Update LLVM for 3.5 rebase (r209712).

Change-Id: I149556c940fb7dc92d075273c87ff584f400941f
diff --git a/lib/Transforms/Utils/AddDiscriminators.cpp b/lib/Transforms/Utils/AddDiscriminators.cpp
index f42635e..196ac79 100644
--- a/lib/Transforms/Utils/AddDiscriminators.cpp
+++ b/lib/Transforms/Utils/AddDiscriminators.cpp
@@ -52,8 +52,6 @@
 // http://wiki.dwarfstd.org/index.php?title=Path_Discriminators
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "add-discriminators"
-
 #include "llvm/Transforms/Scalar.h"
 #include "llvm/IR/BasicBlock.h"
 #include "llvm/IR/Constants.h"
@@ -69,6 +67,8 @@
 
 using namespace llvm;
 
+#define DEBUG_TYPE "add-discriminators"
+
 namespace {
   struct AddDiscriminators : public FunctionPass {
     static char ID; // Pass identification, replacement for typeid
@@ -99,7 +99,7 @@
 
 static bool hasDebugInfo(const Function &F) {
   NamedMDNode *CUNodes = F.getParent()->getNamedMetadata("llvm.dbg.cu");
-  return CUNodes != 0;
+  return CUNodes != nullptr;
 }
 
 /// \brief Assign DWARF discriminators.
@@ -154,10 +154,15 @@
 /// file and line location as I2. This new lexical block will have a
 /// different discriminator number than I1.
 bool AddDiscriminators::runOnFunction(Function &F) {
-  // No need to do anything if there is no debug info for this function.
   // If the function has debug information, but the user has disabled
   // discriminators, do nothing.
-  if (!hasDebugInfo(F) || NoDiscriminators) return false;
+  // Simlarly, if the function has no debug info, do nothing.
+  // Finally, if this module is built with dwarf versions earlier than 4,
+  // do nothing (discriminator support is a DWARF 4 feature).
+  if (NoDiscriminators ||
+      !hasDebugInfo(F) ||
+      F.getParent()->getDwarfVersion() < 4)
+    return false;
 
   bool Changed = false;
   Module *M = F.getParent();
diff --git a/lib/Transforms/Utils/Android.mk b/lib/Transforms/Utils/Android.mk
index ab4d8a8..cbd8dd0 100644
--- a/lib/Transforms/Utils/Android.mk
+++ b/lib/Transforms/Utils/Android.mk
@@ -11,6 +11,7 @@
   CloneModule.cpp \
   CmpInstAnalysis.cpp \
   CodeExtractor.cpp \
+  CtorUtils.cpp \
   DemoteRegToStack.cpp \
   GlobalStatus.cpp \
   InlineFunction.cpp \
diff --git a/lib/Transforms/Utils/BasicBlockUtils.cpp b/lib/Transforms/Utils/BasicBlockUtils.cpp
index b3cd5ce..80b7e22 100644
--- a/lib/Transforms/Utils/BasicBlockUtils.cpp
+++ b/lib/Transforms/Utils/BasicBlockUtils.cpp
@@ -68,8 +68,8 @@
 void llvm::FoldSingleEntryPHINodes(BasicBlock *BB, Pass *P) {
   if (!isa<PHINode>(BB->begin())) return;
 
-  AliasAnalysis *AA = 0;
-  MemoryDependenceAnalysis *MemDep = 0;
+  AliasAnalysis *AA = nullptr;
+  MemoryDependenceAnalysis *MemDep = nullptr;
   if (P) {
     AA = P->getAnalysisIfAvailable<AliasAnalysis>();
     MemDep = P->getAnalysisIfAvailable<MemoryDependenceAnalysis>();
@@ -130,7 +130,7 @@
   BasicBlock *OnlySucc = BB;
   for (; SI != SE; ++SI)
     if (*SI != OnlySucc) {
-      OnlySucc = 0;     // There are multiple distinct successors!
+      OnlySucc = nullptr;     // There are multiple distinct successors!
       break;
     }
 
@@ -217,7 +217,7 @@
 ///
 void llvm::ReplaceInstWithInst(BasicBlock::InstListType &BIL,
                                BasicBlock::iterator &BI, Instruction *I) {
-  assert(I->getParent() == 0 &&
+  assert(I->getParent() == nullptr &&
          "ReplaceInstWithInst: Instruction already inserted into basic block!");
 
   // Insert the new instruction into the basic block...
@@ -254,7 +254,7 @@
     // If the successor only has a single pred, split the top of the successor
     // block.
     assert(SP == BB && "CFG broken");
-    SP = NULL;
+    SP = nullptr;
     return SplitBlock(Succ, Succ->begin(), P);
   }
 
@@ -310,7 +310,7 @@
   if (!P) return;
 
   LoopInfo *LI = P->getAnalysisIfAvailable<LoopInfo>();
-  Loop *L = LI ? LI->getLoopFor(OldBB) : 0;
+  Loop *L = LI ? LI->getLoopFor(OldBB) : nullptr;
 
   // If we need to preserve loop analyses, collect some information about how
   // this split will affect loops.
@@ -351,7 +351,7 @@
     // loop). To find this, examine each of the predecessors and determine which
     // loops enclose them, and select the most-nested loop which contains the
     // loop containing the block being split.
-    Loop *InnermostPredLoop = 0;
+    Loop *InnermostPredLoop = nullptr;
     for (ArrayRef<BasicBlock*>::iterator
            i = Preds.begin(), e = Preds.end(); i != e; ++i) {
       BasicBlock *Pred = *i;
@@ -384,51 +384,68 @@
                            ArrayRef<BasicBlock*> Preds, BranchInst *BI,
                            Pass *P, bool HasLoopExit) {
   // Otherwise, create a new PHI node in NewBB for each PHI node in OrigBB.
-  AliasAnalysis *AA = P ? P->getAnalysisIfAvailable<AliasAnalysis>() : 0;
+  AliasAnalysis *AA = P ? P->getAnalysisIfAvailable<AliasAnalysis>() : nullptr;
+  SmallPtrSet<BasicBlock *, 16> PredSet(Preds.begin(), Preds.end());
   for (BasicBlock::iterator I = OrigBB->begin(); isa<PHINode>(I); ) {
     PHINode *PN = cast<PHINode>(I++);
 
     // Check to see if all of the values coming in are the same.  If so, we
     // don't need to create a new PHI node, unless it's needed for LCSSA.
-    Value *InVal = 0;
+    Value *InVal = nullptr;
     if (!HasLoopExit) {
       InVal = PN->getIncomingValueForBlock(Preds[0]);
-      for (unsigned i = 1, e = Preds.size(); i != e; ++i)
-        if (InVal != PN->getIncomingValueForBlock(Preds[i])) {
-          InVal = 0;
+      for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
+        if (!PredSet.count(PN->getIncomingBlock(i)))
+          continue;
+        if (!InVal)
+          InVal = PN->getIncomingValue(i);
+        else if (InVal != PN->getIncomingValue(i)) {
+          InVal = nullptr;
           break;
         }
+      }
     }
 
     if (InVal) {
       // If all incoming values for the new PHI would be the same, just don't
       // make a new PHI.  Instead, just remove the incoming values from the old
       // PHI.
-      for (unsigned i = 0, e = Preds.size(); i != e; ++i) {
-        // Explicitly check the BB index here to handle duplicates in Preds.
-        int Idx = PN->getBasicBlockIndex(Preds[i]);
-        if (Idx >= 0)
-          PN->removeIncomingValue(Idx, false);
-      }
-    } else {
-      // If the values coming into the block are not the same, we need a PHI.
-      // Create the new PHI node, insert it into NewBB at the end of the block
-      PHINode *NewPHI =
-        PHINode::Create(PN->getType(), Preds.size(), PN->getName() + ".ph", BI);
-      if (AA) AA->copyValue(PN, NewPHI);
 
-      // Move all of the PHI values for 'Preds' to the new PHI.
-      for (unsigned i = 0, e = Preds.size(); i != e; ++i) {
-        Value *V = PN->removeIncomingValue(Preds[i], false);
-        NewPHI->addIncoming(V, Preds[i]);
-      }
+      // NOTE! This loop walks backwards for a reason! First off, this minimizes
+      // the cost of removal if we end up removing a large number of values, and
+      // second off, this ensures that the indices for the incoming values
+      // aren't invalidated when we remove one.
+      for (int64_t i = PN->getNumIncomingValues() - 1; i >= 0; --i)
+        if (PredSet.count(PN->getIncomingBlock(i)))
+          PN->removeIncomingValue(i, false);
 
-      InVal = NewPHI;
+      // Add an incoming value to the PHI node in the loop for the preheader
+      // edge.
+      PN->addIncoming(InVal, NewBB);
+      continue;
     }
 
-    // Add an incoming value to the PHI node in the loop for the preheader
-    // edge.
-    PN->addIncoming(InVal, NewBB);
+    // If the values coming into the block are not the same, we need a new
+    // PHI.
+    // Create the new PHI node, insert it into NewBB at the end of the block
+    PHINode *NewPHI =
+        PHINode::Create(PN->getType(), Preds.size(), PN->getName() + ".ph", BI);
+    if (AA)
+      AA->copyValue(PN, NewPHI);
+
+    // NOTE! This loop walks backwards for a reason! First off, this minimizes
+    // the cost of removal if we end up removing a large number of values, and
+    // second off, this ensures that the indices for the incoming values aren't
+    // invalidated when we remove one.
+    for (int64_t i = PN->getNumIncomingValues() - 1; i >= 0; --i) {
+      BasicBlock *IncomingBB = PN->getIncomingBlock(i);
+      if (PredSet.count(IncomingBB)) {
+        Value *V = PN->removeIncomingValue(i, false);
+        NewPHI->addIncoming(V, IncomingBB);
+      }
+    }
+
+    PN->addIncoming(NewPHI, NewBB);
   }
 }
 
@@ -542,7 +559,7 @@
     e = pred_end(OrigBB);
   }
 
-  BasicBlock *NewBB2 = 0;
+  BasicBlock *NewBB2 = nullptr;
   if (!NewBB2Preds.empty()) {
     // Create another basic block for the rest of OrigBB's predecessors.
     NewBB2 = BasicBlock::Create(OrigBB->getContext(),
@@ -607,7 +624,7 @@
   for (User::op_iterator i = NewRet->op_begin(), e = NewRet->op_end();
        i != e; ++i) {
     Value *V = *i;
-    Instruction *NewBC = 0;
+    Instruction *NewBC = nullptr;
     if (BitCastInst *BCI = dyn_cast<BitCastInst>(V)) {
       // Return value might be bitcasted. Clone and insert it before the
       // return instruction.
@@ -724,32 +741,32 @@
 Value *llvm::GetIfCondition(BasicBlock *BB, BasicBlock *&IfTrue,
                              BasicBlock *&IfFalse) {
   PHINode *SomePHI = dyn_cast<PHINode>(BB->begin());
-  BasicBlock *Pred1 = NULL;
-  BasicBlock *Pred2 = NULL;
+  BasicBlock *Pred1 = nullptr;
+  BasicBlock *Pred2 = nullptr;
 
   if (SomePHI) {
     if (SomePHI->getNumIncomingValues() != 2)
-      return NULL;
+      return nullptr;
     Pred1 = SomePHI->getIncomingBlock(0);
     Pred2 = SomePHI->getIncomingBlock(1);
   } else {
     pred_iterator PI = pred_begin(BB), PE = pred_end(BB);
     if (PI == PE) // No predecessor
-      return NULL;
+      return nullptr;
     Pred1 = *PI++;
     if (PI == PE) // Only one predecessor
-      return NULL;
+      return nullptr;
     Pred2 = *PI++;
     if (PI != PE) // More than two predecessors
-      return NULL;
+      return nullptr;
   }
 
   // We can only handle branches.  Other control flow will be lowered to
   // branches if possible anyway.
   BranchInst *Pred1Br = dyn_cast<BranchInst>(Pred1->getTerminator());
   BranchInst *Pred2Br = dyn_cast<BranchInst>(Pred2->getTerminator());
-  if (Pred1Br == 0 || Pred2Br == 0)
-    return 0;
+  if (!Pred1Br || !Pred2Br)
+    return nullptr;
 
   // Eliminate code duplication by ensuring that Pred1Br is conditional if
   // either are.
@@ -759,7 +776,7 @@
     // required anyway, we stand no chance of eliminating it, so the xform is
     // probably not profitable.
     if (Pred1Br->isConditional())
-      return 0;
+      return nullptr;
 
     std::swap(Pred1, Pred2);
     std::swap(Pred1Br, Pred2Br);
@@ -769,8 +786,8 @@
     // The only thing we have to watch out for here is to make sure that Pred2
     // doesn't have incoming edges from other blocks.  If it does, the condition
     // doesn't dominate BB.
-    if (Pred2->getSinglePredecessor() == 0)
-      return 0;
+    if (!Pred2->getSinglePredecessor())
+      return nullptr;
 
     // If we found a conditional branch predecessor, make sure that it branches
     // to BB and Pred2Br.  If it doesn't, this isn't an "if statement".
@@ -785,7 +802,7 @@
     } else {
       // We know that one arm of the conditional goes to BB, so the other must
       // go somewhere unrelated, and this must not be an "if statement".
-      return 0;
+      return nullptr;
     }
 
     return Pred1Br->getCondition();
@@ -795,12 +812,12 @@
   // BB.  Don't panic!  If both blocks only have a single (identical)
   // predecessor, and THAT is a conditional branch, then we're all ok!
   BasicBlock *CommonPred = Pred1->getSinglePredecessor();
-  if (CommonPred == 0 || CommonPred != Pred2->getSinglePredecessor())
-    return 0;
+  if (CommonPred == nullptr || CommonPred != Pred2->getSinglePredecessor())
+    return nullptr;
 
   // Otherwise, if this is a conditional branch, then we can use it!
   BranchInst *BI = dyn_cast<BranchInst>(CommonPred->getTerminator());
-  if (BI == 0) return 0;
+  if (!BI) return nullptr;
 
   assert(BI->isConditional() && "Two successors but not conditional?");
   if (BI->getSuccessor(0) == Pred1) {
diff --git a/lib/Transforms/Utils/BreakCriticalEdges.cpp b/lib/Transforms/Utils/BreakCriticalEdges.cpp
index 76ebb9f..80bd516 100644
--- a/lib/Transforms/Utils/BreakCriticalEdges.cpp
+++ b/lib/Transforms/Utils/BreakCriticalEdges.cpp
@@ -15,7 +15,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "break-crit-edges"
 #include "llvm/Transforms/Scalar.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/Statistic.h"
@@ -30,6 +29,8 @@
 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "break-crit-edges"
+
 STATISTIC(NumBroken, "Number of blocks inserted");
 
 namespace {
@@ -141,7 +142,7 @@
                                     Pass *P, bool MergeIdenticalEdges,
                                     bool DontDeleteUselessPhis,
                                     bool SplitLandingPads) {
-  if (!isCriticalEdge(TI, SuccNum, MergeIdenticalEdges)) return 0;
+  if (!isCriticalEdge(TI, SuccNum, MergeIdenticalEdges)) return nullptr;
 
   assert(!isa<IndirectBrInst>(TI) &&
          "Cannot split critical edge from IndirectBrInst");
@@ -151,7 +152,7 @@
 
   // Splitting the critical edge to a landing pad block is non-trivial. Don't do
   // it in this generic function.
-  if (DestBB->isLandingPad()) return 0;
+  if (DestBB->isLandingPad()) return nullptr;
 
   // Create a new basic block, linking it into the CFG.
   BasicBlock *NewBB = BasicBlock::Create(TI->getContext(),
@@ -207,15 +208,15 @@
 
 
   // If we don't have a pass object, we can't update anything...
-  if (P == 0) return NewBB;
+  if (!P) return NewBB;
 
   DominatorTreeWrapperPass *DTWP =
       P->getAnalysisIfAvailable<DominatorTreeWrapperPass>();
-  DominatorTree *DT = DTWP ? &DTWP->getDomTree() : 0;
+  DominatorTree *DT = DTWP ? &DTWP->getDomTree() : nullptr;
   LoopInfo *LI = P->getAnalysisIfAvailable<LoopInfo>();
 
   // If we have nothing to update, just return.
-  if (DT == 0 && LI == 0)
+  if (!DT && !LI)
     return NewBB;
 
   // Now update analysis information.  Since the only predecessor of NewBB is
@@ -251,7 +252,7 @@
     //
     if (TINode) {       // Don't break unreachable code!
       DomTreeNode *NewBBNode = DT->addNewBlock(NewBB, TIBB);
-      DomTreeNode *DestBBNode = 0;
+      DomTreeNode *DestBBNode = nullptr;
 
       // If NewBBDominatesDestBB hasn't been computed yet, do so with DT.
       if (!OtherPreds.empty()) {
diff --git a/lib/Transforms/Utils/BuildLibCalls.cpp b/lib/Transforms/Utils/BuildLibCalls.cpp
index 82384a1..be00b69 100644
--- a/lib/Transforms/Utils/BuildLibCalls.cpp
+++ b/lib/Transforms/Utils/BuildLibCalls.cpp
@@ -27,7 +27,8 @@
 
 /// CastToCStr - Return V if it is an i8*, otherwise cast it to i8*.
 Value *llvm::CastToCStr(Value *V, IRBuilder<> &B) {
-  return B.CreateBitCast(V, B.getInt8PtrTy(), "cstr");
+  unsigned AS = V->getType()->getPointerAddressSpace();
+  return B.CreateBitCast(V, B.getInt8PtrTy(AS), "cstr");
 }
 
 /// EmitStrLen - Emit a call to the strlen function to the builder, for the
@@ -35,7 +36,7 @@
 Value *llvm::EmitStrLen(Value *Ptr, IRBuilder<> &B, const DataLayout *TD,
                         const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::strlen))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS[2];
@@ -64,7 +65,7 @@
 Value *llvm::EmitStrNLen(Value *Ptr, Value *MaxLen, IRBuilder<> &B,
                          const DataLayout *TD, const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::strnlen))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS[2];
@@ -94,7 +95,7 @@
 Value *llvm::EmitStrChr(Value *Ptr, char C, IRBuilder<> &B,
                         const DataLayout *TD, const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::strchr))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   Attribute::AttrKind AVs[2] = { Attribute::ReadOnly, Attribute::NoUnwind };
@@ -120,7 +121,7 @@
                          IRBuilder<> &B, const DataLayout *TD,
                          const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::strncmp))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS[3];
@@ -153,7 +154,7 @@
                         const DataLayout *TD, const TargetLibraryInfo *TLI,
                         StringRef Name) {
   if (!TLI->has(LibFunc::strcpy))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS[2];
@@ -177,7 +178,7 @@
                          IRBuilder<> &B, const DataLayout *TD,
                          const TargetLibraryInfo *TLI, StringRef Name) {
   if (!TLI->has(LibFunc::strncpy))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS[2];
@@ -204,7 +205,7 @@
                            IRBuilder<> &B, const DataLayout *TD,
                            const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::memcpy_chk))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS;
@@ -232,7 +233,7 @@
                         Value *Len, IRBuilder<> &B, const DataLayout *TD,
                         const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::memchr))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS;
@@ -260,7 +261,7 @@
                         Value *Len, IRBuilder<> &B, const DataLayout *TD,
                         const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::memcmp))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS[3];
@@ -347,7 +348,7 @@
 Value *llvm::EmitPutChar(Value *Char, IRBuilder<> &B, const DataLayout *TD,
                          const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::putchar))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   Value *PutChar = M->getOrInsertFunction("putchar", B.getInt32Ty(),
@@ -369,7 +370,7 @@
 Value *llvm::EmitPutS(Value *Str, IRBuilder<> &B, const DataLayout *TD,
                       const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::puts))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS[2];
@@ -393,7 +394,7 @@
 Value *llvm::EmitFPutC(Value *Char, Value *File, IRBuilder<> &B,
                        const DataLayout *TD, const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::fputc))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS[2];
@@ -426,7 +427,7 @@
 Value *llvm::EmitFPutS(Value *Str, Value *File, IRBuilder<> &B,
                        const DataLayout *TD, const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::fputs))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS[3];
@@ -459,7 +460,7 @@
                         IRBuilder<> &B, const DataLayout *TD,
                         const TargetLibraryInfo *TLI) {
   if (!TLI->has(LibFunc::fwrite))
-    return 0;
+    return nullptr;
 
   Module *M = B.GetInsertBlock()->getParent()->getParent();
   AttributeSet AS[3];
diff --git a/lib/Transforms/Utils/BypassSlowDivision.cpp b/lib/Transforms/Utils/BypassSlowDivision.cpp
index 1f517d0..f2d5e07 100644
--- a/lib/Transforms/Utils/BypassSlowDivision.cpp
+++ b/lib/Transforms/Utils/BypassSlowDivision.cpp
@@ -15,7 +15,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "bypass-slow-division"
 #include "llvm/Transforms/Utils/BypassSlowDivision.h"
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/IR/Function.h"
@@ -24,6 +23,8 @@
 
 using namespace llvm;
 
+#define DEBUG_TYPE "bypass-slow-division"
+
 namespace {
   struct DivOpInfo {
     bool SignedOp;
@@ -53,11 +54,11 @@
     }
 
     static DivOpInfo getEmptyKey() {
-      return DivOpInfo(false, 0, 0);
+      return DivOpInfo(false, nullptr, nullptr);
     }
 
     static DivOpInfo getTombstoneKey() {
-      return DivOpInfo(true, 0, 0);
+      return DivOpInfo(true, nullptr, nullptr);
     }
 
     static unsigned getHashValue(const DivOpInfo &Val) {
diff --git a/lib/Transforms/Utils/CMakeLists.txt b/lib/Transforms/Utils/CMakeLists.txt
index dac2090..e10ca90 100644
--- a/lib/Transforms/Utils/CMakeLists.txt
+++ b/lib/Transforms/Utils/CMakeLists.txt
@@ -5,6 +5,7 @@
   BreakCriticalEdges.cpp
   BuildLibCalls.cpp
   BypassSlowDivision.cpp
+  CtorUtils.cpp
   CloneFunction.cpp
   CloneModule.cpp
   CmpInstAnalysis.cpp
diff --git a/lib/Transforms/Utils/CloneFunction.cpp b/lib/Transforms/Utils/CloneFunction.cpp
index a199086..5c8f20d 100644
--- a/lib/Transforms/Utils/CloneFunction.cpp
+++ b/lib/Transforms/Utils/CloneFunction.cpp
@@ -159,7 +159,7 @@
   for (DISubprogram Subprogram : Finder.subprograms()) {
     if (Subprogram.describes(F)) return Subprogram;
   }
-  return NULL;
+  return nullptr;
 }
 
 // Add an operand to an existing MDNode. The new operand will be added at the
@@ -359,7 +359,7 @@
       // If the condition was a known constant in the callee...
       ConstantInt *Cond = dyn_cast<ConstantInt>(BI->getCondition());
       // Or is a known constant in the caller...
-      if (Cond == 0) {
+      if (!Cond) {
         Value *V = VMap[BI->getCondition()];
         Cond = dyn_cast_or_null<ConstantInt>(V);
       }
@@ -375,7 +375,7 @@
   } else if (const SwitchInst *SI = dyn_cast<SwitchInst>(OldTI)) {
     // If switching on a value known constant in the caller.
     ConstantInt *Cond = dyn_cast<ConstantInt>(SI->getCondition());
-    if (Cond == 0) { // Or known constant after constant prop in the callee...
+    if (!Cond) { // Or known constant after constant prop in the callee...
       Value *V = VMap[SI->getCondition()];
       Cond = dyn_cast_or_null<ConstantInt>(V);
     }
@@ -454,7 +454,7 @@
        BI != BE; ++BI) {
     Value *V = VMap[BI];
     BasicBlock *NewBB = cast_or_null<BasicBlock>(V);
-    if (NewBB == 0) continue;  // Dead block.
+    if (!NewBB) continue;  // Dead block.
 
