diff --git a/clang/lib/AST/DeclBase.cpp b/clang/lib/AST/DeclBase.cpp
@@ -1115,7 +1115,9 @@ int64_t Decl::getID() const {
 
 const FunctionType *Decl::getFunctionType(bool BlocksToo) const {
   QualType Ty;
-  if (const auto *D = dyn_cast<ValueDecl>(this))
+  if (isa<BindingDecl>(this))
+    return nullptr;
+  else if (const auto *D = dyn_cast<ValueDecl>(this))
     Ty = D->getType();
   else if (const auto *D = dyn_cast<TypedefNameDecl>(this))
     Ty = D->getUnderlyingType();

diff --git a/clang/lib/AST/Expr.cpp b/clang/lib/AST/Expr.cpp
@@ -103,7 +103,7 @@ const Expr *Expr::skipRValueSubobjectAdjustments(
       }
     } else if (const auto *ME = dyn_cast<MemberExpr>(E)) {
       if (!ME->isArrow()) {
-        assert(ME->getBase()->getType()->isRecordType());
+        assert(ME->getBase()->getType()->getAsRecordDecl());
         if (const auto *Field = dyn_cast<FieldDecl>(ME->getMemberDecl())) {
           if (!Field->isBitField() && !Field->getType()->isReferenceType()) {
             E = ME->getBase();
@@ -3893,9 +3893,14 @@ namespace {
     }
 
     void VisitCXXBindTemporaryExpr(const CXXBindTemporaryExpr *E) {
-      if (E->getTemporary()->getDestructor()->isTrivial()) {
-        Inherited::VisitStmt(E);
-        return;
+      // Destructor of the temporary might be null if destructor declaration
+      // is not valid.
+      if (const CXXDestructorDecl *DtorDecl =
+              E->getTemporary()->getDestructor()) {
+        if (DtorDecl->isTrivial()) {
+          Inherited::VisitStmt(E);
+          return;
+        }
       }
 
       NonTrivial = true;

diff --git a/clang/lib/AST/ExprClassification.cpp b/clang/lib/AST/ExprClassification.cpp
@@ -216,8 +216,13 @@ static Cl::Kinds ClassifyInternal(ASTContext &Ctx, const Expr *E) {
     return ClassifyInternal(Ctx,
                  cast<SubstNonTypeTemplateParmExpr>(E)->getReplacement());
 
-  case Expr::PackIndexingExprClass:
+  case Expr::PackIndexingExprClass: {
+    // A pack-index-expression always expands to an id-expression.
+    // Consider it as an LValue expression.
+    if (cast<PackIndexingExpr>(E)->isInstantiationDependent())
+      return Cl::CL_LValue;
     return ClassifyInternal(Ctx, cast<PackIndexingExpr>(E)->getSelectedExpr());
+  }
 
     // C, C++98 [expr.sub]p1: The result is an lvalue of type "T".
     // C++11 (DR1213): in the case of an array operand, the result is an lvalue

diff --git a/clang/lib/AST/ExprConstant.cpp b/clang/lib/AST/ExprConstant.cpp
@@ -2706,7 +2706,11 @@ static bool checkFloatingPointResult(EvalInfo &Info, const Expr *E,
 static bool HandleFloatToFloatCast(EvalInfo &Info, const Expr *E,
                                    QualType SrcType, QualType DestType,
                                    APFloat &Result) {
-  assert(isa<CastExpr>(E) || isa<CompoundAssignOperator>(E));
+  assert((isa<CastExpr>(E) || isa<CompoundAssignOperator>(E) ||
+          isa<ConvertVectorExpr>(E)) &&
+         "HandleFloatToFloatCast has been checked with only CastExpr, "
+         "CompoundAssignOperator and ConvertVectorExpr. Please either validate "
+         "the new expression or address the root cause of this usage.");
   llvm::RoundingMode RM = getActiveRoundingMode(Info, E);
   APFloat::opStatus St;
   APFloat Value = Result;
@@ -9237,9 +9241,10 @@ bool PointerExprEvaluator::VisitCastExpr(const CastExpr *E) {
       bool HasValidResult = !Result.InvalidBase && !Result.Designator.Invalid &&
                             !Result.IsNullPtr;
       bool VoidPtrCastMaybeOK =
-          HasValidResult &&
-          Info.Ctx.hasSameUnqualifiedType(Result.Designator.getType(Info.Ctx),
-                                          E->getType()->getPointeeType());
+          Result.IsNullPtr ||
+          (HasValidResult &&
+           Info.Ctx.hasSimilarType(Result.Designator.getType(Info.Ctx),
+                                   E->getType()->getPointeeType()));
       // 1. We'll allow it in std::allocator::allocate, and anything which that
       //    calls.
       // 2. HACK 2022-03-28: Work around an issue with libstdc++'s
@@ -10709,8 +10714,11 @@ namespace {
     bool VisitUnaryImag(const UnaryOperator *E);
     bool VisitBinaryOperator(const BinaryOperator *E);
     bool VisitUnaryOperator(const UnaryOperator *E);
+    bool VisitConvertVectorExpr(const ConvertVectorExpr *E);
+    bool VisitShuffleVectorExpr(const ShuffleVectorExpr *E);
+
     // FIXME: Missing: conditional operator (for GNU
-    //                 conditional select), shufflevector, ExtVectorElementExpr
+    //                 conditional select), ExtVectorElementExpr
   };
 } // end anonymous namespace
 
@@ -10961,6 +10969,122 @@ bool VectorExprEvaluator::VisitUnaryOperator(const UnaryOperator *E) {
   return Success(APValue(ResultElements.data(), ResultElements.size()), E);
 }
 
+static bool handleVectorElementCast(EvalInfo &Info, const FPOptions FPO,
+                                    const Expr *E, QualType SourceTy,
+                                    QualType DestTy, APValue const &Original,
+                                    APValue &Result) {
+  if (SourceTy->isIntegerType()) {
+    if (DestTy->isRealFloatingType()) {
+      Result = APValue(APFloat(0.0));
+      return HandleIntToFloatCast(Info, E, FPO, SourceTy, Original.getInt(),
+                                  DestTy, Result.getFloat());
+    }
+    if (DestTy->isIntegerType()) {
+      Result = APValue(
+          HandleIntToIntCast(Info, E, DestTy, SourceTy, Original.getInt()));
+      return true;
+    }
+  } else if (SourceTy->isRealFloatingType()) {
+    if (DestTy->isRealFloatingType()) {
+      Result = Original;
+      return HandleFloatToFloatCast(Info, E, SourceTy, DestTy,
+                                    Result.getFloat());
+    }
+    if (DestTy->isIntegerType()) {
+      Result = APValue(APSInt());
+      return HandleFloatToIntCast(Info, E, SourceTy, Original.getFloat(),
+                                  DestTy, Result.getInt());
+    }
+  }
+
+  Info.FFDiag(E, diag::err_convertvector_constexpr_unsupported_vector_cast)
+      << SourceTy << DestTy;
+  return false;
+}
+
+bool VectorExprEvaluator::VisitConvertVectorExpr(const ConvertVectorExpr *E) {
+  APValue Source;
+  QualType SourceVecType = E->getSrcExpr()->getType();
+  if (!EvaluateAsRValue(Info, E->getSrcExpr(), Source))
+    return false;
+
+  QualType DestTy = E->getType()->castAs<VectorType>()->getElementType();
+  QualType SourceTy = SourceVecType->castAs<VectorType>()->getElementType();
+
+  const FPOptions FPO = E->getFPFeaturesInEffect(Info.Ctx.getLangOpts());
+
+  auto SourceLen = Source.getVectorLength();
+  SmallVector<APValue, 4> ResultElements;
+  ResultElements.reserve(SourceLen);
+  for (unsigned EltNum = 0; EltNum < SourceLen; ++EltNum) {
+    APValue Elt;
+    if (!handleVectorElementCast(Info, FPO, E, SourceTy, DestTy,
+                                 Source.getVectorElt(EltNum), Elt))
+      return false;
+    ResultElements.push_back(std::move(Elt));
+  }
+
+  return Success(APValue(ResultElements.data(), ResultElements.size()), E);
+}
+
+static bool handleVectorShuffle(EvalInfo &Info, const ShuffleVectorExpr *E,
+                                QualType ElemType, APValue const &VecVal1,
+                                APValue const &VecVal2, unsigned EltNum,
+                                APValue &Result) {
+  unsigned const TotalElementsInInputVector1 = VecVal1.getVectorLength();
+  unsigned const TotalElementsInInputVector2 = VecVal2.getVectorLength();
+
+  APSInt IndexVal = E->getShuffleMaskIdx(Info.Ctx, EltNum);
+  int64_t index = IndexVal.getExtValue();
+  // The spec says that -1 should be treated as undef for optimizations,
+  // but in constexpr we'd have to produce an APValue::Indeterminate,
+  // which is prohibited from being a top-level constant value. Emit a
+  // diagnostic instead.
+  if (index == -1) {
+    Info.FFDiag(
+        E, diag::err_shufflevector_minus_one_is_undefined_behavior_constexpr)
+        << EltNum;
+    return false;
+  }
+
+  if (index < 0 ||
+      index >= TotalElementsInInputVector1 + TotalElementsInInputVector2)
+    llvm_unreachable("Out of bounds shuffle index");
+
+  if (index >= TotalElementsInInputVector1)
+    Result = VecVal2.getVectorElt(index - TotalElementsInInputVector1);
+  else
+    Result = VecVal1.getVectorElt(index);
+  return true;
+}
+
+bool VectorExprEvaluator::VisitShuffleVectorExpr(const ShuffleVectorExpr *E) {
+  APValue VecVal1;
+  const Expr *Vec1 = E->getExpr(0);
+  if (!EvaluateAsRValue(Info, Vec1, VecVal1))
+    return false;
+  APValue VecVal2;
+  const Expr *Vec2 = E->getExpr(1);
+  if (!EvaluateAsRValue(Info, Vec2, VecVal2))
+    return false;
+
+  VectorType const *DestVecTy = E->getType()->castAs<VectorType>();
+  QualType DestElTy = DestVecTy->getElementType();
+
+  auto TotalElementsInOutputVector = DestVecTy->getNumElements();
+
+  SmallVector<APValue, 4> ResultElements;
+  ResultElements.reserve(TotalElementsInOutputVector);
+  for (unsigned EltNum = 0; EltNum < TotalElementsInOutputVector; ++EltNum) {
+    APValue Elt;
+    if (!handleVectorShuffle(Info, E, DestElTy, VecVal1, VecVal2, EltNum, Elt))
+      return false;
+    ResultElements.push_back(std::move(Elt));
+  }
+
+  return Success(APValue(ResultElements.data(), ResultElements.size()), E);
+}
+
 //===----------------------------------------------------------------------===//
 // Array Evaluation
 //===----------------------------------------------------------------------===//

diff --git a/clang/lib/AST/Interp/ByteCodeExprGen.cpp b/clang/lib/AST/Interp/ByteCodeExprGen.cpp
diff --git a/clang/lib/AST/Interp/ByteCodeExprGen.h b/clang/lib/AST/Interp/ByteCodeExprGen.h
@@ -181,6 +181,7 @@ class ByteCodeExprGen : public ConstStmtVisitor<ByteCodeExprGen<Emitter>, bool>,
   bool visitVarDecl(const VarDecl *VD);
   /// Visit an APValue.
   bool visitAPValue(const APValue &Val, PrimType ValType, const Expr *E);
+  bool visitAPValueInitializer(const APValue &Val, const Expr *E);
 
   /// Visits an expression and converts it to a boolean.
   bool visitBool(const Expr *E);
@@ -224,7 +225,8 @@ class ByteCodeExprGen : public ConstStmtVisitor<ByteCodeExprGen<Emitter>, bool>,
     return this->emitFinishInitPop(I);
   }
 
-  bool visitInitList(ArrayRef<const Expr *> Inits, const Expr *E);
+  bool visitInitList(ArrayRef<const Expr *> Inits, const Expr *ArrayFiller,
+                     const Expr *E);
   bool visitArrayElemInit(unsigned ElemIndex, const Expr *Init);
 
   /// Creates a local primitive value.

diff --git a/clang/lib/AST/Interp/ByteCodeStmtGen.cpp b/clang/lib/AST/Interp/ByteCodeStmtGen.cpp
@@ -292,8 +292,9 @@ bool ByteCodeStmtGen<Emitter>::visitStmt(const Stmt *S) {
   case Stmt::GCCAsmStmtClass:
   case Stmt::MSAsmStmtClass:
   case Stmt::GotoStmtClass:
-  case Stmt::LabelStmtClass:
     return this->emitInvalid(S);
+  case Stmt::LabelStmtClass:
+    return this->visitStmt(cast<LabelStmt>(S)->getSubStmt());
   default: {
     if (auto *Exp = dyn_cast<Expr>(S))
       return this->discard(Exp);
@@ -332,7 +333,8 @@ bool ByteCodeStmtGen<Emitter>::visitCompoundStmt(
 template <class Emitter>
 bool ByteCodeStmtGen<Emitter>::visitDeclStmt(const DeclStmt *DS) {
   for (auto *D : DS->decls()) {
-    if (isa<StaticAssertDecl, TagDecl, TypedefNameDecl, UsingEnumDecl>(D))
+    if (isa<StaticAssertDecl, TagDecl, TypedefNameDecl, UsingEnumDecl,
+            FunctionDecl>(D))
       continue;
 
     const auto *VD = dyn_cast<VarDecl>(D);

diff --git a/clang/lib/AST/Interp/Disasm.cpp b/clang/lib/AST/Interp/Disasm.cpp
@@ -200,7 +200,7 @@ LLVM_DUMP_METHOD void Descriptor::dump(llvm::raw_ostream &OS) const {
     OS << " primitive";
 
   if (isZeroSizeArray())
-    OS << " zero-size-arrary";
+    OS << " zero-size-array";
   else if (isUnknownSizeArray())
     OS << " unknown-size-array";
 

diff --git a/clang/lib/AST/Interp/Interp.h b/clang/lib/AST/Interp/Interp.h
@@ -1935,10 +1935,15 @@ template <PrimType NameL, PrimType NameR>
 inline bool Shr(InterpState &S, CodePtr OpPC) {
   using LT = typename PrimConv<NameL>::T;
   using RT = typename PrimConv<NameR>::T;
-  const auto &RHS = S.Stk.pop<RT>();
+  auto RHS = S.Stk.pop<RT>();
   const auto &LHS = S.Stk.pop<LT>();
   const unsigned Bits = LHS.bitWidth();
 
+  // OpenCL 6.3j: shift values are effectively % word size of LHS.
+  if (S.getLangOpts().OpenCL)
+    RT::bitAnd(RHS, RT::from(LHS.bitWidth() - 1, RHS.bitWidth()),
+               RHS.bitWidth(), &RHS);
+
   if (!CheckShift(S, OpPC, LHS, RHS, Bits))
     return false;
 
@@ -1960,10 +1965,15 @@ template <PrimType NameL, PrimType NameR>
 inline bool Shl(InterpState &S, CodePtr OpPC) {
   using LT = typename PrimConv<NameL>::T;
   using RT = typename PrimConv<NameR>::T;
-  const auto &RHS = S.Stk.pop<RT>();
+  auto RHS = S.Stk.pop<RT>();
   const auto &LHS = S.Stk.pop<LT>();
   const unsigned Bits = LHS.bitWidth();
 
+  // OpenCL 6.3j: shift values are effectively % word size of LHS.
+  if (S.getLangOpts().OpenCL)
+    RT::bitAnd(RHS, RT::from(LHS.bitWidth() - 1, RHS.bitWidth()),
+               RHS.bitWidth(), &RHS);
+
   if (!CheckShift(S, OpPC, LHS, RHS, Bits))
     return false;
 
@@ -2080,6 +2090,24 @@ inline bool ArrayElemPop(InterpState &S, CodePtr OpPC, uint32_t Index) {
   return true;
 }
 
