#include "llvm/CodeGen/Passes.h"
#include "llvm/ADT/DenseMap.h"
+#include "llvm/ADT/SetVector.h"
+#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/ADT/SmallSet.h"
+#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/Analysis/InstructionSimplify.h"
+#include "llvm/Analysis/MemoryLocation.h"
#include "llvm/Analysis/TargetLibraryInfo.h"
#include "llvm/Analysis/TargetTransformInfo.h"
+#include "llvm/Analysis/ValueTracking.h"
#include "llvm/IR/CallSite.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/GetElementPtrTypeIterator.h"
#include "llvm/IR/IRBuilder.h"
#include "llvm/IR/InlineAsm.h"
+#include "llvm/IR/InstIterator.h"
+#include "llvm/IR/InstrTypes.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/MDBuilder.h"
+#include "llvm/IR/NoFolder.h"
#include "llvm/IR/PatternMatch.h"
#include "llvm/IR/Statepoint.h"
#include "llvm/IR/ValueHandle.h"
"computations were sunk");
STATISTIC(NumExtsMoved, "Number of [s|z]ext instructions combined with loads");
STATISTIC(NumExtUses, "Number of uses of [s|z]ext instructions optimized");
+STATISTIC(NumAndsAdded,
+ "Number of and mask instructions added to form ext loads");
+STATISTIC(NumAndUses, "Number of uses of and mask instructions optimized");
STATISTIC(NumRetsDup, "Number of return instructions duplicated");
STATISTIC(NumDbgValueMoved, "Number of debug value instructions moved");
STATISTIC(NumSelectsExpanded, "Number of selects turned into branches");
class TypePromotionTransaction;
class CodeGenPrepare : public FunctionPass {
- /// TLI - Keep a pointer of a TargetLowering to consult for determining
- /// transformation profitability.
const TargetMachine *TM;
const TargetLowering *TLI;
const TargetTransformInfo *TTI;
const TargetLibraryInfo *TLInfo;
- /// CurInstIterator - As we scan instructions optimizing them, this is the
- /// next instruction to optimize. Xforms that can invalidate this should
- /// update it.
+ /// As we scan instructions optimizing them, this is the next instruction
+ /// to optimize. Transforms that can invalidate this should update it.
BasicBlock::iterator CurInstIterator;
/// Keeps track of non-local addresses that have been sunk into a block.
/// promotion for the current function.
InstrToOrigTy PromotedInsts;
- /// ModifiedDT - If CFG is modified in anyway.
+ /// True if CFG is modified in any way.
bool ModifiedDT;
- /// OptSize - True if optimizing for size.
+ /// True if optimizing for size.
bool OptSize;
/// DataLayout for the Function being processed.
}
private:
- bool EliminateFallThrough(Function &F);
- bool EliminateMostlyEmptyBlocks(Function &F);
- bool CanMergeBlocks(const BasicBlock *BB, const BasicBlock *DestBB) const;
- void EliminateMostlyEmptyBlock(BasicBlock *BB);
- bool OptimizeBlock(BasicBlock &BB, bool& ModifiedDT);
- bool OptimizeInst(Instruction *I, bool& ModifiedDT);
- bool OptimizeMemoryInst(Instruction *I, Value *Addr,
+ bool eliminateFallThrough(Function &F);
+ bool eliminateMostlyEmptyBlocks(Function &F);
+ bool canMergeBlocks(const BasicBlock *BB, const BasicBlock *DestBB) const;
+ void eliminateMostlyEmptyBlock(BasicBlock *BB);
+ bool optimizeBlock(BasicBlock &BB, bool& ModifiedDT);
+ bool optimizeInst(Instruction *I, bool& ModifiedDT);
+ bool optimizeMemoryInst(Instruction *I, Value *Addr,
Type *AccessTy, unsigned AS);
- bool OptimizeInlineAsmInst(CallInst *CS);
- bool OptimizeCallInst(CallInst *CI, bool& ModifiedDT);
- bool MoveExtToFormExtLoad(Instruction *&I);
- bool OptimizeExtUses(Instruction *I);
- bool OptimizeSelectInst(SelectInst *SI);
- bool OptimizeShuffleVectorInst(ShuffleVectorInst *SI);
- bool OptimizeExtractElementInst(Instruction *Inst);
- bool DupRetToEnableTailCallOpts(BasicBlock *BB);
- bool PlaceDbgValues(Function &F);
+ bool optimizeInlineAsmInst(CallInst *CS);
+ bool optimizeCallInst(CallInst *CI, bool& ModifiedDT);
+ bool moveExtToFormExtLoad(Instruction *&I);
+ bool optimizeExtUses(Instruction *I);
+ bool optimizeLoadExt(LoadInst *I);
+ bool optimizeSelectInst(SelectInst *SI);
+ bool optimizeShuffleVectorInst(ShuffleVectorInst *SI);
+ bool optimizeSwitchInst(SwitchInst *CI);
+ bool optimizeExtractElementInst(Instruction *Inst);
+ bool dupRetToEnableTailCallOpts(BasicBlock *BB);
+ bool placeDbgValues(Function &F);
bool sinkAndCmp(Function &F);
- bool ExtLdPromotion(TypePromotionTransaction &TPT, LoadInst *&LI,
+ bool extLdPromotion(TypePromotionTransaction &TPT, LoadInst *&LI,
Instruction *&Inst,
const SmallVectorImpl<Instruction *> &Exts,
unsigned CreatedInstCost);
return new CodeGenPrepare(TM);
}
+namespace {
+
+bool StoreAddressDependOnValue(StoreInst* SI, Value* DepVal);
+Value* GetUntaintedAddress(Value* CurrentAddress);
+
+// The depth we trace down a variable to look for its dependence set.
+const unsigned kDependenceDepth = 4;
+
+// Recursively looks for variables that 'Val' depends on at the given depth
+// 'Depth', and adds them in 'DepSet'. If 'InsertOnlyLeafNodes' is true, only
+// inserts the leaf node values; otherwise, all visited nodes are included in
+// 'DepSet'. Note that constants will be ignored.
+template <typename SetType>
+void recursivelyFindDependence(SetType* DepSet, Value* Val,
+ bool InsertOnlyLeafNodes = false,
+ unsigned Depth = kDependenceDepth) {
+ if (Val == nullptr) {
+ return;
+ }
+ if (!InsertOnlyLeafNodes && !isa<Constant>(Val)) {
+ DepSet->insert(Val);
+ }
+ if (Depth == 0) {
+ // Cannot go deeper. Insert the leaf nodes.
+ if (InsertOnlyLeafNodes && !isa<Constant>(Val)) {
+ DepSet->insert(Val);
+ }
+ return;
+ }
+
+ // Go one step further to explore the dependence of the operands.
+ Instruction* I = nullptr;
+ if ((I = dyn_cast<Instruction>(Val))) {
+ if (isa<LoadInst>(I)) {
+ // A load is considerd the leaf load of the dependence tree. Done.
+ DepSet->insert(Val);
+ return;
+ } else if (I->isBinaryOp()) {
+ BinaryOperator* I = dyn_cast<BinaryOperator>(Val);
+ Value *Op0 = I->getOperand(0), *Op1 = I->getOperand(1);
+ recursivelyFindDependence(DepSet, Op0, Depth - 1);
+ recursivelyFindDependence(DepSet, Op1, Depth - 1);
+ } else if (I->isCast()) {
+ Value* Op0 = I->getOperand(0);
+ recursivelyFindDependence(DepSet, Op0, Depth - 1);
+ } else if (I->getOpcode() == Instruction::Select) {
+ Value* Op0 = I->getOperand(0);
+ Value* Op1 = I->getOperand(1);
+ Value* Op2 = I->getOperand(2);
+ recursivelyFindDependence(DepSet, Op0, Depth - 1);
+ recursivelyFindDependence(DepSet, Op1, Depth - 1);
+ recursivelyFindDependence(DepSet, Op2, Depth - 1);
+ } else if (I->getOpcode() == Instruction::GetElementPtr) {
+ for (unsigned i = 0; i < I->getNumOperands(); i++) {
+ recursivelyFindDependence(DepSet, I->getOperand(i), Depth - 1);
+ }
+ } else if (I->getOpcode() == Instruction::Store) {
+ auto* SI = dyn_cast<StoreInst>(Val);
+ recursivelyFindDependence(DepSet, SI->getPointerOperand(), Depth - 1);
+ recursivelyFindDependence(DepSet, SI->getValueOperand(), Depth - 1);
+ } else {
+ Value* Op0 = nullptr;
+ Value* Op1 = nullptr;
+ switch (I->getOpcode()) {
+ case Instruction::ICmp:
+ case Instruction::FCmp: {
+ Op0 = I->getOperand(0);
+ Op1 = I->getOperand(1);
+ recursivelyFindDependence(DepSet, Op0, Depth - 1);
+ recursivelyFindDependence(DepSet, Op1, Depth - 1);
+ break;
+ }
+ default: {
+ // Be conservative. Add it and be done with it.
+ DepSet->insert(Val);
+ return;
+ }
+ }
+ }
+ } else if (isa<Constant>(Val)) {
+ // Not interested in constant values. Done.
+ return;
+ } else {
+ // Be conservative. Add it and be done with it.
+ DepSet->insert(Val);
+ return;
+ }
+}
+
+// Helper function to create a Cast instruction.
+Value* createCast(IRBuilder<true, NoFolder>& Builder, Value* DepVal,
+ Type* TargetIntegerType) {
+ Instruction::CastOps CastOp = Instruction::BitCast;
+ switch (DepVal->getType()->getTypeID()) {
+ case Type::IntegerTyID: {
+ CastOp = Instruction::SExt;
+ break;
+ }
+ case Type::FloatTyID:
+ case Type::DoubleTyID: {
+ CastOp = Instruction::FPToSI;
+ break;
+ }
+ case Type::PointerTyID: {
+ CastOp = Instruction::PtrToInt;
+ break;
+ }
+ default: { break; }
+ }
+
+ return Builder.CreateCast(CastOp, DepVal, TargetIntegerType);
+}
+
+// Given a value, if it's a tainted address, this function returns the
+// instruction that ORs the "dependence value" with the "original address".
+// Otherwise, returns nullptr. This instruction is the first OR instruction
+// where one of its operand is an AND instruction with an operand being 0.
+//
+// E.g., it returns '%4 = or i32 %3, %2' given 'CurrentAddress' is '%5'.
+// %0 = load i32, i32* @y, align 4, !tbaa !1
+// %cmp = icmp ne i32 %0, 42 // <== this is like the condition
+// %1 = sext i1 %cmp to i32
+// %2 = ptrtoint i32* @x to i32
+// %3 = and i32 %1, 0
+// %4 = or i32 %3, %2
+// %5 = inttoptr i32 %4 to i32*
+// store i32 1, i32* %5, align 4
+Instruction* getOrAddress(Value* CurrentAddress) {
+ // Is it a cast from integer to pointer type.
+ Instruction* OrAddress = nullptr;
+ Instruction* AndDep = nullptr;
+ Instruction* CastToInt = nullptr;
+ Value* ActualAddress = nullptr;
+ Constant* ZeroConst = nullptr;
+
+ const Instruction* CastToPtr = dyn_cast<Instruction>(CurrentAddress);
+ if (CastToPtr && CastToPtr->getOpcode() == Instruction::IntToPtr) {
+ // Is it an OR instruction: %1 = or %and, %actualAddress.
+ if ((OrAddress = dyn_cast<Instruction>(CastToPtr->getOperand(0))) &&
+ OrAddress->getOpcode() == Instruction::Or) {
+ // The first operand should be and AND instruction.
+ AndDep = dyn_cast<Instruction>(OrAddress->getOperand(0));
+ if (AndDep && AndDep->getOpcode() == Instruction::And) {
+ // Also make sure its first operand of the "AND" is 0, or the "AND" is
+ // marked explicitly by "NoInstCombine".
+ if ((ZeroConst = dyn_cast<Constant>(AndDep->getOperand(1))) &&
+ ZeroConst->isNullValue()) {
+ return OrAddress;
+ }
+ }
+ }
+ }
+ // Looks like it's not been tainted.
+ return nullptr;
+}
+
+// Given a value, if it's a tainted address, this function returns the
+// instruction that taints the "dependence value". Otherwise, returns nullptr.
+// This instruction is the last AND instruction where one of its operand is 0.
+// E.g., it returns '%3' given 'CurrentAddress' is '%5'.
+// %0 = load i32, i32* @y, align 4, !tbaa !1
+// %cmp = icmp ne i32 %0, 42 // <== this is like the condition
+// %1 = sext i1 %cmp to i32
+// %2 = ptrtoint i32* @x to i32
+// %3 = and i32 %1, 0
+// %4 = or i32 %3, %2
+// %5 = inttoptr i32 %4 to i32*
+// store i32 1, i32* %5, align 4
+Instruction* getAndDependence(Value* CurrentAddress) {
+ // If 'CurrentAddress' is tainted, get the OR instruction.
+ auto* OrAddress = getOrAddress(CurrentAddress);
+ if (OrAddress == nullptr) {
+ return nullptr;
+ }
+
+ // No need to check the operands.
+ auto* AndDepInst = dyn_cast<Instruction>(OrAddress->getOperand(0));
+ assert(AndDepInst);
+ return AndDepInst;
+}
+
+// Given a value, if it's a tainted address, this function returns
+// the "dependence value", which is the first operand in the AND instruction.
+// E.g., it returns '%1' given 'CurrentAddress' is '%5'.
+// %0 = load i32, i32* @y, align 4, !tbaa !1
+// %cmp = icmp ne i32 %0, 42 // <== this is like the condition
+// %1 = sext i1 %cmp to i32
+// %2 = ptrtoint i32* @x to i32
+// %3 = and i32 %1, 0
+// %4 = or i32 %3, %2
+// %5 = inttoptr i32 %4 to i32*
+// store i32 1, i32* %5, align 4
+Value* getDependence(Value* CurrentAddress) {
+ auto* AndInst = getAndDependence(CurrentAddress);
+ if (AndInst == nullptr) {
+ return nullptr;
+ }
+ return AndInst->getOperand(0);
+}
+
+// Given an address that has been tainted, returns the only condition it depends
+// on, if any; otherwise, returns nullptr.
+Value* getConditionDependence(Value* Address) {
+ auto* Dep = getDependence(Address);
+ if (Dep == nullptr) {
+ // 'Address' has not been dependence-tainted.
+ return nullptr;
+ }
+
+ Value* Operand = Dep;
+ while (true) {
+ auto* Inst = dyn_cast<Instruction>(Operand);
+ if (Inst == nullptr) {
+ // Non-instruction type does not have condition dependence.
+ return nullptr;
+ }
+ if (Inst->getOpcode() == Instruction::ICmp) {
+ return Inst;
+ } else {
+ if (Inst->getNumOperands() != 1) {
+ return nullptr;
+ } else {
+ Operand = Inst->getOperand(0);
+ }
+ }
+ }
+}
+
+// Conservatively decides whether the dependence set of 'Val1' includes the
+// dependence set of 'Val2'. If 'ExpandSecondValue' is false, we do not expand
+// 'Val2' and use that single value as its dependence set.
+// If it returns true, it means the dependence set of 'Val1' includes that of
+// 'Val2'; otherwise, it only means we cannot conclusively decide it.
+bool dependenceSetInclusion(Value* Val1, Value* Val2,
+ int Val1ExpandLevel = 2 * kDependenceDepth,
+ int Val2ExpandLevel = kDependenceDepth) {
+ typedef SmallSet<Value*, 8> IncludingSet;
+ typedef SmallSet<Value*, 4> IncludedSet;
+
+ IncludingSet DepSet1;
+ IncludedSet DepSet2;
+ // Look for more depths for the including set.
+ recursivelyFindDependence(&DepSet1, Val1, false /*Insert all visited nodes*/,
+ Val1ExpandLevel);
+ recursivelyFindDependence(&DepSet2, Val2, true /*Only insert leaf nodes*/,
+ Val2ExpandLevel);
+
+ auto set_inclusion = [](IncludingSet FullSet, IncludedSet Subset) {
+ for (auto* Dep : Subset) {
+ if (0 == FullSet.count(Dep)) {
+ return false;
+ }
+ }
+ return true;
+ };
+ bool inclusion = set_inclusion(DepSet1, DepSet2);
+ DEBUG(dbgs() << "[dependenceSetInclusion]: " << inclusion << "\n");
+ DEBUG(dbgs() << "Including set for: " << *Val1 << "\n");
+ DEBUG(for (const auto* Dep : DepSet1) { dbgs() << "\t\t" << *Dep << "\n"; });
+ DEBUG(dbgs() << "Included set for: " << *Val2 << "\n");
+ DEBUG(for (const auto* Dep : DepSet2) { dbgs() << "\t\t" << *Dep << "\n"; });
+
+ return inclusion;
+}
+
+// Recursively iterates through the operands spawned from 'DepVal'. If there
+// exists a single value that 'DepVal' only depends on, we call that value the
+// root dependence of 'DepVal' and return it. Otherwise, return 'DepVal'.
+Value* getRootDependence(Value* DepVal) {
+ SmallSet<Value*, 8> DepSet;
+ for (unsigned depth = kDependenceDepth; depth > 0; --depth) {
+ recursivelyFindDependence(&DepSet, DepVal, true /*Only insert leaf nodes*/,
+ depth);
+ if (DepSet.size() == 1) {
+ return *DepSet.begin();
+ }
+ DepSet.clear();
+ }
+ return DepVal;
+}
+
+// This function actually taints 'DepVal' to the address to 'SI'. If the
+// address
+// of 'SI' already depends on whatever 'DepVal' depends on, this function
+// doesn't do anything and returns false. Otherwise, returns true.
+//
+// This effect forces the store and any stores that comes later to depend on
+// 'DepVal'. For example, we have a condition "cond", and a store instruction
+// "s: STORE addr, val". If we want "s" (and any later store) to depend on
+// "cond", we do the following:
+// %conv = sext i1 %cond to i32
+// %addrVal = ptrtoint i32* %addr to i32
+// %andCond = and i32 conv, 0;
+// %orAddr = or i32 %andCond, %addrVal;
+// %NewAddr = inttoptr i32 %orAddr to i32*;
+//
+// This is a more concrete example:
+// ------
+// %0 = load i32, i32* @y, align 4, !tbaa !1
+// %cmp = icmp ne i32 %0, 42 // <== this is like the condition
+// %1 = sext i1 %cmp to i32
+// %2 = ptrtoint i32* @x to i32
+// %3 = and i32 %1, 0
+// %4 = or i32 %3, %2
+// %5 = inttoptr i32 %4 to i32*
+// store i32 1, i32* %5, align 4
+bool taintStoreAddress(StoreInst* SI, Value* DepVal,
+ const char* calling_func = __builtin_FUNCTION()) {
+ DEBUG(dbgs() << "Called from " << calling_func << '\n');
+ // Set the insertion point right after the 'DepVal'.
