Merge pull request llvm#332 from AMD-Lightning-Internal/upstream_merge_202501272321

merge main into amd-staging

Author: saiislam

clang/cmake/caches/Fuchsia-stage2.cmake

@@ -190,6 +190,10 @@ foreach(target aarch64-unknown-linux-gnu;armv7-unknown-linux-gnueabihf;i386-unkn
     set(RUNTIMES_${target}_LLVM_TOOLS_DIR "${CMAKE_BINARY_DIR}/bin" CACHE BOOL "")
     set(RUNTIMES_${target}_LLVM_ENABLE_RUNTIMES "compiler-rt;libcxx;libcxxabi;libunwind" CACHE STRING "")
 
+    # Enable FatLTO for Linux and baremetal runtimes
+    set(RUNTIMES_${target}_LLVM_ENABLE_LTO ON CACHE BOOL "")
+    set(RUNTIMES_${target}_LLVM_ENABLE_FATLTO ON CACHE BOOL "")
+
     # Use .build-id link.
     list(APPEND RUNTIME_BUILD_ID_LINK "${target}")
   endif()
@@ -272,6 +276,10 @@ if(FUCHSIA_SDK)
     set(RUNTIMES_${target}+asan+noexcept_LIBCXXABI_ENABLE_EXCEPTIONS OFF CACHE BOOL "")
     set(RUNTIMES_${target}+asan+noexcept_LIBCXX_ENABLE_EXCEPTIONS OFF CACHE BOOL "")
 
+    # Enable FatLTO for Fuchsia runtimes
+    set(RUNTIMES_${target}_LLVM_ENABLE_LTO ON CACHE BOOL "")
+    set(RUNTIMES_${target}_LLVM_ENABLE_FATLTO ON CACHE BOOL "")
+
     # Use .build-id link.
     list(APPEND RUNTIME_BUILD_ID_LINK "${target}")
   endforeach()
@@ -369,6 +377,10 @@ foreach(target armv6m-none-eabi;armv7m-none-eabi;armv7em-none-eabi;armv8m.main-n
   set(RUNTIMES_${target}_LLVM_INCLUDE_TESTS OFF CACHE BOOL "")
   set(RUNTIMES_${target}_LLVM_ENABLE_ASSERTIONS OFF CACHE BOOL "")
   set(RUNTIMES_${target}_LLVM_ENABLE_RUNTIMES "libc;libcxx" CACHE STRING "")
+
+  # Enable FatLTO for baremetal runtimes
+  set(RUNTIMES_${target}_LLVM_ENABLE_LTO ON CACHE BOOL "")
+  set(RUNTIMES_${target}_LLVM_ENABLE_FATLTO ON CACHE BOOL "")
 endforeach()
 
 foreach(target riscv32-unknown-elf)
@@ -420,6 +432,10 @@ foreach(target riscv32-unknown-elf)
   set(RUNTIMES_${target}_LLVM_INCLUDE_TESTS OFF CACHE BOOL "")
   set(RUNTIMES_${target}_LLVM_ENABLE_ASSERTIONS OFF CACHE BOOL "")
   set(RUNTIMES_${target}_LLVM_ENABLE_RUNTIMES "libc;libcxx" CACHE STRING "")
+
+  # Enable FatLTO for baremetal runtimes
+  set(RUNTIMES_${target}_LLVM_ENABLE_LTO ON CACHE BOOL "")
+  set(RUNTIMES_${target}_LLVM_ENABLE_FATLTO ON CACHE BOOL "")
 endforeach()
 
 set(LLVM_BUILTIN_TARGETS "${BUILTIN_TARGETS}" CACHE STRING "")

clang/docs/LanguageExtensions.rst

@@ -514,9 +514,7 @@ available in all language modes.
 __nullptr
 ---------
 
-``__nullptr`` is an alternate spelling for ``nullptr``, but is also available in
-C++ modes prior to C++11. Note that it's currently not availbale in C despite
-C23 having support for ``nullptr``.
+``__nullptr`` is an alternate spelling for ``nullptr``. It is available in all C and C++ language modes.
 
 __signed, __signed__
 --------------------

clang/docs/ReleaseNotes.rst

@@ -396,6 +396,7 @@ C Language Changes
 
 - Extend clang's ``<limits.h>`` to define ``LONG_LONG_*`` macros for Android's bionic.
 - Macro ``__STDC_NO_THREADS__`` is no longer necessary for MSVC 2022 1939 and later.
+- Exposed the the ``__nullptr`` keyword as an alias for ``nullptr`` in all C language modes.
 
