1 //===- InstCombineLoadStoreAlloca.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 visit functions for load, store and alloca.
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "InstCombineInternal.h"
15 #include "llvm/ADT/Statistic.h"
16 #include "llvm/Analysis/Loads.h"
17 #include "llvm/IR/DataLayout.h"
18 #include "llvm/IR/LLVMContext.h"
19 #include "llvm/IR/IntrinsicInst.h"
20 #include "llvm/IR/MDBuilder.h"
21 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
22 #include "llvm/Transforms/Utils/Local.h"
23 using namespace llvm;
24 
25 #define DEBUG_TYPE "instcombine"
26 
27 STATISTIC(NumDeadStore,    "Number of dead stores eliminated");
28 STATISTIC(NumGlobalCopies, "Number of allocas copied from constant global");
29 
30 /// pointsToConstantGlobal - Return true if V (possibly indirectly) points to
31 /// some part of a constant global variable.  This intentionally only accepts
32 /// constant expressions because we can't rewrite arbitrary instructions.
pointsToConstantGlobal(Value * V)33 static bool pointsToConstantGlobal(Value *V) {
34   if (GlobalVariable *GV = dyn_cast<GlobalVariable>(V))
35     return GV->isConstant();
36 
37   if (ConstantExpr *CE = dyn_cast<ConstantExpr>(V)) {
38     if (CE->getOpcode() == Instruction::BitCast ||
39         CE->getOpcode() == Instruction::AddrSpaceCast ||
40         CE->getOpcode() == Instruction::GetElementPtr)
41       return pointsToConstantGlobal(CE->getOperand(0));
42   }
43   return false;
44 }
45 
46 /// isOnlyCopiedFromConstantGlobal - Recursively walk the uses of a (derived)
47 /// pointer to an alloca.  Ignore any reads of the pointer, return false if we
48 /// see any stores or other unknown uses.  If we see pointer arithmetic, keep
49 /// track of whether it moves the pointer (with IsOffset) but otherwise traverse
50 /// the uses.  If we see a memcpy/memmove that targets an unoffseted pointer to
51 /// the alloca, and if the source pointer is a pointer to a constant global, we
52 /// can optimize this.
53 static bool
isOnlyCopiedFromConstantGlobal(Value * V,MemTransferInst * & TheCopy,SmallVectorImpl<Instruction * > & ToDelete)54 isOnlyCopiedFromConstantGlobal(Value *V, MemTransferInst *&TheCopy,
55                                SmallVectorImpl<Instruction *> &ToDelete) {
56   // We track lifetime intrinsics as we encounter them.  If we decide to go
57   // ahead and replace the value with the global, this lets the caller quickly
58   // eliminate the markers.
59 
60   SmallVector<std::pair<Value *, bool>, 35> ValuesToInspect;
61   ValuesToInspect.push_back(std::make_pair(V, false));
62   while (!ValuesToInspect.empty()) {
63     auto ValuePair = ValuesToInspect.pop_back_val();
64     const bool IsOffset = ValuePair.second;
65     for (auto &U : ValuePair.first->uses()) {
66       Instruction *I = cast<Instruction>(U.getUser());
67 
68       if (LoadInst *LI = dyn_cast<LoadInst>(I)) {
69         // Ignore non-volatile loads, they are always ok.
70         if (!LI->isSimple()) return false;
71         continue;
72       }
73 
74       if (isa<BitCastInst>(I) || isa<AddrSpaceCastInst>(I)) {
75         // If uses of the bitcast are ok, we are ok.
76         ValuesToInspect.push_back(std::make_pair(I, IsOffset));
77         continue;
78       }
79       if (GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(I)) {
80         // If the GEP has all zero indices, it doesn't offset the pointer. If it
81         // doesn't, it does.
82         ValuesToInspect.push_back(
83             std::make_pair(I, IsOffset || !GEP->hasAllZeroIndices()));
84         continue;
85       }
86 
87       if (auto CS = CallSite(I)) {
88         // If this is the function being called then we treat it like a load and
89         // ignore it.
90         if (CS.isCallee(&U))
91           continue;
92 
93         // Inalloca arguments are clobbered by the call.
94         unsigned ArgNo = CS.getArgumentNo(&U);
95         if (CS.isInAllocaArgument(ArgNo))
96           return false;
97 
98         // If this is a readonly/readnone call site, then we know it is just a
99         // load (but one that potentially returns the value itself), so we can
100         // ignore it if we know that the value isn't captured.
101         if (CS.onlyReadsMemory() &&
102             (CS.getInstruction()->use_empty() || CS.doesNotCapture(ArgNo)))
103           continue;
104 
105         // If this is being passed as a byval argument, the caller is making a
106         // copy, so it is only a read of the alloca.
107         if (CS.isByValArgument(ArgNo))
108           continue;
109       }
110 
111       // Lifetime intrinsics can be handled by the caller.
112       if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(I)) {
113         if (II->getIntrinsicID() == Intrinsic::lifetime_start ||
114             II->getIntrinsicID() == Intrinsic::lifetime_end) {
115           assert(II->use_empty() && "Lifetime markers have no result to use!");
116           ToDelete.push_back(II);
117           continue;
118         }
119       }
120 
121       // If this is isn't our memcpy/memmove, reject it as something we can't
122       // handle.
123       MemTransferInst *MI = dyn_cast<MemTransferInst>(I);
124       if (!MI)
125         return false;
126 
127       // If the transfer is using the alloca as a source of the transfer, then
128       // ignore it since it is a load (unless the transfer is volatile).
129       if (U.getOperandNo() == 1) {
130         if (MI->isVolatile()) return false;
131         continue;
132       }
133 
134       // If we already have seen a copy, reject the second one.
135       if (TheCopy) return false;
136 
137       // If the pointer has been offset from the start of the alloca, we can't
138       // safely handle this.
139       if (IsOffset) return false;
140 
141       // If the memintrinsic isn't using the alloca as the dest, reject it.
142       if (U.getOperandNo() != 0) return false;
143 
144       // If the source of the memcpy/move is not a constant global, reject it.
145       if (!pointsToConstantGlobal(MI->getSource()))
146         return false;
147 
148       // Otherwise, the transform is safe.  Remember the copy instruction.
149       TheCopy = MI;
150     }
151   }
152   return true;
153 }
154 
155 /// isOnlyCopiedFromConstantGlobal - Return true if the specified alloca is only
156 /// modified by a copy from a constant global.  If we can prove this, we can
157 /// replace any uses of the alloca with uses of the global directly.
