1 //===-- LICM.cpp - Loop Invariant Code Motion Pass ------------------------===//
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 pass performs loop invariant code motion, attempting to remove as much
11 // code from the body of a loop as possible. It does this by either hoisting
12 // code into the preheader block, or by sinking code to the exit blocks if it is
13 // safe. This pass also promotes must-aliased memory locations in the loop to
14 // live in registers, thus hoisting and sinking "invariant" loads and stores.
16 // This pass uses alias analysis for two purposes:
18 // 1. Moving loop invariant loads and calls out of loops. If we can determine
19 // that a load or call inside of a loop never aliases anything stored to,
20 // we can hoist it or sink it like any other instruction.
21 // 2. Scalar Promotion of Memory - If there is a store instruction inside of
22 // the loop, we try to move the store to happen AFTER the loop instead of
23 // inside of the loop. This can only happen if a few conditions are true:
24 // A. The pointer stored through is loop invariant
25 // B. There are no stores or loads in the loop which _may_ alias the
26 // pointer. There are no calls in the loop which mod/ref the pointer.
27 // If these conditions are true, we can promote the loads and stores in the
28 // loop of the pointer to use a temporary alloca'd variable. We then use
29 // the mem2reg functionality to construct the appropriate SSA form for the
32 //===----------------------------------------------------------------------===//
34 #define DEBUG_TYPE "licm"
35 #include "llvm/Transforms/Scalar.h"
36 #include "llvm/Constants.h"
37 #include "llvm/DerivedTypes.h"
38 #include "llvm/Instructions.h"
39 #include "llvm/Target/TargetData.h"
40 #include "llvm/Analysis/LoopInfo.h"
41 #include "llvm/Analysis/LoopPass.h"
42 #include "llvm/Analysis/AliasAnalysis.h"
43 #include "llvm/Analysis/AliasSetTracker.h"
44 #include "llvm/Analysis/Dominators.h"
45 #include "llvm/Analysis/ScalarEvolution.h"
46 #include "llvm/Transforms/Utils/PromoteMemToReg.h"
47 #include "llvm/Support/CFG.h"
48 #include "llvm/Support/Compiler.h"
49 #include "llvm/Support/CommandLine.h"
50 #include "llvm/Support/Debug.h"
51 #include "llvm/ADT/Statistic.h"
55 STATISTIC(NumSunk , "Number of instructions sunk out of loop");
56 STATISTIC(NumHoisted , "Number of instructions hoisted out of loop");
57 STATISTIC(NumMovedLoads, "Number of load insts hoisted or sunk");
58 STATISTIC(NumMovedCalls, "Number of call insts hoisted or sunk");
59 STATISTIC(NumPromoted , "Number of memory locations promoted to registers");
63 DisablePromotion("disable-licm-promotion", cl::Hidden,
64 cl::desc("Disable memory promotion in LICM pass"));
66 struct VISIBILITY_HIDDEN LICM : public LoopPass {
67 static char ID; // Pass identification, replacement for typeid
68 LICM() : LoopPass((intptr_t)&ID) {}
70 virtual bool runOnLoop(Loop *L, LPPassManager &LPM);
72 /// This transformation requires natural loop information & requires that
73 /// loop preheaders be inserted into the CFG...
75 virtual void getAnalysisUsage(AnalysisUsage &AU) const {
77 AU.addRequiredID(LoopSimplifyID);
78 AU.addRequired<LoopInfo>();
79 AU.addRequired<DominatorTree>();
80 AU.addRequired<DominanceFrontier>(); // For scalar promotion (mem2reg)
81 AU.addRequired<AliasAnalysis>();
82 AU.addPreserved<ScalarEvolution>();
83 AU.addPreserved<DominanceFrontier>();
86 bool doFinalization() {
87 LoopToAliasMap.clear();
92 // Various analyses that we use...
93 AliasAnalysis *AA; // Current AliasAnalysis information
94 LoopInfo *LI; // Current LoopInfo
95 DominatorTree *DT; // Dominator Tree for the current Loop...
96 DominanceFrontier *DF; // Current Dominance Frontier
98 // State that is updated as we process loops
99 bool Changed; // Set to true when we change anything.
