diff --git a/clang/lib/APINotes/APINotesReader.cpp b/clang/lib/APINotes/APINotesReader.cpp
diff --git a/clang/lib/AST/Decl.cpp b/clang/lib/AST/Decl.cpp
@@ -4534,7 +4534,7 @@ unsigned FunctionDecl::getODRHash() {
   }
 
   class ODRHash Hash;
-  Hash.AddFunctionDecl(this, /*SkipBody=*/shouldSkipCheckingODR());
+  Hash.AddFunctionDecl(this);
   setHasODRHash(true);
   ODRHash = Hash.CalculateHash();
   return ODRHash;

diff --git a/clang/lib/AST/DeclBase.cpp b/clang/lib/AST/DeclBase.cpp
@@ -1106,11 +1106,6 @@ bool Decl::isFromExplicitGlobalModule() const {
   return getOwningModule() && getOwningModule()->isExplicitGlobalModule();
 }
 
-bool Decl::shouldSkipCheckingODR() const {
-  return getASTContext().getLangOpts().SkipODRCheckInGMF &&
-         isFromExplicitGlobalModule();
-}
-
 static Decl::Kind getKind(const Decl *D) { return D->getKind(); }
 static Decl::Kind getKind(const DeclContext *DC) { return DC->getDeclKind(); }
 

diff --git a/clang/lib/AST/Interp/ByteCodeExprGen.cpp b/clang/lib/AST/Interp/ByteCodeExprGen.cpp
@@ -262,7 +262,7 @@ bool ByteCodeExprGen<Emitter>::VisitCastExpr(const CastExpr *CE) {
       return this->discard(SubExpr);
 
     std::optional<PrimType> FromT = classify(SubExpr->getType());
-    std::optional<PrimType> ToT = classifyPrim(CE->getType());
+    std::optional<PrimType> ToT = classify(CE->getType());
     if (!FromT || !ToT)
       return false;
 

diff --git a/clang/lib/AST/StmtProfile.cpp b/clang/lib/AST/StmtProfile.cpp
@@ -2071,13 +2071,31 @@ StmtProfiler::VisitLambdaExpr(const LambdaExpr *S) {
   }
 
   CXXRecordDecl *Lambda = S->getLambdaClass();
-  ID.AddInteger(Lambda->getODRHash());
-
   for (const auto &Capture : Lambda->captures()) {
     ID.AddInteger(Capture.getCaptureKind());
     if (Capture.capturesVariable())
       VisitDecl(Capture.getCapturedVar());
   }
+
+  // Profiling the body of the lambda may be dangerous during deserialization.
+  // So we'd like only to profile the signature here.
+  ODRHash Hasher;
+  // FIXME: We can't get the operator call easily by
+  // `CXXRecordDecl::getLambdaCallOperator()` if we're in deserialization.
+  // So we have to do something raw here.
+  for (auto *SubDecl : Lambda->decls()) {
+    FunctionDecl *Call = nullptr;
+    if (auto *FTD = dyn_cast<FunctionTemplateDecl>(SubDecl))
+      Call = FTD->getTemplatedDecl();
+    else if (auto *FD = dyn_cast<FunctionDecl>(SubDecl))
+      Call = FD;
+
+    if (!Call)
+      continue;
+
+    Hasher.AddFunctionDecl(Call, /*SkipBody=*/true);
+  }
+  ID.AddInteger(Hasher.CalculateHash());
 }
 
 void

diff --git a/clang/lib/Analysis/FlowSensitive/DataflowEnvironment.cpp b/clang/lib/Analysis/FlowSensitive/DataflowEnvironment.cpp
@@ -508,6 +508,11 @@ class ResultObjectVisitor : public RecursiveASTVisitor<ResultObjectVisitor> {
         isa<CXXStdInitializerListExpr>(E)) {
       return;
     }
+    if (auto *Op = dyn_cast<BinaryOperator>(E);
+        Op && Op->getOpcode() == BO_Cmp) {
+      // Builtin `<=>` returns a `std::strong_ordering` object.
+      return;
+    }
 
     if (auto *InitList = dyn_cast<InitListExpr>(E)) {
       if (!InitList->isSemanticForm())

diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -18194,7 +18194,8 @@ Value *CodeGenFunction::EmitHLSLBuiltinExpr(unsigned BuiltinID,
     Value *Op0 = EmitScalarExpr(E->getArg(0));
     return Builder.CreateIntrinsic(
         /*ReturnType=*/llvm::Type::getInt1Ty(getLLVMContext()),
-        Intrinsic::dx_any, ArrayRef<Value *>{Op0}, nullptr, "dx.any");
+        CGM.getHLSLRuntime().getAnyIntrinsic(), ArrayRef<Value *>{Op0}, nullptr,
+        "hlsl.any");
   }
   case Builtin::BI__builtin_hlsl_elementwise_clamp: {
     Value *OpX = EmitScalarExpr(E->getArg(0));

diff --git a/clang/lib/CodeGen/CGCall.cpp b/clang/lib/CodeGen/CGCall.cpp
@@ -4694,11 +4694,11 @@ 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();
@@ -4707,14 +4707,15 @@ 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 =
@@ -729,6 +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;
   }
 
@@ -765,9 +774,16 @@ 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();
@@ -992,6 +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.
@@ -1281,17 +1299,8 @@ 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/CGDecl.cpp b/clang/lib/CodeGen/CGDecl.cpp
@@ -2216,8 +2216,11 @@ void CodeGenFunction::pushDestroyAndDeferDeactivation(
 void CodeGenFunction::pushDestroyAndDeferDeactivation(
     CleanupKind cleanupKind, Address addr, QualType type, Destroyer *destroyer,
     bool useEHCleanupForArray) {
-  pushCleanupAndDeferDeactivation<DestroyObject>(
-      cleanupKind, addr, type, destroyer, 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) {

diff --git a/clang/lib/CodeGen/CGHLSLRuntime.h b/clang/lib/CodeGen/CGHLSLRuntime.h
@@ -73,6 +73,7 @@ class CGHLSLRuntime {
   //===----------------------------------------------------------------------===//
 
   GENERATE_HLSL_INTRINSIC_FUNCTION(All, all)
+  GENERATE_HLSL_INTRINSIC_FUNCTION(Any, any)
   GENERATE_HLSL_INTRINSIC_FUNCTION(ThreadId, thread_id)
 
   //===----------------------------------------------------------------------===//

diff --git a/clang/lib/CodeGen/CGRecordLayoutBuilder.cpp b/clang/lib/CodeGen/CGRecordLayoutBuilder.cpp
@@ -41,10 +41,11 @@ namespace {
 ///   contains enough information to determine where the runs break.  Microsoft
 ///   and Itanium follow different rules and use different codepaths.
 /// * It is desired that, when possible, bitfields use the appropriate iN type
-///   when lowered to llvm types.  For example unsigned x : 24 gets lowered to
+///   when lowered to llvm types. For example unsigned x : 24 gets lowered to
 ///   i24.  This isn't always possible because i24 has storage size of 32 bit
-///   and if it is possible to use that extra byte of padding we must use
-///   [i8 x 3] instead of i24.  The function clipTailPadding does this.
+///   and if it is possible to use that extra byte of padding we must use [i8 x
+///   3] instead of i24. This is computed when accumulating bitfields in
+///   accumulateBitfields.
 ///   C++ examples that require clipping:
 ///   struct { int a : 24; char b; }; // a must be clipped, b goes at offset 3
 ///   struct A { int a : 24; ~A(); }; // a must be clipped because:
@@ -62,11 +63,7 @@ namespace {
 ///   that the tail padding is not used in the complete class.) However,
 ///   because LLVM reads from the complete type it can generate incorrect code
 ///   if we do not clip the tail padding off of the bitfield in the complete
-///   layout.  This introduces a somewhat awkward extra unnecessary clip stage.
-///   The location of the clip is stored internally as a sentinel of type
-///   SCISSOR.  If LLVM were updated to read base types (which it probably
-///   should because locations of things such as VBases are bogus in the llvm
-///   type anyway) then we could eliminate the SCISSOR.
+///   layout.
 /// * Itanium allows nearly empty primary virtual bases.  These bases don't get
 ///   get their own storage because they're laid out as part of another base
 ///   or at the beginning of the structure.  Determining if a VBase actually
@@ -200,9 +197,7 @@ struct CGRecordLowering {
                      const CXXRecordDecl *Query) const;
   void calculateZeroInit();
   CharUnits calculateTailClippingOffset(bool isNonVirtualBaseType) const;
-  /// Lowers bitfield storage types to I8 arrays for bitfields with tail
-  /// padding that is or can potentially be used.
-  void clipTailPadding();
+  void checkBitfieldClipping() const;
   /// Determines if we need a packed llvm struct.
   void determinePacked(bool NVBaseType);
   /// Inserts padding everywhere it's needed.
@@ -305,7 +300,7 @@ void CGRecordLowering::lower(bool NVBaseType) {
   }
   llvm::stable_sort(Members);
   Members.push_back(StorageInfo(Size, getIntNType(8)));
-  clipTailPadding();
+  checkBitfieldClipping();
   determinePacked(NVBaseType);
   insertPadding();
   Members.pop_back();
@@ -531,6 +526,7 @@ CGRecordLowering::accumulateBitFields(bool isNonVirtualBaseType,
   // available padding characters.
   RecordDecl::field_iterator BestEnd = Begin;
   CharUnits BestEndOffset;
+  bool BestClipped; // Whether the representation must be in a byte array.
 
   for (;;) {
     // AtAlignedBoundary is true iff Field is the (potential) start of a new
@@ -593,10 +589,9 @@ CGRecordLowering::accumulateBitFields(bool isNonVirtualBaseType,
         // this is the best seen so far.
         BestEnd = Field;
         BestEndOffset = BeginOffset + AccessSize;
-        if (Types.getCodeGenOpts().FineGrainedBitfieldAccesses)
-          // Fine-grained access, so no merging of spans.
-          InstallBest = true;
-        else if (!BitSizeSinceBegin)
+        // Assume clipped until proven not below.
+        BestClipped = true;
+        if (!BitSizeSinceBegin)
           // A zero-sized initial span -- this will install nothing and reset
           // for another.
           InstallBest = true;
@@ -624,6 +619,12 @@ CGRecordLowering::accumulateBitFields(bool isNonVirtualBaseType,
             // The access unit is not at a naturally aligned offset within the
             // structure.
             InstallBest = true;
+
+          if (InstallBest && BestEnd == Field)
+            // We're installing the first span, whose clipping was presumed
+            // above. Compute it correctly.
+            if (getSize(Type) == AccessSize)
+              BestClipped = false;
         }
 
         if (!InstallBest) {
@@ -656,11 +657,15 @@ CGRecordLowering::accumulateBitFields(bool isNonVirtualBaseType,
             // access unit.
             BestEndOffset = BeginOffset + TypeSize;
             BestEnd = Field;
+            BestClipped = false;
           }
 
           if (Barrier)
             // The next field is a barrier that we cannot merge across.
             InstallBest = true;
+          else if (Types.getCodeGenOpts().FineGrainedBitfieldAccesses)
+            // Fine-grained access, so no merging of spans.
+            InstallBest = true;
           else
             // Otherwise, we're not installing. Update the bit size
             // of the current span to go all the way to LimitOffset, which is
@@ -679,7 +684,17 @@ CGRecordLowering::accumulateBitFields(bool isNonVirtualBaseType,
         // Add the storage member for the access unit to the record. The
         // bitfields get the offset of their storage but come afterward and
         // remain there after a stable sort.
-        llvm::Type *Type = getIntNType(Context.toBits(AccessSize));
+        llvm::Type *Type;
+        if (BestClipped) {
+          assert(getSize(getIntNType(Context.toBits(AccessSize))) >
+                     AccessSize &&
+                 "Clipped access need not be clipped");
+          Type = getByteArrayType(AccessSize);
+        } else {
+          Type = getIntNType(Context.toBits(AccessSize));
+          assert(getSize(Type) == AccessSize &&
+                 "Unclipped access must be clipped");
+        }
         Members.push_back(StorageInfo(BeginOffset, Type));
         for (; Begin != BestEnd; ++Begin)
           if (!Begin->isZeroLengthBitField(Context))
@@ -934,32 +949,21 @@ void CGRecordLowering::calculateZeroInit() {
   }
 }
 
-void CGRecordLowering::clipTailPadding() {
-  std::vector<MemberInfo>::iterator Prior = Members.begin();
-  CharUnits Tail = getSize(Prior->Data);
-  for (std::vector<MemberInfo>::iterator Member = Prior + 1,
-                                         MemberEnd = Members.end();
-       Member != MemberEnd; ++Member) {
+// Verify accumulateBitfields computed the correct storage representations.
+void CGRecordLowering::checkBitfieldClipping() const {
+#ifndef NDEBUG
+  auto Tail = CharUnits::Zero();
+  for (const auto &M : Members) {
     // Only members with data and the scissor can cut into tail padding.
-    if (!Member->Data && Member->Kind != MemberInfo::Scissor)
+    if (!M.Data && M.Kind != MemberInfo::Scissor)
       continue;
-    if (Member->Offset < Tail) {
-      assert(Prior->Kind == MemberInfo::Field &&
-             "Only storage fields have tail padding!");
-      if (!Prior->FD || Prior->FD->isBitField())
-        Prior->Data = getByteArrayType(bitsToCharUnits(llvm::alignTo(
-            cast<llvm::IntegerType>(Prior->Data)->getIntegerBitWidth(), 8)));
-      else {
-        assert(Prior->FD->hasAttr<NoUniqueAddressAttr>() &&
-               "should not have reused this field's tail padding");
-        Prior->Data = getByteArrayType(
-            Context.getTypeInfoDataSizeInChars(Prior->FD->getType()).Width);
-      }
-    }
-    if (Member->Data)
-      Prior = Member;
-    Tail = Prior->Offset + getSize(Prior->Data);
+
+    assert(M.Offset >= Tail && "Bitfield access unit is not clipped");
+    Tail = M.Offset;
+    if (M.Data)
+      Tail += getSize(M.Data);
   }
+#endif
 }
 
 void CGRecordLowering::determinePacked(bool NVBaseType) {

diff --git a/clang/lib/CodeGen/CodeGenFunction.h b/clang/lib/CodeGen/CodeGenFunction.h
@@ -957,7 +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

diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -3952,9 +3952,20 @@ bool CodeGenModule::shouldEmitFunction(GlobalDecl GD) {
   // behavior may break ABI compatibility of the current unit.
   if (const Module *M = F->getOwningModule();
       M && M->getTopLevelModule()->isNamedModule() &&
-      getContext().getCurrentNamedModule() != M->getTopLevelModule() &&
-      !F->hasAttr<AlwaysInlineAttr>())
-    return false;
+      getContext().getCurrentNamedModule() != M->getTopLevelModule()) {
+    // There are practices to mark template member function as always-inline
+    // and mark the template as extern explicit instantiation but not give
+    // the definition for member function. So we have to emit the function
+    // from explicitly instantiation with always-inline.
+    //
+    // See https://github.com/llvm/llvm-project/issues/86893 for details.
+    //
+    // TODO: Maybe it is better to give it a warning if we call a non-inline
+    // function from other module units which is marked as always-inline.
+    if (!F->isTemplateInstantiation() || !F->hasAttr<AlwaysInlineAttr>()) {
+      return false;
+    }
+  }
 
   if (F->hasAttr<NoInlineAttr>())
     return false;

diff --git a/clang/lib/Parse/ParseDecl.cpp b/clang/lib/Parse/ParseDecl.cpp
@@ -2380,10 +2380,6 @@ Parser::DeclGroupPtrTy Parser::ParseDeclGroup(ParsingDeclSpec &DS,
       if (getLangOpts().CPlusPlus23) {
         auto &LastRecord = Actions.ExprEvalContexts.back();
         LastRecord.InLifetimeExtendingContext = true;
-
-        // Materialize non-`cv void` prvalue temporaries in discarded
-        // expressions. These materialized temporaries may be lifetime-extented.
-        LastRecord.InMaterializeTemporaryObjectContext = true;
       }
 
       if (getLangOpts().OpenMP)

diff --git a/clang/lib/Sema/CMakeLists.txt b/clang/lib/Sema/CMakeLists.txt
@@ -1,5 +1,6 @@
 set(LLVM_LINK_COMPONENTS
   Core
+  Demangle
   FrontendHLSL
   FrontendOpenMP
   MC

diff --git a/clang/lib/Sema/SemaDeclAttr.cpp b/clang/lib/Sema/SemaDeclAttr.cpp
@@ -45,6 +45,7 @@
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/STLForwardCompat.h"
 #include "llvm/ADT/StringExtras.h"
+#include "llvm/Demangle/Demangle.h"
 #include "llvm/IR/Assumptions.h"
 #include "llvm/MC/MCSectionMachO.h"
 #include "llvm/Support/Error.h"
@@ -1983,6 +1984,36 @@ static void handleWeakRefAttr(Sema &S, Decl *D, const ParsedAttr &AL) {
   D->addAttr(::new (S.Context) WeakRefAttr(S.Context, AL));
 }
 
