same length,
+      you only need two codes (1+3).  Hmm.
+  10. In the tree reconstruction algorithm, Code = Code + Increment
+      only if BitLength(i) is not zero.  (Pretty obvious.)
+  11. Correction: 4 Bits: # of Bit Length codes - 4     (4 - 19)
+  12. Note: length code 284 can represent 227-258, but length code 285
+      really is 258.  The last length deserves its own, short code
+      since it gets used a lot in very redundant files.  The length
+      258 is special since 258 - 3 (the min match length) is 255.
+  13. The literal/length and distance code bit lengths are read as a
+      single stream of lengths.  It is possible (and advantageous) for
+      a repeat code (16, 17, or 18) to go across the boundary between
+      the two sets of lengths.
+ */
+
+
+local void inflate_blocks_reset(s, z, c)
+inflate_blocks_statef *s;
+z_stream *z;
+uLongf *c;
+{
+  if (s->checkfn != Z_NULL)
+    *c = s->check;
+  if (s->mode == BTREE || s->mode == DTREE)
+    ZFREE(z, s->sub.trees.blens, s->sub.trees.nblens * sizeof(uInt));
+  if (s->mode == CODES)
+  {
+    inflate_codes_free(s->sub.decode.codes, z);
+    inflate_trees_free(s->sub.decode.td, z);
+    inflate_trees_free(s->sub.decode.tl, z);
+  }
+  s->mode = TYPE;
+  s->bitk = 0;
+  s->bitb = 0;
+  s->read = s->write = s->window;
+  if (s->checkfn != Z_NULL)
+    s->check = (*s->checkfn)(0L, Z_NULL, 0);
+  Trace((stderr, "inflate:   blocks reset\n"));
+}
+
+
+local inflate_blocks_statef *inflate_blocks_new(z, c, w)
+z_stream *z;
+check_func c;
+uInt w;
+{
+  inflate_blocks_statef *s;
+
+  if ((s = (inflate_blocks_statef *)ZALLOC
+       (z,1,sizeof(struct inflate_blocks_state))) == Z_NULL)
+    return s;
+  if ((s->window = (Bytef *)ZALLOC(z, 1, w)) == Z_NULL)
+  {
+    ZFREE(z, s, sizeof(struct inflate_blocks_state));
+    return Z_NULL;
+  }
+  s->end = s->window + w;
+  s->checkfn = c;
+  s->mode = TYPE;
+  Trace((stderr, "inflate:   blocks allocated\n"));
+  inflate_blocks_reset(s, z, &s->check);
+  return s;
+}
+
+
+local int inflate_blocks(s, z, r)
+inflate_blocks_statef *s;
+z_stream *z;
+int r;
+{
+  uInt t;               /* temporary storage */
+  uLong b;              /* bit buffer */
+  uInt k;               /* bits in bit buffer */
+  Bytef *p;             /* input data pointer */
+  uInt n;               /* bytes available there */
+  Bytef *q;             /* output window write pointer */
+  uInt m;               /* bytes to end of window or read pointer */
+
+  /* copy input/output information to locals (UPDATE macro restores) */
+  LOAD
+
+  /* process input based on current state */
+  while (1) switch (s->mode)
+  {
+    case TYPE:
+      NEEDBITS(3)
+      t = (uInt)b & 7;
+      s->last = t & 1;
+      switch (t >> 1)
+      {
+        case 0:                         /* stored */
+          Trace((stderr, "inflate:     stored block%s\n",
+                 s->last ? " (last)" : ""));
+          DUMPBITS(3)
+          t = k & 7;                    /* go to byte boundary */
+          DUMPBITS(t)
+          s->mode = LENS;               /* get length of stored block */
+          break;
+        case 1:                         /* fixed */
+          Trace((stderr, "inflate:     fixed codes block%s\n",
+                 s->last ? " (last)" : ""));
+          {
+            uInt bl, bd;
+            inflate_huft *tl, *td;
+
+            inflate_trees_fixed(&bl, &bd, &tl, &td);
+            s->sub.decode.codes = inflate_codes_new(bl, bd, tl, td, z);
+            if (s->sub.decode.codes == Z_NULL)
+            {
+              r = Z_MEM_ERROR;
+              LEAVE
+            }
+            s->sub.decode.tl = Z_NULL;  /* don't try to free these */
+            s->sub.decode.td = Z_NULL;
+          }
+          DUMPBITS(3)
+          s->mode = CODES;
+          break;
+        case 2:                         /* dynamic */
+          Trace((stderr, "inflate:     dynamic codes block%s\n",
+                 s->last ? " (last)" : ""));
+          DUMPBITS(3)
+          s->mode = TABLE;
+          break;
+        case 3:                         /* illegal */
+          DUMPBITS(3)
+          s->mode = BADB;
+          z->msg = "invalid block type";
+          r = Z_DATA_ERROR;
+          LEAVE
+      }
+      break;
+    case LENS:
+      NEEDBITS(32)
+      if (((~b) >> 16) != (b & 0xffff))
+      {
+        s->mode = BADB;
+        z->msg = "invalid stored block lengths";
+        r = Z_DATA_ERROR;
+        LEAVE
+      }
+      s->sub.left = (uInt)b & 0xffff;
+      b = k = 0;                      /* dump bits */
+      Tracev((stderr, "inflate:       stored length %u\n", s->sub.left));
+      s->mode = s->sub.left ? STORED : TYPE;
+      break;
+    case STORED:
+      if (n == 0)
+        LEAVE
+      NEEDOUT
+      t = s->sub.left;
+      if (t > n) t = n;
+      if (t > m) t = m;
+      zmemcpy(q, p, t);
+      p += t;  n -= t;
+      q += t;  m -= t;
+      if ((s->sub.left -= t) != 0)
+        break;
+      Tracev((stderr, "inflate:       stored end, %lu total out\n",
+              z->total_out + (q >= s->read ? q - s->read :
+              (s->end - s->read) + (q - s->window))));
+      s->mode = s->last ? DRY : TYPE;
+      break;
+    case TABLE:
+      NEEDBITS(14)
+      s->sub.trees.table = t = (uInt)b & 0x3fff;
+#ifndef PKZIP_BUG_WORKAROUND
+      if ((t & 0x1f) > 29 || ((t >> 5) & 0x1f) > 29)
+      {
+        s->mode = BADB;
+        z->msg = "too many length or distance symbols";
+        r = Z_DATA_ERROR;
+        LEAVE
+      }
+#endif
+      t = 258 + (t & 0x1f) + ((t >> 5) & 0x1f);
+      if (t < 19)
+        t = 19;
+      if ((s->sub.trees.blens = (uIntf*)ZALLOC(z, t, sizeof(uInt))) == Z_NULL)
+      {
+        r = Z_MEM_ERROR;
+        LEAVE
+      }
+      s->sub.trees.nblens = t;
+      DUMPBITS(14)
+      s->sub.trees.index = 0;
+      Tracev((stderr, "inflate:       table sizes ok\n"));
+      s->mode = BTREE;
+    case BTREE:
+      while (s->sub.trees.index < 4 + (s->sub.trees.table >> 10))
+      {
+        NEEDBITS(3)
+        s->sub.trees.blens[border[s->sub.trees.index++]] = (uInt)b & 7;
+        DUMPBITS(3)
+      }
+      while (s->sub.trees.index < 19)
+        s->sub.trees.blens[border[s->sub.trees.index++]] = 0;
+      s->sub.trees.bb = 7;
+      t = inflate_trees_bits(s->sub.trees.blens, &s->sub.trees.bb,
+                             &s->sub.trees.tb, z);
+      if (t != Z_OK)
+      {
+        r = t;
+        if (r == Z_DATA_ERROR)
+          s->mode = BADB;
+        LEAVE
+      }
+      s->sub.trees.index = 0;
+      Tracev((stderr, "inflate:       bits tree ok\n"));
+      s->mode = DTREE;
+    case DTREE:
+      while (t = s->sub.trees.table,
+             s->sub.trees.index < 258 + (t & 0x1f) + ((t >> 5) & 0x1f))
+      {
+        inflate_huft *h;
+        uInt i, j, c;
+
+        t = s->sub.trees.bb;
+        NEEDBITS(t)
+        h = s->sub.trees.tb + ((uInt)b & inflate_mask[t]);
+        t = h->word.what.Bits;
+        c = h->more.Base;
+        if (c < 16)
+        {
+          DUMPBITS(t)
+          s->sub.trees.blens[s->sub.trees.index++] = c;
+        }
+        else /* c == 16..18 */
+        {
+          i = c == 18 ? 7 : c - 14;
+          j = c == 18 ? 11 : 3;
+          NEEDBITS(t + i)
+          DUMPBITS(t)
+          j += (uInt)b & inflate_mask[i];
+          DUMPBITS(i)
+          i = s->sub.trees.index;
+          t = s->sub.trees.table;
+          if (i + j > 258 + (t & 0x1f) + ((t >> 5) & 0x1f) ||
+              (c == 16 && i < 1))
+          {
+            s->mode = BADB;
+            z->msg = "invalid bit length repeat";
+            r = Z_DATA_ERROR;
+            LEAVE
+          }
+          c = c == 16 ? s->sub.trees.blens[i - 1] : 0;
+          do {
+            s->sub.trees.blens[i++] = c;
+          } while (--j);
+          s->sub.trees.index = i;
+        }
+      }
+      inflate_trees_free(s->sub.trees.tb, z);
+      s->sub.trees.tb = Z_NULL;
+      {
+        uInt bl, bd;
+        inflate_huft *tl, *td;
+        inflate_codes_statef *c;
+
+        bl = 9;         /* must be <= 9 for lookahead assumptions */
+        bd = 6;         /* must be <= 9 for lookahead assumptions */
+        t = s->sub.trees.table;
+        t = inflate_trees_dynamic(257 + (t & 0x1f), 1 + ((t >> 5) & 0x1f),
+                                  s->sub.trees.blens, &bl, &bd, &tl, &td, z);
+        if (t != Z_OK)
+        {
+          if (t == (uInt)Z_DATA_ERROR)
+            s->mode = BADB;
+          r = t;
+          LEAVE
+        }
+        Tracev((stderr, "inflate:       trees ok\n"));
+        if ((c = inflate_codes_new(bl, bd, tl, td, z)) == Z_NULL)
+        {
+          inflate_trees_free(td, z);
+          inflate_trees_free(tl, z);
+          r = Z_MEM_ERROR;
+          LEAVE
+        }
+        ZFREE(z, s->sub.trees.blens, s->sub.trees.nblens * sizeof(uInt));
+        s->sub.decode.codes = c;
+        s->sub.decode.tl = tl;
+        s->sub.decode.td = td;
+      }
+      s->mode = CODES;
+    case CODES:
+      UPDATE
+      if ((r = inflate_codes(s, z, r)) != Z_STREAM_END)
+        return inflate_flush(s, z, r);
+      r = Z_OK;
+      inflate_codes_free(s->sub.decode.codes, z);
+      inflate_trees_free(s->sub.decode.td, z);
+      inflate_trees_free(s->sub.decode.tl, z);
+      LOAD
+      Tracev((stderr, "inflate:       codes end, %lu total out\n",
+              z->total_out + (q >= s->read ? q - s->read :
+              (s->end - s->read) + (q - s->window))));
+      if (!s->last)
+      {
+        s->mode = TYPE;
+        break;
+      }
+      if (k > 7)              /* return unused byte, if any */
+      {
+        Assert(k < 16, "inflate_codes grabbed too many bytes")
+        k -= 8;
+        n++;
+        p--;                    /* can always return one */
+      }
+      s->mode = DRY;
+    case DRY:
+      FLUSH
+      if (s->read != s->write)
+        LEAVE
+      s->mode = DONEB;
+    case DONEB:
+      r = Z_STREAM_END;
+      LEAVE
+    case BADB:
+      r = Z_DATA_ERROR;
+      LEAVE
+    default:
+      r = Z_STREAM_ERROR;
+      LEAVE
+  }
+}
+
+
+local int inflate_blocks_free(s, z, c)
+inflate_blocks_statef *s;
+z_stream *z;
+uLongf *c;
+{
+  inflate_blocks_reset(s, z, c);
+  ZFREE(z, s->window, s->end - s->window);
+  ZFREE(z, s, sizeof(struct inflate_blocks_state));
+  Trace((stderr, "inflate:   blocks freed\n"));
+  return Z_OK;
+}
+
+/*
+ * This subroutine adds the data at next_in/avail_in to the output history
+ * without performing any output.  The output buffer must be "caught up";
+ * i.e. no pending output (hence s->read equals s->write), and the state must
+ * be BLOCKS (i.e. we should be willing to see the start of a series of
+ * BLOCKS).  On exit, the output will also be caught up, and the checksum
+ * will have been updated if need be.
+ */
+local int inflate_addhistory(s, z)
+inflate_blocks_statef *s;
+z_stream *z;
+{
+    uLong b;              /* bit buffer */  /* NOT USED HERE */
+    uInt k;               /* bits in bit buffer */ /* NOT USED HERE */
+    uInt t;               /* temporary storage */
+    Bytef *p;             /* input data pointer */
+    uInt n;               /* bytes available there */
+    Bytef *q;             /* output window write pointer */
+    uInt m;               /* bytes to end of window or read pointer */
+
+    if (s->read != s->write)
+	return Z_STREAM_ERROR;
+    if (s->mode != TYPE)
+	return Z_DATA_ERROR;
+
+    /* we're ready to rock */
+    LOAD
+    /* while there is input ready, copy to output buffer, moving
+     * pointers as needed.
+     */
+    while (n) {
+	t = n;  /* how many to do */
+	/* is there room until end of buffer? */
+	if (t > m) t = m;
+	/* update check information */
+	if (s->checkfn != Z_NULL)
+	    s->check = (*s->checkfn)(s->check, q, t);
+	zmemcpy(q, p, t);
+	q += t;
+	p += t;
+	n -= t;
+	z->total_out += t;
+	s->read = q;    /* drag read pointer forward */
+/*      WRAP  */ 	/* expand WRAP macro by hand to handle s->read */
+	if (q == s->end) {
+	    s->read = q = s->window;
+	    m = WAVAIL;
+	}
+    }
+    UPDATE
+    return Z_OK;
+}
+
+
+/*
+ * At the end of a Deflate-compressed PPP packet, we expect to have seen
+ * a `stored' block type value but not the (zero) length bytes.
+ */
+local int inflate_packet_flush(s)
+    inflate_blocks_statef *s;
+{
+    if (s->mode != LENS)
+	return Z_DATA_ERROR;
+    s->mode = TYPE;
+    return Z_OK;
+}
+
+
+/*+++++*/
+/* inftrees.c -- generate Huffman trees for efficient decoding
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* simplify the use of the inflate_huft type with some defines */
+#define base more.Base
+#define next more.Next
+#define exop word.what.Exop
+#define bits word.what.Bits
+
+
+local int huft_build OF((
+    uIntf *,            /* code lengths in bits */
+    uInt,               /* number of codes */
+    uInt,               /* number of "simple" codes */
+    uIntf *,            /* list of base values for non-simple codes */
+    uIntf *,            /* list of extra bits for non-simple codes */
+    inflate_huft * FAR*,/* result: starting table */
+    uIntf *,            /* maximum lookup bits (returns actual) */
+    z_stream *));       /* for zalloc function */
+
+local voidpf falloc OF((
+    voidpf,             /* opaque pointer (not used) */
+    uInt,               /* number of items */
+    uInt));             /* size of item */
+
+local void ffree OF((
+    voidpf q,           /* opaque pointer (not used) */
+    voidpf p,           /* what to free (not used) */
+    uInt n));		/* number of bytes (not used) */
+
+/* Tables for deflate from PKZIP's appnote.txt. */
+local uInt cplens[] = { /* Copy lengths for literal codes 257..285 */
+        3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31,
+        35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0};
+        /* actually lengths - 2; also see note #13 above about 258 */
+local uInt cplext[] = { /* Extra bits for literal codes 257..285 */
+        0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2,
+        3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0, 192, 192}; /* 192==invalid */
+local uInt cpdist[] = { /* Copy offsets for distance codes 0..29 */
+        1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193,
+        257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145,
+        8193, 12289, 16385, 24577};
+local uInt cpdext[] = { /* Extra bits for distance codes */
+        0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6,
+        7, 7, 8, 8, 9, 9, 10, 10, 11, 11,
+        12, 12, 13, 13};
+
+/*
+   Huffman code decoding is performed using a multi-level table lookup.
+   The fastest way to decode is to simply build a lookup table whose
+   size is determined by the longest code.  However, the time it takes
+   to build this table can also be a factor if the data being decoded
+   is not very long.  The most common codes are necessarily the
+   shortest codes, so those codes dominate the decoding time, and hence
+   the speed.  The idea is you can have a shorter table that decodes the
+   shorter, more probable codes, and then point to subsidiary tables for
+   the longer codes.  The time it costs to decode the longer codes is
+   then traded against the time it takes to make longer tables.
+
+   This results of this trade are in the variables lbits and dbits
+   below.  lbits is the number of bits the first level table for literal/
+   length codes can decode in one step, and dbits is the same thing for
+   the distance codes.  Subsequent tables are also less than or equal to
+   those sizes.  These values may be adjusted either when all of the
+   codes are shorter than that, in which case the longest code length in
+   bits is used, or when the shortest code is *longer* than the requested
+   table size, in which case the length of the shortest code in bits is
+   used.
+
+   There are two different values for the two tables, since they code a
+   different number of possibilities each.  The literal/length table
+   codes 286 possible values, or in a flat code, a little over eight
+   bits.  The distance table codes 30 possible values, or a little less
+   than five bits, flat.  The optimum values for speed end up being
+   about one bit more than those, so lbits is 8+1 and dbits is 5+1.
+   The optimum values may differ though from machine to machine, and
+   possibly even between compilers.  Your mileage may vary.
+ */
+
+
+/* If BMAX needs to be larger than 16, then h and x[] should be uLong. */
+#define BMAX 15         /* maximum bit length of any code */
+#define N_MAX 288       /* maximum number of codes in any set */
+
+#ifdef DEBUG_ZLIB
+  uInt inflate_hufts;
+#endif
+
+local int huft_build(b, n, s, d, e, t, m, zs)
+uIntf *b;               /* code lengths in bits (all assumed <= BMAX) */
+uInt n;                 /* number of codes (assumed <= N_MAX) */
+uInt s;                 /* number of simple-valued codes (0..s-1) */
+uIntf *d;               /* list of base values for non-simple codes */
+uIntf *e;               /* list of extra bits for non-simple codes */  
+inflate_huft * FAR *t;  /* result: starting table */
+uIntf *m;               /* maximum lookup bits, returns actual */
+z_stream *zs;           /* for zalloc function */
+/* Given a list of code lengths and a maximum table size, make a set of
+   tables to decode that set of codes.  Return Z_OK on success, Z_BUF_ERROR
+   if the given code set is incomplete (the tables are still built in this
+   case), Z_DATA_ERROR if the input is invalid (all zero length codes or an
+   over-subscribed set of lengths), or Z_MEM_ERROR if not enough memory. */
+{
+
+  uInt a;                       /* counter for codes of length k */
+  uInt c[BMAX+1];               /* bit length count table */
+  uInt f;                       /* i repeats in table every f entries */
+  int g;                        /* maximum code length */
+  int h;                        /* table level */
+  register uInt i;              /* counter, current code */
+  register uInt j;              /* counter */
+  register int k;               /* number of bits in current code */
+  int l;                        /* bits per table (returned in m) */
+  register uIntf *p;            /* pointer into c[], b[], or v[] */
+  inflate_huft *q;              /* points to current table */
+  struct inflate_huft_s r;      /* table entry for structure assignment */
+  inflate_huft *u[BMAX];        /* table stack */
+  uInt v[N_MAX];                /* values in order of bit length */
+  register int w;               /* bits before this table == (l * h) */
+  uInt x[BMAX+1];               /* bit offsets, then code stack */
+  uIntf *xp;                    /* pointer into x */
+  int y;                        /* number of dummy codes added */
+  uInt z;                       /* number of entries in current table */
+
+
+  /* Generate counts for each bit length */
+  p = c;
+#define C0 *p++ = 0;
+#define C2 C0 C0 C0 C0
+#define C4 C2 C2 C2 C2
+  C4                            /* clear c[]--assume BMAX+1 is 16 */
+  p = b;  i = n;
+  do {
+    c[*p++]++;                  /* assume all entries <= BMAX */
+  } while (--i);
+  if (c[0] == n)                /* null input--all zero length codes */
+  {
+    *t = (inflate_huft *)Z_NULL;
+    *m = 0;
+    return Z_OK;
+  }
+
+
+  /* Find minimum and maximum length, bound *m by those */
+  l = *m;
+  for (j = 1; j <= BMAX; j++)
+    if (c[j])
+      break;
+  k = j;                        /* minimum code length */
+  if ((uInt)l < j)
+    l = j;
+  for (i = BMAX; i; i--)
+    if (c[i])
+      break;
+  g = i;                        /* maximum code length */
+  if ((uInt)l > i)
+    l = i;
+  *m = l;
+
+
+  /* Adjust last length count to fill out codes, if needed */
+  for (y = 1 << j; j < i; j++, y <<= 1)
+    if ((y -= c[j]) < 0)
+      return Z_DATA_ERROR;
+  if ((y -= c[i]) < 0)
+    return Z_DATA_ERROR;
+  c[i] += y;
+
+
+  /* Generate starting offsets into the value table for each length */
+  x[1] = j = 0;
+  p = c + 1;  xp = x + 2;
+  while (--i) {                 /* note that i == g from above */
+    *xp++ = (j += *p++);
+  }
+
+
+  /* Make a table of values in order of bit lengths */
+  p = b;  i = 0;
+  do {
+    if ((j = *p++) != 0)
+      v[x[j]++] = i;
+  } while (++i < n);
+
+
+  /* Generate the Huffman codes and for each, make the table entries */
+  x[0] = i = 0;                 /* first Huffman code is zero */
+  p = v;                        /* grab values in bit order */
+  h = -1;                       /* no tables yet--level -1 */
+  w = -l;                       /* bits decoded == (l * h) */
+  u[0] = (inflate_huft *)Z_NULL;        /* just to keep compilers happy */
+  q = (inflate_huft *)Z_NULL;   /* ditto */
+  z = 0;                        /* ditto */
+
+  /* go through the bit lengths (k already is bits in shortest code) */
+  for (; k <= g; k++)
+  {
+    a = c[k];
+    while (a--)
+    {
+      /* here i is the Huffman code of length k bits for value *p */
+      /* make tables up to required level */
+      while (k > w + l)
+      {
+        h++;
+        w += l;                 /* previous table always l bits */
+
+        /* compute minimum size table less than or equal to l bits */
+        z = (z = g - w) > (uInt)l ? l : z;      /* table size upper limit */
+        if ((f = 1 << (j = k - w)) > a + 1)     /* try a k-w bit table */
+        {                       /* too few codes for k-w bit table */
+          f -= a + 1;           /* deduct codes from patterns left */
+          xp = c + k;
+          if (j < z)
+            while (++j < z)     /* try smaller tables up to z bits */
+            {
+              if ((f <<= 1) <= *++xp)
+                break;          /* enough codes to use up j bits */
+              f -= *xp;         /* else deduct codes from patterns */
+            }
+        }
+        z = 1 << j;             /* table entries for j-bit table */
+
+        /* allocate and link in new table */
+        if ((q = (inflate_huft *)ZALLOC
+             (zs,z + 1,sizeof(inflate_huft))) == Z_NULL)
+        {
+          if (h)
+            inflate_trees_free(u[0], zs);
+          return Z_MEM_ERROR;   /* not enough memory */
+        }
+	q->word.Nalloc = z + 1;
+#ifdef DEBUG_ZLIB
+        inflate_hufts += z + 1;
+#endif
+        *t = q + 1;             /* link to list for huft_free() */
+        *(t = &(q->next)) = Z_NULL;
+        u[h] = ++q;             /* table starts after link */
+
+        /* connect to last table, if there is one */
+        if (h)
+        {
+          x[h] = i;             /* save pattern for backing up */
+          r.bits = (Byte)l;     /* bits to dump before this table */
+          r.exop = (Byte)j;     /* bits in this table */
+          r.next = q;           /* pointer to this table */
+          j = i >> (w - l);     /* (get around Turbo C bug) */
+          u[h-1][j] = r;        /* connect to last table */
+        }
+      }
+
+      /* set up table entry in r */
+      r.bits = (Byte)(k - w);
+      if (p >= v + n)
+        r.exop = 128 + 64;      /* out of values--invalid code */
+      else if (*p < s)
+      {
+        r.exop = (Byte)(*p < 256 ? 0 : 32 + 64);     /* 256 is end-of-block */
+        r.base = *p++;          /* simple code is just the value */
+      }
+      else
+      {
+        r.exop = (Byte)e[*p - s] + 16 + 64; /* non-simple--look up in lists */
+        r.base = d[*p++ - s];
+      }
+
+      /* fill code-like entries with r */
+      f = 1 << (k - w);
+      for (j = i >> w; j < z; j += f)
+        q[j] = r;
+
+      /* backwards increment the k-bit code i */
+      for (j = 1 << (k - 1); i & j; j >>= 1)
+        i ^= j;
+      i ^= j;
+
+      /* backup over finished tables */
+      while ((i & ((1 << w) - 1)) != x[h])
+      {
+        h--;                    /* don't need to update q */
+        w -= l;
+      }
+    }
+  }
+
+
+  /* Return Z_BUF_ERROR if we were given an incomplete table */
+  return y != 0 && g != 1 ? Z_BUF_ERROR : Z_OK;
+}
+
+
+local int inflate_trees_bits(c, bb, tb, z)
+uIntf *c;               /* 19 code lengths */
+uIntf *bb;              /* bits tree desired/actual depth */
+inflate_huft * FAR *tb; /* bits tree result */
+z_stream *z;            /* for zfree function */
+{
+  int r;
+
+  r = huft_build(c, 19, 19, (uIntf*)Z_NULL, (uIntf*)Z_NULL, tb, bb, z);
+  if (r == Z_DATA_ERROR)
+    z->msg = "oversubscribed dynamic bit lengths tree";
+  else if (r == Z_BUF_ERROR)
+  {
+    inflate_trees_free(*tb, z);
+    z->msg = "incomplete dynamic bit lengths tree";
+    r = Z_DATA_ERROR;
+  }
+  return r;
+}
+
+
+local int inflate_trees_dynamic(nl, nd, c, bl, bd, tl, td, z)
+uInt nl;                /* number of literal/length codes */
+uInt nd;                /* number of distance codes */
+uIntf *c;               /* that many (total) code lengths */
+uIntf *bl;              /* literal desired/actual bit depth */
+uIntf *bd;              /* distance desired/actual bit depth */
+inflate_huft * FAR *tl; /* literal/length tree result */
+inflate_huft * FAR *td; /* distance tree result */
+z_stream *z;            /* for zfree function */
+{
+  int r;
+
+  /* build literal/length tree */
+  if ((r = huft_build(c, nl, 257, cplens, cplext, tl, bl, z)) != Z_OK)
+  {
+    if (r == Z_DATA_ERROR)
+      z->msg = "oversubscribed literal/length tree";
+    else if (r == Z_BUF_ERROR)
+    {
+      inflate_trees_free(*tl, z);
+      z->msg = "incomplete literal/length tree";
+      r = Z_DATA_ERROR;
+    }
+    return r;
+  }
+
+  /* build distance tree */
+  if ((r = huft_build(c + nl, nd, 0, cpdist, cpdext, td, bd, z)) != Z_OK)
+  {
+    if (r == Z_DATA_ERROR)
+      z->msg = "oversubscribed literal/length tree";
+    else if (r == Z_BUF_ERROR) {
+#ifdef PKZIP_BUG_WORKAROUND
+      r = Z_OK;
+    }
+#else
+      inflate_trees_free(*td, z);
+      z->msg = "incomplete literal/length tree";
+      r = Z_DATA_ERROR;
+    }
+    inflate_trees_free(*tl, z);
+    return r;
+#endif
+  }
+
+  /* done */
+  return Z_OK;
+}
+
+
+/* build fixed tables only once--keep them here */
+local int fixed_lock = 0;
+local int fixed_built = 0;
+#define FIXEDH 530      /* number of hufts used by fixed tables */
+local uInt fixed_left = FIXEDH;
+local inflate_huft fixed_mem[FIXEDH];
+local uInt fixed_bl;
+local uInt fixed_bd;
+local inflate_huft *fixed_tl;
+local inflate_huft *fixed_td;
+
+
+local voidpf falloc(q, n, s)
+voidpf q;        /* opaque pointer (not used) */
+uInt n;         /* number of items */
+uInt s;         /* size of item */
+{
+  Assert(s == sizeof(inflate_huft) && n <= fixed_left,
+         "inflate_trees falloc overflow");
+  if (q) s++; /* to make some compilers happy */
+  fixed_left -= n;
+  return (voidpf)(fixed_mem + fixed_left);
+}
+
+
+local void ffree(q, p, n)
+voidpf q;
+voidpf p;
+uInt n;
+{
+  Assert(0, "inflate_trees ffree called!");
+  if (q) q = p; /* to make some compilers happy */
+}
+
+
+local int inflate_trees_fixed(bl, bd, tl, td)
+uIntf *bl;               /* literal desired/actual bit depth */
+uIntf *bd;               /* distance desired/actual bit depth */
+inflate_huft * FAR *tl;  /* literal/length tree result */
+inflate_huft * FAR *td;  /* distance tree result */
+{
+  /* build fixed tables if not built already--lock out other instances */
+  while (++fixed_lock > 1)
+    fixed_lock--;
+  if (!fixed_built)
+  {
+    int k;              /* temporary variable */
+    unsigned c[288];    /* length list for huft_build */
+    z_stream z;         /* for falloc function */
+
+    /* set up fake z_stream for memory routines */
+    z.zalloc = falloc;
+    z.zfree = ffree;
+    z.opaque = Z_NULL;
+
+    /* literal table */
+    for (k = 0; k < 144; k++)
+      c[k] = 8;
+    for (; k < 256; k++)
+      c[k] = 9;
+    for (; k < 280; k++)
+      c[k] = 7;
+    for (; k < 288; k++)
+      c[k] = 8;
+    fixed_bl = 7;
+    huft_build(c, 288, 257, cplens, cplext, &fixed_tl, &fixed_bl, &z);
+
+    /* distance table */
+    for (k = 0; k < 30; k++)
+      c[k] = 5;
+    fixed_bd = 5;
+    huft_build(c, 30, 0, cpdist, cpdext, &fixed_td, &fixed_bd, &z);
+
+    /* done */
+    fixed_built = 1;
+  }
+  fixed_lock--;
+  *bl = fixed_bl;
+  *bd = fixed_bd;
+  *tl = fixed_tl;
+  *td = fixed_td;
+  return Z_OK;
+}
+
+
+local int inflate_trees_free(t, z)
+inflate_huft *t;        /* table to free */
+z_stream *z;            /* for zfree function */
+/* Free the malloc'ed tables built by huft_build(), which makes a linked
+   list of the tables it made, with the links in a dummy first entry of
+   each table. */
+{
+  register inflate_huft *p, *q;
+
+  /* Go through linked list, freeing from the malloced (t[-1]) address. */
+  p = t;
+  while (p != Z_NULL)
+  {
+    q = (--p)->next;
+    ZFREE(z, p, p->word.Nalloc * sizeof(inflate_huft));
+    p = q;
+  } 
+  return Z_OK;
+}
+
+/*+++++*/
+/* infcodes.c -- process literals and length/distance pairs
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* simplify the use of the inflate_huft type with some defines */
+#define base more.Base
+#define next more.Next
+#define exop word.what.Exop
+#define bits word.what.Bits
+
+/* inflate codes private state */
+struct inflate_codes_state {
+
+  /* mode */
+  enum {        /* waiting for "i:"=input, "o:"=output, "x:"=nothing */
+      START,    /* x: set up for LEN */
+      LEN,      /* i: get length/literal/eob next */
+      LENEXT,   /* i: getting length extra (have base) */
+      DIST,     /* i: get distance next */
+      DISTEXT,  /* i: getting distance extra */
+      COPY,     /* o: copying bytes in window, waiting for space */
+      LIT,      /* o: got literal, waiting for output space */
+      WASH,     /* o: got eob, possibly still output waiting */
+      END,      /* x: got eob and all data flushed */
+      BADCODE}  /* x: got error */
+    mode;               /* current inflate_codes mode */
+
+  /* mode dependent information */
+  uInt len;
+  union {
+    struct {
+      inflate_huft *tree;       /* pointer into tree */
+      uInt need;                /* bits needed */
+    } code;             /* if LEN or DIST, where in tree */
+    uInt lit;           /* if LIT, literal */
+    struct {
+      uInt get;                 /* bits to get for extra */
+      uInt dist;                /* distance back to copy from */
+    } copy;             /* if EXT or COPY, where and how much */
+  } sub;                /* submode */
+
+  /* mode independent information */
+  Byte lbits;           /* ltree bits decoded per branch */
+  Byte dbits;           /* dtree bits decoder per branch */
+  inflate_huft *ltree;          /* literal/length/eob tree */
+  inflate_huft *dtree;          /* distance tree */
+
+};
+
+
+local inflate_codes_statef *inflate_codes_new(bl, bd, tl, td, z)
+uInt bl, bd;
+inflate_huft *tl, *td;
+z_stream *z;
+{
+  inflate_codes_statef *c;
+
+  if ((c = (inflate_codes_statef *)
+       ZALLOC(z,1,sizeof(struct inflate_codes_state))) != Z_NULL)
+  {
+    c->mode = START;
+    c->lbits = (Byte)bl;
+    c->dbits = (Byte)bd;
+    c->ltree = tl;
+    c->dtree = td;
+    Tracev((stderr, "inflate:       codes new\n"));
+  }
+  return c;
+}
+
+
+local int inflate_codes(s, z, r)
+inflate_blocks_statef *s;
+z_stream *z;
+int r;
+{
+  uInt j;               /* temporary storage */
+  inflate_huft *t;      /* temporary pointer */
+  uInt e;               /* extra bits or operation */
+  uLong b;              /* bit buffer */
+  uInt k;               /* bits in bit buffer */
+  Bytef *p;             /* input data pointer */
+  uInt n;               /* bytes available there */
+  Bytef *q;             /* output window write pointer */
+  uInt m;               /* bytes to end of window or read pointer */
+  Bytef *f;             /* pointer to copy strings from */
+  inflate_codes_statef *c = s->sub.decode.codes;  /* codes state */
+
+  /* copy input/output information to locals (UPDATE macro restores) */
+  LOAD
+
+  /* process input and output based on current state */
+  while (1) switch (c->mode)
+  {             /* waiting for "i:"=input, "o:"=output, "x:"=nothing */
+    case START:         /* x: set up for LEN */
+#ifndef SLOW
+      if (m >= 258 && n >= 10)
+      {
+        UPDATE
+        r = inflate_fast(c->lbits, c->dbits, c->ltree, c->dtree, s, z);
+        LOAD
+        if (r != Z_OK)
+        {
+          c->mode = r == Z_STREAM_END ? WASH : BADCODE;
+          break;
+        }
+      }
+#endif /* !SLOW */
+      c->sub.code.need = c->lbits;
+      c->sub.code.tree = c->ltree;
+      c->mode = LEN;
+    case LEN:           /* i: get length/literal/eob next */
+      j = c->sub.code.need;
+      NEEDBITS(j)
+      t = c->sub.code.tree + ((uInt)b & inflate_mask[j]);
+      DUMPBITS(t->bits)
+      e = (uInt)(t->exop);
+      if (e == 0)               /* literal */
+      {
+        c->sub.lit = t->base;
+        Tracevv((stderr, t->base >= 0x20 && t->base < 0x7f ?
+                 "inflate:         literal '%c'\n" :
+                 "inflate:         literal 0x%02x\n", t->base));
+        c->mode = LIT;
+        break;
+      }
+      if (e & 16)               /* length */
+      {
+        c->sub.copy.get = e & 15;
+        c->len = t->base;
+        c->mode = LENEXT;
+        break;
+      }
+      if ((e & 64) == 0)        /* next table */
+      {
+        c->sub.code.need = e;
+        c->sub.code.tree = t->next;
+        break;
+      }
+      if (e & 32)               /* end of block */
+      {
+        Tracevv((stderr, "inflate:         end of block\n"));
+        c->mode = WASH;
+        break;
+      }
+      c->mode = BADCODE;        /* invalid code */
+      z->msg = "invalid literal/length code";
+      r = Z_DATA_ERROR;
+      LEAVE
+    case LENEXT:        /* i: getting length extra (have base) */
+      j = c->sub.copy.get;
+      NEEDBITS(j)
+      c->len += (uInt)b & inflate_mask[j];
+      DUMPBITS(j)
+      c->sub.code.need = c->dbits;
+      c->sub.code.tree = c->dtree;
+      Tracevv((stderr, "inflate:         length %u\n", c->len));
+      c->mode = DIST;
+    case DIST:          /* i: get distance next */
+      j = c->sub.code.need;
+      NEEDBITS(j)
+      t = c->sub.code.tree + ((uInt)b & inflate_mask[j]);
+      DUMPBITS(t->bits)
+      e = (uInt)(t->exop);
+      if (e & 16)               /* distance */
+      {
+        c->sub.copy.get = e & 15;
+        c->sub.copy.dist = t->base;
+        c->mode = DISTEXT;
+        break;
+      }
+      if ((e & 64) == 0)        /* next table */
+      {
+        c->sub.code.need = e;
+        c->sub.code.tree = t->next;
+        break;
+      }
+      c->mode = BADCODE;        /* invalid code */
+      z->msg = "invalid distance code";
+      r = Z_DATA_ERROR;
+      LEAVE
+    case DISTEXT:       /* i: getting distance extra */
+      j = c->sub.copy.get;
+      NEEDBITS(j)
+      c->sub.copy.dist += (uInt)b & inflate_mask[j];
+      DUMPBITS(j)
+      Tracevv((stderr, "inflate:         distance %u\n", c->sub.copy.dist));
+      c->mode = COPY;
+    case COPY:          /* o: copying bytes in window, waiting for space */
+#ifndef __TURBOC__ /* Turbo C bug for following expression */
+      f = (uInt)(q - s->window) < c->sub.copy.dist ?
+          s->end - (c->sub.copy.dist - (q - s->window)) :
+          q - c->sub.copy.dist;
+#else
+      f = q - c->sub.copy.dist;
+      if ((uInt)(q - s->window) < c->sub.copy.dist)
+        f = s->end - (c->sub.copy.dist - (q - s->window));
+#endif
+      while (c->len)
+      {
+        NEEDOUT
+        OUTBYTE(*f++)
+        if (f == s->end)
+          f = s->window;
+        c->len--;
+      }
+      c->mode = START;
+      break;
+    case LIT:           /* o: got literal, waiting for output space */
+      NEEDOUT
+      OUTBYTE(c->sub.lit)
+      c->mode = START;
+      break;
+    case WASH:          /* o: got eob, possibly more output */
+      FLUSH
+      if (s->read != s->write)
+        LEAVE
+      c->mode = END;
+    case END:
+      r = Z_STREAM_END;
+      LEAVE
+    case BADCODE:       /* x: got error */
+      r = Z_DATA_ERROR;
+      LEAVE
+    default:
+      r = Z_STREAM_ERROR;
+      LEAVE
+  }
+}
+
+
+local void inflate_codes_free(c, z)
+inflate_codes_statef *c;
+z_stream *z;
+{
+  ZFREE(z, c, sizeof(struct inflate_codes_state));
+  Tracev((stderr, "inflate:       codes free\n"));
+}
+
+/*+++++*/
+/* inflate_util.c -- data and routines common to blocks and codes
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* copy as much as possible from the sliding window to the output area */
+local int inflate_flush(s, z, r)
+inflate_blocks_statef *s;
+z_stream *z;
+int r;
+{
+  uInt n;
+  Bytef *p, *q;
+
+  /* local copies of source and destination pointers */
+  p = z->next_out;
+  q = s->read;
+
+  /* compute number of bytes to copy as far as end of window */
+  n = (uInt)((q <= s->write ? s->write : s->end) - q);
+  if (n > z->avail_out) n = z->avail_out;
+  if (n && r == Z_BUF_ERROR) r = Z_OK;
+
+  /* update counters */
+  z->avail_out -= n;
+  z->total_out += n;
+
+  /* update check information */
+  if (s->checkfn != Z_NULL)
+    s->check = (*s->checkfn)(s->check, q, n);
+
+  /* copy as far as end of window */
+  zmemcpy(p, q, n);
+  p += n;
+  q += n;
+
+  /* see if more to copy at beginning of window */
+  if (q == s->end)
+  {
+    /* wrap pointers */
+    q = s->window;
+    if (s->write == s->end)
+      s->write = s->window;
+
+    /* compute bytes to copy */
+    n = (uInt)(s->write - q);
+    if (n > z->avail_out) n = z->avail_out;
+    if (n && r == Z_BUF_ERROR) r = Z_OK;
+
+    /* update counters */
+    z->avail_out -= n;
+    z->total_out += n;
+
+    /* update check information */
+    if (s->checkfn != Z_NULL)
+      s->check = (*s->checkfn)(s->check, q, n);
+
+    /* copy */
+    zmemcpy(p, q, n);
+    p += n;
+    q += n;
+  }
+
+  /* update pointers */
+  z->next_out = p;
+  s->read = q;
+
+  /* done */
+  return r;
+}
+
+
+/*+++++*/
+/* inffast.c -- process literals and length/distance pairs fast
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* simplify the use of the inflate_huft type with some defines */
+#define base more.Base
+#define next more.Next
+#define exop word.what.Exop
+#define bits word.what.Bits
+
+/* macros for bit input with no checking and for returning unused bytes */
+#define GRABBITS(j) {while(k<(j)){b|=((uLong)NEXTBYTE)<<k;k+=8;}}
+#define UNGRAB {n+=(c=k>>3);p-=c;k&=7;}
+
+/* Called with number of bytes left to write in window at least 258
+   (the maximum string length) and number of input bytes available
+   at least ten.  The ten bytes are six bytes for the longest length/
+   distance pair plus four bytes for overloading the bit buffer. */
+
+local int inflate_fast(bl, bd, tl, td, s, z)
+uInt bl, bd;
+inflate_huft *tl, *td;
+inflate_blocks_statef *s;
+z_stream *z;
+{
+  inflate_huft *t;      /* temporary pointer */
+  uInt e;               /* extra bits or operation */
+  uLong b;              /* bit buffer */
+  uInt k;               /* bits in bit buffer */
+  Bytef *p;             /* input data pointer */
+  uInt n;               /* bytes available there */
+  Bytef *q;             /* output window write pointer */
+  uInt m;               /* bytes to end of window or read pointer */
+  uInt ml;              /* mask for literal/length tree */
+  uInt md;              /* mask for distance tree */
+  uInt c;               /* bytes to copy */
+  uInt d;               /* distance back to copy from */
+  Bytef *r;             /* copy source pointer */
+
+  /* load input, output, bit values */
+  LOAD
+
+  /* initialize masks */
+  ml = inflate_mask[bl];
+  md = inflate_mask[bd];
+
+  /* do until not enough input or output space for fast loop */
+  do {                          /* assume called with m >= 258 && n >= 10 */
+    /* get literal/length code */
+    GRABBITS(20)                /* max bits for literal/length code */
+    if ((e = (t = tl + ((uInt)b & ml))->exop) == 0)
+    {
+      DUMPBITS(t->bits)
+      Tracevv((stderr, t->base >= 0x20 && t->base < 0x7f ?
+                "inflate:         * literal '%c'\n" :
+                "inflate:         * literal 0x%02x\n", t->base));
+      *q++ = (Byte)t->base;
+      m--;
+      continue;
+    }
+    do {
+      DUMPBITS(t->bits)
+      if (e & 16)
+      {
+        /* get extra bits for length */
+        e &= 15;
+        c = t->base + ((uInt)b & inflate_mask[e]);
+        DUMPBITS(e)
+        Tracevv((stderr, "inflate:         * length %u\n", c));
+
+        /* decode distance base of block to copy */
+        GRABBITS(15);           /* max bits for distance code */
+        e = (t = td + ((uInt)b & md))->exop;
+        do {
+          DUMPBITS(t->bits)
+          if (e & 16)
+          {
+            /* get extra bits to add to distance base */
+            e &= 15;
+            GRABBITS(e)         /* get extra bits (up to 13) */
+            d = t->base + ((uInt)b & inflate_mask[e]);
+            DUMPBITS(e)
+            Tracevv((stderr, "inflate:         * distance %u\n", d));
+
+            /* do the copy */
+            m -= c;
+            if ((uInt)(q - s->window) >= d)     /* offset before dest */
+            {                                   /*  just copy */
+              r = q - d;
+              *q++ = *r++;  c--;        /* minimum count is three, */
+              *q++ = *r++;  c--;        /*  so unroll loop a little */
+            }
+            else                        /* else offset after destination */
+            {
+              e = d - (q - s->window);  /* bytes from offset to end */
+              r = s->end - e;           /* pointer to offset */
+              if (c > e)                /* if source crosses, */
+              {
+                c -= e;                 /* copy to end of window */
+                do {
+                  *q++ = *r++;
+                } while (--e);
+                r = s->window;          /* copy rest from start of window */
+              }
+            }
+            do {                        /* copy all or what's left */
+              *q++ = *r++;
+            } while (--c);
+            break;
+          }
+          else if ((e & 64) == 0)
+            e = (t = t->next + ((uInt)b & inflate_mask[e]))->exop;
+          else
+          {
+            z->msg = "invalid distance code";
+            UNGRAB
+            UPDATE
+            return Z_DATA_ERROR;
+          }
+        } while (1);
+        break;
+      }
+      if ((e & 64) == 0)
+      {
+        if ((e = (t = t->next + ((uInt)b & inflate_mask[e]))->exop) == 0)
+        {
+          DUMPBITS(t->bits)
+          Tracevv((stderr, t->base >= 0x20 && t->base < 0x7f ?
+                    "inflate:         * literal '%c'\n" :
+                    "inflate:         * literal 0x%02x\n", t->base));
+          *q++ = (Byte)t->base;
+          m--;
+          break;
+        }
+      }
+      else if (e & 32)
+      {
+        Tracevv((stderr, "inflate:         * end of block\n"));
+        UNGRAB
+        UPDATE
+        return Z_STREAM_END;
+      }
+      else
+      {
+        z->msg = "invalid literal/length code";
+        UNGRAB
+        UPDATE
+        return Z_DATA_ERROR;
+      }
+    } while (1);
+  } while (m >= 258 && n >= 10);
+
+  /* not enough input or output--restore pointers and return */
+  UNGRAB
+  UPDATE
+  return Z_OK;
+}
+
+
+/*+++++*/
+/* zutil.c -- target dependent utility functions for the compression library
+ * Copyright (C) 1995 Jean-loup Gailly.
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* From: zutil.c,v 1.8 1995/05/03 17:27:12 jloup Exp */
+
+char *zlib_version = ZLIB_VERSION;
+
+char *z_errmsg[] = {
+"stream end",          /* Z_STREAM_END    1 */
+"",                    /* Z_OK            0 */
+"file error",          /* Z_ERRNO        (-1) */
+"stream error",        /* Z_STREAM_ERROR (-2) */
+"data error",          /* Z_DATA_ERROR   (-3) */
+"insufficient memory", /* Z_MEM_ERROR    (-4) */
+"buffer error",        /* Z_BUF_ERROR    (-5) */
+""};
+
+
+/*+++++*/
+/* adler32.c -- compute the Adler-32 checksum of a data stream
+ * Copyright (C) 1995 Mark Adler
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* From: adler32.c,v 1.6 1995/05/03 17:27:08 jloup Exp */
+
+#define BASE 65521L /* largest prime smaller than 65536 */
+#define NMAX 5552
+/* NMAX is the largest n such that 255n(n+1)/2 + (n+1)(BASE-1) <= 2^32-1 */
+
+#define DO1(buf)  {s1 += *buf++; s2 += s1;}
+#define DO2(buf)  DO1(buf); DO1(buf);
+#define DO4(buf)  DO2(buf); DO2(buf);
+#define DO8(buf)  DO4(buf); DO4(buf);
+#define DO16(buf) DO8(buf); DO8(buf);
+
+/* ========================================================================= */
+uLong adler32(adler, buf, len)
+    uLong adler;
+    Bytef *buf;
+    uInt len;
+{
+    unsigned long s1 = adler & 0xffff;
+    unsigned long s2 = (adler >> 16) & 0xffff;
+    int k;
+
+    if (buf == Z_NULL) return 1L;
+
+    while (len > 0) {
+        k = len < NMAX ? len : NMAX;
+        len -= k;
+        while (k >= 16) {
+            DO16(buf);
+            k -= 16;
+        }
+        if (k != 0) do {
+            DO1(buf);
+        } while (--k);
+        s1 %= BASE;
+        s2 %= BASE;
+    }
+    return (s2 << 16) | s1;
+}
diff --git a/arch/ppc/coffboot/zlib.h b/arch/ppc/coffboot/zlib.h
new file mode 100644
index 000000000..f4ab77617
--- /dev/null
+++ b/arch/ppc/coffboot/zlib.h
@@ -0,0 +1,432 @@
+/*	$Id: zlib.h,v 1.1 1997/08/30 04:51:49 ralf Exp $	*/
+
+/*
+ * This file is derived from zlib.h and zconf.h from the zlib-0.95
+ * distribution by Jean-loup Gailly and Mark Adler, with some additions
+ * by Paul Mackerras to aid in implementing Deflate compression and
+ * decompression for PPP packets.
+ */
+
+/*
+ *  ==FILEVERSION 960122==
+ *
+ * This marker is used by the Linux installation script to determine
+ * whether an up-to-date version of this file is already installed.
+ */
+
+/* zlib.h -- interface of the 'zlib' general purpose compression library
+  version 0.95, Aug 16th, 1995.
+
+  Copyright (C) 1995 Jean-loup Gailly and Mark Adler
+
+  This software is provided 'as-is', without any express or implied
+  warranty.  In no event will the authors be held liable for any damages
+  arising from the use of this software.
+
+  Permission is granted to anyone to use this software for any purpose,
+  including commercial applications, and to alter it and redistribute it
+  freely, subject to the following restrictions:
+
+  1. The origin of this software must not be misrepresented; you must not
+     claim that you wrote the original software. If you use this software
+     in a product, an acknowledgment in the product documentation would be
+     appreciated but is not required.
+  2. Altered source versions must be plainly marked as such, and must not be
+     misrepresented as being the original software.
+  3. This notice may not be removed or altered from any source distribution.
+
+  Jean-loup Gailly        Mark Adler
+  gzip@prep.ai.mit.edu    madler@alumni.caltech.edu
+ */
+
+#ifndef _ZLIB_H
+#define _ZLIB_H
+
+/* #include "zconf.h" */	/* included directly here */
+
+/* zconf.h -- configuration of the zlib compression library
+ * Copyright (C) 1995 Jean-loup Gailly.
+ * For conditions of distribution and use, see copyright notice in zlib.h 
+ */
+
+/* From: zconf.h,v 1.12 1995/05/03 17:27:12 jloup Exp */
+
+/*
+     The library does not install any signal handler. It is recommended to
+  add at least a handler for SIGSEGV when decompressing; the library checks
+  the consistency of the input data whenever possible but may go nuts
+  for some forms of corrupted input.
+ */
+
+/*
+ * Compile with -DMAXSEG_64K if the alloc function cannot allocate more
+ * than 64k bytes at a time (needed on systems with 16-bit int).
+ * Compile with -DUNALIGNED_OK if it is OK to access shorts or ints
+ * at addresses which are not a multiple of their size.
+ * Under DOS, -DFAR=far or -DFAR=__far may be needed.
+ */
+
+#ifndef STDC
+#  if defined(MSDOS) || defined(__STDC__) || defined(__cplusplus)
+#    define STDC
+#  endif
+#endif
+
+#ifdef	__MWERKS__ /* Metrowerks CodeWarrior declares fileno() in unix.h */
+#  include <unix.h>
+#endif
+
+/* Maximum value for memLevel in deflateInit2 */
+#ifndef MAX_MEM_LEVEL
+#  ifdef MAXSEG_64K
+#    define MAX_MEM_LEVEL 8
+#  else
+#    define MAX_MEM_LEVEL 9
+#  endif
+#endif
+
+#ifndef FAR
+#  define FAR
+#endif
+
+/* Maximum value for windowBits in deflateInit2 and inflateInit2 */
+#ifndef MAX_WBITS
+#  define MAX_WBITS   15 /* 32K LZ77 window */
+#endif
+
+/* The memory requirements for deflate are (in bytes):
+            1 << (windowBits+2)   +  1 << (memLevel+9)
+ that is: 128K for windowBits=15  +  128K for memLevel = 8  (default values)
+ plus a few kilobytes for small objects. For example, if you want to reduce
+ the default memory requirements from 256K to 128K, compile with
+     make CFLAGS="-O -DMAX_WBITS=14 -DMAX_MEM_LEVEL=7"
+ Of course this will generally degrade compression (there's no free lunch).
+
+   The memory requirements for inflate are (in bytes) 1 << windowBits
+ that is, 32K for windowBits=15 (default value) plus a few kilobytes
+ for small objects.
+*/
+
+                        /* Type declarations */
+
+#ifndef OF /* function prototypes */
+#  ifdef STDC
+#    define OF(args)  args
+#  else
+#    define OF(args)  ()
+#  endif
+#endif
+
+typedef unsigned char  Byte;  /* 8 bits */
+typedef unsigned int   uInt;  /* 16 bits or more */
+typedef unsigned long  uLong; /* 32 bits or more */
+
+typedef Byte FAR Bytef;
+typedef char FAR charf;
+typedef int FAR intf;
+typedef uInt FAR uIntf;
+typedef uLong FAR uLongf;
+
+#ifdef STDC
+   typedef void FAR *voidpf;
+   typedef void     *voidp;
+#else
+   typedef Byte FAR *voidpf;
+   typedef Byte     *voidp;
+#endif
+
+/* end of original zconf.h */
+
+#define ZLIB_VERSION "0.95P"
+
+/* 
+     The 'zlib' compression library provides in-memory compression and
+  decompression functions, including integrity checks of the uncompressed
+  data.  This version of the library supports only one compression method
+  (deflation) but other algorithms may be added later and will have the same
+  stream interface.
+
+     For compression the application must provide the output buffer and
+  may optionally provide the input buffer for optimization. For decompression,
+  the application must provide the input buffer and may optionally provide
+  the output buffer for optimization.
+
+     Compression can be done in a single step if the buffers are large
+  enough (for example if an input file is mmap'ed), or can be done by
+  repeated calls of the compression function.  In the latter case, the
+  application must provide more input and/or consume the output
+  (providing more output space) before each call.
+*/
+
+typedef voidpf (*alloc_func) OF((voidpf opaque, uInt items, uInt size));
+typedef void   (*free_func)  OF((voidpf opaque, voidpf address, uInt nbytes));
+
+struct internal_state;
+
+typedef struct z_stream_s {
+    Bytef    *next_in;  /* next input byte */
+    uInt     avail_in;  /* number of bytes available at next_in */
+    uLong    total_in;  /* total nb of input bytes read so far */
+
+    Bytef    *next_out; /* next output byte should be put there */
+    uInt     avail_out; /* remaining free space at next_out */
+    uLong    total_out; /* total nb of bytes output so far */
+
+    char     *msg;      /* last error message, NULL if no error */
+    struct internal_state FAR *state; /* not visible by applications */
+
+    alloc_func zalloc;  /* used to allocate the internal state */
+    free_func  zfree;   /* used to free the internal state */
+    voidp      opaque;  /* private data object passed to zalloc and zfree */
+
+    Byte     data_type; /* best guess about the data type: ascii or binary */
+
+} z_stream;
+
+/*
+   The application must update next_in and avail_in when avail_in has
+   dropped to zero. It must update next_out and avail_out when avail_out
+   has dropped to zero. The application must initialize zalloc, zfree and
+   opaque before calling the init function. All other fields are set by the
+   compression library and must not be updated by the application.
+
+   The opaque value provided by the application will be passed as the first
+   parameter for calls of zalloc and zfree. This can be useful for custom
+   memory management. The compression library attaches no meaning to the
+   opaque value.
+
+   zalloc must return Z_NULL if there is not enough memory for the object.
+   On 16-bit systems, the functions zalloc and zfree must be able to allocate
+   exactly 65536 bytes, but will not be required to allocate more than this
+   if the symbol MAXSEG_64K is defined (see zconf.h). WARNING: On MSDOS,
+   pointers returned by zalloc for objects of exactly 65536 bytes *must*
+   have their offset normalized to zero. The default allocation function
+   provided by this library ensures this (see zutil.c). To reduce memory
+   requirements and avoid any allocation of 64K objects, at the expense of
+   compression ratio, compile the library with -DMAX_WBITS=14 (see zconf.h).
+
+   The fields total_in and total_out can be used for statistics or
+   progress reports. After compression, total_in holds the total size of
+   the uncompressed data and may be saved for use in the decompressor
+   (particularly if the decompressor wants to decompress everything in
+   a single step).
+*/
+
+                        /* constants */
+
+#define Z_NO_FLUSH      0
+#define Z_PARTIAL_FLUSH 1
+#define Z_FULL_FLUSH    2
+#define Z_SYNC_FLUSH    3 /* experimental: partial_flush + byte align */
+#define Z_FINISH        4
+#define Z_PACKET_FLUSH	5
+/* See deflate() below for the usage of these constants */
+
+#define Z_OK            0
+#define Z_STREAM_END    1
+#define Z_ERRNO        (-1)
+#define Z_STREAM_ERROR (-2)
+#define Z_DATA_ERROR   (-3)
+#define Z_MEM_ERROR    (-4)
+#define Z_BUF_ERROR    (-5)
+/* error codes for the compression/decompression functions */
+
+#define Z_BEST_SPEED             1
+#define Z_BEST_COMPRESSION       9
+#define Z_DEFAULT_COMPRESSION  (-1)
+/* compression levels */
+
+#define Z_FILTERED            1
+#define Z_HUFFMAN_ONLY        2
+#define Z_DEFAULT_STRATEGY    0
+
+#define Z_BINARY   0
+#define Z_ASCII    1
+#define Z_UNKNOWN  2
+/* Used to set the data_type field */
+
+#define Z_NULL  0  /* for initializing zalloc, zfree, opaque */
+
+extern char *zlib_version;
+/* The application can compare zlib_version and ZLIB_VERSION for consistency.
+   If the first character differs, the library code actually used is
+   not compatible with the zlib.h header file used by the application.
+ */
+
+                        /* basic functions */
+
+extern int inflateInit OF((z_stream *strm));
+/* 
+     Initializes the internal stream state for decompression. The fields
+   zalloc and zfree must be initialized before by the caller.  If zalloc and
+   zfree are set to Z_NULL, inflateInit updates them to use default allocation
+   functions.
+
+     inflateInit returns Z_OK if success, Z_MEM_ERROR if there was not
+   enough memory.  msg is set to null if there is no error message.
+   inflateInit does not perform any decompression: this will be done by
+   inflate().
+*/
+
+
+extern int inflate OF((z_stream *strm, int flush));
+/*
+  Performs one or both of the following actions:
+
+  - Decompress more input starting at next_in and update next_in and avail_in
+    accordingly. If not all input can be processed (because there is not
+    enough room in the output buffer), next_in is updated and processing
+    will resume at this point for the next call of inflate().
+
+  - Provide more output starting at next_out and update next_out and avail_out
+    accordingly.  inflate() always provides as much output as possible
+    (until there is no more input data or no more space in the output buffer).
+
+  Before the call of inflate(), the application should ensure that at least
+  one of the actions is possible, by providing more input and/or consuming
+  more output, and updating the next_* and avail_* values accordingly.
+  The application can consume the uncompressed output when it wants, for
+  example when the output buffer is full (avail_out == 0), or after each
+  call of inflate().
+
+    If the parameter flush is set to Z_PARTIAL_FLUSH or Z_PACKET_FLUSH,
+  inflate flushes as much output as possible to the output buffer. The
+  flushing behavior of inflate is not specified for values of the flush
+  parameter other than Z_PARTIAL_FLUSH, Z_PACKET_FLUSH or Z_FINISH, but the
+  current implementation actually flushes as much output as possible
+  anyway.  For Z_PACKET_FLUSH, inflate checks that once all the input data
+  has been consumed, it is expecting to see the length field of a stored
+  block; if not, it returns Z_DATA_ERROR.
+
+    inflate() should normally be called until it returns Z_STREAM_END or an
+  error. However if all decompression is to be performed in a single step
+  (a single call of inflate), the parameter flush should be set to
+  Z_FINISH. In this case all pending input is processed and all pending
+  output is flushed; avail_out must be large enough to hold all the
+  uncompressed data. (The size of the uncompressed data may have been saved
+  by the compressor for this purpose.) The next operation on this stream must
+  be inflateEnd to deallocate the decompression state. The use of Z_FINISH
+  is never required, but can be used to inform inflate that a faster routine
+  may be used for the single inflate() call.
+
+    inflate() returns Z_OK if some progress has been made (more input
+  processed or more output produced), Z_STREAM_END if the end of the
+  compressed data has been reached and all uncompressed output has been
+  produced, Z_DATA_ERROR if the input data was corrupted, Z_STREAM_ERROR if
+  the stream structure was inconsistent (for example if next_in or next_out
+  was NULL), Z_MEM_ERROR if there was not enough memory, Z_BUF_ERROR if no
+  progress is possible or if there was not enough room in the output buffer
+  when Z_FINISH is used. In the Z_DATA_ERROR case, the application may then
+  call inflateSync to look for a good compression block.  */
+
+
+extern int inflateEnd OF((z_stream *strm));
+/*
+     All dynamically allocated data structures for this stream are freed.
+   This function discards any unprocessed input and does not flush any
+   pending output.
+
+     inflateEnd returns Z_OK if success, Z_STREAM_ERROR if the stream state
+   was inconsistent. In the error case, msg may be set but then points to a
+   static string (which must not be deallocated).
+*/
+
+                        /* advanced functions */
+
+extern int inflateInit2 OF((z_stream *strm,
+                            int  windowBits));
+/*   
+     This is another version of inflateInit with more compression options. The
+   fields next_out, zalloc and zfree must be initialized before by the caller.
+
+     The windowBits parameter is the base two logarithm of the maximum window
+   size (the size of the history buffer).  It should be in the range 8..15 for
+   this version of the library (the value 16 will be allowed soon). The
+   default value is 15 if inflateInit is used instead. If a compressed stream
+   with a larger window size is given as input, inflate() will return with
+   the error code Z_DATA_ERROR instead of trying to allocate a larger window.
+
+     If next_out is not null, the library will use this buffer for the history
+   buffer; the buffer must either be large enough to hold the entire output
+   data, or have at least 1<<windowBits bytes.  If next_out is null, the
+   library will allocate its own buffer (and leave next_out null). next_in
+   need not be provided here but must be provided by the application for the
+   next call of inflate().
+
+     If the history buffer is provided by the application, next_out must
+   never be changed by the application since the decompressor maintains
+   history information inside this buffer from call to call; the application
+   can only reset next_out to the beginning of the history buffer when
+   avail_out is zero and all output has been consumed.
+
+      inflateInit2 returns Z_OK if success, Z_MEM_ERROR if there was
+   not enough memory, Z_STREAM_ERROR if a parameter is invalid (such as
+   windowBits < 8). msg is set to null if there is no error message.
+   inflateInit2 does not perform any decompression: this will be done by
+   inflate().
+*/
+
+extern int inflateSync OF((z_stream *strm));
+/* 
+    Skips invalid compressed data until the special marker (see deflate()
+  above) can be found, or until all available input is skipped. No output
+  is provided.
+
+    inflateSync returns Z_OK if the special marker has been found, Z_BUF_ERROR
+  if no more input was provided, Z_DATA_ERROR if no marker has been found,
+  or Z_STREAM_ERROR if the stream structure was inconsistent. In the success
+  case, the application may save the current current value of total_in which
+  indicates where valid compressed data was found. In the error case, the
+  application may repeatedly call inflateSync, providing more input each time,
+  until success or end of the input data.
+*/
+
+extern int inflateReset OF((z_stream *strm));
+/*
+     This function is equivalent to inflateEnd followed by inflateInit,
+   but does not free and reallocate all the internal decompression state.
+   The stream will keep attributes that may have been set by inflateInit2.
+
+      inflateReset returns Z_OK if success, or Z_STREAM_ERROR if the source
+   stream state was inconsistent (such as zalloc or state being NULL).
+*/
+
+extern int inflateIncomp OF((z_stream *strm));
+/*
+     This function adds the data at next_in (avail_in bytes) to the output
+   history without performing any output.  There must be no pending output,
+   and the decompressor must be expecting to see the start of a block.
+   Calling this function is equivalent to decompressing a stored block
+   containing the data at next_in (except that the data is not output).
+*/
+
+                        /* checksum functions */
+
+/*
+     This function is not related to compression but is exported
+   anyway because it might be useful in applications using the
+   compression library.
+*/
+
+extern uLong adler32 OF((uLong adler, Bytef *buf, uInt len));
+
+/*
+     Update a running Adler-32 checksum with the bytes buf[0..len-1] and
+   return the updated checksum. If buf is NULL, this function returns
+   the required initial value for the checksum.
+   An Adler-32 checksum is almost as reliable as a CRC32 but can be computed
+   much faster. Usage example:
+
+     uLong adler = adler32(0L, Z_NULL, 0);
+
+     while (read_buffer(buffer, length) != EOF) {
+       adler = adler32(adler, buffer, length);
+     }
+     if (adler != original_adler) error();
+*/
+
+#ifndef _Z_UTIL_H
+    struct internal_state {int dummy;}; /* hack for buggy compilers */
+#endif
+
+#endif /* _ZLIB_H */
diff --git a/arch/ppc/config.in b/arch/ppc/config.in
index 9df35fa24..1cea7ce3f 100644
--- a/arch/ppc/config.in
+++ b/arch/ppc/config.in
@@ -1,36 +1,44 @@
-#
+# $Id: config.in,v 1.19 1997/09/04 01:54:26 davem Exp $
 # For a description of the syntax of this configuration file,
 # see the Configure script.
 #
 mainmenu_name "Linux/PowerPC Kernel Configuration"
 
-if [ "`uname`" != "Linux" ]; then
+mainmenu_option next_comment
+comment 'Platform support'
+define_bool CONFIG_PPC y
+
+if [ "`uname`" != "Linux" -o "`uname -m`" != "ppc" ]; then
   define_bool CONFIG_CROSSCOMPILE y
 else
   define_bool CONFIG_NATIVE y
 fi
 
-bool 'Build PowerMac Kernel (not PReP)?' CONFIG_PMAC
-bool 'Build PReP Kernel (not PowerMac)?' CONFIG_PREP
-bool 'Prompt for development and/or incomplete code/drivers' CONFIG_EXPERIMENTAL
-bool 'Used Harddrive LED on IBM 83x workstations as heartbeat?' CONFIG_HEARTBEAT
-bool 'Used PowerPC specific powersaving?' CONFIG_POWERSAVING
+bool 'Build PowerMac Kernel (not PReP or CHRP)?' CONFIG_PMAC
+bool 'Build PReP Kernel (not PowerMac or CHRP)?' CONFIG_PREP
+bool 'Build CHRP Kernel (not PReP or PowerMac)?' CONFIG_CHRP
+
 choice 'Processor type' \
-	"Common		CONFIG_MCOMMON \
-	 601		CONFIG_M601	\
-	 603		CONFIG_M603	\
-	 604		CONFIG_M604" Common
+	"Common		CONFIG_COMMON \
+	 601		CONFIG_601	\
+	 603		CONFIG_603	\
+	 604		CONFIG_604" Common
+endmenu
+
+mainmenu_option next_comment
+comment 'General setup'
 
+bool 'Prompt for development and/or incomplete code/drivers' CONFIG_EXPERIMENTAL
 bool 'Enable loadable module support' CONFIG_MODULES
 if [ "$CONFIG_MODULES" = "y" ]; then
   bool 'Set version information on all symbols for modules' CONFIG_MODVERSIONS
   bool 'Kernel daemon support (e.g. autoload of modules)' CONFIG_KERNELD
 fi
 
-
-mainmenu_option next_comment
 define_bool CONFIG_PCI y
-bool 'PCI bridge optimization (experimental)' CONFIG_PCI_OPTIMIZE
+if [ "$CONFIG_PREP" = "y" ]; then
+  bool 'PCI bridge optimization' CONFIG_PCI_OPTIMIZE
+fi
 bool 'Networking support' CONFIG_NET
 bool 'Sysctl support' CONFIG_SYSCTL
 bool 'System V IPC' CONFIG_SYSVIPC
@@ -39,8 +47,24 @@ bool 'System V IPC' CONFIG_SYSVIPC
 define_bool CONFIG_BINFMT_ELF y
 define_bool CONFIG_KERNEL_ELF y
 tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
-tristate 'Kernel support for JAVA binaries' CONFIG_BINFMT_JAVA
+tristate 'Kernel support for JAVA binaries (obsolete)' CONFIG_BINFMT_JAVA
+
+if [ "$CONFIG_PMAC" = "y" ]; then
+  define_bool CONFIG_PMAC_CONSOLE y
+  define_bool CONFIG_MAC_KEYBOARD y
+  define_bool CONFIG_MAC_FLOPPY	y
+  bool 'Support for Open Firmware device tree in /proc' CONFIG_PROC_DEVICETREE
+  bool 'Include xmon kernel debugger' CONFIG_XMON
+fi
+
+if [ "$CONFIG_PMAC_CONSOLE" = "y" ]; then
+  bool 'Support for ATI Mach64 display cards' CONFIG_ATY_VIDEO
+  bool 'Support for IMS Twin Turbo display card' CONFIG_IMSTT_VIDEO
+else
+  define_bool CONFIG_VGA_CONSOLE y
+fi
 
+endmenu
 
 source drivers/pnp/Config.in
 source drivers/block/Config.in
@@ -83,6 +107,7 @@ fi
 endmenu
 
 source fs/Config.in
+
 source drivers/char/Config.in
 
 mainmenu_option next_comment
@@ -93,12 +118,12 @@ if [ "$CONFIG_SOUND" != "n" ]; then
 fi
 endmenu
 
-mainmenu_option next_comment
+#mainmenu_option next_comment
 #comment 'Kernel hacking'
 #bool 'Debug kmalloc/kfree' CONFIG_DEBUG_MALLOC
 #bool 'Kernel profiling support' CONFIG_PROFILE
 #if [ "$CONFIG_PROFILE" = "y" ]; then
 #  int ' Profile shift count' CONFIG_PROFILE_SHIFT 2
 #fi
-endmenu
+#endmenu
 
diff --git a/arch/ppc/defconfig b/arch/ppc/defconfig
new file mode 100644
index 000000000..ea39e83ea
--- /dev/null
+++ b/arch/ppc/defconfig
@@ -0,0 +1,250 @@
+#
+# Automatically generated by make menuconfig: don't edit
+#
+
+#
+# Platform support
+#
+CONFIG_PPC=y
+CONFIG_NATIVE=y
+# CONFIG_PMAC is not set
+CONFIG_PREP=y
+CONFIG_MCOMMON=y
+
+#
+# General setup
+#
+CONFIG_EXPERIMENTAL=y
+CONFIG_MODULES=y
+CONFIG_MODVERSIONS=y
+CONFIG_KERNELD=y
+CONFIG_PCI=y
+CONFIG_PCI_OPTIMIZE=y
+CONFIG_NET=y
+CONFIG_SYSCTL=y
+CONFIG_SYSVIPC=y
+CONFIG_BINFMT_ELF=y
+CONFIG_KERNEL_ELF=y
+# CONFIG_BINFMT_MISC is not set
+# CONFIG_BINFMT_JAVA is not set
+CONFIG_VGA_CONSOLE=y
+
+#
+# Plug and Play support
+#
+# CONFIG_PNP is not set
+
+#
+# Floppy, IDE, and other block devices
+#
+CONFIG_BLK_DEV_FD=y
+CONFIG_BLK_DEV_IDE=y
+# CONFIG_BLK_DEV_HD_IDE is not set
+CONFIG_BLK_DEV_IDEDISK=y
+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
+# CONFIG_BLK_DEV_CMD640 is not set
+# CONFIG_BLK_DEV_RZ1000 is not set
+# CONFIG_BLK_DEV_TRITON is not set
+# CONFIG_IDE_CHIPSETS is not set
+CONFIG_BLK_DEV_LOOP=y
+# CONFIG_BLK_DEV_MD is not set
+CONFIG_BLK_DEV_RAM=y
+CONFIG_BLK_DEV_INITRD=y
+# CONFIG_BLK_DEV_XD is not set
+# CONFIG_BLK_DEV_EZ is not set
+# CONFIG_BLK_DEV_HD is not set
+
+#
+# SCSI support
+#
+CONFIG_SCSI=y
+CONFIG_BLK_DEV_SD=y
+CONFIG_CHR_DEV_ST=y
+CONFIG_BLK_DEV_SR=y
+CONFIG_BLK_DEV_SR_VENDOR=y
+# CONFIG_CHR_DEV_SG is not set
+# CONFIG_SCSI_MULTI_LUN is not set
+CONFIG_SCSI_CONSTANTS=y
+
+#
+# SCSI low-level drivers
+#
+# CONFIG_SCSI_7000FASST 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_BUSLOGIC is not set
+# CONFIG_SCSI_DTC3280 is not set
+# CONFIG_SCSI_EATA_DMA is not set
+# CONFIG_SCSI_EATA_PIO is not set
+# CONFIG_SCSI_EATA is not set
+# CONFIG_SCSI_FUTURE_DOMAIN is not set
+# CONFIG_SCSI_GENERIC_NCR5380 is not set
+# CONFIG_SCSI_NCR53C406A is not set
+# CONFIG_SCSI_NCR53C7xx is not set
+CONFIG_SCSI_NCR53C8XX=y
+# CONFIG_SCSI_NCR53C8XX_NVRAM_DETECT is not set
+CONFIG_SCSI_NCR53C8XX_TAGGED_QUEUE=y
+CONFIG_SCSI_NCR53C8XX_IOMAPPED=y
+CONFIG_SCSI_NCR53C8XX_MAX_TAGS=4
+CONFIG_SCSI_NCR53C8XX_SYNC=5
+# CONFIG_SCSI_NCR53C8XX_SYMBIOS_COMPAT is not set
+# CONFIG_SCSI_PPA is not set
+# CONFIG_SCSI_PAS16 is not set
+# CONFIG_SCSI_QLOGIC_FAS is not set
+# CONFIG_SCSI_QLOGIC_ISP is not set
+# CONFIG_SCSI_SEAGATE is not set
+# CONFIG_SCSI_DC390T is not set
+# CONFIG_SCSI_T128 is not set
+# CONFIG_SCSI_U14_34F is not set
+# CONFIG_SCSI_ULTRASTOR is not set
+# CONFIG_SCSI_MESH is not set
+# CONFIG_SCSI_MAC53C94 is not set
+
+#
+# Network device support
+#
+
+#
+# Networking options
+#
+# CONFIG_NETLINK is not set
+# CONFIG_FIREWALL is not set
+# CONFIG_NET_ALIAS is not set
+CONFIG_INET=y
+# CONFIG_IP_MULTICAST is not set
+# CONFIG_IP_ACCT is not set
+# CONFIG_IP_ROUTER is not set
+# CONFIG_NET_IPIP is not set
+# CONFIG_SYN_COOKIES is not set
+# CONFIG_XTP is not set
+# CONFIG_INET_PCTCP is not set
+# CONFIG_INET_RARP is not set
+CONFIG_PATH_MTU_DISCOVERY=y
+# CONFIG_IP_NOSR is not set
+CONFIG_SKB_LARGE=y
+# CONFIG_IPV6 is not set
+# CONFIG_IPX is not set
+# CONFIG_ATALK is not set
+# CONFIG_AX25 is not set
+# CONFIG_X25 is not set
+# CONFIG_LAPB is not set
+# CONFIG_BRIDGE is not set
+# CONFIG_LLC is not set
+# CONFIG_WAN_ROUTER is not set
+CONFIG_NETDEVICES=y
+# CONFIG_ARCNET is not set
+# CONFIG_DUMMY is not set
+# CONFIG_EQUALIZER is not set
+CONFIG_NET_ETHERNET=y
+CONFIG_NET_VENDOR_3COM=y
+# CONFIG_EL1 is not set
+# CONFIG_EL2 is not set
+# CONFIG_ELPLUS is not set
+# CONFIG_EL16 is not set
+CONFIG_EL3=y
+# CONFIG_VORTEX is not set
+CONFIG_LANCE=y
+# CONFIG_NET_VENDOR_SMC is not set
+# CONFIG_NET_VENDOR_RACAL is not set
+# CONFIG_NET_ISA is not set
+CONFIG_NET_EISA=y
+CONFIG_PCNET32=y
+# CONFIG_AC3200 is not set
+# CONFIG_APRICOT is not set
+# CONFIG_CS89x0 is not set
+CONFIG_DE4X5=y
+# CONFIG_DEC_ELCP is not set
+# CONFIG_DGRS is not set
+# CONFIG_EEXPRESS_PRO100 is not set
+# CONFIG_TLAN is not set
+# CONFIG_ES3210 is not set
+# CONFIG_ZNET is not set
+# CONFIG_NET_POCKET is not set
+# CONFIG_FDDI is not set
+# CONFIG_DLCI is not set
+# CONFIG_PLIP is not set
+CONFIG_PPP=y
+# CONFIG_NET_RADIO is not set
+# CONFIG_SLIP is not set
+# CONFIG_TR is not set
+# CONFIG_SHAPER is not set
+
+#
+# ISDN subsystem
+#
+# CONFIG_ISDN is not set
+
+#
+# CD-ROM drivers (not for SCSI or IDE/ATAPI drives)
+#
+# CONFIG_CD_NO_IDESCSI is not set
+
+#
+# Filesystems
+#
+# CONFIG_QUOTA is not set
+# CONFIG_MINIX_FS is not set
+CONFIG_EXT2_FS=y
+CONFIG_FAT_FS=y
+CONFIG_MSDOS_FS=y
+# CONFIG_VFAT_FS is not set
+# CONFIG_UMSDOS_FS is not set
+CONFIG_PROC_FS=y
+CONFIG_NFS_FS=y
+# CONFIG_ROOT_NFS is not set
+CONFIG_NFSD=y
+CONFIG_SUNRPC=y
+CONFIG_LOCKD=y
+# CONFIG_SMB_FS is not set
+CONFIG_ISO9660_FS=y
+# CONFIG_HPFS_FS is not set
+# CONFIG_SYSV_FS is not set
+# CONFIG_AFFS_FS is not set
+# CONFIG_ROMFS_FS is not set
+# CONFIG_AUTOFS_FS is not set
+# CONFIG_UFS_FS is not set
+CONFIG_MAC_PARTITION=y
+
+#
+# Character devices
+#
+CONFIG_VT=y
+CONFIG_VT_CONSOLE=y
+# CONFIG_SOFTCURSOR is not set
+CONFIG_SERIAL=y
+CONFIG_SERIAL_EXTENDED=y
+# CONFIG_SERIAL_MANY_PORTS is not set
+# CONFIG_SERIAL_SHARE_IRQ is not set
+# CONFIG_SERIAL_MULTIPORT is not set
+# CONFIG_HUB6 is not set
+CONFIG_SERIAL_CONSOLE=y
+# CONFIG_SERIAL_NONSTANDARD is not set
+# CONFIG_PRINTER is not set
+CONFIG_MOUSE=y
+# CONFIG_ATIXL_BUSMOUSE is not set
+# CONFIG_BUSMOUSE is not set
+# CONFIG_MS_BUSMOUSE is not set
+CONFIG_PSMOUSE=y
+# CONFIG_82C710_MOUSE is not set
+# CONFIG_PC110_PAD is not set
+# CONFIG_UMISC is not set
+# CONFIG_QIC02_TAPE is not set
+# CONFIG_FTAPE is not set
+# CONFIG_APM is not set
+# CONFIG_WATCHDOG is not set
+# CONFIG_RTC is not set
+# CONFIG_NVRAM is not set
+# CONFIG_JOYSTICK is not set
+
+#
+# Sound
+#
+# CONFIG_SOUND is not set
diff --git a/arch/ppc/ignore b/arch/ppc/ignore
new file mode 100644
index 000000000..693fff61a
--- /dev/null
+++ b/arch/ppc/ignore
@@ -0,0 +1,65 @@
+.config
+compile.h
+.version
+.objects
+.blurb
+*.cort
+*.paul
+*.old
+.defines
+version.h
+find_name
+checks
+#*
+.objects
+.object_files
+System.map
+asm
+.menuconfig*
+CVS
+ppc_defs.h
+mk_defs
+mkprep
+*.s
+.depend
+.hdepend
+*~
+*.o
+znetboot
+zvmlinux
+vmlinux
+zImage
+hack-coff
+coffboot
+vmlinux.coff
+.depend
+.cvsignore
+RCS
+SCCS
+CVS.adm
+RCSLOG
+cvslog.*
+tags
+TAGS
+.make.state
+.nse_depinfo
+*~
+#*
+.#*
+,*
+_$*
+*$
+*.old
+*.bak
+*.BAK
+*.orig
+*.rej
+*.a
+*.olb
+*.o
+*.obj
+*.so
+*.exe
+*.Z
+*.elc
+*.ln
diff --git a/arch/ppc/kernel/Makefile b/arch/ppc/kernel/Makefile
index 941cd9e38..6fd50a6fb 100644
--- a/arch/ppc/kernel/Makefile
+++ b/arch/ppc/kernel/Makefile
@@ -7,39 +7,21 @@
 #
 # Note 2! The CFLAGS definitions are now in the main makefile...
 
-.c.s:
-	$(CC) $(CFLAGS) -S $<
-.s.o:
-	$(AS) $(ASFLAGS) -o $*.o $<
-.c.o:
-	$(CC) $(CFLAGS) -c $<
-.S.s:
-	$(CPP) $(CFLAGS) -D__ASSEMBLY__ $< -o $*.s
 .S.o:
-	$(CPP) $(CFLAGS) -D__ASSEMBLY__ $< -o $*.s
-	$(AS) $(ASFLAGS) -o $*.o $*.s
-	rm $*.s
+	$(CC) -D__ASSEMBLY__ -c $< -o $*.o
 
-HOST_CC = gcc
-
-OBJS  = misc.o port_io.o pci.o traps.o process.o \
-	signal.o syscalls.o ptrace.o ksyms.o irq.o bitops.o strcase.o ppc_htab.o
+O_TARGET := kernel.o
+O_OBJS := misc.o traps.o process.o signal.o syscalls.o \
+	  align.o ptrace.o irq.o openpic.o bitops.o ppc_htab.o idle.o \
+	  time.o prep_time.o pmac_time.o chrp_time.o \
+	  setup.o prep_setup.o pmac_setup.o pmac_support.o chrp_setup.o \
+	  pci.o prep_pci.o pmac_pci.o chrp_pci.o \
+	  residual.o prom.o
+OX_OBJS := ppc_ksyms.o
 
 all: head.o kernel.o
-head.o: head.S $(TOPDIR)/include/linux/tasks.h ppc_defs.h
-
-ifeq ($(CONFIG_PREP),y)
-OBJS += prep_setup.o prep_time.o
-endif
-
-ifeq ($(CONFIG_PMAC),y)
-OBJS += pmac_setup.o pmac_support.o align.o pmac_time.o
-endif
-
-ifeq ($(CONFIG_MODULES),y)
-OBJS = ksyms.o
-endif
 
+head.o: head.S $(TOPDIR)/include/linux/tasks.h ppc_defs.h
 
 ppc_defs.h: mk_defs.c ppc_defs.head \
 		$(TOPDIR)/include/asm/mmu.h \
@@ -52,25 +34,7 @@ ppc_defs.h: mk_defs.c ppc_defs.head \
 	rm mk_defs.s
 
 checks: checks.c
-	$(HOSTCC) ${CFLAGS} -o checks checks.c
-	checks
-
-kernel.o: $(OBJS)
-	$(LD) -r -o kernel.o $(OBJS)
+	$(HOSTCC) -fno-builtin -I$(TOPDIR)/include -D__KERNEL__ -o checks checks.c
+	./checks
 
-fastdep:
-	$(TOPDIR)/scripts/mkdep *.[Sch] > .depend
-
-dep:
-	$(CPP) -M *.S *.c > .depend
-
-modules: 
-
-dummy:
-
-#
-# include a dependency file if one exists
-#
-ifeq (.depend,$(wildcard .depend))
-include .depend
-endif
+include $(TOPDIR)/Rules.make
diff --git a/arch/ppc/kernel/align.c b/arch/ppc/kernel/align.c
new file mode 100644
index 000000000..39cb04f77
--- /dev/null
+++ b/arch/ppc/kernel/align.c
@@ -0,0 +1,290 @@
+/*
+ * align.c - handle alignment exceptions for the Power PC.
+ *
+ * Paul Mackerras	August 1996.
+ * Copyright (C) 1996 Paul Mackerras	(paulus@cs.anu.edu.au).
+ */
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <asm/ptrace.h>
+#include <asm/processor.h>
+#include <asm/uaccess.h>
+#include <asm/system.h>
+
+struct aligninfo {
+	unsigned char len;
+	unsigned char flags;
+};
+
+#define INVALID	{ 0, 0 }
+
+#define LD	1	/* load */
+#define ST	2	/* store */
+#define	SE	4	/* sign-extend value */
+#define F	8	/* to/from fp regs */
+#define U	0x10	/* update index register */
+#define M	0x20	/* multiple load/store */
+#define S	0x40	/* single-precision fp, or byte-swap value */
+#define HARD	0x80	/* string, stwcx. */
+
+/*
+ * The PowerPC stores certain bits of the instruction that caused the
+ * alignment exception in the DSISR register.  This array maps those
+ * bits to information about the operand length and what the
+ * instruction would do.
+ */
+static struct aligninfo aligninfo[128] = {
+	{ 4, LD },		/* 00 0 0000: lwz / lwarx */
+	INVALID,		/* 00 0 0001 */
+	{ 4, ST },		/* 00 0 0010: stw */
+	INVALID,		/* 00 0 0011 */
+	{ 2, LD },		/* 00 0 0100: lhz */
+	{ 2, LD+SE },		/* 00 0 0101: lha */
+	{ 2, ST },		/* 00 0 0110: sth */
+	{ 4, LD+M },		/* 00 0 0111: lmw */
+	{ 4, LD+F+S },		/* 00 0 1000: lfs */
+	{ 8, LD+F },		/* 00 0 1001: lfd */
+	{ 4, ST+F+S },		/* 00 0 1010: stfs */
+	{ 8, ST+F },		/* 00 0 1011: stfd */
+	INVALID,		/* 00 0 1100 */
+	INVALID,		/* 00 0 1101 */
+	INVALID,		/* 00 0 1110 */
+	INVALID,		/* 00 0 1111 */
+	{ 4, LD+U },		/* 00 1 0000: lwzu */
+	INVALID,		/* 00 1 0001 */
+	{ 4, ST+U },		/* 00 1 0010: stwu */
+	INVALID,		/* 00 1 0011 */
+	{ 2, LD+U },		/* 00 1 0100: lhzu */
+	{ 2, LD+SE+U },		/* 00 1 0101: lhau */
+	{ 2, ST+U },		/* 00 1 0110: sthu */
+	{ 4, ST+M },		/* 00 1 0111: stmw */
+	{ 4, LD+F+S+U },	/* 00 1 1000: lfsu */
+	{ 8, LD+F+U },		/* 00 1 1001: lfdu */
+	{ 4, ST+F+S+U },	/* 00 1 1010: stfsu */
+	{ 8, ST+F+U },		/* 00 1 1011: stfdu */
+	INVALID,		/* 00 1 1100 */
+	INVALID,		/* 00 1 1101 */
+	INVALID,		/* 00 1 1110 */
+	INVALID,		/* 00 1 1111 */
+	INVALID,		/* 01 0 0000 */
+	INVALID,		/* 01 0 0001 */
+	INVALID,		/* 01 0 0010 */
+	INVALID,		/* 01 0 0011 */
+	INVALID,		/* 01 0 0100 */
+	INVALID,		/* 01 0 0101: lwax?? */
+	INVALID,		/* 01 0 0110 */
+	INVALID,		/* 01 0 0111 */
+	{ 0, LD+HARD },		/* 01 0 1000: lswx */
+	{ 0, LD+HARD },		/* 01 0 1001: lswi */
+	{ 0, ST+HARD },		/* 01 0 1010: stswx */
+	{ 0, ST+HARD },		/* 01 0 1011: stswi */
+	INVALID,		/* 01 0 1100 */
+	INVALID,		/* 01 0 1101 */
+	INVALID,		/* 01 0 1110 */
+	INVALID,		/* 01 0 1111 */
+	INVALID,		/* 01 1 0000 */
+	INVALID,		/* 01 1 0001 */
+	INVALID,		/* 01 1 0010 */
+	INVALID,		/* 01 1 0011 */
+	INVALID,		/* 01 1 0100 */
+	INVALID,		/* 01 1 0101: lwaux?? */
+	INVALID,		/* 01 1 0110 */
+	INVALID,		/* 01 1 0111 */
+	INVALID,		/* 01 1 1000 */
+	INVALID,		/* 01 1 1001 */
+	INVALID,		/* 01 1 1010 */
+	INVALID,		/* 01 1 1011 */
+	INVALID,		/* 01 1 1100 */
+	INVALID,		/* 01 1 1101 */
+	INVALID,		/* 01 1 1110 */
+	INVALID,		/* 01 1 1111 */
+	INVALID,		/* 10 0 0000 */
+	INVALID,		/* 10 0 0001 */
+	{ 0, ST+HARD },		/* 10 0 0010: stwcx. */
+	INVALID,		/* 10 0 0011 */
+	INVALID,		/* 10 0 0100 */
+	INVALID,		/* 10 0 0101 */
+	INVALID,		/* 10 0 0110 */
+	INVALID,		/* 10 0 0111 */
+	{ 4, LD+S },		/* 10 0 1000: lwbrx */
+	INVALID,		/* 10 0 1001 */
+	{ 4, ST+S },		/* 10 0 1010: stwbrx */
+	INVALID,		/* 10 0 1011 */
+	{ 2, LD+S },		/* 10 0 1100: lhbrx */
+	INVALID,		/* 10 0 1101 */
+	{ 2, ST+S },		/* 10 0 1110: sthbrx */
+	INVALID,		/* 10 0 1111 */
+	INVALID,		/* 10 1 0000 */
+	INVALID,		/* 10 1 0001 */
+	INVALID,		/* 10 1 0010 */
+	INVALID,		/* 10 1 0011 */
+	INVALID,		/* 10 1 0100 */
+	INVALID,		/* 10 1 0101 */
+	INVALID,		/* 10 1 0110 */
+	INVALID,		/* 10 1 0111 */
+	INVALID,		/* 10 1 1000 */
+	INVALID,		/* 10 1 1001 */
+	INVALID,		/* 10 1 1010 */
+	INVALID,		/* 10 1 1011 */
+	INVALID,		/* 10 1 1100 */
+	INVALID,		/* 10 1 1101 */
+	INVALID,		/* 10 1 1110 */
+	{ 0, ST+HARD },		/* 10 1 1111: dcbz */
+	{ 4, LD },		/* 11 0 0000: lwzx */
+	INVALID,		/* 11 0 0001 */
+	{ 4, ST },		/* 11 0 0010: stwx */
+	INVALID,		/* 11 0 0011 */
+	{ 2, LD },		/* 11 0 0100: lhzx */
+	{ 2, LD+SE },		/* 11 0 0101: lhax */
+	{ 2, ST },		/* 11 0 0110: sthx */
+	INVALID,		/* 11 0 0111 */
+	{ 4, LD+F+S },		/* 11 0 1000: lfsx */
+	{ 8, LD+F },		/* 11 0 1001: lfdx */
+	{ 4, ST+F+S },		/* 11 0 1010: stfsx */
+	{ 8, ST+F },		/* 11 0 1011: stfdx */
+	INVALID,		/* 11 0 1100 */
+	INVALID,		/* 11 0 1101 */
+	INVALID,		/* 11 0 1110 */
+	INVALID,		/* 11 0 1111 */
+	{ 4, LD+U },		/* 11 1 0000: lwzux */
+	INVALID,		/* 11 1 0001 */
+	{ 4, ST+U },		/* 11 1 0010: stwux */
+	INVALID,		/* 11 1 0011 */
+	{ 2, LD+U },		/* 11 1 0100: lhzux */
+	{ 2, LD+SE+U },		/* 11 1 0101: lhaux */
+	{ 2, ST+U },		/* 11 1 0110: sthux */
+	INVALID,		/* 11 1 0111 */
+	{ 4, LD+F+S+U },	/* 11 1 1000: lfsux */
+	{ 8, LD+F+U },		/* 11 1 1001: lfdux */
+	{ 4, ST+F+S+U },	/* 11 1 1010: stfsux */
+	{ 8, ST+F+U },		/* 11 1 1011: stfdux */
+	INVALID,		/* 11 1 1100 */
+	INVALID,		/* 11 1 1101 */
+	INVALID,		/* 11 1 1110 */
+	INVALID,		/* 11 1 1111 */
+};
+
+#define SWAP(a, b)	(t = (a), (a) = (b), (b) = t)
+
+int
+fix_alignment(struct pt_regs *regs)
+{
+	int instr, nb, flags;
+	int i, t;
+	int reg, areg;
+	unsigned char *addr;
+	union {
+		long l;
+		float f;
+		double d;
+		unsigned char v[8];
+	} data;
+
+	instr = (regs->dsisr >> 10) & 0x7f;
+	nb = aligninfo[instr].len;
+	if (nb == 0)
+		return 0;	/* too hard or invalid instruction bits */
+	flags = aligninfo[instr].flags;
+	addr = (unsigned char *) regs->dar;
+	reg = (regs->dsisr >> 5) & 0x1f;	/* source/dest register */
+
+	/* Verify the address of the operand */
+	if (user_mode(regs)) {
+		if (verify_area((flags & ST? VERIFY_WRITE: VERIFY_READ), addr, nb))
+			return -EFAULT;	/* bad address */
+	}
+
+	if ((flags & F) && last_task_used_math == current)
+		giveup_fpu();
+
+	if (flags & M)
+		return 0;		/* too hard for now */
+
+	/* If we read the operand, copy it in */
+	if (flags & LD) {
+		if (nb == 2) {
+			data.v[0] = data.v[1] = 0;
+			if (__get_user(data.v[2], addr)
+			    || __get_user(data.v[3], addr+1))
+				return -EFAULT;
+		} else {
+			for (i = 0; i < nb; ++i)
+				if (__get_user(data.v[i], addr+i))
+					return -EFAULT;
+		}
+	}
+
+	switch (flags & ~U) {
+	case LD+SE:
+		if (data.v[2] >= 0x80)
+			data.v[0] = data.v[1] = -1;
+		/* fall through */
+	case LD:
+		regs->gpr[reg] = data.l;
+		break;
+	case LD+S:
+		if (nb == 2) {
+			SWAP(data.v[2], data.v[3]);
+		} else {
+			SWAP(data.v[0], data.v[3]);
+			SWAP(data.v[1], data.v[2]);
+		}
+		regs->gpr[reg] = data.l;
+		break;
+	case ST:
+		data.l = regs->gpr[reg];
+		break;
+	case ST+S:
+		data.l = regs->gpr[reg];
+		if (nb == 2) {
+			SWAP(data.v[2], data.v[3]);
+		} else {
+			SWAP(data.v[0], data.v[3]);
+			SWAP(data.v[1], data.v[2]);
+		}
+		break;
+	case LD+F:
+		current->tss.fpr[reg] = data.d;
+		break;
+	case ST+F:
+		data.d = current->tss.fpr[reg];
+		break;
+	/* these require some floating point conversions... */
+	/* note that giveup_fpu enables the FPU for the kernel */
+	/* we'd like to use the assignment, but we have to compile
+	 * the kernel with -msoft-float so it doesn't use the
+	 * fp regs for copying 8-byte objects. */
+	case LD+F+S:
+		giveup_fpu();
+		cvt_fd(&data.f, &current->tss.fpr[reg]);
+		/* current->tss.fpr[reg] = data.f; */
+		break;
+	case ST+F+S:
+		giveup_fpu();
+		cvt_df(&current->tss.fpr[reg], &data.f);
+		/* data.f = current->tss.fpr[reg]; */
+		break;
+	default:
+		printk("align: can't handle flags=%x\n", flags);
+		return 0;
+	}
+
+	if (flags & ST) {
+		if (nb == 2) {
+			if (__put_user(data.v[2], addr)
+			    || __put_user(data.v[3], addr+1))
+				return -EFAULT;
+		} else {
+			for (i = 0; i < nb; ++i)
+				if (__put_user(data.v[i], addr+i))
+					return -EFAULT;
+		}
+	}
+
+	if (flags & U) {
+		areg = regs->dsisr & 0x1f;	/* register to update */
+		regs->gpr[areg] = regs->dar;
+	}
+
+	return 1;
+}
diff --git a/arch/ppc/kernel/bitops.c b/arch/ppc/kernel/bitops.c
new file mode 100644
index 000000000..fb5a19e3a
--- /dev/null
+++ b/arch/ppc/kernel/bitops.c
@@ -0,0 +1,201 @@
+/*
+ * Copyright (C) 1996 Paul Mackerras.
+ */
+
+#include <linux/kernel.h>
+#include <asm/bitops.h>
+
+/*
+ * I left these here since the problems with "cc" make it difficult to keep
+ * them in bitops.h -- Cort
+ */
+void set_bit(int nr, volatile void *addr)
+{
+	unsigned int t;
+	unsigned int mask = 1 << (nr & 0x1f);
+	volatile unsigned int *p = ((volatile unsigned int *)addr) + (nr >> 5);
+
+	if ((unsigned long)addr & 3)
+		printk(KERN_ERR "set_bit(%x, %p)\n", nr, addr);
+	__asm__ __volatile__("\n\
+1:	lwarx	%0,0,%2
+	or	%0,%0,%1
+	stwcx.	%0,0,%2
+	bne	1b"
+	: "=&r" (t)		/*, "=m" (*p)*/
+	: "r" (mask), "r" (p)
+	: "cc");
+}
+
+void clear_bit(int nr, volatile void *addr)
+{
+	unsigned int t;
+	unsigned int mask = 1 << (nr & 0x1f);
+	volatile unsigned int *p = ((volatile unsigned int *)addr) + (nr >> 5);
+
+	if ((unsigned long)addr & 3)
+		printk(KERN_ERR "clear_bit(%x, %p)\n", nr, addr);
+	__asm__ __volatile__("\n\
+1:	lwarx	%0,0,%2
+	andc	%0,%0,%1
+	stwcx.	%0,0,%2
+	bne	1b"
+	: "=&r" (t)		/*, "=m" (*p)*/
+	: "r" (mask), "r" (p)
+	: "cc");
+}
+
+void change_bit(int nr, volatile void *addr)
+{
+	unsigned int t;
+	unsigned int mask = 1 << (nr & 0x1f);
+	volatile unsigned int *p = ((volatile unsigned int *)addr) + (nr >> 5);
+
+	if ((unsigned long)addr & 3)
+		printk(KERN_ERR "change_bit(%x, %p)\n", nr, addr);
+	__asm__ __volatile__("\n\
+1:	lwarx	%0,0,%2
+	xor	%0,%0,%1
+	stwcx.	%0,0,%2
+	bne	1b"
+	: "=&r" (t)		/*, "=m" (*p)*/
+	: "r" (mask), "r" (p)
+	: "cc");
+}
+
+int test_and_set_bit(int nr, volatile void *addr)
+{
+	unsigned int old, t;
+	unsigned int mask = 1 << (nr & 0x1f);
+	volatile unsigned int *p = ((volatile unsigned int *)addr) + (nr >> 5);
+
+	if ((unsigned long)addr & 3)
+		printk(KERN_ERR "test_and_set_bit(%x, %p)\n", nr, addr);
+	__asm__ __volatile__("\n\
+1:	lwarx	%0,0,%3
+	or	%1,%0,%2
+	stwcx.	%1,0,%3
+	bne	1b"
+	: "=&r" (old), "=&r" (t)	/*, "=m" (*p)*/
+	: "r" (mask), "r" (p)
+	: "cc");
+
+	return (old & mask) != 0;
+}
+
+int test_and_clear_bit(int nr, volatile void *addr)
+{
+	unsigned int old, t;
+	unsigned int mask = 1 << (nr & 0x1f);
+	volatile unsigned int *p = ((volatile unsigned int *)addr) + (nr >> 5);
+
+	if ((unsigned long)addr & 3)
+		printk(KERN_ERR "test_and_clear_bit(%x, %p)\n", nr, addr);
+	__asm__ __volatile__("\n\
+1:	lwarx	%0,0,%3
+	andc	%1,%0,%2
+	stwcx.	%1,0,%3
+	bne	1b"
+	: "=&r" (old), "=&r" (t)	/*, "=m" (*p)*/
+	: "r" (mask), "r" (p)
+	: "cc");
+
+	return (old & mask) != 0;
+}
+
+int test_and_change_bit(int nr, volatile void *addr)
+{
+	unsigned int old, t;
+	unsigned int mask = 1 << (nr & 0x1f);
+	volatile unsigned int *p = ((volatile unsigned int *)addr) + (nr >> 5);
+
+	if ((unsigned long)addr & 3)
+		printk(KERN_ERR "test_and_change_bit(%x, %p)\n", nr, addr);
+	__asm__ __volatile__("\n\
+1:	lwarx	%0,0,%3
+	xor	%1,%0,%2
+	stwcx.	%1,0,%3
+	bne	1b"
+	: "=&r" (old), "=&r" (t)	/*, "=m" (*p)*/
+	: "r" (mask), "r" (p)
+	: "cc");
+
+	return (old & mask) != 0;
+}
+
+/* I put it in bitops.h -- Cort */
+#if 0
+int ffz(unsigned int x)
+{
+	int n;
+
+	x = ~x & (x+1);		/* set LS zero to 1, other bits to 0 */
+	__asm__ ("cntlzw %0,%1" : "=r" (n) : "r" (x));
+	return 31 - n;
+}
+
+/*
+ * This implementation of find_{first,next}_zero_bit was stolen from
+ * Linus' asm-alpha/bitops.h.
+ */
+
+int find_first_zero_bit(void * addr, int size)
+{
+	unsigned int * p = ((unsigned int *) addr);
+	unsigned int result = 0;
+	unsigned int tmp;
+
+	if (size == 0)
+		return 0;
+	while (size & ~31UL) {
+		if (~(tmp = *(p++)))
+			goto found_middle;
+		result += 32;
+		size -= 32;
+	}
+	if (!size)
+		return result;
+	tmp = *p;
+	tmp |= ~0UL << size;
+found_middle:
+	return result + ffz(tmp);
+}
+
+/*
+ * Find next zero bit in a bitmap reasonably efficiently..
+ */
+int find_next_zero_bit(void * addr, int size, int 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++);
+		tmp |= ~0UL >> (32-offset);
+		if (size < 32)
+			goto found_first;
+		if (~tmp)
+			goto found_middle;
+		size -= 32;
+		result += 32;
+	}
+	while (size & ~31UL) {
+		if (~(tmp = *(p++)))
+			goto found_middle;
+		result += 32;
+		size -= 32;
+	}
+	if (!size)
+		return result;
+	tmp = *p;
+found_first:
+	tmp |= ~0UL << size;
+found_middle:
+	return result + ffz(tmp);
+}
+#endif
diff --git a/arch/ppc/kernel/checks.c b/arch/ppc/kernel/checks.c
new file mode 100644
index 000000000..312229552
--- /dev/null
+++ b/arch/ppc/kernel/checks.c
@@ -0,0 +1,57 @@
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/ptrace.h>
+#include <linux/malloc.h>
+#include <linux/user.h>
+#include <linux/a.out.h>
+#include <linux/config.h>
+
+#include <asm/pgtable.h>
+#include <asm/uaccess.h>
+#include <asm/system.h>
+#include <asm/io.h>
+#include <asm/smp_lock.h>
+
+/*
+ * Do various before compile checks of data structures
+ * -- Cort
+ */
+int main(void)
+{
+	int ret = 0;
+#if 0
+	if ( sizeof(struct thread_struct) % 16 )
+	{
+		printf("Thread struct is not modulo 16 bytes: "
+			"%d bytes total, %d bytes off\n",
+			sizeof(struct thread_struct),
+			sizeof(struct thread_struct)%16);
+		ret = -1;
+	}
+#endif	
+
+	if ( sizeof(struct pt_regs) % 16 )
+	{
+		printf("pt_regs struct is not modulo 16 bytes: "
+			"%d bytes total, %d bytes off\n",
+			sizeof(struct pt_regs),
+			sizeof(struct pt_regs)%16);
+		ret = -1;
+		
+	}
+
+	printf("Task size        : %d bytes\n"
+	       "Tss size         : %d bytes\n"
+	       "pt_regs size     : %d bytes\n"
+	       "Kernel stack size: %d bytes\n",
+	       sizeof(struct task_struct), sizeof(struct thread_struct),
+	       sizeof(struct pt_regs),
+	       sizeof(union task_union) - sizeof(struct task_struct));
+	return ret;
+}
diff --git a/arch/ppc/kernel/chrp_pci.c b/arch/ppc/kernel/chrp_pci.c
new file mode 100644
index 000000000..8af9ba9d6
--- /dev/null
+++ b/arch/ppc/kernel/chrp_pci.c
@@ -0,0 +1,156 @@
+/*
+ * CHRP pci routines.
+ */
+
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/bios32.h>
+#include <linux/delay.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <linux/openpic.h>
+
+#include <asm/io.h>
+#include <asm/irq.h>
+#include <asm/hydra.h>
+
+/* LongTrail */
+#define pci_config_addr(bus, dev, offset) \
+	(0xfec00000 | ((bus)<<16) | ((dev)<<8) | (offset))
+
+int chrp_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));
+    if (offset == PCI_INTERRUPT_LINE) {
+	/* PCI interrupts are controlled by the OpenPIC */
+	if (*val)
+	    *val = openpic_to_irq(*val);
+    }
+    return PCIBIOS_SUCCESSFUL;
+}
+
+int chrp_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_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_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_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;
+}
+
+int chrp_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn,
+				    unsigned char offset, unsigned int val)
+{
+    if (bus > 7)
+	return PCIBIOS_DEVICE_NOT_FOUND;
+    out_le32((unsigned int *)pci_config_addr(bus, dev_fn, offset), val);
+    return PCIBIOS_SUCCESSFUL;
+}
+
+int chrp_pcibios_find_device(unsigned short vendor, unsigned short dev_id,
+			     unsigned short index, unsigned char *bus_ptr,
+			     unsigned char *dev_fn_ptr)
+{
+    int num, devfn;
+    unsigned int x, vendev;
+
+    if (vendor == 0xffff)
+	return PCIBIOS_BAD_VENDOR_ID;
+    vendev = (dev_id << 16) + vendor;
+    num = 0;
+    for (devfn = 0;  devfn < 32;  devfn++) {
+	chrp_pcibios_read_config_dword(0, devfn<<3, PCI_VENDOR_ID, &x);
+	if (x == vendev) {
+	    if (index == num) {
+		*bus_ptr = 0;
+		*dev_fn_ptr = devfn<<3;
+		return PCIBIOS_SUCCESSFUL;
+	    }
+	    ++num;
+	}
+    }
+    return PCIBIOS_DEVICE_NOT_FOUND;
+}
+
+int chrp_pcibios_find_class(unsigned int class_code, unsigned short index,
+			    unsigned char *bus_ptr, unsigned char *dev_fn_ptr)
+{
+    int devnr, x, num;
+
+    num = 0;
+    for (devnr = 0;  devnr < 32;  devnr++) {
+	chrp_pcibios_read_config_dword(0, devnr<<3, PCI_CLASS_REVISION, &x);
+	if ((x>>8) == class_code) {
+	    if (index == num) {
+		*bus_ptr = 0;
+		*dev_fn_ptr = devnr<<3;
+		return PCIBIOS_SUCCESSFUL;
+	    }
+	    ++num;
+	}
+    }
+    return PCIBIOS_DEVICE_NOT_FOUND;
+}
+
+__initfunc(volatile struct Hydra *find_hydra(void))
+{
+    u_char bus, dev;
+    volatile struct Hydra *hydra = 0;
+    if (chrp_pcibios_find_device(PCI_VENDOR_ID_APPLE,
+				 PCI_DEVICE_ID_APPLE_HYDRA, 0, &bus, &dev)
+	== PCIBIOS_SUCCESSFUL)
+	chrp_pcibios_read_config_dword(bus, dev, PCI_BASE_ADDRESS_0,
+				       (unsigned int *)&hydra);
+    return hydra;
+}
+
+__initfunc(void hydra_post_openpic_init(void))
+{
+    openpic_set_sense(HYDRA_INT_SCSI_DMA, 0);
+    openpic_set_sense(HYDRA_INT_SCCA_TX_DMA, 0);
+    openpic_set_sense(HYDRA_INT_SCCA_RX_DMA, 0);
+    openpic_set_sense(HYDRA_INT_SCCB_TX_DMA, 0);
+    openpic_set_sense(HYDRA_INT_SCCB_RX_DMA, 0);
+    openpic_set_sense(HYDRA_INT_SCSI, 1);
+    openpic_set_sense(HYDRA_INT_SCCA, 1);
+    openpic_set_sense(HYDRA_INT_SCCB, 1);
+    openpic_set_sense(HYDRA_INT_VIA, 1);
+    openpic_set_sense(HYDRA_INT_ADB, 1);
+    openpic_set_sense(HYDRA_INT_ADB_NMI, 0);
+}
diff --git a/arch/ppc/kernel/chrp_setup.c b/arch/ppc/kernel/chrp_setup.c
new file mode 100644
index 000000000..0a5544b96
--- /dev/null
+++ b/arch/ppc/kernel/chrp_setup.c
@@ -0,0 +1,180 @@
+/*
+ *  linux/arch/ppc/kernel/setup.c
+ *
+ *  Copyright (C) 1995  Linus Torvalds
+ *  Adapted from 'alpha' version by Gary Thomas
+ *  Modified by Cort Dougan (cort@cs.nmt.edu)
+ */
+
+/*
+ * bootup setup stuff..
+ */
+
+#include <linux/config.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/ptrace.h>
+#include <linux/malloc.h>
+#include <linux/user.h>
+#include <linux/a.out.h>
+#include <linux/tty.h>
+#include <linux/major.h>
+#include <linux/interrupt.h>
+#include <linux/reboot.h>
+#include <linux/init.h>
+#include <linux/blk.h>
+#include <linux/ioport.h>
+
+#include <asm/mmu.h>
+#include <asm/processor.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+
+/* for the mac fs */
+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;
+
+unsigned long empty_zero_page[1024];
+extern unsigned char aux_device_present;
+
+#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
+
+
+extern char saved_command_line[256];
+
+long TotalMemory;
+
+int
+chrp_get_cpuinfo(char *buffer)
+{
+	int pvr = _get_PVR();
+	int len;
+	char *model;
+
+	switch (pvr>>16)
+	{
+	case 1:
+		model = "601";
+		break;
+	case 3:
+		model = "603";
+		break;
+	case 4:
+		model = "604";
+		break;
+	case 6:
+		model = "603e";
+		break;
+	case 7:
+		model = "603ev";
+		break;
+	case 9:
+		model = "604e";
+		break;
+	default:
+		model = "unknown";
+		break;
+	}
+
+	len = sprintf(buffer, "PowerPC %s rev %d.%d\n", model,
+		      (pvr & 0xff00) >> 8, pvr & 0xff);
+
+	len += sprintf(buffer+len, "bogomips\t: %lu.%02lu\n",
+		       (loops_per_sec+2500)/500000,
+		       ((loops_per_sec+2500)/5000) % 100);
+
+#if 0
+	/*
+	 * Ooh's and aah's info about zero'd pages in idle task
+	 */
+	{
+		extern unsigned int zerocount, zerototal, zeropage_hits,zeropage_calls;
+		len += sprintf(buffer+len,"zero pages\t: total %u (%uKb) "
+			       "current: %u (%uKb) hits: %u/%u (%lu%%)\n",
+			       zerototal, (zerototal*PAGE_SIZE)>>10,
+			       zerocount, (zerocount*PAGE_SIZE)>>10,
+			       zeropage_hits,zeropage_calls,
+			       /* : 1 below is so we don't div by zero */
+			       (zeropage_hits*100) /
+				    ((zeropage_calls)?zeropage_calls:1));
+	}
+#endif
+	return len;
+}
+
+__initfunc(void
+chrp_setup_arch(char **cmdline_p, unsigned long * memory_start_p,
+	   unsigned long * memory_end_p))
+{
+	extern char cmd_line[];
+	extern char _etext[], _edata[], _end[];
+	extern int panic_timeout;
+
+	/* Save unparsed command line copy for /proc/cmdline */
+	strcpy( saved_command_line, cmd_line );
+	*cmdline_p = cmd_line;
+
+	*memory_start_p = (unsigned long) Hash+Hash_size;
+	(unsigned long *)*memory_end_p = (unsigned long *)(TotalMemory+KERNELBASE);
+
+	/* init to some ~sane value until calibrate_delay() runs */
+	loops_per_sec = 50000000;
+	
+	/* reboot on panic */	
+	panic_timeout = 180;
+	
+	init_task.mm->start_code = PAGE_OFFSET;
+	init_task.mm->end_code = (unsigned long) _etext;
+	init_task.mm->end_data = (unsigned long) _edata;
+	init_task.mm->brk = (unsigned long) _end;	
+	
+	aux_device_present = 0xaa;
+	
+	switch ( _machine )
+	{
+	case _MACH_chrp:
+		ROOT_DEV = to_kdev_t(0x0801); /* sda1 */
+		break;
+	}
+	
+#ifdef CONFIG_BLK_DEV_RAM
+#if 0
+	ROOT_DEV = to_kdev_t(0x0200); /* floppy */
+	rd_prompt = 1;
+	rd_doload = 1;
+	rd_image_start = 0;
+#endif
+	/* initrd_start and size are setup by boot/head.S and kernel/head.S */
+	if ( initrd_start )
+	{
+		if (initrd_end > *memory_end_p)
+		{
+			printk("initrd extends beyond end of memory "
+			       "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
+			       initrd_end,*memory_end_p);
+			initrd_start = 0;
+		}
+	}
+#endif
+
+	printk("Boot arguments: %s\n", cmd_line);
+	
+	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");
+}
diff --git a/arch/ppc/kernel/chrp_time.c b/arch/ppc/kernel/chrp_time.c
new file mode 100644
index 000000000..bf58953cd
--- /dev/null
+++ b/arch/ppc/kernel/chrp_time.c
@@ -0,0 +1,132 @@
+/*
+ *  linux/arch/i386/kernel/time.c
+ *
+ *  Copyright (C) 1991, 1992, 1995  Linus Torvalds
+ *
+ * Adapted for PowerPC (PreP) by Gary Thomas
+ * Modified by Cort Dougan (cort@cs.nmt.edu)
+ *  copied and modified from intel version
+ *
+ */
+#include <linux/config.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/param.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <linux/interrupt.h>
+#include <linux/timex.h>
+#include <linux/kernel_stat.h>
+#include <linux/mc146818rtc.h>
+
+#include <asm/segment.h>
+#include <asm/io.h>
+#include <asm/processor.h>
+#include <asm/nvram.h>
+
+#include "time.h"
+
+int chrp_cmos_clock_read(int addr)
+{
+	outb(addr>>8, NVRAM_AS1);
+	outb(addr, NVRAM_AS0);
+	return (inb(NVRAM_DATA));
+}
+
+void chrp_cmos_clock_write(unsigned long val, int addr)
+{
+	outb(addr>>8, NVRAM_AS1);
+	outb(addr, NVRAM_AS0);
+	outb(val,NVRAM_DATA);
+	return;
+}
+
+/*
+ * Set the hardware clock. -- Cort
+ */
+int chrp_set_rtc_time(unsigned long nowtime)
+{
+	unsigned char save_control, save_freq_select;
+	struct rtc_time tm;
+
+	to_tm(nowtime, &tm);
+
+	save_control = chrp_cmos_clock_read(RTC_CONTROL); /* tell the clock it's being set */
+
+	chrp_cmos_clock_write((save_control|RTC_SET), RTC_CONTROL);
+
+	save_freq_select = chrp_cmos_clock_read(RTC_FREQ_SELECT); /* stop and reset prescaler */
+	
+	chrp_cmos_clock_write((save_freq_select|RTC_DIV_RESET2), RTC_FREQ_SELECT);
+
+        tm.tm_year -= 1900;
+	if (!(save_control & RTC_DM_BINARY) || RTC_ALWAYS_BCD) {
+		BIN_TO_BCD(tm.tm_sec);
+		BIN_TO_BCD(tm.tm_min);
+		BIN_TO_BCD(tm.tm_hour);
+		BIN_TO_BCD(tm.tm_mon);
+		BIN_TO_BCD(tm.tm_mday);
+		BIN_TO_BCD(tm.tm_year);
+	}
+	chrp_cmos_clock_write(tm.tm_sec,RTC_SECONDS);
+	chrp_cmos_clock_write(tm.tm_min,RTC_MINUTES);
+	chrp_cmos_clock_write(tm.tm_hour,RTC_HOURS);
+	chrp_cmos_clock_write(tm.tm_mon,RTC_MONTH);
+	chrp_cmos_clock_write(tm.tm_mday,RTC_DAY_OF_MONTH);
+	chrp_cmos_clock_write(tm.tm_year,RTC_YEAR);
+	
+	/* The following flags have to be released exactly in this order,
+	 * otherwise the DS12887 (popular MC146818A clone with integrated
+	 * battery and quartz) will not reset the oscillator and will not
+	 * update precisely 500 ms later. You won't find this mentioned in
+	 * the Dallas Semiconductor data sheets, but who believes data
+	 * sheets anyway ...                           -- Markus Kuhn
+	 */
+	chrp_cmos_clock_write(save_control, RTC_CONTROL);
+	chrp_cmos_clock_write(save_freq_select, RTC_FREQ_SELECT);
+
+	if ( (time_state == TIME_ERROR) || (time_state == TIME_BAD) )
+		time_state = TIME_OK;
+	return 0;
+}
+
+unsigned long chrp_get_rtc_time(void)
+{
+	unsigned int year, mon, day, hour, min, sec;
+	int i;
+
+	/* The Linux interpretation of the CMOS clock register contents:
+	 * When the Update-In-Progress (UIP) flag goes from 1 to 0, the
+	 * RTC registers show the second which has precisely just started.
+	 * Let's hope other operating systems interpret the RTC the same way.
+	 */
+	/* read RTC exactly on falling edge of update flag */
+	for (i = 0 ; i < 1000000 ; i++)	/* may take up to 1 second... */
+		if (chrp_cmos_clock_read(RTC_FREQ_SELECT) & RTC_UIP)
+			break;
+	for (i = 0 ; i < 1000000 ; i++)	/* must try at least 2.228 ms */
+		if (!(chrp_cmos_clock_read(RTC_FREQ_SELECT) & RTC_UIP))
+			break;
+	do { /* Isn't this overkill ? UIP above should guarantee consistency */
+		sec = chrp_cmos_clock_read(RTC_SECONDS);
+		min = chrp_cmos_clock_read(RTC_MINUTES);
+		hour = chrp_cmos_clock_read(RTC_HOURS);
+		day = chrp_cmos_clock_read(RTC_DAY_OF_MONTH);
+		mon = chrp_cmos_clock_read(RTC_MONTH);
+		year = chrp_cmos_clock_read(RTC_YEAR);
+	} while (sec != chrp_cmos_clock_read(RTC_SECONDS));
+	if (!(chrp_cmos_clock_read(RTC_CONTROL) & RTC_DM_BINARY) || RTC_ALWAYS_BCD)
+	  {
+	    BCD_TO_BIN(sec);
+	    BCD_TO_BIN(min);
+	    BCD_TO_BIN(hour);
+	    BCD_TO_BIN(day);
+	    BCD_TO_BIN(mon);
+	    BCD_TO_BIN(year);
+	  }
+	if ((year += 1900) < 1970)
+		year += 100;
+	return mktime(year, mon, day, hour, min, sec);
+}
+
diff --git a/arch/ppc/kernel/head.S b/arch/ppc/kernel/head.S
index 25f2dd8a0..3038e8cc1 100644
--- a/arch/ppc/kernel/head.S
+++ b/arch/ppc/kernel/head.S
@@ -45,14 +45,13 @@
 
 #define TOPHYS(x)	(x - KERNELBASE)
 
-
 /* this is a very kludgey way of loading up the BATs on the
    prep system.  I'll kill this horrible macro and write
    something clean when I have a chance -- Cort
  */	
 #define LOAD_BATS(RA,RB) \
 	mfspr	RA,PVR		; \
-	srwi	r5,r5,16	; \
+	srwi	RA,RA,16	; \
 	cmpi	0,RA,1		; \
 	beq	199f		; \
 	/* load bats for 60x */	; \
@@ -132,21 +131,10 @@
 	mtspr	DBAT3L,RB	; \
 200:
 	
-
-
-	
 	.text
 	.globl	_stext
 _stext:
 
-#ifdef CONFIG_PREP	
-	. = 0x100
-_GLOBAL(HardReset)
-	b	_start
-
-#endif /* CONFIG_PREP */
-
-#ifdef CONFIG_PMAC
 /*
  * _start is defined this way because the XCOFF loader in the OpenFirmware
  * on the powermac expects the entry point to be a procedure descriptor.
@@ -156,7 +144,7 @@ _GLOBAL(HardReset)
 _start:
 	.long	TOPHYS(__start),0,0
 
-/*
+/* PMAC
  * Enter here with the kernel text, data and bss loaded starting at
  * 0, running with virtual == physical mapping.
  * r5 points to the prom entry point (the client interface handler
@@ -165,41 +153,71 @@ _start:
  * pointer (r1) points to just below the end of the half-meg region
  * from 0x380000 - 0x400000, which is mapped in already.
  */
+/* PREP
+ * This is jumped to on prep systems right after the kernel is relocated
+ * to its proper place in memory by the boot loader.  The expected layout
+ * of the regs is:	
+ *   r3: ptr to residual data
+ *   r4: initrd_start or if no initrd then 0
+ *   r5: initrd_end - unused if r4 is 0
+ *   r6: Start of command line string
+ *   r7: End of command line string
+ *
+ * This just gets a minimal mmu environment setup so we can call
+ * start_here() to do the real work.
+ * -- Cort	
+ */
+	
 	.globl	__start
 __start:
 
 /*
- * Use the first pair of BAT registers to map the 1st 8MB
+ * Use the first pair of BAT registers to map the 1st 16MB
  * of RAM to KERNELBASE.
  */
-	mfspr	r9,PVR
-	rlwinm	r9,r9,16,16,31		/* r9 = 1 for 601, 4 for 604 */
-	cmpi	0,r9,1
-	lis	r7,KERNELBASE@h
-	bne	4f
-	ori	r7,r7,4			/* set up BAT registers for 601 */
-	li	r8,0x7f
-	b	5f
-4:	ori	r7,r7,0xff		/* set up BAT registers for 604 */
-	li	r8,2
-	mtspr	DBAT0U,r7
-	mtspr	DBAT0L,r8
-5:	mtspr	IBAT0U,r7
-	mtspr	IBAT0L,r8
-	isync
-
+        mfspr   r9,PVR
+        rlwinm  r9,r9,16,16,31          /* r9 = 1 for 601, 4 for 604 */
+        cmpi    0,r9,1
+        lis     r11,KERNELBASE@h
+        bne     4f
+        ori     r11,r11,4               /* set up BAT registers for 601 */
+        li      r8,0x7f
+        ori     r11,r11,4               /* set up BAT registers for 601 */
+        li      r8,0x7f
+        oris    r9,r11,0x800000@h       /* set up BAT reg for 2nd 8M */
+        oris    r10,r8,0x800000@h       /* set up BAT reg for 2nd 8M */
+        mtspr   IBAT1U,r9
+        mtspr   IBAT1L,r10
+        b       5f
+4:      ori     r11,r11,0x1ff           /* set up BAT registers for 604 */
+        li      r8,2
+        mtspr   DBAT0U,r11
+        mtspr   DBAT0L,r8
+5:      mtspr   IBAT0U,r11
+        mtspr   IBAT0L,r8
+        isync
 /*
- * Now we have the 1st 8M of RAM mapped at KERNELBASE, so we can
- * refer to addresses of data items, procedures, etc. normally.
+ * we now have the 1st 16M of ram mapped with the bats.
+ * prep needs the mmu to be turned on here, but pmac already has it on.
+ * this shouldn't bother the pmac since it just gets turned on again
+ * as we jump to our code at KERNELBASE. -- Cort
  */
-	lis	r7,start_here@ha	/* jump up to our copy at KERNELBASE */
-	addi	r7,r7,start_here@l
-	mtlr	r7
-	blr
-#endif /* CONFIG_PMAC */
-
-
-
+	mfmsr	r0
+	ori	r0,r0,MSR_DR|MSR_IR
+	mtspr	SRR1,r0
+	lis	r0,start_here@h
+	ori	r0,r0,start_here@l
+	mtspr	SRR0,r0
+	SYNC
+	rfi				/* enables MMU */
+		
+/*
+ * GCC sometimes accesses words at negative offsets from the stack
+ * pointer, although the SysV ABI says it shouldn't.  To cope with
+ * this, we leave this much untouched space on the stack on exception
+ * entry.
+ */
+#define STACK_UNDERHEAD	64
 	
 /*
  * Macros for storing registers into and loading registers from
@@ -286,10 +304,8 @@ label:						\
 	.long	hdlr;				\
 	.long	int_return
 
-#ifndef CONFIG_PREP
 /* System reset */
 	STD_EXCEPTION(0x100, Reset, UnknownException)
-#endif /* ndef CONFIG_PREP */
 
 /* Machine check */
 	STD_EXCEPTION(0x200, MachineCheck, MachineCheckException)
@@ -319,7 +335,7 @@ DataAccess:
 
 /* Instruction access exception */
 	. = 0x400
-InstructionAccess:	
+InstructionAccess:
 	EXCEPTION_PROLOG
 	andis.	r0,r23,0x4000		/* no pte found? */
 	beq	1f			/* if so, try to put a PTE */
@@ -338,7 +354,7 @@ InstructionAccess:
 	.long	int_return
 
 /* External interrupt */
-	STD_EXCEPTION(0x500, HardwareInterrupt, handle_IRQ)
+	STD_EXCEPTION(0x500, HardwareInterrupt, do_IRQ)
 
 /* Alignment exception */
 	. = 0x600
@@ -376,21 +392,7 @@ FPUnavailable:
 	.long	KernelFP
 	.long	int_return
 
-/* Decrementer */
-#ifdef CONFIG_PREP
-/* - ignored for now... */
-_ORG(0x0900)
-	mtspr	SPRG0,r1
-	lis	r1,0x7FFF
-	ori	r1,r1,0xFFFF
-	mtspr	DEC,r1
-	mfspr	r1,SPRG0
-	rfi
-#endif /* CONFIG_PREP */
-#ifdef CONFIG_PMAC
 	STD_EXCEPTION(0x900, Decrementer, timer_interrupt)
-#endif /* CONFIG_PMAC */
-
 	STD_EXCEPTION(0xa00, Trap_0a, UnknownException)
 	STD_EXCEPTION(0xb00, Trap_0b, UnknownException)
 
@@ -596,11 +598,6 @@ transfer_to_handler:
 	andi.	r23,r23,MSR_PR
 	mfspr	r23,SPRG3		/* if from user, fix up tss */
 	beq	2f
-#ifdef CONFIG_PMAC	
-	lwz	r24,GPR1(r21)
-	stw	r22,LAST_PC(r23)
-	stw	r24,USER_STACK(r23)
-#endif /* CONFIG_PMAC */
 	addi	r24,r1,STACK_FRAME_OVERHEAD
 	stw	r24,PT_REGS(r23)
 2:	addi	r2,r23,-TSS		/* set r2 to current */
@@ -630,7 +627,7 @@ load_up_fpu:
 	REST_32FPRS(0, r5)
 
 /* use last_task_used_math instead of fpu_tss */
-	lis	r3,last_task_used_math@h/*a*/
+	lis	r3,last_task_used_math@ha
 	addis	r3,r3,-KERNELBASE@h
 	subi	r4,r5,TSS
 	addis	r4,r4,KERNELBASE@h
@@ -654,7 +651,7 @@ load_up_fpu:
 	REST_GPR(20, r21)
 	REST_2GPRS(22, r21)
 	lwz	r21,GPR21(r21)
- 	SYNC
+	SYNC
 	rfi
 
 /*
@@ -682,14 +679,15 @@ Hash_msk = (((1 << Hash_bits) - 1) * 64)
 	.globl	hash_page
 hash_page:
 	/* Get PTE (linux-style) and check access */
-	lwz	r5,PG_TABLES(r5)	/* task's page tables */
-	lis	r2,-KERNELBASE@h
-	add	r5,r5,r2		/* convert to phys addr */
+	lwz	r5,MM-TSS(r5)		
+	addis	r5,r5,-KERNELBASE@h	/* get physical current->mm */
+	lwz	r5,PGD(r5)		/* get current->mm->pgd */	
+	addis	r5,r5,-KERNELBASE@h	/* convert to phys addr */
 	rlwimi	r5,r3,12,20,29		/* insert top 10 bits of address */
 	lwz	r5,0(r5)		/* get pmd entry */
 	rlwinm.	r5,r5,0,0,19		/* extract address of pte page */
 	beqlr-				/* return if no mapping */
-	add	r2,r5,r2
+	addis	r2,r5,-KERNELBASE@h
 	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 */
@@ -819,58 +817,62 @@ start_here:
 	rlwinm	r9,r9,16,16,31
 	cmpi	0,r9,1
 	beq	4f			/* not needed for 601 */
-	mfspr	r7,HID0
-	andi.	r0,r7,HID0_DCE
-	ori	r7,r7,HID0_ICE|HID0_DCE
-	ori	r8,r7,HID0_ICFI
+	mfspr	r11,HID0
+	andi.	r0,r11,HID0_DCE
+	ori	r11,r11,HID0_ICE|HID0_DCE
+	ori	r8,r11,HID0_ICFI
 	bne	3f			/* don't invalidate the D-cache */
 	ori	r8,r8,HID0_DCI		/* unless it wasn't enabled */
-3:	sync
-	mtspr	HID0,r8			/* enable and invalidate caches */
+3:
+	/* turn on dpm for 603 */
+	cmpi	0,r9,3
+	bne	10f
+	oris	r11,r11,HID0_DPM@h
+10:
 	sync
-	mtspr	HID0,r7			/* enable caches */
+	mtspr	HID0,r8			/* enable and invalidate caches */
+	mtspr	HID0,r11		/* enable caches */
 	sync
 	isync
 	cmpi	0,r9,4			/* check for 604 */
 	cmpi	1,r9,9			/* or 604e */
 	cror	2,2,6
 	bne	4f
-	ori	r7,r7,HID0_SIED|HID0_BHTE /* for 604[e], enable */
-	mtspr	HID0,r7			/* superscalar exec & br history tbl */
+	ori	r11,r11,HID0_SIED|HID0_BHTE /* for 604[e], enable */
+	mtspr	HID0,r11		/* superscalar exec & br history tbl */
 4:
 	/* ptr to current */
 	lis	r2,init_task_union@h
 	ori	r2,r2,init_task_union@l
 	/* ptr to phys current tss */
-	addis	r3,r2,-KERNELBASE@h
-	addi	r3,r3,TSS	/* init task's TSS */
-	mtspr	SPRG3,r3
+	addis	r11,r2,-KERNELBASE@h
+	addi	r11,r11,TSS	/* init task's TSS */
+	mtspr	SPRG3,r11
 	/* stack */
 	addi	r1,r2,TASK_UNION_SIZE
 	li	r0,0
 	stwu	r0,-STACK_FRAME_OVERHEAD(r1)
 
 	/* Clear out the BSS */
-	lis	r7,_end@ha
-	addi	r7,r7,_end@l
+	lis	r11,_end@ha
+	addi	r11,r11,_end@l
 	lis	r8,__bss_start@ha
 	addi	r8,r8,__bss_start@l
-	subf	r7,r8,r7
-	addi	r7,r7,3
-	rlwinm.	r7,r7,30,2,31
+	subf	r11,r8,r11
+	addi	r11,r11,3
+	rlwinm.	r11,r11,30,2,31
 	beq	2f
 	addi	r8,r8,-4
-	mtctr	r7
+	mtctr	r11
 	li	r0,0
 3:	stwu	r0,4(r8)
 	bdnz	3b
 2:
 /*
- * Initialize the prom stuff (powermacs only) and the MMU.
+ * Initialize the prom stuff and the MMU.
  */
-#ifdef CONFIG_PMAC
+	bl	identify_machine
 	bl	prom_init
-#endif /* CONFIG_PMAC */
 	bl	MMU_init
 
 /*
@@ -889,11 +891,6 @@ start_here:
 	rfi
 /* Load up the kernel context */
 2:
-#ifdef CONFIG_PREP
-	/* reload the bats now that MMU_init() has setup them up -- Cort */
-	LOAD_BATS(r3,r0)
-#endif
-
 	SYNC			/* Force all PTE updates to finish */
 	tlbia			/* Clear all TLB entries */
 	mtspr	SDR1,r6
@@ -905,8 +902,19 @@ start_here:
 	addi	r3,r3,1		/* increment VSID */
 	addis	r4,r4,0x1000	/* address of next segment */
 	bdnz	3b
-#ifdef CONFIG_PMAC
-	li	r0,0		/* zot the BATs */
+
+	lis	r3,_machine@ha
+	addis	r3,r3,-KERNELBASE@h
+	lwz	r0,_machine@l(r3)
+	cmpi	0,r0,_MACH_Pmac
+	beq	99f
+	
+/* on prep reload the bats now that MMU_init() has setup them up -- Cort */
+	LOAD_BATS(r3,r14)
+	b	100f
+	
+/* on pmac clear the bats out */	
+99:	li	r0,0		/* zot the BATs */
 #if 1
 	mtspr	IBAT0U,r0
 	mtspr	IBAT0L,r0
@@ -925,7 +933,7 @@ start_here:
 	mtspr	IBAT3L,r0
 	mtspr	DBAT3U,r0
 	mtspr	DBAT3L,r0
-#endif
+100:	
 /* Now turn on the MMU for real! */
 	li	r4,MSR_KERNEL
 	lis	r3,start_kernel@h
@@ -934,60 +942,29 @@ start_here:
 	mtspr	SRR1,r4
 	rfi			/* enable MMU and jump to start_kernel */
 
-#ifdef CONFIG_PREP
-/*
- * This is jumped to on prep systems right after the kernel is relocated
- * to its proper place in memory by the boot loader.  The expected layout
- * of the regs is:	
- *   R3: End of image
- *   R4: Start of image - 0x400
- *   R11: Start of command line string
- *   R12: End of command line string
- *
- * This just gets a minimal mmu environment setup so we can call
- * start_here() to do the real work.
- * -- Cort	
- */
-	.globl __start
-__start:	
-	.globl _start
-_start:	
-	lis	r7,0xF000		/* To mask upper 4 bits */
-/* save pointer to residual data */
-	lis	r1,resptr@h
-	ori	r1,r1,resptr@l
-	addis	r1,r1,-KERNELBASE@h
-	stw	r3,0(r1)
-/* save argument string */
-	li	r0,0		/* Null terminate string */
-	stb	r0,0(r12)
-	lis	r1,cmd_line@h
-	ori	r1,r1,cmd_line@l
-	addis	r1,r1,-KERNELBASE@h
-	subi	r1,r1,1
-	subi	r11,r11,1
-00:	lbzu	r0,1(r11)
-	cmpi	0,r0,0
-	stbu	r0,1(r1)
-	bne	00b
-/* setup the msr with sane values */
-	li	r0,MSR_
-	mtmsr	r0
-/* turn on the mmu with bats covering kernel enough to get started */
-	LOAD_BATS(r3,r0)
+
+	.globl reset_SDR1
+reset_SDR1:
+	lis	r6,_SDR1@ha
+	lwz	r6,_SDR1@l(r6)
 	mfmsr	r3
-	ori	r3,r3,MSR_DR|MSR_IR
+	li	r4,MSR_IR|MSR_DR
+	andc	r3,r3,r4
+	lis	r4,2f@h
+	addis	r4,r4,-KERNELBASE@h
+	ori	r4,r4,2f@l
+	mtspr	SRR0,r4
 	mtspr	SRR1,r3
-	lis	r3,10f@h
-	ori	r3,r3,10f@l
+	rfi
+2:	/* load new SDR1 */
+	tlbia			
+	mtspr	SDR1,r6
+	/* turn the mmu back on */
+	li	r4,MSR_KERNEL
+	mflr	r3
 	mtspr	SRR0,r3
-	SYNC
-	rfi				/* enables MMU */
-10:	lis	r7,start_here@ha	/* jump up to our copy at KERNELBASE */
-	addi	r7,r7,start_here@l
-	mtlr	r7
-	blr
-#endif /* CONFIG_PREP */
+	mtspr	SRR1,r4
+	rfi
 	
 /*
  * FP unavailable trap from kernel - print a message, but let
@@ -1022,31 +999,23 @@ giveup_fpu_unmapped:
 giveup_fpu:
 	li	r6,0
 1:
-	addis	r3,r6,last_task_used_math@h/*a*/
+	addis	r3,r6,last_task_used_math@ha
 	lwz	r4,last_task_used_math@l(r3)
-#if 0		
-	addis	r3,r6,fpu_tss@ha
-	lwz	r4,fpu_tss@l(r3)
-#endif	
 	mfmsr	r5
 	ori	r5,r5,MSR_FP
 	SYNC
 	mtmsr	r5			/* enable use of fpu now */
 	SYNC
 	cmpi	0,r4,0
+	beqlr-				/* if no previous owner, done */
 	add	r4,r4,r6
-	beqlr				/* if no previous owner, done */
 	addi	r4,r4,TSS	        /* want TSS of last_task_used_math */
 	li	r5,0
 	stw	r5,last_task_used_math@l(r3)
-#if 0	
-	stw	r5,fpu_tss@l(r3)
-#endif
 	SAVE_32FPRS(0, r4)
 	mffs	fr0
 	stfd	fr0,TSS_FPSCR-4(r4)
 	lwz	r5,PT_REGS(r4)
-	lwz	r5,PT_REGS(r4)
 	add	r5,r5,r6
 	lwz	r3,_MSR-STACK_FRAME_OVERHEAD(r5)
 	li	r4,MSR_FP
@@ -1268,7 +1237,7 @@ int_return:
 	cmpi	0,r4,0
 	beq+	1f
 	addi	r3,r1,STACK_FRAME_OVERHEAD
-	bl	handle_IRQ
+	bl	do_IRQ
 	b	3b
 1:	lis	r4,bh_mask@ha
 	lwz	r4,bh_mask@l(r4)
@@ -1341,7 +1310,7 @@ _GLOBAL(fake_interrupt)
 	li	r0,0x0fac
 	stw	r0,TRAP(r1)
 	addi	r3,r1,STACK_FRAME_OVERHEAD
-	bl	handle_IRQ
+	bl	do_IRQ
 	addi	r1,r1,INT_FRAME_SIZE+STACK_UNDERHEAD
 	lwz	r0,4(r1)
 	mtlr	r0
@@ -1384,11 +1353,11 @@ _GLOBAL(flush_instruction_cache)
  * and invalidate the corresponding instruction cache blocks.
  * This is a no-op on the 601.
  *
- * store_cache_range(unsigned long start, unsigned long stop)
+ * flush_icache_range(unsigned long start, unsigned long stop)
  */
 CACHE_LINE_SIZE = 32
 LG_CACHE_LINE_SIZE = 5
-_GLOBAL(store_cache_range)
+_GLOBAL(flush_icache_range)
 	mfspr	r5,PVR
 	rlwinm	r5,r5,16,16,31
 	cmpi	0,r5,1
@@ -1521,7 +1490,6 @@ _GLOBAL(flush_hash_page)
 _GLOBAL(__main)
 	blr
 
-#ifdef CONFIG_PMAC
 /*
  * These exception handlers are used when we have called a prom
  * routine after we have taken over the exception vectors and MMU.
@@ -1716,7 +1684,6 @@ enter_prom:
 	lwz	r31,28(r1)
 	lwz	r1,0(r1)
 	blr
-#endif
 
 /*
  * We put a few things here that have to be page-aligned.
@@ -1726,7 +1693,6 @@ enter_prom:
 	.data
 	.globl	sdata
 sdata:
-	.space	2*4096
 	.globl	empty_zero_page
 empty_zero_page:
 	.space	4096
@@ -1735,7 +1701,6 @@ empty_zero_page:
 swapper_pg_dir:
 	.space	4096	
 
