Merge with Linux 2.4.2.

19 files changed, 547 insertions, 24 deletions

diff --git a/arch/alpha/kernel/Makefile b/arch/alpha/kernel/Makefile
index abc04cca2..08ec1d613 100644
--- a/arch/alpha/kernel/Makefile
+++ b/arch/alpha/kernel/Makefile
@@ -30,9 +30,7 @@ ifdef CONFIG_VGA_HOSE
 obj-y	 += console.o
 endif
 
-
 obj-$(CONFIG_SMP)    += smp.o irq_smp.o
-
 obj-$(CONFIG_PCI)    += pci.o pci_iommu.o
 
 ifdef CONFIG_ALPHA_GENERIC
@@ -76,7 +74,7 @@ obj-y    += sys_eb64p.o
 endif
 
 obj-$(CONFIG_ALPHA_EIGER) += sys_eiger.o
-obj-$(CONFIG_ALPHA_JENSEN) += sys_jensen.o
+obj-$(CONFIG_ALPHA_JENSEN) += sys_jensen.o pci-noop.o
 obj-$(CONFIG_ALPHA_MIATA) += sys_miata.o
 obj-$(CONFIG_ALPHA_MIKASA) += sys_mikasa.o
 obj-$(CONFIG_ALPHA_NAUTILUS) += sys_nautilus.o
diff --git a/arch/alpha/kernel/alpha_ksyms.c b/arch/alpha/kernel/alpha_ksyms.c
index 17285ac26..d7bf13ec3 100644
--- a/arch/alpha/kernel/alpha_ksyms.c
+++ b/arch/alpha/kernel/alpha_ksyms.c
@@ -98,9 +98,13 @@ EXPORT_SYMBOL(__memcpy);
 EXPORT_SYMBOL(__memset);
 EXPORT_SYMBOL(__memsetw);
 EXPORT_SYMBOL(__constant_c_memset);
+EXPORT_SYMBOL(copy_page);
+EXPORT_SYMBOL(clear_page);
 
 EXPORT_SYMBOL(__direct_map_base);
 EXPORT_SYMBOL(__direct_map_size);
+
+#ifdef CONFIG_PCI
 EXPORT_SYMBOL(pci_alloc_consistent);
 EXPORT_SYMBOL(pci_free_consistent);
 EXPORT_SYMBOL(pci_map_single);
@@ -108,6 +112,7 @@ EXPORT_SYMBOL(pci_unmap_single);
 EXPORT_SYMBOL(pci_map_sg);
 EXPORT_SYMBOL(pci_unmap_sg);
 EXPORT_SYMBOL(pci_dma_supported);
+#endif
 
 EXPORT_SYMBOL(dump_thread);
 EXPORT_SYMBOL(dump_fpu);
@@ -166,6 +171,7 @@ EXPORT_SYMBOL(__down_failed_interruptible);
 EXPORT_SYMBOL(__up_wakeup);
 EXPORT_SYMBOL(down);
 EXPORT_SYMBOL(down_interruptible);
+EXPORT_SYMBOL(down_trylock);
 EXPORT_SYMBOL(up);
 EXPORT_SYMBOL(__down_read_failed);
 EXPORT_SYMBOL(__down_write_failed);
diff --git a/arch/alpha/kernel/irq.c b/arch/alpha/kernel/irq.c
index 080e48e43..825eaf2d5 100644
--- a/arch/alpha/kernel/irq.c
+++ b/arch/alpha/kernel/irq.c
@@ -18,7 +18,7 @@
 #include <linux/signal.h>
 #include <linux/sched.h>
 #include <linux/interrupt.h>
-#include <linux/malloc.h>
+#include <linux/slab.h>
 #include <linux/random.h>
 #include <linux/init.h>
 #include <linux/irq.h>
diff --git a/arch/alpha/kernel/irq_alpha.c b/arch/alpha/kernel/irq_alpha.c
index 59f102496..91e99f573 100644
--- a/arch/alpha/kernel/irq_alpha.c
+++ b/arch/alpha/kernel/irq_alpha.c
@@ -18,7 +18,7 @@
 unsigned long __irq_attempt[NR_IRQS];
 #endif
 
