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
359
360
361
362
363
364
365
366
367
368
369
370
371
|
/* Wrapper implementations of vector math functions.
Copyright (C) 2014-2016 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/>. */
/* SSE2 ISA version as wrapper to scalar. */
.macro WRAPPER_IMPL_SSE2 callee
subq $40, %rsp
cfi_adjust_cfa_offset(40)
movaps %xmm0, (%rsp)
call JUMPTARGET(\callee)
movss %xmm0, 16(%rsp)
movss 4(%rsp), %xmm0
call JUMPTARGET(\callee)
movss %xmm0, 20(%rsp)
movss 8(%rsp), %xmm0
call JUMPTARGET(\callee)
movss %xmm0, 24(%rsp)
movss 12(%rsp), %xmm0
call JUMPTARGET(\callee)
movss 16(%rsp), %xmm3
movss 20(%rsp), %xmm2
movss 24(%rsp), %xmm1
movss %xmm0, 28(%rsp)
unpcklps %xmm1, %xmm3
unpcklps %xmm0, %xmm2
unpcklps %xmm2, %xmm3
movaps %xmm3, %xmm0
addq $40, %rsp
cfi_adjust_cfa_offset(-40)
ret
.endm
/* 2 argument SSE2 ISA version as wrapper to scalar. */
.macro WRAPPER_IMPL_SSE2_ff callee
subq $56, %rsp
cfi_adjust_cfa_offset(56)
movaps %xmm0, (%rsp)
movaps %xmm1, 16(%rsp)
call JUMPTARGET(\callee)
movss %xmm0, 32(%rsp)
movss 4(%rsp), %xmm0
movss 20(%rsp), %xmm1
call JUMPTARGET(\callee)
movss %xmm0, 36(%rsp)
movss 8(%rsp), %xmm0
movss 24(%rsp), %xmm1
call JUMPTARGET(\callee)
movss %xmm0, 40(%rsp)
movss 12(%rsp), %xmm0
movss 28(%rsp), %xmm1
call JUMPTARGET(\callee)
movss 32(%rsp), %xmm3
movss 36(%rsp), %xmm2
movss 40(%rsp), %xmm1
movss %xmm0, 44(%rsp)
unpcklps %xmm1, %xmm3
unpcklps %xmm0, %xmm2
unpcklps %xmm2, %xmm3
movaps %xmm3, %xmm0
addq $56, %rsp
cfi_adjust_cfa_offset(-56)
ret
.endm
/* 3 argument SSE2 ISA version as wrapper to scalar. */
.macro WRAPPER_IMPL_SSE2_fFF callee
pushq %rbp
cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0)
pushq %rbx
cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbx, 0)
movq %rdi, %rbp
movq %rsi, %rbx
subq $40, %rsp
cfi_adjust_cfa_offset(40)
leaq 24(%rsp), %rsi
leaq 28(%rsp), %rdi
movaps %xmm0, (%rsp)
call JUMPTARGET(\callee)
leaq 24(%rsp), %rsi
leaq 28(%rsp), %rdi
movss 28(%rsp), %xmm0
movss %xmm0, 0(%rbp)
movaps (%rsp), %xmm1
movss 24(%rsp), %xmm0
movss %xmm0, (%rbx)
movaps %xmm1, %xmm0
shufps $85, %xmm1, %xmm0
call JUMPTARGET(\callee)
movss 28(%rsp), %xmm0
leaq 24(%rsp), %rsi
movss %xmm0, 4(%rbp)
leaq 28(%rsp), %rdi
movaps (%rsp), %xmm1
movss 24(%rsp), %xmm0
movss %xmm0, 4(%rbx)
movaps %xmm1, %xmm0
unpckhps %xmm1, %xmm0
call JUMPTARGET(\callee)
movaps (%rsp), %xmm1
leaq 24(%rsp), %rsi
leaq 28(%rsp), %rdi
movss 28(%rsp), %xmm0
shufps $255, %xmm1, %xmm1
movss %xmm0, 8(%rbp)
movss 24(%rsp), %xmm0