-#ifdef CONFIG_PREP
 /*
  * This space gets a copy of optional info passed to us by the bootstrap
  * Used to pass parameters into the kernel like root=/dev/sda1, etc.
@@ -1743,5 +1708,4 @@ swapper_pg_dir:
 	.globl	cmd_line
 cmd_line:
 	.space	512
-#endif /* CONFIG_PREP */
 
diff --git a/arch/ppc/kernel/idle.c b/arch/ppc/kernel/idle.c
new file mode 100644
index 000000000..bfe1a4146
--- /dev/null
+++ b/arch/ppc/kernel/idle.c
@@ -0,0 +1,279 @@
+/*
+ * $Id: idle.c,v 1.4 1997/08/23 22:46:01 cort Exp $
+ *
+ * Idle daemon for PowerPC.  Idle daemon will handle any action
+ * that needs to be taken when the system becomes idle.
+ *
+ * Written by Cort Dougan (cort@cs.nmt.edu)
+ *
+ * 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.
+ */
+#define __KERNEL_SYSCALLS__
+
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/ptrace.h>
+#include <linux/malloc.h>
+#include <linux/config.h>
+
+#include <asm/pgtable.h>
+#include <asm/uaccess.h>
+#include <asm/system.h>
+#include <asm/io.h>
+#include <asm/smp_lock.h>
+#include <asm/processor.h>
+
+int zero_paged(void *unused);
+int power_saved(void *unused);
+
+asmlinkage int sys_idle(void)
+{
+	int ret = -EPERM;
+
+	if (current->pid != 0)
+		goto out;
+
+	/*
+	 * want one per cpu since it would be nice to have all
+	 * processors who aren't doing anything
+	 * zero-ing pages since this daemon is lock-free
+	 * -- Cort
+	 */
+	kernel_thread(zero_paged, NULL, 0);
+	/* no powersaving modes on 601 */
+	/*if(  (_get_PVR()>>16) != 1 )
+		kernel_thread(power_saved, NULL, 0);*/
+
+	/* endless loop with no priority at all */
+	current->priority = -100;
+	current->counter = -100;
+	for (;;)
+	{
+		schedule();
+	}
+	ret = 0;
+out:
+	return ret;
+}
+
+/*
+ * vars for idle task zero'ing out pages
+ */
+unsigned long zero_list = 0;	/* head linked list of pre-zero'd pages */
+unsigned long bytecount = 0;	/* pointer into the currently being zero'd page */
+unsigned long zerocount = 0;	/* # currently pre-zero'd pages */
+unsigned long zerototal = 0;	/* # pages zero'd over time -- for ooh's and ahhh's */
+unsigned long pageptr = 0;	/* current page being zero'd */
+unsigned long zeropage_hits = 0;/* # zero'd pages request that we've done */
+unsigned long zeropage_calls = 0;/* # zero'd pages request that've been made */
+static struct wait_queue * page_zerod_wait = NULL;
+#define PAGE_THRESHOLD 96       /* how many pages to keep pre-zero'd */
+
+/*
+ * Returns a pre-zero'd page from the list otherwise returns
+ * NULL.
+ */
+unsigned long get_prezerod_page(void)
+{
+	unsigned long page;
+
+	atomic_inc((atomic_t *)&zeropage_calls);
+	if ( zero_list )
+	{
+		/* atomically remove this page from the list */
+		asm (	"101:lwarx  %1,0,%2\n"  /* reserve zero_list */
+			"    lwz    %0,0(%1)\n" /* get next -- new zero_list */
+			"    stwcx. %0,0,%2\n"  /* update zero_list */
+			"    bne-   101b\n"     /* if lost reservation try again */
+			: "=&r" (zero_list), "=&r" (page)
+			: "r" (&zero_list)
+			: "cc" );
+		/* we can update zerocount after the fact since it is not
+		 * used for anything but control of a loop which doesn't
+		 * matter since it won't effect anything if it zero's one
+		 * less page -- Cort
+		 */
+		atomic_inc((atomic_t *)&zeropage_hits);
+		atomic_dec((atomic_t *)&zerocount);
+		wake_up(&page_zerod_wait);
+		resched_force();
+		
+		/* zero out the pointer to next in the page */
+		*(unsigned long *)page = 0;
+		return page;
+	}
+	return 0;
+}
+
+/*
+ * Experimental stuff to zero out pages in the idle task
+ * to speed up get_free_pages() -- Cort
+ * Zero's out pages until we need to resched or
+ * we've reached the limit of zero'd pages.
+ */
+
+int zero_paged(void *unused)
+{
+	extern pte_t *get_pte( struct mm_struct *mm, unsigned long address );
+	pgd_t *dir;
+	pmd_t *pmd;
+	pte_t *pte;
+
+	sprintf(current->comm, "zero_paged (idle)");
+	current->blocked = ~0UL;
+	
+	printk("Started zero_paged\n");
+	
+	__sti();
+	while ( 1 )
+	{
+		/* don't want to be pre-empted by swapper or power_saved */
+		current->priority = -98;
+		current->counter = -98;
+		/* we don't want to run until we have something to do */
+		while ( zerocount >= PAGE_THRESHOLD )
+			sleep_on(&page_zerod_wait);
+		/*
+		 * Mark a page as reserved so we can mess with it
+		 * If we're interrupted we keep this page and our place in it
+		 * since we validly hold it and it's reserved for us.
+		 */
+		pageptr = __get_free_pages(GFP_ATOMIC, 0, 0 );
+		if ( !pageptr )
+		{
+			printk("!pageptr in zero_paged\n");
+			goto retry;
+		}
+		
+		if ( resched_needed() )
+			schedule();
+		
+		/*
+		 * Make the page no cache so we don't blow our cache with 0's
+		 */
+		dir = pgd_offset( init_task.mm, pageptr );
+		if (dir)
+		{
+			pmd = pmd_offset(dir, pageptr & PAGE_MASK);
+			if (pmd && pmd_present(*pmd))
+			{
+				pte = pte_offset(pmd, pageptr & PAGE_MASK);
+				if (pte && pte_present(*pte))
+				{			
+					pte_uncache(*pte);
+					flush_tlb_page(find_vma(init_task.mm,pageptr),pageptr);
+				}
+			}
+		}
+	
+		/*
+		 * Important here to not take time away from real processes.
+		 */
+		for ( bytecount = 0; bytecount < PAGE_SIZE ; bytecount += 4 )
+		{
+			if ( resched_needed() )
+				schedule();
+			*(unsigned long *)(bytecount + pageptr) = 0;
+		}
+		
+		/*
+		 * If we finished zero-ing out a page add this page to
+		 * the zero_list atomically -- we can't use
+		 * down/up since we can't sleep in idle.
+		 * Disabling interrupts is also a bad idea since we would
+		 * steal time away from real processes.
+		 * We can also have several zero_paged's running
+		 * on different processors so we can't interfere with them.
+		 * So we update the list atomically without locking it.
+		 * -- Cort
+		 */
+		/* turn cache on for this page */
+		pte_cache(*pte);
+		flush_tlb_page(find_vma(init_task.mm,pageptr),pageptr);
+		
+		/* atomically add this page to the list */
+		asm (	"101:lwarx  %0,0,%1\n"  /* reserve zero_list */
+			"    stw    %0,0(%2)\n" /* update *pageptr */
+#ifdef __SMP__
+			"    sync\n"            /* let store settle */
+#endif			
+			"    mr     %0,%2\n"    /* update zero_list in reg */
+			"    stwcx. %2,0,%1\n"  /* update zero_list in mem */
+			"    bne-   101b\n"     /* if lost reservation try again */
+			: "=&r" (zero_list)
+			: "r" (&zero_list), "r" (pageptr)
+			: "cc" );
+		/*
+		 * This variable is used in the above loop and nowhere
+		 * else so the worst that could happen is we would
+		 * zero out one more or one less page than we want
+		 * per processor on the machine.  This is because
+		 * we could add our page to the list but not have
+		 * zerocount updated yet when another processor
+		 * reads it.  -- Cort
+		 */
+		atomic_inc((atomic_t *)&zerocount);
+		atomic_inc((atomic_t *)&zerototal);
+retry:	
+		schedule();
+	}
+}
+
+int power_saved(void *unused)
+{
+	unsigned long msr, hid0;
+	sprintf(current->comm, "power_saved (idle)");
+	current->blocked = ~0UL;
+	
+	printk("Power saving daemon started\n");
+	
+	__sti();
+	while (1)
+	{
+		/* don't want to be pre-empted by swapper */
+		current->priority = -99;
+		current->counter = -99;
+		/* go ahead and wakeup page_zerod() */
+		wake_up(&page_zerod_wait);
+		schedule();
+		asm volatile(
+			/* clear powersaving modes and set nap mode */
+			"mfspr %3,1008 \n\t"
+			"andc  %3,%3,%4 \n\t"
+			"or    %3,%3,%5 \n\t"
+			"mtspr 1008,%3 \n\t"
+			/* enter the mode */
+			"mfmsr %0 \n\t"
+			"oris  %0,%0,%2 \n\t"
+			"sync \n\t"
+			"mtmsr %0 \n\t"
+			"isync \n\t"
+			: "=&r" (msr)
+			: "0" (msr), "i" (MSR_POW>>16),
+			"r" (hid0),
+			"r" (HID0_DOZE|HID0_NAP|HID0_SLEEP),
+			"r" (HID0_NAP));
+		/*
+		 * The ibm carolina spec says that the eagle memory
+		 * controller will detect the need for a snoop
+		 * and wake up the processor so we don't need to
+		 * check for cache operations that need to be
+		 * snooped.  The ppc book says the run signal
+		 * must be asserted while napping for this though.
+		 *
+		 * Paul, what do you know about the pmac here?
+		 * -- Cort
+		 */
+		schedule();
+	}
+}
+
diff --git a/arch/ppc/kernel/irq.c b/arch/ppc/kernel/irq.c
index 171bfaa5c..da3c53697 100644
--- a/arch/ppc/kernel/irq.c
+++ b/arch/ppc/kernel/irq.c
@@ -1,13 +1,13 @@
 /*
  *  arch/ppc/kernel/irq.c
  *
- *  Power Macintosh version
- *    Copyright (C) 1996 Paul Mackerras (paulus@cs.anu.edu.au)
- *
  *  Derived from arch/i386/kernel/irq.c
  *    Copyright (C) 1992 Linus Torvalds
  *  Adapted from arch/i386 by Gary Thomas
- *  Modified by Cort Dougan (cort@cs.nmt.edu) 
+ *    Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
+ *  Updated and modified by Cort Dougan (cort@cs.nmt.edu)
+ *  Adapted for Power Macintosh by Paul Mackerras
+ *    Copyright (C) 1996 Paul Mackerras (paulus@cs.anu.edu.au)
  *
  * This file contains the code used by various IRQ handling routines:
  * asking for different IRQ's should be done through these routines
@@ -15,11 +15,7 @@
  * shouldn't result in any weird surprises, and installing new handlers
  * should be easier.
  */
-	
-/*
- * IRQ's are in fact implemented a bit like signal handlers for the kernel.
- * Naturally it's not a 1:1 relation, but there are similarities.
- */
+
 
 #include <linux/ptrace.h>
 #include <linux/errno.h>
@@ -30,121 +26,164 @@
 #include <linux/interrupt.h>
 #include <linux/timex.h>
 #include <linux/config.h>
-
+#include <linux/init.h>
+#include <linux/malloc.h>
+#include <linux/openpic.h>
+#include <linux/pci.h>
+#include <linux/bios32.h>
+#include <linux/openpic.h>
+
+#include <asm/hydra.h>
 #include <asm/system.h>
 #include <asm/io.h>
 #include <asm/irq.h>
 #include <asm/bitops.h>
 
-#define IRQ_FLAG	((unsigned *)0xf3000020)
-#define IRQ_ENABLE	((unsigned *)0xf3000024)
-#define IRQ_ACK		((unsigned *)0xf3000028)
-#define IRQ_LEVEL	((unsigned *)0xf300002c)
-
-#define KEYBOARD_IRQ	20	/* irq number for command-power interrupt */
-
-#undef SHOW_IRQ 1
+#undef SHOW_IRQ
+#define OPENPIC_DEBUG
 
 unsigned lost_interrupts = 0;
-
 unsigned int local_irq_count[NR_CPUS];
-static struct irqaction irq_action[32];
+static struct irqaction irq_action[NR_IRQS];
+static int spurious_interrupts = 0;
+int __ppc_bh_counter;
+static unsigned int cached_irq_mask = 0xffffffff;
+static void no_action(int cpl, void *dev_id, struct pt_regs *regs) { }
+
+#define cached_21	(((char *)(&cached_irq_mask))[3])
+#define cached_A1	(((char *)(&cached_irq_mask))[2])
 
 /*
- * This contains the irq mask for both irq controllers
+ * These are set to the appropriate functions by init_IRQ()
  */
-static unsigned int cached_irq_mask = 0xffff;
+void (*mask_and_ack_irq)(int irq_nr);
+void (*set_irq_mask)(int irq_nr);
 
-#define cached_21	(((char *)(&cached_irq_mask))[0])
-#define cached_A1	(((char *)(&cached_irq_mask))[1])
 
+/* prep */
+#define PREP_IRQ_MASK	(((unsigned int)cached_A1)<<8) | (unsigned int)cached_21
+extern unsigned long route_pci_interrupts(void);
 
-int __ppc_bh_counter;
+/* pmac */
+#define IRQ_FLAG	((unsigned *)0xf3000020)
+#define IRQ_ENABLE	((unsigned *)0xf3000024)
+#define IRQ_ACK		((unsigned *)0xf3000028)
+#define IRQ_LEVEL	((unsigned *)0xf300002c)
+#define KEYBOARD_IRQ	20	/* irq number for command-power interrupt */
+#define PMAC_IRQ_MASK	(~ld_le32(IRQ_ENABLE))
 
-void *null_handler(int,void *,struct pt_regs *);
+/* chrp */
+volatile struct Hydra *Hydra = NULL;
 
-/*
- * disable and enable intrs in software.  This is used
- * from the non-realtime parts of Linux to disable interrupts.
- * The realtime part disables/enables intrs in the hardware.
- * -- Cort
- */
-unsigned long soft_intr_enable = 1;
-void _soft_cli(void)
+void i8259_mask_and_ack_irq(int irq_nr)
 {
-	soft_intr_enable = 0;
+	spin_lock(&irq_controller_lock);
+	cached_irq_mask |= 1 << irq_nr;
+	if (irq_nr > 7) {
+		inb(0xA1);	/* DUMMY */
+		outb(cached_A1,0xA1);
+		outb(0x62,0x20);	/* Specific EOI to cascade */
+		/*outb(0x20,0xA0);*/
+		outb(0x60|(irq_nr-8), 0xA0); /* specific eoi */
+	} else {
+		inb(0x21);	/* DUMMY */
+		outb(cached_21,0x21);
+		/*outb(0x20,0x20);*/
+		outb(0x60|irq_nr,0x20); /* specific eoi */
+		  
+	}
+	spin_unlock(&irq_controller_lock);
 }
 
-void _soft_sti(void)
+void pmac_mask_and_ack_irq(int irq_nr)
 {
-	soft_intr_enable = 1;
-	if ( lost_interrupts )
-	{
-		printk("lost_interrupts from _soft_sti() %x\n",lost_interrupts);
-		fake_interrupt();
-	}
+	spin_lock(&irq_controller_lock);
+	cached_irq_mask |= 1 << irq_nr;
+	out_le32(IRQ_ENABLE, ld_le32(IRQ_ENABLE) & ~(1 << irq_nr));
+	out_le32(IRQ_ACK, 1U << irq_nr);
+	spin_unlock(&irq_controller_lock);
 }
 
-void *
-null_handler(int a, void *b, struct pt_regs *regs)
+void chrp_mask_and_ack_irq(int irq_nr)
 {
-	/*printk("irq.c: null_handler() called.  Should not have happened.\n");*/
+	/* spinlocks are done by i8259_mask_and_ack() - Cort */
+	if (is_8259_irq(irq_nr))
+	    i8259_mask_and_ack_irq(irq_nr);
+	openpic_eoi(0);	
 }
 
-void
-disable_irq(unsigned int irq_nr)
-{
-	int s = _disable_interrupts();
-	unsigned char mask;
 
-#ifdef CONFIG_PMAC
-	out_le32(IRQ_ENABLE, ld_le32(IRQ_ENABLE) & ~(1 << irq_nr));
-#else /* CONFIG_PMAC */
-	mask = 1 << (irq_nr & 7);
-	if (irq_nr < 8)
-	{
-		cached_21 |= mask;
-		outb(cached_21,0x21);
-	} else
-	{
-		cached_A1 |= mask;
+void i8259_set_irq_mask(int irq_nr)
+{
+	if (irq_nr > 7) {
 		outb(cached_A1,0xA1);
-	}	
-#endif	/* CONFIG_PMAC */
-	_enable_interrupts(s);
+	} else {
+		outb(cached_21,0x21);
+	}
 }
 
-void
-enable_irq(unsigned int irq_nr)
+void pmac_set_irq_mask(int irq_nr)
 {
-	int s = _disable_interrupts();
-#ifdef CONFIG_PMAC
-	unsigned bit = 1U << irq_nr;
-	
-	out_le32(IRQ_ENABLE, ld_le32(IRQ_ENABLE) & ~(1 << irq_nr));
-	out_le32(IRQ_ENABLE, ld_le32(IRQ_ENABLE) | bit);
-
+	/* this could be being enabled or disabled - so use cached_irq_mask */
+	out_le32(IRQ_ENABLE, ~cached_irq_mask /* enable all unmasked */ );
 	/*
 	 * Unfortunately, setting the bit in the enable register
 	 * when the device interrupt is already on *doesn't* set
 	 * the bit in the flag register or request another interrupt.
 	 */
-	if ((ld_le32(IRQ_LEVEL) & bit) && !(ld_le32(IRQ_FLAG) & bit))
-		lost_interrupts |= bit;
-#else /* CONFIG_PMAC */
-	if (irq_nr < 8) {
-		cached_21 &= ~(1 << (irq_nr & 7));
-		outb(cached_21,0x21);
+	if ((ld_le32(IRQ_LEVEL) & (1UL<<irq_nr)) && !(ld_le32(IRQ_FLAG) & (1UL<<irq_nr)))
+		lost_interrupts |= (1UL<<irq_nr);
+}
+
+void chrp_set_irq_mask(int irq_nr)
+{
+	if (is_8259_irq(irq_nr)) {
+		i8259_set_irq_mask(irq_nr);
 	} else
 	{
-		cached_A1 &= ~(1 << (irq_nr-8 & 7));
-		outb(cached_A1,0xA1);
-	}	
-#endif	/* CONFIG_PMAC */
+		/* disable? */
+		if ( cached_irq_mask & (1UL<<irq_nr) )
+			openpic_disable_irq(irq_to_openpic(irq_nr));
+		/* enable */
+		else
+			openpic_disable_irq(irq_to_openpic(irq_nr));
+	}
+}
+
+/*
+ * These have to be protected by the spinlock
+ * before being called.
+ */
+static inline void mask_irq(unsigned int irq_nr)
+{
+	cached_irq_mask |= 1 << irq_nr;
+	set_irq_mask(irq_nr);
+}
 
-	_enable_interrupts(s);
+static inline void unmask_irq(unsigned int irq_nr)
+{
+	cached_irq_mask &= ~(1 << irq_nr);
+	set_irq_mask(irq_nr);
 }
 
+void disable_irq(unsigned int irq_nr)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&irq_controller_lock, flags);
+	mask_irq(irq_nr);
+	spin_unlock_irqrestore(&irq_controller_lock, flags);
+	synchronize_irq();
+}
+
+void enable_irq(unsigned int irq_nr)
+{
+	unsigned long flags;
+
+	spin_lock_irqsave(&irq_controller_lock, flags);
+	unmask_irq(irq_nr);
+	spin_unlock_irqrestore(&irq_controller_lock, flags);
+}
 
 int get_irq_list(char *buf)
 {
@@ -153,7 +192,7 @@ int get_irq_list(char *buf)
 
 	for (i = 0 ; i < NR_IRQS ; i++) {
 		action = irq_action + i;
-		if (!action || !action->handler) 
+		if (!action || !action->handler)
 			continue;
 		len += sprintf(buf+len, "%2d: %10u   %s",
 			i, kstat.interrupts[i], action->name);
@@ -162,152 +201,143 @@ int get_irq_list(char *buf)
 		}
 		len += sprintf(buf+len, "\n");
 	}
-/*
- *	Linus - should you add NMI counts here ?????
- */
 #ifdef __SMP_PROF__
 	len+=sprintf(buf+len, "IPI: %8lu received\n",
 		ipi_count);
-#endif		
+#endif
+	len += sprintf(buf+len, "99: %10u    spurious or short\n",
+		       spurious_interrupts);
 	return len;
 }
 
-asmlinkage void handle_IRQ(struct pt_regs *regs)
+asmlinkage void do_IRQ(struct pt_regs *regs)
 {
 	int irq;
-	unsigned bits;
+	unsigned long bits;
 	struct irqaction *action;
 	int cpu = smp_processor_id();
+	int status;
 
 	hardirq_enter(cpu);
 
-#ifdef CONFIG_PMAC 
-	bits = ld_le32(IRQ_FLAG) | lost_interrupts;
-	lost_interrupts = 0;
-	
-	for (irq = NR_IRQS; irq >= 0; --irq)
-		if (bits & (1U << irq))
-			break;
-#else /* CONFIG_PMAC */
-#if 1
-	if ( lost_interrupts )
-	{
-		irq = ffz(~lost_interrupts);
-		lost_interrupts &= ~irq;
-		goto retry;
-	}
-	outb(0x0C, 0x20);
-	irq = inb(0x20) & 7;
-	if (irq == 2)
+
+	/*
+	 * I'll put this ugly mess of code into a function
+	 * such as get_pending_irq() or some such clear thing
+	 * so we don't have a switch in the irq code and
+	 * the chrp code is merged a bit with the prep.
+	 * -- Cort
+	 */
+	switch ( _machine )
 	{
-retry_cascade:		
-		outb(0x0C, 0xA0);
-		irq = inb(0xA0);
-		/* if no intr left */
-		if ( !(irq & 128 ) )
+	case _MACH_Pmac:
+		bits = ld_le32(IRQ_FLAG) | lost_interrupts;
+		lost_interrupts = 0;
+		for (irq = NR_IRQS - 1; irq >= 0; --irq)
+			if (bits & (1U << irq))
+				break;
+		break;
+	case _MACH_chrp:
+		irq = openpic_irq(0);		
+		if (irq == IRQ_8259_CASCADE)
+		{
+			/*
+			 * This magic address generates a PCI IACK cycle.
+			 * 
+			 * This should go in the above mask/ack code soon. -- Cort
+			 */
+			irq = (*(volatile unsigned char *)0xfec80000) & 0x0f;
+		}
+		else if (irq >= 64)
+		{
+			/*
+			 *  OpenPIC interrupts >64 will be used for other purposes
+			 *  like interprocessor interrupts and hardware errors
+			 */
+#ifdef OPENPIC_DEBUG
+			printk("OpenPIC interrupt %d\n", irq);
+#endif
+			if (irq==99)
+				spurious_interrupts++;
+		}
+		else {
+			/*
+			 *  Here we should process IPI timer
+			 *  for now the interrupt is dismissed.
+			 */
 			goto out;
-		irq = (irq&7) + 8;
-	}
-retry:
+		}
+		break;
+	case _MACH_IBM:
+	case _MACH_Motorola:
+#if 1
+		outb(0x0C, 0x20);
+		irq = inb(0x20) & 7;
+		if (irq == 2)
+		{
+retry_cascade:
+			outb(0x0C, 0xA0);
+			irq = inb(0xA0);
+			/* if no intr left */
+			if ( !(irq & 128 ) )
+				goto out;
+			irq = (irq&7) + 8;
+		}
+		bits = 1UL << irq;
 #else
-	/* get the irr from the intr controller */
-	outb(0x0A, 0x20);
-	bits = inb(0x20);
-	/* handle cascade */
-	if ( bits  ) 
-	{
-		bits &= 4;
-		outb(0x0A, 0xA0);
-		bits = inb(0xA0)<<8;
+		/*
+		 * get the isr from the intr controller since
+		 * the bit in the irr has been cleared
+		 */
+		outb(0x0a, 0x20);
+		bits = inb(0x20)&0xff;
+		/* handle cascade */
+		if ( bits & 4 )
+		{
+			bits &= ~4UL;
+			outb(0x0a, 0xA0);
+			bits |= inb(0xA0)<<8;
+		}
+		/* ignore masked irqs */
+		bits &= ~cached_irq_mask;
+#endif		
+		break;
 	}
-	/* get lost interrupts */
-	bits |= lost_interrupts;
-	/* save intrs that are masked out */
-	lost_interrupts = bits & cached_irq_mask;
-	/* get rid of intrs being masked */
-	bits &= ~cached_irq_mask;
-        /* non-specifc eoi */
-	outb(0x20,0x20);
-	if ( bits & 0xff00 )
-		outb(0x20,0xA0);
-	
-	printk("bits %04X lost %04X mask %04x\n",
-	       bits, lost_interrupts,cached_irq_mask);
-	
-	for (irq = NR_IRQS; irq >= 0; --irq)
-		if (bits & (1U << irq))
-			break;
-#endif
-#endif	/* CONFIG_PMAC */
 	
 	if (irq < 0) {
-		printk("Bogus interrupt from PC = %lx, irq %d\n",regs->nip,irq);
+		printk("Bogus interrupt from PC = %lx\n", regs->nip);
 		goto out;
 	}
 
-#ifdef CONFIG_PMAC 
-	out_le32(IRQ_ACK, 1U << irq);
-#else /* CONFIG_PMAC */
-        /* mask out the irq while handling it */	
-	disable_irq(irq); 
-	/*
-	 * send eoi to interrupt controller right away or lower
-	 * priority intrs would be ignored even if with intrs enabled
-	 */
-	if (irq > 7)
-	{
-		outb(0xE0|(irq-8), 0xA0);
-		outb(0xE2, 0x20);
-	} else
-	{
-		outb(0xE0|irq, 0x20);
-	}
-#endif /* !CONFIG_PMAC */
+	mask_and_ack_irq(irq);
 
-	/*
-	 * now that we've acked the irq, if intrs are disabled in software
-	 * we're in the real-time system and non-rt linux has disabled them
-	 * so we just queue it up and return -- Cort
-	 */
-	if ( ! soft_intr_enable ) 
-	{
-		lost_interrupts |= 1UL << irq;
-		/* can't printk - kernel expects intrs off! */
-		/*printk("irq %d while intrs soft disabled\n", irq);*/
-		goto out;
-	}
-	
-	action = irq + irq_action;
+	status = 0;
+	action = irq_action + irq;
 	kstat.interrupts[irq]++;
-	if (action->handler) {
-		action->handler(irq, action->dev_id, regs);
-		_disable_interrupts(); /* in case the handler turned them on */
+	if ( action && action->handler)
+	{
+		if (!(action->flags & SA_INTERRUPT))
+			__sti();
+		status |= action->flags;
+			action->handler(irq, action->dev_id, regs);
+		/*if (status & SA_SAMPLE_RANDOM)
+			add_interrupt_randomness(irq);*/
+		__cli(); /* in case the handler turned them on */
+		spin_lock(&irq_controller_lock);
+		unmask_irq(irq);
+		spin_unlock(&irq_controller_lock);
 	} else {
+		if ( irq == 7 ) /* i8259 gives us irq 7 on 'short' intrs */
+			spurious_interrupts++;
 		disable_irq( irq );
 	}
-#ifdef CONFIG_PREP
-	/* re-enable if the interrupt was good and isn't one-shot */
-	if ( action->handler && !(action->flags & SA_ONESHOT) )
-		enable_irq(irq);
+	
 	/* make sure we don't miss any cascade intrs due to eoi-ing irq 2 */
-	if ( irq > 7 )
+	if ( is_prep && (irq > 7) )
 		goto retry_cascade;
-#endif
-	
-	hardirq_exit(cpu);
-	/*
-	 * This should be conditional: we should really get
-	 * a return code from the irq handler to tell us
-	 * whether the handler wants us to do software bottom
-	 * half handling or not..
-	 */
-	if (1)
-		if (bh_active & bh_mask)
-			do_bottom_half();
-        return;
+	/* do_bottom_half is called if necessary from int_return in head.S */
 out:
 	hardirq_exit(cpu);
-
 }
 
 int request_irq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *),
@@ -315,13 +345,13 @@ int request_irq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *)
 {
 	struct irqaction * action;
 	unsigned long flags;
-	
-#ifdef SHOW_IRQ	
+
+#ifdef SHOW_IRQ
 	printk("request_irq(): irq %d handler %08x name %s dev_id %04x\n",
 	       irq,handler,devname,dev_id);
 #endif /* SHOW_IRQ */
-	
-	if (irq > NR_IRQS)
+
+	if (irq >= NR_IRQS)
 		return -EINVAL;
 	action = irq + irq_action;
 	if (action->handler)
@@ -339,17 +369,16 @@ int request_irq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *)
 	restore_flags(flags);
 	return 0;
 }
-		
 void free_irq(unsigned int irq, void *dev_id)
 {
 	struct irqaction * action = irq + irq_action;
 	unsigned long flags;
 
-#ifdef SHOW_IRQ	
+#ifdef SHOW_IRQ
 	printk("free_irq(): irq %d dev_id %04x\n", irq, dev_id);
 #endif /* SHOW_IRQ */
-	
-	if (irq > NR_IRQS) {
+
+	if (irq >= NR_IRQS) {
 		printk("Trying to free IRQ%d\n",irq);
 		return;
 	}
@@ -370,114 +399,119 @@ void free_irq(unsigned int irq, void *dev_id)
 
 unsigned long probe_irq_on (void)
 {
-	unsigned int i, irqs = 0, irqmask;
-	unsigned long delay;
-
-	/* first, snaffle up any unassigned irqs */
-	for (i = 15; i > 0; i--) {
-		if (!request_irq(i, null_handler, SA_ONESHOT, "probe", NULL)) {
-			enable_irq(i);
-			irqs |= (1 << i);
-		}
-	}
-
-	/* wait for spurious interrupts to mask themselves out again */
-	for (delay = jiffies + 2; delay > jiffies; );	/* min 10ms delay */
-
-	/* now filter out any obviously spurious interrupts */
-	irqmask = (((unsigned int)cached_A1)<<8) | (unsigned int)cached_21;
-	for (i = 15; i > 0; i--) {
-		if (irqs & (1 << i) & irqmask) {
-			irqs ^= (1 << i);
-			free_irq(i, NULL);
-		}
-	}
-#ifdef DEBUG
-	printk("probe_irq_on:  irqs=0x%04x irqmask=0x%04x\n", irqs, irqmask);
-#endif
-	return irqs;
+	return 0;
 }
 
 int probe_irq_off (unsigned long irqs)
 {
-	unsigned int i, irqmask;
-
-	irqmask = (((unsigned int)cached_A1)<<8) | (unsigned int)cached_21;
-	for (i = 15; i > 0; i--) {
-		if (irqs & (1 << i)) {
-			free_irq(i, NULL);
-		}
-	}
-#ifdef SHOW_IRQ
-	printk("probe_irq_off: irqs=0x%04x irqmask=0x%04x\n", irqs, irqmask);
-#endif
-	irqs &= irqmask;
-	if (!irqs)
-		return 0;
-	i = ffz(~irqs);
-	if (irqs != (irqs & (1 << i)))
-		i = -i;
-	return i;
+	return 0;
 }
 
-void init_IRQ(void)
+__initfunc(static void i8259_init(void))
 {
-#ifdef CONFIG_PMAC  
-	extern void xmon_irq(int, void *, struct pt_regs *);
-
-	*IRQ_ENABLE = 0;
-	request_irq(KEYBOARD_IRQ, xmon_irq, 0, "NMI", 0);
-#else /* CONFIG_PMAC */
-	/* Initialize interrupt controllers */
+	/* init master interrupt controller */
 	outb(0x11, 0x20); /* Start init sequence */
 	outb(0x40, 0x21); /* Vector base */
-#if 1
-	outb(0x04, 0x21); /* edge tiggered, Cascade (slave) on IRQ2 */
-#else
+	/*outb(0x04, 0x21);*/ /* edge tiggered, Cascade (slave) on IRQ2 */
 	outb(0x0C, 0x21); /* level triggered, Cascade (slave) on IRQ2 */
-#endif	
-	
 	outb(0x01, 0x21); /* Select 8086 mode */
 	outb(0xFF, 0x21); /* Mask all */
-
+	
+	/* init slave interrupt controller */
 	outb(0x11, 0xA0); /* Start init sequence */
 	outb(0x48, 0xA1); /* Vector base */
-#if 1
-	outb(0x02, 0xA1); /* edge triggered, Cascade (slave) on IRQ2 */
-#else
+	/*outb(0x02, 0xA1);*/ /* edge triggered, Cascade (slave) on IRQ2 */
 	outb(0x0A, 0x21); /* level triggered, Cascade (slave) on IRQ2 */
-#endif
 	outb(0x01, 0xA1); /* Select 8086 mode */
 	outb(0xFF, 0xA1); /* Mask all */
-
-	/*
-	 * Program level mode for irq's that 'can' be level triggered.
-	 * This does not effect irq's 0,1,2 and 8 since they must be
-	 * edge triggered. This is not the PIC controller.  The default
-	 * here is for edge trigger mode.  -- Cort
-	 */
-#if 0	
-	outb(0xf8, 0x4d0); /* level triggered */
-	outb(0xff, 0x4d1);
-#endif
-#if 0	
-	outb(0x00, 0x4D0); /* All edge triggered */
-	outb(0xCF, 0x4D1); /* Trigger mode */
-#endif
 	outb(cached_A1, 0xA1);
 	outb(cached_21, 0x21);
-	if (request_irq(2, null_handler, SA_INTERRUPT, "cascade", NULL))
-		printk("Unable to get IRQ2 for cascade\n");
+	if (request_irq(2, no_action, SA_INTERRUPT, "cascade", NULL) != 0)
+		panic("Could not allocate cascade IRQ!");
 	enable_irq(2);  /* Enable cascade interrupt */
-	
-#define TIMER0_COUNT 0x40
-#define TIMER_CONTROL 0x43
-	/* set timer to periodic mode */
-	outb_p(0x34,TIMER_CONTROL);		/* binary, mode 2, LSB/MSB, ch 0 */
-	/* set the clock to ~100 Hz */
-	outb_p(LATCH & 0xff , TIMER0_COUNT);	/* LSB */
-	outb(LATCH >> 8 , TIMER0_COUNT);	/* MSB */
-
-	route_pci_interrupts();
-#endif	/* CONFIG_PMAC */
+}
+
+__initfunc(void init_IRQ(void))
+{
+	extern void xmon_irq(int, void *, struct pt_regs *);
+
+	switch (_machine)
+	{
+	case _MACH_Pmac:
+		mask_and_ack_irq = pmac_mask_and_ack_irq;
+		set_irq_mask = pmac_set_irq_mask;
+		
+		*IRQ_ENABLE = 0;
+#ifdef CONFIG_XMON
+		request_irq(KEYBOARD_IRQ, xmon_irq, 0, "NMI", 0);
+#endif	/* CONFIG_XMON */
+		break;
+	case _MACH_Motorola:
+	case _MACH_IBM:
+		mask_and_ack_irq = i8259_mask_and_ack_irq;
+		set_irq_mask = i8259_set_irq_mask;
+		
+		i8259_init();
+		route_pci_interrupts();
+		/*
+		 * According to the Carolina spec from ibm irq's 0,1,2, and 8
+		 * must be edge triggered.  Also, the pci intrs must be level
+		 * triggered and _only_ isa intrs can be level sensitive
+		 * which are 3-7,9-12,14-15. 13 is special - it can be level.
+		 *
+		 * power on default is 0's in both regs - all edge.
+		 *
+		 * These edge/level control regs allow edge/level status
+		 * to be decided on a irq basis instead of on a PIC basis.
+		 * It's still pretty ugly.
+		 * - Cort
+		 */
+		{
+			unsigned char irq_mode1 = 0, irq_mode2 = 0;
+			irq_mode1 = 0; /* to get rid of compiler warnings */
+			/*
+			 * On Carolina, irq 15 and 13 must be level (scsi/ide/net).
+			 */
+			if ( _machine == _MACH_IBM )
+				irq_mode2 |= 0xa0;
+			/*
+			 * Sound on the Powerstack reportedly needs to be edge triggered
+			 */
+			if ( _machine == _MACH_Motorola )
+			{
+				/*irq_mode2 &= ~0x04L;
+				outb( irq_mode1 , 0x4d0 );
+				outb( irq_mode2 , 0x4d1 );*/
+			}
+
+		}
+		break;
+	case _MACH_chrp:
+		mask_and_ack_irq = chrp_mask_and_ack_irq;
+		set_irq_mask = chrp_set_irq_mask;
+		if ((Hydra = find_hydra())) {
+			printk("Hydra Mac I/O at %p\n", Hydra);
+			out_le32(&Hydra->Feature_Control, HYDRA_FC_SCC_CELL_EN |
+				 HYDRA_FC_SCSI_CELL_EN |
+				 HYDRA_FC_SCCA_ENABLE |
+				 HYDRA_FC_SCCB_ENABLE |
+				 HYDRA_FC_ARB_BYPASS |
+				 HYDRA_FC_MPIC_ENABLE |
+				 HYDRA_FC_SLOW_SCC_PCLK |
+				 HYDRA_FC_MPIC_IS_MASTER);
+			OpenPIC = (volatile struct OpenPIC *)&Hydra->OpenPIC;
+		} else if (!OpenPIC /* && find_xxx */) {
+			printk("Unknown openpic implementation\n");
+			/* other OpenPIC implementations */
+			/* ... */
+ 		}
+		if (OpenPIC)
+		    openpic_init();
+		else
+		    panic("No OpenPIC found");
+		if (Hydra)
+		    hydra_post_openpic_init();
+		i8259_init();
+		break;
+	}
 }
diff --git a/arch/ppc/kernel/misc.S b/arch/ppc/kernel/misc.S
index 430cd7a5d..9c6d013b6 100644
--- a/arch/ppc/kernel/misc.S
+++ b/arch/ppc/kernel/misc.S
@@ -64,6 +64,7 @@ _GLOBAL(_hard_sti)
 	mtmsr	r3		/* Update machine state */
 	blr
 
+#if 0
 /*
  * Restore 'flags'
  *	__restore_flags(long val)
@@ -79,11 +80,13 @@ _GLOBAL(__restore_flags)
 	mtmsr	r3
 	isync
 	blr
-
+#endif
+	
 /*
  * We were about to enable interrupts but we have to simulate
  * some interrupts that were lost by enable_irq first.
  */
+	.globl do_lost_interrupts
 do_lost_interrupts:
 	stwu	r1,-16(r1)
 	mflr	r0
@@ -243,7 +246,6 @@ _GLOBAL(_outsl)
 	bdnz	00b
 	blr	
 
-#ifdef CONFIG_PMAC
 _GLOBAL(ide_insw)
 	mtctr	r5
 	subi	r4,r4,2
@@ -259,7 +261,6 @@ _GLOBAL(ide_outsw)
 	sthx	r5,0,r3	
 	bdnz	00b
 	blr	
-#endif
 
 /*
  * Extended precision shifts
@@ -298,7 +299,7 @@ _GLOBAL(abs)
 _GLOBAL(_get_SP)
 	mr	r3,r1		/* Close enough */
 	blr
-	
+
 _GLOBAL(_get_PVR)
 	mfspr	r3,PVR
 	blr
@@ -308,6 +309,7 @@ cvt_fd:
 	lfs	0,0(r3)
 	stfd	0,0(r4)
 	blr
+
 /*
  * Fetch the current SR register
  *   get_SR(int index)
@@ -317,7 +319,6 @@ _GLOBAL(get_SR)
 	mr	r3,r4
 	blr
 
-	
 _GLOBAL(cvt_df)
 cvt_df:
 	lfd	0,0(r3)
diff --git a/arch/ppc/kernel/mk_defs.c b/arch/ppc/kernel/mk_defs.c
index 3f404a37a..80f10c59e 100644
--- a/arch/ppc/kernel/mk_defs.c
+++ b/arch/ppc/kernel/mk_defs.c
@@ -29,7 +29,6 @@
 void
 main(void)
 {
-	/*DEFINE(KERNELBASE, KERNELBASE);*/
 	DEFINE(STATE, offsetof(struct task_struct, state));
 	DEFINE(NEXT_TASK, offsetof(struct task_struct, next_task));
 	DEFINE(COUNTER, offsetof(struct task_struct, counter));
@@ -37,54 +36,20 @@ main(void)
 	DEFINE(SIGNAL, offsetof(struct task_struct, signal));
 	DEFINE(TSS, offsetof(struct task_struct, tss));
 	DEFINE(KSP, offsetof(struct thread_struct, ksp));
-	DEFINE(PG_TABLES, offsetof(struct thread_struct, pg_tables));
-#ifdef CONFIG_PMAC	
-	DEFINE(LAST_PC, offsetof(struct thread_struct, last_pc));
-	DEFINE(USER_STACK, offsetof(struct thread_struct, user_stack));
-#endif	
+	/*DEFINE(PG_TABLES, offsetof(struct thread_struct, pg_tables));*/
+	DEFINE(MM, offsetof(struct task_struct, mm));
+	DEFINE(PGD, offsetof(struct mm_struct, pgd));
 	DEFINE(LAST_SYSCALL, offsetof(struct thread_struct, last_syscall));
 	DEFINE(PT_REGS, offsetof(struct thread_struct, regs));
 	DEFINE(PF_TRACESYS, PF_TRACESYS);
 	DEFINE(TASK_FLAGS, offsetof(struct task_struct, flags));
 	DEFINE(TSS_FPR0, offsetof(struct thread_struct, fpr[0]));
-#if 0
-	DEFINE(TSS_FPR1, offsetof(struct thread_struct, fpr[1]));
-	DEFINE(TSS_FPR2, offsetof(struct thread_struct, fpr[2]));
-	DEFINE(TSS_FPR3, offsetof(struct thread_struct, fpr[3]));
-	DEFINE(TSS_FPR4, offsetof(struct thread_struct, fpr[4]));
-	DEFINE(TSS_FPR5, offsetof(struct thread_struct, fpr[5]));
-	DEFINE(TSS_FPR6, offsetof(struct thread_struct, fpr[6]));
-	DEFINE(TSS_FPR7, offsetof(struct thread_struct, fpr[7]));
-	DEFINE(TSS_FPR8, offsetof(struct thread_struct, fpr[8]));
-	DEFINE(TSS_FPR9, offsetof(struct thread_struct, fpr[9]));
-	DEFINE(TSS_FPR10, offsetof(struct thread_struct, fpr[10]));
-	DEFINE(TSS_FPR11, offsetof(struct thread_struct, fpr[11]));
-	DEFINE(TSS_FPR12, offsetof(struct thread_struct, fpr[12]));
-	DEFINE(TSS_FPR13, offsetof(struct thread_struct, fpr[13]));
-	DEFINE(TSS_FPR14, offsetof(struct thread_struct, fpr[14]));
-	DEFINE(TSS_FPR15, offsetof(struct thread_struct, fpr[15]));
-	DEFINE(TSS_FPR16, offsetof(struct thread_struct, fpr[16]));
-	DEFINE(TSS_FPR17, offsetof(struct thread_struct, fpr[17]));
-	DEFINE(TSS_FPR18, offsetof(struct thread_struct, fpr[18]));
-	DEFINE(TSS_FPR19, offsetof(struct thread_struct, fpr[19]));
-	DEFINE(TSS_FPR20, offsetof(struct thread_struct, fpr[20]));
-	DEFINE(TSS_FPR21, offsetof(struct thread_struct, fpr[21]));
-	DEFINE(TSS_FPR22, offsetof(struct thread_struct, fpr[22]));
-	DEFINE(TSS_FPR23, offsetof(struct thread_struct, fpr[23]));
-	DEFINE(TSS_FPR24, offsetof(struct thread_struct, fpr[24]));
-	DEFINE(TSS_FPR25, offsetof(struct thread_struct, fpr[25]));
-	DEFINE(TSS_FPR26, offsetof(struct thread_struct, fpr[26]));
-	DEFINE(TSS_FPR27, offsetof(struct thread_struct, fpr[27]));
-	DEFINE(TSS_FPR28, offsetof(struct thread_struct, fpr[28]));
-	DEFINE(TSS_FPR29, offsetof(struct thread_struct, fpr[29]));
-	DEFINE(TSS_FPR30, offsetof(struct thread_struct, fpr[30]));
-	DEFINE(TSS_FPR31, offsetof(struct thread_struct, fpr[31]));
-#endif
 	DEFINE(TSS_FPSCR, offsetof(struct thread_struct, fpscr));
 	/* Interrupt register frame */
 	DEFINE(TASK_UNION_SIZE, sizeof(union task_union));
 	DEFINE(STACK_FRAME_OVERHEAD, STACK_FRAME_OVERHEAD);
 	DEFINE(INT_FRAME_SIZE, STACK_FRAME_OVERHEAD + sizeof(struct pt_regs));
+	/* in fact we only use gpr0 - gpr9 and gpr20 - gpr23 */
 	DEFINE(GPR0, STACK_FRAME_OVERHEAD+offsetof(struct pt_regs, gpr[0]));
 	DEFINE(GPR1, STACK_FRAME_OVERHEAD+offsetof(struct pt_regs, gpr[1]));
 	DEFINE(GPR2, STACK_FRAME_OVERHEAD+offsetof(struct pt_regs, gpr[2]));
diff --git a/arch/ppc/kernel/openpic.c b/arch/ppc/kernel/openpic.c
new file mode 100644
index 000000000..7eb5e5aa9
--- /dev/null
+++ b/arch/ppc/kernel/openpic.c
@@ -0,0 +1,546 @@
+/*
+ *  arch/ppc/kernel/openpic.c -- 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.
+ */
+
+
+/*
+ *  Note: Interprocessor Interrupt (IPI) and Timer support is incomplete
+ */
+
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/openpic.h>
+#include <asm/ptrace.h>
+#include <asm/signal.h>
+#include <asm/io.h>
+#include <asm/irq.h>
+
+
+#define REGISTER_DEBUG
+#undef REGISTER_DEBUG
+
+
+#define VEC_TIMER	0x40	/* and up */
+#define VEC_IPI		0x50	/* and up */
+#define VEC_SOURCE	0x10	/* and up */
+#define VEC_SPURIOUS	99
+
+
+volatile struct OpenPIC *OpenPIC;
+
+static u_int Version;
+static u_int NumProcessors;
+static u_int NumSources;
+static u_int VendorID;
+static u_int DeviceID;
+static u_int Stepping;
+static u_int TimerFrequency;
+
+
+    /*
+     *  Accesses to the current processor's registers
+     */
+
+#ifndef __powerpc__
+#define THIS_CPU		Private
+#define CHECK_THIS_CPU		do {} while (0)
+#else
+#define THIS_CPU		Processor[cpu]
+#define CHECK_THIS_CPU		check_arg_cpu(cpu)
+#endif
+
+
+    /*
+     *  Sanity checks
+     */
+
+#if 1
+#define check_arg_ipi(ipi) \
+    if (ipi < 0 || ipi >= OPENPIC_NUM_IPI) \
+	printk("openpic.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);
+#define check_arg_vec(vec) \
+    if (vec < 0 || vec >= OPENPIC_NUM_VECTORS) \
+	printk("openpic.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);
+#define check_arg_irq(irq) \
+    if (irq < 0 || irq >= NumSources) \
+	printk("openpic.c:%d: illegal irq %d\n", __LINE__, irq);
+#define check_arg_cpu(cpu) \
+    if (cpu < 0 || cpu >= NumProcessors) \
+	printk("openpic.c:%d: illegal cpu %d\n", __LINE__, cpu);
+#else
+#define check_arg_ipi(ipi)	do {} while (0)
+#define check_arg_timer(timer)	do {} while (0)
+#define check_arg_vec(vec)	do {} while (0)
+#define check_arg_pri(pri)	do {} while (0)
+#define check_arg_irq(irq)	do {} while (0)
+#define check_arg_cpu(cpu)	do {} while (0)
+#endif
+
+
+    /*
+     *  Dummy interrupt handler
+     */
+
+static void no_action(int ir1, void *dev, struct pt_regs *regs)
+{}
+
+
+    /*
+     *  I/O functions
+     */
+
+#ifdef __i386__
+static inline u_int ld_le32(volatile u_int *addr)
+{
+    return *addr;
+}
+
+static inline void out_le32(volatile u_int *addr, u_int val)
+{
+    *addr = val;
+}
+#endif
+
+static inline u_int openpic_read(volatile u_int *addr)
+{
+    u_int val;
+
+    val = ld_le32(addr);
+#ifdef REGISTER_DEBUG
+    printk("openpic_read(0x%08x) = 0x%08x\n", (u_int)addr, val);
+#endif
+    return val;
+}
+
+static inline void openpic_write(volatile u_int *addr, u_int val)
+{
+#ifdef REGISTER_DEBUG
+    printk("openpic_write(0x%08x, 0x%08x)\n", (u_int)addr, val);
+#endif
+    out_le32(addr, val);
+}
+
+
+static inline u_int openpic_readfield(volatile u_int *addr, u_int mask)
+{
+    u_int val = openpic_read(addr);
+    return val & mask;
+}
+
+static inline void openpic_writefield(volatile u_int *addr, u_int mask,
+				      u_int field)
+{
+    u_int val = openpic_read(addr);
+    openpic_write(addr, (val & ~mask) | (field & mask));
+}
+
+static inline void openpic_clearfield(volatile u_int *addr, u_int mask)
+{
+    openpic_writefield(addr, mask, 0);
+}
+
+static inline void openpic_setfield(volatile u_int *addr, u_int mask)
+{
+    openpic_writefield(addr, mask, mask);
+}
+
+
+    /*
+     *  Update a Vector/Priority register in a safe manner. The interrupt will
+     *  be disabled.
+     */
+
+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 */
+    while (openpic_read(addr) & OPENPIC_ACTIVITY);
+    openpic_writefield(addr, mask | OPENPIC_MASK, field | OPENPIC_MASK);
+}
+
+
+/* -------- Global Operations ---------------------------------------------- */
+
+
+    /*
+     *  Initialize the OpenPIC
+     */
+
+void openpic_init(void)
+{
+    u_int t, i;
+    const char *version, *vendor, *device;
+
+    if (!OpenPIC) {
+	printk("No OpenPIC present\n");
+	return;
+    }
+
+    t = openpic_read(&OpenPIC->Global.Feature_Reporting0);
+    Version = t & OPENPIC_FEATURE_VERSION_MASK;
+    NumProcessors = ((t & OPENPIC_FEATURE_LAST_PROCESSOR_MASK) >>
+		     OPENPIC_FEATURE_LAST_PROCESSOR_SHIFT) + 1;
+    NumSources = ((t & OPENPIC_FEATURE_LAST_SOURCE_MASK) >>
+		  OPENPIC_FEATURE_LAST_SOURCE_SHIFT) + 1;
+
+    switch (Version) {
+	case 1:
+	    version = "1.0";
+	    break;
+	case 2:
+	    version = "1.2";
+	    break;
+	default:
+	    version = "?.?";
+	    break;
+    }
+    printk("OpenPIC Version %s (%d CPUs and %d IRQ sources) at %p\n", version,
+	   NumProcessors, NumSources, OpenPIC);
+
+    t = openpic_read(&OpenPIC->Global.Vendor_Identification);
+    VendorID = t & OPENPIC_VENDOR_ID_VENDOR_ID_MASK;
+    DeviceID = (t & OPENPIC_VENDOR_ID_DEVICE_ID_MASK) >>
+	       OPENPIC_VENDOR_ID_DEVICE_ID_SHIFT;
+    Stepping = (t & OPENPIC_VENDOR_ID_STEPPING_MASK) >>
+	       OPENPIC_VENDOR_ID_STEPPING_SHIFT;
+    switch (VendorID) {
+	case OPENPIC_VENDOR_ID_APPLE:
+	    vendor = "Apple";
+	    break;
+	default:
+	    vendor = "Unknown";
+	    break;
+    }
+    switch (DeviceID) {
+	case OPENPIC_DEVICE_ID_APPLE_HYDRA:
+	    device = "Hydra";
+	    break;
+	default:
+	    device = "Unknown";
+	    break;
+    }
+    printk("OpenPIC Vendor %d (%s), Device %d (%s), Stepping %d\n", VendorID,
+	   vendor, DeviceID, device, Stepping);
+
+    TimerFrequency = openpic_read(&OpenPIC->Global.Timer_Frequency);
+    printk("OpenPIC timer frequency is ");
+    if (TimerFrequency)
+	printk("%d Hz\n", TimerFrequency);
+    else
+	printk("not set\n");
+
+    /* Initialize timer interrupts */
+    for (i = 0; i < OPENPIC_NUM_TIMERS; i++) {
+	/* Disabled, Priority 0 */
+	openpic_inittimer(i, 0, VEC_TIMER+i);
+	/* No processor */
+	openpic_maptimer(i, 0);
+    }
+
+    /* Initialize IPI interrupts */
+    for (i = 0; i < OPENPIC_NUM_IPI; i++) {
+	/* Disabled, Priority 0 */
+	openpic_initipi(i, 0, VEC_IPI+i);
+    }
+
+    /* Initialize external interrupts */
+    /* SIOint (8259 cascade) is special */
+    openpic_initirq(0, 8, VEC_SOURCE, 1, 1); /* 0,1 gives interrupt storm */
+    /* Processor 0 */
+    openpic_mapirq(0, 1<<0);
+    for (i = 1; i < NumSources; i++) {
+	/* Enabled, Priority 8 */
+	openpic_initirq(i, 8, VEC_SOURCE+i, 0, 1);
+	/* Processor 0 */
+	openpic_mapirq(i, 1<<0);
+    }
+
+    /* Initialize the spurious interrupt */
+    openpic_set_spurious(VEC_SPURIOUS);
+
+    if (request_irq(IRQ_8259_CASCADE, no_action, SA_INTERRUPT,
+		    "OpenPIC cascade", NULL))
+	printk("Unable to get OpenPIC IRQ 0 for cascade\n");
+    openpic_set_priority(0, 0);
+    openpic_disable_8259_pass_through();
+}
+
+
+    /*
+     *  Reset the OpenPIC
+     */
+
+void openpic_reset(void)
+{
+    openpic_setfield(&OpenPIC->Global.Global_Configuration0,
+    		       OPENPIC_CONFIG_RESET);
+}
+
+
+    /*
+     *  Enable/disable 8259 Pass Through Mode
+     */
+
+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)
+{
+    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)
+{
+    u_int vec;
+
+    check_arg_cpu(cpu);
+    vec = openpic_readfield(&OpenPIC->THIS_CPU.Interrupt_Acknowledge,
+			    OPENPIC_VECTOR_MASK);
+#if 0
+if (vec != 22 /* SCSI */)
+printk("++openpic_irq: %d\n", vec);
+#endif
+    return vec;
+}
+#endif
+
+
+    /*
+     *  Signal end of interrupt (EOI) processing
+     */
+
+#ifndef __powerpc__
+void openpic_eoi(void)
+#else
+void openpic_eoi(u_int cpu)
+#endif
+{
+#if 0
+printk("++openpic_eoi:\n");
+#endif
+    check_arg_cpu(cpu);
+    openpic_write(&OpenPIC->THIS_CPU.EOI, 0);
+}
+
+
+    /*
+     *  Get/set the current task priority
+     */
+
+#ifndef __powerpc__
+u_int openpic_get_priority(void)
+#else
+u_int openpic_get_priority(u_int cpu)
+#endif
+{
+    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
+{
+    CHECK_THIS_CPU;
+    check_arg_pri(pri);
+    openpic_writefield(&OpenPIC->THIS_CPU.Current_Task_Priority,
+    		       OPENPIC_CURRENT_TASK_PRIORITY_MASK, pri);
+}
+
+    /*
+     *  Get/set the spurious vector
+     */
+
+u_int openpic_get_spurious(void)
+{
+    return openpic_readfield(&OpenPIC->Global.Spurious_Vector,
+    			     OPENPIC_VECTOR_MASK);
+}
+
+void openpic_set_spurious(u_int vec)
+{
+    check_arg_vec(vec);
+    openpic_writefield(&OpenPIC->Global.Spurious_Vector, OPENPIC_VECTOR_MASK,
+    		       vec);
+}
+
+
+    /*
+     *  Initialize one or more CPUs
+     */
+
+void openpic_init_processor(u_int cpumask)
+{
+    openpic_write(&OpenPIC->Global.Processor_Initialization, cpumask);
+}
+
+
+/* -------- Interprocessor Interrupts -------------------------------------- */
+
+
+    /*
+     *  Initialize an interprocessor interrupt (and disable it)
+     *
+     *  ipi: OpenPIC interprocessor interrupt number
+     *  pri: interrupt source priority
+     *  vec: the vector it will produce
+     */
+
+void openpic_initipi(u_int ipi, u_int pri, u_int vec)
+{
+    check_arg_timer(ipi);
+    check_arg_pri(pri);
+    check_arg_vec(vec);
+    openpic_safe_writefield(&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
+     */
+
+#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
+{
+    CHECK_THIS_CPU;
+    check_arg_ipi(ipi);
+    openpic_write(&OpenPIC->THIS_CPU.IPI_Dispatch(ipi), cpumask);
+}
+
+
+/* -------- Timer Interrupts ----------------------------------------------- */
+
+
+    /*
+     *  Initialize a timer interrupt (and disable it)
+     *
+     *  timer: OpenPIC timer number
+     *  pri: interrupt source priority
+     *  vec: the vector it will produce
+     */
+
+void openpic_inittimer(u_int timer, u_int pri, u_int vec)
+{
+    check_arg_timer(timer);
+    check_arg_pri(pri);
+    check_arg_vec(vec);
+    openpic_safe_writefield(&OpenPIC->Global.Timer[timer].Vector_Priority,
+    			    OPENPIC_PRIORITY_MASK | OPENPIC_VECTOR_MASK,
+    			    (pri << OPENPIC_PRIORITY_SHIFT) | vec);
+}
+
+
+    /*
+     *  Map a timer interrupt to one or more CPUs
+     */
+
+void openpic_maptimer(u_int timer, u_int cpumask)
+{
+    check_arg_timer(timer);
+    openpic_write(&OpenPIC->Global.Timer[timer].Destination, cpumask);
+}
+
+
+/* -------- Interrupt Sources ---------------------------------------------- */
+
+
+    /*
+     *  Enable/disable an interrupt source
+     */
+
+void openpic_enable_irq(u_int irq)
+{
+    check_arg_irq(irq);
+    openpic_clearfield(&OpenPIC->Source[irq].Vector_Priority, OPENPIC_MASK);
+}
+
+void openpic_disable_irq(u_int irq)
+{
+    check_arg_irq(irq);
+    openpic_setfield(&OpenPIC->Source[irq].Vector_Priority, OPENPIC_MASK);
+}
+
+
+    /*
+     *  Initialize an interrupt source (and disable it!)
+     *
+     *  irq: OpenPIC interrupt number
+     *  pri: interrupt source priority
+     *  vec: the vector it will produce
+     *  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)
+{
+    check_arg_irq(irq);
+    check_arg_pri(pri);
+    check_arg_vec(vec);
+    openpic_safe_writefield(&OpenPIC->Source[irq].Vector_Priority,
+    			    OPENPIC_PRIORITY_MASK | OPENPIC_VECTOR_MASK |
+    			    OPENPIC_SENSE_POLARITY | OPENPIC_SENSE_LEVEL,
+    			    (pri << OPENPIC_PRIORITY_SHIFT) | vec |
+			    (pol ? OPENPIC_SENSE_POLARITY : 0) |
+			    (sense ? OPENPIC_SENSE_LEVEL : 0));
+}
+
+
+    /*
+     *  Map an interrupt source to one or more CPUs
+     */
+
+void openpic_mapirq(u_int irq, u_int cpumask)
+{
+    check_arg_irq(irq);
+    openpic_write(&OpenPIC->Source[irq].Destination, cpumask);
+}
+
+
+    /*
+     *  Set the sense for an interrupt source (and disable it!)
+     *
+     *  sense: 1 for level, 0 for edge
+     */
+
+void openpic_set_sense(u_int irq, int sense)
+{
+    check_arg_irq(irq);
+    openpic_safe_writefield(&OpenPIC->Source[irq].Vector_Priority,
+    			    OPENPIC_SENSE_LEVEL,
+			    (sense ? OPENPIC_SENSE_LEVEL : 0));
+}
+
+
diff --git a/arch/ppc/kernel/pci-bridge.c b/arch/ppc/kernel/pci-bridge.c
new file mode 100644
index 000000000..0e4420340
--- /dev/null
+++ b/arch/ppc/kernel/pci-bridge.c
@@ -0,0 +1,428 @@
+/*
+ * Support for PCI bridges found on Power Macintoshes.
+ * At present the "bandit" and "chaos" bridges are supported.
+ * Fortunately you access configuration space in the same
+ * way with either bridge.
+ *
+ * Copyright (C) 1997 Paul Mackerras (paulus@cs.anu.edu.au)
+ *
+ * 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.
+ */
+
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/bios32.h>
+#include <linux/delay.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/pci-bridge.h>
+
+struct bridge_data {
+	volatile unsigned int *cfg_addr;
+	volatile unsigned char *cfg_data;
+	void *io_base;
+	int bus_number;
+	int max_bus;
+	struct bridge_data *next;
+	struct device_node *node;
+};
+
+static struct bridge_data **bridges, *bridge_list;
+static int max_bus;
+
+static void add_bridges(struct device_node *dev, unsigned long *mem_ptr);
+
+/*
+ * Magic constants for enabling cache coherency in the bandit/PSX bridge.
+ */
+#define APPLE_VENDID	0x106b
+#define BANDIT_DEVID	1
+#define BANDIT_REVID	3
+
+#define BANDIT_DEVNUM	11
+#define BANDIT_MAGIC	0x50
+#define BANDIT_COHERENT	0x40
+
+/*
+ * 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.
+ */
+static void init_bandit(struct bridge_data *bp)
+{
+	unsigned int vendev, magic;
+	int rev;
+
+	/* read the word at offset 0 in config space for device 11 */
+	out_le32(bp->cfg_addr, (1UL << BANDIT_DEVNUM) + PCI_VENDOR_ID);
+	udelay(2);
+	vendev = in_le32((volatile unsigned int *)bp->cfg_data);
+	if (vendev != (BANDIT_DEVID << 16) + APPLE_VENDID) {
+		printk(KERN_WARNING "bandit isn't? (%x)\n", vendev);
+		return;
+	}
+
+	/* read the revision id */
+	out_le32(bp->cfg_addr, (1UL << BANDIT_DEVNUM) + PCI_REVISION_ID);
+	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);
+
+	/* read the word at offset 0x50 */
+	out_le32(bp->cfg_addr, (1UL << BANDIT_DEVNUM) + BANDIT_MAGIC);
+	udelay(2);
+	magic = in_le32((volatile unsigned int *)bp->cfg_data);
+	if ((magic & BANDIT_COHERENT) != 0)
+		return;
+	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);
+}
+
+unsigned long pmac_find_bridges(unsigned long mem_start, unsigned long mem_end)
+{
+	int bus;
+	struct bridge_data *bridge;
+
+	bridge_list = 0;
+	max_bus = 0;
+	add_bridges(find_devices("bandit"), &mem_start);
+	add_bridges(find_devices("chaos"), &mem_start);
+	bridges = (struct bridge_data **) mem_start;
+	mem_start += (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;
+
+	return mem_start;
+}
+
+static void add_bridges(struct device_node *dev, unsigned long *mem_ptr)
+{
+	int *bus_range;
+	int len;
+	struct bridge_data *bp;
+
+	for (; dev != NULL; dev = dev->next) {
+		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->name, dev->addrs[0].address);
+		bp = (struct bridge_data *) *mem_ptr;
+		*mem_ptr += sizeof(struct bridge_data);
+		bp->cfg_addr = (volatile unsigned int *)
+			(dev->addrs[0].address + 0x800000);
+		bp->cfg_data = (volatile unsigned char *)
+			(dev->addrs[0].address + 0xc00000);
+		bp->io_base = (void *) dev->addrs[0].address;
+		ioremap(dev->addrs[0].address, 0x800000);
+		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);
+	}
+}
+
+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;
+}
+
+int pci_device_loc(struct device_node *dev, unsigned char *bus_ptr,
+		   unsigned char *devfn_ptr)
+{
+	unsigned int *reg;
+	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;
+	}
+	*bus_ptr = reg[0] >> 16;
+	*devfn_ptr = reg[0] >> 8;
+	return 0;
+}
+
+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 {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + (offset & ~3) + 1);
+	}
+	udelay(2);
+	*val = in_8(bp->cfg_data + (offset & 3));
+
+	if (offset == PCI_INTERRUPT_LINE) {
+		/*
+		 * Open Firmware often doesn't initialize this
+		 * register properly, so we find the node and see
+		 * if it has an AAPL,interrupts property.
+		 */
+		struct device_node *node;
+		unsigned int *reg;
+
+		for (node = bp->node->child; node != 0; node = node->sibling) {
+			reg = (unsigned int *) get_property(node, "reg", 0);
+			if (reg == 0 || ((reg[0] >> 8) & 0xff) != dev_fn)
+				continue;
+			/* this is the node, see if it has interrupts */
+			if (node->n_intrs > 0)
+				*val = node->intrs[0];
+			break;
+		}
+	}
+
+	return PCIBIOS_SUCCESSFUL;
+}
+
+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 || (offset & 1) != 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 {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + (offset & ~3) + 1);
+	}
+	udelay(2);
+	*val = in_le16((volatile unsigned short *)(bp->cfg_data + (offset & 3)));
+	return PCIBIOS_SUCCESSFUL;
+}
+
+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 || (offset & 3) != 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);
+	} else {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + offset + 1);
+	}
+	udelay(2);
+	*val = in_le32((volatile unsigned int *)bp->cfg_data);
+	return PCIBIOS_SUCCESSFUL;
+}
+
+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) {
+		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 {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + (offset & ~3) + 1);
+	}
+	udelay(2);
+	out_8(bp->cfg_data + (offset & 3), val);
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pmac_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn,
+				   unsigned char offset, unsigned short val)
+{
+	struct bridge_data *bp;
+
+	if (bus > max_bus || (bp = bridges[bus]) == 0 || (offset & 1) != 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 {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + (offset & ~3) + 1);
+	}
+	udelay(2);
+	out_le16((volatile unsigned short *)(bp->cfg_data + (offset & 3)), val);
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pmac_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn,
+				    unsigned char offset, unsigned int val)
+{
+	struct bridge_data *bp;
+
+	if (bus > max_bus || (bp = bridges[bus]) == 0 || (offset & 3) != 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);
+	} else {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + offset + 1);
+	}
+	udelay(2);
+	out_le32((volatile unsigned int *)bp->cfg_data, val);
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pmac_pcibios_find_device(unsigned short vendor, unsigned short dev_id,
+			     unsigned short index, unsigned char *bus_ptr,
+			     unsigned char *dev_fn_ptr)
+{
+	int bus, unit, fn, num, devfn;
+	unsigned int x, vendev;
+	unsigned char h;
+
+	if (vendor == 0xffff)
+		return PCIBIOS_BAD_VENDOR_ID;
+	vendev = (dev_id << 16) + vendor;
+	num = 0;
+	for (bus = 0; bus <= max_bus; ++bus) {
+		if (bridges[bus] == 0)
+			continue;
+		unit = fn = 0;
+		if (bus == bridges[bus]->bus_number)
+			unit = 11;
+		while (unit < 32) {
+			devfn = PCI_DEVFN(unit, fn);
+			if (pcibios_read_config_dword(bus, devfn,
+						      PCI_VENDOR_ID, &x)
+			    == PCIBIOS_SUCCESSFUL && x == vendev) {
+				if (index == num) {
+					*bus_ptr = bus;
+					*dev_fn_ptr = devfn;
+					return PCIBIOS_SUCCESSFUL;
+				}
+				++num;
+			}
+			if (fn != 0) {
+				if (++fn >= 8) {
+					++unit;
+					fn = 0;
+				}
+				continue;
+			}
+			if (pcibios_read_config_byte(bus, devfn,
+						     PCI_HEADER_TYPE, &h)
+			    == PCIBIOS_SUCCESSFUL && (h & 0x80) != 0)
+				++fn;
+			else
+				++unit;
+		}
+	}
+	return PCIBIOS_DEVICE_NOT_FOUND;
+}
+
+int pmac_pcibios_find_class(unsigned int class_code, unsigned short index,
+			    unsigned char *bus_ptr, unsigned char *dev_fn_ptr)
+{
+	int bus, unit, fn, num, devfn;
+	unsigned int x;
+	unsigned char h;
+
+	num = 0;
+	for (bus = 0; bus <= max_bus; ++bus) {
+		if (bridges[bus] == 0)
+			continue;
+		unit = fn = 0;
+		if (bus == bridges[bus]->bus_number)
+			unit = 11;
+		while (unit < 32) {
+			devfn = PCI_DEVFN(unit, fn);
+			if (pcibios_read_config_dword(bus, devfn,
+						      PCI_CLASS_REVISION, &x)
+			    == PCIBIOS_SUCCESSFUL && (x >> 8) == class_code) {
+				if (index == num) {
+					*bus_ptr = bus;
+					*dev_fn_ptr = devfn;
+					return PCIBIOS_SUCCESSFUL;
+				}
+				++num;
+			}
+			if (fn != 0) {
+				if (++fn >= 8) {
+					++unit;
+					fn = 0;
+				}
+				continue;
+			}
+			if (pcibios_read_config_byte(bus, devfn,
+						     PCI_HEADER_TYPE, &h)
+			    == PCIBIOS_SUCCESSFUL && (h & 0x80) != 0)
+				++fn;
+			else
+				++unit;
+		}
+	}
+	return PCIBIOS_DEVICE_NOT_FOUND;
+}
+
+__initfunc(unsigned long route_pci_interrupts(void))
+{
+	return 0;
+}
diff --git a/arch/ppc/kernel/pci.c b/arch/ppc/kernel/pci.c
index 612a24bff..49f2f37be 100644
--- a/arch/ppc/kernel/pci.c
+++ b/arch/ppc/kernel/pci.c
@@ -1,630 +1,194 @@
 /*
- * PCI support
- * -- rough emulation of "PCI BIOS" functions
- *
- * Note: these are very motherboard specific!  Some way needs to
- * be worked out to handle the differences.
+ * $Id: pci.c,v 1.12 1997/08/27 05:05:28 cort Exp $
+ * Common pmac/prep/chrp pci routines. -- Cort
  */
 
-#include <linux/config.h>
-#include <linux/types.h>
-#include <linux/bios32.h>
+#include <linux/kernel.h>
 #include <linux/pci.h>
+/*#include <linux/bios32.h>*/
+#include <linux/delay.h>
+#include <linux/string.h>
+#include <linux/init.h>
 
-#include <asm/byteorder.h>
-#include <asm/io.h>
-#include <asm/ptrace.h>
 #include <asm/processor.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/pci-bridge.h>
 
-/*
- * PCI interrupt configuration.  This is motherboard specific.
- */
-/* Which PCI interrupt line does a given device [slot] use? */
-/* Note: This really should be two dimensional based in slot/pin used */
-unsigned char *Motherboard_map;
-unsigned char *Motherboard_map_name;
-
-/* How is the 82378 PIRQ mapping setup? */
-unsigned char *Motherboard_routes;
-
-/* Tables for known hardware */   
-
-/* Motorola PowerStack */
-static char Blackhawk_pci_IRQ_map[16] =
-{
-  	0,	/* Slot 0  - unused */
-  	0,	/* Slot 1  - unused */
-  	0,	/* Slot 2  - unused */
-  	0,	/* Slot 3  - unused */
-  	0,	/* Slot 4  - unused */
-  	0,	/* Slot 5  - unused */
-  	0,	/* Slot 6  - unused */
-  	0,	/* Slot 7  - unused */
-  	0,	/* Slot 8  - unused */
-  	0,	/* Slot 9  - unused */
-  	0,	/* Slot 10 - unused */
-  	0,	/* Slot 11 - unused */
-  	3,	/* Slot 12 - SCSI */
-  	0,	/* Slot 13 - unused */
-  	1,	/* Slot 14 - Ethernet */
-  	0,	/* Slot 15 - unused */
-};
-
-static char Blackhawk_pci_IRQ_routes[] =
-{
-   	0,	/* Line 0 - Unused */
-   	9,	/* Line 1 */
-   	11,	/* Line 2 */
-   	14,	/* Line 3 */
-   	15	/* Line 4 */
-};
-   
-/* Motorola MVME16xx */
-static char Genesis_pci_IRQ_map[16] =
-{
-  	0,	/* Slot 0  - unused */
-  	0,	/* Slot 1  - unused */
-  	0,	/* Slot 2  - unused */
-  	0,	/* Slot 3  - unused */
-  	0,	/* Slot 4  - unused */
-  	0,	/* Slot 5  - unused */
-  	0,	/* Slot 6  - unused */
-  	0,	/* Slot 7  - unused */
-  	0,	/* Slot 8  - unused */
-  	0,	/* Slot 9  - unused */
-  	0,	/* Slot 10 - unused */
-  	0,	/* Slot 11 - unused */
-  	3,	/* Slot 12 - SCSI */
-  	0,	/* Slot 13 - unused */
-  	1,	/* Slot 14 - Ethernet */
-  	0,	/* Slot 15 - unused */
-};
-
-static char Genesis_pci_IRQ_routes[] =
-{
-   	0,	/* Line 0 - Unused */
-   	10,	/* Line 1 */
-   	11,	/* Line 2 */
-   	14,	/* Line 3 */
-   	15	/* Line 4 */
-};
-   
-/* Motorola Series-E */
-static char Comet_pci_IRQ_map[16] =
-{
-  	0,	/* Slot 0  - unused */
-  	0,	/* Slot 1  - unused */
-  	0,	/* Slot 2  - unused */
-  	0,	/* Slot 3  - unused */
-  	0,	/* Slot 4  - unused */
-  	0,	/* Slot 5  - unused */
-  	0,	/* Slot 6  - unused */
-  	0,	/* Slot 7  - unused */
-  	0,	/* Slot 8  - unused */
-  	0,	/* Slot 9  - unused */
-  	0,	/* Slot 10 - unused */
-  	0,	/* Slot 11 - unused */
-  	3,	/* Slot 12 - SCSI */
-  	0,	/* Slot 13 - unused */
-  	1,	/* Slot 14 - Ethernet */
-  	0,	/* Slot 15 - unused */
-};
-
-static char Comet_pci_IRQ_routes[] =
-{
-   	0,	/* Line 0 - Unused */
-   	10,	/* Line 1 */
-   	11,	/* Line 2 */
-   	14,	/* Line 3 */
-   	15	/* Line 4 */
-};
-
-/* BeBox */
-static char BeBox_pci_IRQ_map[16] =
-{
-  	0,	/* Slot 0  - unused */
-  	0,	/* Slot 1  - unused */
-  	0,	/* Slot 2  - unused */
-  	0,	/* Slot 3  - unused */
-  	0,	/* Slot 4  - unused */
-  	0,	/* Slot 5  - unused */
-  	0,	/* Slot 6  - unused */
-  	0,	/* Slot 7  - unused */
-  	0,	/* Slot 8  - unused */
-  	0,	/* Slot 9  - unused */
-  	0,	/* Slot 10 - unused */
-  	0,	/* Slot 11 - unused */
-  	16,	/* Slot 12 - SCSI */
-  	0,	/* Slot 13 - unused */
-  	0,	/* Slot 14 - unused */
-  	0,	/* Slot 15 - unused */
-};
-
-static char BeBox_pci_IRQ_routes[] =
-{
-   	0,	/* Line 0 - Unused */
-   	9,	/* Line 1 */
-   	11,	/* Line 2 */
-   	14,	/* Line 3 */
-   	15	/* Line 4 */
-};
-
-/* IBM Nobis */
-static char Nobis_pci_IRQ_map[16] =
-{
-  	0,	/* Slot 0  - unused */
-  	0,	/* Slot 1  - unused */
-  	0,	/* Slot 2  - unused */
-  	0,	/* Slot 3  - unused */
-  	0,	/* Slot 4  - unused */
-  	0,	/* Slot 5  - unused */
-  	0,	/* Slot 6  - unused */
-  	0,	/* Slot 7  - unused */
-  	0,	/* Slot 8  - unused */
-  	0,	/* Slot 9  - unused */
-  	0,	/* Slot 10 - unused */
-  	0,	/* Slot 11 - unused */
-  	3,	/* Slot 12 - SCSI */
-  	0,	/* Slot 13 - unused */
-  	0,	/* Slot 14 - unused */
-  	0,	/* Slot 15 - unused */
-};
-
-static char Nobis_pci_IRQ_routes[] =
-{
-   	0,	/* Line 0 - Unused */
-   	13,	/* Line 1 */
-   	13,	/* Line 2 */
-   	13,	/* Line 3 */
-   	13      /* Line 4 */
-};
-
+unsigned long io_base;
+unsigned long pci_dram_offset;
 
 /*
- * ibm 830 (and 850?).
- * This is actually based on the Carolina motherboard
- * -- Cort
- */
-static char ibm8xx_pci_IRQ_map[23] = {
-        0, /* Slot 0  - unused */
-        0, /* Slot 1  - unused */
-        0, /* Slot 2  - unused */
-        0, /* Slot 3  - unused */
-        0, /* Slot 4  - unused */
-        0, /* Slot 5  - unused */
-        0, /* Slot 6  - unused */
-        0, /* Slot 7  - unused */
-        0, /* Slot 8  - unused */
-        0, /* Slot 9  - unused */
-        0, /* Slot 10 - unused */
-        0, /* Slot 11 - FireCoral */
-        4, /* Slot 12 - Ethernet  PCIINTD# */
-        2, /* Slot 13 - PCI Slot #2 */
-        2, /* Slot 14 - S3 Video PCIINTD# */
-        0, /* Slot 15 - onboard SCSI (INDI) [1] */
-        3, /* Slot 16 - NCR58C810 RS6000 Only PCIINTC# */
-        0, /* Slot 17 - unused */
-        2, /* Slot 18 - PCI Slot 2 PCIINTx# (See below) */
-        0, /* Slot 19 - unused */
-        0, /* Slot 20 - unused */
-        0, /* Slot 21 - unused */
-        2, /* Slot 22 - PCI slot 1 PCIINTx# (See below) */
-};
-static char ibm8xx_pci_IRQ_routes[] = {
-        0,      /* Line 0 - unused */
-        13,     /* Line 1 */
-        10,     /* Line 2 */
-        15,     /* Line 3 */
-        15,     /* Line 4 */
-};
-/* This just changes the PCI slots & onboard SCSI + S3 to IRQ10, but
- * it really needs some logic to set them to unique IRQ's, or even
- * add some logic to the drivers to ask an irq.c routine to re-map
- * the IRQ if it needs one to itself...
+ * It would be nice if we could create a include/asm/pci.h and have just
+ * function ptrs for all these in there, but that isn't the case.
+ * We have a function, pcibios_*() which calls the function ptr ptr_pcibios_*()
+ * which has been setup by pcibios_init().  This is all to avoid a check
+ * for pmac/prep every time we call one of these.  It should also make the move
+ * to a include/asm/pcibios.h easier, we can drop the ptr_ on these functions
+ * and create pci.h
+ *   -- Cort
  */
-static char Carolina_PIRQ_routes[] = {
-   0xad,	/* INTB# = 10, INTA# = 13 */
-   0xff		/* INTD# = 15, INTC# = 15 */
-};
-/* We have to turn on LEVEL mode for changed IRQ's */
-/* All PCI IRQ's need to be level mode, so this should be something
- * other than hard-coded as well... IRQ's are individually mappable
- * to either edge or level.
- */
-#define CAROLINA_IRQ_EDGE_MASK_LO   0x00  /* IRQ's 0-7  */
-#define CAROLINA_IRQ_EDGE_MASK_HI   0xA4  /* IRQ's 8-15 [10,13,15] */
-#define PCI_DEVICE_ID_IBM_CORAL		0x000a
-
-#undef  PCI_DEBUG
-
-#ifdef PCI_STATS
-int PCI_conversions[2];
-#endif
-
-
-unsigned long pcibios_fixup(unsigned long mem_start, unsigned long mem_end)
-{
-	return mem_start;
+int (*ptr_pcibios_read_config_byte)(unsigned char bus, unsigned char dev_fn,
+			     unsigned char offset, unsigned char *val);
+int (*ptr_pcibios_read_config_word)(unsigned char bus, unsigned char dev_fn,
+			     unsigned char offset, unsigned short *val);
+int (*ptr_pcibios_read_config_dword)(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned int *val);
+int (*ptr_pcibios_write_config_byte)(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned char val);
+int (*ptr_pcibios_write_config_word)(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned short val);
+int (*ptr_pcibios_write_config_dword)(unsigned char bus, unsigned char dev_fn,
+			       unsigned char offset, unsigned int val);
+int (*ptr_pcibios_find_device)(unsigned short vendor, unsigned short dev_id,
+			unsigned short index, unsigned char *bus_ptr,
+			unsigned char *dev_fn_ptr);
+int (*ptr_pcibios_find_class)(unsigned int class_code, unsigned short index,
+		       unsigned char *bus_ptr, unsigned char *dev_fn_ptr);
+
+extern int pmac_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn,
+			     unsigned char offset, unsigned char *val);
+extern int pmac_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn,
+			     unsigned char offset, unsigned short *val);
+extern int pmac_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned int *val);
+extern int pmac_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned char val);
+extern int pmac_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned short val);
+extern int pmac_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn,
+			       unsigned char offset, unsigned int val);
+extern int pmac_pcibios_find_device(unsigned short vendor, unsigned short dev_id,
+			unsigned short index, unsigned char *bus_ptr,
+			unsigned char *dev_fn_ptr);
+extern int pmac_pcibios_find_class(unsigned int class_code, unsigned short index,
+		       unsigned char *bus_ptr, unsigned char *dev_fn_ptr);
+
+extern int chrp_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn,
+			     unsigned char offset, unsigned char *val);
+extern int chrp_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn,
+			     unsigned char offset, unsigned short *val);
+extern int chrp_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned int *val);
+extern int chrp_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned char val);
+extern int chrp_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned short val);
+extern int chrp_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn,
+			       unsigned char offset, unsigned int val);
+extern int chrp_pcibios_find_device(unsigned short vendor, unsigned short dev_id,
+			unsigned short index, unsigned char *bus_ptr,
+			unsigned char *dev_fn_ptr);
+extern int chrp_pcibios_find_class(unsigned int class_code, unsigned short index,
+		       unsigned char *bus_ptr, unsigned char *dev_fn_ptr);
+
+extern int prep_pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn,
+			     unsigned char offset, unsigned char *val);
+extern int prep_pcibios_read_config_word(unsigned char bus, unsigned char dev_fn,
+			     unsigned char offset, unsigned short *val);
+extern int prep_pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned int *val);
+extern int prep_pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned char val);
+extern int prep_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned short val);
+extern int prep_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn,
+			       unsigned char offset, unsigned int val);
+extern int prep_pcibios_find_device(unsigned short vendor, unsigned short dev_id,
+			unsigned short index, unsigned char *bus_ptr,
+			unsigned char *dev_fn_ptr);
+extern int prep_pcibios_find_class(unsigned int class_code, unsigned short index,
+		       unsigned char *bus_ptr, unsigned char *dev_fn_ptr);
+
+
+int pcibios_read_config_byte(unsigned char bus, unsigned char dev_fn,
+			     unsigned char offset, unsigned char *val)
+{
+	return ptr_pcibios_read_config_byte(bus,dev_fn,offset,val);
 }
-
-int
-pcibios_present (void)
+int pcibios_read_config_word(unsigned char bus, unsigned char dev_fn,
+			     unsigned char offset, unsigned short *val)
 {
-#ifdef PCI_DEBUG	
-	printk("PCI [BIOS] present?\n");
-#endif	
-	return (1);
+	return ptr_pcibios_read_config_word(bus,dev_fn,offset,val);
 }
-
-int
-pcibios_read_config_dword (unsigned char bus,
-			   unsigned char dev, unsigned char offset, unsigned int *val)
+int pcibios_read_config_dword(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned int *val)
 {
-	unsigned long _val;
-	unsigned long *ptr;
-	dev >>= 3;
-#ifdef PCI_DEBUG	
-	printk("PCI Read config dword[%d.%d.%x] = ", bus, dev, offset);
-#endif	
-	if ((bus != 0) || (dev < 11) || (dev > 16))
-	{
-		*val = 0xFFFFFFFF;
-		return PCIBIOS_DEVICE_NOT_FOUND;
-	} else
-	{
-		ptr = (unsigned long *)(0x80800000 | (1<<dev) | offset);
-#ifdef PCI_DEBUG	
-		printk("[%x] ", ptr);
-#endif		
-		_val = le32_to_cpu(*ptr);
-	}
-#ifdef PCI_DEBUG	
-	printk("%x\n", _val);
-#endif	
-	*val = _val;
-	return PCIBIOS_SUCCESSFUL;
+	return ptr_pcibios_read_config_dword(bus,dev_fn,offset,val);
 }
-
-int
-pcibios_read_config_word (unsigned char bus,
-			  unsigned char dev, unsigned char offset, unsigned short *val)
+int pcibios_write_config_byte(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned char val)
 {
-	unsigned short _val;
-	unsigned short *ptr;
-	dev >>= 3;
-#ifdef PCI_DEBUG
-	printk("PCI Read config word[%d.%d.%x] = ", bus, dev, offset);
-#endif	
-	if ((bus != 0) || (dev < 11) || (dev > 16))
-	{
-		*val = (unsigned short)0xFFFFFFFF;
-		return PCIBIOS_DEVICE_NOT_FOUND;
-	} else
-	{
-		ptr = (unsigned short *)(0x80800000 | (1<<dev) | offset);
-#ifdef PCI_DEBUG
-		printk("[%x] ", ptr);
-#endif		
-		_val = le16_to_cpu(*ptr);
-	}
-#ifdef PCI_DEBUG
-	printk("%x\n", _val);
-#endif		
-	*val = _val;
-	return PCIBIOS_SUCCESSFUL;
+	return ptr_pcibios_write_config_byte(bus,dev_fn,offset,val);
 }
-
-int
-pcibios_read_config_byte (unsigned char bus,
-			  unsigned char dev, unsigned char offset, unsigned char *val)
+int pcibios_write_config_word(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned short val)
 {
-	unsigned char _val;
-	volatile unsigned char *ptr;
-	dev >>= 3;
-	/* Note: the configuration registers don't always have this right! */
-	if (offset == PCI_INTERRUPT_LINE)
-	{
-		if (Motherboard_map[dev] <= 4)
-		{
-			*val = Motherboard_routes[Motherboard_map[dev]];
-			/*printk("dev %d map %d route %d\n",
-			  dev,Motherboard_map[dev],
-			  Motherboard_routes[Motherboard_map[dev]]);*/
-		} else
-		{ /* Pseudo interrupts [for BeBox] */
-			*val = Motherboard_map[dev];
-		}
-#ifdef PCI_DEBUG	
-		printk("PCI Read Interrupt Line[%d.%d] = %d\n", bus, dev, *val);
-#endif		
-		return PCIBIOS_SUCCESSFUL;
-	}
-#ifdef PCI_DEBUG	
-	printk("PCI Read config byte[%d.%d.%x] = ", bus, dev, offset);
-#endif		
-	if ((bus != 0) || (dev < 11) || (dev > 16))
-	{
-		*val = 0xFFFFFFFF;
-		return PCIBIOS_DEVICE_NOT_FOUND;
-	} else
-	{
-		ptr = (unsigned char *)(0x80800000 | (1<<dev) | offset ^ 1);
-#ifdef PCI_DEBUG	
-		printk("[%x] ", ptr);
-#endif		
-		_val = *ptr;
-	}
-#ifdef PCI_DEBUG	
-	printk("%x\n", _val);
-#endif
-	*val = _val;
-	return PCIBIOS_SUCCESSFUL;
+	return ptr_pcibios_write_config_word(bus,dev_fn,offset,val);
 }
-
-int
-pcibios_write_config_dword (unsigned char bus,
-			    unsigned char dev, unsigned char offset, unsigned int val)
+int pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn,
+			       unsigned char offset, unsigned int val)
 {
-	unsigned long _val;
-	unsigned long *ptr;
-	dev >>= 3;
-	_val = le32_to_cpu(val);
-#ifdef PCI_DEBUG	
-	printk("PCI Write config dword[%d.%d.%x] = %x\n", bus, dev, offset, _val);
-#endif		
-	if ((bus != 0) || (dev < 11) || (dev > 16))
-	{
-		return PCIBIOS_DEVICE_NOT_FOUND;
-	} else
-	{
-		ptr = (unsigned long *)(0x80800000 | (1<<dev) | offset);
-		*ptr = _val;
-	}
-	return PCIBIOS_SUCCESSFUL;
+	return ptr_pcibios_write_config_dword(bus,dev_fn,offset,val);
 }
-
-int
-pcibios_write_config_word (unsigned char bus,
-			   unsigned char dev, unsigned char offset, unsigned short val)
+int pcibios_find_device(unsigned short vendor, unsigned short dev_id,
+			unsigned short index, unsigned char *bus_ptr,
+			unsigned char *dev_fn_ptr)
 {
-	unsigned short _val;
-	unsigned short *ptr;
-	dev >>= 3;
-	_val = le16_to_cpu(val);
-#ifdef PCI_DEBUG	
-	printk("PCI Write config word[%d.%d.%x] = %x\n", bus, dev, offset, _val);
-#endif		
-	if ((bus != 0) || (dev < 11) || (dev > 16))
-	{
-		return PCIBIOS_DEVICE_NOT_FOUND;
-	} else
-	{
-		ptr = (unsigned short *)(0x80800000 | (1<<dev) | offset);
-		*ptr = _val;
-	}
-	return PCIBIOS_SUCCESSFUL;
+	return ptr_pcibios_find_device(vendor,dev_id,index,bus_ptr,dev_fn_ptr);
 }
-
-int
-pcibios_write_config_byte (unsigned char bus,
-			   unsigned char dev, unsigned char offset, unsigned char val)
+int pcibios_find_class(unsigned int class_code, unsigned short index,
+		       unsigned char *bus_ptr, unsigned char *dev_fn_ptr)
 {
-	unsigned char _val;
-	unsigned char *ptr;
-	dev >>= 3;
-	_val = val;
-#ifdef PCI_DEBUG	
-	printk("PCI Write config byte[%d.%d.%x] = %x\n", bus, dev, offset, _val);
-#endif		
-	if ((bus != 0) || (dev < 11) || (dev > 16))
-	{
-		return PCIBIOS_DEVICE_NOT_FOUND;
-	} else
-	{
-		ptr = (unsigned char *)(0x80800000 | (1<<dev) | offset ^ 1);
-		*ptr = _val;
-	}
-	return PCIBIOS_SUCCESSFUL;
+	return ptr_pcibios_find_class(class_code,index,bus_ptr,dev_fn_ptr);
 }
 
-int
-pcibios_find_device (unsigned short vendor, unsigned short device_id,
-		     unsigned short index, unsigned char *bus,
-		     unsigned char *dev)
+int pcibios_present(void)
 {
-	unsigned long w, desired = (device_id << 16) | vendor;
-	int devnr;
-
-	if (vendor == 0xffff) {
-		return PCIBIOS_BAD_VENDOR_ID;
-	}
-
-	for (devnr = 11;  devnr < 16;  devnr++)
-	{
-		pcibios_read_config_dword(0, devnr<<3, PCI_VENDOR_ID, &w);
-		if (w == desired) {
-			if (index == 0) {
-				*bus = 0;
-				*dev = devnr<<3;
-				return PCIBIOS_SUCCESSFUL;
-			}
-			--index;
-		}
-	}
-	return PCIBIOS_DEVICE_NOT_FOUND;
+	return 1;
 }
 
-int
-pcibios_find_class (unsigned int class_code, unsigned short index, 
-		    unsigned char *bus, unsigned char *dev)
-{
-	int dev_nr, class, indx;
-	indx = 0;
-#ifdef PCI_DEBUG	
-	printk("pcibios_find_class - class: %x, index: %x", class_code, index);
-#endif	
-	for (dev_nr = 11;  dev_nr < 16;  dev_nr++)
-	{
-		pcibios_read_config_dword(0, dev_nr<<3, PCI_CLASS_REVISION, &class);
-		if ((class>>8) == class_code)
-		{
-			if (index == indx)
-			{
-				*bus = 0;
-				*dev = dev_nr<<3;
-#ifdef PCI_DEBUG
-				printk(" - device: %x\n", dev_nr);
-#endif	
-				return (0);
-			}
-			indx++;
-		}
-	}
-#ifdef PCI_DEBUG
-	printk(" - not found\n");
-#endif	
-	return PCIBIOS_DEVICE_NOT_FOUND;
-}    
-
-const char *pcibios_strerror(int error)
-{
-	static char buf[32];
-	switch (error)
-	{	case PCIBIOS_SUCCESSFUL:
-		return ("PCI BIOS: no error");
-	case PCIBIOS_FUNC_NOT_SUPPORTED:
-		return ("PCI BIOS: function not supported");
-	case PCIBIOS_BAD_VENDOR_ID:
-		return ("PCI BIOS: bad vendor ID");
-	case PCIBIOS_DEVICE_NOT_FOUND:
-		return ("PCI BIOS: device not found");
-	case PCIBIOS_BAD_REGISTER_NUMBER:
-		return ("PCI BIOS: bad register number");
-	case PCIBIOS_SET_FAILED:
-		return ("PCI BIOS: set failed");
-	case PCIBIOS_BUFFER_TOO_SMALL:
-		return ("PCI BIOS: buffer too small");
-	default:
-		sprintf(buf, "PCI BIOS: invalid error #%d", error);
-		return(buf);
+__initfunc(unsigned long
+pcibios_init(unsigned long mem_start,unsigned long mem_end))
+{
+	switch (_machine) {
+	    case _MACH_Motorola:
+	    case _MACH_IBM:
+		ptr_pcibios_read_config_byte = prep_pcibios_read_config_byte;
+		ptr_pcibios_read_config_word = prep_pcibios_read_config_word;
+		ptr_pcibios_read_config_dword = prep_pcibios_read_config_dword;
+		ptr_pcibios_write_config_byte = prep_pcibios_write_config_byte;
+		ptr_pcibios_write_config_word = prep_pcibios_write_config_word;
+		ptr_pcibios_write_config_dword = prep_pcibios_write_config_dword;
+		ptr_pcibios_find_device = prep_pcibios_find_device;
+		ptr_pcibios_find_class = prep_pcibios_find_class;
+		break;
+	    case _MACH_Pmac:
+		ptr_pcibios_read_config_byte = pmac_pcibios_read_config_byte;
+		ptr_pcibios_read_config_word = pmac_pcibios_read_config_word;
+		ptr_pcibios_read_config_dword = pmac_pcibios_read_config_dword;
+		ptr_pcibios_write_config_byte = pmac_pcibios_write_config_byte;
+		ptr_pcibios_write_config_word = pmac_pcibios_write_config_word;
+		ptr_pcibios_write_config_dword = pmac_pcibios_write_config_dword;
+		ptr_pcibios_find_device = pmac_pcibios_find_device;
+		ptr_pcibios_find_class = pmac_pcibios_find_class;
+		break;
+	    case _MACH_chrp:
+		ptr_pcibios_read_config_byte = chrp_pcibios_read_config_byte;
+		ptr_pcibios_read_config_word = chrp_pcibios_read_config_word;
+		ptr_pcibios_read_config_dword = chrp_pcibios_read_config_dword;
+		ptr_pcibios_write_config_byte = chrp_pcibios_write_config_byte;
+		ptr_pcibios_write_config_word = chrp_pcibios_write_config_word;
+		ptr_pcibios_write_config_dword = chrp_pcibios_write_config_dword;
+		ptr_pcibios_find_device = chrp_pcibios_find_device;
+		ptr_pcibios_find_class = chrp_pcibios_find_class;
+		break;
 	}
-}
-
-/*
- * Note: This routine has to access the PCI configuration space
- * for the PCI bridge chip (Intel 82378).
- */
-unsigned long pcibios_init(unsigned long mem_start,unsigned long mem_end)
-{
 	return mem_start;
 }
 
-unsigned long route_pci_interrupts(void)
+__initfunc(unsigned long
+pcibios_fixup(unsigned long mem_start, unsigned long mem_end))
 {
-	unsigned char *ibc_pirq = (unsigned char *)0x80800860;
-	unsigned char *ibc_pcicon = (unsigned char *)0x80800840;
-	extern unsigned long isBeBox[];
-	int i;
-	
-	if ( _machine == _MACH_Motorola)
-	{ 
-		switch (inb(0x800) & 0xF0)
-		{
-		case 0x10: /* MVME16xx */
-			Motherboard_map_name = "Genesis";
-			Motherboard_map = Genesis_pci_IRQ_map;
-			Motherboard_routes = Genesis_pci_IRQ_routes;
-			break;
-		case 0x20: /* Series E */
-			Motherboard_map_name = "Series E";
-			Motherboard_map = Comet_pci_IRQ_map;
-			Motherboard_routes = Comet_pci_IRQ_routes;
-			break;
-		case 0x40: /* PowerStack */
-		default: /* Can't hurt, can it? */
-			Motherboard_map_name = "Blackhawk (Powerstack)";
-			Motherboard_map = Blackhawk_pci_IRQ_map;
-			Motherboard_routes = Blackhawk_pci_IRQ_routes;
-			break;
-		}
-	} else
-	{
-		if ( _machine == _MACH_IBM )
-		{
-			unsigned char pl_id;
-			unsigned long flags;
-			unsigned index;
-			unsigned char fn, bus;
-			unsigned int addr;
-			unsigned char dma_mode, ide_mode;
-			int i;
-			
-			Motherboard_map_name = "IBM 8xx (Carolina)";
-			Motherboard_map = ibm8xx_pci_IRQ_map;
-			Motherboard_routes = ibm8xx_pci_IRQ_routes;
-ll_printk("before loop\n");			
-			
-			for (index = 0;
-			     !pcibios_find_device (PCI_VENDOR_ID_IBM, 
-						   PCI_DEVICE_ID_IBM_CORAL, 
-						   index, &bus, &fn); ++index)
-			{
-				pcibios_read_config_dword(bus, fn, 0x10, &addr);
-				addr &= ~0x3;
-				outb(0x26, addr);
-				dma_mode = inb(addr+4);
-				outb(0x25, addr);
-				ide_mode = inb(addr+4);
-				/*printk("CORAL I/O at 0x%x, DMA mode: %x, IDE mode: %x", 
-				       addr, dma_mode, ide_mode);*/
-				/* Make CDROM non-DMA */
-				ide_mode = (ide_mode & 0x0F) | 0x20;
-				outb(0x25, addr);
-				outb(ide_mode, addr+4);
-				dma_mode = dma_mode & ~0x80;
-				outb(0x26, addr);
-				outb(dma_mode, addr+4);
-				outb(0x26, addr);
-				dma_mode = inb(addr+4);
-				outb(0x25, addr);
-				ide_mode = inb(addr+4);
-				/*printk("=> DMA mode: %x, IDE mode: %x\n", 
-				       dma_mode, ide_mode);*/
-			}
-			
-			/* Setup the PCI INT mappings for the Carolina */
-			/* These are PCI Interrupt Route Control [1|2] Register */
-			outb(Carolina_PIRQ_routes[0], 0x0890);
-			outb(Carolina_PIRQ_routes[1], 0x0891);
-			
-			pl_id=inb(0x0852);
-			/*printk("CPU Planar ID is %#0x\n", pl_id);*/
-			
-			if (pl_id == 0x0C) {
-				/* INDI */
-				Motherboard_map[12] = 1;
-			}
-ll_printk("before edge/level\n");			
-#if 0			
-			/*printk("Changing IRQ mode\n");*/
-			pl_id=inb(0x04d0);
-			/*printk("Low mask is %#0x\n", pl_id);*/
-			outb(pl_id|CAROLINA_IRQ_EDGE_MASK_LO, 0x04d0);
-			
-			pl_id=inb(0x04d1);
-			/*printk("Hi mask is  %#0x\n", pl_id);*/
-			outb(pl_id|CAROLINA_IRQ_EDGE_MASK_HI, 0x04d1);
-			pl_id=inb(0x04d1);
-			/*printk("Hi mask now %#0x\n", pl_id);*/
-#endif			
-		}
-	}
-	
-	/* Set up mapping from slots */
-	for (i = 1;  i <= 4;  i++)
-	{
-		ibc_pirq[i-1] = Motherboard_routes[i];
-	}
-	/* Enable PCI interrupts */
-	*ibc_pcicon |= 0x20;
+	return mem_start;
 }
diff --git a/arch/ppc/kernel/pmac_pci.c b/arch/ppc/kernel/pmac_pci.c
new file mode 100644
index 000000000..a48385b3f
--- /dev/null
+++ b/arch/ppc/kernel/pmac_pci.c
@@ -0,0 +1,423 @@
+/*
+ * Support for PCI bridges found on Power Macintoshes.
+ * At present the "bandit" and "chaos" bridges are supported.
+ * Fortunately you access configuration space in the same
+ * way with either bridge.
+ *
+ * Copyright (C) 1997 Paul Mackerras (paulus@cs.anu.edu.au)
+ *
+ * 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.
+ */
+
+#include <linux/kernel.h>
+#include <linux/pci.h>
+#include <linux/bios32.h>
+#include <linux/delay.h>
+#include <linux/string.h>
+#include <linux/init.h>
+#include <asm/io.h>
+#include <asm/prom.h>
+#include <asm/pci-bridge.h>
+
+struct bridge_data {
+	volatile unsigned int *cfg_addr;
+	volatile unsigned char *cfg_data;
+	void *io_base;
+	int bus_number;
+	int max_bus;
+	struct bridge_data *next;
+	struct device_node *node;
+};
+
+static struct bridge_data **bridges, *bridge_list;
+static int max_bus;
+
+static void add_bridges(struct device_node *dev, unsigned long *mem_ptr);
+
+/*
+ * Magic constants for enabling cache coherency in the bandit/PSX bridge.
+ */
+#define APPLE_VENDID	0x106b
+#define BANDIT_DEVID	1
+#define BANDIT_REVID	3
+
+#define BANDIT_DEVNUM	11
+#define BANDIT_MAGIC	0x50
+#define BANDIT_COHERENT	0x40
+
+/*
+ * 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.
+ */
+static void init_bandit(struct bridge_data *bp)
+{
+	unsigned int vendev, magic;
+	int rev;
+
+	/* read the word at offset 0 in config space for device 11 */
+	out_le32(bp->cfg_addr, (1UL << BANDIT_DEVNUM) + PCI_VENDOR_ID);
+	udelay(2);
+	vendev = in_le32((volatile unsigned int *)bp->cfg_data);
+	if (vendev != (BANDIT_DEVID << 16) + APPLE_VENDID) {
+		printk(KERN_WARNING "bandit isn't? (%x)\n", vendev);
+		return;
+	}
+
+	/* read the revision id */
+	out_le32(bp->cfg_addr, (1UL << BANDIT_DEVNUM) + PCI_REVISION_ID);
+	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);
+
+	/* read the word at offset 0x50 */
+	out_le32(bp->cfg_addr, (1UL << BANDIT_DEVNUM) + BANDIT_MAGIC);
+	udelay(2);
+	magic = in_le32((volatile unsigned int *)bp->cfg_data);
+	if ((magic & BANDIT_COHERENT) != 0)
+		return;
+	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);
+}
+
+unsigned long pmac_find_bridges(unsigned long mem_start, unsigned long mem_end)
+{
+	int bus;
+	struct bridge_data *bridge;
+
+	bridge_list = 0;
+	max_bus = 0;
+	add_bridges(find_devices("bandit"), &mem_start);
+	add_bridges(find_devices("chaos"), &mem_start);
+	bridges = (struct bridge_data **) mem_start;
+	mem_start += (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;
+
+	return mem_start;
+}
+
+static void add_bridges(struct device_node *dev, unsigned long *mem_ptr)
+{
+	int *bus_range;
+	int len;
+	struct bridge_data *bp;
+
+	for (; dev != NULL; dev = dev->next) {
+		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->name, dev->addrs[0].address);
+		bp = (struct bridge_data *) *mem_ptr;
+		*mem_ptr += sizeof(struct bridge_data);
+		bp->cfg_addr = (volatile unsigned int *)
+			(dev->addrs[0].address + 0x800000);
+		bp->cfg_data = (volatile unsigned char *)
+			(dev->addrs[0].address + 0xc00000);
+		bp->io_base = (void *) dev->addrs[0].address;
+		ioremap(dev->addrs[0].address, 0x800000);
+		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);
+	}
+}
+
+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;
+}
+
+int pci_device_loc(struct device_node *dev, unsigned char *bus_ptr,
+		   unsigned char *devfn_ptr)
+{
+	unsigned int *reg;
+	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;
+	}
+	*bus_ptr = reg[0] >> 16;
+	*devfn_ptr = reg[0] >> 8;
+	return 0;
+}
+
+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 {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + (offset & ~3) + 1);
+	}
+	udelay(2);
+	*val = in_8(bp->cfg_data + (offset & 3));
+
+	if (offset == PCI_INTERRUPT_LINE) {
+		/*
+		 * Open Firmware often doesn't initialize this
+		 * register properly, so we find the node and see
+		 * if it has an AAPL,interrupts property.
+		 */
+		struct device_node *node;
+		unsigned int *reg;
+
+		for (node = bp->node->child; node != 0; node = node->sibling) {
+			reg = (unsigned int *) get_property(node, "reg", 0);
+			if (reg == 0 || ((reg[0] >> 8) & 0xff) != dev_fn)
+				continue;
+			/* this is the node, see if it has interrupts */
+			if (node->n_intrs > 0)
+				*val = node->intrs[0];
+			break;
+		}
+	}
+
+	return PCIBIOS_SUCCESSFUL;
+}
+
+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 || (offset & 1) != 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 {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + (offset & ~3) + 1);
+	}
+	udelay(2);
+	*val = in_le16((volatile unsigned short *)(bp->cfg_data + (offset & 3)));
+	return PCIBIOS_SUCCESSFUL;
+}
+
+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 || (offset & 3) != 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);
+	} else {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + offset + 1);
+	}
+	udelay(2);
+	*val = in_le32((volatile unsigned int *)bp->cfg_data);
+	return PCIBIOS_SUCCESSFUL;
+}
+
+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) {
+		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 {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + (offset & ~3) + 1);
+	}
+	udelay(2);
+	out_8(bp->cfg_data + (offset & 3), val);
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pmac_pcibios_write_config_word(unsigned char bus, unsigned char dev_fn,
+			      unsigned char offset, unsigned short val)
+{
+	struct bridge_data *bp;
+
+	if (bus > max_bus || (bp = bridges[bus]) == 0 || (offset & 1) != 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 {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + (offset & ~3) + 1);
+	}
+	udelay(2);
+	out_le16((volatile unsigned short *)(bp->cfg_data + (offset & 3)), val);
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pmac_pcibios_write_config_dword(unsigned char bus, unsigned char dev_fn,
+			       unsigned char offset, unsigned int val)
+{
+	struct bridge_data *bp;
+
+	if (bus > max_bus || (bp = bridges[bus]) == 0 || (offset & 3) != 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);
+	} else {
+		out_le32(bp->cfg_addr, (dev_fn << 8) + offset + 1);
+	}
+	udelay(2);
+	out_le32((volatile unsigned int *)bp->cfg_data, val);
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pmac_pcibios_find_device(unsigned short vendor, unsigned short dev_id,
+			unsigned short index, unsigned char *bus_ptr,
+			unsigned char *dev_fn_ptr)
+{
+	int bus, unit, fn, num, devfn;
+	unsigned int x, vendev;
+	unsigned char h;
+
+	if (vendor == 0xffff)
+		return PCIBIOS_BAD_VENDOR_ID;
+	vendev = (dev_id << 16) + vendor;
+	num = 0;
+	for (bus = 0; bus <= max_bus; ++bus) {
+		if (bridges[bus] == 0)
+			continue;
+		unit = fn = 0;
+		if (bus == bridges[bus]->bus_number)
+			unit = 11;
+		while (unit < 32) {
+			devfn = PCI_DEVFN(unit, fn);
+			if (pcibios_read_config_dword(bus, devfn,
+						      PCI_VENDOR_ID, &x)
+			    == PCIBIOS_SUCCESSFUL && x == vendev) {
+				if (index == num) {
+					*bus_ptr = bus;
+					*dev_fn_ptr = devfn;
+					return PCIBIOS_SUCCESSFUL;
+				}
+				++num;
+			}
+			if (fn != 0) {
+				if (++fn >= 8) {
+					++unit;
+					fn = 0;
+				}
+				continue;
+			}
+			if (pcibios_read_config_byte(bus, devfn,
+						     PCI_HEADER_TYPE, &h)
+			    == PCIBIOS_SUCCESSFUL && (h & 0x80) != 0)
+				++fn;
+			else
+				++unit;
+		}
+	}
+	return PCIBIOS_DEVICE_NOT_FOUND;
+}
+
+int pmac_pcibios_find_class(unsigned int class_code, unsigned short index,
+		       unsigned char *bus_ptr, unsigned char *dev_fn_ptr)
+{
+	int bus, unit, fn, num, devfn;
+	unsigned int x;
+	unsigned char h;
+
+	num = 0;
+	for (bus = 0; bus <= max_bus; ++bus) {
+		if (bridges[bus] == 0)
+			continue;
+		unit = fn = 0;
+		if (bus == bridges[bus]->bus_number)
+			unit = 11;
+		while (unit < 32) {
+			devfn = PCI_DEVFN(unit, fn);
+			if (pcibios_read_config_dword(bus, devfn,
+						      PCI_CLASS_REVISION, &x)
+			    == PCIBIOS_SUCCESSFUL && (x >> 8) == class_code) {
+				if (index == num) {
+					*bus_ptr = bus;
+					*dev_fn_ptr = devfn;
+					return PCIBIOS_SUCCESSFUL;
+				}
+				++num;
+			}
+			if (fn != 0) {
+				if (++fn >= 8) {
+					++unit;
+					fn = 0;
+				}
+				continue;
+			}
+			if (pcibios_read_config_byte(bus, devfn,
+						     PCI_HEADER_TYPE, &h)
+			    == PCIBIOS_SUCCESSFUL && (h & 0x80) != 0)
+				++fn;
+			else
+				++unit;
+		}
+	}
+	return PCIBIOS_DEVICE_NOT_FOUND;
+}
diff --git a/arch/ppc/kernel/pmac_setup.c b/arch/ppc/kernel/pmac_setup.c
new file mode 100644
index 000000000..74a8ff92d
--- /dev/null
+++ b/arch/ppc/kernel/pmac_setup.c
@@ -0,0 +1,271 @@
+/*
+ *  linux/arch/ppc/kernel/setup.c
+ *
+ *  PowerPC version 
+ *    Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
+ *
+ *  Adapted for Power Macintosh by Paul Mackerras
+ *    Copyright (C) 1996 Paul Mackerras (paulus@cs.anu.edu.au)
+ *
+ *  Derived from "arch/alpha/kernel/setup.c"
+ *    Copyright (C) 1995 Linus Torvalds
+ *
+ *  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.
+ *
+ */
+
+/*
+ * bootup setup stuff..
+ */
+
+#include <linux/config.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/ptrace.h>
+#include <linux/malloc.h>
+#include <linux/user.h>
+#include <linux/a.out.h>
+#include <linux/tty.h>
+#include <linux/string.h>
+#include <linux/delay.h>
+#include <linux/ioport.h>
+#include <linux/major.h>
+#include <asm/prom.h>
+#include <asm/system.h>
+#include <asm/pgtable.h>
+#include <asm/io.h>
+#include <asm/ide.h>
+#include <asm/pci-bridge.h>
+#include "time.h"
+
+/*
+ * A magic address and value to put into it on machines with the
+ * "ohare" I/O controller.  This makes the IDE CD work on Starmaxes.
+ * Contributed by Harry Eaton.
+ */
+#define OMAGICPLACE	((volatile unsigned *) 0xf3000038)
+#define OMAGICCONT	0xbeff7a
+
+extern int root_mountflags;
+
+extern char command_line[];
+extern char saved_command_line[256];
+
+unsigned char drive_info;
+
+#define DEFAULT_ROOT_DEVICE 0x0801	/* sda1 - slightly silly choice */
+
+extern unsigned long find_available_memory(void);
+
+void pmac_setup_arch(char **cmdline_p,
+	unsigned long * memory_start_p, unsigned long * memory_end_p)
+{
+	extern unsigned long *end_of_DRAM;
+	struct device_node *cpu;
+	int *fp;
+
+	strcpy(saved_command_line, command_line);
+	*cmdline_p = command_line;
+
+	*memory_start_p = find_available_memory();
+	*memory_end_p = (unsigned long) end_of_DRAM;
+
+	set_prom_callback();
+
+	*memory_start_p = copy_device_tree(*memory_start_p, *memory_end_p);
+
+	/* Set loops_per_sec to a half-way reasonable value,
+	   for use until calibrate_delay gets called. */
+	cpu = find_type_devices("cpu");
+	if (cpu != 0) {
+		fp = (int *) get_property(cpu, "clock-frequency", NULL);
+		if (fp != 0) {
+			switch (_get_PVR() >> 16) {
+			case 4:		/* 604 */
+			case 9:		/* 604e */
+			case 20:	/* 620 */
+				loops_per_sec = *fp;
+				break;
+			default:	/* 601, 603, etc. */
+				loops_per_sec = *fp / 2;
+			}
+		} else
+			loops_per_sec = 50000000;
+	}
+}
+
+char *bootpath;
+char bootdevice[256];
+void *boot_host;
+int boot_target;
+int boot_part;
+kdev_t boot_dev;
+
+unsigned long
+powermac_init(unsigned long mem_start, unsigned long mem_end)
+{
+	struct device_node *chosen_np, *ohare_np;
+
+	mem_start = pmac_find_bridges(mem_start, mem_end);
+	ohare_np = find_devices("ohare");
+	if (ohare_np != NULL) {
+		printk(KERN_INFO "Twiddling the magic ohare bits\n");
+		out_le32(OMAGICPLACE, OMAGICCONT);
+	}
+	pmac_nvram_init();
+	via_cuda_init();
+	pmac_read_rtc_time();
+	pmac_find_display();
+	bootpath = NULL;
+	chosen_np = find_devices("chosen");
+	if (chosen_np != NULL)
+		bootpath = (char *) get_property(chosen_np, "bootpath", NULL);
+	if (bootpath != NULL) {
+		/*
+		 * There's a bug in the prom.  (Why am I not surprised.)
+		 * If you pass a path like scsi/sd@1:0 to canon, it returns
+		 * something like /bandit@F2000000/gc@10/53c94@10000/sd@0,0
+		 * That is, the scsi target number doesn't get preserved.
+		 */
+		call_prom("canon", 3, 1, bootpath, bootdevice, sizeof(bootdevice));
+	}
+	return mem_start;
+}
+
+void
+note_scsi_host(struct device_node *node, void *host)
+{
+	int l;
+	char *p;
+
+	l = strlen(node->full_name);
+	if (strncmp(node->full_name, bootdevice, l) == 0
+	    && (bootdevice[l] == '/' || bootdevice[l] == 0)) {
+		boot_host = host;
+		p = strstr(bootpath, "/sd@");
+		if (p != NULL) {
+			p += 4;
+			boot_target = simple_strtoul(p, NULL, 10);
+			p = strchr(p, ':');
+			if (p != NULL)
+				boot_part = simple_strtoul(p + 1, NULL, 10);
+		}
+	}
+}
+
+void find_boot_device(void)
+{
+	int dev;
+
+	if (kdev_t_to_nr(ROOT_DEV) != 0)
+		return;
+	ROOT_DEV = to_kdev_t(DEFAULT_ROOT_DEVICE);
+	if (boot_host == NULL)
+		return;
+#ifdef CONFIG_SCSI
+	dev = sd_find_target(boot_host, boot_target);
+	if (dev == 0)
+		return;
+	boot_dev = to_kdev_t(dev + boot_part);
+#endif
+	/* XXX should cope with booting from IDE also */
+}
+
+void note_bootable_part(kdev_t dev, int part)
+{
+	static int found_boot = 0;
+
+	if (!found_boot) {
+		find_boot_device();
+		found_boot = 1;
+	}
+	if (dev == boot_dev) {
+		ROOT_DEV = MKDEV(MAJOR(dev), MINOR(dev) + part);
+		boot_dev = NODEV;
+		printk(" (root)");
+	}
+}
+
+void pmac_ide_init_hwif_ports(ide_ioreg_t *p, ide_ioreg_t base, int *irq)
+{
+	struct device_node *np;
+	int i;
+	static struct device_node *atas;
+	static int atas_valid;
+
+	*p = 0;
+	*irq = 0;
+	if (!atas_valid) {
+		atas = find_devices("ATA");
+		atas_valid = 1;
+	}
+	for (i = (int)base, np = atas; i > 0 && np != NULL; --i, np = np->next)
+		;
+	if (np == NULL)
+		return;
+	if (np->n_addrs == 0) {
+		printk("ide: no addresses for device %s\n", np->full_name);
+		return;
+	}
+	if (np->n_intrs == 0) {
+		printk("ide: no intrs for device %s, using 13\n",
+		       np->full_name);
+		*irq = 13;
+	} else {
+		*irq = np->intrs[0];
+	}
+	base = (unsigned long) ioremap(np->addrs[0].address, 0x200);
+	for (i = 0; i < 8; ++i)
+		*p++ = base + i * 0x10;
+	*p = base + 0x160;
+}
+
+int
+pmac_get_cpuinfo(char *buffer)
+{
+	int pvr = _get_PVR();
+	char *model;
+	struct device_node *cpu;
+	int l, *fp;
+
+	l = 0;
+	cpu = find_type_devices("cpu");
+	if (cpu != 0) {
+		fp = (int *) get_property(cpu, "clock-frequency", NULL);
+		if (fp != 0)
+			l += sprintf(buffer, "%dMHz ", *fp / 1000000);
+	}
+
+	switch (pvr>>16) {
+	case 1:
+		model = "601";
+		break;
+	case 3:
+		model = "603";
+		break;
+	case 4:
+		model = "604";
+		break;
+	case 6:
+		model = "603e";
+		break;
+	case 7:
+		model = "603ev";
+		break;
+	case 9:
+		model = "604e";
+		break;
+	default:
+		model = "unknown";
+		break;
+	}
+	return l + sprintf(buffer+l, "PowerPC %s rev %d.%d\n", model,
+			   (pvr & 0xff00) >> 8, pvr & 0xff);
+}
diff --git a/arch/ppc/kernel/pmac_support.c b/arch/ppc/kernel/pmac_support.c
new file mode 100644
index 000000000..3c1895088
--- /dev/null
+++ b/arch/ppc/kernel/pmac_support.c
@@ -0,0 +1,69 @@
+/*
+ * Miscellaneous procedures for dealing with the PowerMac hardware.
+ */
+#include <linux/kernel.h>
+#include <linux/stddef.h>
+#include <linux/reboot.h>
+#include <linux/nvram.h>
+#include <asm/ptrace.h>
+#include <asm/io.h>
+#include <asm/cuda.h>
+#include <asm/system.h>
+#include <asm/prom.h>
+
+/*
+ * Read and write the non-volatile RAM on PowerMacs.
+ */
+static int nvram_naddrs;
+static volatile unsigned char *nvram_addr;
+static volatile unsigned char *nvram_data;
+
+void pmac_nvram_init(void)
+{
+	struct device_node *dp;
+
+	dp = find_devices("nvram");
+	if (dp == NULL) {
+		printk(KERN_ERR "Can't find NVRAM device\n");
+		nvram_naddrs = 0;
+		return;
+	}
+	nvram_naddrs = dp->n_addrs;
+	if (nvram_naddrs == 1) {
+		nvram_data = ioremap(dp->addrs[0].address, dp->addrs[0].size);
+	} else if (nvram_naddrs == 2) {
+		nvram_addr = ioremap(dp->addrs[0].address, dp->addrs[0].size);
+		nvram_data = ioremap(dp->addrs[1].address, dp->addrs[1].size);
+	} else {
+		printk(KERN_ERR "Don't know how to access NVRAM with %d addresses\n",
+		       nvram_naddrs);
+	}
+}
+
+unsigned char nvram_read_byte(int addr)
+{
+	switch (nvram_naddrs) {
+	case 1:
+		return nvram_data[(addr & 0x1fff) << 4];
+	case 2:
+		*nvram_addr = addr >> 5;
+		eieio();
+		return nvram_data[(addr & 0x1f) << 4];
+	}
+	return 0;
+}
+
+void nvram_write_byte(unsigned char val, int addr)
+{
+	switch (nvram_naddrs) {
+	case 1:
+		nvram_data[(addr & 0x1fff) << 4] = val;
+		break;
+	case 2:
+		*nvram_addr = addr >> 5;
+		eieio();
+		nvram_data[(addr & 0x1f) << 4] = val;
+		break;
+	}
+	eieio();
+}
diff --git a/arch/ppc/kernel/pmac_time.c b/arch/ppc/kernel/pmac_time.c
new file mode 100644
index 000000000..69a17321d
--- /dev/null
+++ b/arch/ppc/kernel/pmac_time.c
@@ -0,0 +1,87 @@
+/*
+ * Support for periodic interrupts (100 per second) and for getting
+ * the current time from the RTC on Power Macintoshes.
+ *
+ * At present, we use the decrementer register in the 601 CPU
+ * for our periodic interrupts.  This will probably have to be
+ * changed for other processors.
+ *
+ * Paul Mackerras	August 1996.
+ * Copyright (C) 1996 Paul Mackerras.
+ */
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/param.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <asm/cuda.h>
+#include <asm/prom.h>
+#include <asm/system.h>
+
+#include "time.h"
+
+
+/* Apparently the RTC stores seconds since 1 Jan 1904 */
+#define RTC_OFFSET	2082844800
+
+/*
+ * Query the OF and get the decr frequency.
+ * This was taken from the pmac time_init() when merging the prep/pmac
+ * time functions.
+ */
+void pmac_calibrate_decr(void)
+{
+	struct device_node *cpu;
+	int freq, *fp, divisor;
+
+	/*
+	 * The cpu node should have a timebase-frequency property
+	 * to tell us the rate at which the decrementer counts.
+	 */
+	cpu = find_type_devices("cpu");
+	if (cpu == 0)
+		panic("can't find cpu node in time_init");
+	fp = (int *) get_property(cpu, "timebase-frequency", NULL);
+	if (fp == 0)
+		panic("can't get cpu timebase frequency");
+	freq = *fp * 60;	/* try to make freq/1e6 an integer */
+	divisor = 60;
+	printk("time_init: decrementer frequency = %d/%d\n",
+	       freq, divisor);
+	decrementer_count = freq / HZ / divisor;
+	count_period_num = divisor;
+	count_period_den = freq / 1000000;
+}
+
+unsigned long
+pmac_get_rtc_time(void)
+{
+	struct cuda_request req;
+
+	/* Get the time from the RTC */
+	cuda_request(&req, NULL, 2, CUDA_PACKET, CUDA_GET_TIME);
+	while (!req.got_reply)
+		cuda_poll();
+	if (req.reply_len != 7)
+		printk(KERN_ERR "pmac_get_rtc_time: got %d byte reply\n",
+		      req.reply_len);
+	return (req.reply[3] << 24) + (req.reply[4] << 16)
+		+ (req.reply[5] << 8) + req.reply[6] - RTC_OFFSET;
+}
+
+int pmac_set_rtc_time(unsigned long nowtime)
+{
+	return 0;
+}
+
+/*
+ * We can't do this in time_init, because via_cuda_init hasn't
+ * been called at that stage.
+ */
+void
+pmac_read_rtc_time(void)
+{
+	xtime.tv_sec = pmac_get_rtc_time();
+	xtime.tv_usec = 0;
+}
diff --git a/arch/ppc/kernel/ppc_asm.tmpl b/arch/ppc/kernel/ppc_asm.tmpl
index e31dea266..7036f4a62 100644
--- a/arch/ppc/kernel/ppc_asm.tmpl
+++ b/arch/ppc/kernel/ppc_asm.tmpl
@@ -187,7 +187,3 @@ n:
 #define REST_8FPRS(n, base)	REST_4FPRS(n, base); REST_4FPRS(n+4, base)
 #define REST_16FPRS(n, base)	REST_8FPRS(n, base); REST_8FPRS(n+8, base)
 #define REST_32FPRS(n, base)	REST_16FPRS(n, base); REST_16FPRS(n+16, base)
-
-/* Missing instructions */
-#define bdne	bc 0,2,
-
diff --git a/arch/ppc/kernel/ppc_defs.head b/arch/ppc/kernel/ppc_defs.head
new file mode 100644
index 000000000..fbe926459
--- /dev/null
+++ b/arch/ppc/kernel/ppc_defs.head
@@ -0,0 +1,3 @@
+/*
+ * WARNING! This file is automatically generated - DO NOT EDIT!
+ */
diff --git a/arch/ppc/kernel/ppc_htab.c b/arch/ppc/kernel/ppc_htab.c
new file mode 100644
index 000000000..ea2a073eb
--- /dev/null
+++ b/arch/ppc/kernel/ppc_htab.c
@@ -0,0 +1,221 @@
+/*
+ * $Id: ppc_htab.c,v 1.7 1997/08/24 19:33:32 cort Exp $
+ *
+ * PowerPC hash table management proc entry.  Will show information
+ * about the current hash table and will allow changes to it.
+ *
+ * Written by Cort Dougan (cort@cs.nmt.edu)
+ *
+ * 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.
+ */
+
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/proc_fs.h>
+#include <linux/stat.h>
+
+#include <asm/uaccess.h>
+#include <asm/bitops.h>
+#include <asm/mmu.h>
+#include <asm/processor.h>
+#include <asm/residual.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+
+static long ppc_htab_read(struct inode * inode, struct file * file,
+			   char * buf, unsigned long nbytes);
+static long ppc_htab_write(struct inode * inode, struct file * file,
+			    const char * buffer, unsigned long count);
+static long long ppc_htab_lseek(struct inode * inode, struct file * file, 
+				 long long offset, int orig);
+
+extern PTE *Hash, *Hash_end;
+extern unsigned long Hash_size, Hash_mask;
+extern unsigned long _SDR1;
+
+static struct file_operations ppc_htab_operations = {
+    ppc_htab_lseek,	/* lseek   */
+    ppc_htab_read,	/* read	   */
+    ppc_htab_write,	/* write   */
+    NULL,		/* readdir */
+    NULL,		/* poll    */
+    NULL,		/* ioctl   */
+    NULL,		/* mmap	   */
+    NULL,		/* no special open code	   */
+    NULL,		/* no special release code */
+    NULL		/* can't fsync */
+};
+
+/*
+ * proc files can do almost nothing..
+ */
+struct inode_operations proc_ppc_htab_inode_operations = {
+    &ppc_htab_operations,  /* default proc file-ops */
+    NULL,	    /* create	   */
+    NULL,	    /* lookup	   */
+    NULL,	    /* link	   */
+    NULL,	    /* unlink	   */
+    NULL,	    /* symlink	   */
+    NULL,	    /* mkdir	   */
+    NULL,	    /* rmdir	   */
+    NULL,	    /* mknod	   */
+    NULL,	    /* rename	   */
+    NULL,	    /* readlink	   */
+    NULL,	    /* follow_link */
+    NULL,	    /* readpage	   */
+    NULL,	    /* writepage   */
+    NULL,	    /* bmap	   */
+    NULL,	    /* truncate	   */
+    NULL	    /* permission  */
+};
+
+
+/*
+ * print some useful info about the hash table.  This function
+ * is _REALLY_ slow (see the nested for loops below) but nothing
+ * in here should be really timing critical. -- Cort
+ */
+static long ppc_htab_read(struct inode * inode, struct file * file,
+			   char * buf, unsigned long nbytes)
+{
+	int n = 0, valid;
+	unsigned int kptes = 0, overflow = 0, uptes = 0;
+	PTE *ptr;
+	struct task_struct *p;
+	char buffer[128];
+	
+	if (nbytes < 0)
+		return -EINVAL;
+
+	/*
+	 * compute user/kernel pte's table this info can be
+	 * misleading since there can be valid (v bit set) entries
+	 * in the table but their vsid is used by no process (mm->context)
+	 * due to the way tlb invalidation is handled on the ppc
+	 * -- Cort
+	 */
+	for ( ptr = Hash ; ptr < Hash_end ; ptr += sizeof(PTE))
+	{
+		if (ptr->v)
+		{
+			/* make sure someone is using this context/vsid */
+			for_each_task(p)
+			{
+				if ( (ptr->vsid >> 4) == p->mm->context )
+				{
+					valid = 1;
+					break;
+				}
+			}
+			if ( !valid )
+				continue;
+			/* user not allowed read or write */
+			if (ptr->pp == PP_RWXX)
+				kptes++;
+			else
+				uptes++;
+			if (ptr->h == 1)
+				overflow++;
+		}
+	}
+	
+	n += sprintf( buffer,
+		      "Size\t\t: %luKb\n"
+		      "Buckets\t\t: %lu\n"
+ 		      "Address\t\t: %08lx\n"
+		      "Entries\t\t: %lu\n"
+		      "User ptes\t: %u\n"
+		      "Kernel ptes\t: %u\n"
+		      "Overflows\t: %u\n"
+		      "Percent full\t: %%%lu\n",
+                      (unsigned long)(Hash_size>>10),
+		      (Hash_size/(sizeof(PTE)*8)),
+		      (unsigned long)Hash,
+		      Hash_size/sizeof(PTE),
+                      uptes,
+		      kptes,
+		      overflow,
+		      ((kptes+uptes)*100) / (Hash_size/sizeof(PTE))
+		);
+
+	if (file->f_pos >= strlen(buffer))
+		return 0;
+	if (n > strlen(buffer) - file->f_pos)
+		n = strlen(buffer) - file->f_pos;
+	copy_to_user(buf, buffer + file->f_pos, n);
+	file->f_pos += n;
+	return n;
+}
+
+/*
+ * Can't _yet_ adjust the hash table size while running. -- Cort
+ */
+static long
+ppc_htab_write(struct inode * inode, struct file * file,
+		const char * buffer, unsigned long count)
+{
+	unsigned long size;
+	extern void reset_SDR1(void);
+	
+	if ( current->uid != 0 )
+		return -EACCES;
+	
+	/* only know how to set size right now */
+	if ( strncmp( buffer, "size ", 5) )
+		return -EINVAL;
+
+	size = simple_strtoul( &buffer[5], NULL, 10 );
+	
+	/* only allow to shrink */
+	if ( size >= Hash_size>>10 )
+		return -EINVAL;
+
+	/* minimum size of htab */
+	if ( size < 64 )
+		return -EINVAL;
+	
+	/* make sure it's a multiple of 64k */
+	if ( size % 64 )
+		return -EINVAL;
+	
+	printk("Hash table resize to %luk\n", size);
+	/*
+	 * We need to rehash all kernel entries for the new htab size.
+	 * Kernel only since we do a flush_tlb_all().  Since it's kernel
+	 * we only need to bother with vsids 0-15.  To avoid problems of
+	 * clobbering un-rehashed values we put the htab at a new spot
+	 * and put everything there.
+	 * -- Cort
+	 */
+	Hash_size = size<<10;
+	Hash_mask = (Hash_size >> 6) - 1;
+        _SDR1 = __pa(Hash) | (Hash_mask >> 10);
+	flush_tlb_all();
+
+	reset_SDR1();
+	printk("done\n");
+	return count;
+}
+
+
+static long long
+ppc_htab_lseek(struct inode * inode, struct file * file, 
+				 long long offset, int orig)
+{
+    switch (orig) {
+    case 0:
+	file->f_pos = offset;
+	return(file->f_pos);
+    case 1:
+	file->f_pos += offset;
+	return(file->f_pos);
+    case 2:
+	return(-EINVAL);
+    default:
+	return(-EINVAL);
+    }
+}
+
diff --git a/arch/ppc/kernel/ppc_ksyms.c b/arch/ppc/kernel/ppc_ksyms.c
new file mode 100644
index 000000000..fd2ad89ff
--- /dev/null
+++ b/arch/ppc/kernel/ppc_ksyms.c
@@ -0,0 +1,140 @@
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/smp.h>
+#include <linux/elfcore.h>
+#include <linux/sched.h>
+#include <linux/string.h>
+#include <linux/bios32.h>
+
+#include <asm/semaphore.h>
+#include <asm/processor.h>
+#include <asm/uaccess.h>
+#include <asm/io.h>
+#include <asm/atomic.h>
+#include <asm/bitops.h>
+#include <asm/checksum.h>
+#include <asm/pgtable.h>
+#include <asm/cuda.h>
+#include <asm/prom.h>
+#include <asm/system.h>
+#include <asm/pci-bridge.h>
+
+void transfer_to_handler();
+void int_return();
+void syscall_trace();
+void do_IRQ();
+void MachineCheckException();
+void AlignmentException();
+void ProgramCheckException();
+void SingleStepException();
+void FloatingPointCheckException();
+void sys_sigreturn();
+extern unsigned lost_interrupts;
+extern void do_lost_interrupts(unsigned long);
+
+EXPORT_SYMBOL(do_signal);
+EXPORT_SYMBOL(syscall_trace);
+EXPORT_SYMBOL(transfer_to_handler);
+EXPORT_SYMBOL(int_return);
+EXPORT_SYMBOL(do_IRQ);
+EXPORT_SYMBOL(init_task_union);
+EXPORT_SYMBOL(MachineCheckException);
+EXPORT_SYMBOL(AlignmentException);
+EXPORT_SYMBOL(ProgramCheckException);
+EXPORT_SYMBOL(SingleStepException);
+EXPORT_SYMBOL(sys_sigreturn);
+EXPORT_SYMBOL(lost_interrupts);
+EXPORT_SYMBOL(do_lost_interrupts);
+
+EXPORT_SYMBOL(atomic_add);
+EXPORT_SYMBOL(atomic_sub);
+EXPORT_SYMBOL(atomic_inc);
+EXPORT_SYMBOL(atomic_inc_return);
+EXPORT_SYMBOL(atomic_dec);
+EXPORT_SYMBOL(atomic_dec_return);
+EXPORT_SYMBOL(atomic_dec_and_test);
+
+EXPORT_SYMBOL(set_bit);
+EXPORT_SYMBOL(clear_bit);
+EXPORT_SYMBOL(change_bit);
+EXPORT_SYMBOL(test_and_set_bit);
+EXPORT_SYMBOL(test_and_clear_bit);
+EXPORT_SYMBOL(test_and_change_bit);
+#if 0
+EXPORT_SYMBOL(ffz);
+EXPORT_SYMBOL(find_first_zero_bit);
+EXPORT_SYMBOL(find_next_zero_bit);
+#endif
+
+EXPORT_SYMBOL(strcpy);
+EXPORT_SYMBOL(strncpy);
+EXPORT_SYMBOL(strcat);
+EXPORT_SYMBOL(strncat);
+EXPORT_SYMBOL(strchr);
+EXPORT_SYMBOL(strrchr);
+EXPORT_SYMBOL(strpbrk);
+EXPORT_SYMBOL(strtok);
+EXPORT_SYMBOL(strstr);
+EXPORT_SYMBOL(strlen);
+EXPORT_SYMBOL(strnlen);
+EXPORT_SYMBOL(strspn);
+EXPORT_SYMBOL(strcmp);
+EXPORT_SYMBOL(strncmp);
+EXPORT_SYMBOL(memset);
+EXPORT_SYMBOL(memcpy);
+EXPORT_SYMBOL(memmove);
+EXPORT_SYMBOL(memscan);
+EXPORT_SYMBOL(memcmp);
+
+/* EXPORT_SYMBOL(csum_partial); already in net/netsyms.c */
+EXPORT_SYMBOL(csum_partial_copy_generic);
+EXPORT_SYMBOL(ip_fast_csum);
+EXPORT_SYMBOL(csum_tcpudp_magic);
+
+EXPORT_SYMBOL(__copy_tofrom_user);
+EXPORT_SYMBOL(__clear_user);
+EXPORT_SYMBOL(__strncpy_from_user);
+EXPORT_SYMBOL(strlen_user);
+
+/*
+EXPORT_SYMBOL(inb);
+EXPORT_SYMBOL(inw);
+EXPORT_SYMBOL(inl);
+EXPORT_SYMBOL(outb);
+EXPORT_SYMBOL(outw);
+EXPORT_SYMBOL(outl);
+EXPORT_SYMBOL(outsl);*/
+
+EXPORT_SYMBOL(_insw);
+EXPORT_SYMBOL(_outsw);
+EXPORT_SYMBOL(_insl);
+EXPORT_SYMBOL(_outsl);
+EXPORT_SYMBOL(ioremap);
+
+EXPORT_SYMBOL(start_thread);
+
+EXPORT_SYMBOL(__down_interruptible);
+
+EXPORT_SYMBOL(__cli);
+EXPORT_SYMBOL(__sti);
+/*EXPORT_SYMBOL(__restore_flags);*/
+EXPORT_SYMBOL(_disable_interrupts);
+EXPORT_SYMBOL(_enable_interrupts);
+EXPORT_SYMBOL(flush_instruction_cache);
+EXPORT_SYMBOL(_get_PVR);
+EXPORT_SYMBOL(giveup_fpu);
+EXPORT_SYMBOL(flush_icache_range);
+EXPORT_SYMBOL(xchg_u32);
+
+EXPORT_SYMBOL(cuda_request);
+EXPORT_SYMBOL(cuda_send_request);
+EXPORT_SYMBOL(adb_register);
+EXPORT_SYMBOL(abort);
+EXPORT_SYMBOL(call_prom);
+EXPORT_SYMBOL(find_devices);
+EXPORT_SYMBOL(find_type_devices);
+EXPORT_SYMBOL(find_path_device);
+EXPORT_SYMBOL(get_property);
+EXPORT_SYMBOL(pci_io_base);
+EXPORT_SYMBOL(pci_device_loc);
+EXPORT_SYMBOL(note_scsi_host);
diff --git a/arch/ppc/kernel/prep_pci.c b/arch/ppc/kernel/prep_pci.c
new file mode 100644
index 000000000..ffacd94c9
--- /dev/null
+++ b/arch/ppc/kernel/prep_pci.c
@@ -0,0 +1,436 @@
+/*
+ * $Id: prep_pci.c,v 1.7 1997/08/23 22:46:02 cort Exp $
+ * PReP pci functions.
+ * Originally by Gary Thomas
+ * rewritten and updated by Cort Dougan (cort@cs.nmt.edu)
+ *
+ * The motherboard routes/maps will disappear shortly. -- Cort
+ */
+
+#include <linux/config.h>
+#include <linux/types.h>
+#include <linux/bios32.h>
+#include <linux/pci.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+
+#include <asm/byteorder.h>
+#include <asm/io.h>
+#include <asm/ptrace.h>
+#include <asm/processor.h>
+
+#define MAX_DEVNR 22
+
+/* Which PCI interrupt line does a given device [slot] use? */
+/* Note: This really should be two dimensional based in slot/pin used */
+unsigned char *Motherboard_map;
+unsigned char *Motherboard_map_name;
+
+/* How is the 82378 PIRQ mapping setup? */
+unsigned char *Motherboard_routes;
+
+/* Tables for known hardware */   
+
+/* Motorola PowerStack */
+static char Blackhawk_pci_IRQ_map[16] =
+{
+  	0,	/* Slot 0  - unused */
+  	0,	/* Slot 1  - unused */
+  	0,	/* Slot 2  - unused */
+  	0,	/* Slot 3  - unused */
+  	0,	/* Slot 4  - unused */
+  	0,	/* Slot 5  - unused */
+  	0,	/* Slot 6  - unused */
+  	0,	/* Slot 7  - unused */
+  	0,	/* Slot 8  - unused */
+  	0,	/* Slot 9  - unused */
+  	0,	/* Slot 10 - unused */
+  	0,	/* Slot 11 - unused */
+  	3,	/* Slot 12 - SCSI */
+  	0,	/* Slot 13 - unused */
+  	1,	/* Slot 14 - Ethernet */
+  	0,	/* Slot 15 - unused */
+};
+
+static char Blackhawk_pci_IRQ_routes[] =
+{
+   	0,	/* Line 0 - Unused */
+   	9,	/* Line 1 */
+   	11,	/* Line 2 */
+   	14,	/* Line 3 */
+   	15	/* Line 4 */
+};
+   
+/* Motorola MVME16xx */
+static char Genesis_pci_IRQ_map[16] =
+{
+  	0,	/* Slot 0  - unused */
+  	0,	/* Slot 1  - unused */
+  	0,	/* Slot 2  - unused */
+  	0,	/* Slot 3  - unused */
+  	0,	/* Slot 4  - unused */
+  	0,	/* Slot 5  - unused */
+  	0,	/* Slot 6  - unused */
+  	0,	/* Slot 7  - unused */
+  	0,	/* Slot 8  - unused */
+  	0,	/* Slot 9  - unused */
+  	0,	/* Slot 10 - unused */
+  	0,	/* Slot 11 - unused */
+  	3,	/* Slot 12 - SCSI */
+  	0,	/* Slot 13 - unused */
+  	1,	/* Slot 14 - Ethernet */
+  	0,	/* Slot 15 - unused */
+};
+
+static char Genesis_pci_IRQ_routes[] =
+{
+   	0,	/* Line 0 - Unused */
+   	10,	/* Line 1 */
+   	11,	/* Line 2 */
+   	14,	/* Line 3 */
+   	15	/* Line 4 */
+};
+   
+/* Motorola Series-E */
+static char Comet_pci_IRQ_map[16] =
+{
+  	0,	/* Slot 0  - unused */
+  	0,	/* Slot 1  - unused */
+  	0,	/* Slot 2  - unused */
+  	0,	/* Slot 3  - unused */
+  	0,	/* Slot 4  - unused */
+  	0,	/* Slot 5  - unused */
+  	0,	/* Slot 6  - unused */
+  	0,	/* Slot 7  - unused */
+  	0,	/* Slot 8  - unused */
+  	0,	/* Slot 9  - unused */
+  	0,	/* Slot 10 - unused */
+  	0,	/* Slot 11 - unused */
+  	3,	/* Slot 12 - SCSI */
+  	0,	/* Slot 13 - unused */
+  	1,	/* Slot 14 - Ethernet */
+  	0,	/* Slot 15 - unused */
+};
+
+static char Comet_pci_IRQ_routes[] =
+{
+   	0,	/* Line 0 - Unused */
+   	10,	/* Line 1 */
+   	11,	/* Line 2 */
+   	14,	/* Line 3 */
+   	15	/* Line 4 */
+};
+
+/*
+ * ibm 830 (and 850?).
+ * This is actually based on the Carolina motherboard
+ * -- Cort
+ */
+static char ibm8xx_pci_IRQ_map[23] = {
+        0, /* Slot 0  - unused */
+        0, /* Slot 1  - unused */
+        0, /* Slot 2  - unused */
+        0, /* Slot 3  - unused */
+        0, /* Slot 4  - unused */
+        0, /* Slot 5  - unused */
+        0, /* Slot 6  - unused */
+        0, /* Slot 7  - unused */
+        0, /* Slot 8  - unused */
+        0, /* Slot 9  - unused */
+        0, /* Slot 10 - unused */
+        0, /* Slot 11 - FireCoral */
+        4, /* Slot 12 - Ethernet  PCIINTD# */
+        2, /* Slot 13 - PCI Slot #2 */
+        2, /* Slot 14 - S3 Video PCIINTD# */
+        0, /* Slot 15 - onboard SCSI (INDI) [1] */
+        3, /* Slot 16 - NCR58C810 RS6000 Only PCIINTC# */
+        0, /* Slot 17 - unused */
+        2, /* Slot 18 - PCI Slot 2 PCIINTx# (See below) */
+        0, /* Slot 19 - unused */
+        0, /* Slot 20 - unused */
+        0, /* Slot 21 - unused */
+        2, /* Slot 22 - PCI slot 1 PCIINTx# (See below) */
+};
+static char ibm8xx_pci_IRQ_routes[] = {
+        0,      /* Line 0 - unused */
+        13,     /* Line 1 */
+        10,     /* Line 2 */
+        15,     /* Line 3 */
+        15,     /* Line 4 */
+};
+
+/* IBM Nobis and 850 */
+static char Nobis_pci_IRQ_map[23] ={
+        0, /* Slot 0  - unused */
+        0, /* Slot 1  - unused */
+        0, /* Slot 2  - unused */
+        0, /* Slot 3  - unused */
+        0, /* Slot 4  - unused */
+        0, /* Slot 5  - unused */
+        0, /* Slot 6  - unused */
+        0, /* Slot 7  - unused */
+        0, /* Slot 8  - unused */
+        0, /* Slot 9  - unused */
+        0, /* Slot 10 - unused */
+        0, /* Slot 11 - unused */
+        3, /* Slot 12 - SCSI */
+        0, /* Slot 13 - unused */
+        0, /* Slot 14 - unused */
+        0, /* Slot 15 - unused */
+};
+
+static char Nobis_pci_IRQ_routes[] = {
+        0, /* Line 0 - Unused */
+        13, /* Line 1 */
+        13, /* Line 2 */
+        13, /* Line 3 */
+        13      /* Line 4 */
+};
+
+/* We have to turn on LEVEL mode for changed IRQ's */
+/* All PCI IRQ's need to be level mode, so this should be something
+ * other than hard-coded as well... IRQ's are individually mappable
+ * to either edge or level.
+ */
+#define CAROLINA_IRQ_EDGE_MASK_LO   0x00  /* IRQ's 0-7  */
+#define CAROLINA_IRQ_EDGE_MASK_HI   0xA4  /* IRQ's 8-15 [10,13,15] */
+
+int
+prep_pcibios_read_config_dword (unsigned char bus,
+			   unsigned char dev, unsigned char offset, unsigned int *val)
+{
+	unsigned long _val;
+	unsigned long *ptr;
+	dev >>= 3;
+	
+	if ((bus != 0) || (dev > MAX_DEVNR))
+	{
+		*val = 0xFFFFFFFF;
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	} else
+	{
+		ptr = (unsigned long *)(0x80800000 | (1<<dev) | offset);
+		_val = le32_to_cpu(*ptr);
+	}
+	*val = _val;
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int
+prep_pcibios_read_config_word (unsigned char bus,
+			  unsigned char dev, unsigned char offset, unsigned short *val)
+{
+	unsigned short _val;
+	unsigned short *ptr;
+	dev >>= 3;
+	if ((bus != 0) || (dev > MAX_DEVNR))
+	{
+		*val = (unsigned short)0xFFFFFFFF;
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	} else
+	{
+		ptr = (unsigned short *)(0x80800000 | (1<<dev) | offset);
+		_val = le16_to_cpu(*ptr);
+	}
+	*val = _val;
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int
+prep_pcibios_read_config_byte (unsigned char bus,
+			  unsigned char dev, unsigned char offset, unsigned char *val)
+{
+	unsigned char _val;
+	volatile unsigned char *ptr;
+	dev >>= 3;
+	/* Note: the configuration registers don't always have this right! */
+	if (offset == PCI_INTERRUPT_LINE)
+	{
+		*val = Motherboard_routes[Motherboard_map[dev]];
+/*printk("dev %d map %d route %d on board %d\n",
+  dev,Motherboard_map[dev],
+  Motherboard_routes[Motherboard_map[dev]],
+  *(unsigned char *)(0x80800000 | (1<<dev) | (offset ^ 1)));*/
+		return PCIBIOS_SUCCESSFUL;
+	}
+	if ((bus != 0) || (dev > MAX_DEVNR))
+	{
+		*(unsigned long *)val = (unsigned long) 0xFFFFFFFF;
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	} else
+	{
+		ptr = (unsigned char *)(0x80800000 | (1<<dev) | (offset ^ 1));
+		_val = *ptr;
+	}
+	*val = _val;
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int
+prep_pcibios_write_config_dword (unsigned char bus,
+			    unsigned char dev, unsigned char offset, unsigned int val)
+{
+	unsigned long _val;
+	unsigned long *ptr;
+	dev >>= 3;
+	_val = le32_to_cpu(val);
+	if ((bus != 0) || (dev > MAX_DEVNR))
+	{
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	} else
+	{
+		ptr = (unsigned long *)(0x80800000 | (1<<dev) | offset);
+		*ptr = _val;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int
+prep_pcibios_write_config_word (unsigned char bus,
+			   unsigned char dev, unsigned char offset, unsigned short val)
+{
+	unsigned short _val;
+	unsigned short *ptr;
+	dev >>= 3;
+	_val = le16_to_cpu(val);
+	if ((bus != 0) || (dev > MAX_DEVNR))
+	{
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	} else
+	{
+		ptr = (unsigned short *)(0x80800000 | (1<<dev) | offset);
+		*ptr = _val;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int
+prep_pcibios_write_config_byte (unsigned char bus,
+			   unsigned char dev, unsigned char offset, unsigned char val)
+{
+	unsigned char _val;
+	unsigned char *ptr;
+	dev >>= 3;
+	_val = val;
+	if ((bus != 0) || (dev > MAX_DEVNR))
+	{
+		return PCIBIOS_DEVICE_NOT_FOUND;
+	} else
+	{
+		ptr = (unsigned char *)(0x80800000 | (1<<dev) | (offset^1));
+		*ptr = _val;
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int prep_pcibios_find_device (unsigned short vendor, unsigned short device_id,
+			   unsigned short index, unsigned char *bus,
+			   unsigned char *devfn)
+{
+    unsigned int curr = 0;
+    struct pci_dev *dev;
+/*printk("pcibios_find_device(): vendor %04x devid %04x index %d\n",    
+       vendor,device_id,index);*/
+    for (dev = pci_devices; dev; dev = dev->next) {
+/*printk("   dev->vendor %04x dev->device %04x\n",
+       dev->vendor,dev->device);*/
+	if (dev->vendor == vendor && dev->device == device_id) {
+	    if (curr == index) {
+		*devfn = dev->devfn;
+		*bus = dev->bus->number;
+		return PCIBIOS_SUCCESSFUL;
+	    }
+	    ++curr;
+	}
+    }
+    return PCIBIOS_DEVICE_NOT_FOUND;
+}
+
+/*
+ * Given the class, find the n'th instance of that device
+ * in the system.
+ */
+int prep_pcibios_find_class (unsigned int class_code, unsigned short index,
+			  unsigned char *bus, unsigned char *devfn)
+{
+    unsigned int curr = 0;
+    struct pci_dev *dev;
+
+    for (dev = pci_devices; dev; dev = dev->next) {
+	if (dev->class == class_code) {
+	    if (curr == index) {
+		*devfn = dev->devfn;
+		*bus = dev->bus->number;
+		return PCIBIOS_SUCCESSFUL;
+	    }
+	    ++curr;
+	}
+    }
+    return PCIBIOS_DEVICE_NOT_FOUND;
+}
+
+__initfunc(unsigned long route_pci_interrupts(void))
+{
+	unsigned char *ibc_pirq = (unsigned char *)0x80800860;
+	unsigned char *ibc_pcicon = (unsigned char *)0x80800840;
+	int i;
+	
+	if ( _machine == _MACH_Motorola)
+	{ 
+		switch (inb(0x800) & 0xF0)
+		{
+		case 0x10: /* MVME16xx */
+			Motherboard_map_name = "Genesis";
+			Motherboard_map = Genesis_pci_IRQ_map;
+			Motherboard_routes = Genesis_pci_IRQ_routes;
+			break;
+		case 0x20: /* Series E */
+			Motherboard_map_name = "Series E";
+			Motherboard_map = Comet_pci_IRQ_map;
+			Motherboard_routes = Comet_pci_IRQ_routes;
+			break;
+		case 0x40: /* PowerStack */
+		default: /* Can't hurt, can it? */
+			Motherboard_map_name = "Blackhawk (Powerstack)";
+			Motherboard_map = Blackhawk_pci_IRQ_map;
+			Motherboard_routes = Blackhawk_pci_IRQ_routes;
+			break;
+		}
+	} else if ( _machine == _MACH_IBM )
+	{
+		unsigned char pl_id;
+		
+		if (inb(0x0852) == 0xFF) {
+			Motherboard_map_name = "IBM 850/860 Portable\n";
+			Motherboard_map = Nobis_pci_IRQ_map;
+			Motherboard_routes = Nobis_pci_IRQ_routes;
+		} else {
+			Motherboard_map_name = "IBM 8xx (Carolina)";
+			Motherboard_map = ibm8xx_pci_IRQ_map;
+			Motherboard_routes = ibm8xx_pci_IRQ_routes;
+		}
+		/*printk("Changing IRQ mode\n");*/
+		pl_id=inb(0x04d0);
+		/*printk("Low mask is %#0x\n", pl_id);*/
+		outb(pl_id|CAROLINA_IRQ_EDGE_MASK_LO, 0x04d0);
+		
+		pl_id=inb(0x04d1);
+		/*printk("Hi mask is  %#0x\n", pl_id);*/
+		outb(pl_id|CAROLINA_IRQ_EDGE_MASK_HI, 0x04d1);
+		pl_id=inb(0x04d1);
+		/*printk("Hi mask now %#0x\n", pl_id);*/
+	} else
+	{
+		printk("No known machine pci routing!\n");
+		return -1;
+	}
+	
+	/* Set up mapping from slots */
+	for (i = 1;  i <= 4;  i++)
+	{
+		ibc_pirq[i-1] = Motherboard_routes[i];
+	}
+	/* Enable PCI interrupts */
+	*ibc_pcicon |= 0x20;
+	return 0;
+}
diff --git a/arch/ppc/kernel/prep_setup.c b/arch/ppc/kernel/prep_setup.c
new file mode 100644
index 000000000..18e013964
--- /dev/null
+++ b/arch/ppc/kernel/prep_setup.c
@@ -0,0 +1,294 @@
+/*
+ *  linux/arch/ppc/kernel/setup.c
+ *
+ *  Copyright (C) 1995  Linus Torvalds
+ *  Adapted from 'alpha' version by Gary Thomas
+ *  Modified by Cort Dougan (cort@cs.nmt.edu)
+ */
+
+/*
+ * bootup setup stuff..
+ */
+
+#include <linux/config.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/ptrace.h>
+#include <linux/malloc.h>
+#include <linux/user.h>
+#include <linux/a.out.h>
+#include <linux/tty.h>
+#include <linux/major.h>
+#include <linux/interrupt.h>
+#include <linux/reboot.h>
+#include <linux/init.h>
+#include <linux/blk.h>
+#include <linux/ioport.h>
+
+#include <asm/mmu.h>
+#include <asm/processor.h>
+#include <asm/residual.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/ide.h>
+
+/* for the mac fs */
+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;
+
+unsigned long empty_zero_page[1024];
+extern unsigned char aux_device_present;
+
+#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
+
+
+extern char saved_command_line[256];
+
+void prep_ide_init_hwif_ports (ide_ioreg_t *p, ide_ioreg_t base, int *irq)
+{
+	ide_ioreg_t port = base;
+	int i = 8;
+
+	while (i--)
+		*p++ = port++;
+	*p++ = base + 0x206;
+	if (irq != NULL)
+		*irq = 0;
+}
+
+
+int
+prep_get_cpuinfo(char *buffer)
+{
+	extern char *Motherboard_map_name;
+	extern RESIDUAL res;
+	int i;
+	int pvr = _get_PVR();
+	int len;
+	char *model;
+  
+	switch (pvr>>16)
+	{
+	case 1:
+		model = "601";
+		break;
+	case 3:
+		model = "603";
+		break;
+	case 4:
+		model = "604";
+		break;
+	case 6:
+		model = "603e";
+		break;
+	case 7:
+		model = "603ev";
+		break;
+	default:
+		model = "unknown";
+		break;
+	}
+  
+#ifdef __SMP__
+#define CD(X)		(cpu_data[n].X)  
+#else
+#define CD(X) (X)
+#define CPUN 0
+#endif
+  
+	len = sprintf(buffer,"processor\t: %d\n"
+		      "cpu\t\t: %s\n"
+		      "revision\t: %d.%d\n"
+		      "upgrade\t\t: %s\n"
+		      "clock\t\t: %dMHz\n"
+		      "bus clock\t: %dMHz\n"
+		      "machine\t\t: %s (sn %s)\n"
+		      "pci map\t\t: %s\n",
+		      CPUN,
+		      model,
+		      MAJOR(pvr), MINOR(pvr),
+		      (inb(IBM_EQUIP_PRESENT) & 2) ? "not upgrade" : "upgrade",
+		      (res.VitalProductData.ProcessorHz > 1024) ?
+		      res.VitalProductData.ProcessorHz>>20 :
+		      res.VitalProductData.ProcessorHz,
+		      (res.VitalProductData.ProcessorBusHz > 1024) ?
+		      res.VitalProductData.ProcessorBusHz>>20 :
+		      res.VitalProductData.ProcessorBusHz,
+		      res.VitalProductData.PrintableModel,
+		      res.VitalProductData.Serial,
+		      Motherboard_map_name
+		);
+  
+	/* print info about SIMMs */
+	len += sprintf(buffer+len,"simms\t\t: ");
+	for ( i = 0 ; (res.ActualNumMemories) && (i < MAX_MEMS) ; i++ )
+	{
+		if ( res.Memories[i].SIMMSize != 0 )
+			len += sprintf(buffer+len,"%d:%dM ",i,
+				       (res.Memories[i].SIMMSize > 1024) ?
+				       res.Memories[i].SIMMSize>>20 :
+				       res.Memories[i].SIMMSize);
+	}
+	len += sprintf(buffer+len,"\n");
+
+	/* TLB */
+	len += sprintf(buffer+len,"tlb\t\t:");
+	switch(res.VitalProductData.TLBAttrib)
+	{
+	case CombinedTLB:
+		len += sprintf(buffer+len," %d entries\n",
+			       res.VitalProductData.TLBSize);
+		break;
+	case SplitTLB:
+		len += sprintf(buffer+len," (split I/D) %d/%d entries\n",
+			       res.VitalProductData.I_TLBSize,
+			       res.VitalProductData.D_TLBSize);
+		break;
+	case NoneTLB:
+		len += sprintf(buffer+len," not present\n");
+		break;
+	}
+
+	/* L1 */
+	len += sprintf(buffer+len,"l1\t\t: ");
+	switch(res.VitalProductData.CacheAttrib)
+	{
+	case CombinedCAC:
+		len += sprintf(buffer+len,"%dkB LineSize\n",
+			       res.VitalProductData.CacheSize,
+			       res.VitalProductData.CacheLineSize);
+		break;
+	case SplitCAC:
+		len += sprintf(buffer+len,"(split I/D) %dkB/%dkB Linesize %dB/%dB\n",
+			       res.VitalProductData.I_CacheSize,
+			       res.VitalProductData.D_CacheSize,
+			       res.VitalProductData.D_CacheLineSize,
+			       res.VitalProductData.D_CacheLineSize);
+		break;
+	case NoneCAC:
+		len += sprintf(buffer+len,"not present\n");
+		break;
+	}
+
+	/* L2 */
+	if ( (inb(IBM_EQUIP_PRESENT) & 1) == 0) /* l2 present */
+	{
+		len += sprintf(buffer+len,"l2\t\t: %dkB %s\n",
+			       ((inb(IBM_L2_STATUS) >> 5) & 1) ? 512 : 256,
+			       (inb(IBM_SYS_CTL) & 64) ? "enabled" : "disabled");
+	}
+	else
+	{
+		len += sprintf(buffer+len,"l2\t\t: not present\n");
+	}
+
+
+	len += sprintf(buffer+len, "bogomips\t: %lu.%02lu\n",
+		       CD(loops_per_sec+2500)/500000,
+		       (CD(loops_per_sec+2500)/5000) % 100);
+  
+	/*
+	 * Ooh's and aah's info about zero'd pages in idle task
+	 */ 
+	{
+		extern unsigned int zerocount, zerototal, zeropage_hits,zeropage_calls;
+		len += sprintf(buffer+len,"zero pages\t: total %u (%uKb) "
+			       "current: %u (%uKb) hits: %u/%u (%lu%%)\n",
+			       zerototal, (zerototal*PAGE_SIZE)>>10,
+			       zerocount, (zerocount*PAGE_SIZE)>>10,
+			       zeropage_hits,zeropage_calls,
+			       /* : 1 below is so we don't div by zero */
+			       (zeropage_hits*100) /
+			            ((zeropage_calls)?zeropage_calls:1));
+	}
+	return len;
+}
+
+__initfunc(void
+prep_setup_arch(char **cmdline_p, unsigned long * memory_start_p,
+	   unsigned long * memory_end_p))
+{
+	extern char cmd_line[];
+	extern char _etext[], _edata[], _end[];
+	extern int panic_timeout;
+	unsigned char reg;
+
+	/* Save unparsed command line copy for /proc/cmdline */
+	strcpy( saved_command_line, cmd_line );
+	*cmdline_p = cmd_line;
+  
+	*memory_start_p = (unsigned long) Hash+Hash_size;
+	(unsigned long *)*memory_end_p = (unsigned long *)(res.TotalMemory+KERNELBASE);
+
+	/* init to some ~sane value until calibrate_delay() runs */
+	loops_per_sec = 50000000;
+	
+	/* reboot on panic */	
+	panic_timeout = 180;
+	
+	init_task.mm->start_code = PAGE_OFFSET;
+	init_task.mm->end_code = (unsigned long) _etext;
+	init_task.mm->end_data = (unsigned long) _edata;
+	init_task.mm->brk = (unsigned long) _end;	
+	
+	aux_device_present = 0xaa;
+	/* Set up floppy in PS/2 mode */
+	outb(0x09, SIO_CONFIG_RA);
+	reg = inb(SIO_CONFIG_RD);
+	reg = (reg & 0x3F) | 0x40;
+	outb(reg, SIO_CONFIG_RD);
+	outb(reg, SIO_CONFIG_RD);	/* Have to write twice to change! */
+	
+	switch ( _machine )
+	{
+	case _MACH_IBM:
+		ROOT_DEV = to_kdev_t(0x0301); /* hda1 */
+		break;
+	case _MACH_Motorola:
+		ROOT_DEV = to_kdev_t(0x0801); /* sda1 */
+		break;
+	}
+	
+#ifdef CONFIG_BLK_DEV_RAM
+#if 0
+	ROOT_DEV = to_kdev_t(0x0200); /* floppy */  
+	rd_prompt = 1;
+	rd_doload = 1;
+	rd_image_start = 0;
+#endif
+	/* initrd_start and size are setup by boot/head.S and kernel/head.S */
+	if ( initrd_start )
+	{
+		if (initrd_end > *memory_end_p)
+		{
+			printk("initrd extends beyond end of memory "
+			       "(0x%08lx > 0x%08lx)\ndisabling initrd\n",
+			       initrd_end,*memory_end_p);
+			initrd_start = 0;
+		}
+	}
+#endif
+
+	printk("Boot arguments: %s\n", cmd_line);
+
+	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");
+}
diff --git a/arch/ppc/kernel/prep_time.c b/arch/ppc/kernel/prep_time.c
new file mode 100644
index 000000000..da0151ede
--- /dev/null
+++ b/arch/ppc/kernel/prep_time.c
@@ -0,0 +1,239 @@
+/*
+ *  linux/arch/i386/kernel/time.c
+ *
+ *  Copyright (C) 1991, 1992, 1995  Linus Torvalds
+ *
+ * Adapted for PowerPC (PreP) by Gary Thomas
+ * Modified by Cort Dougan (cort@cs.nmt.edu)
+ *  copied and modified from intel version
+ *
+ */
+#include <linux/config.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/param.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <linux/interrupt.h>
+#include <linux/timex.h>
+#include <linux/kernel_stat.h>
+
+#include <asm/segment.h>
+#include <asm/io.h>
+#include <asm/processor.h>
+#include <asm/nvram.h>
+
+#include "time.h"
+
+/*
+ * The motorola uses the m48t18 rtc (includes DS1643) whose registers
+ * are at a higher end of nvram (1ff8-1fff) than the ibm mc146818
+ * rtc (ds1386) which has regs at addr 0-d).  The intel gets
+ * past this because the bios emulates the mc146818.
+ *
+ * Why in the world did they have to use different clocks?
+ *
+ * Right now things are hacked to check which machine we're on then
+ * use the appropriate macro.  This is very very ugly and I should
+ * probably have a function that checks which machine we're on then
+ * does things correctly transparently or a function pointer which
+ * is setup at boot time to use the correct addresses.
+ * -- Cort
+ */
+/*
+ * translate from mc146818 to m48t18  addresses
+ */
+unsigned int clock_transl[] = { MOTO_RTC_SECONDS,0 /* alarm */,
+		       MOTO_RTC_MINUTES,0 /* alarm */,
+		       MOTO_RTC_HOURS,0 /* alarm */,                 /*  4,5 */
+		       MOTO_RTC_DAY_OF_WEEK,
+		       MOTO_RTC_DAY_OF_MONTH,
+		       MOTO_RTC_MONTH,
+		       MOTO_RTC_YEAR,                    /* 9 */
+		       MOTO_RTC_CONTROLA, MOTO_RTC_CONTROLB /* 10,11 */
+};
+
+int prep_cmos_clock_read(int addr)
+{
+	if ( _machine == _MACH_IBM )
+		return CMOS_READ(addr);
+	else if ( _machine == _MACH_Motorola )
+	{
+		outb(clock_transl[addr]>>8, NVRAM_AS1);
+		outb(clock_transl[addr], NVRAM_AS0);
+		return (inb(NVRAM_DATA));
+	}
+
+	printk("Unknown machine in prep_cmos_clock_read()!\n");
+	return -1;
+}
+
+void prep_cmos_clock_write(unsigned long val, int addr)
+{
+	if ( _machine == _MACH_IBM )
+	{
+		CMOS_WRITE(val,addr);
+		return;
+	}
+	else if ( _machine == _MACH_Motorola )
+	{
+		outb(clock_transl[addr]>>8, NVRAM_AS1);
+		outb(clock_transl[addr], NVRAM_AS0);
+		outb(val,NVRAM_DATA);
+		return;
+	}
+	printk("Unknown machine in prep_cmos_clock_write()!\n");
+}
+
+/*
+ * Set the hardware clock. -- Cort
+ */
+int prep_set_rtc_time(unsigned long nowtime)
+{
+	unsigned char save_control, save_freq_select;
+	struct rtc_time tm;
+
+	to_tm(nowtime, &tm);
+
+	save_control = prep_cmos_clock_read(RTC_CONTROL); /* tell the clock it's being set */
+
+	prep_cmos_clock_write((save_control|RTC_SET), RTC_CONTROL);
+
+	save_freq_select = prep_cmos_clock_read(RTC_FREQ_SELECT); /* stop and reset prescaler */
+	
+	prep_cmos_clock_write((save_freq_select|RTC_DIV_RESET2), RTC_FREQ_SELECT);
+
+        tm.tm_year -= 1900;
+	if (!(save_control & RTC_DM_BINARY) || RTC_ALWAYS_BCD) {
+		BIN_TO_BCD(tm.tm_sec);
+		BIN_TO_BCD(tm.tm_min);
+		BIN_TO_BCD(tm.tm_hour);
+		BIN_TO_BCD(tm.tm_mon);
+		BIN_TO_BCD(tm.tm_mday);
+		BIN_TO_BCD(tm.tm_year);
+	}
+	prep_cmos_clock_write(tm.tm_sec,RTC_SECONDS);
+	prep_cmos_clock_write(tm.tm_min,RTC_MINUTES);
+	prep_cmos_clock_write(tm.tm_hour,RTC_HOURS);
+	prep_cmos_clock_write(tm.tm_mon,RTC_MONTH);
+	prep_cmos_clock_write(tm.tm_mday,RTC_DAY_OF_MONTH);
+	prep_cmos_clock_write(tm.tm_year,RTC_YEAR);
+	
+	/* The following flags have to be released exactly in this order,
+	 * otherwise the DS12887 (popular MC146818A clone with integrated
+	 * battery and quartz) will not reset the oscillator and will not
+	 * update precisely 500 ms later. You won't find this mentioned in
+	 * the Dallas Semiconductor data sheets, but who believes data
+	 * sheets anyway ...                           -- Markus Kuhn
+	 */
+	prep_cmos_clock_write(save_control, RTC_CONTROL);
+	prep_cmos_clock_write(save_freq_select, RTC_FREQ_SELECT);
+
+	if ( (time_state == TIME_ERROR) || (time_state == TIME_BAD) )
+		time_state = TIME_OK;
+	return 0;
+}
+
+unsigned long prep_get_rtc_time(void)
+{
+	unsigned int year, mon, day, hour, min, sec;
+	int i;
+
+	/* The Linux interpretation of the CMOS clock register contents:
+	 * When the Update-In-Progress (UIP) flag goes from 1 to 0, the
+	 * RTC registers show the second which has precisely just started.
+	 * Let's hope other operating systems interpret the RTC the same way.
+	 */
+	/* read RTC exactly on falling edge of update flag */
+	for (i = 0 ; i < 1000000 ; i++)	/* may take up to 1 second... */
+		if (prep_cmos_clock_read(RTC_FREQ_SELECT) & RTC_UIP)
+			break;
+	for (i = 0 ; i < 1000000 ; i++)	/* must try at least 2.228 ms */
+		if (!(prep_cmos_clock_read(RTC_FREQ_SELECT) & RTC_UIP))
+			break;
+	do { /* Isn't this overkill ? UIP above should guarantee consistency */
+		sec = prep_cmos_clock_read(RTC_SECONDS);
+		min = prep_cmos_clock_read(RTC_MINUTES);
+		hour = prep_cmos_clock_read(RTC_HOURS);
+		day = prep_cmos_clock_read(RTC_DAY_OF_MONTH);
+		mon = prep_cmos_clock_read(RTC_MONTH);
+		year = prep_cmos_clock_read(RTC_YEAR);
+	} while (sec != prep_cmos_clock_read(RTC_SECONDS));
+	if (!(prep_cmos_clock_read(RTC_CONTROL) & RTC_DM_BINARY) || RTC_ALWAYS_BCD)
+	  {
+	    BCD_TO_BIN(sec);
+	    BCD_TO_BIN(min);
+	    BCD_TO_BIN(hour);
+	    BCD_TO_BIN(day);
+	    BCD_TO_BIN(mon);
+	    BCD_TO_BIN(year);
+	  }
+	if ((year += 1900) < 1970)
+		year += 100;
+	return mktime(year, mon, day, hour, min, sec);
+}
+
+#if 0
+void time_init(void)
+{
+	void (*irq_handler)(int, void *,struct pt_regs *);
+	
+	xtime.tv_sec = prep_get_rtc_time();
+	xtime.tv_usec = 0;
+	
+	prep_calibrate_decr();
+	
+	/* If we have the CPU hardware time counters, use them */
+	irq_handler = timer_interrupt;
+	if (request_irq(TIMER_IRQ, irq_handler, 0, "timer", NULL) != 0)
+		panic("Could not allocate timer IRQ!");
+}
+
+/*
+ * timer_interrupt() needs to keep up the real-time clock,
+ * as well as call the "do_timer()" routine every clocktick
+ */
+static inline void timer_interrupt(int irq, void *dev, struct pt_regs * regs)
+{
+	prep_calibrate_decr_handler(irq,dev,regs);
+	do_timer(regs);
+
+	/* update the hw clock if:
+	 * the time is marked out of sync (TIME_ERROR)
+	 * or ~11 minutes have expired since the last update -- Cort
+	 * If we have an externally synchronized Linux clock, then update
+	 * CMOS clock accordingly every ~11 minutes. prep_set_rtc_mmss() has to be
+	 * called as close as possible to 500 ms before the new second starts.
+	 */
+	if ( time_state == TIME_BAD ||
+	     xtime.tv_sec > last_rtc_update + 660 )
+	/*if (time_state != TIME_BAD && xtime.tv_sec > last_rtc_update + 660 &&
+	    xtime.tv_usec > 500000 - (tick >> 1) &&
+	    xtime.tv_usec < 500000 + (tick >> 1))*/
+		if (prep_set_rtc_time(xtime.tv_sec) == 0)
+			last_rtc_update = xtime.tv_sec;
+		else
+			last_rtc_update = xtime.tv_sec - 600; /* do it again in 60 s */
+
+
+#ifdef CONFIG_HEARTBEAT
+	/* use hard disk LED as a heartbeat instead -- much more useful
+	   for debugging -- Cort */
+	switch(kstat.interrupts[0] % 101)
+	{
+	/* act like an actual heart beat -- ie thump-thump-pause... */
+	case 0:
+	case 20:
+		outb(1,IBM_HDD_LED);
+		break;
+	case 7:
+	case 27:
+		outb(0,IBM_HDD_LED);
+		break;
+	case 100:
+		break;
+	}
+#endif
+}
+#endif
diff --git a/arch/ppc/kernel/process.c b/arch/ppc/kernel/process.c
index f24872063..5e194fedb 100644
--- a/arch/ppc/kernel/process.c
+++ b/arch/ppc/kernel/process.c
@@ -7,7 +7,7 @@
  *  Derived from "arch/i386/kernel/process.c"
  *    Copyright (C) 1995  Linus Torvalds
  *
- *  Modified by Cort Dougan (cort@cs.nmt.edu) and
+ *  Updated and modified by Cort Dougan (cort@cs.nmt.edu) and
  *  Paul Mackerras (paulus@cs.anu.edu.au)
  *
  *  This program is free software; you can redistribute it and/or
@@ -28,22 +28,22 @@
 #include <linux/ptrace.h>
 #include <linux/malloc.h>
 #include <linux/user.h>
-#include <linux/a.out.h>
+#include <linux/elf.h>
 #include <linux/config.h>
+#include <linux/elf.h>
 