158 static MemTransferInst *
isOnlyCopiedFromConstantGlobal(AllocaInst * AI,SmallVectorImpl<Instruction * > & ToDelete)159 isOnlyCopiedFromConstantGlobal(AllocaInst *AI,
160                                SmallVectorImpl<Instruction *> &ToDelete) {
161   MemTransferInst *TheCopy = nullptr;
162   if (isOnlyCopiedFromConstantGlobal(AI, TheCopy, ToDelete))
163     return TheCopy;
164   return nullptr;
165 }
166 
simplifyAllocaArraySize(InstCombiner & IC,AllocaInst & AI)167 static Instruction *simplifyAllocaArraySize(InstCombiner &IC, AllocaInst &AI) {
168   // Check for array size of 1 (scalar allocation).
169   if (!AI.isArrayAllocation()) {
170     // i32 1 is the canonical array size for scalar allocations.
171     if (AI.getArraySize()->getType()->isIntegerTy(32))
172       return nullptr;
173 
174     // Canonicalize it.
175     Value *V = IC.Builder->getInt32(1);
176     AI.setOperand(0, V);
177     return &AI;
178   }
179 
180   // Convert: alloca Ty, C - where C is a constant != 1 into: alloca [C x Ty], 1
181   if (const ConstantInt *C = dyn_cast<ConstantInt>(AI.getArraySize())) {
182     Type *NewTy = ArrayType::get(AI.getAllocatedType(), C->getZExtValue());
183     AllocaInst *New = IC.Builder->CreateAlloca(NewTy, nullptr, AI.getName());
184     New->setAlignment(AI.getAlignment());
185 
186     // Scan to the end of the allocation instructions, to skip over a block of
187     // allocas if possible...also skip interleaved debug info
188     //
189     BasicBlock::iterator It = New;
190     while (isa<AllocaInst>(*It) || isa<DbgInfoIntrinsic>(*It))
191       ++It;
192 
193     // Now that I is pointing to the first non-allocation-inst in the block,
194     // insert our getelementptr instruction...
195     //
196     Type *IdxTy = IC.getDataLayout().getIntPtrType(AI.getType());
197     Value *NullIdx = Constant::getNullValue(IdxTy);
198     Value *Idx[2] = {NullIdx, NullIdx};
199     Instruction *GEP =
200         GetElementPtrInst::CreateInBounds(New, Idx, New->getName() + ".sub");
201     IC.InsertNewInstBefore(GEP, *It);
202 
203     // Now make everything use the getelementptr instead of the original
204     // allocation.
205     return IC.ReplaceInstUsesWith(AI, GEP);
206   }
207 
208   if (isa<UndefValue>(AI.getArraySize()))
209     return IC.ReplaceInstUsesWith(AI, Constant::getNullValue(AI.getType()));
210 
211   // Ensure that the alloca array size argument has type intptr_t, so that
212   // any casting is exposed early.
213   Type *IntPtrTy = IC.getDataLayout().getIntPtrType(AI.getType());
214   if (AI.getArraySize()->getType() != IntPtrTy) {
215     Value *V = IC.Builder->CreateIntCast(AI.getArraySize(), IntPtrTy, false);
216     AI.setOperand(0, V);
217     return &AI;
218   }
219 
220   return nullptr;
221 }
222 
visitAllocaInst(AllocaInst & AI)223 Instruction *InstCombiner::visitAllocaInst(AllocaInst &AI) {
224   if (auto *I = simplifyAllocaArraySize(*this, AI))
225     return I;
226 
227   if (AI.getAllocatedType()->isSized()) {
228     // If the alignment is 0 (unspecified), assign it the preferred alignment.
229     if (AI.getAlignment() == 0)
230       AI.setAlignment(DL.getPrefTypeAlignment(AI.getAllocatedType()));
231 
232     // Move all alloca's of zero byte objects to the entry block and merge them
233     // together.  Note that we only do this for alloca's, because malloc should
234     // allocate and return a unique pointer, even for a zero byte allocation.
235     if (DL.getTypeAllocSize(AI.getAllocatedType()) == 0) {
236       // For a zero sized alloca there is no point in doing an array allocation.
237       // This is helpful if the array size is a complicated expression not used
238       // elsewhere.
239       if (AI.isArrayAllocation()) {
240         AI.setOperand(0, ConstantInt::get(AI.getArraySize()->getType(), 1));
241         return &AI;
242       }
243 
244       // Get the first instruction in the entry block.
245       BasicBlock &EntryBlock = AI.getParent()->getParent()->getEntryBlock();
246       Instruction *FirstInst = EntryBlock.getFirstNonPHIOrDbg();
247       if (FirstInst != &AI) {
248         // If the entry block doesn't start with a zero-size alloca then move
249         // this one to the start of the entry block.  There is no problem with
250         // dominance as the array size was forced to a constant earlier already.
251         AllocaInst *EntryAI = dyn_cast<AllocaInst>(FirstInst);
252         if (!EntryAI || !EntryAI->getAllocatedType()->isSized() ||
253             DL.getTypeAllocSize(EntryAI->getAllocatedType()) != 0) {
254           AI.moveBefore(FirstInst);
255           return &AI;
256         }
257 
258         // If the alignment of the entry block alloca is 0 (unspecified),
259         // assign it the preferred alignment.
260         if (EntryAI->getAlignment() == 0)
261           EntryAI->setAlignment(
262               DL.getPrefTypeAlignment(EntryAI->getAllocatedType()));
263         // Replace this zero-sized alloca with the one at the start of the entry
264         // block after ensuring that the address will be aligned enough for both
265         // types.
266         unsigned MaxAlign = std::max(EntryAI->getAlignment(),
267                                      AI.getAlignment());
268         EntryAI->setAlignment(MaxAlign);
269         if (AI.getType() != EntryAI->getType())
270           return new BitCastInst(EntryAI, AI.getType());
271         return ReplaceInstUsesWith(AI, EntryAI);
272       }
273     }
274   }
275 
276   if (AI.getAlignment()) {
277     // Check to see if this allocation is only modified by a memcpy/memmove from
278     // a constant global whose alignment is equal to or exceeds that of the
279     // allocation.  If this is the case, we can change all users to use
280     // the constant global instead.  This is commonly produced by the CFE by
281     // constructs like "void foo() { int A[] = {1,2,3,4,5,6,7,8,9...}; }" if 'A'
282     // is only subsequently read.
