[CUDA] Minor cuda-related driver fixes.

[tra]

Summary

• Only the last of the --cuda-host-only/--cuda-device-only options has effect.
• Added test cases.
• Fixed printing of empty action lists for -ccc-print-phases

[echristo]

Yuck. No better ideas off the top of my head other than diagnosing and erroring though.

(Is that a possibility? :)

-eric

[tra]

In D12892#246699, @echristo wrote:

Yuck. No better ideas off the top of my head other than diagnosing and erroring though.

(Is that a possibility? :)

I need to be able to override preceding options, so no error is intentional.

[echristo]

Sure. Seems reasonable.

Thanks!

-eric

cfe/trunk/lib/Driver/Driver.cpp

if (InputAction *IA = dyn_cast<InputAction>(A)) {
     ActionList *AL;
     if (CudaHostAction *CHA = dyn_cast<CudaHostAction>(A)) {
       os << "{" << PrintActions1(C, *CHA->begin(), Ids) << "}"
          << ", gpu binaries ";
       AL = &CHA->getDeviceActions();
     } else
       AL = &A->getInputs();
 
+    if (AL->size()) {
       const char *Prefix = "{";
       for (Action *PreRequisite : *AL) {
         os << Prefix << PrintActions1(C, PreRequisite, Ids);
         Prefix = ", ";
       }
       os << "}";
+    } else
+      os << "{}";
   }
 
   unsigned Id = Ids.size();
   Ids[A] = Id;
   llvm::errs() << Id << ": " << os.str() << ", "
                << types::getTypeName(A->getType()) << "\n";
 
   return Id;

 // 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 std::unique_ptr<Action>
 buildCudaActions(const Driver &D, const ToolChain &TC, DerivedArgList &Args,
                  const Arg *InputArg, std::unique_ptr<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))
+    return std::unique_ptr<Action>(
+        new CudaHostAction(std::move(HostAction), {}));
 
   // 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)) {
       A->claim();
       if (GpuArchNames.insert(A->getValue()).second)
buildCudaActions(const Driver &D, const ToolChain &TC, DerivedArgList &Args,
   // Figure out which NVPTX triple to use for device-side compilation based on
   // whether host is 64-bit.
   const char *DeviceTriple = TC.getTriple().isArch64Bit()
                                  ? "nvptx64-nvidia-cuda"
                                  : "nvptx-nvidia-cuda";
 
   // Figure out what to do with device actions -- pass them as inputs to the
   // host action or run each of them independently.
-  bool DeviceOnlyCompilation = Args.hasArg(options::OPT_cuda_device_only);
+  bool DeviceOnlyCompilation = PartialCompilationArg != nullptr;
   if (PartialCompilation || DeviceOnlyCompilation) {
     // 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)) {
if (InitialPhase > FinalPhase) {
       else
         Diag(clang::diag::warn_drv_input_file_unused)
             << InputArg->getAsString(Args) << getPhaseName(InitialPhase)
             << !!FinalPhaseArg
             << (FinalPhaseArg ? FinalPhaseArg->getOption().getName() : "");
       continue;
     }
 
-    phases::ID CudaInjectionPhase;
-    bool InjectCuda = (InputType == types::TY_CUDA &&
-                       !Args.hasArg(options::OPT_cuda_host_only));
-    CudaInjectionPhase = FinalPhase;
-    for (auto &Phase : PL)
+    phases::ID CudaInjectionPhase = FinalPhase;
+    for (const auto &Phase : PL)
       if (Phase <= FinalPhase && Phase == phases::Compile) {
         CudaInjectionPhase = Phase;
         break;
       }
 
     // Build the pipeline for this file.
     std::unique_ptr<Action> Current(new InputAction(*InputArg, InputType));
     for (SmallVectorImpl<phases::ID>::iterator i = PL.begin(), e = PL.end();
for (SmallVectorImpl<phases::ID>::iterator i = PL.begin(), e = PL.end();
       // encode this in the steps because the intermediate type depends on
       // arguments. Just special case here.
       if (Phase == phases::Assemble && Current->getType() != types::TY_PP_Asm)
         continue;
 
       // Otherwise construct the appropriate action.
       Current = ConstructPhaseAction(TC, Args, Phase, std::move(Current));
 
-      if (InjectCuda && Phase == CudaInjectionPhase) {
+      if (InputType == types::TY_CUDA && Phase == CudaInjectionPhase) {
         Current = buildCudaActions(*this, TC, Args, InputArg,
                                    std::move(Current), Actions);
         if (!Current)
           break;
       }
 
       if (Current->getType() == types::TY_Nothing)
         break;

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

 // RUN: %clang -### -target x86_64-linux-gnu %s 2>&1 \
 // Compile device-side to PTX assembly and make sure we use it on the host side
 // RUN:   | FileCheck -check-prefix CUDA-D1 -check-prefix CUDA-D1NS\
 // Then compile host side and incorporate device code.
 // RUN:   -check-prefix CUDA-H -check-prefix CUDA-H-I1 \
 // Then link things.
 // RUN:   -check-prefix CUDA-L %s
 
-// Verify that -cuda-no-device disables device-side compilation and linking
+// Verify that --cuda-host-only disables device-side compilation and linking
 // RUN: %clang -### -target x86_64-linux-gnu --cuda-host-only %s 2>&1 \
 // Make sure we didn't run device-side compilation.
 // RUN:   | FileCheck -check-prefix CUDA-ND \
 // Then compile host side and make sure we don't attempt to incorporate GPU code.
 // RUN:    -check-prefix CUDA-H -check-prefix CUDA-H-NI \
 // Linking is allowed to happen, even if we're missing GPU code.
 // RUN:    -check-prefix CUDA-L %s
 
-// Verify that -cuda-no-host disables host-side compilation and linking
+// Same test as above, but with preceeding --cuda-device-only to make
+// sure only last option has effect.
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-device-only --cuda-host-only %s 2>&1 \
+// Make sure we didn't run device-side compilation.
+// RUN:   | FileCheck -check-prefix CUDA-ND \
+// Then compile host side and make sure we don't attempt to incorporate GPU code.
+// RUN:    -check-prefix CUDA-H -check-prefix CUDA-H-NI \
+// Linking is allowed to happen, even if we're missing GPU code.
+// RUN:    -check-prefix CUDA-L %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 \
 // Compile device-side to PTX assembly
 // RUN:   | FileCheck -check-prefix CUDA-D1 -check-prefix CUDA-D1NS\
 // Make sure there are no host cmpilation or linking.
 // RUN:   -check-prefix CUDA-NH -check-prefix CUDA-NL %s
 
+// Same test as above, but with preceeding --cuda-host-only to make
+// sure only last option has effect.
+// RUN: %clang -### -target x86_64-linux-gnu --cuda-host-only --cuda-device-only %s 2>&1 \
+// Compile device-side to PTX assembly
+// RUN:   | FileCheck -check-prefix CUDA-D1 -check-prefix CUDA-D1NS\
+// Make sure there are no host cmpilation or linking.
+// RUN:   -check-prefix CUDA-NH -check-prefix CUDA-NL %s
+
 // Verify that with -S we compile host and device sides to assembly
 // and incorporate device code on the host side.
 // RUN: %clang -### -target x86_64-linux-gnu -S -c %s 2>&1 \
 // Compile device-side to PTX assembly
 // RUN:   | FileCheck -check-prefix CUDA-D1 -check-prefix CUDA-D1NS\
 // Then compile host side and incorporate GPU code.
 // RUN:  -check-prefix CUDA-H -check-prefix CUDA-H-I1 \
 // Make sure we don't link anything.