+// Mark alias/ifunc target as used. Due to name mangling, we look up the
+// demangled name ignoring parameters (not supported by microsoftDemangle
+// https://github.com/llvm/llvm-project/issues/88825). This should handle the
+// majority of use cases while leaving namespace scope names unmarked.
+static void markUsedForAliasOrIfunc(Sema &S, Decl *D, const ParsedAttr &AL,
+                                    StringRef Str) {
+  std::unique_ptr<char, llvm::FreeDeleter> Demangled;
+  if (S.getASTContext().getCXXABIKind() != TargetCXXABI::Microsoft)
+    Demangled.reset(llvm::itaniumDemangle(Str, /*ParseParams=*/false));
+  std::unique_ptr<MangleContext> MC(S.Context.createMangleContext());
+  SmallString<256> Name;
+
+  const DeclarationNameInfo Target(
+      &S.Context.Idents.get(Demangled ? Demangled.get() : Str), AL.getLoc());
+  LookupResult LR(S, Target, Sema::LookupOrdinaryName);
+  if (S.LookupName(LR, S.TUScope)) {
+    for (NamedDecl *ND : LR) {
+      if (MC->shouldMangleDeclName(ND)) {
+        llvm::raw_svector_ostream Out(Name);
+        Name.clear();
+        MC->mangleName(GlobalDecl(ND), Out);
+      } else {
+        Name = ND->getIdentifier()->getName();
+      }
+      if (Name == Str)
+        ND->markUsed(S.Context);
+    }
+  }
+}
+
 static void handleIFuncAttr(Sema &S, Decl *D, const ParsedAttr &AL) {
   StringRef Str;
   if (!S.checkStringLiteralArgumentAttr(AL, 0, Str))
@@ -1995,6 +2026,7 @@ static void handleIFuncAttr(Sema &S, Decl *D, const ParsedAttr &AL) {
     return;
   }
 
+  markUsedForAliasOrIfunc(S, D, AL, Str);
   D->addAttr(::new (S.Context) IFuncAttr(S.Context, AL, Str));
 }
 
@@ -2029,17 +2061,7 @@ static void handleAliasAttr(Sema &S, Decl *D, const ParsedAttr &AL) {
     }
   }
 
-  // Mark target used to prevent unneeded-internal-declaration warnings.
-  if (!S.LangOpts.CPlusPlus) {
-    // FIXME: demangle Str for C++, as the attribute refers to the mangled
-    // linkage name, not the pre-mangled identifier.
-    const DeclarationNameInfo target(&S.Context.Idents.get(Str), AL.getLoc());
-    LookupResult LR(S, target, Sema::LookupOrdinaryName);
-    if (S.LookupQualifiedName(LR, S.getCurLexicalContext()))
-      for (NamedDecl *ND : LR)
-        ND->markUsed(S.Context);
-  }
-
+  markUsedForAliasOrIfunc(S, D, AL, Str);
   D->addAttr(::new (S.Context) AliasAttr(S.Context, AL, Str));
 }
 

diff --git a/clang/lib/Sema/SemaExpr.cpp b/clang/lib/Sema/SemaExpr.cpp
@@ -6314,7 +6314,6 @@ ExprResult Sema::BuildCXXDefaultArgExpr(SourceLocation CallLoc,
       // Pass down lifetime extending flag, and collect temporaries in
       // CreateMaterializeTemporaryExpr when we rewrite the call argument.
       keepInLifetimeExtendingContext();
-      keepInMaterializeTemporaryObjectContext();
       EnsureImmediateInvocationInDefaultArgs Immediate(*this);
       ExprResult Res;
       runWithSufficientStackSpace(CallLoc, [&] {
@@ -7740,7 +7739,8 @@ ExprResult Sema::BuildResolvedCallExpr(Expr *Fn, NamedDecl *NDecl,
   }
 
   if (CXXMethodDecl *Method = dyn_cast_or_null<CXXMethodDecl>(FDecl))
-    if (Method->isImplicitObjectMemberFunction())
+    if (!isa<RequiresExprBodyDecl>(CurContext) &&
+        Method->isImplicitObjectMemberFunction())
       return ExprError(Diag(LParenLoc, diag::err_member_call_without_object)
                        << Fn->getSourceRange() << 0);
 
@@ -18679,9 +18679,9 @@ void Sema::PopExpressionEvaluationContext() {
   // Append the collected materialized temporaries into previous context before
   // exit if the previous also is a lifetime extending context.
   auto &PrevRecord = ExprEvalContexts[ExprEvalContexts.size() - 2];
-  if (getLangOpts().CPlusPlus23 && isInLifetimeExtendingContext() &&
-      PrevRecord.InLifetimeExtendingContext && !ExprEvalContexts.empty()) {
-    auto &PrevRecord = ExprEvalContexts[ExprEvalContexts.size() - 2];
+  if (getLangOpts().CPlusPlus23 && Rec.InLifetimeExtendingContext &&
+      PrevRecord.InLifetimeExtendingContext &&
+      !Rec.ForRangeLifetimeExtendTemps.empty()) {
     PrevRecord.ForRangeLifetimeExtendTemps.append(
         Rec.ForRangeLifetimeExtendTemps);
   }

diff --git a/clang/lib/Sema/SemaExprCXX.cpp b/clang/lib/Sema/SemaExprCXX.cpp
@@ -8429,7 +8429,7 @@ ExprResult Sema::IgnoredValueConversions(Expr *E) {
     // unnecessary temporary objects. If we skip this step, IR generation is
     // able to synthesize the storage for itself in the aggregate case, and
     // adding the extra node to the AST is just clutter.
-    if (isInMaterializeTemporaryObjectContext() && getLangOpts().CPlusPlus17 &&
+    if (isInLifetimeExtendingContext() && getLangOpts().CPlusPlus17 &&
         E->isPRValue() && !E->getType()->isVoidType()) {
       ExprResult Res = TemporaryMaterializationConversion(E);
       if (Res.isInvalid())

diff --git a/clang/lib/Sema/SemaOverload.cpp b/clang/lib/Sema/SemaOverload.cpp
@@ -14395,9 +14395,16 @@ Sema::CreateOverloadedUnaryOp(SourceLocation OpLoc, UnaryOperatorKind Opc,
   ArrayRef<Expr *> ArgsArray(Args, NumArgs);
 
   if (Input->isTypeDependent()) {
+    ExprValueKind VK = ExprValueKind::VK_PRValue;
+    // [C++26][expr.unary.op][expr.pre.incr]
+    // The * operator yields an lvalue of type
+    // The pre/post increment operators yied an lvalue.
+    if (Opc == UO_PreDec || Opc == UO_PreInc || Opc == UO_Deref)
+      VK = VK_LValue;
+
     if (Fns.empty())
-      return UnaryOperator::Create(Context, Input, Opc, Context.DependentTy,
-                                   VK_PRValue, OK_Ordinary, OpLoc, false,
+      return UnaryOperator::Create(Context, Input, Opc, Context.DependentTy, VK,
+                                   OK_Ordinary, OpLoc, false,
                                    CurFPFeatureOverrides());
 
     CXXRecordDecl *NamingClass = nullptr; // lookup ignores member operators
@@ -14406,7 +14413,7 @@ Sema::CreateOverloadedUnaryOp(SourceLocation OpLoc, UnaryOperatorKind Opc,
     if (Fn.isInvalid())
       return ExprError();
     return CXXOperatorCallExpr::Create(Context, Op, Fn.get(), ArgsArray,
-                                       Context.DependentTy, VK_PRValue, OpLoc,
+                                       Context.DependentTy, VK, OpLoc,
                                        CurFPFeatureOverrides());
   }
 

diff --git a/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp b/clang/lib/Sema/SemaTemplateInstantiateDecl.cpp
@@ -5483,7 +5483,6 @@ void Sema::InstantiateVariableInitializer(
         *this, Sema::ExpressionEvaluationContext::PotentiallyEvaluated, Var);
 
     keepInLifetimeExtendingContext();
-    keepInMaterializeTemporaryObjectContext();
     // Instantiate the initializer.
     ExprResult Init;
 

diff --git a/clang/lib/Sema/TreeTransform.h b/clang/lib/Sema/TreeTransform.h
@@ -4177,7 +4177,6 @@ ExprResult TreeTransform<Derived>::TransformInitializer(Expr *Init,
       getSema(), EnterExpressionEvaluationContext::InitList,
       Construct->isListInitialization());
 
-  getSema().keepInLifetimeExtendingContext();
   getSema().keepInLifetimeExtendingContext();
   SmallVector<Expr*, 8> NewArgs;
   bool ArgChanged = false;
@@ -8752,10 +8751,6 @@ TreeTransform<Derived>::TransformCXXForRangeStmt(CXXForRangeStmt *S) {
   if (getSema().getLangOpts().CPlusPlus23) {
     auto &LastRecord = getSema().ExprEvalContexts.back();
     LastRecord.InLifetimeExtendingContext = true;
-
-    // Materialize non-`cv void` prvalue temporaries in discarded
-    // expressions. These materialized temporaries may be lifetime-extented.
-    LastRecord.InMaterializeTemporaryObjectContext = true;
   }
   StmtResult Init =
       S->getInit() ? getDerived().TransformStmt(S->getInit()) : StmtResult();

diff --git a/clang/lib/Serialization/ASTReader.cpp b/clang/lib/Serialization/ASTReader.cpp
@@ -915,10 +915,9 @@ ASTSelectorLookupTrait::ReadKey(const unsigned char* d, unsigned) {
   using namespace llvm::support;
 
   SelectorTable &SelTable = Reader.getContext().Selectors;
-  unsigned N =
-      endian::readNext<uint16_t, llvm::endianness::little, unaligned>(d);
+  unsigned N = endian::readNext<uint16_t, llvm::endianness::little>(d);
   const IdentifierInfo *FirstII = Reader.getLocalIdentifier(
-      F, endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d));
+      F, endian::readNext<uint32_t, llvm::endianness::little>(d));
   if (N == 0)
     return SelTable.getNullarySelector(FirstII);
   else if (N == 1)
@@ -928,7 +927,7 @@ ASTSelectorLookupTrait::ReadKey(const unsigned char* d, unsigned) {
   Args.push_back(FirstII);
   for (unsigned I = 1; I != N; ++I)
     Args.push_back(Reader.getLocalIdentifier(
-        F, endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d)));
+        F, endian::readNext<uint32_t, llvm::endianness::little>(d)));
 
   return SelTable.getSelector(N, Args.data());
 }
@@ -941,11 +940,11 @@ ASTSelectorLookupTrait::ReadData(Selector, const unsigned char* d,
   data_type Result;
 
   Result.ID = Reader.getGlobalSelectorID(
-      F, endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d));
+      F, endian::readNext<uint32_t, llvm::endianness::little>(d));
   unsigned FullInstanceBits =
-      endian::readNext<uint16_t, llvm::endianness::little, unaligned>(d);
+      endian::readNext<uint16_t, llvm::endianness::little>(d);
   unsigned FullFactoryBits =
-      endian::readNext<uint16_t, llvm::endianness::little, unaligned>(d);
+      endian::readNext<uint16_t, llvm::endianness::little>(d);
   Result.InstanceBits = FullInstanceBits & 0x3;
   Result.InstanceHasMoreThanOneDecl = (FullInstanceBits >> 2) & 0x1;
   Result.FactoryBits = FullFactoryBits & 0x3;
@@ -956,16 +955,14 @@ ASTSelectorLookupTrait::ReadData(Selector, const unsigned char* d,
   // Load instance methods
   for (unsigned I = 0; I != NumInstanceMethods; ++I) {
     if (ObjCMethodDecl *Method = Reader.GetLocalDeclAs<ObjCMethodDecl>(
-            F,
-            endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d)))
+            F, endian::readNext<uint32_t, llvm::endianness::little>(d)))
       Result.Instance.push_back(Method);
   }
 
   // Load factory methods
   for (unsigned I = 0; I != NumFactoryMethods; ++I) {
     if (ObjCMethodDecl *Method = Reader.GetLocalDeclAs<ObjCMethodDecl>(
-            F,
-            endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d)))
+            F, endian::readNext<uint32_t, llvm::endianness::little>(d)))
       Result.Factory.push_back(Method);
   }
 
@@ -1009,8 +1006,7 @@ static bool readBit(unsigned &Bits) {
 IdentID ASTIdentifierLookupTrait::ReadIdentifierID(const unsigned char *d) {
   using namespace llvm::support;
 
-  unsigned RawID =
-      endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d);
+  unsigned RawID = endian::readNext<uint32_t, llvm::endianness::little>(d);
   return Reader.getGlobalIdentifierID(F, RawID >> 1);
 }
 
@@ -1028,8 +1024,7 @@ IdentifierInfo *ASTIdentifierLookupTrait::ReadData(const internal_key_type& k,
                                                    unsigned DataLen) {
   using namespace llvm::support;
 
-  unsigned RawID =
-      endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d);
+  unsigned RawID = endian::readNext<uint32_t, llvm::endianness::little>(d);
   bool IsInteresting = RawID & 0x01;
 
   // Wipe out the "is interesting" bit.
@@ -1053,9 +1048,8 @@ IdentifierInfo *ASTIdentifierLookupTrait::ReadData(const internal_key_type& k,
   }
 
   unsigned ObjCOrBuiltinID =
-      endian::readNext<uint16_t, llvm::endianness::little, unaligned>(d);
-  unsigned Bits =
-      endian::readNext<uint16_t, llvm::endianness::little, unaligned>(d);
+      endian::readNext<uint16_t, llvm::endianness::little>(d);
+  unsigned Bits = endian::readNext<uint16_t, llvm::endianness::little>(d);
   bool CPlusPlusOperatorKeyword = readBit(Bits);
   bool HasRevertedTokenIDToIdentifier = readBit(Bits);
   bool Poisoned = readBit(Bits);
@@ -1084,7 +1078,7 @@ IdentifierInfo *ASTIdentifierLookupTrait::ReadData(const internal_key_type& k,
   // definition.
   if (HadMacroDefinition) {
     uint32_t MacroDirectivesOffset =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d);
+        endian::readNext<uint32_t, llvm::endianness::little>(d);
     DataLen -= 4;
 
