/
InstructionCombining.cpp
4376 lines (3868 loc) · 173 KB
/
InstructionCombining.cpp
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
//===- InstructionCombining.cpp - Combine multiple instructions -----------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// InstructionCombining - Combine instructions to form fewer, simple
// instructions. This pass does not modify the CFG. This pass is where
// algebraic simplification happens.
//
// This pass combines things like:
// %Y = add i32 %X, 1
// %Z = add i32 %Y, 1
// into:
// %Z = add i32 %X, 2
//
// This is a simple worklist driven algorithm.
//
// This pass guarantees that the following canonicalizations are performed on
// the program:
// 1. If a binary operator has a constant operand, it is moved to the RHS
// 2. Bitwise operators with constant operands are always grouped so that
// shifts are performed first, then or's, then and's, then xor's.
// 3. Compare instructions are converted from <,>,<=,>= to ==,!= if possible
// 4. All cmp instructions on boolean values are replaced with logical ops
// 5. add X, X is represented as (X*2) => (X << 1)
// 6. Multiplies with a power-of-two constant argument are transformed into
// shifts.
// ... etc.
//
//===----------------------------------------------------------------------===//
#include "InstCombineInternal.h"
#include "llvm-c/Initialization.h"
#include "llvm-c/Transforms/InstCombine.h"
#include "llvm/ADT/APInt.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/None.h"
#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/ADT/TinyPtrVector.h"
#include "llvm/Analysis/AliasAnalysis.h"
#include "llvm/Analysis/AssumptionCache.h"
#include "llvm/Analysis/BasicAliasAnalysis.h"
#include "llvm/Analysis/BlockFrequencyInfo.h"
#include "llvm/Analysis/CFG.h"
#include "llvm/Analysis/ConstantFolding.h"
#include "llvm/Analysis/EHPersonalities.h"
#include "llvm/Analysis/GlobalsModRef.h"
#include "llvm/Analysis/InstructionSimplify.h"
#include "llvm/Analysis/LazyBlockFrequencyInfo.h"
#include "llvm/Analysis/LoopInfo.h"
#include "llvm/Analysis/MemoryBuiltins.h"
#include "llvm/Analysis/OptimizationRemarkEmitter.h"
#include "llvm/Analysis/ProfileSummaryInfo.h"
#include "llvm/Analysis/TargetFolder.h"
#include "llvm/Analysis/TargetLibraryInfo.h"
#include "llvm/Analysis/TargetTransformInfo.h"
#include "llvm/Analysis/ValueTracking.h"
#include "llvm/Analysis/VectorUtils.h"
#include "llvm/IR/BasicBlock.h"
#include "llvm/IR/CFG.h"
#include "llvm/IR/Constant.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DIBuilder.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/GetElementPtrTypeIterator.h"
#include "llvm/IR/IRBuilder.h"
#include "llvm/IR/InstrTypes.h"
#include "llvm/IR/Instruction.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/Intrinsics.h"
#include "llvm/IR/LegacyPassManager.h"
#include "llvm/IR/Metadata.h"
#include "llvm/IR/Operator.h"
#include "llvm/IR/PassManager.h"
#include "llvm/IR/PatternMatch.h"
#include "llvm/IR/Type.h"
#include "llvm/IR/Use.h"
#include "llvm/IR/User.h"
#include "llvm/IR/Value.h"
#include "llvm/IR/ValueHandle.h"
#include "llvm/InitializePasses.h"
#include "llvm/Pass.h"
#include "llvm/Support/CBindingWrapping.h"
#include "llvm/Support/Casting.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Compiler.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/DebugCounter.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/KnownBits.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Transforms/InstCombine/InstCombine.h"
#include "llvm/Transforms/Utils/Local.h"
#include <algorithm>
#include <cassert>
#include <cstdint>
#include <memory>
#include <string>
#include <utility>
#define DEBUG_TYPE "instcombine"
#include "llvm/Transforms/Utils/InstructionWorklist.h"
using namespace llvm;
using namespace llvm::PatternMatch;
STATISTIC(NumWorklistIterations,
"Number of instruction combining iterations performed");
STATISTIC(NumCombined , "Number of insts combined");
STATISTIC(NumConstProp, "Number of constant folds");
STATISTIC(NumDeadInst , "Number of dead inst eliminated");
STATISTIC(NumSunkInst , "Number of instructions sunk");
STATISTIC(NumExpand, "Number of expansions");
STATISTIC(NumFactor , "Number of factorizations");
STATISTIC(NumReassoc , "Number of reassociations");
DEBUG_COUNTER(VisitCounter, "instcombine-visit",
"Controls which instructions are visited");
// FIXME: these limits eventually should be as low as 2.
