1 //===- ScalarReplAggregates.cpp - Scalar Replacement of Aggregates --------===//
3 // The LLVM Compiler Infrastructure
5 // This file was developed by the LLVM research group and is distributed under
6 // the University of Illinois Open Source License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This transformation implements the well known scalar replacement of
11 // aggregates transformation. This xform breaks up alloca instructions of
12 // aggregate type (structure or array) into individual alloca instructions for
13 // each member (if possible). Then, if possible, it transforms the individual
14 // alloca instructions into nice clean scalar SSA form.
16 // This combines a simple SRoA algorithm with the Mem2Reg algorithm because
17 // often interact, especially for C++ programs. As such, iterating between
18 // SRoA, then Mem2Reg until we run out of things to promote works well.
20 //===----------------------------------------------------------------------===//
22 #include "llvm/Transforms/Scalar.h"
23 #include "llvm/Constants.h"
24 #include "llvm/DerivedTypes.h"
25 #include "llvm/Function.h"
26 #include "llvm/Pass.h"
27 #include "llvm/Instructions.h"
28 #include "llvm/Analysis/Dominators.h"
29 #include "llvm/Target/TargetData.h"
30 #include "llvm/Transforms/Utils/PromoteMemToReg.h"
31 #include "llvm/Support/Debug.h"
32 #include "llvm/Support/GetElementPtrTypeIterator.h"
33 #include "llvm/Support/MathExtras.h"
34 #include "llvm/Support/Compiler.h"
35 #include "llvm/ADT/Statistic.h"
36 #include "llvm/ADT/StringExtras.h"
40 Statistic NumReplaced("scalarrepl", "Number of allocas broken up");
41 Statistic NumPromoted("scalarrepl", "Number of allocas promoted");
42 Statistic NumConverted("scalarrepl",
43 "Number of aggregates converted to scalar");
45 struct VISIBILITY_HIDDEN SROA : public FunctionPass {
46 bool runOnFunction(Function &F);
48 bool performScalarRepl(Function &F);
49 bool performPromotion(Function &F);
51 // getAnalysisUsage - This pass does not require any passes, but we know it
52 // will not alter the CFG, so say so.
53 virtual void getAnalysisUsage(AnalysisUsage &AU) const {
54 AU.addRequired<DominatorTree>();
55 AU.addRequired<DominanceFrontier>();
56 AU.addRequired<TargetData>();
61 int isSafeElementUse(Value *Ptr);
62 int isSafeUseOfAllocation(Instruction *User);
63 int isSafeAllocaToScalarRepl(AllocationInst *AI);
64 void CanonicalizeAllocaUsers(AllocationInst *AI);
65 AllocaInst *AddNewAlloca(Function &F, const Type *Ty, AllocationInst *Base);
67 const Type *CanConvertToScalar(Value *V, bool &IsNotTrivial);
68 void ConvertToScalar(AllocationInst *AI, const Type *Ty);
69 void ConvertUsesToScalar(Value *Ptr, AllocaInst *NewAI, unsigned Offset);
72 RegisterPass<SROA> X("scalarrepl", "Scalar Replacement of Aggregates");
75 // Public interface to the ScalarReplAggregates pass
76 FunctionPass *llvm::createScalarReplAggregatesPass() { return new SROA(); }
79 bool SROA::runOnFunction(Function &F) {
80 bool Changed = performPromotion(F);
82 bool LocalChange = performScalarRepl(F);
83 if (!LocalChange) break; // No need to repromote if no scalarrepl
85 LocalChange = performPromotion(F);
86 if (!LocalChange) break; // No need to re-scalarrepl if no promotion
93 bool SROA::performPromotion(Function &F) {
94 std::vector<AllocaInst*> Allocas;
95 const TargetData &TD = getAnalysis<TargetData>();
96 DominatorTree &DT = getAnalysis<DominatorTree>();
97 DominanceFrontier &DF = getAnalysis<DominanceFrontier>();
99 BasicBlock &BB = F.getEntryBlock(); // Get the entry node for the function
101 bool Changed = false;
106 // Find allocas that are safe to promote, by looking at all instructions in
108 for (BasicBlock::iterator I = BB.begin(), E = --BB.end(); I != E; ++I)
109 if (AllocaInst *AI = dyn_cast<AllocaInst>(I)) // Is it an alloca?