+ Instruction* Inst = nullptr;
+ IRBuilder<true, NoFolder> Builder(SI);
+ BasicBlock* BB = SI->getParent();
+ Value* Address = SI->getPointerOperand();
+ Type* TargetIntegerType =
+ IntegerType::get(Address->getContext(),
+ BB->getModule()->getDataLayout().getPointerSizeInBits());
+
+ // Does SI's address already depends on whatever 'DepVal' depends on?
+ if (StoreAddressDependOnValue(SI, DepVal)) {
+ return false;
+ }
+
+ // Figure out if there's a root variable 'DepVal' depends on. For example, we
+ // can extract "getelementptr inbounds %struct, %struct* %0, i64 0, i32 123"
+ // to be "%struct* %0" since all other operands are constant.
+ DepVal = getRootDependence(DepVal);
+
+ // Is this already a dependence-tainted store?
+ Value* OldDep = getDependence(Address);
+ if (OldDep) {
+ // The address of 'SI' has already been tainted. Just need to absorb the
+ // DepVal to the existing dependence in the address of SI.
+ Instruction* AndDep = getAndDependence(Address);
+ IRBuilder<true, NoFolder> Builder(AndDep);
+ Value* NewDep = nullptr;
+ if (DepVal->getType() == AndDep->getType()) {
+ NewDep = Builder.CreateAnd(OldDep, DepVal);
+ } else {
+ NewDep = Builder.CreateAnd(
+ OldDep, createCast(Builder, DepVal, TargetIntegerType));
+ }
+
+ auto* NewDepInst = dyn_cast<Instruction>(NewDep);
+
+ // Use the new AND instruction as the dependence
+ AndDep->setOperand(0, NewDep);
+ return true;
+ }
+
+ // SI's address has not been tainted. Now taint it with 'DepVal'.
+ Value* CastDepToInt = createCast(Builder, DepVal, TargetIntegerType);
+ Value* PtrToIntCast = Builder.CreatePtrToInt(Address, TargetIntegerType);
+ Value* AndDepVal =
+ Builder.CreateAnd(CastDepToInt, ConstantInt::get(TargetIntegerType, 0));
+ auto AndInst = dyn_cast<Instruction>(AndDepVal);
+ // XXX-comment: The original IR InstCombiner would change our and instruction
+ // to a select and then the back end optimize the condition out. We attach a
+ // flag to instructions and set it here to inform the InstCombiner to not to
+ // touch this and instruction at all.
+ Value* OrAddr = Builder.CreateOr(AndDepVal, PtrToIntCast);
+ Value* NewAddr = Builder.CreateIntToPtr(OrAddr, Address->getType());
+
+ DEBUG(dbgs() << "[taintStoreAddress]\n"
+ << "Original store: " << *SI << '\n');
+ SI->setOperand(1, NewAddr);
+
+ // Debug output.
+ DEBUG(dbgs() << "\tTargetIntegerType: " << *TargetIntegerType << '\n'
+ << "\tCast dependence value to integer: " << *CastDepToInt
+ << '\n'
+ << "\tCast address to integer: " << *PtrToIntCast << '\n'
+ << "\tAnd dependence value: " << *AndDepVal << '\n'
+ << "\tOr address: " << *OrAddr << '\n'
+ << "\tCast or instruction to address: " << *NewAddr << "\n\n");
+
+ return true;
+}
+
+// Looks for the previous store in the if block --- 'BrBB', which makes the
+// speculative store 'StoreToHoist' safe.
+Value* getSpeculativeStoreInPrevBB(StoreInst* StoreToHoist, BasicBlock* BrBB) {
+ assert(StoreToHoist && "StoreToHoist must be a real store");
+
+ Value* StorePtr = StoreToHoist->getPointerOperand();
+
+ // Look for a store to the same pointer in BrBB.
+ for (BasicBlock::reverse_iterator RI = BrBB->rbegin(), RE = BrBB->rend();
+ RI != RE; ++RI) {
+ Instruction* CurI = &*RI;
+
+ StoreInst* SI = dyn_cast<StoreInst>(CurI);
+ // Found the previous store make sure it stores to the same location.
+ // XXX-update: If the previous store's original untainted address are the
+ // same as 'StorePtr', we are also good to hoist the store.
+ if (SI && (SI->getPointerOperand() == StorePtr ||
+ GetUntaintedAddress(SI->getPointerOperand()) == StorePtr)) {
+ // Found the previous store, return its value operand.
+ return SI;
+ }
+ }
+
+ assert(false &&
+ "We should not reach here since this store is safe to speculate");
+}
+
+// XXX-comment: Returns true if it changes the code, false otherwise (the branch
+// condition already depends on 'DepVal'.
+bool taintConditionalBranch(BranchInst* BI, Value* DepVal) {
+ assert(BI->isConditional());
+ auto* Cond = BI->getOperand(0);
+ if (dependenceSetInclusion(Cond, DepVal)) {
+ // The dependence/ordering is self-evident.
+ return false;
+ }
+
+ IRBuilder<true, NoFolder> Builder(BI);
+ auto* AndDep =
+ Builder.CreateAnd(DepVal, ConstantInt::get(DepVal->getType(), 0));
+ auto* TruncAndDep =
+ Builder.CreateTrunc(AndDep, IntegerType::get(DepVal->getContext(), 1));
+ auto* OrCond = Builder.CreateOr(TruncAndDep, Cond);
+ BI->setOperand(0, OrCond);
+
+ // Debug output.
+ DEBUG(dbgs() << "\tTainted branch condition:\n" << *BI->getParent());
+
+ return true;
+}
+
+bool ConditionalBranchDependsOnValue(BranchInst* BI, Value* DepVal) {
+ assert(BI->isConditional());
+ auto* Cond = BI->getOperand(0);
+ return dependenceSetInclusion(Cond, DepVal);
+}
+
+// XXX-update: For a relaxed load 'LI', find the first immediate atomic store or
+// the first conditional branch. Returns nullptr if there's no such immediately
+// following store/branch instructions, which we can only enforce the load with
+// 'acquire'.
+Instruction* findFirstStoreCondBranchInst(LoadInst* LI) {
+ // In some situations, relaxed loads can be left as is:
+ // 1. The relaxed load is used to calculate the address of the immediate
+ // following store;
+ // 2. The relaxed load is used as a condition in the immediate following
+ // condition, and there are no stores in between. This is actually quite
+ // common. E.g.,
+ // int r1 = x.load(relaxed);
+ // if (r1 != 0) {
+ // y.store(1, relaxed);
+ // }
+ // However, in this function, we don't deal with them directly. Instead, we
+ // just find the immediate following store/condition branch and return it.
+
+ auto* BB = LI->getParent();
+ auto BE = BB->end();
+ auto BBI = BasicBlock::iterator(LI);
+ BBI++;
+ while (true) {
+ for (; BBI != BE; BBI++) {
+ auto* Inst = dyn_cast<Instruction>(&*BBI);
+ if (Inst == nullptr) {
+ continue;
+ }
+ if (Inst->getOpcode() == Instruction::Store) {
+ return Inst;
+ } else if (Inst->getOpcode() == Instruction::Br) {
+ auto* BrInst = dyn_cast<BranchInst>(Inst);
+ if (BrInst->isConditional()) {
+ return Inst;
+ } else {
+ // Reinitialize iterators with the destination of the unconditional
+ // branch.
+ BB = BrInst->getSuccessor(0);
+ BBI = BB->begin();
+ BE = BB->end();
+ break;
+ }
+ }
+ }
+ if (BBI == BE) {
+ return nullptr;
+ }
+ }
+}
+
+// XXX-comment: Returns whether the code has been changed.
+bool taintMonotonicLoads(const SmallVector<LoadInst*, 1>& MonotonicLoadInsts) {
+ bool Changed = false;
+ for (auto* LI : MonotonicLoadInsts) {
+ auto* FirstInst = findFirstStoreCondBranchInst(LI);
+ if (FirstInst == nullptr) {
+ // We don't seem to be able to taint a following store/conditional branch
+ // instruction. Simply make it acquire.
+ DEBUG(dbgs() << "[RelaxedLoad]: Transformed to acquire load\n"
+ << *LI << "\n");
+ LI->setOrdering(Acquire);
+ Changed = true;
+ continue;
+ }
+ // Taint 'FirstInst', which could be a store or a condition branch
+ // instruction.
+ if (FirstInst->getOpcode() == Instruction::Store) {
+ Changed |= taintStoreAddress(dyn_cast<StoreInst>(FirstInst), LI);
+ } else if (FirstInst->getOpcode() == Instruction::Br) {
+ Changed |= taintConditionalBranch(dyn_cast<BranchInst>(FirstInst), LI);
+ } else {
+ assert(false && "findFirstStoreCondBranchInst() should return a "
+ "store/condition branch instruction");
+ }
+ }
+ return Changed;
+}
+
+// Inserts a fake conditional branch right after the instruction 'SplitInst',
+// and the branch condition is 'Condition'. 'SplitInst' will be placed in the
+// newly created block.
+void AddFakeConditionalBranch(Instruction* SplitInst, Value* Condition) {
+ auto* BB = SplitInst->getParent();
+ TerminatorInst* ThenTerm = nullptr;
+ TerminatorInst* ElseTerm = nullptr;
+ SplitBlockAndInsertIfThenElse(Condition, SplitInst, &ThenTerm, &ElseTerm);
+ assert(ThenTerm && ElseTerm &&
+ "Then/Else terminators cannot be empty after basic block spliting");
+ auto* ThenBB = ThenTerm->getParent();
+ auto* ElseBB = ElseTerm->getParent();
+ auto* TailBB = ThenBB->getSingleSuccessor();
+ assert(TailBB && "Tail block cannot be empty after basic block spliting");
+
+ ThenBB->disableCanEliminateBlock();
+ ThenBB->disableCanEliminateBlock();
+ TailBB->disableCanEliminateBlock();
+ ThenBB->setName(BB->getName() + "Then.Fake");
+ ElseBB->setName(BB->getName() + "Else.Fake");
+ DEBUG(dbgs() << "Add fake conditional branch:\n"
+ << "Then Block:\n"
+ << *ThenBB << "Else Block:\n"
+ << *ElseBB << "\n");
+}
+
+// Returns true if the code is changed, and false otherwise.
+void TaintRelaxedLoads(LoadInst* LI) {
+ // For better performance, we can add a "AND X 0" instruction before the
+ // condition.
+ auto* FirstInst = findFirstStoreCondBranchInst(LI);
+ Instruction* InsertPoint = nullptr;
+ if (FirstInst == nullptr) {
+ InsertPoint = LI->getParent()->getTerminator();
+ InsertPoint = LI->getNextNode();
+ } else {
+ InsertPoint = LI->getNextNode();
+ }
+ IRBuilder<true, NoFolder> Builder(InsertPoint);
+ auto* AndZero = dyn_cast<Instruction>(
+ Builder.CreateAnd(LI, Constant::getNullValue(LI->getType())));
+ auto* FakeCondition = dyn_cast<Instruction>(Builder.CreateICmp(
+ CmpInst::ICMP_NE, AndZero, Constant::getNullValue(LI->getType())));
+ AddFakeConditionalBranch(FakeCondition->getNextNode(), FakeCondition);
+}
+
+// XXX-comment: Returns whether the code has been changed.
+bool AddFakeConditionalBranchAfterMonotonicLoads(
+ const SmallVector<LoadInst*, 1>& MonotonicLoadInsts) {
+ bool Changed = false;
+ for (auto* LI : MonotonicLoadInsts) {
+ auto* FirstInst = findFirstStoreCondBranchInst(LI);
+ if (FirstInst != nullptr) {
+ if (FirstInst->getOpcode() == Instruction::Store) {
+ if (StoreAddressDependOnValue(dyn_cast<StoreInst>(FirstInst), LI)) {
+ continue;
+ }
+ } else if (FirstInst->getOpcode() == Instruction::Br) {
+ if (ConditionalBranchDependsOnValue(dyn_cast<BranchInst>(FirstInst),
+ LI)) {
+ continue;
+ }
+ } else {
+ dbgs() << "FirstInst=" << *FirstInst << "\n";
+ assert(false && "findFirstStoreCondBranchInst() should return a "
+ "store/condition branch instruction");
+ }
+ }
+
+ // We really need to process the relaxed load now.
+ StoreInst* SI = nullptr;;
+ if (FirstInst && (SI = dyn_cast<StoreInst>(FirstInst))) {
+ // For immediately coming stores, taint the address of the store.
+ taintStoreAddress(SI, LI);
+ } else {
+ // For immediately coming branch, directly add a fake branch.
+ TaintRelaxedLoads(LI);
+ Changed = true;
+ }
+ }
+ return Changed;
+}
+
+/**** Implementations of public methods for dependence tainting ****/
+Value* GetUntaintedAddress(Value* CurrentAddress) {
+ auto* OrAddress = getOrAddress(CurrentAddress);
+ if (OrAddress == nullptr) {
+ // Is it tainted by a select instruction?
+ auto* Inst = dyn_cast<Instruction>(CurrentAddress);
+ if (nullptr != Inst && Inst->getOpcode() == Instruction::Select) {
+ // A selection instruction.
+ if (Inst->getOperand(1) == Inst->getOperand(2)) {
+ return Inst->getOperand(1);
+ }
+ }
+
+ return CurrentAddress;
+ }
+ Value* ActualAddress = nullptr;
+
+ auto* CastToInt = dyn_cast<Instruction>(OrAddress->getOperand(1));
+ if (CastToInt && CastToInt->getOpcode() == Instruction::PtrToInt) {
+ return CastToInt->getOperand(0);
+ } else {
+ // This should be a IntToPtr constant expression.
+ ConstantExpr* PtrToIntExpr =
+ dyn_cast<ConstantExpr>(OrAddress->getOperand(1));
+ if (PtrToIntExpr && PtrToIntExpr->getOpcode() == Instruction::PtrToInt) {
+ return PtrToIntExpr->getOperand(0);
+ }
+ }
+
+ // Looks like it's not been dependence-tainted. Returns itself.
+ return CurrentAddress;
+}
+
+MemoryLocation GetUntaintedMemoryLocation(StoreInst* SI) {
+ AAMDNodes AATags;
+ SI->getAAMetadata(AATags);
+ const auto& DL = SI->getModule()->getDataLayout();
+ const auto* OriginalAddr = GetUntaintedAddress(SI->getPointerOperand());
+ DEBUG(if (OriginalAddr != SI->getPointerOperand()) {
+ dbgs() << "[GetUntaintedMemoryLocation]\n"
+ << "Storing address: " << *SI->getPointerOperand()
+ << "\nUntainted address: " << *OriginalAddr << "\n";
+ });
+ return MemoryLocation(OriginalAddr,
+ DL.getTypeStoreSize(SI->getValueOperand()->getType()),
+ AATags);
+}
+
+bool TaintDependenceToStore(StoreInst* SI, Value* DepVal) {
+ if (dependenceSetInclusion(SI, DepVal)) {
+ return false;
+ }
+
+ bool tainted = taintStoreAddress(SI, DepVal);
+ assert(tainted);
+ return tainted;
+}
+
+bool TaintDependenceToStoreAddress(StoreInst* SI, Value* DepVal) {
+ if (dependenceSetInclusion(SI->getPointerOperand(), DepVal)) {
+ return false;
+ }
+
+ bool tainted = taintStoreAddress(SI, DepVal);
+ assert(tainted);
+ return tainted;
+}
+
+bool CompressTaintedStore(BasicBlock* BB) {
+ // This function looks for windows of adajcent stores in 'BB' that satisfy the
+ // following condition (and then do optimization):
+ // *Addr(d1) = v1, d1 is a condition and is the only dependence the store's
+ // address depends on && Dep(v1) includes Dep(d1);
+ // *Addr(d2) = v2, d2 is a condition and is the only dependnece the store's
+ // address depends on && Dep(v2) includes Dep(d2) &&
+ // Dep(d2) includes Dep(d1);
+ // ...
+ // *Addr(dN) = vN, dN is a condition and is the only dependence the store's
+ // address depends on && Dep(dN) includes Dep(d"N-1").
+ //
+ // As a result, Dep(dN) includes [Dep(d1) V ... V Dep(d"N-1")], so we can
+ // safely transform the above to the following. In between these stores, we
+ // can omit untainted stores to the same address 'Addr' since they internally
+ // have dependence on the previous stores on the same address.
+ // =>
+ // *Addr = v1
+ // *Addr = v2
+ // *Addr(d3) = v3
+ for (auto BI = BB->begin(), BE = BB->end(); BI != BE; BI++) {
+ // Look for the first store in such a window of adajacent stores.
+ auto* FirstSI = dyn_cast<StoreInst>(&*BI);
+ if (!FirstSI) {
+ continue;
+ }
+
+ // The first store in the window must be tainted.
+ auto* UntaintedAddress = GetUntaintedAddress(FirstSI->getPointerOperand());
+ if (UntaintedAddress == FirstSI->getPointerOperand()) {
+ continue;
+ }
+
+ // The first store's address must directly depend on and only depend on a
+ // condition.
+ auto* FirstSIDepCond = getConditionDependence(FirstSI->getPointerOperand());
+ if (nullptr == FirstSIDepCond) {
+ continue;
+ }
+
+ // Dep(first store's storing value) includes Dep(tainted dependence).
+ if (!dependenceSetInclusion(FirstSI->getValueOperand(), FirstSIDepCond)) {
+ continue;
+ }
+
+ // Look for subsequent stores to the same address that satisfy the condition
+ // of "compressing the dependence".
+ SmallVector<StoreInst*, 8> AdajacentStores;
+ AdajacentStores.push_back(FirstSI);
+ auto BII = BasicBlock::iterator(FirstSI);
+ for (BII++; BII != BE; BII++) {
+ auto* CurrSI = dyn_cast<StoreInst>(&*BII);
+ if (!CurrSI) {
+ if (BII->mayHaveSideEffects()) {
+ // Be conservative. Instructions with side effects are similar to
+ // stores.
+ break;
+ }
+ continue;
+ }
+
+ auto* OrigAddress = GetUntaintedAddress(CurrSI->getPointerOperand());
+ auto* CurrSIDepCond = getConditionDependence(CurrSI->getPointerOperand());
+ // All other stores must satisfy either:
+ // A. 'CurrSI' is an untainted store to the same address, or
+ // B. the combination of the following 5 subconditions:
+ // 1. Tainted;
+ // 2. Untainted address is the same as the group's address;
+ // 3. The address is tainted with a sole value which is a condition;
+ // 4. The storing value depends on the condition in 3.
