diff options
| author | Nicola Zaghen <nicola.zaghen@imgtec.com> | 2018-05-14 12:53:11 +0000 |
|---|---|---|
| committer | Nicola Zaghen <nicola.zaghen@imgtec.com> | 2018-05-14 12:53:11 +0000 |
| commit | d34e60ca8532511acb8c93ef26297e349fbec86a (patch) | |
| tree | 1a095bc8694498d94232e81b95c1da05d462d3ec /llvm/lib/Transforms/Scalar | |
| parent | affbc99bea94e77f7ebccd8ba887e33051bd04ee (diff) | |
| download | bcm5719-llvm-d34e60ca8532511acb8c93ef26297e349fbec86a.tar.gz bcm5719-llvm-d34e60ca8532511acb8c93ef26297e349fbec86a.zip | |
Rename DEBUG macro to LLVM_DEBUG.
The DEBUG() macro is very generic so it might clash with other projects.
The renaming was done as follows:
- git grep -l 'DEBUG' | xargs sed -i 's/\bDEBUG\s\?(/LLVM_DEBUG(/g'
- git diff -U0 master | ../clang/tools/clang-format/clang-format-diff.py -i -p1 -style LLVM
- Manual change to APInt
- Manually chage DOCS as regex doesn't match it.
In the transition period the DEBUG() macro is still present and aliased
to the LLVM_DEBUG() one.
Differential Revision: https://reviews.llvm.org/D43624
llvm-svn: 332240
Diffstat (limited to 'llvm/lib/Transforms/Scalar')
46 files changed, 1271 insertions, 1175 deletions
diff --git a/llvm/lib/Transforms/Scalar/ADCE.cpp b/llvm/lib/Transforms/Scalar/ADCE.cpp index 03a44849645..ce09a477b5f 100644 --- a/llvm/lib/Transforms/Scalar/ADCE.cpp +++ b/llvm/lib/Transforms/Scalar/ADCE.cpp @@ -298,8 +298,8 @@ void AggressiveDeadCodeElimination::initialize() { auto &Info = BlockInfo[BB]; // Real function return if (isa<ReturnInst>(Info.Terminator)) { - DEBUG(dbgs() << "post-dom root child is a return: " << BB->getName() - << '\n';); + LLVM_DEBUG(dbgs() << "post-dom root child is a return: " << BB->getName() + << '\n';); continue; } @@ -356,7 +356,7 @@ void AggressiveDeadCodeElimination::markLiveInstructions() { // where we need to mark the inputs as live. while (!Worklist.empty()) { Instruction *LiveInst = Worklist.pop_back_val(); - DEBUG(dbgs() << "work live: "; LiveInst->dump();); + LLVM_DEBUG(dbgs() << "work live: "; LiveInst->dump();); for (Use &OI : LiveInst->operands()) if (Instruction *Inst = dyn_cast<Instruction>(OI)) @@ -378,7 +378,7 @@ void AggressiveDeadCodeElimination::markLive(Instruction *I) { if (Info.Live) return; - DEBUG(dbgs() << "mark live: "; I->dump()); + LLVM_DEBUG(dbgs() << "mark live: "; I->dump()); Info.Live = true; Worklist.push_back(I); @@ -402,7 +402,7 @@ void AggressiveDeadCodeElimination::markLive(Instruction *I) { void AggressiveDeadCodeElimination::markLive(BlockInfoType &BBInfo) { if (BBInfo.Live) return; - DEBUG(dbgs() << "mark block live: " << BBInfo.BB->getName() << '\n'); + LLVM_DEBUG(dbgs() << "mark block live: " << BBInfo.BB->getName() << '\n'); BBInfo.Live = true; if (!BBInfo.CFLive) { BBInfo.CFLive = true; @@ -463,7 +463,7 @@ void AggressiveDeadCodeElimination::markLiveBranchesFromControlDependences() { if (BlocksWithDeadTerminators.empty()) return; - DEBUG({ + LLVM_DEBUG({ dbgs() << "new live blocks:\n"; for (auto *BB : NewLiveBlocks) dbgs() << "\t" << BB->getName() << '\n'; @@ -487,7 +487,7 @@ void AggressiveDeadCodeElimination::markLiveBranchesFromControlDependences() { // Dead terminators which control live blocks are now marked live. for (auto *BB : IDFBlocks) { - DEBUG(dbgs() << "live control in: " << BB->getName() << '\n'); + LLVM_DEBUG(dbgs() << "live control in: " << BB->getName() << '\n'); markLive(BB->getTerminator()); } } @@ -501,7 +501,7 @@ bool AggressiveDeadCodeElimination::removeDeadInstructions() { // Updates control and dataflow around dead blocks updateDeadRegions(); - DEBUG({ + LLVM_DEBUG({ for (Instruction &I : instructions(F)) { // Check if the instruction is alive. if (isLive(&I)) @@ -555,7 +555,7 @@ bool AggressiveDeadCodeElimination::removeDeadInstructions() { // A dead region is the set of dead blocks with a common live post-dominator. void AggressiveDeadCodeElimination::updateDeadRegions() { - DEBUG({ + LLVM_DEBUG({ dbgs() << "final dead terminator blocks: " << '\n'; for (auto *BB : BlocksWithDeadTerminators) dbgs() << '\t' << BB->getName() @@ -607,8 +607,9 @@ void AggressiveDeadCodeElimination::updateDeadRegions() { // It might have happened that the same successor appeared multiple times // and the CFG edge wasn't really removed. if (Succ != PreferredSucc->BB) { - DEBUG(dbgs() << "ADCE: (Post)DomTree edge enqueued for deletion" - << BB->getName() << " -> " << Succ->getName() << "\n"); + LLVM_DEBUG(dbgs() << "ADCE: (Post)DomTree edge enqueued for deletion" + << BB->getName() << " -> " << Succ->getName() + << "\n"); DeletedEdges.push_back({DominatorTree::Delete, BB, Succ}); } } @@ -652,7 +653,7 @@ void AggressiveDeadCodeElimination::makeUnconditional(BasicBlock *BB, InstInfo[PredTerm].Live = true; return; } - DEBUG(dbgs() << "making unconditional " << BB->getName() << '\n'); + LLVM_DEBUG(dbgs() << "making unconditional " << BB->getName() << '\n'); NumBranchesRemoved += 1; IRBuilder<> Builder(PredTerm); auto *NewTerm = Builder.CreateBr(Target); diff --git a/llvm/lib/Transforms/Scalar/AlignmentFromAssumptions.cpp b/llvm/lib/Transforms/Scalar/AlignmentFromAssumptions.cpp index b84528271a7..fa7bcec677f 100644 --- a/llvm/lib/Transforms/Scalar/AlignmentFromAssumptions.cpp +++ b/llvm/lib/Transforms/Scalar/AlignmentFromAssumptions.cpp @@ -98,8 +98,8 @@ static unsigned getNewAlignmentDiff(const SCEV *DiffSCEV, const SCEV *DiffAlign = SE->getMulExpr(DiffAlignDiv, AlignSCEV); const SCEV *DiffUnitsSCEV = SE->getMinusSCEV(DiffAlign, DiffSCEV); - DEBUG(dbgs() << "\talignment relative to " << *AlignSCEV << " is " << - *DiffUnitsSCEV << " (diff: " << *DiffSCEV << ")\n"); + LLVM_DEBUG(dbgs() << "\talignment relative to " << *AlignSCEV << " is " + << *DiffUnitsSCEV << " (diff: " << *DiffSCEV << ")\n"); if (const SCEVConstant *ConstDUSCEV = dyn_cast<SCEVConstant>(DiffUnitsSCEV)) { @@ -139,12 +139,12 @@ static unsigned getNewAlignment(const SCEV *AASCEV, const SCEV *AlignSCEV, // address. This address is displaced by the provided offset. DiffSCEV = SE->getMinusSCEV(DiffSCEV, OffSCEV); - DEBUG(dbgs() << "AFI: alignment of " << *Ptr << " relative to " << - *AlignSCEV << " and offset " << *OffSCEV << - " using diff " << *DiffSCEV << "\n"); + LLVM_DEBUG(dbgs() << "AFI: alignment of " << *Ptr << " relative to " + << *AlignSCEV << " and offset " << *OffSCEV + << " using diff " << *DiffSCEV << "\n"); unsigned NewAlignment = getNewAlignmentDiff(DiffSCEV, AlignSCEV, SE); - DEBUG(dbgs() << "\tnew alignment: " << NewAlignment << "\n"); + LLVM_DEBUG(dbgs() << "\tnew alignment: " << NewAlignment << "\n"); if (NewAlignment) { return NewAlignment; @@ -160,8 +160,8 @@ static unsigned getNewAlignment(const SCEV *AASCEV, const SCEV *AlignSCEV, const SCEV *DiffStartSCEV = DiffARSCEV->getStart(); const SCEV *DiffIncSCEV = DiffARSCEV->getStepRecurrence(*SE); - DEBUG(dbgs() << "\ttrying start/inc alignment using start " << - *DiffStartSCEV << " and inc " << *DiffIncSCEV << "\n"); + LLVM_DEBUG(dbgs() << "\ttrying start/inc alignment using start " + << *DiffStartSCEV << " and inc " << *DiffIncSCEV << "\n"); // Now compute the new alignment using the displacement to the value in the // first iteration, and also the alignment using the per-iteration delta. @@ -170,26 +170,26 @@ static unsigned getNewAlignment(const SCEV *AASCEV, const SCEV *AlignSCEV, NewAlignment = getNewAlignmentDiff(DiffStartSCEV, AlignSCEV, SE); unsigned NewIncAlignment = getNewAlignmentDiff(DiffIncSCEV, AlignSCEV, SE); - DEBUG(dbgs() << "\tnew start alignment: " << NewAlignment << "\n"); - DEBUG(dbgs() << "\tnew inc alignment: " << NewIncAlignment << "\n"); + LLVM_DEBUG(dbgs() << "\tnew start alignment: " << NewAlignment << "\n"); + LLVM_DEBUG(dbgs() << "\tnew inc alignment: " << NewIncAlignment << "\n"); if (!NewAlignment || !NewIncAlignment) { return 0; } else if (NewAlignment > NewIncAlignment) { if (NewAlignment % NewIncAlignment == 0) { - DEBUG(dbgs() << "\tnew start/inc alignment: " << - NewIncAlignment << "\n"); + LLVM_DEBUG(dbgs() << "\tnew start/inc alignment: " << NewIncAlignment + << "\n"); return NewIncAlignment; } } else if (NewIncAlignment > NewAlignment) { if (NewIncAlignment % NewAlignment == 0) { - DEBUG(dbgs() << "\tnew start/inc alignment: " << - NewAlignment << "\n"); + LLVM_DEBUG(dbgs() << "\tnew start/inc alignment: " << NewAlignment + << "\n"); return NewAlignment; } } else if (NewIncAlignment == NewAlignment) { - DEBUG(dbgs() << "\tnew start/inc alignment: " << - NewAlignment << "\n"); + LLVM_DEBUG(dbgs() << "\tnew start/inc alignment: " << NewAlignment + << "\n"); return NewAlignment; } } @@ -339,7 +339,7 @@ bool AlignmentFromAssumptionsPass::processAssumption(CallInst *ACall) { unsigned NewDestAlignment = getNewAlignment(AASCEV, AlignSCEV, OffSCEV, MI->getDest(), SE); - DEBUG(dbgs() << "\tmem inst: " << NewDestAlignment << "\n";); + LLVM_DEBUG(dbgs() << "\tmem inst: " << NewDestAlignment << "\n";); if (NewDestAlignment > MI->getDestAlignment()) { MI->setDestAlignment(NewDestAlignment); ++NumMemIntAlignChanged; @@ -351,7 +351,7 @@ bool AlignmentFromAssumptionsPass::processAssumption(CallInst *ACall) { unsigned NewSrcAlignment = getNewAlignment(AASCEV, AlignSCEV, OffSCEV, MTI->getSource(), SE); - DEBUG(dbgs() << "\tmem trans: " << NewSrcAlignment << "\n";); + LLVM_DEBUG(dbgs() << "\tmem trans: " << NewSrcAlignment << "\n";); if (NewSrcAlignment > MTI->getSourceAlignment()) { MTI->setSourceAlignment(NewSrcAlignment); diff --git a/llvm/lib/Transforms/Scalar/BDCE.cpp b/llvm/lib/Transforms/Scalar/BDCE.cpp index c39e7443030..dd763847501 100644 --- a/llvm/lib/Transforms/Scalar/BDCE.cpp +++ b/llvm/lib/Transforms/Scalar/BDCE.cpp @@ -100,7 +100,7 @@ static bool bitTrackingDCE(Function &F, DemandedBits &DB) { // For live instructions that have all dead bits, first make them dead by // replacing all uses with something else. Then, if they don't need to // remain live (because they have side effects, etc.) we can remove them. - DEBUG(dbgs() << "BDCE: Trivializing: " << I << " (all bits dead)\n"); + LLVM_DEBUG(dbgs() << "BDCE: Trivializing: " << I << " (all bits dead)\n"); clearAssumptionsOfUsers(&I, DB); diff --git a/llvm/lib/Transforms/Scalar/CallSiteSplitting.cpp b/llvm/lib/Transforms/Scalar/CallSiteSplitting.cpp index dac52a6e3b7..098ad920a76 100644 --- a/llvm/lib/Transforms/Scalar/CallSiteSplitting.cpp +++ b/llvm/lib/Transforms/Scalar/CallSiteSplitting.cpp @@ -316,7 +316,7 @@ static void splitCallSite( if (!IsMustTailCall && !Instr->use_empty()) CallPN = PHINode::Create(Instr->getType(), Preds.size(), "phi.call"); - DEBUG(dbgs() << "split call-site : " << *Instr << " into \n"); + LLVM_DEBUG(dbgs() << "split call-site : " << *Instr << " into \n"); assert(Preds.size() == 2 && "The ValueToValueMaps array has size 2."); // ValueToValueMapTy is neither copy nor moveable, so we use a simple array @@ -344,8 +344,8 @@ static void splitCallSite( ++ArgNo; } } - DEBUG(dbgs() << " " << *NewCI << " in " << SplitBlock->getName() - << "\n"); + LLVM_DEBUG(dbgs() << " " << *NewCI << " in " << SplitBlock->getName() + << "\n"); if (CallPN) CallPN->addIncoming(NewCI, SplitBlock); diff --git a/llvm/lib/Transforms/Scalar/ConstantHoisting.cpp b/llvm/lib/Transforms/Scalar/ConstantHoisting.cpp index 470e687a722..7f710761611 100644 --- a/llvm/lib/Transforms/Scalar/ConstantHoisting.cpp +++ b/llvm/lib/Transforms/Scalar/ConstantHoisting.cpp @@ -132,8 +132,8 @@ bool ConstantHoistingLegacyPass::runOnFunction(Function &Fn) { if (skipFunction(Fn)) return false; - DEBUG(dbgs() << "********** Begin Constant Hoisting **********\n"); - DEBUG(dbgs() << "********** Function: " << Fn.getName() << '\n'); + LLVM_DEBUG(dbgs() << "********** Begin Constant Hoisting **********\n"); + LLVM_DEBUG(dbgs() << "********** Function: " << Fn.getName() << '\n'); bool MadeChange = Impl.runImpl(Fn, getAnalysis<TargetTransformInfoWrapperPass>().getTTI(Fn), @@ -144,11 +144,11 @@ bool ConstantHoistingLegacyPass::runOnFunction(Function &Fn) { Fn.getEntryBlock()); if (MadeChange) { - DEBUG(dbgs() << "********** Function after Constant Hoisting: " - << Fn.getName() << '\n'); - DEBUG(dbgs() << Fn); + LLVM_DEBUG(dbgs() << "********** Function after Constant Hoisting: " + << Fn.getName() << '\n'); + LLVM_DEBUG(dbgs() << Fn); } - DEBUG(dbgs() << "********** End Constant Hoisting **********\n"); + LLVM_DEBUG(dbgs() << "********** End Constant Hoisting **********\n"); return MadeChange; } @@ -364,14 +364,13 @@ void ConstantHoistingPass::collectConstantCandidates( Itr->second = ConstCandVec.size() - 1; } ConstCandVec[Itr->second].addUser(Inst, Idx, Cost); - DEBUG(if (isa<ConstantInt>(Inst->getOperand(Idx))) - dbgs() << "Collect constant " << *ConstInt << " from " << *Inst + LLVM_DEBUG(if (isa<ConstantInt>(Inst->getOperand(Idx))) dbgs() + << "Collect constant " << *ConstInt << " from " << *Inst << " with cost " << Cost << '\n'; - else - dbgs() << "Collect constant " << *ConstInt << " indirectly from " - << *Inst << " via " << *Inst->getOperand(Idx) << " with cost " - << Cost << '\n'; - ); + else dbgs() << "Collect constant " << *ConstInt + << " indirectly from " << *Inst << " via " + << *Inst->getOperand(Idx) << " with cost " << Cost + << '\n';); } } @@ -501,20 +500,21 @@ ConstantHoistingPass::maximizeConstantsInRange(ConstCandVecType::iterator S, return NumUses; } - DEBUG(dbgs() << "== Maximize constants in range ==\n"); + LLVM_DEBUG(dbgs() << "== Maximize constants in range ==\n"); int MaxCost = -1; for (auto ConstCand = S; ConstCand != E; ++ConstCand) { auto Value = ConstCand->ConstInt->getValue(); Type *Ty = ConstCand->ConstInt->getType(); int Cost = 0; NumUses += ConstCand->Uses.size(); - DEBUG(dbgs() << "= Constant: " << ConstCand->ConstInt->getValue() << "\n"); + LLVM_DEBUG(dbgs() << "= Constant: " << ConstCand->ConstInt->getValue() + << "\n"); for (auto User : ConstCand->Uses) { unsigned Opcode = User.Inst->getOpcode(); unsigned OpndIdx = User.OpndIdx; Cost += TTI->getIntImmCost(Opcode, OpndIdx, Value, Ty); - DEBUG(dbgs() << "Cost: " << Cost << "\n"); + LLVM_DEBUG(dbgs() << "Cost: " << Cost << "\n"); for (auto C2 = S; C2 != E; ++C2) { Optional<APInt> Diff = calculateOffsetDiff( @@ -524,18 +524,18 @@ ConstantHoistingPass::maximizeConstantsInRange(ConstCandVecType::iterator S, const int ImmCosts = TTI->getIntImmCodeSizeCost(Opcode, OpndIdx, Diff.getValue(), Ty); Cost -= ImmCosts; - DEBUG(dbgs() << "Offset " << Diff.getValue() << " " - << "has penalty: " << ImmCosts << "\n" - << "Adjusted cost: " << Cost << "\n"); + LLVM_DEBUG(dbgs() << "Offset " << Diff.getValue() << " " + << "has penalty: " << ImmCosts << "\n" + << "Adjusted cost: " << Cost << "\n"); } } } - DEBUG(dbgs() << "Cumulative cost: " << Cost << "\n"); + LLVM_DEBUG(dbgs() << "Cumulative cost: " << Cost << "\n"); if (Cost > MaxCost) { MaxCost = Cost; MaxCostItr = ConstCand; - DEBUG(dbgs() << "New candidate: " << MaxCostItr->ConstInt->getValue() - << "\n"); + LLVM_DEBUG(dbgs() << "New candidate: " << MaxCostItr->ConstInt->getValue() + << "\n"); } } return NumUses; @@ -641,19 +641,20 @@ void ConstantHoistingPass::emitBaseConstants(Instruction *Base, Mat = BinaryOperator::Create(Instruction::Add, Base, Offset, "const_mat", InsertionPt); - DEBUG(dbgs() << "Materialize constant (" << *Base->getOperand(0) - << " + " << *Offset << ") in BB " - << Mat->getParent()->getName() << '\n' << *Mat << '\n'); + LLVM_DEBUG(dbgs() << "Materialize constant (" << *Base->getOperand(0) + << " + " << *Offset << ") in BB " + << Mat->getParent()->getName() << '\n' + << *Mat << '\n'); Mat->setDebugLoc(ConstUser.Inst->getDebugLoc()); } Value *Opnd = ConstUser.Inst->getOperand(ConstUser.OpndIdx); // Visit constant integer. if (isa<ConstantInt>(Opnd)) { - DEBUG(dbgs() << "Update: " << *ConstUser.Inst << '\n'); + LLVM_DEBUG(dbgs() << "Update: " << *ConstUser.Inst << '\n'); if (!updateOperand(ConstUser.Inst, ConstUser.OpndIdx, Mat) && Offset) Mat->eraseFromParent(); - DEBUG(dbgs() << "To : " << *ConstUser.Inst << '\n'); + LLVM_DEBUG(dbgs() << "To : " << *ConstUser.Inst << '\n'); return; } @@ -669,13 +670,13 @@ void ConstantHoistingPass::emitBaseConstants(Instruction *Base, ClonedCastInst->insertAfter(CastInst); // Use the same debug location as the original cast instruction. ClonedCastInst->setDebugLoc(CastInst->getDebugLoc()); - DEBUG(dbgs() << "Clone instruction: " << *CastInst << '\n' - << "To : " << *ClonedCastInst << '\n'); + LLVM_DEBUG(dbgs() << "Clone instruction: " << *CastInst << '\n' + << "To : " << *ClonedCastInst << '\n'); } - DEBUG(dbgs() << "Update: " << *ConstUser.Inst << '\n'); + LLVM_DEBUG(dbgs() << "Update: " << *ConstUser.Inst << '\n'); updateOperand(ConstUser.Inst, ConstUser.OpndIdx, ClonedCastInst); - DEBUG(dbgs() << "To : " << *ConstUser.Inst << '\n'); + LLVM_DEBUG(dbgs() << "To : " << *ConstUser.Inst << '\n'); return; } @@ -689,15 +690,15 @@ void ConstantHoistingPass::emitBaseConstants(Instruction *Base, // Use the same debug location as the instruction we are about to update. ConstExprInst->setDebugLoc(ConstUser.Inst->getDebugLoc()); - DEBUG(dbgs() << "Create instruction: " << *ConstExprInst << '\n' - << "From : " << *ConstExpr << '\n'); - DEBUG(dbgs() << "Update: " << *ConstUser.Inst << '\n'); + LLVM_DEBUG(dbgs() << "Create instruction: " << *ConstExprInst << '\n' + << "From : " << *ConstExpr << '\n'); + LLVM_DEBUG(dbgs() << "Update: " << *ConstUser.Inst << '\n'); if (!updateOperand(ConstUser.Inst, ConstUser.OpndIdx, ConstExprInst)) { ConstExprInst->eraseFromParent(); if (Offset) Mat->eraseFromParent(); } - DEBUG(dbgs() << "To : " << *ConstUser.Inst << '\n'); + LLVM_DEBUG(dbgs() << "To : " << *ConstUser.Inst << '\n'); return; } } @@ -720,9 +721,9 @@ bool ConstantHoistingPass::emitBaseConstants() { Base->setDebugLoc(IP->getDebugLoc()); - DEBUG(dbgs() << "Hoist constant (" << *ConstInfo.BaseConstant - << ") to BB " << IP->getParent()->getName() << '\n' - << *Base << '\n'); + LLVM_DEBUG(dbgs() << "Hoist constant (" << *ConstInfo.BaseConstant + << ") to BB " << IP->getParent()->getName() << '\n' + << *Base << '\n'); // Emit materialization code for all rebased constants. unsigned Uses = 0; diff --git a/llvm/lib/Transforms/Scalar/CorrelatedValuePropagation.cpp b/llvm/lib/Transforms/Scalar/CorrelatedValuePropagation.cpp index a85daaa0780..15df482652c 100644 --- a/llvm/lib/Transforms/Scalar/CorrelatedValuePropagation.cpp +++ b/llvm/lib/Transforms/Scalar/CorrelatedValuePropagation.cpp @@ -228,7 +228,7 @@ static bool processPHI(PHINode *P, LazyValueInfo *LVI, V = SI->getTrueValue(); } - DEBUG(dbgs() << "CVP: Threading PHI over " << *SI << '\n'); + LLVM_DEBUG(dbgs() << "CVP: Threading PHI over " << *SI << '\n'); } P->setIncomingValue(i, V); diff --git a/llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp b/llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp index 1d56a6e8aaa..fd2cacf461b 100644 --- a/llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp +++ b/llvm/lib/Transforms/Scalar/DeadStoreElimination.cpp @@ -420,9 +420,10 @@ static OverwriteResult isOverwrite(const MemoryLocation &Later, // Insert our part of the overlap into the map. auto &IM = IOL[DepWrite]; - DEBUG(dbgs() << "DSE: Partial overwrite: Earlier [" << EarlierOff << ", " << - int64_t(EarlierOff + Earlier.Size) << ") Later [" << - LaterOff << ", " << int64_t(LaterOff + Later.Size) << ")\n"); + LLVM_DEBUG(dbgs() << "DSE: Partial overwrite: Earlier [" << EarlierOff + << ", " << int64_t(EarlierOff + Earlier.Size) + << ") Later [" << LaterOff << ", " + << int64_t(LaterOff + Later.Size) << ")\n"); // Make sure that we only insert non-overlapping intervals and combine // adjacent intervals. The intervals are stored in the map with the ending @@ -459,11 +460,11 @@ static OverwriteResult isOverwrite(const MemoryLocation &Later, ILI = IM.begin(); if (ILI->second <= EarlierOff && ILI->first >= int64_t(EarlierOff + Earlier.Size)) { - DEBUG(dbgs() << "DSE: Full overwrite from partials: Earlier [" << - EarlierOff << ", " << - int64_t(EarlierOff + Earlier.Size) << - ") Composite Later [" << - ILI->second << ", " << ILI->first << ")\n"); + LLVM_DEBUG(dbgs() << "DSE: Full overwrite from partials: Earlier [" + << EarlierOff << ", " + << int64_t(EarlierOff + Earlier.Size) + << ") Composite Later [" << ILI->second << ", " + << ILI->first << ")\n"); ++NumCompletePartials; return OW_Complete; } @@ -474,10 +475,11 @@ static OverwriteResult isOverwrite(const MemoryLocation &Later, if (EnablePartialStoreMerging && LaterOff >= EarlierOff && int64_t(EarlierOff + Earlier.Size) > LaterOff && uint64_t(LaterOff - EarlierOff) + Later.Size <= Earlier.Size) { - DEBUG(dbgs() << "DSE: Partial overwrite an earlier load [" << EarlierOff - << ", " << int64_t(EarlierOff + Earlier.Size) - << ") by a later store [" << LaterOff << ", " - << int64_t(LaterOff + Later.Size) << ")\n"); + LLVM_DEBUG(dbgs() << "DSE: Partial overwrite an earlier load [" + << EarlierOff << ", " + << int64_t(EarlierOff + Earlier.Size) + << ") by a later store [" << LaterOff << ", " + << int64_t(LaterOff + Later.Size) << ")\n"); // TODO: Maybe come up with a better name? return OW_PartialEarlierWithFullLater; } @@ -677,8 +679,9 @@ static bool handleFree(CallInst *F, AliasAnalysis *AA, if (!AA->isMustAlias(F->getArgOperand(0), DepPointer)) break; - DEBUG(dbgs() << "DSE: Dead Store to soon to be freed memory:\n DEAD: " - << *Dependency << '\n'); + LLVM_DEBUG( + dbgs() << "DSE: Dead Store to soon to be freed memory:\n DEAD: " + << *Dependency << '\n'); // DCE instructions only used to calculate that store. BasicBlock::iterator BBI(Dependency); @@ -787,15 +790,16 @@ static bool handleEndBlock(BasicBlock &BB, AliasAnalysis *AA, if (AllDead) { Instruction *Dead = &*BBI; - DEBUG(dbgs() << "DSE: Dead Store at End of Block:\n DEAD: " - << *Dead << "\n Objects: "; - for (SmallVectorImpl<Value *>::iterator I = Pointers.begin(), - E = Pointers.end(); I != E; ++I) { - dbgs() << **I; - if (std::next(I) != E) - dbgs() << ", "; - } - dbgs() << '\n'); + LLVM_DEBUG(dbgs() << "DSE: Dead Store at End of Block:\n DEAD: " + << *Dead << "\n Objects: "; + for (SmallVectorImpl<Value *>::iterator I = Pointers.begin(), + E = Pointers.end(); + I != E; ++I) { + dbgs() << **I; + if (std::next(I) != E) + dbgs() << ", "; + } dbgs() + << '\n'); // DCE instructions only used to calculate that store. deleteDeadInstruction(Dead, &BBI, *MD, *TLI, IOL, InstrOrdering, &DeadStackObjects); @@ -807,8 +811,8 @@ static bool handleEndBlock(BasicBlock &BB, AliasAnalysis *AA, // Remove any dead non-memory-mutating instructions. if (isInstructionTriviallyDead(&*BBI, TLI)) { - DEBUG(dbgs() << "DSE: Removing trivially dead instruction:\n DEAD: " - << *&*BBI << '\n'); + LLVM_DEBUG(dbgs() << "DSE: Removing trivially dead instruction:\n DEAD: " + << *&*BBI << '\n'); deleteDeadInstruction(&*BBI, &BBI, *MD, *TLI, IOL, InstrOrdering, &DeadStackObjects); ++NumFastOther; MadeChange = true; @@ -917,10 +921,10 @@ static bool tryToShorten(Instruction *EarlierWrite, int64_t &EarlierOffset, return false; } - DEBUG(dbgs() << "DSE: Remove Dead Store:\n OW " - << (IsOverwriteEnd ? "END" : "BEGIN") << ": " << *EarlierWrite - << "\n KILLER (offset " << LaterOffset << ", " << EarlierSize - << ")\n"); + LLVM_DEBUG(dbgs() << "DSE: Remove Dead Store:\n OW " + << (IsOverwriteEnd ? "END" : "BEGIN") << ": " + << *EarlierWrite << "\n KILLER (offset " << LaterOffset + << ", " << EarlierSize << ")\n"); Value *EarlierWriteLength = EarlierIntrinsic->getLength(); Value *TrimmedLength = @@ -1025,8 +1029,9 @@ static bool eliminateNoopStore(Instruction *Inst, BasicBlock::iterator &BBI, if (SI->getPointerOperand() == DepLoad->getPointerOperand() && isRemovable(SI) && memoryIsNotModifiedBetween(DepLoad, SI, AA)) { - DEBUG(dbgs() << "DSE: Remove Store Of Load from same pointer:\n LOAD: " - << *DepLoad << "\n STORE: " << *SI << '\n'); + LLVM_DEBUG( + dbgs() << "DSE: Remove Store Of Load from same pointer:\n LOAD: " + << *DepLoad << "\n STORE: " << *SI << '\n'); deleteDeadInstruction(SI, &BBI, *MD, *TLI, IOL, InstrOrdering); ++NumRedundantStores; @@ -1042,7 +1047,7 @@ static bool eliminateNoopStore(Instruction *Inst, BasicBlock::iterator &BBI, if (UnderlyingPointer && isCallocLikeFn(UnderlyingPointer, TLI) && memoryIsNotModifiedBetween(UnderlyingPointer, SI, AA)) { - DEBUG( + LLVM_DEBUG( dbgs() << "DSE: Remove null store to the calloc'ed object:\n DEAD: " << *Inst << "\n OBJECT: " << *UnderlyingPointer << '\n'); @@ -1173,8 +1178,8 @@ static bool eliminateDeadStores(BasicBlock &BB, AliasAnalysis *AA, OverwriteResult OR = isOverwrite(Loc, DepLoc, DL, *TLI, DepWriteOffset, InstWriteOffset, DepWrite, IOL, *AA); if (OR == OW_Complete) { - DEBUG(dbgs() << "DSE: Remove Dead Store:\n DEAD: " - << *DepWrite << "\n KILLER: " << *Inst << '\n'); + LLVM_DEBUG(dbgs() << "DSE: Remove Dead Store:\n DEAD: " << *DepWrite + << "\n KILLER: " << *Inst << '\n'); // Delete the store and now-dead instructions that feed it. deleteDeadInstruction(DepWrite, &BBI, *MD, *TLI, IOL, &InstrOrdering); @@ -1232,9 +1237,9 @@ static bool eliminateDeadStores(BasicBlock &BB, AliasAnalysis *AA, // store, shifted appropriately. APInt Merged = (EarlierValue & ~Mask) | (LaterValue << LShiftAmount); - DEBUG(dbgs() << "DSE: Merge Stores:\n Earlier: " << *DepWrite - << "\n Later: " << *Inst - << "\n Merged Value: " << Merged << '\n'); + LLVM_DEBUG(dbgs() << "DSE: Merge Stores:\n Earlier: " << *DepWrite + << "\n Later: " << *Inst + << "\n Merged Value: " << Merged << '\n'); auto *SI = new StoreInst( ConstantInt::get(Earlier->getValueOperand()->getType(), Merged), diff --git a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp index 4380812968a..d19ceab3889 100644 --- a/llvm/lib/Transforms/Scalar/EarlyCSE.cpp +++ b/llvm/lib/Transforms/Scalar/EarlyCSE.cpp @@ -728,11 +728,11 @@ bool EarlyCSE::processNode(DomTreeNode *Node) { ? ConstantInt::getTrue(BB->getContext()) : ConstantInt::getFalse(BB->getContext()); AvailableValues.insert(CondInst, TorF); - DEBUG(dbgs() << "EarlyCSE CVP: Add conditional value for '" - << CondInst->getName() << "' as " << *TorF << " in " - << BB->getName() << "\n"); + LLVM_DEBUG(dbgs() << "EarlyCSE CVP: Add conditional value for '" + << CondInst->getName() << "' as " << *TorF << " in " + << BB->getName() << "\n"); if (!DebugCounter::shouldExecute(CSECounter)) { - DEBUG(dbgs() << "Skipping due to debug counter\n"); + LLVM_DEBUG(dbgs() << "Skipping due to debug counter\n"); } else { // Replace all dominated uses with the known value. if (unsigned Count = replaceDominatedUsesWith( @@ -758,9 +758,9 @@ bool EarlyCSE::processNode(DomTreeNode *Node) { // Dead instructions should just be removed. if (isInstructionTriviallyDead(Inst, &TLI)) { - DEBUG(dbgs() << "EarlyCSE DCE: " << *Inst << '\n'); + LLVM_DEBUG(dbgs() << "EarlyCSE DCE: " << *Inst << '\n'); if (!DebugCounter::shouldExecute(CSECounter)) { - DEBUG(dbgs() << "Skipping due to debug counter\n"); + LLVM_DEBUG(dbgs() << "Skipping due to debug counter\n"); continue; } salvageDebugInfo(*Inst); @@ -779,16 +779,17 @@ bool EarlyCSE::processNode(DomTreeNode *Node) { auto *CondI = dyn_cast<Instruction>(cast<CallInst>(Inst)->getArgOperand(0)); if (CondI && SimpleValue::canHandle(CondI)) { - DEBUG(dbgs() << "EarlyCSE considering assumption: " << *Inst << '\n'); + LLVM_DEBUG(dbgs() << "EarlyCSE considering assumption: " << *Inst + << '\n'); AvailableValues.insert(CondI, ConstantInt::getTrue(BB->getContext())); } else - DEBUG(dbgs() << "EarlyCSE skipping assumption: " << *Inst << '\n'); + LLVM_DEBUG(dbgs() << "EarlyCSE skipping assumption: " << *Inst << '\n'); continue; } // Skip sideeffect intrinsics, for the same reason as assume intrinsics. if (match(Inst, m_Intrinsic<Intrinsic::sideeffect>())) { - DEBUG(dbgs() << "EarlyCSE skipping sideeffect: " << *Inst << '\n'); + LLVM_DEBUG(dbgs() << "EarlyCSE skipping sideeffect: " << *Inst << '\n'); continue; } @@ -826,7 +827,8 @@ bool EarlyCSE::processNode(DomTreeNode *Node) { // Is the condition known to be true? if (isa<ConstantInt>(KnownCond) && cast<ConstantInt>(KnownCond)->isOne()) { - DEBUG(dbgs() << "EarlyCSE removing guard: " << *Inst << '\n'); + LLVM_DEBUG(dbgs() + << "EarlyCSE removing guard: " << *Inst << '\n'); removeMSSA(Inst); Inst->eraseFromParent(); Changed = true; @@ -851,9 +853,10 @@ bool EarlyCSE::processNode(DomTreeNode *Node) { // If the instruction can be simplified (e.g. X+0 = X) then replace it with // its simpler value. if (Value *V = SimplifyInstruction(Inst, SQ)) { - DEBUG(dbgs() << "EarlyCSE Simplify: " << *Inst << " to: " << *V << '\n'); + LLVM_DEBUG(dbgs() << "EarlyCSE Simplify: " << *Inst << " to: " << *V + << '\n'); if (!DebugCounter::shouldExecute(CSECounter)) { - DEBUG(dbgs() << "Skipping due to debug counter\n"); + LLVM_DEBUG(dbgs() << "Skipping due to debug counter\n"); } else { bool Killed = false; if (!Inst->use_empty()) { @@ -877,9 +880,10 @@ bool EarlyCSE::processNode(DomTreeNode *Node) { if (SimpleValue::canHandle(Inst)) { // See if the instruction has an available value. If so, use it. if (Value *V = AvailableValues.lookup(Inst)) { - DEBUG(dbgs() << "EarlyCSE CSE: " << *Inst << " to: " << *V << '\n'); + LLVM_DEBUG(dbgs() << "EarlyCSE CSE: " << *Inst << " to: " << *V + << '\n'); if (!DebugCounter::shouldExecute(CSECounter)) { - DEBUG(dbgs() << "Skipping due to debug counter\n"); + LLVM_DEBUG(dbgs() << "Skipping due to debug counter\n"); continue; } if (auto *I = dyn_cast<Instruction>(V)) @@ -937,10 +941,10 @@ bool EarlyCSE::processNode(DomTreeNode *Node) { InVal.DefInst, Inst))) { Value *Op = getOrCreateResult(InVal.DefInst, Inst->getType()); if (Op != nullptr) { - DEBUG(dbgs() << "EarlyCSE CSE LOAD: " << *Inst - << " to: " << *InVal.DefInst << '\n'); + LLVM_DEBUG(dbgs() << "EarlyCSE CSE LOAD: " << *Inst + << " to: " << *InVal.DefInst << '\n'); if (!DebugCounter::shouldExecute(CSECounter)) { - DEBUG(dbgs() << "Skipping due to debug counter\n"); + LLVM_DEBUG(dbgs() << "Skipping due to debug counter\n"); continue; } if (!Inst->use_empty()) @@ -980,10 +984,10 @@ bool EarlyCSE::processNode(DomTreeNode *Node) { if (InVal.first != nullptr && isSameMemGeneration(InVal.second, CurrentGeneration, InVal.first, Inst)) { - DEBUG(dbgs() << "EarlyCSE CSE CALL: " << *Inst - << " to: " << *InVal.first << '\n'); + LLVM_DEBUG(dbgs() << "EarlyCSE CSE CALL: " << *Inst + << " to: " << *InVal.first << '\n'); if (!DebugCounter::shouldExecute(CSECounter)) { - DEBUG(dbgs() << "Skipping due to debug counter\n"); + LLVM_DEBUG(dbgs() << "Skipping due to debug counter\n"); continue; } if (!Inst->use_empty()) @@ -1036,9 +1040,9 @@ bool EarlyCSE::processNode(DomTreeNode *Node) { MemInst.getPointerOperand() || MSSA) && "can't have an intervening store if not using MemorySSA!"); - DEBUG(dbgs() << "EarlyCSE DSE (writeback): " << *Inst << '\n'); + LLVM_DEBUG(dbgs() << "EarlyCSE DSE (writeback): " << *Inst << '\n'); if (!DebugCounter::shouldExecute(CSECounter)) { - DEBUG(dbgs() << "Skipping due to debug counter\n"); + LLVM_DEBUG(dbgs() << "Skipping due to debug counter\n"); continue; } removeMSSA(Inst); @@ -1071,10 +1075,10 @@ bool EarlyCSE::processNode(DomTreeNode *Node) { !LastStoreMemInst.isVolatile() && "Violated invariant"); if (LastStoreMemInst.isMatchingMemLoc(MemInst)) { - DEBUG(dbgs() << "EarlyCSE DEAD STORE: " << *LastStore - << " due to: " << *Inst << '\n'); + LLVM_DEBUG(dbgs() << "EarlyCSE DEAD STORE: " << *LastStore + << " due to: " << *Inst << '\n'); if (!DebugCounter::shouldExecute(CSECounter)) { - DEBUG(dbgs() << "Skipping due to debug counter\n"); + LLVM_DEBUG(dbgs() << "Skipping due to debug counter\n"); } else { removeMSSA(LastStore); LastStore->eraseFromParent(); diff --git a/llvm/lib/Transforms/Scalar/Float2Int.cpp b/llvm/lib/Transforms/Scalar/Float2Int.cpp index b105ece8dc7..f2828e80bc5 100644 --- a/llvm/lib/Transforms/Scalar/Float2Int.cpp +++ b/llvm/lib/Transforms/Scalar/Float2Int.cpp @@ -138,7 +138,7 @@ void Float2IntPass::findRoots(Function &F, SmallPtrSet<Instruction*,8> &Roots) { // Helper - mark I as having been traversed, having range R. void Float2IntPass::seen(Instruction *I, ConstantRange R) { - DEBUG(dbgs() << "F2I: " << *I << ":" << R << "\n"); + LLVM_DEBUG(dbgs() << "F2I: " << *I << ":" << R << "\n"); auto IT = SeenInsts.find(I); if (IT != SeenInsts.end()) IT->second = std::move(R); @@ -359,7 +359,7 @@ bool Float2IntPass::validateAndTransform() { for (User *U : I->users()) { Instruction *UI = dyn_cast<Instruction>(U); if (!UI || SeenInsts.find(UI) == SeenInsts.end()) { - DEBUG(dbgs() << "F2I: Failing because of " << *U << "\n"); + LLVM_DEBUG(dbgs() << "F2I: Failing because of " << *U << "\n"); Fail = true; break; } @@ -380,7 +380,7 @@ bool Float2IntPass::validateAndTransform() { // lower limits, plus one so it can be signed. unsigned MinBW = std::max(R.getLower().getMinSignedBits(), R.getUpper().getMinSignedBits()) + 1; - DEBUG(dbgs() << "F2I: MinBitwidth=" << MinBW << ", R: " << R << "\n"); + LLVM_DEBUG(dbgs() << "F2I: MinBitwidth=" << MinBW << ", R: " << R << "\n"); // If we've run off the realms of the exactly representable integers, // the floating point result will differ from an integer approximation. @@ -391,11 +391,12 @@ bool Float2IntPass::validateAndTransform() { unsigned MaxRepresentableBits = APFloat::semanticsPrecision(ConvertedToTy->getFltSemantics()) - 1; if (MinBW > MaxRepresentableBits) { - DEBUG(dbgs() << "F2I: Value not guaranteed to be representable!\n"); + LLVM_DEBUG(dbgs() << "F2I: Value not guaranteed to be representable!\n"); continue; } if (MinBW > 64) { - DEBUG(dbgs() << "F2I: Value requires more than 64 bits to represent!\n"); + LLVM_DEBUG( + dbgs() << "F2I: Value requires more than 64 bits to represent!