Vendor import of llvm trunk r302418:

https://llvm.org/svn/llvm-project/llvm/trunk@302418

docs/Lexicon.rst

@@ -38,6 +38,13 @@ B
 **BB Vectorization**
     Basic-Block Vectorization
 
+**BDCE**
+    Bit-tracking dead code elimination. Some bit-wise instructions (shifts,
+    ands, ors, etc.) "kill" some of their input bits -- that is, they make it
+    such that those bits can be either zero or one without affecting control or
+    data flow of a program. The BDCE pass removes instructions that only
+    compute these dead bits.
+
 **BURS**
     Bottom Up Rewriting System --- A method of instruction selection for code
     generation.  An example is the `BURG

docs/MIRLangRef.rst

@@ -78,6 +78,8 @@ Simplifying MIR files
 The MIR code coming out of ``-stop-after``/``-stop-before`` is very verbose;
 Tests are more accessible and future proof when simplified:
 
+- Use the ``-simplify-mir`` option with llc.
+
 - Machine function attributes often have default values or the test works just
   as well with default values. Typical candidates for this are: `alignment:`,
   `exposesReturnsTwice`, `legalized`, `regBankSelected`, `selected`.

examples/Kaleidoscope/BuildingAJIT/Chapter1/toy.cpp

@@ -1092,7 +1092,7 @@ Function *FunctionAST::codegen() {
   TheFunction->eraseFromParent();
 
   if (P.isBinaryOp())
-    BinopPrecedence.erase(Proto->getOperatorName());
+    BinopPrecedence.erase(P.getOperatorName());
   return nullptr;
 }
 

examples/Kaleidoscope/BuildingAJIT/Chapter2/toy.cpp

@@ -1092,7 +1092,7 @@ Function *FunctionAST::codegen() {
   TheFunction->eraseFromParent();
 
   if (P.isBinaryOp())
-    BinopPrecedence.erase(Proto->getOperatorName());
+    BinopPrecedence.erase(P.getOperatorName());
   return nullptr;
 }
 

examples/Kaleidoscope/BuildingAJIT/Chapter3/toy.cpp

@@ -1092,7 +1092,7 @@ Function *FunctionAST::codegen() {
   TheFunction->eraseFromParent();
 
   if (P.isBinaryOp())
-    BinopPrecedence.erase(Proto->getOperatorName());
+    BinopPrecedence.erase(P.getOperatorName());
   return nullptr;
 }
 

examples/Kaleidoscope/Chapter6/toy.cpp

@@ -932,7 +932,7 @@ Function *FunctionAST::codegen() {
   TheFunction->eraseFromParent();
 
   if (P.isBinaryOp())
-    BinopPrecedence.erase(Proto->getOperatorName());
+    BinopPrecedence.erase(P.getOperatorName());
   return nullptr;
 }
 

examples/Kaleidoscope/Chapter7/toy.cpp

@@ -1099,7 +1099,7 @@ Function *FunctionAST::codegen() {
   TheFunction->eraseFromParent();
 
   if (P.isBinaryOp())
-    BinopPrecedence.erase(Proto->getOperatorName());
+    BinopPrecedence.erase(P.getOperatorName());
   return nullptr;
 }
 

examples/Kaleidoscope/Chapter8/toy.cpp

@@ -1097,7 +1097,7 @@ Function *FunctionAST::codegen() {
   TheFunction->eraseFromParent();
 
   if (P.isBinaryOp())
-    BinopPrecedence.erase(Proto->getOperatorName());
+    BinopPrecedence.erase(P.getOperatorName());
   return nullptr;
 }
 

include/llvm/ADT/APInt.h

@@ -842,6 +842,7 @@ public:
   ///
   /// \returns *this
   APInt &operator*=(const APInt &RHS);
+  APInt &operator*=(uint64_t RHS);
 
   /// \brief Addition assignment operator.
   ///
@@ -2043,6 +2044,16 @@ inline APInt operator-(uint64_t LHS, APInt b) {
   return b;
 }
 
+inline APInt operator*(APInt a, uint64_t RHS) {
+  a *= RHS;
+  return a;
+}
+
+inline APInt operator*(uint64_t LHS, APInt b) {
+  b *= LHS;
+  return b;
+}
+
 
 namespace APIntOps {
 

include/llvm/ADT/BitVector.h

@@ -217,7 +217,7 @@ public:
     unsigned BitPos = Prev % BITWORD_SIZE;
     BitWord Copy = Bits[WordPos];
     // Mask off previous bits.
-    Copy &= ~0UL << BitPos;
+    Copy &= maskTrailingZeros<BitWord>(BitPos);
 
     if (Copy != 0)
       return WordPos * BITWORD_SIZE + countTrailingZeros(Copy);
@@ -229,7 +229,7 @@ public:
     return -1;
   }
 
-  /// find_next_unset - Returns the index of the next usnet bit following the
+  /// find_next_unset - Returns the index of the next unset bit following the
   /// "Prev" bit.  Returns -1 if all remaining bits are set.
   int find_next_unset(unsigned Prev) const {
     ++Prev;
@@ -253,7 +253,34 @@ public:
     return -1;
   }
 
-  /// clear - Clear all bits.
+  /// find_prev - Returns the index of the first set bit that precedes the
+  /// the bit at \p PriorTo.  Returns -1 if all previous bits are unset.
+  int find_prev(unsigned PriorTo) {
+    if (PriorTo == 0)
+      return -1;
+
+    --PriorTo;
+
+    unsigned WordPos = PriorTo / BITWORD_SIZE;
+    unsigned BitPos = PriorTo % BITWORD_SIZE;
+    BitWord Copy = Bits[WordPos];
+    // Mask off next bits.
+    Copy &= maskTrailingOnes<BitWord>(BitPos + 1);
+
+    if (Copy != 0)
+      return (WordPos + 1) * BITWORD_SIZE - countLeadingZeros(Copy) - 1;
+
+    // Check previous words.
+    for (unsigned i = 1; i <= WordPos; ++i) {
+      unsigned Index = WordPos - i;
+      if (Bits[Index] == 0)
+        continue;
+      return (Index + 1) * BITWORD_SIZE - countLeadingZeros(Bits[Index]) - 1;
+    }
+    return -1;
+  }
+
+  /// clear - Removes all bits from the bitvector. Does not change capacity.
   void clear() {
     Size = 0;
   }

include/llvm/ADT/SmallBitVector.h

@@ -278,6 +278,24 @@ public:
     return getPointer()->find_next_unset(Prev);
   }
 
+  /// find_prev - Returns the index of the first set bit that precedes the
+  /// the bit at \p PriorTo.  Returns -1 if all previous bits are unset.
+  int find_prev(unsigned PriorTo) const {
+    if (isSmall()) {
+      if (PriorTo == 0)
+        return -1;
+
+      --PriorTo;
+      uintptr_t Bits = getSmallBits();
+      Bits &= maskTrailingOnes<uintptr_t>(PriorTo + 1);
+      if (Bits == 0)
+        return -1;
+
+      return NumBaseBits - countLeadingZeros(Bits) - 1;
+    }
+    return getPointer()->find_prev(PriorTo);
+  }
+
   /// Clear all bits.
   void clear() {
     if (!isSmall())

include/llvm/Analysis/LoopInfoImpl.h

@@ -220,8 +220,8 @@ void LoopBase<BlockT, LoopT>::verifyLoop() const {
     BI = df_ext_begin(getHeader(), VisitSet),
     BE = df_ext_end(getHeader(), VisitSet);
 
-  // Keep track of the number of BBs visited.
-  unsigned NumVisited = 0;
+  // Keep track of the BBs visited.
+  SmallPtrSet<BlockT*, 8> VisitedBBs;
 
   // Check the individual blocks.
   for ( ; BI != BE; ++BI) {
@@ -259,10 +259,18 @@ void LoopBase<BlockT, LoopT>::verifyLoop() const {
     assert(BB != &getHeader()->getParent()->front() &&
            "Loop contains function entry block!");
 
-    NumVisited++;
+    VisitedBBs.insert(BB);
   }
 
-  assert(NumVisited == getNumBlocks() && "Unreachable block in loop");
+  if (VisitedBBs.size() != getNumBlocks()) {
+    dbgs() << "The following blocks are unreachable in the loop: ";
+    for (auto BB : Blocks) {
+      if (!VisitedBBs.count(BB)) {
+        dbgs() << *BB << "\n";
+      }
+    }
+    assert(false && "Unreachable block in loop");
+  }
 
   // Check the subloops.
   for (iterator I = begin(), E = end(); I != E; ++I)

include/llvm/Analysis/ProfileSummaryInfo.h

@@ -54,6 +54,18 @@ public:
   ProfileSummaryInfo(Module &M) : M(M) {}
   ProfileSummaryInfo(ProfileSummaryInfo &&Arg)
       : M(Arg.M), Summary(std::move(Arg.Summary)) {}
+
+  /// Handle the invalidation of this information.
+  ///
+  /// When used as a result of \c ProfileSummaryAnalysis this method will be
+  /// called when the module this was computed for changes. Since profile
+  /// summary is immutable after it is annotated on the module, we return false
+  /// here.
+  bool invalidate(Module &, const PreservedAnalyses &,
+                  ModuleAnalysisManager::Invalidator &) {
+    return false;
+  }
+
   /// Returns the profile count for \p CallInst.
   static Optional<uint64_t> getProfileCount(const Instruction *CallInst,
                                             BlockFrequencyInfo *BFI);

include/llvm/Analysis/ScalarEvolution.h

@@ -782,13 +782,13 @@ private:
 
   /// Set the memoized range for the given SCEV.
   const ConstantRange &setRange(const SCEV *S, RangeSignHint Hint,
-                                const ConstantRange &CR) {
+                                ConstantRange &&CR) {
     DenseMap<const SCEV *, ConstantRange> &Cache =
         Hint == HINT_RANGE_UNSIGNED ? UnsignedRanges : SignedRanges;
 
-    auto Pair = Cache.insert({S, CR});
+    auto Pair = Cache.try_emplace(S, std::move(CR));
     if (!Pair.second)
-      Pair.first->second = CR;
+      Pair.first->second = std::move(CR);
     return Pair.first->second;
   }
 
@@ -816,6 +816,10 @@ private:
   /// Helper function called from createNodeForPHI.
   const SCEV *createAddRecFromPHI(PHINode *PN);
 
+  /// A helper function for createAddRecFromPHI to handle simple cases.
+  const SCEV *createSimpleAffineAddRec(PHINode *PN, Value *BEValueV,
+                                            Value *StartValueV);
+
   /// Helper function called from createNodeForPHI.
   const SCEV *createNodeFromSelectLikePHI(PHINode *PN);
 
@@ -1565,7 +1569,7 @@ public:
   /// delinearization).
   void findArrayDimensions(SmallVectorImpl<const SCEV *> &Terms,
                            SmallVectorImpl<const SCEV *> &Sizes,
-                           const SCEV *ElementSize) const;
+                           const SCEV *ElementSize);
 
   void print(raw_ostream &OS) const;
   void verify() const;

include/llvm/Analysis/TargetLibraryInfo.def

@@ -1115,6 +1115,9 @@ TLI_DEFINE_STRING_INTERNAL("vsprintf")
 /// int vsscanf(const char *s, const char *format, va_list arg);
 TLI_DEFINE_ENUM_INTERNAL(vsscanf)
 TLI_DEFINE_STRING_INTERNAL("vsscanf")
+/// size_t wcslen (const wchar_t* wcs);
+TLI_DEFINE_ENUM_INTERNAL(wcslen)
+TLI_DEFINE_STRING_INTERNAL("wcslen")
 /// ssize_t write(int fildes, const void *buf, size_t nbyte);
 TLI_DEFINE_ENUM_INTERNAL(write)
 TLI_DEFINE_STRING_INTERNAL("write")

include/llvm/CodeGen/AsmPrinter.h

@@ -226,6 +226,7 @@ public:
     FUNCTION_EXIT = 1,
     TAIL_CALL = 2,
     LOG_ARGS_ENTER = 3,
+    CUSTOM_EVENT = 4,
   };
 
   // The table will contain these structs that point to the sled, the function
@@ -242,7 +243,7 @@ public:
   };
 
   // All the sleds to be emitted.
-  std::vector<XRayFunctionEntry> Sleds;
+  SmallVector<XRayFunctionEntry, 4> Sleds;
 
   // Helper function to record a given XRay sled.
   void recordSled(MCSymbol *Sled, const MachineInstr &MI, SledKind Kind);

include/llvm/CodeGen/FastISel.h

@@ -506,6 +506,7 @@ protected:
   bool selectCast(const User *I, unsigned Opcode);
   bool selectExtractValue(const User *I);
   bool selectInsertValue(const User *I);
+  bool selectXRayCustomEvent(const CallInst *II);
 
 private:
   /// \brief Handle PHI nodes in successor blocks.

include/llvm/CodeGen/FunctionLoweringInfo.h

@@ -249,7 +249,7 @@ public:
   void AddLiveOutRegInfo(unsigned Reg, unsigned NumSignBits,
                          const KnownBits &Known) {
     // Only install this information if it tells us something.
-    if (NumSignBits == 1 && Known.Zero == 0 && Known.One == 0)
+    if (NumSignBits == 1 && Known.isUnknown())
       return;
 
     LiveOutRegInfo.grow(Reg);

include/llvm/CodeGen/GlobalISel/IRTranslator.h

@@ -78,7 +78,7 @@ private:
   /// this function.
   DenseMap<const AllocaInst *, int> FrameIndices;
 
-  /// Methods for translating form LLVM IR to MachineInstr.
+  /// \name Methods for translating form LLVM IR to MachineInstr.
   /// \see ::translate for general information on the translate methods.
   /// @{
 

include/llvm/CodeGen/GlobalISel/MachineIRBuilder.h

@@ -45,7 +45,7 @@ class MachineIRBuilder {
   /// Debug location to be set to any instruction we create.
   DebugLoc DL;
 
-  /// Fields describing the insertion point.
+  /// \name Fields describing the insertion point.
   /// @{
   MachineBasicBlock *MBB;
   MachineBasicBlock::iterator II;
@@ -84,7 +84,7 @@ public:
   void setInsertPt(MachineBasicBlock &MBB, MachineBasicBlock::iterator II);
   /// @}
 
-  /// Setters for the insertion point.
+  /// \name Setters for the insertion point.
   /// @{
   /// Set the MachineFunction where to build instructions.
   void setMF(MachineFunction &);
@@ -98,7 +98,7 @@ public:
   void setInstr(MachineInstr &MI);
   /// @}
 
-  /// Control where instructions we create are recorded (typically for
+  /// \name Control where instructions we create are recorded (typically for
   /// visiting again later during legalization).
   /// @{
   void recordInsertions(std::function<void(MachineInstr *)> InsertedInstr);

include/llvm/CodeGen/GlobalISel/RegBankSelect.h

@@ -309,7 +309,7 @@ public:
       Impossible
     };
 
-    /// Convenient types for a list of insertion points.
+    /// \name Convenient types for a list of insertion points.
     /// @{
     typedef SmallVector<std::unique_ptr<InsertPoint>, 2> InsertionPoints;
     typedef InsertionPoints::iterator insertpt_iterator;
@@ -341,7 +341,7 @@ public:
                        const TargetRegisterInfo &TRI, Pass &P,
                        RepairingKind Kind = RepairingKind::Insert);
 
-    /// Getters.
+    /// \name Getters.
     /// @{
     RepairingKind getKind() const { return Kind; }
     unsigned getOpIdx() const { return OpIdx; }
@@ -349,7 +349,7 @@ public:
     bool hasSplit() { return HasSplit; }
     /// @}
 
-    /// Overloaded methods to add an insertion point.
+    /// \name Overloaded methods to add an insertion point.
     /// @{
     /// Add a MBBInsertionPoint to the list of InsertPoints.
     void addInsertPoint(MachineBasicBlock &MBB, bool Beginning);
@@ -362,7 +362,7 @@ public:
     void addInsertPoint(InsertPoint &Point);
     /// @}
 
-    /// Accessors related to the insertion points.
+    /// \name Accessors related to the insertion points.
     /// @{
     insertpt_iterator begin() { return InsertPoints.begin(); }
     insertpt_iterator end() { return InsertPoints.end(); }
@@ -561,7 +561,7 @@ private:
 
   /// Find the best mapping for \p MI from \p PossibleMappings.
   /// \return a reference on the best mapping in \p PossibleMappings.
-  RegisterBankInfo::InstructionMapping &
+  const RegisterBankInfo::InstructionMapping &
   findBestMapping(MachineInstr &MI,
                   RegisterBankInfo::InstructionMappings &PossibleMappings,
                   SmallVectorImpl<RepairingPlacement> &RepairPts);

include/llvm/CodeGen/GlobalISel/RegisterBankInfo.h

@@ -264,7 +264,7 @@ public:
   /// Convenient type to represent the alternatives for mapping an
   /// instruction.
   /// \todo When we move to TableGen this should be an array ref.
-  typedef SmallVector<InstructionMapping, 4> InstructionMappings;
+  typedef SmallVector<const InstructionMapping *, 4> InstructionMappings;
 
   /// Helper class used to get/create the virtual registers that will be used
   /// to replace the MachineOperand when applying a mapping.
@@ -310,7 +310,7 @@ public:
     OperandsMapper(MachineInstr &MI, const InstructionMapping &InstrMapping,
                    MachineRegisterInfo &MRI);
 
-    /// Getters.
+    /// \name Getters.
     /// @{
     /// The MachineInstr being remapped.
     MachineInstr &getMI() const { return MI; }
@@ -378,15 +378,23 @@ protected:
 
   /// Keep dynamically allocated PartialMapping in a separate map.
   /// This shouldn't be needed when everything gets TableGen'ed.
-  mutable DenseMap<unsigned, std::unique_ptr<const PartialMapping>> MapOfPartialMappings;
+  mutable DenseMap<unsigned, std::unique_ptr<const PartialMapping>>
+      MapOfPartialMappings;
 
   /// Keep dynamically allocated ValueMapping in a separate map.
   /// This shouldn't be needed when everything gets TableGen'ed.
-  mutable DenseMap<unsigned, std::unique_ptr<const ValueMapping> > MapOfValueMappings;
+  mutable DenseMap<unsigned, std::unique_ptr<const ValueMapping>>
+      MapOfValueMappings;
 
   /// Keep dynamically allocated array of ValueMapping in a separate map.
   /// This shouldn't be needed when everything gets TableGen'ed.
-  mutable DenseMap<unsigned, std::unique_ptr<ValueMapping[]>> MapOfOperandsMappings;
+  mutable DenseMap<unsigned, std::unique_ptr<ValueMapping[]>>
+      MapOfOperandsMappings;
+
+  /// Keep dynamically allocated InstructionMapping in a separate map.
+  /// This shouldn't be needed when everything gets TableGen'ed.
+  mutable DenseMap<unsigned, std::unique_ptr<const InstructionMapping>>
+      MapOfInstructionMappings;
 
   /// Create a RegisterBankInfo that can accomodate up to \p NumRegBanks
   /// RegisterBank instances.
@@ -425,14 +433,14 @@ protected:
   ///   register, a register class, or a register bank.
   /// In other words, this method will likely fail to find a mapping for
   /// any generic opcode that has not been lowered by target specific code.
-  InstructionMapping getInstrMappingImpl(const MachineInstr &MI) const;
+  const InstructionMapping &getInstrMappingImpl(const MachineInstr &MI) const;
 
   /// Get the uniquely generated PartialMapping for the
   /// given arguments.
   const PartialMapping &getPartialMapping(unsigned StartIdx, unsigned Length,
                                           const RegisterBank &RegBank) const;
 
-  /// Methods to get a uniquely generated ValueMapping.
+  /// \name Methods to get a uniquely generated ValueMapping.
   /// @{
 
   /// The most common ValueMapping consists of a single PartialMapping.
@@ -445,7 +453,7 @@ protected:
                                       unsigned NumBreakDowns) const;
   /// @}
 
-  /// Methods to get a uniquely generated array of ValueMapping.
+  /// \name Methods to get a uniquely generated array of ValueMapping.
   /// @{
 
   /// Get the uniquely generated array of ValueMapping for the
@@ -478,6 +486,33 @@ protected:
       std::initializer_list<const ValueMapping *> OpdsMapping) const;
   /// @}
 
+  /// \name Methods to get a uniquely generated InstructionMapping.
+  /// @{
+
+private:
+  /// Method to get a uniquely generated InstructionMapping.
+  const InstructionMapping &
+  getInstructionMappingImpl(bool IsInvalid, unsigned ID = InvalidMappingID,
+                            unsigned Cost = 0,
+                            const ValueMapping *OperandsMapping = nullptr,
+                            unsigned NumOperands = 0) const;
+
+public:
+  /// Method to get a uniquely generated InstructionMapping.
+  const InstructionMapping &
+  getInstructionMapping(unsigned ID, unsigned Cost,
+                        const ValueMapping *OperandsMapping,
+                        unsigned NumOperands) const {
+    return getInstructionMappingImpl(/*IsInvalid*/ false, ID, Cost,
+                                     OperandsMapping, NumOperands);
+  }
+
+  /// Method to get a uniquely generated invalid InstructionMapping.
+  const InstructionMapping &getInvalidInstructionMapping() const {
+    return getInstructionMappingImpl(/*IsInvalid*/ true);
+  }
+  /// @}
+
   /// Get the register bank for the \p OpIdx-th operand of \p MI form
   /// the encoding constraints, if any.
   ///
@@ -603,7 +638,8 @@ public:
   ///
   /// \note If returnedVal does not verify MI, this would probably mean
   /// that the target does not support that instruction.
-  virtual InstructionMapping getInstrMapping(const MachineInstr &MI) const;
+  virtual const InstructionMapping &
+  getInstrMapping(const MachineInstr &MI) const;
 
   /// Get the alternative mappings for \p MI.
   /// Alternative in the sense different from getInstrMapping.

include/llvm/CodeGen/MIRPrinter.h

@@ -17,9 +17,11 @@
 
 namespace llvm {
 
+class MachineBasicBlock;
 class MachineFunction;
 class Module;
 class raw_ostream;
+template <typename T> class SmallVectorImpl;
 
 /// Print LLVM IR using the MIR serialization format to the given output stream.
 void printMIR(raw_ostream &OS, const Module &M);
@@ -28,6 +30,17 @@ void printMIR(raw_ostream &OS, const Module &M);
 /// output stream.
 void printMIR(raw_ostream &OS, const MachineFunction &MF);
 
+/// Determine a possible list of successors of a basic block based on the
+/// basic block machine operand being used inside the block. This should give
+/// you the correct list of successor blocks in most cases except for things
+/// like jump tables where the basic block references can't easily be found.
+/// The MIRPRinter will skip printing successors if they match the result of
+/// this funciton and the parser will use this function to construct a list if
+/// it is missing.
+void guessSuccessors(const MachineBasicBlock &MBB,
+                     SmallVectorImpl<MachineBasicBlock*> &Successors,
+                     bool &IsFallthrough);
+
 } // end namespace llvm
 
 #endif

include/llvm/CodeGen/MachineFrameInfo.h

@@ -520,6 +520,14 @@ public:
   bool hasTailCall() const { return HasTailCall; }
   void setHasTailCall() { HasTailCall = true; }
 
+  /// Computes the maximum size of a callframe and the AdjustsStack property.
+  /// This only works for targets defining
+  /// TargetInstrInfo::getCallFrameSetupOpcode(), getCallFrameDestroyOpcode(),
+  /// and getFrameSize().
+  /// This is usually computed by the prologue epilogue inserter but some
+  /// targets may call this to compute it earlier.
+  void computeMaxCallFrameSize(const MachineFunction &MF);
+
   /// Return the maximum size of a call frame that must be
   /// allocated for an outgoing function call.  This is only available if
   /// CallFrameSetup/Destroy pseudo instructions are used by the target, and

include/llvm/CodeGen/MachineModuleInfo.h

@@ -116,7 +116,7 @@ class MachineModuleInfo : public ImmutablePass {
 
   // TODO: Ideally, what we'd like is to have a switch that allows emitting 
   // synchronous (precise at call-sites only) CFA into .eh_frame. However,
-  // even under this switch, we'd like .debug_frame to be precise when using.
+  // even under this switch, we'd like .debug_frame to be precise when using
   // -g. At this moment, there's no way to specify that some CFI directives
   // go into .eh_frame only, while others go into .debug_frame only.
 

include/llvm/DebugInfo/CodeView/TypeDatabase.h

@@ -21,7 +21,7 @@ namespace llvm {
 namespace codeview {
 class TypeDatabase {
 public:
-  TypeDatabase() : TypeNameStorage(Allocator) {}
+  explicit TypeDatabase(uint32_t ExpectedSize);
 
   /// Gets the type index for the next type record.
   TypeIndex getNextTypeIndex() const;

include/llvm/DebugInfo/DWARF/DWARFContext.h

@@ -310,6 +310,11 @@ class DWARFContextInMemory : public DWARFContext {
 
   StringRef *MapSectionToMember(StringRef Name);
 
+  /// If Sec is compressed section, decompresses and updates its contents
+  /// provided by Data. Otherwise leaves it unchanged.
+  Error maybeDecompress(const object::SectionRef &Sec, StringRef Name,
+                        StringRef &Data);
+
 public:
   DWARFContextInMemory(const object::ObjectFile &Obj,
     const LoadedObjectInfo *L = nullptr);

include/llvm/DebugInfo/DWARF/DWARFFormValue.h

@@ -39,20 +39,18 @@ public:
 
 private:
   struct ValueType {
-    ValueType() {
-      uval = 0;
-    }
+    ValueType() { uval = 0; }
 
     union {
       uint64_t uval;
       int64_t sval;
-      const char* cstr;
+      const char *cstr;
     };
-    const uint8_t* data = nullptr;
+    const uint8_t *data = nullptr;
   };
 
-  dwarf::Form Form; // Form for this value.
-  ValueType Value; // Contains all data for the form.
+  dwarf::Form Form;             // Form for this value.
+  ValueType Value;              // Contains all data for the form.
   const DWARFUnit *U = nullptr; // Remember the DWARFUnit at extract time.
 
 public:
@@ -84,7 +82,7 @@ public:
                     const DWARFUnit *U);
 
   bool isInlinedCStr() const {
-    return Value.data != nullptr && Value.data == (const uint8_t*)Value.cstr;
+    return Value.data != nullptr && Value.data == (const uint8_t *)Value.cstr;
   }
 
   /// getAsFoo functions below return the extracted value as Foo if only
@@ -135,45 +133,45 @@ public:
                                             uint8_t AddrSize,
                                             llvm::dwarf::DwarfFormat Format);
 
-  /// Skip a form in \p debug_info_data at offset specified by \p offset_ptr.
+  /// Skip a form in \p DebugInfoData at offset specified by \p OffsetPtr.
   ///
   /// Skips the bytes for this form in the debug info and updates the offset.
   ///
-  /// \param debug_info_data the .debug_info data to use to skip the value.
-  /// \param offset_ptr a reference to the offset that will be updated.
+  /// \param DebugInfoData the .debug_info data to use to skip the value.
+  /// \param OffsetPtr a reference to the offset that will be updated.
   /// \param U the DWARFUnit to use when skipping the form in case the form
   /// size differs according to data in the DWARFUnit.
   /// \returns true on success, false if the form was not skipped.
-  bool skipValue(DataExtractor debug_info_data, uint32_t *offset_ptr,
+  bool skipValue(DataExtractor DebugInfoData, uint32_t *OffsetPtr,
                  const DWARFUnit *U) const;
 
-  /// Skip a form in \p debug_info_data at offset specified by \p offset_ptr.
+  /// Skip a form in \p DebugInfoData at offset specified by \p OffsetPtr.
   ///
   /// Skips the bytes for this form in the debug info and updates the offset.
   ///
-  /// \param form the DW_FORM enumeration that indicates the form to skip.
-  /// \param debug_info_data the .debug_info data to use to skip the value.
-  /// \param offset_ptr a reference to the offset that will be updated.
+  /// \param Form the DW_FORM enumeration that indicates the form to skip.
+  /// \param DebugInfoData the .debug_info data to use to skip the value.
+  /// \param OffsetPtr a reference to the offset that will be updated.
   /// \param U the DWARFUnit to use when skipping the form in case the form
   /// size differs according to data in the DWARFUnit.
   /// \returns true on success, false if the form was not skipped.
-  static bool skipValue(dwarf::Form form, DataExtractor debug_info_data,
-                        uint32_t *offset_ptr, const DWARFUnit *U);
+  static bool skipValue(dwarf::Form Form, DataExtractor DebugInfoData,
+                        uint32_t *OffsetPtr, const DWARFUnit *U);
 
-  /// Skip a form in \p debug_info_data at offset specified by \p offset_ptr.
+  /// Skip a form in \p DebugInfoData at offset specified by \p OffsetPtr.
   ///
   /// Skips the bytes for this form in the debug info and updates the offset.
   ///
-  /// \param form the DW_FORM enumeration that indicates the form to skip.
-  /// \param debug_info_data the .debug_info data to use to skip the value.
-  /// \param offset_ptr a reference to the offset that will be updated.
+  /// \param Form the DW_FORM enumeration that indicates the form to skip.
+  /// \param DebugInfoData the .debug_info data to use to skip the value.
+  /// \param OffsetPtr a reference to the offset that will be updated.
   /// \param Version DWARF version number.
   /// \param AddrSize size of an address in bytes.
   /// \param Format enum value from llvm::dwarf::DwarfFormat.
   /// \returns true on success, false if the form was not skipped.
-  static bool skipValue(dwarf::Form form, DataExtractor debug_info_data,
-                        uint32_t *offset_ptr, uint16_t Version,
-                        uint8_t AddrSize, llvm::dwarf::DwarfFormat Format);
+  static bool skipValue(dwarf::Form Form, DataExtractor DebugInfoData,
+                        uint32_t *OffsetPtr, uint16_t Version, uint8_t AddrSize,
+                        llvm::dwarf::DwarfFormat Format);
 
 private:
   void dumpString(raw_ostream &OS) const;
@@ -181,149 +179,146 @@ private:
 
 namespace dwarf {
 
-  /// Take an optional DWARFFormValue and try to extract a string value from it.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \returns an optional value that contains a value if the form value
-  /// was valid and was a string.
-  inline Optional<const char*> toString(const Optional<DWARFFormValue>& V) {
-    if (V)
-      return V->getAsCString();
-    return None;
-  }
-  
-  /// Take an optional DWARFFormValue and extract a string value from it.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \param Default the default value to return in case of failure.
-  /// \returns the string value or Default if the V doesn't have a value or the
-  /// form value's encoding wasn't a string.
-  inline const char*
-  toString(const Optional<DWARFFormValue>& V, const char *Default) {
-    return toString(V).getValueOr(Default);
-  }
+/// Take an optional DWARFFormValue and try to extract a string value from it.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \returns an optional value that contains a value if the form value
+/// was valid and was a string.
+inline Optional<const char *> toString(const Optional<DWARFFormValue> &V) {
+  if (V)
+    return V->getAsCString();
+  return None;
+}
 
-  /// Take an optional DWARFFormValue and try to extract an unsigned constant.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \returns an optional value that contains a value if the form value
-  /// was valid and has a unsigned constant form.
-  inline Optional<uint64_t> toUnsigned(const Optional<DWARFFormValue>& V) {
-    if (V)
-      return V->getAsUnsignedConstant();
-    return None;
-  }
-  
-  /// Take an optional DWARFFormValue and extract a unsigned constant.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \param Default the default value to return in case of failure.
-  /// \returns the extracted unsigned value or Default if the V doesn't have a
-  /// value or the form value's encoding wasn't an unsigned constant form.
-  inline uint64_t
-  toUnsigned(const Optional<DWARFFormValue>& V, uint64_t Default) {
-    return toUnsigned(V).getValueOr(Default);
-  }
-  
-  /// Take an optional DWARFFormValue and try to extract an reference.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \returns an optional value that contains a value if the form value
-  /// was valid and has a reference form.
-  inline Optional<uint64_t> toReference(const Optional<DWARFFormValue>& V) {
-    if (V)
-      return V->getAsReference();
-    return None;
-  }
-  
-  /// Take an optional DWARFFormValue and extract a reference.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \param Default the default value to return in case of failure.
-  /// \returns the extracted reference value or Default if the V doesn't have a
-  /// value or the form value's encoding wasn't a reference form.
-  inline uint64_t
-  toReference(const Optional<DWARFFormValue>& V, uint64_t Default) {
-    return toReference(V).getValueOr(Default);
-  }
-  
-  /// Take an optional DWARFFormValue and try to extract an signed constant.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \returns an optional value that contains a value if the form value
-  /// was valid and has a signed constant form.
-  inline Optional<int64_t> toSigned(const Optional<DWARFFormValue>& V) {
-    if (V)
-      return V->getAsSignedConstant();
-    return None;
-  }
+/// Take an optional DWARFFormValue and extract a string value from it.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \param Default the default value to return in case of failure.
+/// \returns the string value or Default if the V doesn't have a value or the
+/// form value's encoding wasn't a string.
+inline const char *toString(const Optional<DWARFFormValue> &V,
+                            const char *Default) {
+  return toString(V).getValueOr(Default);
+}
 
-  /// Take an optional DWARFFormValue and extract a signed integer.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \param Default the default value to return in case of failure.
-  /// \returns the extracted signed integer value or Default if the V doesn't
-  /// have a value or the form value's encoding wasn't a signed integer form.
-  inline int64_t
-  toSigned(const Optional<DWARFFormValue>& V, int64_t Default) {
-    return toSigned(V).getValueOr(Default);
-  }
+/// Take an optional DWARFFormValue and try to extract an unsigned constant.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \returns an optional value that contains a value if the form value
+/// was valid and has a unsigned constant form.
+inline Optional<uint64_t> toUnsigned(const Optional<DWARFFormValue> &V) {
+  if (V)
+    return V->getAsUnsignedConstant();
+  return None;
+}
 
-  /// Take an optional DWARFFormValue and try to extract an address.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \returns an optional value that contains a value if the form value
-  /// was valid and has a address form.
-  inline Optional<uint64_t> toAddress(const Optional<DWARFFormValue>& V) {
-    if (V)
-      return V->getAsAddress();
-    return None;
-  }
+/// Take an optional DWARFFormValue and extract a unsigned constant.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \param Default the default value to return in case of failure.
+/// \returns the extracted unsigned value or Default if the V doesn't have a
+/// value or the form value's encoding wasn't an unsigned constant form.
+inline uint64_t toUnsigned(const Optional<DWARFFormValue> &V,
+                           uint64_t Default) {
+  return toUnsigned(V).getValueOr(Default);
+}
 