+template <PrimType Name, class T = typename PrimConv<Name>::T>
+inline bool CopyArray(InterpState &S, CodePtr OpPC, uint32_t SrcIndex, uint32_t DestIndex, uint32_t Size) {
+  const auto &SrcPtr = S.Stk.pop<Pointer>();
+  const auto &DestPtr = S.Stk.peek<Pointer>();
+
+  for (uint32_t I = 0; I != Size; ++I) {
+    const Pointer &SP = SrcPtr.atIndex(SrcIndex + I);
+
+    if (!CheckLoad(S, OpPC, SP))
+      return false;
+
+    const Pointer &DP = DestPtr.atIndex(DestIndex + I);
+    DP.deref<T>() = SP.deref<T>();
+    DP.initialize();
+  }
+  return true;
+}
+
 /// Just takes a pointer and checks if it's an incomplete
 /// array type.
 inline bool ArrayDecay(InterpState &S, CodePtr OpPC) {

diff --git a/clang/lib/AST/Interp/Opcodes.td b/clang/lib/AST/Interp/Opcodes.td
@@ -376,7 +376,11 @@ def ArrayElem : Opcode {
   let HasGroup = 1;
 }
 
-
+def CopyArray : Opcode {
+  let Args = [ArgUint32, ArgUint32, ArgUint32];
+  let Types = [AllTypeClass];
+  let HasGroup = 1;
+}
 
 //===----------------------------------------------------------------------===//
 // Direct field accessors

diff --git a/clang/lib/AST/Interp/Program.cpp b/clang/lib/AST/Interp/Program.cpp
@@ -173,7 +173,8 @@ std::optional<unsigned> Program::createGlobal(const ValueDecl *VD,
   if (const auto *Var = dyn_cast<VarDecl>(VD)) {
     IsStatic = Context::shouldBeGloballyIndexed(VD);
     IsExtern = Var->hasExternalStorage();
-  } else if (isa<UnnamedGlobalConstantDecl, MSGuidDecl>(VD)) {
+  } else if (isa<UnnamedGlobalConstantDecl, MSGuidDecl,
+                 TemplateParamObjectDecl>(VD)) {
     IsStatic = true;
     IsExtern = false;
   } else {

diff --git a/clang/lib/AST/OpenACCClause.cpp b/clang/lib/AST/OpenACCClause.cpp
@@ -134,35 +134,67 @@ OpenACCNumGangsClause *OpenACCNumGangsClause::Create(const ASTContext &C,
   return new (Mem) OpenACCNumGangsClause(BeginLoc, LParenLoc, IntExprs, EndLoc);
 }
 
+OpenACCPrivateClause *OpenACCPrivateClause::Create(const ASTContext &C,
+                                                   SourceLocation BeginLoc,
+                                                   SourceLocation LParenLoc,
+                                                   ArrayRef<Expr *> VarList,
+                                                   SourceLocation EndLoc) {
+  void *Mem = C.Allocate(
+      OpenACCPrivateClause::totalSizeToAlloc<Expr *>(VarList.size()));
+  return new (Mem) OpenACCPrivateClause(BeginLoc, LParenLoc, VarList, EndLoc);
+}
+
 //===----------------------------------------------------------------------===//
 //  OpenACC clauses printing methods
 //===----------------------------------------------------------------------===//
+
+void OpenACCClausePrinter::printExpr(const Expr *E) {
+  E->printPretty(OS, nullptr, Policy, 0);
+}
+
 void OpenACCClausePrinter::VisitDefaultClause(const OpenACCDefaultClause &C) {
   OS << "default(" << C.getDefaultClauseKind() << ")";
 }
 
 void OpenACCClausePrinter::VisitIfClause(const OpenACCIfClause &C) {
-  OS << "if(" << C.getConditionExpr() << ")";
+  OS << "if(";
+  printExpr(C.getConditionExpr());
+  OS << ")";
 }
 
 void OpenACCClausePrinter::VisitSelfClause(const OpenACCSelfClause &C) {
   OS << "self";
-  if (const Expr *CondExpr = C.getConditionExpr())
-    OS << "(" << CondExpr << ")";
+  if (const Expr *CondExpr = C.getConditionExpr()) {
+    OS << "(";
+    printExpr(CondExpr);
+    OS << ")";
+  }
 }
 
 void OpenACCClausePrinter::VisitNumGangsClause(const OpenACCNumGangsClause &C) {
   OS << "num_gangs(";
-  llvm::interleaveComma(C.getIntExprs(), OS);
+  llvm::interleaveComma(C.getIntExprs(), OS,
+                        [&](const Expr *E) { printExpr(E); });
   OS << ")";
 }
 
 void OpenACCClausePrinter::VisitNumWorkersClause(
     const OpenACCNumWorkersClause &C) {
-  OS << "num_workers(" << C.getIntExpr() << ")";
+  OS << "num_workers(";
+  printExpr(C.getIntExpr());
+  OS << ")";
 }
 
 void OpenACCClausePrinter::VisitVectorLengthClause(
     const OpenACCVectorLengthClause &C) {
-  OS << "vector_length(" << C.getIntExpr() << ")";
+  OS << "vector_length(";
+  printExpr(C.getIntExpr());
+  OS << ")";
+}
+
+void OpenACCClausePrinter::VisitPrivateClause(const OpenACCPrivateClause &C) {
+  OS << "private(";
+  llvm::interleaveComma(C.getVarList(), OS,
+                        [&](const Expr *E) { printExpr(E); });
+  OS << ")";
 }
diff --git a/clang/lib/AST/StmtPrinter.cpp b/clang/lib/AST/StmtPrinter.cpp
@@ -1148,9 +1148,10 @@ void StmtPrinter::VisitOpenACCComputeConstruct(OpenACCComputeConstruct *S) {
 
   if (!S->clauses().empty()) {
     OS << ' ';
-    OpenACCClausePrinter Printer(OS);
+    OpenACCClausePrinter Printer(OS, Policy);
     Printer.VisitClauseList(S->clauses());
   }
+  OS << '\n';
 
   PrintStmt(S->getStructuredBlock());
 }

diff --git a/clang/lib/AST/StmtProfile.cpp b/clang/lib/AST/StmtProfile.cpp
@@ -2509,6 +2509,12 @@ void OpenACCClauseProfiler::VisitNumWorkersClause(
   Profiler.VisitStmt(Clause.getIntExpr());
 }
 
+void OpenACCClauseProfiler::VisitPrivateClause(
+    const OpenACCPrivateClause &Clause) {
+  for (auto *E : Clause.getVarList())
+    Profiler.VisitStmt(E);
+}
+
 void OpenACCClauseProfiler::VisitVectorLengthClause(
     const OpenACCVectorLengthClause &Clause) {
   assert(Clause.hasIntExpr() &&

diff --git a/clang/lib/AST/TextNodeDumper.cpp b/clang/lib/AST/TextNodeDumper.cpp
@@ -401,6 +401,7 @@ void TextNodeDumper::Visit(const OpenACCClause *C) {
     case OpenACCClauseKind::Self:
     case OpenACCClauseKind::NumGangs:
     case OpenACCClauseKind::NumWorkers:
+    case OpenACCClauseKind::Private:
     case OpenACCClauseKind::VectorLength:
       // The condition expression will be printed as a part of the 'children',
       // but print 'clause' here so it is clear what is happening from the dump.

diff --git a/clang/lib/AST/Type.cpp b/clang/lib/AST/Type.cpp
@@ -2510,6 +2510,18 @@ bool Type::isSveVLSBuiltinType() const {
   return false;
 }
 
+QualType Type::getSizelessVectorEltType(const ASTContext &Ctx) const {
+  assert(isSizelessVectorType() && "Must be sizeless vector type");
+  // Currently supports SVE and RVV
+  if (isSVESizelessBuiltinType())
+    return getSveEltType(Ctx);
+
+  if (isRVVSizelessBuiltinType())
+    return getRVVEltType(Ctx);
+
+  llvm_unreachable("Unhandled type");
+}
+
 QualType Type::getSveEltType(const ASTContext &Ctx) const {
   assert(isSveVLSBuiltinType() && "unsupported type!");
 

diff --git a/clang/lib/Analysis/ExprMutationAnalyzer.cpp b/clang/lib/Analysis/ExprMutationAnalyzer.cpp
@@ -235,15 +235,17 @@ const Stmt *ExprMutationAnalyzer::Analyzer::findMutationMemoized(
   if (Memoized != MemoizedResults.end())
     return Memoized->second;
 
+  // Assume Exp is not mutated before analyzing Exp.
+  MemoizedResults[Exp] = nullptr;
   if (isUnevaluated(Exp))
-    return MemoizedResults[Exp] = nullptr;
+    return nullptr;
 
   for (const auto &Finder : Finders) {
     if (const Stmt *S = (this->*Finder)(Exp))
       return MemoizedResults[Exp] = S;
   }
 
-  return MemoizedResults[Exp] = nullptr;
+  return nullptr;
 }
 
 const Stmt *

diff --git a/clang/lib/Basic/CMakeLists.txt b/clang/lib/Basic/CMakeLists.txt
@@ -130,6 +130,9 @@ add_clang_library(clangBasic
   DEPENDS
   omp_gen
   ClangDriverOptions
+  # These generated headers are included transitively.
+  ARMTargetParserTableGen
+  AArch64TargetParserTableGen
   )
 
 target_link_libraries(clangBasic

diff --git a/clang/lib/Basic/Targets/WebAssembly.cpp b/clang/lib/Basic/Targets/WebAssembly.cpp
@@ -47,6 +47,7 @@ bool WebAssemblyTargetInfo::hasFeature(StringRef Feature) const {
   return llvm::StringSwitch<bool>(Feature)
       .Case("simd128", SIMDLevel >= SIMD128)
       .Case("relaxed-simd", SIMDLevel >= RelaxedSIMD)
+      .Case("half-precision", HasHalfPrecision)
       .Case("nontrapping-fptoint", HasNontrappingFPToInt)
       .Case("sign-ext", HasSignExt)
       .Case("exception-handling", HasExceptionHandling)
@@ -99,6 +100,8 @@ void WebAssemblyTargetInfo::getTargetDefines(const LangOptions &Opts,
     Builder.defineMacro("__wasm_extended_const__");
   if (HasMultiMemory)
     Builder.defineMacro("__wasm_multimemory__");
+  if (HasHalfPrecision)
+    Builder.defineMacro("__wasm_half_precision__");
 
   Builder.defineMacro("__GCC_HAVE_SYNC_COMPARE_AND_SWAP_1");
   Builder.defineMacro("__GCC_HAVE_SYNC_COMPARE_AND_SWAP_2");
@@ -147,19 +150,26 @@ void WebAssemblyTargetInfo::setFeatureEnabled(llvm::StringMap<bool> &Features,
 bool WebAssemblyTargetInfo::initFeatureMap(
     llvm::StringMap<bool> &Features, DiagnosticsEngine &Diags, StringRef CPU,
     const std::vector<std::string> &FeaturesVec) const {
-  if (CPU == "bleeding-edge") {
+  auto addGenericFeatures = [&]() {
+    Features["multivalue"] = true;
+    Features["mutable-globals"] = true;
+    Features["sign-ext"] = true;
+  };
+  auto addBleedingEdgeFeatures = [&]() {
+    addGenericFeatures();
     Features["atomics"] = true;
     Features["bulk-memory"] = true;
     Features["multimemory"] = true;
-    Features["mutable-globals"] = true;
     Features["nontrapping-fptoint"] = true;
     Features["reference-types"] = true;
-    Features["sign-ext"] = true;
     Features["tail-call"] = true;
+    Features["half-precision"] = true;
     setSIMDLevel(Features, SIMD128, true);
-  } else if (CPU == "generic") {
-    Features["mutable-globals"] = true;
-    Features["sign-ext"] = true;
+  };
+  if (CPU == "generic") {
+    addGenericFeatures();
+  } else if (CPU == "bleeding-edge") {
+    addBleedingEdgeFeatures();
   }
 
   return TargetInfo::initFeatureMap(Features, Diags, CPU, FeaturesVec);
@@ -216,6 +226,15 @@ bool WebAssemblyTargetInfo::handleTargetFeatures(
       HasBulkMemory = false;
       continue;
     }
+    if (Feature == "+half-precision") {
+      SIMDLevel = std::max(SIMDLevel, SIMD128);
+      HasHalfPrecision = true;
+      continue;
+    }
+    if (Feature == "-half-precision") {
+      HasHalfPrecision = false;
+      continue;
+    }
     if (Feature == "+atomics") {
       HasAtomics = true;
       continue;

diff --git a/clang/lib/Basic/Targets/WebAssembly.h b/clang/lib/Basic/Targets/WebAssembly.h
@@ -64,6 +64,7 @@ class LLVM_LIBRARY_VISIBILITY WebAssemblyTargetInfo : public TargetInfo {
   bool HasReferenceTypes = false;
   bool HasExtendedConst = false;
   bool HasMultiMemory = false;
+  bool HasHalfPrecision = false;
 
   std::string ABI;
 

diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -3885,9 +3885,12 @@ RValue CodeGenFunction::EmitBuiltinExpr(const GlobalDecl GD, unsigned BuiltinID,
   }
 
   case Builtin::BI__builtin_reduce_max: {
-    auto GetIntrinsicID = [](QualType QT) {
+    auto GetIntrinsicID = [this](QualType QT) {
       if (auto *VecTy = QT->getAs<VectorType>())
         QT = VecTy->getElementType();
+      else if (QT->isSizelessVectorType())
+        QT = QT->getSizelessVectorEltType(CGM.getContext());
+
       if (QT->isSignedIntegerType())
         return llvm::Intrinsic::vector_reduce_smax;
       if (QT->isUnsignedIntegerType())
@@ -3900,9 +3903,12 @@ RValue CodeGenFunction::EmitBuiltinExpr(const GlobalDecl GD, unsigned BuiltinID,
   }
 
   case Builtin::BI__builtin_reduce_min: {
-    auto GetIntrinsicID = [](QualType QT) {
+    auto GetIntrinsicID = [this](QualType QT) {
       if (auto *VecTy = QT->getAs<VectorType>())
         QT = VecTy->getElementType();
+      else if (QT->isSizelessVectorType())
+        QT = QT->getSizelessVectorEltType(CGM.getContext());
+
       if (QT->isSignedIntegerType())
         return llvm::Intrinsic::vector_reduce_smin;
       if (QT->isUnsignedIntegerType())

diff --git a/clang/lib/CodeGen/CGCUDANV.cpp b/clang/lib/CodeGen/CGCUDANV.cpp
@@ -424,6 +424,33 @@ void CGNVCUDARuntime::emitDeviceStubBodyNew(CodeGenFunction &CGF,
       CGM.CreateRuntimeFunction(FTy, LaunchKernelName);
   CGF.EmitCall(FI, CGCallee::forDirect(cudaLaunchKernelFn), ReturnValueSlot(),
                LaunchKernelArgs);
+
+  // To prevent CUDA device stub functions from being merged by ICF in MSVC
+  // environment, create an unique global variable for each kernel and write to
+  // the variable in the device stub.
+  if (CGM.getContext().getTargetInfo().getCXXABI().isMicrosoft() &&
+      !CGF.getLangOpts().HIP) {
+    llvm::Function *KernelFunction = llvm::cast<llvm::Function>(Kernel);
+    std::string GlobalVarName = (KernelFunction->getName() + ".id").str();
+
+    llvm::GlobalVariable *HandleVar =
+        CGM.getModule().getNamedGlobal(GlobalVarName);
+    if (!HandleVar) {
+      HandleVar = new llvm::GlobalVariable(
+          CGM.getModule(), CGM.Int8Ty,
+          /*Constant=*/false, KernelFunction->getLinkage(),
+          llvm::ConstantInt::get(CGM.Int8Ty, 0), GlobalVarName);
+      HandleVar->setDSOLocal(KernelFunction->isDSOLocal());
+      HandleVar->setVisibility(KernelFunction->getVisibility());
+      if (KernelFunction->hasComdat())
+        HandleVar->setComdat(CGM.getModule().getOrInsertComdat(GlobalVarName));
+    }
+
+    CGF.Builder.CreateAlignedStore(llvm::ConstantInt::get(CGM.Int8Ty, 1),
+                                   HandleVar, CharUnits::One(),
+                                   /*IsVolatile=*/true);
+  }
+
   CGF.EmitBranch(EndBlock);
 
