X-Git-Url: http://demsky.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTransforms%2FScalar%2FLoopStrengthReduce.cpp;h=14274296de7da2bebf3a69d6370dcf64203a639b;hb=7d9f2b93a356aa89186522bd61c5c565718ff555;hp=a5611ff11311bb0b216fc628bd5211cd8d293ab1;hpb=466f37befba81bea849a08f21d0a80fb6070ab8f;p=oota-llvm.git diff --git a/lib/Transforms/Scalar/LoopStrengthReduce.cpp b/lib/Transforms/Scalar/LoopStrengthReduce.cpp index a5611ff1131..14274296de7 100644 --- a/lib/Transforms/Scalar/LoopStrengthReduce.cpp +++ b/lib/Transforms/Scalar/LoopStrengthReduce.cpp @@ -17,6 +17,40 @@ // available on the target, and it performs a variety of other optimizations // related to loop induction variables. // +// Terminology note: this code has a lot of handling for "post-increment" or +// "post-inc" users. This is not talking about post-increment addressing modes; +// it is instead talking about code like this: +// +// %i = phi [ 0, %entry ], [ %i.next, %latch ] +// ... +// %i.next = add %i, 1 +// %c = icmp eq %i.next, %n +// +// The SCEV for %i is {0,+,1}<%L>. The SCEV for %i.next is {1,+,1}<%L>, however +// it's useful to think about these as the same register, with some uses using +// the value of the register before the add and some using // it after. In this +// example, the icmp is a post-increment user, since it uses %i.next, which is +// the value of the induction variable after the increment. The other common +// case of post-increment users is users outside the loop. +// +// TODO: More sophistication in the way Formulae are generated and filtered. +// +// TODO: Handle multiple loops at a time. +// +// TODO: Should TargetLowering::AddrMode::BaseGV be changed to a ConstantExpr +// instead of a GlobalValue? +// +// TODO: When truncation is free, truncate ICmp users' operands to make it a +// smaller encoding (on x86 at least). +// +// TODO: When a negated register is used by an add (such as in a list of +// multiple base registers, or as the increment expression in an addrec), +// we may not actually need both reg and (-1 * reg) in registers; the +// negation can be implemented by using a sub instead of an add. The +// lack of support for taking this into consideration when making +// register pressure decisions is partly worked around by the "Special" +// use kind. +// //===----------------------------------------------------------------------===// #define DEBUG_TYPE "loop-reduce" @@ -26,208 +60,434 @@ #include "llvm/IntrinsicInst.h" #include "llvm/DerivedTypes.h" #include "llvm/Analysis/IVUsers.h" +#include "llvm/Analysis/Dominators.h" #include "llvm/Analysis/LoopPass.h" #include "llvm/Analysis/ScalarEvolutionExpander.h" -#include "llvm/Transforms/Utils/AddrModeMatcher.h" #include "llvm/Transforms/Utils/BasicBlockUtils.h" #include "llvm/Transforms/Utils/Local.h" -#include "llvm/ADT/Statistic.h" +#include "llvm/ADT/SmallBitVector.h" +#include "llvm/ADT/SetVector.h" +#include "llvm/ADT/DenseSet.h" #include "llvm/Support/Debug.h" -#include "llvm/Support/CommandLine.h" #include "llvm/Support/ValueHandle.h" #include "llvm/Support/raw_ostream.h" #include "llvm/Target/TargetLowering.h" #include using namespace llvm; -STATISTIC(NumReduced , "Number of IV uses strength reduced"); -STATISTIC(NumInserted, "Number of PHIs inserted"); -STATISTIC(NumVariable, "Number of PHIs with variable strides"); -STATISTIC(NumEliminated, "Number of strides eliminated"); -STATISTIC(NumShadow, "Number of Shadow IVs optimized"); -STATISTIC(NumImmSunk, "Number of common expr immediates sunk into uses"); -STATISTIC(NumLoopCond, "Number of loop terminating conds optimized"); -STATISTIC(NumCountZero, "Number of count iv optimized to count toward zero"); +namespace { + +/// RegSortData - This class holds data which is used to order reuse candidates. +class RegSortData { +public: + /// UsedByIndices - This represents the set of LSRUse indices which reference + /// a particular register. + SmallBitVector UsedByIndices; + + RegSortData() {} + + void print(raw_ostream &OS) const; + void dump() const; +}; + +} + +void RegSortData::print(raw_ostream &OS) const { + OS << "[NumUses=" << UsedByIndices.count() << ']'; +} -static cl::opt EnableFullLSRMode("enable-full-lsr", - cl::init(false), - cl::Hidden); +void RegSortData::dump() const { + print(errs()); errs() << '\n'; +} namespace { - struct BasedUser; +/// RegUseTracker - Map register candidates to information about how they are +/// used. +class RegUseTracker { + typedef DenseMap RegUsesTy; - /// IVInfo - This structure keeps track of one IV expression inserted during - /// StrengthReduceStridedIVUsers. It contains the stride, the common base, as - /// well as the PHI node and increment value created for rewrite. - struct IVExpr { - const SCEV *Stride; - const SCEV *Base; - PHINode *PHI; + RegUsesTy RegUses; + SmallVector RegSequence; - IVExpr(const SCEV *const stride, const SCEV *const base, PHINode *phi) - : Stride(stride), Base(base), PHI(phi) {} - }; +public: + void CountRegister(const SCEV *Reg, size_t LUIdx); + + bool isRegUsedByUsesOtherThan(const SCEV *Reg, size_t LUIdx) const; + + const SmallBitVector &getUsedByIndices(const SCEV *Reg) const; + + void clear(); + + typedef SmallVectorImpl::iterator iterator; + typedef SmallVectorImpl::const_iterator const_iterator; + iterator begin() { return RegSequence.begin(); } + iterator end() { return RegSequence.end(); } + const_iterator begin() const { return RegSequence.begin(); } + const_iterator end() const { return RegSequence.end(); } +}; + +} + +void +RegUseTracker::CountRegister(const SCEV *Reg, size_t LUIdx) { + std::pair Pair = + RegUses.insert(std::make_pair(Reg, RegSortData())); + RegSortData &RSD = Pair.first->second; + if (Pair.second) + RegSequence.push_back(Reg); + RSD.UsedByIndices.resize(std::max(RSD.UsedByIndices.size(), LUIdx + 1)); + RSD.UsedByIndices.set(LUIdx); +} + +bool +RegUseTracker::isRegUsedByUsesOtherThan(const SCEV *Reg, size_t LUIdx) const { + if (!RegUses.count(Reg)) return false; + const SmallBitVector &UsedByIndices = + RegUses.find(Reg)->second.UsedByIndices; + int i = UsedByIndices.find_first(); + if (i == -1) return false; + if ((size_t)i != LUIdx) return true; + return UsedByIndices.find_next(i) != -1; +} + +const SmallBitVector &RegUseTracker::getUsedByIndices(const SCEV *Reg) const { + RegUsesTy::const_iterator I = RegUses.find(Reg); + assert(I != RegUses.end() && "Unknown register!"); + return I->second.UsedByIndices; +} + +void RegUseTracker::clear() { + RegUses.clear(); + RegSequence.clear(); +} + +namespace { + +/// Formula - This class holds information that describes a formula for +/// computing satisfying a use. It may include broken-out immediates and scaled +/// registers. +struct Formula { + /// AM - This is used to represent complex addressing, as well as other kinds + /// of interesting uses. + TargetLowering::AddrMode AM; + + /// BaseRegs - The list of "base" registers for this use. When this is + /// non-empty, AM.HasBaseReg should be set to true. + SmallVector BaseRegs; + + /// ScaledReg - The 'scaled' register for this use. This should be non-null + /// when AM.Scale is not zero. + const SCEV *ScaledReg; + + Formula() : ScaledReg(0) {} + + void InitialMatch(const SCEV *S, Loop *L, + ScalarEvolution &SE, DominatorTree &DT); + + unsigned getNumRegs() const; + const Type *getType() const; + + bool referencesReg(const SCEV *S) const; + bool hasRegsUsedByUsesOtherThan(size_t LUIdx, + const RegUseTracker &RegUses) const; + + void print(raw_ostream &OS) const; + void dump() const; +}; + +} + +/// DoInitialMatch - Recursion helper for InitialMatch. +static void DoInitialMatch(const SCEV *S, Loop *L, + SmallVectorImpl &Good, + SmallVectorImpl &Bad, + ScalarEvolution &SE, DominatorTree &DT) { + // Collect expressions which properly dominate the loop header. + if (S->properlyDominates(L->getHeader(), &DT)) { + Good.push_back(S); + return; + } - /// IVsOfOneStride - This structure keeps track of all IV expression inserted - /// during StrengthReduceStridedIVUsers for a particular stride of the IV. - struct IVsOfOneStride { - std::vector IVs; + // Look at add operands. + if (const SCEVAddExpr *Add = dyn_cast(S)) { + for (SCEVAddExpr::op_iterator I = Add->op_begin(), E = Add->op_end(); + I != E; ++I) + DoInitialMatch(*I, L, Good, Bad, SE, DT); + return; + } - void addIV(const SCEV *const Stride, const SCEV *const Base, PHINode *PHI) { - IVs.push_back(IVExpr(Stride, Base, PHI)); + // Look at addrec operands. + if (const SCEVAddRecExpr *AR = dyn_cast(S)) + if (!AR->getStart()->isZero()) { + DoInitialMatch(AR->getStart(), L, Good, Bad, SE, DT); + DoInitialMatch(SE.getAddRecExpr(SE.getIntegerSCEV(0, AR->getType()), + AR->getStepRecurrence(SE), + AR->getLoop()), + L, Good, Bad, SE, DT); + return; } - }; - class LoopStrengthReduce : public LoopPass { - IVUsers *IU; - ScalarEvolution *SE; - bool Changed; - - /// IVsByStride - Keep track of all IVs that have been inserted for a - /// particular stride. - std::map IVsByStride; - - /// DeadInsts - Keep track of instructions we may have made dead, so that - /// we can remove them after we are done working. - SmallVector DeadInsts; - - /// TLI - Keep a pointer of a TargetLowering to consult for determining - /// transformation profitability. - const TargetLowering *TLI; - - public: - static char ID; // Pass ID, replacement for typeid - explicit LoopStrengthReduce(const TargetLowering *tli = NULL) : - LoopPass(&ID), TLI(tli) {} - - bool runOnLoop(Loop *L, LPPassManager &LPM); - - virtual void getAnalysisUsage(AnalysisUsage &AU) const { - // We split critical edges, so we change the CFG. However, we do update - // many analyses if they are around. - AU.addPreservedID(LoopSimplifyID); - AU.addPreserved("loops"); - AU.addPreserved("domfrontier"); - AU.addPreserved("domtree"); - - AU.addRequiredID(LoopSimplifyID); - AU.addRequired(); - AU.addPreserved(); - AU.addRequired(); - AU.addPreserved(); + // Handle a multiplication by -1 (negation) if it didn't fold. + if (const SCEVMulExpr *Mul = dyn_cast(S)) + if (Mul->getOperand(0)->isAllOnesValue()) { + SmallVector Ops(Mul->op_begin()+1, Mul->op_end()); + const SCEV *NewMul = SE.getMulExpr(Ops); + + SmallVector MyGood; + SmallVector MyBad; + DoInitialMatch(NewMul, L, MyGood, MyBad, SE, DT); + const SCEV *NegOne = SE.getSCEV(ConstantInt::getAllOnesValue( + SE.getEffectiveSCEVType(NewMul->getType()))); + for (SmallVectorImpl::const_iterator I = MyGood.begin(), + E = MyGood.end(); I != E; ++I) + Good.push_back(SE.getMulExpr(NegOne, *I)); + for (SmallVectorImpl::const_iterator I = MyBad.begin(), + E = MyBad.end(); I != E; ++I) + Bad.push_back(SE.getMulExpr(NegOne, *I)); + return; } - private: - void OptimizeIndvars(Loop *L); - - /// OptimizeLoopTermCond - Change loop terminating condition to use the - /// postinc iv when possible. - void OptimizeLoopTermCond(Loop *L); - - /// OptimizeShadowIV - If IV is used in a int-to-float cast - /// inside the loop then try to eliminate the cast opeation. - void OptimizeShadowIV(Loop *L); - - /// OptimizeMax - Rewrite the loop's terminating condition - /// if it uses a max computation. - ICmpInst *OptimizeMax(Loop *L, ICmpInst *Cond, - IVStrideUse* &CondUse); - - /// OptimizeLoopCountIV - If, after all sharing of IVs, the IV used for - /// deciding when to exit the loop is used only for that purpose, try to - /// rearrange things so it counts down to a test against zero. - bool OptimizeLoopCountIV(Loop *L); - bool OptimizeLoopCountIVOfStride(const SCEV* &Stride, - IVStrideUse* &CondUse, Loop *L); - - /// StrengthReduceIVUsersOfStride - Strength reduce all of the users of a - /// single stride of IV. All of the users may have different starting - /// values, and this may not be the only stride. - void StrengthReduceIVUsersOfStride(const SCEV *Stride, - IVUsersOfOneStride &Uses, - Loop *L); - void StrengthReduceIVUsers(Loop *L); - - ICmpInst *ChangeCompareStride(Loop *L, ICmpInst *Cond, - IVStrideUse* &CondUse, - const SCEV* &CondStride, - bool PostPass = false); - - bool FindIVUserForCond(ICmpInst *Cond, IVStrideUse *&CondUse, - const SCEV* &CondStride); - bool RequiresTypeConversion(const Type *Ty, const Type *NewTy); - const SCEV *CheckForIVReuse(bool, bool, bool, const SCEV *, - IVExpr&, const Type*, - const std::vector& UsersToProcess); - bool ValidScale(bool, int64_t, - const std::vector& UsersToProcess); - bool ValidOffset(bool, int64_t, int64_t, - const std::vector& UsersToProcess); - const SCEV *CollectIVUsers(const SCEV *Stride, - IVUsersOfOneStride &Uses, - Loop *L, - bool &AllUsesAreAddresses, - bool &AllUsesAreOutsideLoop, - std::vector &UsersToProcess); - bool StrideMightBeShared(const SCEV *Stride, Loop *L, bool CheckPreInc); - bool ShouldUseFullStrengthReductionMode( - const std::vector &UsersToProcess, - const Loop *L, - bool AllUsesAreAddresses, - const SCEV *Stride); - void PrepareToStrengthReduceFully( - std::vector &UsersToProcess, - const SCEV *Stride, - const SCEV *CommonExprs, - const Loop *L, - SCEVExpander &PreheaderRewriter); - void PrepareToStrengthReduceFromSmallerStride( - std::vector &UsersToProcess, - Value *CommonBaseV, - const IVExpr &ReuseIV, - Instruction *PreInsertPt); - void PrepareToStrengthReduceWithNewPhi( - std::vector &UsersToProcess, - const SCEV *Stride, - const SCEV *CommonExprs, - Value *CommonBaseV, - Instruction *IVIncInsertPt, - const Loop *L, - SCEVExpander &PreheaderRewriter); - - void DeleteTriviallyDeadInstructions(); - }; + // Ok, we can't do anything interesting. Just stuff the whole thing into a + // register and hope for the best. + Bad.push_back(S); } -char LoopStrengthReduce::ID = 0; -static RegisterPass -X("loop-reduce", "Loop Strength Reduction"); +/// InitialMatch - Incorporate loop-variant parts of S into this Formula, +/// attempting to keep all loop-invariant and loop-computable values in a +/// single base register. +void Formula::InitialMatch(const SCEV *S, Loop *L, + ScalarEvolution &SE, DominatorTree &DT) { + SmallVector Good; + SmallVector Bad; + DoInitialMatch(S, L, Good, Bad, SE, DT); + if (!Good.empty()) { + BaseRegs.push_back(SE.getAddExpr(Good)); + AM.HasBaseReg = true; + } + if (!Bad.empty()) { + BaseRegs.push_back(SE.getAddExpr(Bad)); + AM.HasBaseReg = true; + } +} -Pass *llvm::createLoopStrengthReducePass(const TargetLowering *TLI) { - return new LoopStrengthReduce(TLI); +/// getNumRegs - Return the total number of register operands used by this +/// formula. This does not include register uses implied by non-constant +/// addrec strides. +unsigned Formula::getNumRegs() const { + return !!ScaledReg + BaseRegs.size(); } -/// DeleteTriviallyDeadInstructions - If any of the instructions is the -/// specified set are trivially dead, delete them and see if this makes any of -/// their operands subsequently dead. -void LoopStrengthReduce::DeleteTriviallyDeadInstructions() { - while (!DeadInsts.empty()) { - Instruction *I = dyn_cast_or_null(DeadInsts.pop_back_val()); +/// getType - Return the type of this formula, if it has one, or null +/// otherwise. This type is meaningless except for the bit size. +const Type *Formula::getType() const { + return !BaseRegs.empty() ? BaseRegs.front()->getType() : + ScaledReg ? ScaledReg->getType() : + AM.BaseGV ? AM.BaseGV->getType() : + 0; +} - if (I == 0 || !isInstructionTriviallyDead(I)) - continue; +/// referencesReg - Test if this formula references the given register. +bool Formula::referencesReg(const SCEV *S) const { + return S == ScaledReg || + std::find(BaseRegs.begin(), BaseRegs.end(), S) != BaseRegs.end(); +} - for (User::op_iterator OI = I->op_begin(), E = I->op_end(); OI != E; ++OI) - if (Instruction *U = dyn_cast(*OI)) { - *OI = 0; - if (U->use_empty()) - DeadInsts.push_back(U); +/// hasRegsUsedByUsesOtherThan - Test whether this formula uses registers +/// which are used by uses other than the use with the given index. +bool Formula::hasRegsUsedByUsesOtherThan(size_t LUIdx, + const RegUseTracker &RegUses) const { + if (ScaledReg) + if (RegUses.isRegUsedByUsesOtherThan(ScaledReg, LUIdx)) + return true; + for (SmallVectorImpl::const_iterator I = BaseRegs.begin(), + E = BaseRegs.end(); I != E; ++I) + if (RegUses.isRegUsedByUsesOtherThan(*I, LUIdx)) + return true; + return false; +} + +void Formula::print(raw_ostream &OS) const { + bool First = true; + if (AM.BaseGV) { + if (!First) OS << " + "; else First = false; + WriteAsOperand(OS, AM.BaseGV, /*PrintType=*/false); + } + if (AM.BaseOffs != 0) { + if (!First) OS << " + "; else First = false; + OS << AM.BaseOffs; + } + for (SmallVectorImpl::const_iterator I = BaseRegs.begin(), + E = BaseRegs.end(); I != E; ++I) { + if (!First) OS << " + "; else First = false; + OS << "reg(" << **I << ')'; + } + if (AM.Scale != 0) { + if (!First) OS << " + "; else First = false; + OS << AM.Scale << "*reg("; + if (ScaledReg) + OS << *ScaledReg; + else + OS << ""; + OS << ')'; + } +} + +void Formula::dump() const { + print(errs()); errs() << '\n'; +} + +/// isAddRecSExtable - Return true if the given addrec can be sign-extended +/// without changing its value. +static bool isAddRecSExtable(const SCEVAddRecExpr *AR, ScalarEvolution &SE) { + const Type *WideTy = + IntegerType::get(SE.getContext(), + SE.getTypeSizeInBits(AR->getType()) + 1); + return isa(SE.getSignExtendExpr(AR, WideTy)); +} + +/// isAddSExtable - Return true if the given add can be sign-extended +/// without changing its value. +static bool isAddSExtable(const SCEVAddExpr *A, ScalarEvolution &SE) { + const Type *WideTy = + IntegerType::get(SE.getContext(), + SE.getTypeSizeInBits(A->getType()) + 1); + return isa(SE.getSignExtendExpr(A, WideTy)); +} + +/// isMulSExtable - Return true if the given add can be sign-extended +/// without changing its value. +static bool isMulSExtable(const SCEVMulExpr *A, ScalarEvolution &SE) { + const Type *WideTy = + IntegerType::get(SE.getContext(), + SE.getTypeSizeInBits(A->getType()) + 1); + return isa(SE.getSignExtendExpr(A, WideTy)); +} + +/// getExactSDiv - Return an expression for LHS /s RHS, if it can be determined +/// and if the remainder is known to be zero, or null otherwise. If +/// IgnoreSignificantBits is true, expressions like (X * Y) /s Y are simplified +/// to Y, ignoring that the multiplication may overflow, which is useful when +/// the result will be used in a context where the most significant bits are +/// ignored. +static const SCEV *getExactSDiv(const SCEV *LHS, const SCEV *RHS, + ScalarEvolution &SE, + bool IgnoreSignificantBits = false) { + // Handle the trivial case, which works for any SCEV type. + if (LHS == RHS) + return SE.getIntegerSCEV(1, LHS->getType()); + + // Handle x /s -1 as x * -1, to give ScalarEvolution a chance to do some + // folding. + if (RHS->isAllOnesValue()) + return SE.getMulExpr(LHS, RHS); + + // Check for a division of a constant by a constant. + if (const SCEVConstant *C = dyn_cast(LHS)) { + const SCEVConstant *RC = dyn_cast(RHS); + if (!RC) + return 0; + if (C->getValue()->getValue().srem(RC->getValue()->getValue()) != 0) + return 0; + return SE.getConstant(C->getValue()->getValue() + .sdiv(RC->getValue()->getValue())); + } + + // Distribute the sdiv over addrec operands, if the addrec doesn't overflow. + if (const SCEVAddRecExpr *AR = dyn_cast(LHS)) { + if (IgnoreSignificantBits || isAddRecSExtable(AR, SE)) { + const SCEV *Start = getExactSDiv(AR->getStart(), RHS, SE, + IgnoreSignificantBits); + if (!Start) return 0; + const SCEV *Step = getExactSDiv(AR->getStepRecurrence(SE), RHS, SE, + IgnoreSignificantBits); + if (!Step) return 0; + return SE.getAddRecExpr(Start, Step, AR->getLoop()); + } + } + + // Distribute the sdiv over add operands, if the add doesn't overflow. + if (const SCEVAddExpr *Add = dyn_cast(LHS)) { + if (IgnoreSignificantBits || isAddSExtable(Add, SE)) { + SmallVector Ops; + for (SCEVAddExpr::op_iterator I = Add->op_begin(), E = Add->op_end(); + I != E; ++I) { + const SCEV *Op = getExactSDiv(*I, RHS, SE, + IgnoreSignificantBits); + if (!Op) return 0; + Ops.push_back(Op); } + return SE.getAddExpr(Ops); + } + } - I->eraseFromParent(); - Changed = true; + // Check for a multiply operand that we can pull RHS out of. + if (const SCEVMulExpr *Mul = dyn_cast(LHS)) + if (IgnoreSignificantBits || isMulSExtable(Mul, SE)) { + SmallVector Ops; + bool Found = false; + for (SCEVMulExpr::op_iterator I = Mul->op_begin(), E = Mul->op_end(); + I != E; ++I) { + if (!Found) + if (const SCEV *Q = getExactSDiv(*I, RHS, SE, + IgnoreSignificantBits)) { + Ops.push_back(Q); + Found = true; + continue; + } + Ops.push_back(*I); + } + return Found ? SE.getMulExpr(Ops) : 0; + } + + // Otherwise we don't know. + return 0; +} + +/// ExtractImmediate - If S involves the addition of a constant integer value, +/// return that integer value, and mutate S to point to a new SCEV with that +/// value excluded. +static int64_t ExtractImmediate(const SCEV *&S, ScalarEvolution &SE) { + if (const SCEVConstant *C = dyn_cast(S)) { + if (C->getValue()->getValue().getMinSignedBits() <= 64) { + S = SE.getIntegerSCEV(0, C->getType()); + return C->getValue()->getSExtValue(); + } + } else if (const SCEVAddExpr *Add = dyn_cast(S)) { + SmallVector NewOps(Add->op_begin(), Add->op_end()); + int64_t Result = ExtractImmediate(NewOps.front(), SE); + S = SE.getAddExpr(NewOps); + return Result; + } else if (const SCEVAddRecExpr *AR = dyn_cast(S)) { + SmallVector NewOps(AR->op_begin(), AR->op_end()); + int64_t Result = ExtractImmediate(NewOps.front(), SE); + S = SE.getAddRecExpr(NewOps, AR->getLoop()); + return Result; } + return 0; +} + +/// ExtractSymbol - If S involves the addition of a GlobalValue address, +/// return that symbol, and mutate S to point to a new SCEV with that +/// value excluded. +static GlobalValue *ExtractSymbol(const SCEV *&S, ScalarEvolution &SE) { + if (const SCEVUnknown *U = dyn_cast(S)) { + if (GlobalValue *GV = dyn_cast(U->getValue())) { + S = SE.getIntegerSCEV(0, GV->getType()); + return GV; + } + } else if (const SCEVAddExpr *Add = dyn_cast(S)) { + SmallVector NewOps(Add->op_begin(), Add->op_end()); + GlobalValue *Result = ExtractSymbol(NewOps.back(), SE); + S = SE.getAddExpr(NewOps); + return Result; + } else if (const SCEVAddRecExpr *AR = dyn_cast(S)) { + SmallVector NewOps(AR->op_begin(), AR->op_end()); + GlobalValue *Result = ExtractSymbol(NewOps.front(), SE); + S = SE.getAddRecExpr(NewOps, AR->getLoop()); + return Result; + } + return 0; } /// isAddressUse - Returns true if the specified instruction is using the @@ -276,1776 +536,833 @@ static const Type *getAccessType(const Instruction *Inst) { break; } } + + // All pointers have the same requirements, so canonicalize them to an + // arbitrary pointer type to minimize variation. + if (const PointerType *PTy = dyn_cast(AccessTy)) + AccessTy = PointerType::get(IntegerType::get(PTy->getContext(), 1), + PTy->getAddressSpace()); + return AccessTy; } -namespace { - /// BasedUser - For a particular base value, keep information about how we've - /// partitioned the expression so far. - struct BasedUser { - /// Base - The Base value for the PHI node that needs to be inserted for - /// this use. As the use is processed, information gets moved from this - /// field to the Imm field (below). BasedUser values are sorted by this - /// field. - const SCEV *Base; - - /// Inst - The instruction using the induction variable. - Instruction *Inst; - - /// OperandValToReplace - The operand value of Inst to replace with the - /// EmittedBase. - Value *OperandValToReplace; - - /// Imm - The immediate value that should be added to the base immediately - /// before Inst, because it will be folded into the imm field of the - /// instruction. This is also sometimes used for loop-variant values that - /// must be added inside the loop. - const SCEV *Imm; - - /// Phi - The induction variable that performs the striding that - /// should be used for this user. - PHINode *Phi; - - // isUseOfPostIncrementedValue - True if this should use the - // post-incremented version of this IV, not the preincremented version. - // This can only be set in