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
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=thumbv8.1m.main-none-none-eabi -mattr=+mve.fp -enable-arm-maskedldst %s -o - | FileCheck %s
; i32
; Expand
define arm_aapcs_vfpcc void @ptr_v2i32(<2 x i32> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v2i32:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vmov r2, s0
; CHECK-NEXT: ldrd r0, r1, [r0]
; CHECK-NEXT: str r2, [r0]
; CHECK-NEXT: vmov r0, s2
; CHECK-NEXT: str r0, [r1]
; CHECK-NEXT: bx lr
entry:
%offs = load <2 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v2i32.v2p0(<2 x i32> %v, <2 x ptr> %offs, i32 4, <2 x i1> <i1 true, i1 true>)
ret void
}
; VSTRW.32 Qd, [offs, 0]
define arm_aapcs_vfpcc void @ptr_v4i32(<4 x i32> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v4i32:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vldrw.u32 q1, [r0]
; CHECK-NEXT: vstrw.32 q0, [q1]
; CHECK-NEXT: bx lr
entry:
%offs = load <4 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> %v, <4 x ptr> %offs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>)
ret void
}
; Expand
define arm_aapcs_vfpcc void @ptr_v8i32(<8 x i32> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v8i32:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r6, r7, lr}
; CHECK-NEXT: push {r4, r5, r6, r7, lr}
; CHECK-NEXT: vldrw.u32 q2, [r0]
; CHECK-NEXT: vmov r3, r4, d0
; CHECK-NEXT: vmov r1, r2, d4
; CHECK-NEXT: vmov lr, r12, d5
; CHECK-NEXT: vldrw.u32 q2, [r0, #16]
; CHECK-NEXT: vmov r0, r5, d1
; CHECK-NEXT: str r3, [r1]
; CHECK-NEXT: vmov r1, r7, d4
; CHECK-NEXT: str r4, [r2]
; CHECK-NEXT: vmov r2, r4, d5
; CHECK-NEXT: str.w r0, [lr]
; CHECK-NEXT: vmov r0, r3, d2
; CHECK-NEXT: str.w r5, [r12]
; CHECK-NEXT: vmov r5, r6, d3
; CHECK-NEXT: str r0, [r1]
; CHECK-NEXT: str r3, [r7]
; CHECK-NEXT: str r5, [r2]
; CHECK-NEXT: str r6, [r4]
; CHECK-NEXT: pop {r4, r5, r6, r7, pc}
entry:
%offs = load <8 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v8i32.v8p0(<8 x i32> %v, <8 x ptr> %offs, i32 4, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
ret void
}
; Expand
define arm_aapcs_vfpcc void @ptr_v16i32(<16 x i32> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v16i32:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r6, r7, lr}
; CHECK-NEXT: push {r4, r5, r6, r7, lr}
; CHECK-NEXT: .pad #4
; CHECK-NEXT: sub sp, #4
; CHECK-NEXT: .vsave {d8, d9, d10, d11, d12, d13}
; CHECK-NEXT: vpush {d8, d9, d10, d11, d12, d13}
; CHECK-NEXT: vldrw.u32 q4, [r0]
; CHECK-NEXT: vmov r3, r4, d0
; CHECK-NEXT: vldrw.u32 q5, [r0, #32]
; CHECK-NEXT: vldrw.u32 q6, [r0, #16]
; CHECK-NEXT: vmov r1, r2, d8
; CHECK-NEXT: vmov lr, r12, d9
; CHECK-NEXT: vldrw.u32 q4, [r0, #48]
; CHECK-NEXT: vmov r0, r5, d1
; CHECK-NEXT: str r3, [r1]
; CHECK-NEXT: vmov r1, r3, d12
