merge main into amd-staging (#615)

Author: z1-cciauto

clang/include/clang/Basic/BuiltinsX86.td

@@ -4117,15 +4117,15 @@ let Features = "avx512f", Attributes = [NoThrow, Const, Constexpr, RequiredVecto
   def selectpd_512 : X86Builtin<"_Vector<8, double>(unsigned char, _Vector<8, double>, _Vector<8, double>)">;
 }
 
-let Features = "avx512fp16", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in {
+let Features = "avx512fp16", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
   def selectsh_128 : X86Builtin<"_Vector<8, _Float16>(unsigned char, _Vector<8, _Float16>, _Vector<8, _Float16>)">;
 }
 
-let Features = "avx512bf16", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in {
+let Features = "avx512bf16", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
   def selectsbf_128 : X86Builtin<"_Vector<8, __bf16>(unsigned char, _Vector<8, __bf16>, _Vector<8, __bf16>)">;
 }
 
-let Features = "avx512f", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in {
+let Features = "avx512f", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in {
   def selectss_128 : X86Builtin<"_Vector<4, float>(unsigned char, _Vector<4, float>, _Vector<4, float>)">;
   def selectsd_128 : X86Builtin<"_Vector<2, double>(unsigned char, _Vector<2, double>, _Vector<2, double>)">;
 }

clang/include/clang/Basic/DiagnosticDriverKinds.td

@@ -129,13 +129,23 @@ def err_drv_bad_offload_arch_combo : Error<
   "invalid offload arch combinations: '%0' and '%1' (for a specific processor, "
   "a feature should either exist in all offload archs, or not exist in any "
   "offload archs)">;
+def err_drv_unsupported_option_for_offload_arch_req_feature : Error<
+  "'%0' option for offload arch '%1' is not currently supported "
+  "there. Use it with an offload arch containing '%2' instead">;
 def warn_drv_unsupported_option_for_offload_arch_req_feature : Warning<
   "ignoring '%0' option for offload arch '%1' as it is not currently supported "
   "there. Use it with an offload arch containing '%2' instead">,
   InGroup<OptionIgnored>;
 def warn_drv_unsupported_option_for_target : Warning<
   "ignoring '%0' option as it is not currently supported for target '%1'">,
   InGroup<OptionIgnored>;
+def err_drv_unsupported_option_for_target : Error<
+  "'%0' option is not currently supported for target '%1'">;
+def warn_drv_unsupported_option_part_for_target : Warning<
+  "ignoring '%0' in '%1' option as it is not currently supported for target '%2'">,
+  InGroup<OptionIgnored>;
+def err_drv_unsupported_option_part_for_target : Error<
+  "'%0' in '%1' option is not currently supported for target '%2'">;
 def warn_drv_invalid_argument_for_flang : Warning<
   "'%0' is not valid for Fortran">,
   InGroup<OptionIgnored>;

clang/include/clang/CIR/Dialect/Builder/CIRBaseBuilder.h

@@ -131,6 +131,14 @@ class CIRBaseBuilderTy : public mlir::OpBuilder {
     return cir::IntType::get(getContext(), n, false);
   }
 
+  static unsigned getCIRIntOrFloatBitWidth(mlir::Type eltTy) {
+    if (auto intType = mlir::dyn_cast<cir::IntTypeInterface>(eltTy))
+      return intType.getWidth();
+    if (auto floatType = mlir::dyn_cast<cir::FPTypeInterface>(eltTy))
+      return floatType.getWidth();
+
+    llvm_unreachable("Unsupported type in getCIRIntOrFloatBitWidth");
+  }
   cir::IntType getSIntNTy(int n) {
     return cir::IntType::get(getContext(), n, true);
   }
@@ -565,6 +573,16 @@ class CIRBaseBuilderTy : public mlir::OpBuilder {
     return cir::CmpOp::create(*this, loc, getBoolTy(), kind, lhs, rhs);
   }
 