-  /// Take an optional DWARFFormValue and extract a address.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \param Default the default value to return in case of failure.
-  /// \returns the extracted address value or Default if the V doesn't have a
-  /// value or the form value's encoding wasn't an address form.
-  inline uint64_t
-  toAddress(const Optional<DWARFFormValue>& V, uint64_t Default) {
-    return toAddress(V).getValueOr(Default);
-  }
+/// Take an optional DWARFFormValue and try to extract an reference.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \returns an optional value that contains a value if the form value
+/// was valid and has a reference form.
+inline Optional<uint64_t> toReference(const Optional<DWARFFormValue> &V) {
+  if (V)
+    return V->getAsReference();
+  return None;
+}
 
-  /// Take an optional DWARFFormValue and try to extract an section offset.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \returns an optional value that contains a value if the form value
-  /// was valid and has a section offset form.
-  inline Optional<uint64_t> toSectionOffset(const Optional<DWARFFormValue>& V) {
-    if (V)
-      return V->getAsSectionOffset();
-    return None;
-  }
+/// Take an optional DWARFFormValue and extract a reference.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \param Default the default value to return in case of failure.
+/// \returns the extracted reference value or Default if the V doesn't have a
+/// value or the form value's encoding wasn't a reference form.
+inline uint64_t toReference(const Optional<DWARFFormValue> &V,
+                            uint64_t Default) {
+  return toReference(V).getValueOr(Default);
+}
 
-  /// Take an optional DWARFFormValue and extract a section offset.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \param Default the default value to return in case of failure.
-  /// \returns the extracted section offset value or Default if the V doesn't
-  /// have a value or the form value's encoding wasn't a section offset form.
-  inline uint64_t
-  toSectionOffset(const Optional<DWARFFormValue>& V, uint64_t Default) {
-    return toSectionOffset(V).getValueOr(Default);
-  }
+/// Take an optional DWARFFormValue and try to extract an signed constant.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \returns an optional value that contains a value if the form value
+/// was valid and has a signed constant form.
+inline Optional<int64_t> toSigned(const Optional<DWARFFormValue> &V) {
+  if (V)
+    return V->getAsSignedConstant();
+  return None;
+}
 
-  /// Take an optional DWARFFormValue and try to extract block data.
-  ///
-  /// \param V and optional DWARFFormValue to attempt to extract the value from.
-  /// \returns an optional value that contains a value if the form value
-  /// was valid and has a block form.
-  inline Optional<ArrayRef<uint8_t>>
-  toBlock(const Optional<DWARFFormValue>& V) {
-    if (V)
-      return V->getAsBlock();
-    return None;
-  }
+/// Take an optional DWARFFormValue and extract a signed integer.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \param Default the default value to return in case of failure.
+/// \returns the extracted signed integer value or Default if the V doesn't
+/// have a value or the form value's encoding wasn't a signed integer form.
+inline int64_t toSigned(const Optional<DWARFFormValue> &V, int64_t Default) {
+  return toSigned(V).getValueOr(Default);
+}
+
+/// Take an optional DWARFFormValue and try to extract an address.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \returns an optional value that contains a value if the form value
+/// was valid and has a address form.
+inline Optional<uint64_t> toAddress(const Optional<DWARFFormValue> &V) {
+  if (V)
+    return V->getAsAddress();
+  return None;
+}
+
+/// Take an optional DWARFFormValue and extract a address.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \param Default the default value to return in case of failure.
+/// \returns the extracted address value or Default if the V doesn't have a
+/// value or the form value's encoding wasn't an address form.
+inline uint64_t toAddress(const Optional<DWARFFormValue> &V, uint64_t Default) {
+  return toAddress(V).getValueOr(Default);
+}
+
+/// Take an optional DWARFFormValue and try to extract an section offset.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \returns an optional value that contains a value if the form value
+/// was valid and has a section offset form.
+inline Optional<uint64_t> toSectionOffset(const Optional<DWARFFormValue> &V) {
+  if (V)
+    return V->getAsSectionOffset();
+  return None;
+}
+
+/// Take an optional DWARFFormValue and extract a section offset.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \param Default the default value to return in case of failure.
+/// \returns the extracted section offset value or Default if the V doesn't
+/// have a value or the form value's encoding wasn't a section offset form.
+inline uint64_t toSectionOffset(const Optional<DWARFFormValue> &V,
+                                uint64_t Default) {
+  return toSectionOffset(V).getValueOr(Default);
+}
+
+/// Take an optional DWARFFormValue and try to extract block data.
+///
+/// \param V and optional DWARFFormValue to attempt to extract the value from.
+/// \returns an optional value that contains a value if the form value
+/// was valid and has a block form.
+inline Optional<ArrayRef<uint8_t>> toBlock(const Optional<DWARFFormValue> &V) {
+  if (V)
+    return V->getAsBlock();
+  return None;
+}
 
 } // end namespace dwarf
 

include/llvm/DebugInfo/PDB/Native/DbiModuleDescriptor.h

@@ -53,14 +53,6 @@ private:
   const ModuleInfoHeader *Layout = nullptr;
 };
 
-struct ModuleInfoEx {
-  ModuleInfoEx(const DbiModuleDescriptor &Info) : Info(Info) {}
-  ModuleInfoEx(const ModuleInfoEx &Ex) = default;
-
-  DbiModuleDescriptor Info;
-  std::vector<StringRef> SourceFiles;
-};
-
 } // end namespace pdb
 
 template <> struct VarStreamArrayExtractor<pdb::DbiModuleDescriptor> {

include/llvm/DebugInfo/PDB/Native/DbiModuleList.h

@@ -0,0 +1,114 @@
+//===- DbiModuleList.h - PDB module information list ------------*- C++ -*-===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_DEBUGINFO_PDB_RAW_DBIMODULELIST_H
+#define LLVM_DEBUGINFO_PDB_RAW_DBIMODULELIST_H
+
+#include "llvm/ADT/StringRef.h"
+#include "llvm/ADT/iterator.h"
+#include "llvm/DebugInfo/PDB/Native/DbiModuleDescriptor.h"
+#include "llvm/Support/BinaryStreamArray.h"
+#include "llvm/Support/BinaryStreamRef.h"
+#include "llvm/Support/Endian.h"
+#include "llvm/Support/Error.h"
+#include <cstdint>
+#include <vector>
+
+namespace llvm {
+namespace pdb {
+
+class DbiModuleList;
+struct FileInfoSubstreamHeader;
+
+class DbiModuleSourceFilesIterator
+    : public iterator_facade_base<DbiModuleSourceFilesIterator,
+                                  std::random_access_iterator_tag, StringRef> {
+  typedef iterator_facade_base<DbiModuleSourceFilesIterator,
+                               std::random_access_iterator_tag, StringRef>
+      BaseType;
+
+public:
+  DbiModuleSourceFilesIterator(const DbiModuleList &Modules, uint32_t Modi,
+                               uint16_t Filei);
+  DbiModuleSourceFilesIterator() = default;
+  DbiModuleSourceFilesIterator &
+  operator=(const DbiModuleSourceFilesIterator &R) = default;
+
+  bool operator==(const DbiModuleSourceFilesIterator &R) const;
+
+  const StringRef &operator*() const { return ThisValue; }
+  StringRef &operator*() { return ThisValue; }
+
+  bool operator<(const DbiModuleSourceFilesIterator &RHS) const;
+  std::ptrdiff_t operator-(const DbiModuleSourceFilesIterator &R) const;
+  DbiModuleSourceFilesIterator &operator+=(std::ptrdiff_t N);
+  DbiModuleSourceFilesIterator &operator-=(std::ptrdiff_t N);
+
+private:
+  void setValue();
+
+  bool isEnd() const;
+  bool isCompatible(const DbiModuleSourceFilesIterator &R) const;
+  bool isUniversalEnd() const;
+
+  StringRef ThisValue;
+  const DbiModuleList *Modules{nullptr};
+  uint32_t Modi{0};
+  uint16_t Filei{0};
+};
+
+class DbiModuleList {
+  friend DbiModuleSourceFilesIterator;
+
+public:
+  Error initialize(BinaryStreamRef ModInfo, BinaryStreamRef FileInfo);
+
+  Expected<StringRef> getFileName(uint32_t Index) const;
+  uint32_t getModuleCount() const;
+  uint32_t getSourceFileCount() const;
+  uint16_t getSourceFileCount(uint32_t Modi) const;
+
+  iterator_range<DbiModuleSourceFilesIterator>
+  source_files(uint32_t Modi) const;
+
+  DbiModuleDescriptor getModuleDescriptor(uint32_t Modi) const;
+
+private:
+  Error initializeModInfo(BinaryStreamRef ModInfo);
+  Error initializeFileInfo(BinaryStreamRef FileInfo);
+
+  VarStreamArray<DbiModuleDescriptor> Descriptors;
+
+  FixedStreamArray<support::little32_t> FileNameOffsets;
+  FixedStreamArray<support::ulittle16_t> ModFileCountArray;
+
+  // For each module, there are multiple filenames, which can be obtained by
+  // knowing the index of the file.  Given the index of the file, one can use
+  // that as an offset into the FileNameOffsets array, which contains the
+  // absolute offset of the file name in NamesBuffer.  Thus, for each module
+  // we store the first index in the FileNameOffsets array for this module.
+  // The number of files for the corresponding module is stored in
+  // ModFileCountArray.
+  std::vector<uint32_t> ModuleInitialFileIndex;
+
+  // In order to provide random access into the Descriptors array, we iterate it
+  // once up front to find the offsets of the individual items and store them in
+  // this array.
+  std::vector<uint32_t> ModuleDescriptorOffsets;
+
+  const FileInfoSubstreamHeader *FileInfoHeader = nullptr;
+
+  BinaryStreamRef ModInfoSubstream;
+  BinaryStreamRef FileInfoSubstream;
+  BinaryStreamRef NamesBuffer;
+};
+}
+}
+
+#endif // LLVM_DEBUGINFO_PDB_RAW_DBIMODULELIST_H

include/llvm/DebugInfo/PDB/Native/DbiStream.h

@@ -13,6 +13,7 @@
 #include "llvm/DebugInfo/CodeView/ModuleDebugFragment.h"
 #include "llvm/DebugInfo/MSF/MappedBlockStream.h"
 #include "llvm/DebugInfo/PDB/Native/DbiModuleDescriptor.h"
+#include "llvm/DebugInfo/PDB/Native/DbiModuleList.h"
 #include "llvm/DebugInfo/PDB/Native/PDBStringTable.h"
 #include "llvm/DebugInfo/PDB/Native/RawConstants.h"
 #include "llvm/DebugInfo/PDB/Native/RawTypes.h"
@@ -68,9 +69,7 @@ public:
   /// not present, returns InvalidStreamIndex.
   uint32_t getDebugStreamIndex(DbgHeaderType Type) const;
 
-  ArrayRef<ModuleInfoEx> modules() const;
-
-  Expected<StringRef> getFileNameForIndex(uint32_t Index) const;
+  const DbiModuleList &modules() const;
 
   FixedStreamArray<object::coff_section> getSectionHeaders();
 
@@ -80,27 +79,22 @@ public:
   void visitSectionContributions(ISectionContribVisitor &Visitor) const;
 
 private:
-  Error initializeModInfoArray();
   Error initializeSectionContributionData();
   Error initializeSectionHeadersData();
   Error initializeSectionMapData();
-  Error initializeFileInfo();
   Error initializeFpoRecords();
 
   PDBFile &Pdb;
   std::unique_ptr<msf::MappedBlockStream> Stream;
 
-  std::vector<ModuleInfoEx> ModuleInfos;
   PDBStringTable ECNames;
 
-  BinaryStreamRef ModInfoSubstream;
   BinaryStreamRef SecContrSubstream;
   BinaryStreamRef SecMapSubstream;
-  BinaryStreamRef FileInfoSubstream;
   BinaryStreamRef TypeServerMapSubstream;
   BinaryStreamRef ECSubstream;
 
-  BinaryStreamRef NamesBuffer;
+  DbiModuleList Modules;
 
   FixedStreamArray<support::ulittle16_t> DbgStreams;
 
@@ -108,7 +102,6 @@ private:
   FixedStreamArray<SectionContrib> SectionContribs;
   FixedStreamArray<SectionContrib2> SectionContribs2;
   FixedStreamArray<SecMapEntry> SectionMap;
-  FixedStreamArray<support::little32_t> FileNameOffsets;
 
   std::unique_ptr<msf::MappedBlockStream> SectionHeaderStream;
   FixedStreamArray<object::coff_section> SectionHeaders;

include/llvm/DebugInfo/PDB/Native/NativeCompilandSymbol.h

@@ -18,7 +18,7 @@ namespace pdb {
 
 class NativeCompilandSymbol : public NativeRawSymbol {
 public:
-  NativeCompilandSymbol(NativeSession &Session, const ModuleInfoEx &MI);
+  NativeCompilandSymbol(NativeSession &Session, DbiModuleDescriptor MI);
   PDB_SymType getSymTag() const override;
   bool isEditAndContinueEnabled() const override;
   uint32_t getLexicalParentId() const override;
@@ -26,7 +26,7 @@ public:
   std::string getName() const override;
 
 private:
-  ModuleInfoEx Module;
+  DbiModuleDescriptor Module;
 };
 
 } // namespace pdb

include/llvm/DebugInfo/PDB/Native/NativeEnumModules.h

@@ -16,13 +16,13 @@
 namespace llvm {
 namespace pdb {
 
+class DbiModuleList;
 class NativeSession;
 
 class NativeEnumModules : public IPDBEnumChildren<PDBSymbol> {
 public:
-  explicit NativeEnumModules(NativeSession &Session,
-                             ArrayRef<ModuleInfoEx> Modules,
-                             uint32_t Index = 0);
+  NativeEnumModules(NativeSession &Session, const DbiModuleList &Modules,
+                    uint32_t Index = 0);
 
   uint32_t getChildCount() const override;
   std::unique_ptr<PDBSymbol> getChildAtIndex(uint32_t Index) const override;
@@ -32,7 +32,7 @@ public:
 
 private:
   NativeSession &Session;
-  ArrayRef<ModuleInfoEx> Modules;
+  const DbiModuleList &Modules;
   uint32_t Index;
 };
 }

include/llvm/DebugInfo/PDB/Native/RawTypes.h

@@ -211,7 +211,7 @@ struct ModInfoFlags {
 };
 
 /// The header preceeding each entry in the Module Info substream of the DBI
-/// stream.
+/// stream.  Corresponds to the type MODI in the reference implementation.
 struct ModuleInfoHeader {
   /// Currently opened module. This field is a pointer in the reference
   /// implementation, but that won't work on 64-bit systems, and anyway it
@@ -243,9 +243,12 @@ struct ModuleInfoHeader {
   /// Padding so the next field is 4-byte aligned.
   char Padding1[2];
 
-  /// Array of [0..NumFiles) DBI name buffer offsets.  This field is a pointer
-  /// in the reference implementation, but as with `Mod`, we ignore it for now
-  /// since it is unused.
+  /// Array of [0..NumFiles) DBI name buffer offsets.  In the reference
+  /// implementation this field is a pointer.  But since you can't portably
+  /// serialize a pointer, on 64-bit platforms they copy all the values except
+  /// this one into the 32-bit version of the struct and use that for
+  /// serialization.  Regardless, this field is unused, it is only there to
+  /// store a pointer that can be accessed at runtime.
   support::ulittle32_t FileNameOffs;
 
   /// Name Index for src file name

include/llvm/DebugInfo/PDB/Native/TpiStream.h

@@ -40,12 +40,12 @@ public:
 
   uint32_t TypeIndexBegin() const;
   uint32_t TypeIndexEnd() const;
-  uint32_t NumTypeRecords() const;
+  uint32_t getNumTypeRecords() const;
   uint16_t getTypeHashStreamIndex() const;
   uint16_t getTypeHashStreamAuxIndex() const;
 
   uint32_t getHashKeySize() const;
-  uint32_t NumHashBuckets() const;
+  uint32_t getNumHashBuckets() const;
   FixedStreamArray<support::ulittle32_t> getHashValues() const;
   FixedStreamArray<TypeIndexOffset> getTypeIndexOffsets() const;
   HashTable &getHashAdjusters();
@@ -55,8 +55,6 @@ public:
   Error commit();
 
 private:
-  Error verifyHashValues();
-
   const PDBFile &Pdb;
   std::unique_ptr<msf::MappedBlockStream> Stream;
 

include/llvm/ExecutionEngine/Orc/RPCSerialization.h

@@ -355,7 +355,7 @@ public:
                                             std::move(Deserialize)));
       KeyName = &I->first;
     }
-    
+
     {
       assert(KeyName != nullptr && "No keyname pointer");
       std::lock_guard<std::recursive_mutex> Lock(SerializersMutex);
@@ -370,7 +370,7 @@ public:
         };
     }
   }
-  
+
   static Error serialize(ChannelT &C, Error &&Err) {
     std::lock_guard<std::recursive_mutex> Lock(SerializersMutex);
 

include/llvm/ExecutionEngine/RuntimeDyldChecker.h

@@ -10,6 +10,8 @@
 #ifndef LLVM_EXECUTIONENGINE_RUNTIMEDYLDCHECKER_H
 #define LLVM_EXECUTIONENGINE_RUNTIMEDYLDCHECKER_H
 
+#include "llvm/ADT/Optional.h"
+
 #include <cstdint>
 #include <memory>
 #include <string>
@@ -97,6 +99,10 @@ public:
                                                   StringRef SectionName,
                                                   bool LocalAddress);
 
+  /// \brief If there is a section at the given local address, return its load
+  ///        address, otherwise return none.
+  Optional<uint64_t> getSectionLoadAddress(void *LocalAddress) const;
+
 private:
   std::unique_ptr<RuntimeDyldCheckerImpl> Impl;
 };

include/llvm/IR/Attributes.h

@@ -244,7 +244,8 @@ public:
   std::pair<unsigned, Optional<unsigned>> getAllocSizeArgs() const;
   std::string getAsString(bool InAttrGrp = false) const;
 
-  typedef const Attribute *iterator;
+  using iterator = const Attribute *;
+
   iterator begin() const;
   iterator end() const;
 };
@@ -479,7 +480,7 @@ public:
   /// \brief Return the attributes at the index as a string.
   std::string getAsString(unsigned Index, bool InAttrGrp = false) const;
 
-  typedef ArrayRef<Attribute>::iterator iterator;
+  using iterator = ArrayRef<Attribute>::iterator;
 
   iterator begin(unsigned Slot) const;
   iterator end(unsigned Slot) const;
@@ -662,11 +663,11 @@ public:
   bool empty() const { return Attrs.none(); }
 
   // Iterators for target-dependent attributes.
-  typedef std::pair<std::string, std::string>                td_type;
-  typedef std::map<std::string, std::string>::iterator       td_iterator;
-  typedef std::map<std::string, std::string>::const_iterator td_const_iterator;
-  typedef iterator_range<td_iterator>                        td_range;
-  typedef iterator_range<td_const_iterator>                  td_const_range;
+  using td_type = std::pair<std::string, std::string>;
+  using td_iterator = std::map<std::string, std::string>::iterator;
+  using td_const_iterator = std::map<std::string, std::string>::const_iterator;
+  using td_range = iterator_range<td_iterator>;
+  using td_const_range = iterator_range<td_const_iterator>;
 
   td_iterator td_begin()             { return TargetDepAttrs.begin(); }
   td_iterator td_end()               { return TargetDepAttrs.end(); }

include/llvm/IR/BasicBlock.h

@@ -21,6 +21,7 @@
 #include "llvm/IR/SymbolTableListTraits.h"
 #include "llvm/IR/Value.h"
 #include "llvm/Support/CBindingWrapping.h"
+#include "llvm/Support/Compiler.h"
 #include "llvm-c/Types.h"
 #include <cassert>
 #include <cstddef>
@@ -31,7 +32,9 @@ class CallInst;
 class Function;
 class LandingPadInst;
 class LLVMContext;
+class Module;
 class TerminatorInst;
+class ValueSymbolTable;
 
 /// \brief LLVM Basic Block Representation
 ///
@@ -51,7 +54,7 @@ class TerminatorInst;
 class BasicBlock : public Value, // Basic blocks are data objects also
                    public ilist_node_with_parent<BasicBlock, Function> {
 public:
-  typedef SymbolTableList<Instruction> InstListType;
+  using InstListType = SymbolTableList<Instruction>;
 
 private:
   friend class BlockAddress;
@@ -80,10 +83,10 @@ public:
   LLVMContext &getContext() const;
 
   /// Instruction iterators...
-  typedef InstListType::iterator iterator;
-  typedef InstListType::const_iterator const_iterator;
-  typedef InstListType::reverse_iterator reverse_iterator;
-  typedef InstListType::const_reverse_iterator const_reverse_iterator;
+  using iterator = InstListType::iterator;
+  using const_iterator = InstListType::const_iterator;
+  using reverse_iterator = InstListType::reverse_iterator;
+  using const_reverse_iterator = InstListType::const_reverse_iterator;
 
   /// \brief Creates a new BasicBlock.
   ///

include/llvm/IR/CFG.h

@@ -37,9 +37,9 @@ namespace llvm {
 template <class Ptr, class USE_iterator> // Predecessor Iterator
 class PredIterator : public std::iterator<std::forward_iterator_tag,
                                           Ptr, ptrdiff_t, Ptr*, Ptr*> {
-  typedef std::iterator<std::forward_iterator_tag, Ptr, ptrdiff_t, Ptr*,
-                                                                    Ptr*> super;
-  typedef PredIterator<Ptr, USE_iterator> Self;
+  using super =
+      std::iterator<std::forward_iterator_tag, Ptr, ptrdiff_t, Ptr*, Ptr*>;
+  using Self = PredIterator<Ptr, USE_iterator>;
   USE_iterator It;
 
   inline void advancePastNonTerminators() {
@@ -49,8 +49,8 @@ class PredIterator : public std::iterator<std::forward_iterator_tag,
   }
 
 public:
-  typedef typename super::pointer pointer;
-  typedef typename super::reference reference;
+  using pointer = typename super::pointer;
+  using reference = typename super::reference;
 
   PredIterator() = default;
   explicit inline PredIterator(Ptr *bb) : It(bb->user_begin()) {
@@ -90,11 +90,11 @@ public:
   }
 };
 
-typedef PredIterator<BasicBlock, Value::user_iterator> pred_iterator;
-typedef PredIterator<const BasicBlock,
-                     Value::const_user_iterator> const_pred_iterator;
-typedef iterator_range<pred_iterator> pred_range;
-typedef iterator_range<const_pred_iterator> pred_const_range;
+using pred_iterator = PredIterator<BasicBlock, Value::user_iterator>;
+using const_pred_iterator =
+    PredIterator<const BasicBlock, Value::const_user_iterator>;
+using pred_range = iterator_range<pred_iterator>;
+using pred_const_range = iterator_range<const_pred_iterator>;
 
 inline pred_iterator pred_begin(BasicBlock *BB) { return pred_iterator(BB); }
 inline const_pred_iterator pred_begin(const BasicBlock *BB) {
@@ -118,12 +118,12 @@ inline pred_const_range predecessors(const BasicBlock *BB) {
 // BasicBlock succ_iterator helpers
 //===----------------------------------------------------------------------===//
 
-typedef TerminatorInst::SuccIterator<TerminatorInst *, BasicBlock>
-    succ_iterator;
-typedef TerminatorInst::SuccIterator<const TerminatorInst *, const BasicBlock>
-    succ_const_iterator;
-typedef iterator_range<succ_iterator> succ_range;
-typedef iterator_range<succ_const_iterator> succ_const_range;
+using succ_iterator =
+    TerminatorInst::SuccIterator<TerminatorInst *, BasicBlock>;
+using succ_const_iterator =
+    TerminatorInst::SuccIterator<const TerminatorInst *, const BasicBlock>;
+using succ_range = iterator_range<succ_iterator>;
+using succ_const_range = iterator_range<succ_const_iterator>;
 
 inline succ_iterator succ_begin(BasicBlock *BB) {
   return succ_iterator(BB->getTerminator());
@@ -160,8 +160,8 @@ struct isPodLike<TerminatorInst::SuccIterator<T, U>> {
 // graph of basic blocks...
 
 template <> struct GraphTraits<BasicBlock*> {
-  typedef BasicBlock *NodeRef;
-  typedef succ_iterator ChildIteratorType;
+  using NodeRef = BasicBlock *;
+  using ChildIteratorType = succ_iterator;
 
   static NodeRef getEntryNode(BasicBlock *BB) { return BB; }
   static ChildIteratorType child_begin(NodeRef N) { return succ_begin(N); }
@@ -169,8 +169,8 @@ template <> struct GraphTraits<BasicBlock*> {
 };
 
 template <> struct GraphTraits<const BasicBlock*> {
-  typedef const BasicBlock *NodeRef;
-  typedef succ_const_iterator ChildIteratorType;
+  using NodeRef = const BasicBlock *;
+  using ChildIteratorType = succ_const_iterator;
 
   static NodeRef getEntryNode(const BasicBlock *BB) { return BB; }
 
@@ -184,16 +184,18 @@ template <> struct GraphTraits<const BasicBlock*> {
 // instead of the successor edges.
 //
 template <> struct GraphTraits<Inverse<BasicBlock*>> {
-  typedef BasicBlock *NodeRef;
-  typedef pred_iterator ChildIteratorType;
+  using NodeRef = BasicBlock *;
+  using ChildIteratorType = pred_iterator;
+
   static NodeRef getEntryNode(Inverse<BasicBlock *> G) { return G.Graph; }
   static ChildIteratorType child_begin(NodeRef N) { return pred_begin(N); }
   static ChildIteratorType child_end(NodeRef N) { return pred_end(N); }
 };
 
 template <> struct GraphTraits<Inverse<const BasicBlock*>> {
-  typedef const BasicBlock *NodeRef;
-  typedef const_pred_iterator ChildIteratorType;
+  using NodeRef = const BasicBlock *;
+  using ChildIteratorType = const_pred_iterator;
+
   static NodeRef getEntryNode(Inverse<const BasicBlock *> G) { return G.Graph; }
   static ChildIteratorType child_begin(NodeRef N) { return pred_begin(N); }
   static ChildIteratorType child_end(NodeRef N) { return pred_end(N); }
@@ -211,7 +213,7 @@ template <> struct GraphTraits<Function*> : public GraphTraits<BasicBlock*> {
   static NodeRef getEntryNode(Function *F) { return &F->getEntryBlock(); }
 
   // nodes_iterator/begin/end - Allow iteration over all nodes in the graph
-  typedef pointer_iterator<Function::iterator> nodes_iterator;
+  using nodes_iterator = pointer_iterator<Function::iterator>;
 
   static nodes_iterator nodes_begin(Function *F) {
     return nodes_iterator(F->begin());
@@ -228,7 +230,7 @@ template <> struct GraphTraits<const Function*> :
   static NodeRef getEntryNode(const Function *F) { return &F->getEntryBlock(); }
 
   // nodes_iterator/begin/end - Allow iteration over all nodes in the graph
-  typedef pointer_iterator<Function::const_iterator> nodes_iterator;
+  using nodes_iterator = pointer_iterator<Function::const_iterator>;
 
   static nodes_iterator nodes_begin(const Function *F) {
     return nodes_iterator(F->begin());

include/llvm/IR/CallSite.h

@@ -207,7 +207,7 @@ public:
 
   /// The type of iterator to use when looping over actual arguments at this
   /// call site.
-  typedef IterTy arg_iterator;
+  using arg_iterator = IterTy;
 
   iterator_range<IterTy> args() const {
     return make_range(arg_begin(), arg_end());
@@ -231,7 +231,7 @@ public:
 
   /// Type of iterator to use when looping over data operands at this call site
   /// (see below).
-  typedef IterTy data_operand_iterator;
+  using data_operand_iterator = IterTy;
 
   /// data_operands_begin/data_operands_end - Return iterators iterating over
   /// the call / invoke argument list and bundle operands.  For invokes, this is

include/llvm/IR/CallingConv.h

@@ -1,4 +1,4 @@
-//===-- llvm/CallingConv.h - LLVM Calling Conventions -----------*- C++ -*-===//
+//===- llvm/CallingConv.h - LLVM Calling Conventions ------------*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
@@ -20,8 +20,9 @@ namespace llvm {
 /// the well-known calling conventions.
 ///
 namespace CallingConv {
+
   /// LLVM IR allows to use arbitrary numbers as calling convention identifiers.
-  typedef unsigned ID;
+  using ID = unsigned;
 
   /// A set of enums which specify the assigned numeric values for known llvm
   /// calling conventions.
@@ -203,8 +204,9 @@ namespace CallingConv {
     /// The highest possible calling convention ID. Must be some 2^k - 1.
     MaxID = 1023
   };
-} // End CallingConv namespace
 
-} // End llvm namespace
+} // end namespace CallingConv
 
-#endif
+} // end namespace llvm
+
+#endif // LLVM_IR_CALLINGCONV_H

include/llvm/IR/ConstantRange.h

@@ -41,7 +41,7 @@ namespace llvm {
 class MDNode;
 
 /// This class represents a range of values.
-class ConstantRange {
+class LLVM_NODISCARD ConstantRange {
   APInt Lower, Upper;
 
 public:
@@ -167,7 +167,10 @@ public:
   APInt getSetSize() const;
 
   /// Compare set size of this range with the range CR.
-  bool isSizeStrictlySmallerThanOf(const ConstantRange &CR) const;
+  bool isSizeStrictlySmallerThan(const ConstantRange &CR) const;
+
+  // Compare set size of this range with Value.
+  bool isSizeLargerThan(uint64_t MaxSize) const;
 
   /// Return the largest unsigned value contained in the ConstantRange.
   APInt getUnsignedMax() const;

include/llvm/IR/DataLayout.h

@@ -1,4 +1,4 @@
-//===--------- llvm/DataLayout.h - Data size & alignment info ---*- C++ -*-===//
+//===- llvm/DataLayout.h - Data size & alignment info -----------*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
@@ -20,27 +20,32 @@
 #ifndef LLVM_IR_DATALAYOUT_H
 #define LLVM_IR_DATALAYOUT_H
 
+#include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/STLExtras.h"
+#include "llvm/ADT/StringRef.h"
 #include "llvm/IR/DerivedTypes.h"
 #include "llvm/IR/Type.h"
 #include "llvm/Pass.h"
-#include "llvm/Support/DataTypes.h"
+#include "llvm/Support/Casting.h"
+#include "llvm/Support/ErrorHandling.h"
+#include "llvm/Support/MathExtras.h"
+#include <cassert>
+#include <cstdint>
+#include <string>
 
 // This needs to be outside of the namespace, to avoid conflict with llvm-c
 // decl.
-typedef struct LLVMOpaqueTargetData *LLVMTargetDataRef;
+using LLVMTargetDataRef = struct LLVMOpaqueTargetData *;
 
 namespace llvm {
 
-class Value;
-class StructType;
-class StructLayout;
-class Triple;
 class GlobalVariable;
 class LLVMContext;
-template<typename T>
-class ArrayRef;
+class Module;
+class StructLayout;
+class Triple;
+class Value;
 
 /// Enum used to categorize the alignment types stored by LayoutAlignElem
 enum AlignTypeEnum {
@@ -72,6 +77,7 @@ struct LayoutAlignElem {
 
   static LayoutAlignElem get(AlignTypeEnum align_type, unsigned abi_align,
                              unsigned pref_align, uint32_t bit_width);
+
   bool operator==(const LayoutAlignElem &rhs) const;
 };
 
@@ -90,6 +96,7 @@ struct PointerAlignElem {
   /// Initializer
   static PointerAlignElem get(uint32_t AddressSpace, unsigned ABIAlign,
                               unsigned PrefAlign, uint32_t TypeByteWidth);
+
   bool operator==(const PointerAlignElem &rhs) const;
 };
 
@@ -121,7 +128,7 @@ private:
 
   /// \brief Primitive type alignment data. This is sorted by type and bit
   /// width during construction.
-  typedef SmallVector<LayoutAlignElem, 16> AlignmentsTy;
+  using AlignmentsTy = SmallVector<LayoutAlignElem, 16>;
   AlignmentsTy Alignments;
 
   AlignmentsTy::const_iterator
@@ -136,7 +143,7 @@ private:
   /// \brief The string representation used to create this DataLayout
   std::string StringRepresentation;
 
-  typedef SmallVector<PointerAlignElem, 8> PointersTy;
+  using PointersTy = SmallVector<PointerAlignElem, 8>;
   PointersTy Pointers;
 
   PointersTy::const_iterator
@@ -147,7 +154,7 @@ private:
   PointersTy::iterator findPointerLowerBound(uint32_t AddressSpace);
 
   // The StructType -> StructLayout map.
-  mutable void *LayoutMap;
+  mutable void *LayoutMap = nullptr;
 
   /// Pointers in these address spaces are non-integral, and don't have a
   /// well-defined bitwise representation.
@@ -172,16 +179,16 @@ private:
 
 public:
   /// Constructs a DataLayout from a specification string. See reset().
-  explicit DataLayout(StringRef LayoutDescription) : LayoutMap(nullptr) {
+  explicit DataLayout(StringRef LayoutDescription) {
     reset(LayoutDescription);
   }
 
   /// Initialize target data from properties stored in the module.
   explicit DataLayout(const Module *M);
 
-  void init(const Module *M);
+  DataLayout(const DataLayout &DL) { *this = DL; }
 
-  DataLayout(const DataLayout &DL) : LayoutMap(nullptr) { *this = DL; }
+  ~DataLayout(); // Not virtual, do not subclass this class
 
   DataLayout &operator=(const DataLayout &DL) {
     clear();
@@ -200,7 +207,7 @@ public:
   bool operator==(const DataLayout &Other) const;
   bool operator!=(const DataLayout &Other) const { return !(*this == Other); }
 
-  ~DataLayout(); // Not virtual, do not subclass this class
+  void init(const Module *M);
 
   /// Parse a data layout string (with fallback to default values).
   void reset(StringRef LayoutDescription);
@@ -489,6 +496,7 @@ class StructLayout {
   unsigned IsPadded : 1;
   unsigned NumElements : 31;
   uint64_t MemberOffsets[1]; // variable sized array!
+
 public:
   uint64_t getSizeInBytes() const { return StructSize; }
 
@@ -515,6 +523,7 @@ public:
 
 private:
   friend class DataLayout; // Only DataLayout can create this class
+
   StructLayout(StructType *ST, const DataLayout &DL);
 };
 
@@ -560,6 +569,6 @@ inline uint64_t DataLayout::getTypeSizeInBits(Type *Ty) const {
   }
 }
 
-} // End llvm namespace
+} // end namespace llvm
 
-#endif
+#endif // LLVM_IR_DATALAYOUT_H

include/llvm/IR/DebugInfo.h

@@ -21,17 +21,12 @@
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/iterator_range.h"
 #include "llvm/IR/DebugInfoMetadata.h"
-#include "llvm/Support/Casting.h"
-#include "llvm/Support/Dwarf.h"
-#include "llvm/Support/ErrorHandling.h"
-#include <iterator>
 
 namespace llvm {
-class Module;
+
 class DbgDeclareInst;
 class DbgValueInst;
-template <typename K, typename V, typename KeyInfoT, typename BucketT>
-class DenseMap;
+class Module;
 