-/* Hack minimum IPL during interupt processing for broken hardware.  */
+/* Hack minimum IPL during interrupt processing for broken hardware.  */
 #ifdef CONFIG_ALPHA_BROKEN_IRQ_MASK
 int __min_ipl;
 #endif
diff --git a/arch/alpha/kernel/osf_sys.c b/arch/alpha/kernel/osf_sys.c
index cd28b07fc..6159457d9 100644
--- a/arch/alpha/kernel/osf_sys.c
+++ b/arch/alpha/kernel/osf_sys.c
@@ -19,7 +19,7 @@
 #include <linux/stddef.h>
 #include <linux/unistd.h>
 #include <linux/ptrace.h>
-#include <linux/malloc.h>
+#include <linux/slab.h>
 #include <linux/user.h>
 #include <linux/a.out.h>
 #include <linux/utsname.h>
@@ -74,8 +74,10 @@ asmlinkage int osf_set_program_attributes(
 	mm = current->mm;
 	mm->end_code = bss_start + bss_len;
 	mm->brk = bss_start + bss_len;
+#if 0
 	printk("set_program_attributes(%lx %lx %lx %lx)\n",
 		text_start, text_len, bss_start, bss_len);
+#endif
 	unlock_kernel();
 	return 0;
 }
diff --git a/arch/alpha/kernel/pci-noop.c b/arch/alpha/kernel/pci-noop.c
new file mode 100644
index 000000000..d530ebb82
--- /dev/null
+++ b/arch/alpha/kernel/pci-noop.c
@@ -0,0 +1,104 @@
+/*
+ *	linux/arch/alpha/kernel/pci-noop.c
+ *
+ * Stub PCI interfaces for Jensen-specific kernels.
+ */
+
+#include <linux/pci.h>
+#include <linux/init.h>
+#include <linux/bootmem.h>
+#include <linux/errno.h>
+#include <linux/sched.h>
+
+#include "proto.h"
+
+
+/*
+ * The PCI controler list.
+ */
+
+struct pci_controler *hose_head, **hose_tail = &hose_head;
+struct pci_controler *pci_isa_hose;
+
+
+struct pci_controler * __init
+alloc_pci_controler(void)
+{
+	struct pci_controler *hose;
+
+	hose = alloc_bootmem(sizeof(*hose));
+
+	*hose_tail = hose;
+	hose_tail = &hose->next;
+
+	return hose;
+}
+
+struct resource * __init
+alloc_resource(void)
+{
+	struct resource *res;
+
+	res = alloc_bootmem(sizeof(*res));
+
+	return res;
+}
+
+asmlinkage long
+sys_pciconfig_iobase(long which, unsigned long bus, unsigned long dfn)
+{
+	struct pci_controler *hose;
+	struct pci_dev *dev;
+
+	/* from hose or from bus.devfn */
+	if (which & IOBASE_FROM_HOSE) {
+		for (hose = hose_head; hose; hose = hose->next) 
+			if (hose->index == bus)
+				break;
+		if (!hose)
+			return -ENODEV;
+	} else {
+		/* Special hook for ISA access.  */
+		if (bus == 0 && dfn == 0)
+			hose = pci_isa_hose;
+		else
+			return -ENODEV;
+	}
+
+	switch (which & ~IOBASE_FROM_HOSE) {
+	case IOBASE_HOSE:
+		return hose->index;
+	case IOBASE_SPARSE_MEM:
+		return hose->sparse_mem_base;
+	case IOBASE_DENSE_MEM:
+		return hose->dense_mem_base;
+	case IOBASE_SPARSE_IO:
+		return hose->sparse_io_base;
+	case IOBASE_DENSE_IO:
+		return hose->dense_io_base;
+	case IOBASE_ROOT_BUS:
+		return hose->bus->number;
+	}
+
+	return -EOPNOTSUPP;
+}
+
+asmlinkage long
+sys_pciconfig_read(unsigned long bus, unsigned long dfn,
+		   unsigned long off, unsigned long len, void *buf)
+{
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+	else
+		return -ENODEV;
+}
+
+asmlinkage long
+sys_pciconfig_write(unsigned long bus, unsigned long dfn,
+		    unsigned long off, unsigned long len, void *buf)
+{
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+	else
+		return -ENODEV;
+}
diff --git a/arch/alpha/kernel/process.c b/arch/alpha/kernel/process.c
index 508e278b0..2dd505e10 100644
--- a/arch/alpha/kernel/process.c
+++ b/arch/alpha/kernel/process.c
@@ -18,7 +18,7 @@
 #include <linux/stddef.h>
 #include <linux/unistd.h>
 #include <linux/ptrace.h>
-#include <linux/malloc.h>
+#include <linux/slab.h>
 #include <linux/user.h>
 #include <linux/a.out.h>
 #include <linux/utsname.h>
@@ -416,22 +416,20 @@ dump_fpu(struct pt_regs * regs, elf_fpregset_t *r)
  * Don't do this at home.
  */
 asmlinkage int
-sys_execve(unsigned long a0, unsigned long a1, unsigned long a2,
-	unsigned long a3, unsigned long a4, unsigned long a5,
-	struct pt_regs regs)
+sys_execve(char *ufilename, char **argv, char **envp,
+	   unsigned long a3, unsigned long a4, unsigned long a5,
+	   struct pt_regs regs)
 {
 	int error;
-	char * filename;
+	char *filename;
 
