X-Git-Url: http://demsky.eecs.uci.edu/git/?a=blobdiff_plain;f=lib%2FTransforms%2FScalar%2FDeadStoreElimination.cpp;h=09c01d314124e65ff3c7767db50d7f99eeb93ee9;hb=040056fd11693ffc41ce9b777281c71705d0dc1f;hp=e29bad53c19cb9abc8f70421e58396e760c9382b;hpb=da16d8c00bb52e37967dc7a18a4bd39f16f10650;p=oota-llvm.git

diff --git a/lib/Transforms/Scalar/DeadStoreElimination.cpp b/lib/Transforms/Scalar/DeadStoreElimination.cpp
index e29bad53c19..09c01d31412 100644
--- a/lib/Transforms/Scalar/DeadStoreElimination.cpp
+++ b/lib/Transforms/Scalar/DeadStoreElimination.cpp
@@ -1,10 +1,10 @@
-//===- DeadStoreElimination.cpp - Dead Store Elimination ------------------===//
-// 
+//===- DeadStoreElimination.cpp - Fast Dead Store Elimination -------------===//
+//
 //                     The LLVM Compiler Infrastructure
 //
-// This file was developed by the LLVM research group and is distributed under
-// the University of Illinois Open Source License. See LICENSE.TXT for details.
-// 
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
 //===----------------------------------------------------------------------===//
 //
 // This file implements a trivial dead store elimination that only considers
@@ -15,154 +15,559 @@
 //
 //===----------------------------------------------------------------------===//
 
+#define DEBUG_TYPE "dse"
 #include "llvm/Transforms/Scalar.h"
-#include "llvm/DerivedTypes.h"
+#include "llvm/Constants.h"
 #include "llvm/Function.h"
 #include "llvm/Instructions.h"
+#include "llvm/IntrinsicInst.h"
+#include "llvm/Pass.h"
+#include "llvm/ADT/SmallPtrSet.h"
+#include "llvm/ADT/Statistic.h"
 #include "llvm/Analysis/AliasAnalysis.h"
-#include "llvm/Analysis/AliasSetTracker.h"
+#include "llvm/Analysis/Dominators.h"
+#include "llvm/Analysis/MemoryBuiltins.h"
+#include "llvm/Analysis/MemoryDependenceAnalysis.h"
 #include "llvm/Target/TargetData.h"
 #include "llvm/Transforms/Utils/Local.h"
-#include "llvm/ADT/SetVector.h"
-#include "llvm/ADT/Statistic.h"
 using namespace llvm;
 
-namespace {
-  Statistic<> NumStores("dse", "Number of stores deleted");
-  Statistic<> NumOther ("dse", "Number of other instrs removed");
+STATISTIC(NumFastStores, "Number of stores deleted");
+STATISTIC(NumFastOther , "Number of other instrs removed");
 
+namespace {
   struct DSE : public FunctionPass {
+    TargetData *TD;
+
+    static char ID; // Pass identification, replacement for typeid
+    DSE() : FunctionPass(&ID) {}
 
     virtual bool runOnFunction(Function &F) {
       bool Changed = false;
+      
+      DominatorTree &DT = getAnalysis<DominatorTree>();
+      
       for (Function::iterator I = F.begin(), E = F.end(); I != E; ++I)
-        Changed |= runOnBasicBlock(*I);
+        // Only check non-dead blocks.  Dead blocks may have strange pointer
+        // cycles that will confuse alias analysis.
+        if (DT.isReachableFromEntry(I))
+          Changed |= runOnBasicBlock(*I);
       return Changed;
     }
     
     bool runOnBasicBlock(BasicBlock &BB);
+    bool handleFreeWithNonTrivialDependency(Instruction *F, MemDepResult Dep);
+    bool handleEndBlock(BasicBlock &BB);
+    bool RemoveUndeadPointers(Value *Ptr, uint64_t killPointerSize,
+                              BasicBlock::iterator &BBI,
+                              SmallPtrSet<Value*, 64> &deadPointers);
+    void DeleteDeadInstruction(Instruction *I,
+                               SmallPtrSet<Value*, 64> *deadPointers = 0);
     
