forked from llvm/llvm-project
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathInstCombineMulDivRem.cpp
1668 lines (1444 loc) · 62.7 KB
/
InstCombineMulDivRem.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
//===- InstCombineMulDivRem.cpp -------------------------------------------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// This file implements the visit functions for mul, fmul, sdiv, udiv, fdiv,
// srem, urem, frem.
//
//===----------------------------------------------------------------------===//
#include "InstCombineInternal.h"
#include "llvm/ADT/APFloat.h"
#include "llvm/ADT/APInt.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/Analysis/InstructionSimplify.h"
#include "llvm/IR/BasicBlock.h"
#include "llvm/IR/Constant.h"
#include "llvm/IR/Constants.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/Operator.h"
#include "llvm/IR/PatternMatch.h"
#include "llvm/IR/Type.h"
#include "llvm/IR/Value.h"
#include "llvm/Support/Casting.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/KnownBits.h"
#include "llvm/Transforms/InstCombine/InstCombiner.h"
#include "llvm/Transforms/Utils/BuildLibCalls.h"
#include <cassert>
#include <cstddef>
#include <cstdint>
#include <utility>
#define DEBUG_TYPE "instcombine"
#include "llvm/Transforms/Utils/InstructionWorklist.h"
using namespace llvm;
using namespace PatternMatch;
/// The specific integer value is used in a context where it is known to be
/// non-zero. If this allows us to simplify the computation, do so and return
/// the new operand, otherwise return null.
static Value *simplifyValueKnownNonZero(Value *V, InstCombinerImpl &IC,
Instruction &CxtI) {
// If V has multiple uses, then we would have to do more analysis to determine
// if this is safe. For example, the use could be in dynamically unreached
// code.
if (!V->hasOneUse()) return nullptr;
bool MadeChange = false;
// ((1 << A) >>u B) --> (1 << (A-B))
// Because V cannot be zero, we know that B is less than A.
Value *A = nullptr, *B = nullptr, *One = nullptr;
if (match(V, m_LShr(m_OneUse(m_Shl(m_Value(One), m_Value(A))), m_Value(B))) &&
match(One, m_One())) {
A = IC.Builder.CreateSub(A, B);
return IC.Builder.CreateShl(One, A);
}
// (PowerOfTwo >>u B) --> isExact since shifting out the result would make it
// inexact. Similarly for <<.
BinaryOperator *I = dyn_cast<BinaryOperator>(V);
if (I && I->isLogicalShift() &&
IC.isKnownToBeAPowerOfTwo(I->getOperand(0), false, 0, &CxtI)) {
// We know that this is an exact/nuw shift and that the input is a
// non-zero context as well.
if (Value *V2 = simplifyValueKnownNonZero(I->getOperand(0), IC, CxtI)) {
IC.replaceOperand(*I, 0, V2);
MadeChange = true;
}
if (I->getOpcode() == Instruction::LShr && !I->isExact()) {
I->setIsExact();
MadeChange = true;
}
if (I->getOpcode() == Instruction::Shl && !I->hasNoUnsignedWrap()) {
I->setHasNoUnsignedWrap();
MadeChange = true;
}
}
// TODO: Lots more we could do here:
// If V is a phi node, we can call this on each of its operands.
// "select cond, X, 0" can simplify to "X".
return MadeChange ? V : nullptr;
}
// TODO: This is a specific form of a much more general pattern.
// We could detect a select with any binop identity constant, or we
// could use SimplifyBinOp to see if either arm of the select reduces.
// But that needs to be done carefully and/or while removing potential
// reverse canonicalizations as in InstCombiner::foldSelectIntoOp().
static Value *foldMulSelectToNegate(BinaryOperator &I,
InstCombiner::BuilderTy &Builder) {
Value *Cond, *OtherOp;
// mul (select Cond, 1, -1), OtherOp --> select Cond, OtherOp, -OtherOp
// mul OtherOp, (select Cond, 1, -1) --> select Cond, OtherOp, -OtherOp
if (match(&I, m_c_Mul(m_OneUse(m_Select(m_Value(Cond), m_One(), m_AllOnes())),
m_Value(OtherOp)))) {
bool HasAnyNoWrap = I.hasNoSignedWrap() || I.hasNoUnsignedWrap();
Value *Neg = Builder.CreateNeg(OtherOp, "", false, HasAnyNoWrap);
return Builder.CreateSelect(Cond, OtherOp, Neg);
}
// mul (select Cond, -1, 1), OtherOp --> select Cond, -OtherOp, OtherOp
// mul OtherOp, (select Cond, -1, 1) --> select Cond, -OtherOp, OtherOp
if (match(&I, m_c_Mul(m_OneUse(m_Select(m_Value(Cond), m_AllOnes(), m_One())),
m_Value(OtherOp)))) {
bool HasAnyNoWrap = I.hasNoSignedWrap() || I.hasNoUnsignedWrap();
Value *Neg = Builder.CreateNeg(OtherOp, "", false, HasAnyNoWrap);
return Builder.CreateSelect(Cond, Neg, OtherOp);
}
// fmul (select Cond, 1.0, -1.0), OtherOp --> select Cond, OtherOp, -OtherOp
// fmul OtherOp, (select Cond, 1.0, -1.0) --> select Cond, OtherOp, -OtherOp
if (match(&I, m_c_FMul(m_OneUse(m_Select(m_Value(Cond), m_SpecificFP(1.0),
m_SpecificFP(-1.0))),
m_Value(OtherOp)))) {
IRBuilder<>::FastMathFlagGuard FMFGuard(Builder);
Builder.setFastMathFlags(I.getFastMathFlags());
return Builder.CreateSelect(Cond, OtherOp, Builder.CreateFNeg(OtherOp));
}
// fmul (select Cond, -1.0, 1.0), OtherOp --> select Cond, -OtherOp, OtherOp
// fmul OtherOp, (select Cond, -1.0, 1.0) --> select Cond, -OtherOp, OtherOp
if (match(&I, m_c_FMul(m_OneUse(m_Select(m_Value(Cond), m_SpecificFP(-1.0),
m_SpecificFP(1.0))),
m_Value(OtherOp)))) {
IRBuilder<>::FastMathFlagGuard FMFGuard(Builder);
Builder.setFastMathFlags(I.getFastMathFlags());
return Builder.CreateSelect(Cond, Builder.CreateFNeg(OtherOp), OtherOp);
}
return nullptr;
}
Instruction *InstCombinerImpl::visitMul(BinaryOperator &I) {
if (Value *V = SimplifyMulInst(I.getOperand(0), I.getOperand(1),
SQ.getWithInstruction(&I)))
return replaceInstUsesWith(I, V);
if (SimplifyAssociativeOrCommutative(I))
return &I;
if (Instruction *X = foldVectorBinop(I))
return X;
if (Instruction *Phi = foldBinopWithPhiOperands(I))
return Phi;
if (Value *V = SimplifyUsingDistributiveLaws(I))
return replaceInstUsesWith(I, V);
Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
unsigned BitWidth = I.getType()->getScalarSizeInBits();
// X * -1 == 0 - X
if (match(Op1, m_AllOnes())) {
BinaryOperator *BO = BinaryOperator::CreateNeg(Op0, I.getName());
if (I.hasNoSignedWrap())
BO->setHasNoSignedWrap();
return BO;
}
// Also allow combining multiply instructions on vectors.