     // Add the new block to the new function.
     NewFunc->getBasicBlockList().push_back(NewBB);
diff --git a/lib/Transforms/Utils/CloneModule.cpp b/lib/Transforms/Utils/CloneModule.cpp
index 64df089..eb67db1 100644
--- a/lib/Transforms/Utils/CloneModule.cpp
+++ b/lib/Transforms/Utils/CloneModule.cpp
@@ -47,8 +47,8 @@
     GlobalVariable *GV = new GlobalVariable(*New, 
                                             I->getType()->getElementType(),
                                             I->isConstant(), I->getLinkage(),
-                                            (Constant*) 0, I->getName(),
-                                            (GlobalVariable*) 0,
+                                            (Constant*) nullptr, I->getName(),
+                                            (GlobalVariable*) nullptr,
                                             I->getThreadLocalMode(),
                                             I->getType()->getAddressSpace());
     GV->copyAttributesFrom(I);
@@ -67,8 +67,10 @@
   // Loop over the aliases in the module
   for (Module::const_alias_iterator I = M->alias_begin(), E = M->alias_end();
        I != E; ++I) {
-    GlobalAlias *GA = new GlobalAlias(I->getType(), I->getLinkage(),
-                                      I->getName(), NULL, New);
+    auto *PTy = cast<PointerType>(I->getType());
+    auto *GA =
+        GlobalAlias::create(PTy->getElementType(), PTy->getAddressSpace(),
+                            I->getLinkage(), I->getName(), New);
     GA->copyAttributesFrom(I);
     VMap[I] = GA;
   }
@@ -105,8 +107,8 @@
   for (Module::const_alias_iterator I = M->alias_begin(), E = M->alias_end();
        I != E; ++I) {
     GlobalAlias *GA = cast<GlobalAlias>(VMap[I]);
-    if (const Constant *C = I->getAliasee())
-      GA->setAliasee(MapValue(C, VMap));
+    if (const GlobalObject *C = I->getAliasee())
+      GA->setAliasee(cast<GlobalObject>(MapValue(C, VMap)));
   }
 
   // And named metadata....
diff --git a/lib/Transforms/Utils/CmpInstAnalysis.cpp b/lib/Transforms/Utils/CmpInstAnalysis.cpp
index 8fa412a..3b15a0a 100644
--- a/lib/Transforms/Utils/CmpInstAnalysis.cpp
+++ b/lib/Transforms/Utils/CmpInstAnalysis.cpp
@@ -84,7 +84,7 @@
     case 7: // True.
       return ConstantInt::get(CmpInst::makeCmpResultType(LHS->getType()), 1);
   }
-  return NULL;
+  return nullptr;
 }
 
 /// PredicatesFoldable - Return true if both predicates match sign or if at
diff --git a/lib/Transforms/Utils/CodeExtractor.cpp b/lib/Transforms/Utils/CodeExtractor.cpp
index b814842..e70a7d6 100644
--- a/lib/Transforms/Utils/CodeExtractor.cpp
+++ b/lib/Transforms/Utils/CodeExtractor.cpp
@@ -38,6 +38,8 @@
 #include <set>
 using namespace llvm;
 
+#define DEBUG_TYPE "code-extractor"
+
 // Provide a command-line option to aggregate function arguments into a struct
 // for functions produced by the code extractor. This is useful when converting
 // extracted functions to pthread-based code, as only one argument (void*) can
@@ -118,7 +120,7 @@
 }
 
 CodeExtractor::CodeExtractor(BasicBlock *BB, bool AggregateArgs)
-  : DT(0), AggregateArgs(AggregateArgs||AggregateArgsOpt),
+  : DT(nullptr), AggregateArgs(AggregateArgs||AggregateArgsOpt),
     Blocks(buildExtractionBlockSet(BB)), NumExitBlocks(~0U) {}
 
 CodeExtractor::CodeExtractor(ArrayRef<BasicBlock *> BBs, DominatorTree *DT,
@@ -410,7 +412,7 @@
        return P->getIncomingBlock(U);
   }
 
-  return 0;
+  return nullptr;
 }
 
 /// emitCallAndSwitchStatement - This method sets up the caller side by adding
@@ -438,14 +440,14 @@
       StructValues.push_back(*i);
     } else {
       AllocaInst *alloca =
-        new AllocaInst((*i)->getType(), 0, (*i)->getName()+".loc",
+        new AllocaInst((*i)->getType(), nullptr, (*i)->getName()+".loc",
                        codeReplacer->getParent()->begin()->begin());
       ReloadOutputs.push_back(alloca);
       params.push_back(alloca);
     }
   }
 
-  AllocaInst *Struct = 0;
+  AllocaInst *Struct = nullptr;
   if (AggregateArgs && (inputs.size() + outputs.size() > 0)) {
     std::vector<Type*> ArgTypes;
     for (ValueSet::iterator v = StructValues.begin(),
@@ -455,7 +457,7 @@
     // Allocate a struct at the beginning of this function
     Type *StructArgTy = StructType::get(newFunction->getContext(), ArgTypes);
     Struct =
-      new AllocaInst(StructArgTy, 0, "structArg",
+      new AllocaInst(StructArgTy, nullptr, "structArg",
                      codeReplacer->getParent()->begin()->begin());
     params.push_back(Struct);
 
@@ -484,7 +486,7 @@
 
   // Reload the outputs passed in by reference
   for (unsigned i = 0, e = outputs.size(); i != e; ++i) {
-    Value *Output = 0;
+    Value *Output = nullptr;
     if (AggregateArgs) {
       Value *Idx[2];
       Idx[0] = Constant::getNullValue(Type::getInt32Ty(Context));
@@ -537,7 +539,7 @@
                                          newFunction);
           unsigned SuccNum = switchVal++;
 
-          Value *brVal = 0;
+          Value *brVal = nullptr;
           switch (NumExitBlocks) {
           case 0:
           case 1: break;  // No value needed.
@@ -633,7 +635,7 @@
 
     // Check if the function should return a value
     if (OldFnRetTy->isVoidTy()) {
-      ReturnInst::Create(Context, 0, TheSwitch);  // Return void
+      ReturnInst::Create(Context, nullptr, TheSwitch);  // Return void
     } else if (OldFnRetTy == TheSwitch->getCondition()->getType()) {
       // return what we have
       ReturnInst::Create(Context, TheSwitch->getCondition(), TheSwitch);
@@ -685,7 +687,7 @@
 
 Function *CodeExtractor::extractCodeRegion() {
   if (!isEligible())
-    return 0;
+    return nullptr;
 
   ValueSet inputs, outputs;
 
diff --git a/lib/Transforms/Utils/CtorUtils.cpp b/lib/Transforms/Utils/CtorUtils.cpp
new file mode 100644
index 0000000..a359424
--- /dev/null
+++ b/lib/Transforms/Utils/CtorUtils.cpp
@@ -0,0 +1,183 @@
+//===- CtorUtils.cpp - Helpers for working with global_ctors ----*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file defines functions that are used to process llvm.global_ctors.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Transforms/Utils/CtorUtils.h"
+#include "llvm/IR/Constants.h"
+#include "llvm/IR/Function.h"
+#include "llvm/IR/GlobalVariable.h"
+#include "llvm/IR/Instructions.h"
+#include "llvm/IR/Module.h"
+#include "llvm/Support/Debug.h"
+
+#define DEBUG_TYPE "ctor_utils"
+
+namespace llvm {
+
+namespace {
+/// Given a specified llvm.global_ctors list, install the
+/// specified array.
+void installGlobalCtors(GlobalVariable *GCL,
+                        const std::vector<Function *> &Ctors) {
+  // If we made a change, reassemble the initializer list.
+  Constant *CSVals[3];
+
+  StructType *StructTy =
+      cast<StructType>(GCL->getType()->getElementType()->getArrayElementType());
+
+  // Create the new init list.
+  std::vector<Constant *> CAList;
+  for (Function *F : Ctors) {
+    Type *Int32Ty = Type::getInt32Ty(GCL->getContext());
+    if (F) {
+      CSVals[0] = ConstantInt::get(Int32Ty, 65535);
+      CSVals[1] = F;
+    } else {
+      CSVals[0] = ConstantInt::get(Int32Ty, 0x7fffffff);
+      CSVals[1] = Constant::getNullValue(StructTy->getElementType(1));
+    }
+    // FIXME: Only allow the 3-field form in LLVM 4.0.
+    size_t NumElts = StructTy->getNumElements();
+    if (NumElts > 2)
+      CSVals[2] = Constant::getNullValue(StructTy->getElementType(2));
+    CAList.push_back(
+        ConstantStruct::get(StructTy, makeArrayRef(CSVals, NumElts)));
+  }
+
+  // Create the array initializer.
+  Constant *CA =
+      ConstantArray::get(ArrayType::get(StructTy, CAList.size()), CAList);
+
+  // If we didn't change the number of elements, don't create a new GV.
+  if (CA->getType() == GCL->getInitializer()->getType()) {
+    GCL->setInitializer(CA);
+    return;
+  }
+
+  // Create the new global and insert it next to the existing list.
+  GlobalVariable *NGV =
+      new GlobalVariable(CA->getType(), GCL->isConstant(), GCL->getLinkage(),
+                         CA, "", GCL->getThreadLocalMode());
+  GCL->getParent()->getGlobalList().insert(GCL, NGV);
+  NGV->takeName(GCL);
+
+  // Nuke the old list, replacing any uses with the new one.
+  if (!GCL->use_empty()) {
+    Constant *V = NGV;
+    if (V->getType() != GCL->getType())
+      V = ConstantExpr::getBitCast(V, GCL->getType());
+    GCL->replaceAllUsesWith(V);
+  }
+  GCL->eraseFromParent();
+}
+
+/// Given a llvm.global_ctors list that we can understand,
+/// return a list of the functions and null terminator as a vector.
+std::vector<Function*> parseGlobalCtors(GlobalVariable *GV) {
+  if (GV->getInitializer()->isNullValue())
+    return std::vector<Function *>();
+  ConstantArray *CA = cast<ConstantArray>(GV->getInitializer());
+  std::vector<Function *> Result;
+  Result.reserve(CA->getNumOperands());
+  for (User::op_iterator i = CA->op_begin(), e = CA->op_end(); i != e; ++i) {
+    ConstantStruct *CS = cast<ConstantStruct>(*i);
+    Result.push_back(dyn_cast<Function>(CS->getOperand(1)));
+  }
+  return Result;
+}
+
+/// Find the llvm.global_ctors list, verifying that all initializers have an
+/// init priority of 65535.
+GlobalVariable *findGlobalCtors(Module &M) {
+  GlobalVariable *GV = M.getGlobalVariable("llvm.global_ctors");
+  if (!GV)
+    return nullptr;
+
+  // Verify that the initializer is simple enough for us to handle. We are
+  // only allowed to optimize the initializer if it is unique.
+  if (!GV->hasUniqueInitializer())
+    return nullptr;
+
+  if (isa<ConstantAggregateZero>(GV->getInitializer()))
+    return GV;
+  ConstantArray *CA = cast<ConstantArray>(GV->getInitializer());
+
+  for (User::op_iterator i = CA->op_begin(), e = CA->op_end(); i != e; ++i) {
+    if (isa<ConstantAggregateZero>(*i))
+      continue;
+    ConstantStruct *CS = cast<ConstantStruct>(*i);
+    if (isa<ConstantPointerNull>(CS->getOperand(1)))
+      continue;
+
+    // Must have a function or null ptr.
+    if (!isa<Function>(CS->getOperand(1)))
+      return nullptr;
+
+    // Init priority must be standard.
+    ConstantInt *CI = cast<ConstantInt>(CS->getOperand(0));
+    if (CI->getZExtValue() != 65535)
+      return nullptr;
+  }
+
+  return GV;
+}
+} // namespace
+
+/// Call "ShouldRemove" for every entry in M's global_ctor list and remove the
+/// entries for which it returns true.  Return true if anything changed.
+bool optimizeGlobalCtorsList(Module &M,
+                             function_ref<bool(Function *)> ShouldRemove) {
+  GlobalVariable *GlobalCtors = findGlobalCtors(M);
+  if (!GlobalCtors)
+    return false;
+
+  std::vector<Function *> Ctors = parseGlobalCtors(GlobalCtors);
+  if (Ctors.empty())
+    return false;
+
+  bool MadeChange = false;
+
+  // Loop over global ctors, optimizing them when we can.
+  for (unsigned i = 0; i != Ctors.size(); ++i) {
+    Function *F = Ctors[i];
+    // Found a null terminator in the middle of the list, prune off the rest of
+    // the list.
+    if (!F) {
+      if (i != Ctors.size() - 1) {
+        Ctors.resize(i + 1);
+        MadeChange = true;
+      }
+      break;
+    }
+    DEBUG(dbgs() << "Optimizing Global Constructor: " << *F << "\n");
+
+    // We cannot simplify external ctor functions.
+    if (F->empty())
+      continue;
+
+    // If we can evaluate the ctor at compile time, do.
+    if (ShouldRemove(F)) {
+      Ctors.erase(Ctors.begin() + i);
+      MadeChange = true;
+      --i;
+      continue;
+    }
+  }
+
+  if (!MadeChange)
+    return false;
+
+  installGlobalCtors(GlobalCtors, Ctors);
+  return true;
+}
+
+} // End llvm namespace
diff --git a/lib/Transforms/Utils/DemoteRegToStack.cpp b/lib/Transforms/Utils/DemoteRegToStack.cpp
index ac6926f..9972b22 100644
--- a/lib/Transforms/Utils/DemoteRegToStack.cpp
+++ b/lib/Transforms/Utils/DemoteRegToStack.cpp
@@ -25,17 +25,17 @@
                                    Instruction *AllocaPoint) {
   if (I.use_empty()) {
     I.eraseFromParent();
-    return 0;
+    return nullptr;
   }
 
   // Create a stack slot to hold the value.
   AllocaInst *Slot;
   if (AllocaPoint) {
-    Slot = new AllocaInst(I.getType(), 0,
+    Slot = new AllocaInst(I.getType(), nullptr,
                           I.getName()+".reg2mem", AllocaPoint);
   } else {
     Function *F = I.getParent()->getParent();
-    Slot = new AllocaInst(I.getType(), 0, I.getName()+".reg2mem",
+    Slot = new AllocaInst(I.getType(), nullptr, I.getName()+".reg2mem",
                           F->getEntryBlock().begin());
   }
 
@@ -56,7 +56,7 @@
       for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i)
         if (PN->getIncomingValue(i) == &I) {
           Value *&V = Loads[PN->getIncomingBlock(i)];
-          if (V == 0) {
+          if (!V) {
             // Insert the load into the predecessor block
             V = new LoadInst(Slot, I.getName()+".reload", VolatileLoads,
                              PN->getIncomingBlock(i)->getTerminator());
@@ -110,17 +110,17 @@
 AllocaInst *llvm::DemotePHIToStack(PHINode *P, Instruction *AllocaPoint) {
   if (P->use_empty()) {
     P->eraseFromParent();
-    return 0;
+    return nullptr;
   }
 
   // Create a stack slot to hold the value.
   AllocaInst *Slot;
   if (AllocaPoint) {
-    Slot = new AllocaInst(P->getType(), 0,
+    Slot = new AllocaInst(P->getType(), nullptr,
                           P->getName()+".reg2mem", AllocaPoint);
   } else {
     Function *F = P->getParent()->getParent();
-    Slot = new AllocaInst(P->getType(), 0, P->getName()+".reg2mem",
+    Slot = new AllocaInst(P->getType(), nullptr, P->getName()+".reg2mem",
                           F->getEntryBlock().begin());
   }
 
diff --git a/lib/Transforms/Utils/FlattenCFG.cpp b/lib/Transforms/Utils/FlattenCFG.cpp
index 39c80f8..51ead40 100644
--- a/lib/Transforms/Utils/FlattenCFG.cpp
+++ b/lib/Transforms/Utils/FlattenCFG.cpp
@@ -11,7 +11,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "flattencfg"
 #include "llvm/Transforms/Utils/Local.h"
 #include "llvm/ADT/SmallPtrSet.h"
 #include "llvm/Analysis/AliasAnalysis.h"
@@ -22,16 +21,19 @@
 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "flattencfg"
+
 namespace {
 class FlattenCFGOpt {
   AliasAnalysis *AA;
   /// \brief Use parallel-and or parallel-or to generate conditions for
   /// conditional branches.
-  bool FlattenParallelAndOr(BasicBlock *BB, IRBuilder<> &Builder, Pass *P = 0);
+  bool FlattenParallelAndOr(BasicBlock *BB, IRBuilder<> &Builder,
+                            Pass *P = nullptr);
   /// \brief If \param BB is the merge block of an if-region, attempt to merge
   /// the if-region with an adjacent if-region upstream if two if-regions
   /// contain identical instructions.
-  bool MergeIfRegion(BasicBlock *BB, IRBuilder<> &Builder, Pass *P = 0);
+  bool MergeIfRegion(BasicBlock *BB, IRBuilder<> &Builder, Pass *P = nullptr);
   /// \brief Compare a pair of blocks: \p Block1 and \p Block2, which
   /// are from two if-regions whose entry blocks are \p Head1 and \p
   /// Head2.  \returns true if \p Block1 and \p Block2 contain identical
@@ -126,9 +128,9 @@
   if (PHI)
     return false; // For simplicity, avoid cases containing PHI nodes.
 
-  BasicBlock *LastCondBlock = NULL;
-  BasicBlock *FirstCondBlock = NULL;
-  BasicBlock *UnCondBlock = NULL;
+  BasicBlock *LastCondBlock = nullptr;
+  BasicBlock *FirstCondBlock = nullptr;
+  BasicBlock *UnCondBlock = nullptr;
   int Idx = -1;
 
   // Check predecessors of \param BB.
diff --git a/lib/Transforms/Utils/GlobalStatus.cpp b/lib/Transforms/Utils/GlobalStatus.cpp
index e9ebc45..12057e4 100644
--- a/lib/Transforms/Utils/GlobalStatus.cpp
+++ b/lib/Transforms/Utils/GlobalStatus.cpp
@@ -61,7 +61,7 @@
     } else if (const Instruction *I = dyn_cast<Instruction>(UR)) {
       if (!GS.HasMultipleAccessingFunctions) {
         const Function *F = I->getParent()->getParent();
-        if (GS.AccessingFunction == 0)
+        if (!GS.AccessingFunction)
           GS.AccessingFunction = F;
         else if (GS.AccessingFunction != F)
           GS.HasMultipleAccessingFunctions = true;
@@ -176,6 +176,6 @@
 
 GlobalStatus::GlobalStatus()
     : IsCompared(false), IsLoaded(false), StoredType(NotStored),
-      StoredOnceValue(0), AccessingFunction(0),
+      StoredOnceValue(nullptr), AccessingFunction(nullptr),
       HasMultipleAccessingFunctions(false), HasNonInstructionUser(false),
       Ordering(NotAtomic) {}
diff --git a/lib/Transforms/Utils/InlineFunction.cpp b/lib/Transforms/Utils/InlineFunction.cpp
index 86def3e..e01d0c3 100644
--- a/lib/Transforms/Utils/InlineFunction.cpp
+++ b/lib/Transforms/Utils/InlineFunction.cpp
@@ -19,6 +19,7 @@
 #include "llvm/Analysis/InstructionSimplify.h"
 #include "llvm/IR/Attributes.h"
 #include "llvm/IR/CallSite.h"
+#include "llvm/IR/CFG.h"
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/DataLayout.h"
 #include "llvm/IR/DebugInfo.h"
@@ -51,8 +52,8 @@
 
   public:
     InvokeInliningInfo(InvokeInst *II)
-      : OuterResumeDest(II->getUnwindDest()), InnerResumeDest(0),
-        CallerLPad(0), InnerEHValuesPHI(0) {
+      : OuterResumeDest(II->getUnwindDest()), InnerResumeDest(nullptr),
+        CallerLPad(nullptr), InnerEHValuesPHI(nullptr) {
       // If there are PHI nodes in the unwind destination block, we need to keep
       // track of which values came into them from the invoke before removing
       // the edge from this block.
@@ -289,13 +290,13 @@
 
     ValueToValueMapTy::iterator VMI = VMap.find(OrigCall);
     // Only copy the edge if the call was inlined!
-    if (VMI == VMap.end() || VMI->second == 0)
+    if (VMI == VMap.end() || VMI->second == nullptr)
       continue;
     
     // If the call was inlined, but then constant folded, there is no edge to
     // add.  Check for this case.
     Instruction *NewCall = dyn_cast<Instruction>(VMI->second);
-    if (NewCall == 0) continue;
+    if (!NewCall) continue;
 
     // Remember that this call site got inlined for the client of
     // InlineFunction.
@@ -306,7 +307,7 @@
     // happens, set the callee of the new call site to a more precise
     // destination.  This can also happen if the call graph node of the caller
     // was just unnecessarily imprecise.
-    if (I->second->getFunction() == 0)
+    if (!I->second->getFunction())
       if (Function *F = CallSite(NewCall).getCalledFunction()) {
         // Indirect call site resolved to direct call.
         CallerNode->addCalledFunction(CallSite(NewCall), CG[F]);
@@ -322,13 +323,44 @@
   CallerNode->removeCallEdgeFor(CS);
 }
 
+static void HandleByValArgumentInit(Value *Dst, Value *Src, Module *M,
+                                    BasicBlock *InsertBlock,
+                                    InlineFunctionInfo &IFI) {
+  LLVMContext &Context = Src->getContext();
+  Type *VoidPtrTy = Type::getInt8PtrTy(Context);
+  Type *AggTy = cast<PointerType>(Src->getType())->getElementType();
+  Type *Tys[3] = { VoidPtrTy, VoidPtrTy, Type::getInt64Ty(Context) };
+  Function *MemCpyFn = Intrinsic::getDeclaration(M, Intrinsic::memcpy, Tys);
+  IRBuilder<> builder(InsertBlock->begin());
+  Value *DstCast = builder.CreateBitCast(Dst, VoidPtrTy, "tmp");
+  Value *SrcCast = builder.CreateBitCast(Src, VoidPtrTy, "tmp");
+
+  Value *Size;
+  if (IFI.DL == nullptr)
+    Size = ConstantExpr::getSizeOf(AggTy);
+  else
+    Size = ConstantInt::get(Type::getInt64Ty(Context),
+                            IFI.DL->getTypeStoreSize(AggTy));
+
+  // Always generate a memcpy of alignment 1 here because we don't know
+  // the alignment of the src pointer.  Other optimizations can infer
+  // better alignment.
+  Value *CallArgs[] = {
+    DstCast, SrcCast, Size,
+    ConstantInt::get(Type::getInt32Ty(Context), 1),
+    ConstantInt::getFalse(Context) // isVolatile
+  };
+  builder.CreateCall(MemCpyFn, CallArgs);
+}
+
 /// HandleByValArgument - When inlining a call site that has a byval argument,
 /// we have to make the implicit memcpy explicit by adding it.
 static Value *HandleByValArgument(Value *Arg, Instruction *TheCall,
                                   const Function *CalledFunc,
                                   InlineFunctionInfo &IFI,
                                   unsigned ByValAlignment) {
-  Type *AggTy = cast<PointerType>(Arg->getType())->getElementType();
+  PointerType *ArgTy = cast<PointerType>(Arg->getType());
+  Type *AggTy = ArgTy->getElementType();
 
