/
SIInstrInfo.h
1339 lines (1035 loc) · 46.9 KB
/
SIInstrInfo.h
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
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
//===- SIInstrInfo.h - SI Instruction Info Interface ------------*- C++ -*-===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
/// \file
/// Interface definition for SIInstrInfo.
//
//===----------------------------------------------------------------------===//
#ifndef LLVM_LIB_TARGET_AMDGPU_SIINSTRINFO_H
#define LLVM_LIB_TARGET_AMDGPU_SIINSTRINFO_H
#include "AMDGPUMIRFormatter.h"
#include "MCTargetDesc/AMDGPUMCTargetDesc.h"
#include "SIRegisterInfo.h"
#include "Utils/AMDGPUBaseInfo.h"
#include "llvm/ADT/SetVector.h"
#include "llvm/CodeGen/TargetInstrInfo.h"
#include "llvm/CodeGen/TargetSchedule.h"
#define GET_INSTRINFO_HEADER
#include "AMDGPUGenInstrInfo.inc"
namespace llvm {
class APInt;
class GCNSubtarget;
class LiveVariables;
class MachineDominatorTree;
class MachineRegisterInfo;
class RegScavenger;
class TargetRegisterClass;
class ScheduleHazardRecognizer;
/// Mark the MMO of a uniform load if there are no potentially clobbering stores
/// on any path from the start of an entry function to this load.
static const MachineMemOperand::Flags MONoClobber =
MachineMemOperand::MOTargetFlag1;
class SIInstrInfo final : public AMDGPUGenInstrInfo {
private:
const SIRegisterInfo RI;
const GCNSubtarget &ST;
TargetSchedModel SchedModel;
mutable std::unique_ptr<AMDGPUMIRFormatter> Formatter;
// The inverse predicate should have the negative value.
enum BranchPredicate {
INVALID_BR = 0,
SCC_TRUE = 1,
SCC_FALSE = -1,
VCCNZ = 2,
VCCZ = -2,
EXECNZ = -3,
EXECZ = 3
};
using SetVectorType = SmallSetVector<MachineInstr *, 32>;
static unsigned getBranchOpcode(BranchPredicate Cond);
static BranchPredicate getBranchPredicate(unsigned Opcode);
public:
unsigned buildExtractSubReg(MachineBasicBlock::iterator MI,
MachineRegisterInfo &MRI,
MachineOperand &SuperReg,
const TargetRegisterClass *SuperRC,
unsigned SubIdx,
const TargetRegisterClass *SubRC) const;
MachineOperand buildExtractSubRegOrImm(MachineBasicBlock::iterator MI,
MachineRegisterInfo &MRI,
MachineOperand &SuperReg,
const TargetRegisterClass *SuperRC,
unsigned SubIdx,
const TargetRegisterClass *SubRC) const;
private:
void swapOperands(MachineInstr &Inst) const;
std::pair<bool, MachineBasicBlock *>
moveScalarAddSub(SetVectorType &Worklist, MachineInstr &Inst,
MachineDominatorTree *MDT = nullptr) const;
