[CUDA] Define __CUDACC_DEBUG__ when compiling device code in debug mode

[tambre]

Summary

NVCC defines CUDACC_DEBUG when compiling device code in debug mode.
Let's do the same to help with compatibility between the two and possibly enable users to select debug-friendly code.

Implements PR45596.

[tra]

clang/lib/Driver/ToolChains/Cuda.cpp

632

Built-in preprocessor macros are defined in clang/lib/Frontend/InitPreprocessor.cpp, so the change should ideally go there.

Also, it's not clear what exactly does the macro mean. I am not convinced that clang must set this macro. Clang is not 1:1 compatible with NVCC anyways, and I'd rather not add quirks unless they are absolutely necessary. In this case, it can be easily added by the user.

[tambre]

Abandoning as there are good reasons to not add this.

clang/lib/Driver/ToolChains/Cuda.cpp

632

Agreed.

clang/lib/Driver/ToolChains/Cuda.cpp

if (DeviceOffloadingKind == Action::OFK_Cuda) {
     if (DriverArgs.hasFlag(options::OPT_fgpu_rdc, options::OPT_fno_gpu_rdc,
                            false))
       CC1Args.push_back("-fgpu-rdc");
   }
 
   if (DriverArgs.hasArg(options::OPT_nogpulib))
     return;
 
+  // NVCC defines __CUDACC_DEBUG__ for device code when compiling with debug
+  // info.
+  if (mustEmitDebugInfo(DriverArgs) == EmitSameDebugInfoAsHost)
+    CC1Args.push_back("-D__CUDACC_DEBUG__");
+
   std::string LibDeviceFile = CudaInstallation.getLibDeviceFile(GpuArch);
 
   if (LibDeviceFile.empty()) {
     if (DeviceOffloadingKind == Action::OFK_OpenMP &&
         DriverArgs.hasArg(options::OPT_S))
       return;
 
     getDriver().Diag(diag::err_drv_no_cuda_libdevice) << GpuArch;

clang/test/Driver/cuda-options.cu

 
 // e) --cuda-include-ptx=all overrides preceding --no-cuda-include-ptx=sm_XX
 // RUN: %clang -### -target x86_64-linux-gnu \
 // RUN:   --cuda-gpu-arch=sm_35 --cuda-gpu-arch=sm_30 \
 // RUN:   --no-cuda-include-ptx=sm_30 --cuda-include-ptx=all \
 // RUN:   -c %s 2>&1 \
 // RUN: | FileCheck -check-prefixes FATBIN-COMMON,PTX-SM35,PTX-SM30 %s
 
+// Verify that -g adds __CUDACC_DEBUG__ define.
+// RUN: %clang -### -target x86_64-linux-gnu -g %s 2>&1 \
+// RUN: | FileCheck -check-prefix DEBUG %s
 
 // ARCH-SM20: "-cc1"{{.*}}"-target-cpu" "sm_20"
 // NOARCH-SM20-NOT: "-cc1"{{.*}}"-target-cpu" "sm_20"
 // ARCH-SM30: "-cc1"{{.*}}"-target-cpu" "sm_30"
 // NOARCH-SM30-NOT: "-cc1"{{.*}}"-target-cpu" "sm_30"
 // ARCH-SM35: "-cc1"{{.*}}"-target-cpu" "sm_35"
 // NOARCH-SM35-NOT: "-cc1"{{.*}}"-target-cpu" "sm_35"
 // ARCHALLERROR: error: Unsupported CUDA gpu architecture: all

 // FATBIN-COMMON:fatbinary
 // FATBIN-COMMON: "--create" "[[FATBINARY:[^"]*]]"
 // FATBIN-COMMON: "--image=profile=sm_30,file=
 // PTX-SM30: "--image=profile=compute_30,file=
 // NOPTX-SM30-NOT: "--image=profile=compute_30,file=
 // FATBIN-COMMON: "--image=profile=sm_35,file=
 // PTX-SM35: "--image=profile=compute_35,file=
 // NOPTX-SM35-NOT: "--image=profile=compute_35,file=
+
+// DEBUG: -D__CUDACC_DEBUG__