   // If the called function is readonly, then it could not mutate the caller's
   // copy of the byval'd memory.  In this case, it is safe to elide the copy and
@@ -349,11 +381,7 @@
     // Otherwise, we have to make a memcpy to get a safe alignment.  This is bad
     // for code quality, but rarely happens and is required for correctness.
   }
-  
-  LLVMContext &Context = Arg->getContext();
 
-  Type *VoidPtrTy = Type::getInt8PtrTy(Context);
-  
   // Create the alloca.  If we have DataLayout, use nice alignment.
   unsigned Align = 1;
   if (IFI.DL)
@@ -366,32 +394,9 @@
   
   Function *Caller = TheCall->getParent()->getParent(); 
   
-  Value *NewAlloca = new AllocaInst(AggTy, 0, Align, Arg->getName(), 
+  Value *NewAlloca = new AllocaInst(AggTy, nullptr, Align, Arg->getName(), 
                                     &*Caller->begin()->begin());
-  // Emit a memcpy.
-  Type *Tys[3] = {VoidPtrTy, VoidPtrTy, Type::getInt64Ty(Context)};
-  Function *MemCpyFn = Intrinsic::getDeclaration(Caller->getParent(),
-                                                 Intrinsic::memcpy, 
-                                                 Tys);
-  Value *DestCast = new BitCastInst(NewAlloca, VoidPtrTy, "tmp", TheCall);
-  Value *SrcCast = new BitCastInst(Arg, VoidPtrTy, "tmp", TheCall);
-  
-  Value *Size;
-  if (IFI.DL == 0)
-    Size = ConstantExpr::getSizeOf(AggTy);
-  else
-    Size = ConstantInt::get(Type::getInt64Ty(Context),
-                            IFI.DL->getTypeStoreSize(AggTy));
-  
-  // Always generate a memcpy of alignment 1 here because we don't know
-  // the alignment of the src pointer.  Other optimizations can infer
-  // better alignment.
-  Value *CallArgs[] = {
-    DestCast, SrcCast, Size,
-    ConstantInt::get(Type::getInt32Ty(Context), 1),
-    ConstantInt::getFalse(Context) // isVolatile
-  };
-  IRBuilder<>(TheCall).CreateCall(MemCpyFn, CallArgs);
+  IFI.StaticAllocas.push_back(cast<AllocaInst>(NewAlloca));
   
   // Uses of the argument in the function should use our new alloca
   // instead.
@@ -417,8 +422,10 @@
 // hasLifetimeMarkers - Check whether the given alloca already has
 // lifetime.start or lifetime.end intrinsics.
 static bool hasLifetimeMarkers(AllocaInst *AI) {
-  Type *Int8PtrTy = Type::getInt8PtrTy(AI->getType()->getContext());
-  if (AI->getType() == Int8PtrTy)
+  Type *Ty = AI->getType();
+  Type *Int8PtrTy = Type::getInt8PtrTy(Ty->getContext(),
+                                       Ty->getPointerAddressSpace());
+  if (Ty == Int8PtrTy)
     return isUsedByLifetimeMarker(AI);
 
   // Do a scan to find all the casts to i8*.
@@ -472,6 +479,33 @@
   }
 }
 
+/// Returns a musttail call instruction if one immediately precedes the given
+/// return instruction with an optional bitcast instruction between them.
+static CallInst *getPrecedingMustTailCall(ReturnInst *RI) {
+  Instruction *Prev = RI->getPrevNode();
+  if (!Prev)
+    return nullptr;
+
+  if (Value *RV = RI->getReturnValue()) {
+    if (RV != Prev)
+      return nullptr;
+
+    // Look through the optional bitcast.
+    if (auto *BI = dyn_cast<BitCastInst>(Prev)) {
+      RV = BI->getOperand(0);
+      Prev = BI->getPrevNode();
+      if (!Prev || RV != Prev)
+        return nullptr;
+    }
+  }
+
+  if (auto *CI = dyn_cast<CallInst>(Prev)) {
+    if (CI->isMustTailCall())
+      return CI;
+  }
+  return nullptr;
+}
+
 /// InlineFunction - This function inlines the called function into the basic
 /// block of the caller.  This returns false if it is not possible to inline
 /// this call.  The program is still in a well defined state if this occurs
@@ -491,15 +525,10 @@
   IFI.reset();
   
   const Function *CalledFunc = CS.getCalledFunction();
-  if (CalledFunc == 0 ||          // Can't inline external function or indirect
+  if (!CalledFunc ||              // Can't inline external function or indirect
       CalledFunc->isDeclaration() || // call, or call to a vararg function!
       CalledFunc->getFunctionType()->isVarArg()) return false;
 
-  // If the call to the callee is not a tail call, we must clear the 'tail'
-  // flags on any calls that we inline.
-  bool MustClearTailCallFlags =
-    !(isa<CallInst>(TheCall) && cast<CallInst>(TheCall)->isTailCall());
-
   // If the call to the callee cannot throw, set the 'nounwind' flag on any
   // calls that we inline.
   bool MarkNoUnwind = CS.doesNotThrow();
@@ -519,7 +548,7 @@
   }
 
   // Get the personality function from the callee if it contains a landing pad.
-  Value *CalleePersonality = 0;
+  Value *CalleePersonality = nullptr;
   for (Function::const_iterator I = CalledFunc->begin(), E = CalledFunc->end();
        I != E; ++I)
     if (const InvokeInst *II = dyn_cast<InvokeInst>(I->getTerminator())) {
@@ -562,6 +591,8 @@
 
   { // Scope to destroy VMap after cloning.
     ValueToValueMapTy VMap;
+    // Keep a list of pair (dst, src) to emit byval initializations.
+    SmallVector<std::pair<Value*, Value*>, 4> ByValInit;
 
     assert(CalledFunc->arg_size() == CS.arg_size() &&
            "No varargs calls can be inlined!");
@@ -581,11 +612,8 @@
       if (CS.isByValArgument(ArgNo)) {
         ActualArg = HandleByValArgument(ActualArg, TheCall, CalledFunc, IFI,
                                         CalledFunc->getParamAlignment(ArgNo+1));
- 
-        // Calls that we inline may use the new alloca, so we need to clear
-        // their 'tail' flags if HandleByValArgument introduced a new alloca and
-        // the callee has calls.
-        MustClearTailCallFlags |= ActualArg != *AI;
+        if (ActualArg != *AI)
+          ByValInit.push_back(std::make_pair(ActualArg, (Value*) *AI));
       }
 
       VMap[I] = ActualArg;
@@ -602,6 +630,11 @@
     // Remember the first block that is newly cloned over.
     FirstNewBlock = LastBlock; ++FirstNewBlock;
 
+    // Inject byval arguments initialization.
+    for (std::pair<Value*, Value*> &Init : ByValInit)
+      HandleByValArgumentInit(Init.first, Init.second, Caller->getParent(),
+                              FirstNewBlock, IFI);
+
     // Update the callgraph if requested.
     if (IFI.CG)
       UpdateCallGraphAfterInlining(CS, FirstNewBlock, VMap, IFI);
@@ -619,7 +652,7 @@
     for (BasicBlock::iterator I = FirstNewBlock->begin(),
          E = FirstNewBlock->end(); I != E; ) {
       AllocaInst *AI = dyn_cast<AllocaInst>(I++);
-      if (AI == 0) continue;
+      if (!AI) continue;
       
       // If the alloca is now dead, remove it.  This often occurs due to code
       // specialization.
@@ -651,6 +684,45 @@
     }
   }
 
+  bool InlinedMustTailCalls = false;
+  if (InlinedFunctionInfo.ContainsCalls) {
+    CallInst::TailCallKind CallSiteTailKind = CallInst::TCK_None;
+    if (CallInst *CI = dyn_cast<CallInst>(TheCall))
+      CallSiteTailKind = CI->getTailCallKind();
+
+    for (Function::iterator BB = FirstNewBlock, E = Caller->end(); BB != E;
+         ++BB) {
+      for (Instruction &I : *BB) {
+        CallInst *CI = dyn_cast<CallInst>(&I);
+        if (!CI)
+          continue;
+
+        // We need to reduce the strength of any inlined tail calls.  For
+        // musttail, we have to avoid introducing potential unbounded stack
+        // growth.  For example, if functions 'f' and 'g' are mutually recursive
+        // with musttail, we can inline 'g' into 'f' so long as we preserve
+        // musttail on the cloned call to 'f'.  If either the inlined call site
+        // or the cloned call site is *not* musttail, the program already has
+        // one frame of stack growth, so it's safe to remove musttail.  Here is
+        // a table of example transformations:
+        //
+        //    f -> musttail g -> musttail f  ==>  f -> musttail f
+        //    f -> musttail g ->     tail f  ==>  f ->     tail f
+        //    f ->          g -> musttail f  ==>  f ->          f
+        //    f ->          g ->     tail f  ==>  f ->          f
+        CallInst::TailCallKind ChildTCK = CI->getTailCallKind();
+        ChildTCK = std::min(CallSiteTailKind, ChildTCK);
+        CI->setTailCallKind(ChildTCK);
+        InlinedMustTailCalls |= CI->isMustTailCall();
+
+        // Calls inlined through a 'nounwind' call site should be marked
+        // 'nounwind'.
+        if (MarkNoUnwind)
+          CI->setDoesNotThrow();
+      }
+    }
+  }
+
   // Leave lifetime markers for the static alloca's, scoping them to the
   // function we just inlined.
   if (InsertLifetime && !IFI.StaticAllocas.empty()) {
@@ -664,7 +736,7 @@
         continue;
 
       // Try to determine the size of the allocation.
-      ConstantInt *AllocaSize = 0;
+      ConstantInt *AllocaSize = nullptr;
       if (ConstantInt *AIArraySize =
           dyn_cast<ConstantInt>(AI->getArraySize())) {
         if (IFI.DL) {
@@ -683,9 +755,12 @@
       }
 
       builder.CreateLifetimeStart(AI, AllocaSize);
-      for (unsigned ri = 0, re = Returns.size(); ri != re; ++ri) {
-        IRBuilder<> builder(Returns[ri]);
-        builder.CreateLifetimeEnd(AI, AllocaSize);
+      for (ReturnInst *RI : Returns) {
+        // Don't insert llvm.lifetime.end calls between a musttail call and a
+        // return.  The return kills all local allocas.
+        if (InlinedMustTailCalls && getPrecedingMustTailCall(RI))
+          continue;
+        IRBuilder<>(RI).CreateLifetimeEnd(AI, AllocaSize);
       }
     }
   }
@@ -704,33 +779,56 @@
 
     // Insert a call to llvm.stackrestore before any return instructions in the
     // inlined function.
-    for (unsigned i = 0, e = Returns.size(); i != e; ++i) {
-      IRBuilder<>(Returns[i]).CreateCall(StackRestore, SavedPtr);
+    for (ReturnInst *RI : Returns) {
+      // Don't insert llvm.stackrestore calls between a musttail call and a
+      // return.  The return will restore the stack pointer.
+      if (InlinedMustTailCalls && getPrecedingMustTailCall(RI))
+        continue;
+      IRBuilder<>(RI).CreateCall(StackRestore, SavedPtr);
     }
   }
 
-  // If we are inlining tail call instruction through a call site that isn't
-  // marked 'tail', we must remove the tail marker for any calls in the inlined
-  // code.  Also, calls inlined through a 'nounwind' call site should be marked
-  // 'nounwind'.
-  if (InlinedFunctionInfo.ContainsCalls &&
-      (MustClearTailCallFlags || MarkNoUnwind)) {
-    for (Function::iterator BB = FirstNewBlock, E = Caller->end();
-         BB != E; ++BB)
-      for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E; ++I)
-        if (CallInst *CI = dyn_cast<CallInst>(I)) {
-          if (MustClearTailCallFlags)
-            CI->setTailCall(false);
-          if (MarkNoUnwind)
-            CI->setDoesNotThrow();
-        }
-  }
-
   // If we are inlining for an invoke instruction, we must make sure to rewrite
   // any call instructions into invoke instructions.
   if (InvokeInst *II = dyn_cast<InvokeInst>(TheCall))
     HandleInlinedInvoke(II, FirstNewBlock, InlinedFunctionInfo);
 
+  // Handle any inlined musttail call sites.  In order for a new call site to be
+  // musttail, the source of the clone and the inlined call site must have been
+  // musttail.  Therefore it's safe to return without merging control into the
+  // phi below.
+  if (InlinedMustTailCalls) {
+    // Check if we need to bitcast the result of any musttail calls.
+    Type *NewRetTy = Caller->getReturnType();
+    bool NeedBitCast = !TheCall->use_empty() && TheCall->getType() != NewRetTy;
+
+    // Handle the returns preceded by musttail calls separately.
+    SmallVector<ReturnInst *, 8> NormalReturns;
+    for (ReturnInst *RI : Returns) {
+      CallInst *ReturnedMustTail = getPrecedingMustTailCall(RI);
+      if (!ReturnedMustTail) {
+        NormalReturns.push_back(RI);
+        continue;
+      }
+      if (!NeedBitCast)
+        continue;
+
+      // Delete the old return and any preceding bitcast.
+      BasicBlock *CurBB = RI->getParent();
+      auto *OldCast = dyn_cast_or_null<BitCastInst>(RI->getReturnValue());
+      RI->eraseFromParent();
+      if (OldCast)
+        OldCast->eraseFromParent();
+
+      // Insert a new bitcast and return with the right type.
+      IRBuilder<> Builder(CurBB);
+      Builder.CreateRet(Builder.CreateBitCast(ReturnedMustTail, NewRetTy));
+    }
+
+    // Leave behind the normal returns so we can merge control flow.
+    std::swap(Returns, NormalReturns);
+  }
+
   // If we cloned in _exactly one_ basic block, and if that block ends in a
   // return instruction, we splice the body of the inlined callee directly into
   // the calling basic block.
@@ -774,7 +872,7 @@
   // "starter" and "ender" blocks.  How we accomplish this depends on whether
   // this is an invoke instruction or a call instruction.
   BasicBlock *AfterCallBB;
-  BranchInst *CreatedBranchToNormalDest = NULL;
+  BranchInst *CreatedBranchToNormalDest = nullptr;
   if (InvokeInst *II = dyn_cast<InvokeInst>(TheCall)) {
 
     // Add an unconditional branch to make this look like the CallInst case...
@@ -813,7 +911,7 @@
   // any users of the original call/invoke instruction.
   Type *RTy = CalledFunc->getReturnType();
 
-  PHINode *PHI = 0;
+  PHINode *PHI = nullptr;
   if (Returns.size() > 1) {
     // The PHI node should go at the front of the new basic block to merge all
     // possible incoming values.
@@ -886,6 +984,11 @@
   // Since we are now done with the Call/Invoke, we can delete it.
   TheCall->eraseFromParent();
 
+  // If we inlined any musttail calls and the original return is now
+  // unreachable, delete it.  It can only contain a bitcast and ret.
+  if (InlinedMustTailCalls && pred_begin(AfterCallBB) == pred_end(AfterCallBB))
+    AfterCallBB->eraseFromParent();
+
   // We should always be able to fold the entry block of the function into the
   // single predecessor of the block...
   assert(cast<BranchInst>(Br)->isUnconditional() && "splitBasicBlock broken!");
diff --git a/lib/Transforms/Utils/IntegerDivision.cpp b/lib/Transforms/Utils/IntegerDivision.cpp
index e73a543..9f91eeb 100644
--- a/lib/Transforms/Utils/IntegerDivision.cpp
+++ b/lib/Transforms/Utils/IntegerDivision.cpp
@@ -14,7 +14,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "integer-division"
 #include "llvm/Transforms/Utils/IntegerDivision.h"
 #include "llvm/IR/Function.h"
 #include "llvm/IR/IRBuilder.h"
@@ -24,6 +23,8 @@
 
 using namespace llvm;
 
+#define DEBUG_TYPE "integer-division"
+
 /// Generate code to compute the remainder of two signed integers. Returns the
 /// remainder, which will have the sign of the dividend. Builder's insert point
 /// should be pointing where the caller wants code generated, e.g. at the srem
diff --git a/lib/Transforms/Utils/LCSSA.cpp b/lib/Transforms/Utils/LCSSA.cpp
index d538175..51a3d9c 100644
--- a/lib/Transforms/Utils/LCSSA.cpp
+++ b/lib/Transforms/Utils/LCSSA.cpp
@@ -27,7 +27,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "lcssa"
 #include "llvm/Transforms/Scalar.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/Statistic.h"
@@ -44,6 +43,8 @@
 #include "llvm/Transforms/Utils/SSAUpdater.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "lcssa"
+
 STATISTIC(NumLCSSA, "Number of live out of a loop variables");
 
 /// Return true if the specified block is in the list.
@@ -267,8 +268,6 @@
   }
 
 private:
-  bool processLoop(Loop &L);
-
   void verifyAnalysis() const override;
 };
 }
diff --git a/lib/Transforms/Utils/Local.cpp b/lib/Transforms/Utils/Local.cpp
index 9d0be8b..aedd787 100644
--- a/lib/Transforms/Utils/Local.cpp
+++ b/lib/Transforms/Utils/Local.cpp
@@ -43,6 +43,8 @@
 #include "llvm/Support/raw_ostream.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "local"
+
 STATISTIC(NumRemoved, "Number of unreachable basic blocks removed");
 
 //===----------------------------------------------------------------------===//
@@ -159,7 +161,7 @@
       // Otherwise, check to see if the switch only branches to one destination.
       // We do this by reseting "TheOnlyDest" to null when we find two non-equal
       // destinations.
-      if (i.getCaseSuccessor() != TheOnlyDest) TheOnlyDest = 0;
+      if (i.getCaseSuccessor() != TheOnlyDest) TheOnlyDest = nullptr;
     }
 
     if (CI && !TheOnlyDest) {
@@ -180,7 +182,7 @@
         // Found case matching a constant operand?
         BasicBlock *Succ = SI->getSuccessor(i);
         if (Succ == TheOnlyDest)
-          TheOnlyDest = 0;  // Don't modify the first branch to TheOnlyDest
+          TheOnlyDest = nullptr; // Don't modify the first branch to TheOnlyDest
         else
           Succ->removePredecessor(BB);
       }
@@ -233,7 +235,7 @@
 
       for (unsigned i = 0, e = IBI->getNumDestinations(); i != e; ++i) {
         if (IBI->getDestination(i) == TheOnlyDest)
-          TheOnlyDest = 0;
+          TheOnlyDest = nullptr;
         else
           IBI->getDestination(i)->removePredecessor(IBI->getParent());
       }
@@ -331,7 +333,7 @@
     // dead as we go.
     for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) {
       Value *OpV = I->getOperand(i);
-      I->setOperand(i, 0);
+      I->setOperand(i, nullptr);
 
       if (!OpV->use_empty()) continue;
 
@@ -894,24 +896,26 @@
     return PrefAlign;
   }
 
-  if (GlobalValue *GV = dyn_cast<GlobalValue>(V)) {
+  if (auto *GO = dyn_cast<GlobalObject>(V)) {
     // If there is a large requested alignment and we can, bump up the alignment
     // of the global.
-    if (GV->isDeclaration()) return Align;
+    if (GO->isDeclaration())
+      return Align;
     // If the memory we set aside for the global may not be the memory used by
     // the final program then it is impossible for us to reliably enforce the
     // preferred alignment.
-    if (GV->isWeakForLinker()) return Align;
+    if (GO->isWeakForLinker())
+      return Align;
 
-    if (GV->getAlignment() >= PrefAlign)
-      return GV->getAlignment();
+    if (GO->getAlignment() >= PrefAlign)
+      return GO->getAlignment();
     // We can only increase the alignment of the global if it has no alignment
     // specified or if it is not assigned a section.  If it is assigned a
     // section, the global could be densely packed with other objects in the
     // section, increasing the alignment could cause padding issues.
-    if (!GV->hasSection() || GV->getAlignment() == 0)
-      GV->setAlignment(PrefAlign);
-    return GV->getAlignment();
+    if (!GO->hasSection() || GO->getAlignment() == 0)
+      GO->setAlignment(PrefAlign);
+    return GO->getAlignment();
   }
 
   return Align;
@@ -928,7 +932,7 @@
   unsigned BitWidth = DL ? DL->getPointerTypeSizeInBits(V->getType()) : 64;
 
   APInt KnownZero(BitWidth, 0), KnownOne(BitWidth, 0);
-  ComputeMaskedBits(V, KnownZero, KnownOne, DL);
+  computeKnownBits(V, KnownZero, KnownOne, DL);
   unsigned TrailZ = KnownZero.countTrailingOnes();
 
   // Avoid trouble with ridiculously large TrailZ values, such as
@@ -981,10 +985,10 @@
   if (LdStHasDebugValue(DIVar, SI))
     return true;
 
-  Instruction *DbgVal = NULL;
+  Instruction *DbgVal = nullptr;
   // If an argument is zero extended then use argument directly. The ZExt
   // may be zapped by an optimization pass in future.
-  Argument *ExtendedArg = NULL;
+  Argument *ExtendedArg = nullptr;
   if (ZExtInst *ZExt = dyn_cast<ZExtInst>(SI->getOperand(0)))
     ExtendedArg = dyn_cast<Argument>(ZExt->getOperand(0));
   if (SExtInst *SExt = dyn_cast<SExtInst>(SI->getOperand(0)))
@@ -993,14 +997,7 @@
     DbgVal = Builder.insertDbgValueIntrinsic(ExtendedArg, 0, DIVar, SI);
   else
     DbgVal = Builder.insertDbgValueIntrinsic(SI->getOperand(0), 0, DIVar, SI);
-
-  // Propagate any debug metadata from the store onto the dbg.value.
-  DebugLoc SIDL = SI->getDebugLoc();
-  if (!SIDL.isUnknown())
-    DbgVal->setDebugLoc(SIDL);
-  // Otherwise propagate debug metadata from dbg.declare.
-  else
-    DbgVal->setDebugLoc(DDI->getDebugLoc());
+  DbgVal->setDebugLoc(DDI->getDebugLoc());
   return true;
 }
 
@@ -1020,17 +1017,16 @@
   Instruction *DbgVal =
     Builder.insertDbgValueIntrinsic(LI->getOperand(0), 0,
                                     DIVar, LI);
-
-  // Propagate any debug metadata from the store onto the dbg.value.
-  DebugLoc LIDL = LI->getDebugLoc();
-  if (!LIDL.isUnknown())
-    DbgVal->setDebugLoc(LIDL);
-  // Otherwise propagate debug metadata from dbg.declare.
-  else
-    DbgVal->setDebugLoc(DDI->getDebugLoc());
+  DbgVal->setDebugLoc(DDI->getDebugLoc());
   return true;
 }
 