110 if (isAllocaPromotable(AI, TD))
111 Allocas.push_back(AI);
113 if (Allocas.empty()) break;
115 PromoteMemToReg(Allocas, DT, DF, TD);
116 NumPromoted += Allocas.size();
123 // performScalarRepl - This algorithm is a simple worklist driven algorithm,
124 // which runs on all of the malloc/alloca instructions in the function, removing
125 // them if they are only used by getelementptr instructions.
127 bool SROA::performScalarRepl(Function &F) {
128 std::vector<AllocationInst*> WorkList;
130 // Scan the entry basic block, adding any alloca's and mallocs to the worklist
131 BasicBlock &BB = F.getEntryBlock();
132 for (BasicBlock::iterator I = BB.begin(), E = BB.end(); I != E; ++I)
133 if (AllocationInst *A = dyn_cast<AllocationInst>(I))
134 WorkList.push_back(A);
136 // Process the worklist
137 bool Changed = false;
138 while (!WorkList.empty()) {
139 AllocationInst *AI = WorkList.back();
142 // If we can turn this aggregate value (potentially with casts) into a
143 // simple scalar value that can be mem2reg'd into a register value.
144 bool IsNotTrivial = false;
145 if (const Type *ActualType = CanConvertToScalar(AI, IsNotTrivial))
146 if (IsNotTrivial && ActualType != Type::VoidTy) {
147 ConvertToScalar(AI, ActualType);
152 // We cannot transform the allocation instruction if it is an array
153 // allocation (allocations OF arrays are ok though), and an allocation of a
154 // scalar value cannot be decomposed at all.
156 if (AI->isArrayAllocation() ||
157 (!isa<StructType>(AI->getAllocatedType()) &&
158 !isa<ArrayType>(AI->getAllocatedType()))) continue;
160 // Check that all of the users of the allocation are capable of being
162 switch (isSafeAllocaToScalarRepl(AI)) {
163 default: assert(0 && "Unexpected value!");
164 case 0: // Not safe to scalar replace.
166 case 1: // Safe, but requires cleanup/canonicalizations first
167 CanonicalizeAllocaUsers(AI);
168 case 3: // Safe to scalar replace.
172 DOUT << "Found inst to xform: " << *AI;
175 std::vector<AllocaInst*> ElementAllocas;
176 if (const StructType *ST = dyn_cast<StructType>(AI->getAllocatedType())) {
177 ElementAllocas.reserve(ST->getNumContainedTypes());
178 for (unsigned i = 0, e = ST->getNumContainedTypes(); i != e; ++i) {
179 AllocaInst *NA = new AllocaInst(ST->getContainedType(i), 0,
181 AI->getName() + "." + utostr(i), AI);
182 ElementAllocas.push_back(NA);
183 WorkList.push_back(NA); // Add to worklist for recursive processing
186 const ArrayType *AT = cast<ArrayType>(AI->getAllocatedType());
187 ElementAllocas.reserve(AT->getNumElements());
188 const Type *ElTy = AT->getElementType();
189 for (unsigned i = 0, e = AT->getNumElements(); i != e; ++i) {
190 AllocaInst *NA = new AllocaInst(ElTy, 0, AI->getAlignment(),
191 AI->getName() + "." + utostr(i), AI);
192 ElementAllocas.push_back(NA);
193 WorkList.push_back(NA); // Add to worklist for recursive processing
197 // Now that we have created the alloca instructions that we want to use,
198 // expand the getelementptr instructions to use them.
