//
//===----------------------------------------------------------------------===//
-#define DEBUG_TYPE "codegenprepare"
#include "llvm/CodeGen/Passes.h"
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Target/TargetLibraryInfo.h"
#include "llvm/Target/TargetLowering.h"
+#include "llvm/Target/TargetSubtargetInfo.h"
#include "llvm/Transforms/Utils/BasicBlockUtils.h"
#include "llvm/Transforms/Utils/BuildLibCalls.h"
#include "llvm/Transforms/Utils/BypassSlowDivision.h"
using namespace llvm;
using namespace llvm::PatternMatch;
+#define DEBUG_TYPE "codegenprepare"
+
STATISTIC(NumBlocksElim, "Number of blocks eliminated");
STATISTIC(NumPHIsElim, "Number of trivial PHIs eliminated");
STATISTIC(NumGEPsElim, "Number of GEPs converted to casts");
STATISTIC(NumRetsDup, "Number of return instructions duplicated");
STATISTIC(NumDbgValueMoved, "Number of debug value instructions moved");
STATISTIC(NumSelectsExpanded, "Number of selects turned into branches");
+STATISTIC(NumAndCmpsMoved, "Number of and/cmp's pushed into branches");
static cl::opt<bool> DisableBranchOpts(
"disable-cgp-branch-opts", cl::Hidden, cl::init(false),
"disable-cgp-select2branch", cl::Hidden, cl::init(false),
cl::desc("Disable select to branch conversion."));
+static cl::opt<bool> AddrSinkUsingGEPs(
+ "addr-sink-using-gep", cl::Hidden, cl::init(false),
+ cl::desc("Address sinking in CGP using GEPs."));
+
+static cl::opt<bool> EnableAndCmpSinking(
+ "enable-andcmp-sinking", cl::Hidden, cl::init(true),
+ cl::desc("Enable sinkinig and/cmp into branches."));
+
namespace {
typedef SmallPtrSet<Instruction *, 16> SetOfInstrs;
typedef DenseMap<Instruction *, Type *> InstrToOrigTy;
public:
static char ID; // Pass identification, replacement for typeid
- explicit CodeGenPrepare(const TargetMachine *TM = 0)
- : FunctionPass(ID), TM(TM), TLI(0) {
+ explicit CodeGenPrepare(const TargetMachine *TM = nullptr)
+ : FunctionPass(ID), TM(TM), TLI(nullptr) {
initializeCodeGenPreparePass(*PassRegistry::getPassRegistry());
}
bool runOnFunction(Function &F) override;
bool OptimizeMemoryInst(Instruction *I, Value *Addr, Type *AccessTy);
bool OptimizeInlineAsmInst(CallInst *CS);
bool OptimizeCallInst(CallInst *CI);
- bool SinkExtExpand(CastInst *I);
bool MoveExtToFormExtLoad(Instruction *I);
bool OptimizeExtUses(Instruction *I);
bool OptimizeSelectInst(SelectInst *SI);
bool OptimizeShuffleVectorInst(ShuffleVectorInst *SI);
bool DupRetToEnableTailCallOpts(BasicBlock *BB);
bool PlaceDbgValues(Function &F);
+ bool sinkAndCmp(Function &F);
};
}
char CodeGenPrepare::ID = 0;
-static void *initializeCodeGenPreparePassOnce(PassRegistry &Registry) {
- initializeTargetLibraryInfoPass(Registry);
- PassInfo *PI = new PassInfo(
- "Optimize for code generation", "codegenprepare", &CodeGenPrepare::ID,
- PassInfo::NormalCtor_t(callDefaultCtor<CodeGenPrepare>), false, false,
- PassInfo::TargetMachineCtor_t(callTargetMachineCtor<CodeGenPrepare>));
- Registry.registerPass(*PI, true);
- return PI;
-}
-
-void llvm::initializeCodeGenPreparePass(PassRegistry &Registry) {
- CALL_ONCE_INITIALIZATION(initializeCodeGenPreparePassOnce)
-}
+INITIALIZE_TM_PASS(CodeGenPrepare, "codegenprepare",
+ "Optimize for code generation", false, false)
FunctionPass *llvm::createCodeGenPreparePass(const TargetMachine *TM) {
return new CodeGenPrepare(TM);
}
bool CodeGenPrepare::runOnFunction(Function &F) {
+ if (skipOptnoneFunction(F))
+ return false;
+
bool EverMadeChange = false;
// Clear per function information.
InsertedTruncsSet.clear();
PromotedInsts.clear();
ModifiedDT = false;
- if (TM) TLI = TM->getTargetLowering();
+ if (TM)
+ TLI = TM->getSubtargetImpl()->getTargetLowering();
TLInfo = &getAnalysis<TargetLibraryInfo>();
DominatorTreeWrapperPass *DTWP =
getAnalysisIfAvailable<DominatorTreeWrapperPass>();
- DT = DTWP ? &DTWP->getDomTree() : 0;
+ DT = DTWP ? &DTWP->getDomTree() : nullptr;
OptSize = F.getAttributes().hasAttribute(AttributeSet::FunctionIndex,
Attribute::OptimizeForSize);
// find a node corresponding to the value.
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
+ // users. Do this before OptimizeBlock -> OptimizeInst ->
+ // OptimizeCmpExpression, which perturbs the pattern being searched for.
+ if (!DisableBranchOpts)
+ EverMadeChange |= sinkAndCmp(F);
+
bool MadeChange = true;
while (MadeChange) {
MadeChange = false;
return MadeChange;
}
+/// isExtractBitsCandidateUse - Check if the candidates could
+/// be combined with shift instruction, which includes:
+/// 1. Truncate instruction
+/// 2. And instruction and the imm is a mask of the low bits:
+/// imm & (imm+1) == 0
+static bool isExtractBitsCandidateUse(Instruction *User) {
+ if (!isa<TruncInst>(User)) {
+ if (User->getOpcode() != Instruction::And ||
+ !isa<ConstantInt>(User->getOperand(1)))
+ return false;
+
+ const APInt &Cimm = cast<ConstantInt>(User->getOperand(1))->getValue();
+
+ if ((Cimm & (Cimm + 1)).getBoolValue())
+ return false;
+ }
+ return true;
+}
+
+/// SinkShiftAndTruncate - 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,
+ const TargetLowering &TLI) {
+ BasicBlock *UserBB = User->getParent();
+ DenseMap<BasicBlock *, CastInst *> InsertedTruncs;
+ TruncInst *TruncI = dyn_cast<TruncInst>(User);
+ bool MadeChange = false;
+
+ for (Value::user_iterator TruncUI = TruncI->user_begin(),
+ TruncE = TruncI->user_end();
+ TruncUI != TruncE;) {
+
+ Use &TruncTheUse = TruncUI.getUse();
+ Instruction *TruncUser = cast<Instruction>(*TruncUI);
+ // Preincrement use iterator so we don't invalidate it.