static constexpr unsigned InstCombineDefaultMaxIterations = 1000;
#ifndef NDEBUG
static constexpr unsigned InstCombineDefaultInfiniteLoopThreshold = 100;
#else
static constexpr unsigned InstCombineDefaultInfiniteLoopThreshold = 1000;
#endif
static cl::opt<bool>
EnableCodeSinking("instcombine-code-sinking", cl::desc("Enable code sinking"),
cl::init(true));
static cl::opt<unsigned> LimitMaxIterations(
"instcombine-max-iterations",
cl::desc("Limit the maximum number of instruction combining iterations"),
cl::init(InstCombineDefaultMaxIterations));
static cl::opt<unsigned> InfiniteLoopDetectionThreshold(
"instcombine-infinite-loop-threshold",
cl::desc("Number of instruction combining iterations considered an "
"infinite loop"),
cl::init(InstCombineDefaultInfiniteLoopThreshold), cl::Hidden);
static cl::opt<unsigned>
MaxArraySize("instcombine-maxarray-size", cl::init(1024),
cl::desc("Maximum array size considered when doing a combine"));
// FIXME: Remove this flag when it is no longer necessary to convert
// llvm.dbg.declare to avoid inaccurate debug info. Setting this to false
// increases variable availability at the cost of accuracy. Variables that
// cannot be promoted by mem2reg or SROA will be described as living in memory
// for their entire lifetime. However, passes like DSE and instcombine can
// delete stores to the alloca, leading to misleading and inaccurate debug
// information. This flag can be removed when those passes are fixed.
static cl::opt<unsigned> ShouldLowerDbgDeclare("instcombine-lower-dbg-declare",
cl::Hidden, cl::init(true));
Optional<Instruction *>
InstCombiner::targetInstCombineIntrinsic(IntrinsicInst &II) {
// Handle target specific intrinsics
if (II.getCalledFunction()->isTargetIntrinsic()) {
return TTI.instCombineIntrinsic(*this, II);
}
return None;
}
Optional<Value *> InstCombiner::targetSimplifyDemandedUseBitsIntrinsic(
IntrinsicInst &II, APInt DemandedMask, KnownBits &Known,
bool &KnownBitsComputed) {
// Handle target specific intrinsics
if (II.getCalledFunction()->isTargetIntrinsic()) {
return TTI.simplifyDemandedUseBitsIntrinsic(*this, II, DemandedMask, Known,
KnownBitsComputed);
}
return None;
}
Optional<Value *> InstCombiner::targetSimplifyDemandedVectorEltsIntrinsic(
IntrinsicInst &II, APInt DemandedElts, APInt &UndefElts, APInt &UndefElts2,
APInt &UndefElts3,
std::function<void(Instruction *, unsigned, APInt, APInt &)>
SimplifyAndSetOp) {
// Handle target specific intrinsics
if (II.getCalledFunction()->isTargetIntrinsic()) {
return TTI.simplifyDemandedVectorEltsIntrinsic(
*this, II, DemandedElts, UndefElts, UndefElts2, UndefElts3,
SimplifyAndSetOp);
}
return None;
}
Value *InstCombinerImpl::EmitGEPOffset(User *GEP) {
return llvm::EmitGEPOffset(&Builder, DL, GEP);
}
/// Legal integers and common types are considered desirable. This is used to
/// avoid creating instructions with types that may not be supported well by the
/// the backend.
/// NOTE: This treats i8, i16 and i32 specially because they are common
/// types in frontend languages.
bool InstCombinerImpl::isDesirableIntType(unsigned BitWidth) const {
switch (BitWidth) {
case 8:
case 16:
case 32:
return true;
default:
return DL.isLegalInteger(BitWidth);
}
}
/// Return true if it is desirable to convert an integer computation from a
/// given bit width to a new bit width.
/// We don't want to convert from a legal to an illegal type or from a smaller
/// to a larger illegal type. A width of '1' is always treated as a desirable
/// type because i1 is a fundamental type in IR, and there are many specialized
/// optimizations for i1 types. Common/desirable widths are equally treated as
/// legal to convert to, in order to open up more combining opportunities.
bool InstCombinerImpl::shouldChangeType(unsigned FromWidth,
unsigned ToWidth) const {
bool FromLegal = FromWidth == 1 || DL.isLegalInteger(FromWidth);
bool ToLegal = ToWidth == 1 || DL.isLegalInteger(ToWidth);
// Convert to desirable widths even if they are not legal types.
// Only shrink types, to prevent infinite loops.
if (ToWidth < FromWidth && isDesirableIntType(ToWidth))
return true;
// If this is a legal integer from type, and the result would be an illegal
// type, don't do the transformation.
if (FromLegal && !ToLegal)
return false;
// Otherwise, if both are illegal, do not increase the size of the result. We
// do allow things like i160 -> i64, but not i64 -> i160.
if (!FromLegal && !ToLegal && ToWidth > FromWidth)
return false;
return true;
}
/// Return true if it is desirable to convert a computation from 'From' to 'To'.
/// We don't want to convert from a legal to an illegal type or from a smaller
/// to a larger illegal type. i1 is always treated as a legal type because it is
/// a fundamental type in IR, and there are many specialized optimizations for
/// i1 types.
bool InstCombinerImpl::shouldChangeType(Type *From, Type *To) const {
// TODO: This could be extended to allow vectors. Datalayout changes might be
// needed to properly support that.
if (!From->isIntegerTy() || !To->isIntegerTy())
return false;
unsigned FromWidth = From->getPrimitiveSizeInBits();
unsigned ToWidth = To->getPrimitiveSizeInBits();
return shouldChangeType(FromWidth, ToWidth);
}
// Return true, if No Signed Wrap should be maintained for I.