\n"); continue; } @@ -490,7 +491,7 @@ void Float2IntPass::cleanup() { } bool Float2IntPass::runImpl(Function &F) { - DEBUG(dbgs() << "F2I: Looking at function " << F.getName() << "\n"); + LLVM_DEBUG(dbgs() << "F2I: Looking at function " << F.getName() << "\n"); // Clear out all state. ECs = EquivalenceClasses<Instruction*>(); SeenInsts.clear(); diff --git a/llvm/lib/Transforms/Scalar/GVN.cpp b/llvm/lib/Transforms/Scalar/GVN.cpp index 59b87e9a77d..ff3c2f9fbd2 100644 --- a/llvm/lib/Transforms/Scalar/GVN.cpp +++ b/llvm/lib/Transforms/Scalar/GVN.cpp @@ -784,9 +784,10 @@ Value *AvailableValue::MaterializeAdjustedValue(LoadInst *LI, if (Res->getType() != LoadTy) { Res = getStoreValueForLoad(Res, Offset, LoadTy, InsertPt, DL); - DEBUG(dbgs() << "GVN COERCED NONLOCAL VAL:\nOffset: " << Offset << " " - << *getSimpleValue() << '\n' - << *Res << '\n' << "\n\n\n"); + LLVM_DEBUG(dbgs() << "GVN COERCED NONLOCAL VAL:\nOffset: " << Offset + << " " << *getSimpleValue() << '\n' + << *Res << '\n' + << "\n\n\n"); } } else if (isCoercedLoadValue()) { LoadInst *Load = getCoercedLoadValue(); @@ -800,20 +801,21 @@ Value *AvailableValue::MaterializeAdjustedValue(LoadInst *LI, // but then there all of the operations based on it would need to be // rehashed. Just leave the dead load around. gvn.getMemDep().removeInstruction(Load); - DEBUG(dbgs() << "GVN COERCED NONLOCAL LOAD:\nOffset: " << Offset << " " - << *getCoercedLoadValue() << '\n' - << *Res << '\n' - << "\n\n\n"); + LLVM_DEBUG(dbgs() << "GVN COERCED NONLOCAL LOAD:\nOffset: " << Offset + << " " << *getCoercedLoadValue() << '\n' + << *Res << '\n' + << "\n\n\n"); } } else if (isMemIntrinValue()) { Res = getMemInstValueForLoad(getMemIntrinValue(), Offset, LoadTy, InsertPt, DL); - DEBUG(dbgs() << "GVN COERCED NONLOCAL MEM INTRIN:\nOffset: " << Offset - << " " << *getMemIntrinValue() << '\n' - << *Res << '\n' << "\n\n\n"); + LLVM_DEBUG(dbgs() << "GVN COERCED NONLOCAL MEM INTRIN:\nOffset: " << Offset + << " " << *getMemIntrinValue() << '\n' + << *Res << '\n' + << "\n\n\n"); } else { assert(isUndefValue() && "Should be UndefVal"); - DEBUG(dbgs() << "GVN COERCED NONLOCAL Undef:\n";); + LLVM_DEBUG(dbgs() << "GVN COERCED NONLOCAL Undef:\n";); return UndefValue::get(LoadTy); } assert(Res && "failed to materialize?"); @@ -915,13 +917,11 @@ bool GVN::AnalyzeLoadAvailability(LoadInst *LI, MemDepResult DepInfo, } } // Nothing known about this clobber, have to be conservative - DEBUG( - // fast print dep, using operator<< on instruction is too slow. - dbgs() << "GVN: load "; - LI->printAsOperand(dbgs()); - Instruction *I = DepInfo.getInst(); - dbgs() << " is clobbered by " << *I << '\n'; - ); + LLVM_DEBUG( + // fast print dep, using operator<< on instruction is too slow. + dbgs() << "GVN: load "; LI->printAsOperand(dbgs()); + Instruction *I = DepInfo.getInst(); + dbgs() << " is clobbered by " << *I << '\n';); if (ORE->allowExtraAnalysis(DEBUG_TYPE)) reportMayClobberedLoad(LI, DepInfo, DT, ORE); @@ -979,12 +979,10 @@ bool GVN::AnalyzeLoadAvailability(LoadInst *LI, MemDepResult DepInfo, } // Unknown def - must be conservative - DEBUG( - // fast print dep, using operator<< on instruction is too slow. - dbgs() << "GVN: load "; - LI->printAsOperand(dbgs()); - dbgs() << " has unknown def " << *DepInst << '\n'; - ); + LLVM_DEBUG( + // fast print dep, using operator<< on instruction is too slow. + dbgs() << "GVN: load "; LI->printAsOperand(dbgs()); + dbgs() << " has unknown def " << *DepInst << '\n';); return false; } @@ -1114,9 +1112,9 @@ bool GVN::PerformLoadPRE(LoadInst *LI, AvailValInBlkVect &ValuesPerBlock, // If any predecessor block is an EH pad that does not allow non-PHI // instructions before the terminator, we can't PRE the load. if (Pred->getTerminator()->isEHPad()) { - DEBUG(dbgs() - << "COULD NOT PRE LOAD BECAUSE OF AN EH PAD PREDECESSOR '" - << Pred->getName() << "': " << *LI << '\n'); + LLVM_DEBUG( + dbgs() << "COULD NOT PRE LOAD BECAUSE OF AN EH PAD PREDECESSOR '" + << Pred->getName() << "': " << *LI << '\n'); return false; } @@ -1126,15 +1124,16 @@ bool GVN::PerformLoadPRE(LoadInst *LI, AvailValInBlkVect &ValuesPerBlock, if (Pred->getTerminator()->getNumSuccessors() != 1) { if (isa<IndirectBrInst>(Pred->getTerminator())) { - DEBUG(dbgs() << "COULD NOT PRE LOAD BECAUSE OF INDBR CRITICAL EDGE '" - << Pred->getName() << "': " << *LI << '\n'); + LLVM_DEBUG( + dbgs() << "COULD NOT PRE LOAD BECAUSE OF INDBR CRITICAL EDGE '" + << Pred->getName() << "': " << *LI << '\n'); return false; } if (LoadBB->isEHPad()) { - DEBUG(dbgs() - << "COULD NOT PRE LOAD BECAUSE OF AN EH PAD CRITICAL EDGE '" - << Pred->getName() << "': " << *LI << '\n'); + LLVM_DEBUG( + dbgs() << "COULD NOT PRE LOAD BECAUSE OF AN EH PAD CRITICAL EDGE '" + << Pred->getName() << "': " << *LI << '\n'); return false; } @@ -1162,8 +1161,8 @@ bool GVN::PerformLoadPRE(LoadInst *LI, AvailValInBlkVect &ValuesPerBlock, BasicBlock *NewPred = splitCriticalEdges(OrigPred, LoadBB); assert(!PredLoads.count(OrigPred) && "Split edges shouldn't be in map!"); PredLoads[NewPred] = nullptr; - DEBUG(dbgs() << "Split critical edge " << OrigPred->getName() << "->" - << LoadBB->getName() << '\n'); + LLVM_DEBUG(dbgs() << "Split critical edge " << OrigPred->getName() << "->" + << LoadBB->getName() << '\n'); } // Check if the load can safely be moved to all the unavailable predecessors. @@ -1187,8 +1186,8 @@ bool GVN::PerformLoadPRE(LoadInst *LI, AvailValInBlkVect &ValuesPerBlock, // If we couldn't find or insert a computation of this phi translated value, // we fail PRE. if (!LoadPtr) { - DEBUG(dbgs() << "COULDN'T INSERT PHI TRANSLATED VALUE OF: " - << *LI->getPointerOperand() << "\n"); + LLVM_DEBUG(dbgs() << "COULDN'T INSERT PHI TRANSLATED VALUE OF: " + << *LI->getPointerOperand() << "\n"); CanDoPRE = false; break; } @@ -1209,10 +1208,10 @@ bool GVN::PerformLoadPRE(LoadInst *LI, AvailValInBlkVect &ValuesPerBlock, // Okay, we can eliminate this load by inserting a reload in the predecessor // and using PHI construction to get the value in the other predecessors, do // it. - DEBUG(dbgs() << "GVN REMOVING PRE LOAD: " << *LI << '\n'); - DEBUG(if (!NewInsts.empty()) - dbgs() << "INSERTED " << NewInsts.size() << " INSTS: " - << *NewInsts.back() << '\n'); + LLVM_DEBUG(dbgs() << "GVN REMOVING PRE LOAD: " << *LI << '\n'); + LLVM_DEBUG(if (!NewInsts.empty()) dbgs() + << "INSERTED " << NewInsts.size() << " INSTS: " << *NewInsts.back() + << '\n'); // Assign value numbers to the new instructions. for (Instruction *I : NewInsts) { @@ -1263,7 +1262,7 @@ bool GVN::PerformLoadPRE(LoadInst *LI, AvailValInBlkVect &ValuesPerBlock, ValuesPerBlock.push_back(AvailableValueInBlock::get(UnavailablePred, NewLoad)); MD->invalidateCachedPointerInfo(LoadPtr); - DEBUG(dbgs() << "GVN INSERTED " << *NewLoad << '\n'); + LLVM_DEBUG(dbgs() << "GVN INSERTED " << *NewLoad << '\n'); } // Perform PHI construction. @@ -1321,11 +1320,8 @@ bool GVN::processNonLocalLoad(LoadInst *LI) { // clobber in the current block. Reject this early. if (NumDeps == 1 && !Deps[0].getResult().isDef() && !Deps[0].getResult().isClobber()) { - DEBUG( - dbgs() << "GVN: non-local load "; - LI->printAsOperand(dbgs()); - dbgs() << " has unknown dependencies\n"; - ); + LLVM_DEBUG(dbgs() << "GVN: non-local load "; LI->printAsOperand(dbgs()); + dbgs() << " has unknown dependencies\n";); return false; } @@ -1354,7 +1350,7 @@ bool GVN::processNonLocalLoad(LoadInst *LI) { // load, then it is fully redundant and we can use PHI insertion to compute // its value. Insert PHIs and remove the fully redundant value now. if (UnavailableBlocks.empty()) { - DEBUG(dbgs() << "GVN REMOVING NONLOCAL LOAD: " << *LI << '\n'); + LLVM_DEBUG(dbgs() << "GVN REMOVING NONLOCAL LOAD: " << *LI << '\n'); // Perform PHI construction. Value *V = ConstructSSAForLoadSet(LI, ValuesPerBlock, *this); @@ -1507,12 +1503,10 @@ bool GVN::processLoad(LoadInst *L) { // Only handle the local case below if (!Dep.isDef() && !Dep.isClobber()) { // This might be a NonFuncLocal or an Unknown - DEBUG( - // fast print dep, using operator<< on instruction is too slow. - dbgs() << "GVN: load "; - L->printAsOperand(dbgs()); - dbgs() << " has unknown dependence\n"; - ); + LLVM_DEBUG( + // fast print dep, using operator<< on instruction is too slow. + dbgs() << "GVN: load "; L->printAsOperand(dbgs()); + dbgs() << " has unknown dependence\n";); return false; } @@ -1696,8 +1690,8 @@ bool GVN::replaceOperandsWithConsts(Instruction *Instr) const { if (it != ReplaceWithConstMap.end()) { assert(!isa<Constant>(Operand) && "Replacing constants with constants is invalid"); - DEBUG(dbgs() << "GVN replacing: " << *Operand << " with " << *it->second - << " in instruction " << *Instr << '\n'); + LLVM_DEBUG(dbgs() << "GVN replacing: " << *Operand << " with " + << *it->second << " in instruction " << *Instr << '\n'); Instr->setOperand(OpNum, it->second); Changed = true; } @@ -2039,7 +2033,7 @@ bool GVN::runImpl(Function &F, AssumptionCache &RunAC, DominatorTree &RunDT, unsigned Iteration = 0; while (ShouldContinue) { - DEBUG(dbgs() << "GVN iteration: " << Iteration << "\n"); + LLVM_DEBUG(dbgs() << "GVN iteration: " << Iteration << "\n"); ShouldContinue = iterateOnFunction(F); Changed |= ShouldContinue; ++Iteration; @@ -2105,10 +2099,10 @@ bool GVN::processBlock(BasicBlock *BB) { const Instruction *MaybeFirstICF = FirstImplicitControlFlowInsts.lookup(BB); for (auto *I : InstrsToErase) { assert(I->getParent() == BB && "Removing instruction from wrong block?"); - DEBUG(dbgs() << "GVN removed: " << *I << '\n'); + LLVM_DEBUG(dbgs() << "GVN removed: " << *I << '\n'); salvageDebugInfo(*I); if (MD) MD->removeInstruction(I); - DEBUG(verifyRemoved(I)); + LLVM_DEBUG(verifyRemoved(I)); if (MaybeFirstICF == I) { // We have erased the first ICF in block. The map needs to be updated. InvalidateImplicitCF = true; @@ -2290,7 +2284,7 @@ bool GVN::performScalarPRE(Instruction *CurInst) { PREInstr = CurInst->clone(); if (!performScalarPREInsertion(PREInstr, PREPred, CurrentBlock, ValNo)) { // If we failed insertion, make sure we remove the instruction. - DEBUG(verifyRemoved(PREInstr)); + LLVM_DEBUG(verifyRemoved(PREInstr)); PREInstr->deleteValue(); return false; } @@ -2328,10 +2322,10 @@ bool GVN::performScalarPRE(Instruction *CurInst) { VN.erase(CurInst); removeFromLeaderTable(ValNo, CurInst, CurrentBlock); - DEBUG(dbgs() << "GVN PRE removed: " << *CurInst << '\n'); + LLVM_DEBUG(dbgs() << "GVN PRE removed: " << *CurInst << '\n'); if (MD) MD->removeInstruction(CurInst); - DEBUG(verifyRemoved(CurInst)); + LLVM_DEBUG(verifyRemoved(CurInst)); bool InvalidateImplicitCF = FirstImplicitControlFlowInsts.lookup(CurInst->getParent()) == CurInst; // FIXME: Intended to be markInstructionForDeletion(CurInst), but it causes diff --git a/llvm/lib/Transforms/Scalar/GVNHoist.cpp b/llvm/lib/Transforms/Scalar/GVNHoist.cpp index a8b33d19d17..6994d964a9e 100644 --- a/llvm/lib/Transforms/Scalar/GVNHoist.cpp +++ b/llvm/lib/Transforms/Scalar/GVNHoist.cpp @@ -622,7 +622,7 @@ private: // Iterate in reverse order to keep lower ranked values on the top. for (std::pair<VNType, Instruction *> &VI : reverse(it1->second)) { // Get the value of instruction I - DEBUG(dbgs() << "\nPushing on stack: " << *VI.second); + LLVM_DEBUG(dbgs() << "\nPushing on stack: " << *VI.second); RenameStack[VI.first].push_back(VI.second); } } @@ -636,7 +636,7 @@ private: if (P == CHIBBs.end()) { continue; } - DEBUG(dbgs() << "\nLooking at CHIs in: " << Pred->getName();); + LLVM_DEBUG(dbgs() << "\nLooking at CHIs in: " << Pred->getName();); // A CHI is found (BB -> Pred is an edge in the CFG) // Pop the stack until Top(V) = Ve. auto &VCHI = P->second; @@ -651,9 +651,9 @@ private: DT->properlyDominates(Pred, si->second.back()->getParent())) { C.Dest = BB; // Assign the edge C.I = si->second.pop_back_val(); // Assign the argument - DEBUG(dbgs() << "\nCHI Inserted in BB: " << C.Dest->getName() - << *C.I << ", VN: " << C.VN.first << ", " - << C.VN.second); + LLVM_DEBUG(dbgs() + << "\nCHI Inserted in BB: " << C.Dest->getName() << *C.I + << ", VN: " << C.VN.first << ", " << C.VN.second); } // Move to next CHI of a different value It = std::find_if(It, VCHI.end(), @@ -798,8 +798,8 @@ private: // Ignore spurious PDFs. if (DT->properlyDominates(IDFB, V[i]->getParent())) { OutValue[IDFB].push_back(C); - DEBUG(dbgs() << "\nInsertion a CHI for BB: " << IDFB->getName() - << ", for Insn: " << *V[i]); + LLVM_DEBUG(dbgs() << "\nInsertion a CHI for BB: " << IDFB->getName() + << ", for Insn: " << *V[i]); } } } diff --git a/llvm/lib/Transforms/Scalar/GVNSink.cpp b/llvm/lib/Transforms/Scalar/GVNSink.cpp index 4368b68582e..9d8a6d1c552 100644 --- a/llvm/lib/Transforms/Scalar/GVNSink.cpp +++ b/llvm/lib/Transforms/Scalar/GVNSink.cpp @@ -561,7 +561,8 @@ public: GVNSink() = default; bool run(Function &F) { - DEBUG(dbgs() << "GVNSink: running on function @" << F.getName() << "\n"); + LLVM_DEBUG(dbgs() << "GVNSink: running on function @" << F.getName() + << "\n"); unsigned NumSunk = 0; ReversePostOrderTraversal<Function*> RPOT(&F); @@ -629,15 +630,15 @@ Optional<SinkingInstructionCandidate> GVNSink::analyzeInstructionForSinking( LockstepReverseIterator &LRI, unsigned &InstNum, unsigned &MemoryInstNum, ModelledPHISet &NeededPHIs, SmallPtrSetImpl<Value *> &PHIContents) { auto Insts = *LRI; - DEBUG(dbgs() << " -- Analyzing instruction set: [\n"; for (auto *I - : Insts) { + LLVM_DEBUG(dbgs() << " -- Analyzing instruction set: [\n"; for (auto *I + : Insts) { I->dump(); } dbgs() << " ]\n";); DenseMap<uint32_t, unsigned> VNums; for (auto *I : Insts) { uint32_t N = VN.lookupOrAdd(I); - DEBUG(dbgs() << " VN=" << Twine::utohexstr(N) << " for" << *I << "\n"); + LLVM_DEBUG(dbgs() << " VN=" << Twine::utohexstr(N) << " for" << *I << "\n"); if (N == ~0U) return None; VNums[N]++; @@ -749,8 +750,8 @@ Optional<SinkingInstructionCandidate> GVNSink::analyzeInstructionForSinking( } unsigned GVNSink::sinkBB(BasicBlock *BBEnd) { - DEBUG(dbgs() << "GVNSink: running on basic block "; - BBEnd->printAsOperand(dbgs()); dbgs() << "\n"); + LLVM_DEBUG(dbgs() << "GVNSink: running on basic block "; + BBEnd->printAsOperand(dbgs()); dbgs() << "\n"); SmallVector<BasicBlock *, 4> Preds; for (auto *B : predecessors(BBEnd)) { auto *T = B->getTerminator(); @@ -794,23 +795,23 @@ unsigned GVNSink::sinkBB(BasicBlock *BBEnd) { Candidates.begin(), Candidates.end(), [](const SinkingInstructionCandidate &A, const SinkingInstructionCandidate &B) { return A > B; }); - DEBUG(dbgs() << " -- Sinking candidates:\n"; for (auto &C - : Candidates) dbgs() - << " " << C << "\n";); + LLVM_DEBUG(dbgs() << " -- Sinking candidates:\n"; for (auto &C + : Candidates) dbgs() + << " " << C << "\n";); // Pick the top candidate, as long it is positive! if (Candidates.empty() || Candidates.front().Cost <= 0) return 0; auto C = Candidates.front(); - DEBUG(dbgs() << " -- Sinking: " << C << "\n"); + LLVM_DEBUG(dbgs() << " -- Sinking: " << C << "\n"); BasicBlock *InsertBB = BBEnd; if (C.Blocks.size() < NumOrigPreds) { - DEBUG(dbgs() << " -- Splitting edge to "; BBEnd->printAsOperand(dbgs()); - dbgs() << "\n"); + LLVM_DEBUG(dbgs() << " -- Splitting edge to "; + BBEnd->printAsOperand(dbgs()); dbgs() << "\n"); InsertBB = SplitBlockPredecessors(BBEnd, C.Blocks, ".gvnsink.split"); if (!InsertBB) { - DEBUG(dbgs() << " -- FAILED to split edge!\n"); + LLVM_DEBUG(dbgs() << " -- FAILED to split edge!\n"); // Edge couldn't be split. return 0; } diff --git a/llvm/lib/Transforms/Scalar/GuardWidening.cpp b/llvm/lib/Transforms/Scalar/GuardWidening.cpp index 55d2f6ee81c..ad1598d7b8b 100644 --- a/llvm/lib/Transforms/Scalar/GuardWidening.cpp +++ b/llvm/lib/Transforms/Scalar/GuardWidening.cpp @@ -302,9 +302,9 @@ bool GuardWideningImpl::eliminateGuardViaWidening( for (auto *Candidate : make_range(I, E)) { auto Score = computeWideningScore(GuardInst, GuardInstLoop, Candidate, CurLoop); - DEBUG(dbgs() << "Score between " << *GuardInst->getArgOperand(0) - << " and " << *Candidate->getArgOperand(0) << " is " - << scoreTypeToString(Score) << "\n"); + LLVM_DEBUG(dbgs() << "Score between " << *GuardInst->getArgOperand(0) + << " and " << *Candidate->getArgOperand(0) << " is " + << scoreTypeToString(Score) << "\n"); if (Score > BestScoreSoFar) { BestScoreSoFar = Score; BestSoFar = Candidate; @@ -313,15 +313,16 @@ bool GuardWideningImpl::eliminateGuardViaWidening( } if (BestScoreSoFar == WS_IllegalOrNegative) { - DEBUG(dbgs() << "Did not eliminate guard " << *GuardInst << "\n"); + LLVM_DEBUG(dbgs() << "Did not eliminate guard " << *GuardInst << "\n"); return false; } assert(BestSoFar != GuardInst && "Should have never visited same guard!"); assert(DT.dominates(BestSoFar, GuardInst) && "Should be!"); - DEBUG(dbgs() << "Widening " << *GuardInst << " into " << *BestSoFar - << " with score " << scoreTypeToString(BestScoreSoFar) << "\n"); + LLVM_DEBUG(dbgs() << "Widening " << *GuardInst << " into " << *BestSoFar + << " with score " << scoreTypeToString(BestScoreSoFar) + << "\n"); widenGuard(BestSoFar, GuardInst->getArgOperand(0)); GuardInst->setArgOperand(0, ConstantInt::getTrue(GuardInst->getContext())); EliminatedGuards.push_back(GuardInst); diff --git a/llvm/lib/Transforms/Scalar/IndVarSimplify.cpp b/llvm/lib/Transforms/Scalar/IndVarSimplify.cpp index ae0c3eb243e..86b0269e1f7 100644 --- a/llvm/lib/Transforms/Scalar/IndVarSimplify.cpp +++ b/llvm/lib/Transforms/Scalar/IndVarSimplify.cpp @@ -210,8 +210,8 @@ bool IndVarSimplify::isValidRewrite(Value *FromVal, Value *ToVal) { if (FromBase == ToBase) return true; - DEBUG(dbgs() << "INDVARS: GEP rewrite bail out " - << *FromBase << " != " << *ToBase << "\n"); + LLVM_DEBUG(dbgs() << "INDVARS: GEP rewrite bail out " << *FromBase + << " != " << *ToBase << "\n"); return false; } @@ -653,8 +653,9 @@ void IndVarSimplify::rewriteLoopExitValues(Loop *L, SCEVExpander &Rewriter) { Value *ExitVal = expandSCEVIfNeeded(Rewriter, ExitValue, L, Inst, PN->getType()); - DEBUG(dbgs() << "INDVARS: RLEV: AfterLoopVal = " << *ExitVal << '\n' - << " LoopVal = " << *Inst << "\n"); + LLVM_DEBUG(dbgs() << "INDVARS: RLEV: AfterLoopVal = " << *ExitVal + << '\n' + << " LoopVal = " << *Inst << "\n"); if (!isValidRewrite(Inst, ExitVal)) { DeadInsts.push_back(ExitVal); @@ -1084,7 +1085,7 @@ Instruction *WidenIV::cloneBitwiseIVUser(NarrowIVDefUse DU) { Instruction *NarrowDef = DU.NarrowDef; Instruction *WideDef = DU.WideDef; - DEBUG(dbgs() << "Cloning bitwise IVUser: " << *NarrowUse << "\n"); + LLVM_DEBUG(dbgs() << "Cloning bitwise IVUser: " << *NarrowUse << "\n"); // Replace NarrowDef operands with WideDef. Otherwise, we don't know anything // about the narrow operand yet so must insert a [sz]ext. It is probably loop @@ -1115,7 +1116,7 @@ Instruction *WidenIV::cloneArithmeticIVUser(NarrowIVDefUse DU, Instruction *NarrowDef = DU.NarrowDef; Instruction *WideDef = DU.WideDef; - DEBUG(dbgs() << "Cloning arithmetic IVUser: " << *NarrowUse << "\n"); + LLVM_DEBUG(dbgs() << "Cloning arithmetic IVUser: " << *NarrowUse << "\n"); unsigned IVOpIdx = (NarrowUse->getOperand(0) == NarrowDef) ? 0 : 1; @@ -1315,8 +1316,8 @@ WidenIV::WidenedRecTy WidenIV::getWideRecurrence(NarrowIVDefUse DU) { /// This IV user cannot be widen. Replace this use of the original narrow IV /// with a truncation of the new wide IV to isolate and eliminate the narrow IV. static void truncateIVUse(NarrowIVDefUse DU, DominatorTree *DT, LoopInfo *LI) { - DEBUG(dbgs() << "INDVARS: Truncate IV " << *DU.WideDef - << " for user " << *DU.NarrowUse << "\n"); + LLVM_DEBUG(dbgs() << "INDVARS: Truncate IV " << *DU.WideDef << " for user " + << *DU.NarrowUse << "\n"); IRBuilder<> Builder( getInsertPointForUses(DU.NarrowUse, DU.NarrowDef, DT, LI)); Value *Trunc = Builder.CreateTrunc(DU.WideDef, DU.NarrowDef->getType()); @@ -1396,8 +1397,8 @@ Instruction *WidenIV::widenIVUse(NarrowIVDefUse DU, SCEVExpander &Rewriter) { Value *Trunc = Builder.CreateTrunc(WidePhi, DU.NarrowDef->getType()); UsePhi->replaceAllUsesWith(Trunc); DeadInsts.emplace_back(UsePhi); - DEBUG(dbgs() << "INDVARS: Widen lcssa phi " << *UsePhi - << " to " << *WidePhi << "\n"); + LLVM_DEBUG(dbgs() << "INDVARS: Widen lcssa phi " << *UsePhi << " to " + << *WidePhi << "\n"); } return nullptr; } @@ -1428,15 +1429,16 @@ Instruction *WidenIV::widenIVUse(NarrowIVDefUse DU, SCEVExpander &Rewriter) { // A wider extend was hidden behind a narrower one. This may induce // another round of IV widening in which the intermediate IV becomes // dead. It should be very rare. - DEBUG(dbgs() << "INDVARS: New IV " << *WidePhi - << " not wide enough to subsume " << *DU.NarrowUse << "\n"); + LLVM_DEBUG(dbgs() << "INDVARS: New IV " << *WidePhi + << " not wide enough to subsume " << *DU.NarrowUse + << "\n"); DU.NarrowUse->replaceUsesOfWith(DU.NarrowDef, DU.WideDef); NewDef = DU.NarrowUse; } } if (NewDef != DU.NarrowUse) { - DEBUG(dbgs() << "INDVARS: eliminating " << *DU.NarrowUse - << " replaced by " << *DU.WideDef << "\n"); + LLVM_DEBUG(dbgs() << "INDVARS: eliminating " << *DU.NarrowUse + << " replaced by " << *DU.WideDef << "\n"); ++NumElimExt; DU.NarrowUse->replaceAllUsesWith(NewDef); DeadInsts.emplace_back(DU.NarrowUse); @@ -1491,8 +1493,9 @@ Instruction *WidenIV::widenIVUse(NarrowIVDefUse DU, SCEVExpander &Rewriter) { // absolutely guarantee it. Hence the following failsafe check. In rare cases // where it fails, we simply throw away the newly created wide use. if (WideAddRec.first != SE->getSCEV(WideUse)) { - DEBUG(dbgs() << "Wide use expression mismatch: " << *WideUse - << ": " << *SE->getSCEV(WideUse) << " != " << *WideAddRec.first << "\n"); + LLVM_DEBUG(dbgs() << "Wide use expression mismatch: " << *WideUse << ": " + << *SE->getSCEV(WideUse) << " != " << *WideAddRec.first + << "\n"); DeadInsts.emplace_back(WideUse); return nullptr; } @@ -1597,7 +1600,7 @@ PHINode *WidenIV::createWideIV(SCEVExpander &Rewriter) { WideInc->setDebugLoc(OrigInc->getDebugLoc()); } - DEBUG(dbgs() << "Wide IV: " << *WidePhi << "\n"); + LLVM_DEBUG(dbgs() << "Wide IV: " << *WidePhi << "\n"); ++NumWidened; // Traverse the def-use chain using a worklist starting at the original IV. @@ -2231,12 +2234,12 @@ linearFunctionTestReplace(Loop *L, else P = ICmpInst::ICMP_EQ; - DEBUG(dbgs() << "INDVARS: Rewriting loop exit condition to:\n" - << " LHS:" << *CmpIndVar << '\n' - << " op:\t" - << (P == ICmpInst::ICMP_NE ? "!=" : "==") << "\n" - << " RHS:\t" << *ExitCnt << "\n" - << " IVCount:\t" << *IVCount << "\n"); + LLVM_DEBUG(dbgs() << "INDVARS: Rewriting loop exit condition to:\n" + << " LHS:" << *CmpIndVar << '\n' + << " op:\t" << (P == ICmpInst::ICMP_NE ? "!=" : "==") + << "\n" + << " RHS:\t" << *ExitCnt << "\n" + << " IVCount:\t" << *IVCount << "\n"); IRBuilder<> Builder(BI); @@ -2272,7 +2275,7 @@ linearFunctionTestReplace(Loop *L, NewLimit = Start + Count; ExitCnt = ConstantInt::get(CmpIndVar->getType(), NewLimit); - DEBUG(dbgs() << " Widen RHS:\t" << *ExitCnt << "\n"); + LLVM_DEBUG(dbgs() << " Widen RHS:\t" << *ExitCnt << "\n"); } else { // We try to extend trip count first. If that doesn't work we truncate IV. // Zext(trunc(IV)) == IV implies equivalence of the following two: diff --git a/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp b/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp index 79004ddba18..92dc21456e5 100644 --- a/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp +++ b/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp @@ -716,12 +716,13 @@ static bool isSafeDecreasingBound(const SCEV *Start, assert(SE.isKnownNegative(Step) && "expecting negative step"); - DEBUG(dbgs() << "irce: isSafeDecreasingBound with:\n"); - DEBUG(dbgs() << "irce: Start: " << *Start << "\n"); - DEBUG(dbgs() << "irce: Step: " << *Step << "\n"); - DEBUG(dbgs() << "irce: BoundSCEV: " << *BoundSCEV << "\n"); - DEBUG(dbgs() << "irce: Pred: " << ICmpInst::getPredicateName(Pred) << "\n"); - DEBUG(dbgs() << "irce: LatchExitBrIdx: " << LatchBrExitIdx << "\n"); + LLVM_DEBUG(dbgs() << "irce: isSafeDecreasingBound with:\n"); + LLVM_DEBUG(dbgs() << "irce: Start: " << *Start << "\n"); + LLVM_DEBUG(dbgs() << "irce: Step: " << *Step << "\n"); + LLVM_DEBUG(dbgs() << "irce: BoundSCEV: " << *BoundSCEV << "\n"); + LLVM_DEBUG(dbgs() << "irce: Pred: " << ICmpInst::getPredicateName(Pred) + << "\n"); + LLVM_DEBUG(dbgs() << "irce: LatchExitBrIdx: " << LatchBrExitIdx << "\n"); bool IsSigned = ICmpInst::isSigned(Pred); // The predicate that we need to check that the induction variable lies @@ -763,12 +764,13 @@ static bool isSafeIncreasingBound(const SCEV *Start, if (!SE.isAvailableAtLoopEntry(BoundSCEV, L)) return false; - DEBUG(dbgs() << "irce: isSafeIncreasingBound with:\n"); - DEBUG(dbgs() << "irce: Start: " << *Start << "\n"); - DEBUG(dbgs() << "irce: Step: " << *Step << "\n"); - DEBUG(dbgs() << "irce: BoundSCEV: " << *BoundSCEV << "\n"); - DEBUG(dbgs() << "irce: Pred: " << ICmpInst::getPredicateName(Pred) << "\n"); - DEBUG(dbgs() << "irce: LatchExitBrIdx: " << LatchBrExitIdx << "\n"); + LLVM_DEBUG(dbgs() << "irce: isSafeIncreasingBound with:\n"); + LLVM_DEBUG(dbgs() << "irce: Start: " << *Start << "\n"); + LLVM_DEBUG(dbgs() << "irce: Step: " << *Step << "\n"); + LLVM_DEBUG(dbgs() << "irce: BoundSCEV: " << *BoundSCEV << "\n"); + LLVM_DEBUG(dbgs() << "irce: Pred: " << ICmpInst::getPredicateName(Pred) + << "\n"); + LLVM_DEBUG(dbgs() << "irce: LatchExitBrIdx: " << LatchBrExitIdx << "\n"); bool IsSigned = ICmpInst::isSigned(Pred); // The predicate that we need to check that the induction variable lies @@ -1473,7 +1475,7 @@ bool LoopConstrainer::run() { bool IsSignedPredicate = MainLoopStructure.IsSignedPredicate; Optional<SubRanges> MaybeSR = calculateSubRanges(IsSignedPredicate); if (!MaybeSR.hasValue()) { - DEBUG(dbgs() << "irce: could not compute subranges\n"); + LLVM_DEBUG(dbgs() << "irce: could not compute subranges\n"); return false; } @@ -1509,17 +1511,18 @@ bool LoopConstrainer::run() { IsSignedPredicate)) ExitPreLoopAtSCEV = SE.getAddExpr(*SR.HighLimit, MinusOneS); else { - DEBUG(dbgs() << "irce: could not prove no-overflow when computing " - << "preloop exit limit. HighLimit = " << *(*SR.HighLimit) - << "\n"); + LLVM_DEBUG(dbgs() << "irce: could not prove no-overflow when computing " + << "preloop exit limit. HighLimit = " + << *(*SR.HighLimit) << "\n"); return false; } } if (!isSafeToExpandAt(ExitPreLoopAtSCEV, InsertPt, SE)) { - DEBUG(dbgs() << "irce: could not prove that it is safe to expand the" - << " preloop exit limit " << *ExitPreLoopAtSCEV - << " at block " << InsertPt->getParent()->getName() << "\n"); + LLVM_DEBUG(dbgs() << "irce: could not prove that it is safe to expand the" + << " preloop exit limit " << *ExitPreLoopAtSCEV + << " at block " << InsertPt->getParent()->getName() + << "\n"); return false; } @@ -1537,17 +1540,18 @@ bool LoopConstrainer::run() { IsSignedPredicate)) ExitMainLoopAtSCEV = SE.getAddExpr(*SR.LowLimit, MinusOneS); else { - DEBUG(dbgs() << "irce: could not prove no-overflow when computing " - << "mainloop exit limit. LowLimit = " << *(*SR.LowLimit) - << "\n"); + LLVM_DEBUG(dbgs() << "irce: could not prove no-overflow when computing " + << "mainloop exit limit. LowLimit = " + << *(*SR.LowLimit) << "\n"); return false; } } if (!isSafeToExpandAt(ExitMainLoopAtSCEV, InsertPt, SE)) { - DEBUG(dbgs() << "irce: could not prove that it is safe to expand the" - << " main loop exit limit " << *ExitMainLoopAtSCEV - << " at block " << InsertPt->getParent()->getName() << "\n"); + LLVM_DEBUG(dbgs() << "irce: could not prove that it is safe to expand the" + << " main loop exit limit " << *ExitMainLoopAtSCEV + << " at block " << InsertPt->getParent()->getName() + << "\n"); return false; } @@ -1826,13 +1830,13 @@ bool IRCELegacyPass::runOnLoop(Loop *L, LPPassManager &LPM) { bool InductiveRangeCheckElimination::run( Loop *L, function_ref<void(Loop *, bool)> LPMAddNewLoop) { if (L->getBlocks().size() >= LoopSizeCutoff) { - DEBUG(dbgs() << "irce: giving up constraining loop, too large\n"); + LLVM_DEBUG(dbgs() << "irce: giving up constraining loop, too large\n"); return false; } BasicBlock *Preheader = L->getLoopPreheader(); if (!Preheader) { - DEBUG(dbgs() << "irce: loop has no preheader, leaving\n"); + LLVM_DEBUG(dbgs() << "irce: loop has no preheader, leaving\n"); return false; } @@ -1855,7 +1859,7 @@ bool InductiveRangeCheckElimination::run( IRC.print(OS); }; - DEBUG(PrintRecognizedRangeChecks(dbgs())); + LLVM_DEBUG(PrintRecognizedRangeChecks(dbgs())); if (PrintRangeChecks) PrintRecognizedRangeChecks(errs()); @@ -1864,8 +1868,8 @@ bool InductiveRangeCheckElimination::run( Optional<LoopStructure> MaybeLoopStructure = LoopStructure::parseLoopStructure(SE, BPI, *L, FailureReason); if (!MaybeLoopStructure.hasValue()) { - DEBUG(dbgs() << "irce: could not parse loop structure: " << FailureReason - << "\n";); + LLVM_DEBUG(dbgs() << "irce: could not parse loop structure: " + << FailureReason << "\n";); return false; } LoopStructure LS = MaybeLoopStructure.getValue(); @@ -1915,7 +1919,7 @@ bool InductiveRangeCheckElimination::run( L->print(dbgs()); }; - DEBUG(PrintConstrainedLoopInfo()); + LLVM_DEBUG(PrintConstrainedLoopInfo()); if (PrintChangedLoops) PrintConstrainedLoopInfo(); diff --git a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp index 454ea254b88..e6fea56f264 100644 --- a/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp +++ b/llvm/lib/Transforms/Scalar/InferAddressSpaces.cpp @@ -653,13 +653,13 @@ void InferAddressSpaces::inferAddressSpaces( // Tries to update the address space of the stack top according to the // address spaces of its operands. - DEBUG(dbgs() << "Updating the address space of\n " << *V << '\n'); + LLVM_DEBUG(dbgs() << "Updating the address space of\n " << *V << '\n'); Optional<unsigned> NewAS = updateAddressSpace(*V, *InferredAddrSpace); if (!NewAS.hasValue()) continue; // If any updates are made, grabs its users to the worklist because // their address spaces can also be possibly updated. - DEBUG(dbgs() << " to " << NewAS.getValue() << '\n'); + LLVM_DEBUG(dbgs() << " to " << NewAS.getValue() << '\n'); (*InferredAddrSpace)[V] = NewAS.getValue(); for (Value *User : V->users()) { @@ -901,15 +901,15 @@ bool InferAddressSpaces::rewriteWithNewAddressSpaces( if (NewV == nullptr) continue; - DEBUG(dbgs() << "Replacing the uses of " << *V - << "\n with\n " << *NewV << '\n'); + LLVM_DEBUG(dbgs() << "Replacing the uses of " << *V << "\n with\n " + << *NewV << '\n'); if (Constant *C = dyn_cast<Constant>(V)) { Constant *Replace = ConstantExpr::getAddrSpaceCast(cast<Constant>(NewV), C->getType()); if (C != Replace) { - DEBUG(dbgs() << "Inserting replacement const cast: " - << Replace << ": " << *Replace << '\n'); + LLVM_DEBUG(dbgs() << "Inserting replacement const cast: " << Replace + << ": " << *Replace << '\n'); C->replaceAllUsesWith(Replace); V = Replace; } diff --git a/llvm/lib/Transforms/Scalar/JumpThreading.cpp b/llvm/lib/Transforms/Scalar/JumpThreading.cpp index 5f379253ea5..cf9919ebf9d 100644 --- a/llvm/lib/Transforms/Scalar/JumpThreading.cpp +++ b/llvm/lib/Transforms/Scalar/JumpThreading.cpp @@ -340,7 +340,7 @@ bool JumpThreadingPass::runImpl(Function &F, TargetLibraryInfo *TLI_, DeferredDominance *DDT_, bool HasProfileData_, std::unique_ptr<BlockFrequencyInfo> BFI_, std::unique_ptr<BranchProbabilityInfo> BPI_) { - DEBUG(dbgs() << "Jump threading on function '" << F.getName() << "'\n"); + LLVM_DEBUG(dbgs() << "Jump threading on function '" << F.getName() << "'\n"); TLI = TLI_; LVI = LVI_; AA = AA_; @@ -386,8 +386,9 @@ bool JumpThreadingPass::runImpl(Function &F, TargetLibraryInfo *TLI_, if (pred_empty(&BB)) { // When ProcessBlock makes BB unreachable it doesn't bother to fix up // the instructions in it. We must remove BB to prevent invalid IR. - DEBUG(dbgs() << " JT: Deleting dead block '" << BB.getName() - << "' with terminator: " << *BB.getTerminator() << '\n'); + LLVM_DEBUG(dbgs() << " JT: Deleting dead block '" << BB.getName() + << "' with terminator: " << *BB.getTerminator() + << '\n'); LoopHeaders.erase(&BB); LVI->eraseBlock(&BB); DeleteDeadBlock(&BB, DDT); @@ -1084,8 +1085,8 @@ bool JumpThreadingPass::ProcessBlock(BasicBlock *BB) { Updates.push_back({DominatorTree::Delete, BB, Succ}); } - DEBUG(dbgs() << " In block '" << BB->getName() - << "' folding undef terminator: " << *BBTerm << '\n'); + LLVM_DEBUG(dbgs() << " In block '" << BB->getName() + << "' folding undef terminator: " << *BBTerm << '\n'); BranchInst::Create(BBTerm->getSuccessor(BestSucc), BBTerm); BBTerm->eraseFromParent(); DDT->applyUpdates(Updates); @@ -1096,8 +1097,9 @@ bool JumpThreadingPass::ProcessBlock(BasicBlock *BB) { // terminator to an unconditional branch. This can occur due to threading in // other blocks. if (getKnownConstant(Condition, Preference)) { - DEBUG(dbgs() << " In block '" << BB->getName() - << "' folding terminator: " << *BB->getTerminator() << '\n'); + LLVM_DEBUG(dbgs() << " In block '" << BB->getName() + << "' folding terminator: " << *BB->getTerminator() + << '\n'); ++NumFolds; ConstantFoldTerminator(BB, true, nullptr, DDT); return true; @@ -1574,12 +1576,12 @@ bool JumpThreadingPass::ProcessThreadableEdges(Value *Cond, BasicBlock *BB, assert(!PredValues.empty() && "ComputeValueKnownInPredecessors returned true with no values"); - DEBUG(dbgs() << "IN BB: " << *BB; - for (const auto &PredValue : PredValues) { - dbgs() << " BB '" << BB->getName() << "': FOUND condition = " - << *PredValue.first - << " for pred '" << PredValue.second->getName() << "'.\n"; - }); + LLVM_DEBUG(dbgs() << "IN BB: " << *BB; for (const auto &PredValue + : PredValues) { + dbgs() << " BB '" << BB->getName() + << "': FOUND condition = " << *PredValue.first << " for pred '" + << PredValue.second->getName() << "'.\n"; + }); // Decide what we want to thread through. Convert our list of known values to // a list of known destinations for each pred. This also discards duplicate @@ -1901,15 +1903,15 @@ bool JumpThreadingPass::ThreadEdge(BasicBlock *BB, BasicBlock *SuccBB) { // If threading to the same block as we come from, we would infinite loop. if (SuccBB == BB) { - DEBUG(dbgs() << " Not threading across BB '" << BB->getName() - << "' - would thread to self!\n"); + LLVM_DEBUG(dbgs() << " Not threading across BB '" << BB->getName() + << "' - would thread to self!\n"); return false; } // If threading this would thread across a loop header, don't thread the edge. // See the comments above FindLoopHeaders for justifications and caveats. if (LoopHeaders.count(BB) || LoopHeaders.count(SuccBB)) { - DEBUG({ + LLVM_DEBUG({ bool BBIsHeader = LoopHeaders.count(BB); bool SuccIsHeader = LoopHeaders.count(SuccBB); dbgs() << " Not threading across " @@ -1923,8 +1925,8 @@ bool JumpThreadingPass::ThreadEdge(BasicBlock *BB, unsigned JumpThreadCost = getJumpThreadDuplicationCost(BB, BB->getTerminator(), BBDupThreshold); if (JumpThreadCost > BBDupThreshold) { - DEBUG(dbgs() << " Not threading BB '" << BB->getName() - << "' - Cost is too high: " << JumpThreadCost << "\n"); + LLVM_DEBUG(dbgs() << " Not threading BB '" << BB->getName() + << "' - Cost is too high: " << JumpThreadCost << "\n"); return false; } @@ -1933,16 +1935,16 @@ bool JumpThreadingPass::ThreadEdge(BasicBlock *BB, if (PredBBs.size() == 1) PredBB = PredBBs[0]; else { - DEBUG(dbgs() << " Factoring out " << PredBBs.size() - << " common predecessors.