Merge pull request #515 from AMD-Lightning-Internal/amd/merge/upstream_merge_20250205180955

merge main into amd-staging

Author: ronlieb

clang/docs/ReleaseNotes.rst

@@ -112,6 +112,8 @@ Removed Compiler Flags
 Attribute Changes in Clang
 --------------------------
 
+- The ``no_sanitize`` attribute now accepts both ``gnu`` and ``clang`` names.
+
 Improvements to Clang's diagnostics
 -----------------------------------
 

clang/include/clang/AST/OpenMPClause.h

@@ -9451,7 +9451,8 @@ struct TargetOMPContext final : public llvm::omp::OMPContext {
   TargetOMPContext(ASTContext &ASTCtx,
                    std::function<void(StringRef)> &&DiagUnknownTrait,
                    const FunctionDecl *CurrentFunctionDecl,
-                   ArrayRef<llvm::omp::TraitProperty> ConstructTraits);
+                   ArrayRef<llvm::omp::TraitProperty> ConstructTraits,
+                   int DeviceNum);
 
   virtual ~TargetOMPContext() = default;
 

clang/include/clang/Basic/Attr.td

@@ -380,6 +380,13 @@ class Clang<string name, bit allowInC = 1, int version = 1>
   bit AllowInC = allowInC;
 }
 
+// This spelling combines the spellings of GCC and Clang for cases where the
+// spellings are equivalent for compile compatibility.
+class ClangGCC<string name, bit allowInC = 1, int version = 1>
+    : Spelling<name, "ClangGCC", version> {
+  bit AllowInC = allowInC;
+}
+
 // HLSL Annotation spellings
 class HLSLAnnotation<string name> : Spelling<name, "HLSLAnnotation">;
 
@@ -3677,7 +3684,7 @@ def X86ForceAlignArgPointer : InheritableAttr, TargetSpecificAttr<TargetAnyX86>
 }
 
 def NoSanitize : InheritableAttr {
-  let Spellings = [Clang<"no_sanitize">];
+  let Spellings = [ClangGCC<"no_sanitize">];
   let Args = [VariadicStringArgument<"Sanitizers">];
   let Subjects = SubjectList<[Function, ObjCMethod, GlobalVar], ErrorDiag>;
   let Documentation = [NoSanitizeDocs];

clang/include/clang/Driver/Options.td

@@ -934,7 +934,9 @@ def W_Joined : Joined<["-"], "W">, Group<W_Group>,
 def Xanalyzer : Separate<["-"], "Xanalyzer">,
   HelpText<"Pass <arg> to the static analyzer">, MetaVarName<"<arg>">,
   Group<StaticAnalyzer_Group>;
-def Xarch__ : JoinedAndSeparate<["-"], "Xarch_">, Flags<[NoXarchOption]>;
+def Xarch__ : JoinedAndSeparate<["-"], "Xarch_">, Flags<[NoXarchOption]>,
+  HelpText<"Pass <arg> to the compiliation if the target matches <arch>">,
+  MetaVarName<"<arch> <arg>">;
 def Xarch_host : Separate<["-"], "Xarch_host">, Flags<[NoXarchOption]>,
   HelpText<"Pass <arg> to the CUDA/HIP host compilation">, MetaVarName<"<arg>">;
 def Xarch_device : Separate<["-"], "Xarch_device">, Flags<[NoXarchOption]>,
@@ -1120,8 +1122,8 @@ def fno_convergent_functions : Flag<["-"], "fno-convergent-functions">,
 
 // Common offloading options
 let Group = offload_Group in {
-def offload_arch_EQ : Joined<["--"], "offload-arch=">, Flags<[NoXarchOption]>,
-  Visibility<[ClangOption, FlangOption]>,
+def offload_arch_EQ : Joined<["--"], "offload-arch=">,
+  Visibility<[ClangOption, FlangOption]>, Flags<[NoXarchOption]>,
   HelpText<"Specify an offloading device architecture for CUDA, HIP, or OpenMP. (e.g. sm_35). "
            "If 'native' is used the compiler will detect locally installed architectures. "
            "For HIP offloading, the device architecture can be followed by target ID features "

clang/include/clang/Sema/SemaOpenMP.h

@@ -849,6 +849,9 @@ class SemaOpenMP : public SemaBase {
       ArrayRef<OMPInteropInfo> AppendArgs, SourceLocation AdjustArgsLoc,
       SourceLocation AppendArgsLoc, SourceRange SR);
 
