[OpenMP] Expand short verisions of OpenMP offloading triples

[jhuber6]

Summary

The OpenMP offloading libraries are built with fixed triples and linked
in during compile time. This would cause un-helpful errors if the user
passed in the wrong expansion of the triple used for the bitcode
library. because we only support these triples for OpenMP offloading we
can normalize them to the full verion used in the bitcode library.

[jdoerfert]

Can we have a test? Similar to a/the test introduced with the libdevice handling.

[jhuber6]

Adding tests.

[jdoerfert]

LG

[JonChesterfield]

Wrong fix. The user typing -fopenmp-targets=amdgcn is an error, we shouldn't swallow errors here and hope the backend does something sane with the result. We should rewrite that in clang to the right triple.

This may 'work', if the correct triple is the one which wins the conflict, provided nothing triple-dependent has already happened in the IR, and we never target another OS, but that's much less robust than setting the HSA triple directly.

[arsenm]

llvm/lib/Linker/IRMover.cpp

1482 ↗

(On Diff #401252)

I'm a bit disturbed the linker is special casing bitcode file names

[jhuber6]

Changing approach to simply expand the triple where we parse it for OpenMP.

[JonChesterfield]

LG, thanks!

clang/lib/Driver/Driver.cpp

778

Might be worth mentioning in the comment that this is essentially a user convenience feature. We're not so much normalising, as treating fopenmp-target=nvtpx64 as an abbreviation for nvptx64-nvidia-cuda

clang/lib/Driver/Driver.cpp

if (OpenMPTargets->getNumValues()) {
       }
 
       if (HasValidOpenMPRuntime) {
         llvm::StringMap<const char *> FoundNormalizedTriples;
         for (const char *Val : OpenMPTargets->getValues()) {
           llvm::Triple TT(Val);
           std::string NormalizedName = TT.normalize();
 
+          // We want to expand the shortened versions of the triples passed in to
+          // the values used for the bitcode libraries for convenience.
+          if (TT.getVendor() == llvm::Triple::UnknownVendor ||
+              TT.getOS() == llvm::Triple::UnknownOS) {
+            if (TT.getArch() == llvm::Triple::nvptx)
+              TT = llvm::Triple("nvptx-nvidia-cuda");
+            else if (TT.getArch() == llvm::Triple::nvptx64)
+              TT = llvm::Triple("nvptx64-nvidia-cuda");
+            else if (TT.getArch() == llvm::Triple::amdgcn)
+              TT = llvm::Triple("amdgcn-amd-amdhsa");
+          }
+
           // Make sure we don't have a duplicate triple.
           auto Duplicate = FoundNormalizedTriples.find(NormalizedName);
           if (Duplicate != FoundNormalizedTriples.end()) {
             Diag(clang::diag::warn_drv_omp_offload_target_duplicate)
                 << Val << Duplicate->second;
             continue;
           }
 

clang/test/Driver/fat_archive_nvptx.cpp

 // REQUIRES: clang-driver
 // REQUIRES: x86-registered-target
 // REQUIRES: nvptx-registered-target
 
 // See the steps to create a fat archive are given at the end of the file.
 
 // Given a FatArchive, clang-offload-bundler should be called to create a
 // device specific archive, which should be passed to clang-nvlink-wrapper.
-// RUN: %clang -O2 -### -fopenmp -fopenmp-targets=nvptx64 %s -L%S/Inputs/openmp_static_device_link -lFatArchive 2>&1 | FileCheck %s
-// CHECK: clang{{.*}}"-cc1"{{.*}}"-triple" "nvptx64"{{.*}}"-target-cpu" "[[GPU:sm_[0-9]+]]"{{.*}}"-o" "[[HOSTBC:.*.s]]" "-x" "c++"{{.*}}.cpp
-// CHECK: clang-offload-bundler" "-unbundle" "-type=a" "-inputs={{.*}}/Inputs/openmp_static_device_link/libFatArchive.a" "-targets=openmp-nvptx64-[[GPU]]" "-outputs=[[DEVICESPECIFICARCHIVE:.*.a]]" "-allow-missing-bundles"
+// RUN: %clang -O2 -### -fopenmp -fopenmp-targets=nvptx64-nvidia-cuda %s -L%S/Inputs/openmp_static_device_link -lFatArchive 2>&1 | FileCheck %s
+// CHECK: clang{{.*}}"-cc1"{{.*}}"-triple" "nvptx64-nvidia-cuda"{{.*}}"-target-cpu" "[[GPU:sm_[0-9]+]]"{{.*}}"-o" "[[HOSTBC:.*.s]]" "-x" "c++"{{.*}}.cpp
+// CHECK: clang-offload-bundler" "-unbundle" "-type=a" "-inputs={{.*}}/Inputs/openmp_static_device_link/libFatArchive.a" "-targets=openmp-nvptx64-nvidia-cuda-[[GPU]]" "-outputs=[[DEVICESPECIFICARCHIVE:.*.a]]" "-allow-missing-bundles"
 // CHECK: clang-nvlink-wrapper{{.*}}"-o" "{{.*}}.out" "-arch" "[[GPU]]" "{{.*}}[[DEVICESPECIFICARCHIVE]]"
 // expected-no-diagnostics
 
