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
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
|
/* memcmp with SSE2
Copyright (C) 2009-2013 Free Software Foundation, Inc.
Contributed by Intel Corporation.
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>
.text
ENTRY (memcmp)
test %rdx, %rdx
jz L(finz)
cmpq $1, %rdx
jle L(finr1b)
subq %rdi, %rsi
movq %rdx, %r10
cmpq $32, %r10
jge L(gt32)
/* Handle small chunks and last block of less than 32 bytes. */
L(small):
testq $1, %r10
jz L(s2b)
movzbl (%rdi), %eax
movzbl (%rdi, %rsi), %edx
subq $1, %r10
je L(finz1)
addq $1, %rdi
subl %edx, %eax
jnz L(exit)
L(s2b):
testq $2, %r10
jz L(s4b)
movzwl (%rdi), %eax
movzwl (%rdi, %rsi), %edx
subq $2, %r10
je L(fin2_7)
addq $2, %rdi
cmpl %edx, %eax
jnz L(fin2_7)
L(s4b):
testq $4, %r10
jz L(s8b)
movl (%rdi), %eax
movl (%rdi, %rsi), %edx
subq $4, %r10
je L(fin2_7)
addq $4, %rdi
cmpl %edx, %eax
jnz L(fin2_7)
L(s8b):
testq $8, %r10
jz L(s16b)
movq (%rdi), %rax
movq (%rdi, %rsi), %rdx
subq $8, %r10
je L(fin2_7)
addq $8, %rdi
cmpq %rdx, %rax
jnz L(fin2_7)
L(s16b):
movdqu (%rdi), %xmm1
movdqu (%rdi, %rsi), %xmm0
pcmpeqb %xmm0, %xmm1
pmovmskb %xmm1, %edx
xorl %eax, %eax
subl $0xffff, %edx
jz L(finz)
bsfl %edx, %ecx
leaq (%rdi, %rcx), %rcx
movzbl (%rcx), %eax
movzbl (%rsi, %rcx), %edx
jmp L(finz1)
.p2align 4,, 4
L(finr1b):
movzbl (%rdi), %eax
movzbl (%rsi), %edx
L(finz1):
subl %edx, %eax
L(exit):
ret
.p2align 4,, 4
L(fin2_7):
cmpq %rdx, %rax
jz L(finz)
movq %rax, %r11
subq %rdx, %r11
bsfq %r11, %rcx
sarq $3, %rcx
salq $3, %rcx
sarq %cl, %rax
movzbl %al, %eax
sarq %cl, %rdx
movzbl %dl, %edx
subl %edx, %eax
ret
.p2align 4,, 4
L(finz):
xorl %eax, %eax
ret
/* For blocks bigger than 32 bytes
1. Advance one of the addr pointer to be 16B aligned.
2. Treat the case of both addr pointers aligned to 16B
separately to avoid movdqu.
3. Handle any blocks of greater than 64 consecutive bytes with
unrolling to reduce branches.
4. At least one addr pointer is 16B aligned, use memory version
of pcmbeqb.
*/
.p2align 4,, 4
L(gt32):
movq %rdx, %r11
addq %rdi, %r11
movq %rdi, %r8
andq $15, %r8
jz L(16am)
/* Both pointers may be misaligned. */
movdqu (%rdi), %xmm1
movdqu (%rdi, %rsi), %xmm0
pcmpeqb %xmm0, %xmm1
pmovmskb %xmm1, %edx
subl $0xffff, %edx
jnz L(neq)
neg %r8
leaq 16(%rdi, %r8), %rdi
L(16am):
/* Handle two 16B aligned pointers separately. */
testq $15, %rsi
jz L(ATR)
testq $16, %rdi
jz L(A32)
movdqu (%rdi, %rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
L(A32):
movq %r11, %r10
andq $-32, %r10
cmpq %r10, %rdi
jge L(mt16)
/* Pre-unroll to be ready for unrolled 64B loop. */
testq $32, %rdi
jz L(A64)
movdqu (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
movdqu (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
L(A64):
movq %r11, %r10
andq $-64, %r10
cmpq %r10, %rdi
jge L(mt32)
L(A64main):
movdqu (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
movdqu (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
movdqu (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
movdqu (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
cmpq %rdi, %r10
jne L(A64main)
L(mt32):
movq %r11, %r10
andq $-32, %r10
cmpq %r10, %rdi
jge L(mt16)
L(A32main):
movdqu (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
movdqu (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
cmpq %rdi, %r10
jne L(A32main)
L(mt16):
subq %rdi, %r11
je L(finz)
movq %r11, %r10
jmp L(small)
.p2align 4,, 4
L(neq):
bsfl %edx, %ecx
movzbl (%rdi, %rcx), %eax
addq %rdi, %rsi
movzbl (%rsi,%rcx), %edx
jmp L(finz1)
.p2align 4,, 4
L(ATR):
movq %r11, %r10
andq $-32, %r10
cmpq %r10, %rdi
jge L(mt16)
testq $16, %rdi
jz L(ATR32)
movdqa (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
cmpq %rdi, %r10
je L(mt16)
L(ATR32):
movq %r11, %r10
andq $-64, %r10
testq $32, %rdi
jz L(ATR64)
movdqa (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
movdqa (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
L(ATR64):
cmpq %rdi, %r10
je L(mt32)
L(ATR64main):
movdqa (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
movdqa (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
movdqa (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
movdqa (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
cmpq %rdi, %r10
jne L(ATR64main)
movq %r11, %r10
andq $-32, %r10
cmpq %r10, %rdi
jge L(mt16)
L(ATR32res):
movdqa (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
movdqa (%rdi,%rsi), %xmm0
pcmpeqb (%rdi), %xmm0
pmovmskb %xmm0, %edx
subl $0xffff, %edx
jnz L(neq)
addq $16, %rdi
cmpq %r10, %rdi
jne L(ATR32res)
subq %rdi, %r11
je L(finz)
movq %r11, %r10
jmp L(small)
/* Align to 16byte to improve instruction fetch. */
.p2align 4,, 4
END(memcmp)
#undef bcmp
weak_alias (memcmp, bcmp)
libc_hidden_builtin_def (memcmp)
|
