merge main into amd-staging (#459)

Author: z1-cciauto

clang/include/clang/Basic/LangOptions.def

@@ -217,6 +217,7 @@ LANGOPT(OpenCLPipes              , 1, 0, NotCompatible, "OpenCL pipes language c
 LANGOPT(NativeHalfType    , 1, 0, NotCompatible, "Native half type support")
 LANGOPT(NativeHalfArgsAndReturns, 1, 0, NotCompatible, "Native half args and returns")
 LANGOPT(HalfArgsAndReturns, 1, 0, NotCompatible, "half args and returns")
+LANGOPT(NativeInt16Type   , 1, 1, NotCompatible, "Native int 16 type support")
 LANGOPT(CUDA              , 1, 0, NotCompatible, "CUDA")
 LANGOPT(HIP               , 1, 0, NotCompatible, "HIP")
 LANGOPT(OpenMP            , 32, 0, NotCompatible, "OpenMP support and version of OpenMP (31, 40 or 45)")

clang/include/clang/Driver/Options.td

@@ -8804,6 +8804,11 @@ def fobjc_subscripting_legacy_runtime : Flag<["-"], "fobjc-subscripting-legacy-r
 def vtordisp_mode_EQ : Joined<["-"], "vtordisp-mode=">,
   HelpText<"Control vtordisp placement on win32 targets">,
   MarshallingInfoInt<LangOpts<"VtorDispMode">, "1">;
+def fnative_int16_type : Flag<["-"], "fnative-int16-type">,
+  HelpText<"Use 16 bit integer types">,
+  // This option is implied unless we are in HLSL lang mode
+  ImpliedByAnyOf<[!strconcat("!", hlsl.KeyPath)]>,
+  MarshallingInfoFlag<LangOpts<"NativeInt16Type">>;
 def fnative_half_type: Flag<["-"], "fnative-half-type">,
   HelpText<"Use the native half type for __fp16 instead of promoting to float">,
   MarshallingInfoFlag<LangOpts<"NativeHalfType">>,
@@ -9811,7 +9816,7 @@ def emit_pristine_llvm : DXCFlag<"emit-pristine-llvm">,
   HelpText<"Emit pristine LLVM IR from the frontend by not running any LLVM passes at all."
            "Same as -S + -emit-llvm + -disable-llvm-passes.">;
 def fcgl : DXCFlag<"fcgl">, Alias<emit_pristine_llvm>;
-def enable_16bit_types : DXCFlag<"enable-16bit-types">, Alias<fnative_half_type>,
+def enable_16bit_types : DXCFlag<"enable-16bit-types">,
   HelpText<"Enable 16-bit types and disable min precision types."
            "Available in HLSL 2018 and shader model 6.2.">;
 def fdx_rootsignature_version :

clang/lib/CIR/CodeGen/CIRGenBuiltinX86.cpp

@@ -771,14 +771,6 @@ mlir::Value CIRGenFunction::emitX86BuiltinExpr(unsigned builtinID,
   case X86::BI_WriteBarrier:
   case X86::BI_AddressOfReturnAddress:
   case X86::BI__stosb:
-  case X86::BI__builtin_ia32_t2rpntlvwz0_internal:
-  case X86::BI__builtin_ia32_t2rpntlvwz0rs_internal:
-  case X86::BI__builtin_ia32_t2rpntlvwz0t1_internal:
-  case X86::BI__builtin_ia32_t2rpntlvwz0rst1_internal:
-  case X86::BI__builtin_ia32_t2rpntlvwz1_internal:
-  case X86::BI__builtin_ia32_t2rpntlvwz1rs_internal:
-  case X86::BI__builtin_ia32_t2rpntlvwz1t1_internal:
-  case X86::BI__builtin_ia32_t2rpntlvwz1rst1_internal:
   case X86::BI__ud2:
   case X86::BI__int2c:
   case X86::BI__readfsbyte:

clang/lib/CodeGen/CGOpenMPRuntimeGPU.cpp

@@ -2635,7 +2635,7 @@ llvm::Value *CGOpenMPRuntimeGPU::getGPUBlockID(CodeGenFunction &CGF) {
   CGBuilderTy &Bld = CGF.Builder;
   llvm::Function *F =
       CGF.CGM.getIntrinsic(llvm::Intrinsic::amdgcn_workgroup_id_x);
-  return Bld.CreateCall(F, std::nullopt, "gpu_block_id");
+  return Bld.CreateCall(F, {}, "gpu_block_id");
 }
 
 llvm::Value *CGOpenMPRuntimeGPU::getGPUNumBlocks(CodeGenFunction &CGF) {

clang/lib/Driver/ToolChains/Clang.cpp

@@ -3752,6 +3752,7 @@ static void RenderHLSLOptions(const ArgList &Args, ArgStringList &CmdArgs,
       options::OPT_emit_obj,
       options::OPT_disable_llvm_passes,
       options::OPT_fnative_half_type,
+      options::OPT_fnative_int16_type,
       options::OPT_hlsl_entrypoint,
       options::OPT_fdx_rootsignature_define,
       options::OPT_fdx_rootsignature_version,

clang/lib/Driver/ToolChains/HLSL.cpp

@@ -498,6 +498,15 @@ HLSLToolChain::TranslateArgs(const DerivedArgList &Args, StringRef BoundArch,
       continue;
     }
 
