#define DEBUG_TYPE "early-cse"
#include "llvm/Transforms/Scalar.h"
-#include "llvm/Instructions.h"
-#include "llvm/Pass.h"
+#include "llvm/ADT/Hashing.h"
+#include "llvm/ADT/ScopedHashTable.h"
+#include "llvm/ADT/Statistic.h"
#include "llvm/Analysis/Dominators.h"
#include "llvm/Analysis/InstructionSimplify.h"
-#include "llvm/Target/TargetData.h"
-#include "llvm/Transforms/Utils/Local.h"
+#include "llvm/IR/DataLayout.h"
+#include "llvm/IR/Instructions.h"
+#include "llvm/Pass.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/RecyclingAllocator.h"
-#include "llvm/ADT/ScopedHashTable.h"
-#include "llvm/ADT/Statistic.h"
+#include "llvm/Target/TargetLibraryInfo.h"
+#include "llvm/Transforms/Utils/Local.h"
+#include <deque>
using namespace llvm;
STATISTIC(NumSimplify, "Number of instructions simplified or DCE'd");
STATISTIC(NumCSE, "Number of instructions CSE'd");
STATISTIC(NumCSELoad, "Number of load instructions CSE'd");
STATISTIC(NumCSECall, "Number of call instructions CSE'd");
+STATISTIC(NumDSE, "Number of trivial dead stores removed");
static unsigned getHash(const void *V) {
return DenseMapInfo<const void*>::getHashValue(V);
}
//===----------------------------------------------------------------------===//
-// SimpleValue
+// SimpleValue
//===----------------------------------------------------------------------===//
namespace {
/// scoped hash table.
struct SimpleValue {
Instruction *Inst;
-
+
SimpleValue(Instruction *I) : Inst(I) {
assert((isSentinel() || canHandle(I)) && "Inst can't be handled!");
}
-
+
bool isSentinel() const {
return Inst == DenseMapInfo<Instruction*>::getEmptyKey() ||
Inst == DenseMapInfo<Instruction*>::getTombstoneKey();
}
-
+
static bool canHandle(Instruction *Inst) {
+ // This can only handle non-void readnone functions.
+ if (CallInst *CI = dyn_cast<CallInst>(Inst))
+ return CI->doesNotAccessMemory() && !CI->getType()->isVoidTy();
return isa<CastInst>(Inst) || isa<BinaryOperator>(Inst) ||
isa<GetElementPtrInst>(Inst) || isa<CmpInst>(Inst) ||
isa<SelectInst>(Inst) || isa<ExtractElementInst>(Inst) ||
unsigned DenseMapInfo<SimpleValue>::getHashValue(SimpleValue Val) {
Instruction *Inst = Val.Inst;
-
// Hash in all of the operands as pointers.
- unsigned Res = 0;
- for (unsigned i = 0, e = Inst->getNumOperands(); i != e; ++i)
- Res ^= getHash(Inst->getOperand(i)) << i;
+ if (BinaryOperator* BinOp = dyn_cast<BinaryOperator>(Inst)) {
+ Value *LHS = BinOp->getOperand(0);
+ Value *RHS = BinOp->getOperand(1);
+ if (BinOp->isCommutative() && BinOp->getOperand(0) > BinOp->getOperand(1))
+ std::swap(LHS, RHS);
- if (CastInst *CI = dyn_cast<CastInst>(Inst))
- Res ^= getHash(CI->getType());
- else if (CmpInst *CI = dyn_cast<CmpInst>(Inst))
- Res ^= CI->getPredicate();
- else if (const ExtractValueInst *EVI = dyn_cast<ExtractValueInst>(Inst)) {
- for (ExtractValueInst::idx_iterator I = EVI->idx_begin(),
- E = EVI->idx_end(); I != E; ++I)
- Res ^= *I;
- } else if (const InsertValueInst *IVI = dyn_cast<InsertValueInst>(Inst)) {
- for (InsertValueInst::idx_iterator I = IVI->idx_begin(),
- E = IVI->idx_end(); I != E; ++I)
- Res ^= *I;
- } else {
- // nothing extra to hash in.