+  /// Called on device_num selector in context selectors.
+  void ActOnOpenMPDeviceNum(Expr *DeviceNumExpr);
+
   OMPClause *ActOnOpenMPSingleExprClause(OpenMPClauseKind Kind, Expr *Expr,
                                          SourceLocation StartLoc,
                                          SourceLocation LParenLoc,
@@ -1410,6 +1413,13 @@ class SemaOpenMP : public SemaBase {
 
   void handleOMPAssumeAttr(Decl *D, const ParsedAttr &AL);
 
+  /// Setter and getter functions for device_num.
+  void setOpenMPDeviceNum(int Num);
+
+  int getOpenMPDeviceNum() const;
+
+  void setOpenMPDeviceNumID(StringRef ID);
+
 private:
   void *VarDataSharingAttributesStack;
 
@@ -1480,6 +1490,12 @@ class SemaOpenMP : public SemaBase {
 
   /// All `omp assumes` we encountered so far.
   SmallVector<OMPAssumeAttr *, 4> OMPAssumeGlobal;
+
+  /// Device number specified by the context selector.
+  int DeviceNum = -1;
+
+  /// Device number identifier specified by the context selector.
+  StringRef DeviceNumID;
 };
 
 } // namespace clang

clang/lib/AST/OpenMPClause.cpp

@@ -2927,9 +2927,13 @@ llvm::raw_ostream &clang::operator<<(llvm::raw_ostream &OS,
 TargetOMPContext::TargetOMPContext(
     ASTContext &ASTCtx, std::function<void(StringRef)> &&DiagUnknownTrait,
     const FunctionDecl *CurrentFunctionDecl,
-    ArrayRef<llvm::omp::TraitProperty> ConstructTraits)
+    ArrayRef<llvm::omp::TraitProperty> ConstructTraits, int DeviceNum)
     : OMPContext(ASTCtx.getLangOpts().OpenMPIsTargetDevice,
-                 ASTCtx.getTargetInfo().getTriple()),
+                 ASTCtx.getTargetInfo().getTriple(),
+                 ASTCtx.getLangOpts().OMPTargetTriples.empty()
+                     ? llvm::Triple()
+                     : ASTCtx.getLangOpts().OMPTargetTriples[0],
+                 DeviceNum),
       FeatureValidityCheck([&](StringRef FeatureName) {
         return ASTCtx.getTargetInfo().isValidFeatureName(FeatureName);
       }),

clang/lib/Driver/Driver.cpp

@@ -3418,7 +3418,9 @@ class OffloadingActionBuilder final {
       // Collect all offload arch parameters, removing duplicates.
       std::set<StringRef> GpuArchs;
       bool Error = false;
-      for (Arg *A : Args) {
+      const ToolChain &TC = *ToolChains.front();
+      for (Arg *A : C.getArgsForToolChain(&TC, /*BoundArch=*/"",
+                                          AssociatedOffloadKind)) {
         if (!(A->getOption().matches(options::OPT_offload_arch_EQ) ||
               A->getOption().matches(options::OPT_no_offload_arch_EQ)))
           continue;
@@ -3429,7 +3431,6 @@ class OffloadingActionBuilder final {
               ArchStr == "all") {
             GpuArchs.clear();
           } else if (ArchStr == "native") {
-            const ToolChain &TC = *ToolChains.front();
             auto GPUsOrErr = ToolChains.front()->getSystemGPUArchs(Args);
             if (!GPUsOrErr) {
               TC.getDriver().Diag(diag::err_drv_undetermined_gpu_arch)

clang/lib/Driver/ToolChain.cpp

@@ -1659,7 +1659,8 @@ void ToolChain::TranslateXarchArgs(
       A->getOption().matches(options::OPT_Xarch_host))
     ValuePos = 0;
 
-  unsigned Index = Args.getBaseArgs().MakeIndex(A->getValue(ValuePos));
+  const InputArgList &BaseArgs = Args.getBaseArgs();
+  unsigned Index = BaseArgs.MakeIndex(A->getValue(ValuePos));
   unsigned Prev = Index;
   std::unique_ptr<llvm::opt::Arg> XarchArg(Opts.ParseOneArg(Args, Index));
 
