LLVM  8.0.0svn
InstCombineAndOrXor.cpp
Go to the documentation of this file.
1 //===- InstCombineAndOrXor.cpp --------------------------------------------===//
2 //
3 // The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 // This file implements the visitAnd, visitOr, and visitXor functions.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "InstCombineInternal.h"
18 #include "llvm/IR/ConstantRange.h"
19 #include "llvm/IR/Intrinsics.h"
20 #include "llvm/IR/PatternMatch.h"
21 using namespace llvm;
22 using namespace PatternMatch;
23 
24 #define DEBUG_TYPE "instcombine"
25 
26 /// Similar to getICmpCode but for FCmpInst. This encodes a fcmp predicate into
27 /// a four bit mask.
28 static unsigned getFCmpCode(FCmpInst::Predicate CC) {
30  "Unexpected FCmp predicate!");
31  // Take advantage of the bit pattern of FCmpInst::Predicate here.
32  // U L G E
33  static_assert(FCmpInst::FCMP_FALSE == 0, ""); // 0 0 0 0
34  static_assert(FCmpInst::FCMP_OEQ == 1, ""); // 0 0 0 1
35  static_assert(FCmpInst::FCMP_OGT == 2, ""); // 0 0 1 0
36  static_assert(FCmpInst::FCMP_OGE == 3, ""); // 0 0 1 1
37  static_assert(FCmpInst::FCMP_OLT == 4, ""); // 0 1 0 0
38  static_assert(FCmpInst::FCMP_OLE == 5, ""); // 0 1 0 1
39  static_assert(FCmpInst::FCMP_ONE == 6, ""); // 0 1 1 0
40  static_assert(FCmpInst::FCMP_ORD == 7, ""); // 0 1 1 1
41  static_assert(FCmpInst::FCMP_UNO == 8, ""); // 1 0 0 0
42  static_assert(FCmpInst::FCMP_UEQ == 9, ""); // 1 0 0 1
43  static_assert(FCmpInst::FCMP_UGT == 10, ""); // 1 0 1 0
44  static_assert(FCmpInst::FCMP_UGE == 11, ""); // 1 0 1 1
45  static_assert(FCmpInst::FCMP_ULT == 12, ""); // 1 1 0 0
46  static_assert(FCmpInst::FCMP_ULE == 13, ""); // 1 1 0 1
47  static_assert(FCmpInst::FCMP_UNE == 14, ""); // 1 1 1 0
48  static_assert(FCmpInst::FCMP_TRUE == 15, ""); // 1 1 1 1
49  return CC;
50 }
51 
52 /// This is the complement of getICmpCode, which turns an opcode and two
53 /// operands into either a constant true or false, or a brand new ICmp
54 /// instruction. The sign is passed in to determine which kind of predicate to
55 /// use in the new icmp instruction.
56 static Value *getNewICmpValue(unsigned Code, bool Sign, Value *LHS, Value *RHS,
57  InstCombiner::BuilderTy &Builder) {
58  ICmpInst::Predicate NewPred;
59  if (Constant *TorF = getPredForICmpCode(Code, Sign, LHS->getType(), NewPred))
60  return TorF;
61  return Builder.CreateICmp(NewPred, LHS, RHS);
62 }
63 
64 /// This is the complement of getFCmpCode, which turns an opcode and two
65 /// operands into either a FCmp instruction, or a true/false constant.
66 static Value *getFCmpValue(unsigned Code, Value *LHS, Value *RHS,
67  InstCombiner::BuilderTy &Builder) {
68  const auto Pred = static_cast<FCmpInst::Predicate>(Code);
69  assert(FCmpInst::FCMP_FALSE <= Pred && Pred <= FCmpInst::FCMP_TRUE &&
70  "Unexpected FCmp predicate!");
71  if (Pred == FCmpInst::FCMP_FALSE)
73  if (Pred == FCmpInst::FCMP_TRUE)
75  return Builder.CreateFCmp(Pred, LHS, RHS);
76 }
77 
78 /// Transform BITWISE_OP(BSWAP(A),BSWAP(B)) or
79 /// BITWISE_OP(BSWAP(A), Constant) to BSWAP(BITWISE_OP(A, B))
80 /// \param I Binary operator to transform.
81 /// \return Pointer to node that must replace the original binary operator, or
82 /// null pointer if no transformation was made.
84  InstCombiner::BuilderTy &Builder) {
85  assert(I.isBitwiseLogicOp() && "Unexpected opcode for bswap simplifying");
86 
87  Value *OldLHS = I.getOperand(0);
88  Value *OldRHS = I.getOperand(1);
89 
90  Value *NewLHS;
91  if (!match(OldLHS, m_BSwap(m_Value(NewLHS))))
92  return nullptr;
93 
94  Value *NewRHS;
95  const APInt *C;
96 
97  if (match(OldRHS, m_BSwap(m_Value(NewRHS)))) {
98  // OP( BSWAP(x), BSWAP(y) ) -> BSWAP( OP(x, y) )
99  if (!OldLHS->hasOneUse() && !OldRHS->hasOneUse())
100  return nullptr;
101  // NewRHS initialized by the matcher.
102  } else if (match(OldRHS, m_APInt(C))) {
103  // OP( BSWAP(x), CONSTANT ) -> BSWAP( OP(x, BSWAP(CONSTANT) ) )
104  if (!OldLHS->hasOneUse())
105  return nullptr;
106  NewRHS = ConstantInt::get(I.getType(), C->byteSwap());
107  } else
108  return nullptr;
109 
110  Value *BinOp = Builder.CreateBinOp(I.getOpcode(), NewLHS, NewRHS);
111  Function *F = Intrinsic::getDeclaration(I.getModule(), Intrinsic::bswap,
112  I.getType());
113  return Builder.CreateCall(F, BinOp);
114 }
115 
116 /// This handles expressions of the form ((val OP C1) & C2). Where
117 /// the Op parameter is 'OP', OpRHS is 'C1', and AndRHS is 'C2'.
118 Instruction *InstCombiner::OptAndOp(BinaryOperator *Op,
119  ConstantInt *OpRHS,
120  ConstantInt *AndRHS,
121  BinaryOperator &TheAnd) {
122  Value *X = Op->getOperand(0);
123 
124  switch (Op->getOpcode()) {
125  default: break;
126  case Instruction::Add:
127  if (Op->hasOneUse()) {
128  // Adding a one to a single bit bit-field should be turned into an XOR
129  // of the bit. First thing to check is to see if this AND is with a
130  // single bit constant.
131  const APInt &AndRHSV = AndRHS->getValue();
132 
133  // If there is only one bit set.
134  if (AndRHSV.isPowerOf2()) {
135  // Ok, at this point, we know that we are masking the result of the
136  // ADD down to exactly one bit. If the constant we are adding has
137  // no bits set below this bit, then we can eliminate the ADD.
138  const APInt& AddRHS = OpRHS->getValue();
139 
140  // Check to see if any bits below the one bit set in AndRHSV are set.
141  if ((AddRHS & (AndRHSV - 1)).isNullValue()) {
142  // If not, the only thing that can effect the output of the AND is
143  // the bit specified by AndRHSV. If that bit is set, the effect of
144  // the XOR is to toggle the bit. If it is clear, then the ADD has
145  // no effect.
146  if ((AddRHS & AndRHSV).isNullValue()) { // Bit is not set, noop
147  TheAnd.setOperand(0, X);
148  return &TheAnd;
149  } else {
150  // Pull the XOR out of the AND.
151  Value *NewAnd = Builder.CreateAnd(X, AndRHS);
152  NewAnd->takeName(Op);
153  return BinaryOperator::CreateXor(NewAnd, AndRHS);
154  }
155  }
156  }
157  }
158  break;
159  }
160  return nullptr;
161 }
162 
163 /// Emit a computation of: (V >= Lo && V < Hi) if Inside is true, otherwise
164 /// (V < Lo || V >= Hi). This method expects that Lo <= Hi. IsSigned indicates
165 /// whether to treat V, Lo, and Hi as signed or not.
166 Value *InstCombiner::insertRangeTest(Value *V, const APInt &Lo, const APInt &Hi,
167  bool isSigned, bool Inside) {
168  assert((isSigned ? Lo.sle(Hi) : Lo.ule(Hi)) &&
169  "Lo is not <= Hi in range emission code!");
170 
171  Type *Ty = V->getType();
172  if (Lo == Hi)
173  return Inside ? ConstantInt::getFalse(Ty) : ConstantInt::getTrue(Ty);
174 
175  // V >= Min && V < Hi --> V < Hi
176  // V < Min || V >= Hi --> V >= Hi
178  if (isSigned ? Lo.isMinSignedValue() : Lo.isMinValue()) {
179  Pred = isSigned ? ICmpInst::getSignedPredicate(Pred) : Pred;
180  return Builder.CreateICmp(Pred, V, ConstantInt::get(Ty, Hi));
181  }
182 
183  // V >= Lo && V < Hi --> V - Lo u< Hi - Lo
184  // V < Lo || V >= Hi --> V - Lo u>= Hi - Lo
185  Value *VMinusLo =
186  Builder.CreateSub(V, ConstantInt::get(Ty, Lo), V->getName() + ".off");
187  Constant *HiMinusLo = ConstantInt::get(Ty, Hi - Lo);
188  return Builder.CreateICmp(Pred, VMinusLo, HiMinusLo);
189 }
190 
191 /// Classify (icmp eq (A & B), C) and (icmp ne (A & B), C) as matching patterns
192 /// that can be simplified.
193 /// One of A and B is considered the mask. The other is the value. This is
194 /// described as the "AMask" or "BMask" part of the enum. If the enum contains
195 /// only "Mask", then both A and B can be considered masks. If A is the mask,
196 /// then it was proven that (A & C) == C. This is trivial if C == A or C == 0.
197 /// If both A and C are constants, this proof is also easy.
198 /// For the following explanations, we assume that A is the mask.
199 ///
200 /// "AllOnes" declares that the comparison is true only if (A & B) == A or all
201 /// bits of A are set in B.
202 /// Example: (icmp eq (A & 3), 3) -> AMask_AllOnes
203 ///
204 /// "AllZeros" declares that the comparison is true only if (A & B) == 0 or all
205 /// bits of A are cleared in B.
206 /// Example: (icmp eq (A & 3), 0) -> Mask_AllZeroes
207 ///
208 /// "Mixed" declares that (A & B) == C and C might or might not contain any
209 /// number of one bits and zero bits.
210 /// Example: (icmp eq (A & 3), 1) -> AMask_Mixed
211 ///
212 /// "Not" means that in above descriptions "==" should be replaced by "!=".
213 /// Example: (icmp ne (A & 3), 3) -> AMask_NotAllOnes
214 ///
215 /// If the mask A contains a single bit, then the following is equivalent:
216 /// (icmp eq (A & B), A) equals (icmp ne (A & B), 0)
217 /// (icmp ne (A & B), A) equals (icmp eq (A & B), 0)
227  BMask_Mixed = 256,
229 };
230 
231 /// Return the set of patterns (from MaskedICmpType) that (icmp SCC (A & B), C)
232 /// satisfies.
233 static unsigned getMaskedICmpType(Value *A, Value *B, Value *C,
234  ICmpInst::Predicate Pred) {
235  ConstantInt *ACst = dyn_cast<ConstantInt>(A);
236  ConstantInt *BCst = dyn_cast<ConstantInt>(B);
237  ConstantInt *CCst = dyn_cast<ConstantInt>(C);
238  bool IsEq = (Pred == ICmpInst::ICMP_EQ);
239  bool IsAPow2 = (ACst && !ACst->isZero() && ACst->getValue().isPowerOf2());
240  bool IsBPow2 = (BCst && !BCst->isZero() && BCst->getValue().isPowerOf2());
241  unsigned MaskVal = 0;
242  if (CCst && CCst->isZero()) {
243  // if C is zero, then both A and B qualify as mask
244  MaskVal |= (IsEq ? (Mask_AllZeros | AMask_Mixed | BMask_Mixed)
246  if (IsAPow2)
247  MaskVal |= (IsEq ? (AMask_NotAllOnes | AMask_NotMixed)
248  : (AMask_AllOnes | AMask_Mixed));
249  if (IsBPow2)
250  MaskVal |= (IsEq ? (BMask_NotAllOnes | BMask_NotMixed)
251  : (BMask_AllOnes | BMask_Mixed));
252  return MaskVal;
253  }
254 
255  if (A == C) {
256  MaskVal |= (IsEq ? (AMask_AllOnes | AMask_Mixed)
258  if (IsAPow2)
259  MaskVal |= (IsEq ? (Mask_NotAllZeros | AMask_NotMixed)
260  : (Mask_AllZeros | AMask_Mixed));
261  } else if (ACst && CCst && ConstantExpr::getAnd(ACst, CCst) == CCst) {
262  MaskVal |= (IsEq ? AMask_Mixed : AMask_NotMixed);
263  }
264 
265  if (B == C) {
266  MaskVal |= (IsEq ? (BMask_AllOnes | BMask_Mixed)
268  if (IsBPow2)
269  MaskVal |= (IsEq ? (Mask_NotAllZeros | BMask_NotMixed)
270  : (Mask_AllZeros | BMask_Mixed));
271  } else if (BCst && CCst && ConstantExpr::getAnd(BCst, CCst) == CCst) {
272  MaskVal |= (IsEq ? BMask_Mixed : BMask_NotMixed);
273  }
274 
275  return MaskVal;
276 }
277 
278 /// Convert an analysis of a masked ICmp into its equivalent if all boolean
279 /// operations had the opposite sense. Since each "NotXXX" flag (recording !=)
280 /// is adjacent to the corresponding normal flag (recording ==), this just
281 /// involves swapping those bits over.
282 static unsigned conjugateICmpMask(unsigned Mask) {
283  unsigned NewMask;
284  NewMask = (Mask & (AMask_AllOnes | BMask_AllOnes | Mask_AllZeros |
286  << 1;
287 
288  NewMask |= (Mask & (AMask_NotAllOnes | BMask_NotAllOnes | Mask_NotAllZeros |
290  >> 1;
291 
292  return NewMask;
293 }
294 
295 // Adapts the external decomposeBitTestICmp for local use.
296 static bool decomposeBitTestICmp(Value *LHS, Value *RHS, CmpInst::Predicate &Pred,
297  Value *&X, Value *&Y, Value *&Z) {
298  APInt Mask;
299  if (!llvm::decomposeBitTestICmp(LHS, RHS, Pred, X, Mask))
300  return false;
301 
302  Y = ConstantInt::get(X->getType(), Mask);
303  Z = ConstantInt::get(X->getType(), 0);
304  return true;
305 }
306 
307 /// Handle (icmp(A & B) ==/!= C) &/| (icmp(A & D) ==/!= E).
308 /// Return the pattern classes (from MaskedICmpType) for the left hand side and
309 /// the right hand side as a pair.
310 /// LHS and RHS are the left hand side and the right hand side ICmps and PredL
311 /// and PredR are their predicates, respectively.
312 static
315  Value *&D, Value *&E, ICmpInst *LHS,
316  ICmpInst *RHS,
317  ICmpInst::Predicate &PredL,
318  ICmpInst::Predicate &PredR) {
319  // vectors are not (yet?) supported. Don't support pointers either.
320  if (!LHS->getOperand(0)->getType()->isIntegerTy() ||
321  !RHS->getOperand(0)->getType()->isIntegerTy())
322  return None;
323 
324  // Here comes the tricky part:
325  // LHS might be of the form L11 & L12 == X, X == L21 & L22,
326  // and L11 & L12 == L21 & L22. The same goes for RHS.
327  // Now we must find those components L** and R**, that are equal, so
328  // that we can extract the parameters A, B, C, D, and E for the canonical
329  // above.
330  Value *L1 = LHS->getOperand(0);
331  Value *L2 = LHS->getOperand(1);
332  Value *L11, *L12, *L21, *L22;
333  // Check whether the icmp can be decomposed into a bit test.
334  if (decomposeBitTestICmp(L1, L2, PredL, L11, L12, L2)) {
335  L21 = L22 = L1 = nullptr;
336  } else {
337  // Look for ANDs in the LHS icmp.
338  if (!match(L1, m_And(m_Value(L11), m_Value(L12)))) {
339  // Any icmp can be viewed as being trivially masked; if it allows us to
340  // remove one, it's worth it.
341  L11 = L1;
342  L12 = Constant::getAllOnesValue(L1->getType());
343  }
344 
345  if (!match(L2, m_And(m_Value(L21), m_Value(L22)))) {
346  L21 = L2;
347  L22 = Constant::getAllOnesValue(L2->getType());
348  }
349  }
350 
351  // Bail if LHS was a icmp that can't be decomposed into an equality.
352  if (!ICmpInst::isEquality(PredL))
353  return None;
354 
355  Value *R1 = RHS->getOperand(0);
356  Value *R2 = RHS->getOperand(1);
357  Value *R11, *R12;
358  bool Ok = false;
359  if (decomposeBitTestICmp(R1, R2, PredR, R11, R12, R2)) {
360  if (R11 == L11 || R11 == L12 || R11 == L21 || R11 == L22) {
361  A = R11;
362  D = R12;
363  } else if (R12 == L11 || R12 == L12 || R12 == L21 || R12 == L22) {
364  A = R12;
365  D = R11;
366  } else {
367  return None;
368  }
369  E = R2;
370  R1 = nullptr;
371  Ok = true;
372  } else {
373  if (!match(R1, m_And(m_Value(R11), m_Value(R12)))) {
374  // As before, model no mask as a trivial mask if it'll let us do an
375  // optimization.
376  R11 = R1;
377  R12 = Constant::getAllOnesValue(R1->getType());
378  }
379 
380  if (R11 == L11 || R11 == L12 || R11 == L21 || R11 == L22) {
381  A = R11;
382  D = R12;
383  E = R2;
384  Ok = true;
385  } else if (R12 == L11 || R12 == L12 || R12 == L21 || R12 == L22) {
386  A = R12;
387  D = R11;
388  E = R2;
389  Ok = true;
390  }
391  }
392 
393  // Bail if RHS was a icmp that can't be decomposed into an equality.
394  if (!ICmpInst::isEquality(PredR))
395  return None;
396 
397  // Look for ANDs on the right side of the RHS icmp.
398  if (!Ok) {
399  if (!match(R2, m_And(m_Value(R11), m_Value(R12)))) {
400  R11 = R2;
401  R12 = Constant::getAllOnesValue(R2->getType());
402  }
403 
404  if (R11 == L11 || R11 == L12 || R11 == L21 || R11 == L22) {
405  A = R11;
406  D = R12;
407  E = R1;
408  Ok = true;
409  } else if (R12 == L11 || R12 == L12 || R12 == L21 || R12 == L22) {
410  A = R12;
411  D = R11;
412  E = R1;
413  Ok = true;
414  } else {
415  return None;
416  }
417  }
418  if (!Ok)
419  return None;
420 
421  if (L11 == A) {
422  B = L12;
423  C = L2;
424  } else if (L12 == A) {
425  B = L11;
426  C = L2;
427  } else if (L21 == A) {
428  B = L22;
429  C = L1;
430  } else if (L22 == A) {
431  B = L21;
432  C = L1;
433  }
434 
435  unsigned LeftType = getMaskedICmpType(A, B, C, PredL);
436  unsigned RightType = getMaskedICmpType(A, D, E, PredR);
437  return Optional<std::pair<unsigned, unsigned>>(std::make_pair(LeftType, RightType));
438 }
439 
440 /// Try to fold (icmp(A & B) ==/!= C) &/| (icmp(A & D) ==/!= E) into a single
441 /// (icmp(A & X) ==/!= Y), where the left-hand side is of type Mask_NotAllZeros
442 /// and the right hand side is of type BMask_Mixed. For example,
443 /// (icmp (A & 12) != 0) & (icmp (A & 15) == 8) -> (icmp (A & 15) == 8).
445  ICmpInst *LHS, ICmpInst *RHS, bool IsAnd,
446  Value *A, Value *B, Value *C, Value *D, Value *E,
449  // We are given the canonical form:
450  // (icmp ne (A & B), 0) & (icmp eq (A & D), E).
451  // where D & E == E.
452  //
453  // If IsAnd is false, we get it in negated form:
454  // (icmp eq (A & B), 0) | (icmp ne (A & D), E) ->
455  // !((icmp ne (A & B), 0) & (icmp eq (A & D), E)).
456  //
457  // We currently handle the case of B, C, D, E are constant.
458  //
459  ConstantInt *BCst = dyn_cast<ConstantInt>(B);
460  if (!BCst)
461  return nullptr;
462  ConstantInt *CCst = dyn_cast<ConstantInt>(C);
463  if (!CCst)
464  return nullptr;
465  ConstantInt *DCst = dyn_cast<ConstantInt>(D);
466  if (!DCst)
467  return nullptr;
468  ConstantInt *ECst = dyn_cast<ConstantInt>(E);
469  if (!ECst)
470  return nullptr;
471 
473 
474  // Update E to the canonical form when D is a power of two and RHS is
475  // canonicalized as,
476  // (icmp ne (A & D), 0) -> (icmp eq (A & D), D) or
477  // (icmp ne (A & D), D) -> (icmp eq (A & D), 0).
478  if (PredR != NewCC)
479  ECst = cast<ConstantInt>(ConstantExpr::getXor(DCst, ECst));
480 
481  // If B or D is zero, skip because if LHS or RHS can be trivially folded by
482  // other folding rules and this pattern won't apply any more.
483  if (BCst->getValue() == 0 || DCst->getValue() == 0)
484  return nullptr;
485 
486  // If B and D don't intersect, ie. (B & D) == 0, no folding because we can't
487  // deduce anything from it.
488  // For example,
489  // (icmp ne (A & 12), 0) & (icmp eq (A & 3), 1) -> no folding.
490  if ((BCst->getValue() & DCst->getValue()) == 0)
491  return nullptr;
492 
493  // If the following two conditions are met:
494  //
495  // 1. mask B covers only a single bit that's not covered by mask D, that is,
496  // (B & (B ^ D)) is a power of 2 (in other words, B minus the intersection of
497  // B and D has only one bit set) and,
498  //
499  // 2. RHS (and E) indicates that the rest of B's bits are zero (in other
500  // words, the intersection of B and D is zero), that is, ((B & D) & E) == 0
501  //
502  // then that single bit in B must be one and thus the whole expression can be
503  // folded to
504  // (A & (B | D)) == (B & (B ^ D)) | E.