     Reader.addPendingMacro(II, &F, MacroDirectivesOffset);
@@ -1098,8 +1092,7 @@ IdentifierInfo *ASTIdentifierLookupTrait::ReadData(const internal_key_type& k,
     SmallVector<uint32_t, 4> DeclIDs;
     for (; DataLen > 0; DataLen -= 4)
       DeclIDs.push_back(Reader.getGlobalDeclID(
-          F,
-          endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d)));
+          F, endian::readNext<uint32_t, llvm::endianness::little>(d)));
     Reader.SetGloballyVisibleDecls(II, DeclIDs);
   }
 
@@ -1169,7 +1162,7 @@ ASTDeclContextNameLookupTrait::ReadFileRef(const unsigned char *&d) {
   using namespace llvm::support;
 
   uint32_t ModuleFileID =
-      endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d);
+      endian::readNext<uint32_t, llvm::endianness::little>(d);
   return Reader.getLocalModuleFile(F, ModuleFileID);
 }
 
@@ -1189,18 +1182,15 @@ ASTDeclContextNameLookupTrait::ReadKey(const unsigned char *d, unsigned) {
   case DeclarationName::CXXLiteralOperatorName:
   case DeclarationName::CXXDeductionGuideName:
     Data = (uint64_t)Reader.getLocalIdentifier(
-        F, endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d));
+        F, endian::readNext<uint32_t, llvm::endianness::little>(d));
     break;
   case DeclarationName::ObjCZeroArgSelector:
   case DeclarationName::ObjCOneArgSelector:
   case DeclarationName::ObjCMultiArgSelector:
-    Data =
-        (uint64_t)Reader
-            .getLocalSelector(
-                F,
-                endian::readNext<uint32_t, llvm::endianness::little, unaligned>(
-                    d))
-            .getAsOpaquePtr();
+    Data = (uint64_t)Reader
+               .getLocalSelector(
+                   F, endian::readNext<uint32_t, llvm::endianness::little>(d))
+               .getAsOpaquePtr();
     break;
   case DeclarationName::CXXOperatorName:
     Data = *d++; // OverloadedOperatorKind
@@ -1223,8 +1213,7 @@ void ASTDeclContextNameLookupTrait::ReadDataInto(internal_key_type,
   using namespace llvm::support;
 
   for (unsigned NumDecls = DataLen / 4; NumDecls; --NumDecls) {
-    uint32_t LocalID =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d);
+    uint32_t LocalID = endian::readNext<uint32_t, llvm::endianness::little>(d);
     Val.insert(Reader.getGlobalDeclID(F, LocalID));
   }
 }
@@ -2033,10 +2022,9 @@ HeaderFileInfoTrait::ReadKey(const unsigned char *d, unsigned) {
   using namespace llvm::support;
 
   internal_key_type ikey;
-  ikey.Size =
-      off_t(endian::readNext<uint64_t, llvm::endianness::little, unaligned>(d));
-  ikey.ModTime = time_t(
-      endian::readNext<uint64_t, llvm::endianness::little, unaligned>(d));
+  ikey.Size = off_t(endian::readNext<uint64_t, llvm::endianness::little>(d));
+  ikey.ModTime =
+      time_t(endian::readNext<uint64_t, llvm::endianness::little>(d));
   ikey.Filename = (const char *)d;
   ikey.Imported = true;
   return ikey;
@@ -2064,9 +2052,9 @@ HeaderFileInfoTrait::ReadData(internal_key_ref key, const unsigned char *d,
   HFI.DirInfo = (Flags >> 1) & 0x07;
   HFI.IndexHeaderMapHeader = Flags & 0x01;
   HFI.ControllingMacroID = Reader.getGlobalIdentifierID(
-      M, endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d));
+      M, endian::readNext<uint32_t, llvm::endianness::little>(d));
   if (unsigned FrameworkOffset =
-          endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d)) {
+          endian::readNext<uint32_t, llvm::endianness::little>(d)) {
     // The framework offset is 1 greater than the actual offset,
     // since 0 is used as an indicator for "no framework name".
     StringRef FrameworkName(FrameworkStrings + FrameworkOffset - 1);
@@ -2077,7 +2065,7 @@ HeaderFileInfoTrait::ReadData(internal_key_ref key, const unsigned char *d,
          "Wrong data length in HeaderFileInfo deserialization");
   while (d != End) {
     uint32_t LocalSMID =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d);
+        endian::readNext<uint32_t, llvm::endianness::little>(d);
     auto HeaderRole = static_cast<ModuleMap::ModuleHeaderRole>(LocalSMID & 7);
     LocalSMID >>= 3;
 
@@ -4085,9 +4073,8 @@ void ASTReader::ReadModuleOffsetMap(ModuleFile &F) const {
     // how it goes...
     using namespace llvm::support;
     ModuleKind Kind = static_cast<ModuleKind>(
-        endian::readNext<uint8_t, llvm::endianness::little, unaligned>(Data));
-    uint16_t Len =
-        endian::readNext<uint16_t, llvm::endianness::little, unaligned>(Data);
+        endian::readNext<uint8_t, llvm::endianness::little>(Data));
+    uint16_t Len = endian::readNext<uint16_t, llvm::endianness::little>(Data);
     StringRef Name = StringRef((const char*)Data, Len);
     Data += Len;
     ModuleFile *OM = (Kind == MK_PrebuiltModule || Kind == MK_ExplicitModule ||
@@ -4103,21 +4090,21 @@ void ASTReader::ReadModuleOffsetMap(ModuleFile &F) const {
     }
 
     SourceLocation::UIntTy SLocOffset =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(Data);
+        endian::readNext<uint32_t, llvm::endianness::little>(Data);
     uint32_t IdentifierIDOffset =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(Data);
+        endian::readNext<uint32_t, llvm::endianness::little>(Data);
     uint32_t MacroIDOffset =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(Data);
+        endian::readNext<uint32_t, llvm::endianness::little>(Data);
     uint32_t PreprocessedEntityIDOffset =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(Data);
+        endian::readNext<uint32_t, llvm::endianness::little>(Data);
     uint32_t SubmoduleIDOffset =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(Data);
+        endian::readNext<uint32_t, llvm::endianness::little>(Data);
     uint32_t SelectorIDOffset =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(Data);
+        endian::readNext<uint32_t, llvm::endianness::little>(Data);
     uint32_t DeclIDOffset =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(Data);
+        endian::readNext<uint32_t, llvm::endianness::little>(Data);
     uint32_t TypeIndexOffset =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(Data);
+        endian::readNext<uint32_t, llvm::endianness::little>(Data);
 
     auto mapOffset = [&](uint32_t Offset, uint32_t BaseOffset,
                          RemapBuilder &Remap) {
@@ -9798,7 +9785,7 @@ void ASTReader::finishPendingActions() {
             !NonConstDefn->isLateTemplateParsed() &&
             // We only perform ODR checks for decls not in the explicit
             // global module fragment.
-            !FD->shouldSkipCheckingODR() &&
+            !shouldSkipCheckingODR(FD) &&
             FD->getODRHash() != NonConstDefn->getODRHash()) {
           if (!isa<CXXMethodDecl>(FD)) {
             PendingFunctionOdrMergeFailures[FD].push_back(NonConstDefn);

diff --git a/clang/lib/Serialization/ASTReaderDecl.cpp b/clang/lib/Serialization/ASTReaderDecl.cpp
@@ -826,7 +826,7 @@ void ASTDeclReader::VisitEnumDecl(EnumDecl *ED) {
       Reader.mergeDefinitionVisibility(OldDef, ED);
       // We don't want to check the ODR hash value for declarations from global
       // module fragment.
-      if (!ED->shouldSkipCheckingODR() &&
+      if (!shouldSkipCheckingODR(ED) &&
           OldDef->getODRHash() != ED->getODRHash())
         Reader.PendingEnumOdrMergeFailures[OldDef].push_back(ED);
     } else {
@@ -868,7 +868,7 @@ void ASTDeclReader::VisitRecordDecl(RecordDecl *RD) {
   VisitRecordDeclImpl(RD);
   // We should only reach here if we're in C/Objective-C. There is no
   // global module fragment.
-  assert(!RD->shouldSkipCheckingODR());
+  assert(!shouldSkipCheckingODR(RD));
   RD->setODRHash(Record.readInt());
 
   // Maintain the invariant of a redeclaration chain containing only
@@ -2155,7 +2155,7 @@ void ASTDeclReader::MergeDefinitionData(
   }
 
   // We don't want to check ODR for decls in the global module fragment.
-  if (MergeDD.Definition->shouldSkipCheckingODR())
+  if (shouldSkipCheckingODR(MergeDD.Definition))
     return;
 
   if (D->getODRHash() != MergeDD.ODRHash) {
@@ -3530,7 +3530,7 @@ ASTDeclReader::FindExistingResult ASTDeclReader::findExisting(NamedDecl *D) {
   // same template specialization into the same CXXRecordDecl.
   auto MergedDCIt = Reader.MergedDeclContexts.find(D->getLexicalDeclContext());
   if (MergedDCIt != Reader.MergedDeclContexts.end() &&
-      !D->shouldSkipCheckingODR() && MergedDCIt->second == D->getDeclContext())
+      !shouldSkipCheckingODR(D) && MergedDCIt->second == D->getDeclContext())
     Reader.PendingOdrMergeChecks.push_back(D);
 
   return FindExistingResult(Reader, D, /*Existing=*/nullptr,

diff --git a/clang/lib/Serialization/ASTWriter.cpp b/clang/lib/Serialization/ASTWriter.cpp
@@ -6188,7 +6188,7 @@ void ASTRecordWriter::AddCXXDefinitionData(const CXXRecordDecl *D) {
 
   BitsPacker DefinitionBits;
 
-  bool ShouldSkipCheckingODR = D->shouldSkipCheckingODR();
+  bool ShouldSkipCheckingODR = shouldSkipCheckingODR(D);
   DefinitionBits.addBit(ShouldSkipCheckingODR);
 
 #define FIELD(Name, Width, Merge)                                              \

diff --git a/clang/lib/Serialization/ASTWriterDecl.cpp b/clang/lib/Serialization/ASTWriterDecl.cpp
@@ -526,7 +526,7 @@ void ASTDeclWriter::VisitEnumDecl(EnumDecl *D) {
   BitsPacker EnumDeclBits;
   EnumDeclBits.addBits(D->getNumPositiveBits(), /*BitWidth=*/8);
   EnumDeclBits.addBits(D->getNumNegativeBits(), /*BitWidth=*/8);
-  bool ShouldSkipCheckingODR = D->shouldSkipCheckingODR();
+  bool ShouldSkipCheckingODR = shouldSkipCheckingODR(D);
   EnumDeclBits.addBit(ShouldSkipCheckingODR);
   EnumDeclBits.addBit(D->isScoped());
   EnumDeclBits.addBit(D->isScopedUsingClassTag());
@@ -552,7 +552,7 @@ void ASTDeclWriter::VisitEnumDecl(EnumDecl *D) {
       !D->isTopLevelDeclInObjCContainer() &&
       !CXXRecordDecl::classofKind(D->getKind()) &&
       !D->getIntegerTypeSourceInfo() && !D->getMemberSpecializationInfo() &&
-      !needsAnonymousDeclarationNumber(D) && !D->shouldSkipCheckingODR() &&
+      !needsAnonymousDeclarationNumber(D) && !shouldSkipCheckingODR(D) &&
       D->getDeclName().getNameKind() == DeclarationName::Identifier)
     AbbrevToUse = Writer.getDeclEnumAbbrev();
 
@@ -718,7 +718,7 @@ void ASTDeclWriter::VisitFunctionDecl(FunctionDecl *D) {
   // FIXME: stable encoding
   FunctionDeclBits.addBits(llvm::to_underlying(D->getLinkageInternal()), 3);
   FunctionDeclBits.addBits((uint32_t)D->getStorageClass(), /*BitWidth=*/3);
-  bool ShouldSkipCheckingODR = D->shouldSkipCheckingODR();
+  bool ShouldSkipCheckingODR = shouldSkipCheckingODR(D);
   FunctionDeclBits.addBit(ShouldSkipCheckingODR);
   FunctionDeclBits.addBit(D->isInlineSpecified());
   FunctionDeclBits.addBit(D->isInlined());
@@ -1559,7 +1559,7 @@ void ASTDeclWriter::VisitCXXMethodDecl(CXXMethodDecl *D) {
       D->getFirstDecl() == D->getMostRecentDecl() && !D->isInvalidDecl() &&
       !D->hasAttrs() && !D->isTopLevelDeclInObjCContainer() &&
       D->getDeclName().getNameKind() == DeclarationName::Identifier &&
-      !D->shouldSkipCheckingODR() && !D->hasExtInfo() &&
+      !shouldSkipCheckingODR(D) && !D->hasExtInfo() &&
       !D->isExplicitlyDefaulted()) {
     if (D->getTemplatedKind() == FunctionDecl::TK_NonTemplate ||
         D->getTemplatedKind() == FunctionDecl::TK_FunctionTemplate ||

diff --git a/clang/lib/Serialization/GlobalModuleIndex.cpp b/clang/lib/Serialization/GlobalModuleIndex.cpp
@@ -89,10 +89,8 @@ class IdentifierIndexReaderTrait {
   static std::pair<unsigned, unsigned>
   ReadKeyDataLength(const unsigned char*& d) {
     using namespace llvm::support;
-    unsigned KeyLen =
-        endian::readNext<uint16_t, llvm::endianness::little, unaligned>(d);
-    unsigned DataLen =
-        endian::readNext<uint16_t, llvm::endianness::little, unaligned>(d);
+    unsigned KeyLen = endian::readNext<uint16_t, llvm::endianness::little>(d);
+    unsigned DataLen = endian::readNext<uint16_t, llvm::endianness::little>(d);
     return std::make_pair(KeyLen, DataLen);
   }
 
@@ -113,8 +111,7 @@ class IdentifierIndexReaderTrait {
 
     data_type Result;
     while (DataLen > 0) {
-      unsigned ID =
-          endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d);
+      unsigned ID = endian::readNext<uint32_t, llvm::endianness::little>(d);
       Result.push_back(ID);
       DataLen -= 4;
     }
@@ -514,8 +511,7 @@ namespace {
       // The first bit indicates whether this identifier is interesting.
       // That's all we care about.
       using namespace llvm::support;
-      unsigned RawID =
-          endian::readNext<uint32_t, llvm::endianness::little, unaligned>(d);
+      unsigned RawID = endian::readNext<uint32_t, llvm::endianness::little>(d);
       bool IsInteresting = RawID & 0x01;
       return std::make_pair(k, IsInteresting);
     }

diff --git a/clang/lib/Serialization/MultiOnDiskHashTable.h b/clang/lib/Serialization/MultiOnDiskHashTable.h
@@ -200,11 +200,11 @@ template<typename Info> class MultiOnDiskHashTable {
     storage_type Ptr = Data;
 
     uint32_t BucketOffset =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(Ptr);
+        endian::readNext<uint32_t, llvm::endianness::little>(Ptr);
 
     // Read the list of overridden files.
     uint32_t NumFiles =
-        endian::readNext<uint32_t, llvm::endianness::little, unaligned>(Ptr);
+        endian::readNext<uint32_t, llvm::endianness::little>(Ptr);
     // FIXME: Add a reserve() to TinyPtrVector so that we don't need to make
     // an additional copy.
     llvm::SmallVector<file_type, 16> OverriddenFiles;

diff --git a/clang/lib/StaticAnalyzer/Checkers/BuiltinFunctionChecker.cpp b/clang/lib/StaticAnalyzer/Checkers/BuiltinFunctionChecker.cpp
@@ -6,7 +6,11 @@
 //
 //===----------------------------------------------------------------------===//
 //
-// This checker evaluates clang builtin functions.
+// This checker evaluates "standalone" clang builtin functions that are not
+// just special-cased variants of well-known non-builtin functions.
+// Builtin functions like __builtin_memcpy and __builtin_alloca should be
+// evaluated by the same checker that handles their non-builtin variant to
+// ensure that the two variants are handled consistently.
 //
 //===----------------------------------------------------------------------===//
 
