1 //===- InstCombineLoadStoreAlloca.cpp -------------------------------------===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This file implements the visit functions for load, store and alloca.
12 //===----------------------------------------------------------------------===//
14 #include "InstCombine.h"
15 #include "llvm/IntrinsicInst.h"
16 #include "llvm/Analysis/Loads.h"
17 #include "llvm/Target/TargetData.h"
18 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
19 #include "llvm/Transforms/Utils/Local.h"
20 #include "llvm/ADT/Statistic.h"
23 STATISTIC(NumDeadStore, "Number of dead stores eliminated");
25 Instruction *InstCombiner::visitAllocaInst(AllocaInst &AI) {
26 // Ensure that the alloca array size argument has type intptr_t, so that
27 // any casting is exposed early.
29 Type *IntPtrTy = TD->getIntPtrType(AI.getContext());
30 if (AI.getArraySize()->getType() != IntPtrTy) {
31 Value *V = Builder->CreateIntCast(AI.getArraySize(),
38 // Convert: alloca Ty, C - where C is a constant != 1 into: alloca [C x Ty], 1
39 if (AI.isArrayAllocation()) { // Check C != 1
40 if (const ConstantInt *C = dyn_cast<ConstantInt>(AI.getArraySize())) {
42 ArrayType::get(AI.getAllocatedType(), C->getZExtValue());
43 assert(isa<AllocaInst>(AI) && "Unknown type of allocation inst!");
44 AllocaInst *New = Builder->CreateAlloca(NewTy, 0, AI.getName());
45 New->setAlignment(AI.getAlignment());
47 // Scan to the end of the allocation instructions, to skip over a block of
48 // allocas if possible...also skip interleaved debug info
50 BasicBlock::iterator It = New;
51 while (isa<AllocaInst>(*It) || isa<DbgInfoIntrinsic>(*It)) ++It;
53 // Now that I is pointing to the first non-allocation-inst in the block,
54 // insert our getelementptr instruction...
56 Value *NullIdx =Constant::getNullValue(Type::getInt32Ty(AI.getContext()));
61 GetElementPtrInst::CreateInBounds(New, Idx, New->getName()+".sub");
62 InsertNewInstBefore(GEP, *It);
64 // Now make everything use the getelementptr instead of the original
66 return ReplaceInstUsesWith(AI, GEP);
67 } else if (isa<UndefValue>(AI.getArraySize())) {
68 return ReplaceInstUsesWith(AI, Constant::getNullValue(AI.getType()));
72 if (TD && isa<AllocaInst>(AI) && AI.getAllocatedType()->isSized()) {
73 // If alloca'ing a zero byte object, replace the alloca with a null pointer.
74 // Note that we only do this for alloca's, because malloc should allocate
75 // and return a unique pointer, even for a zero byte allocation.
76 if (TD->getTypeAllocSize(AI.getAllocatedType()) == 0)
77 return ReplaceInstUsesWith(AI, Constant::getNullValue(AI.getType()));
79 // If the alignment is 0 (unspecified), assign it the preferred alignment.
80 if (AI.getAlignment() == 0)
81 AI.setAlignment(TD->getPrefTypeAlignment(AI.getAllocatedType()));
88 /// InstCombineLoadCast - Fold 'load (cast P)' -> cast (load P)' when possible.
89 static Instruction *InstCombineLoadCast(InstCombiner &IC, LoadInst &LI,
90 const TargetData *TD) {
91 User *CI = cast<User>(LI.getOperand(0));
92 Value *CastOp = CI->getOperand(0);
94 PointerType *DestTy = cast<PointerType>(CI->getType());
95 Type *DestPTy = DestTy->getElementType();
96 if (PointerType *SrcTy = dyn_cast<PointerType>(CastOp->getType())) {
98 // If the address spaces don't match, don't eliminate the cast.