505  //
506  // For example,
507  // (icmp ne (A & 12), 0) & (icmp eq (A & 7), 1) -> (icmp eq (A & 15), 9)
508  // (icmp ne (A & 15), 0) & (icmp eq (A & 7), 0) -> (icmp eq (A & 15), 8)
509  if ((((BCst->getValue() & DCst->getValue()) & ECst->getValue()) == 0) &&
510  (BCst->getValue() & (BCst->getValue() ^ DCst->getValue())).isPowerOf2()) {
511  APInt BorD = BCst->getValue() | DCst->getValue();
512  APInt BandBxorDorE = (BCst->getValue() & (BCst->getValue() ^ DCst->getValue())) |
513  ECst->getValue();
514  Value *NewMask = ConstantInt::get(BCst->getType(), BorD);
515  Value *NewMaskedValue = ConstantInt::get(BCst->getType(), BandBxorDorE);
516  Value *NewAnd = Builder.CreateAnd(A, NewMask);
517  return Builder.CreateICmp(NewCC, NewAnd, NewMaskedValue);
518  }
519 
520  auto IsSubSetOrEqual = [](ConstantInt *C1, ConstantInt *C2) {
521  return (C1->getValue() & C2->getValue()) == C1->getValue();
522  };
523  auto IsSuperSetOrEqual = [](ConstantInt *C1, ConstantInt *C2) {
524  return (C1->getValue() & C2->getValue()) == C2->getValue();
525  };
526 
527  // In the following, we consider only the cases where B is a superset of D, B
528  // is a subset of D, or B == D because otherwise there's at least one bit
529  // covered by B but not D, in which case we can't deduce much from it, so
530  // no folding (aside from the single must-be-one bit case right above.)
531  // For example,
532  // (icmp ne (A & 14), 0) & (icmp eq (A & 3), 1) -> no folding.
533  if (!IsSubSetOrEqual(BCst, DCst) && !IsSuperSetOrEqual(BCst, DCst))
534  return nullptr;
535 
536  // At this point, either B is a superset of D, B is a subset of D or B == D.
537 
538  // If E is zero, if B is a subset of (or equal to) D, LHS and RHS contradict
539  // and the whole expression becomes false (or true if negated), otherwise, no
540  // folding.
541  // For example,
542  // (icmp ne (A & 3), 0) & (icmp eq (A & 7), 0) -> false.
543  // (icmp ne (A & 15), 0) & (icmp eq (A & 3), 0) -> no folding.
544  if (ECst->isZero()) {
545  if (IsSubSetOrEqual(BCst, DCst))
546  return ConstantInt::get(LHS->getType(), !IsAnd);
547  return nullptr;
548  }
549 
550  // At this point, B, D, E aren't zero and (B & D) == B, (B & D) == D or B ==
551  // D. If B is a superset of (or equal to) D, since E is not zero, LHS is
552  // subsumed by RHS (RHS implies LHS.) So the whole expression becomes
553  // RHS. For example,
554  // (icmp ne (A & 255), 0) & (icmp eq (A & 15), 8) -> (icmp eq (A & 15), 8).
555  // (icmp ne (A & 15), 0) & (icmp eq (A & 15), 8) -> (icmp eq (A & 15), 8).
556  if (IsSuperSetOrEqual(BCst, DCst))
557  return RHS;
558  // Otherwise, B is a subset of D. If B and E have a common bit set,
559  // ie. (B & E) != 0, then LHS is subsumed by RHS. For example.
560  // (icmp ne (A & 12), 0) & (icmp eq (A & 15), 8) -> (icmp eq (A & 15), 8).
561  assert(IsSubSetOrEqual(BCst, DCst) && "Precondition due to above code");
562  if ((BCst->getValue() & ECst->getValue()) != 0)
563  return RHS;
564  // Otherwise, LHS and RHS contradict and the whole expression becomes false
565  // (or true if negated.) For example,
566  // (icmp ne (A & 7), 0) & (icmp eq (A & 15), 8) -> false.
567  // (icmp ne (A & 6), 0) & (icmp eq (A & 15), 8) -> false.
568  return ConstantInt::get(LHS->getType(), !IsAnd);
569 }
570 
571 /// Try to fold (icmp(A & B) ==/!= 0) &/| (icmp(A & D) ==/!= E) into a single
572 /// (icmp(A & X) ==/!= Y), where the left-hand side and the right hand side
573 /// aren't of the common mask pattern type.
575  ICmpInst *LHS, ICmpInst *RHS, bool IsAnd,
576  Value *A, Value *B, Value *C, Value *D, Value *E,
578  unsigned LHSMask, unsigned RHSMask,
581  "Expected equality predicates for masked type of icmps.");
582  // Handle Mask_NotAllZeros-BMask_Mixed cases.
583  // (icmp ne/eq (A & B), C) &/| (icmp eq/ne (A & D), E), or
584  // (icmp eq/ne (A & B), C) &/| (icmp ne/eq (A & D), E)
585  // which gets swapped to
586  // (icmp ne/eq (A & D), E) &/| (icmp eq/ne (A & B), C).
587  if (!IsAnd) {
588  LHSMask = conjugateICmpMask(LHSMask);
589  RHSMask = conjugateICmpMask(RHSMask);
590  }
591  if ((LHSMask & Mask_NotAllZeros) && (RHSMask & BMask_Mixed)) {
593  LHS, RHS, IsAnd, A, B, C, D, E,
594  PredL, PredR, Builder)) {
595  return V;
596  }
597  } else if ((LHSMask & BMask_Mixed) && (RHSMask & Mask_NotAllZeros)) {
599  RHS, LHS, IsAnd, A, D, E, B, C,
600  PredR, PredL, Builder)) {
601  return V;
602  }
603  }
604  return nullptr;
605 }
606 
607 /// Try to fold (icmp(A & B) ==/!= C) &/| (icmp(A & D) ==/!= E)
608 /// into a single (icmp(A & X) ==/!= Y).
609 static Value *foldLogOpOfMaskedICmps(ICmpInst *LHS, ICmpInst *RHS, bool IsAnd,
611  Value *A = nullptr, *B = nullptr, *C = nullptr, *D = nullptr, *E = nullptr;
612  ICmpInst::Predicate PredL = LHS->getPredicate(), PredR = RHS->getPredicate();
614  getMaskedTypeForICmpPair(A, B, C, D, E, LHS, RHS, PredL, PredR);
615  if (!MaskPair)
616  return nullptr;
618  "Expected equality predicates for masked type of icmps.");
619  unsigned LHSMask = MaskPair->first;
620  unsigned RHSMask = MaskPair->second;
621  unsigned Mask = LHSMask & RHSMask;
622  if (Mask == 0) {
623  // Even if the two sides don't share a common pattern, check if folding can
624  // still happen.
626  LHS, RHS, IsAnd, A, B, C, D, E, PredL, PredR, LHSMask, RHSMask,
627  Builder))
628  return V;
629  return nullptr;
630  }
631 
632  // In full generality:
633  // (icmp (A & B) Op C) | (icmp (A & D) Op E)
634  // == ![ (icmp (A & B) !Op C) & (icmp (A & D) !Op E) ]
635  //
636  // If the latter can be converted into (icmp (A & X) Op Y) then the former is
637  // equivalent to (icmp (A & X) !Op Y).
638  //
639  // Therefore, we can pretend for the rest of this function that we're dealing
640  // with the conjunction, provided we flip the sense of any comparisons (both
641  // input and output).
642 
643  // In most cases we're going to produce an EQ for the "&&" case.
645  if (!IsAnd) {
646  // Convert the masking analysis into its equivalent with negated
647  // comparisons.
648  Mask = conjugateICmpMask(Mask);
649  }
650 
651  if (Mask & Mask_AllZeros) {
652  // (icmp eq (A & B), 0) & (icmp eq (A & D), 0)
653  // -> (icmp eq (A & (B|D)), 0)
654  Value *NewOr = Builder.CreateOr(B, D);
655  Value *NewAnd = Builder.CreateAnd(A, NewOr);
656  // We can't use C as zero because we might actually handle
657  // (icmp ne (A & B), B) & (icmp ne (A & D), D)
658  // with B and D, having a single bit set.
659  Value *Zero = Constant::getNullValue(A->getType());
660  return Builder.CreateICmp(NewCC, NewAnd, Zero);
661  }
662  if (Mask & BMask_AllOnes) {
663  // (icmp eq (A & B), B) & (icmp eq (A & D), D)
664  // -> (icmp eq (A & (B|D)), (B|D))
665  Value *NewOr = Builder.CreateOr(B, D);
666  Value *NewAnd = Builder.CreateAnd(A, NewOr);
667  return Builder.CreateICmp(NewCC, NewAnd, NewOr);
668  }
669  if (Mask & AMask_AllOnes) {
670  // (icmp eq (A & B), A) & (icmp eq (A & D), A)
671  // -> (icmp eq (A & (B&D)), A)
672  Value *NewAnd1 = Builder.CreateAnd(B, D);
673  Value *NewAnd2 = Builder.CreateAnd(A, NewAnd1);
674  return Builder.CreateICmp(NewCC, NewAnd2, A);
675  }
676 
677  // Remaining cases assume at least that B and D are constant, and depend on
678  // their actual values. This isn't strictly necessary, just a "handle the
679  // easy cases for now" decision.
680  ConstantInt *BCst = dyn_cast<ConstantInt>(B);
681  if (!BCst)
682  return nullptr;
683  ConstantInt *DCst = dyn_cast<ConstantInt>(D);
684  if (!DCst)
685  return nullptr;
686 
687  if (Mask & (Mask_NotAllZeros | BMask_NotAllOnes)) {
688  // (icmp ne (A & B), 0) & (icmp ne (A & D), 0) and
689  // (icmp ne (A & B), B) & (icmp ne (A & D), D)
690  // -> (icmp ne (A & B), 0) or (icmp ne (A & D), 0)
691  // Only valid if one of the masks is a superset of the other (check "B&D" is
692  // the same as either B or D).
693  APInt NewMask = BCst->getValue() & DCst->getValue();
694 
695  if (NewMask == BCst->getValue())
696  return LHS;
697  else if (NewMask == DCst->getValue())
698  return RHS;
699  }
700 
701  if (Mask & AMask_NotAllOnes) {
702  // (icmp ne (A & B), B) & (icmp ne (A & D), D)
703  // -> (icmp ne (A & B), A) or (icmp ne (A & D), A)
704  // Only valid if one of the masks is a superset of the other (check "B|D" is
705  // the same as either B or D).
706  APInt NewMask = BCst->getValue() | DCst->getValue();
707 
708  if (NewMask == BCst->getValue())
709  return LHS;
710  else if (NewMask == DCst->getValue())
711  return RHS;
712  }
713 
714  if (Mask & BMask_Mixed) {
715  // (icmp eq (A & B), C) & (icmp eq (A & D), E)
716  // We already know that B & C == C && D & E == E.
717  // If we can prove that (B & D) & (C ^ E) == 0, that is, the bits of
718  // C and E, which are shared by both the mask B and the mask D, don't
719  // contradict, then we can transform to
720  // -> (icmp eq (A & (B|D)), (C|E))
721  // Currently, we only handle the case of B, C, D, and E being constant.
722  // We can't simply use C and E because we might actually handle
723  // (icmp ne (A & B), B) & (icmp eq (A & D), D)
724  // with B and D, having a single bit set.
725  ConstantInt *CCst = dyn_cast<ConstantInt>(C);
726  if (!CCst)
727  return nullptr;
728  ConstantInt *ECst = dyn_cast<ConstantInt>(E);
729  if (!ECst)
730  return nullptr;
731  if (PredL != NewCC)
732  CCst = cast<ConstantInt>(ConstantExpr::getXor(BCst, CCst));
733  if (PredR != NewCC)
734  ECst = cast<ConstantInt>(ConstantExpr::getXor(DCst, ECst));
735 
736  // If there is a conflict, we should actually return a false for the
737  // whole construct.
738  if (((BCst->getValue() & DCst->getValue()) &
739  (CCst->getValue() ^ ECst->getValue())).getBoolValue())
740  return ConstantInt::get(LHS->getType(), !IsAnd);
741 
742  Value *NewOr1 = Builder.CreateOr(B, D);
743  Value *NewOr2 = ConstantExpr::getOr(CCst, ECst);
744  Value *NewAnd = Builder.CreateAnd(A, NewOr1);
745  return Builder.CreateICmp(NewCC, NewAnd, NewOr2);
746  }
747 
748  return nullptr;
749 }
750 
751 /// Try to fold a signed range checked with lower bound 0 to an unsigned icmp.
752 /// Example: (icmp sge x, 0) & (icmp slt x, n) --> icmp ult x, n
753 /// If \p Inverted is true then the check is for the inverted range, e.g.
754 /// (icmp slt x, 0) | (icmp sgt x, n) --> icmp ugt x, n
756  bool Inverted) {
757  // Check the lower range comparison, e.g. x >= 0
758  // InstCombine already ensured that if there is a constant it's on the RHS.
759  ConstantInt *RangeStart = dyn_cast<ConstantInt>(Cmp0->getOperand(1));
760  if (!RangeStart)
761  return nullptr;
762 
763  ICmpInst::Predicate Pred0 = (Inverted ? Cmp0->getInversePredicate() :
764  Cmp0->getPredicate());
765 
766  // Accept x > -1 or x >= 0 (after potentially inverting the predicate).
767  if (!((Pred0 == ICmpInst::ICMP_SGT && RangeStart->isMinusOne()) ||
768  (Pred0 == ICmpInst::ICMP_SGE && RangeStart->isZero())))
769  return nullptr;
770 
771  ICmpInst::Predicate Pred1 = (Inverted ? Cmp1->getInversePredicate() :
772  Cmp1->getPredicate());
773 
774  Value *Input = Cmp0->getOperand(0);
775  Value *RangeEnd;
776  if (Cmp1->getOperand(0) == Input) {
777  // For the upper range compare we have: icmp x, n
778  RangeEnd = Cmp1->getOperand(1);
779  } else if (Cmp1->getOperand(1) == Input) {
780  // For the upper range compare we have: icmp n, x
781  RangeEnd = Cmp1->getOperand(0);
782  Pred1 = ICmpInst::getSwappedPredicate(Pred1);
783  } else {
784  return nullptr;
785  }
786 
787  // Check the upper range comparison, e.g. x < n
788  ICmpInst::Predicate NewPred;
789  switch (Pred1) {
790  case ICmpInst::ICMP_SLT: NewPred = ICmpInst::ICMP_ULT; break;
791  case ICmpInst::ICMP_SLE: NewPred = ICmpInst::ICMP_ULE; break;
792  default: return nullptr;
793  }
794 
795  // This simplification is only valid if the upper range is not negative.
796  KnownBits Known = computeKnownBits(RangeEnd, /*Depth=*/0, Cmp1);
797  if (!Known.isNonNegative())
798  return nullptr;
799 
800  if (Inverted)
801  NewPred = ICmpInst::getInversePredicate(NewPred);
802 
803  return Builder.CreateICmp(NewPred, Input, RangeEnd);
804 }
805 
806 static Value *
808  bool JoinedByAnd,
809  InstCombiner::BuilderTy &Builder) {
810  Value *X = LHS->getOperand(0);
811  if (X != RHS->getOperand(0))
812  return nullptr;
813 
814  const APInt *C1, *C2;
815  if (!match(LHS->getOperand(1), m_APInt(C1)) ||
816  !match(RHS->getOperand(1), m_APInt(C2)))
817  return nullptr;
818 
819  // We only handle (X != C1 && X != C2) and (X == C1 || X == C2).
820  ICmpInst::Predicate Pred = LHS->getPredicate();
821  if (Pred != RHS->getPredicate())
822  return nullptr;
823  if (JoinedByAnd && Pred != ICmpInst::ICMP_NE)
824  return nullptr;
825  if (!JoinedByAnd && Pred != ICmpInst::ICMP_EQ)
826  return nullptr;
827 
828  // The larger unsigned constant goes on the right.
829  if (C1->ugt(*C2))
830  std::swap(C1, C2);
831 
832  APInt Xor = *C1 ^ *C2;
833  if (Xor.isPowerOf2()) {
834  // If LHSC and RHSC differ by only one bit, then set that bit in X and
835  // compare against the larger constant:
836  // (X == C1 || X == C2) --> (X | (C1 ^ C2)) == C2
837  // (X != C1 && X != C2) --> (X | (C1 ^ C2)) != C2
838  // We choose an 'or' with a Pow2 constant rather than the inverse mask with
839  // 'and' because that may lead to smaller codegen from a smaller constant.
840  Value *Or = Builder.CreateOr(X, ConstantInt::get(X->getType(), Xor));
841  return Builder.CreateICmp(Pred, Or, ConstantInt::get(X->getType(), *C2));
842  }
843 
844  // Special case: get the ordering right when the values wrap around zero.
845  // Ie, we assumed the constants were unsigned when swapping earlier.
846  if (C1->isNullValue() && C2->isAllOnesValue())
847  std::swap(C1, C2);
848 
849  if (*C1 == *C2 - 1) {
850  // (X == 13 || X == 14) --> X - 13 <=u 1
851  // (X != 13 && X != 14) --> X - 13 >u 1
852  // An 'add' is the canonical IR form, so favor that over a 'sub'.
853  Value *Add = Builder.CreateAdd(X, ConstantInt::get(X->getType(), -(*C1)));
854  auto NewPred = JoinedByAnd ? ICmpInst::ICMP_UGT : ICmpInst::ICMP_ULE;
855  return Builder.CreateICmp(NewPred, Add, ConstantInt::get(X->getType(), 1));
856  }
857 
858  return nullptr;
859 }
860 
861 // Fold (iszero(A & K1) | iszero(A & K2)) -> (A & (K1 | K2)) != (K1 | K2)
862 // Fold (!iszero(A & K1) & !iszero(A & K2)) -> (A & (K1 | K2)) == (K1 | K2)
863 Value *InstCombiner::foldAndOrOfICmpsOfAndWithPow2(ICmpInst *LHS, ICmpInst *RHS,
864  bool JoinedByAnd,
865  Instruction &CxtI) {
866  ICmpInst::Predicate Pred = LHS->getPredicate();
867  if (Pred != RHS->getPredicate())
868  return nullptr;
869  if (JoinedByAnd && Pred != ICmpInst::ICMP_NE)
870  return nullptr;
871  if (!JoinedByAnd && Pred != ICmpInst::ICMP_EQ)
872  return nullptr;
873 
874  // TODO support vector splats
875  ConstantInt *LHSC = dyn_cast<ConstantInt>(LHS->getOperand(1));
876  ConstantInt *RHSC = dyn_cast<ConstantInt>(RHS->getOperand(1));
877  if (!LHSC || !RHSC || !LHSC->isZero() || !RHSC->isZero())
878  return nullptr;
879 
880  Value *A, *B, *C, *D;
881  if (match(LHS->getOperand(0), m_And(m_Value(A), m_Value(B))) &&
882  match(RHS->getOperand(0), m_And(m_Value(C), m_Value(D)))) {
883  if (A == D || B == D)
884  std::swap(C, D);
885  if (B == C)
886  std::swap(A, B);
887 
888  if (A == C &&
889  isKnownToBeAPowerOfTwo(B, false, 0, &CxtI) &&
890  isKnownToBeAPowerOfTwo(D, false, 0, &CxtI)) {
891  Value *Mask = Builder.CreateOr(B, D);
892  Value *Masked = Builder.CreateAnd(A, Mask);
893  auto NewPred = JoinedByAnd ? ICmpInst::ICMP_EQ : ICmpInst::ICMP_NE;
894  return Builder.CreateICmp(NewPred, Masked, Mask);
895  }
896  }
897 
898  return nullptr;
899 }
900 
901 /// General pattern:
902 /// X & Y
903 ///
904 /// Where Y is checking that all the high bits (covered by a mask 4294967168)
905 /// are uniform, i.e. %arg & 4294967168 can be either 4294967168 or 0
906 /// Pattern can be one of:
907 /// %t = add i32 %arg, 128
908 /// %r = icmp ult i32 %t, 256
909 /// Or
910 /// %t0 = shl i32 %arg, 24
911 /// %t1 = ashr i32 %t0, 24
912 /// %r = icmp eq i32 %t1, %arg
913 /// Or
914 /// %t0 = trunc i32 %arg to i8
915 /// %t1 = sext i8 %t0 to i32
916 /// %r = icmp eq i32 %t1, %arg
917 /// This pattern is a signed truncation check.
918 ///
919 /// And X is checking that some bit in that same mask is zero.
920 /// I.e. can be one of:
921 /// %r = icmp sgt i32 %arg, -1
922 /// Or
923 /// %t = and i32 %arg, 2147483648
924 /// %r = icmp eq i32 %t, 0
925 ///
926 /// Since we are checking that all the bits in that mask are the same,
927 /// and a particular bit is zero, what we are really checking is that all the
928 /// masked bits are zero.
929 /// So this should be transformed to:
930 /// %r = icmp ult i32 %arg, 128
932  Instruction &CxtI,
933  InstCombiner::BuilderTy &Builder) {
934  assert(CxtI.getOpcode() == Instruction::And);
935 
936  // Match icmp ult (add %arg, C01), C1 (C1 == C01 << 1; powers of two)
937  auto tryToMatchSignedTruncationCheck = [](ICmpInst *ICmp, Value *&X,
938  APInt &SignBitMask) -> bool {
939  CmpInst::Predicate Pred;
940  const APInt *I01, *I1; // powers of two; I1 == I01 << 1
941  if (!(match(ICmp,
942  m_ICmp(Pred, m_Add(m_Value(X), m_Power2(I01)), m_Power2(I1))) &&
943  Pred == ICmpInst::ICMP_ULT && I1->ugt(*I01) && I01->shl(1) == *I1))
944  return false;
945  // Which bit is the new sign bit as per the 'signed truncation' pattern?
946  SignBitMask = *I01;
947  return true;
948  };
949 
950  // One icmp needs to be 'signed truncation check'.
951  // We need to match this first, else we will mismatch commutative cases.
952  Value *X1;
953  APInt HighestBit;
954  ICmpInst *OtherICmp;
955  if (tryToMatchSignedTruncationCheck(ICmp1, X1, HighestBit))
956  OtherICmp = ICmp0;
957  else if (tryToMatchSignedTruncationCheck(ICmp0, X1, HighestBit))
958  OtherICmp = ICmp1;
959  else
960  return nullptr;
961 
962  assert(HighestBit.isPowerOf2() && "expected to be power of two (non-zero)");
963 
964  // Try to match/decompose into: icmp eq (X & Mask), 0
965  auto tryToDecompose = [](ICmpInst *ICmp, Value *&X,
966  APInt &UnsetBitsMask) -> bool {
967  CmpInst::Predicate Pred = ICmp->getPredicate();
968  // Can it be decomposed into icmp eq (X & Mask), 0 ?
969  if (llvm::decomposeBitTestICmp(ICmp->getOperand(0), ICmp->getOperand(1),
970  Pred, X, UnsetBitsMask,
971  /*LookThruTrunc=*/false) &&
972  Pred == ICmpInst::ICMP_EQ)
973  return true;
974  // Is it icmp eq (X & Mask), 0 already?
