about summary refs log tree commit diff
path: root/sysdeps/x86_64/multiarch/strcmp-sse2-unaligned.S
blob: 20b65fa77566b4447ea64ce4bd2d28a9f33e2c67 (plain) (blame)
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
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
/* strcmp with unaligned loads
   Copyright (C) 2013-2015 Free Software Foundation, Inc.
   This file is part of the GNU C Library.

   The GNU C Library is free software; you can redistribute it and/or
   modify it under the terms of the GNU Lesser General Public
   License as published by the Free Software Foundation; either
   version 2.1 of the License, or (at your option) any later version.

   The GNU C Library is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
   Lesser General Public License for more details.

   You should have received a copy of the GNU Lesser General Public
   License along with the GNU C Library; if not, see
   <http://www.gnu.org/licenses/>.  */

#include "sysdep.h"

ENTRY ( __strcmp_sse2_unaligned)
	movl	%edi, %eax
	xorl	%edx, %edx
	pxor	%xmm7, %xmm7
	orl	%esi, %eax
	andl	$4095, %eax
	cmpl	$4032, %eax
	jg	L(cross_page)
	movdqu	(%rdi), %xmm1
	movdqu	(%rsi), %xmm0
	pcmpeqb	%xmm1, %xmm0
	pminub	%xmm1, %xmm0
	pxor	%xmm1, %xmm1
	pcmpeqb	%xmm1, %xmm0
	pmovmskb	%xmm0, %eax
	testq	%rax, %rax
	je	L(next_48_bytes)
L(return):
	bsfq	%rax, %rdx
	movzbl	(%rdi, %rdx), %eax
	movzbl	(%rsi, %rdx), %edx
	subl	%edx, %eax
	ret

	.p2align 4
L(next_48_bytes):
	movdqu	16(%rdi), %xmm6
	movdqu	16(%rsi), %xmm3
	movdqu	32(%rdi), %xmm5
	pcmpeqb	%xmm6, %xmm3
	movdqu	32(%rsi), %xmm2
	pminub	%xmm6, %xmm3
	pcmpeqb	%xmm1, %xmm3
	movdqu	48(%rdi), %xmm4
	pcmpeqb	%xmm5, %xmm2
	pmovmskb	%xmm3, %edx
	movdqu	48(%rsi), %xmm0
	pminub	%xmm5, %xmm2
	pcmpeqb	%xmm1, %xmm2
	pcmpeqb	%xmm4, %xmm0
	pmovmskb	%xmm2, %eax
	salq	$16, %rdx
	pminub	%xmm4, %xmm0
	pcmpeqb	%xmm1, %xmm0
	salq	$32, %rax
	orq	%rdx, %rax
	pmovmskb	%xmm0, %ecx
	movq	%rcx, %rdx
	salq	$48, %rdx
	orq	%rdx, %rax
	jne	L(return)
L(main_loop_header):
	leaq	64(%rdi), %rdx
	movl	$4096, %ecx
	pxor	%xmm9, %xmm9
	andq	$-64, %rdx
	subq	%rdi, %rdx
	leaq	(%rdi, %rdx), %rax
	addq	%rsi, %rdx
	movq	%rdx, %rsi
	andl	$4095, %esi
	subq	%rsi, %rcx
	shrq	$6, %rcx
	movq	%rcx, %rsi
	jmp	L(loop_start)