99 if (DestTy->getAddressSpace() != SrcTy->getAddressSpace())
102 Type *SrcPTy = SrcTy->getElementType();
104 if (DestPTy->isIntegerTy() || DestPTy->isPointerTy() ||
105 DestPTy->isVectorTy()) {
106 // If the source is an array, the code below will not succeed. Check to
107 // see if a trivial 'gep P, 0, 0' will help matters. Only do this for
109 if (ArrayType *ASrcTy = dyn_cast<ArrayType>(SrcPTy))
110 if (Constant *CSrc = dyn_cast<Constant>(CastOp))
111 if (ASrcTy->getNumElements() != 0) {
113 Idxs[0] = Constant::getNullValue(Type::getInt32Ty(LI.getContext()));
115 CastOp = ConstantExpr::getGetElementPtr(CSrc, Idxs);
116 SrcTy = cast<PointerType>(CastOp->getType());
117 SrcPTy = SrcTy->getElementType();
120 if (IC.getTargetData() &&
121 (SrcPTy->isIntegerTy() || SrcPTy->isPointerTy() ||
122 SrcPTy->isVectorTy()) &&
123 // Do not allow turning this into a load of an integer, which is then
124 // casted to a pointer, this pessimizes pointer analysis a lot.
125 (SrcPTy->isPointerTy() == LI.getType()->isPointerTy()) &&
126 IC.getTargetData()->getTypeSizeInBits(SrcPTy) ==
127 IC.getTargetData()->getTypeSizeInBits(DestPTy)) {
129 // Okay, we are casting from one integer or pointer type to another of
130 // the same size. Instead of casting the pointer before the load, cast
131 // the result of the loaded value.
133 IC.Builder->CreateLoad(CastOp, LI.isVolatile(), CI->getName());
134 NewLoad->setAlignment(LI.getAlignment());
135 NewLoad->setAtomic(LI.getOrdering(), LI.getSynchScope());
136 // Now cast the result of the load.
137 return new BitCastInst(NewLoad, LI.getType());
144 Instruction *InstCombiner::visitLoadInst(LoadInst &LI) {
145 Value *Op = LI.getOperand(0);
147 // Attempt to improve the alignment.
149 unsigned KnownAlign =
150 getOrEnforceKnownAlignment(Op, TD->getPrefTypeAlignment(LI.getType()),TD);
151 unsigned LoadAlign = LI.getAlignment();
152 unsigned EffectiveLoadAlign = LoadAlign != 0 ? LoadAlign :
153 TD->getABITypeAlignment(LI.getType());
155 if (KnownAlign > EffectiveLoadAlign)
156 LI.setAlignment(KnownAlign);
157 else if (LoadAlign == 0)
158 LI.setAlignment(EffectiveLoadAlign);
161 // load (cast X) --> cast (load X) iff safe.
162 if (isa<CastInst>(Op))
163 if (Instruction *Res = InstCombineLoadCast(*this, LI, TD))
166 // None of the following transforms are legal for volatile/atomic loads.
167 // FIXME: Some of it is okay for atomic loads; needs refactoring.
168 if (!LI.isSimple()) return 0;
170 // Do really simple store-to-load forwarding and load CSE, to catch cases
171 // where there are several consecutive memory accesses to the same location,
172 // separated by a few arithmetic operations.
173 BasicBlock::iterator BBI = &LI;
174 if (Value *AvailableVal = FindAvailableLoadedValue(Op, LI.getParent(), BBI,6))
175 return ReplaceInstUsesWith(LI, AvailableVal);
177 // load(gep null, ...) -> unreachable
178 if (GetElementPtrInst *GEPI = dyn_cast<GetElementPtrInst>(Op)) {
179 const Value *GEPI0 = GEPI->getOperand(0);
180 // TODO: Consider a target hook for valid address spaces for this xform.
181 if (isa<ConstantPointerNull>(GEPI0) && GEPI->getPointerAddressSpace() == 0){
182 // Insert a new store to null instruction before the load to indicate
183 // that this code is not reachable. We do this instead of inserting
184 // an unreachable instruction directly because we cannot modify the
186 new StoreInst(UndefValue::get(LI.getType()),
187 Constant::getNullValue(Op->getType()), &LI);
188 return ReplaceInstUsesWith(LI, UndefValue::get(LI.getType()));
192 // load null/undef -> unreachable
193 // TODO: Consider a target hook for valid address spaces for this xform.