; CHECK-NEXT: str r4, [r2]
; CHECK-NEXT: vmov r2, r7, d13
; CHECK-NEXT: str.w r0, [lr]
; CHECK-NEXT: vmov r0, r4, d2
; CHECK-NEXT: str.w r5, [r12]
; CHECK-NEXT: vmov r5, r6, d3
; CHECK-NEXT: str r0, [r1]
; CHECK-NEXT: vmov r0, r1, d10
; CHECK-NEXT: str r4, [r3]
; CHECK-NEXT: vmov r3, r4, d11
; CHECK-NEXT: str r5, [r2]
; CHECK-NEXT: vmov r2, r5, d4
; CHECK-NEXT: str r6, [r7]
; CHECK-NEXT: vmov r7, r6, d5
; CHECK-NEXT: str r2, [r0]
; CHECK-NEXT: vmov r0, r2, d8
; CHECK-NEXT: str r5, [r1]
; CHECK-NEXT: vmov r1, r5, d9
; CHECK-NEXT: str r7, [r3]
; CHECK-NEXT: vmov r3, r7, d6
; CHECK-NEXT: str r6, [r4]
; CHECK-NEXT: vmov r6, r4, d7
; CHECK-NEXT: str r3, [r0]
; CHECK-NEXT: str r7, [r2]
; CHECK-NEXT: str r6, [r1]
; CHECK-NEXT: str r4, [r5]
; CHECK-NEXT: vpop {d8, d9, d10, d11, d12, d13}
; CHECK-NEXT: add sp, #4
; CHECK-NEXT: pop {r4, r5, r6, r7, pc}
entry:
%offs = load <16 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v16i32.v16p0(<16 x i32> %v, <16 x ptr> %offs, i32 4, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
ret void
}
; f32
; Expand
define arm_aapcs_vfpcc void @ptr_v2f32(<2 x float> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v2f32:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: ldrd r0, r1, [r0]
; CHECK-NEXT: vstr s0, [r0]
; CHECK-NEXT: vstr s1, [r1]
; CHECK-NEXT: bx lr
entry:
%offs = load <2 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v2f32.v2p0(<2 x float> %v, <2 x ptr> %offs, i32 4, <2 x i1> <i1 true, i1 true>)
ret void
}
; VSTRW.32 Qd, [offs, 0]
define arm_aapcs_vfpcc void @ptr_v4f32(<4 x float> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v4f32:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vldrw.u32 q1, [r0]
; CHECK-NEXT: vstrw.32 q0, [q1]
; CHECK-NEXT: bx lr
entry:
%offs = load <4 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v4f32.v4p0(<4 x float> %v, <4 x ptr> %offs, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>)
ret void
}
; Expand
define arm_aapcs_vfpcc void @ptr_v8f32(<8 x float> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v8f32:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r7, lr}
; CHECK-NEXT: push {r4, r5, r7, lr}
; CHECK-NEXT: vldrw.u32 q2, [r0]
; CHECK-NEXT: vmov r1, lr, d4
; CHECK-NEXT: vmov r3, r12, d5
; CHECK-NEXT: vldrw.u32 q2, [r0, #16]
; CHECK-NEXT: vmov r0, r2, d4
; CHECK-NEXT: vmov r4, r5, d5
; CHECK-NEXT: vstr s0, [r1]
; CHECK-NEXT: vstr s1, [lr]
; CHECK-NEXT: vstr s2, [r3]
; CHECK-NEXT: vstr s3, [r12]
; CHECK-NEXT: vstr s4, [r0]
; CHECK-NEXT: vstr s5, [r2]
; CHECK-NEXT: vstr s6, [r4]
; CHECK-NEXT: vstr s7, [r5]
; CHECK-NEXT: pop {r4, r5, r7, pc}
entry:
%offs = load <8 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v8f32.v8p0(<8 x float> %v, <8 x ptr> %offs, i32 4, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
ret void
}
; i16
; Expand.