void lowerSelect(SetVectorType &Worklist, MachineInstr &Inst,
MachineDominatorTree *MDT = nullptr) const;
void lowerScalarAbs(SetVectorType &Worklist,
MachineInstr &Inst) const;
void lowerScalarXnor(SetVectorType &Worklist,
MachineInstr &Inst) const;
void splitScalarNotBinop(SetVectorType &Worklist,
MachineInstr &Inst,
unsigned Opcode) const;
void splitScalarBinOpN2(SetVectorType &Worklist,
MachineInstr &Inst,
unsigned Opcode) const;
void splitScalar64BitUnaryOp(SetVectorType &Worklist,
MachineInstr &Inst, unsigned Opcode,
bool Swap = false) const;
void splitScalar64BitAddSub(SetVectorType &Worklist, MachineInstr &Inst,
MachineDominatorTree *MDT = nullptr) const;
void splitScalar64BitBinaryOp(SetVectorType &Worklist, MachineInstr &Inst,
unsigned Opcode,
MachineDominatorTree *MDT = nullptr) const;
void splitScalar64BitXnor(SetVectorType &Worklist, MachineInstr &Inst,
MachineDominatorTree *MDT = nullptr) const;
void splitScalar64BitBCNT(SetVectorType &Worklist,
MachineInstr &Inst) const;
void splitScalar64BitBFE(SetVectorType &Worklist,
MachineInstr &Inst) const;
void movePackToVALU(SetVectorType &Worklist,
MachineRegisterInfo &MRI,
MachineInstr &Inst) const;
void addUsersToMoveToVALUWorklist(Register Reg, MachineRegisterInfo &MRI,
SetVectorType &Worklist) const;
void addSCCDefUsersToVALUWorklist(MachineOperand &Op,
MachineInstr &SCCDefInst,
SetVectorType &Worklist,
Register NewCond = Register()) const;
void addSCCDefsToVALUWorklist(MachineInstr *SCCUseInst,
SetVectorType &Worklist) const;
const TargetRegisterClass *
getDestEquivalentVGPRClass(const MachineInstr &Inst) const;
bool checkInstOffsetsDoNotOverlap(const MachineInstr &MIa,
const MachineInstr &MIb) const;
Register findUsedSGPR(const MachineInstr &MI, int OpIndices[3]) const;
protected:
bool swapSourceModifiers(MachineInstr &MI,
MachineOperand &Src0, unsigned Src0OpName,
MachineOperand &Src1, unsigned Src1OpName) const;
MachineInstr *commuteInstructionImpl(MachineInstr &MI, bool NewMI,
unsigned OpIdx0,
unsigned OpIdx1) const override;
public:
enum TargetOperandFlags {
MO_MASK = 0xf,
MO_NONE = 0,
// MO_GOTPCREL -> symbol@GOTPCREL -> R_AMDGPU_GOTPCREL.
MO_GOTPCREL = 1,
// MO_GOTPCREL32_LO -> symbol@gotpcrel32@lo -> R_AMDGPU_GOTPCREL32_LO.
MO_GOTPCREL32 = 2,
MO_GOTPCREL32_LO = 2,
// MO_GOTPCREL32_HI -> symbol@gotpcrel32@hi -> R_AMDGPU_GOTPCREL32_HI.
MO_GOTPCREL32_HI = 3,
// MO_REL32_LO -> symbol@rel32@lo -> R_AMDGPU_REL32_LO.
MO_REL32 = 4,
MO_REL32_LO = 4,
// MO_REL32_HI -> symbol@rel32@hi -> R_AMDGPU_REL32_HI.