special cases, such as the terminating setcc - // instruction for a loop and uses outside the loop that are dominated by - // the loop. - bool isUseOfPostIncrementedValue; - - BasedUser(IVStrideUse &IVSU, ScalarEvolution *se) - : Base(IVSU.getOffset()), Inst(IVSU.getUser()), - OperandValToReplace(IVSU.getOperandValToReplace()), - Imm(se->getIntegerSCEV(0, Base->getType())), - isUseOfPostIncrementedValue(IVSU.isUseOfPostIncrementedValue()) {} - - // Once we rewrite the code to insert the new IVs we want, update the - // operands of Inst to use the new expression 'NewBase', with 'Imm' added - // to it. - void RewriteInstructionToUseNewBase(const SCEV *NewBase, - Instruction *InsertPt, - SCEVExpander &Rewriter, Loop *L, Pass *P, - SmallVectorImpl &DeadInsts, - ScalarEvolution *SE); - - Value *InsertCodeForBaseAtPosition(const SCEV *NewBase, - const Type *Ty, - SCEVExpander &Rewriter, - Instruction *IP, - ScalarEvolution *SE); - void dump() const; - }; -} +/// DeleteTriviallyDeadInstructions - If any of the instructions is the +/// specified set are trivially dead, delete them and see if this makes any of +/// their operands subsequently dead. +static bool +DeleteTriviallyDeadInstructions(SmallVectorImpl &DeadInsts) { + bool Changed = false; -void BasedUser::dump() const { - dbgs() << " Base=" << *Base; - dbgs() << " Imm=" << *Imm; - dbgs() << " Inst: " << *Inst; -} + while (!DeadInsts.empty()) { + Instruction *I = dyn_cast_or_null(DeadInsts.pop_back_val()); -Value *BasedUser::InsertCodeForBaseAtPosition(const SCEV *NewBase, - const Type *Ty, - SCEVExpander &Rewriter, - Instruction *IP, - ScalarEvolution *SE) { - Value *Base = Rewriter.expandCodeFor(NewBase, 0, IP); + if (I == 0 || !isInstructionTriviallyDead(I)) + continue; - // Wrap the base in a SCEVUnknown so that ScalarEvolution doesn't try to - // re-analyze it. - const SCEV *NewValSCEV = SE->getUnknown(Base); + for (User::op_iterator OI = I->op_begin(), E = I->op_end(); OI != E; ++OI) + if (Instruction *U = dyn_cast(*OI)) { + *OI = 0; + if (U->use_empty()) + DeadInsts.push_back(U); + } - // Always emit the immediate into the same block as the user. - NewValSCEV = SE->getAddExpr(NewValSCEV, Imm); + I->eraseFromParent(); + Changed = true; + } - return Rewriter.expandCodeFor(NewValSCEV, Ty, IP); + return Changed; } +namespace { -// Once we rewrite the code to insert the new IVs we want, update the -// operands of Inst to use the new expression 'NewBase', with 'Imm' added -// to it. NewBasePt is the last instruction which contributes to the -// value of NewBase in the case that it's a diffferent instruction from -// the PHI that NewBase is computed from, or null otherwise. -// -void BasedUser::RewriteInstructionToUseNewBase(const SCEV *NewBase, - Instruction *NewBasePt, - SCEVExpander &Rewriter, Loop *L, Pass *P, - SmallVectorImpl &DeadInsts, - ScalarEvolution *SE) { - if (!isa(Inst)) { - // By default, insert code at the user instruction. - BasicBlock::iterator InsertPt = Inst; - - // However, if the Operand is itself an instruction, the (potentially - // complex) inserted code may be shared by many users. Because of this, we - // want to emit code for the computation of the operand right before its old - // computation. This is usually safe, because we obviously used to use the - // computation when it was computed in its current block. However, in some - // cases (e.g. use of a post-incremented induction variable) the NewBase - // value will be pinned to live somewhere after the original computation. - // In this case, we have to back off. - // - // If this is a use outside the loop (which means after, since it is based - // on a loop indvar) we use the post-incremented value, so that we don't - // artificially make the preinc value live out the bottom of the loop. - if (!isUseOfPostIncrementedValue && L->contains(Inst)) { - if (NewBasePt && isa(OperandValToReplace)) { - InsertPt = NewBasePt; - ++InsertPt; - } else if (Instruction *OpInst - = dyn_cast(OperandValToReplace)) { - InsertPt = OpInst; - while (isa(InsertPt)) ++InsertPt; - } - } - Value *NewVal = InsertCodeForBaseAtPosition(NewBase, - OperandValToReplace->getType(), - Rewriter, InsertPt, SE); - // Replace the use of the operand Value with the new Phi we just created. - Inst->replaceUsesOfWith(OperandValToReplace, NewVal); - - DEBUG(dbgs() << " Replacing with "); - DEBUG(WriteAsOperand(dbgs(), NewVal, /*PrintType=*/false)); - DEBUG(dbgs() << ", which has value " << *NewBase << " plus IMM " - << *Imm << "\n"); - return; - } +/// Cost - This class is used to measure and compare candidate formulae. +class Cost { + /// TODO: Some of these could be merged. Also, a lexical ordering + /// isn't always optimal. + unsigned NumRegs; + unsigned AddRecCost; + unsigned NumIVMuls; + unsigned NumBaseAdds; + unsigned ImmCost; + unsigned SetupCost; + +public: + Cost() + : NumRegs(0), AddRecCost(0), NumIVMuls(0), NumBaseAdds(0), ImmCost(0), + SetupCost(0) {} + + unsigned getNumRegs() const { return NumRegs; } + + bool operator<(const Cost &Other) const; + + void Loose(); + + void RateFormula(const Formula &F, + SmallPtrSet &Regs, + const DenseSet &VisitedRegs, + const Loop *L, + const SmallVectorImpl &Offsets, + ScalarEvolution &SE, DominatorTree &DT); + + void print(raw_ostream &OS) const; + void dump() const; + +private: + void RateRegister(const SCEV *Reg, + SmallPtrSet &Regs, + const Loop *L, + ScalarEvolution &SE, DominatorTree &DT); + void RatePrimaryRegister(const SCEV *Reg, + SmallPtrSet &Regs, + const Loop *L, + ScalarEvolution &SE, DominatorTree &DT); +}; - // PHI nodes are more complex. We have to insert one copy of the NewBase+Imm - // expression into each operand block that uses it. Note that PHI nodes can - // have multiple entries for the same predecessor. We use a map to make sure - // that a PHI node only has a single Value* for each predecessor (which also - // prevents us from inserting duplicate code in some blocks). - DenseMap InsertedCode; - PHINode *PN = cast(Inst); - for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) { - if (PN->getIncomingValue(i) == OperandValToReplace) { - // If the original expression is outside the loop, put the replacement - // code in the same place as the original expression, - // which need not be an immediate predecessor of this PHI. This way we - // need only one copy of it even if it is referenced multiple times in - // the PHI. We don't do this when the original expression is inside the - // loop because multiple copies sometimes do useful sinking of code in - // that case(?). - Instruction *OldLoc = dyn_cast(OperandValToReplace); - BasicBlock *PHIPred = PN->getIncomingBlock(i); - if (L->contains(OldLoc)) { - // If this is a critical edge, split the edge so that we do not insert - // the code on all predecessor/successor paths. We do this unless this - // is the canonical backedge for this loop, as this can make some - // inserted code be in an illegal position. - if (e != 1 && PHIPred->getTerminator()->getNumSuccessors() > 1 && - !isa(PHIPred->getTerminator()) && - (PN->getParent() != L->getHeader() || !L->contains(PHIPred))) { - - // First step, split the critical edge. - BasicBlock *NewBB = SplitCriticalEdge(PHIPred, PN->getParent(), - P, false); - - // Next step: move the basic block. In particular, if the PHI node - // is outside of the loop, and PredTI is in the loop, we want to - // move the block to be immediately before the PHI block, not - // immediately after PredTI. - if (L->contains(PHIPred) && !L->contains(PN)) - NewBB->moveBefore(PN->getParent()); - - // Splitting the edge can reduce the number of PHI entries we have. - e = PN->getNumIncomingValues(); - PHIPred = NewBB; - i = PN->getBasicBlockIndex(PHIPred); - } - } - Value *&Code = InsertedCode[PHIPred]; - if (!Code) { - // Insert the code into the end of the predecessor block. - Instruction *InsertPt = (L->contains(OldLoc)) ? - PHIPred->getTerminator() : - OldLoc->getParent()->getTerminator(); - Code = InsertCodeForBaseAtPosition(NewBase, PN->getType(), - Rewriter, InsertPt, SE); - - DEBUG(dbgs() << " Changing PHI use to "); - DEBUG(WriteAsOperand(dbgs(), Code, /*PrintType=*/false)); - DEBUG(dbgs() << ", which has value " << *NewBase << " plus IMM " - << *Imm << "\n"); - } +} + +/// RateRegister - Tally up interesting quantities from the given register. +void Cost::RateRegister(const SCEV *Reg, + SmallPtrSet &Regs, + const Loop *L, + ScalarEvolution &SE, DominatorTree &DT) { + if (const SCEVAddRecExpr *AR = dyn_cast(Reg)) { + if (AR->getLoop() == L) + AddRecCost += 1; /// TODO: This should be a function of the stride. + + // If this is an addrec for a loop that's already been visited by LSR, + // don't second-guess its addrec phi nodes. LSR isn't currently smart + // enough to reason about more than one loop at a time. Consider these + // registers free and leave them alone. + else if (L->contains(AR->getLoop()) || + (!AR->getLoop()->contains(L) && + DT.dominates(L->getHeader(), AR->getLoop()->getHeader()))) { + for (BasicBlock::iterator I = AR->getLoop()->getHeader()->begin(); + PHINode *PN = dyn_cast(I); ++I) + if (SE.isSCEVable(PN->getType()) && + (SE.getEffectiveSCEVType(PN->getType()) == + SE.getEffectiveSCEVType(AR->getType())) && + SE.getSCEV(PN) == AR) + return; - // Replace the use of the operand Value with the new Phi we just created. - PN->setIncomingValue(i, Code); - Rewriter.clear(); + // If this isn't one of the addrecs that the loop already has, it + // would require a costly new phi and add. TODO: This isn't + // precisely modeled right now. + ++NumBaseAdds; + if (!Regs.count(AR->getStart())) + RateRegister(AR->getStart(), Regs, L, SE, DT); } - } - // PHI node might have become a constant value after SplitCriticalEdge. - DeadInsts.push_back(Inst); + // Add the step value register, if it needs one. + // TODO: The non-affine case isn't precisely modeled here. + if (!AR->isAffine() || !isa(AR->getOperand(1))) + if (!Regs.count(AR->getStart())) + RateRegister(AR->getOperand(1), Regs, L, SE, DT); + } + ++NumRegs; + + // Rough heuristic; favor registers which don't require extra setup + // instructions in the preheader. + if (!isa(Reg) && + !isa(Reg) && + !(isa(Reg) && + (isa(cast(Reg)->getStart()) || + isa(cast(Reg)->getStart())))) + ++SetupCost; } +/// RatePrimaryRegister - Record this register in the set. If we haven't seen it +/// before, rate it. +void Cost::RatePrimaryRegister(const SCEV *Reg, + SmallPtrSet &Regs, + const Loop *L, + ScalarEvolution &SE, DominatorTree &DT) { + if (Regs.insert(Reg)) + RateRegister(Reg, Regs, L, SE, DT); +} -/// fitsInAddressMode - Return true if V can be subsumed within an addressing -/// mode, and does not need to be put in a register first. -static bool fitsInAddressMode(const SCEV *V, const Type *AccessTy, - const TargetLowering *TLI, bool HasBaseReg) { - if (const SCEVConstant *SC = dyn_cast(V)) { - int64_t VC = SC->getValue()->getSExtValue(); - if (TLI) { - TargetLowering::AddrMode AM; - AM.BaseOffs = VC; - AM.HasBaseReg = HasBaseReg; - return TLI->isLegalAddressingMode(AM, AccessTy); - } else { - // Defaults to PPC. PPC allows a sign-extended 16-bit immediate field. - return (VC > -(1 << 16) && VC < (1 << 16)-1); +void Cost::RateFormula(const Formula &F, + SmallPtrSet &Regs, + const DenseSet &VisitedRegs, + const Loop *L, + const SmallVectorImpl &Offsets, + ScalarEvolution &SE, DominatorTree &DT) { + // Tally up the registers. + if (const SCEV *ScaledReg = F.ScaledReg) { + if (VisitedRegs.count(ScaledReg)) { + Loose(); + return; } + RatePrimaryRegister(ScaledReg, Regs, L, SE, DT); } - - if (const SCEVUnknown *SU = dyn_cast(V)) - if (GlobalValue *GV = dyn_cast(SU->getValue())) { - if (TLI) { - TargetLowering::AddrMode AM; - AM.BaseGV = GV; - AM.HasBaseReg = HasBaseReg; - return TLI->isLegalAddressingMode(AM, AccessTy); - } else { - // Default: assume global addresses are not legal. - } + for (SmallVectorImpl::const_iterator I = F.BaseRegs.begin(), + E = F.BaseRegs.end(); I != E; ++I) { + const SCEV *BaseReg = *I; + if (VisitedRegs.count(BaseReg)) { + Loose(); + return; } + RatePrimaryRegister(BaseReg, Regs, L, SE, DT); - return false; + NumIVMuls += isa(BaseReg) && + BaseReg->hasComputableLoopEvolution(L); + } + + if (F.BaseRegs.size() > 1) + NumBaseAdds += F.BaseRegs.size() - 1; + + // Tally up the non-zero immediates. + for (SmallVectorImpl::const_iterator I = Offsets.begin(), + E = Offsets.end(); I != E; ++I) { + int64_t Offset = (uint64_t)*I + F.AM.BaseOffs; + if (F.AM.BaseGV) + ImmCost += 64; // Handle symbolic values conservatively. + // TODO: This should probably be the pointer size. + else if (Offset != 0) + ImmCost += APInt(64, Offset, true).getMinSignedBits(); + } } -/// MoveLoopVariantsToImmediateField - Move any subexpressions from Val that are -/// loop varying to the Imm operand. -static void MoveLoopVariantsToImmediateField(const SCEV *&Val, const SCEV *&Imm, - Loop *L, ScalarEvolution *SE) { - if (Val->isLoopInvariant(L)) return; // Nothing to do. +/// Loose - Set this cost to a loosing value. +void Cost::Loose() { + NumRegs = ~0u; + AddRecCost = ~0u; + NumIVMuls = ~0u; + NumBaseAdds = ~0u; + ImmCost = ~0u; + SetupCost = ~0u; +} - if (const SCEVAddExpr *SAE = dyn_cast(Val)) { - SmallVector NewOps; - NewOps.reserve(SAE->getNumOperands()); +/// operator< - Choose the lower cost. +bool Cost::operator<(const Cost &Other) const { + if (NumRegs != Other.NumRegs) + return NumRegs < Other.NumRegs; + if (AddRecCost != Other.AddRecCost) + return AddRecCost < Other.AddRecCost; + if (NumIVMuls != Other.NumIVMuls) + return NumIVMuls < Other.NumIVMuls; + if (NumBaseAdds != Other.NumBaseAdds) + return NumBaseAdds < Other.NumBaseAdds; + if (ImmCost != Other.ImmCost) + return ImmCost < Other.ImmCost; + if (SetupCost != Other.SetupCost) + return SetupCost < Other.SetupCost; + return false; +} - for (unsigned i = 0; i != SAE->getNumOperands(); ++i) - if (!SAE->getOperand(i)->isLoopInvariant(L)) { - // If this is a loop-variant expression, it must stay in the immediate - // field of the expression. - Imm = SE->getAddExpr(Imm, SAE->getOperand(i)); - } else { - NewOps.push_back(SAE->getOperand(i)); - } +void Cost::print(raw_ostream &OS) const { + OS << NumRegs << " reg" << (NumRegs == 1 ? "" : "s"); + if (AddRecCost != 0) + OS << ", with addrec cost " << AddRecCost; + if (NumIVMuls != 0) + OS << ", plus " << NumIVMuls << " IV mul" << (NumIVMuls == 1 ? "" : "s"); + if (NumBaseAdds != 0) + OS << ", plus " << NumBaseAdds << " base add" + << (NumBaseAdds == 1 ? "" : "s"); + if (ImmCost != 0) + OS << ", plus " << ImmCost << " imm cost"; + if (SetupCost != 0) + OS << ", plus " << SetupCost << " setup cost"; +} - if (NewOps.empty()) - Val = SE->getIntegerSCEV(0, Val->getType()); - else - Val = SE->getAddExpr(NewOps); - } else if (const SCEVAddRecExpr *SARE = dyn_cast(Val)) { - // Try to pull immediates out of the start value of nested addrec's. - const SCEV *Start = SARE->getStart(); - MoveLoopVariantsToImmediateField(Start, Imm, L, SE); - - SmallVector Ops(SARE->op_begin(), SARE->op_end()); - Ops[0] = Start; - Val = SE->getAddRecExpr(Ops, SARE->getLoop()); - } else { - // Otherwise, all of Val is variant, move the whole thing over. - Imm = SE->getAddExpr(Imm, Val); - Val = SE->getIntegerSCEV(0, Val->getType()); - } +void Cost::dump() const { + print(errs()); errs() << '\n'; } +namespace { -/// MoveImmediateValues - Look at Val, and pull out any additions of constants -/// that can fit into the immediate field of instructions in the target. -/// Accumulate these immediate values into the Imm value. -static void MoveImmediateValues(const TargetLowering *TLI, - const Type *AccessTy, - const SCEV *&Val, const SCEV *&Imm, - bool isAddress, Loop *L, - ScalarEvolution *SE) { - if (const SCEVAddExpr *SAE = dyn_cast(Val)) { - SmallVector NewOps; - NewOps.reserve(SAE->getNumOperands()); +/// LSRFixup - An operand value in an instruction which is to be replaced +/// with some equivalent, possibly strength-reduced, replacement. +struct LSRFixup { + /// UserInst - The instruction which will be updated. + Instruction *UserInst; - for (unsigned i = 0; i != SAE->getNumOperands(); ++i) { - const SCEV *NewOp = SAE->getOperand(i); - MoveImmediateValues(TLI, AccessTy, NewOp, Imm, isAddress, L, SE); + /// OperandValToReplace - The operand of the instruction which will + /// be replaced. The operand may be used more than once; every instance + /// will be replaced. + Value *OperandValToReplace; - if (!NewOp->isLoopInvariant(L)) { - // If this is a loop-variant expression, it must stay in the immediate - // field of the expression. - Imm = SE->getAddExpr(Imm, NewOp); - } else { - NewOps.push_back(NewOp); - } - } + /// PostIncLoop - If this user is to use the post-incremented value of an + /// induction variable, this variable is non-null and holds the loop + /// associated with the induction variable. + const Loop *PostIncLoop; - if (NewOps.empty()) - Val = SE->getIntegerSCEV(0, Val->getType()); - else - Val = SE->getAddExpr(NewOps); - return; - } else if (const SCEVAddRecExpr *SARE = dyn_cast(Val)) { - // Try to pull immediates out of the start value of nested addrec's. - const SCEV *Start = SARE->getStart(); - MoveImmediateValues(TLI, AccessTy, Start, Imm, isAddress, L, SE); - - if (Start != SARE->getStart()) { - SmallVector Ops(SARE->op_begin(), SARE->op_end()); - Ops[0] = Start; - Val = SE->getAddRecExpr(Ops, SARE->getLoop()); - } - return; - } else if (const SCEVMulExpr *SME = dyn_cast(Val)) { - // Transform "8 * (4 + v)" -> "32 + 8*V" if "32" fits in the immed field. - if (isAddress && - fitsInAddressMode(SME->getOperand(0), AccessTy, TLI, false) && - SME->getNumOperands() == 2 && SME->isLoopInvariant(L)) { - - const SCEV *SubImm = SE->getIntegerSCEV(0, Val->getType()); - const SCEV *NewOp = SME->getOperand(1); - MoveImmediateValues(TLI, AccessTy, NewOp, SubImm, isAddress, L, SE); - - // If we extracted something out of the subexpressions, see if we can - // simplify this! - if (NewOp != SME->getOperand(1)) { - // Scale SubImm up by "8". If the result is a target constant, we are - // good. - SubImm = SE->getMulExpr(SubImm, SME->getOperand(0)); - if (fitsInAddressMode(SubImm, AccessTy, TLI, false)) { - // Accumulate the immediate. - Imm = SE->getAddExpr(Imm, SubImm); - - // Update what is left of 'Val'. - Val = SE->getMulExpr(SME->getOperand(0), NewOp); - return; - } - } - } - } + /// LUIdx - The index of the LSRUse describing the expression which + /// this fixup needs, minus an offset (below). + size_t LUIdx; - // Loop-variant expressions must stay in the immediate field of the - // expression. - if ((isAddress && fitsInAddressMode(Val, AccessTy, TLI, false)) || - !Val->isLoopInvariant(L)) { - Imm = SE->getAddExpr(Imm, Val); - Val = SE->getIntegerSCEV(0, Val->getType()); - return; + /// Offset - A constant offset to be added to the LSRUse expression. + /// This allows multiple fixups to share the same LSRUse with different + /// offsets, for example in an unrolled loop. + int64_t Offset; + + LSRFixup(); + + void print(raw_ostream &OS) const; + void dump() const; +}; + +} + +LSRFixup::LSRFixup() + : UserInst(0), OperandValToReplace(0), PostIncLoop(0), + LUIdx(~size_t(0)), Offset(0) {} + +void LSRFixup::print(raw_ostream &OS) const { + OS << "UserInst="; + // Store is common and interesting enough to be worth special-casing. + if (StoreInst *Store = dyn_cast(UserInst)) { + OS << "store "; + WriteAsOperand(OS, Store->getOperand(0), /*PrintType=*/false); + } else if (UserInst->getType()->isVoidTy()) + OS << UserInst->getOpcodeName(); + else + WriteAsOperand(OS, UserInst, /*PrintType=*/false); + + OS << ", OperandValToReplace="; + WriteAsOperand(OS, OperandValToReplace, /*PrintType=*/false); + + if (PostIncLoop) { + OS << ", PostIncLoop="; + WriteAsOperand(OS, PostIncLoop->getHeader(), /*PrintType=*/false); } - // Otherwise, no immediates to move. + if (LUIdx != ~size_t(0)) + OS << ", LUIdx=" << LUIdx; + + if (Offset != 0) + OS << ", Offset=" << Offset; } -static void MoveImmediateValues(const TargetLowering *TLI, - Instruction *User, - const SCEV *&Val, const SCEV *&Imm, - bool isAddress, Loop *L, - ScalarEvolution *SE) { - const Type *AccessTy = getAccessType(User); - MoveImmediateValues(TLI, AccessTy, Val, Imm, isAddress, L, SE); +void LSRFixup::dump() const { + print(errs()); errs() << '\n'; } -/// SeparateSubExprs - Decompose Expr into all of the subexpressions that are -/// added together. This is used to reassociate common addition subexprs -/// together for maximal sharing when rewriting bases. -static void SeparateSubExprs(SmallVector &SubExprs, - const SCEV *Expr, - ScalarEvolution *SE) { - if (const SCEVAddExpr *AE = dyn_cast(Expr)) { - for (unsigned j = 0, e = AE->getNumOperands(); j != e; ++j) - SeparateSubExprs(SubExprs, AE->getOperand(j), SE); - } else if (const SCEVAddRecExpr *SARE = dyn_cast(Expr)) { - const SCEV *Zero = SE->getIntegerSCEV(0, Expr->getType()); - if (SARE->getOperand(0) == Zero) { - SubExprs.push_back(Expr); - } else { - // Compute the addrec with zero as its base. - SmallVector Ops(SARE->op_begin(), SARE->op_end()); - Ops[0] = Zero; // Start with zero base. - SubExprs.push_back(SE->getAddRecExpr(Ops, SARE->getLoop())); +namespace { +/// UniquifierDenseMapInfo - A DenseMapInfo implementation for holding +/// DenseMaps and DenseSets of sorted SmallVectors of const SCEV*. +struct UniquifierDenseMapInfo { + static SmallVector getEmptyKey() { + SmallVector V; + V.push_back(reinterpret_cast(-1)); + return V; + } - SeparateSubExprs(SubExprs, SARE->getOperand(0), SE); - } - } else if (!Expr->isZero()) { - // Do not add zero. - SubExprs.push_back(Expr); - } -} - -// This is logically local to the following function, but C++ says we have -// to make it file scope. -struct SubExprUseData { unsigned Count; bool notAllUsesAreFree; }; - -/// RemoveCommonExpressionsFromUseBases - Look through all of the Bases of all -/// the Uses, removing any common subexpressions, except that if all such -/// subexpressions can be folded into an addressing mode for all uses inside -/// the loop (this case is referred to as "free" in comments herein) we do -/// not remove anything. This looks for things like (a+b+c) and -/// (a+c+d) and computes the common (a+c) subexpression. The common expression -/// is *removed* from the Bases and returned. -static const SCEV * -RemoveCommonExpressionsFromUseBases(std::vector &Uses, - ScalarEvolution *SE, Loop *L, - const TargetLowering *TLI) { - unsigned NumUses = Uses.size(); - - // Only one use? This is a very common case, so we handle it specially and - // cheaply. - const SCEV *Zero = SE->getIntegerSCEV(0, Uses[0].Base->getType()); - const SCEV *Result = Zero; - const SCEV *FreeResult = Zero; - if (NumUses == 1) { - // If the use is inside the loop, use its base, regardless of what it is: - // it is clearly shared across all the IV's. If the use is outside the loop - // (which means after it) we don't want to factor anything *into* the loop, - // so just use 0 as the base. - if (L->contains(Uses[0].Inst)) - std::swap(Result, Uses[0].Base); + static SmallVector getTombstoneKey() { + SmallVector V; + V.push_back(reinterpret_cast(-2)); + return V; + } + + static unsigned getHashValue(const SmallVector &V) { + unsigned Result = 0; + for (SmallVectorImpl::const_iterator I = V.begin(), + E = V.end(); I != E; ++I) + Result ^= DenseMapInfo::getHashValue(*I); return Result; } - // To find common subexpressions, count how many of Uses use each expression. - // If any subexpressions are used Uses.size() times, they are common. - // Also track whether all uses of each expression can be moved into an - // an addressing mode "for free"; such expressions are left within the loop. - // struct SubExprUseData { unsigned Count; bool notAllUsesAreFree; }; - std::map SubExpressionUseData; - - // UniqueSubExprs - Keep track of all of the subexpressions we see in the - // order we see them. - SmallVector UniqueSubExprs; - - SmallVector SubExprs; - unsigned NumUsesInsideLoop = 0; - for (unsigned