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
128
129
130
131
132
133
134
135
136
137
138
|
/*
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-priv.h>
#if defined(rgo_priv_fastimpl)
__asm__ (
".global rgo_memcpy\n"
"rgo_memcpy:\n"
/*
void const * in
sus_typ_usz num
void * out
*/
#if defined(sus_arch_amd64)
/* 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(sus_archfeat_avx)
".big256cpy:\n"
"cmpq $0x20,%rsi\n"
"jl .big128cpy\n"
"vmovups (%rdi),%ymm0\n"
"vmovups %ymm0,(%rdx)\n"
"addq $0x20,%rdi\n"
"addq $0x20,%rdx\n"
"subq $0x20,%rsi\n"
"jmp .big256cpy\n"
#endif
".big128cpy:\n"
"cmpq $0x10,%rsi\n"
"jl .wrdcpy\n"
"movdqu (%rdi),%xmm0\n"
"movdqu %xmm0,(%rdx)\n"
"addq $0x10,%rdi\n"
"addq $0x10,%rdx\n"
"subq $0x10,%rsi\n"
"jmp .big128cpy\n"
".wrdcpy:\n"
"cmpq $0x8,%rsi\n"
"jl .bytecpy\n"
"movq (%rdi),%rcx\n"
"movq %rcx,(%rdx)\n"
"addq $0x8,%rdi\n"
"addq $0x8,%rdx\n"
"subq $0x8,%rsi\n"
"jmp .wrdcpy\n"
".bytecpy:\n"
"testq %rsi,%rsi\n"
"jz .done\n"
"movb (%rdi),%cl\n"
"movb %cl,(%rdx)\n"
"incq %rdi\n"
"incq %rdx\n"
"decq %rsi\n"
"jmp .bytecpy\n"
".done:\n"
"ret\n"
#elif defined(sus_arch_ia32)
/* eax: Address of the current input element. */
"movl 0x4(%esp),%eax\n"
/* ecx: Number of remaining elements. */
"movl 0x8(%esp),%ecx\n"
/* edx: Address of the current output element. */
"movl 0xC(%esp),%edx\n"
/* ebx: Current element. */
"pushl %ebx\n" /* ebx must be restored. */
/* xmm0: Current element. */
/* ymm0: Current element. */
#if defined(sus_archfeat_avx)
".big256cpy:\n"
"cmpl $0x20,%ecx\n"
#if defined(sus_archfeat_sse)
"jl .big128cpy\n"
#else
"jl .wrdcpy\n"
#endif
"vmovdqu (%eax),%ymm0\n"
"vmovdqu %ymm0,(%edx)\n"
"addl $0x20,%eax\n"
"addl $0x20,%edx\n"
"subl $0x20,%ecx\n"
"jmp .big256cpy\n"
#endif
#if defined(sus_archfeat_sse)
".big128cpy:\n"
"cmpl $0x10,%ecx\n"
"jl .wrdcpy\n"
#if defined(sus_archfeat_sse2)
"movdqu (%eax),%xmm0\n"
"movdqu %xmm0,(%edx)\n"
#else
"movups (%eax),%xmm0\n"
"movups %xmm0,(%edx)\n"
#endif
"addl $0x10,%eax\n"
"addl $0x10,%edx\n"
"subl $0x10,%ecx\n"
"jmp .big128cpy\n"
#endif
".wrdcpy:\n"
"cmpl $0x4,%ecx\n"
"jl .bytecpy\n"
"movl (%eax),%ebx\n"
"movl %ebx,(%edx)\n"
"addl $0x4,%eax\n"
"addl $0x4,%edx\n"
"subl $0x4,%ecx\n"
"jmp .wrdcpy\n"
".bytecpy:\n"
"testl %ecx,%ecx\n"
"jz .done\n"
"movb (%eax),%bl\n"
"movb %bl,(%edx)\n"
"incl %eax\n"
"incl %edx\n"
"decl %ecx\n"
"jmp .bytecpy\n"
".done:\n"
"popl %ebx\n"
"ret\n"
#endif
);
#else
void rgo_memcpy(void const * const sus_restr _in,sus_typ_usz const _num,void * const sus_restr _out) {
sus_typ_u8 const * in = (sus_typ_u8 const *)_in;
sus_typ_u8 * sus_restr out = (sus_typ_u8 *)_out;
sus_typ_u8 const * const afterbuf = in + _num;
for (;in != afterbuf;++in,++out) {*out = *in;}
}
#endif
|