	.p2align 4
L(loop):
	addq	$64, %rax
	addq	$64, %rdx
L(loop_start):
	testq	%rsi, %rsi
	leaq	-1(%rsi), %rsi
	je	L(loop_cross_page)
L(back_to_loop):
	movdqu	(%rdx), %xmm0
	movdqu	16(%rdx), %xmm1
	movdqa	(%rax), %xmm2
	movdqa	16(%rax), %xmm3
	pcmpeqb	%xmm2, %xmm0
	movdqu	32(%rdx), %xmm5
	pcmpeqb	%xmm3, %xmm1
	pminub	%xmm2, %xmm0
	movdqu	48(%rdx), %xmm6
	pminub	%xmm3, %xmm1
	movdqa	32(%rax), %xmm2
	pminub	%xmm1, %xmm0
	movdqa	48(%rax), %xmm3
	pcmpeqb	%xmm2, %xmm5
	pcmpeqb	%xmm3, %xmm6
	pminub	%xmm2, %xmm5
	pminub	%xmm3, %xmm6
	pminub	%xmm5, %xmm0
	pminub	%xmm6, %xmm0
	pcmpeqb	%xmm7, %xmm0
	pmovmskb	%xmm0, %ecx
	testl	%ecx, %ecx
	je	L(loop)
	pcmpeqb	%xmm7, %xmm5
	movdqu	(%rdx), %xmm0
	pcmpeqb	%xmm7, %xmm1
	movdqa	(%rax), %xmm2
	pcmpeqb	%xmm2, %xmm0
	pminub	%xmm2, %xmm0
	pcmpeqb	%xmm7, %xmm6
	pcmpeqb	%xmm7, %xmm0
	pmovmskb	%xmm1, %ecx
	pmovmskb	%xmm5, %r8d
	pmovmskb	%xmm0, %edi
	salq	$16, %rcx
	salq	$32, %r8
	pmovmskb	%xmm6, %esi
	orq	%r8, %rcx
	orq	%rdi, %rcx
	salq	$48, %rsi
	orq	%rsi, %rcx
	bsfq	%rcx, %rcx
	movzbl	(%rax, %rcx), %eax
	movzbl	(%rdx, %rcx), %edx
	subl	%edx, %eax
	ret

	.p2align 4
L(loop_cross_page):
	xor	%r10, %r10
	movq	%rdx, %r9
	and	$63, %r9
	subq	%r9, %r10

	movdqa	(%rdx, %r10), %xmm0
	movdqa	16(%rdx, %r10), %xmm1
	movdqu	(%rax, %r10), %xmm2
	movdqu	16(%rax, %r10), %xmm3
	pcmpeqb	%xmm2, %xmm0
	movdqa	32(%rdx, %r10), %xmm5
	pcmpeqb	%xmm3, %xmm1
	pminub	%xmm2, %xmm0
	movdqa	48(%rdx, %r10), %xmm6
	pminub	%xmm3, %xmm1
	movdqu	32(%rax, %r10), %xmm2
	movdqu	48(%rax, %r10), %xmm3
	pcmpeqb	%xmm2, %xmm5
	pcmpeqb	%xmm3, %xmm6
	pminub	%xmm2, %xmm5
	pminub	%xmm3, %xmm6

	pcmpeqb	%xmm7, %xmm0
	pcmpeqb	%xmm7, %xmm1
	pcmpeqb	%xmm7, %xmm5
	pcmpeqb	%xmm7, %xmm6

	pmovmskb	%xmm1, %ecx
	pmovmskb	%xmm5, %r8d
	pmovmskb	%xmm0, %edi
	salq	$16, %rcx
	salq	$32, %r8
	pmovmskb	%xmm6, %esi
	orq	%r8, %rdi
	orq	%rcx, %rdi
	salq	$48, %rsi
	orq	%rsi, %rdi
	movq	%r9, %rcx
	movq	$63, %rsi
	shrq	%cl, %rdi
	test	%rdi, %rdi
	je	L(back_to_loop)
	bsfq	%rdi, %rcx
	movzbl	(%rax, %rcx), %eax
	movzbl	(%rdx, %rcx), %edx
	subl	%edx, %eax
	ret

	.p2align 4
L(cross_page_loop):
	cmpb	%cl, %al
	jne	L(different)
	addq	$1, %rdx
	cmpq	$64, %rdx
	je	L(main_loop_header)
L(cross_page):
	movzbl	(%rdi, %rdx), %eax
	movzbl	(%rsi, %rdx), %ecx
	testb	%al, %al
	jne	L(cross_page_loop)
	xorl	%eax, %eax
L(different):
	subl	%ecx, %eax
	ret
END (__strcmp_sse2_unaligned)