 #ifndef HEADER
 #define HEADER
 
 #define N 10
 
void func_present(float* in, float* out, unsigned n){
   }
 }
 *************************************************
 1. Compile source file(s) to generate object file(s)
     clang -O2 -fopenmp -fopenmp-targets=amdgcn-amd-amdhsa -Xopenmp-target=amdgcn-amd-amdhsa -march=gfx906 -c func_1.c -o func_1_gfx906.o
     clang -O2 -fopenmp -fopenmp-targets=amdgcn-amd-amdhsa -Xopenmp-target=amdgcn-amd-amdhsa -march=gfx908 -c func_1.c -o func_1_gfx908.o
     clang -O2 -fopenmp -fopenmp-targets=amdgcn-amd-amdhsa -Xopenmp-target=amdgcn-amd-amdhsa -march=gfx906 -c func_2.c -o func_2_gfx906.o
     clang -O2 -fopenmp -fopenmp-targets=amdgcn-amd-amdhsa -Xopenmp-target=amdgcn-amd-amdhsa -march=gfx908 -c func_2.c -o func_2_gfx908.o
-    clang -O2 -fopenmp -fopenmp-targets=nvptx64 -c func_1.c -o func_1_nvptx.o
-    clang -O2 -fopenmp -fopenmp-targets=nvptx64 -c func_2.c -o func_2_nvptx.o
+    clang -O2 -fopenmp -fopenmp-targets=nvptx64-nvidia-cuda -c func_1.c -o func_1_nvptx.o
+    clang -O2 -fopenmp -fopenmp-targets=nvptx64-nvidia-cuda -c func_2.c -o func_2_nvptx.o
 
 2. Create a fat archive by combining all the object file(s)
     llvm-ar cr libFatArchive.a func_1_gfx906.o func_1_gfx908.o func_2_gfx906.o func_2_gfx908.o func_1_nvptx.o func_2_nvptx.o
 ************************************************/

openmp/libomptarget/DeviceRTL/CMakeLists.txt

add_custom_command(TARGET ${bclib_target_name} POST_BUILD
                     ${LIBOMPTARGET_LIBRARY_DIR})
 
   # Install bitcode library under the lib destination folder.
   install(FILES ${CMAKE_CURRENT_BINARY_DIR}/${bclib_name} DESTINATION "${OPENMP_INSTALL_LIBDIR}")
 endfunction()
 
 # Generate a Bitcode library for all the compute capabilities the user requested
 foreach(sm ${nvptx_sm_list})
-  compileDeviceRTLLibrary(sm_${sm} nvptx -target nvptx64 -Xclang -target-feature -Xclang +ptx61 "-D__CUDA_ARCH__=${sm}0")
+  compileDeviceRTLLibrary(sm_${sm} nvptx -target nvptx64-nvidia-cuda -Xclang -target-feature -Xclang +ptx61 "-D__CUDA_ARCH__=${sm}0")
 endforeach()
 
 foreach(mcpu ${amdgpu_mcpus})
   compileDeviceRTLLibrary(${mcpu} amdgpu -target amdgcn-amd-amdhsa -D__AMDGCN__ -fvisibility=default -nogpulib)
 endforeach()