+    if (A->getOption().getID() == options::OPT_enable_16bit_types) {
+      // Translate -enable-16bit-types into -fnative-half-type and
+      // -fnative-int16-type
+      DAL->AddFlagArg(nullptr, Opts.getOption(options::OPT_fnative_half_type));
+      DAL->AddFlagArg(nullptr, Opts.getOption(options::OPT_fnative_int16_type));
+      A->claim();
+      continue;
+    }
+
     DAL->append(A);
   }
 

clang/lib/Frontend/CompilerInvocation.cpp

@@ -4729,7 +4729,8 @@ bool CompilerInvocation::ParseLangArgs(LangOptions &Opts, ArgList &Args,
         // Validate that if fnative-half-type is given, that
         // the language standard is at least hlsl2018, and that
         // the target shader model is at least 6.2.
-        if (Args.getLastArg(OPT_fnative_half_type)) {
+        if (Args.getLastArg(OPT_fnative_half_type) ||
+            Args.getLastArg(OPT_fnative_int16_type)) {
           const LangStandard &Std =
               LangStandard::getLangStandardForKind(Opts.LangStd);
           if (!(Opts.LangStd >= LangStandard::lang_hlsl2018 &&
@@ -4743,12 +4744,16 @@ bool CompilerInvocation::ParseLangArgs(LangOptions &Opts, ArgList &Args,
           Diags.Report(diag::err_drv_hlsl_bad_shader_unsupported)
               << VulkanEnv << T.getOSName() << T.str();
         }
-        if (Args.getLastArg(OPT_fnative_half_type)) {
+        if (Args.getLastArg(OPT_fnative_half_type) ||
+            Args.getLastArg(OPT_fnative_int16_type)) {
+          const char *Str = Args.getLastArg(OPT_fnative_half_type)
+                                ? "-fnative-half-type"
+                                : "-fnative-int16-type";
           const LangStandard &Std =
               LangStandard::getLangStandardForKind(Opts.LangStd);
           if (!(Opts.LangStd >= LangStandard::lang_hlsl2018))
             Diags.Report(diag::err_drv_hlsl_16bit_types_unsupported)
-                << "-fnative-half-type" << false << Std.getName();
+                << Str << false << Std.getName();
         }
       } else {
         llvm_unreachable("expected DXIL or SPIR-V target");

clang/lib/Frontend/InitPreprocessor.cpp

@@ -399,7 +399,7 @@ static void InitializeStandardPredefinedMacros(const TargetInfo &TI,
     Builder.defineMacro("__HLSL_202y",
                         Twine((unsigned)LangOptions::HLSLLangStd::HLSL_202y));
 
-    if (LangOpts.NativeHalfType)
+    if (LangOpts.NativeHalfType && LangOpts.NativeInt16Type)
       Builder.defineMacro("__HLSL_ENABLE_16_BIT", "1");
 
     // Shader target information

clang/lib/Parse/ParseDecl.cpp

@@ -4248,6 +4248,13 @@ void Parser::ParseDeclarationSpecifiers(
 
     // type-specifier
     case tok::kw_short:
+      if (!getLangOpts().NativeInt16Type) {
+        Diag(Tok, diag::err_unknown_typename) << Tok.getName();
+        DS.SetTypeSpecError();
+        DS.SetRangeEnd(Tok.getLocation());
+        ConsumeToken();
+        goto DoneWithDeclSpec;
+      }
       isInvalid = DS.SetTypeSpecWidth(TypeSpecifierWidth::Short, Loc, PrevSpec,
                                       DiagID, Policy);
       break;

clang/lib/Sema/SemaCodeComplete.cpp

@@ -10208,6 +10208,24 @@ void SemaCodeCompletion::CodeCompletePreprocessorDirective(bool InConditional) {
   Builder.AddPlaceholderChunk("message");
   Results.AddResult(Builder.TakeString());
 
+  if (getLangOpts().C23) {
+    // #embed "file"
+    Builder.AddTypedTextChunk("embed");
+    Builder.AddChunk(CodeCompletionString::CK_HorizontalSpace);
+    Builder.AddTextChunk("\"");
+    Builder.AddPlaceholderChunk("file");
+    Builder.AddTextChunk("\"");
+    Results.AddResult(Builder.TakeString());
+
+    // #embed <file>
+    Builder.AddTypedTextChunk("embed");
+    Builder.AddChunk(CodeCompletionString::CK_HorizontalSpace);
+    Builder.AddTextChunk("<");
+    Builder.AddPlaceholderChunk("file");
+    Builder.AddTextChunk(">");
+    Results.AddResult(Builder.TakeString());
+  }
+
   // Note: #ident and #sccs are such crazy anachronisms that we don't provide
   // completions for them. And __include_macros is a Clang-internal extension
   // that we don't want to encourage anyone to use.