@@ -80,25 +84,6 @@ bool BuiltinFunctionChecker::evalCall(const CallEvent &Call,
     return true;
   }
 
-  case Builtin::BI__builtin_alloca_with_align:
-  case Builtin::BI__builtin_alloca: {
-    SValBuilder &SVB = C.getSValBuilder();
-    const loc::MemRegionVal R =
-        SVB.getAllocaRegionVal(CE, C.getLocationContext(), C.blockCount());
-
-    // Set the extent of the region in bytes. This enables us to use the SVal
-    // of the argument directly. If we saved the extent in bits, it'd be more
-    // difficult to reason about values like symbol*8.
-    auto Size = Call.getArgSVal(0);
-    if (auto DefSize = Size.getAs<DefinedOrUnknownSVal>()) {
-      // This `getAs()` is mostly paranoia, because core.CallAndMessage reports
-      // undefined function arguments (unless it's disabled somehow).
-      state = setDynamicExtent(state, R.getRegion(), *DefSize, SVB);
-    }
-    C.addTransition(state->BindExpr(CE, LCtx, R));
-    return true;
-  }
-
   case Builtin::BI__builtin_dynamic_object_size:
   case Builtin::BI__builtin_object_size:
   case Builtin::BI__builtin_constant_p: {

diff --git a/clang/lib/StaticAnalyzer/Checkers/MallocChecker.cpp b/clang/lib/StaticAnalyzer/Checkers/MallocChecker.cpp
@@ -401,10 +401,11 @@ class MallocChecker
   };
 
   const CallDescriptionMap<CheckFn> FreeingMemFnMap{
-      {{{"free"}, 1}, &MallocChecker::checkFree},
-      {{{"if_freenameindex"}, 1}, &MallocChecker::checkIfFreeNameIndex},
-      {{{"kfree"}, 1}, &MallocChecker::checkFree},
-      {{{"g_free"}, 1}, &MallocChecker::checkFree},
+      {{CDM::CLibrary, {"free"}, 1}, &MallocChecker::checkFree},
+      {{CDM::CLibrary, {"if_freenameindex"}, 1},
+       &MallocChecker::checkIfFreeNameIndex},
+      {{CDM::CLibrary, {"kfree"}, 1}, &MallocChecker::checkFree},
+      {{CDM::CLibrary, {"g_free"}, 1}, &MallocChecker::checkFree},
   };
 
   bool isFreeingCall(const CallEvent &Call) const;
@@ -413,41 +414,46 @@ class MallocChecker
   friend class NoOwnershipChangeVisitor;
 
   CallDescriptionMap<CheckFn> AllocatingMemFnMap{
-      {{{"alloca"}, 1}, &MallocChecker::checkAlloca},
-      {{{"_alloca"}, 1}, &MallocChecker::checkAlloca},
-      {{{"malloc"}, 1}, &MallocChecker::checkBasicAlloc},
-      {{{"malloc"}, 3}, &MallocChecker::checkKernelMalloc},
-      {{{"calloc"}, 2}, &MallocChecker::checkCalloc},
-      {{{"valloc"}, 1}, &MallocChecker::checkBasicAlloc},
+      {{CDM::CLibrary, {"alloca"}, 1}, &MallocChecker::checkAlloca},
+      {{CDM::CLibrary, {"_alloca"}, 1}, &MallocChecker::checkAlloca},
+      // The line for "alloca" also covers "__builtin_alloca", but the
+      // _with_align variant must be listed separately because it takes an
+      // extra argument:
+      {{CDM::CLibrary, {"__builtin_alloca_with_align"}, 2},
+       &MallocChecker::checkAlloca},
+      {{CDM::CLibrary, {"malloc"}, 1}, &MallocChecker::checkBasicAlloc},
+      {{CDM::CLibrary, {"malloc"}, 3}, &MallocChecker::checkKernelMalloc},
+      {{CDM::CLibrary, {"calloc"}, 2}, &MallocChecker::checkCalloc},
+      {{CDM::CLibrary, {"valloc"}, 1}, &MallocChecker::checkBasicAlloc},
       {{CDM::CLibrary, {"strndup"}, 2}, &MallocChecker::checkStrdup},
       {{CDM::CLibrary, {"strdup"}, 1}, &MallocChecker::checkStrdup},
-      {{{"_strdup"}, 1}, &MallocChecker::checkStrdup},
-      {{{"kmalloc"}, 2}, &MallocChecker::checkKernelMalloc},
-      {{{"if_nameindex"}, 1}, &MallocChecker::checkIfNameIndex},
+      {{CDM::CLibrary, {"_strdup"}, 1}, &MallocChecker::checkStrdup},
+      {{CDM::CLibrary, {"kmalloc"}, 2}, &MallocChecker::checkKernelMalloc},
+      {{CDM::CLibrary, {"if_nameindex"}, 1}, &MallocChecker::checkIfNameIndex},
       {{CDM::CLibrary, {"wcsdup"}, 1}, &MallocChecker::checkStrdup},
       {{CDM::CLibrary, {"_wcsdup"}, 1}, &MallocChecker::checkStrdup},
-      {{{"g_malloc"}, 1}, &MallocChecker::checkBasicAlloc},
-      {{{"g_malloc0"}, 1}, &MallocChecker::checkGMalloc0},
-      {{{"g_try_malloc"}, 1}, &MallocChecker::checkBasicAlloc},
-      {{{"g_try_malloc0"}, 1}, &MallocChecker::checkGMalloc0},
-      {{{"g_memdup"}, 2}, &MallocChecker::checkGMemdup},
-      {{{"g_malloc_n"}, 2}, &MallocChecker::checkGMallocN},
-      {{{"g_malloc0_n"}, 2}, &MallocChecker::checkGMallocN0},
-      {{{"g_try_malloc_n"}, 2}, &MallocChecker::checkGMallocN},
-      {{{"g_try_malloc0_n"}, 2}, &MallocChecker::checkGMallocN0},
+      {{CDM::CLibrary, {"g_malloc"}, 1}, &MallocChecker::checkBasicAlloc},
+      {{CDM::CLibrary, {"g_malloc0"}, 1}, &MallocChecker::checkGMalloc0},
+      {{CDM::CLibrary, {"g_try_malloc"}, 1}, &MallocChecker::checkBasicAlloc},
+      {{CDM::CLibrary, {"g_try_malloc0"}, 1}, &MallocChecker::checkGMalloc0},
+      {{CDM::CLibrary, {"g_memdup"}, 2}, &MallocChecker::checkGMemdup},
+      {{CDM::CLibrary, {"g_malloc_n"}, 2}, &MallocChecker::checkGMallocN},
+      {{CDM::CLibrary, {"g_malloc0_n"}, 2}, &MallocChecker::checkGMallocN0},
+      {{CDM::CLibrary, {"g_try_malloc_n"}, 2}, &MallocChecker::checkGMallocN},
+      {{CDM::CLibrary, {"g_try_malloc0_n"}, 2}, &MallocChecker::checkGMallocN0},
   };
 
   CallDescriptionMap<CheckFn> ReallocatingMemFnMap{
-      {{{"realloc"}, 2},
+      {{CDM::CLibrary, {"realloc"}, 2},
        std::bind(&MallocChecker::checkRealloc, _1, _2, _3, false)},
-      {{{"reallocf"}, 2},
+      {{CDM::CLibrary, {"reallocf"}, 2},
        std::bind(&MallocChecker::checkRealloc, _1, _2, _3, true)},
-      {{{"g_realloc"}, 2},
+      {{CDM::CLibrary, {"g_realloc"}, 2},
        std::bind(&MallocChecker::checkRealloc, _1, _2, _3, false)},
-      {{{"g_try_realloc"}, 2},
+      {{CDM::CLibrary, {"g_try_realloc"}, 2},
        std::bind(&MallocChecker::checkRealloc, _1, _2, _3, false)},
-      {{{"g_realloc_n"}, 3}, &MallocChecker::checkReallocN},
-      {{{"g_try_realloc_n"}, 3}, &MallocChecker::checkReallocN},
+      {{CDM::CLibrary, {"g_realloc_n"}, 3}, &MallocChecker::checkReallocN},
+      {{CDM::CLibrary, {"g_try_realloc_n"}, 3}, &MallocChecker::checkReallocN},
 
       // NOTE: the following CallDescription also matches the C++ standard
       // library function std::getline(); the callback will filter it out.
@@ -1259,9 +1265,6 @@ static bool isStandardRealloc(const CallEvent &Call) {
   assert(FD);
   ASTContext &AC = FD->getASTContext();
 
-  if (isa<CXXMethodDecl>(FD))
-    return false;
-
   return FD->getDeclaredReturnType().getDesugaredType(AC) == AC.VoidPtrTy &&
          FD->getParamDecl(0)->getType().getDesugaredType(AC) == AC.VoidPtrTy &&
          FD->getParamDecl(1)->getType().getDesugaredType(AC) ==
@@ -1273,9 +1276,6 @@ static bool isGRealloc(const CallEvent &Call) {
   assert(FD);
   ASTContext &AC = FD->getASTContext();
 
-  if (isa<CXXMethodDecl>(FD))
-    return false;
-
   return FD->getDeclaredReturnType().getDesugaredType(AC) == AC.VoidPtrTy &&
          FD->getParamDecl(0)->getType().getDesugaredType(AC) == AC.VoidPtrTy &&
          FD->getParamDecl(1)->getType().getDesugaredType(AC) ==
@@ -1284,14 +1284,14 @@ static bool isGRealloc(const CallEvent &Call) {
 
 void MallocChecker::checkRealloc(const CallEvent &Call, CheckerContext &C,
                                  bool ShouldFreeOnFail) const {
-  // HACK: CallDescription currently recognizes non-standard realloc functions
-  // as standard because it doesn't check the type, or wether its a non-method
-  // function. This should be solved by making CallDescription smarter.
-  // Mind that this came from a bug report, and all other functions suffer from
-  // this.
-  // https://bugs.llvm.org/show_bug.cgi?id=46253
+  // Ignore calls to functions whose type does not match the expected type of
+  // either the standard realloc or g_realloc from GLib.
+  // FIXME: Should we perform this kind of checking consistently for each
+  // function? If yes, then perhaps extend the `CallDescription` interface to
+  // handle this.
   if (!isStandardRealloc(Call) && !isGRealloc(Call))
     return;
+
   ProgramStateRef State = C.getState();
   State = ReallocMemAux(C, Call, ShouldFreeOnFail, State, AF_Malloc);
   State = ProcessZeroAllocCheck(Call, 1, State);
@@ -1842,9 +1842,18 @@ static ProgramStateRef MallocUpdateRefState(CheckerContext &C, const Expr *E,
     return nullptr;
 
   SymbolRef Sym = RetVal->getAsLocSymbol();
+
   // This is a return value of a function that was not inlined, such as malloc()
   // or new(). We've checked that in the caller. Therefore, it must be a symbol.
   assert(Sym);
+  // FIXME: In theory this assertion should fail for `alloca()` calls (because
+  // `AllocaRegion`s are not symbolic); but in practice this does not happen.
+  // As the current code appears to work correctly, I'm not touching this issue
+  // now, but it would be good to investigate and clarify this.
+  // Also note that perhaps the special `AllocaRegion` should be replaced by
+  // `SymbolicRegion` (or turned into a subclass of `SymbolicRegion`) to enable
+  // proper tracking of memory allocated by `alloca()` -- and after that change
+  // this assertion would become valid again.
 
   // Set the symbol's state to Allocated.
   return State->set<RegionState>(Sym, RefState::getAllocated(Family, E));

diff --git a/clang/lib/Tooling/DependencyScanning/ModuleDepCollector.cpp b/clang/lib/Tooling/DependencyScanning/ModuleDepCollector.cpp
@@ -154,6 +154,26 @@ void ModuleDepCollector::addOutputPaths(CowCompilerInvocation &CI,
   }
 }
 
+void dependencies::resetBenignCodeGenOptions(frontend::ActionKind ProgramAction,
+                                             const LangOptions &LangOpts,
+                                             CodeGenOptions &CGOpts) {
+  // TODO: Figure out better way to set options to their default value.
+  if (ProgramAction == frontend::GenerateModule) {
+    CGOpts.MainFileName.clear();
+    CGOpts.DwarfDebugFlags.clear();
+  }
+  if (ProgramAction == frontend::GeneratePCH ||
+      (ProgramAction == frontend::GenerateModule && !LangOpts.ModulesCodegen)) {
+    CGOpts.DebugCompilationDir.clear();
+    CGOpts.CoverageCompilationDir.clear();
+    CGOpts.CoverageDataFile.clear();
+    CGOpts.CoverageNotesFile.clear();
+    CGOpts.ProfileInstrumentUsePath.clear();
+    CGOpts.SampleProfileFile.clear();
+    CGOpts.ProfileRemappingFile.clear();
+  }
+}
+
 static CowCompilerInvocation
 makeCommonInvocationForModuleBuild(CompilerInvocation CI) {
   CI.resetNonModularOptions();
@@ -167,18 +187,8 @@ makeCommonInvocationForModuleBuild(CompilerInvocation CI) {
   // LLVM options are not going to affect the AST
   CI.getFrontendOpts().LLVMArgs.clear();
 
-  // TODO: Figure out better way to set options to their default value.
-  CI.getCodeGenOpts().MainFileName.clear();
-  CI.getCodeGenOpts().DwarfDebugFlags.clear();
-  if (!CI.getLangOpts().ModulesCodegen) {
-    CI.getCodeGenOpts().DebugCompilationDir.clear();
-    CI.getCodeGenOpts().CoverageCompilationDir.clear();
-    CI.getCodeGenOpts().CoverageDataFile.clear();
-    CI.getCodeGenOpts().CoverageNotesFile.clear();
-    CI.getCodeGenOpts().ProfileInstrumentUsePath.clear();
-    CI.getCodeGenOpts().SampleProfileFile.clear();
-    CI.getCodeGenOpts().ProfileRemappingFile.clear();
-  }
+  resetBenignCodeGenOptions(frontend::GenerateModule, CI.getLangOpts(),
+                            CI.getCodeGenOpts());
 
   // Map output paths that affect behaviour to "-" so their existence is in the
   // context hash. The final path will be computed in addOutputPaths.
@@ -342,6 +352,8 @@ static bool needsModules(FrontendInputFile FIF) {
 
 void ModuleDepCollector::applyDiscoveredDependencies(CompilerInvocation &CI) {
   CI.clearImplicitModuleBuildOptions();
+  resetBenignCodeGenOptions(CI.getFrontendOpts().ProgramAction,
+                            CI.getLangOpts(), CI.getCodeGenOpts());
 
   if (llvm::any_of(CI.getFrontendOpts().Inputs, needsModules)) {
     Preprocessor &PP = ScanInstance.getPreprocessor();

diff --git a/clang/test/AST/Interp/vectors.cpp b/clang/test/AST/Interp/vectors.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang_cc1 -fexperimental-new-constant-interpreter -verify=expected,both %s
 // RUN: %clang_cc1 -verify=ref,both %s
 