194 if (isa<UndefValue>(Op) ||
195 (isa<ConstantPointerNull>(Op) && LI.getPointerAddressSpace() == 0)) {
196 // Insert a new store to null instruction before the load to indicate that
197 // this code is not reachable. We do this instead of inserting an
198 // unreachable instruction directly because we cannot modify the CFG.
199 new StoreInst(UndefValue::get(LI.getType()),
200 Constant::getNullValue(Op->getType()), &LI);
201 return ReplaceInstUsesWith(LI, UndefValue::get(LI.getType()));
204 // Instcombine load (constantexpr_cast global) -> cast (load global)
205 if (ConstantExpr *CE = dyn_cast<ConstantExpr>(Op))
207 if (Instruction *Res = InstCombineLoadCast(*this, LI, TD))
210 if (Op->hasOneUse()) {
211 // Change select and PHI nodes to select values instead of addresses: this
212 // helps alias analysis out a lot, allows many others simplifications, and
213 // exposes redundancy in the code.
215 // Note that we cannot do the transformation unless we know that the
216 // introduced loads cannot trap! Something like this is valid as long as
217 // the condition is always false: load (select bool %C, int* null, int* %G),
218 // but it would not be valid if we transformed it to load from null
221 if (SelectInst *SI = dyn_cast<SelectInst>(Op)) {
222 // load (select (Cond, &V1, &V2)) --> select(Cond, load &V1, load &V2).
223 unsigned Align = LI.getAlignment();
224 if (isSafeToLoadUnconditionally(SI->getOperand(1), SI, Align, TD) &&
225 isSafeToLoadUnconditionally(SI->getOperand(2), SI, Align, TD)) {
226 LoadInst *V1 = Builder->CreateLoad(SI->getOperand(1),
227 SI->getOperand(1)->getName()+".val");
228 LoadInst *V2 = Builder->CreateLoad(SI->getOperand(2),
229 SI->getOperand(2)->getName()+".val");
230 V1->setAlignment(Align);
231 V2->setAlignment(Align);
232 return SelectInst::Create(SI->getCondition(), V1, V2);
235 // load (select (cond, null, P)) -> load P
236 if (Constant *C = dyn_cast<Constant>(SI->getOperand(1)))
237 if (C->isNullValue()) {
238 LI.setOperand(0, SI->getOperand(2));
242 // load (select (cond, P, null)) -> load P
243 if (Constant *C = dyn_cast<Constant>(SI->getOperand(2)))
244 if (C->isNullValue()) {
245 LI.setOperand(0, SI->getOperand(1));
253 /// InstCombineStoreToCast - Fold store V, (cast P) -> store (cast V), P
254 /// when possible. This makes it generally easy to do alias analysis and/or
255 /// SROA/mem2reg of the memory object.
256 static Instruction *InstCombineStoreToCast(InstCombiner &IC, StoreInst &SI) {
257 User *CI = cast<User>(SI.getOperand(1));
258 Value *CastOp = CI->getOperand(0);
260 Type *DestPTy = cast<PointerType>(CI->getType())->getElementType();
261 PointerType *SrcTy = dyn_cast<PointerType>(CastOp->getType());
262 if (SrcTy == 0) return 0;
264 Type *SrcPTy = SrcTy->getElementType();
266 if (!DestPTy->isIntegerTy() && !DestPTy->isPointerTy())
269 /// NewGEPIndices - If SrcPTy is an aggregate type, we can emit a "noop gep"
270 /// to its first element. This allows us to handle things like:
271 /// store i32 xxx, (bitcast {foo*, float}* %P to i32*)
273 SmallVector<Value*, 4> NewGEPIndices;
275 // If the source is an array, the code below will not succeed. Check to
276 // see if a trivial 'gep P, 0, 0' will help matters. Only do this for
278 if (SrcPTy->isArrayTy() || SrcPTy->isStructTy()) {
279 // Index through pointer.