 #include <asm/pgtable.h>
 #include <asm/uaccess.h>
 #include <asm/system.h>
 #include <asm/io.h>
 #include <asm/smp_lock.h>
+#include <asm/processor.h>
 
-int dump_fpu(void);
+int dump_fpu(struct pt_regs *regs, elf_fpregset_t *fpregs);
 void switch_to(struct task_struct *, struct task_struct *);
-void print_backtrace(unsigned long *);
-void show_regs(struct pt_regs * regs);
-void inline zero_paged(void);
 extern unsigned long _get_SP(void);
 
+
 #undef SHOW_TASK_SWITCHES 1
 #undef CHECK_STACK 1
 #undef IDLE_ZERO 1
@@ -59,7 +59,7 @@ task_top(struct task_struct *tsk)
 {
 	return ((unsigned long)tsk) + sizeof(struct task_struct);
 }
-	
+
 static struct vm_area_struct init_mmap = INIT_MMAP;
 static struct fs_struct init_fs = INIT_FS;
 static struct files_struct init_files = INIT_FILES;
@@ -69,9 +69,12 @@ struct mm_struct init_mm = INIT_MM;
 union task_union init_task_union = { INIT_TASK };
 
 int
-dump_fpu(void)
+dump_fpu(struct pt_regs *regs, elf_fpregset_t *fpregs)
 {
-	return (1);
+	if (last_task_used_math == current)
+		giveup_fpu();
+	memcpy(fpregs, &current->tss.fpr[0], sizeof(*fpregs));
+	return 1;
 }
 
 /* check to make sure the kernel stack is healthy */
@@ -80,7 +83,6 @@ int check_stack(struct task_struct *tsk)
 	unsigned long stack_top = kernel_stack_top(tsk);
 	unsigned long tsk_top = task_top(tsk);
 	int ret = 0;
-	unsigned long *i;
 
 #if 0	
 	/* check tss magic */
@@ -98,7 +100,7 @@ int check_stack(struct task_struct *tsk)
 	if ( (tsk->tss.ksp > stack_top) || (tsk->tss.ksp < tsk_top) )
 	{
 		printk("stack out of bounds: %s/%d\n"
-		       " tsk_top %08x ksp %08x stack_top %08x\n",
+		       " tsk_top %08lx ksp %08lx stack_top %08lx\n",
 		       tsk->comm,tsk->pid,
 		       tsk_top, tsk->tss.ksp, stack_top);
 		ret |= 2;
@@ -108,7 +110,7 @@ int check_stack(struct task_struct *tsk)
 	if ( (tsk == current) && ((_get_SP() > stack_top ) || (_get_SP() < tsk_top)) )
 	{
 		printk("current stack ptr out of bounds: %s/%d\n"
-		       " tsk_top %08x sp %08x stack_top %08x\n",
+		       " tsk_top %08lx sp %08lx stack_top %08lx\n",
 		       current->comm,current->pid,
 		       tsk_top, _get_SP(), stack_top);
 		ret |= 4;
@@ -143,15 +145,12 @@ switch_to(struct task_struct *prev, struct task_struct *new)
 {
 	struct thread_struct *new_tss, *old_tss;
 	int s = _disable_interrupts();
-	struct pt_regs *regs = (struct pt_regs *)(new->tss.ksp+STACK_FRAME_OVERHEAD);
 
 #if CHECK_STACK
 	check_stack(prev);
 	check_stack(new);
 #endif
-        /* turn off fpu for task last to run */
-	/*prev->tss.regs->msr &= ~MSR_FP;*/
-	
+
 #ifdef SHOW_TASK_SWITCHES
 	printk("%s/%d (%x) -> %s/%d (%x) ctx %x\n",
 	       prev->comm,prev->pid,prev->tss.regs->nip,
@@ -160,289 +159,29 @@ switch_to(struct task_struct *prev, struct task_struct *new)
 	new_tss = &new->tss;
 	old_tss = &current->tss;
 	_switch(old_tss, new_tss, new->mm->context);
-        /* turn off fpu for task last to run */	
 	_enable_interrupts(s);
 }
-	
-#include <linux/mc146818rtc.h>
-asmlinkage int sys_debug(long a, long b, long c, long d, long e, long f,struct pt_regs *regs)
-{
-#if 1
-	struct task_struct *p;
-	printk("sys_debug(): r3 %x r4 %x r5 %x r6 %x\n", a,b,c,d);
-	printk("last %x\n", last_task_used_math);
-	printk("cur %x regs %x/%x tss %x/%x\n",
-	       current, current->tss.regs,regs,&current->tss,current->tss);
-	for_each_task(p)
-	{
-		if ((long)p < KERNELBASE)
-		{
-			printk("nip %x lr %x r3 %x\n", regs->nip,regs->link,a);
-			print_mm_info();
-			__cli();
-			while(1);
-		}
-	}
-	return regs->gpr[3];
-#endif
-#if 0
-	/* set the time in the cmos clock */
-	unsigned long hwtime, nowtime;
-	struct rtc_time tm;
-	
-	hwtime = get_cmos_time();
-	to_tm(hwtime, &tm);
-	printk("hw: H:M:S M/D/Y %02d:%02d:%02d %d/%d/%d\n", 
-	       tm.tm_hour, tm.tm_min, tm.tm_sec, tm.tm_mon,
-	       tm.tm_mday, tm.tm_year);
-	return;
-#endif
-}
 
-/*
- * vars for idle task zero'ing out pages
- */
-unsigned long zero_list = 0;	/* head linked list of pre-zero'd pages */
-unsigned long bytecount = 0;	/* pointer into the currently being zero'd page */
-unsigned long zerocount = 0;	/* # currently pre-zero'd pages */
-unsigned long zerototal = 0;	/* # pages zero'd over time -- for ooh's and ahhh's */
-unsigned long pageptr = 0;	/* current page being zero'd */
-unsigned long zeropage_hits = 0;/* # zero'd pages request that we've done */
-
-/*
- * Returns a pre-zero'd page from the list otherwise returns
- * NULL.
- */
-unsigned long get_prezerod_page(void)
+asmlinkage int sys_debug(long a, long b, long c, long d, long e, long f,struct pt_regs *regs)
 {
-	unsigned long page;
-	unsigned long s;
-
-	if ( zero_list )
-	{
-		/* atomically remove this page from the list */
-		asm (	"101:lwarx  %1,0,%2\n"  /* reserve zero_list */
-			"    lwz    %0,0(%1)\n" /* get next -- new zero_list */
-			"    stwcx. %0,0,%2\n"  /* update zero_list */
-			"    bne-   101b\n"     /* if lost reservation try again */
-			: "=&r" (zero_list), "=&r" (page)
-			: "r" (&zero_list)
-			: "cc" );
-		/* we can update zerocount after the fact since it is not
-		 * used for anything but control of a loop which doesn't
-		 * matter since it won't effect anything if it zero's one
-		 * less page -- Cort
-		 */
-		atomic_inc((atomic_t *)&zeropage_hits);
-		atomic_dec((atomic_t *)&zerocount);
-		/* zero out the pointer to next in the page */
-		*(unsigned long *)page = 0;
-		return page;
-	}
 	return 0;
 }
 
-/*
- * Experimental stuff to zero out pages in the idle task
- * to speed up get_free_pages() -- Cort
- * Zero's out pages until we need to resched or
- * we've reached the limit of zero'd pages.
- */
-void inline zero_paged(void)
-{
-	extern pte_t *get_pte( struct mm_struct *mm, unsigned long address );
-	unsigned long tmp;
-	pte_t ptep;
-	pgd_t *dir;
-	pmd_t *pmd;
-	pte_t *pte;
-
-	sprintf(current->comm, "zero_paged");
-	printk("Started zero_paged\n");
-	/* want priority over idle task and powerd */
-	current->priority = -98;
-	current->counter = -98;
-	__sti();
-	
-	while ( zerocount < 128 )
-	{
-		/*
-		 * Mark a page as reserved so we can mess with it
-		 * If we're interrupted we keep this page and our place in it
-		 * since we validly hold it and it's reserved for us.
-		 */
-		pageptr = __get_free_pages(GFP_ATOMIC, 0, 0 );
-		if ( !pageptr )
-		{
-			printk("!pageptr in zero_paged\n");
-			goto retry;
-		}
-		
-		if ( need_resched )
-			schedule();
-		
-		/*
-		 * Make the page no cache so we don't blow our cache with 0's
-		 */
-		dir = pgd_offset( init_task.mm, pageptr );
-		if (dir)
-		{
-			pmd = pmd_offset(dir, pageptr & PAGE_MASK);
-			if (pmd && pmd_present(*pmd))
-			{
-				pte = pte_offset(pmd, pageptr & PAGE_MASK);
-				if (pte && pte_present(*pte))
-				{			
-					pte_uncache(*pte);
-					flush_tlb_page(find_vma(init_task.mm,pageptr),pageptr);
-				}
-			}
-		}
-	
-		/*
-		 * Important here to not take time away from real processes.
-		 */
-		for ( bytecount = 0; bytecount < PAGE_SIZE ; bytecount += 4 )
-		{
-			if ( need_resched )
-				schedule();
-			*(unsigned long *)(bytecount + pageptr) = 0;
-		}
-		
-		/*
-		 * If we finished zero-ing out a page add this page to
-		 * the zero_list atomically -- we can't use
-		 * down/up since we can't sleep in idle.
-		 * Disabling interrupts is also a bad idea since we would
-		 * steal time away from real processes.
-		 * We can also have several zero_paged's running
-		 * on different processors so we can't interfere with them.
-		 * So we update the list atomically without locking it.
-		 * -- Cort
-		 */
-		/* turn cache on for this page */
-		pte_cache(*pte);
-		flush_tlb_page(find_vma(init_task.mm,pageptr),pageptr);
-		
-		/* atomically add this page to the list */
-		asm (	"101:lwarx  %0,0,%1\n"  /* reserve zero_list */
-			"    stw    %0,0(%2)\n" /* update *pageptr */
-#ifdef __SMP__
-			"    sync\n"            /* let store settle */
-#endif			
-			"    mr     %0,%2\n"    /* update zero_list in reg */
-			"    stwcx. %2,0,%1\n"  /* update zero_list in mem */
-			"    bne-   101b\n"     /* if lost reservation try again */
-			: "=&r" (zero_list)
-			: "r" (&zero_list), "r" (pageptr)
-			: "cc" );
-		/*
-		 * This variable is used in the above loop and nowhere
-		 * else so the worst that could happen is we would
-		 * zero out one more or one less page than we want
-		 * per processor on the machine.  This is because
-		 * we could add our page to the list but not have
-		 * zerocount updated yet when another processor
-		 * reads it.  -- Cort
-		 */
-		atomic_inc((atomic_t *)&zerocount);
-		atomic_inc((atomic_t *)&zerototal);
-retry:	
-		schedule();
-	}
-}
-
-void powerd(void)
-{
-	unsigned long msr, hid0;
-
-	sprintf(current->comm, "powerd");
-	__sti();
-	while (1)
-	{
-		/* want priority over idle task 'swapper' -- Cort */
-		current->priority = -99;
-		current->counter = -99;
-		asm volatile(
-			/* clear powersaving modes and set nap mode */
-			"mfspr %3,1008 \n\t"
-			"andc  %3,%3,%4 \n\t"
-			"or    %3,%3,%5 \n\t"
-			"mtspr 1008,%3 \n\t"
-			/* enter the mode */
-			"mfmsr %0 \n\t"
-			"oris  %0,%0,%2 \n\t"
-			"sync \n\t"
-			"mtmsr %0 \n\t"
-			"isync \n\t"
-			: "=&r" (msr)
-			: "0" (msr), "i" (MSR_POW>>16),
-			"r" (hid0),
-			"r" (HID0_DOZE|HID0_NAP|HID0_SLEEP),
-			"r" (HID0_NAP));
-		if ( need_resched )
-			schedule();
-		/*
-		 * The ibm carolina spec says that the eagle memory
-		 * controller will detect the need for a snoop
-		 * and wake up the processor so we don't need to
-		 * check for cache operations that need to be
-		 * snooped.  The ppc book says the run signal
-		 * must be asserted while napping for this though.
-		 * -- Cort
-		 */
-	}
-}
-	
-asmlinkage int sys_idle(void)
-{
-	int ret = -EPERM;
-	if (current->pid != 0)
-		goto out;
-
-#ifdef IDLE_ZERO
-	/*
-	 * want one per cpu since it would be nice to have all
-	 * processors who aren't doing anything
-	 * zero-ing pages since this daemon is lock-free
-	 * -- Cort
-	 */
-	kernel_thread(zero_paged, NULL, 0);
-#endif /* IDLE_ZERO */
-
-#ifdef CONFIG_POWERSAVING
-	/* no powersaving modes on 601 - one per processor */
-	if(  (_get_PVR()>>16) != 1 )
-		kernel_thread(powerd, NULL, 0);
-#endif /* CONFIG_POWERSAVING */
-	
-	/* endless loop with no priority at all */
-	current->priority = -100;
-	current->counter = -100;
-	for (;;) 
-	{
-		schedule();
-	}
-	ret = 0;
-out:
-	return ret;
-}
-
 void show_regs(struct pt_regs * regs)
 {
 	int i;
 
-	printk("NIP: %08X XER: %08X LR: %08X REGS: %08X TRAP: %04x\n",
+	printk("NIP: %08lX XER: %08lX LR: %08lX REGS: %p TRAP: %04lx\n",
 	       regs->nip, regs->xer, regs->link, regs,regs->trap);
-	printk("MSR: %08x EE: %01x PR: %01x FP: %01x ME: %01x IR/DR: %01x%01x\n",
+	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);
-	printk("TASK = %x[%d] '%s' mm->pgd %08X ",
+	printk("TASK = %p[%d] '%s' mm->pgd %p ",
 	       current, current->pid, current->comm, current->mm->pgd);
-	printk("Last syscall: %d ", current->tss.last_syscall);
-	printk("\nlast math %08X\n", last_task_used_math);
+	printk("Last syscall: %ld ", current->tss.last_syscall);
+	printk("\nlast math %p\n", last_task_used_math);
 	for (i = 0;  i < 32;  i++)
 	{
 		long r;
@@ -451,9 +190,9 @@ void show_regs(struct pt_regs * regs)
 			printk("GPR%02d: ", i);
 		}
 
-		if ( get_user(r, &(regs->gpr[i])) )
+		if ( __get_user(r, &(regs->gpr[i])) )
 		    goto out;
-		printk("%08X ", r);
+		printk("%08lX ", r);
 		if ((i % 8) == 7)
 		{
 			printk("\n");
@@ -480,14 +219,14 @@ release_thread(struct task_struct *t)
 }
 
 /*
-  * Copy a thread..
-  */
+ * Copy a thread..
+ */
 int
 copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
 	    struct task_struct * p, struct pt_regs * regs)
 {
-	int i;
 	struct pt_regs * childregs;
+
 	/* Copy registers */
 	childregs = ((struct pt_regs *)
 		     ((unsigned long)p + sizeof(union task_union)
@@ -497,13 +236,13 @@ copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
 	if ((childregs->msr & MSR_PR) == 0)
 		childregs->gpr[2] = (unsigned long) p;	/* `current' in new task */
 	childregs->gpr[3] = 0;  /* Result from fork() */
-	p->tss.ksp = (unsigned long)(childregs) - STACK_FRAME_OVERHEAD;
-	p->tss.regs = (struct pt_regs *)(childregs);
-	if (usp >= (unsigned long)regs)
-	{ /* Stack is in kernel space - must adjust */
-		childregs->gpr[1] = (long)(childregs+1);
-	} else
-	{ /* Provided stack is in user space */
+	p->tss.ksp = (unsigned long) childregs - STACK_FRAME_OVERHEAD;
+	p->tss.regs = childregs;
+	if (usp >= (unsigned long) regs) {
+		/* Stack is in kernel space - must adjust */
+		childregs->gpr[1] = (unsigned long)(childregs + 1);
+	} else {
+		/* Provided stack is in user space */
 		childregs->gpr[1] = usp;
 	}
 
@@ -511,24 +250,75 @@ copy_thread(int nr, unsigned long clone_flags, unsigned long usp,
 
 	/*
 	 * copy fpu info - assume lazy fpu switch now always
-	 * this should really be conditional on whether or
-	 * not the process has used the fpu
 	 *  -- Cort
 	 */
 	if ( last_task_used_math == current )
 		giveup_fpu();
-	
+
 	memcpy(&p->tss.fpr, &current->tss.fpr, sizeof(p->tss.fpr));
 	p->tss.fpscr = current->tss.fpscr;
 	childregs->msr &= ~MSR_FP;
-	
+
 	return 0;
 }
 
+/*
+ * XXX ld.so expects the auxiliary table to start on
+ * a 16-byte boundary, so we have to find it and
+ * move it up. :-(
+ */
+static inline void shove_aux_table(unsigned long sp)
+{
+	int argc;
+	char *p;
+	unsigned long e;
+	unsigned long aux_start, offset;
+
+	if (__get_user(argc, (int *)sp))
+		return;
+	sp += sizeof(int) + (argc + 1) * sizeof(char *);
+	/* skip over the environment pointers */
+	do {
+		if (__get_user(p, (char **)sp))
+			return;
+		sp += sizeof(char *);
+	} while (p != NULL);
+	aux_start = sp;
+	/* skip to the end of the auxiliary table */
+	do {
+		if (__get_user(e, (unsigned long *)sp))
+			return;
+		sp += 2 * sizeof(unsigned long);
+	} while (e != AT_NULL);
+	offset = ((aux_start + 15) & ~15) - aux_start;
+	if (offset != 0) {
+		do {
+			sp -= sizeof(unsigned long);
+			if (__get_user(e, (unsigned long *)sp)
+			    || __put_user(e, (unsigned long *)(sp + offset)))
+				return;
+		} while (sp > aux_start);
+	}
+}
+
+/*
+ * Set up a thread for executing a new program
+ */
+void start_thread(struct pt_regs *regs, unsigned long nip, unsigned long sp)
+{
+	set_fs(USER_DS);
+	regs->nip = nip;
+	regs->gpr[1] = sp;
+	regs->msr = MSR_USER;
+	shove_aux_table(sp);
+}
+
+
 asmlinkage int sys_fork(int p1, int p2, int p3, int p4, int p5, int p6,
 			struct pt_regs *regs)
 {
 	int ret;
+
 	lock_kernel();
 	ret = do_fork(SIGCHLD, regs->gpr[1], regs);
 	unlock_kernel();
@@ -541,26 +331,26 @@ asmlinkage int sys_execve(unsigned long a0, unsigned long a1, unsigned long a2,
 {
 	int error;
 	char * filename;
-	filename = (int) getname((char *) a0);
+	
+	filename = getname((char *) a0);
 	error = PTR_ERR(filename);
-	if(IS_ERR(filename))
+	if (IS_ERR(filename))
 		goto out;
 	if ( last_task_used_math == current )
 		last_task_used_math = NULL;
 	error = do_execve(filename, (char **) a1, (char **) a2, regs);
-
 	putname(filename);
-
 out:
 	unlock_kernel();
 	return error;
 }
 
-asmlinkage int sys_clone(int p1, int p2, int p3, int p4, int p5, int p6, struct pt_regs *regs)
+asmlinkage int sys_clone(int p1, int p2, int p3, int p4, int p5, int p6,
+			 struct pt_regs *regs)
 {
 	unsigned long clone_flags = p1;
 	int res;
-	
+
 	lock_kernel();
 	res = do_fork(clone_flags, regs->gpr[1], regs);
 	unlock_kernel();
@@ -571,34 +361,23 @@ void
 print_backtrace(unsigned long *sp)
 {
 	int cnt = 0;
-	int i;
+	unsigned long i;
+
 	printk("Call backtrace: ");
-	while ( !get_user(i, sp) && i)
-	{
-		if ( get_user( i, &sp[1] ) )
-			return;
-		printk("%08X ", i);
-		if ( get_user( (ulong)sp, sp) )
-		    return;
-		if (cnt == 6 ) cnt = 7; /* wraparound early -- Cort */
-		if (++cnt == 8)
-		{
+	while (sp) {
+		if (__get_user( i, &sp[1] ))
+			break;
+		if (cnt++ % 7 == 0)
 			printk("\n");
-		}
+		printk("%08lX ", i);
 		if (cnt > 32) break;
+		if (__get_user(sp, (unsigned long **)sp))
+			break;
 	}
 	printk("\n");
 }
 
-inline void start_thread(struct pt_regs * regs,
-                         unsigned long eip, unsigned long esp)
-{
-	set_fs(USER_DS);
-	regs->nip = eip;
-	regs->gpr[1] = esp;
-	regs->msr = MSR_USER;
-}
-
+#if 0
 /*
  * Low level print for debugging - Cort
  */
@@ -651,3 +430,4 @@ void ll_puts(const char *s)
 	orig_x = x;
 	orig_y = y;
 }
+#endif /* CONFIG_PREP */
diff --git a/arch/ppc/kernel/prom.c b/arch/ppc/kernel/prom.c
new file mode 100644
index 000000000..c7a566a14
--- /dev/null
+++ b/arch/ppc/kernel/prom.c
@@ -0,0 +1,524 @@
+/*
+ * Procedures for interfacing to the Open Firmware PROM on
+ * Power Macintosh computers.
+ *
+ * In particular, we are interested in the device tree
+ * and in using some of its services (exit, write to stdout).
+ *
+ * Paul Mackerras	August 1996.
+ * Copyright (C) 1996 Paul Mackerras.
+ */
+
+#include <stdarg.h>
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/string.h>
+#include <linux/blk.h>
+#include <asm/prom.h>
+#include <asm/page.h>
+#include <asm/processor.h>
+
+#define getpromprop(node, name, buf, len)	\
+	((int)call_prom("getprop", 4, 1, (node), (name), (buf), (len)))
+
+ihandle prom_stdout;
+ihandle prom_chosen;
+
+char command_line[256];
+int screen_initialized = 0;
+
+char prom_display_path[128];
+
+struct prom_args {
+	const char *service;
+	int nargs;
+	int nret;
+	void *args[10];
+} prom_args;
+
+struct pci_address {
+	unsigned a_hi;
+	unsigned a_mid;
+	unsigned a_lo;
+};
+
+struct pci_reg_property {
+	struct pci_address addr;
+	unsigned size_hi;
+	unsigned size_lo;
+};
+
+struct pci_range {
+	struct pci_address addr;
+	unsigned phys;
+	unsigned size_hi;
+	unsigned size_lo;
+};
+
+void (*prom_entry)(void *);
+extern int prom_trashed;
+
+static int prom_callback(struct prom_args *);
+static unsigned long inspect_node(phandle, struct device_node *, unsigned long,
+				  unsigned long, unsigned long);
+static void check_display(void);
+static int prom_next_node(phandle *);
+
+extern int pmac_display_supported(const char *);
+extern void enter_prom(void *);
+
+void
+prom_exit()
+{
+	struct prom_args args;
+
+	args.service = "exit";
+	args.nargs = 0;
+	args.nret = 0;
+	enter_prom(&args);
+	for (;;)			/* should never get here */
+		;
+}
+
+void *
+call_prom(const char *service, int nargs, int nret, ...)
+{
+	va_list list;
+	int i;
+
+	if (prom_trashed)
+		panic("prom called after its memory was reclaimed");
+	prom_args.service = service;
+	prom_args.nargs = nargs;
+	prom_args.nret = nret;
+	va_start(list, nret);
+	for (i = 0; i < nargs; ++i)
+		prom_args.args[i] = va_arg(list, void *);
+	va_end(list);
+	for (i = 0; i < nret; ++i)
+		prom_args.args[i + nargs] = 0;
+	enter_prom(&prom_args);
+	return prom_args.args[nargs];
+}
+
+void
+prom_print(const char *msg)
+{
+	const char *p, *q;
+	const char *crlf = "\r\n";
+
+	if (screen_initialized)
+		return;
+	for (p = msg; *p != 0; p = q) {
+		for (q = p; *q != 0 && *q != '\n'; ++q)
+			;
+		if (q > p)
+			call_prom("write", 3, 1, prom_stdout, p, q - p);
+		if (*q != 0) {
+			++q;
+			call_prom("write", 3, 1, prom_stdout, crlf, 2);
+		}
+	}
+}
+
+/*
+ * We enter here early on, when the Open Firmware prom is still
+ * handling exceptions and the MMU hash table for us.
+ */
+void
+prom_init(char *params, int unused, void (*pp)(void *))
+{
+	/* First get a handle for the stdout device */
+	if ( ! have_of() )
+		return;
+	prom_entry = pp;
+	prom_chosen = call_prom("finddevice", 1, 1, "/chosen");
+	if (prom_chosen == (void *)-1)
+		prom_exit();
+	call_prom("getprop", 4, 1, prom_chosen, "stdout", &prom_stdout,
+		  (void *) sizeof(prom_stdout));
+
+	/*
+	 * If we were booted via quik, params points to the physical address
+	 * of the command-line parameters.
+	 * If we were booted from an xcoff image (i.e. netbooted or
+	 * booted from floppy), we get the command line from the bootargs
+	 * property of the /chosen node.  If an initial ramdisk is present,
+	 * params and unused are used for initrd_start and initrd_size,
+	 * otherwise they contain 0xdeadbeef.  
+	 */
+	command_line[0] = 0;
+	if ((unsigned long) params >= 0x4000
+	    && (unsigned long) params < 0x800000
+	    && unused == 0) {
+		strncpy(command_line, params+KERNELBASE, sizeof(command_line));
+	} else {
+#ifdef CONFIG_BLK_DEV_INITRD
+		if ((unsigned long) params - KERNELBASE < 0x800000
+		    && unused != 0 && unused != 0xdeadbeef) {
+			initrd_start = (unsigned long) params;
+			initrd_end = initrd_start + unused;
+			ROOT_DEV = MKDEV(RAMDISK_MAJOR, 0);
+		}
+#endif
+		call_prom("getprop", 4, 1, prom_chosen, "bootargs",
+			  command_line, sizeof(command_line));
+	}
+	command_line[sizeof(command_line) - 1] = 0;
+
+	check_display();
+}
+
+/*
+ * If we have a display that we don't know how to drive,
+ * we will want to try to execute OF's open method for it
+ * later.  However, OF may fall over if we do that after
+ * we've taken over the MMU and done set_prom_callback.
+ * So we check whether we will need to open the display,
+ * and if so, open it now.
+ */
+static void
+check_display()
+{
+	phandle node;
+	ihandle ih;
+	char type[16], name[64], path[128];
+
+	for (node = 0; prom_next_node(&node); ) {
+		type[0] = 0;
+		getpromprop(node, "device_type", type, sizeof(type));
+		if (strcmp(type, "display") != 0)
+			continue;
+		name[0] = 0;
+		getpromprop(node, "name", name, sizeof(name));
+		if (pmac_display_supported(name))
+			/* we have a supported display */
+			return;
+	}
+	printk(KERN_INFO "No supported display found\n");
+	for (node = 0; prom_next_node(&node); ) {
+		type[0] = 0;
+		getpromprop(node, "device_type", type, sizeof(type));
+		if (strcmp(type, "display") != 0)
+			continue;
+		/* It seems OF doesn't null-terminate the path :-( */
+		memset(path, 0, sizeof(path));
+		if ((int) call_prom("package-to-path", 3, 1,
+				    node, path, sizeof(path) - 1) < 0) {
+			printk(KERN_WARNING "can't get path for display %p\n",
+			       node);
+			continue;
+		}
+		ih = call_prom("open", 1, 1, path);
+		if (ih == 0 || ih == (ihandle) -1) {
+			printk(KERN_WARNING "couldn't open display %s\n",
+			       path);
+			continue;
+		}
+		printk(KERN_INFO "Opened display device %s using "
+		       "Open Firmware\n", path);
+		strcpy(prom_display_path, path);
+		break;
+	}
+}
+
+static int
+prom_next_node(phandle *nodep)
+{
+	phandle node;
+
+	if ((node = *nodep) != 0
+	    && (*nodep = call_prom("child", 1, 1, node)) != 0)
+		return 1;
+	if ((*nodep = call_prom("peer", 1, 1, node)) != 0)
+		return 1;
+	for (;;) {
+		if ((node = call_prom("parent", 1, 1, node)) == 0)
+			return 0;
+		if ((*nodep = call_prom("peer", 1, 1, node)) != 0)
+			return 1;
+	}
+}
+
+/*
+ * Callback routine for the PROM to call us.
+ * No services are implemented yet :-)
+ */
+static int
+prom_callback(struct prom_args *argv)
+{
+	printk("uh oh, prom callback '%s' (%d/%d)\n", argv->service,
+	       argv->nargs, argv->nret);
+	return -1;
+}
+
+/*
+ * Register a callback with the Open Firmware PROM so it can ask
+ * us to map/unmap memory, etc.
+ */
+void
+set_prom_callback()
+{
+	call_prom("set-callback", 1, 1, prom_callback);
+}
+
+void
+abort()
+{
+#ifdef CONFIG_XMON
+	xmon(0);
+#endif
+	prom_exit();
+}
+
+/*
+ * Make a copy of the device tree from the PROM.
+ */
+
+static struct device_node *allnodes;
+static struct device_node **allnextp;
+
+#define ALIGN(x) (((x) + sizeof(unsigned long)-1) & -sizeof(unsigned long))
+
+unsigned long
+copy_device_tree(unsigned long mem_start, unsigned long mem_end)
+{
+	phandle root;
+
+	root = call_prom("peer", 1, 1, (phandle)0);
+	if (root == (phandle)0)
+		panic("couldn't get device tree root\n");
+	allnextp = &allnodes;
+	mem_start = inspect_node(root, 0, 0, mem_start, mem_end);
+	*allnextp = 0;
+	return mem_start;
+}
+
+static unsigned long
+inspect_node(phandle node, struct device_node *dad, unsigned long base_address,
+	     unsigned long mem_start, unsigned long mem_end)
+{
+	struct reg_property *reg, *rp;
+	struct pci_reg_property *pci_addrs;
+	int l, i;
+	phandle child;
+	struct device_node *np;
+	struct property *pp, **prev_propp;
+	char *prev_name;
+
+	np = (struct device_node *) mem_start;
+	mem_start += sizeof(struct device_node);
+	memset(np, 0, sizeof(*np));
+	np->node = node;
+	*allnextp = np;
+	allnextp = &np->allnext;
+	np->parent = dad;
+	if (dad != 0) {
+		/* we temporarily use the `next' field as `last_child'. */
+		if (dad->next == 0)
+			dad->child = np;
+		else
+			dad->next->sibling = np;
+		dad->next = np;
+	}
+
+	/* get and store all properties */
+	prev_propp = &np->properties;
+	prev_name = 0;
+	for (;;) {
+		pp = (struct property *) mem_start;
+		pp->name = (char *) (pp + 1);
+		if ((int) call_prom("nextprop", 3, 1, node, prev_name,
+				    pp->name) <= 0)
+			break;
+		mem_start = ALIGN((unsigned long)pp->name
+				  + strlen(pp->name) + 1);
+		pp->value = (unsigned char *) mem_start;
+		pp->length = (int)
+			call_prom("getprop", 4, 1, node, pp->name, pp->value,
+				  mem_end - mem_start);
+		if (pp->length < 0)
+			panic("hey, where did property %s go?", pp->name);
+		mem_start = ALIGN(mem_start + pp->length);
+		prev_name = pp->name;
+		*prev_propp = pp;
+		prev_propp = &pp->next;
+	}
+	*prev_propp = 0;
+
+	np->name = get_property(np, "name", 0);
+	np->type = get_property(np, "device_type", 0);
+
+	/* get all the device addresses and interrupts */
+	reg = (struct reg_property *) mem_start;
+	pci_addrs = (struct pci_reg_property *)
+		get_property(np, "assigned-addresses", &l);
+	i = 0;
+	if (pci_addrs != 0) {
+		while ((l -= sizeof(struct pci_reg_property)) >= 0) {
+			/* XXX assumes PCI addresses mapped 1-1 to physical */
+			reg[i].address = pci_addrs[i].addr.a_lo;
+			reg[i].size = pci_addrs[i].size_lo;
+			++i;
+		}
+	} else {
+		rp = (struct reg_property *) get_property(np, "reg", &l);
+		if (rp != 0) {
+			while ((l -= sizeof(struct reg_property)) >= 0) {
+				reg[i].address = rp[i].address + base_address;
+				reg[i].size = rp[i].size;
+				++i;
+			}
+		}
+	}
+	if (i > 0) {
+		np->addrs = reg;
+		np->n_addrs = i;
+		mem_start += i * sizeof(struct reg_property);
+	}
+
+	np->intrs = (int *) get_property(np, "AAPL,interrupts", &l);
+	if (np->intrs != 0)
+		np->n_intrs = l / sizeof(int);
+
+	/* get the node's full name */
+	l = (int) call_prom("package-to-path", 3, 1, node,
+			    (char *) mem_start, mem_end - mem_start);
+	if (l >= 0) {
+		np->full_name = (char *) mem_start;
+		np->full_name[l] = 0;
+		mem_start = ALIGN(mem_start + l + 1);
+	}
+
+	if (np->type != 0 && strcmp(np->type, "dbdma") == 0 && np->n_addrs > 0)
+		base_address = np->addrs[0].address;
+
+	child = call_prom("child", 1, 1, node);
+	while (child != (void *)0) {
+		mem_start = inspect_node(child, np, base_address,
+					 mem_start, mem_end);
+		child = call_prom("peer", 1, 1, child);
+	}
+
+	return mem_start;
+}
+
+/*
+ * Construct a return a list of the device_nodes with a given name.
+ */
+struct device_node *
+find_devices(const char *name)
+{
+	struct device_node *head, **prevp, *np;
+
+	prevp = &head;
+	for (np = allnodes; np != 0; np = np->allnext) {
+		if (np->name != 0 && strcasecmp(np->name, name) == 0) {
+			*prevp = np;
+			prevp = &np->next;
+		}
+	}
+	*prevp = 0;
+	return head;
+}
+
+/*
+ * Construct a return a list of the device_nodes with a given type.
+ */
+struct device_node *
+find_type_devices(const char *type)
+{
+	struct device_node *head, **prevp, *np;
+
+	prevp = &head;
+	for (np = allnodes; np != 0; np = np->allnext) {
+		if (np->type != 0 && strcasecmp(np->type, type) == 0) {
+			*prevp = np;
+			prevp = &np->next;
+		}
+	}
+	*prevp = 0;
+	return head;
+}
+
+/*
+ * Find the device_node with a given full_name.
+ */
+struct device_node *
+find_path_device(const char *path)
+{
+	struct device_node *np;
+
+	for (np = allnodes; np != 0; np = np->allnext)
+		if (np->full_name != 0 && strcasecmp(np->full_name, path) == 0)
+			return np;
+	return NULL;
+}
+
+/*
+ * Find a property with a given name for a given node
+ * and return the value.
+ */
+unsigned char *
+get_property(struct device_node *np, const char *name, int *lenp)
+{
+	struct property *pp;
+
+	for (pp = np->properties; pp != 0; pp = pp->next)
+		if (strcmp(pp->name, name) == 0) {
+			if (lenp != 0)
+				*lenp = pp->length;
+			return pp->value;
+		}
+	return 0;
+}
+
+void
+print_properties(struct device_node *np)
+{
+	struct property *pp;
+	char *cp;
+	int i, n;
+
+	for (pp = np->properties; pp != 0; pp = pp->next) {
+		printk(KERN_INFO "%s", pp->name);
+		for (i = strlen(pp->name); i < 16; ++i)
+			printk(" ");
+		cp = (char *) pp->value;
+		for (i = pp->length; i > 0; --i, ++cp)
+			if ((i > 1 && (*cp < 0x20 || *cp > 0x7e))
+			    || (i == 1 && *cp != 0))
+				break;
+		if (i == 0 && pp->length > 1) {
+			/* looks like a string */
+			printk(" %s\n", (char *) pp->value);
+		} else {
+			/* dump it in hex */
+			n = pp->length;
+			if (n > 64)
+				n = 64;
+			if (pp->length % 4 == 0) {
+				unsigned int *p = (unsigned int *) pp->value;
+
+				n /= 4;
+				for (i = 0; i < n; ++i) {
+					if (i != 0 && (i % 4) == 0)
+						printk("\n                ");
+					printk(" %08x", *p++);
+				}
+			} else {
+				unsigned char *bp = pp->value;
+
+				for (i = 0; i < n; ++i) {
+					if (i != 0 && (i % 16) == 0)
+						printk("\n                ");
+					printk(" %02x", *bp++);
+				}
+			}
+			printk("\n");
+			if (pp->length > 64)
+				printk("                 ... (length = %d)\n",
+				       pp->length);
+		}
+	}
+}
diff --git a/arch/ppc/kernel/ptrace.c b/arch/ppc/kernel/ptrace.c
index a5ec62d20..8a5839d1f 100644
--- a/arch/ppc/kernel/ptrace.c
+++ b/arch/ppc/kernel/ptrace.c
@@ -171,8 +171,11 @@ repeat:
 		goto repeat;
 	}
 /* this is a hack for non-kernel-mapped video buffers and similar */
-	if (MAP_NR(page) < max_mapnr)
-		*(unsigned long *) (page + (addr & ~PAGE_MASK)) = data;
+	if (MAP_NR(page) < max_mapnr) {
+		unsigned long phys_addr = page + (addr & ~PAGE_MASK);
+		*(unsigned long *) phys_addr = data;
+		flush_icache_range(phys_addr, phys_addr+4);
+	}
 /* we're bypassing pagetables, so we have to set the dirty bit ourselves */
 /* this should also re-instate whatever read-only mode there was before */
 	set_pte(pgtable, pte_mkdirty(mk_pte(page, vma->vm_page_prot)));
@@ -364,7 +367,7 @@ asmlinkage int sys_ptrace(long request, long pid, long addr, long data)
 		case PTRACE_PEEKUSR: {
 			unsigned long tmp;
 			
-			if ((addr & 3) || addr < 0 || addr >= sizeof(struct user)) {
+			if ((addr & 3) || addr < 0 || addr > (PT_FPSCR << 2)) {
 				ret = -EIO;
 				goto out;
 			}
@@ -378,13 +381,11 @@ asmlinkage int sys_ptrace(long request, long pid, long addr, long data)
 			if (addr < PT_FPR0) {
 				tmp = get_reg(child, addr);
 			}
-#if 1
-			else if (addr >= PT_FPR0 && addr < PT_FPR0 + 64) {
+			else if (addr >= PT_FPR0 && addr <= PT_FPSCR) {
 				if (last_task_used_math == child)
 					giveup_fpu();
 				tmp = ((long *)child->tss.fpr)[addr - PT_FPR0];
 			}
-#endif
 			else
 				ret = -EIO;
 			if (!ret)
@@ -400,7 +401,7 @@ asmlinkage int sys_ptrace(long request, long pid, long addr, long data)
 
 		case PTRACE_POKEUSR: /* write the word at location addr in the USER area */
 			ret = -EIO;
-			if ((addr & 3) || addr < 0 || addr >= sizeof(struct user))
+			if ((addr & 3) || addr < 0 || addr >= ((PT_FPR0 + 64) << 2))
 				goto out;
 
 			addr = addr >> 2; /* temporary hack. */
@@ -508,7 +509,7 @@ asmlinkage void syscall_trace(void)
 		goto out;
 	current->exit_code = SIGTRAP;
 	current->state = TASK_STOPPED;
-	notify_parent(current);
+	notify_parent(current, SIGCHLD);
 	schedule();
 	/*
 	 * this isn't the same as continuing with a signal, but it will do
diff --git a/arch/ppc/kernel/residual.c b/arch/ppc/kernel/residual.c
new file mode 100644
index 000000000..4084f9f62
--- /dev/null
+++ b/arch/ppc/kernel/residual.c
@@ -0,0 +1,144 @@
+/*
+ * $Id: residual.c,v 1.2 1997/08/25 06:54:56 cort Exp $
+ *
+ * Code to deal with the PReP residual data.
+ *
+ * Written by: Cort Dougan (cort@cs.nmt.edu)
+ */
+
+#include <linux/config.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/ptrace.h>
+#include <linux/malloc.h>
+#include <linux/user.h>
+#include <linux/a.out.h>
+#include <linux/tty.h>
+#include <linux/major.h>
+#include <linux/interrupt.h>
+#include <linux/reboot.h>
+#include <linux/init.h>
+#include <linux/blk.h>
+#include <linux/ioport.h>
+#include <linux/pci.h>
+
+#include <asm/mmu.h>
+#include <asm/processor.h>
+#include <asm/residual.h>
+#include <asm/io.h>
+#include <asm/pgtable.h>
+#include <asm/ide.h>
+#include <asm/pnp.h>
+
+
+/*
+ * Spit out some info about residual data
+ */
+void print_residual_device_info(void)
+{
+	int i;
+	union _PnP_TAG_PACKET *pkt;
+	PPC_DEVICE *dev;
+#define did dev->DeviceId
+	
+	/* make sure we have residual data first */
+	if ( res.ResidualLength == 0 )
+		return;
+	
+	printk("Residual: %ld devices\n", res.ActualNumDevices);
+	for ( i = 0;
+	      i < res.ActualNumDevices ;
+	      i++)
+	{
+		dev = &res.Devices[i];
+		/*
+		 * pci devices
+		 */
+		if ( did.BusId & PCIDEVICE )
+		{
+			printk("PCI Device:");
+			/* unknown vendor */
+			if ( !strncmp( "Unknown", pci_strvendor(did.DevId>>16), 7) )
+				printk(" id %08lx types %d/%d", did.DevId,
+				       did.BaseType, did.SubType);
+			/* known vendor */
+			else
+				printk(" %s %s",
+				       pci_strvendor(did.DevId>>16),
+				       pci_strdev(did.DevId>>16,
+						  did.DevId&0xffff)
+					);
+			
+			if ( did.BusId & PNPISADEVICE )
+			{
+				printk(" pnp:");
+				/* get pnp info on the device */
+				pkt = (union _PnP_TAG_PACKET *)
+					&res.DevicePnPHeap[dev->AllocatedOffset];
+				for (; pkt->S1_Pack.Tag != DF_END_TAG;
+				     pkt++ )
+				{
+					if ( (pkt->S1_Pack.Tag == S4_Packet) ||
+					     (pkt->S1_Pack.Tag == S4_Packet_flags) )
+						printk(" irq %02x%02x",
+						       pkt->S4_Pack.IRQMask[0],
+						       pkt->S4_Pack.IRQMask[1]);
+				}
+			}
+			printk("\n");
+			continue;
+		}
+		/*
+		 * isa devices
+		 */
+		if ( did.BusId & ISADEVICE )
+		{
+			printk("ISA Device: basetype: %d subtype: %d",
+			       did.BaseType, did.SubType);
+			printk("\n");
+			continue;
+		}		
+		/*
+		 * eisa devices
+		 */
+		if ( did.BusId & EISADEVICE )
+		{
+			printk("EISA Device: basetype: %d subtype: %d",
+			       did.BaseType, did.SubType);
+			printk("\n");
+			continue;
+		}		
+		/*
+		 * proc bus devices
+		 */
+		if ( did.BusId & PROCESSORDEVICE )
+		{
+			printk("ProcBus Device: basetype: %d subtype: %d",
+			       did.BaseType, did.SubType);
+			printk("\n");
+			continue;
+		}
+		/*
+		 * pcmcia devices
+		 */
+		if ( did.BusId & PCMCIADEVICE )
+		{
+			printk("PCMCIA Device: basetype: %d subtype: %d",
+			       did.BaseType, did.SubType);
+			printk("\n");
+			continue;
+		}		
+		printk("Unknown bus access device: busid %lx\n",
+		       did.BusId);
+	}
+}
+
+
+
+
+
+
diff --git a/arch/ppc/kernel/setup.c b/arch/ppc/kernel/setup.c
index e69de29bb..f3769cad0 100644
--- a/arch/ppc/kernel/setup.c
+++ b/arch/ppc/kernel/setup.c
@@ -0,0 +1,268 @@
+/*
+ * $Id: setup.c,v 1.16 1997/08/27 22:06:54 cort Exp $
+ * Common prep/pmac boot and setup code.
+ */
+
+#include <linux/config.h>
+#include <linux/string.h>
+#include <linux/sched.h>
+#include <linux/init.h>
+#include <linux/reboot.h>
+#include <linux/openpic.h>
+
+#include <asm/cuda.h>
+#include <asm/residual.h>
+#include <asm/io.h>
+#include <asm/ide.h>
+
+char saved_command_line[256];
+unsigned char aux_device_present;
+
+/* copy of the residual data */
+RESIDUAL res;
+int _machine;
+
+/*
+ * Perhaps we can put the pmac screen_info[] here
+ * on pmac as well so we don't need the ifdef's.
+ * Until we get multiple-console support in here
+ * that is.  -- Cort
+ */ 
+#if defined(CONFIG_CHRP) || defined(CONFIG_PREP )
+struct screen_info screen_info = {
+	0, 25,			/* orig-x, orig-y */
+	{ 0, 0 },		/* unused */
+	0,			/* orig-video-page */
+	0,			/* orig-video-mode */
+	80,			/* orig-video-cols */
+	0,0,0,			/* ega_ax, ega_bx, ega_cx */
+	25,			/* orig-video-lines */
+	1,			/* orig-video-isVGA */
+	16			/* orig-video-points */
+};
+
+/*
+ * I really need to add multiple-console support... -- Cort
+ */
+int pmac_display_supported(char *name)
+{
+	return 0;
+}
+int sd_find_target(void *a, int b)
+{
+	return 0;
+}
+void pmac_find_display(void)
+{
+}
+
+#endif
+
+/*
+ * Find out what kind of machine we're on and save any data we need
+ * from the early boot process (devtree is copied on pmac by prom_init() )
+ */
+unsigned long identify_machine(unsigned long r3, unsigned long r4, unsigned long r5,
+			       unsigned long r6, unsigned long r7)
+{
+	extern unsigned long initrd_start, initrd_end;
+	extern char cmd_line[256];
+#ifdef CONFIG_PMAC /* cheat for now - perhaps a check for OF could tell us */
+	_machine = _MACH_Pmac;
+#endif /* CONFIG_PMAC */
+#ifdef CONFIG_PREP
+	/* make a copy of residual data */
+	if ( r3 )
+		memcpy( (void *)&res,(void *)(r3+KERNELBASE), sizeof(RESIDUAL) );
+	if (!strncmp(res.VitalProductData.PrintableModel,"IBM",3))
+		_machine = _MACH_IBM;
+	else
+		_machine = _MACH_Motorola;
+#endif /* CONFIG_PREP */
+#ifdef CONFIG_CHRP 
+	_machine = _MACH_chrp;
+#endif /* CONFIG_CHRP */
+
+	switch (_machine)
+	{
+	case _MACH_Pmac:
+		io_base = 0;
+		pci_dram_offset = 0;
+		break;
+	case _MACH_IBM:
+	case _MACH_Motorola:
+		io_base = 0x80000000;
+		pci_dram_offset = 0x80000000;
+#ifdef CONFIG_BLK_DEV_RAM
+		/* take care of initrd if we have one */
+		if ( r4 )
+		{
+			initrd_start = r4 + KERNELBASE;
+			initrd_end = r5 + KERNELBASE;
+		}
+#endif /* CONFIG_BLK_DEV_RAM */
+		/* take care of cmd line */
+		if ( r6 )
+		{
+			
+			*(char *)(r7+KERNELBASE) = 0;
+			strcpy(cmd_line, (char *)(r6+KERNELBASE));
+		}
+		break;
+	case _MACH_chrp:
+		/* LongTrail */
+		io_base = 0xf8000000;
+		pci_dram_offset = 0;
+		/* take care of initrd if we have one */
+		if ( r4 ) {
+			initrd_start = r4 + KERNELBASE;
+			initrd_end = r5 + KERNELBASE;
+		}
+		/* take care of cmd line */
+		if ( r6 ) {
+			*(char *)(r7+KERNELBASE) = 0;
+			strcpy(cmd_line, (char *)(r6+KERNELBASE));
+		}
+		break;
+	default:
+		printk("Unknown machine type in identify_machine!\n");
+	}
+	return 0;
+}
+
+/* cmd is ignored for now... */
+void machine_restart(char *cmd)
+{
+	struct cuda_request req;
+	unsigned long flags;
+	unsigned long i = 10000;
+
+	switch(_machine)
+	{
+	case _MACH_Pmac:
+		cuda_request(&req, NULL, 2, CUDA_PACKET, CUDA_RESET_SYSTEM);
+		for (;;)
+			cuda_poll();
+		break;
+	case _MACH_IBM:
+	case _MACH_Motorola:
+		_disable_interrupts();
+		
+		/* set exception prefix high - to the prom */
+		save_flags( flags );
+		restore_flags( flags|MSR_IP );
+		
+		/* make sure bit 0 (reset) is a 0 */
+		outb( inb(0x92) & ~1L , 0x92 );
+		/* signal a reset to system control port A - soft reset */
+		outb( inb(0x92) | 1 , 0x92 );
+		
+		while ( i != 0 ) i++;
+		panic("restart failed\n");
+		break;
+
+	case _MACH_chrp:
+		openpic_init_processor(1<<0);
+		break;
+	}
+}
+
+void machine_power_off(void)
+{
+	struct cuda_request req;
+
+	if ( _machine == _MACH_Pmac )
+	{
+		cuda_request(&req, NULL, 2, CUDA_PACKET, CUDA_POWERDOWN);
+		for (;;)
+			cuda_poll();
+	}
+	else /* prep or chrp */
+	{
+		machine_restart(NULL);
+	}
+}
+
+void machine_halt(void)
+{
+	if ( _machine == _MACH_Pmac )
+	{
+#if 0
+		prom_exit();		/* doesn't work because prom is trashed */
+#else
+		machine_power_off();	/* for now */
+#endif
+	}
+	else /* prep or chrp */
+		machine_restart(NULL);
+
+}
+
+void ide_init_hwif_ports (ide_ioreg_t *p, ide_ioreg_t base, int *irq)
+{
+	if ( _machine == _MACH_Pmac )
+		pmac_ide_init_hwif_ports(p,base,irq);
+	else /* prep */
+		prep_ide_init_hwif_ports(p,base,irq);
+	
+}
+
+/*
+ * Will merge more into here later  -- Cort
+ */
+int get_cpuinfo(char *buffer)
+{
+	extern int pmac_get_cpuinfo(char *);
+	extern int chrp_get_cpuinfo(char *);	
+	extern int prep_get_cpuinfo(char *);
+	
+
+	switch (_machine)
+	{
+	case _MACH_Pmac:
+		return pmac_get_cpuinfo(buffer);
+		break;
+	case _MACH_Motorola:
+	case _MACH_IBM:
+		return prep_get_cpuinfo(buffer);
+		break;
+	case _MACH_chrp:
+		return chrp_get_cpuinfo(buffer);
+		break;
+	}
+	printk("Unknown machine %d in get_cpuinfo()\n",_machine);
+	return 0;
+}
+
+
+__initfunc(unsigned long
+bios32_init(unsigned long memory_start, unsigned long memory_end))
+{
+	return memory_start;
+}
+
+__initfunc(void setup_arch(char **cmdline_p,
+	unsigned long * memory_start_p, unsigned long * memory_end_p))
+{
+	extern void pmac_setup_arch(char **, unsigned long *, unsigned long *);
+	extern void chrp_setup_arch(char **, unsigned long *, unsigned long *);
+	extern void prep_setup_arch(char **, unsigned long *, unsigned long *);
+	
+	switch (_machine)
+	{
+	case _MACH_Pmac:
+		pmac_setup_arch(cmdline_p,memory_start_p,memory_end_p);
+		break;
+	case _MACH_Motorola:
+	case _MACH_IBM:
+		prep_setup_arch(cmdline_p,memory_start_p,memory_end_p);
+		break;
+	case _MACH_chrp:
+		return chrp_setup_arch(cmdline_p,memory_start_p,memory_end_p);
+		break;
+	}
+	printk("Unknown machine %d in setup_arch()\n",_machine);
+}
+
+
+
diff --git a/arch/ppc/kernel/signal.c b/arch/ppc/kernel/signal.c
index 2f602059c..3151d266a 100644
--- a/arch/ppc/kernel/signal.c
+++ b/arch/ppc/kernel/signal.c
@@ -24,6 +24,7 @@
 #include <linux/wait.h>
 #include <linux/ptrace.h>
 #include <linux/unistd.h>
+#include <linux/elf.h>
 #include <asm/uaccess.h>
 
 #define _S(nr) (1<<((nr)-1))
@@ -36,6 +37,10 @@
 #define PAUSE_AFTER_SIGNAL
 #undef  PAUSE_AFTER_SIGNAL
 
+#ifndef MIN
+#define MIN(a,b) (((a) < (b)) ? (a) : (b))
+#endif
+
 asmlinkage int sys_waitpid(pid_t pid,unsigned long * stat_addr, int options);
 
 /*
@@ -71,32 +76,52 @@ printk("Task: %x[%d] - SIGSUSPEND at %x, Mask: %x\n", current, current->pid, reg
 	}
 }
 
+/* 
+ * These are the flags in the MSR that the user is allowed to change
+ * by modifying the saved value of the MSR on the stack.  SE and BE
+ * should not be in this list since gdb may want to change these.  I.e,
+ * you should be able to step out of a signal handler to see what
+ * instruction executes next after the signal handler completes.
+ * Alternately, if you stepped into a signal handler, you should be
+ * able to continue 'til the next breakpoint from within the signal
+ * handler, even if the handler returns.
+ */
+#define MSR_USERCHANGE	(MSR_FE0 | MSR_FE1)
+
 /*
  * This sets regs->esp even though we don't actually use sigstacks yet..
  */
 asmlinkage int sys_sigreturn(struct pt_regs *regs)
 {
-	struct sigcontext_struct *sc;
-	struct pt_regs *int_regs;
-	int signo, ret;
+	struct sigcontext_struct *sc, sigctx;
+	int ret;
+	elf_gregset_t saved_regs;  /* an array of ELF_NGREG unsigned longs */
 
-#if 1 	
-	if (verify_area(VERIFY_READ, (void *) regs->gpr[1], sizeof(sc))
-	    || (regs->gpr[1] >= KERNELBASE))
+	sc = (struct sigcontext_struct *)(regs->gpr[1] + __SIGNAL_FRAMESIZE);
+	if (copy_from_user(&sigctx, sc, sizeof(sigctx)))
 		goto badframe;
-#endif	
-	sc = (struct sigcontext_struct *)(regs->gpr[1]+STACK_FRAME_OVERHEAD);
-	get_user(current->blocked, &sc->oldmask);
-	current->blocked &= _BLOCKABLE;
-	get_user(int_regs, &sc->regs);
-	get_user(signo, &sc->signal);
+	current->blocked = sigctx.oldmask & _BLOCKABLE;
 	sc++;			/* Pop signal 'context' */
 #ifdef DEBUG_SIGNALS
-printk("Sig return - Regs: %x, sc: %x, sig: %d\n", int_regs, sc, signo);
+	printk("Sig return - Regs: %p, sc: %p, sig: %d\n", sigctx.regs, sc,
+	       sigctx.signal);
 #endif
-	if (sc == (struct sigcontext_struct *)(int_regs)) {
-		/* Last stacked signal */
-		memcpy(regs, int_regs, sizeof(*regs));
+	if (sc == (struct sigcontext_struct *)(sigctx.regs)) {
+		/* Last stacked signal - restore registers */
+		if (last_task_used_math == current)
+			giveup_fpu();
+		if (copy_from_user(saved_regs, sigctx.regs, sizeof(saved_regs)))
+			goto badframe;
+		saved_regs[PT_MSR] = (regs->msr & ~MSR_USERCHANGE)
+			| (saved_regs[PT_MSR] & MSR_USERCHANGE);
+		memcpy(regs, saved_regs, 
+		       MIN(sizeof(elf_gregset_t),sizeof(struct pt_regs)));
+
+		if (copy_from_user(current->tss.fpr,
+				   (unsigned long *)sigctx.regs + ELF_NGREG,
+				   ELF_NFPREG * sizeof(double)))
+			goto badframe;
+
 		if (regs->trap == 0x0C00 /* System Call! */ &&
 		    ((int)regs->result == -ERESTARTNOHAND ||
 		     (int)regs->result == -ERESTARTSYS ||
@@ -106,15 +131,17 @@ printk("Sig return - Regs: %x, sc: %x, sig: %d\n", int_regs, sc, signo);
 			regs->result = 0;
 		}
 		ret = regs->result;
+
 	} else {
 		/* More signals to go */
-		regs->gpr[1] = (unsigned long)sc - STACK_FRAME_OVERHEAD;
-		get_user(regs->gpr[3], &sc->signal);
-		get_user(int_regs, (struct pt_regs **) &sc->regs);
-		regs->gpr[4] = (unsigned long) int_regs;
-		regs->link = (unsigned long) (int_regs+1);
-		get_user(regs->nip, &sc->handler);
-		ret = regs->gpr[3];
+		regs->gpr[1] = (unsigned long)sc - __SIGNAL_FRAMESIZE;
+		if (copy_from_user(&sigctx, sc, sizeof(sigctx)))
+			goto badframe;
+		regs->gpr[3] = ret = sigctx.signal;
+		regs->gpr[4] = (unsigned long) sigctx.regs;
+		regs->link = regs->gpr[4] + ELF_NGREG * sizeof(unsigned long)
+			+ ELF_NFPREG * sizeof(double);
+		regs->nip = sigctx.handler;
 	}
 	return ret;
 
@@ -142,6 +169,7 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs * regs)
 	unsigned long *frame = NULL;
 	unsigned long *trampoline;
 	unsigned long *regs_ptr;
+	double *fpregs_ptr;
 	unsigned long nip = 0;
 	unsigned long signr;
 	struct sigcontext_struct *sc;
@@ -164,7 +192,7 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs * regs)
 		if ((current->flags & PF_PTRACED) && signr != SIGKILL) {
 			current->exit_code = signr;
 			current->state = TASK_STOPPED;
-			notify_parent(current);
+			notify_parent(current, SIGCHLD);
 			schedule();
 			if (!(signr = current->exit_code))
 				continue;
@@ -204,7 +232,7 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs * regs)
 				current->exit_code = signr;
 				if (!(current->p_pptr->sig->action[SIGCHLD-1].sa_flags &
 						SA_NOCLDSTOP))
-					notify_parent(current);
+					notify_parent(current, SIGCHLD);
 				schedule();
 				continue;
 
@@ -257,18 +285,33 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs * regs)
 	nip = regs->nip;
 	frame = (unsigned long *) regs->gpr[1];
 
-	/* Build trampoline code on stack */
-	frame -= 2;
+	/*
+	 * Build trampoline code on stack, and save gp and fp regs.
+	 * The 56 word hole is because programs using the rs6000/xcoff
+	 * style calling sequence can save up to 19 gp regs and 18 fp regs
+	 * on the stack before decrementing sp.
+	 */
+	frame -= 2 + 56;
 	trampoline = frame;
-	/* verify stack is valid for writing regs struct */
-	if (verify_area(VERIFY_WRITE,(void *)frame, sizeof(long)*2+sizeof(*regs))
-	    || ((unsigned long) frame >= KERNELBASE ))
-		goto badframe;
-	put_user(0x38007777UL, trampoline);      /* li r0,0x7777 */
-	put_user(0x44000002UL, trampoline+1);    /* sc           */
-	frame -= sizeof(*regs) / sizeof(long);
+	frame -= ELF_NFPREG * sizeof(double) / sizeof(unsigned long);
+	fpregs_ptr = (double *) frame;
+	frame -= ELF_NGREG;
 	regs_ptr = frame;
-	copy_to_user(regs_ptr, regs, sizeof(*regs));
+	/* verify stack is valid for writing to */
+	if (verify_area(VERIFY_WRITE, frame,
+			(ELF_NGREG + 2) * sizeof(long)
+			+ ELF_NFPREG * sizeof(double)))
+		goto badframe;
+	if (last_task_used_math == current)
+		giveup_fpu();
+	if (__copy_to_user(regs_ptr, regs, 
+	                   MIN(sizeof(elf_gregset_t),sizeof(struct pt_regs)))
+	    || __copy_to_user(fpregs_ptr, current->tss.fpr,
+			      ELF_NFPREG * sizeof(double))
+	    || __put_user(0x38007777UL, trampoline)	/* li r0,0x7777 */
+	    || __put_user(0x44000002UL, trampoline+1))	/* sc           */
+		goto badframe;
+
 	signr = 1;
 	sa = current->sig->action;
 
@@ -279,24 +322,29 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs * regs)
 			continue;
 
 		frame -= sizeof(struct sigcontext_struct) / sizeof(long);
-		if (verify_area(VERIFY_WRITE,(void *)frame,
-				sizeof(struct sigcontext_struct)/sizeof(long)))
+		if (verify_area(VERIFY_WRITE, frame,
+				sizeof(struct sigcontext_struct)))
 			goto badframe;
 		sc = (struct sigcontext_struct *)frame;
 		nip = (unsigned long) sa->sa_handler;
 		if (sa->sa_flags & SA_ONESHOT)
 			sa->sa_handler = NULL;
-		put_user(nip, &sc->handler);
-		put_user(oldmask, &sc->oldmask); /* was current->blocked */
-		put_user(regs_ptr, &sc->regs);
-		put_user(signr, &sc->signal);
+		if (__put_user(nip, &sc->handler)
+		    || __put_user(oldmask, &sc->oldmask)
+		    || __put_user(regs_ptr, &sc->regs)
+		    || __put_user(signr, &sc->signal))
+			goto badframe;
 		current->blocked |= sa->sa_mask;
 		regs->gpr[3] = signr;
-		regs->gpr[4] = (unsigned long)regs_ptr;
+		regs->gpr[4] = (unsigned long) regs_ptr;
 	}
+
+	frame -= __SIGNAL_FRAMESIZE / sizeof(unsigned long);
+	if (put_user(regs->gpr[1], frame))
+		goto badframe;
 	regs->link = (unsigned long)trampoline;
 	regs->nip = nip;
-	regs->gpr[1] = (unsigned long)sc - STACK_FRAME_OVERHEAD;
+	regs->gpr[1] = (unsigned long) frame;
 
 	/* The DATA cache must be flushed here to insure coherency */
 	/* between the DATA & INSTRUCTION caches.  Since we just */
@@ -305,8 +353,8 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs * regs)
 	/* cache for new data, we have to force the data to go on to */
 	/* memory and flush the instruction cache to force it to look */
 	/* there.  The following function performs this magic */
-	store_cache_range((unsigned long) trampoline,
-			  (unsigned long) (trampoline + 2));
+	flush_icache_range((unsigned long) trampoline,
+			   (unsigned long) (trampoline + 2));
 	return 1;
 
 badframe:
diff --git a/arch/ppc/kernel/syscalls.c b/arch/ppc/kernel/syscalls.c
index bfdccdee6..b7bac7a6a 100644
--- a/arch/ppc/kernel/syscalls.c
+++ b/arch/ppc/kernel/syscalls.c
@@ -1,15 +1,25 @@
 /*
  * linux/arch/ppc/kernel/sys_ppc.c
  *
+ *  PowerPC version 
+ *    Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
+ *
+ * Derived from "arch/i386/kernel/sys_i386.c"
  * Adapted from the i386 version by Gary Thomas
  * Modified by Cort Dougan (cort@cs.nmt.edu)
+ * and Paul Mackerras (paulus@cs.anu.edu.au).
  *
  * This file contains various random system calls that
  * have a non-standard calling sequence on the Linux/PPC
  * platform.
+ *
+ *  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.
+ *
  */
 
-#include <linux/config.h>
 #include <linux/errno.h>
 #include <linux/sched.h>
 #include <linux/mm.h>
@@ -20,11 +30,11 @@
 #include <linux/shm.h>
 #include <linux/stat.h>
 #include <linux/mman.h>
+#include <linux/sys.h>
 #include <linux/ipc.h>
 #include <asm/uaccess.h>
 #include <asm/ipc.h>
 
-
 void
 check_bugs(void)
 {
@@ -32,32 +42,32 @@ check_bugs(void)
 
 asmlinkage int sys_ioperm(unsigned long from, unsigned long num, int on)
 {
-	printk("sys_ioperm()\n");
+	printk(KERN_ERR "sys_ioperm()\n");
 	return -EIO;
 }
 
 int sys_iopl(int a1, int a2, int a3, int a4)
 {
 	lock_kernel();
-	printk( "sys_iopl(%x, %x, %x, %x)!\n", a1, a2, a3, a4);
+	printk(KERN_ERR "sys_iopl(%x, %x, %x, %x)!\n", a1, a2, a3, a4);
 	unlock_kernel();
-	return (ENOSYS);
+	return (-ENOSYS);
 }
 
 int sys_vm86(int a1, int a2, int a3, int a4)
 {
 	lock_kernel();
-	printk( "sys_vm86(%x, %x, %x, %x)!\n", a1, a2, a3, a4);
+	printk(KERN_ERR "sys_vm86(%x, %x, %x, %x)!\n", a1, a2, a3, a4);
 	unlock_kernel();
-	return (ENOSYS);
+	return (-ENOSYS);
 }
 
 int sys_modify_ldt(int a1, int a2, int a3, int a4)
 {
 	lock_kernel();
-	printk( "sys_modify_ldt(%x, %x, %x, %x)!\n", a1, a2, a3, a4);
+	printk(KERN_ERR "sys_modify_ldt(%x, %x, %x, %x)!\n", a1, a2, a3, a4);
 	unlock_kernel();
-	return (ENOSYS);
+	return (-ENOSYS);
 }
 
 /*
@@ -65,7 +75,8 @@ int sys_modify_ldt(int a1, int a2, int a3, int a4)
  *
  * This is really horribly ugly.
  */
-asmlinkage int sys_ipc (uint call, int first, int second, int third, void *ptr, long fifth)
+asmlinkage int 
+sys_ipc (uint call, int first, int second, int third, void *ptr, long fifth)
 {
 	int version, ret;
 
@@ -73,138 +84,113 @@ asmlinkage int sys_ipc (uint call, int first, int second, int third, void *ptr,
 	version = call >> 16; /* hack for backward compatibility */
 	call &= 0xffff;
 
-	if (call <= SEMCTL)
-		switch (call) {
-		case SEMOP:
-			ret = sys_semop (first, (struct sembuf *)ptr, second);
-			goto out;
-		case SEMGET:
-			ret = sys_semget (first, second, third);
-			goto out;
-		case SEMCTL: {
-			union semun fourth;
-			ret = -EINVAL;
-			if (!ptr)
-				goto out;
-			ret = -EFAULT;
-			if (get_user(fourth.__pad, (void **) ptr))
-				goto out;	
-			ret = sys_semctl (first, second, third, fourth);
-			goto out;
-		}
-		default:
-			ret = -EINVAL;
-			goto out;
+	ret = -EINVAL;
+	switch (call) {
+	case SEMOP:
+		ret = sys_semop (first, (struct sembuf *)ptr, second);
+		break;
+	case SEMGET:
+		ret = sys_semget (first, second, third);
+		break;
+	case SEMCTL: {
+		union semun fourth;
+
+		if (!ptr)
+			break;
+		if ((ret = verify_area (VERIFY_READ, ptr, sizeof(long)))
+		    || (ret = get_user(fourth.__pad, (void **)ptr)))
+			break;
+		ret = sys_semctl (first, second, third, fourth);
+		break;
 		}
-	if (call <= MSGCTL) 
-		switch (call) {
-		case MSGSND:
-			ret = sys_msgsnd (first, (struct msgbuf *) ptr, 
-					  second, third);
-			goto out;
-		case MSGRCV:
-			switch (version) {
-			case 0: {
-				struct ipc_kludge tmp;
-				ret = -EINVAL;
-				if (!ptr)
-					goto out;
-				ret = -EFAULT;
-				if (copy_from_user(&tmp,(struct ipc_kludge *) ptr, 
-						   sizeof (tmp)))
-					goto out; 	
-				ret = sys_msgrcv (first, tmp.msgp, second, tmp.msgtyp, third);
-				goto out;
-			}
-			case 1: default:
-				ret = sys_msgrcv (first, (struct msgbuf *) ptr, second, fifth, third);
-				goto out;
+	case MSGSND:
+		ret = sys_msgsnd (first, (struct msgbuf *) ptr, second, third);
+		break;
+	case MSGRCV:
+		switch (version) {
+		case 0: {
+			struct ipc_kludge tmp;
+
+			if (!ptr)
+				break;
+			if ((ret = verify_area (VERIFY_READ, ptr, sizeof(tmp)))
+			    || (ret = copy_from_user(&tmp,
+						(struct ipc_kludge *) ptr,
+						sizeof (tmp))))
+				break;
+			ret = sys_msgrcv (first, tmp.msgp, second, tmp.msgtyp,
+					  third);
+			break;
 			}
-		case MSGGET:
-			ret = sys_msgget ((key_t) first, second);
-			goto out;
-		case MSGCTL:
-			ret = sys_msgctl (first, second, (struct msqid_ds *) ptr);
-			goto out;
 		default:
-			ret = -EINVAL;
-			goto out;
+			ret = sys_msgrcv (first, (struct msgbuf *) ptr,
+					  second, fifth, third);
+			break;
 		}
-	if (call <= SHMCTL) 
-		switch (call) {
-		case SHMAT:
-			switch (version) {
-			case 0: default: {
-				ulong raddr;
-				ret = sys_shmat (first, (char *) ptr, second, &raddr);
-				if (ret)
-					goto out;
-				ret = put_user (raddr, (ulong *) third);
-				goto out;
+		break;
+	case MSGGET:
+		ret = sys_msgget ((key_t) first, second);
+		break;
+	case MSGCTL:
+		ret = sys_msgctl (first, second, (struct msqid_ds *) ptr);
+		break;
+	case SHMAT:
+		switch (version) {
+		default: {
+			ulong raddr;
+
+			if ((ret = verify_area(VERIFY_WRITE, (ulong*) third,
+					       sizeof(ulong))))
+				break;
+			ret = sys_shmat (first, (char *) ptr, second, &raddr);
+			if (ret)
+				break;
+			ret = put_user (raddr, (ulong *) third);
+			break;
 			}
-			case 1:	/* iBCS2 emulator entry point */
-				ret = -EINVAL;
-				if (get_fs() != get_ds())
-					goto out;
-				ret = sys_shmat (first, (char *) ptr, second, (ulong *) third);
-				goto out;
-			}
-		case SHMDT: 
-			ret = sys_shmdt ((char *)ptr);
-			goto out;
-		case SHMGET:
-			ret = sys_shmget (first, second, third);
-			goto out;
-		case SHMCTL:
-			ret = sys_shmctl (first, second, (struct shmid_ds *) ptr);
-			goto out;
-		default:
-			ret = -EINVAL;
-			goto out;
+		case 1:	/* iBCS2 emulator entry point */
+			if (get_fs() != get_ds())
+				break;
+			ret = sys_shmat (first, (char *) ptr, second,
+					 (ulong *) third);
+			break;
 		}
-	else
-		ret = -EINVAL;
-out:
-	unlock_kernel();
-	return ret;
-}
-
+		break;
+	case SHMDT: 
+		ret = sys_shmdt ((char *)ptr);
+		break;
+	case SHMGET:
+		ret = sys_shmget (first, second, third);
+		break;
+	case SHMCTL:
+		ret = sys_shmctl (first, second, (struct shmid_ds *) ptr);
+		break;
+	}
 
-#ifndef CONFIG_MODULES
-void
-scsi_register_module(void)
-{
-	lock_kernel();
-	panic("scsi_register_module");
-	unlock_kernel();
-}
-
-void
-scsi_unregister_module(void)
-{
-	lock_kernel();
-	panic("scsi_unregister_module");
 	unlock_kernel();
+	return ret;
 }
-#endif
 
 /*
  * sys_pipe() is the normal C calling standard for creating
  * a pipe. It's not the way unix traditionally does this, though.
  */
-asmlinkage int sys_pipe(unsigned long * fildes)
+asmlinkage int sys_pipe(int *fildes)
 {
 	int fd[2];
 	int error;
 
-	error = verify_area(VERIFY_WRITE,fildes,8);
+	error = verify_area(VERIFY_WRITE, fildes, 8);
 	if (error)
 		return error;
+	lock_kernel();
 	error = do_pipe(fd);
+	unlock_kernel();
 	if (error)
 		return error;
-	put_user(fd[0],0+fildes);
-	put_user(fd[1],1+fildes);
+	if (__put_user(fd[0],0+fildes)
+	    || __put_user(fd[1],1+fildes))
+		return -EFAULT;	/* should we close the fds? */
 	return 0;
 }
 
@@ -213,13 +199,18 @@ asmlinkage unsigned long sys_mmap(unsigned long addr, size_t len,
 				  unsigned long fd, off_t offset)
 {
 	struct file * file = NULL;
+	int ret = -EBADF;
+
+	lock_kernel();
 	if (!(flags & MAP_ANONYMOUS)) {
 		if (fd >= NR_OPEN || !(file = current->files->fd[fd]))
-			return -EBADF;
+			goto out;
 	}
 	flags &= ~(MAP_EXECUTABLE | MAP_DENYWRITE);
-
-	return do_mmap(file, addr, len, prot, flags, offset);
+	ret = do_mmap(file, addr, len, prot, flags, offset);
+out:
+	unlock_kernel();
+	return ret;
 }
 
 extern asmlinkage int sys_select(int, fd_set *, fd_set *, fd_set *, struct timeval *);
@@ -233,15 +224,15 @@ extern asmlinkage int sys_select(int, fd_set *, fd_set *, fd_set *, struct timev
 asmlinkage int 
 ppc_select(int n, fd_set *inp, fd_set *outp, fd_set *exp, struct timeval *tvp)
 {
-	int err;
 	if ( (unsigned long)n >= 4096 )
 	{
 		unsigned long *buffer = (unsigned long *)n;
-		if ( get_user(n, buffer) ||
-		     get_user(inp,buffer+1) ||
-		     get_user(outp,buffer+2) ||
-		     get_user(exp,buffer+3) ||
-		     get_user(tvp,buffer+4) )
+		if (verify_area(VERIFY_READ, buffer, 5*sizeof(unsigned long))
+		    || __get_user(n, buffer)
+		    || __get_user(inp, ((fd_set **)(buffer+1)))
+		    || __get_user(outp, ((fd_set **)(buffer+2)))
+		    || __get_user(exp, ((fd_set **)(buffer+3)))
+		    || __get_user(tvp, ((struct timeval **)(buffer+4))))
 			return -EFAULT;
 	}
 	return sys_select(n, inp, outp, exp, tvp);
diff --git a/arch/ppc/kernel/time.c b/arch/ppc/kernel/time.c
index e69de29bb..fc172d7a7 100644
--- a/arch/ppc/kernel/time.c
+++ b/arch/ppc/kernel/time.c
@@ -0,0 +1,268 @@
+/*
+ * $Id: time.c,v 1.10 1997/08/27 22:06:56 cort Exp $
+ * Common time routines among all ppc machines.
+ *
+ * Written by Cort Dougan (cort@cs.nmt.edu) to merge
+ * Paul Mackerras' version and mine for PReP and Pmac.
+ */
+
+#include <linux/config.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/param.h>
+#include <linux/string.h>
+#include <linux/mm.h>
+#include <linux/interrupt.h>
+#include <linux/timex.h>
+#include <linux/kernel_stat.h>
+#include <linux/mc146818rtc.h>
+#include <linux/time.h>
+
+#include <asm/segment.h>
+#include <asm/io.h>
+#include <asm/processor.h>
+#include <asm/nvram.h>
+
+#include "time.h"
+
+/* this is set to the appropriate pmac/prep/chrp func in init_IRQ() */
+int (*set_rtc_time)(unsigned long);
+
+/* keep track of when we need to update the rtc */
+unsigned long last_rtc_update = 0;
+
+/* The decrementer counts down by 128 every 128ns on a 601. */
+#define DECREMENTER_COUNT_601	(1000000000 / HZ)
+#define COUNT_PERIOD_NUM_601	1
+#define COUNT_PERIOD_DEN_601	1000
+
+unsigned decrementer_count;	/* count value for 1e6/HZ microseconds */
+unsigned count_period_num;	/* 1 decrementer count equals */
+unsigned count_period_den;	/* count_period_num / count_period_den us */
+
+/* Accessor functions for the decrementer register. */
+inline unsigned long
+get_dec(void)
+{
+	int ret;
+
+	asm volatile("mfspr %0,22" : "=r" (ret) :);
+	return ret;
+}
+
+inline void
+set_dec(int val)
+{
+	asm volatile("mtspr 22,%0" : : "r" (val));
+}
+
+/*
+ * timer_interrupt - gets called when the decrementer overflows,
+ * with interrupts disabled.
+ * We set it up to overflow again in 1/HZ seconds.
+ */
+void timer_interrupt(struct pt_regs * regs)
+{
+	int dval, d;
+	while ((dval = get_dec()) < 0) {
+		/*
+		 * Wait for the decrementer to change, then jump
+		 * in and add decrementer_count to its value
+		 * (quickly, before it changes again!)
+		 */
+		while ((d = get_dec()) == dval)
+			;
+		set_dec(d + decrementer_count);
+		do_timer(regs);
+		/*
+		 * update the rtc when needed
+		 */
+		if ( xtime.tv_sec > last_rtc_update + 660 )
+			if (set_rtc_time(xtime.tv_sec) == 0)
+				last_rtc_update = xtime.tv_sec;
+			else
+				last_rtc_update = xtime.tv_sec - 600; /* do it again in 60 s */
+	}
+}
+
+/*
+ * This version of gettimeofday has microsecond resolution.
+ */
+void do_gettimeofday(struct timeval *tv)
+{
+	unsigned long flags;
+
+	save_flags(flags);
+	cli();
+	*tv = xtime;
+	tv->tv_usec += (decrementer_count - get_dec())
+	    * count_period_num / count_period_den;
+	if (tv->tv_usec >= 1000000) {
+		tv->tv_usec -= 1000000;
+		tv->tv_sec++;
+	}
+	restore_flags(flags);
+}
+
+void do_settimeofday(struct timeval *tv)
+{
+	unsigned long flags;
+	int frac_tick;
+	
+	last_rtc_update = 0; /* so the rtc gets updated soon */
+	
+	frac_tick = tv->tv_usec % (1000000 / HZ);
+	save_flags(flags);
+	cli();
+	xtime.tv_sec = tv->tv_sec;
+	xtime.tv_usec = tv->tv_usec - frac_tick;
+	set_dec(frac_tick * count_period_den / count_period_num);
+	restore_flags(flags);
+}
+
+
+void
+time_init(void)
+{
+	/* pmac hasn't yet called via_cuda_init() */
+	if ( _machine != _MACH_Pmac )
+	{
+
+		if ( _machine == _MACH_chrp )
+			xtime.tv_sec = chrp_get_rtc_time();
+		else /* assume prep */
+			xtime.tv_sec = prep_get_rtc_time();
+		xtime.tv_usec = 0;
+		/*
+		 * mark the rtc/on-chip timer as in sync
+		 * so we don't update right away
+		 */
+		last_rtc_update = xtime.tv_sec;
+	}
+
+	if ((_get_PVR() >> 16) == 1) {
+		/* 601 processor: dec counts down by 128 every 128ns */
+		decrementer_count = DECREMENTER_COUNT_601;
+		count_period_num = COUNT_PERIOD_NUM_601;
+		count_period_den = COUNT_PERIOD_DEN_601;
+	}
+
+	switch (_machine)
+	{
+	case _MACH_Pmac:
+		pmac_calibrate_decr();
+		set_rtc_time = pmac_set_rtc_time;
+		break;
+	case _MACH_IBM:
+	case _MACH_Motorola:
+		prep_calibrate_decr();
+		set_rtc_time = prep_set_rtc_time;
+		break;
+	case _MACH_chrp:
+		chrp_calibrate_decr();
+		set_rtc_time = chrp_set_rtc_time;
+		break;
+	}
+	set_dec(decrementer_count);
+}
+
+/*
+ * Uses the on-board timer to calibrate the on-chip decrementer register
+ * for prep systems.  On the pmac the OF tells us what the frequency is
+ * but on prep we have to figure it out.
+ * -- Cort
+ */
+int calibrate_done = 0;
+volatile int *done_ptr = &calibrate_done;
+void prep_calibrate_decr(void)
+{
+	unsigned long flags;
+	
+	save_flags(flags);
+	
+#define TIMER0_COUNT 0x40
+#define TIMER_CONTROL 0x43
+	/* set timer to periodic mode */
+	outb_p(0x34,TIMER_CONTROL);/* binary, mode 2, LSB/MSB, ch 0 */
+	/* set the clock to ~100 Hz */
+	outb_p(LATCH & 0xff , TIMER0_COUNT);	/* LSB */
+	outb(LATCH >> 8 , TIMER0_COUNT);	/* MSB */
+	
+	if (request_irq(0, prep_calibrate_decr_handler, 0, "timer", NULL) != 0)
+		panic("Could not allocate timer IRQ!");
+	__sti();
+	while ( ! *done_ptr ) /* nothing */; /* wait for calibrate */
+        restore_flags(flags);
+	free_irq( 0, NULL);
+}
+
+void prep_calibrate_decr_handler(int irq, void *dev, struct pt_regs * regs)
+{
+	int freq, divisor;
+	static unsigned long t1 = 0, t2 = 0;
+
+	if ( !t1 )
+		t1 = get_dec();
+	else if (!t2)
+	{
+		t2 = get_dec();
+		t2 = t1-t2;  /* decr's in 1/HZ */
+		t2 = t2*HZ;  /* # decrs in 1s - thus in Hz */
+		freq = t2 * 60;	/* try to make freq/1e6 an integer */
+		divisor = 60;
+		printk("time_init: decrementer frequency = %d/%d (%luMHz)\n",
+		       freq, divisor,t2>>20);
+		decrementer_count = freq / HZ / divisor;
+		count_period_num = divisor;
+		count_period_den = freq / 1000000;
+		*done_ptr = 1;
+	}
+}
+
+void chrp_calibrate_decr(void)
+{
+	int freq, fp, divisor;
+
+	fp = 16666000;		/* hardcoded for now */
+	freq = fp*60;	/* try to make freq/1e6 an integer */
+        divisor = 60;
+        printk("time_init: decrementer frequency = %d/%d\n", freq, divisor);
+        decrementer_count = freq / HZ / divisor;
+        count_period_num = divisor;
+        count_period_den = freq / 1000000;
+}
+
+
+/* Converts Gregorian date to seconds since 1970-01-01 00:00:00.
+ * Assumes input in normal date format, i.e. 1980-12-31 23:59:59
+ * => year=1980, mon=12, day=31, hour=23, min=59, sec=59.
+ *
+ * [For the Julian calendar (which was used in Russia before 1917,
+ * Britain & colonies before 1752, anywhere else before 1582,
+ * and is still in use by some communities) leave out the
+ * -year/100+year/400 terms, and add 10.]
+ *
+ * This algorithm was first published by Gauss (I think).
+ *
+ * WARNING: this function will overflow on 2106-02-07 06:28:16 on
+ * machines were long is 32-bit! (However, as time_t is signed, we
+ * will already get problems at other places on 2038-01-19 03:14:08)
+ */
+inline unsigned long mktime(unsigned int year, unsigned int mon,
+				   unsigned int day, unsigned int hour,
+				   unsigned int min, unsigned int sec)
+{
+	
+	if (0 >= (int) (mon -= 2)) {	/* 1..12 -> 11,12,1..10 */
+		mon += 12;	/* Puts Feb last since it has leap day */
+		year -= 1;
+	}
+	return (((
+		(unsigned long)(year/4 - year/100 + year/400 + 367*mon/12 + day) +
+		year*365 - 719499
+		)*24 + hour /* now have hours */
+		)*60 + min /* now have minutes */
+		)*60 + sec; /* finally seconds */
+}
+
diff --git a/arch/ppc/kernel/time.h b/arch/ppc/kernel/time.h
new file mode 100644
index 000000000..9ce5921dc
--- /dev/null
+++ b/arch/ppc/kernel/time.h
@@ -0,0 +1,74 @@
+/*
+ * $Id: time.h,v 1.5 1997/08/27 22:06:58 cort Exp $
+ * Common time prototypes and such for all ppc machines.
+ *
+ * Written by Cort Dougan (cort@cs.nmt.edu) to merge
+ * Paul Mackerras' version and mine for PReP and Pmac.
+ */
+
+#include <linux/mc146818rtc.h>
+
+/* time.c */
+__inline__ unsigned long get_dec(void);
+__inline__ void set_dec(int val);
+void prep_calibrate_decr_handler(int, void *,struct pt_regs *);
+void prep_calibrate_decr(void);
+void pmac_calibrate_decr(void);
+void chrp_calibrate_decr(void);
+extern unsigned decrementer_count;
+extern unsigned count_period_num;
+extern unsigned count_period_den;
+extern unsigned long mktime(unsigned int, unsigned int,unsigned int,
+				   unsigned int, unsigned int, unsigned int);
+
+/* pmac/prep/chrp_time.c */
+unsigned long prep_get_rtc_time(void);
+unsigned long pmac_get_rtc_time(void);
+unsigned long chrp_get_rtc_time(void);
+int prep_set_rtc_time(unsigned long nowtime);
+int pmac_set_rtc_time(unsigned long nowtime);
+int chrp_set_rtc_time(unsigned long nowtime);
+void pmac_read_rtc_time(void);
+
+#define TICK_SIZE tick
+#define FEBRUARY	2
+#define	STARTOFTIME	1970
+#define SECDAY		86400L
+#define SECYR		(SECDAY * 365)
+#define	leapyear(year)		((year) % 4 == 0)
+#define	days_in_year(a) 	(leapyear(a) ? 366 : 365)
+#define	days_in_month(a) 	(month_days[(a) - 1])
+
+static int      month_days[12] = {
+	31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31
+};
+
+extern void inline to_tm(int tim, struct rtc_time * tm)
+{
+	register int    i;
+	register long   hms, day;
+
+	day = tim / SECDAY;
+	hms = tim % SECDAY;
+
+	/* Hours, minutes, seconds are easy */
+	tm->tm_hour = hms / 3600;
+	tm->tm_min = (hms % 3600) / 60;
+	tm->tm_sec = (hms % 3600) % 60;
+
+	/* Number of years in days */
+	for (i = STARTOFTIME; day >= days_in_year(i); i++)
+		day -= days_in_year(i);
+	tm->tm_year = i;
+
+	/* Number of months in days left */
+	if (leapyear(tm->tm_year))
+		days_in_month(FEBRUARY) = 29;
+	for (i = 1; day >= days_in_month(i); i++)
+		day -= days_in_month(i);
+	days_in_month(FEBRUARY) = 28;
+	tm->tm_mon = i;
+
+	/* Days are what is left over (+1) from all that. */
+	tm->tm_mday = day + 1;
+}
diff --git a/arch/ppc/kernel/traps.c b/arch/ppc/kernel/traps.c
index 84ce1c5ca..edfcb4d63 100644
--- a/arch/ppc/kernel/traps.c
+++ b/arch/ppc/kernel/traps.c
@@ -1,9 +1,15 @@
 /*
  *  linux/arch/ppc/kernel/traps.c
  *
- *  Copyright (C) 1995  Gary Thomas
- *  Adapted for PowerPC by Gary Thomas
+ *  Copyright (C) 1995-1996  Gary Thomas (gdt@linuxppc.org)
+ *
+ *  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.
+ *
  *  Modified by Cort Dougan (cort@cs.nmt.edu)
+ *  and Paul Mackerras (paulus@cs.anu.edu.au)
  */
 
 /*
@@ -24,9 +30,21 @@
 #include <linux/config.h>
 
 #include <asm/pgtable.h>
-#include <asm/segment.h>
+#include <asm/uaccess.h>
 #include <asm/system.h>
 #include <asm/io.h>
+#include <asm/processor.h>
+
+extern int fix_alignment(struct pt_regs *);
+extern void bad_page_fault(struct pt_regs *, unsigned long);
+
+#ifdef CONFIG_XMON
+extern int xmon_bpt(struct pt_regs *regs);
+extern int xmon_sstep(struct pt_regs *regs);
+extern void xmon(struct pt_regs *regs);
+extern int xmon_iabr_match(struct pt_regs *regs);
+extern void (*xmon_fault_handler)(struct pt_regs *regs);
+#endif
 
 /*
  * Trap & Exception support
@@ -43,40 +61,53 @@ _exception(int signr, struct pt_regs *regs)
 	if (!user_mode(regs))
 	{
 		show_regs(regs);
-		print_backtrace(regs->gpr[1]);
-		panic("Exception in kernel pc %x signal %d",regs->nip,signr);
+		print_backtrace((unsigned long *)regs->gpr[1]);
+#ifdef CONFIG_XMON
+		xmon(regs);
+#endif
+		panic("Exception in kernel pc %lx signal %d",regs->nip,signr);
 	}
 	force_sig(signr, current);
 }
 
+void
 MachineCheckException(struct pt_regs *regs)
 {
 	if ( !user_mode(regs) )
 	{
+#ifdef CONFIG_XMON
+		if (xmon_fault_handler) {
+			xmon_fault_handler(regs);
+			return;
+		}
+#endif
 		printk("Machine check in kernel mode.\n");
 		printk("Caused by (from msr): ");
-		printk("regs %08x ",regs);
+		printk("regs %p ",regs);
 		switch( regs->msr & 0x0000F000)
 		{
 		case (1<<12) :
 			printk("Machine check signal - probably due to mm fault\n"
 				"with mmu off\n");
-		break;
+			break;
 		case (1<<13) :
 			printk("Transfer error ack signal\n");
-		break;
+			break;
 		case (1<<14) :
 			printk("Data parity signal\n");
-		break;
+			break;
 		case (1<<15) :
 			printk("Address parity signal\n");
-		break;
+			break;
 		default:
 			printk("Unknown values in msr\n");
 		}
 		show_regs(regs);
-		print_backtrace(regs->gpr[1]);
-		panic("");
+		print_backtrace((unsigned long *)regs->gpr[1]);
+#ifdef CONFIG_XMON
+		xmon(regs);
+#endif
+		panic("machine check");
 	}
 	_exception(SIGSEGV, regs);	
 }
@@ -105,33 +136,71 @@ RunModeException(struct pt_regs *regs)
 	_exception(SIGTRAP, regs);	
 }
 
+void
 ProgramCheckException(struct pt_regs *regs)
 {
-	if (current->flags & PF_PTRACED)
+	if (regs->msr & 0x100000) {
+		/* IEEE FP exception */
+		_exception(SIGFPE, regs);
+	} else if (regs->msr & 0x20000) {
+		/* trap exception */
+#ifdef CONFIG_XMON
+		if (xmon_bpt(regs))
+			return;
+#endif
 		_exception(SIGTRAP, regs);
-	else
+	} else {
 		_exception(SIGILL, regs);
+	}
 }
 
+void
 SingleStepException(struct pt_regs *regs)
 {
 	regs->msr &= ~MSR_SE;  /* Turn off 'trace' bit */
+#ifdef CONFIG_XMON
+	if (xmon_sstep(regs))
+		return;
+#endif
 	_exception(SIGTRAP, regs);	
 }
 
+void
 AlignmentException(struct pt_regs *regs)
 {
+	int fixed;
+
+	if (last_task_used_math == current)
+		giveup_fpu();
+	fixed = fix_alignment(regs);
+	if (fixed == 1) {
+		regs->nip += 4;	/* skip over emulated instruction */
+		return;
+	}
+	if (fixed == -EFAULT) {
+		/* fixed == -EFAULT means the operand address was bad */
+		bad_page_fault(regs, regs->dar);
+		return;
+	}
 	_exception(SIGBUS, regs);	
 }
 
+void
+PromException(struct pt_regs *regs, int trap)
+{
+	regs->trap = trap;
+#ifdef CONFIG_XMON
+	xmon(regs);
+#endif
+	printk("Exception %lx in prom at PC: %lx, SR: %lx\n",
+	       regs->trap, regs->nip, regs->msr);
+	/* probably should turn up the toes here */
+}
+
+void
 trace_syscall(struct pt_regs *regs)
 {
-	static int count;
-	printk("Task: %08X(%d), PC: %08X/%08X, Syscall: %3d, Result: %s%d\n",
+	printk("Task: %p(%d), PC: %08lX/%08lX, Syscall: %3ld, Result: %s%ld\n",
 	       current, current->pid, regs->nip, regs->link, regs->gpr[0],
 	       regs->ccr&0x10000000?"Error=":"", regs->gpr[3]);
-	if (++count == 20)
-	{
-		count = 0;
-	}
 }
diff --git a/arch/ppc/lib/Makefile b/arch/ppc/lib/Makefile
index 68b33f047..8b84ce2ae 100644
--- a/arch/ppc/lib/Makefile
+++ b/arch/ppc/lib/Makefile
@@ -1,38 +1,11 @@
-.c.s:
-	$(CC) $(CFLAGS) -S $<
-.s.o:
-	$(AS) $(ASFLAGS) -o $*.o $<
-.c.o:
-	$(CC) $(CFLAGS) -c $<
-.S.s:
-	$(CPP) $(CFLAGS) -D__ASSEMBLY__ $< -o $*.s
-.S.o:
-	$(CPP) $(CFLAGS) -D__ASSEMBLY__ $< -o $*.s
-	$(AS) $(ASFLAGS) -o $*.o $*.s
-	rm $*.s
-
-HOST_CC = gcc
-
-L_TARGET = lib.o
-L_OBJS  = checksum.o cksum_support.o string.o
-
-${L_TARGET}: $(L_OBJS)
-	$(LD) -r -o ${L_TARGET} $(L_OBJS)
-
-
-fastdep:
-	$(TOPDIR)/scripts/mkdep *.[Sch] > .depend
-
-dep:
-	$(CPP) -M *.S *.c > .depend
+#
+# Makefile for ppc-specific library files..
+#
 
-modules: 
+.S.o:
+	$(CC) -D__ASSEMBLY__ -c $< -o $*.o
 
-dummy:
+O_TARGET = lib.o
+O_OBJS  = checksum.o string.o strcase.o
 
-#
-# include a dependency file if one exists
-#
-ifeq (.depend,$(wildcard .depend))
-include .depend
-endif
+include $(TOPDIR)/Rules.make
diff --git a/arch/ppc/lib/checksum.S b/arch/ppc/lib/checksum.S
new file mode 100644
index 000000000..6c53d50ff
--- /dev/null
+++ b/arch/ppc/lib/checksum.S
@@ -0,0 +1,192 @@
+/*
+ * This file contains assembly-language implementations
+ * of IP-style 1's complement checksum routines.
+ *	
+ *    Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
+ *
+ *  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.
+ *
+ * Severely hacked about by Paul Mackerras (paulus@cs.anu.edu.au).
+ */
+
+#include <linux/sys.h>
+#include <asm/errno.h>
+#include "../kernel/ppc_asm.tmpl"
+
+_TEXT()
+
+/*
+ * ip_fast_csum(buf, len) -- Optimized for IP header
+ * len is in words and is always >= 5.
+ */
+_GLOBAL(ip_fast_csum)
+	lwz	r0,0(r3)
+	lwzu	r5,4(r3)
+	addi	r4,r4,-2
+	addc	r0,r0,r5
+	mtctr	r4
+1:	lwzu	r4,4(r3)
+	adde	r0,r0,r4
+	bdnz	1b
+	addze	r0,r0		/* add in final carry */
+	rlwinm	r3,r0,16,0,31	/* fold two halves together */
+	add	r3,r0,r3
+	not	r3,r3
+	srwi	r3,r3,16
+	blr
+
+/*
+ * Compute checksum of TCP or UDP pseudo-header:
+ *   csum_tcpudp_magic(saddr, daddr, len, proto, sum)
+ */	
+_GLOBAL(csum_tcpudp_magic)
+	rlwimi	r5,r6,16,0,15	/* put proto in upper half of len */
+	addc	r0,r3,r4	/* add 4 32-bit words together */
+	adde	r0,r0,r5
+	adde	r0,r0,r7
+	addze	r0,r0		/* add in final carry */
+	rlwinm	r3,r0,16,0,31	/* fold two halves together */
+	add	r3,r0,r3
+	not	r3,r3
+	srwi	r3,r3,16
+	blr
+
+/*
+ * computes the checksum of a memory block at buff, length len,
+ * and adds in "sum" (32-bit)
+ *
+ * csum_partial(buff, len, sum)
+ */
+_GLOBAL(csum_partial)
+	addic	r0,r5,0
+	subi	r3,r3,4
+	srwi.	r6,r4,2
+	beq	3f		/* if we're doing < 4 bytes */
+	andi.	r5,r3,2		/* Align buffer to longword boundary */
+	beq+	1f
+	lhz	r5,4(r3)	/* do 2 bytes to get aligned */
+	addi	r3,r3,2
+	subi	r4,r4,2
+	addc	r0,r0,r5
+	srwi.	r6,r4,2		/* # words to do */
+	beq	3f
+1:	mtctr	r6
+2:	lwzu	r5,4(r3)	/* the bdnz has zero overhead, so it should */
+	adde	r0,r0,r5	/* be unnecessary to unroll this loop */
+	bdnz	2b
+	andi.	r4,r4,3
+3:	cmpi	0,r4,2
+	blt+	4f
+	lhz	r5,4(r3)
+	addi	r3,r3,2
+	subi	r4,r4,2
+	adde	r0,r0,r5
+4:	cmpi	0,r4,1
+	bne+	5f
+	lbz	r5,4(r3)
+	slwi	r5,r5,8		/* Upper byte of word */
+	adde	r0,r0,r5
+5:	addze	r3,r0		/* add in final carry */
+	blr
+
+/*
+ * Computes the checksum of a memory block at src, length len,
+ * and adds in "sum" (32-bit), while copying the block to dst.
+ * If an access exception occurs on src or dst, it stores -EFAULT
+ * to *src_err or *dst_err respectively, and (for an error on
+ * src) zeroes the rest of dst.
+ *
+ * csum_partial_copy_generic(src, dst, len, sum, src_err, dst_err)
+ */
+_GLOBAL(csum_partial_copy_generic)
+	addic	r0,r6,0
+	subi	r3,r3,4
+	subi	r4,r4,4
+	srwi.	r6,r5,2
+	beq	3f		/* if we're doing < 4 bytes */
+	andi.	r9,r4,2		/* Align dst to longword boundary */
+	beq+	1f
+81:	lhz	r6,4(r3)	/* do 2 bytes to get aligned */
+	addi	r3,r3,2
+	subi	r5,r5,2
+91:	sth	r6,4(r4)
+	addi	r4,r4,2
+	addc	r0,r0,r6
+	srwi.	r6,r5,2		/* # words to do */
+	beq	3f
+1:	mtctr	r6
+82:	lwzu	r6,4(r3)	/* the bdnz has zero overhead, so it should */
+92:	stwu	r6,4(r4)	/* be unnecessary to unroll this loop */
+	adde	r0,r0,r6
+	bdnz	82b
+	andi.	r5,r5,3
+3:	cmpi	0,r5,2
+	blt+	4f
+83:	lhz	r6,4(r3)
+	addi	r3,r3,2
+	subi	r5,r5,2
+93:	sth	r6,4(r4)
+	addi	r4,r4,2
+	adde	r0,r0,r6
+4:	cmpi	0,r5,1
+	bne+	5f
+84:	lbz	r6,4(r3)
+94:	stb	r6,4(r4)
+	slwi	r6,r6,8		/* Upper byte of word */
+	adde	r0,r0,r6
+5:	addze	r3,r0		/* add in final carry */
+	blr
+
+.section .fixup,"ax"
+
+src_error_1:
+	li	r6,0
+	subi	r5,r5,2
+95:	sth	r6,4(r4)
+	addi	r4,r4,2
+	srwi.	r6,r5,2
+	beq	3f
+	mtctr	r6
+src_error_2:
+	li	r6,0
+96:	stwu	r6,4(r4)
+	bdnz	96b
+3:	andi.	r5,r5,3
+	beq	src_error
+src_error_3:
+	li	r6,0
+	mtctr	r5
+	addi	r4,r4,3
+97:	stbu	r6,1(r4)
+	bdnz	97b
+src_error:
+	cmpi	0,r7,0
+	beq	1f
+	li	r6,-EFAULT
+	stw	r6,0(r7)
+1:	addze	r3,r0
+	blr
+
+dst_error:
+	cmpi	0,r8,0
+	beq	1f
+	li	r6,-EFAULT
+	stw	r6,0(r8)
+1:	addze	r3,r0
+	blr
+
+.section __ex_table,"a"
+	.long	81b,src_error_1
+	.long	91b,dst_error
+	.long	82b,src_error_2
+	.long	92b,dst_error
+	.long	83b,src_error_3
+	.long	93b,dst_error
+	.long	84b,src_error_3
+	.long	94b,dst_error
+	.long	95b,dst_error
+	.long	96b,dst_error
+	.long	97b,dst_error
diff --git a/arch/ppc/lib/strcase.c b/arch/ppc/lib/strcase.c
new file mode 100644
index 000000000..a9e40bce9
--- /dev/null
+++ b/arch/ppc/lib/strcase.c
@@ -0,0 +1,12 @@
+#include <linux/ctype.h>
+
+int strcasecmp(const char *s1, const char *s2)
+{
+	int c1, c2;
+
+	do {
+		c1 = tolower(*s1++);
+		c2 = tolower(*s2++);
+	} while (c1 == c2 && c1 != 0);
+	return c1 - c2;
+}
diff --git a/arch/ppc/lib/string.S b/arch/ppc/lib/string.S
new file mode 100644
index 000000000..96378dec1
--- /dev/null
+++ b/arch/ppc/lib/string.S
@@ -0,0 +1,345 @@
+/*
+ * String handling functions for PowerPC.
+ *
+ * Copyright (C) 1996 Paul Mackerras.
+ *
+ * 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.
+ */
+#include "../kernel/ppc_asm.tmpl"
+#include <asm/errno.h>
+
+	.globl	strcpy
+strcpy:
+	addi	r5,r3,-1
+	addi	r4,r4,-1
+1:	lbzu	r0,1(r4)
+	cmpwi	0,r0,0
+	stbu	r0,1(r5)
+	bne	1b
+	blr
+
+	.globl	strncpy
+strncpy:
+	cmpwi	0,r5,0
+	beqlr
+	mtctr	r5
+	addi	r6,r3,-1
+	addi	r4,r4,-1
+1:	lbzu	r0,1(r4)
+	cmpwi	0,r0,0
+	stbu	r0,1(r6)
+	bdnzf	2,1b		/* dec ctr, branch if ctr != 0 && !cr0.eq */
+	blr
+
+	.globl	strcat
+strcat:
+	addi	r5,r3,-1
+	addi	r4,r4,-1
+1:	lbzu	r0,1(r5)
+	cmpwi	0,r0,0
+	bne	1b
+	addi	r5,r5,-1
+1:	lbzu	r0,1(r4)
+	cmpwi	0,r0,0
+	stbu	r0,1(r5)
+	bne	1b
+	blr
+
+	.globl	strcmp
+strcmp:
+	addi	r5,r3,-1
+	addi	r4,r4,-1
+1:	lbzu	r3,1(r5)
+	cmpwi	1,r3,0
+	lbzu	r0,1(r4)
+	subf.	r3,r0,r3
+	beqlr	1
+	beq	1b
+	blr
+
+	.globl	strlen
+strlen:
+	addi	r4,r3,-1
+1:	lbzu	r0,1(r4)
+	cmpwi	0,r0,0
+	bne	1b
+	subf	r3,r3,r4
+	blr
+
+	.globl	memset
+memset:
+	rlwimi	r4,r4,8,16,23
+	rlwimi	r4,r4,16,0,15
+	addi	r6,r3,-4
+	cmplwi	0,r5,4
+	blt	7f
+	stwu	r4,4(r6)
+	beqlr
+	andi.	r0,r6,3
+	add	r5,r0,r5
+	subf	r6,r0,r6
+	rlwinm	r0,r5,32-2,2,31
+	mtctr	r0
+	bdz	6f
+1:	stwu	r4,4(r6)
+	bdnz	1b
+6:	andi.	r5,r5,3
+7:	cmpwi	0,r5,0
+	beqlr
+	mtctr	r5
+	addi	r6,r6,3
+8:	stbu	r4,1(r6)
+	bdnz	8b
+	blr
+
+	.globl	bcopy
+bcopy:
+	mr	r6,r3
+	mr	r3,r4
+	mr	r4,r6
+	b	memcpy
+
+	.globl	memmove
+memmove:
+	cmplw	0,r3,r4
+	bgt	backwards_memcpy
+	/* fall through */
+
+	.globl	memcpy
+memcpy:
+	rlwinm.	r7,r5,32-3,3,31		/* r0 = r5 >> 3 */
+	addi	r6,r3,-4
+	addi	r4,r4,-4
+	beq	2f			/* if less than 8 bytes to do */
+	andi.	r0,r6,3			/* get dest word aligned */
+	mtctr	r7
+	bne	5f
+1:	lwz	r7,4(r4)
+	lwzu	r8,8(r4)
+	stw	r7,4(r6)
+	stwu	r8,8(r6)
+	bdnz	1b
+	andi.	r5,r5,7
+2:	cmplwi	0,r5,4
+	blt	3f
+	lwzu	r0,4(r4)
+	addi	r5,r5,-4
+	stwu	r0,4(r6)
+3:	cmpwi	0,r5,0
+	beqlr
+	mtctr	r5
+	addi	r4,r4,3
+	addi	r6,r6,3
+4:	lbzu	r0,1(r4)
+	stbu	r0,1(r6)
+	bdnz	4b
+	blr
+5:	subfic	r0,r0,4
+	mtctr	r0
+6:	lbz	r7,4(r4)
+	addi	r4,r4,1
+	stb	r7,4(r6)
+	addi	r6,r6,1
+	bdnz	6b
+	subf	r5,r0,r5
+	rlwinm.	r7,r5,32-3,3,31
+	beq	2b
+	mtctr	r7
+	b	1b
+
+	.globl	backwards_memcpy
+backwards_memcpy:
+	rlwinm.	r7,r5,32-3,3,31		/* r0 = r5 >> 3 */
+	add	r6,r3,r5
+	add	r4,r4,r5
+	beq	2f
+	andi.	r0,r6,3
+	mtctr	r7
+	bne	5f
+1:	lwz	r7,-4(r4)
+	lwzu	r8,-8(r4)
+	stw	r7,-4(r6)
+	stwu	r8,-8(r6)
+	bdnz	1b
+	andi.	r5,r5,7
+2:	cmplwi	0,r5,4
+	blt	3f
+	lwzu	r0,-4(r4)
+	subi	r5,r5,4
+	stwu	r0,-4(r6)
+3:	cmpwi	0,r5,0
+	beqlr
+	mtctr	r5
+4:	lbzu	r0,-1(r4)
+	stbu	r0,-1(r6)
+	bdnz	4b
+	blr
+5:	mtctr	r0
+6:	lbzu	r7,-1(r4)
+	stbu	r7,-1(r6)
+	bdnz	6b
+	subf	r5,r0,r5
+	rlwinm.	r7,r5,32-3,3,31
+	beq	2b
+	mtctr	r7
+	b	1b
+	
+	.globl	memcmp
+memcmp:
+	cmpwi	0,r5,0
+	ble-	2f
+	mtctr	r5
+	addi	r6,r3,-1
+	addi	r4,r4,-1
+1:	lbzu	r3,1(r6)
+	lbzu	r0,1(r4)
+	subf.	r3,r0,r3
+	bdnzt	2,1b
+	blr
+2:	li	r3,0
+	blr
+
+	.global	memchr
+memchr:
+	cmpwi	0,r5,0
+	ble-	2f
+	mtctr	r5
+	addi	r3,r3,-1
+1:	lbzu	r0,1(r3)
+	cmpw	0,r0,r4
+	bdnzf	2,1b
+	beqlr
+2:	li	r3,0
+	blr
+
+	.globl	__copy_tofrom_user
+__copy_tofrom_user:
+	rlwinm.	r7,r5,32-3,3,31		/* r0 = r5 >> 3 */
+	addi	r6,r3,-4
+	addi	r4,r4,-4
+	li	r3,0			/* success return value */
+	beq	2f			/* if less than 8 bytes to do */
+	andi.	r0,r6,3			/* get dest word aligned */
+	mtctr	r7
+	bne	5f
+1:	lwz	r7,4(r4)
+11:	lwzu	r8,8(r4)
+12:	stw	r7,4(r6)
+13:	stwu	r8,8(r6)
+	bdnz	1b
+	andi.	r5,r5,7
+2:	cmplwi	0,r5,4
+	blt	3f
+14:	lwzu	r0,4(r4)
+	addi	r5,r5,-4
+15:	stwu	r0,4(r6)
+3:	cmpwi	0,r5,0
+	beqlr
+	mtctr	r5
+	addi	r4,r4,3
+	addi	r6,r6,3
+4:	lbzu	r0,1(r4)
+16:	stbu	r0,1(r6)
+	bdnz	4b
+	blr
+5:	subfic	r0,r0,4
+	mtctr	r0
+6:	lbz	r7,4(r4)
+	addi	r4,r4,1
+17:	stb	r7,4(r6)
+	addi	r6,r6,1
+	bdnz	6b
+	subf	r5,r0,r5
+	rlwinm.	r7,r5,32-3,3,31
+	beq	2b
+	mtctr	r7
+	b	1b
+99:	li	r3,-EFAULT
+	blr
+.section __ex_table,"a"
+	.align	2
+	.long	1b,99b
+	.long	11b,99b
+	.long	12b,99b
+	.long	13b,99b
+	.long	14b,99b
+	.long	15b,99b
+	.long	4b,99b
+	.long	16b,99b
+	.long	6b,99b
+	.long	17b,99b
+.text
+
+	.globl	__clear_user
+__clear_user:
+	addi	r6,r3,-4
+	li	r3,0
+	li	r5,0
+	cmplwi	0,r4,4
+	blt	7f
+11:	stwu	r5,4(r6)
+	beqlr
+	andi.	r0,r6,3
+	add	r4,r0,r4
+	subf	r6,r0,r6
+	rlwinm	r0,r4,32-2,2,31
+	mtctr	r0
+	bdz	6f
+1:	stwu	r5,4(r6)
+	bdnz	1b
+6:	andi.	r4,r4,3
+7:	cmpwi	0,r4,0
+	beqlr
+	mtctr	r4
+	addi	r6,r6,3
+8:	stbu	r5,1(r6)
+	bdnz	8b
+	blr
+99:	li	r3,-EFAULT
+	blr
+.section __ex_table,"a"
+	.align	2
+	.long	11b,99b
+	.long	1b,99b
+	.long	8b,99b
+.text
+
+	.globl	__strncpy_from_user
+__strncpy_from_user:
+	addi	r6,r3,-1
+	addi	r4,r4,-1
+	cmpwi	0,r5,0
+	beq	2f
+	mtctr	r5
+1:	lbzu	r0,1(r4)
+	cmpwi	0,r0,0
+	stbu	r0,1(r6)
+	bdnzf	2,1b		/* dec ctr, branch if ctr != 0 && !cr0.eq */
+	beq	3f
+2:	addi	r6,r6,1
+3:	subf	r3,r3,r6
+	blr
+99:	li	r3,-EFAULT
+	blr
+.section __ex_table,"a"
+	.align	2
+	.long	1b,99b
+.text
+
+	.globl	strlen_user
+strlen_user:
+	addi	r4,r3,-1
+1:	lbzu	r0,1(r4)
+	cmpwi	0,r0,0
+	bne	1b
+	subf	r3,r3,r4
+	addi	r3,r3,1
+	blr
+99:	li	r3,0
+	blr
+.section __ex_table,"a"
+	.align	2
+	.long	1b,99b
diff --git a/arch/ppc/mkdiff b/arch/ppc/mkdiff
new file mode 100644
index 000000000..304eb9428
--- /dev/null
+++ b/arch/ppc/mkdiff
@@ -0,0 +1,8 @@
+#!/bin/bash
+
+N=`basename $PWD`
+date=`date +'%y%m%d'`
+cd ../
+mkdir -p dist
+echo Diff of: $N against $N.ORIG '->' $N-$date-ppc.patch 
+diff -uNr -X $N/arch/ppc/ignore $N.ORIG $N > dist/$N-$date-ppc.patch
diff --git a/arch/ppc/mkdist b/arch/ppc/mkdist
new file mode 100644
index 000000000..80408b741
--- /dev/null
+++ b/arch/ppc/mkdist
@@ -0,0 +1,13 @@
+#!/bin/bash
+
+V=`egrep ^VERSION Makefile | awk '{print $3}'`
+P=`egrep ^PATCHLEVEL Makefile | awk '{print $3}'`
+S=`egrep ^SUBLEVEL Makefile | awk '{print $3}'`
+date=`date +'%y%m%d'`
+
+echo zImage-$V.$P.$S-$date
+echo System.map-$V.$P.$S-$date
+
+rcp zImage charon:ppc/dist/kernel-images/zImage-$V.$P.$S-$date
+rcp System.map charon:ppc/dist/kernel-images/System.map-$V.$P.$S-$date
+
diff --git a/arch/ppc/mktar b/arch/ppc/mktar
new file mode 100644
index 000000000..a50112ff6
--- /dev/null
+++ b/arch/ppc/mktar
@@ -0,0 +1,9 @@
+#!/bin/bash
+
+V=`egrep ^VERSION Makefile | awk '{print $3}'`
+P=`egrep ^PATCHLEVEL Makefile | awk '{print $3}'`
+S=`egrep ^SUBLEVEL Makefile | awk '{print $3}'`
+date=`date +'%y%m%d'`
+cd ../
+mkdir -p dist
+tar -vzcf ppclinux-$V.$P.$S-$date.tar.gz -X linux/arch/ppc/ignore linux
diff --git a/arch/ppc/mm/Makefile b/arch/ppc/mm/Makefile
index 13aeab6ca..2b07c7227 100644
--- a/arch/ppc/mm/Makefile
+++ b/arch/ppc/mm/Makefile
@@ -7,28 +7,7 @@
 #
 # Note 2! The CFLAGS definition is now in the main makefile...
 