+
+ ++TruncUI;
+
+ int ISDOpcode = TLI.InstructionOpcodeToISD(TruncUser->getOpcode());
+ if (!ISDOpcode)
+ continue;
+
+ // If the use is actually a legal node, there will not be an
+ // implicit truncate.
+ // FIXME: always querying the result type is just an
+ // approximation; some nodes' legality is determined by the
+ // operand or other means. There's no good way to find out though.
+ if (TLI.isOperationLegalOrCustom(ISDOpcode,
+ EVT::getEVT(TruncUser->getType(), true)))
+ continue;
+
+ // Don't bother for PHI nodes.
+ if (isa<PHINode>(TruncUser))
+ continue;
+
+ BasicBlock *TruncUserBB = TruncUser->getParent();
+
+ if (UserBB == TruncUserBB)
+ continue;
+
+ BinaryOperator *&InsertedShift = InsertedShifts[TruncUserBB];
+ CastInst *&InsertedTrunc = InsertedTruncs[TruncUserBB];
+
+ if (!InsertedShift && !InsertedTrunc) {
+ BasicBlock::iterator InsertPt = TruncUserBB->getFirstInsertionPt();
+ // Sink the shift
+ if (ShiftI->getOpcode() == Instruction::AShr)
+ InsertedShift =
+ BinaryOperator::CreateAShr(ShiftI->getOperand(0), CI, "", InsertPt);
+ else
+ InsertedShift =
+ BinaryOperator::CreateLShr(ShiftI->getOperand(0), CI, "", InsertPt);
+
+ // Sink the trunc
+ BasicBlock::iterator TruncInsertPt = TruncUserBB->getFirstInsertionPt();
+ TruncInsertPt++;
+
+ InsertedTrunc = CastInst::Create(TruncI->getOpcode(), InsertedShift,
+ TruncI->getType(), "", TruncInsertPt);
+
+ MadeChange = true;
+
+ TruncTheUse = InsertedTrunc;
+ }
+ }
+ 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:
+/// BB1:
+/// %x.extract.shift = lshr i64 %arg1, 32
+/// BB2:
+/// %x.extract.trunc = trunc i64 %x.extract.shift to i16
+/// ==>
+///
+/// BB2:
+/// %x.extract.shift.1 = lshr i64 %arg1, 32
+/// %x.extract.trunc = trunc i64 %x.extract.shift.1 to i16
+///
+/// CodeGen will recoginze the pattern in BB2 and generate BitExtract
+/// instruction.
+/// Return true if any changes are made.
+static bool OptimizeExtractBits(BinaryOperator *ShiftI, ConstantInt *CI,
+ const TargetLowering &TLI) {
+ BasicBlock *DefBB = ShiftI->getParent();
+
+ /// Only insert instructions in each block once.
+ DenseMap<BasicBlock *, BinaryOperator *> InsertedShifts;
+
+ bool shiftIsLegal = TLI.isTypeLegal(TLI.getValueType(ShiftI->getType()));
+
+ bool MadeChange = false;
+ for (Value::user_iterator UI = ShiftI->user_begin(), E = ShiftI->user_end();
+ UI != E;) {
+ Use &TheUse = UI.getUse();
+ Instruction *User = cast<Instruction>(*UI);
+ // Preincrement use iterator so we don't invalidate it.
+ ++UI;
+
+ // Don't bother for PHI nodes.
+ if (isa<PHINode>(User))
+ continue;
+
+ if (!isExtractBitsCandidateUse(User))
+ continue;
+
+ BasicBlock *UserBB = User->getParent();
+
+ if (UserBB == DefBB) {
+ // If the shift and truncate instruction are in the same BB. The use of
+ // the truncate(TruncUse) may still introduce another truncate if not
+ // legal. In this case, we would like to sink both shift and truncate
+ // instruction to the BB of TruncUse.
+ // for example:
+ // BB1:
+ // i64 shift.result = lshr i64 opnd, imm
+ // trunc.result = trunc shift.result to i16
+ //
+ // BB2:
+ // ----> We will have an implicit truncate here if the architecture does
+ // not have i16 compare.
+ // cmp i16 trunc.result, opnd2
+ //
+ if (isa<TruncInst>(User) && shiftIsLegal
+ // If the type of the truncate is legal, no trucate will be
+ // introduced in other basic blocks.
+ && (!TLI.isTypeLegal(TLI.getValueType(User->getType()))))
+ MadeChange =
+ SinkShiftAndTruncate(ShiftI, User, CI, InsertedShifts, TLI);
+
+ continue;
+ }
+ // If we have already inserted a shift into this block, use it.
+ BinaryOperator *&InsertedShift = InsertedShifts[UserBB];
+
+ if (!InsertedShift) {
+ BasicBlock::iterator InsertPt = UserBB->getFirstInsertionPt();
+
+ if (ShiftI->getOpcode() == Instruction::AShr)
+ InsertedShift =
+ BinaryOperator::CreateAShr(ShiftI->getOperand(0), CI, "", InsertPt);
+ else
+ 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;
+}
+
namespace {
class CodeGenPrepareFortifiedLibCalls : public SimplifyFortifiedLibCalls {
protected:
// happens.
WeakVH IterHandle(CurInstIterator);
- replaceAndRecursivelySimplify(CI, RetVal, TLI ? TLI->getDataLayout() : 0,
- TLInfo, ModifiedDT ? 0 : DT);
+ replaceAndRecursivelySimplify(CI, RetVal,
+ TLI ? TLI->getDataLayout() : nullptr,
+ TLInfo, ModifiedDT ? nullptr : DT);
// If the iterator instruction was recursively deleted, start over at the
// start of the block.
}
// From here on out we're working with named functions.
- if (CI->getCalledFunction() == 0) return false;
+ if (!CI->getCalledFunction()) return false;
// We'll need DataLayout from here on out.