\n"); + LLVM_DEBUG(dbgs() << " Factoring out " << PredBBs.size() + << " common predecessors.\n"); PredBB = SplitBlockPreds(BB, PredBBs, ".thr_comm"); } // And finally, do it! - DEBUG(dbgs() << " Threading edge from '" << PredBB->getName() << "' to '" - << SuccBB->getName() << "' with cost: " << JumpThreadCost - << ", across block:\n " - << *BB << "\n"); + LLVM_DEBUG(dbgs() << " Threading edge from '" << PredBB->getName() + << "' to '" << SuccBB->getName() + << "' with cost: " << JumpThreadCost + << ", across block:\n " << *BB << "\n"); if (DDT->pending()) LVI->disableDT(); @@ -2235,17 +2237,17 @@ bool JumpThreadingPass::DuplicateCondBranchOnPHIIntoPred( // cause us to transform this into an irreducible loop, don't do this. // See the comments above FindLoopHeaders for justifications and caveats. if (LoopHeaders.count(BB)) { - DEBUG(dbgs() << " Not duplicating loop header '" << BB->getName() - << "' into predecessor block '" << PredBBs[0]->getName() - << "' - it might create an irreducible loop!\n"); + LLVM_DEBUG(dbgs() << " Not duplicating loop header '" << BB->getName() + << "' into predecessor block '" << PredBBs[0]->getName() + << "' - it might create an irreducible loop!\n"); return false; } unsigned DuplicationCost = getJumpThreadDuplicationCost(BB, BB->getTerminator(), BBDupThreshold); if (DuplicationCost > BBDupThreshold) { - DEBUG(dbgs() << " Not duplicating BB '" << BB->getName() - << "' - Cost is too high: " << DuplicationCost << "\n"); + LLVM_DEBUG(dbgs() << " Not duplicating BB '" << BB->getName() + << "' - Cost is too high: " << DuplicationCost << "\n"); return false; } @@ -2255,17 +2257,18 @@ bool JumpThreadingPass::DuplicateCondBranchOnPHIIntoPred( if (PredBBs.size() == 1) PredBB = PredBBs[0]; else { - DEBUG(dbgs() << " Factoring out " << PredBBs.size() - << " common predecessors.\n"); + LLVM_DEBUG(dbgs() << " Factoring out " << PredBBs.size() + << " common predecessors.\n"); PredBB = SplitBlockPreds(BB, PredBBs, ".thr_comm"); } Updates.push_back({DominatorTree::Delete, PredBB, BB}); // Okay, we decided to do this! Clone all the instructions in BB onto the end // of PredBB. - DEBUG(dbgs() << " Duplicating block '" << BB->getName() << "' into end of '" - << PredBB->getName() << "' to eliminate branch on phi. Cost: " - << DuplicationCost << " block is:" << *BB << "\n"); + LLVM_DEBUG(dbgs() << " Duplicating block '" << BB->getName() + << "' into end of '" << PredBB->getName() + << "' to eliminate branch on phi. Cost: " + << DuplicationCost << " block is:" << *BB << "\n"); // Unless PredBB ends with an unconditional branch, split the edge so that we // can just clone the bits from BB into the end of the new PredBB. @@ -2357,7 +2360,7 @@ bool JumpThreadingPass::DuplicateCondBranchOnPHIIntoPred( if (UsesToRename.empty()) continue; - DEBUG(dbgs() << "JT: Renaming non-local uses of: " << I << "\n"); + LLVM_DEBUG(dbgs() << "JT: Renaming non-local uses of: " << I << "\n"); // We found a use of I outside of BB. Rename all uses of I that are outside // its block to be uses of the appropriate PHI node etc. See ValuesInBlocks @@ -2368,7 +2371,7 @@ bool JumpThreadingPass::DuplicateCondBranchOnPHIIntoPred( while (!UsesToRename.empty()) SSAUpdate.RewriteUse(*UsesToRename.pop_back_val()); - DEBUG(dbgs() << "\n"); + LLVM_DEBUG(dbgs() << "\n"); } // PredBB no longer jumps to BB, remove entries in the PHI node for the edge @@ -2658,8 +2661,8 @@ bool JumpThreadingPass::ThreadGuard(BasicBlock *BB, IntrinsicInst *Guard, BasicBlock *UnguardedBlock = DuplicateInstructionsInSplitBetween( BB, PredUnguardedBlock, Guard, UnguardedMapping); assert(UnguardedBlock && "Could not create the unguarded block?"); - DEBUG(dbgs() << "Moved guard " << *Guard << " to block " - << GuardedBlock->getName() << "\n"); + LLVM_DEBUG(dbgs() << "Moved guard " << *Guard << " to block " + << GuardedBlock->getName() << "\n"); // DuplicateInstructionsInSplitBetween inserts a new block "BB.split" between // PredBB and BB. We need to perform two inserts and one delete for each of // the above calls to update Dominators. diff --git a/llvm/lib/Transforms/Scalar/LICM.cpp b/llvm/lib/Transforms/Scalar/LICM.cpp index b9d10d9f4c8..8c0747782d7 100644 --- a/llvm/lib/Transforms/Scalar/LICM.cpp +++ b/llvm/lib/Transforms/Scalar/LICM.cpp @@ -392,7 +392,7 @@ bool llvm::sinkRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI, // If the instruction is dead, we would try to sink it because it isn't // used in the loop, instead, just delete it. if (isInstructionTriviallyDead(&I, TLI)) { - DEBUG(dbgs() << "LICM deleting dead inst: " << I << '\n'); + LLVM_DEBUG(dbgs() << "LICM deleting dead inst: " << I << '\n'); salvageDebugInfo(I); ++II; CurAST->deleteValue(&I); @@ -461,7 +461,8 @@ bool llvm::hoistRegion(DomTreeNode *N, AliasAnalysis *AA, LoopInfo *LI, // just fold it. if (Constant *C = ConstantFoldInstruction( &I, I.getModule()->getDataLayout(), TLI)) { - DEBUG(dbgs() << "LICM folding inst: " << I << " --> " << *C << '\n'); + LLVM_DEBUG(dbgs() << "LICM folding inst: " << I << " --> " << *C + << '\n'); CurAST->copyValue(&I, C); I.replaceAllUsesWith(C); if (isInstructionTriviallyDead(&I, TLI)) { @@ -927,7 +928,7 @@ static void splitPredecessorsOfLoopExit(PHINode *PN, DominatorTree *DT, static bool sink(Instruction &I, LoopInfo *LI, DominatorTree *DT, const Loop *CurLoop, LoopSafetyInfo *SafetyInfo, OptimizationRemarkEmitter *ORE, bool FreeInLoop) { - DEBUG(dbgs() << "LICM sinking instruction: " << I << "\n"); + LLVM_DEBUG(dbgs() << "LICM sinking instruction: " << I << "\n"); ORE->emit([&]() { return OptimizationRemark(DEBUG_TYPE, "InstSunk", &I) << "sinking " << ore::NV("Inst", &I); @@ -1029,8 +1030,8 @@ static bool hoist(Instruction &I, const DominatorTree *DT, const Loop *CurLoop, const LoopSafetyInfo *SafetyInfo, OptimizationRemarkEmitter *ORE) { auto *Preheader = CurLoop->getLoopPreheader(); - DEBUG(dbgs() << "LICM hoisting to " << Preheader->getName() << ": " << I - << "\n"); + LLVM_DEBUG(dbgs() << "LICM hoisting to " << Preheader->getName() << ": " << I + << "\n"); ORE->emit([&]() { return OptimizationRemark(DEBUG_TYPE, "Hoisted", &I) << "hoisting " << ore::NV("Inst", &I); @@ -1410,8 +1411,8 @@ bool llvm::promoteLoopAccessesToScalars( return false; // Otherwise, this is safe to promote, lets do it! - DEBUG(dbgs() << "LICM: Promoting value stored to in loop: " << *SomePtr - << '\n'); + LLVM_DEBUG(dbgs() << "LICM: Promoting value stored to in loop: " << *SomePtr + << '\n'); ORE->emit([&]() { return OptimizationRemark(DEBUG_TYPE, "PromoteLoopAccessesToScalar", LoopUses[0]) diff --git a/llvm/lib/Transforms/Scalar/LoopDataPrefetch.cpp b/llvm/lib/Transforms/Scalar/LoopDataPrefetch.cpp index c804115f415..3b41b5d96c8 100644 --- a/llvm/lib/Transforms/Scalar/LoopDataPrefetch.cpp +++ b/llvm/lib/Transforms/Scalar/LoopDataPrefetch.cpp @@ -244,9 +244,9 @@ bool LoopDataPrefetch::runOnLoop(Loop *L) { if (ItersAhead > getMaxPrefetchIterationsAhead()) return MadeChange; - DEBUG(dbgs() << "Prefetching " << ItersAhead - << " iterations ahead (loop size: " << LoopSize << ") in " - << L->getHeader()->getParent()->getName() << ": " << *L); + LLVM_DEBUG(dbgs() << "Prefetching " << ItersAhead + << " iterations ahead (loop size: " << LoopSize << ") in " + << L->getHeader()->getParent()->getName() << ": " << *L); SmallVector<std::pair<Instruction *, const SCEVAddRecExpr *>, 16> PrefLoads; for (const auto BB : L->blocks()) { @@ -320,8 +320,8 @@ bool LoopDataPrefetch::runOnLoop(Loop *L) { ConstantInt::get(I32, MemI->mayReadFromMemory() ? 0 : 1), ConstantInt::get(I32, 3), ConstantInt::get(I32, 1)}); ++NumPrefetches; - DEBUG(dbgs() << " Access: " << *PtrValue << ", SCEV: " << *LSCEV - << "\n"); + LLVM_DEBUG(dbgs() << " Access: " << *PtrValue << ", SCEV: " << *LSCEV + << "\n"); ORE->emit([&]() { return OptimizationRemark(DEBUG_TYPE, "Prefetched", MemI) << "prefetched memory access"; diff --git a/llvm/lib/Transforms/Scalar/LoopDeletion.cpp b/llvm/lib/Transforms/Scalar/LoopDeletion.cpp index 15cd1086f20..d412025d7e9 100644 --- a/llvm/lib/Transforms/Scalar/LoopDeletion.cpp +++ b/llvm/lib/Transforms/Scalar/LoopDeletion.cpp @@ -142,14 +142,15 @@ static LoopDeletionResult deleteLoopIfDead(Loop *L, DominatorTree &DT, // of trouble. BasicBlock *Preheader = L->getLoopPreheader(); if (!Preheader || !L->hasDedicatedExits()) { - DEBUG(dbgs() - << "Deletion requires Loop with preheader and dedicated exits.\n"); + LLVM_DEBUG( + dbgs() + << "Deletion requires Loop with preheader and dedicated exits.\n"); return LoopDeletionResult::Unmodified; } // We can't remove loops that contain subloops. If the subloops were dead, // they would already have been removed in earlier executions of this pass. if (L->begin() != L->end()) { - DEBUG(dbgs() << "Loop contains subloops.\n"); + LLVM_DEBUG(dbgs() << "Loop contains subloops.\n"); return LoopDeletionResult::Unmodified; } @@ -157,7 +158,7 @@ static LoopDeletionResult deleteLoopIfDead(Loop *L, DominatorTree &DT, BasicBlock *ExitBlock = L->getUniqueExitBlock(); if (ExitBlock && isLoopNeverExecuted(L)) { - DEBUG(dbgs() << "Loop is proven to never execute, delete it!"); + LLVM_DEBUG(dbgs() << "Loop is proven to never execute, delete it!"); // Set incoming value to undef for phi nodes in the exit block. for (PHINode &P : ExitBlock->phis()) { std::fill(P.incoming_values().begin(), P.incoming_values().end(), @@ -178,13 +179,13 @@ static LoopDeletionResult deleteLoopIfDead(Loop *L, DominatorTree &DT, // block will be branched to, or trying to preserve the branching logic in // a loop invariant manner. if (!ExitBlock) { - DEBUG(dbgs() << "Deletion requires single exit block\n"); + LLVM_DEBUG(dbgs() << "Deletion requires single exit block\n"); return LoopDeletionResult::Unmodified; } // Finally, we have to check that the loop really is dead. bool Changed = false; if (!isLoopDead(L, SE, ExitingBlocks, ExitBlock, Changed, Preheader)) { - DEBUG(dbgs() << "Loop is not invariant, cannot delete.\n"); + LLVM_DEBUG(dbgs() << "Loop is not invariant, cannot delete.\n"); return Changed ? LoopDeletionResult::Modified : LoopDeletionResult::Unmodified; } @@ -193,12 +194,12 @@ static LoopDeletionResult deleteLoopIfDead(Loop *L, DominatorTree &DT, // They could be infinite, in which case we'd be changing program behavior. const SCEV *S = SE.getMaxBackedgeTakenCount(L); if (isa<SCEVCouldNotCompute>(S)) { - DEBUG(dbgs() << "Could not compute SCEV MaxBackedgeTakenCount.\n"); + LLVM_DEBUG(dbgs() << "Could not compute SCEV MaxBackedgeTakenCount.\n"); return Changed ? LoopDeletionResult::Modified : LoopDeletionResult::Unmodified; } - DEBUG(dbgs() << "Loop is invariant, delete it!"); + LLVM_DEBUG(dbgs() << "Loop is invariant, delete it!"); deleteDeadLoop(L, &DT, &SE, &LI); ++NumDeleted; @@ -209,8 +210,8 @@ PreservedAnalyses LoopDeletionPass::run(Loop &L, LoopAnalysisManager &AM, LoopStandardAnalysisResults &AR, LPMUpdater &Updater) { - DEBUG(dbgs() << "Analyzing Loop for deletion: "); - DEBUG(L.dump()); + LLVM_DEBUG(dbgs() << "Analyzing Loop for deletion: "); + LLVM_DEBUG(L.dump()); std::string LoopName = L.getName(); auto Result = deleteLoopIfDead(&L, AR.DT, AR.SE, AR.LI); if (Result == LoopDeletionResult::Unmodified) @@ -255,8 +256,8 @@ bool LoopDeletionLegacyPass::runOnLoop(Loop *L, LPPassManager &LPM) { ScalarEvolution &SE = getAnalysis<ScalarEvolutionWrapperPass>().getSE(); LoopInfo &LI = getAnalysis<LoopInfoWrapperPass>().getLoopInfo(); - DEBUG(dbgs() << "Analyzing Loop for deletion: "); - DEBUG(L->dump()); + LLVM_DEBUG(dbgs() << "Analyzing Loop for deletion: "); + LLVM_DEBUG(L->dump()); LoopDeletionResult Result = deleteLoopIfDead(L, DT, SE, LI); diff --git a/llvm/lib/Transforms/Scalar/LoopDistribute.cpp b/llvm/lib/Transforms/Scalar/LoopDistribute.cpp index a4da0940e33..06083a4f508 100644 --- a/llvm/lib/Transforms/Scalar/LoopDistribute.cpp +++ b/llvm/lib/Transforms/Scalar/LoopDistribute.cpp @@ -362,9 +362,11 @@ public: std::tie(LoadToPart, NewElt) = LoadToPartition.insert(std::make_pair(Inst, PartI)); if (!NewElt) { - DEBUG(dbgs() << "Merging partitions due to this load in multiple " - << "partitions: " << PartI << ", " - << LoadToPart->second << "\n" << *Inst << "\n"); + LLVM_DEBUG(dbgs() + << "Merging partitions due to this load in multiple " + << "partitions: " << PartI << ", " << LoadToPart->second + << "\n" + << *Inst << "\n"); auto PartJ = I; do { @@ -602,7 +604,7 @@ public: const SmallVectorImpl<Dependence> &Dependences) { Accesses.append(Instructions.begin(), Instructions.end()); - DEBUG(dbgs() << "Backward dependences:\n"); + LLVM_DEBUG(dbgs() << "Backward dependences:\n"); for (auto &Dep : Dependences) if (Dep.isPossiblyBackward()) { // Note that the designations source and destination follow the program @@ -611,7 +613,7 @@ public: ++Accesses[Dep.Source].NumUnsafeDependencesStartOrEnd; --Accesses[Dep.Destination].NumUnsafeDependencesStartOrEnd; - DEBUG(Dep.print(dbgs(), 2, Instructions)); + LLVM_DEBUG(Dep.print(dbgs(), 2, Instructions)); } } @@ -632,8 +634,9 @@ public: bool processLoop(std::function<const LoopAccessInfo &(Loop &)> &GetLAA) { assert(L->empty() && "Only process inner loops."); - DEBUG(dbgs() << "\nLDist: In \"" << L->getHeader()->getParent()->getName() - << "\" checking " << *L << "\n"); + LLVM_DEBUG(dbgs() << "\nLDist: In \"" + << L->getHeader()->getParent()->getName() + << "\" checking " << *L << "\n"); if (!L->getExitBlock()) return fail("MultipleExitBlocks", "multiple exit blocks"); @@ -705,7 +708,7 @@ public: for (auto *Inst : DefsUsedOutside) Partitions.addToNewNonCyclicPartition(Inst); - DEBUG(dbgs() << "Seeded partitions:\n" << Partitions); + LLVM_DEBUG(dbgs() << "Seeded partitions:\n" << Partitions); if (Partitions.getSize() < 2) return fail("CantIsolateUnsafeDeps", "cannot isolate unsafe dependencies"); @@ -713,20 +716,20 @@ public: // Run the merge heuristics: Merge non-cyclic adjacent partitions since we // should be able to vectorize these together. Partitions.mergeBeforePopulating(); - DEBUG(dbgs() << "\nMerged partitions:\n" << Partitions); + LLVM_DEBUG(dbgs() << "\nMerged partitions:\n" << Partitions); if (Partitions.getSize() < 2) return fail("CantIsolateUnsafeDeps", "cannot isolate unsafe dependencies"); // Now, populate the partitions with non-memory operations. Partitions.populateUsedSet(); - DEBUG(dbgs() << "\nPopulated partitions:\n" << Partitions); + LLVM_DEBUG(dbgs() << "\nPopulated partitions:\n" << Partitions); // In order to preserve original lexical order for loads, keep them in the // partition that we set up in the MemoryInstructionDependences loop. if (Partitions.mergeToAvoidDuplicatedLoads()) { - DEBUG(dbgs() << "\nPartitions merged to ensure unique loads:\n" - << Partitions); + LLVM_DEBUG(dbgs() << "\nPartitions merged to ensure unique loads:\n" + << Partitions); if (Partitions.getSize() < 2) return fail("CantIsolateUnsafeDeps", "cannot isolate unsafe dependencies"); @@ -740,7 +743,7 @@ public: return fail("TooManySCEVRuntimeChecks", "too many SCEV run-time checks needed.\n"); - DEBUG(dbgs() << "\nDistributing loop: " << *L << "\n"); + LLVM_DEBUG(dbgs() << "\nDistributing loop: " << *L << "\n"); // We're done forming the partitions set up the reverse mapping from // instructions to partitions. Partitions.setupPartitionIdOnInstructions(); @@ -759,8 +762,8 @@ public: RtPtrChecking); if (!Pred.isAlwaysTrue() || !Checks.empty()) { - DEBUG(dbgs() << "\nPointers:\n"); - DEBUG(LAI->getRuntimePointerChecking()->printChecks(dbgs(), Checks)); + LLVM_DEBUG(dbgs() << "\nPointers:\n"); + LLVM_DEBUG(LAI->getRuntimePointerChecking()->printChecks(dbgs(), Checks)); LoopVersioning LVer(*LAI, L, LI, DT, SE, false); LVer.setAliasChecks(std::move(Checks)); LVer.setSCEVChecks(LAI->getPSE().getUnionPredicate()); @@ -775,8 +778,8 @@ public: // Now, we remove the instruction from each loop that don't belong to that // partition. Partitions.removeUnusedInsts(); - DEBUG(dbgs() << "\nAfter removing unused Instrs:\n"); - DEBUG(Partitions.printBlocks()); + LLVM_DEBUG(dbgs() << "\nAfter removing unused Instrs:\n"); + LLVM_DEBUG(Partitions.printBlocks()); if (LDistVerify) { LI->verify(*DT); @@ -798,7 +801,7 @@ public: LLVMContext &Ctx = F->getContext(); bool Forced = isForced().getValueOr(false); - DEBUG(dbgs() << "Skipping; " << Message << "\n"); + LLVM_DEBUG(dbgs() << "Skipping; " << Message << "\n"); // With Rpass-missed report that distribution failed. ORE->emit([&]() { diff --git a/llvm/lib/Transforms/Scalar/LoopIdiomRecognize.cpp b/llvm/lib/Transforms/Scalar/LoopIdiomRecognize.cpp index 5bfb44bc151..abd6a81da1e 100644 --- a/llvm/lib/Transforms/Scalar/LoopIdiomRecognize.cpp +++ b/llvm/lib/Transforms/Scalar/LoopIdiomRecognize.cpp @@ -310,9 +310,9 @@ bool LoopIdiomRecognize::runOnCountableLoop() { SmallVector<BasicBlock *, 8> ExitBlocks; CurLoop->getUniqueExitBlocks(ExitBlocks); - DEBUG(dbgs() << "loop-idiom Scanning: F[" - << CurLoop->getHeader()->getParent()->getName() << "] Loop %" - << CurLoop->getHeader()->getName() << "\n"); + LLVM_DEBUG(dbgs() << "loop-idiom Scanning: F[" + << CurLoop->getHeader()->getParent()->getName() + << "] Loop %" << CurLoop->getHeader()->getName() << "\n"); bool MadeChange = false; @@ -936,8 +936,9 @@ bool LoopIdiomRecognize::processLoopStridedStore( NewCall = Builder.CreateCall(MSP, {BasePtr, PatternPtr, NumBytes}); } - DEBUG(dbgs() << " Formed memset: " << *NewCall << "\n" - << " from store to: " << *Ev << " at: " << *TheStore << "\n"); + LLVM_DEBUG(dbgs() << " Formed memset: " << *NewCall << "\n" + << " from store to: " << *Ev << " at: " << *TheStore + << "\n"); NewCall->setDebugLoc(TheStore->getDebugLoc()); // Okay, the memset has been formed. Zap the original store and anything that @@ -1067,9 +1068,10 @@ bool LoopIdiomRecognize::processLoopStoreOfLoopLoad(StoreInst *SI, } NewCall->setDebugLoc(SI->getDebugLoc()); - DEBUG(dbgs() << " Formed memcpy: " << *NewCall << "\n" - << " from load ptr=" << *LoadEv << " at: " << *LI << "\n" - << " from store ptr=" << *StoreEv << " at: " << *SI << "\n"); + LLVM_DEBUG(dbgs() << " Formed memcpy: " << *NewCall << "\n" + << " from load ptr=" << *LoadEv << " at: " << *LI << "\n" + << " from store ptr=" << *StoreEv << " at: " << *SI + << "\n"); // Okay, the memcpy has been formed. Zap the original store and anything that // feeds into it. @@ -1085,9 +1087,9 @@ bool LoopIdiomRecognize::avoidLIRForMultiBlockLoop(bool IsMemset, bool IsLoopMemset) { if (ApplyCodeSizeHeuristics && CurLoop->getNumBlocks() > 1) { if (!CurLoop->getParentLoop() && (!IsMemset || !IsLoopMemset)) { - DEBUG(dbgs() << " " << CurLoop->getHeader()->getParent()->getName() - << " : LIR " << (IsMemset ? "Memset" : "Memcpy") - << " avoided: multi-block top-level loop\n"); + LLVM_DEBUG(dbgs() << " " << CurLoop->getHeader()->getParent()->getName() + << " : LIR " << (IsMemset ? "Memset" : "Memcpy") + << " avoided: multi-block top-level loop\n"); return true; } } diff --git a/llvm/lib/Transforms/Scalar/LoopInterchange.cpp b/llvm/lib/Transforms/Scalar/LoopInterchange.cpp index 74c8a5a543e..9addb04e81b 100644 --- a/llvm/lib/Transforms/Scalar/LoopInterchange.cpp +++ b/llvm/lib/Transforms/Scalar/LoopInterchange.cpp @@ -77,8 +77,8 @@ static const unsigned MaxLoopNestDepth = 10; static void printDepMatrix(CharMatrix &DepMatrix) { for (auto &Row : DepMatrix) { for (auto D : Row) - DEBUG(dbgs() << D << " "); - DEBUG(dbgs() << "\n"); + LLVM_DEBUG(dbgs() << D << " "); + LLVM_DEBUG(dbgs() << "\n"); } } #endif @@ -107,8 +107,8 @@ static bool populateDependencyMatrix(CharMatrix &DepMatrix, unsigned Level, } } - DEBUG(dbgs() << "Found " << MemInstr.size() - << " Loads and Stores to analyze\n"); + LLVM_DEBUG(dbgs() << "Found " << MemInstr.size() + << " Loads and Stores to analyze\n"); ValueVector::iterator I, IE, J, JE; @@ -125,11 +125,11 @@ static bool populateDependencyMatrix(CharMatrix &DepMatrix, unsigned Level, // Track Output, Flow, and Anti dependencies. if (auto D = DI->depends(Src, Dst, true)) { assert(D->isOrdered() && "Expected an output, flow or anti dep."); - DEBUG(StringRef DepType = - D->isFlow() ? "flow" : D->isAnti() ? "anti" : "output"; - dbgs() << "Found " << DepType - << " dependency between Src and Dst\n" - << " Src:" << *Src << "\n Dst:" << *Dst << '\n'); + LLVM_DEBUG(StringRef DepType = + D->isFlow() ? "flow" : D->isAnti() ? "anti" : "output"; + dbgs() << "Found " << DepType + << " dependency between Src and Dst\n" + << " Src:" << *Src << "\n Dst:" << *Dst << '\n'); unsigned Levels = D->getLevels(); char Direction; for (unsigned II = 1; II <= Levels; ++II) { @@ -169,8 +169,8 @@ static bool populateDependencyMatrix(CharMatrix &DepMatrix, unsigned Level, DepMatrix.push_back(Dep); if (DepMatrix.size() > MaxMemInstrCount) { - DEBUG(dbgs() << "Cannot handle more than " << MaxMemInstrCount - << " dependencies inside loop\n"); + LLVM_DEBUG(dbgs() << "Cannot handle more than " << MaxMemInstrCount + << " dependencies inside loop\n"); return false; } } @@ -272,9 +272,9 @@ static bool isLegalToInterChangeLoops(CharMatrix &DepMatrix, } static void populateWorklist(Loop &L, SmallVector<LoopVector, 8> &V) { - DEBUG(dbgs() << "Calling populateWorklist on Func: " - << L.getHeader()->getParent()->getName() << " Loop: %" - << L.getHeader()->getName() << '\n'); + LLVM_DEBUG(dbgs() << "Calling populateWorklist on Func: " + << L.getHeader()->getParent()->getName() << " Loop: %" + << L.getHeader()->getName() << '\n'); LoopVector LoopList; Loop *CurrentLoop = &L; const std::vector<Loop *> *Vec = &CurrentLoop->getSubLoops(); @@ -478,7 +478,7 @@ struct LoopInterchange : public FunctionPass { for (Loop *L : *LI) populateWorklist(*L, Worklist); - DEBUG(dbgs() << "Worklist size = " << Worklist.size() << "\n"); + LLVM_DEBUG(dbgs() << "Worklist size = " << Worklist.size() << "\n"); bool Changed = true; while (!Worklist.empty()) { LoopVector LoopList = Worklist.pop_back_val(); @@ -491,15 +491,15 @@ struct LoopInterchange : public FunctionPass { for (Loop *L : LoopList) { const SCEV *ExitCountOuter = SE->getBackedgeTakenCount(L); if (ExitCountOuter == SE->getCouldNotCompute()) { - DEBUG(dbgs() << "Couldn't compute backedge count\n"); + LLVM_DEBUG(dbgs() << "Couldn't compute backedge count\n"); return false; } if (L->getNumBackEdges() != 1) { - DEBUG(dbgs() << "NumBackEdges is not equal to 1\n"); + LLVM_DEBUG(dbgs() << "NumBackEdges is not equal to 1\n"); return false; } if (!L->getExitingBlock()) { - DEBUG(dbgs() << "Loop doesn't have unique exit block\n"); + LLVM_DEBUG(dbgs() << "Loop doesn't have unique exit block\n"); return false; } } @@ -516,37 +516,38 @@ struct LoopInterchange : public FunctionPass { bool Changed = false; unsigned LoopNestDepth = LoopList.size(); if (LoopNestDepth < 2) { - DEBUG(dbgs() << "Loop doesn't contain minimum nesting level.\n"); + LLVM_DEBUG(dbgs() << "Loop doesn't contain minimum nesting level.\n"); return false; } if (LoopNestDepth > MaxLoopNestDepth) { - DEBUG(dbgs() << "Cannot handle loops of depth greater than " - << MaxLoopNestDepth << "\n"); + LLVM_DEBUG(dbgs() << "Cannot handle loops of depth greater than " + << MaxLoopNestDepth << "\n"); return false; } if (!isComputableLoopNest(LoopList)) { - DEBUG(dbgs() << "Not valid loop candidate for interchange\n"); + LLVM_DEBUG(dbgs() << "Not valid loop candidate for interchange\n"); return false; } - DEBUG(dbgs() << "Processing LoopList of size = " << LoopNestDepth << "\n"); + LLVM_DEBUG(dbgs() << "Processing LoopList of size = " << LoopNestDepth + << "\n"); CharMatrix DependencyMatrix; Loop *OuterMostLoop = *(LoopList.begin()); if (!populateDependencyMatrix(DependencyMatrix, LoopNestDepth, OuterMostLoop, DI)) { - DEBUG(dbgs() << "Populating dependency matrix failed\n"); + LLVM_DEBUG(dbgs() << "Populating dependency matrix failed\n"); return false; } #ifdef DUMP_DEP_MATRICIES - DEBUG(dbgs() << "Dependence before interchange\n"); + LLVM_DEBUG(dbgs() << "Dependence before interchange\n"); printDepMatrix(DependencyMatrix); #endif // Get the Outermost loop exit. BasicBlock *LoopNestExit = OuterMostLoop->getExitBlock(); if (!LoopNestExit) { - DEBUG(dbgs() << "OuterMostLoop needs an unique exit block"); + LLVM_DEBUG(dbgs() << "OuterMostLoop needs an unique exit block"); return false; } @@ -563,7 +564,7 @@ struct LoopInterchange : public FunctionPass { // Update the DependencyMatrix interChangeDependencies(DependencyMatrix, i, i - 1); #ifdef DUMP_DEP_MATRICIES - DEBUG(dbgs() << "Dependence after interchange\n"); + LLVM_DEBUG(dbgs() << "Dependence after interchange\n"); printDepMatrix(DependencyMatrix); #endif Changed |= Interchanged; @@ -574,21 +575,21 @@ struct LoopInterchange : public FunctionPass { bool processLoop(LoopVector LoopList, unsigned InnerLoopId, unsigned OuterLoopId, BasicBlock *LoopNestExit, std::vector<std::vector<char>> &DependencyMatrix) { - DEBUG(dbgs() << "Processing Inner Loop Id = " << InnerLoopId - << " and OuterLoopId = " << OuterLoopId << "\n"); + LLVM_DEBUG(dbgs() << "Processing Inner Loop Id = " << InnerLoopId + << " and OuterLoopId = " << OuterLoopId << "\n"); Loop *InnerLoop = LoopList[InnerLoopId]; Loop *OuterLoop = LoopList[OuterLoopId]; LoopInterchangeLegality LIL(OuterLoop, InnerLoop, SE, LI, DT, PreserveLCSSA, ORE); if (!LIL.canInterchangeLoops(InnerLoopId, OuterLoopId, DependencyMatrix)) { - DEBUG(dbgs() << "Not interchanging loops. Cannot prove legality.\n"); + LLVM_DEBUG(dbgs() << "Not interchanging loops. Cannot prove legality.\n"); return false; } - DEBUG(dbgs() << "Loops are legal to interchange\n"); + LLVM_DEBUG(dbgs() << "Loops are legal to interchange\n"); LoopInterchangeProfitability LIP(OuterLoop, InnerLoop, SE, ORE); if (!LIP.isProfitable(InnerLoopId, OuterLoopId, DependencyMatrix)) { - DEBUG(dbgs() << "Interchanging loops not profitable.\n"); + LLVM_DEBUG(dbgs() << "Interchanging loops not profitable.\n"); return false; } @@ -602,7 +603,7 @@ struct LoopInterchange : public FunctionPass { LoopInterchangeTransform LIT(OuterLoop, InnerLoop, SE, LI, DT, LoopNestExit, LIL.hasInnerLoopReduction()); LIT.transform(); - DEBUG(dbgs() << "Loops interchanged.\n"); + LLVM_DEBUG(dbgs() << "Loops interchanged.\n"); LoopsInterchanged++; return true; } @@ -651,7 +652,7 @@ bool LoopInterchangeLegality::tightlyNested(Loop *OuterLoop, Loop *InnerLoop) { BasicBlock *InnerLoopPreHeader = InnerLoop->getLoopPreheader(); BasicBlock *OuterLoopLatch = OuterLoop->getLoopLatch(); - DEBUG(dbgs() << "Checking if loops are tightly nested\n"); + LLVM_DEBUG(dbgs() << "Checking if loops are tightly nested\n"); // A perfectly nested loop will not have any branch in between the outer and // inner block i.e. outer header will branch to either inner preheader and @@ -665,14 +666,14 @@ bool LoopInterchangeLegality::tightlyNested(Loop *OuterLoop, Loop *InnerLoop) { if (Succ != InnerLoopPreHeader && Succ != OuterLoopLatch) return false; - DEBUG(dbgs() << "Checking instructions in Loop header and Loop latch\n"); + LLVM_DEBUG(dbgs() << "Checking instructions in Loop header and Loop latch\n"); // We do not have any basic block in between now make sure the outer header // and outer loop latch doesn't contain any unsafe instructions. if (containsUnsafeInstructionsInHeader(OuterLoopHeader) || containsUnsafeInstructionsInLatch(OuterLoopLatch)) return false; - DEBUG(dbgs() << "Loops are perfectly nested\n"); + LLVM_DEBUG(dbgs() << "Loops are perfectly nested\n"); // We have a perfect loop nest. return true; } @@ -714,7 +715,7 @@ bool LoopInterchangeLegality::findInductionAndReductions( else if (RecurrenceDescriptor::isReductionPHI(&PHI, L, RD)) Reductions.push_back(&PHI); else { - DEBUG( + LLVM_DEBUG( dbgs() << "Failed to recognize PHI as an induction or reduction.\n"); return false; } @@ -750,8 +751,9 @@ bool LoopInterchangeLegality::currentLimitations() { OuterLoop->getExitingBlock() != OuterLoop->getLoopLatch() || !isa<BranchInst>(InnerLoopLatch->getTerminator()) || !isa<BranchInst>(OuterLoop->getLoopLatch()->getTerminator())) { - DEBUG(dbgs() << "Loops where the latch is not the exiting block are not" - << " supported currently.\n"); + LLVM_DEBUG( + dbgs() << "Loops where the latch is not the exiting block are not" + << " supported currently.\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "ExitingNotLatch", OuterLoop->getStartLoc(), @@ -766,8 +768,9 @@ bool LoopInterchangeLegality::currentLimitations() { SmallVector<PHINode *, 8> Inductions; SmallVector<PHINode *, 8> Reductions; if (!findInductionAndReductions(InnerLoop, Inductions, Reductions)) { - DEBUG(dbgs() << "Only inner loops with induction or reduction PHI nodes " - << "are supported currently.\n"); + LLVM_DEBUG( + dbgs() << "Only inner loops with induction or reduction PHI nodes " + << "are supported currently.\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "UnsupportedPHIInner", InnerLoop->getStartLoc(), @@ -780,8 +783,9 @@ bool LoopInterchangeLegality::currentLimitations() { // TODO: Currently we handle only loops with 1 induction variable. if (Inductions.size() != 1) { - DEBUG(dbgs() << "We currently only support loops with 1 induction variable." - << "Failed to interchange due to current limitation\n"); + LLVM_DEBUG( + dbgs() << "We currently only support loops with 1 induction variable." + << "Failed to interchange due to current limitation\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "MultiInductionInner", InnerLoop->getStartLoc(), @@ -797,8 +801,9 @@ bool LoopInterchangeLegality::currentLimitations() { InnerInductionVar = Inductions.pop_back_val(); Reductions.clear(); if (!findInductionAndReductions(OuterLoop, Inductions, Reductions)) { - DEBUG(dbgs() << "Only outer loops with induction or reduction PHI nodes " - << "are supported currently.\n"); + LLVM_DEBUG( + dbgs() << "Only outer loops with induction or reduction PHI nodes " + << "are supported currently.\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "UnsupportedPHIOuter", OuterLoop->getStartLoc(), @@ -812,8 +817,8 @@ bool LoopInterchangeLegality::currentLimitations() { // Outer loop cannot have reduction because then loops will not be tightly // nested. if (!Reductions.empty()) { - DEBUG(dbgs() << "Outer loops with reductions are not supported " - << "currently.\n"); + LLVM_DEBUG(dbgs() << "Outer loops with reductions are not supported " + << "currently.\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "ReductionsOuter", OuterLoop->getStartLoc(), @@ -825,8 +830,8 @@ bool LoopInterchangeLegality::currentLimitations() { } // TODO: Currently we handle only loops with 1 induction variable. if (Inductions.size() != 1) { - DEBUG(dbgs() << "Loops with more than 1 induction variables are not " - << "supported currently.\n"); + LLVM_DEBUG(dbgs() << "Loops with more than 1 induction variables are not " + << "supported currently.\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "MultiIndutionOuter", OuterLoop->getStartLoc(), @@ -839,7 +844,7 @@ bool LoopInterchangeLegality::currentLimitations() { // TODO: Triangular loops are not handled for now. if (!isLoopStructureUnderstood(InnerInductionVar)) { - DEBUG(dbgs() << "Loop structure not understood by pass\n"); + LLVM_DEBUG(dbgs() << "Loop structure not understood by pass\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "UnsupportedStructureInner", InnerLoop->getStartLoc(), @@ -852,7 +857,8 @@ bool LoopInterchangeLegality::currentLimitations() { // TODO: We only handle LCSSA PHI's corresponding to reduction for now. BasicBlock *InnerExit = InnerLoop->getExitBlock(); if (!containsSafePHI(InnerExit, false)) { - DEBUG(dbgs() << "Can only handle LCSSA PHIs in inner loops currently.\n"); + LLVM_DEBUG( + dbgs() << "Can only handle LCSSA PHIs in inner loops currently.\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "NoLCSSAPHIOuterInner", InnerLoop->getStartLoc(), @@ -882,8 +888,9 @@ bool LoopInterchangeLegality::currentLimitations() { dyn_cast<Instruction>(InnerInductionVar->getIncomingValue(0)); if (!InnerIndexVarInc) { - DEBUG(dbgs() << "Did not find an instruction to increment the induction " - << "variable.\n"); + LLVM_DEBUG( + dbgs() << "Did not find an instruction to increment the induction " + << "variable.\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "NoIncrementInInner", InnerLoop->getStartLoc(), @@ -907,8 +914,8 @@ bool LoopInterchangeLegality::currentLimitations() { // We found an instruction. If this is not induction variable then it is not // safe to split this loop latch. if (!I.isIdenticalTo(InnerIndexVarInc)) { - DEBUG(dbgs() << "Found unsupported instructions between induction " - << "variable increment and branch.\n"); + LLVM_DEBUG(dbgs() << "Found unsupported instructions between induction " + << "variable increment and branch.\n"); ORE->emit([&]() { return OptimizationRemarkMissed( DEBUG_TYPE, "UnsupportedInsBetweenInduction", @@ -925,7 +932,7 @@ bool LoopInterchangeLegality::currentLimitations() { // The loop latch ended and we didn't find the induction variable return as // current limitation. if (!FoundInduction) { - DEBUG(dbgs() << "Did not find the induction variable.\n"); + LLVM_DEBUG(dbgs() << "Did not find the induction variable.\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "NoIndutionVariable", InnerLoop->getStartLoc(), @@ -978,9 +985,9 @@ bool LoopInterchangeLegality::canInterchangeLoops(unsigned InnerLoopId, unsigned OuterLoopId, CharMatrix &DepMatrix) { if (!isLegalToInterChangeLoops(DepMatrix, InnerLoopId, OuterLoopId)) { - DEBUG(dbgs() << "Failed interchange InnerLoopId = " << InnerLoopId - << " and OuterLoopId = " << OuterLoopId - << " due to dependence\n"); + LLVM_DEBUG(dbgs() << "Failed interchange InnerLoopId = " << InnerLoopId + << " and OuterLoopId = " << OuterLoopId + << " due to dependence\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "Dependence", InnerLoop->getStartLoc(), @@ -996,8 +1003,9 @@ bool LoopInterchangeLegality::canInterchangeLoops(unsigned InnerLoopId, // readnone functions do not prevent interchanging. if (CI->doesNotReadMemory()) continue; - DEBUG(dbgs() << "Loops with call instructions cannot be interchanged " - << "safely."); + LLVM_DEBUG( + dbgs() << "Loops with call instructions cannot be interchanged " + << "safely."); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "CallInst", CI->getDebugLoc(), @@ -1033,13 +1041,13 @@ bool LoopInterchangeLegality::canInterchangeLoops(unsigned InnerLoopId, // TODO: The loops could not be interchanged due to current limitations in the // transform module. if (currentLimitations()) { - DEBUG(dbgs() << "Not legal because of current transform limitation\n"); + LLVM_DEBUG(dbgs() << "Not legal because of current transform limitation\n"); return false; } // Check if the loops are tightly nested. if (!tightlyNested(OuterLoop, InnerLoop)) { - DEBUG(dbgs() << "Loops not tightly nested\n"); + LLVM_DEBUG(dbgs() << "Loops not tightly nested\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "NotTightlyNested", InnerLoop->getStartLoc(), @@ -1051,7 +1059,7 @@ bool LoopInterchangeLegality::canInterchangeLoops(unsigned InnerLoopId, } if (!areLoopExitPHIsSupported(OuterLoop, InnerLoop)) { - DEBUG(dbgs() << "Found unsupported PHI nodes in outer loop exit.