-.c.o:
-	$(CC) $(CFLAGS) -c $<
-.s.o:
-	$(AS) -o $*.o $<
-.c.s:
-	$(CC) $(CFLAGS) -S $<
+O_TARGET := mm.o
+O_OBJS	= fault.o init.o extable.o
 
-OBJS	= fault.o init.o extable.o
-
-mm.o: $(OBJS)
-	$(LD) -r -o mm.o $(OBJS)
-
-modules:
-
-dep:
-
-fastdep:
-	$(TOPDIR)/scripts/mkdep *.[Sch] > .depend
-
-#
-# include a dependency file if one exists
-#
-ifeq (.depend,$(wildcard .depend))
-include .depend
-endif
+include $(TOPDIR)/Rules.make
diff --git a/arch/ppc/mm/extable.c b/arch/ppc/mm/extable.c
new file mode 100644
index 000000000..77e3171bc
--- /dev/null
+++ b/arch/ppc/mm/extable.c
@@ -0,0 +1,65 @@
+/*
+ * linux/arch/ppc/mm/extable.c
+ *
+ * from linux/arch/i386/mm/extable.c
+ */
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <asm/uaccess.h>
+
+extern const struct exception_table_entry __start___ex_table[];
+extern const struct exception_table_entry __stop___ex_table[];
+
+static inline unsigned long
+search_one_table(const struct exception_table_entry *first,
+		 const struct exception_table_entry *last,
+		 unsigned long value)
+{
+	const struct exception_table_entry *mid;
+	for ( mid = first; mid < last; mid++)
+	{
+		if ( mid->insn == value )
+			return mid->fixup;
+	}
+	return 0;
+#if 0	
+        while (first <= last) {
+		const struct exception_table_entry *mid;
+		long diff;
+		mid = (last - first) / 2 + first;
+		diff = mid->insn - value;
+                if (diff == 0)
+                        return mid->fixup;
+                else if (diff < 0)
+                        first = mid+1;
+                else
+                        last = mid-1;
+        }
+#endif	
+        return 0;
+}
+
+unsigned long
+search_exception_table(unsigned long addr)
+{
+	unsigned long ret;
+
+#if 1 /*ndef CONFIG_MODULES*/
+	/* There is only the kernel to search.  */
+	ret = search_one_table(__start___ex_table, __stop___ex_table-1, addr);
+	if (ret) return ret;
+#else
+	/* The kernel is the last "module" -- no need to treat it special.  */
+	struct module *mp;
+	for (mp = module_list; mp != NULL; mp = mp->next) {
+		if (mp->ex_table_start == NULL)
+			continue;
+		ret = search_one_table(mp->ex_table_start,
+				       mp->ex_table_end - 1, addr);
+		if (ret) return ret;
+	}
+#endif
+
+	return 0;
+}
diff --git a/arch/ppc/mm/fault.c b/arch/ppc/mm/fault.c
index fea722780..67b94d809 100644
--- a/arch/ppc/mm/fault.c
+++ b/arch/ppc/mm/fault.c
@@ -1,8 +1,12 @@
 /*
  *  arch/ppc/mm/fault.c
  *
- *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
- *  Ported to PPC by Gary Thomas
+ *  PowerPC version 
+ *    Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org)
+ *
+ *  Derived from "arch/i386/mm/fault.c"
+ *    Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
+ *
  *  Modified by Cort Dougan and Paul Mackerras.
  *
  *  This program is free software; you can redistribute it and/or
@@ -26,40 +30,42 @@
 
 #include <asm/page.h>
 #include <asm/pgtable.h>
+#include <asm/mmu.h>
 #include <asm/mmu_context.h>
+#include <asm/system.h>
+#include <asm/uaccess.h>
 
-#ifdef CONFIG_PMAC
+#ifdef CONFIG_XMON
+extern void xmon(struct pt_regs *);
 extern void (*xmon_fault_handler)(void);
-#endif
-
-/* the linux norm for the function name is show_regs() so
-   make it call dump_regs() on the mac -- Cort */
-#ifdef CONFIG_PMAC 
-#define show_regs dump_regs
+extern int xmon_dabr_match(struct pt_regs *);
+int xmon_kernel_faults;
 #endif
 
 extern void die_if_kernel(char *, struct pt_regs *, long);
 void bad_page_fault(struct pt_regs *, unsigned long);
 void do_page_fault(struct pt_regs *, unsigned long, unsigned long);
-void print_pte(struct _PTE);
 
-void do_page_fault(struct pt_regs *regs, unsigned long address, unsigned long error_code)     
+/*
+ * The error_code parameter is DSISR for a data fault, SRR1 for
+ * an instruction fault.
+ */
+void do_page_fault(struct pt_regs *regs, unsigned long address,
+		   unsigned long error_code)
 {
-	struct task_struct *tsk = current;
-	extern unsigned _end[];
 	struct vm_area_struct * vma;
 	struct mm_struct *mm = current->mm;
-	pgd_t *dir;
-	pmd_t *pmd;
-	pte_t *pte;
-	
-	/*printk("do_page_fault() %s/%d addr %x nip %x regs %x error %x\n",
-	       current->comm,current->pid,address,regs->nip,regs,error_code);*/
-#ifdef CONFIG_PMAC
+
+#ifdef CONFIG_XMON
 	if (xmon_fault_handler && regs->trap == 0x300) {
 		xmon_fault_handler();
 		return;
 	}
+	if (error_code & 0x00400000) {
+		/* DABR match */
+		if (xmon_dabr_match(regs))
+			return;
+	}
 #endif
 	if (in_interrupt()) {
 		static int complained;
@@ -68,14 +74,18 @@ void do_page_fault(struct pt_regs *regs, unsigned long address, unsigned long er
 			printk("page fault in interrupt handler, addr=%lx\n",
 			       address);
 			show_regs(regs);
+#ifdef CONFIG_XMON
+			if (xmon_kernel_faults)
+				xmon(regs);
+#endif
 		}
 	}
-	if (current == NULL)
-		goto bad_area;
-	
-do_page:
+	if (current == NULL) {
+		bad_page_fault(regs, address);
+		return;
+	}
 	down(&mm->mmap_sem);
-	vma = find_vma(tsk->mm, address);
+	vma = find_vma(mm, address);
 	if (!vma)
 		goto bad_area;
 	if (vma->vm_start <= address)
@@ -84,7 +94,7 @@ do_page:
 		goto bad_area;
 	if (expand_stack(vma, address))
 		goto bad_area;
-  
+
 good_area:
 	if (error_code & 0xb5700000)
 		/* an error such as lwarx to I/O controller space,
@@ -98,67 +108,45 @@ good_area:
 	/* a read */
 	} else {
 		/* protection fault */
-		if ( error_code & 0x08000000 )
+		if (error_code & 0x08000000)
 			goto bad_area;
 		if (!(vma->vm_flags & (VM_READ | VM_EXEC)))
 			goto bad_area;
 	}
 	handle_mm_fault(current, vma, address, error_code & 0x02000000);
-        up(&mm->mmap_sem);
-	/*printk("do_page_fault() return %s/%d addr %x msr %x\n",
-	      current->comm,current->pid,address,regs->msr);*/
-	/* not needed since flush_page_to_ram() works */
-#if 0
-	flush_page(address);
-#endif
+	up(&mm->mmap_sem);
 	return;
-  
+
 bad_area:
-	up(&current->mm->mmap_sem);
+	up(&mm->mmap_sem);
 	bad_page_fault(regs, address);
 }
 
-
 void
 bad_page_fault(struct pt_regs *regs, unsigned long address)
 {
-	extern unsigned int probingmem;
-	struct task_struct *tsk = current;
 	unsigned long fixup;
-	
+
+	if (user_mode(regs)) {
+		force_sig(SIGSEGV, current);
+		return;
+	}
 	
 	/* Are we prepared to handle this fault?  */
 	if ((fixup = search_exception_table(regs->nip)) != 0) {
-		if ( user_mode(regs) )
-			printk("Exception from user mode\n");
-#if 0    
-		printk(KERN_DEBUG "Exception at %lx (%lx)\n", regs->nip, fixup);
-#endif    
 		regs->nip = fixup;
 		return;
 	}
 
-	if ( user_mode(regs) )
-	{
-		force_sig(SIGSEGV, tsk);
-		return;
-	}
-  
-bad_kernel_access:
-	/* make sure it's not a bootup probe test */
-	if ( probingmem )
-	{
-		probingmem = 0;
-		return;
-	}
 	/* kernel has accessed a bad area */
 	show_regs(regs);
-	print_backtrace( regs->gpr[1] );
-#ifdef CONFIG_PMAC
-	xmon(regs);
+	print_backtrace( (unsigned long *)regs->gpr[1] );
+#ifdef CONFIG_XMON
+	if (xmon_kernel_faults)
+		xmon(regs);
 #endif
-	panic("kernel access of bad area\n pc %x address %X tsk %s/%d",
-	      regs->nip,address,tsk->comm,tsk->pid);
+	panic("kernel access of bad area pc %lx lr %lx address %lX tsk %s/%d",
+	      regs->nip,regs->link,address,current->comm,current->pid);
 }
 
 unsigned long va_to_phys(unsigned long address)
@@ -189,6 +177,10 @@ unsigned long va_to_phys(unsigned long address)
 	return (0);
 }
 
+#if 0
+/*
+ * Misc debugging functions.  Please leave them here. -- Cort
+ */
 void print_pte(struct _PTE p)
 {
 	printk(
@@ -210,7 +202,8 @@ unsigned long htab_phys_to_va(unsigned long address)
 	for ( ptr = Hash ; ptr < Hash_end ; ptr++ )
 	{
 		if ( ptr->rpn == (address>>12) )
-			printk("phys %08X -> va ???\n",
+			printk("phys %08lX -> va ???\n",
 			       address);
 	}
 }
+#endif
diff --git a/arch/ppc/mm/init.c b/arch/ppc/mm/init.c
index d46975996..8503401b4 100644
--- a/arch/ppc/mm/init.c
+++ b/arch/ppc/mm/init.c
@@ -31,54 +31,58 @@
 #include <linux/mm.h>
 #include <linux/swap.h>
 #include <linux/stddef.h>
-#ifdef CONFIG_PMAC
 #include <asm/prom.h>
-#endif
 #include <asm/io.h>
 #include <asm/mmu_context.h>
 #include <asm/pgtable.h>
 #include <asm/mmu.h>
-#ifdef CONFIG_PREP
 #include <asm/residual.h>
+#ifdef CONFIG_BLK_DEV_INITRD
+#include <linux/blk.h>		/* for initrd_* */
 #endif
 
+int prom_trashed;
 int next_mmu_context;
+unsigned long _SDR1;
+PTE *Hash, *Hash_end;
+unsigned long Hash_size, Hash_mask;
+unsigned long *end_of_DRAM;
+int mem_init_done;
 extern pgd_t swapper_pg_dir[];
 extern char _start[], _end[];
 extern char etext[], _stext[];
-/* References to section boundaries */
 extern char __init_begin, __init_end;
+extern RESIDUAL res;
 
-extern void die_if_kernel(char *,struct pt_regs *,long);
-extern void show_net_buffers(void);
-extern unsigned long *find_end_of_memory(void);
+/* Hardwired MMU segments */
+#if defined(CONFIG_PREP) || defined(CONFIG_PMAC)
+#define MMU_SEGMENT_1		0x80000000
+#define MMU_SEGMENT_2		0xc0000000
+#endif /* CONFIG_PREP || CONFIG_PMAC */
+#ifdef CONFIG_CHRP
+#define MMU_SEGMENT_1		0xf0000000	/* LongTrail */
+#define MMU_SEGMENT_2		0xc0000000
+#endif /* CONFIG_CHRP */
 
-#undef MAP_RAM_WITH_SEGREGS 1
 
-#ifdef CONFIG_PMAC
 void *find_mem_piece(unsigned, unsigned);
 static void mapin_ram(void);
 static void inherit_prom_translations(void);
-#endif
-#ifdef CONFIG_PREP
-inline void MMU_invalidate_page(struct mm_struct *mm, unsigned long va);
-int inline MMU_hash_page(struct task_struct *,unsigned long,pte *);     
-#endif
-
 static void hash_init(void);
 static void *MMU_get_page(void);
-void map_page(struct thread_struct *, unsigned long va,
+void map_page(struct task_struct *, unsigned long va,
 		     unsigned long pa, int flags);
+extern void die_if_kernel(char *,struct pt_regs *,long);
+extern void show_net_buffers(void);
+extern unsigned long *find_end_of_memory(void);
 
-PTE *Hash, *Hash_end;
-unsigned long Hash_size, Hash_mask;
-unsigned long *end_of_DRAM;
-int mem_init_done;
+/*
+ * this tells the prep system to map all of ram with the segregs
+ * instead of the bats.  I'd like to get this to apply to the
+ * pmac as well then have everything use the bats -- Cort
+ */
+#undef MAP_RAM_WITH_SEGREGS 1 
 
-#ifdef CONFIG_PREP
-#ifdef HASHSTATS
-extern unsigned long evicts;
-#endif /* HASHSTATS */
 /*
  * these are used to setup the initial page tables
  * They can waste up to an entire page since the
@@ -88,8 +92,6 @@ extern unsigned long evicts;
 unsigned int probingmem = 0;
 unsigned int mmu_pages_count = 0;
 char mmu_pages[(MAX_MMU_PAGES+1)*PAGE_SIZE];
-unsigned long _TotalMemory;
-#endif /* CONFIG_PREP */
 
 /*
  * BAD_PAGE is the page that is used for page faults when linux
@@ -120,7 +122,6 @@ pte_t __bad_page(void)
 	return pte_mkdirty(mk_pte(empty_bad_page, PAGE_SHARED));
 }
 
-#ifdef CONFIG_PMAC
 #define MAX_MEM_REGIONS	32
 phandle memory_pkg;
 
@@ -128,7 +129,6 @@ struct mem_pieces {
 	int n_regions;
 	struct reg_property regions[MAX_MEM_REGIONS];
 };
-
 struct mem_pieces phys_mem;
 struct mem_pieces phys_avail;
 struct mem_pieces prom_mem;
@@ -138,7 +138,6 @@ static void remove_mem_piece(struct mem_pieces *, unsigned, unsigned, int);
 static void print_mem_pieces(struct mem_pieces *);
 
 unsigned long avail_start;
-int prom_trashed;
 
 /*
  * Read in a property describing some pieces of memory.
@@ -272,10 +271,10 @@ find_mem_piece(unsigned size, unsigned align)
  * Our text, data, bss use something over 1MB, starting at 0.
  * Open Firmware may be using 1MB at the 4MB point.
  */
-unsigned long *find_end_of_memory(void)
+unsigned long *pmac_find_end_of_memory(void)
 {
 	unsigned long a, total;
-	unsigned long h, kstart, ksize;
+	unsigned long kstart, ksize;
 	extern char _stext[], _end[];
 	int i;
 
@@ -325,20 +324,6 @@ unsigned long *find_end_of_memory(void)
 	remove_mem_piece(&phys_avail, kstart, ksize, 0);
 	remove_mem_piece(&prom_mem, kstart, ksize, 0);
 
-	/*
-	 * Allow 64k of hash table for every 16MB of memory,
-	 * up to a maximum of 2MB.
-	 */
-	for (h = 64<<10; h < total / 256 && h < 2<<20; h *= 2)
-		;
-	Hash_size = h;
-	Hash_mask = (h >> 6) - 1;
-
-	/* Find some memory for the hash table. */
-	Hash = find_mem_piece(Hash_size, Hash_size);
-	printk("Total memory = %ldMB; using %ldkB for hash table (at %p)\n",
-	       total >> 20, Hash_size >> 10, Hash);
-
 	return __va(total);
 }
 
@@ -353,6 +338,9 @@ unsigned long find_available_memory(void)
 	int i;
 	unsigned long a, free;
 	unsigned long start, end;
+	
+	if ( _machine != _MACH_Pmac )
+		return 0;
 
 	free = 0;
 	for (i = 0; i < phys_avail.n_regions - 1; ++i) {
@@ -366,7 +354,6 @@ unsigned long find_available_memory(void)
 	avail_start = (unsigned long) __va(a);
 	return avail_start;
 }
-#endif /* CONFIG_PMAC */
 
 void show_mem(void)
 {
@@ -399,14 +386,21 @@ void show_mem(void)
 	       "Ctx", "Ctx<<4", "Last Sys", "pc", "task");
 	for_each_task(p)
 	{	
-		printk("%-8.8s %3d %3d %8d %8d %8d %c%08x %08x",
+		printk("%-8.8s %3d %3d %8ld %8ld %8ld %c%08lx %08lx ",
 		       p->comm,p->pid,
 		       p->mm->count,p->mm->context,
 		       p->mm->context<<4, p->tss.last_syscall,
 		       user_mode(p->tss.regs) ? 'u' : 'k', p->tss.regs->nip,
-		       p);
+		       (ulong)p);
 		if ( p == current )
-			printk(" current");
+			printk("current");
+		if ( p == last_task_used_math )
+		{
+			if ( p == current )
+				printk(",");
+			printk("last math");
+		}
+
 		printk("\n");
 	}
 }
@@ -434,76 +428,68 @@ unsigned long paging_init(unsigned long start_mem, unsigned long end_mem)
 void mem_init(unsigned long start_mem, unsigned long end_mem)
 {
 	unsigned long addr;
-#ifdef CONFIG_PMAC
 	int i;
-	unsigned long lim;
-#endif
+	unsigned long a, lim;
 	int codepages = 0;
 	int datapages = 0;
 	int initpages = 0;
 
 	end_mem &= PAGE_MASK;
 	high_memory = (void *) end_mem;
-	num_physpages = max_mapnr = MAP_NR(high_memory);
-	
-	/* clear the zero-page */
-	memset(empty_zero_page, 0, PAGE_SIZE);
-	
+	max_mapnr = MAP_NR(high_memory);
+	num_physpages = max_mapnr;	/* RAM is assumed contiguous */
+
 	/* mark usable pages in the mem_map[] */
 	start_mem = PAGE_ALIGN(start_mem);
-	
-#ifdef CONFIG_PMAC
-	remove_mem_piece(&phys_avail, __pa(avail_start),
-			 start_mem - avail_start, 1);
-
-	for (a = KERNELBASE ; a < end_mem; a += PAGE_SIZE)
-		set_bit(PG_reserved, &mem_map[MAP_NR(a)].flags);
-
-	for (i = 0; i < phys_avail.n_regions; ++i) {
-		a = (unsigned long) __va(phys_avail.regions[i].address);
-		lim = a + phys_avail.regions[i].size;
-		a = PAGE_ALIGN(a);
-		for (; a < lim; a += PAGE_SIZE) {
-			clear_bit(PG_reserved, &mem_map[MAP_NR(a)].flags);
-			mem_map[MAP_NR(a)].count = 1;
-			free_page(a);
+
+	if ( _machine == _MACH_Pmac )
+	{
+		remove_mem_piece(&phys_avail, __pa(avail_start),
+				 start_mem - avail_start, 1);
+		
+		for (addr = KERNELBASE ; addr < end_mem; addr += PAGE_SIZE)
+			set_bit(PG_reserved, &mem_map[MAP_NR(addr)].flags);
+		
+		for (i = 0; i < phys_avail.n_regions; ++i) {
+			a = (unsigned long) __va(phys_avail.regions[i].address);
+			lim = a + phys_avail.regions[i].size;
+			a = PAGE_ALIGN(a);
+			for (; a < lim; a += PAGE_SIZE)
+				clear_bit(PG_reserved, &mem_map[MAP_NR(a)].flags);
 		}
-	}
-	phys_avail.n_regions = 0;
-
-	/* free the prom's memory */
-	for (i = 0; i < prom_mem.n_regions; ++i) {
-		a = (unsigned long) __va(prom_mem.regions[i].address);
-		lim = a + prom_mem.regions[i].size;
-		a = PAGE_ALIGN(a);
-		for (; a < lim; a += PAGE_SIZE) {
-			clear_bit(PG_reserved, &mem_map[MAP_NR(a)].flags);
-			mem_map[MAP_NR(a)].count = 1;
-			free_page(a);
+		phys_avail.n_regions = 0;
+		
+		/* free the prom's memory */
+		for (i = 0; i < prom_mem.n_regions; ++i) {
+			a = (unsigned long) __va(prom_mem.regions[i].address);
+			lim = a + prom_mem.regions[i].size;
+			a = PAGE_ALIGN(a);
+			for (; a < lim; a += PAGE_SIZE)
+				clear_bit(PG_reserved, &mem_map[MAP_NR(a)].flags);
 		}
+		prom_trashed = 1;
 	}
-	prom_trashed = 1;
-#endif /* CONFIG_PMAC */
-
-#ifdef CONFIG_PREP
-	/* mark mem used by kernel as reserved, mark other unreserved */
-	for (addr = PAGE_OFFSET ; addr < end_mem; addr += PAGE_SIZE)
+	else /* prep */
 	{
-		/* skip hash table gap */
-		if ( (addr > (ulong)_end) && (addr < (ulong)Hash))
-			continue;
-		if ( addr < (ulong) /*Hash_end*/ start_mem )
-			set_bit(PG_reserved, &mem_map[MAP_NR(addr)].flags);
-		else
-			clear_bit(PG_reserved, &mem_map[MAP_NR(addr)].flags);
+		/* mark mem used by kernel as reserved, mark other unreserved */
+		for (addr = PAGE_OFFSET ; addr < end_mem; addr += PAGE_SIZE)
+		{
+			/* skip hash table gap */
+			if ( (addr > (ulong)_end) && (addr < (ulong)Hash))
+				continue;
+			if ( addr < (ulong) /*Hash_end*/ start_mem )
+				set_bit(PG_reserved, &mem_map[MAP_NR(addr)].flags);
+			else
+				clear_bit(PG_reserved, &mem_map[MAP_NR(addr)].flags);
+		}
 	}
 
 	for (addr = PAGE_OFFSET; addr < end_mem; addr += PAGE_SIZE) {
-		if(PageReserved(mem_map + MAP_NR(addr))) {
+		if (PageReserved(mem_map + MAP_NR(addr))) {
 			if (addr < (ulong) etext)
 				codepages++;
-			/*else if((addr >= (unsigned long)&__init_begin && addr < (unsigned long)&__init_end))
-                                initpages++;*/
+			else if((addr >= (unsigned long)&__init_begin && addr < (unsigned long)&__init_end))
+                                initpages++;
                         else if (addr < (ulong) start_mem)
 				datapages++;
 			continue;
@@ -515,9 +501,8 @@ void mem_init(unsigned long start_mem, unsigned long end_mem)
 #endif /* CONFIG_BLK_DEV_INITRD */
 			free_page(addr);
 	}
-	
-#endif /* CONFIG_PREP */
-        printk("Memory: %luk available (%dk kernel code, %dk data, %dk init) [%08lx,%08lx]\n",
+
+        printk("Memory: %luk available (%dk kernel code, %dk data, %dk init) [%08x,%08lx]\n",
 	       (unsigned long) nr_free_pages << (PAGE_SHIFT-10),
 	       codepages << (PAGE_SHIFT-10),
 	       datapages << (PAGE_SHIFT-10), 
@@ -534,7 +519,6 @@ void mem_init(unsigned long start_mem, unsigned long end_mem)
  */
 void free_initmem(void)
 {
-	unsigned long addr;
 	unsigned long a;
 	unsigned long num_freed_pages = 0;
 
@@ -548,16 +532,14 @@ void free_initmem(void)
 		num_freed_pages++;
 	}
 
-#if 0
-	addr = (unsigned long)(&__init_begin);
-	for (; addr < (unsigned long)(&__init_end); addr += PAGE_SIZE) {
-		num_freed_pages++;
-		mem_map[MAP_NR(addr)].flags &= ~(1 << PG_reserved);
-		mem_map[MAP_NR(addr)].count = 1;
-		free_page(addr);
+	a = (unsigned long)(&__init_begin);
+	for (; a < (unsigned long)(&__init_end); a += PAGE_SIZE) {
+		mem_map[MAP_NR(a)].flags &= ~(1 << PG_reserved);
+		atomic_set(&mem_map[MAP_NR(a)].count, 1);
+		free_page(a);
 	}
-#endif	
-	printk ("Freeing unused kernel memory: %dk freed\n",
+
+	printk ("Freeing unused kernel memory: %ldk freed\n",
 		(num_freed_pages * PAGE_SIZE) >> 10);
 }
 
@@ -576,29 +558,23 @@ void si_meminfo(struct sysinfo *val)
 		val->totalram++;
 		if (!atomic_read(&mem_map[i].count))
 			continue;
-		val->sharedram += atomic_read(&mem_map[i].count)-1;
+		val->sharedram += atomic_read(&mem_map[i].count) - 1;
 	}
 	val->totalram <<= PAGE_SHIFT;
 	val->sharedram <<= PAGE_SHIFT;
 	return;
 }
 
-/* Kernel MMU setup & lowest level hardware support */
-
-unsigned long _SDR1;		/* Hardware SDR1 image */
-
-#ifdef CONFIG_PREP
-
 BAT BAT0 =
 {
   {
-    0x80000000>>17, 	/* bepi */
+    MMU_SEGMENT_1>>17, 	/* bepi */
     BL_256M,		/* bl */
     1,			/* vs -- supervisor mode valid */
     1,			/* vp -- user mode valid */
   },
   {
-    0x80000000>>17,		/* brpn */
+    MMU_SEGMENT_1>>17,		/* brpn */
     1,			/* write-through */
     1,			/* cache-inhibited */
     0,			/* memory coherence */
@@ -609,13 +585,13 @@ BAT BAT0 =
 BAT BAT1 =
 {
   {
-    0xC0000000>>17, 	/* bepi */
+    MMU_SEGMENT_2>>17, 	/* bepi */
     BL_256M,		/* bl */
     1,			/* vs */
     1,			/* vp */
   },
   {
-    0xC0000000>>17,		/* brpn */
+    MMU_SEGMENT_2>>17,		/* brpn */
     1,			/* w */
     1,			/* i (cache disabled) */
     0,			/* m */
@@ -635,7 +611,11 @@ BAT BAT2 =
     0x00000000>>17,		/* brpn */
     0,			/* w */
     0,			/* i */
+#ifdef __SMP__    
     1,			/* m */
+#else    
+    0,			/* m */
+#endif    
     0,			/* g */
     BPP_RW			/* pp */
   }
@@ -676,13 +656,13 @@ P601_BAT BAT0_601 =
 P601_BAT BAT1_601 =
 {
   {
-    0xC0000000>>17, 	/* bepi */
+    MMU_SEGMENT_2>>17, 	/* bepi */
     1,1,0, /* wim */
     1, 0, /* vs, vp */
     BPP_RW, /* pp */
   },
   {
-    0xC0000000>>17,		/* brpn */
+    MMU_SEGMENT_2>>17,		/* brpn */
     1, /* v */
     BL_8M, /* bl */
   }
@@ -724,69 +704,24 @@ P601_BAT BAT3_601 =
  * this will likely stay seperate from the pmac.
  * -- Cort
  */
-unsigned long *find_end_of_memory(void)
+unsigned long *prep_find_end_of_memory(void)
 {
-	extern RESIDUAL res;
-	extern unsigned long resptr;
-	int i, p;
-	unsigned long h;
-
-	/* copy residual data */
-	if ( resptr )
-		memcpy( &res, (void *)(resptr+KERNELBASE), sizeof(RESIDUAL) );
-	else
-		/* clearing bss probably clears this but... */
-		memset( &res, sizeof(RESIDUAL), 0 );
-	_TotalMemory = res.TotalMemory;
+	int i;
 	
-	/* this really has nothing to do with the mmu_init() but is
-	   necessary for early setup -- Cort */
-	if (!strncmp(res.VitalProductData.PrintableModel,"IBM",3))
-	{
-		_machine = _MACH_IBM;
-	}
-	else
-		_machine = _MACH_Motorola;
-  
-	/* setup the hash table */
-	if (_TotalMemory == 0 )
+	if (res.TotalMemory == 0 )
 	{
 		/*
 		 * I need a way to probe the amount of memory if the residual
 		 * data doesn't contain it. -- Cort
 		 */
 		printk("Ramsize from residual data was 0 -- Probing for value\n");
-		_TotalMemory = 0x03000000;
-		printk("Ramsize default to be %dM\n", _TotalMemory>>20);
+		res.TotalMemory = 0x03000000;
+		printk("Ramsize default to be %ldM\n", res.TotalMemory>>20);
 	}
-	
-#if 0
-	/* linux has trouble with > 64M ram -- Cort */
-	if ( _TotalMemory > 0x04000000 /* 64M */ )
-	{
-		printk("Only using first 64M of ram.\n");
-		_TotalMemory = 0x04000000;
-	}
-#endif	
-	
-	/* setup the bat2 mapping to cover physical ram */
-	BAT2.batu.bl = 0x1; /* 256k mapping */
-	for ( h = 256*1024 /* 256k */ ; (h <= _TotalMemory) && (h <= 256*1024*1024);
-	      h *= 2 )
-		BAT2.batu.bl = (BAT2.batu.bl << 1) | BAT2.batu.bl;
-	/*
-	 * Allow 64k of hash table for every 16MB of memory,
-	 * up to a maximum of 2MB.
-	 */
-	for (h = 64<<10; h < _TotalMemory / 256 && h < 2<<20; h *= 2)
-		;
-	Hash_size = h;
-	Hash_mask = (h >> 6) - 1;
-	
-	/* align htab on a Hash_size boundry above _end[] */
-	Hash = (PTE *)_ALIGN( (unsigned long)&_end, Hash_size);
-	memset(Hash, Hash_size, 0 );
 
+	/* NOTE: everything below here is moving to mapin_ram() */
+
+	
 	/*
 	 * if this is a 601, we can only map sizes of 8M with the BAT's
 	 * so we have to map what we can't map with the bats with the segregs
@@ -799,11 +734,11 @@ unsigned long *find_end_of_memory(void)
 	if ( _get_PVR() == 1 )
 	{
 		/* map in rest of ram with seg regs */
-		if ( _TotalMemory > 0x01000000 /* 16M */)
+		if ( res.TotalMemory > 0x01000000 /* 16M */)
 		{
 			for (i = KERNELBASE+0x01000000;
-			     i < KERNELBASE+_TotalMemory;  i += PAGE_SIZE)
-				map_page(&init_task.tss, i, __pa(i),
+			     i < KERNELBASE+res.TotalMemory;  i += PAGE_SIZE)
+				map_page(&init_task, i, __pa(i),
 					 _PAGE_PRESENT| _PAGE_RW|_PAGE_DIRTY|_PAGE_ACCESSED);
 		}
 	}
@@ -815,25 +750,21 @@ unsigned long *find_end_of_memory(void)
 	memset(&BAT2_601, sizeof(BAT2), 0); /* in case we're on a 601 */
 	memset(&BAT3_601, sizeof(BAT2), 0);
 	/* map all of ram for kernel with segregs */
-	for (i = KERNELBASE;  i < KERNELBASE+_TotalMemory;  i += PAGE_SIZE)
+	for (i = KERNELBASE;  i < KERNELBASE+res.TotalMemory;  i += PAGE_SIZE)
 	{
 		if ( i < (unsigned long)etext )
-			map_page(&init_task.tss, i, __pa(i),
+			map_page(&init_task, i, __pa(i),
 				 _PAGE_PRESENT/*| _PAGE_RW*/|_PAGE_DIRTY|_PAGE_ACCESSED);
 		else
-			map_page(&init_task.tss, i, __pa(i),
+			map_page(&init_task, i, __pa(i),
 				 _PAGE_PRESENT| _PAGE_RW|_PAGE_DIRTY|_PAGE_ACCESSED);
 	}
 #endif /* MAP_RAM_WITH_SEGREGS */
 	
-	printk("Total memory = %ldMB; using %ldkB for hash table (at %p)\n",
-	       _TotalMemory >> 20, Hash_size >> 10, Hash);
-	return ((unsigned long *)_TotalMemory);
+	return (__va(res.TotalMemory));
 }
-#endif /* CONFIG_PREP */
 
 
-#ifdef CONFIG_PMAC
 /*
  * Map in all of physical memory starting at KERNELBASE.
  */
@@ -847,17 +778,39 @@ static void mapin_ram()
     int i;
     unsigned long v, p, s, f;
 
-    v = KERNELBASE;
-    for (i = 0; i < phys_mem.n_regions; ++i) {
-	p = phys_mem.regions[i].address;
-	for (s = 0; s < phys_mem.regions[i].size; s += PAGE_SIZE) {
-	    f = _PAGE_PRESENT | _PAGE_ACCESSED;
-	    if ((char *) v < _stext || (char *) v >= etext)
-		f |= _PAGE_RW | _PAGE_DIRTY | _PAGE_HWWRITE;
-	    map_page(&init_task.tss, v, p, f);
-	    v += PAGE_SIZE;
-	    p += PAGE_SIZE;
-	}
+    if ( _machine == _MACH_Pmac )
+    {
+	    v = KERNELBASE;
+	    for (i = 0; i < phys_mem.n_regions; ++i) {
+		    p = phys_mem.regions[i].address;
+		    for (s = 0; s < phys_mem.regions[i].size; s += PAGE_SIZE) {
+			    f = _PAGE_PRESENT | _PAGE_ACCESSED;
+			    if ((char *) v < _stext || (char *) v >= etext)
+				    f |= _PAGE_RW | _PAGE_DIRTY | _PAGE_HWWRITE;
+			    else
+				    /* On the powerpc, no user access forces R/W kernel access */
+				    f |= _PAGE_USER;
+			    map_page(&init_task, v, p, f);
+			    v += PAGE_SIZE;
+			    p += PAGE_SIZE;
+		    }
+	    }
+    }
+    else /* prep */
+    {
+	    /* setup the bat2 mapping to cover physical ram */
+	    BAT2.batu.bl = 0x1; /* 256k mapping */
+	    for ( f = 256*1024 /* 256k */ ;
+		  (f <= res.TotalMemory) && (f <= 256*1024*1024);
+		  f *= 2 )
+		    BAT2.batu.bl = (BAT2.batu.bl << 1) | BAT2.batu.bl;
+	    /*
+	     * let ibm get to the device mem from user mode since
+	     * the X for them needs it right now -- Cort
+	     */
+	    if ( _machine == _MACH_IBM )
+		    BAT0.batu.vp = BAT1.batu.vp = 1;
+	    
     }
 }
 
@@ -889,7 +842,7 @@ static void inherit_prom_translations()
 
 	for (tp = prom_translations, i = 0; i < n_translations; ++i, ++tp) {
 		/* ignore stuff mapped down low */
-		if (tp->virt < 0x10000000)
+		if (tp->virt < 0x10000000 && tp->phys < 0x10000000)
 			continue;
 		/* map PPC mmu flags to linux mm flags */
 		f = (tp->flags & (_PAGE_NO_CACHE | _PAGE_WRITETHRU
@@ -900,13 +853,12 @@ static void inherit_prom_translations()
 		p = tp->phys;
 		n = tp->size;
 		for (; n != 0; n -= PAGE_SIZE) {
-			map_page(&init_task.tss, v, p, f);
+			map_page(&init_task, v, p, f);
 			v += PAGE_SIZE;
 			p += PAGE_SIZE;
 		}
 	}
 }
-#endif
 
 /*
  * Initialize the hash table and patch the instructions in head.S.
@@ -914,10 +866,29 @@ static void inherit_prom_translations()
 static void hash_init(void)
 {
 	int Hash_bits;
+	unsigned long h;
 
 	extern unsigned int hash_page_patch_A[], hash_page_patch_B[],
 		hash_page_patch_C[];
 
+	/*
+	 * Allow 64k of hash table for every 16MB of memory,
+	 * up to a maximum of 2MB.
+	 */
+	for (h = 64<<10; h < (ulong)__pa(end_of_DRAM) / 256 && h < 2<<20; h *= 2)
+		;
+	Hash_size = h;
+	Hash_mask = (h >> 6) - 1;
+
+	/* Find some memory for the hash table. */
+	if ( is_prep )
+		/* align htab on a Hash_size boundry above _end[] */
+		Hash = (PTE *)_ALIGN( (unsigned long)&_end, Hash_size);
+	else /* pmac */
+		Hash = find_mem_piece(Hash_size, Hash_size);
+	
+	printk("Total memory = %ldMB; using %ldkB for hash table (at %p)\n",
+	       __pa(end_of_DRAM) >> 20, Hash_size >> 10, Hash);
 	memset(Hash, 0, Hash_size);
 	Hash_end = (PTE *) ((unsigned long)Hash + Hash_size);
 
@@ -943,8 +914,8 @@ static void hash_init(void)
 	 * out from the data cache and invalidated in the instruction
 	 * cache, on those machines with split caches.
 	 */
-	store_cache_range((unsigned long) hash_page_patch_A,
-			  (unsigned long) (hash_page_patch_C + 1));
+	flush_icache_range((unsigned long) hash_page_patch_A,
+			   (unsigned long) (hash_page_patch_C + 1));
 }
 
 
@@ -958,71 +929,93 @@ static void hash_init(void)
 void
 MMU_init(void)
 {
-        end_of_DRAM = find_end_of_memory();
+	if ( _machine == _MACH_Pmac )
+		end_of_DRAM = pmac_find_end_of_memory();
+	else /* prep and chrp */
+		end_of_DRAM = prep_find_end_of_memory();
+		
         hash_init();
         _SDR1 = __pa(Hash) | (Hash_mask >> 10);
-#ifdef CONFIG_PMAC
-        /* Force initial page tables */
-	/* this done by INIT_TSS in processor.h on prep -- Cort */
-        init_task.tss.pg_tables = (unsigned long *)swapper_pg_dir;
 
 	/* Map in all of RAM starting at KERNELBASE */
 	mapin_ram();
-	/* Copy mappings from the prom */
-	inherit_prom_translations();
-#endif /* CONFIG_PMAC */
+	if ( _machine == _MACH_Pmac )
+		/* Copy mappings from the prom */
+		inherit_prom_translations();
 }
 
 static void *
 MMU_get_page()
 {
-  void *p;
-  
-  if (mem_init_done) {
-    p = (void *) __get_free_page(GFP_KERNEL);
-    if (p == 0)
-      panic("couldn't get a page in MMU_get_page");
-  } else {
-#ifdef CONFIG_PREP
-    mmu_pages_count++;
-    if ( mmu_pages_count > MAX_MMU_PAGES )
-      printk("out of mmu pages!\n");
-    p = (pte *)(PAGE_ALIGN((unsigned long)mmu_pages)+
-		(mmu_pages_count+PAGE_SIZE));
-#endif
-#ifdef CONFIG_PMAC
-    p = find_mem_piece(PAGE_SIZE, PAGE_SIZE);
-#endif
-  }
-  memset(p, 0, PAGE_SIZE);
-  return p;
+	void *p;
+
+	if (mem_init_done) {
+		p = (void *) __get_free_page(GFP_KERNEL);
+		if (p == 0)
+			panic("couldn't get a page in MMU_get_page");
+	} else {
+		if ( is_prep || (_machine == _MACH_chrp) )
+		{
+			mmu_pages_count++;
+			if ( mmu_pages_count > MAX_MMU_PAGES )
+				printk("out of mmu pages!\n");
+			p = (pte *)(PAGE_ALIGN((unsigned long)mmu_pages)+
+				    (mmu_pages_count+PAGE_SIZE));
+		}
+		else /* pmac */
+		{
+			p = find_mem_piece(PAGE_SIZE, PAGE_SIZE);
+		}
+	}
+	memset(p, 0, PAGE_SIZE);
+	return p;
 }
 
-#ifdef CONFIG_PMAC
 void *
 ioremap(unsigned long addr, unsigned long size)
 {
 	unsigned long p, end = addr + size;
 
+	/*
+	 * BAT mappings on prep cover this already so don't waste
+	 * space with it. -- Cort
+	 */
+	if ( is_prep )
+		if ( ((addr >= 0xc0000000) && (end < (0xc0000000+(256<<20)))) ||
+		     ((addr >= 0x80000000) && (end < (0x80000000+(256<<20)))) )
+			return (void *)addr;
 	for (p = addr & PAGE_MASK; p < end; p += PAGE_SIZE)
-		map_page(&init_task.tss, p, p, pgprot_val(PAGE_KERNEL_CI) | _PAGE_GUARDED);
+		map_page(&init_task, p, p, pgprot_val(PAGE_KERNEL_CI) | _PAGE_GUARDED);
 	return (void *) addr;
 }
-#endif
+
+extern void iounmap(unsigned long *addr)
+{
+	/*
+	 * BAT mappings on prep cover this already so don't waste
+	 * space with it. -- Cort
+	 */
+	if ( is_prep )
+		if ( (((unsigned long)addr >= 0xc0000000) && ((unsigned long)addr < (0xc0000000+(256<<20)))) ||
+		     (((unsigned long)addr >= 0x80000000) && ((unsigned long)addr < (0x80000000+(256<<20)))) )
+			return;
+	/* else unmap it */
+}
 
 void
-map_page(struct thread_struct *tss, unsigned long va,
+map_page(struct task_struct *tsk, unsigned long va,
 	 unsigned long pa, int flags)
 {
 	pmd_t *pd;
 	pte_t *pg;
+	
 
-	if (tss->pg_tables == NULL) {
+	if (tsk->mm->pgd == NULL) {
 		/* Allocate upper level page map */
-		tss->pg_tables = (unsigned long *) MMU_get_page();
+		tsk->mm->pgd = (pgd_t *) MMU_get_page();
 	}
 	/* Use upper 10 bits of VA to index the first level map */
-	pd = (pmd_t *) (tss->pg_tables + (va >> PGDIR_SHIFT));
+	pd = (pmd_t *) (tsk->mm->pgd + (va >> PGDIR_SHIFT));
 	if (pmd_none(*pd)) {
 		/* Need to allocate second-level table */
 		pg = (pte_t *) MMU_get_page();
@@ -1084,26 +1077,6 @@ flush_tlb_mm(struct mm_struct *mm)
 	}
 }
 
-
-void
-flush_tlb_page(struct vm_area_struct *vma, unsigned long vmaddr)
-{
-	unsigned vsid;
-	
-	if ( vmaddr < TASK_SIZE) {
-		/*vsid = (vma->vm_mm->context << 4) | (vmaddr >> 28);*/
-		flush_hash_page(vma->vm_mm->context/*vsid*/, vmaddr);
-		/* this is needed on prep at the moment -- don't know why
-		   -- Cort*/
-		MMU_invalidate_page(vma->vm_mm,vmaddr);
-	}
-	else
-	{
-		/*printk("flush_tlb_page() vmaddr > TASK_SIZE %08x\n", vmaddr);*/
-	}
-}
-
-
 /* for each page addr in the range, call MMU_invalidate_page()
    if the range is very large and the hash table is small it might be faster to
    do a search of the hash table and just invalidate pages that are in the range
@@ -1113,15 +1086,10 @@ flush_tlb_page(struct vm_area_struct *vma, unsigned long vmaddr)
 void
 flush_tlb_range(struct mm_struct *mm, unsigned long start, unsigned long end)
 {
-        start &= PAGE_MASK;
+	start &= PAGE_MASK;
 	for (; start < end && start < TASK_SIZE; start += PAGE_SIZE)
 	{
-		/*flush_hash_page(VSID_FROM_CONTEXT( start>>28, mm->context),
-				start );*/
 		flush_hash_page(mm->context, start);
-		/* this is needed on prep at the moment -- don't know why
-		   -- Cort*/
-		MMU_invalidate_page(mm,start);
 	}
 }
 
@@ -1135,78 +1103,18 @@ void
 mmu_context_overflow(void)
 {
 	struct task_struct *tsk;
-	int nr;
 
-	printk(KERN_INFO "mmu_context_overflow\n");
-	for (nr = 0; nr < NR_TASKS; ++nr) {
-		tsk = task[nr];
-		if (tsk && tsk->mm)
+	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);
-	_tlbia();
 	next_mmu_context = 0;
+	/* make sure current always has a context */
+	current->mm->context = MUNGE_CONTEXT(++next_mmu_context);
+	set_context(current->mm->context);
 }
 
-#ifdef CONFIG_PREP
-/*
- * it's not a simple matter to get rid of these and switch to the
- * ones paul is using. it will take some time and thought  -- Cort
- */
-inline void MMU_invalidate_page(struct mm_struct *mm, unsigned long va)
-{
-	int hash, page_index, segment, i, h, _h, api, vsid, perms;
-	PTE *_pte, *slot;
-	int flags = 0;
-	page_index = ((int)va & 0x0FFFF000) >> 12;
-	segment = (unsigned int)va >> 28;
-	api = page_index >> 10;
-	vsid = VSID_FROM_CONTEXT(segment,mm->context);
-	for (_h = 0;  _h < 2;  _h++)
-	{
-		hash = page_index ^ vsid;		
-		if (_h)
-		{
-			hash = ~hash;  /* Secondary hash uses ones-complement */
-		}
-		hash &= 0x3FF | (Hash_mask /*<< 10*/);
-		hash *= 8;  /* Eight entries / hash bucket */
-		_pte = &Hash[hash];
-		for (i = 0;  i < 8;  i++, _pte++)
-		{
-			if (_pte->v && _pte->vsid == vsid && _pte->h == _h && _pte->api == api)
-			{ /* Found it! */
-				_tlbie(va); /* Clear TLB */
-				if (_pte->r) flags |= _PAGE_ACCESSED;
-				if (_pte->c) flags |= _PAGE_DIRTY;
-				_pte->v = 0;
-				return /*(flags)*/;
-			}
-		}
-	}
-	_tlbie(va);
-	return /*(flags)*/;
-}
-#endif
-
-#include <asm/mmu.h>
-void print_mm_info(void)
-{
-	struct _SEGREG s;
-	long a;
-	struct _BATU bu;
-	struct _BATL bl;
-	unsigned long i;
-
-	for ( i = 0x70000000 ; i <= 0x90000000 ; i+= 0x10000000 )
-	{
-		a = get_SR(i);
-		memcpy(&s,&a,4);
-		printk("sr %2d t:%1d ks:%d kp:%d n:%d vsid:%x  %x\n",
-		       i>>28, s.t, s.ks, s.kp, s.n, s.vsid, a);
-	}
-
-	asm("mfspr %0,532; mfspr %1, 533\n" : "=r" (bu), "=r" (bl));
-	printk("bat2 bepi: %0x vs: %1x vp: %1x wimg: %x%x%x%x pp: %1x\n",
-       	       bu.bepi<<17, bu.vs, bu.vp, bl.w, bl.i, bl.m, bl.g, bl.pp);
-}
diff --git a/arch/ppc/pmac_defconfig b/arch/ppc/pmac_defconfig
new file mode 100644
index 000000000..5f34efcd2
--- /dev/null
+++ b/arch/ppc/pmac_defconfig
@@ -0,0 +1,274 @@
+#
+# Automatically generated make config: don't edit
+#
+
+#
+# Platform support
+#
+CONFIG_NATIVE=y
+CONFIG_PMAC=y
+# CONFIG_PREP is not set
+CONFIG_MCOMMON=y
+# CONFIG_M601 is not set
+# CONFIG_M603 is not set
+# CONFIG_M604 is not set
+
+#
+# General setup
+#
+CONFIG_EXPERIMENTAL=y
+CONFIG_MODULES=y
+# CONFIG_MODVERSIONS is not set
+CONFIG_KERNELD=y
+CONFIG_PCI=y
+CONFIG_NET=y
+CONFIG_SYSCTL=y
+CONFIG_SYSVIPC=y
+CONFIG_BINFMT_ELF=y
+CONFIG_KERNEL_ELF=y
+CONFIG_BINFMT_MISC=m
+CONFIG_BINFMT_JAVA=m
+CONFIG_PMAC_CONSOLE=y
+CONFIG_MAC_KEYBOARD=y
+CONFIG_MAC_FLOPPY=y
+CONFIG_PROC_DEVICETREE=y
+CONFIG_XMON=y
+CONFIG_ATY_VIDEO=y
+CONFIG_IMSTT_VIDEO=y
+
+#
+# Plug and Play support
+#
+# CONFIG_PNP is not set
+
+#
+# Floppy, IDE, and other block devices
+#
+# CONFIG_BLK_DEV_FD is not set
+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_IDEDISK=y
+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
+# CONFIG_BLK_DEV_CMD640 is not set
+# CONFIG_BLK_DEV_RZ1000 is not set
+# CONFIG_BLK_DEV_TRITON is not set
+# CONFIG_IDE_CHIPSETS is not set
+
+#
+# Additional Block Devices
+#
+CONFIG_BLK_DEV_LOOP=m
+# CONFIG_BLK_DEV_MD is not set
+CONFIG_BLK_DEV_RAM=y
+CONFIG_BLK_DEV_INITRD=y
+# CONFIG_BLK_DEV_XD is not set
+# CONFIG_BLK_DEV_EZ is not set
+# CONFIG_BLK_DEV_HD is not set
+
+#
+# SCSI support
+#
+CONFIG_SCSI=y
+
+#
+# SCSI support type (disk, tape, CD-ROM)
+#
+CONFIG_BLK_DEV_SD=y
+CONFIG_CHR_DEV_ST=y
+CONFIG_BLK_DEV_SR=y
+CONFIG_BLK_DEV_SR_VENDOR=y
+CONFIG_CHR_DEV_SG=y
+
+#
+# Some SCSI devices (e.g. CD jukebox) support multiple LUNs
+#
+# CONFIG_SCSI_MULTI_LUN is not set
+CONFIG_SCSI_CONSTANTS=y
+
+#
+# SCSI low-level drivers
+#
+# CONFIG_SCSI_7000FASST 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_BUSLOGIC is not set
+# CONFIG_SCSI_DTC3280 is not set
+# CONFIG_SCSI_EATA_DMA is not set
+# CONFIG_SCSI_EATA_PIO is not set
+# CONFIG_SCSI_EATA is not set
+# CONFIG_SCSI_FUTURE_DOMAIN is not set
+# CONFIG_SCSI_GENERIC_NCR5380 is not set
+# CONFIG_SCSI_NCR53C406A is not set
+# CONFIG_SCSI_NCR53C7xx is not set
+# CONFIG_SCSI_NCR53C8XX is not set
+# CONFIG_SCSI_PPA is not set
+# CONFIG_SCSI_PAS16 is not set
+# CONFIG_SCSI_QLOGIC_FAS is not set
+# CONFIG_SCSI_QLOGIC_ISP is not set
+# CONFIG_SCSI_SEAGATE is not set
+# CONFIG_SCSI_DC390T is not set
+# CONFIG_SCSI_T128 is not set
+# CONFIG_SCSI_U14_34F is not set
+# CONFIG_SCSI_ULTRASTOR is not set
+CONFIG_SCSI_MESH=y
+CONFIG_SCSI_MESH_SYNC_RATE=10
+CONFIG_SCSI_MAC53C94=y
+CONFIG_SCSI_QLOGIC_PMAC=m
+
+#
+# Network device support
+#
+
+#
+# Networking options
+#
+# CONFIG_NETLINK is not set
+# CONFIG_FIREWALL is not set
+CONFIG_NET_ALIAS=y
+CONFIG_INET=y
+CONFIG_IP_MULTICAST=y
+# CONFIG_IP_ACCT is not set
+# CONFIG_IP_ROUTER is not set
+CONFIG_NET_IPIP=m
+# CONFIG_IP_MROUTE is not set
+CONFIG_IP_ALIAS=y
+# CONFIG_SYN_COOKIES is not set
+
+#
+# (it is safe to leave these untouched)
+#
+# CONFIG_INET_PCTCP is not set
+CONFIG_INET_RARP=y
+CONFIG_PATH_MTU_DISCOVERY=y
+CONFIG_IP_NOSR=y
+CONFIG_SKB_LARGE=y
+# CONFIG_IPV6 is not set
+
+#
+#  
+#
+# CONFIG_IPX is not set
+# CONFIG_ATALK is not set
+# CONFIG_AX25 is not set
+# CONFIG_X25 is not set
+# CONFIG_LAPB is not set
+# CONFIG_BRIDGE is not set
+# CONFIG_LLC is not set
+# CONFIG_WAN_ROUTER is not set
+CONFIG_NETDEVICES=y
+# CONFIG_ARCNET is not set
+# CONFIG_DUMMY is not set
+# CONFIG_EQUALIZER is not set
+CONFIG_NET_ETHERNET=y
+CONFIG_MACE=y
+CONFIG_DEC_ELCP=m
+# CONFIG_NET_VENDOR_3COM is not set
+# CONFIG_LANCE is not set
+# CONFIG_NET_VENDOR_SMC is not set
+# CONFIG_NET_ISA is not set
+# CONFIG_NET_EISA is not set
+# CONFIG_NET_POCKET is not set
+# CONFIG_FDDI is not set
+# CONFIG_DLCI is not set
+# CONFIG_PLIP is not set
+CONFIG_PPP=m
+
+#
+# CCP compressors for PPP are only built as modules.
+#
+# CONFIG_NET_RADIO is not set
+# CONFIG_SLIP is not set
+# CONFIG_TR is not set
+# CONFIG_SHAPER is not set
+
+#
+# ISDN subsystem
+#
+# CONFIG_ISDN is not set
+
+#
+# CD-ROM drivers (not for SCSI or IDE/ATAPI drives)
+#
+# CONFIG_CD_NO_IDESCSI is not set
+
+#
+# Filesystems
+#
+# CONFIG_QUOTA is not set
+CONFIG_MINIX_FS=m
+CONFIG_EXT2_FS=y
+# CONFIG_FAT_FS is not set
+# CONFIG_MSDOS_FS is not set
+# CONFIG_VFAT_FS is not set
+# CONFIG_UMSDOS_FS is not set
+CONFIG_PROC_FS=y
+CONFIG_NFS_FS=y
+# CONFIG_ROOT_NFS is not set
+# CONFIG_NFSD is not set
+CONFIG_SUNRPC=y
+CONFIG_LOCKD=y
+# CONFIG_SMB_FS is not set
+CONFIG_ISO9660_FS=y
+# CONFIG_HPFS_FS is not set
+# CONFIG_SYSV_FS is not set
+# CONFIG_AFFS_FS is not set
+# CONFIG_ROMFS_FS is not set
+CONFIG_AUTOFS_FS=y
+# CONFIG_UFS_FS is not set
+CONFIG_MAC_PARTITION=y
+
+#
+# Character devices
+#
+CONFIG_VT=y
+CONFIG_VT_CONSOLE=y
+CONFIG_SERIAL=y
+# CONFIG_SERIAL_EXTENDED is not set
+# CONFIG_SERIAL_NONSTANDARD is not set
+# CONFIG_PRINTER is not set
+# CONFIG_MOUSE is not set
+# CONFIG_UMISC is not set
+# CONFIG_QIC02_TAPE is not set
+# CONFIG_FTAPE is not set
+# CONFIG_APM is not set
+# CONFIG_WATCHDOG is not set
+# CONFIG_RTC is not set
+CONFIG_NVRAM=y
+# CONFIG_JOYSTICK is not set
+
+#
+# Sound
+#
+CONFIG_SOUND=m
+# CONFIG_PAS is not set
+# CONFIG_SB is not set
+# CONFIG_ADLIB is not set
+# CONFIG_GUS is not set
+# CONFIG_MPU401 is not set
+# CONFIG_PSS is not set
+# CONFIG_GUS16 is not set
+# CONFIG_GUSMAX is not set
+# CONFIG_MSS is not set
+# CONFIG_SSCAPE is not set
+# CONFIG_TRIX is not set
+# CONFIG_MAD16 is not set
+# CONFIG_CS4232 is not set
+# CONFIG_MAUI is not set
+# CONFIG_YM3812 is not set
+CONFIG_LOWLEVEL_SOUND=y
+# CONFIG_ACI_MIXER is not set
+# CONFIG_AWE32_SYNTH is not set
+# CONFIG_AEDSP16 is not set
+CONFIG_AWACS=y
diff --git a/arch/ppc/prep_defconfig b/arch/ppc/prep_defconfig
new file mode 100644
index 000000000..ea39e83ea
--- /dev/null
+++ b/arch/ppc/prep_defconfig
@@ -0,0 +1,250 @@
+#
+# Automatically generated by make menuconfig: don't edit
+#
+
+#
+# Platform support
+#
+CONFIG_PPC=y
+CONFIG_NATIVE=y
+# CONFIG_PMAC is not set
+CONFIG_PREP=y
+CONFIG_MCOMMON=y
+
+#
+# General setup
+#
+CONFIG_EXPERIMENTAL=y
+CONFIG_MODULES=y
+CONFIG_MODVERSIONS=y
+CONFIG_KERNELD=y
+CONFIG_PCI=y
+CONFIG_PCI_OPTIMIZE=y
+CONFIG_NET=y
+CONFIG_SYSCTL=y
+CONFIG_SYSVIPC=y
+CONFIG_BINFMT_ELF=y
+CONFIG_KERNEL_ELF=y
+# CONFIG_BINFMT_MISC is not set
+# CONFIG_BINFMT_JAVA is not set
+CONFIG_VGA_CONSOLE=y
+
+#
+# Plug and Play support
+#
+# CONFIG_PNP is not set
+
+#
+# Floppy, IDE, and other block devices
+#
+CONFIG_BLK_DEV_FD=y
+CONFIG_BLK_DEV_IDE=y
+# CONFIG_BLK_DEV_HD_IDE is not set
+CONFIG_BLK_DEV_IDEDISK=y
+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
+# CONFIG_BLK_DEV_CMD640 is not set
+# CONFIG_BLK_DEV_RZ1000 is not set
+# CONFIG_BLK_DEV_TRITON is not set
+# CONFIG_IDE_CHIPSETS is not set
+CONFIG_BLK_DEV_LOOP=y
+# CONFIG_BLK_DEV_MD is not set
+CONFIG_BLK_DEV_RAM=y
+CONFIG_BLK_DEV_INITRD=y
+# CONFIG_BLK_DEV_XD is not set
+# CONFIG_BLK_DEV_EZ is not set
+# CONFIG_BLK_DEV_HD is not set
+
+#
+# SCSI support
+#
+CONFIG_SCSI=y
+CONFIG_BLK_DEV_SD=y
+CONFIG_CHR_DEV_ST=y
+CONFIG_BLK_DEV_SR=y
+CONFIG_BLK_DEV_SR_VENDOR=y
+# CONFIG_CHR_DEV_SG is not set
+# CONFIG_SCSI_MULTI_LUN is not set
+CONFIG_SCSI_CONSTANTS=y
+
+#
+# SCSI low-level drivers
+#
+# CONFIG_SCSI_7000FASST 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_BUSLOGIC is not set
+# CONFIG_SCSI_DTC3280 is not set
+# CONFIG_SCSI_EATA_DMA is not set
+# CONFIG_SCSI_EATA_PIO is not set
+# CONFIG_SCSI_EATA is not set
+# CONFIG_SCSI_FUTURE_DOMAIN is not set
+# CONFIG_SCSI_GENERIC_NCR5380 is not set
+# CONFIG_SCSI_NCR53C406A is not set
+# CONFIG_SCSI_NCR53C7xx is not set
+CONFIG_SCSI_NCR53C8XX=y
+# CONFIG_SCSI_NCR53C8XX_NVRAM_DETECT is not set
+CONFIG_SCSI_NCR53C8XX_TAGGED_QUEUE=y
+CONFIG_SCSI_NCR53C8XX_IOMAPPED=y
+CONFIG_SCSI_NCR53C8XX_MAX_TAGS=4
+CONFIG_SCSI_NCR53C8XX_SYNC=5
+# CONFIG_SCSI_NCR53C8XX_SYMBIOS_COMPAT is not set
+# CONFIG_SCSI_PPA is not set
+# CONFIG_SCSI_PAS16 is not set
+# CONFIG_SCSI_QLOGIC_FAS is not set
+# CONFIG_SCSI_QLOGIC_ISP is not set
+# CONFIG_SCSI_SEAGATE is not set
+# CONFIG_SCSI_DC390T is not set
+# CONFIG_SCSI_T128 is not set
+# CONFIG_SCSI_U14_34F is not set
+# CONFIG_SCSI_ULTRASTOR is not set
+# CONFIG_SCSI_MESH is not set
+# CONFIG_SCSI_MAC53C94 is not set
+
+#
+# Network device support
+#
+
+#
+# Networking options
+#
+# CONFIG_NETLINK is not set
+# CONFIG_FIREWALL is not set
+# CONFIG_NET_ALIAS is not set
+CONFIG_INET=y
+# CONFIG_IP_MULTICAST is not set
+# CONFIG_IP_ACCT is not set
+# CONFIG_IP_ROUTER is not set
+# CONFIG_NET_IPIP is not set
+# CONFIG_SYN_COOKIES is not set
+# CONFIG_XTP is not set
+# CONFIG_INET_PCTCP is not set
+# CONFIG_INET_RARP is not set
+CONFIG_PATH_MTU_DISCOVERY=y
+# CONFIG_IP_NOSR is not set
+CONFIG_SKB_LARGE=y
+# CONFIG_IPV6 is not set
+# CONFIG_IPX is not set
+# CONFIG_ATALK is not set
+# CONFIG_AX25 is not set
+# CONFIG_X25 is not set
+# CONFIG_LAPB is not set
+# CONFIG_BRIDGE is not set
+# CONFIG_LLC is not set
+# CONFIG_WAN_ROUTER is not set
+CONFIG_NETDEVICES=y
+# CONFIG_ARCNET is not set
+# CONFIG_DUMMY is not set
+# CONFIG_EQUALIZER is not set
+CONFIG_NET_ETHERNET=y
+CONFIG_NET_VENDOR_3COM=y
+# CONFIG_EL1 is not set
+# CONFIG_EL2 is not set
+# CONFIG_ELPLUS is not set
+# CONFIG_EL16 is not set
+CONFIG_EL3=y
+# CONFIG_VORTEX is not set
+CONFIG_LANCE=y
+# CONFIG_NET_VENDOR_SMC is not set
+# CONFIG_NET_VENDOR_RACAL is not set
+# CONFIG_NET_ISA is not set
+CONFIG_NET_EISA=y
+CONFIG_PCNET32=y
+# CONFIG_AC3200 is not set
+# CONFIG_APRICOT is not set
+# CONFIG_CS89x0 is not set
+CONFIG_DE4X5=y
+# CONFIG_DEC_ELCP is not set
+# CONFIG_DGRS is not set
+# CONFIG_EEXPRESS_PRO100 is not set
+# CONFIG_TLAN is not set
+# CONFIG_ES3210 is not set
+# CONFIG_ZNET is not set
+# CONFIG_NET_POCKET is not set
+# CONFIG_FDDI is not set
+# CONFIG_DLCI is not set
+# CONFIG_PLIP is not set
+CONFIG_PPP=y
+# CONFIG_NET_RADIO is not set
+# CONFIG_SLIP is not set
+# CONFIG_TR is not set
+# CONFIG_SHAPER is not set
+
+#
+# ISDN subsystem
+#
+# CONFIG_ISDN is not set
+
+#
+# CD-ROM drivers (not for SCSI or IDE/ATAPI drives)
+#
+# CONFIG_CD_NO_IDESCSI is not set
+
+#
+# Filesystems
+#
+# CONFIG_QUOTA is not set
+# CONFIG_MINIX_FS is not set
+CONFIG_EXT2_FS=y
+CONFIG_FAT_FS=y
+CONFIG_MSDOS_FS=y
+# CONFIG_VFAT_FS is not set
+# CONFIG_UMSDOS_FS is not set
+CONFIG_PROC_FS=y
+CONFIG_NFS_FS=y
+# CONFIG_ROOT_NFS is not set
+CONFIG_NFSD=y
+CONFIG_SUNRPC=y
+CONFIG_LOCKD=y
+# CONFIG_SMB_FS is not set
+CONFIG_ISO9660_FS=y
+# CONFIG_HPFS_FS is not set
+# CONFIG_SYSV_FS is not set
+# CONFIG_AFFS_FS is not set
+# CONFIG_ROMFS_FS is not set
+# CONFIG_AUTOFS_FS is not set
+# CONFIG_UFS_FS is not set
+CONFIG_MAC_PARTITION=y
+
+#
+# Character devices
+#
+CONFIG_VT=y
+CONFIG_VT_CONSOLE=y
+# CONFIG_SOFTCURSOR is not set
+CONFIG_SERIAL=y
+CONFIG_SERIAL_EXTENDED=y
+# CONFIG_SERIAL_MANY_PORTS is not set
+# CONFIG_SERIAL_SHARE_IRQ is not set
+# CONFIG_SERIAL_MULTIPORT is not set
+# CONFIG_HUB6 is not set
+CONFIG_SERIAL_CONSOLE=y
+# CONFIG_SERIAL_NONSTANDARD is not set
+# CONFIG_PRINTER is not set
+CONFIG_MOUSE=y
+# CONFIG_ATIXL_BUSMOUSE is not set
+# CONFIG_BUSMOUSE is not set
+# CONFIG_MS_BUSMOUSE is not set
+CONFIG_PSMOUSE=y
+# CONFIG_82C710_MOUSE is not set
+# CONFIG_PC110_PAD is not set
+# CONFIG_UMISC is not set
+# CONFIG_QIC02_TAPE is not set
+# CONFIG_FTAPE is not set
+# CONFIG_APM is not set
+# CONFIG_WATCHDOG is not set
+# CONFIG_RTC is not set
+# CONFIG_NVRAM is not set
+# CONFIG_JOYSTICK is not set
+
+#
+# Sound
+#
+# CONFIG_SOUND is not set
diff --git a/arch/ppc/vmlinux.lds b/arch/ppc/vmlinux.lds
new file mode 100644
index 000000000..1a51a4a7e
--- /dev/null
+++ b/arch/ppc/vmlinux.lds
@@ -0,0 +1,85 @@
+OUTPUT_ARCH(powerpc)
+SEARCH_DIR(/lib); SEARCH_DIR(/usr/lib); SEARCH_DIR(/usr/local/lib); SEARCH_DIR(/usr/local/powerpc-any-elf/lib);
+/* Do we need any of these for elf?
+   __DYNAMIC = 0;    */
+SECTIONS
+{
+  /* Read-only sections, merged into text segment: */
+  . = + SIZEOF_HEADERS;
+  .interp : { *(.interp) }
+  .hash          : { *(.hash)		}
+  .dynsym        : { *(.dynsym)		}
+  .dynstr        : { *(.dynstr)		}
+  .rel.text      : { *(.rel.text)		}
+  .rela.text     : { *(.rela.text) 	}
+  .rel.data      : { *(.rel.data)		}
+  .rela.data     : { *(.rela.data) 	}
+  .rel.rodata    : { *(.rel.rodata) 	}
+  .rela.rodata   : { *(.rela.rodata) 	}
+  .rel.got       : { *(.rel.got)		}
+  .rela.got      : { *(.rela.got)		}
+  .rel.ctors     : { *(.rel.ctors)	}
+  .rela.ctors    : { *(.rela.ctors)	}
+  .rel.dtors     : { *(.rel.dtors)	}
+  .rela.dtors    : { *(.rela.dtors)	}
+  .rel.bss       : { *(.rel.bss)		}
+  .rela.bss      : { *(.rela.bss)		}
+  .rel.plt       : { *(.rel.plt)		}
+  .rela.plt      : { *(.rela.plt)		}
+/*  .init          : { *(.init)	} =0*/
+  .plt : { *(.plt) }
+  .text      :
+  {
+    *(.text)
+    *(.fixup)
+    *(.got1)
+  }
+  _etext = .;
+  PROVIDE (etext = .);
+  .rodata    :
+  {
+    *(.rodata)
+    *(.rodata1)
+  }
+  .fini      : { *(.fini)    } =0
+  .ctors     : { *(.ctors)   }
+  .dtors     : { *(.dtors)   }
+  /* Read-write section, merged into data segment: */
+  . = (. + 0x0FFF) & 0xFFFFF000;
+  .data    :
+  {
+    *(.data)
+    *(.data1)
+    *(.sdata)
+    *(.sdata2)
+    *(.got.plt) *(.got)
+    *(.dynamic)
+    CONSTRUCTORS
+  }
+  _edata  =  .;
+  PROVIDE (edata = .);
+
+  .fixup   : { *(.fixup) }
+  __start___ex_table = .;
+  __ex_table : { *(__ex_table) }
+  __stop___ex_table = .;
+
+  . = ALIGN(4096);
+  __init_begin = .;
+  .text.init : { *(.text.init) }
+  .data.init : { *(.data.init) }
+  . = ALIGN(4096);
+  __init_end = .;
+
+  __bss_start = .;
+  .bss       :
+  {
+   *(.sbss) *(.scommon)
+   *(.dynbss)
+   *(.bss)
+   *(COMMON)
+  }
+  _end = . ;
+  PROVIDE (end = .);
+}
+
diff --git a/arch/sparc/ap1000/aplib.c b/arch/sparc/ap1000/aplib.c
index 55a778ed7..8bd67e530 100644
--- a/arch/sparc/ap1000/aplib.c
+++ b/arch/sparc/ap1000/aplib.c
@@ -455,7 +455,8 @@ static inline int aplib_poll(unsigned counter)
 
 	while (counter == aplib->rbuf_flag1 + aplib->rbuf_flag2) {
 		tnet_check_completion();
-		if (need_resched) break;
+		if (resched_needed())
+			break;
 		if (current->signal & ~current->blocked) break;
 	}
 	return 0;
diff --git a/arch/sparc/ap1000/mpp.c b/arch/sparc/ap1000/mpp.c
index 84ef7f28e..3465e311b 100644
--- a/arch/sparc/ap1000/mpp.c
+++ b/arch/sparc/ap1000/mpp.c
@@ -28,7 +28,7 @@ static int last_task = 0;
 void mpp_schedule(struct cap_request *req)
 {
 	mpp_current_task = req->data[0];
-	need_resched = 1;
+	resched_force();
 	mark_bh(TQUEUE_BH);
 }
 
diff --git a/arch/sparc/ap1000/msc.c b/arch/sparc/ap1000/msc.c
index 242df954c..fc8aaf6e5 100644
--- a/arch/sparc/ap1000/msc.c
+++ b/arch/sparc/ap1000/msc.c
@@ -338,7 +338,7 @@ static inline void qbmful_interrupt(void)
 #endif
 		MSC_OUT(MSC_INTR, AP_SET_INTR_MASK << MSC_INTR_QBMFUL_SH);
 		intr_mask |= (AP_INTR_REQ << MSC_INTR_QBMFUL_SH);
-		need_resched = 1;
+		resched_force();
 		block_parallel_tasks = 1;
 		mark_bh(TQUEUE_BH);
 	}
diff --git a/arch/sparc/ap1000/tnet.c b/arch/sparc/ap1000/tnet.c
index 14a29ae7a..293be0d5b 100644
--- a/arch/sparc/ap1000/tnet.c
+++ b/arch/sparc/ap1000/tnet.c
@@ -613,7 +613,7 @@ void tnet_send_ip(int cid,struct sk_buff *skb)
 
 static void reschedule(void)
 {
-	need_resched = 1;  
+	resched_force();
 	mark_bh(TQUEUE_BH);
 }
 
diff --git a/arch/sparc/config.in b/arch/sparc/config.in
index 69ed62422..fa8267db1 100644
--- a/arch/sparc/config.in
+++ b/arch/sparc/config.in
@@ -1,4 +1,4 @@
-# $Id: config.in,v 1.36 1997/06/17 03:54:47 davem Exp $
+# $Id: config.in,v 1.38 1997/09/04 01:54:33 davem Exp $
 # For a description of the syntax of this configuration file,
 # see the Configure script.
 #
@@ -53,9 +53,9 @@ bool 'System V IPC' CONFIG_SYSVIPC
 bool 'Sysctl support' CONFIG_SYSCTL
 tristate 'Kernel support for a.out binaries' CONFIG_BINFMT_AOUT
 tristate 'Kernel support for ELF binaries' CONFIG_BINFMT_ELF
+tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
 if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
-  tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
-  tristate 'Kernel support for JAVA binaries' CONFIG_BINFMT_JAVA
+  tristate 'Kernel support for JAVA binaries (obsolete)' CONFIG_BINFMT_JAVA
 fi
 endmenu
 
@@ -146,6 +146,10 @@ fi
 source fs/Config.in
 
 mainmenu_option next_comment
+comment 'Watchdog'
+
+bool 'Software watchdog' CONFIG_SOFT_WATCHDOG
+mainmenu_option next_comment
 comment 'Kernel hacking'
 
 bool 'Kernel profiling support' CONFIG_PROFILE
diff --git a/arch/sparc/defconfig b/arch/sparc/defconfig
index 80e114585..434d95d2b 100644
--- a/arch/sparc/defconfig
+++ b/arch/sparc/defconfig
@@ -49,6 +49,7 @@ SUN_FB_CREATOR=y
 #
 CONFIG_SUN_OPENPROMIO=m
 CONFIG_SUN_MOSTEK_RTC=y
+# CONFIG_SAB82532 is not set
 # CONFIG_SUN_BPP is not set
 # CONFIG_SUN_VIDEOPIX is not set
 
@@ -105,6 +106,7 @@ CONFIG_IP_MASQUERADE=y
 CONFIG_IP_ALIAS=m
 # CONFIG_ARPD is not set
 # CONFIG_SYN_COOKIES is not set
+# CONFIG_XTP is not set
 
 #
 # (it is safe to leave these untouched)
@@ -184,7 +186,7 @@ CONFIG_MINIX_FS=m
 CONFIG_EXT2_FS=y
 CONFIG_FAT_FS=m
 CONFIG_MSDOS_FS=m
-CONFIG_VFAT_FS=m
+# CONFIG_VFAT_FS is not set
 # CONFIG_UMSDOS_FS is not set
 CONFIG_PROC_FS=y
 CONFIG_NFS_FS=y
@@ -207,6 +209,7 @@ CONFIG_AMIGA_PARTITION=y
 CONFIG_UFS_FS=y
 CONFIG_BSD_DISKLABEL=y
 CONFIG_SMD_DISKLABEL=y
+# CONFIG_MAC_PARTITION is not set
 
 #
 # Kernel hacking
diff --git a/arch/sparc/kernel/devices.c b/arch/sparc/kernel/devices.c
index 3843a7aa8..b9c6495cf 100644
--- a/arch/sparc/kernel/devices.c
+++ b/arch/sparc/kernel/devices.c
@@ -53,8 +53,33 @@ device_scan(unsigned long mem_start))
 			}
 		};
 		if(cpu_ctr == 0) {
+			if (sparc_cpu_model == sun4d) {
+				scan = prom_getchild(prom_root_node);
+				for (scan = prom_searchsiblings(scan, "cpu-unit"); scan;
+				     scan = prom_searchsiblings(prom_getsibling(scan), "cpu-unit")) {
+					int node = prom_getchild(scan);
+					
+					prom_getstring(node, "device_type", node_str, sizeof(node_str));
+					if (strcmp(node_str, "cpu") == 0) {
+						prom_getproperty(node, "cpu-id", (char *) &thismid, sizeof(thismid));
+						if (cpu_ctr < NCPUS) {
+							cpu_nds[cpu_ctr] = node;
+							linux_cpus[cpu_ctr].prom_node = node;
+							linux_cpus[cpu_ctr].mid = thismid;
+						}
+						prom_printf("Found CPU %d <node=%08lx,mid=%d>\n",
+							    cpu_ctr, (unsigned long) node,
+							    thismid);
+						cpu_ctr++;
+					}
+				}
+			}
+			if (cpu_ctr > NCPUS)
+				cpu_ctr = NCPUS;
+		}
+		if(cpu_ctr == 0) {
 			printk("No CPU nodes found, cannot continue.\n");
-			/* Probably a sun4d or sun4e, Sun is trying to trick us ;-) */
+			/* Probably a sun4e, Sun is trying to trick us ;-) */
 			halt();
 		}
 		printk("Found %d CPU prom device tree node(s).\n", cpu_ctr);
diff --git a/arch/sparc/kernel/head.S b/arch/sparc/kernel/head.S
index 90965e26e..5742453ff 100644
--- a/arch/sparc/kernel/head.S
+++ b/arch/sparc/kernel/head.S
@@ -1,4 +1,4 @@
-/* $Id: head.S,v 1.82 1997/05/01 01:40:38 davem Exp $
+/* $Id: head.S,v 1.83 1997/08/28 11:10:39 jj Exp $
  * head.S: The initial boot code for the Sparc port of Linux.
  *
  * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
@@ -827,22 +827,18 @@ got_prop:
 
 		cmp	%l1, 'c'		! We already know we are not
 		be	1f			! on a plain sun4 because of
-		 nop				! the check for 0x4000 in %o0
-
-		cmp	%l1, 'm'		! at start
+		 				! the check for 0x4000 in %o0
+		 cmp	%l1, 'm'		! at start
 		be	1f
-		 nop
-
-		cmp	%l1, 'e'
+		 cmp	%l1, 'd'
+		be	1f
+		 cmp	%l1, 'e'
 		be	no_sun4e_here		! Could be a sun4e.
 		 nop
-
-		b	no_sun4u_here		! AIEEE, a V9 sun4u...
+		b	no_sun4u_here		! AIEEE, a V9 sun4u... Get our BIG BROTHER kernel :))
 		 nop
 
-
-1:
-		set	C_LABEL(cputypval), %l1
+1:		set	C_LABEL(cputypval), %l1
 		ldub	[%l1 + 0x4], %l1
 		cmp	%l1, 'm'		! Test for sun4d, sun4e ?
 		be	sun4m_init
diff --git a/arch/sparc/kernel/process.c b/arch/sparc/kernel/process.c
index 95eed6287..ff4a3f192 100644
--- a/arch/sparc/kernel/process.c
+++ b/arch/sparc/kernel/process.c
@@ -1,4 +1,4 @@
-/*  $Id: process.c,v 1.99 1997/07/17 02:20:13 davem Exp $
+/*  $Id: process.c,v 1.100 1997/08/10 04:49:23 davem Exp $
  *  linux/arch/sparc/kernel/process.c
  *
  *  Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
@@ -105,6 +105,8 @@ out:
 /* This is being executed in task 0 'user space'. */
 int cpu_idle(void *unused)
 {
+	extern volatile int smp_commenced;
+
 	current->priority = -100;
 	while(1) {
 		/*
@@ -118,7 +120,8 @@ int cpu_idle(void *unused)
 		}
 		/* endless idle loop with no priority at all */
 		current->counter = -100;
-		schedule();
+		if(!smp_commenced || resched_needed())
+			schedule();
 	}
 }
 
diff --git a/arch/sparc/kernel/ptrace.c b/arch/sparc/kernel/ptrace.c
index 732b3006d..933a5acfa 100644
--- a/arch/sparc/kernel/ptrace.c
+++ b/arch/sparc/kernel/ptrace.c
@@ -886,7 +886,7 @@ asmlinkage void syscall_trace(void)
 	current->exit_code = SIGTRAP;
 	current->state = TASK_STOPPED;
 	current->tss.flags ^= MAGIC_CONSTANT;
-	notify_parent(current);
+	notify_parent(current, SIGCHLD);
 	schedule();
 	/*
 	 * this isn't the same as continuing with a signal, but it will do
diff --git a/arch/sparc/kernel/rtrap.S b/arch/sparc/kernel/rtrap.S
index d8d160114..8d46d487e 100644
--- a/arch/sparc/kernel/rtrap.S
+++ b/arch/sparc/kernel/rtrap.S
@@ -1,4 +1,4 @@
-/* $Id: rtrap.S,v 1.46 1997/04/01 02:21:48 davem Exp $
+/* $Id: rtrap.S,v 1.47 1997/08/10 04:49:24 davem Exp $
  * rtrap.S: Return from Sparc trap low-level code.
  *
  * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
@@ -65,12 +65,14 @@ C_LABEL(ret_trap_lockless_ipi):
 
 	wr	%t_psr, 0x0, %psr
 	b	ret_trap_kernel
-	 nop
+	 mov	1, %o0
 
 1:
+	ld	[%curptr + AOFF_task_processor], %o1
 	ld	[%twin_tmp1 + %lo(C_LABEL(need_resched))], %g2
+	sll	%o0, %o1, %o0
 
-	cmp	%g2, 0
+	andcc	%g2, %o0, %g0
 	be	signal_p
 	 nop
 
diff --git a/arch/sparc/kernel/signal.c b/arch/sparc/kernel/signal.c
index fd5fa634e..fbb5578fa 100644
--- a/arch/sparc/kernel/signal.c
+++ b/arch/sparc/kernel/signal.c
@@ -1,4 +1,4 @@
-/*  $Id: signal.c,v 1.74 1997/05/15 19:57:09 davem Exp $
+/*  $Id: signal.c,v 1.75 1997/08/05 19:19:26 davem Exp $
  *  linux/arch/sparc/kernel/signal.c
  *
  *  Copyright (C) 1991, 1992  Linus Torvalds
@@ -722,7 +722,7 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs * regs,
 			/* This happens to be SMP safe so no need to
 			 * grab master kernel lock even in this case.
 			 */
-			notify_parent(current);
+			notify_parent(current, SIGCHLD);
 			schedule();
 			if (!(signr = current->exit_code))
 				continue;
@@ -773,7 +773,7 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs * regs,
 				/* notify_parent() is SMP safe */
 				if(!(current->p_pptr->sig->action[SIGCHLD-1].sa_flags &
 				     SA_NOCLDSTOP))
-					notify_parent(current);
+					notify_parent(current, SIGCHLD);
 				schedule();
 				continue;
 
diff --git a/arch/sparc/kernel/smp.c b/arch/sparc/kernel/smp.c
index beef2df14..17931fd55 100644
--- a/arch/sparc/kernel/smp.c
+++ b/arch/sparc/kernel/smp.c
@@ -87,7 +87,7 @@ volatile int smp_process_available=0;
 #define SMP_PRINTK(x)
 #endif
 
-static volatile int smp_commenced = 0;
+volatile int smp_commenced = 0;
 
 static char smp_buf[512];
 
@@ -558,7 +558,7 @@ void smp_flush_sig_insns(struct mm_struct *mm, unsigned long insn_addr)
 /* Reschedule call back. */
 void smp_reschedule_irq(void)
 {
-	need_resched=1;
+	resched_force();
 }
 
 /* Running cross calls. */
@@ -623,7 +623,7 @@ void smp_percpu_timer_interrupt(struct pt_regs *regs)
 
 			if(--current->counter < 0) {
 				current->counter = 0;
-				need_resched = 1;
+				resched_force();
 			}
 
 			spin_lock(&ticker_lock);
diff --git a/arch/sparc/mm/srmmu.c b/arch/sparc/mm/srmmu.c
index ad9aa8953..8c3358a64 100644
--- a/arch/sparc/mm/srmmu.c
+++ b/arch/sparc/mm/srmmu.c
@@ -1,4 +1,4 @@
-/* $Id: srmmu.c,v 1.149 1997/07/20 05:59:34 davem Exp $
+/* $Id: srmmu.c,v 1.151 1997/08/28 11:10:54 jj Exp $
  * srmmu.c:  SRMMU specific routines for memory management.
  *
  * Copyright (C) 1995 David S. Miller  (davem@caip.rutgers.edu)
@@ -2103,14 +2103,18 @@ unsigned long srmmu_paging_init(unsigned long start_mem, unsigned long end_mem)
 	sparc_iobase_vaddr = 0xfd000000;    /* 16MB of IOSPACE on all sun4m's. */
 	physmem_mapped_contig = 0;	    /* for init.c:taint_real_pages()   */
 
-	/* Find the number of contexts on the srmmu. */
-	cpunode = prom_getchild(prom_root_node);
-	num_contexts = 0;
-	while((cpunode = prom_getsibling(cpunode)) != 0) {
-		prom_getstring(cpunode, "device_type", node_str, sizeof(node_str));
-		if(!strcmp(node_str, "cpu")) {
-			num_contexts = prom_getintdefault(cpunode, "mmu-nctx", 0x8);
-			break;
+	if (sparc_cpu_model == sun4d)
+		num_contexts = 65536; /* We now it is Viking */
+	else {
+		/* Find the number of contexts on the srmmu. */
+		cpunode = prom_getchild(prom_root_node);
+		num_contexts = 0;
+		while((cpunode = prom_getsibling(cpunode)) != 0) {
+			prom_getstring(cpunode, "device_type", node_str, sizeof(node_str));
+			if(!strcmp(node_str, "cpu")) {
+				num_contexts = prom_getintdefault(cpunode, "mmu-nctx", 0x8);
+				break;
+			}
 		}
 	}
 
@@ -2601,7 +2605,7 @@ __initfunc(static void init_swift(void))
 }
 
 /* turbosparc.S */
-extern void turbosparc_flush_cache_all();
+extern void turbosparc_flush_cache_all(void);
 extern void turbosparc_flush_sig_insns(struct mm_struct *mm, unsigned long insn_addr);
 
 static void poke_turbosparc(void)
diff --git a/arch/sparc64/Makefile b/arch/sparc64/Makefile
index b8cf06878..20b90a651 100644
--- a/arch/sparc64/Makefile
+++ b/arch/sparc64/Makefile
@@ -1,4 +1,4 @@
-# $Id: Makefile,v 1.20 1997/07/11 11:05:29 jj Exp $
+# $Id: Makefile,v 1.22 1997/08/29 15:51:53 jj Exp $
 # sparc64/Makefile
 #
 # Makefile for the architecture dependent flags and dependencies on the
@@ -17,7 +17,7 @@ LD		= sparc64-linux-ld
 NM		= sparc64-linux-nm
 AR		= sparc64-linux-ar
 RANLIB		= sparc64-linux-ranlib
-ELF2AOUT64	= elf2aout64
+ELFTOAOUT	= elftoaout
 
 #
 # Uncomment the first CFLAGS if you are doing kgdb source level
@@ -34,13 +34,21 @@ HEAD := arch/sparc64/kernel/head.o arch/sparc64/kernel/init_task.o
 SUBDIRS := $(SUBDIRS) arch/sparc64/kernel arch/sparc64/lib arch/sparc64/mm \
 	arch/sparc64/prom
 
+ifneq ($(CONFIG_SOLARIS_EMUL),n)
+  SUBDIRS += arch/sparc64/solaris
+endif
+
 CORE_FILES := arch/sparc64/kernel/kernel.o arch/sparc64/mm/mm.o $(CORE_FILES)
 
+ifeq ($(CONFIG_SOLARIS_EMUL),y)
+  CORE_FILES += arch/sparc64/solaris/solaris.o
+endif
+
 LIBS := $(TOPDIR)/lib/lib.a $(LIBS) $(TOPDIR)/arch/sparc64/prom/promlib.a \
 	$(TOPDIR)/arch/sparc64/lib/lib.a
 
 vmlinux.aout: vmlinux
-	$(ELF2AOUT64) -o $(TOPDIR)/vmlinux.aout $(TOPDIR)/vmlinux
+	$(ELFTOAOUT) -o $(TOPDIR)/vmlinux.aout $(TOPDIR)/vmlinux
 
 archclean:
 	rm -f $(TOPDIR)/vmlinux.aout
diff --git a/arch/sparc64/boot/Makefile b/arch/sparc64/boot/Makefile
index ed3fc6cdb..4f2b9ead3 100644
--- a/arch/sparc64/boot/Makefile
+++ b/arch/sparc64/boot/Makefile
@@ -1,4 +1,4 @@
-# $Id: Makefile,v 1.1 1997/07/18 06:26:30 ralf Exp $
+# $Id: Makefile,v 1.3 1997/08/29 11:08:34 davem Exp $
 # Makefile for the Sparc64 boot stuff.
 #
 # Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
diff --git a/arch/sparc64/config.in b/arch/sparc64/config.in
index fcbac5c1a..c3c438dc6 100644
--- a/arch/sparc64/config.in
+++ b/arch/sparc64/config.in
@@ -1,4 +1,4 @@
-# $Id: config.in,v 1.9 1997/07/04 11:33:05 davem Exp $
+# $Id: config.in,v 1.17 1997/09/04 01:54:43 davem Exp $
 # For a description of the syntax of this configuration file,
 # see the Configure script.
 #
@@ -43,6 +43,7 @@ else
 	define_bool CONFIG_SUN_CONSOLE y
 	define_bool CONFIG_SUN_AUXIO y
 	define_bool CONFIG_SUN_IO y
+	define_bool CONFIG_PCI y
 	source drivers/sbus/char/Config.in
 fi
 
@@ -57,10 +58,11 @@ if [ "$CONFIG_SPARC32_COMPAT" != "n" ]; then
   bool 'Kernel support for 32-bit (ie. SunOS) a.out binaries' CONFIG_BINFMT_AOUT32
 fi
 if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
-  tristate 'Kernel support for JAVA binaries' CONFIG_BINFMT_JAVA
+  tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
+  tristate 'Kernel support for JAVA binaries (obsolete)' CONFIG_BINFMT_JAVA
 fi
 if [ "$CONFIG_EXPERIMENTAL" = "y" ]; then
-  tristate 'Kernel support for MISC binaries' CONFIG_BINFMT_MISC
+  tristate 'Solaris binary emulation' CONFIG_SOLARIS_EMUL
 fi
 endmenu
 
@@ -83,6 +85,19 @@ fi
 
 tristate 'Loopback device support' CONFIG_BLK_DEV_LOOP
 
+if [ "$CONFIG_PCI" = "y" ]; then
+	tristate 'Ultra/PCI IDE disk/cdrom/tape/floppy support' CONFIG_BLK_DEV_IDE
+	if [ "$CONFIG_BLK_DEV_IDE" = "y" ]; then
+	  dep_tristate '   Include IDE/ATA-2 DISK support' CONFIG_BLK_DEV_IDEDISK $CONFIG_BLK_DEV_IDE
+	  dep_tristate '   Include IDE/ATAPI CDROM support' CONFIG_BLK_DEV_IDECD $CONFIG_BLK_DEV_IDE
+	  dep_tristate '   Include IDE/ATAPI TAPE support' CONFIG_BLK_DEV_IDETAPE $CONFIG_BLK_DEV_IDE
+	  dep_tristate '   Include IDE/ATAPI FLOPPY support' CONFIG_BLK_DEV_IDEFLOPPY $CONFIG_BLK_DEV_IDE
+	  dep_tristate '   SCSI emulation support' CONFIG_BLK_DEV_IDESCSI $CONFIG_BLK_DEV_IDE
+	  define_bool CONFIG_IDE_CHIPSETS y
+	  define_bool CONFIG_BLK_DEV_NS87415 y
+	fi
+fi
+
 endmenu
 
 if [ "$CONFIG_NET" = "y" ]; then
@@ -116,6 +131,21 @@ if [ "$CONFIG_SCSI" != "n" ]; then
 
 	bool 'Sparc ESP Scsi Driver' CONFIG_SCSI_SUNESP $CONFIG_SCSI
 	tristate 'PTI Qlogic,ISP Driver' CONFIG_SCSI_QLOGICPTI $CONFIG_SCSI
+
+	if [ "$CONFIG_PCI" != "n" ]; then
+	  dep_tristate 'Adaptec AIC7xxx support' CONFIG_SCSI_AIC7XXX $CONFIG_SCSI
+	  if [ "$CONFIG_SCSI_AIC7XXX" != "n" ]; then
+	    bool '   Enable tagged command queueing' CONFIG_AIC7XXX_TAGGED_QUEUEING Y
+	    dep_tristate '   Override driver defaults for commands per LUN' CONFIG_OVERRIDE_CMDS N
+	    if [ "$CONFIG_OVERRIDE_CMDS" != "n" ]; then
+	      int  '   Maximum number of commands per LUN' CONFIG_AIC7XXX_CMDS_PER_LUN 8
+	    fi
+	    bool '   Enable SCB paging' CONFIG_AIC7XXX_PAGE_ENABLE N
+	    bool '   Collect statistics to report in /proc' CONFIG_AIC7XXX_PROC_STATS N
+	    int  '   Delay in seconds after SCSI bus reset' CONFIG_AIC7XXX_RESET_DELAY 15
+	  fi
+	fi
+
 	endmenu
 fi
 endmenu
@@ -141,6 +171,9 @@ if [ "$CONFIG_NET" = "y" ]; then
 		tristate 'Sun Happy Meal 10/100baseT support' CONFIG_HAPPYMEAL
 		tristate 'Sun QuadEthernet support' CONFIG_SUNQE
 		tristate 'MyriCOM Gigabit Ethernet support' CONFIG_MYRI_SBUS
+		if [ "$CONFIG_PCI" = "y" ]; then
+			tristate 'Generic DECchip & DIGITAL EtherWORKS PCI/EISA' CONFIG_DE4X5
+		fi
 #		bool 'FDDI driver support' CONFIG_FDDI
 #		if [ "$CONFIG_FDDI" = "y" ]; then
 #		fi
diff --git a/arch/sparc64/defconfig b/arch/sparc64/defconfig
index c9c92b987..85958cd6e 100644
--- a/arch/sparc64/defconfig
+++ b/arch/sparc64/defconfig
@@ -29,6 +29,7 @@ CONFIG_SUN_KEYBOARD=y
 CONFIG_SUN_CONSOLE=y
 CONFIG_SUN_AUXIO=y
 CONFIG_SUN_IO=y
+CONFIG_PCI=y
 
 #
 # SBUS Frame Buffer support
@@ -49,6 +50,7 @@ SUN_FB_CREATOR=y
 #
 CONFIG_SUN_OPENPROMIO=m
 CONFIG_SUN_MOSTEK_RTC=y
+# CONFIG_SAB82532 is not set
 # CONFIG_SUN_BPP is not set
 # CONFIG_SUN_VIDEOPIX is not set
 CONFIG_SUN_OPENPROMFS=m
@@ -61,17 +63,26 @@ CONFIG_BINFMT_ELF32=y
 CONFIG_BINFMT_AOUT32=y
 CONFIG_BINFMT_JAVA=m
 CONFIG_BINFMT_MISC=m
+CONFIG_SOLARIS_EMUL=m
 
 #
 # Floppy, IDE, and other block devices
 #
-# CONFIG_BLK_DEV_FD is not set
+CONFIG_BLK_DEV_FD=y
 CONFIG_BLK_DEV_MD=y
 CONFIG_MD_LINEAR=m
 CONFIG_MD_STRIPED=m
 CONFIG_BLK_DEV_RAM=y
 CONFIG_BLK_DEV_INITRD=y
 CONFIG_BLK_DEV_LOOP=m
+CONFIG_BLK_DEV_IDE=y
+CONFIG_BLK_DEV_IDEDISK=y
+CONFIG_BLK_DEV_IDECD=y
+CONFIG_BLK_DEV_IDETAPE=m
+CONFIG_BLK_DEV_IDEFLOPPY=m
+# CONFIG_BLK_DEV_IDESCSI is not set
+CONFIG_IDE_CHIPSETS=y
+CONFIG_BLK_DEV_NS87415=y
 
 #
 # Networking options
@@ -85,12 +96,13 @@ CONFIG_INET=y
 # CONFIG_IP_ROUTER is not set
 # CONFIG_NET_IPIP is not set
 # CONFIG_SYN_COOKIES is not set
+# CONFIG_XTP is not set
 
 #
 # (it is safe to leave these untouched)
 #
 # CONFIG_INET_PCTCP is not set
-# CONFIG_INET_RARP is not set
+CONFIG_INET_RARP=m
 CONFIG_PATH_MTU_DISCOVERY=y
 CONFIG_IP_NOSR=y
 CONFIG_SKB_LARGE=y
@@ -136,6 +148,12 @@ CONFIG_SCSI_CONSTANTS=y
 #
 CONFIG_SCSI_SUNESP=y
 CONFIG_SCSI_QLOGICPTI=m
+CONFIG_SCSI_AIC7XXX=y
+# CONFIG_AIC7XXX_TAGGED_QUEUEING is not set
+# CONFIG_OVERRIDE_CMDS is not set
+# CONFIG_AIC7XXX_PAGE_ENABLE is not set
+# CONFIG_AIC7XXX_PROC_STATS is not set
+CONFIG_AIC7XXX_RESET_DELAY=5
 
 #
 # Network device support
@@ -155,6 +173,7 @@ CONFIG_SUNLANCE=y
 CONFIG_HAPPYMEAL=y
 CONFIG_SUNQE=m
 CONFIG_MYRI_SBUS=m
+CONFIG_DE4X5=y
 
 #
 # Filesystems
@@ -187,6 +206,7 @@ CONFIG_AMIGA_PARTITION=y
 CONFIG_UFS_FS=m
 CONFIG_BSD_DISKLABEL=y
 CONFIG_SMD_DISKLABEL=y
+# CONFIG_MAC_PARTITION is not set
 
 #
 # Kernel hacking
diff --git a/arch/sparc64/kernel/Makefile b/arch/sparc64/kernel/Makefile
index aecb9fd47..2b07a0e15 100644
--- a/arch/sparc64/kernel/Makefile
+++ b/arch/sparc64/kernel/Makefile
@@ -1,4 +1,4 @@
-# $Id: Makefile,v 1.30 1997/07/24 14:48:04 davem Exp $
+# $Id: Makefile,v 1.34 1997/08/12 04:12:36 ecd Exp $
 # Makefile for the linux kernel.
 #
 # Note! Dependencies are done automagically by 'make dep', which also
@@ -29,9 +29,10 @@ all: kernel.o head.o init_task.o
 
 O_TARGET := kernel.o
 O_OBJS   := process.o setup.o cpu.o idprom.o \
-	    systbls.o traps.o devices.o auxio.o ioport.o \
+	    traps.o devices.o auxio.o ioport.o \
 	    irq.o ptrace.o time.o sys_sparc.o signal.o \
-	    unaligned.o sys_sunos32.o sunos_ioctl32.o
+	    unaligned.o sys_sunos32.o sunos_ioctl32.o \
+	    central.o psycho.o ebus.o
 OX_OBJS  := sparc64_ksyms.o
 
 ifdef SMP
diff --git a/arch/sparc64/kernel/auxio.c b/arch/sparc64/kernel/auxio.c
index 00e5f2722..91365af85 100644
--- a/arch/sparc64/kernel/auxio.c
+++ b/arch/sparc64/kernel/auxio.c
@@ -3,15 +3,21 @@
  * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
  */
 
+#include <linux/config.h>
 #include <linux/stddef.h>
 #include <linux/kernel.h>
 #include <linux/sched.h>
 #include <linux/smp.h>
 #include <linux/init.h>
+#include <linux/delay.h>
+#include <linux/ioport.h>
+
 #include <asm/oplib.h>
 #include <asm/io.h>
 #include <asm/auxio.h>
 #include <asm/sbus.h>
+#include <asm/ebus.h>
+#include <asm/fhc.h>
 
 /* Probe and map in the Auxiliary I/O register */
 unsigned char *auxio_register;
@@ -31,9 +37,39 @@ __initfunc(void auxio_probe(void))
         }
 
 	if (!sdev) {
+#ifdef CONFIG_PCI
+		struct linux_ebus *ebus;
+		struct linux_ebus_device *edev = 0;
+		unsigned long led_auxio;
+
+		for_all_ebusdev(edev, ebus)
+			if (!strcmp(edev->prom_name, "auxio"))
+				break;
+
+		if (edev) {
+			if (check_region(edev->base_address[0],
+					 sizeof(unsigned int))) {
+				prom_printf("%s: Can't get region %lx, %d\n",
+					    __FUNCTION__, edev->base_address[0],
+					    sizeof(unsigned int));
+				prom_halt();
+			}
+			request_region(edev->base_address[0],
+				       sizeof(unsigned int), "LED auxio");
+
+			led_auxio = edev->base_address[0];
+			outl(0x01, led_auxio);
+			return;
+		}
+#endif
+		if(central_bus) {
+			auxio_register = NULL;
+			return;
+		}
 		prom_printf("Cannot find auxio node, cannot continue...\n");
 		prom_halt();
 	}
+
 	prom_getproperty(sdev->prom_node, "reg", (char *) auxregs, sizeof(auxregs));
 	prom_apply_sbus_ranges(sdev->my_bus, auxregs, 0x1, sdev);
 	/* Map the register both read and write */
diff --git a/arch/sparc64/kernel/binfmt_elf32.c b/arch/sparc64/kernel/binfmt_elf32.c
index 9ab2b7aca..ef41e3f7a 100644
--- a/arch/sparc64/kernel/binfmt_elf32.c
+++ b/arch/sparc64/kernel/binfmt_elf32.c
@@ -8,6 +8,9 @@
 
 #define elf_check_arch(x)	(((x) == EM_SPARC) || ((x) == EM_SPARC32PLUS))
 
+#define ELF_ET_DYN_BASE         0x60000000
+
+
 #include <asm/processor.h>
 #include <linux/module.h>
 #include <linux/config.h>
diff --git a/arch/sparc64/kernel/central.c b/arch/sparc64/kernel/central.c
new file mode 100644
index 000000000..817a8ecd3
--- /dev/null
+++ b/arch/sparc64/kernel/central.c
@@ -0,0 +1,129 @@
+/* $Id: central.c,v 1.4 1997/08/19 14:17:49 jj Exp $
+ * central.c: Central FHC driver for Sunfire/Starfire/Wildfire.
+ *
+ * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
+ */
+
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/string.h>
+
+#include <asm/page.h>
+#include <asm/fhc.h>
+
+struct linux_central *central_bus = NULL;
+
+static inline unsigned long long_align(unsigned long addr)
+{
+	return ((addr + (sizeof(unsigned long) - 1)) &
+		~(sizeof(unsigned long) - 1));
+}
+
+extern void prom_central_ranges_init(int cnode, struct linux_central *central);
+extern void prom_fhc_ranges_init(int fnode, struct linux_fhc *fhc);
+
+unsigned long central_probe(unsigned long memory_start)
+{
+	struct linux_prom_registers fpregs[6];
+	struct linux_fhc *fhc;
+	char namebuf[128];
+	int cnode, fnode, err;
+
+	printk("CENTRAL: ");
+	cnode = prom_finddevice("/central");
+	if(cnode == 0 || cnode == -1) {
+		printk("no central found.\n");
+		return memory_start;
+	}
+	prom_printf("CENTRAL: found central PROM node.\n");
+	printk("found central PROM node.\n");
+
+	/* Ok we got one, grab some memory for software state. */
+	memory_start = long_align(memory_start);
+	central_bus = (struct linux_central *) (memory_start);
+
+	prom_printf("CENTRAL: central_bus[%p] ", central_bus);
+	memory_start += sizeof(struct linux_central);
+	memory_start = long_align(memory_start);
+	fhc = (struct linux_fhc *)(memory_start);
+	memory_start += sizeof(struct linux_fhc);
+	memory_start = long_align(memory_start);
+
+	prom_printf("fhc[%p] ", fhc);
+
+	/* First init central. */
+	central_bus->child = fhc;
+	central_bus->prom_node = cnode;
+
+	prom_getstring(cnode, "name", namebuf, sizeof(namebuf));
+	strcpy(central_bus->prom_name, namebuf);
+
+	prom_printf("init_central_ranges ");
+	prom_central_ranges_init(cnode, central_bus);
+
+	/* And then central's FHC. */
+	fhc->next = NULL;
+	fhc->parent = central_bus;
+	fnode = prom_searchsiblings(prom_getchild(cnode), "fhc");
+	if(fnode == 0 || fnode == -1) {
+		prom_printf("Critical error, central board lacks fhc.\n");
+		prom_halt();
+	}
+	fhc->prom_node = fnode;
+	prom_getstring(fnode, "name", namebuf, sizeof(namebuf));
+	strcpy(fhc->prom_name, namebuf);
+
+	prom_printf("cnode[%x] fnode[%x] init_fhc_ranges\n", cnode, fnode);
+	prom_fhc_ranges_init(fnode, fhc);
+
+	/* Finally, map in FHC register set.  (From the prtconf dumps
+	 * I have seen on Ex000 boxes only the central ranges need to
+	 * be applied to the fhc internal register set) -DaveM
+	 */
+	err = prom_getproperty(fnode, "reg", (char *)&fpregs[0], sizeof(fpregs));
+	if(err == -1) {
+		prom_printf("CENTRAL: fatal error, cannot get fhc regs.\n");
+		prom_halt();
+	}
+	prom_apply_central_ranges(central_bus, &fpregs[0], 6);
+	prom_printf("CENTRAL: FHC_REGS[(%08x,%08x) (%08x,%08x) "
+		    "(%08x,%08x) (%08x,%08x) (%08x,%08x) (%08x,%08x)]\n",
+		    fpregs[0].which_io, fpregs[0].phys_addr,
+		    fpregs[1].which_io, fpregs[1].phys_addr,
+		    fpregs[2].which_io, fpregs[2].phys_addr,
+		    fpregs[3].which_io, fpregs[3].phys_addr,
+		    fpregs[4].which_io, fpregs[4].phys_addr,
+		    fpregs[5].which_io, fpregs[5].phys_addr);
+	fhc->fhc_regs.pregs = (struct fhc_internal_regs *)
+		__va((((unsigned long)fpregs[0].which_io)<<32) |
+		     (((unsigned long)fpregs[0].phys_addr)));
+	fhc->fhc_regs.ireg = (struct fhc_ign_reg *)
+		__va((((unsigned long)fpregs[1].which_io)<<32) |
+		     (((unsigned long)fpregs[1].phys_addr)));
+	fhc->fhc_regs.ffregs = (struct fhc_fanfail_regs *)
+		__va((((unsigned long)fpregs[2].which_io)<<32) |
+		     (((unsigned long)fpregs[2].phys_addr)));
+	fhc->fhc_regs.sregs = (struct fhc_system_regs *)
+		__va((((unsigned long)fpregs[3].which_io)<<32) |
+		     (((unsigned long)fpregs[3].phys_addr)));
+	fhc->fhc_regs.uregs = (struct fhc_uart_regs *)
+		__va((((unsigned long)fpregs[4].which_io)<<32) |
+		     (((unsigned long)fpregs[4].phys_addr)));
+	fhc->fhc_regs.tregs = (struct fhc_tod_regs *)
+		__va((((unsigned long)fpregs[5].which_io)<<32) |
+		     (((unsigned long)fpregs[5].phys_addr)));
+	prom_printf("CENTRAL: FHC_REGS[%p %p %p %p %p %p]\n",
+		    fhc->fhc_regs.pregs, fhc->fhc_regs.ireg,
+		    fhc->fhc_regs.ffregs, fhc->fhc_regs.sregs,
+		    fhc->fhc_regs.uregs, fhc->fhc_regs.tregs);
+
+	prom_printf("CENTRAL: reading FHC_ID register... ");
+	err = fhc->fhc_regs.pregs->fhc_id;
+	prom_printf("VALUE[%x]\n", err);
+	printk("FHC Version[%x] PartID[%x] Manufacturer[%x]\n",
+	       ((err & FHC_ID_VERS) >> 28),
+	       ((err & FHC_ID_PARTID) >> 12),
+	       ((err & FHC_ID_MANUF) >> 1));
+
+	return memory_start;
+}
diff --git a/arch/sparc64/kernel/cpu.c b/arch/sparc64/kernel/cpu.c
index 2c96a83e9..d009f39d8 100644
--- a/arch/sparc64/kernel/cpu.c
+++ b/arch/sparc64/kernel/cpu.c
@@ -46,8 +46,13 @@ struct cpu_iu_info linux_sparc_chips[] = {
 
 #define NSPARCCHIPS  (sizeof(linux_sparc_chips)/sizeof(struct cpu_iu_info))
 
-char *sparc_cpu_type[NCPUS] = { "cpu-oops", "cpu-oops1", "cpu-oops2", "cpu-oops3" };
-char *sparc_fpu_type[NCPUS] = { "fpu-oops", "fpu-oops1", "fpu-oops2", "fpu-oops3" };
+#ifdef __SMP__
+char *sparc_cpu_type[NR_CPUS] = { "cpu-oops", "cpu-oops1", "cpu-oops2", "cpu-oops3" };
+char *sparc_fpu_type[NR_CPUS] = { "fpu-oops", "fpu-oops1", "fpu-oops2", "fpu-oops3" };
+#else
+char *sparc_cpu_type[NR_CPUS] = { "cpu-oops", };
+char *sparc_fpu_type[NR_CPUS] = { "fpu-oops", };
+#endif
 
 unsigned int fsr_storage;
 
diff --git a/arch/sparc64/kernel/devices.c b/arch/sparc64/kernel/devices.c
index 9327058a8..24ca3ff10 100644
--- a/arch/sparc64/kernel/devices.c
+++ b/arch/sparc64/kernel/devices.c
@@ -17,13 +17,14 @@ struct prom_cpuinfo linux_cpus[NR_CPUS];
 int linux_num_cpus = 0;
 
 extern void cpu_probe(void);
+extern unsigned long central_probe(unsigned long);
 
 __initfunc(unsigned long
 device_scan(unsigned long mem_start))
 {
 	char node_str[128];
 	int nd, prom_node_cpu, thismid;
-	int cpu_nds[NCPUS];  /* One node for each cpu */
+	int cpu_nds[NR_CPUS];  /* One node for each cpu */
 	int cpu_ctr = 0;
 
 	prom_getstring(prom_root_node, "device_type", node_str, sizeof(node_str));
@@ -43,11 +44,14 @@ device_scan(unsigned long mem_start))
 			if(strcmp(node_str, "cpu") == 0) {
 				cpu_nds[cpu_ctr] = scan;
 				linux_cpus[cpu_ctr].prom_node = scan;
-				prom_getproperty(scan, "mid", (char *) &thismid, sizeof(thismid));
+				prom_getproperty(scan, "upa-portid",
+						 (char *) &thismid, sizeof(thismid));
 				linux_cpus[cpu_ctr].mid = thismid;
 				prom_printf("Found CPU %d <node=%08x,mid=%d>\n",
 					    cpu_ctr, (unsigned) scan,
 					    thismid);
+				printk("Found CPU %d <node=%08x,mid=%d>\n",
+				       cpu_ctr, (unsigned) scan, thismid);
 				cpu_ctr++;
 			}
 		};
@@ -62,5 +66,5 @@ device_scan(unsigned long mem_start))
 	linux_num_cpus = cpu_ctr;
 
 	cpu_probe();
-	return mem_start;
+	return central_probe(mem_start);
 }
diff --git a/arch/sparc64/kernel/dtlb_miss.S b/arch/sparc64/kernel/dtlb_miss.S
index b034ef407..04efb1cec 100644
--- a/arch/sparc64/kernel/dtlb_miss.S
+++ b/arch/sparc64/kernel/dtlb_miss.S
@@ -1,4 +1,4 @@
-/* $Id: dtlb_miss.S,v 1.12 1997/06/26 12:47:08 jj Exp $
+/* $Id: dtlb_miss.S,v 1.13 1997/08/14 19:27:15 davem Exp $
  * dtlb_miss.S:	Data TLB miss code, this is included directly
  *              into the trap table.
  *
@@ -62,7 +62,7 @@
   /*0x40*/	sllx		%g1, 22, %g5			! This is now physical page + PAGE_OFFSET
   /*0x44*/	brgez,pn	%g5, 4f				! If >= 0, then walk down page tables
   /*0x48*/	 sethi		%uhi(KERN_HIGHBITS), %g1	! Construct PTE ^ PAGE_OFFSET
-  /*0x4c*/	andcc		%g3, 0x80, %g0			! Slick trick...
+  /*0x4c*/	andcc		%g3, 0x400, %g0			! Slick trick...
   /*0x50*/	sllx		%g1, 32, %g1			! Move high bits up
   /*0x54*/	or		%g1, (KERN_LOWBITS), %g1	! Assume not IO
   /*0x58*/	bne,a,pn	%icc, 5f			! Is it an IO page?
diff --git a/arch/sparc64/kernel/dtlb_prot.S b/arch/sparc64/kernel/dtlb_prot.S
index 2cf372ca9..55e86c887 100644
--- a/arch/sparc64/kernel/dtlb_prot.S
+++ b/arch/sparc64/kernel/dtlb_prot.S
@@ -1,4 +1,4 @@
-/* $Id: dtlb_prot.S,v 1.12 1997/05/18 10:04:43 davem Exp $
+/* $Id: dtlb_prot.S,v 1.14 1997/08/03 09:07:00 davem Exp $
  * dtlb_prot.S:	Data TLB protection code, this is included directly
  *              into the trap table.
  *
@@ -36,17 +36,17 @@
 				/* ICACHE line 3 */
   /*0x40*/	add		%g2, 7, %g5				! Compute mask
   /*0x44*/	andn		%g4, %g5, %g4				! Mask page
-  /*0x48*/	or		%g4, 0x10, %g4				! 2ndary Context
-  /*0x4c*/	stxa		%g0, [%g4] ASI_DMMU_DEMAP		! TLB flush page
-  /*0x50*/	membar		#Sync					! Synchronize
-  /*0x54*/	stxa		%g3, [%g1] ASI_PHYS_USE_EC		! Update sw PTE
-  /*0x58*/	stxa		%g3, [%g0] ASI_DTLB_DATA_IN		! TLB load
-  /*0x5c*/	retry							! Trap return
+  /*0x48*/	mov		TLB_SFSR, %g5				! read SFSR
+  /*0x4c*/	ldxa		[%g5] ASI_DMMU, %g5			! from DMMU for
+  /*0x50*/	and		%g5, 0x10, %g5				! context bit
+  /*0x54*/	or		%g4, %g5, %g4				! for prot trap
+1:/*0x58*/	stxa		%g0, [%g4] ASI_DMMU_DEMAP		! TLB flush page
+  /*0x5c*/	membar		#Sync					! Synchronize
 
 				/* ICACHE line 4 */
-  /*0x60*/	nop
-  /*0x64*/	nop
-  /*0x68*/	nop
+  /*0x60*/	stxa		%g3, [%g1] ASI_PHYS_USE_EC		! Update sw PTE
+  /*0x64*/	stxa		%g3, [%g0] ASI_DTLB_DATA_IN		! TLB load
+  /*0x68*/	retry							! Trap return
   /*0x6c*/	nop
   /*0x70*/	nop
   /*0x74*/	nop
diff --git a/arch/sparc64/kernel/ebus.c b/arch/sparc64/kernel/ebus.c
new file mode 100644
index 000000000..f484cfef8
--- /dev/null
+++ b/arch/sparc64/kernel/ebus.c
@@ -0,0 +1,326 @@
+/* $Id: ebus.c,v 1.8 1997/09/05 22:59:39 ecd Exp $
+ * ebus.c: PCI to EBus bridge device.
+ *
+ * Copyright (C) 1997  Eddie C. Dost  (ecd@skynet.be)
+ */
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/string.h>
+
+#include <asm/system.h>
+#include <asm/page.h>
+#include <asm/pbm.h>
+#include <asm/ebus.h>
+#include <asm/oplib.h>
+#include <asm/bpp.h>
+
+#undef DEBUG_FILL_EBUS_DEV
+
+struct linux_ebus *ebus_chain = 0;
+
+extern void prom_ebus_ranges_init(struct linux_ebus *);
+extern unsigned long pci_console_init(unsigned long memory_start);
+
+#ifdef CONFIG_SUN_OPENPROMIO
+extern int openprom_init(void);
+#endif
+#ifdef CONFIG_SUN_MOSTEK_RTC
+extern int rtc_init(void);
+#endif
+#ifdef CONFIG_SPARCAUDIO
+extern int sparcaudio_init(void);
+#endif
+#ifdef CONFIG_SUN_AUXIO
+extern void auxio_probe(void);
+#endif
+
+extern unsigned int psycho_irq_build(unsigned int full_ino);
+
+static inline unsigned long
+ebus_alloc(unsigned long *memory_start, size_t size)
+{
+	unsigned long mem;
+
+	*memory_start = (*memory_start + 7) & ~(7);
+	mem = *memory_start;
+	*memory_start += size;
+	return mem;
+}
+
+__initfunc(void fill_ebus_child(int node, struct linux_ebus_child *dev))
+{
+	int regs[PROMREG_MAX];
+	int irqs[PROMREG_MAX];
+	char lbuf[128];
+	int i, len;
+
+	dev->prom_node = node;
+	prom_getstring(node, "name", lbuf, sizeof(lbuf));
+	strcpy(dev->prom_name, lbuf);
+
+	len = prom_getproperty(node, "reg", (void *)regs, sizeof(regs));
+	dev->num_addrs = len / sizeof(regs[0]);
+
+	for (i = 0; i < dev->num_addrs; i++) {
+		if (regs[i] >= dev->parent->num_addrs) {
+			prom_printf("UGH: property for %s was %d, need < %d\n",
+				    dev->prom_name, len, dev->parent->num_addrs);
+			panic(__FUNCTION__);
+		}
+		dev->base_address[i] = dev->parent->base_address[regs[i]];
+	}
+
+	len = prom_getproperty(node, "interrupts", (char *)&irqs, sizeof(irqs));
+	if ((len == -1) || (len == 0)) {
+		dev->num_irqs = 0;
+	} else {
+		dev->num_irqs = len / sizeof(irqs[0]);
+		for (i = 0; i < dev->num_irqs; i++)
+			dev->irqs[i] = psycho_irq_build(irqs[i]);
+	}
+
+#ifdef DEBUG_FILL_EBUS_DEV
+	printk("child '%s': address%s\n", dev->prom_name,
+	       dev->num_addrs > 1 ? "es" : "");
+	for (i = 0; i < dev->num_addrs; i++)
+		printk("        %016lx\n", dev->base_address[i]);
+	if (dev->num_irqs) {
+		printk("        IRQ%s", dev->num_irqs > 1 ? "s" : "");
+		for (i = 0; i < dev->num_irqs; i++)
+			printk(" %08x", dev->irqs[i]);
+		printk("\n");
+	}
+#endif
+}
+
+__initfunc(unsigned long fill_ebus_device(int node, struct linux_ebus_device *dev,
+					  unsigned long memory_start))
+{
+	struct linux_prom_registers regs[PROMREG_MAX];
+	struct linux_ebus_child *child;
+	int irqs[PROMINTR_MAX];
+	char lbuf[128];
+	int i, n, len;
+
+	dev->prom_node = node;
+	prom_getstring(node, "name", lbuf, sizeof(lbuf));
+	strcpy(dev->prom_name, lbuf);
+
+	len = prom_getproperty(node, "reg", (void *)regs, sizeof(regs));
+	if (len % sizeof(struct linux_prom_registers)) {
+		prom_printf("UGH: proplen for %s was %d, need multiple of %d\n",
+			    dev->prom_name, len,
+			    (int)sizeof(struct linux_prom_registers));
+		panic(__FUNCTION__);
+	}
+	dev->num_addrs = len / sizeof(struct linux_prom_registers);
+
+	for (i = 0; i < dev->num_addrs; i++) {
+		n = (regs[i].which_io - 0x10) >> 2;
+
+		dev->base_address[i] = dev->parent->self->base_address[n];
+		dev->base_address[i] += (unsigned long)regs[i].phys_addr;
+	}
+
+	len = prom_getproperty(node, "interrupts", (char *)&irqs, sizeof(irqs));
+	if ((len == -1) || (len == 0)) {
+		dev->num_irqs = 0;
+	} else {
+		dev->num_irqs = len / sizeof(irqs[0]);
+		for (i = 0; i < dev->num_irqs; i++)
+			dev->irqs[i] = psycho_irq_build(irqs[i]);
+	}
+
+#ifdef DEBUG_FILL_EBUS_DEV
+	printk("'%s': address%s\n", dev->prom_name,
+	       dev->num_addrs > 1 ? "es" : "");
+	for (i = 0; i < dev->num_addrs; i++)
+		printk("  %016lx\n", dev->base_address[i]);
+	if (dev->num_irqs) {
+		printk("  IRQ%s", dev->num_irqs > 1 ? "s" : "");
+		for (i = 0; i < dev->num_irqs; i++)
+			printk(" %08x", dev->irqs[i]);
+		printk("\n");
+	}
+#endif
+	if ((node = prom_getchild(node))) {
+		dev->children = (struct linux_ebus_child *)
+			ebus_alloc(&memory_start, sizeof(struct linux_ebus_child));
+
+		child = dev->children;
+		child->next = 0;
+		child->parent = dev;
+		fill_ebus_child(node, child);
+
+		while ((node = prom_getsibling(node))) {
+			child->next = (struct linux_ebus_child *)
+				ebus_alloc(&memory_start, sizeof(struct linux_ebus_child));
+
+			child = child->next;
+			child->next = 0;
+			child->parent = dev;
+			fill_ebus_child(node, child);
+		}
+	}
+
+	return memory_start;
+}
+
+__initfunc(unsigned long ebus_init(unsigned long memory_start,
+				   unsigned long memory_end))
+{
+	struct linux_prom_pci_registers regs[PROMREG_MAX];
+	struct linux_pbm_info *pbm;
+	struct linux_ebus_device *dev;
+	struct linux_ebus *ebus;
+	struct pci_dev *pdev;
+	struct pcidev_cookie *cookie;
+	char lbuf[128];
+	unsigned long addr, *base;
+	unsigned short pci_command;
+	int nd, len, ebusnd;
+	int reg, rng, nreg;
+	int num_ebus = 0;
+
+	if (!pcibios_present())
+		return memory_start;
+
+	for (pdev = pci_devices; pdev; pdev = pdev->next) {
+		if ((pdev->vendor == PCI_VENDOR_ID_SUN) &&
+		    (pdev->device == PCI_DEVICE_ID_SUN_EBUS))
+			break;
+	}
+	if (!pdev) {
+		printk("ebus: No EBus's found.\n");
+		return memory_start;
+	}
+
+	cookie = pdev->sysdata;
+	ebusnd = cookie->prom_node;
+
+	ebus_chain = ebus = (struct linux_ebus *)
+			ebus_alloc(&memory_start, sizeof(struct linux_ebus));
+	ebus->next = 0;
+
+	while (ebusnd) {
+		printk("ebus%d:\n", num_ebus);
+
+		prom_getstring(ebusnd, "name", lbuf, sizeof(lbuf));
+		ebus->prom_node = ebusnd;
+		strcpy(ebus->prom_name, lbuf);
+
+		ebus->self = pdev;
+		ebus->parent = pbm = cookie->pbm;
+
+		/* Enable BUS Master. */
+		pcibios_read_config_word(pdev->bus->number, pdev->devfn,
+					 PCI_COMMAND, &pci_command);
+		pci_command |= PCI_COMMAND_MASTER;
+		pcibios_write_config_word(pdev->bus->number, pdev->devfn,
+					  PCI_COMMAND, pci_command);
+
+		len = prom_getproperty(ebusnd, "reg", (void *)regs,
+				       sizeof(regs));
+		if (len == 0 || len == -1) {
+			prom_printf("%s: can't find reg property\n",
+				    __FUNCTION__);
+			prom_halt();
+		}
+		nreg = len / sizeof(struct linux_prom_pci_registers);
+
+		base = &ebus->self->base_address[0];
+		for (reg = 0; reg < nreg; reg++) {
+			if (!(regs[reg].phys_hi & 0x03000000))
+				continue;
+
+			for (rng = 0; rng < pbm->num_pbm_ranges; rng++) {
+				struct linux_prom_pci_ranges *rp =
+						&pbm->pbm_ranges[rng];
+
+				if ((rp->child_phys_hi ^ regs[reg].phys_hi)
+								& 0x03000000)
+					continue;
+
+				addr = (u64)regs[reg].phys_lo;
+				addr += (u64)regs[reg].phys_mid << 32UL;
+				addr += (u64)rp->parent_phys_lo;
+				addr += (u64)rp->parent_phys_hi << 32UL;
+				*base++ = (unsigned long)__va(addr);
+
+				break;
+			}
+		}
+
+		prom_ebus_ranges_init(ebus);
+
+		nd = prom_getchild(ebusnd);
+		ebus->devices = (struct linux_ebus_device *)
+			ebus_alloc(&memory_start, sizeof(struct linux_ebus_device));
+
+		dev = ebus->devices;
+		dev->next = 0;
+		dev->children = 0;
+		dev->parent = ebus;
+		memory_start = fill_ebus_device(nd, dev, memory_start);
+
+		while ((nd = prom_getsibling(nd))) {
+			dev->next = (struct linux_ebus_device *)
+				ebus_alloc(&memory_start, sizeof(struct linux_ebus_device));
+
+			dev = dev->next;
+			dev->next = 0;
+			dev->children = 0;
+			dev->parent = ebus;
+			memory_start = fill_ebus_device(nd, dev, memory_start);
+		}
+
+		for (pdev = pdev->next; pdev; pdev = pdev->next) {
+			if ((pdev->vendor == PCI_VENDOR_ID_SUN) &&
+			    (pdev->device == PCI_DEVICE_ID_SUN_EBUS))
+				break;
+		}
+		if (!pdev)
+			break;
+
+		cookie = pdev->sysdata;
+		ebusnd = cookie->prom_node;
+
+		ebus->next = (struct linux_ebus *)
+			ebus_alloc(&memory_start, sizeof(struct linux_ebus));
+		ebus = ebus->next;
+		ebus->next = 0;
+		++num_ebus;
+	}
+
+	memory_start = pci_console_init(memory_start);
+
+#ifdef CONFIG_SUN_OPENPROMIO
+	openprom_init();
+#endif
+#ifdef CONFIG_SUN_MOSTEK_RTC
+	rtc_init();
+#endif
+#ifdef CONFIG_SPARCAUDIO
+	sparcaudio_init();
+#endif
+#ifdef CONFIG_SUN_BPP
+	bpp_init();
+#endif
+#ifdef CONFIG_SUN_AUXIO
+	if (sparc_cpu_model == sun4u)
+		auxio_probe();
+#endif
+#ifdef __sparc_v9__
+	if (sparc_cpu_model == sun4u) {
+		extern void sun4u_start_timers(void);
+		extern void clock_probe(void);
+
+		sun4u_start_timers();
+		clock_probe();
+	}
+#endif
+	return memory_start;
+}
diff --git a/arch/sparc64/kernel/entry.S b/arch/sparc64/kernel/entry.S
index 425c2d873..a6e2d6da7 100644
--- a/arch/sparc64/kernel/entry.S
+++ b/arch/sparc64/kernel/entry.S
@@ -1,9 +1,9 @@
-/* $Id: entry.S,v 1.51 1997/07/24 12:15:04 davem Exp $
+/* $Id: entry.S,v 1.65 1997/08/29 15:51:29 jj Exp $
  * arch/sparc64/kernel/entry.S:  Sparc64 trap low-level entry points.
  *
  * Copyright (C) 1995,1997 David S. Miller (davem@caip.rutgers.edu)
- * Copyright (C) 1996 Eddie C. Dost   (ecd@skynet.be)
- * Copyright (C) 1996 Miguel de Icaza (miguel@nuclecu.unam.mx)
+ * Copyright (C) 1996 Eddie C. Dost        (ecd@skynet.be)
+ * Copyright (C) 1996 Miguel de Icaza      (miguel@nuclecu.unam.mx)
  * Copyright (C) 1996,1997 Jakub Jelinek   (jj@sunsite.mff.cuni.cz)
  */
 
@@ -17,6 +17,7 @@
 #include <asm/page.h>
 #include <asm/signal.h>
 #include <asm/pgtable.h>
+#include <asm/processor.h>
 
 /* #define SYSCALL_TRACING */
 
@@ -50,41 +51,39 @@ sparc64_dtlb_prot_catch:
 
 	bgu,a,pn	%icc, winfix_trampoline
 	 rdpr		%tpc, %g3
+	sethi		%hi(109f), %g7
 	ba,pt		%xcc, etrap
-	 rd		%pc, %g7
+109:	 or		%g7, %lo(109b), %g7
 	b,pt		%xcc, 1f
 	 mov		1, %o2
+	.align		32
 sparc64_dtlb_refbit_catch:
 	srlx		%g5, 9, %g4
 	and		%g4, ((_PAGE_PRESENT | _PAGE_READ) >> 9), %g4
-
 	cmp		%g4, ((_PAGE_PRESENT | _PAGE_READ) >> 9)
 	be,a,pt		%xcc, 2f
 	 mov		1, %g4
 	wr		%g0, ASI_DMMU, %asi
 	rdpr		%pstate, %g1
 	wrpr		%g1, PSTATE_AG|PSTATE_MG, %pstate
+
 	rdpr		%tl, %g3
 	ldxa		[%g0 + TLB_TAG_ACCESS] %asi, %g5
-
 	cmp		%g3, 1
 	bgu,pn		%icc, winfix_trampoline
 	 rdpr		%tpc, %g3
+	sethi		%hi(109f), %g7
 	b,pt		%xcc, etrap
-	 rd		%pc, %g7
+109:	 or		%g7, %lo(109b), %g7
+
 	clr		%o2
 1:	srlx		%l5, PAGE_SHIFT, %o1
 	add		%sp, STACK_BIAS + REGWIN_SZ, %o0
-
 	call		do_sparc64_fault
 	 sllx		%o1, PAGE_SHIFT, %o1
 	b,pt		%xcc, rtrap
 	 clr		%l6
-	nop
-	nop
-	nop
-	nop
-
+	.align		32
 sparc64_itlb_refbit_catch:
 	srlx		%g5, 9, %g4
 	and		%g4, ((_PAGE_PRESENT | _PAGE_READ) >> 9), %g4
@@ -95,17 +94,21 @@ sparc64_itlb_refbit_catch:
 	wrpr		%g1, PSTATE_AG|PSTATE_MG, %pstate
 	rdpr		%tpc, %g5
 
+	sethi		%hi(109f), %g7
 	b,pt		%xcc, etrap
-	 rd		%pc, %g7
+109:	 or		%g7, %lo(109b), %g7
 	b,pt		%xcc, 1b
 	 clr		%o2
+
+	.align		32
 2:	sllx		%g4, 63, %g4				! _PAGE_VALID
 	or		%g5, _PAGE_ACCESSED, %g5
 	or		%g5, %g4, %g5
 	stxa		%g5, [%g3 + %g1] ASI_PHYS_USE_EC	! store new PTE
-
 	stxa		%g5, [%g0] ASI_DTLB_DATA_IN		! TLB load
 	retry
+
+	.align		32
 3:	sllx		%g4, 63, %g4				! _PAGE_VALID
 	or		%g5, _PAGE_ACCESSED, %g5
 	or		%g5, %g4, %g5
@@ -113,24 +116,33 @@ sparc64_itlb_refbit_catch:
 	stxa		%g5, [%g0] ASI_ITLB_DATA_IN		! TLB load
 	retry
 
+#define FPDIS_OFF (((PAGE_SIZE<<1)-((64*4)+(2*8))) & ~(64 - 1))
 	/* This is trivial with the new code... */
 	.align		32
 	.globl		do_fpdis
 do_fpdis:
-	wr		%g0, FPRS_FEF, %fprs
-	ldx		[%g6 + AOFF_task_flags], %g2
-	sethi		%hi(0x00100000), %g4		! XXX PF_USEDFPU
-	andcc		%g2, %g4, %g0
-
-	bne,a,pt	%xcc, fpload_fromkstk
-	 sethi		%hi((((PAGE_SIZE<<1)-((64*4)+(2*8))) & ~(64 - 1))), %g2
-	fzero		%f0
-	fzero		%f2
+	ldx		[%g6 + AOFF_task_tss + AOFF_thread_flags], %g5		! Load	Group
+	sethi		%hi(TSTATE_PEF), %g4					! IEU0
+	sethi		%hi(FPDIS_OFF), %g3					! IEU1
+	wr		%g0, FPRS_FEF, %fprs					! LSU	Group+4bubbles
+	andcc		%g5, SPARC_FLAG_USEDFPU, %g0				! IEU1	Group
+	or		%g3, %lo(FPDIS_OFF), %g2				! IEU0
+	sethi		%hi(empty_zero_page), %g1				! IEU0	Group
+	add		%g6, %g2, %g2						! IEU1
+	be,a,pn		%icc, 1f						! CTI
+	 clr		%g7							! IEU0	Group
+	add		%g2, 0x100, %g1						! IEU1
+	ldx		[%g2 + 0x108], %g7					! Load
+1:	andcc		%g5, SPARC_FLAG_USEDFPUL, %g0				! IEU1	Group
+	bne,pn		%icc, 2f						! CTI
+	 fzero		%f0							! FPA
+	andcc		%g5, SPARC_FLAG_USEDFPUU, %g0				! IEU1  Group
+	bne,pn		%icc, 1f						! CTI
+	 fzero		%f2							! FPA
 	faddd		%f0, %f2, %f4
 	fmuld		%f0, %f2, %f6
 	faddd		%f0, %f2, %f8
 	fmuld		%f0, %f2, %f10
-
 	faddd		%f0, %f2, %f12
 	fmuld		%f0, %f2, %f14
 	faddd		%f0, %f2, %f16
@@ -139,7 +151,6 @@ do_fpdis:
 	fmuld		%f0, %f2, %f22
 	faddd		%f0, %f2, %f24
 	fmuld		%f0, %f2, %f26
-
 	faddd		%f0, %f2, %f28
 	fmuld		%f0, %f2, %f30
 	faddd		%f0, %f2, %f32
@@ -148,54 +159,92 @@ do_fpdis:
 	fmuld		%f0, %f2, %f38
 	faddd		%f0, %f2, %f40
 	fmuld		%f0, %f2, %f42
-
 	faddd		%f0, %f2, %f44
 	fmuld		%f0, %f2, %f46
-	ldx		[%g6 + AOFF_task_flags], %g2
 	faddd		%f0, %f2, %f48
 	fmuld		%f0, %f2, %f50
-	or		%g2, %g4, %g2
 	faddd		%f0, %f2, %f52
 	fmuld		%f0, %f2, %f54
-
-	stx		%g2, [%g6 + AOFF_task_flags]
 	faddd		%f0, %f2, %f56
-	sethi		%hi(empty_zero_page), %g3
 	fmuld		%f0, %f2, %f58
-
-	faddd		%f0, %f2, %f60
-	ldx		[%g3], %fsr			! wheee, empty_zero_page
+	b,pt		%xcc, fpdis_exit2
+	 faddd		%f0, %f2, %f60
+1:	mov		SECONDARY_CONTEXT, %g3
+	faddd		%f0, %f2, %f4
+	fmuld		%f0, %f2, %f6
+	ldxa		[%g3] ASI_DMMU, %g5
+	stxa		%g0, [%g3] ASI_DMMU
+	faddd		%f0, %f2, %f8
+	fmuld		%f0, %f2, %f10
+	flush		%g2
+	wr		%g0, ASI_BLK_S, %asi	! grrr, where is ASI_BLK_NUCLEUS 8-(
+	membar		#StoreLoad | #LoadLoad
+	ldda		[%g2 + 0x080] %asi, %f32
+	ldda		[%g2 + 0x0c0] %asi, %f48
+	faddd		%f0, %f2, %f12
+	fmuld		%f0, %f2, %f14
+	faddd		%f0, %f2, %f16
+	fmuld		%f0, %f2, %f18
+	faddd		%f0, %f2, %f20
+	fmuld		%f0, %f2, %f22
+	faddd		%f0, %f2, %f24
+	fmuld		%f0, %f2, %f26
+	faddd		%f0, %f2, %f28
+	fmuld		%f0, %f2, %f30
 	b,pt		%xcc, fpdis_exit
-	 wr		%g0, 0, %gsr
-
-fpload_fromkstk:
-	or		%g2, %lo((((PAGE_SIZE<<1)-((64*4)+(2*8))) & ~(64 - 1))), %g2
-	add		%g6, %g2, %g2
+	 membar		#Sync
+2:	andcc		%g5, SPARC_FLAG_USEDFPUU, %g0
+	bne,pt		%icc, 3f
+	 fzero		%f32
 	mov		SECONDARY_CONTEXT, %g3
+	fzero		%f34
+	ldxa		[%g3] ASI_DMMU, %g5
+	stxa		%g0, [%g3] ASI_DMMU
+	faddd		%f32, %f34, %f36
+	fmuld		%f32, %f34, %f38
+	flush		%g2
+	wr		%g0, ASI_BLK_S, %asi	! grrr, where is ASI_BLK_NUCLEUS 8-(
+	membar		#StoreLoad | #LoadLoad
+	ldda		[%g2 + 0x000] %asi, %f0
+	ldda		[%g2 + 0x040] %asi, %f16
+	faddd		%f32, %f34, %f40
+	fmuld		%f32, %f34, %f42
+	faddd		%f32, %f34, %f44
+	fmuld		%f32, %f34, %f46
+	faddd		%f32, %f34, %f48
+	fmuld		%f32, %f34, %f50
+	faddd		%f32, %f34, %f52
+	fmuld		%f32, %f34, %f54
+	faddd		%f32, %f34, %f56
+	fmuld		%f32, %f34, %f58
+	faddd		%f32, %f34, %f60
+	fmuld		%f32, %f34, %f62
+	b,pt		%xcc, fpdis_exit
+	 membar		#Sync
+3:	mov		SECONDARY_CONTEXT, %g3
+	ldxa		[%g3] ASI_DMMU, %g5
 	stxa		%g0, [%g3] ASI_DMMU
 	flush		%g2
 	wr		%g0, ASI_BLK_S, %asi	! grrr, where is ASI_BLK_NUCLEUS 8-(
 	membar		#StoreLoad | #LoadLoad
-
 	ldda		[%g2 + 0x000] %asi, %f0
 	ldda		[%g2 + 0x040] %asi, %f16
 	ldda		[%g2 + 0x080] %asi, %f32
 	ldda		[%g2 + 0x0c0] %asi, %f48
-	ldx		[%g2 + 0x100], %fsr
-	ldx		[%g2 + 0x108], %g2
 	membar		#Sync
-	wr		%g2, 0, %gsr
 fpdis_exit:
+	stxa		%g5, [%g3] ASI_DMMU
+	flush		%g2
+fpdis_exit2:
+	wr		%g7, 0, %gsr
+	ldx		[%g1], %fsr
 	rdpr		%tstate, %g3
-	sethi		%hi(TSTATE_PEF), %g4
 	or		%g3, %g4, %g3		! anal...
 	wrpr		%g3, %tstate
+	wr		%g0, FPRS_FEF, %fprs	! clean DU/DL bits
 	retry
 
-#ifdef __SMP__
-	/* Note check out head.h, this code isn't even used for UP,
-	 * for SMP things will be different.  In particular the data
-	 * registers for cross calls will be:
+	/* The registers for cross calls will be:
 	 *
 	 * DATA 0: [low 32-bits]  Address of function to call, jmp to this
 	 *         [high 32-bits] MMU Context Argument 0, place in %g5
@@ -205,11 +254,17 @@ fpdis_exit:
 	 * With this method we can do most of the cross-call tlb/cache
 	 * flushing very quickly.
 	 */
+	.data
+	.align  8
+	.globl	ivec_spurious_cookie
+ivec_spurious_cookie:	.xword	0
+
+	.text
 	.align	32
-	.globl	do_ivec, do_ivec_return
+	.globl	do_ivec
 do_ivec:
-	ldxa		[%g0] ASI_INTR_RECEIVE, %g1
-	andcc		%g1, 0x20, %g0
+	ldxa		[%g0] ASI_INTR_RECEIVE, %g5
+	andcc		%g5, 0x20, %g0
 	be,pn		%xcc, do_ivec_return
 	 mov		0x40, %g2
 
@@ -223,39 +278,45 @@ do_ivec:
 	sllx		%g3, 3, %g3
 	ldx		[%g1 + %g3], %g2
 	brz,pn		%g2, do_ivec_spurious
-	 nop
+	 sethi		%hi(0x80000000), %g5
+
+	or		%g2, %g5, %g2
+	stx		%g2, [%g1 + %g3]
 
 	/* No branches, worse case we don't know about this interrupt
 	 * yet, so we would just write a zero into the softint register
 	 * which is completely harmless.
 	 */
 	wr		%g2, 0x0, %set_softint
-
 do_ivec_return:
-	/* Acknowledge the UPA */
 	stxa		%g0, [%g0] ASI_INTR_RECEIVE
 	membar		#Sync
 	retry
 do_ivec_xcall:
 	srlx		%g3, 32, %g5
 	add		%g2, 0x10, %g2
-	sra		%g3, 0, %g3
+	srl		%g3, 0, %g3
 	ldxa		[%g2] ASI_UDB_INTR_R, %g6
 	add		%g2, 0x10, %g2
+	ldxa		[%g2] ASI_UDB_INTR_R, %g7
+	stxa		%g0, [%g0] ASI_INTR_RECEIVE
 	jmpl		%g3, %g0
-	 ldxa		[%g2] ASI_UDB_INTR_R, %g7
+	 membar		#Sync
 do_ivec_spurious:
+	srl		%g3, 3, %g3
+	sethi		%hi(ivec_spurious_cookie), %g2
+	stx		%g3, [%g2 + %lo(ivec_spurious_cookie)]
 	stxa		%g0, [%g0] ASI_INTR_RECEIVE
 	membar		#Sync
-	rdpr		%pstate, %g1
-	wrpr		%g1, PSTATE_IG | PSTATE_AG, %pstate
+	rdpr		%pstate, %g5
+	wrpr		%g5, PSTATE_IG | PSTATE_AG, %pstate
+	sethi		%hi(109f), %g7
 	ba,pt		%xcc, etrap
-	 rd		%pc, %g7
+109:	 or		%g7, %lo(109b), %g7
 	call		report_spurious_ivec
 	 add		%sp, STACK_BIAS + REGWIN_SZ, %o0
 	ba,pt		%xcc, rtrap
 	 clr		%l6
-#endif /* __SMP__ */
 
 	.globl		getcc, setcc
 getcc:
@@ -359,8 +420,9 @@ floppy_overrun:
 floppy_dosoftint:
 	rdpr		%pil, %g2
 	wrpr		%g0, 15, %pil
+	sethi		%hi(109f), %g7
 	b,pt		%xcc, etrap_irq
-	 rd		%pc, %g7
+109:	 or		%g7, %lo(109b), %g7
 
 	mov		11, %o0
 	mov		0, %o1
@@ -373,10 +435,8 @@ floppy_dosoftint:
 #endif /* CONFIG_BLK_DEV_FD */
 
 	/* XXX Here is stuff we still need to write... -DaveM XXX */
-	.globl		indirect_syscall, netbsd_syscall, solaris_syscall
-indirect_syscall:
+	.globl		netbsd_syscall
 netbsd_syscall:
-solaris_syscall:
 	retl
 	 nop
 
@@ -386,8 +446,9 @@ do_mna:
 	cmp		%g3, 1
 	bgu,a,pn	%icc, winfix_mna
 	 rdpr		%tpc, %g3
+	sethi		%hi(109f), %g7
 	ba,pt		%xcc, etrap
-	 rd		%pc, %g7
+109:	 or		%g7, %lo(109b), %g7
 	call		mem_address_unaligned
 	 add		%sp, STACK_BIAS + REGWIN_SZ, %o0
 	ba,pt		%xcc, rtrap
@@ -573,14 +634,8 @@ sys_fork:
 sys_vfork:	mov		SIGCHLD, %o0
 		clr		%o1
 sys_clone:	mov		%o7, %l5
-/*???*/		save		%sp, -REGWIN_SZ, %sp
-		flushw
-/*???*/		restore		%g0, %g0, %g0
-		rdpr		%cwp, %o4
 		add		%sp, STACK_BIAS + REGWIN_SZ, %o2
-
 		movrz		%o1, %fp, %o1
-		stx		%o4, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G0]
 		call		do_fork
 		 mov		%l5, %o7
 #ifdef __SMP__
@@ -611,9 +666,9 @@ linux_sparc_syscall:
 	cmp		%g1, NR_SYSCALLS			! IEU1	Group
 	bgeu,pn		%xcc, linux_sparc_ni_syscall		! CTI
 	 mov		%i0, %o0				! IEU0
-	sll		%g1, 3, %l4				! IEU0	Group
+	sll		%g1, 2, %l4				! IEU0	Group
 	mov		%i1, %o1				! IEU1
-	ldx		[%l7 + %l4], %l7			! Load
+	lduw		[%l7 + %l4], %l7			! Load
 syscall_is_too_hard:
 	mov		%i2, %o2				! IEU0	Group
 	ldx		[%curptr + AOFF_task_flags], %l5	! Load
diff --git a/arch/sparc64/kernel/etrap.S b/arch/sparc64/kernel/etrap.S
index 4daf30e21..e10480454 100644
--- a/arch/sparc64/kernel/etrap.S
+++ b/arch/sparc64/kernel/etrap.S
@@ -1,4 +1,4 @@
-/* $Id: etrap.S,v 1.30 1997/06/30 10:31:37 jj Exp $
+/* $Id: etrap.S,v 1.37 1997/08/21 09:13:18 davem Exp $
  * etrap.S: Preparing for entry into the kernel on Sparc V9.
  *
  * Copyright (C) 1996, 1997 David S. Miller (davem@caip.rutgers.edu)
@@ -15,118 +15,127 @@
 #define		FPUREG_SZ		((64 * 4) + (2 * 8))
 #define		TASK_REGOFF		((((PAGE_SIZE<<1)-FPUREG_SZ)&~(64-1)) - \
 					 TRACEREG_SZ-REGWIN_SZ)
+#define		FPU_OFF			(STACK_BIAS + REGWIN_SZ + TRACEREG_SZ)
+
+/*
+ * On entry, %g7 is return address - 0x4.
+ * %g4 and %g5 will be preserved %l4 and %l5 respectively.
+ */
 