{
Value *NewOp;
Constant *C1, *C2;
const APInt *IVal;
if (match(&I, m_Mul(m_Shl(m_Value(NewOp), m_Constant(C2)),
m_Constant(C1))) &&
match(C1, m_APInt(IVal))) {
// ((X << C2)*C1) == (X * (C1 << C2))
Constant *Shl = ConstantExpr::getShl(C1, C2);
BinaryOperator *Mul = cast<BinaryOperator>(I.getOperand(0));
BinaryOperator *BO = BinaryOperator::CreateMul(NewOp, Shl);
if (I.hasNoUnsignedWrap() && Mul->hasNoUnsignedWrap())
BO->setHasNoUnsignedWrap();
if (I.hasNoSignedWrap() && Mul->hasNoSignedWrap() &&
Shl->isNotMinSignedValue())
BO->setHasNoSignedWrap();
return BO;
}
if (match(&I, m_Mul(m_Value(NewOp), m_Constant(C1)))) {
// Replace X*(2^C) with X << C, where C is either a scalar or a vector.
if (Constant *NewCst = ConstantExpr::getExactLogBase2(C1)) {
BinaryOperator *Shl = BinaryOperator::CreateShl(NewOp, NewCst);
if (I.hasNoUnsignedWrap())
Shl->setHasNoUnsignedWrap();
if (I.hasNoSignedWrap()) {
const APInt *V;
if (match(NewCst, m_APInt(V)) && *V != V->getBitWidth() - 1)
Shl->setHasNoSignedWrap();
}
return Shl;
}
}
}
if (Op0->hasOneUse() && match(Op1, m_NegatedPower2())) {
// Interpret X * (-1<<C) as (-X) * (1<<C) and try to sink the negation.
// The "* (1<<C)" thus becomes a potential shifting opportunity.
if (Value *NegOp0 = Negator::Negate(/*IsNegation*/ true, Op0, *this))
return BinaryOperator::CreateMul(
NegOp0, ConstantExpr::getNeg(cast<Constant>(Op1)), I.getName());
}
if (Instruction *FoldedMul = foldBinOpIntoSelectOrPhi(I))
return FoldedMul;
if (Value *FoldedMul = foldMulSelectToNegate(I, Builder))
return replaceInstUsesWith(I, FoldedMul);
// Simplify mul instructions with a constant RHS.
if (isa<Constant>(Op1)) {
// Canonicalize (X+C1)*CI -> X*CI+C1*CI.
Value *X;
Constant *C1;
if (match(Op0, m_OneUse(m_Add(m_Value(X), m_Constant(C1))))) {
Value *Mul = Builder.CreateMul(C1, Op1);
// Only go forward with the transform if C1*CI simplifies to a tidier
// constant.
if (!match(Mul, m_Mul(m_Value(), m_Value())))
return BinaryOperator::CreateAdd(Builder.CreateMul(X, Op1), Mul);
}
}
// abs(X) * abs(X) -> X * X
// nabs(X) * nabs(X) -> X * X
if (Op0 == Op1) {
Value *X, *Y;
SelectPatternFlavor SPF = matchSelectPattern(Op0, X, Y).Flavor;
if (SPF == SPF_ABS || SPF == SPF_NABS)
return BinaryOperator::CreateMul(X, X);
if (match(Op0, m_Intrinsic<Intrinsic::abs>(m_Value(X))))
return BinaryOperator::CreateMul(X, X);
}
// -X * C --> X * -C
Value *X, *Y;
Constant *Op1C;
if (match(Op0, m_Neg(m_Value(X))) && match(Op1, m_Constant(Op1C)))
return BinaryOperator::CreateMul(X, ConstantExpr::getNeg(Op1C));
// -X * -Y --> X * Y
if (match(Op0, m_Neg(m_Value(X))) && match(Op1, m_Neg(m_Value(Y)))) {
auto *NewMul = BinaryOperator::CreateMul(X, Y);
if (I.hasNoSignedWrap() &&
cast<OverflowingBinaryOperator>(Op0)->hasNoSignedWrap() &&
cast<OverflowingBinaryOperator>(Op1)->hasNoSignedWrap())
NewMul->setHasNoSignedWrap();
return NewMul;
}
// -X * Y --> -(X * Y)
// X * -Y --> -(X * Y)
if (match(&I, m_c_Mul(m_OneUse(m_Neg(m_Value(X))), m_Value(Y))))
return BinaryOperator::CreateNeg(Builder.CreateMul(X, Y));
// (X / Y) * Y = X - (X % Y)
// (X / Y) * -Y = (X % Y) - X
{
Value *Y = Op1;
BinaryOperator *Div = dyn_cast<BinaryOperator>(Op0);
if (!Div || (Div->getOpcode() != Instruction::UDiv &&
Div->getOpcode() != Instruction::SDiv)) {
Y = Op0;
Div = dyn_cast<BinaryOperator>(Op1);
}
Value *Neg = dyn_castNegVal(Y);
if (Div && Div->hasOneUse() &&
(Div->getOperand(1) == Y || Div->getOperand(1) == Neg) &&
(Div->getOpcode() == Instruction::UDiv ||
Div->getOpcode() == Instruction::SDiv)) {
Value *X = Div->getOperand(0), *DivOp1 = Div->getOperand(1);
// If the division is exact, X % Y is zero, so we end up with X or -X.
