[CUDA][HIP] Support '--offload-arch=native' for the new driver

[jhuber6]

Summary

This patch applies the same handling for the `--offload-arch=native'
string to the new driver. The support for OpenMP will require some extra
logic to infer the triples from the derived architecture strings.

Depends on D141051

[tra]

clang/test/Driver/amdgpu-hip-system-arch.c

17

Nit: I'd reformat it a bit to make it painfully obvious that it's the combination of --offload-arch=native and --amdgpu-arch-tool=%t/amdgpu_arch_fail that makes the test hermetic.
Put them on a RUN line of their own?

[jhuber6]

clang/test/Driver/amdgpu-hip-system-arch.c

17

Will do.

[arsenm]

clang/lib/Driver/Driver.cpp

4286

should explicitly handle the different triple cases, rather than assume ! NVPTX is AMDGPU

4286

why not just take the canonical triple arch name

[jhuber6]

clang/lib/Driver/Driver.cpp

4286

Yeah I should probably just use the triple. I was trying to keep the old tests the same but I should just change them.

[jhuber6]

Change error to use canonical arch string from the triple.

clang/lib/Driver/Driver.cpp

if (Arg->getOption().matches(options::OPT_Xopenmp_target_EQ) &&
       ExtractedArg = getOpts().ParseOneArg(Args, Index);
       Arg = ExtractedArg.get();
     }
 
     // Add or remove the seen architectures in order of appearance. If an
     // invalid architecture is given we simply exit.
     if (Arg->getOption().matches(options::OPT_offload_arch_EQ)) {
       for (StringRef Arch : llvm::split(Arg->getValue(), ",")) {
+        if (Arch == "native") {
+          auto GPUsOrErr = TC->getSystemGPUArchs(Args);
+          if (!GPUsOrErr) {
+            TC->getDriver().Diag(diag::err_drv_undetermined_gpu_arch)
+                << llvm::Triple::getArchTypeName(TC->getArch())
+                << llvm::toString(GPUsOrErr.takeError()) << "--offload-arch";
+            continue;
+          }
+
+          for (auto ArchStr : *GPUsOrErr)
+            Archs.insert(
+                getCanonicalArchString(C, Args, ArchStr, TC->getTriple()));
+        } else {
           StringRef ArchStr =
               getCanonicalArchString(C, Args, Arch, TC->getTriple());
           if (ArchStr.empty())
             return Archs;
           Archs.insert(ArchStr);
         }
+      }
     } else if (Arg->getOption().matches(options::OPT_no_offload_arch_EQ)) {
       for (StringRef Arch : llvm::split(Arg->getValue(), ",")) {
         if (Arch == "all") {
           Archs.clear();
         } else {
           StringRef ArchStr =
               getCanonicalArchString(C, Args, Arch, TC->getTriple());
           if (ArchStr.empty())

clang/test/Driver/amdgpu-hip-system-arch.c

 // REQUIRES: system-linux
 // REQUIRES: x86-registered-target
 // REQUIRES: amdgpu-registered-target
 // REQUIRES: shell
 
 // RUN: mkdir -p %t
 // RUN: cp %S/Inputs/amdgpu-arch/amdgpu_arch_fail %t/
 // RUN: cp %S/Inputs/amdgpu-arch/amdgpu_arch_gfx906 %t/
 // RUN: echo '#!/bin/sh' > %t/amdgpu_arch_empty
 // RUN: chmod +x %t/amdgpu_arch_fail
 // RUN: chmod +x %t/amdgpu_arch_gfx906
 // RUN: chmod +x %t/amdgpu_arch_empty
 
 // case when amdgpu-arch returns nothing or fails
 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-arch=native --amdgpu-arch-tool=%t/amdgpu_arch_fail -x hip %s 2>&1 \
 // RUN:   | FileCheck %s --check-prefix=NO-OUTPUT-ERROR
+// RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-new-driver --offload-arch=native --amdgpu-arch-tool=%t/amdgpu_arch_fail -x hip %s 2>&1 \
+// RUN:   | FileCheck %s --check-prefix=NO-OUTPUT-ERROR
 // NO-OUTPUT-ERROR: error: cannot determine amdgcn architecture{{.*}}; consider passing it via '--offload-arch'
 
 // case when amdgpu-arch does not return anything with successful execution
 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-arch=native --amdgpu-arch-tool=%t/amdgpu_arch_empty -x hip %s 2>&1 \
 // RUN:   | FileCheck %s --check-prefix=EMPTY-OUTPUT
+// RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-new-driver --offload-arch=native --amdgpu-arch-tool=%t/amdgpu_arch_empty -x hip %s 2>&1 \
+// RUN:   | FileCheck %s --check-prefix=EMPTY-OUTPUT
 // EMPTY-OUTPUT: error: cannot determine amdgcn architecture: No AMD GPU detected in the system; consider passing it via '--offload-arch'
 