define arm_aapcs_vfpcc void @ptr_i16(<8 x i16> %v, ptr %offptr) {
; CHECK-LABEL: ptr_i16:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r6, lr}
; CHECK-NEXT: push {r4, r5, r6, lr}
; CHECK-NEXT: vldrw.u32 q1, [r0]
; CHECK-NEXT: vmov.u16 r6, q0[0]
; CHECK-NEXT: vmov r1, r2, d2
; CHECK-NEXT: vmov r3, r12, d3
; CHECK-NEXT: vldrw.u32 q1, [r0, #16]
; CHECK-NEXT: vmov r0, lr, d2
; CHECK-NEXT: vmov r4, r5, d3
; CHECK-NEXT: strh r6, [r1]
; CHECK-NEXT: vmov.u16 r1, q0[1]
; CHECK-NEXT: strh r1, [r2]
; CHECK-NEXT: vmov.u16 r1, q0[2]
; CHECK-NEXT: strh r1, [r3]
; CHECK-NEXT: vmov.u16 r1, q0[3]
; CHECK-NEXT: strh.w r1, [r12]
; CHECK-NEXT: vmov.u16 r1, q0[4]
; CHECK-NEXT: strh r1, [r0]
; CHECK-NEXT: vmov.u16 r0, q0[5]
; CHECK-NEXT: strh.w r0, [lr]
; CHECK-NEXT: vmov.u16 r0, q0[6]
; CHECK-NEXT: strh r0, [r4]
; CHECK-NEXT: vmov.u16 r0, q0[7]
; CHECK-NEXT: strh r0, [r5]
; CHECK-NEXT: pop {r4, r5, r6, pc}
entry:
%offs = load <8 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> %v, <8 x ptr> %offs, i32 2, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
ret void
}
; Expand
define arm_aapcs_vfpcc void @ptr_v2i16_trunc(<2 x i32> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v2i16_trunc:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vmov r2, s0
; CHECK-NEXT: ldrd r0, r1, [r0]
; CHECK-NEXT: strh r2, [r0]
; CHECK-NEXT: vmov r0, s2
; CHECK-NEXT: strh r0, [r1]
; CHECK-NEXT: bx lr
entry:
%offs = load <2 x ptr>, ptr %offptr, align 4
%ext = trunc <2 x i32> %v to <2 x i16>
call void @llvm.masked.scatter.v2i16.v2p0(<2 x i16> %ext, <2 x ptr> %offs, i32 2, <2 x i1> <i1 true, i1 true>)
ret void
}
define arm_aapcs_vfpcc void @ptr_v4i16_trunc(<4 x i32> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v4i16_trunc:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vldrw.u32 q1, [r0]
; CHECK-NEXT: movs r0, #0
; CHECK-NEXT: vstrh.32 q0, [r0, q1]
; CHECK-NEXT: bx lr
entry:
%offs = load <4 x ptr>, ptr %offptr, align 4
%ext = trunc <4 x i32> %v to <4 x i16>
call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> %ext, <4 x ptr> %offs, i32 2, <4 x i1> <i1 true, i1 true, i1 true, i1 true>)
ret void
}
define arm_aapcs_vfpcc void @ptr_v4i16_dup(i32 %v, <4 x ptr> %offs) {
; CHECK-LABEL: ptr_v4i16_dup:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vdup.32 q1, r0
; CHECK-NEXT: movs r1, #0
; CHECK-NEXT: vmovlb.u16 q1, q1
; CHECK-NEXT: vstrh.32 q1, [r1, q0]
; CHECK-NEXT: bx lr
entry:
%ext = trunc i32 %v to i16
%splatinsert = insertelement <4 x i16> poison, i16 %ext, i32 0
%splat = shufflevector <4 x i16> %splatinsert, <4 x i16> poison, <4 x i32> zeroinitializer
call void @llvm.masked.scatter.v4i16.v4p0(<4 x i16> %splat, <4 x ptr> %offs, i32 2, <4 x i1> <i1 true, i1 true, i1 true, i1 true>)
ret void
}
; Expand
define arm_aapcs_vfpcc void @ptr_v8i16_trunc(<8 x i32> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v8i16_trunc:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r6, r7, lr}
; CHECK-NEXT: push {r4, r5, r6, r7, lr}
; CHECK-NEXT: vldrw.u32 q2, [r0]