 /// \brief Find subprogram that is enclosing this scope.
 DISubprogram *getDISubprogram(const MDNode *Scope);
@@ -95,13 +90,13 @@ private:
   bool addScope(DIScope *Scope);
 
 public:
-  typedef SmallVectorImpl<DICompileUnit *>::const_iterator
-      compile_unit_iterator;
-  typedef SmallVectorImpl<DISubprogram *>::const_iterator subprogram_iterator;
-  typedef SmallVectorImpl<DIGlobalVariableExpression *>::const_iterator
-      global_variable_expression_iterator;
-  typedef SmallVectorImpl<DIType *>::const_iterator type_iterator;
-  typedef SmallVectorImpl<DIScope *>::const_iterator scope_iterator;
+  using compile_unit_iterator =
+      SmallVectorImpl<DICompileUnit *>::const_iterator;
+  using subprogram_iterator = SmallVectorImpl<DISubprogram *>::const_iterator;
+  using global_variable_expression_iterator =
+      SmallVectorImpl<DIGlobalVariableExpression *>::const_iterator;
+  using type_iterator = SmallVectorImpl<DIType *>::const_iterator;
+  using scope_iterator = SmallVectorImpl<DIScope *>::const_iterator;
 
   iterator_range<compile_unit_iterator> compile_units() const {
     return make_range(CUs.begin(), CUs.end());
@@ -140,4 +135,4 @@ private:
 
 } // end namespace llvm
 
-#endif
+#endif // LLVM_IR_DEBUGINFO_H

include/llvm/IR/Dominators.h

@@ -42,7 +42,7 @@ extern template void Calculate<Function, Inverse<BasicBlock *>>(
     DominatorTreeBaseByGraphTraits<GraphTraits<Inverse<BasicBlock *>>> &DT,
     Function &F);
 
-typedef DomTreeNodeBase<BasicBlock> DomTreeNode;
+using DomTreeNode = DomTreeNodeBase<BasicBlock>;
 
 class BasicBlockEdge {
   const BasicBlock *Start;
@@ -70,7 +70,7 @@ public:
 };
 
 template <> struct DenseMapInfo<BasicBlockEdge> {
-  typedef DenseMapInfo<const BasicBlock *> BBInfo;
+  using BBInfo = DenseMapInfo<const BasicBlock *>;
 
   static unsigned getHashValue(const BasicBlockEdge *V);
 
@@ -113,7 +113,7 @@ template <> struct DenseMapInfo<BasicBlockEdge> {
 /// preceding statements; this is stated only to assist human understanding.
 class DominatorTree : public DominatorTreeBase<BasicBlock> {
 public:
-  typedef DominatorTreeBase<BasicBlock> Base;
+  using Base = DominatorTreeBase<BasicBlock>;
 
   DominatorTree() : DominatorTreeBase<BasicBlock>(false) {}
   explicit DominatorTree(Function &F) : DominatorTreeBase<BasicBlock>(false) {
@@ -168,9 +168,9 @@ public:
 // iterable by generic graph iterators.
 
 template <class Node, class ChildIterator> struct DomTreeGraphTraitsBase {
-  typedef Node *NodeRef;
-  typedef ChildIterator ChildIteratorType;
-  typedef df_iterator<Node *, df_iterator_default_set<Node*>> nodes_iterator;
+  using NodeRef = Node *;
+  using ChildIteratorType = ChildIterator;
+  using nodes_iterator = df_iterator<Node *, df_iterator_default_set<Node*>>;
 
   static NodeRef getEntryNode(NodeRef N) { return N; }
   static ChildIteratorType child_begin(NodeRef N) { return N->begin(); }
@@ -212,7 +212,7 @@ class DominatorTreeAnalysis : public AnalysisInfoMixin<DominatorTreeAnalysis> {
 
 public:
   /// \brief Provide the result typedef for this analysis pass.
-  typedef DominatorTree Result;
+  using Result = DominatorTree;
 
   /// \brief Run the analysis pass over a function and produce a dominator tree.
   DominatorTree run(Function &F, FunctionAnalysisManager &);

include/llvm/IR/Function.h

@@ -466,7 +466,6 @@ public:
 
   /// @brief Determine if the parameter or return value is marked with NoAlias
   /// attribute.
-  /// @param n The parameter to check. 1 is the first parameter, 0 is the return
   bool returnDoesNotAlias() const {
     return AttributeSets.hasAttribute(AttributeList::ReturnIndex,
                                       Attribute::NoAlias);

include/llvm/IR/InlineAsm.h

@@ -95,7 +95,7 @@ public:
     isClobber           // '~x'
   };
 
-  typedef std::vector<std::string> ConstraintCodeVector;
+  using ConstraintCodeVector = std::vector<std::string>;
 
   struct SubConstraintInfo {
     /// MatchingInput - If this is not -1, this is an output constraint where an
@@ -112,9 +112,9 @@ public:
     SubConstraintInfo() = default;
   };
 
-  typedef std::vector<SubConstraintInfo> SubConstraintInfoVector;
+  using SubConstraintInfoVector = std::vector<SubConstraintInfo>;
   struct ConstraintInfo;
-  typedef std::vector<ConstraintInfo> ConstraintInfoVector;
+  using ConstraintInfoVector = std::vector<ConstraintInfo>;
 
   struct ConstraintInfo {
     /// Type - The basic type of the constraint: input/output/clobber

include/llvm/IR/InstIterator.h

@@ -31,20 +31,20 @@ namespace llvm {
 // inst_iterator and const_inst_iterator's.
 //
 template <class BB_t, class BB_i_t, class BI_t, class II_t> class InstIterator {
-  typedef BB_t BBty;
-  typedef BB_i_t BBIty;
-  typedef BI_t BIty;
-  typedef II_t IIty;
+  using BBty = BB_t;
+  using BBIty = BB_i_t;
+  using BIty = BI_t;
+  using IIty = II_t;
   BB_t *BBs; // BasicBlocksType
   BB_i_t BB; // BasicBlocksType::iterator
   BI_t BI;   // BasicBlock::iterator
 
 public:
-  typedef std::bidirectional_iterator_tag iterator_category;
-  typedef IIty                            value_type;
-  typedef signed                        difference_type;
-  typedef IIty*                           pointer;
-  typedef IIty&                           reference;
+  using iterator_category = std::bidirectional_iterator_tag;
+  using value_type = IIty;
+  using difference_type = signed;
+  using pointer = IIty *;
+  using reference = IIty &;
 
   // Default constructor
   InstIterator() = default;
@@ -119,13 +119,15 @@ private:
   }
 };
 
-typedef InstIterator<SymbolTableList<BasicBlock>, Function::iterator,
-                     BasicBlock::iterator, Instruction> inst_iterator;
-typedef InstIterator<const SymbolTableList<BasicBlock>,
-                     Function::const_iterator, BasicBlock::const_iterator,
-                     const Instruction> const_inst_iterator;
-typedef iterator_range<inst_iterator> inst_range;
-typedef iterator_range<const_inst_iterator> const_inst_range;
+using inst_iterator =
+    InstIterator<SymbolTableList<BasicBlock>, Function::iterator,
+                 BasicBlock::iterator, Instruction>;
+using const_inst_iterator =
+    InstIterator<const SymbolTableList<BasicBlock>,
+                 Function::const_iterator, BasicBlock::const_iterator,
+                 const Instruction>;
+using inst_range = iterator_range<inst_iterator>;
+using const_inst_range = iterator_range<const_inst_iterator>;
 
 inline inst_iterator inst_begin(Function *F) { return inst_iterator(*F); }
 inline inst_iterator inst_end(Function *F)   { return inst_iterator(*F, true); }

include/llvm/IR/InstrTypes.h

@@ -1,4 +1,4 @@
-//===-- llvm/InstrTypes.h - Important Instruction subclasses ----*- C++ -*-===//
+//===- llvm/InstrTypes.h - Important Instruction subclasses -----*- C++ -*-===//
 //
 //                     The LLVM Compiler Infrastructure
 //
@@ -29,7 +29,9 @@
 #include "llvm/IR/Instruction.h"
 #include "llvm/IR/LLVMContext.h"
 #include "llvm/IR/OperandTraits.h"
+#include "llvm/IR/Type.h"
 #include "llvm/IR/User.h"
+#include "llvm/IR/Value.h"
 #include "llvm/Support/Casting.h"
 #include "llvm/Support/ErrorHandling.h"
 #include <algorithm>
@@ -114,17 +116,17 @@ public:
   template <class Term, class BB> // Successor Iterator
   class SuccIterator : public std::iterator<std::random_access_iterator_tag, BB,
                                             int, BB *, BB *> {
-    typedef std::iterator<std::random_access_iterator_tag, BB, int, BB *, BB *>
-        super;
+    using super =
+        std::iterator<std::random_access_iterator_tag, BB, int, BB *, BB *>;
 
   public:
-    typedef typename super::pointer pointer;
-    typedef typename super::reference reference;
+    using pointer = typename super::pointer;
+    using reference = typename super::reference;
 
   private:
     Term TermInst;
     unsigned idx;
-    typedef SuccIterator<Term, BB> Self;
+    using Self = SuccIterator<Term, BB>;
 
     inline bool index_is_valid(unsigned idx) {
       return idx < TermInst->getNumSuccessors();
@@ -260,11 +262,11 @@ public:
     }
   };
 
-  typedef SuccIterator<TerminatorInst *, BasicBlock> succ_iterator;
-  typedef SuccIterator<const TerminatorInst *, const BasicBlock>
-      succ_const_iterator;
-  typedef iterator_range<succ_iterator> succ_range;
-  typedef iterator_range<succ_const_iterator> succ_const_range;
+  using succ_iterator = SuccIterator<TerminatorInst *, BasicBlock>;
+  using succ_const_iterator =
+      SuccIterator<const TerminatorInst *, const BasicBlock>;
+  using succ_range = iterator_range<succ_iterator>;
+  using succ_const_range = iterator_range<succ_const_iterator>;
 
 private:
   inline succ_iterator succ_begin() { return succ_iterator(this); }
@@ -341,14 +343,16 @@ DEFINE_TRANSPARENT_OPERAND_ACCESSORS(UnaryInstruction, Value)
 
 class BinaryOperator : public Instruction {
 protected:
-  void init(BinaryOps iType);
   BinaryOperator(BinaryOps iType, Value *S1, Value *S2, Type *Ty,
                  const Twine &Name, Instruction *InsertBefore);
   BinaryOperator(BinaryOps iType, Value *S1, Value *S2, Type *Ty,
                  const Twine &Name, BasicBlock *InsertAtEnd);
 
+  void init(BinaryOps iType);
+
   // Note: Instruction needs to be a friend here to call cloneImpl.
   friend class Instruction;
+
   BinaryOperator *cloneImpl() const;
 
 public:
@@ -1125,8 +1129,6 @@ DEFINE_TRANSPARENT_OPERAND_ACCESSORS(CmpInst, Value)
 //===----------------------------------------------------------------------===//
 class FuncletPadInst : public Instruction {
 private:
-  void init(Value *ParentPad, ArrayRef<Value *> Args, const Twine &NameStr);
-
   FuncletPadInst(const FuncletPadInst &CPI);
 
   explicit FuncletPadInst(Instruction::FuncletPadOps Op, Value *ParentPad,
@@ -1136,11 +1138,14 @@ private:
                           ArrayRef<Value *> Args, unsigned Values,
                           const Twine &NameStr, BasicBlock *InsertAtEnd);
 
+  void init(Value *ParentPad, ArrayRef<Value *> Args, const Twine &NameStr);
+
 protected:
   // Note: Instruction needs to be a friend here to call cloneImpl.
   friend class Instruction;
   friend class CatchPadInst;
   friend class CleanupPadInst;
+
   FuncletPadInst *cloneImpl() const;
 
 public:
@@ -1261,7 +1266,8 @@ public:
 
   ArrayRef<InputTy> inputs() const { return Inputs; }
 
-  typedef typename std::vector<InputTy>::const_iterator input_iterator;
+  using input_iterator = typename std::vector<InputTy>::const_iterator;
+
   size_t input_size() const { return Inputs.size(); }
   input_iterator input_begin() const { return Inputs.begin(); }
   input_iterator input_end() const { return Inputs.end(); }
@@ -1269,8 +1275,8 @@ public:
   StringRef getTag() const { return Tag; }
 };
 
-typedef OperandBundleDefT<Value *> OperandBundleDef;
-typedef OperandBundleDefT<const Value *> ConstOperandBundleDef;
+using OperandBundleDef = OperandBundleDefT<Value *>;
+using ConstOperandBundleDef = OperandBundleDefT<const Value *>;
 
 /// \brief A mixin to add operand bundle functionality to llvm instruction
 /// classes.
@@ -1553,8 +1559,8 @@ protected:
     return OperandBundleUse(BOI.Tag, Inputs);
   }
 
-  typedef BundleOpInfo *bundle_op_iterator;
-  typedef const BundleOpInfo *const_bundle_op_iterator;
+  using bundle_op_iterator = BundleOpInfo *;
+  using const_bundle_op_iterator = const BundleOpInfo *;
 
   /// \brief Return the start of the list of BundleOpInfo instances associated
   /// with this OperandBundleUser.
@@ -1654,6 +1660,6 @@ protected:
   }
 };
 
-} // end llvm namespace
+} // end namespace llvm
 
 #endif // LLVM_IR_INSTRTYPES_H

include/llvm/IR/Intrinsics.td

@@ -795,6 +795,14 @@ def int_type_checked_load : Intrinsic<[llvm_ptr_ty, llvm_i1_ty],
 def int_load_relative: Intrinsic<[llvm_ptr_ty], [llvm_ptr_ty, llvm_anyint_ty],
                                  [IntrReadMem, IntrArgMemOnly]>;
 
+// Xray intrinsics
+//===----------------------------------------------------------------------===//
+// Custom event logging for x-ray.
+// Takes a pointer to a string and the length of the string.
+def int_xray_customevent : Intrinsic<[], [llvm_ptr_ty, llvm_i32_ty],
+                                     [NoCapture<0>, ReadOnly<0>, IntrWriteMem]>;
+//===----------------------------------------------------------------------===//
+
 //===------ Memory intrinsics with element-wise atomicity guarantees ------===//
 //
 

include/llvm/IR/IntrinsicsARM.td

@@ -22,12 +22,26 @@ let TargetPrefix = "arm" in {  // All intrinsics start with "llvm.arm.".
 // and return value are essentially chains, used to force ordering during ISel.
 def int_arm_space : Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
 
+// 16-bit multiplications
+def int_arm_smulbb : GCCBuiltin<"__builtin_arm_smulbb">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_smulbt : GCCBuiltin<"__builtin_arm_smulbt">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_smultb : GCCBuiltin<"__builtin_arm_smultb">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_smultt : GCCBuiltin<"__builtin_arm_smultt">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_smulwb : GCCBuiltin<"__builtin_arm_smulwb">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_smulwt : GCCBuiltin<"__builtin_arm_smulwt">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+
 //===----------------------------------------------------------------------===//
 // Saturating Arithmetic
 
 def int_arm_qadd : GCCBuiltin<"__builtin_arm_qadd">,
     Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
-    [IntrNoMem, Commutative]>;
+    [Commutative, IntrNoMem]>;
 def int_arm_qsub : GCCBuiltin<"__builtin_arm_qsub">,
     Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
 def int_arm_ssat : GCCBuiltin<"__builtin_arm_ssat">,
@@ -35,6 +49,176 @@ def int_arm_ssat : GCCBuiltin<"__builtin_arm_ssat">,
 def int_arm_usat : GCCBuiltin<"__builtin_arm_usat">,
     Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
 
+// Accumulating multiplications
+def int_arm_smlabb : GCCBuiltin<"__builtin_arm_smlabb">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+    [IntrNoMem]>;
+def int_arm_smlabt : GCCBuiltin<"__builtin_arm_smlabt">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+    [IntrNoMem]>;
+def int_arm_smlatb : GCCBuiltin<"__builtin_arm_smlatb">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+    [IntrNoMem]>;
+def int_arm_smlatt : GCCBuiltin<"__builtin_arm_smlatt">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+    [IntrNoMem]>;
+def int_arm_smlawb : GCCBuiltin<"__builtin_arm_smlawb">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+    [IntrNoMem]>;
+def int_arm_smlawt : GCCBuiltin<"__builtin_arm_smlawt">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+    [IntrNoMem]>;
+
+// Parallel 16-bit saturation
+def int_arm_ssat16 : GCCBuiltin<"__builtin_arm_ssat16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_usat16 : GCCBuiltin<"__builtin_arm_usat16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+
+// Packing and unpacking
+def int_arm_sxtab16 : GCCBuiltin<"__builtin_arm_sxtab16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_sxtb16 : GCCBuiltin<"__builtin_arm_sxtb16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uxtab16 : GCCBuiltin<"__builtin_arm_uxtab16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uxtb16 : GCCBuiltin<"__builtin_arm_uxtb16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty], [IntrNoMem]>;
+
+// Parallel selection, reads the GE flags.
+def int_arm_sel : GCCBuiltin<"__builtin_arm_sel">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrReadMem]>;
+
+// Parallel 8-bit addition and subtraction
+def int_arm_qadd8  : GCCBuiltin<"__builtin_arm_qadd8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_qsub8  : GCCBuiltin<"__builtin_arm_qsub8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+// Writes to the GE bits.
+def int_arm_sadd8  : GCCBuiltin<"__builtin_arm_sadd8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+def int_arm_shadd8  : GCCBuiltin<"__builtin_arm_shadd8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_shsub8  : GCCBuiltin<"__builtin_arm_shsub8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+// Writes to the GE bits.
+def int_arm_ssub8  : GCCBuiltin<"__builtin_arm_ssub8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+// Writes to the GE bits.
+def int_arm_uadd8  : GCCBuiltin<"__builtin_arm_uadd8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+def int_arm_uhadd8  : GCCBuiltin<"__builtin_arm_uhadd8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uhsub8  : GCCBuiltin<"__builtin_arm_uhsub8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uqadd8  : GCCBuiltin<"__builtin_arm_uqadd8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uqsub8  : GCCBuiltin<"__builtin_arm_uqsub8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+// Writes to the GE bits.
+def int_arm_usub8  : GCCBuiltin<"__builtin_arm_usub8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+
+// Sum of 8-bit absolute differences
+def int_arm_usad8  : GCCBuiltin<"__builtin_arm_usad8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_usada8  : GCCBuiltin<"__builtin_arm_usada8">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+              [IntrNoMem]>;
+
+// Parallel 16-bit addition and subtraction
+def int_arm_qadd16  : GCCBuiltin<"__builtin_arm_qadd16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_qasx  : GCCBuiltin<"__builtin_arm_qasx">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_qsax  : GCCBuiltin<"__builtin_arm_qsax">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_qsub16  : GCCBuiltin<"__builtin_arm_qsub16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+// Writes to the GE bits.
+def int_arm_sadd16  : GCCBuiltin<"__builtin_arm_sadd16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+// Writes to the GE bits.
+def int_arm_sasx  : GCCBuiltin<"__builtin_arm_sasx">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+def int_arm_shadd16  : GCCBuiltin<"__builtin_arm_shadd16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_shasx  : GCCBuiltin<"__builtin_arm_shasx">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_shsax  : GCCBuiltin<"__builtin_arm_shsax">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_shsub16  : GCCBuiltin<"__builtin_arm_shsub16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+// Writes to the GE bits.
+def int_arm_ssax  : GCCBuiltin<"__builtin_arm_ssax">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+// Writes to the GE bits.
+def int_arm_ssub16  : GCCBuiltin<"__builtin_arm_ssub16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+// Writes to the GE bits.
+def int_arm_uadd16  : GCCBuiltin<"__builtin_arm_uadd16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+// Writes to the GE bits.
+def int_arm_uasx  : GCCBuiltin<"__builtin_arm_uasx">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+def int_arm_uhadd16  : GCCBuiltin<"__builtin_arm_uhadd16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uhasx  : GCCBuiltin<"__builtin_arm_uhasx">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uhsax  : GCCBuiltin<"__builtin_arm_uhsax">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uhsub16  : GCCBuiltin<"__builtin_arm_uhsub16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uqadd16  : GCCBuiltin<"__builtin_arm_uqadd16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uqasx  : GCCBuiltin<"__builtin_arm_uqasx">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uqsax  : GCCBuiltin<"__builtin_arm_uqsax">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_uqsub16  : GCCBuiltin<"__builtin_arm_uqsub16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+// Writes to the GE bits.
+def int_arm_usax  : GCCBuiltin<"__builtin_arm_usax">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+// Writes to the GE bits.
+def int_arm_usub16  : GCCBuiltin<"__builtin_arm_usub16">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], []>;
+
+// Parallel 16-bit multiplication
+def int_arm_smlad : GCCBuiltin<"__builtin_arm_smlad">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+              [IntrNoMem]>;
+def int_arm_smladx : GCCBuiltin<"__builtin_arm_smladx">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+              [IntrNoMem]>;
+def int_arm_smlald : GCCBuiltin<"__builtin_arm_smlald">,
+    Intrinsic<[llvm_i64_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i64_ty],
+              [IntrNoMem]>;
+def int_arm_smlaldx : GCCBuiltin<"__builtin_arm_smlaldx">,
+    Intrinsic<[llvm_i64_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i64_ty],
+              [IntrNoMem]>;
+def int_arm_smlsd : GCCBuiltin<"__builtin_arm_smlsd">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+              [IntrNoMem]>;
+def int_arm_smlsdx : GCCBuiltin<"__builtin_arm_smlsdx">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+              [IntrNoMem]>;
+def int_arm_smlsld : GCCBuiltin<"__builtin_arm_smlsld">,
+    Intrinsic<[llvm_i64_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i64_ty],
+              [IntrNoMem]>;
+def int_arm_smlsldx : GCCBuiltin<"__builtin_arm_smlsldx">,
+    Intrinsic<[llvm_i64_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i64_ty],
+              [IntrNoMem]>;
+def int_arm_smuad : GCCBuiltin<"__builtin_arm_smuad">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_smuadx : GCCBuiltin<"__builtin_arm_smuadx">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_smusd : GCCBuiltin<"__builtin_arm_smusd">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+def int_arm_smusdx : GCCBuiltin<"__builtin_arm_smusdx">,
+    Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], [IntrNoMem]>;
+
+
 //===----------------------------------------------------------------------===//
 // Load, Store and Clear exclusive
 

include/llvm/IR/ModuleSummaryIndex.h

@@ -45,58 +45,54 @@ struct CalleeInfo {
   }
 };
 
-/// Struct to hold value either by GUID or GlobalValue*. Values in combined
-/// indexes as well as indirect calls are GUIDs, all others are GlobalValues.
+class GlobalValueSummary;
+
+typedef std::vector<std::unique_ptr<GlobalValueSummary>> GlobalValueSummaryList;
+
+struct GlobalValueSummaryInfo {
+  /// The GlobalValue corresponding to this summary. This is only used in
+  /// per-module summaries.
+  const GlobalValue *GV = nullptr;
+
+  /// List of global value summary structures for a particular value held
+  /// in the GlobalValueMap. Requires a vector in the case of multiple
+  /// COMDAT values of the same name.
+  GlobalValueSummaryList SummaryList;
+};
+
+/// Map from global value GUID to corresponding summary structures. Use a
+/// std::map rather than a DenseMap so that pointers to the map's value_type
+/// (which are used by ValueInfo) are not invalidated by insertion. Also it will
+/// likely incur less overhead, as the value type is not very small and the size
+/// of the map is unknown, resulting in inefficiencies due to repeated
+/// insertions and resizing.
+typedef std::map<GlobalValue::GUID, GlobalValueSummaryInfo>
+    GlobalValueSummaryMapTy;
+
+/// Struct that holds a reference to a particular GUID in a global value
+/// summary.
 struct ValueInfo {
-  /// The value representation used in this instance.
-  enum ValueInfoKind {
-    VI_GUID,
-    VI_Value,
-  };
+  const GlobalValueSummaryMapTy::value_type *Ref = nullptr;
+  ValueInfo() = default;
+  ValueInfo(const GlobalValueSummaryMapTy::value_type *Ref) : Ref(Ref) {}
+  operator bool() const { return Ref; }
 
-  /// Union of the two possible value types.
-  union ValueUnion {
-    GlobalValue::GUID Id;
-    const GlobalValue *GV;
-    ValueUnion(GlobalValue::GUID Id) : Id(Id) {}
-    ValueUnion(const GlobalValue *GV) : GV(GV) {}
-  };
-
-  /// The value being represented.
-  ValueUnion TheValue;
-  /// The value representation.
-  ValueInfoKind Kind;
-  /// Constructor for a GUID value
-  ValueInfo(GlobalValue::GUID Id = 0) : TheValue(Id), Kind(VI_GUID) {}
-  /// Constructor for a GlobalValue* value
-  ValueInfo(const GlobalValue *V) : TheValue(V), Kind(VI_Value) {}
-  /// Accessor for GUID value
-  GlobalValue::GUID getGUID() const {
-    assert(Kind == VI_GUID && "Not a GUID type");
-    return TheValue.Id;
+  GlobalValue::GUID getGUID() const { return Ref->first; }
+  const GlobalValue *getValue() const { return Ref->second.GV; }
+  ArrayRef<std::unique_ptr<GlobalValueSummary>> getSummaryList() const {
+    return Ref->second.SummaryList;
   }
-  /// Accessor for GlobalValue* value
-  const GlobalValue *getValue() const {
-    assert(Kind == VI_Value && "Not a Value type");
-    return TheValue.GV;
-  }
-  bool isGUID() const { return Kind == VI_GUID; }
 };
 
 template <> struct DenseMapInfo<ValueInfo> {
-  static inline ValueInfo getEmptyKey() { return ValueInfo((GlobalValue *)-1); }
+  static inline ValueInfo getEmptyKey() {
+    return ValueInfo((GlobalValueSummaryMapTy::value_type *)-1);
+  }
   static inline ValueInfo getTombstoneKey() {
-    return ValueInfo((GlobalValue *)-2);
-  }
-  static bool isEqual(ValueInfo L, ValueInfo R) {
-    if (L.isGUID() != R.isGUID())
-      return false;
-    return L.isGUID() ? (L.getGUID() == R.getGUID())
-                      : (L.getValue() == R.getValue());
-  }
-  static unsigned getHashValue(ValueInfo I) {
-    return I.isGUID() ? I.getGUID() : (uintptr_t)I.getValue();
+    return ValueInfo((GlobalValueSummaryMapTy::value_type *)-2);
   }
+  static bool isEqual(ValueInfo L, ValueInfo R) { return L.Ref == R.Ref; }
+  static unsigned getHashValue(ValueInfo I) { return (uintptr_t)I.Ref; }
 };
 
 /// \brief Function and variable summary information to aid decisions and
@@ -483,19 +479,6 @@ struct TypeIdSummary {
 /// 160 bits SHA1
 typedef std::array<uint32_t, 5> ModuleHash;
 
-/// List of global value summary structures for a particular value held
-/// in the GlobalValueMap. Requires a vector in the case of multiple
-/// COMDAT values of the same name.
-typedef std::vector<std::unique_ptr<GlobalValueSummary>> GlobalValueSummaryList;
-
-/// Map from global value GUID to corresponding summary structures.
-/// Use a std::map rather than a DenseMap since it will likely incur
-/// less overhead, as the value type is not very small and the size
-/// of the map is unknown, resulting in inefficiencies due to repeated
-/// insertions and resizing.
-typedef std::map<GlobalValue::GUID, GlobalValueSummaryList>
-    GlobalValueSummaryMapTy;
-
 /// Type used for iterating through the global value summary map.
 typedef GlobalValueSummaryMapTy::const_iterator const_gvsummary_iterator;
 typedef GlobalValueSummaryMapTy::iterator gvsummary_iterator;
@@ -532,6 +515,11 @@ private:
   // YAML I/O support.
   friend yaml::MappingTraits<ModuleSummaryIndex>;
 
+  GlobalValueSummaryMapTy::value_type *
+  getOrInsertValuePtr(GlobalValue::GUID GUID) {
+    return &*GlobalValueMap.emplace(GUID, GlobalValueSummaryInfo{}).first;
+  }
+
 public:
   gvsummary_iterator begin() { return GlobalValueMap.begin(); }
   const_gvsummary_iterator begin() const { return GlobalValueMap.begin(); }
@@ -539,21 +527,22 @@ public:
   const_gvsummary_iterator end() const { return GlobalValueMap.end(); }
   size_t size() const { return GlobalValueMap.size(); }
 
-  /// Get the list of global value summary objects for a given value name.
-  const GlobalValueSummaryList &getGlobalValueSummaryList(StringRef ValueName) {
-    return GlobalValueMap[GlobalValue::getGUID(ValueName)];
+  /// Return a ValueInfo for GUID if it exists, otherwise return ValueInfo().
+  ValueInfo getValueInfo(GlobalValue::GUID GUID) const {
+    auto I = GlobalValueMap.find(GUID);
+    return ValueInfo(I == GlobalValueMap.end() ? nullptr : &*I);
   }
 
-  /// Get the list of global value summary objects for a given value name.
-  const const_gvsummary_iterator
-  findGlobalValueSummaryList(StringRef ValueName) const {
-    return GlobalValueMap.find(GlobalValue::getGUID(ValueName));
+  /// Return a ValueInfo for \p GUID.
+  ValueInfo getOrInsertValueInfo(GlobalValue::GUID GUID) {
+    return ValueInfo(getOrInsertValuePtr(GUID));
   }
 
-  /// Get the list of global value summary objects for a given value GUID.
-  const const_gvsummary_iterator
-  findGlobalValueSummaryList(GlobalValue::GUID ValueGUID) const {
-    return GlobalValueMap.find(ValueGUID);
+  /// Return a ValueInfo for \p GV and mark it as belonging to GV.
+  ValueInfo getOrInsertValueInfo(const GlobalValue *GV) {
+    auto VP = getOrInsertValuePtr(GV->getGUID());
+    VP->second.GV = GV;
+    return ValueInfo(VP);
   }
 
   /// Return the GUID for \p OriginalId in the OidGuidMap.
@@ -565,17 +554,18 @@ public:
   /// Add a global value summary for a value of the given name.
   void addGlobalValueSummary(StringRef ValueName,
                              std::unique_ptr<GlobalValueSummary> Summary) {
-    addOriginalName(GlobalValue::getGUID(ValueName),
-                    Summary->getOriginalName());
-    GlobalValueMap[GlobalValue::getGUID(ValueName)].push_back(
-        std::move(Summary));
+    addGlobalValueSummary(getOrInsertValueInfo(GlobalValue::getGUID(ValueName)),
+                          std::move(Summary));
   }
 
-  /// Add a global value summary for a value of the given GUID.
-  void addGlobalValueSummary(GlobalValue::GUID ValueGUID,
+  /// Add a global value summary for the given ValueInfo.
+  void addGlobalValueSummary(ValueInfo VI,
                              std::unique_ptr<GlobalValueSummary> Summary) {
-    addOriginalName(ValueGUID, Summary->getOriginalName());
-    GlobalValueMap[ValueGUID].push_back(std::move(Summary));
+    addOriginalName(VI.getGUID(), Summary->getOriginalName());
+    // Here we have a notionally const VI, but the value it points to is owned
+    // by the non-const *this.
+    const_cast<GlobalValueSummaryMapTy::value_type *>(VI.Ref)
+        ->second.SummaryList.push_back(std::move(Summary));
   }
 
   /// Add an original name for the value of the given GUID.
@@ -593,16 +583,16 @@ public:
   /// not found.
   GlobalValueSummary *findSummaryInModule(GlobalValue::GUID ValueGUID,
                                           StringRef ModuleId) const {
-    auto CalleeInfoList = findGlobalValueSummaryList(ValueGUID);
-    if (CalleeInfoList == end()) {
+    auto CalleeInfo = getValueInfo(ValueGUID);
+    if (!CalleeInfo) {
       return nullptr; // This function does not have a summary
     }
     auto Summary =
-        llvm::find_if(CalleeInfoList->second,
+        llvm::find_if(CalleeInfo.getSummaryList(),
                       [&](const std::unique_ptr<GlobalValueSummary> &Summary) {
                         return Summary->modulePath() == ModuleId;
                       });
-    if (Summary == CalleeInfoList->second.end())
+    if (Summary == CalleeInfo.getSummaryList().end())
       return nullptr;
     return Summary->get();
   }

include/llvm/IR/ModuleSummaryIndexYAML.h

@@ -201,7 +201,7 @@ template <> struct CustomMappingTraits<GlobalValueSummaryMapTy> {
     for (auto &FSum : FSums) {
       GlobalValueSummary::GVFlags GVFlags(GlobalValue::ExternalLinkage, false,
                                           false);
-      Elem.push_back(llvm::make_unique<FunctionSummary>(
+      Elem.SummaryList.push_back(llvm::make_unique<FunctionSummary>(
           GVFlags, 0, ArrayRef<ValueInfo>{},
           ArrayRef<FunctionSummary::EdgeTy>{}, std::move(FSum.TypeTests),
           std::move(FSum.TypeTestAssumeVCalls),
@@ -213,7 +213,7 @@ template <> struct CustomMappingTraits<GlobalValueSummaryMapTy> {
   static void output(IO &io, GlobalValueSummaryMapTy &V) {
     for (auto &P : V) {
       std::vector<FunctionSummaryYaml> FSums;
-      for (auto &Sum : P.second) {
+      for (auto &Sum : P.second.SummaryList) {
         if (auto *FSum = dyn_cast<FunctionSummary>(Sum.get()))
           FSums.push_back(FunctionSummaryYaml{
               FSum->type_tests(), FSum->type_test_assume_vcalls(),

include/llvm/MC/ConstantPools.h

@@ -63,6 +63,8 @@ public:
 
   // Return true if the constant pool is empty
   bool empty();
+
+  void clearCache();
 };
 
 class AssemblerConstantPools {
@@ -86,6 +88,7 @@ class AssemblerConstantPools {
 public:
   void emitAll(MCStreamer &Streamer);
   void emitForCurrentSection(MCStreamer &Streamer);
+  void clearCacheForCurrentSection(MCStreamer &Streamer);
   const MCExpr *addEntry(MCStreamer &Streamer, const MCExpr *Expr,
                          unsigned Size, SMLoc Loc);
 

include/llvm/Object/COFF.h

@@ -20,7 +20,9 @@
 #include "llvm/Object/Binary.h"
 #include "llvm/Object/Error.h"
 #include "llvm/Object/ObjectFile.h"
+#include "llvm/Support/BinaryByteStream.h"
 #include "llvm/Support/COFF.h"
+#include "llvm/Support/ConvertUTF.h"
 #include "llvm/Support/Endian.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/ErrorOr.h"
@@ -40,6 +42,7 @@ class DelayImportDirectoryEntryRef;
 class ExportDirectoryEntryRef;
 class ImportDirectoryEntryRef;
 class ImportedSymbolRef;
+class ResourceSectionRef;
 
 using import_directory_iterator = content_iterator<ImportDirectoryEntryRef>;
 using delay_import_directory_iterator =
@@ -623,6 +626,26 @@ struct coff_base_reloc_block_entry {
   int getOffset() const { return Data & ((1 << 12) - 1); }
 };
 