i = 0; i != NumUses; ++i) { - // If the user is outside the loop, just ignore it for base computation. - // Since the user is outside the loop, it must be *after* the loop (if it - // were before, it could not be based on the loop IV). We don't want users - // after the loop to affect base computation of values *inside* the loop, - // because we can always add their offsets to the result IV after the loop - // is done, ensuring we get good code inside the loop. - if (!L->contains(Uses[i].Inst)) - continue; - NumUsesInsideLoop++; + static bool isEqual(const SmallVector &LHS, + const SmallVector &RHS) { + return LHS == RHS; + } +}; + +/// LSRUse - This class holds the state that LSR keeps for each use in +/// IVUsers, as well as uses invented by LSR itself. It includes information +/// about what kinds of things can be folded into the user, information about +/// the user itself, and information about how the use may be satisfied. +/// TODO: Represent multiple users of the same expression in common? +class LSRUse { + DenseSet, UniquifierDenseMapInfo> Uniquifier; + +public: + /// KindType - An enum for a kind of use, indicating what types of + /// scaled and immediate operands it might support. + enum KindType { + Basic, ///< A normal use, with no folding. + Special, ///< A special case of basic, allowing -1 scales. + Address, ///< An address use; folding according to TargetLowering + ICmpZero ///< An equality icmp with both operands folded into one. + // TODO: Add a generic icmp too? + }; - // If the base is zero (which is common), return zero now, there are no - // CSEs we can find. - if (Uses[i].Base == Zero) return Zero; + KindType Kind; + const Type *AccessTy; - // If this use is as an address we may be able to put CSEs in the addressing - // mode rather than hoisting them. - bool isAddrUse = isAddressUse(Uses[i].Inst, Uses[i].OperandValToReplace); - // We may need the AccessTy below, but only when isAddrUse, so compute it - // only in that case. - const Type *AccessTy = 0; - if (isAddrUse) - AccessTy = getAccessType(Uses[i].Inst); - - // Split the expression into subexprs. - SeparateSubExprs(SubExprs, Uses[i].Base, SE); - // Add one to SubExpressionUseData.Count for each subexpr present, and - // if the subexpr is not a valid immediate within an addressing mode use, - // set SubExpressionUseData.notAllUsesAreFree. We definitely want to - // hoist these out of the loop (if they are common to all uses). - for (unsigned j = 0, e = SubExprs.size(); j != e; ++j) { - if (++SubExpressionUseData[SubExprs[j]].Count == 1) - UniqueSubExprs.push_back(SubExprs[j]); - if (!isAddrUse || !fitsInAddressMode(SubExprs[j], AccessTy, TLI, false)) - SubExpressionUseData[SubExprs[j]].notAllUsesAreFree = true; - } - SubExprs.clear(); - } - - // Now that we know how many times each is used, build Result. Iterate over - // UniqueSubexprs so that we have a stable ordering. - for (unsigned i = 0, e = UniqueSubExprs.size(); i != e; ++i) { - std::map::iterator I = - SubExpressionUseData.find(UniqueSubExprs[i]); - assert(I != SubExpressionUseData.end() && "Entry not found?"); - if (I->second.Count == NumUsesInsideLoop) { // Found CSE! - if (I->second.notAllUsesAreFree) - Result = SE->getAddExpr(Result, I->first); - else - FreeResult = SE->getAddExpr(FreeResult, I->first); - } else - // Remove non-cse's from SubExpressionUseData. - SubExpressionUseData.erase(I); - } - - if (FreeResult != Zero) { - // We have some subexpressions that can be subsumed into addressing - // modes in every use inside the loop. However, it's possible that - // there are so many of them that the combined FreeResult cannot - // be subsumed, or that the target cannot handle both a FreeResult - // and a Result in the same instruction (for example because it would - // require too many registers). Check this. - for (unsigned i=0; icontains(Uses[i].Inst)) - continue; - // We know this is an addressing mode use; if there are any uses that - // are not, FreeResult would be Zero. - const Type *AccessTy = getAccessType(Uses[i].Inst); - if (!fitsInAddressMode(FreeResult, AccessTy, TLI, Result!=Zero)) { - // FIXME: could split up FreeResult into pieces here, some hoisted - // and some not. There is no obvious advantage to this. - Result = SE->getAddExpr(Result, FreeResult); - FreeResult = Zero; - break; - } - } - } + SmallVector Offsets; + int64_t MinOffset; + int64_t MaxOffset; - // If we found no CSE's, return now. - if (Result == Zero) return Result; + /// AllFixupsOutsideLoop - This records whether all of the fixups using this + /// LSRUse are outside of the loop, in which case some special-case heuristics + /// may be used. + bool AllFixupsOutsideLoop; - // If we still have a FreeResult, remove its subexpressions from - // SubExpressionUseData. This means they will remain in the use Bases. - if (FreeResult != Zero) { - SeparateSubExprs(SubExprs, FreeResult, SE); - for (unsigned j = 0, e = SubExprs.size(); j != e; ++j) { - std::map::iterator I = - SubExpressionUseData.find(SubExprs[j]); - SubExpressionUseData.erase(I); - } - SubExprs.clear(); - } + /// Formulae - A list of ways to build a value that can satisfy this user. + /// After the list is populated, one of these is selected heuristically and + /// used to formulate a replacement for OperandValToReplace in UserInst. + SmallVector Formulae; - // Otherwise, remove all of the CSE's we found from each of the base values. - for (unsigned i = 0; i != NumUses; ++i) { - // Uses outside the loop don't necessarily include the common base, but - // the final IV value coming into those uses does. Instead of trying to - // remove the pieces of the common base, which might not be there, - // subtract off the base to compensate for this. - if (!L->contains(Uses[i].Inst)) { - Uses[i].Base = SE->getMinusSCEV(Uses[i].Base, Result); - continue; - } + /// Regs - The set of register candidates used by all formulae in this LSRUse. + SmallPtrSet Regs; - // Split the expression into subexprs. - SeparateSubExprs(SubExprs, Uses[i].Base, SE); + LSRUse(KindType K, const Type *T) : Kind(K), AccessTy(T), + MinOffset(INT64_MAX), + MaxOffset(INT64_MIN), + AllFixupsOutsideLoop(true) {} - // Remove any common subexpressions. - for (unsigned j = 0, e = SubExprs.size(); j != e; ++j) - if (SubExpressionUseData.count(SubExprs[j])) { - SubExprs.erase(SubExprs.begin()+j); - --j; --e; - } + bool InsertFormula(const Formula &F); + + void check() const; + + void print(raw_ostream &OS) const; + void dump() const; +}; + +/// InsertFormula - If the given formula has not yet been inserted, add it to +/// the list, and return true. Return false otherwise. +bool LSRUse::InsertFormula(const Formula &F) { + SmallVector Key = F.BaseRegs; + if (F.ScaledReg) Key.push_back(F.ScaledReg); + // Unstable sort by host order ok, because this is only used for uniquifying. + std::sort(Key.begin(), Key.end()); + + if (!Uniquifier.insert(Key).second) + return false; + + // Using a register to hold the value of 0 is not profitable. + assert((!F.ScaledReg || !F.ScaledReg->isZero()) && + "Zero allocated in a scaled register!"); +#ifndef NDEBUG + for (SmallVectorImpl::const_iterator I = + F.BaseRegs.begin(), E = F.BaseRegs.end(); I != E; ++I) + assert(!(*I)->isZero() && "Zero allocated in a base register!"); +#endif + + // Add the formula to the list. + Formulae.push_back(F); + + // Record registers now being used by this use. + if (F.ScaledReg) Regs.insert(F.ScaledReg); + Regs.insert(F.BaseRegs.begin(), F.BaseRegs.end()); - // Finally, add the non-shared expressions together. - if (SubExprs.empty()) - Uses[i].Base = Zero; + return true; +} + +void LSRUse::print(raw_ostream &OS) const { + OS << "LSR Use: Kind="; + switch (Kind) { + case Basic: OS << "Basic"; break; + case Special: OS << "Special"; break; + case ICmpZero: OS << "ICmpZero"; break; + case Address: + OS << "Address of "; + if (AccessTy->isPointerTy()) + OS << "pointer"; // the full pointer type could be really verbose else - Uses[i].Base = SE->getAddExpr(SubExprs); - SubExprs.clear(); + OS << *AccessTy; + } + + OS << ", Offsets={"; + for (SmallVectorImpl::const_iterator I = Offsets.begin(), + E = Offsets.end(); I != E; ++I) { + OS << *I; + if (next(I) != E) + OS << ','; } + OS << '}'; - return Result; + if (AllFixupsOutsideLoop) + OS << ", all-fixups-outside-loop"; } -/// ValidScale - Check whether the given Scale is valid for all loads and -/// stores in UsersToProcess. -/// -bool LoopStrengthReduce::ValidScale(bool HasBaseReg, int64_t Scale, - const std::vector& UsersToProcess) { - if (!TLI) - return true; +void LSRUse::dump() const { + print(errs()); errs() << '\n'; +} - for (unsigned i = 0, e = UsersToProcess.size(); i!=e; ++i) { - // If this is a load or other access, pass the type of the access in. - const Type *AccessTy = - Type::getVoidTy(UsersToProcess[i].Inst->getContext()); - if (isAddressUse(UsersToProcess[i].Inst, - UsersToProcess[i].OperandValToReplace)) - AccessTy = getAccessType(UsersToProcess[i].Inst); - else if (isa(UsersToProcess[i].Inst)) - continue; +/// isLegalUse - Test whether the use described by AM is "legal", meaning it can +/// be completely folded into the user instruction at isel time. This includes +/// address-mode folding and special icmp tricks. +static bool isLegalUse(const TargetLowering::AddrMode &AM, + LSRUse::KindType Kind, const Type *AccessTy, + const TargetLowering *TLI) { + switch (Kind) { + case LSRUse::Address: + // If we have low-level target information, ask the target if it can + // completely fold this address. + if (TLI) return TLI->isLegalAddressingMode(AM, AccessTy); + + // Otherwise, just guess that reg+reg addressing is legal. + return !AM.BaseGV && AM.BaseOffs == 0 && AM.Scale <= 1; + + case LSRUse::ICmpZero: + // There's not even a target hook for querying whether it would be legal to + // fold a GV into an ICmp. + if (AM.BaseGV) + return false; - TargetLowering::AddrMode AM; - if (const SCEVConstant *SC = dyn_cast(UsersToProcess[i].Imm)) - AM.BaseOffs = SC->getValue()->getSExtValue(); - AM.HasBaseReg = HasBaseReg || !UsersToProcess[i].Base->isZero(); - AM.Scale = Scale; + // ICmp only has two operands; don't allow more than two non-trivial parts. + if (AM.Scale != 0 && AM.HasBaseReg && AM.BaseOffs != 0) + return false; - // If load[imm+r*scale] is illegal, bail out. - if (!TLI->isLegalAddressingMode(AM, AccessTy)) + // ICmp only supports no scale or a -1 scale, as we can "fold" a -1 scale by + // putting the scaled register in the other operand of the icmp. + if (AM.Scale != 0 && AM.Scale != -1) return false; - } - return true; -} -/// ValidOffset - Check whether the given Offset is valid for all loads and -/// stores in UsersToProcess. -/// -bool LoopStrengthReduce::ValidOffset(bool HasBaseReg, - int64_t Offset, - int64_t Scale, - const std::vector& UsersToProcess) { - if (!TLI) + // If we have low-level target information, ask the target if it can fold an + // integer immediate on an icmp. + if (AM.BaseOffs != 0) { + if (TLI) return TLI->isLegalICmpImmediate(-AM.BaseOffs); + return false; + } + return true; - for (unsigned i=0, e = UsersToProcess.size(); i!=e; ++i) { - // If this is a load or other access, pass the type of the access in. - const Type *AccessTy = - Type::getVoidTy(UsersToProcess[i].Inst->getContext()); - if (isAddressUse(UsersToProcess[i].Inst, - UsersToProcess[i].OperandValToReplace)) - AccessTy = getAccessType(UsersToProcess[i].Inst); - else if (isa(UsersToProcess[i].Inst)) - continue; + case LSRUse::Basic: + // Only handle single-register values. + return !AM.BaseGV && AM.Scale == 0 && AM.BaseOffs == 0; + + case LSRUse::Special: + // Only handle -1 scales, or no scale. + return AM.Scale == 0 || AM.Scale == -1; + } - TargetLowering::AddrMode AM; - if (const SCEVConstant *SC = dyn_cast(UsersToProcess[i].Imm)) - AM.BaseOffs = SC->getValue()->getSExtValue(); - AM.BaseOffs = (uint64_t)AM.BaseOffs + (uint64_t)Offset; - AM.HasBaseReg = HasBaseReg || !UsersToProcess[i].Base->isZero(); - AM.Scale = Scale; + return false; +} - // If load[imm+r*scale] is illegal, bail out. - if (!TLI->isLegalAddressingMode(AM, AccessTy)) +static bool isLegalUse(TargetLowering::AddrMode AM, + int64_t MinOffset, int64_t MaxOffset, + LSRUse::KindType Kind, const Type *AccessTy, + const TargetLowering *TLI) { + // Check for overflow. + if (((int64_t)((uint64_t)AM.BaseOffs + MinOffset) > AM.BaseOffs) != + (MinOffset > 0)) + return false; + AM.BaseOffs = (uint64_t)AM.BaseOffs + MinOffset; + if (isLegalUse(AM, Kind, AccessTy, TLI)) { + AM.BaseOffs = (uint64_t)AM.BaseOffs - MinOffset; + // Check for overflow. + if (((int64_t)((uint64_t)AM.BaseOffs + MaxOffset) > AM.BaseOffs) != + (MaxOffset > 0)) return false; + AM.BaseOffs = (uint64_t)AM.BaseOffs + MaxOffset; + return isLegalUse(AM, Kind, AccessTy, TLI); } - return true; + return false; } -/// RequiresTypeConversion - Returns true if converting Ty1 to Ty2 is not -/// a nop. -bool LoopStrengthReduce::RequiresTypeConversion(const Type *Ty1, - const Type *Ty2) { - if (Ty1 == Ty2) - return false; - Ty1 = SE->getEffectiveSCEVType(Ty1); - Ty2 = SE->getEffectiveSCEVType(Ty2); - if (Ty1 == Ty2) - return false; - if (Ty1->canLosslesslyBitCastTo(Ty2)) - return false; - if (TLI && TLI->isTruncateFree(Ty1, Ty2)) - return false; - return true; +static bool isAlwaysFoldable(int64_t BaseOffs, + GlobalValue *BaseGV, + bool HasBaseReg, + LSRUse::KindType Kind, const Type *AccessTy, + const TargetLowering *TLI) { + // Fast-path: zero is always foldable. + if (BaseOffs == 0 && !BaseGV) return true; + + // Conservatively, create an address with an immediate and a + // base and a scale. + TargetLowering::AddrMode AM; + AM.BaseOffs = BaseOffs; + AM.BaseGV = BaseGV; + AM.HasBaseReg = HasBaseReg; + AM.Scale = Kind == LSRUse::ICmpZero ? -1 : 1; + + return isLegalUse(AM, Kind, AccessTy, TLI); } -/// CheckForIVReuse - Returns the multiple if the stride is the multiple -/// of a previous stride and it is a legal value for the target addressing -/// mode scale component and optional base reg. This allows the users of -/// this stride to be rewritten as prev iv * factor. It returns 0 if no -/// reuse is possible. Factors can be negative on same targets, e.g. ARM. -/// -/// If all uses are outside the loop, we don't require that all multiplies -/// be folded into the addressing mode, nor even that the factor be constant; -/// a multiply (executed once) outside the loop is better than another IV -/// within. Well, usually. -const SCEV *LoopStrengthReduce::CheckForIVReuse(bool HasBaseReg, - bool AllUsesAreAddresses, - bool AllUsesAreOutsideLoop, - const SCEV *Stride, - IVExpr &IV, const Type *Ty, - const std::vector& UsersToProcess) { - if (const SCEVConstant *SC = dyn_cast(Stride)) { - int64_t SInt = SC->getValue()->getSExtValue(); - for (unsigned NewStride = 0, e = IU->StrideOrder.size(); - NewStride != e; ++NewStride) { - std::map::iterator SI = - IVsByStride.find(IU->StrideOrder[NewStride]); - if (SI == IVsByStride.end() || !isa(SI->first)) - continue; - // The other stride has no uses, don't reuse it. - std::map::iterator UI = - IU->IVUsesByStride.find(IU->StrideOrder[NewStride]); - if (UI->second->Users.empty()) - continue; - int64_t SSInt = cast(SI->first)->getValue()->getSExtValue(); - if (SI->first != Stride && - (unsigned(abs64(SInt)) < SSInt || (SInt % SSInt) != 0)) - continue; - int64_t Scale = SInt / SSInt; - // Check that this stride is valid for all the types used for loads and - // stores; if it can be used for some and not others, we might as well use - // the original stride everywhere, since we have to create the IV for it - // anyway. If the scale is 1, then we don't need to worry about folding - // multiplications. - if (Scale == 1 || - (AllUsesAreAddresses && - ValidScale(HasBaseReg, Scale, UsersToProcess))) { - // Prefer to reuse an IV with a base of zero. - for (std::vector::iterator II = SI->second.IVs.begin(), - IE = SI->second.IVs.end(); II != IE; ++II) - // Only reuse previous IV if it would not require a type conversion - // and if the base difference can be folded. - if (II->Base->isZero() && - !RequiresTypeConversion(II->Base->getType(), Ty)) { - IV = *II; - return SE->getIntegerSCEV(Scale, Stride->getType()); - } - // Otherwise, settle for an IV with a foldable base. - if (AllUsesAreAddresses) - for (std::vector::iterator II = SI->second.IVs.begin(), - IE = SI->second.IVs.end(); II != IE; ++II) - // Only reuse previous IV if it would not require a type conversion - // and if the base difference can be folded. - if (SE->getEffectiveSCEVType(II->Base->getType()) == - SE->getEffectiveSCEVType(Ty) && - isa(II->Base)) { - int64_t Base = - cast(II->Base)->getValue()->getSExtValue(); - if (Base > INT32_MIN && Base <= INT32_MAX && - ValidOffset(HasBaseReg, -Base * Scale, - Scale, UsersToProcess)) { - IV = *II; - return SE->getIntegerSCEV(Scale, Stride->getType()); - } - } - } - } - } else if (AllUsesAreOutsideLoop) { - // Accept nonconstant strides here; it is really really right to substitute - // an existing IV if we can. - for (unsigned NewStride = 0, e = IU->StrideOrder.size(); - NewStride != e; ++NewStride) { - std::map::iterator SI = - IVsByStride.find(IU->StrideOrder[NewStride]); - if (SI == IVsByStride.end() || !isa(SI->first)) - continue; - int64_t SSInt = cast(SI->first)->getValue()->getSExtValue(); - if (SI->first != Stride && SSInt != 1) - continue; - for (std::vector::iterator II = SI->second.IVs.begin(), - IE = SI->second.IVs.end(); II != IE; ++II) - // Accept nonzero base here. - // Only reuse previous IV if it would not require a type conversion. - if (!RequiresTypeConversion(II->Base->getType(), Ty)) { - IV = *II; - return Stride; - } - } - // Special case, old IV is -1*x and this one is x. Can treat this one as - // -1*old. - for (unsigned NewStride = 0, e = IU->StrideOrder.size(); - NewStride != e; ++NewStride) { - std::map::iterator SI = - IVsByStride.find(IU->StrideOrder[NewStride]); - if (SI == IVsByStride.end()) - continue; - if (const SCEVMulExpr *ME = dyn_cast(SI->first)) - if (const SCEVConstant *SC = dyn_cast(ME->getOperand(0))) - if (Stride == ME->getOperand(1) && - SC->getValue()->getSExtValue() == -1LL) - for (std::vector::iterator II = SI->second.IVs.begin(), - IE = SI->second.IVs.end(); II != IE; ++II) - // Accept nonzero base here. - // Only reuse previous IV if it would not require type conversion. - if (!RequiresTypeConversion(II->Base->getType(), Ty)) { - IV = *II; - return SE->getIntegerSCEV(-1LL, Stride->getType()); - } - } - } - return SE->getIntegerSCEV(0, Stride->getType()); -} - -/// PartitionByIsUseOfPostIncrementedValue - Simple boolean predicate that -/// returns true if Val's isUseOfPostIncrementedValue is true. -static bool PartitionByIsUseOfPostIncrementedValue(const BasedUser &Val) { - return Val.isUseOfPostIncrementedValue; -} - -/// isNonConstantNegative - Return true if the specified scev is negated, but -/// not a constant. -static bool isNonConstantNegative(const SCEV *Expr) { - const SCEVMulExpr *Mul = dyn_cast(Expr); - if (!Mul) return false; - - // If there is a constant factor, it will be first. - const SCEVConstant *SC = dyn_cast(Mul->getOperand(0)); - if (!SC) return false; - - // Return true if the value is negative, this matches things like (-42 * V). - return SC->getValue()->getValue().isNegative(); -} - -/// CollectIVUsers - Transform our list of users and offsets to a bit more -/// complex table. In this new vector, each 'BasedUser' contains 'Base', the -/// base of the strided accesses, as well as the old information from Uses. We -/// progressively move information from the Base field to the Imm field, until -/// we eventually have the full access expression to rewrite the use. -const SCEV *LoopStrengthReduce::CollectIVUsers(const SCEV *Stride, - IVUsersOfOneStride &Uses, - Loop *L, - bool &AllUsesAreAddresses, - bool &AllUsesAreOutsideLoop, - std::vector &UsersToProcess) { - // FIXME: Generalize to non-affine IV's. - if (!Stride->isLoopInvariant(L)) - return SE->getIntegerSCEV(0, Stride->getType()); - - UsersToProcess.reserve(Uses.Users.size()); - for (ilist::iterator I = Uses.Users.begin(), - E = Uses.Users.end(); I != E; ++I) { - UsersToProcess.push_back(BasedUser(*I, SE)); - - // Move any loop variant operands from the offset field to the immediate - // field of the use, so that we don't try to use something before it is - // computed. - MoveLoopVariantsToImmediateField(UsersToProcess.back().Base, - UsersToProcess.back().Imm, L, SE); - assert(UsersToProcess.back().Base->isLoopInvariant(L) && - "Base value is not loop invariant!"); - } - - // We now have a whole bunch of uses of like-strided induction variables, but - // they might all have different bases. We want to emit one PHI node for this - // stride which we fold as many common expressions (between the IVs) into as - // possible. Start by identifying the common expressions in the base values - // for the strides (e.g. if we have "A+C+B" and "A+B+D" as our bases, find - // "A+B"), emit it to the preheader, then remove the expression from the - // UsersToProcess base values. - const SCEV *CommonExprs = - RemoveCommonExpressionsFromUseBases(UsersToProcess, SE, L, TLI); - - // Next, figure out what we can represent in the immediate fields of - // instructions. If we can represent anything there, move it to the imm - // fields of the BasedUsers. We do this so that it increases the commonality - // of the remaining uses. - unsigned NumPHI = 0; - bool HasAddress = false; - for (unsigned i = 0, e = UsersToProcess.size(); i != e; ++i) { - // If the user is not in the current loop, this means it is using the exit - // value of the IV. Do not put anything in the base, make sure it's all in - // the immediate field to allow as much factoring as possible. - if (!L->contains(UsersToProcess[i].Inst)) { - UsersToProcess[i].Imm = SE->getAddExpr(UsersToProcess[i].Imm, - UsersToProcess[i].Base); - UsersToProcess[i].Base = - SE->getIntegerSCEV(0, UsersToProcess[i].Base->getType()); - } else { - // Not all uses are outside the loop. - AllUsesAreOutsideLoop = false; - - // Addressing modes can be folded into loads and stores. Be careful that - // the store is through the expression, not of the expression though. - bool isPHI = false; - bool isAddress = isAddressUse(UsersToProcess[i].Inst, - UsersToProcess[i].OperandValToReplace); - if (isa(UsersToProcess[i].Inst)) { - isPHI = true; - ++NumPHI; - } - - if (isAddress) - HasAddress = true; - - // If this use isn't an address, then not all uses are addresses. - if (!isAddress && !isPHI) - AllUsesAreAddresses = false; +static bool isAlwaysFoldable(const SCEV *S, + int64_t MinOffset, int64_t MaxOffset, + bool HasBaseReg, + LSRUse::KindType Kind, const Type *AccessTy, + const TargetLowering *TLI, + ScalarEvolution &SE) { + // Fast-path: zero is always foldable. + if (S->isZero()) return true; + + // Conservatively, create an address with an immediate and a + // base and a scale. + int64_t BaseOffs = ExtractImmediate(S, SE); + GlobalValue *BaseGV = ExtractSymbol(S, SE); + + // If there's anything else involved, it's not foldable. + if (!S->isZero()) return false; + + // Fast-path: zero is always foldable. + if (BaseOffs == 0 && !BaseGV) return true; + + // Conservatively, create an address with an immediate and a + // base and a scale. + TargetLowering::AddrMode AM; + AM.BaseOffs = BaseOffs; + AM.BaseGV = BaseGV; + AM.HasBaseReg = HasBaseReg; + AM.Scale = Kind == LSRUse::ICmpZero ? -1 : 1; + + return isLegalUse(AM, MinOffset, MaxOffset, Kind, AccessTy, TLI); +} - MoveImmediateValues(TLI, UsersToProcess[i].Inst, UsersToProcess[i].Base, - UsersToProcess[i].Imm, isAddress, L, SE); - } +/// FormulaSorter - This class implements an ordering for formulae which sorts +/// the by their standalone cost. +class FormulaSorter { + /// These two sets are kept empty, so that we compute standalone costs. + DenseSet VisitedRegs; + SmallPtrSet Regs; + Loop *L; + LSRUse *LU; + ScalarEvolution &SE; + DominatorTree &DT; + +public: + FormulaSorter(Loop *l, LSRUse &lu, ScalarEvolution &se, DominatorTree &dt) + : L(l), LU(&lu), SE(se), DT(dt) {} + + bool operator()(const Formula &A, const Formula &B) { + Cost CostA; + CostA.RateFormula(A, Regs, VisitedRegs, L, LU->Offsets, SE, DT); + Regs.clear(); + Cost CostB; + CostB.RateFormula(B, Regs, VisitedRegs, L, LU->Offsets, SE, DT); + Regs.clear(); + return CostA < CostB; + } +}; + +/// LSRInstance - This class holds state for the main loop strength reduction +/// logic. +class LSRInstance { + IVUsers &IU; + ScalarEvolution &SE; + DominatorTree &DT; + const TargetLowering *const TLI; + Loop *const L; + bool Changed; + + /// IVIncInsertPos - This is the insert position that the current loop's + /// induction variable increment should be placed. In simple loops, this is + /// the latch block's terminator. But in more complicated cases, this is a + /// position which will dominate all the in-loop post-increment users. + Instruction *IVIncInsertPos; + + /// Factors - Interesting factors between use strides. + SmallSetVector Factors; + + /// Types - Interesting use types, to facilitate truncation reuse. + SmallSetVector Types; + + /// Fixups - The list of operands which are to be replaced. + SmallVector Fixups; + + /// Uses - The list of interesting uses. + SmallVector Uses; + + /// RegUses - Track which uses use which register candidates. + RegUseTracker RegUses; + + void OptimizeShadowIV(); + bool FindIVUserForCond(ICmpInst *Cond, IVStrideUse *&CondUse); + ICmpInst *OptimizeMax(ICmpInst *Cond, IVStrideUse* &CondUse); + bool OptimizeLoopTermCond(); + + void CollectInterestingTypesAndFactors(); + void CollectFixupsAndInitialFormulae(); + + LSRFixup &getNewFixup() { + Fixups.push_back(LSRFixup()); + return Fixups.back(); } - // If one of the use is a PHI node and all other uses are addresses, still - // allow iv reuse. Essentially we are trading one constant multiplication - // for one fewer iv. - if (NumPHI > 1) - AllUsesAreAddresses = false; - - // There are no in-loop address uses. - if (AllUsesAreAddresses && (!HasAddress && !AllUsesAreOutsideLoop)) - AllUsesAreAddresses = false; + // Support for sharing of LSRUses between LSRFixups. + typedef DenseMap UseMapTy; + UseMapTy UseMap; + + bool reconcileNewOffset(LSRUse &LU, int64_t NewOffset, + LSRUse::KindType Kind, const Type *AccessTy); + + std::pair getUse(const SCEV *&Expr, + LSRUse::KindType Kind, + const Type *AccessTy); + +public: + void InsertInitialFormula(const SCEV *S, LSRUse &LU, size_t LUIdx); + void InsertSupplementalFormula(const SCEV *S, LSRUse &LU, size_t LUIdx); + void CountRegisters(const Formula &F, size_t LUIdx); + bool InsertFormula(LSRUse &LU, unsigned LUIdx, const Formula &F); + + void CollectLoopInvariantFixupsAndFormulae(); + + void GenerateReassociations(LSRUse &LU, unsigned LUIdx, Formula Base, + unsigned Depth = 0); + void GenerateCombinations(LSRUse &LU, unsigned LUIdx, Formula Base); + void GenerateSymbolicOffsets(LSRUse &LU, unsigned LUIdx, Formula Base); + void GenerateConstantOffsets(LSRUse &LU, unsigned LUIdx, Formula Base); + void GenerateICmpZeroScales(LSRUse &LU, unsigned LUIdx, Formula Base); + void GenerateScales(LSRUse &LU, unsigned LUIdx, Formula Base); + void GenerateTruncates(LSRUse &LU, unsigned LUIdx, Formula Base); + void GenerateCrossUseConstantOffsets(); + void GenerateAllReuseFormulae(); + + void FilterOutUndesirableDedicatedRegisters(); + void NarrowSearchSpaceUsingHeuristics(); + + void SolveRecurse(SmallVectorImpl &Solution, + Cost &SolutionCost, + SmallVectorImpl &Workspace, + const Cost &CurCost, + const SmallPtrSet &CurRegs, + DenseSet &VisitedRegs) const; + void Solve(SmallVectorImpl &Solution) const; + + Value *Expand(const LSRFixup &LF, + const Formula &F, + BasicBlock::iterator IP, + SCEVExpander &Rewriter, + SmallVectorImpl &DeadInsts) const; + void RewriteForPHI(PHINode *PN, const LSRFixup &LF, + const Formula &F, + SCEVExpander &Rewriter, + SmallVectorImpl &DeadInsts, + Pass *P) const; + void Rewrite(const LSRFixup &LF, + const Formula &F, + SCEVExpander &Rewriter, + SmallVectorImpl &DeadInsts, + Pass *P) const; + void ImplementSolution(const SmallVectorImpl &Solution, + Pass *P); + + LSRInstance(const TargetLowering *tli, Loop *l, Pass *P); + + bool getChanged() const { return Changed; } + + void print_factors_and_types(raw_ostream &OS) const; + void print_fixups(raw_ostream &OS) const; + void print_uses(raw_ostream &OS) const; + void print(raw_ostream &OS) const; + void dump() const; +}; - return CommonExprs; } -/// ShouldUseFullStrengthReductionMode - Test whether full strength-reduction -/// is valid and profitable for the given set of users of a stride. In -/// full strength-reduction mode, all addresses at the current stride are -/// strength-reduced all the way down to pointer arithmetic. -/// -bool LoopStrengthReduce::ShouldUseFullStrengthReductionMode( - const std::vector &UsersToProcess, - const Loop *L, - bool AllUsesAreAddresses, - const SCEV *Stride) { - if (!EnableFullLSRMode) - return false; - - // The heuristics below aim to avoid increasing register pressure, but - // fully strength-reducing all the addresses increases the number of - // add instructions, so don't do this when optimizing for size. - // TODO: If the loop is large, the savings due to simpler addresses - // may oughtweight the costs of the extra increment instructions. - if (L->getHeader()->getParent()->hasFnAttr(Attribute::OptimizeForSize)) - return false; +/// OptimizeShadowIV - If IV is used in a int-to-float cast +/// inside the loop then try to eliminate the cast operation. +void LSRInstance::OptimizeShadowIV() { + const SCEV *BackedgeTakenCount = SE.getBackedgeTakenCount(L); + if (isa(BackedgeTakenCount)) + return; - // TODO: For now, don't do full strength reduction if there could - // potentially be greater-stride multiples of the current stride - // which could reuse the current stride IV. - if (IU->StrideOrder.back() != Stride) - return false; + for (IVUsers::const_iterator UI = IU.begin(), E = IU.end(); + UI != E; /* empty */) { + IVUsers::const_iterator CandidateUI = UI; + ++UI; + Instruction *ShadowUse = CandidateUI->getUser(); + const Type *DestTy = NULL; - // Iterate through the uses to find conditions that automatically rule out - // full-lsr mode. - for (unsigned i = 0, e = UsersToProcess.size(); i != e; ) { - const SCEV *Base = UsersToProcess[i].Base; - const SCEV *Imm = UsersToProcess[i].Imm; - // If any users have a loop-variant component, they can't be fully - // strength-reduced. - if (Imm && !Imm->isLoopInvariant(L)) - return false; - // If there are to users with the same base and the difference between - // the two Imm values can't be folded into the address, full - // strength reduction would increase register pressure. - do { - const SCEV *CurImm = UsersToProcess[i].Imm; - if ((CurImm || Imm) && CurImm != Imm) { - if (!CurImm) CurImm = SE->getIntegerSCEV(0, Stride->getType()); - if (!Imm) Imm = SE->getIntegerSCEV(0, Stride->getType()); - const Instruction *Inst = UsersToProcess[i].Inst; - const Type *AccessTy = getAccessType(Inst); - const SCEV *Diff = SE->getMinusSCEV(UsersToProcess[i].Imm, Imm); - if (!Diff->isZero() && - (!AllUsesAreAddresses || - !fitsInAddressMode(Diff, AccessTy, TLI, /*HasBaseReg=*/true))) - return false; - } - } while (++i != e && Base == UsersToProcess[i].Base); - } + /* If shadow use is a int->float cast then insert a second IV + to eliminate this cast. - // If there's exactly one user in this stride, fully strength-reducing it - // won't increase register pressure. If it's starting from a non-zero base, - // it'll be simpler this way. - if (UsersToProcess.size() == 1 && !UsersToProcess[0].Base->isZero()) - return true; + for (unsigned i = 0; i < n; ++i) + foo((double)i); - // Otherwise, if there are any users in this stride that don't require - // a register for their base, full strength-reduction will increase - // register pressure. - for (unsigned i = 0, e = UsersToProcess.size(); i != e; ++i) - if (UsersToProcess[i].Base->isZero()) - return false; + is transformed into - // Otherwise, go for it. - return true; -} + double d = 0.0; + for (unsigned i = 0; i < n; ++i, ++d) + foo(d); + */ + if (UIToFPInst *UCast = dyn_cast(CandidateUI->getUser())) + DestTy = UCast->getDestTy(); + else if (SIToFPInst *SCast = dyn_cast(CandidateUI->getUser())) + DestTy = SCast->getDestTy(); + if (!DestTy) continue; -/// InsertAffinePhi Create and insert a PHI node for an induction variable -/// with the specified start and step values in the specified loop. -/// -/// If NegateStride is true, the stride should be negated by using a -/// subtract instead of an add. -/// -/// Return the created phi node. -/// -static PHINode *InsertAffinePhi(const SCEV *Start, const SCEV *Step, - Instruction *IVIncInsertPt, - const Loop *L, - SCEVExpander &Rewriter) { - assert(Start->isLoopInvariant(L) && "New PHI start is not loop invariant!"); - assert(Step->isLoopInvariant(L) && "New PHI stride is not loop invariant!"); - - BasicBlock *Header = L->getHeader(); - BasicBlock *Preheader = L->getLoopPreheader(); - BasicBlock *LatchBlock = L->getLoopLatch(); - const Type *Ty = Start->getType(); - Ty = Rewriter.SE.getEffectiveSCEVType(Ty); - - PHINode *PN = PHINode::Create(Ty, "lsr.iv", Header->begin()); - PN->addIncoming(Rewriter.expandCodeFor(Start, Ty, Preheader->getTerminator()), - Preheader); - - // If the stride is negative, insert a sub instead of an add for the - // increment. - bool isNegative = isNonConstantNegative(Step); - const SCEV *IncAmount = Step; - if (isNegative) - IncAmount = Rewriter.SE.getNegativeSCEV(Step); - - // Insert an add instruction right before the terminator corresponding - // to the back-edge or just before the only use. The location is determined - // by the caller and passed in as IVIncInsertPt. - Value *StepV = Rewriter.expandCodeFor(IncAmount, Ty, - Preheader->getTerminator()); - Instruction *IncV; - if (isNegative) { - IncV = BinaryOperator::CreateSub(PN, StepV, "lsr.iv.next", - IVIncInsertPt); - } else { - IncV = BinaryOperator::CreateAdd(PN, StepV, "lsr.iv.next", - IVIncInsertPt); - } - if (!isa(StepV)) ++NumVariable; - - PN->addIncoming(IncV, LatchBlock); - - ++NumInserted; - return PN; -} - -static void SortUsersToProcess(std::vector &UsersToProcess) { - // We want to emit code for users inside the loop first. To do this, we - // rearrange BasedUser so that the entries at the end have - // isUseOfPostIncrementedValue = false, because we pop off the end of the - // vector (so we handle them first). - std::partition(UsersToProcess.begin(), UsersToProcess.end(), - PartitionByIsUseOfPostIncrementedValue); - - // Sort this by base, so that things with the same base are handled - // together. By partitioning first and stable-sorting later, we are - // guaranteed that within each base we will pop off users from within the - // loop before users outside of the loop with a particular base. - // - // We would like to use stable_sort here, but we can't. The problem is that - // const SCEV *'s don't have a deterministic ordering w.r.t to each other, so - // we don't have anything to do a '<' comparison on. Because we think the - // number of uses is small, do a horrible bubble sort which just relies on - // ==. - for (unsigned i = 0, e = UsersToProcess.size(); i != e; ++i) { - // Get a base value. - const SCEV *Base = UsersToProcess[i].Base; - - // Compact everything with this base to be consecutive with this one. - for (unsigned j = i+1; j != e; ++j) { - if (UsersToProcess[j].Base == Base) { - std::swap(UsersToProcess[i+1], UsersToProcess[j]); - ++i; - } + if (TLI) { + // If target does not support DestTy natively then do not apply + // this transformation. + EVT DVT = TLI->getValueType(DestTy); + if (!TLI->isTypeLegal(DVT)) continue; } - } -} -/// PrepareToStrengthReduceFully - Prepare to fully strength-reduce -/// UsersToProcess, meaning lowering addresses all the way down to direct -/// pointer arithmetic. -/// -void -LoopStrengthReduce::PrepareToStrengthReduceFully( - std::vector &UsersToProcess, - const SCEV *Stride, - const SCEV *CommonExprs, - const Loop *L, - SCEVExpander &PreheaderRewriter) { - DEBUG(dbgs() << " Fully reducing all users\n"); - - // Rewrite the UsersToProcess records, creating a separate PHI for each - // unique Base value. - Instruction *IVIncInsertPt = L->getLoopLatch()->getTerminator(); - for (unsigned i = 0, e = UsersToProcess.size(); i != e; ) { - // TODO: The uses are grouped by base, but not sorted. We arbitrarily - // pick the first Imm value here to start with, and adjust it for the - // other uses. - const SCEV *Imm = UsersToProcess[i].Imm; - const SCEV *Base = UsersToProcess[i].Base; - const SCEV *Start = SE->getAddExpr(CommonExprs, Base, Imm); - PHINode *Phi = InsertAffinePhi(Start, Stride, IVIncInsertPt, L, - PreheaderRewriter); - // Loop over all the users with the same base. - do { - UsersToProcess[i].Base = SE->getIntegerSCEV(0, Stride->getType()); - UsersToProcess[i].Imm = SE->getMinusSCEV(UsersToProcess[i].Imm, Imm); - UsersToProcess[i].Phi = Phi; - assert(UsersToProcess[i].Imm->isLoopInvariant(L) && - "ShouldUseFullStrengthReductionMode should reject this!"); - } while (++i != e && Base == UsersToProcess[i].Base); - } -} - -/// FindIVIncInsertPt - Return the location to insert the increment instruction. -/// If the only use if a use of postinc value, (must be the loop termination -/// condition), then insert it just before the use. -static Instruction *FindIVIncInsertPt(std::vector &UsersToProcess, - const Loop *L) { - if (UsersToProcess.size() == 1 && - UsersToProcess[0].isUseOfPostIncrementedValue && - L->contains(UsersToProcess[0].Inst)) - return UsersToProcess[0].Inst; - return L->getLoopLatch()->getTerminator(); -} - -/// PrepareToStrengthReduceWithNewPhi - Insert a new induction variable for the -/// given users to share. -/// -void -LoopStrengthReduce::PrepareToStrengthReduceWithNewPhi( - std::vector &UsersToProcess, - const SCEV *Stride, - const SCEV *CommonExprs, - Value *CommonBaseV, - Instruction *IVIncInsertPt, - const Loop *L, - SCEVExpander &PreheaderRewriter) { - DEBUG(dbgs() << " Inserting new PHI:\n"); - - PHINode *Phi = InsertAffinePhi(SE->getUnknown(CommonBaseV), - Stride, IVIncInsertPt, L, - PreheaderRewriter); - - // Remember this in case a later stride is multiple of this. - IVsByStride[Stride].addIV(Stride, CommonExprs, Phi); - - // All the users will share this new IV. - for (unsigned i = 0, e = UsersToProcess.size(); i != e; ++i) - UsersToProcess[i].Phi = Phi; - - DEBUG(dbgs() << " IV="); - DEBUG(WriteAsOperand(dbgs(), Phi, /*PrintType=*/false)); - DEBUG(dbgs() << "\n"); -} - -/// PrepareToStrengthReduceFromSmallerStride - Prepare for the given users to -/// reuse an induction variable with a stride that is a factor of the current -/// induction variable. -/// -void -LoopStrengthReduce::PrepareToStrengthReduceFromSmallerStride( - std::vector &UsersToProcess, - Value *CommonBaseV, - const IVExpr &ReuseIV, - Instruction *PreInsertPt) { - DEBUG(dbgs() << " Rewriting in terms of existing IV of STRIDE " - << *ReuseIV.Stride << " and BASE " << *ReuseIV.Base << "\n"); - - // All the users will share the reused IV. - for (unsigned i = 0, e = UsersToProcess.size(); i != e; ++i) - UsersToProcess[i].Phi = ReuseIV.PHI; - - Constant *C = dyn_cast(CommonBaseV); - if (C && - (!C->isNullValue() && - !fitsInAddressMode(SE->getUnknown(CommonBaseV), CommonBaseV->getType(), - TLI, false))) - // We want the common base emitted into the preheader! This is just - // using cast as a copy so BitCast (no-op cast) is appropriate - CommonBaseV = new BitCastInst(CommonBaseV, CommonBaseV->getType(), - "commonbase", PreInsertPt); -} - -static bool IsImmFoldedIntoAddrMode(GlobalValue *GV, int64_t Offset, - const Type *AccessTy, - std::vector &UsersToProcess, - const TargetLowering *TLI) { - SmallVector AddrModeInsts; - for (unsigned i = 0, e = UsersToProcess.size(); i != e; ++i) { - if (UsersToProcess[i].isUseOfPostIncrementedValue) - continue; - ExtAddrMode AddrMode = - AddressingModeMatcher::Match(UsersToProcess[i].OperandValToReplace, - AccessTy, UsersToProcess[i].Inst, - AddrModeInsts, *TLI); - if (GV && GV != AddrMode.BaseGV) - return false; - if (Offset && !AddrMode.BaseOffs) - // FIXME: How to accurate check it's immediate offset is folded. - return false; - AddrModeInsts.clear(); - } - return true; -} + PHINode *PH = dyn_cast(ShadowUse->getOperand(0)); + if (!PH) continue; + if (PH->getNumIncomingValues() != 2) continue; -/// StrengthReduceIVUsersOfStride - Strength reduce all of the users of a single -/// stride of IV. All of the users may have different starting values, and this -/// may not be the only stride. -void -LoopStrengthReduce::StrengthReduceIVUsersOfStride(const SCEV *Stride, - IVUsersOfOneStride &Uses, - Loop *L) { - // If all the users are moved to another stride, then there is nothing to do. - if (Uses.Users.empty()) - return; + const Type *SrcTy = PH->getType(); + int Mantissa = DestTy->getFPMantissaWidth(); + if (Mantissa == -1) continue; + if ((int)SE.getTypeSizeInBits(SrcTy) > Mantissa) + continue; - // Keep track if every use in UsersToProcess is an address. If they all are, - // we may be able to rewrite the entire collection of them in terms of a - // smaller-stride IV. - bool AllUsesAreAddresses = true; - - // Keep track if every use of a single stride is outside the loop. If so, - // we want to be more aggressive about reusing a smaller-stride IV; a - // multiply outside the loop is better than another IV inside. Well, usually. - bool AllUsesAreOutsideLoop = true; - - // Transform our list of users and offsets to a bit more complex table. In - // this new vector, each 'BasedUser' contains 'Base' the base of the strided - // access as well as the old information from Uses. We progressively move - // information from the Base field to the Imm field until we eventually have - // the full access expression to rewrite the use. - std::vector UsersToProcess; - const SCEV *CommonExprs = CollectIVUsers(Stride, Uses, L, AllUsesAreAddresses, - AllUsesAreOutsideLoop, - UsersToProcess); - - // Sort the UsersToProcess array so that users with common bases are - // next to each other. - SortUsersToProcess(UsersToProcess); - - // If we managed to find some expressions in common, we'll need to carry - // their value in a register and add it in for each use. This will take up - // a register operand, which potentially restricts what stride values are - // valid. - bool HaveCommonExprs = !CommonExprs->isZero(); - const Type *ReplacedTy = CommonExprs->getType(); - - // If all uses are addresses, consider sinking the immediate part of the - // common expression back into uses if they can fit in the immediate fields. - if (TLI && HaveCommonExprs && AllUsesAreAddresses) { - const SCEV *NewCommon = CommonExprs; - const SCEV *Imm = SE->getIntegerSCEV(0, ReplacedTy); - MoveImmediateValues(TLI, Type::getVoidTy( - L->getLoopPreheader()->getContext()), - NewCommon, Imm, true, L, SE); - if (!Imm->isZero()) { - bool DoSink = true; - - // If the immediate part of the common expression is a GV, check if it's - // possible to fold it into the target addressing mode. - GlobalValue *GV = 0; - if (const SCEVUnknown *SU = dyn_cast(Imm)) - GV = dyn_cast(SU->getValue()); - int64_t Offset = 0; - if (const SCEVConstant *SC = dyn_cast(Imm)) - Offset = SC->getValue()->getSExtValue(); - if (GV || Offset) - // Pass VoidTy as the AccessTy to be conservative, because - // there could be multiple access types among all the uses. - DoSink = IsImmFoldedIntoAddrMode(GV, Offset, - Type::getVoidTy(L->getLoopPreheader()->getContext()), - UsersToProcess, TLI); - - if (DoSink) { - DEBUG(dbgs() << " Sinking " << *Imm << " back down into uses\n"); - for (unsigned i = 0, e = UsersToProcess.size(); i != e; ++i) - UsersToProcess[i].Imm = SE->getAddExpr(UsersToProcess[i].Imm, Imm); - CommonExprs = NewCommon; - HaveCommonExprs = !CommonExprs->isZero(); - ++NumImmSunk; - } + unsigned Entry, Latch; + if (PH->getIncomingBlock(0) == L->getLoopPreheader()) { + Entry = 0; + Latch = 1; + } else { + Entry = 1; + Latch = 0; } - } - - // Now that we know what we need to do, insert the PHI node itself. - // - DEBUG(dbgs() << "LSR: Examining IVs of TYPE " << *ReplacedTy << " of STRIDE " - << *Stride << ":\n" - << " Common base: " << *CommonExprs << '\n'); - SCEVExpander Rewriter(*SE); - SCEVExpander PreheaderRewriter(*SE); + ConstantInt *Init = dyn_cast(PH->getIncomingValue(Entry)); + if (!Init) continue; + Constant *NewInit = ConstantFP::get(DestTy, Init->getZExtValue()); - BasicBlock *Preheader = L->getLoopPreheader(); - Instruction *PreInsertPt = Preheader->getTerminator(); - BasicBlock *LatchBlock = L->getLoopLatch(); - Instruction *IVIncInsertPt = LatchBlock->getTerminator(); - - Value *CommonBaseV = Constant::getNullValue(ReplacedTy); - - const SCEV *RewriteFactor = SE->getIntegerSCEV(0, ReplacedTy); - IVExpr ReuseIV(SE->getIntegerSCEV(0, - Type::getInt32Ty(Preheader->getContext())), - SE->getIntegerSCEV(0, - Type::getInt32Ty(Preheader->getContext())), - 0); - - // Choose a strength-reduction strategy and prepare for it by creating - // the necessary PHIs and adjusting the bookkeeping. - if (ShouldUseFullStrengthReductionMode(UsersToProcess, L, - AllUsesAreAddresses, Stride)) { - PrepareToStrengthReduceFully(UsersToProcess, Stride, CommonExprs, L, - PreheaderRewriter); - } else { - // Emit the initial base value into the loop preheader. - CommonBaseV = PreheaderRewriter.expandCodeFor(CommonExprs, ReplacedTy, - PreInsertPt); - - // If all uses are addresses, check if it is possible to reuse an IV. The - // new IV must have a stride that is a multiple of the old stride; the - // multiple must be a number that can be encoded in the scale field of the - // target addressing mode; and we must have a valid instruction after this - // substitution, including the immediate field, if any. - RewriteFactor = CheckForIVReuse(HaveCommonExprs, AllUsesAreAddresses, - AllUsesAreOutsideLoop, - Stride, ReuseIV, ReplacedTy, - UsersToProcess); - if (!RewriteFactor->isZero()) - PrepareToStrengthReduceFromSmallerStride(UsersToProcess, CommonBaseV, - ReuseIV, PreInsertPt); - else { - IVIncInsertPt = FindIVIncInsertPt(UsersToProcess, L); - PrepareToStrengthReduceWithNewPhi(UsersToProcess, Stride, CommonExprs, - CommonBaseV, IVIncInsertPt, - L, PreheaderRewriter); - } - } - - // Process all the users now, replacing their strided uses with - // strength-reduced forms. This outer loop handles all bases, the inner - // loop handles all users of a particular base. - while (!UsersToProcess.empty()) { - const SCEV *Base = UsersToProcess.back().Base; - Instruction *Inst = UsersToProcess.back().Inst; - - // Emit the code for Base into the preheader. - Value *BaseV = 0; - if (!Base->isZero()) { - BaseV = PreheaderRewriter.expandCodeFor(Base, 0, PreInsertPt); - - DEBUG(dbgs() << " INSERTING code for BASE = " << *Base << ":"); - if (BaseV->hasName()) - DEBUG(dbgs() << " Result value name = %" << BaseV->getName()); - DEBUG(dbgs() << "\n"); - - // If BaseV is a non-zero constant, make sure that it gets inserted into - // the preheader, instead of being forward substituted into the uses. We - // do this by forcing a BitCast (noop cast) to be inserted into the - // preheader in this case. - if (!fitsInAddressMode(Base, getAccessType(Inst), TLI, false) && - isa(BaseV)) { - // We want this constant emitted into the preheader! This is just - // using cast as a copy so BitCast (no-op cast) is appropriate - BaseV = new BitCastInst(BaseV, BaseV->getType(), "preheaderinsert", - PreInsertPt); - } - } + BinaryOperator *Incr = + dyn_cast(PH->getIncomingValue(Latch)); + if (!Incr) continue; + if (Incr->getOpcode() != Instruction::Add + && Incr->getOpcode() != Instruction::Sub) + continue; - // Emit the code to add the immediate offset to the Phi value, just before - // the instructions that we identified as using this stride and base. - do { - // FIXME: Use emitted users to emit