-    void DeleteDeadInstructionChains(Instruction *I,
-                                     SetVector<Instruction*> &DeadInsts);
 
     // getAnalysisUsage - We require post dominance frontiers (aka Control
     // Dependence Graph)
     virtual void getAnalysisUsage(AnalysisUsage &AU) const {
       AU.setPreservesCFG();
-      AU.addRequired<TargetData>();
+      AU.addRequired<DominatorTree>();
       AU.addRequired<AliasAnalysis>();
+      AU.addRequired<MemoryDependenceAnalysis>();
+      AU.addPreserved<DominatorTree>();
       AU.addPreserved<AliasAnalysis>();
+      AU.addPreserved<MemoryDependenceAnalysis>();
     }
+
+    unsigned getPointerSize(Value *V) const;
   };
-  RegisterOpt<DSE> X("dse", "Dead Store Elimination");
 }
 
+char DSE::ID = 0;
+static RegisterPass<DSE> X("dse", "Dead Store Elimination");
+
 FunctionPass *llvm::createDeadStoreEliminationPass() { return new DSE(); }
 
-bool DSE::runOnBasicBlock(BasicBlock &BB) {
-  TargetData &TD = getAnalysis<TargetData>();
-  AliasAnalysis &AA = getAnalysis<AliasAnalysis>();
-  AliasSetTracker KillLocs(AA);
-
-  // If this block ends in a return, unwind, unreachable, and eventually
-  // tailcall, then all allocas are dead at its end.
-  if (BB.getTerminator()->getNumSuccessors() == 0) {
-    BasicBlock *Entry = BB.getParent()->begin();
-    for (BasicBlock::iterator I = Entry->begin(), E = Entry->end(); I != E; ++I)
-      if (AllocaInst *AI = dyn_cast<AllocaInst>(I)) {
-        unsigned Size = ~0U;
-        if (!AI->isArrayAllocation() &&
-            AI->getType()->getElementType()->isSized())
-          Size = TD.getTypeSize(AI->getType()->getElementType());
-        KillLocs.add(AI, Size);
-      }
+/// doesClobberMemory - Does this instruction clobber (write without reading)
+/// some memory?
+static bool doesClobberMemory(Instruction *I) {
+  if (isa<StoreInst>(I))
+    return true;
+  if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(I)) {
+    switch (II->getIntrinsicID()) {
+    default:
+      return false;
+    case Intrinsic::memset:
+    case Intrinsic::memmove:
+    case Intrinsic::memcpy:
+    case Intrinsic::init_trampoline:
+    case Intrinsic::lifetime_end:
+      return true;
+    }
   }
+  return false;
+}
+
+/// isElidable - If the value of this instruction and the memory it writes to is
+/// unused, may we delete this instrtction?
+static bool isElidable(Instruction *I) {
+  assert(doesClobberMemory(I));
+  if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(I))
+    return II->getIntrinsicID() != Intrinsic::lifetime_end;
+  if (StoreInst *SI = dyn_cast<StoreInst>(I))
+    return !SI->isVolatile();
+  return true;
+}
+
+/// getPointerOperand - Return the pointer that is being clobbered.
+static Value *getPointerOperand(Instruction *I) {
+  assert(doesClobberMemory(I));
+  if (StoreInst *SI = dyn_cast<StoreInst>(I))
+    return SI->getPointerOperand();
+  if (MemIntrinsic *MI = dyn_cast<MemIntrinsic>(I))
+    return MI->getOperand(1);
+  
+  switch (cast<IntrinsicInst>(I)->getIntrinsicID()) {
+  default: assert(false && "Unexpected intrinsic!");
+  case Intrinsic::init_trampoline:
+    return I->getOperand(1);
+  case Intrinsic::lifetime_end:
+    return I->getOperand(2);
+  }
+}
+
+/// getStoreSize - Return the length in bytes of the write by the clobbering
+/// instruction. If variable or unknown, returns -1.
+static unsigned getStoreSize(Instruction *I, const TargetData *TD) {
+  assert(doesClobberMemory(I));
+  if (StoreInst *SI = dyn_cast<StoreInst>(I)) {
+    if (!TD) return -1u;
+    return TD->getTypeStoreSize(SI->getOperand(0)->getType());
+  }
+
+  Value *Len;
+  if (MemIntrinsic *MI = dyn_cast<MemIntrinsic>(I)) {
+    Len = MI->getLength();
+  } else {
+    switch (cast<IntrinsicInst>(I)->getIntrinsicID()) {
+    default: assert(false && "Unexpected intrinsic!");
+    case Intrinsic::init_trampoline:
+      return -1u;
+    case Intrinsic::lifetime_end:
+      Len = I->getOperand(1);
+      break;
+    }
+  }
+  if (ConstantInt *LenCI = dyn_cast<ConstantInt>(Len))
+    if (!LenCI->isAllOnesValue())
+      return LenCI->getZExtValue();
+  return -1u;
+}
+
+/// isStoreAtLeastAsWideAs - Return true if the size of the store in I1 is
+/// greater than or equal to the store in I2.  This returns false if we don't
+/// know.
+///
+static bool isStoreAtLeastAsWideAs(Instruction *I1, Instruction *I2,
+                                   const TargetData *TD) {
+  const Type *I1Ty = getPointerOperand(I1)->getType();
+  const Type *I2Ty = getPointerOperand(I2)->getType();
+  
+  // Exactly the same type, must have exactly the same size.
+  if (I1Ty == I2Ty) return true;
+  
+  int I1Size = getStoreSize(I1, TD);
+  int I2Size = getStoreSize(I2, TD);
+  
+  return I1Size != -1 && I2Size != -1 && I1Size >= I2Size;
+}
 