- assert((isa<BinaryOperator>(Inst) || isa<GetElementPtrInst>(Inst) ||
- isa<SelectInst>(Inst) || isa<ExtractElementInst>(Inst) ||
- isa<InsertElementInst>(Inst) || isa<ShuffleVectorInst>(Inst)) &&
- "Invalid/unknown instruction");
+ if (isa<OverflowingBinaryOperator>(BinOp)) {
+ // Hash the overflow behavior
+ unsigned Overflow =
+ BinOp->hasNoSignedWrap() * OverflowingBinaryOperator::NoSignedWrap |
+ BinOp->hasNoUnsignedWrap() * OverflowingBinaryOperator::NoUnsignedWrap;
+ return hash_combine(BinOp->getOpcode(), Overflow, LHS, RHS);
+ }
+
+ return hash_combine(BinOp->getOpcode(), LHS, RHS);
+ }
+
+ if (CmpInst *CI = dyn_cast<CmpInst>(Inst)) {
+ Value *LHS = CI->getOperand(0);
+ Value *RHS = CI->getOperand(1);
+ CmpInst::Predicate Pred = CI->getPredicate();
+ if (Inst->getOperand(0) > Inst->getOperand(1)) {
+ std::swap(LHS, RHS);
+ Pred = CI->getSwappedPredicate();
+ }
+ return hash_combine(Inst->getOpcode(), Pred, LHS, RHS);
}
+ if (CastInst *CI = dyn_cast<CastInst>(Inst))
+ return hash_combine(CI->getOpcode(), CI->getType(), CI->getOperand(0));
+
+ if (const ExtractValueInst *EVI = dyn_cast<ExtractValueInst>(Inst))
+ return hash_combine(EVI->getOpcode(), EVI->getOperand(0),
+ hash_combine_range(EVI->idx_begin(), EVI->idx_end()));
+
+ if (const InsertValueInst *IVI = dyn_cast<InsertValueInst>(Inst))
+ return hash_combine(IVI->getOpcode(), IVI->getOperand(0),
+ IVI->getOperand(1),
+ hash_combine_range(IVI->idx_begin(), IVI->idx_end()));
+
+ assert((isa<CallInst>(Inst) || isa<BinaryOperator>(Inst) ||
+ isa<GetElementPtrInst>(Inst) || isa<SelectInst>(Inst) ||
+ isa<ExtractElementInst>(Inst) || isa<InsertElementInst>(Inst) ||
+ isa<ShuffleVectorInst>(Inst)) && "Invalid/unknown instruction");
+
// Mix in the opcode.
- return (Res << 1) ^ Inst->getOpcode();
+ return hash_combine(Inst->getOpcode(),
+ hash_combine_range(Inst->value_op_begin(),
+ Inst->value_op_end()));
}
bool DenseMapInfo<SimpleValue>::isEqual(SimpleValue LHS, SimpleValue RHS) {
if (LHS.isSentinel() || RHS.isSentinel())
return LHSI == RHSI;
-
+
if (LHSI->getOpcode() != RHSI->getOpcode()) return false;
- return LHSI->isIdenticalTo(RHSI);
+ if (LHSI->isIdenticalTo(RHSI)) return true;
+
+ // If we're not strictly identical, we still might be a commutable instruction
+ if (BinaryOperator *LHSBinOp = dyn_cast<BinaryOperator>(LHSI)) {
+ if (!LHSBinOp->isCommutative())
+ return false;
+
+ assert(isa<BinaryOperator>(RHSI)
+ && "same opcode, but different instruction type?");
+ BinaryOperator *RHSBinOp = cast<BinaryOperator>(RHSI);
+
+ // Check overflow attributes
+ if (isa<OverflowingBinaryOperator>(LHSBinOp)) {
+ assert(isa<OverflowingBinaryOperator>(RHSBinOp)
+ && "same opcode, but different operator type?");
+ if (LHSBinOp->hasNoUnsignedWrap() != RHSBinOp->hasNoUnsignedWrap() ||
+ LHSBinOp->hasNoSignedWrap() != RHSBinOp->hasNoSignedWrap())
+ return false;
+ }
+
+ // Commuted equality
+ return LHSBinOp->getOperand(0) == RHSBinOp->getOperand(1) &&
+ LHSBinOp->getOperand(1) == RHSBinOp->getOperand(0);
+ }
+ if (CmpInst *LHSCmp = dyn_cast<CmpInst>(LHSI)) {
+ assert(isa<CmpInst>(RHSI)
+ && "same opcode, but different instruction type?");
+ CmpInst *RHSCmp = cast<CmpInst>(RHSI);
+ // Commuted equality
+ return LHSCmp->getOperand(0) == RHSCmp->getOperand(1) &&
+ LHSCmp->getOperand(1) == RHSCmp->getOperand(0) &&
+ LHSCmp->getSwappedPredicate() == RHSCmp->getPredicate();
+ }
+
+ return false;
}
//===----------------------------------------------------------------------===//
-// CallValue
+// CallValue
//===----------------------------------------------------------------------===//
namespace {
/// the scoped hash table.