975  const APInt *Mask;
976  if (match(ICmp, m_ICmp(Pred, m_And(m_Value(X), m_APInt(Mask)), m_Zero())) &&
977  Pred == ICmpInst::ICMP_EQ) {
978  UnsetBitsMask = *Mask;
979  return true;
980  }
981  return false;
982  };
983 
984  // And the other icmp needs to be decomposable into a bit test.
985  Value *X0;
986  APInt UnsetBitsMask;
987  if (!tryToDecompose(OtherICmp, X0, UnsetBitsMask))
988  return nullptr;
989 
990  assert(!UnsetBitsMask.isNullValue() && "empty mask makes no sense.");
991 
992  // Are they working on the same value?
993  Value *X;
994  if (X1 == X0) {
995  // Ok as is.
996  X = X1;
997  } else if (match(X0, m_Trunc(m_Specific(X1)))) {
998  UnsetBitsMask = UnsetBitsMask.zext(X1->getType()->getScalarSizeInBits());
999  X = X1;
1000  } else
1001  return nullptr;
1002 
1003  // So which bits should be uniform as per the 'signed truncation check'?
1004  // (all the bits starting with (i.e. including) HighestBit)
1005  APInt SignBitsMask = ~(HighestBit - 1U);
1006 
1007  // UnsetBitsMask must have some common bits with SignBitsMask,
1008  if (!UnsetBitsMask.intersects(SignBitsMask))
1009  return nullptr;
1010 
1011  // Does UnsetBitsMask contain any bits outside of SignBitsMask?
1012  if (!UnsetBitsMask.isSubsetOf(SignBitsMask)) {
1013  APInt OtherHighestBit = (~UnsetBitsMask) + 1U;
1014  if (!OtherHighestBit.isPowerOf2())
1015  return nullptr;
1016  HighestBit = APIntOps::umin(HighestBit, OtherHighestBit);
1017  }
1018  // Else, if it does not, then all is ok as-is.
1019 
1020  // %r = icmp ult %X, SignBit
1021  return Builder.CreateICmpULT(X, ConstantInt::get(X->getType(), HighestBit),
1022  CxtI.getName() + ".simplified");
1023 }
1024 
1025 /// Fold (icmp)&(icmp) if possible.
1026 Value *InstCombiner::foldAndOfICmps(ICmpInst *LHS, ICmpInst *RHS,
1027  Instruction &CxtI) {
1028  // Fold (!iszero(A & K1) & !iszero(A & K2)) -> (A & (K1 | K2)) == (K1 | K2)
1029  // if K1 and K2 are a one-bit mask.
1030  if (Value *V = foldAndOrOfICmpsOfAndWithPow2(LHS, RHS, true, CxtI))
1031  return V;
1032 
1033  ICmpInst::Predicate PredL = LHS->getPredicate(), PredR = RHS->getPredicate();
1034 
1035  // (icmp1 A, B) & (icmp2 A, B) --> (icmp3 A, B)
1036  if (predicatesFoldable(PredL, PredR)) {
1037  if (LHS->getOperand(0) == RHS->getOperand(1) &&
1038  LHS->getOperand(1) == RHS->getOperand(0))
1039  LHS->swapOperands();
1040  if (LHS->getOperand(0) == RHS->getOperand(0) &&
1041  LHS->getOperand(1) == RHS->getOperand(1)) {
1042  Value *Op0 = LHS->getOperand(0), *Op1 = LHS->getOperand(1);
1043  unsigned Code = getICmpCode(LHS) & getICmpCode(RHS);
1044  bool IsSigned = LHS->isSigned() || RHS->isSigned();
1045  return getNewICmpValue(Code, IsSigned, Op0, Op1, Builder);
1046  }
1047  }
1048 
1049  // handle (roughly): (icmp eq (A & B), C) & (icmp eq (A & D), E)
1050  if (Value *V = foldLogOpOfMaskedICmps(LHS, RHS, true, Builder))
1051  return V;
1052 
1053  // E.g. (icmp sge x, 0) & (icmp slt x, n) --> icmp ult x, n
1054  if (Value *V = simplifyRangeCheck(LHS, RHS, /*Inverted=*/false))
1055  return V;
1056 
1057  // E.g. (icmp slt x, n) & (icmp sge x, 0) --> icmp ult x, n
1058  if (Value *V = simplifyRangeCheck(RHS, LHS, /*Inverted=*/false))
1059  return V;
1060 
1061  if (Value *V = foldAndOrOfEqualityCmpsWithConstants(LHS, RHS, true, Builder))
1062  return V;
1063 
1064  if (Value *V = foldSignedTruncationCheck(LHS, RHS, CxtI, Builder))
1065  return V;
1066 
1067  // This only handles icmp of constants: (icmp1 A, C1) & (icmp2 B, C2).
1068  Value *LHS0 = LHS->getOperand(0), *RHS0 = RHS->getOperand(0);
1069  ConstantInt *LHSC = dyn_cast<ConstantInt>(LHS->getOperand(1));
1070  ConstantInt *RHSC = dyn_cast<ConstantInt>(RHS->getOperand(1));
1071  if (!LHSC || !RHSC)
1072  return nullptr;
1073 
1074  if (LHSC == RHSC && PredL == PredR) {
1075  // (icmp ult A, C) & (icmp ult B, C) --> (icmp ult (A|B), C)
1076  // where C is a power of 2 or
1077  // (icmp eq A, 0) & (icmp eq B, 0) --> (icmp eq (A|B), 0)
1078  if ((PredL == ICmpInst::ICMP_ULT && LHSC->getValue().isPowerOf2()) ||
1079  (PredL == ICmpInst::ICMP_EQ && LHSC->isZero())) {
1080  Value *NewOr = Builder.CreateOr(LHS0, RHS0);
1081  return Builder.CreateICmp(PredL, NewOr, LHSC);
1082  }
1083  }
1084 
1085  // (trunc x) == C1 & (and x, CA) == C2 -> (and x, CA|CMAX) == C1|C2
1086  // where CMAX is the all ones value for the truncated type,
1087  // iff the lower bits of C2 and CA are zero.
1088  if (PredL == ICmpInst::ICMP_EQ && PredL == PredR && LHS->hasOneUse() &&
1089  RHS->hasOneUse()) {
1090  Value *V;
1091  ConstantInt *AndC, *SmallC = nullptr, *BigC = nullptr;
1092 
1093  // (trunc x) == C1 & (and x, CA) == C2
1094  // (and x, CA) == C2 & (trunc x) == C1
1095  if (match(RHS0, m_Trunc(m_Value(V))) &&
1096  match(LHS0, m_And(m_Specific(V), m_ConstantInt(AndC)))) {
1097  SmallC = RHSC;
1098  BigC = LHSC;
1099  } else if (match(LHS0, m_Trunc(m_Value(V))) &&
1100  match(RHS0, m_And(m_Specific(V), m_ConstantInt(AndC)))) {
1101  SmallC = LHSC;
1102  BigC = RHSC;
1103  }
1104 
1105  if (SmallC && BigC) {
1106  unsigned BigBitSize = BigC->getType()->getBitWidth();
1107  unsigned SmallBitSize = SmallC->getType()->getBitWidth();
1108 
1109  // Check that the low bits are zero.
1110  APInt Low = APInt::getLowBitsSet(BigBitSize, SmallBitSize);
1111  if ((Low & AndC->getValue()).isNullValue() &&
1112  (Low & BigC->getValue()).isNullValue()) {
1113  Value *NewAnd = Builder.CreateAnd(V, Low | AndC->getValue());
1114  APInt N = SmallC->getValue().zext(BigBitSize) | BigC->getValue();
1115  Value *NewVal = ConstantInt::get(AndC->getType()->getContext(), N);
1116  return Builder.CreateICmp(PredL, NewAnd, NewVal);
1117  }
1118  }
1119  }
1120 
1121  // From here on, we only handle:
1122  // (icmp1 A, C1) & (icmp2 A, C2) --> something simpler.
1123  if (LHS0 != RHS0)
1124  return nullptr;
1125 
1126  // ICMP_[US][GL]E X, C is folded to ICMP_[US][GL]T elsewhere.
1127  if (PredL == ICmpInst::ICMP_UGE || PredL == ICmpInst::ICMP_ULE ||
1128  PredR == ICmpInst::ICMP_UGE || PredR == ICmpInst::ICMP_ULE ||
1129  PredL == ICmpInst::ICMP_SGE || PredL == ICmpInst::ICMP_SLE ||
1130  PredR == ICmpInst::ICMP_SGE || PredR == ICmpInst::ICMP_SLE)
1131  return nullptr;
1132 
1133  // We can't fold (ugt x, C) & (sgt x, C2).
1134  if (!predicatesFoldable(PredL, PredR))
1135  return nullptr;
1136 
1137  // Ensure that the larger constant is on the RHS.
1138  bool ShouldSwap;
1139  if (CmpInst::isSigned(PredL) ||
1140  (ICmpInst::isEquality(PredL) && CmpInst::isSigned(PredR)))
1141  ShouldSwap = LHSC->getValue().sgt(RHSC->getValue());
1142  else
1143  ShouldSwap = LHSC->getValue().ugt(RHSC->getValue());
1144 
1145  if (ShouldSwap) {
1146  std::swap(LHS, RHS);
1147  std::swap(LHSC, RHSC);
1148  std::swap(PredL, PredR);
1149  }
1150 
1151  // At this point, we know we have two icmp instructions
1152  // comparing a value against two constants and and'ing the result
1153  // together. Because of the above check, we know that we only have
1154  // icmp eq, icmp ne, icmp [su]lt, and icmp [SU]gt here. We also know
1155  // (from the icmp folding check above), that the two constants
1156  // are not equal and that the larger constant is on the RHS
1157  assert(LHSC != RHSC && "Compares not folded above?");
1158 
1159  switch (PredL) {
1160  default:
1161  llvm_unreachable("Unknown integer condition code!");
1162  case ICmpInst::ICMP_NE:
1163  switch (PredR) {
1164  default:
1165  llvm_unreachable("Unknown integer condition code!");
1166  case ICmpInst::ICMP_ULT:
1167  if (LHSC == SubOne(RHSC)) // (X != 13 & X u< 14) -> X < 13
1168  return Builder.CreateICmpULT(LHS0, LHSC);
1169  if (LHSC->isZero()) // (X != 0 & X u< 14) -> X-1 u< 13
1170  return insertRangeTest(LHS0, LHSC->getValue() + 1, RHSC->getValue(),
1171  false, true);
1172  break; // (X != 13 & X u< 15) -> no change
1173  case ICmpInst::ICMP_SLT:
1174  if (LHSC == SubOne(RHSC)) // (X != 13 & X s< 14) -> X < 13
1175  return Builder.CreateICmpSLT(LHS0, LHSC);
1176  break; // (X != 13 & X s< 15) -> no change
1177  case ICmpInst::ICMP_NE:
1178  // Potential folds for this case should already be handled.
1179  break;
1180  }
1181  break;
1182  case ICmpInst::ICMP_UGT:
1183  switch (PredR) {
1184  default:
1185  llvm_unreachable("Unknown integer condition code!");
1186  case ICmpInst::ICMP_NE:
1187  if (RHSC == AddOne(LHSC)) // (X u> 13 & X != 14) -> X u> 14
1188  return Builder.CreateICmp(PredL, LHS0, RHSC);
1189  break; // (X u> 13 & X != 15) -> no change
1190  case ICmpInst::ICMP_ULT: // (X u> 13 & X u< 15) -> (X-14) <u 1
1191  return insertRangeTest(LHS0, LHSC->getValue() + 1, RHSC->getValue(),
1192  false, true);
1193  }
1194  break;
1195  case ICmpInst::ICMP_SGT:
1196  switch (PredR) {
1197  default:
1198  llvm_unreachable("Unknown integer condition code!");
1199  case ICmpInst::ICMP_NE:
1200  if (RHSC == AddOne(LHSC)) // (X s> 13 & X != 14) -> X s> 14
1201  return Builder.CreateICmp(PredL, LHS0, RHSC);
1202  break; // (X s> 13 & X != 15) -> no change
1203  case ICmpInst::ICMP_SLT: // (X s> 13 & X s< 15) -> (X-14) s< 1
1204  return insertRangeTest(LHS0, LHSC->getValue() + 1, RHSC->getValue(), true,
1205  true);
1206  }
1207  break;
1208  }
1209 
1210  return nullptr;
1211 }
1212 
1213 Value *InstCombiner::foldLogicOfFCmps(FCmpInst *LHS, FCmpInst *RHS, bool IsAnd) {
1214  Value *LHS0 = LHS->getOperand(0), *LHS1 = LHS->getOperand(1);
1215  Value *RHS0 = RHS->getOperand(0), *RHS1 = RHS->getOperand(1);
1216  FCmpInst::Predicate PredL = LHS->getPredicate(), PredR = RHS->getPredicate();
1217 
1218  if (LHS0 == RHS1 && RHS0 == LHS1) {
1219  // Swap RHS operands to match LHS.
1220  PredR = FCmpInst::getSwappedPredicate(PredR);
1221  std::swap(RHS0, RHS1);
1222  }
1223 
1224  // Simplify (fcmp cc0 x, y) & (fcmp cc1 x, y).
1225  // Suppose the relation between x and y is R, where R is one of
1226  // U(1000), L(0100), G(0010) or E(0001), and CC0 and CC1 are the bitmasks for
1227  // testing the desired relations.
1228  //
1229  // Since (R & CC0) and (R & CC1) are either R or 0, we actually have this:
1230  // bool(R & CC0) && bool(R & CC1)
1231  // = bool((R & CC0) & (R & CC1))
1232  // = bool(R & (CC0 & CC1)) <= by re-association, commutation, and idempotency
1233  //
1234  // Since (R & CC0) and (R & CC1) are either R or 0, we actually have this:
1235  // bool(R & CC0) || bool(R & CC1)
1236  // = bool((R & CC0) | (R & CC1))
1237  // = bool(R & (CC0 | CC1)) <= by reversed distribution (contribution? ;)
1238  if (LHS0 == RHS0 && LHS1 == RHS1) {
1239  unsigned FCmpCodeL = getFCmpCode(PredL);
1240  unsigned FCmpCodeR = getFCmpCode(PredR);
1241  unsigned NewPred = IsAnd ? FCmpCodeL & FCmpCodeR : FCmpCodeL | FCmpCodeR;
1242  return getFCmpValue(NewPred, LHS0, LHS1, Builder);
1243  }
1244 
1245  if ((PredL == FCmpInst::FCMP_ORD && PredR == FCmpInst::FCMP_ORD && IsAnd) ||
1246  (PredL == FCmpInst::FCMP_UNO && PredR == FCmpInst::FCMP_UNO && !IsAnd)) {
1247  if (LHS0->getType() != RHS0->getType())
1248  return nullptr;
1249 
1250  // FCmp canonicalization ensures that (fcmp ord/uno X, X) and
1251  // (fcmp ord/uno X, C) will be transformed to (fcmp X, +0.0).
1252  if (match(LHS1, m_PosZeroFP()) && match(RHS1, m_PosZeroFP()))
1253  // Ignore the constants because they are obviously not NANs:
1254  // (fcmp ord x, 0.0) & (fcmp ord y, 0.0) -> (fcmp ord x, y)
1255  // (fcmp uno x, 0.0) | (fcmp uno y, 0.0) -> (fcmp uno x, y)
1256  return Builder.CreateFCmp(PredL, LHS0, RHS0);
1257  }
1258 
1259  return nullptr;
1260 }
1261 
1262 /// Match De Morgan's Laws:
1263 /// (~A & ~B) == (~(A | B))
1264 /// (~A | ~B) == (~(A & B))
1266  InstCombiner::BuilderTy &Builder) {
1267  auto Opcode = I.getOpcode();
1268  assert((Opcode == Instruction::And || Opcode == Instruction::Or) &&
1269  "Trying to match De Morgan's Laws with something other than and/or");
1270 
1271  // Flip the logic operation.
1272  Opcode = (Opcode == Instruction::And) ? Instruction::Or : Instruction::And;
1273 
1274  Value *A, *B;
1275  if (match(I.getOperand(0), m_OneUse(m_Not(m_Value(A)))) &&
1276  match(I.getOperand(1), m_OneUse(m_Not(m_Value(B)))) &&
1277  !IsFreeToInvert(A, A->hasOneUse()) &&
1278  !IsFreeToInvert(B, B->hasOneUse())) {
1279  Value *AndOr = Builder.CreateBinOp(Opcode, A, B, I.getName() + ".demorgan");
1280  return BinaryOperator::CreateNot(AndOr);
1281  }
1282 
1283  return nullptr;
1284 }
1285 
1286 bool InstCombiner::shouldOptimizeCast(CastInst *CI) {
1287  Value *CastSrc = CI->getOperand(0);
1288 
1289  // Noop casts and casts of constants should be eliminated trivially.
1290  if (CI->getSrcTy() == CI->getDestTy() || isa<Constant>(CastSrc))
1291  return false;
1292 
1293  // If this cast is paired with another cast that can be eliminated, we prefer
1294  // to have it eliminated.
1295  if (const auto *PrecedingCI = dyn_cast<CastInst>(CastSrc))
1296  if (isEliminableCastPair(PrecedingCI, CI))
1297  return false;
1298 
1299  return true;
1300 }
1301 
1302 /// Fold {and,or,xor} (cast X), C.
1304  InstCombiner::BuilderTy &Builder) {
1305  Constant *C = dyn_cast<Constant>(Logic.getOperand(1));
1306  if (!C)
1307  return nullptr;
1308 
1309  auto LogicOpc = Logic.getOpcode();
1310  Type *DestTy = Logic.getType();
1311  Type *SrcTy = Cast->getSrcTy();
1312 
1313  // Move the logic operation ahead of a zext or sext if the constant is
1314  // unchanged in the smaller source type. Performing the logic in a smaller
1315  // type may provide more information to later folds, and the smaller logic
1316  // instruction may be cheaper (particularly in the case of vectors).
1317  Value *X;
1318  if (match(Cast, m_OneUse(m_ZExt(m_Value(X))))) {
1319  Constant *TruncC = ConstantExpr::getTrunc(C, SrcTy);
1320  Constant *ZextTruncC = ConstantExpr::getZExt(TruncC, DestTy);
1321  if (ZextTruncC == C) {
1322  // LogicOpc (zext X), C --> zext (LogicOpc X, C)
1323  Value *NewOp = Builder.CreateBinOp(LogicOpc, X, TruncC);
1324  return new ZExtInst(NewOp, DestTy);
1325  }
1326  }
1327 
1328  if (match(Cast, m_OneUse(m_SExt(m_Value(X))))) {
1329  Constant *TruncC = ConstantExpr::getTrunc(C, SrcTy);
1330  Constant *SextTruncC = ConstantExpr::getSExt(TruncC, DestTy);
1331  if (SextTruncC == C) {
1332  // LogicOpc (sext X), C --> sext (LogicOpc X, C)
1333  Value *NewOp = Builder.CreateBinOp(LogicOpc, X, TruncC);
1334  return new SExtInst(NewOp, DestTy);
1335  }
1336  }
1337 
1338  return nullptr;
1339 }
1340 
1341 /// Fold {and,or,xor} (cast X), Y.
1342 Instruction *InstCombiner::foldCastedBitwiseLogic(BinaryOperator &I) {
1343  auto LogicOpc = I.getOpcode();
1344  assert(I.isBitwiseLogicOp() && "Unexpected opcode for bitwise logic folding");
1345 
1346  Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
1347  CastInst *Cast0 = dyn_cast<CastInst>(Op0);
1348  if (!Cast0)
1349  return nullptr;
1350 
1351  // This must be a cast from an integer or integer vector source type to allow
1352  // transformation of the logic operation to the source type.
1353  Type *DestTy = I.getType();
1354  Type *SrcTy = Cast0->getSrcTy();
1355  if (!SrcTy->isIntOrIntVectorTy())
1356  return nullptr;
1357 
1358  if (Instruction *Ret = foldLogicCastConstant(I, Cast0, Builder))
1359  return Ret;
1360 
1361  CastInst *Cast1 = dyn_cast<CastInst>(Op1);
1362  if (!Cast1)
1363  return nullptr;
1364 
1365  // Both operands of the logic operation are casts. The casts must be of the
1366  // same type for reduction.
1367  auto CastOpcode = Cast0->getOpcode();
1368  if (CastOpcode != Cast1->getOpcode() || SrcTy != Cast1->getSrcTy())
1369  return nullptr;
1370 
1371  Value *Cast0Src = Cast0->getOperand(0);
1372  Value *Cast1Src = Cast1->getOperand(0);
1373 
1374  // fold logic(cast(A), cast(B)) -> cast(logic(A, B))
1375  if (shouldOptimizeCast(Cast0) && shouldOptimizeCast(Cast1)) {
1376  Value *NewOp = Builder.CreateBinOp(LogicOpc, Cast0Src, Cast1Src,
1377  I.getName());
1378  return CastInst::Create(CastOpcode, NewOp, DestTy);
1379  }
1380 
1381  // For now, only 'and'/'or' have optimizations after this.
1382  if (LogicOpc == Instruction::Xor)
1383  return nullptr;
1384 
1385  // If this is logic(cast(icmp), cast(icmp)), try to fold this even if the
1386  // cast is otherwise not optimizable. This happens for vector sexts.
1387  ICmpInst *ICmp0 = dyn_cast<ICmpInst>(Cast0Src);
1388  ICmpInst *ICmp1 = dyn_cast<ICmpInst>(Cast1Src);
1389  if (ICmp0 && ICmp1) {
1390  Value *Res = LogicOpc == Instruction::And ? foldAndOfICmps(ICmp0, ICmp1, I)
1391  : foldOrOfICmps(ICmp0, ICmp1, I);
1392  if (Res)
1393  return CastInst::Create(CastOpcode, Res, DestTy);
1394  return nullptr;
1395  }
1396 
1397  // If this is logic(cast(fcmp), cast(fcmp)), try to fold this even if the
1398  // cast is otherwise not optimizable. This happens for vector sexts.
1399  FCmpInst *FCmp0 = dyn_cast<FCmpInst>(Cast0Src);
1400  FCmpInst *FCmp1 = dyn_cast<FCmpInst>(Cast1Src);
1401  if (FCmp0 && FCmp1)
1402  if (Value *R = foldLogicOfFCmps(FCmp0, FCmp1, LogicOpc == Instruction::And))
1403  return CastInst::Create(CastOpcode, R, DestTy);
1404 
1405  return nullptr;
1406 }
1407 
1409  InstCombiner::BuilderTy &Builder) {
1410  assert(I.getOpcode() == Instruction::And);
1411  Value *Op0 = I.getOperand(0);
1412  Value *Op1 = I.getOperand(1);
1413  Value *A, *B;
1414 
1415  // Operand complexity canonicalization guarantees that the 'or' is Op0.