   CGF.EmitBlock(EndBlock);

diff --git a/clang/lib/CodeGen/CGCall.cpp b/clang/lib/CodeGen/CGCall.cpp
@@ -4698,11 +4698,11 @@ void CodeGenFunction::EmitCallArg(CallArgList &args, const Expr *E,
     AggValueSlot Slot = args.isUsingInAlloca()
         ? createPlaceholderSlot(*this, type) : CreateAggTemp(type, "agg.tmp");
 
-    bool DestroyedInCallee = true, NeedsEHCleanup = true;
+    bool DestroyedInCallee = true, NeedsCleanup = true;
     if (const auto *RD = type->getAsCXXRecordDecl())
       DestroyedInCallee = RD->hasNonTrivialDestructor();
     else
-      NeedsEHCleanup = needsEHCleanup(type.isDestructedType());
+      NeedsCleanup = type.isDestructedType();
 
     if (DestroyedInCallee)
       Slot.setExternallyDestructed();
@@ -4711,14 +4711,15 @@ void CodeGenFunction::EmitCallArg(CallArgList &args, const Expr *E,
     RValue RV = Slot.asRValue();
     args.add(RV, type);
 
-    if (DestroyedInCallee && NeedsEHCleanup) {
+    if (DestroyedInCallee && NeedsCleanup) {
       // Create a no-op GEP between the placeholder and the cleanup so we can
       // RAUW it successfully.  It also serves as a marker of the first
       // instruction where the cleanup is active.
-      pushFullExprCleanup<DestroyUnpassedArg>(EHCleanup, Slot.getAddress(),
-                                              type);
+      pushFullExprCleanup<DestroyUnpassedArg>(NormalAndEHCleanup,
+                                              Slot.getAddress(), type);
       // This unreachable is a temporary marker which will be removed later.
-      llvm::Instruction *IsActive = Builder.CreateUnreachable();
+      llvm::Instruction *IsActive =
+          Builder.CreateFlagLoad(llvm::Constant::getNullValue(Int8PtrTy));
       args.addArgCleanupDeactivation(EHStack.stable_begin(), IsActive);
     }
     return;

diff --git a/clang/lib/CodeGen/CGCleanup.cpp b/clang/lib/CodeGen/CGCleanup.cpp
@@ -634,12 +634,19 @@ static void destroyOptimisticNormalEntry(CodeGenFunction &CGF,
 /// Pops a cleanup block.  If the block includes a normal cleanup, the
 /// current insertion point is threaded through the cleanup, as are
 /// any branch fixups on the cleanup.
-void CodeGenFunction::PopCleanupBlock(bool FallthroughIsBranchThrough) {
+void CodeGenFunction::PopCleanupBlock(bool FallthroughIsBranchThrough,
+                                      bool ForDeactivation) {
   assert(!EHStack.empty() && "cleanup stack is empty!");
   assert(isa<EHCleanupScope>(*EHStack.begin()) && "top not a cleanup!");
   EHCleanupScope &Scope = cast<EHCleanupScope>(*EHStack.begin());
   assert(Scope.getFixupDepth() <= EHStack.getNumBranchFixups());
 
+  // If we are deactivating a normal cleanup, we need to pretend that the
+  // fallthrough is unreachable. We restore this IP before returning.
+  CGBuilderTy::InsertPoint NormalDeactivateOrigIP;
+  if (ForDeactivation && (Scope.isNormalCleanup() || !getLangOpts().EHAsynch)) {
+    NormalDeactivateOrigIP = Builder.saveAndClearIP();
+  }
   // Remember activation information.
   bool IsActive = Scope.isActive();
   Address NormalActiveFlag =
@@ -667,7 +674,8 @@ void CodeGenFunction::PopCleanupBlock(bool FallthroughIsBranchThrough) {
 
   // - whether there's a fallthrough
   llvm::BasicBlock *FallthroughSource = Builder.GetInsertBlock();
-  bool HasFallthrough = (FallthroughSource != nullptr && IsActive);
+  bool HasFallthrough =
+      FallthroughSource != nullptr && (IsActive || HasExistingBranches);
 
   // Branch-through fall-throughs leave the insertion point set to the
   // end of the last cleanup, which points to the current scope.  The
@@ -692,7 +700,11 @@ void CodeGenFunction::PopCleanupBlock(bool FallthroughIsBranchThrough) {
 
   // If we have a prebranched fallthrough into an inactive normal
   // cleanup, rewrite it so that it leads to the appropriate place.
-  if (Scope.isNormalCleanup() && HasPrebranchedFallthrough && !IsActive) {
+  if (Scope.isNormalCleanup() && HasPrebranchedFallthrough &&
+      !RequiresNormalCleanup) {
+    // FIXME: Come up with a program which would need forwarding prebranched
+    // fallthrough and add tests. Otherwise delete this and assert against it.
+    assert(!IsActive);
     llvm::BasicBlock *prebranchDest;
 
     // If the prebranch is semantically branching through the next
@@ -724,6 +736,8 @@ void CodeGenFunction::PopCleanupBlock(bool FallthroughIsBranchThrough) {
     EHStack.popCleanup(); // safe because there are no fixups
     assert(EHStack.getNumBranchFixups() == 0 ||
            EHStack.hasNormalCleanups());
+    if (NormalDeactivateOrigIP.isSet())
+      Builder.restoreIP(NormalDeactivateOrigIP);
     return;
   }
 
@@ -760,11 +774,19 @@ void CodeGenFunction::PopCleanupBlock(bool FallthroughIsBranchThrough) {
   if (!RequiresNormalCleanup) {
     // Mark CPP scope end for passed-by-value Arg temp
     //   per Windows ABI which is "normally" Cleanup in callee
-    if (IsEHa && getInvokeDest() && Builder.GetInsertBlock()) {
-      if (Personality.isMSVCXXPersonality())
+    if (IsEHa && getInvokeDest()) {
+      // If we are deactivating a normal cleanup then we don't have a
+      // fallthrough. Restore original IP to emit CPP scope ends in the correct
+      // block.
+      if (NormalDeactivateOrigIP.isSet())
+        Builder.restoreIP(NormalDeactivateOrigIP);
+      if (Personality.isMSVCXXPersonality() && Builder.GetInsertBlock())
         EmitSehCppScopeEnd();
+      if (NormalDeactivateOrigIP.isSet())
+        NormalDeactivateOrigIP = Builder.saveAndClearIP();
     }
     destroyOptimisticNormalEntry(*this, Scope);
+    Scope.MarkEmitted();
     EHStack.popCleanup();
   } else {
     // If we have a fallthrough and no other need for the cleanup,
@@ -781,6 +803,7 @@ void CodeGenFunction::PopCleanupBlock(bool FallthroughIsBranchThrough) {
       }
 
       destroyOptimisticNormalEntry(*this, Scope);
+      Scope.MarkEmitted();
       EHStack.popCleanup();
 
       EmitCleanup(*this, Fn, cleanupFlags, NormalActiveFlag);
@@ -916,6 +939,7 @@ void CodeGenFunction::PopCleanupBlock(bool FallthroughIsBranchThrough) {
       }
 
       // IV.  Pop the cleanup and emit it.
+      Scope.MarkEmitted();
       EHStack.popCleanup();
       assert(EHStack.hasNormalCleanups() == HasEnclosingCleanups);
 
@@ -984,6 +1008,8 @@ void CodeGenFunction::PopCleanupBlock(bool FallthroughIsBranchThrough) {
     }
   }
 
+  if (NormalDeactivateOrigIP.isSet())
+    Builder.restoreIP(NormalDeactivateOrigIP);
   assert(EHStack.hasNormalCleanups() || EHStack.getNumBranchFixups() == 0);
 
   // Emit the EH cleanup if required.
@@ -1143,25 +1169,6 @@ void CodeGenFunction::EmitBranchThroughCleanup(JumpDest Dest) {
   Builder.ClearInsertionPoint();
 }
 
-static bool IsUsedAsNormalCleanup(EHScopeStack &EHStack,
-                                  EHScopeStack::stable_iterator C) {
-  // If we needed a normal block for any reason, that counts.
-  if (cast<EHCleanupScope>(*EHStack.find(C)).getNormalBlock())
-    return true;
-
-  // Check whether any enclosed cleanups were needed.
-  for (EHScopeStack::stable_iterator
-         I = EHStack.getInnermostNormalCleanup();
-         I != C; ) {
-    assert(C.strictlyEncloses(I));
-    EHCleanupScope &S = cast<EHCleanupScope>(*EHStack.find(I));
-    if (S.getNormalBlock()) return true;
-    I = S.getEnclosingNormalCleanup();
-  }
-
-  return false;
-}
-
 static bool IsUsedAsEHCleanup(EHScopeStack &EHStack,
                               EHScopeStack::stable_iterator cleanup) {
   // If we needed an EH block for any reason, that counts.
@@ -1210,8 +1217,7 @@ static void SetupCleanupBlockActivation(CodeGenFunction &CGF,
   // Calculate whether the cleanup was used:
 
   //   - as a normal cleanup
-  if (Scope.isNormalCleanup() &&
-      (isActivatedInConditional || IsUsedAsNormalCleanup(CGF.EHStack, C))) {
+  if (Scope.isNormalCleanup()) {
     Scope.setTestFlagInNormalCleanup();
     needFlag = true;
   }
@@ -1224,13 +1230,16 @@ static void SetupCleanupBlockActivation(CodeGenFunction &CGF,
   }
 
   // If it hasn't yet been used as either, we're done.
-  if (!needFlag) return;
+  if (!needFlag)
+    return;
 
   Address var = Scope.getActiveFlag();
   if (!var.isValid()) {
+    CodeGenFunction::AllocaTrackerRAII AllocaTracker(CGF);
     var = CGF.CreateTempAlloca(CGF.Builder.getInt1Ty(), CharUnits::One(),
                                "cleanup.isactive");
     Scope.setActiveFlag(var);
+    Scope.AddAuxAllocas(AllocaTracker.Take());
 
     assert(dominatingIP && "no existing variable and no dominating IP!");
 
@@ -1273,17 +1282,8 @@ void CodeGenFunction::DeactivateCleanupBlock(EHScopeStack::stable_iterator C,
   // to the current RunCleanupsScope.
   if (C == EHStack.stable_begin() &&
       CurrentCleanupScopeDepth.strictlyEncloses(C)) {
-    // Per comment below, checking EHAsynch is not really necessary
-    // it's there to assure zero-impact w/o EHAsynch option
-    if (!Scope.isNormalCleanup() && getLangOpts().EHAsynch) {
-      PopCleanupBlock();
-    } else {
-      // If it's a normal cleanup, we need to pretend that the
-      // fallthrough is unreachable.
-      CGBuilderTy::InsertPoint SavedIP = Builder.saveAndClearIP();
-      PopCleanupBlock();
-      Builder.restoreIP(SavedIP);
-    }
+    PopCleanupBlock(/*FallthroughIsBranchThrough=*/false,
+                    /*ForDeactivation=*/true);
     return;
   }
 

diff --git a/clang/lib/CodeGen/CGCleanup.h b/clang/lib/CodeGen/CGCleanup.h
@@ -16,8 +16,11 @@
 #include "EHScopeStack.h"
 
 #include "Address.h"
+#include "llvm/ADT/STLExtras.h"
+#include "llvm/ADT/SetVector.h"
 #include "llvm/ADT/SmallPtrSet.h"
 #include "llvm/ADT/SmallVector.h"
+#include "llvm/IR/Instruction.h"
 
 namespace llvm {
 class BasicBlock;
@@ -266,6 +269,51 @@ class alignas(8) EHCleanupScope : public EHScope {
   };
   mutable struct ExtInfo *ExtInfo;
 
+  /// Erases auxillary allocas and their usages for an unused cleanup.
+  /// Cleanups should mark these allocas as 'used' if the cleanup is
+  /// emitted, otherwise these instructions would be erased.
+  struct AuxillaryAllocas {
+    SmallVector<llvm::Instruction *, 1> AuxAllocas;
+    bool used = false;
+
+    // Records a potentially unused instruction to be erased later.
+    void Add(llvm::AllocaInst *Alloca) { AuxAllocas.push_back(Alloca); }
+
+    // Mark all recorded instructions as used. These will not be erased later.
+    void MarkUsed() {
+      used = true;
+      AuxAllocas.clear();
+    }
+
+    ~AuxillaryAllocas() {
+      if (used)
+        return;
+      llvm::SetVector<llvm::Instruction *> Uses;
+      for (auto *Inst : llvm::reverse(AuxAllocas))
+        CollectUses(Inst, Uses);
+      // Delete uses in the reverse order of insertion.
+      for (auto *I : llvm::reverse(Uses))
+        I->eraseFromParent();
+    }
+
+  private:
+    void CollectUses(llvm::Instruction *I,
+                     llvm::SetVector<llvm::Instruction *> &Uses) {
+      if (!I || !Uses.insert(I))
+        return;
+      for (auto *User : I->users())
+        CollectUses(cast<llvm::Instruction>(User), Uses);
+    }
+  };
+  mutable struct AuxillaryAllocas *AuxAllocas;
+
+  AuxillaryAllocas &getAuxillaryAllocas() {
+    if (!AuxAllocas) {
+      AuxAllocas = new struct AuxillaryAllocas();
+    }
+    return *AuxAllocas;
+  }
+
   /// The number of fixups required by enclosing scopes (not including
   /// this one).  If this is the top cleanup scope, all the fixups
   /// from this index onwards belong to this scope.
@@ -298,7 +346,7 @@ class alignas(8) EHCleanupScope : public EHScope {
                  EHScopeStack::stable_iterator enclosingEH)
       : EHScope(EHScope::Cleanup, enclosingEH),
         EnclosingNormal(enclosingNormal), NormalBlock(nullptr),
-        ActiveFlag(Address::invalid()), ExtInfo(nullptr),
+        ActiveFlag(Address::invalid()), ExtInfo(nullptr), AuxAllocas(nullptr),
         FixupDepth(fixupDepth) {
     CleanupBits.IsNormalCleanup = isNormal;
     CleanupBits.IsEHCleanup = isEH;
@@ -312,8 +360,15 @@ class alignas(8) EHCleanupScope : public EHScope {
   }
 
   void Destroy() {
+    if (AuxAllocas)
+      delete AuxAllocas;
     delete ExtInfo;
   }
+  void AddAuxAllocas(llvm::SmallVector<llvm::AllocaInst *> Allocas) {
+    for (auto *Alloca : Allocas)
+      getAuxillaryAllocas().Add(Alloca);
+  }
+  void MarkEmitted() { getAuxillaryAllocas().MarkUsed(); }
   // Objects of EHCleanupScope are not destructed. Use Destroy().
   ~EHCleanupScope() = delete;
 

diff --git a/clang/lib/CodeGen/CGDebugInfo.cpp b/clang/lib/CodeGen/CGDebugInfo.cpp
@@ -1373,6 +1373,8 @@ llvm::DIType *CGDebugInfo::CreateType(const TemplateSpecializationType *Ty,
   SourceLocation Loc = AliasDecl->getLocation();
 
   if (CGM.getCodeGenOpts().DebugTemplateAlias &&
+      // FIXME: This is a workaround for the issue
+      //        https://github.com/llvm/llvm-project/issues/89774
       // The TemplateSpecializationType doesn't contain any instantiation
       // information; dependent template arguments can't be resolved. For now,
       // fall back to DW_TAG_typedefs for template aliases that are

diff --git a/clang/lib/CodeGen/CGDecl.cpp b/clang/lib/CodeGen/CGDecl.cpp
@@ -19,6 +19,7 @@
 #include "CodeGenFunction.h"
 #include "CodeGenModule.h"
 #include "ConstantEmitter.h"
+#include "EHScopeStack.h"
 #include "PatternInit.h"
 #include "TargetInfo.h"
 #include "clang/AST/ASTContext.h"
@@ -35,6 +36,7 @@
 #include "llvm/Analysis/ValueTracking.h"
 #include "llvm/IR/DataLayout.h"
 #include "llvm/IR/GlobalVariable.h"
+#include "llvm/IR/Instructions.h"
 #include "llvm/IR/Intrinsics.h"
 #include "llvm/IR/Type.h"
 #include <optional>
@@ -2201,6 +2203,27 @@ void CodeGenFunction::pushDestroy(CleanupKind cleanupKind, Address addr,
                                      destroyer, useEHCleanupForArray);
 }
 
