vendor/llvm/llvm-trunk-r301441

8 files changed, 134 insertions, 73 deletions

diff --git a/lib/Transforms/Utils/BypassSlowDivision.cpp b/lib/Transforms/Utils/BypassSlowDivision.cpp
index 1cfe3bd53648..7ffdad597a9b 100644
--- a/lib/Transforms/Utils/BypassSlowDivision.cpp
+++ b/lib/Transforms/Utils/BypassSlowDivision.cpp
@@ -22,6 +22,7 @@
 #include "llvm/IR/Function.h"
 #include "llvm/IR/IRBuilder.h"
 #include "llvm/IR/Instructions.h"
+#include "llvm/Support/KnownBits.h"
 #include "llvm/Transforms/Utils/Local.h"
 
 using namespace llvm;
@@ -256,14 +257,14 @@ ValueRange FastDivInsertionTask::getValueRange(Value *V,
   unsigned HiBits = LongLen - ShortLen;
 
   const DataLayout &DL = SlowDivOrRem->getModule()->getDataLayout();
-  APInt Zeros(LongLen, 0), Ones(LongLen, 0);
+  KnownBits Known(LongLen);
 
-  computeKnownBits(V, Zeros, Ones, DL);
+  computeKnownBits(V, Known, DL);
 
-  if (Zeros.countLeadingOnes() >= HiBits)
+  if (Known.Zero.countLeadingOnes() >= HiBits)
     return VALRNG_KNOWN_SHORT;
 
-  if (Ones.countLeadingZeros() < HiBits)
+  if (Known.One.countLeadingZeros() < HiBits)
     return VALRNG_LIKELY_LONG;
 
   // Long integer divisions are often used in hashtable implementations. It's
diff --git a/lib/Transforms/Utils/CodeExtractor.cpp b/lib/Transforms/Utils/CodeExtractor.cpp
index 82552684b832..ed72099ec3ed 100644
--- a/lib/Transforms/Utils/CodeExtractor.cpp
+++ b/lib/Transforms/Utils/CodeExtractor.cpp
@@ -73,24 +73,26 @@ bool CodeExtractor::isBlockValidForExtraction(const BasicBlock &BB) {
 }
 
 /// \brief Build a set of blocks to extract if the input blocks are viable.
-template <typename IteratorT>
-static SetVector<BasicBlock *> buildExtractionBlockSet(IteratorT BBBegin,
-                                                       IteratorT BBEnd) {
+static SetVector<BasicBlock *>
+buildExtractionBlockSet(ArrayRef<BasicBlock *> BBs, DominatorTree *DT) {
+  assert(!BBs.empty() && "The set of blocks to extract must be non-empty");
   SetVector<BasicBlock *> Result;
 
-  assert(BBBegin != BBEnd);
-
   // Loop over the blocks, adding them to our set-vector, and aborting with an
   // empty set if we encounter invalid blocks.
-  do {
-    if (!Result.insert(*BBBegin))
-      llvm_unreachable("Repeated basic blocks in extraction input");
+  for (BasicBlock *BB : BBs) {
 
-    if (!CodeExtractor::isBlockValidForExtraction(**BBBegin)) {
+    // If this block is dead, don't process it.
+    if (DT && !DT->isReachableFromEntry(BB))
+      continue;
+
+    if (!Result.insert(BB))
+      llvm_unreachable("Repeated basic blocks in extraction input");
+    if (!CodeExtractor::isBlockValidForExtraction(*BB)) {
       Result.clear();
       return Result;
     }
-  } while (++BBBegin != BBEnd);
+  }
 
 #ifndef NDEBUG
   for (SetVector<BasicBlock *>::iterator I = std::next(Result.begin()),
@@ -106,23 +108,17 @@ static SetVector<BasicBlock *> buildExtractionBlockSet(IteratorT BBBegin,
   return Result;
 }
 
