[CUDA] Add --cuda-compile-host-device, which overrides --cuda-host-only and --cuda-device-only.

[jlebar]

Summary

This completes the flag's tristate, letting you override it at will on
the command line.

[tra]

Help strings seem to be backwards.
LGTM otherwise.

include/clang/Driver/Options.td

378 ↗

(On Diff #54143)

for *device*

381 ↗

(On Diff #54143)

for *host*

[jlebar]

Fix help text.

[jlebar]

Wow, oops. Thank you.

cfe/trunk/include/clang/Driver/Options.td

 def cpp_precomp : Flag<["-"], "cpp-precomp">, Group<clang_ignored_f_Group>;
 def current__version : JoinedOrSeparate<["-"], "current_version">;
 def cxx_isystem : JoinedOrSeparate<["-"], "cxx-isystem">, Group<clang_i_Group>,
   HelpText<"Add directory to the C++ SYSTEM include search path">, Flags<[CC1Option]>,
   MetaVarName<"<directory>">;
 def c : Flag<["-"], "c">, Flags<[DriverOption]>,
   HelpText<"Only run preprocess, compile, and assemble steps">;
 def cuda_device_only : Flag<["--"], "cuda-device-only">,
-  HelpText<"Do device-side CUDA compilation only">;
+  HelpText<"Compile CUDA code for device only">;
+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 cuda_host_only : Flag<["--"], "cuda-host-only">,
-  HelpText<"Do host-side CUDA compilation only">;
 def cuda_noopt_device_debug : Flag<["--"], "cuda-noopt-device-debug">,
   HelpText<"Enable device-side debug info generation. Disables ptxas optimizations.">;
 def cuda_path_EQ : Joined<["--"], "cuda-path=">, Group<i_Group>,
   HelpText<"CUDA installation path">;
 def fcuda_flush_denormals_to_zero : Flag<["-"], "fcuda-flush-denormals-to-zero">,
   Flags<[CC1Option]>, HelpText<"Flush denormal floating point values to zero in CUDA device mode.">;
 def fno_cuda_flush_denormals_to_zero : Flag<["-"], "fno-cuda-flush-denormals-to-zero">;
 def dA : Flag<["-"], "dA">, Group<d_Group>;

cfe/trunk/lib/Driver/Driver.cpp

 // appropriate GPU arch name. In case of partial (i.e preprocessing
 // only) or device-only compilation, each device action is added to /p
 // Actions and /p Current is released. Otherwise the function creates
 // and returns a new CudaHostAction which wraps /p Current and device
 // side actions.
 static Action *buildCudaActions(Compilation &C, DerivedArgList &Args,
                                 const Arg *InputArg, Action *HostAction,
                                 ActionList &Actions) {
-  Arg *PartialCompilationArg = Args.getLastArg(options::OPT_cuda_host_only,
-                                               options::OPT_cuda_device_only);
-  // Host-only compilation case.
-  if (PartialCompilationArg &&
-      PartialCompilationArg->getOption().matches(options::OPT_cuda_host_only))
+  Arg *PartialCompilationArg = Args.getLastArg(
+      options::OPT_cuda_host_only, options::OPT_cuda_device_only,
+      options::OPT_cuda_compile_host_device);
+  bool CompileHostOnly =
+      PartialCompilationArg &&
+      PartialCompilationArg->getOption().matches(options::OPT_cuda_host_only);
+  bool CompileDeviceOnly =
+      PartialCompilationArg &&
+      PartialCompilationArg->getOption().matches(options::OPT_cuda_device_only);
+
+  if (CompileHostOnly)
     return C.MakeAction<CudaHostAction>(HostAction, ActionList());
 
   // Collect all cuda_gpu_arch parameters, removing duplicates.
   SmallVector<const char *, 4> GpuArchList;
   llvm::StringSet<> GpuArchNames;
   for (Arg *A : Args) {
     if (!A->getOption().matches(options::OPT_cuda_gpu_arch_EQ))
       continue;
static Action *buildCudaActions(Compilation &C, DerivedArgList &Args,
   // Check whether any of device actions stopped before they could generate PTX.
   bool PartialCompilation =
       llvm::any_of(CudaDeviceActions, [](const Action *a) {
         return a->getKind() != Action::AssembleJobClass;
       });
 
   // Figure out what to do with device actions -- pass them as inputs to the
   // host action or run each of them independently.
-  bool DeviceOnlyCompilation = PartialCompilationArg != nullptr;
-  if (PartialCompilation || DeviceOnlyCompilation) {
+  if (PartialCompilation || CompileDeviceOnly) {
     // In case of partial or device-only compilation results of device actions
     // are not consumed by the host action device actions have to be added to
     // top-level actions list with AtTopLevel=true and run independently.
 
     // -o is ambiguous if we have more than one top-level action.
     if (Args.hasArg(options::OPT_o) &&
-        (!DeviceOnlyCompilation || GpuArchList.size() > 1)) {
+        (!CompileDeviceOnly || GpuArchList.size() > 1)) {
       C.getDriver().Diag(
           clang::diag::err_drv_output_argument_with_multiple_files);
       return nullptr;
     }
 
     for (unsigned I = 0, E = GpuArchList.size(); I != E; ++I)
       Actions.push_back(C.MakeAction<CudaDeviceAction>(CudaDeviceActions[I],
                                                        GpuArchList[I],
                                                        /* AtTopLevel */ true));
     // Kill host action in case of device-only compilation.
-    if (DeviceOnlyCompilation)
+    if (CompileDeviceOnly)
       return nullptr;
     return HostAction;
   }
 