if (Div->isExact()) {
if (DivOp1 == Y)
return replaceInstUsesWith(I, X);
return BinaryOperator::CreateNeg(X);
}
auto RemOpc = Div->getOpcode() == Instruction::UDiv ? Instruction::URem
: Instruction::SRem;
Value *Rem = Builder.CreateBinOp(RemOpc, X, DivOp1);
if (DivOp1 == Y)
return BinaryOperator::CreateSub(X, Rem);
return BinaryOperator::CreateSub(Rem, X);
}
}
/// i1 mul -> i1 and.
if (I.getType()->isIntOrIntVectorTy(1))
return BinaryOperator::CreateAnd(Op0, Op1);
// X*(1 << Y) --> X << Y
// (1 << Y)*X --> X << Y
{
Value *Y;
BinaryOperator *BO = nullptr;
bool ShlNSW = false;
if (match(Op0, m_Shl(m_One(), m_Value(Y)))) {
BO = BinaryOperator::CreateShl(Op1, Y);
ShlNSW = cast<ShlOperator>(Op0)->hasNoSignedWrap();
} else if (match(Op1, m_Shl(m_One(), m_Value(Y)))) {
BO = BinaryOperator::CreateShl(Op0, Y);
ShlNSW = cast<ShlOperator>(Op1)->hasNoSignedWrap();
}
if (BO) {
if (I.hasNoUnsignedWrap())
BO->setHasNoUnsignedWrap();
if (I.hasNoSignedWrap() && ShlNSW)
BO->setHasNoSignedWrap();
return BO;
}
}
// (zext bool X) * (zext bool Y) --> zext (and X, Y)
// (sext bool X) * (sext bool Y) --> zext (and X, Y)
// Note: -1 * -1 == 1 * 1 == 1 (if the extends match, the result is the same)
if (((match(Op0, m_ZExt(m_Value(X))) && match(Op1, m_ZExt(m_Value(Y)))) ||
(match(Op0, m_SExt(m_Value(X))) && match(Op1, m_SExt(m_Value(Y))))) &&
X->getType()->isIntOrIntVectorTy(1) && X->getType() == Y->getType() &&
(Op0->hasOneUse() || Op1->hasOneUse() || X == Y)) {
Value *And = Builder.CreateAnd(X, Y, "mulbool");
return CastInst::Create(Instruction::ZExt, And, I.getType());
}
// (sext bool X) * (zext bool Y) --> sext (and X, Y)
// (zext bool X) * (sext bool Y) --> sext (and X, Y)
// Note: -1 * 1 == 1 * -1 == -1
if (((match(Op0, m_SExt(m_Value(X))) && match(Op1, m_ZExt(m_Value(Y)))) ||
(match(Op0, m_ZExt(m_Value(X))) && match(Op1, m_SExt(m_Value(Y))))) &&
X->getType()->isIntOrIntVectorTy(1) && X->getType() == Y->getType() &&
(Op0->hasOneUse() || Op1->hasOneUse())) {
Value *And = Builder.CreateAnd(X, Y, "mulbool");
return CastInst::Create(Instruction::SExt, And, I.getType());
}
// (zext bool X) * Y --> X ? Y : 0
// Y * (zext bool X) --> X ? Y : 0
if (match(Op0, m_ZExt(m_Value(X))) && X->getType()->isIntOrIntVectorTy(1))
return SelectInst::Create(X, Op1, ConstantInt::get(I.getType(), 0));
if (match(Op1, m_ZExt(m_Value(X))) && X->getType()->isIntOrIntVectorTy(1))
return SelectInst::Create(X, Op0, ConstantInt::get(I.getType(), 0));
// (sext bool X) * C --> X ? -C : 0
Constant *ImmC;
if (match(Op0, m_SExt(m_Value(X))) && X->getType()->isIntOrIntVectorTy(1) &&
match(Op1, m_ImmConstant(ImmC))) {
Constant *NegC = ConstantExpr::getNeg(ImmC);
return SelectInst::Create(X, NegC, ConstantInt::getNullValue(I.getType()));
}
// (lshr X, 31) * Y --> (ashr X, 31) & Y
// Y * (lshr X, 31) --> (ashr X, 31) & Y
// TODO: We are not checking one-use because the elimination of the multiply
// is better for analysis?