-/// \brief Helper to call buildExtractionBlockSet with an ArrayRef.
-static SetVector<BasicBlock *>
-buildExtractionBlockSet(ArrayRef<BasicBlock *> BBs) {
-  return buildExtractionBlockSet(BBs.begin(), BBs.end());
-}
-
 CodeExtractor::CodeExtractor(ArrayRef<BasicBlock *> BBs, DominatorTree *DT,
                              bool AggregateArgs, BlockFrequencyInfo *BFI,
                              BranchProbabilityInfo *BPI)
     : DT(DT), AggregateArgs(AggregateArgs || AggregateArgsOpt), BFI(BFI),
-      BPI(BPI), Blocks(buildExtractionBlockSet(BBs)), NumExitBlocks(~0U) {}
+      BPI(BPI), Blocks(buildExtractionBlockSet(BBs, DT)), NumExitBlocks(~0U) {}
 
 CodeExtractor::CodeExtractor(DominatorTree &DT, Loop &L, bool AggregateArgs,
                              BlockFrequencyInfo *BFI,
                              BranchProbabilityInfo *BPI)
     : DT(&DT), AggregateArgs(AggregateArgs || AggregateArgsOpt), BFI(BFI),
-      BPI(BPI), Blocks(buildExtractionBlockSet(L.getBlocks())),
+      BPI(BPI), Blocks(buildExtractionBlockSet(L.getBlocks(), &DT)),
       NumExitBlocks(~0U) {}
 
 /// definedInRegion - Return true if the specified value is defined in the
@@ -194,9 +190,7 @@ void CodeExtractor::severSplitPHINodes(BasicBlock *&Header) {
   // containing PHI nodes merging values from outside of the region, and a
   // second that contains all of the code for the block and merges back any
   // incoming values from inside of the region.
-  BasicBlock::iterator AfterPHIs = Header->getFirstNonPHI()->getIterator();
-  BasicBlock *NewBB = Header->splitBasicBlock(AfterPHIs,
-                                              Header->getName()+".ce");
+  BasicBlock *NewBB = llvm::SplitBlock(Header, Header->getFirstNonPHI(), DT);
 
   // We only want to code extract the second block now, and it becomes the new
   // header of the region.
@@ -205,11 +199,6 @@ void CodeExtractor::severSplitPHINodes(BasicBlock *&Header) {
   Blocks.insert(NewBB);
   Header = NewBB;
 
-  // Okay, update dominator sets. The blocks that dominate the new one are the
-  // blocks that dominate TIBB plus the new block itself.
-  if (DT)
-    DT->splitBlock(NewBB);
-
   // Okay, now we need to adjust the PHI nodes and any branches from within the
   // region to go to the new header block instead of the old header block.
   if (NumPredsFromRegion) {
@@ -224,12 +213,14 @@ void CodeExtractor::severSplitPHINodes(BasicBlock *&Header) {
 
     // Okay, everything within the region is now branching to the right block, we
     // just have to update the PHI nodes now, inserting PHI nodes into NewBB.
+    BasicBlock::iterator AfterPHIs;
     for (AfterPHIs = OldPred->begin(); isa<PHINode>(AfterPHIs); ++AfterPHIs) {
       PHINode *PN = cast<PHINode>(AfterPHIs);
       // Create a new PHI node in the new region, which has an incoming value
       // from OldPred of PN.
       PHINode *NewPN = PHINode::Create(PN->getType(), 1 + NumPredsFromRegion,
                                        PN->getName() + ".ce", &NewBB->front());
+      PN->replaceAllUsesWith(NewPN);
       NewPN->addIncoming(PN, OldPred);
 
       // Loop over all of the incoming value in PN, moving them to NewPN if they
diff --git a/lib/Transforms/Utils/Local.cpp b/lib/Transforms/Utils/Local.cpp
index 8c5442762643..d3002c5fb750 100644
--- a/lib/Transforms/Utils/Local.cpp
+++ b/lib/Transforms/Utils/Local.cpp
@@ -45,6 +45,7 @@
 #include "llvm/IR/PatternMatch.h"
 #include "llvm/IR/ValueHandle.h"
 #include "llvm/Support/Debug.h"
+#include "llvm/Support/KnownBits.h"
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/raw_ostream.h"
 using namespace llvm;
@@ -1038,9 +1039,9 @@ unsigned llvm::getOrEnforceKnownAlignment(Value *V, unsigned PrefAlign,
          "getOrEnforceKnownAlignment expects a pointer!");
   unsigned BitWidth = DL.getPointerTypeSizeInBits(V->getType());
 