-  // PotentiallyDeadInsts - Deleting dead stores from the program can make other
-  // instructions die if they were only used as operands to stores.  Keep track
-  // of the operands to stores so that we can try deleting them at the end of
-  // the traversal.
-  SetVector<Instruction*> PotentiallyDeadInsts;
+bool DSE::runOnBasicBlock(BasicBlock &BB) {
+  MemoryDependenceAnalysis &MD = getAnalysis<MemoryDependenceAnalysis>();
+  TD = getAnalysisIfAvailable<TargetData>();
 
   bool MadeChange = false;
-  for (BasicBlock::iterator BBI = BB.end(); BBI != BB.begin(); ) {
-    Instruction *I = --BBI;   // Keep moving iterator backwards
+  
+  // Do a top-down walk on the BB.
+  for (BasicBlock::iterator BBI = BB.begin(), BBE = BB.end(); BBI != BBE; ) {
+    Instruction *Inst = BBI++;
     
-    // If this is a free instruction, it makes the free'd location dead!
-    if (FreeInst *FI = dyn_cast<FreeInst>(I)) {
-      // Free instructions make any stores to the free'd location dead.
-      KillLocs.add(FI);
+    // If we find a store or a free, get its memory dependence.
+    if (!doesClobberMemory(Inst) && !isFreeCall(Inst))
+      continue;
+    
+    MemDepResult InstDep = MD.getDependency(Inst);
+    
+    // Ignore non-local stores.
+    // FIXME: cross-block DSE would be fun. :)
+    if (InstDep.isNonLocal()) continue;
+  
+    // Handle frees whose dependencies are non-trivial.
+    if (isFreeCall(Inst)) {
+      MadeChange |= handleFreeWithNonTrivialDependency(Inst, InstDep);
       continue;
     }
+    
+    // If not a definite must-alias dependency, ignore it.
+    if (!InstDep.isDef())
+      continue;
+    
+    // If this is a store-store dependence, then the previous store is dead so
+    // long as this store is at least as big as it.
+    if (doesClobberMemory(InstDep.getInst())) {
+      Instruction *DepStore = InstDep.getInst();
+      if (isStoreAtLeastAsWideAs(Inst, DepStore, TD) &&
+          isElidable(DepStore)) {
+        // Delete the store and now-dead instructions that feed it.
+        DeleteDeadInstruction(DepStore);
+        NumFastStores++;
+        MadeChange = true;
 