; CHECK-NEXT: vmov r3, r4, d0
; CHECK-NEXT: vmov r1, r2, d4
; CHECK-NEXT: vmov lr, r12, d5
; CHECK-NEXT: vldrw.u32 q2, [r0, #16]
; CHECK-NEXT: vmov r0, r5, d1
; CHECK-NEXT: strh r3, [r1]
; CHECK-NEXT: vmov r1, r7, d4
; CHECK-NEXT: strh r4, [r2]
; CHECK-NEXT: vmov r2, r4, d5
; CHECK-NEXT: strh.w r0, [lr]
; CHECK-NEXT: vmov r0, r3, d2
; CHECK-NEXT: strh.w r5, [r12]
; CHECK-NEXT: vmov r5, r6, d3
; CHECK-NEXT: strh r0, [r1]
; CHECK-NEXT: strh r3, [r7]
; CHECK-NEXT: strh r5, [r2]
; CHECK-NEXT: strh r6, [r4]
; CHECK-NEXT: pop {r4, r5, r6, r7, pc}
entry:
%offs = load <8 x ptr>, ptr %offptr, align 4
%ext = trunc <8 x i32> %v to <8 x i16>
call void @llvm.masked.scatter.v8i16.v8p0(<8 x i16> %ext, <8 x ptr> %offs, i32 2, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
ret void
}
; f16
; Expand.
define arm_aapcs_vfpcc void @ptr_f16(<8 x half> %v, ptr %offptr) {
; CHECK-LABEL: ptr_f16:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vldrw.u32 q2, [r0]
; CHECK-NEXT: vldrw.u32 q1, [r0, #16]
; CHECK-NEXT: vmov r0, r1, d4
; CHECK-NEXT: vstr.16 s0, [r0]
; CHECK-NEXT: vmovx.f16 s0, s0
; CHECK-NEXT: vstr.16 s0, [r1]
; CHECK-NEXT: vmov r0, r1, d5
; CHECK-NEXT: vmovx.f16 s0, s1
; CHECK-NEXT: vstr.16 s1, [r0]
; CHECK-NEXT: vstr.16 s0, [r1]
; CHECK-NEXT: vmov r0, r1, d2
; CHECK-NEXT: vmovx.f16 s0, s2
; CHECK-NEXT: vstr.16 s2, [r0]
; CHECK-NEXT: vstr.16 s0, [r1]
; CHECK-NEXT: vmov r0, r1, d3
; CHECK-NEXT: vmovx.f16 s0, s3
; CHECK-NEXT: vstr.16 s3, [r0]
; CHECK-NEXT: vstr.16 s0, [r1]
; CHECK-NEXT: bx lr
entry:
%offs = load <8 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v8f16.v8p0(<8 x half> %v, <8 x ptr> %offs, i32 2, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
ret void
}
define arm_aapcs_vfpcc void @ptr_v4f16(<4 x half> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v4f16:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vldrw.u32 q1, [r0]
; CHECK-NEXT: vmov r0, r1, d2
; CHECK-NEXT: vstr.16 s0, [r0]
; CHECK-NEXT: vmovx.f16 s0, s0
; CHECK-NEXT: vstr.16 s0, [r1]
; CHECK-NEXT: vmov r0, r1, d3
; CHECK-NEXT: vmovx.f16 s0, s1
; CHECK-NEXT: vstr.16 s1, [r0]
; CHECK-NEXT: vstr.16 s0, [r1]
; CHECK-NEXT: bx lr
entry:
%offs = load <4 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v4f16.v4p0(<4 x half> %v, <4 x ptr> %offs, i32 2, <4 x i1> <i1 true, i1 true, i1 true, i1 true>)
ret void
}
define arm_aapcs_vfpcc void @ptr_v4f16_dup(half %v, <4 x ptr> %offs) {
; CHECK-LABEL: ptr_v4f16_dup:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vmov r0, r1, d2
; CHECK-NEXT: vmov r2, r3, d3
; CHECK-NEXT: vstr.16 s0, [r0]
; CHECK-NEXT: vstr.16 s0, [r1]
; CHECK-NEXT: vstr.16 s0, [r2]
; CHECK-NEXT: vstr.16 s0, [r3]
; CHECK-NEXT: bx lr
entry:
%splatinsert = insertelement <4 x half> poison, half %v, i32 0
%splat = shufflevector <4 x half> %splatinsert, <4 x half> poison, <4 x i32> zeroinitializer
call void @llvm.masked.scatter.v4f16.v4p0(<4 x half> %splat, <4 x ptr> %offs, i32 2, <4 x i1> <i1 true, i1 true, i1 true, i1 true>)
ret void
}
; i8
; Expand.