100 BasicBlock *Preheader; // The preheader block of the current loop...
101 Loop *CurLoop; // The current loop we are working on...
102 AliasSetTracker *CurAST; // AliasSet information for the current loop...
103 std::map<Loop *, AliasSetTracker *> LoopToAliasMap;
105 /// SinkRegion - Walk the specified region of the CFG (defined by all blocks
106 /// dominated by the specified block, and that are in the current loop) in
107 /// reverse depth first order w.r.t the DominatorTree. This allows us to
108 /// visit uses before definitions, allowing us to sink a loop body in one
109 /// pass without iteration.
111 void SinkRegion(DomTreeNode *N);
113 /// HoistRegion - Walk the specified region of the CFG (defined by all
114 /// blocks dominated by the specified block, and that are in the current
115 /// loop) in depth first order w.r.t the DominatorTree. This allows us to
116 /// visit definitions before uses, allowing us to hoist a loop body in one
117 /// pass without iteration.
119 void HoistRegion(DomTreeNode *N);
121 /// inSubLoop - Little predicate that returns true if the specified basic
122 /// block is in a subloop of the current one, not the current one itself.
124 bool inSubLoop(BasicBlock *BB) {
125 assert(CurLoop->contains(BB) && "Only valid if BB is IN the loop");
126 for (Loop::iterator I = CurLoop->begin(), E = CurLoop->end(); I != E; ++I)
127 if ((*I)->contains(BB))
128 return true; // A subloop actually contains this block!
132 /// isExitBlockDominatedByBlockInLoop - This method checks to see if the
133 /// specified exit block of the loop is dominated by the specified block
134 /// that is in the body of the loop. We use these constraints to
135 /// dramatically limit the amount of the dominator tree that needs to be
137 bool isExitBlockDominatedByBlockInLoop(BasicBlock *ExitBlock,
138 BasicBlock *BlockInLoop) const {
139 // If the block in the loop is the loop header, it must be dominated!
140 BasicBlock *LoopHeader = CurLoop->getHeader();
141 if (BlockInLoop == LoopHeader)
144 DomTreeNode *BlockInLoopNode = DT->getNode(BlockInLoop);
145 DomTreeNode *IDom = DT->getNode(ExitBlock);
147 // Because the exit block is not in the loop, we know we have to get _at
148 // least_ its immediate dominator.
150 // Get next Immediate Dominator.
151 IDom = IDom->getIDom();
153 // If we have got to the header of the loop, then the instructions block
154 // did not dominate the exit node, so we can't hoist it.
155 if (IDom->getBlock() == LoopHeader)
158 } while (IDom != BlockInLoopNode);
163 /// sink - When an instruction is found to only be used outside of the loop,
164 /// this function moves it to the exit blocks and patches up SSA form as
167 void sink(Instruction &I);
169 /// hoist - When an instruction is found to only use loop invariant operands
170 /// that is safe to hoist, this instruction is called to do the dirty work.
172 void hoist(Instruction &I);
174 /// isSafeToExecuteUnconditionally - Only sink or hoist an instruction if it
175 /// is not a trapping instruction or if it is a trapping instruction and is
176 /// guaranteed to execute.
178 bool isSafeToExecuteUnconditionally(Instruction &I);
180 /// pointerInvalidatedByLoop - Return true if the body of this loop may
181 /// store into the memory location pointed to by V.
183 bool pointerInvalidatedByLoop(Value *V, unsigned Size) {
184 // Check to see if any of the basic blocks in CurLoop invalidate *V.
185 return CurAST->getAliasSetForPointer(V, Size).isMod();
188 bool canSinkOrHoistInst(Instruction &I);
189 bool isLoopInvariantInst(Instruction &I);
190 bool isNotUsedInLoop(Instruction &I);
192 /// PromoteValuesInLoop - Look at the stores in the loop and promote as many
193 /// to scalars as we can.