- const DataLayout *TD = TLI ? TLI->getDataLayout() : 0;
+ const DataLayout *TD = TLI ? TLI->getDataLayout() : nullptr;
if (!TD) return false;
// Lower all default uses of _chk calls. This is very similar
if (!RI)
return false;
- PHINode *PN = 0;
- BitCastInst *BCI = 0;
+ PHINode *PN = nullptr;
+ BitCastInst *BCI = nullptr;
Value *V = RI->getReturnValue();
if (V) {
BCI = dyn_cast<BitCastInst>(V);
struct ExtAddrMode : public TargetLowering::AddrMode {
Value *BaseReg;
Value *ScaledReg;
- ExtAddrMode() : BaseReg(0), ScaledReg(0) {}
+ ExtAddrMode() : BaseReg(nullptr), ScaledReg(nullptr) {}
void print(raw_ostream &OS) const;
void dump() const;
NeedPlus = true;
}
- if (BaseOffs)
- OS << (NeedPlus ? " + " : "") << BaseOffs, NeedPlus = true;
+ if (BaseOffs) {
+ OS << (NeedPlus ? " + " : "")
+ << BaseOffs;
+ NeedPlus = true;
+ }
if (BaseReg) {
OS << (NeedPlus ? " + " : "")
/// \brief Build a truncate instruction.
class TruncBuilder : public TypePromotionAction {
+ Value *Val;
public:
/// \brief Build a truncate instruction of \p Opnd producing a \p Ty
/// result.
/// trunc Opnd to Ty.
TruncBuilder(Instruction *Opnd, Type *Ty) : TypePromotionAction(Opnd) {
IRBuilder<> Builder(Opnd);
- Inst = cast<Instruction>(Builder.CreateTrunc(Opnd, Ty, "promoted"));
- DEBUG(dbgs() << "Do: TruncBuilder: " << *Inst << "\n");
+ Val = Builder.CreateTrunc(Opnd, Ty, "promoted");
+ DEBUG(dbgs() << "Do: TruncBuilder: " << *Val << "\n");
}
- /// \brief Get the built instruction.
- Instruction *getBuiltInstruction() { return Inst; }
+ /// \brief Get the built value.
+ Value *getBuiltValue() { return Val; }
/// \brief Remove the built instruction.
void undo() override {
- DEBUG(dbgs() << "Undo: TruncBuilder: " << *Inst << "\n");
- Inst->eraseFromParent();
+ DEBUG(dbgs() << "Undo: TruncBuilder: " << *Val << "\n");
+ if (Instruction *IVal = dyn_cast<Instruction>(Val))
+ IVal->eraseFromParent();
}
};
/// \brief Build a sign extension instruction.
class SExtBuilder : public TypePromotionAction {
+ Value *Val;
public:
/// \brief Build a sign extension instruction of \p Opnd producing a \p Ty
/// result.
/// sext Opnd to Ty.
SExtBuilder(Instruction *InsertPt, Value *Opnd, Type *Ty)
- : TypePromotionAction(Inst) {
+ : TypePromotionAction(InsertPt) {
+ IRBuilder<> Builder(InsertPt);
+ Val = Builder.CreateSExt(Opnd, Ty, "promoted");
+ DEBUG(dbgs() << "Do: SExtBuilder: " << *Val << "\n");
+ }
+
+ /// \brief Get the built value.
+ Value *getBuiltValue() { return Val; }
+
+ /// \brief Remove the built instruction.
+ void undo() override {
+ DEBUG(dbgs() << "Undo: SExtBuilder: " << *Val << "\n");
+ if (Instruction *IVal = dyn_cast<Instruction>(Val))
+ IVal->eraseFromParent();
+ }
+ };
+
+ /// \brief Build a zero extension instruction.
+ class ZExtBuilder : public TypePromotionAction {
+ Value *Val;
+ public:
+ /// \brief Build a zero extension instruction of \p Opnd producing a \p Ty
+ /// result.
+ /// zext Opnd to Ty.
+ ZExtBuilder(Instruction *InsertPt, Value *Opnd, Type *Ty)
+ : TypePromotionAction(InsertPt) {
IRBuilder<> Builder(InsertPt);
- Inst = cast<Instruction>(Builder.CreateSExt(Opnd, Ty, "promoted"));
- DEBUG(dbgs() << "Do: SExtBuilder: " << *Inst << "\n");
+ Val = Builder.CreateZExt(Opnd, Ty, "promoted");
+ DEBUG(dbgs() << "Do: ZExtBuilder: " << *Val << "\n");
}
- /// \brief Get the built instruction.
- Instruction *getBuiltInstruction() { return Inst; }
+ /// \brief Get the built value.
+ Value *getBuiltValue() { return Val; }
/// \brief Remove the built instruction.
void undo() override {
- DEBUG(dbgs() << "Undo: SExtBuilder: " << *Inst << "\n");
- Inst->eraseFromParent();
+ DEBUG(dbgs() << "Undo: ZExtBuilder: " << *Val << "\n");
+ if (Instruction *IVal = dyn_cast<Instruction>(Val))
+ IVal->eraseFromParent();
}
};
public:
/// \brief Remove all reference of \p Inst and optinally replace all its
/// uses with New.
- /// \pre If !Inst->use_empty(), then New != NULL
- InstructionRemover(Instruction *Inst, Value *New = NULL)
+ /// \pre If !Inst->use_empty(), then New != nullptr
+ InstructionRemover(Instruction *Inst, Value *New = nullptr)
: TypePromotionAction(Inst), Inserter(Inst), Hider(Inst),
- Replacer(NULL) {
+ Replacer(nullptr) {
if (New)
Replacer = new UsesReplacer(Inst, New);
DEBUG(dbgs() << "Do: InstructionRemover: " << *Inst << "\n");
/// Same as Instruction::setOperand.
void setOperand(Instruction *Inst, unsigned Idx, Value *NewVal);
/// Same as Instruction::eraseFromParent.
- void eraseInstruction(Instruction *Inst, Value *NewVal = NULL);
+ void eraseInstruction(Instruction *Inst, Value *NewVal = nullptr);
/// Same as Value::replaceAllUsesWith.
void replaceAllUsesWith(Instruction *Inst, Value *New);
/// Same as Value::mutateType.
void mutateType(Instruction *Inst, Type *NewTy);
/// Same as IRBuilder::createTrunc.