MO_REL32_HI = 5,
MO_FAR_BRANCH_OFFSET = 6,
MO_ABS32_LO = 8,
MO_ABS32_HI = 9,
};
explicit SIInstrInfo(const GCNSubtarget &ST);
const SIRegisterInfo &getRegisterInfo() const {
return RI;
}
const GCNSubtarget &getSubtarget() const {
return ST;
}
bool isReallyTriviallyReMaterializable(const MachineInstr &MI) const override;
bool isIgnorableUse(const MachineOperand &MO) const override;
bool areLoadsFromSameBasePtr(SDNode *Load1, SDNode *Load2,
int64_t &Offset1,
int64_t &Offset2) const override;
bool getMemOperandsWithOffsetWidth(
const MachineInstr &LdSt,
SmallVectorImpl<const MachineOperand *> &BaseOps, int64_t &Offset,
bool &OffsetIsScalable, unsigned &Width,
const TargetRegisterInfo *TRI) const final;
bool shouldClusterMemOps(ArrayRef<const MachineOperand *> BaseOps1,
ArrayRef<const MachineOperand *> BaseOps2,
unsigned NumLoads, unsigned NumBytes) const override;
bool shouldScheduleLoadsNear(SDNode *Load0, SDNode *Load1, int64_t Offset0,
int64_t Offset1, unsigned NumLoads) const override;
void copyPhysReg(MachineBasicBlock &MBB, MachineBasicBlock::iterator MI,
const DebugLoc &DL, MCRegister DestReg, MCRegister SrcReg,
bool KillSrc) const override;
void materializeImmediate(MachineBasicBlock &MBB,
MachineBasicBlock::iterator MI, const DebugLoc &DL,
Register DestReg, int64_t Value) const;
const TargetRegisterClass *getPreferredSelectRegClass(
unsigned Size) const;
Register insertNE(MachineBasicBlock *MBB,
MachineBasicBlock::iterator I, const DebugLoc &DL,
Register SrcReg, int Value) const;
Register insertEQ(MachineBasicBlock *MBB,
MachineBasicBlock::iterator I, const DebugLoc &DL,
Register SrcReg, int Value) const;
void storeRegToStackSlot(MachineBasicBlock &MBB,
MachineBasicBlock::iterator MI, Register SrcReg,
bool isKill, int FrameIndex,
const TargetRegisterClass *RC,
const TargetRegisterInfo *TRI,
Register VReg) const override;
void loadRegFromStackSlot(MachineBasicBlock &MBB,
MachineBasicBlock::iterator MI, Register DestReg,
int FrameIndex, const TargetRegisterClass *RC,
const TargetRegisterInfo *TRI,
Register VReg) const override;
bool expandPostRAPseudo(MachineInstr &MI) const override;
// Splits a V_MOV_B64_DPP_PSEUDO opcode into a pair of v_mov_b32_dpp
// instructions. Returns a pair of generated instructions.
// Can split either post-RA with physical registers or pre-RA with
// virtual registers. In latter case IR needs to be in SSA form and
// and a REG_SEQUENCE is produced to define original register.
std::pair<MachineInstr*, MachineInstr*>
expandMovDPP64(MachineInstr &MI) const;
// Returns an opcode that can be used to move a value to a \p DstRC
// register. If there is no hardware instruction that can store to \p
// DstRC, then AMDGPU::COPY is returned.
unsigned getMovOpcode(const TargetRegisterClass *DstRC) const;
const MCInstrDesc &getIndirectRegWriteMovRelPseudo(unsigned VecSize,
unsigned EltSize,
bool IsSGPR) const;
const MCInstrDesc &getIndirectGPRIDXPseudo(unsigned VecSize,
bool IsIndirectSrc) const;
LLVM_READONLY
int commuteOpcode(unsigned Opc) const;
LLVM_READONLY
inline int commuteOpcode(const MachineInstr &MI) const {
return commuteOpcode(MI.getOpcode());
}
bool findCommutedOpIndices(const MachineInstr &MI, unsigned &SrcOpIdx1,
unsigned &SrcOpIdx2) const override;
bool findCommutedOpIndices(MCInstrDesc Desc, unsigned & SrcOpIdx0,
unsigned & SrcOpIdx1) const;
bool isBranchOffsetInRange(unsigned BranchOpc,
int64_t BrOffset) const override;
MachineBasicBlock *getBranchDestBlock(const MachineInstr &MI) const override;
/// Return whether the block terminate with divergent branch.
/// Note this only work before lowering the pseudo control flow instructions.