+ // 5. The condition in 3 depends on the previous stores dependence
+ // condition.
+
+ // Condition A. Should ignore this store directly.
+ if (OrigAddress == CurrSI->getPointerOperand() &&
+ OrigAddress == UntaintedAddress) {
+ continue;
+ }
+ // Check condition B.
+ Value* Cond = nullptr;
+ if (OrigAddress == CurrSI->getPointerOperand() ||
+ OrigAddress != UntaintedAddress || CurrSIDepCond == nullptr ||
+ !dependenceSetInclusion(CurrSI->getValueOperand(), CurrSIDepCond)) {
+ // Check condition 1, 2, 3 & 4.
+ break;
+ }
+
+ // Check condition 5.
+ StoreInst* PrevSI = AdajacentStores[AdajacentStores.size() - 1];
+ auto* PrevSIDepCond = getConditionDependence(PrevSI->getPointerOperand());
+ assert(PrevSIDepCond &&
+ "Store in the group must already depend on a condtion");
+ if (!dependenceSetInclusion(CurrSIDepCond, PrevSIDepCond)) {
+ break;
+ }
+
+ AdajacentStores.push_back(CurrSI);
+ }
+
+ if (AdajacentStores.size() == 1) {
+ // The outer loop should keep looking from the next store.
+ continue;
+ }
+
+ // Now we have such a group of tainted stores to the same address.
+ DEBUG(dbgs() << "[CompressTaintedStore]\n");
+ DEBUG(dbgs() << "Original BB\n");
+ DEBUG(dbgs() << *BB << '\n');
+ auto* LastSI = AdajacentStores[AdajacentStores.size() - 1];
+ for (unsigned i = 0; i < AdajacentStores.size() - 1; ++i) {
+ auto* SI = AdajacentStores[i];
+
+ // Use the original address for stores before the last one.
+ SI->setOperand(1, UntaintedAddress);
+
+ DEBUG(dbgs() << "Store address has been reversed: " << *SI << '\n';);
+ }
+ // XXX-comment: Try to make the last store use fewer registers.
+ // If LastSI's storing value is a select based on the condition with which
+ // its address is tainted, transform the tainted address to a select
+ // instruction, as follows:
+ // r1 = Select Cond ? A : B
+ // r2 = Cond & 0
+ // r3 = Addr | r2
+ // *r3 = r1
+ // ==>
+ // r1 = Select Cond ? A : B
+ // r2 = Select Cond ? Addr : Addr
+ // *r2 = r1
+ // The idea is that both Select instructions depend on the same condition,
+ // so hopefully the backend can generate two cmov instructions for them (and
+ // this saves the number of registers needed).
+ auto* LastSIDep = getConditionDependence(LastSI->getPointerOperand());
+ auto* LastSIValue = dyn_cast<Instruction>(LastSI->getValueOperand());
+ if (LastSIValue && LastSIValue->getOpcode() == Instruction::Select &&
+ LastSIValue->getOperand(0) == LastSIDep) {
+ // XXX-comment: Maybe it's better for us to just leave it as an and/or
+ // dependence pattern.
+ // /*
+ IRBuilder<true, NoFolder> Builder(LastSI);
+ auto* Address =
+ Builder.CreateSelect(LastSIDep, UntaintedAddress, UntaintedAddress);
+ LastSI->setOperand(1, Address);
+ DEBUG(dbgs() << "The last store becomes :" << *LastSI << "\n\n";);
+ // */
+ }
+ }
+
+ return true;
+}
+
+bool PassDependenceToStore(Value* OldAddress, StoreInst* NewStore) {
+ Value* OldDep = getDependence(OldAddress);
+ // Return false when there's no dependence to pass from the OldAddress.
+ if (!OldDep) {
+ return false;
+ }
+
+ // No need to pass the dependence to NewStore's address if it already depends
+ // on whatever 'OldAddress' depends on.
+ if (StoreAddressDependOnValue(NewStore, OldDep)) {
+ return false;
+ }
+ return taintStoreAddress(NewStore, OldAddress);
+}
+
+SmallSet<Value*, 8> FindDependence(Value* Val) {
+ SmallSet<Value*, 8> DepSet;
+ recursivelyFindDependence(&DepSet, Val, true /*Only insert leaf nodes*/);
+ return DepSet;
+}
+
+bool StoreAddressDependOnValue(StoreInst* SI, Value* DepVal) {
+ return dependenceSetInclusion(SI->getPointerOperand(), DepVal);
+}
+
+bool StoreDependOnValue(StoreInst* SI, Value* Dep) {
+ return dependenceSetInclusion(SI, Dep);
+}
+
+} // namespace
+
+
+
bool CodeGenPrepare::runOnFunction(Function &F) {
+ bool EverMadeChange = false;
+
if (skipOptnoneFunction(F))
return false;
DL = &F.getParent()->getDataLayout();
- bool EverMadeChange = false;
// Clear per function information.
InsertedInsts.clear();
PromotedInsts.clear();
if (!OptSize && TLI && TLI->isSlowDivBypassed()) {
const DenseMap<unsigned int, unsigned int> &BypassWidths =
TLI->getBypassSlowDivWidths();
- for (Function::iterator I = F.begin(); I != F.end(); I++)
- EverMadeChange |= bypassSlowDivision(F, I, BypassWidths);
+ BasicBlock* BB = &*F.begin();
+ while (BB != nullptr) {
+ // bypassSlowDivision may create new BBs, but we don't want to reapply the
+ // optimization to those blocks.
+ BasicBlock* Next = BB->getNextNode();
+ EverMadeChange |= bypassSlowDivision(BB, BypassWidths);
+ BB = Next;
+ }
}
// Eliminate blocks that contain only PHI nodes and an
// unconditional branch.
- EverMadeChange |= EliminateMostlyEmptyBlocks(F);
+ EverMadeChange |= eliminateMostlyEmptyBlocks(F);
// llvm.dbg.value is far away from the value then iSel may not be able
// handle it properly. iSel will drop llvm.dbg.value if it can not
// find a node corresponding to the value.
- EverMadeChange |= PlaceDbgValues(F);
+ EverMadeChange |= placeDbgValues(F);
// If there is a mask, compare against zero, and branch that can be combined
// into a single target instruction, push the mask and compare into branch
while (MadeChange) {
MadeChange = false;
for (Function::iterator I = F.begin(); I != F.end(); ) {
- BasicBlock *BB = I++;
+ BasicBlock *BB = &*I++;
bool ModifiedDTOnIteration = false;
- MadeChange |= OptimizeBlock(*BB, ModifiedDTOnIteration);
+ MadeChange |= optimizeBlock(*BB, ModifiedDTOnIteration);
// Restart BB iteration if the dominator tree of the Function was changed
if (ModifiedDTOnIteration)
// Merge pairs of basic blocks with unconditional branches, connected by
// a single edge.
if (EverMadeChange || MadeChange)
- MadeChange |= EliminateFallThrough(F);
+ MadeChange |= eliminateFallThrough(F);
EverMadeChange |= MadeChange;
}
EverMadeChange |= simplifyOffsetableRelocate(*I);
}
+ // XXX-comment: Delay dealing with relaxed loads in this function to avoid
+ // further changes done by other passes (e.g., SimplifyCFG).
+ // Collect all the relaxed loads.
+ SmallVector<LoadInst*, 1> MonotonicLoadInsts;
+ for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I) {
+ if (I->isAtomic()) {
+ switch (I->getOpcode()) {
+ case Instruction::Load: {
+ auto* LI = dyn_cast<LoadInst>(&*I);
+ if (LI->getOrdering() == Monotonic) {
+ MonotonicLoadInsts.push_back(LI);
+ }
+ break;
+ }
+ default: {
+ break;
+ }
+ }
+ }
+ }
+ EverMadeChange |=
+ AddFakeConditionalBranchAfterMonotonicLoads(MonotonicLoadInsts);
+
return EverMadeChange;
}
-/// EliminateFallThrough - Merge basic blocks which are connected
-/// by a single edge, where one of the basic blocks has a single successor
-/// pointing to the other basic block, which has a single predecessor.
-bool CodeGenPrepare::EliminateFallThrough(Function &F) {
+/// Merge basic blocks which are connected by a single edge, where one of the
+/// basic blocks has a single successor pointing to the other basic block,
+/// which has a single predecessor.
+bool CodeGenPrepare::eliminateFallThrough(Function &F) {
bool Changed = false;
// Scan all of the blocks in the function, except for the entry block.
for (Function::iterator I = std::next(F.begin()), E = F.end(); I != E;) {
- BasicBlock *BB = I++;
+ BasicBlock *BB = &*I++;
// If the destination block has a single pred, then this is a trivial
// edge, just collapse it.
BasicBlock *SinglePred = BB->getSinglePredecessor();
BB->moveBefore(&BB->getParent()->getEntryBlock());
// We have erased a block. Update the iterator.
- I = BB;
+ I = BB->getIterator();
}
}
return Changed;
}
-/// EliminateMostlyEmptyBlocks - eliminate blocks that contain only PHI nodes,
-/// debug info directives, and an unconditional branch. Passes before isel
-/// (e.g. LSR/loopsimplify) often split edges in ways that are non-optimal for
-/// isel. Start by eliminating these blocks so we can split them the way we
-/// want them.
-bool CodeGenPrepare::EliminateMostlyEmptyBlocks(Function &F) {
+/// Eliminate blocks that contain only PHI nodes, debug info directives, and an
+/// unconditional branch. Passes before isel (e.g. LSR/loopsimplify) often split
+/// edges in ways that are non-optimal for isel. Start by eliminating these
+/// blocks so we can split them the way we want them.
+bool CodeGenPrepare::eliminateMostlyEmptyBlocks(Function &F) {
bool MadeChange = false;
// Note that this intentionally skips the entry block.
for (Function::iterator I = std::next(F.begin()), E = F.end(); I != E;) {
- BasicBlock *BB = I++;
-
+ BasicBlock *BB = &*I++;
// If this block doesn't end with an uncond branch, ignore it.
BranchInst *BI = dyn_cast<BranchInst>(BB->getTerminator());
if (!BI || !BI->isUnconditional())
// If the instruction before the branch (skipping debug info) isn't a phi
// node, then other stuff is happening here.
- BasicBlock::iterator BBI = BI;
+ BasicBlock::iterator BBI = BI->getIterator();
if (BBI != BB->begin()) {
--BBI;
while (isa<DbgInfoIntrinsic>(BBI)) {
if (DestBB == BB)
continue;
- if (!CanMergeBlocks(BB, DestBB))
+ if (!canMergeBlocks(BB, DestBB))
continue;
- EliminateMostlyEmptyBlock(BB);
+ eliminateMostlyEmptyBlock(BB);
MadeChange = true;
}
return MadeChange;
}
-/// CanMergeBlocks - Return true if we can merge BB into DestBB if there is a
-/// single uncond branch between them, and BB contains no other non-phi
+/// Return true if we can merge BB into DestBB if there is a single
+/// unconditional branch between them, and BB contains no other non-phi
/// instructions.
-bool CodeGenPrepare::CanMergeBlocks(const BasicBlock *BB,
+bool CodeGenPrepare::canMergeBlocks(const BasicBlock *BB,
const BasicBlock *DestBB) const {
// We only want to eliminate blocks whose phi nodes are used by phi nodes in
// the successor. If there are more complex condition (e.g. preheaders),
}
-/// EliminateMostlyEmptyBlock - Eliminate a basic block that have only phi's and
-/// an unconditional branch in it.
-void CodeGenPrepare::EliminateMostlyEmptyBlock(BasicBlock *BB) {
+/// Eliminate a basic block that has only phi's and an unconditional branch in
+/// it.
+void CodeGenPrepare::eliminateMostlyEmptyBlock(BasicBlock *BB) {
BranchInst *BI = cast<BranchInst>(BB->getTerminator());
BasicBlock *DestBB = BI->getSuccessor(0);
// Computes a map of base pointer relocation instructions to corresponding
// derived pointer relocation instructions given a vector of all relocate calls
static void computeBaseDerivedRelocateMap(
- const SmallVectorImpl<User *> &AllRelocateCalls,
- DenseMap<IntrinsicInst *, SmallVector<IntrinsicInst *, 2>> &
- RelocateInstMap) {
+ const SmallVectorImpl<GCRelocateInst *> &AllRelocateCalls,
+ DenseMap<GCRelocateInst *, SmallVector<GCRelocateInst *, 2>>
+ &RelocateInstMap) {
// Collect information in two maps: one primarily for locating the base object
// while filling the second map; the second map is the final structure holding
// a mapping between Base and corresponding Derived relocate calls
- DenseMap<std::pair<unsigned, unsigned>, IntrinsicInst *> RelocateIdxMap;
- for (auto &U : AllRelocateCalls) {
- GCRelocateOperands ThisRelocate(U);
- IntrinsicInst *I = cast<IntrinsicInst>(U);
- auto K = std::make_pair(ThisRelocate.getBasePtrIndex(),
- ThisRelocate.getDerivedPtrIndex());
- RelocateIdxMap.insert(std::make_pair(K, I));
+ DenseMap<std::pair<unsigned, unsigned>, GCRelocateInst *> RelocateIdxMap;
+ for (auto *ThisRelocate : AllRelocateCalls) {
+ auto K = std::make_pair(ThisRelocate->getBasePtrIndex(),
+ ThisRelocate->getDerivedPtrIndex());
+ RelocateIdxMap.insert(std::make_pair(K, ThisRelocate));
}
for (auto &Item : RelocateIdxMap) {
std::pair<unsigned, unsigned> Key = Item.first;
// Base relocation: nothing to insert
continue;
- IntrinsicInst *I = Item.second;
+ GCRelocateInst *I = Item.second;
auto BaseKey = std::make_pair(Key.first, Key.first);
// We're iterating over RelocateIdxMap so we cannot modify it.
// Takes a RelocatedBase (base pointer relocation instruction) and Targets to
// replace, computes a replacement, and affects it.
static bool
-simplifyRelocatesOffABase(IntrinsicInst *RelocatedBase,
- const SmallVectorImpl<IntrinsicInst *> &Targets) {
+simplifyRelocatesOffABase(GCRelocateInst *RelocatedBase,
+ const SmallVectorImpl<GCRelocateInst *> &Targets) {
bool MadeChange = false;
- for (auto &ToReplace : Targets) {
- GCRelocateOperands MasterRelocate(RelocatedBase);
- GCRelocateOperands ThisRelocate(ToReplace);
-
- assert(ThisRelocate.getBasePtrIndex() == MasterRelocate.getBasePtrIndex() &&
+ for (GCRelocateInst *ToReplace : Targets) {
+ assert(ToReplace->getBasePtrIndex() == RelocatedBase->getBasePtrIndex() &&
"Not relocating a derived object of the original base object");
- if (ThisRelocate.getBasePtrIndex() == ThisRelocate.getDerivedPtrIndex()) {
+ if (ToReplace->getBasePtrIndex() == ToReplace->getDerivedPtrIndex()) {
// A duplicate relocate call. TODO: coalesce duplicates.
continue;
}
- Value *Base = ThisRelocate.getBasePtr();
- auto Derived = dyn_cast<GetElementPtrInst>(ThisRelocate.getDerivedPtr());
+ if (RelocatedBase->getParent() != ToReplace->getParent()) {
+ // Base and derived relocates are in different basic blocks.
+ // In this case transform is only valid when base dominates derived
+ // relocate. However it would be too expensive to check dominance
+ // for each such relocate, so we skip the whole transformation.
+ continue;
+ }
+
+ Value *Base = ToReplace->getBasePtr();
+ auto Derived = dyn_cast<GetElementPtrInst>(ToReplace->getDerivedPtr());
if (!Derived || Derived->getPointerOperand() != Base)
continue;
// In this case, we can not find the bitcast any more. So we insert a new bitcast
// no matter there is already one or not. In this way, we can handle all cases, and
// the extra bitcast should be optimized away in later passes.
- Instruction *ActualRelocatedBase = RelocatedBase;
+ Value *ActualRelocatedBase = RelocatedBase;
if (RelocatedBase->getType() != Base->getType()) {
ActualRelocatedBase =
- cast<Instruction>(Builder.CreateBitCast(RelocatedBase, Base->getType()));
+ Builder.CreateBitCast(RelocatedBase, Base->getType());
}
Value *Replacement = Builder.CreateGEP(
Derived->getSourceElementType(), ActualRelocatedBase, makeArrayRef(OffsetV));
- Instruction *ReplacementInst = cast<Instruction>(Replacement);
Replacement->takeName(ToReplace);
// If the newly generated derived pointer's type does not match the original derived
// pointer's type, cast the new derived pointer to match it. Same reasoning as above.
- Instruction *ActualReplacement = ReplacementInst;
- if (ReplacementInst->getType() != ToReplace->getType()) {
+ Value *ActualReplacement = Replacement;
+ if (Replacement->getType() != ToReplace->getType()) {
ActualReplacement =
- cast<Instruction>(Builder.CreateBitCast(ReplacementInst, ToReplace->getType()));
+ Builder.CreateBitCast(Replacement, ToReplace->getType());
}
ToReplace->replaceAllUsesWith(ActualReplacement);
ToReplace->eraseFromParent();
// %val = load %ptr'
bool CodeGenPrepare::simplifyOffsetableRelocate(Instruction &I) {
bool MadeChange = false;
- SmallVector<User *, 2> AllRelocateCalls;
+ SmallVector<GCRelocateInst *, 2> AllRelocateCalls;
for (auto *U : I.users())
- if (isGCRelocate(dyn_cast<Instruction>(U)))
+ if (GCRelocateInst *Relocate = dyn_cast<GCRelocateInst>(U))
// Collect all the relocate calls associated with a statepoint
- AllRelocateCalls.push_back(U);
+ AllRelocateCalls.push_back(Relocate);
// We need atleast one base pointer relocation + one derived pointer
// relocation to mangle
// RelocateInstMap is a mapping from the base relocate instruction to the
// corresponding derived relocate instructions
- DenseMap<IntrinsicInst *, SmallVector<IntrinsicInst *, 2>> RelocateInstMap;
+ DenseMap<GCRelocateInst *, SmallVector<GCRelocateInst *, 2>> RelocateInstMap;
computeBaseDerivedRelocateMap(AllRelocateCalls, RelocateInstMap);
if (RelocateInstMap.empty())
return false;
// Preincrement use iterator so we don't invalidate it.
++UI;
+ // If the block selected to receive the cast is an EH pad that does not
+ // allow non-PHI instructions before the terminator, we can't sink the
+ // cast.