-    if (!isa<StoreInst>(I) || cast<StoreInst>(I)->isVolatile()) {
-      // If this is a non-store instruction, it makes everything referenced no
-      // longer killed.  Remove anything aliased from the alias set tracker.
-      KillLocs.remove(I);
+        // DeleteDeadInstruction can delete the current instruction in loop
+        // cases, reset BBI.
+        BBI = Inst;
+        if (BBI != BB.begin())
+          --BBI;
+        continue;
+      }
+    }
+    
+    if (!isElidable(Inst))
       continue;
+    
+    // If we're storing the same value back to a pointer that we just
+    // loaded from, then the store can be removed.
+    if (StoreInst *SI = dyn_cast<StoreInst>(Inst)) {
+      if (LoadInst *DepLoad = dyn_cast<LoadInst>(InstDep.getInst())) {
+        if (SI->getPointerOperand() == DepLoad->getPointerOperand() &&
+            SI->getOperand(0) == DepLoad) {
+          // DeleteDeadInstruction can delete the current instruction.  Save BBI
+          // in case we need it.
+          WeakVH NextInst(BBI);
+          
+          DeleteDeadInstruction(SI);
+          
+          if (NextInst == 0)  // Next instruction deleted.
+            BBI = BB.begin();
+          else if (BBI != BB.begin())  // Revisit this instruction if possible.
+            --BBI;
+          NumFastStores++;
+          MadeChange = true;
+          continue;
+        }
+      }
     }
+    
+    // If this is a lifetime end marker, we can throw away the store.
+    if (IntrinsicInst *II = dyn_cast<IntrinsicInst>(InstDep.getInst())) {
+      if (II->getIntrinsicID() == Intrinsic::lifetime_end) {
+        // Delete the store and now-dead instructions that feed it.
+        // DeleteDeadInstruction can delete the current instruction.  Save BBI
+        // in case we need it.
+        WeakVH NextInst(BBI);
+        
+        DeleteDeadInstruction(Inst);
+        
+        if (NextInst == 0)  // Next instruction deleted.
+          BBI = BB.begin();
+        else if (BBI != BB.begin())  // Revisit this instruction if possible.
+          --BBI;
+        NumFastStores++;
+        MadeChange = true;
+        continue;
+      }
+    }
+  }
+  
+  // If this block ends in a return, unwind, or unreachable, all allocas are
+  // dead at its end, which means stores to them are also dead.
+  if (BB.getTerminator()->getNumSuccessors() == 0)
+    MadeChange |= handleEndBlock(BB);
+  
+  return MadeChange;
+}
+
+/// handleFreeWithNonTrivialDependency - Handle frees of entire structures whose
+/// dependency is a store to a field of that structure.
+bool DSE::handleFreeWithNonTrivialDependency(Instruction *F, MemDepResult Dep) {
+  AliasAnalysis &AA = getAnalysis<AliasAnalysis>();
+  
+  Instruction *Dependency = Dep.getInst();
+  if (!Dependency || !doesClobberMemory(Dependency) || !isElidable(Dependency))
+    return false;
+  
+  Value *DepPointer = getPointerOperand(Dependency)->getUnderlyingObject();
+
+  // Check for aliasing.
+  if (AA.alias(F->getOperand(1), 1, DepPointer, 1) !=
+         AliasAnalysis::MustAlias)
+    return false;
+  
+  // DCE instructions only used to calculate that store
+  DeleteDeadInstruction(Dependency);
+  NumFastStores++;
+  return true;
+}
 