283     SmallVector<Instruction *, 4> ToDelete;
284     if (MemTransferInst *Copy = isOnlyCopiedFromConstantGlobal(&AI, ToDelete)) {
285       unsigned SourceAlign = getOrEnforceKnownAlignment(
286           Copy->getSource(), AI.getAlignment(), DL, &AI, AC, DT);
287       if (AI.getAlignment() <= SourceAlign) {
288         DEBUG(dbgs() << "Found alloca equal to global: " << AI << '\n');
289         DEBUG(dbgs() << "  memcpy = " << *Copy << '\n');
290         for (unsigned i = 0, e = ToDelete.size(); i != e; ++i)
291           EraseInstFromFunction(*ToDelete[i]);
292         Constant *TheSrc = cast<Constant>(Copy->getSource());
293         Constant *Cast
294           = ConstantExpr::getPointerBitCastOrAddrSpaceCast(TheSrc, AI.getType());
295         Instruction *NewI = ReplaceInstUsesWith(AI, Cast);
296         EraseInstFromFunction(*Copy);
297         ++NumGlobalCopies;
298         return NewI;
299       }
300     }
301   }
302 
303   // At last, use the generic allocation site handler to aggressively remove
304   // unused allocas.
305   return visitAllocSite(AI);
306 }
307 
308 /// \brief Helper to combine a load to a new type.
309 ///
310 /// This just does the work of combining a load to a new type. It handles
311 /// metadata, etc., and returns the new instruction. The \c NewTy should be the
312 /// loaded *value* type. This will convert it to a pointer, cast the operand to
313 /// that pointer type, load it, etc.
314 ///
315 /// Note that this will create all of the instructions with whatever insert
316 /// point the \c InstCombiner currently is using.
combineLoadToNewType(InstCombiner & IC,LoadInst & LI,Type * NewTy)317 static LoadInst *combineLoadToNewType(InstCombiner &IC, LoadInst &LI, Type *NewTy) {
318   Value *Ptr = LI.getPointerOperand();
319   unsigned AS = LI.getPointerAddressSpace();
320   SmallVector<std::pair<unsigned, MDNode *>, 8> MD;
321   LI.getAllMetadata(MD);
322 
323   LoadInst *NewLoad = IC.Builder->CreateAlignedLoad(
324       IC.Builder->CreateBitCast(Ptr, NewTy->getPointerTo(AS)),
325       LI.getAlignment(), LI.getName());
326   MDBuilder MDB(NewLoad->getContext());
327   for (const auto &MDPair : MD) {
328     unsigned ID = MDPair.first;
329     MDNode *N = MDPair.second;
330     // Note, essentially every kind of metadata should be preserved here! This
331     // routine is supposed to clone a load instruction changing *only its type*.
332     // The only metadata it makes sense to drop is metadata which is invalidated
333     // when the pointer type changes. This should essentially never be the case
334     // in LLVM, but we explicitly switch over only known metadata to be
335     // conservatively correct. If you are adding metadata to LLVM which pertains
336     // to loads, you almost certainly want to add it here.
337     switch (ID) {
338     case LLVMContext::MD_dbg:
339     case LLVMContext::MD_tbaa:
340     case LLVMContext::MD_prof:
341     case LLVMContext::MD_fpmath:
342     case LLVMContext::MD_tbaa_struct:
343     case LLVMContext::MD_invariant_load:
344     case LLVMContext::MD_alias_scope:
345     case LLVMContext::MD_noalias:
346     case LLVMContext::MD_nontemporal:
347     case LLVMContext::MD_mem_parallel_loop_access:
348       // All of these directly apply.
349       NewLoad->setMetadata(ID, N);
350       break;
351 
352     case LLVMContext::MD_nonnull:
353       // This only directly applies if the new type is also a pointer.
354       if (NewTy->isPointerTy()) {
355         NewLoad->setMetadata(ID, N);
356         break;
357       }
358       // If it's integral now, translate it to !range metadata.
359       if (NewTy->isIntegerTy()) {
360         auto *ITy = cast<IntegerType>(NewTy);
361         auto *NullInt = ConstantExpr::getPtrToInt(
362             ConstantPointerNull::get(cast<PointerType>(Ptr->getType())), ITy);
363         auto *NonNullInt =
364             ConstantExpr::getAdd(NullInt, ConstantInt::get(ITy, 1));
365         NewLoad->setMetadata(LLVMContext::MD_range,
366                              MDB.createRange(NonNullInt, NullInt));
367       }
368       break;
369 
370     case LLVMContext::MD_range:
371       // FIXME: It would be nice to propagate this in some way, but the type
372       // conversions make it hard. If the new type is a pointer, we could
373       // translate it to !nonnull metadata.
374       break;
375     }
376   }
377   return NewLoad;
378 }
379 
380 /// \brief Combine a store to a new type.
381 ///
382 /// Returns the newly created store instruction.
combineStoreToNewValue(InstCombiner & IC,StoreInst & SI,Value * V)383 static StoreInst *combineStoreToNewValue(InstCombiner &IC, StoreInst &SI, Value *V) {
384   Value *Ptr = SI.getPointerOperand();
385   unsigned AS = SI.getPointerAddressSpace();
386   SmallVector<std::pair<unsigned, MDNode *>, 8> MD;
387   SI.getAllMetadata(MD);
388 
389   StoreInst *NewStore = IC.Builder->CreateAlignedStore(
390       V, IC.Builder->CreateBitCast(Ptr, V->getType()->getPointerTo(AS)),
391       SI.getAlignment());
392   for (const auto &MDPair : MD) {
393     unsigned ID = MDPair.first;
394     MDNode *N = MDPair.second;
395     // Note, essentially every kind of metadata should be preserved here! This
396     // routine is supposed to clone a store instruction changing *only its
397     // type*. The only metadata it makes sense to drop is metadata which is
398     // invalidated when the pointer type changes. This should essentially
399     // never be the case in LLVM, but we explicitly switch over only known
400     // metadata to be conservatively correct. If you are adding metadata to
401     // LLVM which pertains to stores, you almost certainly want to add it
402     // here.
403     switch (ID) {
404     case LLVMContext::MD_dbg:
405     case LLVMContext::MD_tbaa:
406     case LLVMContext::MD_prof:
407     case LLVMContext::MD_fpmath:
408     case LLVMContext::MD_tbaa_struct:
409     case LLVMContext::MD_alias_scope:
410     case LLVMContext::MD_noalias:
411     case LLVMContext::MD_nontemporal:
412     case LLVMContext::MD_mem_parallel_loop_access:
413       // All of these directly apply.
