Post-commit fix of #164794 reported at https://github.com/llvm/llvm-project/pull/164794#issuecomment-3536253493 `LLVM_LIBRARY_OUTPUT_INTDIR` and `LLVM_RUNTIME_OUTPUT_INTDIR` is used by `AddLLVM.cmake` as output directories. Unless we are in a bootstrapping-build, It must not point to directories found by `find_package(LLVM)` which may be read-only directories. MLIR for instance sets thesese variables to its own build output directory, so should the runtimes.
Summary: This is missing because I forgot to add it.
Summary:
Currently we build the OpenMP device runtime as part of the `offload/`
project. This is problematic because it has several restrictions when
compared to the normal offloading runtime. It can only be built with an
up-to-date clang and we need to set the target appropriately. Currently
we hack around this by creating the compiler invocation manually, but
this patch moves it into a separate runtimes build.
This follows the same build we use for libc, libc++, compiler-rt, and
flang-rt. This also moves it from `offload/` into `openmp/` because it
is still the `openmp/` runtime and I feel it is more appropriate. We do
want a generic `offload/` library at some point, but it would be trivial
to then add that as a separate library now that we have the
infrastructure that makes adding these new libraries trivial.
This most importantly will require that users update their build
configs, mostly adding the following lines at a minimum. I was debating
whether or not I should 'auto-upgrade' this, but I just went with a
warning.
```
-DLLVM_RUNTIME_TARGETS='default;amdgcn-amd-amdhsa;nvptx64-nvidia-cuda' \
-DRUNTIMES_nvptx64-nvidia-cuda_LLVM_ENABLE_RUNTIMES=openmp \
-DRUNTIMES_amdgcn-amd-amdhsa_LLVM_ENABLE_RUNTIMES=openmp \
```
This also changed where the `.bc` version of the library lives, but it's
still created.
Reduce the number of paralell tests run to align with the typical number of VMIDs provided by the kernel driver.
A CMake change included in CMake 4.0 makes `AIX` into a variable
(similar to `APPLE`, etc.)
https://gitlab.kitware.com/cmake/cmake/-/commit/ff03db6657c38c8cf992877ea66174c33d0bcb0b
However, `${CMAKE_SYSTEM_NAME}` unfortunately also expands exactly to
`AIX` and `if` auto-expands variable names in CMake. That means you get
a double expansion if you write:
`if (${CMAKE_SYSTEM_NAME} MATCHES "AIX")`
which becomes:
`if (AIX MATCHES "AIX")`
which is as if you wrote:
`if (ON MATCHES "AIX")`
You can prevent this by quoting the expansion of "${CMAKE_SYSTEM_NAME}",
due to policy
[CMP0054](https://cmake.org/cmake/help/latest/policy/CMP0054.html#policy:CMP0054)
which is on by default in 4.0+. Most of the LLVM CMake already does
this, but this PR fixes the remaining cases where we do not.
Added GPU test job limit to make it consistent with current config https://github.com/llvm/llvm-zorg/blob/main/buildbot/osuosl/master/config/builders.py#L2027C31-L2027C77
An upcoming libc4GPU buildbot will be using this CMake cache file for its build configuration.
PR #134713, which landed as 79cb6f05da37, causes this on my test systems: ``` -- Building AMDGPU plugin for dlopened libhsa -- Not generating AMDGPU tests, no supported devices detected. Use 'LIBOMPTARGET_FORCE_AMDGPU_TESTS' to override. -- Building CUDA plugin for dlopened libcuda -- Not generating NVIDIA tests, no supported devices detected. Use 'LIBOMPTARGET_FORCE_NVIDIA_TESTS' to override. ``` The problem is it cannot locate amdgpu-arch and nvptx-arch. This patch enables it to. I suspect there is more cleanup to do here. amdgpu-arch and nvptx-arch do not appear to exist as cmake targets anymore, but there is still cmake code here that looks for those targets.
Enable the LLVM_ENABLE_RUNTIMES=flang-rt build of the Fortran runtime for the amdgpu-offload-* buildbots. This pre-population cmake cache files is referred to by the llvm-zorg annotated builder factory [script](https://github.com/llvm/llvm-zorg/blob/872f477610d83821c9f1368c969006789b21011b/zorg/buildbot/builders/annotated/amdgpu-offload-cmake.py#L26). The corresponding change in llvm-zorg is llvm/llvm-zorg#402 This reverts commit e296fb8ff6255b97db9ff6cd941acc730164b38f. The worker of amdgpu-offload-rhel-8-cmake-build-only has been updated with a newer version of Ninja that supports Fortran.