Description
Describe the bug
You explicitly removed the ARM build option:
jehammond@brewster:/local/jehammond/DPCPP/llvm$ rm -f /local/home/jehammond/DPCPP/llvm/build/CMakeCache.txt ; python ./buildbot/configure.py --cuda --arm
usage: configure.py [-h] [-n BUILD_NUM] [-b BRANCH] [-d BASE_BRANCH] [-r PR_NUM] [-w BUILDER_DIR] [-s SRC_DIR]
[-o OBJ_DIR] [--l0-headers L0_HEADER_DIR] [--l0-loader L0_LOADER] [-t BUILD_TYPE] [--cuda] [--hip]
[--hip-platform {AMD,NVIDIA}] [--host-target HOST_TARGET] [--enable-esimd-emulator]
[--enable-all-llvm-targets] [--no-assertions] [--docs] [--werror] [--shared-libs]
[--cmake-opt CMAKE_OPT] [--cmake-gen CMAKE_GEN] [--use-libcxx]
[--libcxx-include LIBCXX_INCLUDE_PATH] [--libcxx-library LIBCXX_LIBRARY_PATH] [--use-lld]
[--llvm-external-projects LLVM_EXTERNAL_PROJECTS] [--ci-defaults] [--enable-plugin ENABLE_PLUGIN]
[--disable-fusion] [--add_security_flags {none,default,sanitize}]
configure.py: error: unrecognized arguments: --arm
The supposed replacement described in #6620 does not work:
jehammond@brewster:/local/jehammond/DPCPP/llvm$ rm -f /local/home/jehammond/DPCPP/llvm/build/CMakeCache.txt ; python ./buildbot/configure.py --cuda --host-target=AArch64
args:Namespace(add_security_flags=None, base_branch=None, branch=None, build_number=None, build_type='Release', builder_dir=None, ci_defaults=False, cmake_gen='Ninja', cmake_opt=None, cuda=True, disable_fusion=False, docs=False, enable_all_llvm_targets=False, enable_esimd_emulator=False, enable_plugin=None, hip=False, hip_platform='AMD', host_target='AArch64', l0_headers=None, l0_loader=None, libcxx_include=None, libcxx_library=None, llvm_external_projects=None, no_assertions=False, obj_dir=None, pr_number=None, shared_libs=False, src_dir=None, use_libcxx=False, use_lld=False, werror=False)
[Cmake Command]: cmake -G Ninja -DCMAKE_BUILD_TYPE=Release -DLLVM_ENABLE_ASSERTIONS=ON -DLLVM_TARGETS_TO_BUILD=AArch64;NVPTX -DLLVM_EXTERNAL_PROJECTS=sycl;llvm-spirv;opencl;xpti;xptifw;libdevice;sycl-fusion -DLLVM_EXTERNAL_SYCL_SOURCE_DIR=/local/home/jehammond/DPCPP/llvm/sycl -DLLVM_EXTERNAL_LLVM_SPIRV_SOURCE_DIR=/local/home/jehammond/DPCPP/llvm/llvm-spirv -DLLVM_EXTERNAL_XPTI_SOURCE_DIR=/local/home/jehammond/DPCPP/llvm/xpti -DXPTI_SOURCE_DIR=/local/home/jehammond/DPCPP/llvm/xpti -DLLVM_EXTERNAL_XPTIFW_SOURCE_DIR=/local/home/jehammond/DPCPP/llvm/xptifw -DLLVM_EXTERNAL_LIBDEVICE_SOURCE_DIR=/local/home/jehammond/DPCPP/llvm/libdevice -DLLVM_EXTERNAL_SYCL_FUSION_SOURCE_DIR=/local/home/jehammond/DPCPP/llvm/sycl-fusion -DLLVM_ENABLE_PROJECTS=clang;sycl;llvm-spirv;opencl;xpti;xptifw;libdevice;sycl-fusion;libclc -DLIBCLC_TARGETS_TO_BUILD=;nvptx64--nvidiacl -DLIBCLC_GENERATE_REMANGLED_VARIANTS=ON -DSYCL_BUILD_PI_HIP_PLATFORM=AMD -DLLVM_BUILD_TOOLS=ON -DSYCL_ENABLE_WERROR=OFF -DCMAKE_INSTALL_PREFIX=/local/home/jehammond/DPCPP/llvm/build/install -DSYCL_INCLUDE_TESTS=ON -DLLVM_ENABLE_DOXYGEN=OFF -DLLVM_ENABLE_SPHINX=OFF -DBUILD_SHARED_LIBS=OFF -DSYCL_ENABLE_XPTI_TRACING=ON -DLLVM_ENABLE_LLD=OFF -DXPTI_ENABLE_WERROR=OFF -DSYCL_CLANG_EXTRA_FLAGS= -DSYCL_ENABLE_PLUGINS=level_zero;cuda;opencl -DSYCL_ENABLE_KERNEL_FUSION=ON -DBUG_REPORT_URL=https://github.com/intel/llvm/issues /local/home/jehammond/DPCPP/llvm/llvm
/lib64/ld-linux-x86-64.so.2: No such file or directory
Why is it even necessary to say the host is AArch64 when the host is AArch64? This is not a problem in upstream, is it?
To Reproduce
Please describe the steps to reproduce the behavior:
Clone the repo and run python ./buildbot/configure.py --cuda, potentially with --host-target=AArch64.
Environment (please complete the following information):
-
OS: [e.g Windows/Linux]
Linux -
Target device and vendor: [e.g. Intel GPU]
AArch64 -
DPC++ version: [e.g. commit hash or output of
clang++ --version]
4b321e9 -
Dependencies version: [e.g. low-level runtime versions (like NEO 20.04)]
N/A
Additional context
I personally made this work (#2333, #2322) and am annoyed that people are changing things without properly testing them to see that the changes actually work.