other users. - BasedUser &User = UsersToProcess.back(); + /* Initialize new IV, double d = 0.0 in above example. */ + ConstantInt *C = NULL; + if (Incr->getOperand(0) == PH) + C = dyn_cast(Incr->getOperand(1)); + else if (Incr->getOperand(1) == PH) + C = dyn_cast(Incr->getOperand(0)); + else + continue; - DEBUG(dbgs() << " Examining "); - if (User.isUseOfPostIncrementedValue) - DEBUG(dbgs() << "postinc"); - else - DEBUG(dbgs() << "preinc"); - DEBUG(dbgs() << " use "); - DEBUG(WriteAsOperand(dbgs(), UsersToProcess.back().OperandValToReplace, - /*PrintType=*/false)); - DEBUG(dbgs() << " in Inst: " << *User.Inst << '\n'); - - // If this instruction wants to use the post-incremented value, move it - // after the post-inc and use its value instead of the PHI. - Value *RewriteOp = User.Phi; - if (User.isUseOfPostIncrementedValue) { - RewriteOp = User.Phi->getIncomingValueForBlock(LatchBlock); - // If this user is in the loop, make sure it is the last thing in the - // loop to ensure it is dominated by the increment. In case it's the - // only use of the iv, the increment instruction is already before the - // use. - if (L->contains(User.Inst) && User.Inst != IVIncInsertPt) - User.Inst->moveBefore(IVIncInsertPt); - } + if (!C) continue; - const SCEV *RewriteExpr = SE->getUnknown(RewriteOp); + // Ignore negative constants, as the code below doesn't handle them + // correctly. TODO: Remove this restriction. + if (!C->getValue().isStrictlyPositive()) continue; - if (SE->getEffectiveSCEVType(RewriteOp->getType()) != - SE->getEffectiveSCEVType(ReplacedTy)) { - assert(SE->getTypeSizeInBits(RewriteOp->getType()) > - SE->getTypeSizeInBits(ReplacedTy) && - "Unexpected widening cast!"); - RewriteExpr = SE->getTruncateExpr(RewriteExpr, ReplacedTy); - } + /* Add new PHINode. */ + PHINode *NewPH = PHINode::Create(DestTy, "IV.S.", PH); - // If we had to insert new instructions for RewriteOp, we have to - // consider that they may not have been able to end up immediately - // next to RewriteOp, because non-PHI instructions may never precede - // PHI instructions in a block. In this case, remember where the last - // instruction was inserted so that if we're replacing a different - // PHI node, we can use the later point to expand the final - // RewriteExpr. - Instruction *NewBasePt = dyn_cast(RewriteOp); - if (RewriteOp == User.Phi) NewBasePt = 0; - - // Clear the SCEVExpander's expression map so that we are guaranteed - // to have the code emitted where we expect it. - Rewriter.clear(); - - // If we are reusing the iv, then it must be multiplied by a constant - // factor to take advantage of the addressing mode scale component. - if (!RewriteFactor->isZero()) { - // If we're reusing an IV with a nonzero base (currently this happens - // only when all reuses are outside the loop) subtract that base here. - // The base has been used to initialize the PHI node but we don't want - // it here. - if (!ReuseIV.Base->isZero()) { - const SCEV *typedBase = ReuseIV.Base; - if (SE->getEffectiveSCEVType(RewriteExpr->getType()) != - SE->getEffectiveSCEVType(ReuseIV.Base->getType())) { - // It's possible the original IV is a larger type than the new IV, - // in which case we have to truncate the Base. We checked in - // RequiresTypeConversion that this is valid. - assert(SE->getTypeSizeInBits(RewriteExpr->getType()) < - SE->getTypeSizeInBits(ReuseIV.Base->getType()) && - "Unexpected lengthening conversion!"); - typedBase = SE->getTruncateExpr(ReuseIV.Base, - RewriteExpr->getType()); - } - RewriteExpr = SE->getMinusSCEV(RewriteExpr, typedBase); - } + /* create new increment. '++d' in above example. */ + Constant *CFP = ConstantFP::get(DestTy, C->getZExtValue()); + BinaryOperator *NewIncr = + BinaryOperator::Create(Incr->getOpcode() == Instruction::Add ? + Instruction::FAdd : Instruction::FSub, + NewPH, CFP, "IV.S.next.", Incr); - // Multiply old variable, with base removed, by new scale factor. - RewriteExpr = SE->getMulExpr(RewriteFactor, - RewriteExpr); - - // The common base is emitted in the loop preheader. But since we - // are reusing an IV, it has not been used to initialize the PHI node. - // Add it to the expression used to rewrite the uses. - // When this use is outside the loop, we earlier subtracted the - // common base, and are adding it back here. Use the same expression - // as before, rather than CommonBaseV, so DAGCombiner will zap it. - if (!CommonExprs->isZero()) { - if (L->contains(User.Inst)) - RewriteExpr = SE->getAddExpr(RewriteExpr, - SE->getUnknown(CommonBaseV)); - else - RewriteExpr = SE->getAddExpr(RewriteExpr, CommonExprs); - } - } + NewPH->addIncoming(NewInit, PH->getIncomingBlock(Entry)); + NewPH->addIncoming(NewIncr, PH->getIncomingBlock(Latch)); - // Now that we know what we need to do, insert code before User for the - // immediate and any loop-variant expressions. - if (BaseV) - // Add BaseV to the PHI value if needed. - RewriteExpr = SE->getAddExpr(RewriteExpr, SE->getUnknown(BaseV)); - - User.RewriteInstructionToUseNewBase(RewriteExpr, NewBasePt, - Rewriter, L, this, - DeadInsts, SE); - - // Mark old value we replaced as possibly dead, so that it is eliminated - // if we just replaced the last use of that value. - DeadInsts.push_back(User.OperandValToReplace); - - UsersToProcess.pop_back(); - ++NumReduced; - - // If there are any more users to process with the same base, process them - // now. We sorted by base above, so we just have to check the last elt. - } while (!UsersToProcess.empty() && UsersToProcess.back().Base == Base); - // TODO: Next, find out which base index is the most common, pull it out. - } - - // IMPORTANT TODO: Figure out how to partition the IV's with this stride, but - // different starting values, into different PHIs. -} - -void LoopStrengthReduce::StrengthReduceIVUsers(Loop *L) { - // Note: this processes each stride/type pair individually. All users - // passed into StrengthReduceIVUsersOfStride have the same type AND stride. - // Also, note that we iterate over IVUsesByStride indirectly by using - // StrideOrder. This extra layer of indirection makes the ordering of - // strides deterministic - not dependent on map order. - for (unsigned Stride = 0, e = IU->StrideOrder.size(); Stride != e; ++Stride) { - std::map::iterator SI = - IU->IVUsesByStride.find(IU->StrideOrder[Stride]); - assert(SI != IU->IVUsesByStride.end() && "Stride doesn't exist!"); - // FIXME: Generalize to non-affine IV's. - if (!SI->first->isLoopInvariant(L)) - continue; - StrengthReduceIVUsersOfStride(SI->first, *SI->second, L); + /* Remove cast operation */ + ShadowUse->replaceAllUsesWith(NewPH); + ShadowUse->eraseFromParent(); + break; } } /// FindIVUserForCond - If Cond has an operand that is an expression of an IV, /// set the IV user and stride information and return true, otherwise return /// false. -bool LoopStrengthReduce::FindIVUserForCond(ICmpInst *Cond, - IVStrideUse *&CondUse, - const SCEV* &CondStride) { - for (unsigned Stride = 0, e = IU->StrideOrder.size(); - Stride != e && !CondUse; ++Stride) { - std::map::iterator SI = - IU->IVUsesByStride.find(IU->StrideOrder[Stride]); - assert(SI != IU->IVUsesByStride.end() && "Stride doesn't exist!"); - - for (ilist::iterator UI = SI->second->Users.begin(), - E = SI->second->Users.end(); UI != E; ++UI) - if (UI->getUser() == Cond) { - // NOTE: we could handle setcc instructions with multiple uses here, but - // InstCombine does it as well for simple uses, it's not clear that it - // occurs enough in real life to handle. - CondUse = UI; - CondStride = SI->first; - return true; - } - } - return false; -} - -namespace { - // Constant strides come first which in turns are sorted by their absolute - // values. If absolute values are the same, then positive strides comes first. - // e.g. - // 4, -1, X, 1, 2 ==> 1, -1, 2, 4, X - struct StrideCompare { - const ScalarEvolution *SE; - explicit StrideCompare(const ScalarEvolution *se) : SE(se) {} - - bool operator()(const SCEV *LHS, const SCEV *RHS) { - const SCEVConstant *LHSC = dyn_cast(LHS); - const SCEVConstant *RHSC = dyn_cast(RHS); - if (LHSC && RHSC) { - int64_t LV = LHSC->getValue()->getSExtValue(); - int64_t RV = RHSC->getValue()->getSExtValue(); - uint64_t ALV = (LV < 0) ? -LV : LV; - uint64_t ARV = (RV < 0) ? -RV : RV; - if (ALV == ARV) { - if (LV != RV) - return LV > RV; - } else { - return ALV < ARV; - } - - // If it's the same value but different type, sort by bit width so - // that we emit larger induction variables before smaller - // ones, letting the smaller be re-written in terms of larger ones. - return SE->getTypeSizeInBits(RHS->getType()) < - SE->getTypeSizeInBits(LHS->getType()); - } - return LHSC && !RHSC; - } - }; -} - -/// ChangeCompareStride - If a loop termination compare instruction is the only -/// use of its stride, and the comparison is against a constant value, try to -/// eliminate the stride by moving the compare instruction to another stride and -/// changing its constant operand accordingly. E.g. -/// -/// loop: -/// ... -/// v1 = v1 + 3 -/// v2 = v2 + 1 -/// if (v2 < 10) goto loop -/// => -/// loop: -/// ... -/// v1 = v1 + 3 -/// if (v1 < 30) goto loop -ICmpInst *LoopStrengthReduce::ChangeCompareStride(Loop *L, ICmpInst *Cond, - IVStrideUse* &CondUse, - const SCEV* &CondStride, - bool PostPass) { - // If there's only one stride in the loop, there's nothing to do here. - if (IU->StrideOrder.size() < 2) - return Cond; - - // If there are other users of the condition's stride, don't bother trying to - // change the condition because the stride will still remain. - std::map::iterator I = - IU->IVUsesByStride.find(CondStride); - if (I == IU->IVUsesByStride.end()) - return Cond; - - if (I->second->Users.size() > 1) { - for (ilist::iterator II = I->second->Users.begin(), - EE = I->second->Users.end(); II != EE; ++II) { - if (II->getUser() == Cond) - continue; - if (!isInstructionTriviallyDead(II->getUser())) - return Cond; - } - } - - // Only handle constant strides for now. - const SCEVConstant *SC = dyn_cast(CondStride); - if (!SC) return Cond; - - ICmpInst::Predicate Predicate = Cond->getPredicate(); - int64_t CmpSSInt = SC->getValue()->getSExtValue(); - unsigned BitWidth = SE->getTypeSizeInBits(CondStride->getType()); - uint64_t SignBit = 1ULL << (BitWidth-1); - const Type *CmpTy = Cond->getOperand(0)->getType(); - const Type *NewCmpTy = NULL; - unsigned TyBits = SE->getTypeSizeInBits(CmpTy); - unsigned NewTyBits = 0; - const SCEV *NewStride = NULL; - Value *NewCmpLHS = NULL; - Value *NewCmpRHS = NULL; - int64_t Scale = 1; - const SCEV *NewOffset = SE->getIntegerSCEV(0, CmpTy); - - if (ConstantInt *C = dyn_cast(Cond->getOperand(1))) { - int64_t CmpVal = C->getValue().getSExtValue(); - - // Check the relevant induction variable for conformance to the pattern. - const SCEV *IV = SE->getSCEV(Cond->getOperand(0)); - const SCEVAddRecExpr *AR = dyn_cast(IV); - if (!AR || !AR->isAffine()) - return Cond; - - const SCEVConstant *StartC = dyn_cast(AR->getStart()); - // Check stride constant and the comparision constant signs to detect - // overflow. - if (StartC) { - if ((StartC->getValue()->getSExtValue() < CmpVal && CmpSSInt < 0) || - (StartC->getValue()->getSExtValue() > CmpVal && CmpSSInt > 0)) - return Cond; - } else { - // More restrictive check for the other cases. - if ((CmpVal & SignBit) != (CmpSSInt & SignBit)) - return Cond; - } - - // Look for a suitable stride / iv as replacement. - for (unsigned i = 0, e = IU->StrideOrder.size(); i != e; ++i) { - std::map::iterator SI = - IU->IVUsesByStride.find(IU->StrideOrder[i]); - if (!isa(SI->first) || SI->second->Users.empty()) - continue; - int64_t SSInt = cast(SI->first)->getValue()->getSExtValue(); - if (SSInt == CmpSSInt || - abs64(SSInt) < abs64(CmpSSInt) || - (SSInt % CmpSSInt) != 0) - continue; - - Scale = SSInt / CmpSSInt; - int64_t NewCmpVal = CmpVal * Scale; - - // If old icmp value fits in icmp immediate field, but the new one doesn't - // try something else. - if (TLI && - TLI->isLegalICmpImmediate(CmpVal) && - !TLI->isLegalICmpImmediate(NewCmpVal)) - continue; - - APInt Mul = APInt(BitWidth*2, CmpVal, true); - Mul = Mul * APInt(BitWidth*2, Scale, true); - // Check for overflow. - if (!Mul.isSignedIntN(BitWidth)) - continue; - // Check for overflow in the stride's type too. - if (!Mul.isSignedIntN(SE->getTypeSizeInBits(SI->first->getType()))) - continue; - - // Watch out for overflow. - if (ICmpInst::isSigned(Predicate) && - (CmpVal & SignBit) != (NewCmpVal & SignBit)) - continue; - - // Pick the best iv to use trying to avoid a cast. - NewCmpLHS = NULL; - for (ilist::iterator UI = SI->second->Users.begin(), - E = SI->second->Users.end(); UI != E; ++UI) { - Value *Op = UI->getOperandValToReplace(); - - // If the IVStrideUse implies a cast, check for an actual cast which - // can be used to find the original IV expression. - if (SE->getEffectiveSCEVType(Op->getType()) != - SE->getEffectiveSCEVType(SI->first->getType())) { - CastInst *CI = dyn_cast(Op); - // If it's not a simple cast, it's complicated. - if (!CI) - continue; - // If it's a cast from a type other than the stride type, - // it's complicated. - if (CI->getOperand(0)->getType() != SI->first->getType()) - continue; - // Ok, we found the IV expression in the stride's type. - Op = CI->getOperand(0); - } - - NewCmpLHS = Op; - if (NewCmpLHS->getType() == CmpTy) - break; - } - if (!NewCmpLHS) - continue; - - NewCmpTy = NewCmpLHS->getType(); - NewTyBits = SE->getTypeSizeInBits(NewCmpTy); - const Type *NewCmpIntTy = IntegerType::get(Cond->getContext(), NewTyBits); - if (RequiresTypeConversion(NewCmpTy, CmpTy)) { - // Check if it is possible to rewrite it using - // an iv / stride of a smaller integer type. - unsigned Bits = NewTyBits; - if (ICmpInst::isSigned(Predicate)) - --Bits; - uint64_t Mask = (1ULL << Bits) - 1; - if (((uint64_t)NewCmpVal & Mask) != (uint64_t)NewCmpVal) - continue; - } - - // Don't rewrite if use offset is non-constant and the new type is - // of a different type. - // FIXME: too conservative? - if (NewTyBits != TyBits && !isa(CondUse->getOffset())) - continue; - - if (!PostPass) { - bool AllUsesAreAddresses = true; - bool AllUsesAreOutsideLoop = true; - std::vector UsersToProcess; - const SCEV *CommonExprs = CollectIVUsers(SI->first, *SI->second, L, - AllUsesAreAddresses, - AllUsesAreOutsideLoop, - UsersToProcess); - // Avoid rewriting the compare instruction with an iv of new stride - // if it's likely the new stride uses will be rewritten using the - // stride of the compare instruction. - if (AllUsesAreAddresses && - ValidScale(!CommonExprs->isZero(), Scale, UsersToProcess)) - continue; - } - - // Avoid rewriting the compare instruction with an iv which has - // implicit extension or truncation built into it. - // TODO: This is over-conservative. - if (SE->getTypeSizeInBits(CondUse->getOffset()->getType()) != TyBits) - continue; - - // If scale is negative, use swapped predicate unless it's testing - // for equality. - if (Scale < 0 && !Cond->isEquality()) - Predicate = ICmpInst::getSwappedPredicate(Predicate); - - NewStride = IU->StrideOrder[i]; - if (!isa(NewCmpTy)) - NewCmpRHS = ConstantInt::get(NewCmpTy, NewCmpVal); - else { - Constant *CI = ConstantInt::get(NewCmpIntTy, NewCmpVal); - NewCmpRHS = ConstantExpr::getIntToPtr(CI, NewCmpTy); - } - NewOffset = TyBits == NewTyBits - ? SE->getMulExpr(CondUse->getOffset(), - SE->getConstant(CmpTy, Scale)) - : SE->getConstant(NewCmpIntTy, - cast(CondUse->getOffset())->getValue() - ->getSExtValue()*Scale); - break; +bool LSRInstance::FindIVUserForCond(ICmpInst *Cond, + IVStrideUse *&CondUse) { + for (IVUsers::iterator UI = IU.begin(), E = IU.end(); UI != E; ++UI) + if (UI->getUser() == Cond) { + // NOTE: we could handle setcc instructions with multiple uses here, but + // InstCombine does it as well for simple uses, it's not clear that it + // occurs enough in real life to handle. + CondUse = UI; + return true; } - } - - // Forgo this transformation if it the increment happens to be - // unfortunately positioned after the condition, and the condition - // has multiple uses which prevent it from being moved immediately - // before the branch. See - // test/Transforms/LoopStrengthReduce/change-compare-stride-trickiness-*.ll - // for an example of this situation. - if (!Cond->hasOneUse()) { - for (BasicBlock::iterator I = Cond, E = Cond->getParent()->end(); - I != E; ++I) - if (I == NewCmpLHS) - return Cond; - } - - if (NewCmpRHS) { - // Create a new compare instruction using new stride / iv. - ICmpInst *OldCond = Cond; - // Insert new compare instruction. - Cond = new ICmpInst(OldCond, Predicate, NewCmpLHS, NewCmpRHS, - L->getHeader()->getName() + ".termcond"); - - DEBUG(dbgs() << " Change compare stride in Inst " << *OldCond); - DEBUG(dbgs() << " to " << *Cond << '\n'); - - // Remove the old compare instruction. The old indvar is probably dead too. - DeadInsts.push_back(CondUse->getOperandValToReplace()); - OldCond->replaceAllUsesWith(Cond); - OldCond->eraseFromParent(); - - IU->IVUsesByStride[NewStride]->addUser(NewOffset, Cond, NewCmpLHS); - CondUse = &IU->IVUsesByStride[NewStride]->Users.back(); - CondStride = NewStride; - ++NumEliminated; - Changed = true; - } - - return Cond; + return false; } /// OptimizeMax - Rewrite the loop's terminating condition if it uses @@ -2088,7 +1405,7 @@ ICmpInst *LoopStrengthReduce::ChangeCompareStride(Loop *L, ICmpInst *Cond, /// are designed around them. The most obvious example of this is the /// LoopInfo analysis, which doesn't remember trip count values. It /// expects to be able to rediscover the trip count each time it is -/// needed, and it does this using a simple analyis that only succeeds if +/// needed, and it does this using a simple analysis that only succeeds if /// the loop has a canonical induction variable. /// /// However, when it comes time to generate code, the maximum operation @@ -2098,8 +1415,7 @@ ICmpInst *LoopStrengthReduce::ChangeCompareStride(Loop *L, ICmpInst *Cond, /// rewriting their conditions from ICMP_NE back to ICMP_SLT, and deleting /// the instructions for the maximum computation. /// -ICmpInst *LoopStrengthReduce::OptimizeMax(Loop *L, ICmpInst *Cond, - IVStrideUse* &CondUse) { +ICmpInst *LSRInstance::OptimizeMax(ICmpInst *Cond, IVStrideUse* &CondUse) { // Check that the loop matches the pattern we're looking for. if (Cond->getPredicate() != CmpInst::ICMP_EQ && Cond->getPredicate() != CmpInst::ICMP_NE) @@ -2108,19 +1424,19 @@ ICmpInst *LoopStrengthReduce::OptimizeMax(Loop *L, ICmpInst *Cond, SelectInst *Sel = dyn_cast(Cond->getOperand(1)); if (!Sel || !Sel->hasOneUse()) return Cond; - const SCEV *BackedgeTakenCount = SE->getBackedgeTakenCount(L); + const SCEV *BackedgeTakenCount = SE.getBackedgeTakenCount(L); if (isa(BackedgeTakenCount)) return Cond; - const SCEV *One = SE->getIntegerSCEV(1, BackedgeTakenCount->getType()); + const SCEV *One = SE.getIntegerSCEV(1, BackedgeTakenCount->getType()); // Add one to the backedge-taken count to get the trip count. - const SCEV *IterationCount = SE->getAddExpr(BackedgeTakenCount, One); + const SCEV *IterationCount = SE.getAddExpr(BackedgeTakenCount, One); // Check for a max calculation that matches the pattern. if (!isa(IterationCount) && !isa(IterationCount)) return Cond; const SCEVNAryExpr *Max = cast(IterationCount); - if (Max != SE->getSCEV(Sel)) return Cond; + if (Max != SE.getSCEV(Sel)) return Cond; // To handle a max with more than two operands, this optimization would // require additional checking and setup. @@ -2130,14 +1446,13 @@ ICmpInst *LoopStrengthReduce::OptimizeMax(Loop *L, ICmpInst *Cond, const SCEV *MaxLHS = Max->getOperand(0); const SCEV *MaxRHS = Max->getOperand(1); if (!MaxLHS || MaxLHS != One) return Cond; - // Check the relevant induction variable for conformance to // the pattern. - const SCEV *IV = SE->getSCEV(Cond->getOperand(0)); + const SCEV *IV = SE.getSCEV(Cond->getOperand(0)); const SCEVAddRecExpr *AR = dyn_cast(IV); if (!AR || !AR->isAffine() || AR->getStart() != One || - AR->getStepRecurrence(*SE) != One) + AR->getStepRecurrence(SE) != One) return Cond; assert(AR->getLoop() == L && @@ -2146,9 +1461,9 @@ ICmpInst *LoopStrengthReduce::OptimizeMax(Loop *L, ICmpInst *Cond, // Check the right operand of the select, and remember it, as it will // be used in the new comparison instruction. Value *NewRHS = 0; - if (SE->getSCEV(Sel->getOperand(1)) == MaxRHS) + if (SE.getSCEV(Sel->getOperand(1)) == MaxRHS) NewRHS = Sel->getOperand(1); - else if (SE->getSCEV(Sel->getOperand(2)) == MaxRHS) + else if (SE.getSCEV(Sel->getOperand(2)) == MaxRHS) NewRHS = Sel->getOperand(2); if (!NewRHS) return Cond; @@ -2175,249 +1490,20 @@ ICmpInst *LoopStrengthReduce::OptimizeMax(Loop *L, ICmpInst *Cond, return NewCond; } -/// OptimizeShadowIV - If IV is used in a int-to-float cast -/// inside the loop then try to eliminate the cast opeation. -void LoopStrengthReduce::OptimizeShadowIV(Loop *L) { - - const SCEV *BackedgeTakenCount = SE->getBackedgeTakenCount(L); - if (isa(BackedgeTakenCount)) - return; - - for (unsigned Stride = 0, e = IU->StrideOrder.size(); Stride != e; - ++Stride) { - std::map::iterator SI = - IU->IVUsesByStride.find(IU->StrideOrder[Stride]); - assert(SI != IU->IVUsesByStride.end() && "Stride doesn't exist!"); - if (!isa(SI->first)) - continue; - - for (ilist::iterator UI = SI->second->Users.begin(), - E = SI->second->Users.end(); UI != E; /* empty */) { - ilist::iterator CandidateUI = UI; - ++UI; - Instruction *ShadowUse = CandidateUI->getUser(); - const Type *DestTy = NULL; - - /* If shadow use is a int->float cast then insert a second IV - to eliminate this cast. - - for (unsigned i = 0; i < n; ++i) - foo((double)i); - - is transformed into - - double d = 0.0; - for (unsigned i = 0; i < n; ++i, ++d) - foo(d); - */ - if (UIToFPInst *UCast = dyn_cast(CandidateUI->getUser())) - DestTy = UCast->getDestTy(); - else if (SIToFPInst *SCast = dyn_cast(CandidateUI->getUser())) - DestTy = SCast->getDestTy(); - if (!DestTy) continue; - - if (TLI) { - // If target does not support DestTy natively then do not apply - // this transformation. - EVT DVT = TLI->getValueType(DestTy); - if (!TLI->isTypeLegal(DVT)) continue; - } - - PHINode *PH = dyn_cast(ShadowUse->getOperand(0)); - if (!PH) continue; - if (PH->getNumIncomingValues() != 2) continue; - - const Type *SrcTy = PH->getType(); - int Mantissa = DestTy->getFPMantissaWidth(); - if (Mantissa == -1) continue; - if ((int)SE->getTypeSizeInBits(SrcTy) > Mantissa) - continue; - - unsigned Entry, Latch; - if (PH->getIncomingBlock(0) == L->getLoopPreheader()) { - Entry = 0; - Latch = 1; - } else { - Entry = 1; - Latch = 0; - } - - ConstantInt *Init = dyn_cast(PH->getIncomingValue(Entry)); - if (!Init) continue; - Constant *NewInit = ConstantFP::get(DestTy, Init->getZExtValue()); - - BinaryOperator *Incr = - dyn_cast(PH->getIncomingValue(Latch)); - if (!Incr) continue; - if (Incr->getOpcode() != Instruction::Add - && Incr->getOpcode() != Instruction::Sub) - continue; - - /* Initialize new IV, double d = 0.0 in above example. */ - ConstantInt *C = NULL; - if (Incr->getOperand(0) == PH) - C = dyn_cast(Incr->getOperand(1)); - else if (Incr->getOperand(1) == PH) - C = dyn_cast(Incr->getOperand(0)); - else - continue; - - if (!C) continue; - - // Ignore negative constants, as the code below doesn't handle them - // correctly. TODO: Remove this restriction. - if (!C->getValue().isStrictlyPositive()) continue; - - /* Add new PHINode. */ - PHINode *NewPH = PHINode::Create(DestTy, "IV.S.", PH); - - /* create new increment. '++d' in above example. */ - Constant *CFP = ConstantFP::get(DestTy, C->getZExtValue()); - BinaryOperator *NewIncr = - BinaryOperator::Create(Incr->getOpcode() == Instruction::Add ? - Instruction::FAdd : Instruction::FSub, - NewPH, CFP, "IV.S.next.", Incr); - - NewPH->addIncoming(NewInit, PH->getIncomingBlock(Entry)); - NewPH->addIncoming(NewIncr, PH->getIncomingBlock(Latch)); - - /* Remove cast operation */ - ShadowUse->replaceAllUsesWith(NewPH); - ShadowUse->eraseFromParent(); - NumShadow++; - break; - } - } -} - -/// OptimizeIndvars - Now that IVUsesByStride is set up with all of the indvar -/// uses in the loop, look to see if we can eliminate some, in favor of using -/// common indvars for the different uses. -void LoopStrengthReduce::OptimizeIndvars(Loop *L) { - // TODO: implement optzns here. - - OptimizeShadowIV(L); -} - -bool LoopStrengthReduce::StrideMightBeShared(const SCEV* Stride, Loop *L, - bool CheckPreInc) { - int64_t SInt = cast(Stride)->getValue()->getSExtValue(); - for (unsigned i = 0, e = IU->StrideOrder.size(); i != e; ++i) { - std::map::iterator SI = - IU->IVUsesByStride.find(IU->StrideOrder[i]); - const SCEV *Share = SI->first; - if (!isa(SI->first) || Share == Stride) - continue; - int64_t SSInt = cast(Share)->getValue()->getSExtValue(); - if (SSInt == SInt) - return true; // This can definitely be reused. - if (unsigned(abs64(SSInt)) < SInt || (SSInt % SInt) != 0) - continue; - int64_t Scale = SSInt / SInt; - bool AllUsesAreAddresses = true; - bool AllUsesAreOutsideLoop = true; - std::vector UsersToProcess; - const SCEV *CommonExprs = CollectIVUsers(SI->first, *SI->second, L, - AllUsesAreAddresses, - AllUsesAreOutsideLoop, - UsersToProcess); - if (AllUsesAreAddresses && - ValidScale(!CommonExprs->isZero(), Scale, UsersToProcess)) { - if (!CheckPreInc) - return true; - // Any pre-inc iv use? - IVUsersOfOneStride &StrideUses = *IU->IVUsesByStride[Share]; - for (ilist::iterator I = StrideUses.Users.begin(), - E = StrideUses.Users.end(); I != E; ++I) { - if (!I->isUseOfPostIncrementedValue()) - return true; - } - } - } - return false; -} - -/// isUsedByExitBranch - Return true if icmp is used by a loop terminating -/// conditional branch or it's and / or with other conditions before being used -/// as the condition. -static bool isUsedByExitBranch(ICmpInst *Cond, Loop *L) { - BasicBlock *CondBB = Cond->getParent(); - if (!L->isLoopExiting(CondBB)) - return false; - BranchInst *TermBr = dyn_cast(CondBB->getTerminator()); - if (!TermBr || !TermBr->isConditional()) - return false; - - Value *User = *Cond->use_begin(); - Instruction *UserInst = dyn_cast(User); - while (UserInst && - (UserInst->getOpcode() == Instruction::And || - UserInst->getOpcode() == Instruction::Or)) { - if (!UserInst->hasOneUse() || UserInst->getParent() != CondBB) - return false; - User = *User->use_begin(); - UserInst = dyn_cast(User); - } - return User == TermBr; -} - -static bool ShouldCountToZero(ICmpInst *Cond, IVStrideUse* &CondUse, - ScalarEvolution *SE, Loop *L, - const TargetLowering *TLI = 0) { - if (!L->contains(Cond)) - return false; - - if (!isa(CondUse->getOffset())) - return false; - - // Handle only tests for equality for the moment. - if (!Cond->isEquality() || !Cond->hasOneUse()) - return false; - if (!isUsedByExitBranch(Cond, L)) - return false; - - Value *CondOp0 = Cond->getOperand(0); - const SCEV *IV = SE->getSCEV(CondOp0); - const SCEVAddRecExpr *AR = dyn_cast(IV); - if (!AR || !AR->isAffine()) - return false; - - const SCEVConstant *SC = dyn_cast(AR->getStepRecurrence(*SE)); - if (!SC || SC->getValue()->getSExtValue() < 0) - // If it's already counting down, don't do anything. - return false; - - // If the RHS of the comparison is not an loop invariant, the rewrite - // cannot be done. Also bail out if it's already comparing against a zero. - // If we are checking this before cmp stride optimization, check if it's - // comparing against a already legal immediate. - Value *RHS = Cond->getOperand(1); - ConstantInt *RHSC = dyn_cast(RHS); - if (!L->isLoopInvariant(RHS) || - (RHSC && RHSC->isZero()) || - (RHSC && TLI && TLI->isLegalICmpImmediate(RHSC->getSExtValue()))) - return false; - - // Make sure the IV is only used for counting. Value may be preinc or - // postinc; 2 uses in either case. - if (!CondOp0->hasNUses(2)) - return false; - - return true; -} - /// OptimizeLoopTermCond - Change loop terminating condition to use the /// postinc iv when possible. -void LoopStrengthReduce::OptimizeLoopTermCond(Loop *L) { +bool +LSRInstance::OptimizeLoopTermCond() { + SmallPtrSet PostIncs; + BasicBlock *LatchBlock = L->getLoopLatch(); - bool LatchExit = L->isLoopExiting(LatchBlock); SmallVector ExitingBlocks; L->getExitingBlocks(ExitingBlocks); for (unsigned i = 0, e = ExitingBlocks.size(); i != e; ++i) { BasicBlock *ExitingBlock = ExitingBlocks[i]; - // Finally, get the terminating condition for the loop if possible. If we + // Get the terminating condition for the loop if possible. If we // can, we want to change it to use a post-incremented version of its // induction variable, to allow coalescing the live ranges for the IV into // one register value. @@ -2431,297 +1517,1758 @@ void LoopStrengthReduce::OptimizeLoopTermCond(Loop *L) { // Search IVUsesByStride to find Cond's IVUse if there is one. IVStrideUse *CondUse = 0; - const SCEV *CondStride = 0; ICmpInst *Cond = cast(TermBr->getCondition()); - if (!FindIVUserForCond(Cond, CondUse, CondStride)) + if (!FindIVUserForCond(Cond, CondUse)) continue; - // If the latch block is exiting and it's not a single block loop, it's - // not safe to use postinc iv in other exiting blocks. FIXME: overly - // conservative? How about icmp stride optimization? - bool UsePostInc = !(e > 1 && LatchExit && ExitingBlock != LatchBlock); - if (UsePostInc && ExitingBlock != LatchBlock) { - if (!Cond->hasOneUse()) - // See below, we don't want the condition to be cloned. - UsePostInc = false; - else { - // If exiting block is the latch block, we know it's safe and profitable - // to transform the icmp to use post-inc iv. Otherwise do so only if it - // would not reuse another iv and its iv would be reused by other uses. - // We are optimizing for the case where the icmp is the only use of the - // iv. - IVUsersOfOneStride &StrideUses = *IU->IVUsesByStride[CondStride]; - for (ilist::iterator I = StrideUses.Users.begin(), - E = StrideUses.Users.end(); I != E; ++I) { - if (I->getUser() == Cond) - continue; - if (!I->isUseOfPostIncrementedValue()) { - UsePostInc = false; - break; - } - } - } - - // If iv for the stride might be shared and any of the users use pre-inc - // iv might be used, then it's not safe to use post-inc iv. - if (UsePostInc && - isa(CondStride) && - StrideMightBeShared(CondStride, L, true)) - UsePostInc = false; - } - // If the trip count is computed in terms of a max (due to ScalarEvolution // being unable to find a sufficient guard, for example), change the loop // comparison to use SLT or ULT instead of NE. - Cond = OptimizeMax(L, Cond, CondUse); - - // If possible, change stride and operands of the compare instruction to - // eliminate one stride. However, avoid rewriting the compare instruction - // with an iv of new stride if it's likely the new stride uses will be - // rewritten using the stride of the compare instruction. - if (ExitingBlock == LatchBlock && isa(CondStride)) { - // If the condition stride is a constant and it's the only use, we might - // want to optimize it first by turning it to count toward zero. - if (!StrideMightBeShared(CondStride, L, false) && - !ShouldCountToZero(Cond, CondUse, SE, L, TLI)) - Cond = ChangeCompareStride(L, Cond, CondUse, CondStride); - } - - if (!UsePostInc) + // One consequence of doing this now is that it disrupts the count-down + // optimization. That's not always a bad thing though, because in such + // cases it may still be worthwhile to avoid a max. + Cond = OptimizeMax(Cond, CondUse); + + // If this exiting block dominates the latch block, it may also use + // the post-inc value if it won't be shared with other uses. + // Check for dominance. + if (!DT.dominates(ExitingBlock, LatchBlock)) continue; + // Conservatively avoid trying to use the post-inc value in non-latch + // exits if there may be pre-inc users in intervening blocks. + if (LatchBlock != ExitingBlock) + for (IVUsers::const_iterator UI = IU.begin(), E = IU.end(); UI != E; ++UI) + // Test if the use is reachable from the exiting block. This dominator + // query is a conservative approximation of reachability. + if (&*UI != CondUse && + !DT.properlyDominates(UI->getUser()->getParent(), ExitingBlock)) { + // Conservatively assume there may be reuse if the quotient of their + // strides could be a legal scale. + const SCEV *A = CondUse->getStride(); + const SCEV *B = UI->getStride(); + if (SE.getTypeSizeInBits(A->getType()) != + SE.getTypeSizeInBits(B->getType())) { + if (SE.getTypeSizeInBits(A->getType()) > + SE.getTypeSizeInBits(B->getType())) + B = SE.getSignExtendExpr(B, A->getType()); + else + A = SE.getSignExtendExpr(A, B->getType()); + } + if (const SCEVConstant *D = + dyn_cast_or_null(getExactSDiv(B, A, SE))) { + // Stride of one or negative one can have reuse with non-addresses. + if (D->getValue()->isOne() || + D->getValue()->isAllOnesValue()) + goto decline_post_inc; + // Avoid weird situations. + if (D->getValue()->getValue().getMinSignedBits() >= 64 || + D->getValue()->getValue().isMinSignedValue()) + goto decline_post_inc; + // Without TLI, assume that any stride might be valid, and so any + // use might be shared. + if (!TLI) + goto decline_post_inc; + // Check for possible scaled-address reuse. + const Type *AccessTy = getAccessType(UI->getUser()); + TargetLowering::AddrMode AM; + AM.Scale = D->getValue()->getSExtValue(); + if (TLI->isLegalAddressingMode(AM, AccessTy)) + goto decline_post_inc; + AM.Scale = -AM.Scale; + if (TLI->isLegalAddressingMode(AM, AccessTy)) + goto decline_post_inc; + } + } + DEBUG(dbgs() << " Change loop exiting icmp to use postinc iv: " - << *Cond << '\n'); + << *Cond << '\n'); // It's possible for the setcc instruction to be anywhere in the loop, and // possible for it to have multiple users. If it is not immediately before // the exiting block branch, move it. - if (&*++BasicBlock::iterator(Cond) != (Instruction*)TermBr) { - if (Cond->hasOneUse()) { // Condition has a single use, just move it. + if (&*++BasicBlock::iterator(Cond) != TermBr) { + if (Cond->hasOneUse()) { Cond->moveBefore(TermBr); } else { - // Otherwise, clone the terminating condition and insert into the - // loopend. + // Clone the terminating condition and insert into the loopend. + ICmpInst *OldCond = Cond; Cond = cast(Cond->clone()); Cond->setName(L->getHeader()->getName() + ".termcond"); ExitingBlock->getInstList().insert(TermBr, Cond); // Clone the IVUse, as the old use still exists! - IU->IVUsesByStride[CondStride]->addUser(CondUse->getOffset(), Cond, - CondUse->getOperandValToReplace()); - CondUse = &IU->IVUsesByStride[CondStride]->Users.back(); + CondUse = &IU.AddUser(CondUse->getStride(), CondUse->getOffset(), + Cond, CondUse->getOperandValToReplace()); + TermBr->replaceUsesOfWith(OldCond, Cond); } } // If we get to here, we know that we can transform the setcc instruction to // use the post-incremented version of the IV, allowing us to coalesce the // live ranges for the IV correctly. - CondUse->setOffset(SE->getMinusSCEV(CondUse->getOffset(), CondStride)); + CondUse->setOffset(SE.getMinusSCEV(CondUse->getOffset(), + CondUse->getStride())); CondUse->setIsUseOfPostIncrementedValue(true); Changed = true; - ++NumLoopCond; + PostIncs.insert(Cond); + decline_post_inc:; } -} - -bool LoopStrengthReduce::OptimizeLoopCountIVOfStride(const SCEV* &Stride, - IVStrideUse* &CondUse, - Loop *L) { - // If the only use is an icmp of a loop exiting conditional branch, then - // attempt the optimization. - BasedUser User = BasedUser(*CondUse, SE); - assert(isa(User.Inst) && "Expecting an ICMPInst!"); - ICmpInst *Cond = cast(User.Inst); - // Less strict check now that compare stride optimization is done. - if (!ShouldCountToZero(Cond, CondUse, SE, L)) - return false; + // Determine an insertion point for the loop induction variable increment. It + // must dominate all the post-inc comparisons we just set up, and it must + // dominate the loop latch edge. + IVIncInsertPos = L->getLoopLatch()->getTerminator(); + for (SmallPtrSet::const_iterator I = PostIncs.begin(), + E = PostIncs.end(); I != E; ++I) { + BasicBlock *BB = + DT.findNearestCommonDominator(IVIncInsertPos->getParent(), + (*I)->getParent()); + if (BB == (*I)->getParent()) + IVIncInsertPos = *I; + else if (BB != IVIncInsertPos->getParent()) + IVIncInsertPos = BB->getTerminator(); + } - Value *CondOp0 = Cond->getOperand(0); - PHINode *PHIExpr = dyn_cast(CondOp0); - Instruction *Incr; - if (!PHIExpr) { - // Value tested is postinc. Find the phi node. - Incr = dyn_cast(CondOp0); - // FIXME: Just use User.OperandValToReplace here? - if (!Incr || Incr->getOpcode() != Instruction::Add) - return false; + return Changed; +} - PHIExpr = dyn_cast(Incr->getOperand(0)); - if (!PHIExpr) - return false; - // 1 use for preinc value, the increment. - if (!PHIExpr->hasOneUse()) +bool +LSRInstance::reconcileNewOffset(LSRUse &LU, int64_t NewOffset, + LSRUse::KindType Kind, const Type *AccessTy) { + int64_t NewMinOffset = LU.MinOffset; + int64_t NewMaxOffset = LU.MaxOffset; + const Type *NewAccessTy = AccessTy; + + // Check for a mismatched kind. It's tempting to collapse mismatched kinds to + // something conservative, however this can pessimize in the case that one of + // the uses will have all its uses outside the loop, for example. + if (LU.Kind != Kind) + return false; + // Conservatively assume HasBaseReg is true for now. + if (NewOffset < LU.MinOffset) { + if (!isAlwaysFoldable(LU.MaxOffset - NewOffset, 0, /*HasBaseReg=*/true, + Kind, AccessTy, TLI)) return false; - } else { - assert(isa(CondOp0) && - "Unexpected loop exiting counting instruction sequence!"); - PHIExpr = cast(CondOp0); - // Value tested is preinc. Find the increment. - // A CmpInst is not a BinaryOperator; we depend on this. - Instruction::use_iterator UI = PHIExpr->use_begin(); - Incr = dyn_cast(UI); - if (!Incr) - Incr = dyn_cast(++UI); - // One use for postinc value, the phi. Unnecessarily conservative? - if (!Incr || !Incr->hasOneUse() || Incr->getOpcode() != Instruction::Add) + NewMinOffset = NewOffset; + } else if (NewOffset > LU.MaxOffset) { + if (!isAlwaysFoldable(NewOffset - LU.MinOffset, 0, /*HasBaseReg=*/true, + Kind, AccessTy, TLI)) return false; + NewMaxOffset = NewOffset; } + // Check for a mismatched access type, and fall back conservatively as needed. + if (Kind == LSRUse::Address && AccessTy != LU.AccessTy) + NewAccessTy = Type::getVoidTy(AccessTy->getContext()); + + // Update the use. + LU.MinOffset = NewMinOffset; + LU.MaxOffset = NewMaxOffset; + LU.AccessTy = NewAccessTy; + if (NewOffset != LU.Offsets.back()) + LU.Offsets.push_back(NewOffset); + return true; +} - // Replace the increment with a decrement. - DEBUG(dbgs() << "LSR: Examining use "); - DEBUG(WriteAsOperand(dbgs(), CondOp0, /*PrintType=*/false)); - DEBUG(dbgs() << " in Inst: " << *Cond << '\n'); - BinaryOperator *Decr = BinaryOperator::Create(Instruction::Sub, - Incr->getOperand(0), Incr->getOperand(1), "tmp", Incr); - Incr->replaceAllUsesWith(Decr); - Incr->eraseFromParent(); - - // Substitute endval-startval for the original startval, and 0 for the - // original endval. Since we're only testing for equality this is OK even - // if the computation wraps around. - BasicBlock *Preheader = L->getLoopPreheader(); - Instruction *PreInsertPt = Preheader->getTerminator(); - unsigned InBlock = L->contains(PHIExpr->getIncomingBlock(0)) ? 1 : 0; - Value *StartVal = PHIExpr->getIncomingValue(InBlock); - Value *EndVal = Cond->getOperand(1); - DEBUG(dbgs() << " Optimize loop counting iv to count down [" - << *EndVal << " .. " << *StartVal << "]\n"); - - // FIXME: check for case where both are constant. - Constant* Zero = ConstantInt::get(Cond->getOperand(1)->getType(), 0); - BinaryOperator *NewStartVal = BinaryOperator::Create(Instruction::Sub, - EndVal, StartVal, "tmp", PreInsertPt); - PHIExpr->setIncomingValue(InBlock, NewStartVal); - Cond->setOperand(1, Zero); - DEBUG(dbgs() << " New icmp: " << *Cond << "\n"); - - int64_t SInt = cast(Stride)->getValue()->getSExtValue(); - const SCEV *NewStride = 0; - bool Found = false; - for (unsigned i = 0, e = IU->StrideOrder.size(); i != e; ++i) { - const SCEV *OldStride = IU->StrideOrder[i]; - if (const SCEVConstant *SC = dyn_cast(OldStride)) - if (SC->getValue()->getSExtValue() == -SInt) { - Found = true; - NewStride = OldStride; - break; - } +/// getUse - Return an LSRUse index and an offset value for a fixup which +/// needs the given expression, with the given kind and optional access type. +/// Either reuse an existing use or create a new one, as needed. +std::pair +LSRInstance::getUse(const SCEV *&Expr, + LSRUse::KindType Kind, const Type *AccessTy) { + const SCEV *Copy = Expr; + int64_t Offset = ExtractImmediate(Expr, SE); + + // Basic uses can't accept any offset, for example. + if (!isAlwaysFoldable(Offset, 0, /*HasBaseReg=*/true, Kind, AccessTy, TLI)) { + Expr = Copy; + Offset = 0; } - if (!Found) - NewStride = SE->getIntegerSCEV(-SInt, Stride->getType()); - IU->AddUser(NewStride, CondUse->getOffset(), Cond, Cond->getOperand(0)); - IU->IVUsesByStride[Stride]->removeUser(CondUse); + std::pair P = + UseMap.insert(std::make_pair(Expr, 0)); + if (!P.second) { + // A use already existed with this base. + size_t LUIdx = P.first->second; + LSRUse &LU = Uses[LUIdx]; + if (reconcileNewOffset(LU, Offset, Kind, AccessTy)) + // Reuse this use. + return std::make_pair(LUIdx, Offset); + } - CondUse = &IU->IVUsesByStride[NewStride]->Users.back(); - Stride = NewStride; + // Create a new use. + size_t LUIdx = Uses.size(); + P.first->second = LUIdx; + Uses.push_back(LSRUse(Kind, AccessTy)); + LSRUse &LU = Uses[LUIdx]; - ++NumCountZero; + // We don't need to track redundant offsets, but we don't need to go out + // of our way here to avoid them. + if (LU.Offsets.empty() || Offset != LU.Offsets.back()) + LU.Offsets.push_back(Offset); - return true; + LU.MinOffset = Offset; + LU.MaxOffset = Offset; + return std::make_pair(LUIdx, Offset); } -/// OptimizeLoopCountIV - If, after all sharing of IVs, the IV used for deciding -/// when to exit the loop is used only for that purpose, try to rearrange things -/// so it counts down to a test against zero. -bool LoopStrengthReduce::OptimizeLoopCountIV(Loop *L) { - bool ThisChanged = false; - for (unsigned i = 0, e = IU->StrideOrder.size(); i != e; ++i) { - const SCEV *Stride = IU->StrideOrder[i]; - std::map::iterator SI = - IU->IVUsesByStride.find(Stride); - assert(SI != IU->IVUsesByStride.end() && "Stride doesn't exist!"); - // FIXME: Generalize to non-affine IV's. - if (!SI->first->isLoopInvariant(L)) - continue; - // If stride is a constant and it has an icmpinst use, check if we can - // optimize the loop to count down. - if (isa(Stride) && SI->second->Users.size() == 1) { - Instruction *User = SI->second->Users.begin()->getUser(); - if (!isa(User)) - continue; - const SCEV *CondStride = Stride; - IVStrideUse *Use = &*SI->second->Users.begin(); - if (!OptimizeLoopCountIVOfStride(CondStride, Use, L)) - continue; - ThisChanged = true; +void LSRInstance::CollectInterestingTypesAndFactors() { + SmallSetVector Strides; - // Now check if it's possible to reuse this iv for other stride uses. - for (unsigned j = 0, ee = IU->StrideOrder.size(); j != ee; ++j) { - const SCEV *SStride = IU->StrideOrder[j]; - if (SStride == CondStride) - continue; - std::map::iterator SII = - IU->IVUsesByStride.find(SStride); - assert(SII != IU->IVUsesByStride.end() && "Stride doesn't exist!"); - // FIXME: Generalize to non-affine IV's. - if (!SII->first->isLoopInvariant(L)) - continue; - // FIXME: Rewrite other stride using CondStride. + // Collect interesting types and strides. + for (IVUsers::const_iterator UI = IU.begin(), E = IU.end(); UI != E; ++UI) { + const SCEV *Stride = UI->getStride(); + + // Collect interesting types. + Types.insert(SE.getEffectiveSCEVType(Stride->getType())); + + // Add the stride for this loop. + Strides.insert(Stride); + + // Add strides for other mentioned loops. + for (const SCEVAddRecExpr *AR = dyn_cast(UI->getOffset()); + AR; AR = dyn_cast(AR->getStart())) + Strides.insert(AR->getStepRecurrence(SE)); + } + + // Compute interesting factors from the set of interesting strides. + for (SmallSetVector::const_iterator + I = Strides.begin(), E = Strides.end(); I != E; ++I) + for (SmallSetVector::const_iterator NewStrideIter = + next(I); NewStrideIter != E; ++NewStrideIter) { + const SCEV *OldStride = *I; + const SCEV *NewStride = *NewStrideIter; + + if (SE.getTypeSizeInBits(OldStride->getType()) != + SE.getTypeSizeInBits(NewStride->getType())) { + if (SE.getTypeSizeInBits(OldStride->getType()) > + SE.getTypeSizeInBits(NewStride->getType())) + NewStride = SE.getSignExtendExpr(NewStride, OldStride->getType()); + else + OldStride = SE.getSignExtendExpr(OldStride, NewStride->getType()); + } + if (const SCEVConstant *Factor = + dyn_cast_or_null(getExactSDiv(NewStride, OldStride, + SE, true))) { + if (Factor->getValue()->getValue().getMinSignedBits() <= 64) + Factors.insert(Factor->getValue()->getValue().getSExtValue()); + } else if (const SCEVConstant *Factor = + dyn_cast_or_null(getExactSDiv(OldStride, + NewStride, + SE, true))) { + if (Factor->getValue()->getValue().getMinSignedBits() <= 64) + Factors.insert(Factor->getValue()->getValue().getSExtValue()); + } + } + + // If all uses use the same type, don't bother looking for truncation-based + // reuse. + if (Types.size() == 1) + Types.clear(); + + DEBUG(print_factors_and_types(dbgs())); +} + +void LSRInstance::CollectFixupsAndInitialFormulae() { + for (IVUsers::const_iterator UI = IU.begin(), E = IU.end(); UI != E; ++UI) { + // Record the uses. + LSRFixup &LF = getNewFixup(); + LF.UserInst = UI->getUser(); + LF.OperandValToReplace = UI->getOperandValToReplace(); + if (UI->isUseOfPostIncrementedValue()) + LF.PostIncLoop = L; + + LSRUse::KindType Kind = LSRUse::Basic; + const Type *AccessTy = 0; + if (isAddressUse(LF.UserInst, LF.OperandValToReplace)) { + Kind = LSRUse::Address; + AccessTy = getAccessType(LF.UserInst); + } + + const SCEV *S = IU.getCanonicalExpr(*UI); + + // Equality (== and !