// The No Signed Wrap flag can be kept if the operation "B (I.getOpcode) C",
// where both B and C should be ConstantInts, results in a constant that does
// not overflow. This function only handles the Add and Sub opcodes. For
// all other opcodes, the function conservatively returns false.
static bool maintainNoSignedWrap(BinaryOperator &I, Value *B, Value *C) {
auto *OBO = dyn_cast<OverflowingBinaryOperator>(&I);
if (!OBO || !OBO->hasNoSignedWrap())
return false;
// We reason about Add and Sub Only.
Instruction::BinaryOps Opcode = I.getOpcode();
if (Opcode != Instruction::Add && Opcode != Instruction::Sub)
return false;
const APInt *BVal, *CVal;
if (!match(B, m_APInt(BVal)) || !match(C, m_APInt(CVal)))
return false;
bool Overflow = false;
if (Opcode == Instruction::Add)
(void)BVal->sadd_ov(*CVal, Overflow);
else
(void)BVal->ssub_ov(*CVal, Overflow);
return !Overflow;
}
static bool hasNoUnsignedWrap(BinaryOperator &I) {
auto *OBO = dyn_cast<OverflowingBinaryOperator>(&I);
return OBO && OBO->hasNoUnsignedWrap();
}
static bool hasNoSignedWrap(BinaryOperator &I) {
auto *OBO = dyn_cast<OverflowingBinaryOperator>(&I);
return OBO && OBO->hasNoSignedWrap();
}
/// Conservatively clears subclassOptionalData after a reassociation or
/// commutation. We preserve fast-math flags when applicable as they can be
/// preserved.
static void ClearSubclassDataAfterReassociation(BinaryOperator &I) {
FPMathOperator *FPMO = dyn_cast<FPMathOperator>(&I);
if (!FPMO) {
I.clearSubclassOptionalData();
return;
}
FastMathFlags FMF = I.getFastMathFlags();
I.clearSubclassOptionalData();
I.setFastMathFlags(FMF);
}
/// Combine constant operands of associative operations either before or after a
/// cast to eliminate one of the associative operations:
/// (op (cast (op X, C2)), C1) --> (cast (op X, op (C1, C2)))
/// (op (cast (op X, C2)), C1) --> (op (cast X), op (C1, C2))
static bool simplifyAssocCastAssoc(BinaryOperator *BinOp1,
InstCombinerImpl &IC) {
auto *Cast = dyn_cast<CastInst>(BinOp1->getOperand(0));
if (!Cast || !Cast->hasOneUse())
return false;
// TODO: Enhance logic for other casts and remove this check.
auto CastOpcode = Cast->getOpcode();
if (CastOpcode != Instruction::ZExt)
return false;
// TODO: Enhance logic for other BinOps and remove this check.
if (!BinOp1->isBitwiseLogicOp())
return false;
auto AssocOpcode = BinOp1->getOpcode();
auto *BinOp2 = dyn_cast<BinaryOperator>(Cast->getOperand(0));
if (!BinOp2 || !BinOp2->hasOneUse() || BinOp2->getOpcode() != AssocOpcode)
return false;
Constant *C1, *C2;
if (!match(BinOp1->getOperand(1), m_Constant(C1)) ||
!match(BinOp2->getOperand(1), m_Constant(C2)))
return false;
// TODO: This assumes a zext cast.
// Eg, if it was a trunc, we'd cast C1 to the source type because casting C2
// to the destination type might lose bits.
// Fold the constants together in the destination type:
// (op (cast (op X, C2)), C1) --> (op (cast X), FoldedC)
Type *DestTy = C1->getType();
Constant *CastC2 = ConstantExpr::getCast(CastOpcode, C2, DestTy);
Constant *FoldedC = ConstantExpr::get(AssocOpcode, C1, CastC2);
IC.replaceOperand(*Cast, 0, BinOp2->getOperand(0));
IC.replaceOperand(*BinOp1, 1, FoldedC);
return true;
}
// Simplifies IntToPtr/PtrToInt RoundTrip Cast To BitCast.
// inttoptr ( ptrtoint (x) ) --> x
Value *InstCombinerImpl::simplifyIntToPtrRoundTripCast(Value *Val) {
auto *IntToPtr = dyn_cast<IntToPtrInst>(Val);
if (IntToPtr && DL.getPointerTypeSizeInBits(IntToPtr->getDestTy()) ==
DL.getTypeSizeInBits(IntToPtr->getSrcTy())) {
auto *PtrToInt = dyn_cast<PtrToIntInst>(IntToPtr->getOperand(0));
Type *CastTy = IntToPtr->getDestTy();
if (PtrToInt &&
CastTy->getPointerAddressSpace() ==
PtrToInt->getSrcTy()->getPointerAddressSpace() &&
DL.getPointerTypeSizeInBits(PtrToInt->getSrcTy()) ==
DL.getTypeSizeInBits(PtrToInt->getDestTy())) {
return CastInst::CreateBitOrPointerCast(PtrToInt->getOperand(0), CastTy,
"", PtrToInt);
}
}
return nullptr;
}
/// This performs a few simplifications for operators that are associative or
/// commutative:
///
/// Commutative operators:
///
/// 1. Order operands such that they are listed from right (least complex) to
/// left (most complex). This puts constants before unary operators before
/// binary operators.