200 while (!AI->use_empty()) {
201 Instruction *User = cast<Instruction>(AI->use_back());
202 GetElementPtrInst *GEPI = cast<GetElementPtrInst>(User);
203 // We now know that the GEP is of the form: GEP <ptr>, 0, <cst>
205 (unsigned)cast<ConstantInt>(GEPI->getOperand(2))->getZExtValue();
207 assert(Idx < ElementAllocas.size() && "Index out of range?");
208 AllocaInst *AllocaToUse = ElementAllocas[Idx];
211 if (GEPI->getNumOperands() == 3) {
212 // Do not insert a new getelementptr instruction with zero indices, only
213 // to have it optimized out later.
214 RepValue = AllocaToUse;
216 // We are indexing deeply into the structure, so we still need a
217 // getelement ptr instruction to finish the indexing. This may be
218 // expanded itself once the worklist is rerun.
220 std::string OldName = GEPI->getName(); // Steal the old name.
221 std::vector<Value*> NewArgs;
222 NewArgs.push_back(Constant::getNullValue(Type::IntTy));
223 NewArgs.insert(NewArgs.end(), GEPI->op_begin()+3, GEPI->op_end());
225 RepValue = new GetElementPtrInst(AllocaToUse, NewArgs, OldName, GEPI);
228 // Move all of the users over to the new GEP.
229 GEPI->replaceAllUsesWith(RepValue);
230 // Delete the old GEP
231 GEPI->eraseFromParent();
234 // Finally, delete the Alloca instruction
235 AI->getParent()->getInstList().erase(AI);
243 /// isSafeElementUse - Check to see if this use is an allowed use for a
244 /// getelementptr instruction of an array aggregate allocation.
246 int SROA::isSafeElementUse(Value *Ptr) {
247 for (Value::use_iterator I = Ptr->use_begin(), E = Ptr->use_end();
249 Instruction *User = cast<Instruction>(*I);
250 switch (User->getOpcode()) {
251 case Instruction::Load: break;
252 case Instruction::Store:
253 // Store is ok if storing INTO the pointer, not storing the pointer
254 if (User->getOperand(0) == Ptr) return 0;
256 case Instruction::GetElementPtr: {
257 GetElementPtrInst *GEP = cast<GetElementPtrInst>(User);
258 if (GEP->getNumOperands() > 1) {
259 if (!isa<Constant>(GEP->getOperand(1)) ||
260 !cast<Constant>(GEP->getOperand(1))->isNullValue())
261 return 0; // Using pointer arithmetic to navigate the array...
263 if (!isSafeElementUse(GEP)) return 0;
267 DOUT << " Transformation preventing inst: " << *User;
271 return 3; // All users look ok :)
274 /// AllUsersAreLoads - Return true if all users of this value are loads.
275 static bool AllUsersAreLoads(Value *Ptr) {
276 for (Value::use_iterator I = Ptr->use_begin(), E = Ptr->use_end();
278 if (cast<Instruction>(*I)->getOpcode() != Instruction::Load)
283 /// isSafeUseOfAllocation - Check to see if this user is an allowed use for an
284 /// aggregate allocation.
286 int SROA::isSafeUseOfAllocation(Instruction *User) {
287 if (!isa<GetElementPtrInst>(User)) return 0;
289 GetElementPtrInst *GEPI = cast<GetElementPtrInst>(User);
290 gep_type_iterator I = gep_type_begin(GEPI), E = gep_type_end(GEPI);
292 // The GEP is not safe to transform if not of the form "GEP <ptr>, 0, <cst>".
294 I.getOperand() != Constant::getNullValue(I.getOperand()->getType()))
298 if (I == E) return 0; // ran out of GEP indices??
300 // If this is a use of an array allocation, do a bit more checking for sanity.
301 if (const ArrayType *AT = dyn_cast<ArrayType>(*I)) {
302 uint64_t NumElements = AT->getNumElements();
304 if (isa<ConstantInt>(I.getOperand())) {
305 // Check to make sure that index falls within the array. If not,
306 // something funny is going on, so we won't do the optimization.