- Instruction *createTrunc(Instruction *Opnd, Type *Ty);
+ Value *createTrunc(Instruction *Opnd, Type *Ty);
/// Same as IRBuilder::createSExt.
- Instruction *createSExt(Instruction *Inst, Value *Opnd, Type *Ty);
+ Value *createSExt(Instruction *Inst, Value *Opnd, Type *Ty);
+ /// Same as IRBuilder::createZExt.
+ Value *createZExt(Instruction *Inst, Value *Opnd, Type *Ty);
/// Same as Instruction::moveBefore.
void moveBefore(Instruction *Inst, Instruction *Before);
/// @}
- ~TypePromotionTransaction();
-
private:
/// The ordered list of actions made so far.
- SmallVector<TypePromotionAction *, 16> Actions;
- typedef SmallVectorImpl<TypePromotionAction *>::iterator CommitPt;
+ SmallVector<std::unique_ptr<TypePromotionAction>, 16> Actions;
+ typedef SmallVectorImpl<std::unique_ptr<TypePromotionAction>>::iterator CommitPt;
};
void TypePromotionTransaction::setOperand(Instruction *Inst, unsigned Idx,
Value *NewVal) {
Actions.push_back(
- new TypePromotionTransaction::OperandSetter(Inst, Idx, NewVal));
+ make_unique<TypePromotionTransaction::OperandSetter>(Inst, Idx, NewVal));
}
void TypePromotionTransaction::eraseInstruction(Instruction *Inst,
Value *NewVal) {
Actions.push_back(
- new TypePromotionTransaction::InstructionRemover(Inst, NewVal));
+ make_unique<TypePromotionTransaction::InstructionRemover>(Inst, NewVal));
}
void TypePromotionTransaction::replaceAllUsesWith(Instruction *Inst,
Value *New) {
- Actions.push_back(new TypePromotionTransaction::UsesReplacer(Inst, New));
+ Actions.push_back(make_unique<TypePromotionTransaction::UsesReplacer>(Inst, New));
}
void TypePromotionTransaction::mutateType(Instruction *Inst, Type *NewTy) {
- Actions.push_back(new TypePromotionTransaction::TypeMutator(Inst, NewTy));
+ Actions.push_back(make_unique<TypePromotionTransaction::TypeMutator>(Inst, NewTy));
+}
+
+Value *TypePromotionTransaction::createTrunc(Instruction *Opnd,
+ Type *Ty) {
+ std::unique_ptr<TruncBuilder> Ptr(new TruncBuilder(Opnd, Ty));
+ Value *Val = Ptr->getBuiltValue();
+ Actions.push_back(std::move(Ptr));
+ return Val;
}
-Instruction *TypePromotionTransaction::createTrunc(Instruction *Opnd,
- Type *Ty) {
- TruncBuilder *TB = new TruncBuilder(Opnd, Ty);
- Actions.push_back(TB);
- return TB->getBuiltInstruction();
+Value *TypePromotionTransaction::createSExt(Instruction *Inst,
+ Value *Opnd, Type *Ty) {
+ std::unique_ptr<SExtBuilder> Ptr(new SExtBuilder(Inst, Opnd, Ty));
+ Value *Val = Ptr->getBuiltValue();
+ Actions.push_back(std::move(Ptr));
+ return Val;
}
-Instruction *TypePromotionTransaction::createSExt(Instruction *Inst,
- Value *Opnd, Type *Ty) {
- SExtBuilder *SB = new SExtBuilder(Inst, Opnd, Ty);
- Actions.push_back(SB);
- return SB->getBuiltInstruction();
+Value *TypePromotionTransaction::createZExt(Instruction *Inst,
+ Value *Opnd, Type *Ty) {
+ std::unique_ptr<ZExtBuilder> Ptr(new ZExtBuilder(Inst, Opnd, Ty));
+ Value *Val = Ptr->getBuiltValue();
+ Actions.push_back(std::move(Ptr));
+ return Val;
}
void TypePromotionTransaction::moveBefore(Instruction *Inst,
Instruction *Before) {
Actions.push_back(
- new TypePromotionTransaction::InstructionMoveBefore(Inst, Before));
+ make_unique<TypePromotionTransaction::InstructionMoveBefore>(Inst, Before));
}
TypePromotionTransaction::ConstRestorationPt
TypePromotionTransaction::getRestorationPoint() const {
- return Actions.rbegin() != Actions.rend() ? *Actions.rbegin() : NULL;
+ return !Actions.empty() ? Actions.back().get() : nullptr;
}
void TypePromotionTransaction::commit() {
for (CommitPt It = Actions.begin(), EndIt = Actions.end(); It != EndIt;
- ++It) {
+ ++It)
(*It)->commit();
- delete *It;
- }
Actions.clear();
}
void TypePromotionTransaction::rollback(
TypePromotionTransaction::ConstRestorationPt Point) {
- while (!Actions.empty() && Point != (*Actions.rbegin())) {
- TypePromotionAction *Curr = Actions.pop_back_val();
+ while (!Actions.empty() && Point != Actions.back().get()) {
+ std::unique_ptr<TypePromotionAction> Curr = Actions.pop_back_val();
Curr->undo();
- delete Curr;
}
}
-TypePromotionTransaction::~TypePromotionTransaction() {
- for (CommitPt It = Actions.begin(), EndIt = Actions.end(); It != EndIt; ++It)
- delete *It;
- Actions.clear();
-}
-
/// \brief A helper class for matching addressing modes.
///
/// This encapsulates the logic for matching the target-legal addressing modes.
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 = NULL);
+ bool *MovedAway = nullptr);
bool IsProfitableToFoldIntoAddressingMode(Instruction *I,
ExtAddrMode &AMBefore,
ExtAddrMode &AMAfter);
// Okay, we decided that we can add ScaleReg+Scale to AddrMode. Check now
// to see if ScaleReg is actually X+C. If so, we can turn this into adding
// X*Scale + C*Scale to addr mode.
- ConstantInt *CI = 0; Value *AddLHS = 0;
+ ConstantInt *CI = nullptr; Value *AddLHS = nullptr;
if (isa<Instruction>(ScaleReg) && // not a constant expr.
match(ScaleReg, m_Add(m_Value(AddLHS), m_ConstantInt(CI)))) {
TestAddrMode.ScaledReg = AddLHS;
static bool MightBeFoldableInst(Instruction *I) {
switch (I->getOpcode()) {
case Instruction::BitCast:
+ case Instruction::AddrSpaceCast:
// Don't touch identity bitcasts.
if (I->getType() == I->getOperand(0)->getType())
return false;
}
/// \brief Utility function to promote the operand of \p SExt when this
- /// operand is a promotable trunc or sext.