414       NewStore->setMetadata(ID, N);
415       break;
416 
417     case LLVMContext::MD_invariant_load:
418     case LLVMContext::MD_nonnull:
419     case LLVMContext::MD_range:
420       // These don't apply for stores.
421       break;
422     }
423   }
424 
425   return NewStore;
426 }
427 
428 /// \brief Combine loads to match the type of value their uses after looking
429 /// through intervening bitcasts.
430 ///
431 /// The core idea here is that if the result of a load is used in an operation,
432 /// we should load the type most conducive to that operation. For example, when
433 /// loading an integer and converting that immediately to a pointer, we should
434 /// instead directly load a pointer.
435 ///
436 /// However, this routine must never change the width of a load or the number of
437 /// loads as that would introduce a semantic change. This combine is expected to
438 /// be a semantic no-op which just allows loads to more closely model the types
439 /// of their consuming operations.
440 ///
441 /// Currently, we also refuse to change the precise type used for an atomic load
442 /// or a volatile load. This is debatable, and might be reasonable to change
443 /// later. However, it is risky in case some backend or other part of LLVM is
444 /// relying on the exact type loaded to select appropriate atomic operations.
combineLoadToOperationType(InstCombiner & IC,LoadInst & LI)445 static Instruction *combineLoadToOperationType(InstCombiner &IC, LoadInst &LI) {
446   // FIXME: We could probably with some care handle both volatile and atomic
447   // loads here but it isn't clear that this is important.
448   if (!LI.isSimple())
449     return nullptr;
450 
451   if (LI.use_empty())
452     return nullptr;
453 
454   Type *Ty = LI.getType();
455   const DataLayout &DL = IC.getDataLayout();
456 
457   // Try to canonicalize loads which are only ever stored to operate over
458   // integers instead of any other type. We only do this when the loaded type
459   // is sized and has a size exactly the same as its store size and the store
460   // size is a legal integer type.
461   if (!Ty->isIntegerTy() && Ty->isSized() &&
462       DL.isLegalInteger(DL.getTypeStoreSizeInBits(Ty)) &&
463       DL.getTypeStoreSizeInBits(Ty) == DL.getTypeSizeInBits(Ty)) {
464     if (std::all_of(LI.user_begin(), LI.user_end(), [&LI](User *U) {
465           auto *SI = dyn_cast<StoreInst>(U);
466           return SI && SI->getPointerOperand() != &LI;
467         })) {
468       LoadInst *NewLoad = combineLoadToNewType(
469           IC, LI,
470           Type::getIntNTy(LI.getContext(), DL.getTypeStoreSizeInBits(Ty)));
471       // Replace all the stores with stores of the newly loaded value.
472       for (auto UI = LI.user_begin(), UE = LI.user_end(); UI != UE;) {
473         auto *SI = cast<StoreInst>(*UI++);
474         IC.Builder->SetInsertPoint(SI);
475         combineStoreToNewValue(IC, *SI, NewLoad);
476         IC.EraseInstFromFunction(*SI);
477       }
478       assert(LI.use_empty() && "Failed to remove all users of the load!");
479       // Return the old load so the combiner can delete it safely.
480       return &LI;
481     }
482   }
483 
484   // Fold away bit casts of the loaded value by loading the desired type.
485   if (LI.hasOneUse())
486     if (auto *BC = dyn_cast<BitCastInst>(LI.user_back())) {
487       LoadInst *NewLoad = combineLoadToNewType(IC, LI, BC->getDestTy());
488       BC->replaceAllUsesWith(NewLoad);
489       IC.EraseInstFromFunction(*BC);
490       return &LI;
491     }
492 
493   // FIXME: We should also canonicalize loads of vectors when their elements are
494   // cast to other types.
495   return nullptr;
496 }
497 
498 // If we can determine that all possible objects pointed to by the provided
499 // pointer value are, not only dereferenceable, but also definitively less than
500 // or equal to the provided maximum size, then return true. Otherwise, return
501 // false (constant global values and allocas fall into this category).
502 //
503 // FIXME: This should probably live in ValueTracking (or similar).
isObjectSizeLessThanOrEq(Value * V,uint64_t MaxSize,const DataLayout & DL)504 static bool isObjectSizeLessThanOrEq(Value *V, uint64_t MaxSize,
505                                      const DataLayout &DL) {
506   SmallPtrSet<Value *, 4> Visited;
507   SmallVector<Value *, 4> Worklist(1, V);
508 
509   do {
510     Value *P = Worklist.pop_back_val();
511     P = P->stripPointerCasts();
512 
513     if (!Visited.insert(P).second)
514       continue;
515 
516     if (SelectInst *SI = dyn_cast<SelectInst>(P)) {
517       Worklist.push_back(SI->getTrueValue());
518       Worklist.push_back(SI->getFalseValue());
519       continue;
520     }
521 
522     if (PHINode *PN = dyn_cast<PHINode>(P)) {
523       for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
524         Worklist.push_back(PN->getIncomingValue(i));
525       continue;
526     }
527 
528     if (GlobalAlias *GA = dyn_cast<GlobalAlias>(P)) {
529       if (GA->mayBeOverridden())
530         return false;
531       Worklist.push_back(GA->getAliasee());
532       continue;
533     }
534 
535     // If we know how big this object is, and it is less than MaxSize, continue
536     // searching. Otherwise, return false.
537     if (AllocaInst *AI = dyn_cast<AllocaInst>(P)) {
538       if (!AI->getAllocatedType()->isSized())
539         return false;
540 
541       ConstantInt *CS = dyn_cast<ConstantInt>(AI->getArraySize());
542       if (!CS)
543         return false;
544 
545       uint64_t TypeSize = DL.getTypeAllocSize(AI->getAllocatedType());
546       // Make sure that, even if the multiplication below would wrap as an
547       // uint64_t, we still do the right thing.
548       if ((CS->getValue().zextOrSelf(128)*APInt(128, TypeSize)).ugt(MaxSize))
549         return false;
550       continue;
551     }
552 
553     if (GlobalVariable *GV = dyn_cast<GlobalVariable>(P)) {
554       if (!GV->hasDefinitiveInitializer() || !GV->isConstant())
555         return false;
556 
557       uint64_t InitSize = DL.getTypeAllocSize(GV->getType()->getElementType());
558       if (InitSize > MaxSize)
559         return false;
560       continue;
561     }
562 
563     return false;
564   } while (!Worklist.empty());
565 
566   return true;
567 }
568 
569 // If we're indexing into an object of a known size, and the outer index is
570 // not a constant, but having any value but zero would lead to undefined
571 // behavior, replace it with zero.