308 if (cast<ConstantInt>(GEPI->getOperand(2))->getZExtValue() >= NumElements)
311 // We cannot scalar repl this level of the array unless any array
312 // sub-indices are in-range constants. In particular, consider:
313 // A[0][i]. We cannot know that the user isn't doing invalid things like
314 // allowing i to index an out-of-range subscript that accesses A[1].
316 // Scalar replacing *just* the outer index of the array is probably not
317 // going to be a win anyway, so just give up.
318 for (++I; I != E && (isa<ArrayType>(*I) || isa<PackedType>(*I)); ++I) {
319 uint64_t NumElements;
320 if (const ArrayType *SubArrayTy = dyn_cast<ArrayType>(*I))
321 NumElements = SubArrayTy->getNumElements();
323 NumElements = cast<PackedType>(*I)->getNumElements();
325 if (!isa<ConstantInt>(I.getOperand())) return 0;
326 if (cast<ConstantInt>(I.getOperand())->getZExtValue() >= NumElements)
331 // If this is an array index and the index is not constant, we cannot
332 // promote... that is unless the array has exactly one or two elements in
333 // it, in which case we CAN promote it, but we have to canonicalize this
334 // out if this is the only problem.
335 if ((NumElements == 1 || NumElements == 2) &&
336 AllUsersAreLoads(GEPI))
337 return 1; // Canonicalization required!
342 // If there are any non-simple uses of this getelementptr, make sure to reject
344 return isSafeElementUse(GEPI);
347 /// isSafeStructAllocaToScalarRepl - Check to see if the specified allocation of
348 /// an aggregate can be broken down into elements. Return 0 if not, 3 if safe,
349 /// or 1 if safe after canonicalization has been performed.
351 int SROA::isSafeAllocaToScalarRepl(AllocationInst *AI) {
352 // Loop over the use list of the alloca. We can only transform it if all of
353 // the users are safe to transform.
356 for (Value::use_iterator I = AI->use_begin(), E = AI->use_end();
358 isSafe &= isSafeUseOfAllocation(cast<Instruction>(*I));
360 DOUT << "Cannot transform: " << *AI << " due to user: " << **I;
364 // If we require cleanup, isSafe is now 1, otherwise it is 3.
368 /// CanonicalizeAllocaUsers - If SROA reported that it can promote the specified
369 /// allocation, but only if cleaned up, perform the cleanups required.
370 void SROA::CanonicalizeAllocaUsers(AllocationInst *AI) {
371 // At this point, we know that the end result will be SROA'd and promoted, so
372 // we can insert ugly code if required so long as sroa+mem2reg will clean it
374 for (Value::use_iterator UI = AI->use_begin(), E = AI->use_end();
376 GetElementPtrInst *GEPI = cast<GetElementPtrInst>(*UI++);
377 gep_type_iterator I = gep_type_begin(GEPI);
380 if (const ArrayType *AT = dyn_cast<ArrayType>(*I)) {
381 uint64_t NumElements = AT->getNumElements();
383 if (!isa<ConstantInt>(I.getOperand())) {
384 if (NumElements == 1) {
385 GEPI->setOperand(2, Constant::getNullValue(Type::IntTy));
387 assert(NumElements == 2 && "Unhandled case!");
388 // All users of the GEP must be loads. At each use of the GEP, insert
389 // two loads of the appropriate indexed GEP and select between them.
390 Value *IsOne = BinaryOperator::createSetNE(I.getOperand(),
391 Constant::getNullValue(I.getOperand()->getType()),
393 // Insert the new GEP instructions, which are properly indexed.
394 std::vector<Value*> Indices(GEPI->op_begin()+1, GEPI->op_end());
395 Indices[1] = Constant::getNullValue(Type::IntTy);
396 Value *ZeroIdx = new GetElementPtrInst(GEPI->getOperand(0), Indices,
397 GEPI->getName()+".0", GEPI);
398 Indices[1] = ConstantInt::get(Type::IntTy, 1);
399 Value *OneIdx = new GetElementPtrInst(GEPI->getOperand(0), Indices,
400 GEPI->getName()+".1", GEPI);
401 // Replace all loads of the variable index GEP with loads from both
402 // indexes and a select.