280 Constant *Zero = Constant::getNullValue(Type::getInt32Ty(SI.getContext()));
281 NewGEPIndices.push_back(Zero);
284 if (StructType *STy = dyn_cast<StructType>(SrcPTy)) {
285 if (!STy->getNumElements()) /* Struct can be empty {} */
287 NewGEPIndices.push_back(Zero);
288 SrcPTy = STy->getElementType(0);
289 } else if (ArrayType *ATy = dyn_cast<ArrayType>(SrcPTy)) {
290 NewGEPIndices.push_back(Zero);
291 SrcPTy = ATy->getElementType();
297 SrcTy = PointerType::get(SrcPTy, SrcTy->getAddressSpace());
300 if (!SrcPTy->isIntegerTy() && !SrcPTy->isPointerTy())
303 // If the pointers point into different address spaces or if they point to
304 // values with different sizes, we can't do the transformation.
305 if (!IC.getTargetData() ||
306 SrcTy->getAddressSpace() !=
307 cast<PointerType>(CI->getType())->getAddressSpace() ||
308 IC.getTargetData()->getTypeSizeInBits(SrcPTy) !=
309 IC.getTargetData()->getTypeSizeInBits(DestPTy))
312 // Okay, we are casting from one integer or pointer type to another of
313 // the same size. Instead of casting the pointer before
314 // the store, cast the value to be stored.
316 Value *SIOp0 = SI.getOperand(0);
317 Instruction::CastOps opcode = Instruction::BitCast;
318 Type* CastSrcTy = SIOp0->getType();
319 Type* CastDstTy = SrcPTy;
320 if (CastDstTy->isPointerTy()) {
321 if (CastSrcTy->isIntegerTy())
322 opcode = Instruction::IntToPtr;
323 } else if (CastDstTy->isIntegerTy()) {
324 if (SIOp0->getType()->isPointerTy())
325 opcode = Instruction::PtrToInt;
328 // SIOp0 is a pointer to aggregate and this is a store to the first field,
329 // emit a GEP to index into its first field.
330 if (!NewGEPIndices.empty())
331 CastOp = IC.Builder->CreateInBoundsGEP(CastOp, NewGEPIndices);
333 NewCast = IC.Builder->CreateCast(opcode, SIOp0, CastDstTy,
334 SIOp0->getName()+".c");
335 SI.setOperand(0, NewCast);
336 SI.setOperand(1, CastOp);
340 /// equivalentAddressValues - Test if A and B will obviously have the same
341 /// value. This includes recognizing that %t0 and %t1 will have the same
342 /// value in code like this:
343 /// %t0 = getelementptr \@a, 0, 3
344 /// store i32 0, i32* %t0
345 /// %t1 = getelementptr \@a, 0, 3
346 /// %t2 = load i32* %t1
348 static bool equivalentAddressValues(Value *A, Value *B) {
349 // Test if the values are trivially equivalent.
350 if (A == B) return true;
352 // Test if the values come form identical arithmetic instructions.
353 // This uses isIdenticalToWhenDefined instead of isIdenticalTo because
354 // its only used to compare two uses within the same basic block, which
355 // means that they'll always either have the same value or one of them
356 // will have an undefined value.
357 if (isa<BinaryOperator>(A) ||
360 isa<GetElementPtrInst>(A))
361 if (Instruction *BI = dyn_cast<Instruction>(B))
362 if (cast<Instruction>(A)->isIdenticalToWhenDefined(BI))
365 // Otherwise they may not be equivalent.
369 Instruction *InstCombiner::visitStoreInst(StoreInst &SI) {
370 Value *Val = SI.getOperand(0);
371 Value *Ptr = SI.getOperand(1);
373 // Attempt to improve the alignment.
375 unsigned KnownAlign =
376 getOrEnforceKnownAlignment(Ptr, TD->getPrefTypeAlignment(Val->getType()),
378 unsigned StoreAlign = SI.getAlignment();
379 unsigned EffectiveStoreAlign = StoreAlign != 0 ? StoreAlign :
380 TD->getABITypeAlignment(Val->getType());
382 if (KnownAlign > EffectiveStoreAlign)
383 SI.setAlignment(KnownAlign);
384 else if (StoreAlign == 0)
385 SI.setAlignment(EffectiveStoreAlign);
388 // Don't hack volatile/atomic stores.