+/// Determine whether this alloca is either a VLA or an array.
+static bool isArray(AllocaInst *AI) {
+  return AI->isArrayAllocation() ||
+    AI->getType()->getElementType()->isArrayTy();
+}
+
 /// LowerDbgDeclare - Lowers llvm.dbg.declare intrinsics into appropriate set
 /// of llvm.dbg.value intrinsics.
 bool llvm::LowerDbgDeclare(Function &F) {
@@ -1049,20 +1045,26 @@
     AllocaInst *AI = dyn_cast_or_null<AllocaInst>(DDI->getAddress());
     // If this is an alloca for a scalar variable, insert a dbg.value
     // at each load and store to the alloca and erase the dbg.declare.
-    if (AI && !AI->isArrayAllocation()) {
-
-      // We only remove the dbg.declare intrinsic if all uses are
-      // converted to dbg.value intrinsics.
-      bool RemoveDDI = true;
+    // The dbg.values allow tracking a variable even if it is not
+    // stored on the stack, while the dbg.declare can only describe
+    // the stack slot (and at a lexical-scope granularity). Later
+    // passes will attempt to elide the stack slot.
+    if (AI && !isArray(AI)) {
       for (User *U : AI->users())
         if (StoreInst *SI = dyn_cast<StoreInst>(U))
           ConvertDebugDeclareToDebugValue(DDI, SI, DIB);
         else if (LoadInst *LI = dyn_cast<LoadInst>(U))
           ConvertDebugDeclareToDebugValue(DDI, LI, DIB);
-        else
-          RemoveDDI = false;
-      if (RemoveDDI)
-        DDI->eraseFromParent();
+        else if (CallInst *CI = dyn_cast<CallInst>(U)) {
+	  // This is a call by-value or some other instruction that
+	  // takes a pointer to the variable. Insert a *value*
+	  // intrinsic that describes the alloca.
+	  auto DbgVal =
+	    DIB.insertDbgValueIntrinsic(AI, 0,
+					DIVariable(DDI->getVariable()), CI);
+	  DbgVal->setDebugLoc(DDI->getDebugLoc());
+	}
+      DDI->eraseFromParent();
     }
   }
   return true;
@@ -1076,7 +1078,7 @@
       if (DbgDeclareInst *DDI = dyn_cast<DbgDeclareInst>(U))
         return DDI;
 
-  return 0;
+  return nullptr;
 }
 
 bool llvm::replaceDbgDeclareForAlloca(AllocaInst *AI, Value *NewAllocaAddress,
diff --git a/lib/Transforms/Utils/LoopSimplify.cpp b/lib/Transforms/Utils/LoopSimplify.cpp
index 47083ea..f7787da 100644
--- a/lib/Transforms/Utils/LoopSimplify.cpp
+++ b/lib/Transforms/Utils/LoopSimplify.cpp
@@ -37,7 +37,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "loop-simplify"
 #include "llvm/Transforms/Scalar.h"
 #include "llvm/ADT/DepthFirstIterator.h"
 #include "llvm/ADT/SetOperations.h"
@@ -63,6 +62,8 @@
 #include "llvm/Transforms/Utils/LoopUtils.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "loop-simplify"
+
 STATISTIC(NumInserted, "Number of pre-header or exit blocks inserted");
 STATISTIC(NumNested  , "Number of nested loops split out");
 
@@ -85,7 +86,7 @@
 
   // Figure out *which* outside block to put this after.  Prefer an outside
   // block that neighbors a BB actually in the loop.
-  BasicBlock *FoundBB = 0;
+  BasicBlock *FoundBB = nullptr;
   for (unsigned i = 0, e = SplitPreds.size(); i != e; ++i) {
     Function::iterator BBI = SplitPreds[i];
     if (++BBI != NewBB->getParent()->end() &&
@@ -119,7 +120,7 @@
       // If the loop is branched to from an indirect branch, we won't
       // be able to fully transform the loop, because it prohibits
       // edge splitting.
-      if (isa<IndirectBrInst>(P->getTerminator())) return 0;
+      if (isa<IndirectBrInst>(P->getTerminator())) return nullptr;
 
       // Keep track of it.
       OutsideBlocks.push_back(P);
@@ -160,14 +161,14 @@
     BasicBlock *P = *I;
     if (L->contains(P)) {
       // Don't do this if the loop is exited via an indirect branch.
-      if (isa<IndirectBrInst>(P->getTerminator())) return 0;
+      if (isa<IndirectBrInst>(P->getTerminator())) return nullptr;
 
       LoopBlocks.push_back(P);
     }
   }
 
   assert(!LoopBlocks.empty() && "No edges coming in from outside the loop?");
-  BasicBlock *NewExitBB = 0;
+  BasicBlock *NewExitBB = nullptr;
 
   if (Exit->isLandingPad()) {
     SmallVector<BasicBlock*, 2> NewBBs;
@@ -211,7 +212,7 @@
   for (BasicBlock::iterator I = L->getHeader()->begin(); isa<PHINode>(I); ) {
     PHINode *PN = cast<PHINode>(I);
     ++I;
-    if (Value *V = SimplifyInstruction(PN, 0, 0, DT)) {
+    if (Value *V = SimplifyInstruction(PN, nullptr, nullptr, DT)) {
       // This is a degenerate PHI already, don't modify it!
       PN->replaceAllUsesWith(V);
       if (AA) AA->deleteValue(PN);
@@ -226,7 +227,7 @@
         // We found something tasty to remove.
         return PN;
   }
-  return 0;
+  return nullptr;
 }
 
 /// \brief If this loop has multiple backedges, try to pull one of them out into
@@ -253,14 +254,14 @@
                                 LoopInfo *LI, ScalarEvolution *SE, Pass *PP) {
   // Don't try to separate loops without a preheader.
   if (!Preheader)
-    return 0;
+    return nullptr;
 
   // The header is not a landing pad; preheader insertion should ensure this.
   assert(!L->getHeader()->isLandingPad() &&
          "Can't insert backedge to landing pad");
 
   PHINode *PN = findPHIToPartitionLoops(L, AA, DT);
-  if (PN == 0) return 0;  // No known way to partition.
+  if (!PN) return nullptr;  // No known way to partition.
 
   // Pull out all predecessors that have varying values in the loop.  This
   // handles the case when a PHI node has multiple instances of itself as
@@ -271,7 +272,7 @@
         !L->contains(PN->getIncomingBlock(i))) {
       // We can't split indirectbr edges.
       if (isa<IndirectBrInst>(PN->getIncomingBlock(i)->getTerminator()))
-        return 0;
+        return nullptr;
       OuterLoopPreds.push_back(PN->getIncomingBlock(i));
     }
   }
@@ -362,7 +363,7 @@
 
   // Unique backedge insertion currently depends on having a preheader.
   if (!Preheader)
-    return 0;
+    return nullptr;
 
   // The header is not a landing pad; preheader insertion should ensure this.
   assert(!Header->isLandingPad() && "Can't insert backedge to landing pad");
@@ -374,7 +375,7 @@
 
     // Indirectbr edges cannot be split, so we must fail if we find one.
     if (isa<IndirectBrInst>(P->getTerminator()))
-      return 0;
+      return nullptr;
 
     if (P != Preheader) BackedgeBlocks.push_back(P);
   }
@@ -403,7 +404,7 @@
     // preheader over to the new PHI node.
     unsigned PreheaderIdx = ~0U;
     bool HasUniqueIncomingValue = true;
-    Value *UniqueValue = 0;
+    Value *UniqueValue = nullptr;
     for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
       BasicBlock *IBB = PN->getIncomingBlock(i);
       Value *IV = PN->getIncomingValue(i);
@@ -412,7 +413,7 @@
       } else {
         NewPN->addIncoming(IV, IBB);
         if (HasUniqueIncomingValue) {
-          if (UniqueValue == 0)
+          if (!UniqueValue)
             UniqueValue = IV;
           else if (UniqueValue != IV)
             HasUniqueIncomingValue = false;
@@ -609,7 +610,7 @@
   PHINode *PN;
   for (BasicBlock::iterator I = L->getHeader()->begin();
        (PN = dyn_cast<PHINode>(I++)); )
-    if (Value *V = SimplifyInstruction(PN, 0, 0, DT)) {
+    if (Value *V = SimplifyInstruction(PN, nullptr, nullptr, DT)) {
       if (AA) AA->deleteValue(PN);
       if (SE) SE->forgetValue(PN);
       PN->replaceAllUsesWith(V);
@@ -653,7 +654,8 @@
         if (Inst == CI)
           continue;
         if (!L->makeLoopInvariant(Inst, AnyInvariant,
-                                 Preheader ? Preheader->getTerminator() : 0)) {
+                                  Preheader ? Preheader->getTerminator()
+                                            : nullptr)) {
           AllInvariant = false;
           break;
         }
@@ -761,12 +763,6 @@
 
     /// verifyAnalysis() - Verify LoopSimplifyForm's guarantees.
     void verifyAnalysis() const override;
-
-  private:
-    bool ProcessLoop(Loop *L);
-    BasicBlock *RewriteLoopExitBlock(Loop *L, BasicBlock *Exit);
-    Loop *SeparateNestedLoop(Loop *L, BasicBlock *Preheader);
-    BasicBlock *InsertUniqueBackedgeBlock(Loop *L, BasicBlock *Preheader);
   };
 }
 
@@ -782,7 +778,7 @@
 char &llvm::LoopSimplifyID = LoopSimplify::ID;
 Pass *llvm::createLoopSimplifyPass() { return new LoopSimplify(); }
 
-/// runOnLoop - Run down all loops in the CFG (recursively, but we could do
+/// runOnFunction - Run down all loops in the CFG (recursively, but we could do
 /// it in any convenient order) inserting preheaders...
 ///
 bool LoopSimplify::runOnFunction(Function &F) {
diff --git a/lib/Transforms/Utils/LoopUnroll.cpp b/lib/Transforms/Utils/LoopUnroll.cpp
index d2dfc20..d953e30 100644
--- a/lib/Transforms/Utils/LoopUnroll.cpp
+++ b/lib/Transforms/Utils/LoopUnroll.cpp
@@ -16,7 +16,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "loop-unroll"
 #include "llvm/Transforms/Utils/UnrollLoop.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Analysis/InstructionSimplify.h"
@@ -25,6 +24,8 @@
 #include "llvm/Analysis/ScalarEvolution.h"
 #include "llvm/IR/BasicBlock.h"
 #include "llvm/IR/Dominators.h"
+#include "llvm/IR/DiagnosticInfo.h"
+#include "llvm/IR/LLVMContext.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
@@ -34,6 +35,8 @@
 #include "llvm/Transforms/Utils/SimplifyIndVar.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "loop-unroll"
+
 // TODO: Should these be here or in LoopUnroll?
 STATISTIC(NumCompletelyUnrolled, "Number of loops completely unrolled");
 STATISTIC(NumUnrolled, "Number of loops unrolled (completely or otherwise)");
@@ -68,10 +71,10 @@
   // pred, and if there is only one distinct successor of the predecessor, and
   // if there are no PHI nodes.
   BasicBlock *OnlyPred = BB->getSinglePredecessor();
-  if (!OnlyPred) return 0;
+  if (!OnlyPred) return nullptr;
 
   if (OnlyPred->getTerminator()->getNumSuccessors() != 1)
-    return 0;
+    return nullptr;
 
   DEBUG(dbgs() << "Merging: " << *BB << "into: " << *OnlyPred);
 
@@ -227,20 +230,33 @@
       (unsigned)GreatestCommonDivisor64(Count, TripMultiple);
   }
 
+  // Report the unrolling decision.
+  DebugLoc LoopLoc = L->getStartLoc();
+  Function *F = Header->getParent();
+  LLVMContext &Ctx = F->getContext();
+
   if (CompletelyUnroll) {
     DEBUG(dbgs() << "COMPLETELY UNROLLING loop %" << Header->getName()
           << " with trip count " << TripCount << "!\n");
+    emitOptimizationRemark(Ctx, DEBUG_TYPE, *F, LoopLoc,
+                           Twine("completely unrolled loop with ") +
+                               Twine(TripCount) + " iterations");
   } else {
     DEBUG(dbgs() << "UNROLLING loop %" << Header->getName()
           << " by " << Count);
+    Twine DiagMsg("unrolled loop by a factor of " + Twine(Count));
     if (TripMultiple == 0 || BreakoutTrip != TripMultiple) {
       DEBUG(dbgs() << " with a breakout at trip " << BreakoutTrip);
+      DiagMsg.concat(" with a breakout at trip " + Twine(BreakoutTrip));
     } else if (TripMultiple != 1) {
       DEBUG(dbgs() << " with " << TripMultiple << " trips per branch");
+      DiagMsg.concat(" with " + Twine(TripMultiple) + " trips per branch");
     } else if (RuntimeTripCount) {
       DEBUG(dbgs() << " with run-time trip count");
+      DiagMsg.concat(" with run-time trip count");
     }
     DEBUG(dbgs() << "!\n");
+    emitOptimizationRemark(Ctx, DEBUG_TYPE, *F, LoopLoc, DiagMsg);
   }
 
   bool ContinueOnTrue = L->contains(BI->getSuccessor(0));
@@ -411,7 +427,7 @@
     }
   }
 
-  DominatorTree *DT = 0;
+  DominatorTree *DT = nullptr;
   if (PP) {
     // FIXME: Reconstruct dom info, because it is not preserved properly.
     // Incrementally updating domtree after loop unrolling would be easy.
@@ -458,7 +474,7 @@
 
   Loop *OuterL = L->getParentLoop();
   // Remove the loop from the LoopPassManager if it's completely removed.
-  if (CompletelyUnroll && LPM != NULL)
+  if (CompletelyUnroll && LPM != nullptr)
     LPM->deleteLoopFromQueue(L);
 
   // If we have a pass and a DominatorTree we should re-simplify impacted loops
@@ -470,7 +486,7 @@
       OuterL = L;
     if (OuterL) {
       ScalarEvolution *SE = PP->getAnalysisIfAvailable<ScalarEvolution>();
-      simplifyLoop(OuterL, DT, LI, PP, /*AliasAnalysis*/ 0, SE);
+      simplifyLoop(OuterL, DT, LI, PP, /*AliasAnalysis*/ nullptr, SE);
       formLCSSARecursively(*OuterL, *DT, SE);
     }
   }
diff --git a/lib/Transforms/Utils/LoopUnrollRuntime.cpp b/lib/Transforms/Utils/LoopUnrollRuntime.cpp
index d801d5f..5bef091 100644
--- a/lib/Transforms/Utils/LoopUnrollRuntime.cpp
+++ b/lib/Transforms/Utils/LoopUnrollRuntime.cpp
@@ -21,7 +21,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "loop-unroll"
 #include "llvm/Transforms/Utils/UnrollLoop.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Analysis/LoopIterator.h"
@@ -37,6 +36,8 @@
 
 using namespace llvm;
 
+#define DEBUG_TYPE "loop-unroll"
+
 STATISTIC(NumRuntimeUnrolled,
           "Number of loops unrolled with run-time trip counts");
 
@@ -58,7 +59,7 @@
                           BasicBlock *OrigPH, BasicBlock *NewPH,
                           ValueToValueMapTy &LVMap, Pass *P) {
   BasicBlock *Latch = L->getLoopLatch();
-  assert(Latch != 0 && "Loop must have a latch");
+  assert(Latch && "Loop must have a latch");
 
   // Create a PHI node for each outgoing value from the original loop
   // (which means it is an outgoing value from the prolog code too).
@@ -110,7 +111,7 @@
     new ICmpInst(InsertPt, ICmpInst::ICMP_ULT, TripCount,
                  ConstantInt::get(TripCount->getType(), Count));
   BasicBlock *Exit = L->getUniqueExitBlock();
-  assert(Exit != 0 && "Loop must have a single exit block only");
+  assert(Exit && "Loop must have a single exit block only");
   // Split the exit to maintain loop canonicalization guarantees
   SmallVector<BasicBlock*, 4> Preds(pred_begin(Exit), pred_end(Exit));
   if (!Exit->isLandingPad()) {
@@ -232,7 +233,7 @@
 
   // Make sure the loop is in canonical form, and there is a single
   // exit block only.
-  if (!L->isLoopSimplifyForm() || L->getUniqueExitBlock() == 0)
+  if (!L->isLoopSimplifyForm() || !L->getUniqueExitBlock())
     return false;
 
   // Use Scalar Evolution to compute the trip count.  This allows more
@@ -240,7 +241,7 @@
   if (!LPM)
     return false;
   ScalarEvolution *SE = LPM->getAnalysisIfAvailable<ScalarEvolution>();
-  if (SE == 0)
+  if (!SE)
     return false;
 
   // Only unroll loops with a computable trip count and the trip count needs
@@ -301,7 +302,7 @@
   ValueToValueMapTy LVMap;
   Function *F = Header->getParent();
   // These variables are used to update the CFG links in each iteration
-  BasicBlock *CompareBB = 0;
+  BasicBlock *CompareBB = nullptr;
   BasicBlock *LastLoopBB = PH;
   // Get an ordered list of blocks in the loop to help with the ordering of the
   // cloned blocks in the prolog code
diff --git a/lib/Transforms/Utils/LowerExpectIntrinsic.cpp b/lib/Transforms/Utils/LowerExpectIntrinsic.cpp
index 3e61289..ff89e74 100644
--- a/lib/Transforms/Utils/LowerExpectIntrinsic.cpp
+++ b/lib/Transforms/Utils/LowerExpectIntrinsic.cpp
@@ -11,7 +11,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "lower-expect-intrinsic"
 #include "llvm/Transforms/Scalar.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/IR/BasicBlock.h"
@@ -29,6 +28,8 @@
 
 using namespace llvm;
 
+#define DEBUG_TYPE "lower-expect-intrinsic"
+
 STATISTIC(IfHandled, "Number of 'expect' intrinsic instructions handled");
 
 static cl::opt<uint32_t>
diff --git a/lib/Transforms/Utils/LowerInvoke.cpp b/lib/Transforms/Utils/LowerInvoke.cpp
index b1f758e..66d57b0 100644
--- a/lib/Transforms/Utils/LowerInvoke.cpp
+++ b/lib/Transforms/Utils/LowerInvoke.cpp
@@ -14,7 +14,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "lowerinvoke"
 #include "llvm/Transforms/Scalar.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/Statistic.h"
@@ -25,6 +24,8 @@
 #include "llvm/Support/CommandLine.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "lowerinvoke"
+
 STATISTIC(NumInvokes, "Number of invokes replaced");
 
 namespace {
diff --git a/lib/Transforms/Utils/LowerSwitch.cpp b/lib/Transforms/Utils/LowerSwitch.cpp
index 6fb7410..9ef694c 100644
--- a/lib/Transforms/Utils/LowerSwitch.cpp
+++ b/lib/Transforms/Utils/LowerSwitch.cpp
@@ -27,6 +27,8 @@
 #include <algorithm>
 using namespace llvm;
 
+#define DEBUG_TYPE "lower-switch"
+
 namespace {
   /// LowerSwitch Pass - Replace all SwitchInst instructions with chained branch
   /// instructions.
@@ -51,7 +53,8 @@
       Constant* High;
       BasicBlock* BB;
 
-      CaseRange(Constant *low = 0, Constant *high = 0, BasicBlock *bb = 0) :
+      CaseRange(Constant *low = nullptr, Constant *high = nullptr,
+                BasicBlock *bb = nullptr) :
         Low(low), High(high), BB(bb) { }
     };
 
@@ -182,7 +185,7 @@
   F->getBasicBlockList().insert(++FI, NewLeaf);
 
   // Emit comparison
-  ICmpInst* Comp = NULL;
+  ICmpInst* Comp = nullptr;
   if (Leaf.Low == Leaf.High) {
     // Make the seteq instruction...
     Comp = new ICmpInst(*NewLeaf, ICmpInst::ICMP_EQ, Val,
diff --git a/lib/Transforms/Utils/Mem2Reg.cpp b/lib/Transforms/Utils/Mem2Reg.cpp
index a188ac5..189caa7 100644
--- a/lib/Transforms/Utils/Mem2Reg.cpp
+++ b/lib/Transforms/Utils/Mem2Reg.cpp
@@ -12,7 +12,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "mem2reg"
 #include "llvm/Transforms/Scalar.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/IR/Dominators.h"
@@ -22,6 +21,8 @@
 #include "llvm/Transforms/Utils/UnifyFunctionExitNodes.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "mem2reg"
+
 STATISTIC(NumPromoted, "Number of alloca's promoted");
 
 namespace {
diff --git a/lib/Transforms/Utils/ModuleUtils.cpp b/lib/Transforms/Utils/ModuleUtils.cpp
index ff6e6f9..d9dbbca 100644
--- a/lib/Transforms/Utils/ModuleUtils.cpp
+++ b/lib/Transforms/Utils/ModuleUtils.cpp
@@ -24,16 +24,16 @@
                                 Module &M, Function *F, int Priority) {
   IRBuilder<> IRB(M.getContext());
   FunctionType *FnTy = FunctionType::get(IRB.getVoidTy(), false);
-  StructType *Ty = StructType::get(
-      IRB.getInt32Ty(), PointerType::getUnqual(FnTy), NULL);
-
-  Constant *RuntimeCtorInit = ConstantStruct::get(
-      Ty, IRB.getInt32(Priority), F, NULL);
 
   // Get the current set of static global constructors and add the new ctor
   // to the list.
   SmallVector<Constant *, 16> CurrentCtors;
-  if (GlobalVariable * GVCtor = M.getNamedGlobal(Array)) {
+  StructType *EltTy;
+  if (GlobalVariable *GVCtor = M.getNamedGlobal(Array)) {
+    // If there is a global_ctors array, use the existing struct type, which can
+    // have 2 or 3 fields.
+    ArrayType *ATy = cast<ArrayType>(GVCtor->getType()->getElementType());
+    EltTy = cast<StructType>(ATy->getElementType());
     if (Constant *Init = GVCtor->getInitializer()) {
       unsigned n = Init->getNumOperands();
       CurrentCtors.reserve(n + 1);
@@ -41,13 +41,26 @@
         CurrentCtors.push_back(cast<Constant>(Init->getOperand(i)));
     }
     GVCtor->eraseFromParent();
+  } else {
+    // Use a simple two-field struct if there isn't one already.
+    EltTy = StructType::get(IRB.getInt32Ty(), PointerType::getUnqual(FnTy),
+                            nullptr);
   }
 
+  // Build a 2 or 3 field global_ctor entry.  We don't take a comdat key.
+  Constant *CSVals[3];
+  CSVals[0] = IRB.getInt32(Priority);
+  CSVals[1] = F;
+  // FIXME: Drop support for the two element form in LLVM 4.0.
+  if (EltTy->getNumElements() >= 3)
+    CSVals[2] = llvm::Constant::getNullValue(IRB.getInt8PtrTy());
+  Constant *RuntimeCtorInit =
+      ConstantStruct::get(EltTy, makeArrayRef(CSVals, EltTy->getNumElements()));
+
   CurrentCtors.push_back(RuntimeCtorInit);
 
   // Create a new initializer.
-  ArrayType *AT = ArrayType::get(RuntimeCtorInit->getType(),
-                                 CurrentCtors.size());
+  ArrayType *AT = ArrayType::get(EltTy, CurrentCtors.size());
   Constant *NewInit = ConstantArray::get(AT, CurrentCtors);
 
   // Create the new global variable and replace all uses of
diff --git a/lib/Transforms/Utils/PromoteMemoryToRegister.cpp b/lib/Transforms/Utils/PromoteMemoryToRegister.cpp
index 25fab89..06d73fe 100644
--- a/lib/Transforms/Utils/PromoteMemoryToRegister.cpp
+++ b/lib/Transforms/Utils/PromoteMemoryToRegister.cpp
@@ -25,7 +25,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "mem2reg"
 #include "llvm/Transforms/Utils/PromoteMemToReg.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/DenseMap.h"
@@ -51,6 +50,8 @@
 #include <queue>
 using namespace llvm;
 
+#define DEBUG_TYPE "mem2reg"
+
 STATISTIC(NumLocalPromoted, "Number of alloca's promoted within one block");
 STATISTIC(NumSingleStore,   "Number of alloca's promoted with a single store");
 STATISTIC(NumDeadAlloca,    "Number of dead alloca's removed");
@@ -59,6 +60,7 @@
 bool llvm::isAllocaPromotable(const AllocaInst *AI) {
   // FIXME: If the memory unit is of pointer or integer type, we can permit
   // assignments to subsections of the memory unit.
+  unsigned AS = AI->getType()->getAddressSpace();
 