\n"); + LLVM_DEBUG(dbgs() << "Found unsupported PHI nodes in outer loop exit.\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "UnsupportedExitPHI", OuterLoop->getStartLoc(), @@ -1145,7 +1153,7 @@ bool LoopInterchangeProfitability::isProfitable(unsigned InnerLoopId, // of induction variables in the instruction and allows reordering if number // of bad orders is more than good. int Cost = getInstrOrderCost(); - DEBUG(dbgs() << "Cost = " << Cost << "\n"); + LLVM_DEBUG(dbgs() << "Cost = " << Cost << "\n"); if (Cost < -LoopInterchangeCostThreshold) return true; @@ -1258,10 +1266,10 @@ bool LoopInterchangeTransform::transform() { if (InnerLoop->getSubLoops().empty()) { BasicBlock *InnerLoopPreHeader = InnerLoop->getLoopPreheader(); - DEBUG(dbgs() << "Calling Split Inner Loop\n"); + LLVM_DEBUG(dbgs() << "Calling Split Inner Loop\n"); PHINode *InductionPHI = getInductionVariable(InnerLoop, SE); if (!InductionPHI) { - DEBUG(dbgs() << "Failed to find the point to split loop latch \n"); + LLVM_DEBUG(dbgs() << "Failed to find the point to split loop latch \n"); return false; } @@ -1279,16 +1287,16 @@ bool LoopInterchangeTransform::transform() { // incremented/decremented. // TODO: This splitting logic may not work always. Fix this. splitInnerLoopLatch(InnerIndexVar); - DEBUG(dbgs() << "splitInnerLoopLatch done\n"); + LLVM_DEBUG(dbgs() << "splitInnerLoopLatch done\n"); // Splits the inner loops phi nodes out into a separate basic block. splitInnerLoopHeader(); - DEBUG(dbgs() << "splitInnerLoopHeader done\n"); + LLVM_DEBUG(dbgs() << "splitInnerLoopHeader done\n"); } Transformed |= adjustLoopLinks(); if (!Transformed) { - DEBUG(dbgs() << "adjustLoopLinks failed\n"); + LLVM_DEBUG(dbgs() << "adjustLoopLinks failed\n"); return false; } @@ -1322,8 +1330,8 @@ void LoopInterchangeTransform::splitInnerLoopHeader() { } } - DEBUG(dbgs() << "Output of splitInnerLoopHeader InnerLoopHeaderSucc & " - "InnerLoopHeader\n"); + LLVM_DEBUG(dbgs() << "Output of splitInnerLoopHeader InnerLoopHeaderSucc & " + "InnerLoopHeader\n"); } /// Move all instructions except the terminator from FromBB right before @@ -1370,7 +1378,7 @@ static void updateSuccessor(BranchInst *BI, BasicBlock *OldBB, } bool LoopInterchangeTransform::adjustLoopBranches() { - DEBUG(dbgs() << "adjustLoopBranches called\n"); + LLVM_DEBUG(dbgs() << "adjustLoopBranches called\n"); std::vector<DominatorTree::UpdateType> DTUpdates; // Adjust the loop preheader diff --git a/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp b/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp index a7c27662aa0..7f882191d1f 100644 --- a/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp +++ b/llvm/lib/Transforms/Scalar/LoopLoadElimination.cpp @@ -285,9 +285,11 @@ public: Candidates.remove_if([&](const StoreToLoadForwardingCandidate &Cand) { if (LoadToSingleCand[Cand.Load] != &Cand) { - DEBUG(dbgs() << "Removing from candidates: \n" << Cand - << " The load may have multiple stores forwarding to " - << "it\n"); + LLVM_DEBUG( + dbgs() << "Removing from candidates: \n" + << Cand + << " The load may have multiple stores forwarding to " + << "it\n"); return true; } return false; @@ -395,8 +397,9 @@ public: return false; }); - DEBUG(dbgs() << "\nPointer Checks (count: " << Checks.size() << "):\n"); - DEBUG(LAI.getRuntimePointerChecking()->printChecks(dbgs(), Checks)); + LLVM_DEBUG(dbgs() << "\nPointer Checks (count: " << Checks.size() + << "):\n"); + LLVM_DEBUG(LAI.getRuntimePointerChecking()->printChecks(dbgs(), Checks)); return Checks; } @@ -440,8 +443,8 @@ public: /// Top-level driver for each loop: find store->load forwarding /// candidates, add run-time checks and perform transformation. bool processLoop() { - DEBUG(dbgs() << "\nIn \"" << L->getHeader()->getParent()->getName() - << "\" checking " << *L << "\n"); + LLVM_DEBUG(dbgs() << "\nIn \"" << L->getHeader()->getParent()->getName() + << "\" checking " << *L << "\n"); // Look for store-to-load forwarding cases across the // backedge. E.g.: @@ -480,7 +483,7 @@ public: SmallVector<StoreToLoadForwardingCandidate, 4> Candidates; unsigned NumForwarding = 0; for (const StoreToLoadForwardingCandidate Cand : StoreToLoadDependences) { - DEBUG(dbgs() << "Candidate " << Cand); + LLVM_DEBUG(dbgs() << "Candidate " << Cand); // Make sure that the stored values is available everywhere in the loop in // the next iteration. @@ -499,9 +502,10 @@ public: continue; ++NumForwarding; - DEBUG(dbgs() - << NumForwarding - << ". Valid store-to-load forwarding across the loop backedge\n"); + LLVM_DEBUG( + dbgs() + << NumForwarding + << ". Valid store-to-load forwarding across the loop backedge\n"); Candidates.push_back(Cand); } if (Candidates.empty()) @@ -514,25 +518,26 @@ public: // Too many checks are likely to outweigh the benefits of forwarding. if (Checks.size() > Candidates.size() * CheckPerElim) { - DEBUG(dbgs() << "Too many run-time checks needed.\n"); + LLVM_DEBUG(dbgs() << "Too many run-time checks needed.\n"); return false; } if (LAI.getPSE().getUnionPredicate().getComplexity() > LoadElimSCEVCheckThreshold) { - DEBUG(dbgs() << "Too many SCEV run-time checks needed.\n"); + LLVM_DEBUG(dbgs() << "Too many SCEV run-time checks needed.\n"); return false; } if (!Checks.empty() || !LAI.getPSE().getUnionPredicate().isAlwaysTrue()) { if (L->getHeader()->getParent()->optForSize()) { - DEBUG(dbgs() << "Versioning is needed but not allowed when optimizing " - "for size.\n"); + LLVM_DEBUG( + dbgs() << "Versioning is needed but not allowed when optimizing " + "for size.\n"); return false; } if (!L->isLoopSimplifyForm()) { - DEBUG(dbgs() << "Loop is not is loop-simplify form"); + LLVM_DEBUG(dbgs() << "Loop is not is loop-simplify form"); return false; } diff --git a/llvm/lib/Transforms/Scalar/LoopPredication.cpp b/llvm/lib/Transforms/Scalar/LoopPredication.cpp index 6102890bc5e..561ceea1d88 100644 --- a/llvm/lib/Transforms/Scalar/LoopPredication.cpp +++ b/llvm/lib/Transforms/Scalar/LoopPredication.cpp @@ -411,11 +411,11 @@ LoopPredication::generateLoopLatchCheck(Type *RangeCheckType) { if (!NewLatchCheck.IV) return None; NewLatchCheck.Limit = SE->getTruncateExpr(LatchCheck.Limit, RangeCheckType); - DEBUG(dbgs() << "IV of type: " << *LatchType - << "can be represented as range check type:" << *RangeCheckType - << "\n"); - DEBUG(dbgs() << "LatchCheck.IV: " << *NewLatchCheck.IV << "\n"); - DEBUG(dbgs() << "LatchCheck.Limit: " << *NewLatchCheck.Limit << "\n"); + LLVM_DEBUG(dbgs() << "IV of type: " << *LatchType + << "can be represented as range check type:" + << *RangeCheckType << "\n"); + LLVM_DEBUG(dbgs() << "LatchCheck.IV: " << *NewLatchCheck.IV << "\n"); + LLVM_DEBUG(dbgs() << "LatchCheck.Limit: " << *NewLatchCheck.Limit << "\n"); return NewLatchCheck; } @@ -448,15 +448,15 @@ Optional<Value *> LoopPredication::widenICmpRangeCheckIncrementingLoop( SE->getMinusSCEV(LatchStart, SE->getOne(Ty))); if (!CanExpand(GuardStart) || !CanExpand(GuardLimit) || !CanExpand(LatchLimit) || !CanExpand(RHS)) { - DEBUG(dbgs() << "Can't expand limit check!\n"); + LLVM_DEBUG(dbgs() << "Can't expand limit check!\n"); return None; } auto LimitCheckPred = ICmpInst::getFlippedStrictnessPredicate(LatchCheck.Pred); - DEBUG(dbgs() << "LHS: " << *LatchLimit << "\n"); - DEBUG(dbgs() << "RHS: " << *RHS << "\n"); - DEBUG(dbgs() << "Pred: " << LimitCheckPred << "\n"); + LLVM_DEBUG(dbgs() << "LHS: " << *LatchLimit << "\n"); + LLVM_DEBUG(dbgs() << "RHS: " << *RHS << "\n"); + LLVM_DEBUG(dbgs() << "Pred: " << LimitCheckPred << "\n"); Instruction *InsertAt = Preheader->getTerminator(); auto *LimitCheck = @@ -475,16 +475,16 @@ Optional<Value *> LoopPredication::widenICmpRangeCheckDecrementingLoop( const SCEV *LatchLimit = LatchCheck.Limit; if (!CanExpand(GuardStart) || !CanExpand(GuardLimit) || !CanExpand(LatchLimit)) { - DEBUG(dbgs() << "Can't expand limit check!\n"); + LLVM_DEBUG(dbgs() << "Can't expand limit check!\n"); return None; } // The decrement of the latch check IV should be the same as the // rangeCheckIV. auto *PostDecLatchCheckIV = LatchCheck.IV->getPostIncExpr(*SE); if (RangeCheck.IV != PostDecLatchCheckIV) { - DEBUG(dbgs() << "Not the same. PostDecLatchCheckIV: " - << *PostDecLatchCheckIV - << " and RangeCheckIV: " << *RangeCheck.IV << "\n"); + LLVM_DEBUG(dbgs() << "Not the same. PostDecLatchCheckIV: " + << *PostDecLatchCheckIV + << " and RangeCheckIV: " << *RangeCheck.IV << "\n"); return None; } @@ -508,8 +508,8 @@ Optional<Value *> LoopPredication::widenICmpRangeCheckDecrementingLoop( Optional<Value *> LoopPredication::widenICmpRangeCheck(ICmpInst *ICI, SCEVExpander &Expander, IRBuilder<> &Builder) { - DEBUG(dbgs() << "Analyzing ICmpInst condition:\n"); - DEBUG(ICI->dump()); + LLVM_DEBUG(dbgs() << "Analyzing ICmpInst condition:\n"); + LLVM_DEBUG(ICI->dump()); // parseLoopStructure guarantees that the latch condition is: // ++i <pred> latchLimit, where <pred> is u<, u<=, s<, or s<=. @@ -517,34 +517,34 @@ Optional<Value *> LoopPredication::widenICmpRangeCheck(ICmpInst *ICI, // i u< guardLimit auto RangeCheck = parseLoopICmp(ICI); if (!RangeCheck) { - DEBUG(dbgs() << "Failed to parse the loop latch condition!\n"); + LLVM_DEBUG(dbgs() << "Failed to parse the loop latch condition!\n"); return None; } - DEBUG(dbgs() << "Guard check:\n"); - DEBUG(RangeCheck->dump()); + LLVM_DEBUG(dbgs() << "Guard check:\n"); + LLVM_DEBUG(RangeCheck->dump()); if (RangeCheck->Pred != ICmpInst::ICMP_ULT) { - DEBUG(dbgs() << "Unsupported range check predicate(" << RangeCheck->Pred - << ")!\n"); + LLVM_DEBUG(dbgs() << "Unsupported range check predicate(" + << RangeCheck->Pred << ")!\n"); return None; } auto *RangeCheckIV = RangeCheck->IV; if (!RangeCheckIV->isAffine()) { - DEBUG(dbgs() << "Range check IV is not affine!\n"); + LLVM_DEBUG(dbgs() << "Range check IV is not affine!\n"); return None; } auto *Step = RangeCheckIV->getStepRecurrence(*SE); // We cannot just compare with latch IV step because the latch and range IVs // may have different types. if (!isSupportedStep(Step)) { - DEBUG(dbgs() << "Range check and latch have IVs different steps!\n"); + LLVM_DEBUG(dbgs() << "Range check and latch have IVs different steps!\n"); return None; } auto *Ty = RangeCheckIV->getType(); auto CurrLatchCheckOpt = generateLoopLatchCheck(Ty); if (!CurrLatchCheckOpt) { - DEBUG(dbgs() << "Failed to generate a loop latch check " - "corresponding to range type: " - << *Ty << "\n"); + LLVM_DEBUG(dbgs() << "Failed to generate a loop latch check " + "corresponding to range type: " + << *Ty << "\n"); return None; } @@ -555,7 +555,7 @@ Optional<Value *> LoopPredication::widenICmpRangeCheck(ICmpInst *ICI, CurrLatchCheck.IV->getStepRecurrence(*SE)->getType() && "Range and latch steps should be of same type!"); if (Step != CurrLatchCheck.IV->getStepRecurrence(*SE)) { - DEBUG(dbgs() << "Range and latch have different step values!\n"); + LLVM_DEBUG(dbgs() << "Range and latch have different step values!\n"); return None; } @@ -571,8 +571,8 @@ Optional<Value *> LoopPredication::widenICmpRangeCheck(ICmpInst *ICI, bool LoopPredication::widenGuardConditions(IntrinsicInst *Guard, SCEVExpander &Expander) { - DEBUG(dbgs() << "Processing guard:\n"); - DEBUG(Guard->dump()); + LLVM_DEBUG(dbgs() << "Processing guard:\n"); + LLVM_DEBUG(Guard->dump()); IRBuilder<> Builder(cast<Instruction>(Preheader->getTerminator())); @@ -625,7 +625,7 @@ bool LoopPredication::widenGuardConditions(IntrinsicInst *Guard, LastCheck = Builder.CreateAnd(LastCheck, Check); Guard->setOperand(0, LastCheck); - DEBUG(dbgs() << "Widened checks = " << NumWidened << "\n"); + LLVM_DEBUG(dbgs() << "Widened checks = " << NumWidened << "\n"); return true; } @@ -634,7 +634,7 @@ Optional<LoopPredication::LoopICmp> LoopPredication::parseLoopLatchICmp() { BasicBlock *LoopLatch = L->getLoopLatch(); if (!LoopLatch) { - DEBUG(dbgs() << "The loop doesn't have a single latch!\n"); + LLVM_DEBUG(dbgs() << "The loop doesn't have a single latch!\n"); return None; } @@ -645,7 +645,7 @@ Optional<LoopPredication::LoopICmp> LoopPredication::parseLoopLatchICmp() { if (!match(LoopLatch->getTerminator(), m_Br(m_ICmp(Pred, m_Value(LHS), m_Value(RHS)), TrueDest, FalseDest))) { - DEBUG(dbgs() << "Failed to match the latch terminator!\n"); + LLVM_DEBUG(dbgs() << "Failed to match the latch terminator!\n"); return None; } assert((TrueDest == L->getHeader() || FalseDest == L->getHeader()) && @@ -655,20 +655,20 @@ Optional<LoopPredication::LoopICmp> LoopPredication::parseLoopLatchICmp() { auto Result = parseLoopICmp(Pred, LHS, RHS); if (!Result) { - DEBUG(dbgs() << "Failed to parse the loop latch condition!\n"); + LLVM_DEBUG(dbgs() << "Failed to parse the loop latch condition!\n"); return None; } // Check affine first, so if it's not we don't try to compute the step // recurrence. if (!Result->IV->isAffine()) { - DEBUG(dbgs() << "The induction variable is not affine!\n"); + LLVM_DEBUG(dbgs() << "The induction variable is not affine!\n"); return None; } auto *Step = Result->IV->getStepRecurrence(*SE); if (!isSupportedStep(Step)) { - DEBUG(dbgs() << "Unsupported loop stride(" << *Step << ")!\n"); + LLVM_DEBUG(dbgs() << "Unsupported loop stride(" << *Step << ")!\n"); return None; } @@ -684,8 +684,8 @@ Optional<LoopPredication::LoopICmp> LoopPredication::parseLoopLatchICmp() { }; if (IsUnsupportedPredicate(Step, Result->Pred)) { - DEBUG(dbgs() << "Unsupported loop latch predicate(" << Result->Pred - << ")!\n"); + LLVM_DEBUG(dbgs() << "Unsupported loop latch predicate(" << Result->Pred + << ")!\n"); return None; } return Result; @@ -751,11 +751,11 @@ bool LoopPredication::isLoopProfitableToPredicate() { // less than one, can invert the definition of profitable loop predication. float ScaleFactor = LatchExitProbabilityScale; if (ScaleFactor < 1) { - DEBUG( + LLVM_DEBUG( dbgs() << "Ignored user setting for loop-predication-latch-probability-scale: " << LatchExitProbabilityScale << "\n"); - DEBUG(dbgs() << "The value is set to 1.0\n"); + LLVM_DEBUG(dbgs() << "The value is set to 1.0\n"); ScaleFactor = 1.0; } const auto LatchProbabilityThreshold = @@ -778,8 +778,8 @@ bool LoopPredication::isLoopProfitableToPredicate() { bool LoopPredication::runOnLoop(Loop *Loop) { L = Loop; - DEBUG(dbgs() << "Analyzing "); - DEBUG(L->dump()); + LLVM_DEBUG(dbgs() << "Analyzing "); + LLVM_DEBUG(L->dump()); Module *M = L->getHeader()->getModule(); @@ -800,11 +800,11 @@ bool LoopPredication::runOnLoop(Loop *Loop) { return false; LatchCheck = *LatchCheckOpt; - DEBUG(dbgs() << "Latch check:\n"); - DEBUG(LatchCheck.dump()); + LLVM_DEBUG(dbgs() << "Latch check:\n"); + LLVM_DEBUG(LatchCheck.dump()); if (!isLoopProfitableToPredicate()) { - DEBUG(dbgs()<< "Loop not profitable to predicate!\n"); + LLVM_DEBUG(dbgs() << "Loop not profitable to predicate!\n"); return false; } // Collect all the guards into a vector and process later, so as not diff --git a/llvm/lib/Transforms/Scalar/LoopRerollPass.cpp b/llvm/lib/Transforms/Scalar/LoopRerollPass.cpp index 1f693467647..83dd196f20e 100644 --- a/llvm/lib/Transforms/Scalar/LoopRerollPass.cpp +++ b/llvm/lib/Transforms/Scalar/LoopRerollPass.cpp @@ -644,14 +644,14 @@ void LoopReroll::collectPossibleIVs(Loop *L, if (IncSCEV->getValue()->isZero() || AInt.uge(MaxInc)) continue; IVToIncMap[&*I] = IncSCEV->getValue()->getSExtValue(); - DEBUG(dbgs() << "LRR: Possible IV: " << *I << " = " << *PHISCEV - << "\n"); + LLVM_DEBUG(dbgs() << "LRR: Possible IV: " << *I << " = " << *PHISCEV + << "\n"); if (isLoopControlIV(L, &*I)) { assert(!LoopControlIV && "Found two loop control only IV"); LoopControlIV = &(*I); - DEBUG(dbgs() << "LRR: Possible loop control only IV: " << *I << " = " - << *PHISCEV << "\n"); + LLVM_DEBUG(dbgs() << "LRR: Possible loop control only IV: " << *I + << " = " << *PHISCEV << "\n"); } else PossibleIVs.push_back(&*I); } @@ -718,8 +718,8 @@ void LoopReroll::collectPossibleReductions(Loop *L, if (!SLR.valid()) continue; - DEBUG(dbgs() << "LRR: Possible reduction: " << *I << " (with " << - SLR.size() << " chained instructions)\n"); + LLVM_DEBUG(dbgs() << "LRR: Possible reduction: " << *I << " (with " + << SLR.size() << " chained instructions)\n"); Reductions.addSLR(SLR); } } @@ -857,7 +857,8 @@ collectPossibleRoots(Instruction *Base, std::map<int64_t,Instruction*> &Roots) { BaseUsers.push_back(II); continue; } else { - DEBUG(dbgs() << "LRR: Aborting due to non-instruction: " << *I << "\n"); + LLVM_DEBUG(dbgs() << "LRR: Aborting due to non-instruction: " << *I + << "\n"); return false; } } @@ -879,7 +880,7 @@ collectPossibleRoots(Instruction *Base, std::map<int64_t,Instruction*> &Roots) { // away. if (BaseUsers.size()) { if (Roots.find(0) != Roots.end()) { - DEBUG(dbgs() << "LRR: Multiple roots found for base - aborting!\n"); + LLVM_DEBUG(dbgs() << "LRR: Multiple roots found for base - aborting!\n"); return false; } Roots[0] = Base; @@ -895,9 +896,9 @@ collectPossibleRoots(Instruction *Base, std::map<int64_t,Instruction*> &Roots) { if (KV.first == 0) continue; if (!KV.second->hasNUses(NumBaseUses)) { - DEBUG(dbgs() << "LRR: Aborting - Root and Base #users not the same: " - << "#Base=" << NumBaseUses << ", #Root=" << - KV.second->getNumUses() << "\n"); + LLVM_DEBUG(dbgs() << "LRR: Aborting - Root and Base #users not the same: " + << "#Base=" << NumBaseUses + << ", #Root=" << KV.second->getNumUses() << "\n"); return false; } } @@ -1025,13 +1026,14 @@ bool LoopReroll::DAGRootTracker::findRoots() { // Ensure all sets have the same size. if (RootSets.empty()) { - DEBUG(dbgs() << "LRR: Aborting because no root sets found!\n"); + LLVM_DEBUG(dbgs() << "LRR: Aborting because no root sets found!\n"); return false; } for (auto &V : RootSets) { if (V.Roots.empty() || V.Roots.size() != RootSets[0].Roots.size()) { - DEBUG(dbgs() - << "LRR: Aborting because not all root sets have the same size\n"); + LLVM_DEBUG( + dbgs() + << "LRR: Aborting because not all root sets have the same size\n"); return false; } } @@ -1039,13 +1041,14 @@ bool LoopReroll::DAGRootTracker::findRoots() { Scale = RootSets[0].Roots.size() + 1; if (Scale > IL_MaxRerollIterations) { - DEBUG(dbgs() << "LRR: Aborting - too many iterations found. " - << "#Found=" << Scale << ", #Max=" << IL_MaxRerollIterations - << "\n"); + LLVM_DEBUG(dbgs() << "LRR: Aborting - too many iterations found. " + << "#Found=" << Scale + << ", #Max=" << IL_MaxRerollIterations << "\n"); return false; } - DEBUG(dbgs() << "LRR: Successfully found roots: Scale=" << Scale << "\n"); + LLVM_DEBUG(dbgs() << "LRR: Successfully found roots: Scale=" << Scale + << "\n"); return true; } @@ -1079,7 +1082,7 @@ bool LoopReroll::DAGRootTracker::collectUsedInstructions(SmallInstructionSet &Po // While we're here, check the use sets are the same size. if (V.size() != VBase.size()) { - DEBUG(dbgs() << "LRR: Aborting - use sets are different sizes\n"); + LLVM_DEBUG(dbgs() << "LRR: Aborting - use sets are different sizes\n"); return false; } @@ -1236,17 +1239,17 @@ bool LoopReroll::DAGRootTracker::validate(ReductionTracker &Reductions) { // set. for (auto &KV : Uses) { if (KV.second.count() != 1 && !isIgnorableInst(KV.first)) { - DEBUG(dbgs() << "LRR: Aborting - instruction is not used in 1 iteration: " - << *KV.first << " (#uses=" << KV.second.count() << ")\n"); + LLVM_DEBUG( + dbgs() << "LRR: Aborting - instruction is not used in 1 iteration: " + << *KV.first << " (#uses=" << KV.second.count() << ")\n"); return false; } } - DEBUG( - for (auto &KV : Uses) { - dbgs() << "LRR: " << KV.second.find_first() << "\t" << *KV.first << "\n"; - } - ); + LLVM_DEBUG(for (auto &KV + : Uses) { + dbgs() << "LRR: " << KV.second.find_first() << "\t" << *KV.first << "\n"; + }); for (unsigned Iter = 1; Iter < Scale; ++Iter) { // In addition to regular aliasing information, we need to look for @@ -1305,8 +1308,8 @@ bool LoopReroll::DAGRootTracker::validate(ReductionTracker &Reductions) { if (TryIt == Uses.end() || TryIt == RootIt || instrDependsOn(TryIt->first, RootIt, TryIt)) { - DEBUG(dbgs() << "LRR: iteration root match failed at " << *BaseInst << - " vs. " << *RootInst << "\n"); + LLVM_DEBUG(dbgs() << "LRR: iteration root match failed at " + << *BaseInst << " vs. " << *RootInst << "\n"); return false; } @@ -1342,8 +1345,8 @@ bool LoopReroll::DAGRootTracker::validate(ReductionTracker &Reductions) { // root instruction, does not also belong to the base set or the set of // some other root instruction. if (RootIt->second.count() > 1) { - DEBUG(dbgs() << "LRR: iteration root match failed at " << *BaseInst << - " vs. " << *RootInst << " (prev. case overlap)\n"); + LLVM_DEBUG(dbgs() << "LRR: iteration root match failed at " << *BaseInst + << " vs. " << *RootInst << " (prev. case overlap)\n"); return false; } @@ -1353,8 +1356,9 @@ bool LoopReroll::DAGRootTracker::validate(ReductionTracker &Reductions) { if (RootInst->mayReadFromMemory()) for (auto &K : AST) { if (K.aliasesUnknownInst(RootInst, *AA)) { - DEBUG(dbgs() << "LRR: iteration root match failed at " << *BaseInst << - " vs. " << *RootInst << " (depends on future store)\n"); + LLVM_DEBUG(dbgs() << "LRR: iteration root match failed at " + << *BaseInst << " vs. " << *RootInst + << " (depends on future store)\n"); return false; } } @@ -1367,9 +1371,9 @@ bool LoopReroll::DAGRootTracker::validate(ReductionTracker &Reductions) { !isSafeToSpeculativelyExecute(BaseInst)) || (!isUnorderedLoadStore(RootInst) && !isSafeToSpeculativelyExecute(RootInst)))) { - DEBUG(dbgs() << "LRR: iteration root match failed at " << *BaseInst << - " vs. " << *RootInst << - " (side effects prevent reordering)\n"); + LLVM_DEBUG(dbgs() << "LRR: iteration root match failed at " << *BaseInst + << " vs. " << *RootInst + << " (side effects prevent reordering)\n"); return false; } @@ -1420,8 +1424,9 @@ bool LoopReroll::DAGRootTracker::validate(ReductionTracker &Reductions) { BaseInst->getOperand(!j) == Op2) { Swapped = true; } else { - DEBUG(dbgs() << "LRR: iteration root match failed at " << *BaseInst - << " vs. " << *RootInst << " (operand " << j << ")\n"); + LLVM_DEBUG(dbgs() + << "LRR: iteration root match failed at " << *BaseInst + << " vs. " << *RootInst << " (operand " << j << ")\n"); return false; } } @@ -1434,8 +1439,8 @@ bool LoopReroll::DAGRootTracker::validate(ReductionTracker &Reductions) { hasUsesOutsideLoop(BaseInst, L)) || (!PossibleRedLastSet.count(RootInst) && hasUsesOutsideLoop(RootInst, L))) { - DEBUG(dbgs() << "LRR: iteration root match failed at " << *BaseInst << - " vs. " << *RootInst << " (uses outside loop)\n"); + LLVM_DEBUG(dbgs() << "LRR: iteration root match failed at " << *BaseInst + << " vs. " << *RootInst << " (uses outside loop)\n"); return false; } @@ -1452,8 +1457,8 @@ bool LoopReroll::DAGRootTracker::validate(ReductionTracker &Reductions) { "Mismatched set sizes!"); } - DEBUG(dbgs() << "LRR: Matched all iteration increments for " << - *IV << "\n"); + LLVM_DEBUG(dbgs() << "LRR: Matched all iteration increments for " << *IV + << "\n"); return true; } @@ -1465,7 +1470,7 @@ void LoopReroll::DAGRootTracker::replace(const SCEV *IterCount) { J != JE;) { unsigned I = Uses[&*J].find_first(); if (I > 0 && I < IL_All) { - DEBUG(dbgs() << "LRR: removing: " << *J << "\n"); + LLVM_DEBUG(dbgs() << "LRR: removing: " << *J << "\n"); J++->eraseFromParent(); continue; } @@ -1618,17 +1623,17 @@ bool LoopReroll::ReductionTracker::validateSelected() { int Iter = PossibleRedIter[J]; if (Iter != PrevIter && Iter != PrevIter + 1 && !PossibleReds[i].getReducedValue()->isAssociative()) { - DEBUG(dbgs() << "LRR: Out-of-order non-associative reduction: " << - J << "\n"); + LLVM_DEBUG(dbgs() << "LRR: Out-of-order non-associative reduction: " + << J << "\n"); return false; } if (Iter != PrevIter) { if (Count != BaseCount) { - DEBUG(dbgs() << "LRR: Iteration " << PrevIter << - " reduction use count " << Count << - " is not equal to the base use count " << - BaseCount << "\n"); + LLVM_DEBUG(dbgs() + << "LRR: Iteration " << PrevIter << " reduction use count " + << Count << " is not equal to the base use count " + << BaseCount << "\n"); return false; } @@ -1724,8 +1729,8 @@ bool LoopReroll::reroll(Instruction *IV, Loop *L, BasicBlock *Header, if (!DAGRoots.findRoots()) return false; - DEBUG(dbgs() << "LRR: Found all root induction increments for: " << - *IV << "\n"); + LLVM_DEBUG(dbgs() << "LRR: Found all root induction increments for: " << *IV + << "\n"); if (!DAGRoots.validate(Reductions)) return false; @@ -1753,9 +1758,9 @@ bool LoopReroll::runOnLoop(Loop *L, LPPassManager &LPM) { PreserveLCSSA = mustPreserveAnalysisID(LCSSAID); BasicBlock *Header = L->getHeader(); - DEBUG(dbgs() << "LRR: F[" << Header->getParent()->getName() << - "] Loop %" << Header->getName() << " (" << - L->getNumBlocks() << " block(s))\n"); + LLVM_DEBUG(dbgs() << "LRR: F[" << Header->getParent()->getName() << "] Loop %" + << Header->getName() << " (" << L->getNumBlocks() + << " block(s))\n"); // For now, we'll handle only single BB loops. if (L->getNumBlocks() > 1) @@ -1766,8 +1771,8 @@ bool LoopReroll::runOnLoop(Loop *L, LPPassManager &LPM) { const SCEV *LIBETC = SE->getBackedgeTakenCount(L); const SCEV *IterCount = SE->getAddExpr(LIBETC, SE->getOne(LIBETC->getType())); - DEBUG(dbgs() << "\n Before Reroll:\n" << *(L->getHeader()) << "\n"); - DEBUG(dbgs() << "LRR: iteration count = " << *IterCount << "\n"); + LLVM_DEBUG(dbgs() << "\n Before Reroll:\n" << *(L->getHeader()) << "\n"); + LLVM_DEBUG(dbgs() << "LRR: iteration count = " << *IterCount << "\n"); // First, we need to find the induction variable with respect to which we can // reroll (there may be several possible options). @@ -1777,7 +1782,7 @@ bool LoopReroll::runOnLoop(Loop *L, LPPassManager &LPM) { collectPossibleIVs(L, PossibleIVs); if (PossibleIVs.empty()) { - DEBUG(dbgs() << "LRR: No possible IVs found\n"); + LLVM_DEBUG(dbgs() << "LRR: No possible IVs found\n"); return false; } @@ -1792,7 +1797,7 @@ bool LoopReroll::runOnLoop(Loop *L, LPPassManager &LPM) { Changed = true; break; } - DEBUG(dbgs() << "\n After Reroll:\n" << *(L->getHeader()) << "\n"); + LLVM_DEBUG(dbgs() << "\n After Reroll:\n" << *(L->getHeader()) << "\n"); // Trip count of L has changed so SE must be re-evaluated. if (Changed) diff --git a/llvm/lib/Transforms/Scalar/LoopSink.cpp b/llvm/lib/Transforms/Scalar/LoopSink.cpp index a2983e60b8f..731506d9107 100644 --- a/llvm/lib/Transforms/Scalar/LoopSink.cpp +++ b/llvm/lib/Transforms/Scalar/LoopSink.cpp @@ -224,11 +224,11 @@ static bool sinkInstruction(Loop &L, Instruction &I, } // Replaces uses of I with IC in blocks dominated by N replaceDominatedUsesWith(&I, IC, DT, N); - DEBUG(dbgs() << "Sinking a clone of " << I << " To: " << N->getName() - << '\n'); + LLVM_DEBUG(dbgs() << "Sinking a clone of " << I << " To: " << N->getName() + << '\n'); NumLoopSunkCloned++; } - DEBUG(dbgs() << "Sinking " << I << " To: " << MoveBB->getName() << '\n'); + LLVM_DEBUG(dbgs() << "Sinking " << I << " To: " << MoveBB->getName() << '\n'); NumLoopSunk++; I.moveBefore(&*MoveBB->getFirstInsertionPt()); diff --git a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp index 4c0b3cc808c..b46dc749390 100644 --- a/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp +++ b/llvm/lib/Transforms/Scalar/LoopStrengthReduce.cpp @@ -2424,8 +2424,8 @@ LSRInstance::OptimizeLoopTermCond() { } } - DEBUG(dbgs() << " Change loop exiting icmp to use postinc iv: " - << *Cond << '\n'); + LLVM_DEBUG(dbgs() << " Change loop exiting icmp to use postinc iv: " + << *Cond << '\n'); // It's possible for the setcc instruction to be anywhere in the loop, and // possible for it to have multiple users. If it is not immediately before @@ -2666,7 +2666,7 @@ void LSRInstance::CollectInterestingTypesAndFactors() { if (Types.size() == 1) Types.clear(); - DEBUG(print_factors_and_types(dbgs())); + LLVM_DEBUG(print_factors_and_types(dbgs())); } /// Helper for CollectChains that finds an IV operand (computed by an AddRec in @@ -2797,10 +2797,9 @@ isProfitableChain(IVChain &Chain, SmallPtrSetImpl<Instruction*> &Users, return false; if (!Users.empty()) { - DEBUG(dbgs() << "Chain: " << *Chain.Incs[0].UserInst << " users:\n"; - for (Instruction *Inst : Users) { - dbgs() << " " << *Inst << "\n"; - }); + LLVM_DEBUG(dbgs() << "Chain: " << *Chain.Incs[0].UserInst << " users:\n"; + for (Instruction *Inst + : Users) { dbgs() << " " << *Inst << "\n"; }); return false; } assert(!Chain.Incs.empty() && "empty IV chains are not allowed"); @@ -2853,8 +2852,8 @@ isProfitableChain(IVChain &Chain, SmallPtrSetImpl<Instruction*> &Users, // the stride. cost -= NumReusedIncrements; - DEBUG(dbgs() << "Chain: " << *Chain.Incs[0].UserInst << " Cost: " << cost - << "\n"); + LLVM_DEBUG(dbgs() << "Chain: " << *Chain.Incs[0].UserInst << " Cost: " << cost + << "\n"); return cost < 0; } @@ -2907,7 +2906,7 @@ void LSRInstance::ChainInstruction(Instruction *UserInst, Instruction *IVOper, if (isa<PHINode>(UserInst)) return; if (NChains >= MaxChains && !StressIVChain) { - DEBUG(dbgs() << "IV Chain Limit\n"); + LLVM_DEBUG(dbgs() << "IV Chain Limit\n"); return; } LastIncExpr = OperExpr; @@ -2920,11 +2919,11 @@ void LSRInstance::ChainInstruction(Instruction *UserInst, Instruction *IVOper, IVChainVec.push_back(IVChain(IVInc(UserInst, IVOper, LastIncExpr), OperExprBase)); ChainUsersVec.resize(NChains); - DEBUG(dbgs() << "IV Chain#" << ChainIdx << " Head: (" << *UserInst - << ") IV=" << *LastIncExpr << "\n"); + LLVM_DEBUG(dbgs() << "IV Chain#" << ChainIdx << " Head: (" << *UserInst + << ") IV=" << *LastIncExpr << "\n"); } else { - DEBUG(dbgs() << "IV Chain#" << ChainIdx << " Inc: (" << *UserInst - << ") IV+" << *LastIncExpr << "\n"); + LLVM_DEBUG(dbgs() << "IV Chain#" << ChainIdx << " Inc: (" << *UserInst + << ") IV+" << *LastIncExpr << "\n"); // Add this IV user to the end of the chain. IVChainVec[ChainIdx].add(IVInc(UserInst, IVOper, LastIncExpr)); } @@ -2994,7 +2993,7 @@ void LSRInstance::ChainInstruction(Instruction *UserInst, Instruction *IVOper, /// loop latch. This will discover chains on side paths, but requires /// maintaining multiple copies of the Chains state. void LSRInstance::CollectChains() { - DEBUG(dbgs() << "Collecting IV Chains.\n"); + LLVM_DEBUG(dbgs() << "Collecting IV Chains.\n"); SmallVector<ChainUsers, 8> ChainUsersVec; SmallVector<BasicBlock *,8> LatchPath; @@ -3063,10 +3062,10 @@ void LSRInstance::CollectChains() { void LSRInstance::FinalizeChain(IVChain &Chain) { assert(!Chain.Incs.empty() && "empty IV chains are not allowed"); - DEBUG(dbgs() << "Final Chain: " << *Chain.Incs[0].UserInst << "\n"); + LLVM_DEBUG(dbgs() << "Final Chain: " << *Chain.Incs[0].UserInst << "\n"); for (const IVInc &Inc : Chain) { - DEBUG(dbgs() << " Inc: " << *Inc.UserInst << "\n"); + LLVM_DEBUG(dbgs() << " Inc: " << *Inc.UserInst << "\n"); auto UseI = find(Inc.UserInst->operands(), Inc.IVOperand); assert(UseI != Inc.UserInst->op_end() && "cannot find IV operand"); IVIncSet.insert(UseI); @@ -3123,11 +3122,11 @@ void LSRInstance::GenerateIVChain(const IVChain &Chain, SCEVExpander &Rewriter, } if (IVOpIter == IVOpEnd) { // Gracefully give up on this chain. - DEBUG(dbgs() << "Concealed chain head: " << *Head.UserInst << "\n"); + LLVM_DEBUG(dbgs() << "Concealed chain head: " << *Head.UserInst << "\n"); return; } - DEBUG(dbgs() << "Generate chain at: " << *IVSrc << "\n"); + LLVM_DEBUG(dbgs() << "Generate chain at: " << *IVSrc << "\n"); Type *IVTy = IVSrc->getType(); Type *IntTy = SE.getEffectiveSCEVType(IVTy); const SCEV *LeftOverExpr = nullptr; @@ -3203,7 +3202,7 @@ void LSRInstance::CollectFixupsAndInitialFormulae() { find(UserInst->operands(), U.getOperandValToReplace()); assert(UseI != UserInst->op_end() && "cannot find IV operand"); if (IVIncSet.count(UseI)) { - DEBUG(dbgs() << "Use is in profitable chain: " << **UseI << '\n'); + LLVM_DEBUG(dbgs() << "Use is in profitable chain: " << **UseI << '\n'); continue; } @@ -3279,7 +3278,7 @@ void LSRInstance::CollectFixupsAndInitialFormulae() { } } - DEBUG(print_fixups(dbgs())); + LLVM_DEBUG(print_fixups(dbgs())); } /// Insert a formula for the given expression into the given use, separating out @@ -3995,10 +3994,11 @@ void LSRInstance::GenerateCrossUseConstantOffsets() { if (Imms.size() == 1) continue; - DEBUG(dbgs() << "Generating cross-use offsets for " << *Reg << ':'; - for (const auto &Entry : Imms) - dbgs() << ' ' << Entry.first; - dbgs() << '\n'); + LLVM_DEBUG(dbgs() << "Generating cross-use offsets for " << *Reg << ':'; + for (const auto &Entry + : Imms) dbgs() + << ' ' << Entry.first; + dbgs() << '\n'); // Examine each offset. for (ImmMapTy::const_iterator J = Imms.begin(), JE = Imms.end(); @@ -4010,7 +4010,8 @@ void LSRInstance::GenerateCrossUseConstantOffsets() { if (!isa<SCEVConstant>(OrigReg) && UsedByIndicesMap[Reg].count() == 1) { - DEBUG(dbgs() << "Skipping cross-use reuse for " << *OrigReg << '\n'); + LLVM_DEBUG(dbgs() << "Skipping cross-use reuse for " << *OrigReg + << '\n'); continue; } @@ -4159,9 +4160,9 @@ LSRInstance::GenerateAllReuseFormulae() { GenerateCrossUseConstantOffsets(); - DEBUG(dbgs() << "\n" - "After generating reuse formulae:\n"; - print_uses(dbgs())); + LLVM_DEBUG(dbgs() << "\n" + "After generating reuse formulae:\n"; + print_uses(dbgs())); } /// If there are multiple formulae with the same set of registers used @@ -4183,7 +4184,8 @@ void LSRInstance::FilterOutUndesirableDedicatedRegisters() { for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { LSRUse &LU = Uses[LUIdx]; - DEBUG(dbgs() << "Filtering for use "; LU.print(dbgs()); dbgs() << '\n'); + LLVM_DEBUG(dbgs() << "Filtering for use "; LU.print(dbgs()); + dbgs() << '\n'); bool Any = false; for (size_t FIdx = 0, NumForms = LU.Formulae.size(); @@ -4207,8 +4209,8 @@ void LSRInstance::FilterOutUndesirableDedicatedRegisters() { // as the basis of rediscovering the desired formula that uses an AddRec // corresponding to the existing phi. Once all formulae have been // generated, these initial losers may be pruned. - DEBUG(dbgs() << " Filtering loser "; F.print(dbgs()); - dbgs() << "\n"); + LLVM_DEBUG(dbgs() << " Filtering loser "; F.print(dbgs()); + dbgs() << "\n"); } else { SmallVector<const SCEV *, 4> Key; @@ -4235,10 +4237,10 @@ void LSRInstance::FilterOutUndesirableDedicatedRegisters() { CostBest.RateFormula(TTI, Best, Regs, VisitedRegs, L, SE, DT, LU); if (CostF.isLess(CostBest, TTI)) std::swap(F, Best); - DEBUG(dbgs() << " Filtering out formula "; F.print(dbgs()); - dbgs() << "\n" - " in favor of formula "; Best.print(dbgs()); - dbgs() << '\n'); + LLVM_DEBUG(dbgs() << " Filtering out formula "; F.print(dbgs()); + dbgs() << "\n" + " in favor of formula "; + Best.print(dbgs()); dbgs() << '\n'); } #ifndef NDEBUG ChangedFormulae = true; @@ -4257,11 +4259,11 @@ void LSRInstance::FilterOutUndesirableDedicatedRegisters() { BestFormulae.clear(); } - DEBUG(if (ChangedFormulae) { - dbgs() << "\n" - "After filtering out undesirable candidates:\n"; - print_uses(dbgs()); - }); + LLVM_DEBUG(if (ChangedFormulae) { + dbgs() << "\n" + "After filtering out undesirable candidates:\n"; + print_uses(dbgs()); + }); } // This is a rough guess that seems to work fairly well. @@ -4290,11 +4292,11 @@ size_t LSRInstance::EstimateSearchSpaceComplexity() const { /// register pressure); remove it to simplify the system. void LSRInstance::NarrowSearchSpaceByDetectingSupersets() { if (EstimateSearchSpaceComplexity() >= ComplexityLimit) { - DEBUG(dbgs() << "The search space is too complex.