define arm_aapcs_vfpcc void @ptr_i8(<16 x i8> %v, ptr %offptr) {
; CHECK-LABEL: ptr_i8:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r6, r7, r8, r9, lr}
; CHECK-NEXT: push.w {r4, r5, r6, r7, r8, r9, lr}
; CHECK-NEXT: vldrw.u32 q1, [r0]
; CHECK-NEXT: vldrw.u32 q2, [r0, #16]
; CHECK-NEXT: vldrw.u32 q3, [r0, #32]
; CHECK-NEXT: vmov.u8 r6, q0[0]
; CHECK-NEXT: vmov r1, r2, d2
; CHECK-NEXT: vmov.u8 r5, q0[4]
; CHECK-NEXT: vmov r3, r12, d3
; CHECK-NEXT: vldrw.u32 q1, [r0, #48]
; CHECK-NEXT: vmov lr, r4, d4
; CHECK-NEXT: vmov.u8 r7, q0[6]
; CHECK-NEXT: vmov r0, r8, d5
; CHECK-NEXT: strb r6, [r1]
; CHECK-NEXT: vmov.u8 r1, q0[1]
; CHECK-NEXT: strb r1, [r2]
; CHECK-NEXT: vmov.u8 r6, q0[2]
; CHECK-NEXT: vmov r1, r9, d6
; CHECK-NEXT: strb r6, [r3]
; CHECK-NEXT: vmov.u8 r3, q0[3]
; CHECK-NEXT: vmov.u8 r2, q0[8]
; CHECK-NEXT: strb.w r3, [r12]
; CHECK-NEXT: vmov r3, r6, d7
; CHECK-NEXT: strb.w r5, [lr]
; CHECK-NEXT: vmov.u8 r5, q0[5]
; CHECK-NEXT: strb r5, [r4]
; CHECK-NEXT: vmov r5, r4, d2
; CHECK-NEXT: strb r7, [r0]
; CHECK-NEXT: vmov.u8 r0, q0[7]
; CHECK-NEXT: strb.w r0, [r8]
; CHECK-NEXT: vmov r0, r7, d3
; CHECK-NEXT: strb r2, [r1]
; CHECK-NEXT: vmov.u8 r1, q0[9]
; CHECK-NEXT: strb.w r1, [r9]
; CHECK-NEXT: vmov.u8 r1, q0[10]
; CHECK-NEXT: strb r1, [r3]
; CHECK-NEXT: vmov.u8 r1, q0[11]
; CHECK-NEXT: strb r1, [r6]
; CHECK-NEXT: vmov.u8 r1, q0[12]
; CHECK-NEXT: strb r1, [r5]
; CHECK-NEXT: vmov.u8 r1, q0[13]
; CHECK-NEXT: strb r1, [r4]
; CHECK-NEXT: vmov.u8 r1, q0[14]
; CHECK-NEXT: strb r1, [r0]
; CHECK-NEXT: vmov.u8 r0, q0[15]
; CHECK-NEXT: strb r0, [r7]
; CHECK-NEXT: pop.w {r4, r5, r6, r7, r8, r9, pc}
entry:
%offs = load <16 x ptr>, ptr %offptr, align 4
call void @llvm.masked.scatter.v16i8.v16p0(<16 x i8> %v, <16 x ptr> %offs, i32 2, <16 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
ret void
}
; Expand
define arm_aapcs_vfpcc void @ptr_v8i8_trunc16(<8 x i16> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v8i8_trunc16:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r6, lr}
; CHECK-NEXT: push {r4, r5, r6, lr}
; CHECK-NEXT: vldrw.u32 q1, [r0]
; CHECK-NEXT: vmov.u16 r6, q0[0]
; CHECK-NEXT: vmov r1, r2, d2
; CHECK-NEXT: vmov r3, r12, d3
; CHECK-NEXT: vldrw.u32 q1, [r0, #16]
; CHECK-NEXT: vmov r0, lr, d2
; CHECK-NEXT: vmov r4, r5, d3