+ /// operand is a promotable trunc or sext or zext.
/// \p PromotedInsts maps the instructions to their type before promotion.
/// \p CreatedInsts[out] contains how many non-free instructions have been
/// created to promote the operand of SExt.
/// Should never be called directly.
/// \return The promoted value which is used instead of SExt.
- static Value *promoteOperandForTruncAndSExt(Instruction *SExt,
- TypePromotionTransaction &TPT,
- InstrToOrigTy &PromotedInsts,
- unsigned &CreatedInsts);
+ static Value *promoteOperandForTruncAndAnyExt(Instruction *SExt,
+ TypePromotionTransaction &TPT,
+ InstrToOrigTy &PromotedInsts,
+ unsigned &CreatedInsts);
/// \brief Utility function to promote the operand of \p SExt when this
/// operand is promotable and is not a supported trunc or sext.
bool TypePromotionHelper::canGetThrough(const Instruction *Inst,
Type *ConsideredSExtType,
const InstrToOrigTy &PromotedInsts) {
- // We can always get through sext.
- if (isa<SExtInst>(Inst))
+ // We can always get through sext or zext.
+ if (isa<SExtInst>(Inst) || isa<ZExtInst>(Inst))
return true;
// We can get through binary operator, if it is legal. In other words, the
// get through.
// If it, check we can get through.
if (!SExtOpnd || !canGetThrough(SExtOpnd, SExtTy, PromotedInsts))
- return NULL;
+ return nullptr;
// Do not promote if the operand has been added by codegenprepare.
// Otherwise, it means we are undoing an optimization that is likely to be
// redone, thus causing potential infinite loop.
if (isa<TruncInst>(SExtOpnd) && InsertedTruncs.count(SExtOpnd))
- return NULL;
+ return nullptr;
// SExt or Trunc instructions.
// Return the related handler.
- if (isa<SExtInst>(SExtOpnd) || isa<TruncInst>(SExtOpnd))
- return promoteOperandForTruncAndSExt;
+ if (isa<SExtInst>(SExtOpnd) || isa<TruncInst>(SExtOpnd) ||
+ isa<ZExtInst>(SExtOpnd))
+ return promoteOperandForTruncAndAnyExt;
// Regular instruction.
// Abort early if we will have to insert non-free instructions.
if (!SExtOpnd->hasOneUse() &&
!TLI.isTruncateFree(SExtTy, SExtOpnd->getType()))
- return NULL;
+ return nullptr;
return promoteOperandForOther;
}
-Value *TypePromotionHelper::promoteOperandForTruncAndSExt(
+Value *TypePromotionHelper::promoteOperandForTruncAndAnyExt(
llvm::Instruction *SExt, TypePromotionTransaction &TPT,
InstrToOrigTy &PromotedInsts, unsigned &CreatedInsts) {
// By construction, the operand of SExt is an instruction. Otherwise we cannot
// get through it and this method should not be called.
Instruction *SExtOpnd = cast<Instruction>(SExt->getOperand(0));
- // Replace sext(trunc(opnd)) or sext(sext(opnd))
- // => sext(opnd).
- TPT.setOperand(SExt, 0, SExtOpnd->getOperand(0));
+ Value *ExtVal = SExt;
+ if (isa<ZExtInst>(SExtOpnd)) {
+ // Replace sext(zext(opnd))
+ // => zext(opnd).
+ Value *ZExt =
+ TPT.createZExt(SExt, SExtOpnd->getOperand(0), SExt->getType());
+ TPT.replaceAllUsesWith(SExt, ZExt);
+ TPT.eraseInstruction(SExt);
+ ExtVal = ZExt;
+ } else {
+ // Replace sext(trunc(opnd)) or sext(sext(opnd))
+ // => sext(opnd).
+ TPT.setOperand(SExt, 0, SExtOpnd->getOperand(0));
+ }
CreatedInsts = 0;
// Remove dead code.
if (SExtOpnd->use_empty())
TPT.eraseInstruction(SExtOpnd);
- // Check if the sext is still needed.
- if (SExt->getType() != SExt->getOperand(0)->getType())
- return SExt;
+ // Check if the extension is still needed.
+ Instruction *ExtInst = dyn_cast<Instruction>(ExtVal);
+ if (!ExtInst || ExtInst->getType() != ExtInst->getOperand(0)->getType())
+ return ExtVal;
- // At this point we have: sext ty opnd to ty.
- // Reassign the uses of SExt to the opnd and remove SExt.
- Value *NextVal = SExt->getOperand(0);
- TPT.eraseInstruction(SExt, NextVal);
+ // At this point we have: ext ty opnd to ty.
+ // Reassign the uses of ExtInst to the opnd and remove ExtInst.
+ Value *NextVal = ExtInst->getOperand(0);
+ TPT.eraseInstruction(ExtInst, NextVal);
return NextVal;
}
// All its uses, but SExt, will need to use a truncated value of the
// promoted version.
// Create the truncate now.
- Instruction *Trunc = TPT.createTrunc(SExt, SExtOpnd->getType());
- Trunc->removeFromParent();
- // Insert it just after the definition.
- Trunc->insertAfter(SExtOpnd);
+ Value *Trunc = TPT.createTrunc(SExt, SExtOpnd->getType());
+ if (Instruction *ITrunc = dyn_cast<Instruction>(Trunc)) {
+ ITrunc->removeFromParent();
+ // Insert it just after the definition.
+ ITrunc->insertAfter(SExtOpnd);
+ }
TPT.replaceAllUsesWith(SExtOpnd, Trunc);
// Restore the operand of SExt (which has been replace by the previous call
if (!SExtForOpnd) {
// If yes, create a new one.
DEBUG(dbgs() << "More operands to sext\n");
- SExtForOpnd = TPT.createSExt(SExt, Opnd, SExt->getType());
+ SExtForOpnd =
+ cast<Instruction>(TPT.createSExt(SExt, Opnd, SExt->getType()));
++CreatedInsts;
}
TPT.moveBefore(SExtForOpnd, SExtOpnd);
TPT.setOperand(SExtOpnd, OpIdx, SExtForOpnd);
// If more sext are required, new instructions will have to be created.