\n"); + LLVM_DEBUG(dbgs() << "The search space is too complex.\n"); - DEBUG(dbgs() << "Narrowing the search space by eliminating formulae " - "which use a superset of registers used by other " - "formulae.\n"); + LLVM_DEBUG(dbgs() << "Narrowing the search space by eliminating formulae " + "which use a superset of registers used by other " + "formulae.\n"); for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { LSRUse &LU = Uses[LUIdx]; @@ -4312,7 +4314,8 @@ void LSRInstance::NarrowSearchSpaceByDetectingSupersets() { NewF.BaseRegs.erase(NewF.BaseRegs.begin() + (I - F.BaseRegs.begin())); if (LU.HasFormulaWithSameRegs(NewF)) { - DEBUG(dbgs() << " Deleting "; F.print(dbgs()); dbgs() << '\n'); + LLVM_DEBUG(dbgs() << " Deleting "; F.print(dbgs()); + dbgs() << '\n'); LU.DeleteFormula(F); --i; --e; @@ -4327,8 +4330,8 @@ void LSRInstance::NarrowSearchSpaceByDetectingSupersets() { NewF.BaseRegs.erase(NewF.BaseRegs.begin() + (I - F.BaseRegs.begin())); if (LU.HasFormulaWithSameRegs(NewF)) { - DEBUG(dbgs() << " Deleting "; F.print(dbgs()); - dbgs() << '\n'); + LLVM_DEBUG(dbgs() << " Deleting "; F.print(dbgs()); + dbgs() << '\n'); LU.DeleteFormula(F); --i; --e; @@ -4343,8 +4346,7 @@ void LSRInstance::NarrowSearchSpaceByDetectingSupersets() { LU.RecomputeRegs(LUIdx, RegUses); } - DEBUG(dbgs() << "After pre-selection:\n"; - print_uses(dbgs())); + LLVM_DEBUG(dbgs() << "After pre-selection:\n"; print_uses(dbgs())); } } @@ -4354,9 +4356,10 @@ void LSRInstance::NarrowSearchSpaceByCollapsingUnrolledCode() { if (EstimateSearchSpaceComplexity() < ComplexityLimit) return; - DEBUG(dbgs() << "The search space is too complex.\n" - "Narrowing the search space by assuming that uses separated " - "by a constant offset will use the same registers.\n"); + LLVM_DEBUG( + dbgs() << "The search space is too complex.\n" + "Narrowing the search space by assuming that uses separated " + "by a constant offset will use the same registers.\n"); // This is especially useful for unrolled loops. @@ -4374,7 +4377,7 @@ void LSRInstance::NarrowSearchSpaceByCollapsingUnrolledCode() { LU.Kind, LU.AccessTy)) continue; - DEBUG(dbgs() << " Deleting use "; LU.print(dbgs()); dbgs() << '\n'); + LLVM_DEBUG(dbgs() << " Deleting use "; LU.print(dbgs()); dbgs() << '\n'); LUThatHas->AllFixupsOutsideLoop &= LU.AllFixupsOutsideLoop; @@ -4382,7 +4385,7 @@ void LSRInstance::NarrowSearchSpaceByCollapsingUnrolledCode() { for (LSRFixup &Fixup : LU.Fixups) { Fixup.Offset += F.BaseOffset; LUThatHas->pushFixup(Fixup); - DEBUG(dbgs() << "New fixup has offset " << Fixup.Offset << '\n'); + LLVM_DEBUG(dbgs() << "New fixup has offset " << Fixup.Offset << '\n'); } // Delete formulae from the new use which are no longer legal. @@ -4391,8 +4394,7 @@ void LSRInstance::NarrowSearchSpaceByCollapsingUnrolledCode() { Formula &F = LUThatHas->Formulae[i]; if (!isLegalUse(TTI, LUThatHas->MinOffset, LUThatHas->MaxOffset, LUThatHas->Kind, LUThatHas->AccessTy, F)) { - DEBUG(dbgs() << " Deleting "; F.print(dbgs()); - dbgs() << '\n'); + LLVM_DEBUG(dbgs() << " Deleting "; F.print(dbgs()); dbgs() << '\n'); LUThatHas->DeleteFormula(F); --i; --e; @@ -4411,7 +4413,7 @@ void LSRInstance::NarrowSearchSpaceByCollapsingUnrolledCode() { } } - DEBUG(dbgs() << "After pre-selection:\n"; print_uses(dbgs())); + LLVM_DEBUG(dbgs() << "After pre-selection:\n"; print_uses(dbgs())); } /// Call FilterOutUndesirableDedicatedRegisters again, if necessary, now that @@ -4419,15 +4421,14 @@ void LSRInstance::NarrowSearchSpaceByCollapsingUnrolledCode() { /// eliminate. void LSRInstance::NarrowSearchSpaceByRefilteringUndesirableDedicatedRegisters(){ if (EstimateSearchSpaceComplexity() >= ComplexityLimit) { - DEBUG(dbgs() << "The search space is too complex.\n"); + LLVM_DEBUG(dbgs() << "The search space is too complex.\n"); - DEBUG(dbgs() << "Narrowing the search space by re-filtering out " - "undesirable dedicated registers.\n"); + LLVM_DEBUG(dbgs() << "Narrowing the search space by re-filtering out " + "undesirable dedicated registers.\n"); FilterOutUndesirableDedicatedRegisters(); - DEBUG(dbgs() << "After pre-selection:\n"; - print_uses(dbgs())); + LLVM_DEBUG(dbgs() << "After pre-selection:\n"; print_uses(dbgs())); } } @@ -4444,9 +4445,10 @@ void LSRInstance::NarrowSearchSpaceByFilterFormulaWithSameScaledReg() { if (EstimateSearchSpaceComplexity() < ComplexityLimit) return; - DEBUG(dbgs() << "The search space is too complex.\n" - "Narrowing the search space by choosing the best Formula " - "from the Formulae with the same Scale and ScaledReg.\n"); + LLVM_DEBUG( + dbgs() << "The search space is too complex.\n" + "Narrowing the search space by choosing the best Formula " + "from the Formulae with the same Scale and ScaledReg.\n"); // Map the "Scale * ScaledReg" pair to the best formula of current LSRUse. using BestFormulaeTy = DenseMap<std::pair<const SCEV *, int64_t>, size_t>; @@ -4460,7 +4462,8 @@ void LSRInstance::NarrowSearchSpaceByFilterFormulaWithSameScaledReg() { for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { LSRUse &LU = Uses[LUIdx]; - DEBUG(dbgs() << "Filtering for use "; LU.print(dbgs()); dbgs() << '\n'); + LLVM_DEBUG(dbgs() << "Filtering for use "; LU.print(dbgs()); + dbgs() << '\n'); // Return true if Formula FA is better than Formula FB. auto IsBetterThan = [&](Formula &FA, Formula &FB) { @@ -4504,10 +4507,10 @@ void LSRInstance::NarrowSearchSpaceByFilterFormulaWithSameScaledReg() { Formula &Best = LU.Formulae[P.first->second]; if (IsBetterThan(F, Best)) std::swap(F, Best); - DEBUG(dbgs() << " Filtering out formula "; F.print(dbgs()); - dbgs() << "\n" - " in favor of formula "; - Best.print(dbgs()); dbgs() << '\n'); + LLVM_DEBUG(dbgs() << " Filtering out formula "; F.print(dbgs()); + dbgs() << "\n" + " in favor of formula "; + Best.print(dbgs()); dbgs() << '\n'); #ifndef NDEBUG ChangedFormulae = true; #endif @@ -4523,7 +4526,7 @@ void LSRInstance::NarrowSearchSpaceByFilterFormulaWithSameScaledReg() { BestFormulae.clear(); } - DEBUG(if (ChangedFormulae) { + LLVM_DEBUG(if (ChangedFormulae) { dbgs() << "\n" "After filtering out undesirable candidates:\n"; print_uses(dbgs()); @@ -4582,7 +4585,7 @@ void LSRInstance::NarrowSearchSpaceByDeletingCostlyFormulas() { // Used in each formula of a solution (in example above this is reg(c)). // We can skip them in calculations. SmallPtrSet<const SCEV *, 4> UniqRegs; - DEBUG(dbgs() << "The search space is too complex.\n"); + LLVM_DEBUG(dbgs() << "The search space is too complex.\n"); // Map each register to probability of not selecting DenseMap <const SCEV *, float> RegNumMap; @@ -4602,7 +4605,8 @@ void LSRInstance::NarrowSearchSpaceByDeletingCostlyFormulas() { RegNumMap.insert(std::make_pair(Reg, PNotSel)); } - DEBUG(dbgs() << "Narrowing the search space by deleting costly formulas\n"); + LLVM_DEBUG( + dbgs() << "Narrowing the search space by deleting costly formulas\n"); // Delete formulas where registers number expectation is high. for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { @@ -4644,26 +4648,25 @@ void LSRInstance::NarrowSearchSpaceByDeletingCostlyFormulas() { MinIdx = i; } } - DEBUG(dbgs() << " The formula "; LU.Formulae[MinIdx].print(dbgs()); - dbgs() << " with min reg num " << FMinRegNum << '\n'); + LLVM_DEBUG(dbgs() << " The formula "; LU.Formulae[MinIdx].print(dbgs()); + dbgs() << " with min reg num " << FMinRegNum << '\n'); if (MinIdx != 0) std::swap(LU.Formulae[MinIdx], LU.Formulae[0]); while (LU.Formulae.size() != 1) { - DEBUG(dbgs() << " Deleting "; LU.Formulae.back().print(dbgs()); - dbgs() << '\n'); + LLVM_DEBUG(dbgs() << " Deleting "; LU.Formulae.back().print(dbgs()); + dbgs() << '\n'); LU.Formulae.pop_back(); } LU.RecomputeRegs(LUIdx, RegUses); assert(LU.Formulae.size() == 1 && "Should be exactly 1 min regs formula"); Formula &F = LU.Formulae[0]; - DEBUG(dbgs() << " Leaving only "; F.print(dbgs()); dbgs() << '\n'); + LLVM_DEBUG(dbgs() << " Leaving only "; F.print(dbgs()); dbgs() << '\n'); // When we choose the formula, the regs become unique. UniqRegs.insert(F.BaseRegs.begin(), F.BaseRegs.end()); if (F.ScaledReg) UniqRegs.insert(F.ScaledReg); } - DEBUG(dbgs() << "After pre-selection:\n"; - print_uses(dbgs())); + LLVM_DEBUG(dbgs() << "After pre-selection:\n"; print_uses(dbgs())); } /// Pick a register which seems likely to be profitable, and then in any use @@ -4676,7 +4679,7 @@ void LSRInstance::NarrowSearchSpaceByPickingWinnerRegs() { while (EstimateSearchSpaceComplexity() >= ComplexityLimit) { // Ok, we have too many of formulae on our hands to conveniently handle. // Use a rough heuristic to thin out the list. - DEBUG(dbgs() << "The search space is too complex.\n"); + LLVM_DEBUG(dbgs() << "The search space is too complex.\n"); // Pick the register which is used by the most LSRUses, which is likely // to be a good reuse register candidate. @@ -4697,8 +4700,8 @@ void LSRInstance::NarrowSearchSpaceByPickingWinnerRegs() { } } - DEBUG(dbgs() << "Narrowing the search space by assuming " << *Best - << " will yield profitable reuse.\n"); + LLVM_DEBUG(dbgs() << "Narrowing the search space by assuming " << *Best + << " will yield profitable reuse.\n"); Taken.insert(Best); // In any use with formulae which references this register, delete formulae @@ -4711,7 +4714,7 @@ void LSRInstance::NarrowSearchSpaceByPickingWinnerRegs() { for (size_t i = 0, e = LU.Formulae.size(); i != e; ++i) { Formula &F = LU.Formulae[i]; if (!F.referencesReg(Best)) { - DEBUG(dbgs() << " Deleting "; F.print(dbgs()); dbgs() << '\n'); + LLVM_DEBUG(dbgs() << " Deleting "; F.print(dbgs()); dbgs() << '\n'); LU.DeleteFormula(F); --e; --i; @@ -4725,8 +4728,7 @@ void LSRInstance::NarrowSearchSpaceByPickingWinnerRegs() { LU.RecomputeRegs(LUIdx, RegUses); } - DEBUG(dbgs() << "After pre-selection:\n"; - print_uses(dbgs())); + LLVM_DEBUG(dbgs() << "After pre-selection:\n"; print_uses(dbgs())); } } @@ -4808,11 +4810,11 @@ void LSRInstance::SolveRecurse(SmallVectorImpl<const Formula *> &Solution, if (F.getNumRegs() == 1 && Workspace.size() == 1) VisitedRegs.insert(F.ScaledReg ? F.ScaledReg : F.BaseRegs[0]); } else { - DEBUG(dbgs() << "New best at "; NewCost.print(dbgs()); - dbgs() << ".\n Regs:"; - for (const SCEV *S : NewRegs) - dbgs() << ' ' << *S; - dbgs() << '\n'); + LLVM_DEBUG(dbgs() << "New best at "; NewCost.print(dbgs()); + dbgs() << ".\n Regs:"; for (const SCEV *S + : NewRegs) dbgs() + << ' ' << *S; + dbgs() << '\n'); SolutionCost = NewCost; Solution = Workspace; @@ -4837,22 +4839,22 @@ void LSRInstance::Solve(SmallVectorImpl<const Formula *> &Solution) const { SolveRecurse(Solution, SolutionCost, Workspace, CurCost, CurRegs, VisitedRegs); if (Solution.empty()) { - DEBUG(dbgs() << "\nNo Satisfactory Solution\n"); + LLVM_DEBUG(dbgs() << "\nNo Satisfactory Solution\n"); return; } // Ok, we've now made all our decisions. - DEBUG(dbgs() << "\n" - "The chosen solution requires "; SolutionCost.print(dbgs()); - dbgs() << ":\n"; - for (size_t i = 0, e = Uses.size(); i != e; ++i) { - dbgs() << " "; - Uses[i].print(dbgs()); - dbgs() << "\n" - " "; - Solution[i]->print(dbgs()); - dbgs() << '\n'; - }); + LLVM_DEBUG(dbgs() << "\n" + "The chosen solution requires "; + SolutionCost.print(dbgs()); dbgs() << ":\n"; + for (size_t i = 0, e = Uses.size(); i != e; ++i) { + dbgs() << " "; + Uses[i].print(dbgs()); + dbgs() << "\n" + " "; + Solution[i]->print(dbgs()); + dbgs() << '\n'; + }); assert(Solution.size() == Uses.size() && "Malformed solution!"); } @@ -5326,7 +5328,8 @@ LSRInstance::LSRInstance(Loop *L, IVUsers &IU, ScalarEvolution &SE, for (const IVStrideUse &U : IU) { if (++NumUsers > MaxIVUsers) { (void)U; - DEBUG(dbgs() << "LSR skipping loop, too many IV Users in " << U << "\n"); + LLVM_DEBUG(dbgs() << "LSR skipping loop, too many IV Users in " << U + << "\n"); return; } // Bail out if we have a PHI on an EHPad that gets a value from a @@ -5359,9 +5362,9 @@ LSRInstance::LSRInstance(Loop *L, IVUsers &IU, ScalarEvolution &SE, } #endif // DEBUG - DEBUG(dbgs() << "\nLSR on loop "; - L->getHeader()->printAsOperand(dbgs(), /*PrintType=*/false); - dbgs() << ":\n"); + LLVM_DEBUG(dbgs() << "\nLSR on loop "; + L->getHeader()->printAsOperand(dbgs(), /*PrintType=*/false); + dbgs() << ":\n"); // First, perform some low-level loop optimizations. OptimizeShadowIV(); @@ -5372,7 +5375,7 @@ LSRInstance::LSRInstance(Loop *L, IVUsers &IU, ScalarEvolution &SE, // Skip nested loops until we can model them better with formulae. if (!L->empty()) { - DEBUG(dbgs() << "LSR skipping outer loop " << *L << "\n"); + LLVM_DEBUG(dbgs() << "LSR skipping outer loop " << *L << "\n"); return; } @@ -5383,8 +5386,8 @@ LSRInstance::LSRInstance(Loop *L, IVUsers &IU, ScalarEvolution &SE, CollectLoopInvariantFixupsAndFormulae(); assert(!Uses.empty() && "IVUsers reported at least one use"); - DEBUG(dbgs() << "LSR found " << Uses.size() << " uses:\n"; - print_uses(dbgs())); + LLVM_DEBUG(dbgs() << "LSR found " << Uses.size() << " uses:\n"; + print_uses(dbgs())); // Now use the reuse data to generate a bunch of interesting ways // to formulate the values needed for the uses. diff --git a/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp b/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp index 822f880f222..bbd1d3baa92 100644 --- a/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp +++ b/llvm/lib/Transforms/Scalar/LoopUnrollPass.cpp @@ -406,9 +406,9 @@ static Optional<EstimatedUnrollCost> analyzeLoopUnrollCost( // First accumulate the cost of this instruction. if (!Cost.IsFree) { UnrolledCost += TTI.getUserCost(I); - DEBUG(dbgs() << "Adding cost of instruction (iteration " << Iteration - << "): "); - DEBUG(I->dump()); + LLVM_DEBUG(dbgs() << "Adding cost of instruction (iteration " + << Iteration << "): "); + LLVM_DEBUG(I->dump()); } // We must count the cost of every operand which is not free, @@ -443,14 +443,14 @@ static Optional<EstimatedUnrollCost> analyzeLoopUnrollCost( assert(L->isLCSSAForm(DT) && "Must have loops in LCSSA form to track live-out values."); - DEBUG(dbgs() << "Starting LoopUnroll profitability analysis...\n"); + LLVM_DEBUG(dbgs() << "Starting LoopUnroll profitability analysis...\n"); // Simulate execution of each iteration of the loop counting instructions, // which would be simplified. // Since the same load will take different values on different iterations, // we literally have to go through all loop's iterations. for (unsigned Iteration = 0; Iteration < TripCount; ++Iteration) { - DEBUG(dbgs() << " Analyzing iteration " << Iteration << "\n"); + LLVM_DEBUG(dbgs() << " Analyzing iteration " << Iteration << "\n"); // Prepare for the iteration by collecting any simplified entry or backedge // inputs. @@ -525,10 +525,10 @@ static Optional<EstimatedUnrollCost> analyzeLoopUnrollCost( // If unrolled body turns out to be too big, bail out. if (UnrolledCost > MaxUnrolledLoopSize) { - DEBUG(dbgs() << " Exceeded threshold.. exiting.\n" - << " UnrolledCost: " << UnrolledCost - << ", MaxUnrolledLoopSize: " << MaxUnrolledLoopSize - << "\n"); + LLVM_DEBUG(dbgs() << " Exceeded threshold.. exiting.\n" + << " UnrolledCost: " << UnrolledCost + << ", MaxUnrolledLoopSize: " << MaxUnrolledLoopSize + << "\n"); return None; } } @@ -581,8 +581,8 @@ static Optional<EstimatedUnrollCost> analyzeLoopUnrollCost( // If we found no optimization opportunities on the first iteration, we // won't find them on later ones too. if (UnrolledCost == RolledDynamicCost) { - DEBUG(dbgs() << " No opportunities found.. exiting.\n" - << " UnrolledCost: " << UnrolledCost << "\n"); + LLVM_DEBUG(dbgs() << " No opportunities found.. exiting.\n" + << " UnrolledCost: " << UnrolledCost << "\n"); return None; } } @@ -603,9 +603,9 @@ static Optional<EstimatedUnrollCost> analyzeLoopUnrollCost( } } - DEBUG(dbgs() << "Analysis finished:\n" - << "UnrolledCost: " << UnrolledCost << ", " - << "RolledDynamicCost: " << RolledDynamicCost << "\n"); + LLVM_DEBUG(dbgs() << "Analysis finished:\n" + << "UnrolledCost: " << UnrolledCost << ", " + << "RolledDynamicCost: " << RolledDynamicCost << "\n"); return {{UnrolledCost, RolledDynamicCost}}; } @@ -808,8 +808,8 @@ static bool computeUnrollCount( if (TripCount) { UP.Partial |= ExplicitUnroll; if (!UP.Partial) { - DEBUG(dbgs() << " will not try to unroll partially because " - << "-unroll-allow-partial not given\n"); + LLVM_DEBUG(dbgs() << " will not try to unroll partially because " + << "-unroll-allow-partial not given\n"); UP.Count = 0; return false; } @@ -896,8 +896,9 @@ static bool computeUnrollCount( // Reduce count based on the type of unrolling and the threshold values. UP.Runtime |= PragmaEnableUnroll || PragmaCount > 0 || UserUnrollCount; if (!UP.Runtime) { - DEBUG(dbgs() << " will not try to unroll loop with runtime trip count " - << "-unroll-runtime not given\n"); + LLVM_DEBUG( + dbgs() << " will not try to unroll loop with runtime trip count " + << "-unroll-runtime not given\n"); UP.Count = 0; return false; } @@ -917,12 +918,13 @@ static bool computeUnrollCount( if (!UP.AllowRemainder && UP.Count != 0 && (TripMultiple % UP.Count) != 0) { while (UP.Count != 0 && TripMultiple % UP.Count != 0) UP.Count >>= 1; - DEBUG(dbgs() << "Remainder loop is restricted (that could architecture " - "specific or because the loop contains a convergent " - "instruction), so unroll count must divide the trip " - "multiple, " - << TripMultiple << ". Reducing unroll count from " - << OrigCount << " to " << UP.Count << ".\n"); + LLVM_DEBUG( + dbgs() << "Remainder loop is restricted (that could architecture " + "specific or because the loop contains a convergent " + "instruction), so unroll count must divide the trip " + "multiple, " + << TripMultiple << ". Reducing unroll count from " << OrigCount + << " to " << UP.Count << ".\n"); using namespace ore; @@ -944,7 +946,8 @@ static bool computeUnrollCount( if (UP.Count > UP.MaxCount) UP.Count = UP.MaxCount; - DEBUG(dbgs() << " partially unrolling with count: " << UP.Count << "\n"); + LLVM_DEBUG(dbgs() << " partially unrolling with count: " << UP.Count + << "\n"); if (UP.Count < 2) UP.Count = 0; return ExplicitUnroll; @@ -957,12 +960,13 @@ static LoopUnrollResult tryToUnrollLoop( Optional<unsigned> ProvidedCount, Optional<unsigned> ProvidedThreshold, Optional<bool> ProvidedAllowPartial, Optional<bool> ProvidedRuntime, Optional<bool> ProvidedUpperBound, Optional<bool> ProvidedAllowPeeling) { - DEBUG(dbgs() << "Loop Unroll: F[" << L->getHeader()->getParent()->getName() - << "] Loop %" << L->getHeader()->getName() << "\n"); + LLVM_DEBUG(dbgs() << "Loop Unroll: F[" + << L->getHeader()->getParent()->getName() << "] Loop %" + << L->getHeader()->getName() << "\n"); if (HasUnrollDisablePragma(L)) return LoopUnrollResult::Unmodified; if (!L->isLoopSimplifyForm()) { - DEBUG( + LLVM_DEBUG( dbgs() << " Not unrolling loop which is not in loop-simplify form.\n"); return LoopUnrollResult::Unmodified; } @@ -984,14 +988,14 @@ static LoopUnrollResult tryToUnrollLoop( unsigned LoopSize = ApproximateLoopSize(L, NumInlineCandidates, NotDuplicatable, Convergent, TTI, EphValues, UP.BEInsns); - DEBUG(dbgs() << " Loop Size = " << LoopSize << "\n"); + LLVM_DEBUG(dbgs() << " Loop Size = " << LoopSize << "\n"); if (NotDuplicatable) { - DEBUG(dbgs() << " Not unrolling loop which contains non-duplicatable" - << " instructions.\n"); + LLVM_DEBUG(dbgs() << " Not unrolling loop which contains non-duplicatable" + << " instructions.\n"); return LoopUnrollResult::Unmodified; } if (NumInlineCandidates != 0) { - DEBUG(dbgs() << " Not unrolling loop with inlinable calls.\n"); + LLVM_DEBUG(dbgs() << " Not unrolling loop with inlinable calls.\n"); return LoopUnrollResult::Unmodified; } diff --git a/llvm/lib/Transforms/Scalar/LoopUnswitch.cpp b/llvm/lib/Transforms/Scalar/LoopUnswitch.cpp index 91a30088fa0..b530f7c5dbd 100644 --- a/llvm/lib/Transforms/Scalar/LoopUnswitch.cpp +++ b/llvm/lib/Transforms/Scalar/LoopUnswitch.cpp @@ -298,9 +298,9 @@ bool LUAnalysisCache::countLoop(const Loop *L, const TargetTransformInfo &TTI, MaxSize -= Props.SizeEstimation * Props.CanBeUnswitchedCount; if (Metrics.notDuplicatable) { - DEBUG(dbgs() << "NOT unswitching loop %" - << L->getHeader()->getName() << ", contents cannot be " - << "duplicated!\n"); + LLVM_DEBUG(dbgs() << "NOT unswitching loop %" << L->getHeader()->getName() + << ", contents cannot be " + << "duplicated!\n"); return false; } } @@ -856,20 +856,20 @@ bool LoopUnswitch::UnswitchIfProfitable(Value *LoopCond, Constant *Val, TerminatorInst *TI) { // Check to see if it would be profitable to unswitch current loop. if (!BranchesInfo.CostAllowsUnswitching()) { - DEBUG(dbgs() << "NOT unswitching loop %" - << currentLoop->getHeader()->getName() - << " at non-trivial condition '" << *Val - << "' == " << *LoopCond << "\n" - << ". Cost too high.\n"); + LLVM_DEBUG(dbgs() << "NOT unswitching loop %" + << currentLoop->getHeader()->getName() + << " at non-trivial condition '" << *Val + << "' == " << *LoopCond << "\n" + << ". Cost too high.\n"); return false; } if (hasBranchDivergence && getAnalysis<DivergenceAnalysis>().isDivergent(LoopCond)) { - DEBUG(dbgs() << "NOT unswitching loop %" - << currentLoop->getHeader()->getName() - << " at non-trivial condition '" << *Val - << "' == " << *LoopCond << "\n" - << ". Condition is divergent.\n"); + LLVM_DEBUG(dbgs() << "NOT unswitching loop %" + << currentLoop->getHeader()->getName() + << " at non-trivial condition '" << *Val + << "' == " << *LoopCond << "\n" + << ". Condition is divergent.\n"); return false; } @@ -970,11 +970,11 @@ void LoopUnswitch::EmitPreheaderBranchOnCondition(Value *LIC, Constant *Val, void LoopUnswitch::UnswitchTrivialCondition(Loop *L, Value *Cond, Constant *Val, BasicBlock *ExitBlock, TerminatorInst *TI) { - DEBUG(dbgs() << "loop-unswitch: Trivial-Unswitch loop %" - << loopHeader->getName() << " [" << L->getBlocks().size() - << " blocks] in Function " - << L->getHeader()->getParent()->getName() << " on cond: " << *Val - << " == " << *Cond << "\n"); + LLVM_DEBUG(dbgs() << "loop-unswitch: Trivial-Unswitch loop %" + << loopHeader->getName() << " [" << L->getBlocks().size() + << " blocks] in Function " + << L->getHeader()->getParent()->getName() + << " on cond: " << *Val << " == " << *Cond << "\n"); // First step, split the preheader, so that we know that there is a safe place // to insert the conditional branch. We will change loopPreheader to have a @@ -1196,10 +1196,10 @@ void LoopUnswitch::SplitExitEdges(Loop *L, void LoopUnswitch::UnswitchNontrivialCondition(Value *LIC, Constant *Val, Loop *L, TerminatorInst *TI) { Function *F = loopHeader->getParent(); - DEBUG(dbgs() << "loop-unswitch: Unswitching loop %" - << loopHeader->getName() << " [" << L->getBlocks().size() - << " blocks] in Function " << F->getName() - << " when '" << *Val << "' == " << *LIC << "\n"); + LLVM_DEBUG(dbgs() << "loop-unswitch: Unswitching loop %" + << loopHeader->getName() << " [" << L->getBlocks().size() + << " blocks] in Function " << F->getName() << " when '" + << *Val << "' == " << *LIC << "\n"); if (auto *SEWP = getAnalysisIfAvailable<ScalarEvolutionWrapperPass>()) SEWP->getSE().forgetLoop(L); @@ -1355,7 +1355,7 @@ static void RemoveFromWorklist(Instruction *I, static void ReplaceUsesOfWith(Instruction *I, Value *V, std::vector<Instruction*> &Worklist, Loop *L, LPPassManager *LPM) { - DEBUG(dbgs() << "Replace with '" << *V << "': " << *I << "\n"); + LLVM_DEBUG(dbgs() << "Replace with '" << *V << "': " << *I << "\n"); // Add uses to the worklist, which may be dead now. for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) @@ -1524,7 +1524,7 @@ void LoopUnswitch::SimplifyCode(std::vector<Instruction*> &Worklist, Loop *L) { // Simple DCE. if (isInstructionTriviallyDead(I)) { - DEBUG(dbgs() << "Remove dead instruction '" << *I << "\n"); + LLVM_DEBUG(dbgs() << "Remove dead instruction '" << *I << "\n"); // Add uses to the worklist, which may be dead now. for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) @@ -1557,8 +1557,8 @@ void LoopUnswitch::SimplifyCode(std::vector<Instruction*> &Worklist, Loop *L) { if (!SinglePred) continue; // Nothing to do. assert(SinglePred == Pred && "CFG broken"); - DEBUG(dbgs() << "Merging blocks: " << Pred->getName() << " <- " - << Succ->getName() << "\n"); + LLVM_DEBUG(dbgs() << "Merging blocks: " << Pred->getName() << " <- " + << Succ->getName() << "\n"); // Resolve any single entry PHI nodes in Succ. while (PHINode *PN = dyn_cast<PHINode>(Succ->begin())) diff --git a/llvm/lib/Transforms/Scalar/LoopVersioningLICM.cpp b/llvm/lib/Transforms/Scalar/LoopVersioningLICM.cpp index e0e2c1938aa..06e86081e8a 100644 --- a/llvm/lib/Transforms/Scalar/LoopVersioningLICM.cpp +++ b/llvm/lib/Transforms/Scalar/LoopVersioningLICM.cpp @@ -246,48 +246,47 @@ private: bool LoopVersioningLICM::legalLoopStructure() { // Loop must be in loop simplify form. if (!CurLoop->isLoopSimplifyForm()) { - DEBUG( - dbgs() << " loop is not in loop-simplify form.\n"); + LLVM_DEBUG(dbgs() << " loop is not in loop-simplify form.\n"); return false; } // Loop should be innermost loop, if not return false. if (!CurLoop->getSubLoops().empty()) { - DEBUG(dbgs() << " loop is not innermost\n"); + LLVM_DEBUG(dbgs() << " loop is not innermost\n"); return false; } // Loop should have a single backedge, if not return false. if (CurLoop->getNumBackEdges() != 1) { - DEBUG(dbgs() << " loop has multiple backedges\n"); + LLVM_DEBUG(dbgs() << " loop has multiple backedges\n"); return false; } // Loop must have a single exiting block, if not return false. if (!CurLoop->getExitingBlock()) { - DEBUG(dbgs() << " loop has multiple exiting block\n"); + LLVM_DEBUG(dbgs() << " loop has multiple exiting block\n"); return false; } // We only handle bottom-tested loop, i.e. loop in which the condition is // checked at the end of each iteration. With that we can assume that all // instructions in the loop are executed the same number of times. if (CurLoop->getExitingBlock() != CurLoop->getLoopLatch()) { - DEBUG(dbgs() << " loop is not bottom tested\n"); + LLVM_DEBUG(dbgs() << " loop is not bottom tested\n"); return false; } // Parallel loops must not have aliasing loop-invariant memory accesses. // Hence we don't need to version anything in this case. if (CurLoop->isAnnotatedParallel()) { - DEBUG(dbgs() << " Parallel loop is not worth versioning\n"); + LLVM_DEBUG(dbgs() << " Parallel loop is not worth versioning\n"); return false; } // Loop depth more then LoopDepthThreshold are not allowed if (CurLoop->getLoopDepth() > LoopDepthThreshold) { - DEBUG(dbgs() << " loop depth is more then threshold\n"); + LLVM_DEBUG(dbgs() << " loop depth is more then threshold\n"); return false; } // We need to be able to compute the loop trip count in order // to generate the bound checks. const SCEV *ExitCount = SE->getBackedgeTakenCount(CurLoop); if (ExitCount == SE->getCouldNotCompute()) { - DEBUG(dbgs() << " loop does not has trip count\n"); + LLVM_DEBUG(dbgs() << " loop does not has trip count\n"); return false; } return true; @@ -335,18 +334,18 @@ bool LoopVersioningLICM::legalLoopMemoryAccesses() { } // Ensure types should be of same type. if (!TypeSafety) { - DEBUG(dbgs() << " Alias tracker type safety failed!\n"); + LLVM_DEBUG(dbgs() << " Alias tracker type safety failed!\n"); return false; } // Ensure loop body shouldn't be read only. if (!HasMod) { - DEBUG(dbgs() << " No memory modified in loop body\n"); + LLVM_DEBUG(dbgs() << " No memory modified in loop body\n"); return false; } // Make sure alias set has may alias case. // If there no alias memory ambiguity, return false. if (!HasMayAlias) { - DEBUG(dbgs() << " No ambiguity in memory access.\n"); + LLVM_DEBUG(dbgs() << " No ambiguity in memory access.\n"); return false; } return true; @@ -362,12 +361,12 @@ bool LoopVersioningLICM::instructionSafeForVersioning(Instruction *I) { assert(I != nullptr && "Null instruction found!"); // Check function call safety if (isa<CallInst>(I) && !AA->doesNotAccessMemory(CallSite(I))) { - DEBUG(dbgs() << " Unsafe call site found.\n"); + LLVM_DEBUG(dbgs() << " Unsafe call site found.\n"); return false; } // Avoid loops with possiblity of throw if (I->mayThrow()) { - DEBUG(dbgs() << " May throw instruction found in loop body\n"); + LLVM_DEBUG(dbgs() << " May throw instruction found in loop body\n"); return false; } // If current instruction is load instructions @@ -375,7 +374,7 @@ bool LoopVersioningLICM::instructionSafeForVersioning(Instruction *I) { if (I->mayReadFromMemory()) { LoadInst *Ld = dyn_cast<LoadInst>(I); if (!Ld || !Ld->isSimple()) { - DEBUG(dbgs() << " Found a non-simple load.\n"); + LLVM_DEBUG(dbgs() << " Found a non-simple load.\n"); return false; } LoadAndStoreCounter++; @@ -389,7 +388,7 @@ bool LoopVersioningLICM::instructionSafeForVersioning(Instruction *I) { else if (I->mayWriteToMemory()) { StoreInst *St = dyn_cast<StoreInst>(I); if (!St || !St->isSimple()) { - DEBUG(dbgs() << " Found a non-simple store.\n"); + LLVM_DEBUG(dbgs() << " Found a non-simple store.\n"); return false; } LoadAndStoreCounter++; @@ -428,13 +427,14 @@ bool LoopVersioningLICM::legalLoopInstructions() { LAI = &LAA->getInfo(CurLoop); // Check LoopAccessInfo for need of runtime check. if (LAI->getRuntimePointerChecking()->getChecks().empty()) { - DEBUG(dbgs() << " LAA: Runtime check not found !!\n"); + LLVM_DEBUG(dbgs() << " LAA: Runtime check not found !!\n"); return false; } // Number of runtime-checks should be less then RuntimeMemoryCheckThreshold if (LAI->getNumRuntimePointerChecks() > VectorizerParams::RuntimeMemoryCheckThreshold) { - DEBUG(dbgs() << " LAA: Runtime checks are more than threshold !!\n"); + LLVM_DEBUG( + dbgs() << " LAA: Runtime checks are more than threshold !!\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "RuntimeCheck", CurLoop->getStartLoc(), @@ -448,23 +448,25 @@ bool LoopVersioningLICM::legalLoopInstructions() { } // Loop should have at least one invariant load or store instruction. if (!InvariantCounter) { - DEBUG(dbgs() << " Invariant not found !!\n"); + LLVM_DEBUG(dbgs() << " Invariant not found !!\n"); return false; } // Read only loop not allowed. if (IsReadOnlyLoop) { - DEBUG(dbgs() << " Found a read-only loop!\n"); + LLVM_DEBUG(dbgs() << " Found a read-only loop!\n"); return false; } // Profitablity check: // Check invariant threshold, should be in limit. if (InvariantCounter * 100 < InvariantThreshold * LoadAndStoreCounter) { - DEBUG(dbgs() - << " Invariant load & store are less then defined threshold\n"); - DEBUG(dbgs() << " Invariant loads & stores: " - << ((InvariantCounter * 100) / LoadAndStoreCounter) << "%\n"); - DEBUG(dbgs() << " Invariant loads & store threshold: " - << InvariantThreshold << "%\n"); + LLVM_DEBUG( + dbgs() + << " Invariant load & store are less then defined threshold\n"); + LLVM_DEBUG(dbgs() << " Invariant loads & stores: " + << ((InvariantCounter * 100) / LoadAndStoreCounter) + << "%\n"); + LLVM_DEBUG(dbgs() << " Invariant loads & store threshold: " + << InvariantThreshold << "%\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "InvariantThreshold", CurLoop->getStartLoc(), @@ -497,16 +499,16 @@ bool LoopVersioningLICM::isLoopAlreadyVisited() { /// Return true if legal else returns false. bool LoopVersioningLICM::isLegalForVersioning() { using namespace ore; - DEBUG(dbgs() << "Loop: " << *CurLoop); + LLVM_DEBUG(dbgs() << "Loop: " << *CurLoop); // Make sure not re-visiting same loop again. if (isLoopAlreadyVisited()) { - DEBUG( + LLVM_DEBUG( dbgs() << " Revisiting loop in LoopVersioningLICM not allowed.