+ if (UserBB->getTerminator()->isEHPad())
+ continue;
+
// If this user is in the same block as the cast, don't change the cast.
if (UserBB == DefBB) continue;
if (!InsertedCast) {
BasicBlock::iterator InsertPt = UserBB->getFirstInsertionPt();
- InsertedCast =
- CastInst::Create(CI->getOpcode(), CI->getOperand(0), CI->getType(), "",
- InsertPt);
+ assert(InsertPt != UserBB->end());
+ InsertedCast = CastInst::Create(CI->getOpcode(), CI->getOperand(0),
+ CI->getType(), "", &*InsertPt);
}
// Replace a use of the cast with a use of the new cast.
return MadeChange;
}
-/// OptimizeNoopCopyExpression - If the specified cast instruction is a noop
-/// copy (e.g. it's casting from one pointer type to another, i32->i8 on PPC),
-/// sink it into user blocks to reduce the number of virtual
-/// registers that must be created and coalesced.
+/// If the specified cast instruction is a noop copy (e.g. it's casting from
+/// one pointer type to another, i32->i8 on PPC), sink it into user blocks to
+/// reduce the number of virtual registers that must be created and coalesced.
///
/// Return true if any changes are made.
///
return SinkCast(CI);
}
-/// CombineUAddWithOverflow - try to combine CI into a call to the
-/// llvm.uadd.with.overflow intrinsic if possible.
+/// Try to combine CI into a call to the llvm.uadd.with.overflow intrinsic if
+/// possible.
///
/// Return true if any changes were made.
static bool CombineUAddWithOverflow(CmpInst *CI) {
assert(*AddI->user_begin() == CI && "expected!");
#endif
- Module *M = CI->getParent()->getParent()->getParent();
+ Module *M = CI->getModule();
Value *F = Intrinsic::getDeclaration(M, Intrinsic::uadd_with_overflow, Ty);
auto *InsertPt = AddI->hasOneUse() ? CI : AddI;
return true;
}
-/// SinkCmpExpression - Sink the given CmpInst into user blocks to reduce
-/// the number of virtual registers that must be created and coalesced. This is
-/// a clear win except on targets with multiple condition code registers
-/// (PowerPC), where it might lose; some adjustment may be wanted there.
+/// Sink the given CmpInst into user blocks to reduce the number of virtual
+/// registers that must be created and coalesced. This is a clear win except on
+/// targets with multiple condition code registers (PowerPC), where it might
+/// lose; some adjustment may be wanted there.
///
/// Return true if any changes are made.
static bool SinkCmpExpression(CmpInst *CI) {
BasicBlock *DefBB = CI->getParent();
- /// InsertedCmp - Only insert a cmp in each block once.
+ /// Only insert a cmp in each block once.
DenseMap<BasicBlock*, CmpInst*> InsertedCmps;
bool MadeChange = false;
if (!InsertedCmp) {
BasicBlock::iterator InsertPt = UserBB->getFirstInsertionPt();
+ assert(InsertPt != UserBB->end());
InsertedCmp =
- CmpInst::Create(CI->getOpcode(),
- CI->getPredicate(), CI->getOperand(0),
- CI->getOperand(1), "", InsertPt);
+ CmpInst::Create(CI->getOpcode(), CI->getPredicate(),
+ CI->getOperand(0), CI->getOperand(1), "", &*InsertPt);
}
// Replace a use of the cmp with a use of the new cmp.
return false;
}
-/// isExtractBitsCandidateUse - Check if the candidates could
-/// be combined with shift instruction, which includes:
+/// Check if the candidates could be combined with a shift instruction, which
+/// includes:
/// 1. Truncate instruction
/// 2. And instruction and the imm is a mask of the low bits:
/// imm & (imm+1) == 0
return true;
}
-/// SinkShiftAndTruncate - sink both shift and truncate instruction
-/// to the use of truncate's BB.
+/// Sink both shift and truncate instruction to the use of truncate's BB.
static bool
SinkShiftAndTruncate(BinaryOperator *ShiftI, Instruction *User, ConstantInt *CI,
DenseMap<BasicBlock *, BinaryOperator *> &InsertedShifts,
if (!InsertedShift && !InsertedTrunc) {
BasicBlock::iterator InsertPt = TruncUserBB->getFirstInsertionPt();
+ assert(InsertPt != TruncUserBB->end());
// Sink the shift
if (ShiftI->getOpcode() == Instruction::AShr)
- InsertedShift =
- BinaryOperator::CreateAShr(ShiftI->getOperand(0), CI, "", InsertPt);
+ InsertedShift = BinaryOperator::CreateAShr(ShiftI->getOperand(0), CI,
+ "", &*InsertPt);
else
- InsertedShift =
- BinaryOperator::CreateLShr(ShiftI->getOperand(0), CI, "", InsertPt);
+ InsertedShift = BinaryOperator::CreateLShr(ShiftI->getOperand(0), CI,
+ "", &*InsertPt);
// Sink the trunc
BasicBlock::iterator TruncInsertPt = TruncUserBB->getFirstInsertionPt();
TruncInsertPt++;
+ assert(TruncInsertPt != TruncUserBB->end());
InsertedTrunc = CastInst::Create(TruncI->getOpcode(), InsertedShift,
- TruncI->getType(), "", TruncInsertPt);
+ TruncI->getType(), "", &*TruncInsertPt);
MadeChange = true;
return MadeChange;
}
-/// OptimizeExtractBits - sink the shift *right* instruction into user blocks if
-/// the uses could potentially be combined with this shift instruction and
-/// generate BitExtract instruction. It will only be applied if the architecture
-/// supports BitExtract instruction. Here is an example:
+/// Sink the shift *right* instruction into user blocks if the uses could
+/// potentially be combined with this shift instruction and generate BitExtract
+/// instruction. It will only be applied if the architecture supports BitExtract
+/// instruction. Here is an example:
/// BB1:
/// %x.extract.shift = lshr i64 %arg1, 32
/// BB2:
if (!InsertedShift) {
BasicBlock::iterator InsertPt = UserBB->getFirstInsertionPt();
+ assert(InsertPt != UserBB->end());
if (ShiftI->getOpcode() == Instruction::AShr)
- InsertedShift =
- BinaryOperator::CreateAShr(ShiftI->getOperand(0), CI, "", InsertPt);
+ InsertedShift = BinaryOperator::CreateAShr(ShiftI->getOperand(0), CI,
+ "", &*InsertPt);
else
- InsertedShift =
- BinaryOperator::CreateLShr(ShiftI->getOperand(0), CI, "", InsertPt);
+ InsertedShift = BinaryOperator::CreateLShr(ShiftI->getOperand(0), CI,
+ "", &*InsertPt);
+
+ MadeChange = true;
+ }
+
+ // Replace a use of the shift with a use of the new shift.
+ TheUse = InsertedShift;
+ }
+
+ // If we removed all uses, nuke the shift.
+ if (ShiftI->use_empty())
+ ShiftI->eraseFromParent();
+
+ return MadeChange;
+}
+
+// Translate a masked load intrinsic like
+// <16 x i32 > @llvm.masked.load( <16 x i32>* %addr, i32 align,
+// <16 x i1> %mask, <16 x i32> %passthru)
+// to a chain of basic blocks, with loading element one-by-one if
+// the appropriate mask bit is set
+//
+// %1 = bitcast i8* %addr to i32*
+// %2 = extractelement <16 x i1> %mask, i32 0
+// %3 = icmp eq i1 %2, true
+// br i1 %3, label %cond.load, label %else
+//
+//cond.load: ; preds = %0
+// %4 = getelementptr i32* %1, i32 0
+// %5 = load i32* %4
+// %6 = insertelement <16 x i32> undef, i32 %5, i32 0
+// br label %else
+//
+//else: ; preds = %0, %cond.load
+// %res.phi.else = phi <16 x i32> [ %6, %cond.load ], [ undef, %0 ]
+// %7 = extractelement <16 x i1> %mask, i32 1
+// %8 = icmp eq i1 %7, true
+// br i1 %8, label %cond.load1, label %else2
+//
+//cond.load1: ; preds = %else
+// %9 = getelementptr i32* %1, i32 1
+// %10 = load i32* %9
+// %11 = insertelement <16 x i32> %res.phi.else, i32 %10, i32 1
+// br label %else2
+//
+//else2: ; preds = %else, %cond.load1
+// %res.phi.else3 = phi <16 x i32> [ %11, %cond.load1 ], [ %res.phi.else, %else ]
+// %12 = extractelement <16 x i1> %mask, i32 2
+// %13 = icmp eq i1 %12, true
+// br i1 %13, label %cond.load4, label %else5
+//
+static void ScalarizeMaskedLoad(CallInst *CI) {
+ Value *Ptr = CI->getArgOperand(0);
+ Value *Alignment = CI->getArgOperand(1);
+ Value *Mask = CI->getArgOperand(2);
+ Value *Src0 = CI->getArgOperand(3);
+
+ unsigned AlignVal = cast<ConstantInt>(Alignment)->getZExtValue();
+ VectorType *VecType = dyn_cast<VectorType>(CI->getType());
+ assert(VecType && "Unexpected return type of masked load intrinsic");
+
+ Type *EltTy = CI->getType()->getVectorElementType();
+
+ IRBuilder<> Builder(CI->getContext());
+ Instruction *InsertPt = CI;
+ BasicBlock *IfBlock = CI->getParent();
+ BasicBlock *CondBlock = nullptr;
+ BasicBlock *PrevIfBlock = CI->getParent();
+
+ Builder.SetInsertPoint(InsertPt);
+ Builder.SetCurrentDebugLocation(CI->getDebugLoc());
+
+ // Short-cut if the mask is all-true.
+ bool IsAllOnesMask = isa<Constant>(Mask) &&
+ cast<Constant>(Mask)->isAllOnesValue();
+
+ if (IsAllOnesMask) {
+ Value *NewI = Builder.CreateAlignedLoad(Ptr, AlignVal);
+ CI->replaceAllUsesWith(NewI);
+ CI->eraseFromParent();
+ return;
+ }
+
+ // Adjust alignment for the scalar instruction.
+ AlignVal = std::min(AlignVal, VecType->getScalarSizeInBits()/8);
+ // Bitcast %addr fron i8* to EltTy*
+ Type *NewPtrType =
+ EltTy->getPointerTo(cast<PointerType>(Ptr->getType())->getAddressSpace());
+ Value *FirstEltPtr = Builder.CreateBitCast(Ptr, NewPtrType);
+ unsigned VectorWidth = VecType->getNumElements();
+
+ Value *UndefVal = UndefValue::get(VecType);
+
+ // The result vector
+ Value *VResult = UndefVal;
+
+ if (isa<ConstantVector>(Mask)) {
+ for (unsigned Idx = 0; Idx < VectorWidth; ++Idx) {
+ if (cast<ConstantVector>(Mask)->getOperand(Idx)->isNullValue())
+ continue;
+ Value *Gep =
+ Builder.CreateInBoundsGEP(EltTy, FirstEltPtr, Builder.getInt32(Idx));
+ LoadInst* Load = Builder.CreateAlignedLoad(Gep, AlignVal);
+ VResult = Builder.CreateInsertElement(VResult, Load,
+ Builder.getInt32(Idx));
+ }
+ Value *NewI = Builder.CreateSelect(Mask, VResult, Src0);
+ CI->replaceAllUsesWith(NewI);
+ CI->eraseFromParent();
+ return;
+ }
+
+ PHINode *Phi = nullptr;
+ Value *PrevPhi = UndefVal;
+
+ for (unsigned Idx = 0; Idx < VectorWidth; ++Idx) {
+
+ // Fill the "else" block, created in the previous iteration
+ //
+ // %res.phi.else3 = phi <16 x i32> [ %11, %cond.load1 ], [ %res.phi.else, %else ]
+ // %mask_1 = extractelement <16 x i1> %mask, i32 Idx
+ // %to_load = icmp eq i1 %mask_1, true
+ // br i1 %to_load, label %cond.load, label %else
+ //
+ if (Idx > 0) {
+ Phi = Builder.CreatePHI(VecType, 2, "res.phi.else");
+ Phi->addIncoming(VResult, CondBlock);
+ Phi->addIncoming(PrevPhi, PrevIfBlock);
+ PrevPhi = Phi;
+ VResult = Phi;
+ }
+
+ Value *Predicate = Builder.CreateExtractElement(Mask, Builder.getInt32(Idx));
+ Value *Cmp = Builder.CreateICmp(ICmpInst::ICMP_EQ, Predicate,
+ ConstantInt::get(Predicate->getType(), 1));
+
+ // Create "cond" block
+ //
+ // %EltAddr = getelementptr i32* %1, i32 0
+ // %Elt = load i32* %EltAddr
+ // VResult = insertelement <16 x i32> VResult, i32 %Elt, i32 Idx
+ //
+ CondBlock = IfBlock->splitBasicBlock(InsertPt->getIterator(), "cond.load");
+ Builder.SetInsertPoint(InsertPt);
+
+ Value *Gep =
+ Builder.CreateInBoundsGEP(EltTy, FirstEltPtr, Builder.getInt32(Idx));
+ LoadInst *Load = Builder.CreateAlignedLoad(Gep, AlignVal);
+ VResult = Builder.CreateInsertElement(VResult, Load, Builder.getInt32(Idx));
+
+ // Create "else" block, fill it in the next iteration
+ BasicBlock *NewIfBlock =
+ CondBlock->splitBasicBlock(InsertPt->getIterator(), "else");
+ Builder.SetInsertPoint(InsertPt);
+ Instruction *OldBr = IfBlock->getTerminator();
+ BranchInst::Create(CondBlock, NewIfBlock, Cmp, OldBr);
+ OldBr->eraseFromParent();
+ PrevIfBlock = IfBlock;
+ IfBlock = NewIfBlock;
+ }
+
+ Phi = Builder.CreatePHI(VecType, 2, "res.phi.select");
+ Phi->addIncoming(VResult, CondBlock);
+ Phi->addIncoming(PrevPhi, PrevIfBlock);
+ Value *NewI = Builder.CreateSelect(Mask, Phi, Src0);
+ CI->replaceAllUsesWith(NewI);
+ CI->eraseFromParent();
+}
+
+// Translate a masked store intrinsic, like
+// void @llvm.masked.store(<16 x i32> %src, <16 x i32>* %addr, i32 align,
+// <16 x i1> %mask)
+// to a chain of basic blocks, that stores element one-by-one if
+// the appropriate mask bit is set
+//
+// %1 = bitcast i8* %addr to i32*
+// %2 = extractelement <16 x i1> %mask, i32 0
+// %3 = icmp eq i1 %2, true
+// br i1 %3, label %cond.store, label %else
+//
+// cond.store: ; preds = %0
+// %4 = extractelement <16 x i32> %val, i32 0
+// %5 = getelementptr i32* %1, i32 0
+// store i32 %4, i32* %5
+// br label %else
+//
+// else: ; preds = %0, %cond.store
+// %6 = extractelement <16 x i1> %mask, i32 1
+// %7 = icmp eq i1 %6, true
+// br i1 %7, label %cond.store1, label %else2
+//
+// cond.store1: ; preds = %else
+// %8 = extractelement <16 x i32> %val, i32 1
+// %9 = getelementptr i32* %1, i32 1
+// store i32 %8, i32* %9
+// br label %else2
+// . . .
+static void ScalarizeMaskedStore(CallInst *CI) {
+ Value *Src = CI->getArgOperand(0);
+ Value *Ptr = CI->getArgOperand(1);
+ Value *Alignment = CI->getArgOperand(2);
+ Value *Mask = CI->getArgOperand(3);
+
+ unsigned AlignVal = cast<ConstantInt>(Alignment)->getZExtValue();
+ VectorType *VecType = dyn_cast<VectorType>(Src->getType());
+ assert(VecType && "Unexpected data type in masked store intrinsic");
+
+ Type *EltTy = VecType->getElementType();
+
+ IRBuilder<> Builder(CI->getContext());
+ Instruction *InsertPt = CI;
+ BasicBlock *IfBlock = CI->getParent();
+ Builder.SetInsertPoint(InsertPt);
+ Builder.SetCurrentDebugLocation(CI->getDebugLoc());
+
+ // Short-cut if the mask is all-true.
+ bool IsAllOnesMask = isa<Constant>(Mask) &&
+ cast<Constant>(Mask)->isAllOnesValue();
+
+ if (IsAllOnesMask) {
+ Builder.CreateAlignedStore(Src, Ptr, AlignVal);
+ CI->eraseFromParent();
+ return;
+ }
+
+ // Adjust alignment for the scalar instruction.
+ AlignVal = std::max(AlignVal, VecType->getScalarSizeInBits()/8);
+ // Bitcast %addr fron i8* to EltTy*
+ Type *NewPtrType =
+ EltTy->getPointerTo(cast<PointerType>(Ptr->getType())->getAddressSpace());
+ Value *FirstEltPtr = Builder.CreateBitCast(Ptr, NewPtrType);
+ unsigned VectorWidth = VecType->getNumElements();
+
+ if (isa<ConstantVector>(Mask)) {
+ for (unsigned Idx = 0; Idx < VectorWidth; ++Idx) {
+ if (cast<ConstantVector>(Mask)->getOperand(Idx)->isNullValue())
+ continue;
+ Value *OneElt = Builder.CreateExtractElement(Src, Builder.getInt32(Idx));
+ Value *Gep =
+ Builder.CreateInBoundsGEP(EltTy, FirstEltPtr, Builder.getInt32(Idx));
+ Builder.CreateAlignedStore(OneElt, Gep, AlignVal);
+ }
+ CI->eraseFromParent();
+ return;
+ }
+
+ for (unsigned Idx = 0; Idx < VectorWidth; ++Idx) {
+
+ // Fill the "else" block, created in the previous iteration
+ //
+ // %mask_1 = extractelement <16 x i1> %mask, i32 Idx
+ // %to_store = icmp eq i1 %mask_1, true
+ // br i1 %to_store, label %cond.store, label %else
+ //
+ Value *Predicate = Builder.CreateExtractElement(Mask, Builder.getInt32(Idx));
+ Value *Cmp = Builder.CreateICmp(ICmpInst::ICMP_EQ, Predicate,
+ ConstantInt::get(Predicate->getType(), 1));
+
+ // Create "cond" block
+ //
+ // %OneElt = extractelement <16 x i32> %Src, i32 Idx
+ // %EltAddr = getelementptr i32* %1, i32 0
+ // %store i32 %OneElt, i32* %EltAddr
+ //
+ BasicBlock *CondBlock =
+ IfBlock->splitBasicBlock(InsertPt->getIterator(), "cond.store");
+ Builder.SetInsertPoint(InsertPt);
- MadeChange = true;
- }
+ Value *OneElt = Builder.CreateExtractElement(Src, Builder.getInt32(Idx));
+ Value *Gep =
+ Builder.CreateInBoundsGEP(EltTy, FirstEltPtr, Builder.getInt32(Idx));
+ Builder.CreateAlignedStore(OneElt, Gep, AlignVal);
- // Replace a use of the shift with a use of the new shift.