///
/// Associative operators:
///
/// 2. Transform: "(A op B) op C" ==> "A op (B op C)" if "B op C" simplifies.
/// 3. Transform: "A op (B op C)" ==> "(A op B) op C" if "A op B" simplifies.
///
/// Associative and commutative operators:
///
/// 4. Transform: "(A op B) op C" ==> "(C op A) op B" if "C op A" simplifies.
/// 5. Transform: "A op (B op C)" ==> "B op (C op A)" if "C op A" simplifies.
/// 6. Transform: "(A op C1) op (B op C2)" ==> "(A op B) op (C1 op C2)"
/// if C1 and C2 are constants.
bool InstCombinerImpl::SimplifyAssociativeOrCommutative(BinaryOperator &I) {
Instruction::BinaryOps Opcode = I.getOpcode();
bool Changed = false;
do {
// Order operands such that they are listed from right (least complex) to
// left (most complex). This puts constants before unary operators before
// binary operators.
if (I.isCommutative() && getComplexity(I.getOperand(0)) <
getComplexity(I.getOperand(1)))
Changed = !I.swapOperands();
BinaryOperator *Op0 = dyn_cast<BinaryOperator>(I.getOperand(0));
BinaryOperator *Op1 = dyn_cast<BinaryOperator>(I.getOperand(1));
if (I.isAssociative()) {
// Transform: "(A op B) op C" ==> "A op (B op C)" if "B op C" simplifies.
if (Op0 && Op0->getOpcode() == Opcode) {
Value *A = Op0->getOperand(0);
Value *B = Op0->getOperand(1);
Value *C = I.getOperand(1);
// Does "B op C" simplify?
if (Value *V = SimplifyBinOp(Opcode, B, C, SQ.getWithInstruction(&I))) {
// It simplifies to V. Form "A op V".
replaceOperand(I, 0, A);
replaceOperand(I, 1, V);
bool IsNUW = hasNoUnsignedWrap(I) && hasNoUnsignedWrap(*Op0);
bool IsNSW = maintainNoSignedWrap(I, B, C) && hasNoSignedWrap(*Op0);
// Conservatively clear all optional flags since they may not be
// preserved by the reassociation. Reset nsw/nuw based on the above
// analysis.
ClearSubclassDataAfterReassociation(I);
// Note: this is only valid because SimplifyBinOp doesn't look at
// the operands to Op0.
if (IsNUW)
I.setHasNoUnsignedWrap(true);
if (IsNSW)
I.setHasNoSignedWrap(true);
Changed = true;
++NumReassoc;
continue;
}
}
// Transform: "A op (B op C)" ==> "(A op B) op C" if "A op B" simplifies.
if (Op1 && Op1->getOpcode() == Opcode) {
Value *A = I.getOperand(0);
Value *B = Op1->getOperand(0);
Value *C = Op1->getOperand(1);
// Does "A op B" simplify?
if (Value *V = SimplifyBinOp(Opcode, A, B, SQ.getWithInstruction(&I))) {
// It simplifies to V. Form "V op C".
replaceOperand(I, 0, V);
replaceOperand(I, 1, C);
// Conservatively clear the optional flags, since they may not be
// preserved by the reassociation.
ClearSubclassDataAfterReassociation(I);
Changed = true;
++NumReassoc;
continue;
}
}
}
if (I.isAssociative() && I.isCommutative()) {
if (simplifyAssocCastAssoc(&I, *this)) {
Changed = true;
++NumReassoc;
continue;
}
// Transform: "(A op B) op C" ==> "(C op A) op B" if "C op A" simplifies.
if (Op0 && Op0->getOpcode() == Opcode) {
Value *A = Op0->getOperand(0);
Value *B = Op0->getOperand(1);
Value *C = I.getOperand(1);
// Does "C op A" simplify?
if (Value *V = SimplifyBinOp(Opcode, C, A, SQ.getWithInstruction(&I))) {
// It simplifies to V. Form "V op B".
replaceOperand(I, 0, V);
replaceOperand(I, 1, B);
// Conservatively clear the optional flags, since they may not be
// preserved by the reassociation.
ClearSubclassDataAfterReassociation(I);
Changed = true;
++NumReassoc;
continue;
}
}
// Transform: "A op (B op C)" ==> "B op (C op A)" if "C op A" simplifies.
if (Op1 && Op1->getOpcode() == Opcode) {
Value *A = I.getOperand(0);
Value *B = Op1->getOperand(0);
Value *C = Op1->getOperand(1);
// Does "C op A" simplify?
if (Value *V = SimplifyBinOp(Opcode, C, A, SQ.getWithInstruction(&I))) {
// It simplifies to V. Form "B op V".
replaceOperand(I, 0, B);
replaceOperand(I, 1, V);
// Conservatively clear the optional flags, since they may not be
// preserved by the reassociation.
ClearSubclassDataAfterReassociation(I);
Changed = true;
++NumReassoc;
continue;
}
}
// Transform: "(A op C1) op (B op C2)" ==> "(A op B) op (C1 op C2)"
// if C1 and C2 are constants.