+// Pushes a destroy and defers its deactivation until its
+// CleanupDeactivationScope is exited.
+void CodeGenFunction::pushDestroyAndDeferDeactivation(
+    QualType::DestructionKind dtorKind, Address addr, QualType type) {
+  assert(dtorKind && "cannot push destructor for trivial type");
+
+  CleanupKind cleanupKind = getCleanupKind(dtorKind);
+  pushDestroyAndDeferDeactivation(
+      cleanupKind, addr, type, getDestroyer(dtorKind), cleanupKind & EHCleanup);
+}
+
+void CodeGenFunction::pushDestroyAndDeferDeactivation(
+    CleanupKind cleanupKind, Address addr, QualType type, Destroyer *destroyer,
+    bool useEHCleanupForArray) {
+  llvm::Instruction *DominatingIP =
+      Builder.CreateFlagLoad(llvm::Constant::getNullValue(Int8PtrTy));
+  pushDestroy(cleanupKind, addr, type, destroyer, useEHCleanupForArray);
+  DeferredDeactivationCleanupStack.push_back(
+      {EHStack.stable_begin(), DominatingIP});
+}
+
 void CodeGenFunction::pushStackRestore(CleanupKind Kind, Address SPMem) {
   EHStack.pushCleanup<CallStackRestore>(Kind, SPMem);
 }
@@ -2217,39 +2240,48 @@ void CodeGenFunction::pushLifetimeExtendedDestroy(CleanupKind cleanupKind,
   // If we're not in a conditional branch, we don't need to bother generating a
   // conditional cleanup.
   if (!isInConditionalBranch()) {
-    // Push an EH-only cleanup for the object now.
     // FIXME: When popping normal cleanups, we need to keep this EH cleanup
     // around in case a temporary's destructor throws an exception.
-    if (cleanupKind & EHCleanup)
-      EHStack.pushCleanup<DestroyObject>(
-          static_cast<CleanupKind>(cleanupKind & ~NormalCleanup), addr, type,
-          destroyer, useEHCleanupForArray);
 
+    // Add the cleanup to the EHStack. After the full-expr, this would be
+    // deactivated before being popped from the stack.
+    pushDestroyAndDeferDeactivation(cleanupKind, addr, type, destroyer,
+                                    useEHCleanupForArray);
+
+    // Since this is lifetime-extended, push it once again to the EHStack after
+    // the full expression.
     return pushCleanupAfterFullExprWithActiveFlag<DestroyObject>(
-        cleanupKind, Address::invalid(), addr, type, destroyer, useEHCleanupForArray);
+        cleanupKind, Address::invalid(), addr, type, destroyer,
+        useEHCleanupForArray);
   }
 
   // Otherwise, we should only destroy the object if it's been initialized.
-  // Re-use the active flag and saved address across both the EH and end of
-  // scope cleanups.
 
-  using SavedType = typename DominatingValue<Address>::saved_type;
   using ConditionalCleanupType =
       EHScopeStack::ConditionalCleanup<DestroyObject, Address, QualType,
                                        Destroyer *, bool>;
-
-  Address ActiveFlag = createCleanupActiveFlag();
-  SavedType SavedAddr = saveValueInCond(addr);
-
-  if (cleanupKind & EHCleanup) {
-    EHStack.pushCleanup<ConditionalCleanupType>(
-        static_cast<CleanupKind>(cleanupKind & ~NormalCleanup), SavedAddr, type,
-        destroyer, useEHCleanupForArray);
-    initFullExprCleanupWithFlag(ActiveFlag);
-  }
-
+  DominatingValue<Address>::saved_type SavedAddr = saveValueInCond(addr);
+
+  // Remember to emit cleanup if we branch-out before end of full-expression
+  // (eg: through stmt-expr or coro suspensions).
+  AllocaTrackerRAII DeactivationAllocas(*this);
+  Address ActiveFlagForDeactivation = createCleanupActiveFlag();
+
+  pushCleanupAndDeferDeactivation<ConditionalCleanupType>(
+      cleanupKind, SavedAddr, type, destroyer, useEHCleanupForArray);
+  initFullExprCleanupWithFlag(ActiveFlagForDeactivation);
+  EHCleanupScope &cleanup = cast<EHCleanupScope>(*EHStack.begin());
+  // Erase the active flag if the cleanup was not emitted.
+  cleanup.AddAuxAllocas(std::move(DeactivationAllocas).Take());
+
+  // Since this is lifetime-extended, push it once again to the EHStack after
+  // the full expression.
+  // The previous active flag would always be 'false' due to forced deferred
+  // deactivation. Use a separate flag for lifetime-extension to correctly
+  // remember if this branch was taken and the object was initialized.
+  Address ActiveFlagForLifetimeExt = createCleanupActiveFlag();
   pushCleanupAfterFullExprWithActiveFlag<ConditionalCleanupType>(
-      cleanupKind, ActiveFlag, SavedAddr, type, destroyer,
+      cleanupKind, ActiveFlagForLifetimeExt, SavedAddr, type, destroyer,
       useEHCleanupForArray);
 }
 
@@ -2442,9 +2474,9 @@ namespace {
   };
 } // end anonymous namespace
 
-/// pushIrregularPartialArrayCleanup - Push an EH cleanup to destroy
-/// already-constructed elements of the given array.  The cleanup
-/// may be popped with DeactivateCleanupBlock or PopCleanupBlock.
+/// pushIrregularPartialArrayCleanup - Push a NormalAndEHCleanup to
+/// destroy already-constructed elements of the given array.  The cleanup may be
+/// popped with DeactivateCleanupBlock or PopCleanupBlock.
 ///
 /// \param elementType - the immediate element type of the array;
 ///   possibly still an array type
@@ -2453,10 +2485,9 @@ void CodeGenFunction::pushIrregularPartialArrayCleanup(llvm::Value *arrayBegin,
                                                        QualType elementType,
                                                        CharUnits elementAlign,
                                                        Destroyer *destroyer) {
-  pushFullExprCleanup<IrregularPartialArrayDestroy>(EHCleanup,
-                                                    arrayBegin, arrayEndPointer,
-                                                    elementType, elementAlign,
-                                                    destroyer);
+  pushFullExprCleanup<IrregularPartialArrayDestroy>(
+      NormalAndEHCleanup, arrayBegin, arrayEndPointer, elementType,
+      elementAlign, destroyer);
 }
 
 /// pushRegularPartialArrayCleanup - Push an EH cleanup to destroy

diff --git a/clang/lib/CodeGen/CGExpr.cpp b/clang/lib/CodeGen/CGExpr.cpp
@@ -115,10 +115,16 @@ RawAddress CodeGenFunction::CreateTempAlloca(llvm::Type *Ty, CharUnits Align,
 llvm::AllocaInst *CodeGenFunction::CreateTempAlloca(llvm::Type *Ty,
                                                     const Twine &Name,
                                                     llvm::Value *ArraySize) {
+  llvm::AllocaInst *Alloca;
   if (ArraySize)
-    return Builder.CreateAlloca(Ty, ArraySize, Name);
-  return new llvm::AllocaInst(Ty, CGM.getDataLayout().getAllocaAddrSpace(),
-                              ArraySize, Name, AllocaInsertPt);
+    Alloca = Builder.CreateAlloca(Ty, ArraySize, Name);
+  else
+    Alloca = new llvm::AllocaInst(Ty, CGM.getDataLayout().getAllocaAddrSpace(),
+                                  ArraySize, Name, AllocaInsertPt);
+  if (Allocas) {
+    Allocas->Add(Alloca);
+  }
+  return Alloca;
 }
 
 /// CreateDefaultAlignTempAlloca - This creates an alloca with the

diff --git a/clang/lib/CodeGen/CGExprAgg.cpp b/clang/lib/CodeGen/CGExprAgg.cpp
@@ -15,6 +15,7 @@
 #include "CodeGenFunction.h"
 #include "CodeGenModule.h"
 #include "ConstantEmitter.h"
+#include "EHScopeStack.h"
 #include "TargetInfo.h"
 #include "clang/AST/ASTContext.h"
 #include "clang/AST/Attr.h"
@@ -24,6 +25,7 @@
 #include "llvm/IR/Constants.h"
 #include "llvm/IR/Function.h"
 #include "llvm/IR/GlobalVariable.h"
+#include "llvm/IR/Instruction.h"
 #include "llvm/IR/IntrinsicInst.h"
 #include "llvm/IR/Intrinsics.h"
 using namespace clang;
@@ -558,24 +560,27 @@ void AggExprEmitter::EmitArrayInit(Address DestPtr, llvm::ArrayType *AType,
   // For that, we'll need an EH cleanup.
   QualType::DestructionKind dtorKind = elementType.isDestructedType();
   Address endOfInit = Address::invalid();
-  EHScopeStack::stable_iterator cleanup;
-  llvm::Instruction *cleanupDominator = nullptr;
-  if (CGF.needsEHCleanup(dtorKind)) {
+  CodeGenFunction::CleanupDeactivationScope deactivation(CGF);
+
+  if (dtorKind) {
+    CodeGenFunction::AllocaTrackerRAII allocaTracker(CGF);
     // In principle we could tell the cleanup where we are more
     // directly, but the control flow can get so varied here that it
     // would actually be quite complex.  Therefore we go through an
     // alloca.
+    llvm::Instruction *dominatingIP =
+        Builder.CreateFlagLoad(llvm::ConstantInt::getNullValue(CGF.Int8PtrTy));
     endOfInit = CGF.CreateTempAlloca(begin->getType(), CGF.getPointerAlign(),
                                      "arrayinit.endOfInit");
-    cleanupDominator = Builder.CreateStore(begin, endOfInit);
+    Builder.CreateStore(begin, endOfInit);
     CGF.pushIrregularPartialArrayCleanup(begin, endOfInit, elementType,
                                          elementAlign,
                                          CGF.getDestroyer(dtorKind));
-    cleanup = CGF.EHStack.stable_begin();
+    cast<EHCleanupScope>(*CGF.EHStack.find(CGF.EHStack.stable_begin()))
+        .AddAuxAllocas(allocaTracker.Take());
 
-  // Otherwise, remember that we didn't need a cleanup.
-  } else {
-    dtorKind = QualType::DK_none;
+    CGF.DeferredDeactivationCleanupStack.push_back(
+        {CGF.EHStack.stable_begin(), dominatingIP});
   }
 
   llvm::Value *one = llvm::ConstantInt::get(CGF.SizeTy, 1);
@@ -671,9 +676,6 @@ void AggExprEmitter::EmitArrayInit(Address DestPtr, llvm::ArrayType *AType,
 
     CGF.EmitBlock(endBB);
   }
-
-  // Leave the partial-array cleanup if we entered one.
-  if (dtorKind) CGF.DeactivateCleanupBlock(cleanup, cleanupDominator);
 }
 
 //===----------------------------------------------------------------------===//
@@ -1374,9 +1376,8 @@ AggExprEmitter::VisitLambdaExpr(LambdaExpr *E) {
   LValue SlotLV = CGF.MakeAddrLValue(Slot.getAddress(), E->getType());
 
   // We'll need to enter cleanup scopes in case any of the element
-  // initializers throws an exception.
-  SmallVector<EHScopeStack::stable_iterator, 16> Cleanups;
-  llvm::Instruction *CleanupDominator = nullptr;
+  // initializers throws an exception or contains branch out of the expressions.
+  CodeGenFunction::CleanupDeactivationScope scope(CGF);
 
   CXXRecordDecl::field_iterator CurField = E->getLambdaClass()->field_begin();
   for (LambdaExpr::const_capture_init_iterator i = E->capture_init_begin(),
@@ -1395,28 +1396,12 @@ AggExprEmitter::VisitLambdaExpr(LambdaExpr *E) {
     if (QualType::DestructionKind DtorKind =
             CurField->getType().isDestructedType()) {
       assert(LV.isSimple());
-      if (CGF.needsEHCleanup(DtorKind)) {
-        if (!CleanupDominator)
-          CleanupDominator = CGF.Builder.CreateAlignedLoad(
-              CGF.Int8Ty,
-              llvm::Constant::getNullValue(CGF.Int8PtrTy),
-              CharUnits::One()); // placeholder
-
-        CGF.pushDestroy(EHCleanup, LV.getAddress(CGF), CurField->getType(),
-                        CGF.getDestroyer(DtorKind), false);
-        Cleanups.push_back(CGF.EHStack.stable_begin());
-      }
+      if (DtorKind)
+        CGF.pushDestroyAndDeferDeactivation(
+            NormalAndEHCleanup, LV.getAddress(CGF), CurField->getType(),
+            CGF.getDestroyer(DtorKind), false);
     }
   }
-
-  // Deactivate all the partial cleanups in reverse order, which
-  // generally means popping them.
-  for (unsigned i = Cleanups.size(); i != 0; --i)
-    CGF.DeactivateCleanupBlock(Cleanups[i-1], CleanupDominator);
-
-  // Destroy the placeholder if we made one.
-  if (CleanupDominator)
-    CleanupDominator->eraseFromParent();
 }
 
 void AggExprEmitter::VisitExprWithCleanups(ExprWithCleanups *E) {
@@ -1705,14 +1690,7 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
   // We'll need to enter cleanup scopes in case any of the element
   // initializers throws an exception.
   SmallVector<EHScopeStack::stable_iterator, 16> cleanups;
-  llvm::Instruction *cleanupDominator = nullptr;
-  auto addCleanup = [&](const EHScopeStack::stable_iterator &cleanup) {
-    cleanups.push_back(cleanup);
-    if (!cleanupDominator) // create placeholder once needed
-      cleanupDominator = CGF.Builder.CreateAlignedLoad(
-          CGF.Int8Ty, llvm::Constant::getNullValue(CGF.Int8PtrTy),
-          CharUnits::One());
-  };
+  CodeGenFunction::CleanupDeactivationScope DeactivateCleanups(CGF);
 
   unsigned curInitIndex = 0;
 
@@ -1735,10 +1713,8 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
       CGF.EmitAggExpr(InitExprs[curInitIndex++], AggSlot);
 
       if (QualType::DestructionKind dtorKind =
-              Base.getType().isDestructedType()) {
-        CGF.pushDestroy(dtorKind, V, Base.getType());
-        addCleanup(CGF.EHStack.stable_begin());
-      }
+              Base.getType().isDestructedType())
+        CGF.pushDestroyAndDeferDeactivation(dtorKind, V, Base.getType());
     }
   }
 