195 void PromoteValuesInLoop();
197 /// FindPromotableValuesInLoop - Check the current loop for stores to
198 /// definite pointers, which are not loaded and stored through may aliases.
199 /// If these are found, create an alloca for the value, add it to the
200 /// PromotedValues list, and keep track of the mapping from value to
203 void FindPromotableValuesInLoop(
204 std::vector<std::pair<AllocaInst*, Value*> > &PromotedValues,
205 std::map<Value*, AllocaInst*> &Val2AlMap);
209 RegisterPass<LICM> X("licm", "Loop Invariant Code Motion");
212 LoopPass *llvm::createLICMPass() { return new LICM(); }
214 /// Hoist expressions out of the specified loop...
216 bool LICM::runOnLoop(Loop *L, LPPassManager &LPM) {
219 // Get our Loop and Alias Analysis information...
220 LI = &getAnalysis<LoopInfo>();
221 AA = &getAnalysis<AliasAnalysis>();
222 DF = &getAnalysis<DominanceFrontier>();
223 DT = &getAnalysis<DominatorTree>();
225 CurAST = new AliasSetTracker(*AA);
226 // Collect Alias info from subloops
227 for (Loop::iterator LoopItr = L->begin(), LoopItrE = L->end();
228 LoopItr != LoopItrE; ++LoopItr) {
229 Loop *InnerL = *LoopItr;
230 AliasSetTracker *InnerAST = LoopToAliasMap[InnerL];
231 assert (InnerAST && "Where is my AST?");
233 // What if InnerLoop was modified by other passes ?
234 CurAST->add(*InnerAST);
239 // Get the preheader block to move instructions into...
240 Preheader = L->getLoopPreheader();
241 assert(Preheader&&"Preheader insertion pass guarantees we have a preheader!");
243 // Loop over the body of this loop, looking for calls, invokes, and stores.
244 // Because subloops have already been incorporated into AST, we skip blocks in
247 for (std::vector<BasicBlock*>::const_iterator I = L->getBlocks().begin(),
248 E = L->getBlocks().end(); I != E; ++I)
249 if (LI->getLoopFor(*I) == L) // Ignore blocks in subloops...
250 CurAST->add(**I); // Incorporate the specified basic block
252 // We want to visit all of the instructions in this loop... that are not parts
253 // of our subloops (they have already had their invariants hoisted out of
254 // their loop, into this loop, so there is no need to process the BODIES of
257 // Traverse the body of the loop in depth first order on the dominator tree so
258 // that we are guaranteed to see definitions before we see uses. This allows
259 // us to sink instructions in one pass, without iteration. AFter sinking
260 // instructions, we perform another pass to hoist them out of the loop.
262 SinkRegion(DT->getNode(L->getHeader()));
263 HoistRegion(DT->getNode(L->getHeader()));
265 // Now that all loop invariants have been removed from the loop, promote any
266 // memory references to scalars that we can...
267 if (!DisablePromotion)
268 PromoteValuesInLoop();
270 // Clear out loops state information for the next iteration
274 LoopToAliasMap[L] = CurAST;
278 /// SinkRegion - Walk the specified region of the CFG (defined by all blocks
279 /// dominated by the specified block, and that are in the current loop) in
280 /// reverse depth first order w.r.t the DominatorTree. This allows us to visit
281 /// uses before definitions, allowing us to sink a loop body in one pass without
284 void LICM::SinkRegion(DomTreeNode *N) {
285 assert(N != 0 && "Null dominator tree node?");
286 BasicBlock *BB = N->getBlock();
288 // If this subregion is not in the top level loop at all, exit.
289 if (!CurLoop->contains(BB)) return;
291 // We are processing blocks in reverse dfo, so process children first...
292 const std::vector<DomTreeNode*> &Children = N->getChildren();
293 for (unsigned i = 0, e = Children.size(); i != e; ++i)
294 SinkRegion(Children[i]);
296 // Only need to process the contents of this block if it is not part of a
297 // subloop (which would already have been processed).