 		.text
 		.align			32
+etrap_priv:	or			%g1, %g3, %g1						! IEU0		Group
+		rd			%fprs, %g3						! Single	Group+4bubbles
+		sub			%sp, REGWIN_SZ + TRACEREG_SZ - STACK_BIAS, %g2		! IEU0		Group
+		andcc			%g3, FPRS_FEF, %g0					! IEU1
+		add			%g2, REGWIN_SZ + TRACEREG_SZ - FPUREG_SZ, %g3		! IEU0		Group
+		be,pt			%icc, 1f						! CTI
+		 andcc			%g1, TSTATE_PRIV, %g0					! IEU1
+		andn			%g3, (64 - 1), %g3					! IEU0		Group
+		ba,pt			%xcc, 1f						! CTI
+		 sub			%g3, REGWIN_SZ + TRACEREG_SZ, %g2			! IEU0		Group
+		
+		.align			32
 		.globl			etrap, etrap_irq, etraptl1
 
-etrap:		rdpr			%pil, %g2
-etrap_irq:	rdpr			%tstate, %g1
-		sllx			%g2, 20, %g2
-		or			%g1, %g2, %g1
-		andcc			%g1, TSTATE_PRIV, %g0
-		bne,pn			%xcc, etrap_maybe_fpu
-		 sub			%sp, REGWIN_SZ + TRACEREG_SZ - STACK_BIAS, %g2
-		sethi			%hi(TASK_REGOFF), %g2
-
-		or			%g2, %lo(TASK_REGOFF), %g2
-		add			%g6, %g2, %g2
-etrap_maybe_fpu:rd			%fprs, %g3
-		brnz,pn			%g3, etrap_save_fpu
-		 st			%g0, [%g2 + REGWIN_SZ + PT_V9_FPRS]
-etrap_after_fpu:rdpr			%tpc, %g3
-		stx			%g1, [%g2 + REGWIN_SZ + PT_V9_TSTATE]
-		rdpr			%tnpc, %g1
-
-		stx			%g3, [%g2 + REGWIN_SZ + PT_V9_TPC]
-		rd			%y, %g3
-		stx			%g1, [%g2 + REGWIN_SZ + PT_V9_TNPC]
-		st			%g3, [%g2 + REGWIN_SZ + PT_V9_Y]
-		save			%g2, -STACK_BIAS, %sp	! The ordering here is
-		rdpr			%pstate, %g1		! critical, see winfixup
-		bne,pn			%xcc, 2f
-		 rdpr			%canrestore, %g3
-
-		rdpr			%wstate, %g2
-		wrpr			%g0, 7, %cleanwin
-		wrpr			%g0, 0, %canrestore
-		sll			%g2, 3, %g2
-		wrpr			%g3, 0, %otherwin
-		wrpr			%g2, 0, %wstate
-		wr			%g0, ASI_DMMU, %asi
-		ldxa			[%g0 + PRIMARY_CONTEXT] %asi, %g2
-
-		stxa			%g0, [%g0 + PRIMARY_CONTEXT] %asi
-		stxa			%g2, [%g0 + SECONDARY_CONTEXT] %asi
-		flush			%g6
-2:		wrpr			%g0, 0x0, %tl
-		or			%g1, 0, %l1
-		add			%g4, 0, %l4
-		or			%g5, 0, %l5
-		add			%g7, 0, %l2
-
-		or			%g6, 0, %l6
-		wrpr			%l1, (PSTATE_AG|PSTATE_RMO), %pstate
-		stx			%g1, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G1]
-		stx			%g2, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G2]
-		stx			%g3, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G3]
-		stx			%g4, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G4]
-		stx			%g5, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G5]
-		stx			%g6, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G6]
-
-		stx			%g7, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G7]
-		stx			%i0, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I0]
-		stx			%i1, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I1]
-		stx			%i2, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I2]
-		stx			%i3, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I3]
-		stx			%i4, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I4]
-		sethi			%uhi(PAGE_OFFSET), %g4
-		stx			%i5, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I5]
-
-		stx			%i6, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I6]
-		sllx			%g4, 32, %g4
-		stx			%i7, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I7]
-		wrpr			%l1, (PSTATE_IE|PSTATE_AG|PSTATE_RMO), %pstate
-		jmpl			%l2 + 0x4, %g0
-		 mov			%l6, %g6
-etrap_save_fpu:	and			%g3, FPRS_FEF, %g3
-		brz,pn			%g3, 2f
-
-		 nop
-		be,a,pt			%xcc, 3f
-		 add			%g2, (TRACEREG_SZ + REGWIN_SZ), %g2
-		wr			%g0, ASI_BLK_P, %asi
-		add			%g2, ((TRACEREG_SZ+REGWIN_SZ)-FPUREG_SZ), %g2
-		andn			%g2, (64 - 1), %g2
-1:		st			%g3, [%g2 - 0x4 /*REGWIN_SZ + PT_V9_FPRS*/]
-		rd			%gsr, %g3
-
-		stx			%fsr, [%g2 + 0x100]
-		stx			%g3,  [%g2 + 0x108]
-		membar			#StoreStore | #LoadStore
-		stda			%f0,  [%g2 + 0x000] %asi
-		stda			%f16, [%g2 + 0x040] %asi
-		stda			%f32, [%g2 + 0x080] %asi
-		stda			%f48, [%g2 + 0x0c0] %asi
-		membar			#Sync
-
-		sub			%g2, (TRACEREG_SZ + REGWIN_SZ), %g2
-2:		b,pt			%xcc, etrap_after_fpu
-		 wr			%g0, 0, %fprs
-3:		/* Because Ultra lacks ASI_BLK_NUCLEUS a hack has to take place. */
-		mov			SECONDARY_CONTEXT, %g3
-		stxa			%g0, [%g3] ASI_DMMU
-		flush			%g2
-		wr			%g0, ASI_BLK_S, %asi
-		nop
+etrap:		rdpr			%pil, %g2						! Single 	Group
+etrap_irq:	rdpr			%tstate, %g1						! Single 	Group
+		sllx			%g2, 20, %g3						! IEU0		Group
+		andcc			%g1, TSTATE_PRIV, %g0					! IEU1
+		bne,pn			%xcc, etrap_priv					! CTI
+		 sethi			%hi(TASK_REGOFF), %g2					! IEU0		Group
+		or			%g1, %g3, %g1						! IEU1
+		or			%g2, %lo(TASK_REGOFF), %g2				! IEU0		Group
+		add			%g6, %g2, %g2						! IEU0		Group
+1:		rdpr			%tpc, %g3						! Single	Group
+		stx			%g1, [%g2 + REGWIN_SZ + PT_V9_TSTATE]			! Store		Group
+		rdpr			%tnpc, %g1						! Single	Group
+		stx			%g3, [%g2 + REGWIN_SZ + PT_V9_TPC]			! Store		Group
+		rd			%y, %g3							! Single	Group+4bubbles
+		stx			%g1, [%g2 + REGWIN_SZ + PT_V9_TNPC]			! Store		Group
+		st			%g3, [%g2 + REGWIN_SZ + PT_V9_Y]			! Store		Group
+		save			%g2, -STACK_BIAS, %sp	! The ordering here is		! Single	Group
+		rdpr			%pstate, %g1		! critical, see winfixup	! Single	Group+9bubbles
+		bne,pn			%xcc, 2f						! CTI		Group
+		 sethi			%hi(TSTATE_PEF), %l2					! IEU0
+		mov			PRIMARY_CONTEXT, %l4					! IEU1
+		rdpr			%canrestore, %g3					! Single	Group+4bubbles
+		rdpr			%wstate, %g2						! Single	Group+4bubbles
+		wrpr			%g0, 7, %cleanwin					! Single	Group+4bubbles
+		wrpr			%g0, 0, %canrestore					! Single	Group+4bubbles
+		sll			%g2, 3, %g2						! IEU0		Group
+		mov			SECONDARY_CONTEXT, %l5					! IEU1
+		wrpr			%g3, 0, %otherwin					! Single	Group+4bubbles
+		wrpr			%g2, 0, %wstate						! Single	Group+4bubbles
+		rdpr			%tstate, %l3						! Single	Group
+		ldxa			[%l4] ASI_DMMU, %g2					! Load		Group
+		stxa			%g0, [%l4] ASI_DMMU					! Store		Group
+		stxa			%g2, [%l5] ASI_DMMU					! Store		Group
+		flush			%g6							! Single	Group+9bubbles
+		andcc			%l3, %l2, %g0						! IEU1		Group
+		be,a,pt			%icc, 6f						! CTI
+		 st			%g0, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_FPRS]	! Store
+		rd			%fprs, %l0						! Single	Group+4bubbles
+		andcc			%l0, FPRS_FEF, %g0					! IEU1		Group
+		be,pn			%icc, 6f						! CTI
+		 st			%l0, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_FPRS]	! Store
+		ld			[%g6 + AOFF_task_tss + AOFF_thread_flags], %l4		! Load		Group
+		stx			%fsr, [%sp + FPU_OFF + 0x100] 				! Single	Group
+		or			%l4, %l0, %l4						! IEU0		Group
+		ba,pt			%xcc, 3f						! CTI
+		 st			%l4, [%g6 + AOFF_task_tss + AOFF_thread_flags]		! Store
+2:		rd			%fprs, %l0						! Single	Group+4bubbles
+		andcc			%l0, FPRS_FEF, %g0					! IEU1		Group
+		be,pn			%icc, 6f						! CTI
+		 st			%l0, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_FPRS]	! Store
+		stx			%fsr, [%sp + FPU_OFF + 0x100] 				! Single	Group
+3:		rd			%gsr, %l7						! Single	Group+4bubbles
+		cmp			%l0, FPRS_FEF						! IEU1		Group
+		be,pn			%icc, 6f						! CTI
+		 stx			%l7, [%sp + FPU_OFF + 0x108]				! Store		
+		wr			%g0, ASI_BLK_P, %asi					! Singe		Group+4bubbles
+		andcc			%l0, FPRS_DL, %g0					! IEU1		Group
+		be,pn			%icc, 4f						! CTI
+		 membar			#StoreStore | #LoadStore				! Memory	
+		stda			%f0, [%sp + FPU_OFF + 0x000] %asi			! Store		Group
+		stda			%f16, [%sp + FPU_OFF + 0x040] %asi			! Store		Group
+		andcc			%l0, FPRS_DU, %g0					! IEU1
+		be,pn			%icc, 5f						! CTI
+		 nop										! IEU0		Group
+4:		stda			%f32, [%sp + FPU_OFF + 0x080] %asi			! Store		Group
+		stda			%f48, [%sp + FPU_OFF + 0x0c0] %asi			! Store		Group
+5:		membar			#Sync							! Memory	
+6:		wr			%g0, 0x0, %fprs						! Single	Group+4bubbles
+		wrpr			%g0, 0x0, %tl						! Single	Group+4bubbles
+		mov			%g1, %l1						! IEU0		Group
+		mov			%g4, %l4						! IEU1
+		mov			%g5, %l5						! IEU0		Group
+		mov			%g7, %l2						! IEU1
+		mov			%g6, %l6						! IEU0		Group
+		wrpr			%l1, (PSTATE_AG|PSTATE_RMO), %pstate			! Single	Group+4bubbles
+		stx			%g1, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G1]		! Store		Group
+		stx			%g2, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G2]		! Store		Group
+		stx			%g3, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G3]		! Store		Group
+		stx			%g4, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G4]		! Store		Group
+		stx			%g5, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G5]		! Store		Group
+		stx			%g6, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G6]		! Store		Group
+		stx			%g7, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_G7]		! Store		Group
+		stx			%i0, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I0]		! Store		Group
+		stx			%i1, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I1]		! Store		Group
+		stx			%i2, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I2]		! Store		Group
+		stx			%i3, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I3]		! Store		Group
+		stx			%i4, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I4]		! Store		Group
+		sethi			%uhi(PAGE_OFFSET), %g4					! IEU0
+		stx			%i5, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I5]		! Store		Group
+		stx			%i6, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I6]		! Store		Group
+		sllx			%g4, 32, %g4						! IEU0
+		stx			%i7, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I7]		! Store		Group
+		wrpr			%l1, (PSTATE_IE|PSTATE_AG|PSTATE_RMO), %pstate		! Single	Group+4bubbles
+		jmpl			%l2 + 0x4, %g0						! CTI		Group
+		 mov			%l6, %g6						! IEU0
 
-		b,pt			%xcc, 1b
-		 mov			FPRS_FEF, %g3
-		nop
-etraptl1:	rdpr			%tstate, %g1
-		sub			%sp, REGWIN_SZ + TRACEREG_SZ - STACK_BIAS, %g2
-		ba,pt			%xcc, etrap_maybe_fpu
-		 andcc			%g1, TSTATE_PRIV, %g0
-		nop
+etraptl1:	rdpr			%tstate, %g1						! Single	Group+4bubbles
+		ba,pt			%xcc, etrap_priv					! CTI		Group
+		 clr			%g3							! IEU0
 #undef TASK_REGOFF
 #undef FPUREG_SZ
diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S
index 0ed975aff..43f950b25 100644
--- a/arch/sparc64/kernel/head.S
+++ b/arch/sparc64/kernel/head.S
@@ -1,4 +1,4 @@
-/* $Id: head.S,v 1.43 1997/07/07 03:05:25 davem Exp $
+/* $Id: head.S,v 1.46 1997/08/08 08:33:30 jj Exp $
  * head.S: Initial boot code for the Sparc64 port of Linux.
  *
  * Copyright (C) 1996,1997 David S. Miller (davem@caip.rutgers.edu)
@@ -21,6 +21,7 @@
 #include <asm/processor.h>
 #include <asm/lsu.h>
 #include <asm/head.h>
+#include <asm/ttable.h>
 	
 /* This section from from _start to sparc64_boot_end should fit into
  * 0x0000.0000.0040.4000 to 0x0000.0000.0040.8000 and will be sharing space
@@ -89,6 +90,28 @@ sparc64_boot:
          */
 	wrpr    %g0, (PSTATE_PRIV|PSTATE_PEF|PSTATE_IE), %pstate
 
+#ifdef __SMP__
+	/* Ugly but necessary... */
+	sethi	%hi(KERNBASE), %g7
+	sethi	%hi(sparc64_cpu_startup), %g5
+	or	%g5, %lo(sparc64_cpu_startup), %g5
+	sub	%g5, %g7, %g5
+	sethi	%hi(sparc64_cpu_startup_end), %g6
+	or	%g6, %lo(sparc64_cpu_startup_end), %g6
+	sub	%g6, %g7, %g6
+	sethi	%hi(smp_trampoline), %g3
+	or	%g3, %lo(smp_trampoline), %g3
+	sub	%g3, %g7, %g3
+1:	ldx	[%g5], %g1
+	stx	%g1, [%g3]
+	membar	#StoreStore
+	flush	%g3
+	add	%g5, 8, %g5
+	cmp	%g5, %g6
+	blu,pt	%xcc, 1b
+	 add	%g3, 8, %g3
+#endif
+
 create_mappings:
 	/* %g5 holds the tlb data */
         sethi   %uhi(_PAGE_VALID | _PAGE_SZ4MB), %g5
@@ -340,7 +363,7 @@ setup_tba:
 	mov	0x40, %g2			/* INTR data 0 register */
 
 	/* Ok, we're done setting up all the state our trap mechanims needs,
-	 * now get back into normal globals and let the PROM know what it up.
+	 * now get back into normal globals and let the PROM know what is up.
 	 */
 	wrpr	%g0, %g0, %wstate
 	wrpr	%o1, PSTATE_IE, %pstate
@@ -374,6 +397,7 @@ bootup_kernel_stack:
 ! 0x0000000000408000
 
 #include "ttable.S"
+#include "systbls.S"
 #include "etrap.S"
 #include "rtrap.S"
 #include "winfixup.S"
diff --git a/arch/sparc64/kernel/ioctl32.c b/arch/sparc64/kernel/ioctl32.c
index 81eb45e42..af88ca4b6 100644
--- a/arch/sparc64/kernel/ioctl32.c
+++ b/arch/sparc64/kernel/ioctl32.c
@@ -1,4 +1,4 @@
-/* $Id: ioctl32.c,v 1.14 1997/07/17 06:21:12 davem Exp $
+/* $Id: ioctl32.c,v 1.18 1997/09/06 02:25:13 davem Exp $
  * ioctl32.c: Conversion between 32bit and 64bit native ioctls.
  *
  * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
@@ -23,6 +23,7 @@
 #include <linux/vt.h>
 #include <linux/fs.h>
 #include <linux/fd.h>
+#include <linux/if_ppp.h>
 
 #include <asm/types.h>
 #include <asm/uaccess.h>
@@ -102,6 +103,7 @@ static inline int dev_ifconf(unsigned int fd, u32 arg)
 
 	if (copy_from_user(&ifc32, (struct ifconf32 *)A(arg), sizeof(struct ifconf32)))
 		return -EFAULT;
+
 	ifc.ifc_len = ((ifc32.ifc_len / sizeof (struct ifreq32)) + 1) * sizeof (struct ifreq);
 	ifc.ifc_buf = kmalloc (ifc.ifc_len, GFP_KERNEL);
 	if (!ifc.ifc_buf) return -ENOMEM;
@@ -145,7 +147,8 @@ static inline int dev_ifsioc(unsigned int fd, unsigned int cmd, u32 arg)
 	unsigned long old_fs;
 	int err;
 	
-	if (cmd == SIOCSIFMAP) {
+	switch (cmd) {
+	case SIOCSIFMAP:
 		if (copy_from_user(&ifr, (struct ifreq32 *)A(arg), sizeof(ifr.ifr_name)) ||
 		    __get_user(ifr.ifr_map.mem_start, &(((struct ifreq32 *)A(arg))->ifr_ifru.ifru_map.mem_start)) ||
 		    __get_user(ifr.ifr_map.mem_end, &(((struct ifreq32 *)A(arg))->ifr_ifru.ifru_map.mem_end)) ||
@@ -154,9 +157,20 @@ static inline int dev_ifsioc(unsigned int fd, unsigned int cmd, u32 arg)
 		    __get_user(ifr.ifr_map.dma, &(((struct ifreq32 *)A(arg))->ifr_ifru.ifru_map.dma)) ||
 		    __get_user(ifr.ifr_map.port, &(((struct ifreq32 *)A(arg))->ifr_ifru.ifru_map.port)))
 			return -EFAULT;
-	} else {
+		break;
+	case SIOCGPPPSTATS:
+	case SIOCGPPPCSTATS:
+	case SIOCGPPPVER:
+		if (copy_from_user(&ifr, (struct ifreq32 *)A(arg), sizeof(struct ifreq32)))
+			return -EFAULT;
+		ifr.ifr_data = (__kernel_caddr_t)get_free_page(GFP_KERNEL);
+		if (!ifr.ifr_data)
+			return -EAGAIN;
+		break;
+	default:
 		if (copy_from_user(&ifr, (struct ifreq32 *)A(arg), sizeof(struct ifreq32)))
 			return -EFAULT;
+		break;
 	}
 	old_fs = get_fs();
 	set_fs (KERNEL_DS);
@@ -177,6 +191,25 @@ static inline int dev_ifsioc(unsigned int fd, unsigned int cmd, u32 arg)
 			if (copy_to_user((struct ifreq32 *)A(arg), &ifr, sizeof(struct ifreq32)))
 				return -EFAULT;
 			break;
+		case SIOCGPPPSTATS:
+		case SIOCGPPPCSTATS:
+		case SIOCGPPPVER:
+		{
+			u32 data;
+			int len;
+
+			__get_user(data, &(((struct ifreq32 *)A(arg))->ifr_ifru.ifru_data));
+			if(cmd == SIOCGPPPVER)
+				len = strlen(PPP_VERSION) + 1;
+			else if(cmd == SIOCGPPPCSTATS)
+				len = sizeof(struct ppp_comp_stats);
+			else
+				len = sizeof(struct ppp_stats);
+
+			if (copy_to_user((char *)A(data), ifr.ifr_data, len))
+				return -EFAULT;
+			break;
+		}
 		case SIOCGIFMAP:
 			if (copy_to_user((struct ifreq32 *)A(arg), &ifr, sizeof(ifr.ifr_name)) ||
 			    __put_user(ifr.ifr_map.mem_start, &(((struct ifreq32 *)A(arg))->ifr_ifru.ifru_map.mem_start)) ||
@@ -481,6 +514,74 @@ static inline int fbiogscursor(unsigned int fd, unsigned int cmd, u32 arg)
 	return ret;
 }
 	
+static int hdio_ioctl_trans(unsigned int fd, unsigned int cmd, u32 arg)
+{
+	unsigned long old_fs = get_fs();
+	unsigned long kval;
+	unsigned int *uvp;
+	int error;
+
+	set_fs(KERNEL_DS);
+	error = sys_ioctl(fd, cmd, (long)&kval);
+	set_fs(old_fs);
+
+	if(error == 0) {
+		uvp = (unsigned int *)A(arg);
+		if(put_user(kval, uvp))
+			error = -EFAULT;
+	}
+	return error;
+}
+
+struct ppp_option_data32 {
+	__kernel_caddr_t32	ptr;
+	__u32			length;
+	int			transmit;
+};
+#define PPPIOCSCOMPRESS32	_IOW('t', 77, struct ppp_option_data32)
+
+static int ppp_ioctl(unsigned int fd, unsigned int cmd, u32 arg)
+{
+	unsigned long old_fs = get_fs();
+	struct ppp_option_data32 data32;
+	struct ppp_option_data data;
+	int err;
+
+	switch (cmd) {
+	case PPPIOCSCOMPRESS32:
+		if (copy_from_user(&data32, (struct ppp_option_data32 *)A(arg), sizeof(struct ppp_option_data32)))
+			return -EFAULT;
+		data.ptr = kmalloc (data32.length, GFP_KERNEL);
+		if (!data.ptr)
+			return -ENOMEM;
+		if (copy_from_user(data.ptr, (__u8 *)A(data32.ptr), data32.length)) {
+			err = -EFAULT;
+			goto out;
+		}
+		data.length = data32.length;
+		data.transmit = data32.transmit;
+		break;
+	default:
+		printk("ppp_ioctl: Unknown cmd fd(%d) cmd(%08x) arg(%08x)\n",
+		       (int)fd, (unsigned int)cmd, (unsigned int)arg);
+		return -EINVAL;
+	}
+	old_fs = get_fs();
+	set_fs (KERNEL_DS);
+	err = sys_ioctl (fd, cmd, (unsigned long)&data);
+	set_fs (old_fs);
+	if (err)
+		goto out;
+	switch (cmd) {
+	case PPPIOCSCOMPRESS32:
+	default:
+		break;
+	}
+out:
+	kfree(data.ptr);
+	return err;
+}
+
 asmlinkage int sys32_ioctl(unsigned int fd, unsigned int cmd, u32 arg)
 {
 	struct file * filp;
@@ -527,6 +628,9 @@ asmlinkage int sys32_ioctl(unsigned int fd, unsigned int cmd, u32 arg)
 	case SIOCSIFDSTADDR:
 	case SIOCGIFNETMASK:
 	case SIOCSIFNETMASK:
+	case SIOCGPPPSTATS:
+	case SIOCGPPPCSTATS:
+	case SIOCGPPPVER:
 		error = dev_ifsioc(fd, cmd, arg);
 		goto out;
 		
@@ -557,6 +661,16 @@ asmlinkage int sys32_ioctl(unsigned int fd, unsigned int cmd, u32 arg)
 		error = fbiogscursor(fd, cmd, arg);
 		goto out;
 
+	case HDIO_GET_KEEPSETTINGS:
+	case HDIO_GET_UNMASKINTR:
+	case HDIO_GET_DMA:
+	case HDIO_GET_32BIT:
+	case HDIO_GET_MULTCOUNT:
+	case HDIO_GET_NOWERR:
+	case HDIO_GET_NICE:
+		error = hdio_ioctl_trans(fd, cmd, arg);
+		goto out;
+
 	/* List here exlicitly which ioctl's are known to have
 	 * compatable types passed or none at all...
 	 */
@@ -619,6 +733,23 @@ asmlinkage int sys32_ioctl(unsigned int fd, unsigned int cmd, u32 arg)
 	case FIBMAP:
 	case FIGETBSZ:
 	
+	/* 0x03 -- HD/IDE ioctl's used by hdparm and friends.
+	 *         Some need translations, these do not.
+	 */
+	case HDIO_GET_IDENTITY:
+	case HDIO_SET_DMA:
+	case HDIO_SET_KEEPSETTINGS:
+	case HDIO_SET_UNMASKINTR:
+	case HDIO_SET_NOWERR:
+	case HDIO_SET_32BIT:
+	case HDIO_SET_MULTCOUNT:
+	case HDIO_DRIVE_CMD:
+	case HDIO_SET_PIO_MODE:
+	case HDIO_SCAN_HWIF:
+	case HDIO_SET_NICE:
+	case BLKROSET:
+	case BLKROGET:
+
 	/* 0x02 -- Floppy ioctls */
 	case FDSETEMSGTRESH:
 	case FDFLUSH:
@@ -729,11 +860,40 @@ asmlinkage int sys32_ioctl(unsigned int fd, unsigned int cmd, u32 arg)
 	case SIOCSARP:
 	case SIOCGARP:
 	case SIOCDARP:
+	case OLD_SIOCSARP:
+	case OLD_SIOCGARP:
+	case OLD_SIOCDARP:
+	case SIOCSRARP:
+	case SIOCGRARP:
+	case SIOCDRARP:
 	case SIOCADDDLCI:
 	case SIOCDELDLCI:
+
+	/* PPP stuff */
+	case PPPIOCGFLAGS:
+	case PPPIOCSFLAGS:
+	case PPPIOCGASYNCMAP:
+	case PPPIOCSASYNCMAP:
+	case PPPIOCGUNIT:
+	case PPPIOCGRASYNCMAP:
+	case PPPIOCSRASYNCMAP:
+	case PPPIOCGMRU:
+	case PPPIOCSMRU:
+	case PPPIOCSMAXCID:
+	case PPPIOCGXASYNCMAP:
+	case PPPIOCSXASYNCMAP:
+	case PPPIOCXFERUNIT:
+	case PPPIOCGNPMODE:
+	case PPPIOCSNPMODE:
+	case PPPIOCGDEBUG:
+	case PPPIOCSDEBUG:
+	case PPPIOCGIDLE:
 		error = sys_ioctl (fd, cmd, (unsigned long)arg);
 		goto out;
-		break;
+
+	case PPPIOCSCOMPRESS32:
+		error = ppp_ioctl (fd, cmd, (unsigned long)arg);
+		goto out;
 
 	default:
 		printk("sys32_ioctl: Unknown cmd fd(%d) cmd(%08x) arg(%08x)\n",
diff --git a/arch/sparc64/kernel/ioport.c b/arch/sparc64/kernel/ioport.c
index 7d1580b39..7ae12df11 100644
--- a/arch/sparc64/kernel/ioport.c
+++ b/arch/sparc64/kernel/ioport.c
@@ -1,4 +1,4 @@
-/* $Id: ioport.c,v 1.11 1997/07/22 06:14:04 davem Exp $
+/* $Id: ioport.c,v 1.13 1997/08/18 01:20:22 davem Exp $
  * ioport.c:  Simple io mapping allocator.
  *
  * Copyright (C) 1995,1996 David S. Miller (davem@caip.rutgers.edu)
@@ -45,7 +45,7 @@ void *sparc_alloc_io (u32 address, void *virtual, int len, char *name,
 		      u32 bus_type, int rdonly)
 {
 	unsigned long vaddr, base_address;
-	unsigned long addr = ((unsigned long) address) + (((unsigned long) bus_type) << 32);
+	unsigned long addr = ((unsigned long)address) + (((unsigned long)bus_type)<<32);
 	unsigned long offset = (addr & (~PAGE_MASK));
 
 	if (virtual) {
@@ -64,7 +64,12 @@ void *sparc_alloc_io (u32 address, void *virtual, int len, char *name,
 		/* Tell Linux resource manager about the mapping */
 		request_region ((vaddr | offset), len, name);
 	} else {
-		return __va(addr);
+		unsigned long vaddr = (unsigned long) __va(addr);
+
+		if(!check_region(vaddr, len))
+			request_region(vaddr, len, name);
+
+		return (void *) vaddr;
 	}
 
 	base_address = vaddr;
@@ -83,11 +88,11 @@ void sparc_free_io (void *virtual, int len)
 	unsigned long vaddr = (unsigned long) virtual & PAGE_MASK;
 	unsigned long plen = (((unsigned long)virtual & ~PAGE_MASK) + len + PAGE_SIZE-1) & PAGE_MASK;
 	
+	release_region(vaddr, plen);
+
 	if (((unsigned long)virtual) >= PAGE_OFFSET + 0x10000000000UL)
 		return;
 
-	release_region(vaddr, plen);
-
 	for (; plen != 0;) {
 		plen -= PAGE_SIZE;
 		unmapioaddr(vaddr + plen);
diff --git a/arch/sparc64/kernel/irq.c b/arch/sparc64/kernel/irq.c
index f76c27c57..6f4c9dfdf 100644
--- a/arch/sparc64/kernel/irq.c
+++ b/arch/sparc64/kernel/irq.c
@@ -1,4 +1,4 @@
-/* $Id: irq.c,v 1.19 1997/07/24 12:15:04 davem Exp $
+/* $Id: irq.c,v 1.39 1997/08/31 03:11:18 davem Exp $
  * irq.c: UltraSparc IRQ handling/init/registry.
  *
  * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
@@ -28,8 +28,18 @@
 #include <asm/hardirq.h>
 #include <asm/softirq.h>
 
+#ifdef CONFIG_PCI
+#include <linux/pci.h>
+#include <asm/pbm.h>
+#endif
+
 /* Internal flag, should not be visible elsewhere at all. */
-#define SA_SYSIO_MASKED		0x100
+#define SA_IMAP_MASKED		0x100
+
+#ifdef __SMP__
+void distribute_irqs(void);
+static int irqs_have_been_distributed = 0;
+#endif
 
 /* UPA nodes send interrupt packet to UltraSparc with first data reg value
  * low 5 bits holding the IRQ identifier being delivered.  We must translate
@@ -37,10 +47,39 @@
  * make things even more swift we store the complete mask here.
  */
 
-#define NUM_IVECS	2048	/* XXX may need more on sunfire/wildfire */
+#define NUM_HARD_IVECS	2048
+#define NUM_IVECS	(NUM_HARD_IVECS + 64)	/* For SMP IRQ distribution alg. */
 
 unsigned long ivector_to_mask[NUM_IVECS];
 
+struct ino_bucket {
+	struct ino_bucket *next;
+	unsigned int ino;
+	unsigned int *imap;
+	unsigned int *iclr;
+};
+
+#define INO_HASHSZ	(NUM_HARD_IVECS >> 2)
+#define NUM_INO_STATIC	4
+static struct ino_bucket *ino_hash[INO_HASHSZ] = { NULL, };
+static struct ino_bucket static_ino_buckets[NUM_INO_STATIC];
+static int static_ino_bucket_count = 0;
+
+static inline struct ino_bucket *__ino_lookup(unsigned int hash, unsigned int ino)
+{
+	struct ino_bucket *ret = ino_hash[hash];
+
+	for(ret = ino_hash[hash]; ret && ret->ino != ino; ret = ret->next)
+		;
+
+	return ret;
+}
+
+static inline struct ino_bucket *ino_lookup(unsigned int ino)
+{
+	return __ino_lookup((ino & (INO_HASHSZ - 1)), ino);
+}
+
 /* This is based upon code in the 32-bit Sparc kernel written mostly by
  * David Redman (djhr@tadpole.co.uk).
  */
@@ -73,15 +112,22 @@ int get_irq_list(char *buf)
 		}
 		len += sprintf(buf + len, "\n");
 	}
+#if 0
+#ifdef CONFIG_PCI
+	len += sprintf(buf + len, "ISTAT: PCI[%016lx] OBIO[%016lx]\n",
+		       psycho_root->psycho_regs->pci_istate,
+		       psycho_root->psycho_regs->obio_istate);
+#endif
+#endif
 	return len;
 }
 
-/* INO number to Sparc PIL level. */
-unsigned char ino_to_pil[] = {
-	0, 1, 2, 3, 5, 7, 8, 9,		/* SBUS slot 0 */
-	0, 1, 2, 3, 5, 7, 8, 9,		/* SBUS slot 1 */
-	0, 1, 2, 3, 5, 7, 8, 9,		/* SBUS slot 2 */
-	0, 1, 2, 3, 5, 7, 8, 9,		/* SBUS slot 3 */
+/* SBUS SYSIO INO number to Sparc PIL level. */
+unsigned char sysio_ino_to_pil[] = {
+	0, 1, 2, 7, 5, 7, 8, 9,		/* SBUS slot 0 */
+	0, 1, 2, 7, 5, 7, 8, 9,		/* SBUS slot 1 */
+	0, 1, 2, 7, 5, 7, 8, 9,		/* SBUS slot 2 */
+	0, 1, 2, 7, 5, 7, 8, 9,		/* SBUS slot 3 */
 	3, /* Onboard SCSI */
 	5, /* Onboard Ethernet */
 /*XXX*/	8, /* Onboard BPP */
@@ -112,7 +158,7 @@ unsigned char ino_to_pil[] = {
  */
 #define offset(x) ((unsigned long)(&(((struct sysio_regs *)0)->x)))
 #define bogon     ((unsigned long) -1)
-static unsigned long irq_offsets[] = {
+static unsigned long sysio_irq_offsets[] = {
 /* SBUS Slot 0 --> 3, level 1 --> 7 */
 offset(imap_slot0),offset(imap_slot0),offset(imap_slot0),offset(imap_slot0),
 offset(imap_slot0),offset(imap_slot0),offset(imap_slot0),offset(imap_slot0),
@@ -134,29 +180,29 @@ offset(imap_pmgmt),
 
 #undef bogon
 
-#define NUM_IRQ_ENTRIES (sizeof(irq_offsets) / sizeof(irq_offsets[0]))
+#define NUM_SYSIO_OFFSETS (sizeof(sysio_irq_offsets) / sizeof(sysio_irq_offsets[0]))
 
-/* Convert an "interrupts" property IRQ level to an SBUS/SYSIO
- * Interrupt Mapping register pointer, or NULL if none exists.
+/* XXX Old compatability cruft, get rid of me when all drivers have been
+ * XXX converted to dcookie registry calls... -DaveM
  */
-static unsigned int *irq_to_imap(unsigned int irq)
+static unsigned int *sysio_irq_to_imap(unsigned int irq)
 {
 	unsigned long offset;
 	struct sysio_regs *sregs;
 
 	if((irq == 14) ||
-	   (irq >= NUM_IRQ_ENTRIES) ||
-	   ((offset = irq_offsets[irq]) == ((unsigned long)-1)))
+	   (irq >= NUM_SYSIO_OFFSETS) ||
+	   ((offset = sysio_irq_offsets[irq]) == ((unsigned long)-1)))
 		return NULL;
 	sregs = SBus_chain->iommu->sysio_regs;
 	offset += ((unsigned long) sregs);
-	return ((unsigned int *)offset) + 1;
+	return ((unsigned int *)offset);
 }
 
 /* Convert Interrupt Mapping register pointer to assosciated
- * Interrupt Clear register pointer.
+ * Interrupt Clear register pointer, SYSIO specific version.
  */
-static unsigned int *imap_to_iclr(unsigned int *imap)
+static unsigned int *sysio_imap_to_iclr(unsigned int *imap)
 {
 	unsigned long diff;
 
@@ -166,32 +212,68 @@ static unsigned int *imap_to_iclr(unsigned int *imap)
 
 #undef offset
 
-/* For non-SBUS IRQ's we do nothing, else we must enable them in the
- * appropriate SYSIO interrupt map registers.
+#ifdef CONFIG_PCI
+/* PCI PSYCHO INO number to Sparc PIL level. */
+unsigned char psycho_ino_to_pil[] = {
+	7, 5, 5, 2,			/* PCI A slot 0  Int A, B, C, D */
+	7, 5, 5, 2,			/* PCI A slot 1  Int A, B, C, D */
+	0, 0, 0, 0,
+	0, 0, 0, 0,
+	6, 4, 3, 1,			/* PCI B slot 0  Int A, B, C, D */
+	6, 4, 3, 1,			/* PCI B slot 1  Int A, B, C, D */
+	6, 4, 3, 1,			/* PCI B slot 2  Int A, B, C, D */
+	6, 4, 3, 1,			/* PCI B slot 3  Int A, B, C, D */
+	3,  /* SCSI */
+	5,  /* Ethernet */
+	8,  /* Parallel Port */
+	13, /* Audio Record */
+	14, /* Audio Playback */
+	15, /* PowerFail */
+	12, /* Keyboard/Mouse/Serial */
+	11, /* Floppy */
+	2,  /* Spare Hardware */
+	12, /* Keyboard */
+	4,  /* Mouse */
+	12, /* Serial */
+	10, /* Timer 0 */
+	11, /* Timer 1 */
+	15, /* Uncorrectable ECC */
+	15, /* Correctable ECC */
+	15, /* PCI Bus A Error */
+	15, /* PCI Bus B Error */
+	1, /* Power Management */
+};
+
+/* INO number to IMAP register offset for PSYCHO external IRQ's.
  */
-void enable_irq(unsigned int irq)
+#define psycho_offset(x) ((unsigned long)(&(((struct psycho_regs *)0)->x)))
+
+#define psycho_imap_offset(ino)							\
+	((ino & 0x20) ? (psycho_offset(imap_scsi) + (((ino) & 0x1f) << 3)) :	\
+			(psycho_offset(imap_a_slot0) + (((ino) & 0x3c) << 1)))
+
+#define psycho_iclr_offset(ino)							\
+	((ino & 0x20) ? (psycho_offset(iclr_scsi) + (((ino) & 0x1f) << 3)) :	\
+			(psycho_offset(iclr_a_slot0[0]) + (((ino) & 0x1f) << 3)))
+
+#endif
+
+/* Now these are always passed a true fully specified sun4u INO. */
+void enable_irq(unsigned int ino)
 {
+	struct ino_bucket *bucket;
 	unsigned long tid;
 	unsigned int *imap;
 
-	/* If this is for the tick interrupt, just ignore, note
-	 * that this is the one and only locally generated interrupt
-	 * source, all others come from external sources (essentially
-	 * any UPA device which is an interruptor).  (actually, on
-	 * second thought Ultra can generate local interrupts for
-	 * async memory errors and we may setup handlers for those
-	 * at some point as well)
-	 *
-	 * XXX See commentary below in request_irq() this assumption
-	 * XXX is broken and needs to be fixed.
-	 */
-	if(irq == 14)
+#ifdef CONFIG_PCI
+	if(PCI_IRQ_P(ino))
+		ino &= (PCI_IRQ_IGN | PCI_IRQ_INO);
+#endif
+	bucket = ino_lookup(ino);
+	if(!bucket)
 		return;
 
-	/* Check for bogons. */
-	imap = irq_to_imap(irq);
-	if(imap == NULL)
-		goto do_the_stb_watoosi;
+	imap = bucket->imap;
 
 	/* We send it to our UPA MID, for SMP this will be different. */
 	__asm__ __volatile__("ldxa [%%g0] %1, %0" : "=r" (tid) : "i" (ASI_UPA_CONFIG));
@@ -202,28 +284,27 @@ void enable_irq(unsigned int irq)
 	 * Register, the hardware just mirrors that value here.
 	 * However for Graphics and UPA Slave devices the full
 	 * SYSIO_IMAP_INR field can be set by the programmer here.
-	 * (XXX we will have to handle those for FFB etc. XXX)
+	 *
+	 * Things like FFB can now be handled via the dcookie mechanism.
 	 */
 	*imap = SYSIO_IMAP_VALID | (tid & SYSIO_IMAP_TID);
-	return;
-
-do_the_stb_watoosi:
-	printk("Cannot enable irq(%d), doing the \"STB Watoosi\" instead.", irq);
-	panic("Trying to enable bogon IRQ");
 }
 
-void disable_irq(unsigned int irq)
+/* This now gets passed true ino's as well. */
+void disable_irq(unsigned int ino)
 {
+	struct ino_bucket *bucket;
 	unsigned int *imap;
 
-	/* XXX Grrr, I know this is broken... */
-	if(irq == 14)
+#ifdef CONFIG_PCI
+	if(PCI_IRQ_P(ino))
+		ino &= (PCI_IRQ_IGN | PCI_IRQ_INO);
+#endif
+	bucket = ino_lookup(ino);
+	if(!bucket)
 		return;
 
-	/* Check for bogons. */
-	imap = irq_to_imap(irq);
-	if(imap == NULL)
-		goto do_the_stb_watoosi;
+	imap = bucket->imap;
 
 	/* NOTE: We do not want to futz with the IRQ clear registers
 	 *       and move the state to IDLE, the SCSI code does call
@@ -231,34 +312,254 @@ void disable_irq(unsigned int irq)
 	 *       SCSI adapter driver code.  Thus we'd lose interrupts.
 	 */
 	*imap &= ~(SYSIO_IMAP_VALID);
-	return;
+}
+
+static void get_irq_translations(int *cpu_irq, int *ivindex_fixup,
+				 unsigned int **imap, unsigned int **iclr,
+				 void *busp, unsigned long flags,
+				 unsigned int irq)
+{
+	if(*cpu_irq != -1 && *imap != NULL && *iclr != NULL)
+		return;
+
+	if(*cpu_irq != -1 || *imap != NULL || *iclr != NULL || busp == NULL) {
+		printk("get_irq_translations: Partial specification, this is bad.\n");
+		printk("get_irq_translations: cpu_irq[%d] imap[%p] iclr[%p] busp[%p]\n",
+		       *cpu_irq, *imap, *iclr, busp);
+		panic("Bad IRQ translations...");
+	}
+
+	if(SA_BUS(flags) == SA_SBUS) {
+		struct linux_sbus *sbusp = busp;
+		struct sysio_regs *sregs = sbusp->iommu->sysio_regs;
+		unsigned long offset;
 
-do_the_stb_watoosi:
-	printk("Cannot disable irq(%d), doing the \"STB Watoosi\" instead.", irq);
-	panic("Trying to enable bogon IRQ");
+		*cpu_irq = sysio_ino_to_pil[irq];
+		if(*cpu_irq == 0) {
+			printk("get_irq_translations: Bad SYSIO INO[%x]\n", irq);
+			panic("Bad SYSIO IRQ translations...");
+		}
+		offset = sysio_irq_offsets[irq];
+		if(offset == ((unsigned long)-1)) {
+			printk("get_irq_translations: Bad SYSIO INO[%x] cpu[%d]\n",
+			       irq, *cpu_irq);
+			panic("BAD SYSIO IRQ offset...");
+		}
+		offset += ((unsigned long)sregs);
+		*imap = ((unsigned int *)offset);
+
+		/* SYSIO inconsistancy.  For external SLOTS, we have to select
+		 * the right ICLR register based upon the lower SBUS irq level
+		 * bits.
+		 */
+		if(irq >= 0x20) {
+			*iclr = sysio_imap_to_iclr(*imap);
+		} else {
+			unsigned long iclraddr;
+			int sbus_slot = (irq & 0x18)>>3;
+			int sbus_level = irq & 0x7;
+
+			switch(sbus_slot) {
+			case 0:
+				*iclr = &sregs->iclr_slot0;
+				break;
+			case 1:
+				*iclr = &sregs->iclr_slot1;
+				break;
+			case 2:
+				*iclr = &sregs->iclr_slot2;
+				break;
+			case 3:
+				*iclr = &sregs->iclr_slot3;
+				break;
+			};
+
+			iclraddr = (unsigned long) *iclr;
+			iclraddr += ((sbus_level - 1) * 8);
+			*iclr = (unsigned int *) iclraddr;
+
+#if 0 /* DEBUGGING */
+			printk("SYSIO_FIXUP: slot[%x] level[%x] iclr[%p] ",
+			       sbus_slot, sbus_level, *iclr);
+#endif
+
+			/* Also, make sure this is accounted for in ivindex
+			 * computations done by the caller.
+			 */
+			*ivindex_fixup = sbus_level;
+		}
+		return;
+	}
+#ifdef CONFIG_PCI
+	if(SA_BUS(flags) == SA_PCI) {
+		struct pci_bus *pbusp = busp;
+		struct linux_pbm_info *pbm = pbusp->sysdata;
+		struct psycho_regs *pregs = pbm->parent->psycho_regs;
+		unsigned long offset;
+
+		*cpu_irq = psycho_ino_to_pil[irq & 0x3f];
+		if(*cpu_irq == 0) {
+			printk("get_irq_translations: Bad PSYCHO INO[%x]\n", irq);
+			panic("Bad PSYCHO IRQ translations...");
+		}
+		offset = psycho_imap_offset(irq);
+		if(offset == ((unsigned long)-1)) {
+			printk("get_irq_translations: Bad PSYCHO INO[%x] cpu[%d]\n",
+			       irq, *cpu_irq);
+			panic("Bad PSYCHO IRQ offset...");
+		}
+		offset += ((unsigned long)pregs);
+		*imap = ((unsigned int *)offset) + 1;
+		*iclr = (unsigned int *)
+			(((unsigned long)pregs) + psycho_imap_offset(irq));
+		return;
+	}
+#endif
+#if 0	/* XXX More to do before we can use this. -DaveM */
+	if(SA_BUS(flags) == SA_FHC) {
+		struct fhc_bus *fbusp = busp;
+		struct fhc_regs *fregs = fbusp->regs;
+		unsigned long offset;
+
+		*cpu_irq = fhc_ino_to_pil[irq];
+		if(*cpu_irq == 0) {
+			printk("get_irq_translations: Bad FHC INO[%x]\n", irq);
+			panic("Bad FHC IRQ translations...");
+		}
+		offset = fhc_irq_offset[*cpu_irq];
+		if(offset == ((unsigned long)-1)) {
+			printk("get_irq_translations: Bad FHC INO[%x] cpu[%d]\n",
+			       irq, *cpu_irq);
+			panic("Bad FHC IRQ offset...");
+		}
+		offset += ((unsigned long)pregs);
+		*imap = (((unsigned int *)offset)+1);
+		*iclr = fhc_imap_to_iclr(*imap);
+		return;
+	}
+#endif
+	printk("get_irq_translations: IRQ register for unknown bus type.\n");
+	printk("get_irq_translations: BUS[%lx] IRQ[%x]\n",
+	       SA_BUS(flags), irq);
+	panic("Bad IRQ bus type...");
+}
+
+#ifdef CONFIG_PCI
+static void pci_irq_frobnicate(int *cpu_irq, int *ivindex_fixup,
+			       unsigned int **imap, unsigned int **iclr,
+			       unsigned int irq)
+{
+	struct linux_psycho *psycho = psycho_root;
+	struct psycho_regs *pregs = psycho->psycho_regs;
+	unsigned long addr, imoff;
+
+	addr = (unsigned long) &pregs->imap_a_slot0;
+	imoff = (irq & PCI_IRQ_IMAP_OFF) >> PCI_IRQ_IMAP_OFF_SHFT;
+	addr = addr + imoff;
+
+	*imap = ((unsigned int *)addr) + 1;
+
+	addr = (unsigned long) pregs;
+	addr += psycho_iclr_offset(irq & (PCI_IRQ_INO));
+	*iclr = ((unsigned int *)addr) + 1;
+
+	*cpu_irq = psycho_ino_to_pil[irq & (PCI_IRQ_INO)];
+	if(*cpu_irq == 0) {
+		printk("get_irq_translations: BAD PSYCHO INO[%x]\n", irq);
+		panic("Bad PSYCHO IRQ frobnication...");
+	}
+
+	/* IVINDEX fixup only needed for PCI slot irq lines. */
+	if(!(irq & 0x20))
+		*ivindex_fixup = irq & 0x03;
+}
+#endif
+
+/* Once added, they are never removed. */
+static struct ino_bucket *add_ino_hash(unsigned int ivindex,
+				       unsigned int *imap, unsigned int *iclr,
+				       unsigned long flags)
+{
+	struct ino_bucket *new = NULL, **hashp;
+	unsigned int hash = (ivindex & (INO_HASHSZ - 1));
+
+	new = __ino_lookup(hash, ivindex);
+	if(new)
+		return new;
+	if(flags & SA_STATIC_ALLOC) {
+		if(static_ino_bucket_count < NUM_INO_STATIC)
+			new = &static_ino_buckets[static_ino_bucket_count++];
+		else
+			printk("Request for ino bucket SA_STATIC_ALLOC failed "
+			       "using kmalloc\n");
+	}
+	if(new == NULL)
+		new = kmalloc(sizeof(struct ino_bucket), GFP_KERNEL);
+	if(new) {
+		hashp = &ino_hash[hash];
+		new->imap = imap;
+		new->iclr = iclr;
+		new->ino  = ivindex;
+		new->next = *hashp;
+		*hashp = new;
+	}
+	return new;
 }
 
 int request_irq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *),
-		unsigned long irqflags, const char *name, void *dev_cookie)
+		unsigned long irqflags, const char *name, void *dev_id)
 {
 	struct irqaction *action, *tmp = NULL;
+	struct devid_cookie *dcookie = NULL;
+	struct ino_bucket *bucket = NULL;
 	unsigned long flags;
-	unsigned int cpu_irq, *imap, *iclr;
+	unsigned int *imap, *iclr;
+	void *bus_id = NULL;
+	int ivindex, ivindex_fixup, cpu_irq = -1;
 	
-	/* XXX This really is not the way to do it, the "right way"
-	 * XXX is to have drivers set SA_SBUS or something like that
-	 * XXX in irqflags and we base our decision here on whether
-	 * XXX that flag bit is set or not.
-	 */
-	if(irq == 14)
-		cpu_irq = irq;
-	else
-		cpu_irq = ino_to_pil[irq];
-
 	if(!handler)
 	    return -EINVAL;
 
-	imap = irq_to_imap(irq);
+	imap = iclr = NULL;
+
+	ivindex_fixup = 0;
+#ifdef CONFIG_PCI
+	if(PCI_IRQ_P(irq)) {
+		pci_irq_frobnicate(&cpu_irq, &ivindex_fixup, &imap, &iclr, irq);
+	} else
+#endif
+	if(irqflags & SA_DCOOKIE) {
+		if(!dev_id) {
+			printk("request_irq: SA_DCOOKIE but dev_id is NULL!\n");
+			panic("Bogus irq registry.");
+		}
+		dcookie		= dev_id;
+		dev_id		= dcookie->real_dev_id;
+		cpu_irq		= dcookie->pil;
+		imap		= dcookie->imap;
+		iclr		= dcookie->iclr;
+		bus_id		= dcookie->bus_cookie;
+		get_irq_translations(&cpu_irq, &ivindex_fixup, &imap,
+				     &iclr, bus_id, irqflags, irq);
+	} else {
+		/* XXX NOTE: This code is maintained for compatability until I can
+		 * XXX       verify that all drivers sparc64 will use are updated
+		 * XXX       to use the new IRQ registry dcookie interface.  -DaveM
+		 */
+		if(irq == 14)
+			cpu_irq = irq;
+		else
+			cpu_irq = sysio_ino_to_pil[irq];
+		imap = sysio_irq_to_imap(irq);
+		if(!imap) {
+			printk("request_irq: BAD, null imap for old style "
+			       "irq registry IRQ[%x].\n", irq);
+			panic("Bad IRQ registery...");
+		}
+		iclr = sysio_imap_to_iclr(imap);
+	}
+	ivindex = (*imap & (SYSIO_IMAP_IGN | SYSIO_IMAP_INO));
+	ivindex += ivindex_fixup;
 
 	action = *(cpu_irq + irq_action);
 	if(action) {
@@ -297,52 +598,67 @@ int request_irq(unsigned int irq, void (*handler)(int, void *, struct pt_regs *)
 		return -ENOMEM;
 	}
 
-	if(imap) {
-		int ivindex = (*imap & (SYSIO_IMAP_IGN | SYSIO_IMAP_INO));
+	bucket = add_ino_hash(ivindex, imap, iclr, irqflags);
+	if(!bucket) {
+		kfree(action);
+		restore_flags(flags);
+		return -ENOMEM;
+	}
 
-		ivector_to_mask[ivindex] = (1<<cpu_irq);
-		iclr = imap_to_iclr(imap);
-		action->mask = (unsigned long) iclr;
-		irqflags |= SA_SYSIO_MASKED;
-	} else {
-		action->mask = 0;
+	ivector_to_mask[ivindex] = (1 << cpu_irq);
+
+	if(dcookie) {
+		dcookie->ret_ino = ivindex;
+		dcookie->ret_pil = cpu_irq;
 	}
 
+	action->mask = (unsigned long) bucket;
 	action->handler = handler;
-	action->flags = irqflags;
+	action->flags = irqflags | SA_IMAP_MASKED;
 	action->name = name;
 	action->next = NULL;
-	action->dev_id = dev_cookie;
+	action->dev_id = dev_id;
 
 	if(tmp)
 		tmp->next = action;
 	else
 		*(cpu_irq + irq_action) = action;
 
-	enable_irq(irq);
+	enable_irq(ivindex);
 	restore_flags(flags);
+#ifdef __SMP__
+	if(irqs_have_been_distributed)
+		distribute_irqs();
+#endif
 	return 0;
 }
 
-void free_irq(unsigned int irq, void *dev_cookie)
+void free_irq(unsigned int irq, void *dev_id)
 {
 	struct irqaction *action;
 	struct irqaction *tmp = NULL;
 	unsigned long flags;
 	unsigned int cpu_irq;
+	int ivindex = -1;
 
-	if(irq == 14)
+	if(irq == 14) {
 		cpu_irq = irq;
-	else
-		cpu_irq = ino_to_pil[irq];
+	} else {
+#ifdef CONFIG_PCI
+		if(PCI_IRQ_P(irq))
+			cpu_irq = psycho_ino_to_pil[irq & PCI_IRQ_INO];
+		else
+#endif
+			cpu_irq = sysio_ino_to_pil[irq];
+	}
 	action = *(cpu_irq + irq_action);
 	if(!action->handler) {
 		printk("Freeing free IRQ %d\n", irq);
 		return;
 	}
-	if(dev_cookie) {
+	if(dev_id) {
 		for( ; action; action = action->next) {
-			if(action->dev_id == dev_cookie)
+			if(action->dev_id == dev_id)
 				break;
 			tmp = action;
 		}
@@ -351,7 +667,7 @@ void free_irq(unsigned int irq, void *dev_cookie)
 			return;
 		}
 	} else if(action->flags & SA_SHIRQ) {
-		printk("Trying to free shared IRQ %d with NULL device cookie\n", irq);
+		printk("Trying to free shared IRQ %d with NULL device ID\n", irq);
 		return;
 	}
 
@@ -367,29 +683,37 @@ void free_irq(unsigned int irq, void *dev_cookie)
 	else
 		*(cpu_irq + irq_action) = action->next;
 
-	if(action->flags & SA_SYSIO_MASKED) {
-		unsigned int *imap = irq_to_imap(irq);
-		if(imap != NULL)
-			ivector_to_mask[*imap & (SYSIO_IMAP_IGN | SYSIO_IMAP_INO)] = 0;
+	if(action->flags & SA_IMAP_MASKED) {
+		struct ino_bucket *bucket = (struct ino_bucket *)action->mask;
+		unsigned int *imap = bucket->imap;
+
+		if(imap != NULL) {
+			ivindex = bucket->ino;
+			ivector_to_mask[ivindex] = 0;
+		}
 		else
 			printk("free_irq: WHeee, SYSIO_MASKED yet no imap reg.\n");
 	}
 
 	kfree(action);
-	if(!*(cpu_irq + irq_action))
-		disable_irq(irq);
+	if(ivindex != -1)
+		disable_irq(ivindex);
 
 	restore_flags(flags);
 }
 
-/* Per-processor IRQ locking depth, both SMP and non-SMP code use this. */
-unsigned int local_irq_count[NR_CPUS];
+/* Only uniprocessor needs this IRQ locking depth, on SMP it lives in the per-cpu
+ * structure for cache reasons.
+ */
+#ifndef __SMP__
+unsigned int local_irq_count;
+#endif
 
 #ifndef __SMP__
 int __sparc64_bh_counter = 0;
 
-#define irq_enter(cpu, irq)	(local_irq_count[cpu]++)
-#define irq_exit(cpu, irq)	(local_irq_count[cpu]--)
+#define irq_enter(cpu, irq)	(local_irq_count++)
+#define irq_exit(cpu, irq)	(local_irq_count--)
 
 #else
 
@@ -407,18 +731,31 @@ spinlock_t global_bh_lock = SPIN_LOCK_UNLOCKED;
 /* Global IRQ locking depth. */
 atomic_t global_irq_count = ATOMIC_INIT(0);
 
-static inline void wait_on_irq(int cpu)
+static unsigned long previous_irqholder;
+
+#undef INIT_STUCK
+#define INIT_STUCK 100000000
+
+#undef STUCK
+#define STUCK \
+if (!--stuck) {printk("wait_on_irq CPU#%d stuck at %08lx, waiting for %08lx (local=%d, global=%d)\n", cpu, where, previous_irqholder, local_count, atomic_read(&global_irq_count)); stuck = INIT_STUCK; }
+
+static inline void wait_on_irq(int cpu, unsigned long where)
 {
-	int local_count = local_irq_count[cpu];
+	int stuck = INIT_STUCK;
+	int local_count = local_irq_count;
 
 	while(local_count != atomic_read(&global_irq_count)) {
 		atomic_sub(local_count, &global_irq_count);
 		spin_unlock(&global_irq_lock);
 		for(;;) {
+			STUCK;
+			membar("#StoreLoad | #LoadLoad");
 			if (atomic_read(&global_irq_count))
 				continue;
-			if (*((unsigned char *)&global_irq_lock))
+			if (*((volatile unsigned char *)&global_irq_lock))
 				continue;
+			membar("#LoadLoad | #LoadStore");
 			if (spin_trylock(&global_irq_lock))
 				break;
 		}
@@ -426,25 +763,41 @@ static inline void wait_on_irq(int cpu)
 	}
 }
 
-static inline void get_irqlock(int cpu)
+#undef INIT_STUCK
+#define INIT_STUCK 10000000
+
+#undef STUCK
+#define STUCK \
+if (!--stuck) {printk("get_irqlock stuck at %08lx, waiting for %08lx\n", where, previous_irqholder); stuck = INIT_STUCK;}
+
+static inline void get_irqlock(int cpu, unsigned long where)
 {
+	int stuck = INIT_STUCK;
+
 	if (!spin_trylock(&global_irq_lock)) {
+		membar("#StoreLoad | #LoadLoad");
 		if ((unsigned char) cpu == global_irq_holder)
 			return;
 		do {
-			barrier();
+			do {
+				STUCK;
+				membar("#LoadLoad");
+			} while(*((volatile unsigned char *)&global_irq_lock));
 		} while (!spin_trylock(&global_irq_lock));
 	}
-	wait_on_irq(cpu);
+	wait_on_irq(cpu, where);
 	global_irq_holder = cpu;
+	previous_irqholder = where;
 }
 
 void __global_cli(void)
 {
 	int cpu = smp_processor_id();
+	unsigned long where;
 
+	__asm__ __volatile__("mov %%i7, %0" : "=r" (where));
 	__cli();
-	get_irqlock(cpu);
+	get_irqlock(cpu, where);
 }
 
 void __global_sti(void)
@@ -453,11 +806,6 @@ void __global_sti(void)
 	__sti();
 }
 
-unsigned long __global_save_flags(void)
-{
-	return global_irq_holder == (unsigned char) smp_processor_id();
-}
-
 void __global_restore_flags(unsigned long flags)
 {
 	if (flags & 1) {
@@ -472,15 +820,24 @@ void __global_restore_flags(unsigned long flags)
 	}
 }
 
+#undef INIT_STUCK
+#define INIT_STUCK 200000000
+
+#undef STUCK
+#define STUCK \
+if (!--stuck) {printk("irq_enter stuck (irq=%d, cpu=%d, global=%d)\n",irq,cpu,global_irq_holder); stuck = INIT_STUCK;}
+
 void irq_enter(int cpu, int irq)
 {
+	int stuck = INIT_STUCK;
+
 	hardirq_enter(cpu);
-	barrier();
-	while (*((unsigned char *)&global_irq_lock)) {
+	while (*((volatile unsigned char *)&global_irq_lock)) {
 		if ((unsigned char) cpu == global_irq_holder)
 			printk("irq_enter: Frosted Lucky Charms, "
 			       "they're magically delicious!\n");
-		barrier();
+		STUCK;
+		membar("#LoadLoad");
 	}
 }
 
@@ -492,8 +849,7 @@ void irq_exit(int cpu, int irq)
 
 void synchronize_irq(void)
 {
-	int cpu = smp_processor_id();
-	int local_count = local_irq_count[cpu];
+	int local_count = local_irq_count;
 	unsigned long flags;
 
 	if (local_count != atomic_read(&global_irq_count)) {
@@ -506,9 +862,13 @@ void synchronize_irq(void)
 
 void report_spurious_ivec(struct pt_regs *regs)
 {
-	printk("IVEC: Spurious interrupt vector received at (%016lx)\n",
-	       regs->tpc);
-	return;
+	extern unsigned long ivec_spurious_cookie;
+	static int times = 0;
+
+	printk("IVEC: Spurious interrupt vector (%016lx) received at (%016lx)\n",
+	       ivec_spurious_cookie, regs->tpc);
+	if(times++ > 1)
+		prom_halt();
 }
 
 void unexpected_irq(int irq, void *dev_cookie, struct pt_regs *regs)
@@ -547,13 +907,28 @@ void handler_irq(int irq, struct pt_regs *regs)
 	irq_enter(cpu, irq);
 	action = *(irq + irq_action);
 	kstat.interrupts[irq]++;
-	do {
-		if(!action || !action->handler)
-			unexpected_irq(irq, 0, regs);
-		action->handler(irq, action->dev_id, regs);
-		if(action->flags & SA_SYSIO_MASKED)
-			*((unsigned int *)action->mask) = SYSIO_ICLR_IDLE;
-	} while((action = action->next) != NULL);
+	if(!action) {
+		unexpected_irq(irq, 0, regs);
+	} else {
+		do {
+			struct ino_bucket *bucket = NULL;
+			unsigned int ino = 0;
+
+			if(action->flags & SA_IMAP_MASKED) {
+				bucket = (struct ino_bucket *)action->mask;
+
+				ino = bucket->ino;
+				if(!(ivector_to_mask[ino] & 0x80000000))
+					continue;
+			}
+
+			action->handler(irq, action->dev_id, regs);
+			if(bucket) {
+				ivector_to_mask[ino] &= ~(0x80000000);
+				*(bucket->iclr) = SYSIO_ICLR_IDLE;
+			}
+		} while((action = action->next) != NULL);
+	}
 	irq_exit(cpu, irq);
 }
 
@@ -567,7 +942,7 @@ void sparc_floppy_irq(int irq, void *dev_cookie, struct pt_regs *regs)
 
 	irq_enter(cpu, irq);
 	floppy_interrupt(irq, dev_cookie, regs);
-	if(action->flags & SA_SYSIO_MASKED)
+	if(action->flags & SA_IMAP_MASKED)
 		*((unsigned int *)action->mask) = SYSIO_ICLR_IDLE;
 	irq_exit(cpu, irq);
 }
@@ -595,7 +970,7 @@ static void install_fast_irq(unsigned int cpu_irq,
 	insns[0] = SPARC_BRANCH(((unsigned long) handler),
 				((unsigned long)&insns[0]));
 	insns[1] = SPARC_NOP;
-	__asm__ __volatile__("flush %0" : : "r" (ttent));
+	__asm__ __volatile__("membar #StoreStore; flush %0" : : "r" (ttent));
 }
 
 int request_fast_irq(unsigned int irq,
@@ -605,6 +980,7 @@ int request_fast_irq(unsigned int irq,
 	struct irqaction *action;
 	unsigned long flags;
 	unsigned int cpu_irq, *imap, *iclr;
+	int ivindex = -1;
 
 	/* XXX This really is not the way to do it, the "right way"
 	 * XXX is to have drivers set SA_SBUS or something like that
@@ -616,11 +992,11 @@ int request_fast_irq(unsigned int irq,
 	 */
 	if(irq == 14)
 		return -EINVAL;
-	cpu_irq = ino_to_pil[irq];
+	cpu_irq = sysio_ino_to_pil[irq];
 
 	if(!handler)
 		return -EINVAL;
-	imap = irq_to_imap(irq);
+	imap = sysio_irq_to_imap(irq);
 	action = *(cpu_irq + irq_action);
 	if(action) {
 		if(action->flags & SA_SHIRQ)
@@ -648,12 +1024,12 @@ int request_fast_irq(unsigned int irq,
 	install_fast_irq(cpu_irq, handler);
 
 	if(imap) {
-		int ivindex = (*imap & (SYSIO_IMAP_IGN | SYSIO_IMAP_INO));
-
+		ivindex = (*imap & (SYSIO_IMAP_IGN | SYSIO_IMAP_INO));
 		ivector_to_mask[ivindex] = (1 << cpu_irq);
-		iclr = imap_to_iclr(imap);
+		iclr = sysio_imap_to_iclr(imap);
 		action->mask = (unsigned long) iclr;
-		irqflags |= SA_SYSIO_MASKED;
+		irqflags |= SA_IMAP_MASKED;
+		add_ino_hash(ivindex, imap, iclr, irqflags);
 	} else
 		action->mask = 0;
 
@@ -665,7 +1041,9 @@ int request_fast_irq(unsigned int irq,
 
 	*(cpu_irq + irq_action) = action;
 
-	enable_irq(irq);
+	if(ivindex != -1)
+		enable_irq(ivindex);
+
 	restore_flags(flags);
 	return 0;
 }
@@ -675,31 +1053,27 @@ int request_fast_irq(unsigned int irq,
  */
 unsigned long probe_irq_on(void)
 {
-  return 0;
+	return 0;
 }
 
 int probe_irq_off(unsigned long mask)
 {
-  return 0;
+	return 0;
 }
 
 struct sun5_timer *linux_timers = NULL;
 
-/* This is called from sbus_init() to get the jiffies timer going.
- * We need to call this after there exists a valid SBus_chain so
- * that the IMAP/ICLR registers can be accessed.
- *
- * XXX That is because the whole startup sequence is broken.  I will
- * XXX fix it all up very soon.  -DaveM
- */
+/* This is gets the master level10 timer going. */
 void init_timers(void (*cfunc)(int, void *, struct pt_regs *))
 {
 	struct linux_prom64_registers pregs[3];
+	struct devid_cookie dcookie;
+	unsigned int *imap, *iclr;
 	u32 pirqs[2];
 	int node, err;
 
 	node = prom_finddevice("/counter-timer");
-	if(node == 0) {
+	if(node == 0 || node == -1) {
 		prom_printf("init_timers: Cannot find counter-timer PROM node.\n");
 		prom_halt();
 	}
@@ -715,13 +1089,22 @@ void init_timers(void (*cfunc)(int, void *, struct pt_regs *))
 		prom_halt();
 	}
 	linux_timers = (struct sun5_timer *) __va(pregs[0].phys_addr);
+	iclr = (((unsigned int *)__va(pregs[1].phys_addr))+1);
+	imap = (((unsigned int *)__va(pregs[2].phys_addr))+1);
 
 	/* Shut it up first. */
 	linux_timers->limit0 = 0;
 
 	/* Register IRQ handler. */
-	err = request_irq(pirqs[0] & 0x3f, /* XXX Fix this for big Enterprise XXX */
-			  cfunc, (SA_INTERRUPT | SA_STATIC_ALLOC), "timer", NULL);
+	dcookie.real_dev_id = NULL;
+	dcookie.imap = imap;
+	dcookie.iclr = iclr;
+	dcookie.pil = 10;
+	dcookie.bus_cookie = NULL;
+
+	err = request_irq(pirqs[0], cfunc,
+			  (SA_DCOOKIE | SA_INTERRUPT | SA_STATIC_ALLOC),
+			  "timer", &dcookie);
 
 	if(err) {
 		prom_printf("Serious problem, cannot register timer interrupt\n");
@@ -825,6 +1208,52 @@ void enable_prom_timer(void)
 	prom_timers->count0 = 0;
 }
 
+#ifdef __SMP__
+/* Called from smp_commence, when we know how many cpus are in the system
+ * and can have device IRQ's directed at them.
+ */
+void distribute_irqs(void)
+{
+	unsigned long flags;
+	int cpu, level;
+
+	printk("SMP: redistributing interrupts...\n");
+	save_and_cli(flags);
+	cpu = 0;
+	for(level = 0; level < NR_IRQS; level++) {
+		struct irqaction *p = irq_action[level];
+
+		while(p) {
+			if(p->flags & SA_IMAP_MASKED) {
+				struct ino_bucket *bucket = (struct ino_bucket *)p->mask;
+				unsigned int *imap = bucket->imap;
+				unsigned int val;
+				unsigned long tid = linux_cpus[cpu].mid << 9;
+
+				val = *imap;
+				*imap = SYSIO_IMAP_VALID | (tid & SYSIO_IMAP_TID);
+
+				printk("SMP: Redirecting IGN[%x] INO[%x] "
+				       "to cpu %d [%s]\n",
+				       (val & SYSIO_IMAP_IGN) >> 6,
+				       (val & SYSIO_IMAP_INO), cpu,
+				       p->name);
+
+				cpu += 1;
+				while(!(cpu_present_map & (1UL << cpu))) {
+					cpu += 1;
+					if(cpu >= smp_num_cpus)
+						cpu = 0;
+				}
+			}
+			p = p->next;
+		}
+	}
+	restore_flags(flags);
+	irqs_have_been_distributed = 1;
+}
+#endif
+
 __initfunc(void init_IRQ(void))
 {
 	int i;
diff --git a/arch/sparc64/kernel/process.c b/arch/sparc64/kernel/process.c
index afd5af8d0..35d4d606d 100644
--- a/arch/sparc64/kernel/process.c
+++ b/arch/sparc64/kernel/process.c
@@ -1,4 +1,4 @@
-/*  $Id: process.c,v 1.31 1997/07/24 12:15:05 davem Exp $
+/*  $Id: process.c,v 1.42 1997/08/19 14:17:55 jj Exp $
  *  arch/sparc64/kernel/process.c
  *
  *  Copyright (C) 1995, 1996 David S. Miller (davem@caip.rutgers.edu)
@@ -27,18 +27,21 @@
 #include <linux/a.out.h>
 #include <linux/config.h>
 #include <linux/reboot.h>
+#include <linux/delay.h>
 
 #include <asm/oplib.h>
 #include <asm/uaccess.h>
 #include <asm/system.h>
 #include <asm/page.h>
 #include <asm/pgtable.h>
-#include <asm/delay.h>
 #include <asm/processor.h>
 #include <asm/pstate.h>
 #include <asm/elf.h>
 #include <asm/fpumacro.h>
 
+#define PGTCACHE_HIGH_WATER		50
+#define PGTCACHE_LOW_WATER		25
+
 #ifndef __SMP__
 
 /*
@@ -53,6 +56,16 @@ asmlinkage int sys_idle(void)
 	current->priority = -100;
 	current->counter = -100;
 	for (;;) {
+		if(pgtable_cache_size > PGTCACHE_LOW_WATER) {
+			do {
+				if(pgd_quicklist)
+					free_page((unsigned long) get_pgd_fast());
+				if(pmd_quicklist)
+					free_page((unsigned long) get_pmd_fast());
+				if(pte_quicklist)
+					free_page((unsigned long) get_pte_fast());
+			} while(pgtable_cache_size > PGTCACHE_HIGH_WATER);
+		}
 		run_task_queue(&tq_scheduler);
 		schedule();
 	}
@@ -68,13 +81,26 @@ asmlinkage int cpu_idle(void)
 {
 	current->priority = -100;
 	while(1) {
+		if(pgtable_cache_size > PGTCACHE_LOW_WATER) {
+			do {
+				if(pgd_quicklist)
+					free_page((unsigned long) get_pgd_fast());
+				if(pmd_quicklist)
+					free_page((unsigned long) get_pmd_fast());
+				if(pte_quicklist)
+					free_page((unsigned long) get_pte_fast());
+			} while(pgtable_cache_size > PGTCACHE_HIGH_WATER);
+		}
 		if(tq_scheduler) {
 			lock_kernel();
 			run_task_queue(&tq_scheduler);
 			unlock_kernel();
 		}
+		barrier();
 		current->counter = -100;
-		schedule();
+		if(resched_needed())
+			schedule();
+		barrier();
 	}
 }
 
@@ -251,8 +277,20 @@ void show_stackframe32(struct sparc_stackf32 *sf)
 	} while ((size -= sizeof(unsigned)));
 }
 
-void show_regs(struct pt_regs * regs)
+#ifdef __SMP__
+static spinlock_t regdump_lock = SPIN_LOCK_UNLOCKED;
+#endif
+
+void __show_regs(struct pt_regs * regs)
 {
+#ifdef __SMP__
+	unsigned long flags;
+
+	spin_lock_irqsave(&regdump_lock, flags);
+	printk("CPU[%d]: local_irq_count[%ld] global_irq_count[%d]\n",
+	       smp_processor_id(), local_irq_count,
+	       atomic_read(&global_irq_count));
+#endif
         printk("TSTATE: %016lx TPC: %016lx TNPC: %016lx Y: %08x\n", regs->tstate,
 	       regs->tpc, regs->tnpc, regs->y);
 	printk("g0: %016lx g1: %016lx g2: %016lx g3: %016lx\n",
@@ -268,6 +306,21 @@ void show_regs(struct pt_regs * regs)
 	       regs->u_regs[12], regs->u_regs[13], regs->u_regs[14],
 	       regs->u_regs[15]);
 	show_regwindow(regs);
+#ifdef __SMP__
+	spin_unlock_irqrestore(&regdump_lock, flags);
+#endif
+}
+
+void show_regs(struct pt_regs *regs)
+{
+	__show_regs(regs);
+#ifdef __SMP__
+	{
+		extern void smp_report_regs(void);
+
+		smp_report_regs();
+	}
+#endif
 }
 
 void show_regs32(struct pt_regs32 *regs)
@@ -332,17 +385,22 @@ void flush_thread(void)
 
 	/* No new signal delivery by default. */
 	current->tss.new_signal = 0;
-	current->flags &= ~PF_USEDFPU;
+	current->tss.flags &= ~(SPARC_FLAG_USEDFPU | SPARC_FLAG_USEDFPUL |
+			        SPARC_FLAG_USEDFPUU);
 	
 	/* Now, this task is no longer a kernel thread. */
 	current->tss.current_ds = USER_DS;
 	if(current->tss.flags & SPARC_FLAG_KTHREAD) {
+		extern spinlock_t scheduler_lock;
+
 		current->tss.flags &= ~SPARC_FLAG_KTHREAD;
 
 		/* exec_mmap() set context to NO_CONTEXT, here is
 		 * where we grab a new one.
 		 */
+		spin_lock(&scheduler_lock);
 		get_mmu_context(current);
+		spin_unlock(&scheduler_lock);
 	}
 	current->tss.ctx = current->mm->context & 0x1fff;
 	spitfire_set_secondary_context (current->tss.ctx);
@@ -437,10 +495,14 @@ void fault_in_user_windows(struct pt_regs *regs)
 			struct reg_window *rwin = &tp->reg_window[window];
 
 			if(copy_to_user((char *)sp, rwin, winsize))
-				do_exit(SIGILL);
+				goto barf;
 		} while(window--);
 	}
 	current->tss.w_saved = 0;
+	return;
+barf:
+	lock_kernel();
+	do_exit(SIGILL);
 }
 
 /* Copy a Sparc thread.  The fork() return value conventions
@@ -483,7 +545,7 @@ int copy_thread(int nr, unsigned long clone_flags, unsigned long sp,
 	p->tss.kpc = ((unsigned long) ret_from_syscall) - 0x8;
 #endif
 	p->tss.kregs = (struct pt_regs *)(child_trap_frame+sizeof(struct reg_window));
-	p->tss.cwp = regs->u_regs[UREG_G0];
+	p->tss.cwp = (regs->tstate + 1) & TSTATE_CWP;
 	if(regs->tstate & TSTATE_PRIV) {
 		p->tss.kregs->u_regs[UREG_FP] = p->tss.ksp;
 		p->tss.flags |= SPARC_FLAG_KTHREAD;
@@ -581,6 +643,7 @@ asmlinkage int sparc_execve(struct pt_regs *regs)
 	if(!error) {
 		fprs_write(0);
 		regs->fprs = 0;
+		regs->tstate &= ~TSTATE_PEF;
 	}
 out:
 	unlock_kernel();
diff --git a/arch/sparc64/kernel/psycho.c b/arch/sparc64/kernel/psycho.c
new file mode 100644
index 000000000..8aa1c342b
--- /dev/null
+++ b/arch/sparc64/kernel/psycho.c
@@ -0,0 +1,1417 @@
+/* $Id: psycho.c,v 1.22 1997/08/31 03:51:40 davem Exp $
+ * psycho.c: Ultra/AX U2P PCI controller support.
+ *
+ * Copyright (C) 1997 David S. Miller (davem@caipfs.rutgers.edu)
+ */
+
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/types.h>
+
+#include <asm/ebus.h>
+#include <asm/sbus.h> /* for sanity check... */
+
+#ifndef CONFIG_PCI
+
+int pcibios_present(void)
+{
+	return 0;
+}
+
+asmlinkage int sys_pciconfig_read(unsigned long bus,
+				  unsigned long dfn,
+				  unsigned long off,
+				  unsigned long len,
+				  unsigned char *buf)
+{
+	return 0;
+}
+
+asmlinkage int sys_pciconfig_write(unsigned long bus,
+				   unsigned long dfn,
+				   unsigned long off,
+				   unsigned long len,
+				   unsigned char *buf)
+{
+	return 0;
+}
+
+#else
+
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/bios32.h>
+#include <linux/pci.h>
+
+#include <asm/io.h>
+#include <asm/oplib.h>
+#include <asm/pbm.h>
+#include <asm/uaccess.h>
+
+struct linux_psycho *psycho_root = NULL;
+
+/* This is used to make the scan_bus in the generic PCI code be
+ * a nop, as we need to control the actual bus probing sequence.
+ * After that we leave it on of course.
+ */
+static int pci_probe_enable = 0;
+
+static inline unsigned long long_align(unsigned long addr)
+{
+	return ((addr + (sizeof(unsigned long) - 1)) &
+		~(sizeof(unsigned long) - 1));
+}
+
+static unsigned long psycho_iommu_init(struct linux_psycho *psycho,
+				       unsigned long memory_start)
+{
+	unsigned long tsbbase = PAGE_ALIGN(memory_start);
+	unsigned long control, i;
+	unsigned long *iopte;
+
+	memory_start = (tsbbase + ((32 * 1024) * 8));
+	iopte = (unsigned long *)tsbbase;
+
+	for(i = 0; i < (65536 / 2); i++) {
+		*iopte = (IOPTE_VALID | IOPTE_64K |
+			  IOPTE_CACHE | IOPTE_WRITE);
+		*iopte |= (i << 16);
+		iopte++;
+	}
+
+	psycho->psycho_regs->iommu_tsbbase = __pa(tsbbase);
+
+	control = psycho->psycho_regs->iommu_control;
+	control &= ~(IOMMU_CTRL_TSBSZ);
+	control |= (IOMMU_TSBSZ_32K | IOMMU_CTRL_TBWSZ | IOMMU_CTRL_ENAB);
+	psycho->psycho_regs->iommu_control = control;
+
+	return memory_start;
+}
+
+extern void prom_pbm_ranges_init(int node, struct linux_pbm_info *pbm);
+
+unsigned long pcibios_init(unsigned long memory_start, unsigned long memory_end)
+{
+	struct linux_prom64_registers pr_regs[3];
+	char namebuf[128];
+	u32 portid;
+	int node;
+
+	printk("PSYCHO: Probing for controllers.\n");
+
+	memory_start = long_align(memory_start);
+	node = prom_getchild(prom_root_node);
+	while((node = prom_searchsiblings(node, "pci")) != 0) {
+		struct linux_psycho *psycho = (struct linux_psycho *)memory_start;
+		struct linux_psycho *search;
+		struct linux_pbm_info *pbm = NULL;
+		u32 busrange[2];
+		int err, is_pbm_a;
+
+		portid = prom_getintdefault(node, "upa-portid", 0xff);
+		for(search = psycho_root; search; search = search->next) {
+			if(search->upa_portid == portid) {
+				psycho = search;
+
+				/* This represents _this_ instance, so it's
+				 * which ever one does _not_ have the prom node
+				 * info filled in yet.
+				 */
+				is_pbm_a = (psycho->pbm_A.prom_node == 0);
+				goto other_pbm;
+			}
+		}
+
+		memory_start = long_align(memory_start + sizeof(struct linux_psycho));
+
+		memset(psycho, 0, sizeof(*psycho));
+
+		psycho->next = psycho_root;
+		psycho_root = psycho;
+
+		psycho->upa_portid = portid;
+
+		/* Map in PSYCHO register set and report the presence of this PSYCHO. */
+		err = prom_getproperty(node, "reg",
+				       (char *)&pr_regs[0], sizeof(pr_regs));
+		if(err == 0 || err == -1) {
+			prom_printf("PSYCHO: Error, cannot get U2P registers "
+				    "from PROM.\n");
+			prom_halt();
+		}
+
+		/* Third REG in property is base of entire PSYCHO register space. */
+		psycho->psycho_regs = sparc_alloc_io((pr_regs[2].phys_addr & 0xffffffff),
+						     NULL, sizeof(struct psycho_regs),
+						     "PSYCHO Registers",
+						     (pr_regs[2].phys_addr >> 32), 0);
+		if(psycho->psycho_regs == NULL) {
+			prom_printf("PSYCHO: Error, cannot map PSYCHO "
+				    "main registers.\n");
+			prom_halt();
+		}
+
+		printk("PSYCHO: Found controller, main regs at %p\n",
+		       psycho->psycho_regs);
+#if 0
+		printk("PSYCHO: Interrupt retry [%016lx]\n",
+		       psycho->psycho_regs->irq_retry);
+#endif
+		psycho->psycho_regs->irq_retry = 0xff;
+
+		/* Now map in PCI config space for entire PSYCHO. */
+		psycho->pci_config_space =
+			sparc_alloc_io(((pr_regs[2].phys_addr & 0xffffffff)+0x01000000),
+				       NULL, 0x01000000,
+				       "PCI Config Space",
+				       (pr_regs[2].phys_addr >> 32), 0);
+		if(psycho->pci_config_space == NULL) {
+			prom_printf("PSYCHO: Error, cannot map PCI config space.\n");
+			prom_halt();
+		}
+
+		/* Report some more info. */
+		printk("PSYCHO: PCI config space at %p\n", psycho->pci_config_space);
+
+		memory_start = psycho_iommu_init(psycho, memory_start);
+
+		is_pbm_a = ((pr_regs[0].phys_addr & 0x6000) == 0x2000);
+
+		/* Enable arbitration for all PCI slots. */
+		psycho->psycho_regs->pci_a_control |= 0x3f;
+		psycho->psycho_regs->pci_b_control |= 0x3f;
+
+	other_pbm:
+		if(is_pbm_a)
+			pbm = &psycho->pbm_A;
+		else
+			pbm = &psycho->pbm_B;
+
+		pbm->parent = psycho;
+		pbm->IO_assignments = NULL;
+		pbm->MEM_assignments = NULL;
+		pbm->prom_node = node;
+
+		prom_getstring(node, "name", namebuf, sizeof(namebuf));
+		strcpy(pbm->prom_name, namebuf);
+
+		/* Now the ranges. */
+		prom_pbm_ranges_init(node, pbm);
+
+		/* Finally grab the pci bus root array for this pbm after
+		 * having found the bus range existing under it.
+		 */
+		err = prom_getproperty(node, "bus-range",
+				       (char *)&busrange[0], sizeof(busrange));
+		if(err == 0 || err == -1) {
+			prom_printf("PSYCHO: Error, cannot get PCI bus range.\n");
+			prom_halt();
+		}
+		pbm->pci_first_busno = busrange[0];
+		pbm->pci_last_busno = busrange[1];
+		memset(&pbm->pci_bus, 0, sizeof(struct pci_bus));
+
+		node = prom_getsibling(node);
+		if(!node)
+			break;
+	}
+
+	/* Last minute sanity check. */
+	if(psycho_root == NULL && SBus_chain == NULL) {
+		prom_printf("Fatal error, neither SBUS nor PCI bus found.\n");
+		prom_halt();
+	}
+
+	return memory_start;
+}
+
+int pcibios_present(void)
+{
+	return psycho_root != NULL;
+}
+
+int pcibios_find_device (unsigned short vendor, unsigned short device_id,
+			 unsigned short index, unsigned char *bus,
+			 unsigned char *devfn)
+{
+	unsigned int curr = 0;
+	struct pci_dev *dev;
+
+	for (dev = pci_devices; dev; dev = dev->next) {
+		if (dev->vendor == vendor && dev->device == device_id) {
+			if (curr == index) {
+				*devfn = dev->devfn;
+				*bus = dev->bus->number;
+				return PCIBIOS_SUCCESSFUL;
+			}
+			++curr;
+		}
+	}
+	return PCIBIOS_DEVICE_NOT_FOUND;
+}
+
+int pcibios_find_class (unsigned int class_code, unsigned short index,
+			unsigned char *bus, unsigned char *devfn)
+{
+	unsigned int curr = 0;
+	struct pci_dev *dev;
+
+	for (dev = pci_devices; dev; dev = dev->next) {
+		if (dev->class == class_code) {
+			if (curr == index) {
+				*devfn = dev->devfn;
+				*bus = dev->bus->number;
+				return PCIBIOS_SUCCESSFUL;
+			}
+			++curr;
+		}
+	}
+	return PCIBIOS_DEVICE_NOT_FOUND;
+}
+
+static inline struct pci_vma *pci_find_vma(struct linux_pbm_info *pbm,
+					   unsigned long start,
+					   int io)
+{
+	struct pci_vma *vp = (io ? pbm->IO_assignments : pbm->MEM_assignments);
+
+	while(vp) {
+		if(vp->end > start)
+			break;
+		vp = vp->next;
+	}
+	return vp;
+}
+
+static inline void pci_add_vma(struct linux_pbm_info *pbm, struct pci_vma *new, int io)
+{
+	struct pci_vma *vp = (io ? pbm->IO_assignments : pbm->MEM_assignments);
+
+	if(!vp) {
+		new->next = NULL;
+		if(io)
+			pbm->IO_assignments = new;
+		else
+			pbm->MEM_assignments = new;
+	} else {
+		struct pci_vma *prev = NULL;
+
+		while(vp && (vp->end < new->end)) {
+			prev = vp;
+			vp = vp->next;
+		}
+		new->next = vp;
+		if(!prev) {
+			if(io)
+				pbm->IO_assignments = new;
+			else
+				pbm->MEM_assignments = new;
+		} else {
+			prev->next = new;
+		}
+
+		/* Check for programming errors. */
+		if(vp &&
+		   ((vp->start >= new->start && vp->start < new->end) ||
+		    ((vp->end - 1) >= new->start && (vp->end - 1) < new->end))) {
+			prom_printf("pci_add_vma: Wheee, overlapping %s PCI vma's\n",
+				    io ? "IO" : "MEM");
+			prom_printf("pci_add_vma: vp[%016lx:%016lx] "
+				    "new[%016lx:%016lx]\n",
+				    vp->start, vp->end,
+				    new->start, new->end);
+		}
+	}
+}
+
+static unsigned long *pci_alloc_arena = NULL;
+
+static inline void pci_init_alloc_init(unsigned long *mstart)
+{
+	pci_alloc_arena = mstart;
+}
+
+static inline void pci_init_alloc_fini(void)
+{
+	pci_alloc_arena = NULL;
+}
+
+static void *pci_init_alloc(int size)
+{
+	unsigned long start = long_align(*pci_alloc_arena);
+	void *mp = (void *)start;
+
+	if(!pci_alloc_arena) {
+		prom_printf("pci_init_alloc: pci_vma arena not init'd\n");
+		prom_halt();
+	}
+	start += size;
+	*pci_alloc_arena = start;
+	return mp;
+}
+
+static inline struct pci_vma *pci_vma_alloc(void)
+{
+	return pci_init_alloc(sizeof(struct pci_vma));
+}
+
+static inline struct pcidev_cookie *pci_devcookie_alloc(void)
+{
+	return pci_init_alloc(sizeof(struct pcidev_cookie));
+}
+
+static void pbm_probe(struct linux_pbm_info *pbm, unsigned long *mstart)
+{
+	struct pci_bus *pbus = &pbm->pci_bus;
+
+	/* PSYCHO PBM's include child PCI bridges in bus-range property,
+	 * but we don't scan each of those ourselves, Linux generic PCI
+	 * probing code will find child bridges and link them into this
+	 * pbm's root PCI device hierarchy.
+	 */
+	pbus->number = pbm->pci_first_busno;
+	pbus->sysdata = pbm;
+	pbus->subordinate = pci_scan_bus(pbus, mstart);
+}
+
+static int pdev_to_pnode_sibtraverse(struct linux_pbm_info *pbm,
+				     struct pci_dev *pdev,
+				     int node)
+{
+	struct linux_prom_pci_registers pregs[PROMREG_MAX];
+	int err;
+
+	while(node) {
+		int child;
+
+		child = prom_getchild(node);
+		if(child != 0 && child != -1) {
+			int res;
+
+			res = pdev_to_pnode_sibtraverse(pbm, pdev, child);
+			if(res != 0 && res != -1)
+				return res;
+		}
+		err = prom_getproperty(node, "reg", (char *)&pregs[0], sizeof(pregs));
+		if(err != 0 && err != -1) {
+			u32 devfn = (pregs[0].phys_hi >> 8) & 0xff;
+
+			if(devfn == pdev->devfn)
+				return node; /* Match */
+		}
+
+		node = prom_getsibling(node);
+	}
+	return 0;
+}
+
+static void pdev_cookie_fillin(struct linux_pbm_info *pbm, struct pci_dev *pdev)
+{
+	struct pcidev_cookie *pcp;
+	int node = prom_getchild(pbm->prom_node);
+
+	node = pdev_to_pnode_sibtraverse(pbm, pdev, node);
+	if(node == 0)
+		node = -1;
+	pcp = pci_devcookie_alloc();
+	pcp->pbm = pbm;
+	pcp->prom_node = node;
+	pdev->sysdata = pcp;
+}
+
+static void fill_in_pbm_cookies(struct linux_pbm_info *pbm)
+{
+	struct pci_bus *pbtmp, *pbus = &pbm->pci_bus;
+	struct pci_dev *pdev;
+
+	for(pbtmp = pbus->children; pbtmp; pbtmp = pbtmp->children)
+		pbtmp->sysdata = pbm;
+
+	for( ; pbus; pbus = pbus->children)
+		for(pdev = pbus->devices; pdev; pdev = pdev->sibling)
+			pdev_cookie_fillin(pbm, pdev);
+}
+
+/* #define RECORD_ASSIGNMENTS_DEBUG */
+
+/* Walk PROM device tree under PBM, looking for 'assigned-address'
+ * properties, and recording them in pci_vma's linked in via
+ * PBM->assignments.
+ */
+static int gimme_ebus_assignments(int node, struct linux_prom_pci_registers *aregs)
+{
+	struct linux_prom_ebus_ranges erng[PROMREG_MAX];
+	int err, iter;
+
+	err = prom_getproperty(node, "ranges", (char *)&erng[0], sizeof(erng));
+	if(err == 0 || err == -1) {
+		prom_printf("EBUS: fatal error, no range property.\n");
+		prom_halt();
+	}
+	err = (err / sizeof(struct linux_prom_ebus_ranges));
+	for(iter = 0; iter < err; iter++) {
+		struct linux_prom_ebus_ranges *ep = &erng[iter];
+		struct linux_prom_pci_registers *ap = &aregs[iter];
+
+		ap->phys_hi = ep->parent_phys_hi;
+		ap->phys_mid = ep->parent_phys_mid;
+		ap->phys_lo = ep->parent_phys_lo;
+	}
+	return err;
+}
+
+static void assignment_process(struct linux_pbm_info *pbm, int node)
+{
+	struct linux_prom_pci_registers aregs[PROMREG_MAX];
+	char pname[256];
+	int err, iter, numa;
+
+	err = prom_getproperty(node, "name", (char *)&pname[0], sizeof(pname));
+	if(strncmp(pname, "ebus", 4) == 0) {
+		numa = gimme_ebus_assignments(node, &aregs[0]);
+	} else {
+		err = prom_getproperty(node, "assigned-addresses",
+				       (char *)&aregs[0], sizeof(aregs));
+
+		/* No assignments, nothing to do. */
+		if(err == 0 || err == -1)
+			return;
+
+		numa = (err / sizeof(struct linux_prom_pci_ranges));
+	}
+
+	for(iter = 0; iter < numa; iter++) {
+		struct linux_prom_pci_registers *ap = &aregs[iter];
+		struct pci_vma *vp;
+		int space, breg, io;
+
+		space = (ap->phys_hi >> 24) & 3;
+		if(space != 1 && space != 2)
+			continue;
+		io = (space == 1);
+
+		breg = (ap->phys_hi & 0xff);
+		if(breg == PCI_ROM_ADDRESS)
+			continue;
+
+		vp = pci_vma_alloc();
+
+		/* XXX Means we don't support > 32-bit range of
+		 * XXX PCI MEM space, PSYCHO/PBM does not support it
+		 * XXX either due to it's layout so...
+		 */
+		vp->start = ap->phys_lo;
+		vp->end = vp->start + ap->size_lo;
+		vp->base_reg = breg;
+
+		/* Sanity */
+		if(io && (vp->end & ~(0xffff))) {
+			prom_printf("assignment_process: Out of range PCI I/O "
+				    "[%08lx:%08lx]\n", vp->start, vp->end);
+			prom_halt();
+		}
+
+		pci_add_vma(pbm, vp, io);
+	}
+}
+
+static void assignment_walk_siblings(struct linux_pbm_info *pbm, int node)
+{
+	while(node) {
+		int child = prom_getchild(node);
+		if(child)
+			assignment_walk_siblings(pbm, child);
+
+		assignment_process(pbm, node);
+
+		node = prom_getsibling(node);
+	}
+}
+
+static void record_assignments(struct linux_pbm_info *pbm)
+{
+	assignment_walk_siblings(pbm, prom_getchild(pbm->prom_node));
+}
+
+/* #define FIXUP_REGS_DEBUG */
+
+static void fixup_regs(struct pci_dev *pdev,
+		       struct linux_pbm_info *pbm,
+		       struct linux_prom_pci_registers *pregs,
+		       int nregs,
+		       struct linux_prom_pci_registers *assigned,
+		       int numaa)
+{
+	int preg, rng;
+	int IO_seen = 0;
+	int MEM_seen = 0;
+
+	for(preg = 0; preg < nregs; preg++) {
+		struct linux_prom_pci_registers *ap = NULL;
+		int bustype = (pregs[preg].phys_hi >> 24) & 0x3;
+		int bsreg, brindex;
+		u64 pci_addr;
+
+		if(bustype == 0) {
+			/* Config space cookie, nothing to do. */
+			if(preg != 0)
+				prom_printf("fixup_doit: strange, config space not 0\n");
+			continue;
+		} else if(bustype == 3) {
+			/* XXX add support for this... */
+			prom_printf("fixup_doit: Warning, ignoring 64-bit PCI "
+				    "memory space, tell DaveM.\n");
+			continue;
+		}
+		bsreg = (pregs[preg].phys_hi & 0xff);
+
+		/* We can safely ignore these. */
+		if(bsreg == PCI_ROM_ADDRESS)
+			continue;
+
+		/* Sanity */
+		if((bsreg < PCI_BASE_ADDRESS_0) ||
+		   (bsreg > (PCI_BASE_ADDRESS_5 + 4)) ||
+		   (bsreg & 3)) {
+			prom_printf("fixup_doit: Warning, ignoring bogus basereg [%x]\n",
+				    bsreg);
+			continue;
+		}
+
+		brindex = (bsreg - PCI_BASE_ADDRESS_0) >> 2;
+		if(numaa) {
+			int r;
+
+			for(r = 0; r < numaa; r++) {
+				int abreg;
+
+				abreg = (assigned[r].phys_hi & 0xff);
+				if(abreg == bsreg) {
+					ap = &assigned[r];
+					break;
+				}
+			}
+		}
+
+		/* Now construct UPA physical address. */
+		pci_addr  = (((u64)pregs[preg].phys_mid) << 32UL);
+		pci_addr |= (((u64)pregs[preg].phys_lo));
+
+		if(ap) {
+			pci_addr += ((u64)ap->phys_lo);
+			pci_addr += (((u64)ap->phys_mid) << 32UL);
+		}
+
+		/* Final step, apply PBM range. */
+		for(rng = 0; rng < pbm->num_pbm_ranges; rng++) {
+			struct linux_prom_pci_ranges *rp = &pbm->pbm_ranges[rng];
+			int space = (rp->child_phys_hi >> 24) & 3;
+
+			if(space == bustype) {
+				pci_addr += ((u64)rp->parent_phys_lo);
+				pci_addr += (((u64)rp->parent_phys_hi) << 32UL);
+				break;
+			}
+		}
+		if(rng == pbm->num_pbm_ranges) {
+			/* AIEEE */
+			prom_printf("fixup_doit: YIEEE, cannot find PBM ranges\n");
+		}
+		pdev->base_address[brindex] = (unsigned long)__va(pci_addr);
+
+		/* Preserve I/O space bit. */
+		if(bustype == 0x1) {
+			pdev->base_address[brindex] |= 1;
+			IO_seen = 1;
+		} else {
+			MEM_seen = 1;
+		}
+	}
+
+	/* Now handle assignments PROM did not take care of. */
+	if(nregs) {
+		int breg;
+
+		for(breg = PCI_BASE_ADDRESS_0; breg <= PCI_BASE_ADDRESS_5; breg += 4) {
+			unsigned int rtmp, ridx = ((breg - PCI_BASE_ADDRESS_0) >> 2);
+			unsigned int base = (unsigned int)pdev->base_address[ridx];
+			struct pci_vma *vp;
+			u64 pci_addr;
+			int io;
+
+			if(pdev->base_address[ridx] > PAGE_OFFSET)
+				continue;
+
+			io = (base & PCI_BASE_ADDRESS_SPACE)==PCI_BASE_ADDRESS_SPACE_IO;
+			base &= ~((io ?
+				   PCI_BASE_ADDRESS_IO_MASK :
+				   PCI_BASE_ADDRESS_MEM_MASK));
+			vp = pci_find_vma(pbm, base, io);
+			if(!vp || vp->start > base) {
+				unsigned int size, new_base;
+
+				pcibios_read_config_dword(pdev->bus->number,
+							  pdev->devfn,
+							  breg, &rtmp);
+				pcibios_write_config_dword(pdev->bus->number,
+							   pdev->devfn,
+							   breg, 0xffffffff);
+				pcibios_read_config_dword(pdev->bus->number,
+							  pdev->devfn,
+							  breg, &size);
+				if(io)
+					size &= ~1;
+				size = (~(size) + 1);
+				if(!size)
+					continue;
+
+				new_base = 0;
+				for(vp=pci_find_vma(pbm,new_base,io); ; vp=vp->next) {
+					if(!vp || new_base + size <= vp->start)
+						break;
+					new_base = (vp->end + (size - 1)) & ~(size-1);
+				}
+				if(vp && (new_base + size > vp->start)) {
+					prom_printf("PCI: Impossible full %s space.\n",
+						    (io ? "IO" : "MEM"));
+					prom_halt();
+				}
+				vp = pci_vma_alloc();
+				vp->start = new_base;
+				vp->end = vp->start + size;
+				vp->base_reg = breg;
+
+				/* Sanity */
+				if(io && vp->end & ~(0xffff)) {
+					prom_printf("PCI: Out of range PCI I/O "
+						    "[%08lx:%08lx] during fixup\n",
+						    vp->start, vp->end);
+					prom_halt();
+				}
+				pci_add_vma(pbm, vp, io);
+
+				rtmp = new_base;
+				if(io)
+					rtmp |= (rtmp & PCI_BASE_ADDRESS_IO_MASK);
+				else
+					rtmp |= (rtmp & PCI_BASE_ADDRESS_MEM_MASK);
+				pcibios_write_config_dword(pdev->bus->number,
+							   pdev->devfn,
+							   breg, rtmp);
+
+				/* Apply PBM ranges and update pci_dev. */
+				pci_addr = new_base;
+				for(rng = 0; rng < pbm->num_pbm_ranges; rng++) {
+					struct linux_prom_pci_ranges *rp;
+					int rspace;
+
+					rp = &pbm->pbm_ranges[rng];
+					rspace = (rp->child_phys_hi >> 24) & 3;
+					if(io && rspace != 1)
+						continue;
+					else if(!io && rspace != 2)
+						continue;
+					pci_addr += ((u64)rp->parent_phys_lo);
+					pci_addr += (((u64)rp->parent_phys_hi)<<32UL);
+					break;
+				}
+				if(rng == pbm->num_pbm_ranges) {
+					/* AIEEE */
+					prom_printf("fixup_doit: YIEEE, cannot find "
+						    "PBM ranges\n");
+				}
+				pdev->base_address[ridx] = (unsigned long)__va(pci_addr);
+
+				/* Preserve I/O space bit. */
+				if(io) {
+					pdev->base_address[ridx] |= 1;
+					IO_seen = 1;
+				} else {
+					MEM_seen = 1;
+				}
+			}
+		}
+	}
+	if(IO_seen || MEM_seen) {
+		unsigned int l;
+
+		pcibios_read_config_dword(pdev->bus->number,
+					  pdev->devfn,
+					  PCI_COMMAND, &l);
+#ifdef FIXUP_REGS_DEBUG
+		prom_printf("[");
+#endif
+		if(IO_seen) {
+#ifdef FIXUP_REGS_DEBUG
+			prom_printf("IO ");
+#endif
+			l |= PCI_COMMAND_IO;
+		}
+		if(MEM_seen) {
+#ifdef FIXUP_REGS_DEBUG
+			prom_printf("MEM");
+#endif
+			l |= PCI_COMMAND_MEMORY;
+		}
+#ifdef FIXUP_REGS_DEBUG
+		prom_printf("]");
+#endif
+		pcibios_write_config_dword(pdev->bus->number,
+					   pdev->devfn,
+					   PCI_COMMAND, l);
+	}
+
+#ifdef FIXUP_REGS_DEBUG
+	prom_printf("REG_FIXUP[%s]: ", pci_strdev(pdev->vendor, pdev->device));
+	for(preg = 0; preg < 6; preg++) {
+		if(pdev->base_address[preg] != 0)
+			prom_printf("%d[%016lx] ", preg, pdev->base_address[preg]);
+	}
+	prom_printf("\n");
+#endif
+}
+
+#define imap_offset(__member) \
+	((unsigned long)(&(((struct psycho_regs *)0)->__member)))
+
+static unsigned long psycho_pcislot_imap_offset(unsigned long ino)
+{
+	unsigned int bus, slot;
+
+	bus = (ino & 0x10) >> 4;
+	slot = (ino & 0x0c) >> 2;
+
+	if(bus == 0) {
+		/* Perform a sanity check, we might as well.
+		 * PBM A only has 2 PCI slots.
+		 */
+		if(slot > 1) {
+			prom_printf("pcislot_imap: Bogus slot on PBM A (%ld)\n", slot);
+			prom_halt();
+		}
+		if(slot == 0)
+			return imap_offset(imap_a_slot0);
+		else
+			return imap_offset(imap_a_slot1);
+	} else {
+		switch(slot) {
+		case 0:
+			return imap_offset(imap_b_slot0);
+		case 1:
+			return imap_offset(imap_b_slot1);
+		case 2:
+			return imap_offset(imap_b_slot2);
+		case 3:
+			return imap_offset(imap_b_slot3);
+		default:
+			prom_printf("pcislot_imap: IMPOSSIBLE [%d:%d]\n",
+				    bus, slot);
+			prom_halt();
+			return 0; /* Make gcc happy */
+		};
+	}
+}
+
+/* Exported for EBUS probing layer. */
+unsigned int psycho_irq_build(unsigned int full_ino)
+{
+	unsigned long imap_off, ign, ino;
+
+	ign = (full_ino & PSYCHO_IMAP_IGN) >> 6;
+	ino = (full_ino & PSYCHO_IMAP_INO);
+
+	/* Compute IMAP register offset, generic IRQ layer figures out
+	 * the ICLR register address as this is simple given the 32-bit
+	 * irq number and IMAP register address.
+	 */
+	if((ino & 0x20) == 0)
+		imap_off = psycho_pcislot_imap_offset(ino);
+	else {
+		switch(ino) {
+		case 0x20:
+			/* Onboard SCSI. */
+			imap_off = imap_offset(imap_scsi);
+			break;
+
+		case 0x21:
+			/* Onboard Ethernet (ie. CheerIO/HME) */
+			imap_off = imap_offset(imap_eth);
+			break;
+
+		case 0x22:
+			/* Onboard Parallel Port */
+			imap_off = imap_offset(imap_bpp);
+			break;
+
+		case 0x23:
+			/* Audio Record */
+			imap_off = imap_offset(imap_au_rec);
+			break;
+
+		case 0x24:
+			/* Audio Play */
+			imap_off = imap_offset(imap_au_play);
+			break;
+
+		case 0x25:
+			/* Power Fail */
+			imap_off = imap_offset(imap_pfail);
+			break;
+
+		case 0x26:
+			/* Onboard KBD/MOUSE/SERIAL */
+			imap_off = imap_offset(imap_kms);
+			break;
+
+		case 0x27:
+			/* Floppy (ie. fdthree) */
+			imap_off = imap_offset(imap_flpy);
+			break;
+
+		case 0x28:
+			/* Spare HW INT */
+			imap_off = imap_offset(imap_shw);
+			break;
+
+		case 0x29:
+			/* Onboard Keyboard (only) */
+			imap_off = imap_offset(imap_kbd);
+			break;
+
+		case 0x2a:
+			/* Onboard Mouse (only) */
+			imap_off = imap_offset(imap_ms);
+			break;
+
+		case 0x2b:
+			/* Onboard Serial (only) */
+			imap_off = imap_offset(imap_ser);
+			break;
+
+		case 0x32:
+			/* Power Management */
+			imap_off = imap_offset(imap_pmgmt);
+			break;
+
+		default:
+			/* We don't expect anything else.  The other possible
+			 * values are not found in PCI device nodes, and are
+			 * so hardware specific that they should use DCOOKIE's
+			 * anyways.
+			 */
+			prom_printf("psycho_irq_build: Wacky INO [%x]\n", ino);
+			prom_halt();
+		};
+	}
+	imap_off -= imap_offset(imap_a_slot0);
+
+	return pci_irq_encode(imap_off, 0 /* XXX */, ign, ino);
+}
+
+/* #define FIXUP_IRQ_DEBUG */
+
+static void fixup_irq(struct pci_dev *pdev,
+		      struct linux_pbm_info *pbm,
+		      int node)
+{
+	unsigned int prom_irq, portid = pbm->parent->upa_portid;
+	unsigned char pci_irq_line = pdev->irq;
+	int err;
+
+#ifdef FIXUP_IRQ_DEBUG
+	printk("fixup_irq[%s:%s]: ",
+	       pci_strvendor(pdev->vendor),
+	       pci_strdev(pdev->vendor, pdev->device));
+#endif
+	err = prom_getproperty(node, "interrupts", (void *)&prom_irq, sizeof(prom_irq));
+	if(err == 0 || err == -1) {
+		prom_printf("fixup_irq: No interrupts property for dev[%s:%s]\n",
+			    pci_strvendor(pdev->vendor),
+			    pci_strdev(pdev->vendor, pdev->device));
+		prom_halt();
+	}
+
+	/* See if fully specified already (ie. for onboard devices like hme) */
+	if(((prom_irq & PSYCHO_IMAP_IGN) >> 6) == pbm->parent->upa_portid) {
+		pdev->irq = psycho_irq_build(prom_irq);
+#ifdef FIXUP_IRQ_DEBUG
+		printk("fully specified prom_irq[%x] pdev->irq[%x]",
+		       prom_irq, pdev->irq);
+#endif
+	} else {
+		unsigned int bus, slot, line;
+
+		bus = (pbm == &pbm->parent->pbm_B) ? (1 << 4) : 0;
+		line = (pci_irq_line) & 3;
+
+		/* Slot determination is only slightly complex.  Handle
+		 * the easy case first.
+		 */
+		if(pdev->bus->number == pbm->pci_first_busno) {
+			if(pbm == &pbm->parent->pbm_A)
+				slot = (pdev->devfn >> 3) - 1;
+			else
+				slot = ((pdev->devfn >> 3) >> 1) - 1;
+		} else {
+			/* Underneath a bridge, use slot number of parent
+			 * bridge.
+			 */
+			slot = (pdev->bus->self->devfn >> 3) - 1;
+
+			/* Use low slot number bits of child as IRQ line. */
+			line = ((pdev->devfn >> 3) & 3);
+		}
+		slot = (slot << 2);
+
+		pdev->irq = psycho_irq_build((((portid << 6) & PSYCHO_IMAP_IGN) |
+					     (bus | slot | line)));
+#ifdef FIXUP_IRQ_DEBUG
+		do {
+			unsigned char iline, ipin;
+
+			(void)pcibios_read_config_byte(pdev->bus->number,
+						       pdev->devfn,
+						       PCI_INTERRUPT_PIN,
+						       &ipin);
+			(void)pcibios_read_config_byte(pdev->bus->number,
+						       pdev->devfn,
+						       PCI_INTERRUPT_LINE,
+						       &iline);
+			printk("FIXED portid[%x] bus[%x] slot[%x] line[%x] irq[%x] "
+			       "iline[%x] ipin[%x] prom_irq[%x]",
+			       portid, bus>>4, slot>>2, line, pdev->irq,
+			       iline, ipin, prom_irq);
+		} while(0);
+#endif
+	}
+#ifdef FIXUP_IRQ_DEBUG
+	printk("\n");
+#endif
+}
+
+static void fixup_doit(struct pci_dev *pdev,
+		       struct linux_pbm_info *pbm,
+		       struct linux_prom_pci_registers *pregs,
+		       int nregs,
+		       int node)
+{
+	struct linux_prom_pci_registers assigned[PROMREG_MAX];
+	int numaa, err;
+
+	/* Get assigned addresses, if any. */
+	err = prom_getproperty(node, "assigned-addresses",
+			       (char *)&assigned[0], sizeof(assigned));
+	if(err == 0 || err == -1)
+		numaa = 0;
+	else
+		numaa = (err / sizeof(struct linux_prom_pci_registers));
+
+	/* First, scan and fixup base registers. */
+	fixup_regs(pdev, pbm, pregs, nregs, &assigned[0], numaa);
+
+	/* Next, fixup interrupt numbers. */
+	fixup_irq(pdev, pbm, node);
+}
+
+static void fixup_pci_dev(struct pci_dev *pdev,
+			  struct pci_bus *pbus,
+			  struct linux_pbm_info *pbm)
+{
+	struct linux_prom_pci_registers pregs[PROMREG_MAX];
+	struct pcidev_cookie *pcp = pdev->sysdata;
+	int node, nregs, err;
+
+	/* If this is a PCI bridge, we must program it. */
+	if(pdev->class >> 8 == PCI_CLASS_BRIDGE_PCI) {
+		unsigned short cmd;
+
+		/* First, enable bus mastering. */
+		pcibios_read_config_word(pdev->bus->number,
+					 pdev->devfn,
+					 PCI_COMMAND, &cmd);
+		cmd |= PCI_COMMAND_MASTER;
+		pcibios_write_config_word(pdev->bus->number,
+					  pdev->devfn,
+					  PCI_COMMAND, cmd);
+
+		/* Now, set cache line size to 64-bytes. */
+		pcibios_write_config_byte(pdev->bus->number,
+					  pdev->devfn,
+					  PCI_CACHE_LINE_SIZE, 64);
+	}
+
+	/* Ignore if this is one of the PBM's, EBUS, or a
+	 * sub-bridge underneath the PBM.  We only need to fixup
+	 * true devices.
+	 */
+	if((pdev->class >> 8 == PCI_CLASS_BRIDGE_PCI) ||
+	   (pdev->class >> 8 == PCI_CLASS_BRIDGE_HOST) ||
+	   (pdev->class >> 8 == PCI_CLASS_BRIDGE_OTHER) ||
+	   (pcp == NULL))
+		return;
+
+	node = pcp->prom_node;
+
+	err = prom_getproperty(node, "reg", (char *)&pregs[0], sizeof(pregs));
+	if(err == 0 || err == -1) {
+		prom_printf("Cannot find REG for pci_dev\n");
+		prom_halt();
+	}
+
+	nregs = (err / sizeof(pregs[0]));
+
+	fixup_doit(pdev, pbm, &pregs[0], nregs, node);
+}
+
+static void fixup_pci_bus(struct pci_bus *pbus, struct linux_pbm_info *pbm)
+{
+	struct pci_dev *pdev;
+
+	for(pdev = pbus->devices; pdev; pdev = pdev->sibling)
+		fixup_pci_dev(pdev, pbus, pbm);
+
+	for(pbus = pbus->children; pbus; pbus = pbus->children)
+		fixup_pci_bus(pbus, pbm);
+}
+
+static void fixup_addr_irq(struct linux_pbm_info *pbm)
+{
+	struct pci_bus *pbus = &pbm->pci_bus;
+
+	/* Work through top level devices (not bridges, those and their
+	 * devices are handled specially in the next loop).
+	 */
+	fixup_pci_bus(pbus, pbm);
+}
+
+/* Walk all PCI devices probes, fixing up base registers and IRQ registers.
+ * We use OBP for most of this work.
+ */
+static void psycho_final_fixup(struct linux_psycho *psycho)
+{
+	/* Second, fixup base address registers and IRQ lines... */
+	fixup_addr_irq(&psycho->pbm_A);
+	fixup_addr_irq(&psycho->pbm_B);
+
+#if 0
+	prom_halt();
+#endif
+}
+
+unsigned long pcibios_fixup(unsigned long memory_start, unsigned long memory_end)
+{
+	struct linux_psycho *psycho = psycho_root;
+
+	pci_probe_enable = 1;
+
+	/* XXX Really this should be per-PSYCHO, but the config space
+	 * XXX reads and writes give us no way to know which PSYCHO
+	 * XXX in which the config space reads should occur.
+	 * XXX
+	 * XXX Further thought says that we cannot change this generic
+	 * XXX interface, else we'd break xfree86 and other parts of the
+	 * XXX kernel (but whats more important is breaking userland for
+	 * XXX the ix86/Alpha/etc. people).  So we should define our own
+	 * XXX internal extension initially, we can compile our own user
+	 * XXX apps that need to get at PCI configuration space.
+	 */
+
+	/* Probe busses under PBM A. */
+	pbm_probe(&psycho->pbm_A, &memory_start);
+
+	/* Probe busses under PBM B. */
+	pbm_probe(&psycho->pbm_B, &memory_start);
+
+	pci_init_alloc_init(&memory_start);
+
+	/* Walk all PCI devices found.  For each device, and
+	 * PCI bridge which is not one of the PSYCHO PBM's, fill in the
+	 * sysdata with a pointer to the PBM (for pci_bus's) or
+	 * a pci_dev cookie (PBM+PROM_NODE, for pci_dev's).
+	 */
+	fill_in_pbm_cookies(&psycho->pbm_A);
+	fill_in_pbm_cookies(&psycho->pbm_B);
+
+	/* See what OBP has taken care of already. */
+	record_assignments(&psycho->pbm_A);
+	record_assignments(&psycho->pbm_B);
+
+	/* Now, fix it all up. */
+	psycho_final_fixup(psycho);
+
+	pci_init_alloc_fini();
+
+	return ebus_init(memory_start, memory_end);
+}
+
+/* "PCI: The emerging standard..." 8-( */
+volatile int pci_poke_in_progress = 0;
+volatile int pci_poke_faulted = 0;
+
+/* XXX Current PCI support code is broken, it assumes one master PCI config
+ * XXX space exists, on Ultra we can have many of them, especially with
+ * XXX 'dual-pci' boards on Sunfire/Starfire/Wildfire.
+ */
+static char *pci_mkaddr(unsigned char bus, unsigned char device_fn,
+			unsigned char where)
+{
+	unsigned long ret = (unsigned long) psycho_root->pci_config_space;
+
+	ret |= (1 << 24);
+	ret |= ((bus & 0xff) << 16);
+	ret |= ((device_fn & 0xff) << 8);
+	ret |= (where & 0xfc);
+	return (unsigned char *)ret;
+}
+
+static inline int out_of_range(unsigned char bus, unsigned char device_fn)
+{
+	return ((bus == 0 && PCI_SLOT(device_fn) > 4) ||
+		(bus == 1 && PCI_SLOT(device_fn) > 6) ||
+		(pci_probe_enable == 0));
+}
+
+int pcibios_read_config_byte (unsigned char bus, unsigned char device_fn,
+			      unsigned char where, unsigned char *value)
+{
+	unsigned char *addr = pci_mkaddr(bus, device_fn, where);
+	unsigned int word, trapped;
+
+	*value = 0xff;
+
+	if(out_of_range(bus, device_fn))
+		return PCIBIOS_SUCCESSFUL;
+
+	pci_poke_in_progress = 1;
+	pci_poke_faulted = 0;
+	__asm__ __volatile__("membar #Sync\n\t"
+			     "lduwa [%1] %2, %0\n\t"
+			     "membar #Sync"
+			     : "=r" (word)
+			     : "r" (addr), "i" (ASI_PL));
+	pci_poke_in_progress = 0;
+	trapped = pci_poke_faulted;
+	pci_poke_faulted = 0;
+	if(!trapped) {
+		switch(where & 3) {
+		case 0:
+			*value = word & 0xff;
+			break;
+		case 1:
+			*value = (word >> 8) & 0xff;
+			break;
+		case 2:
+			*value = (word >> 16) & 0xff;
+			break;
+		case 3:
+			*value = (word >> 24) & 0xff;
+			break;
+		};
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pcibios_read_config_word (unsigned char bus, unsigned char device_fn,
+			      unsigned char where, unsigned short *value)
+{
+	unsigned short *addr = (unsigned short *)pci_mkaddr(bus, device_fn, where);
+	unsigned int word, trapped;
+
+	*value = 0xffff;
+
+	if(out_of_range(bus, device_fn))
+		return PCIBIOS_SUCCESSFUL;
+
+	pci_poke_in_progress = 1;
+	pci_poke_faulted = 0;
+	__asm__ __volatile__("membar #Sync\n\t"
+			     "lduwa [%1] %2, %0\n\t"
+			     "membar #Sync"
+			     : "=r" (word)
+			     : "r" (addr), "i" (ASI_PL));
+	pci_poke_in_progress = 0;
+	trapped = pci_poke_faulted;
+	pci_poke_faulted = 0;
+	if(!trapped) {
+		switch(where & 3) {
+		case 0:
+			*value = word & 0xffff;
+			break;
+		case 2:
+			*value = (word >> 16) & 0xffff;
+			break;
+		default:
+			printk("pcibios_read_config_word: misaligned "
+			       "reg [%x]\n", where);
+			break;
+		};
+	}
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pcibios_read_config_dword (unsigned char bus, unsigned char device_fn,
+			       unsigned char where, unsigned int *value)
+{
+	unsigned int *addr = (unsigned int *)pci_mkaddr(bus, device_fn, where);
+	unsigned int word, trapped;
+
+	*value = 0xffffffff;
+
+	if(out_of_range(bus, device_fn))
+		return PCIBIOS_SUCCESSFUL;
+
+	pci_poke_in_progress = 1;
+	pci_poke_faulted = 0;
+	__asm__ __volatile__("membar #Sync\n\t"
+			     "lduwa [%1] %2, %0\n\t"
+			     "membar #Sync"
+			     : "=r" (word)
+			     : "r" (addr), "i" (ASI_PL));
+	pci_poke_in_progress = 0;
+	trapped = pci_poke_faulted;
+	pci_poke_faulted = 0;
+	if(!trapped)
+		*value = word;
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pcibios_write_config_byte (unsigned char bus, unsigned char device_fn,
+			       unsigned char where, unsigned char value)
+{
+	unsigned char *addr = pci_mkaddr(bus, device_fn, where);
+
+	if(out_of_range(bus, device_fn))
+		return PCIBIOS_SUCCESSFUL;
+
+	pci_poke_in_progress = 1;
+
+	/* Endianness doesn't matter but we have to get the memory
+	 * barriers in there so...
+	 */
+	__asm__ __volatile__("membar #Sync\n\t"
+			     "stba %0, [%1] %2\n\t"
+			     "membar #Sync\n\t"
+			     : /* no outputs */
+			     : "r" (value), "r" (addr), "i" (ASI_PL));
+	pci_poke_in_progress = 0;
+
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pcibios_write_config_word (unsigned char bus, unsigned char device_fn,
+			       unsigned char where, unsigned short value)
+{
+	unsigned short *addr = (unsigned short *)pci_mkaddr(bus, device_fn, where);
+
+	if(out_of_range(bus, device_fn))
+		return PCIBIOS_SUCCESSFUL;
+
+	pci_poke_in_progress = 1;
+	__asm__ __volatile__("membar #Sync\n\t"
+			     "stha %0, [%1] %2\n\t"
+			     "membar #Sync\n\t"
+			     : /* no outputs */
+			     : "r" (value), "r" (addr), "i" (ASI_PL));
+	pci_poke_in_progress = 0;
+	return PCIBIOS_SUCCESSFUL;
+}
+
+int pcibios_write_config_dword (unsigned char bus, unsigned char device_fn,
+				unsigned char where, unsigned int value)
+{
+	unsigned int *addr = (unsigned int *)pci_mkaddr(bus, device_fn, where);
+
+	if(out_of_range(bus, device_fn))
+		return PCIBIOS_SUCCESSFUL;
+
+	pci_poke_in_progress = 1;
+	__asm__ __volatile__("membar #Sync\n\t"
+			     "stwa %0, [%1] %2\n\t"
+			     "membar #Sync"
+			     : /* no outputs */
+			     : "r" (value), "r" (addr), "i" (ASI_PL));
+	pci_poke_in_progress = 0;
+	return PCIBIOS_SUCCESSFUL;
+}
+
+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;
+
+	lock_kernel();
+	switch(len) {
+	case 1:
+		pcibios_read_config_byte(bus, dfn, off, &ubyte);
+		put_user(ubyte, buf);
+		break;
+	case 2:
+		pcibios_read_config_word(bus, dfn, off, &ushort);
+		put_user(ushort, buf);
+		break;
+	case 4:
+		pcibios_read_config_dword(bus, dfn, off, &uint);
+		put_user(uint, buf);
+		break;
+
+	default:
+		err = -EINVAL;
+		break;
+	};
+	unlock_kernel();
+
+	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;
+
+	lock_kernel();
+	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;
+
+	};
+	unlock_kernel();
+
+	return err;
+}
+
+#endif
diff --git a/arch/sparc64/kernel/ptrace.c b/arch/sparc64/kernel/ptrace.c
index ac91df894..3df35ef14 100644
--- a/arch/sparc64/kernel/ptrace.c
+++ b/arch/sparc64/kernel/ptrace.c
@@ -609,10 +609,6 @@ asmlinkage void do_ptrace(struct pt_regs *regs)
 		unsigned long tmp;
 		int res;
 
-#if 0
-		/* XXX Find out what is really going on. */
-		flush_cache_all();
-#endif
 		/* Non-word alignment _not_ allowed on Sparc. */
 		if (current->tss.flags & SPARC_FLAG_32BIT) {
 			unsigned int x;
@@ -1055,7 +1051,7 @@ asmlinkage void syscall_trace(void)
 	current->exit_code = SIGTRAP;
 	current->state = TASK_STOPPED;
 	current->tss.flags ^= MAGIC_CONSTANT;
-	notify_parent(current);
+	notify_parent(current, SIGCHLD);
 	schedule();
 	/*
 	 * this isn't the same as continuing with a signal, but it will do
@@ -1067,9 +1063,9 @@ asmlinkage void syscall_trace(void)
 		current->pid, current->exit_code);
 #endif
 	if (current->exit_code) {
-		/* spin_lock_irq(&current->sigmask_lock); */
+		spin_lock_irq(&current->sigmask_lock);
 		current->signal |= (1 << (current->exit_code - 1));
-		/* spin_unlock_irq(&current->sigmask_lock); */
+		spin_unlock_irq(&current->sigmask_lock);
 	}
 
 	current->exit_code = 0;
diff --git a/arch/sparc64/kernel/rtrap.S b/arch/sparc64/kernel/rtrap.S
index 9f087a969..a3ac093f3 100644
--- a/arch/sparc64/kernel/rtrap.S
+++ b/arch/sparc64/kernel/rtrap.S
@@ -1,4 +1,4 @@
-/* $Id: rtrap.S,v 1.28 1997/06/30 10:31:39 jj Exp $
+/* $Id: rtrap.S,v 1.33 1997/08/21 09:13:22 davem Exp $
  * rtrap.S: Preparing for return from trap on Sparc V9.
  *
  * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
@@ -36,13 +36,29 @@ rtrap:		sethi			%hi(bh_active), %l2
 		andn			%l1, %l4, %l1
 		be,pt			%icc, to_user
 		 andn			%l7, PSTATE_IE, %l7
-rt_continue:	ld			[%sp + PTREGS_OFF + PT_V9_FPRS], %l2
-		ld			[%g6 + AOFF_task_tss + AOFF_thread_ctx], %l0
-		ldx			[%sp + PTREGS_OFF + PT_V9_G1], %g1
 
-		brnz,pn			%l2, rt_fpu_restore
-		 ldx			[%sp + PTREGS_OFF + PT_V9_G2], %g2
-rt_after_fpu:	ldx			[%sp + PTREGS_OFF + PT_V9_G3], %g3
+		ld			[%sp + PTREGS_OFF + PT_V9_FPRS], %l2
+		andcc			%l2, FPRS_FEF, %g0
+		be,pt			%icc, rt_continue
+		 and			%l2, FPRS_DL, %l6
+		wr			%g0, FPRS_FEF, %fprs
+		ldx			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x108], %g5
+		membar			#StoreLoad | #LoadLoad
+		brz,pn			%l6, 1f
+		 wr			%g0, ASI_BLK_P, %asi
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x000] %asi, %f0
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x040] %asi, %f16
+1:		andcc			%l2, FPRS_DU, %g0
+		be,pn			%icc, 1f
+		 wr			%g5, 0, %gsr
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x080] %asi, %f32
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x0c0] %asi, %f48
+1:		membar			#Sync
+		ldx			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x100], %fsr
+rt_continue:	ld			[%g6 + AOFF_task_tss + AOFF_thread_ctx], %l0
+		ldx			[%sp + PTREGS_OFF + PT_V9_G1], %g1
+		ldx			[%sp + PTREGS_OFF + PT_V9_G2], %g2
+		ldx			[%sp + PTREGS_OFF + PT_V9_G3], %g3
 		mov			%g6, %l6
 		ldx			[%sp + PTREGS_OFF + PT_V9_G4], %g4
 		ldx			[%sp + PTREGS_OFF + PT_V9_G5], %g5
@@ -69,10 +85,12 @@ rt_after_fpu:	ldx			[%sp + PTREGS_OFF + PT_V9_G3], %g3
 
 		wrpr			%l1, %g0, %tstate
 		wrpr			%l2, %g0, %tpc
+		wrpr			%o2, %g0, %tnpc
 		mov			PRIMARY_CONTEXT, %l7
 		brnz,pn			%l3, kern_rtt
-		 wrpr			%o2, %g0, %tnpc
+		 mov			SECONDARY_CONTEXT, %l5
 		stxa			%l0, [%l7] ASI_DMMU
+		stxa			%l0, [%l5] ASI_DMMU
 		flush			%l6
 		rdpr			%wstate, %l1
 
@@ -88,10 +106,14 @@ rt_after_fpu:	ldx			[%sp + PTREGS_OFF + PT_V9_G3], %g3
 		retry
 kern_rtt:	restore
 		retry
-to_user:	sethi			%hi(need_resched), %l0
-		ld			[%l0 + %lo(need_resched)], %l0
+to_user:	lduw			[%g6 + AOFF_task_processor], %o0
+		mov			1, %o1
+		sethi			%hi(need_resched), %l0
+		ldx			[%l0 + %lo(need_resched)], %l0
+		sllx			%o1, %o0, %o1
 		wrpr			%l7, PSTATE_IE, %pstate
-		brz,pt			%l0, check_signal
+		andcc			%o1, %l0, %g0
+		be,pt			%xcc, check_signal
 		 ldx			[%g6 + AOFF_task_signal], %l0
 
 		call			schedule
@@ -99,36 +121,123 @@ to_user:	sethi			%hi(need_resched), %l0
 		ldx			[%g6 + AOFF_task_signal], %l0
 		nop
 check_signal:	ldx			[%g6 + AOFF_task_blocked], %o0
+		ld			[%sp + PTREGS_OFF + PT_V9_FPRS], %l2
 		andncc			%l0, %o0, %g0
 		be,pt			%xcc, check_user_wins
 		 ldx			[%g6 + AOFF_task_tss + AOFF_thread_w_saved], %o2
-
 		mov			%l5, %o2  
 		mov			%l6, %o3
 		call			do_signal
 		 add			%sp, STACK_BIAS + REGWIN_SZ, %o1
 		ldx			[%g6 + AOFF_task_tss + AOFF_thread_w_saved], %o2
 		clr			%l6
-check_user_wins:
-		brz,pt			%o2, rt_continue
-		 nop
+check_user_wins:brz,pt			%o2, 1f
+		 sethi			%hi(TSTATE_PEF), %o3
 
 		call			fault_in_user_windows
 		 add			%sp, STACK_BIAS + REGWIN_SZ, %o0
-		ba,a,pt			%xcc, rt_continue
-rt_fpu_restore:	wr			%g0, FPRS_FEF, %fprs
-		add			%sp, PTREGS_OFF + TRACEREG_SZ, %g4
-		wr			%g0, ASI_BLK_P, %asi
 
+		sethi			%hi(TSTATE_PEF), %o3
+1:		andcc			%l2, FPRS_FEF, %g0
+		be,a,pt			%icc, rt_continue
+		 andn			%l1, %o3, %l1		! If fprs.FEF is not set, disable tstate.PEF
+		ldx			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x108], %o3
+		ld			[%g6 + AOFF_task_tss + AOFF_thread_flags], %l2
+		wr			%g0, FPRS_FEF, %fprs
+		wr			%o3, 0, %gsr
+		andcc			%l2, SPARC_FLAG_USEDFPUL, %g0
+		bne,pn			%icc, 2f
+		 andcc			%l2, SPARC_FLAG_USEDFPUU, %g0
+		fzero			%f0
+		bne,pn			%icc, 1f
+		 fzero			%f2
+		faddd			%f0, %f2, %f4
+		fmuld			%f0, %f2, %f6
+		faddd			%f0, %f2, %f8
+		fmuld			%f0, %f2, %f10
+		faddd			%f0, %f2, %f12
+		fmuld			%f0, %f2, %f14
+		faddd			%f0, %f2, %f16
+		fmuld			%f0, %f2, %f18
+		faddd			%f0, %f2, %f20
+		fmuld			%f0, %f2, %f22
+		faddd			%f0, %f2, %f24
+		fmuld			%f0, %f2, %f26
+		faddd			%f0, %f2, %f28
+		fmuld			%f0, %f2, %f30
+		faddd			%f0, %f2, %f32
+		fmuld			%f0, %f2, %f34
+		faddd			%f0, %f2, %f36
+		fmuld			%f0, %f2, %f38
+		faddd			%f0, %f2, %f40
+		fmuld			%f0, %f2, %f42
+		faddd			%f0, %f2, %f44
+		fmuld			%f0, %f2, %f46
+		faddd			%f0, %f2, %f48
+		fmuld			%f0, %f2, %f50
+		faddd			%f0, %f2, %f52
+		fmuld			%f0, %f2, %f54
+		faddd			%f0, %f2, %f56
+		fmuld			%f0, %f2, %f58
+		faddd			%f0, %f2, %f60
+		ldx			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x100], %fsr
+		ba,pt			%xcc, rt_continue
+		 wr			%g0, FPRS_FEF, %fprs
+1:		wr			%g0, ASI_BLK_P, %asi
 		membar			#StoreLoad | #LoadLoad
-		ldda			[%g4 + 0x000] %asi, %f0
-		ldda			[%g4 + 0x040] %asi, %f16
-		ldda			[%g4 + 0x080] %asi, %f32
-		ldda			[%g4 + 0x0c0] %asi, %f48
-		ldx			[%g4 + 0x100], %fsr
-		ldx			[%g4 + 0x108], %g3
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x080] %asi, %f32
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x0c0] %asi, %f48
+		faddd			%f0, %f2, %f4
+		fmuld			%f0, %f2, %f6
+		faddd			%f0, %f2, %f8
+		fmuld			%f0, %f2, %f10
+		faddd			%f0, %f2, %f12
+		fmuld			%f0, %f2, %f14
+		faddd			%f0, %f2, %f16
+		fmuld			%f0, %f2, %f18
+		faddd			%f0, %f2, %f20
+		fmuld			%f0, %f2, %f22
+		faddd			%f0, %f2, %f24
+		fmuld			%f0, %f2, %f26
+		faddd			%f0, %f2, %f28
+		fmuld			%f0, %f2, %f30
+		membar			#Sync
+		ldx			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x100], %fsr
+		ba,pt			%xcc, rt_continue
+		 wr			%g0, FPRS_FEF, %fprs
+2:		membar			#StoreLoad | #LoadLoad
+		andcc			%l2, SPARC_FLAG_USEDFPUU, %g0
+		bne,pt			%icc, 3f
+		 wr			%g0, ASI_BLK_P, %asi
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x000] %asi, %f0
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x040] %asi, %f16
+		fzero			%f32
+		fzero			%f34
+		faddd			%f32, %f34, %f36
+		fmuld			%f32, %f34, %f38
+		faddd			%f32, %f34, %f40
+		fmuld			%f32, %f34, %f42
+		faddd			%f32, %f34, %f44
+		fmuld			%f32, %f34, %f46
+		faddd			%f32, %f34, %f48
+		fmuld			%f32, %f34, %f50
+		faddd			%f32, %f34, %f52
+		fmuld			%f32, %f34, %f54
+		faddd			%f32, %f34, %f56
+		fmuld			%f32, %f34, %f58
+		faddd			%f32, %f34, %f60
+		fmuld			%f32, %f34, %f62
+		membar			#Sync
+		ldx			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x100], %fsr
+		ba,pt			%xcc, rt_continue
+		 wr			%g0, FPRS_FEF, %fprs
+3:		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x000] %asi, %f0
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x040] %asi, %f16
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x080] %asi, %f32
+		ldda			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x0c0] %asi, %f48
 		membar			#Sync
+		ldx			[%sp + PTREGS_OFF + TRACEREG_SZ + 0x100], %fsr
+		ba,pt			%xcc, rt_continue
+		 wr			%g0, FPRS_FEF, %fprs
 
-		b,pt			%xcc, rt_after_fpu
-		 wr			%g3, 0, %gsr
 #undef PTREGS_OFF
diff --git a/arch/sparc64/kernel/setup.c b/arch/sparc64/kernel/setup.c
index 7eb47e976..147c6e55d 100644
--- a/arch/sparc64/kernel/setup.c
+++ b/arch/sparc64/kernel/setup.c
@@ -1,4 +1,4 @@
-/*  $Id: setup.c,v 1.11 1997/07/24 12:15:05 davem Exp $
+/*  $Id: setup.c,v 1.12 1997/08/28 02:23:19 ecd Exp $
  *  linux/arch/sparc64/kernel/setup.c
  *
  *  Copyright (C) 1995,1996  David S. Miller (davem@caip.rutgers.edu)
@@ -233,8 +233,6 @@ extern unsigned ramdisk_size;
 
 extern int root_mountflags;
 
-extern void register_console(void (*proc)(const char *));
-
 char saved_command_line[256];
 char reboot_command[256];
 
diff --git a/arch/sparc64/kernel/signal.c b/arch/sparc64/kernel/signal.c
index cfc55fc2e..87241f8e3 100644
--- a/arch/sparc64/kernel/signal.c
+++ b/arch/sparc64/kernel/signal.c
@@ -1,4 +1,4 @@
-/*  $Id: signal.c,v 1.20 1997/07/14 03:10:28 davem Exp $
+/*  $Id: signal.c,v 1.24 1997/09/02 20:53:03 davem Exp $
  *  arch/sparc64/kernel/signal.c
  *
  *  Copyright (C) 1991, 1992  Linus Torvalds
@@ -54,12 +54,12 @@ asmlinkage void sparc64_set_context(struct pt_regs *regs)
 	if(tp->w_saved						||
 	   (((unsigned long)ucp) & (sizeof(unsigned long)-1))	||
 	   (!__access_ok((unsigned long)ucp, sizeof(*ucp))))
-		do_exit(SIGSEGV);
+		goto do_sigsegv;
 	grp = &ucp->uc_mcontext.mc_gregs;
 	__get_user(pc, &((*grp)[MC_PC]));
 	__get_user(npc, &((*grp)[MC_NPC]));
 	if((pc | npc) & 3)
-		do_exit(SIGSEGV);
+		goto do_sigsegv;
 	if(regs->u_regs[UREG_I1]) {
 		__get_user(current->blocked, &ucp->uc_sigmask);
 		current->blocked &= _BLOCKABLE;
@@ -67,7 +67,7 @@ asmlinkage void sparc64_set_context(struct pt_regs *regs)
 	regs->tpc = pc;
 	regs->tnpc = npc;
 	__get_user(regs->y, &((*grp)[MC_Y]));
-	__get_user(tstate, &((*grp)[MC_Y]));
+	__get_user(tstate, &((*grp)[MC_TSTATE]));
 	regs->tstate &= ~(TSTATE_ICC | TSTATE_XCC);
 	regs->tstate |= (tstate & (TSTATE_ICC | TSTATE_XCC));
 	__get_user(regs->u_regs[UREG_G1], (&(*grp)[MC_G1]));
@@ -94,12 +94,23 @@ asmlinkage void sparc64_set_context(struct pt_regs *regs)
 	__get_user(fenab, &(ucp->uc_mcontext.mc_fpregs.mcfpu_enab));
 	if(fenab) {
 		unsigned long *fpregs = (unsigned long *)(regs+1);
-		copy_from_user(fpregs, &(ucp->uc_mcontext.mc_fpregs.mcfpu_fregs),
-			       (sizeof(unsigned long) * 32));
+		unsigned long fprs;
+		__get_user(fprs, &(ucp->uc_mcontext.mc_fpregs.mcfpu_fprs));
+		if (fprs & FPRS_DL)
+			copy_from_user(fpregs, &(ucp->uc_mcontext.mc_fpregs.mcfpu_fregs),
+			       	       (sizeof(unsigned int) * 32));
+		if (fprs & FPRS_DU)
+			copy_from_user(fpregs+16, ((unsigned long *)&(ucp->uc_mcontext.mc_fpregs.mcfpu_fregs))+16,
+			       	       (sizeof(unsigned int) * 32));
 		__get_user(fpregs[32], &(ucp->uc_mcontext.mc_fpregs.mcfpu_fsr));
 		__get_user(fpregs[33], &(ucp->uc_mcontext.mc_fpregs.mcfpu_gsr));
-		regs->fprs = FPRS_FEF;
+		regs->fprs = fprs;
+		regs->tstate |= TSTATE_PEF;
 	}
+	return;
+do_sigsegv:
+	lock_kernel();
+	do_exit(SIGSEGV);
 }
 
 asmlinkage void sparc64_get_context(struct pt_regs *regs)
@@ -109,11 +120,11 @@ asmlinkage void sparc64_get_context(struct pt_regs *regs)
 	mc_gregset_t *grp;
 	mcontext_t *mcp;
 	unsigned long fp, i7;
-	unsigned char fenab = (current->flags & PF_USEDFPU);
+	unsigned char fenab = (current->tss.flags & SPARC_FLAG_USEDFPU);
 
 	synchronize_user_stack();
 	if(tp->w_saved || clear_user(ucp, sizeof(*ucp)))
-		do_exit(SIGSEGV);
+		goto do_sigsegv;
 	mcp = &ucp->uc_mcontext;
 	grp = &mcp->mc_gregs;
 
@@ -150,12 +161,30 @@ asmlinkage void sparc64_get_context(struct pt_regs *regs)
 	__put_user(fenab, &(mcp->mc_fpregs.mcfpu_enab));
 	if(fenab) {
 		unsigned long *fpregs = (unsigned long *)(regs+1);
-		copy_to_user(&(mcp->mc_fpregs.mcfpu_fregs), fpregs,
-			     (sizeof(unsigned long) * 32));
+		unsigned long fprs;
+		
+		fprs = (regs->fprs & FPRS_FEF) | 
+		       (current->tss.flags & (SPARC_FLAG_USEDFPUL | SPARC_FLAG_USEDFPUU));
+		if (fprs & FPRS_DL)
+			copy_to_user(&(mcp->mc_fpregs.mcfpu_fregs), fpregs,
+				     (sizeof(unsigned int) * 32));
+		else
+			clear_user(&(mcp->mc_fpregs.mcfpu_fregs),
+				     (sizeof(unsigned int) * 32));
+		if (fprs & FPRS_DU)
+			copy_to_user(((unsigned long *)&(mcp->mc_fpregs.mcfpu_fregs))+16, fpregs+16,
+				     (sizeof(unsigned int) * 32));
+		else
+			clear_user(((unsigned long *)&(mcp->mc_fpregs.mcfpu_fregs))+16,
+				     (sizeof(unsigned int) * 32));
 		__put_user(fpregs[32], &(mcp->mc_fpregs.mcfpu_fsr));
 		__put_user(fpregs[33], &(mcp->mc_fpregs.mcfpu_gsr));
-		__put_user(FPRS_FEF, &(mcp->mc_fpregs.mcfpu_fprs));
+		__put_user(fprs, &(mcp->mc_fpregs.mcfpu_fprs));
 	}
+	return;
+do_sigsegv:
+	lock_kernel();
+	do_exit(SIGSEGV);
 }
 