-// both-no-diagnostics
+// ref-no-diagnostics
 
 typedef int __attribute__((vector_size(16))) VI4;
 constexpr VI4 A = {1,2,3,4};
@@ -20,3 +20,9 @@ namespace Vector {
   }
   constexpr auto v2 = g(4);
 }
+
+/// FIXME: We need to support BitCasts between vector types.
+namespace {
+  typedef float __attribute__((vector_size(16))) VI42;
+  constexpr VI42 A2 = A; // expected-error {{must be initialized by a constant expression}}
+}
diff --git a/clang/test/AST/ast-dump-attr-json.cpp b/clang/test/AST/ast-dump-attr-json.cpp
@@ -46,6 +46,7 @@ __thread __attribute__ ((tls_model ("local-exec"))) int tls_model_var;
 // CHECK-NEXT:    "tokLen": 11
 // CHECK-NEXT:   }
 // CHECK-NEXT:  },
+// CHECK-NEXT:  "isUsed": true,
 // CHECK-NEXT:  "name": "global_decl",
 // CHECK-NEXT:  "mangledName": "global_decl",
 // CHECK-NEXT:  "type": {

diff --git a/clang/test/Analysis/Inputs/system-header-simulator-cxx.h b/clang/test/Analysis/Inputs/system-header-simulator-cxx.h
@@ -1106,6 +1106,7 @@ using ostream = basic_ostream<char>;
 extern std::ostream cout;
 
 ostream &operator<<(ostream &, const string &);
+
 #if __cplusplus >= 202002L
 template <class T>
 ostream &operator<<(ostream &, const std::unique_ptr<T> &);
@@ -1122,11 +1123,12 @@ istream &getline(istream &, string &, char);
 istream &getline(istream &, string &);
 } // namespace std
 
-#ifdef TEST_INLINABLE_ALLOCATORS
 namespace std {
   void *malloc(size_t);
   void free(void *);
-}
+} // namespace std
+
+#ifdef TEST_INLINABLE_ALLOCATORS
 void* operator new(std::size_t size, const std::nothrow_t&) throw() { return std::malloc(size); }
 void* operator new[](std::size_t size, const std::nothrow_t&) throw() { return std::malloc(size); }
 void operator delete(void* ptr, const std::nothrow_t&) throw() { std::free(ptr); }

diff --git a/clang/test/Analysis/cxx-uninitialized-object-ptr-ref.cpp b/clang/test/Analysis/cxx-uninitialized-object-ptr-ref.cpp
@@ -1,9 +1,9 @@
-// RUN: %clang_analyze_cc1 -analyzer-checker=core,optin.cplusplus.UninitializedObject \
+// RUN: %clang_analyze_cc1 -analyzer-checker=core,unix.Malloc,optin.cplusplus.UninitializedObject \
 // RUN:   -analyzer-config optin.cplusplus.UninitializedObject:Pedantic=true -DPEDANTIC \
 // RUN:   -analyzer-config optin.cplusplus.UninitializedObject:CheckPointeeInitialization=true \
 // RUN:   -std=c++11 -verify  %s
 
-// RUN: %clang_analyze_cc1 -analyzer-checker=core,optin.cplusplus.UninitializedObject \
+// RUN: %clang_analyze_cc1 -analyzer-checker=core,unix.Malloc,optin.cplusplus.UninitializedObject \
 // RUN:   -analyzer-config optin.cplusplus.UninitializedObject:CheckPointeeInitialization=true \
 // RUN:   -std=c++11 -verify  %s
 
@@ -316,7 +316,10 @@ void fCyclicPointerTest2() {
 
 // Void pointer tests are mainly no-crash tests.
 
-void *malloc(int size);
+typedef __typeof(sizeof(int)) size_t;
+
+void *calloc(size_t nmemb, size_t size);
+void free(void *p);
 
 class VoidPointerTest1 {
   void *vptr;
@@ -328,8 +331,9 @@ class VoidPointerTest1 {
 };
 
 void fVoidPointerTest1() {
-  void *vptr = malloc(sizeof(int));
+  void *vptr = calloc(1, sizeof(int));
   VoidPointerTest1(vptr, char());
+  free(vptr);
 }
 
 class VoidPointerTest2 {
@@ -342,8 +346,9 @@ class VoidPointerTest2 {
 };
 
 void fVoidPointerTest2() {
-  void *vptr = malloc(sizeof(int));
+  void *vptr = calloc(1, sizeof(int));
   VoidPointerTest2(&vptr, char());
+  free(vptr);
 }
 
 class VoidPointerRRefTest1 {
@@ -359,8 +364,9 @@ upon returning to the caller.  This will be a dangling reference}}
 };
 
 void fVoidPointerRRefTest1() {
-  void *vptr = malloc(sizeof(int));
+  void *vptr = calloc(1, sizeof(int));
   VoidPointerRRefTest1(vptr, char());
+  free(vptr);
 }
 
 class VoidPointerRRefTest2 {
@@ -376,8 +382,9 @@ upon returning to the caller.  This will be a dangling reference}}
 };
 
 void fVoidPointerRRefTest2() {
-  void *vptr = malloc(sizeof(int));
+  void *vptr = calloc(1, sizeof(int));
   VoidPointerRRefTest2(&vptr, char());
+  free(vptr);
 }
 
 class VoidPointerLRefTest {
@@ -393,8 +400,9 @@ upon returning to the caller.  This will be a dangling reference}}
 };
 
 void fVoidPointerLRefTest() {
-  void *vptr = malloc(sizeof(int));
+  void *vptr = calloc(1, sizeof(int));
   VoidPointerLRefTest(vptr, char());
+  free(vptr);
 }
 
 struct CyclicVoidPointerTest {

diff --git a/clang/test/Analysis/exercise-ps.c b/clang/test/Analysis/exercise-ps.c
@@ -1,5 +1,5 @@
 // RUN: %clang_analyze_cc1 %s -verify -Wno-error=implicit-function-declaration \
-// RUN:   -analyzer-checker=core \
+// RUN:   -analyzer-checker=core,unix.Malloc \
 // RUN:   -analyzer-config core.CallAndMessage:ArgPointeeInitializedness=true
 //
 // Just exercise the analyzer on code that has at one point caused issues

diff --git a/clang/test/Analysis/explain-svals.cpp b/clang/test/Analysis/explain-svals.cpp
@@ -1,7 +1,7 @@
 // RUN: %clang_analyze_cc1 -triple i386-apple-darwin10 -verify %s \
-// RUN:   -analyzer-checker=core.builtin \
 // RUN:   -analyzer-checker=debug.ExprInspection \
 // RUN:   -analyzer-checker=unix.cstring \
+// RUN:   -analyzer-checker=unix.Malloc \
 // RUN:   -analyzer-config display-checker-name=false
 
 typedef unsigned long size_t;

diff --git a/clang/test/Analysis/malloc-std-namespace.cpp b/clang/test/Analysis/malloc-std-namespace.cpp
@@ -0,0 +1,24 @@
+// RUN: %clang_analyze_cc1 -analyzer-checker=core,unix.Malloc -verify -analyzer-output=text %s
+
+// This file tests that unix.Malloc can handle C++ code where e.g. malloc and
+// free are declared within the namespace 'std' by the header <cstdlib>.
+
+#include "Inputs/system-header-simulator-cxx.h"
+
+void leak() {
+  int *p = static_cast<int*>(std::malloc(sizeof(int))); // expected-note{{Memory is allocated}}
+} // expected-warning{{Potential leak of memory pointed to by 'p'}}
+  // expected-note@-1{{Potential leak of memory pointed to by 'p'}}
+
+void no_leak() {
+  int *p = static_cast<int*>(std::malloc(sizeof(int)));
+  std::free(p); // no-warning
+}
+
+void invalid_free() {
+  int i;
+  int *p = &i;
+  //expected-note@+2{{Argument to free() is the address of the local variable 'i', which is not memory allocated by malloc()}}
+  //expected-warning@+1{{Argument to free() is the address of the local variable 'i', which is not memory allocated by malloc()}}
+  std::free(p);
+}
diff --git a/clang/test/Analysis/malloc.c b/clang/test/Analysis/malloc.c
@@ -740,6 +740,17 @@ void allocaFree(void) {
   free(p); // expected-warning {{Memory allocated by alloca() should not be deallocated}}
 }
 
+void allocaFreeBuiltin(void) {
+  int *p = __builtin_alloca(sizeof(int));
+  free(p); // expected-warning {{Memory allocated by alloca() should not be deallocated}}
+}
+
+void allocaFreeBuiltinAlign(void) {
+  int *p = __builtin_alloca_with_align(sizeof(int), 64);
+  free(p); // expected-warning {{Memory allocated by alloca() should not be deallocated}}
+}
+
+
 int* mallocEscapeRet(void) {
   int *p = malloc(12);
   return p; // no warning

diff --git a/clang/test/Analysis/malloc.cpp b/clang/test/Analysis/malloc.cpp
@@ -214,3 +214,14 @@ void *realloc(void **ptr, size_t size) { realloc(ptr, size); } // no-crash
 namespace pr46253_paramty2{
 void *realloc(void *ptr, int size) { realloc(ptr, size); } // no-crash
 } // namespace pr46253_paramty2
+
+namespace pr81597 {
+struct S {};
+struct T {
+  void free(const S& s);
+};
+void f(T& t) {
+  S s;
+  t.free(s); // no-warning: This is not the free you are looking for...
+}
+} // namespace pr81597
diff --git a/clang/test/Analysis/stack-addr-ps.c b/clang/test/Analysis/stack-addr-ps.c
@@ -1,4 +1,4 @@
-// RUN: %clang_analyze_cc1 -analyzer-checker=core -fblocks -verify %s
+// RUN: %clang_analyze_cc1 -analyzer-checker=core,unix.Malloc -fblocks -verify %s
 
 int* f1(void) {
   int x = 0;

diff --git a/clang/test/Analysis/stackaddrleak.c b/clang/test/Analysis/stackaddrleak.c
@@ -1,5 +1,5 @@
-// RUN: %clang_analyze_cc1 -analyzer-checker=core -verify -std=c99 -Dbool=_Bool -Wno-bool-conversion %s
-// RUN: %clang_analyze_cc1 -analyzer-checker=core -verify -x c++ -Wno-bool-conversion %s
+// RUN: %clang_analyze_cc1 -analyzer-checker=core,unix.Malloc -verify -std=c99 -Dbool=_Bool -Wno-bool-conversion %s
+// RUN: %clang_analyze_cc1 -analyzer-checker=core,unix.Malloc -verify -x c++ -Wno-bool-conversion %s
 
 typedef __INTPTR_TYPE__ intptr_t;
 char const *p;

diff --git a/clang/test/CXX/over/over.built/ast.cpp b/clang/test/CXX/over/over.built/ast.cpp
@@ -4,11 +4,11 @@ struct A{};
 
 template <typename T, typename U>
 auto Test(T* pt, U* pu) {
-  // CHECK: UnaryOperator {{.*}} '<dependent type>' prefix '*'
+  // CHECK: UnaryOperator {{.*}} '<dependent type>' lvalue prefix '*'
   // CHECK-NEXT: DeclRefExpr {{.*}} 'T *' lvalue ParmVar {{.*}} 'pt' 'T *'
   (void)*pt;
 
-  // CHECK: UnaryOperator {{.*}} '<dependent type>' prefix '++'
+  // CHECK: UnaryOperator {{.*}} '<dependent type>' lvalue prefix '++'
   // CHECK-NEXT: DeclRefExpr {{.*}} 'T *' lvalue ParmVar {{.*}} 'pt' 'T *'
   (void)(++pt);
 

diff --git a/clang/test/ClangScanDeps/removed-args.c b/clang/test/ClangScanDeps/removed-args.c
@@ -93,3 +93,31 @@
 // CHECK-NOT:          "-fmodules-prune-interval=
 // CHECK-NOT:          "-fmodules-prune-after=
 // CHECK:            ],
+
+// Check for removed args for PCH invocations.
+
+// RUN: split-file %s %t
+// RUN: sed "s|DIR|%/t|g" %t/cdb-pch.json.template > %t/cdb-pch.json
+// RUN: clang-scan-deps -compilation-database %t/cdb-pch.json -format experimental-full > %t/result-pch.json
+// RUN: cat %t/result-pch.json | sed 's:\\\\\?:/:g' | FileCheck %s -DPREFIX=%/t -check-prefix=PCH
+//
+// PCH-NOT:          "-fdebug-compilation-dir="
+// PCH-NOT:          "-fcoverage-compilation-dir="
+// PCH-NOT:          "-coverage-notes-file
+// PCH-NOT:          "-coverage-data-file
+// PCH-NOT:          "-fprofile-instrument-use-path
+// PCH-NOT:          "-include"
+// PCH-NOT:          "-fmodules-cache-path=
+// PCH-NOT:          "-fmodules-validate-once-per-build-session"
+// PCH-NOT:          "-fbuild-session-timestamp=
+// PCH-NOT:          "-fmodules-prune-interval=
+// PCH-NOT:          "-fmodules-prune-after=
+
+//--- cdb-pch.json.template
+[
+  {
+    "directory": "DIR",
+    "command": "clang -x c-header DIR/header.h -fmodules -fimplicit-module-maps -fmodules-cache-path=DIR/cache -fdebug-compilation-dir=DIR/debug -fcoverage-compilation-dir=DIR/coverage -ftest-coverage -fprofile-instr-use=DIR/tu.profdata -o DIR/header.h.pch -serialize-diagnostics DIR/header.h.pch.diag ",
+    "file": "DIR/header.h.pch"
+  }
+]
diff --git a/clang/test/CodeGen/bitfield-access-unit.c b/clang/test/CodeGen/bitfield-access-unit.c
@@ -222,6 +222,24 @@ struct G {
 // LAYOUT-DWN32-NEXT: <CGBitFieldInfo Offset:{{[0-9]+}} Size:7 IsSigned:1 StorageSize:8 StorageOffset:4
 // CHECK-NEXT: ]>
 
+struct __attribute__((aligned(8))) H {
+  char a;
+  unsigned b : 24; // on expensive alignment we want this to stay 24
+  unsigned c __attribute__((aligned(8))); // Think 'long long' or lp64 ptr
+} h;
+// CHECK-LABEL: LLVMType:%struct.H =
+// LAYOUT-FLEX-SAME: type <{ i8, i32, [3 x i8], i32, [4 x i8] }>
+// LAYOUT-STRICT-SAME: type { i8, [3 x i8], [4 x i8], i32, [4 x i8] }
+// LAYOUT-DWN32-FLEX-SAME: type <{ i8, i32, [3 x i8], i32, [4 x i8] }>
+// LAYOUT-DWN32-STRICT-SAME: type { i8, [3 x i8], [4 x i8], i32, [4 x i8] }
+// CHECK: BitFields:[
+// LAYOUT-FLEX-NEXT: <CGBitFieldInfo Offset:{{[0-9]+}} Size:24 IsSigned:0 StorageSize:32 StorageOffset:1
+// LAYOUT-STRICT-NEXT: <CGBitFieldInfo Offset:{{[0-9]+}} Size:24 IsSigned:0 StorageSize:24 StorageOffset:1
+
+// LAYOUT-DWN32-FLEX-NEXT: <CGBitFieldInfo Offset:{{[0-9]+}} Size:24 IsSigned:0 StorageSize:32 StorageOffset:1
+// LAYOUT-DWN32-STRICT-NEXT: <CGBitFieldInfo Offset:{{[0-9]+}} Size:24 IsSigned:0 StorageSize:24 StorageOffset:1
+// CHECK-NEXT: ]>
+
 #if _LP64
 struct A64 {
   int a : 16;

diff --git a/clang/test/CodeGenCXX/control-flow-in-stmt-expr.cpp b/clang/test/CodeGenCXX/control-flow-in-stmt-expr.cpp
@@ -1,4 +1,5 @@
-// RUN: %clang_cc1 --std=c++20 -triple x86_64-linux-gnu -emit-llvm %s -o - | FileCheck %s
+// RUN: %clang_cc1 --std=c++20 -fexceptions -triple x86_64-linux-gnu -emit-llvm %s -o - | FileCheck -check-prefixes=EH %s
+// RUN: %clang_cc1 --std=c++20 -triple x86_64-linux-gnu -emit-llvm %s -o - | FileCheck -check-prefixes=NOEH,CHECK %s
 
 struct Printy {
   Printy(const char *name) : name(name) {}
@@ -349,6 +350,34 @@ void NewArrayInit() {
   // CHECK-NEXT:    br label %return
 }
 