 // case when amdgpu-arch returns a gfx906 GPU.
 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-arch=native --amdgpu-arch-tool=%t/amdgpu_arch_gfx906 -x hip %s 2>&1 \
 // RUN:   | FileCheck %s --check-prefix=ARCH-GFX906
+// RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-new-driver --offload-arch=native --amdgpu-arch-tool=%t/amdgpu_arch_gfx906 -x hip %s 2>&1 \
+// RUN:   | FileCheck %s --check-prefix=ARCH-GFX906
 // ARCH-GFX906: "-cc1" "-triple" "amdgcn-amd-amdhsa"{{.*}}"-target-cpu" "gfx906"

clang/test/Driver/nvptx-cuda-system-arch.c

 // REQUIRES: system-linux
 // REQUIRES: x86-registered-target
 // REQUIRES: nvptx-registered-target
 // REQUIRES: shell
 
 // RUN: mkdir -p %t
 // RUN: cp %S/Inputs/nvptx-arch/nvptx_arch_fail %t/
 // RUN: cp %S/Inputs/nvptx-arch/nvptx_arch_sm_70 %t/
 // RUN: echo '#!/bin/sh' > %t/nvptx_arch_empty
 // RUN: chmod +x %t/nvptx_arch_fail
 // RUN: chmod +x %t/nvptx_arch_sm_70
 // RUN: chmod +x %t/nvptx_arch_empty
 
 // case when nvptx-arch returns nothing or fails
 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-arch=native --nvptx-arch-tool=%t/nvptx_arch_fail -x cuda %s 2>&1 \
 // RUN:   | FileCheck %s --check-prefix=NO-OUTPUT-ERROR
+// RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-new-driver --offload-arch=native --nvptx-arch-tool=%t/nvptx_arch_fail -x cuda %s 2>&1 \
+// RUN:   | FileCheck %s --check-prefix=NO-OUTPUT-ERROR
 // NO-OUTPUT-ERROR: error: cannot determine nvptx64 architecture{{.*}}; consider passing it via '--offload-arch'
 
 // case when nvptx-arch does not return anything with successful execution
 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-arch=native --nvptx-arch-tool=%t/nvptx_arch_empty -x cuda %s 2>&1 \
 // RUN:   | FileCheck %s --check-prefix=EMPTY-OUTPUT
+// RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-new-driver --offload-arch=native --nvptx-arch-tool=%t/nvptx_arch_empty -x cuda %s 2>&1 \
+// RUN:   | FileCheck %s --check-prefix=EMPTY-OUTPUT
 // EMPTY-OUTPUT: error: cannot determine nvptx64 architecture: No NVIDIA GPU detected in the system; consider passing it via '--offload-arch'
 
 // case when nvptx-arch does not return anything with successful execution
 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-arch=native --nvptx-arch-tool=%t/nvptx_arch_sm_70 -x cuda %s 2>&1 \
 // RUN:   | FileCheck %s --check-prefix=ARCH-sm_70
+// RUN:   %clang -### --target=x86_64-unknown-linux-gnu -nogpulib --offload-arch=native --offload-new-driver --nvptx-arch-tool=%t/nvptx_arch_sm_70 -x cuda %s 2>&1 \
+// RUN:   | FileCheck %s --check-prefix=ARCH-sm_70
 // ARCH-sm_70: "-cc1" "-triple" "nvptx64-nvidia-cuda"{{.*}}"-target-cpu" "sm_70"

clang/test/Driver/openmp-offload-infer.c

 // CHECK-ARCH-BINDINGS: "nvptx64-nvidia-cuda" - "clang", inputs: ["[[INPUT]]", "[[HOST_BC]]"], output: "[[DEVICE_BC_SM_70:.*]]"
 // CHECK-ARCH-BINDINGS: "nvptx64-nvidia-cuda" - "NVPTX::Assembler", inputs: ["[[DEVICE_BC_SM_70]]"], output: "[[DEVICE_OBJ_SM_70:.*]]"
 // CHECK-ARCH-BINDINGS: "x86_64-unknown-linux-gnu" - "Offload::Packager", inputs: ["[[DEVICE_OBJ_SM_52]]", "[[DEVICE_OBJ_SM_70]]"], output: "[[BINARY:.+]]"
 // CHECK-ARCH-BINDINGS: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[HOST_BC]]", "[[BINARY]]"], output: "[[HOST_OBJ:.*]]"
 // CHECK-ARCH-BINDINGS: "x86_64-unknown-linux-gnu" - "Offload::Linker", inputs: ["[[HOST_OBJ]]"], output: "a.out"
 
 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp \
 // RUN:     --offload-arch=sm_70 --offload-arch=gfx908 --offload-arch=native \
+// RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp \
+// RUN:     --offload-arch=sm_70 --offload-arch=gfx908 --offload-arch=skylake \
 // RUN:     -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-FAILED
 
-// CHECK-FAILED: error: failed to deduce triple for target architecture 'native'; specify the triple using '-fopenmp-targets' and '-Xopenmp-target' instead.
+// CHECK-FAILED: error: failed to deduce triple for target architecture 'skylake'; specify the triple using '-fopenmp-targets' and '-Xopenmp-target' instead.
 
 // RUN:   %clang -### --target=x86_64-unknown-linux-gnu -ccc-print-bindings -fopenmp=libomp \
 // RUN:     --offload-arch=sm_70 --offload-arch=gfx908 -fno-openmp \
 // RUN:     -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-DISABLED
 
 // CHECK-DISABLED-NOT: "nvptx64-nvidia-cuda" - "clang",