 /* 
@@ -233,11 +262,19 @@ static inline void
 restore_fpu_state(struct pt_regs *regs, __siginfo_fpu_t *fpu)
 {
 	unsigned long *fpregs = (unsigned long *)(regs+1);
-	copy_from_user(fpregs, &fpu->si_float_regs[0],
-		       (sizeof(unsigned int) * 64));
+	unsigned long fprs;
+	
+	__get_user(fprs, &fpu->si_fprs);
+	if (fprs & FPRS_DL)
+		copy_from_user(fpregs, &fpu->si_float_regs[0],
+		       	       (sizeof(unsigned int) * 32));
+	if (fprs & FPRS_DU)
+		copy_from_user(fpregs+16, &fpu->si_float_regs[32],
+		       	       (sizeof(unsigned int) * 32));
 	__get_user(fpregs[32], &fpu->si_fsr);
 	__get_user(fpregs[33], &fpu->si_gsr);
-	regs->fprs = FPRS_FEF;
+	regs->fprs = fprs;
+	regs->tstate |= TSTATE_PEF;
 }
 
 void do_sigreturn(struct pt_regs *regs)
@@ -304,11 +341,26 @@ static inline void
 save_fpu_state(struct pt_regs *regs, __siginfo_fpu_t *fpu)
 {
 	unsigned long *fpregs = (unsigned long *)(regs+1);
-	copy_to_user(&fpu->si_float_regs[0], fpregs,
-		     (sizeof(unsigned int) * 64));
+	unsigned long fprs;
+	
+	fprs = (regs->fprs & FPRS_FEF) | 
+		(current->tss.flags & (SPARC_FLAG_USEDFPUL | SPARC_FLAG_USEDFPUU));
+	if (fprs & FPRS_DL)
+		copy_to_user(&fpu->si_float_regs[0], fpregs,
+		     	     (sizeof(unsigned int) * 32));
+	else
+		clear_user(&fpu->si_float_regs[0],
+		     	     (sizeof(unsigned int) * 32));
+	if (fprs & FPRS_DU)
+		copy_to_user(&fpu->si_float_regs[32], fpregs+16,
+		     	     (sizeof(unsigned int) * 32));
+	else
+		clear_user(&fpu->si_float_regs[32],
+		     	     (sizeof(unsigned int) * 32));
 	__put_user(fpregs[32], &fpu->si_fsr);
 	__put_user(fpregs[33], &fpu->si_gsr);
-	regs->fprs = 0;
+	__put_user(fprs, &fpu->si_fprs);
+	regs->tstate &= ~TSTATE_PEF;
 }
 
 static inline void
@@ -321,7 +373,7 @@ new_setup_frame(struct sigaction *sa, struct pt_regs *regs,
 	/* 1. Make sure everything is clean */
 	synchronize_user_stack();
 	sigframe_size = NF_ALIGNEDSZ;
-	if (!(current->flags & PF_USEDFPU))
+	if (!(current->tss.flags & SPARC_FLAG_USEDFPU))
 		sigframe_size -= sizeof(__siginfo_fpu_t);
 
 	sf = (struct new_signal_frame *)
@@ -339,7 +391,7 @@ new_setup_frame(struct sigaction *sa, struct pt_regs *regs,
 	/* 2. Save the current process state */
 	copy_to_user(&sf->info.si_regs, regs, sizeof (*regs));
 
-	if (current->flags & PF_USEDFPU) {
+	if (current->tss.flags & SPARC_FLAG_USEDFPU) {
 		save_fpu_state(regs, &sf->fpu_state);
 		__put_user((u64)&sf->fpu_state, &sf->fpu_save);
 	} else {
@@ -452,7 +504,7 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs * regs,
 		if ((current->flags & PF_PTRACED) && signr != SIGKILL) {
 			current->exit_code = signr;
 			current->state = TASK_STOPPED;
-			notify_parent(current);
+			notify_parent(current, SIGCHLD);
 			schedule();
 			if (!(signr = current->exit_code))
 				continue;
@@ -498,7 +550,7 @@ asmlinkage int do_signal(unsigned long oldmask, struct pt_regs * regs,
 				current->exit_code = signr;
 				if(!(current->p_pptr->sig->action[SIGCHLD-1].sa_flags &
 				     SA_NOCLDSTOP))
-					notify_parent(current);
+					notify_parent(current, SIGCHLD);
 				schedule();
 				continue;
 
diff --git a/arch/sparc64/kernel/signal32.c b/arch/sparc64/kernel/signal32.c
index 5135c2ae5..16cec2a72 100644
--- a/arch/sparc64/kernel/signal32.c
+++ b/arch/sparc64/kernel/signal32.c
@@ -1,4 +1,4 @@
-/*  $Id: signal32.c,v 1.26 1997/07/14 03:10:31 davem Exp $
+/*  $Id: signal32.c,v 1.30 1997/08/29 15:51:33 jj Exp $
  *  arch/sparc64/kernel/signal32.c
  *
  *  Copyright (C) 1991, 1992  Linus Torvalds
@@ -117,9 +117,17 @@ asmlinkage void _sigpause32_common(unsigned int set, struct pt_regs *regs)
 static inline void restore_fpu_state32(struct pt_regs *regs, __siginfo_fpu_t *fpu)
 {
 	unsigned long *fpregs = (unsigned long *)(regs + 1);
-	copy_from_user(fpregs, &fpu->si_float_regs[0], (sizeof(unsigned int) * 64));
+	unsigned long fprs;
+	
+	__get_user(fprs, &fpu->si_fprs);
+	if (fprs & FPRS_DL)
+		copy_from_user(fpregs, &fpu->si_float_regs[0], (sizeof(unsigned int) * 32));
+	if (fprs & FPRS_DU)
+		copy_from_user(fpregs+16, &fpu->si_float_regs[32], (sizeof(unsigned int) * 32));
 	__get_user(fpregs[32], &fpu->si_fsr);
 	__get_user(fpregs[33], &fpu->si_gsr);
+	regs->fprs = fprs;
+	regs->tstate |= TSTATE_PEF;
 }
 
 void do_new_sigreturn32(struct pt_regs *regs)
@@ -169,8 +177,10 @@ void do_new_sigreturn32(struct pt_regs *regs)
 	regs->tstate &= ~(TSTATE_ICC);
 	regs->tstate |= psr_to_tstate_icc(psr);
 
+#if 0
 	if (psr & PSR_EF)
-		regs->fprs = FPRS_FEF;
+		regs->tstate |= TSTATE_PEF;
+#endif		
 
 	__get_user(fpu_save, &sf->fpu_save);
 	if (fpu_save)
@@ -273,7 +283,7 @@ setup_frame32(struct sigaction *sa, unsigned long pc, unsigned long npc,
 	__put_user(pc, &sc->sigc_pc);
 	__put_user(npc, &sc->sigc_npc);
 	psr = tstate_to_psr (regs->tstate);
-	if(current->flags & PF_USEDFPU)
+	if(current->tss.flags & SPARC_FLAG_USEDFPU)
 		psr |= PSR_EF;
 	__put_user(psr, &sc->sigc_psr);
 	__put_user(regs->u_regs[UREG_G1], &sc->sigc_g1);
@@ -318,10 +328,22 @@ setup_frame32(struct sigaction *sa, unsigned long pc, unsigned long npc,
 static inline void save_fpu_state32(struct pt_regs *regs, __siginfo_fpu_t *fpu)
 {
 	unsigned long *fpregs = (unsigned long *)(regs+1);
-	copy_to_user(&fpu->si_float_regs[0], fpregs, (sizeof(unsigned int) * 64));
+	unsigned long fprs;
+	
+	fprs = (regs->fprs & FPRS_FEF) |
+		(current->tss.flags & (SPARC_FLAG_USEDFPUL | SPARC_FLAG_USEDFPUU));
+	if (fprs & FPRS_DL)
+		copy_to_user(&fpu->si_float_regs[0], fpregs, (sizeof(unsigned int) * 32));
+	else
+		clear_user(&fpu->si_float_regs[0], (sizeof(unsigned int) * 32));
+	if (fprs & FPRS_DU)
+		copy_to_user(&fpu->si_float_regs[32], fpregs+16, (sizeof(unsigned int) * 32));
+	else
+		clear_user(&fpu->si_float_regs[32], (sizeof(unsigned int) * 32));
 	__put_user(fpregs[32], &fpu->si_fsr);
 	__put_user(fpregs[33], &fpu->si_gsr);
-	regs->fprs = 0;
+	__put_user(fprs, &fpu->si_fprs);
+	regs->tstate &= ~TSTATE_PEF;
 }
 
 static inline void new_setup_frame32(struct sigaction *sa, struct pt_regs *regs,
@@ -335,7 +357,7 @@ static inline void new_setup_frame32(struct sigaction *sa, struct pt_regs *regs,
 	/* 1. Make sure everything is clean */
 	synchronize_user_stack();
 	sigframe_size = NF_ALIGNEDSZ;
-	if (!(current->flags & PF_USEDFPU))
+	if (!(current->tss.flags & SPARC_FLAG_USEDFPU))
 		sigframe_size -= sizeof(__siginfo_fpu_t);
 
 	regs->u_regs[UREG_FP] &= 0x00000000ffffffffUL;
@@ -362,7 +384,7 @@ static inline void new_setup_frame32(struct sigaction *sa, struct pt_regs *regs,
 	__put_user(regs->tnpc, &sf->info.si_regs.npc);
 	__put_user(regs->y, &sf->info.si_regs.y);
 	psr = tstate_to_psr (regs->tstate);
-	if(current->flags & PF_USEDFPU)
+	if(current->tss.flags & SPARC_FLAG_USEDFPU)
 		psr |= PSR_EF;
 	__put_user(psr, &sf->info.si_regs.psr);
 	for (i = 0; i < 16; i++)
@@ -469,7 +491,7 @@ setup_svr4_frame32(struct sigaction *sa, unsigned long pc, unsigned long npc,
 	__put_user(regs->tpc, &((*gr) [SVR4_PC]));
 	__put_user(regs->tnpc, &((*gr) [SVR4_NPC]));
 	psr = tstate_to_psr (regs->tstate);
-	if(current->flags & PF_USEDFPU)
+	if(current->tss.flags & SPARC_FLAG_USEDFPU)
 		psr |= PSR_EF;
 	__put_user(psr, &((*gr) [SVR4_PSR]));
 	__put_user(regs->y, &((*gr) [SVR4_Y]));
@@ -488,7 +510,7 @@ setup_svr4_frame32(struct sigaction *sa, unsigned long pc, unsigned long npc,
 	/* Save the currently window file: */
 
 	/* 1. Link sfp->uc->gwins to our windows */
-	__put_user(gw, &mc->gwin);
+	__put_user((u32)(long)gw, &mc->gwin);
 	    
 	/* 2. Number of windows to restore at setcontext (): */
 	__put_user(current->tss.w_saved, &gw->count);
@@ -506,9 +528,9 @@ setup_svr4_frame32(struct sigaction *sa, unsigned long pc, unsigned long npc,
 	 */
 #if 0	 
 	for(window = 0; window < current->tss.w_saved; window++) {
-		__put_user((int *) &(gw->win [window]), &gw->winptr [window]);
+		__put_user((int *) &(gw->win [window]), (int **)gw->winptr +window );
 		copy_to_user(&gw->win [window], &current->tss.reg_window [window], sizeof (svr4_rwindow_t));
-		__put_user(0, gw->winptr [window]);
+		__put_user(0, (int *)gw->winptr + window);
 	}
 #endif	
 
@@ -546,7 +568,7 @@ setup_svr4_frame32(struct sigaction *sa, unsigned long pc, unsigned long npc,
 }
 
 asmlinkage int
-svr4_getcontext32(svr4_ucontext_t *uc, struct pt_regs *regs)
+svr4_getcontext(svr4_ucontext_t *uc, struct pt_regs *regs)
 {
 	svr4_gregset_t  *gr;
 	svr4_mcontext_t *mc;
@@ -555,6 +577,7 @@ svr4_getcontext32(svr4_ucontext_t *uc, struct pt_regs *regs)
 	synchronize_user_stack();
 	if (current->tss.w_saved){
 		printk ("Uh oh, w_saved is not zero (%ld)\n", current->tss.w_saved);
+		lock_kernel();
 		do_exit (SIGSEGV);
 	}
 	if(clear_user(uc, sizeof (*uc)))
@@ -571,7 +594,7 @@ svr4_getcontext32(svr4_ucontext_t *uc, struct pt_regs *regs)
 	__put_user(regs->tpc, &uc->mcontext.greg [SVR4_PC]);
 	__put_user(regs->tnpc, &uc->mcontext.greg [SVR4_NPC]);
 	__put_user((tstate_to_psr(regs->tstate) |
-		    ((current->flags & PF_USEDFPU) ? PSR_EF : 0)),
+		    ((current->tss.flags & SPARC_FLAG_USEDFPU) ? PSR_EF : 0)),
 		   &uc->mcontext.greg [SVR4_PSR]);
         __put_user(regs->y, &uc->mcontext.greg [SVR4_Y]);
 	
@@ -594,7 +617,7 @@ svr4_getcontext32(svr4_ucontext_t *uc, struct pt_regs *regs)
 
 
 /* Set the context for a svr4 application, this is Solaris way to sigreturn */
-asmlinkage int svr4_setcontext32(svr4_ucontext_t *c, struct pt_regs *regs)
+asmlinkage int svr4_setcontext(svr4_ucontext_t *c, struct pt_regs *regs)
 {
 	struct thread_struct *tp = &current->tss;
 	svr4_gregset_t  *gr;
@@ -639,9 +662,10 @@ asmlinkage int svr4_setcontext32(svr4_ucontext_t *c, struct pt_regs *regs)
 	__get_user(psr, &((*gr) [SVR4_PSR]));
 	regs->tstate &= ~(TSTATE_ICC);
 	regs->tstate |= psr_to_tstate_icc(psr);
+#if 0	
 	if(psr & PSR_EF)
-		regs->fprs = FPRS_FEF;
-
+		regs->tstate |= TSTATE_PEF;
+#endif
 	/* Restore g[1..7] and o[0..7] registers */
 	for (i = 0; i < 7; i++)
 		__get_user(regs->u_regs[UREG_G1+i], (&(*gr)[SVR4_G1])+i);
@@ -718,7 +742,7 @@ asmlinkage int do_signal32(unsigned long oldmask, struct pt_regs * regs,
 		if ((current->flags & PF_PTRACED) && signr != SIGKILL) {
 			current->exit_code = signr;
 			current->state = TASK_STOPPED;
-			notify_parent(current);
+			notify_parent(current, SIGCHLD);
 			schedule();
 			if (!(signr = current->exit_code))
 				continue;
@@ -764,7 +788,7 @@ asmlinkage int do_signal32(unsigned long oldmask, struct pt_regs * regs,
 				current->exit_code = signr;
 				if(!(current->p_pptr->sig->action[SIGCHLD-1].sa_flags &
 				     SA_NOCLDSTOP))
-					notify_parent(current);
+					notify_parent(current, SIGCHLD);
 				schedule();
 				continue;
 
diff --git a/arch/sparc64/kernel/smp.c b/arch/sparc64/kernel/smp.c
index 77ccf40a0..8dd471be6 100644
--- a/arch/sparc64/kernel/smp.c
+++ b/arch/sparc64/kernel/smp.c
@@ -10,11 +10,12 @@
 #include <linux/smp_lock.h>
 #include <linux/interrupt.h>
 #include <linux/kernel_stat.h>
+#include <linux/delay.h>
 
+#include <asm/head.h>
 #include <asm/ptrace.h>
 #include <asm/atomic.h>
 
-#include <asm/delay.h>
 #include <asm/irq.h>
 #include <asm/page.h>
 #include <asm/pgtable.h>
@@ -33,9 +34,9 @@ volatile int smp_processors_ready = 0;
 unsigned long cpu_present_map = 0;
 int smp_num_cpus = 1;
 int smp_threads_ready = 0;
-volatile unsigned long cpu_callin_map[NR_CPUS] = {0,};
 
-struct cpuinfo_sparc cpu_data[NR_CPUS];
+struct cpuinfo_sparc cpu_data[NR_CPUS] __attribute__ ((aligned (64)));
+
 static unsigned char boot_cpu_id = 0;
 static int smp_activated = 0;
 
@@ -44,8 +45,6 @@ volatile int cpu_logical_map[NR_CPUS];
 
 struct klock_info klock_info = { KLOCK_CLEAR, 0 };
 
-static volatile int smp_commenced = 0;
-
 void smp_setup(char *str, int *ints)
 {
 	/* XXX implement me XXX */
@@ -68,45 +67,64 @@ char *smp_info(void)
 
 void smp_store_cpu_info(int id)
 {
-	cpu_data[id].udelay_val = loops_per_sec;
+	cpu_data[id].udelay_val			= loops_per_sec;
+	cpu_data[id].irq_count			= 0;
+	cpu_data[id].last_tlbversion_seen	= tlb_context_cache & CTX_VERSION_MASK;
+	cpu_data[id].pgcache_size		= 0;
+	cpu_data[id].pgd_cache			= NULL;
+	cpu_data[id].pmd_cache			= NULL;
+	cpu_data[id].pte_cache			= NULL;
 }
 
+extern void distribute_irqs(void);
+
 void smp_commence(void)
 {
-	flush_cache_all();
-	flush_tlb_all();
-	smp_commenced = 1;
-	flush_cache_all();
-	flush_tlb_all();
+	distribute_irqs();
 }
 
 static void smp_setup_percpu_timer(void);
 
 static volatile unsigned long callin_flag = 0;
 
+extern void inherit_locked_prom_mappings(int save_p);
+extern void cpu_probe(void);
+
 void smp_callin(void)
 {
 	int cpuid = hard_smp_processor_id();
 
-	flush_cache_all();
-	flush_tlb_all();
+	inherit_locked_prom_mappings(0);
+
+	__flush_cache_all();
+	__flush_tlb_all();
+
+	cpu_probe();
+
+	/* Master did this already, now is the time for us to do it. */
+	__asm__ __volatile__("
+	sethi	%%hi(0x80000000), %%g1
+	sllx	%%g1, 32, %%g1
+	rd	%%tick, %%g2
+	add	%%g2, 6, %%g2
+	andn	%%g2, %%g1, %%g2
+	wrpr	%%g2, 0, %%tick
+"	: /* no outputs */
+	: /* no inputs */
+	: "g1", "g2");
 
 	smp_setup_percpu_timer();
 
+	__sti();
+
 	calibrate_delay();
 	smp_store_cpu_info(cpuid);
 	callin_flag = 1;
 	__asm__ __volatile__("membar #Sync\n\t"
 			     "flush  %%g6" : : : "memory");
 
-	while(!task[cpuid])
-		barrier();
-	current = task[cpuid];
-
-	while(!smp_commenced)
-		barrier();
-
-	__sti();
+	while(!smp_processors_ready)
+		membar("#LoadLoad");
 }
 
 extern int cpu_idle(void *unused);
@@ -130,18 +148,22 @@ void cpu_panic(void)
 	panic("SMP bolixed\n");
 }
 
+static void smp_tickoffset_init(void);
+
 extern struct prom_cpuinfo linux_cpus[NR_CPUS];
-extern unsigned long sparc64_cpu_startup;
+
+extern unsigned long smp_trampoline;
 
 void smp_boot_cpus(void)
 {
 	int cpucount = 0, i;
 
 	printk("Entering UltraSMPenguin Mode...\n");
+	smp_tickoffset_init();
 	__sti();
 	cpu_present_map = 0;
 	for(i = 0; i < linux_num_cpus; i++)
-		cpu_present_map |= (1 << i);
+		cpu_present_map |= (1UL << i);
 	for(i = 0; i < NR_CPUS; i++) {
 		cpu_number_map[i] = -1;
 		cpu_logical_map[i] = -1;
@@ -160,23 +182,24 @@ void smp_boot_cpus(void)
 		if(i == boot_cpu_id)
 			continue;
 
-		if(cpu_present_map & (1 << i)) {
+		if(cpu_present_map & (1UL << i)) {
+			unsigned long entry = (unsigned long)(&smp_trampoline);
 			struct task_struct *p;
 			int timeout;
 
+			entry -= KERNBASE;
 			kernel_thread(start_secondary, NULL, CLONE_PID);
 			p = task[++cpucount];
 			p->processor = i;
+			callin_flag = 0;
 			prom_startcpu(linux_cpus[i].prom_node,
-				      ((unsigned long)&sparc64_cpu_startup),
-				      ((unsigned long)p));
+				      entry, ((unsigned long)p));
 			for(timeout = 0; timeout < 5000000; timeout++) {
-				if(cpu_callin_map[i])
+				if(callin_flag)
 					break;
 				udelay(100);
 			}
-			if(cpu_callin_map[i]) {
-				/* XXX fix this */
+			if(callin_flag) {
 				cpu_number_map[i] = i;
 				cpu_logical_map[i] = i;
 			} else {
@@ -184,19 +207,19 @@ void smp_boot_cpus(void)
 				printk("Processor %d is stuck.\n", i);
 			}
 		}
-		if(!(cpu_callin_map[i])) {
-			cpu_present_map &= ~(1 << i);
+		if(!callin_flag) {
+			cpu_present_map &= ~(1UL << i);
 			cpu_number_map[i] = -1;
 		}
 	}
 	if(cpucount == 0) {
 		printk("Error: only one processor found.\n");
-		cpu_present_map = (1 << smp_processor_id());
+		cpu_present_map = (1UL << smp_processor_id());
 	} else {
 		unsigned long bogosum = 0;
 
 		for(i = 0; i < NR_CPUS; i++) {
-			if(cpu_present_map & (1 << i))
+			if(cpu_present_map & (1UL << i))
 				bogosum += cpu_data[i].udelay_val;
 		}
 		printk("Total of %d processors activated (%lu.%02lu BogoMIPS).\n",
@@ -207,27 +230,39 @@ void smp_boot_cpus(void)
 		smp_num_cpus = cpucount + 1;
 	}
 	smp_processors_ready = 1;
+	membar("#StoreStore | #StoreLoad");
 }
 
-/* XXX deprecated interface... */
+/* We don't even need to do anything, the only generic message pass done
+ * anymore is to stop all cpus during a panic().  When the user drops to
+ * the PROM prompt, the firmware will send the other cpu's it's MONDO
+ * vector anyways, so doing anything special here is pointless.
+ *
+ * This whole thing should go away anyways...
+ */
 void smp_message_pass(int target, int msg, unsigned long data, int wait)
 {
-	printk("smp_message_pass() called, this is bad, spinning.\n");
-	__sti();
-	while(1)
-		barrier();
 }
 
+/* #define XCALL_DEBUG */
+
 static inline void xcall_deliver(u64 data0, u64 data1, u64 data2, u64 pstate, int cpu)
 {
-	u64 result, target = (cpu_number_map[cpu] << 14) | 0x70;
-
+	u64 result, target = (((unsigned long)linux_cpus[cpu].mid) << 14) | 0x70;
+	int stuck;
+
+#ifdef XCALL_DEBUG
+	printk("CPU[%d]: xcall(data[%016lx:%016lx:%016lx],tgt[%016lx])\n",
+	       smp_processor_id(), data0, data1, data2, target);
+#endif
+again:
 	__asm__ __volatile__("
 	wrpr	%0, %1, %%pstate
 	wr	%%g0, %2, %%asi
 	stxa	%3, [0x40] %%asi
 	stxa	%4, [0x50] %%asi
 	stxa	%5, [0x60] %%asi
+	membar	#Sync
 	stxa	%%g0, [%6] %%asi
 	membar	#Sync"
 	: /* No outputs */
@@ -235,27 +270,46 @@ static inline void xcall_deliver(u64 data0, u64 data1, u64 data2, u64 pstate, in
 	  "r" (data0), "r" (data1), "r" (data2), "r" (target));
 
 	/* NOTE: PSTATE_IE is still clear. */
+	stuck = 100000;
 	do {
 		__asm__ __volatile__("ldxa [%%g0] %1, %0"
 			: "=r" (result)
 			: "i" (ASI_INTR_DISPATCH_STAT));
+		if(result == 0) {
+			__asm__ __volatile__("wrpr %0, 0x0, %%pstate"
+					     : : "r" (pstate));
+			return;
+		}
+		stuck -= 1;
+		if(stuck == 0)
+			break;
 	} while(result & 0x1);
 	__asm__ __volatile__("wrpr %0, 0x0, %%pstate"
 			     : : "r" (pstate));
-	if(result & 0x2)
-		panic("Penguin NACK's master!");
+	if(stuck == 0) {
+#ifdef XCALL_DEBUG
+		printk("CPU[%d]: mondo stuckage result[%016lx]\n",
+		       smp_processor_id(), result);
+#endif
+	} else {
+#ifdef XCALL_DEBUG
+		printk("CPU[%d]: Penguin %d NACK's master.\n", smp_processor_id(), cpu);
+#endif
+		udelay(2);
+		goto again;
+	}
 }
 
 void smp_cross_call(unsigned long *func, u32 ctx, u64 data1, u64 data2)
 {
 	if(smp_processors_ready) {
-		unsigned long mask = (cpu_present_map & ~(1<<smp_processor_id()));
+		unsigned long mask = (cpu_present_map & ~(1UL<<smp_processor_id()));
 		u64 pstate, data0 = (((u64)ctx)<<32 | (((u64)func) & 0xffffffff));
 		int i, ncpus = smp_num_cpus;
 
 		__asm__ __volatile__("rdpr %%pstate, %0" : "=r" (pstate));
 		for(i = 0; i < ncpus; i++) {
-			if(mask & (1 << i))
+			if(mask & (1UL << i))
 				xcall_deliver(data0, data1, data2, pstate, i);
 		}
 		/* NOTE: Caller runs local copy on master. */
@@ -266,7 +320,14 @@ extern unsigned long xcall_flush_tlb_page;
 extern unsigned long xcall_flush_tlb_mm;
 extern unsigned long xcall_flush_tlb_range;
 extern unsigned long xcall_flush_tlb_all;
+extern unsigned long xcall_tlbcachesync;
 extern unsigned long xcall_flush_cache_all;
+extern unsigned long xcall_report_regs;
+
+void smp_report_regs(void)
+{
+	smp_cross_call(&xcall_report_regs, 0, 0, 0);
+}
 
 void smp_flush_cache_all(void)
 {
@@ -280,11 +341,33 @@ void smp_flush_tlb_all(void)
 	__flush_tlb_all();
 }
 
+static void smp_cross_call_avoidance(struct mm_struct *mm)
+{
+	spin_lock(&scheduler_lock);
+	get_new_mmu_context(mm, &tlb_context_cache);
+	mm->cpu_vm_mask = (1UL << smp_processor_id());
+	if(current->tss.current_ds) {
+		u32 ctx = mm->context & 0x1fff;
+
+		current->tss.ctx = ctx;
+		spitfire_set_secondary_context(ctx);
+		__asm__ __volatile__("flush %g6");
+	}
+	spin_unlock(&scheduler_lock);
+}
+
 void smp_flush_tlb_mm(struct mm_struct *mm)
 {
 	u32 ctx = mm->context & 0x1fff;
-	if(mm->cpu_vm_mask != (1UL << smp_processor_id()))
-		smp_cross_call(&xcall_flush_tlb_mm, ctx, 0, 0);
+
+	if(mm == current->mm && mm->count == 1) {
+		if(mm->cpu_vm_mask == (1UL << smp_processor_id()))
+			goto local_flush_and_out;
+		return smp_cross_call_avoidance(mm);
+	}
+	smp_cross_call(&xcall_flush_tlb_mm, ctx, 0, 0);
+
+local_flush_and_out:
 	__flush_tlb_mm(ctx);
 }
 
@@ -292,22 +375,101 @@ void smp_flush_tlb_range(struct mm_struct *mm, unsigned long start,
 			 unsigned long end)
 {
 	u32 ctx = mm->context & 0x1fff;
-	if(mm->cpu_vm_mask != (1UL << smp_processor_id()))
-		smp_cross_call(&xcall_flush_tlb_range, ctx, start, end);
+
+	if(mm == current->mm && mm->count == 1) {
+		if(mm->cpu_vm_mask == (1UL << smp_processor_id()))
+			goto local_flush_and_out;
+		return smp_cross_call_avoidance(mm);
+	}
+	smp_cross_call(&xcall_flush_tlb_range, ctx, start, end);
+
+local_flush_and_out:
 	__flush_tlb_range(ctx, start, end);
 }
 
-void smp_flush_tlb_page(struct vm_area_struct *vma, unsigned long page)
+void smp_flush_tlb_page(struct mm_struct *mm, unsigned long page)
 {
-	struct mm_struct *mm = vma->vm_mm;
 	u32 ctx = mm->context & 0x1fff;
 
-	if(mm->cpu_vm_mask != (1UL << smp_processor_id()))
-		smp_cross_call(&xcall_flush_tlb_page, ctx, page, 0);
+	if(mm == current->mm && mm->count == 1) {
+		if(mm->cpu_vm_mask == (1UL << smp_processor_id()))
+			goto local_flush_and_out;
+		return smp_cross_call_avoidance(mm);
+	}
+#if 0 /* XXX Disabled until further notice... */
+	else if(mm != current->mm && mm->count == 1) {
+		/* Try to handle two special cases to avoid cross calls
+		 * in common scenerios where we are swapping process
+		 * pages out.
+		 */
+		if((mm->context ^ tlb_context_cache) & CTX_VERSION_MASK)
+			return; /* It's dead, nothing to do. */
+		if(mm->cpu_vm_mask == (1UL << smp_processor_id()))
+			goto local_flush_and_out;
+	}
+#endif
+	smp_cross_call(&xcall_flush_tlb_page, ctx, page, 0);
+
+local_flush_and_out:
 	__flush_tlb_page(ctx, page);
 }
 
-static spinlock_t ticker_lock = SPIN_LOCK_UNLOCKED;
+/* CPU capture. */
+#define CAPTURE_DEBUG
+extern unsigned long xcall_capture;
+
+static atomic_t smp_capture_depth = ATOMIC_INIT(0);
+static atomic_t smp_capture_registry = ATOMIC_INIT(0);
+static unsigned long penguins_are_doing_time = 0;
+
+void smp_capture(void)
+{
+	int result = atomic_add_return(1, &smp_capture_depth);
+
+	membar("#StoreStore | #LoadStore");
+	if(result == 1) {
+		int ncpus = smp_num_cpus;
+
+#ifdef CAPTURE_DEBUG
+		printk("CPU[%d]: Sending penguins to jail...", smp_processor_id());
+#endif
+		penguins_are_doing_time = 1;
+		membar("#StoreStore | #LoadStore");
+		atomic_inc(&smp_capture_registry);
+		smp_cross_call(&xcall_capture, 0, 0, 0);
+		while(atomic_read(&smp_capture_registry) != ncpus)
+			membar("#LoadLoad");
+#ifdef CAPTURE_DEBUG
+		printk("done\n");
+#endif
+	}
+}
+
+void smp_release(void)
+{
+	if(atomic_dec_and_test(&smp_capture_depth)) {
+#ifdef CAPTURE_DEBUG
+		printk("CPU[%d]: Giving pardon to imprisoned penguins\n",
+		       smp_processor_id());
+#endif
+		penguins_are_doing_time = 0;
+		membar("#StoreStore | #StoreLoad");
+		atomic_dec(&smp_capture_registry);
+	}
+}
+
+/* Imprisoned penguins run with %pil == 15, but PSTATE_IE set, so they
+ * can service tlb flush xcalls...
+ */
+void smp_penguin_jailcell(void)
+{
+	flushw_user();
+	atomic_inc(&smp_capture_registry);
+	membar("#StoreLoad | #StoreStore");
+	while(penguins_are_doing_time)
+		membar("#LoadLoad");
+	atomic_dec(&smp_capture_registry);
+}
 
 static inline void sparc64_do_profile(unsigned long pc)
 {
@@ -317,59 +479,100 @@ static inline void sparc64_do_profile(unsigned long pc)
 		pc -= (unsigned long) &_stext;
 		pc >>= prof_shift;
 
-		spin_lock(&ticker_lock);
-		if(pc < prof_len)
-			prof_buffer[pc]++;
-		else
-			prof_buffer[prof_len - 1]++;
-		spin_unlock(&ticker_lock);
+		if(pc >= prof_len)
+			pc = prof_len - 1;
+		atomic_inc((atomic_t *)&prof_buffer[pc]);
 	}
 }
 
-unsigned int prof_multiplier[NR_CPUS];
-unsigned int prof_counter[NR_CPUS];
+static unsigned long real_tick_offset, current_tick_offset;
+
+#define prof_multiplier(__cpu)		cpu_data[(__cpu)].multiplier
+#define prof_counter(__cpu)		cpu_data[(__cpu)].counter
 
 extern void update_one_process(struct task_struct *p, unsigned long ticks,
 			       unsigned long user, unsigned long system);
 
 void smp_percpu_timer_interrupt(struct pt_regs *regs)
 {
+	unsigned long compare, tick;
 	int cpu = smp_processor_id();
 	int user = user_mode(regs);
 
-	/* XXX clear_profile_irq(cpu); */
-	if(!user)
-		sparc64_do_profile(regs->tpc);
-	if(!--prof_counter[cpu]) {
-		if(current->pid) {
-			update_one_process(current, 1, user, !user);
-			if(--current->counter < 0) {
-				current->counter = 0;
-				need_resched = 1;
-			}
-
-			spin_lock(&ticker_lock);
-			if(user) {
-				if(current->priority < DEF_PRIORITY)
-					kstat.cpu_nice++;
-				else
-					kstat.cpu_user++;
-			} else {
-				kstat.cpu_system++;
+	clear_softint((1UL << 0));
+	do {
+		if(!user)
+			sparc64_do_profile(regs->tpc);
+		if(!--prof_counter(cpu)) {
+			if(current->pid) {
+				unsigned int *inc_me;
+
+				update_one_process(current, 1, user, !user);
+				if(--current->counter < 0) {
+					current->counter = 0;
+					resched_force();
+				}
+
+				if(user) {
+					if(current->priority < DEF_PRIORITY)
+						inc_me = &kstat.cpu_nice;
+					else
+						inc_me = &kstat.cpu_user;
+				} else {
+					inc_me = &kstat.cpu_system;
+				}
+				atomic_inc((atomic_t *)inc_me);
 			}
-			spin_unlock(&ticker_lock);
+			prof_counter(cpu) = prof_multiplier(cpu);
 		}
-		prof_counter[cpu] = prof_multiplier[cpu];
-	}
+		__asm__ __volatile__("rd	%%tick_cmpr, %0\n\t"
+				     "add	%0, %2, %0\n\t"
+				     "wr	%0, 0x0, %%tick_cmpr\n\t"
+				     "rd	%%tick, %1"
+				     : "=&r" (compare), "=r" (tick)
+				     : "r" (current_tick_offset));
+	} while (tick >= compare);
 }
 
 static void smp_setup_percpu_timer(void)
 {
-	/* XXX implement me */
+	int cpu = smp_processor_id();
+
+	prof_counter(cpu) = prof_multiplier(cpu) = 1;
+
+	__asm__ __volatile__("rd	%%tick, %%g1\n\t"
+			     "add	%%g1, %0, %%g1\n\t"
+			     "wr	%%g1, 0x0, %%tick_cmpr"
+			     : /* no outputs */
+			     : "r" (current_tick_offset)
+			     : "g1");
+}
+
+static void smp_tickoffset_init(void)
+{
+	int node;
+
+	node = linux_cpus[0].prom_node;
+	real_tick_offset = prom_getint(node, "clock-frequency");
+	real_tick_offset = real_tick_offset / HZ;
+	current_tick_offset = real_tick_offset;
 }
 
 int setup_profiling_timer(unsigned int multiplier)
 {
-	/* XXX implement me */
+	unsigned long flags;
+	int i;
+
+	if((!multiplier) || (real_tick_offset / multiplier) < 1000)
+		return -EINVAL;
+
+	save_and_cli(flags);
+	for(i = 0; i < NR_CPUS; i++) {
+		if(cpu_present_map & (1UL << i))
+			prof_multiplier(i) = multiplier;
+	}
+	current_tick_offset = (real_tick_offset / multiplier);
+	restore_flags(flags);
+
 	return 0;
 }
diff --git a/arch/sparc64/kernel/sparc64_ksyms.c b/arch/sparc64/kernel/sparc64_ksyms.c
index 990202bac..c1ceacc3f 100644
--- a/arch/sparc64/kernel/sparc64_ksyms.c
+++ b/arch/sparc64/kernel/sparc64_ksyms.c
@@ -1,4 +1,4 @@
-/* $Id: sparc64_ksyms.c,v 1.11 1997/07/14 23:58:20 davem Exp $
+/* $Id: sparc64_ksyms.c,v 1.21 1997/09/03 12:29:07 jj Exp $
  * arch/sparc64/kernel/sparc64_ksyms.c: Sparc64 specific ksyms support.
  *
  * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
@@ -11,6 +11,7 @@
 #include <linux/module.h>
 #include <linux/types.h>
 #include <linux/string.h>
+#include <linux/in6.h>
 
 #include <asm/oplib.h>
 #include <asm/delay.h>
@@ -28,11 +29,13 @@
 #include <asm/ptrace.h>
 #include <asm/user.h>
 #include <asm/uaccess.h>
+#include <asm/checksum.h>
 #ifdef CONFIG_SBUS
 #include <asm/sbus.h>
 #include <asm/dma.h>
 #endif
 #include <asm/a.out.h>
+#include <asm/svr4.h>
 
 struct poll {
 	int fd;
@@ -45,18 +48,36 @@ extern unsigned long sunos_mmap(unsigned long, unsigned long, unsigned long,
 void _sigpause_common (unsigned int set, struct pt_regs *);
 extern void *__bzero_1page(void *);
 extern void *__bzero(void *, size_t);
+extern void *__bzero_noasi(void *, size_t);
 extern void *__memscan_zero(void *, size_t);
 extern void *__memscan_generic(void *, int, size_t);
 extern int __memcmp(const void *, const void *, __kernel_size_t);
 extern int __strncmp(const char *, const char *, __kernel_size_t);
-extern unsigned int __csum_partial_copy_sparc_generic (const char *, char *);
 extern char saved_command_line[];
-
+extern char *getname32(u32 name);
+extern void linux_sparc_syscall(void);
+extern void rtrap(void);
+extern void show_regs(struct pt_regs *);
+extern void solaris_syscall(void);
+extern void syscall_trace(void);
+extern u32 sunos_sys_table[], sys_call_table32[];
+extern void tl0_solaris(void);
+extern void sys_sigsuspend(void);
+extern int sys_getppid(void);
+extern int svr4_getcontext(svr4_ucontext_t *uc, struct pt_regs *regs);
+extern int svr4_setcontext(svr4_ucontext_t *uc, struct pt_regs *regs);
+extern int sys_ioctl(unsigned int fd, unsigned int cmd, unsigned long arg);
+extern int sys32_ioctl(unsigned int fd, unsigned int cmd, u32 arg);
+                
 extern void bcopy (const char *, char *, int);
 extern int __ashrdi3(int, int);
 
 extern void dump_thread(struct pt_regs *, struct user *);
 
+#ifdef __SMP__
+extern spinlock_t scheduler_lock;
+#endif
+
 /* One thing to note is that the way the symbols of the mul/div
  * support routines are named is a mess, they all start with
  * a '.' which makes it a bitch to export, here is the trick:
@@ -70,7 +91,17 @@ __attribute__((section("__ksymtab"))) =				\
 
 /* used by various drivers */
 #ifdef __SMP__
+EXPORT_SYMBOL(scheduler_lock);
+EXPORT_SYMBOL(global_bh_lock);
 EXPORT_SYMBOL(klock_info);
+EXPORT_SYMBOL(global_irq_holder);
+EXPORT_SYMBOL(synchronize_irq);
+EXPORT_SYMBOL(cpu_data);
+EXPORT_SYMBOL_PRIVATE(global_cli);
+EXPORT_SYMBOL_PRIVATE(global_sti);
+EXPORT_SYMBOL_PRIVATE(global_restore_flags);
+#else
+EXPORT_SYMBOL(local_irq_count);
 #endif
 EXPORT_SYMBOL_PRIVATE(_lock_kernel);
 EXPORT_SYMBOL_PRIVATE(_unlock_kernel);
@@ -81,12 +112,13 @@ EXPORT_SYMBOL(mstk48t02_regs);
 EXPORT_SYMBOL(request_fast_irq);
 EXPORT_SYMBOL(sparc_alloc_io);
 EXPORT_SYMBOL(sparc_free_io);
-EXPORT_SYMBOL(local_irq_count);
 EXPORT_SYMBOL(__sparc64_bh_counter);
 EXPORT_SYMBOL(sparc_ultra_unmapioaddr);
 EXPORT_SYMBOL(mmu_get_scsi_sgl);
 EXPORT_SYMBOL(mmu_get_scsi_one);
 EXPORT_SYMBOL(sparc_dvma_malloc);
+EXPORT_SYMBOL(mmu_release_scsi_one);
+EXPORT_SYMBOL(mmu_release_scsi_sgl);
 #if CONFIG_SBUS
 EXPORT_SYMBOL(SBus_chain);
 EXPORT_SYMBOL(dma_chain);
@@ -139,6 +171,25 @@ EXPORT_SYMBOL(strtok);
 EXPORT_SYMBOL(strstr);
 EXPORT_SYMBOL(strspn);
 
+#ifdef CONFIG_SOLARIS_EMUL_MODULE
+EXPORT_SYMBOL(getname32);
+EXPORT_SYMBOL(linux_sparc_syscall);
+EXPORT_SYMBOL(rtrap);
+EXPORT_SYMBOL(show_regs);
+EXPORT_SYMBOL(solaris_syscall);
+EXPORT_SYMBOL(syscall_trace);
+EXPORT_SYMBOL(sunos_sys_table);
+EXPORT_SYMBOL(sys_call_table32);
+EXPORT_SYMBOL(tl0_solaris);
+EXPORT_SYMBOL(sys_sigsuspend);
+EXPORT_SYMBOL(sys_getppid);
+EXPORT_SYMBOL(svr4_getcontext);
+EXPORT_SYMBOL(svr4_setcontext);
+EXPORT_SYMBOL(linux_cpus);
+EXPORT_SYMBOL(sys_ioctl);
+EXPORT_SYMBOL(sys32_ioctl);
+#endif
+
 /* Special internal versions of library functions. */
 EXPORT_SYMBOL(__memcpy);
 EXPORT_SYMBOL(__memset);
@@ -150,12 +201,13 @@ EXPORT_SYMBOL(__memcmp);
 EXPORT_SYMBOL(__strncmp);
 EXPORT_SYMBOL(__memmove);
 
-EXPORT_SYMBOL(__csum_partial_copy_sparc_generic);
+EXPORT_SYMBOL(csum_partial_copy_sparc64);
 
 /* Moving data to/from userspace. */
 EXPORT_SYMBOL(__copy_to_user);
 EXPORT_SYMBOL(__copy_from_user);
 EXPORT_SYMBOL(__strncpy_from_user);
+EXPORT_SYMBOL(__bzero_noasi);
 
 /* No version information on this, heavily used in inline asm,
  * and will always be 'void __ret_efault(void)'.
diff --git a/arch/sparc64/kernel/sys32.S b/arch/sparc64/kernel/sys32.S
index ca0faec5d..37c541755 100644
--- a/arch/sparc64/kernel/sys32.S
+++ b/arch/sparc64/kernel/sys32.S
@@ -1,4 +1,4 @@
-/* $Id: sys32.S,v 1.2 1997/07/25 01:50:47 ralf Exp $
+/* $Id: sys32.S,v 1.3 1997/08/22 20:11:47 davem Exp $
  * sys32.S: I-cache tricks for 32-bit compatability layer simple
  *          conversions.
  *
@@ -73,7 +73,7 @@ sys32_mremap:
 	.globl		sys32_readlink, sys32_unlink, sys32_rmdir, sys32_symlink
 	.globl		sys32_link, sys32_rename, sys32_truncate, sys32_ftruncate
 	.globl		sys32_chroot, sys32_chmod, sys32_chown, sys32_creat
-	.globl		sys32_mkdir, sys32_mknod, sys32_utimes, sys32_ustat
+	.globl		sys32_mkdir, sys32_mknod, sys32_ustat
 sys32_read:
 	srl		%o1, 0, %o1
 	mov		%o7, %g1
@@ -202,12 +202,6 @@ sys32_mknod:
 	srl		%o2, 16, %o2
 	call		sys_mknod
 	 mov		%g1, %o7
-sys32_utimes:
-	srl		%o0, 0, %o0
-	mov		%o7, %g1
-	srl		%o1, 0, %o1
-	call		sys_utimes
-	 mov		%g1, %o7
 sys32_ustat:
 	srl		%o1, 0, %o1
 	mov		%o7, %g1
@@ -285,24 +279,9 @@ sys32_getsockopt:
 	call		sys_setsockopt
 	 mov		%g1, %o7
 
-	.align		32
-	.globl		sys32_gettimeofday, sys32_settimeofday
-sys32_gettimeofday:
-	srl		%o0, 0, %o0
-	mov		%o7, %g1
-	srl		%o1, 0, %o1
-	call		sys_gettimeofday
-	 mov		%g1, %o7
-sys32_settimeofday:
-	srl		%o0, 0, %o0
-	mov		%o7, %g1
-	srl		%o1, 0, %o1
-	call		sys_settimeofday
-	 mov		%g1, %o7
-
 	.globl		sys32_bdflush, sys32_uselib, sys32_umount, sys32_syslog
-	.globl		sys32_personality, sys32_waitpid, sys32_getitimer
-	.globl		sys32_setitimer, sys32_sched_setscheduler
+	.globl		sys32_personality, sys32_waitpid
+	.globl		sys32_sched_setscheduler
 	.globl		sys32_sched_setparam, sys32_sched_getparam, sys32_signal
 	.globl		sys32_reboot, sys32_acct, sys32_newuname, sys32_olduname
 	.globl		sys32_sethostname, sys32_gethostname, sys32_setdomainname
@@ -338,17 +317,6 @@ sys32_waitpid:
 	mov		%o7, %g1
 	call		sys_waitpid
 	 mov		%g1, %o7
-sys32_getitimer:
-	srl		%o1, 0, %o1
-	mov		%o7, %g1
-	call		sys_getitimer
-	 mov		%g1, %o7
-sys32_setitimer:
-	srl		%o1, 0, %o1
-	mov		%o7, %g1
-	srl		%o2, 0, %o2
-	call		sys_setitimer
-	 mov		%g1, %o7
 sys32_sched_setscheduler:
 	srl		%o2, 0, %o2
 	mov		%o7, %g1
diff --git a/arch/sparc64/kernel/sys_sparc.c b/arch/sparc64/kernel/sys_sparc.c
index c827df7a1..0ec6de167 100644
--- a/arch/sparc64/kernel/sys_sparc.c
+++ b/arch/sparc64/kernel/sys_sparc.c
@@ -1,4 +1,4 @@
-/* $Id: sys_sparc.c,v 1.2 1997/07/05 09:52:34 davem Exp $
+/* $Id: sys_sparc.c,v 1.5 1997/09/03 12:29:05 jj Exp $
  * linux/arch/sparc64/kernel/sys_sparc.c
  *
  * This file contains various random system calls that
@@ -191,9 +191,12 @@ out:
 asmlinkage unsigned long
 c_sys_nis_syscall (struct pt_regs *regs)
 {
+	static int count=0;
 	lock_kernel();
-	printk ("Unimplemented SPARC system call %ld\n",regs->u_regs[1]);
-	show_regs (regs);
+	if (++count <= 20) { /* Don't make the system unusable, if someone goes stuck */
+		printk ("Unimplemented SPARC system call %ld\n",regs->u_regs[1]);
+		show_regs (regs);
+	}
 	unlock_kernel();
 	return -ENOSYS;
 }
@@ -220,46 +223,36 @@ asmlinkage int
 sparc_sigaction (int signum, const struct sigaction *action, struct sigaction *oldaction)
 {
 	struct sigaction new_sa, *p;
-	int err = -EINVAL;
 
-	lock_kernel();
 	if(signum < 0) {
 		current->tss.new_signal = 1;
 		signum = -signum;
 	}
-
 	if (signum<1 || signum>32)
-		goto out;
+		return -EINVAL;
 	p = signum - 1 + current->sig->action;
 	if (action) {
-		err = -EINVAL;
 		if (signum==SIGKILL || signum==SIGSTOP)
-			goto out;
-		err = -EFAULT;
+			return -EINVAL;
 		if(copy_from_user(&new_sa, action, sizeof(struct sigaction)))
-			goto out;	
+			return -EFAULT;
 		if (new_sa.sa_handler != SIG_DFL && new_sa.sa_handler != SIG_IGN) {
-			err = verify_area(VERIFY_READ, new_sa.sa_handler, 1);
+			int err = verify_area(VERIFY_READ, new_sa.sa_handler, 1);
 			if (err)
-				goto out;
+				return err;
 		}
 	}
-
 	if (oldaction) {
-		err = -EFAULT;
 		if (copy_to_user(oldaction, p, sizeof(struct sigaction)))
-			goto out;	
+			return -EFAULT;
 	}
-
 	if (action) {
+		spin_lock_irq(&current->sig->siglock);
 		*p = new_sa;
 		check_pending(signum);
+		spin_unlock_irq(&current->sig->siglock);
 	}
-
-	err = 0;
-out:
-	unlock_kernel();
-	return err;
+	return 0;
 }
 
 /* only AP+ systems have sys_aplib */
@@ -267,3 +260,15 @@ asmlinkage int sys_aplib(void)
 {
 	return -ENOSYS;
 }
+
+asmlinkage int solaris_syscall(struct pt_regs *regs)
+{
+	lock_kernel();
+	regs->tpc = regs->tnpc;
+	regs->tnpc += 4;
+	printk ("For Solaris binary emulation you need solaris module loaded\n");
+	show_regs (regs);
+	send_sig(SIGSEGV, current, 1);
+	unlock_kernel();
+	return 0;
+}
diff --git a/arch/sparc64/kernel/sys_sparc32.c b/arch/sparc64/kernel/sys_sparc32.c
index b6ca9448c..e2c199f60 100644
--- a/arch/sparc64/kernel/sys_sparc32.c
+++ b/arch/sparc64/kernel/sys_sparc32.c
@@ -1,4 +1,4 @@
-/* $Id: sys_sparc32.c,v 1.44 1997/07/20 09:18:47 davem Exp $
+/* $Id: sys_sparc32.c,v 1.55 1997/09/04 01:54:51 davem Exp $
  * sys_sparc32.c: Conversion between 32bit and 64bit native syscalls.
  *
  * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
@@ -44,6 +44,7 @@
 #include <asm/ipc.h>
 #include <asm/uaccess.h>
 #include <asm/fpumacro.h>
+#include <asm/semaphore.h>
 
 /* As gcc will warn about casting u32 to some ptr, we have to cast it to
  * unsigned long first, and that's what is A() for.
@@ -53,6 +54,33 @@
  */
 #define A(x) ((unsigned long)x)
  
+extern char * getname_quicklist;
+extern int getname_quickcount;
+extern struct semaphore getname_quicklock;
+extern int kerneld_msqid;
+
+/* Tuning: increase locality by reusing same pages again...
+ * if getname_quicklist becomes too long on low memory machines, either a limit
+ * should be added or after a number of cycles some pages should
+ * be released again ...
+ */
+static inline char * get_page(void)
+{
+	char * res;
+	down(&getname_quicklock);
+	res = getname_quicklist;
+	if (res) {
+		getname_quicklist = *(char**)res;
+		getname_quickcount--;
+	}
+	else
+		res = (char*)__get_free_page(GFP_KERNEL);
+	up(&getname_quicklock);
+	return res;
+}
+
+#define putname32 putname
+
 /* In order to reduce some races, while at the same time doing additional
  * checking and hopefully speeding things up, we copy filenames to the
  * kernel data space before using them..
@@ -74,44 +102,67 @@ static inline int do_getname32(u32 filename, char *page)
 	return retval;
 }
 
-/* This is a single page for faster getname.
- *   If the page is available when entering getname, use it.
- *   If the page is not available, call __get_free_page instead.
- * This works even though do_getname can block (think about it).
- * -- Michael Chastain, based on idea of Linus Torvalds, 1 Dec 1996.
- * We don't use the common getname/putname from namei.c, so that
- * this still works well, as every routine which calls getname32
- * will then call getname, then putname and then putname32.
- */
-static unsigned long name_page_cache32 = 0;
+char * getname32(u32 filename)
+{
+	char *tmp, *result;
+
+	result = ERR_PTR(-ENOMEM);
+	tmp = get_page();
+	if (tmp)  {
+		int retval = do_getname32(filename, tmp);
 
-void putname32(char * name)
+		result = tmp;
+		if (retval < 0) {
+			putname32(tmp);
+			result = ERR_PTR(retval);
+		}
+	}
+	return result;
+}
+
+/* 32-bit timeval and related flotsam.  */
+
+struct timeval32
 {
-	if (name_page_cache32 == 0)
-		name_page_cache32 = (unsigned long) name;
-	else
-		free_page((unsigned long) name);
+    int tv_sec, tv_usec;
+};
+
+struct itimerval32
+{
+    struct timeval32 it_interval;
+    struct timeval32 it_value;
+};
+
+static inline long get_tv32(struct timeval *o, struct timeval32 *i)
+{
+	return (!access_ok(VERIFY_READ, tv32, sizeof(*tv32)) ||
+		(__get_user(o->tv_sec, &i->tv_sec) |
+		 __get_user(o->tv_usec, &i->tv_usec)));
 }
 
-int getname32(u32 filename, char **result)
+static inline long put_tv32(struct timeval32 *o, struct timeval *i)
 {
-	unsigned long page;
-	int retval;
+	return (!access_ok(VERIFY_WRITE, o, sizeof(*o)) ||
+		(__put_user(i->tv_sec, &o->tv_sec) |
+		 __put_user(i->tv_usec, &o->tv_usec)));
+}
 
-	page = name_page_cache32;
-	name_page_cache32 = 0;
-	if (!page) {
-		page = __get_free_page(GFP_KERNEL);
-		if (!page)
-			return -ENOMEM;
-	}
+static inline long get_it32(struct itimerval *o, struct itimerval32 *i)
+{
+	return (!access_ok(VERIFY_READ, i32, sizeof(*i32)) ||
+		(__get_user(o->it_interval.tv_sec, &i->it_interval.tv_sec) |
+		 __get_user(o->it_interval.tv_usec, &i->it_interval.tv_usec) |
+		 __get_user(o->it_value.tv_sec, &i->it_value.tv_sec) |
+		 __get_user(o->it_value.tv_usec, &i->it_value.tv_usec)));
+}
 
-	retval = do_getname32(filename, (char *) page);
-	if (retval < 0)
-		putname32( (char *) page );
-	else
-		*result = (char *) page;
-	return retval;
+static inline long put_it32(struct itimerval32 *o, struct itimerval *i)
+{
+	return (!access_ok(VERIFY_WRITE, i32, sizeof(*i32)) ||
+		(__put_user(i->it_interval.tv_sec, &o->it_interval.tv_sec) |
+		 __put_user(i->it_interval.tv_usec, &o->it_interval.tv_usec) |
+		 __put_user(i->it_value.tv_sec, &o->it_value.tv_sec) |
+		 __put_user(i->it_value.tv_usec, &o->it_value.tv_usec)));
 }
 
 extern asmlinkage int sys_ioperm(unsigned long from, unsigned long num, int on);
@@ -266,14 +317,25 @@ asmlinkage int sys32_ipc (u32 call, int first, int second, int third, u32 ptr, u
 		switch (call) {
 		case MSGSND:
 			{
-				struct msgbuf *p = kmalloc (second + sizeof (struct msgbuf), GFP_KERNEL);
+				struct msgbuf *p = kmalloc (second + sizeof (struct msgbuf) + 4, GFP_KERNEL);
 				
 				if (!p) err = -ENOMEM;
 				else {
-					if (get_user(p->mtype, &(((struct msgbuf32 *)A(ptr))->mtype)) ||
-					    __copy_from_user(p->mtext, &(((struct msgbuf32 *)A(ptr))->mtext), second))
-						err = -EFAULT;
-					else {
+					err = 0;
+					if (first == kerneld_msqid) {
+						*(int *)p->mtext = 0;
+						if (get_user(p->mtype, &(((struct msgbuf32 *)A(ptr))->mtype)) ||
+						    __copy_from_user(&p->mtext[4], &(((struct msgbuf32 *)A(ptr))->mtext[0]), 4) ||
+						    __copy_from_user(&p->mtext[8], &(((struct msgbuf32 *)A(ptr))->mtext[4]), second-4))
+							err = -EFAULT;
+						else
+							second += 4;
+					} else {
+						if (get_user(p->mtype, &(((struct msgbuf32 *)A(ptr))->mtype)) ||
+						    __copy_from_user(p->mtext, &(((struct msgbuf32 *)A(ptr))->mtext), second))
+							err = -EFAULT;
+					}
+					if (!err) {
 						unsigned long old_fs = get_fs();
 						set_fs (KERNEL_DS);
 						err = sys_msgsnd (first, p, second, third);
@@ -287,6 +349,7 @@ asmlinkage int sys32_ipc (u32 call, int first, int second, int third, u32 ptr, u
 			{
 				struct msgbuf *p;
 				unsigned long old_fs;
+				long msgtyp = fifth;
 				
 				if (!version) {
 					struct ipc_kludge tmp;
@@ -297,20 +360,35 @@ asmlinkage int sys32_ipc (u32 call, int first, int second, int third, u32 ptr, u
 					if(copy_from_user(&tmp,(struct ipc_kludge *)A(ptr), sizeof (tmp)))
 						goto out;
 					ptr = tmp.msgp;
-					fifth = tmp.msgtyp;
+					msgtyp = tmp.msgtyp;
 				}
-				p = kmalloc (second + sizeof (struct msgbuf), GFP_KERNEL);
+
+				p = kmalloc (second + sizeof (struct msgbuf) + 4, GFP_KERNEL);
 				if (!p) {
 					err = -EFAULT;
 					goto out;
 				}
+
 				old_fs = get_fs();
 				set_fs (KERNEL_DS);
-				err = sys_msgrcv (first, p, second, fifth, third);
+				err = sys_msgrcv (first, p, second + 4, msgtyp, third);
 				set_fs (old_fs);
-				if (put_user (p->mtype, &(((struct msgbuf32 *)A(ptr))->mtype)) ||
-				    __copy_to_user(&(((struct msgbuf32 *)A(ptr))->mtext), p->mtext, second))
-					err = -EFAULT;
+
+				if (err < 0)
+					goto out;
+
+				if (first == kerneld_msqid) {
+					if (put_user (p->mtype, &(((struct msgbuf32 *)A(ptr))->mtype)) ||
+					    __copy_to_user(&(((struct msgbuf32 *)A(ptr))->mtext[0]), &p->mtext[4], 4) ||
+					    __copy_to_user(&(((struct msgbuf32 *)A(ptr))->mtext[4]), &p->mtext[8], err-8))
+						err = -EFAULT;
+					else
+						err -= 4;
+				} else {
+					if (put_user (p->mtype, &(((struct msgbuf32 *)A(ptr))->mtype)) ||
+					    __copy_to_user(&(((struct msgbuf32 *)A(ptr))->mtext), p->mtext, err))
+						err = -EFAULT;
+				}
 				kfree (p);
 				goto out;
 			}
@@ -472,8 +550,9 @@ asmlinkage int sys32_ipc (u32 call, int first, int second, int third, u32 ptr, u
 			err = -EINVAL;
 			goto out;
 		}
-	else
-		err = -EINVAL;
+
+	err = -EINVAL;
+
 out:
 	unlock_kernel();
 	return err;
@@ -565,8 +644,9 @@ asmlinkage int sys32_quotactl(int cmd, u32 special, int id, u32 addr)
 		return sys_quotactl(cmd, (const char *)A(special),
 				    id, (caddr_t)A(addr));
 	}
-	err = getname32 (special, &spec);
-	if (err) return err;
+	spec = getname32 (special);
+	err = PTR_ERR(spec);
+	if (IS_ERR(spec)) return err;
 	old_fs = get_fs ();
 	set_fs (KERNEL_DS);
 	err = sys_quotactl(cmd, (const char *)spec, id, (caddr_t)A(addr));
@@ -608,8 +688,9 @@ asmlinkage int sys32_statfs(u32 path, u32 buf)
 	unsigned long old_fs = get_fs();
 	char *pth;
 	
-	ret = getname32 (path, &pth);
-	if (!ret) {
+	pth = getname32 (path);
+	ret = PTR_ERR(pth);
+	if (!IS_ERR(pth)) {
 		set_fs (KERNEL_DS);
 		ret = sys_statfs((const char *)pth, &s);
 		set_fs (old_fs);
@@ -651,8 +732,9 @@ asmlinkage int sys32_utime(u32 filename, u32 times)
 	if (get_user (t.actime, &(((struct utimbuf32 *)A(times))->actime)) ||
 	    __get_user (t.modtime, &(((struct utimbuf32 *)A(times))->modtime)))
 		return -EFAULT;
-	ret = getname32 (filename, &filenam);
-	if (!ret) {
+	filenam = getname32 (filename);
+	ret = PTR_ERR(filenam);
+	if (!IS_ERR(filenam)) {
 		old_fs = get_fs();
 		set_fs (KERNEL_DS); 
 		ret = sys_utime(filenam, &t);
@@ -1010,7 +1092,7 @@ asmlinkage int sys32_select(int n, u32 inp, u32 outp, u32 exp, u32 tvp)
 	lock_kernel ();
 	p = (char *)__get_free_page (GFP_KERNEL);
 	if (!p)
-		goto out;
+		goto out_nofree;
 
 	q = (u32 *)p;
 	Inp = (u32 *)A(inp);
@@ -1019,7 +1101,13 @@ asmlinkage int sys32_select(int n, u32 inp, u32 outp, u32 exp, u32 tvp)
 
 	ret = -EFAULT;
 
-	nn = (n + (8 * sizeof(unsigned long)) - 1) / (8 * sizeof (unsigned long));
+	nn = (n + (8 * sizeof(long)) - 1) / (8 * sizeof(long));
+	if (inp && verify_area(VERIFY_WRITE, Inp, nn*sizeof(long)))
+		goto out;
+	if (outp && verify_area(VERIFY_WRITE, Outp, nn*sizeof(long)))
+		goto out;
+	if (exp && verify_area(VERIFY_WRITE, Exp, nn*sizeof(long)))
+		goto out;
 	for (i = 0; i < nn; i++, Inp += 2, Outp += 2, Exp += 2, q += 2) {
 		if(inp && (__get_user (q[1], Inp) || __get_user (q[0], Inp+1)))
 			goto out;
@@ -1033,7 +1121,7 @@ asmlinkage int sys32_select(int n, u32 inp, u32 outp, u32 exp, u32 tvp)
 
 	ktvp = NULL;
 	if(tvp) {
-		if(copy_from_user(&kern_tv, (struct timeval *)A(tvp), sizeof(*ktvp)))
+		if (get_tv32(&kern_tv, (struct timeval32 *)A(tvp)))
 			goto out;
 		ktvp = &kern_tv;
 	}
@@ -1048,8 +1136,12 @@ asmlinkage int sys32_select(int n, u32 inp, u32 outp, u32 exp, u32 tvp)
 			 ktvp);
 	set_fs (old_fs);
 
-	if(tvp && !(current->personality & STICKY_TIMEOUTS))
-		copy_to_user((struct timeval *)A(tvp), &kern_tv, sizeof(*ktvp));
+	if(tvp && !(current->personality & STICKY_TIMEOUTS)) {
+		if (put_tv32((struct timeval32 *)A(tvp), &kern_tv)) {
+			ret = -EFAULT;
+			goto out;
+		}
+	}
 
 	q = (u32 *)p;
 	Inp = (u32 *)A(inp);
@@ -1079,6 +1171,7 @@ asmlinkage int sys32_select(int n, u32 inp, u32 outp, u32 exp, u32 tvp)
 	}
 out:
 	free_page ((unsigned long)p);
+out_nofree:
 	unlock_kernel();
 	return ret;
 }
@@ -1111,8 +1204,9 @@ asmlinkage int sys32_newstat(u32 filename, u32 statbuf)
 	char *filenam;
 	unsigned long old_fs = get_fs();
 	
-	ret = getname32 (filename, &filenam);
-	if (!ret) {
+	filenam = getname32 (filename);
+	ret = PTR_ERR(filenam);
+	if (!IS_ERR(filenam)) {
 		set_fs (KERNEL_DS);
 		ret = sys_newstat(filenam, &s);
 		set_fs (old_fs);
@@ -1132,8 +1226,9 @@ asmlinkage int sys32_newlstat(u32 filename, u32 statbuf)
 	char *filenam;
 	unsigned long old_fs = get_fs();
 	
-	ret = getname32 (filename, &filenam);
-	if (!ret) {
+	filenam = getname32 (filename);
+	ret = PTR_ERR(filenam);
+	if (!IS_ERR(filenam)) {
 		set_fs (KERNEL_DS);
 		ret = sys_newlstat(filenam, &s);
 		set_fs (old_fs);
@@ -1239,12 +1334,12 @@ static void *do_smb_super_data_conv(void *raw_data)
 	struct smb_mount_data *s = (struct smb_mount_data *)raw_data;
 	struct smb_mount_data32 *s32 = (struct smb_mount_data32 *)raw_data;
 
-	s->dir_mode = s32->dir_mode;
-	s->file_mode = s32->file_mode;
-	s->gid = s32->gid;
-	s->uid = s32->uid;
-	memmove (&s->addr, &s32->addr, (((long)&s->uid) - ((long)&s->addr)));
+	s->version = s32->version;
 	s->mounted_uid = s32->mounted_uid;
+	s->uid = s32->uid;
+	s->gid = s32->gid;
+	s->file_mode = s32->file_mode;
+	s->dir_mode = s32->dir_mode;
 	return raw_data;
 }
 
@@ -1344,8 +1439,8 @@ asmlinkage int sys32_mount(u32 dev_name, u32 dir_name, u32 type, u32 new_flags,
 }
 
 struct rusage32 {
-        struct timeval ru_utime;
-        struct timeval ru_stime;
+        struct timeval32 ru_utime;
+        struct timeval32 ru_stime;
         s32    ru_maxrss;
         s32    ru_ixrss;
         s32    ru_idrss;
@@ -1703,7 +1798,7 @@ struct timex32 {
 	s32 constant;
 	s32 precision;
 	s32 tolerance;
-	struct timeval time;
+	struct timeval32 time;
 	s32 tick;
 	s32 ppsfreq;
 	s32 jitter;
@@ -2110,54 +2205,45 @@ asmlinkage int sparc32_sigaction (int signum, u32 action, u32 oldaction)
 {
 	struct sigaction32 new_sa, old_sa;
 	struct sigaction *p;
-	int err = -EINVAL;
 
-	lock_kernel();
 	if(signum < 0) {
 		current->tss.new_signal = 1;
 		signum = -signum;
 	}
-
 	if (signum<1 || signum>32)
-		goto out;
+		return -EINVAL;
 	p = signum - 1 + current->sig->action;
 	if (action) {
-		err = -EINVAL;
 		if (signum==SIGKILL || signum==SIGSTOP)
-			goto out;
-		err = -EFAULT;
+			return -EINVAL;
 		if(copy_from_user(&new_sa, A(action), sizeof(struct sigaction32)))
-			goto out;
+			return -EFAULT;
 		if (((__sighandler_t)A(new_sa.sa_handler)) != SIG_DFL && 
 		    ((__sighandler_t)A(new_sa.sa_handler)) != SIG_IGN) {
-			err = verify_area(VERIFY_READ, (__sighandler_t)A(new_sa.sa_handler), 1);
+			int err = verify_area(VERIFY_READ,
+					      (__sighandler_t)A(new_sa.sa_handler), 1);
 			if (err)
-				goto out;
+				return err;
 		}
 	}
-
 	if (oldaction) {
-		err = -EFAULT;
 		old_sa.sa_handler = (unsigned)(u64)(p->sa_handler);
 		old_sa.sa_mask = (sigset_t32)(p->sa_mask);
 		old_sa.sa_flags = (unsigned)(p->sa_flags);
 		old_sa.sa_restorer = (unsigned)(u64)(p->sa_restorer);
 		if (copy_to_user(A(oldaction), &old_sa, sizeof(struct sigaction32)))
-			goto out;	
+			return -EFAULT;
 	}
-
 	if (action) {
+		spin_lock_irq(&current->sig->siglock);
 		p->sa_handler = (__sighandler_t)A(new_sa.sa_handler);
 		p->sa_mask = (sigset_t)(new_sa.sa_mask);
 		p->sa_flags = new_sa.sa_flags;
 		p->sa_restorer = (void (*)(void))A(new_sa.sa_restorer);
 		check_pending(signum);
+		spin_unlock_irq(&current->sig->siglock);
 	}
-
-	err = 0;
-out:
-	unlock_kernel();
-	return err;
+	return 0;
 }
 
 /*
@@ -2298,7 +2384,7 @@ asmlinkage int sparc32_execve(struct pt_regs *regs)
                 base = 1;
 
 	lock_kernel();
-        filename = getname((char *)(unsigned long)(u32)regs->u_regs[base + UREG_I0]);
+        filename = getname((char *)A((u32)regs->u_regs[base + UREG_I0]));
 	error = PTR_ERR(filename);
         if(IS_ERR(filename))
                 goto out;
@@ -2310,6 +2396,7 @@ asmlinkage int sparc32_execve(struct pt_regs *regs)
 	if(!error) {
 		fprs_write(0);
 		regs->fprs = 0;
+		regs->tstate &= ~TSTATE_PEF;
 	}
 out:
 	unlock_kernel();
@@ -2369,8 +2456,12 @@ asmlinkage int sys32_query_module(u32 name_user, int which, u32 buf, __kernel_si
 	case QM_MODULES:
 	case QM_REFS:
 	case QM_DEPS:
-		if (name_user && (ret = getname32 (name_user, &usernam)))
-			return ret;
+		if (name_user) {
+			usernam = getname32 (name_user);
+			ret = PTR_ERR(usernam);
+			if (IS_ERR(usernam))
+				return ret;
+		}
 		buff = kmalloc (bufsiz, GFP_KERNEL);
 		if (!buff) {
 			if (name_user) putname32 (usernam);
@@ -2431,8 +2522,12 @@ qmsym_toshort:
 		if (name_user) putname32 (usernam);
 		return ret;
 	case QM_INFO:
-		if (name_user && (ret = getname32 (name_user, &usernam)))
-			return ret;
+		if (name_user) {
+			usernam = getname32 (name_user);
+			ret = PTR_ERR(usernam);
+			if (IS_ERR(usernam))
+				return ret;
+		}
 		set_fs (KERNEL_DS);
 		ret = sys_query_module (usernam, which, (char *)&mi, sizeof (mi), &val);
 		set_fs (old_fs);
@@ -2720,9 +2815,7 @@ static int nfs_getfh32_res_trans(union nfsctl_res *kres, union nfsctl_res32 *res
 	return 0;
 }
 
-extern asmlinkage int sys_nfsservctl(int cmd,
-				     struct nfsctl_arg *arg,
-				     union nfsctl_res *resp);
+extern asmlinkage int sys_nfsservctl(int cmd, void *arg, void *resp);
 
 int asmlinkage sys32_nfsservctl(int cmd, u32 u_argp, u32 u_resp)
 {
@@ -2793,3 +2886,106 @@ done:
 		kfree(kres);
 	return err;
 }
+
+/* Translations due to time_t size differences.  Which affects all
+   sorts of things, like timeval and itimerval.  */
+
+extern struct timezone sys_tz;
+extern int do_sys_settimeofday(struct timeval *tv, struct timezone *tz);
+
+asmlinkage int sys32_gettimeofday(u32 tv, u32 tz)
+{
+	if (tv) {
+		struct timeval ktv;
+		do_gettimeofday(&ktv);
+		if (put_tv32((struct timeval32 *)A(tv), &ktv))
+			return -EFAULT;
+	}
+	if (tz) {
+		if (copy_to_user((void*)A(tz), &sys_tz, sizeof(sys_tz)))
+			return -EFAULT;
+	}
+	return 0;
+}
+
+asmlinkage int sys32_settimeofday(u32 tv, u32 tz)
+{
+	struct timeval ktv;
+	struct timezone ktz;
+
+ 	if (tv) {
+		if (get_tv32(&ktv, (struct timeval32 *)A(tv)))
+			return -EFAULT;
+	}
+	if (tz) {
+		if (copy_from_user(&ktz, (void*)A(tz), sizeof(ktz)))
+			return -EFAULT;
+	}
+
+	return do_sys_settimeofday(tv ? &ktv : NULL, tz ? &ktz : NULL);
+}
+
+extern int do_getitimer(int which, struct itimerval *value);
+
+asmlinkage int sys32_getitimer(int which, u32 it)
+{
+	struct itimerval kit;
+	int error;
+
+	error = do_getitimer(which, &kit);
+	if (!error && put_it32((struct itimerval32 *)A(it), &kit))
+		error = -EFAULT;
+
+	return error;
+}
+
+extern int do_setitimer(int which, struct itimerval *, struct itimerval *);
+
+asmlinkage int sys32_setitimer(int which, u32 in, u32 out)
+{
+	struct itimerval kin, kout;
+	int error;
+
+	if (in) {
+		if (get_it32(&kin, (struct itimerval32 *)A(in)))
+			return -EFAULT;
+	} else
+		memset(&kin, 0, sizeof(kin));
+
+	error = do_setitimer(which, &kin, out ? &kout : NULL);
+	if (error || !out)
+		return error;
+	if (put_it32((struct itimerval32 *)A(out), &kout))
+		return -EFAULT;
+
+	return 0;
+
+}
+
+asmlinkage int sys_utimes(char *, struct timeval *);
+
+asmlinkage int sys32_utimes(u32 filename, u32 tvs)
+{
+	char *kfilename;
+	struct timeval ktvs[2];
+	unsigned long old_fs;
+	int ret;
+
+	kfilename = getname32(filename);
+	ret = PTR_ERR(kfilename);
+	if (!IS_ERR(kfilename)) {
+		if (tvs) {
+			if (get_tv32(&ktvs[0], (struct timeval32 *)A(tvs)) ||
+			    get_tv32(&ktvs[1], 1+(struct timeval32 *)A(tvs)))
+				return -EFAULT;
+		}
+
+		old_fs = get_fs();
+		set_fs(KERNEL_DS);
+		ret = sys_utimes(kfilename, &ktvs[0]);
+		set_fs(old_fs);
+
+		putname32(kfilename);
+	}
+	return ret;
+}
diff --git a/arch/sparc64/kernel/systbls.S b/arch/sparc64/kernel/systbls.S
index eda0ff326..9b00175a3 100644
--- a/arch/sparc64/kernel/systbls.S
+++ b/arch/sparc64/kernel/systbls.S
@@ -1,4 +1,4 @@
-/* $Id: systbls.S,v 1.21 1997/07/05 07:09:17 davem Exp $
+/* $Id: systbls.S,v 1.24 1997/08/22 20:12:06 davem Exp $
  * systbls.S: System call entry point tables for OS compatibility.
  *            The native Linux system call table lives here also.
  *
@@ -10,213 +10,214 @@
  * Copyright (C) 1995 Adrian M. Rodriguez (adrian@remus.rutgers.edu)
  */
 
-	.data
-	.align	8
+	.text
+	.align	1024
 
 	/* First, the 32-bit Linux native syscall table. */
 
 	.globl sys_call_table32
 sys_call_table32:
-/*0*/	.xword sys_setup, sys_exit, sys_fork, sys32_read, sys32_write
-/*5*/	.xword sys32_open, sys_close, sys32_wait4, sys32_creat, sys32_link
-/*10*/  .xword sys32_unlink, sunos_execv, sys32_chdir, sys_nis_syscall, sys32_mknod
-/*15*/	.xword sys32_chmod, sys32_chown, sparc32_brk, sys_nis_syscall, sys32_lseek
-/*20*/	.xword sys_getpid, sys_nis_syscall, sys_nis_syscall, sys_setuid, sys_getuid
-/*25*/	.xword sys32_time, sys_ptrace, sys_alarm, sys_nis_syscall, sys_pause
-/*30*/	.xword sys32_utime, sys_stty, sys_gtty, sys32_access, sys_nice
-	.xword sys_ftime, sys_sync, sys_kill, sys32_newstat, sys_nis_syscall
-/*40*/	.xword sys32_newlstat, sys_dup, sys_pipe, sys32_times, sys_profil
-	.xword sys_nis_syscall, sys_setgid, sys_getgid, sys32_signal, sys_geteuid
-/*50*/	.xword sys_getegid, sys32_acct, sys_nis_syscall, sys_nis_syscall, sys32_ioctl
-	.xword sys32_reboot, sys_nis_syscall, sys32_symlink, sys32_readlink, sys32_execve
-/*60*/	.xword sys_umask, sys32_chroot, sys32_newfstat, sys_nis_syscall, sys_getpagesize
-	.xword sys32_msync, sys_vfork, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-/*70*/	.xword sys_nis_syscall, sys32_mmap, sys_nis_syscall, sys32_munmap, sys32_mprotect
-	.xword sys_nis_syscall, sys_vhangup, sys_nis_syscall, sys_nis_syscall, sys32_getgroups
-/*80*/	.xword sys32_setgroups, sys_getpgrp, sys_nis_syscall, sys32_setitimer, sys_nis_syscall
-	.xword sys32_swapon, sys32_getitimer, sys_nis_syscall, sys32_sethostname, sys_nis_syscall
-/*90*/	.xword sys_dup2, sys_nis_syscall, sys32_fcntl, sys32_select, sys_nis_syscall
-	.xword sys_fsync, sys_setpriority, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-/*100*/	.xword sys_getpriority, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-/*110*/	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-	.xword sys_nis_syscall, sys32_gettimeofday, sys32_getrusage, sys_nis_syscall, sys_nis_syscall
-/*120*/	.xword sys32_readv, sys32_writev, sys32_settimeofday, sys_fchown, sys_fchmod
-	.xword sys_nis_syscall, sys32_setreuid, sys_setregid, sys32_rename, sys32_truncate
-/*130*/	.xword sys32_ftruncate, sys_flock, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-	.xword sys_nis_syscall, sys32_mkdir, sys32_rmdir, sys_nis_syscall, sys_nis_syscall
-/*140*/	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys32_getrlimit
-	.xword sys32_setrlimit, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-/*150*/	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-	.xword sys_nis_syscall, sys_nis_syscall, sys32_statfs, sys32_fstatfs, sys32_umount
-/*160*/	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys32_setdomainname, sys_nis_syscall
-	.xword sys32_quotactl, sys_nis_syscall, sys32_mount, sys32_ustat, sys_nis_syscall
-/*170*/	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys32_getdents
-	.xword sys_setsid, sys_fchdir, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-/*180*/	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys32_sigpending, sys32_query_module
-	.xword sys_setpgid, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys32_newuname
-/*190*/	.xword sys32_init_module, sys32_personality, sys_prof, sys_break, sys_lock
-	.xword sys_mpx, sys_ulimit, sys_getppid, sparc32_sigaction, sys_sgetmask
-/*200*/	.xword sys_ssetmask, sys_sigsuspend, sys32_newlstat, sys32_uselib, old32_readdir
-	.xword sys_nis_syscall, sys32_socketcall, sys32_syslog, sys32_olduname, sys_nis_syscall
-/*210*/	.xword sys_idle, sys_nis_syscall, sys32_waitpid, sys32_swapoff, sys32_sysinfo
-	.xword sys32_ipc, sys_sigreturn, sys_clone, sys_nis_syscall, sys32_adjtimex
-/*220*/	.xword sys32_sigprocmask, sys32_create_module, sys32_delete_module, sys32_get_kernel_syms, sys_getpgid
-	.xword sys32_bdflush, sys32_sysfs, sys_nis_syscall, sys_setfsuid, sys_setfsgid
-/*230*/	.xword sys32_llseek, sys32_time, sys_nis_syscall, sys_stime, sys_nis_syscall
-	.xword sys_nis_syscall, sys32_llseek, sys32_mlock, sys32_munlock, sys_mlockall
-/*240*/	.xword sys_munlockall, sys32_sched_setparam, sys32_sched_getparam, sys_nis_syscall, sys_nis_syscall
-	.xword sys_nis_syscall, sys_sched_get_priority_max, sys_sched_get_priority_min, sys32_sched_rr_get_interval, sys32_nanosleep
-/*250*/	.xword sys32_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys32_nfsservctl
-	.xword sys_aplib, sys_nis_syscall
+/*0*/	.word sys_setup, sys_exit, sys_fork, sys32_read, sys32_write
+/*5*/	.word sys32_open, sys_close, sys32_wait4, sys32_creat, sys32_link
+/*10*/  .word sys32_unlink, sunos_execv, sys32_chdir, sys_nis_syscall, sys32_mknod
+/*15*/	.word sys32_chmod, sys32_chown, sparc32_brk, sys_nis_syscall, sys32_lseek
+/*20*/	.word sys_getpid, sys_nis_syscall, sys_nis_syscall, sys_setuid, sys_getuid
+/*25*/	.word sys32_time, sys_ptrace, sys_alarm, sys_nis_syscall, sys_pause
+/*30*/	.word sys32_utime, sys_stty, sys_gtty, sys32_access, sys_nice
+	.word sys_ftime, sys_sync, sys_kill, sys32_newstat, sys_nis_syscall
+/*40*/	.word sys32_newlstat, sys_dup, sys_pipe, sys32_times, sys_profil
+	.word sys_nis_syscall, sys_setgid, sys_getgid, sys32_signal, sys_geteuid
+/*50*/	.word sys_getegid, sys32_acct, sys_nis_syscall, sys_nis_syscall, sys32_ioctl
+	.word sys32_reboot, sys_nis_syscall, sys32_symlink, sys32_readlink, sys32_execve
+/*60*/	.word sys_umask, sys32_chroot, sys32_newfstat, sys_nis_syscall, sys_getpagesize
+	.word sys32_msync, sys_vfork, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+/*70*/	.word sys_nis_syscall, sys32_mmap, sys_nis_syscall, sys32_munmap, sys32_mprotect
+	.word sys_nis_syscall, sys_vhangup, sys_nis_syscall, sys_nis_syscall, sys32_getgroups
+/*80*/	.word sys32_setgroups, sys_getpgrp, sys_nis_syscall, sys32_setitimer, sys_nis_syscall
+	.word sys32_swapon, sys32_getitimer, sys_nis_syscall, sys32_sethostname, sys_nis_syscall
+/*90*/	.word sys_dup2, sys_nis_syscall, sys32_fcntl, sys32_select, sys_nis_syscall
+	.word sys_fsync, sys_setpriority, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+/*100*/	.word sys_getpriority, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+/*110*/	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+	.word sys_nis_syscall, sys32_gettimeofday, sys32_getrusage, sys_nis_syscall, sys_nis_syscall
+/*120*/	.word sys32_readv, sys32_writev, sys32_settimeofday, sys_fchown, sys_fchmod
+	.word sys_nis_syscall, sys32_setreuid, sys_setregid, sys32_rename, sys32_truncate
+/*130*/	.word sys32_ftruncate, sys_flock, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+	.word sys_nis_syscall, sys32_mkdir, sys32_rmdir, sys_nis_syscall, sys_nis_syscall
+/*140*/	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys32_getrlimit
+	.word sys32_setrlimit, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+/*150*/	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+	.word sys_nis_syscall, sys_nis_syscall, sys32_statfs, sys32_fstatfs, sys32_umount
+/*160*/	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys32_setdomainname, sys_nis_syscall
+	.word sys32_quotactl, sys_nis_syscall, sys32_mount, sys32_ustat, sys_nis_syscall
+/*170*/	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys32_getdents
+	.word sys_setsid, sys_fchdir, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+/*180*/	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys32_sigpending, sys32_query_module
+	.word sys_setpgid, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys32_newuname
+/*190*/	.word sys32_init_module, sys32_personality, sys_prof, sys_break, sys_lock
+	.word sys_mpx, sys_ulimit, sys_getppid, sparc32_sigaction, sys_sgetmask
+/*200*/	.word sys_ssetmask, sys_sigsuspend, sys32_newlstat, sys32_uselib, old32_readdir
+	.word sys_nis_syscall, sys32_socketcall, sys32_syslog, sys32_olduname, sys_nis_syscall
+/*210*/	.word sys_idle, sys_nis_syscall, sys32_waitpid, sys32_swapoff, sys32_sysinfo
+	.word sys32_ipc, sys_sigreturn, sys_clone, sys_nis_syscall, sys32_adjtimex
+/*220*/	.word sys32_sigprocmask, sys32_create_module, sys32_delete_module, sys32_get_kernel_syms, sys_getpgid
+	.word sys32_bdflush, sys32_sysfs, sys_nis_syscall, sys_setfsuid, sys_setfsgid
+/*230*/	.word sys32_llseek, sys32_time, sys_nis_syscall, sys_stime, sys_nis_syscall
+	.word sys_nis_syscall, sys32_llseek, sys32_mlock, sys32_munlock, sys_mlockall
+/*240*/	.word sys_munlockall, sys32_sched_setparam, sys32_sched_getparam, sys_nis_syscall, sys_nis_syscall
+	.word sys_nis_syscall, sys_sched_get_priority_max, sys_sched_get_priority_min, sys32_sched_rr_get_interval, sys32_nanosleep
+/*250*/	.word sys32_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys32_nfsservctl
+	.word sys_aplib
 
 	/* Now the 64-bit native Linux syscall table. */
 
+	.align	1024
 	.globl sys_call_table64, sys_call_table
 sys_call_table64:
 sys_call_table:
-/*0*/	.xword sys_setup, sys_exit, sys_fork, sys_read, sys_write
-/*5*/	.xword sys_open, sys_close, sys_wait4, sys_creat, sys_link
-/*10*/  .xword sys_unlink, sunos_execv, sys_chdir, sys_nis_syscall, sys_mknod
-/*15*/	.xword sys_chmod, sys_chown, sparc_brk, sys_nis_syscall, sys_lseek
-/*20*/	.xword sys_getpid, sys_nis_syscall, sys_nis_syscall, sys_setuid, sys_getuid
-/*25*/	.xword sys_time, sys_ptrace, sys_alarm, sys_nis_syscall, sys_pause
-/*30*/	.xword sys_utime, sys_stty, sys_gtty, sys_access, sys_nice
-	.xword sys_ftime, sys_sync, sys_kill, sys_newstat, sys_nis_syscall
-/*40*/	.xword sys_newlstat, sys_dup, sys_pipe, sys_times, sys_profil
-	.xword sys_nis_syscall, sys_setgid, sys_getgid, sys_signal, sys_geteuid
-/*50*/	.xword sys_getegid, sys_acct, sys_nis_syscall, sys_nis_syscall, sys_ioctl
-	.xword sys_reboot, sys_nis_syscall, sys_symlink, sys_readlink, sys_execve
-/*60*/	.xword sys_umask, sys_chroot, sys_newfstat, sys_nis_syscall, sys_getpagesize
-	.xword sys_nis_syscall, sys_vfork, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-/*70*/	.xword sys_nis_syscall, sys_mmap, sys_nis_syscall, sys_munmap, sys_mprotect
-	.xword sys_nis_syscall, sys_vhangup, sys_nis_syscall, sys_nis_syscall, sys_getgroups
-/*80*/	.xword sys_setgroups, sys_getpgrp, sys_nis_syscall, sys_setitimer, sys_nis_syscall
-	.xword sys_swapon, sys_getitimer, sys_nis_syscall, sys_sethostname, sys_nis_syscall
-/*90*/	.xword sys_dup2, sys_nis_syscall, sys_fcntl, sys_select, sys_nis_syscall
-	.xword sys_fsync, sys_setpriority, sys_socket, sys_connect, sys_accept
-/*100*/	.xword sys_getpriority, sys_send, sys_recv, sys_nis_syscall, sys_bind
-	.xword sys_setsockopt, sys_listen, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-/*110*/	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_recvmsg, sys_sendmsg
-	.xword sys_nis_syscall, sys_gettimeofday, sys_getrusage, sys_getsockopt, sys_nis_syscall
-/*120*/	.xword sys_readv, sys_writev, sys_settimeofday, sys_fchown, sys_fchmod
-	.xword sys_recvfrom, sys_setreuid, sys_setregid, sys_rename, sys_truncate
-/*130*/	.xword sys_ftruncate, sys_flock, sys_nis_syscall, sys_sendto, sys_shutdown
-	.xword sys_socketpair, sys_mkdir, sys_rmdir, sys_nis_syscall, sys_nis_syscall
-/*140*/	.xword sys_nis_syscall, sys_getpeername, sys_nis_syscall, sys_nis_syscall, sys_getrlimit
-	.xword sys_setrlimit, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-/*150*/	.xword sys_getsockname, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-	.xword sys_nis_syscall, sys_nis_syscall, sys_statfs, sys_fstatfs, sys_umount
-/*160*/	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_setdomainname, sys_nis_syscall
-	.xword sys_quotactl, sys_nis_syscall, sys_mount, sys_ustat, sys_nis_syscall
-/*170*/	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_getdents
-	.xword sys_setsid, sys_fchdir, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
-/*180*/	.xword sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_sigpending, sys_query_module
-	.xword sys_setpgid, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_newuname
-/*190*/	.xword sys_init_module, sys_personality, sys_prof, sys_break, sys_lock
-	.xword sys_mpx, sys_ulimit, sys_getppid, sparc_sigaction, sys_sgetmask
-/*200*/	.xword sys_ssetmask, sys_sigsuspend, sys_newlstat, sys_uselib, sys_nis_syscall
-	.xword sys_nis_syscall, sys_nis_syscall, sys_syslog, sys_nis_syscall, sys_nis_syscall
-/*210*/	.xword sys_idle, sys_nis_syscall, sys_waitpid, sys_swapoff, sys_sysinfo
-	.xword sys_ipc, sys_sigreturn, sys_clone, sys_nis_syscall, sys_adjtimex
-/*220*/	.xword sys_sigprocmask, sys_create_module, sys_delete_module, sys_get_kernel_syms, sys_getpgid
-	.xword sys_bdflush, sys_sysfs, sys_nis_syscall, sys_setfsuid, sys_setfsgid
-/*230*/	.xword sys_llseek, sys_time, sys_nis_syscall, sys_stime, sys_nis_syscall
-	.xword sys_nis_syscall, sys_llseek, sys_mlock, sys_munlock, sys_mlockall
-/*240*/	.xword sys_munlockall, sys_sched_setparam, sys_sched_getparam, sys_nis_syscall, sys_nis_syscall
-	.xword sys_nis_syscall, sys_sched_get_priority_max, sys_sched_get_priority_min, sys_sched_rr_get_interval, sys_nanosleep
-/*250*/	.xword sys_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys_nfsservctl
-	.xword sys_aplib, sys_nis_syscall
+/*0*/	.word sys_setup, sys_exit, sys_fork, sys_read, sys_write
+/*5*/	.word sys_open, sys_close, sys_wait4, sys_creat, sys_link
+/*10*/  .word sys_unlink, sunos_execv, sys_chdir, sys_nis_syscall, sys_mknod
+/*15*/	.word sys_chmod, sys_chown, sparc_brk, sys_nis_syscall, sys_lseek
+/*20*/	.word sys_getpid, sys_nis_syscall, sys_nis_syscall, sys_setuid, sys_getuid
+/*25*/	.word sys_time, sys_ptrace, sys_alarm, sys_nis_syscall, sys_pause
+/*30*/	.word sys_utime, sys_stty, sys_gtty, sys_access, sys_nice
+	.word sys_ftime, sys_sync, sys_kill, sys_newstat, sys_nis_syscall
+/*40*/	.word sys_newlstat, sys_dup, sys_pipe, sys_times, sys_profil
+	.word sys_nis_syscall, sys_setgid, sys_getgid, sys_signal, sys_geteuid
+/*50*/	.word sys_getegid, sys_acct, sys_nis_syscall, sys_nis_syscall, sys_ioctl
+	.word sys_reboot, sys_nis_syscall, sys_symlink, sys_readlink, sys_execve
+/*60*/	.word sys_umask, sys_chroot, sys_newfstat, sys_nis_syscall, sys_getpagesize
+	.word sys_nis_syscall, sys_vfork, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+/*70*/	.word sys_nis_syscall, sys_mmap, sys_nis_syscall, sys_munmap, sys_mprotect
+	.word sys_nis_syscall, sys_vhangup, sys_nis_syscall, sys_nis_syscall, sys_getgroups
+/*80*/	.word sys_setgroups, sys_getpgrp, sys_nis_syscall, sys_setitimer, sys_nis_syscall
+	.word sys_swapon, sys_getitimer, sys_nis_syscall, sys_sethostname, sys_nis_syscall
+/*90*/	.word sys_dup2, sys_nis_syscall, sys_fcntl, sys_select, sys_nis_syscall
+	.word sys_fsync, sys_setpriority, sys_socket, sys_connect, sys_accept
+/*100*/	.word sys_getpriority, sys_send, sys_recv, sys_nis_syscall, sys_bind
+	.word sys_setsockopt, sys_listen, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+/*110*/	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_recvmsg, sys_sendmsg
+	.word sys_nis_syscall, sys_gettimeofday, sys_getrusage, sys_getsockopt, sys_nis_syscall
+/*120*/	.word sys_readv, sys_writev, sys_settimeofday, sys_fchown, sys_fchmod
+	.word sys_recvfrom, sys_setreuid, sys_setregid, sys_rename, sys_truncate
+/*130*/	.word sys_ftruncate, sys_flock, sys_nis_syscall, sys_sendto, sys_shutdown
+	.word sys_socketpair, sys_mkdir, sys_rmdir, sys_nis_syscall, sys_nis_syscall
+/*140*/	.word sys_nis_syscall, sys_getpeername, sys_nis_syscall, sys_nis_syscall, sys_getrlimit
+	.word sys_setrlimit, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+/*150*/	.word sys_getsockname, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+	.word sys_nis_syscall, sys_nis_syscall, sys_statfs, sys_fstatfs, sys_umount
+/*160*/	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_setdomainname, sys_nis_syscall
+	.word sys_quotactl, sys_nis_syscall, sys_mount, sys_ustat, sys_nis_syscall
+/*170*/	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_getdents
+	.word sys_setsid, sys_fchdir, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall
+/*180*/	.word sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_sigpending, sys_query_module
+	.word sys_setpgid, sys_nis_syscall, sys_nis_syscall, sys_nis_syscall, sys_newuname
+/*190*/	.word sys_init_module, sys_personality, sys_prof, sys_break, sys_lock
+	.word sys_mpx, sys_ulimit, sys_getppid, sparc_sigaction, sys_sgetmask
+/*200*/	.word sys_ssetmask, sys_sigsuspend, sys_newlstat, sys_uselib, sys_nis_syscall
+	.word sys_nis_syscall, sys_nis_syscall, sys_syslog, sys_nis_syscall, sys_nis_syscall
+/*210*/	.word sys_idle, sys_nis_syscall, sys_waitpid, sys_swapoff, sys_sysinfo
+	.word sys_ipc, sys_sigreturn, sys_clone, sys_nis_syscall, sys_adjtimex
+/*220*/	.word sys_sigprocmask, sys_create_module, sys_delete_module, sys_get_kernel_syms, sys_getpgid
+	.word sys_bdflush, sys_sysfs, sys_nis_syscall, sys_setfsuid, sys_setfsgid
+/*230*/	.word sys_llseek, sys_time, sys_nis_syscall, sys_stime, sys_nis_syscall
+	.word sys_nis_syscall, sys_llseek, sys_mlock, sys_munlock, sys_mlockall
+/*240*/	.word sys_munlockall, sys_sched_setparam, sys_sched_getparam, sys_nis_syscall, sys_nis_syscall
+	.word sys_nis_syscall, sys_sched_get_priority_max, sys_sched_get_priority_min, sys_sched_rr_get_interval, sys_nanosleep
+/*250*/	.word sys_mremap, sys_sysctl, sys_getsid, sys_fdatasync, sys_nfsservctl
+	.word sys_aplib
 
 	/* Now the 32-bit SunOS syscall table. */
 
-	.align 8
+	.align 1024
 	.globl sunos_sys_table
 sunos_sys_table:
-/*0*/	.xword sunos_indir, sys_exit, sys_fork
-	.xword sunos_read, sunos_write, sunos_open
-	.xword sys_close, sunos_wait4, sys32_creat
-	.xword sys32_link, sys32_unlink, sunos_execv
-	.xword sys32_chdir, sunos_nosys, sys32_mknod
-	.xword sys32_chmod, sys32_chown, sunos_brk
-	.xword sunos_nosys, sys32_lseek, sunos_getpid
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_getuid, sunos_nosys, sys_ptrace
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sys32_access, sunos_nosys, sunos_nosys
-	.xword sys_sync, sys_kill, sys32_newstat
-	.xword sunos_nosys, sys32_newlstat, sys_dup
-	.xword sys_pipe, sunos_nosys, sys_profil
-	.xword sunos_nosys, sunos_nosys, sunos_getgid
-	.xword sunos_nosys, sunos_nosys
-/*50*/	.xword sunos_nosys, sys32_acct, sunos_nosys
-	.xword sunos_mctl, sunos_ioctl, sys32_reboot
-	.xword sunos_nosys, sys32_symlink, sys32_readlink
-	.xword sys32_execve, sys_umask, sys32_chroot
-	.xword sys32_newfstat, sunos_nosys, sys_getpagesize
-	.xword sys32_msync, sys_vfork, sunos_nosys
-	.xword sunos_nosys, sunos_sbrk, sunos_sstk
-	.xword sunos_mmap, sunos_vadvise, sys32_munmap
-	.xword sys32_mprotect, sunos_madvise, sys_vhangup
-	.xword sunos_nosys, sunos_mincore, sys32_getgroups
-	.xword sys32_setgroups, sys_getpgrp, sunos_setpgrp
-	.xword sys32_setitimer, sunos_nosys, sys32_swapon
-	.xword sys32_getitimer, sys32_gethostname, sys32_sethostname
-	.xword sunos_getdtablesize, sys_dup2, sunos_nop
-	.xword sys32_fcntl, sunos_select, sunos_nop
-	.xword sys_fsync, sys_setpriority, sys_socket
-	.xword sys32_connect, sunos_accept
-/*100*/	.xword sys_getpriority, sunos_send, sunos_recv
-	.xword sunos_nosys, sys32_bind, sunos_setsockopt
-	.xword sys_listen, sunos_nosys, sunos_sigaction
-	.xword sunos_sigblock, sunos_sigsetmask, sys_sigpause
-	.xword sys32_sigstack, sys32_recvmsg, sys32_sendmsg
-	.xword sunos_nosys, sys_gettimeofday, sys32_getrusage
-	.xword sunos_getsockopt, sunos_nosys, sunos_readv
-	.xword sunos_writev, sys_settimeofday, sys_fchown
-	.xword sys_fchmod, sys32_recvfrom, sys32_setreuid
-	.xword sys_setregid, sys32_rename, sys32_truncate
-	.xword sys32_ftruncate, sys_flock, sunos_nosys
-	.xword sys32_sendto, sys_shutdown, sys_socketpair
-	.xword sys32_mkdir, sys32_rmdir, sys32_utimes
-	.xword sys_sigreturn, sunos_nosys, sys32_getpeername
-	.xword sunos_gethostid, sunos_nosys, sys32_getrlimit
-	.xword sys32_setrlimit, sunos_killpg, sunos_nosys
-	.xword sunos_nosys, sunos_nosys
-/*150*/	.xword sys32_getsockname, sunos_nosys, sunos_nosys
-	.xword sys32_poll, sunos_nosys, sunos_nosys
-	.xword sunos_getdirentries, sys32_statfs, sys32_fstatfs
-	.xword sys32_umount, sunos_nosys, sunos_nosys
-	.xword sunos_getdomainname, sys32_setdomainname
-	.xword sunos_nosys, sys32_quotactl, sunos_nosys
-	.xword sunos_mount, sys32_ustat, sunos_semsys
-	.xword sunos_nosys, sunos_shmsys, sunos_audit
-	.xword sunos_nosys, sunos_getdents, sys_setsid
-	.xword sys_fchdir, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sys32_sigpending, sunos_nosys
-	.xword sys_setpgid, sunos_pathconf, sunos_fpathconf
-	.xword sunos_sysconf, sunos_uname, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-/*200*/	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys
-/*250*/	.xword sunos_nosys, sunos_nosys, sunos_nosys
-	.xword sunos_nosys, sunos_nosys, sys_aplib
+/*0*/	.word sunos_indir, sys_exit, sys_fork
+	.word sunos_read, sunos_write, sunos_open
+	.word sys_close, sunos_wait4, sys32_creat
+	.word sys32_link, sys32_unlink, sunos_execv
+	.word sys32_chdir, sunos_nosys, sys32_mknod
+	.word sys32_chmod, sys32_chown, sunos_brk
+	.word sunos_nosys, sys32_lseek, sunos_getpid
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_getuid, sunos_nosys, sys_ptrace
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sys32_access, sunos_nosys, sunos_nosys
+	.word sys_sync, sys_kill, sys32_newstat
+	.word sunos_nosys, sys32_newlstat, sys_dup
+	.word sys_pipe, sunos_nosys, sys_profil
+	.word sunos_nosys, sunos_nosys, sunos_getgid
+	.word sunos_nosys, sunos_nosys
+/*50*/	.word sunos_nosys, sys32_acct, sunos_nosys
+	.word sunos_mctl, sunos_ioctl, sys32_reboot
+	.word sunos_nosys, sys32_symlink, sys32_readlink
+	.word sys32_execve, sys_umask, sys32_chroot
+	.word sys32_newfstat, sunos_nosys, sys_getpagesize
+	.word sys32_msync, sys_vfork, sunos_nosys
+	.word sunos_nosys, sunos_sbrk, sunos_sstk
+	.word sunos_mmap, sunos_vadvise, sys32_munmap
+	.word sys32_mprotect, sunos_madvise, sys_vhangup
+	.word sunos_nosys, sunos_mincore, sys32_getgroups
+	.word sys32_setgroups, sys_getpgrp, sunos_setpgrp
+	.word sys32_setitimer, sunos_nosys, sys32_swapon
+	.word sys32_getitimer, sys32_gethostname, sys32_sethostname
+	.word sunos_getdtablesize, sys_dup2, sunos_nop
+	.word sys32_fcntl, sunos_select, sunos_nop
+	.word sys_fsync, sys_setpriority, sys_socket
+	.word sys32_connect, sunos_accept
+/*100*/	.word sys_getpriority, sunos_send, sunos_recv
+	.word sunos_nosys, sys32_bind, sunos_setsockopt
+	.word sys_listen, sunos_nosys, sunos_sigaction
+	.word sunos_sigblock, sunos_sigsetmask, sys_sigpause
+	.word sys32_sigstack, sys32_recvmsg, sys32_sendmsg
+	.word sunos_nosys, sys32_gettimeofday, sys32_getrusage
+	.word sunos_getsockopt, sunos_nosys, sunos_readv
+	.word sunos_writev, sys32_settimeofday, sys_fchown
+	.word sys_fchmod, sys32_recvfrom, sys32_setreuid
+	.word sys_setregid, sys32_rename, sys32_truncate
+	.word sys32_ftruncate, sys_flock, sunos_nosys
+	.word sys32_sendto, sys_shutdown, sys_socketpair
+	.word sys32_mkdir, sys32_rmdir, sys32_utimes
+	.word sys_sigreturn, sunos_nosys, sys32_getpeername
+	.word sunos_gethostid, sunos_nosys, sys32_getrlimit
+	.word sys32_setrlimit, sunos_killpg, sunos_nosys
+	.word sunos_nosys, sunos_nosys
+/*150*/	.word sys32_getsockname, sunos_nosys, sunos_nosys
+	.word sys32_poll, sunos_nosys, sunos_nosys
+	.word sunos_getdirentries, sys32_statfs, sys32_fstatfs
+	.word sys32_umount, sunos_nosys, sunos_nosys
+	.word sunos_getdomainname, sys32_setdomainname
+	.word sunos_nosys, sys32_quotactl, sunos_nosys
+	.word sunos_mount, sys32_ustat, sunos_semsys
+	.word sunos_nosys, sunos_shmsys, sunos_audit
+	.word sunos_nosys, sunos_getdents, sys_setsid
+	.word sys_fchdir, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sys32_sigpending, sunos_nosys
+	.word sys_setpgid, sunos_pathconf, sunos_fpathconf
+	.word sunos_sysconf, sunos_uname, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+/*200*/	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys
+/*250*/	.word sunos_nosys, sunos_nosys, sunos_nosys
+	.word sunos_nosys, sunos_nosys, sys_aplib
diff --git a/arch/sparc64/kernel/time.c b/arch/sparc64/kernel/time.c
index df6a1c05e..8b0152231 100644
--- a/arch/sparc64/kernel/time.c
+++ b/arch/sparc64/kernel/time.c
@@ -1,4 +1,4 @@
-/* $Id: time.c,v 1.5 1997/07/23 11:32:06 davem Exp $
+/* $Id: time.c,v 1.12 1997/08/22 20:12:13 davem Exp $
  * time.c: UltraSparc timer and TOD clock support.
  *
  * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
@@ -8,6 +8,7 @@
  * Copyright (C) 1996 Thomas K. Dyas (tdyas@eden.rutgers.edu)
  */
 
+#include <linux/config.h>
 #include <linux/errno.h>
 #include <linux/sched.h>
 #include <linux/kernel.h>
@@ -17,12 +18,17 @@
 #include <linux/interrupt.h>
 #include <linux/timex.h>
 #include <linux/init.h>
+#include <linux/ioport.h>
 
 #include <asm/oplib.h>
 #include <asm/mostek.h>
 #include <asm/timer.h>
 #include <asm/irq.h>
 #include <asm/io.h>
+#include <asm/sbus.h>
+#include <asm/fhc.h>
+#include <asm/pbm.h>
+#include <asm/ebus.h>
 
 struct mostek48t02 *mstk48t02_regs = 0;
 struct mostek48t08 *mstk48t08_regs = 0;
@@ -155,33 +161,55 @@ static int has_low_battery(void)
 
 
 /* Probe for the real time clock chip. */
-__initfunc(static void clock_probe(void))
+__initfunc(static void set_system_time(void))
 {
-	struct linux_prom_registers clk_reg[2];
-	char model[128];
-	int node, sbusnd, err;
-
-	/* XXX HACK HACK HACK, delete me soon */
-	struct linux_prom_ranges XXX_sbus_ranges[PROMREG_MAX];
-	int XXX_sbus_nranges;
+	unsigned int year, mon, day, hour, min, sec;
+	struct mostek48t02 *mregs;
 
-	node = prom_getchild(prom_root_node);
-	sbusnd = prom_searchsiblings(node, "sbus");
-	node = prom_getchild(sbusnd);
+	do_get_fast_time = do_gettimeofday;
 
-	if(node == 0 || node == -1) {
-		prom_printf("clock_probe: Serious problem can't find sbus PROM node.\n");
+	mregs = mstk48t02_regs;
+	if(!mregs) {
+		prom_printf("Something wrong, clock regs not mapped yet.\n");
 		prom_halt();
+	}		
+
+	mregs->creg |= MSTK_CREG_READ;
+	sec = MSTK_REG_SEC(mregs);
+	min = MSTK_REG_MIN(mregs);
+	hour = MSTK_REG_HOUR(mregs);
+	day = MSTK_REG_DOM(mregs);
+	mon = MSTK_REG_MONTH(mregs);
+	year = MSTK_CVT_YEAR( MSTK_REG_YEAR(mregs) );
+	xtime.tv_sec = mktime(year, mon, day, hour, min, sec);
+	xtime.tv_usec = 0;
+	mregs->creg &= ~MSTK_CREG_READ;
+}
+
+__initfunc(void clock_probe(void))
+{
+	struct linux_prom_registers clk_reg[2];
+	char model[128];
+	int node, busnd = -1, err;
+
+	if(central_bus != NULL) {
+		busnd = central_bus->child->prom_node;
+	}
+#ifdef CONFIG_PCI
+	else if (ebus_chain != NULL) {
+		busnd = ebus_chain->prom_node;
+	}
+#endif
+	else {
+		busnd = SBus_chain->prom_node;
 	}
 
-	/* XXX FIX ME */
-	err = prom_getproperty(sbusnd, "ranges", (char *) XXX_sbus_ranges,
-			       sizeof(XXX_sbus_ranges));
-	if(err == -1) {
-		prom_printf("clock_probe: Cannot get XXX sbus ranges\n");
+	if(busnd == -1) {
+		prom_printf("clock_probe: problem, cannot find bus to search.\n");
 		prom_halt();
 	}
-	XXX_sbus_nranges = (err / sizeof(struct linux_prom_ranges));
+
+	node = prom_getchild(busnd);
 
 	while(1) {
 		prom_getstring(node, "model", model, sizeof(model));
@@ -199,12 +227,48 @@ __initfunc(static void clock_probe(void))
 		err = prom_getproperty(node, "reg", (char *)clk_reg,
 				       sizeof(clk_reg));
 		if(err == -1) {
-			prom_printf("clock_probe: Cannot make Mostek\n");
+			prom_printf("clock_probe: Cannot get Mostek reg property\n");
 			prom_halt();
 		}
 
-		/* XXX fix me badly */
-		prom_adjust_regs(clk_reg, 1, XXX_sbus_ranges, XXX_sbus_nranges);
+		if(central_bus) {
+			prom_apply_fhc_ranges(central_bus->child, clk_reg, 1);
+			prom_apply_central_ranges(central_bus, clk_reg, 1);
+		}
+#ifdef CONFIG_PCI
+		else if (ebus_chain) {
+			struct linux_ebus_device *edev;
+
+			for_each_ebusdev(edev, ebus_chain)
+				if (edev->prom_node == node)
+					break;
+			if (!edev) {
+				prom_printf("%s: Mostek not probed by EBUS\n",
+					    __FUNCTION__);
+				prom_halt();
+			}
+
+			if (check_region(edev->base_address[0],
+					 sizeof(struct mostek48t59))) {
+				prom_printf("%s: Can't get region %lx, %d\n",
+					    __FUNCTION__, edev->base_address[0],
+					    sizeof(struct mostek48t59));
+				prom_halt();
+			}
+			request_region(edev->base_address[0],
+				       sizeof(struct mostek48t59), "clock");
+
+			mstk48t59_regs = (struct mostek48t59 *)
+						edev->base_address[0];
+			mstk48t02_regs = &mstk48t59_regs->regs;
+			break;
+		}
+#endif
+		else {
+			prom_adjust_regs(clk_reg, 1,
+					 SBus_chain->sbus_ranges,
+					 SBus_chain->num_sbus_ranges);
+		}
 
 		if(model[5] == '0' && model[6] == '2') {
 			mstk48t02_regs = (struct mostek48t02 *)
@@ -234,6 +298,8 @@ __initfunc(static void clock_probe(void))
 	/* Kick start the clock if it is completely stopped. */
 	if (mstk48t02_regs->sec & MSTK_STOP)
 		kick_start_clock();
+
+	set_system_time();
 }
 
 #ifndef BCD_TO_BIN
@@ -246,29 +312,10 @@ __initfunc(static void clock_probe(void))
 
 __initfunc(void time_init(void))
 {
-	unsigned int year, mon, day, hour, min, sec;
-	struct mostek48t02 *mregs;
-
-	do_get_fast_time = do_gettimeofday;
-
-	clock_probe();
-
-	mregs = mstk48t02_regs;
-	if(!mregs) {
-		prom_printf("Something wrong, clock regs not mapped yet.\n");
-		prom_halt();
-	}		
-
-	mregs->creg |= MSTK_CREG_READ;
-	sec = MSTK_REG_SEC(mregs);
-	min = MSTK_REG_MIN(mregs);
-	hour = MSTK_REG_HOUR(mregs);
-	day = MSTK_REG_DOM(mregs);
-	mon = MSTK_REG_MONTH(mregs);
-	year = MSTK_CVT_YEAR( MSTK_REG_YEAR(mregs) );
-	xtime.tv_sec = mktime(year, mon, day, hour, min, sec);
-	xtime.tv_usec = 0;
-	mregs->creg &= ~MSTK_CREG_READ;
+	/* clock_probe() is now done at end of sbus_init on sparc64
+	 * so that both sbus and fhc bus information is probed and
+	 * available.
+	 */
 }
 
 extern void init_timers(void (*func)(int, void *, struct pt_regs *));
@@ -301,22 +348,29 @@ void do_gettimeofday(struct timeval *tv)
 	/* Load doubles must be used on xtime so that what we get
 	 * is guarenteed to be atomic, this is why we can run this
 	 * with interrupts on full blast.  Don't touch this... -DaveM
+	 *
+	 * Note with time_t changes to the timeval type, I must now use
+	 * nucleus atomic quad 128-bit loads.
 	 */
 	__asm__ __volatile__("
 	sethi	%hi(linux_timers), %o1
 	sethi	%hi(xtime), %g2
 	ldx	[%o1 + %lo(linux_timers)], %g3
-1:	ldd	[%g2 + %lo(xtime)], %o4
+	or	%g2, %lo(xtime), %g2
+1:	ldda	[%g2] 0x24, %o4
+	membar	#LoadLoad | #MemIssue
 	ldx	[%g3], %o1
-	ldd	[%g2 + %lo(xtime)], %o2
+	membar	#LoadLoad | #MemIssue
+	ldda	[%g2] 0x24, %o2
+	membar	#LoadLoad
 	xor	%o4, %o2, %o2
 	xor	%o5, %o3, %o3
 	orcc	%o2, %o3, %g0
-	bne,pn	%icc, 1b
+	bne,pn	%xcc, 1b
 	 cmp	%o1, 0
 	bge,pt	%icc, 1f
 	 sethi	%hi(tick), %o3
-	ld	[%o3 + %lo(tick)], %o3
+	ldx	[%o3 + %lo(tick)], %o3
 	sethi	%hi(0x1fffff), %o2
 	or	%o2, %lo(0x1fffff), %o2
 	add	%o5, %o3, %o5
@@ -325,12 +379,12 @@ void do_gettimeofday(struct timeval *tv)
 	sethi	%hi(1000000), %o2
 	or	%o2, %lo(1000000), %o2
 	cmp	%o5, %o2
-	bl,a,pn	%icc, 1f
-	 st	%o4, [%o0 + 0x0]
+	bl,a,pn	%xcc, 1f
+	 stx	%o4, [%o0 + 0x0]
 	add	%o4, 0x1, %o4
 	sub	%o5, %o2, %o5
-	st	%o4, [%o0 + 0x0]
-1:	st	%o5, [%o0 + 0x4]");
+	stx	%o4, [%o0 + 0x0]
+1:	stx	%o5, [%o0 + 0x8]");
 }
 
 void do_settimeofday(struct timeval *tv)
diff --git a/arch/sparc64/kernel/trampoline.S b/arch/sparc64/kernel/trampoline.S
new file mode 100644
index 000000000..cefcb6ba3
--- /dev/null
+++ b/arch/sparc64/kernel/trampoline.S
@@ -0,0 +1,207 @@
+/* $Id: trampoline.S,v 1.2 1997/08/30 04:53:05 ralf Exp $
+ * trampoline.S: Jump start slave processors on sparc64.
+ *
+ * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
+ */
+
+#include <asm/head.h>
+#include <asm/asi.h>
+#include <asm/lsu.h>
+#include <asm/pstate.h>
+#include <asm/page.h>
+#include <asm/pgtable.h>
+#include <asm/spitfire.h>
+#include <asm/asm_offsets.h>
+
+	.data
+	.align		8
+	.globl		smp_trampoline
+smp_trampoline:		.skip	0x300
+
+	.text
+	.align		8
+	.globl		sparc64_cpu_startup, sparc64_cpu_startup_end
+sparc64_cpu_startup:
+	flushw
+	mov	(LSU_CONTROL_IC | LSU_CONTROL_DC | LSU_CONTROL_IM | LSU_CONTROL_DM), %g1
+	stxa	%g1, [%g0] ASI_LSU_CONTROL
+	membar	#Sync
+	wrpr	%g0, (PSTATE_PRIV | PSTATE_PEF | PSTATE_IE), %pstate
+	wrpr	%g0, 15, %pil
+
+	sethi	%uhi(PAGE_OFFSET), %g4
+	sllx	%g4, 32, %g4
+
+	/* XXX Buggy PROM... */
+	srl	%o0, 0, %g6
+	add	%g6, %g4, %g6
+
+	sethi	%uhi(_PAGE_VALID | _PAGE_SZ4MB), %g5
+	sllx	%g5, 32, %g5
+	or	%g5, (_PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W | _PAGE_G), %g5
+
+	sethi	%uhi(_PAGE_PADDR), %g3
+	or	%g3, %ulo(_PAGE_PADDR), %g3
+	sllx	%g3, 32, %g3
+	sethi	%hi(_PAGE_PADDR), %g7
+	or	%g7, %lo(_PAGE_PADDR), %g7
+	or	%g3, %g7, %g3
+
+	clr	%l0
+	set	0x1fff, %l2
+	rd	%pc, %l3
+	andn	%l3, %l2, %g2
+1:	ldxa	[%l0] ASI_ITLB_TAG_READ, %g1
+	nop
+	nop
+	nop
+	andn	%g1, %l2, %g1
+	cmp	%g1, %g2
+	be,a,pn	%xcc, 2f
+	 ldxa	[%l0] ASI_ITLB_DATA_ACCESS, %g1
+	cmp	%l0, (63 << 3)
+	blu,pt	%xcc, 1b
+	 add	%l0, (1 << 3), %l0
+
+2:	nop
+	nop
+	nop
+	and	%g1, %g3, %g1
+	sub	%g1, %g2, %g1
+	or	%g5, %g1, %g5
+	clr	%l0
+	sethi	%hi(KERNBASE), %g3
+	sethi	%hi(KERNBASE<<1), %g7
+	mov	TLB_TAG_ACCESS, %l7
+1:	ldxa	[%l0] ASI_ITLB_TAG_READ, %g1
+	nop
+	nop
+	nop
+	andn	%g1, %l2, %g1
+	cmp	%g1, %g3
+	blu,pn	%xcc, 2f
+	 cmp	%g1, %g7
+	bgeu,pn	%xcc, 2f
+	 nop
+	stxa	%g0, [%l7] ASI_IMMU
+	stxa	%g0, [%l0] ASI_ITLB_DATA_ACCESS
+2:	cmp	%l0, (63 << 3)
+	blu,pt	%xcc, 1b
+	 add	%l0, (1 << 3), %l0
+
+	nop
+	nop
+	nop
+	clr	%l0
+1:	ldxa	[%l0] ASI_DTLB_TAG_READ, %g1
+	nop
+	nop
+	nop
+	andn	%g1, %l2, %g1
+	cmp	%g1, %g3
+	blu,pn	%xcc, 2f
+	 cmp	%g1, %g7
+	bgeu,pn	%xcc, 2f
+	 nop
+	stxa	%g0, [%l7] ASI_DMMU
+	stxa	%g0, [%l0] ASI_DTLB_DATA_ACCESS
+2:	cmp	%l0, (63 << 3)
+	blu,pt	%xcc, 1b
+	 add	%l0, (1 << 3), %l0
+
+	nop
+	nop
+	nop
+	sethi	%hi(KERNBASE), %g3
+	mov	(63 << 3), %g7
+	stxa	%g3, [%l7] ASI_DMMU
+	stxa	%g5, [%g7] ASI_DTLB_DATA_ACCESS
+	membar	#Sync
+	stxa	%g3, [%l7] ASI_IMMU
+	stxa	%g5, [%g7] ASI_ITLB_DATA_ACCESS
+	membar	#Sync
+	flush	%g3
+	membar	#Sync
+	b,pt	%xcc, 1f
+	 nop
+1:	set	bounce, %g2
+	jmpl	%g2 + %g0, %g0
+	 nop
+
+bounce:
+	mov	PRIMARY_CONTEXT, %g7
+	stxa	%g0, [%g7] ASI_DMMU
+	membar	#Sync
+	mov	SECONDARY_CONTEXT, %g7
+	stxa	%g0, [%g7] ASI_DMMU
+	membar	#Sync
+
+	mov	TLB_TAG_ACCESS, %g2
+	stxa	%g3, [%g2] ASI_IMMU
+	stxa	%g3, [%g2] ASI_DMMU
+
+	mov	(63 << 3), %g7
+	ldxa	[%g7] ASI_ITLB_DATA_ACCESS, %g1
+	andn	%g1, (_PAGE_G), %g1
+	stxa	%g1, [%g7] ASI_ITLB_DATA_ACCESS
+	membar	#Sync
+
+	ldxa	[%g7] ASI_DTLB_DATA_ACCESS, %g1
+	andn	%g1, (_PAGE_G), %g1
+	stxa	%g1, [%g7] ASI_DTLB_DATA_ACCESS
+	membar	#Sync
+
+	flush	%g3
+	membar	#Sync
+
+	mov	1, %g5
+	sllx	%g5, (PAGE_SHIFT + 1), %g5
+	sub	%g5, (REGWIN_SZ + STACK_BIAS), %g5
+	add	%g6, %g5, %sp
+	mov	0, %fp
+
+	wrpr	%g0, 0, %wstate
+	wrpr	%g0, 0, %tl
+
+	/* Setup the trap globals, then we can resurface. */
+	rdpr	%pstate, %o1
+	mov	%g6, %o2
+	wrpr	%o1, (PSTATE_AG | PSTATE_IE), %pstate
+	sethi	%hi(sparc64_ttable_tl0), %g5
+	wrpr	%g5, %tba
+	mov	%o2, %g6
+
+	wrpr	%o1, (PSTATE_MG | PSTATE_IE), %pstate
+	sethi	%hi(0x1ff8), %g2
+	or	%g2, %lo(0x1ff8), %g2
+	ldx	[%o2 + AOFF_task_mm], %g6
+	ldx	[%g6 + AOFF_mm_pgd], %g6
+	clr	%g7
+
+	wrpr	%o1, (PSTATE_IG | PSTATE_IE), %pstate
+	sethi	%hi(ivector_to_mask), %g5
+	or	%g5, %lo(ivector_to_mask), %g1
+	mov	0x40, %g2
+
+	wrpr	%g0, 0, %wstate
+	wrpr	%o1, PSTATE_IE, %pstate
+
+	mov	TSB_REG, %o4
+	mov	1, %o5
+	stxa	%o5, [%o4] ASI_DMMU
+	stxa	%o5, [%o4] ASI_IMMU
+	membar	#Sync
+
+	or	%o1, PSTATE_IE, %o1
+	wrpr	%o1, 0, %pstate
+
+	call	smp_callin
+	 nop
+	call	cpu_idle
+	 mov	0, %o0
+	call	cpu_panic
+	 nop
+1:	b,a,pt	%xcc, 1b
+
+	.align		8
+sparc64_cpu_startup_end:
diff --git a/arch/sparc64/kernel/traps.c b/arch/sparc64/kernel/traps.c
index ac3e79958..1ffd43730 100644
--- a/arch/sparc64/kernel/traps.c
+++ b/arch/sparc64/kernel/traps.c
@@ -1,4 +1,4 @@
-/* $Id: traps.c,v 1.29 1997/07/05 09:52:38 davem Exp $
+/* $Id: traps.c,v 1.31 1997/08/11 14:35:33 davem Exp $
  * arch/sparc64/kernel/traps.c
  *
  * Copyright (C) 1995,1997 David S. Miller (davem@caip.rutgers.edu)
@@ -25,6 +25,7 @@
 #include <asm/unistd.h>
 #include <asm/uaccess.h>
 #include <asm/fpumacro.h>
+#include <asm/lsu.h>
 
 /* #define SYSCALL_TRACING */
 /* #define VERBOSE_SYSCALL_TRACING */
@@ -194,8 +195,55 @@ void data_access_exception (struct pt_regs *regs)
 	send_sig(SIGSEGV, current, 1);
 }
 
+#ifdef CONFIG_PCI
+/* This is really pathetic... */
+/* #define DEBUG_PCI_POKES */
+extern volatile int pci_poke_in_progress;
+extern volatile int pci_poke_faulted;
+#endif
+
 void do_dae(struct pt_regs *regs)
 {
+#ifdef CONFIG_PCI
+#ifdef DEBUG_PCI_POKES
+	prom_printf(" (POKE ");
+#endif
+	if(pci_poke_in_progress) {
+		unsigned long va;
+#ifdef DEBUG_PCI_POKES
+		prom_printf("tpc[%016lx] tnpc[%016lx] ",
+			    regs->tpc, regs->tnpc);
+#endif
+		pci_poke_faulted = 1;
+		regs->tnpc = regs->tpc + 4;
+
+
+#ifdef DEBUG_PCI_POKES
+		prom_printf("PCI) ");
+		/* prom_halt(); */
+#endif
+		/* Re-enable I/D caches, Ultra turned them off. */
+		for(va =  0; va < (PAGE_SIZE << 1); va += 32) {
+			spitfire_put_icache_tag(va, 0x0);
+			spitfire_put_dcache_tag(va, 0x0);
+		}
+		__asm__ __volatile__("flush %%g6\n\t"
+				     "membar #Sync\n\t"
+				     "stxa %0, [%%g0] %1\n\t"
+				     "membar #Sync"
+				     : /* no outputs */
+				     : "r" (LSU_CONTROL_IC | LSU_CONTROL_DC |
+					    LSU_CONTROL_IM | LSU_CONTROL_DM),
+				       "i" (ASI_LSU_CONTROL)
+				     : "memory");
+		return;
+	}
+#ifdef DEBUG_PCI_POKES
+	prom_printf("USER) ");
+	prom_printf("tpc[%016lx] tnpc[%016lx]\n");
+	prom_halt();
+#endif
+#endif
 	send_sig(SIGSEGV, current, 1);
 }
 
@@ -215,11 +263,9 @@ void do_fpe_common(struct pt_regs *regs)
 		regs->tpc = regs->tnpc;
 		regs->tnpc += 4;
 	} else {
-		lock_kernel();
 		current->tss.sig_address = regs->tpc;
 		current->tss.sig_desc = SUBSIG_FPERROR;
 		send_sig(SIGFPE, current, 1);
-		unlock_kernel();
 	}
 }
 
@@ -288,6 +334,7 @@ void die_if_kernel(char *str, struct pt_regs *regs)
 	}
 	printk("Instruction DUMP:");
 	instruction_dump ((unsigned int *) regs->tpc);
+	lock_kernel(); /* Or else! */
 	if(regs->tstate & TSTATE_PRIV)
 		do_exit(SIGKILL);
 	do_exit(SIGSEGV);
@@ -298,13 +345,11 @@ void do_illegal_instruction(struct pt_regs *regs)
 	unsigned long pc = regs->tpc;
 	unsigned long tstate = regs->tstate;
 
-	lock_kernel();
 	if(tstate & TSTATE_PRIV)
 		die_if_kernel("Kernel illegal instruction", regs);
 	current->tss.sig_address = pc;
 	current->tss.sig_desc = SUBSIG_ILLINST;
 	send_sig(SIGILL, current, 1);
-	unlock_kernel();
 }
 
 void mem_address_unaligned(struct pt_regs *regs)
@@ -333,19 +378,16 @@ void do_privact(struct pt_regs *regs)
 	current->tss.sig_address = regs->tpc;
 	current->tss.sig_desc = SUBSIG_PRIVINST;
 	send_sig(SIGILL, current, 1);
-	unlock_kernel();
 }
 
 void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 			 unsigned long tstate)
 {
-	lock_kernel();
 	if(tstate & TSTATE_PRIV)
 		die_if_kernel("Penguin instruction from Penguin mode??!?!", regs);
 	current->tss.sig_address = pc;
 	current->tss.sig_desc = SUBSIG_PRIVINST;
 	send_sig(SIGILL, current, 1);
-	unlock_kernel();
 }
 
 /* XXX User may want to be allowed to do this. XXX */
@@ -353,7 +395,6 @@ void do_priv_instruction(struct pt_regs *regs, unsigned long pc, unsigned long n
 void do_memaccess_unaligned(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 			    unsigned long tstate)
 {
-	lock_kernel();
 	if(regs->tstate & TSTATE_PRIV) {
 		printk("KERNEL MNA at pc %016lx npc %016lx called by %016lx\n", pc, npc,
 		       regs->u_regs[UREG_RETPC]);
@@ -363,15 +404,12 @@ void do_memaccess_unaligned(struct pt_regs *regs, unsigned long pc, unsigned lon
 	current->tss.sig_address = pc;
 	current->tss.sig_desc = SUBSIG_PRIVINST;
 	send_sig(SIGBUS, current, 1);
-	unlock_kernel();
 }
 
 void handle_hw_divzero(struct pt_regs *regs, unsigned long pc, unsigned long npc,
 		       unsigned long psr)
 {
-	lock_kernel();
 	send_sig(SIGILL, current, 1);
-	unlock_kernel();
 }
 
 /* Trap level 1 stuff or other traps we should never see... */
diff --git a/arch/sparc64/kernel/ttable.S b/arch/sparc64/kernel/ttable.S
index 73bda96d9..bf00cb231 100644
--- a/arch/sparc64/kernel/ttable.S
+++ b/arch/sparc64/kernel/ttable.S
@@ -1,4 +1,4 @@
-/* $Id: ttable.S,v 1.18 1997/07/05 09:52:41 davem Exp $
+/* $Id: ttable.S,v 1.20 1997/08/29 15:51:39 jj Exp $
  * ttable.S: Sparc V9 Trap Table(s) with SpitFire extensions.
  *
  * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
@@ -44,7 +44,8 @@ tl0_irq5:	TRAP_IRQ(handler_irq, 5)  TRAP_IRQ(handler_irq, 6)
 tl0_irq7:	TRAP_IRQ(handler_irq, 7)  TRAP_IRQ(handler_irq, 8)
 tl0_irq9:	TRAP_IRQ(handler_irq, 9)  TRAP_IRQ(handler_irq, 10)
 tl0_irq11:	TRAP_IRQ(handler_irq, 11) TRAP_IRQ(handler_irq, 12)
-tl0_irq13:	TRAP_IRQ(handler_irq, 13) TRAP_IRQ(handler_irq, 14)
+tl0_irq13:	TRAP_IRQ(handler_irq, 13)
+tl0_itick:	TRAP_TICK
 tl0_irq15:	TRAP_IRQ(handler_irq, 15)
 tl0_resv050:	BTRAP(0x50) BTRAP(0x51) BTRAP(0x52) BTRAP(0x53) BTRAP(0x54) BTRAP(0x55)
 tl0_resv056:	BTRAP(0x56) BTRAP(0x57) BTRAP(0x58) BTRAP(0x59) BTRAP(0x5a) BTRAP(0x5b)
@@ -99,6 +100,7 @@ tl0_bkpt:	BREAKPOINT_TRAP
 tl0_resv102:	BTRAP(0x102)
 tl0_flushw:	FLUSH_WINDOW_TRAP
 tl0_resv104:	BTRAP(0x104) BTRAP(0x105) BTRAP(0x106) BTRAP(0x107)
+		.globl tl0_solaris
 tl0_solaris:	SOLARIS_SYSCALL_TRAP
 tl0_netbsd:	NETBSD_SYSCALL_TRAP
 tl0_resv10a:	BTRAP(0x10a) BTRAP(0x10b) BTRAP(0x10c) BTRAP(0x10d) BTRAP(0x10e)
diff --git a/arch/sparc64/kernel/unaligned.c b/arch/sparc64/kernel/unaligned.c
index f66889195..3aea13953 100644
--- a/arch/sparc64/kernel/unaligned.c
+++ b/arch/sparc64/kernel/unaligned.c
@@ -1,4 +1,4 @@
-/* $Id: unaligned.c,v 1.1 1997/07/18 06:26:45 ralf Exp $
+/* $Id: unaligned.c,v 1.4 1997/08/19 15:25:11 jj Exp $
  * unaligned.c: Unaligned load/store trap handling with special
  *              cases for the kernel to do them more quickly.
  *
@@ -205,19 +205,19 @@ __asm__ __volatile__ (								\
 	"stx	%%g7, [%0 + 8]\n"						\
 "0:\n\n\t"									\
 	".section __ex_table\n\t"						\
-	".xword	4b, " #errh "\n\t"						\
-	".xword	5b, " #errh "\n\t"						\
-	".xword	6b, " #errh "\n\t"						\
-	".xword	7b, " #errh "\n\t"						\
-	".xword	8b, " #errh "\n\t"						\
-	".xword	9b, " #errh "\n\t"						\
-	".xword	10b, " #errh "\n\t"						\
-	".xword	11b, " #errh "\n\t"						\
-	".xword	12b, " #errh "\n\t"						\
-	".xword	13b, " #errh "\n\t"						\
-	".xword	14b, " #errh "\n\t"						\
-	".xword	15b, " #errh "\n\t"						\
-	".xword	16b, " #errh "\n\n\t"						\
+	".word	4b, " #errh "\n\t"						\
+	".word	5b, " #errh "\n\t"						\
+	".word	6b, " #errh "\n\t"						\
+	".word	7b, " #errh "\n\t"						\
+	".word	8b, " #errh "\n\t"						\
+	".word	9b, " #errh "\n\t"						\
+	".word	10b, " #errh "\n\t"						\
+	".word	11b, " #errh "\n\t"						\
+	".word	12b, " #errh "\n\t"						\
+	".word	13b, " #errh "\n\t"						\
+	".word	14b, " #errh "\n\t"						\
+	".word	15b, " #errh "\n\t"						\
+	".word	16b, " #errh "\n\n\t"						\
 	".previous\n\t"								\
 	: : "r" (dest_reg), "r" (size), "r" (saddr), "r" (is_signed), "r" (asi)	\
 	: "l1", "l2", "g7", "g1", "cc");					\
@@ -259,20 +259,20 @@ __asm__ __volatile__ (								\
 "17:\t"	"stba	%%l1, [%0 + 1] %%asi\n"						\
 "0:\n\n\t"									\
 	".section __ex_table\n\t"						\
-	".xword	4b, " #errh "\n\t"						\
-	".xword	5b, " #errh "\n\t"						\
-	".xword	6b, " #errh "\n\t"						\
-	".xword	7b, " #errh "\n\t"						\
-	".xword	8b, " #errh "\n\t"						\
-	".xword	9b, " #errh "\n\t"						\
-	".xword	10b, " #errh "\n\t"						\
-	".xword	11b, " #errh "\n\t"						\
-	".xword	12b, " #errh "\n\t"						\
-	".xword	13b, " #errh "\n\t"						\
-	".xword	14b, " #errh "\n\t"						\
-	".xword	15b, " #errh "\n\t"						\
-	".xword	16b, " #errh "\n\t"						\
-	".xword	17b, " #errh "\n\n\t"						\
+	".word	4b, " #errh "\n\t"						\
+	".word	5b, " #errh "\n\t"						\
+	".word	6b, " #errh "\n\t"						\
+	".word	7b, " #errh "\n\t"						\
+	".word	8b, " #errh "\n\t"						\
+	".word	9b, " #errh "\n\t"						\
+	".word	10b, " #errh "\n\t"						\
+	".word	11b, " #errh "\n\t"						\
+	".word	12b, " #errh "\n\t"						\
+	".word	13b, " #errh "\n\t"						\
+	".word	14b, " #errh "\n\t"						\
+	".word	15b, " #errh "\n\t"						\
+	".word	16b, " #errh "\n\t"						\
+	".word	17b, " #errh "\n\n\t"						\
 	".previous\n\t"								\
 	: : "r" (dst_addr), "r" (size), "r" (src_val), "r" (asi)		\
 	: "l1", "l2", "g7", "g1", "cc");					\
diff --git a/arch/sparc64/kernel/winfixup.S b/arch/sparc64/kernel/winfixup.S
index f2c714eae..0ebf92767 100644
--- a/arch/sparc64/kernel/winfixup.S
+++ b/arch/sparc64/kernel/winfixup.S
@@ -1,4 +1,4 @@
-/* $Id: winfixup.S,v 1.16 1997/07/13 20:02:42 davem Exp $
+/* $Id: winfixup.S,v 1.19 1997/08/08 08:33:37 jj Exp $
  *
  * winfixup.S: Handle cases where user stack pointer is found to be bogus.
  *
@@ -143,8 +143,9 @@ spill_fixup:
 	retry
 window_scheisse_from_user_common:
 	wrpr		%g1, %cwp
+	sethi		%hi(109f), %g7
 	ba,pt		%xcc, etrap
-	 rd		%pc, %g7
+109:	 or		%g7, %lo(109b), %g7
 window_scheisse_merge:
 	srlx		%l5, PAGE_SHIFT, %o1
 
@@ -244,8 +245,9 @@ spill_fixup_mna:
 	retry
 window_mna_from_user_common:
 	wrpr		%g1, %cwp
+	sethi		%hi(109f), %g7
 	ba,pt		%xcc, etrap
-	 rd		%pc, %g7
+109:	 or		%g7, %lo(109b), %g7
 window_mna_merge:
 	call		mem_address_unaligned
 	 add		%sp, STACK_BIAS + REGWIN_SZ, %o0
diff --git a/arch/sparc64/lib/Makefile b/arch/sparc64/lib/Makefile
index 3da21c606..9f8729ee5 100644
--- a/arch/sparc64/lib/Makefile
+++ b/arch/sparc64/lib/Makefile
@@ -1,12 +1,12 @@
-# $Id: Makefile,v 1.13 1997/07/16 10:12:03 jj Exp $
+# $Id: Makefile,v 1.15 1997/08/19 03:11:50 davem Exp $
 # Makefile for Sparc library files..
 #
 
-CFLAGS := $(CFLAGS) -ansi
+CFLAGS := $(CFLAGS)
 