// TODO: Should we canonicalize to '(X < 0) ? Y : 0' instead? That would be
// more similar to what we're doing above.
const APInt *C;
if (match(Op0, m_LShr(m_Value(X), m_APInt(C))) && *C == C->getBitWidth() - 1)
return BinaryOperator::CreateAnd(Builder.CreateAShr(X, *C), Op1);
if (match(Op1, m_LShr(m_Value(X), m_APInt(C))) && *C == C->getBitWidth() - 1)
return BinaryOperator::CreateAnd(Builder.CreateAShr(X, *C), Op0);
// ((ashr X, 31) | 1) * X --> abs(X)
// X * ((ashr X, 31) | 1) --> abs(X)
if (match(&I, m_c_BinOp(m_Or(m_AShr(m_Value(X),
m_SpecificIntAllowUndef(BitWidth - 1)),
m_One()),
m_Deferred(X)))) {
Value *Abs = Builder.CreateBinaryIntrinsic(
Intrinsic::abs, X,
ConstantInt::getBool(I.getContext(), I.hasNoSignedWrap()));
Abs->takeName(&I);
return replaceInstUsesWith(I, Abs);
}
if (Instruction *Ext = narrowMathIfNoOverflow(I))
return Ext;
bool Changed = false;
if (!I.hasNoSignedWrap() && willNotOverflowSignedMul(Op0, Op1, I)) {
Changed = true;
I.setHasNoSignedWrap(true);
}
if (!I.hasNoUnsignedWrap() && willNotOverflowUnsignedMul(Op0, Op1, I)) {
Changed = true;
I.setHasNoUnsignedWrap(true);
}
return Changed ? &I : nullptr;
}
Instruction *InstCombinerImpl::foldFPSignBitOps(BinaryOperator &I) {
BinaryOperator::BinaryOps Opcode = I.getOpcode();
assert((Opcode == Instruction::FMul || Opcode == Instruction::FDiv) &&
"Expected fmul or fdiv");
Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
Value *X, *Y;
// -X * -Y --> X * Y
// -X / -Y --> X / Y
if (match(Op0, m_FNeg(m_Value(X))) && match(Op1, m_FNeg(m_Value(Y))))
return BinaryOperator::CreateWithCopiedFlags(Opcode, X, Y, &I);
// fabs(X) * fabs(X) -> X * X
// fabs(X) / fabs(X) -> X / X
if (Op0 == Op1 && match(Op0, m_FAbs(m_Value(X))))
return BinaryOperator::CreateWithCopiedFlags(Opcode, X, X, &I);
// fabs(X) * fabs(Y) --> fabs(X * Y)
// fabs(X) / fabs(Y) --> fabs(X / Y)
if (match(Op0, m_FAbs(m_Value(X))) && match(Op1, m_FAbs(m_Value(Y))) &&
(Op0->hasOneUse() || Op1->hasOneUse())) {
IRBuilder<>::FastMathFlagGuard FMFGuard(Builder);
Builder.setFastMathFlags(I.getFastMathFlags());
Value *XY = Builder.CreateBinOp(Opcode, X, Y);
Value *Fabs = Builder.CreateUnaryIntrinsic(Intrinsic::fabs, XY);
Fabs->takeName(&I);
return replaceInstUsesWith(I, Fabs);
}
return nullptr;
}
Instruction *InstCombinerImpl::visitFMul(BinaryOperator &I) {
if (Value *V = SimplifyFMulInst(I.getOperand(0), I.getOperand(1),
I.getFastMathFlags(),
SQ.getWithInstruction(&I)))
return replaceInstUsesWith(I, V);
if (SimplifyAssociativeOrCommutative(I))
return &I;
if (Instruction *X = foldVectorBinop(I))
return X;
if (Instruction *Phi = foldBinopWithPhiOperands(I))
return Phi;
if (Instruction *FoldedMul = foldBinOpIntoSelectOrPhi(I))
return FoldedMul;
if (Value *FoldedMul = foldMulSelectToNegate(I, Builder))
return replaceInstUsesWith(I, FoldedMul);
if (Instruction *R = foldFPSignBitOps(I))
return R;
// X * -1.0 --> -X
Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
if (match(Op1, m_SpecificFP(-1.0)))
return UnaryOperator::CreateFNegFMF(Op0, &I);
// -X * C --> X * -C
Value *X, *Y;
Constant *C;
if (match(Op0, m_FNeg(m_Value(X))) && match(Op1, m_Constant(C)))
return BinaryOperator::CreateFMulFMF(X, ConstantExpr::getFNeg(C), &I);
// (select A, B, C) * (select A, D, E) --> select A, (B*D), (C*E)
if (Value *V = SimplifySelectsFeedingBinaryOp(I, Op0, Op1))
return replaceInstUsesWith(I, V);
if (I.hasAllowReassoc()) {
// Reassociate constant RHS with another constant to form constant
// expression.
if (match(Op1, m_Constant(C)) && C->isFiniteNonZeroFP()) {
Constant *C1;
if (match(Op0, m_OneUse(m_FDiv(m_Constant(C1), m_Value(X))))) {
// (C1 / X) * C --> (C * C1) / X
Constant *CC1 = ConstantExpr::getFMul(C, C1);
if (CC1->isNormalFP())
return BinaryOperator::CreateFDivFMF(CC1, X, &I);
}
if (match(Op0, m_FDiv(m_Value(X), m_Constant(C1)))) {
// (X / C1) * C --> X * (C / C1)
Constant *CDivC1 = ConstantExpr::getFDiv(C, C1);
if (CDivC1->isNormalFP())
return BinaryOperator::CreateFMulFMF(X, CDivC1, &I);
// If the constant was a denormal, try reassociating differently.