=) ICmps are special. We can rewrite (i == N) as + // (N - i == 0), and this allows (N - i) to be the expression that we work + // with rather than just N or i, so we can consider the register + // requirements for both N and i at the same time. Limiting this code to + // equality icmps is not a problem because all interesting loops use + // equality icmps, thanks to IndVarSimplify. + if (ICmpInst *CI = dyn_cast(LF.UserInst)) + if (CI->isEquality()) { + // Swap the operands if needed to put the OperandValToReplace on the + // left, for consistency. + Value *NV = CI->getOperand(1); + if (NV == LF.OperandValToReplace) { + CI->setOperand(1, CI->getOperand(0)); + CI->setOperand(0, NV); + } + + // x == y --> x - y == 0 + const SCEV *N = SE.getSCEV(NV); + if (N->isLoopInvariant(L)) { + Kind = LSRUse::ICmpZero; + S = SE.getMinusSCEV(N, S); + } + + // -1 and the negations of all interesting strides (except the negation + // of -1) are now also interesting. + for (size_t i = 0, e = Factors.size(); i != e; ++i) + if (Factors[i] != -1) + Factors.insert(-(uint64_t)Factors[i]); + Factors.insert(-1); } + + // Set up the initial formula for this use. + std::pair P = getUse(S, Kind, AccessTy); + LF.LUIdx = P.first; + LF.Offset = P.second; + LSRUse &LU = Uses[LF.LUIdx]; + LU.AllFixupsOutsideLoop &= !L->contains(LF.UserInst); + + // If this is the first use of this LSRUse, give it a formula. + if (LU.Formulae.empty()) { + InsertInitialFormula(S, LU, LF.LUIdx); + CountRegisters(LU.Formulae.back(), LF.LUIdx); } } - Changed |= ThisChanged; - return ThisChanged; + DEBUG(print_fixups(dbgs())); +} + +void +LSRInstance::InsertInitialFormula(const SCEV *S, LSRUse &LU, size_t LUIdx) { + Formula F; + F.InitialMatch(S, L, SE, DT); + bool Inserted = InsertFormula(LU, LUIdx, F); + assert(Inserted && "Initial formula already exists!"); (void)Inserted; } -bool LoopStrengthReduce::runOnLoop(Loop *L, LPPassManager &LPM) { - IU = &getAnalysis(); - SE = &getAnalysis(); - Changed = false; +void +LSRInstance::InsertSupplementalFormula(const SCEV *S, + LSRUse &LU, size_t LUIdx) { + Formula F; + F.BaseRegs.push_back(S); + F.AM.HasBaseReg = true; + bool Inserted = InsertFormula(LU, LUIdx, F); + assert(Inserted && "Supplemental formula already exists!"); (void)Inserted; +} - // If LoopSimplify form is not available, stay out of trouble. - if (!L->getLoopPreheader() || !L->getLoopLatch()) +/// CountRegisters - Note which registers are used by the given formula, +/// updating RegUses. +void LSRInstance::CountRegisters(const Formula &F, size_t LUIdx) { + if (F.ScaledReg) + RegUses.CountRegister(F.ScaledReg, LUIdx); + for (SmallVectorImpl::const_iterator I = F.BaseRegs.begin(), + E = F.BaseRegs.end(); I != E; ++I) + RegUses.CountRegister(*I, LUIdx); +} + +/// InsertFormula - If the given formula has not yet been inserted, add it to +/// the list, and return true. Return false otherwise. +bool LSRInstance::InsertFormula(LSRUse &LU, unsigned LUIdx, const Formula &F) { + if (!LU.InsertFormula(F)) return false; - if (!IU->IVUsesByStride.empty()) { - DEBUG(dbgs() << "\nLSR on \"" << L->getHeader()->getParent()->getName() - << "\" "; - L->print(dbgs())); + CountRegisters(F, LUIdx); + return true; +} + +/// CollectLoopInvariantFixupsAndFormulae - Check for other uses of +/// loop-invariant values which we're tracking. These other uses will pin these +/// values in registers, making them less profitable for elimination. +/// TODO: This currently misses non-constant addrec step registers. +/// TODO: Should this give more weight to users inside the loop? +void +LSRInstance::CollectLoopInvariantFixupsAndFormulae() { + SmallVector Worklist(RegUses.begin(), RegUses.end()); + SmallPtrSet Inserted; + + while (!Worklist.empty()) { + const SCEV *S = Worklist.pop_back_val(); + + if (const SCEVNAryExpr *N = dyn_cast(S)) + Worklist.insert(Worklist.end(), N->op_begin(), N->op_end()); + else if (const SCEVCastExpr *C = dyn_cast(S)) + Worklist.push_back(C->getOperand()); + else if (const SCEVUDivExpr *D = dyn_cast(S)) { + Worklist.push_back(D->getLHS()); + Worklist.push_back(D->getRHS()); + } else if (const SCEVUnknown *U = dyn_cast(S)) { + if (!Inserted.insert(U)) continue; + const Value *V = U->getValue(); + if (const Instruction *Inst = dyn_cast(V)) + if (L->contains(Inst)) continue; + for (Value::use_const_iterator UI = V->use_begin(), UE = V->use_end(); + UI != UE; ++UI) { + const Instruction *UserInst = dyn_cast(*UI); + // Ignore non-instructions. + if (!UserInst) + continue; + // Ignore instructions in other functions (as can happen with + // Constants). + if (UserInst->getParent()->getParent() != L->getHeader()->getParent()) + continue; + // Ignore instructions not dominated by the loop. + const BasicBlock *UseBB = !isa(UserInst) ? + UserInst->getParent() : + cast(UserInst)->getIncomingBlock( + PHINode::getIncomingValueNumForOperand(UI.getOperandNo())); + if (!DT.dominates(L->getHeader(), UseBB)) + continue; + // Ignore uses which are part of other SCEV expressions, to avoid + // analyzing them multiple times. + if (SE.isSCEVable(UserInst->getType()) && + !isa(SE.getSCEV(const_cast(UserInst)))) + continue; + // Ignore icmp instructions which are already being analyzed. + if (const ICmpInst *ICI = dyn_cast(UserInst)) { + unsigned OtherIdx = !UI.getOperandNo(); + Value *OtherOp = const_cast(ICI->getOperand(OtherIdx)); + if (SE.getSCEV(OtherOp)->hasComputableLoopEvolution(L)) + continue; + } + + LSRFixup &LF = getNewFixup(); + LF.UserInst = const_cast(UserInst); + LF.OperandValToReplace = UI.getUse(); + std::pair P = getUse(S, LSRUse::Basic, 0); + LF.LUIdx = P.first; + LF.Offset = P.second; + LSRUse &LU = Uses[LF.LUIdx]; + LU.AllFixupsOutsideLoop &= L->contains(LF.UserInst); + InsertSupplementalFormula(U, LU, LF.LUIdx); + CountRegisters(LU.Formulae.back(), Uses.size() - 1); + break; + } + } + } +} + +/// CollectSubexprs - Split S into subexpressions which can be pulled out into +/// separate registers. If C is non-null, multiply each subexpression by C. +static void CollectSubexprs(const SCEV *S, const SCEVConstant *C, + SmallVectorImpl &Ops, + ScalarEvolution &SE) { + if (const SCEVAddExpr *Add = dyn_cast(S)) { + // Break out add operands. + for (SCEVAddExpr::op_iterator I = Add->op_begin(), E = Add->op_end(); + I != E; ++I) + CollectSubexprs(*I, C, Ops, SE); + return; + } else if (const SCEVAddRecExpr *AR = dyn_cast(S)) { + // Split a non-zero base out of an addrec. + if (!AR->getStart()->isZero()) { + CollectSubexprs(SE.getAddRecExpr(SE.getIntegerSCEV(0, AR->getType()), + AR->getStepRecurrence(SE), + AR->getLoop()), C, Ops, SE); + CollectSubexprs(AR->getStart(), C, Ops, SE); + return; + } + } else if (const SCEVMulExpr *Mul = dyn_cast(S)) { + // Break (C * (a + b + c)) into C*a + C*b + C*c. + if (Mul->getNumOperands() == 2) + if (const SCEVConstant *Op0 = + dyn_cast(Mul->getOperand(0))) { + CollectSubexprs(Mul->getOperand(1), + C ? cast(SE.getMulExpr(C, Op0)) : Op0, + Ops, SE); + return; + } + } + + // Otherwise use the value itself. + Ops.push_back(C ? SE.getMulExpr(C, S) : S); +} + +/// GenerateReassociations - Split out subexpressions from adds and the bases of +/// addrecs. +void LSRInstance::GenerateReassociations(LSRUse &LU, unsigned LUIdx, + Formula Base, + unsigned Depth) { + // Arbitrarily cap recursion to protect compile time. + if (Depth >= 3) return; + + for (size_t i = 0, e = Base.BaseRegs.size(); i != e; ++i) { + const SCEV *BaseReg = Base.BaseRegs[i]; + + SmallVector AddOps; + CollectSubexprs(BaseReg, 0, AddOps, SE); + if (AddOps.size() == 1) continue; + + for (SmallVectorImpl::const_iterator J = AddOps.begin(), + JE = AddOps.end(); J != JE; ++J) { + // Don't pull a constant into a register if the constant could be folded + // into an immediate field. + if (isAlwaysFoldable(*J, LU.MinOffset, LU.MaxOffset, + Base.getNumRegs() > 1, + LU.Kind, LU.AccessTy, TLI, SE)) + continue; + + // Collect all operands except *J. + SmallVector InnerAddOps; + for (SmallVectorImpl::const_iterator K = AddOps.begin(), + KE = AddOps.end(); K != KE; ++K) + if (K != J) + InnerAddOps.push_back(*K); + + // Don't leave just a constant behind in a register if the constant could + // be folded into an immediate field. + if (InnerAddOps.size() == 1 && + isAlwaysFoldable(InnerAddOps[0], LU.MinOffset, LU.MaxOffset, + Base.getNumRegs() > 1, + LU.Kind, LU.AccessTy, TLI, SE)) + continue; + + Formula F = Base; + F.BaseRegs[i] = SE.getAddExpr(InnerAddOps); + F.BaseRegs.push_back(*J); + if (InsertFormula(LU, LUIdx, F)) + // If that formula hadn't been seen before, recurse to find more like + // it. + GenerateReassociations(LU, LUIdx, LU.Formulae.back(), Depth+1); + } + } +} + +/// GenerateCombinations - Generate a formula consisting of all of the +/// loop-dominating registers added into a single register. +void LSRInstance::GenerateCombinations(LSRUse &LU, unsigned LUIdx, + Formula Base) { + // This method is only interesting on a plurality of registers. + if (Base.BaseRegs.size() <= 1) return; + + Formula F = Base; + F.BaseRegs.clear(); + SmallVector Ops; + for (SmallVectorImpl::const_iterator + I = Base.BaseRegs.begin(), E = Base.BaseRegs.end(); I != E; ++I) { + const SCEV *BaseReg = *I; + if (BaseReg->properlyDominates(L->getHeader(), &DT) && + !BaseReg->hasComputableLoopEvolution(L)) + Ops.push_back(BaseReg); + else + F.BaseRegs.push_back(BaseReg); + } + if (Ops.size() > 1) { + const SCEV *Sum = SE.getAddExpr(Ops); + // TODO: If Sum is zero, it probably means ScalarEvolution missed an + // opportunity to fold something. For now, just ignore such cases + // rather than proceed with zero in a register. + if (!Sum->isZero()) { + F.BaseRegs.push_back(Sum); + (void)InsertFormula(LU, LUIdx, F); + } + } +} + +/// GenerateSymbolicOffsets - Generate reuse formulae using symbolic offsets. +void LSRInstance::GenerateSymbolicOffsets(LSRUse &LU, unsigned LUIdx, + Formula Base) { + // We can't add a symbolic offset if the address already contains one. + if (Base.AM.BaseGV) return; + + for (size_t i = 0, e = Base.BaseRegs.size(); i != e; ++i) { + const SCEV *G = Base.BaseRegs[i]; + GlobalValue *GV = ExtractSymbol(G, SE); + if (G->isZero() || !GV) + continue; + Formula F = Base; + F.AM.BaseGV = GV; + if (!isLegalUse(F.AM, LU.MinOffset, LU.MaxOffset, + LU.Kind, LU.AccessTy, TLI)) + continue; + F.BaseRegs[i] = G; + (void)InsertFormula(LU, LUIdx, F); + } +} + +/// GenerateConstantOffsets - Generate reuse formulae using symbolic offsets. +void LSRInstance::GenerateConstantOffsets(LSRUse &LU, unsigned LUIdx, + Formula Base) { + // TODO: For now, just add the min and max offset, because it usually isn't + // worthwhile looking at everything inbetween. + SmallVector Worklist; + Worklist.push_back(LU.MinOffset); + if (LU.MaxOffset != LU.MinOffset) + Worklist.push_back(LU.MaxOffset); + + for (size_t i = 0, e = Base.BaseRegs.size(); i != e; ++i) { + const SCEV *G = Base.BaseRegs[i]; + + for (SmallVectorImpl::const_iterator I = Worklist.begin(), + E = Worklist.end(); I != E; ++I) { + Formula F = Base; + F.AM.BaseOffs = (uint64_t)Base.AM.BaseOffs - *I; + if (isLegalUse(F.AM, LU.MinOffset - *I, LU.MaxOffset - *I, + LU.Kind, LU.AccessTy, TLI)) { + F.BaseRegs[i] = SE.getAddExpr(G, SE.getIntegerSCEV(*I, G->getType())); + + (void)InsertFormula(LU, LUIdx, F); + } + } + + int64_t Imm = ExtractImmediate(G, SE); + if (G->isZero() || Imm == 0) + continue; + Formula F = Base; + F.AM.BaseOffs = (uint64_t)F.AM.BaseOffs + Imm; + if (!isLegalUse(F.AM, LU.MinOffset, LU.MaxOffset, + LU.Kind, LU.AccessTy, TLI)) + continue; + F.BaseRegs[i] = G; + (void)InsertFormula(LU, LUIdx, F); + } +} + +/// GenerateICmpZeroScales - For ICmpZero, check to see if we can scale up +/// the comparison. For example, x == y -> x*c == y*c. +void LSRInstance::GenerateICmpZeroScales(LSRUse &LU, unsigned LUIdx, + Formula Base) { + if (LU.Kind != LSRUse::ICmpZero) return; + + // Determine the integer type for the base formula. + const Type *IntTy = Base.getType(); + if (!IntTy) return; + if (SE.getTypeSizeInBits(IntTy) > 64) return; + + // Don't do this if there is more than one offset. + if (LU.MinOffset != LU.MaxOffset) return; + + assert(!Base.AM.BaseGV && "ICmpZero use is not legal!"); + + // Check each interesting stride. + for (SmallSetVector::const_iterator + I = Factors.begin(), E = Factors.end(); I != E; ++I) { + int64_t Factor = *I; + Formula F = Base; + + // Check that the multiplication doesn't overflow. + if (F.AM.BaseOffs == INT64_MIN && Factor == -1) + continue; + F.AM.BaseOffs = (uint64_t)Base.AM.BaseOffs * Factor; + if (F.AM.BaseOffs / Factor != Base.AM.BaseOffs) + continue; + + // Check that multiplying with the use offset doesn't overflow. + int64_t Offset = LU.MinOffset; + if (Offset == INT64_MIN && Factor == -1) + continue; + Offset = (uint64_t)Offset * Factor; + if (Offset / Factor != LU.MinOffset) + continue; + + // Check that this scale is legal. + if (!isLegalUse(F.AM, Offset, Offset, LU.Kind, LU.AccessTy, TLI)) + continue; + + // Compensate for the use having MinOffset built into it. + F.AM.BaseOffs = (uint64_t)F.AM.BaseOffs + Offset - LU.MinOffset; + + const SCEV *FactorS = SE.getIntegerSCEV(Factor, IntTy); + + // Check that multiplying with each base register doesn't overflow. + for (size_t i = 0, e = F.BaseRegs.size(); i != e; ++i) { + F.BaseRegs[i] = SE.getMulExpr(F.BaseRegs[i], FactorS); + if (getExactSDiv(F.BaseRegs[i], FactorS, SE) != Base.BaseRegs[i]) + goto next; + } + + // Check that multiplying with the scaled register doesn't overflow. + if (F.ScaledReg) { + F.ScaledReg = SE.getMulExpr(F.ScaledReg, FactorS); + if (getExactSDiv(F.ScaledReg, FactorS, SE) != Base.ScaledReg) + continue; + } + + // If we make it here and it's legal, add it. + (void)InsertFormula(LU, LUIdx, F); + next:; + } +} + +/// GenerateScales - Generate stride factor reuse formulae by making use of +/// scaled-offset address modes, for example. +void LSRInstance::GenerateScales(LSRUse &LU, unsigned LUIdx, + Formula Base) { + // Determine the integer type for the base formula. + const Type *IntTy = Base.getType(); + if (!IntTy) return; + + // If this Formula already has a scaled register, we can't add another one. + if (Base.AM.Scale != 0) return; + + // Check each interesting stride. + for (SmallSetVector::const_iterator + I = Factors.begin(), E = Factors.end(); I != E; ++I) { + int64_t Factor = *I; + + Base.AM.Scale = Factor; + Base.AM.HasBaseReg = Base.BaseRegs.size() > 1; + // Check whether this scale is going to be legal. + if (!isLegalUse(Base.AM, LU.MinOffset, LU.MaxOffset, + LU.Kind, LU.AccessTy, TLI)) { + // As a special-case, handle special out-of-loop Basic users specially. + // TODO: Reconsider this special case. + if (LU.Kind == LSRUse::Basic && + isLegalUse(Base.AM, LU.MinOffset, LU.MaxOffset, + LSRUse::Special, LU.AccessTy, TLI) && + LU.AllFixupsOutsideLoop) + LU.Kind = LSRUse::Special; + else + continue; + } + // For an ICmpZero, negating a solitary base register won't lead to + // new solutions. + if (LU.Kind == LSRUse::ICmpZero && + !Base.AM.HasBaseReg && Base.AM.BaseOffs == 0 && !Base.AM.BaseGV) + continue; + // For each addrec base reg, apply the scale, if possible. + for (size_t i = 0, e = Base.BaseRegs.size(); i != e; ++i) + if (const SCEVAddRecExpr *AR = + dyn_cast(Base.BaseRegs[i])) { + const SCEV *FactorS = SE.getIntegerSCEV(Factor, IntTy); + if (FactorS->isZero()) + continue; + // Divide out the factor, ignoring high bits, since we'll be + // scaling the value back up in the end. + if (const SCEV *Quotient = getExactSDiv(AR, FactorS, SE, true)) { + // TODO: This could be optimized to avoid all the copying. + Formula F = Base; + F.ScaledReg = Quotient; + std::swap(F.BaseRegs[i], F.BaseRegs.back()); + F.BaseRegs.pop_back(); + (void)InsertFormula(LU, LUIdx, F); + } + } + } +} + +/// GenerateTruncates - Generate reuse formulae from different IV types. +void LSRInstance::GenerateTruncates(LSRUse &LU, unsigned LUIdx, + Formula Base) { + // This requires TargetLowering to tell us which truncates are free. + if (!TLI) return; + + // Don't bother truncating symbolic values. + if (Base.AM.BaseGV) return; + + // Determine the integer type for the base formula. + const Type *DstTy = Base.getType(); + if (!DstTy) return; + DstTy = SE.getEffectiveSCEVType(DstTy); + + for (SmallSetVector::const_iterator + I = Types.begin(), E = Types.end(); I != E; ++I) { + const Type *SrcTy = *I; + if (SrcTy != DstTy && TLI->isTruncateFree(SrcTy, DstTy)) { + Formula F = Base; + + if (F.ScaledReg) F.ScaledReg = SE.getAnyExtendExpr(F.ScaledReg, *I); + for (SmallVectorImpl::iterator J = F.BaseRegs.begin(), + JE = F.BaseRegs.end(); J != JE; ++J) + *J = SE.getAnyExtendExpr(*J, SrcTy); + + // TODO: This assumes we've done basic processing on all uses and + // have an idea what the register usage is. + if (!F.hasRegsUsedByUsesOtherThan(LUIdx, RegUses)) + continue; + + (void)InsertFormula(LU, LUIdx, F); + } + } +} + +namespace { + +/// WorkItem - Helper class for GenerateCrossUseConstantOffsets. It's used to +/// defer modifications so that the search phase doesn't have to worry about +/// the data structures moving underneath it. +struct WorkItem { + size_t LUIdx; + int64_t Imm; + const SCEV *OrigReg; + + WorkItem(size_t LI, int64_t I, const SCEV *R) + : LUIdx(LI), Imm(I), OrigReg(R) {} + + void print(raw_ostream &OS) const; + void dump() const; +}; + +} + +void WorkItem::print(raw_ostream &OS) const { + OS << "in formulae referencing " << *OrigReg << " in use " << LUIdx + << " , add offset " << Imm; +} + +void WorkItem::dump() const { + print(errs()); errs() << '\n'; +} + +/// GenerateCrossUseConstantOffsets - Look for registers which are a constant +/// distance apart and try to form reuse opportunities between them. +void LSRInstance::GenerateCrossUseConstantOffsets() { + // Group the registers by their value without any added constant offset. + typedef std::map ImmMapTy; + typedef DenseMap RegMapTy; + RegMapTy Map; + DenseMap UsedByIndicesMap; + SmallVector Sequence; + for (RegUseTracker::const_iterator I = RegUses.begin(), E = RegUses.end(); + I != E; ++I) { + const SCEV *Reg = *I; + int64_t Imm = ExtractImmediate(Reg, SE); + std::pair Pair = + Map.insert(std::make_pair(Reg, ImmMapTy())); + if (Pair.second) + Sequence.push_back(Reg); + Pair.first->second.insert(std::make_pair(Imm, *I)); + UsedByIndicesMap[Reg] |= RegUses.getUsedByIndices(*I); + } + + // Now examine each set of registers with the same base value. Build up + // a list of work to do and do the work in a separate step so that we're + // not adding formulae and register counts while we're searching. + SmallVector WorkItems; + SmallSet, 32> UniqueItems; + for (SmallVectorImpl::const_iterator I = Sequence.begin(), + E = Sequence.end(); I != E; ++I) { + const SCEV *Reg = *I; + const ImmMapTy &Imms = Map.find(Reg)->second; + + // It's not worthwhile looking for reuse if there's only one offset. + if (Imms.size() == 1) + continue; + + DEBUG(dbgs() << "Generating cross-use offsets for " << *Reg << ':'; + for (ImmMapTy::const_iterator J = Imms.begin(), JE = Imms.end(); + J != JE; ++J) + dbgs() << ' ' << J->first; + dbgs() << '\n'); + + // Examine each offset. + for (ImmMapTy::const_iterator J = Imms.begin(), JE = Imms.end(); + J != JE; ++J) { + const SCEV *OrigReg = J->second; + + int64_t JImm = J->first; + const SmallBitVector &UsedByIndices = RegUses.getUsedByIndices(OrigReg); + + if (!isa(OrigReg) && + UsedByIndicesMap[Reg].count() == 1) { + DEBUG(dbgs() << "Skipping cross-use reuse for " << *OrigReg << '\n'); + continue; + } + + // Conservatively examine offsets between this orig reg a few selected + // other orig regs. + ImmMapTy::const_iterator OtherImms[] = { + Imms.begin(), prior(Imms.end()), + Imms.upper_bound((Imms.begin()->first + prior(Imms.end())->first) / 2) + }; + for (size_t i = 0, e = array_lengthof(OtherImms); i != e; ++i) { + ImmMapTy::const_iterator M = OtherImms[i]; + if (M == J || M == JE) continue; + + // Compute the difference between the two. + int64_t Imm = (uint64_t)JImm - M->first; + for (int LUIdx = UsedByIndices.find_first(); LUIdx != -1; + LUIdx = UsedByIndices.find_next(LUIdx)) + // Make a memo of this use, offset, and register tuple. + if (UniqueItems.insert(std::make_pair(LUIdx, Imm))) + WorkItems.push_back(WorkItem(LUIdx, Imm, OrigReg)); + } + } + } + + Map.clear(); + Sequence.clear(); + UsedByIndicesMap.clear(); + UniqueItems.clear(); + + // Now iterate through the worklist and add new formulae. + for (SmallVectorImpl::const_iterator I = WorkItems.begin(), + E = WorkItems.end(); I != E; ++I) { + const WorkItem &WI = *I; + size_t LUIdx = WI.LUIdx; + LSRUse &LU = Uses[LUIdx]; + int64_t Imm = WI.Imm; + const SCEV *OrigReg = WI.OrigReg; + + const Type *IntTy = SE.getEffectiveSCEVType(OrigReg->getType()); + const SCEV *NegImmS = SE.getSCEV(ConstantInt::get(IntTy, -(uint64_t)Imm)); + unsigned BitWidth = SE.getTypeSizeInBits(IntTy); + + // TODO: Use a more targeted data structure. + for (size_t L = 0, LE = LU.Formulae.size(); L != LE; ++L) { + Formula F = LU.Formulae[L]; + // Use the immediate in the scaled register. + if (F.ScaledReg == OrigReg) { + int64_t Offs = (uint64_t)F.AM.BaseOffs + + Imm * (uint64_t)F.AM.Scale; + // Don't create 50 + reg(-50). + if (F.referencesReg(SE.getSCEV( + ConstantInt::get(IntTy, -(uint64_t)Offs)))) + continue; + Formula NewF = F; + NewF.AM.BaseOffs = Offs; + if (!isLegalUse(NewF.AM, LU.MinOffset, LU.MaxOffset, + LU.Kind, LU.AccessTy, TLI)) + continue; + NewF.ScaledReg = SE.getAddExpr(NegImmS, NewF.ScaledReg); + + // If the new scale is a constant in a register, and adding the constant + // value to the immediate would produce a value closer to zero than the + // immediate itself, then the formula isn't worthwhile. + if (const SCEVConstant *C = dyn_cast(NewF.ScaledReg)) + if (C->getValue()->getValue().isNegative() != + (NewF.AM.BaseOffs < 0) && + (C->getValue()->getValue().abs() * APInt(BitWidth, F.AM.Scale)) + .ule(APInt(BitWidth, NewF.AM.BaseOffs).abs())) + continue; + + // OK, looks good. + (void)InsertFormula(LU, LUIdx, NewF); + } else { + // Use the immediate in a base register. + for (size_t N = 0, NE = F.BaseRegs.size(); N != NE; ++N) { + const SCEV *BaseReg = F.BaseRegs[N]; + if (BaseReg != OrigReg) + continue; + Formula NewF = F; + NewF.AM.BaseOffs = (uint64_t)NewF.AM.BaseOffs + Imm; + if (!isLegalUse(NewF.AM, LU.MinOffset, LU.MaxOffset, + LU.Kind, LU.AccessTy, TLI)) + continue; + NewF.BaseRegs[N] = SE.getAddExpr(NegImmS, BaseReg); + + // If the new formula has a constant in a register, and adding the + // constant value to the immediate would produce a value closer to + // zero than the immediate itself, then the formula isn't worthwhile. + for (SmallVectorImpl::const_iterator + J = NewF.BaseRegs.begin(), JE = NewF.BaseRegs.end(); + J != JE; ++J) + if (const SCEVConstant *C = dyn_cast(*J)) + if (C->getValue()->getValue().isNegative() != + (NewF.AM.BaseOffs < 0) && + C->getValue()->getValue().abs() + .ule(APInt(BitWidth, NewF.AM.BaseOffs).abs())) + goto skip_formula; + + // Ok, looks good. + (void)InsertFormula(LU, LUIdx, NewF); + break; + skip_formula:; + } + } + } + } +} + +/// GenerateAllReuseFormulae - Generate formulae for each use. +void +LSRInstance::GenerateAllReuseFormulae() { + // This is split into multiple loops so that hasRegsUsedByUsesOtherThan + // queries are more precise. + for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { + LSRUse &LU = Uses[LUIdx]; + for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) + GenerateReassociations(LU, LUIdx, LU.Formulae[i]); + for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) + GenerateCombinations(LU, LUIdx, LU.Formulae[i]); + } + for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { + LSRUse &LU = Uses[LUIdx]; + for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) + GenerateSymbolicOffsets(LU, LUIdx, LU.Formulae[i]); + for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) + GenerateConstantOffsets(LU, LUIdx, LU.Formulae[i]); + for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) + GenerateICmpZeroScales(LU, LUIdx, LU.Formulae[i]); + for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) + GenerateScales(LU, LUIdx, LU.Formulae[i]); + } + for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { + LSRUse &LU = Uses[LUIdx]; + for (size_t i = 0, f = LU.Formulae.size(); i != f; ++i) + GenerateTruncates(LU, LUIdx, LU.Formulae[i]); + } + + GenerateCrossUseConstantOffsets(); +} + +/// If their are multiple formulae with the same set of registers used +/// by other uses, pick the best one and delete the others. +void LSRInstance::FilterOutUndesirableDedicatedRegisters() { +#ifndef NDEBUG + bool Changed = false; +#endif + + // Collect the best formula for each unique set of shared registers. This + // is reset for each use. + typedef DenseMap, size_t, UniquifierDenseMapInfo> + BestFormulaeTy; + BestFormulaeTy BestFormulae; + + for (size_t LUIdx = 0, NumUses = Uses.size(); LUIdx != NumUses; ++LUIdx) { + LSRUse &LU = Uses[LUIdx]; + FormulaSorter Sorter(L, LU, SE, DT); + + // Clear out the set of used regs; it will be recomputed. + LU.Regs.clear(); + + for (size_t FIdx = 0, NumForms = LU.Formulae.size(); + FIdx != NumForms; ++FIdx) { + Formula &F = LU.Formulae[FIdx]; + + SmallVector Key; + for (SmallVectorImpl::const_iterator J = F.BaseRegs.begin(), + JE = F.BaseRegs.end(); J != JE; ++J) { + const SCEV *Reg = *J; + if (RegUses.isRegUsedByUsesOtherThan(Reg, LUIdx)) + Key.push_back(Reg); + } + if (F.ScaledReg && + RegUses.isRegUsedByUsesOtherThan(F.ScaledReg, LUIdx)) + Key.push_back(F.ScaledReg); + // Unstable sort by host order ok, because this is only used for + // uniquifying. + std::sort(Key.begin(), Key.end()); + + std::pair P = + BestFormulae.insert(std::make_pair(Key, FIdx)); + if (!P.second) { + Formula &Best = LU.Formulae[P.first->second]; + if (Sorter.operator()(F, Best)) + std::swap(F, Best); + DEBUG(dbgs() << "Filtering out "; F.print(dbgs()); + dbgs() << "\n" + " in favor of "; Best.print(dbgs()); + dbgs() << '\n'); +#ifndef NDEBUG + Changed = true; +#endif + std::swap(F, LU.Formulae.back()); + LU.Formulae.pop_back(); + --FIdx; + --NumForms; + continue; + } + if (F.ScaledReg) LU.Regs.insert(F.ScaledReg); + LU.Regs.insert(F.BaseRegs.begin(), F.BaseRegs.end()); + } + BestFormulae.clear(); + } + + DEBUG(if (Changed) { + dbgs() << "\n" + "After filtering out undesirable candidates:\n"; + print_uses(dbgs()); + }); +} + +/// NarrowSearchSpaceUsingHeuristics - If there are an extraordinary number of +/// formulae to choose from, use some rough heuristics to prune down the number +/// of formulae. This keeps the main solver from taking an extraordinary amount +/// of time in some worst-case scenarios. +void LSRInstance::NarrowSearchSpaceUsingHeuristics() { + // This is a rough guess that seems to work fairly well. + const size_t Limit = UINT16_MAX; + + SmallPtrSet Taken; + for (;;) { + // Estimate the worst-case number of solutions we might consider. We almost + // never consider this many solutions because we prune the search space, + // but the pruning isn't always sufficient. + uint32_t Power = 1; + for (SmallVectorImpl::const_iterator I = Uses.begin(), + E = Uses.end(); I != E; ++I) { + size_t FSize = I->Formulae.size(); + if (FSize >= Limit) { + Power = Limit; + break; + } + Power *= FSize; + if (Power >= Limit) + break; + } + if (Power < Limit) + break; + + // Ok, we have too many of formulae on our hands to conveniently handle. + // Use a rough heuristic to thin out the list. + + // Pick the register which is used by the most LSRUses, which is likely + // to be a good reuse register candidate. + const SCEV *Best = 0; + unsigned BestNum = 0; + for (RegUseTracker::const_iterator I = RegUses.begin(), E = RegUses.end(); + I != E; ++I) { + const SCEV *Reg = *I; + if (Taken.count(Reg)) + continue; + if (!Best) + Best = Reg; + else { + unsigned Count = RegUses.getUsedByIndices(Reg).count(); + if (Count > BestNum) { + Best = Reg; + BestNum = Count; + } + } + } + + DEBUG(dbgs() << "Narrowing the search space by assuming " << *Best + << " will yield profitable reuse.