//===-- LoopUtils.cpp - Loop Utility functions -------------------------===// // // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. // See https://llvm.org/LICENSE.txt for license information. // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception // //===----------------------------------------------------------------------===// // // This file defines common loop utility functions. // //===----------------------------------------------------------------------===// #include "llvm/Transforms/Utils/LoopUtils.h" #include "llvm/ADT/DenseSet.h" #include "llvm/ADT/Optional.h" #include "llvm/ADT/PriorityWorklist.h" #include "llvm/ADT/ScopeExit.h" #include "llvm/ADT/SetVector.h" #include "llvm/ADT/SmallPtrSet.h" #include "llvm/ADT/SmallVector.h" #include "llvm/Analysis/AliasAnalysis.h" #include "llvm/Analysis/BasicAliasAnalysis.h" #include "llvm/Analysis/DomTreeUpdater.h" #include "llvm/Analysis/GlobalsModRef.h" #include "llvm/Analysis/InstSimplifyFolder.h" #include "llvm/Analysis/InstructionSimplify.h" #include "llvm/Analysis/LoopAccessAnalysis.h" #include "llvm/Analysis/LoopInfo.h" #include "llvm/Analysis/LoopPass.h" #include "llvm/Analysis/MemorySSA.h" #include "llvm/Analysis/MemorySSAUpdater.h" #include "llvm/Analysis/MustExecute.h" #include "llvm/Analysis/ScalarEvolution.h" #include "llvm/Analysis/ScalarEvolutionAliasAnalysis.h" #include "llvm/Analysis/ScalarEvolutionExpressions.h" #include "llvm/Analysis/TargetTransformInfo.h" #include "llvm/Analysis/ValueTracking.h" #include "llvm/IR/DIBuilder.h" #include "llvm/IR/Dominators.h" #include "llvm/IR/Instructions.h" #include "llvm/IR/IntrinsicInst.h" #include "llvm/IR/MDBuilder.h" #include "llvm/IR/Module.h" #include "llvm/IR/Operator.h" #include "llvm/IR/PatternMatch.h" #include "llvm/IR/ValueHandle.h" #include "llvm/InitializePasses.h" #include "llvm/Pass.h" #include "llvm/Support/Debug.h" #include "llvm/Support/KnownBits.h" #include "llvm/Transforms/Utils/BasicBlockUtils.h" #include "llvm/Transforms/Utils/Local.h" #include "llvm/Transforms/Utils/ScalarEvolutionExpander.h" using namespace llvm; using namespace llvm::PatternMatch; #define DEBUG_TYPE "loop-utils" static const char *LLVMLoopDisableNonforced = "llvm.loop.disable_nonforced"; static const char *LLVMLoopDisableLICM = "llvm.licm.disable"; bool llvm::formDedicatedExitBlocks(Loop *L, DominatorTree *DT, LoopInfo *LI, MemorySSAUpdater *MSSAU, bool PreserveLCSSA) { bool Changed = false; // We re-use a vector for the in-loop predecesosrs. SmallVector InLoopPredecessors; auto RewriteExit = [&](BasicBlock *BB) { assert(InLoopPredecessors.empty() && "Must start with an empty predecessors list!"); auto Cleanup = make_scope_exit([&] { InLoopPredecessors.clear(); }); // See if there are any non-loop predecessors of this exit block and // keep track of the in-loop predecessors. bool IsDedicatedExit = true; for (auto *PredBB : predecessors(BB)) if (L->contains(PredBB)) { if (isa(PredBB->getTerminator())) // We cannot rewrite exiting edges from an indirectbr. return false; if (isa(PredBB->getTerminator())) // We cannot rewrite exiting edges from a callbr. return false; InLoopPredecessors.push_back(PredBB); } else { IsDedicatedExit = false; } assert(!InLoopPredecessors.empty() && "Must have *some* loop predecessor!"); // Nothing to do if this is already a dedicated exit. if (IsDedicatedExit) return false; auto *NewExitBB = SplitBlockPredecessors( BB, InLoopPredecessors, ".loopexit", DT, LI, MSSAU, PreserveLCSSA); if (!NewExitBB) LLVM_DEBUG( dbgs() << "WARNING: Can't create a dedicated exit block for loop: " << *L << "\n"); else LLVM_DEBUG(dbgs() << "LoopSimplify: Creating dedicated exit block " << NewExitBB->getName() << "\n"); return true; }; // Walk the exit blocks directly rather than building up a data structure for // them, but only visit each one once. SmallPtrSet Visited; for (auto *BB : L->blocks()) for (auto *SuccBB : successors(BB)) { // We're looking for exit blocks so skip in-loop successors. if (L->contains(SuccBB)) continue; // Visit each exit block exactly once. if (!Visited.insert(SuccBB).second) continue; Changed |= RewriteExit(SuccBB); } return Changed; } /// Returns the instructions that use values defined in the loop. SmallVector llvm::findDefsUsedOutsideOfLoop(Loop *L) { SmallVector UsedOutside; for (auto *Block : L->getBlocks()) // FIXME: I believe that this could use copy_if if the Inst reference could // be adapted into a pointer. for (auto &Inst : *Block) { auto Users = Inst.users(); if (any_of(Users, [&](User *U) { auto *Use = cast(U); return !L->contains(Use->getParent()); })) UsedOutside.push_back(&Inst); } return UsedOutside; } void llvm::getLoopAnalysisUsage(AnalysisUsage &AU) { // By definition, all loop passes need the LoopInfo analysis and the // Dominator tree it depends on. Because they all participate in the loop // pass manager, they must also preserve these. AU.addRequired(); AU.addPreserved(); AU.addRequired(); AU.addPreserved(); // We must also preserve LoopSimplify and LCSSA. We locally access their IDs // here because users shouldn't directly get them from this header. extern char &LoopSimplifyID; extern char &LCSSAID; AU.addRequiredID(LoopSimplifyID); AU.addPreservedID(LoopSimplifyID); AU.addRequiredID(LCSSAID); AU.addPreservedID(LCSSAID); // This is used in the LPPassManager to perform LCSSA verification on passes // which preserve lcssa form AU.addRequired(); AU.addPreserved(); // Loop passes are designed to run inside of a loop pass manager which means // that any function analyses they require must be required by the first loop // pass in the manager (so that it is computed before the loop pass manager // runs) and preserved by all loop pasess in the manager. To make this // reasonably robust, the set needed for most loop passes is maintained here. // If your loop pass requires an analysis not listed here, you will need to // carefully audit the loop pass manager nesting structure that results. AU.addRequired(); AU.addPreserved(); AU.addPreserved(); AU.addPreserved(); AU.addPreserved(); AU.addRequired(); AU.addPreserved(); // FIXME: When all loop passes preserve MemorySSA, it can be required and // preserved here instead of the individual handling in each pass. } /// Manually defined generic "LoopPass" dependency initialization. This is used /// to initialize the exact set of passes from above in \c /// getLoopAnalysisUsage. It can be used within a loop pass's initialization /// with: /// /// INITIALIZE_PASS_DEPENDENCY(LoopPass) /// /// As-if "LoopPass" were a pass. void llvm::initializeLoopPassPass(PassRegistry &Registry) { INITIALIZE_PASS_DEPENDENCY(DominatorTreeWrapperPass) INITIALIZE_PASS_DEPENDENCY(LoopInfoWrapperPass) INITIALIZE_PASS_DEPENDENCY(LoopSimplify) INITIALIZE_PASS_DEPENDENCY(LCSSAWrapperPass) INITIALIZE_PASS_DEPENDENCY(AAResultsWrapperPass) INITIALIZE_PASS_DEPENDENCY(BasicAAWrapperPass) INITIALIZE_PASS_DEPENDENCY(GlobalsAAWrapperPass) INITIALIZE_PASS_DEPENDENCY(SCEVAAWrapperPass) INITIALIZE_PASS_DEPENDENCY(ScalarEvolutionWrapperPass) INITIALIZE_PASS_DEPENDENCY(MemorySSAWrapperPass) } /// Create MDNode for input string. static MDNode *createStringMetadata(Loop *TheLoop, StringRef Name, unsigned V) { LLVMContext &Context = TheLoop->getHeader()->getContext(); Metadata *MDs[] = { MDString::get(Context, Name), ConstantAsMetadata::get(ConstantInt::get(Type::getInt32Ty(Context), V))}; return