-  APInt KnownZero(BitWidth, 0), KnownOne(BitWidth, 0);
-  computeKnownBits(V, KnownZero, KnownOne, DL, 0, AC, CxtI, DT);
-  unsigned TrailZ = KnownZero.countTrailingOnes();
+  KnownBits Known(BitWidth);
+  computeKnownBits(V, Known, DL, 0, AC, CxtI, DT);
+  unsigned TrailZ = Known.Zero.countTrailingOnes();
 
   // Avoid trouble with ridiculously large TrailZ values, such as
   // those computed from a null pointer.
@@ -1268,21 +1269,37 @@ static void appendOffset(SmallVectorImpl<uint64_t> &Ops, int64_t Offset) {
   }
 }
 
-/// Prepend \p DIExpr with a deref and offset operation.
+enum { WithStackValue = true };
+
+/// Prepend \p DIExpr with a deref and offset operation and optionally turn it
+/// into a stack value.
 static DIExpression *prependDIExpr(DIBuilder &Builder, DIExpression *DIExpr,
-                                   bool Deref, int64_t Offset) {
-  if (!Deref && !Offset)
+                                   bool Deref, int64_t Offset = 0,
+                                   bool StackValue = false) {
+  if (!Deref && !Offset && !StackValue)
     return DIExpr;
-  // Create a copy of the original DIDescriptor for user variable, prepending
-  // "deref" operation to a list of address elements, as new llvm.dbg.declare
-  // will take a value storing address of the memory for variable, not
-  // alloca itself.
-  SmallVector<uint64_t, 4> Ops;
+
+  SmallVector<uint64_t, 8> Ops;
+  appendOffset(Ops, Offset);
   if (Deref)
     Ops.push_back(dwarf::DW_OP_deref);
-  appendOffset(Ops, Offset);
   if (DIExpr)
-    Ops.append(DIExpr->elements_begin(), DIExpr->elements_end());
+    for (auto Op : DIExpr->expr_ops()) {
+      // A DW_OP_stack_value comes at the end, but before a DW_OP_LLVM_fragment.
+      if (StackValue) {
+        if (Op.getOp() == dwarf::DW_OP_stack_value)
+          StackValue = false;
+        else if (Op.getOp() == dwarf::DW_OP_LLVM_fragment) {
+          Ops.push_back(dwarf::DW_OP_stack_value);
+          StackValue = false;
+        }
+      }
+      Ops.push_back(Op.getOp());
+      for (unsigned I = 0; I < Op.getNumArgs(); ++I)
+        Ops.push_back(Op.getArg(I));
+    }
+  if (StackValue)
+    Ops.push_back(dwarf::DW_OP_stack_value);
   return Builder.createExpression(Ops);
 }
 
