[PATCH] x86_64: Undo the earlier changes to remove unrolled copy/mems…

…et functions

They cause quite bad performance regressions on Netburst
This is temporary until we can get new optimized functions
for these CPUs.

This undoes changes that were done in 2.6.15 and in 2.6.16-rc1,
essentially bringing the code back to 2.6.14 level. Only change
is I renamed the X86_FEATURE_K8_C flag to X86_FEATURE_REP_GOOD
and fixed the check for the flag and also fixed some comments.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

arch/x86_64/kernel/setup.c

@@ -877,6 +877,7 @@ static void __init amd_detect_cmp(struct cpuinfo_x86 *c)
 static int __init init_amd(struct cpuinfo_x86 *c)
 {
 	int r;
+	unsigned level;
 
 #ifdef CONFIG_SMP
 	unsigned long value;
@@ -899,6 +900,11 @@ static int __init init_amd(struct cpuinfo_x86 *c)
 	   3DNow is IDd by bit 31 in extended CPUID (1*32+31) anyway */
 	clear_bit(0*32+31, &c->x86_capability);
 
+	/* On C+ stepping K8 rep microcode works well for copy/memset */
+	level = cpuid_eax(1);
+	if (c->x86 == 15 && ((level >= 0x0f48 && level < 0x0f50) || level >= 0x0f58))
+		set_bit(X86_FEATURE_REP_GOOD, &c->x86_capability);
+
 	r = get_model_name(c);
 	if (!r) { 
 		switch (c->x86) { 

arch/x86_64/lib/clear_page.S

@@ -5,8 +5,46 @@
 	.globl clear_page
 	.p2align 4
 clear_page:
+	xorl   %eax,%eax
+	movl   $4096/64,%ecx
+	.p2align 4
+.Lloop:
+	decl	%ecx
+#define PUT(x) movq %rax,x*8(%rdi)
+	movq %rax,(%rdi)
+	PUT(1)
+	PUT(2)
+	PUT(3)
+	PUT(4)
+	PUT(5)
+	PUT(6)
+	PUT(7)
+	leaq	64(%rdi),%rdi
+	jnz	.Lloop
+	nop
+	ret
+clear_page_end:
+
+	/* Some CPUs run faster using the string instructions.
+	   It is also a lot simpler. Use this when possible */
+
+#include <asm/cpufeature.h>
+
+	.section .altinstructions,"a"
+	.align 8
+	.quad  clear_page
+	.quad  clear_page_c
+	.byte  X86_FEATURE_REP_GOOD
+	.byte  clear_page_end-clear_page
+	.byte  clear_page_c_end-clear_page_c
+	.previous
+
+	.section .altinstr_replacement,"ax"
+clear_page_c:
 	movl $4096/8,%ecx
 	xorl %eax,%eax
 	rep 
 	stosq
 	ret
+clear_page_c_end:
+	.previous

arch/x86_64/lib/copy_page.S

@@ -8,7 +8,94 @@
 	.globl copy_page
 	.p2align 4
 copy_page:
+	subq	$3*8,%rsp
+	movq	%rbx,(%rsp)
+	movq	%r12,1*8(%rsp)
+	movq	%r13,2*8(%rsp)
+
+	movl	$(4096/64)-5,%ecx
+	.p2align 4
+.Loop64:
+  	dec     %rcx
+
+	movq        (%rsi), %rax
+	movq      8 (%rsi), %rbx
+	movq     16 (%rsi), %rdx
+	movq     24 (%rsi), %r8
+	movq     32 (%rsi), %r9
+	movq     40 (%rsi), %r10
+	movq     48 (%rsi), %r11
+	movq     56 (%rsi), %r12
+
+	prefetcht0 5*64(%rsi)
+
+	movq     %rax,    (%rdi)
+	movq     %rbx,  8 (%rdi)
+	movq     %rdx, 16 (%rdi)
+	movq     %r8,  24 (%rdi)
+	movq     %r9,  32 (%rdi)
+	movq     %r10, 40 (%rdi)
+	movq     %r11, 48 (%rdi)
+	movq     %r12, 56 (%rdi)
+
+	leaq    64 (%rsi), %rsi
+	leaq    64 (%rdi), %rdi
+
+	jnz     .Loop64
+
+	movl	$5,%ecx
+	.p2align 4
+.Loop2:
+	decl   %ecx
+
+	movq        (%rsi), %rax
+	movq      8 (%rsi), %rbx
+	movq     16 (%rsi), %rdx
+	movq     24 (%rsi), %r8
+	movq     32 (%rsi), %r9
+	movq     40 (%rsi), %r10
+	movq     48 (%rsi), %r11
+	movq     56 (%rsi), %r12
+
+	movq     %rax,    (%rdi)
+	movq     %rbx,  8 (%rdi)
+	movq     %rdx, 16 (%rdi)
+	movq     %r8,  24 (%rdi)
+	movq     %r9,  32 (%rdi)
+	movq     %r10, 40 (%rdi)
+	movq     %r11, 48 (%rdi)
+	movq     %r12, 56 (%rdi)
+
+	leaq	64(%rdi),%rdi
+	leaq	64(%rsi),%rsi
+
+	jnz	.Loop2
+
+	movq	(%rsp),%rbx
+	movq	1*8(%rsp),%r12
+	movq	2*8(%rsp),%r13
+	addq	$3*8,%rsp
+	ret
+
+	/* Some CPUs run faster using the string copy instructions.
+	   It is also a lot simpler. Use this when possible */
+
+#include <asm/cpufeature.h>
+
+	.section .altinstructions,"a"
+	.align 8
+	.quad  copy_page
+	.quad  copy_page_c
+	.byte  X86_FEATURE_REP_GOOD
+	.byte  copy_page_c_end-copy_page_c
+	.byte  copy_page_c_end-copy_page_c
+	.previous
+
+	.section .altinstr_replacement,"ax"
+copy_page_c:
 	movl $4096/8,%ecx
 	rep 
 	movsq 
 	ret
+copy_page_c_end:
+	.previous