bool hasDivergentBranch(const MachineBasicBlock *MBB) const;
void insertIndirectBranch(MachineBasicBlock &MBB,
MachineBasicBlock &NewDestBB,
MachineBasicBlock &RestoreBB, const DebugLoc &DL,
int64_t BrOffset, RegScavenger *RS) const override;
bool analyzeBranchImpl(MachineBasicBlock &MBB,
MachineBasicBlock::iterator I,
MachineBasicBlock *&TBB,
MachineBasicBlock *&FBB,
SmallVectorImpl<MachineOperand> &Cond,
bool AllowModify) const;
bool analyzeBranch(MachineBasicBlock &MBB, MachineBasicBlock *&TBB,
MachineBasicBlock *&FBB,
SmallVectorImpl<MachineOperand> &Cond,
bool AllowModify = false) const override;
unsigned removeBranch(MachineBasicBlock &MBB,
int *BytesRemoved = nullptr) const override;
unsigned insertBranch(MachineBasicBlock &MBB, MachineBasicBlock *TBB,
MachineBasicBlock *FBB, ArrayRef<MachineOperand> Cond,
const DebugLoc &DL,
int *BytesAdded = nullptr) const override;
bool reverseBranchCondition(
SmallVectorImpl<MachineOperand> &Cond) const override;
bool canInsertSelect(const MachineBasicBlock &MBB,
ArrayRef<MachineOperand> Cond, Register DstReg,
Register TrueReg, Register FalseReg, int &CondCycles,
int &TrueCycles, int &FalseCycles) const override;
void insertSelect(MachineBasicBlock &MBB,
MachineBasicBlock::iterator I, const DebugLoc &DL,
Register DstReg, ArrayRef<MachineOperand> Cond,
Register TrueReg, Register FalseReg) const override;
void insertVectorSelect(MachineBasicBlock &MBB,
MachineBasicBlock::iterator I, const DebugLoc &DL,
Register DstReg, ArrayRef<MachineOperand> Cond,
Register TrueReg, Register FalseReg) const;
bool analyzeCompare(const MachineInstr &MI, Register &SrcReg,
Register &SrcReg2, int64_t &CmpMask,
int64_t &CmpValue) const override;
bool optimizeCompareInstr(MachineInstr &CmpInstr, Register SrcReg,
Register SrcReg2, int64_t CmpMask, int64_t CmpValue,
const MachineRegisterInfo *MRI) const override;
bool
areMemAccessesTriviallyDisjoint(const MachineInstr &MIa,
const MachineInstr &MIb) const override;
static bool isFoldableCopy(const MachineInstr &MI);
void removeModOperands(MachineInstr &MI) const;
bool FoldImmediate(MachineInstr &UseMI, MachineInstr &DefMI, Register Reg,
MachineRegisterInfo *MRI) const final;
unsigned getMachineCSELookAheadLimit() const override { return 500; }
MachineInstr *convertToThreeAddress(MachineInstr &MI, LiveVariables *LV,
LiveIntervals *LIS) const override;
bool isSchedulingBoundary(const MachineInstr &MI,
const MachineBasicBlock *MBB,
const MachineFunction &MF) const override;
static bool isSALU(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SALU;
}
bool isSALU(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SALU;
}
static bool isVALU(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::VALU;
}
bool isVALU(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::VALU;
}
static bool isVMEM(const MachineInstr &MI) {
return isMUBUF(MI) || isMTBUF(MI) || isMIMG(MI);
}
bool isVMEM(uint16_t Opcode) const {
return isMUBUF(Opcode) || isMTBUF(Opcode) || isMIMG(Opcode);
}
static bool isSOP1(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SOP1;
}
bool isSOP1(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SOP1;
}
static bool isSOP2(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SOP2;
}
bool isSOP2(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SOP2;
}
static bool isSOPC(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SOPC;
}
bool isSOPC(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SOPC;
}
static bool isSOPK(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SOPK;
}
bool isSOPK(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SOPK;
}
static bool isSOPP(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SOPP;
}
bool isSOPP(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SOPP;
}
static bool isPacked(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::IsPacked;
}
bool isPacked(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::IsPacked;
}
static bool isVOP1(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::VOP1;
}
bool isVOP1(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::VOP1;
}
static bool isVOP2(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::VOP2;
}
bool isVOP2(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::VOP2;
}
static bool isVOP3(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::VOP3;
}
bool isVOP3(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::VOP3;
}
static bool isSDWA(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SDWA;
}
bool isSDWA(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SDWA;
}
static bool isVOPC(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::VOPC;
}
bool isVOPC(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::VOPC;
}
static bool isMUBUF(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::MUBUF;
}
bool isMUBUF(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::MUBUF;
}
static bool isMTBUF(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::MTBUF;
}
bool isMTBUF(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::MTBUF;
}
static bool isSMRD(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SMRD;
}
bool isSMRD(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SMRD;
}
bool isBufferSMRD(const MachineInstr &MI) const;
static bool isDS(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::DS;
}
bool isDS(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::DS;
}
bool isAlwaysGDS(uint16_t Opcode) const;
static bool isMIMG(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::MIMG;
}
bool isMIMG(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::MIMG;
}
static bool isGather4(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::Gather4;
}
bool isGather4(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::Gather4;
}
static bool isFLAT(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::FLAT;
}
// Is a FLAT encoded instruction which accesses a specific segment,
// i.e. global_* or scratch_*.