struct CallValue {
Instruction *Inst;
-
+
CallValue(Instruction *I) : Inst(I) {
assert((isSentinel() || canHandle(I)) && "Inst can't be handled!");
}
-
+
bool isSentinel() const {
return Inst == DenseMapInfo<Instruction*>::getEmptyKey() ||
Inst == DenseMapInfo<Instruction*>::getTombstoneKey();
}
-
+
static bool canHandle(Instruction *Inst) {
- if (CallInst *CI = dyn_cast<CallInst>(Inst))
- return CI->onlyReadsMemory();
- return false;
+ // Don't value number anything that returns void.
+ if (Inst->getType()->isVoidTy())
+ return false;
+
+ CallInst *CI = dyn_cast<CallInst>(Inst);
+ if (CI == 0 || !CI->onlyReadsMemory())
+ return false;
+ return true;
}
};
}
template<> struct isPodLike<CallValue> {
static const bool value = true;
};
-
+
template<> struct DenseMapInfo<CallValue> {
static inline CallValue getEmptyKey() {
return DenseMapInfo<Instruction*>::getEmptyKey();
Instruction *Inst = Val.Inst;
// Hash in all of the operands as pointers.
unsigned Res = 0;
- for (unsigned i = 0, e = Inst->getNumOperands(); i != e; ++i)
- Res ^= getHash(Inst->getOperand(i)) << i;
+ for (unsigned i = 0, e = Inst->getNumOperands(); i != e; ++i) {
+ assert(!Inst->getOperand(i)->getType()->isMetadataTy() &&
+ "Cannot value number calls with metadata operands");
+ Res ^= getHash(Inst->getOperand(i)) << (i & 0xF);
+ }
+
// Mix in the opcode.
return (Res << 1) ^ Inst->getOpcode();
}
//===----------------------------------------------------------------------===//
-// EarlyCSE pass.
+// EarlyCSE pass.
//===----------------------------------------------------------------------===//
namespace {
-
+
/// EarlyCSE - This pass does a simple depth-first walk over the dominator
/// tree, eliminating trivially redundant instructions and using instsimplify
/// to canonicalize things as it goes. It is intended to be fast and catch
/// cases.
class EarlyCSE : public FunctionPass {
public:
- const TargetData *TD;
+ const DataLayout *TD;
+ const TargetLibraryInfo *TLI;
DominatorTree *DT;
typedef RecyclingAllocator<BumpPtrAllocator,
ScopedHashTableVal<SimpleValue, Value*> > AllocatorTy;
typedef ScopedHashTable<SimpleValue, Value*, DenseMapInfo<SimpleValue>,
AllocatorTy> ScopedHTType;
-
+
/// AvailableValues - This scoped hash table contains the current values of
/// all of our simple scalar expressions. As we walk down the domtree, we
/// look to see if instructions are in this: if so, we replace them with what
/// we find, otherwise we insert them so that dominated values can succeed in
/// their lookup.
ScopedHTType *AvailableValues;
-
+
/// AvailableLoads - This scoped hash table contains the current values
/// of loads. This allows us to get efficient access to dominating loads when
/// we have a fully redundant load. In addition to the most recent load, we
/// the current generation count. The current generation count is
/// incremented after every possibly writing memory operation, which ensures
/// that we only CSE loads with other loads that have no intervening store.