- TheUse = InsertedShift;
+ // Create "else" block, fill it in the next iteration
+ BasicBlock *NewIfBlock =
+ CondBlock->splitBasicBlock(InsertPt->getIterator(), "else");
+ Builder.SetInsertPoint(InsertPt);
+ Instruction *OldBr = IfBlock->getTerminator();
+ BranchInst::Create(CondBlock, NewIfBlock, Cmp, OldBr);
+ OldBr->eraseFromParent();
+ IfBlock = NewIfBlock;
}
-
- // If we removed all uses, nuke the shift.
- if (ShiftI->use_empty())
- ShiftI->eraseFromParent();
-
- return MadeChange;
+ CI->eraseFromParent();
}
-// ScalarizeMaskedLoad() translates masked load intrinsic, like
-// <16 x i32 > @llvm.masked.load( <16 x i32>* %addr, i32 align,
-// <16 x i1> %mask, <16 x i32> %passthru)
+// Translate a masked gather intrinsic like
+// <16 x i32 > @llvm.masked.gather.v16i32( <16 x i32*> %Ptrs, i32 4,
+// <16 x i1> %Mask, <16 x i32> %Src)
// to a chain of basic blocks, with loading element one-by-one if
// the appropriate mask bit is set
-//
-// %1 = bitcast i8* %addr to i32*
-// %2 = extractelement <16 x i1> %mask, i32 0
-// %3 = icmp eq i1 %2, true
-// br i1 %3, label %cond.load, label %else
-//
-//cond.load: ; preds = %0
-// %4 = getelementptr i32* %1, i32 0
-// %5 = load i32* %4
-// %6 = insertelement <16 x i32> undef, i32 %5, i32 0
-// br label %else
//
-//else: ; preds = %0, %cond.load
-// %res.phi.else = phi <16 x i32> [ %6, %cond.load ], [ undef, %0 ]
-// %7 = extractelement <16 x i1> %mask, i32 1
-// %8 = icmp eq i1 %7, true
-// br i1 %8, label %cond.load1, label %else2
+// % Ptrs = getelementptr i32, i32* %base, <16 x i64> %ind
+// % Mask0 = extractelement <16 x i1> %Mask, i32 0
+// % ToLoad0 = icmp eq i1 % Mask0, true
+// br i1 % ToLoad0, label %cond.load, label %else
//
-//cond.load1: ; preds = %else
-// %9 = getelementptr i32* %1, i32 1
-// %10 = load i32* %9
-// %11 = insertelement <16 x i32> %res.phi.else, i32 %10, i32 1
-// br label %else2
+// cond.load:
+// % Ptr0 = extractelement <16 x i32*> %Ptrs, i32 0
+// % Load0 = load i32, i32* % Ptr0, align 4
+// % Res0 = insertelement <16 x i32> undef, i32 % Load0, i32 0
+// br label %else
//
-//else2: ; preds = %else, %cond.load1
-// %res.phi.else3 = phi <16 x i32> [ %11, %cond.load1 ], [ %res.phi.else, %else ]
-// %12 = extractelement <16 x i1> %mask, i32 2
-// %13 = icmp eq i1 %12, true
-// br i1 %13, label %cond.load4, label %else5
+// else:
+// %res.phi.else = phi <16 x i32>[% Res0, %cond.load], [undef, % 0]
+// % Mask1 = extractelement <16 x i1> %Mask, i32 1
+// % ToLoad1 = icmp eq i1 % Mask1, true
+// br i1 % ToLoad1, label %cond.load1, label %else2
//
-static void ScalarizeMaskedLoad(CallInst *CI) {
- Value *Ptr = CI->getArgOperand(0);
- Value *Src0 = CI->getArgOperand(3);
+// cond.load1:
+// % Ptr1 = extractelement <16 x i32*> %Ptrs, i32 1
+// % Load1 = load i32, i32* % Ptr1, align 4
+// % Res1 = insertelement <16 x i32> %res.phi.else, i32 % Load1, i32 1
+// br label %else2
+// . . .
+// % Result = select <16 x i1> %Mask, <16 x i32> %res.phi.select, <16 x i32> %Src
+// ret <16 x i32> %Result
+static void ScalarizeMaskedGather(CallInst *CI) {
+ Value *Ptrs = CI->getArgOperand(0);
+ Value *Alignment = CI->getArgOperand(1);
Value *Mask = CI->getArgOperand(2);
+ Value *Src0 = CI->getArgOperand(3);
+
VectorType *VecType = dyn_cast<VectorType>(CI->getType());
- Type *EltTy = VecType->getElementType();
assert(VecType && "Unexpected return type of masked load intrinsic");
BasicBlock *CondBlock = nullptr;
BasicBlock *PrevIfBlock = CI->getParent();
Builder.SetInsertPoint(InsertPt);
+ unsigned AlignVal = cast<ConstantInt>(Alignment)->getZExtValue();
Builder.SetCurrentDebugLocation(CI->getDebugLoc());
- // Bitcast %addr fron i8* to EltTy*
- Type *NewPtrType =
- EltTy->getPointerTo(cast<PointerType>(Ptr->getType())->getAddressSpace());
- Value *FirstEltPtr = Builder.CreateBitCast(Ptr, NewPtrType);
Value *UndefVal = UndefValue::get(VecType);
// The result vector
Value *VResult = UndefVal;
+ unsigned VectorWidth = VecType->getNumElements();
+
+ // Shorten the way if the mask is a vector of constants.
+ bool IsConstMask = isa<ConstantVector>(Mask);
+
+ if (IsConstMask) {
+ for (unsigned Idx = 0; Idx < VectorWidth; ++Idx) {
+ if (cast<ConstantVector>(Mask)->getOperand(Idx)->isNullValue())
+ continue;
+ Value *Ptr = Builder.CreateExtractElement(Ptrs, Builder.getInt32(Idx),
+ "Ptr" + Twine(Idx));
+ LoadInst *Load = Builder.CreateAlignedLoad(Ptr, AlignVal,
+ "Load" + Twine(Idx));
+ VResult = Builder.CreateInsertElement(VResult, Load,
+ Builder.getInt32(Idx),
+ "Res" + Twine(Idx));
+ }
+ Value *NewI = Builder.CreateSelect(Mask, VResult, Src0);
+ CI->replaceAllUsesWith(NewI);
+ CI->eraseFromParent();
+ return;
+ }
PHINode *Phi = nullptr;
Value *PrevPhi = UndefVal;
- unsigned VectorWidth = VecType->getNumElements();
for (unsigned Idx = 0; Idx < VectorWidth; ++Idx) {
// Fill the "else" block, created in the previous iteration
//
- // %res.phi.else3 = phi <16 x i32> [ %11, %cond.load1 ], [ %res.phi.else, %else ]
- // %mask_1 = extractelement <16 x i1> %mask, i32 Idx
- // %to_load = icmp eq i1 %mask_1, true
- // br i1 %to_load, label %cond.load, label %else
+ // %Mask1 = extractelement <16 x i1> %Mask, i32 1
+ // %ToLoad1 = icmp eq i1 %Mask1, true
+ // br i1 %ToLoad1, label %cond.load, label %else
//
if (Idx > 0) {
Phi = Builder.CreatePHI(VecType, 2, "res.phi.else");
VResult = Phi;
}
- Value *Predicate = Builder.CreateExtractElement(Mask, Builder.getInt32(Idx));
+ Value *Predicate = Builder.CreateExtractElement(Mask,
+ Builder.getInt32(Idx),
+ "Mask" + Twine(Idx));
Value *Cmp = Builder.CreateICmp(ICmpInst::ICMP_EQ, Predicate,
- ConstantInt::get(Predicate->getType(), 1));
+ ConstantInt::get(Predicate->getType(), 1),
+ "ToLoad" + Twine(Idx));
// Create "cond" block
//
CondBlock = IfBlock->splitBasicBlock(InsertPt, "cond.load");
Builder.SetInsertPoint(InsertPt);
- Value *Gep =
- Builder.CreateInBoundsGEP(EltTy, FirstEltPtr, Builder.getInt32(Idx));
- LoadInst* Load = Builder.CreateLoad(Gep, false);
- VResult = Builder.CreateInsertElement(VResult, Load, Builder.getInt32(Idx));
+ Value *Ptr = Builder.CreateExtractElement(Ptrs, Builder.getInt32(Idx),
+ "Ptr" + Twine(Idx));
+ LoadInst *Load = Builder.CreateAlignedLoad(Ptr, AlignVal,
+ "Load" + Twine(Idx));
+ VResult = Builder.CreateInsertElement(VResult, Load, Builder.getInt32(Idx),
+ "Res" + Twine(Idx));
// Create "else" block, fill it in the next iteration
BasicBlock *NewIfBlock = CondBlock->splitBasicBlock(InsertPt, "else");
CI->eraseFromParent();
}
-// ScalarizeMaskedStore() translates masked store intrinsic, like
-// void @llvm.masked.store(<16 x i32> %src, <16 x i32>* %addr, i32 align,
-// <16 x i1> %mask)
+// Translate a masked scatter intrinsic, like
+// void @llvm.masked.scatter.v16i32(<16 x i32> %Src, <16 x i32*>* %Ptrs, i32 4,
+// <16 x i1> %Mask)
// to a chain of basic blocks, that stores element one-by-one if
-// the appropriate mask bit is set
+// the appropriate mask bit is set.
//
-// %1 = bitcast i8* %addr to i32*
-// %2 = extractelement <16 x i1> %mask, i32 0
-// %3 = icmp eq i1 %2, true
-// br i1 %3, label %cond.store, label %else
+// % Ptrs = getelementptr i32, i32* %ptr, <16 x i64> %ind
+// % Mask0 = extractelement <16 x i1> % Mask, i32 0
+// % ToStore0 = icmp eq i1 % Mask0, true
+// br i1 %ToStore0, label %cond.store, label %else
//
-// cond.store: ; preds = %0
-// %4 = extractelement <16 x i32> %val, i32 0
-// %5 = getelementptr i32* %1, i32 0
-// store i32 %4, i32* %5
-// br label %else
-//
-// else: ; preds = %0, %cond.store
-// %6 = extractelement <16 x i1> %mask, i32 1
-// %7 = icmp eq i1 %6, true
-// br i1 %7, label %cond.store1, label %else2
-//
-// cond.store1: ; preds = %else
-// %8 = extractelement <16 x i32> %val, i32 1
-// %9 = getelementptr i32* %1, i32 1
-// store i32 %8, i32* %9
-// br label %else2
+// cond.store:
+// % Elt0 = extractelement <16 x i32> %Src, i32 0
+// % Ptr0 = extractelement <16 x i32*> %Ptrs, i32 0
+// store i32 %Elt0, i32* % Ptr0, align 4
+// br label %else
+//
+// else:
+// % Mask1 = extractelement <16 x i1> % Mask, i32 1
+// % ToStore1 = icmp eq i1 % Mask1, true
+// br i1 % ToStore1, label %cond.store1, label %else2
+//
+// cond.store1:
+// % Elt1 = extractelement <16 x i32> %Src, i32 1
+// % Ptr1 = extractelement <16 x i32*> %Ptrs, i32 1
+// store i32 % Elt1, i32* % Ptr1, align 4
+// br label %else2
// . . .
-static void ScalarizeMaskedStore(CallInst *CI) {
- Value *Ptr = CI->getArgOperand(1);
+static void ScalarizeMaskedScatter(CallInst *CI) {
Value *Src = CI->getArgOperand(0);
+ Value *Ptrs = CI->getArgOperand(1);
+ Value *Alignment = CI->getArgOperand(2);
Value *Mask = CI->getArgOperand(3);
- VectorType *VecType = dyn_cast<VectorType>(Src->getType());
- Type *EltTy = VecType->getElementType();
-
- assert(VecType && "Unexpected data type in masked store intrinsic");
+ assert(isa<VectorType>(Src->getType()) &&
+ "Unexpected data type in masked scatter intrinsic");
+ assert(isa<VectorType>(Ptrs->getType()) &&
+ isa<PointerType>(Ptrs->getType()->getVectorElementType()) &&
+ "Vector of pointers is expected in masked scatter intrinsic");
IRBuilder<> Builder(CI->getContext());
Instruction *InsertPt = CI;
Builder.SetInsertPoint(InsertPt);
Builder.SetCurrentDebugLocation(CI->getDebugLoc());
- // Bitcast %addr fron i8* to EltTy*
- Type *NewPtrType =
- EltTy->getPointerTo(cast<PointerType>(Ptr->getType())->getAddressSpace());
- Value *FirstEltPtr = Builder.CreateBitCast(Ptr, NewPtrType);
+ unsigned AlignVal = cast<ConstantInt>(Alignment)->getZExtValue();
+ unsigned VectorWidth = Src->getType()->getVectorNumElements();
- unsigned VectorWidth = VecType->getNumElements();
- for (unsigned Idx = 0; Idx < VectorWidth; ++Idx) {
+ // Shorten the way if the mask is a vector of constants.
+ bool IsConstMask = isa<ConstantVector>(Mask);
+ if (IsConstMask) {
+ for (unsigned Idx = 0; Idx < VectorWidth; ++Idx) {
+ if (cast<ConstantVector>(Mask)->getOperand(Idx)->isNullValue())
+ continue;
+ Value *OneElt = Builder.CreateExtractElement(Src, Builder.getInt32(Idx),
+ "Elt" + Twine(Idx));
+ Value *Ptr = Builder.CreateExtractElement(Ptrs, Builder.getInt32(Idx),
+ "Ptr" + Twine(Idx));
+ Builder.CreateAlignedStore(OneElt, Ptr, AlignVal);
+ }
+ CI->eraseFromParent();
+ return;
+ }
+ for (unsigned Idx = 0; Idx < VectorWidth; ++Idx) {
// Fill the "else" block, created in the previous iteration
//
- // %mask_1 = extractelement <16 x i1> %mask, i32 Idx
- // %to_store = icmp eq i1 %mask_1, true
- // br i1 %to_load, label %cond.store, label %else
+ // % Mask1 = extractelement <16 x i1> % Mask, i32 Idx
+ // % ToStore = icmp eq i1 % Mask1, true
+ // br i1 % ToStore, label %cond.store, label %else
//
- Value *Predicate = Builder.CreateExtractElement(Mask, Builder.getInt32(Idx));
- Value *Cmp = Builder.CreateICmp(ICmpInst::ICMP_EQ, Predicate,
- ConstantInt::get(Predicate->getType(), 1));
+ Value *Predicate = Builder.CreateExtractElement(Mask,
+ Builder.getInt32(Idx),
+ "Mask" + Twine(Idx));
+ Value *Cmp =
+ Builder.CreateICmp(ICmpInst::ICMP_EQ, Predicate,
+ ConstantInt::get(Predicate->getType(), 1),
+ "ToStore" + Twine(Idx));
// Create "cond" block
//
- // %OneElt = extractelement <16 x i32> %Src, i32 Idx
- // %EltAddr = getelementptr i32* %1, i32 0
- // %store i32 %OneElt, i32* %EltAddr
+ // % Elt1 = extractelement <16 x i32> %Src, i32 1
+ // % Ptr1 = extractelement <16 x i32*> %Ptrs, i32 1
+ // %store i32 % Elt1, i32* % Ptr1
//
BasicBlock *CondBlock = IfBlock->splitBasicBlock(InsertPt, "cond.store");
Builder.SetInsertPoint(InsertPt);
-
- Value *OneElt = Builder.CreateExtractElement(Src, Builder.getInt32(Idx));
- Value *Gep =
- Builder.CreateInBoundsGEP(EltTy, FirstEltPtr, Builder.getInt32(Idx));
- Builder.CreateStore(OneElt, Gep);
+
+ Value *OneElt = Builder.CreateExtractElement(Src, Builder.getInt32(Idx),
+ "Elt" + Twine(Idx));
+ Value *Ptr = Builder.CreateExtractElement(Ptrs, Builder.getInt32(Idx),
+ "Ptr" + Twine(Idx));
+ Builder.CreateAlignedStore(OneElt, Ptr, AlignVal);
// Create "else" block, fill it in the next iteration
BasicBlock *NewIfBlock = CondBlock->splitBasicBlock(InsertPt, "else");
CI->eraseFromParent();
}
-bool CodeGenPrepare::OptimizeCallInst(CallInst *CI, bool& ModifiedDT) {
+/// If counting leading or trailing zeros is an expensive operation and a zero
+/// input is defined, add a check for zero to avoid calling the intrinsic.
+///
+/// We want to transform:
+/// %z = call i64 @llvm.cttz.i64(i64 %A, i1 false)
+///
+/// into:
+/// entry:
+/// %cmpz = icmp eq i64 %A, 0
+/// br i1 %cmpz, label %cond.end, label %cond.false
+/// cond.false:
+/// %z = call i64 @llvm.cttz.i64(i64 %A, i1 true)
+/// br label %cond.end
+/// cond.end:
+/// %ctz = phi i64 [ 64, %entry ], [ %z, %cond.false ]
+///
+/// If the transform is performed, return true and set ModifiedDT to true.
+static bool despeculateCountZeros(IntrinsicInst *CountZeros,
+ const TargetLowering *TLI,
+ const DataLayout *DL,
+ bool &ModifiedDT) {
+ if (!TLI || !DL)
+ return false;
+
+ // If a zero input is undefined, it doesn't make sense to despeculate that.
+ if (match(CountZeros->getOperand(1), m_One()))
+ return false;
+
+ // If it's cheap to speculate, there's nothing to do.
+ auto IntrinsicID = CountZeros->getIntrinsicID();
+ if ((IntrinsicID == Intrinsic::cttz && TLI->isCheapToSpeculateCttz()) ||
+ (IntrinsicID == Intrinsic::ctlz && TLI->isCheapToSpeculateCtlz()))
+ return false;
+
+ // Only handle legal scalar cases. Anything else requires too much work.
+ Type *Ty = CountZeros->getType();
+ unsigned SizeInBits = Ty->getPrimitiveSizeInBits();
+ if (Ty->isVectorTy() || SizeInBits > DL->getLargestLegalIntTypeSize())
+ return false;
+
+ // The intrinsic will be sunk behind a compare against zero and branch.
+ BasicBlock *StartBlock = CountZeros->getParent();
+ BasicBlock *CallBlock = StartBlock->splitBasicBlock(CountZeros, "cond.false");
+
+ // Create another block after the count zero intrinsic. A PHI will be added
+ // in this block to select the result of the intrinsic or the bit-width
+ // constant if the input to the intrinsic is zero.
+ BasicBlock::iterator SplitPt = ++(BasicBlock::iterator(CountZeros));
+ BasicBlock *EndBlock = CallBlock->splitBasicBlock(SplitPt, "cond.end");
+
+ // Set up a builder to create a compare, conditional branch, and PHI.