-	lock_kernel();
-	filename = getname((char *) a0);
+	filename = getname(ufilename);
 	error = PTR_ERR(filename);
 	if (IS_ERR(filename))
 		goto out;
-	error = do_execve(filename, (char **) a1, (char **) a2, &regs);
+	error = do_execve(filename, argv, envp, &regs);
 	putname(filename);
 out:
-	unlock_kernel();
 	return error;
 }
 
diff --git a/arch/alpha/kernel/ptrace.c b/arch/alpha/kernel/ptrace.c
index 09fcfd787..a919e7c2f 100644
--- a/arch/alpha/kernel/ptrace.c
+++ b/arch/alpha/kernel/ptrace.c
@@ -12,7 +12,7 @@
 #include <linux/errno.h>
 #include <linux/ptrace.h>
 #include <linux/user.h>
-#include <linux/malloc.h>
+#include <linux/slab.h>
 
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
diff --git a/arch/alpha/kernel/setup.c b/arch/alpha/kernel/setup.c
index 2acf56d96..a6443da53 100644
--- a/arch/alpha/kernel/setup.c
+++ b/arch/alpha/kernel/setup.c
@@ -16,7 +16,7 @@
 #include <linux/stddef.h>
 #include <linux/unistd.h>
 #include <linux/ptrace.h>
-#include <linux/malloc.h>
+#include <linux/slab.h>
 #include <linux/user.h>
 #include <linux/a.out.h>
 #include <linux/tty.h>
diff --git a/arch/alpha/kernel/smc37c669.c b/arch/alpha/kernel/smc37c669.c
index 0b2db18ab..3ffb611f2 100644
--- a/arch/alpha/kernel/smc37c669.c
+++ b/arch/alpha/kernel/smc37c669.c
@@ -3,7 +3,7 @@
  */
 #include <linux/kernel.h>
 
-#include <linux/malloc.h>
+#include <linux/slab.h>
 #include <linux/mm.h>
 #include <linux/init.h>
 #include <linux/delay.h>
diff --git a/arch/alpha/kernel/smc37c93x.c b/arch/alpha/kernel/smc37c93x.c
index 5448305a3..b0e15d307 100644
--- a/arch/alpha/kernel/smc37c93x.c
+++ b/arch/alpha/kernel/smc37c93x.c
@@ -5,7 +5,7 @@
 #include <linux/config.h>
 #include <linux/kernel.h>
 