 C2y Feature Support
 ^^^^^^^^^^^^^^^^^^^
@@ -1035,6 +1036,7 @@ Bug Fixes to C++ Support
 - Fixed assertions or false compiler diagnostics in the case of C++ modules for
   lambda functions or inline friend functions defined inside templates (#GH122493).
 - Clang now rejects declaring an alias template with the same name as its template parameter. (#GH123423)
+- Fixed the rejection of valid code when referencing an enumerator of an unscoped enum member with a prior declaration. (#GH124405)
 - Fixed immediate escalation of non-dependent expressions. (#GH123405)
 - Fix type of expression when calling a template which returns an ``__array_rank`` querying a type depending on a
   template parameter. Now, such expression can be used with ``static_assert`` and ``constexpr``. (#GH123498)

clang/include/clang/AST/ASTLambda.h

@@ -17,6 +17,7 @@
 
 #include "clang/AST/DeclCXX.h"
 #include "clang/AST/DeclTemplate.h"
+#include "llvm/Support/Casting.h"
 
 namespace clang {
 inline StringRef getLambdaStaticInvokerName() {
@@ -35,6 +36,12 @@ inline bool isLambdaCallOperator(const DeclContext *DC) {
   return isLambdaCallOperator(cast<CXXMethodDecl>(DC));
 }
 
+inline bool isLambdaMethod(const DeclContext *DC) {
+  if (const auto *MD = dyn_cast_if_present<CXXMethodDecl>(DC))
+    return MD->getParent()->isLambda();
+  return false;
+}
+
 inline bool isLambdaCallWithExplicitObjectParameter(const DeclContext *DC) {
   return isLambdaCallOperator(DC) &&
          cast<CXXMethodDecl>(DC)->isExplicitObjectMemberFunction();

clang/include/clang/AST/DeclCXX.h

@@ -1967,24 +1967,42 @@ class ExplicitSpecifier {
 class CXXDeductionGuideDecl : public FunctionDecl {
   void anchor() override;
 
+public:
+  // Represents the relationship between this deduction guide and the
+  // deduction guide that it was generated from (or lack thereof).
+  // See the SourceDeductionGuide member for more details.
+  enum class SourceDeductionGuideKind : uint8_t {
+    None,
+    Alias,
+  };
+
 private:
   CXXDeductionGuideDecl(ASTContext &C, DeclContext *DC, SourceLocation StartLoc,
                         ExplicitSpecifier ES,
                         const DeclarationNameInfo &NameInfo, QualType T,
                         TypeSourceInfo *TInfo, SourceLocation EndLocation,
                         CXXConstructorDecl *Ctor, DeductionCandidate Kind,
-                        Expr *TrailingRequiresClause)
+                        Expr *TrailingRequiresClause,
+                        const CXXDeductionGuideDecl *GeneratedFrom,
+                        SourceDeductionGuideKind SourceKind)
       : FunctionDecl(CXXDeductionGuide, C, DC, StartLoc, NameInfo, T, TInfo,
                      SC_None, false, false, ConstexprSpecKind::Unspecified,
                      TrailingRequiresClause),
-        Ctor(Ctor), ExplicitSpec(ES) {
+        Ctor(Ctor), ExplicitSpec(ES),
+        SourceDeductionGuide(GeneratedFrom, SourceKind) {
     if (EndLocation.isValid())
       setRangeEnd(EndLocation);
     setDeductionCandidateKind(Kind);
   }
 
   CXXConstructorDecl *Ctor;
   ExplicitSpecifier ExplicitSpec;
+  // The deduction guide, if any, that this deduction guide was generated from,
+  // in the case of alias template deduction. The SourceDeductionGuideKind
+  // member indicates which of these sources applies, or is None otherwise.
+  llvm::PointerIntPair<const CXXDeductionGuideDecl *, 2,
+                       SourceDeductionGuideKind>
+      SourceDeductionGuide;
   void setExplicitSpecifier(ExplicitSpecifier ES) { ExplicitSpec = ES; }
 
 public:
@@ -1997,7 +2015,9 @@ class CXXDeductionGuideDecl : public FunctionDecl {
          TypeSourceInfo *TInfo, SourceLocation EndLocation,
          CXXConstructorDecl *Ctor = nullptr,
          DeductionCandidate Kind = DeductionCandidate::Normal,
-         Expr *TrailingRequiresClause = nullptr);
+         Expr *TrailingRequiresClause = nullptr,
+         const CXXDeductionGuideDecl *SourceDG = nullptr,
+         SourceDeductionGuideKind SK = SourceDeductionGuideKind::None);
 
   static CXXDeductionGuideDecl *CreateDeserialized(ASTContext &C,
                                                    GlobalDeclID ID);
@@ -2017,6 +2037,25 @@ class CXXDeductionGuideDecl : public FunctionDecl {
   /// this is an implicit deduction guide.
   CXXConstructorDecl *getCorrespondingConstructor() const { return Ctor; }
 
