[CUDA,Driver] Added --no-cuda-gpu-arch= option.

[tra]

Summary

This allows us to negate preceding --cuda-gpu-arch=X.
This comes handy when user needs to override default flags set for them by the build system.

[jlebar]

lib/Driver/Driver.cpp

1952

Extra linebreak

1960

We're using std::set, so I don't think this is necessary? You could have a comment on std::set saying that we use it to get a deterministic order, if you want.

test/Driver/cuda-options.cu

154

Extra linebreak

[tra]

Removed sorting and extraneous empty lines.

include/clang/Driver/Options.td

 def cuda_host_only : Flag<["--"], "cuda-host-only">,
   HelpText<"Compile CUDA code for host only.  Has no effect on non-CUDA "
            "compilations.">;
 def cuda_compile_host_device : Flag<["--"], "cuda-compile-host-device">,
   HelpText<"Compile CUDA code for both host and device (default).  Has no "
            "effect on non-CUDA compilations.">;
 def cuda_gpu_arch_EQ : Joined<["--"], "cuda-gpu-arch=">, Flags<[DriverOption]>,
   HelpText<"CUDA GPU architecture (e.g. sm_35).  May be specified more than once.">;
+def no_cuda_gpu_arch_EQ : Joined<["--"], "no-cuda-gpu-arch=">, Flags<[DriverOption]>,
+  HelpText<"Remove GPU architecture (e.g. sm_35) from the list of GPUs to compile for. "
+           "'all' resets the list to its default value.">;
 def cuda_noopt_device_debug : Flag<["--"], "cuda-noopt-device-debug">,
   HelpText<"Enable device-side debug info generation. Disables ptxas optimizations.">;
 def no_cuda_version_check : Flag<["--"], "no-cuda-version-check">,
   HelpText<"Don't error out if the detected version of the CUDA install is "
            "too low for the requested CUDA gpu architecture.">;
 def no_cuda_noopt_device_debug : Flag<["--"], "no-cuda-noopt-device-debug">;
 def cuda_path_EQ : Joined<["--"], "cuda-path=">, Group<i_Group>,
   HelpText<"CUDA installation path">;

lib/Driver/Driver.cpp

bool initialize() override {
       CompileHostOnly = PartialCompilationArg &&
                         PartialCompilationArg->getOption().matches(
                             options::OPT_cuda_host_only);
       CompileDeviceOnly = PartialCompilationArg &&
                           PartialCompilationArg->getOption().matches(
                               options::OPT_cuda_device_only);
 
       // Collect all cuda_gpu_arch parameters, removing duplicates.
-      llvm::SmallSet<CudaArch, 4> GpuArchs;
+      std::set<CudaArch> GpuArchs;
       bool Error = false;
       for (Arg *A : Args) {
-        if (!A->getOption().matches(options::OPT_cuda_gpu_arch_EQ))
+        if (!(A->getOption().matches(options::OPT_cuda_gpu_arch_EQ) ||
+              A->getOption().matches(options::OPT_no_cuda_gpu_arch_EQ)))
           continue;
         A->claim();
 
-        const auto &ArchStr = A->getValue();
+        const StringRef ArchStr = A->getValue();
+        if (A->getOption().matches(options::OPT_no_cuda_gpu_arch_EQ) &&
+            ArchStr == "all") {
+          GpuArchs.clear();
+          continue;
+        }
         CudaArch Arch = StringToCudaArch(ArchStr);
         if (Arch == CudaArch::UNKNOWN) {
           C.getDriver().Diag(clang::diag::err_drv_cuda_bad_gpu_arch) << ArchStr;
           Error = true;
-        } else if (GpuArchs.insert(Arch).second)
-          GpuArchList.push_back(Arch);
+        } else if (A->getOption().matches(options::OPT_cuda_gpu_arch_EQ))
+          GpuArchs.insert(Arch);
+        else if (A->getOption().matches(options::OPT_no_cuda_gpu_arch_EQ))
+          GpuArchs.erase(Arch);
+        else
+          llvm_unreachable("Unexpected option.");
       }
 
-      // Default to sm_20 which is the lowest common denominator for supported
-      // GPUs.
-      // sm_20 code should work correctly, if suboptimally, on all newer GPUs.
+      // Collect list of GPUs remaining in the set.
+      for (CudaArch Arch : GpuArchs)
+        GpuArchList.push_back(Arch);
+
+
+      // Default to sm_20 which is the lowest common denominator for
+      // supported GPUs.  sm_20 code should work correctly, if
+      // suboptimally, on all newer GPUs.
       if (GpuArchList.empty())
         GpuArchList.push_back(CudaArch::SM_20);
+      else
+        // Sort GpuArchList to make compilation order deterministic.
+        std::sort(GpuArchList.begin(), GpuArchList.end());
 
       return Error;
     }
   };
 