@@ -1374,12 +1391,15 @@ void llvm::salvageDebugInfo(Instruction &I) {
       unsigned BitWidth =
           M.getDataLayout().getPointerSizeInBits(GEP->getPointerAddressSpace());
       APInt Offset(BitWidth, 0);
-      // Rewrite a constant GEP into a DIExpression.
+      // Rewrite a constant GEP into a DIExpression.  Since we are performing
+      // arithmetic to compute the variable's *value* in the DIExpression, we
+      // need to mark the expression with a DW_OP_stack_value.
       if (GEP->accumulateConstantOffset(M.getDataLayout(), Offset)) {
         auto *DIExpr = DVI->getExpression();
         DIBuilder DIB(M, /*AllowUnresolved*/ false);
-        // GEP offsets are i32 and thus alwaus fit into an int64_t.
-        DIExpr = prependDIExpr(DIB, DIExpr, NoDeref, Offset.getSExtValue());
+        // GEP offsets are i32 and thus always fit into an int64_t.
+        DIExpr = prependDIExpr(DIB, DIExpr, NoDeref, Offset.getSExtValue(),
+                               WithStackValue);
         DVI->setOperand(0, MDWrap(I.getOperand(0)));
         DVI->setOperand(3, MetadataAsValue::get(I.getContext(), DIExpr));
         DEBUG(dbgs() << "SALVAGE: " << *DVI << '\n');
@@ -1391,7 +1411,7 @@ void llvm::salvageDebugInfo(Instruction &I) {
       // Rewrite the load into DW_OP_deref.
       auto *DIExpr = DVI->getExpression();
       DIBuilder DIB(M, /*AllowUnresolved*/ false);
-      DIExpr = prependDIExpr(DIB, DIExpr, WithDeref, 0);
+      DIExpr = prependDIExpr(DIB, DIExpr, WithDeref);
       DVI->setOperand(0, MDWrap(I.getOperand(0)));
       DVI->setOperand(3, MetadataAsValue::get(I.getContext(), DIExpr));
       DEBUG(dbgs() << "SALVAGE:  " << *DVI << '\n');
diff --git a/lib/Transforms/Utils/LoopUnroll.cpp b/lib/Transforms/Utils/LoopUnroll.cpp
index 3c669ce644e2..43ab725b0769 100644
--- a/lib/Transforms/Utils/LoopUnroll.cpp
+++ b/lib/Transforms/Utils/LoopUnroll.cpp
@@ -318,6 +318,10 @@ bool llvm::UnrollLoop(Loop *L, unsigned Count, unsigned TripCount, bool Force,
     return false;
   }
 
+  // The current loop unroll pass can only unroll loops with a single latch
+  // that's a conditional branch exiting the loop.
+  // FIXME: The implementation can be extended to work with more complicated
+  // cases, e.g. loops with multiple latches.
   BasicBlock *Header = L->getHeader();
   BranchInst *BI = dyn_cast<BranchInst>(LatchBlock->getTerminator());
 
@@ -328,6 +332,16 @@ bool llvm::UnrollLoop(Loop *L, unsigned Count, unsigned TripCount, bool Force,
     return false;
   }
 
+  auto CheckSuccessors = [&](unsigned S1, unsigned S2) {
+    return BI->getSuccessor(S1) == Header && !L->contains(BI->getSuccessor(S2));
+  };
+
+  if (!CheckSuccessors(0, 1) && !CheckSuccessors(1, 0)) {
+    DEBUG(dbgs() << "Can't unroll; only loops with one conditional latch"
+                    " exiting the loop can be unrolled\n");
+    return false;
+  }
+
   if (Header->hasAddressTaken()) {
     // The loop-rotate pass can be helpful to avoid this in many cases.
     DEBUG(dbgs() <<
diff --git a/lib/Transforms/Utils/LowerSwitch.cpp b/lib/Transforms/Utils/LowerSwitch.cpp
index b375d51005d5..8959e77438e9 100644
--- a/lib/Transforms/Utils/LowerSwitch.cpp
+++ b/lib/Transforms/Utils/LowerSwitch.cpp
@@ -403,6 +403,14 @@ void LowerSwitch::processSwitchInst(SwitchInst *SI,
   Value *Val = SI->getCondition();  // The value we are switching on...
   BasicBlock* Default = SI->getDefaultDest();
 
+  // Don't handle unreachable blocks. If there are successors with phis, this
+  // would leave them behind with missing predecessors.
+  if ((CurBlock != &F->getEntryBlock() && pred_empty(CurBlock)) ||
+      CurBlock->getSinglePredecessor() == CurBlock) {
+    DeleteList.insert(CurBlock);
+    return;
+  }
+
   // If there is only the default destination, just branch.
   if (!SI->getNumCases()) {
     BranchInst::Create(Default, CurBlock);
diff --git a/lib/Transforms/Utils/SimplifyCFG.cpp b/lib/Transforms/Utils/SimplifyCFG.cpp
index 2f575b9d5027..f86e97b6cc72 100644
--- a/lib/Transforms/Utils/SimplifyCFG.cpp
+++ b/lib/Transforms/Utils/SimplifyCFG.cpp
@@ -60,6 +60,7 @@
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/KnownBits.h"
 #include "llvm/Support/MathExtras.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
@@ -3055,6 +3056,15 @@ static bool mergeConditionalStores(BranchInst *PBI, BranchInst *QBI) {
   BasicBlock *QFB = QBI->getSuccessor(1);
   BasicBlock *PostBB = QFB->getSingleSuccessor();
 
