1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
|
/*
Copyright 2022 Gabriel Jensen.
This Source Code Form is subject to the terms of the Mozilla Public License, v. 2.0.
If a copy of the MPL was not distributed with this file, You can obtain one at https://mozilla.org/MPL/2.0/.
*/
#include <rgo.h>
.global rgo_memcpy
rgo_memcpy:
/*
void const * in
size_t num
void * out
*/
#if defined(__i386__)
/* eax: Address of the current input element. */
movl 0x4(%esp),%eax
/* ecx: Number of remaining elements. */
movl 0x8(%esp),%ecx
/* edx: Address of the current output element. */
movl 0xC(%esp),%edx
/* ebx: Current element. */
pushl %ebx /* ebx must be restored. */
/* xmm0: Current element. */
/* ymm0: Current element. */
#if defined(__AVX__)
.big256cpy:
cmpl $0x20,%ecx
#if defined(__SSE__)
jl .big128cpy
#else
jl .wrdcpy
#endif
vmovdqu (%eax),%ymm0
vmovdqu %ymm0,(%edx)
addl $0x20,%eax
addl $0x20,%edx
subl $0x20,%ecx
jmp .big256cpy
#endif
#if defined(__SSE__)
.big128cpy:
cmpl $0x10,%ecx
jl .wrdcpy
#if defined(__SSE2__)
movdqu (%eax),%xmm0
movdqu %xmm0,(%edx)
#else
movups (%eax),%xmm0
movups %xmm0,(%edx)
#endif
addl $0x10,%eax
addl $0x10,%edx
subl $0x10,%ecx
jmp .big128cpy
#endif
.wrdcpy:
cmpl $0x4,%ecx
jl .bytecpy
movl (%eax),%ebx
movl %ebx,(%edx)
addl $0x4,%eax
addl $0x4,%edx
subl $0x4,%ecx
jmp .wrdcpy
.bytecpy:
testl %ecx,%ecx
jz .done
movb (%eax),%bl
movb %bl,(%edx)
incl %eax
incl %edx
decl %ecx
jmp .bytecpy
.done:
popl %ebx
ret
#elif defined(__x86_64__)
/* rdi: Address of the current input element. */
/* rsi: Number of remaining elements. */
/* rdx: Address of the current output element. */
/* rcx: Current element. */
/* xmm0: Current element. */
/* ymm0: Current element. */
#if defined(__AVX__)
.big256cpy:
cmpq $0x20,%rsi
jl .big128cpy
vmovups (%rdi),%ymm0
vmovups %ymm0,(%rdx)
addq $0x20,%rdi
addq $0x20,%rdx
subq $0x20,%rsi
jmp .big256cpy
#endif
.big128cpy:
cmpq $0x10,%rsi
jl .wrdcpy
movdqu (%rdi),%xmm0
movdqu %xmm0,(%rdx)
addq $0x10,%rdi
addq $0x10,%rdx
subq $0x10,%rsi
jmp .big128cpy
.wrdcpy:
cmpq $0x8,%rsi
jl .bytecpy
movq (%rdi),%rcx
movq %rcx,(%rdx)
addq $0x8,%rdi
addq $0x8,%rdx
subq $0x8,%rsi
jmp .wrdcpy
.bytecpy:
testq %rsi,%rsi
jz .done
movb (%rdi),%cl
movb %cl,(%rdx)
incq %rdi
incq %rdx
decq %rsi
jmp .bytecpy
.done:
ret
#endif
|