1416  // (A | B) & ~(A & B) --> A ^ B
1417  // (A | B) & ~(B & A) --> A ^ B
1418  if (match(&I, m_BinOp(m_Or(m_Value(A), m_Value(B)),
1419  m_Not(m_c_And(m_Deferred(A), m_Deferred(B))))))
1420  return BinaryOperator::CreateXor(A, B);
1421 
1422  // (A | ~B) & (~A | B) --> ~(A ^ B)
1423  // (A | ~B) & (B | ~A) --> ~(A ^ B)
1424  // (~B | A) & (~A | B) --> ~(A ^ B)
1425  // (~B | A) & (B | ~A) --> ~(A ^ B)
1426  if (Op0->hasOneUse() || Op1->hasOneUse())
1427  if (match(&I, m_BinOp(m_c_Or(m_Value(A), m_Not(m_Value(B))),
1428  m_c_Or(m_Not(m_Deferred(A)), m_Deferred(B)))))
1429  return BinaryOperator::CreateNot(Builder.CreateXor(A, B));
1430 
1431  return nullptr;
1432 }
1433 
1435  InstCombiner::BuilderTy &Builder) {
1436  assert(I.getOpcode() == Instruction::Or);
1437  Value *Op0 = I.getOperand(0);
1438  Value *Op1 = I.getOperand(1);
1439  Value *A, *B;
1440 
1441  // Operand complexity canonicalization guarantees that the 'and' is Op0.
1442  // (A & B) | ~(A | B) --> ~(A ^ B)
1443  // (A & B) | ~(B | A) --> ~(A ^ B)
1444  if (Op0->hasOneUse() || Op1->hasOneUse())
1445  if (match(Op0, m_And(m_Value(A), m_Value(B))) &&
1446  match(Op1, m_Not(m_c_Or(m_Specific(A), m_Specific(B)))))
1447  return BinaryOperator::CreateNot(Builder.CreateXor(A, B));
1448 
1449  // (A & ~B) | (~A & B) --> A ^ B
1450  // (A & ~B) | (B & ~A) --> A ^ B
1451  // (~B & A) | (~A & B) --> A ^ B
1452  // (~B & A) | (B & ~A) --> A ^ B
1453  if (match(Op0, m_c_And(m_Value(A), m_Not(m_Value(B)))) &&
1454  match(Op1, m_c_And(m_Not(m_Specific(A)), m_Specific(B))))
1455  return BinaryOperator::CreateXor(A, B);
1456 
1457  return nullptr;
1458 }
1459 
1460 /// Return true if a constant shift amount is always less than the specified
1461 /// bit-width. If not, the shift could create poison in the narrower type.
1462 static bool canNarrowShiftAmt(Constant *C, unsigned BitWidth) {
1463  if (auto *ScalarC = dyn_cast<ConstantInt>(C))
1464  return ScalarC->getZExtValue() < BitWidth;
1465 
1466  if (C->getType()->isVectorTy()) {
1467  // Check each element of a constant vector.
1468  unsigned NumElts = C->getType()->getVectorNumElements();
1469  for (unsigned i = 0; i != NumElts; ++i) {
1470  Constant *Elt = C->getAggregateElement(i);
1471  if (!Elt)
1472  return false;
1473  if (isa<UndefValue>(Elt))
1474  continue;
1475  auto *CI = dyn_cast<ConstantInt>(Elt);
1476  if (!CI || CI->getZExtValue() >= BitWidth)
1477  return false;
1478  }
1479  return true;
1480  }
1481 
1482  // The constant is a constant expression or unknown.
1483  return false;
1484 }
1485 
1486 /// Try to use narrower ops (sink zext ops) for an 'and' with binop operand and
1487 /// a common zext operand: and (binop (zext X), C), (zext X).
1488 Instruction *InstCombiner::narrowMaskedBinOp(BinaryOperator &And) {
1489  // This transform could also apply to {or, and, xor}, but there are better
1490  // folds for those cases, so we don't expect those patterns here. AShr is not
1491  // handled because it should always be transformed to LShr in this sequence.
1492  // The subtract transform is different because it has a constant on the left.
1493  // Add/mul commute the constant to RHS; sub with constant RHS becomes add.
1494  Value *Op0 = And.getOperand(0), *Op1 = And.getOperand(1);
1495  Constant *C;
1496  if (!match(Op0, m_OneUse(m_Add(m_Specific(Op1), m_Constant(C)))) &&
1497  !match(Op0, m_OneUse(m_Mul(m_Specific(Op1), m_Constant(C)))) &&
1498  !match(Op0, m_OneUse(m_LShr(m_Specific(Op1), m_Constant(C)))) &&
1499  !match(Op0, m_OneUse(m_Shl(m_Specific(Op1), m_Constant(C)))) &&
1500  !match(Op0, m_OneUse(m_Sub(m_Constant(C), m_Specific(Op1)))))
1501  return nullptr;
1502 
1503  Value *X;
1504  if (!match(Op1, m_ZExt(m_Value(X))) || Op1->hasNUsesOrMore(3))
1505  return nullptr;
1506 
1507  Type *Ty = And.getType();
1508  if (!isa<VectorType>(Ty) && !shouldChangeType(Ty, X->getType()))
1509  return nullptr;
1510 
1511  // If we're narrowing a shift, the shift amount must be safe (less than the
1512  // width) in the narrower type. If the shift amount is greater, instsimplify
1513  // usually handles that case, but we can't guarantee/assert it.
1514  Instruction::BinaryOps Opc = cast<BinaryOperator>(Op0)->getOpcode();
1515  if (Opc == Instruction::LShr || Opc == Instruction::Shl)
1517  return nullptr;
1518 
1519  // and (sub C, (zext X)), (zext X) --> zext (and (sub C', X), X)
1520  // and (binop (zext X), C), (zext X) --> zext (and (binop X, C'), X)
1521  Value *NewC = ConstantExpr::getTrunc(C, X->getType());
1522  Value *NewBO = Opc == Instruction::Sub ? Builder.CreateBinOp(Opc, NewC, X)
1523  : Builder.CreateBinOp(Opc, X, NewC);
1524  return new ZExtInst(Builder.CreateAnd(NewBO, X), Ty);
1525 }
1526 
1527 // FIXME: We use commutative matchers (m_c_*) for some, but not all, matches
1528 // here. We should standardize that construct where it is needed or choose some
1529 // other way to ensure that commutated variants of patterns are not missed.
1531  if (Value *V = SimplifyAndInst(I.getOperand(0), I.getOperand(1),
1532  SQ.getWithInstruction(&I)))
1533  return replaceInstUsesWith(I, V);
1534 
1535  if (SimplifyAssociativeOrCommutative(I))
1536  return &I;
1537 
1538  if (Instruction *X = foldVectorBinop(I))
1539  return X;
1540 
1541  // See if we can simplify any instructions used by the instruction whose sole
1542  // purpose is to compute bits we don't care about.
1543  if (SimplifyDemandedInstructionBits(I))
1544  return &I;
1545 
1546  // Do this before using distributive laws to catch simple and/or/not patterns.
1547  if (Instruction *Xor = foldAndToXor(I, Builder))
1548  return Xor;
1549 
1550  // (A|B)&(A|C) -> A|(B&C) etc
1551  if (Value *V = SimplifyUsingDistributiveLaws(I))
1552  return replaceInstUsesWith(I, V);
1553 
1554  if (Value *V = SimplifyBSwap(I, Builder))
1555  return replaceInstUsesWith(I, V);
1556 
1557  Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
1558  const APInt *C;
1559  if (match(Op1, m_APInt(C))) {
1560  Value *X, *Y;
1561  if (match(Op0, m_OneUse(m_LogicalShift(m_One(), m_Value(X)))) &&
1562  C->isOneValue()) {
1563  // (1 << X) & 1 --> zext(X == 0)
1564  // (1 >> X) & 1 --> zext(X == 0)
1565  Value *IsZero = Builder.CreateICmpEQ(X, ConstantInt::get(I.getType(), 0));
1566  return new ZExtInst(IsZero, I.getType());
1567  }
1568 
1569  const APInt *XorC;
1570  if (match(Op0, m_OneUse(m_Xor(m_Value(X), m_APInt(XorC))))) {
1571  // (X ^ C1) & C2 --> (X & C2) ^ (C1&C2)
1572  Constant *NewC = ConstantInt::get(I.getType(), *C & *XorC);
1573  Value *And = Builder.CreateAnd(X, Op1);
1574  And->takeName(Op0);
1575  return BinaryOperator::CreateXor(And, NewC);
1576  }
1577 
1578  const APInt *OrC;
1579  if (match(Op0, m_OneUse(m_Or(m_Value(X), m_APInt(OrC))))) {
1580  // (X | C1) & C2 --> (X & C2^(C1&C2)) | (C1&C2)
1581  // NOTE: This reduces the number of bits set in the & mask, which
1582  // can expose opportunities for store narrowing for scalars.
1583  // NOTE: SimplifyDemandedBits should have already removed bits from C1
1584  // that aren't set in C2. Meaning we can replace (C1&C2) with C1 in
1585  // above, but this feels safer.
1586  APInt Together = *C & *OrC;
1587  Value *And = Builder.CreateAnd(X, ConstantInt::get(I.getType(),
1588  Together ^ *C));
1589  And->takeName(Op0);
1590  return BinaryOperator::CreateOr(And, ConstantInt::get(I.getType(),
1591  Together));
1592  }
1593 
1594  // If the mask is only needed on one incoming arm, push the 'and' op up.
1595  if (match(Op0, m_OneUse(m_Xor(m_Value(X), m_Value(Y)))) ||
1596  match(Op0, m_OneUse(m_Or(m_Value(X), m_Value(Y))))) {
1597  APInt NotAndMask(~(*C));
1598  BinaryOperator::BinaryOps BinOp = cast<BinaryOperator>(Op0)->getOpcode();
1599  if (MaskedValueIsZero(X, NotAndMask, 0, &I)) {
1600  // Not masking anything out for the LHS, move mask to RHS.
1601  // and ({x}or X, Y), C --> {x}or X, (and Y, C)
1602  Value *NewRHS = Builder.CreateAnd(Y, Op1, Y->getName() + ".masked");
1603  return BinaryOperator::Create(BinOp, X, NewRHS);
1604  }
1605  if (!isa<Constant>(Y) && MaskedValueIsZero(Y, NotAndMask, 0, &I)) {
1606  // Not masking anything out for the RHS, move mask to LHS.
1607  // and ({x}or X, Y), C --> {x}or (and X, C), Y
1608  Value *NewLHS = Builder.CreateAnd(X, Op1, X->getName() + ".masked");
1609  return BinaryOperator::Create(BinOp, NewLHS, Y);
1610  }
1611  }
1612 
1613  }
1614 
1615  if (ConstantInt *AndRHS = dyn_cast<ConstantInt>(Op1)) {
1616  const APInt &AndRHSMask = AndRHS->getValue();
1617 
1618  // Optimize a variety of ((val OP C1) & C2) combinations...
1619  if (BinaryOperator *Op0I = dyn_cast<BinaryOperator>(Op0)) {
1620  // ((C1 OP zext(X)) & C2) -> zext((C1-X) & C2) if C2 fits in the bitwidth
1621  // of X and OP behaves well when given trunc(C1) and X.
1622  switch (Op0I->getOpcode()) {
1623  default:
1624  break;
1625  case Instruction::Xor:
1626  case Instruction::Or:
1627  case Instruction::Mul:
1628  case Instruction::Add:
1629  case Instruction::Sub:
1630  Value *X;
1631  ConstantInt *C1;
1632  if (match(Op0I, m_c_BinOp(m_ZExt(m_Value(X)), m_ConstantInt(C1)))) {
1633  if (AndRHSMask.isIntN(X->getType()->getScalarSizeInBits())) {
1634  auto *TruncC1 = ConstantExpr::getTrunc(C1, X->getType());
1635  Value *BinOp;
1636  Value *Op0LHS = Op0I->getOperand(0);
1637  if (isa<ZExtInst>(Op0LHS))
1638  BinOp = Builder.CreateBinOp(Op0I->getOpcode(), X, TruncC1);
1639  else
1640  BinOp = Builder.CreateBinOp(Op0I->getOpcode(), TruncC1, X);
1641  auto *TruncC2 = ConstantExpr::getTrunc(AndRHS, X->getType());
1642  auto *And = Builder.CreateAnd(BinOp, TruncC2);
1643  return new ZExtInst(And, I.getType());
1644  }
1645  }
1646  }
1647 
1648  if (ConstantInt *Op0CI = dyn_cast<ConstantInt>(Op0I->getOperand(1)))
1649  if (Instruction *Res = OptAndOp(Op0I, Op0CI, AndRHS, I))
1650  return Res;
1651  }
1652 
1653  // If this is an integer truncation, and if the source is an 'and' with
1654  // immediate, transform it. This frequently occurs for bitfield accesses.
1655  {
1656  Value *X = nullptr; ConstantInt *YC = nullptr;
1657  if (match(Op0, m_Trunc(m_And(m_Value(X), m_ConstantInt(YC))))) {
1658  // Change: and (trunc (and X, YC) to T), C2
1659  // into : and (trunc X to T), trunc(YC) & C2
1660  // This will fold the two constants together, which may allow
1661  // other simplifications.
1662  Value *NewCast = Builder.CreateTrunc(X, I.getType(), "and.shrunk");
1663  Constant *C3 = ConstantExpr::getTrunc(YC, I.getType());
1664  C3 = ConstantExpr::getAnd(C3, AndRHS);
1665  return BinaryOperator::CreateAnd(NewCast, C3);
1666  }
1667  }
1668  }
1669 
1670  if (Instruction *Z = narrowMaskedBinOp(I))
1671  return Z;
1672 
1673  if (Instruction *FoldedLogic = foldBinOpIntoSelectOrPhi(I))
1674  return FoldedLogic;
1675 
1676  if (Instruction *DeMorgan = matchDeMorgansLaws(I, Builder))
1677  return DeMorgan;
1678 
1679  {
1680  Value *A, *B, *C;
1681  // A & (A ^ B) --> A & ~B
1682  if (match(Op1, m_OneUse(m_c_Xor(m_Specific(Op0), m_Value(B)))))
1683  return BinaryOperator::CreateAnd(Op0, Builder.CreateNot(B));
1684  // (A ^ B) & A --> A & ~B
1685  if (match(Op0, m_OneUse(m_c_Xor(m_Specific(Op1), m_Value(B)))))
1686  return BinaryOperator::CreateAnd(Op1, Builder.CreateNot(B));
1687 
1688  // (A ^ B) & ((B ^ C) ^ A) -> (A ^ B) & ~C
1689  if (match(Op0, m_Xor(m_Value(A), m_Value(B))))
1690  if (match(Op1, m_Xor(m_Xor(m_Specific(B), m_Value(C)), m_Specific(A))))
1691  if (Op1->hasOneUse() || IsFreeToInvert(C, C->hasOneUse()))
1692  return BinaryOperator::CreateAnd(Op0, Builder.CreateNot(C));
1693 
1694  // ((A ^ C) ^ B) & (B ^ A) -> (B ^ A) & ~C
1695  if (match(Op0, m_Xor(m_Xor(m_Value(A), m_Value(C)), m_Value(B))))
1696  if (match(Op1, m_Xor(m_Specific(B), m_Specific(A))))
1697  if (Op0->hasOneUse() || IsFreeToInvert(C, C->hasOneUse()))
1698  return BinaryOperator::CreateAnd(Op1, Builder.CreateNot(C));
1699 
1700  // (A | B) & ((~A) ^ B) -> (A & B)
1701  // (A | B) & (B ^ (~A)) -> (A & B)
1702  // (B | A) & ((~A) ^ B) -> (A & B)
1703  // (B | A) & (B ^ (~A)) -> (A & B)
1704  if (match(Op1, m_c_Xor(m_Not(m_Value(A)), m_Value(B))) &&
1705  match(Op0, m_c_Or(m_Specific(A), m_Specific(B))))
1706  return BinaryOperator::CreateAnd(A, B);
1707 
1708  // ((~A) ^ B) & (A | B) -> (A & B)
1709  // ((~A) ^ B) & (B | A) -> (A & B)
1710  // (B ^ (~A)) & (A | B) -> (A & B)
1711  // (B ^ (~A)) & (B | A) -> (A & B)
1712  if (match(Op0, m_c_Xor(m_Not(m_Value(A)), m_Value(B))) &&
1713  match(Op1, m_c_Or(m_Specific(A), m_Specific(B))))
1714  return BinaryOperator::CreateAnd(A, B);
1715  }
1716 
1717  {
1718  ICmpInst *LHS = dyn_cast<ICmpInst>(Op0);
1719  ICmpInst *RHS = dyn_cast<ICmpInst>(Op1);
1720  if (LHS && RHS)
1721  if (Value *Res = foldAndOfICmps(LHS, RHS, I))
1722  return replaceInstUsesWith(I, Res);
1723 
1724  // TODO: Make this recursive; it's a little tricky because an arbitrary
1725  // number of 'and' instructions might have to be created.
1726  Value *X, *Y;
1727  if (LHS && match(Op1, m_OneUse(m_And(m_Value(X), m_Value(Y))))) {
1728  if (auto *Cmp = dyn_cast<ICmpInst>(X))
1729  if (Value *Res = foldAndOfICmps(LHS, Cmp, I))
1730  return replaceInstUsesWith(I, Builder.CreateAnd(Res, Y));
1731  if (auto *Cmp = dyn_cast<ICmpInst>(Y))
1732  if (Value *Res = foldAndOfICmps(LHS, Cmp, I))
1733  return replaceInstUsesWith(I, Builder.CreateAnd(Res, X));
1734  }
1735  if (RHS && match(Op0, m_OneUse(m_And(m_Value(X), m_Value(Y))))) {
1736  if (auto *Cmp = dyn_cast<ICmpInst>(X))
1737  if (Value *Res = foldAndOfICmps(Cmp, RHS, I))
1738  return replaceInstUsesWith(I, Builder.CreateAnd(Res, Y));
1739  if (auto *Cmp = dyn_cast<ICmpInst>(Y))
1740  if (Value *Res = foldAndOfICmps(Cmp, RHS, I))
1741  return replaceInstUsesWith(I, Builder.CreateAnd(Res, X));
1742  }
1743  }
1744 
1745  if (FCmpInst *LHS = dyn_cast<FCmpInst>(I.getOperand(0)))
1746  if (FCmpInst *RHS = dyn_cast<FCmpInst>(I.getOperand(1)))
1747  if (Value *Res = foldLogicOfFCmps(LHS, RHS, true))
1748  return replaceInstUsesWith(I, Res);
1749 
1750  if (Instruction *CastedAnd = foldCastedBitwiseLogic(I))
1751  return CastedAnd;
1752 
1753  // and(sext(A), B) / and(B, sext(A)) --> A ? B : 0, where A is i1 or <N x i1>.
1754  Value *A;
1755  if (match(Op0, m_OneUse(m_SExt(m_Value(A)))) &&
1756  A->getType()->isIntOrIntVectorTy(1))
1757  return SelectInst::Create(A, Op1, Constant::getNullValue(I.getType()));
1758  if (match(Op1, m_OneUse(m_SExt(m_Value(A)))) &&
1759  A->getType()->isIntOrIntVectorTy(1))
1760  return SelectInst::Create(A, Op0, Constant::getNullValue(I.getType()));
1761 
1762  return nullptr;
1763 }
1764 
1765 Instruction *InstCombiner::matchBSwap(BinaryOperator &Or) {
1766  assert(Or.getOpcode() == Instruction::Or && "bswap requires an 'or'");
1767  Value *Op0 = Or.getOperand(0), *Op1 = Or.getOperand(1);
1768 
1769  // Look through zero extends.
1770  if (Instruction *Ext = dyn_cast<ZExtInst>(Op0))
1771  Op0 = Ext->getOperand(0);
1772 
1773  if (Instruction *Ext = dyn_cast<ZExtInst>(Op1))
1774  Op1 = Ext->getOperand(0);
1775 
1776  // (A | B) | C and A | (B | C) -> bswap if possible.
1777  bool OrOfOrs = match(Op0, m_Or(m_Value(), m_Value())) ||
1778  match(Op1, m_Or(m_Value(), m_Value()));
1779 
1780  // (A >> B) | (C << D) and (A << B) | (B >> C) -> bswap if possible.
1781  bool OrOfShifts = match(Op0, m_LogicalShift(m_Value(), m_Value())) &&
1782  match(Op1, m_LogicalShift(m_Value(), m_Value()));
1783 
1784  // (A & B) | (C & D) -> bswap if possible.
1785  bool OrOfAnds = match(Op0, m_And(m_Value(), m_Value())) &&
1786  match(Op1, m_And(m_Value(), m_Value()));
1787 
1788  // (A << B) | (C & D) -> bswap if possible.
1789  // The bigger pattern here is ((A & C1) << C2) | ((B >> C2) & C1), which is a
1790  // part of the bswap idiom for specific values of C1, C2 (e.g. C1 = 16711935,
1791  // C2 = 8 for i32).
1792  // This pattern can occur when the operands of the 'or' are not canonicalized
1793  // for some reason (not having only one use, for example).
1794  bool OrOfAndAndSh = (match(Op0, m_LogicalShift(m_Value(), m_Value())) &&
1795  match(Op1, m_And(m_Value(), m_Value()))) ||
1796  (match(Op0, m_And(m_Value(), m_Value())) &&
1797  match(Op1, m_LogicalShift(m_Value(), m_Value())));
1798 
1799  if (!OrOfOrs && !OrOfShifts && !OrOfAnds && !OrOfAndAndSh)
1800  return nullptr;
1801 
1803  if (!recognizeBSwapOrBitReverseIdiom(&Or, true, false, Insts))
1804  return nullptr;
1805  Instruction *LastInst = Insts.pop_back_val();
1806  LastInst->removeFromParent();
1807 
1808  for (auto *Inst : Insts)
1809  Worklist.Add(Inst);
1810  return LastInst;
1811 }
1812 
1813 /// If all elements of two constant vectors are 0/-1 and inverses, return true.
1815  unsigned NumElts = C1->getType()->getVectorNumElements();
1816  for (unsigned i = 0; i != NumElts; ++i) {
1817  Constant *EltC1 = C1->getAggregateElement(i);
1818  Constant *EltC2 = C2->getAggregateElement(i);
1819  if (!EltC1 || !EltC2)
1820  return false;
1821 
1822  // One element must be all ones, and the other must be all zeros.
1823  if (!((match(EltC1, m_Zero()) && match(EltC2, m_AllOnes())) ||
1824  (match(EltC2, m_Zero()) && match(EltC1, m_AllOnes()))))
1825  return false;
1826  }
1827  return true;
1828 }
1829 
1830 /// We have an expression of the form (A & C) | (B & D). If A is a scalar or
1831 /// vector composed of all-zeros or all-ones values and is the bitwise 'not' of
1832 /// B, it can be used as the condition operand of a select instruction.