- typedef ScopedHashTable<Value*, std::pair<Value*, unsigned> > LoadHTType;
+ typedef RecyclingAllocator<BumpPtrAllocator,
+ ScopedHashTableVal<Value*, std::pair<Value*, unsigned> > > LoadMapAllocator;
+ typedef ScopedHashTable<Value*, std::pair<Value*, unsigned>,
+ DenseMapInfo<Value*>, LoadMapAllocator> LoadHTType;
LoadHTType *AvailableLoads;
-
+
/// AvailableCalls - This scoped hash table contains the current values
/// of read-only call values. It uses the same generation count as loads.
typedef ScopedHashTable<CallValue, std::pair<Value*, unsigned> > CallHTType;
CallHTType *AvailableCalls;
-
+
/// CurrentGeneration - This is the current generation of the memory value.
unsigned CurrentGeneration;
-
+
static char ID;
explicit EarlyCSE() : FunctionPass(ID) {
initializeEarlyCSEPass(*PassRegistry::getPassRegistry());
bool runOnFunction(Function &F);
private:
-
+
+ // NodeScope - almost a POD, but needs to call the constructors for the
+ // scoped hash tables so that a new scope gets pushed on. These are RAII so
+ // that the scope gets popped when the NodeScope is destroyed.
+ class NodeScope {
+ public:
+ NodeScope(ScopedHTType *availableValues,
+ LoadHTType *availableLoads,
+ CallHTType *availableCalls) :
+ Scope(*availableValues),
+ LoadScope(*availableLoads),
+ CallScope(*availableCalls) {}
+
+ private:
+ NodeScope(const NodeScope&) LLVM_DELETED_FUNCTION;
+ void operator=(const NodeScope&) LLVM_DELETED_FUNCTION;
+
+ ScopedHTType::ScopeTy Scope;
+ LoadHTType::ScopeTy LoadScope;
+ CallHTType::ScopeTy CallScope;
+ };
+
+ // StackNode - contains all the needed information to create a stack for
+ // doing a depth first tranversal of the tree. This includes scopes for
+ // values, loads, and calls as well as the generation. There is a child
+ // iterator so that the children do not need to be store spearately.
+ class StackNode {
+ public:
+ StackNode(ScopedHTType *availableValues,
+ LoadHTType *availableLoads,
+ CallHTType *availableCalls,
+ unsigned cg, DomTreeNode *n,
+ DomTreeNode::iterator child, DomTreeNode::iterator end) :
+ CurrentGeneration(cg), ChildGeneration(cg), Node(n),
+ ChildIter(child), EndIter(end),
+ Scopes(availableValues, availableLoads, availableCalls),
+ Processed(false) {}
+
+ // Accessors.
+ unsigned currentGeneration() { return CurrentGeneration; }
+ unsigned childGeneration() { return ChildGeneration; }
+ void childGeneration(unsigned generation) { ChildGeneration = generation; }
+ DomTreeNode *node() { return Node; }
+ DomTreeNode::iterator childIter() { return ChildIter; }
+ DomTreeNode *nextChild() {
+ DomTreeNode *child = *ChildIter;
+ ++ChildIter;
+ return child;
+ }
+ DomTreeNode::iterator end() { return EndIter; }
+ bool isProcessed() { return Processed; }
+ void process() { Processed = true; }
+
+ private:
+ StackNode(const StackNode&) LLVM_DELETED_FUNCTION;
+ void operator=(const StackNode&) LLVM_DELETED_FUNCTION;
+
+ // Members.
+ unsigned CurrentGeneration;
+ unsigned ChildGeneration;
+ DomTreeNode *Node;
+ DomTreeNode::iterator ChildIter;
+ DomTreeNode::iterator EndIter;
+ NodeScope Scopes;
+ bool Processed;
+ };
+
bool processNode(DomTreeNode *Node);
-
+
// This transformation requires dominator postdominator info
virtual void getAnalysisUsage(AnalysisUsage &AU) const {
AU.addRequired<DominatorTree>();
+ AU.addRequired<TargetLibraryInfo>();
AU.setPreservesCFG();
}
};
INITIALIZE_PASS_BEGIN(EarlyCSE, "early-cse", "Early CSE", false, false)
INITIALIZE_PASS_DEPENDENCY(DominatorTree)
+INITIALIZE_PASS_DEPENDENCY(TargetLibraryInfo)
INITIALIZE_PASS_END(EarlyCSE, "early-cse", "Early CSE", false, false)
bool EarlyCSE::processNode(DomTreeNode *Node) {
- // Define a scope in the scoped hash table. When we are done processing this
- // domtree node and recurse back up to our parent domtree node, this will pop
- // off all the values we install.