// (X / C1) * C --> X / (C1 / C)
Constant *C1DivC = ConstantExpr::getFDiv(C1, C);
if (Op0->hasOneUse() && C1DivC->isNormalFP())
return BinaryOperator::CreateFDivFMF(X, C1DivC, &I);
}
// We do not need to match 'fadd C, X' and 'fsub X, C' because they are
// canonicalized to 'fadd X, C'. Distributing the multiply may allow
// further folds and (X * C) + C2 is 'fma'.
if (match(Op0, m_OneUse(m_FAdd(m_Value(X), m_Constant(C1))))) {
// (X + C1) * C --> (X * C) + (C * C1)
Constant *CC1 = ConstantExpr::getFMul(C, C1);
Value *XC = Builder.CreateFMulFMF(X, C, &I);
return BinaryOperator::CreateFAddFMF(XC, CC1, &I);
}
if (match(Op0, m_OneUse(m_FSub(m_Constant(C1), m_Value(X))))) {
// (C1 - X) * C --> (C * C1) - (X * C)
Constant *CC1 = ConstantExpr::getFMul(C, C1);
Value *XC = Builder.CreateFMulFMF(X, C, &I);
return BinaryOperator::CreateFSubFMF(CC1, XC, &I);
}
}
Value *Z;
if (match(&I, m_c_FMul(m_OneUse(m_FDiv(m_Value(X), m_Value(Y))),
m_Value(Z)))) {
// Sink division: (X / Y) * Z --> (X * Z) / Y
Value *NewFMul = Builder.CreateFMulFMF(X, Z, &I);
return BinaryOperator::CreateFDivFMF(NewFMul, Y, &I);
}
// sqrt(X) * sqrt(Y) -> sqrt(X * Y)
// nnan disallows the possibility of returning a number if both operands are
// negative (in that case, we should return NaN).
if (I.hasNoNaNs() &&
match(Op0, m_OneUse(m_Intrinsic<Intrinsic::sqrt>(m_Value(X)))) &&
match(Op1, m_OneUse(m_Intrinsic<Intrinsic::sqrt>(m_Value(Y))))) {
Value *XY = Builder.CreateFMulFMF(X, Y, &I);
Value *Sqrt = Builder.CreateUnaryIntrinsic(Intrinsic::sqrt, XY, &I);
return replaceInstUsesWith(I, Sqrt);
}
// The following transforms are done irrespective of the number of uses
// for the expression "1.0/sqrt(X)".
// 1) 1.0/sqrt(X) * X -> X/sqrt(X)
// 2) X * 1.0/sqrt(X) -> X/sqrt(X)
// We always expect the backend to reduce X/sqrt(X) to sqrt(X), if it
// has the necessary (reassoc) fast-math-flags.
if (I.hasNoSignedZeros() &&
match(Op0, (m_FDiv(m_SpecificFP(1.0), m_Value(Y)))) &&
match(Y, m_Intrinsic<Intrinsic::sqrt>(m_Value(X))) && Op1 == X)
return BinaryOperator::CreateFDivFMF(X, Y, &I);
if (I.hasNoSignedZeros() &&
match(Op1, (m_FDiv(m_SpecificFP(1.0), m_Value(Y)))) &&
match(Y, m_Intrinsic<Intrinsic::sqrt>(m_Value(X))) && Op0 == X)
return BinaryOperator::CreateFDivFMF(X, Y, &I);
// Like the similar transform in instsimplify, this requires 'nsz' because
// sqrt(-0.0) = -0.0, and -0.0 * -0.0 does not simplify to -0.0.
if (I.hasNoNaNs() && I.hasNoSignedZeros() && Op0 == Op1 &&
Op0->hasNUses(2)) {
// Peek through fdiv to find squaring of square root:
// (X / sqrt(Y)) * (X / sqrt(Y)) --> (X * X) / Y
if (match(Op0, m_FDiv(m_Value(X),
m_Intrinsic<Intrinsic::sqrt>(m_Value(Y))))) {
Value *XX = Builder.CreateFMulFMF(X, X, &I);
return BinaryOperator::CreateFDivFMF(XX, Y, &I);
}
// (sqrt(Y) / X) * (sqrt(Y) / X) --> Y / (X * X)
if (match(Op0, m_FDiv(m_Intrinsic<Intrinsic::sqrt>(m_Value(Y)),
m_Value(X)))) {
Value *XX = Builder.CreateFMulFMF(X, X, &I);
return BinaryOperator::CreateFDivFMF(Y, XX, &I);
}
}
if (I.isOnlyUserOfAnyOperand()) {
// pow(x, y) * pow(x, z) -> pow(x, y + z)
if (match(Op0, m_Intrinsic<Intrinsic::pow>(m_Value(X), m_Value(Y))) &&
match(Op1, m_Intrinsic<Intrinsic::pow>(m_Specific(X), m_Value(Z)))) {
auto *YZ = Builder.CreateFAddFMF(Y, Z, &I);
auto *NewPow = Builder.CreateBinaryIntrinsic(Intrinsic::pow, X, YZ, &I);
return replaceInstUsesWith(I, NewPow);
}
// powi(x, y) * powi(x, z) -> powi(x, y + z)
if (match(Op0, m_Intrinsic<Intrinsic::powi>(m_Value(X), m_Value(Y))) &&
match(Op1, m_Intrinsic<Intrinsic::powi>(m_Specific(X), m_Value(Z))) &&
Y->getType() == Z->getType()) {
auto *YZ = Builder.CreateAdd(Y, Z);
auto *NewPow = Builder.CreateIntrinsic(
Intrinsic::powi, {X->getType(), YZ->getType()}, {X, YZ}, &I);
return replaceInstUsesWith(I, NewPow);
}
// exp(X) * exp(Y) -> exp(X + Y)
if (match(Op0, m_Intrinsic<Intrinsic::exp>(m_Value(X))) &&
match(Op1, m_Intrinsic<Intrinsic::exp>(m_Value(Y)))) {
Value *XY = Builder.CreateFAddFMF(X, Y, &I);
Value *Exp = Builder.CreateUnaryIntrinsic(Intrinsic::exp, XY, &I);
return replaceInstUsesWith(I, Exp);
}
// exp2(X) * exp2(Y) -> exp2(X + Y)
if (match(Op0, m_Intrinsic<Intrinsic::exp2>(m_Value(X))) &&
match(Op1, m_Intrinsic<Intrinsic::exp2>(m_Value(Y)))) {
Value *XY = Builder.CreateFAddFMF(X, Y, &I);
Value *Exp2 = Builder.CreateUnaryIntrinsic(Intrinsic::exp2, XY, &I);
return replaceInstUsesWith(I, Exp2);
}
}
// (X*Y) * X => (X*X) * Y where Y != X
// The purpose is two-fold:
// 1) to form a power expression (of X).