+  /// Get the deduction guide from which this deduction guide was generated,
+  /// if it was generated as part of alias template deduction or from an
+  /// inherited constructor.
+  const CXXDeductionGuideDecl *getSourceDeductionGuide() const {
+    return SourceDeductionGuide.getPointer();
+  }
+
+  void setSourceDeductionGuide(CXXDeductionGuideDecl *DG) {
+    SourceDeductionGuide.setPointer(DG);
+  }
+
+  SourceDeductionGuideKind getSourceDeductionGuideKind() const {
+    return SourceDeductionGuide.getInt();
+  }
+
+  void setSourceDeductionGuideKind(SourceDeductionGuideKind SK) {
+    SourceDeductionGuide.setInt(SK);
+  }
+
   void setDeductionCandidateKind(DeductionCandidate K) {
     FunctionDeclBits.DeductionCandidateKind = static_cast<unsigned char>(K);
   }

clang/include/clang/Basic/TokenKinds.def

@@ -707,7 +707,7 @@ ALIAS("__decltype"       , decltype     , KEYCXX)
 ALIAS("__imag__"         , __imag       , KEYALL)
 ALIAS("__inline"         , inline       , KEYALL)
 ALIAS("__inline__"       , inline       , KEYALL)
-ALIAS("__nullptr"        , nullptr      , KEYCXX)
+ALIAS("__nullptr"        , nullptr      , KEYALL)
 ALIAS("__real__"         , __real       , KEYALL)
 ALIAS("__restrict"       , restrict     , KEYALL)
 ALIAS("__restrict__"     , restrict     , KEYALL)

clang/include/clang/Basic/arm_neon.td

@@ -2141,6 +2141,50 @@ let ArchGuard = "defined(__aarch64__)", TargetGuard = "fp8,neon" in {
   def VCVTN_F8_F16      : VInst<"vcvt_mf8_f16_fpm",      ".(>F)(>F)V",      "mQm">;
 }
 