+struct coff_resource_dir_entry {
+  union {
+    support::ulittle32_t NameOffset;
+    support::ulittle32_t ID;
+    uint32_t getNameOffset() const {
+      return maskTrailingOnes<uint32_t>(31) & NameOffset;
+    }
+  } Identifier;
+  union {
+    support::ulittle32_t DataEntryOffset;
+    support::ulittle32_t SubdirOffset;
+
+    bool isSubDir() const { return SubdirOffset >> 31; }
+    uint32_t value() const {
+      return maskTrailingOnes<uint32_t>(31) & SubdirOffset;
+    }
+
+  } Offset;
+};
+
 struct coff_resource_dir_table {
   support::ulittle32_t Characteristics;
   support::ulittle32_t TimeDateStamp;
@@ -1047,6 +1070,23 @@ private:
   const COFFObjectFile *OwningObject = nullptr;
 };
 
+class ResourceSectionRef {
+public:
+  ResourceSectionRef() = default;
+  explicit ResourceSectionRef(StringRef Ref) : BBS(Ref, support::little) {}
+
+  ErrorOr<ArrayRef<UTF16>> getEntryNameString(const coff_resource_dir_entry &Entry);
+  ErrorOr<const coff_resource_dir_table &>
+  getEntrySubDir(const coff_resource_dir_entry &Entry);
+  ErrorOr<const coff_resource_dir_table &> getBaseTable();
+
+private:
+  BinaryByteStream BBS;
+
+  ErrorOr<const coff_resource_dir_table &> getTableAtOffset(uint32_t Offset);
+  ErrorOr<ArrayRef<UTF16>> getDirStringAtOffset(uint32_t Offset);
+};
+
 // Corresponds to `_FPO_DATA` structure in the PE/COFF spec.
 struct FpoData {
   support::ulittle32_t Offset; // ulOffStart: Offset 1st byte of function code

include/llvm/Object/Wasm.h

@@ -41,10 +41,14 @@ public:
     DEBUG_FUNCTION_NAME,
   };
 
-  WasmSymbol(StringRef Name, SymbolType Type) : Name(Name), Type(Type) {}
+  WasmSymbol(StringRef Name, SymbolType Type, uint32_t Section,
+             uint32_t ElementIndex)
+      : Name(Name), Type(Type), Section(Section), ElementIndex(ElementIndex) {}
 
   StringRef Name;
   SymbolType Type;
+  uint32_t Section;
+  uint32_t ElementIndex;
 };
 
 class WasmSection {

include/llvm/ObjectYAML/WasmYAML.h

@@ -97,6 +97,11 @@ struct DataSegment {
   yaml::BinaryRef Content;
 };
 
+struct NameEntry {
+  uint32_t Index;
+  StringRef Name;
+};
+
 struct Signature {
   Signature() : Form(wasm::WASM_TYPE_FUNC) {}
 
@@ -122,6 +127,11 @@ struct CustomSection : Section {
 
   StringRef Name;
   yaml::BinaryRef Payload;
+
+  // The follow is used by the "name" custom section.
+  // TODO(sbc): Add support for more then just functions names.  The wasm
+  // name section can support multiple sub-sections.
+  std::vector<NameEntry> FunctionNames;
 };
 
 struct TypeSection : Section {
@@ -244,6 +254,7 @@ LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::WasmYAML::Global)
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::WasmYAML::Function)
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::WasmYAML::LocalDecl)
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::WasmYAML::Relocation)
+LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::WasmYAML::NameEntry)
 LLVM_YAML_IS_FLOW_SEQUENCE_VECTOR(uint32_t)
 
 namespace llvm {
@@ -297,6 +308,10 @@ template <> struct MappingTraits<WasmYAML::Relocation> {
   static void mapping(IO &IO, WasmYAML::Relocation &Relocation);
 };
 
+template <> struct MappingTraits<WasmYAML::NameEntry> {
+  static void mapping(IO &IO, WasmYAML::NameEntry &NameEntry);
+};
+
 template <> struct MappingTraits<WasmYAML::LocalDecl> {
   static void mapping(IO &IO, WasmYAML::LocalDecl &LocalDecl);
 };

include/llvm/Support/AArch64TargetParser.def

@@ -20,8 +20,7 @@ AARCH64_ARCH("invalid", AK_INVALID, nullptr, nullptr,
              ARMBuildAttrs::CPUArch::v8_A, FK_NONE, AArch64::AEK_NONE)
 AARCH64_ARCH("armv8-a", AK_ARMV8A, "8-A", "v8", ARMBuildAttrs::CPUArch::v8_A,
              FK_CRYPTO_NEON_FP_ARMV8,
-             (AArch64::AEK_CRC | AArch64::AEK_CRYPTO | AArch64::AEK_FP |
-              AArch64::AEK_SIMD))
+             (AArch64::AEK_CRYPTO | AArch64::AEK_FP | AArch64::AEK_SIMD))
 AARCH64_ARCH("armv8.1-a", AK_ARMV8_1A, "8.1-A", "v8.1a",
              ARMBuildAttrs::CPUArch::v8_A, FK_CRYPTO_NEON_FP_ARMV8,
              (AArch64::AEK_CRC | AArch64::AEK_CRYPTO | AArch64::AEK_FP |
@@ -52,38 +51,37 @@ AARCH64_ARCH_EXT_NAME("ras",     AArch64::AEK_RAS,     "+ras",  "-ras")
 #define AARCH64_CPU_NAME(NAME, ID, DEFAULT_FPU, IS_DEFAULT, DEFAULT_EXT)
 #endif
 AARCH64_CPU_NAME("cortex-a35", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("cortex-a53", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, true,
-                ( AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("cortex-a57", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("cortex-a72", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("cortex-a73", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("cyclone", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_NONE))
 AARCH64_CPU_NAME("exynos-m1", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("exynos-m2", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("exynos-m3", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("falkor", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("kryo", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO))
+                (AArch64::AEK_CRC))
 AARCH64_CPU_NAME("thunderx2t99", AK_ARMV8_1A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_LSE | AArch64::AEK_CRC |
-                 AArch64::AEK_CRYPTO))
+                (AArch64::AEK_NONE))
 AARCH64_CPU_NAME("thunderx", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO | AArch64::AEK_FP | AArch64::AEK_PROFILE))
+                (AArch64::AEK_CRC | AArch64::AEK_PROFILE))
 AARCH64_CPU_NAME("thunderxt88", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO | AArch64::AEK_FP | AArch64::AEK_PROFILE))
+                (AArch64::AEK_CRC | AArch64::AEK_PROFILE))
 AARCH64_CPU_NAME("thunderxt81", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO | AArch64::AEK_FP | AArch64::AEK_PROFILE))
+                (AArch64::AEK_CRC | AArch64::AEK_PROFILE))
 AARCH64_CPU_NAME("thunderxt83", AK_ARMV8A, FK_CRYPTO_NEON_FP_ARMV8, false,
-                (AArch64::AEK_SIMD | AArch64::AEK_CRC | AArch64::AEK_CRYPTO | AArch64::AEK_FP | AArch64::AEK_PROFILE))
+                (AArch64::AEK_CRC | AArch64::AEK_PROFILE))
 // Invalid CPU
 AARCH64_CPU_NAME("invalid", AK_INVALID, FK_INVALID, true, AArch64::AEK_INVALID)
 #undef AARCH64_CPU_NAME

include/llvm/Support/BinaryStreamArray.h

@@ -64,8 +64,10 @@ class VarStreamArrayIterator
 public:
   VarStreamArrayIterator() = default;
   VarStreamArrayIterator(const ArrayType &Array, const WrappedCtx &Ctx,
-                         BinaryStreamRef Stream, bool *HadError = nullptr)
-      : IterRef(Stream), Ctx(&Ctx), Array(&Array), HadError(HadError) {
+                         BinaryStreamRef Stream, bool *HadError = nullptr,
+                         uint32_t Offset = 0)
+      : IterRef(Stream), Ctx(&Ctx), Array(&Array), AbsOffset(Offset),
+        HadError(HadError) {
     if (IterRef.getLength() == 0)
       moveToEnd();
     else {
@@ -115,6 +117,7 @@ public:
     for (unsigned I = 0; I < N; ++I) {
       // We are done with the current record, discard it so that we are
       // positioned at the next record.
+      AbsOffset += ThisLen;
       IterRef = IterRef.drop_front(ThisLen);
       if (IterRef.getLength() == 0) {
         // There is nothing after the current record, we must make this an end
@@ -135,6 +138,8 @@ public:
     return *this;
   }
 
+  uint32_t offset() const { return AbsOffset; }
+
 private:
   void moveToEnd() {
     Array = nullptr;
@@ -152,6 +157,7 @@ private:
   const WrappedCtx *Ctx{nullptr};
   const ArrayType *Array{nullptr};
   uint32_t ThisLen{0};
+  uint32_t AbsOffset{0};
   bool HasError{false};
   bool *HadError{nullptr};
 };
@@ -234,7 +240,7 @@ public:
   /// since the behavior is undefined if \p Offset does not refer to the
   /// beginning of a valid record.
   Iterator at(uint32_t Offset) const {
-    return Iterator(*this, Ctx, Stream.drop_front(Offset), nullptr);
+    return Iterator(*this, Ctx, Stream.drop_front(Offset), nullptr, Offset);
   }
 
   BinaryStreamRef getUnderlyingStream() const { return Stream; }
@@ -338,7 +344,7 @@ private:
 template <typename T>
 class FixedStreamArrayIterator
     : public iterator_facade_base<FixedStreamArrayIterator<T>,
-                                  std::random_access_iterator_tag, T> {
+                                  std::random_access_iterator_tag, const T> {
 
 public:
   FixedStreamArrayIterator(const FixedStreamArray<T> &Array, uint32_t Index)
@@ -352,6 +358,7 @@ public:
   }
 
   const T &operator*() const { return Array[Index]; }
+  const T &operator*() { return Array[Index]; }
 
   bool operator==(const FixedStreamArrayIterator<T> &R) const {
     assert(Array == R.Array);

include/llvm/Support/COFF.h

@@ -152,6 +152,30 @@ namespace COFF {
     IMAGE_FILE_BYTES_REVERSED_HI       = 0x8000
   };
 
+  enum ResourceTypeID {
+    RID_Cursor = 1,
+    RID_Bitmap = 2,
+    RID_Icon = 3,
+    RID_Menu = 4,
+    RID_Dialog = 5,
+    RID_String = 6,
+    RID_FontDir = 7,
+    RID_Font = 8,
+    RID_Accelerator = 9,
+    RID_RCData = 10,
+    RID_MessageTable = 11,
+    RID_Group_Cursor = 12,
+    RID_Group_Icon = 14,
+    RID_Version = 16,
+    RID_DLGInclude = 17,
+    RID_PlugPlay = 19,
+    RID_VXD = 20,
+    RID_AniCursor = 21,
+    RID_AniIcon = 22,
+    RID_HTML = 23,
+    RID_Manifest = 24,
+  };
+
   struct symbol {
     char     Name[NameSize];
     uint32_t Value;
@@ -349,6 +373,26 @@ namespace COFF {
     IMAGE_REL_ARM_BLX23T    = 0x0015
   };
 
+  enum RelocationTypesARM64 {
+    IMAGE_REL_ARM64_ABSOLUTE        = 0x0000,
+    IMAGE_REL_ARM64_ADDR32          = 0x0001,
+    IMAGE_REL_ARM64_ADDR32NB        = 0x0002,
+    IMAGE_REL_ARM64_BRANCH26        = 0x0003,
+    IMAGE_REL_ARM64_PAGEBASE_REL2   = 0x0004,
+    IMAGE_REL_ARM64_REL21           = 0x0005,
+    IMAGE_REL_ARM64_PAGEOFFSET_12A  = 0x0006,
+    IMAGE_REL_ARM64_PAGEOFFSET_12L  = 0x0007,
+    IMAGE_REL_ARM64_SECREL          = 0x0008,
+    IMAGE_REL_ARM64_SECREL_LOW12A   = 0x0009,
+    IMAGE_REL_ARM64_SECREL_HIGH12A  = 0x000A,
+    IMAGE_REL_ARM64_SECREL_LOW12L   = 0x000B,
+    IMAGE_REL_ARM64_TOKEN           = 0x000C,
+    IMAGE_REL_ARM64_SECTION         = 0x000D,
+    IMAGE_REL_ARM64_ADDR64          = 0x000E,
+    IMAGE_REL_ARM64_BRANCH19        = 0x000F,
+    IMAGE_REL_ARM64_BRANCH14        = 0x0010,
+  };
+
   enum COMDATType {
     IMAGE_COMDAT_SELECT_NODUPLICATES = 1,
     IMAGE_COMDAT_SELECT_ANY,

include/llvm/Support/KnownBits.h

@@ -24,6 +24,12 @@ struct KnownBits {
   APInt Zero;
   APInt One;
 
+private:
+  // Internal constructor for creating a ConstantRange from two APInts.
+  KnownBits(APInt Zero, APInt One)
+      : Zero(std::move(Zero)), One(std::move(One)) {}
+
+public:
   // Default construct Zero and One.
   KnownBits() {}
 
@@ -37,6 +43,55 @@ struct KnownBits {
     return Zero.getBitWidth();
   }
 
+  /// Returns true if there is conflicting information.
+  bool hasConflict() const { return Zero.intersects(One); }
+
+  /// Returns true if we know the value of all bits.
+  bool isConstant() const {
+    assert(!hasConflict() && "KnownBits conflict!");
+    return Zero.countPopulation() + One.countPopulation() == getBitWidth();
+  }
+
+  /// Returns the value when all bits have a known value. This just returns One
+  /// with a protective assertion.
+  const APInt &getConstant() const {
+    assert(isConstant() && "Can only get value when all bits are known");
+    return One;
+  }
+
+  /// Returns true if we don't know any bits.
+  bool isUnknown() const { return Zero.isNullValue() && One.isNullValue(); }
+
+  /// Resets the known state of all bits.
+  void resetAll() {
+    Zero.clearAllBits();
+    One.clearAllBits();
+  }
+
+  /// Returns true if value is all zero.
+  bool isZero() const {
+    assert(!hasConflict() && "KnownBits conflict!");
+    return Zero.isAllOnesValue();
+  }
+
+  /// Returns true if value is all one bits.
+  bool isAllOnes() const {
+    assert(!hasConflict() && "KnownBits conflict!");
+    return One.isAllOnesValue();
+  }
+
+  /// Make all bits known to be zero and discard any previous information.
+  void setAllZero() {
+    Zero.setAllBits();
+    One.clearAllBits();
+  }
+
+  /// Make all bits known to be one and discard any previous information.
+  void setAllOnes() {
+    Zero.clearAllBits();
+    One.setAllBits();
+  }
+
   /// Returns true if this value is known to be negative.
   bool isNegative() const { return One.isSignBitSet(); }
 
@@ -54,6 +109,30 @@ struct KnownBits {
     assert(!isNegative() && "Can't make a negative value non-negative");
     Zero.setSignBit();
   }
+
+  /// Truncate the underlying known Zero and One bits. This is equivalent
+  /// to truncating the value we're tracking.
+  KnownBits trunc(unsigned BitWidth) {
+    return KnownBits(Zero.trunc(BitWidth), One.trunc(BitWidth));
+  }
+
+  /// Zero extends the underlying known Zero and One bits. This is equivalent
+  /// to zero extending the value we're tracking.
+  KnownBits zext(unsigned BitWidth) {
+    return KnownBits(Zero.zext(BitWidth), One.zext(BitWidth));
+  }
+
+  /// Sign extends the underlying known Zero and One bits. This is equivalent
+  /// to sign extending the value we're tracking.
+  KnownBits sext(unsigned BitWidth) {
+    return KnownBits(Zero.sext(BitWidth), One.sext(BitWidth));
+  }
+
+  /// Zero extends or truncates the underlying known Zero and One bits. This is
+  /// equivalent to zero extending or truncating the value we're tracking.
+  KnownBits zextOrTrunc(unsigned BitWidth) {
+    return KnownBits(Zero.zextOrTrunc(BitWidth), One.zextOrTrunc(BitWidth));
+  }
 };
 
 } // end namespace llvm

include/llvm/Support/MathExtras.h

@@ -214,6 +214,18 @@ template <typename T> T maskLeadingOnes(unsigned N) {
   return ~maskTrailingOnes<T>(CHAR_BIT * sizeof(T) - N);
 }
 
+/// \brief Create a bitmask with the N right-most bits set to 0, and all other
+/// bits set to 1.  Only unsigned types are allowed.
+template <typename T> T maskTrailingZeros(unsigned N) {
+  return maskLeadingOnes<T>(CHAR_BIT * sizeof(T) - N);
+}
+
+/// \brief Create a bitmask with the N left-most bits set to 0, and all other
+/// bits set to 1.  Only unsigned types are allowed.
+template <typename T> T maskLeadingZeros(unsigned N) {
+  return maskTrailingOnes<T>(CHAR_BIT * sizeof(T) - N);
+}
+
 /// \brief Get the index of the last set bit starting from the least
 ///   significant bit.
 ///

include/llvm/Target/GlobalISel/SelectionDAGCompat.td

@@ -25,25 +25,43 @@ class GINodeEquiv<Instruction i, SDNode node> {
   SDNode Node = node;
 }
 
-def : GINodeEquiv<G_ZEXT, zext>;
+// These are defined in the same order as the G_* instructions.
+def : GINodeEquiv<G_ANYEXT, anyext>;
 def : GINodeEquiv<G_SEXT, sext>;
+def : GINodeEquiv<G_ZEXT, zext>;
+def : GINodeEquiv<G_TRUNC, trunc>;
+def : GINodeEquiv<G_BITCAST, bitconvert>;
+// G_INTTOPTR - SelectionDAG has no equivalent.
+// G_PTRTOINT - SelectionDAG has no equivalent.
+// G_CONSTANT - Not needed since constants aren't operators.
+// G_FCONSTANT - Not needed since constants aren't operators.
 def : GINodeEquiv<G_ADD, add>;
 def : GINodeEquiv<G_SUB, sub>;
 def : GINodeEquiv<G_MUL, mul>;
-
-def : GINodeEquiv<G_OR, or>;
-def : GINodeEquiv<G_XOR, xor>;
-def : GINodeEquiv<G_AND, and>;
-
-def : GINodeEquiv<G_SHL, shl>;
-def : GINodeEquiv<G_LSHR, srl>;
-def : GINodeEquiv<G_ASHR, sra>;
-
 def : GINodeEquiv<G_SDIV, sdiv>;
 def : GINodeEquiv<G_UDIV, udiv>;
 def : GINodeEquiv<G_SREM, srem>;
 def : GINodeEquiv<G_UREM, urem>;
-
+def : GINodeEquiv<G_AND, and>;
+def : GINodeEquiv<G_OR, or>;
+def : GINodeEquiv<G_XOR, xor>;
+def : GINodeEquiv<G_SHL, shl>;
+def : GINodeEquiv<G_LSHR, srl>;
+def : GINodeEquiv<G_ASHR, sra>;
+def : GINodeEquiv<G_SELECT, select>;
+def : GINodeEquiv<G_FNEG, fneg>;
+def : GINodeEquiv<G_FPEXT, fpextend>;
+def : GINodeEquiv<G_FPTRUNC, ftrunc>;
+def : GINodeEquiv<G_FPTOSI, fp_to_sint>;
+def : GINodeEquiv<G_FPTOUI, fp_to_uint>;
+def : GINodeEquiv<G_SITOFP, sint_to_fp>;
+def : GINodeEquiv<G_UITOFP, uint_to_fp>;
+def : GINodeEquiv<G_FADD, fadd>;
+def : GINodeEquiv<G_FSUB, fsub>;
+def : GINodeEquiv<G_FMUL, fmul>;
+def : GINodeEquiv<G_FDIV, fdiv>;
+def : GINodeEquiv<G_FREM, frem>;
+def : GINodeEquiv<G_FPOW, fpow>;
 def : GINodeEquiv<G_BR, br>;
 
 // Specifies the GlobalISel equivalents for SelectionDAG's ComplexPattern.

include/llvm/Target/Target.td

@@ -1002,6 +1002,16 @@ def PATCHABLE_TAIL_CALL : Instruction {
   let hasSideEffects = 1;
   let isReturn = 1;
 }
+def PATCHABLE_EVENT_CALL : Instruction {
+  let OutOperandList = (outs);
+  let InOperandList = (ins ptr_rc:$event, i8imm:$size);
+  let AsmString = "# XRay Custom Event Log.";
+  let usesCustomInserter = 1;
+  let isCall = 1;
+  let mayLoad = 1;
+  let mayStore = 1;
+  let hasSideEffects = 1;
+}
 def FENTRY_CALL : Instruction {
   let OutOperandList = (outs unknown:$dst);
   let InOperandList = (ins variable_ops);

include/llvm/Target/TargetOpcodes.def

@@ -182,6 +182,10 @@ HANDLE_TARGET_OPCODE(PATCHABLE_FUNCTION_EXIT)
 /// PATCHABLE_RET which specifically only works for return instructions.
 HANDLE_TARGET_OPCODE(PATCHABLE_TAIL_CALL)
 
+/// Wraps a logging call and its arguments with nop sleds. At runtime, this can be
+/// patched to insert instrumentation instructions.
+HANDLE_TARGET_OPCODE(PATCHABLE_EVENT_CALL)
+
 /// The following generic opcodes are not supposed to appear after ISel.
 /// This is something we might want to relax, but for now, this is convenient
 /// to produce diagnostics.

include/llvm/Transforms/Instrumentation.h

@@ -177,6 +177,7 @@ struct SanitizerCoverageOptions {
   bool Use8bitCounters = false;
   bool TracePC = false;
   bool TracePCGuard = false;
+  bool NoPrune = false;
 
   SanitizerCoverageOptions() = default;
 };

include/llvm/Transforms/Scalar/Float2Int.h

@@ -31,7 +31,7 @@ public:
 
 private:
   void findRoots(Function &F, SmallPtrSet<Instruction *, 8> &Roots);
-  ConstantRange seen(Instruction *I, ConstantRange R);
+  void seen(Instruction *I, ConstantRange R);
   ConstantRange badRange();
   ConstantRange unknownRange();
   ConstantRange validateRange(ConstantRange R);

lib/Analysis/ConstantFolding.cpp

@@ -701,11 +701,10 @@ Constant *SymbolicallyEvaluateBinop(unsigned Opc, Constant *Op0, Constant *Op1,
       return Op1;
     }
 
-    APInt KnownZero = Known0.Zero | Known1.Zero;
-    APInt KnownOne = Known0.One & Known1.One;
-    if ((KnownZero | KnownOne).isAllOnesValue()) {
-      return ConstantInt::get(Op0->getType(), KnownOne);
-    }
+    Known0.Zero |= Known1.Zero;
+    Known0.One &= Known1.One;
+    if (Known0.isConstant())
+      return ConstantInt::get(Op0->getType(), Known0.getConstant());
   }
 
   // If the constant expr is something like &A[123] - &A[4].f, fold this into a

lib/Analysis/InstructionSimplify.cpp

@@ -1495,36 +1495,87 @@ static Value *simplifyAndOfICmpsWithSameOperands(ICmpInst *Op0, ICmpInst *Op1) {
 
 /// Commuted variants are assumed to be handled by calling this function again
 /// with the parameters swapped.
-static Value *SimplifyAndOfICmps(ICmpInst *Op0, ICmpInst *Op1) {
+static Value *simplifyOrOfICmpsWithSameOperands(ICmpInst *Op0, ICmpInst *Op1) {
+  ICmpInst::Predicate Pred0, Pred1;
+  Value *A ,*B;
+  if (!match(Op0, m_ICmp(Pred0, m_Value(A), m_Value(B))) ||
+      !match(Op1, m_ICmp(Pred1, m_Specific(A), m_Specific(B))))
+    return nullptr;
+
+  // We have (icmp Pred0, A, B) | (icmp Pred1, A, B).
+  // If Op1 is always implied true by Op0, then Op0 is a subset of Op1, and we
+  // can eliminate Op0 from this 'or'.
+  if (ICmpInst::isImpliedTrueByMatchingCmp(Pred0, Pred1))
+    return Op1;
+
+  // Check for any combination of predicates that cover the entire range of
+  // possibilities.
+  if ((Pred0 == ICmpInst::getInversePredicate(Pred1)) ||
+      (Pred0 == ICmpInst::ICMP_NE && ICmpInst::isTrueWhenEqual(Pred1)) ||
+      (Pred0 == ICmpInst::ICMP_SLE && Pred1 == ICmpInst::ICMP_SGE) ||
+      (Pred0 == ICmpInst::ICMP_ULE && Pred1 == ICmpInst::ICMP_UGE))
+    return getTrue(Op0->getType());
+
+  return nullptr;
+}
+
+/// Test if a pair of compares with a shared operand and 2 constants has an
+/// empty set intersection, full set union, or if one compare is a superset of
+/// the other.
+static Value *simplifyAndOrOfICmpsWithConstants(ICmpInst *Cmp0, ICmpInst *Cmp1,
+                                                bool IsAnd) {
+  // Look for this pattern: {and/or} (icmp X, C0), (icmp X, C1)).
+  if (Cmp0->getOperand(0) != Cmp1->getOperand(0))
+    return nullptr;
+
+  const APInt *C0, *C1;
+  if (!match(Cmp0->getOperand(1), m_APInt(C0)) ||
+      !match(Cmp1->getOperand(1), m_APInt(C1)))
+    return nullptr;
+
+  auto Range0 = ConstantRange::makeExactICmpRegion(Cmp0->getPredicate(), *C0);
+  auto Range1 = ConstantRange::makeExactICmpRegion(Cmp1->getPredicate(), *C1);
+
+  // For and-of-comapares, check if the intersection is empty:
+  // (icmp X, C0) && (icmp X, C1) --> empty set --> false
+  if (IsAnd && Range0.intersectWith(Range1).isEmptySet())
+    return getFalse(Cmp0->getType());
+
+  // For or-of-compares, check if the union is full:
+  // (icmp X, C0) || (icmp X, C1) --> full set --> true
+  if (!IsAnd && Range0.unionWith(Range1).isFullSet())
+    return getTrue(Cmp0->getType());
+
+  // Is one range a superset of the other?
+  // If this is and-of-compares, take the smaller set:
+  // (icmp sgt X, 4) && (icmp sgt X, 42) --> icmp sgt X, 42
+  // If this is or-of-compares, take the larger set:
+  // (icmp sgt X, 4) || (icmp sgt X, 42) --> icmp sgt X, 4
+  if (Range0.contains(Range1))
+    return IsAnd ? Cmp1 : Cmp0;
+  if (Range1.contains(Range0))
+    return IsAnd ? Cmp0 : Cmp1;
+
+  return nullptr;
+}
+
+/// Commuted variants are assumed to be handled by calling this function again
+/// with the parameters swapped.
+static Value *simplifyAndOfICmps(ICmpInst *Op0, ICmpInst *Op1) {
   if (Value *X = simplifyUnsignedRangeCheck(Op0, Op1, /*IsAnd=*/true))
     return X;
 
   if (Value *X = simplifyAndOfICmpsWithSameOperands(Op0, Op1))
     return X;
 
-  // FIXME: This should be shared with or-of-icmps.
-  // Look for this pattern: (icmp V, C0) & (icmp V, C1)).
+  if (Value *X = simplifyAndOrOfICmpsWithConstants(Op0, Op1, true))
+    return X;
+
+  // (icmp (add V, C0), C1) & (icmp V, C0)
   Type *ITy = Op0->getType();
   ICmpInst::Predicate Pred0, Pred1;
   const APInt *C0, *C1;
   Value *V;
-  if (match(Op0, m_ICmp(Pred0, m_Value(V), m_APInt(C0))) &&
-      match(Op1, m_ICmp(Pred1, m_Specific(V), m_APInt(C1)))) {
-    // Make a constant range that's the intersection of the two icmp ranges.
-    // If the intersection is empty, we know that the result is false.
-    auto Range0 = ConstantRange::makeExactICmpRegion(Pred0, *C0);
-    auto Range1 = ConstantRange::makeExactICmpRegion(Pred1, *C1);
-    if (Range0.intersectWith(Range1).isEmptySet())
-      return getFalse(ITy);
-
-    // If a range is a superset of the other, the smaller set is all we need.
-    if (Range0.contains(Range1))
-      return Op1;
-    if (Range1.contains(Range0))
-      return Op0;
-  }
-
-  // (icmp (add V, C0), C1) & (icmp V, C0)
   if (!match(Op0, m_ICmp(Pred0, m_Add(m_Value(V), m_APInt(C0)), m_APInt(C1))))
     return nullptr;
 
@@ -1565,6 +1616,103 @@ static Value *SimplifyAndOfICmps(ICmpInst *Op0, ICmpInst *Op1) {
   return nullptr;
 }
 
+/// Commuted variants are assumed to be handled by calling this function again
+/// with the parameters swapped.
+static Value *simplifyOrOfICmps(ICmpInst *Op0, ICmpInst *Op1) {
+  if (Value *X = simplifyUnsignedRangeCheck(Op0, Op1, /*IsAnd=*/false))
+    return X;
+
+  if (Value *X = simplifyOrOfICmpsWithSameOperands(Op0, Op1))
+    return X;
+
+  if (Value *X = simplifyAndOrOfICmpsWithConstants(Op0, Op1, false))
+    return X;
+
+  // (icmp (add V, C0), C1) | (icmp V, C0)
+  ICmpInst::Predicate Pred0, Pred1;
+  const APInt *C0, *C1;
+  Value *V;
+  if (!match(Op0, m_ICmp(Pred0, m_Add(m_Value(V), m_APInt(C0)), m_APInt(C1))))
+    return nullptr;
+
+  if (!match(Op1, m_ICmp(Pred1, m_Specific(V), m_Value())))
+    return nullptr;
+
+  auto *AddInst = cast<BinaryOperator>(Op0->getOperand(0));
+  if (AddInst->getOperand(1) != Op1->getOperand(1))
+    return nullptr;
+
+  Type *ITy = Op0->getType();
+  bool isNSW = AddInst->hasNoSignedWrap();
+  bool isNUW = AddInst->hasNoUnsignedWrap();
+
+  const APInt Delta = *C1 - *C0;
+  if (C0->isStrictlyPositive()) {
+    if (Delta == 2) {
+      if (Pred0 == ICmpInst::ICMP_UGE && Pred1 == ICmpInst::ICMP_SLE)
+        return getTrue(ITy);
+      if (Pred0 == ICmpInst::ICMP_SGE && Pred1 == ICmpInst::ICMP_SLE && isNSW)
+        return getTrue(ITy);
+    }
+    if (Delta == 1) {
+      if (Pred0 == ICmpInst::ICMP_UGT && Pred1 == ICmpInst::ICMP_SLE)
+        return getTrue(ITy);
+      if (Pred0 == ICmpInst::ICMP_SGT && Pred1 == ICmpInst::ICMP_SLE && isNSW)
+        return getTrue(ITy);
+    }
+  }
+  if (C0->getBoolValue() && isNUW) {
+    if (Delta == 2)
+      if (Pred0 == ICmpInst::ICMP_UGE && Pred1 == ICmpInst::ICMP_ULE)
+        return getTrue(ITy);
+    if (Delta == 1)
+      if (Pred0 == ICmpInst::ICMP_UGT && Pred1 == ICmpInst::ICMP_ULE)
+        return getTrue(ITy);
+  }
+
+  return nullptr;
+}
+
+static Value *simplifyPossiblyCastedAndOrOfICmps(ICmpInst *Cmp0, ICmpInst *Cmp1,
+                                                 bool IsAnd, CastInst *Cast) {
+  Value *V =
+      IsAnd ? simplifyAndOfICmps(Cmp0, Cmp1) : simplifyOrOfICmps(Cmp0, Cmp1);
+  if (!V)
+    return nullptr;
+  if (!Cast)
+    return V;
+
+  // If we looked through casts, we can only handle a constant simplification
+  // because we are not allowed to create a cast instruction here.
+  if (auto *C = dyn_cast<Constant>(V))
+    return ConstantExpr::getCast(Cast->getOpcode(), C, Cast->getType());
+
+  return nullptr;
+}
+
+static Value *simplifyAndOrOfICmps(Value *Op0, Value *Op1, bool IsAnd) {
+  // Look through casts of the 'and' operands to find compares.
+  auto *Cast0 = dyn_cast<CastInst>(Op0);
+  auto *Cast1 = dyn_cast<CastInst>(Op1);
+  if (Cast0 && Cast1 && Cast0->getOpcode() == Cast1->getOpcode() &&
+      Cast0->getSrcTy() == Cast1->getSrcTy()) {
+    Op0 = Cast0->getOperand(0);
+    Op1 = Cast1->getOperand(0);
+  }
+
+  auto *Cmp0 = dyn_cast<ICmpInst>(Op0);
+  auto *Cmp1 = dyn_cast<ICmpInst>(Op1);
+  if (!Cmp0 || !Cmp1)
+    return nullptr;
+
+  if (Value *V = simplifyPossiblyCastedAndOrOfICmps(Cmp0, Cmp1, IsAnd, Cast0))
+    return V;
+  if (Value *V = simplifyPossiblyCastedAndOrOfICmps(Cmp1, Cmp0, IsAnd, Cast0))
+    return V;
+
+  return nullptr;
+}
+
 /// Given operands for an And, see if we can fold the result.
 /// If not, this returns null.
 static Value *SimplifyAndInst(Value *Op0, Value *Op1, const SimplifyQuery &Q,
@@ -1615,32 +1763,8 @@ static Value *SimplifyAndInst(Value *Op0, Value *Op1, const SimplifyQuery &Q,
       return Op1;
   }
 
-  if (auto *ICILHS = dyn_cast<ICmpInst>(Op0)) {
-    if (auto *ICIRHS = dyn_cast<ICmpInst>(Op1)) {
-      if (Value *V = SimplifyAndOfICmps(ICILHS, ICIRHS))
-        return V;
-      if (Value *V = SimplifyAndOfICmps(ICIRHS, ICILHS))
-        return V;
-    }
-  }
-
-  // The compares may be hidden behind casts. Look through those and try the
-  // same folds as above.
-  auto *Cast0 = dyn_cast<CastInst>(Op0);
-  auto *Cast1 = dyn_cast<CastInst>(Op1);
-  if (Cast0 && Cast1 && Cast0->getOpcode() == Cast1->getOpcode() &&
-      Cast0->getSrcTy() == Cast1->getSrcTy()) {
-    auto *Cmp0 = dyn_cast<ICmpInst>(Cast0->getOperand(0));
-    auto *Cmp1 = dyn_cast<ICmpInst>(Cast1->getOperand(0));
-    if (Cmp0 && Cmp1) {
-      Instruction::CastOps CastOpc = Cast0->getOpcode();
-      Type *ResultType = Cast0->getType();
-      if (auto *V = dyn_cast_or_null<Constant>(SimplifyAndOfICmps(Cmp0, Cmp1)))
-        return ConstantExpr::getCast(CastOpc, V, ResultType);
-      if (auto *V = dyn_cast_or_null<Constant>(SimplifyAndOfICmps(Cmp1, Cmp0)))
-        return ConstantExpr::getCast(CastOpc, V, ResultType);
-    }
-  }
+  if (Value *V = simplifyAndOrOfICmps(Op0, Op1, true))
+    return V;
 