MDNode::get(Context, MDs); } /// Set input string into loop metadata by keeping other values intact. /// If the string is already in loop metadata update value if it is /// different. void llvm::addStringMetadataToLoop(Loop *TheLoop, const char *StringMD, unsigned V) { SmallVector MDs(1); // If the loop already has metadata, retain it. MDNode *LoopID = TheLoop->getLoopID(); if (LoopID) { for (unsigned i = 1, ie = LoopID->getNumOperands(); i < ie; ++i) { MDNode *Node = cast(LoopID->getOperand(i)); // If it is of form key = value, try to parse it. if (Node->getNumOperands() == 2) { MDString *S = dyn_cast(Node->getOperand(0)); if (S && S->getString().equals(StringMD)) { ConstantInt *IntMD = mdconst::extract_or_null(Node->getOperand(1)); if (IntMD && IntMD->getSExtValue() == V) // It is already in place. Do nothing. return; // We need to update the value, so just skip it here and it will // be added after copying other existed nodes. continue; } } MDs.push_back(Node); } } // Add new metadata. MDs.push_back(createStringMetadata(TheLoop, StringMD, V)); // Replace current metadata node with new one. LLVMContext &Context = TheLoop->getHeader()->getContext(); MDNode *NewLoopID = MDNode::get(Context, MDs); // Set operand 0 to refer to the loop id itself. NewLoopID->replaceOperandWith(0, NewLoopID); TheLoop->setLoopID(NewLoopID); } Optional llvm::getOptionalElementCountLoopAttribute(const Loop *TheLoop) { Optional Width = getOptionalIntLoopAttribute(TheLoop, "llvm.loop.vectorize.width"); if (Width.hasValue()) { Optional IsScalable = getOptionalIntLoopAttribute( TheLoop, "llvm.loop.vectorize.scalable.enable"); return ElementCount::get(*Width, IsScalable.getValueOr(false)); } return None; } Optional llvm::makeFollowupLoopID( MDNode *OrigLoopID, ArrayRef FollowupOptions, const char *InheritOptionsExceptPrefix, bool AlwaysNew) { if (!OrigLoopID) { if (AlwaysNew) return nullptr; return None; } assert(OrigLoopID->getOperand(0) == OrigLoopID); bool InheritAllAttrs = !InheritOptionsExceptPrefix; bool InheritSomeAttrs = InheritOptionsExceptPrefix && InheritOptionsExceptPrefix[0] != '\0'; SmallVector MDs; MDs.push_back(nullptr); bool Changed = false; if (InheritAllAttrs || InheritSomeAttrs) { for (const MDOperand &Existing : drop_begin(OrigLoopID->operands())) { MDNode *Op = cast(Existing.get()); auto InheritThisAttribute = [InheritSomeAttrs, InheritOptionsExceptPrefix](MDNode *Op) { if (!InheritSomeAttrs) return false; // Skip malformatted attribute metadata nodes. if (Op->getNumOperands() == 0) return true; Metadata *NameMD = Op->getOperand(0).get(); if (!isa(NameMD)) return true; StringRef AttrName = cast(NameMD)->getString(); // Do not inherit excluded attributes. return !AttrName.startswith(InheritOptionsExceptPrefix); }; if (InheritThisAttribute(Op)) MDs.push_back(Op); else Changed = true; } } else { // Modified if we dropped at least one attribute. Changed = OrigLoopID->getNumOperands() > 1; } bool HasAnyFollowup = false; for (StringRef OptionName : FollowupOptions) { MDNode *FollowupNode = findOptionMDForLoopID(OrigLoopID, OptionName); if (!FollowupNode) continue; HasAnyFollowup = true; for (const MDOperand &Option : drop_begin(FollowupNode->operands())) { MDs.push_back(Option.get()); Changed = true; } } // Attributes of the followup loop not specified explicity, so signal to the // transformation pass to add suitable attributes. if (!AlwaysNew && !HasAnyFollowup) return None; // If no attributes were added or remove, the previous loop Id can be reused. if (!AlwaysNew && !Changed) return OrigLoopID; // No attributes is equivalent to having no !llvm.loop metadata at all. if (MDs.size() == 1) return nullptr; // Build the new loop ID. MDTuple *FollowupLoopID = MDNode::get(OrigLoopID->getContext(), MDs); FollowupLoopID->replaceOperandWith(0, FollowupLoopID); return FollowupLoopID; } bool llvm::hasDisableAllTransformsHint(const Loop *L) { return getBooleanLoopAttribute(L, LLVMLoopDisableNonforced); } bool llvm::hasDisableLICMTransformsHint(const Loop *L) { return getBooleanLoopAttribute(L, LLVMLoopDisableLICM); } TransformationMode llvm::hasUnrollTransformation(const Loop *L) { if (getBooleanLoopAttribute(L, "llvm.loop.unroll.disable")) return TM_SuppressedByUser; Optional Count = getOptionalIntLoopAttribute(L, "llvm.loop.unroll.count"); if (Count.hasValue()) return Count.getValue() == 1 ? TM_SuppressedByUser : TM_ForcedByUser; if (getBooleanLoopAttribute(L, "llvm.loop.unroll.enable")) return TM_ForcedByUser; if (getBooleanLoopAttribute(L, "llvm.loop.unroll.full")) return TM_ForcedByUser; if (hasDisableAllTransformsHint(L)) return TM_Disable; return TM_Unspecified; } TransformationMode llvm::hasUnrollAndJamTransformation(const Loop *L) { if (getBooleanLoopAttribute(L, "llvm.loop.unroll_and_jam.disable")) return TM_SuppressedByUser; Optional Count = getOptionalIntLoopAttribute(L, "llvm.loop.unroll_and_jam.count"); if (Count.hasValue()) return Count.getValue() == 1 ? TM_SuppressedByUser : TM_ForcedByUser; if (getBooleanLoopAttribute(L, "llvm.loop.unroll_and_jam.enable")) return TM_ForcedByUser; if (hasDisableAllTransformsHint(L)) return TM_Disable; return TM_Unspecified; } TransformationMode llvm::hasVectorizeTransformation(const Loop *L) { Optional Enable = getOptionalBoolLoopAttribute(L, "llvm.loop.vectorize.enable"); if (Enable == false) return TM_SuppressedByUser; Optional VectorizeWidth = getOptionalElementCountLoopAttribute(L); Optional InterleaveCount = getOptionalIntLoopAttribute(L, "llvm.loop.interleave.count"); // 'Forcing' vector width and interleave count to one effectively disables // this tranformation. if (Enable == true && VectorizeWidth && VectorizeWidth->isScalar() && InterleaveCount == 1) return TM_SuppressedByUser; if (getBooleanLoopAttribute(L, "llvm.loop.isvectorized")) return TM_Disable; if (Enable == true) return TM_ForcedByUser; if ((VectorizeWidth && VectorizeWidth->isScalar()) && InterleaveCount == 1) return TM_Disable; if ((VectorizeWidth && VectorizeWidth->isVector()) || InterleaveCount > 1) return TM_Enable; if (hasDisableAllTransformsHint(L)) return TM_Disable; return TM_Unspecified; } TransformationMode llvm::hasDistributeTransformation(const Loop *L) { if (getBooleanLoopAttribute(L, "llvm.loop.distribute.enable")) return TM_ForcedByUser; if (hasDisableAllTransformsHint(L)) return TM_Disable; return TM_Unspecified; } TransformationMode llvm::hasLICMVersioningTransformation(const Loop *L) { if (getBooleanLoopAttribute(L, "llvm.loop.licm_versioning.disable")) return TM_SuppressedByUser; if (hasDisableAllTransformsHint(L)) return TM_Disable; return TM_Unspecified; } /// Does a BFS from a given node to all of its children inside a given loop. /// The returned vector of nodes includes the starting point. SmallVector llvm::collectChildrenInLoop(DomTreeNode *N, const Loop *CurLoop) { SmallVector Worklist; auto AddRegionToWorklist = [&](DomTreeNode *DTN) { // Only include subregions in the top level loop. BasicBlock *BB = DTN->getBlock(); if (CurLoop->contains(BB)) Worklist.push_back(DTN); }; AddRegionToWorklist(N); for (size_t I = 0; I < Worklist.size(); I++) { for (DomTreeNode *Child : Worklist[I]->children()) AddRegionToWorklist(Child); } return Worklist; } void llvm::deleteDeadLoop(Loop *L, DominatorTree *DT, ScalarEvolution *SE, LoopInfo *LI, MemorySSA *MSSA) { assert((!DT || L->isLCSSAForm(*DT)) && "Expected LCSSA!"); auto *Preheader = L->getLoopPreheader(); assert(Preheader && "Preheader should exist!"); std::unique_ptr MSSAU; if (MSSA) MSSAU = std::make_unique(MSSA); // Now that we know the removal is safe, remove the loop by changing the // branch from the preheader to go to the single exit block. // // Because we're deleting a large chunk of code at once, the sequence in which // we remove things is very important to avoid invalidation issues. // Tell ScalarEvolution that the loop is deleted. Do this before // deleting the loop so that ScalarEvolution can look at the loop // to determine what it needs to clean up. if (SE) SE->forgetLoop(L); auto *OldBr = dyn_cast(Preheader->getTerminator()); assert(OldBr && "Preheader must end with a branch"); assert(OldBr->isUnconditional() && "Preheader must have a single successor"); // Connect the preheader to the exit block. Keep the old edge to the header // around to perform the dominator tree update in two separate steps // -- #1 insertion of the edge preheader -> exit and #2 deletion of the edge // preheader -> header. // // // 0. Preheader 1. Preheader 2. Preheader // | | | | // V | V | // Header <--\ | Header <--\ | Header <--\ // | | | | | | | | | | | // | V | | | V | | | V | // | Body --/ | | Body --/ | | Body --/ // V V V V V // Exit Exit Exit // // By doing this is two separate steps we can perform the dominator tree // update without using the batch update API. // // Even when the loop is never executed, we cannot remove the edge from the // source block to the exit block. Consider the case where the unexecuted loop // branches back to an outer loop. If we deleted the loop and removed the edge // coming to this inner loop, this will break the outer loop structure (by // deleting the backedge of the outer loop). If the outer loop is indeed a // non-loop, it will be deleted in a future iteration of loop deletion pass. IRBuilder<> Builder(OldBr); auto *ExitBlock = L->getUniqueExitBlock(); DomTreeUpdater DTU(DT, DomTreeUpdater::UpdateStrategy::Eager); if (ExitBlock) { assert(ExitBlock && "Should have a unique exit block!"); assert(L->hasDedicatedExits() && "Loop should have dedicated exits!"); Builder.CreateCondBr(Builder.getFalse(), L->getHeader(), ExitBlock); // Remove the old branch. The conditional branch becomes a new terminator. OldBr->eraseFromParent(); // Rewrite phis in the exit block to get their inputs from the Preheader // instead of the exiting block. for (PHINode &P : ExitBlock->phis()) { // Set the zero'th element of Phi to be from the preheader and remove all // other incoming values. Given the loop has dedicated exits, all other // incoming values must be from the exiting blocks. int PredIndex = 0; P.setIncomingBlock(PredIndex, Preheader); // Removes all incoming values from all other exiting blocks (including // duplicate values from an exiting block). // Nuke all entries except the zero'th entry which is the preheader entry. // NOTE! We need to remove Incoming Values in the reverse order as done // below, to keep the indices valid for deletion (removeIncomingValues // updates getNumIncomingValues and shifts all values down into the // operand being deleted). for (unsigned i = 0, e = P.getNumIncomingValues() - 1; i != e; ++i) P.removeIncomingValue(e - i, false); assert((P.getNumIncomingValues() == 1 && P.getIncomingBlock(PredIndex) == Preheader) && "Should have exactly one value and that's from the preheader!"); } if (DT) { DTU.applyUpdates({{DominatorTree::Insert, Preheader, ExitBlock}}); if (MSSA) { MSSAU->applyUpdates({{DominatorTree::Insert, Preheader, ExitBlock}}, *DT); if (VerifyMemorySSA) MSSA->verifyMemorySSA(); } } // Disconnect the loop body by branching directly to its exit. Builder.SetInsertPoint(Preheader->getTerminator()); Builder.CreateBr(ExitBlock); // Remove the old branch. Preheader->getTerminator()->eraseFromParent(); } else { assert(L->hasNoExitBlocks() && "Loop should have either zero or one exit blocks."); Builder.SetInsertPoint(OldBr); Builder.CreateUnreachable(); Preheader->getTerminator()->eraseFromParent(); } if (DT) { DTU.applyUpdates({{DominatorTree::Delete, Preheader, L->getHeader()}}); if (MSSA) { MSSAU->applyUpdates({{DominatorTree::Delete, Preheader, L->getHeader()}}, *DT); SmallSetVector DeadBlockSet(L->block_begin(), L->block_end()); MSSAU->removeBlocks(DeadBlockSet); if (VerifyMemorySSA) MSSA->verifyMemorySSA(); } } // Use a map to unique and a vector to guarantee deterministic ordering. llvm::SmallDenseSet, 4> DeadDebugSet; llvm::SmallVector DeadDebugInst; if (ExitBlock) { // Given LCSSA form is satisfied, we should not have users of instructions // within the dead loop outside of the loop. However, LCSSA doesn't take // unreachable uses into account. We handle them here. // We could do it after drop all references (in this case all users in the // loop will be already eliminated and we have less work to do but according // to API doc of User::dropAllReferences only valid operation after dropping // references, is deletion. So let's substitute all usages of // instruction from the loop with undef value of corresponding type first. for (auto *Block : L->blocks()) for (Instruction &I : *Block) { auto *Undef = UndefValue::get(I.getType()); for (Use &U : llvm::make_early_inc_range(I.uses())) { if (auto *Usr = dyn_cast(U.getUser())) if (L->contains(Usr->getParent())) continue; // If we have a DT then we can check that uses outside a loop only in // unreachable block. if (DT) assert(!DT->isReachableFromEntry(U) && "Unexpected user in reachable block"); U.set(Undef); } auto *DVI = dyn_cast(&I); if (!DVI) continue; auto Key = DeadDebugSet.find({DVI->getVariable(), DVI->getExpression()}); if (Key != DeadDebugSet.end()) continue; DeadDebugSet.insert({DVI->getVariable(), DVI->getExpression()}); DeadDebugInst.push_back(DVI); } // After the loop has been deleted all the values defined and modified // inside the loop are going to be unavailable. // Since debug values in the loop have been deleted, inserting an undef // dbg.value truncates the range of any dbg.value before the loop where the // loop used to be. This is particularly important for constant values. DIBuilder DIB(*ExitBlock->getModule()); Instruction *InsertDbgValueBefore = ExitBlock->getFirstNonPHI(); assert(InsertDbgValueBefore && "There should be a non-PHI instruction in exit block, else these " "instructions will have no parent."); for (auto *DVI : DeadDebugInst) DIB.insertDbgValueIntrinsic(UndefValue::get(Builder.getInt32Ty()), DVI->getVariable(), DVI->getExpression(), DVI->getDebugLoc(), InsertDbgValueBefore); } // Remove the block from the reference counting scheme, so that we can // delete it freely later. for (auto *Block : L->blocks()) Block->dropAllReferences(); if (MSSA && VerifyMemorySSA) MSSA->verifyMemorySSA(); if (LI) { // Erase the instructions and the blocks without having to worry // about ordering because we already dropped the references. // NOTE: This iteration is safe because erasing the block does not remove // its entry from the loop's block list. We do that in the next section. for (BasicBlock *BB : L->blocks()) BB->eraseFromParent(); // Finally, the blocks from loopinfo. This has to happen late because // otherwise our loop iterators won't work. SmallPtrSet blocks; blocks.insert(L->block_begin(), L->block_end()); for (BasicBlock *BB : blocks) LI->removeBlock(BB); // The last step is to update LoopInfo now that