-#include <linux/malloc.h>
+#include <linux/slab.h>
 #include <linux/mm.h>
 #include <linux/init.h>
 #include <linux/delay.h>
diff --git a/arch/alpha/kernel/smp.c b/arch/alpha/kernel/smp.c
index e91e77895..4f877b10b 100644
--- a/arch/alpha/kernel/smp.c
+++ b/arch/alpha/kernel/smp.c
@@ -837,7 +837,7 @@ smp_call_function (void (*func) (void *info), void *info, int retry, int wait)
 	atomic_set(&data.unstarted_count, smp_num_cpus - 1);
 	atomic_set(&data.unfinished_count, smp_num_cpus - 1);
 
-	/* Aquire the smp_call_function_data mutex.  */
+	/* Acquire the smp_call_function_data mutex.  */
 	if (pointer_lock(&smp_call_function_data, &data, retry))
 		return -EBUSY;
 
diff --git a/arch/alpha/kernel/sys_ruffian.c b/arch/alpha/kernel/sys_ruffian.c
index c9a2e79a4..0d9713377 100644
--- a/arch/alpha/kernel/sys_ruffian.c
+++ b/arch/alpha/kernel/sys_ruffian.c
@@ -92,14 +92,80 @@ ruffian_kill_arch (int mode)
 #endif
 }
 
+/*
+ *  Interrupt routing:
+ *
+ *		Primary bus
+ *	  IdSel		INTA	INTB	INTC	INTD
+ * 21052   13		  -	  -	  -	  -
+ * SIO	   14		 23	  -	  -	  -
+ * 21143   15		 44	  -	  -	  -
+ * Slot 0  17		 43	 42	 41	 40
+ *
+ *		Secondary bus
+ *	  IdSel		INTA	INTB	INTC	INTD
+ * Slot 0   8 (18)	 19	 18	 17	 16
+ * Slot 1   9 (19)	 31	 30	 29	 28
+ * Slot 2  10 (20)	 27	 26	 25	 24
+ * Slot 3  11 (21)	 39	 38	 37	 36
+ * Slot 4  12 (22)	 35	 34	 33	 32
+ * 53c875  13 (23)	 20	  -	  -	  -
+ *
+ */
+
 static int __init
 ruffian_map_irq(struct pci_dev *dev, u8 slot, u8 pin)
 {
-	/* We don't know anything about the PCI routing, so leave
-	   the IRQ unchanged.  */
-	return dev->irq;
+        static char irq_tab[11][5] __initdata = {
+	      /*INT  INTA INTB INTC INTD */
+		{-1,  -1,  -1,  -1,  -1},  /* IdSel 13,  21052	     */
+		{-1,  -1,  -1,  -1,  -1},  /* IdSel 14,  SIO	     */
+		{44,  44,  44,  44,  44},  /* IdSel 15,  21143	     */
+		{-1,  -1,  -1,  -1,  -1},  /* IdSel 16,  none	     */
+		{43,  43,  42,  41,  40},  /* IdSel 17,  64-bit slot */
+		/* the next 6 are actually on PCI bus 1, across the bridge */
+		{19,  19,  18,  17,  16},  /* IdSel  8,  slot 0	     */
+		{31,  31,  30,  29,  28},  /* IdSel  9,  slot 1	     */
+		{27,  27,  26,  25,  24},  /* IdSel 10,  slot 2	     */
+		{39,  39,  38,  37,  36},  /* IdSel 11,  slot 3	     */
+		{35,  35,  34,  33,  32},  /* IdSel 12,  slot 4	     */
+		{20,  20,  20,  20,  20},  /* IdSel 13,  53c875	     */
+        };
+	const long min_idsel = 13, max_idsel = 23, irqs_per_slot = 5;
+	return COMMON_TABLE_LOOKUP;
 }
 
+static u8 __init
+ruffian_swizzle(struct pci_dev *dev, u8 *pinp)
+{
+	int slot, pin = *pinp;
+
+	if (dev->bus->number == 0) {
+		slot = PCI_SLOT(dev->devfn);
+	}		
+	/* Check for the built-in bridge.  */
+	else if (PCI_SLOT(dev->bus->self->devfn) == 13) {
+		slot = PCI_SLOT(dev->devfn) + 10;
+	}
+	else 
+	{
+		/* Must be a card-based bridge.  */
+		do {
+			if (PCI_SLOT(dev->bus->self->devfn) == 13) {
+				slot = PCI_SLOT(dev->devfn) + 10;
+				break;
+			}
+			pin = bridge_swizzle(pin, PCI_SLOT(dev->devfn));
+
+			/* Move up the chain of bridges.  */
+			dev = dev->bus->self;
+			/* Slot of the next bridge.  */
+			slot = PCI_SLOT(dev->devfn);
+		} while (dev->bus->self);
+	}
+	*pinp = pin;
+	return slot;
+}
 
