[CUDA][HIP] add options --[no-]offload-inc

yxsamliu · yxsamliu · commit 9945ef4a18fc · 2025-06-11T13:26:26.000-04:00
Currently there is only option -nogpuinc for disabling
the default CUDA/HIP wrapper headers. However, there
are situations where -nogpuinc needs to be overriden
for enabling CUDA/HIP wrapper headers. This patch
adds --[no-]offload-inc for that purpose. When both
exist, the last wins. -nogpuinc and -nocudainc are
now alias to --no-offload-inc.
diff --git a/clang/include/clang/Driver/Options.td b/clang/include/clang/Driver/Options.td
@@ -577,7 +577,8 @@ multiclass BoolWOption<string flag_base, KeyPathAndMacro kpm,
 // Works like BoolOption except without marshalling
 multiclass BoolOptionWithoutMarshalling<string prefix = "", string spelling_base,
                                         FlagDef flag1_base, FlagDef flag2_base,
-                                        BothFlags suffix = BothFlags<[]>> {
+                                        BothFlags suffix = BothFlags<[]>,
+                                        list<string> flag_prefix = ["-"]> {
   defvar flag1 = FlagDefExpanded<ApplySuffix<flag1_base, suffix>.Result, prefix,
                                  NAME, spelling_base>;
 
@@ -598,12 +599,12 @@ multiclass BoolOptionWithoutMarshalling<string prefix = "", string spelling_base
 
   defvar implied = !if(flag1.CanBeImplied, flag1, flag2);
 
-  def flag1.RecordName : Flag<["-"], flag1.Spelling>, Flags<flag1.OptionFlags>,
+  def flag1.RecordName : Flag<flag_prefix, flag1.Spelling>, Flags<flag1.OptionFlags>,
                          Visibility<flag1.OptionVisibility>,
                          HelpText<flag1.Help>,
                          ImpliedByAnyOf<implied.ImpliedBy, implied.ValueAsCode>
                          {}
-  def flag2.RecordName : Flag<["-"], flag2.Spelling>, Flags<flag2.OptionFlags>,
+  def flag2.RecordName : Flag<flag_prefix, flag2.Spelling>, Flags<flag2.OptionFlags>,
                          Visibility<flag2.OptionVisibility>,
                          HelpText<flag2.Help>,
                          ImpliedByAnyOf<implied.ImpliedBy, implied.ValueAsCode>
@@ -5742,12 +5743,17 @@ def nobuiltininc : Flag<["-"], "nobuiltininc">,
   Group<IncludePath_Group>,
   HelpText<"Disable builtin #include directories only">,
   MarshallingInfoNegativeFlag<HeaderSearchOpts<"UseBuiltinIncludes">>;
-def nogpuinc : Flag<["-"], "nogpuinc">, Group<IncludePath_Group>,
-  HelpText<"Do not add include paths for CUDA/HIP and"
-  " do not include the default CUDA/HIP wrapper headers">;
+defm offload_inc: BoolOptionWithoutMarshalling<"", "offload-inc",
+  PosFlag<SetTrue, [], [ClangOption], "Add include paths for CUDA/HIP and"
+   " include the default CUDA/HIP wrapper headers (default)">,
+  NegFlag<SetFalse, [], [ClangOption], "Do not add include paths for CUDA/HIP"
+   " and include the default CUDA/HIP wrapper headers">,
+   BothFlags<[]>, ["--"]>, Group<IncludePath_Group>;
+def : Flag<["-"], "nogpuinc">, Alias<no_offload_inc>;
+
 def nohipwrapperinc : Flag<["-"], "nohipwrapperinc">, Group<IncludePath_Group>,
   HelpText<"Do not include the default HIP wrapper headers and include paths">;
-def : Flag<["-"], "nocudainc">, Alias<nogpuinc>;
+def : Flag<["-"], "nocudainc">, Alias<no_offload_inc>;
 def no_offloadlib
     : Flag<["--"], "no-offloadlib">,
       MarshallingInfoFlag<LangOpts<"NoGPULib">>,
diff --git a/clang/lib/Driver/ToolChains/AMDGPU.cpp b/clang/lib/Driver/ToolChains/AMDGPU.cpp
@@ -525,7 +525,8 @@ void RocmInstallationDetector::AddHIPIncludeArgs(const ArgList &DriverArgs,
                     "hipstdpar_lib.hpp"});
   };
 