- SExtForOpnd = NULL;
+ SExtForOpnd = nullptr;
}
if (SExtForOpnd == SExt) {
DEBUG(dbgs() << "Sign extension is useless now\n");
return MatchAddr(AddrInst->getOperand(0), Depth);
return false;
case Instruction::BitCast:
+ case Instruction::AddrSpaceCast:
// BitCast is always a noop, and we can handle it as long as it is
// int->int or pointer->pointer (we don't want int<->fp or something).
if ((AddrInst->getOperand(0)->getType()->isPointerTy() ||
case Instruction::Shl: {
// Can only handle X*C and X << C.
ConstantInt *RHS = dyn_cast<ConstantInt>(AddrInst->getOperand(1));
- if (!RHS) return false;
+ if (!RHS)
+ return false;
int64_t Scale = RHS->getSExtValue();
if (Opcode == Instruction::Shl)
Scale = 1LL << Scale;
return true;
}
case Instruction::SExt: {
+ Instruction *SExt = dyn_cast<Instruction>(AddrInst);
+ if (!SExt)
+ return false;
+
// Try to move this sext out of the way of the addressing mode.
- Instruction *SExt = cast<Instruction>(AddrInst);
// Ask for a method for doing so.
TypePromotionHelper::Action TPH = TypePromotionHelper::getAction(
SExt, InsertedTruncs, TLI, PromotedInsts);
AddrMode.BaseOffs -= CI->getSExtValue();
} else if (GlobalValue *GV = dyn_cast<GlobalValue>(Addr)) {
// If this is a global variable, try to fold it into the addressing mode.
- if (AddrMode.BaseGV == 0) {
+ if (!AddrMode.BaseGV) {
AddrMode.BaseGV = GV;
if (TLI.isLegalAddressingMode(AddrMode, AccessTy))
return true;
- AddrMode.BaseGV = 0;
+ AddrMode.BaseGV = nullptr;
}
} else if (Instruction *I = dyn_cast<Instruction>(Addr)) {
ExtAddrMode BackupAddrMode = AddrMode;
if (TLI.isLegalAddressingMode(AddrMode, AccessTy))
return true;
AddrMode.HasBaseReg = false;
- AddrMode.BaseReg = 0;
+ AddrMode.BaseReg = nullptr;
}
// If the base register is already taken, see if we can do [r+r].
if (TLI.isLegalAddressingMode(AddrMode, AccessTy))
return true;
AddrMode.Scale = 0;
- AddrMode.ScaledReg = 0;
+ AddrMode.ScaledReg = nullptr;
}
// Couldn't match.
TPT.rollback(LastKnownGood);
/// Add the ultimately found memory instructions to MemoryUses.
static bool FindAllMemoryUses(Instruction *I,
SmallVectorImpl<std::pair<Instruction*,unsigned> > &MemoryUses,
- SmallPtrSet<Instruction*, 16> &ConsideredInsts,
+ SmallPtrSetImpl<Instruction*> &ConsideredInsts,
const TargetLowering &TLI) {
// If we already considered this instruction, we're done.
if (!ConsideredInsts.insert(I))
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 == 0 || Val == KnownLive1 || Val == KnownLive2)
+ if (Val == nullptr || Val == KnownLive1 || Val == KnownLive2)
return true;
// All values other than instructions and arguments (e.g. constants) are live.
// 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))
- BaseReg = 0;
+ BaseReg = nullptr;
if (ValueAlreadyLiveAtInst(ScaledReg, AMBefore.BaseReg, AMBefore.ScaledReg))
- ScaledReg = 0;
+ ScaledReg = nullptr;
// If folding this instruction (and it's subexprs) didn't extend any live
// ranges, we're ok with it.
- if (BaseReg == 0 && ScaledReg == 0)
+ if (!BaseReg && !ScaledReg)
return true;
// If all uses of this instruction are ultimately load/store/inlineasm's,
// Use a worklist to iteratively look through PHI nodes, and ensure that
// the addressing mode obtained from the non-PHI roots of the graph
// are equivalent.
- Value *Consensus = 0;
+ Value *Consensus = nullptr;
unsigned NumUsesConsensus = 0;
bool IsNumUsesConsensusValid = false;
SmallVector<Instruction*, 16> AddrModeInsts;
// Break use-def graph loops.
if (!Visited.insert(V)) {
- Consensus = 0;
+ Consensus = nullptr;
break;
}
continue;
}
- Consensus = 0;
+ Consensus = nullptr;
break;
}
Value *&SunkAddr = SunkAddrs[Addr];
if (SunkAddr) {
DEBUG(dbgs() << "CGP: Reusing nonlocal addrmode: " << AddrMode << " for "
- << *MemoryInst);
+ << *MemoryInst << "\n");
if (SunkAddr->getType() != Addr->getType())
SunkAddr = Builder.CreateBitCast(SunkAddr, Addr->getType());
+ } else if (AddrSinkUsingGEPs || (!AddrSinkUsingGEPs.getNumOccurrences() &&
+ TM && TM->getSubtarget<TargetSubtargetInfo>().useAA())) {
+ // By default, we use the GEP-based method when AA is used later. This
+ // prevents new inttoptr/ptrtoint pairs from degrading AA capabilities.
+ DEBUG(dbgs() << "CGP: SINKING nonlocal addrmode: " << AddrMode << " for "
+ << *MemoryInst << "\n");
+ Type *IntPtrTy = TLI->getDataLayout()->getIntPtrType(Addr->getType());
+ Value *ResultPtr = nullptr, *ResultIndex = nullptr;
+
+ // First, find the pointer.
+ if (AddrMode.BaseReg && AddrMode.BaseReg->getType()->isPointerTy()) {
+ ResultPtr = AddrMode.BaseReg;
+ AddrMode.BaseReg = nullptr;
+ }
+
+ if (AddrMode.Scale && AddrMode.ScaledReg->getType()->isPointerTy()) {
+ // We can't add more than one pointer together, nor can we scale a
+ // pointer (both of which seem meaningless).