298 if (inSubLoop(BB)) return;
300 for (BasicBlock::iterator II = BB->end(); II != BB->begin(); ) {
301 Instruction &I = *--II;
303 // Check to see if we can sink this instruction to the exit blocks
304 // of the loop. We can do this if the all users of the instruction are
305 // outside of the loop. In this case, it doesn't even matter if the
306 // operands of the instruction are loop invariant.
308 if (isNotUsedInLoop(I) && canSinkOrHoistInst(I)) {
316 /// HoistRegion - Walk the specified region of the CFG (defined by all blocks
317 /// dominated by the specified block, and that are in the current loop) in depth
318 /// first order w.r.t the DominatorTree. This allows us to visit definitions
319 /// before uses, allowing us to hoist a loop body in one pass without iteration.
321 void LICM::HoistRegion(DomTreeNode *N) {
322 assert(N != 0 && "Null dominator tree node?");
323 BasicBlock *BB = N->getBlock();
325 // If this subregion is not in the top level loop at all, exit.
326 if (!CurLoop->contains(BB)) return;
328 // Only need to process the contents of this block if it is not part of a
329 // subloop (which would already have been processed).
331 for (BasicBlock::iterator II = BB->begin(), E = BB->end(); II != E; ) {
332 Instruction &I = *II++;
334 // Try hoisting the instruction out to the preheader. We can only do this
335 // if all of the operands of the instruction are loop invariant and if it
336 // is safe to hoist the instruction.
338 if (isLoopInvariantInst(I) && canSinkOrHoistInst(I) &&
339 isSafeToExecuteUnconditionally(I))
343 const std::vector<DomTreeNode*> &Children = N->getChildren();
344 for (unsigned i = 0, e = Children.size(); i != e; ++i)
345 HoistRegion(Children[i]);
348 /// canSinkOrHoistInst - Return true if the hoister and sinker can handle this
351 bool LICM::canSinkOrHoistInst(Instruction &I) {
352 // Loads have extra constraints we have to verify before we can hoist them.
353 if (LoadInst *LI = dyn_cast<LoadInst>(&I)) {
354 if (LI->isVolatile())
355 return false; // Don't hoist volatile loads!
357 // Don't hoist loads which have may-aliased stores in loop.
359 if (LI->getType()->isSized())
360 Size = AA->getTargetData().getTypeSize(LI->getType());
361 return !pointerInvalidatedByLoop(LI->getOperand(0), Size);
362 } else if (CallInst *CI = dyn_cast<CallInst>(&I)) {
363 // Handle obvious cases efficiently.
364 if (Function *Callee = CI->getCalledFunction()) {
365 AliasAnalysis::ModRefBehavior Behavior =AA->getModRefBehavior(Callee, CI);
366 if (Behavior == AliasAnalysis::DoesNotAccessMemory)
368 else if (Behavior == AliasAnalysis::OnlyReadsMemory) {
369 // If this call only reads from memory and there are no writes to memory
370 // in the loop, we can hoist or sink the call as appropriate.
371 bool FoundMod = false;
372 for (AliasSetTracker::iterator I = CurAST->begin(), E = CurAST->end();
375 if (!AS.isForwardingAliasSet() && AS.isMod()) {
380 if (!FoundMod) return true;
384 // FIXME: This should use mod/ref information to see if we can hoist or sink
390 // Otherwise these instructions are hoistable/sinkable
391 return isa<BinaryOperator>(I) || isa<CastInst>(I) ||
392 isa<SelectInst>(I) || isa<GetElementPtrInst>(I) || isa<CmpInst>(I) ||
393 isa<InsertElementInst>(I) || isa<ExtractElementInst>(I) ||
394 isa<ShuffleVectorInst>(I);
397 /// isNotUsedInLoop - Return true if the only users of this instruction are
398 /// outside of the loop. If this is true, we can sink the instruction to the
399 /// exit blocks of the loop.
401 bool LICM::isNotUsedInLoop(Instruction &I) {
402 for (Value::use_iterator UI = I.use_begin(), E = I.use_end(); UI != E; ++UI) {
403 Instruction *User = cast<Instruction>(*UI);
404 if (PHINode *PN = dyn_cast<PHINode>(User)) {
405 // PHI node uses occur in predecessor blocks!