-OBJS  = blockops.o locks.o strlen.o strncmp.o \
+OBJS  = PeeCeeI.o blockops.o locks.o strlen.o strncmp.o \
 	memscan.o strncpy_from_user.o strlen_user.o memcmp.o checksum.o \
-	VIScopy.o VISbzero.o VISmemset.o VIScsum.o
+	VIScopy.o VISbzero.o VISmemset.o VIScsum.o VIScsumcopy.o
 
 lib.a: $(OBJS)
 	$(AR) rcs lib.a $(OBJS)
diff --git a/arch/sparc64/lib/PeeCeeI.c b/arch/sparc64/lib/PeeCeeI.c
new file mode 100644
index 000000000..6677f581a
--- /dev/null
+++ b/arch/sparc64/lib/PeeCeeI.c
@@ -0,0 +1,225 @@
+/* $Id: PeeCeeI.c,v 1.3 1997/08/28 23:59:52 davem Exp $
+ * PeeCeeI.c: The emerging standard...
+ *
+ * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
+ */
+
+#include <linux/config.h>
+
+#ifdef CONFIG_PCI
+
+#include <asm/io.h>
+
+void outsb(unsigned long addr, const void *src, unsigned long count)
+{
+	const u8 *p = src;
+
+	while(count--)
+		outb(*p++, addr);
+}
+
+void outsw(unsigned long addr, const void *src, unsigned long count)
+{
+	if(count) {
+		const u16 *ps = src;
+		const u32 *pi;
+
+		if(((u64)src) & 0x2) {
+			outw(*ps++, addr);
+			count--;
+		}
+		pi = (const u32 *)ps;
+		while(count >= 2) {
+			u32 w;
+
+			w = *pi++;
+			outw(w >> 16, addr);
+			outw(w, addr);
+			count -= 2;
+		}
+		ps = (const u16 *)pi;
+		if(count)
+			outw(*ps, addr);
+	}
+}
+
+void outsl(unsigned long addr, const void *src, unsigned long count)
+{
+	if(count) {
+		if((((u64)src) & 0x3) == 0) {
+			const u32 *p = src;
+			while(count--)
+				outl(*p++, addr);
+		} else {
+			const u8 *pb;
+			const u16 *ps = src;
+			u32 l = 0, l2;
+			const u32 *pi;
+
+			switch(((u64)src) & 0x3) {
+			case 0x2:
+				count -= 1;
+				l = *ps++;
+				pi = (const u32 *)ps;
+				while(count--) {
+					l2 = *pi++;
+					outl(((l <<16) | (l2 >> 16)), addr);
+					l = l2;
+				}
+				ps = (const u16 *)pi;
+				outl(((l << 16) | (*ps >> 16)), addr);
+				break;
+
+			case 0x1:
+				count -= 1;
+				pb = src;
+				l = (*pb++ << 16);
+				ps = (const u16 *)pb;
+				l |= *ps++;
+				pi = (const u32 *)ps;
+				while(count--) {
+					l2 = *pi++;
+					outl(((l << 8) | (l2 >> 24)), addr);
+					l = l2;
+				}
+				pb = (const u8 *)pi;
+				outl(((l << 8) | (*pb >> 24)), addr);
+				break;
+
+			case 0x3:
+				count -= 1;
+				pb = src;
+				l = (*pb++ >> 24);
+				pi = (const u32 *)pb;
+				while(count--) {
+					l2 = *pi++;
+					outl(((l << 24) | (l2 >> 8)), addr);
+					l = l2;
+				}
+				ps = (const u16 *)pi;
+				l2 = (*ps++ << 16);
+				pb = (const u8 *)ps;
+				l2 |= (*pb << 8);
+				outl(((l << 24) | (l2 >> 8)), addr);
+				break;
+			}
+		}
+	}
+}
+
+void insb(unsigned long addr, void *dst, unsigned long count)
+{
+	if(count) {
+		u32 *pi;
+		u8 *pb = dst;
+
+		while((((unsigned long)pb) & 0x3) && count--)
+			*pb++ = inb(addr);
+		pi = (u32 *)pb;
+		while(count >= 4) {
+			u32 w;
+
+			w  = (inb(addr) << 24);
+			w |= (inb(addr) << 16);
+			w |= (inb(addr) << 8);
+			w |= inb(addr);
+			*pi++ = w;
+			count -= 4;
+		}
+		pb = (u8 *)pi;
+		while(count--)
+			*pb++ = inb(addr);
+	}
+}
+
+void insw(unsigned long addr, void *dst, unsigned long count)
+{
+	if(count) {
+		u16 *ps = dst;
+		u32 *pi;
+
+		if(((unsigned long)ps) & 0x2) {
+			*ps++ = inw(addr);
+			count--;
+		}
+		pi = (u32 *)ps;
+		while(count >= 2) {
+			u32 w;
+
+			w  = (inw(addr) << 16);
+			w |= inw(addr);
+			*pi++ = w;
+			count -= 2;
+		}
+		ps = (u16 *)pi;
+		if(count)
+			*ps = inw(addr);
+	}
+}
+
+void insl(unsigned long addr, void *dst, unsigned long count)
+{
+	if(count) {
+		if((((unsigned long)dst) & 0x3) == 0) {
+			u32 *pi = dst;
+			while(count--)
+				*pi++ = inl(addr);
+		} else {
+			u32 l = 0, l2, *pi;
+			u16 *ps;
+			u8 *pb;
+
+			switch(((unsigned long)dst) & 3) {
+			case 0x2:
+				ps = dst;
+				count -= 1;
+				l = inl(addr);
+				*ps++ = (l >> 16);
+				pi = (u32 *)ps;
+				while(count--) {
+					l2 = inl(addr);
+					*pi++ = (l << 16) | (l2 >> 16);
+					l = l2;
+				}
+				ps = (u16 *)pi;
+				*ps = (l << 16);
+				break;
+
+			case 0x1:
+				pb = dst;
+				count -= 1;
+				*pb++ = (l >> 24);
+				ps = (u16 *)pb;
+				*ps++ = (l >> 8);
+				pi = (u32 *)ps;
+				while(count--) {
+					l2 = inl(addr);
+					*pi++ = ((l << 24) | (l2 >> 8));
+					l = l2;
+				}
+				pb = (u8 *)pi;
+				*pb = (l >> 8);
+				break;
+
+			case 0x3:
+				pb = (u8 *)dst;
+				count -= 1;
+				l = inl(addr);
+				*pb++ = l >> 24;
+				pi = (u32 *)pb;
+				while(count--) {
+					l2 = inl(addr);
+					*pi++ = ((l >> 24) | (l2 << 8));
+					l = l2;
+				}
+				ps = (u16 *)pi;
+				*ps++ = l >> 8;
+				pb = (u8 *)ps;
+				*pb = l;
+				break;
+			}
+		}
+	}
+}
+
+#endif /* CONFIG_PCI */
diff --git a/arch/sparc64/lib/VISbzero.S b/arch/sparc64/lib/VISbzero.S
index 3c86861fd..ede87843b 100644
--- a/arch/sparc64/lib/VISbzero.S
+++ b/arch/sparc64/lib/VISbzero.S
@@ -1,4 +1,4 @@
-/* $Id: VISbzero.S,v 1.1 1997/07/18 06:26:48 ralf Exp $
+/* $Id: VISbzero.S,v 1.8 1997/08/22 15:54:50 jj Exp $
  * VISbzero.S: High speed clear operations utilizing the UltraSparc
  *        Visual Instruction Set.
  *
@@ -16,8 +16,8 @@
 99:	ba	VISbzerofixup_ret##z;		\
 	 a, b, %o0;				\
 	.section __ex_table;			\
-	.align	8;				\
-	.xword	98b, 99b;			\
+	.align	4;				\
+	.word	98b, 99b;			\
 	.text;					\
 	.align	4;
 #define EXC(x,y,a,b,c...) 			\
@@ -28,15 +28,15 @@
 	ba	VISbzerofixup_ret0;		\
 	 a, b, %o0;				\
 	.section __ex_table;			\
-	.align	8;				\
-	.xword	98b, 99b;			\
+	.align	4;				\
+	.word	98b, 99b;			\
 	.text;					\
 	.align	4;
 #define EXO1(x,y) 				\
 98: 	x,y;					\
 	.section __ex_table;			\
-	.align	8;				\
-	.xword	98b, VISbzerofixup_reto1;	\
+	.align	4;				\
+	.word	98b, VISbzerofixup_reto1;	\
 	.text;					\
 	.align	4;
 #define EX(x,y,a,b) EXN(x,y,a,b,0)
@@ -44,8 +44,8 @@
 #define EX2(x,y,a,b) EXN(x,y,a,b,2)
 #define EXT(start,end,handler) 			\
 	.section __ex_table;			\
-	.align	8;				\
-	.xword	start, 0, end, handler;		\
+	.align	4;				\
+	.word	start, 0, end, handler;		\
 	.text;					\
 	.align	4
 #else
@@ -147,7 +147,7 @@ bzero:
 #else
 	 wr		%g0, ASI_BLK_P, %asi
 #endif
-	membar		#StoreStore | #LoadStore
+	membar		#StoreLoad | #StoreStore | #LoadStore
 	fzero		%f0
 	andcc		%o3, 0xc0, %o2
 	and		%o1, 0x3f, %o1
@@ -180,16 +180,28 @@ bzero:
 #ifdef __KERNEL__
 	wr		%g0, 0, %fprs
 	wr		%g7, 0x0, %asi
+#else
+#ifndef REGS_64BIT
+	wr		%g0, FPRS_FEF, %fprs
 #endif
-	membar		#Sync
+#endif
+	membar		#StoreLoad | #StoreStore
 9:	andcc		%o1, 0xf8, %o2
 	be,pn		%xcc, 13f
 	 andcc		%o1, 7, %o1
+#ifdef __KERNEL__
+14:	sethi		%hi(13f), %o4
+	srl		%o2, 1, %o3
+	sub		%o4, %o3, %o4
+	jmpl		%o4 + %lo(13f), %g0
+	 add		%o0, %o2, %o0
+#else
 14:	rd		%pc, %o4
 	srl		%o2, 1, %o3
 	sub		%o4, %o3, %o4
 	jmpl		%o4 + (13f - 14b), %g0
 	 add		%o0, %o2, %o0
+#endif
 12:	ZERO_BLOCKS(%o0, 0xc8, %g0)
 	ZERO_BLOCKS(%o0, 0x88, %g0)
 	ZERO_BLOCKS(%o0, 0x48, %g0)
diff --git a/arch/sparc64/lib/VIScopy.S b/arch/sparc64/lib/VIScopy.S
index 1429f1658..40b781e73 100644
--- a/arch/sparc64/lib/VIScopy.S
+++ b/arch/sparc64/lib/VIScopy.S
@@ -1,4 +1,4 @@
-/* $Id: VIScopy.S,v 1.1 1997/07/18 06:26:48 ralf Exp $
+/* $Id: VIScopy.S,v 1.14 1997/08/22 15:54:53 jj Exp $
  * VIScopy.S: High speed copy operations utilizing the UltraSparc
  *            Visual Instruction Set.
  *
@@ -42,8 +42,8 @@
 99:	ba	VIScopyfixup_ret;		\
 	 a, b, %o0;				\
 	.section __ex_table;			\
-	.align	8;				\
-	.xword	98b, 99b;			\
+	.align	4;				\
+	.word	98b, 99b;			\
 	.text;					\
 	.align	4;
 #define EX2(x,y,c,d,e,a,b) 			\
@@ -54,37 +54,48 @@
 	ba	VIScopyfixup_ret;		\
 	 a, b, %o0;				\
 	.section __ex_table;			\
-	.align	8;				\
-	.xword	98b, 99b;			\
+	.align	4;				\
+	.word	98b, 99b;			\
 	.text;					\
 	.align	4;
 #define EXO2(x,y) 				\
 98: 	x,y;					\
 	.section __ex_table;			\
-	.align	8;				\
-	.xword	98b, VIScopyfixup_reto2;	\
+	.align	4;				\
+	.word	98b, VIScopyfixup_reto2;	\
 	.text;					\
 	.align	4;
 #define EXVISN(x,y,n) 				\
 98: 	x,y;					\
 	.section __ex_table;			\
-	.align	8;				\
-	.xword	98b, VIScopyfixup_vis##n;	\
+	.align	4;				\
+	.word	98b, VIScopyfixup_vis##n;	\
 	.text;					\
 	.align	4;
 #define EXT(start,end,handler) 			\
 	.section __ex_table;			\
-	.align	8;				\
-	.xword	start, 0, end, handler;		\
+	.align	4;				\
+	.word	start, 0, end, handler;		\
 	.text;					\
 	.align	4;
 #else
-#define FPU_CLEAN_RETL	\
-	retl;		\
+#ifdef REGS_64BIT
+#define FPU_CLEAN_RETL				\
+	retl;					\
 	 mov	%g6, %o0;
-#define FPU_RETL	\
-	retl;		\
+#define FPU_RETL				\
+	retl;					\
 	 mov	%g6, %o0;
+#else
+#define FPU_CLEAN_RETL				\
+	wr	%g0, FPRS_FEF, %fprs;		\
+	retl;					\
+	 mov	%g6, %o0;
+#define FPU_RETL				\
+	wr	%g0, FPRS_FEF, %fprs;		\
+	retl;					\
+	 mov	%g6, %o0;
+#endif
 #define NORMAL_RETL	\
 	retl;		\
 	 mov	%g6, %o0;
@@ -113,8 +124,8 @@
 
 #define MAIN_LOOP_CHUNK(src, dest, fdest, fsrc, len, jmptgt)	\
 	EXVIS(LDBLK		[%src] ASIBLK, %fdest);		\
-	add			%src, 0x40, %src;		\
 	ASI_SETDST_BLK						\
+	add			%src, 0x40, %src;		\
 	add			%dest, 0x40, %dest;		\
 	subcc			%len, 0x40, %len;		\
 	be,pn			%xcc, jmptgt;			\
@@ -303,7 +314,7 @@ copy_page:	wr		%g0, FPRS_FEF, %fprs		! FPU	Group
 		sethi		%hi(8192), %o2			! IEU0	Group
 		mov		ASI_BLK_P, asi_src		! IEU1
 		b,pt		%xcc, dest_is_64byte_aligned	! CTI
-		 mov		ASI_BLK_COMMIT_P, asi_dest	! IEU0	Group
+		 mov		ASI_BLK_P, asi_dest		! IEU0	Group
 
 		.align			32
 		.globl			__copy_from_user
@@ -446,6 +457,13 @@ dest_is_64byte_aligned:
 	EXVIS1(LDBLK		[%o1 + 0x40] ASIBLK, %f16)	! LSU	Group
 	sub			%g7, 0x80, %g7			! IEU0
 	EXVIS(LDBLK		[%o1 + 0x80] ASIBLK, %f32)	! LSU	Group
+#ifdef __KERNEL__
+vispc:	sll			%g2, 9, %g2			! IEU0	Group
+	sethi			%hi(vis00), %g5			! IEU1
+	or			%g5, %lo(vis00), %g5		! IEU0	Group
+	jmpl			%g5 + %g2, %g0			! CTI	Group brk forced
+	 addcc			%o1, 0xc0, %o1			! IEU1	Group
+#else
 								! Clk1	Group 8-(
 								! Clk2	Group 8-(
 								! Clk3	Group 8-(
@@ -455,6 +473,7 @@ vispc:	rd			%pc, %g5			! PDU	Group 8-(
 	sll			%g2, 9, %g2			! IEU0
 	jmpl			%g5 + %g2, %g0			! CTI	Group brk forced
 	 addcc			%o1, 0xc0, %o1			! IEU1	Group
+#endif
 	.align			512		/* OK, here comes the fun part... */
 vis00:FREG_FROB(f0, f2, f4, f6, f8, f10,f12,f14,f16) LOOP_CHUNK1(o1, o0, g7, vis01)
       FREG_FROB(f16,f18,f20,f22,f24,f26,f28,f30,f32) LOOP_CHUNK2(o1, o0, g7, vis02)
@@ -721,20 +740,21 @@ __memcpy_16plus:
 3:	andcc		%o2, 0x70, %g7				! IEU1	Group
 41:	be,pn		%xcc, 80f				! CTI
 	 andcc		%o2, 8, %g0				! IEU1	Group
-								! Clk1 8-(
-								! Clk2 8-(
-								! Clk3 8-(
-								! Clk4 8-(
-79:	rd		%pc, %o5				! PDU	Group
 #ifdef __KERNEL__
+79:	sethi		%hi(80f), %o5				! IEU0
 	sll		%g7, 1, %g5				! IEU0	Group
 	add		%o1, %g7, %o1				! IEU1
 	srl		%g7, 1, %g2				! IEU0  Group
 	sub		%o5, %g5, %o5				! IEU1
 	sub		%o5, %g2, %o5				! IEU0  Group
-	jmpl		%o5 + %lo(80f - 79b), %g0		! CTI	Group brk forced
+	jmpl		%o5 + %lo(80f), %g0			! CTI	Group brk forced
 	 add		%o0, %g7, %o0				! IEU0	Group
 #else
+								! Clk1 8-(
+								! Clk2 8-(
+								! Clk3 8-(
+								! Clk4 8-(
+79:	rd		%pc, %o5				! PDU	Group
 	sll		%g7, 1, %g5				! IEU0	Group
 	add		%o1, %g7, %o1				! IEU1
 	sub		%o5, %g5, %o5				! IEU0  Group
@@ -814,19 +834,20 @@ normal_retl:
 	andcc		%o2, 0x70, %g7				! IEU1
 	be,pn		%xcc, 84f				! CTI
 	 andcc		%o2, 8, %g0				! IEU1	Group
+#ifdef __KERNEL__
+83:	srl		%g7, 1, %g5				! IEU0
+	sethi		%hi(84f), %o5				! IEU0	Group
+	add		%g7, %g5, %g5				! IEU1
+	add		%o1, %g7, %o1				! IEU0	Group
+	sub		%o5, %g5, %o5				! IEU1
+	jmpl		%o5 + %lo(84f), %g0			! CTI	Group brk forced
+	 add		%o0, %g7, %o0				! IEU0	Group
+#else
 								! Clk1 8-(
 								! Clk2 8-(
 								! Clk3 8-(
 								! Clk4 8-(
 83:	rd		%pc, %o5				! PDU	Group
-#ifdef __KERNEL__
-	srl		%g7, 1, %g5				! IEU0	Group
-	add		%g7, %g5, %g5				! IEU0	Group
-	add		%o1, %g7, %o1				! IEU1
-	sub		%o5, %g5, %o5				! IEU0	Group
-	jmpl		%o5 + %lo(84f - 83b), %g0		! CTI	Group brk forced
-	 add		%o0, %g7, %o0				! IEU0	Group
-#else
 	add		%o1, %g7, %o1				! IEU0	Group
 	sub		%o5, %g7, %o5				! IEU1
 	jmpl		%o5 + %lo(84f - 83b), %g0		! CTI	Group brk forced
diff --git a/arch/sparc64/lib/VIScsum.S b/arch/sparc64/lib/VIScsum.S
index 1ccb98759..81b020c49 100644
--- a/arch/sparc64/lib/VIScsum.S
+++ b/arch/sparc64/lib/VIScsum.S
@@ -1,4 +1,4 @@
-/* $Id: VIScsum.S,v 1.1 1997/07/18 06:26:49 ralf Exp $
+/* $Id: VIScsum.S,v 1.2 1997/08/08 08:34:05 jj Exp $
  * VIScsum.S: High bandwidth IP checksumming utilizing the UltraSparc
  *            Visual Instruction Set.
  *
@@ -345,11 +345,19 @@ csum_partial:
 20:	andcc		%o1, 0xf0, %g1		/*  IEU1	Group			*/
 	be,pn		%icc, 23f		/*  CTI					*/
 	 and		%o1, 0xf, %o3		/*  IEU0				*/
-22:	rd		%pc, %g7		/*  LSU		Group			*/
+#ifdef __KERNEL__
+22:	sll		%g1, 1, %o4		/*  IEU0	Group			*/
+	sethi		%hi(23f), %g7		/*  IEU1				*/
+	sub		%g7, %o4, %g7		/*  IEU0	Group			*/
+	jmpl		%g7 + %lo(23f), %g0	/*  CTI		Group brk forced	*/
+	 add		%o0, %g1, %o0		/*  IEU0				*/
+#else
+22:	rd		%pc, %g7		/*  LSU		Group+4bubbles		*/
 	sll		%g1, 1, %o4		/*  IEU0	Group			*/
 	sub		%g7, %o4, %g7		/*  IEU0	Group (regdep)		*/
 	jmpl		%g7 + (23f - 22b), %g0	/*  CTI		Group brk forced	*/
-	add		%o0, %g1, %o0		/*  IEU0				*/
+	 add		%o0, %g1, %o0		/*  IEU0				*/
+#endif
 	CSUM_LASTCHUNK(0xe0)
 	CSUM_LASTCHUNK(0xd0)
 	CSUM_LASTCHUNK(0xc0)
diff --git a/arch/sparc64/lib/VIScsumcopy.S b/arch/sparc64/lib/VIScsumcopy.S
new file mode 100644
index 000000000..efd2bfcd5
--- /dev/null
+++ b/arch/sparc64/lib/VIScsumcopy.S
@@ -0,0 +1,880 @@
+/* $Id: VIScsumcopy.S,v 1.2 1997/08/19 15:25:22 jj Exp $
+ * VIScsumcopy.S: High bandwidth IP checksumming with simultaneous
+ *            copying utilizing the UltraSparc Visual Instruction Set.
+ *
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ *
+ * Based on older sparc32/sparc64 checksum.S, which is:
+ *
+ *      Copyright(C) 1995 Linus Torvalds
+ *      Copyright(C) 1995 Miguel de Icaza
+ *      Copyright(C) 1996,1997 David S. Miller
+ *    derived from:
+ *	  Linux/Alpha checksum c-code
+ *        Linux/ix86 inline checksum assembly
+ *        RFC1071 Computing the Internet Checksum (esp. Jacobsons m68k code)
+ *	  David Mosberger-Tang for optimized reference c-code
+ *	  BSD4.4 portable checksum routine
+ */
+
+#ifdef __sparc_v9__
+#define STACKOFF	0x7ff+128
+#else
+#define STACKOFF	64
+#endif
+
+#ifdef __KERNEL__
+#include <asm/head.h>
+#include <asm/asi.h>
+#include <asm/page.h>
+#else
+#define ASI_P		0x80
+#define ASI_BLK_P	0xf0
+#define FRPS_FEF	0x04
+#define FPRS_DU		0x02
+#define FPRS_DL		0x01
+#endif
+#define ASI_BLK_XOR	(ASI_BLK_P ^ ASI_P)
+
+#define src		o0
+#define dst		o1
+#define	len		o2
+#define sum		o3
+#define x1		g1
+#define x2		g2
+#define x3		g3
+#define x4		g4
+#define x5		g5
+#define x6		g7
+#define x7		o4
+#define x8		o5
+
+/* Dobrou noc, SunSoft engineers. Spete sladce.
+ * This has a couple of tricks in and those
+ * tricks are UltraLinux trade secrets :))
+ * Once AGAIN, the SunSoft engineers are caught
+ * asleep at the keyboard :)).
+ * The main loop does about 20 superscalar cycles
+ * per 64bytes checksummed/copied.
+ */
+
+#define LDBLK(O0)										\
+	ldda		[%src] %asi, %O0 	/*  Load	Group			*/
+
+#define STBLK											\
+	stda		%f48, [%dst] ASI_BLK_P	/*  Store				*/
+
+#define ST(fx,off)										\
+	std		%fx, [%dst + off]	/*  Store				*/
+
+#define SYNC											\
+	membar		#Sync
+
+
+#define DO_THE_TRICK(f0,f2,f4,f6,f8,f10,f12,f14,F0,F2,F4,F6,F8,F10,F12,F14,DUMMY1,A0,A2,A4,A6,A8,A10,A12,A14,B14,DYMMY2,LOAD,STORE1,STORE2,STORE3,STORE4,STORE5,STORE6,STORE7,STORE8,DUMMY3,BRANCH...)	\
+	LOAD					/*  Load	Group			*/;	\
+	faligndata	%A14, %F0, %A14		/*  FPA		Group			*/;	\
+	inc		%x5			/*  IEU0				*/;	\
+	STORE1					/*  Store (optional)			*/;	\
+	faligndata	%F0, %F2, %A0		/*  FPA		Group			*/;	\
+	srl		%x5, 1, %x5		/*  IEU0				*/;	\
+	add		%sum, %x4, %sum		/*  IEU1				*/;	\
+	fpadd32		%F0, %f0, %F0		/*  FPA		Group			*/;	\
+	inc		%x6			/*  IEU0				*/;	\
+	STORE2					/*  Store (optional)			*/;	\
+	faligndata	%F2, %F4, %A2		/*  FPA		Group			*/;	\
+	srl		%x6, 1, %x6		/*  IEU0				*/;	\
+	add		%sum, %x5, %sum		/*  IEU1				*/;	\
+	fpadd32		%F2, %f2, %F2		/*  FPA		Group			*/;	\
+	add		%src, 64, %src		/*  IEU0				*/;	\
+	add		%dst, 64, %dst		/*  IEU1				*/;	\
+	fcmpgt32	%f0, %F0, %x1		/*  FPM		Group			*/;	\
+	inc		%x7			/*  IEU0				*/;	\
+	STORE3					/*  Store (optional)			*/;	\
+	faligndata	%F4, %F6, %A4		/*  FPA					*/;	\
+	srl		%x7, 1, %x7		/*  IEU0	Group			*/;	\
+	add		%sum, %x6, %sum		/*  IEU1				*/;	\
+	fpadd32		%F4, %f4, %F4		/*  FPA					*/;	\
+	fcmpgt32	%f2, %F2, %x2		/*  FPM		Group			*/;	\
+	inc		%x8			/*  IEU0				*/;	\
+	STORE4					/*  Store (optional)			*/;	\
+	faligndata	%F6, %F8, %A6		/*  FPA					*/;	\
+	srl		%x8, 1, %x8		/*  IEU0	Group			*/;	\
+	add		%sum, %x7, %sum		/*  IEU1				*/;	\
+	fpadd32		%F6, %f6, %F6		/*  FPA					*/;	\
+	fcmpgt32	%f4, %F4, %x3		/*  FPM		Group			*/;	\
+	inc		%x1			/*  IEU0				*/;	\
+	STORE5					/*  Store (optional)			*/;	\
+	faligndata	%F8, %F10, %A8		/*  FPA					*/;	\
+	srl		%x1, 1, %x1		/*  IEU0	Group			*/;	\
+	add		%sum, %x8, %sum		/*  IEU1				*/;	\
+	fpadd32		%F8, %f8, %F8		/*  FPA					*/;	\
+	fcmpgt32	%f6, %F6, %x4		/*  FPM		Group			*/;	\
+	inc		%x2			/*  IEU0				*/;	\
+	STORE6					/*  Store (optional)			*/;	\
+	faligndata	%F10, %F12, %A10	/*  FPA					*/;	\
+	srl		%x2, 1, %x2		/*  IEU0	Group			*/;	\
+	add		%sum, %x1, %sum		/*  IEU1				*/;	\
+	fpadd32		%F10, %f10, %F10	/*  FPA					*/;	\
+	fcmpgt32	%f8, %F8, %x5		/*  FPM		Group			*/;	\
+	inc		%x3			/*  IEU0				*/;	\
+	STORE7					/*  Store (optional)			*/;	\
+	faligndata	%F12, %F14, %A12	/*  FPA					*/;	\
+	srl		%x3, 1, %x3		/*  IEU0	Group			*/;	\
+	add		%sum, %x2, %sum		/*  IEU1				*/;	\
+	fpadd32		%F12, %f12, %F12	/*  FPA					*/;	\
+	fcmpgt32	%f10, %F10, %x6		/*  FPM		Group			*/;	\
+	inc		%x4			/*  IEU0				*/;	\
+	STORE8					/*  Store (optional)			*/;	\
+	fmovd		%F14, %B14		/*  FPA					*/;	\
+	srl		%x4, 1, %x4		/*  IEU0	Group			*/;	\
+	add		%sum, %x3, %sum		/*  IEU1				*/;	\
+	fpadd32		%F14, %f14, %F14	/*  FPA					*/;	\
+	fcmpgt32	%f12, %F12, %x7		/*  FPM		Group			*/;	\
+	subcc		%len, 64, %len		/*  IEU1				*/;	\
+	BRANCH					/*  CTI					*/;	\
+	fcmpgt32	%f14, %F14, %x8		/*  FPM		Group			*/;	\
+
+#define END_THE_TRICK(f0,f2,f4,f6,f8,f10,f12,f14,FA,FB,S0,S1,S2,S3,T0,T1,U0,fz) \
+	inc		%x5			/*  IEU0	Group			*/;	\
+	fpadd32		%f2, %f0, %S0		/*  FPA					*/;	\
+	srl		%x5, 1, %x5		/*  IEU0	Group			*/;	\
+	add		%sum, %x4, %sum		/*  IEU1				*/;	\
+	fpadd32		%f6, %f4, %S1		/*  FPA					*/;	\
+	inc		%x6			/*  IEU0	Group			*/;	\
+	add		%sum, %x5, %sum		/*  IEU1				*/;	\
+	fcmpgt32	%f0, %S0, %x1		/*  FPM		Group			*/;	\
+	srl		%x6, 1, %x6		/*  IEU0				*/;	\
+	inc		%x7			/*  IEU1				*/;	\
+	fpadd32		%f10, %f8, %S2		/*  FPA					*/;	\
+	fcmpgt32	%f4, %S1, %x2		/*  FPM		Group			*/;	\
+	srl		%x7, 1, %x7		/*  IEU0				*/;	\
+	add		%sum, %x6, %sum		/*  IEU1				*/;	\
+	fpadd32		%f14, %f12, %S3		/*  FPA					*/;	\
+	inc		%x8			/*  IEU0	Group			*/;	\
+	add		%sum, %x7, %sum		/*  IEU1				*/;	\
+	fzero		%fz			/*  FPA					*/;	\
+	fcmpgt32	%f8, %S2, %x3		/*  FPM		Group			*/;	\
+	srl		%x8, 1, %x8		/*  IEU0				*/;	\
+	inc		%x1			/*  IEU1				*/;	\
+	fpadd32		%S0, %S1, %T0		/*  FPA					*/;	\
+	fcmpgt32	%f12, %S3, %x4		/*  FPM		Group			*/;	\
+	srl		%x1, 1, %x1		/*  IEU0				*/;	\
+	add		%sum, %x8, %sum		/*  IEU1				*/;	\
+	fpadd32		%S2, %S3, %T1		/*  FPA					*/;	\
+	inc		%x2			/*  IEU0	Group			*/;	\
+	add		%sum, %x1, %sum		/*  IEU1				*/;	\
+	fcmpgt32	%S0, %T0, %x5		/*  FPM		Group			*/;	\
+	srl		%x2, 1, %x2		/*  IEU0				*/;	\
+	inc		%x3			/*  IEU1				*/;	\
+	fcmpgt32	%S2, %T1, %x6		/*  FPM		Group			*/;	\
+	srl		%x3, 1, %x3		/*  IEU0				*/;	\
+	add		%sum, %x2, %sum		/*  IEU1				*/;	\
+	inc		%x4			/*  IEU0	Group			*/;	\
+	add		%sum, %x3, %sum		/*  IEU1				*/;	\
+	fcmpgt32	%fz, %f2, %x7		/*  FPM		Group			*/;	\
+	srl		%x4, 1, %x4		/*  IEU0				*/;	\
+	inc		%x5			/*  IEU1				*/;	\
+	fpadd32		%T0, %T1, %U0		/*  FPA					*/;	\
+	fcmpgt32	%fz, %f6, %x8		/*  FPM		Group			*/;	\
+	srl		%x5, 1, %x5		/*  IEU0				*/;	\
+	add		%sum, %x4, %sum		/*  IEU1				*/;	\
+	inc		%x6			/*  IEU0	Group			*/;	\
+	add		%sum, %x5, %sum		/*  IEU1				*/;	\
+	fcmpgt32	%fz, %f10, %x1		/*  FPM		Group			*/;	\
+	srl		%x6, 1, %x6		/*  IEU0				*/;	\
+	inc		%x7			/*  IEU1				*/;	\
+	fcmpgt32	%fz, %f14, %x2		/*  FPM		Group			*/;	\
+	ba,pt		%xcc, ett		/*  CTI					*/;	\
+	 fmovd		%FA, %FB		/*  FPA					*/;	\
+
+#define END_THE_TRICK1(f0,f2,f4,f6,f8,f10,f12,f14,FA,FB) 					\
+	END_THE_TRICK(f0,f2,f4,f6,f8,f10,f12,f14,FA,FB,f48,f50,f52,f54,f56,f58,f60,f62)
+
+#define END_THE_TRICK2(S0,S1,S2,S3,T0,T1,U0,U1,V0,fz) 						\
+	fpadd32		%U0, %U1, %V0		/*  FPA		Group			*/;	\
+	srl		%x7, 1, %x7		/*  IEU0				*/;	\
+	add		%sum, %x6, %sum		/*  IEU1				*/;	\
+	std		%V0, [%sp + STACKOFF]	/*  Store	Group			*/;	\
+	inc		%x8			/*  IEU0				*/;	\
+	sub		%sum, %x7, %sum		/*  IEU1				*/;	\
+	fcmpgt32	%fz, %S1, %x3		/*  FPM		Group			*/;	\
+	srl		%x8, 1, %x8		/*  IEU0				*/;	\
+	inc		%x1			/*  IEU1				*/;	\
+	fcmpgt32	%fz, %S3, %x4		/*  FPM		Group			*/;	\
+	srl		%x1, 1, %x1		/*  IEU0				*/;	\
+	sub		%sum, %x8, %sum		/*  IEU1				*/;	\
+	ldx		[%sp + STACKOFF], %x8	/*  Load	Group			*/;	\
+	inc		%x2			/*  IEU0				*/;	\
+	sub		%sum, %x1, %sum		/*  IEU1				*/;	\
+	fcmpgt32	%fz, %T1, %x5		/*  FPM		Group			*/;	\
+	srl		%x2, 1, %x2		/*  IEU0				*/;	\
+	inc		%x3			/*  IEU1				*/;	\
+	fcmpgt32	%T0, %U0, %x6		/*  FPM		Group			*/;	\
+	srl		%x3, 1, %x3		/*  IEU0				*/;	\
+	sub		%sum, %x2, %sum		/*  IEU1				*/;	\
+	inc		%x4			/*  IEU0	Group			*/;	\
+	sub		%sum, %x3, %sum		/*  IEU1				*/;	\
+	fcmpgt32	%fz, %U1, %x7		/*  FPM		Group			*/;	\
+	srl		%x4, 1, %x4		/*  IEU0				*/;	\
+	inc		%x5			/*  IEU1				*/;	\
+	fcmpgt32	%U0, %V0, %x1		/*  FPM		Group			*/;	\
+	srl		%x5, 1, %x5		/*  IEU0				*/;	\
+	sub		%sum, %x4, %sum		/*  IEU1				*/;	\
+	fcmpgt32	%fz, %V0, %x2		/*  FPM		Group			*/;	\
+	inc		%x6			/*  IEU0				*/;	\
+	sub		%sum, %x5, %sum		/*  IEU1				*/;	\
+	srl		%x6, 1, %x6		/*  IEU0	Group			*/;	\
+	inc		%x7			/*  IEU1				*/;	\
+	srl		%x7, 1, %x7		/*  IEU0	Group			*/;	\
+	add		%sum, %x6, %sum		/*  IEU1				*/;	\
+	inc		%x1			/*  IEU0	Group			*/;	\
+	sub		%sum, %x7, %sum		/*  IEU1				*/;	\
+	srl		%x1, 1, %x1		/*  IEU0	Group			*/;	\
+	inc		%x2			/*  IEU1				*/;	\
+	srl		%x2, 1, %x2		/*  IEU0	Group			*/;	\
+	add		%sum, %x1, %sum		/*  IEU1				*/;	\
+	sub		%sum, %x2, %sum		/*  IEU0	Group			*/;	\
+	addcc		%sum, %x8, %sum		/*  IEU		Group			*/;	\
+	bcs,a,pn	%xcc, 33f		/*  CTI					*/;	\
+	 add		%sum, 1, %sum		/*  IEU0				*/;	\
+33:						/*  That's it				*/;
+
+	.text
+	.globl		csum_partial_copy_vis
+	.align		32
+/* %asi should be either ASI_P or ASI_S for csum_partial_copy resp. csum_partial_copy_from_user */
+/* This assumes that !((%src^%dst)&3) && !((%src|%dst)&1) && %len >= 256 */
+csum_partial_copy_vis:
+	andcc		%dst, 7, %g0		/*  IEU1	Group			*/
+	be,pt		%icc, 4f		/*  CTI					*/
+	 and		%dst, 0x38, %g3		/*  IEU0				*/
+	mov		1, %g5			/*  IEU0	Group			*/
+	andcc		%dst, 2, %g0		/*  IEU1				*/
+	be,pt		%icc, 1f		/*  CTI					*/
+	 and		%dst, 4, %g7		/*  IEU0	Group			*/
+	lduha		[%src] %asi, %g2	/*  Load				*/
+	sub		%len, 2, %len		/*  IEU0	Group			*/
+	add		%dst, 2, %dst		/*  IEU1				*/
+	andcc		%dst, 4, %g7		/*  IEU1	Group			*/
+	sll		%g5, 16, %g5		/*  IEU0				*/
+	sth		%g2, [%dst - 2]		/*  Store	Group			*/
+	sll		%g2, 16, %g2		/*  IEU0	 			*/
+	add		%src, 2, %src		/*  IEU1				*/
+	addcc		%g2, %sum, %sum		/*  IEU1	Group			*/
+	bcs,a,pn	%icc, 1f		/*  CTI					*/
+	 add		%sum, %g5, %sum		/*  IEU0				*/
+1:	lduwa		[%src] %asi, %g2	/*  Load				*/
+	brz,a,pn	%g7, 4f			/*  CTI+IEU1	Group			*/
+	 and		%dst, 0x38, %g3		/*  IEU0				*/
+	add		%dst, 4, %dst		/*  IEU0	Group			*/
+	sub		%len, 4, %len		/*  IEU1				*/
+	addcc		%g2, %sum, %sum		/*  IEU1	Group			*/
+	bcs,a,pn	%icc, 1f		/*  CTI					*/
+	 add		%sum, 1, %sum		/*  IEU0				*/
+1:	and		%dst, 0x38, %g3		/*  IEU0	Group			*/
+	stw		%g2, [%dst - 4]		/*  Store				*/
+	add		%src, 4, %src		/*  IEU1				*/
+4:
+#ifdef __KERNEL__
+	wr		%g0, FPRS_FEF, %fprs	/*  LSU		Group			*/
+#endif
+	mov		%src, %g7		/*  IEU1	Group			*/
+	fzero		%f48			/*  FPA					*/
+	alignaddr	%src, %g0, %src		/*  Single	Group			*/
+	subcc		%g7, %src, %g7		/*  IEU1	Group			*/
+	be,pt		%xcc, 1f		/*  CTI					*/
+	 mov		0x40, %g1		/*  IEU0				*/
+	lduwa		[%src] %asi, %g2	/*  Load	Group			*/
+	subcc		%sum, %g2, %sum		/*  IEU1	Group+load stall	*/
+	bcs,a,pn	%icc, 1f		/*  CTI					*/
+	 sub		%sum, 1, %sum		/*  IEU0				*/
+1:	srl		%sum, 0, %sum		/*  IEU0	Group			*/
+	clr		%g5			/*  IEU1				*/
+	brz,pn		%g3, 3f			/*  CTI+IEU1	Group			*/
+	 sub		%g1, %g3, %g1		/*  IEU0				*/
+	ldda		[%src] %asi, %f0	/*  Load				*/
+	clr		%g3			/*  IEU0	Group			*/
+	andcc		%dst, 8, %g0		/*  IEU1				*/
+	be,pn		%icc, 1f		/*  CTI					*/
+	 ldda		[%src + 8] %asi, %f2	/*  Load	Group			*/
+	add		%src, 8, %src		/*  IEU0				*/
+	sub		%len, 8, %len		/*  IEU1				*/
+	fpadd32		%f0, %f48, %f50		/*  FPA					*/
+	addcc		%dst, 8, %dst		/*  IEU1	Group			*/
+	faligndata	%f0, %f2, %f16		/*  FPA					*/
+	fcmpgt32	%f48, %f50, %g3		/*  FPM		Group			*/
+	fmovd		%f2, %f0		/*  FPA		Group			*/
+	ldda		[%src + 8] %asi, %f2	/*  Load				*/
+	std		%f16, [%dst - 8]	/*  Store				*/
+	fmovd		%f50, %f48		/*  FPA					*/
+1:	andcc		%g1, 0x10, %g0		/*  IEU1	Group			*/
+	be,pn		%icc, 1f		/*  CTI					*/
+	 and		%g1, 0x20, %g1		/*  IEU0				*/
+	fpadd32		%f0, %f48, %f50		/*  FPA					*/
+	ldda		[%src + 16] %asi, %f4	/*  Load	Group			*/
+	add		%src, 16, %src		/*  IEU0				*/
+	add		%dst, 16, %dst		/*  IEU1				*/
+	faligndata	%f0, %f2, %f16		/*  FPA					*/
+	fcmpgt32	%f48, %f50, %g5		/*  FPM		Group			*/
+	sub		%len, 16, %len		/*  IEU0				*/
+	inc		%g3			/*  IEU1				*/
+	std		%f16, [%dst - 16]	/*  Store	Group			*/
+	fpadd32		%f2, %f50, %f48		/*  FPA					*/
+	srl		%g3, 1, %o5		/*  IEU0				*/
+	faligndata	%f2, %f4, %f18		/*  FPA		Group			*/
+	std		%f18, [%dst - 8]	/*  Store				*/
+	fcmpgt32	%f50, %f48, %g3		/*  FPM		Group			*/
+	add		%o5, %sum, %sum		/*  IEU0				*/
+	ldda		[%src + 8] %asi, %f2	/*  Load				*/
+	fmovd		%f4, %f0		/*  FPA					*/
+1:	brz,a,pn	%g1, 4f			/*  CTI+IEU1	Group			*/
+	 rd		%asi, %g2		/*  LSU		Group + 4 bubbles	*/
+	inc		%g5			/*  IEU0				*/
+	fpadd32		%f0, %f48, %f50		/*  FPA					*/
+	ldda		[%src + 16] %asi, %f4	/*  Load	Group			*/
+	srl		%g5, 1, %g5		/*  IEU0				*/
+	add		%dst, 32, %dst		/*  IEU1				*/
+	faligndata	%f0, %f2, %f16		/*  FPA					*/
+	fcmpgt32	%f48, %f50, %o5		/*  FPM		Group			*/
+	inc		%g3			/*  IEU0				*/
+	ldda		[%src + 24] %asi, %f6	/*  Load				*/
+	srl		%g3, 1, %g3		/*  IEU0	Group			*/
+	add		%g5, %sum, %sum		/*  IEU1				*/
+	ldda		[%src + 32] %asi, %f8	/*  Load				*/
+	fpadd32		%f2, %f50, %f48		/*  FPA					*/
+	faligndata	%f2, %f4, %f18		/*  FPA		Group			*/
+	sub		%len, 32, %len		/*  IEU0				*/
+	std		%f16, [%dst - 32]	/*  Store				*/
+	fcmpgt32	%f50, %f48, %o4		/*  FPM		Group			*/
+	inc		%o5			/*  IEU0				*/
+	add		%g3, %sum, %sum		/*  IEU1				*/
+	fpadd32		%f4, %f48, %f50		/*  FPA					*/
+	faligndata	%f4, %f6, %f20		/*  FPA		Group			*/
+	srl		%o5, 1, %o5		/*  IEU0				*/
+	fcmpgt32	%f48, %f50, %g5		/*  FPM		Group			*/
+	add		%o5, %sum, %sum		/*  IEU0				*/
+	std		%f18, [%dst - 24]	/*  Store				*/
+	fpadd32		%f6, %f50, %f48		/*  FPA					*/
+	inc		%o4			/*  IEU0	Group			*/
+	std		%f20, [%dst - 16]	/*  Store				*/
+	add		%src, 32, %src		/*  IEU1				*/
+	faligndata	%f6, %f8, %f22		/*  FPA					*/
+	fcmpgt32	%f50, %f48, %g3		/*  FPM		Group			*/
+	srl		%o4, 1, %o4		/*  IEU0				*/
+	std		%f22, [%dst - 8]	/*  Store				*/	
+	add		%o4, %sum, %sum		/*  IEU0	Group			*/
+3:	rd		%asi, %g2		/*  LSU		Group + 4 bubbles	*/
+#ifdef __KERNEL__
+4:	sethi		%hi(vis0s), %g7		/*  IEU0	Group			*/
+#else
+4:	rd		%pc, %g7		/*  LSU		Group + 4 bubbles	*/
+#endif
+	inc		%g5			/*  IEU0	Group			*/
+	and		%src, 0x38, %o4		/*  IEU1				*/	
+	membar		#StoreLoad		/*  LSU		Group			*/
+	srl		%g5, 1, %g5		/*  IEU0				*/
+	inc		%g3			/*  IEU1				*/
+	sll		%o4, 8, %o4		/*  IEU0	Group			*/
+	sub		%len, 0xc0, %len	/*  IEU1				*/
+	addcc		%g5, %sum, %sum		/*  IEU1	Group			*/
+	srl		%g3, 1, %g3		/*  IEU0				*/
+	add		%g7, %o4, %g7		/*  IEU0	Group			*/
+	add		%g3, %sum, %sum		/*  IEU1				*/
+#ifdef __KERNEL__
+	jmpl		%g7 + %lo(vis0s), %g0	/*  CTI+IEU1	Group			*/
+#else
+	jmpl		%g7 + (vis0s - 4b), %g0	/*  CTI+IEU1	Group			*/
+#endif
+	 fzero		%f32			/*  FPA					*/
+
+	.align		2048
+vis0s:	wr		%g2, ASI_BLK_XOR, %asi	/*  LSU		Group			*/
+	add		%src, 128, %src		/*  IEU0	Group			*/
+	ldda		[%src-128] %asi, %f0	/*  Load	Group			*/
+	ldda		[%src-64] %asi, %f16	/*  Load	Group			*/
+	fmovd		%f48, %f62		/*  FPA		Group			*/
+	faligndata	%f0, %f2, %f48		/*  FPA		Group			*/
+	fcmpgt32	%f32, %f2, %x1		/*  FPM		Group			*/
+	fpadd32		%f0, %f62, %f0		/*  FPA					*/
+	fcmpgt32	%f32, %f4, %x2		/*  FPM		Group			*/
+	faligndata	%f2, %f4, %f50		/*  FPA					*/
+	fcmpgt32	%f62, %f0, %x3		/*  FPM		Group			*/
+	faligndata	%f4, %f6, %f52		/*  FPA					*/
+	fcmpgt32	%f32, %f6, %x4		/*  FPM		Group			*/
+	inc		%x1			/*  IEU0				*/
+	faligndata	%f6, %f8, %f54		/*  FPA					*/
+	fcmpgt32	%f32, %f8, %x5		/*  FPM		Group			*/
+	srl		%x1, 1, %x1		/*  IEU0				*/
+	inc		%x2			/*  IEU1				*/
+	faligndata	%f8, %f10, %f56		/*  FPA					*/
+	fcmpgt32	%f32, %f10, %x6		/*  FPM		Group			*/
+	srl		%x2, 1, %x2		/*  IEU0				*/
+	add		%sum, %x1, %sum		/*  IEU1				*/
+	faligndata	%f10, %f12, %f58	/*  FPA					*/
+	fcmpgt32	%f32, %f12, %x7		/*  FPM		Group			*/
+	inc		%x3			/*  IEU0				*/
+	add		%sum, %x2, %sum		/*  IEU1				*/
+	faligndata	%f12, %f14, %f60	/*  FPA					*/
+	fcmpgt32	%f32, %f14, %x8		/*  FPM		Group			*/
+	srl		%x3, 1, %x3		/*  IEU0				*/
+	inc		%x4			/*  IEU1				*/
+	fmovd		%f14, %f62		/*  FPA					*/
+	srl		%x4, 1, %x4		/*  IEU0	Group			*/
+	add		%sum, %x3, %sum		/*  IEU1				*/
+vis0:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,
+			,f48,f50,f52,f54,f56,f58,f60,f62,f62,								
+			,LDBLK(f32),	STBLK,,,,,,,,									
+			,bcs,pn %icc, vis0e1)
+	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f48,f50,f52,f54,f56,f58,f60,f62,f62,								
+			,LDBLK(f0),	STBLK,,,,,,,,									
+			,bcs,pn %icc, vis0e2)
+	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f48,f50,f52,f54,f56,f58,f60,f62,f62,
+			,LDBLK(f16),	STBLK,,,,,,,,
+			,bcc,pt %icc, vis0)
+vis0e3:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f48,f50,f52,f54,f56,f58,f60,f62,f32,
+			,SYNC,		STBLK,ST(f48,64),ST(f50,8),ST(f52,16),ST(f54,24),ST(f56,32),ST(f58,40),ST(f60,48),
+			,add %dst, 56, %dst; add %len, 192 - 8*8, %len; ba,pt %icc, e2)
+vis0e1:	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f48,f50,f52,f54,f56,f58,f60,f62,f0,
+			,SYNC,		STBLK,ST(f48,64),ST(f50,8),ST(f52,16),ST(f54,24),ST(f56,32),ST(f58,40),ST(f60,48),
+			,add %dst, 56, %dst; add %len, 192 - 8*8, %len; ba,pt %icc, e3)
+vis0e2:	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f48,f50,f52,f54,f56,f58,f60,f62,f16,
+			,SYNC,		STBLK,ST(f48,64),ST(f50,8),ST(f52,16),ST(f54,24),ST(f56,32),ST(f58,40),ST(f60,48),
+			,add %dst, 56, %dst; add %len, 192 - 8*8, %len; ba,pt %icc, e1)
+	.align		2048
+vis1s:	wr		%g2, ASI_BLK_XOR, %asi	/*  LSU		Group			*/
+	add		%src, 128 - 8, %src	/*  IEU0	Group			*/
+	ldda		[%src-128] %asi, %f0	/*  Load	Group			*/
+	ldda		[%src-64] %asi, %f16	/*  Load	Group			*/
+	fmovd		%f48, %f0		/*  FPA		Group			*/
+	fcmpgt32	%f32, %f2, %x2		/*  FPM		Group			*/
+	faligndata	%f2, %f4, %f48		/*  FPA					*/
+	fcmpgt32	%f32, %f4, %x3		/*  FPM		Group			*/
+	faligndata	%f4, %f6, %f50		/*  FPA					*/
+	fcmpgt32	%f32, %f6, %x4		/*  FPM		Group			*/
+	faligndata	%f6, %f8, %f52		/*  FPA					*/
+	fcmpgt32	%f32, %f8, %x5		/*  FPM		Group			*/
+	inc		%x2			/*  IEU1				*/
+	faligndata	%f8, %f10, %f54		/*  FPA					*/
+	fcmpgt32	%f32, %f10, %x6		/*  FPM		Group			*/
+	srl		%x2, 1, %x2		/*  IEU0				*/
+	faligndata	%f10, %f12, %f56	/*  FPA					*/
+	fcmpgt32	%f32, %f12, %x7		/*  FPM		Group			*/
+	inc		%x3			/*  IEU0				*/
+	add		%sum, %x2, %sum		/*  IEU1				*/
+	faligndata	%f12, %f14, %f58	/*  FPA					*/
+	fcmpgt32	%f32, %f14, %x8		/*  FPM		Group			*/
+	srl		%x3, 1, %x3		/*  IEU0				*/
+	inc		%x4			/*  IEU1				*/
+	fmovd		%f14, %f60		/*  FPA					*/
+	srl		%x4, 1, %x4		/*  IEU0	Group			*/
+	add		%sum, %x3, %sum		/*  IEU1				*/
+vis1:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f62,f48,f50,f52,f54,f56,f58,f60,f60,
+			,LDBLK(f32),	,STBLK,,,,,,,
+			,bcs,pn %icc, vis1e1)
+	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f62,f48,f50,f52,f54,f56,f58,f60,f60,
+			,LDBLK(f0),	,STBLK,,,,,,,
+			,bcs,pn %icc, vis1e2)
+	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f62,f48,f50,f52,f54,f56,f58,f60,f60,
+			,LDBLK(f16),	,STBLK,,,,,,,
+			,bcc,pt %icc, vis1)
+vis1e3:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f62,f48,f50,f52,f54,f56,f58,f60,f32,
+			,SYNC,		,STBLK,ST(f48,0),ST(f50,8),ST(f52,16),ST(f54,24),ST(f56,32),ST(f58,40),
+			,add %dst, 48, %dst; add %len, 192 - 7*8, %len; ba,pt %icc, e2)
+vis1e1:	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f62,f48,f50,f52,f54,f56,f58,f60,f0,
+			,SYNC,		,STBLK,ST(f48,0),ST(f50,8),ST(f52,16),ST(f54,24),ST(f56,32),ST(f58,40),
+			,add %dst, 48, %dst; add %len, 192 - 7*8, %len; ba,pt %icc, e3)
+vis1e2:	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f62,f48,f50,f52,f54,f56,f58,f60,f16,
+			,SYNC,		,STBLK,ST(f48,0),ST(f50,8),ST(f52,16),ST(f54,24),ST(f56,32),ST(f58,40),
+			,add %dst, 48, %dst; add %len, 192 - 7*8, %len; ba,pt %icc, e1)
+	.align		2048
+vis2s:	wr		%g2, ASI_BLK_XOR, %asi	/*  LSU		Group			*/
+	add		%src, 128 - 16, %src	/*  IEU0	Group			*/
+	ldda		[%src-128] %asi, %f0	/*  Load	Group			*/
+	ldda		[%src-64] %asi, %f16	/*  Load	Group			*/
+	fmovd		%f48, %f0		/*  FPA		Group			*/	
+	sub		%dst, 64, %dst		/*  IEU0				*/
+	fzero		%f2			/*  FPA		Group			*/
+	fcmpgt32	%f32, %f4, %x3		/*  FPM		Group			*/
+	faligndata	%f4, %f6, %f48		/*  FPA					*/
+	fcmpgt32	%f32, %f6, %x4		/*  FPM		Group			*/
+	faligndata	%f6, %f8, %f50		/*  FPA					*/
+	fcmpgt32	%f32, %f8, %x5		/*  FPM		Group			*/
+	faligndata	%f8, %f10, %f52		/*  FPA					*/
+	fcmpgt32	%f32, %f10, %x6		/*  FPM		Group			*/
+	faligndata	%f10, %f12, %f54	/*  FPA					*/
+	fcmpgt32	%f32, %f12, %x7		/*  FPM		Group			*/
+	inc		%x3			/*  IEU0				*/
+	faligndata	%f12, %f14, %f56	/*  FPA					*/
+	fcmpgt32	%f32, %f14, %x8		/*  FPM		Group			*/
+	srl		%x3, 1, %x3		/*  IEU0				*/
+	inc		%x4			/*  IEU1				*/
+	fmovd		%f14, %f58		/*  FPA					*/
+	srl		%x4, 1, %x4		/*  IEU0	Group			*/
+	add		%sum, %x3, %sum		/*  IEU1				*/
+vis2:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f60,f62,f48,f50,f52,f54,f56,f58,f58,
+			,LDBLK(f32),	,,STBLK,,,,,,
+			,bcs,pn %icc, vis2e1)
+	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f60,f62,f48,f50,f52,f54,f56,f58,f58,
+			,LDBLK(f0),	,,STBLK,,,,,,
+			,bcs,pn %icc, vis2e2)
+	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f60,f62,f48,f50,f52,f54,f56,f58,f58,
+			,LDBLK(f16),	,,STBLK,,,,,,
+			,bcc,pt %icc, vis2)
+vis2e3:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f60,f62,f48,f50,f52,f54,f56,f58,f32,
+			,SYNC,		,,STBLK,ST(f48,64),ST(f50,72),ST(f52,80),ST(f54,88),ST(f56,96),
+			,add %dst, 104, %dst; add %len, 192 - 6*8, %len; ba,pt %icc, e2)
+vis2e1:	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f60,f62,f48,f50,f52,f54,f56,f58,f0,
+			,SYNC,		,,STBLK,ST(f48,64),ST(f50,72),ST(f52,80),ST(f54,88),ST(f56,96),
+			,add %dst, 104, %dst; add %len, 192 - 6*8, %len; ba,pt %icc, e3)
+vis2e2:	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f60,f62,f48,f50,f52,f54,f56,f58,f16,
+			,SYNC,		,,STBLK,ST(f48,64),ST(f50,72),ST(f52,80),ST(f54,88),ST(f56,96),
+			,add %dst, 104, %dst; add %len, 192 - 6*8, %len; ba,pt %icc, e1)
+	.align		2048
+vis3s:	wr		%g2, ASI_BLK_XOR, %asi	/*  LSU		Group			*/
+	add		%src, 128 - 24, %src	/*  IEU0	Group			*/
+	ldda		[%src-128] %asi, %f0	/*  Load	Group			*/
+	ldda		[%src-64] %asi, %f16	/*  Load	Group			*/
+	fmovd		%f48, %f0		/*  FPA		Group			*/
+	sub		%dst, 64, %dst		/*  IEU0				*/
+	fzero		%f2			/*  FPA		Group			*/
+	fzero		%f4			/*  FPA		Group			*/
+	fcmpgt32	%f32, %f6, %x4		/*  FPM		Group			*/
+	faligndata	%f6, %f8, %f48		/*  FPA					*/
+	fcmpgt32	%f32, %f8, %x5		/*  FPM		Group			*/
+	faligndata	%f8, %f10, %f50		/*  FPA					*/
+	fcmpgt32	%f32, %f10, %x6		/*  FPM		Group			*/
+	faligndata	%f10, %f12, %f52	/*  FPA					*/
+	fcmpgt32	%f32, %f12, %x7		/*  FPM		Group			*/
+	faligndata	%f12, %f14, %f54	/*  FPA					*/
+	fcmpgt32	%f32, %f14, %x8		/*  FPM		Group			*/
+	fmovd		%f14, %f56		/*  FPA					*/
+	inc		%x4			/*  IEU0				*/
+	srl		%x4, 1, %x4		/*  IEU0	Group			*/
+vis3:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f58,f60,f62,f48,f50,f52,f54,f56,f56,
+			,LDBLK(f32),	,,,STBLK,,,,,
+			,bcs,pn %icc, vis3e1)
+	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f58,f60,f62,f48,f50,f52,f54,f56,f56,
+			,LDBLK(f0),	,,,STBLK,,,,,
+			,bcs,pn %icc, vis3e2)
+	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f58,f60,f62,f48,f50,f52,f54,f56,f56,
+			,LDBLK(f16),	,,,STBLK,,,,,
+			,bcc,pt %icc, vis3)
+vis3e3:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f58,f60,f62,f48,f50,f52,f54,f56,f32,
+			,SYNC,		,,,STBLK,ST(f48,64),ST(f50,72),ST(f52,80),ST(f54,88),
+			,add %dst, 96, %dst; add %len, 192 - 5*8, %len; ba,pt %icc, e2)
+vis3e1:	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f58,f60,f62,f48,f50,f52,f54,f56,f0,
+			,SYNC,		,,,STBLK,ST(f48,64),ST(f50,72),ST(f52,80),ST(f54,88),
+			,add %dst, 96, %dst; add %len, 192 - 5*8, %len; ba,pt %icc, e3)
+vis3e2:	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f58,f60,f62,f48,f50,f52,f54,f56,f16,
+			,SYNC,		,,,STBLK,ST(f48,64),ST(f50,72),ST(f52,80),ST(f54,88),
+			,add %dst, 96, %dst; add %len, 192 - 5*8, %len; ba,pt %icc, e1)
+	.align		2048
+vis4s:	wr		%g2, ASI_BLK_XOR, %asi	/*  LSU		Group			*/
+	add		%src, 128 - 32, %src	/*  IEU0	Group			*/
+	ldda		[%src-128] %asi, %f0	/*  Load	Group			*/
+	ldda		[%src-64] %asi, %f16	/*  Load	Group			*/
+	fmovd		%f48, %f0		/*  FPA		Group			*/
+	sub		%dst, 64, %dst		/*  IEU0				*/
+	fzero		%f2			/*  FPA		Group			*/
+	fzero		%f4			/*  FPA		Group			*/
+	fzero		%f6			/*  FPA		Group			*/
+	clr		%x4			/*  IEU0				*/
+	fcmpgt32	%f32, %f8, %x5		/*  FPM		Group			*/
+	faligndata	%f8, %f10, %f48		/*  FPA					*/
+	fcmpgt32	%f32, %f10, %x6		/*  FPM		Group			*/
+	faligndata	%f10, %f12, %f50	/*  FPA					*/
+	fcmpgt32	%f32, %f12, %x7		/*  FPM		Group			*/
+	faligndata	%f12, %f14, %f52	/*  FPA					*/
+	fcmpgt32	%f32, %f14, %x8		/*  FPM		Group			*/
+	fmovd		%f14, %f54		/*  FPA					*/
+vis4:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f56,f58,f60,f62,f48,f50,f52,f54,f54,
+			,LDBLK(f32),	,,,,STBLK,,,,
+			,bcs,pn %icc, vis4e1)
+	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f56,f58,f60,f62,f48,f50,f52,f54,f54,
+			,LDBLK(f0),	,,,,STBLK,,,,
+			,bcs,pn %icc, vis4e2)
+	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f56,f58,f60,f62,f48,f50,f52,f54,f54,
+			,LDBLK(f16),	,,,,STBLK,,,,
+			,bcc,pt %icc, vis4)
+vis4e3:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f56,f58,f60,f62,f48,f50,f52,f54,f32,
+			,SYNC,		,,,,STBLK,ST(f48,64),ST(f50,72),ST(f52,80),
+			,add %dst, 88, %dst; add %len, 192 - 4*8, %len; ba,pt %icc, e2)
+vis4e1:	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f56,f58,f60,f62,f48,f50,f52,f54,f0,
+			,SYNC,		,,,,STBLK,ST(f48,64),ST(f50,72),ST(f52,80),
+			,add %dst, 88, %dst; add %len, 192 - 4*8, %len; ba,pt %icc, e3)
+vis4e2:	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f56,f58,f60,f62,f48,f50,f52,f54,f16,
+			,SYNC,		,,,,STBLK,ST(f48,64),ST(f50,72),ST(f52,80),
+			,add %dst, 88, %dst; add %len, 192 - 4*8, %len; ba,pt %icc, e1)
+	.align		2048
+vis5s:	add		%src, 128 - 40, %src	/*  IEU0	Group			*/
+	ldda		[%src-88] %asi, %f10	/*  Load	Group			*/
+	ldda		[%src-80] %asi, %f12	/*  Load	Group			*/
+	ldda		[%src-72] %asi, %f14	/*  Load	Group			*/
+	wr		%g2, ASI_BLK_XOR, %asi	/*  LSU		Group			*/
+	ldda		[%src-64] %asi, %f16	/*  Load	Group			*/
+	fmovd		%f48, %f0		/*  FPA		Group			*/
+	fmuld		%f32, %f32, %f2		/*  FPM					*/
+	clr		%x4			/*  IEU0				*/
+	faddd		%f32, %f32, %f4		/*  FPA		Group			*/
+	fmuld		%f32, %f32, %f6		/*  FPM					*/
+	clr		%x5			/*  IEU0				*/
+	faddd		%f32, %f32, %f8		/*  FPA		Group			*/
+	fcmpgt32	%f32, %f10, %x6		/*  FPM		Group			*/
+	sub		%dst, 64, %dst		/*  IEU0				*/
+	faligndata	%f10, %f12, %f48	/*  FPA					*/
+	fcmpgt32	%f32, %f12, %x7		/*  FPM		Group			*/
+	faligndata	%f12, %f14, %f50	/*  FPA					*/
+	fcmpgt32	%f32, %f14, %x8		/*  FPM		Group			*/
+	fmovd		%f14, %f52		/*  FPA					*/
+vis5:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f54,f56,f58,f60,f62,f48,f50,f52,f52,
+			,LDBLK(f32),	,,,,,STBLK,,,
+			,bcs,pn %icc, vis5e1)
+	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f54,f56,f58,f60,f62,f48,f50,f52,f52,
+			,LDBLK(f0),	,,,,,STBLK,,,
+			,bcs,pn %icc, vis5e2)
+	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f54,f56,f58,f60,f62,f48,f50,f52,f52,
+			,LDBLK(f16),	,,,,,STBLK,,,
+			,bcc,pt %icc, vis5)
+vis5e3:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f54,f56,f58,f60,f62,f48,f50,f52,f32,
+			,SYNC,		,,,,,STBLK,ST(f48,64),ST(f50,72),
+			,add %dst, 80, %dst; add %len, 192 - 3*8, %len; ba,pt %icc, e2)
+vis5e1:	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f54,f56,f58,f60,f62,f48,f50,f52,f0,
+			,SYNC,		,,,,,STBLK,ST(f48,64),ST(f50,72),
+			,add %dst, 80, %dst; add %len, 192 - 3*8, %len; ba,pt %icc, e3)
+vis5e2:	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f54,f56,f58,f60,f62,f48,f50,f52,f16,
+			,SYNC,		,,,,,STBLK,ST(f48,64),ST(f50,72),
+			,add %dst, 80, %dst; add %len, 192 - 3*8, %len; ba,pt %icc, e1)
+	.align		2048
+vis6s:	add		%src, 128 - 48, %src	/*  IEU0	Group			*/
+	ldda		[%src-80] %asi, %f12	/*  Load	Group			*/
+	ldda		[%src-72] %asi, %f14	/*  Load	Group			*/
+	wr		%g2, ASI_BLK_XOR, %asi	/*  LSU		Group			*/
+	ldda		[%src-64] %asi, %f16	/*  Load	Group			*/
+	fmovd		%f48, %f0		/*  FPA		Group			*/
+	fmuld		%f32, %f32, %f2		/*  FPM					*/
+	clr		%x4			/*  IEU0				*/
+	faddd		%f32, %f32, %f4		/*  FPA		Group			*/
+	fmuld		%f32, %f32, %f6		/*  FPM					*/
+	clr		%x5			/*  IEU0				*/
+	faddd		%f32, %f32, %f8		/*  FPA		Group			*/
+	fmuld		%f32, %f32, %f10	/*  FPM					*/
+	clr		%x6			/*  IEU0				*/
+	fcmpgt32	%f32, %f12, %x7		/*  FPM		Group			*/
+	sub		%dst, 64, %dst		/*  IEU0				*/
+	faligndata	%f12, %f14, %f48	/*  FPA					*/
+	fcmpgt32	%f32, %f14, %x8		/*  FPM		Group			*/
+	fmovd		%f14, %f50		/*  FPA					*/
+vis6:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f52,f54,f56,f58,f60,f62,f48,f50,f50,
+			,LDBLK(f32),	,,,,,,STBLK,,
+			,bcs,pn %icc, vis6e1)
+	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f52,f54,f56,f58,f60,f62,f48,f50,f50,
+			,LDBLK(f0),	,,,,,,STBLK,,
+			,bcs,pn %icc, vis6e2)
+	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f52,f54,f56,f58,f60,f62,f48,f50,f50,
+			,LDBLK(f16),	,,,,,,STBLK,,
+			,bcc,pt %icc, vis6)
+vis6e3:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f52,f54,f56,f58,f60,f62,f48,f50,f32,
+			,SYNC,		,,,,,,STBLK,ST(f48,64),
+			,add %dst, 72, %dst; add %len, 192 - 2*8, %len; ba,pt %icc, e2)
+vis6e1:	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f52,f54,f56,f58,f60,f62,f48,f50,f0,
+			,SYNC,		,,,,,,STBLK,ST(f48,64),
+			,add %dst, 72, %dst; add %len, 192 - 2*8, %len; ba,pt %icc, e3)
+vis6e2:	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f52,f54,f56,f58,f60,f62,f48,f50,f16,
+			,SYNC,		,,,,,,STBLK,ST(f48,64),
+			,add %dst, 72, %dst; add %len, 192 - 2*8, %len; ba,pt %icc, e1)
+	.align		2048
+vis7s:	add		%src, 128 - 56, %src	/*  IEU0	Group			*/
+	ldda		[%src-72] %asi, %f14	/*  Load	Group			*/
+	wr		%g2, ASI_BLK_XOR, %asi	/*  LSU		Group			*/
+	ldda		[%src-64] %asi, %f16	/*  Load	Group			*/
+	fmovd		%f48, %f0		/*  FPA		Group			*/
+	fmuld		%f32, %f32, %f2		/*  FPM					*/
+	clr		%x4			/*  IEU0				*/
+	faddd		%f32, %f32, %f4		/*  FPA		Group			*/
+	fmuld		%f32, %f32, %f6		/*  FPM					*/
+	clr		%x5			/*  IEU0				*/
+	faddd		%f32, %f32, %f8		/*  FPA		Group			*/
+	fmuld		%f32, %f32, %f10	/*  FPM					*/
+	clr		%x6			/*  IEU0				*/
+	faddd		%f32, %f32, %f12	/*  FPA		Group			*/
+	clr		%x7			/*  IEU0				*/
+	fcmpgt32	%f32, %f14, %x8		/*  FPM		Group			*/
+	sub		%dst, 64, %dst		/*  IEU0				*/
+	fmovd		%f14, %f48		/*  FPA					*/
+vis7:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f50,f52,f54,f56,f58,f60,f62,f48,f48,
+			,LDBLK(f32),	,,,,,,,STBLK,
+			,bcs,pn %icc, vis7e1)
+	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f50,f52,f54,f56,f58,f60,f62,f48,f48,
+			,LDBLK(f0),	,,,,,,,STBLK,
+			,bcs,pn %icc, vis7e2)
+	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f50,f52,f54,f56,f58,f60,f62,f48,f48,
+			,LDBLK(f16),	,,,,,,,STBLK,
+			,bcc,pt %icc, vis7)
+vis7e3:	DO_THE_TRICK(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f18,f20,f22,f24,f26,f28,f30,	
+			,f50,f52,f54,f56,f58,f60,f62,f48,f32,
+			,SYNC,		,,,,,,,STBLK,
+			,add %dst, 64, %dst; add %len, 192 - 1*8, %len; ba,pt %icc, e2)
+vis7e1:	DO_THE_TRICK(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f34,f36,f38,f40,f42,f44,f46,
+			,f50,f52,f54,f56,f58,f60,f62,f48,f0,
+			,SYNC,		,,,,,,,STBLK,
+			,add %dst, 64, %dst; add %len, 192 - 1*8, %len; ba,pt %icc, e3)
+vis7e2:	DO_THE_TRICK(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f2,f4,f6,f8,f10,f12,f14,	
+			,f50,f52,f54,f56,f58,f60,f62,f48,f16,
+			,SYNC,		,,,,,,,STBLK,
+			,add %dst, 64, %dst; add %len, 192 - 1*8, %len; ba,pt %icc, e1)
+e1:	END_THE_TRICK1(	f0,f2,f4,f6,f8,f10,f12,f14,f16,f6)
+e2:	END_THE_TRICK1(	f16,f18,f20,f22,f24,f26,f28,f30,f32,f6)
+e3:	END_THE_TRICK1(	f32,f34,f36,f38,f40,f42,f44,f46,f0,f6)
+ett:	rd		%gsr, %x3		/*  LSU		Group+4bubbles		*/
+	andcc		%x3, 7, %x3		/*  IEU1	Group			*/
+	add		%dst, 8, %dst		/*  IEU0	Group			*/
+	bne,pn		%icc, 1f		/*  CTI					*/
+	 fzero		%f10			/*  FPA					*/
+	brz,a,pn	%len, 2f		/*  CTI+IEU1	Group			*/
+	 std		%f6, [%dst - 8]		/*  Store				*/
+1:	rd		%asi, %x4		/*  LSU		Group+4bubbles		*/
+	sub		%src, 64, %src		/*  IEU0	Group			*/
+	cmp		%len, 8			/*  IEU1				*/
+	blu,pn		%icc, 3f		/*  CTI					*/
+	 wr		%x4, ASI_BLK_XOR, %asi	/*  LSU		Group+4bubbles		*/
+1:	ldda		[%src] %asi, %f2	/*  Load	Group			*/
+	fpadd32		%f10, %f2, %f12		/*  FPA		Group+load stall	*/
+	add		%src, 8, %src		/*  IEU0				*/
+	add		%dst, 8, %dst		/*  IEU1				*/
+	faligndata	%f6, %f2, %f14		/*  FPA		Group			*/
+	fcmpgt32	%f10, %f12, %x5		/*  FPM		Group			*/
+	std		%f14, [%dst - 16]	/*  Store				*/
+	fmovd		%f2, %f6		/*  FPA					*/
+	fmovd		%f12, %f10		/*  FPA		Group			*/
+	sub		%len, 8, %len		/*  IEU1				*/
+	fzero		%f16			/*  FPA		Group - FPU nop		*/
+	fzero		%f18			/*  FPA		Group - FPU nop		*/
+	inc		%x5			/*  IEU0				*/
+	srl		%x5, 1, %x5		/*  IEU0	Group (regdep)		*/
+	cmp		%len, 8			/*  IEU1				*/
+	bgeu,pt		%icc, 1b		/*  CTI					*/
+	 add		%x5, %sum, %sum		/*  IEU0	Group			*/
+3:	brz,a,pt	%x3, 2f			/*  CTI+IEU1				*/
+	 std		%f6, [%dst - 8]		/*  Store	Group			*/
+	st		%f7, [%dst - 8]		/*  Store	Group			*/
+	sub		%dst, 4, %dst		/*  IEU0				*/
+	add		%len, 4, %len		/*  IEU1				*/
+2:
+#ifdef __KERNEL__
+	sub		%sp, 8, %sp		/*  IEU0	Group			*/
+#endif
+	END_THE_TRICK2(	f48,f50,f52,f54,f56,f58,f60,f10,f12,f62)
+	membar		#Sync			/*  LSU		Group			*/
+#ifdef __KERNEL__
+	wr		%g0, 0, %fprs		/*  LSU		Group			*/
+	add		%sp, 8, %sp		/*  IEU0	Group			*/
+#endif
+23:	brnz,pn		%len, 26f		/*  CTI+IEU1	Group			*/
+24:	 sllx		%sum, 32, %g1		/*  IEU0				*/
+25:	addcc		%sum, %g1, %src		/*  IEU1	Group			*/
+	srlx		%src, 32, %src		/*  IEU0	Group (regdep)		*/
+	bcs,a,pn	%xcc, 1f		/*  CTI					*/
+	 add		%src, 1, %src		/*  IEU1				*/
+#ifndef __KERNEL__
+1:	retl					/*  CTI		Group brk forced	*/
+	 srl		%src, 0, %src		/*  IEU0				*/
+#else
+1:	sethi		%uhi(PAGE_OFFSET), %g4	/*  IEU0	Group			*/
+	retl					/*  CTI		Group brk forced	*/
+	 sllx		%g4, 32, %g4		/*  IEU0				*/
+#endif
+26:	andcc		%len, 8, %g0		/*  IEU1	Group			*/
+	be,pn		%icc, 1f		/*  CTI					*/
+	 lduwa		[%src] %asi, %g3	/*  Load				*/
+	lduwa		[%src+4] %asi, %g2	/*  Load	Group			*/
+	add		%src, 8, %src		/*  IEU0				*/
+	add		%dst, 8, %dst		/*  IEU1				*/
+	sllx		%g3, 32, %g5		/*  IEU0	Group			*/
+	stw		%g3, [%dst - 8]		/*  Store				*/
+	or		%g5, %g2, %g5		/*  IEU0	Group			*/
+	stw		%g2, [%dst - 4]		/*  Store				*/
+	addcc		%g5, %sum, %sum		/*  IEU1	Group			*/
+	bcs,a,pn	%xcc, 1f		/*  CTI					*/
+	 add		%sum, 1, %sum		/*  IEU0				*/
+1:	andcc		%len, 4, %g0		/*  IEU1	Group			*/
+	be,a,pn		%icc, 1f		/*  CTI					*/
+	 clr		%g2			/*  IEU0				*/
+	lduwa		[%src] %asi, %g7	/*  Load				*/
+	add		%src, 4, %src		/*  IEU0	Group			*/
+	add		%dst, 4, %dst		/*  IEU1				*/
+	sllx		%g7, 32, %g2		/*  IEU0	Group			*/
+	stw		%g7, [%dst - 4]		/*  Store				*/
+1:	andcc		%len, 2, %g0		/*  IEU1				*/
+	be,a,pn		%icc, 1f		/*  CTI					*/
+	 clr		%o4			/*  IEU0	Group			*/
+	lduha		[%src] %asi, %g7	/*  Load				*/
+	add		%src, 2, %src		/*  IEU1				*/
+	add		%dst, 2, %dst		/*  IEU0	Group			*/
+	sll		%g7, 16, %o4		/*  IEU0	Group			*/
+	sth		%g7, [%dst - 2]		/*  Store				*/
+1:	andcc		%len, 1, %g0		/*  IEU1				*/
+	be,a,pn		%icc, 1f		/*  CTI					*/
+	 clr		%o5			/*  IEU0	Group			*/
+	lduba		[%src] %asi, %g7	/*  Load				*/
+	sll		%g7, 8, %o5		/*  IEU0	Group			*/
+	stb		%g7, [%dst]		/*  Store				*/
+1:	or		%g2, %o4, %o4		/*  IEU1				*/
+	or		%o5, %o4, %o4		/*  IEU0	Group (regdep)		*/
+	addcc		%o4, %sum, %sum		/*  IEU1	Group (regdep)		*/
+	bcs,a,pn	%xcc, 1f		/*  CTI					*/
+	 add		%sum, 1, %sum		/*  IEU0				*/
+1:	ba,pt		%xcc, 25b		/*  CTI		Group			*/
+	 sllx		%sum, 32, %g1		/*  IEU0				*/
+
+#ifdef __KERNEL__
+end:
+
+	.section	__ex_table
+	.align		4
+	.word		csum_partial_copy_vis, 0, end, cpc_handler
+#endif
diff --git a/arch/sparc64/lib/VISmemset.S b/arch/sparc64/lib/VISmemset.S
index d674f2a6e..4c24931ba 100644
--- a/arch/sparc64/lib/VISmemset.S
+++ b/arch/sparc64/lib/VISmemset.S
@@ -1,4 +1,4 @@
-/* $Id: VISmemset.S,v 1.1 1997/07/18 06:26:49 ralf Exp $
+/* $Id: VISmemset.S,v 1.7 1997/08/22 15:54:56 jj Exp $
  * VISmemset.S: High speed memset operations utilizing the UltraSparc
  *        Visual Instruction Set.
  *
@@ -171,11 +171,22 @@ memset:
 12:
 #ifdef __KERNEL__
 	wr		%g0, 0, %fprs
+#else
+#ifndef REGS_64BIT
+	wr		%g0, FPRS_FEF, %fprs
 #endif
-	membar		#Sync
+#endif
+	membar		#StoreLoad | #StoreStore
 9:	andcc		%o2, 0x78, %g5
 	be,pn		%xcc, 13f
 	 andcc		%o2, 7, %o2
+#ifdef __KERNEL__
+14:	srl		%g5, 1, %o3
+	sethi		%hi(13f), %o4
+	sub		%o4, %o3, %o4
+	jmpl		%o4 + %lo(13f), %g0
+	 add		%o0, %g5, %o0
+#else
 14:	rd		%pc, %o4
 #ifdef REGS_64BIT
 	srl		%g5, 1, %o3
@@ -185,6 +196,7 @@ memset:
 #endif
 	jmpl		%o4 + (13f - 14b), %g0
 	 add		%o0, %g5, %o0
+#endif
 12:	SET_BLOCKS(%o0, 0x68, %o1)
 	SET_BLOCKS(%o0, 0x48, %o1)
 	SET_BLOCKS(%o0, 0x28, %o1)
diff --git a/arch/sparc64/lib/blockops.S b/arch/sparc64/lib/blockops.S
index 59083aa02..7d5b240ad 100644
--- a/arch/sparc64/lib/blockops.S
+++ b/arch/sparc64/lib/blockops.S
@@ -1,4 +1,4 @@
-/* $Id: blockops.S,v 1.10 1997/06/24 17:29:10 jj Exp $
+/* $Id: blockops.S,v 1.11 1997/07/29 09:35:36 davem Exp $
  * arch/sparc64/lib/blockops.S: UltraSparc block zero optimized routines.
  *
  * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
@@ -15,7 +15,7 @@ __bfill64:		/* %o0 = buf, %o1= ptr to pattern */
 	wr		%g0, FPRS_FEF, %fprs		! FPU	Group
 	ldd		[%o1], %f48			! Load	Group
 	wr		%g0, ASI_BLK_P, %asi		! LSU	Group
-	membar		#StoreStore | #LoadStore	! LSU	Group
+	membar		#StoreLoad | #StoreStore | #LoadStore	! LSU	Group
 	mov		32, %g2				! IEU0	Group
 
 	/* Cannot perform real arithmatic on the pattern, that can
@@ -36,7 +36,7 @@ __bfill64:		/* %o0 = buf, %o1= ptr to pattern */
 	subcc		%g2, 1, %g2			! IEU1	Group
 	bne,pt		%icc, 1b			! CTI
 	 add		%o0, 0x100, %o0			! IEU0
-	membar		#Sync				! LSU	Group
+	membar		#StoreLoad | #StoreStore	! LSU	Group
 
 	jmpl		%o7 + 0x8, %g0			! CTI	Group brk forced
 	 wr		%g0, 0, %fprs			! FPU	Group
@@ -56,7 +56,7 @@ __bzero_1page:
 	faddd		%f0, %f2, %f12			! FPA	Group
 	fmuld		%f0, %f2, %f14			! FPM
 	wr		%g0, ASI_BLK_P, %asi		! LSU	Group
-	membar		#StoreStore | #LoadStore	! LSU	Group
+	membar		#StoreLoad | #StoreStore | #LoadStore	! LSU	Group
 1:	stda		%f0, [%o0 + 0x00] %asi		! Store	Group
 	stda		%f0, [%o0 + 0x40] %asi		! Store	Group
 	stda		%f0, [%o0 + 0x80] %asi		! Store	Group
@@ -65,6 +65,6 @@ __bzero_1page:
 	subcc		%g1, 1, %g1			! IEU1
 	bne,pt		%icc, 1b			! CTI
 	 add		%o0, 0x100, %o0			! IEU0	Group
-	membar		#Sync				! LSU	Group
+	membar		#StoreLoad | #StoreStore	! LSU	Group
 	jmpl		%o7 + 0x8, %g0			! CTI	Group brk forced
 	 wr		%g0, 0, %fprs			! FPU	Group
diff --git a/arch/sparc64/lib/checksum.S b/arch/sparc64/lib/checksum.S
index 703370fc6..5f35f136b 100644
--- a/arch/sparc64/lib/checksum.S
+++ b/arch/sparc64/lib/checksum.S
@@ -23,456 +23,456 @@
 	 * are two fold.  Firstly, they cannot pair with jack shit,
 	 * and also they only add in the 32-bit carry condition bit
 	 * into the accumulated sum.  The following is much better.
-	 *
-	 * This should run at max bandwidth for ecache hits, a better
-	 * technique is to use VIS and fpu operations. This is already
-	 * done for csum_partial, needs to be written for the copy stuff
-	 * still.
+	 * For larger chunks we use VIS code, which is faster ;)
 	 */
 
-	.text
-	.globl __csum_partial_copy_start, __csum_partial_copy_end
-__csum_partial_copy_start:
+#define src o0
+#define dst o1
+#define len o2
+#define sum o3
 
+	.text
 	/* I think I have an erection...  Once _AGAIN_ the SunSoft
 	 * engineers are caught asleep at the keyboard, tsk tsk...
 	 */
-#define CSUMCOPY_ECACHE_LOAD(src, off, t0, t1, t2, t3, t4, t5, t6, t7)			\
-	ldxa		[src + off + 0x00] %asi, t0;					\
-	ldxa		[src + off + 0x08] %asi, t1;					\
-	ldxa		[src + off + 0x10] %asi, t2;					\
-	ldxa		[src + off + 0x18] %asi, t3;					\
-	ldxa		[src + off + 0x20] %asi, t4;					\
-	ldxa		[src + off + 0x28] %asi, t5;					\
-	ldxa		[src + off + 0x30] %asi, t6;					\
-	ldxa		[src + off + 0x38] %asi, t7;					\
+#define CSUMCOPY_ECACHE_LOAD(off, t0, t1, t2, t3, t4, t5, t6, t7)			\
+	ldxa		[%src + off + 0x00] %asi, t0;					\
+	ldxa		[%src + off + 0x08] %asi, t1;					\
+	ldxa		[%src + off + 0x10] %asi, t2;					\
+	ldxa		[%src + off + 0x18] %asi, t3;					\
+	ldxa		[%src + off + 0x20] %asi, t4;					\
+	ldxa		[%src + off + 0x28] %asi, t5;					\
+	ldxa		[%src + off + 0x30] %asi, t6;					\
+	ldxa		[%src + off + 0x38] %asi, t7;					\
 	nop; nop; /* DO NOT TOUCH THIS!!!!! */
 
-#define CSUMCOPY_EC_STALIGNED_LDNXT(src, dest, off, sum, t0, t1, t2, t3, t4, t5, t6, t7)\
-	stx		t0, [dest + off - 0x40];					\
-	addcc		sum, t0, sum;							\
+#define CSUMCOPY_EC_STALIGNED_LDNXT(off, t0, t1, t2, t3, t4, t5, t6, t7)		\
+	stx		t0, [%dst + off - 0x40];					\
+	addcc		%sum, t0, %sum;							\
 	bcc,pt		%xcc, 11f;							\
-	 ldxa		[src + off + 0x00] %asi, t0;					\
-	add		sum, 1, sum;							\
-11:	stx		t1, [dest + off - 0x38];					\
-	addcc		sum, t1, sum;							\
+	 ldxa		[%src + off + 0x00] %asi, t0;					\
+	add		%sum, 1, %sum;							\
+11:	stx		t1, [%dst + off - 0x38];					\
+	addcc		%sum, t1, %sum;							\
 	bcc,pt		%xcc, 12f;							\
-	 ldxa		[src + off + 0x08] %asi, t1;					\
-	add		sum, 1, sum;							\
-12:	stx		t2, [dest + off - 0x30];					\
-	addcc		sum, t2, sum;							\
+	 ldxa		[%src + off + 0x08] %asi, t1;					\
+	add		%sum, 1, %sum;							\
+12:	stx		t2, [%dst + off - 0x30];					\
+	addcc		%sum, t2, %sum;							\
 	bcc,pt		%xcc, 13f;							\
-	 ldxa		[src + off + 0x10] %asi, t2;					\
-	add		sum, 1, sum;							\
-13:	stx		t3, [dest + off - 0x28];					\
-	addcc		sum, t3, sum;							\
+	 ldxa		[%src + off + 0x10] %asi, t2;					\
+	add		%sum, 1, %sum;							\
+13:	stx		t3, [%dst + off - 0x28];					\
+	addcc		%sum, t3, %sum;							\
 	bcc,pt		%xcc, 14f;							\
-	 ldxa		[src + off + 0x18] %asi, t3;					\
-	add		sum, 1, sum;							\
-14:	stx		t4, [dest + off - 0x20];					\
-	addcc		sum, t4, sum;							\
+	 ldxa		[%src + off + 0x18] %asi, t3;					\
+	add		%sum, 1, %sum;							\
+14:	stx		t4, [%dst + off - 0x20];					\
+	addcc		%sum, t4, %sum;							\
 	bcc,pt		%xcc, 15f;							\
-	 ldxa		[src + off + 0x20] %asi, t4;					\
-	add		sum, 1, sum;							\
-15:	stx		t5, [dest + off - 0x18];					\
-	addcc		sum, t5, sum;							\
+	 ldxa		[%src + off + 0x20] %asi, t4;					\
+	add		%sum, 1, %sum;							\
+15:	stx		t5, [%dst + off - 0x18];					\
+	addcc		%sum, t5, %sum;							\
 	bcc,pt		%xcc, 16f;							\
-	 ldxa		[src + off + 0x28] %asi, t5;					\
-	add		sum, 1, sum;							\
-16:	stx		t6, [dest + off - 0x10];					\
-	addcc		sum, t6, sum;							\
+	 ldxa		[%src + off + 0x28] %asi, t5;					\
+	add		%sum, 1, %sum;							\
+16:	stx		t6, [%dst + off - 0x10];					\
+	addcc		%sum, t6, %sum;							\
 	bcc,pt		%xcc, 17f;							\
-	 ldxa		[src + off + 0x30] %asi, t6;					\
-	add		sum, 1, sum;							\
-17:	stx		t7, [dest + off - 0x08];					\
-	addcc		sum, t7, sum;							\
+	 ldxa		[%src + off + 0x30] %asi, t6;					\
+	add		%sum, 1, %sum;							\
+17:	stx		t7, [%dst + off - 0x08];					\
+	addcc		%sum, t7, %sum;							\
 	bcc,pt		%xcc, 18f;							\
-	 ldxa		[src + off + 0x38] %asi, t7;					\
-	add		sum, 1, sum;							\
+	 ldxa		[%src + off + 0x38] %asi, t7;					\
+	add		%sum, 1, %sum;							\
 18:
 
-#define CSUMCOPY_EC_STUNALIGN_LDNXT(src, dest, off, sum, t0, t1, t2, t3, t4, t5, t6, t7)\
-	stw		t0, [dest + off - 0x3c];					\
-	addcc		sum, t0, sum;							\
+#define CSUMCOPY_EC_STUNALIGN_LDNXT(off, t0, t1, t2, t3, t4, t5, t6, t7)		\
+	stw		t0, [%dst + off - 0x3c];					\
+	addcc		%sum, t0, %sum;							\
 	srlx		t0, 32, t0;							\
-	stw		t0, [dest + off - 0x40];					\
+	stw		t0, [%dst + off - 0x40];					\
 	bcc,pt		%xcc, 21f;							\
-	 ldxa		[src + off + 0x00] %asi, t0;					\
-	add		sum, 1, sum;							\
-21:	stw		t1, [dest + off - 0x34];					\
-	addcc		sum, t1, sum;							\
+	 ldxa		[%src + off + 0x00] %asi, t0;					\
+	add		%sum, 1, %sum;							\
+21:	stw		t1, [%dst + off - 0x34];					\
+	addcc		%sum, t1, %sum;							\
 	srlx		t1, 32, t1;							\
-	stw		t1, [dest + off - 0x38];					\
+	stw		t1, [%dst + off - 0x38];					\
 	bcc,pt		%xcc, 22f;							\
-	 ldxa		[src + off + 0x08] %asi, t1;					\
-	add		sum, 1, sum;							\
-22:	stw		t2, [dest + off - 0x2c];					\
-	addcc		sum, t2, sum;							\
+	 ldxa		[%src + off + 0x08] %asi, t1;					\
+	add		%sum, 1, %sum;							\
+22:	stw		t2, [%dst + off - 0x2c];					\
+	addcc		%sum, t2, %sum;							\
 	srlx		t2, 32, t2;							\
-	stw		t2, [dest + off - 0x30];					\
+	stw		t2, [%dst + off - 0x30];					\
 	bcc,pt		%xcc, 23f;							\
-	 ldxa		[src + off + 0x10] %asi, t2;					\
-	add		sum, 1, sum;							\
-23:	stw		t3, [dest + off - 0x24];					\
-	addcc		sum, t3, sum;							\
+	 ldxa		[%src + off + 0x10] %asi, t2;					\
+	add		%sum, 1, %sum;							\
+23:	stw		t3, [%dst + off - 0x24];					\
+	addcc		%sum, t3, %sum;							\
 	srlx		t3, 32, t3;							\
-	stw		t3, [dest + off - 0x28];					\
+	stw		t3, [%dst + off - 0x28];					\
 	bcc,pt		%xcc, 24f;							\
-	 ldxa		[src + off + 0x18] %asi, t3;					\
-	add		sum, 1, sum;							\
-24:	stw		t4, [dest + off - 0x1c];					\
-	addcc		sum, t4, sum;							\
+	 ldxa		[%src + off + 0x18] %asi, t3;					\
+	add		%sum, 1, %sum;							\
+24:	stw		t4, [%dst + off - 0x1c];					\
+	addcc		%sum, t4, %sum;							\
 	srlx		t4, 32, t4;							\
-	stw		t4, [dest + off - 0x20];					\
+	stw		t4, [%dst + off - 0x20];					\
 	bcc,pt		%xcc, 25f;							\
-	 ldxa		[src + off + 0x20] %asi, t4;					\
-	add		sum, 1, sum;							\
-25:	stw		t5, [dest + off - 0x14];					\
-	addcc		sum, t5, sum;							\
+	 ldxa		[%src + off + 0x20] %asi, t4;					\
+	add		%sum, 1, %sum;							\
+25:	stw		t5, [%dst + off - 0x14];					\
+	addcc		%sum, t5, %sum;							\
 	srlx		t5, 32, t5;							\
-	stw		t5, [dest + off - 0x18];					\
+	stw		t5, [%dst + off - 0x18];					\
 	bcc,pt		%xcc, 26f;							\
-	 ldxa		[src + off + 0x28] %asi, t5;					\
-	add		sum, 1, sum;							\
-26:	stw		t6, [dest + off - 0x0c];					\
-	addcc		sum, t6, sum;							\
+	 ldxa		[%src + off + 0x28] %asi, t5;					\
+	add		%sum, 1, %sum;							\
+26:	stw		t6, [%dst + off - 0x0c];					\
+	addcc		%sum, t6, %sum;							\
 	srlx		t6, 32, t6;							\
-	stw		t6, [dest + off - 0x10];					\
+	stw		t6, [%dst + off - 0x10];					\
 	bcc,pt		%xcc, 27f;							\
-	 ldxa		[src + off + 0x30] %asi, t6;					\
-	add		sum, 1, sum;							\
-27:	stw		t7, [dest + off - 0x04];					\
-	addcc		sum, t7, sum;							\
+	 ldxa		[%src + off + 0x30] %asi, t6;					\
+	add		%sum, 1, %sum;							\
+27:	stw		t7, [%dst + off - 0x04];					\
+	addcc		%sum, t7, %sum;							\
 	srlx		t7, 32, t7;							\
-	stw		t7, [dest + off - 0x08];					\
+	stw		t7, [%dst + off - 0x08];					\
 	bcc,pt		%xcc, 28f;							\
-	 ldxa		[src + off + 0x38] %asi, t7;					\
-	add		sum, 1, sum;							\
+	 ldxa		[%src + off + 0x38] %asi, t7;					\
+	add		%sum, 1, %sum;							\
 28:
 
-#define CSUMCOPY_EC_STALIGNED(dest, off, sum, t0, t1, t2, t3, t4, t5, t6, t7)		\
-	addcc		sum, t0, sum;							\
+#define CSUMCOPY_EC_STALIGNED(off, t0, t1, t2, t3, t4, t5, t6, t7)			\
+	addcc		%sum, t0, %sum;							\
 	bcc,pt		%xcc, 31f;							\
-	 stx		t0, [dest + off + 0x00];					\
-	add		sum, 1, sum;							\
-31:	addcc		sum, t1, sum;							\
+	 stx		t0, [%dst + off + 0x00];					\
+	add		%sum, 1, %sum;							\
+31:	addcc		%sum, t1, %sum;							\
 	bcc,pt		%xcc, 32f;							\
-	 stx		t1, [dest + off + 0x08];					\
-	add		sum, 1, sum;							\
-32:	addcc		sum, t2, sum;							\
+	 stx		t1, [%dst + off + 0x08];					\
+	add		%sum, 1, %sum;							\
+32:	addcc		%sum, t2, %sum;							\
 	bcc,pt		%xcc, 33f;							\
-	 stx		t2, [dest + off + 0x10];					\
-	add		sum, 1, sum;							\
-33:	addcc		sum, t3, sum;							\
+	 stx		t2, [%dst + off + 0x10];					\
+	add		%sum, 1, %sum;							\
+33:	addcc		%sum, t3, %sum;							\
 	bcc,pt		%xcc, 34f;							\
-	 stx		t3, [dest + off + 0x18];					\
-	add		sum, 1, sum;							\
-34:	addcc		sum, t4, sum;							\
+	 stx		t3, [%dst + off + 0x18];					\
+	add		%sum, 1, %sum;							\
+34:	addcc		%sum, t4, %sum;							\
 	bcc,pt		%xcc, 35f;							\
-	 stx		t4, [dest + off + 0x20];					\
-	add		sum, 1, sum;							\
-35:	addcc		sum, t5, sum;							\
+	 stx		t4, [%dst + off + 0x20];					\
+	add		%sum, 1, %sum;							\
+35:	addcc		%sum, t5, %sum;							\
 	bcc,pt		%xcc, 36f;							\
-	 stx		t5, [dest + off + 0x28];					\
-	add		sum, 1, sum;							\
-36:	addcc		sum, t6, sum;							\
+	 stx		t5, [%dst + off + 0x28];					\
+	add		%sum, 1, %sum;							\
+36:	addcc		%sum, t6, %sum;							\
 	bcc,pt		%xcc, 37f;							\
-	 stx		t6, [dest + off + 0x30];					\
-	add		sum, 1, sum;							\
-37:	addcc		sum, t7, sum;							\
+	 stx		t6, [%dst + off + 0x30];					\
+	add		%sum, 1, %sum;							\
+37:	addcc		%sum, t7, %sum;							\
 	bcc,pt		%xcc, 38f;							\
-	 stx		t7, [dest + off + 0x38];					\
-	add		sum, 1, sum;							\
+	 stx		t7, [%dst + off + 0x38];					\
+	add		%sum, 1, %sum;							\
 38:
 
-#define CSUMCOPY_EC_STUNALIGN(dest, off, sum, t0, t1, t2, t3, t4, t5, t6, t7)		\
-	stw		t0, [dest + off + 0x04];					\
-	addcc		sum, t0, sum;							\
+#define CSUMCOPY_EC_STUNALIGN(off, t0, t1, t2, t3, t4, t5, t6, t7)			\
+	stw		t0, [%dst + off + 0x04];					\
+	addcc		%sum, t0, %sum;							\
 	srlx		t0, 32, t0;							\
 	bcc,pt		%xcc, 41f;							\
-	 stw		t0, [dest + off + 0x00];					\
-	add		sum, 1, sum;							\
-41:	stw		t1, [dest + off + 0x0c];					\
-	addcc		sum, t1, sum;							\
+	 stw		t0, [%dst + off + 0x00];					\
+	add		%sum, 1, %sum;							\
+41:	stw		t1, [%dst + off + 0x0c];					\
+	addcc		%sum, t1, %sum;							\
 	srlx		t1, 32, t1;							\
 	bcc,pt		%xcc, 42f;							\
-	 stw		t1, [dest + off + 0x08];					\
-	add		sum, 1, sum;							\
-42:	stw		t2, [dest + off + 0x14];					\
-	addcc		sum, t2, sum;							\
+	 stw		t1, [%dst + off + 0x08];					\
+	add		%sum, 1, %sum;							\
+42:	stw		t2, [%dst + off + 0x14];					\
+	addcc		%sum, t2, %sum;							\
 	srlx		t2, 32, t2;							\
 	bcc,pt		%xcc, 43f;							\
-	 stw		t2, [dest + off + 0x10];					\
-	add		sum, 1, sum;							\
-43:	stw		t3, [dest + off + 0x1c];					\
-	addcc		sum, t3, sum;							\
+	 stw		t2, [%dst + off + 0x10];					\
+	add		%sum, 1, %sum;							\
+43:	stw		t3, [%dst + off + 0x1c];					\
+	addcc		%sum, t3, %sum;							\
 	srlx		t3, 32, t3;							\
 	bcc,pt		%xcc, 44f;							\
-	 stw		t3, [dest + off + 0x18];					\
-	add		sum, 1, sum;							\
-44:	stw		t4, [dest + off + 0x24];					\
-	addcc		sum, t4, sum;							\
+	 stw		t3, [%dst + off + 0x18];					\
+	add		%sum, 1, %sum;							\
+44:	stw		t4, [%dst + off + 0x24];					\
+	addcc		%sum, t4, %sum;							\
 	srlx		t4, 32, t4;							\
 	bcc,pt		%xcc, 45f;							\
-	 stw		t4, [dest + off + 0x20];					\
-	add		sum, 1, sum;							\
-45:	stw		t5, [dest + off + 0x2c];					\
-	addcc		sum, t5, sum;							\
+	 stw		t4, [%dst + off + 0x20];					\
+	add		%sum, 1, %sum;							\
+45:	stw		t5, [%dst + off + 0x2c];					\
+	addcc		%sum, t5, %sum;							\
 	srlx		t5, 32, t5;							\
 	bcc,pt		%xcc, 46f;							\
-	 stw		t5, [dest + off + 0x28];					\
-	add		sum, 1, sum;							\
-46:	stw		t6, [dest + off + 0x34];					\
-	addcc		sum, t6, sum;							\
+	 stw		t5, [%dst + off + 0x28];					\
+	add		%sum, 1, %sum;							\
+46:	stw		t6, [%dst + off + 0x34];					\
+	addcc		%sum, t6, %sum;							\
 	srlx		t6, 32, t6;							\
 	bcc,pt		%xcc, 47f;							\
-	 stw		t6, [dest + off + 0x30];					\
-	add		sum, 1, sum;							\
-47:	stw		t7, [dest + off + 0x3c];					\
-	addcc		sum, t7, sum;							\
+	 stw		t6, [%dst + off + 0x30];					\
+	add		%sum, 1, %sum;							\
+47:	stw		t7, [%dst + off + 0x3c];					\
+	addcc		%sum, t7, %sum;							\
 	srlx		t7, 32, t7;							\
 	bcc,pt		%xcc, 48f;							\
-	 stw		t7, [dest + off + 0x38];					\
-	add		sum, 1, sum;							\
+	 stw		t7, [%dst + off + 0x38];					\
+	add		%sum, 1, %sum;							\
 48:
 
-#define CSUMCOPY_LASTCHUNK(src, dst, sum, off, t0, t1)					\
-	ldxa		[src - off - 0x08] %asi, t0;					\
-	ldxa		[src - off - 0x00] %asi, t1;					\
+#define CSUMCOPY_LASTCHUNK(off, t0, t1)							\
+	ldxa		[%src - off - 0x08] %asi, t0;					\
+	ldxa		[%src - off - 0x00] %asi, t1;					\
 	nop; nop;									\
-	addcc		t0, sum, sum;							\
-	stw		t0, [dst - off - 0x04];						\
+	addcc		t0, %sum, %sum;							\
+	stw		t0, [%dst - off - 0x04];					\
 	srlx		t0, 32, t0;							\
 	bcc,pt		%xcc, 51f;							\
-	 stw		t0, [dst - off - 0x08];						\
-	add		sum, 1, sum;							\
-51:	addcc		t1, sum, sum;							\
-	stw		t1, [dst - off + 0x04];						\
+	 stw		t0, [%dst - off - 0x08];					\
+	add		%sum, 1, %sum;							\
+51:	addcc		t1, %sum, %sum;							\
+	stw		t1, [%dst - off + 0x04];					\
 	srlx		t1, 32, t1;							\
 	bcc,pt		%xcc, 52f;							\
-	 stw		t1, [dst - off - 0x00];						\
-	add		sum, 1, sum;							\
+	 stw		t1, [%dst - off - 0x00];					\
+	add		%sum, 1, %sum;							\
 52:
 
+cpc_start:
 cc_end_cruft:
-	andcc		%o3, 8, %g0		! IEU1	Group
+	andcc		%g7, 8, %g0		! IEU1	Group
 	be,pn		%icc, 1f		! CTI
-	 and		%o3, 4, %g5		! IEU0
-	ldxa		[%o0 + 0x00] %asi, %g2	! Load	Group
-	add		%o1, 8, %o1		! IEU0
-	add		%o0, 8, %o0		! IEU1
-	addcc		%g2, %g7, %g7		! IEU1	Group + 2 bubbles
-	stw		%g2, [%o1 - 0x04]	! Store
+	 and		%g7, 4, %g5		! IEU0
+	ldxa		[%src + 0x00] %asi, %g2	! Load	Group
+	add		%dst, 8, %dst		! IEU0
+	add		%src, 8, %src		! IEU1
+	addcc		%g2, %sum, %sum		! IEU1	Group + 2 bubbles
+	stw		%g2, [%dst - 0x04]	! Store
 	srlx		%g2, 32, %g2		! IEU0
 	bcc,pt		%xcc, 1f		! CTI	Group
-	 stw		%g2, [%o1 - 0x08]	! Store
-	add		%g7, 1, %g7		! IEU0
+	 stw		%g2, [%dst - 0x08]	! Store
+	add		%sum, 1, %sum		! IEU0
 1:	brz,pt		%g5, 1f			! CTI	Group
 	 clr		%g2			! IEU0
-	lduwa		[%o0 + 0x00] %asi, %g2	! Load
-	add		%o1, 4, %o1		! IEU0	Group
-	add		%o0, 4, %o0		! IEU1
-	stw		%g2, [%o1 - 0x04]	! Store	Group + 2 bubbles
+	lduwa		[%src + 0x00] %asi, %g2	! Load
+	add		%dst, 4, %dst		! IEU0	Group
+	add		%src, 4, %src		! IEU1
+	stw		%g2, [%dst - 0x04]	! Store	Group + 2 bubbles
 	sllx		%g2, 32, %g2		! IEU0
-1:	andcc		%o3, 2, %g0		! IEU1
+1:	andcc		%g7, 2, %g0		! IEU1
 	be,pn		%icc, 1f		! CTI	Group
 	 clr		%o4			! IEU1
-	lduha		[%o0 + 0x00] %asi, %o4	! Load
-	add		%o0, 2, %o0		! IEU0	Group
-	add		%o1, 2, %o1		! IEU1
-	sth		%o4, [%o1 - 0x2]	! Store Group + 2 bubbles
+	lduha		[%src + 0x00] %asi, %o4	! Load
+	add		%src, 2, %src		! IEU0	Group
+	add		%dst, 2, %dst		! IEU1
+	sth		%o4, [%dst - 0x2]	! Store Group + 2 bubbles
 	sll		%o4, 16, %o4		! IEU0
-1:	andcc		%o3, 1, %g0		! IEU1
+1:	andcc		%g7, 1, %g0		! IEU1
 	be,pn		%icc, 1f		! CTI	Group
 	 clr		%o5			! IEU0
-	lduba		[%o0 + 0x00] %asi, %o5	! Load
-	stb		%o5, [%o1 + 0x00]	! Store	Group + 2 bubbles
+	lduba		[%src + 0x00] %asi, %o5	! Load
+	stb		%o5, [%dst + 0x00]	! Store	Group + 2 bubbles
 	sll		%o5, 8, %o5		! IEU0
 1:	or		%g2, %o4, %o4		! IEU1
 	or		%o5, %o4, %o4		! IEU0	Group
-	addcc		%o4, %g7, %g7		! IEU1
+	addcc		%o4, %sum, %sum		! IEU1
 	bcc,pt		%xcc, ccfold		! CTI
 	 sethi		%uhi(PAGE_OFFSET), %g4	! IEU0	Group
 	b,pt		%xcc, ccfold		! CTI
-	 add		%g7, 1, %g7		! IEU1
+	 add		%sum, 1, %sum		! IEU1
 
 cc_fixit:
 	bl,a,pn		%icc, ccte		! CTI
-	 andcc		%g1, 0xf, %o3		! IEU1	Group
-	andcc		%o0, 1, %g0		! IEU1	Group
-	bne,pn		%icc, ccslow		! CTI
-	 andcc		%o0, 2, %g0		! IEU1	Group
+	 andcc		%len, 0xf, %g7		! IEU1	Group
+	andcc		%src, 2, %g0		! IEU1	Group
 	be,pn		%icc, 1f		! CTI
-	 andcc		%o0, 0x4, %g0		! IEU1	Group
-	lduha		[%o0 + 0x00] %asi, %g4	! Load
-	sub		%g1, 2, %g1		! IEU0
-	add		%o0, 2, %o0		! IEU0	Group
-	add		%o1, 2, %o1		! IEU1
+	 andcc		%src, 0x4, %g0		! IEU1	Group
+	lduha		[%src + 0x00] %asi, %g4	! Load
+	sub		%len, 2, %len		! IEU0
+	add		%src, 2, %src		! IEU0	Group
+	add		%dst, 2, %dst		! IEU1
 	sll		%g4, 16, %g3		! IEU0	Group + 1 bubble
-	addcc		%g3, %g7, %g7		! IEU1
+	addcc		%g3, %sum, %sum		! IEU1
 	bcc,pt		%xcc, 0f		! CTI
-	 srl		%g7, 16, %g3		! IEU0	Group
+	 srl		%sum, 16, %g3		! IEU0	Group
 	add		%g3, 1, %g3		! IEU0	4 clocks (mispredict)
-0:	andcc		%o0, 0x4, %g0		! IEU1	Group
-	sth		%g4, [%o1 - 0x2]	! Store
-	sll		%g7, 16, %g7		! IEU0
+0:	andcc		%src, 0x4, %g0		! IEU1	Group
+	sth		%g4, [%dst - 0x2]	! Store
+	sll		%sum, 16, %sum		! IEU0
 	sll		%g3, 16, %g3		! IEU0	Group
-	srl		%g7, 16, %g7		! IEU0	Group
-	or		%g3, %g7, %g7		! IEU0	Group (regdep)
+	srl		%sum, 16, %sum		! IEU0	Group
+	or		%g3, %sum, %sum		! IEU0	Group (regdep)
 1:	be,pt		%icc, cc_dword_aligned	! CTI
-	 andn		%g1, 0xff, %g2		! IEU1
-	lduwa		[%o0 + 0x00] %asi, %g4	! Load	Group
-	sub		%g1, 4, %g1		! IEU0
-	add		%o0, 4, %o0		! IEU1
-	add		%o1, 4, %o1		! IEU0	Group
-	addcc		%g4, %g7, %g7		! IEU1	Group + 1 bubble
-	stw		%g4, [%o1 - 0x4]	! Store
+	 andn		%len, 0xff, %g2		! IEU1
+	lduwa		[%src + 0x00] %asi, %g4	! Load	Group
+	sub		%len, 4, %len		! IEU0
+	add		%src, 4, %src		! IEU1
+	add		%dst, 4, %dst		! IEU0	Group
+	addcc		%g4, %sum, %sum		! IEU1	Group + 1 bubble
+	stw		%g4, [%dst - 0x4]	! Store
 	bcc,pt		%xcc, cc_dword_aligned	! CTI
-	 andn		%g1, 0xff, %g2		! IEU0	Group
+	 andn		%len, 0xff, %g2		! IEU0	Group
 	b,pt		%xcc, cc_dword_aligned	! CTI	4 clocks (mispredict)
-	 add		%g7, 1, %g7		! IEU0
+	 add		%sum, 1, %sum		! IEU0
 
 	.align		32
-	.globl		__csum_partial_copy_sparc_generic, csum_partial_copy
-csum_partial_copy:
-__csum_partial_copy_sparc_generic:		/* %o0=src, %o1=dest, %g1=len, %g7=sum */
-	xorcc		%o0, %o1, %o4		! IEU1	Group
-	srl		%g7, 0, %g7		! IEU0
+	.globl		csum_partial_copy_sparc64
+csum_partial_copy_sparc64:			/* %o0=src, %o1=dest, %o2=len, %o3=sum */
+	xorcc		%src, %dst, %o4		! IEU1	Group
+	srl		%sum, 0, %sum		! IEU0
 	andcc		%o4, 3, %g0		! IEU1	Group
-	srl		%g1, 0, %g1		! IEU0
+	srl		%len, 0, %len		! IEU0
+	bne,pn		%icc, ccslow		! CTI
+	 andcc		%src, 1, %g0		! IEU1	Group
 	bne,pn		%icc, ccslow		! CTI
-	 andcc		%o0, 7, %g0		! IEU1	Group
+	 cmp		%len, 256		! IEU1	Group
+	bgeu,pt		%icc, csum_partial_copy_vis ! CTI
+	 andcc		%src, 7, %g0		! IEU1	Group
 	be,pt		%icc, cc_dword_aligned	! CTI
-	 andn		%g1, 0xff, %g2		! IEU0
+	 andn		%len, 0xff, %g2		! IEU0
 	b,pt		%xcc, cc_fixit		! CTI	Group
-	 cmp		%g1, 6			! IEU1
+	 cmp		%len, 6			! IEU1
 cc_dword_aligned:
 	brz,pn		%g2, 3f			! CTI	Group
-	 andcc		%o1, 4, %g0		! IEU1	Group (brz uses IEU1)
+	 andcc		%dst, 4, %g0		! IEU1	Group (brz uses IEU1)
 	be,pn		%icc, ccdbl + 4		! CTI
-5:	CSUMCOPY_ECACHE_LOAD(       %o0,    0x00,    %o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
-	CSUMCOPY_EC_STUNALIGN_LDNXT(%o0,%o1,0x40,%g7,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
-	CSUMCOPY_EC_STUNALIGN_LDNXT(%o0,%o1,0x80,%g7,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
-	CSUMCOPY_EC_STUNALIGN_LDNXT(%o0,%o1,0xc0,%g7,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
-	CSUMCOPY_EC_STUNALIGN(          %o1,0xc0,%g7,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
+5:	CSUMCOPY_ECACHE_LOAD(       0x00,%o4,%o5,%g2,%g3,%g4,%g5,%g1,%g7)
+	CSUMCOPY_EC_STUNALIGN_LDNXT(0x40,%o4,%o5,%g2,%g3,%g4,%g5,%g1,%g7)
+	CSUMCOPY_EC_STUNALIGN_LDNXT(0x80,%o4,%o5,%g2,%g3,%g4,%g5,%g1,%g7)
+	CSUMCOPY_EC_STUNALIGN_LDNXT(0xc0,%o4,%o5,%g2,%g3,%g4,%g5,%g1,%g7)
+	CSUMCOPY_EC_STUNALIGN(      0xc0,%o4,%o5,%g2,%g3,%g4,%g5,%g1,%g7)
 10:
-	sub		%g1, 256, %g1		! IEU0	Group
-	add		%o0, 256, %o0		! IEU1
-	andncc		%g1, 0xff, %g0		! IEU1	Group
+	sub		%len, 256, %len		! IEU0	Group
+	add		%src, 256, %src		! IEU1
+	andncc		%len, 0xff, %g0		! IEU1	Group
 	bne,pt		%icc, 5b		! CTI
-	 add		%o1, 256, %o1		! IEU0
-3:	andcc		%g1, 0xf0, %o2		! IEU1	Group
+	 add		%dst, 256, %dst		! IEU0
+3:	andcc		%len, 0xf0, %g1		! IEU1	Group
 ccmerge:be,pn		%icc, ccte		! CTI
-	 andcc		%g1, 0xf, %o3		! IEU1	Group
-	sll		%o2, 2, %o4		! IEU0
-13:	rd		%pc, %o5		! LSU	Group + 4 clocks
-	add		%o0, %o2, %o0		! IEU0	Group
-	sub		%o5, %o4, %o5		! IEU1	Group
-	jmpl		%o5 + (12f - 13b), %g0	! CTI	Group brk forced
-	 add		%o1, %o2, %o1		! IEU0	Group
-cctbl:	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0xe8,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0xd8,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0xc8,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0xb8,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0xa8,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x98,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x88,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x78,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x68,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x58,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x48,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x38,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x28,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x18,%g2,%g3)
-	CSUMCOPY_LASTCHUNK(%o0,%o1,%g7,0x08,%g2,%g3)
+	 andcc		%len, 0xf, %g7		! IEU1	Group
+	sll		%g1, 2, %o4		! IEU0
+13:	sethi		%hi(12f), %o5		! IEU0	Group
+	add		%src, %g1, %src		! IEU1	
+	sub		%o5, %o4, %o5		! IEU0	Group
+	jmpl		%o5 + %lo(12f), %g0	! CTI	Group brk forced
+	 add		%dst, %g1, %dst		! IEU0	Group
+cctbl:	CSUMCOPY_LASTCHUNK(0xe8,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0xd8,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0xc8,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0xb8,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0xa8,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0x98,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0x88,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0x78,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0x68,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0x58,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0x48,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0x38,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0x28,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0x18,%g2,%g3)
+	CSUMCOPY_LASTCHUNK(0x08,%g2,%g3)
 12:
-	andcc		%g1, 0xf, %o3		! IEU1	Group
+	andcc		%len, 0xf, %g7		! IEU1	Group
 ccte:	bne,pn		%icc, cc_end_cruft	! CTI
 	 sethi		%uhi(PAGE_OFFSET), %g4	! IEU0
-ccfold:	sllx		%g7, 32, %o0		! IEU0	Group
-	addcc		%g7, %o0, %o0		! IEU1	Group (regdep)
+ccfold:	sllx		%sum, 32, %o0		! IEU0	Group
+	addcc		%sum, %o0, %o0		! IEU1	Group (regdep)
 	srlx		%o0, 32, %o0		! IEU0	Group (regdep)
 	bcs,a,pn	%xcc, 1f		! CTI
 	 add		%o0, 1, %o0		! IEU1	4 clocks (mispredict)
 1:	retl					! CTI	Group brk forced
 	 sllx		%g4, 32,%g4		! IEU0	Group
-ccdbl:	CSUMCOPY_ECACHE_LOAD(       %o0,    0x00,    %o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
-	CSUMCOPY_EC_STALIGNED_LDNXT(%o0,%o1,0x40,%g7,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
-	CSUMCOPY_EC_STALIGNED_LDNXT(%o0,%o1,0x80,%g7,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
-	CSUMCOPY_EC_STALIGNED_LDNXT(%o0,%o1,0xc0,%g7,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
-	CSUMCOPY_EC_STALIGNED(          %o1,0xc0,%g7,%o4,%o5,%g2,%g3,%g4,%g5,%o2,%o3)
+ccdbl:	CSUMCOPY_ECACHE_LOAD(       0x00,%o4,%o5,%g2,%g3,%g4,%g5,%g1,%g7)
+	CSUMCOPY_EC_STALIGNED_LDNXT(0x40,%o4,%o5,%g2,%g3,%g4,%g5,%g1,%g7)
+	CSUMCOPY_EC_STALIGNED_LDNXT(0x80,%o4,%o5,%g2,%g3,%g4,%g5,%g1,%g7)
+	CSUMCOPY_EC_STALIGNED_LDNXT(0xc0,%o4,%o5,%g2,%g3,%g4,%g5,%g1,%g7)
+	CSUMCOPY_EC_STALIGNED(      0xc0,%o4,%o5,%g2,%g3,%g4,%g5,%g1,%g7)
 11:
-	sub		%g1, 256, %g1		! IEU0	Group
-	add		%o0, 256, %o0		! IEU1
-	andncc		%g1, 0xff, %g0		! IEU1	Group	
+	sub		%len, 256, %len		! IEU0	Group
+	add		%src, 256, %src		! IEU1
+	andncc		%len, 0xff, %g0		! IEU1	Group	
 	bne,pt		%icc, ccdbl		! CTI
-	 add		%o1, 256, %o1		! IEU0
+	 add		%dst, 256, %dst		! IEU0
 	b,pt		%xcc, ccmerge		! CTI	Group
-	 andcc		%g1, 0xf0, %o2		! IEU1
+	 andcc		%len, 0xf0, %g1		! IEU1
 
 ccslow:	mov	0, %g5
-	brlez,pn %g1, 4f
-	 andcc	%o0, 1, %o5		
+	brlez,pn %len, 4f
+	 andcc	%src, 1, %o5		
 	be,a,pt	%icc, 1f
-	 srl	%g1, 1, %o3		
-	sub	%g1, 1, %g1	
-	lduba [%o0] %asi, %g5
-	add	%o0, 1, %o0	
-	stb	%g5, [%o1]
-	srl	%g1, 1, %o3
-	add	%o1, 1, %o1
-1:	brz,a,pn %o3, 3f
-	 andcc	%g1, 1, %g0
-	andcc	%o0, 2, %g0	
+	 srl	%len, 1, %g7		
+	sub	%len, 1, %len	
+	lduba [%src] %asi, %g5
+	add	%src, 1, %src	
+	stb	%g5, [%dst]
+	srl	%len, 1, %g7
+	add	%dst, 1, %dst
+1:	brz,a,pn %g7, 3f
+	 andcc	%len, 1, %g0
+	andcc	%src, 2, %g0	
 	be,a,pt	%icc, 1f
-	 srl	%o3, 1, %o3
-	lduha [%o0] %asi, %o4
-	sub	%g1, 2, %g1	
+	 srl	%g7, 1, %g7
+	lduha [%src] %asi, %o4
+	sub	%len, 2, %len	
 	srl	%o4, 8, %g2
-	sub	%o3, 1, %o3	
-	stb	%g2, [%o1]
+	sub	%g7, 1, %g7	
+	stb	%g2, [%dst]
 	add	%o4, %g5, %g5
-	stb	%o4, [%o1 + 1]
-	add	%o0, 2, %o0	
-	srl	%o3, 1, %o3
-	add	%o1, 2, %o1
-1:	brz,a,pn %o3, 2f		
-	 andcc	%g1, 2, %g0
-	lda	[%o0] %asi, %o4
+	stb	%o4, [%dst + 1]
+	add	%src, 2, %src	
+	srl	%g7, 1, %g7
+	add	%dst, 2, %dst
+1:	brz,a,pn %g7, 2f		
+	 andcc	%len, 2, %g0
+	lduwa	[%src] %asi, %o4
 5:	srl	%o4, 24, %g2
 	srl	%o4, 16, %g3
-	stb	%g2, [%o1]
+	stb	%g2, [%dst]
 	srl	%o4, 8, %g2
-	stb	%g3, [%o1 + 1]
-	add	%o0, 4, %o0
-	stb	%g2, [%o1 + 2]
+	stb	%g3, [%dst + 1]
+	add	%src, 4, %src
+	stb	%g2, [%dst + 2]
 	addcc	%o4, %g5, %g5
-	stb	%o4, [%o1 + 3]
-	addc	%g5, %g0, %g5	! I am now to lazy to optimize this (question is if it
-	add	%o1, 4, %o1	! is worthy). Maybe some day - with the sll/srl
-	subcc	%o3, 1, %o3	! tricks
+	stb	%o4, [%dst + 3]
+	addc	%g5, %g0, %g5
+	add	%dst, 4, %dst
+	subcc	%g7, 1, %g7
 	bne,a,pt %icc, 5b
-	 lda [%o0] %asi, %o4
+	 lduwa [%src] %asi, %o4
 	sll	%g5, 16, %g2
 	srl	%g5, 16, %g5
 	srl	%g2, 16, %g2
-	andcc	%g1, 2, %g0
+	andcc	%len, 2, %g0
 	add	%g2, %g5, %g5 
 2:	be,a,pt	%icc, 3f		
-	 andcc	%g1, 1, %g0
-	lduha [%o0] %asi, %o4
-	andcc	%g1, 1, %g0
+	 andcc	%len, 1, %g0
+	lduha [%src] %asi, %o4
+	andcc	%len, 1, %g0
 	srl	%o4, 8, %g2
-	add	%o0, 2, %o0	
-	stb	%g2, [%o1]
+	add	%src, 2, %src	
+	stb	%g2, [%dst]
 	add	%g5, %o4, %g5
-	stb	%o4, [%o1 + 1]
-	add	%o1, 2, %o1
+	stb	%o4, [%dst + 1]
+	add	%dst, 2, %dst
 3:	be,a,pt	%icc, 1f		
 	 sll	%g5, 16, %o4
-	lduba [%o0] %asi, %g2
+	lduba [%src] %asi, %g2
 	sll	%g2, 8, %o4	
-	stb	%g2, [%o1]
+	stb	%g2, [%dst]
 	add	%g5, %o4, %g5
 	sll	%g5, 16, %o4
 1:	addcc	%o4, %g5, %g5
@@ -484,8 +484,22 @@ ccslow:	mov	0, %g5
 	and	%o4, 0xff, %o4
 	sll	%g2, 8, %g2
 	or	%g2, %o4, %g5
-4:	addcc	%g7, %g5, %g7
-	addc	%g0, %g7, %o0
+4:	addcc	%sum, %g5, %sum
+	addc	%g0, %sum, %o0
 	retl	
 	 srl	%o0, 0, %o0
-__csum_partial_copy_end:
+cpc_end:
+
+	.globl	cpc_handler
+cpc_handler:
+	ldx	[%sp + 0x7ff + 128], %g1
+	sub	%g0, EFAULT, %g2
+	brnz,a,pt %g1, 1f
+	 st	%g2, [%g1]
+1:	retl
+	 nop
+
+	.section __ex_table
+	.align  4
+	.word  cpc_start, 0, cpc_end, cpc_handler
+
diff --git a/arch/sparc64/lib/locks.S b/arch/sparc64/lib/locks.S
index 74054f63a..7b2bdba62 100644
--- a/arch/sparc64/lib/locks.S
+++ b/arch/sparc64/lib/locks.S
@@ -1,4 +1,4 @@
-/* $Id: locks.S,v 1.3 1997/07/22 05:51:42 davem Exp $
+/* $Id: locks.S,v 1.5 1997/07/31 05:28:16 davem Exp $
  * locks.S: SMP low-level lock primitives on Sparc64.
  *
  * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
@@ -12,9 +12,9 @@
 	.align		32
 
 ___lk_busy_spin:
-	orcc		%g2, 0, %g0
-	bne,pt		%icc, ___lk_busy_spin
-	 ldub		[%g1 + 0], %g2
+	ldub		[%g1 + 0], %g2
+	brnz,pt		%g2, ___lk_busy_spin
+	 membar		#LoadLoad
 	b,pt		%xcc, 1f
 	 ldstub		[%g1 + 0], %g2
 
@@ -23,16 +23,15 @@ ___lock_kernel:
 	addcc		%g2, -1, %g2
 	rdpr		%pil, %g3
 	bcs,a,pn	%icc, 9f
-	 st		%g2, [%g6 + AOFF_task_lock_depth]
+	 stw		%g2, [%g6 + AOFF_task_lock_depth]
 	wrpr		%g0, 15, %pil
 	ldstub		[%g1 + 0], %g2
-1:	brnz,a,pn	%g2, ___lk_busy_spin
-	 ldub		[%g1 + 0], %g2
+1:	brnz,pn		%g2, ___lk_busy_spin
+	 membar		#StoreLoad | #StoreStore
 	lduw		[%g6 + AOFF_task_processor], %g2
-	membar		#LoadLoad | #LoadStore
 	stb		%g2, [%g1 + 1]
 2:	mov		-1, %g2
-	st		%g2, [%g6 + AOFF_task_lock_depth]
+	stw		%g2, [%g6 + AOFF_task_lock_depth]
 	wrpr		%g3, 0, %pil
 9:	jmpl		%o7 + 0x8, %g0
 	 mov		%g5, %o7
@@ -41,16 +40,16 @@ ___lock_kernel:
 ___lock_reacquire_kernel:
 	rdpr		%pil, %g3
 	wrpr		%g0, 15, %pil
-	st		%g2, [%g6 + AOFF_task_lock_depth]
+	stw		%g2, [%g6 + AOFF_task_lock_depth]
 	ldstub		[%g1 + 0], %g2
 1:	brz,pt		%g2, 3f
-	 ldub		[%g1 + 0], %g2
-2:	brnz,a,pt	%g2, 2b
-	 ldub		[%g1 + 0], %g2
+	 membar		#StoreLoad | #StoreStore
+2:	ldub		[%g1 + 0], %g2
+	brnz,pt		%g2, 2b
+	 membar		#LoadLoad
 	b,pt		%xcc, 1b
 	 ldstub		[%g1 + 0], %g2
 3:	lduw		[%g6 + AOFF_task_processor], %g2
-	membar		#LoadLoad | #LoadStore
 	stb		%g2, [%g1 + 1]
 	wrpr		%g3, 0, %pil
 	jmpl		%o7 + 0x8, %g0
diff --git a/arch/sparc64/lib/strlen_user.S b/arch/sparc64/lib/strlen_user.S
index ef6cee5a6..9ae3d0381 100644
--- a/arch/sparc64/lib/strlen_user.S
+++ b/arch/sparc64/lib/strlen_user.S
@@ -78,11 +78,11 @@ __strlen_user:
          clr    %o0
 
 	.section __ex_table,#alloc
-	.align	8
+	.align	4
 
-	.xword	10b, 30b
-	.xword	11b, 30b
-	.xword	12b, 30b
-	.xword	15b, 30b
-	.xword	13b, 30b
-	.xword	14b, 30b
+	.word	10b, 30b
+	.word	11b, 30b
+	.word	12b, 30b
+	.word	15b, 30b
+	.word	13b, 30b
+	.word	14b, 30b
diff --git a/arch/sparc64/lib/strncpy_from_user.S b/arch/sparc64/lib/strncpy_from_user.S
index 7a5dc768f..e0fb0f09b 100644
--- a/arch/sparc64/lib/strncpy_from_user.S
+++ b/arch/sparc64/lib/strncpy_from_user.S
@@ -49,6 +49,6 @@ __strncpy_from_user:
 	 mov	-EFAULT, %o0
 
 	.section __ex_table,#alloc
-	.align	8
-	.xword	10b, 4b
-	.xword	11b, 4b
+	.align	4
+	.word	10b, 4b
+	.word	11b, 4b
diff --git a/arch/sparc64/mm/fault.c b/arch/sparc64/mm/fault.c
index 5e16e1218..6bc52f3eb 100644
--- a/arch/sparc64/mm/fault.c
+++ b/arch/sparc64/mm/fault.c
@@ -1,4 +1,4 @@
-/* $Id: fault.c,v 1.18 1997/07/17 02:20:56 davem Exp $
+/* $Id: fault.c,v 1.20 1997/08/04 16:16:51 davem Exp $
  * arch/sparc64/mm/fault.c: Page fault handlers for the 64-bit Sparc.
  *
  * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
@@ -217,56 +217,3 @@ bad_area:
 out:
 	unlock_kernel();
 }
-
-void fixup_dcache_alias(struct vm_area_struct *vma, unsigned long address, pte_t pte)
-{
-	struct vm_area_struct *vmaring;
-	struct inode *inode;
-	unsigned long vaddr, offset, start;
-	pgd_t *pgdp;
-	pmd_t *pmdp;
-	pte_t *ptep;
-	int alias_found = 0;
-
-	inode = vma->vm_dentry->d_inode;
-	if(!inode)
-		return;
-
-	offset = (address & PAGE_MASK) - vma->vm_start;
-	vmaring = inode->i_mmap;
-	do {
-		vaddr = vmaring->vm_start + offset;
-
-		/* This conditional is misleading... */
-		if((vaddr ^ address) & PAGE_SIZE) {
-			alias_found++;
-			start = vmaring->vm_start;
-			while(start < vmaring->vm_end) {
-				pgdp = pgd_offset(vmaring->vm_mm, start);
-				if(!pgdp) goto next;
-				pmdp = pmd_offset(pgdp, start);
-				if(!pmdp) goto next;
-				ptep = pte_offset(pmdp, start);
-				if(!ptep) goto next;
-
-				if(pte_val(*ptep) & _PAGE_PRESENT) {
-					flush_cache_page(vmaring, start);
-					*ptep = __pte(pte_val(*ptep) &
-						      ~(_PAGE_CV));
-					flush_tlb_page(vmaring, start);
-				}
-			next:
-				start += PAGE_SIZE;
-			}
-		}
-	} while((vmaring = vmaring->vm_next_share) != NULL);
-
-	if(alias_found && (pte_val(pte) & _PAGE_CV)) {
-		pgdp = pgd_offset(vma->vm_mm, address);
-		pmdp = pmd_offset(pgdp, address);
-		ptep = pte_offset(pmdp, address);
-		flush_cache_page(vma, address);
-		*ptep = __pte(pte_val(*ptep) & ~(_PAGE_CV));
-		flush_tlb_page(vma, address);
-	}
-}
diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c
index 386c8540f..4a7a1bce1 100644
--- a/arch/sparc64/mm/init.c
+++ b/arch/sparc64/mm/init.c
@@ -1,4 +1,4 @@
-/*  $Id: init.c,v 1.40 1997/07/24 16:48:27 davem Exp $
+/*  $Id: init.c,v 1.55 1997/08/24 01:22:29 davem Exp $
  *  arch/sparc64/mm/init.c
  *
  *  Copyright (C) 1996,1997 David S. Miller (davem@caip.rutgers.edu)
@@ -37,6 +37,18 @@ unsigned long tlb_context_cache = CTX_FIRST_VERSION;
 /* References to section boundaries */
 extern char __init_begin, __init_end, etext, __bss_start;
 
+extern void __bfill64(void *, unsigned long *);
+
+static __inline__ void __init_pmd(pmd_t *pmdp)
+{
+	__bfill64((void *)pmdp, &null_pte_table);
+}
+
+static __inline__ void __init_pgd(pgd_t *pgdp)
+{
+	__bfill64((void *)pgdp, &null_pmd_table);
+}
+
 /*
  * BAD_PAGE is the page that is used for page faults when linux
  * is out-of-memory. Older versions of linux just did a
@@ -103,48 +115,72 @@ void show_mem(void)
 
 /* IOMMU support, the ideas are right, the code should be cleaned a bit still... */
 
-/* XXX Also, play with the streaming buffers at some point, both
- * XXX Fusion and Sunfire both have them aparently... -DaveM
- */
-
 /* This keeps track of pages used in sparc_alloc_dvma() invocations. */
 static unsigned long dvma_map_pages[0x10000000 >> 16] = { 0, };
 static unsigned long dvma_pages_current_offset = 0;
 static int dvma_pages_current_index = 0;
 
+/* #define E3000_DEBUG */
+
 __initfunc(unsigned long iommu_init(int iommu_node, unsigned long memory_start,
 				    unsigned long memory_end, struct linux_sbus *sbus))
 {
 	struct iommu_struct *iommu;
 	struct sysio_regs *sregs;
-	struct linux_prom_registers rprop[2];
+	struct linux_prom64_registers rprop;
 	unsigned long impl, vers;
 	unsigned long control, tsbbase;
 	unsigned long *iopte;
+	u32 rlow, rhigh;
 	int err, i;
 
-	err = prom_getproperty(iommu_node, "reg", (char *)rprop,
+#ifdef E3000_DEBUG
+	prom_printf("\niommu_init: [%x:%016lx:%016lx:%p] ",
+		    iommu_node, memory_start, memory_end, sbus);
+#endif
+	err = prom_getproperty(iommu_node, "reg", (char *)&rprop,
 			       sizeof(rprop));
 	if(err == -1) {
 		prom_printf("iommu_init: Cannot map SYSIO control registers.\n");
 		prom_halt();
 	}
-	sregs = (struct sysio_regs *) sparc_alloc_io(rprop[0].phys_addr,
-						     (void *)0,
+	rlow  = (rprop.phys_addr & 0xffffffff);
+	rhigh = (rprop.phys_addr >> 32);
+#ifdef E3000_DEBUG
+	prom_printf("rlow[%08x] rhigh[%08x] ", rlow, rhigh);
+#endif
+	sregs = (struct sysio_regs *) sparc_alloc_io(rlow, (void *)0,
 						     sizeof(struct sysio_regs),
-						     "SYSIO Regs",
-						     rprop[0].which_io, 0x0);
+						     "SYSIO Regs", rhigh, 0x0);
+#ifdef E3000_DEBUG
+	prom_printf("sregs[%p]\n");
+#endif
+	if(!sregs) {
+		prom_printf("iommu_init: Fatal error, sysio regs not mapped\n");
+		prom_halt();
+	}
 
 	memory_start = (memory_start + 7) & ~7;
 	iommu = (struct iommu_struct *) memory_start;
 	memory_start += sizeof(struct iommu_struct);
+
+#ifdef E3000_DEBUG
+	prom_printf("iommu_init: iommu[%p] ", iommu);
+#endif
+
+	spin_lock_init(&iommu->iommu_lock);
 	iommu->sysio_regs = sregs;
 	sbus->iommu = iommu;
 
 	control = sregs->iommu_control;
 	impl = (control & IOMMU_CTRL_IMPL) >> 60;
 	vers = (control & IOMMU_CTRL_VERS) >> 56;
-	printk("IOMMU: IMPL[%x] VERS[%x] SYSIO mapped at %016lx\n",
+#ifdef E3000_DEBUG
+	prom_printf("sreg_control[%08x]\n", control);
+	prom_printf("IOMMU: IMPL[%x] VERS[%x] SYSIO mapped at %016lx\n",
+		    (unsigned int) impl, (unsigned int)vers, (unsigned long) sregs);
+#endif
+	printk("IOMMU(SBUS): IMPL[%x] VERS[%x] SYSIO mapped at %016lx\n",
 	       (unsigned int) impl, (unsigned int)vers, (unsigned long) sregs);
 	
 	control &= ~(IOMMU_CTRL_TSBSZ);
@@ -168,7 +204,8 @@ __initfunc(unsigned long iommu_init(int iommu_node, unsigned long memory_start,
 
 	/* Setup aliased mappings... */
 	for(i = 0; i < (65536 - 4096); i++) {
-		*iopte  = (IOPTE_VALID | IOPTE_64K | IOPTE_CACHE | IOPTE_WRITE);
+		*iopte  = (IOPTE_VALID | IOPTE_64K | IOPTE_STBUF |
+			   IOPTE_CACHE | IOPTE_WRITE);
 		*iopte |= (i << 16);
 		iopte++;
 	}
@@ -177,15 +214,47 @@ __initfunc(unsigned long iommu_init(int iommu_node, unsigned long memory_start,
 	for( ; i < 65536; i++)
 		*iopte++ = 0;
 
+#ifdef E3000_DEBUG
+	prom_printf("IOMMU: pte's mapped, enabling IOMMU... ");
+#endif
 	sregs->iommu_tsbbase = __pa(tsbbase);
 	sregs->iommu_control = control;
 
+#ifdef E3000_DEBUG
+	prom_printf("done\n");
+#endif
+	/* Get the streaming buffer going. */
+	control = sregs->sbuf_control;
+	impl = (control & SYSIO_SBUFCTRL_IMPL) >> 60;
+	vers = (control & SYSIO_SBUFCTRL_REV) >> 56;
+#ifdef E3000_DEBUG
+	prom_printf("IOMMU: enabling streaming buffer, control[%08x]... ",
+		    control);
+#endif
+	printk("IOMMU: Streaming Buffer IMPL[%x] REV[%x] ",
+	       (unsigned int)impl, (unsigned int)vers);
+	printk("FlushFLAG[%p,%016lx] ... ",
+	       (iommu->sbuf_flushflag_va = (unsigned int *)memory_start),
+	       (iommu->sbuf_flushflag_pa = __pa(memory_start)));
+	*(iommu->sbuf_flushflag_va) = 0;
+	memory_start += sizeof(unsigned long); /* yes, unsigned long, for alignment */
+
+	sregs->sbuf_control = (control | SYSIO_SBUFCTRL_SB_EN);
+
+#ifdef E3000_DEBUG
+	prom_printf("done, returning %016lx\n", memory_start);
+#endif
+	printk("ENABLED\n");
+
+	/* Finally enable DVMA arbitration for all devices, just in case. */
+	sregs->sbus_control |= SYSIO_SBCNTRL_AEN;
+
 	return memory_start;
 }
 
-void mmu_map_dma_area(unsigned long addr, int len, __u32 *dvma_addr)
+void mmu_map_dma_area(unsigned long addr, int len, __u32 *dvma_addr,
+		      struct linux_sbus *sbus)
 {
-	struct iommu_struct *iommu = SBus_chain->iommu; /* GROSS ME OUT! */
 	pgd_t *pgdp;
 	pmd_t *pmdp;
 	pte_t *ptep;
@@ -193,6 +262,7 @@ void mmu_map_dma_area(unsigned long addr, int len, __u32 *dvma_addr)
 	/* Find out if we need to grab some pages. */
 	if(!dvma_map_pages[dvma_pages_current_index] ||
 	   ((dvma_pages_current_offset + len) > (1 << 16))) {
+		struct linux_sbus *sbus;
 		unsigned long *iopte;
 		unsigned long newpages = __get_free_pages(GFP_KERNEL, 3, 0);
 		int i;
@@ -212,9 +282,16 @@ void mmu_map_dma_area(unsigned long addr, int len, __u32 *dvma_addr)
 
 		/* Stick it in the IOMMU. */
 		i = (65536 - 4096) + i;
-		iopte = (unsigned long *)(iommu->page_table + i);
-		*iopte  = (IOPTE_VALID | IOPTE_64K | IOPTE_CACHE | IOPTE_WRITE);
-		*iopte |= __pa(newpages);
+		for_each_sbus(sbus) {
+			struct iommu_struct *iommu = sbus->iommu;
+			unsigned long flags;
+
+			spin_lock_irqsave(&iommu->iommu_lock, flags);
+			iopte = (unsigned long *)(iommu->page_table + i);
+			*iopte  = (IOPTE_VALID | IOPTE_64K | IOPTE_CACHE | IOPTE_WRITE);
+			*iopte |= __pa(newpages);
+			spin_unlock_irqrestore(&iommu->iommu_lock, flags);
+		}
 	}
 
 	/* Get this out of the way. */
@@ -258,6 +335,33 @@ __u32 mmu_get_scsi_one(char *vaddr, unsigned long len, struct linux_sbus *sbus)
         return (__u32)0;
 }
 
+void mmu_release_scsi_one(u32 vaddr, unsigned long len, struct linux_sbus *sbus)
+{
+	struct iommu_struct *iommu = sbus->iommu;
+	struct sysio_regs *sregs = iommu->sysio_regs;
+	unsigned long start = (unsigned long) vaddr;
+	unsigned long end = PAGE_ALIGN(start + len);
+	unsigned long flags;
+	unsigned int *sync_word;
+
+	start &= PAGE_MASK;
+
+	spin_lock_irqsave(&iommu->iommu_lock, flags);
+
+	while(start < end) {
+		sregs->sbuf_pflush = start;
+		start += PAGE_SIZE;
+	}
+	sync_word = iommu->sbuf_flushflag_va;
+	sregs->sbuf_fsync = iommu->sbuf_flushflag_pa;
+	membar("#StoreLoad | #MemIssue");
+	while((*sync_word & 0x1) == 0)
+		membar("#LoadLoad");
+	*sync_word = 0;
+
+	spin_unlock_irqrestore(&iommu->iommu_lock, flags);
+}
+
 void mmu_get_scsi_sgl(struct mmu_sglist *sg, int sz, struct linux_sbus *sbus)
 {
 	while(sz >= 0) {
@@ -273,6 +377,36 @@ void mmu_get_scsi_sgl(struct mmu_sglist *sg, int sz, struct linux_sbus *sbus)
 	}
 }
 
+void mmu_release_scsi_sgl(struct mmu_sglist *sg, int sz, struct linux_sbus *sbus)
+{
+	struct iommu_struct *iommu = sbus->iommu;
+	struct sysio_regs *sregs = iommu->sysio_regs;
+	unsigned long flags;
+	unsigned int *sync_word;
+
+	spin_lock_irqsave(&iommu->iommu_lock, flags);
+
+	while(sz >= 0) {
+		unsigned long start = sg[sz].dvma_addr;
+		unsigned long end = PAGE_ALIGN(start + sg[sz].len);
+
+		start &= PAGE_MASK;
+		while(start < end) {
+			sregs->sbuf_pflush = start;
+			start += PAGE_SIZE;
+		}
+		sz--;
+	}
+	sync_word = iommu->sbuf_flushflag_va;
+	sregs->sbuf_fsync = iommu->sbuf_flushflag_pa;
+	membar("#StoreLoad | #MemIssue");
+	while((*sync_word & 0x1) == 0)
+		membar("#LoadLoad");
+	*sync_word = 0;
+
+	spin_unlock_irqrestore(&iommu->iommu_lock, flags);
+}
+
 static char sfmmuinfo[512];
 
 char *mmu_info(void)
@@ -340,7 +474,7 @@ int prom_itlb_ent, prom_dtlb_ent;
 unsigned long prom_itlb_tag, prom_itlb_data;
 unsigned long prom_dtlb_tag, prom_dtlb_data;
 
-static inline void inherit_locked_prom_mappings(void)
+void inherit_locked_prom_mappings(int save_p)
 {
 	int i;
 	int dtlb_seen = 0;
@@ -367,9 +501,12 @@ static inline void inherit_locked_prom_mappings(void)
 		data = spitfire_get_dtlb_data(i);
 		if(!dtlb_seen && (data & _PAGE_L)) {
 			unsigned long tag = spitfire_get_dtlb_tag(i);
-			prom_dtlb_ent = i;
-			prom_dtlb_tag = tag;
-			prom_dtlb_data = data;
+
+			if(save_p) {
+				prom_dtlb_ent = i;
+				prom_dtlb_tag = tag;
+				prom_dtlb_data = data;
+			}
 			__asm__ __volatile__("stxa %%g0, [%0] %1"
 					     : : "r" (TLB_TAG_ACCESS), "i" (ASI_DMMU));
 			membar("#Sync");
@@ -390,9 +527,12 @@ static inline void inherit_locked_prom_mappings(void)
 		data = spitfire_get_itlb_data(i);
 		if(!itlb_seen && (data & _PAGE_L)) {
 			unsigned long tag = spitfire_get_itlb_tag(i);
-			prom_itlb_ent = i;
-			prom_itlb_tag = tag;
-			prom_itlb_data = data;
+
+			if(save_p) {
+				prom_itlb_ent = i;
+				prom_itlb_tag = tag;
+				prom_itlb_data = data;
+			}
 			__asm__ __volatile__("stxa %%g0, [%0] %1"
 					     : : "r" (TLB_TAG_ACCESS), "i" (ASI_IMMU));
 			membar("#Sync");
@@ -443,10 +583,14 @@ void __flush_cache_all(void)
 /* If not locked, zap it. */
 void __flush_tlb_all(void)
 {
-	unsigned long flags;
+	unsigned long pstate;
 	int i;
 
-	save_flags(flags); cli();
+	__asm__ __volatile__("rdpr	%%pstate, %0\n\t"
+			     "wrpr	%0, %1, %%pstate\n\t"
+			     "flushw"
+			     : "=r" (pstate)
+			     : "i" (PSTATE_IE));
 	for(i = 0; i < 64; i++) {
 		if(!(spitfire_get_dtlb_data(i) & _PAGE_L)) {
 			__asm__ __volatile__("stxa %%g0, [%0] %1"
@@ -465,19 +609,67 @@ void __flush_tlb_all(void)
 			membar("#Sync");
 		}
 	}
-	restore_flags(flags);
+	__asm__ __volatile__("wrpr	%0, 0, %%pstate"
+			     : : "r" (pstate));
 }
 
-void get_new_mmu_context(struct mm_struct *mm, unsigned long ctx)
+/* We are always protected by scheduler_lock under SMP. */
+void get_new_mmu_context(struct mm_struct *mm, unsigned long *ctx)
 {
-	if((ctx & ~(CTX_VERSION_MASK)) == 0) {
-		flush_tlb_all();
-		ctx = (ctx & CTX_VERSION_MASK) + CTX_FIRST_VERSION;
-		if(ctx == 1)
-			ctx = CTX_FIRST_VERSION;
+	unsigned int new_ctx = *ctx;
+
+	if((new_ctx & ~(CTX_VERSION_MASK)) == 0) {
+		new_ctx += CTX_FIRST_VERSION;
+		if(new_ctx == 1)
+			new_ctx = CTX_FIRST_VERSION;
+		*ctx = new_ctx;
+		DO_LOCAL_FLUSH(smp_processor_id());
 	}
-	tlb_context_cache = ctx + 1;
-	mm->context = ctx;
+	mm->context = new_ctx;
+	mm->cpu_vm_mask = 0;	/* Callers sets it properly. */
+	(*ctx)++;
+}
+
+#ifndef __SMP__
+unsigned long *pgd_quicklist = NULL;
+unsigned long *pmd_quicklist = NULL;
+unsigned long *pte_quicklist = NULL;
+unsigned long pgtable_cache_size = 0;
+#endif
+
+pgd_t *get_pgd_slow(void)
+{
+	pgd_t *pgd;
+
+	pgd = (pgd_t *) __get_free_page(GFP_KERNEL);
+	if(pgd)
+		__init_pgd(pgd);
+	return pgd;
+}
+
+pmd_t *get_pmd_slow(pgd_t *pgd, unsigned long offset)
+{
+	pmd_t *pmd;
+
+	pmd = (pmd_t *) __get_free_page(GFP_KERNEL);
+	if(pmd) {
+		__init_pmd(pmd);
+		pgd_set(pgd, pmd);
+		return pmd + offset;
+	}
+	return NULL;
+}
+
+pte_t *get_pte_slow(pmd_t *pmd, unsigned long offset)
+{
+	pte_t *pte;
+
+	pte = (pte_t *) get_free_page(GFP_KERNEL);
+	if(pte) {
+		pmd_set(pmd, pte);
+		return pte + offset;
+	}
+	return NULL;
 }
 