   // Only allow direct and non-volatile loads and stores...
   for (const User *U : AI->users()) {
@@ -79,12 +81,12 @@
           II->getIntrinsicID() != Intrinsic::lifetime_end)
         return false;
     } else if (const BitCastInst *BCI = dyn_cast<BitCastInst>(U)) {
-      if (BCI->getType() != Type::getInt8PtrTy(U->getContext()))
+      if (BCI->getType() != Type::getInt8PtrTy(U->getContext(), AS))
         return false;
       if (!onlyUsedByLifetimeMarkers(BCI))
         return false;
     } else if (const GetElementPtrInst *GEPI = dyn_cast<GetElementPtrInst>(U)) {
-      if (GEPI->getType() != Type::getInt8PtrTy(U->getContext()))
+      if (GEPI->getType() != Type::getInt8PtrTy(U->getContext(), AS))
         return false;
       if (!GEPI->hasAllZeroIndices())
         return false;
@@ -114,11 +116,11 @@
   void clear() {
     DefiningBlocks.clear();
     UsingBlocks.clear();
-    OnlyStore = 0;
-    OnlyBlock = 0;
+    OnlyStore = nullptr;
+    OnlyBlock = nullptr;
     OnlyUsedInOneBlock = true;
-    AllocaPointerVal = 0;
-    DbgDeclare = 0;
+    AllocaPointerVal = nullptr;
+    DbgDeclare = nullptr;
   }
 
   /// Scan the uses of the specified alloca, filling in the AllocaInfo used
@@ -146,7 +148,7 @@
       }
 
       if (OnlyUsedInOneBlock) {
-        if (OnlyBlock == 0)
+        if (!OnlyBlock)
           OnlyBlock = User->getParent();
         else if (OnlyBlock != User->getParent())
           OnlyUsedInOneBlock = false;
@@ -162,7 +164,7 @@
 public:
   typedef std::vector<Value *> ValVector;
 
-  RenamePassData() : BB(NULL), Pred(NULL), Values() {}
+  RenamePassData() : BB(nullptr), Pred(nullptr), Values() {}
   RenamePassData(BasicBlock *B, BasicBlock *P, const ValVector &V)
       : BB(B), Pred(P), Values(V) {}
   BasicBlock *BB;
@@ -471,7 +473,8 @@
     // Find the nearest store that has a lower index than this load.
     StoresByIndexTy::iterator I =
         std::lower_bound(StoresByIndex.begin(), StoresByIndex.end(),
-                         std::make_pair(LoadIdx, static_cast<StoreInst *>(0)),
+                         std::make_pair(LoadIdx,
+                                        static_cast<StoreInst *>(nullptr)),
                          less_first());
 
     if (I == StoresByIndex.begin())
@@ -632,7 +635,7 @@
   // and inserting the phi nodes we marked as necessary
   //
   std::vector<RenamePassData> RenamePassWorkList;
-  RenamePassWorkList.push_back(RenamePassData(F.begin(), 0, Values));
+  RenamePassWorkList.push_back(RenamePassData(F.begin(), nullptr, Values));
   do {
     RenamePassData RPD;
     RPD.swap(RenamePassWorkList.back());
@@ -682,7 +685,7 @@
       PHINode *PN = I->second;
 
       // If this PHI node merges one value and/or undefs, get the value.
-      if (Value *V = SimplifyInstruction(PN, 0, 0, &DT)) {
+      if (Value *V = SimplifyInstruction(PN, nullptr, nullptr, &DT)) {
         if (AST && PN->getType()->isPointerTy())
           AST->deleteValue(PN);
         PN->replaceAllUsesWith(V);
@@ -990,7 +993,7 @@
         // Get the next phi node.
         ++PNI;
         APN = dyn_cast<PHINode>(PNI);
-        if (APN == 0)
+        if (!APN)
           break;
 
         // Verify that it is missing entries.  If not, it is not being inserted
diff --git a/lib/Transforms/Utils/SSAUpdater.cpp b/lib/Transforms/Utils/SSAUpdater.cpp
index 28f5c44..3fcb789 100644
--- a/lib/Transforms/Utils/SSAUpdater.cpp
+++ b/lib/Transforms/Utils/SSAUpdater.cpp
@@ -11,7 +11,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "ssaupdater"
 #include "llvm/Transforms/Utils/SSAUpdater.h"
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/TinyPtrVector.h"
@@ -28,20 +27,22 @@
 
 using namespace llvm;
 
+#define DEBUG_TYPE "ssaupdater"
+
 typedef DenseMap<BasicBlock*, Value*> AvailableValsTy;
 static AvailableValsTy &getAvailableVals(void *AV) {
   return *static_cast<AvailableValsTy*>(AV);
 }
 
 SSAUpdater::SSAUpdater(SmallVectorImpl<PHINode*> *NewPHI)
-  : AV(0), ProtoType(0), ProtoName(), InsertedPHIs(NewPHI) {}
+  : AV(nullptr), ProtoType(nullptr), ProtoName(), InsertedPHIs(NewPHI) {}
 
 SSAUpdater::~SSAUpdater() {
   delete static_cast<AvailableValsTy*>(AV);
 }
 
 void SSAUpdater::Initialize(Type *Ty, StringRef Name) {
-  if (AV == 0)
+  if (!AV)
     AV = new AvailableValsTy();
   else
     getAvailableVals(AV).clear();
@@ -54,7 +55,7 @@
 }
 
 void SSAUpdater::AddAvailableValue(BasicBlock *BB, Value *V) {
-  assert(ProtoType != 0 && "Need to initialize SSAUpdater");
+  assert(ProtoType && "Need to initialize SSAUpdater");
   assert(ProtoType == V->getType() &&
          "All rewritten values must have the same type");
   getAvailableVals(AV)[BB] = V;
@@ -90,7 +91,7 @@
   // Otherwise, we have the hard case.  Get the live-in values for each
   // predecessor.
   SmallVector<std::pair<BasicBlock*, Value*>, 8> PredValues;
-  Value *SingularValue = 0;
+  Value *SingularValue = nullptr;
 
   // We can get our predecessor info by walking the pred_iterator list, but it
   // is relatively slow.  If we already have PHI nodes in this block, walk one
@@ -105,7 +106,7 @@
       if (i == 0)
         SingularValue = PredVal;
       else if (PredVal != SingularValue)
-        SingularValue = 0;
+        SingularValue = nullptr;
     }
   } else {
     bool isFirstPred = true;
@@ -119,7 +120,7 @@
         SingularValue = PredVal;
         isFirstPred = false;
       } else if (PredVal != SingularValue)
-        SingularValue = 0;
+        SingularValue = nullptr;
     }
   }
 
@@ -128,7 +129,7 @@
     return UndefValue::get(ProtoType);
 
   // Otherwise, if all the merged values are the same, just use it.
-  if (SingularValue != 0)
+  if (SingularValue)
     return SingularValue;
 
   // Otherwise, we do need a PHI: check to see if we already have one available
@@ -291,7 +292,7 @@
     PHINode *PHI = ValueIsPHI(Val, Updater);
     if (PHI && PHI->getNumIncomingValues() == 0)
       return PHI;
-    return 0;
+    return nullptr;
   }
 
   /// GetPHIValue - For the specified PHI instruction, return the value
@@ -401,7 +402,7 @@
     // the order of these instructions in the block.  If the first use in the
     // block is a load, then it uses the live in value.  The last store defines
     // the live out value.  We handle this by doing a linear scan of the block.
-    Value *StoredValue = 0;
+    Value *StoredValue = nullptr;
     for (BasicBlock::iterator II = BB->begin(), E = BB->end(); II != E; ++II) {
       if (LoadInst *L = dyn_cast<LoadInst>(II)) {
         // If this is a load from an unrelated pointer, ignore it.
diff --git a/lib/Transforms/Utils/SimplifyCFG.cpp b/lib/Transforms/Utils/SimplifyCFG.cpp
index 1e88587..150dbdd 100644
--- a/lib/Transforms/Utils/SimplifyCFG.cpp
+++ b/lib/Transforms/Utils/SimplifyCFG.cpp
@@ -11,7 +11,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "simplifycfg"
 #include "llvm/Transforms/Utils/Local.h"
 #include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/STLExtras.h"
@@ -50,6 +49,8 @@
 using namespace llvm;
 using namespace PatternMatch;
 
+#define DEBUG_TYPE "simplifycfg"
+
 static cl::opt<unsigned>
 PHINodeFoldingThreshold("phi-node-folding-threshold", cl::Hidden, cl::init(1),
    cl::desc("Control the amount of phi node folding to perform (default = 1)"));
@@ -212,6 +213,7 @@
     if (!cast<GEPOperator>(I)->hasAllConstantIndices())
       return UINT_MAX;
     return 1;
+  case Instruction::ExtractValue:
   case Instruction::Load:
   case Instruction::Add:
   case Instruction::Sub:
@@ -272,12 +274,12 @@
   // branch to BB, then it must be in the 'conditional' part of the "if
   // statement".  If not, it definitely dominates the region.
   BranchInst *BI = dyn_cast<BranchInst>(PBB->getTerminator());
-  if (BI == 0 || BI->isConditional() || BI->getSuccessor(0) != BB)
+  if (!BI || BI->isConditional() || BI->getSuccessor(0) != BB)
     return true;
 
   // If we aren't allowing aggressive promotion anymore, then don't consider
   // instructions in the 'if region'.
-  if (AggressiveInsts == 0) return false;
+  if (!AggressiveInsts) return false;
 
   // If we have seen this instruction before, don't count it again.
   if (AggressiveInsts->count(I)) return true;
@@ -332,7 +334,7 @@
           return cast<ConstantInt>
             (ConstantExpr::getIntegerCast(CI, PtrTy, /*isSigned=*/false));
       }
-  return 0;
+  return nullptr;
 }
 
 /// GatherConstantCompares - Given a potentially 'or'd or 'and'd together
@@ -343,7 +345,7 @@
 GatherConstantCompares(Value *V, std::vector<ConstantInt*> &Vals, Value *&Extra,
                        const DataLayout *DL, bool isEQ, unsigned &UsedICmps) {
   Instruction *I = dyn_cast<Instruction>(V);
-  if (I == 0) return 0;
+  if (!I) return nullptr;
 
   // If this is an icmp against a constant, handle this as one of the cases.
   if (ICmpInst *ICI = dyn_cast<ICmpInst>(I)) {
@@ -390,19 +392,19 @@
 
       // If there are a ton of values, we don't want to make a ginormous switch.
       if (Span.getSetSize().ugt(8) || Span.isEmptySet())
-        return 0;
+        return nullptr;
 
       for (APInt Tmp = Span.getLower(); Tmp != Span.getUpper(); ++Tmp)
         Vals.push_back(ConstantInt::get(V->getContext(), Tmp));
       UsedICmps++;
       return hasAdd ? RHSVal : I->getOperand(0);
     }
-    return 0;
+    return nullptr;
   }
 
   // Otherwise, we can only handle an | or &, depending on isEQ.
   if (I->getOpcode() != (isEQ ? Instruction::Or : Instruction::And))
-    return 0;
+    return nullptr;
 
   unsigned NumValsBeforeLHS = Vals.size();
   unsigned UsedICmpsBeforeLHS = UsedICmps;
@@ -420,19 +422,19 @@
 
     // The RHS of the or/and can't be folded in and we haven't used "Extra" yet,
     // set it and return success.
-    if (Extra == 0 || Extra == I->getOperand(1)) {
+    if (Extra == nullptr || Extra == I->getOperand(1)) {
       Extra = I->getOperand(1);
       return LHS;
     }
 
     Vals.resize(NumValsBeforeLHS);
     UsedICmps = UsedICmpsBeforeLHS;
-    return 0;
+    return nullptr;
   }
 
   // If the LHS can't be folded in, but Extra is available and RHS can, try to
   // use LHS as Extra.
-  if (Extra == 0 || Extra == I->getOperand(0)) {
+  if (Extra == nullptr || Extra == I->getOperand(0)) {
     Value *OldExtra = Extra;
     Extra = I->getOperand(0);
     if (Value *RHS = GatherConstantCompares(I->getOperand(1), Vals, Extra, DL,
@@ -442,11 +444,11 @@
     Extra = OldExtra;
   }
 
-  return 0;
+  return nullptr;
 }
 
 static void EraseTerminatorInstAndDCECond(TerminatorInst *TI) {
-  Instruction *Cond = 0;
+  Instruction *Cond = nullptr;
   if (SwitchInst *SI = dyn_cast<SwitchInst>(TI)) {
     Cond = dyn_cast<Instruction>(SI->getCondition());
   } else if (BranchInst *BI = dyn_cast<BranchInst>(TI)) {
@@ -463,7 +465,7 @@
 /// isValueEqualityComparison - Return true if the specified terminator checks
 /// to see if a value is equal to constant integer value.
 Value *SimplifyCFGOpt::isValueEqualityComparison(TerminatorInst *TI) {
-  Value *CV = 0;
+  Value *CV = nullptr;
   if (SwitchInst *SI = dyn_cast<SwitchInst>(TI)) {
     // Do not permit merging of large switch instructions into their
     // predecessors unless there is only one predecessor.
@@ -653,11 +655,11 @@
 
   // Otherwise, TI's block must correspond to some matched value.  Find out
   // which value (or set of values) this is.
-  ConstantInt *TIV = 0;
+  ConstantInt *TIV = nullptr;
   BasicBlock *TIBB = TI->getParent();
   for (unsigned i = 0, e = PredCases.size(); i != e; ++i)
     if (PredCases[i].Dest == TIBB) {
-      if (TIV != 0)
+      if (TIV)
         return false;  // Cannot handle multiple values coming to this block.
       TIV = PredCases[i].Value;
     }
@@ -665,7 +667,7 @@
 
   // Okay, we found the one constant that our value can be if we get into TI's
   // BB.  Find out which successor will unconditionally be branched to.
-  BasicBlock *TheRealDest = 0;
+  BasicBlock *TheRealDest = nullptr;
   for (unsigned i = 0, e = ThisCases.size(); i != e; ++i)
     if (ThisCases[i].Value == TIV) {
       TheRealDest = ThisCases[i].Dest;
@@ -673,7 +675,7 @@
     }
 
   // If not handled by any explicit cases, it is handled by the default case.
-  if (TheRealDest == 0) TheRealDest = ThisDef;
+  if (!TheRealDest) TheRealDest = ThisDef;
 
   // Remove PHI node entries for dead edges.
   BasicBlock *CheckEdge = TheRealDest;
@@ -681,7 +683,7 @@
     if (*SI != CheckEdge)
       (*SI)->removePredecessor(TIBB);
     else
-      CheckEdge = 0;
+      CheckEdge = nullptr;
 
   // Insert the new branch.
   Instruction *NI = Builder.CreateBr(TheRealDest);
@@ -950,10 +952,10 @@
       // Okay, last check.  If BB is still a successor of PSI, then we must
       // have an infinite loop case.  If so, add an infinitely looping block
       // to handle the case to preserve the behavior of the code.
-      BasicBlock *InfLoopBlock = 0;
+      BasicBlock *InfLoopBlock = nullptr;
       for (unsigned i = 0, e = NewSI->getNumSuccessors(); i != e; ++i)
         if (NewSI->getSuccessor(i) == BB) {
-          if (InfLoopBlock == 0) {
+          if (!InfLoopBlock) {
             // Insert it at the end of the function, because it's either code,
             // or it won't matter if it's hot. :)
             InfLoopBlock = BasicBlock::Create(BB->getContext(),
@@ -1099,7 +1101,7 @@
       // These values do not agree.  Insert a select instruction before NT
       // that determines the right value.
       SelectInst *&SI = InsertedSelects[std::make_pair(BB1V, BB2V)];
-      if (SI == 0)
+      if (!SI)
         SI = cast<SelectInst>
           (Builder.CreateSelect(BI->getCondition(), BB1V, BB2V,
                                 BB1V->getName()+"."+BB2V->getName()));
@@ -1144,7 +1146,7 @@
 
   // Gather the PHI nodes in BBEnd.
   std::map<Value*, std::pair<Value*, PHINode*> > MapValueFromBB1ToBB2;
-  Instruction *FirstNonPhiInBBEnd = 0;
+  Instruction *FirstNonPhiInBBEnd = nullptr;
   for (BasicBlock::iterator I = BBEnd->begin(), E = BBEnd->end();
        I != E; ++I) {
     if (PHINode *PN = dyn_cast<PHINode>(I)) {
@@ -1222,7 +1224,7 @@
     // The operands should be either the same or they need to be generated
     // with a PHI node after sinking. We only handle the case where there is
     // a single pair of different operands.
-    Value *DifferentOp1 = 0, *DifferentOp2 = 0;
+    Value *DifferentOp1 = nullptr, *DifferentOp2 = nullptr;
     unsigned Op1Idx = 0;
     for (unsigned I = 0, E = I1->getNumOperands(); I != E; ++I) {
       if (I1->getOperand(I) == I2->getOperand(I))
@@ -1318,11 +1320,11 @@
                                      BasicBlock *StoreBB, BasicBlock *EndBB) {
   StoreInst *StoreToHoist = dyn_cast<StoreInst>(I);
   if (!StoreToHoist)
-    return 0;
+    return nullptr;
 
   // Volatile or atomic.
   if (!StoreToHoist->isSimple())
-    return 0;
+    return nullptr;
 
   Value *StorePtr = StoreToHoist->getPointerOperand();
 
@@ -1334,7 +1336,7 @@
 
     // Could be calling an instruction that effects memory like free().
     if (CurI->mayHaveSideEffects() && !isa<StoreInst>(CurI))
-      return 0;
+      return nullptr;
 
     StoreInst *SI = dyn_cast<StoreInst>(CurI);
     // Found the previous store make sure it stores to the same location.
@@ -1342,10 +1344,10 @@
       // Found the previous store, return its value operand.
       return SI->getValueOperand();
     else if (SI)
-      return 0; // Unknown store.
+      return nullptr; // Unknown store.
   }
 
-  return 0;
+  return nullptr;
 }
 
 /// \brief Speculate a conditional basic block flattening the CFG.
@@ -1411,8 +1413,8 @@
   SmallDenseMap<Instruction *, unsigned, 4> SinkCandidateUseCounts;
 
   unsigned SpeculationCost = 0;
-  Value *SpeculatedStoreValue = 0;
-  StoreInst *SpeculatedStore = 0;
+  Value *SpeculatedStoreValue = nullptr;
+  StoreInst *SpeculatedStore = nullptr;
   for (BasicBlock::iterator BBI = ThenBB->begin(),
                             BBE = std::prev(ThenBB->end());
        BBI != BBE; ++BBI) {
@@ -1620,7 +1622,7 @@
   // constants.
   for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
     ConstantInt *CB = dyn_cast<ConstantInt>(PN->getIncomingValue(i));
-    if (CB == 0 || !CB->getType()->isIntegerTy(1)) continue;
+    if (!CB || !CB->getType()->isIntegerTy(1)) continue;
 
     // Okay, we now know that all edges from PredBB should be revectored to
     // branch to RealDest.
@@ -1745,7 +1747,7 @@
   // If we folded the first phi, PN dangles at this point.  Refresh it.  If
   // we ran out of PHIs then we simplified them all.
   PN = dyn_cast<PHINode>(BB->begin());
-  if (PN == 0) return true;
+  if (!PN) return true;
 
   // Don't fold i1 branches on PHIs which contain binary operators.  These can
   // often be turned into switches and other things.
@@ -1759,11 +1761,11 @@
   // instructions in the predecessor blocks can be promoted as well.  If
   // not, we won't be able to get rid of the control flow, so it's not
   // worth promoting to select instructions.
-  BasicBlock *DomBlock = 0;
+  BasicBlock *DomBlock = nullptr;
   BasicBlock *IfBlock1 = PN->getIncomingBlock(0);
   BasicBlock *IfBlock2 = PN->getIncomingBlock(1);
   if (cast<BranchInst>(IfBlock1->getTerminator())->isConditional()) {
-    IfBlock1 = 0;
+    IfBlock1 = nullptr;
   } else {
     DomBlock = *pred_begin(IfBlock1);
     for (BasicBlock::iterator I = IfBlock1->begin();!isa<TerminatorInst>(I);++I)
@@ -1776,7 +1778,7 @@
   }
 
   if (cast<BranchInst>(IfBlock2->getTerminator())->isConditional()) {
-    IfBlock2 = 0;
+    IfBlock2 = nullptr;
   } else {
     DomBlock = *pred_begin(IfBlock2);
     for (BasicBlock::iterator I = IfBlock2->begin();!isa<TerminatorInst>(I);++I)
@@ -1959,7 +1961,7 @@
 bool llvm::FoldBranchToCommonDest(BranchInst *BI) {
   BasicBlock *BB = BI->getParent();
 
-  Instruction *Cond = 0;
+  Instruction *Cond = nullptr;
   if (BI->isConditional())
     Cond = dyn_cast<Instruction>(BI->getCondition());
   else {
@@ -1985,12 +1987,12 @@
           }
         }
 
-    if (Cond == 0)
+    if (!Cond)
       return false;
   }
 
-  if (Cond == 0 || (!isa<CmpInst>(Cond) && !isa<BinaryOperator>(Cond)) ||
-    Cond->getParent() != BB || !Cond->hasOneUse())
+  if (!Cond || (!isa<CmpInst>(Cond) && !isa<BinaryOperator>(Cond)) ||
+      Cond->getParent() != BB || !Cond->hasOneUse())
   return false;
 
   // Only allow this if the condition is a simple instruction that can be
@@ -2005,7 +2007,7 @@
   // that feeds the branch.  We later ensure that any values that _it_ uses
   // were also live in the predecessor, so that we don't unnecessarily create
   // register pressure or inhibit out-of-order execution.
-  Instruction *BonusInst = 0;
+  Instruction *BonusInst = nullptr;
   if (&*FrontIt != Cond &&
       FrontIt->hasOneUse() && FrontIt->user_back() == Cond &&
       isSafeToSpeculativelyExecute(FrontIt)) {
@@ -2040,7 +2042,7 @@
 
   // Finally, don't infinitely unroll conditional loops.
   BasicBlock *TrueDest  = BI->getSuccessor(0);
-  BasicBlock *FalseDest = (BI->isConditional()) ? BI->getSuccessor(1) : 0;
+  BasicBlock *FalseDest = (BI->isConditional()) ? BI->getSuccessor(1) : nullptr;
   if (TrueDest == BB || FalseDest == BB)
     return false;
 
@@ -2052,7 +2054,7 @@
     // the common successor, verify that the same value flows in from both
     // blocks.
     SmallVector<PHINode*, 4> PHIs;
-    if (PBI == 0 || PBI->isUnconditional() ||
+    if (!PBI || PBI->isUnconditional() ||
         (BI->isConditional() &&
          !SafeToMergeTerminators(BI, PBI)) ||
         (!BI->isConditional() &&
@@ -2142,7 +2144,7 @@
     }
 
     // If we have a bonus inst, clone it into the predecessor block.
-    Instruction *NewBonus = 0;
+    Instruction *NewBonus = nullptr;
     if (BonusInst) {
       NewBonus = BonusInst->clone();
 
@@ -2218,14 +2220,14 @@
                          MDBuilder(BI->getContext()).
                          createBranchWeights(MDWeights));
       } else
-        PBI->setMetadata(LLVMContext::MD_prof, NULL);
+        PBI->setMetadata(LLVMContext::MD_prof, nullptr);
     } else {
       // Update PHI nodes in the common successors.
       for (unsigned i = 0, e = PHIs.size(); i != e; ++i) {
         ConstantInt *PBI_C = cast<ConstantInt>(
           PHIs[i]->getIncomingValueForBlock(PBI->getParent()));
         assert(PBI_C->getType()->isIntegerTy(1));
-        Instruction *MergedCond = 0;
+        Instruction *MergedCond = nullptr;
         if (PBI->getSuccessor(0) == TrueDest) {
           // Create (PBI_Cond and PBI_C) or (!PBI_Cond and BI_Value)
           // PBI_C is true: PBI_Cond or (!PBI_Cond and BI_Value)
@@ -2498,16 +2500,16 @@
   // If TrueBB and FalseBB are equal, only try to preserve one copy of that
   // successor.
   BasicBlock *KeepEdge1 = TrueBB;
-  BasicBlock *KeepEdge2 = TrueBB != FalseBB ? FalseBB : 0;
+  BasicBlock *KeepEdge2 = TrueBB != FalseBB ? FalseBB : nullptr;
 