+ IRBuilder<> Builder(CountZeros->getContext());
+ Builder.SetInsertPoint(StartBlock->getTerminator());
+ Builder.SetCurrentDebugLocation(CountZeros->getDebugLoc());
+
+ // Replace the unconditional branch that was created by the first split with
+ // a compare against zero and a conditional branch.
+ Value *Zero = Constant::getNullValue(Ty);
+ Value *Cmp = Builder.CreateICmpEQ(CountZeros->getOperand(0), Zero, "cmpz");
+ Builder.CreateCondBr(Cmp, EndBlock, CallBlock);
+ StartBlock->getTerminator()->eraseFromParent();
+
+ // Create a PHI in the end block to select either the output of the intrinsic
+ // or the bit width of the operand.
+ Builder.SetInsertPoint(&EndBlock->front());
+ PHINode *PN = Builder.CreatePHI(Ty, 2, "ctz");
+ CountZeros->replaceAllUsesWith(PN);
+ Value *BitWidth = Builder.getInt(APInt(SizeInBits, SizeInBits));
+ PN->addIncoming(BitWidth, StartBlock);
+ PN->addIncoming(CountZeros, CallBlock);
+
+ // We are explicitly handling the zero case, so we can set the intrinsic's
+ // undefined zero argument to 'true'. This will also prevent reprocessing the
+ // intrinsic; we only despeculate when a zero input is defined.
+ CountZeros->setArgOperand(1, Builder.getTrue());
+ ModifiedDT = true;
+ return true;
+}
+
+bool CodeGenPrepare::optimizeCallInst(CallInst *CI, bool& ModifiedDT) {
BasicBlock *BB = CI->getParent();
// Lower inline assembly if we can.
return true;
}
// Sink address computing for memory operands into the block.
- if (OptimizeInlineAsmInst(CI))
+ if (optimizeInlineAsmInst(CI))
return true;
}
// over-aligning global variables that have an explicit section is
// forbidden.
GlobalVariable *GV;
- if ((GV = dyn_cast<GlobalVariable>(Val)) && GV->hasUniqueInitializer() &&
- !GV->hasSection() && GV->getAlignment() < PrefAlign &&
+ if ((GV = dyn_cast<GlobalVariable>(Val)) && GV->canIncreaseAlignment() &&
+ GV->getAlignment() < PrefAlign &&
DL->getTypeAllocSize(GV->getType()->getElementType()) >=
MinSize + Offset2)
GV->setAlignment(PrefAlign);
// Substituting this can cause recursive simplifications, which can
// invalidate our iterator. Use a WeakVH to hold onto it in case this
// happens.
- WeakVH IterHandle(CurInstIterator);
+ WeakVH IterHandle(&*CurInstIterator);
replaceAndRecursivelySimplify(CI, RetVal,
TLInfo, nullptr);
// If the iterator instruction was recursively deleted, start over at the
// start of the block.
- if (IterHandle != CurInstIterator) {
+ if (IterHandle != CurInstIterator.getNodePtrUnchecked()) {
CurInstIterator = BB->begin();
SunkAddrs.clear();
}
}
case Intrinsic::masked_load: {
// Scalarize unsupported vector masked load
- if (!TTI->isLegalMaskedLoad(CI->getType(), 1)) {
+ if (!TTI->isLegalMaskedLoad(CI->getType())) {
ScalarizeMaskedLoad(CI);
ModifiedDT = true;
return true;
return false;
}
case Intrinsic::masked_store: {
- if (!TTI->isLegalMaskedStore(CI->getArgOperand(0)->getType(), 1)) {
+ if (!TTI->isLegalMaskedStore(CI->getArgOperand(0)->getType())) {
ScalarizeMaskedStore(CI);
ModifiedDT = true;
return true;
}
return false;
}
+ case Intrinsic::masked_gather: {
+ if (!TTI->isLegalMaskedGather(CI->getType())) {
+ ScalarizeMaskedGather(CI);
+ ModifiedDT = true;
+ return true;
+ }
+ return false;
+ }
+ case Intrinsic::masked_scatter: {
+ if (!TTI->isLegalMaskedScatter(CI->getArgOperand(0)->getType())) {
+ ScalarizeMaskedScatter(CI);
+ ModifiedDT = true;
+ return true;
+ }
+ return false;
+ }
case Intrinsic::aarch64_stlxr:
case Intrinsic::aarch64_stxr: {
ZExtInst *ExtVal = dyn_cast<ZExtInst>(CI->getArgOperand(0));
II->replaceAllUsesWith(II->getArgOperand(0));
II->eraseFromParent();
return true;
+
+ case Intrinsic::cttz:
+ case Intrinsic::ctlz:
+ // If counting zeros is expensive, try to avoid it.
+ return despeculateCountZeros(II, TLI, DL, ModifiedDT);
}
if (TLI) {
Type *AccessTy;
if (TLI->GetAddrModeArguments(II, PtrOps, AccessTy, AddrSpace))
while (!PtrOps.empty())
- if (OptimizeMemoryInst(II, PtrOps.pop_back_val(), AccessTy, AddrSpace))
+ if (optimizeMemoryInst(II, PtrOps.pop_back_val(), AccessTy, AddrSpace))
return true;
}
}
return false;
}
-/// DupRetToEnableTailCallOpts - Look for opportunities to duplicate return
-/// instructions to the predecessor to enable tail call optimizations. The
-/// case it is currently looking for is:
+/// Look for opportunities to duplicate return instructions to the predecessor
+/// to enable tail call optimizations. The case it is currently looking for is:
/// @code
/// bb0:
/// %tmp0 = tail call i32 @f0()
/// %tmp2 = tail call i32 @f2()
/// ret i32 %tmp2
/// @endcode
-bool CodeGenPrepare::DupRetToEnableTailCallOpts(BasicBlock *BB) {
+bool CodeGenPrepare::dupRetToEnableTailCallOpts(BasicBlock *BB) {
if (!TLI)
return false;
namespace {
-/// ExtAddrMode - This is an extended version of TargetLowering::AddrMode
+/// This is an extended version of TargetLowering::AddrMode
/// which holds actual Value*'s for register values.
struct ExtAddrMode : public TargetLowering::AddrMode {
Value *BaseReg;
public:
/// \brief Record the position of \p Inst.
InsertionHandler(Instruction *Inst) {
- BasicBlock::iterator It = Inst;
+ BasicBlock::iterator It = Inst->getIterator();
HasPrevInstruction = (It != (Inst->getParent()->begin()));
if (HasPrevInstruction)
- Point.PrevInst = --It;
+ Point.PrevInst = &*--It;
else
Point.BB = Inst->getParent();
}
Inst->removeFromParent();
Inst->insertAfter(Point.PrevInst);
} else {
- Instruction *Position = Point.BB->getFirstInsertionPt();
+ Instruction *Position = &*Point.BB->getFirstInsertionPt();
if (Inst->getParent())
Inst->moveBefore(Position);
else
Value *Val = Inst->getOperand(It);
OriginalValues.push_back(Val);
// Set a dummy one.
- // We could use OperandSetter here, but that would implied an overhead
+ // We could use OperandSetter here, but that would imply an overhead
// that we are not willing to pay.
Inst->setOperand(It, UndefValue::get(Val->getType()));
}
unsigned AddrSpace;
Instruction *MemoryInst;
- /// AddrMode - This is the addressing mode that we're building up. This is
+ /// This is the addressing mode that we're building up. This is
/// part of the return value of this addressing mode matching stuff.
ExtAddrMode &AddrMode;
/// The ongoing transaction where every action should be registered.
TypePromotionTransaction &TPT;
- /// IgnoreProfitability - This is set to true when we should not do
- /// profitability checks. When true, IsProfitableToFoldIntoAddressingMode
- /// always returns true.
+ /// This is set to true when we should not do profitability checks.
+ /// When true, IsProfitableToFoldIntoAddressingMode always returns true.
bool IgnoreProfitability;
AddressingModeMatcher(SmallVectorImpl<Instruction *> &AMI,
}
public:
- /// Match - Find the maximal addressing mode that a load/store of V can fold,
+ /// Find the maximal addressing mode that a load/store of V can fold,
/// give an access type of AccessTy. This returns a list of involved
/// instructions in AddrModeInsts.
/// \p InsertedInsts The instructions inserted by other CodeGenPrepare
bool Success = AddressingModeMatcher(AddrModeInsts, TM, AccessTy, AS,
MemoryInst, Result, InsertedInsts,
- PromotedInsts, TPT).MatchAddr(V, 0);
+ PromotedInsts, TPT).matchAddr(V, 0);
(void)Success; assert(Success && "Couldn't select *anything*?");
return Result;
}
private:
- bool MatchScaledValue(Value *ScaleReg, int64_t Scale, unsigned Depth);
- bool MatchAddr(Value *V, unsigned Depth);
- bool MatchOperationAddr(User *Operation, unsigned Opcode, unsigned Depth,
+ bool matchScaledValue(Value *ScaleReg, int64_t Scale, unsigned Depth);
+ bool matchAddr(Value *V, unsigned Depth);
+ bool matchOperationAddr(User *Operation, unsigned Opcode, unsigned Depth,
bool *MovedAway = nullptr);
- bool IsProfitableToFoldIntoAddressingMode(Instruction *I,
+ bool isProfitableToFoldIntoAddressingMode(Instruction *I,
ExtAddrMode &AMBefore,
ExtAddrMode &AMAfter);
- bool ValueAlreadyLiveAtInst(Value *Val, Value *KnownLive1, Value *KnownLive2);
- bool IsPromotionProfitable(unsigned NewCost, unsigned OldCost,
+ bool valueAlreadyLiveAtInst(Value *Val, Value *KnownLive1, Value *KnownLive2);
+ bool isPromotionProfitable(unsigned NewCost, unsigned OldCost,
Value *PromotedOperand) const;
};
-/// MatchScaledValue - Try adding ScaleReg*Scale to the current addressing mode.
+/// Try adding ScaleReg*Scale to the current addressing mode.
/// Return true and update AddrMode if this addr mode is legal for the target,
/// false if not.
-bool AddressingModeMatcher::MatchScaledValue(Value *ScaleReg, int64_t Scale,
+bool AddressingModeMatcher::matchScaledValue(Value *ScaleReg, int64_t Scale,
unsigned Depth) {
// If Scale is 1, then this is the same as adding ScaleReg to the addressing
// mode. Just process that directly.
if (Scale == 1)
- return MatchAddr(ScaleReg, Depth);
+ return matchAddr(ScaleReg, Depth);
// If the scale is 0, it takes nothing to add this.
if (Scale == 0)
return true;
}
-/// MightBeFoldableInst - This is a little filter, which returns true if an
-/// addressing computation involving I might be folded into a load/store
-/// accessing it. This doesn't need to be perfect, but needs to accept at least
+/// This is a little filter, which returns true if an addressing computation
+/// involving I might be folded into a load/store accessing it.
+/// This doesn't need to be perfect, but needs to accept at least
/// the set of instructions that MatchOperationAddr can.
static bool MightBeFoldableInst(Instruction *I) {
switch (I->getOpcode()) {
/// \brief Utility function to determine if \p OpIdx should be promoted when
/// promoting \p Inst.
static bool shouldExtOperand(const Instruction *Inst, int OpIdx) {
- if (isa<SelectInst>(Inst) && OpIdx == 0)
- return false;
- return true;
+ return !(isa<SelectInst>(Inst) && OpIdx == 0);
}
/// \brief Utility function to promote the operand of \p Ext when this
Value *OpndVal = Inst->getOperand(0);
// Check if we can use this operand in the extension.
- // If the type is larger than the result type of the extension,
- // we cannot.
+ // If the type is larger than the result type of the extension, we cannot.
if (!OpndVal->getType()->isIntegerTy() ||
OpndVal->getType()->getIntegerBitWidth() >
ConsideredExtType->getIntegerBitWidth())
else
return false;
- // #2 check that the truncate just drop extended bits.
- if (Inst->getType()->getIntegerBitWidth() >= OpndType->getIntegerBitWidth())
- return true;
-
- return false;
+ // #2 check that the truncate just drops extended bits.
+ return Inst->getType()->getIntegerBitWidth() >=
+ OpndType->getIntegerBitWidth();
}
TypePromotionHelper::Action TypePromotionHelper::getAction(
}
TPT.replaceAllUsesWith(ExtOpnd, Trunc);
- // Restore the operand of Ext (which has been replace by the previous call
+ // Restore the operand of Ext (which has been replaced by the previous call
// to replaceAllUsesWith) to avoid creating a cycle trunc <-> sext.
TPT.setOperand(Ext, 0, ExtOpnd);
}
return ExtOpnd;
}
-/// IsPromotionProfitable - Check whether or not promoting an instruction
-/// to a wider type was profitable.
+/// Check whether or not promoting an instruction to a wider type is profitable.
/// \p NewCost gives the cost of extension instructions created by the
/// promotion.
/// \p OldCost gives the cost of extension instructions before the promotion
/// matched in the addressing mode the promotion.
/// \p PromotedOperand is the value that has been promoted.
/// \return True if the promotion is profitable, false otherwise.
-bool AddressingModeMatcher::IsPromotionProfitable(
+bool AddressingModeMatcher::isPromotionProfitable(
unsigned NewCost, unsigned OldCost, Value *PromotedOperand) const {
DEBUG(dbgs() << "OldCost: " << OldCost << "\tNewCost: " << NewCost << '\n');
// The cost of the new extensions is greater than the cost of the
return isPromotedInstructionLegal(TLI, DL, PromotedOperand);
}
-/// MatchOperationAddr - Given an instruction or constant expr, see if we can
-/// fold the operation into the addressing mode. If so, update the addressing
-/// mode and return true, otherwise return false without modifying AddrMode.
+/// Given an instruction or constant expr, see if we can fold the operation
+/// into the addressing mode. If so, update the addressing mode and return
+/// true, otherwise return false without modifying AddrMode.
/// If \p MovedAway is not NULL, it contains the information of whether or
/// not AddrInst has to be folded into the addressing mode on success.
/// If \p MovedAway == true, \p AddrInst will not be part of the addressing
/// This state can happen when AddrInst is a sext, since it may be moved away.
/// Therefore, AddrInst may not be valid when MovedAway is true and it must
/// not be referenced anymore.
-bool AddressingModeMatcher::MatchOperationAddr(User *AddrInst, unsigned Opcode,
+bool AddressingModeMatcher::matchOperationAddr(User *AddrInst, unsigned Opcode,
unsigned Depth,
bool *MovedAway) {
// Avoid exponential behavior on extremely deep expression trees.
switch (Opcode) {
case Instruction::PtrToInt:
// PtrToInt is always a noop, as we know that the int type is pointer sized.
- return MatchAddr(AddrInst->getOperand(0), Depth);
+ return matchAddr(AddrInst->getOperand(0), Depth);
case Instruction::IntToPtr: {
auto AS = AddrInst->getType()->getPointerAddressSpace();
auto PtrTy = MVT::getIntegerVT(DL.getPointerSizeInBits(AS));
// This inttoptr is a no-op if the integer type is pointer sized.
if (TLI.getValueType(DL, AddrInst->getOperand(0)->getType()) == PtrTy)
- return MatchAddr(AddrInst->getOperand(0), Depth);
+ return matchAddr(AddrInst->getOperand(0), Depth);
return false;
}
case Instruction::BitCast:
// and we don't want to mess around with them. Assume it knows what it
// is doing.
AddrInst->getOperand(0)->getType() != AddrInst->getType())
- return MatchAddr(AddrInst->getOperand(0), Depth);
+ return matchAddr(AddrInst->getOperand(0), Depth);
return false;
case Instruction::AddrSpaceCast: {
unsigned SrcAS
= AddrInst->getOperand(0)->getType()->getPointerAddressSpace();
unsigned DestAS = AddrInst->getType()->getPointerAddressSpace();
if (TLI.isNoopAddrSpaceCast(SrcAS, DestAS))
- return MatchAddr(AddrInst->getOperand(0), Depth);
+ return matchAddr(AddrInst->getOperand(0), Depth);
return false;
}
case Instruction::Add: {
TypePromotionTransaction::ConstRestorationPt LastKnownGood =
TPT.getRestorationPoint();
- if (MatchAddr(AddrInst->getOperand(1), Depth+1) &&
- MatchAddr(AddrInst->getOperand(0), Depth+1))
+ if (matchAddr(AddrInst->getOperand(1), Depth+1) &&
+ matchAddr(AddrInst->getOperand(0), Depth+1))
return true;
// Restore the old addr mode info.
TPT.rollback(LastKnownGood);
// Otherwise this was over-aggressive. Try merging in the LHS then the RHS.
- if (MatchAddr(AddrInst->getOperand(0), Depth+1) &&
- MatchAddr(AddrInst->getOperand(1), Depth+1))
+ if (matchAddr(AddrInst->getOperand(0), Depth+1) &&
+ matchAddr(AddrInst->getOperand(1), Depth+1))
return true;
// Otherwise we definitely can't merge the ADD in.
if (Opcode == Instruction::Shl)
Scale = 1LL << Scale;
- return MatchScaledValue(AddrInst->getOperand(0), Scale, Depth);
+ return matchScaledValue(AddrInst->getOperand(0), Scale, Depth);
}
case Instruction::GetElementPtr: {
// Scan the GEP. We check it if it contains constant offsets and at most
if (ConstantOffset == 0 ||
TLI.isLegalAddressingMode(DL, AddrMode, AccessTy, AddrSpace)) {
// Check to see if we can fold the base pointer in too.
- if (MatchAddr(AddrInst->getOperand(0), Depth+1))
+ if (matchAddr(AddrInst->getOperand(0), Depth+1))
return true;
}
AddrMode.BaseOffs -= ConstantOffset;
AddrMode.BaseOffs += ConstantOffset;
// Match the base operand of the GEP.
- if (!MatchAddr(AddrInst->getOperand(0), Depth+1)) {
+ if (!matchAddr(AddrInst->getOperand(0), Depth+1)) {
// If it couldn't be matched, just stuff the value in a register.
if (AddrMode.HasBaseReg) {
AddrMode = BackupAddrMode;
}
// Match the remaining variable portion of the GEP.
- if (!MatchScaledValue(AddrInst->getOperand(VariableOperand), VariableScale,
+ if (!matchScaledValue(AddrInst->getOperand(VariableOperand), VariableScale,
Depth)) {
// If it couldn't be matched, try stuffing the base into a register
// instead of matching it, and retrying the match of the scale.
AddrMode.HasBaseReg = true;
AddrMode.BaseReg = AddrInst->getOperand(0);
AddrMode.BaseOffs += ConstantOffset;
- if (!MatchScaledValue(AddrInst->getOperand(VariableOperand),
+ if (!matchScaledValue(AddrInst->getOperand(VariableOperand),
VariableScale, Depth)) {
// If even that didn't work, bail.