+ if (ResultPtr || AddrMode.Scale != 1)
+ return false;
+
+ ResultPtr = AddrMode.ScaledReg;
+ AddrMode.Scale = 0;
+ }
+
+ if (AddrMode.BaseGV) {
+ if (ResultPtr)
+ return false;
+
+ ResultPtr = AddrMode.BaseGV;
+ }
+
+ // If the real base value actually came from an inttoptr, then the matcher
+ // will look through it and provide only the integer value. In that case,
+ // use it here.
+ if (!ResultPtr && AddrMode.BaseReg) {
+ ResultPtr =
+ Builder.CreateIntToPtr(AddrMode.BaseReg, Addr->getType(), "sunkaddr");
+ AddrMode.BaseReg = nullptr;
+ } else if (!ResultPtr && AddrMode.Scale == 1) {
+ ResultPtr =
+ Builder.CreateIntToPtr(AddrMode.ScaledReg, Addr->getType(), "sunkaddr");
+ AddrMode.Scale = 0;
+ }
+
+ if (!ResultPtr &&
+ !AddrMode.BaseReg && !AddrMode.Scale && !AddrMode.BaseOffs) {
+ SunkAddr = Constant::getNullValue(Addr->getType());
+ } else if (!ResultPtr) {
+ return false;
+ } else {
+ Type *I8PtrTy =
+ Builder.getInt8PtrTy(Addr->getType()->getPointerAddressSpace());
+
+ // Start with the base register. Do this first so that subsequent address
+ // matching finds it last, which will prevent it from trying to match it
+ // as the scaled value in case it happens to be a mul. That would be
+ // problematic if we've sunk a different mul for the scale, because then
+ // we'd end up sinking both muls.
+ if (AddrMode.BaseReg) {
+ Value *V = AddrMode.BaseReg;
+ if (V->getType() != IntPtrTy)
+ V = Builder.CreateIntCast(V, IntPtrTy, /*isSigned=*/true, "sunkaddr");
+
+ ResultIndex = V;
+ }
+
+ // Add the scale value.
+ if (AddrMode.Scale) {
+ Value *V = AddrMode.ScaledReg;
+ if (V->getType() == IntPtrTy) {
+ // done.
+ } else if (cast<IntegerType>(IntPtrTy)->getBitWidth() <
+ cast<IntegerType>(V->getType())->getBitWidth()) {
+ V = Builder.CreateTrunc(V, IntPtrTy, "sunkaddr");
+ } else {
+ // It is only safe to sign extend the BaseReg if we know that the math
+ // required to create it did not overflow before we extend it. Since
+ // the original IR value was tossed in favor of a constant back when
+ // the AddrMode was created we need to bail out gracefully if widths
+ // do not match instead of extending it.
+ Instruction *I = dyn_cast_or_null<Instruction>(ResultIndex);
+ if (I && (ResultIndex != AddrMode.BaseReg))
+ I->eraseFromParent();
+ return false;
+ }
+
+ if (AddrMode.Scale != 1)
+ V = Builder.CreateMul(V, ConstantInt::get(IntPtrTy, AddrMode.Scale),
+ "sunkaddr");
+ if (ResultIndex)
+ ResultIndex = Builder.CreateAdd(ResultIndex, V, "sunkaddr");
+ else
+ ResultIndex = V;
+ }
+
+ // Add in the Base Offset if present.
+ if (AddrMode.BaseOffs) {
+ Value *V = ConstantInt::get(IntPtrTy, AddrMode.BaseOffs);
+ if (ResultIndex) {
+ // We need to add this separately from the scale above to help with
+ // SDAG consecutive load/store merging.
+ if (ResultPtr->getType() != I8PtrTy)
+ ResultPtr = Builder.CreateBitCast(ResultPtr, I8PtrTy);
+ ResultPtr = Builder.CreateGEP(ResultPtr, ResultIndex, "sunkaddr");
+ }
+
+ ResultIndex = V;
+ }
+
+ if (!ResultIndex) {
+ SunkAddr = ResultPtr;
+ } else {
+ if (ResultPtr->getType() != I8PtrTy)
+ ResultPtr = Builder.CreateBitCast(ResultPtr, I8PtrTy);
+ SunkAddr = Builder.CreateGEP(ResultPtr, ResultIndex, "sunkaddr");
+ }
+
+ if (SunkAddr->getType() != Addr->getType())
+ SunkAddr = Builder.CreateBitCast(SunkAddr, Addr->getType());
+ }
} else {
DEBUG(dbgs() << "CGP: SINKING nonlocal addrmode: " << AddrMode << " for "
- << *MemoryInst);
+ << *MemoryInst << "\n");
Type *IntPtrTy = TLI->getDataLayout()->getIntPtrType(Addr->getType());
- Value *Result = 0;
+ Value *Result = nullptr;
// Start with the base register. Do this first so that subsequent address
// matching finds it last, which will prevent it from trying to match it
cast<IntegerType>(V->getType())->getBitWidth()) {
V = Builder.CreateTrunc(V, IntPtrTy, "sunkaddr");
} else {
- V = Builder.CreateSExt(V, IntPtrTy, "sunkaddr");
+ // It is only safe to sign extend the BaseReg if we know that the math
+ // required to create it did not overflow before we extend it. Since
+ // the original IR value was tossed in favor of a constant back when
+ // the AddrMode was created we need to bail out gracefully if widths
+ // do not match instead of extending it.
+ Instruction *I = dyn_cast_or_null<Instruction>(Result);
+ if (I && (Result != AddrMode.BaseReg))
+ I->eraseFromParent();
+ return false;
}
if (AddrMode.Scale != 1)
V = Builder.CreateMul(V, ConstantInt::get(IntPtrTy, AddrMode.Scale),
Result = V;
}
- if (Result == 0)
+ if (!Result)
SunkAddr = Constant::getNullValue(Addr->getType());
else
SunkAddr = Builder.CreateIntToPtr(Result, Addr->getType(), "sunkaddr");
return MadeChange;
}
-/// SinkExtExpand - Sink a zext or sext into its user blocks if the target type
-/// doesn't fit in one register
-bool CodeGenPrepare::SinkExtExpand(CastInst *CI) {
- if (TLI &&
- TLI->getTypeAction(CI->getContext(), TLI->getValueType(CI->getType())) ==
- TargetLowering::TypeExpandInteger)
- return SinkCast(CI);
- 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.
if (LI->getParent() == I->getParent())
return false;
- // Do not undo the optimization in SinkExtExpand
- if (TLI &&
- TLI->getTypeAction(I->getContext(), TLI->getValueType(I->getType())) ==
- TargetLowering::TypeExpandInteger)
- return false;
-
// If the load has other users and the truncate is not free, this probably
// isn't worthwhile.
if (!LI->hasOneUse() &&
// It is possible for very late stage optimizations (such as SimplifyCFG)
// to introduce PHI nodes too late to be cleaned up. If we detect such a
// trivial PHI, go ahead and zap it here.