+  // Make sure we have a good guess for PostBB. If QTB's only successor is
+  // QFB, then QFB is a better PostBB.
+  if (QTB->getSingleSuccessor() == QFB)
+    PostBB = QFB;
+
+  // If we couldn't find a good PostBB, stop.
+  if (!PostBB)
+    return false;
+
   bool InvertPCond = false, InvertQCond = false;
   // Canonicalize fallthroughs to the true branches.
   if (PFB == QBI->getParent()) {
@@ -3079,8 +3089,7 @@ static bool mergeConditionalStores(BranchInst *PBI, BranchInst *QBI) {
   auto HasOnePredAndOneSucc = [](BasicBlock *BB, BasicBlock *P, BasicBlock *S) {
     return BB->getSinglePredecessor() == P && BB->getSingleSuccessor() == S;
   };
-  if (!PostBB ||
-      !HasOnePredAndOneSucc(PFB, PBI->getParent(), QBI->getParent()) ||
+  if (!HasOnePredAndOneSucc(PFB, PBI->getParent(), QBI->getParent()) ||
       !HasOnePredAndOneSucc(QFB, QBI->getParent(), PostBB))
     return false;
   if ((PTB && !HasOnePredAndOneSucc(PTB, PBI->getParent(), QBI->getParent())) ||
@@ -3746,7 +3755,7 @@ bool SimplifyCFGOpt::SimplifyCommonResume(ResumeInst *RI) {
     if (!isa<DbgInfoIntrinsic>(I))
       return false;
 
-  SmallSet<BasicBlock *, 4> TrivialUnwindBlocks;
+  SmallSetVector<BasicBlock *, 4> TrivialUnwindBlocks;
   auto *PhiLPInst = cast<PHINode>(RI->getValue());
 
   // Check incoming blocks to see if any of them are trivial.
@@ -4359,8 +4368,8 @@ static bool EliminateDeadSwitchCases(SwitchInst *SI, AssumptionCache *AC,
                                      const DataLayout &DL) {
   Value *Cond = SI->getCondition();
   unsigned Bits = Cond->getType()->getIntegerBitWidth();
-  APInt KnownZero(Bits, 0), KnownOne(Bits, 0);
-  computeKnownBits(Cond, KnownZero, KnownOne, DL, 0, AC, SI);
+  KnownBits Known(Bits);
+  computeKnownBits(Cond, Known, DL, 0, AC, SI);
 
   // We can also eliminate cases by determining that their values are outside of
   // the limited range of the condition based on how many significant (non-sign)
@@ -4372,7 +4381,7 @@ static bool EliminateDeadSwitchCases(SwitchInst *SI, AssumptionCache *AC,
   SmallVector<ConstantInt *, 8> DeadCases;
   for (auto &Case : SI->cases()) {
     APInt CaseVal = Case.getCaseValue()->getValue();
-    if ((CaseVal & KnownZero) != 0 || (CaseVal & KnownOne) != KnownOne ||
+    if (Known.Zero.intersects(CaseVal) || !Known.One.isSubsetOf(CaseVal) ||
         (CaseVal.getMinSignedBits() > MaxSignificantBitsInCond)) {
       DeadCases.push_back(Case.getCaseValue());
       DEBUG(dbgs() << "SimplifyCFG: switch case " << CaseVal << " is dead.\n");
@@ -4386,7 +4395,7 @@ static bool EliminateDeadSwitchCases(SwitchInst *SI, AssumptionCache *AC,
   bool HasDefault =
       !isa<UnreachableInst>(SI->getDefaultDest()->getFirstNonPHIOrDbg());
   const unsigned NumUnknownBits =
-      Bits - (KnownZero | KnownOne).countPopulation();
+      Bits - (Known.Zero | Known.One).countPopulation();
   assert(NumUnknownBits <= Bits);
   if (HasDefault && DeadCases.empty() &&
       NumUnknownBits < 64 /* avoid overflow */ &&
diff --git a/lib/Transforms/Utils/SimplifyInstructions.cpp b/lib/Transforms/Utils/SimplifyInstructions.cpp
index f6070868de44..27373427d4f7 100644
--- a/lib/Transforms/Utils/SimplifyInstructions.cpp
+++ b/lib/Transforms/Utils/SimplifyInstructions.cpp
@@ -35,10 +35,8 @@ using namespace llvm;
 