@@ -1815,10 +1791,10 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
     if (QualType::DestructionKind dtorKind
           = field->getType().isDestructedType()) {
       assert(LV.isSimple());
-      if (CGF.needsEHCleanup(dtorKind)) {
-        CGF.pushDestroy(EHCleanup, LV.getAddress(CGF), field->getType(),
-                        CGF.getDestroyer(dtorKind), false);
-        addCleanup(CGF.EHStack.stable_begin());
+      if (dtorKind) {
+        CGF.pushDestroyAndDeferDeactivation(
+            NormalAndEHCleanup, LV.getAddress(CGF), field->getType(),
+            CGF.getDestroyer(dtorKind), false);
         pushedCleanup = true;
       }
     }
@@ -1831,17 +1807,6 @@ void AggExprEmitter::VisitCXXParenListOrInitListExpr(
         if (GEP->use_empty())
           GEP->eraseFromParent();
   }
-
-  // Deactivate all the partial cleanups in reverse order, which
-  // generally means popping them.
-  assert((cleanupDominator || cleanups.empty()) &&
-         "Missing cleanupDominator before deactivating cleanup blocks");
-  for (unsigned i = cleanups.size(); i != 0; --i)
-    CGF.DeactivateCleanupBlock(cleanups[i-1], cleanupDominator);
-
-  // Destroy the placeholder if we made one.
-  if (cleanupDominator)
-    cleanupDominator->eraseFromParent();
 }
 
 void AggExprEmitter::VisitArrayInitLoopExpr(const ArrayInitLoopExpr *E,

diff --git a/clang/lib/CodeGen/CGExprCXX.cpp b/clang/lib/CodeGen/CGExprCXX.cpp
@@ -1008,8 +1008,8 @@ void CodeGenFunction::EmitNewArrayInitializer(
   const Expr *Init = E->getInitializer();
   Address EndOfInit = Address::invalid();
   QualType::DestructionKind DtorKind = ElementType.isDestructedType();
-  EHScopeStack::stable_iterator Cleanup;
-  llvm::Instruction *CleanupDominator = nullptr;
+  CleanupDeactivationScope deactivation(*this);
+  bool pushedCleanup = false;
 
   CharUnits ElementSize = getContext().getTypeSizeInChars(ElementType);
   CharUnits ElementAlign =
@@ -1105,19 +1105,24 @@ void CodeGenFunction::EmitNewArrayInitializer(
     }
 
     // Enter a partial-destruction Cleanup if necessary.
-    if (needsEHCleanup(DtorKind)) {
+    if (DtorKind) {
+      AllocaTrackerRAII AllocaTracker(*this);
       // In principle we could tell the Cleanup where we are more
       // directly, but the control flow can get so varied here that it
       // would actually be quite complex.  Therefore we go through an
       // alloca.
+      llvm::Instruction *DominatingIP =
+          Builder.CreateFlagLoad(llvm::ConstantInt::getNullValue(Int8PtrTy));
       EndOfInit = CreateTempAlloca(BeginPtr.getType(), getPointerAlign(),
                                    "array.init.end");
-      CleanupDominator =
-          Builder.CreateStore(BeginPtr.emitRawPointer(*this), EndOfInit);
       pushIrregularPartialArrayCleanup(BeginPtr.emitRawPointer(*this),
                                        EndOfInit, ElementType, ElementAlign,
                                        getDestroyer(DtorKind));
-      Cleanup = EHStack.stable_begin();
+      cast<EHCleanupScope>(*EHStack.find(EHStack.stable_begin()))
+          .AddAuxAllocas(AllocaTracker.Take());
+      DeferredDeactivationCleanupStack.push_back(
+          {EHStack.stable_begin(), DominatingIP});
+      pushedCleanup = true;
     }
 
     CharUnits StartAlign = CurPtr.getAlignment();
@@ -1164,9 +1169,6 @@ void CodeGenFunction::EmitNewArrayInitializer(
   // initialization.
   llvm::ConstantInt *ConstNum = dyn_cast<llvm::ConstantInt>(NumElements);
   if (ConstNum && ConstNum->getZExtValue() <= InitListElements) {
-    // If there was a Cleanup, deactivate it.
-    if (CleanupDominator)
-      DeactivateCleanupBlock(Cleanup, CleanupDominator);
     return;
   }
 
@@ -1281,24 +1283,22 @@ void CodeGenFunction::EmitNewArrayInitializer(
     Builder.CreateStore(CurPtr.emitRawPointer(*this), EndOfInit);
 
   // Enter a partial-destruction Cleanup if necessary.
-  if (!CleanupDominator && needsEHCleanup(DtorKind)) {
-    llvm::Value *BeginPtrRaw = BeginPtr.emitRawPointer(*this);
-    llvm::Value *CurPtrRaw = CurPtr.emitRawPointer(*this);
-    pushRegularPartialArrayCleanup(BeginPtrRaw, CurPtrRaw, ElementType,
+  if (!pushedCleanup && needsEHCleanup(DtorKind)) {
+    llvm::Instruction *DominatingIP =
+        Builder.CreateFlagLoad(llvm::ConstantInt::getNullValue(Int8PtrTy));
+    pushRegularPartialArrayCleanup(BeginPtr.emitRawPointer(*this),
+                                   CurPtr.emitRawPointer(*this), ElementType,
                                    ElementAlign, getDestroyer(DtorKind));
-    Cleanup = EHStack.stable_begin();
-    CleanupDominator = Builder.CreateUnreachable();
+    DeferredDeactivationCleanupStack.push_back(
+        {EHStack.stable_begin(), DominatingIP});
   }
 
   // Emit the initializer into this element.
   StoreAnyExprIntoOneUnit(*this, Init, Init->getType(), CurPtr,
                           AggValueSlot::DoesNotOverlap);
 
   // Leave the Cleanup if we entered one.
-  if (CleanupDominator) {
-    DeactivateCleanupBlock(Cleanup, CleanupDominator);
-    CleanupDominator->eraseFromParent();
-  }
+  deactivation.ForceDeactivate();
 
   // Advance to the next element by adjusting the pointer type as necessary.
   llvm::Value *NextPtr = Builder.CreateConstInBoundsGEP1_32(

diff --git a/clang/lib/CodeGen/CGExprScalar.cpp b/clang/lib/CodeGen/CGExprScalar.cpp
@@ -2330,7 +2330,7 @@ Value *ScalarExprEmitter::VisitCastExpr(CastExpr *CE) {
     }
 
     // Perform VLAT <-> VLST bitcast through memory.
-    // TODO: since the llvm.experimental.vector.{insert,extract} intrinsics
+    // TODO: since the llvm.vector.{insert,extract} intrinsics
     //       require the element types of the vectors to be the same, we
     //       need to keep this around for bitcasts between VLAT <-> VLST where
     //       the element types of the vectors are not the same, until we figure

diff --git a/clang/lib/CodeGen/CGLoopInfo.cpp b/clang/lib/CodeGen/CGLoopInfo.cpp
@@ -673,8 +673,6 @@ void LoopInfoStack::push(BasicBlock *Header, clang::ASTContext &Ctx,
         setPipelineDisabled(true);
         break;
       case LoopHintAttr::UnrollCount:
-        setUnrollState(LoopAttributes::Disable);
-        break;
       case LoopHintAttr::UnrollAndJamCount:
       case LoopHintAttr::VectorizeWidth:
       case LoopHintAttr::InterleaveCount:

diff --git a/clang/lib/CodeGen/CMakeLists.txt b/clang/lib/CodeGen/CMakeLists.txt
@@ -143,6 +143,9 @@ add_clang_library(clangCodeGen
   DEPENDS
   intrinsics_gen
   ClangDriverOptions
+  # These generated headers are included transitively.
+  ARMTargetParserTableGen
+  AArch64TargetParserTableGen
 
   LINK_LIBS
   clangAST

diff --git a/clang/lib/CodeGen/CodeGenFunction.cpp b/clang/lib/CodeGen/CodeGenFunction.cpp
@@ -91,6 +91,8 @@ CodeGenFunction::CodeGenFunction(CodeGenModule &cgm, bool suppressNewContext)
 
 CodeGenFunction::~CodeGenFunction() {
   assert(LifetimeExtendedCleanupStack.empty() && "failed to emit a cleanup");
+  assert(DeferredDeactivationCleanupStack.empty() &&
+         "missed to deactivate a cleanup");
 
   if (getLangOpts().OpenMP && CurFn)
     CGM.getOpenMPRuntime().functionFinished(*this);
@@ -346,6 +348,10 @@ static void EmitIfUsed(CodeGenFunction &CGF, llvm::BasicBlock *BB) {
 void CodeGenFunction::FinishFunction(SourceLocation EndLoc) {
   assert(BreakContinueStack.empty() &&
          "mismatched push/pop in break/continue stack!");
+  assert(LifetimeExtendedCleanupStack.empty() &&
+         "mismatched push/pop of cleanups in EHStack!");
+  assert(DeferredDeactivationCleanupStack.empty() &&
+         "mismatched activate/deactivate of cleanups!");
 
   bool OnlySimpleReturnStmts = NumSimpleReturnExprs > 0
     && NumSimpleReturnExprs == NumReturnExprs

diff --git a/clang/lib/CodeGen/CodeGenFunction.h b/clang/lib/CodeGen/CodeGenFunction.h
@@ -39,6 +39,7 @@
 #include "llvm/ADT/MapVector.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/Frontend/OpenMP/OMPIRBuilder.h"
+#include "llvm/IR/Instructions.h"
 #include "llvm/IR/ValueHandle.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Transforms/Utils/SanitizerStats.h"
@@ -670,6 +671,51 @@ class CodeGenFunction : public CodeGenTypeCache {
 
   EHScopeStack EHStack;
   llvm::SmallVector<char, 256> LifetimeExtendedCleanupStack;
+
+  // A stack of cleanups which were added to EHStack but have to be deactivated
+  // later before being popped or emitted. These are usually deactivated on
+  // exiting a `CleanupDeactivationScope` scope. For instance, after a
+  // full-expr.
+  //
+  // These are specially useful for correctly emitting cleanups while
+  // encountering branches out of expression (through stmt-expr or coroutine
+  // suspensions).
+  struct DeferredDeactivateCleanup {
+    EHScopeStack::stable_iterator Cleanup;
+    llvm::Instruction *DominatingIP;
+  };
+  llvm::SmallVector<DeferredDeactivateCleanup> DeferredDeactivationCleanupStack;
+
+  // Enters a new scope for capturing cleanups which are deferred to be
+  // deactivated, all of which will be deactivated once the scope is exited.
+  struct CleanupDeactivationScope {
+    CodeGenFunction &CGF;
+    size_t OldDeactivateCleanupStackSize;
+    bool Deactivated;
+    CleanupDeactivationScope(CodeGenFunction &CGF)
+        : CGF(CGF), OldDeactivateCleanupStackSize(
+                        CGF.DeferredDeactivationCleanupStack.size()),
+          Deactivated(false) {}
+
+    void ForceDeactivate() {
+      assert(!Deactivated && "Deactivating already deactivated scope");
+      auto &Stack = CGF.DeferredDeactivationCleanupStack;
+      for (size_t I = Stack.size(); I > OldDeactivateCleanupStackSize; I--) {
+        CGF.DeactivateCleanupBlock(Stack[I - 1].Cleanup,
+                                   Stack[I - 1].DominatingIP);
+        Stack[I - 1].DominatingIP->eraseFromParent();
+      }
+      Stack.resize(OldDeactivateCleanupStackSize);
+      Deactivated = true;
+    }
+
+    ~CleanupDeactivationScope() {
+      if (Deactivated)
+        return;
+      ForceDeactivate();
+    }
+  };
+
   llvm::SmallVector<const JumpDest *, 2> SEHTryEpilogueStack;
 
   llvm::Instruction *CurrentFuncletPad = nullptr;
@@ -875,6 +921,19 @@ class CodeGenFunction : public CodeGenTypeCache {
       new (Buffer + sizeof(Header) + sizeof(T)) RawAddress(ActiveFlag);
   }
 
+  // Push a cleanup onto EHStack and deactivate it later. It is usually
+  // deactivated when exiting a `CleanupDeactivationScope` (for example: after a
+  // full expression).
+  template <class T, class... As>
+  void pushCleanupAndDeferDeactivation(CleanupKind Kind, As... A) {
+    // Placeholder dominating IP for this cleanup.
+    llvm::Instruction *DominatingIP =
+        Builder.CreateFlagLoad(llvm::Constant::getNullValue(Int8PtrTy));
+    EHStack.pushCleanup<T>(Kind, A...);
+    DeferredDeactivationCleanupStack.push_back(
+        {EHStack.stable_begin(), DominatingIP});
+  }
+
   /// Set up the last cleanup that was pushed as a conditional
   /// full-expression cleanup.
   void initFullExprCleanup() {
@@ -898,7 +957,8 @@ class CodeGenFunction : public CodeGenTypeCache {
 
   /// PopCleanupBlock - Will pop the cleanup entry on the stack and
   /// process all branch fixups.
-  void PopCleanupBlock(bool FallThroughIsBranchThrough = false);
+  void PopCleanupBlock(bool FallThroughIsBranchThrough = false,
+                       bool ForDeactivation = false);
 
   /// DeactivateCleanupBlock - Deactivates the given cleanup block.
   /// The block cannot be reactivated.  Pops it if it's the top of the
@@ -926,6 +986,7 @@ class CodeGenFunction : public CodeGenTypeCache {
   class RunCleanupsScope {
     EHScopeStack::stable_iterator CleanupStackDepth, OldCleanupScopeDepth;
     size_t LifetimeExtendedCleanupStackSize;
+    CleanupDeactivationScope DeactivateCleanups;
     bool OldDidCallStackSave;
   protected:
     bool PerformCleanup;
@@ -940,8 +1001,7 @@ class CodeGenFunction : public CodeGenTypeCache {
   public:
     /// Enter a new cleanup scope.
     explicit RunCleanupsScope(CodeGenFunction &CGF)
-      : PerformCleanup(true), CGF(CGF)
-    {
+        : DeactivateCleanups(CGF), PerformCleanup(true), CGF(CGF) {
       CleanupStackDepth = CGF.EHStack.stable_begin();
       LifetimeExtendedCleanupStackSize =
           CGF.LifetimeExtendedCleanupStack.size();
@@ -971,6 +1031,7 @@ class CodeGenFunction : public CodeGenTypeCache {
     void ForceCleanup(std::initializer_list<llvm::Value**> ValuesToReload = {}) {
       assert(PerformCleanup && "Already forced cleanup");
       CGF.DidCallStackSave = OldDidCallStackSave;
+      DeactivateCleanups.ForceDeactivate();
       CGF.PopCleanupBlocks(CleanupStackDepth, LifetimeExtendedCleanupStackSize,
                            ValuesToReload);
       PerformCleanup = false;
@@ -2160,6 +2221,11 @@ class CodeGenFunction : public CodeGenTypeCache {
                      Address addr, QualType type);
   void pushDestroy(CleanupKind kind, Address addr, QualType type,
                    Destroyer *destroyer, bool useEHCleanupForArray);
+  void pushDestroyAndDeferDeactivation(QualType::DestructionKind dtorKind,
+                                       Address addr, QualType type);
+  void pushDestroyAndDeferDeactivation(CleanupKind cleanupKind, Address addr,
+                                       QualType type, Destroyer *destroyer,
+                                       bool useEHCleanupForArray);
   void pushLifetimeExtendedDestroy(CleanupKind kind, Address addr,
                                    QualType type, Destroyer *destroyer,
                                    bool useEHCleanupForArray);
@@ -2698,6 +2764,33 @@ class CodeGenFunction : public CodeGenTypeCache {
                             TBAAAccessInfo *TBAAInfo = nullptr);
   LValue EmitLoadOfPointerLValue(Address Ptr, const PointerType *PtrTy);
 