-    // If this is a non-volatile store instruction, and if it is already in
-    // the stored location is already in the tracker, then this is a dead
-    // store.  We can just delete it here, but while we're at it, we also
-    // delete any trivially dead expression chains.
-    unsigned ValSize = TD.getTypeSize(I->getOperand(0)->getType());
-    Value *Ptr = I->getOperand(1);
-
-    if (AliasSet *AS = KillLocs.getAliasSetForPointerIfExists(Ptr, ValSize))
-      for (AliasSet::iterator ASI = AS->begin(), E = AS->end(); ASI != E; ++ASI)
-        if (ASI.getSize() >= ValSize &&  // Overwriting all of this store.
-            AA.alias(ASI.getPointer(), ASI.getSize(), Ptr, ValSize)
-               == AliasAnalysis::MustAlias) {
-          // If we found a must alias in the killed set, then this store really
-          // is dead.  Remember that the various operands of the store now have
-          // fewer users.  At the end we will see if we can delete any values
-          // that are dead as part of the store becoming dead.
-          if (Instruction *Op = dyn_cast<Instruction>(I->getOperand(0)))
-            PotentiallyDeadInsts.insert(Op);
-          if (Instruction *Op = dyn_cast<Instruction>(Ptr))
-            PotentiallyDeadInsts.insert(Op);
-
-          // Delete it now.
-          ++BBI;                        // Don't invalidate iterator.
-          BB.getInstList().erase(I);    // Nuke the store!
-          ++NumStores;
+/// handleEndBlock - Remove dead stores to stack-allocated locations in the
+/// function end block.  Ex:
+/// %A = alloca i32
+/// ...
+/// store i32 1, i32* %A
+/// ret void
+bool DSE::handleEndBlock(BasicBlock &BB) {
+  AliasAnalysis &AA = getAnalysis<AliasAnalysis>();
+  
+  bool MadeChange = false;
+  
+  // Pointers alloca'd in this function are dead in the end block
+  SmallPtrSet<Value*, 64> deadPointers;
+  
+  // Find all of the alloca'd pointers in the entry block.
+  BasicBlock *Entry = BB.getParent()->begin();
+  for (BasicBlock::iterator I = Entry->begin(), E = Entry->end(); I != E; ++I)
+    if (AllocaInst *AI = dyn_cast<AllocaInst>(I))
+      deadPointers.insert(AI);
+  
+  // Treat byval arguments the same, stores to them are dead at the end of the
+  // function.
+  for (Function::arg_iterator AI = BB.getParent()->arg_begin(),
+       AE = BB.getParent()->arg_end(); AI != AE; ++AI)
+    if (AI->hasByValAttr())
+      deadPointers.insert(AI);
+  
+  // Scan the basic block backwards
+  for (BasicBlock::iterator BBI = BB.end(); BBI != BB.begin(); ){
+    --BBI;
+    
+    // If we find a store whose pointer is dead.
+    if (doesClobberMemory(BBI)) {
+      if (isElidable(BBI)) {
+        // See through pointer-to-pointer bitcasts
+        Value *pointerOperand = getPointerOperand(BBI)->getUnderlyingObject();
+
+        // Alloca'd pointers or byval arguments (which are functionally like
+        // alloca's) are valid candidates for removal.
+        if (deadPointers.count(pointerOperand)) {
+          // DCE instructions only used to calculate that store.
+          Instruction *Dead = BBI;
+          BBI++;
+          DeleteDeadInstruction(Dead, &deadPointers);
+          NumFastStores++;
           MadeChange = true;
-          goto BigContinue;
+          continue;
         }
+      }
+      
+      // Because a memcpy or memmove is also a load, we can't skip it if we
+      // didn't remove it.
+      if (!isa<MemTransferInst>(BBI))
+        continue;
+    }
+    
+    Value *killPointer = 0;
+    uint64_t killPointerSize = ~0UL;
+    
+    // If we encounter a use of the pointer, it is no longer considered dead
+    if (LoadInst *L = dyn_cast<LoadInst>(BBI)) {
+      // However, if this load is unused and not volatile, we can go ahead and
+      // remove it, and not have to worry about it making our pointer undead!
+      if (L->use_empty() && !L->isVolatile()) {
+        BBI++;
+        DeleteDeadInstruction(L, &deadPointers);
+        NumFastOther++;
+        MadeChange = true;
+        continue;
+      }
+      
+      killPointer = L->getPointerOperand();
+    } else if (VAArgInst *V = dyn_cast<VAArgInst>(BBI)) {
+      killPointer = V->getOperand(0);
+    } else if (isa<MemTransferInst>(BBI) &&
+               isa<ConstantInt>(cast<MemTransferInst>(BBI)->getLength())) {
+      killPointer = cast<MemTransferInst>(BBI)->getSource();
+      killPointerSize = cast<ConstantInt>(
+                       cast<MemTransferInst>(BBI)->getLength())->getZExtValue();
+    } else if (AllocaInst *A = dyn_cast<AllocaInst>(BBI)) {
+      deadPointers.erase(A);
+      
+      // Dead alloca's can be DCE'd when we reach them
+      if (A->use_empty()) {
+        BBI++;
+        DeleteDeadInstruction(A, &deadPointers);
+        NumFastOther++;
+        MadeChange = true;
+      }
+      
+      continue;
+    } else if (CallSite::get(BBI).getInstruction() != 0) {
+      // If this call does not access memory, it can't
+      // be undeadifying any of our pointers.
+      CallSite CS = CallSite::get(BBI);
+      if (AA.doesNotAccessMemory(CS))
+        continue;
+      
+      unsigned modRef = 0;
+      unsigned other = 0;
+      
+      // Remove any pointers made undead by the call from the dead set
+      std::vector<Value*> dead;
+      for (SmallPtrSet<Value*, 64>::iterator I = deadPointers.begin(),
+           E = deadPointers.end(); I != E; ++I) {
+        // HACK: if we detect that our AA is imprecise, it's not
+        // worth it to scan the rest of the deadPointers set.  Just
+        // assume that the AA will return ModRef for everything, and
+        // go ahead and bail.
+        if (modRef >= 16 && other == 0) {
+          deadPointers.clear();
+          return MadeChange;
+        }
+        
+        // See if the call site touches it
+        AliasAnalysis::ModRefResult A = AA.getModRefInfo(CS, *I,
+                                                         getPointerSize(*I));
+        
+        if (A == AliasAnalysis::ModRef)
+          modRef++;
+        else
+          other++;
+        
+        if (A == AliasAnalysis::ModRef || A == AliasAnalysis::Ref)
+          dead.push_back(*I);
+      }
 