572 //
573 // For example, if we have:
574 // @f.a = private unnamed_addr constant [1 x i32] [i32 12], align 4
575 // ...
576 // %arrayidx = getelementptr inbounds [1 x i32]* @f.a, i64 0, i64 %x
577 // ... = load i32* %arrayidx, align 4
578 // Then we know that we can replace %x in the GEP with i64 0.
579 //
580 // FIXME: We could fold any GEP index to zero that would cause UB if it were
581 // not zero. Currently, we only handle the first such index. Also, we could
582 // also search through non-zero constant indices if we kept track of the
583 // offsets those indices implied.
canReplaceGEPIdxWithZero(InstCombiner & IC,GetElementPtrInst * GEPI,Instruction * MemI,unsigned & Idx)584 static bool canReplaceGEPIdxWithZero(InstCombiner &IC, GetElementPtrInst *GEPI,
585                                      Instruction *MemI, unsigned &Idx) {
586   if (GEPI->getNumOperands() < 2)
587     return false;
588 
589   // Find the first non-zero index of a GEP. If all indices are zero, return
590   // one past the last index.
591   auto FirstNZIdx = [](const GetElementPtrInst *GEPI) {
592     unsigned I = 1;
593     for (unsigned IE = GEPI->getNumOperands(); I != IE; ++I) {
594       Value *V = GEPI->getOperand(I);
595       if (const ConstantInt *CI = dyn_cast<ConstantInt>(V))
596         if (CI->isZero())
597           continue;
598 
599       break;
600     }
601 
602     return I;
603   };
604 
605   // Skip through initial 'zero' indices, and find the corresponding pointer
606   // type. See if the next index is not a constant.
607   Idx = FirstNZIdx(GEPI);
608   if (Idx == GEPI->getNumOperands())
609     return false;
610   if (isa<Constant>(GEPI->getOperand(Idx)))
611     return false;
612 
613   SmallVector<Value *, 4> Ops(GEPI->idx_begin(), GEPI->idx_begin() + Idx);
614   Type *AllocTy = GetElementPtrInst::getIndexedType(
615       cast<PointerType>(GEPI->getOperand(0)->getType()->getScalarType())
616           ->getElementType(),
617       Ops);
618   if (!AllocTy || !AllocTy->isSized())
619     return false;
620   const DataLayout &DL = IC.getDataLayout();
621   uint64_t TyAllocSize = DL.getTypeAllocSize(AllocTy);
622 
623   // If there are more indices after the one we might replace with a zero, make
624   // sure they're all non-negative. If any of them are negative, the overall
625   // address being computed might be before the base address determined by the
626   // first non-zero index.
627   auto IsAllNonNegative = [&]() {
628     for (unsigned i = Idx+1, e = GEPI->getNumOperands(); i != e; ++i) {
629       bool KnownNonNegative, KnownNegative;
630       IC.ComputeSignBit(GEPI->getOperand(i), KnownNonNegative,
631                         KnownNegative, 0, MemI);
632       if (KnownNonNegative)
633         continue;
634       return false;
635     }
636 
637     return true;
638   };
639 
640   // FIXME: If the GEP is not inbounds, and there are extra indices after the
641   // one we'll replace, those could cause the address computation to wrap
642   // (rendering the IsAllNonNegative() check below insufficient). We can do
643   // better, ignoring zero indicies (and other indicies we can prove small
644   // enough not to wrap).
645   if (Idx+1 != GEPI->getNumOperands() && !GEPI->isInBounds())
646     return false;
647 
648   // Note that isObjectSizeLessThanOrEq will return true only if the pointer is
649   // also known to be dereferenceable.
650   return isObjectSizeLessThanOrEq(GEPI->getOperand(0), TyAllocSize, DL) &&
651          IsAllNonNegative();
652 }
653 
654 // If we're indexing into an object with a variable index for the memory
655 // access, but the object has only one element, we can assume that the index
656 // will always be zero. If we replace the GEP, return it.
657 template <typename T>
replaceGEPIdxWithZero(InstCombiner & IC,Value * Ptr,T & MemI)658 static Instruction *replaceGEPIdxWithZero(InstCombiner &IC, Value *Ptr,
659                                           T &MemI) {
660   if (GetElementPtrInst *GEPI = dyn_cast<GetElementPtrInst>(Ptr)) {
661     unsigned Idx;
662     if (canReplaceGEPIdxWithZero(IC, GEPI, &MemI, Idx)) {
663       Instruction *NewGEPI = GEPI->clone();
664       NewGEPI->setOperand(Idx,
665         ConstantInt::get(GEPI->getOperand(Idx)->getType(), 0));
666       NewGEPI->insertBefore(GEPI);
667       MemI.setOperand(MemI.getPointerOperandIndex(), NewGEPI);
668       return NewGEPI;
669     }
670   }
671 
672   return nullptr;
673 }
674 
visitLoadInst(LoadInst & LI)675 Instruction *InstCombiner::visitLoadInst(LoadInst &LI) {
676   Value *Op = LI.getOperand(0);
677 
678   // Try to canonicalize the loaded type.
679   if (Instruction *Res = combineLoadToOperationType(*this, LI))
680     return Res;
681 
682   // Attempt to improve the alignment.
683   unsigned KnownAlign = getOrEnforceKnownAlignment(
684       Op, DL.getPrefTypeAlignment(LI.getType()), DL, &LI, AC, DT);
685   unsigned LoadAlign = LI.getAlignment();
686   unsigned EffectiveLoadAlign =
687       LoadAlign != 0 ? LoadAlign : DL.getABITypeAlignment(LI.getType());
688 
689   if (KnownAlign > EffectiveLoadAlign)
690     LI.setAlignment(KnownAlign);
691   else if (LoadAlign == 0)
692     LI.setAlignment(EffectiveLoadAlign);
693 
694   // Replace GEP indices if possible.
695   if (Instruction *NewGEPI = replaceGEPIdxWithZero(*this, Op, LI)) {
696       Worklist.Add(NewGEPI);
697       return &LI;
698   }
699 
700   // None of the following transforms are legal for volatile/atomic loads.
701   // FIXME: Some of it is okay for atomic loads; needs refactoring.