 #ifdef BUILDING_FOR_MILO
 /*
@@ -164,6 +230,6 @@ struct alpha_machine_vector ruffian_mv __initmv = {
 	init_pci:		cia_init_pci,
 	kill_arch:		ruffian_kill_arch,
 	pci_map_irq:		ruffian_map_irq,
-	pci_swizzle:		common_swizzle,
+	pci_swizzle:		ruffian_swizzle,
 };
 ALIAS_MV(ruffian)
diff --git a/arch/alpha/kernel/traps.c b/arch/alpha/kernel/traps.c
index 7b2f8be03..ed7d2f68e 100644
--- a/arch/alpha/kernel/traps.c
+++ b/arch/alpha/kernel/traps.c
@@ -1093,7 +1093,9 @@ alpha_ni_syscall(unsigned long a0, unsigned long a1, unsigned long a2,
 {
 	/* We only get here for OSF system calls, minus #112;
 	   the rest go to sys_ni_syscall.  */
+#if 0
 	printk("<sc %ld(%lx,%lx,%lx)>", regs.r0, a0, a1, a2);
+#endif
 	return -ENOSYS;
 }
 
diff --git a/arch/alpha/lib/Makefile b/arch/alpha/lib/Makefile
index 1e3e485b5..d95ace0d7 100644
--- a/arch/alpha/lib/Makefile
+++ b/arch/alpha/lib/Makefile
@@ -42,6 +42,8 @@ OBJS =	__divqu.o __remqu.o __divlu.o __remlu.o \
 	$(ev6)strncpy_from_user.o \
 	$(ev67)strlen_user.o \
 	$(ev6)csum_ipv6_magic.o \
+	$(ev6)clear_page.o \
+	$(ev6)copy_page.o \
 	strcasecmp.o \
 	fpreg.o \
 	callback_srm.o srm_puts.o srm_printk.o
diff --git a/arch/alpha/lib/clear_page.S b/arch/alpha/lib/clear_page.S
new file mode 100644
index 000000000..a221ae266
--- /dev/null
+++ b/arch/alpha/lib/clear_page.S
@@ -0,0 +1,39 @@
+/*
+ * arch/alpha/lib/clear_page.S
+ *
+ * Zero an entire page.
+ */
+
+	.text
+	.align 4
+	.global clear_page
+	.ent clear_page
+clear_page:
+	.prologue 0
+
+	lda	$0,128
+	nop
+	unop
+	nop
+
+1:	stq	$31,0($16)
+	stq	$31,8($16)
+	stq	$31,16($16)
+	stq	$31,24($16)
+
+	stq	$31,32($16)
+	stq	$31,40($16)
+	stq	$31,48($16)
+	subq	$0,1,$0
+
+	stq	$31,56($16)
+	addq	$16,64,$16
+	unop
+	bne	$0,1b
+
+	ret
+	nop
+	unop
+	nop
+
+	.end clear_page
diff --git a/arch/alpha/lib/copy_page.S b/arch/alpha/lib/copy_page.S
new file mode 100644
index 000000000..9f3b97459
--- /dev/null
+++ b/arch/alpha/lib/copy_page.S
@@ -0,0 +1,49 @@
+/*
+ * arch/alpha/lib/copy_page.S
+ *
+ * Copy an entire page.
+ */
+
+	.text
+	.align 4
+	.global copy_page
+	.ent copy_page
+copy_page:
+	.prologue 0
+
+	lda	$18,128
+	nop
+	unop
+	nop
+
+1:	ldq	$0,0($17)
+	ldq	$1,8($17)
+	ldq	$2,16($17)
+	ldq	$3,24($17)
+
+	ldq	$4,32($17)
+	ldq	$5,40($17)
+	ldq	$6,48($17)
+	ldq	$7,56($17)
+
+	stq	$0,0($16)
+	subq	$18,1,$18
+	stq	$1,8($16)
+	addq	$17,64,$17
+
+	stq	$2,16($16)
+	stq	$3,24($16)
+	stq	$4,32($16)
+	stq	$5,40($16)
+
+	stq	$6,48($16)
+	stq	$7,56($16)
+	addq	$16,64,$16
+	bne	$18, 1b
+
+	ret
+	nop
+	unop
+	nop
+
+	.end copy_page
diff --git a/arch/alpha/lib/ev6-clear_page.S b/arch/alpha/lib/ev6-clear_page.S
new file mode 100644
index 000000000..adf4f7be0
--- /dev/null