406 for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
407 if (PN->getIncomingValue(i) == &I)
408 if (CurLoop->contains(PN->getIncomingBlock(i)))
410 } else if (CurLoop->contains(User->getParent())) {
418 /// isLoopInvariantInst - Return true if all operands of this instruction are
419 /// loop invariant. We also filter out non-hoistable instructions here just for
422 bool LICM::isLoopInvariantInst(Instruction &I) {
423 // The instruction is loop invariant if all of its operands are loop-invariant
424 for (unsigned i = 0, e = I.getNumOperands(); i != e; ++i)
425 if (!CurLoop->isLoopInvariant(I.getOperand(i)))
428 // If we got this far, the instruction is loop invariant!
432 /// sink - When an instruction is found to only be used outside of the loop,
433 /// this function moves it to the exit blocks and patches up SSA form as needed.
434 /// This method is guaranteed to remove the original instruction from its
435 /// position, and may either delete it or move it to outside of the loop.
437 void LICM::sink(Instruction &I) {
438 DOUT << "LICM sinking instruction: " << I;
440 std::vector<BasicBlock*> ExitBlocks;
441 CurLoop->getExitBlocks(ExitBlocks);
443 if (isa<LoadInst>(I)) ++NumMovedLoads;
444 else if (isa<CallInst>(I)) ++NumMovedCalls;
448 // The case where there is only a single exit node of this loop is common
449 // enough that we handle it as a special (more efficient) case. It is more
450 // efficient to handle because there are no PHI nodes that need to be placed.
451 if (ExitBlocks.size() == 1) {
452 if (!isExitBlockDominatedByBlockInLoop(ExitBlocks[0], I.getParent())) {
453 // Instruction is not used, just delete it.
454 CurAST->deleteValue(&I);
455 if (!I.use_empty()) // If I has users in unreachable blocks, eliminate.
456 I.replaceAllUsesWith(UndefValue::get(I.getType()));
459 // Move the instruction to the start of the exit block, after any PHI
461 I.removeFromParent();
463 BasicBlock::iterator InsertPt = ExitBlocks[0]->begin();
464 while (isa<PHINode>(InsertPt)) ++InsertPt;
465 ExitBlocks[0]->getInstList().insert(InsertPt, &I);
467 } else if (ExitBlocks.size() == 0) {
468 // The instruction is actually dead if there ARE NO exit blocks.
469 CurAST->deleteValue(&I);
470 if (!I.use_empty()) // If I has users in unreachable blocks, eliminate.
471 I.replaceAllUsesWith(UndefValue::get(I.getType()));
474 // Otherwise, if we have multiple exits, use the PromoteMem2Reg function to
475 // do all of the hard work of inserting PHI nodes as necessary. We convert
476 // the value into a stack object to get it to do this.
478 // Firstly, we create a stack object to hold the value...
481 if (I.getType() != Type::VoidTy) {
482 AI = new AllocaInst(I.getType(), 0, I.getName(),
483 I.getParent()->getParent()->getEntryBlock().begin());
487 // Secondly, insert load instructions for each use of the instruction
488 // outside of the loop.
489 while (!I.use_empty()) {
490 Instruction *U = cast<Instruction>(I.use_back());
492 // If the user is a PHI Node, we actually have to insert load instructions
493 // in all predecessor blocks, not in the PHI block itself!
494 if (PHINode *UPN = dyn_cast<PHINode>(U)) {
495 // Only insert into each predecessor once, so that we don't have
496 // different incoming values from the same block!
497 std::map<BasicBlock*, Value*> InsertedBlocks;
498 for (unsigned i = 0, e = UPN->getNumIncomingValues(); i != e; ++i)
499 if (UPN->getIncomingValue(i) == &I) {
500 BasicBlock *Pred = UPN->getIncomingBlock(i);
501 Value *&PredVal = InsertedBlocks[Pred];
503 // Insert a new load instruction right before the terminator in
504 // the predecessor block.