- ScopedHTType::ScopeTy Scope(*AvailableValues);
-
- // Define a scope for the load values so that anything we add will get
- // popped when we recurse back up to our parent domtree node.
- LoadHTType::ScopeTy LoadScope(*AvailableLoads);
-
- // Define a scope for the call values so that anything we add will get
- // popped when we recurse back up to our parent domtree node.
- CallHTType::ScopeTy CallScope(*AvailableCalls);
-
BasicBlock *BB = Node->getBlock();
-
+
// If this block has a single predecessor, then the predecessor is the parent
// of the domtree node and all of the live out memory values are still current
// in this block. If this block has multiple predecessors, then they could
// predecessors.
if (BB->getSinglePredecessor() == 0)
++CurrentGeneration;
-
+
+ /// LastStore - Keep track of the last non-volatile store that we saw... for
+ /// as long as there in no instruction that reads memory. If we see a store
+ /// to the same location, we delete the dead store. This zaps trivial dead
+ /// stores which can occur in bitfield code among other things.
+ StoreInst *LastStore = 0;
+
bool Changed = false;
// See if any instructions in the block can be eliminated. If so, do it. If
// not, add them to AvailableValues.
for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E; ) {
Instruction *Inst = I++;
-
+
// Dead instructions should just be removed.
- if (isInstructionTriviallyDead(Inst)) {
+ if (isInstructionTriviallyDead(Inst, TLI)) {
DEBUG(dbgs() << "EarlyCSE DCE: " << *Inst << '\n');
Inst->eraseFromParent();
Changed = true;
++NumSimplify;
continue;
}
-
+
// If the instruction can be simplified (e.g. X+0 = X) then replace it with
// its simpler value.
- if (Value *V = SimplifyInstruction(Inst, TD, DT)) {
+ if (Value *V = SimplifyInstruction(Inst, TD, TLI, DT)) {
DEBUG(dbgs() << "EarlyCSE Simplify: " << *Inst << " to: " << *V << '\n');
Inst->replaceAllUsesWith(V);
Inst->eraseFromParent();
++NumSimplify;
continue;
}
-
+
// If this is a simple instruction that we can value number, process it.
if (SimpleValue::canHandle(Inst)) {
// See if the instruction has an available value. If so, use it.
++NumCSE;
continue;
}
-
+
// Otherwise, just remember that this value is available.
AvailableValues->insert(Inst, Inst);
continue;
}
-
+
// If this is a non-volatile load, process it.
if (LoadInst *LI = dyn_cast<LoadInst>(Inst)) {
// Ignore volatile loads.
- if (LI->isVolatile()) continue;
-
+ if (!LI->isSimple()) {
+ LastStore = 0;
+ continue;
+ }
+
// If we have an available version of this load, and if it is the right
// generation, replace this instruction.
std::pair<Value*, unsigned> InVal =
++NumCSELoad;
continue;
}
-
+
// Otherwise, remember that we have this instruction.
AvailableLoads->insert(Inst->getOperand(0),
std::pair<Value*, unsigned>(Inst, CurrentGeneration));
+ LastStore = 0;
continue;
}
-
+
+ // If this instruction may read from memory, forget LastStore.
+ if (Inst->mayReadFromMemory())
+ LastStore = 0;
+
// If this is a read-only call, process it.
if (CallValue::canHandle(Inst)) {
// If we have an available version of this call, and if it is the right
++NumCSECall;
continue;
}
-
+
// Otherwise, remember that we have this instruction.
AvailableCalls->insert(Inst,
std::pair<Value*, unsigned>(Inst, CurrentGeneration));
continue;
}
-
+
// Okay, this isn't something we can CSE at all. Check to see if it is
// something that could modify memory. If so, our available memory values
// cannot be used so bump the generation count.
if (Inst->mayWriteToMemory()) {
++CurrentGeneration;
-
- // Okay, we just invalidated anything we knew about loaded values. Try to
- // salvage *something* by remembering that the stored value is a live
- // version of the pointer. It is safe to forward from volatile stores to
- // non-volatile loads, so we don't have to check for volatility of the
- // store.