403 while (!GEPI->use_empty()) {
404 LoadInst *LI = cast<LoadInst>(GEPI->use_back());
405 Value *Zero = new LoadInst(ZeroIdx, LI->getName()+".0", LI);
406 Value *One = new LoadInst(OneIdx , LI->getName()+".1", LI);
407 Value *R = new SelectInst(IsOne, One, Zero, LI->getName(), LI);
408 LI->replaceAllUsesWith(R);
409 LI->eraseFromParent();
411 GEPI->eraseFromParent();
418 /// MergeInType - Add the 'In' type to the accumulated type so far. If the
419 /// types are incompatible, return true, otherwise update Accum and return
422 /// There are two cases we handle here:
423 /// 1) An effectively integer union, where the pieces are stored into as
424 /// smaller integers (common with byte swap and other idioms).
425 /// 2) A union of a vector and its elements. Here we turn element accesses
426 /// into insert/extract element operations.
427 static bool MergeInType(const Type *In, const Type *&Accum,
428 const TargetData &TD) {
429 // If this is our first type, just use it.
430 const PackedType *PTy;
431 if (Accum == Type::VoidTy || In == Accum) {
433 } else if (In->isIntegral() && Accum->isIntegral()) { // integer union.
434 // Otherwise pick whichever type is larger.
435 if (In->getTypeID() > Accum->getTypeID())
437 } else if (isa<PointerType>(In) && isa<PointerType>(Accum)) {
438 // Pointer unions just stay as one of the pointers.
439 } else if ((PTy = dyn_cast<PackedType>(Accum)) &&
440 PTy->getElementType() == In) {
441 // Accum is a vector, and we are accessing an element: ok.
442 } else if ((PTy = dyn_cast<PackedType>(In)) &&
443 PTy->getElementType() == Accum) {
444 // In is a vector, and accum is an element: ok, remember In.
446 } else if (isa<PointerType>(In) && Accum->isIntegral()) {
447 // Pointer/Integer unions merge together as integers.
448 return MergeInType(TD.getIntPtrType(), Accum, TD);
449 } else if (isa<PointerType>(Accum) && In->isIntegral()) {
450 // Pointer/Integer unions merge together as integers.
451 Accum = TD.getIntPtrType();
452 return MergeInType(In, Accum, TD);
459 /// getUIntAtLeastAsBitAs - Return an unsigned integer type that is at least
460 /// as big as the specified type. If there is no suitable type, this returns
462 const Type *getUIntAtLeastAsBitAs(unsigned NumBits) {
463 if (NumBits > 64) return 0;
464 if (NumBits > 32) return Type::ULongTy;
465 if (NumBits > 16) return Type::UIntTy;
466 if (NumBits > 8) return Type::UShortTy;
467 return Type::UByteTy;
470 /// CanConvertToScalar - V is a pointer. If we can convert the pointee to a
471 /// single scalar integer type, return that type. Further, if the use is not
472 /// a completely trivial use that mem2reg could promote, set IsNotTrivial. If
473 /// there are no uses of this pointer, return Type::VoidTy to differentiate from
476 const Type *SROA::CanConvertToScalar(Value *V, bool &IsNotTrivial) {
477 const Type *UsedType = Type::VoidTy; // No uses, no forced type.
478 const TargetData &TD = getAnalysis<TargetData>();
479 const PointerType *PTy = cast<PointerType>(V->getType());
481 for (Value::use_iterator UI = V->use_begin(), E = V->use_end(); UI!=E; ++UI) {
482 Instruction *User = cast<Instruction>(*UI);
484 if (LoadInst *LI = dyn_cast<LoadInst>(User)) {
485 if (MergeInType(LI->getType(), UsedType, TD))
488 } else if (StoreInst *SI = dyn_cast<StoreInst>(User)) {
489 // Storing the pointer, not the into the value?