   /// OpenMP action builder. The host bitcode is passed to the device frontend
   /// and all the device linked images are passed to the host link phase.
   class OpenMPActionBuilder final : public DeviceActionBuilder {

test/Driver/cuda-options.cu

 // RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only \
 // RUN:   --cuda-compile-host-device %s 2>&1 \
 // RUN: | FileCheck -check-prefix DEVICE -check-prefix DEVICE-NOSAVE \
 // RUN:    -check-prefix HOST -check-prefix INCLUDES-DEVICE \
 // RUN:    -check-prefix LINK %s
 
 // Verify that --cuda-gpu-arch option passes the correct GPU archtecture to
 // device compilation.
-// RUN: %clang -### -target x86_64-linux-gnu --cuda-gpu-arch=sm_35 -c %s 2>&1 \
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-gpu-arch=sm_30 -c %s 2>&1 \
 // RUN: | FileCheck -check-prefix DEVICE -check-prefix DEVICE-NOSAVE \
-// RUN:    -check-prefix DEVICE-SM35 -check-prefix HOST \
+// RUN:    -check-prefix DEVICE-SM30 -check-prefix HOST \
 // RUN:    -check-prefix INCLUDES-DEVICE -check-prefix NOLINK %s
 
 // Verify that there is one device-side compilation per --cuda-gpu-arch args
 // and that all results are included on the host side.
 // RUN: %clang -### -target x86_64-linux-gnu \
 // RUN:   --cuda-gpu-arch=sm_35 --cuda-gpu-arch=sm_30 -c %s 2>&1 \
 // RUN: | FileCheck -check-prefix DEVICE -check-prefix DEVICE-NOSAVE \
-// RUN:    -check-prefix DEVICE2 -check-prefix DEVICE-SM35 \
-// RUN:    -check-prefix DEVICE2-SM30 -check-prefix HOST \
+// RUN:    -check-prefix DEVICE2 -check-prefix DEVICE-SM30 \
+// RUN:    -check-prefix DEVICE2-SM35 -check-prefix HOST \
 // RUN:    -check-prefix HOST-NOSAVE -check-prefix INCLUDES-DEVICE \
 // RUN:    -check-prefix NOLINK %s
 
 // Verify that device-side results are passed to the correct tool when
 // -save-temps is used.
 // RUN: %clang -### -target x86_64-linux-gnu -save-temps -c %s 2>&1 \
 // RUN: | FileCheck -check-prefix DEVICE -check-prefix DEVICE-SAVE \
 // RUN:    -check-prefix HOST -check-prefix HOST-SAVE -check-prefix NOLINK %s
 
 // Verify that device-side results are passed to the correct tool when
 // -fno-integrated-as is used.
 // RUN: %clang -### -target x86_64-linux-gnu -fno-integrated-as -c %s 2>&1 \
 // RUN: | FileCheck -check-prefix DEVICE -check-prefix DEVICE-NOSAVE \
 // RUN:    -check-prefix HOST -check-prefix HOST-NOSAVE \
 // RUN:    -check-prefix HOST-AS -check-prefix NOLINK %s
 