 __initfunc(static void
@@ -595,7 +787,7 @@ paging_init(unsigned long start_mem, unsigned long end_mem))
 	 */
 	pt  = phys_base | _PAGE_VALID | _PAGE_SZ4MB;
 	pt |= _PAGE_CP | _PAGE_CV | _PAGE_P | _PAGE_L | _PAGE_W;
-	save_flags(flags); cli();
+	__save_and_cli(flags);
 	__asm__ __volatile__("
 	stxa	%1, [%0] %3
 	stxa	%2, [%5] %4
@@ -608,15 +800,18 @@ paging_init(unsigned long start_mem, unsigned long end_mem))
 	: "r" (TLB_TAG_ACCESS), "r" (alias_base), "r" (pt),
 	  "i" (ASI_DMMU), "i" (ASI_DTLB_DATA_ACCESS), "r" (61 << 3)
 	: "memory");
-	restore_flags(flags);
+	__restore_flags(flags);
 	
 	/* Now set kernel pgd to upper alias so physical page computations
 	 * work.
 	 */
 	init_mm.pgd += ((shift) / (sizeof(pgd_t *)));
 
-	null_pmd_table = __pa(((unsigned long)&empty_null_pmd_table) + shift);
-	null_pte_table = __pa(((unsigned long)&empty_null_pte_table) + shift);
+	/* The funny offsets are to make page table operations much quicker and
+	 * requite less state, see pgtable.h for gory details.
+	 */
+	null_pmd_table=__pa(((unsigned long)&empty_null_pmd_table)+shift);
+	null_pte_table=__pa(((unsigned long)&empty_null_pte_table)+shift);
 
 	pmdp = (pmd_t *) &empty_null_pmd_table;
 	for(i = 0; i < 1024; i++)
@@ -658,7 +853,7 @@ paging_init(unsigned long start_mem, unsigned long end_mem))
 	flushi((long)&empty_zero_page);
 	membar("#Sync");
 
-	inherit_locked_prom_mappings();
+	inherit_locked_prom_mappings(1);
 	
 	flush_tlb_all();
 	
diff --git a/arch/sparc64/mm/ultra.S b/arch/sparc64/mm/ultra.S
index 18f9a363d..ec96e8871 100644
--- a/arch/sparc64/mm/ultra.S
+++ b/arch/sparc64/mm/ultra.S
@@ -1,4 +1,4 @@
-/* $Id: ultra.S,v 1.9 1997/07/24 12:15:08 davem Exp $
+/* $Id: ultra.S,v 1.3 1997/08/30 04:53:20 ralf Exp $
  * ultra.S: Don't expand these all over the place...
  *
  * Copyright (C) 1997 David S. Miller (davem@caip.rutgers.edu)
@@ -13,28 +13,26 @@
 	.align		32
 	.globl		__flush_tlb_mm, __flush_tlb_range, __flush_tlb_page
 __flush_tlb_mm:		/* %o0 == (mm->context & 0x1fff) */
-	rdpr		%otherwin, %g1
-	brz,pt		%g1, 1f
-	 mov		%o7, %g3
-	call		__flushw_user
-	 clr		%g2
-1:	rdpr		%pil, %g1
-9:	mov		SECONDARY_CONTEXT, %g7
-	wrpr		%g0, 15, %pil
-
-	ldxa		[%g7] ASI_DMMU, %g2
+	mov		SECONDARY_CONTEXT, %g7
+9:	ldxa		[%g7] ASI_DMMU, %g2
 	cmp		%g2, %o0
-	be,pt		%icc, 1f
+	bne,pn		%icc, 1f
 	 mov		0x50, %g3
+	stxa		%g0, [%g3] ASI_DMMU_DEMAP
+	stxa		%g0, [%g3] ASI_IMMU_DEMAP
+	retl
+	 flush		%g6
+1:	rdpr		%pstate, %g1
+	wrpr		%g1, PSTATE_IE, %pstate
 	stxa		%o0, [%g7] ASI_DMMU
-1:	stxa		%g0, [%g3] ASI_DMMU_DEMAP
-	be,pt		%icc, 1f
-	 stxa		%g0, [%g3] ASI_IMMU_DEMAP
-
+	stxa		%g0, [%g3] ASI_DMMU_DEMAP
+	stxa		%g0, [%g3] ASI_IMMU_DEMAP
+	flush		%g6
 	stxa		%g2, [%g7] ASI_DMMU
-1:	wrpr		%g1, 0x0, %pil
+	flush		%g6
 	retl
-	 flush		%g6
+	 wrpr		%g1, 0, %pstate
+	nop
 __flush_tlb_range:	/* %o0 == (mm->context & 0x1fff), %o1 == start, %o2 == end */
 	sethi		%hi(8192 - 1), %g5
 	or		%g5, %lo(8192 - 1), %g5
@@ -43,19 +41,13 @@ __flush_tlb_range:	/* %o0 == (mm->context & 0x1fff), %o1 == start, %o2 == end */
 
 	sub		%o2, %o1, %o3
 	add		%g5, 1, %g5
-	orcc		%o1, 0x50, %o1
+	orcc		%o1, 0x10, %o1
 	srlx		%o3, 13, %o4
-	rdpr		%otherwin, %g1
-	brz,pt		%g1, 1f
-	 mov		%o7, %g3
-	call		__flushw_user
-
-	 clr		%g2
-1:	cmp		%o4, 96
+	cmp		%o4, 96
 	bgu,pn		%icc, 9b
-	 rdpr		%pil, %g1
-	mov		SECONDARY_CONTEXT, %g7
-	wrpr		%g0, 15, %pil
+	 mov		SECONDARY_CONTEXT, %g7
+	rdpr		%pstate, %g1
+	wrpr		%g1, PSTATE_IE, %pstate
 	ldxa		[%g7] ASI_DMMU, %g2
 	cmp		%g2, %o0
 
@@ -66,37 +58,37 @@ __flush_tlb_range:	/* %o0 == (mm->context & 0x1fff), %o1 == start, %o2 == end */
 	stxa		%g0, [%o1 + %o3] ASI_IMMU_DEMAP
 	brnz,pt		%o3, 1b
 	 sub		%o3, %g5, %o3
-	nop
+	flush		%g6
 
-	be,pt		%icc, 1f
-	 wrpr		%g1, 0x0, %pil
+	be,a,pt		%icc, 1f
+	 nop
 	stxa		%g2, [%g7] ASI_DMMU
-1:	retl
-	 flush		%g6
+1:	flush		%g6
+	wrpr		%g1, 0, %pstate
+	retl
+	 nop
 
 	.align		32
 __flush_tlb_page:	/* %o0 == (mm->context & 0x1fff), %o1 == page & PAGE_MASK */
-	rdpr		%otherwin, %g1
-	brz,pt		%g1, 1f
-	 mov		%o7, %g3
-	call		__flushw_user
-	 clr		%g2
-1:	rdpr		%pil, %g1
 	mov		SECONDARY_CONTEXT, %g7
-	wrpr		%g0, 15, %pil
-
 	ldxa		[%g7] ASI_DMMU, %g2
 	cmp		%g2, %o0
 	be,pt		%icc, 1f
 	 or		%o1, 0x10, %g3
+	stxa		%g0, [%g3] ASI_DMMU_DEMAP
+	stxa		%g0, [%g3] ASI_IMMU_DEMAP
+	retl
+	 flush		%g6
+1:	rdpr		%pstate, %g1
+	wrpr		%g1, PSTATE_IE, %pstate
 	stxa		%o0, [%g7] ASI_DMMU
-1:	stxa		%g0, [%g3] ASI_DMMU_DEMAP
-	be,pt		%icc, 1f
-	 stxa		%g0, [%g3] ASI_IMMU_DEMAP
+	stxa		%g0, [%g3] ASI_DMMU_DEMAP
+	stxa		%g0, [%g3] ASI_IMMU_DEMAP
+	flush		%g6
 	stxa		%g2, [%g7] ASI_DMMU
-1:	wrpr		%g1, 0x0, %pil
+	flush		%g6
 	retl
-	 flush		%g6
+	 wrpr		%g1, 0, %pstate
 
 #ifdef __SMP__
 	/* These are all called by the slaves of a cross call, at
@@ -111,50 +103,29 @@ __flush_tlb_page:	/* %o0 == (mm->context & 0x1fff), %o1 == page & PAGE_MASK */
 	 *   %g2	scratch 1
 	 *   %g3	scratch 2
 	 *   %g4	scratch 3
-	 *
-	 * NOTE: We do not acknowledge the UPA until we are done
-	 *       with the service.  This is what tells the master
-	 *       that he can consider the effects of the flush
-	 *       "complete" on this cpu.
 	 */
 	.align		32
-	.globl		xcall_flush_tlb_page
+	.globl		xcall_flush_tlb_page, xcall_flush_tlb_mm, xcall_flush_tlb_range
 xcall_flush_tlb_page:
 	mov		SECONDARY_CONTEXT, %g2
-	nop
+	or		%g6, 0x10, %g4
 	ldxa		[%g2] ASI_DMMU, %g3
-	cmp		%g3, %g5
-	be,pt		%icc, 1f
-	 or		%g6, 0x10, %g4
 	stxa		%g5, [%g2] ASI_DMMU
-1:	stxa		%g0, [%g4] ASI_DMMU_DEMAP
-
-	be,pt		%icc, 1f
-	 stxa		%g0, [%g4] ASI_IMMU_DEMAP
+	stxa		%g0, [%g4] ASI_DMMU_DEMAP
+	stxa		%g0, [%g4] ASI_IMMU_DEMAP
 	stxa		%g3, [%g2] ASI_DMMU
-1:	b,pt		%xcc, do_ivec_return
-	 flush		%g1
+	retry
 
-	.align		32
-	.globl		xcall_flush_tlb_mm
 xcall_flush_tlb_mm:
 	mov		SECONDARY_CONTEXT, %g2
-	nop
+	mov		0x50, %g4
 	ldxa		[%g2] ASI_DMMU, %g3
-	cmp		%g3, %g5
-	be,pt		%icc, 1f
-	 mov		0x50, %g4
 	stxa		%g5, [%g2] ASI_DMMU
-1:	stxa		%g0, [%g4] ASI_DMMU_DEMAP
-
-	be,pt		%icc, 1f
-	 stxa		%g0, [%g4] ASI_IMMU_DEMAP
+	stxa		%g0, [%g4] ASI_DMMU_DEMAP
+	stxa		%g0, [%g4] ASI_IMMU_DEMAP
 	stxa		%g3, [%g2] ASI_DMMU
-1:	b,pt		%xcc, do_ivec_return
-	 flush		%g1
+	retry
 
-	.align		32
-	.globl		xcall_flush_tlb_range
 xcall_flush_tlb_range:
 	sethi		%hi(8192 - 1), %g2
 	or		%g2, %lo(8192 - 1), %g2
@@ -162,26 +133,54 @@ xcall_flush_tlb_range:
 	andn		%g7, %g2, %g7
 	sub		%g7, %g6, %g3
 	add		%g2, 1, %g2
-	orcc		%g6, 0x50, %g6
+	orcc		%g6, 0x10, %g6
 	srlx		%g3, 13, %g4
 
 	cmp		%g4, 96
 	bgu,pn		%icc, xcall_flush_tlb_mm
 	 mov		SECONDARY_CONTEXT, %g4
 	ldxa		[%g4] ASI_DMMU, %g7
-	cmp		%g7, %g5
-	be,pt		%icc, 1f
-	 sub		%g3, %g2, %g3
+	sub		%g3, %g2, %g3
 	stxa		%g5, [%g4] ASI_DMMU
+	nop
+	nop
 
 1:	stxa		%g0, [%g6 + %g3] ASI_DMMU_DEMAP
 	stxa		%g0, [%g6 + %g3] ASI_IMMU_DEMAP
 	brnz,pt		%g3, 1b
 	 sub		%g3, %g2, %g3
-	bne,a,pn	%icc, 1f
-	 stxa		%g7, [%g4] ASI_DMMU
-1:	b,pt		%xcc, do_ivec_return
-	 flush		%g1	
+	stxa		%g7, [%g4] ASI_DMMU
+	retry
+	nop
+	nop
+
+	.globl		xcall_report_regs
+xcall_report_regs:
+	rdpr		%pstate, %g2
+	wrpr		%g2, PSTATE_IG | PSTATE_AG, %pstate
+	rdpr		%pil, %g2
+	wrpr		%g0, 15, %pil
+	sethi		%hi(109f), %g7
+	b,pt		%xcc, etrap_irq
+109:	 or		%g7, %lo(109b), %g7
+	call		__show_regs
+	 add		%sp, STACK_BIAS + REGWIN_SZ, %o0
+	b,pt		%xcc, rtrap
+	 clr		%l6
+
+	.globl		xcall_capture
+xcall_capture:
+	rdpr		%pstate, %g2
+	wrpr		%g2, PSTATE_IG | PSTATE_AG, %pstate
+	rdpr		%pil, %g2
+	wrpr		%g0, 15, %pil
+	sethi		%hi(109f), %g7
+	b,pt		%xcc, etrap_irq
+109:	 or		%g7, %lo(109b), %g7
+	call		smp_penguin_jailcell
+	 nop
+	b,pt		%xcc, rtrap
+	 clr		%l6
 
 	/* These two are not performance critical... */
 	.globl		xcall_flush_tlb_all
@@ -209,8 +208,8 @@ xcall_flush_tlb_all:
 	cmp		%g2, 63
 	ble,pt		%icc, 1b
 	 sll		%g2, 3, %g3
-	b,pt		%xcc, do_ivec_return
-	 flush		%g1
+	flush		%g1
+	retry
 
 	.globl		xcall_flush_cache_all
 xcall_flush_cache_all:
@@ -222,6 +221,6 @@ xcall_flush_cache_all:
 	cmp		%g3, %g2
 	bleu,pt		%xcc, 1b
 	 nop
-	b,pt		%xcc, do_ivec_return
-	 flush		%g1
+	flush		%g1
+	retry
 #endif /* __SMP__ */
diff --git a/arch/sparc64/prom/console.c b/arch/sparc64/prom/console.c
index 40b33da4b..80ba5ec10 100644
--- a/arch/sparc64/prom/console.c
+++ b/arch/sparc64/prom/console.c
@@ -1,4 +1,4 @@
-/* $Id: console.c,v 1.7 1997/07/19 08:28:29 ecd Exp $
+/* $Id: console.c,v 1.8 1997/08/16 08:00:16 davem Exp $
  * console.c: Routines that deal with sending and receiving IO
  *            to/from the current console device using the PROM.
  *
@@ -85,13 +85,13 @@ prom_query_input_device()
 	if(prom_node_has_property(st_p, "keyboard"))
 		return PROMDEV_IKBD;
 	prom_getproperty(st_p, "device_type", propb, sizeof(propb));
-	if(strncmp(propb, "serial", sizeof("serial")))
+	if(strncmp(propb, "serial", 6))
 		return PROMDEV_I_UNK;
 	/* FIXME: Is there any better way how to find out? */	
 	memset(propb, 0, sizeof(propb));
 	st_p = prom_finddevice ("/options");
 	prom_getproperty(st_p, "input-device", propb, sizeof(propb));
-	if (strncmp (propb, "tty", 3) || !propb[3] || propb[4])
+	if (strncmp (propb, "tty", 3) || !propb[3])
 		return PROMDEV_I_UNK;
 	switch (propb[3]) {
 		case 'a': return PROMDEV_ITTYA;
@@ -114,13 +114,13 @@ prom_query_output_device()
 	if (propl >= 0 && propl == sizeof("display") &&
 	    strncmp("display", propb, sizeof("display")) == 0)
 		return PROMDEV_OSCREEN;
-	if(strncmp("serial", propb, sizeof("serial")))
+	if(strncmp("serial", propb, 6))
 		return PROMDEV_O_UNK;
 	/* FIXME: Is there any better way how to find out? */	
 	memset(propb, 0, sizeof(propb));
 	st_p = prom_finddevice ("/options");
 	prom_getproperty(st_p, "output-device", propb, sizeof(propb));
-	if (strncmp (propb, "tty", 3) || !propb[3] || propb[4])
+	if (strncmp (propb, "tty", 3) || !propb[3])
 		return PROMDEV_O_UNK;
 	switch (propb[3]) {
 		case 'a': return PROMDEV_OTTYA;
diff --git a/arch/sparc64/prom/p1275.c b/arch/sparc64/prom/p1275.c
index 0703b5cd7..9c5bce294 100644
--- a/arch/sparc64/prom/p1275.c
+++ b/arch/sparc64/prom/p1275.c
@@ -1,4 +1,4 @@
-/* $Id: p1275.c,v 1.11 1997/07/24 12:15:11 davem Exp $
+/* $Id: p1275.c,v 1.12 1997/07/26 18:39:01 davem Exp $
  * p1275.c: Sun IEEE 1275 PROM low level interface routines
  *
  * Copyright (C) 1996,1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
@@ -64,7 +64,7 @@ long p1275_cmd (char *service, long fmt, ...)
 	long ctx = 0;
 	
 	p = p1275buf.prom_buffer;
-	save_and_cli(flags);
+	__save_and_cli(flags);
 	ctx = spitfire_get_primary_context ();
 	if (ctx) {
 		flushw_user ();
@@ -149,7 +149,7 @@ long p1275_cmd (char *service, long fmt, ...)
 	
 	if (ctx)
 		spitfire_set_primary_context (ctx);
-	restore_flags(flags);
+	__restore_flags(flags);
 	return x;
 }
 
diff --git a/arch/sparc64/prom/ranges.c b/arch/sparc64/prom/ranges.c
index 323539237..83f860d45 100644
--- a/arch/sparc64/prom/ranges.c
+++ b/arch/sparc64/prom/ranges.c
@@ -1,15 +1,21 @@
-/* $Id: ranges.c,v 1.3 1997/03/21 12:33:36 jj Exp $
+/* $Id: ranges.c,v 1.8 1997/08/17 22:39:45 ecd Exp $
  * ranges.c: Handle ranges in newer proms for obio/sbus.
  *
  * Copyright (C) 1995 David S. Miller (davem@caip.rutgers.edu)
  * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
  */
 
+#include <linux/config.h>
 #include <linux/init.h>
 #include <asm/openprom.h>
 #include <asm/oplib.h>
 #include <asm/sbus.h>
+#include <asm/fhc.h>
 #include <asm/system.h>
+#ifdef CONFIG_PCI
+#include <asm/pbm.h>
+#include <asm/ebus.h>
+#endif
 
 struct linux_prom_ranges promlib_obio_ranges[PROMREG_MAX];
 int num_obio_ranges;
@@ -62,6 +68,25 @@ void prom_apply_sbus_ranges(struct linux_sbus *sbus, struct linux_prom_registers
 	}
 }
 
+/* Apply probed fhc ranges to registers passed, if no ranges return. */
+void prom_apply_fhc_ranges(struct linux_fhc *fhc,
+			   struct linux_prom_registers *regs,
+			   int nregs)
+{
+	if(fhc->num_fhc_ranges)
+		prom_adjust_regs(regs, nregs, fhc->fhc_ranges,
+				 fhc->num_fhc_ranges);
+}
+
+/* Apply probed central ranges to registers passed, if no ranges return. */
+void prom_apply_central_ranges(struct linux_central *central,
+			       struct linux_prom_registers *regs, int nregs)
+{
+	if(central->num_central_ranges)
+		prom_adjust_regs(regs, nregs, central->central_ranges,
+				 central->num_central_ranges);
+}
+
 __initfunc(void prom_ranges_init(void))
 {
 }
@@ -78,6 +103,56 @@ __initfunc(void prom_sbus_ranges_init(int iommund, struct linux_sbus *sbus))
 		sbus->num_sbus_ranges = (success/sizeof(struct linux_prom_ranges));
 }
 
+__initfunc(void prom_central_ranges_init(int cnode, struct linux_central *central))
+{
+	int success;
+	
+	central->num_central_ranges = 0;
+	success = prom_getproperty(central->prom_node, "ranges",
+				   (char *) central->central_ranges,
+				   sizeof (central->central_ranges));
+	if (success != -1)
+		central->num_central_ranges = (success/sizeof(struct linux_prom_ranges));
+}
+
+__initfunc(void prom_fhc_ranges_init(int fnode, struct linux_fhc *fhc))
+{
+	int success;
+	
+	fhc->num_fhc_ranges = 0;
+	success = prom_getproperty(fhc->prom_node, "ranges",
+				   (char *) fhc->fhc_ranges,
+				   sizeof (fhc->fhc_ranges));
+	if (success != -1)
+		fhc->num_fhc_ranges = (success/sizeof(struct linux_prom_ranges));
+}
+
+#ifdef CONFIG_PCI
+__initfunc(void prom_ebus_ranges_init(struct linux_ebus *ebus))
+{
+	int success;
+
+	ebus->num_ebus_ranges = 0;
+	success = prom_getproperty(ebus->prom_node, "ranges",
+				   (char *)ebus->ebus_ranges,
+				   sizeof(ebus->ebus_ranges));
+	if (success != -1)
+		ebus->num_ebus_ranges = (success/sizeof(struct linux_prom_ebus_ranges));
+}
+
+__initfunc(void prom_pbm_ranges_init(int pnode, struct linux_pbm_info *pbm))
+{
+	int success;
+
+	pbm->num_pbm_ranges = 0;
+	success = prom_getproperty(pbm->prom_node, "ranges",
+				   (char *)&pbm->pbm_ranges,
+				   sizeof(pbm->pbm_ranges));
+	if(success != -1)
+		pbm->num_pbm_ranges = (success/sizeof(struct linux_prom_pci_ranges));
+}
+#endif
+
 void
 prom_apply_generic_ranges (int node, int parent, struct linux_prom_registers *regs, int nregs)
 {
diff --git a/arch/sparc64/prom/tree.c b/arch/sparc64/prom/tree.c
index ae4baf858..c2b386641 100644
--- a/arch/sparc64/prom/tree.c
+++ b/arch/sparc64/prom/tree.c
@@ -1,4 +1,4 @@
-/* $Id: tree.c,v 1.5 1997/03/24 17:44:01 jj Exp $
+/* $Id: tree.c,v 1.6 1997/08/12 16:32:48 davem Exp $
  * tree.c: Basic device tree traversal/scanning for the Linux
  *         prom library.
  *
@@ -26,7 +26,7 @@ __prom_getchild(int node)
 __inline__ int
 prom_getchild(int node)
 {
-	long cnode;
+	int cnode;
 
 	if(node == -1) return 0;
 	cnode = __prom_getchild(node);
@@ -37,7 +37,7 @@ prom_getchild(int node)
 __inline__ int
 prom_getparent(int node)
 {
-	long cnode;
+	int cnode;
 
 	if(node == -1) return 0;
 	cnode = p1275_cmd ("parent", P1275_INOUT(1, 1), node);
@@ -57,12 +57,12 @@ __prom_getsibling(int node)
 __inline__ int
 prom_getsibling(int node)
 {
-	long sibnode;
+	int sibnode;
 
 	if(node == -1) return 0;
 	sibnode = __prom_getsibling(node);
 	if(sibnode == -1) return 0;
-	return (int)sibnode;
+	return sibnode;
 }
 
 /* Return the length in bytes of property 'prop' at node 'node'.
diff --git a/arch/sparc64/solaris/Makefile b/arch/sparc64/solaris/Makefile
new file mode 100644
index 000000000..56d252962
--- /dev/null
+++ b/arch/sparc64/solaris/Makefile
@@ -0,0 +1,26 @@
+#
+# Makefile for the Solaris binary emulation.
+#
+# 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 definition is now in the main makefile...
+
+O_TARGET := solaris.o
+O_OBJS   := entry64.o fs.o misc.o signal.o systbl.o ioctl.o ipc.o socksys.o
+ifeq ($(CONFIG_SOLARIS_EMUL),m)
+M_OBJS   := $(O_TARGET)
+endif
+
+.S.s:
+	$(CPP) -D__ASSEMBLY__ -ansi $< -o $*.s
+
+.S.o:
+	$(CC) -D__ASSEMBLY__ -ansi -c $< -o $*.o
+
+ifneq ($(CONFIG_SOLARIS_EMUL),y)
+do_it_all:
+endif
+
+include $(TOPDIR)/Rules.make
diff --git a/arch/sparc64/solaris/conv.h b/arch/sparc64/solaris/conv.h
new file mode 100644
index 000000000..c806d3dc1
--- /dev/null
+++ b/arch/sparc64/solaris/conv.h
@@ -0,0 +1,28 @@
+/* $Id: conv.h,v 1.2 1997/09/03 12:29:13 jj Exp $
+ * conv.h: Utility macros for Solaris emulation
+ *
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ */
+ 
+/* #define DEBUG_SOLARIS */
+
+#ifndef __ASSEMBLY__
+
+#include <asm/unistd.h>
+
+/* As gcc will warn about casting u32 to some ptr, we have to cast it to
+ * unsigned long first, and that's what is A() for.
+ * You just do (void *)A(x), instead of having to
+ * type (void *)((unsigned long)x) or instead of just (void *)x, which will
+ * produce warnings.
+ */
+#define A(x) ((unsigned long)x)
+
+extern unsigned sys_call_table[];
+extern unsigned sys_call_table32[];
+extern unsigned sunos_sys_table[];
+
+#define SYS(name) ((long)sys_call_table[__NR_##name])
+#define SUNOS(x) ((long)sunos_sys_table[x])
+
+#endif /* __ASSEMBLY__ */
diff --git a/arch/sparc64/solaris/entry64.S b/arch/sparc64/solaris/entry64.S
new file mode 100644
index 000000000..e983010d1
--- /dev/null
+++ b/arch/sparc64/solaris/entry64.S
@@ -0,0 +1,202 @@
+/* $Id: entry64.S,v 1.3 1997/09/03 12:29:12 jj Exp $
+ * entry64.S:  Solaris syscall emulation entry point.
+ *
+ * Copyright (C) 1996,1997 Jakub Jelinek   (jj@sunsite.mff.cuni.cz)
+ * Copyright (C) 1995,1997 David S. Miller (davem@caip.rutgers.edu)
+ * Copyright (C) 1996 Miguel de Icaza      (miguel@nuclecu.unam.mx)
+ */
+
+#include <linux/errno.h>
+
+#include <asm/head.h>
+#include <asm/asi.h>
+#include <asm/smp.h>
+#include <asm/ptrace.h>
+#include <asm/page.h>
+#include <asm/signal.h>
+#include <asm/pgtable.h>
+#include <asm/processor.h>
+
+#include "conv.h"
+
+#define NR_SYSCALLS	256
+
+	.text
+solaris_syscall_trace:
+	call		syscall_trace
+	 nop
+	mov		%i0, %o0
+	mov		%i1, %o1
+	mov		%i2, %o2
+	mov		%i3, %o3
+	b,pt		%xcc, 2f
+	 mov		%i4, %o4
+
+solaris_sucks:
+/* Solaris is a big system which needs to be able to do all the things
+ * in Inf+1 different ways */
+	add		%i6, 0x5c, %o0
+	mov		%i0, %g1
+	mov		%i1, %i0
+	mov		%i2, %i1
+	srl		%o0, 0, %o0
+	mov		%i3, %i2
+	mov		%i4, %i3
+	mov		%i5, %i4
+	ba,pt		%xcc, solaris_sparc_syscall
+exen:	 lduwa		[%o0] ASI_S, %i5
+
+exenf:	ba,pt		%xcc, solaris_sparc_syscall
+	 clr		%i5
+
+/* For shared binaries, binfmt_elf32 already sets up personality
+   and exec_domain. This is to handle static binaries as well */
+solaris_reg:
+	call		solaris_register
+	 nop
+	ba,pt		%xcc, 1f
+	 mov		%i0, %o0
+
+linux_syscall_for_solaris:
+	sll		%l7, 2, %l4
+	ba,pt		%xcc, 10f
+	 lduw		[%l6 + %l4], %l7
+
+	/* Solaris system calls enter here... */
+	.align	32
+	.globl	solaris_sparc_syscall
+solaris_sparc_syscall:
+#ifdef DEBUG_SOLARIS
+	mov		%g1, %o0
+	call		entry_printk
+	 add		%sp, STACK_BIAS + REGWIN_SZ, %o1
+#endif
+	ldub		[%g6 + AOFF_task_personality + ASIZ_task_personality - 1], %l0
+	cmp		%g1, 255
+	bg,pn		%icc, solaris_unimplemented
+	 srl		%g1, 0, %g1
+	sethi		%hi(solaris_sys_table), %l7
+	brz,pn		%g1, solaris_sucks
+	 mov		%i0, %o0
+	sll		%g1, 2, %l4
+	cmp		%l0, 1
+	bne,pn		%icc, solaris_reg
+1:	 mov		%i1, %o1				
+	lduw		[%l7 + %l4], %l7
+	mov		%i2, %o2				
+	ldx		[%g6 + AOFF_task_flags], %l5
+!	st		%g0, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_FPRS]
+	cmp		%l7, NR_SYSCALLS
+	bleu,a,pn	%xcc, linux_syscall_for_solaris
+	 sethi		%hi(sys_call_table32), %l6
+	andcc		%l7, 1, %g0
+	bne,a,pn	%icc, 10f
+	 add		%sp, STACK_BIAS + REGWIN_SZ, %o0
+10:	mov		%i3, %o3
+	mov		%i4, %o4
+	andn		%l7, 3, %l7
+	andcc		%l5, 0x20, %g0				
+	bne,pn		%icc, solaris_syscall_trace		
+	 mov		%i0, %l5
+2:	call		%l7
+	 mov		%i5, %o5
+ret_from_solaris:
+#ifdef DEBUG_SOLARIS
+	call		exit_printk
+	 mov		%o0, %l0
+	mov		%l0, %o0
+#endif
+	stx		%o0, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I0]
+	ldx		[%g6 + AOFF_task_flags], %l6
+	sra		%o0, 0, %o0
+	mov		%ulo(TSTATE_XCARRY | TSTATE_ICARRY), %g2
+	ldx		[%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TSTATE], %g3
+	cmp		%o0, -ENOIOCTLCMD
+	sllx		%g2, 32, %g2
+	bgeu,pn		%xcc, 1f
+	 andcc		%l6, 0x20, %l6	
+
+	/* System call success, clear Carry condition code. */
+	andn		%g3, %g2, %g3
+	stx		%g3, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TSTATE]	
+	bne,pn		%icc, solaris_syscall_trace2
+	 ldx		[%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TNPC], %l1 ! pc = npc
+	add		%l1, 0x4, %l2				         !npc = npc+4
+	stx		%l1, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TPC]
+	clr		%l6
+	call		rtrap
+	 stx		%l2, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TNPC]
+1:
+	/* System call failure, set Carry condition code.
+	 * Also, get abs(errno) to return to the process.
+	 */
+	sub		%g0, %o0, %o0
+	or		%g3, %g2, %g3
+	cmp		%o0, ERANGE	/* 0-ERANGE are identity mapped */
+	bleu,pt		%icc, 1f
+	 cmp		%o0, EMEDIUMTYPE
+	bgu,pn		%icc, 1f
+	 sethi		%hi(solaris_err_table), %l6
+	sll		%o0, 2, %o0
+	or		%l6, %lo(solaris_err_table), %l6
+	ldsw		[%l6 + %o0], %o0
+1:	stx		%o0, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I0]
+	mov		1, %l6
+	stx		%g3, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TSTATE]
+	bne,pn		%icc, solaris_syscall_trace2
+	 ldx		[%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TNPC], %l1 ! pc = npc
+	add		%l1, 0x4, %l2				         !npc = npc+4
+
+	stx		%l1, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TPC]
+	call		rtrap
+	 stx		%l2, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TNPC]
+solaris_syscall_trace2:
+	call		syscall_trace
+	 add		%l1, 0x4, %l2			/* npc = npc+4 */
+	stx		%l1, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TPC]
+	call		rtrap
+	 stx		%l2, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_TNPC]
+
+	/* This one is tricky, so that's why we do it in assembly */
+	.globl		solaris_sigsuspend
+solaris_sigsuspend:
+	call		do_sol_sigsuspend
+	 nop
+	brlz,pn		%o0, ret_from_solaris
+	 nop
+	call		sys_sigsuspend
+	 stx		%o0, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I0]
+
+	.globl		solaris_getpid
+solaris_getpid:
+	call		sys_getppid	/* This is tricky, so don't do it in assembly */
+	 nop
+	stx		%o0, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I1]
+	b,pt		%xcc, ret_from_solaris
+	 lduw		[%g6 + AOFF_task_pid], %o0
+
+	.globl	solaris_getuid
+solaris_getuid:
+	lduh		[%g6 + AOFF_task_euid], %o1
+	lduh		[%g6 + AOFF_task_uid], %o0
+	b,pt		%xcc, ret_from_solaris
+	 stx		%o1, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I1]
+
+	.globl	solaris_getgid
+solaris_getgid:
+	lduh		[%g6 + AOFF_task_egid], %o1
+	lduh		[%g6 + AOFF_task_gid], %o0
+	b,pt		%xcc, ret_from_solaris
+	 stx		%o1, [%sp + STACK_BIAS + REGWIN_SZ + PT_V9_I1]
+
+	.globl		solaris_unimplemented
+solaris_unimplemented:
+	call		do_sol_unimplemented
+	 add		%sp, STACK_BIAS + REGWIN_SZ, %o0
+	ba,pt		%xcc, ret_from_solaris
+	 nop
+
+	.section	__ex_table,#alloc
+	.align		4
+	.word		exen, exenf
+
diff --git a/arch/sparc64/solaris/fs.c b/arch/sparc64/solaris/fs.c
new file mode 100644
index 000000000..9d0905adc
--- /dev/null
+++ b/arch/sparc64/solaris/fs.c
@@ -0,0 +1,711 @@
+/* $Id: fs.c,v 1.4 1997/09/04 15:46:26 jj Exp $
+ * fs.c: fs related syscall emulation for Solaris
+ *
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ */
+
+#include <linux/types.h>
+#include <linux/fs.h>
+#include <linux/stat.h>
+#include <linux/smp_lock.h>
+#include <linux/limits.h>
+#include <linux/resource.h>
+
+#include <asm/uaccess.h>
+#include <asm/string.h>
+#include <asm/ptrace.h>
+
+#include "conv.h"
+
+extern char * getname32(u32 filename);
+#define putname32 putname
+ 
+#define R4_DEV(DEV) ((DEV & 0xff) | ((DEV & 0xff00) << 10))
+#define R4_MAJOR(DEV) (((DEV) >> 18) & 0x3fff)
+#define R4_MINOR(DEV) ((DEV) & 0x3ffff)
+#define R3_VERSION	1
+#define R4_VERSION	2
+
+typedef struct {
+	s32	tv_sec;
+	s32	tv_nsec;
+} timestruct_t;
+
+struct sol_stat {
+	u32		st_dev;
+	s32		st_pad1[3];     /* network id */
+	u32		st_ino;
+	u32		st_mode;
+	u32		st_nlink;
+	u32		st_uid;
+	u32		st_gid;
+	u32		st_rdev;
+	s32		st_pad2[2];
+	s32		st_size;
+	s32		st_pad3;	/* st_size, off_t expansion */
+	timestruct_t	st_atime;
+	timestruct_t	st_mtime;
+	timestruct_t	st_ctime;
+	s32		st_blksize;
+	s32		st_blocks;
+	char		st_fstype[16];
+	s32		st_pad4[8];     /* expansion area */
+};
+
+#define UFSMAGIC (((unsigned)'u'<<24)||((unsigned)'f'<<16)||((unsigned)'s'<<8))
+
+static inline int putstat(struct sol_stat *ubuf, struct stat *kbuf)
+{
+	if (put_user (R4_DEV(kbuf->st_dev), &ubuf->st_dev)	||
+	    __put_user (kbuf->st_ino, &ubuf->st_ino)		||
+	    __put_user (kbuf->st_mode, &ubuf->st_mode)		||
+	    __put_user (kbuf->st_nlink, &ubuf->st_nlink)	||
+	    __put_user (kbuf->st_uid, &ubuf->st_uid)		||
+	    __put_user (kbuf->st_gid, &ubuf->st_gid)		||
+	    __put_user (R4_DEV(kbuf->st_rdev), &ubuf->st_rdev)	||
+	    __put_user (kbuf->st_size, &ubuf->st_size)		||
+	    __put_user (kbuf->st_atime, &ubuf->st_atime.tv_sec)	||
+	    __put_user (0, &ubuf->st_atime.tv_nsec)		||
+	    __put_user (kbuf->st_mtime, &ubuf->st_mtime.tv_sec)	||
+	    __put_user (0, &ubuf->st_mtime.tv_nsec)		||
+	    __put_user (kbuf->st_ctime, &ubuf->st_ctime.tv_sec)	||
+	    __put_user (0, &ubuf->st_ctime.tv_nsec)		||
+	    __put_user (kbuf->st_blksize, &ubuf->st_blksize)	||
+	    __put_user (kbuf->st_blocks, &ubuf->st_blocks)	||
+	    __put_user (UFSMAGIC, (unsigned *)ubuf->st_fstype))
+		return -EFAULT;
+	return 0;
+}
+
+asmlinkage int solaris_stat(u32 filename, u32 statbuf)
+{
+	int ret;
+	struct stat s;
+	char *filenam;
+	unsigned long old_fs = get_fs();
+	int (*sys_newstat)(char *,struct stat *) = 
+		(int (*)(char *,struct stat *))SYS(stat);
+	
+	filenam = getname32 (filename);
+	ret = PTR_ERR(filenam);
+	if (!IS_ERR(filenam)) {
+		set_fs (KERNEL_DS);
+		ret = sys_newstat(filenam, &s);
+		set_fs (old_fs);
+		putname32 (filenam);
+		if (putstat ((struct sol_stat *)A(statbuf), &s))
+			return -EFAULT;
+	}
+	return ret;
+}
+
+asmlinkage int solaris_xstat(int vers, u32 filename, u32 statbuf)
+{
+	/* Solaris doesn't bother with looking at vers, so we do neither */
+	return solaris_stat(filename, statbuf);
+}
+
+asmlinkage int solaris_lstat(u32 filename, u32 statbuf)
+{
+	int ret;
+	struct stat s;
+	char *filenam;
+	unsigned long old_fs = get_fs();
+	int (*sys_newlstat)(char *,struct stat *) = 
+		(int (*)(char *,struct stat *))SYS(lstat);
+	
+	filenam = getname32 (filename);
+	ret = PTR_ERR(filenam);
+	if (!IS_ERR(filenam)) {
+		set_fs (KERNEL_DS);
+		ret = sys_newlstat(filenam, &s);
+		set_fs (old_fs);
+		putname32 (filenam);
+		if (putstat ((struct sol_stat *)A(statbuf), &s))
+			return -EFAULT;
+	}
+	return ret;
+}
+
+asmlinkage int solaris_lxstat(int vers, u32 filename, u32 statbuf)
+{
+	return solaris_lstat(filename, statbuf);
+}
+
+asmlinkage int solaris_fstat(unsigned int fd, u32 statbuf)
+{
+	int ret;
+	struct stat s;
+	unsigned long old_fs = get_fs();
+	int (*sys_newfstat)(unsigned,struct stat *) = 
+		(int (*)(unsigned,struct stat *))SYS(fstat);
+	
+	set_fs (KERNEL_DS);
+	ret = sys_newfstat(fd, &s);
+	set_fs (old_fs);
+	if (putstat ((struct sol_stat *)A(statbuf), &s))
+		return -EFAULT;
+	return ret;
+}
+
+asmlinkage int solaris_fxstat(int vers, u32 fd, u32 statbuf)
+{
+	return solaris_fstat(fd, statbuf);
+}
+
+asmlinkage int solaris_mknod(u32 path, u32 mode, s32 dev)
+{
+	int (*sys_mknod)(const char *,int,dev_t) = 
+		(int (*)(const char *,int,dev_t))SYS(mknod);
+	int major, minor;
+
+	if ((major = R4_MAJOR(dev)) > 255 || 
+	    (minor = R4_MINOR(dev)) > 255) return -EINVAL;
+	return sys_mknod((const char *)A(path), mode, MKDEV(major,minor));
+}
+
+asmlinkage int solaris_xmknod(int vers, u32 path, u32 mode, s32 dev)
+{
+	return solaris_mknod(path, mode, dev);
+}
+
+/* This statfs thingie probably will go in the near future, but... */
+
+struct sol_statfs {
+	short	f_type;
+	s32	f_bsize;
+	s32	f_frsize;
+	s32	f_blocks;
+	s32	f_bfree;
+	u32	f_files;
+	u32	f_ffree;
+	char	f_fname[6];
+	char	f_fpack[6];
+};
+
+asmlinkage int solaris_statfs(u32 path, u32 buf, int len, int fstype)
+{
+	int ret;
+	struct statfs s;
+	unsigned long old_fs = get_fs();
+	int (*sys_statfs)(const char *,struct statfs *) = 
+		(int (*)(const char *,struct statfs *))SYS(statfs);
+	struct sol_statfs *ss = (struct sol_statfs *)A(buf);
+	
+	if (len != sizeof(struct sol_statfs)) return -EINVAL;
+	if (!fstype) {
+		set_fs (KERNEL_DS);
+		ret = sys_statfs((const char *)A(path), &s);
+		set_fs (old_fs);
+		if (!ret) {
+			if (put_user (s.f_type, &ss->f_type)		||
+			    __put_user (s.f_bsize, &ss->f_bsize)	||
+			    __put_user (0, &ss->f_frsize)		||
+			    __put_user (s.f_blocks, &ss->f_blocks)	||
+			    __put_user (s.f_bfree, &ss->f_bfree)	||
+			    __put_user (s.f_files, &ss->f_files)	||
+			    __put_user (s.f_ffree, &ss->f_ffree)	||
+			    __clear_user (&ss->f_fname, 12))
+				return -EFAULT;
+		}
+		return ret;
+	}
+/* Linux can't stat unmounted filesystems so we
+ * simply lie and claim 100MB of 1GB is free. Sorry.
+ */
+	if (put_user (fstype, &ss->f_type)		||
+	    __put_user (1024, &ss->f_bsize)		||
+	    __put_user (0, &ss->f_frsize)		||
+	    __put_user (1024*1024, &ss->f_blocks)	||
+	    __put_user (100*1024, &ss->f_bfree)		||
+	    __put_user (60000, &ss->f_files)		||
+	    __put_user (50000, &ss->f_ffree)		||
+	    __clear_user (&ss->f_fname, 12))
+		return -EFAULT;
+	return 0;
+}
+
+asmlinkage int solaris_fstatfs(u32 fd, u32 buf, int len, int fstype)
+{
+	int ret;
+	struct statfs s;
+	unsigned long old_fs = get_fs();
+	int (*sys_fstatfs)(unsigned,struct statfs *) = 
+		(int (*)(unsigned,struct statfs *))SYS(fstatfs);
+	struct sol_statfs *ss = (struct sol_statfs *)A(buf);
+	
+	if (len != sizeof(struct sol_statfs)) return -EINVAL;
+	if (!fstype) {
+		set_fs (KERNEL_DS);
+		ret = sys_fstatfs(fd, &s);
+		set_fs (old_fs);
+		if (!ret) {
+			if (put_user (s.f_type, &ss->f_type)		||
+			    __put_user (s.f_bsize, &ss->f_bsize)	||
+			    __put_user (0, &ss->f_frsize)		||
+			    __put_user (s.f_blocks, &ss->f_blocks)	||
+			    __put_user (s.f_bfree, &ss->f_bfree)	||
+			    __put_user (s.f_files, &ss->f_files)	||
+			    __put_user (s.f_ffree, &ss->f_ffree)	||
+			    __clear_user (&ss->f_fname, 12))
+				return -EFAULT;
+		}
+		return ret;
+	}
+	/* Otherwise fstatfs is the same as statfs */
+	return solaris_statfs(0, buf, len, fstype);
+}
+
+struct sol_statvfs {
+	u32	f_bsize;
+	u32	f_frsize;
+	u32	f_blocks;
+	u32	f_bfree;
+	u32	f_bavail;
+	u32	f_files;
+	u32	f_ffree;
+	u32	f_favail;
+	u32	f_fsid;
+	char	f_basetype[16];
+	u32	f_flag;
+	u32	f_namemax;
+	char	f_fstr[32];
+	u32	f_filler[16];
+};
+
+static int report_statvfs(struct inode *inode, u32 buf)
+{
+	struct statfs s;
+	unsigned long old_fs = get_fs();
+	int error;
+	struct sol_statvfs *ss = (struct sol_statvfs *)A(buf);
+			
+	set_fs (KERNEL_DS);
+	error = inode->i_sb->s_op->statfs(inode->i_sb, &s, sizeof(struct statfs));
+	set_fs (old_fs);
+	if (!error) {
+		const char *p = inode->i_sb->s_type->name;
+		int i = 0;
+		int j = strlen (p);
+		
+		if (j > 15) j = 15;
+		if (IS_RDONLY(inode)) i = 1;
+		if (IS_NOSUID(inode)) i |= 2;
+		if (put_user (s.f_bsize, &ss->f_bsize)		||
+		    __put_user (0, &ss->f_frsize)		||
+		    __put_user (s.f_blocks, &ss->f_blocks)	||
+		    __put_user (s.f_bfree, &ss->f_bfree)	||
+		    __put_user (s.f_bavail, &ss->f_bavail)	||
+		    __put_user (s.f_files, &ss->f_files)	||
+		    __put_user (s.f_ffree, &ss->f_ffree)	||
+		    __put_user (s.f_ffree, &ss->f_favail)	||
+		    __put_user (R4_DEV(inode->i_sb->s_dev), &ss->f_fsid) ||
+		    __copy_to_user (ss->f_basetype,p,j)		||
+		    __put_user (0, (char *)&ss->f_basetype[j])	||
+		    __put_user (s.f_namelen, &ss->f_namemax)	||
+		    __put_user (i, &ss->f_flag)			||		    
+		    __clear_user (&ss->f_fstr, 32))
+			return -EFAULT;
+	}
+	return error;
+}
+
+asmlinkage int solaris_statvfs(u32 path, u32 buf)
+{
+	struct dentry * dentry;
+	int error;
+
+	lock_kernel();
+	dentry = namei((const char *)A(path));
+	error = PTR_ERR(dentry);
+	if (!IS_ERR(dentry)) {
+		struct inode * inode = dentry->d_inode;
+
+		error = -ENOSYS;
+		if (inode->i_sb->s_op->statfs)
+			error = report_statvfs(inode, buf);
+		dput(dentry);
+	}
+	unlock_kernel();
+	return error;
+}
+
+asmlinkage int solaris_fstatvfs(unsigned int fd, u32 buf)
+{
+	struct inode * inode;
+	struct dentry * dentry;
+	struct file * file;
+	int error;
+
+	lock_kernel();
+	if (fd >= NR_OPEN || !(file = current->files->fd[fd]))
+		error = -EBADF;
+	else if (!(dentry = file->f_dentry))
+		error = -ENOENT;
+	else if (!(inode = dentry->d_inode))
+		error = -ENOENT;
+	else if (!inode->i_sb)
+		error = -ENODEV;
+	else if (!inode->i_sb->s_op->statfs)
+		error = -ENOSYS;
+	else
+		error = report_statvfs(inode, buf);
+	unlock_kernel();
+	return error;
+}
+
+asmlinkage int solaris_open(u32 filename, int flags, u32 mode)
+{
+	int (*sys_open)(const char *,int,int) = 
+		(int (*)(const char *,int,int))SYS(open);
+	int fl = flags & 0xf;
+	
+	if (flags & 0x8050) fl |= O_SYNC;
+	if (flags & 0x80) fl |= O_NONBLOCK;
+	if (flags & 0x100) fl |= O_CREAT;
+	if (flags & 0x200) fl |= O_TRUNC;
+	if (flags & 0x400) fl |= O_EXCL;
+	if (flags & 0x800) fl |= O_NOCTTY;
+	return sys_open((const char *)A(filename), fl, mode);
+}
+
+#define SOL_F_SETLK	6
+#define SOL_F_SETLKW	7
+#define SOL_F_FREESP    11
+#define SOL_F_ISSTREAM  13
+#define SOL_F_GETLK     14
+#define SOL_F_PRIV      15
+#define SOL_F_NPRIV     16
+#define SOL_F_QUOTACTL  17
+#define SOL_F_BLOCKS    18
+#define SOL_F_BLKSIZE   19
+#define SOL_F_GETOWN    23
+#define SOL_F_SETOWN    24
+
+struct sol_flock {
+	short	l_type;
+	short	l_whence;
+	u32	l_start;
+	u32	l_len;
+	s32	l_sysid;
+	s32	l_pid;
+	s32	l_pad[4];
+};
+
+asmlinkage int solaris_fcntl(unsigned fd, unsigned cmd, u32 arg)
+{
+	int (*sys_fcntl)(unsigned,unsigned,unsigned long) = 
+		(int (*)(unsigned,unsigned,unsigned long))SYS(fcntl);
+	int ret, flags;
+
+	switch (cmd) {
+	case F_DUPFD:
+	case F_GETFD:
+	case F_SETFD: return sys_fcntl(fd, cmd, (unsigned long)arg);
+	case F_GETFL:
+		flags = sys_fcntl(fd, cmd, 0);
+		ret = flags & 0xf;
+		if (flags & O_SYNC) ret |= 0x8050;
+		if (flags & O_NONBLOCK) ret |= 0x80;
+		return ret;
+	case F_SETFL:
+		flags = arg & 0xf;
+		if (arg & 0x8050) flags |= O_SYNC;
+		if (arg & 0x80) flags |= O_NONBLOCK;
+		return sys_fcntl(fd, cmd, (long)flags);
+	case SOL_F_GETLK:
+	case SOL_F_SETLK:
+	case SOL_F_SETLKW:
+		{
+			struct flock f;
+			unsigned long old_fs = get_fs();
+
+			switch (cmd) {
+			case SOL_F_GETLK: cmd = F_GETLK; break;
+			case SOL_F_SETLK: cmd = F_SETLK; break;
+			case SOL_F_SETLKW: cmd = F_SETLKW; break;
+			}
+
+			get_user_ret (f.l_type, &((struct sol_flock *)A(arg))->l_type, -EFAULT);
+			__get_user_ret (f.l_whence, &((struct sol_flock *)A(arg))->l_whence, -EFAULT);
+			__get_user_ret (f.l_start, &((struct sol_flock *)A(arg))->l_start, -EFAULT);
+			__get_user_ret (f.l_len, &((struct sol_flock *)A(arg))->l_len, -EFAULT);
+			__get_user_ret (f.l_pid, &((struct sol_flock *)A(arg))->l_sysid, -EFAULT);
+			set_fs(KERNEL_DS);
+			ret = sys_fcntl(fd, cmd, (unsigned long)&f);
+			set_fs(old_fs);
+			__put_user_ret (f.l_type, &((struct sol_flock *)A(arg))->l_type, -EFAULT);
+			__put_user_ret (f.l_whence, &((struct sol_flock *)A(arg))->l_whence, -EFAULT);
+			__put_user_ret (f.l_start, &((struct sol_flock *)A(arg))->l_start, -EFAULT);
+			__put_user_ret (f.l_len, &((struct sol_flock *)A(arg))->l_len, -EFAULT);
+			__put_user_ret (f.l_pid, &((struct sol_flock *)A(arg))->l_pid, -EFAULT);
+			__put_user_ret (0, &((struct sol_flock *)A(arg))->l_sysid, -EFAULT);
+			return ret;
+		}
+	}
+	return -EINVAL;
+}
+
+asmlinkage int solaris_ulimit(int cmd, int val)
+{
+	switch (cmd) {
+	case 1: /* UL_GETFSIZE - in 512B chunks */
+		return current->rlim[RLIMIT_FSIZE].rlim_cur >> 9;
+	case 2: /* UL_SETFSIZE */
+		if ((unsigned long)val > (LONG_MAX>>9)) return -ERANGE;
+		val <<= 9;
+		lock_kernel();
+		if (val > current->rlim[RLIMIT_FSIZE].rlim_max) {
+			if (!suser()) {
+				unlock_kernel();
+				return -EPERM;
+			}
+			current->rlim[RLIMIT_FSIZE].rlim_max = val;
+		}
+		current->rlim[RLIMIT_FSIZE].rlim_cur = val;
+		unlock_kernel();
+		return 0;
+	case 3: /* UL_GMEMLIM */
+		return current->rlim[RLIMIT_DATA].rlim_cur;
+	case 4: /* UL_GDESLIM */
+		return NR_OPEN;
+	}
+	return -EINVAL;
+}
+
+static int chown_common(struct dentry * dentry, uid_t user, gid_t group)
+{
+	struct inode * inode;
+	struct iattr newattrs;
+	int error;
+
+	error = -ENOENT;
+	if (!(inode = dentry->d_inode)) {
+		printk("chown_common: NULL inode\n");
+		goto out;
+	}
+	error = -EROFS;
+	if (IS_RDONLY(inode))
+		goto out;
+	error = -EPERM;
+	if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
+		goto out;
+	if (user == (uid_t) -1)
+		user = inode->i_uid;
+	if (group == (gid_t) -1)
+		group = inode->i_gid;
+	newattrs.ia_mode = inode->i_mode;
+	newattrs.ia_uid = user;
+	newattrs.ia_gid = group;
+	newattrs.ia_valid =  ATTR_UID | ATTR_GID | ATTR_CTIME;
+	/*
+	 * If the owner has been changed, remove the setuid bit
+	 */
+	if (inode->i_mode & S_ISUID) {
+		newattrs.ia_mode &= ~S_ISUID;
+		newattrs.ia_valid |= ATTR_MODE;
+	}
+	/*
+	 * If the group has been changed, remove the setgid bit
+	 *
+	 * Don't remove the setgid bit if no group execute bit.
+	 * This is a file marked for mandatory locking.
+	 */
+	if (((inode->i_mode & (S_ISGID | S_IXGRP)) == (S_ISGID | S_IXGRP))) {
+		newattrs.ia_mode &= ~S_ISGID;
+		newattrs.ia_valid |= ATTR_MODE;
+	}
+	if (inode->i_sb && inode->i_sb->dq_op) {
+		inode->i_sb->dq_op->initialize(inode, -1);
+		error = -EDQUOT;
+		if (inode->i_sb->dq_op->transfer(inode, &newattrs, 0))
+			goto out;
+		error = notify_change(inode, &newattrs);
+		if (error)
+			inode->i_sb->dq_op->transfer(inode, &newattrs, 1);
+	} else
+		error = notify_change(inode, &newattrs);
+out:
+	return error;
+}
+
+/* Linux chown works like Solaris lchown. Solaris chown does follow symlink */
+asmlinkage int solaris_chown(u32 filename, s32 user, s32 group)
+{
+	struct dentry * dentry;
+	int error;
+
+	lock_kernel();
+	dentry = namei((const char *)A(filename));
+
+	error = PTR_ERR(dentry);
+	if (!IS_ERR(dentry)) {
+		error = chown_common(dentry, user, group);
+		dput(dentry);
+	}
+	unlock_kernel();
+	return error;
+}
+
+/* At least at the time I'm writing this, Linux doesn't have ACLs, so we
+   just fake this */
+asmlinkage int solaris_acl(u32 filename, int cmd, int nentries, u32 aclbufp)
+{
+	return -ENOSYS;
+}
+
+asmlinkage int solaris_facl(unsigned int fd, int cmd, int nentries, u32 aclbufp)
+{
+	return -ENOSYS;
+}
+
+asmlinkage int solaris_pread(int fd, u32 buf, u32 nbyte, s32 offset)
+{
+	off_t temp;
+	int retval;
+	struct file * file;
+	long (*sys_read)(unsigned int, char *, unsigned long) =
+		(long (*)(unsigned int, char *, unsigned long))SYS(read);
+	long (*sys_lseek)(unsigned int, off_t, unsigned int) =
+		(long (*)(unsigned int, off_t, unsigned int))SYS(lseek);
+	
+	lock_kernel();
+	retval = -EBADF;
+	if (fd >= NR_OPEN ||
+            !(file = current->files->fd[fd]))
+		goto bad;
+	temp = file->f_pos;
+	if (temp != offset) {
+		retval = sys_lseek(fd, offset, 0);
+		if (retval < 0) goto bad;
+	}
+	retval = sys_read(fd, (char *)A(buf), nbyte);
+	if (file->f_pos != temp) {
+		if (!retval)
+			retval = sys_lseek(fd, temp, 0);
+		else
+			sys_lseek(fd, temp, 0);
+	}
+bad:
+	unlock_kernel();
+	return retval;
+}
+
+asmlinkage int solaris_pwrite(int fd, u32 buf, u32 nbyte, s32 offset)
+{
+	off_t temp;
+	int retval;
+	struct file * file;
+	long (*sys_write)(unsigned int, char *, unsigned long) =
+		(long (*)(unsigned int, char *, unsigned long))SYS(read);
+	long (*sys_lseek)(unsigned int, off_t, unsigned int) =
+		(long (*)(unsigned int, off_t, unsigned int))SYS(lseek);
+	
+	lock_kernel();
+	retval = -EBADF;
+	if (fd >= NR_OPEN ||
+            !(file = current->files->fd[fd]))
+		goto bad;
+	temp = file->f_pos;
+	if (temp != offset) {
+		retval = sys_lseek(fd, offset, 0);
+		if (retval < 0) goto bad;
+	}
+	retval = sys_write(fd, (char *)A(buf), nbyte);
+	if (file->f_pos != temp) {
+		if (!retval)
+			retval = sys_lseek(fd, temp, 0);
+		else
+			sys_lseek(fd, temp, 0);
+	}
+bad:
+	unlock_kernel();
+	return retval;
+}
+
+/* POSIX.1 names */
+#define _PC_LINK_MAX    1
+#define _PC_MAX_CANON   2
+#define _PC_MAX_INPUT   3
+#define _PC_NAME_MAX    4
+#define _PC_PATH_MAX    5
+#define _PC_PIPE_BUF    6
+#define _PC_NO_TRUNC    7
+#define _PC_VDISABLE    8
+#define _PC_CHOWN_RESTRICTED    9
+/* POSIX.4 names */
+#define _PC_ASYNC_IO    10
+#define _PC_PRIO_IO     11
+#define _PC_SYNC_IO     12
+#define _PC_LAST        12
+
+/* This is not a real and complete implementation yet, just to keep
+ * the easy Solaris binaries happy.
+ */
+asmlinkage int solaris_fpathconf(int fd, int name)
+{
+	int ret;
+
+	switch(name) {
+	case _PC_LINK_MAX:
+		ret = LINK_MAX;
+		break;
+	case _PC_MAX_CANON:
+		ret = MAX_CANON;
+		break;
+	case _PC_MAX_INPUT:
+		ret = MAX_INPUT;
+		break;
+	case _PC_NAME_MAX:
+		ret = NAME_MAX;
+		break;
+	case _PC_PATH_MAX:
+		ret = PATH_MAX;
+		break;
+	case _PC_PIPE_BUF:
+		ret = PIPE_BUF;
+		break;
+	case _PC_CHOWN_RESTRICTED:
+		ret = 1;
+		break;
+	case _PC_NO_TRUNC:
+	case _PC_VDISABLE:
+		ret = 0;
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+	return ret;
+}
+
+asmlinkage int solaris_pathconf(u32 path, int name)
+{
+	return solaris_fpathconf(0, name);
+}
+
+/* solaris_llseek returns long long - quite difficult */
+asmlinkage long solaris_llseek(struct pt_regs *regs, u32 off_hi, u32 off_lo, int whence)
+{
+	int (*sys_llseek)(unsigned int, unsigned long, unsigned long, loff_t *, unsigned int) =
+		(int (*)(unsigned int, unsigned long, unsigned long, loff_t *, unsigned int))SYS(_llseek);
+	int ret;
+	unsigned long old_fs = get_fs();
+	loff_t retval;
+	
+	set_fs(KERNEL_DS);
+	ret = sys_llseek((unsigned int)regs->u_regs[UREG_I0], off_hi, off_lo, &retval, whence);
+	set_fs(old_fs);
+	if (ret < 0) return ret;
+	regs->u_regs[UREG_I1] = (u32)retval;
+	return (retval >> 32);
+}
+
+/* Have to mask out all but lower 3 bits */
+asmlinkage int solaris_access(u32 filename, long mode)
+{
+	int (*sys_access)(const char *, int) = 
+		(int (*)(const char *, int))SYS(access);
+		
+	return sys_access((const char *)A(filename), mode & 7);
+}
diff --git a/arch/sparc64/solaris/ioctl.c b/arch/sparc64/solaris/ioctl.c
new file mode 100644
index 000000000..43b844caf
--- /dev/null
+++ b/arch/sparc64/solaris/ioctl.c
@@ -0,0 +1,316 @@
+/* $Id: ioctl.c,v 1.2 1997/09/04 00:59:22 davem Exp $
+ * ioctl.c: Solaris ioctl emulation.
+ *
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ *
+ * Streams & timod emulation based on code
+ * Copyright (C) 1995, 1996 Mike Jagdis (jaggy@purplet.demon.co.uk)
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/ioctl.h>
+#include <linux/fs.h>
+
+#include <asm/uaccess.h>
+#include <asm/termios.h>
+
+#include "conv.h"
+
+extern char * getname32(u32 filename);
+#define putname32 putname
+
+extern asmlinkage int sys_ioctl(unsigned int fd, unsigned int cmd, 
+	unsigned long arg);
+extern asmlinkage int sys32_ioctl(unsigned int fd, unsigned int cmd,
+	u32 arg);
+asmlinkage int solaris_ioctl(unsigned int fd, unsigned int cmd, u32 arg);
+
+/* termio* stuff {{{ */
+
+struct solaris_termios {
+	u32	c_iflag;
+	u32	c_oflag;
+	u32	c_cflag;
+	u32	c_lflag;
+	u8	c_cc[19];
+};
+
+struct solaris_termio {
+	u16	c_iflag;
+	u16	c_oflag;
+	u16	c_cflag;
+	u16	c_lflag;
+	s8	c_line;
+	u8	c_cc[8];
+};
+
+struct solaris_termiox {
+	u16	x_hflag;
+	u16	x_cflag;
+	u16	x_rflag[5];
+	u16	x_sflag;
+};
+
+static u32 solaris_to_linux_cflag(u32 cflag)
+{
+	cflag &= 0x7fdff000;
+	if (cflag & 0x200000) {
+		int baud = cflag & 0xf;
+		cflag &= ~0x20000f;
+		switch (baud) {
+		case 0: baud = B57600; break;
+		case 1: baud = B76800; break;
+		case 2: baud = B115200; break;
+		case 3: baud = B153600; break;
+		case 4: baud = B230400; break;
+		case 5: baud = B307200; break;
+		case 6: baud = B460800; break;
+		}
+		cflag |= CBAUDEX | baud;
+	}
+	return cflag;
+}
+
+static u32 linux_to_solaris_cflag(u32 cflag)
+{
+	cflag &= ~(CMSPAR | CIBAUD);
+	if (cflag & CBAUDEX) {
+		int baud = cflag & CBAUD;
+		cflag &= ~CBAUD;
+		switch (baud) {
+		case B57600: baud = 0; break;
+		case B76800: baud = 1; break;
+		case B115200: baud = 2; break;
+		case B153600: baud = 3; break;
+		case B230400: baud = 4; break;
+		case B307200: baud = 5; break;
+		case B460800: baud = 6; break;
+		case B614400: baud = 7; break;
+		case B921600: baud = 8; break;
+		/* case B1843200: baud = 9; break; */
+		}
+		cflag |= 0x200000 | baud;
+	}
+	return cflag;
+}
+
+static inline int linux_to_solaris_termio(unsigned int fd, unsigned int cmd, u32 arg)
+{
+	int ret;
+	
+	ret = sys_ioctl(fd, cmd, A(arg));
+	if (!ret) {
+		u32 cflag;
+		
+		if (__get_user (cflag, &((struct solaris_termio *)A(arg))->c_cflag))
+			return -EFAULT;
+		cflag = linux_to_solaris_cflag(cflag);
+		if (__put_user (cflag, &((struct solaris_termio *)A(arg))->c_cflag))
+			return -EFAULT;
+	}
+	return ret;
+}
+
+static int solaris_to_linux_termio(unsigned int fd, unsigned int cmd, u32 arg)
+{
+	int ret;
+	struct solaris_termio s;
+	unsigned long old_fs = get_fs();
+	
+	if (copy_from_user (&s, (struct solaris_termio *)A(arg), sizeof(struct solaris_termio)))
+		return -EFAULT;
+	s.c_cflag = solaris_to_linux_cflag(s.c_cflag);
+	set_fs(KERNEL_DS);
+	ret = sys_ioctl(fd, cmd, (unsigned long)&s);
+	set_fs(old_fs);
+	return ret;
+}
+
+static inline int linux_to_solaris_termios(unsigned int fd, unsigned int cmd, u32 arg)
+{
+	int ret;
+	struct solaris_termios s;
+	unsigned long old_fs = get_fs();
+
+	set_fs(KERNEL_DS);	
+	ret = sys_ioctl(fd, cmd, (unsigned long)&s);
+	set_fs(old_fs);
+	if (!ret) {
+		if (put_user (s.c_iflag, &((struct solaris_termios *)A(arg))->c_iflag) ||
+		    __put_user (s.c_oflag, &((struct solaris_termios *)A(arg))->c_oflag) ||
+		    __put_user (linux_to_solaris_cflag(s.c_cflag), &((struct solaris_termios *)A(arg))->c_cflag) ||
+		    __put_user (s.c_lflag, &((struct solaris_termios *)A(arg))->c_lflag) ||
+		    __copy_to_user (((struct solaris_termios *)A(arg))->c_cc, s.c_cc, 16) ||
+		    __clear_user (((struct solaris_termios *)A(arg))->c_cc + 16, 2))
+			return -EFAULT;
+	}
+	return ret;
+}
+
+static int solaris_to_linux_termios(unsigned int fd, unsigned int cmd, u32 arg)
+{
+	int ret;
+	struct solaris_termios s;
+	unsigned long old_fs = get_fs();
+
+	set_fs(KERNEL_DS);
+	ret = sys_ioctl(fd, TCGETS, (unsigned long)&s);
+	set_fs(old_fs);
+	if (ret) return ret;
+	if (put_user (s.c_iflag, &((struct solaris_termios *)A(arg))->c_iflag) ||
+	    __put_user (s.c_oflag, &((struct solaris_termios *)A(arg))->c_oflag) ||
+	    __put_user (s.c_cflag, &((struct solaris_termios *)A(arg))->c_cflag) ||
+	    __put_user (s.c_lflag, &((struct solaris_termios *)A(arg))->c_lflag) ||
+	    __copy_from_user (s.c_cc, ((struct solaris_termios *)A(arg))->c_cc, 16))
+		return -EFAULT;
+	s.c_cflag = solaris_to_linux_cflag(s.c_cflag);
+	set_fs(KERNEL_DS);
+	ret = sys_ioctl(fd, cmd, (unsigned long)&s);
+	set_fs(old_fs);
+	return ret;
+}
+
+static inline int solaris_T(unsigned int fd, unsigned int cmd, u32 arg)
+{
+	switch (cmd & 0xff) {
+	case 1: /* TCGETA */
+		return linux_to_solaris_termio(fd, TCGETA, arg);
+	case 2: /* TCSETA */
+		return solaris_to_linux_termio(fd, TCSETA, arg);
+	case 3: /* TCSETAW */
+		return solaris_to_linux_termio(fd, TCSETAW, arg);
+	case 4: /* TCSETAF */
+		return solaris_to_linux_termio(fd, TCSETAF, arg);
+	case 5: /* TCSBRK */
+		return sys_ioctl(fd, TCSBRK, arg);
+	case 6: /* TCXONC */
+		return sys_ioctl(fd, TCXONC, arg);
+	case 7: /* TCFLSH */
+		return sys_ioctl(fd, TCFLSH, arg);
+	case 13: /* TCGETS */
+		return linux_to_solaris_termios(fd, TCGETS, arg);
+	case 14: /* TCSETS */
+		return solaris_to_linux_termios(fd, TCSETS, arg);
+	case 15: /* TCSETSW */
+		return solaris_to_linux_termios(fd, TCSETSW, arg);
+	case 16: /* TCSETSF */
+		return solaris_to_linux_termios(fd, TCSETSF, arg);
+	case 103: /* TIOCSWINSZ */
+		return sys_ioctl(fd, TIOCSWINSZ, arg);
+	case 104: /* TIOCGWINSZ */
+		return sys_ioctl(fd, TIOCGWINSZ, arg);
+	}
+	return -ENOSYS;
+}
+
+static inline int solaris_t(unsigned int fd, unsigned int cmd, u32 arg)
+{
+	switch (cmd & 0xff) {
+	case 20: /* TIOCGPGRP */
+		return sys_ioctl(fd, TIOCGPGRP, arg);
+	case 21: /* TIOCSPGRP */
+		return sys_ioctl(fd, TIOCSPGRP, arg);
+	}
+	return -ENOSYS;
+}
+
+/* }}} */
+
+/* A pseudo STREAMS support {{{ */
+
+struct strioctl {
+	int cmd, timeout, len;
+	u32 data;
+};
+
+static inline int solaris_S(unsigned int fd, unsigned int cmd, u32 arg)
+{
+	char *p;
+	int ret;
+	unsigned long old_fs;
+	struct strioctl si;
+	
+	switch (cmd & 0xff) {
+	case 1: /* I_NREAD */
+		return -ENOSYS;
+	case 2: /* I_PUSH */
+		p = getname32 (arg);
+		if (IS_ERR (p))
+			return PTR_ERR(p);
+		putname32 (p);
+		return 0;
+	case 3: /* I_POP */
+		return 0;
+	case 5: /* I_FLUSH */
+		return 0;
+	case 8: /* I_STR */
+		if (copy_from_user (&si, (struct strioctl *)A(arg), sizeof(struct strioctl)))
+			return -EFAULT;
+		switch ((si.cmd >> 8) & 0xff) {
+		case 'T':
+		default:
+			return solaris_ioctl(fd, si.cmd, si.data);
+		}
+	case 9: /* I_SETSIG */
+		return sys_ioctl(fd, FIOSETOWN, current->pid);
+	case 10: /* I_GETSIG */
+		old_fs = get_fs();
+		set_fs(KERNEL_DS);
+		sys_ioctl(fd, FIOGETOWN, (unsigned long)&ret);
+		set_fs(old_fs);
+		if (ret == current->pid) return 0x3ff;
+		else return -EINVAL;
+	case 11: /* I_FIND */
+		p = getname32 (arg);
+		if (IS_ERR (p))
+			return PTR_ERR(p);
+		ret = !strcmp(p, "timod");
+		putname32 (p);
+		return ret;
+	}
+	return -ENOSYS;
+}
+/* }}} */
+
+asmlinkage int solaris_ioctl(unsigned int fd, unsigned int cmd, u32 arg)
+{
+	struct file * filp;
+	int error = -EBADF;
+
+	lock_kernel();
+	if(fd >= NR_OPEN) goto out;
+
+	filp = current->files->fd[fd];
+	if(!filp) goto out;
+
+	if (!filp->f_op || !filp->f_op->ioctl) {
+		error = sys_ioctl (fd, cmd, (unsigned long)arg);
+		goto out;
+	}
+	
+	error = -EFAULT;
+	switch ((cmd >> 8) & 0xff) {
+	case 'S': error = solaris_S(fd, cmd, arg); break;
+	case 'T': error = solaris_T(fd, cmd, arg); break;
+	case 't': error = solaris_t(fd, cmd, arg); break;
+	default:
+		error = -ENOSYS;
+		break;
+	}
+out:
+	if (error == -ENOSYS) {
+		unsigned char c = cmd>>8;
+		
+		if (c < ' ' || c > 126) c = '.';
+		printk("solaris_ioctl: Unknown cmd fd(%d) cmd(%08x '%c') arg(%08x)\n",
+		       (int)fd, (unsigned int)cmd, c, (unsigned int)arg);
+		error = -EINVAL;
+	}
+	unlock_kernel();
+	return error;
+}
diff --git a/arch/sparc64/solaris/ipc.c b/arch/sparc64/solaris/ipc.c
new file mode 100644
index 000000000..132028447
--- /dev/null
+++ b/arch/sparc64/solaris/ipc.c
@@ -0,0 +1,122 @@
+/* $Id: ipc.c,v 1.1 1997/09/03 12:29:29 jj Exp $
+ * ipc.c: Solaris IPC emulation
+ *
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ */
+
+#include <linux/types.h>
+#include <linux/smp_lock.h>
+#include <linux/shm.h>
+#include <linux/sem.h>
+#include <linux/msg.h>
+
+#include <asm/uaccess.h>
+#include <asm/string.h>
+#include <asm/ipc.h>
+
+#include "conv.h"
+
+struct solaris_ipc_perm {
+	s32	uid;
+	s32	gid;
+	s32	cuid;
+	s32	cgid;
+	u32	mode;
+	u32	seq;
+	int	key;
+	s32	pad[4];
+};
+
+struct solaris_shmid_ds {
+	struct solaris_ipc_perm	shm_perm;
+	int			shm_segsz;
+	u32			shm_amp;
+	unsigned short		shm_lkcnt;
+	char			__padxx[2];
+	s32			shm_lpid;
+	s32			shm_cpid;
+	u32			shm_nattch;
+	u32			shm_cnattch;
+	s32			shm_atime;
+	s32			shm_pad1;
+	s32			shm_dtime;
+	s32			shm_pad2;
+	s32			shm_ctime;
+	s32			shm_pad3;
+	unsigned short		shm_cv;
+	char			shm_pad4[2];
+	u32			shm_sptas;
+	s32			shm_pad5[2];
+};
+
+asmlinkage long solaris_shmsys(int cmd, u32 arg1, u32 arg2, u32 arg3)
+{
+	int (*sys_ipc)(unsigned,int,int,unsigned long,void *,long) = 
+		(int (*)(unsigned,int,int,unsigned long,void *,long))SYS(ipc);
+	unsigned long old_fs;
+	unsigned long raddr;
+	int ret;
+		
+	switch (cmd) {
+	case 0: /* shmat */
+		old_fs = get_fs();
+		set_fs(KERNEL_DS);
+		ret = sys_ipc(SHMAT, arg1, arg3 & ~0x4000, (unsigned long)&raddr, (void *)A(arg2), 0);
+		set_fs(old_fs);
+		if (ret >= 0) return (u32)raddr;
+		else return ret;
+	case 1: /* shmctl */
+		switch (arg2) {
+		case 3: /* SHM_LOCK */
+		case 4: /* SHM_UNLOCK */
+			return sys_ipc(SHMCTL, arg1, (arg2 == 3) ? SHM_LOCK : SHM_UNLOCK, 0, NULL, 0);
+		case 10: /* IPC_RMID */
+			return sys_ipc(SHMCTL, arg1, IPC_RMID, 0, NULL, 0);
+		case 11: /* IPC_SET */
+			{
+				struct shmid_ds s;
+				
+				if (get_user (s.shm_perm.uid, &(((struct solaris_shmid_ds *)A(arg3))->shm_perm.uid)) ||
+				    __get_user (s.shm_perm.gid, &(((struct solaris_shmid_ds *)A(arg3))->shm_perm.gid)) || 
+				    __get_user (s.shm_perm.mode, &(((struct solaris_shmid_ds *)A(arg3))->shm_perm.mode)))
+					return -EFAULT;
+				old_fs = get_fs();
+				set_fs(KERNEL_DS);
+				ret = sys_ipc(SHMCTL, arg1, IPC_SET, 0, &s, 0);
+				set_fs(old_fs);
+				return ret;
+			}
+		case 12: /* IPC_STAT */
+			{
+				struct shmid_ds s;
+				
+				old_fs = get_fs();
+				set_fs(KERNEL_DS);
+				ret = sys_ipc(SHMCTL, arg1, IPC_SET, 0, &s, 0);
+				set_fs(old_fs);
+				if (get_user (s.shm_perm.uid, &(((struct solaris_shmid_ds *)A(arg3))->shm_perm.uid)) ||
+				    __get_user (s.shm_perm.gid, &(((struct solaris_shmid_ds *)A(arg3))->shm_perm.gid)) || 
+				    __get_user (s.shm_perm.cuid, &(((struct solaris_shmid_ds *)A(arg3))->shm_perm.cuid)) ||
+				    __get_user (s.shm_perm.cgid, &(((struct solaris_shmid_ds *)A(arg3))->shm_perm.cgid)) || 
+				    __get_user (s.shm_perm.mode, &(((struct solaris_shmid_ds *)A(arg3))->shm_perm.mode)) ||
+				    __get_user (s.shm_perm.seq, &(((struct solaris_shmid_ds *)A(arg3))->shm_perm.seq)) ||
+				    __get_user (s.shm_perm.key, &(((struct solaris_shmid_ds *)A(arg3))->shm_perm.key)) ||
+				    __get_user (s.shm_segsz, &(((struct solaris_shmid_ds *)A(arg3))->shm_segsz)) ||
+				    __get_user (s.shm_lpid, &(((struct solaris_shmid_ds *)A(arg3))->shm_lpid)) ||
+				    __get_user (s.shm_cpid, &(((struct solaris_shmid_ds *)A(arg3))->shm_cpid)) ||
+				    __get_user (s.shm_nattch, &(((struct solaris_shmid_ds *)A(arg3))->shm_nattch)) ||
+				    __get_user (s.shm_atime, &(((struct solaris_shmid_ds *)A(arg3))->shm_atime)) ||
+				    __get_user (s.shm_dtime, &(((struct solaris_shmid_ds *)A(arg3))->shm_dtime)) ||
+				    __get_user (s.shm_ctime, &(((struct solaris_shmid_ds *)A(arg3))->shm_ctime)))
+					return -EFAULT;
+				return ret;
+			}
+		default: return -EINVAL;
+		}
+	case 2: /* shmdt */
+		return sys_ipc(SHMDT, 0, 0, 0, (void *)A(arg1), 0);
+	case 3: /* shmget */
+		return sys_ipc(SHMGET, arg1, arg2, arg3, NULL, 0);
+	}
+	return -EINVAL;
+}
diff --git a/arch/sparc64/solaris/misc.c b/arch/sparc64/solaris/misc.c
new file mode 100644
index 000000000..fdabd0e26
--- /dev/null
+++ b/arch/sparc64/solaris/misc.c
@@ -0,0 +1,463 @@
+/* $Id: misc.c,v 1.4 1997/09/04 14:57:31 jj Exp $
+ * misc.c: Miscelaneous syscall emulation for Solaris
+ *
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ */
+
+#include <linux/module.h> 
+#include <linux/types.h>
+#include <linux/smp_lock.h>
+#include <linux/utsname.h>
+#include <linux/limits.h>
+#include <linux/mm.h>
+#include <linux/smp.h>
+
+#include <asm/uaccess.h>
+#include <asm/string.h>
+#include <asm/oplib.h>
+#include <asm/idprom.h>
+#include <asm/machines.h>
+
+#include "conv.h"
+
+/* Conversion from Linux to Solaris errnos. 0-34 are identity mapped.
+   Some Linux errnos (EPROCLIM, EDOTDOT, ERREMOTE, EUCLEAN, ENOTNAM, 
+   ENAVAIL, EISNAM, EREMOTEIO, ENOMEDIUM, EMEDIUMTYPE) have no Solaris
+   equivalents. I return EINVAL in that case, which is very wrong. If
+   someone suggest a better value for them, you're welcomed.
+   On the other side, Solaris ECANCELED and ENOTSUP have no Linux equivalents,
+   but that doesn't matter here. --jj */
+int solaris_err_table[] = {
+/* 0 */  0, 1, 2, 3, 4, 5, 6, 7, 8, 9,
+/* 10 */  10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
+/* 20 */  20, 21, 22, 23, 24, 25, 26, 27, 28, 29,
+/* 30 */  30, 31, 32, 33, 34, 22, 150, 149, 95, 96,
+/* 40 */  97, 98, 99, 120, 121, 122, 123, 124, 125, 126, 
+/* 50 */ 127, 128, 129, 130, 131, 132, 133, 134, 143, 144,
+/* 60 */ 145, 146, 90, 78, 147, 148, 93, 22, 94, 49,
+/* 70 */ 151, 66, 60, 62, 63, 35, 77, 36, 45, 46, 
+/* 80 */ 64, 22, 67, 68, 69, 70, 71, 74, 22, 82, 
+/* 90 */ 89, 92, 79, 81, 37, 38, 39, 40, 41, 42,
+/* 100 */ 43, 44, 50, 51, 52, 53, 54, 55, 56, 57,
+/* 110 */ 87, 61, 84, 65, 83, 80, 91, 22, 22, 22,
+/* 120 */ 22, 22, 88, 86, 85, 22, 22,
+};
+
+asmlinkage u32 solaris_mmap(u32 addr, u32 len, u32 prot, u32 flags, u32 fd, u32 off)
+{
+	u32 (*sunos_mmap)(u32,u32,u32,u32,u32,u32) = 
+		(u32 (*)(u32,u32,u32,u32,u32,u32))SUNOS(71);
+	u32 ret;
+	
+	ret = sunos_mmap(addr,len,prot,flags,fd,off);
+	/* sunos_mmap sets personality to PER_BSD */
+	current->personality = PER_SVR4;
+	return ret;
+}
+
+asmlinkage int solaris_brk(u32 brk)
+{
+	int (*sunos_brk)(u32) = (int (*)(u32))SUNOS(17);
+	
+	return sunos_brk(brk);
+}
+
+#define set_utsfield(to, from, dotchop, countfrom) {			\
+	char *p; 							\
+	int i, len = (countfrom) ? 					\
+		((sizeof(to) > sizeof(from) ? 				\
+			sizeof(from) : sizeof(to))) : sizeof(to); 	\
+	copy_to_user_ret(to, from, len, -EFAULT); 			\
+	if (dotchop) 							\
+		for (p=from,i=0; *p && *p != '.' && --len; p++,i++); 	\
+	else 								\
+		i = len - 1; 						\
+	__put_user_ret('\0', (char *)(to+i), -EFAULT); 			\
+}
+
+struct sol_uname {
+	char sysname[9];
+	char nodename[9];
+	char release[9];
+	char version[9];
+	char machine[9];
+};
+
+struct sol_utsname {
+	char sysname[257];
+	char nodename[257];
+	char release[257];
+	char version[257];
+	char machine[257];
+};
+
+static char *machine(void)
+{
+	switch (sparc_cpu_model) {
+	case sun4: return "sun4";
+	case sun4c: return "sun4c";
+	case sun4e: return "sun4e";
+	case sun4m: return "sun4m";
+	case sun4d: return "sun4d";
+	case sun4u: return "sun4u";
+	default: return "sparc";
+	}
+}
+
+static char *platform(char *buffer)
+{
+	int i;
+	struct {
+		char *platform;
+		int id_machtype;
+	} platforms [] = {
+		{ "sun4", (SM_SUN4 | SM_4_110) },
+		{ "sun4", (SM_SUN4 | SM_4_260) },
+		{ "sun4", (SM_SUN4 | SM_4_330) },
+		{ "sun4", (SM_SUN4 | SM_4_470) },
+		{ "SUNW,Sun_4_60", (SM_SUN4C | SM_4C_SS1) },
+		{ "SUNW,Sun_4_40", (SM_SUN4C | SM_4C_IPC) },
+		{ "SUNW,Sun_4_65", (SM_SUN4C | SM_4C_SS1PLUS) },
+		{ "SUNW,Sun_4_20", (SM_SUN4C | SM_4C_SLC) },
+		{ "SUNW,Sun_4_75", (SM_SUN4C | SM_4C_SS2) },
+		{ "SUNW,Sun_4_25", (SM_SUN4C | SM_4C_ELC) },
+		{ "SUNW,Sun_4_50", (SM_SUN4C | SM_4C_IPX) },
+		{ "SUNW,Sun_4_600", (SM_SUN4M | SM_4M_SS60) },
+		{ "SUNW,SPARCstation-5", (SM_SUN4M | SM_4M_SS50) },
+		{ "SUNW,SPARCstation-20", (SM_SUN4M | SM_4M_SS40) }
+	};
+
+	*buffer = 0;
+	prom_getproperty(prom_root_node, "name", buffer, 256);
+	if (*buffer) {
+		char *p;
+
+		for (p = buffer; *p; p++)
+			if (*p == '/' || *p == ' ') *p = '_';
+		return buffer;
+	}
+	for (i = 0; i < sizeof (platforms)/sizeof (platforms[0]); i++)
+		if (platforms[i].id_machtype == idprom->id_machtype)
+			return platforms[i].platform;
+	return "sun4c";
+}
+
+static char *serial(char *buffer)
+{
+	int node = prom_getchild(prom_root_node);
+
+	node = prom_searchsiblings(node, "options");
+	*buffer = 0;
+	prom_getproperty(node, "system-board-serial#", buffer, 256);
+	if (!*buffer)
+		return "4512348717234";
+	else
+		return buffer;
+}
+
+asmlinkage int solaris_utssys(u32 buf, u32 flags, int which, u32 buf2)
+{
+	switch (which) {
+	case 0:	/* old uname */
+		/* Lets cheat */
+		set_utsfield(((struct sol_uname *)A(buf))->sysname, 
+			"SunOS", 1, 0);
+		set_utsfield(((struct sol_uname *)A(buf))->nodename, 
+			system_utsname.nodename, 1, 1);
+		set_utsfield(((struct sol_uname *)A(buf))->release, 
+			"2.6", 0, 0);
+		set_utsfield(((struct sol_uname *)A(buf))->version, 
+			"Generic", 0, 0);
+		set_utsfield(((struct sol_uname *)A(buf))->machine, 
+			machine(), 0, 0);
+		return 0;
+	case 2: /* ustat */
+		return -ENOSYS;
+	case 3: /* fusers */
+		return -ENOSYS;
+	default:
+		return -ENOSYS;
+	}
+}
+
+asmlinkage int solaris_utsname(u32 buf)
+{
+	/* Why should we not lie a bit? */
+	set_utsfield(((struct sol_utsname *)A(buf))->sysname, 
+			"SunOS", 0, 0);
+	set_utsfield(((struct sol_utsname *)A(buf))->nodename, 
+			system_utsname.nodename, 1, 1);
+	set_utsfield(((struct sol_utsname *)A(buf))->release, 
+			"5.6", 0, 0);
+	set_utsfield(((struct sol_utsname *)A(buf))->version, 
+			"Generic", 0, 0);
+	set_utsfield(((struct sol_utsname *)A(buf))->machine, 
+			machine(), 0, 0);
+	return 0;
+}
+
+#define SI_SYSNAME		1       /* return name of operating system */
+#define SI_HOSTNAME		2       /* return name of node */
+#define SI_RELEASE		3       /* return release of operating system */
+#define SI_VERSION		4       /* return version field of utsname */
+#define SI_MACHINE		5       /* return kind of machine */
+#define SI_ARCHITECTURE		6       /* return instruction set arch */
+#define SI_HW_SERIAL		7       /* return hardware serial number */
+#define SI_HW_PROVIDER		8       /* return hardware manufacturer */
+#define SI_SRPC_DOMAIN		9       /* return secure RPC domain */
+#define SI_PLATFORM		513     /* return platform identifier */
+
+asmlinkage int solaris_sysinfo(int cmd, u32 buf, s32 count)
+{
+	char *p, *q, *r;
+	char buffer[256];
+	int len;
+	
+	/* Again, we cheat :)) */
+	switch (cmd) {
+	case SI_SYSNAME: r = "SunOS"; break;
+	case SI_HOSTNAME:
+		r = buffer + 256;
+		for (p = system_utsname.nodename, q = buffer; 
+		     q < r && *p && *p != '.'; *q++ = *p++);
+		*q = 0;
+		r = buffer;
+		break;
+	case SI_RELEASE: r = "5.6"; break;
+	case SI_MACHINE: r = machine(); break;
+	case SI_ARCHITECTURE: r = "sparc"; break;
+	case SI_HW_PROVIDER: r = "Sun_Microsystems"; break;
+	case SI_HW_SERIAL: r = serial(buffer); break;
+	case SI_PLATFORM: r = platform(buffer); break;
+	case SI_SRPC_DOMAIN: r = ""; break;
+	case SI_VERSION: r = "Generic"; break;
+	default: return -EINVAL;
+	}
+	len = strlen(r) + 1;
+	if (count < len) {
+		copy_to_user_ret((char *)A(buf), r, count - 1, -EFAULT);
+		__put_user_ret(0, (char *)A(buf) + count - 1, -EFAULT);
+	} else
+		copy_to_user_ret((char *)A(buf), r, len, -EFAULT);
+	return len;
+}
+
+#define	SOLARIS_CONFIG_NGROUPS			2
+#define	SOLARIS_CONFIG_CHILD_MAX		3
+#define	SOLARIS_CONFIG_OPEN_FILES		4
+#define	SOLARIS_CONFIG_POSIX_VER		5
+#define	SOLARIS_CONFIG_PAGESIZE			6
+#define	SOLARIS_CONFIG_CLK_TCK			7
+#define	SOLARIS_CONFIG_XOPEN_VER		8
+#define	SOLARIS_CONFIG_PROF_TCK			10
+#define	SOLARIS_CONFIG_NPROC_CONF		11
+#define	SOLARIS_CONFIG_NPROC_ONLN		12
+#define	SOLARIS_CONFIG_AIO_LISTIO_MAX		13
+#define	SOLARIS_CONFIG_AIO_MAX			14
+#define	SOLARIS_CONFIG_AIO_PRIO_DELTA_MAX	15
+#define	SOLARIS_CONFIG_DELAYTIMER_MAX		16
+#define	SOLARIS_CONFIG_MQ_OPEN_MAX		17
+#define	SOLARIS_CONFIG_MQ_PRIO_MAX		18
+#define	SOLARIS_CONFIG_RTSIG_MAX		19
+#define	SOLARIS_CONFIG_SEM_NSEMS_MAX		20
+#define	SOLARIS_CONFIG_SEM_VALUE_MAX		21
+#define	SOLARIS_CONFIG_SIGQUEUE_MAX		22
+#define	SOLARIS_CONFIG_SIGRT_MIN		23
+#define	SOLARIS_CONFIG_SIGRT_MAX		24
+#define	SOLARIS_CONFIG_TIMER_MAX		25
+#define	SOLARIS_CONFIG_PHYS_PAGES		26
+#define	SOLARIS_CONFIG_AVPHYS_PAGES		27
+
+asmlinkage int solaris_sysconf(int id)
+{
+	switch (id) {
+	case SOLARIS_CONFIG_NGROUPS:	return NGROUPS_MAX;
+	case SOLARIS_CONFIG_CHILD_MAX:	return CHILD_MAX;
+	case SOLARIS_CONFIG_OPEN_FILES:	return OPEN_MAX;
+	case SOLARIS_CONFIG_POSIX_VER:	return 199309;
+	case SOLARIS_CONFIG_PAGESIZE:	return PAGE_SIZE;
+	case SOLARIS_CONFIG_XOPEN_VER:	return 3;
+	case SOLARIS_CONFIG_CLK_TCK:
+	case SOLARIS_CONFIG_PROF_TCK:
+		return prom_getintdefault(
+			linux_cpus[smp_processor_id()].prom_node,
+			"clock-frequency", 167000000);
+#ifdef __SMP__	
+	case SOLARIS_CONFIG_NPROC_CONF:	return NCPUS;
+	case SOLARIS_CONFIG_NPROC_ONLN:	return smp_num_cpus;
+#else
+	case SOLARIS_CONFIG_NPROC_CONF:	return 1;
+	case SOLARIS_CONFIG_NPROC_ONLN:	return 1;
+#endif
+	case SOLARIS_CONFIG_SIGRT_MIN:		return 37;
+	case SOLARIS_CONFIG_SIGRT_MAX:		return 44;
+	case SOLARIS_CONFIG_PHYS_PAGES:
+	case SOLARIS_CONFIG_AVPHYS_PAGES:
+		{
+			struct sysinfo s;
+			
+			si_meminfo(&s);
+			if (id == SOLARIS_CONFIG_PHYS_PAGES)
+				return s.totalram >>= PAGE_SHIFT;
+			else
+				return s.freeram >>= PAGE_SHIFT;
+		}
+	/* XXX support these as well -jj */
+	case SOLARIS_CONFIG_AIO_LISTIO_MAX:	return -EINVAL;
+	case SOLARIS_CONFIG_AIO_MAX:		return -EINVAL;
+	case SOLARIS_CONFIG_AIO_PRIO_DELTA_MAX:	return -EINVAL;
+	case SOLARIS_CONFIG_DELAYTIMER_MAX:	return -EINVAL;
+	case SOLARIS_CONFIG_MQ_OPEN_MAX:	return -EINVAL;
+	case SOLARIS_CONFIG_MQ_PRIO_MAX:	return -EINVAL;
+	case SOLARIS_CONFIG_RTSIG_MAX:		return -EINVAL;
+	case SOLARIS_CONFIG_SEM_NSEMS_MAX:	return -EINVAL;
+	case SOLARIS_CONFIG_SEM_VALUE_MAX:	return -EINVAL;
+	case SOLARIS_CONFIG_SIGQUEUE_MAX:	return -EINVAL;
+	case SOLARIS_CONFIG_TIMER_MAX:		return -EINVAL;
+	default: return -EINVAL;
+	}
+}
+
+asmlinkage int solaris_procids(int cmd, s32 pid, s32 pgid)
+{
+	int ret;
+	
+	switch (cmd) {
+	case 0: /* getpgrp */
+		return current->pgrp;
+	case 1: /* setpgrp */
+		{
+			int (*sys_setpgid)(pid_t,pid_t) =
+				(int (*)(pid_t,pid_t))SYS(setpgid);
+				
+			/* can anyone explain me the difference between
+			   Solaris setpgrp and setsid? */
+			ret = sys_setpgid(0, 0);
+			if (ret) return ret;
+			current->tty = NULL;
+			return current->pgrp;
+		}
+	case 2: /* getsid */
+		{
+			int (*sys_getsid)(pid_t) = (int (*)(pid_t))SYS(getsid);
+			return sys_getsid(pid);
+		}
+	case 3: /* setsid */
+		{
+			int (*sys_setsid)(void) = (int (*)(void))SYS(setsid);
+			return sys_setsid();
+		}
+	case 4: /* getpgid */
+		{
+			int (*sys_getpgid)(pid_t) = (int (*)(pid_t))SYS(getpgid);
+			return sys_getpgid(pid);
+		}
+	case 5: /* setpgid */
+		{
+			int (*sys_setpgid)(pid_t,pid_t) = 
+				(int (*)(pid_t,pid_t))SYS(setpgid);
+			return sys_setpgid(pid,pgid);
+		}
+	}
+	return -EINVAL;
+}
+
+asmlinkage int do_sol_unimplemented(struct pt_regs *regs)
+{
+	printk ("Unimplemented Solaris syscall %d %08x %08x %08x %08x\n", 
+			(int)regs->u_regs[UREG_G1], 
+			(int)regs->u_regs[UREG_I0],
+			(int)regs->u_regs[UREG_I1],
+			(int)regs->u_regs[UREG_I2],
+			(int)regs->u_regs[UREG_I3]);
+	return -ENOSYS;
+}
+
+asmlinkage void solaris_register(void)
+{
+	lock_kernel();
+	current->personality = PER_SVR4;
+	if (current->exec_domain && current->exec_domain->module)
+		__MOD_DEC_USE_COUNT(current->exec_domain->module);
+	current->exec_domain = lookup_exec_domain(current->personality);
+	if (current->exec_domain && current->exec_domain->module)
+		__MOD_INC_USE_COUNT(current->exec_domain->module);
+	unlock_kernel();
+}
+
+extern long solaris_to_linux_signals[], linux_to_solaris_signals[];
+
+struct exec_domain solaris_exec_domain = {
+	"Solaris",
+	(lcall7_func)NULL,
+	1, 1,	/* PER_SVR4 personality */
+	solaris_to_linux_signals,
+	linux_to_solaris_signals,
+#ifdef MODULE
+	&__this_module,
+#else
+	NULL,
+#endif
+	NULL
+};
+
+#ifdef MODULE
+
+MODULE_AUTHOR("Jakub Jelinek (jj@sunsite.mff.cuni.cz)");
+MODULE_DESCRIPTION("Solaris binary emulation module");
+
+#ifdef __sparc_v9__
+extern u32 tl0_solaris[8];
+#define update_ttable(x) 										\
+	tl0_solaris[3] = (((long)(x) - (long)tl0_solaris - 3) >> 2) | 0x40000000;			\
+	__asm__ __volatile__ ("membar #StoreStore; flush %0" : : "r" (&tl0_solaris[3]))
+#else
+#endif	
+
+extern u32 solaris_sparc_syscall[];
+extern u32 solaris_syscall[];
+extern int init_socksys(void);
+extern void cleanup_socksys(void);
+
+int init_module(void)
+{
+	int ret;
+	register_exec_domain(&solaris_exec_domain);
+	if ((ret = init_socksys())) {
+		unregister_exec_domain(&solaris_exec_domain);
+		return ret;
+	}
+	update_ttable(solaris_sparc_syscall);
+	return 0;
+}
+
+void cleanup_module(void)
+{
+	update_ttable(solaris_syscall);
+	cleanup_socksys();
+	unregister_exec_domain(&solaris_exec_domain);
+}
+
+#else
+int init_solaris_emul(void)
+{
+	register_exec_domain(&solaris_exec_domain);
+	init_socksys();
+}
+#endif
+
+#ifdef DEBUG_SOLARIS
+void entry_printk(int sysno, struct pt_regs *regs)
+{
+	printk ("Entering %d\n", sysno);
+	printk ("%08x %08x %08x %08x\n", (int)regs->u_regs[UREG_I0],
+					 (int)regs->u_regs[UREG_I1],
+					 (int)regs->u_regs[UREG_I2],
+					 (int)regs->u_regs[UREG_I3]);
+}
+
+void exit_printk(unsigned long ret)
+{
+	printk ("Returning %016lx\n", ret);
+}
+#endif
diff --git a/arch/sparc64/solaris/signal.c b/arch/sparc64/solaris/signal.c
new file mode 100644
index 000000000..6d3081c03
--- /dev/null
+++ b/arch/sparc64/solaris/signal.c
@@ -0,0 +1,419 @@
+/* $Id: signal.c,v 1.2 1997/09/03 12:29:19 jj Exp $
+ * signal.c: Signal emulation for Solaris
+ *
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ */
+
+#include <linux/types.h>
+#include <linux/smp_lock.h>
+
+#include <asm/uaccess.h>
+#include <asm/svr4.h>
+#include <asm/string.h>
+
+#include "conv.h"
+#include "signal.h"
+
+#define _S(nr) (1L<<((nr)-1))
+
+#define _BLOCKABLE (~(_S(SIGKILL) | _S(SIGSTOP)))
+
+long linux_to_solaris_signals[] = {
+        0,
+	SOLARIS_SIGHUP,		SOLARIS_SIGINT,	
+	SOLARIS_SIGQUIT,	SOLARIS_SIGILL,
+	SOLARIS_SIGTRAP,	SOLARIS_SIGIOT,
+	SOLARIS_SIGEMT,		SOLARIS_SIGFPE,
+	SOLARIS_SIGKILL,	SOLARIS_SIGBUS,
+	SOLARIS_SIGSEGV,	SOLARIS_SIGSYS,
+	SOLARIS_SIGPIPE,	SOLARIS_SIGALRM,
+	SOLARIS_SIGTERM,	SOLARIS_SIGURG,
+	SOLARIS_SIGSTOP,	SOLARIS_SIGTSTP,
+	SOLARIS_SIGCONT,	SOLARIS_SIGCLD,
+	SOLARIS_SIGTTIN,	SOLARIS_SIGTTOU,
+	SOLARIS_SIGPOLL,	SOLARIS_SIGXCPU,
+	SOLARIS_SIGXFSZ,	SOLARIS_SIGVTALRM,
+	SOLARIS_SIGPROF,	SOLARIS_SIGWINCH,
+	SOLARIS_SIGUSR1,	SOLARIS_SIGUSR1,
+	SOLARIS_SIGUSR2,	-1,
+};
+
+long solaris_to_linux_signals[] = {
+        0,
+        SIGHUP,		SIGINT,		SIGQUIT,	SIGILL,
+        SIGTRAP,	SIGIOT,		SIGEMT,		SIGFPE,
+        SIGKILL,	SIGBUS,		SIGSEGV,	SIGSYS,
+        SIGPIPE,	SIGALRM,	SIGTERM,	SIGUSR1,
+        SIGUSR2,	SIGCHLD,	-1,		SIGWINCH,
+        SIGURG,		SIGPOLL,	SIGSTOP,	SIGTSTP,
+        SIGCONT,	SIGTTIN,	SIGTTOU,	SIGVTALRM,
+        SIGPROF,	SIGXCPU,	SIGXFSZ,        -1,
+	-1,		-1,		-1,		-1,
+	-1,		-1,		-1,		-1,
+	-1,		-1,		-1,		-1,
+};
+
+static inline long mapsig(long sig)
+{
+	if ((unsigned long)sig > SOLARIS_NSIGNALS)
+		return -EINVAL;
+	return solaris_to_linux_signals[sig];
+}
+
+asmlinkage int solaris_kill(int pid, int sig)
+{
+	int (*sys_kill)(int,int) = 
+		(int (*)(int,int))SYS(kill);
+	int s = mapsig(sig);
+	
+	if (s < 0) return s;
+	return sys_kill(pid, s);
+}
+
+static long sig_handler(int sig, u32 arg, int one_shot)
+{
+	struct sigaction sa, old;
+	int ret;
+	unsigned long old_fs = get_fs();
+	int (*sys_sigaction)(int,struct sigaction *,struct sigaction *) = 
+		(int (*)(int,struct sigaction *,struct sigaction *))SYS(sigaction);
+	
+	sa.sa_mask = 0L;
+	sa.sa_restorer = NULL;
+	sa.sa_handler = (__sighandler_t)A(arg);
+	sa.sa_flags = 0;
+	if (one_shot) sa.sa_flags = SA_ONESHOT | SA_NOMASK;
+	set_fs (KERNEL_DS);
+	ret = sys_sigaction(sig, &sa, &old);
+	set_fs (old_fs);
+	if (ret < 0) return ret;
+	return (u32)(long)old.sa_handler;
+}
+
+static inline long solaris_signal(int sig, u32 arg)
+{
+	return sig_handler (sig, arg, 1);
+}
+
+static long solaris_sigset(int sig, u32 arg)
+{
+	if (arg != 2) /* HOLD */ {
+		spin_lock_irq(&current->sigmask_lock);
+		current->blocked &= ~_S(sig);
+		spin_unlock_irq(&current->sigmask_lock);
+		return sig_handler (sig, arg, 0);
+	} else {
+		sigset_t n = _S(sig) & _BLOCKABLE;
+		spin_lock_irq(&current->sigmask_lock);
+		current->blocked |= n;
+		spin_unlock_irq(&current->sigmask_lock);
+		return 0;
+	}
+}
+
+static inline long solaris_sighold(int sig)
+{
+	return solaris_sigset(sig, 2);
+}
+
+static inline long solaris_sigrelse(int sig)
+{
+	spin_lock_irq(&current->sigmask_lock);
+	current->blocked &= ~_S(sig);
+	spin_unlock_irq(&current->sigmask_lock);
+	return 0;
+}
+
+static inline long solaris_sigignore(int sig)
+{
+	return sig_handler (sig, (u32)SIG_IGN, 0);
+}
+
+static inline long solaris_sigpause(int sig)
+{
+	printk ("Need to support solaris sigpause\n");
+	return -ENOSYS;
+}
+
+asmlinkage long solaris_sigfunc(int sig, u32 arg)
+{
+	int func = sig & ~0xff;
+	
+	sig = mapsig(sig & 0xff); 
+	if (sig < 0) return sig; 
+	switch (func) {
+	case 0: return solaris_signal(sig, arg); 
+	case 0x100: return solaris_sigset(sig, arg); 
+	case 0x200: return solaris_sighold(sig);
+	case 0x400: return solaris_sigrelse(sig); 
+	case 0x800: return solaris_sigignore(sig); 
+	case 0x1000: return solaris_sigpause(sig);
+	}
+	return -EINVAL;
+}
+
+typedef struct {
+	u32 __sigbits[4];
+} sol_sigset_t;
+
+static inline int mapin(u32 *p, sigset_t *q)
+{
+	int i;
+	u32 x;
+	int sig;
+	
+	*q = 0L;
+	x = p[0];
+	for (i = 1; i <= SOLARIS_NSIGNALS; i++) {
+		if (x & 1) {
+			sig = solaris_to_linux_signals[i];
+			if (sig == -1)
+				return -EINVAL;
+			*q |= 1L << (sig - 1);
+		}
+		x >>= 1;
+		if (i == 32)
+			x = p[1];
+	}
+	return 0;
+}
+
+static inline int mapout(sigset_t *q, u32 *p)
+{
+	int i;
+	sigset_t x;
+	int sig;
+	
+	p[0] = 0;
+	p[1] = 0;
+	x = *q;
+	for (i = 1; i <= 32; i++, x >>= 1) {
+		if (x & 1) {
+			sig = linux_to_solaris_signals[i];
+			if (sig == -1)
+				return -EINVAL;
+			if (sig > 32)
+				p[1] |= 1L << (sig - 33);
+			else
+				p[0] |= 1L << (sig - 1);
+		}
+	}
+	return 0;
+	
+}
+
+asmlinkage int solaris_sigprocmask(int how, u32 in, u32 out)
+{
+	sigset_t in_s, *ins, out_s, *outs;
+	unsigned long old_fs = get_fs();
+	int ret;
+	int (*sys_sigprocmask)(int,sigset_t *,sigset_t *) = 
+		(int (*)(int,sigset_t *,sigset_t *))SYS(sigprocmask);
+	
+	ins = NULL; outs = NULL;
+	if (in) {
+		u32 tmp[2];
+		
+		if (copy_from_user (tmp, (sol_sigset_t *)A(in), 2*sizeof(u32)))
+			return -EFAULT;
+		ins = &in_s;
+		if (mapin (tmp, ins)) return -EINVAL;
+	}
+	if (out) outs = &out_s;
+	set_fs (KERNEL_DS);
+	ret = sys_sigprocmask((how == 3) ? SIG_SETMASK : how, ins, outs);
+	set_fs (old_fs);
+	if (ret) return ret;
+	if (out) {
+		u32 tmp[4];
+		
+		tmp[2] = 0; tmp[3] = 0;
+		if (mapout (outs, tmp)) return -EINVAL;
+		if (copy_to_user((sol_sigset_t *)A(out), tmp, 4*sizeof(u32)))
+			return -EFAULT;
+	}
+	return 0;
+}
+
+asmlinkage long do_sol_sigsuspend(u32 mask)
+{
+	sigset_t s;
+	u32 tmp[2];
+		
+	if (copy_from_user (tmp, (sol_sigset_t *)A(mask), 2*sizeof(u32)))
+		return -EFAULT;
+	if (mapin (tmp, &s)) return -EINVAL;
+	return (long)s;
+}
+
+struct sol_sigaction {
+	int	sa_flags;
+	u32	sa_handler;
+	u32	sa_mask[4];
+	int	sa_resv[2];
+};
+
+asmlinkage int solaris_sigaction(int sig, u32 act, u32 old)
+{
+	u32 tmp, tmp2[4];
+	struct sigaction s, s2;
+	int ret;
+	unsigned long old_fs = get_fs();
+	int (*sys_sigaction)(int,struct sigaction *,struct sigaction *) = 
+		(int (*)(int,struct sigaction *,struct sigaction *))SYS(sigaction);
+	
+	sig = mapsig(sig); 
+	if (sig < 0) {
+		/* We cheat a little bit for Solaris only signals */
+		if (old && clear_user((struct sol_sigaction *)A(old), sizeof(struct sol_sigaction)))
+			return -EFAULT;
+		return 0;
+	}
+	if (act) {
+		if (get_user (tmp, &((struct sol_sigaction *)A(act))->sa_flags))
+			return -EFAULT;
+		s.sa_flags = 0;
+		if (tmp & SOLARIS_SA_ONSTACK) s.sa_flags |= SA_STACK;
+		if (tmp & SOLARIS_SA_RESTART) s.sa_flags |= SA_RESTART;
+		if (tmp & SOLARIS_SA_NODEFER) s.sa_flags |= SA_NOMASK;
+		if (tmp & SOLARIS_SA_RESETHAND) s.sa_flags |= SA_ONESHOT;
+		if (tmp & SOLARIS_SA_NOCLDSTOP) s.sa_flags |= SA_NOCLDSTOP;
+		if (get_user (tmp, &((struct sol_sigaction *)A(act))->sa_handler) ||
+		    copy_from_user (tmp2, &((struct sol_sigaction *)A(act))->sa_mask, 2*sizeof(u32)))
+			return -EFAULT;
+		s.sa_handler = (__sighandler_t)A(tmp);
+		if (mapin (tmp2, &s.sa_mask)) return -EINVAL;
+		s.sa_restorer = 0;
+	}
+	set_fs(KERNEL_DS);
+	ret = sys_sigaction(sig, act ? &s : NULL, old ? &s2 : NULL);
+	set_fs(old_fs);
+	if (ret) return ret;
+	if (old) {
+		if (mapout (&s2.sa_mask, tmp2)) return -EINVAL;
+		tmp = 0; tmp2[2] = 0; tmp2[3] = 0;
+		if (s2.sa_flags & SA_STACK) tmp |= SOLARIS_SA_ONSTACK;
+		if (s2.sa_flags & SA_RESTART) tmp |= SOLARIS_SA_RESTART;
+		if (s2.sa_flags & SA_NOMASK) tmp |= SOLARIS_SA_NODEFER;
+		if (s2.sa_flags & SA_ONESHOT) tmp |= SOLARIS_SA_RESETHAND;
+		if (s2.sa_flags & SA_NOCLDSTOP) tmp |= SOLARIS_SA_NOCLDSTOP;
+		if (put_user (tmp, &((struct sol_sigaction *)A(old))->sa_flags) ||
+		    __put_user ((u32)(long)s2.sa_handler, &((struct sol_sigaction *)A(old))->sa_handler) ||
+		    copy_to_user (&((struct sol_sigaction *)A(old))->sa_mask, tmp2, 4*sizeof(u32)))
+			return -EFAULT;
+	}
+	return 0;
+}
+
+asmlinkage int solaris_sigpending(int which, u32 set)
+{
+	sigset_t s;
+	u32 tmp[4];
+	switch (which) {
+	case 1: /* sigpending */
+		lock_kernel();
+		s = current->blocked & current->signal;
+		unlock_kernel();
+		break;
+	case 2: /* sigfillset - I just set signals which have linux equivalents */
+		s = 0x7fffffff;
+		break;
+	default: return -EINVAL;
+	}
+	if (mapout (&s, tmp)) return -EINVAL;
+	tmp[2] = 0; tmp[3] = 0;
+	if (copy_to_user ((u32 *)A(set), tmp, sizeof(tmp)))
+		return -EFAULT;
+	return 0;
+}
+
+asmlinkage int solaris_wait(u32 stat_loc)
+{
+	int (*sys_wait4)(pid_t,unsigned int *, int, struct rusage *) =
+		(int (*)(pid_t,unsigned int *, int, struct rusage *))SYS(wait4);
+	int ret, status;
+	
+	ret = sys_wait4(-1, (unsigned int *)A(stat_loc), WUNTRACED, NULL);
+	if (ret >= 0 && stat_loc) {
+		if (get_user (status, (unsigned int *)A(stat_loc)))
+			return -EFAULT;
+		if (((status - 1) & 0xffff) < 0xff)
+			status = linux_to_solaris_signals[status & 0x7f] & 0x7f;
+		else if ((status & 0xff) == 0x7f)
+			status = (linux_to_solaris_signals[(status >> 8) & 0xff] << 8) | 0x7f;
+		if (__put_user (status, (unsigned int *)A(stat_loc)))
+			return -EFAULT;
+	}
+	return ret;
+}
+
+asmlinkage int solaris_waitid(int idtype, s32 pid, u32 info, int options)
+{
+	int (*sys_wait4)(pid_t,unsigned int *, int, struct rusage *) =
+		(int (*)(pid_t,unsigned int *, int, struct rusage *))SYS(wait4);
+	int opts, status, ret;
+	
+	switch (idtype) {
+	case 0: /* P_PID */ break;
+	case 1: /* P_PGID */ pid = -pid; break;
+	case 7: /* P_ALL */ pid = -1; break;
+	default: return -EINVAL;
+	}
+	opts = 0;
+	if (options & SOLARIS_WUNTRACED) opts |= WUNTRACED;
+	if (options & SOLARIS_WNOHANG) opts |= WNOHANG;
+	current->state = TASK_RUNNING;
+	ret = sys_wait4(pid, (unsigned int *)A(info), opts, NULL);
+	if (ret < 0) return ret;
+	if (info) {
+		struct sol_siginfo *s = (struct sol_siginfo *)A(info);
+	
+		if (get_user (status, (unsigned int *)A(info))) return -EFAULT;
+		__put_user_ret (SOLARIS_SIGCLD, &s->si_signo, -EFAULT);
+		__put_user_ret (ret, &s->_data._proc._pid, -EFAULT);
+		switch (status & 0xff) {
+		case 0: ret = SOLARIS_CLD_EXITED;
+			status = (status >> 8) & 0xff;
+			break;
+		case 0x7f:
+			status = (status >> 8) & 0xff;
+			switch (status) {
+			case SIGSTOP:
+			case SIGTSTP: ret = SOLARIS_CLD_STOPPED;
+			default: ret = SOLARIS_CLD_EXITED;
+			}
+			status = linux_to_solaris_signals[status];
+			break;
+		default:
+			if (status & 0x80) ret = SOLARIS_CLD_DUMPED;
+			else ret = SOLARIS_CLD_KILLED;
+			status = linux_to_solaris_signals[status & 0x7f];
+			break;
+		}
+		__put_user_ret (ret, &s->si_code, -EFAULT);
+		__put_user_ret (status, &s->_data._proc._pdata._cld._status, -EFAULT);
+	}
+	return 0;
+}
+
+extern int svr4_setcontext(svr4_ucontext_t *c, struct pt_regs *regs);
+extern int svr4_getcontext(svr4_ucontext_t *c, struct pt_regs *regs);
+
+asmlinkage int solaris_context(struct pt_regs *regs)
+{
+	switch ((unsigned)regs->u_regs[UREG_I0]) {
+	case 0: /* getcontext */
+		return svr4_getcontext((svr4_ucontext_t *)(long)(u32)regs->u_regs[UREG_I1], regs);
+	case 1: /* setcontext */
+		return svr4_setcontext((svr4_ucontext_t *)(long)(u32)regs->u_regs[UREG_I1], regs);
+	default:
+		return -EINVAL;
+
+	}
+}
+
+asmlinkage int solaris_sigaltstack(u32 ss, u32 oss)
+{
+/* XXX Implement this soon */
+	return 0;
+}
diff --git a/arch/sparc64/solaris/signal.h b/arch/sparc64/solaris/signal.h
new file mode 100644
index 000000000..2e948c788
--- /dev/null
+++ b/arch/sparc64/solaris/signal.h
@@ -0,0 +1,109 @@
+/* $Id: signal.h,v 1.2 1997/09/03 12:29:21 jj Exp $
+ * signal.h: Signal emulation for Solaris
+ *
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ */
+    
+#define SOLARIS_SIGHUP		1
+#define SOLARIS_SIGINT		2
+#define SOLARIS_SIGQUIT		3
+#define SOLARIS_SIGILL		4
+#define SOLARIS_SIGTRAP		5
+#define SOLARIS_SIGIOT		6
+#define SOLARIS_SIGEMT		7
+#define SOLARIS_SIGFPE		8
+#define SOLARIS_SIGKILL		9
+#define SOLARIS_SIGBUS		10
+#define SOLARIS_SIGSEGV		11
+#define SOLARIS_SIGSYS		12
+#define SOLARIS_SIGPIPE		13
+#define SOLARIS_SIGALRM		14
+#define SOLARIS_SIGTERM		15
+#define SOLARIS_SIGUSR1		16
+#define SOLARIS_SIGUSR2		17
+#define SOLARIS_SIGCLD		18
+#define SOLARIS_SIGPWR		19
+#define SOLARIS_SIGWINCH	20
+#define SOLARIS_SIGURG		21
+#define SOLARIS_SIGPOLL		22
+#define SOLARIS_SIGSTOP		23
+#define SOLARIS_SIGTSTP		24
+#define SOLARIS_SIGCONT		25
+#define SOLARIS_SIGTTIN		26
+#define SOLARIS_SIGTTOU		27
+#define SOLARIS_SIGVTALRM	28
+#define SOLARIS_SIGPROF		29
+#define SOLARIS_SIGXCPU		30
+#define SOLARIS_SIGXFSZ		31
+#define SOLARIS_SIGWAITING	32
+#define SOLARIS_SIGLWP		33
+#define SOLARIS_SIGFREEZE	34
+#define SOLARIS_SIGTHAW		35
+#define SOLARIS_SIGCANCEL	36
+#define SOLARIS_SIGRTMIN	37
+#define SOLARIS_SIGRTMAX	44
+#define SOLARIS_NSIGNALS	44
+
+
+#define SOLARIS_SA_ONSTACK	1
+#define SOLARIS_SA_RESETHAND	2
+#define SOLARIS_SA_RESTART	4
+#define SOLARIS_SA_SIGINFO	8
+#define SOLARIS_SA_NODEFER	16
+#define SOLARIS_SA_NOCLDWAIT	0x10000
+#define SOLARIS_SA_NOCLDSTOP	0x20000
+
+struct sol_siginfo {
+	int	si_signo;
+	int	si_code;
+	int	si_errno;
+	union	{
+		char	pad[128-3*sizeof(int)];
+		struct { 
+			s32	_pid;
+			union {
+				struct {
+					s32	_uid;
+					s32	_value;
+				} _kill;
+				struct {
+					s32	_utime;
+					int	_status;
+					s32	_stime;
+				} _cld;
+			} _pdata;
+		} _proc;
+		struct { /* SIGSEGV, SIGBUS, SIGILL and SIGFPE */
+			u32	_addr;
+			int	_trapno;
+		} _fault;
+		struct { /* SIGPOLL, SIGXFSZ */
+			int	_fd;
+			s32	_band;
+		} _file;
+	} _data;
+};
+
+#define SOLARIS_WUNTRACED	0x04
+#define SOLARIS_WNOHANG		0x40
+#define SOLARIS_WEXITED         0x01
+#define SOLARIS_WTRAPPED        0x02
+#define SOLARIS_WSTOPPED        WUNTRACED
+#define SOLARIS_WCONTINUED      0x08
+#define SOLARIS_WNOWAIT         0x80
+
+#define SOLARIS_TRAP_BRKPT      1
+#define SOLARIS_TRAP_TRACE      2
+#define SOLARIS_CLD_EXITED      1
+#define SOLARIS_CLD_KILLED      2
+#define SOLARIS_CLD_DUMPED      3
+#define SOLARIS_CLD_TRAPPED     4
+#define SOLARIS_CLD_STOPPED     5
+#define SOLARIS_CLD_CONTINUED   6
+#define SOLARIS_POLL_IN         1
+#define SOLARIS_POLL_OUT        2
+#define SOLARIS_POLL_MSG        3
+#define SOLARIS_POLL_ERR        4
+#define SOLARIS_POLL_PRI        5
+#define SOLARIS_POLL_HUP        6
+    
diff --git a/arch/sparc64/solaris/socksys.c b/arch/sparc64/solaris/socksys.c
new file mode 100644
index 000000000..984b1bdb5
--- /dev/null
+++ b/arch/sparc64/solaris/socksys.c
@@ -0,0 +1,127 @@
+/* $Id: socksys.c,v 1.1 1997/09/03 12:29:27 jj Exp $
+ * socksys.c: /dev/inet/ stuff for Solaris emulation.
+ *
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ * Copyright (C) 1995, 1996 Mike Jagdis (jaggy@purplet.demon.co.uk)
+ */
+
+#include <linux/types.h>
+#include <linux/kernel.h>
+#include <linux/sched.h>
+#include <linux/smp.h>
+#include <linux/smp_lock.h>
+#include <linux/ioctl.h>
+#include <linux/fs.h>
+#include <linux/init.h>
+#include <linux/poll.h>
+
+#include <asm/uaccess.h>
+#include <asm/termios.h>
+
+#include "conv.h"
+
+extern asmlinkage int sys_ioctl(unsigned int fd, unsigned int cmd, 
+	unsigned long arg);
+	
+static int af_inet_protocols[] = {
+IPPROTO_ICMP, IPPROTO_ICMP, IPPROTO_IGMP, IPPROTO_IPIP, IPPROTO_TCP,
+IPPROTO_EGP, IPPROTO_PUP, IPPROTO_UDP, IPPROTO_IDP, IPPROTO_RAW,
+0, 0, 0, 0, 0, 0,
+};
+
+static struct file_operations socksys_file_ops = {
+	NULL,		/* lseek */
+	NULL,		/* read */
+	NULL,		/* write */
+	NULL,		/* readdir */
+	NULL,		/* poll */
+	NULL,		/* ioctl */
+	NULL,		/* mmap */
+	NULL,		/* open */
+	NULL,		/* release */
+};
+
+static int socksys_open(struct inode * inode, struct file * filp)
+{
+	int family, type, protocol, fd;
+	int (*sys_socket)(int,int,int) =
+		(int (*)(int,int,int))SUNOS(97);
+	
+	family = ((MINOR(inode->i_rdev) >> 4) & 0xf);
+	switch (family) {
+	case AF_UNIX:
+		type = SOCK_STREAM;
+		protocol = 0;
+		break;
+	case AF_INET:
+		protocol = af_inet_protocols[MINOR(inode->i_rdev) & 0xf];
+		switch (protocol) {
+		case IPPROTO_TCP: type = SOCK_STREAM; break;
+		case IPPROTO_UDP: type = SOCK_DGRAM; break;
+		default: type = SOCK_RAW; break;
+		}
+		break;
+	default:
+		type = SOCK_RAW;
+		protocol = 0;
+		break;
+	}
+	fd = sys_socket(family, type, protocol);
+	if (fd < 0) return fd;
+	return 0;
+}
+
+static int socksys_release(struct inode * inode, struct file * filp)
+{
+	return 0;
+}
+
+static unsigned int socksys_poll(struct file * filp, poll_table * wait)
+{
+	return 0;
+}
+	
+static struct file_operations socksys_fops = {
+	NULL,		/* lseek */
+	NULL,		/* read */
+	NULL,		/* write */
+	NULL,		/* readdir */
+	NULL,		/* poll */
+	NULL,		/* ioctl */
+	NULL,		/* mmap */
+	socksys_open,	/* open */
+	socksys_release,/* release */
+};
+
+__initfunc(int
+init_socksys(void))
+{
+	int ret;
+	int (*sys_socket)(int,int,int) =
+		(int (*)(int,int,int))SUNOS(97);
+	int (*sys_close)(unsigned int) = 
+		(int (*)(unsigned int))SYS(close);
+	
+	ret = register_chrdev (30, "socksys", &socksys_fops);
+	if (ret < 0) {
+		printk ("Couldn't register socksys character device\n");
+		return ret;
+	}
+	ret = sys_socket(AF_INET, SOCK_STREAM, IPPROTO_TCP);
+	if (ret < 0) {
+		printk ("Couldn't create socket\n");
+		return ret;
+	}
+	socksys_file_ops = *current->files->fd[ret]->f_op;
+	sys_close(ret);
+	socksys_file_ops.poll = socksys_poll;
+	socksys_file_ops.release = socksys_release;
+	return 0;
+}
+
+void
+cleanup_socksys(void)
+{
+	if (unregister_chrdev (30, "socksys"))
+		printk ("Couldn't unregister socksys character device\n");
+}
diff --git a/arch/sparc64/solaris/systbl.S b/arch/sparc64/solaris/systbl.S
new file mode 100644
index 000000000..c425fb721
--- /dev/null
+++ b/arch/sparc64/solaris/systbl.S
@@ -0,0 +1,289 @@
+/* $Id: systbl.S,v 1.5 1997/09/04 15:46:24 jj Exp $
+ * systbl.S: System call entry point table for Solaris compatibility.
+ *
+ * Copyright (C) 1997 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
+ * Copyright (C) 1996 Miguel de Icaza (miguel@nuclecu.unam.mx)
+ */
+
+#include <asm/unistd.h>
+
+/* Fall back to sys_call_table32 entry */
+#define CHAIN(name)	__NR_##name
+
+/* Pass pt_regs pointer as first argument */
+#define REGS(name)	name+1
+
+/* Hack till all be implemented */
+#define solaris_getmsg		solaris_unimplemented
+#define solaris_getpmsg		solaris_unimplemented
+#define solaris_hrtsys		solaris_unimplemented
+#define solaris_msgsys		solaris_unimplemented
+#define solaris_putmsg		solaris_unimplemented
+#define solaris_putpmsg		solaris_unimplemented
+#define solaris_semsys		solaris_unimplemented
+
+        .data
+        .align  1024
+	.globl		solaris_sys_table
+solaris_sys_table:
+	.word solaris_unimplemented	/* nosys		0	*/
+	.word CHAIN(exit)		/* exit		d	1	*/
+	.word CHAIN(fork)		/* fork			2	*/
+	.word CHAIN(read)		/* read		dpd	3	*/
+	.word CHAIN(write)		/* write	dpd	4	*/
+	.word solaris_open		/* open		soo	5	*/
+	.word CHAIN(close)		/* close	d	6	*/
+	.word solaris_wait		/* wait		xxx	7	*/
+	.word CHAIN(creat)		/* creat	so	8	*/
+	.word CHAIN(link)		/* link		ss	9	*/
+	.word CHAIN(unlink)		/* unlink	s	10	*/
+	.word solaris_unimplemented	/* exec		sxx	11	*/
+	.word CHAIN(chdir)		/* chdir	s	12	*/
+	.word CHAIN(time)		/* time			13	*/
+	.word solaris_mknod		/* mknod	sox	14	*/
+	.word CHAIN(chmod)		/* chmod	so	15	*/ 
+	.word solaris_chown		/* chown	sdd	16	*/
+	.word solaris_brk		/* brk/break	x	17	*/
+	.word solaris_stat		/* stat		sp	18	*/
+	.word CHAIN(lseek)		/* seek/lseek	ddd	19	*/
+	.word solaris_getpid		/* getpid		20	*/
+	.word solaris_unimplemented	/* mount		21	*/
+	.word CHAIN(umount)		/* umount	s	22	*/
+	.word CHAIN(setuid)		/* setuid	d	23	*/
+	.word solaris_getuid		/* getuid		24	*/
+	.word CHAIN(stime)		/* stime	d	25	*/
+#if 0
+	.word solaris_ptrace		/* ptrace	xdxx	26	*/
+#else
+	.word CHAIN(ptrace)		/* ptrace	xdxx	26	*/
+#endif
+	.word CHAIN(alarm)		/* alarm	d	27	*/
+	.word solaris_fstat		/* fstat	dp	28	*/
+	.word CHAIN(pause)		/* pause		29	*/
+	.word CHAIN(utime)		/* utime	xx	30	*/
+	.word solaris_unimplemented	/* stty			31	*/
+	.word solaris_unimplemented	/* gtty			32	*/
+	.word solaris_access		/* access	so	33	*/
+	.word CHAIN(nice)		/* nice		d	34	*/
+	.word solaris_statfs		/* statfs	spdd	35	*/
+	.word CHAIN(sync)		/* sync			36	*/
+	.word solaris_kill		/* kill		dd	37	*/
+	.word solaris_fstatfs		/* fstatfs	dpdd	38	*/
+	.word solaris_procids		/* pgrpsys	ddd	39	*/
+	.word solaris_unimplemented	/* xenix		40	*/
+	.word CHAIN(dup)		/* dup		d	41	*/
+	.word CHAIN(pipe)		/* pipe			42	*/
+	.word CHAIN(times)		/* times	p	43	*/
+	.word CHAIN(profil)		/* prof		xxxx	44	*/
+	.word solaris_unimplemented	/* lock/plock		45	*/
+	.word CHAIN(setgid)		/* setgid	d	46	*/
+	.word solaris_getgid		/* getgid		47	*/
+	.word solaris_sigfunc		/* sigfunc	xx	48	*/
+	.word REGS(solaris_msgsys)	/* msgsys	dxddd	49	*/
+	.word solaris_unimplemented	/* syssun/3b		50	*/
+	.word CHAIN(acct)		/* acct/sysacct	x	51	*/
+	.word solaris_shmsys		/* shmsys	ddxo	52	*/
+	.word REGS(solaris_semsys)	/* semsys	dddx	53	*/
+	.word solaris_ioctl		/* ioctl	dxx	54	*/
+	.word solaris_unimplemented	/* uadmin	xxx	55	*/
+	.word solaris_unimplemented	/* reserved:exch	56	*/
+	.word solaris_utssys		/* utssys	x	57	*/
+	.word CHAIN(fsync)		/* fsync	d	58	*/
+	.word CHAIN(execve)		/* execv	spp	59	*/
+	.word CHAIN(umask)		/* umask	o	60	*/
+	.word CHAIN(chroot)		/* chroot	s	61	*/
+	.word solaris_fcntl		/* fcntl	dxx	62	*/
+	.word solaris_ulimit		/* ulimit	xx	63	*/
+	.word solaris_unimplemented	/* ?			64	*/
+	.word solaris_unimplemented	/* ?			65	*/
+	.word solaris_unimplemented	/* ?			66	*/
+	.word solaris_unimplemented	/* ?			67	*/
+	.word solaris_unimplemented	/* ?			68	*/
+	.word solaris_unimplemented	/* ?			69	*/
+	.word solaris_unimplemented	/* advfs		70	*/
+	.word solaris_unimplemented	/* unadvfs		71	*/
+	.word solaris_unimplemented	/* rmount		72	*/
+	.word solaris_unimplemented	/* rumount		73	*/
+	.word solaris_unimplemented	/* rfstart		74	*/
+	.word solaris_unimplemented	/* ?			75	*/
+	.word solaris_unimplemented	/* rdebug		76	*/
+	.word solaris_unimplemented	/* rfstop		77	*/
+	.word solaris_unimplemented	/* rfsys		78	*/
+	.word CHAIN(rmdir)		/* rmdir	s	79	*/
+	.word CHAIN(mkdir)		/* mkdir	so	80	*/
+	.word CHAIN(getdents)		/* getdents	dxd	81	*/
+	.word solaris_unimplemented	/* libattach		82	*/
+	.word solaris_unimplemented	/* libdetach		83	*/
+	.word CHAIN(sysfs)		/* sysfs	dxx	84	*/
+	.word REGS(solaris_getmsg)	/* getmsg	dxxx	85	*/
+	.word REGS(solaris_putmsg)	/* putmsg	dxxd	86	*/
+	.word CHAIN(poll)		/* poll		xdd	87	*/
+	.word solaris_lstat		/* lstat	sp	88	*/
+	.word CHAIN(symlink)		/* symlink	ss	89	*/
+	.word CHAIN(readlink)		/* readlink	spd	90	*/
+	.word CHAIN(setgroups)		/* setgroups	dp	91	*/
+	.word CHAIN(getgroups)		/* getgroups	dp	92	*/
+	.word CHAIN(fchmod)		/* fchmod	do	93	*/
+	.word CHAIN(fchown)		/* fchown	ddd	94	*/
+	.word solaris_sigprocmask	/* sigprocmask	dxx	95	*/
+	.word solaris_sigsuspend	/* sigsuspend	x	96	*/
+	.word solaris_sigaltstack	/* sigaltstack	xx	97	*/
+	.word solaris_sigaction		/* sigaction	dxx	98	*/
+	.word solaris_sigpending	/* sigpending	dd	99	*/
+	.word REGS(solaris_context)	/* context		100	*/
+	.word solaris_unimplemented	/* evsys		101	*/
+	.word solaris_unimplemented	/* evtrapret		102	*/
+	.word solaris_statvfs		/* statvfs	sp	103	*/
+	.word solaris_fstatvfs		/* fstatvfs	dp	104	*/
+	.word solaris_unimplemented	/* unknown		105	*/
+	.word solaris_unimplemented	/* nfssys		106	*/
+	.word solaris_waitid		/* waitid	ddxd	107	*/
+	.word solaris_unimplemented	/* sigsendsys	ddd	108	*/
+	.word REGS(solaris_hrtsys)	/* hrtsys	xxx	109	*/
+	.word solaris_unimplemented	/* acancel	dxd	110	*/
+	.word solaris_unimplemented	/* async		111	*/
+	.word solaris_unimplemented	/* priocntlsys		112	*/
+	.word solaris_pathconf		/* pathconf	sd	113	*/
+	.word solaris_unimplemented	/* mincore	xdx	114	*/
+	.word solaris_mmap		/* mmap		xxxxdx	115	*/
+	.word CHAIN(mprotect)		/* mprotect	xdx	116	*/
+	.word CHAIN(munmap)		/* munmap	xd	117	*/
+	.word solaris_fpathconf		/* fpathconf	dd	118	*/
+	.word CHAIN(fork)		/* fork			119	*/
+	.word solaris_unimplemented	/* fchdir	d	120	*/
+	.word CHAIN(readv)		/* readv	dxd	121	*/
+	.word CHAIN(writev)		/* writev	dxd	122	*/
+	.word solaris_xstat		/* xstat	dsx	123	*/
+	.word solaris_lxstat		/* lxstat	dsx	124	*/
+	.word solaris_fxstat		/* fxstat	ddx	125	*/
+	.word solaris_xmknod		/* xmknod	dsox	126	*/
+	.word solaris_unimplemented	/* syslocal	d	127	*/
+	.word solaris_unimplemented	/* setrlimit		128	*/
+	.word solaris_unimplemented	/* getrlimit		129	*/
+	.word CHAIN(chown)		/* lchown	sdd	130	*/
+	.word solaris_unimplemented	/* memcntl		131	*/
+	.word solaris_getpmsg		/* getpmsg	dxxxx	132	*/
+	.word solaris_putpmsg		/* putpmsg	dxxdd	133	*/
+	.word CHAIN(rename)		/* rename	ss	134	*/
+	.word solaris_utsname		/* uname	x	135	*/
+	.word solaris_unimplemented	/* setegid		136	*/
+	.word solaris_sysconf		/* sysconfig	d	137	*/
+	.word solaris_unimplemented	/* adjtime		138	*/
+	.word solaris_sysinfo		/* systeminfo	dsd	139	*/
+	.word solaris_unimplemented	/* ?			140	*/
+	.word solaris_unimplemented	/* seteuid		141	*/
+	.word solaris_unimplemented	/* ?			142	*/
+	.word solaris_unimplemented	/* ?			143	*/
+	.word solaris_unimplemented	/* secsys	dx	144	*/
+	.word solaris_unimplemented	/* filepriv	sdxd	145	*/
+	.word solaris_unimplemented	/* procpriv	dxd	146	*/
+	.word solaris_unimplemented	/* devstat	sdx	147	*/
+	.word solaris_unimplemented	/* aclipc	ddddx	148	*/
+	.word solaris_unimplemented	/* fdevstat	ddx	149	*/
+	.word solaris_unimplemented	/* flvlfile	ddx	150	*/
+	.word solaris_unimplemented	/* lvlfile	sdx	151	*/
+	.word solaris_unimplemented	/* ?			152	*/
+	.word solaris_unimplemented	/* fchroot	d	153	*/
+	.word solaris_unimplemented	/* lvlproc	dx	154	*/
+	.word solaris_unimplemented	/* ?			155	*/
+	.word CHAIN(gettimeofday)	/* gettimeofday	xx	156	*/
+	.word CHAIN(getitimer)		/* getitimer	dx	157	*/
+	.word CHAIN(setitimer)		/* setitimer	dxx	158	*/
+	.word solaris_unimplemented	/* lwp-xxx		159	*/
+	.word solaris_unimplemented	/* lwp-xxx		160	*/
+	.word solaris_unimplemented	/* lwp-xxx		161	*/
+	.word solaris_unimplemented	/* lwp-xxx		162	*/
+	.word solaris_unimplemented	/* lwp-xxx		163	*/
+	.word solaris_unimplemented	/* lwp-xxx		164	*/
+	.word solaris_unimplemented	/* lwp-xxx		165	*/
+	.word solaris_unimplemented	/* lwp-xxx		166	*/
+	.word solaris_unimplemented	/* lwp-xxx		167	*/
+	.word solaris_unimplemented	/* lwp-xxx		168	*/
+	.word solaris_unimplemented	/* lwp-xxx		169	*/
+	.word solaris_unimplemented	/* lwp-xxx		170	*/
+	.word solaris_unimplemented	/* lwp-xxx		171	*/
+	.word solaris_unimplemented	/* lwp-xxx		172	*/
+	.word solaris_pread		/* pread	dpdd	173	*/
+	.word solaris_pwrite		/* pwrite	dpdd	174	*/
+	.word REGS(solaris_llseek)	/* llseek	dLd	175	*/
+	.word solaris_unimplemented	/* lwpself		176	*/
+	.word solaris_unimplemented	/* lwpinfo		177	*/
+	.word solaris_unimplemented	/* lwpprivate		178	*/
+	.word solaris_unimplemented	/* processorbind	179	*/
+	.word solaris_unimplemented	/* processorexbind	180	*/
+	.word solaris_unimplemented	/* 			181	*/
+	.word solaris_unimplemented	/* sync_mailbox		182	*/
+	.word solaris_unimplemented	/* prepblock		183	*/
+	.word solaris_unimplemented	/* block		184	*/
+	.word solaris_acl		/* acl		sddp	185	*/
+	.word solaris_unimplemented	/* unblock		186	*/
+	.word solaris_unimplemented	/* cancelblock		187	*/
+	.word solaris_unimplemented	/* ?			188	*/
+	.word solaris_unimplemented	/* xxxxx		189	*/
+	.word solaris_unimplemented	/* xxxxxe		190	*/
+	.word solaris_unimplemented	/*			191	*/
+	.word solaris_unimplemented	/*			192	*/
+	.word solaris_unimplemented	/*			193	*/
+	.word solaris_unimplemented	/*			194	*/
+	.word solaris_unimplemented	/* 			195	*/
+	.word solaris_unimplemented	/* 			196	*/
+	.word solaris_unimplemented	/* 			197	*/
+	.word solaris_unimplemented	/* 			198	*/
+	.word CHAIN(nanosleep)		/* nanosleep	dd	199	*/
+	.word solaris_facl		/* facl		dddp	200	*/
+	.word solaris_unimplemented	/* 			201	*/
+	.word solaris_unimplemented	/* 			202	*/
+	.word solaris_unimplemented	/* 			203	*/
+	.word solaris_unimplemented	/* 			204	*/
+	.word solaris_unimplemented	/* 			205	*/
+	.word solaris_unimplemented	/* 			206	*/
+	.word solaris_unimplemented	/* 			207	*/
+	.word solaris_unimplemented	/* 			208	*/
+	.word solaris_unimplemented	/* 			209	*/
+	.word solaris_unimplemented	/* 			210	*/
+	.word solaris_unimplemented	/* 			211	*/
+	.word solaris_unimplemented	/* 			212	*/
+	.word solaris_unimplemented	/* 			213	*/
+	.word solaris_unimplemented	/* 			214	*/
+	.word solaris_unimplemented	/* 			215	*/
+	.word solaris_unimplemented	/* 			216	*/
+	.word solaris_unimplemented	/* 			217	*/
+	.word solaris_unimplemented	/* 			218	*/
+	.word solaris_unimplemented	/* 			219	*/
+	.word solaris_unimplemented	/* 			220	*/
+	.word solaris_unimplemented	/* 			221	*/
+	.word solaris_unimplemented	/* 			222	*/
+	.word solaris_unimplemented	/* 			223	*/
+	.word solaris_unimplemented	/* 			224	*/
+	.word solaris_unimplemented	/* 			225	*/
+	.word solaris_unimplemented	/* 			226	*/
+	.word solaris_unimplemented	/* 			227	*/
+	.word solaris_unimplemented	/* 			228	*/
+	.word solaris_unimplemented	/* 			229	*/
+	.word solaris_unimplemented	/* 			230	*/
+	.word solaris_unimplemented	/* 			231	*/
+	.word solaris_unimplemented	/* 			232	*/
+	.word solaris_unimplemented	/* 			233	*/
+	.word solaris_unimplemented	/* 			234	*/
+	.word solaris_unimplemented	/* 			235	*/
+	.word solaris_unimplemented	/* 			236	*/
+	.word solaris_unimplemented	/* 			237	*/
+	.word solaris_unimplemented	/* 			238	*/
+	.word solaris_unimplemented	/* 			239	*/
+	.word solaris_unimplemented	/* 			240	*/
+	.word solaris_unimplemented	/* 			241	*/
+	.word solaris_unimplemented	/* 			242	*/
+	.word solaris_unimplemented	/* 			243	*/
+	.word solaris_unimplemented	/* 			244	*/
+	.word solaris_unimplemented	/* 			245	*/
+	.word solaris_unimplemented	/* 			246	*/
+	.word solaris_unimplemented	/* 			247	*/
+	.word solaris_unimplemented	/* 			248	*/
+	.word solaris_unimplemented	/* 			249	*/
+	.word solaris_unimplemented	/* 			250	*/
+	.word solaris_unimplemented	/* 			251	*/
+	.word solaris_unimplemented	/* 			252	*/
+	.word solaris_unimplemented	/* 			253	*/
+	.word solaris_unimplemented	/* 			254	*/
+	.word solaris_unimplemented	/* 			255	*/
+
diff --git a/arch/sparc64/vmlinux.lds b/arch/sparc64/vmlinux.lds
index eac8314ca..661acc098 100644
--- a/arch/sparc64/vmlinux.lds
+++ b/arch/sparc64/vmlinux.lds
@@ -52,12 +52,14 @@ SECTIONS
    . += 8192;
    empty_bad_pte_table = .;
    . += 8192;
+   empty_bad_page = .;
+   . += 8192;
+   . += 0x40;
    empty_null_pmd_table = .;
    . += 8192;
+   . += 0x40;
    empty_null_pte_table = .;
    . += 8192;
-   empty_bad_page = .;
-   . += 8192;
   }
   _end = . ;
   PROVIDE (end = .);