490 if (SI->getOperand(0) == V) return 0;
492 // NOTE: We could handle storing of FP imms into integers here!
494 if (MergeInType(SI->getOperand(0)->getType(), UsedType, TD))
496 } else if (CastInst *CI = dyn_cast<CastInst>(User)) {
497 if (!isa<PointerType>(CI->getType())) return 0;
499 const Type *SubTy = CanConvertToScalar(CI, IsNotTrivial);
500 if (!SubTy || MergeInType(SubTy, UsedType, TD)) return 0;
501 } else if (GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(User)) {
502 // Check to see if this is stepping over an element: GEP Ptr, int C
503 if (GEP->getNumOperands() == 2 && isa<ConstantInt>(GEP->getOperand(1))) {
504 unsigned Idx = cast<ConstantInt>(GEP->getOperand(1))->getZExtValue();
505 unsigned ElSize = TD.getTypeSize(PTy->getElementType());
506 unsigned BitOffset = Idx*ElSize*8;
507 if (BitOffset > 64 || !isPowerOf2_32(ElSize)) return 0;
510 const Type *SubElt = CanConvertToScalar(GEP, IsNotTrivial);
511 if (SubElt == 0) return 0;
512 if (SubElt != Type::VoidTy && SubElt->isInteger()) {
514 getUIntAtLeastAsBitAs(TD.getTypeSize(SubElt)*8+BitOffset);
515 if (NewTy == 0 || MergeInType(NewTy, UsedType, TD)) return 0;
518 } else if (GEP->getNumOperands() == 3 &&
519 isa<ConstantInt>(GEP->getOperand(1)) &&
520 isa<ConstantInt>(GEP->getOperand(2)) &&
521 cast<Constant>(GEP->getOperand(1))->isNullValue()) {
522 // We are stepping into an element, e.g. a structure or an array:
523 // GEP Ptr, int 0, uint C
524 const Type *AggTy = PTy->getElementType();
525 unsigned Idx = cast<ConstantInt>(GEP->getOperand(2))->getZExtValue();
527 if (const ArrayType *ATy = dyn_cast<ArrayType>(AggTy)) {
528 if (Idx >= ATy->getNumElements()) return 0; // Out of range.
529 } else if (const PackedType *PackedTy = dyn_cast<PackedType>(AggTy)) {
530 // Getting an element of the packed vector.
531 if (Idx >= PackedTy->getNumElements()) return 0; // Out of range.
533 // Merge in the packed type.
534 if (MergeInType(PackedTy, UsedType, TD)) return 0;
536 const Type *SubTy = CanConvertToScalar(GEP, IsNotTrivial);
537 if (SubTy == 0) return 0;
539 if (SubTy != Type::VoidTy && MergeInType(SubTy, UsedType, TD))
542 // We'll need to change this to an insert/extract element operation.
544 continue; // Everything looks ok
546 } else if (isa<StructType>(AggTy)) {
547 // Structs are always ok.
551 const Type *NTy = getUIntAtLeastAsBitAs(TD.getTypeSize(AggTy)*8);
552 if (NTy == 0 || MergeInType(NTy, UsedType, TD)) return 0;
553 const Type *SubTy = CanConvertToScalar(GEP, IsNotTrivial);
554 if (SubTy == 0) return 0;
555 if (SubTy != Type::VoidTy && MergeInType(SubTy, UsedType, TD))
557 continue; // Everything looks ok
561 // Cannot handle this!
569 /// ConvertToScalar - The specified alloca passes the CanConvertToScalar
570 /// predicate and is non-trivial. Convert it to something that can be trivially
571 /// promoted into a register by mem2reg.