+void DestroyInConditionalCleanup() {
+  // EH-LABEL: DestroyInConditionalCleanupv()
+  // NOEH-LABEL: DestroyInConditionalCleanupv()
+  struct A {
+    A() {}
+    ~A() {}
+  };
+
+  struct Value {
+    Value(A) {}
+    ~Value() {}
+  };
+
+  struct V2 {
+    Value K;
+    Value V;
+  };
+  // Verify we use conditional cleanups.
+  (void)(foo() ? V2{A(), A()} : V2{A(), A()});
+  // NOEH:   cond.true:
+  // NOEH:      call void @_ZZ27DestroyInConditionalCleanupvEN1AC1Ev
+  // NOEH:      store ptr %{{.*}}, ptr %cond-cleanup.save
+
+  // EH:   cond.true:
+  // EH:        invoke void @_ZZ27DestroyInConditionalCleanupvEN1AC1Ev
+  // EH:        store ptr %{{.*}}, ptr %cond-cleanup.save
+}
+
 void ArrayInitWithContinue() {
   // CHECK-LABEL: @_Z21ArrayInitWithContinuev
   // Verify that we start to emit the array destructor.
@@ -362,3 +391,19 @@ void ArrayInitWithContinue() {
                      })};
   }
 }
+
+struct [[clang::trivial_abi]] HasTrivialABI {
+  HasTrivialABI();
+  ~HasTrivialABI();
+};
+void AcceptTrivialABI(HasTrivialABI, int);
+void TrivialABI() {
+  // CHECK-LABEL: define dso_local void @_Z10TrivialABIv()
+  AcceptTrivialABI(HasTrivialABI(), ({
+                     if (foo()) return;
+                     // CHECK:      if.then:
+                     // CHECK-NEXT:   call void @_ZN13HasTrivialABID1Ev
+                     // CHECK-NEXT:   br label %return
+                     0;
+                   }));
+}
diff --git a/clang/test/CodeGenCXX/module-funcs-from-imports.cppm b/clang/test/CodeGenCXX/module-funcs-from-imports.cppm
@@ -23,6 +23,21 @@ int func_in_gmf_not_called() {
     return 44;
 }
 
+template <class T>
+class A {
+public:
+    __attribute__((always_inline))
+    inline constexpr int getValue() {
+        return 43;
+    }
+
+    inline constexpr int getValue2() {
+        return 43;
+    }
+};
+
+extern template class A<char>;
+
 //--- M.cppm
 module;
 #include "foo.h"
@@ -47,17 +62,21 @@ int always_inline_func() {
     return 45;
 }
 
+export using ::A;
+
 //--- Use.cpp
 import M;
 int use() {
-    return exported_func() + always_inline_func();
+    A<char> a;
+    return exported_func() + always_inline_func() +
+           a.getValue() + a.getValue2();
 }
 
-// Checks that none of the function (except the always_inline_func) in the importees
-// are generated in the importer's code.
 // CHECK-O0: define{{.*}}_Z3usev(
 // CHECK-O0: declare{{.*}}_ZW1M13exported_funcv(
-// CHECK-O0: define{{.*}}available_externally{{.*}}_ZW1M18always_inline_funcv(
+// CHECK-O0: declare{{.*}}_ZW1M18always_inline_funcv(
+// CHECK-O0: define{{.*}}@_ZN1AIcE8getValueEv(
+// CHECK-O0: declare{{.*}}@_ZN1AIcE9getValue2Ev(
 // CHECK-O0-NOT: func_in_gmf
 // CHECK-O0-NOT: func_in_gmf_not_called
 // CHECK-O0-NOT: non_exported_func
@@ -68,7 +87,9 @@ int use() {
 // O0 to keep consistent ABI.
 // CHECK-O1: define{{.*}}_Z3usev(
 // CHECK-O1: declare{{.*}}_ZW1M13exported_funcv(
-// CHECK-O1: define{{.*}}available_externally{{.*}}_ZW1M18always_inline_funcv(
+// CHECK-O1: declare{{.*}}_ZW1M18always_inline_funcv(
+// CHECK-O1: define{{.*}}@_ZN1AIcE8getValueEv(
+// CHECK-O1: declare{{.*}}@_ZN1AIcE9getValue2Ev(
 // CHECK-O1-NOT: func_in_gmf
 // CHECK-O1-NOT: func_in_gmf_not_called
 // CHECK-O1-NOT: non_exported_func

diff --git a/clang/test/CodeGenHLSL/builtins/any.hlsl b/clang/test/CodeGenHLSL/builtins/any.hlsl
diff --git a/clang/test/Modules/hashing-decls-in-exprs-from-gmf-2.cppm b/clang/test/Modules/hashing-decls-in-exprs-from-gmf-2.cppm
@@ -0,0 +1,44 @@
+// RUN: rm -rf %t
+// RUN: mkdir -p %t
+// RUN: split-file %s %t
+//
+// RUN: %clang_cc1 -std=c++20 -fskip-odr-check-in-gmf %t/A.cppm -emit-module-interface -o %t/A.pcm
+// RUN: %clang_cc1 -std=c++20 -fskip-odr-check-in-gmf %t/test.cpp -fprebuilt-module-path=%t -fsyntax-only -verify
+
+//--- header.h
+#pragma once
+template <class _Tp>
+class Optional {};
+
+template <class _Tp>
+concept C = requires(const _Tp& __t) {
+    []<class _Up>(const Optional<_Up>&) {}(__t);
+};
+
+//--- func.h
+#include "header.h"
+template <C T>
+void func() {}
+
+//--- test_func.h
+#include "func.h"
+
+inline void test_func() {
+    func<Optional<int>>();
+}
+
+//--- A.cppm
+module;
+#include "header.h"
+#include "test_func.h"
+export module A;
+export using ::test_func;
+
+//--- test.cpp
+// expected-no-diagnostics
+import A;
+#include "test_func.h"
+
+void test() {
+    test_func();
+}
diff --git a/clang/test/OpenMP/ompx_attributes_codegen.cpp b/clang/test/OpenMP/ompx_attributes_codegen.cpp
@@ -36,6 +36,5 @@ void func() {
 // NVIDIA: "omp_target_thread_limit"="45"
 // NVIDIA: "omp_target_thread_limit"="17"
 // NVIDIA: !{ptr @__omp_offloading[[HASH1:.*]]_l16, !"maxntidx", i32 20}
-// NVIDIA: !{ptr @__omp_offloading[[HASH2:.*]]_l18, !"minctasm", i32 90}
-// NVIDIA: !{ptr @__omp_offloading[[HASH2]]_l18, !"maxntidx", i32 45}
+// NVIDIA: !{ptr @__omp_offloading[[HASH2:.*]]_l18, !"maxntidx", i32 45}
 // NVIDIA: !{ptr @__omp_offloading[[HASH3:.*]]_l20, !"maxntidx", i32 17}
diff --git a/clang/test/Sema/alias-unused-win.cpp b/clang/test/Sema/alias-unused-win.cpp
@@ -7,7 +7,7 @@ extern "C" {
 static int f(void) { return 42; } // cxx-warning{{unused function 'f'}}
 int g(void) __attribute__((alias("f")));
 
-static int foo [] = { 42, 0xDEAD }; // cxx-warning{{variable 'foo' is not needed and will not be emitted}}
+static int foo [] = { 42, 0xDEAD };
 extern typeof(foo) bar __attribute__((unused, alias("foo")));
 
 static int __attribute__((overloadable)) f0(int x) { return x; } // expected-warning{{unused function 'f0'}}

diff --git a/clang/test/Sema/alias-unused.cpp b/clang/test/Sema/alias-unused.cpp
@@ -14,24 +14,26 @@ extern typeof(foo) bar __attribute__((unused, alias("foo")));
 /// We report a warning in C++ mode because the internal linkage `resolver` gets
 /// mangled as it does not have a language linkage. GCC does not mangle
 /// `resolver` or report a warning.
-static int (*resolver(void))(void) { return f; } // expected-warning{{unused function 'resolver'}}
+static int (*resolver(void))(void) { return f; } // cxx-warning{{unused function 'resolver'}}
 int ifunc(void) __attribute__((ifunc("resolver")));
 
-static int __attribute__((overloadable)) f0(int x) { return x; } // expected-warning{{unused function 'f0'}}
+static int __attribute__((overloadable)) f0(int x) { return x; }
 static float __attribute__((overloadable)) f0(float x) { return x; } // expected-warning{{unused function 'f0'}}
 int g0(void) __attribute__((alias("_ZL2f0i")));
 
 #ifdef __cplusplus
-static int f1() { return 42; } // expected-warning{{unused function 'f1'}}
+static int f1() { return 42; }
 int g1(void) __attribute__((alias("_ZL2f1v")));
 }
 
-static int f2(int) { return 42; } // expected-warning{{unused function 'f2'}}
-static int f2() { return 42; } // expected-warning{{unused function 'f2'}}
+/// We demangle alias/ifunc target and mark all found functions as used.
+
+static int f2(int) { return 42; } // cxx-warning{{unused function 'f2'}}
+static int f2() { return 42; }
 int g2() __attribute__((alias("_ZL2f2v")));
 
-static int (*resolver1())() { return f; } // expected-warning{{unused function 'resolver1'}}
-static int (*resolver1(int))() { return f; } // expected-warning{{unused function 'resolver1'}}
+static int (*resolver1())() { return f; } // cxx-warning{{unused function 'resolver1'}}
+static int (*resolver1(int))() { return f; }
 int ifunc1() __attribute__((ifunc("_ZL9resolver1i")));
 
 /// TODO: We should report "unused function" for f3(int).

diff --git a/clang/test/SemaCXX/PR84020.cpp b/clang/test/SemaCXX/PR84020.cpp
@@ -0,0 +1,23 @@
+// RUN: %clang_cc1 -std=c++20 -verify %s
+// RUN: %clang_cc1 -std=c++23 -verify %s
+// expected-no-diagnostics
+
+struct B {
+    template <typename S>
+    void foo();
+
+    void bar();
+};
+
+template <typename T, typename S>
+struct A : T {
+    auto foo() {
+        static_assert(requires { T::template foo<S>(); });
+        static_assert(requires { T::bar(); });
+    }
+};
+
+int main() {
+    A<B, double> a;
+    a.foo();
+}
diff --git a/clang/test/SemaCXX/overloaded-operator.cpp b/clang/test/SemaCXX/overloaded-operator.cpp
@@ -682,3 +682,13 @@ namespace nw{
   }
 }
 #endif
+
+#if __cplusplus >= 201703L
+namespace GH88329 {
+
+template <auto T> struct A {};
+template <auto T> A<*T> operator *() { return {}; }
+// expected-error@-1 {{overloaded 'operator*' must have at least one parameter of class or enumeration type}}
+}
+
+#endif
diff --git a/clang/unittests/Analysis/FlowSensitive/TransferTest.cpp b/clang/unittests/Analysis/FlowSensitive/TransferTest.cpp
@@ -3098,6 +3098,58 @@ TEST(TransferTest, ResultObjectLocationForCXXOperatorCallExpr) {
       });
 }
 
+// Check that the `std::strong_ordering` object returned by builtin `<=>` has a
+// correctly modeled result object location.
+TEST(TransferTest, ResultObjectLocationForBuiltinSpaceshipOperator) {
+  std::string Code = R"(
+    namespace std {
+      // This is the minimal definition required to get
+      // `Sema::CheckComparisonCategoryType()` to accept this fake.
+      struct strong_ordering {
+        enum class ordering { less, equal, greater };
+        ordering o;
+        static const strong_ordering less;
+        static const strong_ordering equivalent;
+        static const strong_ordering equal;
+        static const strong_ordering greater;
+      };
+
+      inline constexpr strong_ordering strong_ordering::less =
+        { strong_ordering::ordering::less };
+      inline constexpr strong_ordering strong_ordering::equal =
+        { strong_ordering::ordering::equal };
+      inline constexpr strong_ordering strong_ordering::equivalent =
+        { strong_ordering::ordering::equal };
+      inline constexpr strong_ordering strong_ordering::greater =
+        { strong_ordering::ordering::greater };
+    }
+    void target(int i, int j) {
+      auto ordering = i <=> j;
+      // [[p]]
+    }
+  )";
+  using ast_matchers::binaryOperator;
+  using ast_matchers::hasOperatorName;
+  using ast_matchers::match;
+  using ast_matchers::selectFirst;
+  using ast_matchers::traverse;
+  runDataflow(
+      Code,
+      [](const llvm::StringMap<DataflowAnalysisState<NoopLattice>> &Results,
+         ASTContext &ASTCtx) {
+        const Environment &Env = getEnvironmentAtAnnotation(Results, "p");
+
+        auto *Spaceship = selectFirst<BinaryOperator>(
+            "op",
+            match(binaryOperator(hasOperatorName("<=>")).bind("op"), ASTCtx));
+
+        EXPECT_EQ(
+            &Env.getResultObjectLocation(*Spaceship),
+            &getLocForDecl<RecordStorageLocation>(ASTCtx, Env, "ordering"));
+      },
+      LangStandard::lang_cxx20);
+}
+
 TEST(TransferTest, ResultObjectLocationForStdInitializerListExpr) {
   std::string Code = R"(
     namespace std {

diff --git a/compiler-rt/test/sanitizer_common/TestCases/Linux/mmap_56bit_test.c b/compiler-rt/test/sanitizer_common/TestCases/Linux/mmap_56bit_test.c
@@ -1,6 +1,9 @@
-// RUN: %clangxx %s -pie -fPIE -o %t && %run %t
+// RUN: %clang %s -pie -fPIE -o %t && %run %t
 // REQUIRES: x86_64-target-arch
 
+// FIXME: Fails Asan, as expected, with 5lvl page tables.
+// UNSUPPORTED: x86_64-target-arch
+
 #include <assert.h>
 #include <stdio.h>
 #include <sys/mman.h>

diff --git a/flang/include/flang/Optimizer/Dialect/FIROps.td b/flang/include/flang/Optimizer/Dialect/FIROps.td
@@ -3190,4 +3190,36 @@ def fir_CUDADataTransferOp : fir_Op<"cuda_data_transfer", []> {
   }];
 }
 