   // Try some generic simplifications for associative operations.
   if (Value *V = SimplifyAssociativeBinOp(Instruction::And, Op0, Op1, Q,
@@ -1678,86 +1802,6 @@ Value *llvm::SimplifyAndInst(Value *Op0, Value *Op1, const SimplifyQuery &Q) {
   return ::SimplifyAndInst(Op0, Op1, Q, RecursionLimit);
 }
 
-/// Commuted variants are assumed to be handled by calling this function again
-/// with the parameters swapped.
-static Value *simplifyOrOfICmpsWithSameOperands(ICmpInst *Op0, ICmpInst *Op1) {
-  ICmpInst::Predicate Pred0, Pred1;
-  Value *A ,*B;
-  if (!match(Op0, m_ICmp(Pred0, m_Value(A), m_Value(B))) ||
-      !match(Op1, m_ICmp(Pred1, m_Specific(A), m_Specific(B))))
-    return nullptr;
-
-  // We have (icmp Pred0, A, B) | (icmp Pred1, A, B).
-  // If Op1 is always implied true by Op0, then Op0 is a subset of Op1, and we
-  // can eliminate Op0 from this 'or'.
-  if (ICmpInst::isImpliedTrueByMatchingCmp(Pred0, Pred1))
-    return Op1;
-
-  // Check for any combination of predicates that cover the entire range of
-  // possibilities.
-  if ((Pred0 == ICmpInst::getInversePredicate(Pred1)) ||
-      (Pred0 == ICmpInst::ICMP_NE && ICmpInst::isTrueWhenEqual(Pred1)) ||
-      (Pred0 == ICmpInst::ICMP_SLE && Pred1 == ICmpInst::ICMP_SGE) ||
-      (Pred0 == ICmpInst::ICMP_ULE && Pred1 == ICmpInst::ICMP_UGE))
-    return getTrue(Op0->getType());
-
-  return nullptr;
-}
-
-/// Commuted variants are assumed to be handled by calling this function again
-/// with the parameters swapped.
-static Value *SimplifyOrOfICmps(ICmpInst *Op0, ICmpInst *Op1) {
-  if (Value *X = simplifyUnsignedRangeCheck(Op0, Op1, /*IsAnd=*/false))
-    return X;
-
-  if (Value *X = simplifyOrOfICmpsWithSameOperands(Op0, Op1))
-    return X;
-
-  // (icmp (add V, C0), C1) | (icmp V, C0)
-  ICmpInst::Predicate Pred0, Pred1;
-  const APInt *C0, *C1;
-  Value *V;
-  if (!match(Op0, m_ICmp(Pred0, m_Add(m_Value(V), m_APInt(C0)), m_APInt(C1))))
-    return nullptr;
-
-  if (!match(Op1, m_ICmp(Pred1, m_Specific(V), m_Value())))
-    return nullptr;
-
-  auto *AddInst = cast<BinaryOperator>(Op0->getOperand(0));
-  if (AddInst->getOperand(1) != Op1->getOperand(1))
-    return nullptr;
-
-  Type *ITy = Op0->getType();
-  bool isNSW = AddInst->hasNoSignedWrap();
-  bool isNUW = AddInst->hasNoUnsignedWrap();
-
-  const APInt Delta = *C1 - *C0;
-  if (C0->isStrictlyPositive()) {
-    if (Delta == 2) {
-      if (Pred0 == ICmpInst::ICMP_UGE && Pred1 == ICmpInst::ICMP_SLE)
-        return getTrue(ITy);
-      if (Pred0 == ICmpInst::ICMP_SGE && Pred1 == ICmpInst::ICMP_SLE && isNSW)
-        return getTrue(ITy);
-    }
-    if (Delta == 1) {
-      if (Pred0 == ICmpInst::ICMP_UGT && Pred1 == ICmpInst::ICMP_SLE)
-        return getTrue(ITy);
-      if (Pred0 == ICmpInst::ICMP_SGT && Pred1 == ICmpInst::ICMP_SLE && isNSW)
-        return getTrue(ITy);
-    }
-  }
-  if (C0->getBoolValue() && isNUW) {
-    if (Delta == 2)
-      if (Pred0 == ICmpInst::ICMP_UGE && Pred1 == ICmpInst::ICMP_ULE)
-        return getTrue(ITy);
-    if (Delta == 1)
-      if (Pred0 == ICmpInst::ICMP_UGT && Pred1 == ICmpInst::ICMP_ULE)
-        return getTrue(ITy);
-  }
-
-  return nullptr;
-}
-
 /// Given operands for an Or, see if we can fold the result.
 /// If not, this returns null.
 static Value *SimplifyOrInst(Value *Op0, Value *Op1, const SimplifyQuery &Q,
@@ -1826,14 +1870,8 @@ static Value *SimplifyOrInst(Value *Op0, Value *Op1, const SimplifyQuery &Q,
        match(Op1, m_c_And(m_Not(m_Specific(A)), m_Specific(B)))))
     return Op0;
 
-  if (auto *ICILHS = dyn_cast<ICmpInst>(Op0)) {
-    if (auto *ICIRHS = dyn_cast<ICmpInst>(Op1)) {
-      if (Value *V = SimplifyOrOfICmps(ICILHS, ICIRHS))
-        return V;
-      if (Value *V = SimplifyOrOfICmps(ICIRHS, ICILHS))
-        return V;
-    }
-  }
+  if (Value *V = simplifyAndOrOfICmps(Op0, Op1, false))
+    return V;
 
   // Try some generic simplifications for associative operations.
   if (Value *V = SimplifyAssociativeBinOp(Instruction::Or, Op0, Op1, Q,
@@ -4056,20 +4094,13 @@ static Value *SimplifyShuffleVectorInst(Value *Op0, Value *Op1, Constant *Mask,
   unsigned MaskNumElts = Mask->getType()->getVectorNumElements();
   unsigned InVecNumElts = InVecTy->getVectorNumElements();
 
-  auto *Op0Const = dyn_cast<Constant>(Op0);
-  auto *Op1Const = dyn_cast<Constant>(Op1);
-
-  // If all operands are constant, constant fold the shuffle.
-  if (Op0Const && Op1Const)
-    return ConstantFoldShuffleVectorInstruction(Op0Const, Op1Const, Mask);
-
   SmallVector<int, 32> Indices;
   ShuffleVectorInst::getShuffleMask(Mask, Indices);
   assert(MaskNumElts == Indices.size() &&
          "Size of Indices not same as number of mask elements?");
 
-  // If only one of the operands is constant, constant fold the shuffle if the
-  // mask does not select elements from the variable operand.
+  // Canonicalization: If mask does not select elements from an input vector,
+  // replace that input vector with undef.
   bool MaskSelects0 = false, MaskSelects1 = false;
   for (unsigned i = 0; i != MaskNumElts; ++i) {
     if (Indices[i] == -1)
@@ -4079,23 +4110,41 @@ static Value *SimplifyShuffleVectorInst(Value *Op0, Value *Op1, Constant *Mask,
     else
       MaskSelects1 = true;
   }
-  if (!MaskSelects0 && Op1Const)
-    return ConstantFoldShuffleVectorInstruction(UndefValue::get(InVecTy),
-                                                Op1Const, Mask);
-  if (!MaskSelects1 && Op0Const)
-    return ConstantFoldShuffleVectorInstruction(Op0Const,
-                                                UndefValue::get(InVecTy), Mask);
+  if (!MaskSelects0)
+    Op0 = UndefValue::get(InVecTy);
+  if (!MaskSelects1)
+    Op1 = UndefValue::get(InVecTy);
+
+  auto *Op0Const = dyn_cast<Constant>(Op0);
+  auto *Op1Const = dyn_cast<Constant>(Op1);
+
+  // If all operands are constant, constant fold the shuffle.
+  if (Op0Const && Op1Const)
+    return ConstantFoldShuffleVectorInstruction(Op0Const, Op1Const, Mask);
+
+  // Canonicalization: if only one input vector is constant, it shall be the
+  // second one.
+  if (Op0Const && !Op1Const) {
+    std::swap(Op0, Op1);
+    for (int &Idx : Indices) {
+      if (Idx == -1)
+        continue;
+      Idx = Idx < (int)InVecNumElts ? Idx + InVecNumElts : Idx - InVecNumElts;
+      assert(Idx >= 0 && Idx < (int)InVecNumElts * 2 &&
+             "shufflevector mask index out of range");
+    }
+    Mask = ConstantDataVector::get(
+        Mask->getContext(),
+        makeArrayRef(reinterpret_cast<uint32_t *>(Indices.data()),
+                     MaskNumElts));
+  }
 
   // A shuffle of a splat is always the splat itself. Legal if the shuffle's
   // value type is same as the input vectors' type.
   if (auto *OpShuf = dyn_cast<ShuffleVectorInst>(Op0))
-    if (!MaskSelects1 && RetTy == InVecTy &&
+    if (isa<UndefValue>(Op1) && RetTy == InVecTy &&
         OpShuf->getMask()->getSplatValue())
       return Op0;
-  if (auto *OpShuf = dyn_cast<ShuffleVectorInst>(Op1))
-    if (!MaskSelects0 && RetTy == InVecTy &&
-        OpShuf->getMask()->getSplatValue())
-      return Op1;
 
   // Don't fold a shuffle with undef mask elements. This may get folded in a
   // better way using demanded bits or other analysis.
@@ -4595,8 +4644,8 @@ Value *llvm::SimplifyInstruction(Instruction *I, const SimplifyQuery &SQ,
     unsigned BitWidth = I->getType()->getScalarSizeInBits();
     KnownBits Known(BitWidth);
     computeKnownBits(I, Known, Q.DL, /*Depth*/ 0, Q.AC, I, Q.DT, ORE);
-    if ((Known.Zero | Known.One).isAllOnesValue())
-      Result = ConstantInt::get(I->getType(), Known.One);
+    if (Known.isConstant())
+      Result = ConstantInt::get(I->getType(), Known.getConstant());
   }
 
   /// If called on unreachable code, the above logic may report that the

lib/Analysis/LazyValueInfo.cpp

@@ -142,7 +142,7 @@ public:
     return Val;
   }
 
-  ConstantRange getConstantRange() const {
+  const ConstantRange &getConstantRange() const {
     assert(isConstantRange() &&
            "Cannot get the constant-range of a non-constant-range!");
     return Range;
@@ -250,7 +250,7 @@ public:
     if (NewR.isFullSet())
       markOverdefined();
     else
-      markConstantRange(NewR);
+      markConstantRange(std::move(NewR));
   }
 };
 
@@ -1079,8 +1079,8 @@ bool LazyValueInfoImpl::solveBlockValueSelect(LVILatticeVal &BBLV,
   }
 
   if (TrueVal.isConstantRange() && FalseVal.isConstantRange()) {
-    ConstantRange TrueCR = TrueVal.getConstantRange();
-    ConstantRange FalseCR = FalseVal.getConstantRange();
+    const ConstantRange &TrueCR = TrueVal.getConstantRange();
+    const ConstantRange &FalseCR = FalseVal.getConstantRange();
     Value *LHS = nullptr;
     Value *RHS = nullptr;
     SelectPatternResult SPR = matchSelectPattern(SI, LHS, RHS);
@@ -1649,7 +1649,7 @@ Constant *LazyValueInfo::getConstant(Value *V, BasicBlock *BB,
   if (Result.isConstant())
     return Result.getConstant();
   if (Result.isConstantRange()) {
-    ConstantRange CR = Result.getConstantRange();
+    const ConstantRange &CR = Result.getConstantRange();
     if (const APInt *SingleVal = CR.getSingleElement())
       return ConstantInt::get(V->getContext(), *SingleVal);
   }
@@ -1686,7 +1686,7 @@ Constant *LazyValueInfo::getConstantOnEdge(Value *V, BasicBlock *FromBB,
   if (Result.isConstant())
     return Result.getConstant();
   if (Result.isConstantRange()) {
-    ConstantRange CR = Result.getConstantRange();
+    const ConstantRange &CR = Result.getConstantRange();
     if (const APInt *SingleVal = CR.getSingleElement())
       return ConstantInt::get(V->getContext(), *SingleVal);
   }
@@ -1712,7 +1712,7 @@ static LazyValueInfo::Tristate getPredicateResult(unsigned Pred, Constant *C,
     ConstantInt *CI = dyn_cast<ConstantInt>(C);
     if (!CI) return LazyValueInfo::Unknown;
 
-    ConstantRange CR = Result.getConstantRange();
+    const ConstantRange &CR = Result.getConstantRange();
     if (Pred == ICmpInst::ICMP_EQ) {
       if (!CR.contains(CI->getValue()))
         return LazyValueInfo::False;

lib/Analysis/Lint.cpp

@@ -537,7 +537,7 @@ static bool isZero(Value *V, const DataLayout &DL, DominatorTree *DT,
     unsigned BitWidth = V->getType()->getIntegerBitWidth();
     KnownBits Known(BitWidth);
     computeKnownBits(V, Known, DL, 0, AC, dyn_cast<Instruction>(V), DT);
-    return Known.Zero.isAllOnesValue();
+    return Known.isZero();
   }
 
   // Per-component check doesn't work with zeroinitializer
@@ -558,7 +558,7 @@ static bool isZero(Value *V, const DataLayout &DL, DominatorTree *DT,
 
     KnownBits Known(BitWidth);
     computeKnownBits(Elem, Known, DL);
-    if (Known.Zero.isAllOnesValue())
+    if (Known.isZero())
       return true;
   }
 

lib/Analysis/ModuleSummaryAnalysis.cpp

@@ -37,7 +37,8 @@ using namespace llvm;
 // Walk through the operands of a given User via worklist iteration and populate
 // the set of GlobalValue references encountered. Invoked either on an
 // Instruction or a GlobalVariable (which walks its initializer).
-static void findRefEdges(const User *CurUser, SetVector<ValueInfo> &RefEdges,
+static void findRefEdges(ModuleSummaryIndex &Index, const User *CurUser,
+                         SetVector<ValueInfo> &RefEdges,
                          SmallPtrSet<const User *, 8> &Visited) {
   SmallVector<const User *, 32> Worklist;
   Worklist.push_back(CurUser);
@@ -61,7 +62,7 @@ static void findRefEdges(const User *CurUser, SetVector<ValueInfo> &RefEdges,
         // the reference set unless it is a callee. Callees are handled
         // specially by WriteFunction and are added to a separate list.
         if (!(CS && CS.isCallee(&OI)))
-          RefEdges.insert(GV);
+          RefEdges.insert(Index.getOrInsertValueInfo(GV));
         continue;
       }
       Worklist.push_back(Operand);
@@ -198,7 +199,7 @@ computeFunctionSummary(ModuleSummaryIndex &Index, const Module &M,
       if (isa<DbgInfoIntrinsic>(I))
         continue;
       ++NumInsts;
-      findRefEdges(&I, RefEdges, Visited);
+      findRefEdges(Index, &I, RefEdges, Visited);
       auto CS = ImmutableCallSite(&I);
       if (!CS)
         continue;
@@ -239,7 +240,9 @@ computeFunctionSummary(ModuleSummaryIndex &Index, const Module &M,
         // to record the call edge to the alias in that case. Eventually
         // an alias summary will be created to associate the alias and
         // aliasee.
-        CallGraphEdges[cast<GlobalValue>(CalledValue)].updateHotness(Hotness);
+        CallGraphEdges[Index.getOrInsertValueInfo(
+                           cast<GlobalValue>(CalledValue))]
+            .updateHotness(Hotness);
       } else {
         // Skip inline assembly calls.
         if (CI && CI->isInlineAsm())
@@ -254,15 +257,16 @@ computeFunctionSummary(ModuleSummaryIndex &Index, const Module &M,
             ICallAnalysis.getPromotionCandidatesForInstruction(
                 &I, NumVals, TotalCount, NumCandidates);
         for (auto &Candidate : CandidateProfileData)
-          CallGraphEdges[Candidate.Value].updateHotness(
-              getHotness(Candidate.Count, PSI));
+          CallGraphEdges[Index.getOrInsertValueInfo(Candidate.Value)]
+              .updateHotness(getHotness(Candidate.Count, PSI));
       }
     }
 
   // Explicit add hot edges to enforce importing for designated GUIDs for
   // sample PGO, to enable the same inlines as the profiled optimized binary.
   for (auto &I : F.getImportGUIDs())
-    CallGraphEdges[I].updateHotness(CalleeInfo::HotnessType::Hot);
+    CallGraphEdges[Index.getOrInsertValueInfo(I)].updateHotness(
+        CalleeInfo::HotnessType::Hot);
 
   bool NonRenamableLocal = isNonRenamableLocal(F);
   bool NotEligibleForImport =
@@ -288,7 +292,7 @@ computeVariableSummary(ModuleSummaryIndex &Index, const GlobalVariable &V,
                        DenseSet<GlobalValue::GUID> &CantBePromoted) {
   SetVector<ValueInfo> RefEdges;
   SmallPtrSet<const User *, 8> Visited;
-  findRefEdges(&V, RefEdges, Visited);
+  findRefEdges(Index, &V, RefEdges, Visited);
   bool NonRenamableLocal = isNonRenamableLocal(V);
   GlobalValueSummary::GVFlags Flags(V.getLinkage(), NonRenamableLocal,
                                     /* LiveRoot = */ false);
@@ -317,12 +321,9 @@ computeAliasSummary(ModuleSummaryIndex &Index, const GlobalAlias &A,
 
 // Set LiveRoot flag on entries matching the given value name.
 static void setLiveRoot(ModuleSummaryIndex &Index, StringRef Name) {
-  auto SummaryList =
-      Index.findGlobalValueSummaryList(GlobalValue::getGUID(Name));
-  if (SummaryList == Index.end())
-    return;
-  for (auto &Summary : SummaryList->second)
-    Summary->setLiveRoot();
+  if (ValueInfo VI = Index.getValueInfo(GlobalValue::getGUID(Name)))
+    for (auto &Summary : VI.getSummaryList())
+      Summary->setLiveRoot();
 }
 
 ModuleSummaryIndex llvm::buildModuleSummaryIndex(
@@ -446,12 +447,16 @@ ModuleSummaryIndex llvm::buildModuleSummaryIndex(
   }
 
   for (auto &GlobalList : Index) {
-    assert(GlobalList.second.size() == 1 &&
+    // Ignore entries for references that are undefined in the current module.
+    if (GlobalList.second.SummaryList.empty())
+      continue;
+
+    assert(GlobalList.second.SummaryList.size() == 1 &&
            "Expected module's index to have one summary per GUID");
-    auto &Summary = GlobalList.second[0];
+    auto &Summary = GlobalList.second.SummaryList[0];
     bool AllRefsCanBeExternallyReferenced =
         llvm::all_of(Summary->refs(), [&](const ValueInfo &VI) {
-          return !CantBePromoted.count(VI.getValue()->getGUID());
+          return !CantBePromoted.count(VI.getGUID());
         });
     if (!AllRefsCanBeExternallyReferenced) {
       Summary->setNotEligibleToImport();
@@ -461,9 +466,7 @@ ModuleSummaryIndex llvm::buildModuleSummaryIndex(
     if (auto *FuncSummary = dyn_cast<FunctionSummary>(Summary.get())) {
       bool AllCallsCanBeExternallyReferenced = llvm::all_of(
           FuncSummary->calls(), [&](const FunctionSummary::EdgeTy &Edge) {
-            auto GUID = Edge.first.isGUID() ? Edge.first.getGUID()
-                                            : Edge.first.getValue()->getGUID();
-            return !CantBePromoted.count(GUID);
+            return !CantBePromoted.count(Edge.first.getGUID());
           });
       if (!AllCallsCanBeExternallyReferenced)
         Summary->setNotEligibleToImport();

lib/Analysis/ScalarEvolution.cpp

@@ -2970,7 +2970,7 @@ static const APInt gcd(const SCEVConstant *C1, const SCEVConstant *C2) {
   else if (ABW < BBW)
     A = A.zext(BBW);
 
-  return APIntOps::GreatestCommonDivisor(A, B);
+  return APIntOps::GreatestCommonDivisor(std::move(A), std::move(B));
 }
 
 /// Get a canonical unsigned division expression, or something simpler if
@@ -4083,6 +4083,56 @@ static Optional<BinaryOp> MatchBinaryOp(Value *V, DominatorTree &DT) {
   return None;
 }
 
+/// A helper function for createAddRecFromPHI to handle simple cases.
+///
+/// This function tries to find an AddRec expression for the simplest (yet most
+/// common) cases: PN = PHI(Start, OP(Self, LoopInvariant)).
+/// If it fails, createAddRecFromPHI will use a more general, but slow,
+/// technique for finding the AddRec expression.
+const SCEV *ScalarEvolution::createSimpleAffineAddRec(PHINode *PN,
+                                                      Value *BEValueV,
+                                                      Value *StartValueV) {
+  const Loop *L = LI.getLoopFor(PN->getParent());
+  assert(L && L->getHeader() == PN->getParent());
+  assert(BEValueV && StartValueV);
+
+  auto BO = MatchBinaryOp(BEValueV, DT);
+  if (!BO)
+    return nullptr;
+
+  if (BO->Opcode != Instruction::Add)
+    return nullptr;
+
+  const SCEV *Accum = nullptr;
+  if (BO->LHS == PN && L->isLoopInvariant(BO->RHS))
+    Accum = getSCEV(BO->RHS);
+  else if (BO->RHS == PN && L->isLoopInvariant(BO->LHS))
+    Accum = getSCEV(BO->LHS);
+
+  if (!Accum)
+    return nullptr;
+
+  SCEV::NoWrapFlags Flags = SCEV::FlagAnyWrap;
+  if (BO->IsNUW)
+    Flags = setFlags(Flags, SCEV::FlagNUW);
+  if (BO->IsNSW)
+    Flags = setFlags(Flags, SCEV::FlagNSW);
+
+  const SCEV *StartVal = getSCEV(StartValueV);
+  const SCEV *PHISCEV = getAddRecExpr(StartVal, Accum, L, Flags);
+
+  ValueExprMap[SCEVCallbackVH(PN, this)] = PHISCEV;
+
+  // We can add Flags to the post-inc expression only if we
+  // know that it is *undefined behavior* for BEValueV to
+  // overflow.
+  if (auto *BEInst = dyn_cast<Instruction>(BEValueV))
+    if (isLoopInvariant(Accum, L) && isAddRecNeverPoison(BEInst, L))
+      (void)getAddRecExpr(getAddExpr(StartVal, Accum), Accum, L, Flags);
+
+  return PHISCEV;
+}
+
 const SCEV *ScalarEvolution::createAddRecFromPHI(PHINode *PN) {
   const Loop *L = LI.getLoopFor(PN->getParent());
   if (!L || L->getHeader() != PN->getParent())
@@ -4111,10 +4161,16 @@ const SCEV *ScalarEvolution::createAddRecFromPHI(PHINode *PN) {
   if (!BEValueV || !StartValueV)
     return nullptr;
 
-  // While we are analyzing this PHI node, handle its value symbolically.
-  const SCEV *SymbolicName = getUnknown(PN);
   assert(ValueExprMap.find_as(PN) == ValueExprMap.end() &&
          "PHI node already processed?");
+
+  // First, try to find AddRec expression without creating a fictituos symbolic
+  // value for PN.
+  if (auto *S = createSimpleAffineAddRec(PN, BEValueV, StartValueV))
+    return S;
+
+  // Handle PHI node value symbolically.
+  const SCEV *SymbolicName = getUnknown(PN);
   ValueExprMap.insert({SCEVCallbackVH(PN, this), SymbolicName});
 
   // Using this symbolic name for the PHI, analyze the value coming around
@@ -4189,7 +4245,7 @@ const SCEV *ScalarEvolution::createAddRecFromPHI(PHINode *PN) {
         ValueExprMap[SCEVCallbackVH(PN, this)] = PHISCEV;
 
         // We can add Flags to the post-inc expression only if we
-        // know that it us *undefined behavior* for BEValueV to
+        // know that it is *undefined behavior* for BEValueV to
         // overflow.
         if (auto *BEInst = dyn_cast<Instruction>(BEValueV))
           if (isLoopInvariant(Accum, L) && isAddRecNeverPoison(BEInst, L))
@@ -4744,7 +4800,7 @@ ScalarEvolution::getRange(const SCEV *S,
       }
     }
 
-    return setRange(AddRec, SignHint, ConservativeResult);
+    return setRange(AddRec, SignHint, std::move(ConservativeResult));
   }
 
   if (const SCEVUnknown *U = dyn_cast<SCEVUnknown>(S)) {
@@ -4775,10 +4831,10 @@ ScalarEvolution::getRange(const SCEV *S,
                           APInt::getSignedMaxValue(BitWidth).ashr(NS - 1) + 1));
     }
 
-    return setRange(U, SignHint, ConservativeResult);
+    return setRange(U, SignHint, std::move(ConservativeResult));
   }
 
-  return setRange(S, SignHint, ConservativeResult);
+  return setRange(S, SignHint, std::move(ConservativeResult));
 }
 
 // Given a StartRange, Step and MaxBECount for an expression compute a range of
@@ -4786,8 +4842,8 @@ ScalarEvolution::getRange(const SCEV *S,
 // from StartRange and then is changed by Step up to MaxBECount times. Signed
 // argument defines if we treat Step as signed or unsigned.
 static ConstantRange getRangeForAffineARHelper(APInt Step,
-                                               ConstantRange StartRange,
-                                               APInt MaxBECount,
+                                               const ConstantRange &StartRange,
+                                               const APInt &MaxBECount,
                                                unsigned BitWidth, bool Signed) {
   // If either Step or MaxBECount is 0, then the expression won't change, and we
   // just need to return the initial range.
@@ -4826,8 +4882,8 @@ static ConstantRange getRangeForAffineARHelper(APInt Step,
   // if the expression is decreasing and will be increased by Offset otherwise.
   APInt StartLower = StartRange.getLower();
   APInt StartUpper = StartRange.getUpper() - 1;
-  APInt MovedBoundary =
-      Descending ? (StartLower - Offset) : (StartUpper + Offset);
+  APInt MovedBoundary = Descending ? (StartLower - std::move(Offset))
+                                   : (StartUpper + std::move(Offset));
 
   // It's possible that the new minimum/maximum value will fall into the initial
   // range (due to wrap around). This means that the expression can take any
@@ -4835,21 +4891,18 @@ static ConstantRange getRangeForAffineARHelper(APInt Step,
   if (StartRange.contains(MovedBoundary))
     return ConstantRange(BitWidth, /* isFullSet = */ true);
 
-  APInt NewLower, NewUpper;
-  if (Descending) {
-    NewLower = MovedBoundary;
-    NewUpper = StartUpper;
-  } else {
-    NewLower = StartLower;
-    NewUpper = MovedBoundary;
-  }
+  APInt NewLower =
+      Descending ? std::move(MovedBoundary) : std::move(StartLower);
+  APInt NewUpper =
+      Descending ? std::move(StartUpper) : std::move(MovedBoundary);
+  NewUpper += 1;
 
   // If we end up with full range, return a proper full range.
-  if (NewLower == NewUpper + 1)
+  if (NewLower == NewUpper)
     return ConstantRange(BitWidth, /* isFullSet = */ true);
 
   // No overflow detected, return [StartLower, StartUpper + Offset + 1) range.
-  return ConstantRange(NewLower, NewUpper + 1);
+  return ConstantRange(std::move(NewLower), std::move(NewUpper));
 }
 
 ConstantRange ScalarEvolution::getRangeForAffineAR(const SCEV *Start,
@@ -7323,7 +7376,6 @@ SolveQuadraticEquation(const SCEVAddRecExpr *AddRec, ScalarEvolution &SE) {
   const APInt &M = MC->getAPInt();
   const APInt &N = NC->getAPInt();
   APInt Two(BitWidth, 2);
-  APInt Four(BitWidth, 4);
 
   {
     using namespace APIntOps;
@@ -7339,7 +7391,7 @@ SolveQuadraticEquation(const SCEVAddRecExpr *AddRec, ScalarEvolution &SE) {
     // Compute the B^2-4ac term.
     APInt SqrtTerm(B);
     SqrtTerm *= B;
-    SqrtTerm -= Four * (A * C);
+    SqrtTerm -= 4 * (A * C);
 
     if (SqrtTerm.isNegative()) {
       // The loop is provably infinite.
@@ -8887,7 +8939,7 @@ bool ScalarEvolution::isImpliedCondOperandsViaRanges(ICmpInst::Predicate Pred,
   if (!Addend)
     return false;
 
-  APInt ConstFoundRHS = cast<SCEVConstant>(FoundRHS)->getAPInt();
+  const APInt &ConstFoundRHS = cast<SCEVConstant>(FoundRHS)->getAPInt();
 
   // `FoundLHSRange` is the range we know `FoundLHS` to be in by virtue of the
   // antecedent "`FoundLHS` `Pred` `FoundRHS`".
@@ -8899,7 +8951,7 @@ bool ScalarEvolution::isImpliedCondOperandsViaRanges(ICmpInst::Predicate Pred,
 
   // We can also compute the range of values for `LHS` that satisfy the
   // consequent, "`LHS` `Pred` `RHS`":
-  APInt ConstRHS = cast<SCEVConstant>(RHS)->getAPInt();
+  const APInt &ConstRHS = cast<SCEVConstant>(RHS)->getAPInt();
   ConstantRange SatisfyingLHSRange =
       ConstantRange::makeSatisfyingICmpRegion(Pred, ConstRHS);
 
@@ -8924,7 +8976,7 @@ bool ScalarEvolution::doesIVOverflowOnLT(const SCEV *RHS, const SCEV *Stride,
                                 .getSignedMax();
 
     // SMaxRHS + SMaxStrideMinusOne > SMaxValue => overflow!
-    return (MaxValue - MaxStrideMinusOne).slt(MaxRHS);
+    return (std::move(MaxValue) - std::move(MaxStrideMinusOne)).slt(MaxRHS);
   }
 
   APInt MaxRHS = getUnsignedRange(RHS).getUnsignedMax();
@@ -8933,7 +8985,7 @@ bool ScalarEvolution::doesIVOverflowOnLT(const SCEV *RHS, const SCEV *Stride,
                               .getUnsignedMax();
 
   // UMaxRHS + UMaxStrideMinusOne > UMaxValue => overflow!
-  return (MaxValue - MaxStrideMinusOne).ult(MaxRHS);
+  return (std::move(MaxValue) - std::move(MaxStrideMinusOne)).ult(MaxRHS);
 }
 
 bool ScalarEvolution::doesIVOverflowOnGT(const SCEV *RHS, const SCEV *Stride,
@@ -8950,7 +9002,7 @@ bool ScalarEvolution::doesIVOverflowOnGT(const SCEV *RHS, const SCEV *Stride,
                                .getSignedMax();
 
     // SMinRHS - SMaxStrideMinusOne < SMinValue => overflow!
-    return (MinValue + MaxStrideMinusOne).sgt(MinRHS);
+    return (std::move(MinValue) + std::move(MaxStrideMinusOne)).sgt(MinRHS);
   }
 
   APInt MinRHS = getUnsignedRange(RHS).getUnsignedMin();
@@ -8959,7 +9011,7 @@ bool ScalarEvolution::doesIVOverflowOnGT(const SCEV *RHS, const SCEV *Stride,
                             .getUnsignedMax();
 
   // UMinRHS - UMaxStrideMinusOne < UMinValue => overflow!
-  return (MinValue + MaxStrideMinusOne).ugt(MinRHS);
+  return (std::move(MinValue) + std::move(MaxStrideMinusOne)).ugt(MinRHS);
 }
 
 const SCEV *ScalarEvolution::computeBECount(const SCEV *Delta, const SCEV *Step,
@@ -9250,9 +9302,8 @@ const SCEV *SCEVAddRecExpr::getNumIterationsInRange(const ConstantRange &Range,
     // the upper value of the range must be the first possible exit value.
     // If A is negative then the lower of the range is the last possible loop
     // value.  Also note that we already checked for a full range.
-    APInt One(BitWidth,1);
     APInt A = cast<SCEVConstant>(getOperand(1))->getAPInt();
-    APInt End = A.sge(One) ? (Range.getUpper() - One) : Range.getLower();
+    APInt End = A.sge(1) ? (Range.getUpper() - 1) : Range.getLower();
 
     // The exit value should be (End+A)/A.
     APInt ExitVal = (End + A).udiv(A);
@@ -9268,7 +9319,7 @@ const SCEV *SCEVAddRecExpr::getNumIterationsInRange(const ConstantRange &Range,
     // Ensure that the previous value is in the range.  This is a sanity check.
     assert(Range.contains(
            EvaluateConstantChrecAtConstant(this,
-           ConstantInt::get(SE.getContext(), ExitVal - One), SE)->getValue()) &&
+           ConstantInt::get(SE.getContext(), ExitVal - 1), SE)->getValue()) &&
            "Linear scev computation is off in a bad way!");
     return SE.getConstant(ExitValue);
   } else if (isQuadratic()) {
@@ -9574,7 +9625,7 @@ const SCEV *ScalarEvolution::getElementSize(Instruction *Inst) {
 
 void ScalarEvolution::findArrayDimensions(SmallVectorImpl<const SCEV *> &Terms,
                                           SmallVectorImpl<const SCEV *> &Sizes,
-                                          const SCEV *ElementSize) const {
+                                          const SCEV *ElementSize) {
   if (Terms.size() < 1 || !ElementSize)
     return;
 
@@ -9590,7 +9641,7 @@ void ScalarEvolution::findArrayDimensions(SmallVectorImpl<const SCEV *> &Terms,
     });
 
   // Remove duplicates.
-  std::sort(Terms.begin(), Terms.end());
+  array_pod_sort(Terms.begin(), Terms.end());
   Terms.erase(std::unique(Terms.begin(), Terms.end()), Terms.end());
 
   // Put larger terms first.
@@ -9598,13 +9649,11 @@ void ScalarEvolution::findArrayDimensions(SmallVectorImpl<const SCEV *> &Terms,
     return numberOfTerms(LHS) > numberOfTerms(RHS);
   });
 