702   if (!LI.isSimple()) return nullptr;
703 
704   // Do really simple store-to-load forwarding and load CSE, to catch cases
705   // where there are several consecutive memory accesses to the same location,
706   // separated by a few arithmetic operations.
707   BasicBlock::iterator BBI = &LI;
708   if (Value *AvailableVal = FindAvailableLoadedValue(Op, LI.getParent(), BBI,6))
709     return ReplaceInstUsesWith(
710         LI, Builder->CreateBitOrPointerCast(AvailableVal, LI.getType(),
711                                             LI.getName() + ".cast"));
712 
713   // load(gep null, ...) -> unreachable
714   if (GetElementPtrInst *GEPI = dyn_cast<GetElementPtrInst>(Op)) {
715     const Value *GEPI0 = GEPI->getOperand(0);
716     // TODO: Consider a target hook for valid address spaces for this xform.
717     if (isa<ConstantPointerNull>(GEPI0) && GEPI->getPointerAddressSpace() == 0){
718       // Insert a new store to null instruction before the load to indicate
719       // that this code is not reachable.  We do this instead of inserting
720       // an unreachable instruction directly because we cannot modify the
721       // CFG.
722       new StoreInst(UndefValue::get(LI.getType()),
723                     Constant::getNullValue(Op->getType()), &LI);
724       return ReplaceInstUsesWith(LI, UndefValue::get(LI.getType()));
725     }
726   }
727 
728   // load null/undef -> unreachable
729   // TODO: Consider a target hook for valid address spaces for this xform.
730   if (isa<UndefValue>(Op) ||
731       (isa<ConstantPointerNull>(Op) && LI.getPointerAddressSpace() == 0)) {
732     // Insert a new store to null instruction before the load to indicate that
733     // this code is not reachable.  We do this instead of inserting an
734     // unreachable instruction directly because we cannot modify the CFG.
735     new StoreInst(UndefValue::get(LI.getType()),
736                   Constant::getNullValue(Op->getType()), &LI);
737     return ReplaceInstUsesWith(LI, UndefValue::get(LI.getType()));
738   }
739 
740   if (Op->hasOneUse()) {
741     // Change select and PHI nodes to select values instead of addresses: this
742     // helps alias analysis out a lot, allows many others simplifications, and
743     // exposes redundancy in the code.
744     //
745     // Note that we cannot do the transformation unless we know that the
746     // introduced loads cannot trap!  Something like this is valid as long as
747     // the condition is always false: load (select bool %C, int* null, int* %G),
748     // but it would not be valid if we transformed it to load from null
749     // unconditionally.
750     //
751     if (SelectInst *SI = dyn_cast<SelectInst>(Op)) {
752       // load (select (Cond, &V1, &V2))  --> select(Cond, load &V1, load &V2).
753       unsigned Align = LI.getAlignment();
754       if (isSafeToLoadUnconditionally(SI->getOperand(1), SI, Align) &&
755           isSafeToLoadUnconditionally(SI->getOperand(2), SI, Align)) {
756         LoadInst *V1 = Builder->CreateLoad(SI->getOperand(1),
757                                            SI->getOperand(1)->getName()+".val");
758         LoadInst *V2 = Builder->CreateLoad(SI->getOperand(2),
759                                            SI->getOperand(2)->getName()+".val");
760         V1->setAlignment(Align);
761         V2->setAlignment(Align);
762         return SelectInst::Create(SI->getCondition(), V1, V2);
763       }
764 
765       // load (select (cond, null, P)) -> load P
766       if (isa<ConstantPointerNull>(SI->getOperand(1)) &&
767           LI.getPointerAddressSpace() == 0) {
768         LI.setOperand(0, SI->getOperand(2));
769         return &LI;
770       }
771 
772       // load (select (cond, P, null)) -> load P
773       if (isa<ConstantPointerNull>(SI->getOperand(2)) &&
774           LI.getPointerAddressSpace() == 0) {
775         LI.setOperand(0, SI->getOperand(1));
776         return &LI;
777       }
778     }
779   }
780   return nullptr;
781 }
782 
783 /// \brief Combine stores to match the type of value being stored.
784 ///
785 /// The core idea here is that the memory does not have any intrinsic type and
786 /// where we can we should match the type of a store to the type of value being
787 /// stored.
788 ///
789 /// However, this routine must never change the width of a store or the number of
790 /// stores as that would introduce a semantic change. This combine is expected to
791 /// be a semantic no-op which just allows stores to more closely model the types
792 /// of their incoming values.
793 ///
794 /// Currently, we also refuse to change the precise type used for an atomic or
795 /// volatile store. This is debatable, and might be reasonable to change later.
796 /// However, it is risky in case some backend or other part of LLVM is relying
797 /// on the exact type stored to select appropriate atomic operations.
798 ///
799 /// \returns true if the store was successfully combined away. This indicates
800 /// the caller must erase the store instruction. We have to let the caller erase
801 /// the store instruction sas otherwise there is no way to signal whether it was
802 /// combined or not: IC.EraseInstFromFunction returns a null pointer.
combineStoreToValueType(InstCombiner & IC,StoreInst & SI)803 static bool combineStoreToValueType(InstCombiner &IC, StoreInst &SI) {
804   // FIXME: We could probably with some care handle both volatile and atomic
805   // stores here but it isn't clear that this is important.
806   if (!SI.isSimple())
807     return false;
808 
809   Value *V = SI.getValueOperand();
810 
811   // Fold away bit casts of the stored value by storing the original type.
812   if (auto *BC = dyn_cast<BitCastInst>(V)) {
813     V = BC->getOperand(0);
814     combineStoreToNewValue(IC, SI, V);
815     return true;
816   }
817 
818   // FIXME: We should also canonicalize loads of vectors when their elements are
819   // cast to other types.
820   return false;
821 }
822 
unpackStoreToAggregate(InstCombiner & IC,StoreInst & SI)823 static bool unpackStoreToAggregate(InstCombiner &IC, StoreInst &SI) {
824   // FIXME: We could probably with some care handle both volatile and atomic
825   // stores here but it isn't clear that this is important.
826   if (!SI.isSimple())
827     return false;
828 
829   Value *V = SI.getValueOperand();
830   Type *T = V->getType();
831 
832   if (!T->isAggregateType())
833     return false;
834 
835   if (StructType *ST = dyn_cast<StructType>(T)) {
836     // If the struct only have one element, we unpack.