+def fir_CUDAAllocateOp : fir_Op<"cuda_allocate", [AttrSizedOperandSegments,
+    MemoryEffects<[MemAlloc<DefaultResource>]>]> {
+  let summary = "Perform the device allocation of data of an allocatable";
+
+  let description = [{
+    The fir.cuda_allocate operation performs the allocation on the device
+    of the data of an allocatable. The descriptor passed to the operation
+    is initialized before with the standard flang runtime calls.
+  }];
+
+  let arguments = (ins Arg<AnyRefOrBoxType, "", [MemWrite]>:$box,
+                       Arg<Optional<AnyRefOrBoxType>, "", [MemWrite]>:$errmsg,
+                       Optional<AnyIntegerType>:$stream,
+                       Arg<Optional<AnyRefOrBoxType>, "", [MemWrite]>:$pinned,
+                       Arg<Optional<AnyRefOrBoxType>, "", [MemRead]>:$source,
+                       fir_CUDADataAttributeAttr:$cuda_attr,
+                       UnitAttr:$hasStat);
+
+  let results = (outs AnyIntegerType:$stat);
+
+  let assemblyFormat = [{
+    $box `:` qualified(type($box))
+    ( `source` `(` $source^ `:` qualified(type($source) )`)` )?
+    ( `errmsg` `(` $errmsg^ `:` type($errmsg) `)` )?
+    ( `stream` `(` $stream^ `:` type($stream) `)` )?
+    ( `pinned` `(` $pinned^ `:` type($pinned) `)` )?
+    attr-dict `->` type($stat)
+  }];
+
+  let hasVerifier = 1;
+}
+
 #endif
diff --git a/flang/include/flang/Optimizer/Dialect/FIRTypes.td b/flang/include/flang/Optimizer/Dialect/FIRTypes.td
@@ -625,6 +625,7 @@ def AnyRefOrBoxLike : TypeConstraint<Or<[AnyReferenceLike.predicate,
 def AnyRefOrBox : TypeConstraint<Or<[fir_ReferenceType.predicate,
     fir_HeapType.predicate, fir_PointerType.predicate,
     IsBaseBoxTypePred]>, "any reference or box">;
+def AnyRefOrBoxType : Type<AnyRefOrBox.predicate, "any legal ref or box type">;
 
 def AnyShapeLike : TypeConstraint<Or<[fir_ShapeType.predicate,
     fir_ShapeShiftType.predicate]>, "any legal shape type">;

diff --git a/flang/include/flang/Parser/dump-parse-tree.h b/flang/include/flang/Parser/dump-parse-tree.h
@@ -542,6 +542,7 @@ class ParseTreeDumper {
   NODE_ENUM(OmpOrderModifier, Kind)
   NODE(parser, OmpProcBindClause)
   NODE_ENUM(OmpProcBindClause, Type)
+  NODE_ENUM(OmpReductionClause, ReductionModifier)
   NODE(parser, OmpReductionClause)
   NODE(parser, OmpInReductionClause)
   NODE(parser, OmpReductionCombiner)

diff --git a/flang/include/flang/Parser/parse-tree.h b/flang/include/flang/Parser/parse-tree.h
@@ -3552,7 +3552,10 @@ struct OmpReductionOperator {
 //                                         variable-name-list)
 struct OmpReductionClause {
   TUPLE_CLASS_BOILERPLATE(OmpReductionClause);
-  std::tuple<OmpReductionOperator, OmpObjectList> t;
+  ENUM_CLASS(ReductionModifier, Inscan, Task, Default)
+  std::tuple<std::optional<ReductionModifier>, OmpReductionOperator,
+      OmpObjectList>
+      t;
 };
 
 // OMP 5.0 2.19.5.6 in_reduction-clause -> IN_REDUCTION (reduction-identifier:

diff --git a/flang/lib/Lower/OpenMP/Clauses.cpp b/flang/lib/Lower/OpenMP/Clauses.cpp
@@ -1005,12 +1005,28 @@ ProcBind make(const parser::OmpClause::ProcBind &inp,
 Reduction make(const parser::OmpClause::Reduction &inp,
                semantics::SemanticsContext &semaCtx) {
   // inp.v -> parser::OmpReductionClause
-  auto &t0 = std::get<parser::OmpReductionOperator>(inp.v.t);
-  auto &t1 = std::get<parser::OmpObjectList>(inp.v.t);
+  using wrapped = parser::OmpReductionClause;
+
+  CLAUSET_ENUM_CONVERT( //
+      convert, wrapped::ReductionModifier, Reduction::ReductionModifier,
+      // clang-format off
+      MS(Inscan,  Inscan)
+      MS(Task,    Task)
+      MS(Default, Default)
+      // clang-format on
+  );
+
+  auto &t0 =
+      std::get<std::optional<parser::OmpReductionClause::ReductionModifier>>(
+          inp.v.t);
+  auto &t1 = std::get<parser::OmpReductionOperator>(inp.v.t);
+  auto &t2 = std::get<parser::OmpObjectList>(inp.v.t);
   return Reduction{
-      {/*ReductionIdentifiers=*/{makeReductionOperator(t0, semaCtx)},
-       /*ReductionModifier=*/std::nullopt,
-       /*List=*/makeObjects(t1, semaCtx)}};
+      {/*ReductionModifier=*/t0
+           ? std::make_optional<Reduction::ReductionModifier>(convert(*t0))
+           : std::nullopt,
+       /*ReductionIdentifiers=*/{makeReductionOperator(t1, semaCtx)},
+       /*List=*/makeObjects(t2, semaCtx)}};
 }
 
 // Relaxed: empty

diff --git a/flang/lib/Lower/OpenMP/ReductionProcessor.cpp b/flang/lib/Lower/OpenMP/ReductionProcessor.cpp
@@ -574,6 +574,11 @@ void ReductionProcessor::addDeclareReduction(
     llvm::SmallVectorImpl<const Fortran::semantics::Symbol *>
         *reductionSymbols) {
   fir::FirOpBuilder &firOpBuilder = converter.getFirOpBuilder();
+
+  if (std::get<std::optional<omp::clause::Reduction::ReductionModifier>>(
+          reduction.t))
+    TODO(currentLocation, "Reduction modifiers are not supported");
+
   mlir::omp::DeclareReductionOp decl;
   const auto &redOperatorList{
       std::get<omp::clause::Reduction::ReductionIdentifiers>(reduction.t)};

diff --git a/flang/lib/Optimizer/Dialect/FIROps.cpp b/flang/lib/Optimizer/Dialect/FIROps.cpp
@@ -3993,6 +3993,25 @@ mlir::LogicalResult fir::CUDAKernelOp::verify() {
   return mlir::success();
 }
 
+mlir::LogicalResult fir::CUDAAllocateOp::verify() {
+  if (getPinned() && getStream())
+    return emitOpError("pinned and stream cannot appears at the same time");
+  if (!fir::unwrapRefType(getBox().getType()).isa<fir::BaseBoxType>())
+    return emitOpError(
+        "expect box to be a reference to/or a class or box type value");
+  if (getSource() &&
+      !fir::unwrapRefType(getSource().getType()).isa<fir::BaseBoxType>())
+    return emitOpError(
+        "expect source to be a reference to/or a class or box type value");
+  if (getErrmsg() &&
+      !fir::unwrapRefType(getErrmsg().getType()).isa<fir::BoxType>())
+    return emitOpError(
+        "expect errmsg to be a reference to/or a box type value");
+  if (getErrmsg() && !getHasStat())
+    return emitOpError("expect stat attribute when errmsg is provided");
+  return mlir::success();
+}
+
 //===----------------------------------------------------------------------===//
 // FIROpsDialect
 //===----------------------------------------------------------------------===//

diff --git a/flang/lib/Parser/openmp-parsers.cpp b/flang/lib/Parser/openmp-parsers.cpp
@@ -136,6 +136,11 @@ TYPE_PARSER(construct<OmpReductionOperator>(Parser<DefinedOperator>{}) ||
     construct<OmpReductionOperator>(Parser<ProcedureDesignator>{}))
 
 TYPE_PARSER(construct<OmpReductionClause>(
+    maybe(
+        ("INSCAN" >> pure(OmpReductionClause::ReductionModifier::Inscan) ||
+            "TASK" >> pure(OmpReductionClause::ReductionModifier::Task) ||
+            "DEFAULT" >> pure(OmpReductionClause::ReductionModifier::Default)) /
+        ","),
     Parser<OmpReductionOperator>{} / ":", Parser<OmpObjectList>{}))
 
 // OMP 5.0 2.19.5.6 IN_REDUCTION (reduction-identifier: variable-name-list)

diff --git a/flang/lib/Parser/unparse.cpp b/flang/lib/Parser/unparse.cpp
@@ -2090,6 +2090,8 @@ class UnparseVisitor {
     Walk(":", x.step);
   }
   void Unparse(const OmpReductionClause &x) {
+    Walk(std::get<std::optional<OmpReductionClause::ReductionModifier>>(x.t),
+        ",");
     Walk(std::get<OmpReductionOperator>(x.t));
     Put(":");
     Walk(std::get<OmpObjectList>(x.t));
@@ -2727,6 +2729,8 @@ class UnparseVisitor {
   WALK_NESTED_ENUM(OmpScheduleClause, ScheduleType) // OMP schedule-type
   WALK_NESTED_ENUM(OmpDeviceClause, DeviceModifier) // OMP device modifier
   WALK_NESTED_ENUM(OmpDeviceTypeClause, Type) // OMP DEVICE_TYPE
+  WALK_NESTED_ENUM(
+      OmpReductionClause, ReductionModifier) // OMP reduction-modifier
   WALK_NESTED_ENUM(OmpIfClause, DirectiveNameModifier) // OMP directive-modifier
   WALK_NESTED_ENUM(OmpCancelType, Type) // OMP cancel-type
   WALK_NESTED_ENUM(OmpOrderClause, Type) // OMP order-type

diff --git a/flang/lib/Semantics/check-omp-structure.cpp b/flang/lib/Semantics/check-omp-structure.cpp
@@ -2289,7 +2289,7 @@ void OmpStructureChecker::Enter(const parser::OmpClause::Reduction &x) {
 bool OmpStructureChecker::CheckReductionOperators(
     const parser::OmpClause::Reduction &x) {
 
-  const auto &definedOp{std::get<0>(x.v.t)};
+  const auto &definedOp{std::get<parser::OmpReductionOperator>(x.v.t)};
   bool ok = false;
   common::visit(
       common::visitors{

