1 files changed, 2 insertions, 91 deletions
diff --git a/arch/x86_64/lib/memcpy.S b/arch/x86_64/lib/memcpy.S
index c6c46494fef5..92dd80544602 100644
--- a/arch/x86_64/lib/memcpy.S
+++ b/arch/x86_64/lib/memcpy.S
@@ -11,6 +11,8 @@
 * 
 * Output:
 * rax original destination
+ * 
+ * TODO: check best memcpy for PSC
 */     
        .globl __memcpy
@@ -18,95 +20,6 @@
        .p2align 4
 __memcpy:
 memcpy:         
-        pushq %rbx
-        movq %rdi,%rax
-        movl %edx,%ecx
-        shrl $6,%ecx
-        jz .Lhandle_tail
-        
-        .p2align 4
-.Lloop_64:
-        decl %ecx
-        
-        movq (%rsi),%r11
-        movq 8(%rsi),%r8
-        movq %r11,(%rdi)
-        movq %r8,1*8(%rdi)
-        movq 2*8(%rsi),%r9
-        movq 3*8(%rsi),%r10
-        movq %r9,2*8(%rdi)
-        movq %r10,3*8(%rdi)
-                
-        movq 4*8(%rsi),%r11
-        movq 5*8(%rsi),%r8
-        movq %r11,4*8(%rdi)
-        movq %r8,5*8(%rdi)
-        movq 6*8(%rsi),%r9
-        movq 7*8(%rsi),%r10
-        movq %r9,6*8(%rdi)
-        movq %r10,7*8(%rdi)
-        leaq 64(%rsi),%rsi
-        leaq 64(%rdi),%rdi
-        jnz  .Lloop_64
-.Lhandle_tail:
-        movl %edx,%ecx
-        andl $63,%ecx
-        shrl $3,%ecx
-        jz   .Lhandle_7
-        .p2align 4
-.Lloop_8: 
-        decl %ecx
-        movq (%rsi),%r8
-        movq %r8,(%rdi) 
-        leaq 8(%rdi),%rdi
-        leaq 8(%rsi),%rsi
-        jnz  .Lloop_8
-.Lhandle_7:
-        movl %edx,%ecx
-        andl $7,%ecx
-        jz .Lende
-        .p2align 4
-.Lloop_1:
-        movb (%rsi),%r8b
-        movb %r8b,(%rdi) 
-        incq %rdi
-        incq %rsi
-        decl %ecx
-        jnz .Lloop_1
-        
-.Lende:         
-        popq %rbx
-        ret
-.Lfinal:
-        
-        /* C stepping K8 run faster using the string copy instructions.
-           It is also a lot simpler. Use this when possible */
-        
-        .section .altinstructions,"a"
-        .align 8
-        .quad  memcpy
-        .quad  memcpy_c
-        .byte  X86_FEATURE_K8_C
-        .byte  .Lfinal-memcpy
-        .byte  memcpy_c_end-memcpy_c    
-        .previous
-        .section .altinstr_replacement,"ax"
- /* rdi destination
-  * rsi source
-  * rdx count
-  */                    
-memcpy_c:
        movq %rdi,%rax
        movl %edx,%ecx
        shrl $3,%ecx
@@ -117,5 +30,3 @@ memcpy_c:
        rep
        movsb
        ret
-memcpy_c_end:
-        .previous

diff --git a/arch/x86_64/lib/memcpy.S b/arch/x86_64/lib/memcpy.S index c6c46494fef5..92dd80544602 100644 --- a/arch/x86_64/lib/memcpy.S +++ b/arch/x86_64/lib/memcpy.S
@@ -11,6 +11,8 @@
11	*	11	*
12	* Output:	12	* Output:
13	* rax original destination	13	* rax original destination
		14	*
		15	* TODO: check best memcpy for PSC
14	*/	16	*/
15		17
16	.globl __memcpy	18	.globl __memcpy
@@ -18,95 +20,6 @@
18	.p2align 4	20	.p2align 4
19	__memcpy:	21	__memcpy:
20	memcpy:	22	memcpy:
21	pushq %rbx
22	movq %rdi,%rax
23
24	movl %edx,%ecx
25	shrl $6,%ecx
26	jz .Lhandle_tail
27
28	.p2align 4
29	.Lloop_64:
30	decl %ecx
31
32	movq (%rsi),%r11
33	movq 8(%rsi),%r8
34
35	movq %r11,(%rdi)
36	movq %r8,1*8(%rdi)
37
38	movq 2*8(%rsi),%r9
39	movq 3*8(%rsi),%r10
40
41	movq %r9,2*8(%rdi)
42	movq %r10,3*8(%rdi)
43
44	movq 4*8(%rsi),%r11
45	movq 5*8(%rsi),%r8
46
47	movq %r11,4*8(%rdi)
48	movq %r8,5*8(%rdi)
49
50	movq 6*8(%rsi),%r9
51	movq 7*8(%rsi),%r10
52
53	movq %r9,6*8(%rdi)
54	movq %r10,7*8(%rdi)
55
56	leaq 64(%rsi),%rsi
57	leaq 64(%rdi),%rdi
58	jnz .Lloop_64
59
60	.Lhandle_tail:
61	movl %edx,%ecx
62	andl $63,%ecx
63	shrl $3,%ecx
64	jz .Lhandle_7
65	.p2align 4
66	.Lloop_8:
67	decl %ecx
68	movq (%rsi),%r8
69	movq %r8,(%rdi)
70	leaq 8(%rdi),%rdi
71	leaq 8(%rsi),%rsi
72	jnz .Lloop_8
73
74	.Lhandle_7:
75	movl %edx,%ecx
76	andl $7,%ecx
77	jz .Lende
78	.p2align 4
79	.Lloop_1:
80	movb (%rsi),%r8b
81	movb %r8b,(%rdi)
82	incq %rdi
83	incq %rsi
84	decl %ecx
85	jnz .Lloop_1
86
87	.Lende:
88	popq %rbx
89	ret
90	.Lfinal:
91
92	/* C stepping K8 run faster using the string copy instructions.
93	It is also a lot simpler. Use this when possible */
94
95	.section .altinstructions,"a"
96	.align 8
97	.quad memcpy
98	.quad memcpy_c
99	.byte X86_FEATURE_K8_C
100	.byte .Lfinal-memcpy
101	.byte memcpy_c_end-memcpy_c
102	.previous
103
104	.section .altinstr_replacement,"ax"
105	/* rdi destination
106	* rsi source
107	* rdx count
108	*/
109	memcpy_c:
110	movq %rdi,%rax	23	movq %rdi,%rax
111	movl %edx,%ecx	24	movl %edx,%ecx
112	shrl $3,%ecx	25	shrl $3,%ecx
@@ -117,5 +30,3 @@ memcpy_c:
117	rep	30	rep
118	movsb	31	movsb
119	ret	32	ret
120	memcpy_c_end:
121	.previous