+let ArchGuard = "defined(__aarch64__)", TargetGuard = "fp8dot2,neon" in {
+  def VDOT_F16_MF8 : VInst<"vdot_f16_mf8_fpm", "(>F)(>F)..V", "mQm">;
+
+  def VDOT_LANE_F16_MF8  : VInst<"vdot_lane_f16_mf8_fpm",  "(>F)(>F)..IV", "m", [ImmCheck<3, ImmCheck0_3, 0>]>;
+  def VDOT_LANEQ_F16_MF8 : VInst<"vdot_laneq_f16_mf8_fpm", "(>F)(>F).QIV", "m", [ImmCheck<3, ImmCheck0_7, 0>]>;
+
+  def VDOTQ_LANE_F16_MF8  : VInst<"vdot_lane_f16_mf8_fpm",  "(>F)(>F).qIV", "Qm", [ImmCheck<3, ImmCheck0_3, 0>]>;
+  def VDOTQ_LANEQ_F16_MF8 : VInst<"vdot_laneq_f16_mf8_fpm", "(>F)(>F)..IV", "Qm", [ImmCheck<3, ImmCheck0_7, 0>]>;
+}
+
+let ArchGuard = "defined(__aarch64__)", TargetGuard = "fp8dot4,neon" in {
+  def VDOT_F32_MF8 : VInst<"vdot_f32_mf8_fpm", "(>>F)(>>F)..V", "mQm">;
+
+  def VDOT_LANE_F32_MF8  : VInst<"vdot_lane_f32_mf8_fpm",  "(>>F)(>>F)..IV", "m", [ImmCheck<3, ImmCheck0_1, 0>]>;
+  def VDOT_LANEQ_F32_MF8 : VInst<"vdot_laneq_f32_mf8_fpm", "(>>F)(>>F).QIV", "m", [ImmCheck<3, ImmCheck0_3, 0>]>;
+
+  def VDOTQ_LANE_F32_MF8  : VInst<"vdot_lane_f32_mf8_fpm",  "(>>F)(>>F).qIV", "Qm", [ImmCheck<3, ImmCheck0_1, 0>]>;
+  def VDOTQ_LANEQ_F32_MF8 : VInst<"vdot_laneq_f32_mf8_fpm", "(>>F)(>>F)..IV", "Qm", [ImmCheck<3, ImmCheck0_3, 0>]>;
+}
+
+let ArchGuard = "defined(__aarch64__)", TargetGuard = "fp8fma,neon" in {
+  def VMLALB_F16_F8 : VInst<"vmlalb_f16_mf8_fpm", "(>F)(>F)..V", "Qm">;
+  def VMLALT_F16_F8 : VInst<"vmlalt_f16_mf8_fpm", "(>F)(>F)..V", "Qm">;
+
+  def VMLALLBB_F32_F8 : VInst<"vmlallbb_f32_mf8_fpm", "(>>F)(>>F)..V", "Qm">;
+  def VMLALLBT_F32_F8 : VInst<"vmlallbt_f32_mf8_fpm", "(>>F)(>>F)..V", "Qm">;
+  def VMLALLTB_F32_F8 : VInst<"vmlalltb_f32_mf8_fpm", "(>>F)(>>F)..V", "Qm">;
+  def VMLALLTT_F32_F8 : VInst<"vmlalltt_f32_mf8_fpm", "(>>F)(>>F)..V", "Qm">;
+
+  def VMLALB_F16_F8_LANE  : VInst<"vmlalb_lane_f16_mf8_fpm",  "(>F)(>F).qIV", "Qm", [ImmCheck<3, ImmCheck0_7,  0>]>;
+  def VMLALB_F16_F8_LANEQ : VInst<"vmlalb_laneq_f16_mf8_fpm", "(>F)(>F)..IV", "Qm", [ImmCheck<3, ImmCheck0_15, 0>]>;
+  def VMLALT_F16_F8_LANE  : VInst<"vmlalt_lane_f16_mf8_fpm",  "(>F)(>F).qIV", "Qm", [ImmCheck<3, ImmCheck0_7,  0>]>;
+  def VMLALT_F16_F8_LANEQ : VInst<"vmlalt_laneq_f16_mf8_fpm", "(>F)(>F)..IV", "Qm", [ImmCheck<3, ImmCheck0_15, 0>]>;
+
+  def VMLALLBB_F32_F8_LANE  : VInst<"vmlallbb_lane_f32_mf8_fpm",  "(>>F)(>>F).qIV", "Qm", [ImmCheck<3, ImmCheck0_7,  0>]>;
+  def VMLALLBB_F32_F8_LANEQ : VInst<"vmlallbb_laneq_f32_mf8_fpm", "(>>F)(>>F)..IV", "Qm", [ImmCheck<3, ImmCheck0_15, 0>]>;
+  def VMLALLBT_F32_F8_LANE  : VInst<"vmlallbt_lane_f32_mf8_fpm",  "(>>F)(>>F).qIV", "Qm", [ImmCheck<3, ImmCheck0_7,  0>]>;
+  def VMLALLBT_F32_F8_LANEQ : VInst<"vmlallbt_laneq_f32_mf8_fpm", "(>>F)(>>F)..IV", "Qm", [ImmCheck<3, ImmCheck0_15, 0>]>;
+  def VMLALLTB_F32_F8_LANE  : VInst<"vmlalltb_lane_f32_mf8_fpm",  "(>>F)(>>F).qIV", "Qm", [ImmCheck<3, ImmCheck0_7,  0>]>;
+  def VMLALLTB_F32_F8_LANEQ : VInst<"vmlalltb_laneq_f32_mf8_fpm", "(>>F)(>>F)..IV", "Qm", [ImmCheck<3, ImmCheck0_15, 0>]>;
+  def VMLALLTT_F32_F8_LANE  : VInst<"vmlalltt_lane_f32_mf8_fpm",  "(>>F)(>>F).qIV", "Qm", [ImmCheck<3, ImmCheck0_7,  0>]>;
+  def VMLALLTT_F32_F8_LANEQ : VInst<"vmlalltt_laneq_f32_mf8_fpm", "(>>F)(>>F)..IV", "Qm", [ImmCheck<3, ImmCheck0_15, 0>]>;
+}
+
 let ArchGuard = "defined(__aarch64__)", TargetGuard = "neon,faminmax" in {
   def FAMIN : WInst<"vamin", "...", "fhQdQfQh">;
   def FAMAX : WInst<"vamax", "...", "fhQdQfQh">;

clang/include/clang/Basic/arm_neon_incl.td

@@ -302,7 +302,7 @@ class Inst <string n, string p, string t, Operation o, list<ImmCheck> ch = []>{
 class SInst<string n, string p, string t, list<ImmCheck> ch = []> : Inst<n, p, t, OP_NONE, ch> {}
 class IInst<string n, string p, string t, list<ImmCheck> ch = []> : Inst<n, p, t, OP_NONE, ch> {}
 class WInst<string n, string p, string t, list<ImmCheck> ch = []> : Inst<n, p, t, OP_NONE, ch> {}
-class VInst<string n, string p, string t> : Inst<n, p, t, OP_NONE> {}
+class VInst<string n, string p, string t, list<ImmCheck> ch = []> : Inst<n, p, t, OP_NONE, ch> {}
 