// 2) potentially shorten the critical path: After transformation, the
// latency of the instruction Y is amortized by the expression of X*X,
// and therefore Y is in a "less critical" position compared to what it
// was before the transformation.
if (match(Op0, m_OneUse(m_c_FMul(m_Specific(Op1), m_Value(Y)))) &&
Op1 != Y) {
Value *XX = Builder.CreateFMulFMF(Op1, Op1, &I);
return BinaryOperator::CreateFMulFMF(XX, Y, &I);
}
if (match(Op1, m_OneUse(m_c_FMul(m_Specific(Op0), m_Value(Y)))) &&
Op0 != Y) {
Value *XX = Builder.CreateFMulFMF(Op0, Op0, &I);
return BinaryOperator::CreateFMulFMF(XX, Y, &I);
}
}
// log2(X * 0.5) * Y = log2(X) * Y - Y
if (I.isFast()) {
IntrinsicInst *Log2 = nullptr;
if (match(Op0, m_OneUse(m_Intrinsic<Intrinsic::log2>(
m_OneUse(m_FMul(m_Value(X), m_SpecificFP(0.5))))))) {
Log2 = cast<IntrinsicInst>(Op0);
Y = Op1;
}
if (match(Op1, m_OneUse(m_Intrinsic<Intrinsic::log2>(
m_OneUse(m_FMul(m_Value(X), m_SpecificFP(0.5))))))) {
Log2 = cast<IntrinsicInst>(Op1);
Y = Op0;
}
if (Log2) {
Value *Log2 = Builder.CreateUnaryIntrinsic(Intrinsic::log2, X, &I);
Value *LogXTimesY = Builder.CreateFMulFMF(Log2, Y, &I);
return BinaryOperator::CreateFSubFMF(LogXTimesY, Y, &I);
}
}
return nullptr;
}
/// Fold a divide or remainder with a select instruction divisor when one of the
/// select operands is zero. In that case, we can use the other select operand
/// because div/rem by zero is undefined.
bool InstCombinerImpl::simplifyDivRemOfSelectWithZeroOp(BinaryOperator &I) {
SelectInst *SI = dyn_cast<SelectInst>(I.getOperand(1));
if (!SI)
return false;
int NonNullOperand;
if (match(SI->getTrueValue(), m_Zero()))
// div/rem X, (Cond ? 0 : Y) -> div/rem X, Y
NonNullOperand = 2;
else if (match(SI->getFalseValue(), m_Zero()))
// div/rem X, (Cond ? Y : 0) -> div/rem X, Y
NonNullOperand = 1;
else
return false;
// Change the div/rem to use 'Y' instead of the select.
replaceOperand(I, 1, SI->getOperand(NonNullOperand));
// Okay, we know we replace the operand of the div/rem with 'Y' with no
// problem. However, the select, or the condition of the select may have
// multiple uses. Based on our knowledge that the operand must be non-zero,
// propagate the known value for the select into other uses of it, and
// propagate a known value of the condition into its other users.
// If the select and condition only have a single use, don't bother with this,
// early exit.
Value *SelectCond = SI->getCondition();
if (SI->use_empty() && SelectCond->hasOneUse())
return true;
// Scan the current block backward, looking for other uses of SI.
BasicBlock::iterator BBI = I.getIterator(), BBFront = I.getParent()->begin();
Type *CondTy = SelectCond->getType();
while (BBI != BBFront) {
--BBI;
// If we found an instruction that we can't assume will return, so
// information from below it cannot be propagated above it.
if (!isGuaranteedToTransferExecutionToSuccessor(&*BBI))
break;
// Replace uses of the select or its condition with the known values.
for (Use &Op : BBI->operands()) {
if (Op == SI) {
replaceUse(Op, SI->getOperand(NonNullOperand));
Worklist.push(&*BBI);
} else if (Op == SelectCond) {
replaceUse(Op, NonNullOperand == 1 ? ConstantInt::getTrue(CondTy)
: ConstantInt::getFalse(CondTy));
Worklist.push(&*BBI);
}
}
// If we past the instruction, quit looking for it.
if (&*BBI == SI)
SI = nullptr;
if (&*BBI == SelectCond)
SelectCond = nullptr;
// If we ran out of things to eliminate, break out of the loop.
if (!SelectCond && !SI)
break;
}
return true;
}
/// True if the multiply can not be expressed in an int this size.
static bool multiplyOverflows(const APInt &C1, const APInt &C2, APInt &Product,
bool IsSigned) {
bool Overflow;
Product = IsSigned ? C1.smul_ov(C2, Overflow) : C1.umul_ov(C2, Overflow);
return Overflow;
}
/// True if C1 is a multiple of C2. Quotient contains C1/C2.
static bool isMultiple(const APInt &C1, const APInt &C2, APInt &Quotient,
bool IsSigned) {
assert(C1.getBitWidth() == C2.getBitWidth() && "Constant widths not equal");
// Bail if we will divide by zero.
if (C2.isZero())
return false;
// Bail if we would divide INT_MIN by -1.
if (IsSigned && C1.isMinSignedValue() && C2.isAllOnes())
return false;
APInt Remainder(C1.getBitWidth(), /*val=*/0ULL, IsSigned);
if (IsSigned)
APInt::sdivrem(C1, C2, Quotient, Remainder);
else
APInt::udivrem(C1, C2, Quotient, Remainder);
return Remainder.isMinValue();
}
/// This function implements the transforms common to both integer division
/// instructions (udiv and sdiv). It is called by the visitors to those integer
/// division instructions.