+  cir::VecCmpOp createVecCompare(mlir::Location loc, cir::CmpOpKind kind,
+                                 mlir::Value lhs, mlir::Value rhs) {
+    VectorType vecCast = mlir::cast<VectorType>(lhs.getType());
+    IntType integralTy =
+        getSIntNTy(getCIRIntOrFloatBitWidth(vecCast.getElementType()));
+    VectorType integralVecTy =
+        VectorType::get(context, integralTy, vecCast.getSize());
+    return cir::VecCmpOp::create(*this, loc, integralVecTy, kind, lhs, rhs);
+  }
+
   mlir::Value createIsNaN(mlir::Location loc, mlir::Value operand) {
     return createCompare(loc, cir::CmpOpKind::ne, operand, operand);
   }

clang/include/clang/CIR/MissingFeatures.h

@@ -259,6 +259,7 @@ struct MissingFeatures {
   static bool emitBranchThroughCleanup() { return false; }
   static bool emitCheckedInBoundsGEP() { return false; }
   static bool emitCondLikelihoodViaExpectIntrinsic() { return false; }
+  static bool emitConstrainedFPCall() { return false; }
   static bool emitLifetimeMarkers() { return false; }
   static bool emitLValueAlignmentAssumption() { return false; }
   static bool emitNullCheckForDeleteCalls() { return false; }

clang/include/clang/Options/Options.td

@@ -201,6 +201,10 @@ def hlsl_Group : OptionGroup<"<HLSL group>">, Group<f_Group>,
                    DocName<"HLSL options">,
                    Visibility<[ClangOption]>;
 
+def fsan_cov_Group : OptionGroup<"<-fsanitize-coverage group>">,
+                   Group<f_clang_Group>,
+                   DocName<"Sanitizer Coverage options">;
+
 // Feature groups - these take command line options that correspond directly to
 // target specific features and can be translated directly from command line
 // options.
@@ -2413,26 +2417,26 @@ def : Flag<["-"], "fno-sanitize-blacklist">,
   Group<f_clang_Group>, Flags<[HelpHidden]>, Alias<fno_sanitize_ignorelist>;
 