837     if (ST->getNumElements() == 1) {
838       V = IC.Builder->CreateExtractValue(V, 0);
839       combineStoreToNewValue(IC, SI, V);
840       return true;
841     }
842   }
843 
844   return false;
845 }
846 
847 /// equivalentAddressValues - Test if A and B will obviously have the same
848 /// value. This includes recognizing that %t0 and %t1 will have the same
849 /// value in code like this:
850 ///   %t0 = getelementptr \@a, 0, 3
851 ///   store i32 0, i32* %t0
852 ///   %t1 = getelementptr \@a, 0, 3
853 ///   %t2 = load i32* %t1
854 ///
equivalentAddressValues(Value * A,Value * B)855 static bool equivalentAddressValues(Value *A, Value *B) {
856   // Test if the values are trivially equivalent.
857   if (A == B) return true;
858 
859   // Test if the values come form identical arithmetic instructions.
860   // This uses isIdenticalToWhenDefined instead of isIdenticalTo because
861   // its only used to compare two uses within the same basic block, which
862   // means that they'll always either have the same value or one of them
863   // will have an undefined value.
864   if (isa<BinaryOperator>(A) ||
865       isa<CastInst>(A) ||
866       isa<PHINode>(A) ||
867       isa<GetElementPtrInst>(A))
868     if (Instruction *BI = dyn_cast<Instruction>(B))
869       if (cast<Instruction>(A)->isIdenticalToWhenDefined(BI))
870         return true;
871 
872   // Otherwise they may not be equivalent.
873   return false;
874 }
875 
visitStoreInst(StoreInst & SI)876 Instruction *InstCombiner::visitStoreInst(StoreInst &SI) {
877   Value *Val = SI.getOperand(0);
878   Value *Ptr = SI.getOperand(1);
879 
880   // Try to canonicalize the stored type.
881   if (combineStoreToValueType(*this, SI))
882     return EraseInstFromFunction(SI);
883 
884   // Attempt to improve the alignment.
885   unsigned KnownAlign = getOrEnforceKnownAlignment(
886       Ptr, DL.getPrefTypeAlignment(Val->getType()), DL, &SI, AC, DT);
887   unsigned StoreAlign = SI.getAlignment();
888   unsigned EffectiveStoreAlign =
889       StoreAlign != 0 ? StoreAlign : DL.getABITypeAlignment(Val->getType());
890 
891   if (KnownAlign > EffectiveStoreAlign)
892     SI.setAlignment(KnownAlign);
893   else if (StoreAlign == 0)
894     SI.setAlignment(EffectiveStoreAlign);
895 
896   // Try to canonicalize the stored type.
897   if (unpackStoreToAggregate(*this, SI))
898     return EraseInstFromFunction(SI);
899 
900   // Replace GEP indices if possible.
901   if (Instruction *NewGEPI = replaceGEPIdxWithZero(*this, Ptr, SI)) {
902       Worklist.Add(NewGEPI);
903       return &SI;
904   }
905 
906   // Don't hack volatile/atomic stores.
907   // FIXME: Some bits are legal for atomic stores; needs refactoring.
908   if (!SI.isSimple()) return nullptr;
909 
910   // If the RHS is an alloca with a single use, zapify the store, making the
911   // alloca dead.
912   if (Ptr->hasOneUse()) {
913     if (isa<AllocaInst>(Ptr))
914       return EraseInstFromFunction(SI);
915     if (GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(Ptr)) {
916       if (isa<AllocaInst>(GEP->getOperand(0))) {
917         if (GEP->getOperand(0)->hasOneUse())
918           return EraseInstFromFunction(SI);
919       }
920     }
921   }
922 
923   // Do really simple DSE, to catch cases where there are several consecutive
924   // stores to the same location, separated by a few arithmetic operations. This
925   // situation often occurs with bitfield accesses.
926   BasicBlock::iterator BBI = &SI;
927   for (unsigned ScanInsts = 6; BBI != SI.getParent()->begin() && ScanInsts;
928        --ScanInsts) {
929     --BBI;
930     // Don't count debug info directives, lest they affect codegen,
931     // and we skip pointer-to-pointer bitcasts, which are NOPs.
932     if (isa<DbgInfoIntrinsic>(BBI) ||
933         (isa<BitCastInst>(BBI) && BBI->getType()->isPointerTy())) {
934       ScanInsts++;
935       continue;
936     }
937 
938     if (StoreInst *PrevSI = dyn_cast<StoreInst>(BBI)) {
939       // Prev store isn't volatile, and stores to the same location?
940       if (PrevSI->isSimple() && equivalentAddressValues(PrevSI->getOperand(1),
941                                                         SI.getOperand(1))) {
942         ++NumDeadStore;
943         ++BBI;
944         EraseInstFromFunction(*PrevSI);
945         continue;
946       }
947       break;
948     }
949 
950     // If this is a load, we have to stop.  However, if the loaded value is from
951     // the pointer we're loading and is producing the pointer we're storing,
952     // then *this* store is dead (X = load P; store X -> P).
953     if (LoadInst *LI = dyn_cast<LoadInst>(BBI)) {
954       if (LI == Val && equivalentAddressValues(LI->getOperand(0), Ptr) &&
955           LI->isSimple())
956         return EraseInstFromFunction(SI);
957 
958       // Otherwise, this is a load from some other location.  Stores before it
959       // may not be dead.
960       break;
961     }
962 
963     // Don't skip over loads or things that can modify memory.
964     if (BBI->mayWriteToMemory() || BBI->mayReadFromMemory())
965       break;
966   }
967 
968   // store X, null    -> turns into 'unreachable' in SimplifyCFG
969   if (isa<ConstantPointerNull>(Ptr) && SI.getPointerAddressSpace() == 0) {
970     if (!isa<UndefValue>(Val)) {
971       SI.setOperand(0, UndefValue::get(Val->getType()));
972       if (Instruction *U = dyn_cast<Instruction>(Val))
973         Worklist.Add(U);  // Dropped a use.
974     }
975     return nullptr;  // Do not modify these!
976   }
977 
978   // store undef, Ptr -> noop
979   if (isa<UndefValue>(Val))
980     return EraseInstFromFunction(SI);
981 
982   // If this store is the last instruction in the basic block (possibly
983   // excepting debug info instructions), and if the block ends with an
984   // unconditional branch, try to move it to the successor block.