+++ b/arch/alpha/lib/ev6-clear_page.S
@@ -0,0 +1,54 @@
+/*
+ * arch/alpha/lib/ev6-clear_page.S
+ *
+ * Zero an entire page.
+ */
+
+        .text
+        .align 4
+        .global clear_page
+        .ent clear_page
+clear_page:
+        .prologue 0
+
+	lda	$0,128
+	lda	$1,125
+	addq	$16,64,$2
+	addq	$16,128,$3
+
+	addq	$16,192,$17
+	wh64	($16)
+	wh64	($2)
+	wh64	($3)
+
+1:	wh64	($17)
+	stq	$31,0($16)
+	subq	$0,1,$0
+	subq	$1,1,$1
+
+	stq	$31,8($16)
+	stq	$31,16($16)
+	addq	$17,64,$2
+	nop
+
+	stq	$31,24($16)
+	stq	$31,32($16)
+	cmovgt	$1,$2,$17
+	nop
+
+	stq	$31,40($16)
+	stq	$31,48($16)
+	nop
+	nop
+
+	stq	$31,56($16)
+	addq	$16,64,$16
+	nop
+	bne	$0,1b
+
+	ret
+	nop
+	nop
+	nop
+
+	.end clear_page
diff --git a/arch/alpha/lib/ev6-copy_page.S b/arch/alpha/lib/ev6-copy_page.S
new file mode 100644
index 000000000..b789db192
--- /dev/null
+++ b/arch/alpha/lib/ev6-copy_page.S
@@ -0,0 +1,203 @@
+/*
+ * arch/alpha/lib/ev6-copy_page.S
+ *
+ * Copy an entire page.
+ */
+
+/* The following comparison of this routine vs the normal copy_page.S
+   was written by an unnamed ev6 hardware designer and forwarded to me
+   via Steven Hobbs <hobbs@steven.zko.dec.com>.
+ 
+   First Problem: STQ overflows.
+   -----------------------------
+
+	It would be nice if EV6 handled every resource overflow efficiently,
+	but for some it doesn't.  Including store queue overflows.  It causes
+	a trap and a restart of the pipe.
+
+	To get around this we sometimes use (to borrow a term from a VSSAD
+	researcher) "aeration".  The idea is to slow the rate at which the
+	processor receives valid instructions by inserting nops in the fetch
+	path.  In doing so, you can prevent the overflow and actually make
+	the code run faster.  You can, of course, take advantage of the fact
+	that the processor can fetch at most 4 aligned instructions per cycle.
+
+	I inserted enough nops to force it to take 10 cycles to fetch the
+	loop code.  In theory, EV6 should be able to execute this loop in
+	9 cycles but I was not able to get it to run that fast -- the initial
+	conditions were such that I could not reach this optimum rate on
+	(chaotic) EV6.  I wrote the code such that everything would issue
+	in order. 
+
+   Second Problem: Dcache index matches.
+   -------------------------------------
+
+	If you are going to use this routine on random aligned pages, there
+	is a 25% chance that the pages will be at the same dcache indices.
+	This results in many nasty memory traps without care.
+
+	The solution is to schedule the prefetches to avoid the memory
+	conflicts.  I schedule the wh64 prefetches farther ahead of the
+	read prefetches to avoid this problem.
+