Value *A, *B;
Constant *C1, *C2;
if (Op0 && Op1 &&
Op0->getOpcode() == Opcode && Op1->getOpcode() == Opcode &&
match(Op0, m_OneUse(m_BinOp(m_Value(A), m_Constant(C1)))) &&
match(Op1, m_OneUse(m_BinOp(m_Value(B), m_Constant(C2))))) {
bool IsNUW = hasNoUnsignedWrap(I) &&
hasNoUnsignedWrap(*Op0) &&
hasNoUnsignedWrap(*Op1);
BinaryOperator *NewBO = (IsNUW && Opcode == Instruction::Add) ?
BinaryOperator::CreateNUW(Opcode, A, B) :
BinaryOperator::Create(Opcode, A, B);
if (isa<FPMathOperator>(NewBO)) {
FastMathFlags Flags = I.getFastMathFlags();
Flags &= Op0->getFastMathFlags();
Flags &= Op1->getFastMathFlags();
NewBO->setFastMathFlags(Flags);
}
InsertNewInstWith(NewBO, I);
NewBO->takeName(Op1);
replaceOperand(I, 0, NewBO);
replaceOperand(I, 1, ConstantExpr::get(Opcode, C1, C2));
// Conservatively clear the optional flags, since they may not be
// preserved by the reassociation.
ClearSubclassDataAfterReassociation(I);
if (IsNUW)
I.setHasNoUnsignedWrap(true);
Changed = true;
continue;
}
}
// No further simplifications.
return Changed;
} while (true);
}
/// Return whether "X LOp (Y ROp Z)" is always equal to
/// "(X LOp Y) ROp (X LOp Z)".
static bool leftDistributesOverRight(Instruction::BinaryOps LOp,
Instruction::BinaryOps ROp) {
// X & (Y | Z) <--> (X & Y) | (X & Z)
// X & (Y ^ Z) <--> (X & Y) ^ (X & Z)
if (LOp == Instruction::And)
return ROp == Instruction::Or || ROp == Instruction::Xor;
// X | (Y & Z) <--> (X | Y) & (X | Z)
if (LOp == Instruction::Or)
return ROp == Instruction::And;
// X * (Y + Z) <--> (X * Y) + (X * Z)
// X * (Y - Z) <--> (X * Y) - (X * Z)
if (LOp == Instruction::Mul)
return ROp == Instruction::Add || ROp == Instruction::Sub;
return false;
}
/// Return whether "(X LOp Y) ROp Z" is always equal to
/// "(X ROp Z) LOp (Y ROp Z)".
static bool rightDistributesOverLeft(Instruction::BinaryOps LOp,
Instruction::BinaryOps ROp) {
if (Instruction::isCommutative(ROp))
return leftDistributesOverRight(ROp, LOp);
// (X {&|^} Y) >> Z <--> (X >> Z) {&|^} (Y >> Z) for all shifts.
return Instruction::isBitwiseLogicOp(LOp) && Instruction::isShift(ROp);
// TODO: It would be nice to handle division, aka "(X + Y)/Z = X/Z + Y/Z",
// but this requires knowing that the addition does not overflow and other
// such subtleties.
}
/// This function returns identity value for given opcode, which can be used to
/// factor patterns like (X * 2) + X ==> (X * 2) + (X * 1) ==> X * (2 + 1).
static Value *getIdentityValue(Instruction::BinaryOps Opcode, Value *V) {
if (isa<Constant>(V))
return nullptr;
return ConstantExpr::getBinOpIdentity(Opcode, V->getType());
}
/// This function predicates factorization using distributive laws. By default,
/// it just returns the 'Op' inputs. But for special-cases like
/// 'add(shl(X, 5), ...)', this function will have TopOpcode == Instruction::Add
/// and Op = shl(X, 5). The 'shl' is treated as the more general 'mul X, 32' to
/// allow more factorization opportunities.
static Instruction::BinaryOps
getBinOpsForFactorization(Instruction::BinaryOps TopOpcode, BinaryOperator *Op,
Value *&LHS, Value *&RHS) {
assert(Op && "Expected a binary operator");
LHS = Op->getOperand(0);
RHS = Op->getOperand(1);
if (TopOpcode == Instruction::Add || TopOpcode == Instruction::Sub) {
Constant *C;
if (match(Op, m_Shl(m_Value(), m_Constant(C)))) {
// X << C --> X * (1 << C)
RHS = ConstantExpr::getShl(ConstantInt::get(Op->getType(), 1), C);
return Instruction::Mul;
}
// TODO: We can add other conversions e.g. shr => div etc.
}
return Op->getOpcode();
}
/// This tries to simplify binary operations by factorizing out common terms
/// (e. g. "(A*B)+(A*C)" -> "A*(B+C)").