389 // FIXME: Some bits are legal for atomic stores; needs refactoring.
390 if (!SI.isSimple()) return 0;
392 // If the RHS is an alloca with a single use, zapify the store, making the
394 if (Ptr->hasOneUse()) {
395 if (isa<AllocaInst>(Ptr))
396 return EraseInstFromFunction(SI);
397 if (GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(Ptr)) {
398 if (isa<AllocaInst>(GEP->getOperand(0))) {
399 if (GEP->getOperand(0)->hasOneUse())
400 return EraseInstFromFunction(SI);
405 // Do really simple DSE, to catch cases where there are several consecutive
406 // stores to the same location, separated by a few arithmetic operations. This
407 // situation often occurs with bitfield accesses.
408 BasicBlock::iterator BBI = &SI;
409 for (unsigned ScanInsts = 6; BBI != SI.getParent()->begin() && ScanInsts;
412 // Don't count debug info directives, lest they affect codegen,
413 // and we skip pointer-to-pointer bitcasts, which are NOPs.
414 if (isa<DbgInfoIntrinsic>(BBI) ||
415 (isa<BitCastInst>(BBI) && BBI->getType()->isPointerTy())) {
420 if (StoreInst *PrevSI = dyn_cast<StoreInst>(BBI)) {
421 // Prev store isn't volatile, and stores to the same location?
422 if (PrevSI->isSimple() && equivalentAddressValues(PrevSI->getOperand(1),
426 EraseInstFromFunction(*PrevSI);
432 // If this is a load, we have to stop. However, if the loaded value is from
433 // the pointer we're loading and is producing the pointer we're storing,
434 // then *this* store is dead (X = load P; store X -> P).
435 if (LoadInst *LI = dyn_cast<LoadInst>(BBI)) {
436 if (LI == Val && equivalentAddressValues(LI->getOperand(0), Ptr) &&
438 return EraseInstFromFunction(SI);
440 // Otherwise, this is a load from some other location. Stores before it
445 // Don't skip over loads or things that can modify memory.
446 if (BBI->mayWriteToMemory() || BBI->mayReadFromMemory())
450 // store X, null -> turns into 'unreachable' in SimplifyCFG
451 if (isa<ConstantPointerNull>(Ptr) && SI.getPointerAddressSpace() == 0) {
452 if (!isa<UndefValue>(Val)) {
453 SI.setOperand(0, UndefValue::get(Val->getType()));
454 if (Instruction *U = dyn_cast<Instruction>(Val))
455 Worklist.Add(U); // Dropped a use.
457 return 0; // Do not modify these!
460 // store undef, Ptr -> noop
461 if (isa<UndefValue>(Val))
462 return EraseInstFromFunction(SI);
464 // If the pointer destination is a cast, see if we can fold the cast into the
466 if (isa<CastInst>(Ptr))
467 if (Instruction *Res = InstCombineStoreToCast(*this, SI))
469 if (ConstantExpr *CE = dyn_cast<ConstantExpr>(Ptr))
471 if (Instruction *Res = InstCombineStoreToCast(*this, SI))
475 // If this store is the last instruction in the basic block (possibly
476 // excepting debug info instructions), and if the block ends with an
477 // unconditional branch, try to move it to the successor block.
481 } while (isa<DbgInfoIntrinsic>(BBI) ||
482 (isa<BitCastInst>(BBI) && BBI->getType()->isPointerTy()));
483 if (BranchInst *BI = dyn_cast<BranchInst>(BBI))
484 if (BI->isUnconditional())
485 if (SimplifyStoreAtEndOfBlock(SI))
486 return 0; // xform done!
491 /// SimplifyStoreAtEndOfBlock - Turn things like:
492 /// if () { *P = v1; } else { *P = v2 }
493 /// into a phi node with a store in the successor.
495 /// Simplify things like:
496 /// *P = v1; if () { *P = v2; }
497 /// into a phi node with a store in the successor.