 // The following instruction classes are implemented via operators
 // instead of builtins. As such these declarations are only used for

clang/include/clang/Sema/Sema.h

@@ -16,6 +16,7 @@
 
 #include "clang/APINotes/APINotesManager.h"
 #include "clang/AST/ASTFwd.h"
+#include "clang/AST/ASTLambda.h"
 #include "clang/AST/Attr.h"
 #include "clang/AST/AttrIterator.h"
 #include "clang/AST/CharUnits.h"
@@ -13108,14 +13109,16 @@ class Sema final : public SemaBase {
               ? ExpressionEvaluationContext::ImmediateFunctionContext
               : ExpressionEvaluationContext::PotentiallyEvaluated);
       if (FD) {
+        auto &Current = S.currentEvaluationContext();
+        const auto &Parent = S.parentEvaluationContext();
+
         FD->setWillHaveBody(true);
-        S.ExprEvalContexts.back().InImmediateFunctionContext =
+        Current.InImmediateFunctionContext =
             FD->isImmediateFunction() ||
-            S.ExprEvalContexts[S.ExprEvalContexts.size() - 2]
-                .isConstantEvaluated() ||
-            S.ExprEvalContexts[S.ExprEvalContexts.size() - 2]
-                .isImmediateFunctionContext();
-        S.ExprEvalContexts.back().InImmediateEscalatingFunctionContext =
+            (isLambdaMethod(FD) && (Parent.isConstantEvaluated() ||
+                                    Parent.isImmediateFunctionContext()));
+
+        Current.InImmediateEscalatingFunctionContext =
             S.getLangOpts().CPlusPlus20 && FD->isImmediateEscalating();
       } else
         assert(isa<ObjCMethodDecl>(DC));

clang/lib/AST/ASTImporter.cpp

@@ -15,6 +15,7 @@
 #include "clang/AST/ASTContext.h"
 #include "clang/AST/ASTDiagnostic.h"
 #include "clang/AST/ASTImporterSharedState.h"
+#include "clang/AST/ASTLambda.h"
 #include "clang/AST/ASTStructuralEquivalence.h"
 #include "clang/AST/Attr.h"
 #include "clang/AST/Decl.h"
@@ -3729,10 +3730,7 @@ bool ASTNodeImporter::hasReturnTypeDeclaredInside(FunctionDecl *D) {
     if (Importer.FromContext.getLangOpts().CPlusPlus14) // C++14 or later
       return false;
 
-    if (const auto *MD = dyn_cast<CXXMethodDecl>(D))
-      return cast<CXXRecordDecl>(MD->getDeclContext())->isLambda();
-
-    return false;
+    return isLambdaMethod(D);
   };
 
   QualType RetT = FromFPT->getReturnType();
@@ -3999,14 +3997,16 @@ ExpectedDecl ASTNodeImporter::VisitFunctionDecl(FunctionDecl *D) {
         importExplicitSpecifier(Err, Guide->getExplicitSpecifier());
     CXXConstructorDecl *Ctor =
         importChecked(Err, Guide->getCorrespondingConstructor());
+    const CXXDeductionGuideDecl *SourceDG =
+        importChecked(Err, Guide->getSourceDeductionGuide());
     if (Err)
       return std::move(Err);
     if (GetImportedOrCreateDecl<CXXDeductionGuideDecl>(
             ToFunction, D, Importer.getToContext(), DC, ToInnerLocStart, ESpec,
-            NameInfo, T, TInfo, ToEndLoc, Ctor))
+            NameInfo, T, TInfo, ToEndLoc, Ctor,
+            Guide->getDeductionCandidateKind(), TrailingRequiresClause,
+            SourceDG, Guide->getSourceDeductionGuideKind()))
       return ToFunction;
-    cast<CXXDeductionGuideDecl>(ToFunction)
-        ->setDeductionCandidateKind(Guide->getDeductionCandidateKind());
   } else {
     if (GetImportedOrCreateDecl(
             ToFunction, D, Importer.getToContext(), DC, ToInnerLocStart,