X-Git-Url: http://plrg.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FAnalysis%2FLoads.cpp;h=4b2fa3c6505a14a1e0b4a5a1eacc9abe5aa7cab6;hb=dffe9f4fa6bac1ea04bb1d88f1fab87ffac77322;hp=73aa8b49cda584dd9ef4c36618cd959239c29407;hpb=3574eca1b02600bac4e625297f4ecf745f4c4f32;p=oota-llvm.git diff --git a/lib/Analysis/Loads.cpp b/lib/Analysis/Loads.cpp index 73aa8b49cda..4b2fa3c6505 100644 --- a/lib/Analysis/Loads.cpp +++ b/lib/Analysis/Loads.cpp @@ -13,33 +13,39 @@ #include "llvm/Analysis/Loads.h" #include "llvm/Analysis/AliasAnalysis.h" -#include "llvm/DataLayout.h" -#include "llvm/GlobalAlias.h" -#include "llvm/GlobalVariable.h" -#include "llvm/IntrinsicInst.h" -#include "llvm/LLVMContext.h" -#include "llvm/Operator.h" +#include "llvm/Analysis/ValueTracking.h" +#include "llvm/IR/DataLayout.h" +#include "llvm/IR/GlobalAlias.h" +#include "llvm/IR/GlobalVariable.h" +#include "llvm/IR/IntrinsicInst.h" +#include "llvm/IR/LLVMContext.h" +#include "llvm/IR/Module.h" +#include "llvm/IR/Operator.h" using namespace llvm; -/// AreEquivalentAddressValues - Test if A and B will obviously have the same -/// value. This includes recognizing that %t0 and %t1 will have the same +/// \brief Test if A and B will obviously have the same value. +/// +/// This includes recognizing that %t0 and %t1 will have the same /// value in code like this: +/// \code /// %t0 = getelementptr \@a, 0, 3 /// store i32 0, i32* %t0 /// %t1 = getelementptr \@a, 0, 3 /// %t2 = load i32* %t1 +/// \endcode /// static bool AreEquivalentAddressValues(const Value *A, const Value *B) { // Test if the values are trivially equivalent. - if (A == B) return true; + if (A == B) + return true; // Test if the values come from identical arithmetic instructions. // Use isIdenticalToWhenDefined instead of isIdenticalTo because // this function is only used when one address use dominates the // other, which means that they'll always either have the same // value or one of them will have an undefined value. - if (isa(A) || isa(A) || - isa(A) || isa(A)) + if (isa(A) || isa(A) || isa(A) || + isa(A)) if (const Instruction *BI = dyn_cast(B)) if (cast(A)->isIdenticalToWhenDefined(BI)) return true; @@ -48,77 +54,61 @@ static bool AreEquivalentAddressValues(const Value *A, const Value *B) { return false; } -/// getUnderlyingObjectWithOffset - Strip off up to MaxLookup GEPs and -/// bitcasts to get back to the underlying object being addressed, keeping -/// track of the offset in bytes from the GEPs relative to the result. -/// This is closely related to GetUnderlyingObject but is located -/// here to avoid making VMCore depend on DataLayout. -static Value *getUnderlyingObjectWithOffset(Value *V, const DataLayout *TD, - uint64_t &ByteOffset, - unsigned MaxLookup = 6) { - if (!V->getType()->isPointerTy()) - return V; - for (unsigned Count = 0; MaxLookup == 0 || Count < MaxLookup; ++Count) { - if (GEPOperator *GEP = dyn_cast(V)) { - if (!GEP->hasAllConstantIndices()) - return V; - SmallVector Indices(GEP->op_begin() + 1, GEP->op_end()); - ByteOffset += TD->getIndexedOffset(GEP->getPointerOperandType(), - Indices); - V = GEP->getPointerOperand(); - } else if (Operator::getOpcode(V) == Instruction::BitCast) { - V = cast(V)->getOperand(0); - } else if (GlobalAlias *GA = dyn_cast(V)) { - if (GA->mayBeOverridden()) - return V; - V = GA->getAliasee(); - } else { - return V; - } - assert(V->getType()->isPointerTy() && "Unexpected operand type!"); - } - return V; -} - -/// isSafeToLoadUnconditionally - Return true if we know that executing a load -/// from this value cannot trap. If it is not obviously safe to load from the -/// specified pointer, we do a quick local scan of the basic block containing -/// ScanFrom, to