static bool isSegmentSpecificFLAT(const MachineInstr &MI) {
auto Flags = MI.getDesc().TSFlags;
return Flags & (SIInstrFlags::FlatGlobal | SIInstrFlags::FlatScratch);
}
bool isSegmentSpecificFLAT(uint16_t Opcode) const {
auto Flags = get(Opcode).TSFlags;
return Flags & (SIInstrFlags::FlatGlobal | SIInstrFlags::FlatScratch);
}
static bool isFLATGlobal(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::FlatGlobal;
}
bool isFLATGlobal(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::FlatGlobal;
}
static bool isFLATScratch(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::FlatScratch;
}
bool isFLATScratch(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::FlatScratch;
}
// Any FLAT encoded instruction, including global_* and scratch_*.
bool isFLAT(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::FLAT;
}
static bool isEXP(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::EXP;
}
static bool isDualSourceBlendEXP(const MachineInstr &MI) {
if (!isEXP(MI))
return false;
unsigned Target = MI.getOperand(0).getImm();
return Target == AMDGPU::Exp::ET_DUAL_SRC_BLEND0 ||
Target == AMDGPU::Exp::ET_DUAL_SRC_BLEND1;
}
bool isEXP(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::EXP;
}
static bool isAtomicNoRet(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::IsAtomicNoRet;
}
bool isAtomicNoRet(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::IsAtomicNoRet;
}
static bool isAtomicRet(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::IsAtomicRet;
}
bool isAtomicRet(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::IsAtomicRet;
}
static bool isAtomic(const MachineInstr &MI) {
return MI.getDesc().TSFlags & (SIInstrFlags::IsAtomicRet |
SIInstrFlags::IsAtomicNoRet);
}
bool isAtomic(uint16_t Opcode) const {
return get(Opcode).TSFlags & (SIInstrFlags::IsAtomicRet |
SIInstrFlags::IsAtomicNoRet);
}
static bool isWQM(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::WQM;
}
bool isWQM(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::WQM;
}
static bool isDisableWQM(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::DisableWQM;
}
bool isDisableWQM(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::DisableWQM;
}
static bool isVGPRSpill(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::VGPRSpill;
}
bool isVGPRSpill(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::VGPRSpill;
}
static bool isSGPRSpill(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SGPRSpill;
}
bool isSGPRSpill(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SGPRSpill;
}
static bool isDPP(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::DPP;
}
bool isDPP(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::DPP;
}
static bool isTRANS(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::TRANS;
}
bool isTRANS(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::TRANS;
}
static bool isVOP3P(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::VOP3P;
}
bool isVOP3P(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::VOP3P;
}
static bool isVINTRP(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::VINTRP;
}
bool isVINTRP(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::VINTRP;
}
static bool isMAI(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::IsMAI;
}
bool isMAI(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::IsMAI;
}
static bool isMFMA(const MachineInstr &MI) {
return isMAI(MI) && MI.getOpcode() != AMDGPU::V_ACCVGPR_WRITE_B32_e64 &&
MI.getOpcode() != AMDGPU::V_ACCVGPR_READ_B32_e64;
}
static bool isDOT(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::IsDOT;
}
static bool isWMMA(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::IsWMMA;
}
bool isWMMA(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::IsWMMA;
}
bool isDOT(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::IsDOT;
}
static bool isLDSDIR(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::LDSDIR;
}
bool isLDSDIR(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::LDSDIR;
}
static bool isVINTERP(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::VINTERP;
}
bool isVINTERP(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::VINTERP;
}
static bool isScalarUnit(const MachineInstr &MI) {
return MI.getDesc().TSFlags & (SIInstrFlags::SALU | SIInstrFlags::SMRD);
}
static bool usesVM_CNT(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::VM_CNT;
}
static bool usesLGKM_CNT(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::LGKM_CNT;
}
static bool sopkIsZext(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SOPK_ZEXT;
}
bool sopkIsZext(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SOPK_ZEXT;
}
/// \returns true if this is an s_store_dword* instruction. This is more
/// specific than isSMEM && mayStore.