\n\n"); return false; } // Check loop structure leagality. if (!legalLoopStructure()) { - DEBUG( + LLVM_DEBUG( dbgs() << " Loop structure not suitable for LoopVersioningLICM\n\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "IllegalLoopStruct", @@ -518,14 +520,16 @@ bool LoopVersioningLICM::isLegalForVersioning() { } // Check loop instruction leagality. if (!legalLoopInstructions()) { - DEBUG(dbgs() - << " Loop instructions not suitable for LoopVersioningLICM\n\n"); + LLVM_DEBUG( + dbgs() + << " Loop instructions not suitable for LoopVersioningLICM\n\n"); return false; } // Check loop memory access leagality. if (!legalLoopMemoryAccesses()) { - DEBUG(dbgs() - << " Loop memory access not suitable for LoopVersioningLICM\n\n"); + LLVM_DEBUG( + dbgs() + << " Loop memory access not suitable for LoopVersioningLICM\n\n"); ORE->emit([&]() { return OptimizationRemarkMissed(DEBUG_TYPE, "IllegalLoopMemoryAccess", CurLoop->getStartLoc(), @@ -535,7 +539,7 @@ bool LoopVersioningLICM::isLegalForVersioning() { return false; } // Loop versioning is feasible, return true. - DEBUG(dbgs() << " Loop Versioning found to be beneficial\n\n"); + LLVM_DEBUG(dbgs() << " Loop Versioning found to be beneficial\n\n"); ORE->emit([&]() { return OptimizationRemark(DEBUG_TYPE, "IsLegalForVersioning", CurLoop->getStartLoc(), CurLoop->getHeader()) diff --git a/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp b/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp index e21d09e65a1..437c70745d3 100644 --- a/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp +++ b/llvm/lib/Transforms/Scalar/MemCpyOptimizer.cpp @@ -479,10 +479,10 @@ Instruction *MemCpyOptPass::tryMergingIntoMemset(Instruction *StartInst, AMemSet = Builder.CreateMemSet(StartPtr, ByteVal, Range.End-Range.Start, Alignment); - DEBUG(dbgs() << "Replace stores:\n"; - for (Instruction *SI : Range.TheStores) - dbgs() << *SI << '\n'; - dbgs() << "With: " << *AMemSet << '\n'); + LLVM_DEBUG(dbgs() << "Replace stores:\n"; for (Instruction *SI + : Range.TheStores) dbgs() + << *SI << '\n'; + dbgs() << "With: " << *AMemSet << '\n'); if (!Range.TheStores.empty()) AMemSet->setDebugLoc(Range.TheStores[0]->getDebugLoc()); @@ -603,7 +603,7 @@ static bool moveUp(AliasAnalysis &AA, StoreInst *SI, Instruction *P, // We made it, we need to lift for (auto *I : llvm::reverse(ToLift)) { - DEBUG(dbgs() << "Lifting " << *I << " before " << *P << "\n"); + LLVM_DEBUG(dbgs() << "Lifting " << *I << " before " << *P << "\n"); I->moveBefore(P); } @@ -680,8 +680,8 @@ bool MemCpyOptPass::processStore(StoreInst *SI, BasicBlock::iterator &BBI) { LI->getPointerOperand(), findLoadAlignment(DL, LI), Size, SI->isVolatile()); - DEBUG(dbgs() << "Promoting " << *LI << " to " << *SI - << " => " << *M << "\n"); + LLVM_DEBUG(dbgs() << "Promoting " << *LI << " to " << *SI << " => " + << *M << "\n"); MD->removeInstruction(SI); SI->eraseFromParent(); @@ -770,7 +770,7 @@ bool MemCpyOptPass::processStore(StoreInst *SI, BasicBlock::iterator &BBI) { auto *M = Builder.CreateMemSet(SI->getPointerOperand(), ByteVal, Size, Align, SI->isVolatile()); - DEBUG(dbgs() << "Promoting " << *SI << " to " << *M << "\n"); + LLVM_DEBUG(dbgs() << "Promoting " << *SI << " to " << *M << "\n"); MD->removeInstruction(SI); SI->eraseFromParent(); @@ -1294,8 +1294,8 @@ bool MemCpyOptPass::processMemMove(MemMoveInst *M) { MemoryLocation::getForSource(M))) return false; - DEBUG(dbgs() << "MemCpyOptPass: Optimizing memmove -> memcpy: " << *M - << "\n"); + LLVM_DEBUG(dbgs() << "MemCpyOptPass: Optimizing memmove -> memcpy: " << *M + << "\n"); // If not, then we know we can transform this. Type *ArgTys[3] = { M->getRawDest()->getType(), @@ -1377,9 +1377,9 @@ bool MemCpyOptPass::processByValArgument(CallSite CS, unsigned ArgNo) { TmpCast = new BitCastInst(MDep->getSource(), ByValArg->getType(), "tmpcast", CS.getInstruction()); - DEBUG(dbgs() << "MemCpyOptPass: Forwarding memcpy to byval:\n" - << " " << *MDep << "\n" - << " " << *CS.getInstruction() << "\n"); + LLVM_DEBUG(dbgs() << "MemCpyOptPass: Forwarding memcpy to byval:\n" + << " " << *MDep << "\n" + << " " << *CS.getInstruction() << "\n"); // Otherwise we're good! Update the byval argument. CS.setArgument(ArgNo, TmpCast); diff --git a/llvm/lib/Transforms/Scalar/MergeICmps.cpp b/llvm/lib/Transforms/Scalar/MergeICmps.cpp index bb1b8a53110..57dd2292dfb 100644 --- a/llvm/lib/Transforms/Scalar/MergeICmps.cpp +++ b/llvm/lib/Transforms/Scalar/MergeICmps.cpp @@ -76,25 +76,25 @@ struct BCEAtom { BCEAtom visitICmpLoadOperand(Value *const Val) { BCEAtom Result; if (auto *const LoadI = dyn_cast<LoadInst>(Val)) { - DEBUG(dbgs() << "load\n"); + LLVM_DEBUG(dbgs() << "load\n"); if (LoadI->isUsedOutsideOfBlock(LoadI->getParent())) { - DEBUG(dbgs() << "used outside of block\n"); + LLVM_DEBUG(dbgs() << "used outside of block\n"); return {}; } if (LoadI->isVolatile()) { - DEBUG(dbgs() << "volatile\n"); + LLVM_DEBUG(dbgs() << "volatile\n"); return {}; } Value *const Addr = LoadI->getOperand(0); if (auto *const GEP = dyn_cast<GetElementPtrInst>(Addr)) { - DEBUG(dbgs() << "GEP\n"); + LLVM_DEBUG(dbgs() << "GEP\n"); if (LoadI->isUsedOutsideOfBlock(LoadI->getParent())) { - DEBUG(dbgs() << "used outside of block\n"); + LLVM_DEBUG(dbgs() << "used outside of block\n"); return {}; } const auto &DL = GEP->getModule()->getDataLayout(); if (!isDereferenceablePointer(GEP, DL)) { - DEBUG(dbgs() << "not dereferenceable\n"); + LLVM_DEBUG(dbgs() << "not dereferenceable\n"); // We need to make sure that we can do comparison in any order, so we // require memory to be unconditionnally dereferencable. return {}; @@ -251,13 +251,13 @@ BCECmpBlock visitICmp(const ICmpInst *const CmpI, // If there are any other uses of the comparison, we cannot merge it with // other comparisons as we would create an orphan use of the value. if (!CmpI->hasOneUse()) { - DEBUG(dbgs() << "cmp has several uses\n"); + LLVM_DEBUG(dbgs() << "cmp has several uses\n"); return {}; } if (CmpI->getPredicate() == ExpectedPredicate) { - DEBUG(dbgs() << "cmp " - << (ExpectedPredicate == ICmpInst::ICMP_EQ ? "eq" : "ne") - << "\n"); + LLVM_DEBUG(dbgs() << "cmp " + << (ExpectedPredicate == ICmpInst::ICMP_EQ ? "eq" : "ne") + << "\n"); auto Lhs = visitICmpLoadOperand(CmpI->getOperand(0)); if (!Lhs.Base()) return {}; auto Rhs = visitICmpLoadOperand(CmpI->getOperand(1)); @@ -275,7 +275,7 @@ BCECmpBlock visitCmpBlock(Value *const Val, BasicBlock *const Block, if (Block->empty()) return {}; auto *const BranchI = dyn_cast<BranchInst>(Block->getTerminator()); if (!BranchI) return {}; - DEBUG(dbgs() << "branch\n"); + LLVM_DEBUG(dbgs() << "branch\n"); if (BranchI->isUnconditional()) { // In this case, we expect an incoming value which is the result of the // comparison. This is the last link in the chain of comparisons (note @@ -283,7 +283,7 @@ BCECmpBlock visitCmpBlock(Value *const Val, BasicBlock *const Block, // can be reordered). auto *const CmpI = dyn_cast<ICmpInst>(Val); if (!CmpI) return {}; - DEBUG(dbgs() << "icmp\n"); + LLVM_DEBUG(dbgs() << "icmp\n"); auto Result = visitICmp(CmpI, ICmpInst::ICMP_EQ); Result.CmpI = CmpI; Result.BranchI = BranchI; @@ -292,12 +292,12 @@ BCECmpBlock visitCmpBlock(Value *const Val, BasicBlock *const Block, // In this case, we expect a constant incoming value (the comparison is // chained). const auto *const Const = dyn_cast<ConstantInt>(Val); - DEBUG(dbgs() << "const\n"); + LLVM_DEBUG(dbgs() << "const\n"); if (!Const->isZero()) return {}; - DEBUG(dbgs() << "false\n"); + LLVM_DEBUG(dbgs() << "false\n"); auto *const CmpI = dyn_cast<ICmpInst>(BranchI->getCondition()); if (!CmpI) return {}; - DEBUG(dbgs() << "icmp\n"); + LLVM_DEBUG(dbgs() << "icmp\n"); assert(BranchI->getNumSuccessors() == 2 && "expecting a cond branch"); BasicBlock *const FalseBlock = BranchI->getSuccessor(1); auto Result = visitICmp( @@ -311,12 +311,13 @@ BCECmpBlock visitCmpBlock(Value *const Val, BasicBlock *const Block, static inline void enqueueBlock(std::vector<BCECmpBlock> &Comparisons, BCECmpBlock &Comparison) { - DEBUG(dbgs() << "Block '" << Comparison.BB->getName() << "': Found cmp of " - << Comparison.SizeBits() << " bits between " - << Comparison.Lhs().Base() << " + " << Comparison.Lhs().Offset - << " and " << Comparison.Rhs().Base() << " + " - << Comparison.Rhs().Offset << "\n"); - DEBUG(dbgs() << "\n"); + LLVM_DEBUG(dbgs() << "Block '" << Comparison.BB->getName() + << "': Found cmp of " << Comparison.SizeBits() + << " bits between " << Comparison.Lhs().Base() << " + " + << Comparison.Lhs().Offset << " and " + << Comparison.Rhs().Base() << " + " + << Comparison.Rhs().Offset << "\n"); + LLVM_DEBUG(dbgs() << "\n"); Comparisons.push_back(Comparison); } @@ -367,12 +368,12 @@ BCECmpChain::BCECmpChain(const std::vector<BasicBlock *> &Blocks, PHINode &Phi) Block, Phi.getParent()); Comparison.BB = Block; if (!Comparison.IsValid()) { - DEBUG(dbgs() << "chain with invalid BCECmpBlock, no merge.\n"); + LLVM_DEBUG(dbgs() << "chain with invalid BCECmpBlock, no merge.\n"); return; } if (Comparison.doesOtherWork()) { - DEBUG(dbgs() << "block '" << Comparison.BB->getName() - << "' does extra work besides compare\n"); + LLVM_DEBUG(dbgs() << "block '" << Comparison.BB->getName() + << "' does extra work besides compare\n"); if (Comparisons.empty()) { // This is the initial block in the chain, in case this block does other // work, we can try to split the block and move the irrelevant @@ -388,13 +389,15 @@ BCECmpChain::BCECmpChain(const std::vector<BasicBlock *> &Blocks, PHINode &Phi) // // NOTE: we only handle block with single predecessor for now. if (Comparison.canSplit()) { - DEBUG(dbgs() << "Split initial block '" << Comparison.BB->getName() - << "' that does extra work besides compare\n"); + LLVM_DEBUG(dbgs() + << "Split initial block '" << Comparison.BB->getName() + << "' that does extra work besides compare\n"); Comparison.RequireSplit = true; enqueueBlock(Comparisons, Comparison); } else { - DEBUG(dbgs() << "ignoring initial block '" << Comparison.BB->getName() - << "' that does extra work besides compare\n"); + LLVM_DEBUG(dbgs() + << "ignoring initial block '" << Comparison.BB->getName() + << "' that does extra work besides compare\n"); } continue; } @@ -428,7 +431,7 @@ BCECmpChain::BCECmpChain(const std::vector<BasicBlock *> &Blocks, PHINode &Phi) // It is possible we have no suitable comparison to merge. if (Comparisons.empty()) { - DEBUG(dbgs() << "chain with no BCE basic blocks, no merge\n"); + LLVM_DEBUG(dbgs() << "chain with no BCE basic blocks, no merge\n"); return; } EntryBlock_ = Comparisons[0].BB; @@ -549,7 +552,7 @@ void BCECmpChain::mergeComparisons(ArrayRef<BCECmpBlock> Comparisons, if (C != Comparisons.end()) C->split(EntryBlock_); - DEBUG(dbgs() << "Merging " << Comparisons.size() << " comparisons\n"); + LLVM_DEBUG(dbgs() << "Merging " << Comparisons.size() << " comparisons\n"); const auto TotalSize = std::accumulate(Comparisons.begin(), Comparisons.end(), 0, [](int Size, const BCECmpBlock &C) { @@ -594,17 +597,17 @@ void BCECmpChain::mergeComparisons(ArrayRef<BCECmpBlock> Comparisons, } else { assert(Comparisons.size() == 1); // There are no blocks to merge, but we still need to update the branches. - DEBUG(dbgs() << "Only one comparison, updating branches\n"); + LLVM_DEBUG(dbgs() << "Only one comparison, updating branches\n"); if (NextBBInChain) { if (FirstComparison.BranchI->isConditional()) { - DEBUG(dbgs() << "conditional -> conditional\n"); + LLVM_DEBUG(dbgs() << "conditional -> conditional\n"); // Just update the "true" target, the "false" target should already be // the phi block. assert(FirstComparison.BranchI->getSuccessor(1) == Phi.getParent()); FirstComparison.BranchI->setSuccessor(0, NextBBInChain); Phi.addIncoming(ConstantInt::getFalse(Context), BB); } else { - DEBUG(dbgs() << "unconditional -> conditional\n"); + LLVM_DEBUG(dbgs() << "unconditional -> conditional\n"); // Replace the unconditional branch by a conditional one. FirstComparison.BranchI->eraseFromParent(); IRBuilder<> Builder(BB); @@ -614,14 +617,14 @@ void BCECmpChain::mergeComparisons(ArrayRef<BCECmpBlock> Comparisons, } } else { if (FirstComparison.BranchI->isConditional()) { - DEBUG(dbgs() << "conditional -> unconditional\n"); + LLVM_DEBUG(dbgs() << "conditional -> unconditional\n"); // Replace the conditional branch by an unconditional one. FirstComparison.BranchI->eraseFromParent(); IRBuilder<> Builder(BB); Builder.CreateBr(Phi.getParent()); Phi.addIncoming(FirstComparison.CmpI, BB); } else { - DEBUG(dbgs() << "unconditional -> unconditional\n"); + LLVM_DEBUG(dbgs() << "unconditional -> unconditional\n"); Phi.addIncoming(FirstComparison.CmpI, BB); } } @@ -639,22 +642,22 @@ std::vector<BasicBlock *> getOrderedBlocks(PHINode &Phi, if (CurBlock->hasAddressTaken()) { // Somebody is jumping to the block through an address, all bets are // off. - DEBUG(dbgs() << "skip: block " << BlockIndex - << " has its address taken\n"); + LLVM_DEBUG(dbgs() << "skip: block " << BlockIndex + << " has its address taken\n"); return {}; } Blocks[BlockIndex] = CurBlock; auto *SinglePredecessor = CurBlock->getSinglePredecessor(); if (!SinglePredecessor) { // The block has two or more predecessors. - DEBUG(dbgs() << "skip: block " << BlockIndex - << " has two or more predecessors\n"); + LLVM_DEBUG(dbgs() << "skip: block " << BlockIndex + << " has two or more predecessors\n"); return {}; } if (Phi.getBasicBlockIndex(SinglePredecessor) < 0) { // The block does not link back to the phi. - DEBUG(dbgs() << "skip: block " << BlockIndex - << " does not link back to the phi\n"); + LLVM_DEBUG(dbgs() << "skip: block " << BlockIndex + << " does not link back to the phi\n"); return {}; } CurBlock = SinglePredecessor; @@ -664,9 +667,9 @@ std::vector<BasicBlock *> getOrderedBlocks(PHINode &Phi, } bool processPhi(PHINode &Phi, const TargetLibraryInfo *const TLI) { - DEBUG(dbgs() << "processPhi()\n"); + LLVM_DEBUG(dbgs() << "processPhi()\n"); if (Phi.getNumIncomingValues() <= 1) { - DEBUG(dbgs() << "skip: only one incoming value in phi\n"); + LLVM_DEBUG(dbgs() << "skip: only one incoming value in phi\n"); return false; } // We are looking for something that has the following structure: @@ -690,7 +693,7 @@ bool processPhi(PHINode &Phi, const TargetLibraryInfo *const TLI) { if (isa<ConstantInt>(Phi.getIncomingValue(I))) continue; if (LastBlock) { // There are several non-constant values. - DEBUG(dbgs() << "skip: several non-constant values\n"); + LLVM_DEBUG(dbgs() << "skip: several non-constant values\n"); return false; } if (!isa<ICmpInst>(Phi.getIncomingValue(I)) || @@ -701,7 +704,7 @@ bool processPhi(PHINode &Phi, const TargetLibraryInfo *const TLI) { // producing block more than once. // // This is an uncommon case, so we bail. - DEBUG( + LLVM_DEBUG( dbgs() << "skip: non-constant value not from cmp or not from last block.\n"); return false; @@ -710,11 +713,11 @@ bool processPhi(PHINode &Phi, const TargetLibraryInfo *const TLI) { } if (!LastBlock) { // There is no non-constant block. - DEBUG(dbgs() << "skip: no non-constant block\n"); + LLVM_DEBUG(dbgs() << "skip: no non-constant block\n"); return false; } if (LastBlock->getSingleSuccessor() != Phi.getParent()) { - DEBUG(dbgs() << "skip: last block non-phi successor\n"); + LLVM_DEBUG(dbgs() << "skip: last block non-phi successor\n"); return false; } @@ -724,7 +727,7 @@ bool processPhi(PHINode &Phi, const TargetLibraryInfo *const TLI) { BCECmpChain CmpChain(Blocks, Phi); if (CmpChain.size() < 2) { - DEBUG(dbgs() << "skip: only one compare block\n"); + LLVM_DEBUG(dbgs() << "skip: only one compare block\n"); return false; } @@ -759,7 +762,7 @@ class MergeICmps : public FunctionPass { PreservedAnalyses MergeICmps::runImpl(Function &F, const TargetLibraryInfo *TLI, const TargetTransformInfo *TTI) { - DEBUG(dbgs() << "MergeICmpsPass: " << F.getName() << "\n"); + LLVM_DEBUG(dbgs() << "MergeICmpsPass: " << F.getName() << "\n"); // We only try merging comparisons if the target wants to expand memcmp later. // The rationale is to avoid turning small chains into memcmp calls. diff --git a/llvm/lib/Transforms/Scalar/MergedLoadStoreMotion.cpp b/llvm/lib/Transforms/Scalar/MergedLoadStoreMotion.cpp index 07d45144758..72ab175deca 100644 --- a/llvm/lib/Transforms/Scalar/MergedLoadStoreMotion.cpp +++ b/llvm/lib/Transforms/Scalar/MergedLoadStoreMotion.cpp @@ -180,7 +180,7 @@ bool MergedLoadStoreMotion::isStoreSinkBarrierInRange(const Instruction &Start, /// StoreInst *MergedLoadStoreMotion::canSinkFromBlock(BasicBlock *BB1, StoreInst *Store0) { - DEBUG(dbgs() << "can Sink? : "; Store0->dump(); dbgs() << "\n"); + LLVM_DEBUG(dbgs() << "can Sink? : "; Store0->dump(); dbgs() << "\n"); BasicBlock *BB0 = Store0->getParent(); for (Instruction &Inst : reverse(*BB1)) { auto *Store1 = dyn_cast<StoreInst>(&Inst); @@ -229,9 +229,9 @@ bool MergedLoadStoreMotion::sinkStore(BasicBlock *BB, StoreInst *S0, if (A0 && A1 && A0->isIdenticalTo(A1) && A0->hasOneUse() && (A0->getParent() == S0->getParent()) && A1->hasOneUse() && (A1->getParent() == S1->getParent()) && isa<GetElementPtrInst>(A0)) { - DEBUG(dbgs() << "Sink Instruction into BB \n"; BB->dump(); - dbgs() << "Instruction Left\n"; S0->dump(); dbgs() << "\n"; - dbgs() << "Instruction Right\n"; S1->dump(); dbgs() << "\n"); + LLVM_DEBUG(dbgs() << "Sink Instruction into BB \n"; BB->dump(); + dbgs() << "Instruction Left\n"; S0->dump(); dbgs() << "\n"; + dbgs() << "Instruction Right\n"; S1->dump(); dbgs() << "\n"); // Hoist the instruction. BasicBlock::iterator InsertPt = BB->getFirstInsertionPt(); // Intersect optional metadata. @@ -313,7 +313,7 @@ bool MergedLoadStoreMotion::mergeStores(BasicBlock *T) { break; RBI = Pred0->rbegin(); RBE = Pred0->rend(); - DEBUG(dbgs() << "Search again\n"; Instruction *I = &*RBI; I->dump()); + LLVM_DEBUG(dbgs() << "Search again\n"; Instruction *I = &*RBI; I->dump()); } } return MergedStores; @@ -323,7 +323,7 @@ bool MergedLoadStoreMotion::run(Function &F, AliasAnalysis &AA) { this->AA = &AA; bool Changed = false; - DEBUG(dbgs() << "Instruction Merger\n"); + LLVM_DEBUG(dbgs() << "Instruction Merger\n"); // Merge unconditional branches, allowing PRE to catch more // optimization opportunities. diff --git a/llvm/lib/Transforms/Scalar/NewGVN.cpp b/llvm/lib/Transforms/Scalar/NewGVN.cpp index 369f59ec3c3..0cf9979b40a 100644 --- a/llvm/lib/Transforms/Scalar/NewGVN.cpp +++ b/llvm/lib/Transforms/Scalar/NewGVN.cpp @@ -221,13 +221,13 @@ private: Components.resize(Components.size() + 1); auto &Component = Components.back(); Component.insert(I); - DEBUG(dbgs() << "Component root is " << *I << "\n"); + LLVM_DEBUG(dbgs() << "Component root is " << *I << "\n"); InComponent.insert(I); ValueToComponent[I] = ComponentID; // Pop a component off the stack and label it. while (!Stack.empty() && Root.lookup(Stack.back()) >= OurDFS) { auto *Member = Stack.back(); - DEBUG(dbgs() << "Component member is " << *Member << "\n"); + LLVM_DEBUG(dbgs() << "Component member is " << *Member << "\n"); Component.insert(Member); InComponent.insert(Member); ValueToComponent[Member] = ComponentID; @@ -1068,8 +1068,8 @@ const Expression *NewGVN::checkSimplificationResults(Expression *E, return nullptr; if (auto *C = dyn_cast<Constant>(V)) { if (I) - DEBUG(dbgs() << "Simplified " << *I << " to " - << " constant " << *C << "\n"); + LLVM_DEBUG(dbgs() << "Simplified " << *I << " to " + << " constant " << *C << "\n"); NumGVNOpsSimplified++; assert(isa<BasicExpression>(E) && "We should always have had a basic expression here"); @@ -1077,8 +1077,8 @@ const Expression *NewGVN::checkSimplificationResults(Expression *E, return createConstantExpression(C); } else if (isa<Argument>(V) || isa<GlobalVariable>(V)) { if (I) - DEBUG(dbgs() << "Simplified " << *I << " to " - << " variable " << *V << "\n"); + LLVM_DEBUG(dbgs() << "Simplified " << *I << " to " + << " variable " << *V << "\n"); deleteExpression(E); return createVariableExpression(V); } @@ -1101,8 +1101,8 @@ const Expression *NewGVN::checkSimplificationResults(Expression *E, } if (I) - DEBUG(dbgs() << "Simplified " << *I << " to " - << " expression " << *CC->getDefiningExpr() << "\n"); + LLVM_DEBUG(dbgs() << "Simplified " << *I << " to " + << " expression " << *CC->getDefiningExpr() << "\n"); NumGVNOpsSimplified++; deleteExpression(E); return CC->getDefiningExpr(); @@ -1422,8 +1422,8 @@ NewGVN::performSymbolicLoadCoercion(Type *LoadType, Value *LoadPtr, if (Offset >= 0) { if (auto *C = dyn_cast<Constant>( lookupOperandLeader(DepSI->getValueOperand()))) { - DEBUG(dbgs() << "Coercing load from store " << *DepSI << " to constant " - << *C << "\n"); + LLVM_DEBUG(dbgs() << "Coercing load from store " << *DepSI + << " to constant " << *C << "\n"); return createConstantExpression( getConstantStoreValueForLoad(C, Offset, LoadType, DL)); } @@ -1438,8 +1438,8 @@ NewGVN::performSymbolicLoadCoercion(Type *LoadType, Value *LoadPtr, if (auto *C = dyn_cast<Constant>(lookupOperandLeader(DepLI))) if (auto *PossibleConstant = getConstantLoadValueForLoad(C, Offset, LoadType, DL)) { - DEBUG(dbgs() << "Coercing load from load " << *LI << " to constant " - << *PossibleConstant << "\n"); + LLVM_DEBUG(dbgs() << "Coercing load from load " << *LI + << " to constant " << *PossibleConstant << "\n"); return createConstantExpression(PossibleConstant); } } @@ -1448,8 +1448,8 @@ NewGVN::performSymbolicLoadCoercion(Type *LoadType, Value *LoadPtr, if (Offset >= 0) { if (auto *PossibleConstant = getConstantMemInstValueForLoad(DepMI, Offset, LoadType, DL)) { - DEBUG(dbgs() << "Coercing load from meminst " << *DepMI - << " to constant " << *PossibleConstant << "\n"); + LLVM_DEBUG(dbgs() << "Coercing load from meminst " << *DepMI + << " to constant " << *PossibleConstant << "\n"); return createConstantExpression(PossibleConstant); } } @@ -1530,7 +1530,7 @@ NewGVN::performSymbolicPredicateInfoEvaluation(Instruction *I) const { if (!PI) return nullptr; - DEBUG(dbgs() << "Found predicate info from instruction !\n"); + LLVM_DEBUG(dbgs() << "Found predicate info from instruction !\n"); auto *PWC = dyn_cast<PredicateWithCondition>(PI); if (!PWC) @@ -1570,7 +1570,7 @@ NewGVN::performSymbolicPredicateInfoEvaluation(Instruction *I) const { return nullptr; if (CopyOf != Cmp->getOperand(0) && CopyOf != Cmp->getOperand(1)) { - DEBUG(dbgs() << "Copy is not of any condition operands!\n"); + LLVM_DEBUG(dbgs() << "Copy is not of any condition operands!\n"); return nullptr; } Value *FirstOp = lookupOperandLeader(Cmp->getOperand(0)); @@ -1653,10 +1653,11 @@ bool NewGVN::setMemoryClass(const MemoryAccess *From, CongruenceClass *NewClass) { assert(NewClass && "Every MemoryAccess should be getting mapped to a non-null class"); - DEBUG(dbgs() << "Setting " << *From); - DEBUG(dbgs() << " equivalent to congruence class "); - DEBUG(dbgs() << NewClass->getID() << " with current MemoryAccess leader "); - DEBUG(dbgs() << *NewClass->getMemoryLeader() << "\n"); + LLVM_DEBUG(dbgs() << "Setting " << *From); + LLVM_DEBUG(dbgs() << " equivalent to congruence class "); + LLVM_DEBUG(dbgs() << NewClass->getID() + << " with current MemoryAccess leader "); + LLVM_DEBUG(dbgs() << *NewClass->getMemoryLeader() << "\n"); auto LookupResult = MemoryAccessToClass.find(From); bool Changed = false; @@ -1674,11 +1675,11 @@ bool NewGVN::setMemoryClass(const MemoryAccess *From, OldClass->setMemoryLeader(nullptr); } else { OldClass->setMemoryLeader(getNextMemoryLeader(OldClass)); - DEBUG(dbgs() << "Memory class leader change for class " - << OldClass->getID() << " to " - << *OldClass->getMemoryLeader() - << " due to removal of a memory member " << *From - << "\n"); + LLVM_DEBUG(dbgs() << "Memory class leader change for class " + << OldClass->getID() << " to " + << *OldClass->getMemoryLeader() + << " due to removal of a memory member " << *From + << "\n"); markMemoryLeaderChangeTouched(OldClass); } } @@ -1754,12 +1755,13 @@ NewGVN::performSymbolicPHIEvaluation(ArrayRef<ValPair> PHIOps, // If it has undef at this point, it means there are no-non-undef arguments, // and thus, the value of the phi node must be undef. if (HasUndef) { - DEBUG(dbgs() << "PHI Node " << *I - << " has no non-undef arguments, valuing it as undef\n"); + LLVM_DEBUG( + dbgs() << "PHI Node " << *I + << " has no non-undef arguments, valuing it as undef\n"); return createConstantExpression(UndefValue::get(I->getType())); } - DEBUG(dbgs() << "No arguments of PHI node " << *I << " are live\n"); + LLVM_DEBUG(dbgs() << "No arguments of PHI node " << *I << " are live\n"); deleteExpression(E); return createDeadExpression(); } @@ -1798,8 +1800,8 @@ NewGVN::performSymbolicPHIEvaluation(ArrayRef<ValPair> PHIOps, InstrToDFSNum(AllSameValue) > InstrToDFSNum(I)) return E; NumGVNPhisAllSame++; - DEBUG(dbgs() << "Simplified PHI node " << *I << " to " << *AllSameValue - << "\n"); + LLVM_DEBUG(dbgs() << "Simplified PHI node " << *I << " to " << *AllSameValue + << "\n"); deleteExpression(E); return createVariableOrConstant(AllSameValue); } @@ -2092,7 +2094,7 @@ void NewGVN::markUsersTouched(Value *V) { } void NewGVN::addMemoryUsers(const MemoryAccess *To, MemoryAccess *U) const { - DEBUG(dbgs() << "Adding memory user " << *U << " to " << *To << "\n"); + LLVM_DEBUG(dbgs() << "Adding memory user " << *U << " to " << *To << "\n"); MemoryToUsers[To].insert(U); } @@ -2228,8 +2230,9 @@ void NewGVN::moveMemoryToNewCongruenceClass(Instruction *I, (isa<StoreInst>(I) && NewClass->getStoreCount() == 1)); NewClass->setMemoryLeader(InstMA); // Mark it touched if we didn't just create a singleton - DEBUG(dbgs() << "Memory class leader change for class " << NewClass->getID() - << " due to new memory instruction becoming leader\n"); + LLVM_DEBUG(dbgs() << "Memory class leader change for class " + << NewClass->getID() + << " due to new memory instruction becoming leader\n"); markMemoryLeaderChangeTouched(NewClass); } setMemoryClass(InstMA, NewClass); @@ -2237,10 +2240,10 @@ void NewGVN::moveMemoryToNewCongruenceClass(Instruction *I, if (OldClass->getMemoryLeader() == InstMA) { if (!OldClass->definesNoMemory()) { OldClass->setMemoryLeader(getNextMemoryLeader(OldClass)); - DEBUG(dbgs() << "Memory class leader change for class " - << OldClass->getID() << " to " - << *OldClass->getMemoryLeader() - << " due to removal of old leader " << *InstMA << "\n"); + LLVM_DEBUG(dbgs() << "Memory class leader change for class " + << OldClass->getID() << " to " + << *OldClass->getMemoryLeader() + << " due to removal of old leader " << *InstMA << "\n"); markMemoryLeaderChangeTouched(OldClass); } else OldClass->setMemoryLeader(nullptr); @@ -2277,9 +2280,10 @@ void NewGVN::moveValueToNewCongruenceClass(Instruction *I, const Expression *E, NewClass->setStoredValue(SE->getStoredValue()); markValueLeaderChangeTouched(NewClass); // Shift the new class leader to be the store - DEBUG(dbgs() << "Changing leader of congruence class " - << NewClass->getID() << " from " << *NewClass->getLeader() - << " to " << *SI << " because store joined class\n"); + LLVM_DEBUG(dbgs() << "Changing leader of congruence class " + << NewClass->getID() << " from " + << *NewClass->getLeader() << " to " << *SI + << " because store joined class\n"); // If we changed the leader, we have to mark it changed because we don't // know what it will do to symbolic evaluation. NewClass->setLeader(SI); @@ -2299,8 +2303,8 @@ void NewGVN::moveValueToNewCongruenceClass(Instruction *I, const Expression *E, // See if we destroyed the class or need to swap leaders. if (OldClass->empty() && OldClass != TOPClass) { if (OldClass->getDefiningExpr()) { - DEBUG(dbgs() << "Erasing expression " << *OldClass->getDefiningExpr() - << " from table\n"); + LLVM_DEBUG(dbgs() << "Erasing expression " << *OldClass->getDefiningExpr() + << " from table\n"); // We erase it as an exact expression to make sure we don't just erase an // equivalent one. auto Iter = ExpressionToClass.find_as( @@ -2317,8 +2321,8 @@ void NewGVN::moveValueToNewCongruenceClass(Instruction *I, const Expression *E, // When the leader changes, the value numbering of // everything may change due to symbolization changes, so we need to // reprocess. - DEBUG(dbgs() << "Value class leader change for class " << OldClass->getID() - << "\n"); + LLVM_DEBUG(dbgs() << "Value class leader change for class " + << OldClass->getID() << "\n"); ++NumGVNLeaderChanges; // Destroy the stored value if there are no more stores to represent it. // Note that this is basically clean up for the expression removal that @@ -2381,12 +2385,14 @@ void NewGVN::performCongruenceFinding(Instruction *I, const Expression *E) { "VariableExpression should have been handled already"); EClass = NewClass; - DEBUG(dbgs() << "Created new congruence class for " << *I - << " using expression " << *E << " at " << NewClass->getID() - << " and leader " << *(NewClass->getLeader())); + LLVM_DEBUG(dbgs() << "Created new congruence class for " << *I + << " using expression " << *E << " at " + << NewClass->getID() << " and leader " + << *(NewClass->getLeader())); if (NewClass->getStoredValue()) - DEBUG(dbgs() << " and stored value " << *(NewClass->getStoredValue())); - DEBUG(dbgs() << "\n"); + LLVM_DEBUG(dbgs() << " and stored value " + << *(NewClass->getStoredValue())); + LLVM_DEBUG(dbgs() << "\n"); } else { EClass = lookupResult.first->second; if (isa<ConstantExpression>(E)) @@ -2404,8 +2410,8 @@ void NewGVN::performCongruenceFinding(Instruction *I, const Expression *E) { bool ClassChanged = IClass != EClass; bool LeaderChanged = LeaderChanges.erase(I); if (ClassChanged || LeaderChanged) { - DEBUG(dbgs() << "New class " << EClass->getID() << " for expression " << *E - << "\n"); + LLVM_DEBUG(dbgs() << "New class " << EClass->getID() << " for expression " + << *E << "\n"); if (ClassChanged) { moveValueToNewCongruenceClass(I, E, IClass, EClass); markPhiOfOpsChanged(E); @@ -2443,13 +2449,15 @@ void NewGVN::updateReachableEdge(BasicBlock *From, BasicBlock *To) { if (ReachableEdges.insert({From, To}).second) { // If this block wasn't reachable before, all instructions are touched. if (ReachableBlocks.insert(To).second) { - DEBUG(dbgs() << "Block " << getBlockName(To) << " marked reachable\n"); + LLVM_DEBUG(dbgs() << "Block " << getBlockName(To) + << " marked reachable\n"); const auto &InstRange = BlockInstRange.lookup(To); TouchedInstructions.set(InstRange.first, InstRange.second); } else { - DEBUG(dbgs() << "Block " << getBlockName(To) - << " was reachable, but new edge {" << getBlockName(From) - << "," << getBlockName(To) << "} to it found\n"); + LLVM_DEBUG(dbgs() << "Block " << getBlockName(To) + << " was reachable, but new edge {" + << getBlockName(From) << "," << getBlockName(To) + << "} to it found\n"); // We've made an edge reachable to an existing block, which may // impact predicates. Otherwise, only mark the phi nodes as touched, as @@ -2496,12 +2504,12 @@ void NewGVN::processOutgoingEdges(TerminatorInst *TI, BasicBlock *B) { BasicBlock *FalseSucc = BR->getSuccessor(1); if (CondEvaluated && (CI = dyn_cast<ConstantInt>(CondEvaluated))) { if (CI->isOne()) { - DEBUG(dbgs() << "Condition for Terminator " << *TI - << " evaluated to true\n"); + LLVM_DEBUG(dbgs() << "Condition for Terminator " << *TI + << " evaluated to true\n"); updateReachableEdge(B, TrueSucc); } else if (CI->isZero()) { - DEBUG(dbgs() << "Condition for Terminator " << *TI - << " evaluated to false\n"); + LLVM_DEBUG(dbgs() << "Condition for Terminator " << *TI + << " evaluated to false\n"); updateReachableEdge(B, FalseSucc); } } else { @@ -2686,8 +2694,8 @@ Value *NewGVN::findLeaderForInst(Instruction *TransInst, auto *FoundVal = findPHIOfOpsLeader(E, OrigInst, PredBB); if (!FoundVal) { ExpressionToPhiOfOps[E].insert(OrigInst); - DEBUG(dbgs() << "Cannot find phi of ops operand for " << *TransInst - << " in block " << getBlockName(PredBB) << "\n"); + LLVM_DEBUG(dbgs() << "Cannot find phi of ops operand for " << *TransInst + << " in block " << getBlockName(PredBB) << "\n"); return nullptr; } if (auto *SI = dyn_cast<StoreInst>(FoundVal)) @@ -2736,15 +2744,16 @@ NewGVN::makePossiblePHIOfOps(Instruction *I, auto *ValuePHI = RealToTemp.lookup(Op); if (!ValuePHI) continue; - DEBUG(dbgs() << "Found possible dependent phi of ops\n"); + LLVM_DEBUG(dbgs() << "Found possible dependent phi of ops\n"); Op = ValuePHI; } OpPHI = cast<PHINode>(Op); if (!SamePHIBlock) { SamePHIBlock = getBlockForValue(OpPHI); } else if (SamePHIBlock != getBlockForValue(OpPHI)) { - DEBUG(dbgs() - << "PHIs for operands are not all in the same block, aborting\n"); + LLVM_DEBUG( + dbgs() + << "PHIs for operands are not all in the same block, aborting\n"); return nullptr; } // No point in doing this for one-operand phis. @@ -2812,25 +2821,26 @@ NewGVN::makePossiblePHIOfOps(Instruction *I, } Deps.insert(CurrentDeps.begin(), CurrentDeps.end()); } else { - DEBUG(dbgs() << "Skipping phi of ops operand for incoming block " - << getBlockName(PredBB) - << " because the block is unreachable\n"); + LLVM_DEBUG(dbgs() << "Skipping phi of ops operand for incoming block " + << getBlockName(PredBB) + << " because the block is unreachable\n"); FoundVal = UndefValue::get(I->getType()); RevisitOnReachabilityChange[PHIBlock].set(InstrToDFSNum(I)); } PHIOps.push_back({FoundVal, PredBB}); - DEBUG(dbgs() << "Found phi of ops operand " << *FoundVal << " in " - << getBlockName(PredBB) << "\n"); + LLVM_DEBUG(dbgs() << "Found phi of ops operand " << *FoundVal << " in " + << getBlockName(PredBB) << "\n"); } for (auto Dep : Deps) addAdditionalUsers(Dep, I); sortPHIOps(PHIOps); auto *E = performSymbolicPHIEvaluation(PHIOps, I, PHIBlock); if (isa<ConstantExpression>(E) || isa<VariableExpression>(E)) { - DEBUG(dbgs() - << "Not creating real PHI of ops because it simplified to existing " - "value or constant\n"); + LLVM_DEBUG( + dbgs() + << "Not creating real PHI of ops because it simplified to existing " + "value or constant\n"); return E; } auto *ValuePHI = RealToTemp.lookup(I); @@ -2855,7 +2865,8 @@ NewGVN::makePossiblePHIOfOps(Instruction *I, } } RevisitOnReachabilityChange[PHIBlock].set(InstrToDFSNum(I)); - DEBUG(dbgs() << "Created phi of ops " << *ValuePHI << " for " << *I << "\n"); + LLVM_DEBUG(dbgs() << "Created phi of ops " << *ValuePHI << " for " << *I + << "\n"); return E; } @@ -2927,8 +2938,9 @@ void NewGVN::initializeCongruenceClasses(Function &F) { void NewGVN::cleanupTables() { for (unsigned i = 0, e = CongruenceClasses.size(); i != e; ++i) { - DEBUG(dbgs() << "Congruence class " << CongruenceClasses[i]->getID() - << " has " << CongruenceClasses[i]->size() << " members\n"); + LLVM_DEBUG(dbgs() << "Congruence class " << CongruenceClasses[i]->getID() + << " has " << CongruenceClasses[i]->size() + << " members\n"); // Make sure we delete the congruence class (probably worth switching to // a unique_ptr at some point. delete CongruenceClasses[i]; @@ -2998,7 +3010,7 @@ std::pair<unsigned, unsigned> NewGVN::assignDFSNumbers(BasicBlock *B, // we change its DFS number so that it doesn't get value numbered. if (isInstructionTriviallyDead(&I, TLI)) { InstrDFS[&I] = 0; - DEBUG(dbgs() << "Skipping trivially dead instruction " << I << "\n"); + LLVM_DEBUG(dbgs() << "Skipping trivially dead instruction " << I << "\n"); markInstructionForDeletion(&I); continue; } @@ -3064,9 +3076,10 @@ void NewGVN::valueNumberMemoryPhi(MemoryPhi *MP) { [&AllSameValue](const MemoryAccess *V) { return V == AllSameValue; }); if (AllEqual) - DEBUG(dbgs() << "Memory Phi value numbered to " << *AllSameValue << "\n"); + LLVM_DEBUG(dbgs() << "Memory Phi value numbered to " << *AllSameValue + << "\n"); else - DEBUG(dbgs() << "Memory Phi value numbered to itself\n"); + LLVM_DEBUG(dbgs() << "Memory Phi value numbered to itself\n"); // If it's equal to something, it's in that class. Otherwise, it has to be in // a class where it is the leader (other things may be equivalent to it, but // it needs to start off in its own class, which means it must have been the @@ -3085,7 +3098,7 @@ void NewGVN::valueNumberMemoryPhi(MemoryPhi *MP) { // Value number a single instruction, symbolically evaluating, performing // congruence finding, and updating mappings. void NewGVN::valueNumberInstruction(Instruction *I) { - DEBUG(dbgs() << "Processing instruction " << *I << "\n"); + LLVM_DEBUG(dbgs() << "Processing instruction " << *I << "\n"); if (!I->isTerminator()) { const Expression *Symbolized = nullptr; SmallPtrSet<Value *, 2> Visited; @@ -3271,7 +3284,7 @@ void NewGVN::verifyMemoryCongruency() const { // and redoing the iteration to see if anything changed. void NewGVN::verifyIterationSettled(Function &F) { #ifndef NDEBUG - DEBUG(dbgs() << "Beginning iteration verification\n"); + LLVM_DEBUG(dbgs() << "Beginning iteration verification\n"); if (DebugCounter::isCounterSet(VNCounter)) DebugCounter::setCounterValue(VNCounter, StartingVNCounter); @@ -3389,9 +3402,9 @@ void NewGVN::iterateTouchedInstructions() { // If it's not reachable, erase any touched instructions and move on. if (!BlockReachable) { TouchedInstructions.reset(CurrInstRange.first, CurrInstRange.second); - DEBUG(dbgs() << "Skipping instructions in block " - << getBlockName(CurrBlock) - << " because it is unreachable\n"); + LLVM_DEBUG(dbgs() << "Skipping instructions in block " + << getBlockName(CurrBlock) + << " because it is unreachable\n"); continue; } updateProcessedCount(CurrBlock); @@ -3401,7 +3414,7 @@ void NewGVN::iterateTouchedInstructions() { TouchedInstructions.reset(InstrNum); if (auto *MP = dyn_cast<MemoryPhi>(V)) { - DEBUG(dbgs() << "Processing MemoryPhi " << *MP << "\n"); + LLVM_DEBUG(dbgs() << "Processing MemoryPhi " << *MP << "\n"); valueNumberMemoryPhi(MP); } else if (auto *I = dyn_cast<Instruction>(V)) { valueNumberInstruction(I); @@ -3471,8 +3484,8 @@ bool NewGVN::runGVN() { // Initialize the touched instructions to include the entry block. const auto &InstRange = BlockInstRange.lookup(&F.getEntryBlock()); TouchedInstructions.set(InstRange.first, InstRange.second); - DEBUG(dbgs() << "Block " << getBlockName(&F.getEntryBlock()) - << " marked reachable\n"); + LLVM_DEBUG(dbgs() << "Block " << getBlockName(&F.getEntryBlock()) + << " marked reachable\n"); ReachableBlocks.insert(&F.getEntryBlock()); iterateTouchedInstructions(); @@ -3497,8 +3510,8 @@ bool NewGVN::runGVN() { }; for (auto &BB : make_filter_range(F, UnreachableBlockPred)) { - DEBUG(dbgs() << "We believe block " << getBlockName(&BB) - << " is unreachable\n"); + LLVM_DEBUG(dbgs() << "We believe block " << getBlockName(&BB) + << " is unreachable\n"); deleteInstructionsInBlock(&BB); Changed = true; } @@ -3720,7 +3733,7 @@ static void patchAndReplaceAllUsesWith(Instruction *I, Value *Repl) { } void NewGVN::deleteInstructionsInBlock(BasicBlock *BB) { - DEBUG(dbgs() << " BasicBlock Dead:" << *BB); + LLVM_DEBUG(dbgs() << " BasicBlock Dead:" << *BB); ++NumGVNBlocksDeleted; // Delete the instructions backwards, as it has a reduced likelihood of having @@ -3747,12 +3760,12 @@ void NewGVN::deleteInstructionsInBlock(BasicBlock *BB) { } void NewGVN::markInstructionForDeletion(Instruction *I) { - DEBUG(dbgs() << "Marking " << *I << " for deletion\n"); + LLVM_DEBUG(dbgs() << "Marking " << *I << " for deletion\n"); InstructionsToErase.insert(I); } void NewGVN::replaceInstruction(Instruction *I, Value *V) { - DEBUG(dbgs() << "Replacing " << *I << " with " << *V << "\n"); + LLVM_DEBUG(dbgs() << "Replacing " << *I << " with " << *V << "\n"); patchAndReplaceAllUsesWith(I, V); // We save the actual erasing to avoid invalidating memory // dependencies until we are done with everything. @@ -3878,9 +3891,10 @@ bool NewGVN::eliminateInstructions(Function &F) { auto ReplaceUnreachablePHIArgs = [&](PHINode *PHI, BasicBlock *BB) { for (auto &Operand : PHI->incoming_values()) if (!ReachableEdges.count({PHI->getIncomingBlock(Operand), BB})) { - DEBUG(dbgs() << "Replacing incoming value of " << PHI << " for block " - << getBlockName(PHI->getIncomingBlock(Operand)) - << " with undef due to it being unreachable\n"); + LLVM_DEBUG(dbgs() << "Replacing incoming value of " << PHI + << " for block " + << getBlockName(PHI->getIncomingBlock(Operand)) + << " with undef due to it being unreachable\n"); Operand.set(UndefValue::get(PHI->getType())); } }; @@ -3912,7 +3926,8 @@ bool NewGVN::eliminateInstructions(Function &F) { // Map to store the use counts DenseMap<const Value *, unsigned int> UseCounts; for (auto *CC : reverse(CongruenceClasses)) { - DEBUG(dbgs() << "Eliminating in congruence class " << CC->getID() << "\n"); + LLVM_DEBUG(dbgs() << "Eliminating in congruence class " << CC->getID() + << "\n"); // Track the equivalent store info so we can decide whether to try // dead store elimination. SmallVector<ValueDFS, 8> PossibleDeadStores; @@ -3950,8 +3965,8 @@ bool NewGVN::eliminateInstructions(Function &F) { MembersLeft.insert(Member); continue; } - DEBUG(dbgs() << "Found replacement " << *(Leader) << " for " << *Member - << "\n"); + LLVM_DEBUG(dbgs() << "Found replacement " << *(Leader) << " for " + << *Member << "\n"); auto *I = cast<Instruction>(Member); assert(Leader != I && "About to accidentally remove our leader"); replaceInstruction(I, Leader); @@ -3991,24 +4006,24 @@ bool NewGVN::eliminateInstructions(Function &F) { // remove from temp instruction list. AllTempInstructions.erase(PN); auto *DefBlock = getBlockForValue(Def); - DEBUG(dbgs() << "Inserting fully real phi of ops" << *Def - << " into block " - << getBlockName(getBlockForValue(Def)) << "\n"); + LLVM_DEBUG(dbgs() << "Inserting fully real phi of ops" << *Def + << " into block " + << getBlockName(getBlockForValue(Def)) << "\n"); PN->insertBefore(&DefBlock->front()); Def = PN; NumGVNPHIOfOpsEliminations++; } if (EliminationStack.empty()) { - DEBUG(dbgs() << "Elimination Stack is empty\n"); + LLVM_DEBUG(dbgs() << "Elimination Stack is empty\n"); } else { - DEBUG(dbgs() << "Elimination Stack Top DFS numbers are (" - << EliminationStack.dfs_back().first << "," - << EliminationStack.dfs_back().second << ")\n"); + LLVM_DEBUG(dbgs() << "Elimination Stack Top DFS numbers are (" + << EliminationStack.dfs_back().first << "," + << EliminationStack.dfs_back().second << ")\n"); } - DEBUG(dbgs() << "Current DFS numbers are (" << MemberDFSIn << "," - << MemberDFSOut << ")\n"); + LLVM_DEBUG(dbgs() << "Current DFS numbers are (" << MemberDFSIn << "," + << MemberDFSOut << ")\n"); // First, we see if we are out of scope or empty. If so, // and there equivalences, we try to replace the top of // stack with equivalences (if it's on the stack, it must @@ -4090,8 +4105,9 @@ bool NewGVN::eliminateInstructions(Function &F) { // Don't replace our existing users with ourselves. if (U->get() == DominatingLeader) continue; - DEBUG(dbgs() << "Found replacement " << *DominatingLeader << " for " - << *U->get() << " in " << *(U->getUser()) << "\n"); + LLVM_DEBUG(dbgs() + << "Found replacement " << *DominatingLeader << " for " + << *U->get() << " in " << *(U->getUser()) << "\n"); // If we replaced something in an instruction, handle the patching of // metadata. Skip this if we are replacing predicateinfo with its @@ -4157,8 +4173,8 @@ bool NewGVN::eliminateInstructions(Function &F) { (void)Leader; assert(DT->dominates(Leader->getParent(), Member->getParent())); // Member is dominater by Leader, and thus dead - DEBUG(dbgs() << "Marking dead store " << *Member - << " that is dominated by " << *Leader << "\n"); + LLVM_DEBUG(dbgs() << "Marking dead store " << *Member + << " that is dominated by " << *Leader << "\n"); markInstructionForDeletion(Member); CC->erase(Member); ++NumGVNDeadStores; diff --git a/llvm/lib/Transforms/Scalar/PlaceSafepoints.cpp b/llvm/lib/Transforms/Scalar/PlaceSafepoints.cpp index 6b1ae2c3cab..a06c424f82b 100644 --- a/llvm/lib/Transforms/Scalar/PlaceSafepoints.cpp +++ b/llvm/lib/Transforms/Scalar/PlaceSafepoints.cpp @@ -323,7 +323,7 @@ bool PlaceBackedgeSafepointsImpl::runOnLoop(Loop *L) { // avoiding the runtime cost of the actual safepoint. if (!AllBackedges) { if (mustBeFiniteCountedLoop(L, SE, Pred)) { - DEBUG(dbgs() << "skipping safepoint placement in finite loop\n"); + LLVM_DEBUG(dbgs() << "skipping safepoint placement in finite loop\n"); FiniteExecution++; continue; } @@ -332,7 +332,9 @@ bool PlaceBackedgeSafepointsImpl::runOnLoop(Loop *L) { // Note: This is only semantically legal since we won't do any further // IPO or inlining before the actual call insertion.. If we hadn't, we // might latter loose this call safepoint. - DEBUG(dbgs() << "skipping safepoint placement due to unconditional call\n"); + LLVM_DEBUG( + dbgs() + << "skipping safepoint placement due to unconditional call\n"); CallInLoop++; continue; } @@ -348,7 +350,7 @@ bool PlaceBackedgeSafepointsImpl::runOnLoop(Loop *L) { // variables) and branches to the true header TerminatorInst *Term = Pred->getTerminator(); - DEBUG(dbgs() << "[LSP] terminator instruction: " << *Term); + LLVM_DEBUG(dbgs() << "[LSP] terminator instruction: " << *Term); PollLocations.push_back(Term); } diff --git a/llvm/lib/Transforms/Scalar/Reassociate.cpp b/llvm/lib/Transforms/Scalar/Reassociate.cpp index 0d1d57d6486..bb9802a66ea 100644 --- a/llvm/lib/Transforms/Scalar/Reassociate.cpp +++ b/llvm/lib/Transforms/Scalar/Reassociate.cpp @@ -169,8 +169,8 @@ void ReassociatePass::BuildRankMap(Function &F, // Assign distinct ranks to function arguments. for (auto &Arg : F.args()) { ValueRankMap[&Arg] = ++Rank; - DEBUG(dbgs() << "Calculated Rank[" << Arg.getName() << "] = " << Rank - << "\n"); + LLVM_DEBUG(dbgs() << "Calculated Rank[" << Arg.getName() << "] = " << Rank + << "\n"); } // Traverse basic blocks in ReversePostOrder @@ -210,7 +210,8 @@ unsigned ReassociatePass::getRank(Value *V) { !BinaryOperator::isFNeg(I)) ++Rank; - DEBUG(dbgs() << "Calculated Rank[" << V->getName() << "] = " << Rank << "\n"); + LLVM_DEBUG(dbgs() << "Calculated Rank[" << V->getName() << "] = " << Rank + << "\n"); return ValueRankMap[I] = Rank; } @@ -445,7 +446,7 @@ using RepeatedValue = std::pair<Value*, APInt>; /// type and thus make the expression bigger. static bool LinearizeExprTree(BinaryOperator *I, SmallVectorImpl<RepeatedValue> &Ops) { - DEBUG(dbgs() << "LINEARIZE: " << *I << '\n'); + LLVM_DEBUG(dbgs() << "LINEARIZE: " << *I << '\n'); unsigned Bitwidth = I->getType()->getScalarType()->getPrimitiveSizeInBits(); unsigned Opcode = I->getOpcode(); assert(I->isAssociative() && I->isCommutative() && @@ -494,14 +495,14 @@ static bool LinearizeExprTree(BinaryOperator *I, for (unsigned OpIdx = 0; OpIdx < 2; ++OpIdx) { // Visit operands. Value *Op = I->getOperand(OpIdx); APInt Weight = P.second; // Number of paths to this operand. - DEBUG(dbgs() << "OPERAND: " << *Op << " (" << Weight << ")\n"); + LLVM_DEBUG(dbgs() << "OPERAND: " << *Op << " (" << Weight << ")\n"); assert(!Op->use_empty() && "No uses, so how did we get to it?!"); // If this is a binary operation of the right kind with only one use then // add its operands to the expression. if (BinaryOperator *BO = isReassociableOp(Op, Opcode)) { assert(Visited.insert(Op).second && "Not first visit!"); - DEBUG(dbgs() << "DIRECT ADD: " << *Op << " (" << Weight << ")\n"); + LLVM_DEBUG(dbgs() << "DIRECT ADD: " << *Op << " (" << Weight << ")\n"); Worklist.push_back(std::make_pair(BO, Weight)); continue; } @@ -514,7 +515,8 @@ static bool LinearizeExprTree(BinaryOperator *I, if (!Op->hasOneUse()) { // This value has uses not accounted for by the expression, so it is // not safe to modify. Mark it as being a leaf. - DEBUG(dbgs() << "ADD USES LEAF: " << *Op << " (" << Weight << ")\n"); + LLVM_DEBUG(dbgs() + << "ADD USES LEAF: " << *Op << " (" << Weight << ")\n"); LeafOrder.push_back(Op); Leaves[Op] = Weight; continue; @@ -540,7 +542,7 @@ static bool LinearizeExprTree(BinaryOperator *I, // to the expression, then no longer consider it to be a leaf and add // its operands to the expression. if (BinaryOperator *BO = isReassociableOp(Op, Opcode)) { - DEBUG(dbgs() << "UNLEAF: " << *Op << " (" << It->second << ")\n"); + LLVM_DEBUG(dbgs() << "UNLEAF: " << *Op << " (" << It->second << ")\n"); Worklist.push_back(std::make_pair(BO, It->second)); Leaves.erase(It); continue; @@ -573,9 +575,10 @@ static bool LinearizeExprTree(BinaryOperator *I, if (BinaryOperator *BO = dyn_cast<BinaryOperator>(Op)) if ((Opcode == Instruction::Mul && BinaryOperator::isNeg(BO)) || (Opcode == Instruction::FMul && BinaryOperator::isFNeg(BO))) { - DEBUG(dbgs() << "MORPH LEAF: " << *Op << " (" << Weight << ") TO "); + LLVM_DEBUG(dbgs() + << "MORPH LEAF: " << *Op << " (" << Weight << ") TO "); BO = LowerNegateToMultiply(BO); - DEBUG(dbgs() << *BO << '\n'); + LLVM_DEBUG(dbgs() << *BO << '\n'); Worklist.push_back(std::make_pair(BO, Weight)); Changed = true; continue; @@ -583,7 +586,7 @@ static bool LinearizeExprTree(BinaryOperator *I, // Failed to morph into an expression of the right type. This really is // a leaf. - DEBUG(dbgs() << "ADD LEAF: " << *Op << " (" << Weight << ")\n"); + LLVM_DEBUG(dbgs() << "ADD LEAF: " << *Op << " (" << Weight << ")\n"); assert(!isReassociableOp(Op, Opcode) && "Value was morphed?"); LeafOrder.push_back(Op); Leaves[Op] = Weight; @@ -675,9 +678,9 @@ void ReassociatePass::RewriteExprTree(BinaryOperator *I, if (NewLHS == OldRHS && NewRHS == OldLHS) { // The order of the operands was reversed. Swap them. - DEBUG(dbgs() << "RA: " << *Op << '\n'); + LLVM_DEBUG(dbgs() << "RA: " << *Op << '\n'); Op->swapOperands(); - DEBUG(dbgs() << "TO: " << *Op << '\n'); + LLVM_DEBUG(dbgs() << "TO: " << *Op << '\n'); MadeChange = true; ++NumChanged; break; @@ -685,7 +688,7 @@ void ReassociatePass::RewriteExprTree(BinaryOperator *I, // The new operation differs non-trivially from the original. Overwrite // the old operands with the new ones. - DEBUG(dbgs() << "RA: " << *Op << '\n'); + LLVM_DEBUG(dbgs() << "RA: " << *Op << '\n'); if (NewLHS != OldLHS) { BinaryOperator *BO = isReassociableOp(OldLHS, Opcode); if (BO && !NotRewritable.count(BO)) @@ -698,7 +701,7 @@ void ReassociatePass::RewriteExprTree(BinaryOperator *I, NodesToRewrite.push_back(BO); Op->setOperand(1, NewRHS); } - DEBUG(dbgs() << "TO: " << *Op << '\n'); + LLVM_DEBUG(dbgs() << "TO: " << *Op << '\n'); ExpressionChanged = Op; MadeChange = true; @@ -711,7 +714,7 @@ void ReassociatePass::RewriteExprTree(BinaryOperator *I, // while the right-hand side will be the current element of Ops. Value *NewRHS = Ops[i].Op; if (NewRHS != Op->getOperand(1)) { - DEBUG(dbgs() << "RA: " << *Op << '\n'); + LLVM_DEBUG(dbgs() << "RA: " << *Op << '\n'); if (NewRHS == Op->getOperand(0)) { // The new right-hand side was already present as the left operand. If // we are lucky then swapping the operands will sort out both of them. @@ -724,7 +727,7 @@ void ReassociatePass::RewriteExprTree(BinaryOperator *I, Op->setOperand(1, NewRHS); ExpressionChanged = Op; } - DEBUG(dbgs() << "TO: " << *Op << '\n'); + LLVM_DEBUG(dbgs() << "TO: " << *Op << '\n'); MadeChange = true; ++NumChanged; } @@ -756,9 +759,9 @@ void ReassociatePass::RewriteExprTree(BinaryOperator *I, NewOp = NodesToRewrite.pop_back_val(); } - DEBUG(dbgs() << "RA: " << *Op << '\n'); + LLVM_DEBUG(dbgs() << "RA: " << *Op << '\n'); Op->setOperand(0, NewOp); - DEBUG(dbgs() << "TO: " << *Op << '\n'); + LLVM_DEBUG(dbgs() << "TO: " << *Op << '\n'); ExpressionChanged = Op; MadeChange = true; ++NumChanged; @@ -941,7 +944,7 @@ static BinaryOperator *BreakUpSubtract(Instruction *Sub, Sub->replaceAllUsesWith(New); New->setDebugLoc(Sub->getDebugLoc()); - DEBUG(dbgs() << "Negated: " << *New << '\n'); + LLVM_DEBUG(dbgs() << "Negated: " << *New << '\n'); return New; } @@ -1427,7 +1430,8 @@ Value *ReassociatePass::OptimizeAdd(Instruction *I, ++NumFound; } while (i != Ops.size() && Ops[i].Op == TheOp); - DEBUG(dbgs() << "\nFACTORING [" << NumFound << "]: " << *TheOp << '\n'); + LLVM_DEBUG(dbgs() << "\nFACTORING [" << NumFound << "]: " << *TheOp + << '\n'); ++NumFactor; // Insert a new multiply. @@ -1565,7 +1569,8 @@ Value *ReassociatePass::OptimizeAdd(Instruction *I, // If any factor occurred more than one time, we can pull it out. if (MaxOcc > 1) { - DEBUG(dbgs() << "\nFACTORING [" << MaxOcc << "]: " << *MaxOccVal << '\n'); + LLVM_DEBUG(dbgs() << "\nFACTORING [" << MaxOcc << "]: " << *MaxOccVal + << '\n'); ++NumFactor; // Create a new instruction that uses the MaxOccVal twice. If we don't do @@ -1888,7 +1893,7 @@ void ReassociatePass::RecursivelyEraseDeadInsts(Instruction *I, /// Zap the given instruction, adding interesting operands to the work list. void ReassociatePass::EraseInst(Instruction *I) { assert(isInstructionTriviallyDead(I) && "Trivially dead instructions only!"); - DEBUG(dbgs() << "Erasing dead inst: "; I->dump()); + LLVM_DEBUG(dbgs() << "Erasing dead inst: "; I->dump()); SmallVector<Value*, 8> Ops(I->op_begin(), I->op_end()); // Erase the dead instruction. @@ -2139,7 +2144,7 @@ void ReassociatePass::ReassociateExpression(BinaryOperator *I) { ValueEntry(getRank(E.first), E.first)); } - DEBUG(dbgs() << "RAIn:\t"; PrintOps(I, Ops); dbgs() << '\n'); + LLVM_DEBUG(dbgs() << "RAIn:\t"; PrintOps(I, Ops); dbgs() << '\n'); // Now that we have linearized the tree to a list and have gathered all of // the operands and their ranks, sort the operands by their rank. Use a @@ -2157,7 +2162,7 @@ void ReassociatePass::ReassociateExpression(BinaryOperator *I) { return; // This expression tree simplified to something that isn't a tree, // eliminate it. - DEBUG(dbgs() << "Reassoc to scalar: " << *V << '\n'); + LLVM_DEBUG(dbgs() << "Reassoc to scalar: " << *V << '\n'); I->replaceAllUsesWith(V); if (Instruction *VI = dyn_cast<Instruction>(V)) if (I->getDebugLoc()) @@ -2188,7 +2193,7 @@ void ReassociatePass::ReassociateExpression(BinaryOperator *I) { } } - DEBUG(dbgs() << "RAOut:\t"; PrintOps(I, Ops); dbgs() << '\n'); + LLVM_DEBUG(dbgs() << "RAOut:\t"; PrintOps(I, Ops); dbgs() << '\n'); if (Ops.size() == 1) { if (Ops[0].Op == I) diff --git a/llvm/lib/Transforms/Scalar/RewriteStatepointsForGC.cpp b/llvm/lib/Transforms/Scalar/RewriteStatepointsForGC.cpp index bf851ef2e71..342ba9bc145 100644 --- a/llvm/lib/Transforms/Scalar/RewriteStatepointsForGC.cpp +++ b/llvm/lib/Transforms/Scalar/RewriteStatepointsForGC.cpp @@ -616,8 +616,8 @@ static Value *findBaseDefiningValueCached(Value *I, DefiningValueMapTy &Cache) { Value *&Cached = Cache[I]; if (!Cached) { Cached = findBaseDefiningValue(I).BDV; - DEBUG(dbgs() << "fBDV-cached: " << I->getName() << " -> " - << Cached->getName() << "\n"); + LLVM_DEBUG(dbgs() << "fBDV-cached: " << I->getName() << " -> " + << Cached->getName() << "\n"); } assert(Cache[I] != nullptr); return Cached; @@ -848,9 +848,9 @@ static Value *findBasePointer(Value *I, DefiningValueMapTy &Cache) { } #ifndef NDEBUG - DEBUG(dbgs() << "States after initialization:\n"); + LLVM_DEBUG(dbgs() << "States after initialization:\n"); for (auto Pair : States) { - DEBUG(dbgs() << " " << Pair.second << " for " << *Pair.first << "\n"); + LLVM_DEBUG(dbgs() << " " << Pair.second << " for " << *Pair.first << "\n"); } #endif @@ -923,9 +923,9 @@ static Value *findBasePointer(Value *I, DefiningValueMapTy &Cache) { } #ifndef NDEBUG - DEBUG(dbgs() << "States after meet iteration:\n"); + LLVM_DEBUG(dbgs() << "States after meet iteration:\n"); for (auto Pair : States) { - DEBUG(dbgs() << " " << Pair.second << " for " << *Pair.first << "\n"); + LLVM_DEBUG(dbgs() << " " << Pair.second << " for " << *Pair.first << "\n"); } #endif @@ -1124,10 +1124,11 @@ static Value *findBasePointer(Value *I, DefiningValueMapTy &Cache) { assert(BDV && Base); assert(!isKnownBaseResult(BDV) && "why did it get added?"); - DEBUG(dbgs() << "Updating base value cache" - << " for: " << BDV->getName() << " from: " - << (Cache.count(BDV) ? Cache[BDV]->getName().str() : "none") - << " to: " << Base->getName() << "\n"); + LLVM_DEBUG( + dbgs() << "Updating base value cache" + << " for: " << BDV->getName() << " from: " + << (Cache.count(BDV) ? Cache[BDV]->getName().str() : "none") + << " to: " << Base->getName() << "\n"); if (Cache.count(BDV)) { assert(isKnownBaseResult(Base) && diff --git a/llvm/lib/Transforms/Scalar/SCCP.cpp b/llvm/lib/Transforms/Scalar/SCCP.cpp index 5a6697fd9fe..27c51f07f1f 100644 --- a/llvm/lib/Transforms/Scalar/SCCP.cpp +++ b/llvm/lib/Transforms/Scalar/SCCP.cpp @@ -259,7 +259,7 @@ public: bool MarkBlockExecutable(BasicBlock *BB) { if (!BBExecutable.insert(BB).second) return false; - DEBUG(dbgs() << "Marking Block Executable: " << BB->getName() << '\n'); + LLVM_DEBUG(dbgs() << "Marking Block Executable: " << BB->getName() << '\n'); BBWorkList.push_back(BB); // Add the block to the work list! return true; } @@ -415,7 +415,7 @@ private: // the users of the instruction are updated later. void markConstant(LatticeVal &IV, Value *V, Constant *C) { if (!IV.markConstant(C)) return; - DEBUG(dbgs() << "markConstant: " << *C << ": " << *V << '\n'); + LLVM_DEBUG(dbgs() << "markConstant: " << *C << ": " << *V << '\n'); pushToWorkList(IV, V); } @@ -428,7 +428,7 @@ private: assert(!V->getType()->isStructTy() && "structs should use mergeInValue"); LatticeVal &IV = ValueState[V]; IV.markForcedConstant(C); - DEBUG(dbgs() << "markForcedConstant: " << *C << ": " << *V << '\n'); + LLVM_DEBUG(dbgs() << "markForcedConstant: " << *C << ": " << *V << '\n'); pushToWorkList(IV, V); } @@ -438,11 +438,10 @@ private: void markOverdefined(LatticeVal &IV, Value *V) { if (!IV.markOverdefined()) return; - DEBUG(dbgs() << "markOverdefined: "; - if (auto *F = dyn_cast<Function>(V)) - dbgs() << "Function '" << F->getName() << "'\n"; - else - dbgs() << *V << '\n'); + LLVM_DEBUG(dbgs() << "markOverdefined: "; + if (auto *F = dyn_cast<Function>(V)) dbgs() + << "Function '" << F->getName() << "'\n"; + else dbgs() << *V << '\n'); // Only instructions go on the work list pushToWorkList(IV, V); } @@ -540,8 +539,8 @@ private: // If the destination is already executable, we just made an *edge* // feasible that wasn't before. Revisit the PHI nodes in the block // because they have potentially new operands. - DEBUG(dbgs() << "Marking Edge Executable: " << Source->getName() - << " -> " << Dest->getName() << '\n'); + LLVM_DEBUG(dbgs() << "Marking Edge Executable: " << Source->getName() + << " -> " << Dest->getName() << '\n'); for (PHINode &PN : Dest->phis()) visitPHINode(PN); @@ -612,7 +611,7 @@ private: void visitInstruction(Instruction &I) { // All the instructions we don't do any special handling for just // go to overdefined. - DEBUG(dbgs() << "SCCP: Don't know how to handle: " << I << '\n'); + LLVM_DEBUG(dbgs() << "SCCP: Don't know how to handle: " << I << '\n'); markOverdefined(&I); } }; @@ -699,7 +698,7 @@ void SCCPSolver::getFeasibleSuccessors(TerminatorInst &TI, return; } - DEBUG(dbgs() << "Unknown terminator instruction: " << TI << '\n'); + LLVM_DEBUG(dbgs() << "Unknown terminator instruction: " << TI << '\n'); llvm_unreachable("SCCP: Don't know how to handle this terminator!"); } @@ -759,7 +758,7 @@ bool SCCPSolver::isEdgeFeasible(BasicBlock *From, BasicBlock *To) { return Addr->getBasicBlock() == To; } - DEBUG(dbgs() << "Unknown terminator instruction: " << *TI << '\n'); + LLVM_DEBUG(dbgs() << "Unknown terminator instruction: " << *TI << '\n'); llvm_unreachable("SCCP: Don't know how to handle this terminator!"); } @@ -1260,7 +1259,7 @@ void SCCPSolver::Solve() { while (!OverdefinedInstWorkList.empty()) { Value *I = OverdefinedInstWorkList.pop_back_val(); - DEBUG(dbgs() << "\nPopped off OI-WL: " << *I << '\n'); + LLVM_DEBUG(dbgs() << "\nPopped off OI-WL: " << *I << '\n'); // "I" got into the work list because it either made the transition from // bottom to constant, or to overdefined. @@ -1278,7 +1277,7 @@ void SCCPSolver::Solve() { while (!InstWorkList.empty()) { Value *I = InstWorkList.pop_back_val(); - DEBUG(dbgs() << "\nPopped off I-WL: " << *I << '\n'); + LLVM_DEBUG(dbgs() << "\nPopped off I-WL: " << *I << '\n'); // "I" got into the work list because it made the transition from undef to // constant. @@ -1298,7 +1297,7 @@ void SCCPSolver::Solve() { BasicBlock *BB = BBWorkList.back(); BBWorkList.pop_back(); - DEBUG(dbgs() << "\nPopped off BBWL: " << *BB << '\n'); + LLVM_DEBUG(dbgs() << "\nPopped off BBWL: " << *BB << '\n'); // Notify all instructions in this basic block that they are newly // executable. @@ -1645,9 +1644,9 @@ static bool tryToReplaceWithConstantRange(SCCPSolver &Solver, Value *V) { Constant *C = A.getCompare(Icmp->getPredicate(), Icmp->getType(), B); if (C) { Icmp->replaceAllUsesWith(C); - DEBUG(dbgs() << "Replacing " << *Icmp << " with " << *C - << ", because of range information " << A << " " << B - << "\n"); + LLVM_DEBUG(dbgs() << "Replacing " << *Icmp << " with " << *C + << ", because of range information " << A << " " << B + << "\n"); Icmp->eraseFromParent(); Changed = true; } @@ -1699,12 +1698,12 @@ static bool tryToReplaceWithConstant(SCCPSolver &Solver, Value *V) { if (F) Solver.AddMustTailCallee(F); - DEBUG(dbgs() << " Can\'t treat the result of musttail call : " << *CI - << " as a constant\n"); + LLVM_DEBUG(dbgs() << " Can\'t treat the result of musttail call : " << *CI + << " as a constant\n"); return false; } - DEBUG(dbgs() << " Constant: " << *Const << " = " << *V << '\n'); + LLVM_DEBUG(dbgs() << " Constant: " << *Const << " = " << *V << '\n'); // Replaces all of the uses of a variable with uses of the constant. V->replaceAllUsesWith(Const); @@ -1715,7 +1714,7 @@ static bool tryToReplaceWithConstant(SCCPSolver &Solver, Value *V) { // and return true if the function was modified. static bool runSCCP(Function &F, const DataLayout &DL, const TargetLibraryInfo *TLI) { - DEBUG(dbgs() << "SCCP on function '" << F.getName() << "'\n"); + LLVM_DEBUG(dbgs() << "SCCP on function '" << F.getName() << "'\n"); SCCPSolver Solver(DL, TLI); // Mark the first block of the function as being executable. @@ -1729,7 +1728,7 @@ static bool runSCCP(Function &F, const DataLayout &DL, bool ResolvedUndefs = true; while (ResolvedUndefs) { Solver.Solve(); - DEBUG(dbgs() << "RESOLVING UNDEFs\n"); + LLVM_DEBUG(dbgs() << "RESOLVING UNDEFs\n"); ResolvedUndefs = Solver.ResolvedUndefsIn(F); } @@ -1741,7 +1740,7 @@ static bool runSCCP(Function &F, const DataLayout &DL, for (BasicBlock &BB : F) { if (!Solver.isBlockExecutable(&BB)) { - DEBUG(dbgs() << " BasicBlock Dead:" << BB); + LLVM_DEBUG(dbgs() << " BasicBlock Dead:" << BB); ++NumDeadBlocks; NumInstRemoved += removeAllNonTerminatorAndEHPadInstructions(&BB); @@ -1837,15 +1836,15 @@ static void findReturnsToZap(Function &F, // There is a non-removable musttail call site of this function. Zapping // returns is not allowed. if (Solver.isMustTailCallee(&F)) { - DEBUG(dbgs() << "Can't zap returns of the function : " << F.getName() - << " due to present musttail call of it\n"); + LLVM_DEBUG(dbgs() << "Can't zap returns of the function : " << F.getName() + << " due to present musttail call of it\n"); return; } for (BasicBlock &BB : F) { if (CallInst *CI = BB.getTerminatingMustTailCall()) { - DEBUG(dbgs() << "Can't zap return of the block due to present " - << "musttail call : " << *CI << "\n"); + LLVM_DEBUG(dbgs() << "Can't zap return of the block due to present " + << "musttail call : " << *CI << "\n"); (void)CI; return; } @@ -1900,7 +1899,7 @@ bool llvm::runIPSCCP(Module &M, const DataLayout &DL, while (ResolvedUndefs) { Solver.Solve(); - DEBUG(dbgs() << "RESOLVING UNDEFS\n"); + LLVM_DEBUG(dbgs() << "RESOLVING UNDEFS\n"); ResolvedUndefs = false; for (Function &F : M) ResolvedUndefs |= Solver.ResolvedUndefsIn(F); @@ -1930,7 +1929,7 @@ bool llvm::runIPSCCP(Module &M, const DataLayout &DL, for (Function::iterator BB = F.begin(), E = F.end(); BB != E; ++BB) { if (!Solver.isBlockExecutable(&*BB)) { - DEBUG(dbgs() << " BasicBlock Dead:" << *BB); + LLVM_DEBUG(dbgs() << " BasicBlock Dead:" << *BB); ++NumDeadBlocks; NumInstRemoved += @@ -2028,7 +2027,8 @@ bool llvm::runIPSCCP(Module &M, const DataLayout &DL, GlobalVariable *GV = I->first; assert(!I->second.isOverdefined() && "Overdefined values should have been taken out of the map!"); - DEBUG(dbgs() << "Found that GV '" << GV->getName() << "' is constant!\n"); + LLVM_DEBUG(dbgs() << "Found that GV '" << GV->getName() + << "' is constant!\n"); while (!GV->use_empty()) { StoreInst *SI = cast<StoreInst>(GV->user_back()); SI->eraseFromParent(); diff --git a/llvm/lib/Transforms/Scalar/SROA.cpp b/llvm/lib/Transforms/Scalar/SROA.cpp index b4200da99cc..2db08b72972 100644 --- a/llvm/lib/Transforms/Scalar/SROA.cpp +++ b/llvm/lib/Transforms/Scalar/SROA.cpp @@ -683,11 +683,12 @@ private: // Completely skip uses which have a zero size or start either before or // past the end of the allocation. if (Size == 0 || Offset.uge(AllocSize)) { - DEBUG(dbgs() << "WARNING: Ignoring " << Size << " byte use @" << Offset - << " which has zero size or starts outside of the " - << AllocSize << " byte alloca:\n" - << " alloca: " << AS.AI << "\n" - << " use: " << I << "\n"); + LLVM_DEBUG(dbgs() << "WARNING: Ignoring " << Size << " byte use @" + << Offset + << " which has zero size or starts outside of the " + << AllocSize << " byte alloca:\n" + << " alloca: " << AS.AI << "\n" + << " use: " << I << "\n"); return markAsDead(I); } @@ -702,10 +703,11 @@ private: // them, and so have to record at least the information here. assert(AllocSize >= BeginOffset); // Established above. if (Size > AllocSize - BeginOffset) { - DEBUG(dbgs() << "WARNING: Clamping a " << Size << " byte use @" << Offset - << " to remain within the " << AllocSize << " byte alloca:\n" - << " alloca: " << AS.AI << "\n" - << " use: " << I << "\n"); + LLVM_DEBUG(dbgs() << "WARNING: Clamping a " << Size << " byte use @" + << Offset << " to remain within the " << AllocSize + << " byte alloca:\n" + << " alloca: " << AS.AI << "\n" + << " use: " << I << "\n"); EndOffset = AllocSize; } @@ -805,11 +807,11 @@ private: // FIXME: We should instead consider the pointer to have escaped if this // function is being instrumented for addressing bugs or race conditions. if (Size > AllocSize || Offset.ugt(AllocSize - Size)) { - DEBUG(dbgs() << "WARNING: Ignoring " << Size << " byte store @" << Offset - << " which extends past the end of the " << AllocSize - << " byte alloca:\n" - << " alloca: " << AS.AI << "\n" - << " use: " << SI << "\n"); + LLVM_DEBUG(dbgs() << "WARNING: Ignoring " << Size << " byte store @" + << Offset << " which extends past the end of the " + << AllocSize << " byte alloca:\n" + << " alloca: " << AS.AI << "\n" + << " use: " << SI << "\n"); return markAsDead(SI); } @@ -1236,7 +1238,7 @@ static bool isSafePHIToSpeculate(PHINode &PN) { } static void speculatePHINodeLoads(PHINode &PN) { - DEBUG(dbgs() << " original: " << PN << "\n"); + LLVM_DEBUG(dbgs() << " original: " << PN << "\n"); Type *LoadTy = cast<PointerType>(PN.getType())->getElementType(); IRBuilderTy PHIBuilder(&PN); @@ -1274,7 +1276,7 @@ static void speculatePHINodeLoads(PHINode &PN) { NewPN->addIncoming(Load, Pred); } - DEBUG(dbgs() << " speculated to: " << *NewPN << "\n"); + LLVM_DEBUG(dbgs() << " speculated to: " << *NewPN << "\n"); PN.eraseFromParent(); } @@ -1314,7 +1316,7 @@ static bool isSafeSelectToSpeculate(SelectInst &SI) { } static void speculateSelectInstLoads(SelectInst &SI) { - DEBUG(dbgs() << " original: " << SI << "\n"); + LLVM_DEBUG(dbgs() << " original: " << SI << "\n"); IRBuilderTy IRB(&SI); Value *TV = SI.getTrueValue(); @@ -1345,7 +1347,7 @@ static void speculateSelectInstLoads(SelectInst &SI) { Value *V = IRB.CreateSelect(SI.getCondition(), TL, FL, LI->getName() + ".sroa.speculated"); - DEBUG(dbgs() << " speculated to: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " speculated to: " << *V << "\n"); LI->replaceAllUsesWith(V); LI->eraseFromParent(); } @@ -2068,7 +2070,7 @@ static bool isIntegerWideningViable(Partition &P, Type *AllocaTy, static Value *extractInteger(const DataLayout &DL, IRBuilderTy &IRB, Value *V, IntegerType *Ty, uint64_t Offset, const Twine &Name) { - DEBUG(dbgs() << " start: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " start: " << *V << "\n"); IntegerType *IntTy = cast<IntegerType>(V->getType()); assert(DL.getTypeStoreSize(Ty) + Offset <= DL.getTypeStoreSize(IntTy) && "Element extends past full value"); @@ -2077,13 +2079,13 @@ static Value *extractInteger(const DataLayout &DL, IRBuilderTy &IRB, Value *V, ShAmt = 8 * (DL.getTypeStoreSize(IntTy) - DL.getTypeStoreSize(Ty) - Offset); if (ShAmt) { V = IRB.CreateLShr(V, ShAmt, Name + ".shift"); - DEBUG(dbgs() << " shifted: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " shifted: " << *V << "\n"); } assert(Ty->getBitWidth() <= IntTy->getBitWidth() && "Cannot extract to a larger integer!"); if (Ty != IntTy) { V = IRB.CreateTrunc(V, Ty, Name + ".trunc"); - DEBUG(dbgs() << " trunced: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " trunced: " << *V << "\n"); } return V; } @@ -2094,10 +2096,10 @@ static Value *insertInteger(const DataLayout &DL, IRBuilderTy &IRB, Value *Old, IntegerType *Ty = cast<IntegerType>(V->getType()); assert(Ty->getBitWidth() <= IntTy->getBitWidth() && "Cannot insert a larger integer!"); - DEBUG(dbgs() << " start: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " start: " << *V << "\n"); if (Ty != IntTy) { V = IRB.CreateZExt(V, IntTy, Name + ".ext"); - DEBUG(dbgs() << " extended: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " extended: " << *V << "\n"); } assert(DL.getTypeStoreSize(Ty) + Offset <= DL.getTypeStoreSize(IntTy) && "Element store outside of alloca store"); @@ -2106,15 +2108,15 @@ static Value *insertInteger(const DataLayout &DL, IRBuilderTy &IRB, Value *Old, ShAmt = 8 * (DL.getTypeStoreSize(IntTy) - DL.getTypeStoreSize(Ty) - Offset); if (ShAmt) { V = IRB.CreateShl(V, ShAmt, Name + ".shift"); - DEBUG(dbgs() << " shifted: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " shifted: " << *V << "\n"); } if (ShAmt || Ty->getBitWidth() < IntTy->getBitWidth()) { APInt Mask = ~Ty->getMask().zext(IntTy->getBitWidth()).shl(ShAmt); Old = IRB.CreateAnd(Old, Mask, Name + ".mask"); - DEBUG(dbgs() << " masked: " << *Old << "\n"); + LLVM_DEBUG(dbgs() << " masked: " << *Old << "\n"); V = IRB.CreateOr(Old, V, Name + ".insert"); - DEBUG(dbgs() << " inserted: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " inserted: " << *V << "\n"); } return V; } @@ -2131,7 +2133,7 @@ static Value *extractVector(IRBuilderTy &IRB, Value *V, unsigned BeginIndex, if (NumElements == 1) { V = IRB.CreateExtractElement(V, IRB.getInt32(BeginIndex), Name + ".extract"); - DEBUG(dbgs() << " extract: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " extract: " << *V << "\n"); return V; } @@ -2141,7 +2143,7 @@ static Value *extractVector(IRBuilderTy &IRB, Value *V, unsigned BeginIndex, Mask.push_back(IRB.getInt32(i)); V = IRB.CreateShuffleVector(V, UndefValue::get(V->getType()), ConstantVector::get(Mask), Name + ".extract"); - DEBUG(dbgs() << " shuffle: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " shuffle: " << *V << "\n"); return V; } @@ -2155,7 +2157,7 @@ static Value *insertVector(IRBuilderTy &IRB, Value *Old, Value *V, // Single element to insert. V = IRB.CreateInsertElement(Old, V, IRB.getInt32(BeginIndex), Name + ".insert"); - DEBUG(dbgs() << " insert: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " insert: " << *V << "\n"); return V; } @@ -2180,7 +2182,7 @@ static Value *insertVector(IRBuilderTy &IRB, Value *Old, Value *V, Mask.push_back(UndefValue::get(IRB.getInt32Ty())); V = IRB.CreateShuffleVector(V, UndefValue::get(V->getType()), ConstantVector::get(Mask), Name + ".expand"); - DEBUG(dbgs() << " shuffle: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " shuffle: " << *V << "\n"); Mask.clear(); for (unsigned i = 0; i != VecTy->getNumElements(); ++i) @@ -2188,7 +2190,7 @@ static Value *insertVector(IRBuilderTy &IRB, Value *Old, Value *V, V = IRB.CreateSelect(ConstantVector::get(Mask), V, Old, Name + "blend"); - DEBUG(dbgs() << " blend: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " blend: " << *V << "\n"); return V; } @@ -2291,9 +2293,9 @@ public: IsSplittable = I->isSplittable(); IsSplit = BeginOffset < NewAllocaBeginOffset || EndOffset > NewAllocaEndOffset; - DEBUG(dbgs() << " rewriting " << (IsSplit ? "split " : "")); - DEBUG(AS.printSlice(dbgs(), I, "")); - DEBUG(dbgs() << "\n"); + LLVM_DEBUG(dbgs() << " rewriting " << (IsSplit ? "split " : "")); + LLVM_DEBUG(AS.printSlice(dbgs(), I, "")); + LLVM_DEBUG(dbgs() << "\n"); // Compute the intersecting offset range. assert(BeginOffset < NewAllocaEndOffset); @@ -2323,7 +2325,7 @@ private: // Every instruction which can end up as a user must have a rewrite rule. bool visitInstruction(Instruction &I) { - DEBUG(dbgs() << " !!!! Cannot rewrite: " << I << "\n"); + LLVM_DEBUG(dbgs() << " !!!! Cannot rewrite: " << I << "\n"); llvm_unreachable("No rewrite rule for this instruction!"); } @@ -2427,7 +2429,7 @@ private: } bool visitLoadInst(LoadInst &LI) { - DEBUG(dbgs() << " original: " << LI << "\n"); + LLVM_DEBUG(dbgs() << " original: " << LI << "\n"); Value *OldOp = LI.getOperand(0); assert(OldOp == OldPtr); @@ -2527,7 +2529,7 @@ private: Pass.DeadInsts.insert(&LI); deleteIfTriviallyDead(OldOp); - DEBUG(dbgs() << " to: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *V << "\n"); return !LI.isVolatile() && !IsPtrAdjusted; } @@ -2554,7 +2556,7 @@ private: Store->setAAMetadata(AATags); Pass.DeadInsts.insert(&SI); - DEBUG(dbgs() << " to: " << *Store << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *Store << "\n"); return true; } @@ -2575,12 +2577,12 @@ private: if (AATags) Store->setAAMetadata(AATags); Pass.DeadInsts.insert(&SI); - DEBUG(dbgs() << " to: " << *Store << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *Store << "\n"); return true; } bool visitStoreInst(StoreInst &SI) { - DEBUG(dbgs() << " original: " << SI << "\n"); + LLVM_DEBUG(dbgs() << " original: " << SI << "\n"); Value *OldOp = SI.getOperand(1); assert(OldOp == OldPtr); @@ -2648,7 +2650,7 @@ private: Pass.DeadInsts.insert(&SI); deleteIfTriviallyDead(OldOp); - DEBUG(dbgs() << " to: " << *NewSI << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *NewSI << "\n"); return NewSI->getPointerOperand() == &NewAI && !SI.isVolatile(); } @@ -2682,12 +2684,12 @@ private: /// Compute a vector splat for a given element value. Value *getVectorSplat(Value *V, unsigned NumElements) { V = IRB.CreateVectorSplat(NumElements, V, "vsplat"); - DEBUG(dbgs() << " splat: " << *V << "\n"); + LLVM_DEBUG(dbgs() << " splat: " << *V << "\n"); return V; } bool visitMemSetInst(MemSetInst &II) { - DEBUG(dbgs() << " original: " << II << "\n"); + LLVM_DEBUG(dbgs() << " original: " << II << "\n"); assert(II.getRawDest() == OldPtr); AAMDNodes AATags; @@ -2726,7 +2728,7 @@ private: getSliceAlign(), II.isVolatile()); if (AATags) New->setAAMetadata(AATags); - DEBUG(dbgs() << " to: " << *New << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *New << "\n"); return false; } @@ -2792,7 +2794,7 @@ private: II.isVolatile()); if (AATags) New->setAAMetadata(AATags); - DEBUG(dbgs() << " to: " << *New << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *New << "\n"); return !II.isVolatile(); } @@ -2800,7 +2802,7 @@ private: // Rewriting of memory transfer instructions can be a bit tricky. We break // them into two categories: split intrinsics and unsplit intrinsics. - DEBUG(dbgs() << " original: " << II << "\n"); + LLVM_DEBUG(dbgs() << " original: " << II << "\n"); AAMDNodes AATags; II.getAAMetadata(AATags); @@ -2829,7 +2831,7 @@ private: II.setSourceAlignment(SliceAlign); } - DEBUG(dbgs() << " to: " << II << "\n"); + LLVM_DEBUG(dbgs() << " to: " << II << "\n"); deleteIfTriviallyDead(OldPtr); return false; } @@ -2912,7 +2914,7 @@ private: Size, II.isVolatile()); if (AATags) New->setAAMetadata(AATags); - DEBUG(dbgs() << " to: " << *New << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *New << "\n"); return false; } @@ -2984,14 +2986,14 @@ private: IRB.CreateAlignedStore(Src, DstPtr, DstAlign, II.isVolatile())); if (AATags) Store->setAAMetadata(AATags); - DEBUG(dbgs() << " to: " << *Store << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *Store << "\n"); return !II.isVolatile(); } bool visitIntrinsicInst(IntrinsicInst &II) { assert(II.getIntrinsicID() == Intrinsic::lifetime_start || II.getIntrinsicID() == Intrinsic::lifetime_end); - DEBUG(dbgs() << " original: " << II << "\n"); + LLVM_DEBUG(dbgs() << " original: " << II << "\n"); assert(II.getArgOperand(1) == OldPtr); // Record this instruction for deletion. @@ -3019,13 +3021,13 @@ private: New = IRB.CreateLifetimeEnd(Ptr, Size); (void)New; - DEBUG(dbgs() << " to: " << *New << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *New << "\n"); return true; } bool visitPHINode(PHINode &PN) { - DEBUG(dbgs() << " original: " << PN << "\n"); + LLVM_DEBUG(dbgs() << " original: " << PN << "\n"); assert(BeginOffset >= NewAllocaBeginOffset && "PHIs are unsplittable"); assert(EndOffset <= NewAllocaEndOffset && "PHIs are unsplittable"); @@ -3044,7 +3046,7 @@ private: // Replace the operands which were using the old pointer. std::replace(PN.op_begin(), PN.op_end(), cast<Value>(OldPtr), NewPtr); - DEBUG(dbgs() << " to: " << PN << "\n"); + LLVM_DEBUG(dbgs() << " to: " << PN << "\n"); deleteIfTriviallyDead(OldPtr); // PHIs can't be promoted on their own, but often can be speculated. We @@ -3055,7 +3057,7 @@ private: } bool visitSelectInst(SelectInst &SI) { - DEBUG(dbgs() << " original: " << SI << "\n"); + LLVM_DEBUG(dbgs() << " original: " << SI << "\n"); assert((SI.getTrueValue() == OldPtr || SI.getFalseValue() == OldPtr) && "Pointer isn't an operand!"); assert(BeginOffset >= NewAllocaBeginOffset && "Selects are unsplittable"); @@ -3068,7 +3070,7 @@ private: if (SI.getOperand(2) == OldPtr) SI.setOperand(2, NewPtr); - DEBUG(dbgs() << " to: " << SI << "\n"); + LLVM_DEBUG(dbgs() << " to: " << SI << "\n"); deleteIfTriviallyDead(OldPtr); // Selects can't be promoted on their own, but often can be speculated. We @@ -3104,7 +3106,7 @@ public: /// Rewrite loads and stores through a pointer and all pointers derived from /// it. bool rewrite(Instruction &I) { - DEBUG(dbgs() << " Rewriting FCA loads and stores...\n"); + LLVM_DEBUG(dbgs() << " Rewriting FCA loads and stores...\n"); enqueueUsers(I); bool Changed = false; while (!Queue.empty()) { @@ -3218,7 +3220,7 @@ private: if (AATags) Load->setAAMetadata(AATags); Agg = IRB.CreateInsertValue(Agg, Load, Indices, Name + ".insert"); - DEBUG(dbgs() << " to: " << *Load << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *Load << "\n"); } }; @@ -3228,7 +3230,7 @@ private: return false; // We have an aggregate being loaded, split it apart. - DEBUG(dbgs() << " original: " << LI << "\n"); + LLVM_DEBUG(dbgs() << " original: " << LI << "\n"); AAMDNodes AATags; LI.getAAMetadata(AATags); LoadOpSplitter Splitter(&LI, *U, AATags); @@ -3259,7 +3261,7 @@ private: StoreInst *Store = IRB.CreateStore(ExtractValue, InBoundsGEP); if (AATags) Store->setAAMetadata(AATags); - DEBUG(dbgs() << " to: " << *Store << "\n"); + LLVM_DEBUG(dbgs() << " to: " << *Store << "\n"); } }; @@ -3271,7 +3273,7 @@ private: return false; // We have an aggregate being stored, split it apart. - DEBUG(dbgs() << " original: " << SI << "\n"); + LLVM_DEBUG(dbgs() << " original: " << SI << "\n"); AAMDNodes AATags; SI.getAAMetadata(AATags); StoreOpSplitter Splitter(&SI, *U, AATags); @@ -3470,7 +3472,7 @@ static Type *getTypePartition(const DataLayout &DL, Type *Ty, uint64_t Offset, /// /// \returns true if any changes are made. bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { - DEBUG(dbgs() << "Pre-splitting loads and stores\n"); + LLVM_DEBUG(dbgs() << "Pre-splitting loads and stores\n"); // Track the loads and stores which are candidates for pre-splitting here, in // the order they first appear during the partition scan. These give stable @@ -3502,7 +3504,7 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { // maybe it would make it more principled? SmallPtrSet<LoadInst *, 8> UnsplittableLoads; - DEBUG(dbgs() << " Searching for candidate loads and stores\n"); + LLVM_DEBUG(dbgs() << " Searching for candidate loads and stores\n"); for (auto &P : AS.partitions()) { for (Slice &S : P) { Instruction *I = cast<Instruction>(S.getUse()->getUser()); @@ -3557,7 +3559,7 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { } // Record the initial split. - DEBUG(dbgs() << " Candidate: " << *I << "\n"); + LLVM_DEBUG(dbgs() << " Candidate: " << *I << "\n"); auto &Offsets = SplitOffsetsMap[I]; assert(Offsets.Splits.empty() && "Should not have splits the first time we see an instruction!"); @@ -3617,10 +3619,11 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { if (LoadOffsets.Splits == StoreOffsets.Splits) return false; - DEBUG(dbgs() - << " Mismatched splits for load and store:\n" - << " " << *LI << "\n" - << " " << *SI << "\n"); + LLVM_DEBUG( + dbgs() + << " Mismatched splits for load and store:\n" + << " " << *LI << "\n" + << " " << *SI << "\n"); // We've found a store and load that we need to split // with mismatched relative splits. Just give up on them @@ -3693,7 +3696,7 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { Instruction *BasePtr = cast<Instruction>(LI->getPointerOperand()); IRB.SetInsertPoint(LI); - DEBUG(dbgs() << " Splitting load: " << *LI << "\n"); + LLVM_DEBUG(dbgs() << " Splitting load: " << *LI << "\n"); uint64_t PartOffset = 0, PartSize = Offsets.Splits.front(); int Idx = 0, Size = Offsets.Splits.size(); @@ -3718,9 +3721,9 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { Slice(BaseOffset + PartOffset, BaseOffset + PartOffset + PartSize, &PLoad->getOperandUse(PLoad->getPointerOperandIndex()), /*IsSplittable*/ false)); - DEBUG(dbgs() << " new slice [" << NewSlices.back().beginOffset() - << ", " << NewSlices.back().endOffset() << "): " << *PLoad - << "\n"); + LLVM_DEBUG(dbgs() << " new slice [" << NewSlices.back().beginOffset() + << ", " << NewSlices.back().endOffset() + << "): " << *PLoad << "\n"); // See if we've handled all the splits. if (Idx >= Size) @@ -3740,14 +3743,15 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { StoreInst *SI = cast<StoreInst>(LU); if (!Stores.empty() && SplitOffsetsMap.count(SI)) { DeferredStores = true; - DEBUG(dbgs() << " Deferred splitting of store: " << *SI << "\n"); + LLVM_DEBUG(dbgs() << " Deferred splitting of store: " << *SI + << "\n"); continue; } Value *StoreBasePtr = SI->getPointerOperand(); IRB.SetInsertPoint(SI); - DEBUG(dbgs() << " Splitting store of load: " << *SI << "\n"); + LLVM_DEBUG(dbgs() << " Splitting store of load: " << *SI << "\n"); for (int Idx = 0, Size = SplitLoads.size(); Idx < Size; ++Idx) { LoadInst *PLoad = SplitLoads[Idx]; @@ -3763,7 +3767,7 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { PartPtrTy, StoreBasePtr->getName() + "."), getAdjustedAlignment(SI, PartOffset, DL), /*IsVolatile*/ false); PStore->copyMetadata(*LI, LLVMContext::MD_mem_parallel_loop_access); - DEBUG(dbgs() << " +" << PartOffset << ":" << *PStore << "\n"); + LLVM_DEBUG(dbgs() << " +" << PartOffset << ":" << *PStore << "\n"); } // We want to immediately iterate on any allocas impacted by splitting @@ -3812,7 +3816,7 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { Value *LoadBasePtr = LI->getPointerOperand(); Instruction *StoreBasePtr = cast<Instruction>(SI->getPointerOperand()); - DEBUG(dbgs() << " Splitting store: " << *SI << "\n"); + LLVM_DEBUG(dbgs() << " Splitting store: " << *SI << "\n"); // Check whether we have an already split load. auto SplitLoadsMapI = SplitLoadsMap.find(LI); @@ -3822,7 +3826,7 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { assert(SplitLoads->size() == Offsets.Splits.size() + 1 && "Too few split loads for the number of splits in the store!"); } else { - DEBUG(dbgs() << " of load: " << *LI << "\n"); + LLVM_DEBUG(dbgs() << " of load: " << *LI << "\n"); } uint64_t PartOffset = 0, PartSize = Offsets.Splits.front(); @@ -3862,11 +3866,11 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { Slice(BaseOffset + PartOffset, BaseOffset + PartOffset + PartSize, &PStore->getOperandUse(PStore->getPointerOperandIndex()), /*IsSplittable*/ false)); - DEBUG(dbgs() << " new slice [" << NewSlices.back().beginOffset() - << ", " << NewSlices.back().endOffset() << "): " << *PStore - << "\n"); + LLVM_DEBUG(dbgs() << " new slice [" << NewSlices.back().beginOffset() + << ", " << NewSlices.back().endOffset() + << "): " << *PStore << "\n"); if (!SplitLoads) { - DEBUG(dbgs() << " of split load: " << *PLoad << "\n"); + LLVM_DEBUG(dbgs() << " of split load: " << *PLoad << "\n"); } // See if we've finished all the splits. @@ -3921,10 +3925,10 @@ bool SROA::presplitLoadsAndStores(AllocaInst &AI, AllocaSlices &AS) { // sequence. AS.insert(NewSlices); - DEBUG(dbgs() << " Pre-split slices:\n"); + LLVM_DEBUG(dbgs() << " Pre-split slices:\n"); #ifndef NDEBUG for (auto I = AS.begin(), E = AS.end(); I != E; ++I) - DEBUG(AS.print(dbgs(), I, " ")); + LLVM_DEBUG(AS.print(dbgs(), I, " ")); #endif // Finally, don't try to promote any allocas that new require re-splitting. @@ -4008,9 +4012,9 @@ AllocaInst *SROA::rewritePartition(AllocaInst &AI, AllocaSlices &AS, ++NumNewAllocas; } - DEBUG(dbgs() << "Rewriting alloca partition " - << "[" << P.beginOffset() << "," << P.endOffset() - << ") to: " << *NewAI << "\n"); + LLVM_DEBUG(dbgs() << "Rewriting alloca partition " + << "[" << P.beginOffset() << "," << P.endOffset() + << ") to: " << *NewAI << "\n"); // Track the high watermark on the worklist as it is only relevant for // promoted allocas. We will reset it to this point if the alloca is not in @@ -4269,7 +4273,7 @@ void SROA::clobberUse(Use &U) { /// the slices of the alloca, and then hands it off to be split and /// rewritten as needed. bool SROA::runOnAlloca(AllocaInst &AI) { - DEBUG(dbgs() << "SROA alloca: " << AI << "\n"); + LLVM_DEBUG(dbgs() << "SROA alloca: " << AI << "\n"); ++NumAllocasAnalyzed; // Special case dead allocas, as they're trivial. @@ -4293,7 +4297,7 @@ bool SROA::runOnAlloca(AllocaInst &AI) { // Build the slices using a recursive instruction-visiting builder. AllocaSlices AS(DL, AI); - DEBUG(AS.print(dbgs())); + LLVM_DEBUG(AS.print(dbgs())); if (AS.isEscaped()) return Changed; @@ -4321,11 +4325,11 @@ bool SROA::runOnAlloca(AllocaInst &AI) { Changed |= splitAlloca(AI, AS); - DEBUG(dbgs() << " Speculating PHIs\n"); + LLVM_DEBUG(dbgs() << " Speculating PHIs\n"); while (!SpeculatablePHIs.empty()) speculatePHINodeLoads(*SpeculatablePHIs.pop_back_val()); - DEBUG(dbgs() << " Speculating Selects\n"); + LLVM_DEBUG(dbgs() << " Speculating Selects\n"); while (!SpeculatableSelects.empty()) speculateSelectInstLoads(*SpeculatableSelects.pop_back_val()); @@ -4346,7 +4350,7 @@ bool SROA::deleteDeadInstructions( bool Changed = false; while (!DeadInsts.empty()) { Instruction *I = DeadInsts.pop_back_val(); - DEBUG(dbgs() << "Deleting dead instruction: " << *I << "\n"); + LLVM_DEBUG(dbgs() << "Deleting dead instruction: " << *I << "\n"); // If the instruction is an alloca, find the possible dbg.declare connected // to it, and remove it too. We must do this before calling RAUW or we will @@ -4385,7 +4389,7 @@ bool SROA::promoteAllocas(Function &F) { NumPromoted += PromotableAllocas.size(); - DEBUG(dbgs() << "Promoting allocas with mem2reg...\n"); + LLVM_DEBUG(dbgs() << "Promoting allocas with mem2reg...\n"); PromoteMemToReg(PromotableAllocas, *DT, AC); PromotableAllocas.clear(); return true; @@ -4393,7 +4397,7 @@ bool SROA::promoteAllocas(Function &F) { PreservedAnalyses SROA::runImpl(Function &F, DominatorTree &RunDT, AssumptionCache &RunAC) { - DEBUG(dbgs() << "SROA function: " << F.getName() << "\n"); + LLVM_DEBUG(dbgs() << "SROA function: " << F.getName() << "\n"); C = &F.getContext(); DT = &RunDT; AC = &RunAC; diff --git a/llvm/lib/Transforms/Scalar/SimpleLoopUnswitch.cpp b/llvm/lib/Transforms/Scalar/SimpleLoopUnswitch.cpp index 1cd61bfffc2..09b91781e9d 100644 --- a/llvm/lib/Transforms/Scalar/SimpleLoopUnswitch.cpp +++ b/llvm/lib/Transforms/Scalar/SimpleLoopUnswitch.cpp @@ -184,7 +184,7 @@ static void rewritePHINodesForExitAndUnswitchedBlocks(BasicBlock &ExitBB, static bool unswitchTrivialBranch(Loop &L, BranchInst &BI, DominatorTree &DT, LoopInfo &LI) { assert(BI.isConditional() && "Can only unswitch a conditional branch!"); - DEBUG(dbgs() << " Trying to unswitch branch: " << BI << "\n"); + LLVM_DEBUG(dbgs() << " Trying to unswitch branch: " << BI << "\n"); Value *LoopCond = BI.getCondition(); @@ -212,8 +212,8 @@ static bool unswitchTrivialBranch(Loop &L, BranchInst &BI, DominatorTree &DT, if (!areLoopExitPHIsLoopInvariant(L, *ParentBB, *LoopExitBB)) return false; - DEBUG(dbgs() << " unswitching trivial branch when: " << CondVal - << " == " << LoopCond << "\n"); + LLVM_DEBUG(dbgs() << " unswitching trivial branch when: " << CondVal + << " == " << LoopCond << "\n"); // Split the preheader, so that we know that there is a safe place to insert // the conditional branch. We will change the preheader to have a conditional @@ -292,7 +292,7 @@ static bool unswitchTrivialBranch(Loop &L, BranchInst &BI, DominatorTree &DT, /// branch. Still more cleanup can be done with some simplify-cfg like pass. static bool unswitchTrivialSwitch(Loop &L, SwitchInst &SI, DominatorTree &DT, LoopInfo &LI) { - DEBUG(dbgs() << " Trying to unswitch switch: " << SI << "\n"); + LLVM_DEBUG(dbgs() << " Trying to unswitch switch: " << SI << "\n"); Value *LoopCond = SI.getCondition(); // If this isn't switching on an invariant condition, we can't unswitch it. @@ -316,7 +316,7 @@ static bool unswitchTrivialSwitch(Loop &L, SwitchInst &SI, DominatorTree &DT, else if (ExitCaseIndices.empty()) return false; - DEBUG(dbgs() << " unswitching trivial cases...\n"); + LLVM_DEBUG(dbgs() << " unswitching trivial cases...\n"); SmallVector<std::pair<ConstantInt *, BasicBlock *>, 4> ExitCases; ExitCases.reserve(ExitCaseIndices.size()); @@ -1798,8 +1798,9 @@ unswitchLoop(Loop &L, DominatorTree &DT, LoopInfo &LI, AssumptionCache &AC, if (containsIrreducibleCFG<const BasicBlock *>(RPOT, LI)) return Changed; - DEBUG(dbgs() << "Considering " << UnswitchCandidates.size() - << " non-trivial loop invariant conditions for unswitching.\n"); + LLVM_DEBUG( + dbgs() << "Considering " << UnswitchCandidates.size() + << " non-trivial loop invariant conditions for unswitching.\n"); // Given that unswitching these terminators will require duplicating parts of // the loop, so we need to be able to model that cost. Compute the ephemeral @@ -1835,7 +1836,7 @@ unswitchLoop(Loop &L, DominatorTree &DT, LoopInfo &LI, AssumptionCache &AC, assert(LoopCost >= 0 && "Must not have negative loop costs!"); BBCostMap[BB] = Cost; } - DEBUG(dbgs() << " Total loop cost: " << LoopCost << "\n"); + LLVM_DEBUG(dbgs() << " Total loop cost: " << LoopCost << "\n"); // Now we find the best candidate by searching for the one with the following // properties in order: @@ -1889,8 +1890,8 @@ unswitchLoop(Loop &L, DominatorTree &DT, LoopInfo &LI, AssumptionCache &AC, int BestUnswitchCost; for (TerminatorInst *CandidateTI : UnswitchCandidates) { int CandidateCost = ComputeUnswitchedCost(CandidateTI); - DEBUG(dbgs() << " Computed cost of " << CandidateCost - << " for unswitch candidate: " << *CandidateTI << "\n"); + LLVM_DEBUG(dbgs() << " Computed cost of " << CandidateCost + << " for unswitch candidate: " << *CandidateTI << "\n"); if (!BestUnswitchTI || CandidateCost < BestUnswitchCost) { BestUnswitchTI = CandidateTI; BestUnswitchCost = CandidateCost; @@ -1898,14 +1899,14 @@ unswitchLoop(Loop &L, DominatorTree &DT, LoopInfo &LI, AssumptionCache &AC, } if (BestUnswitchCost < UnswitchThreshold) { - DEBUG(dbgs() << " Trying to unswitch non-trivial (cost = " - << BestUnswitchCost << ") branch: " << *BestUnswitchTI - << "\n"); + LLVM_DEBUG(dbgs() << " Trying to unswitch non-trivial (cost = " + << BestUnswitchCost << ") branch: " << *BestUnswitchTI + << "\n"); Changed |= unswitchInvariantBranch(L, cast<BranchInst>(*BestUnswitchTI), DT, LI, AC, NonTrivialUnswitchCB); } else { - DEBUG(dbgs() << "Cannot unswitch, lowest cost found: " << BestUnswitchCost - << "\n"); + LLVM_DEBUG(dbgs() << "Cannot unswitch, lowest cost found: " + << BestUnswitchCost << "\n"); } return Changed; @@ -1917,7 +1918,8 @@ PreservedAnalyses SimpleLoopUnswitchPass::run(Loop &L, LoopAnalysisManager &AM, Function &F = *L.getHeader()->getParent(); (void)F; - DEBUG(dbgs() << "Unswitching loop in " << F.getName() << ": " << L << "\n"); + LLVM_DEBUG(dbgs() << "Unswitching loop in " << F.getName() << ": " << L + << "\n"); // Save the current loop name in a variable so that we can report it even // after it has been deleted. @@ -1977,7 +1979,8 @@ bool SimpleLoopUnswitchLegacyPass::runOnLoop(Loop *L, LPPassManager &LPM) { Function &F = *L->getHeader()->getParent(); - DEBUG(dbgs() << "Unswitching loop in " << F.getName() << ": " << *L << "\n"); + LLVM_DEBUG(dbgs() << "Unswitching loop in " << F.getName() << ": " << *L + << "\n"); auto &DT = getAnalysis<DominatorTreeWrapperPass>().getDomTree(); auto &LI = getAnalysis<LoopInfoWrapperPass>().getLoopInfo(); diff --git a/llvm/lib/Transforms/Scalar/Sink.cpp b/llvm/lib/Transforms/Scalar/Sink.cpp index 81176288049..ca6b93e0b4a 100644 --- a/llvm/lib/Transforms/Scalar/Sink.cpp +++ b/llvm/lib/Transforms/Scalar/Sink.cpp @@ -187,11 +187,9 @@ static bool SinkInstruction(Instruction *Inst, if (!SuccToSinkTo) return false; - DEBUG(dbgs() << "Sink" << *Inst << " ("; - Inst->getParent()->printAsOperand(dbgs(), false); - dbgs() << " -> "; - SuccToSinkTo->printAsOperand(dbgs(), false); - dbgs() << ")\n"); + LLVM_DEBUG(dbgs() << "Sink" << *Inst << " ("; + Inst->getParent()->printAsOperand(dbgs(), false); dbgs() << " -> "; + SuccToSinkTo->printAsOperand(dbgs(), false); dbgs() << ")\n"); // Move the instruction. Inst->moveBefore(&*SuccToSinkTo->getFirstInsertionPt()); @@ -244,7 +242,7 @@ static bool iterativelySinkInstructions(Function &F, DominatorTree &DT, do { MadeChange = false; - DEBUG(dbgs() << "Sinking iteration " << NumSinkIter << "\n"); + LLVM_DEBUG(dbgs() << "Sinking iteration " << NumSinkIter << "\n"); // Process all basic blocks. for (BasicBlock &I : F) MadeChange |= ProcessBlock(I, DT, LI, AA); diff --git a/llvm/lib/Transforms/Scalar/SpeculateAroundPHIs.cpp b/llvm/lib/Transforms/Scalar/SpeculateAroundPHIs.cpp index 23156d5a4d8..3b0971804e6 100644 --- a/llvm/lib/Transforms/Scalar/SpeculateAroundPHIs.cpp +++ b/llvm/lib/Transforms/Scalar/SpeculateAroundPHIs.cpp @@ -64,7 +64,7 @@ isSafeToSpeculatePHIUsers(PHINode &PN, DominatorTree &DT, // block. We should consider using actual post-dominance here in the // future. if (UI->getParent() != PhiBB) { - DEBUG(dbgs() << " Unsafe: use in a different BB: " << *UI << "\n"); + LLVM_DEBUG(dbgs() << " Unsafe: use in a different BB: " << *UI << "\n"); return false; } @@ -75,7 +75,7 @@ isSafeToSpeculatePHIUsers(PHINode &PN, DominatorTree &DT, // probably change this to do at least a limited scan of the intervening // instructions and allow handling stores in easily proven safe cases. if (mayBeMemoryDependent(*UI)) { - DEBUG(dbgs() << " Unsafe: can't speculate use: " << *UI << "\n"); + LLVM_DEBUG(dbgs() << " Unsafe: can't speculate use: " << *UI << "\n"); return false; } @@ -126,8 +126,8 @@ isSafeToSpeculatePHIUsers(PHINode &PN, DominatorTree &DT, // If when we directly test whether this is safe it fails, bail. if (UnsafeSet.count(OpI) || ParentBB != PhiBB || mayBeMemoryDependent(*OpI)) { - DEBUG(dbgs() << " Unsafe: can't speculate transitive use: " << *OpI - << "\n"); + LLVM_DEBUG(dbgs() << " Unsafe: can't speculate transitive use: " + << *OpI << "\n"); // Record the stack of instructions which reach this node as unsafe // so we prune subsequent searches. UnsafeSet.insert(OpI); @@ -229,7 +229,7 @@ static bool isSafeAndProfitableToSpeculateAroundPHI( NonFreeMat |= MatCost != TTI.TCC_Free; } if (!NonFreeMat) { - DEBUG(dbgs() << " Free: " << PN << "\n"); + LLVM_DEBUG(dbgs() << " Free: " << PN << "\n"); // No profit in free materialization. return false; } @@ -237,7 +237,7 @@ static bool isSafeAndProfitableToSpeculateAroundPHI( // Now check that the uses of this PHI can actually be speculated, // otherwise we'll still have to materialize the PHI value. if (!isSafeToSpeculatePHIUsers(PN, DT, PotentialSpecSet, UnsafeSet)) { - DEBUG(dbgs() << " Unsafe PHI: " << PN << "\n"); + LLVM_DEBUG(dbgs() << " Unsafe PHI: " << PN << "\n"); return false; } @@ -288,9 +288,13 @@ static bool isSafeAndProfitableToSpeculateAroundPHI( // just bail. We're only interested in cases where folding the incoming // constants is at least break-even on all paths. if (FoldedCost > MatCost) { - DEBUG(dbgs() << " Not profitable to fold imm: " << *IncomingC << "\n" - " Materializing cost: " << MatCost << "\n" - " Accumulated folded cost: " << FoldedCost << "\n"); + LLVM_DEBUG(dbgs() << " Not profitable to fold imm: " << *IncomingC + << "\n" + " Materializing cost: " + << MatCost + << "\n" + " Accumulated folded cost: " + << FoldedCost << "\n"); return false; } } @@ -310,8 +314,8 @@ static bool isSafeAndProfitableToSpeculateAroundPHI( "less that its materialized cost, " "the sum must be as well."); - DEBUG(dbgs() << " Cost savings " << (TotalMatCost - TotalFoldedCost) - << ": " << PN << "\n"); + LLVM_DEBUG(dbgs() << " Cost savings " << (TotalMatCost - TotalFoldedCost) + << ": " << PN << "\n"); CostSavingsMap[&PN] = TotalMatCost - TotalFoldedCost; return true; } @@ -489,9 +493,13 @@ findProfitablePHIs(ArrayRef<PHINode *> PNs, // and zero out the cost of everything it depends on. int CostSavings = CostSavingsMap.find(PN)->second; if (SpecCost > CostSavings) { - DEBUG(dbgs() << " Not profitable, speculation cost: " << *PN << "\n" - " Cost savings: " << CostSavings << "\n" - " Speculation cost: " << SpecCost << "\n"); + LLVM_DEBUG(dbgs() << " Not profitable, speculation cost: " << *PN + << "\n" + " Cost savings: " + << CostSavings + << "\n" + " Speculation cost: " + << SpecCost << "\n"); continue; } @@ -545,7 +553,7 @@ static void speculatePHIs(ArrayRef<PHINode *> SpecPNs, SmallPtrSetImpl<Instruction *> &PotentialSpecSet, SmallSetVector<BasicBlock *, 16> &PredSet, DominatorTree &DT) { - DEBUG(dbgs() << " Speculating around " << SpecPNs.size() << " PHIs!\n"); + LLVM_DEBUG(dbgs() << " Speculating around " << SpecPNs.size() << " PHIs!\n"); NumPHIsSpeculated += SpecPNs.size(); // Split any critical edges so that we have a block to hoist into. @@ -558,8 +566,8 @@ static void speculatePHIs(ArrayRef<PHINode *> SpecPNs, CriticalEdgeSplittingOptions(&DT).setMergeIdenticalEdges()); if (NewPredBB) { ++NumEdgesSplit; - DEBUG(dbgs() << " Split critical edge from: " << PredBB->getName() - << "\n"); + LLVM_DEBUG(dbgs() << " Split critical edge from: " << PredBB->getName() + << "\n"); SpecPreds.push_back(NewPredBB); } else { assert(PredBB->getSingleSuccessor() == ParentBB && @@ -593,8 +601,9 @@ static void speculatePHIs(ArrayRef<PHINode *> SpecPNs, int NumSpecInsts = SpecList.size() * SpecPreds.size(); int NumRedundantInsts = NumSpecInsts - SpecList.size(); - DEBUG(dbgs() << " Inserting " << NumSpecInsts << " speculated instructions, " - << NumRedundantInsts << " redundancies\n"); + LLVM_DEBUG(dbgs() << " Inserting " << NumSpecInsts + << " speculated instructions, " << NumRedundantInsts + << " redundancies\n"); NumSpeculatedInstructions += NumSpecInsts; NumNewRedundantInstructions += NumRedundantInsts; @@ -716,7 +725,7 @@ static void speculatePHIs(ArrayRef<PHINode *> SpecPNs, /// true when at least some speculation occurs. static bool tryToSpeculatePHIs(SmallVectorImpl<PHINode *> &PNs, DominatorTree &DT, TargetTransformInfo &TTI) { - DEBUG(dbgs() << "Evaluating phi nodes for speculation:\n"); + LLVM_DEBUG(dbgs() << "Evaluating phi nodes for speculation:\n"); // Savings in cost from speculating around a PHI node. SmallDenseMap<PHINode *, int, 16> CostSavingsMap; @@ -745,7 +754,7 @@ static bool tryToSpeculatePHIs(SmallVectorImpl<PHINode *> &PNs, PNs.end()); // If no PHIs were profitable, skip. if (PNs.empty()) { - DEBUG(dbgs() << " No safe and profitable PHIs found!\n"); + LLVM_DEBUG(dbgs() << " No safe and profitable PHIs found!\n"); return false; } @@ -763,13 +772,13 @@ static bool tryToSpeculatePHIs(SmallVectorImpl<PHINode *> &PNs, // differently. if (isa<IndirectBrInst>(PredBB->getTerminator()) || isa<InvokeInst>(PredBB->getTerminator())) { - DEBUG(dbgs() << " Invalid: predecessor terminator: " << PredBB->getName() - << "\n"); + LLVM_DEBUG(dbgs() << " Invalid: predecessor terminator: " + << PredBB->getName() << "\n"); return false; } } if (PredSet.size() < 2) { - DEBUG(dbgs() << " Unimportant: phi with only one predecessor\n"); + LLVM_DEBUG(dbgs() << " Unimportant: phi with only one predecessor\n"); return false; } diff --git a/llvm/lib/Transforms/Scalar/SpeculativeExecution.cpp b/llvm/lib/Transforms/Scalar/SpeculativeExecution.cpp index a7c308b5987..488c4df2b19 100644 --- a/llvm/lib/Transforms/Scalar/SpeculativeExecution.cpp +++ b/llvm/lib/Transforms/Scalar/SpeculativeExecution.cpp @@ -151,8 +151,8 @@ namespace llvm { bool SpeculativeExecutionPass::runImpl(Function &F, TargetTransformInfo *TTI) { if (OnlyIfDivergentTarget && !TTI->hasBranchDivergence()) { - DEBUG(dbgs() << "Not running SpeculativeExecution because " - "TTI->hasBranchDivergence() is false.\n"); + LLVM_DEBUG(dbgs() << "Not running SpeculativeExecution because " + "TTI->hasBranchDivergence() is false.\n"); return false; } diff --git a/llvm/lib/Transforms/Scalar/StructurizeCFG.cpp b/llvm/lib/Transforms/Scalar/StructurizeCFG.cpp index 25816f70c56..f0ddc30beff 100644 --- a/llvm/lib/Transforms/Scalar/StructurizeCFG.cpp +++ b/llvm/lib/Transforms/Scalar/StructurizeCFG.cpp @@ -489,10 +489,10 @@ void StructurizeCFG::collectInfos() { Visited.clear(); for (RegionNode *RN : reverse(Order)) { - DEBUG(dbgs() << "Visiting: " - << (RN->isSubRegion() ? "SubRegion with entry: " : "") - << RN->getEntry()->getName() << " Loop Depth: " - << LI->getLoopDepth(RN->getEntry()) << "\n"); + LLVM_DEBUG(dbgs() << "Visiting: " + << (RN->isSubRegion() ? "SubRegion with entry: " : "") + << RN->getEntry()->getName() << " Loop Depth: " + << LI->getLoopDepth(RN->getEntry()) << "\n"); // Analyze all the conditions leading to a node gatherPredicates(RN); @@ -901,8 +901,8 @@ static bool hasOnlyUniformBranches(Region *R, unsigned UniformMDKindID, if (!DA.isUniform(Br)) return false; - DEBUG(dbgs() << "BB: " << Br->getParent()->getName() - << " has uniform terminator\n"); + LLVM_DEBUG(dbgs() << "BB: " << Br->getParent()->getName() + << " has uniform terminator\n"); } else { // Explicitly refuse to treat regions as uniform if they have non-uniform // subregions. We cannot rely on DivergenceAnalysis for branches in @@ -943,7 +943,8 @@ bool StructurizeCFG::runOnRegion(Region *R, RGPassManager &RGM) { DA = &getAnalysis<DivergenceAnalysis>(); if (hasOnlyUniformBranches(R, UniformMDKindID, *DA)) { - DEBUG(dbgs() << "Skipping region with uniform control flow: " << *R << '\n'); + LLVM_DEBUG(dbgs() << "Skipping region with uniform control flow: " << *R + << '\n'); // Mark all direct child block terminators as having been treated as // uniform. To account for a possible future in which non-uniform diff --git a/llvm/lib/Transforms/Scalar/TailRecursionElimination.cpp b/llvm/lib/Transforms/Scalar/TailRecursionElimination.cpp index 37ea4375a4c..f8cd6c17a5a 100644 --- a/llvm/lib/Transforms/Scalar/TailRecursionElimination.cpp +++ b/llvm/lib/Transforms/Scalar/TailRecursionElimination.cpp @@ -302,7 +302,7 @@ static bool markTails(Function &F, bool &AllCallsAreTailCalls, if (Visited[CI->getParent()] != ESCAPED) { // If the escape point was part way through the block, calls after the // escape point wouldn't have been put into DeferredTails. - DEBUG(dbgs() << "Marked as tail call candidate: " << *CI << "\n"); + LLVM_DEBUG(dbgs() << "Marked as tail call candidate: " << *CI << "\n"); CI->setTailCall(); Modified = true; } else { @@ -699,8 +699,8 @@ static bool foldReturnAndProcessPred( BranchInst *BI = UncondBranchPreds.pop_back_val(); BasicBlock *Pred = BI->getParent(); if (CallInst *CI = findTRECandidate(BI, CannotTailCallElimCallsMarkedTail, TTI)){ - DEBUG(dbgs() << "FOLDING: " << *BB - << "INTO UNCOND BRANCH PRED: " << *Pred); + LLVM_DEBUG(dbgs() << "FOLDING: " << *BB + << "INTO UNCOND BRANCH PRED: " << *Pred); ReturnInst *RI = FoldReturnIntoUncondBranch(Ret, BB, Pred); // Cleanup: if all predecessors of BB have been eliminated by |