 def fsanitize_coverage : CommaJoined<["-"], "fsanitize-coverage=">,
-  Group<f_clang_Group>,
+  Group<fsan_cov_Group>,
   HelpText<"Specify the type of coverage instrumentation for Sanitizers">;
 def fno_sanitize_coverage : CommaJoined<["-"], "fno-sanitize-coverage=">,
-  Group<f_clang_Group>, Visibility<[ClangOption, CLOption]>,
+  Group<fsan_cov_Group>, Visibility<[ClangOption, CLOption]>,
   HelpText<"Disable features of coverage instrumentation for Sanitizers">,
   Values<"func,bb,edge,indirect-calls,trace-bb,trace-cmp,trace-div,trace-gep,"
          "8bit-counters,trace-pc,trace-pc-guard,no-prune,inline-8bit-counters,"
          "inline-bool-flag">;
 def fsanitize_coverage_allowlist : Joined<["-"], "fsanitize-coverage-allowlist=">,
-    Group<f_clang_Group>, Visibility<[ClangOption, CLOption]>,
+    Group<fsan_cov_Group>, Visibility<[ClangOption, CLOption]>,
     HelpText<"Restrict sanitizer coverage instrumentation exclusively to modules and functions that match the provided special case list, except the blocked ones">,
     MarshallingInfoStringVector<CodeGenOpts<"SanitizeCoverageAllowlistFiles">>;
 def fsanitize_coverage_ignorelist : Joined<["-"], "fsanitize-coverage-ignorelist=">,
-    Group<f_clang_Group>, Visibility<[ClangOption, CLOption]>,
+    Group<fsan_cov_Group>, Visibility<[ClangOption, CLOption]>,
     HelpText<"Disable sanitizer coverage instrumentation for modules and functions "
              "that match the provided special case list, even the allowed ones">,
     MarshallingInfoStringVector<CodeGenOpts<"SanitizeCoverageIgnorelistFiles">>;
 def fsanitize_coverage_stack_depth_callback_min_EQ
     : Joined<["-"], "fsanitize-coverage-stack-depth-callback-min=">,
-      Group<f_clang_Group>,
+      Group<fsan_cov_Group>,
       MetaVarName<"<M>">,
       HelpText<"Use callback for max stack depth tracing with minimum stack "
                "depth M">,
@@ -8068,70 +8072,87 @@ def linker_option : Joined<["--"], "linker-option=">,
   HelpText<"Add linker option">,
   MarshallingInfoStringVector<CodeGenOpts<"LinkerOptions">>;
 def fsanitize_coverage_type : Joined<["-"], "fsanitize-coverage-type=">,
+                              Group<fsan_cov_Group>,
                               HelpText<"Sanitizer coverage type">,
                               MarshallingInfoInt<CodeGenOpts<"SanitizeCoverageType">>;
 def fsanitize_coverage_indirect_calls
     : Flag<["-"], "fsanitize-coverage-indirect-calls">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable sanitizer coverage for indirect calls">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageIndirectCalls">>;
 def fsanitize_coverage_trace_bb
     : Flag<["-"], "fsanitize-coverage-trace-bb">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable basic block tracing in sanitizer coverage">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageTraceBB">>;
 def fsanitize_coverage_trace_cmp
     : Flag<["-"], "fsanitize-coverage-trace-cmp">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable cmp instruction tracing in sanitizer coverage">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageTraceCmp">>;
 def fsanitize_coverage_trace_div
     : Flag<["-"], "fsanitize-coverage-trace-div">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable div instruction tracing in sanitizer coverage">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageTraceDiv">>;
 def fsanitize_coverage_trace_gep
     : Flag<["-"], "fsanitize-coverage-trace-gep">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable gep instruction tracing in sanitizer coverage">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageTraceGep">>;
 def fsanitize_coverage_8bit_counters
     : Flag<["-"], "fsanitize-coverage-8bit-counters">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable frequency counters in sanitizer coverage">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverage8bitCounters">>;
 def fsanitize_coverage_inline_8bit_counters
     : Flag<["-"], "fsanitize-coverage-inline-8bit-counters">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable inline 8-bit counters in sanitizer coverage">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageInline8bitCounters">>;
 def fsanitize_coverage_inline_bool_flag
     : Flag<["-"], "fsanitize-coverage-inline-bool-flag">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable inline bool flag in sanitizer coverage">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageInlineBoolFlag">>;
 def fsanitize_coverage_pc_table
     : Flag<["-"], "fsanitize-coverage-pc-table">,
+      Group<fsan_cov_Group>,
       HelpText<"Create a table of coverage-instrumented PCs">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoveragePCTable">>;
 def fsanitize_coverage_control_flow
     : Flag<["-"], "fsanitize-coverage-control-flow">,
+      Group<fsan_cov_Group>,
       HelpText<"Collect control flow of function">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageControlFlow">>;
 def fsanitize_coverage_trace_pc
     : Flag<["-"], "fsanitize-coverage-trace-pc">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable PC tracing in sanitizer coverage">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageTracePC">>;
 def fsanitize_coverage_trace_pc_guard
     : Flag<["-"], "fsanitize-coverage-trace-pc-guard">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable PC tracing with guard in sanitizer coverage">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageTracePCGuard">>;
 def fsanitize_coverage_no_prune
     : Flag<["-"], "fsanitize-coverage-no-prune">,
+      Group<fsan_cov_Group>,
       HelpText<"Disable coverage pruning (i.e. instrument all blocks/edges)">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageNoPrune">>;
 def fsanitize_coverage_stack_depth
     : Flag<["-"], "fsanitize-coverage-stack-depth">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable max stack depth tracing">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageStackDepth">>;
 def fsanitize_coverage_trace_loads
     : Flag<["-"], "fsanitize-coverage-trace-loads">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable tracing of loads">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageTraceLoads">>;
 def fsanitize_coverage_trace_stores
     : Flag<["-"], "fsanitize-coverage-trace-stores">,
+      Group<fsan_cov_Group>,
       HelpText<"Enable tracing of stores">,
       MarshallingInfoFlag<CodeGenOpts<"SanitizeCoverageTraceStores">>;
 def fexperimental_sanitize_metadata_EQ_covered

clang/lib/AST/ByteCode/Compiler.cpp

@@ -6057,13 +6057,15 @@ bool Compiler<Emitter>::visitSwitchStmt(const SwitchStmt *S) {
                           DefaultLabel);
   if (!this->visitStmt(S->getBody()))
     return false;
+  this->fallthrough(EndLabel);
   this->emitLabel(EndLabel);
 
   return LS.destroyLocals();
 }
 
 template <class Emitter>
 bool Compiler<Emitter>::visitCaseStmt(const CaseStmt *S) {
+  this->fallthrough(CaseLabels[S]);
   this->emitLabel(CaseLabels[S]);
   return this->visitStmt(S->getSubStmt());
 }

clang/lib/AST/ByteCode/InterpBuiltin.cpp

@@ -2838,6 +2838,30 @@ static bool interp__builtin_select(InterpState &S, CodePtr OpPC,
   return true;
 }
 