static bool isScalarStore(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::SCALAR_STORE;
}
bool isScalarStore(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::SCALAR_STORE;
}
static bool isFixedSize(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::FIXED_SIZE;
}
bool isFixedSize(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::FIXED_SIZE;
}
static bool hasFPClamp(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::FPClamp;
}
bool hasFPClamp(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::FPClamp;
}
static bool hasIntClamp(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::IntClamp;
}
uint64_t getClampMask(const MachineInstr &MI) const {
const uint64_t ClampFlags = SIInstrFlags::FPClamp |
SIInstrFlags::IntClamp |
SIInstrFlags::ClampLo |
SIInstrFlags::ClampHi;
return MI.getDesc().TSFlags & ClampFlags;
}
static bool usesFPDPRounding(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::FPDPRounding;
}
bool usesFPDPRounding(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::FPDPRounding;
}
static bool isFPAtomic(const MachineInstr &MI) {
return MI.getDesc().TSFlags & SIInstrFlags::FPAtomic;
}
bool isFPAtomic(uint16_t Opcode) const {
return get(Opcode).TSFlags & SIInstrFlags::FPAtomic;
}
bool isVGPRCopy(const MachineInstr &MI) const {
assert(MI.isCopy());
Register Dest = MI.getOperand(0).getReg();
const MachineFunction &MF = *MI.getParent()->getParent();
const MachineRegisterInfo &MRI = MF.getRegInfo();
return !RI.isSGPRReg(MRI, Dest);
}
bool hasVGPRUses(const MachineInstr &MI) const {
const MachineFunction &MF = *MI.getParent()->getParent();
const MachineRegisterInfo &MRI = MF.getRegInfo();
return llvm::any_of(MI.explicit_uses(),
[&MRI, this](const MachineOperand &MO) {
return MO.isReg() && RI.isVGPR(MRI, MO.getReg());});
}
/// Return true if the instruction modifies the mode register.q
static bool modifiesModeRegister(const MachineInstr &MI);
/// Whether we must prevent this instruction from executing with EXEC = 0.
bool hasUnwantedEffectsWhenEXECEmpty(const MachineInstr &MI) const;
/// Returns true if the instruction could potentially depend on the value of
/// exec. If false, exec dependencies may safely be ignored.
bool mayReadEXEC(const MachineRegisterInfo &MRI, const MachineInstr &MI) const;
bool isInlineConstant(const APInt &Imm) const;
bool isInlineConstant(const APFloat &Imm) const {
return isInlineConstant(Imm.bitcastToAPInt());
}
// Returns true if this non-register operand definitely does not need to be
// encoded as a 32-bit literal. Note that this function handles all kinds of
// operands, not just immediates.