Value *InstCombinerImpl::tryFactorization(BinaryOperator &I,
Instruction::BinaryOps InnerOpcode,
Value *A, Value *B, Value *C,
Value *D) {
assert(A && B && C && D && "All values must be provided");
Value *V = nullptr;
Value *SimplifiedInst = nullptr;
Value *LHS = I.getOperand(0), *RHS = I.getOperand(1);
Instruction::BinaryOps TopLevelOpcode = I.getOpcode();
// Does "X op' Y" always equal "Y op' X"?
bool InnerCommutative = Instruction::isCommutative(InnerOpcode);
// Does "X op' (Y op Z)" always equal "(X op' Y) op (X op' Z)"?
if (leftDistributesOverRight(InnerOpcode, TopLevelOpcode))
// Does the instruction have the form "(A op' B) op (A op' D)" or, in the
// commutative case, "(A op' B) op (C op' A)"?
if (A == C || (InnerCommutative && A == D)) {
if (A != C)
std::swap(C, D);
// Consider forming "A op' (B op D)".
// If "B op D" simplifies then it can be formed with no cost.
V = SimplifyBinOp(TopLevelOpcode, B, D, SQ.getWithInstruction(&I));
// If "B op D" doesn't simplify then only go on if both of the existing
// operations "A op' B" and "C op' D" will be zapped as no longer used.
if (!V && LHS->hasOneUse() && RHS->hasOneUse())
V = Builder.CreateBinOp(TopLevelOpcode, B, D, RHS->getName());
if (V) {
SimplifiedInst = Builder.CreateBinOp(InnerOpcode, A, V);
}
}
// Does "(X op Y) op' Z" always equal "(X op' Z) op (Y op' Z)"?
if (!SimplifiedInst && rightDistributesOverLeft(TopLevelOpcode, InnerOpcode))
// Does the instruction have the form "(A op' B) op (C op' B)" or, in the
// commutative case, "(A op' B) op (B op' D)"?
if (B == D || (InnerCommutative && B == C)) {
if (B != D)
std::swap(C, D);
// Consider forming "(A op C) op' B".
// If "A op C" simplifies then it can be formed with no cost.
V = SimplifyBinOp(TopLevelOpcode, A, C, SQ.getWithInstruction(&I));
// If "A op C" doesn't simplify then only go on if both of the existing
// operations "A op' B" and "C op' D" will be zapped as no longer used.
if (!V && LHS->hasOneUse() && RHS->hasOneUse())
V = Builder.CreateBinOp(TopLevelOpcode, A, C, LHS->getName());
if (V) {
SimplifiedInst = Builder.CreateBinOp(InnerOpcode, V, B);
}
}
if (SimplifiedInst) {
++NumFactor;
SimplifiedInst->takeName(&I);
// Check if we can add NSW/NUW flags to SimplifiedInst. If so, set them.
if (BinaryOperator *BO = dyn_cast<BinaryOperator>(SimplifiedInst)) {
if (isa<OverflowingBinaryOperator>(SimplifiedInst)) {
bool HasNSW = false;
bool HasNUW = false;
if (isa<OverflowingBinaryOperator>(&I)) {
HasNSW = I.hasNoSignedWrap();
HasNUW = I.hasNoUnsignedWrap();
}
if (auto *LOBO = dyn_cast<OverflowingBinaryOperator>(LHS)) {
HasNSW &= LOBO->hasNoSignedWrap();
HasNUW &= LOBO->hasNoUnsignedWrap();
}
if (auto *ROBO = dyn_cast<OverflowingBinaryOperator>(RHS)) {
HasNSW &= ROBO->hasNoSignedWrap();
HasNUW &= ROBO->hasNoUnsignedWrap();
}
if (TopLevelOpcode == Instruction::Add &&
InnerOpcode == Instruction::Mul) {
// We can propagate 'nsw' if we know that
// %Y = mul nsw i16 %X, C
// %Z = add nsw i16 %Y, %X
// =>
// %Z = mul nsw i16 %X, C+1
//
// iff C+1 isn't INT_MIN
const APInt *CInt;
if (match(V, m_APInt(CInt))) {
if (!CInt->isMinSignedValue())
BO->setHasNoSignedWrap(HasNSW);
}
// nuw can be propagated with any constant or nuw value.
BO->setHasNoUnsignedWrap(HasNUW);
}
}
}
}
return SimplifiedInst;
}
/// This tries to simplify binary operations which some other binary operation
/// distributes over either by factorizing out common terms
/// (eg "(A*B)+(A*C)" -> "A*(B+C)") or expanding out if this results in
/// simplifications (eg: "A & (B | C) -> (A&B) | (A&C)" if this is a win).
/// Returns the simplified value, or null if it didn't simplify.
Value *InstCombinerImpl::SimplifyUsingDistributiveLaws(BinaryOperator &I) {
Value *LHS = I.getOperand(0), *RHS = I.getOperand(1);
BinaryOperator *Op0 = dyn_cast<BinaryOperator>(LHS);
BinaryOperator *Op1 = dyn_cast<BinaryOperator>(RHS);
Instruction::BinaryOps TopLevelOpcode = I.getOpcode();
{
// Factorization.