1833 Value *InstCombiner::getSelectCondition(Value *A, Value *B) {
1834  // Step 1: We may have peeked through bitcasts in the caller.
1835  // Exit immediately if we don't have (vector) integer types.
1836  Type *Ty = A->getType();
1837  if (!Ty->isIntOrIntVectorTy() || !B->getType()->isIntOrIntVectorTy())
1838  return nullptr;
1839 
1840  // Step 2: We need 0 or all-1's bitmasks.
1841  if (ComputeNumSignBits(A) != Ty->getScalarSizeInBits())
1842  return nullptr;
1843 
1844  // Step 3: If B is the 'not' value of A, we have our answer.
1845  if (match(A, m_Not(m_Specific(B)))) {
1846  // If these are scalars or vectors of i1, A can be used directly.
1847  if (Ty->isIntOrIntVectorTy(1))
1848  return A;
1849  return Builder.CreateTrunc(A, CmpInst::makeCmpResultType(Ty));
1850  }
1851 
1852  // If both operands are constants, see if the constants are inverse bitmasks.
1853  Constant *AConst, *BConst;
1854  if (match(A, m_Constant(AConst)) && match(B, m_Constant(BConst)))
1855  if (AConst == ConstantExpr::getNot(BConst))
1856  return Builder.CreateZExtOrTrunc(A, CmpInst::makeCmpResultType(Ty));
1857 
1858  // Look for more complex patterns. The 'not' op may be hidden behind various
1859  // casts. Look through sexts and bitcasts to find the booleans.
1860  Value *Cond;
1861  Value *NotB;
1862  if (match(A, m_SExt(m_Value(Cond))) &&
1863  Cond->getType()->isIntOrIntVectorTy(1) &&
1864  match(B, m_OneUse(m_Not(m_Value(NotB))))) {
1865  NotB = peekThroughBitcast(NotB, true);
1866  if (match(NotB, m_SExt(m_Specific(Cond))))
1867  return Cond;
1868  }
1869 
1870  // All scalar (and most vector) possibilities should be handled now.
1871  // Try more matches that only apply to non-splat constant vectors.
1872  if (!Ty->isVectorTy())
1873  return nullptr;
1874 
1875  // If both operands are xor'd with constants using the same sexted boolean
1876  // operand, see if the constants are inverse bitmasks.
1877  // TODO: Use ConstantExpr::getNot()?
1878  if (match(A, (m_Xor(m_SExt(m_Value(Cond)), m_Constant(AConst)))) &&
1879  match(B, (m_Xor(m_SExt(m_Specific(Cond)), m_Constant(BConst)))) &&
1880  Cond->getType()->isIntOrIntVectorTy(1) &&
1881  areInverseVectorBitmasks(AConst, BConst)) {
1882  AConst = ConstantExpr::getTrunc(AConst, CmpInst::makeCmpResultType(Ty));
1883  return Builder.CreateXor(Cond, AConst);
1884  }
1885  return nullptr;
1886 }
1887 
1888 /// We have an expression of the form (A & C) | (B & D). Try to simplify this
1889 /// to "A' ? C : D", where A' is a boolean or vector of booleans.
1890 Value *InstCombiner::matchSelectFromAndOr(Value *A, Value *C, Value *B,
1891  Value *D) {
1892  // The potential condition of the select may be bitcasted. In that case, look
1893  // through its bitcast and the corresponding bitcast of the 'not' condition.
1894  Type *OrigType = A->getType();
1895  A = peekThroughBitcast(A, true);
1896  B = peekThroughBitcast(B, true);
1897  if (Value *Cond = getSelectCondition(A, B)) {
1898  // ((bc Cond) & C) | ((bc ~Cond) & D) --> bc (select Cond, (bc C), (bc D))
1899  // The bitcasts will either all exist or all not exist. The builder will
1900  // not create unnecessary casts if the types already match.
1901  Value *BitcastC = Builder.CreateBitCast(C, A->getType());
1902  Value *BitcastD = Builder.CreateBitCast(D, A->getType());
1903  Value *Select = Builder.CreateSelect(Cond, BitcastC, BitcastD);
1904  return Builder.CreateBitCast(Select, OrigType);
1905  }
1906 
1907  return nullptr;
1908 }
1909 
1910 /// Fold (icmp)|(icmp) if possible.
1911 Value *InstCombiner::foldOrOfICmps(ICmpInst *LHS, ICmpInst *RHS,
1912  Instruction &CxtI) {
1913  // Fold (iszero(A & K1) | iszero(A & K2)) -> (A & (K1 | K2)) != (K1 | K2)
1914  // if K1 and K2 are a one-bit mask.
1915  if (Value *V = foldAndOrOfICmpsOfAndWithPow2(LHS, RHS, false, CxtI))
1916  return V;
1917 
1918  ICmpInst::Predicate PredL = LHS->getPredicate(), PredR = RHS->getPredicate();
1919 
1920  ConstantInt *LHSC = dyn_cast<ConstantInt>(LHS->getOperand(1));
1921  ConstantInt *RHSC = dyn_cast<ConstantInt>(RHS->getOperand(1));
1922 
1923  // Fold (icmp ult/ule (A + C1), C3) | (icmp ult/ule (A + C2), C3)
1924  // --> (icmp ult/ule ((A & ~(C1 ^ C2)) + max(C1, C2)), C3)
1925  // The original condition actually refers to the following two ranges:
1926  // [MAX_UINT-C1+1, MAX_UINT-C1+1+C3] and [MAX_UINT-C2+1, MAX_UINT-C2+1+C3]
1927  // We can fold these two ranges if:
1928  // 1) C1 and C2 is unsigned greater than C3.
1929  // 2) The two ranges are separated.
1930  // 3) C1 ^ C2 is one-bit mask.
1931  // 4) LowRange1 ^ LowRange2 and HighRange1 ^ HighRange2 are one-bit mask.
1932  // This implies all values in the two ranges differ by exactly one bit.
1933 
1934  if ((PredL == ICmpInst::ICMP_ULT || PredL == ICmpInst::ICMP_ULE) &&
1935  PredL == PredR && LHSC && RHSC && LHS->hasOneUse() && RHS->hasOneUse() &&
1936  LHSC->getType() == RHSC->getType() &&
1937  LHSC->getValue() == (RHSC->getValue())) {
1938 
1939  Value *LAdd = LHS->getOperand(0);
1940  Value *RAdd = RHS->getOperand(0);
1941 
1942  Value *LAddOpnd, *RAddOpnd;
1943  ConstantInt *LAddC, *RAddC;
1944  if (match(LAdd, m_Add(m_Value(LAddOpnd), m_ConstantInt(LAddC))) &&
1945  match(RAdd, m_Add(m_Value(RAddOpnd), m_ConstantInt(RAddC))) &&
1946  LAddC->getValue().ugt(LHSC->getValue()) &&
1947  RAddC->getValue().ugt(LHSC->getValue())) {
1948 
1949  APInt DiffC = LAddC->getValue() ^ RAddC->getValue();
1950  if (LAddOpnd == RAddOpnd && DiffC.isPowerOf2()) {
1951  ConstantInt *MaxAddC = nullptr;
1952  if (LAddC->getValue().ult(RAddC->getValue()))
1953  MaxAddC = RAddC;
1954  else
1955  MaxAddC = LAddC;
1956 
1957  APInt RRangeLow = -RAddC->getValue();
1958  APInt RRangeHigh = RRangeLow + LHSC->getValue();
1959  APInt LRangeLow = -LAddC->getValue();
1960  APInt LRangeHigh = LRangeLow + LHSC->getValue();
1961  APInt LowRangeDiff = RRangeLow ^ LRangeLow;
1962  APInt HighRangeDiff = RRangeHigh ^ LRangeHigh;
1963  APInt RangeDiff = LRangeLow.sgt(RRangeLow) ? LRangeLow - RRangeLow
1964  : RRangeLow - LRangeLow;
1965 
1966  if (LowRangeDiff.isPowerOf2() && LowRangeDiff == HighRangeDiff &&
1967  RangeDiff.ugt(LHSC->getValue())) {
1968  Value *MaskC = ConstantInt::get(LAddC->getType(), ~DiffC);
1969 
1970  Value *NewAnd = Builder.CreateAnd(LAddOpnd, MaskC);
1971  Value *NewAdd = Builder.CreateAdd(NewAnd, MaxAddC);
1972  return Builder.CreateICmp(LHS->getPredicate(), NewAdd, LHSC);
1973  }
1974  }
1975  }
1976  }
1977 
1978  // (icmp1 A, B) | (icmp2 A, B) --> (icmp3 A, B)
1979  if (predicatesFoldable(PredL, PredR)) {
1980  if (LHS->getOperand(0) == RHS->getOperand(1) &&
1981  LHS->getOperand(1) == RHS->getOperand(0))
1982  LHS->swapOperands();
1983  if (LHS->getOperand(0) == RHS->getOperand(0) &&
1984  LHS->getOperand(1) == RHS->getOperand(1)) {
1985  Value *Op0 = LHS->getOperand(0), *Op1 = LHS->getOperand(1);
1986  unsigned Code = getICmpCode(LHS) | getICmpCode(RHS);
1987  bool IsSigned = LHS->isSigned() || RHS->isSigned();
1988  return getNewICmpValue(Code, IsSigned, Op0, Op1, Builder);
1989  }
1990  }
1991 
1992  // handle (roughly):
1993  // (icmp ne (A & B), C) | (icmp ne (A & D), E)
1994  if (Value *V = foldLogOpOfMaskedICmps(LHS, RHS, false, Builder))
1995  return V;
1996 
1997  Value *LHS0 = LHS->getOperand(0), *RHS0 = RHS->getOperand(0);
1998  if (LHS->hasOneUse() || RHS->hasOneUse()) {
1999  // (icmp eq B, 0) | (icmp ult A, B) -> (icmp ule A, B-1)
2000  // (icmp eq B, 0) | (icmp ugt B, A) -> (icmp ule A, B-1)
2001  Value *A = nullptr, *B = nullptr;
2002  if (PredL == ICmpInst::ICMP_EQ && LHSC && LHSC->isZero()) {
2003  B = LHS0;
2004  if (PredR == ICmpInst::ICMP_ULT && LHS0 == RHS->getOperand(1))
2005  A = RHS0;
2006  else if (PredR == ICmpInst::ICMP_UGT && LHS0 == RHS0)
2007  A = RHS->getOperand(1);
2008  }
2009  // (icmp ult A, B) | (icmp eq B, 0) -> (icmp ule A, B-1)
2010  // (icmp ugt B, A) | (icmp eq B, 0) -> (icmp ule A, B-1)
2011  else if (PredR == ICmpInst::ICMP_EQ && RHSC && RHSC->isZero()) {
2012  B = RHS0;
2013  if (PredL == ICmpInst::ICMP_ULT && RHS0 == LHS->getOperand(1))
2014  A = LHS0;
2015  else if (PredL == ICmpInst::ICMP_UGT && LHS0 == RHS0)
2016  A = LHS->getOperand(1);
2017  }
2018  if (A && B)
2019  return Builder.CreateICmp(
2021  Builder.CreateAdd(B, ConstantInt::getSigned(B->getType(), -1)), A);
2022  }
2023 
2024  // E.g. (icmp slt x, 0) | (icmp sgt x, n) --> icmp ugt x, n
2025  if (Value *V = simplifyRangeCheck(LHS, RHS, /*Inverted=*/true))
2026  return V;
2027 
2028  // E.g. (icmp sgt x, n) | (icmp slt x, 0) --> icmp ugt x, n
2029  if (Value *V = simplifyRangeCheck(RHS, LHS, /*Inverted=*/true))
2030  return V;
2031 
2032  if (Value *V = foldAndOrOfEqualityCmpsWithConstants(LHS, RHS, false, Builder))
2033  return V;
2034 
2035  // This only handles icmp of constants: (icmp1 A, C1) | (icmp2 B, C2).
2036  if (!LHSC || !RHSC)
2037  return nullptr;
2038 
2039  if (LHSC == RHSC && PredL == PredR) {
2040  // (icmp ne A, 0) | (icmp ne B, 0) --> (icmp ne (A|B), 0)
2041  if (PredL == ICmpInst::ICMP_NE && LHSC->isZero()) {
2042  Value *NewOr = Builder.CreateOr(LHS0, RHS0);
2043  return Builder.CreateICmp(PredL, NewOr, LHSC);
2044  }
2045  }
2046 
2047  // (icmp ult (X + CA), C1) | (icmp eq X, C2) -> (icmp ule (X + CA), C1)
2048  // iff C2 + CA == C1.
2049  if (PredL == ICmpInst::ICMP_ULT && PredR == ICmpInst::ICMP_EQ) {
2050  ConstantInt *AddC;
2051  if (match(LHS0, m_Add(m_Specific(RHS0), m_ConstantInt(AddC))))
2052  if (RHSC->getValue() + AddC->getValue() == LHSC->getValue())
2053  return Builder.CreateICmpULE(LHS0, LHSC);
2054  }
2055 
2056  // From here on, we only handle:
2057  // (icmp1 A, C1) | (icmp2 A, C2) --> something simpler.
2058  if (LHS0 != RHS0)
2059  return nullptr;
2060 
2061  // ICMP_[US][GL]E X, C is folded to ICMP_[US][GL]T elsewhere.
2062  if (PredL == ICmpInst::ICMP_UGE || PredL == ICmpInst::ICMP_ULE ||
2063  PredR == ICmpInst::ICMP_UGE || PredR == ICmpInst::ICMP_ULE ||
2064  PredL == ICmpInst::ICMP_SGE || PredL == ICmpInst::ICMP_SLE ||
2065  PredR == ICmpInst::ICMP_SGE || PredR == ICmpInst::ICMP_SLE)
2066  return nullptr;
2067 
2068  // We can't fold (ugt x, C) | (sgt x, C2).
2069  if (!predicatesFoldable(PredL, PredR))
2070  return nullptr;
2071 
2072  // Ensure that the larger constant is on the RHS.
2073  bool ShouldSwap;
2074  if (CmpInst::isSigned(PredL) ||
2075  (ICmpInst::isEquality(PredL) && CmpInst::isSigned(PredR)))
2076  ShouldSwap = LHSC->getValue().sgt(RHSC->getValue());
2077  else
2078  ShouldSwap = LHSC->getValue().ugt(RHSC->getValue());
2079 
2080  if (ShouldSwap) {
2081  std::swap(LHS, RHS);
2082  std::swap(LHSC, RHSC);
2083  std::swap(PredL, PredR);
2084  }
2085 
2086  // At this point, we know we have two icmp instructions
2087  // comparing a value against two constants and or'ing the result
2088  // together. Because of the above check, we know that we only have
2089  // ICMP_EQ, ICMP_NE, ICMP_LT, and ICMP_GT here. We also know (from the
2090  // icmp folding check above), that the two constants are not
2091  // equal.
2092  assert(LHSC != RHSC && "Compares not folded above?");
2093 
2094  switch (PredL) {
2095  default:
2096  llvm_unreachable("Unknown integer condition code!");
2097  case ICmpInst::ICMP_EQ:
2098  switch (PredR) {
2099  default:
2100  llvm_unreachable("Unknown integer condition code!");
2101  case ICmpInst::ICMP_EQ:
2102  // Potential folds for this case should already be handled.
2103  break;
2104  case ICmpInst::ICMP_UGT: // (X == 13 | X u> 14) -> no change
2105  case ICmpInst::ICMP_SGT: // (X == 13 | X s> 14) -> no change
2106  break;
2107  }
2108  break;
2109  case ICmpInst::ICMP_ULT:
2110  switch (PredR) {
2111  default:
2112  llvm_unreachable("Unknown integer condition code!");
2113  case ICmpInst::ICMP_EQ: // (X u< 13 | X == 14) -> no change
2114  break;
2115  case ICmpInst::ICMP_UGT: // (X u< 13 | X u> 15) -> (X-13) u> 2
2116  assert(!RHSC->isMaxValue(false) && "Missed icmp simplification");
2117  return insertRangeTest(LHS0, LHSC->getValue(), RHSC->getValue() + 1,
2118  false, false);
2119  }
2120  break;
2121  case ICmpInst::ICMP_SLT:
2122  switch (PredR) {
2123  default:
2124  llvm_unreachable("Unknown integer condition code!");
2125  case ICmpInst::ICMP_EQ: // (X s< 13 | X == 14) -> no change
2126  break;
2127  case ICmpInst::ICMP_SGT: // (X s< 13 | X s> 15) -> (X-13) s> 2
2128  assert(!RHSC->isMaxValue(true) && "Missed icmp simplification");
2129  return insertRangeTest(LHS0, LHSC->getValue(), RHSC->getValue() + 1, true,
2130  false);
2131  }
2132  break;
2133  }
2134  return nullptr;
2135 }
2136 
2137 // FIXME: We use commutative matchers (m_c_*) for some, but not all, matches
2138 // here. We should standardize that construct where it is needed or choose some
2139 // other way to ensure that commutated variants of patterns are not missed.
2141  if (Value *V = SimplifyOrInst(I.getOperand(0), I.getOperand(1),
2142  SQ.getWithInstruction(&I)))
2143  return replaceInstUsesWith(I, V);
2144 
2145  if (SimplifyAssociativeOrCommutative(I))
2146  return &I;
2147 
2148  if (Instruction *X = foldVectorBinop(I))
2149  return X;
2150 
2151  // See if we can simplify any instructions used by the instruction whose sole
2152  // purpose is to compute bits we don't care about.
2153  if (SimplifyDemandedInstructionBits(I))
2154  return &I;
2155 
2156  // Do this before using distributive laws to catch simple and/or/not patterns.
2157  if (Instruction *Xor = foldOrToXor(I, Builder))
2158  return Xor;
2159 
2160  // (A&B)|(A&C) -> A&(B|C) etc
2161  if (Value *V = SimplifyUsingDistributiveLaws(I))
2162  return replaceInstUsesWith(I, V);
2163 
2164  if (Value *V = SimplifyBSwap(I, Builder))
2165  return replaceInstUsesWith(I, V);
2166 
2167  if (Instruction *FoldedLogic = foldBinOpIntoSelectOrPhi(I))
2168  return FoldedLogic;
2169 
2170  if (Instruction *BSwap = matchBSwap(I))
2171  return BSwap;
2172 
2173  Value *X, *Y;
2174  const APInt *CV;
2175  if (match(&I, m_c_Or(m_OneUse(m_Xor(m_Value(X), m_APInt(CV))), m_Value(Y))) &&
2176  !CV->isAllOnesValue() && MaskedValueIsZero(Y, *CV, 0, &I)) {
2177  // (X ^ C) | Y -> (X | Y) ^ C iff Y & C == 0
2178  // The check for a 'not' op is for efficiency (if Y is known zero --> ~X).
2179  Value *Or = Builder.CreateOr(X, Y);
2180  return BinaryOperator::CreateXor(Or, ConstantInt::get(I.getType(), *CV));
2181  }
2182 
2183  // (A & C)|(B & D)
2184  Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
2185  Value *A, *B, *C, *D;
2186  if (match(Op0, m_And(m_Value(A), m_Value(C))) &&
2187  match(Op1, m_And(m_Value(B), m_Value(D)))) {
2190  if (C1 && C2) { // (A & C1)|(B & C2)
2191  Value *V1 = nullptr, *V2 = nullptr;
2192  if ((C1->getValue() & C2->getValue()).isNullValue()) {
2193  // ((V | N) & C1) | (V & C2) --> (V|N) & (C1|C2)
2194  // iff (C1&C2) == 0 and (N&~C1) == 0
2195  if (match(A, m_Or(m_Value(V1), m_Value(V2))) &&
2196  ((V1 == B &&
2197  MaskedValueIsZero(V2, ~C1->getValue(), 0, &I)) || // (V|N)
2198  (V2 == B &&
2199  MaskedValueIsZero(V1, ~C1->getValue(), 0, &I)))) // (N|V)
2200  return BinaryOperator::CreateAnd(A,
2201  Builder.getInt(C1->getValue()|C2->getValue()));
2202  // Or commutes, try both ways.
2203  if (match(B, m_Or(m_Value(V1), m_Value(V2))) &&
2204  ((V1 == A &&
2205  MaskedValueIsZero(V2, ~C2->getValue(), 0, &I)) || // (V|N)
2206  (V2 == A &&
2207  MaskedValueIsZero(V1, ~C2->getValue(), 0, &I)))) // (N|V)
2208  return BinaryOperator::CreateAnd(B,
2209  Builder.getInt(C1->getValue()|C2->getValue()));
2210 
2211  // ((V|C3)&C1) | ((V|C4)&C2) --> (V|C3|C4)&(C1|C2)
2212  // iff (C1&C2) == 0 and (C3&~C1) == 0 and (C4&~C2) == 0.
2213  ConstantInt *C3 = nullptr, *C4 = nullptr;
2214  if (match(A, m_Or(m_Value(V1), m_ConstantInt(C3))) &&
2215  (C3->getValue() & ~C1->getValue()).isNullValue() &&
2216  match(B, m_Or(m_Specific(V1), m_ConstantInt(C4))) &&
2217  (C4->getValue() & ~C2->getValue()).isNullValue()) {
2218  V2 = Builder.CreateOr(V1, ConstantExpr::getOr(C3, C4), "bitfield");
2219  return BinaryOperator::CreateAnd(V2,
2220  Builder.getInt(C1->getValue()|C2->getValue()));
2221  }
2222  }
2223 
2224  if (C1->getValue() == ~C2->getValue()) {
2225  Value *X;
2226 
2227  // ((X|B)&C1)|(B&C2) -> (X&C1) | B iff C1 == ~C2
2228  if (match(A, m_c_Or(m_Value(X), m_Specific(B))))
2229  return BinaryOperator::CreateOr(Builder.CreateAnd(X, C1), B);
2230  // (A&C2)|((X|A)&C1) -> (X&C2) | A iff C1 == ~C2
2231  if (match(B, m_c_Or(m_Specific(A), m_Value(X))))
2232  return BinaryOperator::CreateOr(Builder.CreateAnd(X, C2), A);
2233 
2234  // ((X^B)&C1)|(B&C2) -> (X&C1) ^ B iff C1 == ~C2
2235  if (match(A, m_c_Xor(m_Value(X), m_Specific(B))))
2236  return BinaryOperator::CreateXor(Builder.CreateAnd(X, C1), B);
2237  // (A&C2)|((X^A)&C1) -> (X&C2) ^ A iff C1 == ~C2
2238  if (match(B, m_c_Xor(m_Specific(A), m_Value(X))))
2239  return BinaryOperator::CreateXor(Builder.CreateAnd(X, C2), A);
2240  }
2241  }
2242 
2243  // Don't try to form a select if it's unlikely that we'll get rid of at
2244  // least one of the operands. A select is generally more expensive than the
2245  // 'or' that it is replacing.