determine if the address is already accessed. +/// \brief Check if executing a load of this pointer value cannot trap. +/// +/// If it is not obviously safe to load from the specified pointer, we do +/// a quick local scan of the basic block containing \c ScanFrom, to determine +/// if the address is already accessed. +/// +/// This uses the pointee type to determine how many bytes need to be safe to +/// load from the pointer. bool llvm::isSafeToLoadUnconditionally(Value *V, Instruction *ScanFrom, - unsigned Align, const DataLayout *TD) { - uint64_t ByteOffset = 0; + unsigned Align) { + const DataLayout &DL = ScanFrom->getModule()->getDataLayout(); + + // Zero alignment means that the load has the ABI alignment for the target + if (Align == 0) + Align = DL.getABITypeAlignment(V->getType()->getPointerElementType()); + assert(isPowerOf2_32(Align)); + + int64_t ByteOffset = 0; Value *Base = V; - if (TD) - Base = getUnderlyingObjectWithOffset(V, TD, ByteOffset); + Base = GetPointerBaseWithConstantOffset(V, ByteOffset, DL); + + if (ByteOffset < 0) // out of bounds + return false; - Type *BaseType = 0; + Type *BaseType = nullptr; unsigned BaseAlign = 0; if (const AllocaInst *AI = dyn_cast(Base)) { // An alloca is safe to load from as load as it is suitably aligned. BaseType = AI->getAllocatedType(); BaseAlign = AI->getAlignment(); - } else if (const GlobalValue *GV = dyn_cast(Base)) { - // Global variables are safe to load from but their size cannot be - // guaranteed if they are overridden. - if (!isa(GV) && !GV->mayBeOverridden()) { + } else if (const GlobalVariable *GV = dyn_cast(Base)) { + // Global variables are not necessarily safe to load from if they are + // overridden. Their size may change or they may be weak and require a test + // to determine if they were in fact provided. + if (!GV->mayBeOverridden()) { BaseType = GV->getType()->getElementType(); BaseAlign = GV->getAlignment(); } } + PointerType *AddrTy = cast(V->getType()); + uint64_t LoadSize = DL.getTypeStoreSize(AddrTy->getElementType()); + + // If we found a base allocated type from either an alloca or global variable, + // try to see if we are definitively within the allocated region. We need to + // know the size of the base type and the loaded type to do anything in this + // case. if (BaseType && BaseType->isSized()) { - if (TD && BaseAlign == 0) - BaseAlign = TD->getPrefTypeAlignment(BaseType); + if (BaseAlign == 0) + BaseAlign = DL.getPrefTypeAlignment(BaseType); if (Align <= BaseAlign) { - if (!TD) - return true; // Loading directly from an alloca or global is OK. - // Check if the load is within the bounds of the underlying object. - PointerType *AddrTy = cast(V->getType()); - uint64_t LoadSize = TD->getTypeStoreSize(AddrTy->getElementType()); - if (ByteOffset + LoadSize <= TD->getTypeAllocSize(BaseType) && - (Align == 0 || (ByteOffset % Align) == 0)) + if (ByteOffset + LoadSize <= DL.getTypeAllocSize(BaseType) && + ((ByteOffset % Align) == 0)) return true; } } @@ -128,7 +118,12 @@ bool llvm::isSafeToLoadUnconditionally(Value *V, Instruction *ScanFrom, // from/to. If so, the previous load or store would have already trapped, // so there is no harm doing an extra load (also, CSE will later eliminate // the load entirely). - BasicBlock::iterator BBI = ScanFrom, E = ScanFrom->getParent()->begin(); + BasicBlock::iterator BBI = ScanFrom->getIterator(), + E = ScanFrom->getParent()->begin(); + + // We can at least always strip pointer casts even though we can't use the + // base here. + V = V->stripPointerCasts(); while (BBI != E) { --BBI; @@ -139,112 +134,154 @@ bool llvm::isSafeToLoadUnconditionally(Value *V, Instruction *ScanFrom, !isa(BBI)) return false; + Value *AccessedPtr; + unsigned AccessedAlign; if (LoadInst *LI = dyn_cast(BBI)) { - if (AreEquivalentAddressValues(LI->getOperand(0), V)) return true; + AccessedPtr = LI->getPointerOperand(); + AccessedAlign = LI->getAlignment(); } else if (StoreInst *SI = dyn_cast(BBI)) { - if (AreEquivalentAddressValues(SI->getOperand(1), V)) return true; - } + AccessedPtr = SI->getPointerOperand(); + AccessedAlign = SI->getAlignment(); + } else + continue; + + Type *AccessedTy = AccessedPtr->getType()->getPointerElementType(); + if (AccessedAlign == 0) + AccessedAlign = DL.getABITypeAlignment(AccessedTy); + if (AccessedAlign < Align) + continue; + + // Handle trivial cases. + if (AccessedPtr == V) + return true; + + if (AreEquivalentAddressValues(AccessedPtr->stripPointerCasts(), V) && + LoadSize <= DL.getTypeStoreSize(AccessedTy)) + return true; } return false; } -/// FindAvailableLoadedValue - Scan the ScanBB block backwards (starting at the -/// instruction before ScanFrom) checking to see if we have the value at the +/// DefMaxInstsToScan - the default number of maximum instructions +/// to scan in the block, used by FindAvailableLoadedValue(). +/// FindAvailableLoadedValue() was introduced in r60148, to improve jump +/// threading in part by eliminating partially redundant loads. +/// At that point, the value of MaxInstsToScan was already set to '6' +/// without documented explanation. +cl::opt +llvm::DefMaxInstsToScan("available-load-scan-limit", cl::init(6), cl::Hidden, + cl::desc("Use this to specify the default maximum number of instructions " + "to scan backward from a given instruction, when searching for " + "available loaded value")); + +/// \brief Scan the ScanBB block backwards to see if we have the value at the /// memory address *Ptr locally available within a small number of instructions. -/// If the value is available, return it. /// -/// If not, return the iterator for the last validated instruction that the -/// value would be live through. If we scanned the entire block and didn't find -/// something that invalidates *Ptr or provides it, ScanFrom would be left at -/// begin() and this returns null. ScanFrom could also be left +/// The scan starts from \c ScanFrom. \c MaxInstsToScan specifies the maximum +/// instructions to scan in the block. If it is set to \c 0, it will scan the whole +/// block. +/// +/// If the value is available, this function returns it. If not, it returns the +/// iterator for the last validated instruction that the value would be live +/// through. If we scanned the entire block and didn't find something that +/// invalidates \c *Ptr or provides it, \c ScanFrom is left at the last +/// instruction processed and this returns null. /// -/// MaxInstsToScan specifies the maximum instructions to scan in the block. If -/// it is set to 0, it will scan the whole block. You can also optionally -/// specify an alias analysis implementation, which makes this more precise. +/// You can also optionally specify an alias analysis implementation, which +/// makes this more precise. /// -/// If TBAATag is non-null and a load or store is found, the TBAA tag from the -/// load or store is recorded there. If there is no TBAA tag or if no access -/// is found, it is left unmodified. +/// If \c AATags is non-null and a load or store is found, the AA tags from the +/// load or store are recorded there. If there are no AA tags or if no access is +/// found, it is left unmodified. Value *llvm::FindAvailableLoadedValue(Value *Ptr, BasicBlock *ScanBB, BasicBlock::iterator &ScanFrom, unsigned MaxInstsToScan, - AliasAnalysis *AA, - MDNode **TBAATag) { - if (MaxInstsToScan == 0) MaxInstsToScan = ~0U; - - // If we're using alias analysis