+private:
+  struct AllocaTracker {
+    void Add(llvm::AllocaInst *I) { Allocas.push_back(I); }
+    llvm::SmallVector<llvm::AllocaInst *> Take() { return std::move(Allocas); }
+
+  private:
+    llvm::SmallVector<llvm::AllocaInst *> Allocas;
+  };
+  AllocaTracker *Allocas = nullptr;
+
+public:
+  // Captures all the allocas created during the scope of its RAII object.
+  struct AllocaTrackerRAII {
+    AllocaTrackerRAII(CodeGenFunction &CGF)
+        : CGF(CGF), OldTracker(CGF.Allocas) {
+      CGF.Allocas = &Tracker;
+    }
+    ~AllocaTrackerRAII() { CGF.Allocas = OldTracker; }
+
+    llvm::SmallVector<llvm::AllocaInst *> Take() { return Tracker.Take(); }
+
+  private:
+    CodeGenFunction &CGF;
+    AllocaTracker *OldTracker;
+    AllocaTracker Tracker;
+  };
+
   /// CreateTempAlloca - This creates an alloca and inserts it into the entry
   /// block if \p ArraySize is nullptr, otherwise inserts it at the current
   /// insertion point of the builder. The caller is responsible for setting an

diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -73,6 +73,7 @@
 #include "llvm/TargetParser/RISCVISAInfo.h"
 #include "llvm/TargetParser/Triple.h"
 #include "llvm/TargetParser/X86TargetParser.h"
+#include "llvm/Transforms/Utils/BuildLibCalls.h"
 #include <optional>
 
 using namespace clang;
@@ -442,6 +443,11 @@ CodeGenModule::CodeGenModule(ASTContext &C,
       }
     ModuleNameHash = llvm::getUniqueInternalLinkagePostfix(Path);
   }
+
+  // Record mregparm value now so it is visible through all of codegen.
+  if (Context.getTargetInfo().getTriple().getArch() == llvm::Triple::x86)
+    getModule().addModuleFlag(llvm::Module::Error, "NumRegisterParameters",
+                              CodeGenOpts.NumRegisterParameters);
 }
 
 CodeGenModule::~CodeGenModule() {}
@@ -980,11 +986,6 @@ void CodeGenModule::Release() {
       NMD->addOperand(MD);
   }
 
-  // Record mregparm value now so it is visible through rest of codegen.
-  if (Context.getTargetInfo().getTriple().getArch() == llvm::Triple::x86)
-    getModule().addModuleFlag(llvm::Module::Error, "NumRegisterParameters",
-                              CodeGenOpts.NumRegisterParameters);
-
   if (CodeGenOpts.DwarfVersion) {
     getModule().addModuleFlag(llvm::Module::Max, "Dwarf Version",
                               CodeGenOpts.DwarfVersion);
@@ -4781,6 +4782,10 @@ CodeGenModule::CreateRuntimeFunction(llvm::FunctionType *FTy, StringRef Name,
         }
       }
       setDSOLocal(F);
+      // FIXME: We should use CodeGenModule::SetLLVMFunctionAttributes() instead
+      // of trying to approximate the attributes using the LLVM function
+      // signature. This requires revising the API of CreateRuntimeFunction().
+      markRegisterParameterAttributes(F);
     }
   }
 

diff --git a/clang/lib/CodeGen/MicrosoftCXXABI.cpp b/clang/lib/CodeGen/MicrosoftCXXABI.cpp
@@ -1131,9 +1131,15 @@ static bool isTrivialForMSVC(const CXXRecordDecl *RD, QualType Ty,
     return false;
   if (RD->hasNonTrivialCopyAssignment())
     return false;
-  for (const CXXConstructorDecl *Ctor : RD->ctors())
-    if (Ctor->isUserProvided())
-      return false;
+  for (const Decl *D : RD->decls()) {
+    if (auto *Ctor = dyn_cast<CXXConstructorDecl>(D)) {
+      if (Ctor->isUserProvided())
+        return false;
+    } else if (auto *Template = dyn_cast<FunctionTemplateDecl>(D)) {
+      if (isa<CXXConstructorDecl>(Template->getTemplatedDecl()))
+        return false;
+    }
+  }
   if (RD->hasNonTrivialDestructor())
     return false;
   return true;

diff --git a/clang/lib/Driver/CMakeLists.txt b/clang/lib/Driver/CMakeLists.txt
@@ -90,6 +90,9 @@ add_clang_library(clangDriver
 
   DEPENDS
   ClangDriverOptions
+  # These generated headers are included transitively.
+  ARMTargetParserTableGen
+  AArch64TargetParserTableGen
 
   LINK_LIBS
   clangBasic

diff --git a/clang/lib/Driver/Driver.cpp b/clang/lib/Driver/Driver.cpp
@@ -361,6 +361,7 @@ phases::ID Driver::getFinalPhase(const DerivedArgList &DAL,
              (PhaseArg = DAL.getLastArg(options::OPT_rewrite_legacy_objc)) ||
              (PhaseArg = DAL.getLastArg(options::OPT__migrate)) ||
              (PhaseArg = DAL.getLastArg(options::OPT__analyze)) ||
+             (PhaseArg = DAL.getLastArg(options::OPT_emit_cir)) ||
              (PhaseArg = DAL.getLastArg(options::OPT_emit_ast))) {
     FinalPhase = phases::Compile;
 
@@ -4799,6 +4800,8 @@ Action *Driver::ConstructPhaseAction(
       return C.MakeAction<MigrateJobAction>(Input, types::TY_Remap);
     if (Args.hasArg(options::OPT_emit_ast))
       return C.MakeAction<CompileJobAction>(Input, types::TY_AST);
+    if (Args.hasArg(options::OPT_emit_cir))
+      return C.MakeAction<CompileJobAction>(Input, types::TY_CIR);
     if (Args.hasArg(options::OPT_module_file_info))
       return C.MakeAction<CompileJobAction>(Input, types::TY_ModuleFile);
     if (Args.hasArg(options::OPT_verify_pch))

diff --git a/clang/lib/Driver/ToolChain.cpp b/clang/lib/Driver/ToolChain.cpp
@@ -1316,14 +1316,19 @@ bool ToolChain::isFastMathRuntimeAvailable(const ArgList &Args,
   // (to keep the linker options consistent with gcc and clang itself).
   if (Default && !isOptimizationLevelFast(Args)) {
     // Check if -ffast-math or -funsafe-math.
-    Arg *A =
-      Args.getLastArg(options::OPT_ffast_math, options::OPT_fno_fast_math,
-                      options::OPT_funsafe_math_optimizations,
-                      options::OPT_fno_unsafe_math_optimizations);
+    Arg *A = Args.getLastArg(
+        options::OPT_ffast_math, options::OPT_fno_fast_math,
+        options::OPT_funsafe_math_optimizations,
+        options::OPT_fno_unsafe_math_optimizations, options::OPT_ffp_model_EQ);
 
     if (!A || A->getOption().getID() == options::OPT_fno_fast_math ||
         A->getOption().getID() == options::OPT_fno_unsafe_math_optimizations)
       Default = false;
+    if (A && A->getOption().getID() == options::OPT_ffp_model_EQ) {
+      StringRef Model = A->getValue();
+      if (Model != "fast")
+        Default = false;
+    }
   }
 
   // Whatever decision came as a result of the above implicit settings, either

diff --git a/clang/lib/Driver/ToolChains/Clang.cpp b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -2743,13 +2743,11 @@ static void RenderFloatingPointOptions(const ToolChain &TC, const Driver &D,
   StringRef FPExceptionBehavior = "";
   // -ffp-eval-method options: double, extended, source
   StringRef FPEvalMethod = "";
-  const llvm::DenormalMode DefaultDenormalFPMath =
+  llvm::DenormalMode DenormalFPMath =
       TC.getDefaultDenormalModeForType(Args, JA);
-  const llvm::DenormalMode DefaultDenormalFP32Math =
+  llvm::DenormalMode DenormalFP32Math =
       TC.getDefaultDenormalModeForType(Args, JA, &llvm::APFloat::IEEEsingle());
 
-  llvm::DenormalMode DenormalFPMath = DefaultDenormalFPMath;
-  llvm::DenormalMode DenormalFP32Math = DefaultDenormalFP32Math;
   // CUDA and HIP don't rely on the frontend to pass an ffp-contract option.
   // If one wasn't given by the user, don't pass it here.
   StringRef FPContract;
@@ -2899,11 +2897,6 @@ static void RenderFloatingPointOptions(const ToolChain &TC, const Driver &D,
       SignedZeros = true;
       // -fno_fast_math restores default denormal and fpcontract handling
       FPContract = "on";
-      DenormalFPMath = llvm::DenormalMode::getIEEE();
-
-      // FIXME: The target may have picked a non-IEEE default mode here based on
-      // -cl-denorms-are-zero. Should the target consider -fp-model interaction?
-      DenormalFP32Math = llvm::DenormalMode::getIEEE();
 
       StringRef Val = A->getValue();
       if (OFastEnabled && !Val.equals("fast")) {
@@ -3119,12 +3112,7 @@ static void RenderFloatingPointOptions(const ToolChain &TC, const Driver &D,
       ReciprocalMath = false;
       SignedZeros = true;
       ApproxFunc = false;
-      TrappingMath = true;
-      FPExceptionBehavior = "strict";
 
-      // The target may have opted to flush by default, so force IEEE.
-      DenormalFPMath = llvm::DenormalMode::getIEEE();
-      DenormalFP32Math = llvm::DenormalMode::getIEEE();
       if (!JA.isDeviceOffloading(Action::OFK_Cuda) &&
           !JA.isOffloading(Action::OFK_HIP)) {
         if (LastSeenFfpContractOption != "") {
@@ -3154,9 +3142,7 @@ static void RenderFloatingPointOptions(const ToolChain &TC, const Driver &D,
       ReciprocalMath = false;
       ApproxFunc = false;
       SignedZeros = true;
-      // -fno_fast_math restores default denormal and fpcontract handling
-      DenormalFPMath = DefaultDenormalFPMath;
-      DenormalFP32Math = llvm::DenormalMode::getIEEE();
+      // -fno_fast_math restores default fpcontract handling
       if (!JA.isDeviceOffloading(Action::OFK_Cuda) &&
           !JA.isOffloading(Action::OFK_HIP)) {
         if (LastSeenFfpContractOption != "") {
@@ -3171,8 +3157,6 @@ static void RenderFloatingPointOptions(const ToolChain &TC, const Driver &D,
       // subsequent options conflict then emit warning diagnostic.
       if (HonorINFs && HonorNaNs && !AssociativeMath && !ReciprocalMath &&
           SignedZeros && TrappingMath && RoundingFPMath && !ApproxFunc &&
-          DenormalFPMath == llvm::DenormalMode::getIEEE() &&
-          DenormalFP32Math == llvm::DenormalMode::getIEEE() &&
           FPContract.equals("off"))
         // OK: Current Arg doesn't conflict with -ffp-model=strict
         ;
@@ -7262,15 +7246,10 @@ void Clang::ConstructJob(Compilation &C, const JobAction &JA,
   Args.addOptInFlag(CmdArgs, options::OPT_frelaxed_template_template_args,
                     options::OPT_fno_relaxed_template_template_args);
 
-  // -fsized-deallocation is on by default in C++14 onwards and otherwise off
-  // by default.
-  if (Arg *A = Args.getLastArg(options::OPT_fsized_deallocation,
-                               options::OPT_fno_sized_deallocation)) {
-    if (A->getOption().matches(options::OPT_fno_sized_deallocation))
-      CmdArgs.push_back("-fno-sized-deallocation");
-    else
-      CmdArgs.push_back("-fsized-deallocation");
-  }
+  // -fsized-deallocation is off by default, as it is an ABI-breaking change for
+  // most platforms.
+  Args.addOptInFlag(CmdArgs, options::OPT_fsized_deallocation,
+                    options::OPT_fno_sized_deallocation);
 
   // -faligned-allocation is on by default in C++17 onwards and otherwise off
   // by default.
@@ -7509,6 +7488,9 @@ void Clang::ConstructJob(Compilation &C, const JobAction &JA,
   Args.addOptInFlag(CmdArgs, options::OPT_fsafe_buffer_usage_suggestions,
                     options::OPT_fno_safe_buffer_usage_suggestions);
 
+  Args.addOptInFlag(CmdArgs, options::OPT_fexperimental_late_parse_attributes,
+                    options::OPT_fno_experimental_late_parse_attributes);
+
   // Setup statistics file output.
   SmallString<128> StatsFile = getStatsFileName(Args, Output, Input, D);
   if (!StatsFile.empty()) {

diff --git a/clang/lib/Driver/ToolChains/CommonArgs.cpp b/clang/lib/Driver/ToolChains/CommonArgs.cpp
@@ -737,7 +737,7 @@ bool tools::isTLSDESCEnabled(const ToolChain &TC,
   StringRef V = A->getValue();
   bool SupportedArgument = false, EnableTLSDESC = false;
   bool Unsupported = !Triple.isOSBinFormatELF();
-  if (Triple.isRISCV()) {
+  if (Triple.isLoongArch() || Triple.isRISCV()) {
     SupportedArgument = V == "desc" || V == "trad";
     EnableTLSDESC = V == "desc";
   } else if (Triple.isX86()) {
@@ -1191,118 +1191,10 @@ bool tools::addOpenMPRuntime(const Compilation &C, ArgStringList &CmdArgs,
   return true;
 }
 
-/// Determines if --whole-archive is active in the list of arguments.
-static bool isWholeArchivePresent(const ArgList &Args) {
-  bool WholeArchiveActive = false;
-  for (auto *Arg : Args.filtered(options::OPT_Wl_COMMA)) {
-    if (Arg) {
-      for (StringRef ArgValue : Arg->getValues()) {
-        if (ArgValue == "--whole-archive")
-          WholeArchiveActive = true;
-        if (ArgValue == "--no-whole-archive")
-          WholeArchiveActive = false;
-      }
-    }
-  }
-
-  return WholeArchiveActive;
-}
-
-/// Determine if driver is invoked to create a shared object library (-static)
-static bool isSharedLinkage(const ArgList &Args) {
-  return Args.hasArg(options::OPT_shared);
-}
-
-/// Determine if driver is invoked to create a static object library (-shared)
-static bool isStaticLinkage(const ArgList &Args) {
-  return Args.hasArg(options::OPT_static);
-}
-
-/// Add Fortran runtime libs for MSVC
-static void addFortranRuntimeLibsMSVC(const ArgList &Args,
-                                      llvm::opt::ArgStringList &CmdArgs) {
-  unsigned RTOptionID = options::OPT__SLASH_MT;
-  if (auto *rtl = Args.getLastArg(options::OPT_fms_runtime_lib_EQ)) {
-    RTOptionID = llvm::StringSwitch<unsigned>(rtl->getValue())
-                     .Case("static", options::OPT__SLASH_MT)
-                     .Case("static_dbg", options::OPT__SLASH_MTd)
-                     .Case("dll", options::OPT__SLASH_MD)
-                     .Case("dll_dbg", options::OPT__SLASH_MDd)
-                     .Default(options::OPT__SLASH_MT);
-  }
-  switch (RTOptionID) {
-  case options::OPT__SLASH_MT:
-    CmdArgs.push_back("/WHOLEARCHIVE:Fortran_main.static.lib");
-    break;
-  case options::OPT__SLASH_MTd:
-    CmdArgs.push_back("/WHOLEARCHIVE:Fortran_main.static_dbg.lib");
-    break;
-  case options::OPT__SLASH_MD:
-    CmdArgs.push_back("/WHOLEARCHIVE:Fortran_main.dynamic.lib");
-    break;
-  case options::OPT__SLASH_MDd:
-    CmdArgs.push_back("/WHOLEARCHIVE:Fortran_main.dynamic_dbg.lib");
-    break;
-  }
-}
-
-// Add FortranMain runtime lib
-static void addFortranMain(const ToolChain &TC, const ArgList &Args,
-                           llvm::opt::ArgStringList &CmdArgs) {
-  // 0. Shared-library linkage
-  // If we are attempting to link a library, we should not add
-  // -lFortran_main.a to the link line, as the `main` symbol is not
-  // required for a library and should also be provided by one of
-  // the translation units of the code that this shared library
-  // will be linked against eventually.
-  if (isSharedLinkage(Args) || isStaticLinkage(Args)) {
-    return;
-  }
-
-  // 1. MSVC
-  if (TC.getTriple().isKnownWindowsMSVCEnvironment()) {
-    addFortranRuntimeLibsMSVC(Args, CmdArgs);
-    return;
-  }
-
-  // 2. GNU and similar
-  const Driver &D = TC.getDriver();
-  const char *FortranMainLinkFlag = "-lFortran_main";
-
-  // Warn if the user added `-lFortran_main` - this library is an implementation
-  // detail of Flang and should be handled automaticaly by the driver.
-  for (const char *arg : CmdArgs) {
-    if (strncmp(arg, FortranMainLinkFlag, strlen(FortranMainLinkFlag)) == 0)
-      D.Diag(diag::warn_drv_deprecated_custom)
-          << FortranMainLinkFlag
-          << "see the Flang driver documentation for correct usage";
-  }
-
-  // The --whole-archive option needs to be part of the link line to make
-  // sure that the main() function from Fortran_main.a is pulled in by the
-  // linker. However, it shouldn't be used if it's already active.
-  // TODO: Find an equivalent of `--whole-archive` for Darwin and AIX.
-  if (!isWholeArchivePresent(Args) && !TC.getTriple().isMacOSX() &&
-      !TC.getTriple().isOSAIX()) {
-    CmdArgs.push_back("--whole-archive");
-    CmdArgs.push_back(FortranMainLinkFlag);
-    CmdArgs.push_back("--no-whole-archive");
-    return;
-  }
-
-  CmdArgs.push_back(FortranMainLinkFlag);
-}
-
 /// Add Fortran runtime libs
 void tools::addFortranRuntimeLibs(const ToolChain &TC, const ArgList &Args,
                                   llvm::opt::ArgStringList &CmdArgs) {
-  // 1. Link FortranMain
-  // FortranMain depends on FortranRuntime, so needs to be listed first. If
-  // -fno-fortran-main has been passed, skip linking Fortran_main.a
-  if (!Args.hasArg(options::OPT_no_fortran_main))
-    addFortranMain(TC, Args, CmdArgs);
-
-  // 2. Link FortranRuntime and FortranDecimal
+  // Link FortranRuntime and FortranDecimal
   // These are handled earlier on Windows by telling the frontend driver to
   // add the correct libraries to link against as dependents in the object
   // file.