-  if (DriverArgs.hasArg(options::OPT_nogpuinc)) {
+  if (!DriverArgs.hasFlag(options::OPT_offload_inc, options::OPT_no_offload_inc,
+                          true)) {
     if (HasHipStdPar)
       HandleHipStdPar();
 
diff --git a/clang/lib/Driver/ToolChains/Clang.cpp b/clang/lib/Driver/ToolChains/Clang.cpp
@@ -937,7 +937,8 @@ void Clang::AddPreprocessingOptions(Compilation &C, const JobAction &JA,
   // openmp_wrappers folder which contains alternative system headers.
   if (JA.isDeviceOffloading(Action::OFK_OpenMP) &&
       !Args.hasArg(options::OPT_nostdinc) &&
-      !Args.hasArg(options::OPT_nogpuinc) &&
+      Args.hasFlag(options::OPT_offload_inc, options::OPT_no_offload_inc,
+                   true) &&
       getToolChain().getTriple().isGPU()) {
     if (!Args.hasArg(options::OPT_nobuiltininc)) {
       // Add openmp_wrappers/* to our system include path.  This lets us wrap
@@ -1120,7 +1121,8 @@ void Clang::AddPreprocessingOptions(Compilation &C, const JobAction &JA,
   // TODO: This should be moved to `AddClangSystemIncludeArgs` by passing the
   //       OffloadKind as an argument.
   if (!Args.hasArg(options::OPT_nostdinc) &&
-      !Args.hasArg(options::OPT_nogpuinc) &&
+      Args.hasFlag(options::OPT_offload_inc, options::OPT_no_offload_inc,
+                   true) &&
       !Args.hasArg(options::OPT_nobuiltininc)) {
     // Without an offloading language we will include these headers directly.
     // Offloading languages will instead only use the declarations stored in
diff --git a/clang/lib/Driver/ToolChains/Cuda.cpp b/clang/lib/Driver/ToolChains/Cuda.cpp
@@ -302,7 +302,8 @@ void CudaInstallationDetector::AddCudaIncludeArgs(
     CC1Args.push_back(DriverArgs.MakeArgString(P));
   }
 
-  if (DriverArgs.hasArg(options::OPT_nogpuinc))
+  if (!DriverArgs.hasFlag(options::OPT_offload_inc, options::OPT_no_offload_inc,
+                          true))
     return;
 
   if (!isValid()) {
@@ -928,7 +929,8 @@ llvm::DenormalMode CudaToolChain::getDefaultDenormalModeForType(
 void CudaToolChain::AddCudaIncludeArgs(const ArgList &DriverArgs,
                                        ArgStringList &CC1Args) const {
   // Check our CUDA version if we're going to include the CUDA headers.
-  if (!DriverArgs.hasArg(options::OPT_nogpuinc) &&
+  if (DriverArgs.hasFlag(options::OPT_offload_inc, options::OPT_no_offload_inc,
+                         true) &&
       !DriverArgs.hasArg(options::OPT_no_cuda_version_check)) {
     StringRef Arch = DriverArgs.getLastArgValue(options::OPT_march_EQ);
     assert(!Arch.empty() && "Must have an explicit GPU arch.");
@@ -1001,7 +1003,9 @@ void CudaToolChain::AddClangSystemIncludeArgs(const ArgList &DriverArgs,
                                               ArgStringList &CC1Args) const {
   HostTC.AddClangSystemIncludeArgs(DriverArgs, CC1Args);
 
-  if (!DriverArgs.hasArg(options::OPT_nogpuinc) && CudaInstallation.isValid())
+  if (DriverArgs.hasFlag(options::OPT_offload_inc, options::OPT_no_offload_inc,
+                         true) &&
+      CudaInstallation.isValid())
     CC1Args.append(
         {"-internal-isystem",
          DriverArgs.MakeArgString(CudaInstallation.getIncludePath())});
diff --git a/clang/lib/Driver/ToolChains/HIPSPV.cpp b/clang/lib/Driver/ToolChains/HIPSPV.cpp
@@ -187,7 +187,8 @@ void HIPSPVToolChain::AddIAMCUIncludeArgs(const ArgList &Args,
 
 void HIPSPVToolChain::AddHIPIncludeArgs(const ArgList &DriverArgs,
                                         ArgStringList &CC1Args) const {
-  if (DriverArgs.hasArg(options::OPT_nogpuinc))
+  if (!DriverArgs.hasFlag(options::OPT_offload_inc, options::OPT_no_offload_inc,
+                          true))
     return;
 
   StringRef hipPath = DriverArgs.getLastArgValue(options::OPT_hip_path_EQ);
diff --git a/clang/test/Driver/hip-include-path.hip b/clang/test/Driver/hip-include-path.hip
@@ -12,6 +12,10 @@
 // RUN:   -std=c++11 --rocm-path=%S/Inputs/rocm -nogpuinc -nogpulib %s 2>&1 \
 // RUN:   | FileCheck -check-prefixes=COMMON,CLANG,NOHIP %s
 
+// RUN: %clang -c -### --target=x86_64-unknown-linux-gnu --cuda-gpu-arch=gfx900 \
+// RUN:   -std=c++11 --rocm-path=%S/Inputs/rocm --no-offload-inc -nogpulib --offload-inc %s 2>&1 \
+// RUN:   | FileCheck -check-prefixes=COMMON,CLANG,HIP %s
+
 // COMMON-LABEL: "{{[^"]*}}clang{{[^"]*}}" "-cc1"
 // CLANG-SAME: "-internal-isystem" "{{[^"]*}}/lib{{[^"]*}}/clang/{{[^"]*}}/include/cuda_wrappers"
 // NOCLANG-NOT: "{{[^"]*}}/lib{{[^"]*}}/clang/{{[^"]*}}/include/cuda_wrappers"