@@ -1683,8 +1684,31 @@ void ToolChain::TranslateXarchArgs(
     Diags.Report(DiagID) << A->getAsString(Args);
     return;
   }
+
   XarchArg->setBaseArg(A);
   A = XarchArg.release();
+
+  // Linker input arguments require custom handling. The problem is that we
+  // have already constructed the phase actions, so we can not treat them as
+  // "input arguments".
+  if (A->getOption().hasFlag(options::LinkerInput)) {
+    // Convert the argument into individual Zlinker_input_args. Need to do this
+    // manually to avoid memory leaks with the allocated arguments.
+    for (const char *Value : A->getValues()) {
+      auto Opt = Opts.getOption(options::OPT_Zlinker_input);
+      unsigned Index = BaseArgs.MakeIndex(Opt.getName(), Value);
+      auto NewArg =
+          new Arg(Opt, BaseArgs.MakeArgString(Opt.getPrefix() + Opt.getName()),
+                  Index, BaseArgs.getArgString(Index + 1), A);
+
+      DAL->append(NewArg);
+      if (!AllocatedArgs)
+        DAL->AddSynthesizedArg(NewArg);
+      else
+        AllocatedArgs->push_back(NewArg);
+    }
+  }
+
   if (!AllocatedArgs)
     DAL->AddSynthesizedArg(A);
   else
@@ -1708,19 +1732,17 @@ llvm::opt::DerivedArgList *ToolChain::TranslateXarchArgs(
     } else if (A->getOption().matches(options::OPT_Xarch_host)) {
       NeedTrans = !IsDevice;
       Skip = IsDevice;
-    } else if (A->getOption().matches(options::OPT_Xarch__) && IsDevice) {
-      // Do not translate -Xarch_ options for non CUDA/HIP toolchain since
-      // they may need special translation.
-      // Skip this argument unless the architecture matches BoundArch
-      if (BoundArch.empty() || A->getValue(0) != BoundArch)
-        Skip = true;
-      else
-        NeedTrans = true;
+    } else if (A->getOption().matches(options::OPT_Xarch__)) {
+      NeedTrans = A->getValue() == getArchName() ||
+                  (!BoundArch.empty() && A->getValue() == BoundArch);
+      Skip = !NeedTrans;
     }
     if (NeedTrans || Skip)
       Modified = true;
-    if (NeedTrans)
+    if (NeedTrans) {
+      A->claim();
       TranslateXarchArgs(Args, A, DAL, AllocatedArgs);
+    }
     if (!Skip)
       DAL->append(A);
   }

clang/lib/Driver/ToolChains/Darwin.cpp

@@ -2777,30 +2777,6 @@ DerivedArgList *MachO::TranslateArgs(const DerivedArgList &Args,
   // and try to push it down into tool specific logic.
 
   for (Arg *A : Args) {
-    if (A->getOption().matches(options::OPT_Xarch__)) {
-      // Skip this argument unless the architecture matches either the toolchain
-      // triple arch, or the arch being bound.
-      StringRef XarchArch = A->getValue(0);
-      if (!(XarchArch == getArchName() ||
-            (!BoundArch.empty() && XarchArch == BoundArch)))
-        continue;
-
-      Arg *OriginalArg = A;
-      TranslateXarchArgs(Args, A, DAL);
-
-      // Linker input arguments require custom handling. The problem is that we
-      // have already constructed the phase actions, so we can not treat them as
-      // "input arguments".
-      if (A->getOption().hasFlag(options::LinkerInput)) {
-        // Convert the argument into individual Zlinker_input_args.
-        for (const char *Value : A->getValues()) {
-          DAL->AddSeparateArg(
-              OriginalArg, Opts.getOption(options::OPT_Zlinker_input), Value);
-        }
-        continue;
-      }
-    }
-
     // Sob. These is strictly gcc compatible for the time being. Apple
     // gcc translates options twice, which means that self-expanding
     // options add duplicates.

clang/lib/Parse/ParseOpenMP.cpp

@@ -888,7 +888,18 @@ void Parser::parseOMPTraitPropertyKind(OMPTraitProperty &TIProperty,
   TIProperty.Kind = TraitProperty::invalid;
 
   SourceLocation NameLoc = Tok.getLocation();
-  StringRef Name = getNameFromIdOrString(*this, Tok, CONTEXT_TRAIT_LVL);
+  StringRef Name;
+  if (Selector == llvm::omp::TraitSelector::target_device_device_num) {
+    Name = "number";
+    TIProperty.Kind = getOpenMPContextTraitPropertyKind(Set, Selector, Name);
+    ExprResult DeviceNumExprResult = ParseExpression();
+    if (DeviceNumExprResult.isUsable()) {
+      Expr *DeviceNumExpr = DeviceNumExprResult.get();
+      Actions.OpenMP().ActOnOpenMPDeviceNum(DeviceNumExpr);
+    }
+    return;
+  }
+  Name = getNameFromIdOrString(*this, Tok, CONTEXT_TRAIT_LVL);
   if (Name.empty()) {
     Diag(Tok.getLocation(), diag::note_omp_declare_variant_ctx_options)
         << CONTEXT_TRAIT_LVL << listOpenMPContextTraitProperties(Set, Selector);
@@ -918,7 +929,8 @@ void Parser::parseOMPTraitPropertyKind(OMPTraitProperty &TIProperty,
         << "(<property-name>)";
     return;
   }
-  TraitSelector SelectorForName = getOpenMPContextTraitSelectorKind(Name);
+  TraitSelector SelectorForName =
+      getOpenMPContextTraitSelectorKind(Name, SetForName);
   if (SelectorForName != TraitSelector::invalid) {
     Diag(NameLoc, diag::note_omp_declare_variant_ctx_is_a)
         << Name << CONTEXT_SELECTOR_LVL << CONTEXT_TRAIT_LVL;
@@ -935,7 +947,7 @@ void Parser::parseOMPTraitPropertyKind(OMPTraitProperty &TIProperty,
   }
   for (const auto &PotentialSet :
        {TraitSet::construct, TraitSet::user, TraitSet::implementation,
-        TraitSet::device}) {
+        TraitSet::device, TraitSet::target_device}) {
     TraitProperty PropertyForName =
         getOpenMPContextTraitPropertyKind(PotentialSet, Selector, Name);
     if (PropertyForName == TraitProperty::invalid)
@@ -1062,7 +1074,7 @@ void Parser::parseOMPTraitSelectorKind(OMPTraitSelector &TISelector,
     return;
   }
 