-  ScalarEvolution &SE = *const_cast<ScalarEvolution *>(this);
-
   // Try to divide all terms by the element size. If term is not divisible by
   // element size, proceed with the original term.
   for (const SCEV *&Term : Terms) {
     const SCEV *Q, *R;
-    SCEVDivision::divide(SE, Term, ElementSize, &Q, &R);
+    SCEVDivision::divide(*this, Term, ElementSize, &Q, &R);
     if (!Q->isZero())
       Term = Q;
   }
@@ -9613,7 +9662,7 @@ void ScalarEvolution::findArrayDimensions(SmallVectorImpl<const SCEV *> &Terms,
 
   // Remove constant factors.
   for (const SCEV *T : Terms)
-    if (const SCEV *NewT = removeConstantFactors(SE, T))
+    if (const SCEV *NewT = removeConstantFactors(*this, T))
       NewTerms.push_back(NewT);
 
   DEBUG({
@@ -9622,8 +9671,7 @@ void ScalarEvolution::findArrayDimensions(SmallVectorImpl<const SCEV *> &Terms,
         dbgs() << *T << "\n";
     });
 
-  if (NewTerms.empty() ||
-      !findArrayDimensionsRec(SE, NewTerms, Sizes)) {
+  if (NewTerms.empty() || !findArrayDimensionsRec(*this, NewTerms, Sizes)) {
     Sizes.clear();
     return;
   }

lib/Analysis/TargetLibraryInfo.cpp

@@ -1176,6 +1176,10 @@ bool TargetLibraryInfoImpl::isValidProtoForLibFunc(const FunctionType &FTy,
             FTy.getParamType(0)->isPointerTy() &&
             FTy.getParamType(1) == SizeTTy && FTy.getParamType(2) == SizeTTy);
 
+  case LibFunc_wcslen:
+    return (NumParams == 1 && FTy.getParamType(0)->isPointerTy() &&
+            FTy.getReturnType()->isIntegerTy());
+
   case LibFunc::NumLibFuncs:
     break;
   }

lib/Analysis/ValueTracking.cpp

@@ -59,8 +59,8 @@ static cl::opt<bool>
 DontImproveNonNegativePhiBits("dont-improve-non-negative-phi-bits",
                               cl::Hidden, cl::init(true));
 
-/// Returns the bitwidth of the given scalar or pointer type (if unknown returns
-/// 0). For vector types, returns the element type's bitwidth.
+/// Returns the bitwidth of the given scalar or pointer type. For vector types,
+/// returns the element type's bitwidth.
 static unsigned getBitWidth(Type *Ty, const DataLayout &DL) {
   if (unsigned BitWidth = Ty->getScalarSizeInBits())
     return BitWidth;
@@ -342,7 +342,6 @@ static void computeKnownBitsMul(const Value *Op0, const Value *Op1, bool NSW,
   // Also compute a conservative estimate for high known-0 bits.
   // More trickiness is possible, but this is sufficient for the
   // interesting case of alignment computation.
-  Known.One.clearAllBits();
   unsigned TrailZ = Known.Zero.countTrailingOnes() +
                     Known2.Zero.countTrailingOnes();
   unsigned LeadZ =  std::max(Known.Zero.countLeadingOnes() +
@@ -351,7 +350,7 @@ static void computeKnownBitsMul(const Value *Op0, const Value *Op1, bool NSW,
 
   TrailZ = std::min(TrailZ, BitWidth);
   LeadZ = std::min(LeadZ, BitWidth);
-  Known.Zero.clearAllBits();
+  Known.resetAll();
   Known.Zero.setLowBits(TrailZ);
   Known.Zero.setHighBits(LeadZ);
 
@@ -529,15 +528,13 @@ static void computeKnownBitsFromAssume(const Value *V, KnownBits &Known,
 
     if (Arg == V && isValidAssumeForContext(I, Q.CxtI, Q.DT)) {
       assert(BitWidth == 1 && "assume operand is not i1?");
-      Known.Zero.clearAllBits();
-      Known.One.setAllBits();
+      Known.setAllOnes();
       return;
     }
     if (match(Arg, m_Not(m_Specific(V))) &&
         isValidAssumeForContext(I, Q.CxtI, Q.DT)) {
       assert(BitWidth == 1 && "assume operand is not i1?");
-      Known.Zero.setAllBits();
-      Known.One.clearAllBits();
+      Known.setAllZero();
       return;
     }
 
@@ -719,7 +716,7 @@ static void computeKnownBitsFromAssume(const Value *V, KnownBits &Known,
       KnownBits RHSKnown(BitWidth);
       computeKnownBits(A, RHSKnown, Depth+1, Query(Q, I));
 
-      if (RHSKnown.One.isAllOnesValue() || RHSKnown.isNonNegative()) {
+      if (RHSKnown.isAllOnes() || RHSKnown.isNonNegative()) {
         // We know that the sign bit is zero.
         Known.makeNonNegative();
       }
@@ -741,7 +738,7 @@ static void computeKnownBitsFromAssume(const Value *V, KnownBits &Known,
       KnownBits RHSKnown(BitWidth);
       computeKnownBits(A, RHSKnown, Depth+1, Query(Q, I));
 
-      if (RHSKnown.Zero.isAllOnesValue() || RHSKnown.isNegative()) {
+      if (RHSKnown.isZero() || RHSKnown.isNegative()) {
         // We know that the sign bit is one.
         Known.makeNegative();
       }
@@ -776,8 +773,7 @@ static void computeKnownBitsFromAssume(const Value *V, KnownBits &Known,
   // behavior, or we might have a bug in the compiler. We can't assert/crash, so
   // clear out the known bits, try to warn the user, and hope for the best.
   if (Known.Zero.intersects(Known.One)) {
-    Known.Zero.clearAllBits();
-    Known.One.clearAllBits();
+    Known.resetAll();
 
     if (Q.ORE) {
       auto *CxtI = const_cast<Instruction *>(Q.CxtI);
@@ -813,10 +809,8 @@ static void computeKnownBitsFromShiftOperator(
     // If there is conflict between Known.Zero and Known.One, this must be an
     // overflowing left shift, so the shift result is undefined. Clear Known
     // bits so that other code could propagate this undef.
-    if ((Known.Zero & Known.One) != 0) {
-      Known.Zero.clearAllBits();
-      Known.One.clearAllBits();
-    }
+    if ((Known.Zero & Known.One) != 0)
+      Known.resetAll();
 
     return;
   }
@@ -826,8 +820,7 @@ static void computeKnownBitsFromShiftOperator(
   // If the shift amount could be greater than or equal to the bit-width of the LHS, the
   // value could be undef, so we don't know anything about it.
   if ((~Known.Zero).uge(BitWidth)) {
-    Known.Zero.clearAllBits();
-    Known.One.clearAllBits();
+    Known.resetAll();
     return;
   }
 
@@ -839,8 +832,7 @@ static void computeKnownBitsFromShiftOperator(
 
   // It would be more-clearly correct to use the two temporaries for this
   // calculation. Reusing the APInts here to prevent unnecessary allocations.
-  Known.Zero.clearAllBits();
-  Known.One.clearAllBits();
+  Known.resetAll();
 
   // If we know the shifter operand is nonzero, we can sometimes infer more
   // known bits. However this is expensive to compute, so be lazy about it and
@@ -886,10 +878,8 @@ static void computeKnownBitsFromShiftOperator(
   // return anything we'd like, but we need to make sure the sets of known bits
   // stay disjoint (it should be better for some other code to actually
   // propagate the undef than to pick a value here using known bits).
-  if (Known.Zero.intersects(Known.One)) {
-    Known.Zero.clearAllBits();
-    Known.One.clearAllBits();
-  }
+  if (Known.Zero.intersects(Known.One))
+    Known.resetAll();
 }
 
 static void computeKnownBitsFromOperator(const Operator *I, KnownBits &Known,
@@ -924,7 +914,7 @@ static void computeKnownBitsFromOperator(const Operator *I, KnownBits &Known,
                                        m_Value(Y))) ||
          match(I->getOperand(1), m_Add(m_Specific(I->getOperand(0)),
                                        m_Value(Y))))) {
-      Known2.Zero.clearAllBits(); Known2.One.clearAllBits();
+      Known2.resetAll();
       computeKnownBits(Y, Known2, Depth + 1, Q);
       if (Known2.One.countTrailingOnes() > 0)
         Known.Zero.setBit(0);
@@ -965,8 +955,7 @@ static void computeKnownBitsFromOperator(const Operator *I, KnownBits &Known,
     computeKnownBits(I->getOperand(0), Known2, Depth + 1, Q);
     unsigned LeadZ = Known2.Zero.countLeadingOnes();
 
-    Known2.One.clearAllBits();
-    Known2.Zero.clearAllBits();
+    Known2.resetAll();
     computeKnownBits(I->getOperand(1), Known2, Depth + 1, Q);
     unsigned RHSUnknownLeadingOnes = Known2.One.countLeadingZeros();
     if (RHSUnknownLeadingOnes != BitWidth)
@@ -1051,11 +1040,9 @@ static void computeKnownBitsFromOperator(const Operator *I, KnownBits &Known,
     SrcBitWidth = Q.DL.getTypeSizeInBits(SrcTy->getScalarType());
 
     assert(SrcBitWidth && "SrcBitWidth can't be zero");
-    Known.Zero = Known.Zero.zextOrTrunc(SrcBitWidth);
-    Known.One = Known.One.zextOrTrunc(SrcBitWidth);
+    Known = Known.zextOrTrunc(SrcBitWidth);
     computeKnownBits(I->getOperand(0), Known, Depth + 1, Q);
-    Known.Zero = Known.Zero.zextOrTrunc(BitWidth);
-    Known.One = Known.One.zextOrTrunc(BitWidth);
+    Known = Known.zextOrTrunc(BitWidth);
     // Any top bits are known to be zero.
     if (BitWidth > SrcBitWidth)
       Known.Zero.setBitsFrom(SrcBitWidth);
@@ -1076,13 +1063,11 @@ static void computeKnownBitsFromOperator(const Operator *I, KnownBits &Known,
     // Compute the bits in the result that are not present in the input.
     unsigned SrcBitWidth = I->getOperand(0)->getType()->getScalarSizeInBits();
 
-    Known.Zero = Known.Zero.trunc(SrcBitWidth);
-    Known.One = Known.One.trunc(SrcBitWidth);
+    Known = Known.trunc(SrcBitWidth);
     computeKnownBits(I->getOperand(0), Known, Depth + 1, Q);
     // If the sign bit of the input is known set or clear, then we know the
     // top bits of the result.
-    Known.Zero = Known.Zero.sext(BitWidth);
-    Known.One = Known.One.sext(BitWidth);
+    Known = Known.sext(BitWidth);
     break;
   }
   case Instruction::Shl: {
@@ -1202,8 +1187,7 @@ static void computeKnownBitsFromOperator(const Operator *I, KnownBits &Known,
 
     unsigned Leaders = std::max(Known.Zero.countLeadingOnes(),
                                 Known2.Zero.countLeadingOnes());
-    Known.One.clearAllBits();
-    Known.Zero.clearAllBits();
+    Known.resetAll();
     Known.Zero.setHighBits(Leaders);
     break;
   }
@@ -1504,8 +1488,7 @@ void computeKnownBits(const Value *V, KnownBits &Known, unsigned Depth,
   }
   // Null and aggregate-zero are all-zeros.
   if (isa<ConstantPointerNull>(V) || isa<ConstantAggregateZero>(V)) {
-    Known.One.clearAllBits();
-    Known.Zero.setAllBits();
+    Known.setAllZero();
     return;
   }
   // Handle a constant vector by taking the intersection of the known bits of
@@ -1532,8 +1515,7 @@ void computeKnownBits(const Value *V, KnownBits &Known, unsigned Depth,
       Constant *Element = CV->getAggregateElement(i);
       auto *ElementCI = dyn_cast_or_null<ConstantInt>(Element);
       if (!ElementCI) {
-        Known.Zero.clearAllBits();
-        Known.One.clearAllBits();
+        Known.resetAll();
         return;
       }
       Elt = ElementCI->getValue();
@@ -1544,7 +1526,7 @@ void computeKnownBits(const Value *V, KnownBits &Known, unsigned Depth,
   }
 
   // Start out not knowing anything.
-  Known.Zero.clearAllBits(); Known.One.clearAllBits();
+  Known.resetAll();
 
   // We can't imply anything about undefs.
   if (isa<UndefValue>(V))
@@ -1590,13 +1572,7 @@ void computeKnownBits(const Value *V, KnownBits &Known, unsigned Depth,
 /// Convenience wrapper around computeKnownBits.
 void ComputeSignBit(const Value *V, bool &KnownZero, bool &KnownOne,
                     unsigned Depth, const Query &Q) {
-  unsigned BitWidth = getBitWidth(V->getType(), Q.DL);
-  if (!BitWidth) {
-    KnownZero = false;
-    KnownOne = false;
-    return;
-  }
-  KnownBits Bits(BitWidth);
+  KnownBits Bits(getBitWidth(V->getType(), Q.DL));
   computeKnownBits(V, Bits, Depth, Q);
   KnownOne = Bits.isNegative();
   KnownZero = Bits.isNonNegative();
@@ -1847,7 +1823,7 @@ bool isKnownNonZero(const Value *V, unsigned Depth, const Query &Q) {
 
   // shl X, Y != 0 if X is odd.  Note that the value of the shift is undefined
   // if the lowest bit is shifted off the end.
-  if (BitWidth && match(V, m_Shl(m_Value(X), m_Value(Y)))) {
+  if (match(V, m_Shl(m_Value(X), m_Value(Y)))) {
     // shl nuw can't remove any non-zero bits.
     const OverflowingBinaryOperator *BO = cast<OverflowingBinaryOperator>(V);
     if (BO->hasNoUnsignedWrap())
@@ -1906,7 +1882,7 @@ bool isKnownNonZero(const Value *V, unsigned Depth, const Query &Q) {
 
     // If X and Y are both negative (as signed values) then their sum is not
     // zero unless both X and Y equal INT_MIN.
-    if (BitWidth && XKnownNegative && YKnownNegative) {
+    if (XKnownNegative && YKnownNegative) {
       KnownBits Known(BitWidth);
       APInt Mask = APInt::getSignedMaxValue(BitWidth);
       // The sign bit of X is set.  If some other bit is set then X is not equal
@@ -1971,7 +1947,6 @@ bool isKnownNonZero(const Value *V, unsigned Depth, const Query &Q) {
       return true;
   }
 
-  if (!BitWidth) return false;
   KnownBits Known(BitWidth);
   computeKnownBits(V, Known, Depth, Q);
   return Known.One != 0;

lib/Bitcode/Reader/BitcodeReader.cpp

@@ -694,15 +694,16 @@ class ModuleSummaryIndexBitcodeReader : public BitcodeReaderBase {
   /// Used to enable on-demand parsing of the VST.
   uint64_t VSTOffset = 0;
 
-  // Map to save ValueId to GUID association that was recorded in the
+  // Map to save ValueId to ValueInfo association that was recorded in the
   // ValueSymbolTable. It is used after the VST is parsed to convert
   // call graph edges read from the function summary from referencing
-  // callees by their ValueId to using the GUID instead, which is how
+  // callees by their ValueId to using the ValueInfo instead, which is how
   // they are recorded in the summary index being built.
-  // We save a second GUID which is the same as the first one, but ignoring the
-  // linkage, i.e. for value other than local linkage they are identical.
-  DenseMap<unsigned, std::pair<GlobalValue::GUID, GlobalValue::GUID>>
-      ValueIdToCallGraphGUIDMap;
+  // We save a GUID which refers to the same global as the ValueInfo, but
+  // ignoring the linkage, i.e. for values other than local linkage they are
+  // identical.
+  DenseMap<unsigned, std::pair<ValueInfo, GlobalValue::GUID>>
+      ValueIdToValueInfoMap;
 
   /// Map populated during module path string table parsing, from the
   /// module ID to a string reference owned by the index's module
@@ -742,8 +743,8 @@ private:
   Error parseEntireSummary();
   Error parseModuleStringTable();
 
-  std::pair<GlobalValue::GUID, GlobalValue::GUID>
-  getGUIDFromValueId(unsigned ValueId);
+  std::pair<ValueInfo, GlobalValue::GUID>
+  getValueInfoFromValueId(unsigned ValueId);
 
   ModulePathStringTableTy::iterator addThisModulePath();
 };
@@ -4697,11 +4698,11 @@ ModuleSummaryIndexBitcodeReader::addThisModulePath() {
   return TheIndex.addModulePath(ModulePath, ModuleId);
 }
 
-std::pair<GlobalValue::GUID, GlobalValue::GUID>
-ModuleSummaryIndexBitcodeReader::getGUIDFromValueId(unsigned ValueId) {
-  auto VGI = ValueIdToCallGraphGUIDMap.find(ValueId);
-  assert(VGI != ValueIdToCallGraphGUIDMap.end());
-  return VGI->second;
+std::pair<ValueInfo, GlobalValue::GUID>
+ModuleSummaryIndexBitcodeReader::getValueInfoFromValueId(unsigned ValueId) {
+  auto VGI = ValueIdToValueInfoMap[ValueId];
+  assert(VGI.first);
+  return VGI;
 }
 
 void ModuleSummaryIndexBitcodeReader::setValueGUID(
@@ -4716,8 +4717,8 @@ void ModuleSummaryIndexBitcodeReader::setValueGUID(
   if (PrintSummaryGUIDs)
     dbgs() << "GUID " << ValueGUID << "(" << OriginalNameID << ") is "
            << ValueName << "\n";
-  ValueIdToCallGraphGUIDMap[ValueID] =
-      std::make_pair(ValueGUID, OriginalNameID);
+  ValueIdToValueInfoMap[ValueID] =
+      std::make_pair(TheIndex.getOrInsertValueInfo(ValueGUID), OriginalNameID);
 }
 
 // Specialized value symbol table parser used when reading module index
@@ -4795,7 +4796,8 @@ Error ModuleSummaryIndexBitcodeReader::parseValueSymbolTable(
       GlobalValue::GUID RefGUID = Record[1];
       // The "original name", which is the second value of the pair will be
       // overriden later by a FS_COMBINED_ORIGINAL_NAME in the combined index.
-      ValueIdToCallGraphGUIDMap[ValueID] = std::make_pair(RefGUID, RefGUID);
+      ValueIdToValueInfoMap[ValueID] =
+          std::make_pair(TheIndex.getOrInsertValueInfo(RefGUID), RefGUID);
       break;
     }
     }
@@ -4940,7 +4942,7 @@ ModuleSummaryIndexBitcodeReader::makeRefList(ArrayRef<uint64_t> Record) {
   std::vector<ValueInfo> Ret;
   Ret.reserve(Record.size());
   for (uint64_t RefValueId : Record)
-    Ret.push_back(getGUIDFromValueId(RefValueId).first);
+    Ret.push_back(getValueInfoFromValueId(RefValueId).first);
   return Ret;
 }
 
@@ -4950,14 +4952,14 @@ std::vector<FunctionSummary::EdgeTy> ModuleSummaryIndexBitcodeReader::makeCallLi
   Ret.reserve(Record.size());
   for (unsigned I = 0, E = Record.size(); I != E; ++I) {
     CalleeInfo::HotnessType Hotness = CalleeInfo::HotnessType::Unknown;
-    GlobalValue::GUID CalleeGUID = getGUIDFromValueId(Record[I]).first;
+    ValueInfo Callee = getValueInfoFromValueId(Record[I]).first;
     if (IsOldProfileFormat) {
       I += 1; // Skip old callsitecount field
       if (HasProfile)
         I += 1; // Skip old profilecount field
     } else if (HasProfile)
       Hotness = static_cast<CalleeInfo::HotnessType>(Record[++I]);
-    Ret.push_back(FunctionSummary::EdgeTy{CalleeGUID, CalleeInfo{Hotness}});
+    Ret.push_back(FunctionSummary::EdgeTy{Callee, CalleeInfo{Hotness}});
   }
   return Ret;
 }
@@ -5027,7 +5029,8 @@ Error ModuleSummaryIndexBitcodeReader::parseEntireSummary() {
     case bitc::FS_VALUE_GUID: { // [valueid, refguid]
       uint64_t ValueID = Record[0];
       GlobalValue::GUID RefGUID = Record[1];
-      ValueIdToCallGraphGUIDMap[ValueID] = std::make_pair(RefGUID, RefGUID);
+      ValueIdToValueInfoMap[ValueID] =
+          std::make_pair(TheIndex.getOrInsertValueInfo(RefGUID), RefGUID);
       break;
     }
     // FS_PERMODULE: [valueid, flags, instcount, numrefs, numrefs x valueid,
@@ -5068,10 +5071,10 @@ Error ModuleSummaryIndexBitcodeReader::parseEntireSummary() {
       PendingTypeCheckedLoadVCalls.clear();
       PendingTypeTestAssumeConstVCalls.clear();
       PendingTypeCheckedLoadConstVCalls.clear();
-      auto GUID = getGUIDFromValueId(ValueID);
+      auto VIAndOriginalGUID = getValueInfoFromValueId(ValueID);
       FS->setModulePath(addThisModulePath()->first());
-      FS->setOriginalName(GUID.second);
-      TheIndex.addGlobalValueSummary(GUID.first, std::move(FS));
+      FS->setOriginalName(VIAndOriginalGUID.second);
+      TheIndex.addGlobalValueSummary(VIAndOriginalGUID.first, std::move(FS));
       break;
     }
     // FS_ALIAS: [valueid, flags, valueid]
@@ -5091,14 +5094,15 @@ Error ModuleSummaryIndexBitcodeReader::parseEntireSummary() {
       // ownership.
       AS->setModulePath(addThisModulePath()->first());
 
-      GlobalValue::GUID AliaseeGUID = getGUIDFromValueId(AliaseeID).first;
+      GlobalValue::GUID AliaseeGUID =
+          getValueInfoFromValueId(AliaseeID).first.getGUID();
       auto AliaseeInModule =
           TheIndex.findSummaryInModule(AliaseeGUID, ModulePath);
       if (!AliaseeInModule)
         return error("Alias expects aliasee summary to be parsed");
       AS->setAliasee(AliaseeInModule);
 
-      auto GUID = getGUIDFromValueId(ValueID);
+      auto GUID = getValueInfoFromValueId(ValueID);
       AS->setOriginalName(GUID.second);
       TheIndex.addGlobalValueSummary(GUID.first, std::move(AS));
       break;
@@ -5112,7 +5116,7 @@ Error ModuleSummaryIndexBitcodeReader::parseEntireSummary() {
           makeRefList(ArrayRef<uint64_t>(Record).slice(2));
       auto FS = llvm::make_unique<GlobalVarSummary>(Flags, std::move(Refs));
       FS->setModulePath(addThisModulePath()->first());
-      auto GUID = getGUIDFromValueId(ValueID);
+      auto GUID = getValueInfoFromValueId(ValueID);
       FS->setOriginalName(GUID.second);
       TheIndex.addGlobalValueSummary(GUID.first, std::move(FS));
       break;
@@ -5139,7 +5143,7 @@ Error ModuleSummaryIndexBitcodeReader::parseEntireSummary() {
       std::vector<FunctionSummary::EdgeTy> Edges = makeCallList(
           ArrayRef<uint64_t>(Record).slice(CallGraphEdgeStartIndex),
           IsOldProfileFormat, HasProfile);
-      GlobalValue::GUID GUID = getGUIDFromValueId(ValueID).first;
+      ValueInfo VI = getValueInfoFromValueId(ValueID).first;
       auto FS = llvm::make_unique<FunctionSummary>(
           Flags, InstCount, std::move(Refs), std::move(Edges),
           std::move(PendingTypeTests), std::move(PendingTypeTestAssumeVCalls),
@@ -5152,9 +5156,9 @@ Error ModuleSummaryIndexBitcodeReader::parseEntireSummary() {
       PendingTypeTestAssumeConstVCalls.clear();
       PendingTypeCheckedLoadConstVCalls.clear();
       LastSeenSummary = FS.get();
-      LastSeenGUID = GUID;
+      LastSeenGUID = VI.getGUID();
       FS->setModulePath(ModuleIdMap[ModuleId]);
-      TheIndex.addGlobalValueSummary(GUID, std::move(FS));
+      TheIndex.addGlobalValueSummary(VI, std::move(FS));
       break;
     }
     // FS_COMBINED_ALIAS: [valueid, modid, flags, valueid]
@@ -5170,16 +5174,17 @@ Error ModuleSummaryIndexBitcodeReader::parseEntireSummary() {
       LastSeenSummary = AS.get();
       AS->setModulePath(ModuleIdMap[ModuleId]);
 
-      auto AliaseeGUID = getGUIDFromValueId(AliaseeValueId).first;
+      auto AliaseeGUID =
+          getValueInfoFromValueId(AliaseeValueId).first.getGUID();
       auto AliaseeInModule =
           TheIndex.findSummaryInModule(AliaseeGUID, AS->modulePath());
       if (!AliaseeInModule)
         return error("Alias expects aliasee summary to be parsed");
       AS->setAliasee(AliaseeInModule);
 
-      GlobalValue::GUID GUID = getGUIDFromValueId(ValueID).first;
-      LastSeenGUID = GUID;
-      TheIndex.addGlobalValueSummary(GUID, std::move(AS));
+      ValueInfo VI = getValueInfoFromValueId(ValueID).first;
+      LastSeenGUID = VI.getGUID();
+      TheIndex.addGlobalValueSummary(VI, std::move(AS));
       break;
     }
     // FS_COMBINED_GLOBALVAR_INIT_REFS: [valueid, modid, flags, n x valueid]
@@ -5193,9 +5198,9 @@ Error ModuleSummaryIndexBitcodeReader::parseEntireSummary() {
       auto FS = llvm::make_unique<GlobalVarSummary>(Flags, std::move(Refs));
       LastSeenSummary = FS.get();
       FS->setModulePath(ModuleIdMap[ModuleId]);
-      GlobalValue::GUID GUID = getGUIDFromValueId(ValueID).first;
-      LastSeenGUID = GUID;
-      TheIndex.addGlobalValueSummary(GUID, std::move(FS));
+      ValueInfo VI = getValueInfoFromValueId(ValueID).first;
+      LastSeenGUID = VI.getGUID();
+      TheIndex.addGlobalValueSummary(VI, std::move(FS));
       break;
     }
     // FS_COMBINED_ORIGINAL_NAME: [original_name]

lib/Bitcode/Writer/BitcodeWriter.cpp

@@ -156,14 +156,14 @@ public:
       return;
     for (const auto &GUIDSummaryLists : *Index)
       // Examine all summaries for this GUID.
-      for (auto &Summary : GUIDSummaryLists.second)
+      for (auto &Summary : GUIDSummaryLists.second.SummaryList)
         if (auto FS = dyn_cast<FunctionSummary>(Summary.get()))
           // For each call in the function summary, see if the call
           // is to a GUID (which means it is for an indirect call,
           // otherwise we would have a Value for it). If so, synthesize
           // a value id.
           for (auto &CallEdge : FS->calls())
-            if (CallEdge.first.isGUID())
+            if (!CallEdge.first.getValue())
               assignValueId(CallEdge.first.getGUID());
   }
 
@@ -304,7 +304,7 @@ private:
   }
   // Helper to get the valueId for the type of value recorded in VI.
   unsigned getValueId(ValueInfo VI) {
-    if (VI.isGUID())
+    if (!VI.getValue())
       return getValueId(VI.getGUID());
     return VE.getValueID(VI.getValue());
   }
@@ -358,7 +358,7 @@ public:
           Callback(Summary);
     } else {
       for (auto &Summaries : Index)
-        for (auto &Summary : Summaries.second)
+        for (auto &Summary : Summaries.second.SummaryList)
           Callback({Summaries.first, Summary.get()});
     }
   }
@@ -3270,15 +3270,14 @@ void ModuleBitcodeWriter::writePerModuleFunctionSummaryRecord(
 void ModuleBitcodeWriter::writeModuleLevelReferences(
     const GlobalVariable &V, SmallVector<uint64_t, 64> &NameVals,
     unsigned FSModRefsAbbrev) {
-  auto Summaries =
-      Index->findGlobalValueSummaryList(GlobalValue::getGUID(V.getName()));
-  if (Summaries == Index->end()) {
+  auto VI = Index->getValueInfo(GlobalValue::getGUID(V.getName()));
+  if (!VI || VI.getSummaryList().empty()) {
     // Only declarations should not have a summary (a declaration might however
     // have a summary if the def was in module level asm).
     assert(V.isDeclaration());
     return;
   }
-  auto *Summary = Summaries->second.front().get();
+  auto *Summary = VI.getSummaryList()[0].get();
   NameVals.push_back(VE.getValueID(&V));
   GlobalVarSummary *VS = cast<GlobalVarSummary>(Summary);
   NameVals.push_back(getEncodedGVSummaryFlags(VS->flags()));
@@ -3367,15 +3366,14 @@ void ModuleBitcodeWriter::writePerModuleGlobalValueSummary() {
     if (!F.hasName())
       report_fatal_error("Unexpected anonymous function when writing summary");
 
-    auto Summaries =
-        Index->findGlobalValueSummaryList(GlobalValue::getGUID(F.getName()));
-    if (Summaries == Index->end()) {
+    ValueInfo VI = Index->getValueInfo(GlobalValue::getGUID(F.getName()));
+    if (!VI || VI.getSummaryList().empty()) {
       // Only declarations should not have a summary (a declaration might
       // however have a summary if the def was in module level asm).
       assert(F.isDeclaration());
       continue;
     }
-    auto *Summary = Summaries->second.front().get();
+    auto *Summary = VI.getSummaryList()[0].get();
     writePerModuleFunctionSummaryRecord(NameVals, Summary, VE.getValueID(&F),
                                         FSCallsAbbrev, FSCallsProfileAbbrev, F);
   }

lib/CodeGen/AsmPrinter/AsmPrinter.cpp

@@ -2761,37 +2761,63 @@ void AsmPrinter::emitXRayTable() {
 
   auto PrevSection = OutStreamer->getCurrentSectionOnly();
   auto Fn = MF->getFunction();
-  MCSection *Section = nullptr;
+  MCSection *InstMap = nullptr;
+  MCSection *FnSledIndex = nullptr;
   if (MF->getSubtarget().getTargetTriple().isOSBinFormatELF()) {
     if (Fn->hasComdat()) {
-      Section = OutContext.getELFSection("xray_instr_map", ELF::SHT_PROGBITS,
+      InstMap = OutContext.getELFSection("xray_instr_map", ELF::SHT_PROGBITS,
                                          ELF::SHF_ALLOC | ELF::SHF_GROUP, 0,
                                          Fn->getComdat()->getName());
+      FnSledIndex = OutContext.getELFSection("xray_fn_idx", ELF::SHT_PROGBITS,
+                                             ELF::SHF_ALLOC | ELF::SHF_GROUP, 0,
+                                             Fn->getComdat()->getName());
     } else {
-      Section = OutContext.getELFSection("xray_instr_map", ELF::SHT_PROGBITS,
+      InstMap = OutContext.getELFSection("xray_instr_map", ELF::SHT_PROGBITS,
                                          ELF::SHF_ALLOC);
+      FnSledIndex = OutContext.getELFSection("xray_fn_idx", ELF::SHT_PROGBITS,
+                                             ELF::SHF_ALLOC);
     }
   } else if (MF->getSubtarget().getTargetTriple().isOSBinFormatMachO()) {
-    Section = OutContext.getMachOSection("__DATA", "xray_instr_map", 0,
+    InstMap = OutContext.getMachOSection("__DATA", "xray_instr_map", 0,
                                          SectionKind::getReadOnlyWithRel());
+    FnSledIndex = OutContext.getMachOSection("__DATA", "xray_fn_idx", 0,
+                                             SectionKind::getReadOnlyWithRel());
   } else {
     llvm_unreachable("Unsupported target");
   }
 
   // Before we switch over, we force a reference to a label inside the
-  // xray_instr_map section. Since this function is always called just
-  // before the function's end, we assume that this is happening after
-  // the last return instruction.
-
+  // xray_instr_map and xray_fn_idx sections. Since this function is always
+  // called just before the function's end, we assume that this is happening
+  // after the last return instruction. We also use the synthetic label in the
+  // xray_inster_map as a delimeter for the range of sleds for this function in
+  // the index.
   auto WordSizeBytes = MAI->getCodePointerSize();
-  MCSymbol *Tmp = OutContext.createTempSymbol("xray_synthetic_", true);
+  MCSymbol *SledsStart = OutContext.createTempSymbol("xray_synthetic_", true);
+  MCSymbol *IdxRef = OutContext.createTempSymbol("xray_fn_idx_synth_", true);
   OutStreamer->EmitCodeAlignment(16);
-  OutStreamer->EmitSymbolValue(Tmp, WordSizeBytes, false);
-  OutStreamer->SwitchSection(Section);
-  OutStreamer->EmitLabel(Tmp);
+  OutStreamer->EmitSymbolValue(SledsStart, WordSizeBytes, false);
+  OutStreamer->EmitSymbolValue(IdxRef, WordSizeBytes, false);
+
+  // Now we switch to the instrumentation map section. Because this is done
+  // per-function, we are able to create an index entry that will represent the
+  // range of sleds associated with a function.
+  OutStreamer->SwitchSection(InstMap);
+  OutStreamer->EmitLabel(SledsStart);
   for (const auto &Sled : Sleds)
     Sled.emit(WordSizeBytes, OutStreamer.get(), CurrentFnSym);
+  MCSymbol *SledsEnd = OutContext.createTempSymbol("xray_synthetic_end", true);
+  OutStreamer->EmitLabel(SledsEnd);
 
+  // We then emit a single entry in the index per function. We use the symbols
+  // that bound the instrumentation map as the range for a specific function.
+  // Each entry here will be 2 * word size aligned, as we're writing down two
+  // pointers. This should work for both 32-bit and 64-bit platforms.
+  OutStreamer->SwitchSection(FnSledIndex);
+  OutStreamer->EmitCodeAlignment(2 * WordSizeBytes);
+  OutStreamer->EmitLabel(IdxRef);
+  OutStreamer->EmitSymbolValue(SledsStart, WordSizeBytes);
+  OutStreamer->EmitSymbolValue(SledsEnd, WordSizeBytes);
   OutStreamer->SwitchSection(PrevSection);
   Sleds.clear();
 }

lib/CodeGen/AsmPrinter/CodeViewDebug.cpp

@@ -469,7 +469,7 @@ void CodeViewDebug::emitTypeInformation() {
     CommentPrefix += ' ';
   }
 
-  TypeDatabase TypeDB;
+  TypeDatabase TypeDB(TypeTable.records().size());
   CVTypeDumper CVTD(TypeDB);
   TypeTable.ForEachRecord([&](TypeIndex Index, ArrayRef<uint8_t> Record) {
     if (OS.isVerboseAsm()) {
@@ -1705,10 +1705,12 @@ TypeIndex CodeViewDebug::lowerCompleteTypeClass(const DICompositeType *Ty) {
                  SizeInBytes, FullName, Ty->getIdentifier());
   TypeIndex ClassTI = TypeTable.writeKnownType(CR);
 