diff --git a/clang/lib/Driver/ToolChains/Darwin.cpp b/clang/lib/Driver/ToolChains/Darwin.cpp
@@ -2912,68 +2912,16 @@ static bool sdkSupportsBuiltinModules(const Darwin::DarwinPlatformKind &TargetPl
   }
 }
 
-static inline llvm::VersionTuple
-sizedDeallocMinVersion(llvm::Triple::OSType OS) {
-  switch (OS) {
-  default:
-    break;
-  case llvm::Triple::Darwin:
-  case llvm::Triple::MacOSX: // Earliest supporting version is 10.12.
-    return llvm::VersionTuple(10U, 12U);
-  case llvm::Triple::IOS:
-  case llvm::Triple::TvOS: // Earliest supporting version is 10.0.0.
-    return llvm::VersionTuple(10U);
-  case llvm::Triple::WatchOS: // Earliest supporting version is 3.0.0.
-    return llvm::VersionTuple(3U);
-  }
-
-  llvm_unreachable("Unexpected OS");
-}
-
-bool Darwin::isSizedDeallocationUnavailable() const {
-  llvm::Triple::OSType OS;
-
-  if (isTargetMacCatalyst())
-    return TargetVersion < sizedDeallocMinVersion(llvm::Triple::MacOSX);
-  switch (TargetPlatform) {
-  case MacOS: // Earlier than 10.12.
-    OS = llvm::Triple::MacOSX;
-    break;
-  case IPhoneOS:
-    OS = llvm::Triple::IOS;
-    break;
-  case TvOS: // Earlier than 10.0.
-    OS = llvm::Triple::TvOS;
-    break;
-  case WatchOS: // Earlier than 3.0.
-    OS = llvm::Triple::WatchOS;
-    break;
-  case DriverKit:
-  case XROS:
-    // Always available.
-    return false;
-  }
-
-  return TargetVersion < sizedDeallocMinVersion(OS);
-}
-
-void Darwin::addClangTargetOptions(
-    const llvm::opt::ArgList &DriverArgs, llvm::opt::ArgStringList &CC1Args,
-    Action::OffloadKind DeviceOffloadKind) const {
+void Darwin::addClangTargetOptions(const llvm::opt::ArgList &DriverArgs,
+                                   llvm::opt::ArgStringList &CC1Args,
+                                   Action::OffloadKind DeviceOffloadKind) const {
   // Pass "-faligned-alloc-unavailable" only when the user hasn't manually
   // enabled or disabled aligned allocations.
   if (!DriverArgs.hasArgNoClaim(options::OPT_faligned_allocation,
                                 options::OPT_fno_aligned_allocation) &&
       isAlignedAllocationUnavailable())
     CC1Args.push_back("-faligned-alloc-unavailable");
 
-  // Pass "-fno-sized-deallocation" only when the user hasn't manually enabled
-  // or disabled sized deallocations.
-  if (!DriverArgs.hasArgNoClaim(options::OPT_fsized_deallocation,
-                                options::OPT_fno_sized_deallocation) &&
-      isSizedDeallocationUnavailable())
-    CC1Args.push_back("-fno-sized-deallocation");
-
   addClangCC1ASTargetOptions(DriverArgs, CC1Args);
 
   // Enable compatibility mode for NSItemProviderCompletionHandler in

diff --git a/clang/lib/Driver/ToolChains/Darwin.h b/clang/lib/Driver/ToolChains/Darwin.h
@@ -511,10 +511,6 @@ class LLVM_LIBRARY_VISIBILITY Darwin : public MachO {
   /// targeting.
   bool isAlignedAllocationUnavailable() const;
 
-  /// Return true if c++14 sized deallocation functions are not implemented in
-  /// the c++ standard library of the deployment target we are targeting.
-  bool isSizedDeallocationUnavailable() const;
-
   void addClangTargetOptions(const llvm::opt::ArgList &DriverArgs,
                              llvm::opt::ArgStringList &CC1Args,
                              Action::OffloadKind DeviceOffloadKind) const override;

diff --git a/clang/lib/Driver/ToolChains/Flang.cpp b/clang/lib/Driver/ToolChains/Flang.cpp
@@ -282,7 +282,6 @@ static void processVSRuntimeLibrary(const ToolChain &TC, const ArgList &Args,
   assert(TC.getTriple().isKnownWindowsMSVCEnvironment() &&
          "can only add VS runtime library on Windows!");
   // if -fno-fortran-main has been passed, skip linking Fortran_main.a
-  bool LinkFortranMain = !Args.hasArg(options::OPT_no_fortran_main);
   if (TC.getTriple().isKnownWindowsMSVCEnvironment()) {
     CmdArgs.push_back(Args.MakeArgString(
         "--dependent-lib=" + TC.getCompilerRTBasename(Args, "builtins")));
@@ -300,26 +299,20 @@ static void processVSRuntimeLibrary(const ToolChain &TC, const ArgList &Args,
   case options::OPT__SLASH_MT:
     CmdArgs.push_back("-D_MT");
     CmdArgs.push_back("--dependent-lib=libcmt");
-    if (LinkFortranMain)
-      CmdArgs.push_back("--dependent-lib=Fortran_main.static.lib");
     CmdArgs.push_back("--dependent-lib=FortranRuntime.static.lib");
     CmdArgs.push_back("--dependent-lib=FortranDecimal.static.lib");
     break;
   case options::OPT__SLASH_MTd:
     CmdArgs.push_back("-D_MT");
     CmdArgs.push_back("-D_DEBUG");
     CmdArgs.push_back("--dependent-lib=libcmtd");
-    if (LinkFortranMain)
-      CmdArgs.push_back("--dependent-lib=Fortran_main.static_dbg.lib");
     CmdArgs.push_back("--dependent-lib=FortranRuntime.static_dbg.lib");
     CmdArgs.push_back("--dependent-lib=FortranDecimal.static_dbg.lib");
     break;
   case options::OPT__SLASH_MD:
     CmdArgs.push_back("-D_MT");
     CmdArgs.push_back("-D_DLL");
     CmdArgs.push_back("--dependent-lib=msvcrt");
-    if (LinkFortranMain)
-      CmdArgs.push_back("--dependent-lib=Fortran_main.dynamic.lib");
     CmdArgs.push_back("--dependent-lib=FortranRuntime.dynamic.lib");
     CmdArgs.push_back("--dependent-lib=FortranDecimal.dynamic.lib");
     break;
@@ -328,8 +321,6 @@ static void processVSRuntimeLibrary(const ToolChain &TC, const ArgList &Args,
     CmdArgs.push_back("-D_DEBUG");
     CmdArgs.push_back("-D_DLL");
     CmdArgs.push_back("--dependent-lib=msvcrtd");
-    if (LinkFortranMain)
-      CmdArgs.push_back("--dependent-lib=Fortran_main.dynamic_dbg.lib");
     CmdArgs.push_back("--dependent-lib=FortranRuntime.dynamic_dbg.lib");
     CmdArgs.push_back("--dependent-lib=FortranDecimal.dynamic_dbg.lib");
     break;

diff --git a/clang/lib/Driver/ToolChains/Gnu.cpp b/clang/lib/Driver/ToolChains/Gnu.cpp
@@ -1796,9 +1796,7 @@ selectRISCVMultilib(const MultilibSet &RISCVMultilibSet, StringRef Arch,
       }
       auto &MLConfigISAInfo = *MLConfigParseResult;
 
-      const llvm::RISCVISAInfo::OrderedExtensionMap &MLConfigArchExts =
-          MLConfigISAInfo->getExtensions();
-      for (auto MLConfigArchExt : MLConfigArchExts) {
+      for (auto &MLConfigArchExt : MLConfigISAInfo->getExtensions()) {
         auto ExtName = MLConfigArchExt.first;
         NewMultilib.flag(Twine("-", ExtName).str());
 

diff --git a/clang/lib/Driver/ToolChains/ZOS.cpp b/clang/lib/Driver/ToolChains/ZOS.cpp
@@ -36,12 +36,6 @@ void ZOS::addClangTargetOptions(const ArgList &DriverArgs,
   if (!DriverArgs.hasArgNoClaim(options::OPT_faligned_allocation,
                                 options::OPT_fno_aligned_allocation))
     CC1Args.push_back("-faligned-alloc-unavailable");
-
-  // Pass "-fno-sized-deallocation" only when the user hasn't manually enabled
-  // or disabled sized deallocations.
-  if (!DriverArgs.hasArgNoClaim(options::OPT_fsized_deallocation,
-                                options::OPT_fno_sized_deallocation))
-    CC1Args.push_back("-fno-sized-deallocation");
 }
 
 void zos::Assembler::ConstructJob(Compilation &C, const JobAction &JA,

diff --git a/clang/lib/Format/FormatToken.h b/clang/lib/Format/FormatToken.h
@@ -1623,10 +1623,10 @@ struct AdditionalKeywords {
   IdentifierInfo *kw_then;
 
   /// Returns \c true if \p Tok is a keyword or an identifier.
-  bool isWordLike(const FormatToken &Tok) const {
+  bool isWordLike(const FormatToken &Tok, bool IsVerilog = true) const {
     // getIdentifierinfo returns non-null for keywords as well as identifiers.
     return Tok.Tok.getIdentifierInfo() &&
-           !Tok.isOneOf(kw_verilogHash, kw_verilogHashHash, kw_apostrophe);
+           (!IsVerilog || !isVerilogKeywordSymbol(Tok));
   }
 
   /// Returns \c true if \p Tok is a true JavaScript identifier, returns
@@ -1755,6 +1755,10 @@ struct AdditionalKeywords {
     }
   }
 
+  bool isVerilogKeywordSymbol(const FormatToken &Tok) const {
+    return Tok.isOneOf(kw_verilogHash, kw_verilogHashHash, kw_apostrophe);
+  }
+
   bool isVerilogWordOperator(const FormatToken &Tok) const {
     return Tok.isOneOf(kw_before, kw_intersect, kw_dist, kw_iff, kw_inside,
                        kw_with);

diff --git a/clang/lib/Format/TokenAnnotator.cpp b/clang/lib/Format/TokenAnnotator.cpp
@@ -4780,9 +4780,14 @@ bool TokenAnnotator::spaceRequiredBefore(const AnnotatedLine &Line,
   if (Left.Finalized)
     return Right.hasWhitespaceBefore();
 
+  const bool IsVerilog = Style.isVerilog();
+  assert(!IsVerilog || !IsCpp);
+
   // Never ever merge two words.
-  if (Keywords.isWordLike(Right) && Keywords.isWordLike(Left))
+  if (Keywords.isWordLike(Right, IsVerilog) &&
+      Keywords.isWordLike(Left, IsVerilog)) {
     return true;
+  }
 
   // Leave a space between * and /* to avoid C4138 `comment end` found outside
   // of comment.
@@ -4834,10 +4839,8 @@ bool TokenAnnotator::spaceRequiredBefore(const AnnotatedLine &Line,
         Right.is(TT_TemplateOpener)) {
       return true;
     }
-    if (Left.is(tok::identifier) && Right.is(tok::numeric_constant) &&
-        Right.TokenText[0] == '.') {
-      return false;
-    }
+    if (Left.Tok.getIdentifierInfo() && Right.is(tok::numeric_constant))
+      return Right.TokenText[0] != '.';
   } else if (Style.isProto()) {
     if (Right.is(tok::period) &&
         Left.isOneOf(Keywords.kw_optional, Keywords.kw_required,
@@ -5065,12 +5068,10 @@ bool TokenAnnotator::spaceRequiredBefore(const AnnotatedLine &Line,
         Right.is(TT_TemplateOpener)) {
       return true;
     }
-  } else if (Style.isVerilog()) {
+  } else if (IsVerilog) {
     // An escaped identifier ends with whitespace.
-    if (Style.isVerilog() && Left.is(tok::identifier) &&
-        Left.TokenText[0] == '\\') {
+    if (Left.is(tok::identifier) && Left.TokenText[0] == '\\')
       return true;
-    }
     // Add space between things in a primitive's state table unless in a
     // transition like `(0?)`.
     if ((Left.is(TT_VerilogTableItem) &&
@@ -5266,21 +5267,11 @@ bool TokenAnnotator::spaceRequiredBefore(const AnnotatedLine &Line,
     return true;
   }
   if (Left.is(TT_UnaryOperator)) {
-    if (Right.isNot(tok::l_paren)) {
-      // The alternative operators for ~ and ! are "compl" and "not".
-      // If they are used instead, we do not want to combine them with
-      // the token to the right, unless that is a left paren.
-      if (Left.is(tok::exclaim) && Left.TokenText == "not")
-        return true;
-      if (Left.is(tok::tilde) && Left.TokenText == "compl")
-        return true;
-      // Lambda captures allow for a lone &, so "&]" needs to be properly
-      // handled.
-      if (Left.is(tok::amp) && Right.is(tok::r_square))
-        return Style.SpacesInSquareBrackets;
-    }
-    return (Style.SpaceAfterLogicalNot && Left.is(tok::exclaim)) ||
-           Right.is(TT_BinaryOperator);
+    // Lambda captures allow for a lone &, so "&]" needs to be properly
+    // handled.
+    if (Left.is(tok::amp) && Right.is(tok::r_square))
+      return Style.SpacesInSquareBrackets;
+    return Style.SpaceAfterLogicalNot && Left.is(tok::exclaim);
   }
 