985   BBI = &SI;
986   do {
987     ++BBI;
988   } while (isa<DbgInfoIntrinsic>(BBI) ||
989            (isa<BitCastInst>(BBI) && BBI->getType()->isPointerTy()));
990   if (BranchInst *BI = dyn_cast<BranchInst>(BBI))
991     if (BI->isUnconditional())
992       if (SimplifyStoreAtEndOfBlock(SI))
993         return nullptr;  // xform done!
994 
995   return nullptr;
996 }
997 
998 /// SimplifyStoreAtEndOfBlock - Turn things like:
999 ///   if () { *P = v1; } else { *P = v2 }
1000 /// into a phi node with a store in the successor.
1001 ///
1002 /// Simplify things like:
1003 ///   *P = v1; if () { *P = v2; }
1004 /// into a phi node with a store in the successor.
1005 ///
SimplifyStoreAtEndOfBlock(StoreInst & SI)1006 bool InstCombiner::SimplifyStoreAtEndOfBlock(StoreInst &SI) {
1007   BasicBlock *StoreBB = SI.getParent();
1008 
1009   // Check to see if the successor block has exactly two incoming edges.  If
1010   // so, see if the other predecessor contains a store to the same location.
1011   // if so, insert a PHI node (if needed) and move the stores down.
1012   BasicBlock *DestBB = StoreBB->getTerminator()->getSuccessor(0);
1013 
1014   // Determine whether Dest has exactly two predecessors and, if so, compute
1015   // the other predecessor.
1016   pred_iterator PI = pred_begin(DestBB);
1017   BasicBlock *P = *PI;
1018   BasicBlock *OtherBB = nullptr;
1019 
1020   if (P != StoreBB)
1021     OtherBB = P;
1022 
1023   if (++PI == pred_end(DestBB))
1024     return false;
1025 
1026   P = *PI;
1027   if (P != StoreBB) {
1028     if (OtherBB)
1029       return false;
1030     OtherBB = P;
1031   }
1032   if (++PI != pred_end(DestBB))
1033     return false;
1034 
1035   // Bail out if all the relevant blocks aren't distinct (this can happen,
1036   // for example, if SI is in an infinite loop)
1037   if (StoreBB == DestBB || OtherBB == DestBB)
1038     return false;
1039 
1040   // Verify that the other block ends in a branch and is not otherwise empty.
1041   BasicBlock::iterator BBI = OtherBB->getTerminator();
1042   BranchInst *OtherBr = dyn_cast<BranchInst>(BBI);
1043   if (!OtherBr || BBI == OtherBB->begin())
1044     return false;
1045 
1046   // If the other block ends in an unconditional branch, check for the 'if then
1047   // else' case.  there is an instruction before the branch.
1048   StoreInst *OtherStore = nullptr;
1049   if (OtherBr->isUnconditional()) {
1050     --BBI;
1051     // Skip over debugging info.
1052     while (isa<DbgInfoIntrinsic>(BBI) ||
1053            (isa<BitCastInst>(BBI) && BBI->getType()->isPointerTy())) {
1054       if (BBI==OtherBB->begin())
1055         return false;
1056       --BBI;
1057     }
1058     // If this isn't a store, isn't a store to the same location, or is not the
1059     // right kind of store, bail out.
1060     OtherStore = dyn_cast<StoreInst>(BBI);
1061     if (!OtherStore || OtherStore->getOperand(1) != SI.getOperand(1) ||
1062         !SI.isSameOperationAs(OtherStore))
1063       return false;
1064   } else {
1065     // Otherwise, the other block ended with a conditional branch. If one of the
1066     // destinations is StoreBB, then we have the if/then case.
1067     if (OtherBr->getSuccessor(0) != StoreBB &&
1068         OtherBr->getSuccessor(1) != StoreBB)
1069       return false;
1070 
1071     // Okay, we know that OtherBr now goes to Dest and StoreBB, so this is an
1072     // if/then triangle.  See if there is a store to the same ptr as SI that
1073     // lives in OtherBB.
1074     for (;; --BBI) {
1075       // Check to see if we find the matching store.
1076       if ((OtherStore = dyn_cast<StoreInst>(BBI))) {
1077         if (OtherStore->getOperand(1) != SI.getOperand(1) ||
1078             !SI.isSameOperationAs(OtherStore))
1079           return false;
1080         break;
1081       }
1082       // If we find something that may be using or overwriting the stored
1083       // value, or if we run out of instructions, we can't do the xform.
1084       if (BBI->mayReadFromMemory() || BBI->mayWriteToMemory() ||
1085           BBI == OtherBB->begin())
1086         return false;
1087     }
1088 
1089     // In order to eliminate the store in OtherBr, we have to
1090     // make sure nothing reads or overwrites the stored value in
1091     // StoreBB.
1092     for (BasicBlock::iterator I = StoreBB->begin(); &*I != &SI; ++I) {
1093       // FIXME: This should really be AA driven.
1094       if (I->mayReadFromMemory() || I->mayWriteToMemory())
1095         return false;
1096     }
1097   }
1098 
1099   // Insert a PHI node now if we need it.
1100   Value *MergedVal = OtherStore->getOperand(0);
1101   if (MergedVal != SI.getOperand(0)) {
1102     PHINode *PN = PHINode::Create(MergedVal->getType(), 2, "storemerge");
1103     PN->addIncoming(SI.getOperand(0), SI.getParent());
1104     PN->addIncoming(OtherStore->getOperand(0), OtherBB);
1105     MergedVal = InsertNewInstBefore(PN, DestBB->front());
1106   }
1107 
1108   // Advance to a place where it is safe to insert the new store and
1109   // insert it.
1110   BBI = DestBB->getFirstInsertionPt();
1111   StoreInst *NewSI = new StoreInst(MergedVal, SI.getOperand(1),
1112                                    SI.isVolatile(),
1113                                    SI.getAlignment(),
1114                                    SI.getOrdering(),
1115                                    SI.getSynchScope());
1116   InsertNewInstBefore(NewSI, *BBI);
1117   NewSI->setDebugLoc(OtherStore->getDebugLoc());
1118 
1119   // If the two stores had AA tags, merge them.
1120   AAMDNodes AATags;
1121   SI.getAAMetadata(AATags);
1122   if (AATags) {
1123     OtherStore->getAAMetadata(AATags, /* Merge = */ true);
1124     NewSI->setAAMetadata(AATags);
1125   }
1126 
1127   // Nuke the old stores.
1128   EraseInstFromFunction(SI);
1129   EraseInstFromFunction(*OtherStore);
1130   return true;
1131 }
1132