-  StringIdRecord SIDR(TypeIndex(0x0), getFullFilepath(Ty->getFile()));
-  TypeIndex SIDI = TypeTable.writeKnownType(SIDR);
-  UdtSourceLineRecord USLR(ClassTI, SIDI, Ty->getLine());
-  TypeTable.writeKnownType(USLR);
+  if (const auto *File = Ty->getFile()) {
+    StringIdRecord SIDR(TypeIndex(0x0), getFullFilepath(File));
+    TypeIndex SIDI = TypeTable.writeKnownType(SIDR);
+    UdtSourceLineRecord USLR(ClassTI, SIDI, Ty->getLine());
+    TypeTable.writeKnownType(USLR);
+  }
 
   addToUDTs(Ty, ClassTI);
 

lib/CodeGen/BranchFolding.cpp

@@ -1850,8 +1850,8 @@ bool BranchFolder::HoistCommonCodeInSuccs(MachineBasicBlock *MBB) {
     return false;
 
   bool HasDups = false;
-  SmallVector<unsigned, 4> LocalDefs;
-  SmallSet<unsigned, 4> LocalDefsSet;
+  SmallVector<unsigned, 4> LocalDefs, LocalKills;
+  SmallSet<unsigned, 4> ActiveDefsSet, AllDefsSet;
   MachineBasicBlock::iterator TIB = TBB->begin();
   MachineBasicBlock::iterator FIB = FBB->begin();
   MachineBasicBlock::iterator TIE = TBB->end();
@@ -1905,7 +1905,7 @@ bool BranchFolder::HoistCommonCodeInSuccs(MachineBasicBlock *MBB) {
           IsSafe = false;
           break;
         }
-      } else if (!LocalDefsSet.count(Reg)) {
+      } else if (!ActiveDefsSet.count(Reg)) {
         if (Defs.count(Reg)) {
           // Use is defined by the instruction at the point of insertion.
           IsSafe = false;
@@ -1925,18 +1925,22 @@ bool BranchFolder::HoistCommonCodeInSuccs(MachineBasicBlock *MBB) {
     if (!TIB->isSafeToMove(nullptr, DontMoveAcrossStore))
       break;
 
-    // Remove kills from LocalDefsSet, these registers had short live ranges.
+    // Remove kills from ActiveDefsSet, these registers had short live ranges.
     for (const MachineOperand &MO : TIB->operands()) {
       if (!MO.isReg() || !MO.isUse() || !MO.isKill())
         continue;
       unsigned Reg = MO.getReg();
-      if (!Reg || !LocalDefsSet.count(Reg))
+      if (!Reg)
         continue;
+      if (!AllDefsSet.count(Reg)) {
+        LocalKills.push_back(Reg);
+        continue;
+      }
       if (TargetRegisterInfo::isPhysicalRegister(Reg)) {
         for (MCRegAliasIterator AI(Reg, TRI, true); AI.isValid(); ++AI)
-          LocalDefsSet.erase(*AI);
+          ActiveDefsSet.erase(*AI);
       } else {
-        LocalDefsSet.erase(Reg);
+        ActiveDefsSet.erase(Reg);
       }
     }
 
@@ -1948,7 +1952,8 @@ bool BranchFolder::HoistCommonCodeInSuccs(MachineBasicBlock *MBB) {
       if (!Reg || TargetRegisterInfo::isVirtualRegister(Reg))
         continue;
       LocalDefs.push_back(Reg);
-      addRegAndItsAliases(Reg, TRI, LocalDefsSet);
+      addRegAndItsAliases(Reg, TRI, ActiveDefsSet);
+      addRegAndItsAliases(Reg, TRI, AllDefsSet);
     }
 
     HasDups = true;
@@ -1963,17 +1968,22 @@ bool BranchFolder::HoistCommonCodeInSuccs(MachineBasicBlock *MBB) {
   FBB->erase(FBB->begin(), FIB);
 
   // Update livein's.
-  bool AddedLiveIns = false;
+  bool ChangedLiveIns = false;
   for (unsigned i = 0, e = LocalDefs.size(); i != e; ++i) {
     unsigned Def = LocalDefs[i];
-    if (LocalDefsSet.count(Def)) {
+    if (ActiveDefsSet.count(Def)) {
       TBB->addLiveIn(Def);
       FBB->addLiveIn(Def);
-      AddedLiveIns = true;
+      ChangedLiveIns = true;
     }
   }
+  for (unsigned K : LocalKills) {
+    TBB->removeLiveIn(K);
+    FBB->removeLiveIn(K);
+    ChangedLiveIns = true;
+  }
 
-  if (AddedLiveIns) {
+  if (ChangedLiveIns) {
     TBB->sortUniqueLiveIns();
     FBB->sortUniqueLiveIns();
   }

lib/CodeGen/GlobalISel/IRTranslator.cpp

@@ -1108,6 +1108,14 @@ bool IRTranslator::translate(const Constant &C, unsigned Reg) {
     default:
       return false;
     }
+  } else if (auto CV = dyn_cast<ConstantVector>(&C)) {
+    if (CV->getNumOperands() == 1)
+      return translate(*CV->getOperand(0), Reg);
+    SmallVector<unsigned, 4> Ops;
+    for (unsigned i = 0; i < CV->getNumOperands(); ++i) {
+      Ops.push_back(getOrCreateVReg(*CV->getOperand(i)));
+    }
+    EntryBuilder.buildMerge(Reg, Ops);
   } else
     return false;
 
@@ -1199,9 +1207,6 @@ bool IRTranslator::runOnMachineFunction(MachineFunction &CurMF) {
 
   finishPendingPhis();
 
-  auto &TLI = *MF->getSubtarget().getTargetLowering();
-  TLI.finalizeLowering(*MF);
-
   // Merge the argument lowering and constants block with its single
   // successor, the LLVM-IR entry block.  We want the basic block to
   // be maximal.

lib/CodeGen/GlobalISel/InstructionSelect.cpp

@@ -24,6 +24,7 @@
 #include "llvm/IR/Function.h"
 #include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
+#include "llvm/Target/TargetLowering.h"
 #include "llvm/Target/TargetSubtargetInfo.h"
 
 #define DEBUG_TYPE "instruction-select"
@@ -70,8 +71,7 @@ bool InstructionSelect::runOnMachineFunction(MachineFunction &MF) {
   // An optimization remark emitter. Used to report failures.
   MachineOptimizationRemarkEmitter MORE(MF, /*MBFI=*/nullptr);
 
-  // FIXME: freezeReservedRegs is now done in IRTranslator, but there are many
-  // other MF/MFI fields we need to initialize.
+  // FIXME: There are many other MF/MFI fields we need to initialize.
 
 #ifndef NDEBUG
   // Check that our input is fully legal: we require the function to have the
@@ -184,6 +184,9 @@ bool InstructionSelect::runOnMachineFunction(MachineFunction &MF) {
     return false;
   }
 
+  auto &TLI = *MF.getSubtarget().getTargetLowering();
+  TLI.finalizeLowering(MF);
+
   // FIXME: Should we accurately track changes?
   return true;
 }

lib/CodeGen/GlobalISel/Legalizer.cpp

@@ -176,8 +176,13 @@ bool Legalizer::runOnMachineFunction(MachineFunction &MF) {
       unsigned NumNewInsns = 0;
       SmallVector<MachineInstr *, 4> WorkList;
       Helper.MIRBuilder.recordInsertions([&](MachineInstr *MI) {
-        ++NumNewInsns;
-        WorkList.push_back(MI);
+        // Only legalize pre-isel generic instructions.
+        // Legalization process could generate Target specific pseudo
+        // instructions with generic types. Don't record them
+        if (isPreISelGenericOpcode(MI->getOpcode())) {
+          ++NumNewInsns;
+          WorkList.push_back(MI);
+        }
       });
       WorkList.push_back(&*MI);
 

lib/CodeGen/GlobalISel/RegBankSelect.cpp

@@ -213,21 +213,23 @@ uint64_t RegBankSelect::getRepairCost(
   return UINT_MAX;
 }
 
-RegisterBankInfo::InstructionMapping &RegBankSelect::findBestMapping(
+const RegisterBankInfo::InstructionMapping &RegBankSelect::findBestMapping(
     MachineInstr &MI, RegisterBankInfo::InstructionMappings &PossibleMappings,
     SmallVectorImpl<RepairingPlacement> &RepairPts) {
   assert(!PossibleMappings.empty() &&
          "Do not know how to map this instruction");
 
-  RegisterBankInfo::InstructionMapping *BestMapping = nullptr;
+  const RegisterBankInfo::InstructionMapping *BestMapping = nullptr;
   MappingCost Cost = MappingCost::ImpossibleCost();
   SmallVector<RepairingPlacement, 4> LocalRepairPts;
-  for (RegisterBankInfo::InstructionMapping &CurMapping : PossibleMappings) {
-    MappingCost CurCost = computeMapping(MI, CurMapping, LocalRepairPts, &Cost);
+  for (const RegisterBankInfo::InstructionMapping *CurMapping :
+       PossibleMappings) {
+    MappingCost CurCost =
+        computeMapping(MI, *CurMapping, LocalRepairPts, &Cost);
     if (CurCost < Cost) {
       DEBUG(dbgs() << "New best: " << CurCost << '\n');
       Cost = CurCost;
-      BestMapping = &CurMapping;
+      BestMapping = CurMapping;
       RepairPts.clear();
       for (RepairingPlacement &RepairPt : LocalRepairPts)
         RepairPts.emplace_back(std::move(RepairPt));
@@ -237,7 +239,7 @@ RegisterBankInfo::InstructionMapping &RegBankSelect::findBestMapping(
     // If none of the mapping worked that means they are all impossible.
     // Thus, pick the first one and set an impossible repairing point.
     // It will trigger the failed isel mode.
-    BestMapping = &(*PossibleMappings.begin());
+    BestMapping = *PossibleMappings.begin();
     RepairPts.emplace_back(
         RepairingPlacement(MI, 0, *TRI, *this, RepairingPlacement::Impossible));
   } else
@@ -543,10 +545,10 @@ bool RegBankSelect::assignInstr(MachineInstr &MI) {
   // Remember the repairing placement for all the operands.
   SmallVector<RepairingPlacement, 4> RepairPts;
 
-  RegisterBankInfo::InstructionMapping BestMapping;
+  const RegisterBankInfo::InstructionMapping *BestMapping;
   if (OptMode == RegBankSelect::Mode::Fast) {
-    BestMapping = RBI->getInstrMapping(MI);
-    MappingCost DefaultCost = computeMapping(MI, BestMapping, RepairPts);
+    BestMapping = &RBI->getInstrMapping(MI);
+    MappingCost DefaultCost = computeMapping(MI, *BestMapping, RepairPts);
     (void)DefaultCost;
     if (DefaultCost == MappingCost::ImpossibleCost())
       return false;
@@ -555,16 +557,16 @@ bool RegBankSelect::assignInstr(MachineInstr &MI) {
         RBI->getInstrPossibleMappings(MI);
     if (PossibleMappings.empty())
       return false;
-    BestMapping = std::move(findBestMapping(MI, PossibleMappings, RepairPts));
+    BestMapping = &findBestMapping(MI, PossibleMappings, RepairPts);
   }
   // Make sure the mapping is valid for MI.
-  assert(BestMapping.verify(MI) && "Invalid instruction mapping");
+  assert(BestMapping->verify(MI) && "Invalid instruction mapping");
 
-  DEBUG(dbgs() << "Best Mapping: " << BestMapping << '\n');
+  DEBUG(dbgs() << "Best Mapping: " << *BestMapping << '\n');
 
   // After this call, MI may not be valid anymore.
   // Do not use it.
-  return applyMapping(MI, BestMapping, RepairPts);
+  return applyMapping(MI, *BestMapping, RepairPts);
 }
 
 bool RegBankSelect::runOnMachineFunction(MachineFunction &MF) {

lib/CodeGen/GlobalISel/RegisterBankInfo.cpp

@@ -45,6 +45,10 @@ STATISTIC(NumOperandsMappingsCreated,
           "Number of operands mappings dynamically created");
 STATISTIC(NumOperandsMappingsAccessed,
           "Number of operands mappings dynamically accessed");
+STATISTIC(NumInstructionMappingsCreated,
+          "Number of instruction mappings dynamically created");
+STATISTIC(NumInstructionMappingsAccessed,
+          "Number of instruction mappings dynamically accessed");
 
 const unsigned RegisterBankInfo::DefaultMappingID = UINT_MAX;
 const unsigned RegisterBankInfo::InvalidMappingID = UINT_MAX - 1;
@@ -137,7 +141,7 @@ static bool isCopyLike(const MachineInstr &MI) {
          MI.getOpcode() == TargetOpcode::REG_SEQUENCE;
 }
 
-RegisterBankInfo::InstructionMapping
+const RegisterBankInfo::InstructionMapping &
 RegisterBankInfo::getInstrMappingImpl(const MachineInstr &MI) const {
   // For copies we want to walk over the operands and try to find one
   // that has a register bank since the instruction itself will not get
@@ -147,9 +151,6 @@ RegisterBankInfo::getInstrMappingImpl(const MachineInstr &MI) const {
   // is important. The rest is not constrained.
   unsigned NumOperandsForMapping = IsCopyLike ? 1 : MI.getNumOperands();
 
-  RegisterBankInfo::InstructionMapping Mapping(DefaultMappingID, /*Cost*/ 1,
-                                               /*OperandsMapping*/ nullptr,
-                                               NumOperandsForMapping);
   const MachineFunction &MF = *MI.getParent()->getParent();
   const TargetSubtargetInfo &STI = MF.getSubtarget();
   const TargetRegisterInfo &TRI = *STI.getRegisterInfo();
@@ -190,7 +191,7 @@ RegisterBankInfo::getInstrMappingImpl(const MachineInstr &MI) const {
 
         if (!IsCopyLike)
           // MI does not carry enough information to guess the mapping.
-          return InstructionMapping();
+          return getInvalidInstructionMapping();
         continue;
       }
     }
@@ -206,11 +207,13 @@ RegisterBankInfo::getInstrMappingImpl(const MachineInstr &MI) const {
 
   if (IsCopyLike && !CompleteMapping)
     // No way to deduce the type from what we have.
-    return InstructionMapping();
+    return getInvalidInstructionMapping();
 
   assert(CompleteMapping && "Setting an uncomplete mapping");
-  Mapping.setOperandsMapping(getOperandsMapping(OperandsMapping));
-  return Mapping;
+  return getInstructionMapping(
+      DefaultMappingID, /*Cost*/ 1,
+      /*OperandsMapping*/ getOperandsMapping(OperandsMapping),
+      NumOperandsForMapping);
 }
 
 /// Hashing function for PartialMapping.
@@ -320,9 +323,44 @@ const RegisterBankInfo::ValueMapping *RegisterBankInfo::getOperandsMapping(
   return getOperandsMapping(OpdsMapping.begin(), OpdsMapping.end());
 }
 
-RegisterBankInfo::InstructionMapping
+static hash_code
+hashInstructionMapping(unsigned ID, unsigned Cost,
+                       const RegisterBankInfo::ValueMapping *OperandsMapping,
+                       unsigned NumOperands) {
+  return hash_combine(ID, Cost, OperandsMapping, NumOperands);
+}
+
+const RegisterBankInfo::InstructionMapping &
+RegisterBankInfo::getInstructionMappingImpl(
+    bool IsInvalid, unsigned ID, unsigned Cost,
+    const RegisterBankInfo::ValueMapping *OperandsMapping,
+    unsigned NumOperands) const {
+  assert(((IsInvalid && ID == InvalidMappingID && Cost == 0 &&
+           OperandsMapping == nullptr && NumOperands == 0) ||
+          !IsInvalid) &&
+         "Mismatch argument for invalid input");
+  ++NumInstructionMappingsAccessed;
+
+  hash_code Hash =
+      hashInstructionMapping(ID, Cost, OperandsMapping, NumOperands);
+  const auto &It = MapOfInstructionMappings.find(Hash);
+  if (It != MapOfInstructionMappings.end())
+    return *It->second;
+
+  ++NumInstructionMappingsCreated;
+
+  auto &InstrMapping = MapOfInstructionMappings[Hash];
+  if (IsInvalid)
+    InstrMapping = llvm::make_unique<InstructionMapping>();
+  else
+    InstrMapping = llvm::make_unique<InstructionMapping>(
+        ID, Cost, OperandsMapping, NumOperands);
+  return *InstrMapping;
+}
+
+const RegisterBankInfo::InstructionMapping &
 RegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
-  RegisterBankInfo::InstructionMapping Mapping = getInstrMappingImpl(MI);
+  const RegisterBankInfo::InstructionMapping &Mapping = getInstrMappingImpl(MI);
   if (Mapping.isValid())
     return Mapping;
   llvm_unreachable("The target must implement this");
@@ -332,14 +370,14 @@ RegisterBankInfo::InstructionMappings
 RegisterBankInfo::getInstrPossibleMappings(const MachineInstr &MI) const {
   InstructionMappings PossibleMappings;
   // Put the default mapping first.
-  PossibleMappings.push_back(getInstrMapping(MI));
+  PossibleMappings.push_back(&getInstrMapping(MI));
   // Then the alternative mapping, if any.
   InstructionMappings AltMappings = getInstrAlternativeMappings(MI);
-  for (InstructionMapping &AltMapping : AltMappings)
-    PossibleMappings.emplace_back(std::move(AltMapping));
+  for (const InstructionMapping *AltMapping : AltMappings)
+    PossibleMappings.push_back(AltMapping);
 #ifndef NDEBUG
-  for (const InstructionMapping &Mapping : PossibleMappings)
-    assert(Mapping.verify(MI) && "Mapping is invalid");
+  for (const InstructionMapping *Mapping : PossibleMappings)
+    assert(Mapping->verify(MI) && "Mapping is invalid");
 #endif
   return PossibleMappings;
 }

lib/CodeGen/MIRParser/MIParser.cpp

@@ -12,11 +12,13 @@
 //===----------------------------------------------------------------------===//
 
 #include "MIParser.h"
+
 #include "MILexer.h"
 #include "llvm/ADT/StringMap.h"
 #include "llvm/ADT/StringSwitch.h"
 #include "llvm/AsmParser/Parser.h"
 #include "llvm/AsmParser/SlotMapping.h"
+#include "llvm/CodeGen/MIRPrinter.h"
 #include "llvm/CodeGen/MachineBasicBlock.h"
 #include "llvm/CodeGen/MachineFrameInfo.h"
 #include "llvm/CodeGen/MachineFunction.h"
@@ -134,7 +136,8 @@ public:
 
   bool
   parseBasicBlockDefinition(DenseMap<unsigned, MachineBasicBlock *> &MBBSlots);
-  bool parseBasicBlock(MachineBasicBlock &MBB);
+  bool parseBasicBlock(MachineBasicBlock &MBB,
+                       MachineBasicBlock *&AddFalthroughFrom);
   bool parseBasicBlockLiveins(MachineBasicBlock &MBB);
   bool parseBasicBlockSuccessors(MachineBasicBlock &MBB);
 
@@ -518,7 +521,8 @@ bool MIParser::parseBasicBlockSuccessors(MachineBasicBlock &MBB) {
   return false;
 }
 
-bool MIParser::parseBasicBlock(MachineBasicBlock &MBB) {
+bool MIParser::parseBasicBlock(MachineBasicBlock &MBB,
+                               MachineBasicBlock *&AddFalthroughFrom) {
   // Skip the definition.
   assert(Token.is(MIToken::MachineBasicBlockLabel));
   lex();
@@ -538,10 +542,12 @@ bool MIParser::parseBasicBlock(MachineBasicBlock &MBB) {
   //
   // is equivalent to
   //   liveins: %edi, %esi
+  bool ExplicitSuccesors = false;
   while (true) {
     if (Token.is(MIToken::kw_successors)) {
       if (parseBasicBlockSuccessors(MBB))
         return true;
+      ExplicitSuccesors = true;
     } else if (Token.is(MIToken::kw_liveins)) {
       if (parseBasicBlockLiveins(MBB))
         return true;
@@ -557,10 +563,9 @@ bool MIParser::parseBasicBlock(MachineBasicBlock &MBB) {
   // Parse the instructions.
   bool IsInBundle = false;
   MachineInstr *PrevMI = nullptr;
-  while (true) {
-    if (Token.is(MIToken::MachineBasicBlockLabel) || Token.is(MIToken::Eof))
-      return false;
-    else if (consumeIfPresent(MIToken::Newline))
+  while (!Token.is(MIToken::MachineBasicBlockLabel) &&
+         !Token.is(MIToken::Eof)) {
+    if (consumeIfPresent(MIToken::Newline))
       continue;
     if (consumeIfPresent(MIToken::rbrace)) {
       // The first parsing pass should verify that all closing '}' have an
@@ -592,6 +597,22 @@ bool MIParser::parseBasicBlock(MachineBasicBlock &MBB) {
     assert(Token.isNewlineOrEOF() && "MI is not fully parsed");
     lex();
   }
+
+  // Construct successor list by searching for basic block machine operands.
+  if (!ExplicitSuccesors) {
+    SmallVector<MachineBasicBlock*,4> Successors;
+    bool IsFallthrough;
+    guessSuccessors(MBB, Successors, IsFallthrough);
+    for (MachineBasicBlock *Succ : Successors)
+      MBB.addSuccessor(Succ);
+
+    if (IsFallthrough) {
+      AddFalthroughFrom = &MBB;
+    } else {
+      MBB.normalizeSuccProbs();
+    }
+  }
+
   return false;
 }
 
@@ -605,11 +626,18 @@ bool MIParser::parseBasicBlocks() {
   // The first parsing pass should have verified that this token is a MBB label
   // in the 'parseBasicBlockDefinitions' method.
   assert(Token.is(MIToken::MachineBasicBlockLabel));
+  MachineBasicBlock *AddFalthroughFrom = nullptr;
   do {
     MachineBasicBlock *MBB = nullptr;
     if (parseMBBReference(MBB))
       return true;
-    if (parseBasicBlock(*MBB))
+    if (AddFalthroughFrom) {
+      if (!AddFalthroughFrom->isSuccessor(MBB))
+        AddFalthroughFrom->addSuccessor(MBB);
+      AddFalthroughFrom->normalizeSuccProbs();
+      AddFalthroughFrom = nullptr;
+    }
+    if (parseBasicBlock(*MBB, AddFalthroughFrom))
       return true;
     // The method 'parseBasicBlock' should parse the whole block until the next
     // block or the end of file.

lib/CodeGen/MIRPrinter.cpp

@@ -12,7 +12,8 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "MIRPrinter.h"
+#include "llvm/CodeGen/MIRPrinter.h"
+
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SmallBitVector.h"
 #include "llvm/CodeGen/GlobalISel/RegisterBank.h"
@@ -34,6 +35,7 @@
 #include "llvm/MC/MCSymbol.h"
 #include "llvm/Support/Format.h"
 #include "llvm/Support/MemoryBuffer.h"
+#include "llvm/Support/Options.h"
 #include "llvm/Support/YAMLTraits.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetInstrInfo.h"
@@ -42,6 +44,9 @@
 
 using namespace llvm;
 
+static cl::opt<bool> SimplifyMIR("simplify-mir",
+    cl::desc("Leave out unnecessary information when printing MIR"));
+
 namespace {
 
 /// This structure describes how to print out stack object references.
@@ -105,6 +110,9 @@ class MIPrinter {
   const DenseMap<const uint32_t *, unsigned> &RegisterMaskIds;
   const DenseMap<int, FrameIndexOperand> &StackObjectOperandMapping;
 
+  bool canPredictBranchProbabilities(const MachineBasicBlock &MBB) const;
+  bool canPredictSuccessors(const MachineBasicBlock &MBB) const;
+
 public:
   MIPrinter(raw_ostream &OS, ModuleSlotTracker &MST,
             const DenseMap<const uint32_t *, unsigned> &RegisterMaskIds,
@@ -454,6 +462,63 @@ void MIRPrinter::initRegisterMaskIds(const MachineFunction &MF) {
     RegisterMaskIds.insert(std::make_pair(Mask, I++));
 }
 
+void llvm::guessSuccessors(const MachineBasicBlock &MBB,
+                           SmallVectorImpl<MachineBasicBlock*> &Result,
+                           bool &IsFallthrough) {
+  SmallPtrSet<MachineBasicBlock*,8> Seen;
+
+  for (const MachineInstr &MI : MBB) {
+    if (MI.isPHI())
+      continue;
+    for (const MachineOperand &MO : MI.operands()) {
+      if (!MO.isMBB())
+        continue;
+      MachineBasicBlock *Succ = MO.getMBB();
+      auto RP = Seen.insert(Succ);
+      if (RP.second)
+        Result.push_back(Succ);
+    }
+  }
+  MachineBasicBlock::const_iterator I = MBB.getLastNonDebugInstr();
+  IsFallthrough = I == MBB.end() || !I->isBarrier();
+}
+
+bool
+MIPrinter::canPredictBranchProbabilities(const MachineBasicBlock &MBB) const {
+  if (MBB.succ_size() <= 1)
+    return true;
+  if (!MBB.hasSuccessorProbabilities())
+    return true;
+
+  SmallVector<BranchProbability,8> Normalized(MBB.Probs.begin(),
+                                              MBB.Probs.end());
+  BranchProbability::normalizeProbabilities(Normalized.begin(),
+                                            Normalized.end());
+  SmallVector<BranchProbability,8> Equal(Normalized.size());
+  BranchProbability::normalizeProbabilities(Equal.begin(), Equal.end());
+
+  return std::equal(Normalized.begin(), Normalized.end(), Equal.begin());
+}
+
+bool MIPrinter::canPredictSuccessors(const MachineBasicBlock &MBB) const {
+  SmallVector<MachineBasicBlock*,8> GuessedSuccs;
+  bool GuessedFallthrough;
+  guessSuccessors(MBB, GuessedSuccs, GuessedFallthrough);
+  if (GuessedFallthrough) {
+    const MachineFunction &MF = *MBB.getParent();
+    MachineFunction::const_iterator NextI = std::next(MBB.getIterator());
+    if (NextI != MF.end()) {
+      MachineBasicBlock *Next = const_cast<MachineBasicBlock*>(&*NextI);
+      if (!is_contained(GuessedSuccs, Next))
+        GuessedSuccs.push_back(Next);
+    }
+  }
+  if (GuessedSuccs.size() != MBB.succ_size())
+    return false;
+  return std::equal(MBB.succ_begin(), MBB.succ_end(), GuessedSuccs.begin());
+}
+
+
 void MIPrinter::print(const MachineBasicBlock &MBB) {
   assert(MBB.getNumber() >= 0 && "Invalid MBB number");
   OS << "bb." << MBB.getNumber();
@@ -492,13 +557,15 @@ void MIPrinter::print(const MachineBasicBlock &MBB) {
 
   bool HasLineAttributes = false;
   // Print the successors
-  if (!MBB.succ_empty()) {
+  bool canPredictProbs = canPredictBranchProbabilities(MBB);
+  if (!MBB.succ_empty() && (!SimplifyMIR || !canPredictProbs ||
+                            !canPredictSuccessors(MBB))) {
     OS.indent(2) << "successors: ";
     for (auto I = MBB.succ_begin(), E = MBB.succ_end(); I != E; ++I) {
       if (I != MBB.succ_begin())
         OS << ", ";
       printMBBReference(**I);
-      if (MBB.hasSuccessorProbabilities())
+      if (!SimplifyMIR || !canPredictProbs)
         OS << '('
            << format("0x%08" PRIx32, MBB.getSuccProbability(I).getNumerator())
            << ')';

lib/CodeGen/MIRPrintingPass.cpp

@@ -12,7 +12,8 @@
 //
 //===----------------------------------------------------------------------===//
 
-#include "MIRPrinter.h"
+#include "llvm/CodeGen/MIRPrinter.h"
+
 #include "llvm/CodeGen/Passes.h"
 #include "llvm/CodeGen/MachineFunctionPass.h"
 #include "llvm/CodeGen/MIRYamlMapping.h"

lib/CodeGen/MachineFrameInfo.cpp

@@ -19,6 +19,7 @@
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Target/TargetFrameLowering.h"
+#include "llvm/Target/TargetInstrInfo.h"
 #include "llvm/Target/TargetRegisterInfo.h"
 #include "llvm/Target/TargetSubtargetInfo.h"
 #include <cassert>
@@ -175,6 +176,31 @@ unsigned MachineFrameInfo::estimateStackSize(const MachineFunction &MF) const {
   return (unsigned)Offset;
 }
 
+void MachineFrameInfo::computeMaxCallFrameSize(const MachineFunction &MF) {
+  const TargetInstrInfo &TII = *MF.getSubtarget().getInstrInfo();
+  unsigned FrameSetupOpcode = TII.getCallFrameSetupOpcode();
+  unsigned FrameDestroyOpcode = TII.getCallFrameDestroyOpcode();
+  assert(FrameSetupOpcode != ~0u && FrameDestroyOpcode != ~0u &&
+         "Can only compute MaxCallFrameSize if Setup/Destroy opcode are known");
+
+  MaxCallFrameSize = 0;
+  for (const MachineBasicBlock &MBB : MF) {
+    for (const MachineInstr &MI : MBB) {
+      unsigned Opcode = MI.getOpcode();
+      if (Opcode == FrameSetupOpcode || Opcode == FrameDestroyOpcode) {
+        unsigned Size = TII.getFrameSize(MI);
+        MaxCallFrameSize = std::max(MaxCallFrameSize, Size);
+        AdjustsStack = true;
+      } else if (MI.isInlineAsm()) {
+        // Some inline asm's need a stack frame, as indicated by operand 1.
+        unsigned ExtraInfo = MI.getOperand(InlineAsm::MIOp_ExtraInfo).getImm();
+        if (ExtraInfo & InlineAsm::Extra_IsAlignStack)
+          AdjustsStack = true;
+      }
+    }
+  }
+}
+
 void MachineFrameInfo::print(const MachineFunction &MF, raw_ostream &OS) const{
   if (Objects.empty()) return;
 

lib/CodeGen/MachineVerifier.cpp

@@ -188,8 +188,9 @@ namespace {
       return Reg < regsReserved.size() && regsReserved.test(Reg);
     }
 
-    bool isAllocatable(unsigned Reg) {
-      return Reg < TRI->getNumRegs() && MRI->isAllocatable(Reg);
+    bool isAllocatable(unsigned Reg) const {
+      return Reg < TRI->getNumRegs() && TRI->isInAllocatableClass(Reg) &&
+        !regsReserved.test(Reg);
     }
 
     // Analysis information if available
@@ -526,7 +527,8 @@ void MachineVerifier::markReachable(const MachineBasicBlock *MBB) {
 
 void MachineVerifier::visitMachineFunctionBefore() {
   lastIndex = SlotIndex();
-  regsReserved = MRI->getReservedRegs();
+  regsReserved = MRI->reservedRegsFrozen() ? MRI->getReservedRegs()
+                                           : TRI->getReservedRegs(*MF);
 
   if (!MF->empty())
     markReachable(&MF->front());

lib/CodeGen/PrologEpilogInserter.cpp

@@ -277,6 +277,9 @@ void PEI::calculateCallFrameInfo(MachineFunction &Fn) {
           AdjustsStack = true;
       }
 
+  assert(!MFI.isMaxCallFrameSizeComputed() ||
+         (MFI.getMaxCallFrameSize() == MaxCallFrameSize &&
+          MFI.adjustsStack() == AdjustsStack));
   MFI.setAdjustsStack(AdjustsStack);
   MFI.setMaxCallFrameSize(MaxCallFrameSize);
 

lib/CodeGen/SelectionDAG/DAGCombiner.cpp

@@ -6688,6 +6688,9 @@ SDValue DAGCombiner::visitVSELECT(SDNode *N) {
 
     if (isAbs) {
       EVT VT = LHS.getValueType();
+      if (TLI.isOperationLegalOrCustom(ISD::ABS, VT))
+        return DAG.getNode(ISD::ABS, DL, VT, LHS);
+
       SDValue Shift = DAG.getNode(
           ISD::SRA, DL, VT, LHS,
           DAG.getConstant(VT.getScalarSizeInBits() - 1, DL, VT));
@@ -9469,6 +9472,14 @@ SDValue DAGCombiner::visitFMULForFMADistributiveCombine(SDNode *N) {
   return SDValue();
 }
 
+static bool isFMulNegTwo(SDValue &N) {
+  if (N.getOpcode() != ISD::FMUL)
+    return false;
+  if (ConstantFPSDNode *CFP = isConstOrConstSplatFP(N.getOperand(1)))
+    return CFP->isExactlyValue(-2.0);
+  return false;
+}
+
 SDValue DAGCombiner::visitFADD(SDNode *N) {
   SDValue N0 = N->getOperand(0);
   SDValue N1 = N->getOperand(1);
@@ -9507,6 +9518,16 @@ SDValue DAGCombiner::visitFADD(SDNode *N) {
     return DAG.getNode(ISD::FSUB, DL, VT, N1,
                        GetNegatedExpression(N0, DAG, LegalOperations), Flags);
 
+  // fold (fadd A, (fmul B, -2.0)) -> (fsub A, (fadd B, B))
+  // fold (fadd (fmul B, -2.0), A) -> (fsub A, (fadd B, B))
+  if ((isFMulNegTwo(N0) && N0.hasOneUse()) ||
+      (isFMulNegTwo(N1) && N1.hasOneUse())) {
+    bool N1IsFMul = isFMulNegTwo(N1);
+    SDValue AddOp = N1IsFMul ? N1.getOperand(0) : N0.getOperand(0);
+    SDValue Add = DAG.getNode(ISD::FADD, DL, VT, AddOp, AddOp, Flags);
+    return DAG.getNode(ISD::FSUB, DL, VT, N1IsFMul ? N0 : N1, Add, Flags);
+  }
+
   // FIXME: Auto-upgrade the target/function-level option.
   if (Options.NoSignedZerosFPMath || N->getFlags().hasNoSignedZeros()) {
     // fold (fadd A, 0) -> A

lib/CodeGen/SelectionDAG/FastISel.cpp

@@ -861,6 +861,25 @@ bool FastISel::selectPatchpoint(const CallInst *I) {
   return true;
 }
 