we've eliminated this loop. // Note: LoopInfo::erase remove the given loop and relink its subloops with // its parent. While removeLoop/removeChildLoop remove the given loop but // not relink its subloops, which is what we want. if (Loop *ParentLoop = L->getParentLoop()) { Loop::iterator I = find(*ParentLoop, L); assert(I != ParentLoop->end() && "Couldn't find loop"); ParentLoop->removeChildLoop(I); } else { Loop::iterator I = find(*LI, L); assert(I != LI->end() && "Couldn't find loop"); LI->removeLoop(I); } LI->destroy(L); } } static Loop *getOutermostLoop(Loop *L) { while (Loop *Parent = L->getParentLoop()) L = Parent; return L; } void llvm::breakLoopBackedge(Loop *L, DominatorTree &DT, ScalarEvolution &SE, LoopInfo &LI, MemorySSA *MSSA) { auto *Latch = L->getLoopLatch(); assert(Latch && "multiple latches not yet supported"); auto *Header = L->getHeader(); Loop *OutermostLoop = getOutermostLoop(L); SE.forgetLoop(L); std::unique_ptr MSSAU; if (MSSA) MSSAU = std::make_unique(MSSA); // Update the CFG and domtree. We chose to special case a couple of // of common cases for code quality and test readability reasons. [&]() -> void { if (auto *BI = dyn_cast(Latch->getTerminator())) { if (!BI->isConditional()) { DomTreeUpdater DTU(&DT, DomTreeUpdater::UpdateStrategy::Eager); (void)changeToUnreachable(BI, /*PreserveLCSSA*/ true, &DTU, MSSAU.get()); return; } // Conditional latch/exit - note that latch can be shared by inner // and outer loop so the other target doesn't need to an exit if (L->isLoopExiting(Latch)) { // TODO: Generalize ConstantFoldTerminator so that it can be used // here without invalidating LCSSA or MemorySSA. (Tricky case for // LCSSA: header is an exit block of a preceeding sibling loop w/o // dedicated exits.) const unsigned ExitIdx = L->contains(BI->getSuccessor(0)) ? 1 : 0; BasicBlock *ExitBB = BI->getSuccessor(ExitIdx); DomTreeUpdater DTU(&DT, DomTreeUpdater::UpdateStrategy::Eager); Header->removePredecessor(Latch, true); IRBuilder<> Builder(BI); auto *NewBI = Builder.CreateBr(ExitBB); // Transfer the metadata to the new branch instruction (minus the // loop info since this is no longer a loop) NewBI->copyMetadata(*BI, {LLVMContext::MD_dbg, LLVMContext::MD_annotation}); BI->eraseFromParent(); DTU.applyUpdates({{DominatorTree::Delete, Latch, Header}}); if (MSSA) MSSAU->applyUpdates({{DominatorTree::Delete, Latch, Header}}, DT); return; } } // General case. By splitting the backedge, and then explicitly making it // unreachable we gracefully handle corner cases such as switch and invoke // termiantors. auto *BackedgeBB = SplitEdge(Latch, Header, &DT, &LI, MSSAU.get()); DomTreeUpdater DTU(&DT, DomTreeUpdater::UpdateStrategy::Eager); (void)changeToUnreachable(BackedgeBB->getTerminator(), /*PreserveLCSSA*/ true, &DTU, MSSAU.get()); }(); // Erase (and destroy) this loop instance. Handles relinking sub-loops // and blocks within the loop as needed. LI.erase(L); // If the loop we broke had a parent, then changeToUnreachable might have // caused a block to be removed from the parent loop (see loop_nest_lcssa // test case in zero-btc.ll for an example), thus changing the parent's // exit blocks. If that happened, we need to rebuild LCSSA on the outermost // loop which might have a had a block removed. if (OutermostLoop != L) formLCSSARecursively(*OutermostLoop, DT, &LI, &SE); } /// Checks if \p L has an exiting latch branch. There may also be other /// exiting blocks. Returns branch instruction terminating the loop /// latch if above check is successful, nullptr otherwise. static BranchInst *getExpectedExitLoopLatchBranch(Loop *L) { BasicBlock *Latch = L->getLoopLatch(); if (!Latch) return nullptr; BranchInst *LatchBR = dyn_cast(Latch->getTerminator()); if (!LatchBR || LatchBR->getNumSuccessors() != 2 || !L->isLoopExiting(Latch)) return nullptr; assert((LatchBR->getSuccessor(0) == L->getHeader() || LatchBR->getSuccessor(1) == L->getHeader()) && "At least one edge out of the latch must go to the header"); return LatchBR; } /// Return the estimated trip count for any exiting branch which dominates /// the loop latch. static Optional getEstimatedTripCount(BranchInst *ExitingBranch, Loop *L, uint64_t &OrigExitWeight) { // To estimate the number of times the loop body was executed, we want to // know the number of times the backedge was taken, vs. the number of times // we exited the loop. uint64_t LoopWeight, ExitWeight; if (!ExitingBranch->extractProfMetadata(LoopWeight, ExitWeight)) return None; if (L->contains(ExitingBranch->getSuccessor(1))) std::swap(LoopWeight, ExitWeight); if (!ExitWeight) // Don't have a way to return predicated infinite return None; OrigExitWeight = ExitWeight; // Estimated exit count is a ratio of the loop weight by the weight of the // edge exiting the loop, rounded to nearest. uint64_t ExitCount = llvm::divideNearest(LoopWeight, ExitWeight); // Estimated trip count is one plus estimated exit count. return ExitCount + 1; } Optional llvm::getLoopEstimatedTripCount(Loop *L, unsigned *EstimatedLoopInvocationWeight) { // Currently we take the estimate exit count only from the loop latch, // ignoring other exiting blocks. This can overestimate the trip count // if we exit through another exit, but can never underestimate it. // TODO: incorporate information from other exits if (BranchInst *LatchBranch = getExpectedExitLoopLatchBranch(L)) { uint64_t ExitWeight; if (Optional EstTripCount = getEstimatedTripCount(LatchBranch, L, ExitWeight)) { if (EstimatedLoopInvocationWeight) *EstimatedLoopInvocationWeight = ExitWeight; return *EstTripCount; } } return None; } bool llvm::setLoopEstimatedTripCount(Loop *L, unsigned EstimatedTripCount, unsigned EstimatedloopInvocationWeight) { // At the moment, we currently support changing the estimate trip count of // the latch branch only. We could extend this API to manipulate estimated // trip counts for any exit. BranchInst *LatchBranch = getExpectedExitLoopLatchBranch(L); if (!LatchBranch) return false; // Calculate taken and exit weights. unsigned LatchExitWeight = 0; unsigned BackedgeTakenWeight = 0; if (EstimatedTripCount > 0) { LatchExitWeight = EstimatedloopInvocationWeight; BackedgeTakenWeight = (EstimatedTripCount - 1) * LatchExitWeight; } // Make a swap if back edge is taken when condition is "false". if (LatchBranch->getSuccessor(0) != L->getHeader()) std::swap(BackedgeTakenWeight, LatchExitWeight); MDBuilder MDB(LatchBranch->getContext()); // Set/Update profile metadata. LatchBranch->setMetadata( LLVMContext::MD_prof, MDB.createBranchWeights(BackedgeTakenWeight, LatchExitWeight)); return true; } bool llvm::hasIterationCountInvariantInParent(Loop *InnerLoop, ScalarEvolution &SE) { Loop *OuterL = InnerLoop->getParentLoop(); if (!OuterL) return true; // Get the backedge taken count for the inner loop BasicBlock *InnerLoopLatch = InnerLoop->getLoopLatch(); const SCEV *InnerLoopBECountSC = SE.getExitCount(InnerLoop, InnerLoopLatch); if (isa(InnerLoopBECountSC) || !InnerLoopBECountSC->getType()->isIntegerTy()) return false; // Get whether count is invariant to the outer loop ScalarEvolution::LoopDisposition LD = SE.getLoopDisposition(InnerLoopBECountSC, OuterL); if (LD != ScalarEvolution::LoopInvariant) return false; return true; } CmpInst::Predicate llvm::getMinMaxReductionPredicate(RecurKind RK) { switch (RK) { default: llvm_unreachable("Unknown min/max recurrence kind"); case