/// Common integer divide transforms
Instruction *InstCombinerImpl::commonIDivTransforms(BinaryOperator &I) {
if (Instruction *Phi = foldBinopWithPhiOperands(I))
return Phi;
Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
bool IsSigned = I.getOpcode() == Instruction::SDiv;
Type *Ty = I.getType();
// The RHS is known non-zero.
if (Value *V = simplifyValueKnownNonZero(I.getOperand(1), *this, I))
return replaceOperand(I, 1, V);
// Handle cases involving: [su]div X, (select Cond, Y, Z)
// This does not apply for fdiv.
if (simplifyDivRemOfSelectWithZeroOp(I))
return &I;
// If the divisor is a select-of-constants, try to constant fold all div ops:
// C / (select Cond, TrueC, FalseC) --> select Cond, (C / TrueC), (C / FalseC)
// TODO: Adapt simplifyDivRemOfSelectWithZeroOp to allow this and other folds.
if (match(Op0, m_ImmConstant()) &&
match(Op1, m_Select(m_Value(), m_ImmConstant(), m_ImmConstant()))) {
if (Instruction *R = FoldOpIntoSelect(I, cast<SelectInst>(Op1)))
return R;
}
const APInt *C2;
if (match(Op1, m_APInt(C2))) {
Value *X;
const APInt *C1;
// (X / C1) / C2 -> X / (C1*C2)
if ((IsSigned && match(Op0, m_SDiv(m_Value(X), m_APInt(C1)))) ||
(!IsSigned && match(Op0, m_UDiv(m_Value(X), m_APInt(C1))))) {
APInt Product(C1->getBitWidth(), /*val=*/0ULL, IsSigned);
if (!multiplyOverflows(*C1, *C2, Product, IsSigned))
return BinaryOperator::Create(I.getOpcode(), X,
ConstantInt::get(Ty, Product));
}
if ((IsSigned && match(Op0, m_NSWMul(m_Value(X), m_APInt(C1)))) ||
(!IsSigned && match(Op0, m_NUWMul(m_Value(X), m_APInt(C1))))) {
APInt Quotient(C1->getBitWidth(), /*val=*/0ULL, IsSigned);
// (X * C1) / C2 -> X / (C2 / C1) if C2 is a multiple of C1.
if (isMultiple(*C2, *C1, Quotient, IsSigned)) {
auto *NewDiv = BinaryOperator::Create(I.getOpcode(), X,
ConstantInt::get(Ty, Quotient));
NewDiv->setIsExact(I.isExact());
return NewDiv;
}
// (X * C1) / C2 -> X * (C1 / C2) if C1 is a multiple of C2.
if (isMultiple(*C1, *C2, Quotient, IsSigned)) {
auto *Mul = BinaryOperator::Create(Instruction::Mul, X,
ConstantInt::get(Ty, Quotient));
auto *OBO = cast<OverflowingBinaryOperator>(Op0);
Mul->setHasNoUnsignedWrap(!IsSigned && OBO->hasNoUnsignedWrap());
Mul->setHasNoSignedWrap(OBO->hasNoSignedWrap());
return Mul;
}
}
if ((IsSigned && match(Op0, m_NSWShl(m_Value(X), m_APInt(C1))) &&
C1->ult(C1->getBitWidth() - 1)) ||
(!IsSigned && match(Op0, m_NUWShl(m_Value(X), m_APInt(C1))) &&
C1->ult(C1->getBitWidth()))) {
APInt Quotient(C1->getBitWidth(), /*val=*/0ULL, IsSigned);
APInt C1Shifted = APInt::getOneBitSet(
C1->getBitWidth(), static_cast<unsigned>(C1->getZExtValue()));
// (X << C1) / C2 -> X / (C2 >> C1) if C2 is a multiple of 1 << C1.
if (isMultiple(*C2, C1Shifted, Quotient, IsSigned)) {
auto *BO = BinaryOperator::Create(I.getOpcode(), X,
ConstantInt::get(Ty, Quotient));
BO->setIsExact(I.isExact());
return BO;
}
// (X << C1) / C2 -> X * ((1 << C1) / C2) if 1 << C1 is a multiple of C2.
if (isMultiple(C1Shifted, *C2, Quotient, IsSigned)) {
auto *Mul = BinaryOperator::Create(Instruction::Mul, X,
ConstantInt::get(Ty, Quotient));
auto *OBO = cast<OverflowingBinaryOperator>(Op0);
Mul->setHasNoUnsignedWrap(!IsSigned && OBO->hasNoUnsignedWrap());
Mul->setHasNoSignedWrap(OBO->hasNoSignedWrap());
return Mul;
}
}
if (!C2->isZero()) // avoid X udiv 0
if (Instruction *FoldedDiv = foldBinOpIntoSelectOrPhi(I))
return FoldedDiv;
}
if (match(Op0, m_One())) {
assert(!Ty->isIntOrIntVectorTy(1) && "i1 divide not removed?");
if (IsSigned) {
// If Op1 is 0 then it's undefined behaviour, if Op1 is 1 then the
// result is one, if Op1 is -1 then the result is minus one, otherwise
// it's zero.