AddrMode = BackupAddrMode;
ExtAddrMode BackupAddrMode = AddrMode;
unsigned OldSize = AddrModeInsts.size();
- if (!MatchAddr(PromotedOperand, Depth) ||
- // The total of the new cost is equals to the cost of the created
+ if (!matchAddr(PromotedOperand, Depth) ||
+ // The total of the new cost is equal to the cost of the created
// instructions.
- // The total of the old cost is equals to the cost of the extension plus
+ // The total of the old cost is equal to the cost of the extension plus
// what we have saved in the addressing mode.
- !IsPromotionProfitable(CreatedInstsCost,
+ !isPromotionProfitable(CreatedInstsCost,
ExtCost + (AddrModeInsts.size() - OldSize),
PromotedOperand)) {
AddrMode = BackupAddrMode;
return false;
}
-/// MatchAddr - If we can, try to add the value of 'Addr' into the current
-/// addressing mode. If Addr can't be added to AddrMode this returns false and
-/// leaves AddrMode unmodified. This assumes that Addr is either a pointer type
-/// or intptr_t for the target.
+/// If we can, try to add the value of 'Addr' into the current addressing mode.
+/// If Addr can't be added to AddrMode this returns false and leaves AddrMode
+/// unmodified. This assumes that Addr is either a pointer type or intptr_t
+/// for the target.
///
-bool AddressingModeMatcher::MatchAddr(Value *Addr, unsigned Depth) {
+bool AddressingModeMatcher::matchAddr(Value *Addr, unsigned Depth) {
// Start a transaction at this point that we will rollback if the matching
// fails.
TypePromotionTransaction::ConstRestorationPt LastKnownGood =
// Check to see if it is possible to fold this operation.
bool MovedAway = false;
- if (MatchOperationAddr(I, I->getOpcode(), Depth, &MovedAway)) {
- // This instruction may have been move away. If so, there is nothing
+ if (matchOperationAddr(I, I->getOpcode(), Depth, &MovedAway)) {
+ // This instruction may have been moved away. If so, there is nothing
// to check here.
if (MovedAway)
return true;
// *profitable* to do so. We use a simple cost model to avoid increasing
// register pressure too much.
if (I->hasOneUse() ||
- IsProfitableToFoldIntoAddressingMode(I, BackupAddrMode, AddrMode)) {
+ isProfitableToFoldIntoAddressingMode(I, BackupAddrMode, AddrMode)) {
AddrModeInsts.push_back(I);
return true;
}
TPT.rollback(LastKnownGood);
}
} else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(Addr)) {
- if (MatchOperationAddr(CE, CE->getOpcode(), Depth))
+ if (matchOperationAddr(CE, CE->getOpcode(), Depth))
return true;
TPT.rollback(LastKnownGood);
} else if (isa<ConstantPointerNull>(Addr)) {
return false;
}
-/// IsOperandAMemoryOperand - Check to see if all uses of OpVal by the specified
-/// inline asm call are due to memory operands. If so, return true, otherwise
-/// return false.
+/// Check to see if all uses of OpVal by the specified inline asm call are due
+/// to memory operands. If so, return true, otherwise return false.
static bool IsOperandAMemoryOperand(CallInst *CI, InlineAsm *IA, Value *OpVal,
const TargetMachine &TM) {
const Function *F = CI->getParent()->getParent();
return true;
}
-/// FindAllMemoryUses - Recursively walk all the uses of I until we find a
-/// memory use. If we find an obviously non-foldable instruction, return true.
+/// Recursively walk all the uses of I until we find a memory use.
+/// If we find an obviously non-foldable instruction, return true.
/// Add the ultimately found memory instructions to MemoryUses.
static bool FindAllMemoryUses(
Instruction *I,
return false;
}
-/// ValueAlreadyLiveAtInst - Retrn true if Val is already known to be live at
-/// the use site that we're folding it into. If so, there is no cost to
-/// include it in the addressing mode. KnownLive1 and KnownLive2 are two values
-/// that we know are live at the instruction already.
-bool AddressingModeMatcher::ValueAlreadyLiveAtInst(Value *Val,Value *KnownLive1,
+/// Return true if Val is already known to be live at the use site that we're
+/// folding it into. If so, there is no cost to include it in the addressing
+/// mode. KnownLive1 and KnownLive2 are two values that we know are live at the
+/// instruction already.
+bool AddressingModeMatcher::valueAlreadyLiveAtInst(Value *Val,Value *KnownLive1,
Value *KnownLive2) {
// If Val is either of the known-live values, we know it is live!
if (Val == nullptr || Val == KnownLive1 || Val == KnownLive2)
return Val->isUsedInBasicBlock(MemoryInst->getParent());
}
-/// IsProfitableToFoldIntoAddressingMode - It is possible for the addressing
-/// mode of the machine to fold the specified instruction into a load or store
-/// that ultimately uses it. However, the specified instruction has multiple
-/// uses. Given this, it may actually increase register pressure to fold it
-/// into the load. For example, consider this code:
+/// It is possible for the addressing mode of the machine to fold the specified
+/// instruction into a load or store that ultimately uses it.
+/// However, the specified instruction has multiple uses.
+/// Given this, it may actually increase register pressure to fold it
+/// into the load. For example, consider this code:
///
/// X = ...
/// Y = X+1
/// X was live across 'load Z' for other reasons, we actually *would* want to
/// fold the addressing mode in the Z case. This would make Y die earlier.
bool AddressingModeMatcher::
-IsProfitableToFoldIntoAddressingMode(Instruction *I, ExtAddrMode &AMBefore,
+isProfitableToFoldIntoAddressingMode(Instruction *I, ExtAddrMode &AMBefore,
ExtAddrMode &AMAfter) {
if (IgnoreProfitability) return true;
// If the BaseReg or ScaledReg was referenced by the previous addrmode, their
// lifetime wasn't extended by adding this instruction.
- if (ValueAlreadyLiveAtInst(BaseReg, AMBefore.BaseReg, AMBefore.ScaledReg))
+ if (valueAlreadyLiveAtInst(BaseReg, AMBefore.BaseReg, AMBefore.ScaledReg))
BaseReg = nullptr;
- if (ValueAlreadyLiveAtInst(ScaledReg, AMBefore.BaseReg, AMBefore.ScaledReg))
+ if (valueAlreadyLiveAtInst(ScaledReg, AMBefore.BaseReg, AMBefore.ScaledReg))
ScaledReg = nullptr;
// If folding this instruction (and it's subexprs) didn't extend any live
MemoryInst, Result, InsertedInsts,
PromotedInsts, TPT);
Matcher.IgnoreProfitability = true;
- bool Success = Matcher.MatchAddr(Address, 0);
+ bool Success = Matcher.matchAddr(Address, 0);
(void)Success; assert(Success && "Couldn't select *anything*?");
// The match was to check the profitability, the changes made are not
} // end anonymous namespace
-/// IsNonLocalValue - Return true if the specified values are defined in a
+/// Return true if the specified values are defined in a
/// different basic block than BB.
static bool IsNonLocalValue(Value *V, BasicBlock *BB) {
if (Instruction *I = dyn_cast<Instruction>(V))
return false;
}
-/// OptimizeMemoryInst - Load and Store Instructions often have
-/// addressing modes that can do significant amounts of computation. As such,
-/// instruction selection will try to get the load or store to do as much
-/// computation as possible for the program. The problem is that isel can only
-/// see within a single block. As such, we sink as much legal addressing mode
-/// stuff into the block as possible.
+/// Load and Store Instructions often have addressing modes that can do
+/// significant amounts of computation. As such, instruction selection will try
+/// to get the load or store to do as much computation as possible for the
+/// program. The problem is that isel can only see within a single block. As
+/// such, we sink as much legal addressing mode work into the block as possible.
///
/// This method is used to optimize both load/store and inline asms with memory
/// operands.
-bool CodeGenPrepare::OptimizeMemoryInst(Instruction *MemoryInst, Value *Addr,
+bool CodeGenPrepare::optimizeMemoryInst(Instruction *MemoryInst, Value *Addr,
Type *AccessTy, unsigned AddrSpace) {
Value *Repl = Addr;
if (Repl->use_empty()) {
// This can cause recursive deletion, which can invalidate our iterator.
// Use a WeakVH to hold onto it in case this happens.
- WeakVH IterHandle(CurInstIterator);
+ WeakVH IterHandle(&*CurInstIterator);
BasicBlock *BB = CurInstIterator->getParent();
RecursivelyDeleteTriviallyDeadInstructions(Repl, TLInfo);
- if (IterHandle != CurInstIterator) {
+ if (IterHandle != CurInstIterator.getNodePtrUnchecked()) {
// If the iterator instruction was recursively deleted, start over at the
// start of the block.
CurInstIterator = BB->begin();
return true;
}
-/// OptimizeInlineAsmInst - If there are any memory operands, use
-/// OptimizeMemoryInst to sink their address computing into the block when
-/// possible / profitable.
-bool CodeGenPrepare::OptimizeInlineAsmInst(CallInst *CS) {
+/// If there are any memory operands, use OptimizeMemoryInst to sink their
+/// address computing into the block when possible / profitable.
+bool CodeGenPrepare::optimizeInlineAsmInst(CallInst *CS) {
bool MadeChange = false;
const TargetRegisterInfo *TRI =
if (OpInfo.ConstraintType == TargetLowering::C_Memory &&
OpInfo.isIndirect) {
Value *OpVal = CS->getArgOperand(ArgNo++);
- MadeChange |= OptimizeMemoryInst(CS, OpVal, OpVal->getType(), ~0u);
+ MadeChange |= optimizeMemoryInst(CS, OpVal, OpVal->getType(), ~0u);
} else if (OpInfo.Type == InlineAsm::isInput)
ArgNo++;
}
/// %add = add nuw i64 %zext, 4
/// \encode
/// Thanks to the promotion, we can match zext(load i32*) to i64.
-bool CodeGenPrepare::ExtLdPromotion(TypePromotionTransaction &TPT,
+bool CodeGenPrepare::extLdPromotion(TypePromotionTransaction &TPT,
LoadInst *&LI, Instruction *&Inst,
const SmallVectorImpl<Instruction *> &Exts,
unsigned CreatedInstsCost = 0) {
}
// The promotion is profitable.
// Check if it exposes an ext(load).
- (void)ExtLdPromotion(TPT, LI, Inst, NewExts, TotalCreatedInstsCost);
+ (void)extLdPromotion(TPT, LI, Inst, NewExts, TotalCreatedInstsCost);
if (LI && (StressExtLdPromotion || NewCreatedInstsCost <= ExtCost ||
// If we have created a new extension, i.e., now we have two
// extensions. We must make sure one of them is merged with
return false;
}
-/// MoveExtToFormExtLoad - Move a zext or sext fed by a load into the same
-/// basic block as the load, unless conditions are unfavorable. This allows
-/// SelectionDAG to fold the extend into the load.
+/// Move a zext or sext fed by a load into the same basic block as the load,
+/// unless conditions are unfavorable. This allows SelectionDAG to fold the
+/// extend into the load.
/// \p I[in/out] the extension may be modified during the process if some
/// promotions apply.
///
-bool CodeGenPrepare::MoveExtToFormExtLoad(Instruction *&I) {
+bool CodeGenPrepare::moveExtToFormExtLoad(Instruction *&I) {
// Try to promote a chain of computation if it allows to form
// an extended load.
TypePromotionTransaction TPT;
// Look for a load being extended.
LoadInst *LI = nullptr;
Instruction *OldExt = I;
- bool HasPromoted = ExtLdPromotion(TPT, LI, I, Exts);
+ bool HasPromoted = extLdPromotion(TPT, LI, I, Exts);
if (!LI || !I) {
assert(!HasPromoted && !LI && "If we did not match any load instruction "
"the code must remain the same");
return true;
}
-bool CodeGenPrepare::OptimizeExtUses(Instruction *I) {
+bool CodeGenPrepare::optimizeExtUses(Instruction *I) {
BasicBlock *DefBB = I->getParent();
// If the result of a {s|z}ext and its source are both live out, rewrite all
if (!InsertedTrunc) {
BasicBlock::iterator InsertPt = UserBB->getFirstInsertionPt();
- InsertedTrunc = new TruncInst(I, Src->getType(), "", InsertPt);
+ assert(InsertPt != UserBB->end());
+ InsertedTrunc = new TruncInst(I, Src->getType(), "", &*InsertPt);
InsertedInsts.insert(InsertedTrunc);
}
return MadeChange;
}
-/// isFormingBranchFromSelectProfitable - Returns true if a SelectInst should be
-/// turned into an explicit branch.
-static bool isFormingBranchFromSelectProfitable(SelectInst *SI) {
+// Find loads whose uses only use some of the loaded value's bits. Add an "and"
+// just after the load if the target can fold this into one extload instruction,
+// with the hope of eliminating some of the other later "and" instructions using
+// the loaded value. "and"s that are made trivially redundant by the insertion
+// of the new "and" are removed by this function, while others (e.g. those whose
+// path from the load goes through a phi) are left for isel to potentially
+// remove.
+//
+// For example:
+//
+// b0:
+// x = load i32
+// ...
+// b1:
+// y = and x, 0xff
+// z = use y
+//
+// becomes:
+//
+// b0:
+// x = load i32
+// x' = and x, 0xff
+// ...
+// b1:
+// z = use x'
+//
+// whereas:
+//
+// b0:
+// x1 = load i32
+// ...
+// b1:
+// x2 = load i32
+// ...
+// b2:
+// x = phi x1, x2
+// y = and x, 0xff
+//
+// becomes (after a call to optimizeLoadExt for each load):
+//
+// b0:
+// x1 = load i32
+// x1' = and x1, 0xff
+// ...
+// b1:
+// x2 = load i32
+// x2' = and x2, 0xff
+// ...
+// b2:
+// x = phi x1', x2'
+// y = and x, 0xff
+//
+
+bool CodeGenPrepare::optimizeLoadExt(LoadInst *Load) {
+
+ if (!Load->isSimple() ||
+ !(Load->getType()->isIntegerTy() || Load->getType()->isPointerTy()))
+ return false;
+
+ // Skip loads we've already transformed or have no reason to transform.
+ if (Load->hasOneUse()) {
+ User *LoadUser = *Load->user_begin();
+ if (cast<Instruction>(LoadUser)->getParent() == Load->getParent() &&
+ !dyn_cast<PHINode>(LoadUser))
+ return false;
+ }
+
+ // Look at all uses of Load, looking through phis, to determine how many bits
+ // of the loaded value are needed.
+ SmallVector<Instruction *, 8> WorkList;
+ SmallPtrSet<Instruction *, 16> Visited;
+ SmallVector<Instruction *, 8> AndsToMaybeRemove;
+ for (auto *U : Load->users())
+ WorkList.push_back(cast<Instruction>(U));
+
+ EVT LoadResultVT = TLI->getValueType(*DL, Load->getType());
+ unsigned BitWidth = LoadResultVT.getSizeInBits();
+ APInt DemandBits(BitWidth, 0);
+ APInt WidestAndBits(BitWidth, 0);
+
+ while (!WorkList.empty()) {
+ Instruction *I = WorkList.back();
+ WorkList.pop_back();
+
+ // Break use-def graph loops.
+ if (!Visited.insert(I).second)
+ continue;
+
+ // For a PHI node, push all of its users.
+ if (auto *Phi = dyn_cast<PHINode>(I)) {
+ for (auto *U : Phi->users())
+ WorkList.push_back(cast<Instruction>(U));
+ continue;
+ }
+
+ switch (I->getOpcode()) {
+ case llvm::Instruction::And: {
+ auto *AndC = dyn_cast<ConstantInt>(I->getOperand(1));
+ if (!AndC)
+ return false;
+ APInt AndBits = AndC->getValue();
+ DemandBits |= AndBits;
+ // Keep track of the widest and mask we see.
+ if (AndBits.ugt(WidestAndBits))
+ WidestAndBits = AndBits;
+ if (AndBits == WidestAndBits && I->getOperand(0) == Load)
+ AndsToMaybeRemove.push_back(I);
+ break;
+ }
+
+ case llvm::Instruction::Shl: {
+ auto *ShlC = dyn_cast<ConstantInt>(I->getOperand(1));
+ if (!ShlC)
+ return false;
+ uint64_t ShiftAmt = ShlC->getLimitedValue(BitWidth - 1);
+ auto ShlDemandBits = APInt::getAllOnesValue(BitWidth).lshr(ShiftAmt);
+ DemandBits |= ShlDemandBits;
+ break;
+ }
+
+ case llvm::Instruction::Trunc: {
+ EVT TruncVT = TLI->getValueType(*DL, I->getType());
+ unsigned TruncBitWidth = TruncVT.getSizeInBits();
+ auto TruncBits = APInt::getAllOnesValue(TruncBitWidth).zext(BitWidth);
+ DemandBits |= TruncBits;
+ break;
+ }
+
+ default:
+ return false;
+ }
+ }
+
+ uint32_t ActiveBits = DemandBits.getActiveBits();
+ // Avoid hoisting (and (load x) 1) since it is unlikely to be folded by the
+ // target even if isLoadExtLegal says an i1 EXTLOAD is valid. For example,
+ // for the AArch64 target isLoadExtLegal(ZEXTLOAD, i32, i1) returns true, but
+ // (and (load x) 1) is not matched as a single instruction, rather as a LDR
+ // followed by an AND.
+ // TODO: Look into removing this restriction by fixing backends to either
+ // return false for isLoadExtLegal for i1 or have them select this pattern to
+ // a single instruction.
+ //
+ // Also avoid hoisting if we didn't see any ands with the exact DemandBits
+ // mask, since these are the only ands that will be removed by isel.
+ if (ActiveBits <= 1 || !APIntOps::isMask(ActiveBits, DemandBits) ||
+ WidestAndBits != DemandBits)
+ return false;
+
+ LLVMContext &Ctx = Load->getType()->getContext();
+ Type *TruncTy = Type::getIntNTy(Ctx, ActiveBits);
+ EVT TruncVT = TLI->getValueType(*DL, TruncTy);
+
+ // Reject cases that won't be matched as extloads.
+ if (!LoadResultVT.bitsGT(TruncVT) || !TruncVT.isRound() ||
+ !TLI->isLoadExtLegal(ISD::ZEXTLOAD, LoadResultVT, TruncVT))
+ return false;
+
+ IRBuilder<> Builder(Load->getNextNode());
+ auto *NewAnd = dyn_cast<Instruction>(
+ Builder.CreateAnd(Load, ConstantInt::get(Ctx, DemandBits)));
+
+ // Replace all uses of load with new and (except for the use of load in the
+ // new and itself).