Value *A, *B, *C, *D;
Instruction::BinaryOps LHSOpcode, RHSOpcode;
if (Op0)
LHSOpcode = getBinOpsForFactorization(TopLevelOpcode, Op0, A, B);
if (Op1)
RHSOpcode = getBinOpsForFactorization(TopLevelOpcode, Op1, C, D);
// The instruction has the form "(A op' B) op (C op' D)". Try to factorize
// a common term.
if (Op0 && Op1 && LHSOpcode == RHSOpcode)
if (Value *V = tryFactorization(I, LHSOpcode, A, B, C, D))
return V;
// The instruction has the form "(A op' B) op (C)". Try to factorize common
// term.
if (Op0)
if (Value *Ident = getIdentityValue(LHSOpcode, RHS))
if (Value *V = tryFactorization(I, LHSOpcode, A, B, RHS, Ident))
return V;
// The instruction has the form "(B) op (C op' D)". Try to factorize common
// term.
if (Op1)
if (Value *Ident = getIdentityValue(RHSOpcode, LHS))
if (Value *V = tryFactorization(I, RHSOpcode, LHS, Ident, C, D))
return V;
}
// Expansion.
if (Op0 && rightDistributesOverLeft(Op0->getOpcode(), TopLevelOpcode)) {
// The instruction has the form "(A op' B) op C". See if expanding it out
// to "(A op C) op' (B op C)" results in simplifications.
Value *A = Op0->getOperand(0), *B = Op0->getOperand(1), *C = RHS;
Instruction::BinaryOps InnerOpcode = Op0->getOpcode(); // op'
// Disable the use of undef because it's not safe to distribute undef.
auto SQDistributive = SQ.getWithInstruction(&I).getWithoutUndef();
Value *L = SimplifyBinOp(TopLevelOpcode, A, C, SQDistributive);
Value *R = SimplifyBinOp(TopLevelOpcode, B, C, SQDistributive);
// Do "A op C" and "B op C" both simplify?
if (L && R) {
// They do! Return "L op' R".
++NumExpand;
C = Builder.CreateBinOp(InnerOpcode, L, R);
C->takeName(&I);
return C;
}
// Does "A op C" simplify to the identity value for the inner opcode?
if (L && L == ConstantExpr::getBinOpIdentity(InnerOpcode, L->getType())) {
// They do! Return "B op C".
++NumExpand;
C = Builder.CreateBinOp(TopLevelOpcode, B, C);
C->takeName(&I);
return C;
}
// Does "B op C" simplify to the identity value for the inner opcode?
if (R && R == ConstantExpr::getBinOpIdentity(InnerOpcode, R->getType())) {
// They do! Return "A op C".
++NumExpand;
C = Builder.CreateBinOp(TopLevelOpcode, A, C);
C->takeName(&I);
return C;
}
}
if (Op1 && leftDistributesOverRight(TopLevelOpcode, Op1->getOpcode())) {
// The instruction has the form "A op (B op' C)". See if expanding it out
// to "(A op B) op' (A op C)" results in simplifications.
Value *A = LHS, *B = Op1->getOperand(0), *C = Op1->getOperand(1);
Instruction::BinaryOps InnerOpcode = Op1->getOpcode(); // op'
// Disable the use of undef because it's not safe to distribute undef.
auto SQDistributive = SQ.getWithInstruction(&I).getWithoutUndef();
Value *L = SimplifyBinOp(TopLevelOpcode, A, B, SQDistributive);
Value *R = SimplifyBinOp(TopLevelOpcode, A, C, SQDistributive);
// Do "A op B" and "A op C" both simplify?
if (L && R) {
// They do! Return "L op' R".
++NumExpand;
A = Builder.CreateBinOp(InnerOpcode, L, R);
A->takeName(&I);
return A;
}
// Does "A op B" simplify to the identity value for the inner opcode?
if (L && L == ConstantExpr::getBinOpIdentity(InnerOpcode, L->getType())) {
// They do! Return "A op C".
++NumExpand;
A = Builder.CreateBinOp(TopLevelOpcode, A, C);
A->takeName(&I);
return A;
}
// Does "A op C" simplify to the identity value for the inner opcode?
if (R && R == ConstantExpr::getBinOpIdentity(InnerOpcode, R->getType())) {
// They do! Return "A op B".