 STATISTIC(NumSimplified, "Number of redundant instructions removed");
 
-static bool runImpl(Function &F, const DominatorTree *DT,
-                    const TargetLibraryInfo *TLI, AssumptionCache *AC,
+static bool runImpl(Function &F, const SimplifyQuery &SQ,
                     OptimizationRemarkEmitter *ORE) {
-  const DataLayout &DL = F.getParent()->getDataLayout();
   SmallPtrSet<const Instruction *, 8> S1, S2, *ToSimplify = &S1, *Next = &S2;
   bool Changed = false;
 
@@ -56,7 +54,8 @@ static bool runImpl(Function &F, const DominatorTree *DT,
 
         // Don't waste time simplifying unused instructions.
         if (!I->use_empty()) {
-          if (Value *V = SimplifyInstruction(I, DL, TLI, DT, AC, ORE)) {
+          if (Value *V =
+                  SimplifyInstruction(I, SQ.getWithInstruction(I), ORE)) {
             // Mark all uses for resimplification next time round the loop.
             for (User *U : I->users())
               Next->insert(cast<Instruction>(U));
@@ -65,7 +64,7 @@ static bool runImpl(Function &F, const DominatorTree *DT,
             Changed = true;
           }
         }
-        if (RecursivelyDeleteTriviallyDeadInstructions(I, TLI)) {
+        if (RecursivelyDeleteTriviallyDeadInstructions(I, SQ.TLI)) {
           // RecursivelyDeleteTriviallyDeadInstruction can remove more than one
           // instruction, so simply incrementing the iterator does not work.
           // When instructions get deleted re-iterate instead.
@@ -113,8 +112,9 @@ namespace {
           &getAnalysis<AssumptionCacheTracker>().getAssumptionCache(F);
       OptimizationRemarkEmitter *ORE =
           &getAnalysis<OptimizationRemarkEmitterWrapperPass>().getORE();
-
-      return runImpl(F, DT, TLI, AC, ORE);
+      const DataLayout &DL = F.getParent()->getDataLayout();
+      const SimplifyQuery SQ(DL, TLI, DT, AC);
+      return runImpl(F, SQ, ORE);
     }
   };
 }
@@ -141,7 +141,9 @@ PreservedAnalyses InstSimplifierPass::run(Function &F,
   auto &TLI = AM.getResult<TargetLibraryAnalysis>(F);
   auto &AC = AM.getResult<AssumptionAnalysis>(F);
   auto &ORE = AM.getResult<OptimizationRemarkEmitterAnalysis>(F);
-  bool Changed = runImpl(F, &DT, &TLI, &AC, &ORE);
+  const DataLayout &DL = F.getParent()->getDataLayout();
+  const SimplifyQuery SQ(DL, &TLI, &DT, &AC);
+  bool Changed = runImpl(F, SQ, &ORE);
   if (!Changed)
     return PreservedAnalyses::all();
 
diff --git a/lib/Transforms/Utils/SimplifyLibCalls.cpp b/lib/Transforms/Utils/SimplifyLibCalls.cpp
index aa71e3669ea2..2c1c30463a23 100644
--- a/lib/Transforms/Utils/SimplifyLibCalls.cpp
+++ b/lib/Transforms/Utils/SimplifyLibCalls.cpp
@@ -30,6 +30,7 @@
 #include "llvm/IR/Module.h"
 #include "llvm/IR/PatternMatch.h"
 #include "llvm/Support/CommandLine.h"
+#include "llvm/Support/KnownBits.h"
 #include "llvm/Transforms/Utils/BuildLibCalls.h"
 #include "llvm/Transforms/Utils/Local.h"
 