RecurKind::UMin: return CmpInst::ICMP_ULT; case RecurKind::UMax: return CmpInst::ICMP_UGT; case RecurKind::SMin: return CmpInst::ICMP_SLT; case RecurKind::SMax: return CmpInst::ICMP_SGT; case RecurKind::FMin: return CmpInst::FCMP_OLT; case RecurKind::FMax: return CmpInst::FCMP_OGT; } } Value *llvm::createSelectCmpOp(IRBuilderBase &Builder, Value *StartVal, RecurKind RK, Value *Left, Value *Right) { if (auto VTy = dyn_cast(Left->getType())) StartVal = Builder.CreateVectorSplat(VTy->getElementCount(), StartVal); Value *Cmp = Builder.CreateCmp(CmpInst::ICMP_NE, Left, StartVal, "rdx.select.cmp"); return Builder.CreateSelect(Cmp, Left, Right, "rdx.select"); } Value *llvm::createMinMaxOp(IRBuilderBase &Builder, RecurKind RK, Value *Left, Value *Right) { CmpInst::Predicate Pred = getMinMaxReductionPredicate(RK); Value *Cmp = Builder.CreateCmp(Pred, Left, Right, "rdx.minmax.cmp"); Value *Select = Builder.CreateSelect(Cmp, Left, Right, "rdx.minmax.select"); return Select; } // Helper to generate an ordered reduction. Value *llvm::getOrderedReduction(IRBuilderBase &Builder, Value *Acc, Value *Src, unsigned Op, RecurKind RdxKind) { unsigned VF = cast(Src->getType())->getNumElements(); // Extract and apply reduction ops in ascending order: // e.g. ((((Acc + Scl[0]) + Scl[1]) + Scl[2]) + ) ... + Scl[VF-1] Value *Result = Acc; for (unsigned ExtractIdx = 0; ExtractIdx != VF; ++ExtractIdx) { Value *Ext = Builder.CreateExtractElement(Src, Builder.getInt32(ExtractIdx)); if (Op != Instruction::ICmp && Op != Instruction::FCmp) { Result = Builder.CreateBinOp((Instruction::BinaryOps)Op, Result, Ext, "bin.rdx"); } else { assert(RecurrenceDescriptor::isMinMaxRecurrenceKind(RdxKind) && "Invalid min/max"); Result = createMinMaxOp(Builder, RdxKind, Result, Ext); } } return Result; } // Helper to generate a log2 shuffle reduction. Value *llvm::getShuffleReduction(IRBuilderBase &Builder, Value *Src, unsigned Op, RecurKind RdxKind) { unsigned VF = cast(Src->getType())->getNumElements(); // VF is a power of 2 so we can emit the reduction using log2(VF) shuffles // and vector ops, reducing the set of values being computed by half each // round. assert(isPowerOf2_32(VF) && "Reduction emission only supported for pow2 vectors!"); // Note: fast-math-flags flags are controlled by the builder configuration // and are assumed to apply to all generated arithmetic instructions. Other // poison generating flags (nsw/nuw/inbounds/inrange/exact) are not part // of the builder configuration, and since they're not passed explicitly, // will never be relevant here. Note that it would be generally unsound to // propagate these from an intrinsic call to the expansion anyways as we/ // change the order of operations. Value *TmpVec = Src; SmallVector ShuffleMask(VF); for (unsigned i = VF; i != 1; i >>= 1) { // Move the upper half of the vector to the lower half. for (unsigned j = 0; j != i / 2; ++j) ShuffleMask[j] = i / 2 + j; // Fill the rest of the mask with undef. std::fill(&ShuffleMask[i / 2], ShuffleMask.end(), -1); Value *Shuf = Builder.CreateShuffleVector(TmpVec, ShuffleMask, "rdx.shuf"); if (Op != Instruction::ICmp && Op != Instruction::FCmp) { TmpVec = Builder.CreateBinOp((Instruction::BinaryOps)Op, TmpVec, Shuf, "bin.rdx"); } else { assert(RecurrenceDescriptor::isMinMaxRecurrenceKind(RdxKind) && "Invalid min/max"); TmpVec = createMinMaxOp(Builder, RdxKind, TmpVec, Shuf); } } // The result is in the first element of the vector. return Builder.CreateExtractElement(TmpVec, Builder.getInt32(0)); } Value *llvm::createSelectCmpTargetReduction(IRBuilderBase &Builder, const TargetTransformInfo *TTI, Value *Src, const RecurrenceDescriptor &Desc, PHINode *OrigPhi) { assert(RecurrenceDescriptor::isSelectCmpRecurrenceKind( Desc.getRecurrenceKind()) && "Unexpected reduction kind"); Value *InitVal = Desc.getRecurrenceStartValue(); Value *NewVal = nullptr; // First use the original phi to determine the new value we're trying to // select from in the loop. SelectInst *SI = nullptr; for (auto *U : OrigPhi->users()) { if ((SI = dyn_cast(U))) break; } assert(SI && "One user of the original phi should be a select"); if (SI->getTrueValue() == OrigPhi) NewVal = SI->getFalseValue(); else { assert(SI->getFalseValue() == OrigPhi && "At least one input to the select should be the original Phi"); NewVal = SI->getTrueValue(); } // Create a splat vector with the new value and compare this to the vector // we want to reduce. ElementCount EC = cast(Src->getType())->getElementCount(); Value *Right = Builder.CreateVectorSplat(EC, InitVal); Value *Cmp = Builder.CreateCmp(CmpInst::ICMP_NE, Src, Right, "rdx.select.cmp"); // If any predicate is true it means that we want to select the new value. Cmp = Builder.CreateOrReduce(Cmp); return Builder.CreateSelect(Cmp, NewVal, InitVal, "rdx.select"); } Value *llvm::createSimpleTargetReduction(IRBuilderBase &Builder, const TargetTransformInfo *TTI, Value *Src, RecurKind RdxKind) { auto *SrcVecEltTy = cast(Src->getType())->getElementType(); switch (RdxKind) { case RecurKind::Add: return Builder.CreateAddReduce(Src); case RecurKind::Mul: return Builder.CreateMulReduce(Src); case RecurKind::And: return Builder.CreateAndReduce(Src); case RecurKind::Or: return Builder.CreateOrReduce(Src); case RecurKind::Xor: return Builder.CreateXorReduce(Src); case RecurKind::FMulAdd: case RecurKind::FAdd: return Builder.CreateFAddReduce(ConstantFP::getNegativeZero(SrcVecEltTy), Src); case RecurKind::FMul: return Builder.CreateFMulReduce(ConstantFP::get(SrcVecEltTy, 1.0), Src); case RecurKind::SMax: return Builder.CreateIntMaxReduce(Src, true); case RecurKind::SMin: return Builder.CreateIntMinReduce(Src, true); case RecurKind::UMax: return Builder.CreateIntMaxReduce(Src, false); case RecurKind::UMin: return Builder.CreateIntMinReduce(Src, false); case RecurKind::FMax: return Builder.CreateFPMaxReduce(Src); case RecurKind::FMin: return Builder.CreateFPMinReduce(Src); default: llvm_unreachable("Unhandled opcode"); } } Value *llvm::createTargetReduction(IRBuilderBase &B, const TargetTransformInfo *TTI, const RecurrenceDescriptor &Desc, Value *Src, PHINode *OrigPhi) { // TODO: Support in-order reductions based on the recurrence descriptor. // All ops in the reduction inherit fast-math-flags from the recurrence // descriptor. IRBuilderBase::FastMathFlagGuard FMFGuard(B); B.setFastMathFlags(Desc.getFastMathFlags()); RecurKind RK = Desc.getRecurrenceKind(); if (RecurrenceDescriptor::isSelectCmpRecurrenceKind(RK)) return createSelectCmpTargetReduction(B, TTI, Src, Desc, OrigPhi); return createSimpleTargetReduction(B, TTI, Src, RK); } Value *llvm::createOrderedReduction(IRBuilderBase &B, const RecurrenceDescriptor &Desc, Value *Src, Value *Start) { assert((Desc.getRecurrenceKind() == RecurKind::FAdd || Desc.getRecurrenceKind() == RecurKind::FMulAdd) && "Unexpected reduction kind"); assert(Src->getType()->isVectorTy() && "Expected a vector type"); assert(!Start->getType()->isVectorTy() && "Expected a scalar type"); return B.CreateFAddReduce(Start, Src); } void llvm::propagateIRFlags(Value *I, ArrayRef VL, Value *OpValue) { auto *VecOp = dyn_cast(I); if (!VecOp) return; auto *Intersection = (OpValue == nullptr) ? dyn_cast(VL[0]) : dyn_cast(OpValue); if (!Intersection) return; const unsigned Opcode = Intersection->getOpcode(); VecOp->copyIRFlags(Intersection); for (auto *V : VL) { auto *Instr = dyn_cast(V); if (!Instr) continue; if (OpValue == nullptr || Opcode == Instr->getOpcode()) VecOp->andIRFlags(V); } } bool llvm::isKnownNegativeInLoop(const SCEV *S, const Loop *L, ScalarEvolution &SE) { const SCEV *Zero = SE.getZero(S->getType()); return SE.isAvailableAtLoopEntry(S, L) && SE.isLoopEntryGuardedByCond(L, ICmpInst::ICMP_SLT, S, Zero); } bool llvm::isKnownNonNegativeInLoop(const SCEV *S, const Loop *L, ScalarEvolution &SE) { const SCEV *Zero = SE.getZero(S->getType()); return SE.isAvailableAtLoopEntry(S, L) && SE.isLoopEntryGuardedByCond(L, ICmpInst::ICMP_SGE, S, Zero); } bool llvm::cannotBeMinInLoop(const SCEV *S, const Loop *L, ScalarEvolution &SE, bool Signed) { unsigned BitWidth = cast(S->getType())->getBitWidth(); APInt Min = Signed ? APInt::getSignedMinValue(BitWidth) : APInt::getMinValue(BitWidth); auto Predicate = Signed ? ICmpInst::ICMP_SGT : ICmpInst::ICMP_UGT; return SE.isAvailableAtLoopEntry(S, L) && SE.isLoopEntryGuardedByCond(L, Predicate, S, SE.getConstant(Min)); } bool llvm::cannotBeMaxInLoop(const SCEV *S, const Loop *L, ScalarEvolution &SE, bool Signed) { unsigned BitWidth = cast(S->getType())->getBitWidth(); APInt Max = Signed ? APInt::getSignedMaxValue(BitWidth) : APInt::getMaxValue(BitWidth); auto Predicate = Signed ? ICmpInst::ICMP_SLT : ICmpInst::ICMP_ULT; return SE.isAvailableAtLoopEntry(S, L) && SE.isLoopEntryGuardedByCond(L, Predicate, S, SE.getConstant(Max)); } //===----------------------------------------------------------------------===// // rewriteLoopExitValues - Optimize IV users outside the loop. // As a side effect, reduces the amount of IV processing within the loop. //===----------------------------------------------------------------------===// static bool hasHardUserWithinLoop(const Loop *L, const Instruction *I) { SmallPtrSet Visited; SmallVector WorkList; Visited.insert(I); WorkList.push_back(I); while (!WorkList.empty()) { const Instruction *Curr = WorkList.pop_back_val(); // This use is outside the loop, nothing to do. if (!L->contains(Curr)) continue; // Do we assume it is a "hard" use which will not be eliminated easily? if (Curr->mayHaveSideEffects()) return true; // Otherwise, add all its users to worklist. for (auto U : Curr->users()) { auto *UI = cast(U); if (Visited.insert(UI).second) WorkList.push_back(UI); } } return false; } // Collect information about PHI nodes which can be transformed in // rewriteLoopExitValues. struct RewritePhi { PHINode *PN; // For which PHI node is this replacement? unsigned Ith; // For which incoming value? const SCEV *ExpansionSCEV; // The SCEV of the incoming value we are rewriting. Instruction *ExpansionPoint; // Where we'd like to expand that SCEV? bool HighCost; // Is this expansion a high-cost? RewritePhi(PHINode *P, unsigned I, const SCEV *Val, Instruction *ExpansionPt, bool H) : PN(P), Ith(I), ExpansionSCEV(Val), ExpansionPoint(ExpansionPt), HighCost(H) {} }; // Check whether it is possible to delete the loop after rewriting exit // value. If it is possible, ignore ReplaceExitValue and do rewriting // aggressively. static bool canLoopBeDeleted(Loop *L, SmallVector &RewritePhiSet) { BasicBlock *Preheader = L->getLoopPreheader(); // If there is no preheader, the loop will not be deleted. if (!Preheader) return false; // In LoopDeletion pass Loop can be deleted when ExitingBlocks.size() > 1. // We obviate multiple ExitingBlocks case for simplicity. // TODO: If we see testcase with multiple ExitingBlocks can be deleted // after exit value rewriting, we can enhance the logic here. SmallVector ExitingBlocks; L->getExitingBlocks(ExitingBlocks); SmallVector ExitBlocks; L->getUniqueExitBlocks(ExitBlocks); if (ExitBlocks.size() != 1 || ExitingBlocks.size() != 1) return false; BasicBlock *ExitBlock = ExitBlocks[0]; BasicBlock::iterator BI = ExitBlock->begin(); while (PHINode *P = dyn_cast(BI)) { Value *Incoming = P->getIncomingValueForBlock(ExitingBlocks[0]); // If the Incoming value of P is found in RewritePhiSet, we know it // could be rewritten to use a loop invariant value in transformation // phase later. Skip it in the loop invariant check below. bool found = false; for (const RewritePhi &Phi : RewritePhiSet) { unsigned i = Phi.Ith; if (Phi.PN == P && (Phi.PN)->getIncomingValue(i) == Incoming) { found = true; break; } } Instruction *I; if (!found && (I = dyn_cast(Incoming))) if (!L->hasLoopInvariantOperands(I)) return false; ++BI; } for (auto *BB : L->blocks()) if (llvm::any_of(*BB, [](Instruction &I) { return I.mayHaveSideEffects(); })) return false; return true; } int llvm::rewriteLoopExitValues(Loop *L, LoopInfo *LI, TargetLibraryInfo *TLI, ScalarEvolution *SE, const TargetTransformInfo *TTI, SCEVExpander &Rewriter, DominatorTree *DT, ReplaceExitVal ReplaceExitValue, SmallVector &DeadInsts) { // Check a pre-condition. assert(L->isRecursivelyLCSSAForm(*DT, *LI) && "Indvars did not preserve LCSSA!"); SmallVector ExitBlocks; L->getUniqueExitBlocks(ExitBlocks); SmallVector RewritePhiSet; // Find all values that are computed inside the loop, but used outside of it. // Because of LCSSA, these values will only occur in LCSSA PHI Nodes. Scan // the exit blocks of the loop to find them. for (BasicBlock *ExitBB : ExitBlocks) { // If there are no PHI nodes in this exit block, then no values defined // inside the loop are used on this path, skip it. PHINode *PN = dyn_cast(ExitBB->begin()); if (!PN) continue; unsigned NumPreds = PN->getNumIncomingValues(); // Iterate over all of the PHI nodes. BasicBlock::iterator BBI = ExitBB->begin(); while ((PN = dyn_cast(BBI++))) { if (PN->use_empty()) continue; // dead use, don't replace it if (!SE->isSCEVable(PN->getType())) continue; // Iterate over all of the values in all the PHI nodes. for (unsigned i = 0; i != NumPreds; ++i) { // If the value being merged in is not integer or is not defined // in the loop, skip it. Value *InVal = PN->getIncomingValue(i); if (!isa(InVal)) continue; // If this pred is for a subloop, not L itself, skip it. if (LI->getLoopFor(PN->getIncomingBlock(i)) != L) continue; // The Block is in a subloop, skip it. // Check that InVal is defined in the loop. Instruction *Inst = cast(InVal); if (!L->contains(Inst)) continue; // Okay, this instruction has a user outside of the current loop // and varies predictably *inside* the loop. Evaluate the value it // contains when the loop exits, if possible. We prefer to start with // expressions which are true for all exits (so as to maximize // expression reuse by the SCEVExpander), but resort to per-exit // evaluation if that fails. const SCEV *ExitValue = SE->getSCEVAtScope(Inst, L->getParentLoop()); if (isa(ExitValue) || !SE->isLoopInvariant(ExitValue, L) || !isSafeToExpand(ExitValue, *SE)) { // TODO: This should probably be sunk into SCEV in some way; maybe a // getSCEVForExit(SCEV*, L, ExitingBB)? It can be generalized for // most SCEV expressions and other recurrence types (e.g. shift // recurrences). Is there existing code we can reuse? const SCEV *ExitCount = SE->getExitCount(L, PN->getIncomingBlock(i)); if (isa(ExitCount)) continue; if (auto *AddRec = dyn_cast(SE->getSCEV(Inst))) if (AddRec->getLoop() == L) ExitValue = AddRec->evaluateAtIteration(ExitCount, *SE); if (isa(ExitValue) || !SE->isLoopInvariant(ExitValue, L) || !isSafeToExpand(ExitValue, *SE)) continue; } // Computing the value outside of the loop brings no benefit if it is // definitely used inside the loop in a way which