- if (Value *V = SimplifyInstruction(P, TLI ? TLI->getDataLayout() : 0,
+ if (Value *V = SimplifyInstruction(P, TLI ? TLI->getDataLayout() : nullptr,
TLInfo, DT)) {
P->replaceAllUsesWith(V);
P->eraseFromParent();
return true;
if (isa<ZExtInst>(I) || isa<SExtInst>(I)) {
- if (SinkExtExpand(CI))
- return true;
- bool MadeChange = MoveExtToFormExtLoad(I);
- return MadeChange | OptimizeExtUses(I);
+ /// Sink a zext or sext into its user blocks if the target type doesn't
+ /// fit in one register
+ if (TLI && TLI->getTypeAction(CI->getContext(),
+ TLI->getValueType(CI->getType())) ==
+ TargetLowering::TypeExpandInteger) {
+ return SinkCast(CI);
+ } else {
+ bool MadeChange = MoveExtToFormExtLoad(I);
+ return MadeChange | OptimizeExtUses(I);
+ }
}
return false;
}
return false;
}
+ BinaryOperator *BinOp = dyn_cast<BinaryOperator>(I);
+
+ if (BinOp && (BinOp->getOpcode() == Instruction::AShr ||
+ BinOp->getOpcode() == Instruction::LShr)) {
+ ConstantInt *CI = dyn_cast<ConstantInt>(BinOp->getOperand(1));
+ if (TLI && CI && TLI->hasExtractBitsInsn())
+ return OptimizeExtractBits(BinOp, CI, *TLI);
+
+ return false;
+ }
+
if (GetElementPtrInst *GEPI = dyn_cast<GetElementPtrInst>(I)) {
if (GEPI->hasAllZeroIndices()) {
/// The GEP operand must be a pointer, so must its result -> BitCast
bool CodeGenPrepare::PlaceDbgValues(Function &F) {
bool MadeChange = false;
for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I) {
- Instruction *PrevNonDbgInst = NULL;
+ Instruction *PrevNonDbgInst = nullptr;
for (BasicBlock::iterator BI = I->begin(), BE = I->end(); BI != BE;) {
Instruction *Insn = BI; ++BI;
DbgValueInst *DVI = dyn_cast<DbgValueInst>(Insn);
- if (!DVI) {
+ // Leave dbg.values that refer to an alloca alone. These
+ // instrinsics describe the address of a variable (= the alloca)
+ // being taken. They should not be moved next to the alloca
+ // (and to the beginning of the scope), but rather stay close to
+ // where said address is used.
+ if (!DVI || (DVI->getValue() && isa<AllocaInst>(DVI->getValue()))) {
PrevNonDbgInst = Insn;
continue;
}
}
return MadeChange;
}
+
+// If there is a sequence that branches based on comparing a single bit
+// against zero that can be combined into a single instruction, and the
+// target supports folding these into a single instruction, sink the
+// mask and compare into the branch uses. Do this before OptimizeBlock ->
+// OptimizeInst -> OptimizeCmpExpression, which perturbs the pattern being
+// searched for.
+bool CodeGenPrepare::sinkAndCmp(Function &F) {
+ if (!EnableAndCmpSinking)
+ return false;
+ if (!TLI || !TLI->isMaskAndBranchFoldingLegal())
+ return false;
+ bool MadeChange = false;
+ for (Function::iterator I = F.begin(), E = F.end(); I != E; ) {
+ BasicBlock *BB = I++;
+
+ // Does this BB end with the following?
+ // %andVal = and %val, #single-bit-set
+ // %icmpVal = icmp %andResult, 0
+ // br i1 %cmpVal label %dest1, label %dest2"
+ BranchInst *Brcc = dyn_cast<BranchInst>(BB->getTerminator());
+ if (!Brcc || !Brcc->isConditional())
+ continue;
+ ICmpInst *Cmp = dyn_cast<ICmpInst>(Brcc->getOperand(0));
+ if (!Cmp || Cmp->getParent() != BB)
+ continue;
+ ConstantInt *Zero = dyn_cast<ConstantInt>(Cmp->getOperand(1));
+ if (!Zero || !Zero->isZero())
+ continue;
+ Instruction *And = dyn_cast<Instruction>(Cmp->getOperand(0));
+ if (!And || And->getOpcode() != Instruction::And || And->getParent() != BB)
+ continue;
+ ConstantInt* Mask = dyn_cast<ConstantInt>(And->getOperand(1));
+ if (!Mask || !Mask->getUniqueInteger().isPowerOf2())
+ continue;
+ DEBUG(dbgs() << "found and; icmp ?,0; brcc\n"); DEBUG(BB->dump());
+
+ // Push the "and; icmp" for any users that are conditional branches.
+ // Since there can only be one branch use per BB, we don't need to keep
+ // track of which BBs we insert into.
+ for (Value::use_iterator UI = Cmp->use_begin(), E = Cmp->use_end();
+ UI != E; ) {
+ Use &TheUse = *UI;
+ // Find brcc use.
+ BranchInst *BrccUser = dyn_cast<BranchInst>(*UI);
+ ++UI;
+ if (!BrccUser || !BrccUser->isConditional())
+ continue;
+ BasicBlock *UserBB = BrccUser->getParent();
+ if (UserBB == BB) continue;
+ DEBUG(dbgs() << "found Brcc use\n");
+
+ // Sink the "and; icmp" to use.
+ MadeChange = true;
+ BinaryOperator *NewAnd =
+ BinaryOperator::CreateAnd(And->getOperand(0), And->getOperand(1), "",
+ BrccUser);
+ CmpInst *NewCmp =
+ CmpInst::Create(Cmp->getOpcode(), Cmp->getPredicate(), NewAnd, Zero,
+ "", BrccUser);
+ TheUse = NewCmp;
+ ++NumAndCmpsMoved;
+ DEBUG(BrccUser->getParent()->dump());
+ }
+ }
+ return MadeChange;
+}
projects / oota-llvm.git / blobdiff