//
// Some operands like FrameIndexes could resolve to an inline immediate value
// that will not require an additional 4-bytes; this function assumes that it
// will.
bool isInlineConstant(const MachineOperand &MO, uint8_t OperandType) const;
bool isInlineConstant(const MachineOperand &MO,
const MCOperandInfo &OpInfo) const {
return isInlineConstant(MO, OpInfo.OperandType);
}
/// \p returns true if \p UseMO is substituted with \p DefMO in \p MI it would
/// be an inline immediate.
bool isInlineConstant(const MachineInstr &MI,
const MachineOperand &UseMO,
const MachineOperand &DefMO) const {
assert(UseMO.getParent() == &MI);
int OpIdx = MI.getOperandNo(&UseMO);
if (!MI.getDesc().OpInfo || OpIdx >= MI.getDesc().NumOperands) {
return false;
}
return isInlineConstant(DefMO, MI.getDesc().OpInfo[OpIdx]);
}
/// \p returns true if the operand \p OpIdx in \p MI is a valid inline
/// immediate.
bool isInlineConstant(const MachineInstr &MI, unsigned OpIdx) const {
const MachineOperand &MO = MI.getOperand(OpIdx);
return isInlineConstant(MO, MI.getDesc().OpInfo[OpIdx].OperandType);
}
bool isInlineConstant(const MachineInstr &MI, unsigned OpIdx,
const MachineOperand &MO) const {
if (!MI.getDesc().OpInfo || OpIdx >= MI.getDesc().NumOperands)
return false;
if (MI.isCopy()) {
unsigned Size = getOpSize(MI, OpIdx);
assert(Size == 8 || Size == 4);
uint8_t OpType = (Size == 8) ?
AMDGPU::OPERAND_REG_IMM_INT64 : AMDGPU::OPERAND_REG_IMM_INT32;
return isInlineConstant(MO, OpType);
}
return isInlineConstant(MO, MI.getDesc().OpInfo[OpIdx].OperandType);
}
bool isInlineConstant(const MachineOperand &MO) const {
const MachineInstr *Parent = MO.getParent();
return isInlineConstant(*Parent, Parent->getOperandNo(&MO));
}
bool isImmOperandLegal(const MachineInstr &MI, unsigned OpNo,
const MachineOperand &MO) const;
/// Return true if this 64-bit VALU instruction has a 32-bit encoding.
/// This function will return false if you pass it a 32-bit instruction.
bool hasVALU32BitEncoding(unsigned Opcode) const;
/// Returns true if this operand uses the constant bus.
bool usesConstantBus(const MachineRegisterInfo &MRI,
const MachineOperand &MO,
const MCOperandInfo &OpInfo) const;
/// Return true if this instruction has any modifiers.
/// e.g. src[012]_mod, omod, clamp.
bool hasModifiers(unsigned Opcode) const;
bool hasModifiersSet(const MachineInstr &MI,
unsigned OpName) const;
bool hasAnyModifiersSet(const MachineInstr &MI) const;
bool canShrink(const MachineInstr &MI,
const MachineRegisterInfo &MRI) const;
MachineInstr *buildShrunkInst(MachineInstr &MI,
unsigned NewOpcode) const;
bool verifyInstruction(const MachineInstr &MI,
StringRef &ErrInfo) const override;
unsigned getVALUOp(const MachineInstr &MI) const;
/// Return the correct register class for \p OpNo. For target-specific
/// instructions, this will return the register class that has been defined
/// in tablegen. For generic instructions, like REG_SEQUENCE it will return
/// the register class of its machine operand.