   // If the next token is a binary operator or a selector name, we have

diff --git a/clang/lib/Format/UnwrappedLineParser.cpp b/clang/lib/Format/UnwrappedLineParser.cpp
@@ -3944,8 +3944,11 @@ void UnwrappedLineParser::parseRecord(bool ParseAsExpr) {
     switch (FormatTok->Tok.getKind()) {
     case tok::l_paren:
       // We can have macros in between 'class' and the class name.
-      if (!IsNonMacroIdentifier(Previous))
+      if (!IsNonMacroIdentifier(Previous) ||
+          // e.g. `struct macro(a) S { int i; };`
+          Previous->Previous == &InitialToken) {
         parseParens();
+      }
       break;
     case tok::coloncolon:
       break;

diff --git a/clang/lib/Format/WhitespaceManager.cpp b/clang/lib/Format/WhitespaceManager.cpp
@@ -128,11 +128,14 @@ const tooling::Replacements &WhitespaceManager::generateReplacements() {
 void WhitespaceManager::calculateLineBreakInformation() {
   Changes[0].PreviousEndOfTokenColumn = 0;
   Change *LastOutsideTokenChange = &Changes[0];
-  for (unsigned i = 1, e = Changes.size(); i != e; ++i) {
+  for (unsigned I = 1, e = Changes.size(); I != e; ++I) {
+    auto &C = Changes[I];
+    auto &P = Changes[I - 1];
+    auto &PrevTokLength = P.TokenLength;
     SourceLocation OriginalWhitespaceStart =
-        Changes[i].OriginalWhitespaceRange.getBegin();
+        C.OriginalWhitespaceRange.getBegin();
     SourceLocation PreviousOriginalWhitespaceEnd =
-        Changes[i - 1].OriginalWhitespaceRange.getEnd();
+        P.OriginalWhitespaceRange.getEnd();
     unsigned OriginalWhitespaceStartOffset =
         SourceMgr.getFileOffset(OriginalWhitespaceStart);
     unsigned PreviousOriginalWhitespaceEndOffset =
@@ -167,31 +170,28 @@ void WhitespaceManager::calculateLineBreakInformation() {
     // line of the token.
     auto NewlinePos = Text.find_first_of('\n');
     if (NewlinePos == StringRef::npos) {
-      Changes[i - 1].TokenLength = OriginalWhitespaceStartOffset -
-                                   PreviousOriginalWhitespaceEndOffset +
-                                   Changes[i].PreviousLinePostfix.size() +
-                                   Changes[i - 1].CurrentLinePrefix.size();
+      PrevTokLength = OriginalWhitespaceStartOffset -
+                      PreviousOriginalWhitespaceEndOffset +
+                      C.PreviousLinePostfix.size() + P.CurrentLinePrefix.size();
+      if (!P.IsInsideToken)
+        PrevTokLength = std::min(PrevTokLength, P.Tok->ColumnWidth);
     } else {
-      Changes[i - 1].TokenLength =
-          NewlinePos + Changes[i - 1].CurrentLinePrefix.size();
+      PrevTokLength = NewlinePos + P.CurrentLinePrefix.size();
     }
 
     // If there are multiple changes in this token, sum up all the changes until
     // the end of the line.
-    if (Changes[i - 1].IsInsideToken && Changes[i - 1].NewlinesBefore == 0) {
-      LastOutsideTokenChange->TokenLength +=
-          Changes[i - 1].TokenLength + Changes[i - 1].Spaces;
-    } else {
-      LastOutsideTokenChange = &Changes[i - 1];
-    }
+    if (P.IsInsideToken && P.NewlinesBefore == 0)
+      LastOutsideTokenChange->TokenLength += PrevTokLength + P.Spaces;
+    else
+      LastOutsideTokenChange = &P;
 
-    Changes[i].PreviousEndOfTokenColumn =
-        Changes[i - 1].StartOfTokenColumn + Changes[i - 1].TokenLength;
+    C.PreviousEndOfTokenColumn = P.StartOfTokenColumn + PrevTokLength;
 
-    Changes[i - 1].IsTrailingComment =
-        (Changes[i].NewlinesBefore > 0 || Changes[i].Tok->is(tok::eof) ||
-         (Changes[i].IsInsideToken && Changes[i].Tok->is(tok::comment))) &&
-        Changes[i - 1].Tok->is(tok::comment) &&
+    P.IsTrailingComment =
+        (C.NewlinesBefore > 0 || C.Tok->is(tok::eof) ||
+         (C.IsInsideToken && C.Tok->is(tok::comment))) &&
+        P.Tok->is(tok::comment) &&
         // FIXME: This is a dirty hack. The problem is that
         // BreakableLineCommentSection does comment reflow changes and here is
         // the aligning of trailing comments. Consider the case where we reflow

diff --git a/clang/lib/Frontend/CompilerInvocation.cpp b/clang/lib/Frontend/CompilerInvocation.cpp
@@ -2549,6 +2549,7 @@ static const auto &getFrontendActionTable() {
       {frontend::DumpTokens, OPT_dump_tokens},
       {frontend::EmitAssembly, OPT_S},
       {frontend::EmitBC, OPT_emit_llvm_bc},
+      {frontend::EmitCIR, OPT_emit_cir},
       {frontend::EmitHTML, OPT_emit_html},
       {frontend::EmitLLVM, OPT_emit_llvm},
       {frontend::EmitLLVMOnly, OPT_emit_llvm_only},
@@ -2891,6 +2892,8 @@ static bool ParseFrontendArgs(FrontendOptions &Opts, ArgList &Args,
   if (Opts.ProgramAction != frontend::GenerateModule && Opts.IsSystemModule)
     Diags.Report(diag::err_drv_argument_only_allowed_with) << "-fsystem-module"
                                                            << "-emit-module";
+  if (Args.hasArg(OPT_fclangir) || Args.hasArg(OPT_emit_cir))
+    Opts.UseClangIRPipeline = true;
 
   if (Args.hasArg(OPT_aux_target_cpu))
     Opts.AuxTargetCPU = std::string(Args.getLastArgValue(OPT_aux_target_cpu));
@@ -4337,6 +4340,7 @@ static bool isStrictlyPreprocessorAction(frontend::ActionKind Action) {
   case frontend::ASTView:
   case frontend::EmitAssembly:
   case frontend::EmitBC:
+  case frontend::EmitCIR:
   case frontend::EmitHTML:
   case frontend::EmitLLVM:
   case frontend::EmitLLVMOnly:

diff --git a/clang/lib/Frontend/FrontendActions.cpp b/clang/lib/Frontend/FrontendActions.cpp
@@ -272,14 +272,10 @@ bool GenerateModuleInterfaceAction::BeginSourceFileAction(
 std::unique_ptr<ASTConsumer>
 GenerateModuleInterfaceAction::CreateASTConsumer(CompilerInstance &CI,
                                                  StringRef InFile) {
-  CI.getHeaderSearchOpts().ModulesSkipDiagnosticOptions = true;
-  CI.getHeaderSearchOpts().ModulesSkipHeaderSearchPaths = true;
-  CI.getHeaderSearchOpts().ModulesSkipPragmaDiagnosticMappings = true;
-
-  std::vector<std::unique_ptr<ASTConsumer>> Consumers =
-      CreateMultiplexConsumer(CI, InFile);
-  if (Consumers.empty())
-    return nullptr;
+  std::vector<std::unique_ptr<ASTConsumer>> Consumers;
+  Consumers.push_back(std::make_unique<CXX20ModulesGenerator>(
+      CI.getPreprocessor(), CI.getModuleCache(),
+      CI.getFrontendOpts().OutputFile));
 
   if (CI.getFrontendOpts().GenReducedBMI &&
       !CI.getFrontendOpts().ModuleOutputPath.empty()) {

diff --git a/clang/lib/Frontend/InitPreprocessor.cpp b/clang/lib/Frontend/InitPreprocessor.cpp
@@ -703,7 +703,7 @@ static void InitializeCPlusPlusFeatureTestMacros(const LangOptions &LangOpts,
     Builder.defineMacro("__cpp_nested_namespace_definitions", "201411L");
     Builder.defineMacro("__cpp_variadic_using", "201611L");
     Builder.defineMacro("__cpp_aggregate_bases", "201603L");
-    Builder.defineMacro("__cpp_structured_bindings", "201606L");
+    Builder.defineMacro("__cpp_structured_bindings", "202403L");
     Builder.defineMacro("__cpp_nontype_template_args",
                         "201411L"); // (not latest)
     Builder.defineMacro("__cpp_fold_expressions", "201603L");

diff --git a/clang/lib/FrontendTool/ExecuteCompilerInvocation.cpp b/clang/lib/FrontendTool/ExecuteCompilerInvocation.cpp
@@ -53,6 +53,8 @@ CreateFrontendBaseAction(CompilerInstance &CI) {
   case DumpTokens:             return std::make_unique<DumpTokensAction>();
   case EmitAssembly:           return std::make_unique<EmitAssemblyAction>();
   case EmitBC:                 return std::make_unique<EmitBCAction>();
+  case EmitCIR:
+    llvm_unreachable("CIR suppport not built into clang");
   case EmitHTML:               return std::make_unique<HTMLPrintAction>();
   case EmitLLVM:               return std::make_unique<EmitLLVMAction>();
   case EmitLLVMOnly:           return std::make_unique<EmitLLVMOnlyAction>();

diff --git a/clang/lib/Headers/CMakeLists.txt b/clang/lib/Headers/CMakeLists.txt
@@ -335,6 +335,10 @@ set(llvm_libc_wrapper_files
   llvm_libc_wrappers/time.h
 )
 
+set(zos_wrapper_files
+  zos_wrappers/builtins.h
+)
+
 include(GetClangResourceDir)
 get_clang_resource_dir(output_dir PREFIX ${LLVM_LIBRARY_OUTPUT_INTDIR}/.. SUBDIR include)
 set(out_files)
@@ -370,7 +374,7 @@ endfunction(clang_generate_header)
 
 # Copy header files from the source directory to the build directory
 foreach( f ${files} ${cuda_wrapper_files} ${cuda_wrapper_bits_files}
-           ${ppc_wrapper_files} ${openmp_wrapper_files} ${hlsl_files}
+           ${ppc_wrapper_files} ${openmp_wrapper_files} ${zos_wrapper_files} ${hlsl_files}
            ${llvm_libc_wrapper_files})
   copy_header_to_output_dir(${CMAKE_CURRENT_SOURCE_DIR} ${f})
 endforeach( f )
@@ -487,7 +491,7 @@ add_header_target("mips-resource-headers" "${mips_msa_files}")
 add_header_target("ppc-resource-headers" "${ppc_files};${ppc_wrapper_files}")
 add_header_target("ppc-htm-resource-headers" "${ppc_htm_files}")
 add_header_target("riscv-resource-headers" "${riscv_files};${riscv_generated_files}")
-add_header_target("systemz-resource-headers" "${systemz_files}")
+add_header_target("systemz-resource-headers" "${systemz_files};${zos_wrapper_files}")
 add_header_target("ve-resource-headers" "${ve_files}")
 add_header_target("webassembly-resource-headers" "${webassembly_files}")
 add_header_target("x86-resource-headers" "${x86_files}")
@@ -538,6 +542,11 @@ install(
   DESTINATION ${header_install_dir}/openmp_wrappers
   COMPONENT clang-resource-headers)
 
+install(
+  FILES ${zos_wrapper_files}
+  DESTINATION ${header_install_dir}/zos_wrappers
+  COMPONENT clang-resource-headers)
+
 #############################################################
 # Install rules for separate header lists
 install(
@@ -642,6 +651,12 @@ install(
   EXCLUDE_FROM_ALL
   COMPONENT systemz-resource-headers)
 
+install(
+  FILES ${zos_wrapper_files}
+  DESTINATION  ${header_install_dir}/zos_wrappers
+  EXCLUDE_FROM_ALL
+  COMPONENT systemz-resource-headers)
+
 install(
   FILES ${ve_files}
   DESTINATION ${header_install_dir}

diff --git a/clang/lib/Headers/builtins.h b/clang/lib/Headers/builtins.h
@@ -13,4 +13,7 @@
 #ifndef __BUILTINS_H
 #define __BUILTINS_H
 
+#if defined(__MVS__) && __has_include_next(<builtins.h>)
+#include_next <builtins.h>
+#endif /* __MVS__ */
 #endif /* __BUILTINS_H */
diff --git a/clang/lib/Headers/float.h b/clang/lib/Headers/float.h
@@ -10,6 +10,10 @@
 #ifndef __CLANG_FLOAT_H
 #define __CLANG_FLOAT_H
 
+#if defined(__MVS__) && __has_include_next(<float.h>)
+#include_next <float.h>
+#else
+
 /* If we're on MinGW, fall back to the system's float.h, which might have
  * additional definitions provided for Windows.
  * For more details see http://msdn.microsoft.com/en-us/library/y0ybw9fy.aspx
@@ -165,4 +169,5 @@
 #  define FLT16_TRUE_MIN    __FLT16_TRUE_MIN__
 #endif /* __STDC_WANT_IEC_60559_TYPES_EXT__ */
 
+#endif /* __MVS__ */
 #endif /* __CLANG_FLOAT_H */
diff --git a/clang/lib/Headers/inttypes.h b/clang/lib/Headers/inttypes.h
@@ -13,6 +13,9 @@
 #if !defined(_AIX) || !defined(_STD_TYPES_T)
 #define __CLANG_INTTYPES_H
 #endif
+#if defined(__MVS__) && __has_include_next(<inttypes.h>)
+#include_next <inttypes.h>
+#else
 
 #if defined(_MSC_VER) && _MSC_VER < 1800
 #error MSVC does not have inttypes.h prior to Visual Studio 2013
@@ -94,4 +97,5 @@
 #define SCNxFAST32 "x"
 #endif
 
+#endif /* __MVS__ */
 #endif /* __CLANG_INTTYPES_H */
diff --git a/clang/lib/Headers/iso646.h b/clang/lib/Headers/iso646.h
@@ -9,6 +9,9 @@
 
 #ifndef __ISO646_H
 #define __ISO646_H
+#if defined(__MVS__) && __has_include_next(<iso646.h>)
+#include_next <iso646.h>
+#else
 
 #ifndef __cplusplus
 #define and    &&
@@ -24,4 +27,5 @@
 #define xor_eq ^=
 #endif
 
+#endif /* __MVS__ */
 #endif /* __ISO646_H */
diff --git a/clang/lib/Headers/limits.h b/clang/lib/Headers/limits.h
@@ -9,6 +9,10 @@
 #ifndef __CLANG_LIMITS_H
 #define __CLANG_LIMITS_H
 
+#if defined(__MVS__) && __has_include_next(<limits.h>)
+#include_next <limits.h>
+#else
+
 /* The system's limits.h may, in turn, try to #include_next GCC's limits.h.
    Avert this #include_next madness. */
 #if defined __GNUC__ && !defined _GCC_LIMITS_H_
@@ -122,4 +126,5 @@
 #define ULONG_LONG_MAX (__LONG_LONG_MAX__*2ULL+1ULL)
 #endif
 
+#endif /* __MVS__ */
 #endif /* __CLANG_LIMITS_H */
diff --git a/clang/lib/Headers/stdalign.h b/clang/lib/Headers/stdalign.h
@@ -10,6 +10,10 @@
 #ifndef __STDALIGN_H
 #define __STDALIGN_H
 
+#if defined(__MVS__) && __has_include_next(<stdalign.h>)
+#include_next <stdalign.h>
+#else
+
 #if defined(__cplusplus) ||                                                    \
     (defined(__STDC_VERSION__) && __STDC_VERSION__ < 202311L)
 #ifndef __cplusplus
@@ -21,4 +25,5 @@
 #define __alignof_is_defined 1
 #endif /* __STDC_VERSION__ */
 
+#endif /* __MVS__ */
 #endif /* __STDALIGN_H */
diff --git a/clang/lib/Headers/stdarg.h b/clang/lib/Headers/stdarg.h
@@ -33,6 +33,16 @@
     defined(__need_va_arg) || defined(__need___va_copy) ||                     \
     defined(__need_va_copy)
 
+#if defined(__MVS__) && __has_include_next(<stdarg.h>)
+#define __STDARG_H
+#undef __need___va_list
+#undef __need_va_list
+#undef __need_va_arg
+#undef __need___va_copy
+#undef __need_va_copy
+#include_next <stdarg.h>
+
+#else
 #if !defined(__need___va_list) && !defined(__need_va_list) &&                  \
     !defined(__need_va_arg) && !defined(__need___va_copy) &&                   \
     !defined(__need_va_copy)
@@ -76,4 +86,6 @@
 #undef __need_va_copy
 #endif /* defined(__need_va_copy) */
 
+#endif /* __MVS__ */
+
 #endif