+ Load->replaceAllUsesWith(NewAnd);
+ NewAnd->setOperand(0, Load);
+
+ // Remove any and instructions that are now redundant.
+ for (auto *And : AndsToMaybeRemove)
+ // Check that the and mask is the same as the one we decided to put on the
+ // new and.
+ if (cast<ConstantInt>(And->getOperand(1))->getValue() == DemandBits) {
+ And->replaceAllUsesWith(NewAnd);
+ if (&*CurInstIterator == And)
+ CurInstIterator = std::next(And->getIterator());
+ And->eraseFromParent();
+ ++NumAndUses;
+ }
+
+ ++NumAndsAdded;
+ return true;
+}
+
+/// Check if V (an operand of a select instruction) is an expensive instruction
+/// that is only used once.
+static bool sinkSelectOperand(const TargetTransformInfo *TTI, Value *V) {
+ auto *I = dyn_cast<Instruction>(V);
+ // If it's safe to speculatively execute, then it should not have side
+ // effects; therefore, it's safe to sink and possibly *not* execute.
+ return I && I->hasOneUse() && isSafeToSpeculativelyExecute(I) &&
+ TTI->getUserCost(I) >= TargetTransformInfo::TCC_Expensive;
+}
+
+/// Returns true if a SelectInst should be turned into an explicit branch.
+static bool isFormingBranchFromSelectProfitable(const TargetTransformInfo *TTI,
+ SelectInst *SI) {
// FIXME: This should use the same heuristics as IfConversion to determine
// whether a select is better represented as a branch. This requires that
// branch probability metadata is preserved for the select, which is not the
CmpInst *Cmp = dyn_cast<CmpInst>(SI->getCondition());
- // If the branch is predicted right, an out of order CPU can avoid blocking on
- // the compare. Emit cmovs on compares with a memory operand as branches to
- // avoid stalls on the load from memory. If the compare has more than one use
- // there's probably another cmov or setcc around so it's not worth emitting a
- // branch.
- if (!Cmp)
+ // If a branch is predictable, an out-of-order CPU can avoid blocking on its
+ // comparison condition. If the compare has more than one use, there's
+ // probably another cmov or setcc around, so it's not worth emitting a branch.
+ if (!Cmp || !Cmp->hasOneUse())
return false;
Value *CmpOp0 = Cmp->getOperand(0);
Value *CmpOp1 = Cmp->getOperand(1);
- // We check that the memory operand has one use to avoid uses of the loaded
- // value directly after the compare, making branches unprofitable.
- return Cmp->hasOneUse() &&
- ((isa<LoadInst>(CmpOp0) && CmpOp0->hasOneUse()) ||
- (isa<LoadInst>(CmpOp1) && CmpOp1->hasOneUse()));
+ // Emit "cmov on compare with a memory operand" as a branch to avoid stalls
+ // on a load from memory. But if the load is used more than once, do not
+ // change the select to a branch because the load is probably needed
+ // regardless of whether the branch is taken or not.
+ if ((isa<LoadInst>(CmpOp0) && CmpOp0->hasOneUse()) ||
+ (isa<LoadInst>(CmpOp1) && CmpOp1->hasOneUse()))
+ return true;
+
+ // If either operand of the select is expensive and only needed on one side
+ // of the select, we should form a branch.
+ if (sinkSelectOperand(TTI, SI->getTrueValue()) ||
+ sinkSelectOperand(TTI, SI->getFalseValue()))
+ return true;
+
+ return false;
}
/// If we have a SelectInst that will likely profit from branch prediction,
/// turn it into a branch.
-bool CodeGenPrepare::OptimizeSelectInst(SelectInst *SI) {
+bool CodeGenPrepare::optimizeSelectInst(SelectInst *SI) {
bool VectorCond = !SI->getCondition()->getType()->isIntegerTy(1);
// Can we convert the 'select' to CF ?
// We have efficient codegen support for the select instruction.
// Check if it is profitable to keep this 'select'.
if (!TLI->isPredictableSelectExpensive() ||
- !isFormingBranchFromSelectProfitable(SI))
+ !isFormingBranchFromSelectProfitable(TTI, SI))
return false;
}
ModifiedDT = true;
+ // Transform a sequence like this:
+ // start:
+ // %cmp = cmp uge i32 %a, %b
+ // %sel = select i1 %cmp, i32 %c, i32 %d
+ //
+ // Into:
+ // start:
+ // %cmp = cmp uge i32 %a, %b
+ // br i1 %cmp, label %select.true, label %select.false
+ // select.true:
+ // br label %select.end
+ // select.false:
+ // br label %select.end
+ // select.end:
+ // %sel = phi i32 [ %c, %select.true ], [ %d, %select.false ]
+ //
+ // In addition, we may sink instructions that produce %c or %d from
+ // the entry block into the destination(s) of the new branch.
+ // If the true or false blocks do not contain a sunken instruction, that
+ // block and its branch may be optimized away. In that case, one side of the
+ // first branch will point directly to select.end, and the corresponding PHI
+ // predecessor block will be the start block.
+
// First, we split the block containing the select into 2 blocks.
BasicBlock *StartBlock = SI->getParent();
BasicBlock::iterator SplitPt = ++(BasicBlock::iterator(SI));
- BasicBlock *NextBlock = StartBlock->splitBasicBlock(SplitPt, "select.end");
-
- // Create a new block serving as the landing pad for the branch.
- BasicBlock *SmallBlock = BasicBlock::Create(SI->getContext(), "select.mid",
- NextBlock->getParent(), NextBlock);
+ BasicBlock *EndBlock = StartBlock->splitBasicBlock(SplitPt, "select.end");
- // Move the unconditional branch from the block with the select in it into our
- // landing pad block.
+ // Delete the unconditional branch that was just created by the split.
StartBlock->getTerminator()->eraseFromParent();
- BranchInst::Create(NextBlock, SmallBlock);
+
+ // These are the new basic blocks for the conditional branch.
+ // At least one will become an actual new basic block.
+ BasicBlock *TrueBlock = nullptr;
+ BasicBlock *FalseBlock = nullptr;
+
+ // Sink expensive instructions into the conditional blocks to avoid executing
+ // them speculatively.
+ if (sinkSelectOperand(TTI, SI->getTrueValue())) {
+ TrueBlock = BasicBlock::Create(SI->getContext(), "select.true.sink",
+ EndBlock->getParent(), EndBlock);
+ auto *TrueBranch = BranchInst::Create(EndBlock, TrueBlock);
+ auto *TrueInst = cast<Instruction>(SI->getTrueValue());
+ TrueInst->moveBefore(TrueBranch);
+ }
+ if (sinkSelectOperand(TTI, SI->getFalseValue())) {
+ FalseBlock = BasicBlock::Create(SI->getContext(), "select.false.sink",
+ EndBlock->getParent(), EndBlock);
+ auto *FalseBranch = BranchInst::Create(EndBlock, FalseBlock);
+ auto *FalseInst = cast<Instruction>(SI->getFalseValue());
+ FalseInst->moveBefore(FalseBranch);
+ }
+
+ // If there was nothing to sink, then arbitrarily choose the 'false' side
+ // for a new input value to the PHI.
+ if (TrueBlock == FalseBlock) {
+ assert(TrueBlock == nullptr &&
+ "Unexpected basic block transform while optimizing select");
+
+ FalseBlock = BasicBlock::Create(SI->getContext(), "select.false",
+ EndBlock->getParent(), EndBlock);
+ BranchInst::Create(EndBlock, FalseBlock);
+ }
// Insert the real conditional branch based on the original condition.
- BranchInst::Create(NextBlock, SmallBlock, SI->getCondition(), SI);
+ // If we did not create a new block for one of the 'true' or 'false' paths
+ // of the condition, it means that side of the branch goes to the end block
+ // directly and the path originates from the start block from the point of
+ // view of the new PHI.
+ if (TrueBlock == nullptr) {
+ BranchInst::Create(EndBlock, FalseBlock, SI->getCondition(), SI);
+ TrueBlock = StartBlock;
+ } else if (FalseBlock == nullptr) {
+ BranchInst::Create(TrueBlock, EndBlock, SI->getCondition(), SI);
+ FalseBlock = StartBlock;
+ } else {
+ BranchInst::Create(TrueBlock, FalseBlock, SI->getCondition(), SI);
+ }
// The select itself is replaced with a PHI Node.
- PHINode *PN = PHINode::Create(SI->getType(), 2, "", NextBlock->begin());
+ PHINode *PN = PHINode::Create(SI->getType(), 2, "", &EndBlock->front());
PN->takeName(SI);
- PN->addIncoming(SI->getTrueValue(), StartBlock);
- PN->addIncoming(SI->getFalseValue(), SmallBlock);
+ PN->addIncoming(SI->getTrueValue(), TrueBlock);
+ PN->addIncoming(SI->getFalseValue(), FalseBlock);
+
SI->replaceAllUsesWith(PN);
SI->eraseFromParent();
/// (e.g. x86 only introduced "vpsllvd" and friends with AVX2). In these cases
/// it's often worth sinking a shufflevector splat down to its use so that
/// codegen can spot all lanes are identical.
-bool CodeGenPrepare::OptimizeShuffleVectorInst(ShuffleVectorInst *SVI) {
+bool CodeGenPrepare::optimizeShuffleVectorInst(ShuffleVectorInst *SVI) {
BasicBlock *DefBB = SVI->getParent();
// Only do this xform if variable vector shifts are particularly expensive.
if (!InsertedShuffle) {
BasicBlock::iterator InsertPt = UserBB->getFirstInsertionPt();
- InsertedShuffle = new ShuffleVectorInst(SVI->getOperand(0),
- SVI->getOperand(1),
- SVI->getOperand(2), "", InsertPt);
+ assert(InsertPt != UserBB->end());
+ InsertedShuffle =
+ new ShuffleVectorInst(SVI->getOperand(0), SVI->getOperand(1),
+ SVI->getOperand(2), "", &*InsertPt);
}
UI->replaceUsesOfWith(SVI, InsertedShuffle);
return MadeChange;
}
+bool CodeGenPrepare::optimizeSwitchInst(SwitchInst *SI) {
+ if (!TLI || !DL)
+ return false;
+
+ Value *Cond = SI->getCondition();
+ Type *OldType = Cond->getType();
+ LLVMContext &Context = Cond->getContext();
+ MVT RegType = TLI->getRegisterType(Context, TLI->getValueType(*DL, OldType));
+ unsigned RegWidth = RegType.getSizeInBits();
+
+ if (RegWidth <= cast<IntegerType>(OldType)->getBitWidth())
+ return false;
+
+ // If the register width is greater than the type width, expand the condition
+ // of the switch instruction and each case constant to the width of the
+ // register. By widening the type of the switch condition, subsequent
+ // comparisons (for case comparisons) will not need to be extended to the
+ // preferred register width, so we will potentially eliminate N-1 extends,
+ // where N is the number of cases in the switch.
+ auto *NewType = Type::getIntNTy(Context, RegWidth);
+
+ // Zero-extend the switch condition and case constants unless the switch
+ // condition is a function argument that is already being sign-extended.
+ // In that case, we can avoid an unnecessary mask/extension by sign-extending
+ // everything instead.
+ Instruction::CastOps ExtType = Instruction::ZExt;
+ if (auto *Arg = dyn_cast<Argument>(Cond))
+ if (Arg->hasSExtAttr())
+ ExtType = Instruction::SExt;
+
+ auto *ExtInst = CastInst::Create(ExtType, Cond, NewType);
+ ExtInst->insertBefore(SI);
+ SI->setCondition(ExtInst);
+ for (SwitchInst::CaseIt Case : SI->cases()) {
+ APInt NarrowConst = Case.getCaseValue()->getValue();
+ APInt WideConst = (ExtType == Instruction::ZExt) ?
+ NarrowConst.zext(RegWidth) : NarrowConst.sext(RegWidth);
+ Case.setValue(ConstantInt::get(Context, WideConst));
+ }
+
+ return true;
+}
+
namespace {
/// \brief Helper class to promote a scalar operation to a vector one.
/// This class is used to move downward extractelement transition.
/// Some targets can do store(extractelement) with one instruction.
/// Try to push the extractelement towards the stores when the target
/// has this feature and this is profitable.
-bool CodeGenPrepare::OptimizeExtractElementInst(Instruction *Inst) {
+bool CodeGenPrepare::optimizeExtractElementInst(Instruction *Inst) {
unsigned CombineCost = UINT_MAX;
if (DisableStoreExtract || !TLI ||
(!StressStoreExtract &&
return false;
}
-bool CodeGenPrepare::OptimizeInst(Instruction *I, bool& ModifiedDT) {
+bool CodeGenPrepare::optimizeInst(Instruction *I, bool& ModifiedDT) {
// Bail out if we inserted the instruction to prevent optimizations from
// stepping on each other's toes.
if (InsertedInsts.count(I))
TargetLowering::TypeExpandInteger) {
return SinkCast(CI);
} else {
- bool MadeChange = MoveExtToFormExtLoad(I);
- return MadeChange | OptimizeExtUses(I);
+ bool MadeChange = moveExtToFormExtLoad(I);
+ return MadeChange | optimizeExtUses(I);
}
}
return false;
if (LoadInst *LI = dyn_cast<LoadInst>(I)) {
stripInvariantGroupMetadata(*LI);
if (TLI) {
+ bool Modified = optimizeLoadExt(LI);
unsigned AS = LI->getPointerAddressSpace();
- return OptimizeMemoryInst(I, I->getOperand(0), LI->getType(), AS);
+ Modified |= optimizeMemoryInst(I, I->getOperand(0), LI->getType(), AS);
+ return Modified;
}
return false;
}
stripInvariantGroupMetadata(*SI);
if (TLI) {
unsigned AS = SI->getPointerAddressSpace();
- return OptimizeMemoryInst(I, SI->getOperand(1),
+ return optimizeMemoryInst(I, SI->getOperand(1),
SI->getOperand(0)->getType(), AS);
}
return false;
GEPI->replaceAllUsesWith(NC);
GEPI->eraseFromParent();
++NumGEPsElim;
- OptimizeInst(NC, ModifiedDT);
+ optimizeInst(NC, ModifiedDT);
return true;
}
return false;
}
if (CallInst *CI = dyn_cast<CallInst>(I))
- return OptimizeCallInst(CI, ModifiedDT);
+ return optimizeCallInst(CI, ModifiedDT);
if (SelectInst *SI = dyn_cast<SelectInst>(I))
- return OptimizeSelectInst(SI);
+ return optimizeSelectInst(SI);
if (ShuffleVectorInst *SVI = dyn_cast<ShuffleVectorInst>(I))
- return OptimizeShuffleVectorInst(SVI);
+ return optimizeShuffleVectorInst(SVI);
+
+ if (auto *Switch = dyn_cast<SwitchInst>(I))
+ return optimizeSwitchInst(Switch);
if (isa<ExtractElementInst>(I))
- return OptimizeExtractElementInst(I);
+ return optimizeExtractElementInst(I);
return false;
}
+/// Given an OR instruction, check to see if this is a bitreverse
+/// idiom. If so, insert the new intrinsic and return true.
+static bool makeBitReverse(Instruction &I, const DataLayout &DL,
+ const TargetLowering &TLI) {
+ if (!I.getType()->isIntegerTy() ||
+ !TLI.isOperationLegalOrCustom(ISD::BITREVERSE,
+ TLI.getValueType(DL, I.getType(), true)))
+ return false;
+
+ SmallVector<Instruction*, 4> Insts;
+ if (!recognizeBitReverseOrBSwapIdiom(&I, false, true, Insts))
+ return false;
+ Instruction *LastInst = Insts.back();
+ I.replaceAllUsesWith(LastInst);
+ RecursivelyDeleteTriviallyDeadInstructions(&I);
+ return true;
+}
+
// In this pass we look for GEP and cast instructions that are used
// across basic blocks and rewrite them to improve basic-block-at-a-time
// selection.
-bool CodeGenPrepare::OptimizeBlock(BasicBlock &BB, bool& ModifiedDT) {
+bool CodeGenPrepare::optimizeBlock(BasicBlock &BB, bool& ModifiedDT) {
SunkAddrs.clear();
bool MadeChange = false;
CurInstIterator = BB.begin();
while (CurInstIterator != BB.end()) {
- MadeChange |= OptimizeInst(CurInstIterator++, ModifiedDT);
+ MadeChange |= optimizeInst(&*CurInstIterator++, ModifiedDT);
if (ModifiedDT)
return true;
}
- MadeChange |= DupRetToEnableTailCallOpts(&BB);
+ bool MadeBitReverse = true;
+ while (TLI && MadeBitReverse) {
+ MadeBitReverse = false;
+ for (auto &I : reverse(BB)) {
+ if (makeBitReverse(I, *DL, *TLI)) {
+ MadeBitReverse = MadeChange = true;
+ break;
+ }
+ }
+ }
+ MadeChange |= dupRetToEnableTailCallOpts(&BB);
+
return MadeChange;
}
// llvm.dbg.value is far away from the value then iSel may not be able
// handle it properly. iSel will drop llvm.dbg.value if it can not
// find a node corresponding to the value.
-bool CodeGenPrepare::PlaceDbgValues(Function &F) {
+bool CodeGenPrepare::placeDbgValues(Function &F) {
bool MadeChange = false;
for (BasicBlock &BB : F) {
Instruction *PrevNonDbgInst = nullptr;
for (BasicBlock::iterator BI = BB.begin(), BE = BB.end(); BI != BE;) {
- Instruction *Insn = BI++;
+ Instruction *Insn = &*BI++;
DbgValueInst *DVI = dyn_cast<DbgValueInst>(Insn);
// Leave dbg.values that refer to an alloca alone. These
// instrinsics describe the address of a variable (= the alloca)
Instruction *VI = dyn_cast_or_null<Instruction>(DVI->getValue());
if (VI && VI != PrevNonDbgInst && !VI->isTerminator()) {
+ // If VI is a phi in a block with an EHPad terminator, we can't insert
+ // after it.
+ if (isa<PHINode>(VI) && VI->getParent()->getTerminator()->isEHPad())
+ continue;
DEBUG(dbgs() << "Moving Debug Value before :\n" << *DVI << ' ' << *VI);
DVI->removeFromParent();
if (isa<PHINode>(VI))
- DVI->insertBefore(VI->getParent()->getFirstInsertionPt());
+ DVI->insertBefore(&*VI->getParent()->getFirstInsertionPt());
else
DVI->insertAfter(VI);
MadeChange = true;
return false;
bool MadeChange = false;
for (Function::iterator I = F.begin(), E = F.end(); I != E; ) {
- BasicBlock *BB = I++;
+ BasicBlock *BB = &*I++;
// Does this BB end with the following?
// %andVal = and %val, #single-bit-set
projects / oota-llvm.git / blobdiff