   // Then remove the rest.
   for (unsigned I = 0, E = OldTerm->getNumSuccessors(); I != E; ++I) {
     BasicBlock *Succ = OldTerm->getSuccessor(I);
     // Make sure only to keep exactly one copy of each edge.
     if (Succ == KeepEdge1)
-      KeepEdge1 = 0;
+      KeepEdge1 = nullptr;
     else if (Succ == KeepEdge2)
-      KeepEdge2 = 0;
+      KeepEdge2 = nullptr;
     else
       Succ->removePredecessor(OldTerm->getParent());
   }
@@ -2516,7 +2518,7 @@
   Builder.SetCurrentDebugLocation(OldTerm->getDebugLoc());
 
   // Insert an appropriate new terminator.
-  if ((KeepEdge1 == 0) && (KeepEdge2 == 0)) {
+  if (!KeepEdge1 && !KeepEdge2) {
     if (TrueBB == FalseBB)
       // We were only looking for one successor, and it was present.
       // Create an unconditional branch to it.
@@ -2538,7 +2540,7 @@
     // One of the selected values was a successor, but the other wasn't.
     // Insert an unconditional branch to the one that was found;
     // the edge to the one that wasn't must be unreachable.
-    if (KeepEdge1 == 0)
+    if (!KeepEdge1)
       // Only TrueBB was found.
       Builder.CreateBr(TrueBB);
     else
@@ -2639,7 +2641,7 @@
   // 'V' and this block is the default case for the switch.  In this case we can
   // fold the compared value into the switch to simplify things.
   BasicBlock *Pred = BB->getSinglePredecessor();
-  if (Pred == 0 || !isa<SwitchInst>(Pred->getTerminator())) return false;
+  if (!Pred || !isa<SwitchInst>(Pred->getTerminator())) return false;
 
   SwitchInst *SI = cast<SwitchInst>(Pred->getTerminator());
   if (SI->getCondition() != V)
@@ -2681,7 +2683,7 @@
   // the block.
   BasicBlock *SuccBlock = BB->getTerminator()->getSuccessor(0);
   PHINode *PHIUse = dyn_cast<PHINode>(ICI->user_back());
-  if (PHIUse == 0 || PHIUse != &SuccBlock->front() ||
+  if (PHIUse == nullptr || PHIUse != &SuccBlock->front() ||
       isa<PHINode>(++BasicBlock::iterator(PHIUse)))
     return false;
 
@@ -2733,16 +2735,16 @@
 static bool SimplifyBranchOnICmpChain(BranchInst *BI, const DataLayout *DL,
                                       IRBuilder<> &Builder) {
   Instruction *Cond = dyn_cast<Instruction>(BI->getCondition());
-  if (Cond == 0) return false;
+  if (!Cond) return false;
 
 
   // Change br (X == 0 | X == 1), T, F into a switch instruction.
   // If this is a bunch of seteq's or'd together, or if it's a bunch of
   // 'setne's and'ed together, collect them.
-  Value *CompVal = 0;
+  Value *CompVal = nullptr;
   std::vector<ConstantInt*> Values;
   bool TrueWhenEqual = true;
-  Value *ExtraCase = 0;
+  Value *ExtraCase = nullptr;
   unsigned UsedICmps = 0;
 
   if (Cond->getOpcode() == Instruction::Or) {
@@ -2755,7 +2757,7 @@
   }
 
   // If we didn't have a multiply compared value, fail.
-  if (CompVal == 0) return false;
+  if (!CompVal) return false;
 
   // Avoid turning single icmps into a switch.
   if (UsedICmps <= 1)
@@ -3050,7 +3052,7 @@
         // Find the most popular block.
         unsigned MaxPop = 0;
         unsigned MaxIndex = 0;
-        BasicBlock *MaxBlock = 0;
+        BasicBlock *MaxBlock = nullptr;
         for (std::map<BasicBlock*, std::pair<unsigned, unsigned> >::iterator
              I = Popularity.begin(), E = Popularity.end(); I != E; ++I) {
           if (I->second.first > MaxPop ||
@@ -3188,7 +3190,7 @@
   Value *Cond = SI->getCondition();
   unsigned Bits = Cond->getType()->getIntegerBitWidth();
   APInt KnownZero(Bits, 0), KnownOne(Bits, 0);
-  ComputeMaskedBits(Cond, KnownZero, KnownOne);
+  computeKnownBits(Cond, KnownZero, KnownOne);
 
   // Gather dead cases.
   SmallVector<ConstantInt*, 8> DeadCases;
@@ -3241,13 +3243,13 @@
                                               BasicBlock *BB,
                                               int *PhiIndex) {
   if (BB->getFirstNonPHIOrDbg() != BB->getTerminator())
-    return NULL; // BB must be empty to be a candidate for simplification.
+    return nullptr; // BB must be empty to be a candidate for simplification.
   if (!BB->getSinglePredecessor())
-    return NULL; // BB must be dominated by the switch.
+    return nullptr; // BB must be dominated by the switch.
 
   BranchInst *Branch = dyn_cast<BranchInst>(BB->getTerminator());
   if (!Branch || !Branch->isUnconditional())
-    return NULL; // Terminator must be unconditional branch.
+    return nullptr; // Terminator must be unconditional branch.
 
   BasicBlock *Succ = Branch->getSuccessor(0);
 
@@ -3263,7 +3265,7 @@
     return PHI;
   }
 
-  return NULL;
+  return nullptr;
 }
 
 /// ForwardSwitchConditionToPHI - Try to forward the condition of a switch
@@ -3336,12 +3338,12 @@
   if (SelectInst *Select = dyn_cast<SelectInst>(I)) {
     Constant *A = LookupConstant(Select->getCondition(), ConstantPool);
     if (!A)
-      return 0;
+      return nullptr;
     if (A->isAllOnesValue())
       return LookupConstant(Select->getTrueValue(), ConstantPool);
     if (A->isNullValue())
       return LookupConstant(Select->getFalseValue(), ConstantPool);
-    return 0;
+    return nullptr;
   }
 
   SmallVector<Constant *, 4> COps;
@@ -3349,7 +3351,7 @@
     if (Constant *A = LookupConstant(I->getOperand(N), ConstantPool))
       COps.push_back(A);
     else
-      return 0;
+      return nullptr;
   }
 
   if (CmpInst *Cmp = dyn_cast<CmpInst>(I))
@@ -3492,7 +3494,8 @@
              const SmallVectorImpl<std::pair<ConstantInt*, Constant*> >& Values,
                                      Constant *DefaultValue,
                                      const DataLayout *DL)
-    : SingleValue(0), BitMap(0), BitMapElementTy(0), Array(0) {
+    : SingleValue(nullptr), BitMap(nullptr), BitMapElementTy(nullptr),
+      Array(nullptr) {
   assert(Values.size() && "Can't build lookup table without values!");
   assert(TableSize >= Values.size() && "Can't fit values in table!");
 
@@ -3513,7 +3516,7 @@
     TableContents[Idx] = CaseRes;
 
     if (CaseRes != SingleValue)
-      SingleValue = 0;
+      SingleValue = nullptr;
   }
 
   // Fill in any holes in the table with the default result.
@@ -3526,7 +3529,7 @@
     }
 
     if (DefaultValue != SingleValue)
-      SingleValue = 0;
+      SingleValue = nullptr;
   }
 
   // If each element in the table contains the same value, we only need to store
@@ -3696,7 +3699,7 @@
   ConstantInt *MinCaseVal = CI.getCaseValue();
   ConstantInt *MaxCaseVal = CI.getCaseValue();
 
-  BasicBlock *CommonDest = 0;
+  BasicBlock *CommonDest = nullptr;
   typedef SmallVector<std::pair<ConstantInt*, Constant*>, 4> ResultListTy;
   SmallDenseMap<PHINode*, ResultListTy> ResultLists;
   SmallDenseMap<PHINode*, Constant*> DefaultResults;
@@ -3741,8 +3744,8 @@
   SmallVector<std::pair<PHINode*, Constant*>, 4> DefaultResultsList;
   bool HasDefaultResults = false;
   if (TableHasHoles) {
-    HasDefaultResults = GetCaseResults(SI, 0, SI->getDefaultDest(), &CommonDest,
-                                       DefaultResultsList, DL);
+    HasDefaultResults = GetCaseResults(SI, nullptr, SI->getDefaultDest(),
+                                       &CommonDest, DefaultResultsList, DL);
   }
   bool NeedMask = (TableHasHoles && !HasDefaultResults);
   if (NeedMask) {
@@ -4038,8 +4041,8 @@
   // from BI.  We know that the condbr dominates the two blocks, so see if
   // there is any identical code in the "then" and "else" blocks.  If so, we
   // can hoist it up to the branching block.
-  if (BI->getSuccessor(0)->getSinglePredecessor() != 0) {
-    if (BI->getSuccessor(1)->getSinglePredecessor() != 0) {
+  if (BI->getSuccessor(0)->getSinglePredecessor()) {
+    if (BI->getSuccessor(1)->getSinglePredecessor()) {
       if (HoistThenElseCodeToIf(BI))
         return SimplifyCFG(BB, TTI, DL) | true;
     } else {
@@ -4051,7 +4054,7 @@
         if (SpeculativelyExecuteBB(BI, BI->getSuccessor(0)))
           return SimplifyCFG(BB, TTI, DL) | true;
     }
-  } else if (BI->getSuccessor(1)->getSinglePredecessor() != 0) {
+  } else if (BI->getSuccessor(1)->getSinglePredecessor()) {
     // If Successor #0 has multiple preds, we may be able to conditionally
     // execute Successor #1 if it branches to successor #0.
     TerminatorInst *Succ1TI = BI->getSuccessor(1)->getTerminator();
diff --git a/lib/Transforms/Utils/SimplifyIndVar.cpp b/lib/Transforms/Utils/SimplifyIndVar.cpp
index 30f56be..b284e6f 100644
--- a/lib/Transforms/Utils/SimplifyIndVar.cpp
+++ b/lib/Transforms/Utils/SimplifyIndVar.cpp
@@ -13,8 +13,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "indvars"
-
 #include "llvm/Transforms/Utils/SimplifyIndVar.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SmallVector.h"
@@ -34,6 +32,8 @@
 
 using namespace llvm;
 
+#define DEBUG_TYPE "indvars"
+
 STATISTIC(NumElimIdentity, "Number of IV identities eliminated");
 STATISTIC(NumElimOperand,  "Number of IV operands folded into a use");
 STATISTIC(NumElimRem     , "Number of IV remainder operations eliminated");
@@ -56,14 +56,14 @@
 
   public:
     SimplifyIndvar(Loop *Loop, ScalarEvolution *SE, LPPassManager *LPM,
-                   SmallVectorImpl<WeakVH> &Dead, IVUsers *IVU = NULL) :
+                   SmallVectorImpl<WeakVH> &Dead, IVUsers *IVU = nullptr) :
       L(Loop),
       LI(LPM->getAnalysisIfAvailable<LoopInfo>()),
       SE(SE),
       DeadInsts(Dead),
       Changed(false) {
       DataLayoutPass *DLP = LPM->getAnalysisIfAvailable<DataLayoutPass>();
-      DL = DLP ? &DLP->getDataLayout() : 0;
+      DL = DLP ? &DLP->getDataLayout() : nullptr;
       assert(LI && "IV simplification requires LoopInfo");
     }
 
@@ -72,7 +72,7 @@
     /// Iteratively perform simplification on a worklist of users of the
     /// specified induction variable. This is the top-level driver that applies
     /// all simplicitions to users of an IV.
-    void simplifyUsers(PHINode *CurrIV, IVVisitor *V = NULL);
+    void simplifyUsers(PHINode *CurrIV, IVVisitor *V = nullptr);
 
     Value *foldIVUser(Instruction *UseInst, Instruction *IVOperand);
 
@@ -95,25 +95,25 @@
 /// be folded (in case more folding opportunities have been exposed).
 /// Otherwise return null.
 Value *SimplifyIndvar::foldIVUser(Instruction *UseInst, Instruction *IVOperand) {
-  Value *IVSrc = 0;
+  Value *IVSrc = nullptr;
   unsigned OperIdx = 0;
-  const SCEV *FoldedExpr = 0;
+  const SCEV *FoldedExpr = nullptr;
   switch (UseInst->getOpcode()) {
   default:
-    return 0;
+    return nullptr;
   case Instruction::UDiv:
   case Instruction::LShr:
     // We're only interested in the case where we know something about
     // the numerator and have a constant denominator.
     if (IVOperand != UseInst->getOperand(OperIdx) ||
         !isa<ConstantInt>(UseInst->getOperand(1)))
-      return 0;
+      return nullptr;
 
     // Attempt to fold a binary operator with constant operand.
     // e.g. ((I + 1) >> 2) => I >> 2
     if (!isa<BinaryOperator>(IVOperand)
         || !isa<ConstantInt>(IVOperand->getOperand(1)))
-      return 0;
+      return nullptr;
 
     IVSrc = IVOperand->getOperand(0);
     // IVSrc must be the (SCEVable) IV, since the other operand is const.
@@ -124,7 +124,7 @@
       // Get a constant for the divisor. See createSCEV.
       uint32_t BitWidth = cast<IntegerType>(UseInst->getType())->getBitWidth();
       if (D->getValue().uge(BitWidth))
-        return 0;
+        return nullptr;
 
       D = ConstantInt::get(UseInst->getContext(),
                            APInt::getOneBitSet(BitWidth, D->getZExtValue()));
@@ -133,11 +133,11 @@
   }
   // We have something that might fold it's operand. Compare SCEVs.
   if (!SE->isSCEVable(UseInst->getType()))
-    return 0;
+    return nullptr;
 
   // Bypass the operand if SCEV can prove it has no effect.
   if (SE->getSCEV(UseInst) != FoldedExpr)
-    return 0;
+    return nullptr;
 
   DEBUG(dbgs() << "INDVARS: Eliminated IV operand: " << *IVOperand
         << " -> " << *UseInst << '\n');
@@ -283,8 +283,8 @@
     return IVUser;
 
   // Find a branch guarded by the overflow check.
-  BranchInst *Branch = 0;
-  Instruction *AddVal = 0;
+  BranchInst *Branch = nullptr;
+  Instruction *AddVal = nullptr;
   for (User *U : II->users()) {
     if (ExtractValueInst *ExtractInst = dyn_cast<ExtractValueInst>(U)) {
       if (ExtractInst->getNumIndices() != 1)
diff --git a/lib/Transforms/Utils/SimplifyInstructions.cpp b/lib/Transforms/Utils/SimplifyInstructions.cpp
index bbd65f1..33b3637 100644
--- a/lib/Transforms/Utils/SimplifyInstructions.cpp
+++ b/lib/Transforms/Utils/SimplifyInstructions.cpp
@@ -14,7 +14,6 @@
 //
 //===----------------------------------------------------------------------===//
 
-#define DEBUG_TYPE "instsimplify"
 #include "llvm/Transforms/Scalar.h"
 #include "llvm/ADT/DepthFirstIterator.h"
 #include "llvm/ADT/SmallPtrSet.h"
@@ -29,6 +28,8 @@
 #include "llvm/Transforms/Utils/Local.h"
 using namespace llvm;
 
+#define DEBUG_TYPE "instsimplify"
+
 STATISTIC(NumSimplified, "Number of redundant instructions removed");
 
 namespace {
@@ -47,17 +48,18 @@
     bool runOnFunction(Function &F) override {
       const DominatorTreeWrapperPass *DTWP =
           getAnalysisIfAvailable<DominatorTreeWrapperPass>();
-      const DominatorTree *DT = DTWP ? &DTWP->getDomTree() : 0;
+      const DominatorTree *DT = DTWP ? &DTWP->getDomTree() : nullptr;
       DataLayoutPass *DLP = getAnalysisIfAvailable<DataLayoutPass>();
-      const DataLayout *DL = DLP ? &DLP->getDataLayout() : 0;
+      const DataLayout *DL = DLP ? &DLP->getDataLayout() : nullptr;
       const TargetLibraryInfo *TLI = &getAnalysis<TargetLibraryInfo>();
       SmallPtrSet<const Instruction*, 8> S1, S2, *ToSimplify = &S1, *Next = &S2;
       bool Changed = false;
 
       do {
-        for (df_iterator<BasicBlock*> DI = df_begin(&F.getEntryBlock()),
-             DE = df_end(&F.getEntryBlock()); DI != DE; ++DI)
-          for (BasicBlock::iterator BI = DI->begin(), BE = DI->end(); BI != BE;) {
+        for (BasicBlock *BB : depth_first(&F.getEntryBlock()))
+          // Here be subtlety: the iterator must be incremented before the loop
+          // body (not sure why), so a range-for loop won't work here.
+          for (BasicBlock::iterator BI = BB->begin(), BE = BB->end(); BI != BE;) {
             Instruction *I = BI++;
             // The first time through the loop ToSimplify is empty and we try to
             // simplify all instructions.  On later iterations ToSimplify is not
@@ -74,7 +76,15 @@
                 ++NumSimplified;
                 Changed = true;
               }
-            Changed |= RecursivelyDeleteTriviallyDeadInstructions(I, TLI);
+            bool res = RecursivelyDeleteTriviallyDeadInstructions(I, TLI);
+            if (res)  {
+              // RecursivelyDeleteTriviallyDeadInstruction can remove
+              // more than one instruction, so simply incrementing the
+              // iterator does not work. When instructions get deleted
+              // re-iterate instead.
+              BI = BB->begin(); BE = BB->end();
+              Changed |= res;
+            }
           }
 
         // Place the list of instructions to simplify on the next loop iteration
diff --git a/lib/Transforms/Utils/SimplifyLibCalls.cpp b/lib/Transforms/Utils/SimplifyLibCalls.cpp
index b5bc391..3b61bb5 100644
--- a/lib/Transforms/Utils/SimplifyLibCalls.cpp
+++ b/lib/Transforms/Utils/SimplifyLibCalls.cpp
@@ -20,6 +20,7 @@
 #include "llvm/ADT/Triple.h"
 #include "llvm/Analysis/ValueTracking.h"
 #include "llvm/IR/DataLayout.h"
+#include "llvm/IR/DiagnosticInfo.h"
 #include "llvm/IR/Function.h"
 #include "llvm/IR/IRBuilder.h"
 #include "llvm/IR/IntrinsicInst.h"
@@ -75,7 +76,7 @@
 
     // We never change the calling convention.
     if (!ignoreCallingConv() && CI->getCallingConv() != llvm::CallingConv::C)
-      return NULL;
+      return nullptr;
 
     return callOptimizer(CI->getCalledFunction(), CI, B);
   }
@@ -186,14 +187,14 @@
         !FT->getParamType(1)->isPointerTy() ||
         FT->getParamType(2) != DL->getIntPtrType(Context) ||
         FT->getParamType(3) != DL->getIntPtrType(Context))
-      return 0;
+      return nullptr;
 
     if (isFoldable(3, 2, false)) {
       B.CreateMemCpy(CI->getArgOperand(0), CI->getArgOperand(1),
                      CI->getArgOperand(2), 1);
       return CI->getArgOperand(0);
     }
-    return 0;
+    return nullptr;
   }
 };
 
@@ -210,14 +211,14 @@
         !FT->getParamType(1)->isPointerTy() ||
         FT->getParamType(2) != DL->getIntPtrType(Context) ||
         FT->getParamType(3) != DL->getIntPtrType(Context))
-      return 0;
+      return nullptr;
 
     if (isFoldable(3, 2, false)) {
       B.CreateMemMove(CI->getArgOperand(0), CI->getArgOperand(1),
                       CI->getArgOperand(2), 1);
       return CI->getArgOperand(0);
     }
-    return 0;
+    return nullptr;
   }
 };
 
@@ -234,7 +235,7 @@
         !FT->getParamType(1)->isIntegerTy() ||
         FT->getParamType(2) != DL->getIntPtrType(Context) ||
         FT->getParamType(3) != DL->getIntPtrType(Context))
-      return 0;
+      return nullptr;
 
     if (isFoldable(3, 2, false)) {
       Value *Val = B.CreateIntCast(CI->getArgOperand(1), B.getInt8Ty(),
@@ -242,7 +243,7 @@
       B.CreateMemSet(CI->getArgOperand(0), Val, CI->getArgOperand(2), 1);
       return CI->getArgOperand(0);
     }
-    return 0;
+    return nullptr;
   }
 };
 
@@ -260,7 +261,7 @@
         FT->getParamType(0) != FT->getParamType(1) ||
         FT->getParamType(0) != Type::getInt8PtrTy(Context) ||
         FT->getParamType(2) != DL->getIntPtrType(Context))
-      return 0;
+      return nullptr;
 
     Value *Dst = CI->getArgOperand(0), *Src = CI->getArgOperand(1);
     if (Dst == Src)      // __strcpy_chk(x,x)  -> x
@@ -277,10 +278,10 @@
     } else {
       // Maybe we can stil fold __strcpy_chk to __memcpy_chk.
       uint64_t Len = GetStringLength(Src);
-      if (Len == 0) return 0;
+      if (Len == 0) return nullptr;
 
       // This optimization require DataLayout.
-      if (!DL) return 0;
+      if (!DL) return nullptr;
 
       Value *Ret =
 	EmitMemCpyChk(Dst, Src,
@@ -288,7 +289,7 @@
                       CI->getArgOperand(2), B, DL, TLI);
       return Ret;
     }
-    return 0;
+    return nullptr;
   }
 };
 
@@ -306,12 +307,12 @@
         FT->getParamType(0) != FT->getParamType(1) ||
         FT->getParamType(0) != Type::getInt8PtrTy(Context) ||
         FT->getParamType(2) != DL->getIntPtrType(FT->getParamType(0)))
-      return 0;
+      return nullptr;
 
     Value *Dst = CI->getArgOperand(0), *Src = CI->getArgOperand(1);
     if (Dst == Src) {  // stpcpy(x,x)  -> x+strlen(x)
       Value *StrLen = EmitStrLen(Src, B, DL, TLI);
-      return StrLen ? B.CreateInBoundsGEP(Dst, StrLen) : 0;
+      return StrLen ? B.CreateInBoundsGEP(Dst, StrLen) : nullptr;
     }
 
     // If a) we don't have any length information, or b) we know this will
@@ -325,10 +326,10 @@
     } else {
       // Maybe we can stil fold __stpcpy_chk to __memcpy_chk.
       uint64_t Len = GetStringLength(Src);
-      if (Len == 0) return 0;
+      if (Len == 0) return nullptr;
 
       // This optimization require DataLayout.
-      if (!DL) return 0;
+      if (!DL) return nullptr;
 
       Type *PT = FT->getParamType(0);
       Value *LenV = ConstantInt::get(DL->getIntPtrType(PT), Len);
@@ -336,10 +337,10 @@
                                   ConstantInt::get(DL->getIntPtrType(PT),
                                                    Len - 1));
       if (!EmitMemCpyChk(Dst, Src, LenV, CI->getArgOperand(2), B, DL, TLI))
-        return 0;
+        return nullptr;
       return DstEnd;
     }
-    return 0;
+    return nullptr;
   }
 };
 
@@ -357,7 +358,7 @@
         FT->getParamType(0) != Type::getInt8PtrTy(Context) ||
         !FT->getParamType(2)->isIntegerTy() ||
         FT->getParamType(3) != DL->getIntPtrType(Context))
-      return 0;
+      return nullptr;
 
     if (isFoldable(3, 2, false)) {
       Value *Ret = EmitStrNCpy(CI->getArgOperand(0), CI->getArgOperand(1),
@@ -365,7 +366,7 @@
                                Name.substr(2, 7));
       return Ret;
     }
-    return 0;
+    return nullptr;
   }
 };
 
@@ -382,7 +383,7 @@
         FT->getReturnType() != B.getInt8PtrTy() ||
         FT->getParamType(0) != FT->getReturnType() ||
         FT->getParamType(1) != FT->getReturnType())
-      return 0;
+      return nullptr;
 
     // Extract some information from the instruction
     Value *Dst = CI->getArgOperand(0);
@@ -390,7 +391,7 @@
 
     // See if we can get the length of the input string.
     uint64_t Len = GetStringLength(Src);
-    if (Len == 0) return 0;
+    if (Len == 0) return nullptr;
     --Len;  // Unbias length.
 