+   Third Problem: Needs more prefetching.
+   --------------------------------------
+
+	In order to improve the code I added deeper prefetching to take the
+	most advantage of EV6's bandwidth.
+
+	I also prefetched the read stream. Note that adding the read prefetch
+	forced me to add another cycle to the inner-most kernel - up to 11
+	from the original 8 cycles per iteration.  We could improve performance
+	further by unrolling the loop and doing multiple prefetches per cycle.
+
+   I think that the code below will be very robust and fast code for the
+   purposes of copying aligned pages.  It is slower when both source and
+   destination pages are in the dcache, but it is my guess that this is
+   less important than the dcache miss case.  */
+
+
+	.text
+	.align 4
+	.global copy_page
+	.ent copy_page
+copy_page:
+	.prologue 0
+
+	/* Prefetch 5 read cachelines; write-hint 10 cache lines.  */
+	wh64	($16)
+	ldl	$31,0($17)
+	ldl	$31,64($17)
+	lda	$1,1*64($16)
+
+	wh64	($1)
+	ldl	$31,128($17)
+	ldl	$31,192($17)
+	lda	$1,2*64($16)
+
+	wh64	($1)
+	ldl	$31,256($17)
+	lda	$18,118
+	lda	$1,3*64($16)
+
+	wh64	($1)
+	nop
+	lda	$1,4*64($16)
+	lda	$2,5*64($16)
+
+	wh64	($1)
+	wh64	($2)
+	lda	$1,6*64($16)
+	lda	$2,7*64($16)
+
+	wh64	($1)
+	wh64	($2)
+	lda	$1,8*64($16)
+	lda	$2,9*64($16)
+
+	wh64	($1)
+	wh64	($2)
+	lda	$19,10*64($16)
+	nop
+
+	/* Main prefetching/write-hinting loop.  */
+1:	ldq	$0,0($17)
+	ldq	$1,8($17)
+	unop
+	unop
+
+	unop
+	unop
+	ldq	$2,16($17)
+	ldq	$3,24($17)
+
+	ldq	$4,32($17)
+	ldq	$5,40($17)
+	unop
+	unop
+
+	unop
+	unop
+	ldq	$6,48($17)
+	ldq	$7,56($17)
+
+	ldl	$31,320($17)
+	unop
+	unop
+	unop
+
+	/* This gives the extra cycle of aeration above the minimum.  */
+	unop			
+	unop
+	unop
+	unop
+
+	wh64	($19)
+	unop
+	unop
+	unop
+
+	stq	$0,0($16)
+	subq	$18,1,$18
+	stq	$1,8($16)
+	unop
+
+	unop
+	stq	$2,16($16)
+	addq	$17,64,$17
+	stq	$3,24($16)
+
+	stq	$4,32($16)
+	stq	$5,40($16)
+	addq	$19,64,$19
+	unop
+
+	stq	$6,48($16)
+	stq	$7,56($16)
+	addq	$16,64,$16
+	bne	$18, 1b
+
+	/* Prefetch the final 5 cache lines of the read stream.  */
+	lda	$18,10
+	ldl	$31,320($17)
+	ldl	$31,384($17)
+	ldl	$31,448($17)
+
+	ldl	$31,512($17)
+	ldl	$31,576($17)
+	nop
+	nop
+
+	/* Non-prefetching, non-write-hinting cleanup loop for the
+	   final 10 cache lines.  */
+2:	ldq	$0,0($17)
+	ldq	$1,8($17)
+	ldq	$2,16($17)
+	ldq	$3,24($17)
+
+	ldq	$4,32($17)
+	ldq	$5,40($17)
+	ldq	$6,48($17)
+	ldq	$7,56($17)
+
+	stq	$0,0($16)
+	subq	$18,1,$18
+	stq	$1,8($16)
+	addq	$17,64,$17
+
+	stq	$2,16($16)
+	stq	$3,24($16)
+	stq	$4,32($16)
+	stq	$5,40($16)
+
+	stq	$6,48($16)
+	stq	$7,56($16)
+	addq	$16,64,$16
+	bne	$18, 2b
+
+	ret
+	nop
+	unop
+	nop
+
+	.end copy_page