can not be optimized // away. Avoid doing so unless we know we have a value which computes // the ExitValue already. TODO: This should be merged into SCEV // expander to leverage its knowledge of existing expressions. if (ReplaceExitValue != AlwaysRepl && !isa(ExitValue) && !isa(ExitValue) && hasHardUserWithinLoop(L, Inst)) continue; // Check if expansions of this SCEV would count as being high cost. bool HighCost = Rewriter.isHighCostExpansion( ExitValue, L, SCEVCheapExpansionBudget, TTI, Inst); // Note that we must not perform expansions until after // we query *all* the costs, because if we perform temporary expansion // inbetween, one that we might not intend to keep, said expansion // *may* affect cost calculation of the the next SCEV's we'll query, // and next SCEV may errneously get smaller cost. // Collect all the candidate PHINodes to be rewritten. RewritePhiSet.emplace_back(PN, i, ExitValue, Inst, HighCost); } } } // TODO: evaluate whether it is beneficial to change how we calculate // high-cost: if we have SCEV 'A' which we know we will expand, should we // calculate the cost of other SCEV's after expanding SCEV 'A', thus // potentially giving cost bonus to those other SCEV's? bool LoopCanBeDel = canLoopBeDeleted(L, RewritePhiSet); int NumReplaced = 0; // Transformation. for (const RewritePhi &Phi : RewritePhiSet) { PHINode *PN = Phi.PN; // Only do the rewrite when the ExitValue can be expanded cheaply. // If LoopCanBeDel is true, rewrite exit value aggressively. if (ReplaceExitValue == OnlyCheapRepl && !LoopCanBeDel && Phi.HighCost) continue; Value *ExitVal = Rewriter.expandCodeFor( Phi.ExpansionSCEV, Phi.PN->getType(), Phi.ExpansionPoint); LLVM_DEBUG(dbgs() << "rewriteLoopExitValues: AfterLoopVal = " << *ExitVal << '\n' << " LoopVal = " << *(Phi.ExpansionPoint) << "\n"); #ifndef NDEBUG // If we reuse an instruction from a loop which is neither L nor one of // its containing loops, we end up breaking LCSSA form for this loop by // creating a new use of its instruction. if (auto *ExitInsn = dyn_cast(ExitVal)) if (auto *EVL = LI->getLoopFor(ExitInsn->getParent())) if (EVL != L) assert(EVL->contains(L) && "LCSSA breach detected!"); #endif NumReplaced++; Instruction *Inst = cast(PN->getIncomingValue(Phi.Ith)); PN->setIncomingValue(Phi.Ith, ExitVal); // It's necessary to tell ScalarEvolution about this explicitly so that // it can walk the def-use list and forget all SCEVs, as it may not be // watching the PHI itself. Once the new exit value is in place, there // may not be a def-use connection between the loop and every instruction // which got a SCEVAddRecExpr for that loop. SE->forgetValue(PN); // If this instruction is dead now, delete it. Don't do it now to avoid // invalidating iterators. if (isInstructionTriviallyDead(Inst, TLI)) DeadInsts.push_back(Inst); // Replace PN with ExitVal if that is legal and does not break LCSSA. if (PN->getNumIncomingValues() == 1 && LI->replacementPreservesLCSSAForm(PN, ExitVal)) { PN->replaceAllUsesWith(ExitVal); PN->eraseFromParent(); } } // The insertion point instruction may have been deleted; clear it out // so that the rewriter doesn't trip over it later. Rewriter.clearInsertPoint(); return NumReplaced; } /// Set weights for \p UnrolledLoop and \p RemainderLoop based on weights for /// \p OrigLoop. void llvm::setProfileInfoAfterUnrolling(Loop *OrigLoop, Loop *UnrolledLoop, Loop *RemainderLoop, uint64_t UF) { assert(UF > 0 && "Zero unrolled factor is not supported"); assert(UnrolledLoop != RemainderLoop && "Unrolled and Remainder loops are expected to distinct"); // Get number of iterations in the original scalar loop. unsigned OrigLoopInvocationWeight = 0; Optional OrigAverageTripCount = getLoopEstimatedTripCount(OrigLoop, &OrigLoopInvocationWeight); if (!OrigAverageTripCount) return; // Calculate number of iterations in unrolled loop. unsigned UnrolledAverageTripCount = *OrigAverageTripCount / UF; // Calculate number of iterations for remainder loop. unsigned RemainderAverageTripCount = *OrigAverageTripCount % UF; setLoopEstimatedTripCount(UnrolledLoop, UnrolledAverageTripCount, OrigLoopInvocationWeight); setLoopEstimatedTripCount(RemainderLoop, RemainderAverageTripCount, OrigLoopInvocationWeight); } /// Utility that implements appending of loops onto a worklist. /// Loops are added in preorder (analogous for reverse postorder for trees), /// and the worklist is processed LIFO. template void llvm::appendReversedLoopsToWorklist( RangeT &&Loops, SmallPriorityWorklist &Worklist) { // We use an internal worklist to build up the preorder traversal without // recursion. SmallVector PreOrderLoops, PreOrderWorklist; // We walk the initial sequence of loops in reverse because we generally want // to visit defs before uses and the worklist is LIFO. for (Loop *RootL : Loops) { assert(PreOrderLoops.empty() && "Must start with an empty preorder walk."); assert(PreOrderWorklist.empty() && "Must start with an empty preorder walk worklist."); PreOrderWorklist.push_back(RootL); do { Loop *L = PreOrderWorklist.pop_back_val(); PreOrderWorklist.append(L->begin(), L->end()); PreOrderLoops.push_back(L); } while (!PreOrderWorklist.empty()); Worklist.insert(std::move(PreOrderLoops)); PreOrderLoops.clear(); } } template void llvm::appendLoopsToWorklist(RangeT &&Loops, SmallPriorityWorklist &Worklist) { appendReversedLoopsToWorklist(reverse(Loops), Worklist); } template void llvm::appendLoopsToWorklist &>( ArrayRef &Loops, SmallPriorityWorklist &Worklist); template void llvm::appendLoopsToWorklist(Loop &L, SmallPriorityWorklist &Worklist); void llvm::appendLoopsToWorklist(LoopInfo &LI, SmallPriorityWorklist &Worklist) { appendReversedLoopsToWorklist(LI, Worklist); } Loop *llvm::cloneLoop(Loop *L, Loop *PL, ValueToValueMapTy &VM, LoopInfo *LI, LPPassManager *LPM) { Loop &New = *LI->AllocateLoop(); if (PL) PL->addChildLoop(&New); else LI->addTopLevelLoop(&New); if (LPM) LPM->addLoop(New); // Add all of the blocks in L to the new loop. for (BasicBlock *BB : L->blocks()) if (LI->getLoopFor(BB) == L) New.addBasicBlockToLoop(cast(VM[BB]), *LI); // Add all of the subloops to the new loop. for (Loop *I : *L) cloneLoop(I, &New, VM, LI, LPM); return &New; } /// IR Values for the lower and upper bounds of a pointer evolution. We /// need to use value-handles because SCEV expansion can invalidate previously /// expanded values. Thus expansion of a pointer can invalidate the bounds for /// a previous one. struct PointerBounds { TrackingVH Start; TrackingVH End; }; /// Expand code for the lower and upper bound of the pointer group \p CG /// in \p TheLoop. \return the values for the bounds. static PointerBounds expandBounds(const RuntimeCheckingPtrGroup *CG, Loop *TheLoop, Instruction *Loc, SCEVExpander &Exp) { LLVMContext &Ctx = Loc->getContext(); Type *PtrArithTy = Type::getInt8PtrTy(Ctx, CG->AddressSpace); Value *Start = nullptr, *End = nullptr; LLVM_DEBUG(dbgs() << "LAA: Adding RT check for range:\n"); Start = Exp.expandCodeFor(CG->Low, PtrArithTy, Loc); End = Exp.expandCodeFor(CG->High, PtrArithTy, Loc); LLVM_DEBUG(dbgs() << "Start: " << *CG->Low << " End: " << *CG->High << "\n"); return {Start, End}; } /// Turns a collection of checks into a collection of expanded upper and /// lower bounds for both pointers in the check. static SmallVector, 4> expandBounds(const SmallVectorImpl &PointerChecks, Loop *L, Instruction *Loc, SCEVExpander &Exp) { SmallVector, 4> ChecksWithBounds; // Here we're relying on the SCEV Expander's cache to only emit code for the // same bounds once. transform(PointerChecks, std::back_inserter(ChecksWithBounds), [&](const RuntimePointerCheck &Check) { PointerBounds First = expandBounds(Check.first, L, Loc, Exp), Second = expandBounds(Check.second, L, Loc, Exp); return std::make_pair(First, Second); }); return ChecksWithBounds; } Value *llvm::addRuntimeChecks( Instruction *Loc, Loop *TheLoop, const SmallVectorImpl &PointerChecks, SCEVExpander &Exp) { // TODO: Move noalias annotation code from LoopVersioning here and share with LV if possible. // TODO: Pass RtPtrChecking instead of PointerChecks and SE separately, if possible auto ExpandedChecks = expandBounds(PointerChecks, TheLoop, Loc, Exp); LLVMContext &Ctx = Loc->getContext(); IRBuilder ChkBuilder(Ctx, Loc->getModule()->getDataLayout()); ChkBuilder.SetInsertPoint(Loc); // Our instructions might fold to a constant. Value *MemoryRuntimeCheck = nullptr; for (const auto &Check : ExpandedChecks) { const PointerBounds &A = Check.first, &B = Check.second; // Check if two pointers (A and B) conflict where conflict is computed as: // start(A) <= end(B) && start(B) <= end(A) unsigned AS0 = A.Start->getType()->getPointerAddressSpace(); unsigned AS1 = B.Start->getType()->getPointerAddressSpace(); assert((AS0 == B.End->getType()->getPointerAddressSpace()) && (AS1 == A.End->getType()->getPointerAddressSpace()) && "Trying to bounds check pointers with different address spaces"); Type *PtrArithTy0 = Type::getInt8PtrTy(Ctx, AS0); Type *PtrArithTy1 = Type::getInt8PtrTy(Ctx, AS1); Value *Start0 = ChkBuilder.CreateBitCast(A.Start, PtrArithTy0, "bc"); Value *Start1 = ChkBuilder.CreateBitCast(B.Start, PtrArithTy1, "bc"); Value *End0 = ChkBuilder.CreateBitCast(A.End, PtrArithTy1, "bc"); Value *End1 = ChkBuilder.CreateBitCast(B.End, PtrArithTy0, "bc"); // [A|B].Start points to the first accessed byte under base [A|B]. // [A|B].End points to the last accessed byte, plus one. // There is no conflict when the intervals are disjoint: // NoConflict = (B.Start >= A.End) || (A.Start >= B.End) // // bound0 = (B.Start < A.End) // bound1 = (A.Start < B.End) // IsConflict = bound0 & bound1 Value *Cmp0 = ChkBuilder.CreateICmpULT(Start0, End1, "bound0"); Value *Cmp1 = ChkBuilder.CreateICmpULT(Start1, End0, "bound1"); Value *IsConflict = ChkBuilder.CreateAnd(Cmp0, Cmp1, "found.conflict"); if (MemoryRuntimeCheck) { IsConflict = ChkBuilder.CreateOr(MemoryRuntimeCheck, IsConflict, "conflict.rdx"); } MemoryRuntimeCheck = IsConflict; } return MemoryRuntimeCheck; } Optional llvm::hasPartialIVCondition(Loop &L, unsigned MSSAThreshold, MemorySSA &MSSA, AAResults &AA) { auto *TI = dyn_cast(L.getHeader()->getTerminator()); if (!TI || !TI->isConditional()) return {}; auto *CondI = dyn_cast(TI->getCondition()); // The case with the condition outside the loop should already be handled // earlier. if (!CondI || !L.contains(CondI)) return {}; SmallVector InstToDuplicate; InstToDuplicate.push_back(CondI); SmallVector WorkList; WorkList.append(CondI->op_begin(), CondI->op_end()); SmallVector AccessesToCheck; SmallVector AccessedLocs; while (!WorkList.empty()) { Instruction *I = dyn_cast(WorkList.pop_back_val()); if (!I || !L.contains(I)) continue; // TODO: support additional instructions. if (!isa(I) && !isa(I)) return {}; // Do not duplicate volatile and atomic loads. if (auto *LI = dyn_cast(I)) if (LI->isVolatile() || LI->isAtomic()) return {}; InstToDuplicate.push_back(I); if (MemoryAccess *MA = MSSA.getMemoryAccess(I)) { if (auto *MemUse = dyn_cast_or_null(MA)) { // Queue the defining access to check for alias checks. AccessesToCheck.push_back(MemUse->getDefiningAccess()); AccessedLocs.push_back(MemoryLocation::get(I)); } else { // MemoryDefs may clobber the location or may be atomic memory // operations. Bail out. return {}; } } WorkList.append(I->op_begin(), I->op_end()); } if (InstToDuplicate.empty()) return {}; SmallVector ExitingBlocks; L.getExitingBlocks(ExitingBlocks); auto HasNoClobbersOnPath = [&L, &AA, &AccessedLocs, &ExitingBlocks, &InstToDuplicate, MSSAThreshold](BasicBlock *Succ, BasicBlock *Header, SmallVector AccessesToCheck) -> Optional { IVConditionInfo Info; // First, collect all blocks in the loop that are on a patch from Succ // to the header. SmallVector WorkList; WorkList.push_back(Succ); WorkList.push_back(Header); SmallPtrSet Seen; Seen.insert(Header); Info.PathIsNoop &= all_of(*Header, [](Instruction &I) { return !I.mayHaveSideEffects(); }); while (!WorkList.empty()) { BasicBlock *Current = WorkList.pop_back_val(); if (!L.contains(Current)) continue; const auto &SeenIns = Seen.insert(Current); if (!SeenIns.second) continue; Info.PathIsNoop &= all_of( *Current, [](Instruction &I) { return !I.mayHaveSideEffects(); }); WorkList.append(succ_begin(Current), succ_end(Current)); } // Require at least 2 blocks on a path through the loop. This skips // paths that directly exit the loop. if (Seen.size() < 2) return {}; // Next, check if there are any MemoryDefs that are on the path through // the loop (in the Seen set) and they may-alias any of the locations in // AccessedLocs. If that is the case, they may modify the condition and // partial unswitching is not possible. SmallPtrSet SeenAccesses; while (!AccessesToCheck.empty()) { MemoryAccess *Current = AccessesToCheck.pop_back_val(); auto SeenI = SeenAccesses.insert(Current); if (!SeenI.second || !Seen.contains(Current->getBlock())) continue; // Bail out if exceeded the threshold. if (SeenAccesses.size() >= MSSAThreshold) return {}; // MemoryUse are read-only accesses. if (isa(Current)) continue; // For a MemoryDef, check if is aliases any of the location feeding // the original condition. if (auto *CurrentDef = dyn_cast(Current)) { if (any_of(AccessedLocs, [&AA, CurrentDef](MemoryLocation &Loc) { return isModSet( AA.getModRefInfo(CurrentDef->getMemoryInst(), Loc)); })) return {}; } for (Use &U : Current->uses()) AccessesToCheck.push_back(cast(U.getUser())); } // We could also allow loops with known trip counts without mustprogress, // but ScalarEvolution may not be available. Info.PathIsNoop &= isMustProgress(&L); // If the path is considered a no-op so far, check if it reaches a // single exit block without any phis. This ensures no values from the // loop are used outside of the loop. if (Info.PathIsNoop) { for (auto *Exiting : ExitingBlocks) { if (!Seen.contains(Exiting)) continue; for (auto *Succ : successors(Exiting)) { if (L.contains(Succ)) continue; Info.PathIsNoop &= llvm::empty(Succ->phis()) && (!Info.ExitForPath || Info.ExitForPath == Succ); if (!Info.PathIsNoop) break; assert((!Info.ExitForPath || Info.ExitForPath == Succ) && "cannot have multiple exit blocks"); Info.ExitForPath = Succ; } } } if (!Info.ExitForPath) Info.PathIsNoop = false; Info.InstToDuplicate = InstToDuplicate; return Info; }; // If we branch to the same successor, partial unswitching will not be // beneficial. if (TI->getSuccessor(0) == TI->getSuccessor(1)) return {}; if (auto Info = HasNoClobbersOnPath(TI->getSuccessor(0), L.getHeader(), AccessesToCheck)) { Info->KnownValue = ConstantInt::getTrue(TI->getContext()); return Info; } if (auto Info = HasNoClobbersOnPath(TI->getSuccessor(1), L.getHeader(), AccessesToCheck)) { Info->KnownValue = ConstantInt::getFalse(TI->getContext()); return Info; } return {}; }