572 void SROA::ConvertToScalar(AllocationInst *AI, const Type *ActualTy) {
573 DOUT << "CONVERT TO SCALAR: " << *AI << " TYPE = "
574 << *ActualTy << "\n";
577 BasicBlock *EntryBlock = AI->getParent();
578 assert(EntryBlock == &EntryBlock->getParent()->front() &&
579 "Not in the entry block!");
580 EntryBlock->getInstList().remove(AI); // Take the alloca out of the program.
582 if (ActualTy->isInteger())
583 ActualTy = ActualTy->getUnsignedVersion();
585 // Create and insert the alloca.
586 AllocaInst *NewAI = new AllocaInst(ActualTy, 0, AI->getName(),
587 EntryBlock->begin());
588 ConvertUsesToScalar(AI, NewAI, 0);
593 /// ConvertUsesToScalar - Convert all of the users of Ptr to use the new alloca
594 /// directly. This happens when we are converting an "integer union" to a
595 /// single integer scalar, or when we are converting a "vector union" to a
596 /// vector with insert/extractelement instructions.
598 /// Offset is an offset from the original alloca, in bits that need to be
599 /// shifted to the right. By the end of this, there should be no uses of Ptr.
600 void SROA::ConvertUsesToScalar(Value *Ptr, AllocaInst *NewAI, unsigned Offset) {
601 bool isVectorInsert = isa<PackedType>(NewAI->getType()->getElementType());
602 const TargetData &TD = getAnalysis<TargetData>();
603 while (!Ptr->use_empty()) {
604 Instruction *User = cast<Instruction>(Ptr->use_back());
606 if (LoadInst *LI = dyn_cast<LoadInst>(User)) {
607 // The load is a bit extract from NewAI shifted right by Offset bits.
608 Value *NV = new LoadInst(NewAI, LI->getName(), LI);
609 if (NV->getType() != LI->getType()) {
610 if (const PackedType *PTy = dyn_cast<PackedType>(NV->getType())) {
611 // Must be an element access.
612 unsigned Elt = Offset/(TD.getTypeSize(PTy->getElementType())*8);
613 NV = new ExtractElementInst(NV, ConstantInt::get(Type::UIntTy, Elt),
617 assert(NV->getType()->isInteger() && "Unknown promotion!");
618 if (Offset < TD.getTypeSize(NV->getType())*8) {
619 NV = new ShiftInst(Instruction::LShr, NV,
620 ConstantInt::get(Type::UByteTy, Offset),
624 assert((NV->getType()->isInteger() ||
625 isa<PointerType>(NV->getType())) && "Unknown promotion!");
627 NV = CastInst::createInferredCast(NV, LI->getType(), LI->getName(),
631 LI->replaceAllUsesWith(NV);
632 LI->eraseFromParent();
633 } else if (StoreInst *SI = dyn_cast<StoreInst>(User)) {
634 assert(SI->getOperand(0) != Ptr && "Consistency error!");
636 // Convert the stored type to the actual type, shift it left to insert
637 // then 'or' into place.
638 Value *SV = SI->getOperand(0);
639 const Type *AllocaType = NewAI->getType()->getElementType();
640 if (SV->getType() != AllocaType) {
641 Value *Old = new LoadInst(NewAI, NewAI->getName()+".in", SI);
643 if (const PackedType *PTy = dyn_cast<PackedType>(AllocaType)) {
644 // Must be an element insertion.
645 unsigned Elt = Offset/(TD.getTypeSize(PTy->getElementType())*8);
646 SV = new InsertElementInst(Old, SV,
647 ConstantInt::get(Type::UIntTy, Elt),
650 // Always zero extend the value.
651 if (SV->getType()->isSigned())
652 SV = CastInst::createInferredCast(SV,
653 SV->getType()->getUnsignedVersion(), SV->getName(), SI);
654 SV = CastInst::createInferredCast(SV, Old->getType(), SV->getName(),
656 if (Offset && Offset < TD.getTypeSize(SV->getType())*8)
657 SV = new ShiftInst(Instruction::Shl, SV,
658 ConstantInt::get(Type::UByteTy, Offset),
659 SV->getName()+".adj", SI);
660 // Mask out the bits we are about to insert from the old value.