to disambiguate get the size of *Ptr. - uint64_t AccessSize = 0; - if (AA) { - Type *AccessTy = cast(Ptr->getType())->getElementType(); - AccessSize = AA->getTypeStoreSize(AccessTy); - } - + AliasAnalysis *AA, AAMDNodes *AATags) { + if (MaxInstsToScan == 0) + MaxInstsToScan = ~0U; + + Type *AccessTy = cast(Ptr->getType())->getElementType(); + + const DataLayout &DL = ScanBB->getModule()->getDataLayout(); + + // Try to get the store size for the type. + uint64_t AccessSize = DL.getTypeStoreSize(AccessTy); + + Value *StrippedPtr = Ptr->stripPointerCasts(); + while (ScanFrom != ScanBB->begin()) { // We must ignore debug info directives when counting (otherwise they // would affect codegen). - Instruction *Inst = --ScanFrom; + Instruction *Inst = &*--ScanFrom; if (isa(Inst)) continue; // Restore ScanFrom to expected value in case next test succeeds ScanFrom++; - + // Don't scan huge blocks. - if (MaxInstsToScan-- == 0) return 0; - + if (MaxInstsToScan-- == 0) + return nullptr; + --ScanFrom; // If this is a load of Ptr, the loaded value is available. // (This is true even if the load is volatile or atomic, although // those cases are unlikely.) if (LoadInst *LI = dyn_cast(Inst)) - if (AreEquivalentAddressValues(LI->getOperand(0), Ptr)) { - if (TBAATag) *TBAATag = LI->getMetadata(LLVMContext::MD_tbaa); + if (AreEquivalentAddressValues( + LI->getPointerOperand()->stripPointerCasts(), StrippedPtr) && + CastInst::isBitOrNoopPointerCastable(LI->getType(), AccessTy, DL)) { + if (AATags) + LI->getAAMetadata(*AATags); return LI; } - + if (StoreInst *SI = dyn_cast(Inst)) { + Value *StorePtr = SI->getPointerOperand()->stripPointerCasts(); // If this is a store through Ptr, the value is available! // (This is true even if the store is volatile or atomic, although // those cases are unlikely.) - if (AreEquivalentAddressValues(SI->getOperand(1), Ptr)) { - if (TBAATag) *TBAATag = SI->getMetadata(LLVMContext::MD_tbaa); + if (AreEquivalentAddressValues(StorePtr, StrippedPtr) && + CastInst::isBitOrNoopPointerCastable(SI->getValueOperand()->getType(), + AccessTy, DL)) { + if (AATags) + SI->getAAMetadata(*AATags); return SI->getOperand(0); } - - // If Ptr is an alloca and this is a store to a different alloca, ignore - // the store. This is a trivial form of alias analysis that is important - // for reg2mem'd code. - if ((isa(Ptr) || isa(Ptr)) && - (isa(SI->getOperand(1)) || - isa(SI->getOperand(1)))) + + // If both StrippedPtr and StorePtr reach all the way to an alloca or + // global and they are different, ignore the store. This is a trivial form + // of alias analysis that is important for reg2mem'd code. + if ((isa(StrippedPtr) || isa(StrippedPtr)) && + (isa(StorePtr) || isa(StorePtr)) && + StrippedPtr != StorePtr) continue; - + // If we have alias analysis and it says the store won't modify the loaded // value, ignore the store. - if (AA && - (AA->getModRefInfo(SI, Ptr, AccessSize) & AliasAnalysis::Mod) == 0) + if (AA && (AA->getModRefInfo(SI, StrippedPtr, AccessSize) & MRI_Mod) == 0) continue; - + // Otherwise the store that may or may not alias the pointer, bail out. ++ScanFrom; - return 0; + return nullptr; } - + // If this is some other instruction that may clobber Ptr, bail out. if (Inst->mayWriteToMemory()) { // If alias analysis claims that it really won't modify the load, // ignore it. if (AA && - (AA->getModRefInfo(Inst, Ptr, AccessSize) & AliasAnalysis::Mod) == 0) + (AA->getModRefInfo(Inst, StrippedPtr, AccessSize) & MRI_Mod) == 0) continue; - + // May modify the pointer, bail out. ++ScanFrom; - return 0; + return nullptr; } } - + // Got to the start of the block, we didn't find it, but are done for this // block. - return 0; + return nullptr; }