-    // Otherwise, this is a non-dead store just add it to the set of dead
-    // locations.
-    KillLocs.add(cast<StoreInst>(I));
-  BigContinue:;
-  }
+      for (std::vector<Value*>::iterator I = dead.begin(), E = dead.end();
+           I != E; ++I)
+        deadPointers.erase(*I);
+      
+      continue;
+    } else if (isInstructionTriviallyDead(BBI)) {
+      // For any non-memory-affecting non-terminators, DCE them as we reach them
+      Instruction *Inst = BBI;
+      BBI++;
+      DeleteDeadInstruction(Inst, &deadPointers);
+      NumFastOther++;
+      MadeChange = true;
+      continue;
+    }
+    
+    if (!killPointer)
+      continue;
+
+    killPointer = killPointer->getUnderlyingObject();
 
-  while (!PotentiallyDeadInsts.empty()) {
-    Instruction *I = PotentiallyDeadInsts.back();
-    PotentiallyDeadInsts.pop_back();
-    DeleteDeadInstructionChains(I, PotentiallyDeadInsts);
+    // Deal with undead pointers
+    MadeChange |= RemoveUndeadPointers(killPointer, killPointerSize, BBI,
+                                       deadPointers);
   }
+  
   return MadeChange;
 }
 
-void DSE::DeleteDeadInstructionChains(Instruction *I,
-                                      SetVector<Instruction*> &DeadInsts) {
-  // Instruction must be dead.
-  if (!I->use_empty() || !isInstructionTriviallyDead(I)) return;
+/// RemoveUndeadPointers - check for uses of a pointer that make it
+/// undead when scanning for dead stores to alloca's.
+bool DSE::RemoveUndeadPointers(Value *killPointer, uint64_t killPointerSize,
+                               BasicBlock::iterator &BBI,
+                               SmallPtrSet<Value*, 64> &deadPointers) {
+  AliasAnalysis &AA = getAnalysis<AliasAnalysis>();
 
-  // Let the alias analysis know that we have nuked a value.
-  getAnalysis<AliasAnalysis>().deleteValue(I);
+  // If the kill pointer can be easily reduced to an alloca,
+  // don't bother doing extraneous AA queries.
+  if (deadPointers.count(killPointer)) {
+    deadPointers.erase(killPointer);
+    return false;
+  }
+  
+  // A global can't be in the dead pointer set.
+  if (isa<GlobalValue>(killPointer))
+    return false;
+  
+  bool MadeChange = false;
+  
+  SmallVector<Value*, 16> undead;
+  
+  for (SmallPtrSet<Value*, 64>::iterator I = deadPointers.begin(),
+       E = deadPointers.end(); I != E; ++I) {
+    // See if this pointer could alias it
+    AliasAnalysis::AliasResult A = AA.alias(*I, getPointerSize(*I),
+                                            killPointer, killPointerSize);
 