+
if (StoreInst *SI = dyn_cast<StoreInst>(Inst)) {
+ // We do a trivial form of DSE if there are two stores to the same
+ // location with no intervening loads. Delete the earlier store.
+ if (LastStore &&
+ LastStore->getPointerOperand() == SI->getPointerOperand()) {
+ DEBUG(dbgs() << "EarlyCSE DEAD STORE: " << *LastStore << " due to: "
+ << *Inst << '\n');
+ LastStore->eraseFromParent();
+ Changed = true;
+ ++NumDSE;
+ LastStore = 0;
+ continue;
+ }
+
+ // Okay, we just invalidated anything we knew about loaded values. Try
+ // to salvage *something* by remembering that the stored value is a live
+ // version of the pointer. It is safe to forward from volatile stores
+ // to non-volatile loads, so we don't have to check for volatility of
+ // the store.
AvailableLoads->insert(SI->getPointerOperand(),
std::pair<Value*, unsigned>(SI->getValueOperand(), CurrentGeneration));
+
+ // Remember that this was the last store we saw for DSE.
+ if (SI->isSimple())
+ LastStore = SI;
}
}
}
-
- unsigned LiveOutGeneration = CurrentGeneration;
- for (DomTreeNode::iterator I = Node->begin(), E = Node->end(); I != E; ++I) {
- Changed |= processNode(*I);
- // Pop any generation changes off the stack from the recursive walk.
- CurrentGeneration = LiveOutGeneration;
- }
+
return Changed;
}
bool EarlyCSE::runOnFunction(Function &F) {
- TD = getAnalysisIfAvailable<TargetData>();
+ std::deque<StackNode *> nodesToProcess;
+
+ TD = getAnalysisIfAvailable<DataLayout>();
+ TLI = &getAnalysis<TargetLibraryInfo>();
DT = &getAnalysis<DominatorTree>();
-
+
// Tables that the pass uses when walking the domtree.
ScopedHTType AVTable;
AvailableValues = &AVTable;
AvailableLoads = &LoadTable;
CallHTType CallTable;
AvailableCalls = &CallTable;
-
+
CurrentGeneration = 0;
- return processNode(DT->getRootNode());
+ bool Changed = false;
+
+ // Process the root node.
+ nodesToProcess.push_front(
+ new StackNode(AvailableValues, AvailableLoads, AvailableCalls,
+ CurrentGeneration, DT->getRootNode(),
+ DT->getRootNode()->begin(),
+ DT->getRootNode()->end()));
+
+ // Save the current generation.
+ unsigned LiveOutGeneration = CurrentGeneration;
+
+ // Process the stack.
+ while (!nodesToProcess.empty()) {
+ // Grab the first item off the stack. Set the current generation, remove
+ // the node from the stack, and process it.
+ StackNode *NodeToProcess = nodesToProcess.front();
+
+ // Initialize class members.
+ CurrentGeneration = NodeToProcess->currentGeneration();
+
+ // Check if the node needs to be processed.
+ if (!NodeToProcess->isProcessed()) {
+ // Process the node.
+ Changed |= processNode(NodeToProcess->node());
+ NodeToProcess->childGeneration(CurrentGeneration);
+ NodeToProcess->process();
+ } else if (NodeToProcess->childIter() != NodeToProcess->end()) {
+ // Push the next child onto the stack.
+ DomTreeNode *child = NodeToProcess->nextChild();
+ nodesToProcess.push_front(
+ new StackNode(AvailableValues,
+ AvailableLoads,
+ AvailableCalls,
+ NodeToProcess->childGeneration(), child,
+ child->begin(), child->end()));
+ } else {
+ // It has been processed, and there are no more children to process,
+ // so delete it and pop it off the stack.
+ delete NodeToProcess;
+ nodesToProcess.pop_front();
+ }
+ } // while (!nodes...)
+
+ // Reset the current generation.
+ CurrentGeneration = LiveOutGeneration;
+
+ return Changed;
}