505 PredVal = new LoadInst(AI, "", Pred->getTerminator());
506 CurAST->add(cast<LoadInst>(PredVal));
509 UPN->setIncomingValue(i, PredVal);
513 LoadInst *L = new LoadInst(AI, "", U);
514 U->replaceUsesOfWith(&I, L);
519 // Thirdly, insert a copy of the instruction in each exit block of the loop
520 // that is dominated by the instruction, storing the result into the memory
521 // location. Be careful not to insert the instruction into any particular
522 // basic block more than once.
523 std::set<BasicBlock*> InsertedBlocks;
524 BasicBlock *InstOrigBB = I.getParent();
526 for (unsigned i = 0, e = ExitBlocks.size(); i != e; ++i) {
527 BasicBlock *ExitBlock = ExitBlocks[i];
529 if (isExitBlockDominatedByBlockInLoop(ExitBlock, InstOrigBB)) {
530 // If we haven't already processed this exit block, do so now.
531 if (InsertedBlocks.insert(ExitBlock).second) {
532 // Insert the code after the last PHI node...
533 BasicBlock::iterator InsertPt = ExitBlock->begin();
534 while (isa<PHINode>(InsertPt)) ++InsertPt;
536 // If this is the first exit block processed, just move the original
537 // instruction, otherwise clone the original instruction and insert
540 if (InsertedBlocks.size() == 1) {
541 I.removeFromParent();
542 ExitBlock->getInstList().insert(InsertPt, &I);
546 CurAST->copyValue(&I, New);
547 if (!I.getName().empty())
548 New->setName(I.getName()+".le");
549 ExitBlock->getInstList().insert(InsertPt, New);
552 // Now that we have inserted the instruction, store it into the alloca
553 if (AI) new StoreInst(New, AI, InsertPt);
558 // If the instruction doesn't dominate any exit blocks, it must be dead.
559 if (InsertedBlocks.empty()) {
560 CurAST->deleteValue(&I);
564 // Finally, promote the fine value to SSA form.
566 std::vector<AllocaInst*> Allocas;
567 Allocas.push_back(AI);
568 PromoteMemToReg(Allocas, *DT, *DF, CurAST);
573 /// hoist - When an instruction is found to only use loop invariant operands
574 /// that is safe to hoist, this instruction is called to do the dirty work.
576 void LICM::hoist(Instruction &I) {
577 DOUT << "LICM hoisting to " << Preheader->getName() << ": " << I;
579 // Remove the instruction from its current basic block... but don't delete the
581 I.removeFromParent();
583 // Insert the new node in Preheader, before the terminator.
584 Preheader->getInstList().insert(Preheader->getTerminator(), &I);
586 if (isa<LoadInst>(I)) ++NumMovedLoads;
587 else if (isa<CallInst>(I)) ++NumMovedCalls;
592 /// isSafeToExecuteUnconditionally - Only sink or hoist an instruction if it is
593 /// not a trapping instruction or if it is a trapping instruction and is
594 /// guaranteed to execute.
596 bool LICM::isSafeToExecuteUnconditionally(Instruction &Inst) {
597 // If it is not a trapping instruction, it is always safe to hoist.
598 if (!Inst.isTrapping()) return true;
600 // Otherwise we have to check to make sure that the instruction dominates all
601 // of the exit blocks. If it doesn't, then there is a path out of the loop
602 // which does not execute this instruction, so we can't hoist it.
604 // If the instruction is in the header block for the loop (which is very
605 // common), it is always guaranteed to dominate the exit blocks. Since this
606 // is a common case, and can save some work, check it now.
607 if (Inst.getParent() == CurLoop->getHeader())
610 // It's always safe to load from a global or alloca.
611 if (isa<LoadInst>(Inst))
612 if (isa<AllocationInst>(Inst.getOperand(0)) ||
613 isa<GlobalVariable>(Inst.getOperand(0)))
616 // Get the exit blocks for the current loop.