+// Verify that --[no-]cuda-gpu-arch arguments are handled correctly.
+// a) --no-cuda-gpu-arch=X negates preceeding --cuda-gpu-arch=X
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only \
+// RUN:   --cuda-gpu-arch=sm_35 --cuda-gpu-arch=sm_30 \
+// RUN:   --no-cuda-gpu-arch=sm_35 \
+// RUN:   -c %s 2>&1 \
+// RUN: | FileCheck -check-prefixes NOARCH-SM20,ARCH-SM30,NOARCH-SM35 %s
+
+// b) --no-cuda-gpu-arch=X negates more than one preceeding --cuda-gpu-arch=X
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only \
+// RUN:   --cuda-gpu-arch=sm_35 --cuda-gpu-arch=sm_35 --cuda-gpu-arch=sm_30 \
+// RUN:   --no-cuda-gpu-arch=sm_35 \
+// RUN:   -c %s 2>&1 \
+// RUN: | FileCheck -check-prefixes NOARCH-SM20,ARCH-SM30,NOARCH-SM35 %s
+
+// c) if --no-cuda-gpu-arch=X negates all preceeding --cuda-gpu-arch=X
+//    we default to sm_20 -- same as if no --cuda-gpu-arch were passed.
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only \
+// RUN:   --cuda-gpu-arch=sm_35 --cuda-gpu-arch=sm_30 \
+// RUN:   --no-cuda-gpu-arch=sm_35 --no-cuda-gpu-arch=sm_30 \
+// RUN:   -c %s 2>&1 \
+// RUN: | FileCheck -check-prefixes ARCH-SM20,NOARCH-SM30,NOARCH-SM35 %s
+
+// d) --no-cuda-gpu-arch=X is a no-op if there's no preceding --cuda-gpu-arch=X
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only \
+// RUN:   --cuda-gpu-arch=sm_35 --cuda-gpu-arch=sm_30\
+// RUN:   --no-cuda-gpu-arch=sm_50 \
+// RUN:   -c %s 2>&1 \
+// RUN: | FileCheck -check-prefixes NOARCH-SM20,ARCH-SM30,ARCH-SM35 %s
+
+// e) --no-cuda-gpu-arch=X does not affect following --cuda-gpu-arch=X
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only \
+// RUN:   --no-cuda-gpu-arch=sm_35 --no-cuda-gpu-arch=sm_30 \
+// RUN:   --cuda-gpu-arch=sm_35 --cuda-gpu-arch=sm_30 \
+// RUN:   -c %s 2>&1 \
+// RUN: | FileCheck -check-prefixes NOARCH-SM20,ARCH-SM30,ARCH-SM35 %s
+
+// f) --no-cuda-gpu-arch=all negates all preceding --cuda-gpu-arch=X
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only \
+// RUN:   --cuda-gpu-arch=sm_20 --cuda-gpu-arch=sm_30 \
+// RUN:   --no-cuda-gpu-arch=all \
+// RUN:   --cuda-gpu-arch=sm_35 \
+// RUN:   -c %s 2>&1 \
+// RUN: | FileCheck -check-prefixes NOARCH-SM20,NOARCH-SM30,ARCH-SM35 %s
+
+// g) There's no --cuda-gpu-arch=all
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only \
+// RUN:   --cuda-gpu-arch=all \
+// RUN:   -c %s 2>&1 \
+// RUN: | FileCheck -check-prefix ARCHALLERROR %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
+
+
 // Match device-side preprocessor and compiler phases with -save-temps.
 // DEVICE-SAVE: "-cc1" "-triple" "nvptx64-nvidia-cuda"
 // DEVICE-SAVE-SAME: "-aux-triple" "x86_64--linux-gnu"
 // DEVICE-SAVE-SAME: "-fcuda-is-device"
 // DEVICE-SAVE-SAME: "-x" "cuda"
 
 // DEVICE-SAVE: "-cc1" "-triple" "nvptx64-nvidia-cuda"
 // DEVICE-SAVE-SAME: "-aux-triple" "x86_64--linux-gnu"
 // DEVICE-SAVE-SAME: "-fcuda-is-device"
 // DEVICE-SAVE-SAME: "-x" "cuda-cpp-output"
 
 // Match the job that produces PTX assembly.
 // DEVICE: "-cc1" "-triple" "nvptx64-nvidia-cuda"
 // DEVICE-NOSAVE-SAME: "-aux-triple" "x86_64--linux-gnu"
 // DEVICE-SAME: "-fcuda-is-device"
-// DEVICE-SM35-SAME: "-target-cpu" "sm_35"
+// DEVICE-SM30-SAME: "-target-cpu" "sm_30"
 // DEVICE-SAME: "-o" "[[PTXFILE:[^"]*]]"
 // DEVICE-NOSAVE-SAME: "-x" "cuda"
 // DEVICE-SAVE-SAME: "-x" "ir"
 
 // Match the call to ptxas (which assembles PTX to SASS).
 // DEVICE:ptxas
-// DEVICE-SM35-DAG: "--gpu-name" "sm_35"
+// DEVICE-SM30-DAG: "--gpu-name" "sm_30"
 // DEVICE-DAG: "--output-file" "[[CUBINFILE:[^"]*]]"
 // DEVICE-DAG: "[[PTXFILE]]"
 
 // Match another device-side compilation.
 // DEVICE2: "-cc1" "-triple" "nvptx64-nvidia-cuda"
 // DEVICE2-SAME: "-aux-triple" "x86_64--linux-gnu"
 // DEVICE2-SAME: "-fcuda-is-device"
-// DEVICE2-SM30-SAME: "-target-cpu" "sm_30"
+// DEVICE2-SM35-SAME: "-target-cpu" "sm_35"
 // DEVICE2-SAME: "-o" "[[GPUBINARY2:[^"]*]]"
 // DEVICE2-SAME: "-x" "cuda"
 
 // Match no device-side compilation.
 // NODEVICE-NOT: "-cc1" "-triple" "nvptx64-nvidia-cuda"
 // NODEVICE-NOT: "-fcuda-is-device"
 
 // INCLUDES-DEVICE:fatbinary