\n"); + Taken.insert(Best); + + // In any use with formulae which references this register, delete formulae + // which don't reference it. + for (SmallVectorImpl::iterator I = Uses.begin(), + E = Uses.end(); I != E; ++I) { + LSRUse &LU = *I; + if (!LU.Regs.count(Best)) continue; + + // Clear out the set of used regs; it will be recomputed. + LU.Regs.clear(); + + for (size_t i = 0, e = LU.Formulae.size(); i != e; ++i) { + Formula &F = LU.Formulae[i]; + if (!F.referencesReg(Best)) { + DEBUG(dbgs() << " Deleting "; F.print(dbgs()); dbgs() << '\n'); + std::swap(LU.Formulae.back(), F); + LU.Formulae.pop_back(); + --e; + --i; + continue; + } + + if (F.ScaledReg) LU.Regs.insert(F.ScaledReg); + LU.Regs.insert(F.BaseRegs.begin(), F.BaseRegs.end()); + } + } + + DEBUG(dbgs() << "After pre-selection:\n"; + print_uses(dbgs())); + } +} + +/// SolveRecurse - This is the recursive solver. +void LSRInstance::SolveRecurse(SmallVectorImpl &Solution, + Cost &SolutionCost, + SmallVectorImpl &Workspace, + const Cost &CurCost, + const SmallPtrSet &CurRegs, + DenseSet &VisitedRegs) const { + // Some ideas: + // - prune more: + // - use more aggressive filtering + // - sort the formula so that the most profitable solutions are found first + // - sort the uses too + // - search faster: + // - don't compute a cost, and then compare. compare while computing a cost + // and bail early. + // - track register sets with SmallBitVector + + const LSRUse &LU = Uses[Workspace.size()]; + + // If this use references any register that's already a part of the + // in-progress solution, consider it a requirement that a formula must + // reference that register in order to be considered. This prunes out + // unprofitable searching. + SmallSetVector ReqRegs; + for (SmallPtrSet::const_iterator I = CurRegs.begin(), + E = CurRegs.end(); I != E; ++I) + if (LU.Regs.count(*I)) + ReqRegs.insert(*I); + + bool AnySatisfiedReqRegs = false; + SmallPtrSet NewRegs; + Cost NewCost; +retry: + for (SmallVectorImpl::const_iterator I = LU.Formulae.begin(), + E = LU.Formulae.end(); I != E; ++I) { + const Formula &F = *I; + + // Ignore formulae which do not use any of the required registers. + for (SmallSetVector::const_iterator J = ReqRegs.begin(), + JE = ReqRegs.end(); J != JE; ++J) { + const SCEV *Reg = *J; + if ((!F.ScaledReg || F.ScaledReg != Reg) && + std::find(F.BaseRegs.begin(), F.BaseRegs.end(), Reg) == + F.BaseRegs.end()) + goto skip; + } + AnySatisfiedReqRegs = true; + + // Evaluate the cost of the current formula. If it's already worse than + // the current best, prune the search at that point. + NewCost = CurCost; + NewRegs = CurRegs; + NewCost.RateFormula(F, NewRegs, VisitedRegs, L, LU.Offsets, SE, DT); + if (NewCost < SolutionCost) { + Workspace.push_back(&F); + if (Workspace.size() != Uses.size()) { + SolveRecurse(Solution, SolutionCost, Workspace, NewCost, + NewRegs, VisitedRegs); + if (F.getNumRegs() == 1 && Workspace.size() == 1) + VisitedRegs.insert(F.ScaledReg ? F.ScaledReg : F.BaseRegs[0]); + } else { + DEBUG(dbgs() << "New best at "; NewCost.print(dbgs()); + dbgs() << ". Regs:"; + for (SmallPtrSet::const_iterator + I = NewRegs.begin(), E = NewRegs.end(); I != E; ++I) + dbgs() << ' ' << **I; + dbgs() << '\n'); + + SolutionCost = NewCost; + Solution = Workspace; + } + Workspace.pop_back(); + } + skip:; + } + + // If none of the formulae had all of the required registers, relax the + // constraint so that we don't exclude all formulae. + if (!AnySatisfiedReqRegs) { + ReqRegs.clear(); + goto retry; + } +} + +void LSRInstance::Solve(SmallVectorImpl &Solution) const { + SmallVector Workspace; + Cost SolutionCost; + SolutionCost.Loose(); + Cost CurCost; + SmallPtrSet CurRegs; + DenseSet VisitedRegs; + Workspace.reserve(Uses.size()); + + SolveRecurse(Solution, SolutionCost, Workspace, CurCost, + CurRegs, VisitedRegs); + + // Ok, we've now made all our decisions. + DEBUG(dbgs() << "\n" + "The chosen solution requires "; SolutionCost.print(dbgs()); + dbgs() << ":\n"; + for (size_t i = 0, e = Uses.size(); i != e; ++i) { + dbgs() << " "; + Uses[i].print(dbgs()); + dbgs() << "\n" + " "; + Solution[i]->print(dbgs()); + dbgs() << '\n'; + }); +} + +/// getImmediateDominator - A handy utility for the specific DominatorTree +/// query that we need here. +/// +static BasicBlock *getImmediateDominator(BasicBlock *BB, DominatorTree &DT) { + DomTreeNode *Node = DT.getNode(BB); + if (!Node) return 0; + Node = Node->getIDom(); + if (!Node) return 0; + return Node->getBlock(); +} + +Value *LSRInstance::Expand(const LSRFixup &LF, + const Formula &F, + BasicBlock::iterator IP, + SCEVExpander &Rewriter, + SmallVectorImpl &DeadInsts) const { + const LSRUse &LU = Uses[LF.LUIdx]; + + // Then, collect some instructions which we will remain dominated by when + // expanding the replacement. These must be dominated by any operands that + // will be required in the expansion. + SmallVector Inputs; + if (Instruction *I = dyn_cast(LF.OperandValToReplace)) + Inputs.push_back(I); + if (LU.Kind == LSRUse::ICmpZero) + if (Instruction *I = + dyn_cast(cast(LF.UserInst)->getOperand(1))) + Inputs.push_back(I); + if (LF.PostIncLoop) { + if (!L->contains(LF.UserInst)) + Inputs.push_back(L->getLoopLatch()->getTerminator()); + else + Inputs.push_back(IVIncInsertPos); + } + + // Then, climb up the immediate dominator tree as far as we can go while + // still being dominated by the input positions. + for (;;) { + bool AllDominate = true; + Instruction *BetterPos = 0; + BasicBlock *IDom = getImmediateDominator(IP->getParent(), DT); + if (!IDom) break; + Instruction *Tentative = IDom->getTerminator(); + for (SmallVectorImpl::const_iterator I = Inputs.begin(), + E = Inputs.end(); I != E; ++I) { + Instruction *Inst = *I; + if (Inst == Tentative || !DT.dominates(Inst, Tentative)) { + AllDominate = false; + break; + } + if (IDom == Inst->getParent() && + (!BetterPos || DT.dominates(BetterPos, Inst))) + BetterPos = next(BasicBlock::iterator(Inst)); + } + if (!AllDominate) + break; + if (BetterPos) + IP = BetterPos; + else + IP = Tentative; + } + while (isa(IP)) ++IP; + + // Inform the Rewriter if we have a post-increment use, so that it can + // perform an advantageous expansion. + Rewriter.setPostInc(LF.PostIncLoop); + + // This is the type that the user actually needs. + const Type *OpTy = LF.OperandValToReplace->getType(); + // This will be the type that we'll initially expand to. + const Type *Ty = F.getType(); + if (!Ty) + // No type known; just expand directly to the ultimate type. + Ty = OpTy; + else if (SE.getEffectiveSCEVType(Ty) == SE.getEffectiveSCEVType(OpTy)) + // Expand directly to the ultimate type if it's the right size. + Ty = OpTy; + // This is the type to do integer arithmetic in. + const Type *IntTy = SE.getEffectiveSCEVType(Ty); + + // Build up a list of operands to add together to form the full base. + SmallVector Ops; + + // Expand the BaseRegs portion. + for (SmallVectorImpl::const_iterator I = F.BaseRegs.begin(), + E = F.BaseRegs.end(); I != E; ++I) { + const SCEV *Reg = *I; + assert(!Reg->isZero() && "Zero allocated in a base register!"); + + // If we're expanding for a post-inc user for the add-rec's loop, make the + // post-inc adjustment. + const SCEV *Start = Reg; + while (const SCEVAddRecExpr *AR = dyn_cast(Start)) { + if (AR->getLoop() == LF.PostIncLoop) { + Reg = SE.getAddExpr(Reg, AR->getStepRecurrence(SE)); + // If the user is inside the loop, insert the code after the increment + // so that it is dominated by its operand. If the original insert point + // was already dominated by the increment, keep it, because there may + // be loop-variant operands that need to be respected also. + if (L->contains(LF.UserInst) && !DT.dominates(IVIncInsertPos, IP)) + IP = IVIncInsertPos; + break; + } + Start = AR->getStart(); + } + + Ops.push_back(SE.getUnknown(Rewriter.expandCodeFor(Reg, 0, IP))); + } + + // Expand the ScaledReg portion. + Value *ICmpScaledV = 0; + if (F.AM.Scale != 0) { + const SCEV *ScaledS = F.ScaledReg; + + // If we're expanding for a post-inc user for the add-rec's loop, make the + // post-inc adjustment. + if (const SCEVAddRecExpr *AR = dyn_cast(ScaledS)) + if (AR->getLoop() == LF.PostIncLoop) + ScaledS = SE.getAddExpr(ScaledS, AR->getStepRecurrence(SE)); + + if (LU.Kind == LSRUse::ICmpZero) { + // An interesting way of "folding" with an icmp is to use a negated + // scale, which we'll implement by inserting it into the other operand + // of the icmp. + assert(F.AM.Scale == -1 && + "The only scale supported by ICmpZero uses is -1!"); + ICmpScaledV = Rewriter.expandCodeFor(ScaledS, 0, IP); + } else { + // Otherwise just expand the scaled register and an explicit scale, + // which is expected to be matched as part of the address. + ScaledS = SE.getUnknown(Rewriter.expandCodeFor(ScaledS, 0, IP)); + ScaledS = SE.getMulExpr(ScaledS, + SE.getIntegerSCEV(F.AM.Scale, + ScaledS->getType())); + Ops.push_back(ScaledS); + } + } - // Sort the StrideOrder so we process larger strides first. - std::stable_sort(IU->StrideOrder.begin(), IU->StrideOrder.end(), - StrideCompare(SE)); + // Expand the immediate portions. + if (F.AM.BaseGV) + Ops.push_back(SE.getSCEV(F.AM.BaseGV)); + int64_t Offset = (uint64_t)F.AM.BaseOffs + LF.Offset; + if (Offset != 0) { + if (LU.Kind == LSRUse::ICmpZero) { + // The other interesting way of "folding" with an ICmpZero is to use a + // negated immediate. + if (!ICmpScaledV) + ICmpScaledV = ConstantInt::get(IntTy, -Offset); + else { + Ops.push_back(SE.getUnknown(ICmpScaledV)); + ICmpScaledV = ConstantInt::get(IntTy, Offset); + } + } else { + // Just add the immediate values. These again are expected to be matched + // as part of the address. + Ops.push_back(SE.getIntegerSCEV(Offset, IntTy)); + } + } - // Optimize induction variables. Some indvar uses can be transformed to use - // strides that will be needed for other purposes. A common example of this - // is the exit test for the loop, which can often be rewritten to use the - // computation of some other indvar to decide when to terminate the loop. - OptimizeIndvars(L); + // Emit instructions summing all the operands. + const SCEV *FullS = Ops.empty() ? + SE.getIntegerSCEV(0, IntTy) : + SE.getAddExpr(Ops); + Value *FullV = Rewriter.expandCodeFor(FullS, Ty, IP); + + // We're done expanding now, so reset the rewriter. + Rewriter.setPostInc(0); + + // An ICmpZero Formula represents an ICmp which we're handling as a + // comparison against zero. Now that we've expanded an expression for that + // form, update the ICmp's other operand. + if (LU.Kind == LSRUse::ICmpZero) { + ICmpInst *CI = cast(LF.UserInst); + DeadInsts.push_back(CI->getOperand(1)); + assert(!F.AM.BaseGV && "ICmp does not support folding a global value and " + "a scale at the same time!"); + if (F.AM.Scale == -1) { + if (ICmpScaledV->getType() != OpTy) { + Instruction *Cast = + CastInst::Create(CastInst::getCastOpcode(ICmpScaledV, false, + OpTy, false), + ICmpScaledV, OpTy, "tmp", CI); + ICmpScaledV = Cast; + } + CI->setOperand(1, ICmpScaledV); + } else { + assert(F.AM.Scale == 0 && + "ICmp does not support folding a global value and " + "a scale at the same time!"); + Constant *C = ConstantInt::getSigned(SE.getEffectiveSCEVType(OpTy), + -(uint64_t)Offset); + if (C->getType() != OpTy) + C = ConstantExpr::getCast(CastInst::getCastOpcode(C, false, + OpTy, false), + C, OpTy); + + CI->setOperand(1, C); + } + } + + return FullV; +} + +/// RewriteForPHI - Helper for Rewrite. PHI nodes are special because the use +/// of their operands effectively happens in their predecessor blocks, so the +/// expression may need to be expanded in multiple places. +void LSRInstance::RewriteForPHI(PHINode *PN, + const LSRFixup &LF, + const Formula &F, + SCEVExpander &Rewriter, + SmallVectorImpl &DeadInsts, + Pass *P) const { + DenseMap Inserted; + for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) + if (PN->getIncomingValue(i) == LF.OperandValToReplace) { + BasicBlock *BB = PN->getIncomingBlock(i); + + // If this is a critical edge, split the edge so that we do not insert + // the code on all predecessor/successor paths. We do this unless this + // is the canonical backedge for this loop, which complicates post-inc + // users. + if (e != 1 && BB->getTerminator()->getNumSuccessors() > 1 && + !isa(BB->getTerminator()) && + (PN->getParent() != L->getHeader() || !L->contains(BB))) { + // Split the critical edge. + BasicBlock *NewBB = SplitCriticalEdge(BB, PN->getParent(), P); + + // If PN is outside of the loop and BB is in the loop, we want to + // move the block to be immediately before the PHI block, not + // immediately after BB. + if (L->contains(BB) && !L->contains(PN)) + NewBB->moveBefore(PN->getParent()); + + // Splitting the edge can reduce the number of PHI entries we have. + e = PN->getNumIncomingValues(); + BB = NewBB; + i = PN->getBasicBlockIndex(BB); + } + + std::pair::iterator, bool> Pair = + Inserted.insert(std::make_pair(BB, static_cast(0))); + if (!Pair.second) + PN->setIncomingValue(i, Pair.first->second); + else { + Value *FullV = Expand(LF, F, BB->getTerminator(), Rewriter, DeadInsts); + + // If this is reuse-by-noop-cast, insert the noop cast. + const Type *OpTy = LF.OperandValToReplace->getType(); + if (FullV->getType() != OpTy) + FullV = + CastInst::Create(CastInst::getCastOpcode(FullV, false, + OpTy, false), + FullV, LF.OperandValToReplace->getType(), + "tmp", BB->getTerminator()); + + PN->setIncomingValue(i, FullV); + Pair.first->second = FullV; + } + } +} - // Change loop terminating condition to use the postinc iv when possible - // and optimize loop terminating compare. FIXME: Move this after - // StrengthReduceIVUsersOfStride? - OptimizeLoopTermCond(L); +/// Rewrite - Emit instructions for the leading candidate expression for this +/// LSRUse (this is called "expanding"), and update the UserInst to reference +/// the newly expanded value. +void LSRInstance::Rewrite(const LSRFixup &LF, + const Formula &F, + SCEVExpander &Rewriter, + SmallVectorImpl &DeadInsts, + Pass *P) const { + // First, find an insertion point that dominates UserInst. For PHI nodes, + // find the nearest block which dominates all the relevant uses. + if (PHINode *PN = dyn_cast(LF.UserInst)) { + RewriteForPHI(PN, LF, F, Rewriter, DeadInsts, P); + } else { + Value *FullV = Expand(LF, F, LF.UserInst, Rewriter, DeadInsts); + + // If this is reuse-by-noop-cast, insert the noop cast. + const Type *OpTy = LF.OperandValToReplace->getType(); + if (FullV->getType() != OpTy) { + Instruction *Cast = + CastInst::Create(CastInst::getCastOpcode(FullV, false, OpTy, false), + FullV, OpTy, "tmp", LF.UserInst); + FullV = Cast; + } - // FIXME: We can shrink overlarge IV's here. e.g. if the code has - // computation in i64 values and the target doesn't support i64, demote - // the computation to 32-bit if safe. + // Update the user. ICmpZero is handled specially here (for now) because + // Expand may have updated one of the operands of the icmp already, and + // its new value may happen to be equal to LF.OperandValToReplace, in + // which case doing replaceUsesOfWith leads to replacing both operands + // with the same value. TODO: Reorganize this. + if (Uses[LF.LUIdx].Kind == LSRUse::ICmpZero) + LF.UserInst->setOperand(0, FullV); + else + LF.UserInst->replaceUsesOfWith(LF.OperandValToReplace, FullV); + } - // FIXME: Attempt to reuse values across multiple IV's. In particular, we - // could have something like "for(i) { foo(i*8); bar(i*16) }", which should - // be codegened as "for (j = 0;; j+=8) { foo(j); bar(j+j); }" on X86/PPC. - // Need to be careful that IV's are all the same type. Only works for - // intptr_t indvars. + DeadInsts.push_back(LF.OperandValToReplace); +} - // IVsByStride keeps IVs for one particular loop. - assert(IVsByStride.empty() && "Stale entries in IVsByStride?"); +void +LSRInstance::ImplementSolution(const SmallVectorImpl &Solution, + Pass *P) { + // Keep track of instructions we may have made dead, so that + // we can remove them after we are done working. + SmallVector DeadInsts; - StrengthReduceIVUsers(L); + SCEVExpander Rewriter(SE); + Rewriter.disableCanonicalMode(); + Rewriter.setIVIncInsertPos(L, IVIncInsertPos); - // After all sharing is done, see if we can adjust the loop to test against - // zero instead of counting up to a maximum. This is usually faster. - OptimizeLoopCountIV(L); + // Expand the new value definitions and update the users. + for (size_t i = 0, e = Fixups.size(); i != e; ++i) { + size_t LUIdx = Fixups[i].LUIdx; - // We're done analyzing this loop; release all the state we built up for it. - IVsByStride.clear(); + Rewrite(Fixups[i], *Solution[LUIdx], Rewriter, DeadInsts, P); - // Clean up after ourselves - DeleteTriviallyDeadInstructions(); + Changed = true; } + // Clean up after ourselves. This must be done before deleting any + // instructions. + Rewriter.clear(); + + Changed |= DeleteTriviallyDeadInstructions(DeadInsts); +} + +LSRInstance::LSRInstance(const TargetLowering *tli, Loop *l, Pass *P) + : IU(P->getAnalysis()), + SE(P->getAnalysis()), + DT(P->getAnalysis()), + TLI(tli), L(l), Changed(false), IVIncInsertPos(0) { + + // If LoopSimplify form is not available, stay out of trouble. + if (!L->isLoopSimplifyForm()) return; + + // If there's no interesting work to be done, bail early. + if (IU.empty()) return; + + DEBUG(dbgs() << "\nLSR on loop "; + WriteAsOperand(dbgs(), L->getHeader(), /*PrintType=*/false); + dbgs() << ":\n"); + + /// OptimizeShadowIV - If IV is used in a int-to-float cast + /// inside the loop then try to eliminate the cast operation. + OptimizeShadowIV(); + + // Change loop terminating condition to use the postinc iv when possible. + Changed |= OptimizeLoopTermCond(); + + CollectInterestingTypesAndFactors(); + CollectFixupsAndInitialFormulae(); + CollectLoopInvariantFixupsAndFormulae(); + + DEBUG(dbgs() << "LSR found " << Uses.size() << " uses:\n"; + print_uses(dbgs())); + + // Now use the reuse data to generate a bunch of interesting ways + // to formulate the values needed for the uses. + GenerateAllReuseFormulae(); + + DEBUG(dbgs() << "\n" + "After generating reuse formulae:\n"; + print_uses(dbgs())); + + FilterOutUndesirableDedicatedRegisters(); + NarrowSearchSpaceUsingHeuristics(); + + SmallVector Solution; + Solve(Solution); + assert(Solution.size() == Uses.size() && "Malformed solution!"); + + // Release memory that is no longer needed. + Factors.clear(); + Types.clear(); + RegUses.clear(); + +#ifndef NDEBUG + // Formulae should be legal. + for (SmallVectorImpl::const_iterator I = Uses.begin(), + E = Uses.end(); I != E; ++I) { + const LSRUse &LU = *I; + for (SmallVectorImpl::const_iterator J = LU.Formulae.begin(), + JE = LU.Formulae.end(); J != JE; ++J) + assert(isLegalUse(J->AM, LU.MinOffset, LU.MaxOffset, + LU.Kind, LU.AccessTy, TLI) && + "Illegal formula generated!"); + }; +#endif + + // Now that we've decided what we want, make it so. + ImplementSolution(Solution, P); +} + +void LSRInstance::print_factors_and_types(raw_ostream &OS) const { + if (Factors.empty() && Types.empty()) return; + + OS << "LSR has identified the following interesting factors and types: "; + bool First = true; + + for (SmallSetVector::const_iterator + I = Factors.begin(), E = Factors.end(); I != E; ++I) { + if (!First) OS << ", "; + First = false; + OS << '*' << *I; + } + + for (SmallSetVector::const_iterator + I = Types.begin(), E = Types.end(); I != E; ++I) { + if (!First) OS << ", "; + First = false; + OS << '(' << **I << ')'; + } + OS << '\n'; +} + +void LSRInstance::print_fixups(raw_ostream &OS) const { + OS << "LSR is examining the following fixup sites:\n"; + for (SmallVectorImpl::const_iterator I = Fixups.begin(), + E = Fixups.end(); I != E; ++I) { + const LSRFixup &LF = *I; + dbgs() << " "; + LF.print(OS); + OS << '\n'; + } +} + +void LSRInstance::print_uses(raw_ostream &OS) const { + OS << "LSR is examining the following uses:\n"; + for (SmallVectorImpl::const_iterator I = Uses.begin(), + E = Uses.end(); I != E; ++I) { + const LSRUse &LU = *I; + dbgs() << " "; + LU.print(OS); + OS << '\n'; + for (SmallVectorImpl::const_iterator J = LU.Formulae.begin(), + JE = LU.Formulae.end(); J != JE; ++J) { + OS << " "; + J->print(OS); + OS << '\n'; + } + } +} + +void LSRInstance::print(raw_ostream &OS) const { + print_factors_and_types(OS); + print_fixups(OS); + print_uses(OS); +} + +void LSRInstance::dump() const { + print(errs()); errs() << '\n'; +} + +namespace { + +class LoopStrengthReduce : public LoopPass { + /// TLI - Keep a pointer of a TargetLowering to consult for determining + /// transformation profitability. + const TargetLowering *const TLI; + +public: + static char ID; // Pass ID, replacement for typeid + explicit LoopStrengthReduce(const TargetLowering *tli = 0); + +private: + bool runOnLoop(Loop *L, LPPassManager &LPM); + void getAnalysisUsage(AnalysisUsage &AU) const; +}; + +} + +char LoopStrengthReduce::ID = 0; +static RegisterPass +X("loop-reduce", "Loop Strength Reduction"); + +Pass *llvm::createLoopStrengthReducePass(const TargetLowering *TLI) { + return new LoopStrengthReduce(TLI); +} + +LoopStrengthReduce::LoopStrengthReduce(const TargetLowering *tli) + : LoopPass(&ID), TLI(tli) {} + +void LoopStrengthReduce::getAnalysisUsage(AnalysisUsage &AU) const { + // We split critical edges, so we change the CFG. However, we do update + // many analyses if they are around. + AU.addPreservedID(LoopSimplifyID); + AU.addPreserved(); + AU.addPreserved("domfrontier"); + + AU.addRequiredID(LoopSimplifyID); + AU.addRequired(); + AU.addPreserved(); + AU.addRequired(); + AU.addPreserved(); + AU.addRequired(); + AU.addPreserved(); +} + +bool LoopStrengthReduce::runOnLoop(Loop *L, LPPassManager & /*LPM*/) { + bool Changed = false; + + // Run the main LSR transformation. + Changed |= LSRInstance(TLI, L, this).getChanged(); + // At this point, it is worth checking to see if any recurrence PHIs are also // dead, so that we can remove them as well. Changed |= DeleteDeadPHIs(L->getHeader());