617 std::vector<BasicBlock*> ExitBlocks;
618 CurLoop->getExitBlocks(ExitBlocks);
620 // For each exit block, get the DT node and walk up the DT until the
621 // instruction's basic block is found or we exit the loop.
622 for (unsigned i = 0, e = ExitBlocks.size(); i != e; ++i)
623 if (!isExitBlockDominatedByBlockInLoop(ExitBlocks[i], Inst.getParent()))
630 /// PromoteValuesInLoop - Try to promote memory values to scalars by sinking
631 /// stores out of the loop and moving loads to before the loop. We do this by
632 /// looping over the stores in the loop, looking for stores to Must pointers
633 /// which are loop invariant. We promote these memory locations to use allocas
634 /// instead. These allocas can easily be raised to register values by the
635 /// PromoteMem2Reg functionality.
637 void LICM::PromoteValuesInLoop() {
638 // PromotedValues - List of values that are promoted out of the loop. Each
639 // value has an alloca instruction for it, and a canonical version of the
641 std::vector<std::pair<AllocaInst*, Value*> > PromotedValues;
642 std::map<Value*, AllocaInst*> ValueToAllocaMap; // Map of ptr to alloca
644 FindPromotableValuesInLoop(PromotedValues, ValueToAllocaMap);
645 if (ValueToAllocaMap.empty()) return; // If there are values to promote.
648 NumPromoted += PromotedValues.size();
650 std::vector<Value*> PointerValueNumbers;
652 // Emit a copy from the value into the alloca'd value in the loop preheader
653 TerminatorInst *LoopPredInst = Preheader->getTerminator();
654 for (unsigned i = 0, e = PromotedValues.size(); i != e; ++i) {
655 Value *Ptr = PromotedValues[i].second;
657 // If we are promoting a pointer value, update alias information for the
659 Value *LoadValue = 0;
660 if (isa<PointerType>(cast<PointerType>(Ptr->getType())->getElementType())) {
661 // Locate a load or store through the pointer, and assign the same value
662 // to LI as we are loading or storing. Since we know that the value is
663 // stored in this loop, this will always succeed.
664 for (Value::use_iterator UI = Ptr->use_begin(), E = Ptr->use_end();
666 if (LoadInst *LI = dyn_cast<LoadInst>(*UI)) {
669 } else if (StoreInst *SI = dyn_cast<StoreInst>(*UI)) {
670 if (SI->getOperand(1) == Ptr) {
671 LoadValue = SI->getOperand(0);
675 assert(LoadValue && "No store through the pointer found!");
676 PointerValueNumbers.push_back(LoadValue); // Remember this for later.
679 // Load from the memory we are promoting.
680 LoadInst *LI = new LoadInst(Ptr, Ptr->getName()+".promoted", LoopPredInst);
682 if (LoadValue) CurAST->copyValue(LoadValue, LI);
684 // Store into the temporary alloca.
685 new StoreInst(LI, PromotedValues[i].first, LoopPredInst);
688 // Scan the basic blocks in the loop, replacing uses of our pointers with
689 // uses of the allocas in question.
691 const std::vector<BasicBlock*> &LoopBBs = CurLoop->getBlocks();
692 for (std::vector<BasicBlock*>::const_iterator I = LoopBBs.begin(),
693 E = LoopBBs.end(); I != E; ++I) {
694 // Rewrite all loads and stores in the block of the pointer...
695 for (BasicBlock::iterator II = (*I)->begin(), E = (*I)->end();
697 if (LoadInst *L = dyn_cast<LoadInst>(II)) {
698 std::map<Value*, AllocaInst*>::iterator
699 I = ValueToAllocaMap.find(L->getOperand(0));
700 if (I != ValueToAllocaMap.end())
701 L->setOperand(0, I->second); // Rewrite load instruction...
702 } else if (StoreInst *S = dyn_cast<StoreInst>(II)) {
703 std::map<Value*, AllocaInst*>::iterator
704 I = ValueToAllocaMap.find(S->getOperand(1));
705 if (I != ValueToAllocaMap.end())
706 S->setOperand(1, I->second); // Rewrite store instruction...
711 // Now that the body of the loop uses the allocas instead of the original
712 // memory locations, insert code to copy the alloca value back into the
713 // original memory location on all exits from the loop. Note that we only
714 // want to insert one copy of the code in each exit block, though the loop may
715 // exit to the same block more than once.