++NumExpand;
A = Builder.CreateBinOp(TopLevelOpcode, A, B);
A->takeName(&I);
return A;
}
}
return SimplifySelectsFeedingBinaryOp(I, LHS, RHS);
}
Value *InstCombinerImpl::SimplifySelectsFeedingBinaryOp(BinaryOperator &I,
Value *LHS,
Value *RHS) {
Value *A, *B, *C, *D, *E, *F;
bool LHSIsSelect = match(LHS, m_Select(m_Value(A), m_Value(B), m_Value(C)));
bool RHSIsSelect = match(RHS, m_Select(m_Value(D), m_Value(E), m_Value(F)));
if (!LHSIsSelect && !RHSIsSelect)
return nullptr;
FastMathFlags FMF;
BuilderTy::FastMathFlagGuard Guard(Builder);
if (isa<FPMathOperator>(&I)) {
FMF = I.getFastMathFlags();
Builder.setFastMathFlags(FMF);
}
Instruction::BinaryOps Opcode = I.getOpcode();
SimplifyQuery Q = SQ.getWithInstruction(&I);
Value *Cond, *True = nullptr, *False = nullptr;
if (LHSIsSelect && RHSIsSelect && A == D) {
// (A ? B : C) op (A ? E : F) -> A ? (B op E) : (C op F)
Cond = A;
True = SimplifyBinOp(Opcode, B, E, FMF, Q);
False = SimplifyBinOp(Opcode, C, F, FMF, Q);
if (LHS->hasOneUse() && RHS->hasOneUse()) {
if (False && !True)
True = Builder.CreateBinOp(Opcode, B, E);
else if (True && !False)
False = Builder.CreateBinOp(Opcode, C, F);
}
} else if (LHSIsSelect && LHS->hasOneUse()) {
// (A ? B : C) op Y -> A ? (B op Y) : (C op Y)
Cond = A;
True = SimplifyBinOp(Opcode, B, RHS, FMF, Q);
False = SimplifyBinOp(Opcode, C, RHS, FMF, Q);
} else if (RHSIsSelect && RHS->hasOneUse()) {
// X op (D ? E : F) -> D ? (X op E) : (X op F)
Cond = D;
True = SimplifyBinOp(Opcode, LHS, E, FMF, Q);
False = SimplifyBinOp(Opcode, LHS, F, FMF, Q);
}
if (!True || !False)
return nullptr;
Value *SI = Builder.CreateSelect(Cond, True, False);
SI->takeName(&I);
return SI;
}
/// Freely adapt every user of V as-if V was changed to !V.
/// WARNING: only if canFreelyInvertAllUsersOf() said this can be done.
void InstCombinerImpl::freelyInvertAllUsersOf(Value *I) {
for (User *U : I->users()) {
switch (cast<Instruction>(U)->getOpcode()) {
case Instruction::Select: {
auto *SI = cast<SelectInst>(U);
SI->swapValues();
SI->swapProfMetadata();
break;
}
case Instruction::Br:
cast<BranchInst>(U)->swapSuccessors(); // swaps prof metadata too
break;
case Instruction::Xor:
replaceInstUsesWith(cast<Instruction>(*U), I);
break;
default:
llvm_unreachable("Got unexpected user - out of sync with "
"canFreelyInvertAllUsersOf() ?");
}
}
}
/// Given a 'sub' instruction, return the RHS of the instruction if the LHS is a
/// constant zero (which is the 'negate' form).
Value *InstCombinerImpl::dyn_castNegVal(Value *V) const {
Value *NegV;
if (match(V, m_Neg(m_Value(NegV))))
return NegV;
// Constants can be considered to be negated values if they can be folded.
if (ConstantInt *C = dyn_cast<ConstantInt>(V))
return ConstantExpr::getNeg(C);
if (ConstantDataVector *C = dyn_cast<ConstantDataVector>(V))
if (C->getType()->getElementType()->isIntegerTy())
return ConstantExpr::getNeg(C);
if (ConstantVector *CV = dyn_cast<ConstantVector>(V)) {
for (unsigned i = 0, e = CV->getNumOperands(); i != e; ++i) {
Constant *Elt = CV->getAggregateElement(i);
if (!Elt)
return nullptr;
if (isa<UndefValue>(Elt))
continue;
if (!isa<ConstantInt>(Elt))
return nullptr;
}
return ConstantExpr::getNeg(CV);
}
// Negate integer vector splats.
if (auto *CV = dyn_cast<Constant>(V))
if (CV->getType()->isVectorTy() &&
CV->getType()->getScalarType()->isIntegerTy() && CV->getSplatValue())
return ConstantExpr::getNeg(CV);
return nullptr;
}
/// A binop with a constant operand and a sign-extended boolean operand may be
/// converted into a select of constants by applying the binary operation to
/// the constant with the two possible values of the extended boolean (0 or -1).
Instruction *InstCombinerImpl::foldBinopOfSextBoolToSelect(BinaryOperator &BO) {
// TODO: Handle non-commutative binop (constant is operand 0).
// TODO: Handle zext.
// TODO: Peek through 'not' of cast.
Value *BO0 = BO.getOperand(0);
Value *BO1 = BO.getOperand(1);
Value *X;
Constant *C;
if (!match(BO0, m_SExt(m_Value(X))) || !match(BO1, m_ImmConstant(C)) ||
!X->getType()->isIntOrIntVectorTy(1))
return nullptr;
// bo (sext i1 X), C --> select X, (bo -1, C), (bo 0, C)
Constant *Ones = ConstantInt::getAllOnesValue(BO.getType());
Constant *Zero = ConstantInt::getNullValue(BO.getType());
Constant *TVal = ConstantExpr::get(BO.getOpcode(), Ones, C);
Constant *FVal = ConstantExpr::get(BO.getOpcode(), Zero, C);
return SelectInst::Create(X, TVal, FVal);
}
static Value *foldOperationIntoSelectOperand(Instruction &I, Value *SO,
InstCombiner::BuilderTy &Builder) {
if (auto *Cast = dyn_cast<CastInst>(&I))
return Builder.CreateCast(Cast->getOpcode(), SO, I.getType());
if (auto *II = dyn_cast<IntrinsicInst>(&I)) {
assert(canConstantFoldCallTo(II, cast<Function>(II->getCalledOperand())) &&
"Expected constant-foldable intrinsic");