     // Handle the simple, do-nothing case: strcat(x, "") -> x
@@ -398,7 +399,7 @@
       return Dst;
 
     // These optimizations require DataLayout.
-    if (!DL) return 0;
+    if (!DL) return nullptr;
 
     return emitStrLenMemCpy(Src, Dst, Len, B);
   }
@@ -409,7 +410,7 @@
     // memory is to be moved to. We just generate a call to strlen.
     Value *DstLen = EmitStrLen(Dst, B, DL, TLI);
     if (!DstLen)
-      return 0;
+      return nullptr;
 
     // Now that we have the destination's length, we must index into the
     // destination's pointer to get the actual memcpy destination (end of
@@ -434,7 +435,7 @@
         FT->getParamType(0) != FT->getReturnType() ||
         FT->getParamType(1) != FT->getReturnType() ||
         !FT->getParamType(2)->isIntegerTy())
-      return 0;
+      return nullptr;
 
     // Extract some information from the instruction
     Value *Dst = CI->getArgOperand(0);
@@ -445,11 +446,11 @@
     if (ConstantInt *LengthArg = dyn_cast<ConstantInt>(CI->getArgOperand(2)))
       Len = LengthArg->getZExtValue();
     else
-      return 0;
+      return nullptr;
 
     // See if we can get the length of the input string.
     uint64_t SrcLen = GetStringLength(Src);
-    if (SrcLen == 0) return 0;
+    if (SrcLen == 0) return nullptr;
     --SrcLen;  // Unbias length.
 
     // Handle the simple, do-nothing cases:
@@ -458,10 +459,10 @@
     if (SrcLen == 0 || Len == 0) return Dst;
 
     // These optimizations require DataLayout.
-    if (!DL) return 0;
+    if (!DL) return nullptr;
 
     // We don't optimize this case
-    if (Len < SrcLen) return 0;
+    if (Len < SrcLen) return nullptr;
 
     // strncat(x, s, c) -> strcat(x, s)
     // s is constant so the strcat can be optimized further
@@ -478,20 +479,20 @@
         FT->getReturnType() != B.getInt8PtrTy() ||
         FT->getParamType(0) != FT->getReturnType() ||
         !FT->getParamType(1)->isIntegerTy(32))
-      return 0;
+      return nullptr;
 
     Value *SrcStr = CI->getArgOperand(0);
 
     // If the second operand is non-constant, see if we can compute the length
     // of the input string and turn this into memchr.
     ConstantInt *CharC = dyn_cast<ConstantInt>(CI->getArgOperand(1));
-    if (CharC == 0) {
+    if (!CharC) {
       // These optimizations require DataLayout.
-      if (!DL) return 0;
+      if (!DL) return nullptr;
 
       uint64_t Len = GetStringLength(SrcStr);
       if (Len == 0 || !FT->getParamType(1)->isIntegerTy(32))// memchr needs i32.
-        return 0;
+        return nullptr;
 
       return EmitMemChr(SrcStr, CI->getArgOperand(1), // include nul.
                         ConstantInt::get(DL->getIntPtrType(*Context), Len),
@@ -504,7 +505,7 @@
     if (!getConstantStringInfo(SrcStr, Str)) {
       if (DL && CharC->isZero()) // strchr(p, 0) -> p + strlen(p)
         return B.CreateGEP(SrcStr, EmitStrLen(SrcStr, B, DL, TLI), "strchr");
-      return 0;
+      return nullptr;
     }
 
     // Compute the offset, make sure to handle the case when we're searching for
@@ -528,21 +529,21 @@
         FT->getReturnType() != B.getInt8PtrTy() ||
         FT->getParamType(0) != FT->getReturnType() ||
         !FT->getParamType(1)->isIntegerTy(32))
-      return 0;
+      return nullptr;
 
     Value *SrcStr = CI->getArgOperand(0);
     ConstantInt *CharC = dyn_cast<ConstantInt>(CI->getArgOperand(1));
 
     // Cannot fold anything if we're not looking for a constant.
     if (!CharC)
-      return 0;
+      return nullptr;
 
     StringRef Str;
     if (!getConstantStringInfo(SrcStr, Str)) {
       // strrchr(s, 0) -> strchr(s, 0)
       if (DL && CharC->isZero())
         return EmitStrChr(SrcStr, '\0', B, DL, TLI);
-      return 0;
+      return nullptr;
     }
 
     // Compute the offset.
@@ -565,7 +566,7 @@
         !FT->getReturnType()->isIntegerTy(32) ||
         FT->getParamType(0) != FT->getParamType(1) ||
         FT->getParamType(0) != B.getInt8PtrTy())
-      return 0;
+      return nullptr;
 
     Value *Str1P = CI->getArgOperand(0), *Str2P = CI->getArgOperand(1);
     if (Str1P == Str2P)      // strcmp(x,x)  -> 0
@@ -591,14 +592,14 @@
     uint64_t Len2 = GetStringLength(Str2P);
     if (Len1 && Len2) {
       // These optimizations require DataLayout.
-      if (!DL) return 0;
+      if (!DL) return nullptr;
 
       return EmitMemCmp(Str1P, Str2P,
                         ConstantInt::get(DL->getIntPtrType(*Context),
                         std::min(Len1, Len2)), B, DL, TLI);
     }
 
-    return 0;
+    return nullptr;
   }
 };
 
@@ -612,7 +613,7 @@
         FT->getParamType(0) != FT->getParamType(1) ||
         FT->getParamType(0) != B.getInt8PtrTy() ||
         !FT->getParamType(2)->isIntegerTy())
-      return 0;
+      return nullptr;
 
     Value *Str1P = CI->getArgOperand(0), *Str2P = CI->getArgOperand(1);
     if (Str1P == Str2P)      // strncmp(x,x,n)  -> 0
@@ -623,7 +624,7 @@
     if (ConstantInt *LengthArg = dyn_cast<ConstantInt>(CI->getArgOperand(2)))
       Length = LengthArg->getZExtValue();
     else
-      return 0;
+      return nullptr;
 
     if (Length == 0) // strncmp(x,y,0)   -> 0
       return ConstantInt::get(CI->getType(), 0);
@@ -649,7 +650,7 @@
     if (HasStr2 && Str2.empty())  // strncmp(x, "", n) -> *x
       return B.CreateZExt(B.CreateLoad(Str1P, "strcmpload"), CI->getType());
 
-    return 0;
+    return nullptr;
   }
 };
 
@@ -662,18 +663,18 @@
         FT->getReturnType() != FT->getParamType(0) ||
         FT->getParamType(0) != FT->getParamType(1) ||
         FT->getParamType(0) != B.getInt8PtrTy())
-      return 0;
+      return nullptr;
 
     Value *Dst = CI->getArgOperand(0), *Src = CI->getArgOperand(1);
     if (Dst == Src)      // strcpy(x,x)  -> x
       return Src;
 
     // These optimizations require DataLayout.
-    if (!DL) return 0;
+    if (!DL) return nullptr;
 
     // See if we can get the length of the input string.
     uint64_t Len = GetStringLength(Src);
-    if (Len == 0) return 0;
+    if (Len == 0) return nullptr;
 
     // We have enough information to now generate the memcpy call to do the
     // copy for us.  Make a memcpy to copy the nul byte with align = 1.
@@ -692,20 +693,20 @@
         FT->getReturnType() != FT->getParamType(0) ||
         FT->getParamType(0) != FT->getParamType(1) ||
         FT->getParamType(0) != B.getInt8PtrTy())
-      return 0;
+      return nullptr;
 
     // These optimizations require DataLayout.
-    if (!DL) return 0;
+    if (!DL) return nullptr;
 
     Value *Dst = CI->getArgOperand(0), *Src = CI->getArgOperand(1);
     if (Dst == Src) {  // stpcpy(x,x)  -> x+strlen(x)
       Value *StrLen = EmitStrLen(Src, B, DL, TLI);
-      return StrLen ? B.CreateInBoundsGEP(Dst, StrLen) : 0;
+      return StrLen ? B.CreateInBoundsGEP(Dst, StrLen) : nullptr;
     }
 
     // See if we can get the length of the input string.
     uint64_t Len = GetStringLength(Src);
-    if (Len == 0) return 0;
+    if (Len == 0) return nullptr;
 
     Type *PT = FT->getParamType(0);
     Value *LenV = ConstantInt::get(DL->getIntPtrType(PT), Len);
@@ -728,7 +729,7 @@
         FT->getParamType(0) != FT->getParamType(1) ||
         FT->getParamType(0) != B.getInt8PtrTy() ||
         !FT->getParamType(2)->isIntegerTy())
-      return 0;
+      return nullptr;
 
     Value *Dst = CI->getArgOperand(0);
     Value *Src = CI->getArgOperand(1);
@@ -736,7 +737,7 @@
 
     // See if we can get the length of the input string.
     uint64_t SrcLen = GetStringLength(Src);
-    if (SrcLen == 0) return 0;
+    if (SrcLen == 0) return nullptr;
     --SrcLen;
 
     if (SrcLen == 0) {
@@ -749,15 +750,15 @@
     if (ConstantInt *LengthArg = dyn_cast<ConstantInt>(LenOp))
       Len = LengthArg->getZExtValue();
     else
-      return 0;
+      return nullptr;
 
     if (Len == 0) return Dst; // strncpy(x, y, 0) -> x
 
     // These optimizations require DataLayout.
-    if (!DL) return 0;
+    if (!DL) return nullptr;
 
     // Let strncpy handle the zero padding
-    if (Len > SrcLen+1) return 0;
+    if (Len > SrcLen+1) return nullptr;
 
     Type *PT = FT->getParamType(0);
     // strncpy(x, s, c) -> memcpy(x, s, c, 1) [s and c are constant]
@@ -776,7 +777,7 @@
     if (FT->getNumParams() != 1 ||
         FT->getParamType(0) != B.getInt8PtrTy() ||
         !FT->getReturnType()->isIntegerTy())
-      return 0;
+      return nullptr;
 
     Value *Src = CI->getArgOperand(0);
 
@@ -784,11 +785,26 @@
     if (uint64_t Len = GetStringLength(Src))
       return ConstantInt::get(CI->getType(), Len-1);
 
+    // strlen(x?"foo":"bars") --> x ? 3 : 4
+    if (SelectInst *SI = dyn_cast<SelectInst>(Src)) {
+      uint64_t LenTrue = GetStringLength(SI->getTrueValue());
+      uint64_t LenFalse = GetStringLength(SI->getFalseValue());
+      if (LenTrue && LenFalse) {
+        emitOptimizationRemark(*Context, "simplify-libcalls", *Caller,
+                               SI->getDebugLoc(),
+                               "folded strlen(select) to select of constants");
+        return B.CreateSelect(SI->getCondition(),
+                              ConstantInt::get(CI->getType(), LenTrue-1),
+                              ConstantInt::get(CI->getType(), LenFalse-1));
+      }
+    }
+
     // strlen(x) != 0 --> *x != 0
     // strlen(x) == 0 --> *x == 0
     if (isOnlyUsedInZeroEqualityComparison(CI))
       return B.CreateZExt(B.CreateLoad(Src, "strlenfirst"), CI->getType());
-    return 0;
+
+    return nullptr;
   }
 };
 
@@ -800,7 +816,7 @@
         FT->getParamType(0) != B.getInt8PtrTy() ||
         FT->getParamType(1) != FT->getParamType(0) ||
         FT->getReturnType() != FT->getParamType(0))
-      return 0;
+      return nullptr;
 
     StringRef S1, S2;
     bool HasS1 = getConstantStringInfo(CI->getArgOperand(0), S1);
@@ -824,7 +840,7 @@
     if (DL && HasS2 && S2.size() == 1)
       return EmitStrChr(CI->getArgOperand(0), S2[0], B, DL, TLI);
 
-    return 0;
+    return nullptr;
   }
 };
 
@@ -835,7 +851,7 @@
     if ((FT->getNumParams() != 2 && FT->getNumParams() != 3) ||
         !FT->getParamType(0)->isPointerTy() ||
         !FT->getParamType(1)->isPointerTy())
-      return 0;
+      return nullptr;
 
     Value *EndPtr = CI->getArgOperand(1);
     if (isa<ConstantPointerNull>(EndPtr)) {
@@ -844,7 +860,7 @@
       CI->addAttribute(1, Attribute::NoCapture);
     }
 
-    return 0;
+    return nullptr;
   }
 };
 
@@ -856,7 +872,7 @@
         FT->getParamType(0) != B.getInt8PtrTy() ||
         FT->getParamType(1) != FT->getParamType(0) ||
         !FT->getReturnType()->isIntegerTy())
-      return 0;
+      return nullptr;
 
     StringRef S1, S2;
     bool HasS1 = getConstantStringInfo(CI->getArgOperand(0), S1);
@@ -874,7 +890,7 @@
       return ConstantInt::get(CI->getType(), Pos);
     }
 
-    return 0;
+    return nullptr;
   }
 };
 
@@ -886,7 +902,7 @@
         FT->getParamType(0) != B.getInt8PtrTy() ||
         FT->getParamType(1) != FT->getParamType(0) ||
         !FT->getReturnType()->isIntegerTy())
-      return 0;
+      return nullptr;
 
     StringRef S1, S2;
     bool HasS1 = getConstantStringInfo(CI->getArgOperand(0), S1);
@@ -907,7 +923,7 @@
     if (DL && HasS2 && S2.empty())
       return EmitStrLen(CI->getArgOperand(0), B, DL, TLI);
 
-    return 0;
+    return nullptr;
   }
 };
 
@@ -919,7 +935,7 @@
         !FT->getParamType(0)->isPointerTy() ||
         !FT->getParamType(1)->isPointerTy() ||
         !FT->getReturnType()->isPointerTy())
-      return 0;
+      return nullptr;
 
     // fold strstr(x, x) -> x.
     if (CI->getArgOperand(0) == CI->getArgOperand(1))
@@ -929,11 +945,11 @@
     if (DL && isOnlyUsedInEqualityComparison(CI, CI->getArgOperand(0))) {
       Value *StrLen = EmitStrLen(CI->getArgOperand(1), B, DL, TLI);
       if (!StrLen)
-        return 0;
+        return nullptr;
       Value *StrNCmp = EmitStrNCmp(CI->getArgOperand(0), CI->getArgOperand(1),
                                    StrLen, B, DL, TLI);
       if (!StrNCmp)
-        return 0;
+        return nullptr;
       for (auto UI = CI->user_begin(), UE = CI->user_end(); UI != UE;) {
         ICmpInst *Old = cast<ICmpInst>(*UI++);
         Value *Cmp = B.CreateICmp(Old->getPredicate(), StrNCmp,
@@ -969,9 +985,9 @@
     // fold strstr(x, "y") -> strchr(x, 'y').
     if (HasStr2 && ToFindStr.size() == 1) {
       Value *StrChr= EmitStrChr(CI->getArgOperand(0), ToFindStr[0], B, DL, TLI);
-      return StrChr ? B.CreateBitCast(StrChr, CI->getType()) : 0;
+      return StrChr ? B.CreateBitCast(StrChr, CI->getType()) : nullptr;
     }
-    return 0;
+    return nullptr;
   }
 };
 
@@ -982,7 +998,7 @@
     if (FT->getNumParams() != 3 || !FT->getParamType(0)->isPointerTy() ||
         !FT->getParamType(1)->isPointerTy() ||
         !FT->getReturnType()->isIntegerTy(32))
-      return 0;
+      return nullptr;
 
     Value *LHS = CI->getArgOperand(0), *RHS = CI->getArgOperand(1);
 
@@ -991,7 +1007,7 @@
 
     // Make sure we have a constant length.
     ConstantInt *LenC = dyn_cast<ConstantInt>(CI->getArgOperand(2));
-    if (!LenC) return 0;
+    if (!LenC) return nullptr;
     uint64_t Len = LenC->getZExtValue();
 
     if (Len == 0) // memcmp(s1,s2,0) -> 0
@@ -1012,7 +1028,7 @@
         getConstantStringInfo(RHS, RHSStr)) {
       // Make sure we're not reading out-of-bounds memory.
       if (Len > LHSStr.size() || Len > RHSStr.size())
-        return 0;
+        return nullptr;
       // Fold the memcmp and normalize the result.  This way we get consistent
       // results across multiple platforms.
       uint64_t Ret = 0;
@@ -1024,7 +1040,7 @@
       return ConstantInt::get(CI->getType(), Ret);
     }
 
-    return 0;
+    return nullptr;
   }
 };
 
@@ -1032,14 +1048,14 @@
   Value *callOptimizer(Function *Callee, CallInst *CI,
                        IRBuilder<> &B) override {
     // These optimizations require DataLayout.
-    if (!DL) return 0;
+    if (!DL) return nullptr;
 
     FunctionType *FT = Callee->getFunctionType();
     if (FT->getNumParams() != 3 || FT->getReturnType() != FT->getParamType(0) ||
         !FT->getParamType(0)->isPointerTy() ||
         !FT->getParamType(1)->isPointerTy() ||
         FT->getParamType(2) != DL->getIntPtrType(*Context))
-      return 0;
+      return nullptr;
 
     // memcpy(x, y, n) -> llvm.memcpy(x, y, n, 1)
     B.CreateMemCpy(CI->getArgOperand(0), CI->getArgOperand(1),
@@ -1052,14 +1068,14 @@
   Value *callOptimizer(Function *Callee, CallInst *CI,
                        IRBuilder<> &B) override {
     // These optimizations require DataLayout.
-    if (!DL) return 0;
+    if (!DL) return nullptr;
 
     FunctionType *FT = Callee->getFunctionType();
     if (FT->getNumParams() != 3 || FT->getReturnType() != FT->getParamType(0) ||
         !FT->getParamType(0)->isPointerTy() ||
         !FT->getParamType(1)->isPointerTy() ||
         FT->getParamType(2) != DL->getIntPtrType(*Context))
-      return 0;
+      return nullptr;
 
     // memmove(x, y, n) -> llvm.memmove(x, y, n, 1)
     B.CreateMemMove(CI->getArgOperand(0), CI->getArgOperand(1),
@@ -1072,14 +1088,14 @@
   Value *callOptimizer(Function *Callee, CallInst *CI,
                        IRBuilder<> &B) override {
     // These optimizations require DataLayout.
-    if (!DL) return 0;
+    if (!DL) return nullptr;
 
     FunctionType *FT = Callee->getFunctionType();
     if (FT->getNumParams() != 3 || FT->getReturnType() != FT->getParamType(0) ||
         !FT->getParamType(0)->isPointerTy() ||
         !FT->getParamType(1)->isIntegerTy() ||
         FT->getParamType(2) != DL->getIntPtrType(FT->getParamType(0)))
-      return 0;
+      return nullptr;
 
     // memset(p, v, n) -> llvm.memset(p, v, n, 1)
     Value *Val = B.CreateIntCast(CI->getArgOperand(1), B.getInt8Ty(), false);
@@ -1103,21 +1119,21 @@
     FunctionType *FT = Callee->getFunctionType();
     if (FT->getNumParams() != 1 || !FT->getReturnType()->isDoubleTy() ||
         !FT->getParamType(0)->isDoubleTy())
-      return 0;
+      return nullptr;
 
     if (CheckRetType) {
       // Check if all the uses for function like 'sin' are converted to float.
       for (User *U : CI->users()) {
         FPTruncInst *Cast = dyn_cast<FPTruncInst>(U);
-        if (Cast == 0 || !Cast->getType()->isFloatTy())
-          return 0;
+        if (!Cast || !Cast->getType()->isFloatTy())
+          return nullptr;
       }
     }
 
     // If this is something like 'floor((double)floatval)', convert to floorf.
     FPExtInst *Cast = dyn_cast<FPExtInst>(CI->getArgOperand(0));
-    if (Cast == 0 || !Cast->getOperand(0)->getType()->isFloatTy())
-      return 0;
+    if (!Cast || !Cast->getOperand(0)->getType()->isFloatTy())
+      return nullptr;
 
     // floor((double)floatval) -> (double)floorf(floatval)
     Value *V = Cast->getOperand(0);
@@ -1138,15 +1154,15 @@
     if (FT->getNumParams() != 2 || FT->getReturnType() != FT->getParamType(0) ||
         FT->getParamType(0) != FT->getParamType(1) ||
         !FT->getParamType(0)->isFloatingPointTy())
-      return 0;
+      return nullptr;
 
     if (CheckRetType) {
       // Check if all the uses for function like 'fmin/fmax' are converted to
       // float.
       for (User *U : CI->users()) {
         FPTruncInst *Cast = dyn_cast<FPTruncInst>(U);
-        if (Cast == 0 || !Cast->getType()->isFloatTy())
-          return 0;
+        if (!Cast || !Cast->getType()->isFloatTy())
+          return nullptr;
       }
     }
 
@@ -1154,13 +1170,13 @@
     // we convert it to fminf.
     FPExtInst *Cast1 = dyn_cast<FPExtInst>(CI->getArgOperand(0));
     FPExtInst *Cast2 = dyn_cast<FPExtInst>(CI->getArgOperand(1));
-    if (Cast1 == 0 || !Cast1->getOperand(0)->getType()->isFloatTy() ||
-        Cast2 == 0 || !Cast2->getOperand(0)->getType()->isFloatTy())
-      return 0;
+    if (!Cast1 || !Cast1->getOperand(0)->getType()->isFloatTy() ||
+        !Cast2 || !Cast2->getOperand(0)->getType()->isFloatTy())
+      return nullptr;
 
     // fmin((double)floatval1, (double)floatval2)
     //                      -> (double)fmin(floatval1, floatval2)
-    Value *V = NULL;
+    Value *V = nullptr;
     Value *V1 = Cast1->getOperand(0);
     Value *V2 = Cast2->getOperand(0);
     V = EmitBinaryFloatFnCall(V1, V2, Callee->getName(), B,
@@ -1180,7 +1196,7 @@
   CosOpt(bool UnsafeFPShrink) : UnsafeFPLibCallOptimization(UnsafeFPShrink) {}
   Value *callOptimizer(Function *Callee, CallInst *CI,
                        IRBuilder<> &B) override {
-    Value *Ret = NULL;
+    Value *Ret = nullptr;
     if (UnsafeFPShrink && Callee->getName() == "cos" &&
         TLI->has(LibFunc::cosf)) {
       UnaryDoubleFPOpt UnsafeUnaryDoubleFP(true);
@@ -1208,7 +1224,7 @@
   PowOpt(bool UnsafeFPShrink) : UnsafeFPLibCallOptimization(UnsafeFPShrink) {}
   Value *callOptimizer(Function *Callee, CallInst *CI,
                        IRBuilder<> &B) override {
-    Value *Ret = NULL;
+    Value *Ret = nullptr;
     if (UnsafeFPShrink && Callee->getName() == "pow" &&
         TLI->has(LibFunc::powf)) {
       UnaryDoubleFPOpt UnsafeUnaryDoubleFP(true);
@@ -1242,7 +1258,7 @@
     }
 