/// to infer the correct register class base on the other operands.
const TargetRegisterClass *getOpRegClass(const MachineInstr &MI,
unsigned OpNo) const;
/// Return the size in bytes of the operand OpNo on the given
// instruction opcode.
unsigned getOpSize(uint16_t Opcode, unsigned OpNo) const {
const MCOperandInfo &OpInfo = get(Opcode).OpInfo[OpNo];
if (OpInfo.RegClass == -1) {
// If this is an immediate operand, this must be a 32-bit literal.
assert(OpInfo.OperandType == MCOI::OPERAND_IMMEDIATE);
return 4;
}
return RI.getRegSizeInBits(*RI.getRegClass(OpInfo.RegClass)) / 8;
}
/// This form should usually be preferred since it handles operands
/// with unknown register classes.
unsigned getOpSize(const MachineInstr &MI, unsigned OpNo) const {
const MachineOperand &MO = MI.getOperand(OpNo);
if (MO.isReg()) {
if (unsigned SubReg = MO.getSubReg()) {
return RI.getSubRegIdxSize(SubReg) / 8;
}
}
return RI.getRegSizeInBits(*getOpRegClass(MI, OpNo)) / 8;
}
/// Legalize the \p OpIndex operand of this instruction by inserting
/// a MOV. For example:
/// ADD_I32_e32 VGPR0, 15
/// to
/// MOV VGPR1, 15
/// ADD_I32_e32 VGPR0, VGPR1
///
/// If the operand being legalized is a register, then a COPY will be used
/// instead of MOV.
void legalizeOpWithMove(MachineInstr &MI, unsigned OpIdx) const;
/// Check if \p MO is a legal operand if it was the \p OpIdx Operand
/// for \p MI.
bool isOperandLegal(const MachineInstr &MI, unsigned OpIdx,
const MachineOperand *MO = nullptr) const;
/// Check if \p MO would be a valid operand for the given operand
/// definition \p OpInfo. Note this does not attempt to validate constant bus
/// restrictions (e.g. literal constant usage).
bool isLegalVSrcOperand(const MachineRegisterInfo &MRI,
const MCOperandInfo &OpInfo,
const MachineOperand &MO) const;
/// Check if \p MO (a register operand) is a legal register for the
/// given operand description.
bool isLegalRegOperand(const MachineRegisterInfo &MRI,
const MCOperandInfo &OpInfo,
const MachineOperand &MO) const;
/// Legalize operands in \p MI by either commuting it or inserting a
/// copy of src1.
void legalizeOperandsVOP2(MachineRegisterInfo &MRI, MachineInstr &MI) const;
/// Fix operands in \p MI to satisfy constant bus requirements.
void legalizeOperandsVOP3(MachineRegisterInfo &MRI, MachineInstr &MI) const;
/// Copy a value from a VGPR (\p SrcReg) to SGPR. This function can only
/// be used when it is know that the value in SrcReg is same across all
/// threads in the wave.
/// \returns The SGPR register that \p SrcReg was copied to.
Register readlaneVGPRToSGPR(Register SrcReg, MachineInstr &UseMI,
MachineRegisterInfo &MRI) const;
void legalizeOperandsSMRD(MachineRegisterInfo &MRI, MachineInstr &MI) const;
void legalizeOperandsFLAT(MachineRegisterInfo &MRI, MachineInstr &MI) const;
void legalizeGenericOperand(MachineBasicBlock &InsertMBB,
MachineBasicBlock::iterator I,
const TargetRegisterClass *DstRC,
MachineOperand &Op, MachineRegisterInfo &MRI,
const DebugLoc &DL) const;
/// Legalize all operands in this instruction. This function may create new
/// instructions and control-flow around \p MI. If present, \p MDT is
/// updated.
/// \returns A new basic block that contains \p MI if new blocks were created.
MachineBasicBlock *
legalizeOperands(MachineInstr &MI, MachineDominatorTree *MDT = nullptr) const;
/// Change SADDR form of a FLAT \p Inst to its VADDR form if saddr operand
/// was moved to VGPR. \returns true if succeeded.
bool moveFlatAddrToVGPR(MachineInstr &Inst) const;
/// Replace this instruction's opcode with the equivalent VALU
/// opcode. This function will also move the users of \p MI to the
/// VALU if necessary. If present, \p MDT is updated.