Value *Inc = Builder.CreateAdd(Op1, Op0);
Value *Cmp = Builder.CreateICmpULT(Inc, ConstantInt::get(Ty, 3));
return SelectInst::Create(Cmp, Op1, ConstantInt::get(Ty, 0));
} else {
// If Op1 is 0 then it's undefined behaviour. If Op1 is 1 then the
// result is one, otherwise it's zero.
return new ZExtInst(Builder.CreateICmpEQ(Op1, Op0), Ty);
}
}
// See if we can fold away this div instruction.
if (SimplifyDemandedInstructionBits(I))
return &I;
// (X - (X rem Y)) / Y -> X / Y; usually originates as ((X / Y) * Y) / Y
Value *X, *Z;
if (match(Op0, m_Sub(m_Value(X), m_Value(Z)))) // (X - Z) / Y; Y = Op1
if ((IsSigned && match(Z, m_SRem(m_Specific(X), m_Specific(Op1)))) ||
(!IsSigned && match(Z, m_URem(m_Specific(X), m_Specific(Op1)))))
return BinaryOperator::Create(I.getOpcode(), X, Op1);
// (X << Y) / X -> 1 << Y
Value *Y;
if (IsSigned && match(Op0, m_NSWShl(m_Specific(Op1), m_Value(Y))))
return BinaryOperator::CreateNSWShl(ConstantInt::get(Ty, 1), Y);
if (!IsSigned && match(Op0, m_NUWShl(m_Specific(Op1), m_Value(Y))))
return BinaryOperator::CreateNUWShl(ConstantInt::get(Ty, 1), Y);
// X / (X * Y) -> 1 / Y if the multiplication does not overflow.
if (match(Op1, m_c_Mul(m_Specific(Op0), m_Value(Y)))) {
bool HasNSW = cast<OverflowingBinaryOperator>(Op1)->hasNoSignedWrap();
bool HasNUW = cast<OverflowingBinaryOperator>(Op1)->hasNoUnsignedWrap();
if ((IsSigned && HasNSW) || (!IsSigned && HasNUW)) {
replaceOperand(I, 0, ConstantInt::get(Ty, 1));
replaceOperand(I, 1, Y);
return &I;
}
}
return nullptr;
}
static const unsigned MaxDepth = 6;
namespace {
using FoldUDivOperandCb = Instruction *(*)(Value *Op0, Value *Op1,
const BinaryOperator &I,
InstCombinerImpl &IC);
/// Used to maintain state for visitUDivOperand().
struct UDivFoldAction {
/// Informs visitUDiv() how to fold this operand. This can be zero if this
/// action joins two actions together.
FoldUDivOperandCb FoldAction;
/// Which operand to fold.
Value *OperandToFold;
union {
/// The instruction returned when FoldAction is invoked.
Instruction *FoldResult;
/// Stores the LHS action index if this action joins two actions together.
size_t SelectLHSIdx;
};
UDivFoldAction(FoldUDivOperandCb FA, Value *InputOperand)
: FoldAction(FA), OperandToFold(InputOperand), FoldResult(nullptr) {}
UDivFoldAction(FoldUDivOperandCb FA, Value *InputOperand, size_t SLHS)
: FoldAction(FA), OperandToFold(InputOperand), SelectLHSIdx(SLHS) {}
};
} // end anonymous namespace
// X udiv 2^C -> X >> C
static Instruction *foldUDivPow2Cst(Value *Op0, Value *Op1,
const BinaryOperator &I,
InstCombinerImpl &IC) {
Constant *C1 = ConstantExpr::getExactLogBase2(cast<Constant>(Op1));
if (!C1)
llvm_unreachable("Failed to constant fold udiv -> logbase2");
BinaryOperator *LShr = BinaryOperator::CreateLShr(Op0, C1);
if (I.isExact())
LShr->setIsExact();
return LShr;
}
// X udiv (C1 << N), where C1 is "1<<C2" --> X >> (N+C2)
// X udiv (zext (C1 << N)), where C1 is "1<<C2" --> X >> (N+C2)
static Instruction *foldUDivShl(Value *Op0, Value *Op1, const BinaryOperator &I,
InstCombinerImpl &IC) {
Value *ShiftLeft;
if (!match(Op1, m_ZExt(m_Value(ShiftLeft))))
ShiftLeft = Op1;
Constant *CI;
Value *N;
if (!match(ShiftLeft, m_Shl(m_Constant(CI), m_Value(N))))
llvm_unreachable("match should never fail here!");
Constant *Log2Base = ConstantExpr::getExactLogBase2(CI);
if (!Log2Base)
llvm_unreachable("getLogBase2 should never fail here!");
N = IC.Builder.CreateAdd(N, Log2Base);
if (Op1 != ShiftLeft)
N = IC.Builder.CreateZExt(N, Op1->getType());
BinaryOperator *LShr = BinaryOperator::CreateLShr(Op0, N);
if (I.isExact())
LShr->setIsExact();
return LShr;
}
// Recursively visits the possible right hand operands of a udiv
// instruction, seeing through select instructions, to determine if we can
// replace the udiv with something simpler. If we find that an operand is not
// able to simplify the udiv, we abort the entire transformation.
static size_t visitUDivOperand(Value *Op0, Value *Op1, const BinaryOperator &I,
SmallVectorImpl<UDivFoldAction> &Actions,
unsigned Depth = 0) {
// FIXME: assert that Op1 isn't/doesn't contain undef.
// Check to see if this is an unsigned division with an exact power of 2,
// if so, convert to a right shift.
if (match(Op1, m_Power2())) {
Actions.push_back(UDivFoldAction(foldUDivPow2Cst, Op1));
return Actions.size();
}
// X udiv (C1 << N), where C1 is "1<<C2" --> X >> (N+C2)
if (match(Op1, m_Shl(m_Power2(), m_Value())) ||
match(Op1, m_ZExt(m_Shl(m_Power2(), m_Value())))) {
Actions.push_back(UDivFoldAction(foldUDivShl, Op1));
return Actions.size();
}
// The remaining tests are all recursive, so bail out if we hit the limit.
if (Depth++ == MaxDepth)
return 0;
if (SelectInst *SI = dyn_cast<SelectInst>(Op1))
// FIXME: missed optimization: if one of the hands of select is/contains
// undef, just directly pick the other one.