   // If we're not a partial or device-only compilation, we compile each arch to
   // ptx and assemble to cubin, then feed the cubin *and* the ptx into a device
   // "link" action, which uses fatbinary to combine these cubins into one
   // fatbin.  The fatbin is then an input to the host compilation.
void Driver::BuildActions(Compilation &C, DerivedArgList &Args,
   if (FinalPhase == phases::Link && PL.size() == 1) {
     Args.ClaimAllArgs(options::OPT_CompileOnly_Group);
     Args.ClaimAllArgs(options::OPT_cl_compile_Group);
   }
 
   // Claim ignored clang-cl options.
   Args.ClaimAllArgs(options::OPT_cl_ignored_Group);
 
-  // Claim --cuda-host-only arg which may be passed to non-CUDA
-  // compilations and should not trigger warnings there.
+  // Claim --cuda-host-only and --cuda-compile-host-device, which may be passed
+  // to non-CUDA compilations and should not trigger warnings there.
   Args.ClaimAllArgs(options::OPT_cuda_host_only);
+  Args.ClaimAllArgs(options::OPT_cuda_compile_host_device);
 }
 
 Action *Driver::ConstructPhaseAction(Compilation &C, const ArgList &Args,
                                      phases::ID Phase, Action *Input) const {
   llvm::PrettyStackTraceString CrashInfo("Constructing phase actions");
   // Build the appropriate action.
   switch (Phase) {
   case phases::Link:

cfe/trunk/test/Driver/cuda-options.cu

 // RUN:    -check-prefix LINK %s
 
 // Verify that --cuda-host-only disables device-side compilation, but doesn't
 // disable host-side compilation/linking.
 // RUN: %clang -### -target x86_64-linux-gnu --cuda-host-only %s 2>&1 \
 // RUN: | FileCheck -check-prefix NODEVICE -check-prefix HOST \
 // RUN:    -check-prefix NOINCLUDES-DEVICE -check-prefix LINK %s
 
-// Same test as above, but with preceeding --cuda-device-only to make sure only
-// the last option has an effect.
-// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only --cuda-host-only %s 2>&1 \
+// Verify that --cuda-device-only disables host-side compilation and linking.
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only %s 2>&1 \
+// RUN: | FileCheck -check-prefix DEVICE -check-prefix DEVICE-NOSAVE \
+// RUN:    -check-prefix NOHOST -check-prefix NOLINK %s
+
+// Check that the last of --cuda-compile-host-device, --cuda-host-only, and
+// --cuda-device-only wins.
+
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only \
+// RUN:    --cuda-host-only %s 2>&1 \
 // RUN: | FileCheck -check-prefix NODEVICE -check-prefix HOST \
 // RUN:    -check-prefix NOINCLUDES-DEVICE -check-prefix LINK %s
 
-// Verify that --cuda-device-only disables host-side compilation and linking.
-// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only %s 2>&1 \
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-compile-host-device \
+// RUN:    --cuda-host-only %s 2>&1 \
+// RUN: | FileCheck -check-prefix NODEVICE -check-prefix HOST \
+// RUN:    -check-prefix NOINCLUDES-DEVICE -check-prefix LINK %s
+
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-host-only \
+// RUN:    --cuda-device-only %s 2>&1 \
 // RUN: | FileCheck -check-prefix DEVICE -check-prefix DEVICE-NOSAVE \
 // RUN:    -check-prefix NOHOST -check-prefix NOLINK %s
 
-// Same test as above, but with preceeding --cuda-host-only to make sure only
-// the last option has an effect.
-// RUN: %clang -### -target x86_64-linux-gnu --cuda-host-only --cuda-device-only %s 2>&1 \
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-compile-host-device \
+// RUN:    --cuda-device-only %s 2>&1 \
 // RUN: | FileCheck -check-prefix DEVICE -check-prefix DEVICE-NOSAVE \
 // RUN:    -check-prefix NOHOST -check-prefix NOLINK %s
 
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-host-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
+
+// 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: | FileCheck -check-prefix DEVICE -check-prefix DEVICE-NOSAVE \
 // RUN:    -check-prefix DEVICE-SM35 -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

cfe/trunk/test/Driver/cuda-unused-arg-warning.cu

 // Tests that we trigger unused-arg warnings on CUDA flags appropriately.
 
 // REQUIRES: clang-driver
 // REQUIRES: x86-registered-target
 // REQUIRES: nvptx-registered-target
 
-// --cuda-host-only should never trigger unused arg warning.
+// --cuda-host-only and --cuda-compile-host-device should never trigger an
+// unused arg warning.
 // RUN: %clang -### -target x86_64-linux-gnu --cuda-host-only -c %s 2>&1 | \
 // RUN:    FileCheck %s
 // RUN: %clang -### -target x86_64-linux-gnu --cuda-host-only -x c -c %s 2>&1 | \
 // RUN:    FileCheck %s
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-compile-host-device -c %s 2>&1 | \
+// RUN:    FileCheck %s
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-compile-host-device -x c -c %s 2>&1 | \
+// RUN:    FileCheck %s
 
 // --cuda-device-only should warn during non-CUDA compilation.
 // RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only -x c -c %s 2>&1 | \
 // RUN:    FileCheck -check-prefix UNUSED-WARNING %s
 
 // --cuda-device-only should not produce warning compiling CUDA files
 // RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only -c %s 2>&1 | \
 // RUN:    FileCheck -check-prefix NO-UNUSED-WARNING %s
 
 // CHECK-NOT: warning: argument unused during compilation: '--cuda-host-only'
+// CHECK-NOT: warning: argument unused during compilation: '--cuda-compile-host-device'
 // UNUSED-WARNING: warning: argument unused during compilation: '--cuda-device-only'
 // NO-UNUSED-WARNING-NOT: warning: argument unused during compilation: '--cuda-device-only'