-  // See if this made any operands dead.  We do it this way in case the
-  // instruction uses the same operand twice.  We don't want to delete a
-  // value then reference it.
-  while (unsigned NumOps = I->getNumOperands()) {
-    Instruction *Op = dyn_cast<Instruction>(I->getOperand(NumOps-1));
-    I->op_erase(I->op_end()-1);         // Drop from the operand list.
-    
-    if (Op) DeadInsts.insert(Op);       // Attempt to nuke it later.
+    // If it must-alias and a store, we can delete it
+    if (isa<StoreInst>(BBI) && A == AliasAnalysis::MustAlias) {
+      StoreInst *S = cast<StoreInst>(BBI);
+
+      // Remove it!
+      ++BBI;
+      DeleteDeadInstruction(S, &deadPointers);
+      NumFastStores++;
+      MadeChange = true;
+
+      continue;
+
+      // Otherwise, it is undead
+    } else if (A != AliasAnalysis::NoAlias)
+      undead.push_back(*I);
   }
+
+  for (SmallVector<Value*, 16>::iterator I = undead.begin(), E = undead.end();
+       I != E; ++I)
+      deadPointers.erase(*I);
   
-  I->getParent()->getInstList().erase(I);
-  ++NumOther;
+  return MadeChange;
+}
+
+/// DeleteDeadInstruction - Delete this instruction.  Before we do, go through
+/// and zero out all the operands of this instruction.  If any of them become
+/// dead, delete them and the computation tree that feeds them.
+///
+/// If ValueSet is non-null, remove any deleted instructions from it as well.
+///
+void DSE::DeleteDeadInstruction(Instruction *I,
+                                SmallPtrSet<Value*, 64> *ValueSet) {
+  SmallVector<Instruction*, 32> NowDeadInsts;
+  
+  NowDeadInsts.push_back(I);
+  --NumFastOther;
+
+  // Before we touch this instruction, remove it from memdep!
+  MemoryDependenceAnalysis &MDA = getAnalysis<MemoryDependenceAnalysis>();
+  do {
+    Instruction *DeadInst = NowDeadInsts.pop_back_val();
+    
+    ++NumFastOther;
+    
+    // This instruction is dead, zap it, in stages.  Start by removing it from
+    // MemDep, which needs to know the operands and needs it to be in the
+    // function.
+    MDA.removeInstruction(DeadInst);
+    
+    for (unsigned op = 0, e = DeadInst->getNumOperands(); op != e; ++op) {
+      Value *Op = DeadInst->getOperand(op);
+      DeadInst->setOperand(op, 0);
+      
+      // If this operand just became dead, add it to the NowDeadInsts list.
+      if (!Op->use_empty()) continue;
+      
+      if (Instruction *OpI = dyn_cast<Instruction>(Op))
+        if (isInstructionTriviallyDead(OpI))
+          NowDeadInsts.push_back(OpI);
+    }
+    
+    DeadInst->eraseFromParent();
+    
+    if (ValueSet) ValueSet->erase(DeadInst);
+  } while (!NowDeadInsts.empty());
+}
+
+unsigned DSE::getPointerSize(Value *V) const {
+  if (TD) {
+    if (AllocaInst *A = dyn_cast<AllocaInst>(V)) {
+      // Get size information for the alloca
+      if (ConstantInt *C = dyn_cast<ConstantInt>(A->getArraySize()))
+        return C->getZExtValue() * TD->getTypeAllocSize(A->getAllocatedType());
+    } else {
+      assert(isa<Argument>(V) && "Expected AllocaInst or Argument!");
+      const PointerType *PT = cast<PointerType>(V->getType());
+      return TD->getTypeAllocSize(PT->getElementType());
+    }
+  }
+  return ~0U;
 }