499 bool InstCombiner::SimplifyStoreAtEndOfBlock(StoreInst &SI) {
500 BasicBlock *StoreBB = SI.getParent();
502 // Check to see if the successor block has exactly two incoming edges. If
503 // so, see if the other predecessor contains a store to the same location.
504 // if so, insert a PHI node (if needed) and move the stores down.
505 BasicBlock *DestBB = StoreBB->getTerminator()->getSuccessor(0);
507 // Determine whether Dest has exactly two predecessors and, if so, compute
508 // the other predecessor.
509 pred_iterator PI = pred_begin(DestBB);
511 BasicBlock *OtherBB = 0;
516 if (++PI == pred_end(DestBB))
525 if (++PI != pred_end(DestBB))
528 // Bail out if all the relevant blocks aren't distinct (this can happen,
529 // for example, if SI is in an infinite loop)
530 if (StoreBB == DestBB || OtherBB == DestBB)
533 // Verify that the other block ends in a branch and is not otherwise empty.
534 BasicBlock::iterator BBI = OtherBB->getTerminator();
535 BranchInst *OtherBr = dyn_cast<BranchInst>(BBI);
536 if (!OtherBr || BBI == OtherBB->begin())
539 // If the other block ends in an unconditional branch, check for the 'if then
540 // else' case. there is an instruction before the branch.
541 StoreInst *OtherStore = 0;
542 if (OtherBr->isUnconditional()) {
544 // Skip over debugging info.
545 while (isa<DbgInfoIntrinsic>(BBI) ||
546 (isa<BitCastInst>(BBI) && BBI->getType()->isPointerTy())) {
547 if (BBI==OtherBB->begin())
551 // If this isn't a store, isn't a store to the same location, or is not the
552 // right kind of store, bail out.
553 OtherStore = dyn_cast<StoreInst>(BBI);
554 if (!OtherStore || OtherStore->getOperand(1) != SI.getOperand(1) ||
555 !SI.isSameOperationAs(OtherStore))
558 // Otherwise, the other block ended with a conditional branch. If one of the
559 // destinations is StoreBB, then we have the if/then case.
560 if (OtherBr->getSuccessor(0) != StoreBB &&
561 OtherBr->getSuccessor(1) != StoreBB)
564 // Okay, we know that OtherBr now goes to Dest and StoreBB, so this is an
565 // if/then triangle. See if there is a store to the same ptr as SI that
568 // Check to see if we find the matching store.
569 if ((OtherStore = dyn_cast<StoreInst>(BBI))) {
570 if (OtherStore->getOperand(1) != SI.getOperand(1) ||
571 !SI.isSameOperationAs(OtherStore))
575 // If we find something that may be using or overwriting the stored
576 // value, or if we run out of instructions, we can't do the xform.
577 if (BBI->mayReadFromMemory() || BBI->mayWriteToMemory() ||
578 BBI == OtherBB->begin())
582 // In order to eliminate the store in OtherBr, we have to
583 // make sure nothing reads or overwrites the stored value in
585 for (BasicBlock::iterator I = StoreBB->begin(); &*I != &SI; ++I) {
586 // FIXME: This should really be AA driven.
587 if (I->mayReadFromMemory() || I->mayWriteToMemory())
592 // Insert a PHI node now if we need it.
593 Value *MergedVal = OtherStore->getOperand(0);
594 if (MergedVal != SI.getOperand(0)) {
595 PHINode *PN = PHINode::Create(MergedVal->getType(), 2, "storemerge");
596 PN->addIncoming(SI.getOperand(0), SI.getParent());
597 PN->addIncoming(OtherStore->getOperand(0), OtherBB);
598 MergedVal = InsertNewInstBefore(PN, DestBB->front());
601 // Advance to a place where it is safe to insert the new store and
603 BBI = DestBB->getFirstNonPHI();
604 StoreInst *NewSI = new StoreInst(MergedVal, SI.getOperand(1),
609 InsertNewInstBefore(NewSI, *BBI);
610 NewSI->setDebugLoc(OtherStore->getDebugLoc());
612 // Nuke the old stores.
613 EraseInstFromFunction(SI);
614 EraseInstFromFunction(*OtherStore);