     ConstantFP *Op2C = dyn_cast<ConstantFP>(Op2);
-    if (Op2C == 0) return Ret;
+    if (!Op2C) return Ret;
 
     if (Op2C->getValueAPF().isZero())  // pow(x, 0.0) -> 1.0
       return ConstantFP::get(CI->getType(), 1.0);
@@ -1275,7 +1291,7 @@
     if (Op2C->isExactlyValue(-1.0)) // pow(x, -1.0) -> 1.0/x
       return B.CreateFDiv(ConstantFP::get(CI->getType(), 1.0),
                           Op1, "powrecip");
-    return 0;
+    return nullptr;
   }
 };
 
@@ -1283,7 +1299,7 @@
   Exp2Opt(bool UnsafeFPShrink) : UnsafeFPLibCallOptimization(UnsafeFPShrink) {}
   Value *callOptimizer(Function *Callee, CallInst *CI,
                        IRBuilder<> &B) override {
-    Value *Ret = NULL;
+    Value *Ret = nullptr;
     if (UnsafeFPShrink && Callee->getName() == "exp2" &&
         TLI->has(LibFunc::exp2f)) {
       UnaryDoubleFPOpt UnsafeUnaryDoubleFP(true);
@@ -1307,7 +1323,7 @@
       LdExp = LibFunc::ldexp;
 
     if (TLI->has(LdExp)) {
-      Value *LdExpArg = 0;
+      Value *LdExpArg = nullptr;
       if (SIToFPInst *OpC = dyn_cast<SIToFPInst>(Op)) {
         if (OpC->getOperand(0)->getType()->getPrimitiveSizeInBits() <= 32)
           LdExpArg = B.CreateSExt(OpC->getOperand(0), B.getInt32Ty());
@@ -1344,7 +1360,7 @@
     // Make sure the prototype is as expected, otherwise the rest of the
     // function is probably invalid and likely to abort.
     if (!isTrigLibCall(CI))
-      return 0;
+      return nullptr;
 
     Value *Arg = CI->getArgOperand(0);
     SmallVector<CallInst *, 1> SinCalls;
@@ -1362,7 +1378,7 @@
 
     // It's only worthwhile if both sinpi and cospi are actually used.
     if (SinCosCalls.empty() && (SinCalls.empty() || CosCalls.empty()))
-      return 0;
+      return nullptr;
 
     Value *Sin, *Cos, *SinCos;
     insertSinCosCall(B, CI->getCalledFunction(), Arg, IsFloat, Sin, Cos,
@@ -1372,7 +1388,7 @@
     replaceTrigInsts(CosCalls, Cos);
     replaceTrigInsts(SinCosCalls, SinCos);
 
-    return 0;
+    return nullptr;
   }
 
   bool isTrigLibCall(CallInst *CI) {
@@ -1498,7 +1514,7 @@
     if (FT->getNumParams() != 1 ||
         !FT->getReturnType()->isIntegerTy(32) ||
         !FT->getParamType(0)->isIntegerTy())
-      return 0;
+      return nullptr;
 
     Value *Op = CI->getArgOperand(0);
 
@@ -1531,7 +1547,7 @@
     // We require integer(integer) where the types agree.
     if (FT->getNumParams() != 1 || !FT->getReturnType()->isIntegerTy() ||
         FT->getParamType(0) != FT->getReturnType())
-      return 0;
+      return nullptr;
 
     // abs(x) -> x >s -1 ? x : -x
     Value *Op = CI->getArgOperand(0);
@@ -1549,7 +1565,7 @@
     // We require integer(i32)
     if (FT->getNumParams() != 1 || !FT->getReturnType()->isIntegerTy() ||
         !FT->getParamType(0)->isIntegerTy(32))
-      return 0;
+      return nullptr;
 
     // isdigit(c) -> (c-'0') <u 10
     Value *Op = CI->getArgOperand(0);
@@ -1566,7 +1582,7 @@
     // We require integer(i32)
     if (FT->getNumParams() != 1 || !FT->getReturnType()->isIntegerTy() ||
         !FT->getParamType(0)->isIntegerTy(32))
-      return 0;
+      return nullptr;
 
     // isascii(c) -> c <u 128
     Value *Op = CI->getArgOperand(0);
@@ -1582,7 +1598,7 @@
     // We require i32(i32)
     if (FT->getNumParams() != 1 || FT->getReturnType() != FT->getParamType(0) ||
         !FT->getParamType(0)->isIntegerTy(32))
-      return 0;
+      return nullptr;
 
     // toascii(c) -> c & 0x7f
     return B.CreateAnd(CI->getArgOperand(0),
@@ -1612,7 +1628,7 @@
       CI->addAttribute(AttributeSet::FunctionIndex, Attribute::Cold);
     }
 
-    return 0;
+    return nullptr;
   }
 
 protected:
@@ -1649,7 +1665,7 @@
     // Check for a fixed format string.
     StringRef FormatStr;
     if (!getConstantStringInfo(CI->getArgOperand(0), FormatStr))
-      return 0;
+      return nullptr;
 
     // Empty format string -> noop.
     if (FormatStr.empty())  // Tolerate printf's declared void.
@@ -1660,7 +1676,7 @@
     // is used, in general the printf return value is not compatible with either
     // putchar() or puts().
     if (!CI->use_empty())
-      return 0;
+      return nullptr;
 
     // printf("x") -> putchar('x'), even for '%'.
     if (FormatStr.size() == 1) {
@@ -1697,7 +1713,7 @@
         CI->getArgOperand(1)->getType()->isPointerTy()) {
       return EmitPutS(CI->getArgOperand(1), B, DL, TLI);
     }
-    return 0;
+    return nullptr;
   }
 
   Value *callOptimizer(Function *Callee, CallInst *CI,
@@ -1707,7 +1723,7 @@
     if (FT->getNumParams() < 1 || !FT->getParamType(0)->isPointerTy() ||
         !(FT->getReturnType()->isIntegerTy() ||
           FT->getReturnType()->isVoidTy()))
-      return 0;
+      return nullptr;
 
     if (Value *V = optimizeFixedFormatString(Callee, CI, B)) {
       return V;
@@ -1724,7 +1740,7 @@
       B.Insert(New);
       return New;
     }
-    return 0;
+    return nullptr;
   }
 };
 
@@ -1734,7 +1750,7 @@
     // Check for a fixed format string.
     StringRef FormatStr;
     if (!getConstantStringInfo(CI->getArgOperand(1), FormatStr))
-      return 0;
+      return nullptr;
 
     // If we just have a format string (nothing else crazy) transform it.
     if (CI->getNumArgOperands() == 2) {
@@ -1742,10 +1758,10 @@
       // %% -> % in the future if we cared.
       for (unsigned i = 0, e = FormatStr.size(); i != e; ++i)
         if (FormatStr[i] == '%')
-          return 0; // we found a format specifier, bail out.
+          return nullptr; // we found a format specifier, bail out.
 
       // These optimizations require DataLayout.
-      if (!DL) return 0;
+      if (!DL) return nullptr;
 
       // sprintf(str, fmt) -> llvm.memcpy(str, fmt, strlen(fmt)+1, 1)
       B.CreateMemCpy(CI->getArgOperand(0), CI->getArgOperand(1),
@@ -1758,12 +1774,12 @@
     // and have an extra operand.
     if (FormatStr.size() != 2 || FormatStr[0] != '%' ||
         CI->getNumArgOperands() < 3)
-      return 0;
+      return nullptr;
 
     // Decode the second character of the format string.
     if (FormatStr[1] == 'c') {
       // sprintf(dst, "%c", chr) --> *(i8*)dst = chr; *((i8*)dst+1) = 0
-      if (!CI->getArgOperand(2)->getType()->isIntegerTy()) return 0;
+      if (!CI->getArgOperand(2)->getType()->isIntegerTy()) return nullptr;
       Value *V = B.CreateTrunc(CI->getArgOperand(2), B.getInt8Ty(), "char");
       Value *Ptr = CastToCStr(CI->getArgOperand(0), B);
       B.CreateStore(V, Ptr);
@@ -1775,14 +1791,14 @@
 
     if (FormatStr[1] == 's') {
       // These optimizations require DataLayout.
-      if (!DL) return 0;
+      if (!DL) return nullptr;
 
       // sprintf(dest, "%s", str) -> llvm.memcpy(dest, str, strlen(str)+1, 1)
-      if (!CI->getArgOperand(2)->getType()->isPointerTy()) return 0;
+      if (!CI->getArgOperand(2)->getType()->isPointerTy()) return nullptr;
 
       Value *Len = EmitStrLen(CI->getArgOperand(2), B, DL, TLI);
       if (!Len)
-        return 0;
+        return nullptr;
       Value *IncLen = B.CreateAdd(Len,
                                   ConstantInt::get(Len->getType(), 1),
                                   "leninc");
@@ -1791,7 +1807,7 @@
       // The sprintf result is the unincremented number of bytes in the string.
       return B.CreateIntCast(Len, CI->getType(), false);
     }
-    return 0;
+    return nullptr;
   }
 
   Value *callOptimizer(Function *Callee, CallInst *CI,
@@ -1801,7 +1817,7 @@
     if (FT->getNumParams() != 2 || !FT->getParamType(0)->isPointerTy() ||
         !FT->getParamType(1)->isPointerTy() ||
         !FT->getReturnType()->isIntegerTy())
-      return 0;
+      return nullptr;
 
     if (Value *V = OptimizeFixedFormatString(Callee, CI, B)) {
       return V;
@@ -1818,7 +1834,7 @@
       B.Insert(New);
       return New;
     }
-    return 0;
+    return nullptr;
   }
 };
 
@@ -1831,22 +1847,22 @@
     // All the optimizations depend on the format string.
     StringRef FormatStr;
     if (!getConstantStringInfo(CI->getArgOperand(1), FormatStr))
-      return 0;
+      return nullptr;
 
     // Do not do any of the following transformations if the fprintf return
     // value is used, in general the fprintf return value is not compatible
     // with fwrite(), fputc() or fputs().
     if (!CI->use_empty())
-      return 0;
+      return nullptr;
 
     // fprintf(F, "foo") --> fwrite("foo", 3, 1, F)
     if (CI->getNumArgOperands() == 2) {
       for (unsigned i = 0, e = FormatStr.size(); i != e; ++i)
         if (FormatStr[i] == '%')  // Could handle %% -> % if we cared.
-          return 0; // We found a format specifier.
+          return nullptr; // We found a format specifier.
 
       // These optimizations require DataLayout.
-      if (!DL) return 0;
+      if (!DL) return nullptr;
 
       return EmitFWrite(CI->getArgOperand(1),
                         ConstantInt::get(DL->getIntPtrType(*Context),
@@ -1858,22 +1874,22 @@
     // and have an extra operand.
     if (FormatStr.size() != 2 || FormatStr[0] != '%' ||
         CI->getNumArgOperands() < 3)
-      return 0;
+      return nullptr;
 
     // Decode the second character of the format string.
     if (FormatStr[1] == 'c') {
       // fprintf(F, "%c", chr) --> fputc(chr, F)
-      if (!CI->getArgOperand(2)->getType()->isIntegerTy()) return 0;
+      if (!CI->getArgOperand(2)->getType()->isIntegerTy()) return nullptr;
       return EmitFPutC(CI->getArgOperand(2), CI->getArgOperand(0), B, DL, TLI);
     }
 
     if (FormatStr[1] == 's') {
       // fprintf(F, "%s", str) --> fputs(str, F)
       if (!CI->getArgOperand(2)->getType()->isPointerTy())
-        return 0;
+        return nullptr;
       return EmitFPutS(CI->getArgOperand(2), CI->getArgOperand(0), B, DL, TLI);
     }
-    return 0;
+    return nullptr;
   }
 
   Value *callOptimizer(Function *Callee, CallInst *CI,
@@ -1883,7 +1899,7 @@
     if (FT->getNumParams() != 2 || !FT->getParamType(0)->isPointerTy() ||
         !FT->getParamType(1)->isPointerTy() ||
         !FT->getReturnType()->isIntegerTy())
-      return 0;
+      return nullptr;
 
     if (Value *V = optimizeFixedFormatString(Callee, CI, B)) {
       return V;
@@ -1900,7 +1916,7 @@
       B.Insert(New);
       return New;
     }
-    return 0;
+    return nullptr;
   }
 };
 
@@ -1917,12 +1933,12 @@
         !FT->getParamType(2)->isIntegerTy() ||
         !FT->getParamType(3)->isPointerTy() ||
         !FT->getReturnType()->isIntegerTy())
-      return 0;
+      return nullptr;
 
     // Get the element size and count.
     ConstantInt *SizeC = dyn_cast<ConstantInt>(CI->getArgOperand(1));
     ConstantInt *CountC = dyn_cast<ConstantInt>(CI->getArgOperand(2));
-    if (!SizeC || !CountC) return 0;
+    if (!SizeC || !CountC) return nullptr;
     uint64_t Bytes = SizeC->getZExtValue()*CountC->getZExtValue();
 
     // If this is writing zero records, remove the call (it's a noop).
@@ -1934,10 +1950,10 @@
     if (Bytes == 1 && CI->use_empty()) {  // fwrite(S,1,1,F) -> fputc(S[0],F)
       Value *Char = B.CreateLoad(CastToCStr(CI->getArgOperand(0), B), "char");
       Value *NewCI = EmitFPutC(Char, CI->getArgOperand(3), B, DL, TLI);
-      return NewCI ? ConstantInt::get(CI->getType(), 1) : 0;
+      return NewCI ? ConstantInt::get(CI->getType(), 1) : nullptr;
     }
 
-    return 0;
+    return nullptr;
   }
 };
 
@@ -1948,18 +1964,18 @@
     (void) ER.callOptimizer(Callee, CI, B);
 
     // These optimizations require DataLayout.
-    if (!DL) return 0;
+    if (!DL) return nullptr;
 
     // Require two pointers.  Also, we can't optimize if return value is used.
     FunctionType *FT = Callee->getFunctionType();
     if (FT->getNumParams() != 2 || !FT->getParamType(0)->isPointerTy() ||
         !FT->getParamType(1)->isPointerTy() ||
         !CI->use_empty())
-      return 0;
+      return nullptr;
 
     // fputs(s,F) --> fwrite(s,1,strlen(s),F)
     uint64_t Len = GetStringLength(CI->getArgOperand(0));
-    if (!Len) return 0;
+    if (!Len) return nullptr;
     // Known to have no uses (see above).
     return EmitFWrite(CI->getArgOperand(0),
                       ConstantInt::get(DL->getIntPtrType(*Context), Len-1),
@@ -1975,12 +1991,12 @@
     if (FT->getNumParams() < 1 || !FT->getParamType(0)->isPointerTy() ||
         !(FT->getReturnType()->isIntegerTy() ||
           FT->getReturnType()->isVoidTy()))
-      return 0;
+      return nullptr;
 
     // Check for a constant string.
     StringRef Str;
     if (!getConstantStringInfo(CI->getArgOperand(0), Str))
-      return 0;
+      return nullptr;
 
     if (Str.empty() && CI->use_empty()) {
       // puts("") -> putchar('\n')
@@ -1989,7 +2005,7 @@
       return B.CreateIntCast(Res, CI->getType(), true);
     }
 
-    return 0;
+    return nullptr;
   }
 };
 
@@ -2100,7 +2116,7 @@
     case Intrinsic::exp2:
        return &Exp2;
     default:
-       return 0;
+       return nullptr;
     }
   }
 
@@ -2210,7 +2226,7 @@
       case LibFunc::trunc:
         if (hasFloatVersion(FuncName))
           return &UnaryDoubleFP;
-        return 0;
+        return nullptr;
       case LibFunc::acos:
       case LibFunc::acosh:
       case LibFunc::asin:
@@ -2234,16 +2250,16 @@
       case LibFunc::tanh:
         if (UnsafeFPShrink && hasFloatVersion(FuncName))
          return &UnsafeUnaryDoubleFP;
-        return 0;
+        return nullptr;
       case LibFunc::fmin:
       case LibFunc::fmax:
         if (hasFloatVersion(FuncName))
           return &BinaryDoubleFP;
-        return 0;
+        return nullptr;
       case LibFunc::memcpy_chk:
         return &MemCpyChk;
       default:
-        return 0;
+        return nullptr;
       }
   }
 
@@ -2263,7 +2279,7 @@
       return &StrNCpyChk;
   }
 
-  return 0;
+  return nullptr;
 
 }
 
@@ -2273,7 +2289,7 @@
     IRBuilder<> Builder(CI);
     return LCO->optimizeCall(CI, DL, TLI, LCS, Builder);
   }
-  return 0;
+  return nullptr;
 }
 
 LibCallSimplifier::LibCallSimplifier(const DataLayout *DL,
@@ -2287,7 +2303,7 @@
 }
 
 Value *LibCallSimplifier::optimizeCall(CallInst *CI) {
-  if (CI->isNoBuiltin()) return 0;
+  if (CI->isNoBuiltin()) return nullptr;
   return Impl->optimizeCall(CI);
 }
 
diff --git a/lib/Transforms/Utils/SpecialCaseList.cpp b/lib/Transforms/Utils/SpecialCaseList.cpp
index c318560..2c6fcd1 100644
--- a/lib/Transforms/Utils/SpecialCaseList.cpp
+++ b/lib/Transforms/Utils/SpecialCaseList.cpp
@@ -41,7 +41,7 @@
   StringSet<> Strings;
   Regex *RegEx;
 
-  Entry() : RegEx(0) {}
+  Entry() : RegEx(nullptr) {}
 
   bool match(StringRef Query) const {
     return Strings.count(Query) || (RegEx && RegEx->match(Query));
@@ -57,7 +57,7 @@
   std::unique_ptr<MemoryBuffer> File;
   if (error_code EC = MemoryBuffer::getFile(Path, File)) {
     Error = (Twine("Can't open file '") + Path + "': " + EC.message()).str();
-    return 0;
+    return nullptr;
   }
   return create(File.get(), Error);
 }
@@ -66,7 +66,7 @@
     const MemoryBuffer *MB, std::string &Error) {
   std::unique_ptr<SpecialCaseList> SCL(new SpecialCaseList());
   if (!SCL->parse(MB, Error))
-    return 0;
+    return nullptr;
   return SCL.release();
 }
 
diff --git a/lib/Transforms/Utils/UnifyFunctionExitNodes.cpp b/lib/Transforms/Utils/UnifyFunctionExitNodes.cpp
index 560f581..0c2fc0a 100644
--- a/lib/Transforms/Utils/UnifyFunctionExitNodes.cpp
+++ b/lib/Transforms/Utils/UnifyFunctionExitNodes.cpp
@@ -59,7 +59,7 @@
 
   // Then unreachable blocks.
   if (UnreachableBlocks.empty()) {
-    UnreachableBlock = 0;
+    UnreachableBlock = nullptr;
   } else if (UnreachableBlocks.size() == 1) {
     UnreachableBlock = UnreachableBlocks.front();
   } else {
@@ -77,7 +77,7 @@
 
   // Now handle return blocks.
   if (ReturningBlocks.empty()) {
-    ReturnBlock = 0;
+    ReturnBlock = nullptr;
     return false;                          // No blocks return
   } else if (ReturningBlocks.size() == 1) {
     ReturnBlock = ReturningBlocks.front(); // Already has a single return block
@@ -91,9 +91,9 @@
   BasicBlock *NewRetBlock = BasicBlock::Create(F.getContext(),
                                                "UnifiedReturnBlock", &F);
 
-  PHINode *PN = 0;
+  PHINode *PN = nullptr;
   if (F.getReturnType()->isVoidTy()) {
-    ReturnInst::Create(F.getContext(), NULL, NewRetBlock);
+    ReturnInst::Create(F.getContext(), nullptr, NewRetBlock);
   } else {
     // If the function doesn't return void... add a PHI node to the block...
     PN = PHINode::Create(F.getReturnType(), ReturningBlocks.size(),
diff --git a/lib/Transforms/Utils/ValueMapper.cpp b/lib/Transforms/Utils/ValueMapper.cpp
index 457fc80..0f20e6d 100644
--- a/lib/Transforms/Utils/ValueMapper.cpp
+++ b/lib/Transforms/Utils/ValueMapper.cpp
@@ -71,12 +71,12 @@
     // Check all operands to see if any need to be remapped.
     for (unsigned i = 0, e = MD->getNumOperands(); i != e; ++i) {
       Value *OP = MD->getOperand(i);
-      if (OP == 0) continue;
+      if (!OP) continue;
       Value *Mapped_OP = MapValue(OP, VM, Flags, TypeMapper, Materializer);
       // Use identity map if Mapped_Op is null and we can ignore missing
       // entries.
       if (Mapped_OP == OP ||
-          (Mapped_OP == 0 && (Flags & RF_IgnoreMissingEntries)))
+          (Mapped_OP == nullptr && (Flags & RF_IgnoreMissingEntries)))
         continue;
 
       // Ok, at least one operand needs remapping.  
@@ -84,13 +84,13 @@
       Elts.reserve(MD->getNumOperands());
       for (i = 0; i != e; ++i) {
         Value *Op = MD->getOperand(i);
-        if (Op == 0)
-          Elts.push_back(0);
+        if (!Op)
+          Elts.push_back(nullptr);
         else {
           Value *Mapped_Op = MapValue(Op, VM, Flags, TypeMapper, Materializer);
           // Use identity map if Mapped_Op is null and we can ignore missing
           // entries.
-          if (Mapped_Op == 0 && (Flags & RF_IgnoreMissingEntries))
+          if (Mapped_Op == nullptr && (Flags & RF_IgnoreMissingEntries))
             Mapped_Op = Op;
           Elts.push_back(Mapped_Op);
         }
@@ -112,8 +112,8 @@
   // Okay, this either must be a constant (which may or may not be mappable) or
   // is something that is not in the mapping table.
   Constant *C = const_cast<Constant*>(dyn_cast<Constant>(V));
-  if (C == 0)
-    return 0;
+  if (!C)
+    return nullptr;
   
   if (BlockAddress *BA = dyn_cast<BlockAddress>(C)) {
     Function *F = 
@@ -126,7 +126,7 @@
   // Otherwise, we have some other constant to remap.  Start by checking to see
   // if all operands have an identity remapping.
   unsigned OpNo = 0, NumOperands = C->getNumOperands();
-  Value *Mapped = 0;
+  Value *Mapped = nullptr;
   for (; OpNo != NumOperands; ++OpNo) {
     Value *Op = C->getOperand(OpNo);
     Mapped = MapValue(Op, VM, Flags, TypeMapper, Materializer);
@@ -187,7 +187,7 @@
   for (User::op_iterator op = I->op_begin(), E = I->op_end(); op != E; ++op) {
     Value *V = MapValue(*op, VMap, Flags, TypeMapper, Materializer);
     // If we aren't ignoring missing entries, assert that something happened.
-    if (V != 0)
+    if (V)
       *op = V;
     else
       assert((Flags & RF_IgnoreMissingEntries) &&
@@ -199,7 +199,7 @@
     for (unsigned i = 0, e = PN->getNumIncomingValues(); i != e; ++i) {
       Value *V = MapValue(PN->getIncomingBlock(i), VMap, Flags);
       // If we aren't ignoring missing entries, assert that something happened.
-      if (V != 0)
+      if (V)
         PN->setIncomingBlock(i, cast<BasicBlock>(V));
       else
         assert((Flags & RF_IgnoreMissingEntries) &&