+bool FastISel::selectXRayCustomEvent(const CallInst *I) {
+  const auto &Triple = TM.getTargetTriple();
+  if (Triple.getArch() != Triple::x86_64 || !Triple.isOSLinux())
+    return true; // don't do anything to this instruction.
+  SmallVector<MachineOperand, 8> Ops;
+  Ops.push_back(MachineOperand::CreateReg(getRegForValue(I->getArgOperand(0)),
+                                          /*IsDef=*/false));
+  Ops.push_back(MachineOperand::CreateReg(getRegForValue(I->getArgOperand(1)),
+                                          /*IsDef=*/false));
+  MachineInstrBuilder MIB =
+      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DbgLoc,
+              TII.get(TargetOpcode::PATCHABLE_EVENT_CALL));
+  for (auto &MO : Ops)
+    MIB.add(MO);
+  // Insert the Patchable Event Call instruction, that gets lowered properly.
+  return true;
+}
+
+
 /// Returns an AttributeList representing the attributes applied to the return
 /// value of the given call.
 static AttributeList getReturnAttrs(FastISel::CallLoweringInfo &CLI) {
@@ -1252,6 +1271,9 @@ bool FastISel::selectIntrinsicCall(const IntrinsicInst *II) {
   case Intrinsic::experimental_patchpoint_void:
   case Intrinsic::experimental_patchpoint_i64:
     return selectPatchpoint(II);
+
+  case Intrinsic::xray_customevent:
+    return selectXRayCustomEvent(II);
   }
 
   return fastLowerIntrinsicCall(II);

lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp

@@ -402,8 +402,7 @@ FunctionLoweringInfo::GetLiveOutRegInfo(unsigned Reg, unsigned BitWidth) {
 
   if (BitWidth > LOI->Known.getBitWidth()) {
     LOI->NumSignBits = 1;
-    LOI->Known.Zero = LOI->Known.Zero.zextOrTrunc(BitWidth);
-    LOI->Known.One = LOI->Known.One.zextOrTrunc(BitWidth);
+    LOI->Known = LOI->Known.zextOrTrunc(BitWidth);
   }
 
   return LOI;

lib/CodeGen/SelectionDAG/ResourcePriorityQueue.cpp

@@ -67,12 +67,11 @@ ResourcePriorityQueue::ResourcePriorityQueue(SelectionDAGISel *IS)
 unsigned
 ResourcePriorityQueue::numberRCValPredInSU(SUnit *SU, unsigned RCId) {
   unsigned NumberDeps = 0;
-  for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-       I != E; ++I) {
-    if (I->isCtrl())
+  for (SDep &Pred : SU->Preds) {
+    if (Pred.isCtrl())
       continue;
 
-    SUnit *PredSU = I->getSUnit();
+    SUnit *PredSU = Pred.getSUnit();
     const SDNode *ScegN = PredSU->getNode();
 
     if (!ScegN)
@@ -105,12 +104,11 @@ ResourcePriorityQueue::numberRCValPredInSU(SUnit *SU, unsigned RCId) {
 unsigned ResourcePriorityQueue::numberRCValSuccInSU(SUnit *SU,
                                                     unsigned RCId) {
   unsigned NumberDeps = 0;
-  for (SUnit::const_succ_iterator I = SU->Succs.begin(), E = SU->Succs.end();
-       I != E; ++I) {
-    if (I->isCtrl())
+  for (const SDep &Succ : SU->Succs) {
+    if (Succ.isCtrl())
       continue;
 
-    SUnit *SuccSU = I->getSUnit();
+    SUnit *SuccSU = Succ.getSUnit();
     const SDNode *ScegN = SuccSU->getNode();
     if (!ScegN)
       continue;
@@ -142,9 +140,8 @@ unsigned ResourcePriorityQueue::numberRCValSuccInSU(SUnit *SU,
 
 static unsigned numberCtrlDepsInSU(SUnit *SU) {
   unsigned NumberDeps = 0;
-  for (SUnit::const_succ_iterator I = SU->Succs.begin(), E = SU->Succs.end();
-       I != E; ++I)
-    if (I->isCtrl())
+  for (const SDep &Succ : SU->Succs)
+    if (Succ.isCtrl())
       NumberDeps++;
 
   return NumberDeps;
@@ -152,9 +149,8 @@ static unsigned numberCtrlDepsInSU(SUnit *SU) {
 
 static unsigned numberCtrlPredInSU(SUnit *SU) {
   unsigned NumberDeps = 0;
-  for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-       I != E; ++I)
-    if (I->isCtrl())
+  for (SDep &Pred : SU->Preds)
+    if (Pred.isCtrl())
       NumberDeps++;
 
   return NumberDeps;
@@ -212,15 +208,14 @@ bool resource_sort::operator()(const SUnit *LHS, const SUnit *RHS) const {
 /// of SU, return it, otherwise return null.
 SUnit *ResourcePriorityQueue::getSingleUnscheduledPred(SUnit *SU) {
   SUnit *OnlyAvailablePred = nullptr;
-  for (SUnit::const_pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-       I != E; ++I) {
-    SUnit &Pred = *I->getSUnit();
-    if (!Pred.isScheduled) {
+  for (const SDep &Pred : SU->Preds) {
+    SUnit &PredSU = *Pred.getSUnit();
+    if (!PredSU.isScheduled) {
       // We found an available, but not scheduled, predecessor.  If it's the
       // only one we have found, keep track of it... otherwise give up.
-      if (OnlyAvailablePred && OnlyAvailablePred != &Pred)
+      if (OnlyAvailablePred && OnlyAvailablePred != &PredSU)
         return nullptr;
-      OnlyAvailablePred = &Pred;
+      OnlyAvailablePred = &PredSU;
     }
   }
   return OnlyAvailablePred;
@@ -230,9 +225,8 @@ void ResourcePriorityQueue::push(SUnit *SU) {
   // Look at all of the successors of this node.  Count the number of nodes that
   // this node is the sole unscheduled node for.
   unsigned NumNodesBlocking = 0;
-  for (SUnit::const_succ_iterator I = SU->Succs.begin(), E = SU->Succs.end();
-       I != E; ++I)
-    if (getSingleUnscheduledPred(I->getSUnit()) == SU)
+  for (const SDep &Succ : SU->Succs)
+    if (getSingleUnscheduledPred(Succ.getSUnit()) == SU)
       ++NumNodesBlocking;
 
   NumNodesSolelyBlocking[SU->NodeNum] = NumNodesBlocking;
@@ -269,14 +263,13 @@ bool ResourcePriorityQueue::isResourceAvailable(SUnit *SU) {
   // Now see if there are no other dependencies
   // to instructions already in the packet.
   for (unsigned i = 0, e = Packet.size(); i != e; ++i)
-    for (SUnit::const_succ_iterator I = Packet[i]->Succs.begin(),
-         E = Packet[i]->Succs.end(); I != E; ++I) {
+    for (const SDep &Succ : Packet[i]->Succs) {
       // Since we do not add pseudos to packets, might as well
       // ignore order deps.
-      if (I->isCtrl())
+      if (Succ.isCtrl())
         continue;
 
-      if (I->getSUnit() == SU)
+      if (Succ.getSUnit() == SU)
         return false;
     }
 
@@ -499,11 +492,10 @@ void ResourcePriorityQueue::scheduledNode(SUnit *SU) {
         }
       }
     }
-    for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-                              I != E; ++I) {
-      if (I->isCtrl() || (I->getSUnit()->NumRegDefsLeft == 0))
+    for (SDep &Pred : SU->Preds) {
+      if (Pred.isCtrl() || (Pred.getSUnit()->NumRegDefsLeft == 0))
         continue;
-      --I->getSUnit()->NumRegDefsLeft;
+      --Pred.getSUnit()->NumRegDefsLeft;
     }
   }
 
@@ -515,10 +507,9 @@ void ResourcePriorityQueue::scheduledNode(SUnit *SU) {
   // number of live ranges. All others, increase it.
   unsigned NumberNonControlDeps = 0;
 
-  for (SUnit::const_succ_iterator I = SU->Succs.begin(), E = SU->Succs.end();
-                                  I != E; ++I) {
-    adjustPriorityOfUnscheduledPreds(I->getSUnit());
-    if (!I->isCtrl())
+  for (const SDep &Succ : SU->Succs) {
+    adjustPriorityOfUnscheduledPreds(Succ.getSUnit());
+    if (!Succ.isCtrl())
       NumberNonControlDeps++;
   }
 
@@ -595,8 +586,7 @@ SUnit *ResourcePriorityQueue::pop() {
   std::vector<SUnit *>::iterator Best = Queue.begin();
   if (!DisableDFASched) {
     int BestCost = SUSchedulingCost(*Best);
-    for (std::vector<SUnit *>::iterator I = std::next(Queue.begin()),
-           E = Queue.end(); I != E; ++I) {
+    for (auto I = std::next(Queue.begin()), E = Queue.end(); I != E; ++I) {
 
       if (SUSchedulingCost(*I) > BestCost) {
         BestCost = SUSchedulingCost(*I);
@@ -606,8 +596,7 @@ SUnit *ResourcePriorityQueue::pop() {
   }
   // Use default TD scheduling mechanism.
   else {
-    for (std::vector<SUnit *>::iterator I = std::next(Queue.begin()),
-       E = Queue.end(); I != E; ++I)
+    for (auto I = std::next(Queue.begin()), E = Queue.end(); I != E; ++I)
       if (Picker(*Best, *I))
         Best = I;
   }

lib/CodeGen/SelectionDAG/ScheduleDAGFast.cpp

@@ -160,18 +160,17 @@ void ScheduleDAGFast::ReleasePred(SUnit *SU, SDep *PredEdge) {
 
 void ScheduleDAGFast::ReleasePredecessors(SUnit *SU, unsigned CurCycle) {
   // Bottom up: release predecessors
-  for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-       I != E; ++I) {
-    ReleasePred(SU, &*I);
-    if (I->isAssignedRegDep()) {
+  for (SDep &Pred : SU->Preds) {
+    ReleasePred(SU, &Pred);
+    if (Pred.isAssignedRegDep()) {
       // This is a physical register dependency and it's impossible or
       // expensive to copy the register. Make sure nothing that can
       // clobber the register is scheduled between the predecessor and
       // this node.
-      if (!LiveRegDefs[I->getReg()]) {
+      if (!LiveRegDefs[Pred.getReg()]) {
         ++NumLiveRegs;
-        LiveRegDefs[I->getReg()] = I->getSUnit();
-        LiveRegCycles[I->getReg()] = CurCycle;
+        LiveRegDefs[Pred.getReg()] = Pred.getSUnit();
+        LiveRegCycles[Pred.getReg()] = CurCycle;
       }
     }
   }
@@ -191,16 +190,15 @@ void ScheduleDAGFast::ScheduleNodeBottomUp(SUnit *SU, unsigned CurCycle) {
   ReleasePredecessors(SU, CurCycle);
 
   // Release all the implicit physical register defs that are live.
-  for (SUnit::succ_iterator I = SU->Succs.begin(), E = SU->Succs.end();
-       I != E; ++I) {
-    if (I->isAssignedRegDep()) {
-      if (LiveRegCycles[I->getReg()] == I->getSUnit()->getHeight()) {
+  for (SDep &Succ : SU->Succs) {
+    if (Succ.isAssignedRegDep()) {
+      if (LiveRegCycles[Succ.getReg()] == Succ.getSUnit()->getHeight()) {
         assert(NumLiveRegs > 0 && "NumLiveRegs is already zero!");
-        assert(LiveRegDefs[I->getReg()] == SU &&
+        assert(LiveRegDefs[Succ.getReg()] == SU &&
                "Physical register dependency violated?");
         --NumLiveRegs;
-        LiveRegDefs[I->getReg()] = nullptr;
-        LiveRegCycles[I->getReg()] = 0;
+        LiveRegDefs[Succ.getReg()] = nullptr;
+        LiveRegCycles[Succ.getReg()] = 0;
       }
     }
   }
@@ -282,22 +280,20 @@ SUnit *ScheduleDAGFast::CopyAndMoveSuccessors(SUnit *SU) {
     SmallVector<SDep, 4> LoadPreds;
     SmallVector<SDep, 4> NodePreds;
     SmallVector<SDep, 4> NodeSuccs;
-    for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-         I != E; ++I) {
-      if (I->isCtrl())
-        ChainPred = *I;
-      else if (I->getSUnit()->getNode() &&
-               I->getSUnit()->getNode()->isOperandOf(LoadNode))
-        LoadPreds.push_back(*I);
+    for (SDep &Pred : SU->Preds) {
+      if (Pred.isCtrl())
+        ChainPred = Pred;
+      else if (Pred.getSUnit()->getNode() &&
+               Pred.getSUnit()->getNode()->isOperandOf(LoadNode))
+        LoadPreds.push_back(Pred);
       else
-        NodePreds.push_back(*I);
+        NodePreds.push_back(Pred);
     }
-    for (SUnit::succ_iterator I = SU->Succs.begin(), E = SU->Succs.end();
-         I != E; ++I) {
-      if (I->isCtrl())
-        ChainSuccs.push_back(*I);
+    for (SDep &Succ : SU->Succs) {
+      if (Succ.isCtrl())
+        ChainSuccs.push_back(Succ);
       else
-        NodeSuccs.push_back(*I);
+        NodeSuccs.push_back(Succ);
     }
 
     if (ChainPred.getSUnit()) {
@@ -354,21 +350,19 @@ SUnit *ScheduleDAGFast::CopyAndMoveSuccessors(SUnit *SU) {
   NewSU = Clone(SU);
 
   // New SUnit has the exact same predecessors.
-  for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-       I != E; ++I)
-    if (!I->isArtificial())
-      AddPred(NewSU, *I);
+  for (SDep &Pred : SU->Preds)
+    if (!Pred.isArtificial())
+      AddPred(NewSU, Pred);
 
   // Only copy scheduled successors. Cut them from old node's successor
   // list and move them over.
   SmallVector<std::pair<SUnit *, SDep>, 4> DelDeps;
-  for (SUnit::succ_iterator I = SU->Succs.begin(), E = SU->Succs.end();
-       I != E; ++I) {
-    if (I->isArtificial())
+  for (SDep &Succ : SU->Succs) {
+    if (Succ.isArtificial())
       continue;
-    SUnit *SuccSU = I->getSUnit();
+    SUnit *SuccSU = Succ.getSUnit();
     if (SuccSU->isScheduled) {
-      SDep D = *I;
+      SDep D = Succ;
       D.setSUnit(NewSU);
       AddPred(SuccSU, D);
       D.setSUnit(SU);
@@ -399,16 +393,15 @@ void ScheduleDAGFast::InsertCopiesAndMoveSuccs(SUnit *SU, unsigned Reg,
   // Only copy scheduled successors. Cut them from old node's successor
   // list and move them over.
   SmallVector<std::pair<SUnit *, SDep>, 4> DelDeps;
-  for (SUnit::succ_iterator I = SU->Succs.begin(), E = SU->Succs.end();
-       I != E; ++I) {
-    if (I->isArtificial())
+  for (SDep &Succ : SU->Succs) {
+    if (Succ.isArtificial())
       continue;
-    SUnit *SuccSU = I->getSUnit();
+    SUnit *SuccSU = Succ.getSUnit();
     if (SuccSU->isScheduled) {
-      SDep D = *I;
+      SDep D = Succ;
       D.setSUnit(CopyToSU);
       AddPred(SuccSU, D);
-      DelDeps.push_back(std::make_pair(SuccSU, *I));
+      DelDeps.push_back(std::make_pair(SuccSU, Succ));
     }
   }
   for (unsigned i = 0, e = DelDeps.size(); i != e; ++i) {
@@ -479,10 +472,9 @@ bool ScheduleDAGFast::DelayForLiveRegsBottomUp(SUnit *SU,
 
   SmallSet<unsigned, 4> RegAdded;
   // If this node would clobber any "live" register, then it's not ready.
-  for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-       I != E; ++I) {
-    if (I->isAssignedRegDep()) {
-      CheckForLiveRegDef(I->getSUnit(), I->getReg(), LiveRegDefs,
+  for (SDep &Pred : SU->Preds) {
+    if (Pred.isAssignedRegDep()) {
+      CheckForLiveRegDef(Pred.getSUnit(), Pred.getReg(), LiveRegDefs,
                          RegAdded, LRegs, TRI);
     }
   }
@@ -755,9 +747,8 @@ void ScheduleDAGLinearize::Schedule() {
     // Glue user must be scheduled together with the glue operand. So other
     // users of the glue operand must be treated as its users.
     SDNode *ImmGUser = Glue->getGluedUser();
-    for (SDNode::use_iterator ui = Glue->use_begin(), ue = Glue->use_end();
-         ui != ue; ++ui)
-      if (*ui == ImmGUser)
+    for (const SDNode *U : Glue->uses())
+      if (U == ImmGUser)
         --Degree;
     GUser->setNodeId(UDegree + Degree);
     Glue->setNodeId(1);

lib/CodeGen/SelectionDAG/ScheduleDAGRRList.cpp

@@ -520,21 +520,20 @@ FindCallSeqStart(SDNode *N, unsigned &NestLevel, unsigned &MaxNest,
 /// interference on flags.
 void ScheduleDAGRRList::ReleasePredecessors(SUnit *SU) {
   // Bottom up: release predecessors
-  for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-       I != E; ++I) {
-    ReleasePred(SU, &*I);
-    if (I->isAssignedRegDep()) {
+  for (SDep &Pred : SU->Preds) {
+    ReleasePred(SU, &Pred);
+    if (Pred.isAssignedRegDep()) {
       // This is a physical register dependency and it's impossible or
       // expensive to copy the register. Make sure nothing that can
       // clobber the register is scheduled between the predecessor and
       // this node.
-      SUnit *RegDef = LiveRegDefs[I->getReg()]; (void)RegDef;
-      assert((!RegDef || RegDef == SU || RegDef == I->getSUnit()) &&
+      SUnit *RegDef = LiveRegDefs[Pred.getReg()]; (void)RegDef;
+      assert((!RegDef || RegDef == SU || RegDef == Pred.getSUnit()) &&
              "interference on register dependence");
-      LiveRegDefs[I->getReg()] = I->getSUnit();
-      if (!LiveRegGens[I->getReg()]) {
+      LiveRegDefs[Pred.getReg()] = Pred.getSUnit();
+      if (!LiveRegGens[Pred.getReg()]) {
         ++NumLiveRegs;
-        LiveRegGens[I->getReg()] = SU;
+        LiveRegGens[Pred.getReg()] = SU;
       }
     }
   }
@@ -733,15 +732,14 @@ void ScheduleDAGRRList::ScheduleNodeBottomUp(SUnit *SU) {
   ReleasePredecessors(SU);
 
   // Release all the implicit physical register defs that are live.
-  for (SUnit::succ_iterator I = SU->Succs.begin(), E = SU->Succs.end();
-       I != E; ++I) {
-    // LiveRegDegs[I->getReg()] != SU when SU is a two-address node.
-    if (I->isAssignedRegDep() && LiveRegDefs[I->getReg()] == SU) {
+  for (SDep &Succ : SU->Succs) {
+    // LiveRegDegs[Succ.getReg()] != SU when SU is a two-address node.
+    if (Succ.isAssignedRegDep() && LiveRegDefs[Succ.getReg()] == SU) {
       assert(NumLiveRegs > 0 && "NumLiveRegs is already zero!");
       --NumLiveRegs;
-      LiveRegDefs[I->getReg()] = nullptr;
-      LiveRegGens[I->getReg()] = nullptr;
-      releaseInterferences(I->getReg());
+      LiveRegDefs[Succ.getReg()] = nullptr;
+      LiveRegGens[Succ.getReg()] = nullptr;
+      releaseInterferences(Succ.getReg());
     }
   }
   // Release the special call resource dependence, if this is the beginning
@@ -802,17 +800,16 @@ void ScheduleDAGRRList::UnscheduleNodeBottomUp(SUnit *SU) {
   DEBUG(dbgs() << "*** Unscheduling [" << SU->getHeight() << "]: ");
   DEBUG(SU->dump(this));
 
-  for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-       I != E; ++I) {
-    CapturePred(&*I);
-    if (I->isAssignedRegDep() && SU == LiveRegGens[I->getReg()]){
+  for (SDep &Pred : SU->Preds) {
+    CapturePred(&Pred);
+    if (Pred.isAssignedRegDep() && SU == LiveRegGens[Pred.getReg()]){
       assert(NumLiveRegs > 0 && "NumLiveRegs is already zero!");
-      assert(LiveRegDefs[I->getReg()] == I->getSUnit() &&
+      assert(LiveRegDefs[Pred.getReg()] == Pred.getSUnit() &&
              "Physical register dependency violated?");
       --NumLiveRegs;
-      LiveRegDefs[I->getReg()] = nullptr;
-      LiveRegGens[I->getReg()] = nullptr;
-      releaseInterferences(I->getReg());
+      LiveRegDefs[Pred.getReg()] = nullptr;
+      LiveRegGens[Pred.getReg()] = nullptr;
+      releaseInterferences(Pred.getReg());
     }
   }
 
@@ -895,7 +892,7 @@ void ScheduleDAGRRList::RestoreHazardCheckerBottomUp() {
 
   std::vector<SUnit*>::const_iterator I = (Sequence.end() - LookAhead);
   unsigned HazardCycle = (*I)->getHeight();
-  for (std::vector<SUnit*>::const_iterator E = Sequence.end(); I != E; ++I) {
+  for (auto E = Sequence.end(); I != E; ++I) {
     SUnit *SU = *I;
     for (; SU->getHeight() > HazardCycle; ++HazardCycle) {
       HazardRec->RecedeCycle();
@@ -1261,10 +1258,9 @@ DelayForLiveRegsBottomUp(SUnit *SU, SmallVectorImpl<unsigned> &LRegs) {
   //
   // If SU is the currently live definition of the same register that it uses,
   // then we are free to schedule it.
-  for (SUnit::pred_iterator I = SU->Preds.begin(), E = SU->Preds.end();
-       I != E; ++I) {
-    if (I->isAssignedRegDep() && LiveRegDefs[I->getReg()] != SU)
-      CheckForLiveRegDef(I->getSUnit(), I->getReg(), LiveRegDefs.get(),
+  for (SDep &Pred : SU->Preds) {
+    if (Pred.isAssignedRegDep() && LiveRegDefs[Pred.getReg()] != SU)
+      CheckForLiveRegDef(Pred.getSUnit(), Pred.getReg(), LiveRegDefs.get(),
                          RegAdded, LRegs, TRI);
   }
 
@@ -1743,8 +1739,7 @@ protected:
 template<class SF>
 static SUnit *popFromQueueImpl(std::vector<SUnit*> &Q, SF &Picker) {
   std::vector<SUnit *>::iterator Best = Q.begin();
-  for (std::vector<SUnit *>::iterator I = std::next(Q.begin()),
-         E = Q.end(); I != E; ++I)
+  for (auto I = std::next(Q.begin()), E = Q.end(); I != E; ++I)
     if (Picker(*Best, *I))
       Best = I;
   SUnit *V = *Best;

lib/CodeGen/SelectionDAG/SelectionDAG.cpp

@@ -2017,8 +2017,7 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
       if (SrcOp.getValueSizeInBits() != BitWidth) {
         assert(SrcOp.getValueSizeInBits() > BitWidth &&
                "Expected BUILD_VECTOR implicit truncation");
-        Known2.One = Known2.One.trunc(BitWidth);
-        Known2.Zero = Known2.Zero.trunc(BitWidth);
+        Known2 = Known2.trunc(BitWidth);
       }
 
       // Known bits are the values that are shared by every demanded element.
@@ -2045,8 +2044,7 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
       if (M < 0) {
         // For UNDEF elements, we don't know anything about the common state of
         // the shuffle result.
-        Known.One.clearAllBits();
-        Known.Zero.clearAllBits();
+        Known.resetAll();
         DemandedLHS.clearAllBits();
         DemandedRHS.clearAllBits();
         break;
@@ -2219,14 +2217,13 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
     // Also compute a conservative estimate for high known-0 bits.
     // More trickiness is possible, but this is sufficient for the
     // interesting case of alignment computation.
-    Known.One.clearAllBits();
     unsigned TrailZ = Known.Zero.countTrailingOnes() +
                       Known2.Zero.countTrailingOnes();
     unsigned LeadZ =  std::max(Known.Zero.countLeadingOnes() +
                                Known2.Zero.countLeadingOnes(),
                                BitWidth) - BitWidth;
 
-    Known.Zero.clearAllBits();
+    Known.resetAll();
     Known.Zero.setLowBits(std::min(TrailZ, BitWidth));
     Known.Zero.setHighBits(std::min(LeadZ, BitWidth));
     break;
@@ -2377,7 +2374,10 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
     break;
   }
   case ISD::CTPOP: {
-    Known.Zero.setBitsFrom(Log2_32(BitWidth)+1);
+    computeKnownBits(Op.getOperand(0), Known2, DemandedElts, Depth + 1);
+    // If we know some of the bits are zero, they can't be one.
+    unsigned PossibleOnes = BitWidth - Known2.Zero.countPopulation();
+    Known.Zero.setBitsFrom(Log2_32(PossibleOnes) + 1);
     break;
   }
   case ISD::LOAD: {
@@ -2396,24 +2396,20 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
   case ISD::ZERO_EXTEND_VECTOR_INREG: {
     EVT InVT = Op.getOperand(0).getValueType();
     unsigned InBits = InVT.getScalarSizeInBits();
-    Known.Zero = Known.Zero.trunc(InBits);
-    Known.One = Known.One.trunc(InBits);
+    Known = Known.trunc(InBits);
     computeKnownBits(Op.getOperand(0), Known,
                      DemandedElts.zext(InVT.getVectorNumElements()),
                      Depth + 1);
-    Known.Zero = Known.Zero.zext(BitWidth);
-    Known.One = Known.One.zext(BitWidth);
+    Known = Known.zext(BitWidth);
     Known.Zero.setBitsFrom(InBits);
     break;
   }
   case ISD::ZERO_EXTEND: {
     EVT InVT = Op.getOperand(0).getValueType();
     unsigned InBits = InVT.getScalarSizeInBits();
-    Known.Zero = Known.Zero.trunc(InBits);
-    Known.One = Known.One.trunc(InBits);
+    Known = Known.trunc(InBits);
     computeKnownBits(Op.getOperand(0), Known, DemandedElts, Depth + 1);
-    Known.Zero = Known.Zero.zext(BitWidth);
-    Known.One = Known.One.zext(BitWidth);
+    Known = Known.zext(BitWidth);
     Known.Zero.setBitsFrom(InBits);
     break;
   }
@@ -2422,34 +2418,28 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
     EVT InVT = Op.getOperand(0).getValueType();
     unsigned InBits = InVT.getScalarSizeInBits();
 
-    Known.Zero = Known.Zero.trunc(InBits);
-    Known.One = Known.One.trunc(InBits);
+    Known = Known.trunc(InBits);
     computeKnownBits(Op.getOperand(0), Known, DemandedElts, Depth + 1);
 
     // If the sign bit is known to be zero or one, then sext will extend
     // it to the top bits, else it will just zext.
-    Known.Zero = Known.Zero.sext(BitWidth);
-    Known.One = Known.One.sext(BitWidth);
+    Known = Known.sext(BitWidth);
     break;
   }
   case ISD::ANY_EXTEND: {
     EVT InVT = Op.getOperand(0).getValueType();
     unsigned InBits = InVT.getScalarSizeInBits();
-    Known.Zero = Known.Zero.trunc(InBits);
-    Known.One = Known.One.trunc(InBits);
+    Known = Known.trunc(InBits);
     computeKnownBits(Op.getOperand(0), Known, Depth+1);
-    Known.Zero = Known.Zero.zext(BitWidth);
-    Known.One = Known.One.zext(BitWidth);
+    Known = Known.zext(BitWidth);
     break;
   }
   case ISD::TRUNCATE: {
     EVT InVT = Op.getOperand(0).getValueType();
     unsigned InBits = InVT.getScalarSizeInBits();
-    Known.Zero = Known.Zero.zext(InBits);
-    Known.One = Known.One.zext(InBits);
+    Known = Known.zext(InBits);
     computeKnownBits(Op.getOperand(0), Known, DemandedElts, Depth + 1);
-    Known.Zero = Known.Zero.trunc(BitWidth);
-    Known.One = Known.One.trunc(BitWidth);
+    Known = Known.trunc(BitWidth);
     break;
   }
   case ISD::AssertZext: {
@@ -2606,8 +2596,7 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
 
     uint32_t Leaders = std::max(Known.Zero.countLeadingOnes(),
                                 Known2.Zero.countLeadingOnes());
-    Known.One.clearAllBits();
-    Known.Zero.clearAllBits();
+    Known.resetAll();
     Known.Zero.setHighBits(Leaders);
     break;
   }
@@ -2621,8 +2610,7 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
     Known.One = Known.One.getHiBits(Known.One.getBitWidth() - Index * BitWidth);
 
     // Remove high part of known bit mask
-    Known.Zero = Known.Zero.trunc(BitWidth);
-    Known.One = Known.One.trunc(BitWidth);
+    Known = Known.trunc(BitWidth);
     break;
   }
   case ISD::EXTRACT_VECTOR_ELT: {
@@ -2634,10 +2622,8 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
     const unsigned NumSrcElts = VecVT.getVectorNumElements();
     // If BitWidth > EltBitWidth the value is anyext:ed. So we do not know
     // anything about the extended bits.
-    if (BitWidth > EltBitWidth) {
-      Known.Zero = Known.Zero.trunc(EltBitWidth);
-      Known.One = Known.One.trunc(EltBitWidth);
-    }
+    if (BitWidth > EltBitWidth)
+      Known = Known.trunc(EltBitWidth);
     ConstantSDNode *ConstEltNo = dyn_cast<ConstantSDNode>(EltNo);
     if (ConstEltNo && ConstEltNo->getAPIntValue().ult(NumSrcElts)) {
       // If we know the element index, just demand that vector element.
@@ -2648,10 +2634,8 @@ void SelectionDAG::computeKnownBits(SDValue Op, KnownBits &Known,
       // Unknown element index, so ignore DemandedElts and demand them all.
       computeKnownBits(InVec, Known, Depth + 1);
     }
-    if (BitWidth > EltBitWidth) {
-      Known.Zero = Known.Zero.zext(BitWidth);
-      Known.One = Known.One.zext(BitWidth);
-    }
+    if (BitWidth > EltBitWidth)
+      Known = Known.zext(BitWidth);
     break;
   }
   case ISD::INSERT_VECTOR_ELT: {

lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp

@@ -4992,45 +4992,33 @@ SelectionDAGBuilder::visitIntrinsicCall(const CallInst &I, unsigned Intrinsic) {
       SDV = DAG.getConstantDbgValue(Variable, Expression, V, Offset, dl,
                                     SDNodeOrder);
       DAG.AddDbgValue(SDV, nullptr, false);
-    } else {
-      // Do not use getValue() in here; we don't want to generate code at
-      // this point if it hasn't been done yet.
-      SDValue N = NodeMap[V];
-      if (!N.getNode() && isa<Argument>(V))
-        // Check unused arguments map.
-        N = UnusedArgNodeMap[V];
-      if (N.getNode()) {
-        if (!EmitFuncArgumentDbgValue(V, Variable, Expression, dl, Offset,
-                                      false, N)) {
-          SDV = getDbgValue(N, Variable, Expression, Offset, dl, SDNodeOrder);
-          DAG.AddDbgValue(SDV, N.getNode(), false);
-        }
-      } else if (!V->use_empty() ) {
-        // Do not call getValue(V) yet, as we don't want to generate code.
-        // Remember it for later.
-        DanglingDebugInfo DDI(&DI, dl, SDNodeOrder);
-        DanglingDebugInfoMap[V] = DDI;
-      } else {
-        // We may expand this to cover more cases.  One case where we have no
-        // data available is an unreferenced parameter.
-        DEBUG(dbgs() << "Dropping debug info for " << DI << "\n");
-      }
-    }
-
-    // Build a debug info table entry.
-    if (const BitCastInst *BCI = dyn_cast<BitCastInst>(V))
-      V = BCI->getOperand(0);
-    const AllocaInst *AI = dyn_cast<AllocaInst>(V);
-    // Don't handle byval struct arguments or VLAs, for example.
-    if (!AI) {
-      DEBUG(dbgs() << "Dropping debug location info for:\n  " << DI << "\n");
-      DEBUG(dbgs() << "  Last seen at:\n    " << *V << "\n");
       return nullptr;
     }
-    DenseMap<const AllocaInst*, int>::iterator SI =
-      FuncInfo.StaticAllocaMap.find(AI);
-    if (SI == FuncInfo.StaticAllocaMap.end())
-      return nullptr; // VLAs.
+
+    // Do not use getValue() in here; we don't want to generate code at
+    // this point if it hasn't been done yet.
+    SDValue N = NodeMap[V];
+    if (!N.getNode() && isa<Argument>(V)) // Check unused arguments map.
+      N = UnusedArgNodeMap[V];
+    if (N.getNode()) {
+      if (EmitFuncArgumentDbgValue(V, Variable, Expression, dl, Offset, false,
+                                   N))
+        return nullptr;
+      SDV = getDbgValue(N, Variable, Expression, Offset, dl, SDNodeOrder);
+      DAG.AddDbgValue(SDV, N.getNode(), false);
+      return nullptr;
+    }
+
+    if (!V->use_empty() ) {
+      // Do not call getValue(V) yet, as we don't want to generate code.
+      // Remember it for later.
+      DanglingDebugInfo DDI(&DI, dl, SDNodeOrder);
+      DanglingDebugInfoMap[V] = DDI;
+      return nullptr;
+    }
+
+    DEBUG(dbgs() << "Dropping debug location info for:\n  " << DI << "\n");
+    DEBUG(dbgs() << "  Last seen at:\n    " << *V << "\n");
     return nullptr;
   }
 
@@ -5715,7 +5703,37 @@ SelectionDAGBuilder::visitIntrinsicCall(const CallInst &I, unsigned Intrinsic) {
     setValue(&I, N);
     return nullptr;
   }
+  case Intrinsic::xray_customevent: {
+    // Here we want to make sure that the intrinsic behaves as if it has a
+    // specific calling convention, and only for x86_64.
+    // FIXME: Support other platforms later.
+    const auto &Triple = DAG.getTarget().getTargetTriple();
+    if (Triple.getArch() != Triple::x86_64 || !Triple.isOSLinux())
+      return nullptr;
 
+    SDLoc DL = getCurSDLoc();
+    SmallVector<SDValue, 8> Ops;
+
+    // We want to say that we always want the arguments in registers.
+    SDValue LogEntryVal = getValue(I.getArgOperand(0));
+    SDValue StrSizeVal = getValue(I.getArgOperand(1));
+    SDVTList NodeTys = DAG.getVTList(MVT::Other, MVT::Glue);
+    SDValue Chain = getRoot();
+    Ops.push_back(LogEntryVal);
+    Ops.push_back(StrSizeVal);
+    Ops.push_back(Chain);
+
+    // We need to enforce the calling convention for the callsite, so that
+    // argument ordering is enforced correctly, and that register allocation can
+    // see that some registers may be assumed clobbered and have to preserve
+    // them across calls to the intrinsic.
+    MachineSDNode *MN = DAG.getMachineNode(TargetOpcode::PATCHABLE_EVENT_CALL,
+                                           DL, NodeTys, Ops);
+    SDValue patchableNode = SDValue(MN, 0);
+    DAG.setRoot(patchableNode);
+    setValue(&I, patchableNode);
+    return nullptr;
+  }
   case Intrinsic::experimental_deoptimize:
     LowerDeoptimizeCall(&I);
     return nullptr;