717 std::set<BasicBlock*> ProcessedBlocks;
719 std::vector<BasicBlock*> ExitBlocks;
720 CurLoop->getExitBlocks(ExitBlocks);
721 for (unsigned i = 0, e = ExitBlocks.size(); i != e; ++i)
722 if (ProcessedBlocks.insert(ExitBlocks[i]).second) {
723 // Copy all of the allocas into their memory locations.
724 BasicBlock::iterator BI = ExitBlocks[i]->begin();
725 while (isa<PHINode>(*BI))
726 ++BI; // Skip over all of the phi nodes in the block.
727 Instruction *InsertPos = BI;
729 for (unsigned i = 0, e = PromotedValues.size(); i != e; ++i) {
730 // Load from the alloca.
731 LoadInst *LI = new LoadInst(PromotedValues[i].first, "", InsertPos);
733 // If this is a pointer type, update alias info appropriately.
734 if (isa<PointerType>(LI->getType()))
735 CurAST->copyValue(PointerValueNumbers[PVN++], LI);
737 // Store into the memory we promoted.
738 new StoreInst(LI, PromotedValues[i].second, InsertPos);
742 // Now that we have done the deed, use the mem2reg functionality to promote
743 // all of the new allocas we just created into real SSA registers.
745 std::vector<AllocaInst*> PromotedAllocas;
746 PromotedAllocas.reserve(PromotedValues.size());
747 for (unsigned i = 0, e = PromotedValues.size(); i != e; ++i)
748 PromotedAllocas.push_back(PromotedValues[i].first);
749 PromoteMemToReg(PromotedAllocas, *DT, *DF, CurAST);
752 /// FindPromotableValuesInLoop - Check the current loop for stores to definite
753 /// pointers, which are not loaded and stored through may aliases. If these are
754 /// found, create an alloca for the value, add it to the PromotedValues list,
755 /// and keep track of the mapping from value to alloca.
757 void LICM::FindPromotableValuesInLoop(
758 std::vector<std::pair<AllocaInst*, Value*> > &PromotedValues,
759 std::map<Value*, AllocaInst*> &ValueToAllocaMap) {
760 Instruction *FnStart = CurLoop->getHeader()->getParent()->begin()->begin();
762 // Loop over all of the alias sets in the tracker object.
763 for (AliasSetTracker::iterator I = CurAST->begin(), E = CurAST->end();
766 // We can promote this alias set if it has a store, if it is a "Must" alias
767 // set, if the pointer is loop invariant, and if we are not eliminating any
768 // volatile loads or stores.
769 if (!AS.isForwardingAliasSet() && AS.isMod() && AS.isMustAlias() &&
770 !AS.isVolatile() && CurLoop->isLoopInvariant(AS.begin()->first)) {
771 assert(AS.begin() != AS.end() &&
772 "Must alias set should have at least one pointer element in it!");
773 Value *V = AS.begin()->first;
775 // Check that all of the pointers in the alias set have the same type. We
776 // cannot (yet) promote a memory location that is loaded and stored in
778 bool PointerOk = true;
779 for (AliasSet::iterator I = AS.begin(), E = AS.end(); I != E; ++I)
780 if (V->getType() != I->first->getType()) {
786 const Type *Ty = cast<PointerType>(V->getType())->getElementType();
787 AllocaInst *AI = new AllocaInst(Ty, 0, V->getName()+".tmp", FnStart);
788 PromotedValues.push_back(std::make_pair(AI, V));
790 // Update the AST and alias analysis.
791 CurAST->copyValue(V, AI);
793 for (AliasSet::iterator I = AS.begin(), E = AS.end(); I != E; ++I)
794 ValueToAllocaMap.insert(std::make_pair(I->first, AI));
796 DOUT << "LICM: Promoting value: " << *V << "\n";