movss %xmm0, 8(%rbx)
movaps %xmm1, %xmm0
call JUMPTARGET(\callee)
movss 28(%rsp), %xmm0
movss %xmm0, 12(%rbp)
movss 24(%rsp), %xmm0
movss %xmm0, 12(%rbx)
addq $40, %rsp
cfi_adjust_cfa_offset(-40)
popq %rbx
cfi_adjust_cfa_offset (-8)
cfi_restore (%rbx)
popq %rbp
cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp)
ret
.endm
/* AVX/AVX2 ISA version as wrapper to SSE ISA version. */
.macro WRAPPER_IMPL_AVX callee
pushq %rbp
cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp
cfi_def_cfa_register (%rbp)
andq $-32, %rsp
subq $32, %rsp
vextractf128 $1, %ymm0, (%rsp)
vzeroupper
call HIDDEN_JUMPTARGET(\callee)
vmovaps %xmm0, 16(%rsp)
vmovaps (%rsp), %xmm0
call HIDDEN_JUMPTARGET(\callee)
vmovaps %xmm0, %xmm1
vmovaps 16(%rsp), %xmm0
vinsertf128 $1, %xmm1, %ymm0, %ymm0
movq %rbp, %rsp
cfi_def_cfa_register (%rsp)
popq %rbp
cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp)
ret
.endm
/* 2 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */
.macro WRAPPER_IMPL_AVX_ff callee
pushq %rbp
cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp
cfi_def_cfa_register (%rbp)
andq $-32, %rsp
subq $64, %rsp
vextractf128 $1, %ymm0, 16(%rsp)
vextractf128 $1, %ymm1, (%rsp)
vzeroupper
call HIDDEN_JUMPTARGET(\callee)
vmovaps %xmm0, 32(%rsp)
vmovaps 16(%rsp), %xmm0
vmovaps (%rsp), %xmm1
call HIDDEN_JUMPTARGET(\callee)
vmovaps %xmm0, %xmm1
vmovaps 32(%rsp), %xmm0
vinsertf128 $1, %xmm1, %ymm0, %ymm0
movq %rbp, %rsp
cfi_def_cfa_register (%rsp)
popq %rbp
cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp)
ret
.endm
/* 3 argument AVX/AVX2 ISA version as wrapper to SSE ISA version. */
.macro WRAPPER_IMPL_AVX_fFF callee
pushq %rbp
cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp
cfi_def_cfa_register (%rbp)
andq $-32, %rsp
pushq %r13
cfi_adjust_cfa_offset (8)
cfi_rel_offset (%r13, 0)
pushq %r14
cfi_adjust_cfa_offset (8)
cfi_rel_offset (%r14, 0)
subq $48, %rsp
movq %rsi, %r14
vmovaps %ymm0, (%rsp)
movq %rdi, %r13
vmovaps 16(%rsp), %xmm1
vmovaps %xmm1, 32(%rsp)
vzeroupper
vmovaps (%rsp), %xmm0
call HIDDEN_JUMPTARGET(\callee)
vmovaps 32(%rsp), %xmm0
lea (%rsp), %rdi
lea 16(%rsp), %rsi
call HIDDEN_JUMPTARGET(\callee)
vmovaps (%rsp), %xmm0
vmovaps 16(%rsp), %xmm1
vmovaps %xmm0, 16(%r13)
vmovaps %xmm1, 16(%r14)
addq $48, %rsp
popq %r14
cfi_adjust_cfa_offset (-8)
cfi_restore (%r14)
popq %r13
cfi_adjust_cfa_offset (-8)
cfi_restore (%r13)
movq %rbp, %rsp
cfi_def_cfa_register (%rsp)
popq %rbp
cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp)
ret
.endm
/* AVX512 ISA version as wrapper to AVX2 ISA version. */