661 unsigned TotalBits = TD.getTypeSize(SV->getType())*8;
662 unsigned InsertBits = TD.getTypeSize(SI->getOperand(0)->getType())*8;
663 if (TotalBits != InsertBits) {
664 assert(TotalBits > InsertBits);
665 uint64_t Mask = ~(((1ULL << InsertBits)-1) << Offset);
667 Mask = Mask & ((1ULL << TotalBits)-1);
668 Old = BinaryOperator::createAnd(Old,
669 ConstantInt::get(Old->getType(), Mask),
670 Old->getName()+".mask", SI);
671 SV = BinaryOperator::createOr(Old, SV, SV->getName()+".ins", SI);
675 new StoreInst(SV, NewAI, SI);
676 SI->eraseFromParent();
678 } else if (CastInst *CI = dyn_cast<CastInst>(User)) {
679 unsigned NewOff = Offset;
680 const TargetData &TD = getAnalysis<TargetData>();
681 if (TD.isBigEndian() && !isVectorInsert) {
682 // Adjust the pointer. For example, storing 16-bits into a 32-bit
683 // alloca with just a cast makes it modify the top 16-bits.
684 const Type *SrcTy = cast<PointerType>(Ptr->getType())->getElementType();
685 const Type *DstTy = cast<PointerType>(CI->getType())->getElementType();
686 int PtrDiffBits = TD.getTypeSize(SrcTy)*8-TD.getTypeSize(DstTy)*8;
687 NewOff += PtrDiffBits;
689 ConvertUsesToScalar(CI, NewAI, NewOff);
690 CI->eraseFromParent();
691 } else if (GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(User)) {
692 const PointerType *AggPtrTy =
693 cast<PointerType>(GEP->getOperand(0)->getType());
694 const TargetData &TD = getAnalysis<TargetData>();
695 unsigned AggSizeInBits = TD.getTypeSize(AggPtrTy->getElementType())*8;
697 // Check to see if this is stepping over an element: GEP Ptr, int C
698 unsigned NewOffset = Offset;
699 if (GEP->getNumOperands() == 2) {
700 unsigned Idx = cast<ConstantInt>(GEP->getOperand(1))->getZExtValue();
701 unsigned BitOffset = Idx*AggSizeInBits;
703 if (TD.isLittleEndian() || isVectorInsert)
704 NewOffset += BitOffset;
706 NewOffset -= BitOffset;
708 } else if (GEP->getNumOperands() == 3) {
709 // We know that operand #2 is zero.
710 unsigned Idx = cast<ConstantInt>(GEP->getOperand(2))->getZExtValue();
711 const Type *AggTy = AggPtrTy->getElementType();
712 if (const SequentialType *SeqTy = dyn_cast<SequentialType>(AggTy)) {
713 unsigned ElSizeBits = TD.getTypeSize(SeqTy->getElementType())*8;
715 if (TD.isLittleEndian() || isVectorInsert)
716 NewOffset += ElSizeBits*Idx;
718 NewOffset += AggSizeInBits-ElSizeBits*(Idx+1);
719 } else if (const StructType *STy = dyn_cast<StructType>(AggTy)) {
720 unsigned EltBitOffset = TD.getStructLayout(STy)->MemberOffsets[Idx]*8;
722 if (TD.isLittleEndian() || isVectorInsert)
723 NewOffset += EltBitOffset;
725 const PointerType *ElPtrTy = cast<PointerType>(GEP->getType());
726 unsigned ElSizeBits = TD.getTypeSize(ElPtrTy->getElementType())*8;
727 NewOffset += AggSizeInBits-(EltBitOffset+ElSizeBits);
731 assert(0 && "Unsupported operation!");
735 assert(0 && "Unsupported operation!");
738 ConvertUsesToScalar(GEP, NewAI, NewOffset);
739 GEP->eraseFromParent();
741 assert(0 && "Unsupported operation!");