; CHECK-NEXT: strb r6, [r1]
; CHECK-NEXT: vmov.u16 r1, q0[1]
; CHECK-NEXT: strb r1, [r2]
; CHECK-NEXT: vmov.u16 r1, q0[2]
; CHECK-NEXT: strb r1, [r3]
; CHECK-NEXT: vmov.u16 r1, q0[3]
; CHECK-NEXT: strb.w r1, [r12]
; CHECK-NEXT: vmov.u16 r1, q0[4]
; CHECK-NEXT: strb r1, [r0]
; CHECK-NEXT: vmov.u16 r0, q0[5]
; CHECK-NEXT: strb.w r0, [lr]
; CHECK-NEXT: vmov.u16 r0, q0[6]
; CHECK-NEXT: strb r0, [r4]
; CHECK-NEXT: vmov.u16 r0, q0[7]
; CHECK-NEXT: strb r0, [r5]
; CHECK-NEXT: pop {r4, r5, r6, pc}
entry:
%offs = load <8 x ptr>, ptr %offptr, align 4
%ext = trunc <8 x i16> %v to <8 x i8>
call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> %ext, <8 x ptr> %offs, i32 1, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
ret void
}
define arm_aapcs_vfpcc void @ptr_v4i8_trunc32(<4 x i32> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v4i8_trunc32:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: vldrw.u32 q1, [r0]
; CHECK-NEXT: movs r0, #0
; CHECK-NEXT: vstrb.32 q0, [r0, q1]
; CHECK-NEXT: bx lr
entry:
%offs = load <4 x ptr>, ptr %offptr, align 4
%ext = trunc <4 x i32> %v to <4 x i8>
call void @llvm.masked.scatter.v4i8.v4p0(<4 x i8> %ext, <4 x ptr> %offs, i32 1, <4 x i1> <i1 true, i1 true, i1 true, i1 true>)
ret void
}
; Expand
define arm_aapcs_vfpcc void @ptr_v8i8_trunc32(<8 x i32> %v, ptr %offptr) {
; CHECK-LABEL: ptr_v8i8_trunc32:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: .save {r4, r5, r6, r7, lr}
; CHECK-NEXT: push {r4, r5, r6, r7, lr}
; CHECK-NEXT: vldrw.u32 q2, [r0]
; CHECK-NEXT: vmov r3, r4, d0
; CHECK-NEXT: vmov r1, r2, d4
; CHECK-NEXT: vmov lr, r12, d5
; CHECK-NEXT: vldrw.u32 q2, [r0, #16]
; CHECK-NEXT: vmov r0, r5, d1
; CHECK-NEXT: strb r3, [r1]
; CHECK-NEXT: vmov r1, r7, d4
; CHECK-NEXT: strb r4, [r2]
; CHECK-NEXT: vmov r2, r4, d5
; CHECK-NEXT: strb.w r0, [lr]
; CHECK-NEXT: vmov r0, r3, d2
; CHECK-NEXT: strb.w r5, [r12]
; CHECK-NEXT: vmov r5, r6, d3
; CHECK-NEXT: strb r0, [r1]
; CHECK-NEXT: strb r3, [r7]
; CHECK-NEXT: strb r5, [r2]
; CHECK-NEXT: strb r6, [r4]
; CHECK-NEXT: pop {r4, r5, r6, r7, pc}
entry:
%offs = load <8 x ptr>, ptr %offptr, align 4
%ext = trunc <8 x i32> %v to <8 x i8>
call void @llvm.masked.scatter.v8i8.v8p0(<8 x i8> %ext, <8 x ptr> %offs, i32 1, <8 x i1> <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>)
ret void
}
; loops
define void @foo_ptr_p_int32_t(ptr %dest, ptr %src, i32 %n) {
; CHECK-LABEL: foo_ptr_p_int32_t:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: bic r3, r2, #15
; CHECK-NEXT: cmp r3, #1
; CHECK-NEXT: it lt
; CHECK-NEXT: bxlt lr
; CHECK-NEXT: .LBB19_1: @ %vector.body
; CHECK-NEXT: @ =>This Inner Loop Header: Depth=1
; CHECK-NEXT: vldrw.u32 q0, [r1], #16
; CHECK-NEXT: subs r2, #4
; CHECK-NEXT: vptt.i32 ne, q0, zr
; CHECK-NEXT: vldrwt.u32 q1, [r0], #16
; CHECK-NEXT: vstrwt.32 q1, [q0]
; CHECK-NEXT: bne .LBB19_1
; CHECK-NEXT: @ %bb.2: @ %for.end
; CHECK-NEXT: bx lr
entry:
%and = and i32 %n, -16
%cmp11 = icmp sgt i32 %and, 0
br i1 %cmp11, label %vector.body, label %for.end
vector.body: ; preds = %entry, %vector.body
%index = phi i32 [ 0, %entry ], [ %index.next, %vector.body ]
%0 = getelementptr inbounds ptr, ptr %src, i32 %index
%wide.load = load <4 x ptr>, ptr %0, align 4
%1 = icmp ne <4 x ptr> %wide.load, zeroinitializer
%2 = getelementptr inbounds i32, ptr %dest, i32 %index
%wide.masked.load = call <4 x i32> @llvm.masked.load.v4i32.v4p0(ptr %2, i32 4, <4 x i1> %1, <4 x i32> undef)
call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> %wide.masked.load, <4 x ptr> %wide.load, i32 4, <4 x i1> %1)
%index.next = add i32 %index, 4
%3 = icmp eq i32 %index.next, %n
br i1 %3, label %for.end, label %vector.body
for.end: ; preds = %vector.body, %entry
ret void
}
define void @foo_ptr_p_float(ptr %dest, ptr %src, i32 %n) {
; CHECK-LABEL: foo_ptr_p_float:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: bic r3, r2, #15
; CHECK-NEXT: cmp r3, #1
; CHECK-NEXT: it lt
; CHECK-NEXT: bxlt lr
; CHECK-NEXT: .LBB20_1: @ %vector.body
; CHECK-NEXT: @ =>This Inner Loop Header: Depth=1
; CHECK-NEXT: vldrw.u32 q0, [r1], #16
; CHECK-NEXT: subs r2, #4
; CHECK-NEXT: vptt.i32 ne, q0, zr
; CHECK-NEXT: vldrwt.u32 q1, [r0], #16
; CHECK-NEXT: vstrwt.32 q1, [q0]
; CHECK-NEXT: bne .LBB20_1
; CHECK-NEXT: @ %bb.2: @ %for.end
; CHECK-NEXT: bx lr
entry:
%and = and i32 %n, -16
%cmp11 = icmp sgt i32 %and, 0
br i1 %cmp11, label %vector.body, label %for.end
vector.body: ; preds = %entry, %vector.body
%index = phi i32 [ 0, %entry ], [ %index.next, %vector.body ]
%0 = getelementptr inbounds ptr, ptr %src, i32 %index
%wide.load = load <4 x ptr>, ptr %0, align 4