.macro WRAPPER_IMPL_AVX512 callee
pushq %rbp
cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp
cfi_def_cfa_register (%rbp)
andq $-64, %rsp
subq $128, %rsp
/* Below is encoding for vmovups %zmm0, (%rsp). */
.byte 0x62
.byte 0xf1
.byte 0x7c
.byte 0x48
.byte 0x11
.byte 0x04
.byte 0x24
vmovupd (%rsp), %ymm0
call HIDDEN_JUMPTARGET(\callee)
vmovupd %ymm0, 64(%rsp)
vmovupd 32(%rsp), %ymm0
call HIDDEN_JUMPTARGET(\callee)
vmovupd %ymm0, 96(%rsp)
/* Below is encoding for vmovups 64(%rsp), %zmm0. */
.byte 0x62
.byte 0xf1
.byte 0x7c
.byte 0x48
.byte 0x10
.byte 0x44
.byte 0x24
.byte 0x01
movq %rbp, %rsp
cfi_def_cfa_register (%rsp)
popq %rbp
cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp)
ret
.endm
/* 2 argument AVX512 ISA version as wrapper to AVX2 ISA version. */
.macro WRAPPER_IMPL_AVX512_ff callee
pushq %rbp
cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp
cfi_def_cfa_register (%rbp)
andq $-64, %rsp
subq $192, %rsp
/* Below is encoding for vmovups %zmm0, (%rsp). */
.byte 0x62
.byte 0xf1
.byte 0x7c
.byte 0x48
.byte 0x11
.byte 0x04
.byte 0x24
/* Below is encoding for vmovups %zmm1, 64(%rsp). */
.byte 0x62
.byte 0xf1
.byte 0x7c
.byte 0x48
.byte 0x11
.byte 0x4c
.byte 0x24
.byte 0x01
vmovups (%rsp), %ymm0
vmovups 64(%rsp), %ymm1
call HIDDEN_JUMPTARGET(\callee)
vmovups %ymm0, 128(%rsp)
vmovups 32(%rsp), %ymm0
vmovups 96(%rsp), %ymm1
call HIDDEN_JUMPTARGET(\callee)
vmovups %ymm0, 160(%rsp)
/* Below is encoding for vmovups 128(%rsp), %zmm0. */
.byte 0x62
.byte 0xf1
.byte 0x7c
.byte 0x48
.byte 0x10
.byte 0x44
.byte 0x24
.byte 0x02
movq %rbp, %rsp
cfi_def_cfa_register (%rsp)
popq %rbp
cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp)
ret
.endm
/* 3 argument AVX512 ISA version as wrapper to AVX2 ISA version. */
.macro WRAPPER_IMPL_AVX512_fFF callee
pushq %rbp
cfi_adjust_cfa_offset (8)
cfi_rel_offset (%rbp, 0)
movq %rsp, %rbp
cfi_def_cfa_register (%rbp)
andq $-64, %rsp
pushq %r12
pushq %r13
subq $176, %rsp
movq %rsi, %r13
/* Below is encoding for vmovaps %zmm0, (%rsp). */
.byte 0x62
.byte 0xf1
.byte 0x7c
.byte 0x48
.byte 0x29
.byte 0x04
.byte 0x24
movq %rdi, %r12
vmovaps (%rsp), %ymm0
call HIDDEN_JUMPTARGET(\callee)
vmovaps 32(%rsp), %ymm0
lea 64(%rsp), %rdi
lea 96(%rsp), %rsi
call HIDDEN_JUMPTARGET(\callee)
vmovaps 64(%rsp), %ymm0
vmovaps 96(%rsp), %ymm1
vmovaps %ymm0, 32(%r12)
vmovaps %ymm1, 32(%r13)
addq $176, %rsp
popq %r13
popq %r12
movq %rbp, %rsp
cfi_def_cfa_register (%rsp)
popq %rbp
cfi_adjust_cfa_offset (-8)
cfi_restore (%rbp)
ret
.endm
|