2246  if (Op0->hasOneUse() || Op1->hasOneUse()) {
2247  // (Cond & C) | (~Cond & D) -> Cond ? C : D, and commuted variants.
2248  if (Value *V = matchSelectFromAndOr(A, C, B, D))
2249  return replaceInstUsesWith(I, V);
2250  if (Value *V = matchSelectFromAndOr(A, C, D, B))
2251  return replaceInstUsesWith(I, V);
2252  if (Value *V = matchSelectFromAndOr(C, A, B, D))
2253  return replaceInstUsesWith(I, V);
2254  if (Value *V = matchSelectFromAndOr(C, A, D, B))
2255  return replaceInstUsesWith(I, V);
2256  if (Value *V = matchSelectFromAndOr(B, D, A, C))
2257  return replaceInstUsesWith(I, V);
2258  if (Value *V = matchSelectFromAndOr(B, D, C, A))
2259  return replaceInstUsesWith(I, V);
2260  if (Value *V = matchSelectFromAndOr(D, B, A, C))
2261  return replaceInstUsesWith(I, V);
2262  if (Value *V = matchSelectFromAndOr(D, B, C, A))
2263  return replaceInstUsesWith(I, V);
2264  }
2265  }
2266 
2267  // (A ^ B) | ((B ^ C) ^ A) -> (A ^ B) | C
2268  if (match(Op0, m_Xor(m_Value(A), m_Value(B))))
2269  if (match(Op1, m_Xor(m_Xor(m_Specific(B), m_Value(C)), m_Specific(A))))
2270  return BinaryOperator::CreateOr(Op0, C);
2271 
2272  // ((A ^ C) ^ B) | (B ^ A) -> (B ^ A) | C
2273  if (match(Op0, m_Xor(m_Xor(m_Value(A), m_Value(C)), m_Value(B))))
2274  if (match(Op1, m_Xor(m_Specific(B), m_Specific(A))))
2275  return BinaryOperator::CreateOr(Op1, C);
2276 
2277  // ((B | C) & A) | B -> B | (A & C)
2278  if (match(Op0, m_And(m_Or(m_Specific(Op1), m_Value(C)), m_Value(A))))
2279  return BinaryOperator::CreateOr(Op1, Builder.CreateAnd(A, C));
2280 
2281  if (Instruction *DeMorgan = matchDeMorgansLaws(I, Builder))
2282  return DeMorgan;
2283 
2284  // Canonicalize xor to the RHS.
2285  bool SwappedForXor = false;
2286  if (match(Op0, m_Xor(m_Value(), m_Value()))) {
2287  std::swap(Op0, Op1);
2288  SwappedForXor = true;
2289  }
2290 
2291  // A | ( A ^ B) -> A | B
2292  // A | (~A ^ B) -> A | ~B
2293  // (A & B) | (A ^ B)
2294  if (match(Op1, m_Xor(m_Value(A), m_Value(B)))) {
2295  if (Op0 == A || Op0 == B)
2296  return BinaryOperator::CreateOr(A, B);
2297 
2298  if (match(Op0, m_And(m_Specific(A), m_Specific(B))) ||
2299  match(Op0, m_And(m_Specific(B), m_Specific(A))))
2300  return BinaryOperator::CreateOr(A, B);
2301 
2302  if (Op1->hasOneUse() && match(A, m_Not(m_Specific(Op0)))) {
2303  Value *Not = Builder.CreateNot(B, B->getName() + ".not");
2304  return BinaryOperator::CreateOr(Not, Op0);
2305  }
2306  if (Op1->hasOneUse() && match(B, m_Not(m_Specific(Op0)))) {
2307  Value *Not = Builder.CreateNot(A, A->getName() + ".not");
2308  return BinaryOperator::CreateOr(Not, Op0);
2309  }
2310  }
2311 
2312  // A | ~(A | B) -> A | ~B
2313  // A | ~(A ^ B) -> A | ~B
2314  if (match(Op1, m_Not(m_Value(A))))
2315  if (BinaryOperator *B = dyn_cast<BinaryOperator>(A))
2316  if ((Op0 == B->getOperand(0) || Op0 == B->getOperand(1)) &&
2317  Op1->hasOneUse() && (B->getOpcode() == Instruction::Or ||
2318  B->getOpcode() == Instruction::Xor)) {
2319  Value *NotOp = Op0 == B->getOperand(0) ? B->getOperand(1) :
2320  B->getOperand(0);
2321  Value *Not = Builder.CreateNot(NotOp, NotOp->getName() + ".not");
2322  return BinaryOperator::CreateOr(Not, Op0);
2323  }
2324 
2325  if (SwappedForXor)
2326  std::swap(Op0, Op1);
2327 
2328  {
2329  ICmpInst *LHS = dyn_cast<ICmpInst>(Op0);
2330  ICmpInst *RHS = dyn_cast<ICmpInst>(Op1);
2331  if (LHS && RHS)
2332  if (Value *Res = foldOrOfICmps(LHS, RHS, I))
2333  return replaceInstUsesWith(I, Res);
2334 
2335  // TODO: Make this recursive; it's a little tricky because an arbitrary
2336  // number of 'or' instructions might have to be created.
2337  Value *X, *Y;
2338  if (LHS && match(Op1, m_OneUse(m_Or(m_Value(X), m_Value(Y))))) {
2339  if (auto *Cmp = dyn_cast<ICmpInst>(X))
2340  if (Value *Res = foldOrOfICmps(LHS, Cmp, I))
2341  return replaceInstUsesWith(I, Builder.CreateOr(Res, Y));
2342  if (auto *Cmp = dyn_cast<ICmpInst>(Y))
2343  if (Value *Res = foldOrOfICmps(LHS, Cmp, I))
2344  return replaceInstUsesWith(I, Builder.CreateOr(Res, X));
2345  }
2346  if (RHS && match(Op0, m_OneUse(m_Or(m_Value(X), m_Value(Y))))) {
2347  if (auto *Cmp = dyn_cast<ICmpInst>(X))
2348  if (Value *Res = foldOrOfICmps(Cmp, RHS, I))
2349  return replaceInstUsesWith(I, Builder.CreateOr(Res, Y));
2350  if (auto *Cmp = dyn_cast<ICmpInst>(Y))
2351  if (Value *Res = foldOrOfICmps(Cmp, RHS, I))
2352  return replaceInstUsesWith(I, Builder.CreateOr(Res, X));
2353  }
2354  }
2355 
2356  if (FCmpInst *LHS = dyn_cast<FCmpInst>(I.getOperand(0)))
2357  if (FCmpInst *RHS = dyn_cast<FCmpInst>(I.getOperand(1)))
2358  if (Value *Res = foldLogicOfFCmps(LHS, RHS, false))
2359  return replaceInstUsesWith(I, Res);
2360 
2361  if (Instruction *CastedOr = foldCastedBitwiseLogic(I))
2362  return CastedOr;
2363 
2364  // or(sext(A), B) / or(B, sext(A)) --> A ? -1 : B, where A is i1 or <N x i1>.
2365  if (match(Op0, m_OneUse(m_SExt(m_Value(A)))) &&
2366  A->getType()->isIntOrIntVectorTy(1))
2367  return SelectInst::Create(A, ConstantInt::getSigned(I.getType(), -1), Op1);
2368  if (match(Op1, m_OneUse(m_SExt(m_Value(A)))) &&
2369  A->getType()->isIntOrIntVectorTy(1))
2370  return SelectInst::Create(A, ConstantInt::getSigned(I.getType(), -1), Op0);
2371 
2372  // Note: If we've gotten to the point of visiting the outer OR, then the
2373  // inner one couldn't be simplified. If it was a constant, then it won't
2374  // be simplified by a later pass either, so we try swapping the inner/outer
2375  // ORs in the hopes that we'll be able to simplify it this way.
2376  // (X|C) | V --> (X|V) | C
2377  ConstantInt *CI;
2378  if (Op0->hasOneUse() && !isa<ConstantInt>(Op1) &&
2379  match(Op0, m_Or(m_Value(A), m_ConstantInt(CI)))) {
2380  Value *Inner = Builder.CreateOr(A, Op1);
2381  Inner->takeName(Op0);
2382  return BinaryOperator::CreateOr(Inner, CI);
2383  }
2384 
2385  // Change (or (bool?A:B),(bool?C:D)) --> (bool?(or A,C):(or B,D))
2386  // Since this OR statement hasn't been optimized further yet, we hope
2387  // that this transformation will allow the new ORs to be optimized.
2388  {
2389  Value *X = nullptr, *Y = nullptr;
2390  if (Op0->hasOneUse() && Op1->hasOneUse() &&
2391  match(Op0, m_Select(m_Value(X), m_Value(A), m_Value(B))) &&
2392  match(Op1, m_Select(m_Value(Y), m_Value(C), m_Value(D))) && X == Y) {
2393  Value *orTrue = Builder.CreateOr(A, C);
2394  Value *orFalse = Builder.CreateOr(B, D);
2395  return SelectInst::Create(X, orTrue, orFalse);
2396  }
2397  }
2398 
2399  return nullptr;
2400 }
2401 
2402 /// A ^ B can be specified using other logic ops in a variety of patterns. We
2403 /// can fold these early and efficiently by morphing an existing instruction.
2405  InstCombiner::BuilderTy &Builder) {
2406  assert(I.getOpcode() == Instruction::Xor);
2407  Value *Op0 = I.getOperand(0);
2408  Value *Op1 = I.getOperand(1);
2409  Value *A, *B;
2410 
2411  // There are 4 commuted variants for each of the basic patterns.
2412 
2413  // (A & B) ^ (A | B) -> A ^ B
2414  // (A & B) ^ (B | A) -> A ^ B
2415  // (A | B) ^ (A & B) -> A ^ B
2416  // (A | B) ^ (B & A) -> A ^ B
2417  if (match(&I, m_c_Xor(m_And(m_Value(A), m_Value(B)),
2418  m_c_Or(m_Deferred(A), m_Deferred(B))))) {
2419  I.setOperand(0, A);
2420  I.setOperand(1, B);
2421  return &I;
2422  }
2423 
2424  // (A | ~B) ^ (~A | B) -> A ^ B
2425  // (~B | A) ^ (~A | B) -> A ^ B
2426  // (~A | B) ^ (A | ~B) -> A ^ B
2427  // (B | ~A) ^ (A | ~B) -> A ^ B
2428  if (match(&I, m_Xor(m_c_Or(m_Value(A), m_Not(m_Value(B))),
2429  m_c_Or(m_Not(m_Deferred(A)), m_Deferred(B))))) {
2430  I.setOperand(0, A);
2431  I.setOperand(1, B);
2432  return &I;
2433  }
2434 
2435  // (A & ~B) ^ (~A & B) -> A ^ B
2436  // (~B & A) ^ (~A & B) -> A ^ B
2437  // (~A & B) ^ (A & ~B) -> A ^ B
2438  // (B & ~A) ^ (A & ~B) -> A ^ B
2439  if (match(&I, m_Xor(m_c_And(m_Value(A), m_Not(m_Value(B))),
2440  m_c_And(m_Not(m_Deferred(A)), m_Deferred(B))))) {
2441  I.setOperand(0, A);
2442  I.setOperand(1, B);
2443  return &I;
2444  }
2445 
2446  // For the remaining cases we need to get rid of one of the operands.
2447  if (!Op0->hasOneUse() && !Op1->hasOneUse())
2448  return nullptr;
2449 
2450  // (A | B) ^ ~(A & B) -> ~(A ^ B)
2451  // (A | B) ^ ~(B & A) -> ~(A ^ B)
2452  // (A & B) ^ ~(A | B) -> ~(A ^ B)
2453  // (A & B) ^ ~(B | A) -> ~(A ^ B)
2454  // Complexity sorting ensures the not will be on the right side.
2455  if ((match(Op0, m_Or(m_Value(A), m_Value(B))) &&
2456  match(Op1, m_Not(m_c_And(m_Specific(A), m_Specific(B))))) ||
2457  (match(Op0, m_And(m_Value(A), m_Value(B))) &&
2458  match(Op1, m_Not(m_c_Or(m_Specific(A), m_Specific(B))))))
2459  return BinaryOperator::CreateNot(Builder.CreateXor(A, B));
2460 
2461  return nullptr;
2462 }
2463 
2464 Value *InstCombiner::foldXorOfICmps(ICmpInst *LHS, ICmpInst *RHS) {
2465  if (predicatesFoldable(LHS->getPredicate(), RHS->getPredicate())) {
2466  if (LHS->getOperand(0) == RHS->getOperand(1) &&
2467  LHS->getOperand(1) == RHS->getOperand(0))
2468  LHS->swapOperands();
2469  if (LHS->getOperand(0) == RHS->getOperand(0) &&
2470  LHS->getOperand(1) == RHS->getOperand(1)) {
2471  // (icmp1 A, B) ^ (icmp2 A, B) --> (icmp3 A, B)
2472  Value *Op0 = LHS->getOperand(0), *Op1 = LHS->getOperand(1);
2473  unsigned Code = getICmpCode(LHS) ^ getICmpCode(RHS);
2474  bool IsSigned = LHS->isSigned() || RHS->isSigned();
2475  return getNewICmpValue(Code, IsSigned, Op0, Op1, Builder);
2476  }
2477  }
2478 
2479  // TODO: This can be generalized to compares of non-signbits using
2480  // decomposeBitTestICmp(). It could be enhanced more by using (something like)
2481  // foldLogOpOfMaskedICmps().
2482  ICmpInst::Predicate PredL = LHS->getPredicate(), PredR = RHS->getPredicate();
2483  Value *LHS0 = LHS->getOperand(0), *LHS1 = LHS->getOperand(1);
2484  Value *RHS0 = RHS->getOperand(0), *RHS1 = RHS->getOperand(1);
2485  if ((LHS->hasOneUse() || RHS->hasOneUse()) &&
2486  LHS0->getType() == RHS0->getType() &&
2487  LHS0->getType()->isIntOrIntVectorTy()) {
2488  // (X > -1) ^ (Y > -1) --> (X ^ Y) < 0
2489  // (X < 0) ^ (Y < 0) --> (X ^ Y) < 0
2490  if ((PredL == CmpInst::ICMP_SGT && match(LHS1, m_AllOnes()) &&
2491  PredR == CmpInst::ICMP_SGT && match(RHS1, m_AllOnes())) ||
2492  (PredL == CmpInst::ICMP_SLT && match(LHS1, m_Zero()) &&
2493  PredR == CmpInst::ICMP_SLT && match(RHS1, m_Zero()))) {
2494  Value *Zero = ConstantInt::getNullValue(LHS0->getType());
2495  return Builder.CreateICmpSLT(Builder.CreateXor(LHS0, RHS0), Zero);
2496  }
2497  // (X > -1) ^ (Y < 0) --> (X ^ Y) > -1
2498  // (X < 0) ^ (Y > -1) --> (X ^ Y) > -1
2499  if ((PredL == CmpInst::ICMP_SGT && match(LHS1, m_AllOnes()) &&
2500  PredR == CmpInst::ICMP_SLT && match(RHS1, m_Zero())) ||
2501  (PredL == CmpInst::ICMP_SLT && match(LHS1, m_Zero()) &&
2502  PredR == CmpInst::ICMP_SGT && match(RHS1, m_AllOnes()))) {
2503  Value *MinusOne = ConstantInt::getAllOnesValue(LHS0->getType());
2504  return Builder.CreateICmpSGT(Builder.CreateXor(LHS0, RHS0), MinusOne);
2505  }
2506  }
2507 
2508  // Instead of trying to imitate the folds for and/or, decompose this 'xor'
2509  // into those logic ops. That is, try to turn this into an and-of-icmps
2510  // because we have many folds for that pattern.
2511  //
2512  // This is based on a truth table definition of xor:
2513  // X ^ Y --> (X | Y) & !(X & Y)
2514  if (Value *OrICmp = SimplifyBinOp(Instruction::Or, LHS, RHS, SQ)) {
2515  // TODO: If OrICmp is true, then the definition of xor simplifies to !(X&Y).
2516  // TODO: If OrICmp is false, the whole thing is false (InstSimplify?).
2517  if (Value *AndICmp = SimplifyBinOp(Instruction::And, LHS, RHS, SQ)) {
2518  // TODO: Independently handle cases where the 'and' side is a constant.
2519  if (OrICmp == LHS && AndICmp == RHS && RHS->hasOneUse()) {
2520  // (LHS | RHS) & !(LHS & RHS) --> LHS & !RHS
2521  RHS->setPredicate(RHS->getInversePredicate());
2522  return Builder.CreateAnd(LHS, RHS);
2523  }
2524  if (OrICmp == RHS && AndICmp == LHS && LHS->hasOneUse()) {
2525  // !(LHS & RHS) & (LHS | RHS) --> !LHS & RHS
2526  LHS->setPredicate(LHS->getInversePredicate());
2527  return Builder.CreateAnd(LHS, RHS);
2528  }
2529  }
2530  }
2531 
2532  return nullptr;
2533 }
2534 
2535 /// If we have a masked merge, in the canonical form of:
2536 /// (assuming that A only has one use.)
2537 /// | A | |B|
2538 /// ((x ^ y) & M) ^ y
2539 /// | D |
2540 /// * If M is inverted:
2541 /// | D |
2542 /// ((x ^ y) & ~M) ^ y
2543 /// We can canonicalize by swapping the final xor operand
2544 /// to eliminate the 'not' of the mask.
2545 /// ((x ^ y) & M) ^ x
2546 /// * If M is a constant, and D has one use, we transform to 'and' / 'or' ops
2547 /// because that shortens the dependency chain and improves analysis:
2548 /// (x & M) | (y & ~M)
2550  InstCombiner::BuilderTy &Builder) {
2551  Value *B, *X, *D;
2552  Value *M;
2553  if (!match(&I, m_c_Xor(m_Value(B),
2554  m_OneUse(m_c_And(
2556  m_Value(D)),
2557  m_Value(M))))))
2558  return nullptr;
2559 
2560  Value *NotM;
2561  if (match(M, m_Not(m_Value(NotM)))) {
2562  // De-invert the mask and swap the value in B part.
2563  Value *NewA = Builder.CreateAnd(D, NotM);
2564  return BinaryOperator::CreateXor(NewA, X);
2565  }
2566 
2567  Constant *C;
2568  if (D->hasOneUse() && match(M, m_Constant(C))) {
2569  // Unfold.
2570  Value *LHS = Builder.CreateAnd(X, C);
2571  Value *NotC = Builder.CreateNot(C);
2572  Value *RHS = Builder.CreateAnd(B, NotC);
2573  return BinaryOperator::CreateOr(LHS, RHS);
2574  }
2575 
2576  return nullptr;
2577 }
2578 
2579 // Transform
2580 // ~(x ^ y)
2581 // into:
2582 // (~x) ^ y
2583 // or into
2584 // x ^ (~y)
2586  InstCombiner::BuilderTy &Builder) {
2587  Value *X, *Y;
2588  // FIXME: one-use check is not needed in general, but currently we are unable
2589  // to fold 'not' into 'icmp', if that 'icmp' has multiple uses. (D35182)
2590  if (!match(&I, m_Not(m_OneUse(m_Xor(m_Value(X), m_Value(Y))))))
2591  return nullptr;
2592 
2593  // We only want to do the transform if it is free to do.
2594  if (IsFreeToInvert(X, X->hasOneUse())) {
2595  // Ok, good.
2596  } else if (IsFreeToInvert(Y, Y->hasOneUse())) {
2597  std::swap(X, Y);
2598  } else
2599  return nullptr;
2600 
2601  Value *NotX = Builder.CreateNot(X, X->getName() + ".not");
2602  return BinaryOperator::CreateXor(NotX, Y, I.getName() + ".demorgan");
2603 }
2604 
2605 // FIXME: We use commutative matchers (m_c_*) for some, but not all, matches
2606 // here. We should standardize that construct where it is needed or choose some
2607 // other way to ensure that commutated variants of patterns are not missed.
2609  if (Value *V = SimplifyXorInst(I.getOperand(0), I.getOperand(1),
2610  SQ.getWithInstruction(&I)))
2611  return replaceInstUsesWith(I, V);
2612 
2613  if (SimplifyAssociativeOrCommutative(I))
2614  return &I;
2615 
2616  if (Instruction *X = foldVectorBinop(I))
2617  return X;
2618 
2619  if (Instruction *NewXor = foldXorToXor(I, Builder))
2620  return NewXor;
2621 
2622  // (A&B)^(A&C) -> A&(B^C) etc
2623  if (Value *V = SimplifyUsingDistributiveLaws(I))
2624  return replaceInstUsesWith(I, V);
2625 
2626  // See if we can simplify any instructions used by the instruction whose sole
2627  // purpose is to compute bits we don't care about.
2628  if (SimplifyDemandedInstructionBits(I))
2629  return &I;
2630 
2631  if (Value *V = SimplifyBSwap(I, Builder))
2632  return replaceInstUsesWith(I, V);
2633 
2634  Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
2635 
2636  // Fold (X & M) ^ (Y & ~M) -> (X & M) | (Y & ~M)
2637  // This it a special case in haveNoCommonBitsSet, but the computeKnownBits
2638  // calls in there are unnecessary as SimplifyDemandedInstructionBits should
2639  // have already taken care of those cases.
2640  Value *M;
2641  if (match(&I, m_c_Xor(m_c_And(m_Not(m_Value(M)), m_Value()),
2642  m_c_And(m_Deferred(M), m_Value()))))
2643  return BinaryOperator::CreateOr(Op0, Op1);
2644 
2645  // Apply DeMorgan's Law for 'nand' / 'nor' logic with an inverted operand.
2646  Value *X, *Y;
2647 
2648  // We must eliminate the and/or (one-use) for these transforms to not increase
2649  // the instruction count.
2650  // ~(~X & Y) --> (X | ~Y)
2651  // ~(Y & ~X) --> (X | ~Y)
2652  if (match(&I, m_Not(m_OneUse(m_c_And(m_Not(m_Value(X)), m_Value(Y)))))) {
2653  Value *NotY = Builder.CreateNot(Y, Y->getName() + ".not");
2654  return BinaryOperator::CreateOr(X, NotY);
2655  }
2656  // ~(~X | Y) --> (X & ~Y)
2657  // ~(Y | ~X) --> (X & ~Y)
2658  if (match(&I, m_Not(m_OneUse(m_c_Or(m_Not(m_Value(X)), m_Value(Y)))))) {
2659  Value *NotY = Builder.CreateNot(Y, Y->getName() + ".not");
2660  return BinaryOperator::CreateAnd(X, NotY);
2661  }
2662 
2663  if (Instruction *Xor = visitMaskedMerge(I, Builder))
2664  return Xor;
2665 
2666  // Is this a 'not' (~) fed by a binary operator?