%1 = icmp ne <4 x ptr> %wide.load, zeroinitializer
%2 = getelementptr inbounds float, ptr %dest, i32 %index
%wide.masked.load = call <4 x i32> @llvm.masked.load.v4i32.v4p0(ptr %2, i32 4, <4 x i1> %1, <4 x i32> undef)
%3 = bitcast <4 x ptr> %wide.load to <4 x ptr>
call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> %wide.masked.load, <4 x ptr> %3, i32 4, <4 x i1> %1)
%index.next = add i32 %index, 4
%4 = icmp eq i32 %index.next, %n
br i1 %4, label %for.end, label %vector.body
for.end: ; preds = %vector.body, %entry
ret void
}
; VLSTW.u32 Qd, [P, 4]
define arm_aapcs_vfpcc void @qi4(<4 x i32> %v, <4 x ptr> %p) {
; CHECK-LABEL: qi4:
; CHECK: @ %bb.0: @ %entry
; CHECK-NEXT: movs r0, #16
; CHECK-NEXT: vadd.i32 q1, q1, r0
; CHECK-NEXT: vstrw.32 q0, [q1]
; CHECK-NEXT: bx lr
entry:
%g = getelementptr inbounds i32, <4 x ptr> %p, i32 4
call void @llvm.masked.scatter.v4i32.v4p0(<4 x i32> %v, <4 x ptr> %g, i32 4, <4 x i1> <i1 true, i1 true, i1 true, i1 true>)
ret void
}
declare void @llvm.masked.scatter.v2i16.v2p0(<2 x i16>, <2 x ptr>, i32, <2 x i1>)
declare void @llvm.masked.scatter.v2i32.v2p0(<2 x i32>, <2 x ptr>, i32, <2 x i1>)
declare void @llvm.masked.scatter.v2f32.v2p0(<2 x float>, <2 x ptr>, i32, <2 x i1>)
declare void @llvm.masked.scatter.v4i8.v4p0(<4 x i8>, <4 x ptr>, i32, <4 x i1>)
declare void @llvm.masked.scatter.v4i16.v4p0(<4 x i16>, <4 x ptr>, i32, <4 x i1>)
declare void @llvm.masked.scatter.v4f16.v4p0(<4 x half>, <4 x ptr>, i32, <4 x i1>)
declare void @llvm.masked.scatter.v4i32.v4p0(<4 x i32>, <4 x ptr>, i32, <4 x i1>)
declare void @llvm.masked.scatter.v4f32.v4p0(<4 x float>, <4 x ptr>, i32, <4 x i1>)
declare void @llvm.masked.scatter.v8i8.v8p0(<8 x i8>, <8 x ptr>, i32, <8 x i1>)
declare void @llvm.masked.scatter.v8i16.v8p0(<8 x i16>, <8 x ptr>, i32, <8 x i1>)
declare void @llvm.masked.scatter.v8f16.v8p0(<8 x half>, <8 x ptr>, i32, <8 x i1>)
declare void @llvm.masked.scatter.v8i32.v8p0(<8 x i32>, <8 x ptr>, i32, <8 x i1>)
declare void @llvm.masked.scatter.v8f32.v8p0(<8 x float>, <8 x ptr>, i32, <8 x i1>)
declare void @llvm.masked.scatter.v16i8.v16p0(<16 x i8>, <16 x ptr>, i32, <16 x i1>)
declare void @llvm.masked.scatter.v16i32.v16p0(<16 x i32>, <16 x ptr>, i32, <16 x i1>)
declare <4 x i32> @llvm.masked.load.v4i32.v4p0(ptr, i32, <4 x i1>, <4 x i32>)
|