-  TISelector.Kind = getOpenMPContextTraitSelectorKind(Name);
+  TISelector.Kind = getOpenMPContextTraitSelectorKind(Name, Set);
   if (TISelector.Kind != TraitSelector::invalid) {
     if (checkForDuplicates(*this, Name, NameLoc, Seen, CONTEXT_SELECTOR_LVL))
       TISelector.Kind = TraitSelector::invalid;
@@ -1084,7 +1096,7 @@ void Parser::parseOMPTraitSelectorKind(OMPTraitSelector &TISelector,
   }
   for (const auto &PotentialSet :
        {TraitSet::construct, TraitSet::user, TraitSet::implementation,
-        TraitSet::device}) {
+        TraitSet::device, TraitSet::target_device}) {
     TraitProperty PropertyForName = getOpenMPContextTraitPropertyKind(
         PotentialSet, TraitSelector::invalid, Name);
     if (PropertyForName == TraitProperty::invalid)
@@ -1259,7 +1271,8 @@ void Parser::parseOMPTraitSetKind(OMPTraitSet &TISet,
   // It follows diagnosis and helping notes.
   Diag(NameLoc, diag::warn_omp_declare_variant_ctx_not_a_set) << Name;
 
-  TraitSelector SelectorForName = getOpenMPContextTraitSelectorKind(Name);
+  TraitSelector SelectorForName =
+      getOpenMPContextTraitSelectorKind(Name, TISet.Kind);
   if (SelectorForName != TraitSelector::invalid) {
     Diag(NameLoc, diag::note_omp_declare_variant_ctx_is_a)
         << Name << CONTEXT_SELECTOR_LVL << CONTEXT_SELECTOR_SET_LVL;
@@ -1276,7 +1289,7 @@ void Parser::parseOMPTraitSetKind(OMPTraitSet &TISet,
   }
   for (const auto &PotentialSet :
        {TraitSet::construct, TraitSet::user, TraitSet::implementation,
-        TraitSet::device}) {
+        TraitSet::device, TraitSet::target_device}) {
     TraitProperty PropertyForName = getOpenMPContextTraitPropertyKind(
         PotentialSet, TraitSelector::invalid, Name);
     if (PropertyForName == TraitProperty::invalid)
@@ -2253,7 +2266,8 @@ Parser::DeclGroupPtrTy Parser::ParseOpenMPDeclarativeDirectiveWithExtDecl(
     TargetOMPContext OMPCtx(
         ASTCtx, std::move(DiagUnknownTrait),
         /* CurrentFunctionDecl */ nullptr,
-        /* ConstructTraits */ ArrayRef<llvm::omp::TraitProperty>());
+        /* ConstructTraits */ ArrayRef<llvm::omp::TraitProperty>(),
+        Actions.OpenMP().getOpenMPDeviceNum());
 
     if (isVariantApplicableInContext(VMI, OMPCtx, /* DeviceSetOnly */ true)) {
       Actions.OpenMP().ActOnOpenMPBeginDeclareVariant(Loc, TI);
@@ -2805,7 +2819,8 @@ StmtResult Parser::ParseOpenMPDeclarativeOrExecutableDirective(
         };
     TargetOMPContext OMPCtx(ASTContext, std::move(DiagUnknownTrait),
                             /* CurrentFunctionDecl */ nullptr,
-                            ArrayRef<llvm::omp::TraitProperty>());
+                            ArrayRef<llvm::omp::TraitProperty>(),
+                            Actions.OpenMP().getOpenMPDeviceNum());
 
     // A single match is returned for OpenMP 5.0
     int BestIdx = getBestVariantMatchForContext(VMIs, OMPCtx);