+/// Scalar variant of AVX512 predicated select:
+/// Result[i] = (Mask bit 0) ? LHS[i] : RHS[i], but only element 0 may change.
+/// All other elements are taken from RHS.
+static bool interp__builtin_select_scalar(InterpState &S,
+                                          const CallExpr *Call) {
+  unsigned N =
+      Call->getArg(1)->getType()->getAs<VectorType>()->getNumElements();
+
+  const Pointer &W = S.Stk.pop<Pointer>();
+  const Pointer &A = S.Stk.pop<Pointer>();
+  APSInt U = popToAPSInt(S, Call->getArg(0));
+  const Pointer &Dst = S.Stk.peek<Pointer>();
+
+  bool TakeA0 = U.getZExtValue() & 1ULL;
+
+  for (unsigned I = TakeA0; I != N; ++I)
+    Dst.elem<Floating>(I) = W.elem<Floating>(I);
+  if (TakeA0)
+    Dst.elem<Floating>(0) = A.elem<Floating>(0);
+
+  Dst.initializeAllElements();
+  return true;
+}
+
 static bool interp__builtin_blend(InterpState &S, CodePtr OpPC,
                                   const CallExpr *Call) {
   APSInt Mask = popToAPSInt(S, Call->getArg(2));
@@ -4151,6 +4175,11 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call,
       return APInt::getAllOnes(DstBits);
     });
 
+  case clang::X86::BI__builtin_ia32_selectss_128:
+  case clang::X86::BI__builtin_ia32_selectsd_128:
+  case clang::X86::BI__builtin_ia32_selectsh_128:
+  case clang::X86::BI__builtin_ia32_selectsbf_128:
+    return interp__builtin_select_scalar(S, Call);
   case clang::X86::BI__builtin_ia32_vprotbi:
   case clang::X86::BI__builtin_ia32_vprotdi:
   case clang::X86::BI__builtin_ia32_vprotqi:

clang/lib/AST/ExprConstant.cpp

@@ -12202,6 +12202,24 @@ bool VectorExprEvaluator::VisitCallExpr(const CallExpr *E) {
         return Success(APValue(ResultElements.data(), SourceLen), E);
       };
 
+  auto EvalSelectScalar = [&](unsigned Len) -> bool {
+    APSInt Mask;
+    APValue AVal, WVal;
+    if (!EvaluateInteger(E->getArg(0), Mask, Info) ||
+        !EvaluateAsRValue(Info, E->getArg(1), AVal) ||
+        !EvaluateAsRValue(Info, E->getArg(2), WVal))
+      return false;
+
+    bool TakeA0 = (Mask.getZExtValue() & 1u) != 0;
+    SmallVector<APValue, 4> Res;
+    Res.reserve(Len);
+    Res.push_back(TakeA0 ? AVal.getVectorElt(0) : WVal.getVectorElt(0));
+    for (unsigned I = 1; I < Len; ++I)
+      Res.push_back(WVal.getVectorElt(I));
+    APValue V(Res.data(), Res.size());
+    return Success(V, E);
+  };
+
   switch (E->getBuiltinCallee()) {
   default:
     return false;
@@ -12505,6 +12523,13 @@ bool VectorExprEvaluator::VisitCallExpr(const CallExpr *E) {
         return APInt((Src).trunc(DstBits));
       return APInt::getAllOnes(DstBits);
     });
+  case clang::X86::BI__builtin_ia32_selectss_128:
+    return EvalSelectScalar(4);
+  case clang::X86::BI__builtin_ia32_selectsd_128:
+    return EvalSelectScalar(2);
+  case clang::X86::BI__builtin_ia32_selectsh_128:
+  case clang::X86::BI__builtin_ia32_selectsbf_128:
+    return EvalSelectScalar(8);
   case clang::X86::BI__builtin_ia32_pmuldq128:
   case clang::X86::BI__builtin_ia32_pmuldq256:
   case clang::X86::BI__builtin_ia32_pmuldq512: