[AMDGPU] Emit predefined macro `__AMDGCN_CUMODE__`

[yaxunl]

Summary

Predefine __AMDGCN_CUMODE__ as 1 or 0 when compilation assumes CU or WGP modes.

If WGP mode is not supported, ignore -mno-cumode and emit a warning.

This is needed for implementing device functions like __smid (https://github.com/ROCm-Developer-Tools/hipamd/blob/312dff7b794337aa040be0691acc78e9f968a8d2/include/hip/amd_detail/amd_device_functions.h#L957)

[arsenm]

I think exposing whether or not the flag was used is weird/broken, as is including _OPTION in the name. Should just define to whether it's enabled or not

[yaxunl]

In D145343#4170250, @arsenm wrote:

I think exposing whether or not the flag was used is weird/broken, as is including _OPTION in the name. Should just define to whether it's enabled or not

I agree. @b-sumner What do you think?

[b-sumner]

In D145343#4170305, @yaxunl wrote:

In D145343#4170250, @arsenm wrote:

I think exposing whether or not the flag was used is weird/broken, as is including _OPTION in the name. Should just define to whether it's enabled or not

I agree. @b-sumner What do you think?

Applications may need to check if CUMode is set at compile time, and they may also need to check if an old compiler is being used. So an macro that says whether it's enabled or not is fine, and if no macro is defined, then the application can conclude the compiler is old

[tra]

Can __has_feature be used to accomplish what you need?

[arsenm]

We can’t let specific flag usage leak into the semantics. Cu mode is on or off. If someone really cares about supporting older compilers they could always define their own macro

[yaxunl]

Revised by Matt's comments.

[yaxunl]

In D145343#4172643, @tra wrote:

Can __has_feature be used to accomplish what you need?

There are a few issues using __has_feature or similar builtin macros to check target feature, which has been discussed in https://reviews.llvm.org/D125555 . One of the concerns is the stability of target feature name.

Using a predefined macro does not have such concern.

[arsenm]

clang/lib/Basic/Targets/AMDGPU.cpp

318

Why do we sometimes use __ on both sides, and sometimes only leading __?

[yaxunl]

clang/lib/Basic/Targets/AMDGPU.cpp

318

We did not establish a naming convention for pre-defined macros.

The rationale of only prefixing with __ is that:

1. it is shorter

2. many other targets following that for target feature related macros, e.g. https://github.com/llvm/llvm-project/blob/main/clang/test/Preprocessor/arm-target-features.c , https://github.com/llvm/llvm-project/blob/main/clang/test/Preprocessor/riscv-target-features.c

The rationale of adding __ on both sides is:

1. most gcc predefined macros follow this convention https://gcc.gnu.org/onlinedocs/cpp/Common-Predefined-Macros.html

2. some targets follow this convention for target feature related macros, e.g https://github.com/llvm/llvm-project/blob/main/clang/test/Preprocessor/x86_target_features.c

I think it is not too late to establish a naming convention for predefined macros for amdgpu target. I would prefer to adding __ on both sides since

1. most of the existing amdgpu predefined macros follow that. We just need to add __AMDGCN_WAVEFRONT_SIZE__ and phasing out __AMDGCN_WAVEFRONT_SIZE. This will be the least interruptive.

2. we will conform to GCC naming convention for pre-defined macros.

Or maybe we should have an RFC for this topic.

[yaxunl]

rename the macro to be consistent with most existing predefined macros for amdgcn target

[arsenm]

clang/include/clang/Basic/DiagnosticDriverKinds.td

121–123

I'm surprised there already isn't a warning for this?

clang/lib/Driver/ToolChains/CommonArgs.cpp

146

I don't understand the use of "no-cumode". Where is this defined?

[yaxunl]

clang/include/clang/Basic/DiagnosticDriverKinds.td

121–123

I am surprised too. Before this patch, there is no check and no handling of unsupported target features in Clang. clang just blindly adds target features specified in the command line and relies on the backend to handle them properly.

This patch enables Clang to warn and ignore unsupported target features.

clang/lib/Driver/ToolChains/CommonArgs.cpp

146

This function is called by clang driver. The argument TargetFeature is the target feature command line option to be checked, with -m removed, e.g. -mcumode or -mno-cumode passed to this function as "cumode" or "no-cumode".

[arsenm]

clang/lib/Driver/ToolChains/CommonArgs.cpp

146

Can this use something like

if (!DriverArgs.hasFlag(options::OPT_fgpu_sanitize,
                        options::OPT_fno_gpu_sanitize, true))
  return true;

to make it clear this is checking the flag? At a minimum this needs to be renamed to indicate it's a flag and not the actual subtarget feature name

[yaxunl]

revised by Matt's comments

clang/include/clang/Basic/DiagnosticDriverKinds.td

def warn_drv_unsupported_option_for_target : Warning<
   "ignoring '%0' option as it is not currently supported for target '%1'">,
   InGroup<OptionIgnored>;
 def warn_drv_unsupported_option_for_flang : Warning<
   "the argument '%0' is not supported for option '%1'. Mapping to '%1%2'">,
   InGroup<OptionIgnored>;
 def warn_drv_unsupported_diag_option_for_flang : Warning<
   "The warning option '-%0' is not supported">,
   InGroup<OptionIgnored>;
+def warn_drv_unsupported_option_for_processor : Warning<
+  "ignoring '%0' option as it is not currently supported for processor '%1'">,
+  InGroup<OptionIgnored>;
 
 def err_drv_invalid_thread_model_for_target : Error<
   "invalid thread model '%0' in '%1' for this target">;
 def err_drv_invalid_linker_name : Error<
   "invalid linker name in argument '%0'">;
 def err_drv_invalid_rtlib_name : Error<
   "invalid runtime library name in argument '%0'">;
 def err_drv_unsupported_rtlib_for_platform : Error<

clang/lib/Basic/Targets/AMDGPU.h

class LLVM_LIBRARY_VISIBILITY AMDGPUTargetInfo final : public TargetInfo {
   };
   static const LangASMap AMDGPUDefIsGenMap;
   static const LangASMap AMDGPUDefIsPrivMap;
 
   llvm::AMDGPU::GPUKind GPUKind;
   unsigned GPUFeatures;
   unsigned WavefrontSize;
 
+  /// Whether to use cumode or WGP mode. True for cumode. False for WGP mode.
+  bool CUMode;
+
   /// Target ID is device name followed by optional feature name postfixed
   /// by plus or minus sign delimitted by colon, e.g. gfx908:xnack+:sramecc-.
   /// If the target ID contains feature+, map it to true.
   /// If the target ID contains feature-, map it to false.
   /// If the target ID does not contain a feature (default), do not map it.
   llvm::StringMap<bool> OffloadArchFeatures;
   std::string TargetID;
 
public:
   bool handleTargetFeatures(std::vector<std::string> &Features,
                             DiagnosticsEngine &Diags) override {
     auto TargetIDFeatures =
         getAllPossibleTargetIDFeatures(getTriple(), getArchNameAMDGCN(GPUKind));
     for (const auto &F : Features) {
       assert(F.front() == '+' || F.front() == '-');
       if (F == "+wavefrontsize64")
         WavefrontSize = 64;
+      else if (F == "+cumode")
+        CUMode = true;
+      else if (F == "-cumode")
+        CUMode = false;
       bool IsOn = F.front() == '+';
       StringRef Name = StringRef(F).drop_front();
       if (!llvm::is_contained(TargetIDFeatures, Name))
         continue;
       assert(!OffloadArchFeatures.contains(Name));
       OffloadArchFeatures[Name] = IsOn;
     }
     return true;

clang/lib/Basic/Targets/AMDGPU.cpp

 
 #include "AMDGPU.h"
 #include "clang/Basic/Builtins.h"
 #include "clang/Basic/CodeGenOptions.h"
 #include "clang/Basic/Diagnostic.h"
 #include "clang/Basic/LangOptions.h"
 #include "clang/Basic/MacroBuilder.h"
 #include "clang/Basic/TargetBuiltins.h"
-
 using namespace clang;
 using namespace clang::targets;
 
 namespace clang {
 namespace targets {
 
 // If you edit the description strings, make sure you update
 // getPointerWidthV().
AMDGPUTargetInfo::AMDGPUTargetInfo(const llvm::Triple &Triple,
   if (getMaxPointerWidth() == 64) {
     LongWidth = LongAlign = 64;
     SizeType = UnsignedLong;
     PtrDiffType = SignedLong;
     IntPtrType = SignedLong;
   }
 
   MaxAtomicPromoteWidth = MaxAtomicInlineWidth = 64;
+  CUMode = !(GPUFeatures & llvm::AMDGPU::FEATURE_WGP);
 }
 
 void AMDGPUTargetInfo::adjust(DiagnosticsEngine &Diags, LangOptions &Opts) {
   TargetInfo::adjust(Diags, Opts);
   // ToDo: There are still a few places using default address space as private
   // address space in OpenCL, which needs to be cleaned up, then Opts.OpenCL
   // can be removed from the following line.
   setAddressSpaceMap(/*DefaultIsPrivate=*/Opts.OpenCL ||
void AMDGPUTargetInfo::getTargetDefines(const LangOptions &Opts,
   if (hasLDEXPF())
     Builder.defineMacro("__HAS_LDEXPF__");
   if (hasFP64())
     Builder.defineMacro("__HAS_FP64__");
   if (hasFastFMA())
     Builder.defineMacro("FP_FAST_FMA");
 
   Builder.defineMacro("__AMDGCN_WAVEFRONT_SIZE", Twine(WavefrontSize));
+  Builder.defineMacro("__AMDGCN_CUMODE__", Twine(CUMode));
 }
 
 void AMDGPUTargetInfo::setAuxTarget(const TargetInfo *Aux) {
   assert(HalfFormat == Aux->HalfFormat);
   assert(FloatFormat == Aux->FloatFormat);
   assert(DoubleFormat == Aux->DoubleFormat);
 
   // On x86_64 long double is 80-bit extended precision format, which is

clang/lib/Driver/ToolChains/AMDGPU.cpp

if (OptionalGpuArch) {
       }
     }
   }
 
   if (Args.hasFlag(options::OPT_mwavefrontsize64,
                    options::OPT_mno_wavefrontsize64, false))
     Features.push_back("+wavefrontsize64");
 
-  handleTargetFeaturesGroup(
-    Args, Features, options::OPT_m_amdgpu_Features_Group);
+  handleTargetFeaturesGroup(D, Triple, Args, Features,
+                            options::OPT_m_amdgpu_Features_Group);
 }
 
 /// AMDGPU Toolchain
 AMDGPUToolChain::AMDGPUToolChain(const Driver &D, const llvm::Triple &Triple,
                                  const ArgList &Args)
     : Generic_ELF(D, Triple, Args),
       OptionsDefault(
           {{options::OPT_O, "3"}, {options::OPT_cl_std_EQ, "CL1.2"}}) {

clang/lib/Driver/ToolChains/Arch/PPC.cpp

 }
 
 void ppc::getPPCTargetFeatures(const Driver &D, const llvm::Triple &Triple,
                                const ArgList &Args,
                                std::vector<StringRef> &Features) {
   if (Triple.getSubArch() == llvm::Triple::PPCSubArch_spe)
     Features.push_back("+spe");
 
-  handleTargetFeaturesGroup(Args, Features, options::OPT_m_ppc_Features_Group);
+  handleTargetFeaturesGroup(D, Triple, Args, Features,
+                            options::OPT_m_ppc_Features_Group);
 
   ppc::FloatABI FloatABI = ppc::getPPCFloatABI(D, Args);
   if (FloatABI == ppc::FloatABI::Soft)
     Features.push_back("-hard-float");
 
   ppc::ReadGOTPtrMode ReadGOT = ppc::getPPCReadGOTPtrMode(D, Triple, Args);
   if (ReadGOT == ppc::ReadGOTPtrMode::SecurePlt)
     Features.push_back("+secure-plt");

clang/lib/Driver/ToolChains/Arch/RISCV.cpp

void riscv::getRISCVTargetFeatures(const Driver &D, const llvm::Triple &Triple,
   // specified.
   if (Args.hasFlag(options::OPT_msave_restore, options::OPT_mno_save_restore, false))
     Features.push_back("+save-restore");
   else
     Features.push_back("-save-restore");
 
   // Now add any that the user explicitly requested on the command line,
   // which may override the defaults.
-  handleTargetFeaturesGroup(Args, Features, options::OPT_m_riscv_Features_Group);
+  handleTargetFeaturesGroup(D, Triple, Args, Features,
+                            options::OPT_m_riscv_Features_Group);
 }
 
 StringRef riscv::getRISCVABI(const ArgList &Args, const llvm::Triple &Triple) {
   assert(Triple.isRISCV() && "Unexpected triple");
 
   // GCC's logic around choosing a default `-mabi=` is complex. If GCC is not
   // configured using `--with-abi=`, then the logic for the default choice is
   // defined in config.gcc. This function is based on the logic in GCC 9.2.0.

clang/lib/Driver/ToolChains/CommonArgs.h

void getTargetFeatures(const Driver &D, const llvm::Triple &Triple,
                        bool IsAux = false);
 
 /// Iterate \p Args and convert -mxxx to +xxx and -mno-xxx to -xxx and
 /// append it to \p Features.
 ///
 /// Note: Since \p Features may contain default values before calling
 /// this function, or may be appended with entries to override arguments,
 /// entries in \p Features are not unique.
-void handleTargetFeaturesGroup(const llvm::opt::ArgList &Args,
+void handleTargetFeaturesGroup(const Driver &D, const llvm::Triple &Triple,
+                               const llvm::opt::ArgList &Args,
                                std::vector<StringRef> &Features,
                                llvm::opt::OptSpecifier Group);
 
 /// If there are multiple +xxx or -xxx features, keep the last one.
 SmallVector<StringRef> unifyTargetFeatures(ArrayRef<StringRef> Features);
 
 /// Handles the -save-stats option and returns the filename to save statistics
 /// to.

clang/lib/Driver/ToolChains/CommonArgs.cpp

static void renderRemarksHotnessOptions(const ArgList &Args,
 
   if (const Arg *A =
           Args.getLastArg(options::OPT_fdiagnostics_hotness_threshold_EQ))
     CmdArgs.push_back(
         Args.MakeArgString(Twine(PluginOptPrefix) +
                            "opt-remarks-hotness-threshold=" + A->getValue()));
 }
 
+static bool shouldIgnoreUnsupportedTargetFeature(const Arg &TargetFeatureArg,
+                                                 llvm::Triple T,
+                                                 StringRef Processor) {
+  // Warn no-cumode for AMDGCN processors not supporing WGP mode.
+  if (!T.isAMDGPU())
+    return false;
+  auto GPUKind = T.isAMDGCN() ? llvm::AMDGPU::parseArchAMDGCN(Processor)
+                              : llvm::AMDGPU::parseArchR600(Processor);
+  auto GPUFeatures = T.isAMDGCN() ? llvm::AMDGPU::getArchAttrAMDGCN(GPUKind)
+                                  : llvm::AMDGPU::getArchAttrR600(GPUKind);
+  if (GPUFeatures & llvm::AMDGPU::FEATURE_WGP)
+    return false;
+  return TargetFeatureArg.getOption().matches(options::OPT_mno_cumode);
+}
+
 void tools::addPathIfExists(const Driver &D, const Twine &Path,
                             ToolChain::path_list &Paths) {
   if (D.getVFS().exists(Path))
     Paths.push_back(Path.str());
 }
 
-void tools::handleTargetFeaturesGroup(const ArgList &Args,
+void tools::handleTargetFeaturesGroup(const Driver &D,
+                                      const llvm::Triple &Triple,
+                                      const ArgList &Args,
                                       std::vector<StringRef> &Features,
                                       OptSpecifier Group) {
+  std::set<StringRef> Warned;
   for (const Arg *A : Args.filtered(Group)) {
     StringRef Name = A->getOption().getName();
     A->claim();
 
     // Skip over "-m".
     assert(Name.startswith("m") && "Invalid feature name.");
     Name = Name.substr(1);
 
+    auto Proc = getCPUName(D, Args, Triple);
+    if (shouldIgnoreUnsupportedTargetFeature(*A, Triple, Proc)) {
+      if (Warned.count(Name) == 0) {
+        D.getDiags().Report(
+            clang::diag::warn_drv_unsupported_option_for_processor)
+            << A->getAsString(Args) << Proc;
+        Warned.insert(Name);
+      }
+      continue;
+    }
+
     bool IsNegative = Name.startswith("no-");
     if (IsNegative)
       Name = Name.substr(3);
+
     Features.push_back(Args.MakeArgString((IsNegative ? "-" : "+") + Name));
   }
 }
 
 SmallVector<StringRef>
 tools::unifyTargetFeatures(ArrayRef<StringRef> Features) {
   // Only add a feature if it hasn't been seen before starting from the end.
   SmallVector<StringRef> UnifiedFeatures;
case llvm::Triple::amdgcn:
     return getAMDGPUTargetGPU(T, Args);
 
   case llvm::Triple::wasm32:
   case llvm::Triple::wasm64:
     return std::string(getWebAssemblyTargetCPU(Args));
   }
 }
 
-static void getWebAssemblyTargetFeatures(const ArgList &Args,
+static void getWebAssemblyTargetFeatures(const Driver &D,
+                                         const llvm::Triple &Triple,
+                                         const ArgList &Args,
                                          std::vector<StringRef> &Features) {
-  handleTargetFeaturesGroup(Args, Features, options::OPT_m_wasm_Features_Group);
+  handleTargetFeaturesGroup(D, Triple, Args, Features,
+                            options::OPT_m_wasm_Features_Group);
 }
 
 void tools::getTargetFeatures(const Driver &D, const llvm::Triple &Triple,
                               const ArgList &Args, ArgStringList &CmdArgs,
                               bool ForAS, bool IsAux) {
   std::vector<StringRef> Features;
   switch (Triple.getArch()) {
   default:
void tools::getTargetFeatures(const Driver &D, const llvm::Triple &Triple,
   case llvm::Triple::aarch64_be:
     aarch64::getAArch64TargetFeatures(D, Triple, Args, Features, ForAS);
     break;
   case llvm::Triple::x86:
   case llvm::Triple::x86_64:
     x86::getX86TargetFeatures(D, Triple, Args, Features);
     break;
   case llvm::Triple::hexagon:
-    hexagon::getHexagonTargetFeatures(D, Args, Features);
+    hexagon::getHexagonTargetFeatures(D, Triple, Args, Features);
     break;
   case llvm::Triple::wasm32:
   case llvm::Triple::wasm64:
-    getWebAssemblyTargetFeatures(Args, Features);
+    getWebAssemblyTargetFeatures(D, Triple, Args, Features);
     break;
   case llvm::Triple::sparc:
   case llvm::Triple::sparcel:
   case llvm::Triple::sparcv9:
     sparc::getSparcTargetFeatures(D, Args, Features);
     break;
   case llvm::Triple::r600:
   case llvm::Triple::amdgcn:

clang/lib/Driver/ToolChains/Hexagon.h

public:
   virtual void RenderExtraToolArgs(const JobAction &JA,
                                    llvm::opt::ArgStringList &CmdArgs) const;
   void ConstructJob(Compilation &C, const JobAction &JA,
                     const InputInfo &Output, const InputInfoList &Inputs,
                     const llvm::opt::ArgList &TCArgs,
                     const char *LinkingOutput) const override;
 };
 
-void getHexagonTargetFeatures(const Driver &D, const llvm::opt::ArgList &Args,
+void getHexagonTargetFeatures(const Driver &D, const llvm::Triple &Triple,
+                              const llvm::opt::ArgList &Args,
                               std::vector<StringRef> &Features);
 
 } // end namespace hexagon.
 } // end namespace tools
 
 namespace toolchains {
 
 class LLVM_LIBRARY_VISIBILITY HexagonToolChain : public Linux {

clang/lib/Driver/ToolChains/Hexagon.cpp

static void handleHVXTargetFeatures(const Driver &D, const ArgList &Args,
   }
   if (auto F = checkFlagHvxVersion(options::OPT_mhexagon_hvx_ieee_fp,
                                    options::OPT_mno_hexagon_hvx_ieee_fp, 68)) {
     Features.push_back(*F);
   }
 }
 
 // Hexagon target features.
-void hexagon::getHexagonTargetFeatures(const Driver &D, const ArgList &Args,
+void hexagon::getHexagonTargetFeatures(const Driver &D,
+                                       const llvm::Triple &Triple,
+                                       const ArgList &Args,
                                        std::vector<StringRef> &Features) {
-  handleTargetFeaturesGroup(Args, Features,
+  handleTargetFeaturesGroup(D, Triple, Args, Features,
                             options::OPT_m_hexagon_Features_Group);
 
   bool UseLongCalls = false;
   if (Arg *A = Args.getLastArg(options::OPT_mlong_calls,
                                options::OPT_mno_long_calls)) {
     if (A->getOption().matches(options::OPT_mlong_calls))
       UseLongCalls = true;
   }

clang/test/CodeGenHIP/hip-cumode.hip

+// REQUIRES: amdgpu-registered-target
+
+// RUN: %clang -S -o - --offload-arch=gfx906 --cuda-device-only -nogpuinc -nogpulib \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=NOWGP %s
+// RUN: %clang -S -o - --offload-arch=gfx906 --cuda-device-only -nogpuinc -nogpulib -mcumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=NOWGP %s
+// RUN: %clang -S -o - --offload-arch=gfx906 --cuda-device-only -nogpuinc -nogpulib -mno-cumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefixes=NOWGP,WARN-CUMODE %s
+// RUN: %clang -S -o - --offload-arch=gfx1030 --cuda-device-only -nogpuinc -nogpulib \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-OFF %s
+// RUN: %clang -S -o - --offload-arch=gfx1030 --cuda-device-only -nogpuinc -nogpulib -mcumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-ON %s
+// RUN: %clang -S -o - --offload-arch=gfx1030 --cuda-device-only -nogpuinc -nogpulib -mno-cumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-OFF %s
+// WARN-CUMODE: warning: ignoring '-mno-cumode' option as it is not currently supported for processor 'gfx906' [-Woption-ignored]
+// NOWGP-NOT: .amdhsa_workgroup_processor_mode
+// CUMODE-ON: .amdhsa_workgroup_processor_mode 0
+// CUMODE-OFF: .amdhsa_workgroup_processor_mode 1
+
+__attribute__((global)) void kernel() {}

clang/test/Driver/amdgpu-macros.cl

 // RUN: %clang -E -dM -target amdgcn -mcpu=gfx1010 -mwavefrontsize64 \
 // RUN:   -mno-wavefrontsize64 %s 2>&1 | FileCheck --check-prefix=WAVE32 %s
 // RUN: %clang -E -dM -target amdgcn -mcpu=gfx906 -mno-wavefrontsize64 \
 // RUN:   -mwavefrontsize64 %s 2>&1 | FileCheck --check-prefix=WAVE64 %s
 // RUN: %clang -E -dM -target amdgcn -mcpu=gfx1010 -mno-wavefrontsize64 \
 // RUN:   -mwavefrontsize64 %s 2>&1 | FileCheck --check-prefix=WAVE64 %s
 // WAVE64-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
 // WAVE32-DAG: #define __AMDGCN_WAVEFRONT_SIZE 32
+
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx906 \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-ON %s
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx906 -mcumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-ON %s
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx906 -mno-cumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefixes=CUMODE-ON,WARN-CUMODE %s
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx1030 \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-OFF %s
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx1030 -mcumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-ON %s
+// RUN: %clang -E -dM -target amdgcn -mcpu=gfx1030 -mno-cumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-OFF %s
+// WARN-CUMODE-DAG: warning: ignoring '-mno-cumode' option as it is not currently supported for processor 'gfx906' [-Woption-ignored]
+// CUMODE-ON-DAG: #define __AMDGCN_CUMODE__ 1
+// CUMODE-OFF-DAG: #define __AMDGCN_CUMODE__ 0

clang/test/Driver/hip-macros.hip

 // RUN: %clang -E -dM --offload-arch=gfx906 -mno-wavefrontsize64 \
 // RUN:   --cuda-device-only -nogpuinc -nogpulib \
 // RUN:   -mwavefrontsize64 %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
 // RUN: %clang -E -dM --offload-arch=gfx1010 -mno-wavefrontsize64 \
 // RUN:   --cuda-device-only -nogpuinc -nogpulib \
 // RUN:   -mwavefrontsize64 %s 2>&1 | FileCheck --check-prefixes=WAVE64 %s
 // WAVE64-DAG: #define __AMDGCN_WAVEFRONT_SIZE 64
 // WAVE32-DAG: #define __AMDGCN_WAVEFRONT_SIZE 32
+
+// RUN: %clang -E -dM --offload-arch=gfx906 --cuda-device-only -nogpuinc -nogpulib \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-ON %s
+// RUN: %clang -E -dM --offload-arch=gfx906 --cuda-device-only -nogpuinc -nogpulib -mcumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-ON %s
+// RUN: %clang -E -dM --offload-arch=gfx906 --cuda-device-only -nogpuinc -nogpulib -mno-cumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefixes=CUMODE-ON,WARN-CUMODE %s
+// RUN: %clang -E -dM --offload-arch=gfx1030 --cuda-device-only -nogpuinc -nogpulib \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-OFF %s
+// RUN: %clang -E -dM --offload-arch=gfx1030 --cuda-device-only -nogpuinc -nogpulib -mcumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-ON %s
+// RUN: %clang -E -dM --offload-arch=gfx1030 --cuda-device-only -nogpuinc -nogpulib -mno-cumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefix=CUMODE-OFF %s
+
+// Check no duplicate warnings.
+// RUN: %clang -E -dM --offload-arch=gfx906 --cuda-device-only -nogpuinc -nogpulib -mcumode \
+// RUN:   -mno-cumode -mno-cumode \
+// RUN:   %s 2>&1 | FileCheck --check-prefixes=CUMODE-ON,WARN-CUMODE %s
+
+// WARN-CUMODE-DAG: warning: ignoring '-mno-cumode' option as it is not currently supported for processor 'gfx906' [-Woption-ignored]
+// WARN-CUMODE-NOT: warning: ignoring '-mno-cumode' option as it is not currently supported for processor 'gfx906' [-Woption-ignored]
+// CUMODE-ON-DAG: #define __AMDGCN_CUMODE__ 1
+// CUMODE-OFF-DAG: #define __AMDGCN_CUMODE__ 0

llvm/include/llvm/TargetParser/TargetParser.h

enum ArchFeatureKind : uint32_t {
   // Wavefront 32 is available.
   FEATURE_WAVE32 = 1 << 6,
 
   // Xnack is available.
   FEATURE_XNACK = 1 << 7,
 
   // Sram-ecc is available.
   FEATURE_SRAMECC = 1 << 8,
+
+  // WGP mode is supported.
+  FEATURE_WGP = 1 << 9,
 };
 
 StringRef getArchNameAMDGCN(GPUKind AK);
 StringRef getArchNameR600(GPUKind AK);
 StringRef getCanonicalArchName(const Triple &T, StringRef Arch);
 GPUKind parseArchAMDGCN(StringRef CPU);
 GPUKind parseArchR600(StringRef CPU);
 unsigned getArchAttrAMDGCN(GPUKind AK);

llvm/lib/TargetParser/TargetParser.cpp

constexpr GPUInfo AMDGCNGPUs[] = {
   {{"gfx906"},    {"gfx906"},  GK_GFX906,  FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_XNACK|FEATURE_SRAMECC},
   {{"gfx908"},    {"gfx908"},  GK_GFX908,  FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_XNACK|FEATURE_SRAMECC},
   {{"gfx909"},    {"gfx909"},  GK_GFX909,  FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_XNACK},
   {{"gfx90a"},    {"gfx90a"},  GK_GFX90A,  FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_XNACK|FEATURE_SRAMECC},
   {{"gfx90c"},    {"gfx90c"},  GK_GFX90C,  FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_XNACK},
   {{"gfx940"},    {"gfx940"},  GK_GFX940,  FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_XNACK|FEATURE_SRAMECC},
   {{"gfx941"},    {"gfx941"},  GK_GFX941,  FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_XNACK|FEATURE_SRAMECC},
   {{"gfx942"},    {"gfx942"},  GK_GFX942,  FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_XNACK|FEATURE_SRAMECC},
-  {{"gfx1010"},   {"gfx1010"}, GK_GFX1010, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_XNACK},
-  {{"gfx1011"},   {"gfx1011"}, GK_GFX1011, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_XNACK},
-  {{"gfx1012"},   {"gfx1012"}, GK_GFX1012, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_XNACK},
-  {{"gfx1013"},   {"gfx1013"}, GK_GFX1013, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_XNACK},
-  {{"gfx1030"},   {"gfx1030"}, GK_GFX1030, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
-  {{"gfx1031"},   {"gfx1031"}, GK_GFX1031, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
-  {{"gfx1032"},   {"gfx1032"}, GK_GFX1032, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
-  {{"gfx1033"},   {"gfx1033"}, GK_GFX1033, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
-  {{"gfx1034"},   {"gfx1034"}, GK_GFX1034, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
-  {{"gfx1035"},   {"gfx1035"}, GK_GFX1035, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
-  {{"gfx1036"},   {"gfx1036"}, GK_GFX1036, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
-  {{"gfx1100"},   {"gfx1100"}, GK_GFX1100, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
-  {{"gfx1101"},   {"gfx1101"}, GK_GFX1101, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
-  {{"gfx1102"},   {"gfx1102"}, GK_GFX1102, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
-  {{"gfx1103"},   {"gfx1103"}, GK_GFX1103, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32},
+  {{"gfx1010"},   {"gfx1010"}, GK_GFX1010, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_XNACK|FEATURE_WGP},
+  {{"gfx1011"},   {"gfx1011"}, GK_GFX1011, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_XNACK|FEATURE_WGP},
+  {{"gfx1012"},   {"gfx1012"}, GK_GFX1012, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_XNACK|FEATURE_WGP},
+  {{"gfx1013"},   {"gfx1013"}, GK_GFX1013, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_XNACK|FEATURE_WGP},
+  {{"gfx1030"},   {"gfx1030"}, GK_GFX1030, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
+  {{"gfx1031"},   {"gfx1031"}, GK_GFX1031, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
+  {{"gfx1032"},   {"gfx1032"}, GK_GFX1032, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
+  {{"gfx1033"},   {"gfx1033"}, GK_GFX1033, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
+  {{"gfx1034"},   {"gfx1034"}, GK_GFX1034, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
+  {{"gfx1035"},   {"gfx1035"}, GK_GFX1035, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
+  {{"gfx1036"},   {"gfx1036"}, GK_GFX1036, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
+  {{"gfx1100"},   {"gfx1100"}, GK_GFX1100, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
+  {{"gfx1101"},   {"gfx1101"}, GK_GFX1101, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
+  {{"gfx1102"},   {"gfx1102"}, GK_GFX1102, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
+  {{"gfx1103"},   {"gfx1103"}, GK_GFX1103, FEATURE_FAST_FMA_F32|FEATURE_FAST_DENORMAL_F32|FEATURE_WAVE32|FEATURE_WGP},
 };
 
 const GPUInfo *getArchEntry(AMDGPU::GPUKind AK, ArrayRef<GPUInfo> Table) {
   GPUInfo Search = { {""}, {""}, AK, AMDGPU::FEATURE_NONE };
 
   auto I =
       llvm::lower_bound(Table, Search, [](const GPUInfo &A, const GPUInfo &B) {
         return A.Kind < B.Kind;