[Clang] Remove direct linking of offloading runtimes from the arch tools
authorJoseph Huber <jhuber6@vols.utk.edu>
Wed, 17 May 2023 18:20:34 +0000 (13:20 -0500)
committerJoseph Huber <jhuber6@vols.utk.edu>
Wed, 17 May 2023 22:47:49 +0000 (17:47 -0500)
commit66da9ec073ff8dde54b70adcf3b62914769324bf
treedd1d7bc3f5c3085fb3d46ea46259424ba3180171
parent44096e6904e10bb313fef2f6aaff25c25d1325f7
[Clang] Remove direct linking of offloading runtimes from the arch tools

The tools `amdgpu-arch` and `nvptx-arch` are used to query the supported
GPUs on a system to implement features like `--offload-arch=native` as
well as generally being useful for setting up tests. However, we
currently directly link these if they are availible. This patch removes
this because it causes many problems on the user not having the libaries
present or misconfigured at build time. Since these are built
unconditionally we shoudl keep the dependencies away from clang.

Fixes https://github.com/llvm/llvm-project/issues/62784

Reviewed By: ye-luo

Differential Revision: https://reviews.llvm.org/D150807
clang/tools/amdgpu-arch/AMDGPUArch.cpp
clang/tools/amdgpu-arch/CMakeLists.txt
clang/tools/nvptx-arch/CMakeLists.txt
clang/tools/nvptx-arch/NVPTXArch.cpp