diff --git a/flang/test/Fir/cuf-invalid.fir b/flang/test/Fir/cuf-invalid.fir
@@ -0,0 +1,50 @@
+// RUN: fir-opt -split-input-file -verify-diagnostics %s
+
+func.func @_QPsub1() {
+  %0 = fir.alloca !fir.box<!fir.heap<!fir.array<?xf32>>> {bindc_name = "a", uniq_name = "_QFsub1Ea"}
+  %1 = fir.alloca i32
+  %pinned = fir.alloca i1
+  %4:2 = hlfir.declare %0 {cuda_attr = #fir.cuda<device>, fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFsub1Ea"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>)
+  %11 = fir.convert %4#1 : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> !fir.ref<!fir.box<none>>
+  %s = fir.load %1 : !fir.ref<i32>
+  // expected-error@+1{{'fir.cuda_allocate' op pinned and stream cannot appears at the same time}}
+  %13 = fir.cuda_allocate %11 : !fir.ref<!fir.box<none>> stream(%s : i32) pinned(%pinned : !fir.ref<i1>) {cuda_attr = #fir.cuda<device>} -> i32
+  return
+}
+
+// -----
+
+func.func @_QPsub1() {
+  %1 = fir.alloca i32
+  // expected-error@+1{{'fir.cuda_allocate' op expect box to be a reference to/or a class or box type value}}
+  %2 = fir.cuda_allocate %1 : !fir.ref<i32> {cuda_attr = #fir.cuda<device>} -> i32
+  return
+}
+
+// -----
+
+func.func @_QPsub1() {
+  %0 = fir.alloca !fir.box<!fir.heap<!fir.array<?xf32>>> {bindc_name = "a", uniq_name = "_QFsub1Ea"}
+  %4:2 = hlfir.declare %0 {cuda_attr = #fir.cuda<device>, fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFsub1Ea"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>)
+  %c100 = arith.constant 100 : index
+  %7 = fir.alloca !fir.char<1,100> {bindc_name = "msg", uniq_name = "_QFsub1Emsg"}
+  %8:2 = hlfir.declare %7 typeparams %c100 {uniq_name = "_QFsub1Emsg"} : (!fir.ref<!fir.char<1,100>>, index) -> (!fir.ref<!fir.char<1,100>>, !fir.ref<!fir.char<1,100>>)
+  %9 = fir.embox %8#1 : (!fir.ref<!fir.char<1,100>>) -> !fir.box<!fir.char<1,100>>
+  %11 = fir.convert %4#1 : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> !fir.ref<!fir.box<none>>
+  %16 = fir.convert %9 : (!fir.box<!fir.char<1,100>>) -> !fir.box<none>
+  // expected-error@+1{{'fir.cuda_allocate' op expect stat attribute when errmsg is provided}}
+  %13 = fir.cuda_allocate %11 : !fir.ref<!fir.box<none>> errmsg(%16 : !fir.box<none>) {cuda_attr = #fir.cuda<device>} -> i32
+  return
+}
+
+// -----
+
+func.func @_QPsub1() {
+  %0 = fir.alloca !fir.box<!fir.heap<!fir.array<?xf32>>> {bindc_name = "a", uniq_name = "_QFsub1Ea"}
+  %4:2 = hlfir.declare %0 {cuda_attr = #fir.cuda<device>, fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFsub1Ea"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>)
+  %1 = fir.alloca i32
+  %11 = fir.convert %4#1 : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> !fir.ref<!fir.box<none>>
+  // expected-error@+1{{'fir.cuda_allocate' op expect errmsg to be a reference to/or a box type value}}
+  %13 = fir.cuda_allocate %11 : !fir.ref<!fir.box<none>> errmsg(%1 : !fir.ref<i32>) {cuda_attr = #fir.cuda<device>, hasStat} -> i32
+  return
+}
diff --git a/flang/test/Fir/cuf.mlir b/flang/test/Fir/cuf.mlir
@@ -0,0 +1,70 @@
+// RUN: fir-opt --split-input-file %s | fir-opt --split-input-file | FileCheck %s
+
+// Simple round trip test of operations.
+
+func.func @_QPsub1() {
+  %0 = fir.alloca !fir.box<!fir.heap<!fir.array<?xf32>>> {bindc_name = "a", uniq_name = "_QFsub1Ea"}
+  %4:2 = hlfir.declare %0 {cuda_attr = #fir.cuda<device>, fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFsub1Ea"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>)
+  %11 = fir.convert %4#1 : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> !fir.ref<!fir.box<none>>
+  %13 = fir.cuda_allocate %11 : !fir.ref<!fir.box<none>> {cuda_attr = #fir.cuda<device>} -> i32
+  return
+}
+
+// CHECK: fir.cuda_allocate %{{.*}} : !fir.ref<!fir.box<none>> {cuda_attr = #fir.cuda<device>} -> i32
+
+// -----
+
+func.func @_QPsub1() {
+  %0 = fir.alloca !fir.box<!fir.heap<!fir.array<?xf32>>> {bindc_name = "a", uniq_name = "_QFsub1Ea"}
+  %1 = fir.alloca i32
+  %4:2 = hlfir.declare %0 {cuda_attr = #fir.cuda<device>, fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFsub1Ea"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>)
+  %11 = fir.convert %4#1 : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> !fir.ref<!fir.box<none>>
+  %s = fir.load %1 : !fir.ref<i32>
+  %13 = fir.cuda_allocate %11 : !fir.ref<!fir.box<none>> stream(%s : i32) {cuda_attr = #fir.cuda<device>} -> i32
+  return
+}
+
+// CHECK: fir.cuda_allocate %{{.*}} : !fir.ref<!fir.box<none>> stream(%{{.*}} : i32) {cuda_attr = #fir.cuda<device>} -> i32
+
+// -----
+
+func.func @_QPsub1() {
+  %0 = fir.alloca !fir.box<!fir.heap<!fir.array<?xf32>>> {bindc_name = "a", uniq_name = "_QFsub1Ea"}
+  %1 = fir.alloca !fir.box<!fir.heap<!fir.array<?xf32>>> {bindc_name = "b", uniq_name = "_QFsub1Eb"}
+  %4:2 = hlfir.declare %0 {cuda_attr = #fir.cuda<device>, fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFsub1Ea"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>)
+  %5:2 = hlfir.declare %1 {fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFsub1Ea"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>)
+  %11 = fir.convert %4#1 : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> !fir.ref<!fir.box<none>>
+  %12 = fir.convert %5#1 : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> !fir.ref<!fir.box<none>>
+  %13 = fir.cuda_allocate %11 : !fir.ref<!fir.box<none>> source(%12 : !fir.ref<!fir.box<none>>) {cuda_attr = #fir.cuda<device>} -> i32
+  return
+}
+
+// CHECK: fir.cuda_allocate %{{.*}} : !fir.ref<!fir.box<none>> source(%{{.*}} : !fir.ref<!fir.box<none>>) {cuda_attr = #fir.cuda<device>} -> i32
+
+// -----
+
+func.func @_QPsub1() {
+  %0 = fir.alloca !fir.box<!fir.heap<!fir.array<?xf32>>> {bindc_name = "a", uniq_name = "_QFsub1Ea"}
+  %pinned = fir.alloca i1
+  %4:2 = hlfir.declare %0 {cuda_attr = #fir.cuda<device>, fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFsub1Ea"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>)
+  %11 = fir.convert %4#1 : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> !fir.ref<!fir.box<none>>
+  %13 = fir.cuda_allocate %11 : !fir.ref<!fir.box<none>> pinned(%pinned : !fir.ref<i1>) {cuda_attr = #fir.cuda<device>} -> i32
+  return
+}
+
+// CHECK: fir.cuda_allocate %{{.*}} : !fir.ref<!fir.box<none>> pinned(%{{.*}} : !fir.ref<i1>) {cuda_attr = #fir.cuda<device>} -> i32
+
+// -----
+
+func.func @_QPsub1() {
+  %0 = fir.alloca !fir.box<!fir.heap<!fir.array<?xf32>>> {bindc_name = "a", uniq_name = "_QFsub1Ea"}
+  %4:2 = hlfir.declare %0 {cuda_attr = #fir.cuda<device>, fortran_attrs = #fir.var_attrs<allocatable>, uniq_name = "_QFsub1Ea"} : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>, !fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>)
+  %c100 = arith.constant 100 : index
+  %7 = fir.alloca !fir.char<1,100> {bindc_name = "msg", uniq_name = "_QFsub1Emsg"}
+  %8:2 = hlfir.declare %7 typeparams %c100 {uniq_name = "_QFsub1Emsg"} : (!fir.ref<!fir.char<1,100>>, index) -> (!fir.ref<!fir.char<1,100>>, !fir.ref<!fir.char<1,100>>)
+  %9 = fir.embox %8#1 : (!fir.ref<!fir.char<1,100>>) -> !fir.box<!fir.char<1,100>>
+  %11 = fir.convert %4#1 : (!fir.ref<!fir.box<!fir.heap<!fir.array<?xf32>>>>) -> !fir.ref<!fir.box<none>>
+  %16 = fir.convert %9 : (!fir.box<!fir.char<1,100>>) -> !fir.box<none>
+  %13 = fir.cuda_allocate %11 : !fir.ref<!fir.box<none>> errmsg(%16 : !fir.box<none>) {cuda_attr = #fir.cuda<device>, hasStat} -> i32
+  return
+}
diff --git a/flang/test/Lower/OpenMP/Todo/reduction-modifiers.f90 b/flang/test/Lower/OpenMP/Todo/reduction-modifiers.f90
@@ -0,0 +1,13 @@
+! RUN: %not_todo_cmd bbc -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
+! RUN: %not_todo_cmd %flang_fc1 -emit-fir -fopenmp -o - %s 2>&1 | FileCheck %s
+
+! CHECK: not yet implemented: Reduction modifiers are not supported
+
+subroutine foo()
+  integer :: i, j
+  j = 0
+  !$omp do reduction (inscan, *: j)
+  do i = 1, 10
+    j = j + 1
+  end do
+end subroutine
diff --git a/flang/test/Parser/OpenMP/reduction-modifier.f90 b/flang/test/Parser/OpenMP/reduction-modifier.f90
@@ -0,0 +1,20 @@
+! RUN: %flang_fc1 -fdebug-unparse-no-sema -fopenmp %s | FileCheck --ignore-case %s
+! RUN: %flang_fc1 -fdebug-dump-parse-tree-no-sema -fopenmp %s | FileCheck --check-prefix="PARSE-TREE" %s
+
+subroutine foo()
+  integer :: i, j
+  j = 0
+! CHECK: !$OMP DO  REDUCTION(TASK,*:j)
+! PARSE-TREE: | | ExecutionPartConstruct -> ExecutableConstruct -> OpenMPConstruct -> OpenMPLoopConstruct
+! PARSE-TREE: | | | OmpBeginLoopDirective
+! PARSE-TREE: | | | | OmpLoopDirective -> llvm::omp::Directive = do
+! PARSE-TREE: | | | | OmpClauseList -> OmpClause -> Reduction -> OmpReductionClause
+! PARSE-TREE: | | | | | ReductionModifier = Task
+! PARSE-TREE: | | | | | OmpReductionOperator -> DefinedOperator -> IntrinsicOperator = Multiply
+! PARSE-TREE: | | | | | OmpObjectList -> OmpObject -> Designator -> DataRef -> Name = 'j
+  !$omp do reduction (task, *: j)
+  do i = 1, 10
+    j = j + 1
+  end do
+  !$omp end do
+end
diff --git a/libc/hdr/types/CMakeLists.txt b/libc/hdr/types/CMakeLists.txt
@@ -21,3 +21,21 @@ add_proxy_header_library(
   FULL_BUILD_DEPENDS
     libc.include.llvm-libc-types.struct_timespec
 )
+
+add_proxy_header_library(
+  fenv_t
+  HDRS
+    fenv_t.h
+  FULL_BUILD_DEPENDS
+    libc.include.llvm-libc-types.fenv_t
+    libc.incude.fenv
+)
+
+add_proxy_header_library(
+  fexcept_t
+  HDRS
+    fexcept_t.h
+  FULL_BUILD_DEPENDS
+    libc.include.llvm-libc-types.fexcept_t
+    libc.incude.fenv
+)
diff --git a/libc/hdr/types/fenv_t.h b/libc/hdr/types/fenv_t.h
@@ -0,0 +1,22 @@
+//===-- Definition of macros from fenv_t.h --------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_HDR_FENV_T_H
+#define LLVM_LIBC_HDR_FENV_T_H
+
+#ifdef LIBC_FULL_BUILD
+
+#include "include/llvm-libc-types/fenv_t.h"
+
+#else // Overlay mode
+
+#include <fenv.h>
+
+#endif // LLVM_LIBC_FULL_BUILD
+
+#endif // LLVM_LIBC_HDR_FENV_T_H
diff --git a/libc/hdr/types/fexcept_t.h b/libc/hdr/types/fexcept_t.h
@@ -0,0 +1,22 @@
+//===-- Definition of macros from fexcept_t.h -----------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_HDR_FEXCEPT_T_H
+#define LLVM_LIBC_HDR_FEXCEPT_T_H
+
+#ifdef LIBC_FULL_BUILD
+
+#include "include/llvm-libc-types/fexcept_t.h"
+
+#else // Overlay mode
+
+#include <fenv.h>
+
+#endif // LLVM_LIBC_FULL_BUILD
+
+#endif // LLVM_LIBC_HDR_FENV_T_H
diff --git a/libc/src/__support/FPUtil/CMakeLists.txt b/libc/src/__support/FPUtil/CMakeLists.txt
@@ -3,7 +3,7 @@ add_header_library(
   HDRS
     FEnvImpl.h
   DEPENDS
-    libc.include.fenv
+    libc.hdr.types.fenv_t
     libc.hdr.fenv_macros
     libc.hdr.math_macros
     libc.src.__support.macros.attributes

diff --git a/libc/src/__support/FPUtil/aarch64/FEnvImpl.h b/libc/src/__support/FPUtil/aarch64/FEnvImpl.h
@@ -17,10 +17,10 @@
 #endif
 
 #include <arm_acle.h>
-#include <fenv.h>
 #include <stdint.h>
 
 #include "hdr/fenv_macros.h"
+#include "hdr/types/fenv_t.h"
 #include "src/__support/FPUtil/FPBits.h"
 
 namespace LIBC_NAMESPACE {

diff --git a/libc/src/__support/FPUtil/aarch64/fenv_darwin_impl.h b/libc/src/__support/FPUtil/aarch64/fenv_darwin_impl.h
@@ -17,10 +17,10 @@
 #endif
 
 #include <arm_acle.h>
-#include <fenv.h>
 #include <stdint.h>
 
 #include "hdr/fenv_macros.h"
+#include "hdr/types/fenv_t.h"
 #include "src/__support/FPUtil/FPBits.h"
 
 namespace LIBC_NAMESPACE {

diff --git a/libc/src/__support/FPUtil/arm/FEnvImpl.h b/libc/src/__support/FPUtil/arm/FEnvImpl.h
@@ -10,9 +10,9 @@
 #define LLVM_LIBC_SRC___SUPPORT_FPUTIL_ARM_FENVIMPL_H
 
 #include "hdr/fenv_macros.h"
+#include "hdr/types/fenv_t.h"
 #include "src/__support/FPUtil/FPBits.h"
 #include "src/__support/macros/attributes.h" // For LIBC_INLINE
-#include <fenv.h>
 #include <stdint.h>
 
 namespace LIBC_NAMESPACE {

diff --git a/libc/src/__support/FPUtil/riscv/FEnvImpl.h b/libc/src/__support/FPUtil/riscv/FEnvImpl.h
@@ -10,6 +10,7 @@
 #define LLVM_LIBC_SRC___SUPPORT_FPUTIL_RISCV_FENVIMPL_H
 
 #include "hdr/fenv_macros.h"
+#include "hdr/types/fenv_t.h"
 #include "src/__support/FPUtil/FPBits.h"
 #include "src/__support/macros/attributes.h" // For LIBC_INLINE_ASM
 #include "src/__support/macros/config.h"     // For LIBC_INLINE

diff --git a/libc/src/__support/FPUtil/x86_64/FEnvImpl.h b/libc/src/__support/FPUtil/x86_64/FEnvImpl.h
@@ -16,9 +16,9 @@
 #error "Invalid include"
 #endif
 
-#include <fenv.h>
 #include <stdint.h>
 
+#include "hdr/types/fenv_t.h"
 #include "src/__support/macros/sanitizer.h"
 
 namespace LIBC_NAMESPACE {

diff --git a/libc/src/fenv/CMakeLists.txt b/libc/src/fenv/CMakeLists.txt
@@ -5,7 +5,6 @@ add_entrypoint_object(
   HDRS
     fegetround.h
   DEPENDS
-    libc.include.fenv
     libc.src.__support.FPUtil.fenv_impl
   COMPILE_OPTIONS
     -O2
@@ -71,6 +70,7 @@ add_entrypoint_object(
     fegetenv.h
   DEPENDS
     libc.hdr.fenv_macros
+    libc.hdr.types.fenv_t
     libc.src.__support.FPUtil.fenv_impl
   COMPILE_OPTIONS
     -O2
@@ -84,6 +84,7 @@ add_entrypoint_object(
     fesetenv.h
   DEPENDS
     libc.hdr.fenv_macros
+    libc.hdr.types.fenv_t
     libc.src.__support.FPUtil.fenv_impl
   COMPILE_OPTIONS
     -O2
@@ -97,6 +98,7 @@ add_entrypoint_object(
     fegetexceptflag.h
   DEPENDS
     libc.hdr.fenv_macros
+    libc.hdr.types.fexcept_t
     libc.src.__support.FPUtil.fenv_impl
   COMPILE_OPTIONS
     -O2
@@ -123,6 +125,7 @@ add_entrypoint_object(
     fesetexceptflag.h
   DEPENDS
     libc.hdr.fenv_macros
+    libc.hdr.types.fexcept_t
     libc.src.__support.FPUtil.fenv_impl
   COMPILE_OPTIONS
     -O2
@@ -136,6 +139,7 @@ add_entrypoint_object(
     feholdexcept.h
   DEPENDS
     libc.hdr.fenv_macros
+    libc.hdr.types.fenv_t
     libc.src.__support.FPUtil.fenv_impl
   COMPILE_OPTIONS
     -O2
@@ -149,6 +153,7 @@ add_entrypoint_object(
     feupdateenv.h
   DEPENDS
     libc.hdr.fenv_macros
+    libc.hdr.types.fenv_t
     libc.src.__support.FPUtil.fenv_impl
   COMPILE_OPTIONS
     -O2

diff --git a/libc/src/fenv/fegetenv.h b/libc/src/fenv/fegetenv.h
@@ -9,7 +9,7 @@
 #ifndef LLVM_LIBC_SRC_FENV_FEGETENV_H
 #define LLVM_LIBC_SRC_FENV_FEGETENV_H
 
-#include <fenv.h>
+#include "hdr/types/fenv_t.h"
 
 namespace LIBC_NAMESPACE {
 

diff --git a/libc/src/fenv/fegetexceptflag.cpp b/libc/src/fenv/fegetexceptflag.cpp
@@ -7,11 +7,10 @@
 //===----------------------------------------------------------------------===//
 
 #include "src/fenv/fegetexceptflag.h"
+#include "hdr/types/fexcept_t.h"
 #include "src/__support/FPUtil/FEnvImpl.h"
 #include "src/__support/common.h"
 
-#include <fenv.h>
-
 namespace LIBC_NAMESPACE {
 
 LLVM_LIBC_FUNCTION(int, fegetexceptflag, (fexcept_t * flagp, int excepts)) {

diff --git a/libc/src/fenv/feholdexcept.cpp b/libc/src/fenv/feholdexcept.cpp
@@ -7,9 +7,9 @@
 //===----------------------------------------------------------------------===//
 
 #include "src/fenv/feholdexcept.h"
+#include "hdr/types/fenv_t.h"
 #include "src/__support/FPUtil/FEnvImpl.h"
 #include "src/__support/common.h"
-#include <fenv.h>
 
 namespace LIBC_NAMESPACE {
 

diff --git a/libc/src/fenv/feholdexcept.h b/libc/src/fenv/feholdexcept.h
@@ -9,7 +9,7 @@
 #ifndef LLVM_LIBC_SRC_FENV_FEHOLDEXCEPT_H
 #define LLVM_LIBC_SRC_FENV_FEHOLDEXCEPT_H
 
-#include <fenv.h>
+#include "hdr/types/fenv_t.h"
 
 namespace LIBC_NAMESPACE {
 

diff --git a/libc/src/fenv/fesetenv.h b/libc/src/fenv/fesetenv.h
@@ -9,7 +9,7 @@
 #ifndef LLVM_LIBC_SRC_FENV_FESETENV_H
 #define LLVM_LIBC_SRC_FENV_FESETENV_H
 
-#include <fenv.h>
+#include "hdr/types/fenv_t.h"
 
 namespace LIBC_NAMESPACE {
 

diff --git a/libc/src/fenv/fesetexceptflag.cpp b/libc/src/fenv/fesetexceptflag.cpp
@@ -7,9 +7,9 @@
 //===----------------------------------------------------------------------===//
 
 #include "src/fenv/fesetexceptflag.h"
+#include "hdr/types/fexcept_t.h"
 #include "src/__support/FPUtil/FEnvImpl.h"
 #include "src/__support/common.h"
-#include <fenv.h>
 
 namespace LIBC_NAMESPACE {
 

diff --git a/libc/src/fenv/feupdateenv.h b/libc/src/fenv/feupdateenv.h
@@ -9,7 +9,7 @@
 #ifndef LLVM_LIBC_SRC_FENV_FEUPDATEENV_H
 #define LLVM_LIBC_SRC_FENV_FEUPDATEENV_H
 
-#include <fenv.h>
+#include "hdr/types/fenv_t.h"
 
 namespace LIBC_NAMESPACE {
 

diff --git a/libc/test/UnitTest/CMakeLists.txt b/libc/test/UnitTest/CMakeLists.txt
@@ -135,6 +135,7 @@ add_unittest_framework_library(
     LibcTest
     libc.src.__support.FPUtil.fp_bits
     libc.src.__support.FPUtil.fenv_impl
+    libc.hdr.types.fenv_t
 )
 
 add_unittest_framework_library(