2667  BinaryOperator *NotVal;
2668  if (match(&I, m_Not(m_BinOp(NotVal)))) {
2669  if (NotVal->getOpcode() == Instruction::And ||
2670  NotVal->getOpcode() == Instruction::Or) {
2671  // Apply DeMorgan's Law when inverts are free:
2672  // ~(X & Y) --> (~X | ~Y)
2673  // ~(X | Y) --> (~X & ~Y)
2674  if (IsFreeToInvert(NotVal->getOperand(0),
2675  NotVal->getOperand(0)->hasOneUse()) &&
2676  IsFreeToInvert(NotVal->getOperand(1),
2677  NotVal->getOperand(1)->hasOneUse())) {
2678  Value *NotX = Builder.CreateNot(NotVal->getOperand(0), "notlhs");
2679  Value *NotY = Builder.CreateNot(NotVal->getOperand(1), "notrhs");
2680  if (NotVal->getOpcode() == Instruction::And)
2681  return BinaryOperator::CreateOr(NotX, NotY);
2682  return BinaryOperator::CreateAnd(NotX, NotY);
2683  }
2684  }
2685 
2686  // ~(X - Y) --> ~X + Y
2687  if (match(NotVal, m_Sub(m_Value(X), m_Value(Y))))
2688  if (isa<Constant>(X) || NotVal->hasOneUse())
2689  return BinaryOperator::CreateAdd(Builder.CreateNot(X), Y);
2690 
2691  // ~(~X >>s Y) --> (X >>s Y)
2692  if (match(NotVal, m_AShr(m_Not(m_Value(X)), m_Value(Y))))
2693  return BinaryOperator::CreateAShr(X, Y);
2694 
2695  // If we are inverting a right-shifted constant, we may be able to eliminate
2696  // the 'not' by inverting the constant and using the opposite shift type.
2697  // Canonicalization rules ensure that only a negative constant uses 'ashr',
2698  // but we must check that in case that transform has not fired yet.
2699 
2700  // ~(C >>s Y) --> ~C >>u Y (when inverting the replicated sign bits)
2701  Constant *C;
2702  if (match(NotVal, m_AShr(m_Constant(C), m_Value(Y))) &&
2703  match(C, m_Negative()))
2704  return BinaryOperator::CreateLShr(ConstantExpr::getNot(C), Y);
2705 
2706  // ~(C >>u Y) --> ~C >>s Y (when inverting the replicated sign bits)
2707  if (match(NotVal, m_LShr(m_Constant(C), m_Value(Y))) &&
2708  match(C, m_NonNegative()))
2709  return BinaryOperator::CreateAShr(ConstantExpr::getNot(C), Y);
2710 
2711  // ~(X + C) --> -(C + 1) - X
2712  if (match(Op0, m_Add(m_Value(X), m_Constant(C))))
2713  return BinaryOperator::CreateSub(ConstantExpr::getNeg(AddOne(C)), X);
2714  }
2715 
2716  // Use DeMorgan and reassociation to eliminate a 'not' op.
2717  Constant *C1;
2718  if (match(Op1, m_Constant(C1))) {
2719  Constant *C2;
2720  if (match(Op0, m_OneUse(m_Or(m_Not(m_Value(X)), m_Constant(C2))))) {
2721  // (~X | C2) ^ C1 --> ((X & ~C2) ^ -1) ^ C1 --> (X & ~C2) ^ ~C1
2722  Value *And = Builder.CreateAnd(X, ConstantExpr::getNot(C2));
2723  return BinaryOperator::CreateXor(And, ConstantExpr::getNot(C1));
2724  }
2725  if (match(Op0, m_OneUse(m_And(m_Not(m_Value(X)), m_Constant(C2))))) {
2726  // (~X & C2) ^ C1 --> ((X | ~C2) ^ -1) ^ C1 --> (X | ~C2) ^ ~C1
2727  Value *Or = Builder.CreateOr(X, ConstantExpr::getNot(C2));
2728  return BinaryOperator::CreateXor(Or, ConstantExpr::getNot(C1));
2729  }
2730  }
2731 
2732  // not (cmp A, B) = !cmp A, B
2733  CmpInst::Predicate Pred;
2734  if (match(&I, m_Not(m_OneUse(m_Cmp(Pred, m_Value(), m_Value()))))) {
2735  cast<CmpInst>(Op0)->setPredicate(CmpInst::getInversePredicate(Pred));
2736  return replaceInstUsesWith(I, Op0);
2737  }
2738 
2739  {
2740  const APInt *RHSC;
2741  if (match(Op1, m_APInt(RHSC))) {
2742  Value *X;
2743  const APInt *C;
2744  if (RHSC->isSignMask() && match(Op0, m_Sub(m_APInt(C), m_Value(X)))) {
2745  // (C - X) ^ signmask -> (C + signmask - X)
2746  Constant *NewC = ConstantInt::get(I.getType(), *C + *RHSC);
2747  return BinaryOperator::CreateSub(NewC, X);
2748  }
2749  if (RHSC->isSignMask() && match(Op0, m_Add(m_Value(X), m_APInt(C)))) {
2750  // (X + C) ^ signmask -> (X + C + signmask)
2751  Constant *NewC = ConstantInt::get(I.getType(), *C + *RHSC);
2752  return BinaryOperator::CreateAdd(X, NewC);
2753  }
2754 
2755  // (X|C1)^C2 -> X^(C1^C2) iff X&~C1 == 0
2756  if (match(Op0, m_Or(m_Value(X), m_APInt(C))) &&
2757  MaskedValueIsZero(X, *C, 0, &I)) {
2758  Constant *NewC = ConstantInt::get(I.getType(), *C ^ *RHSC);
2759  Worklist.Add(cast<Instruction>(Op0));
2760  I.setOperand(0, X);
2761  I.setOperand(1, NewC);
2762  return &I;
2763  }
2764  }
2765  }
2766 
2767  if (ConstantInt *RHSC = dyn_cast<ConstantInt>(Op1)) {
2768  if (BinaryOperator *Op0I = dyn_cast<BinaryOperator>(Op0)) {
2769  if (ConstantInt *Op0CI = dyn_cast<ConstantInt>(Op0I->getOperand(1))) {
2770  if (Op0I->getOpcode() == Instruction::LShr) {
2771  // ((X^C1) >> C2) ^ C3 -> (X>>C2) ^ ((C1>>C2)^C3)
2772  // E1 = "X ^ C1"
2773  BinaryOperator *E1;
2774  ConstantInt *C1;
2775  if (Op0I->hasOneUse() &&
2776  (E1 = dyn_cast<BinaryOperator>(Op0I->getOperand(0))) &&
2777  E1->getOpcode() == Instruction::Xor &&
2778  (C1 = dyn_cast<ConstantInt>(E1->getOperand(1)))) {
2779  // fold (C1 >> C2) ^ C3
2780  ConstantInt *C2 = Op0CI, *C3 = RHSC;
2781  APInt FoldConst = C1->getValue().lshr(C2->getValue());
2782  FoldConst ^= C3->getValue();
2783  // Prepare the two operands.
2784  Value *Opnd0 = Builder.CreateLShr(E1->getOperand(0), C2);
2785  Opnd0->takeName(Op0I);
2786  cast<Instruction>(Opnd0)->setDebugLoc(I.getDebugLoc());
2787  Value *FoldVal = ConstantInt::get(Opnd0->getType(), FoldConst);
2788 
2789  return BinaryOperator::CreateXor(Opnd0, FoldVal);
2790  }
2791  }
2792  }
2793  }
2794  }
2795 
2796  if (Instruction *FoldedLogic = foldBinOpIntoSelectOrPhi(I))
2797  return FoldedLogic;
2798 
2799  // Y ^ (X | Y) --> X & ~Y
2800  // Y ^ (Y | X) --> X & ~Y
2801  if (match(Op1, m_OneUse(m_c_Or(m_Value(X), m_Specific(Op0)))))
2802  return BinaryOperator::CreateAnd(X, Builder.CreateNot(Op0));
2803  // (X | Y) ^ Y --> X & ~Y
2804  // (Y | X) ^ Y --> X & ~Y
2805  if (match(Op0, m_OneUse(m_c_Or(m_Value(X), m_Specific(Op1)))))
2806  return BinaryOperator::CreateAnd(X, Builder.CreateNot(Op1));
2807 
2808  // Y ^ (X & Y) --> ~X & Y
2809  // Y ^ (Y & X) --> ~X & Y
2810  if (match(Op1, m_OneUse(m_c_And(m_Value(X), m_Specific(Op0)))))
2811  return BinaryOperator::CreateAnd(Op0, Builder.CreateNot(X));
2812  // (X & Y) ^ Y --> ~X & Y
2813  // (Y & X) ^ Y --> ~X & Y
2814  // Canonical form is (X & C) ^ C; don't touch that.
2815  // TODO: A 'not' op is better for analysis and codegen, but demanded bits must
2816  // be fixed to prefer that (otherwise we get infinite looping).
2817  if (!match(Op1, m_Constant()) &&
2818  match(Op0, m_OneUse(m_c_And(m_Value(X), m_Specific(Op1)))))
2819  return BinaryOperator::CreateAnd(Op1, Builder.CreateNot(X));
2820 
2821  Value *A, *B, *C;
2822  // (A ^ B) ^ (A | C) --> (~A & C) ^ B -- There are 4 commuted variants.
2823  if (match(&I, m_c_Xor(m_OneUse(m_Xor(m_Value(A), m_Value(B))),
2824  m_OneUse(m_c_Or(m_Deferred(A), m_Value(C))))))
2825  return BinaryOperator::CreateXor(
2826  Builder.CreateAnd(Builder.CreateNot(A), C), B);
2827 
2828  // (A ^ B) ^ (B | C) --> (~B & C) ^ A -- There are 4 commuted variants.
2829  if (match(&I, m_c_Xor(m_OneUse(m_Xor(m_Value(A), m_Value(B))),
2830  m_OneUse(m_c_Or(m_Deferred(B), m_Value(C))))))
2831  return BinaryOperator::CreateXor(
2832  Builder.CreateAnd(Builder.CreateNot(B), C), A);
2833 
2834  // (A & B) ^ (A ^ B) -> (A | B)
2835  if (match(Op0, m_And(m_Value(A), m_Value(B))) &&
2836  match(Op1, m_c_Xor(m_Specific(A), m_Specific(B))))
2837  return BinaryOperator::CreateOr(A, B);
2838  // (A ^ B) ^ (A & B) -> (A | B)
2839  if (match(Op0, m_Xor(m_Value(A), m_Value(B))) &&
2840  match(Op1, m_c_And(m_Specific(A), m_Specific(B))))
2841  return BinaryOperator::CreateOr(A, B);
2842 
2843  // (A & ~B) ^ ~A -> ~(A & B)
2844  // (~B & A) ^ ~A -> ~(A & B)
2845  if (match(Op0, m_c_And(m_Value(A), m_Not(m_Value(B)))) &&
2846  match(Op1, m_Not(m_Specific(A))))
2847  return BinaryOperator::CreateNot(Builder.CreateAnd(A, B));
2848 
2849  if (auto *LHS = dyn_cast<ICmpInst>(I.getOperand(0)))
2850  if (auto *RHS = dyn_cast<ICmpInst>(I.getOperand(1)))
2851  if (Value *V = foldXorOfICmps(LHS, RHS))
2852  return replaceInstUsesWith(I, V);
2853 
2854  if (Instruction *CastedXor = foldCastedBitwiseLogic(I))
2855  return CastedXor;
2856 
2857  // Canonicalize a shifty way to code absolute value to the common pattern.
2858  // There are 4 potential commuted variants. Move the 'ashr' candidate to Op1.
2859  // We're relying on the fact that we only do this transform when the shift has
2860  // exactly 2 uses and the add has exactly 1 use (otherwise, we might increase
2861  // instructions).
2862  if (Op0->hasNUses(2))
2863  std::swap(Op0, Op1);
2864 
2865  const APInt *ShAmt;
2866  Type *Ty = I.getType();
2867  if (match(Op1, m_AShr(m_Value(A), m_APInt(ShAmt))) &&
2868  Op1->hasNUses(2) && *ShAmt == Ty->getScalarSizeInBits() - 1 &&
2869  match(Op0, m_OneUse(m_c_Add(m_Specific(A), m_Specific(Op1))))) {
2870  // B = ashr i32 A, 31 ; smear the sign bit
2871  // xor (add A, B), B ; add -1 and flip bits if negative
2872  // --> (A < 0) ? -A : A
2873  Value *Cmp = Builder.CreateICmpSLT(A, ConstantInt::getNullValue(Ty));
2874  // Copy the nuw/nsw flags from the add to the negate.
2875  auto *Add = cast<BinaryOperator>(Op0);
2876  Value *Neg = Builder.CreateNeg(A, "", Add->hasNoUnsignedWrap(),
2877  Add->hasNoSignedWrap());
2878  return SelectInst::Create(Cmp, Neg, A);
2879  }
2880 
2881  // Eliminate a bitwise 'not' op of 'not' min/max by inverting the min/max:
2882  //
2883  // %notx = xor i32 %x, -1
2884  // %cmp1 = icmp sgt i32 %notx, %y
2885  // %smax = select i1 %cmp1, i32 %notx, i32 %y
2886  // %res = xor i32 %smax, -1
2887  // =>
2888  // %noty = xor i32 %y, -1
2889  // %cmp2 = icmp slt %x, %noty
2890  // %res = select i1 %cmp2, i32 %x, i32 %noty
2891  //
2892  // Same is applicable for smin/umax/umin.
2893  if (match(Op1, m_AllOnes()) && Op0->hasOneUse()) {
2894  Value *LHS, *RHS;
2895  SelectPatternFlavor SPF = matchSelectPattern(Op0, LHS, RHS).Flavor;
2897  // It's possible we get here before the not has been simplified, so make
2898  // sure the input to the not isn't freely invertible.
2899  if (match(LHS, m_Not(m_Value(X))) && !IsFreeToInvert(X, X->hasOneUse())) {
2900  Value *NotY = Builder.CreateNot(RHS);
2901  return SelectInst::Create(
2902  Builder.CreateICmp(getInverseMinMaxPred(SPF), X, NotY), X, NotY);
2903  }
2904 
2905  // It's possible we get here before the not has been simplified, so make
2906  // sure the input to the not isn't freely invertible.
2907  if (match(RHS, m_Not(m_Value(Y))) && !IsFreeToInvert(Y, Y->hasOneUse())) {
2908  Value *NotX = Builder.CreateNot(LHS);
2909  return SelectInst::Create(
2910  Builder.CreateICmp(getInverseMinMaxPred(SPF), NotX, Y), NotX, Y);
2911  }
2912 
2913  // If both sides are freely invertible, then we can get rid of the xor
2914  // completely.
2915  if (IsFreeToInvert(LHS, !LHS->hasNUsesOrMore(3)) &&
2916  IsFreeToInvert(RHS, !RHS->hasNUsesOrMore(3))) {
2917  Value *NotLHS = Builder.CreateNot(LHS);
2918  Value *NotRHS = Builder.CreateNot(RHS);
2919  return SelectInst::Create(
2920  Builder.CreateICmp(getInverseMinMaxPred(SPF), NotLHS, NotRHS),
2921  NotLHS, NotRHS);
2922  }
2923  }
2924  }
2925 
2926  if (Instruction *NewXor = sinkNotIntoXor(I, Builder))
2927  return NewXor;
2928 
2929  return nullptr;
2930 }
const NoneType None
Definition: None.h:24
BinaryOp_match< LHS, RHS, Instruction::And > m_And(const LHS &L, const RHS &R)
Definition: PatternMatch.h:750
uint64_t CallInst * C
void computeKnownBits(const Value *V, KnownBits &Known, const DataLayout &DL, unsigned Depth=0, AssumptionCache *AC=nullptr, const Instruction *CxtI=nullptr, const DominatorTree *DT=nullptr, OptimizationRemarkEmitter *ORE=nullptr, bool UseInstrInfo=true)
Determine which bits of V are known to be either zero or one and return them in the KnownZero/KnownOn...
IntegerType * getType() const
getType - Specialize the getType() method to always return an IntegerType, which reduces the amount o...
Definition: Constants.h:172
Constant * getPredForICmpCode(unsigned Code, bool Sign, Type *OpTy, CmpInst::Predicate &Pred)
This is the complement of getICmpCode.
cst_pred_ty< is_nonnegative > m_NonNegative()
Match an integer or vector of nonnegative values.
Definition: PatternMatch.h:344
static ConstantInt * getFalse(LLVMContext &Context)
Definition: Constants.cpp:585
class_match< Value > m_Value()
Match an arbitrary value and ignore it.
Definition: PatternMatch.h:72
Value * CreateICmp(CmpInst::Predicate P, Value *LHS, Value *RHS, const Twine &Name="")
Definition: IRBuilder.h:1858
static GCMetadataPrinterRegistry::Add< ErlangGCPrinter > X("erlang", "erlang-compatible garbage collector")
bool isSignMask() const
Check if the APInt&#39;s value is returned by getSignMask.
Definition: APInt.h:473
Type * getSrcTy() const
Return the source type, as a convenience.
Definition: InstrTypes.h:607
static bool IsFreeToInvert(Value *V, bool WillInvertAllUses)
Return true if the specified value is free to invert (apply ~ to).
class_match< CmpInst > m_Cmp()
Matches any compare instruction and ignore it.
Definition: PatternMatch.h:80
static Value * getFCmpValue(unsigned Code, Value *LHS, Value *RHS, InstCombiner::BuilderTy &Builder)
This is the complement of getFCmpCode, which turns an opcode and two operands into either a FCmp inst...
Instruction * visitXor(BinaryOperator &I)
Value * CreateBinOp(Instruction::BinaryOps Opc, Value *LHS, Value *RHS, const Twine &Name="", MDNode *FPMathTag=nullptr)
Definition: IRBuilder.h:1258
ThreeOps_match< Cond, LHS, RHS, Instruction::Select > m_Select(const Cond &C, const LHS &L, const RHS &R)
Matches SelectInst.
static Value * foldSignedTruncationCheck(ICmpInst *ICmp0, ICmpInst *ICmp1, Instruction &CxtI, InstCombiner::BuilderTy &Builder)
General pattern: X & Y.
BinaryOp_match< LHS, RHS, Instruction::Sub > m_Sub(const LHS &L, const RHS &R)
Definition: PatternMatch.h:655
Instruction * visitOr(BinaryOperator &I)
is_zero m_Zero()
Match any null constant or a vector with all elements equal to 0.
Definition: PatternMatch.h:377
static BinaryOperator * CreateNot(Value *Op, const Twine &Name="", Instruction *InsertBefore=nullptr)
static Type * makeCmpResultType(Type *opnd_type)
Create a result type for fcmp/icmp.
Definition: InstrTypes.h:885
This class represents lattice values for constants.
Definition: AllocatorList.h:24
BinaryOps getOpcode() const
Definition: InstrTypes.h:312
BinaryOp_match< LHS, RHS, Instruction::Xor, true > m_c_Xor(const LHS &L, const RHS &R)
Matches an Xor with LHS and RHS in either order.
Value * CreateICmpULT(Value *LHS, Value *RHS, const Twine &Name="")
Definition: IRBuilder.h:1764
Value * CreateXor(Value *LHS, Value *RHS, const Twine &Name="")
Definition: IRBuilder.h:1160
bool decomposeBitTestICmp(Value *LHS, Value *RHS, CmpInst::Predicate &Pred, Value *&X, APInt &Mask, bool LookThroughTrunc=true)
Decompose an icmp into the form ((X & Mask) pred 0) if possible.
static unsigned getFCmpCode(FCmpInst::Predicate CC)
Similar to getICmpCode but for FCmpInst.
This class represents zero extension of integer types.
unsigned getICmpCode(const ICmpInst *ICI, bool InvertPred=false)
Encode a icmp predicate into a three bit mask.
BinaryOp_match< LHS, RHS, Instruction::Mul > m_Mul(const LHS &L, const RHS &R)
Definition: PatternMatch.h:702
static bool areInverseVectorBitmasks(Constant *C1, Constant *C2)
If all elements of two constant vectors are 0/-1 and inverses, return true.
APInt zext(unsigned width) const
Zero extend to a new width.
Definition: APInt.cpp:858
static APInt getLowBitsSet(unsigned numBits, unsigned loBitsSet)
Get a value with low bits set.
Definition: APInt.h:648
class_match< Constant > m_Constant()
Match an arbitrary Constant and ignore it.
Definition: PatternMatch.h:91
m_Intrinsic_Ty< Opnd0 >::Ty m_BSwap(const Opnd0 &Op0)
unsigned less or equal
Definition: InstrTypes.h:668
unsigned less than
Definition: InstrTypes.h:667
BinaryOp_match< LHS, RHS, Instruction::AShr > m_AShr(const LHS &L, const RHS &R)
Definition: PatternMatch.h:780
bool isSubsetOf(const APInt &RHS) const
This operation checks that all bits set in this APInt are also set in RHS.
Definition: APInt.h:1329
0 1 0 0 True if ordered and less than
Definition: InstrTypes.h:648
static SelectInst * Create(Value *C, Value *S1, Value *S2, const Twine &NameStr="", Instruction *InsertBefore=nullptr, Instruction *MDFrom=nullptr)
1 1 1 0 True if unordered or not equal
Definition: InstrTypes.h:658
bool sgt(const APInt &RHS) const
Signed greather than comparison.
Definition: APInt.h:1274
MaskedICmpType
Classify (icmp eq (A & B), C) and (icmp ne (A & B), C) as matching patterns that can be simplified...
static Value * SimplifyBSwap(BinaryOperator &I, InstCombiner::BuilderTy &Builder)
Transform BITWISE_OP(BSWAP(A),BSWAP(B)) or BITWISE_OP(BSWAP(A), Constant) to BSWAP(BITWISE_OP(A, B))
F(f)
This class represents a sign extension of integer types.
#define R2(n)
bool isVectorTy() const
True if this is an instance of VectorType.
Definition: Type.h:230
bool sle(const APInt &RHS) const
Signed less or equal comparison.
Definition: APInt.h:1239
AnyBinaryOp_match< LHS, RHS, true > m_c_BinOp(const LHS &L, const RHS &R)
Matches a BinaryOperator with LHS and RHS in either order.
static bool isBitwiseLogicOp(unsigned Opcode)
Determine if the Opcode is and/or/xor.
Definition: Instruction.h:173
Value * SimplifyOrInst(Value *LHS, Value *RHS, const SimplifyQuery &Q)
Given operands for an Or, fold the result or return null.
BinOpPred_match< LHS, RHS, is_logical_shift_op > m_LogicalShift(const LHS &L, const RHS &R)
Matches logical shift operations.
Definition: PatternMatch.h:942
LLVMContext & getContext() const
Return the LLVMContext in which this type was uniqued.
Definition: Type.h:130
static Constant * getNullValue(Type *Ty)
Constructor to create a &#39;0&#39; constant of arbitrary type.
Definition: Constants.cpp:265
Value * CreateNot(Value *V, const Twine &Name="")
Definition: IRBuilder.h:1294
1 0 0 1 True if unordered or equal
Definition: InstrTypes.h:653
static GCMetadataPrinterRegistry::Add< OcamlGCMetadataPrinter > Y("ocaml", "ocaml 3.10-compatible collector")
bool match(Val *V, const Pattern &P)
Definition: PatternMatch.h:49
1 0 0 0 True if unordered: isnan(X) | isnan(Y)
Definition: InstrTypes.h:652
bool isSigned() const
Definition: InstrTypes.h:812
BinaryOp_match< LHS, RHS, Instruction::Xor > m_Xor(const LHS &L, const RHS &R)
Definition: PatternMatch.h:762
bool isKnownToBeAPowerOfTwo(const Value *V, const DataLayout &DL, bool OrZero=false, unsigned Depth=0, AssumptionCache *AC=nullptr, const Instruction *CxtI=nullptr, const DominatorTree *DT=nullptr, bool UseInstrInfo=true)
Return true if the given value is known to have exactly one bit set when defined. ...
Predicate getInversePredicate() const
For example, EQ -> NE, UGT -> ULE, SLT -> SGE, OEQ -> UNE, UGT -> OLE, OLT -> UGE, etc.
Definition: InstrTypes.h:741
This is the base class for all instructions that perform data casts.
Definition: InstrTypes.h:349
static Value * getNewICmpValue(unsigned Code, bool Sign, Value *LHS, Value *RHS, InstCombiner::BuilderTy &Builder)
This is the complement of getICmpCode, which turns an opcode and two operands into either a constant ...
APInt shl(unsigned shiftAmt) const
Left-shift function.
Definition: APInt.h:993
static Optional< unsigned > getOpcode(ArrayRef< VPValue *> Values)
Returns the opcode of Values or ~0 if they do not all agree.
Definition: VPlanSLP.cpp:197
CastClass_match< OpTy, Instruction::Trunc > m_Trunc(const OpTy &Op)
Matches Trunc.
bool isIntegerTy() const
True if this is an instance of IntegerType.
Definition: Type.h:197
This provides a uniform API for creating instructions and inserting them into a basic block: either a...
Definition: IRBuilder.h:743
static Constant * AddOne(Constant *C)
Add one to a Constant.
0 1 0 1 True if ordered and less than or equal
Definition: InstrTypes.h:649
bool MaskedValueIsZero(const Value *V, const APInt &Mask, const DataLayout &DL, unsigned Depth=0, AssumptionCache *AC=nullptr, const Instruction *CxtI=nullptr, const DominatorTree *DT=nullptr, bool UseInstrInfo=true)
Return true if &#39;V & Mask&#39; is known to be zero.
static Constant * getSExt(Constant *C, Type *Ty, bool OnlyIfReduced=false)
Definition: Constants.cpp:1651
Value * CreateAdd(Value *LHS, Value *RHS, const Twine &Name="", bool HasNUW=false, bool HasNSW=false)
Definition: IRBuilder.h:974
BinaryOp_match< LHS, RHS, Instruction::Add > m_Add(const LHS &L, const RHS &R)
Definition: PatternMatch.h:643
static Constant * getZExt(Constant *C, Type *Ty, bool OnlyIfReduced=false)
Definition: Constants.cpp:1665
static Value * foldLogOpOfMaskedICmps_NotAllZeros_BMask_Mixed(ICmpInst *LHS, ICmpInst *RHS, bool IsAnd, Value *A, Value *B, Value *C, Value *D, Value *E, ICmpInst::Predicate PredL, ICmpInst::Predicate PredR, llvm::InstCombiner::BuilderTy &Builder)
Try to fold (icmp(A & B) ==/!= C) &/| (icmp(A & D) ==/!= E) into a single (icmp(A & X) ==/!= Y)...
static Value * peekThroughBitcast(Value *V, bool OneUseOnly=false)
Return the source operand of a potentially bitcasted value while optionally checking if it has one us...
Instruction::CastOps getOpcode() const
Return the opcode of this CastInst.
Definition: InstrTypes.h:602
Type * getType() const
All values are typed, get the type of this value.
Definition: Value.h:245
CastClass_match< OpTy, Instruction::ZExt > m_ZExt(const OpTy &Op)
Matches ZExt.
This instruction compares its operands according to the predicate given to the constructor.
class_match< ConstantInt > m_ConstantInt()
Match an arbitrary ConstantInt and ignore it.
Definition: PatternMatch.h:83
static Value * foldAndOrOfEqualityCmpsWithConstants(ICmpInst *LHS, ICmpInst *RHS, bool JoinedByAnd, InstCombiner::BuilderTy &Builder)
const APInt & getValue() const
Return the constant as an APInt value reference.
Definition: Constants.h:138
unsigned getOpcode() const
Returns a member of one of the enums like Instruction::Add.
Definition: Instruction.h:126
cstfp_pred_ty< is_pos_zero_fp > m_PosZeroFP()
Match a floating-point positive zero.
Definition: PatternMatch.h:446
bool isMinusOne() const
This function will return true iff every bit in this constant is set to true.
Definition: Constants.h:209
bool isIntOrIntVectorTy() const
Return true if this is an integer type or a vector of integer types.
Definition: Type.h:203
cst_pred_ty< is_power2 > m_Power2()
Match an integer or vector power-of-2.
Definition: PatternMatch.h:386
unsigned getBitWidth() const
Get the number of bits in this IntegerType.
Definition: DerivedTypes.h:66
void takeName(Value *V)
Transfer the name from V to this value.
Definition: Value.cpp:292
Function * getDeclaration(Module *M, ID id, ArrayRef< Type *> Tys=None)
Create or insert an LLVM Function declaration for an intrinsic, and return it.
Definition: Function.cpp:1021
static BinaryOperator * CreateAdd(Value *S1, Value *S2, const Twine &Name, Instruction *InsertBefore, Value *FlagsOp)
Value * getOperand(unsigned i) const
Definition: User.h:170
Value * CreateOr(Value *LHS, Value *RHS, const Twine &Name="")
Definition: IRBuilder.h:1142
Constant * getAggregateElement(unsigned Elt) const
For aggregates (struct/array/vector) return the constant that corresponds to the specified element if...
Definition: Constants.cpp:335
Value * CreateFCmp(CmpInst::Predicate P, Value *LHS, Value *RHS, const Twine &Name="", MDNode *FPMathTag=nullptr)
Definition: IRBuilder.h:1866
OneUse_match< T > m_OneUse(const T &SubPattern)
Definition: PatternMatch.h:63
static Optional< std::pair< unsigned, unsigned > > getMaskedTypeForICmpPair(Value *&A, Value *&B, Value *&C, Value *&D, Value *&E, ICmpInst *LHS, ICmpInst *RHS, ICmpInst::Predicate &PredL, ICmpInst::Predicate &PredR)
Handle (icmp(A & B) ==/!= C) &/| (icmp(A & D) ==/!= E).
unsigned ComputeNumSignBits(const Value *Op, const DataLayout &DL, unsigned Depth=0, AssumptionCache *AC=nullptr, const Instruction *CxtI=nullptr, const DominatorTree *DT=nullptr, bool UseInstrInfo=true)
Return the number of times the sign bit of the register is replicated into the other bits...
BinaryOp_match< LHS, RHS, Instruction::LShr > m_LShr(const LHS &L, const RHS &R)
Definition: PatternMatch.h:774
bool hasNUsesOrMore(unsigned N) const
Return true if this value has N users or more.
Definition: Value.cpp:136
static Instruction * foldOrToXor(BinaryOperator &I, InstCombiner::BuilderTy &Builder)
bool isAllOnesValue() const
Determine if all bits are set.
Definition: APInt.h:396
static GCRegistry::Add< OcamlGC > B("ocaml", "ocaml 3.10-compatible GC")
bool hasNUses(unsigned N) const
Return true if this Value has exactly N users.
Definition: Value.cpp:132
apint_match m_APInt(const APInt *&Res)
Match a ConstantInt or splatted ConstantVector, binding the specified pointer to the contained APInt...
Definition: PatternMatch.h:177
BinaryOp_match< LHS, RHS, Instruction::Add, true > m_c_Add(const LHS &L, const RHS &R)
Matches a Add with LHS and RHS in either order.
static unsigned getMaskedICmpType(Value *A, Value *B, Value *C, ICmpInst::Predicate Pred)
Return the set of patterns (from MaskedICmpType) that (icmp SCC (A & B), C) satisfies.
static bool canNarrowShiftAmt(Constant *C, unsigned BitWidth)
Return true if a constant shift amount is always less than the specified bit-width.
The instances of the Type class are immutable: once they are created, they are never changed...
Definition: Type.h:46
BinaryOp_match< LHS, RHS, Instruction::Or > m_Or(const LHS &L, const RHS &R)
Definition: PatternMatch.h:756
bool ult(const APInt &RHS) const
Unsigned less than comparison.
Definition: APInt.h:1185
static GCRegistry::Add< CoreCLRGC > E("coreclr", "CoreCLR-compatible GC")
This is an important base class in LLVM.
Definition: Constant.h:42
BinaryOp_match< LHS, RHS, Instruction::And, true > m_c_And(const LHS &L, const RHS &R)
Matches an And with LHS and RHS in either order.
static Constant * getAnd(Constant *C1, Constant *C2)
Definition: Constants.cpp:2298
bool isOneValue() const
Determine if this is a value of 1.
Definition: APInt.h:411
cst_pred_ty< is_all_ones > m_AllOnes()
Match an integer or vector with all bits set.
Definition: PatternMatch.h:310
specificval_ty m_Specific(const Value *V)
Match if we have a specific specified value.
Definition: PatternMatch.h:503
bool isMinSignedValue() const
Determine if this is the smallest signed value.
Definition: APInt.h:443
BinaryOp_match< LHS, RHS, Instruction::Shl > m_Shl(const LHS &L, const RHS &R)
Definition: PatternMatch.h:768
This instruction compares its operands according to the predicate given to the constructor.
Predicate
This enumeration lists the possible predicates for CmpInst subclasses.
Definition: InstrTypes.h:642
0 1 1 1 True if ordered (no nans)
Definition: InstrTypes.h:651
class_match< BinaryOperator > m_BinOp()
Match an arbitrary binary operation and ignore it.
Definition: PatternMatch.h:75
static unsigned conjugateICmpMask(unsigned Mask)
Convert an analysis of a masked ICmp into its equivalent if all boolean operations had the opposite s...
bool isIntN(unsigned N) const
Check if this APInt has an N-bits unsigned integer value.
Definition: APInt.h:450
CmpInst::Predicate getInverseMinMaxPred(SelectPatternFlavor SPF)
Return the canonical inverse comparison predicate for the specified minimum/maximum flavor...
static Constant * getNot(Constant *C)
Definition: Constants.cpp:2233
static Constant * getAllOnesValue(Type *Ty)
Definition: Constants.cpp:319
1 1 1 1 Always true (always folded)
Definition: InstrTypes.h:659
void swapOperands()
Exchange the two operands to this instruction in such a way that it does not modify the semantics of ...
static Instruction * matchDeMorgansLaws(BinaryOperator &I, InstCombiner::BuilderTy &Builder)
Match De Morgan&#39;s Laws: (~A & ~B) == (~(A | B)) (~A | ~B) == (~(A & B))
1 1 0 1 True if unordered, less than, or equal
Definition: InstrTypes.h:657
#define llvm_unreachable(msg)
Marks that the current location is not supposed to be reachable.
deferredval_ty< Value > m_Deferred(Value *const &V)
A commutative-friendly version of m_Specific().
Definition: PatternMatch.h:516
APInt lshr(unsigned shiftAmt) const
Logical right-shift function.
Definition: APInt.h:971
signed greater than
Definition: InstrTypes.h:669
Instruction * visitAnd(BinaryOperator &I)
const APInt & umin(const APInt &A, const APInt &B)
Determine the smaller of two APInts considered to be signed.
Definition: APInt.h:2115
CastClass_match< OpTy, Instruction::SExt > m_SExt(const OpTy &Op)
Matches SExt.
static Value * foldLogOpOfMaskedICmpsAsymmetric(ICmpInst *LHS, ICmpInst *RHS, bool IsAnd, Value *A, Value *B, Value *C, Value *D, Value *E, ICmpInst::Predicate PredL, ICmpInst::Predicate PredR, unsigned LHSMask, unsigned RHSMask, llvm::InstCombiner::BuilderTy &Builder)
Try to fold (icmp(A & B) ==/!= 0) &/| (icmp(A & D) ==/!= E) into a single (icmp(A & X) ==/!= Y), where the left-hand ...
0 0 1 0 True if ordered and greater than
Definition: InstrTypes.h:646
cst_pred_ty< is_negative > m_Negative()
Match an integer or vector of negative values.
Definition: PatternMatch.h:332
BinaryOp_match< LHS, RHS, Instruction::Or, true > m_c_Or(const LHS &L, const RHS &R)
Matches an Or with LHS and RHS in either order.
This is the shared class of boolean and integer constants.
Definition: Constants.h:84
SelectPatternFlavor Flavor
unsigned getScalarSizeInBits() const LLVM_READONLY
If this is a vector type, return the getPrimitiveSizeInBits value for the element type...
Definition: Type.cpp:130
This is a &#39;vector&#39; (really, a variable-sized array), optimized for the case when the array is small...
Definition: SmallVector.h:847
1 1 0 0 True if unordered or less than
Definition: InstrTypes.h:656
SelectPatternFlavor
Specific patterns of select instructions we can match.
signed less than
Definition: InstrTypes.h:671
LLVM_NODISCARD T pop_back_val()
Definition: SmallVector.h:381
static Constant * getTrunc(Constant *C, Type *Ty, bool OnlyIfReduced=false)
Definition: Constants.cpp:1637
static GCRegistry::Add< StatepointGC > D("statepoint-example", "an example strategy for statepoint")
static Constant * get(Type *Ty, uint64_t V, bool isSigned=false)
If Ty is a vector type, return a Constant with a splat of the given value.
Definition: Constants.cpp:622
static ConstantInt * getSigned(IntegerType *Ty, int64_t V)
Return a ConstantInt with the specified value for the specified type.
Definition: Constants.cpp:636
Predicate getSignedPredicate() const
For example, EQ->EQ, SLE->SLE, UGT->SGT, etc.
Type * getDestTy() const
Return the destination type, as a convenience.
Definition: InstrTypes.h:609
static ConstantInt * getTrue(LLVMContext &Context)
Definition: Constants.cpp:578
void setPredicate(Predicate P)
Set the predicate for this instruction to the specified value.
Definition: InstrTypes.h:722
void setOperand(unsigned i, Value *Val)
Definition: User.h:175
void swap(llvm::BitVector &LHS, llvm::BitVector &RHS)
Implement std::swap in terms of BitVector swap.
Definition: BitVector.h:941
unsigned getVectorNumElements() const
Definition: DerivedTypes.h:462
signed less or equal
Definition: InstrTypes.h:672
const Module * getModule() const
Return the module owning the function this instruction belongs to or nullptr it the function does not...
Definition: Instruction.cpp:56
Class for arbitrary precision integers.
Definition: APInt.h:70
bool ule(const APInt &RHS) const
Unsigned less or equal comparison.
Definition: APInt.h:1223
bool isPowerOf2() const
Check if this APInt&#39;s value is a power of two greater than zero.
Definition: APInt.h:464
static Instruction * sinkNotIntoXor(BinaryOperator &I, InstCombiner::BuilderTy &Builder)
static BinaryOperator * Create(BinaryOps Op, Value *S1, Value *S2, const Twine &Name=Twine(), Instruction *InsertBefore=nullptr)
Construct a binary instruction, given the opcode and the two operands.
void removeFromParent()
This method unlinks &#39;this&#39; from the containing basic block, but does not delete it.
Definition: Instruction.cpp:64
static Instruction * foldAndToXor(BinaryOperator &I, InstCombiner::BuilderTy &Builder)
static Constant * getNeg(Constant *C, bool HasNUW=false, bool HasNSW=false)
Definition: Constants.cpp:2220
Value * SimplifyAndInst(Value *LHS, Value *RHS, const SimplifyQuery &Q)
Given operands for an And, fold the result or return null.
static Value * foldLogOpOfMaskedICmps(ICmpInst *LHS, ICmpInst *RHS, bool IsAnd, llvm::InstCombiner::BuilderTy &Builder)
Try to fold (icmp(A & B) ==/!= C) &/| (icmp(A & D) ==/!= E) into a single (icmp(A & X) ==/!= Y)...
static CastInst * Create(Instruction::CastOps, Value *S, Type *Ty, const Twine &Name="", Instruction *InsertBefore=nullptr)
Provides a way to construct any of the CastInst subclasses using an opcode instead of the subclass&#39;s ...
bool ugt(const APInt &RHS) const
Unsigned greather than comparison.
Definition: APInt.h:1255
Predicate getPredicate() const
Return the predicate for this instruction.
Definition: InstrTypes.h:717
const DebugLoc & getDebugLoc() const
Return the debug location for this node as a DebugLoc.
Definition: Instruction.h:311
bool intersects(const APInt &RHS) const
This operation tests if there are any pairs of corresponding bits between this APInt and RHS that are...
Definition: APInt.h:1321
static bool isMinOrMax(SelectPatternFlavor SPF)
When implementing this min/max pattern as fcmp; select, does the fcmp have to be ordered?
unsigned greater or equal
Definition: InstrTypes.h:666
StringRef getName() const
Return a constant reference to the value&#39;s name.
Definition: Value.cpp:215
bool isEquality() const
Return true if this predicate is either EQ or NE.
#define I(x, y, z)
Definition: MD5.cpp:58
#define N
static Constant * getOr(Constant *C1, Constant *C2)
Definition: Constants.cpp:2302
bool isZero() const
This is just a convenience method to make client code smaller for a common code.
Definition: Constants.h:193
0 1 1 0 True if ordered and operands are unequal
Definition: InstrTypes.h:650
LLVM_NODISCARD std::enable_if<!is_simple_type< Y >::value, typename cast_retty< X, const Y >::ret_type >::type dyn_cast(const Y &Val)
Definition: Casting.h:323
Value * SimplifyXorInst(Value *LHS, Value *RHS, const SimplifyQuery &Q)
Given operands for an Xor, fold the result or return null.
1 0 1 0 True if unordered or greater than
Definition: InstrTypes.h:654
APInt byteSwap() const
Definition: APInt.cpp:618
Value * CreateAnd(Value *LHS, Value *RHS, const Twine &Name="")
Definition: IRBuilder.h:1124
bool isMinValue() const
Determine if this is the smallest unsigned value.
Definition: APInt.h:437
Value * SimplifyBinOp(unsigned Opcode, Value *LHS, Value *RHS, const SimplifyQuery &Q)
Given operands for a BinaryOperator, fold the result or return null.
assert(ImpDefSCC.getReg()==AMDGPU::SCC &&ImpDefSCC.isDef())
0 0 0 1 True if ordered and equal
Definition: InstrTypes.h:645
LLVM Value Representation.
Definition: Value.h:73
This file provides internal interfaces used to implement the InstCombine.
1 0 1 1 True if unordered, greater than, or equal
Definition: InstrTypes.h:655
SelectPatternResult matchSelectPattern(Value *V, Value *&LHS, Value *&RHS, Instruction::CastOps *CastOp=nullptr, unsigned Depth=0)
Pattern match integer [SU]MIN, [SU]MAX and ABS idioms, returning the kind and providing the out param...
cst_pred_ty< is_one > m_One()
Match an integer 1 or a vector with all elements equal to 1.
Definition: PatternMatch.h:356
std::underlying_type< E >::type Mask()
Get a bitmask with 1s in all places up to the high-order bit of E&#39;s largest value.
Definition: BitmaskEnum.h:81
match_combine_and< LTy, RTy > m_CombineAnd(const LTy &L, const RTy &R)
Combine two pattern matchers matching L && R.
Definition: PatternMatch.h:131
static Instruction * visitMaskedMerge(BinaryOperator &I, InstCombiner::BuilderTy &Builder)
If we have a masked merge, in the canonical form of: (assuming that A only has one use...
bool hasOneUse() const
Return true if there is exactly one user of this value.
Definition: Value.h:413
unsigned greater than
Definition: InstrTypes.h:665
Predicate getSwappedPredicate() const
For example, EQ->EQ, SLE->SGE, ULT->UGT, OEQ->OEQ, ULE->UGE, OLT->OGT, etc.
Definition: InstrTypes.h:757
bool isNonNegative() const
Returns true if this value is known to be non-negative.
Definition: KnownBits.h:99
Value * simplifyRangeCheck(ICmpInst *Cmp0, ICmpInst *Cmp1, bool Inverted)
Try to fold a signed range checked with lower bound 0 to an unsigned icmp.
bool recognizeBSwapOrBitReverseIdiom(Instruction *I, bool MatchBSwaps, bool MatchBitReversals, SmallVectorImpl< Instruction *> &InsertedInsts)
Try to match a bswap or bitreverse idiom.
Definition: Local.cpp:2743
0 0 1 1 True if ordered and greater than or equal
Definition: InstrTypes.h:647
static Instruction * foldLogicCastConstant(BinaryOperator &Logic, CastInst *Cast, InstCombiner::BuilderTy &Builder)
Fold {and,or,xor} (cast X), C.
static Constant * SubOne(Constant *C)
Subtract one from a Constant.
bool predicatesFoldable(CmpInst::Predicate P1, CmpInst::Predicate P2)
Return true if both predicates match sign or if at least one of them is an equality comparison (which...
static Instruction * foldXorToXor(BinaryOperator &I, InstCombiner::BuilderTy &Builder)
A ^ B can be specified using other logic ops in a variety of patterns.
BinaryOp_match< ValTy, cst_pred_ty< is_all_ones >, Instruction::Xor, true > m_Not(const ValTy &V)
Matches a &#39;Not&#39; as &#39;xor V, -1&#39; or &#39;xor -1, V&#39;.
bool isNullValue() const
Determine if all bits are clear.
Definition: APInt.h:406
0 0 0 0 Always false (always folded)
Definition: InstrTypes.h:644
signed greater or equal
Definition: InstrTypes.h:670
CmpClass_match< LHS, RHS, ICmpInst, ICmpInst::Predicate > m_ICmp(ICmpInst::Predicate &Pred, const LHS &L, const RHS &R)
static Constant * getXor(Constant *C1, Constant *C2)
Definition: Constants.cpp:2306
CallInst * CreateCall(Value *Callee, ArrayRef< Value *> Args=None, const Twine &Name="", MDNode *FPMathTag=nullptr)
Definition: IRBuilder.h:1883