@@ -37,10 +38,6 @@ using namespace llvm;
 using namespace PatternMatch;
 
 static cl::opt<bool>
-    ColdErrorCalls("error-reporting-is-cold", cl::init(true), cl::Hidden,
-                   cl::desc("Treat error-reporting calls as cold"));
-
-static cl::opt<bool>
     EnableUnsafeFPShrink("enable-double-float-shrink", cl::Hidden,
                          cl::init(false),
                          cl::desc("Enable unsafe double to float "
@@ -459,11 +456,9 @@ Value *LibCallSimplifier::optimizeStrLen(CallInst *CI, IRBuilder<> &B) {
      
       Value *Offset = GEP->getOperand(2);
       unsigned BitWidth = Offset->getType()->getIntegerBitWidth();
-      APInt KnownZero(BitWidth, 0);
-      APInt KnownOne(BitWidth, 0);
-      computeKnownBits(Offset, KnownZero, KnownOne, DL, 0, nullptr, CI, 
-                       nullptr);
-      KnownZero.flipAllBits();
+      KnownBits Known(BitWidth);
+      computeKnownBits(Offset, Known, DL, 0, nullptr, CI, nullptr);
+      Known.Zero.flipAllBits();
       size_t ArrSize = 
              cast<ArrayType>(GEP->getSourceElementType())->getNumElements();
 
@@ -477,7 +472,7 @@ Value *LibCallSimplifier::optimizeStrLen(CallInst *CI, IRBuilder<> &B) {
       // optimize if we can prove that the program has undefined behavior when 
       // Offset is outside that range. That is the case when GEP->getOperand(0) 
       // is a pointer to an object whose memory extent is NullTermIdx+1.
-      if ((KnownZero.isNonNegative() && KnownZero.ule(NullTermIdx)) || 
+      if ((Known.Zero.isNonNegative() && Known.Zero.ule(NullTermIdx)) || 
           (GEP->isInBounds() && isa<GlobalVariable>(GEP->getOperand(0)) &&
            NullTermIdx == ArrSize - 1))
         return B.CreateSub(ConstantInt::get(CI->getType(), NullTermIdx), 
@@ -846,6 +841,9 @@ static Value *foldMallocMemset(CallInst *Memset, IRBuilder<> &B,
 
   // Is the inner call really malloc()?
   Function *InnerCallee = Malloc->getCalledFunction();
+  if (!InnerCallee)
+    return nullptr;
+
   LibFunc Func;
   if (!TLI.getLibFunc(*InnerCallee, Func) || !TLI.has(Func) ||
       Func != LibFunc_malloc)
@@ -930,6 +928,24 @@ static Value *optimizeUnaryDoubleFP(CallInst *CI, IRBuilder<> &B,
   if (V == nullptr)
     return nullptr;
   
+  // If call isn't an intrinsic, check that it isn't within a function with the
+  // same name as the float version of this call.
+  //
+  // e.g. inline float expf(float val) { return (float) exp((double) val); }
+  //
+  // A similar such definition exists in the MinGW-w64 math.h header file which
+  // when compiled with -O2 -ffast-math causes the generation of infinite loops
+  // where expf is called.
+  if (!Callee->isIntrinsic()) {
+    const Function *F = CI->getFunction();
+    StringRef FName = F->getName();
+    StringRef CalleeName = Callee->getName();
+    if ((FName.size() == (CalleeName.size() + 1)) &&
+        (FName.back() == 'f') &&
+        FName.startswith(CalleeName))
+      return nullptr;
+  }
+
   // Propagate fast-math flags from the existing call to the new call.
   IRBuilder<>::FastMathFlagGuard Guard(B);
   B.setFastMathFlags(CI->getFastMathFlags());
@@ -1632,7 +1648,7 @@ Value *LibCallSimplifier::optimizeErrorReporting(CallInst *CI, IRBuilder<> &B,
 }
 
 static bool isReportingError(Function *Callee, CallInst *CI, int StreamArg) {
-  if (!ColdErrorCalls || !Callee || !Callee->isDeclaration())
+  if (!Callee || !Callee->isDeclaration())
     return false;
 
   if (StreamArg < 0)