Created attachment 680578 [details] build.log -- Performing Test HAVE_STD_IS_TRIVIALLY_COPYABLE -- Performing Test HAVE_STD_IS_TRIVIALLY_COPYABLE - Failed -- Performing Test HAVE_CXX_ATOMICS_WITHOUT_LIB -- Performing Test HAVE_CXX_ATOMICS_WITHOUT_LIB - Failed -- Looking for __atomic_fetch_add_4 in atomic -- Looking for __atomic_fetch_add_4 in atomic - not found CMake Error at cmake/modules/CheckAtomic.cmake:58 (message): Host compiler appears to require libatomic, but cannot find it. Call Stack (most recent call first): cmake/config-ix.cmake:350 (include) CMakeLists.txt:655 (include) -- Configuring incomplete, errors occurred! See also "/Users/sam/prefix/var/tmp/portage/sys-devel/llvm-11.0.0/work/llvm-11.0.0_build-abi_x86_64.amd64/CMakeFiles/CMakeOutput.log". See also "/Users/sam/prefix/var/tmp/portage/sys-devel/llvm-11.0.0/work/llvm-11.0.0_build-abi_x86_64.amd64/CMakeFiles/CMakeError.log". * ERROR: sys-devel/llvm-11.0.0::gentoo_prefix failed (configure phase): * cmake failed
Created attachment 680581 [details] CMakeOutput.log
Created attachment 680584 [details] CMakeError.log
Oh, Fabian already noted this: https://bugs.gentoo.org/758167#c51.
Yes, but a bug is actually good, since this is x86_64 and not arm64. So, my thinking was, one needs compiler_rt. Next is, that compiler_rt needs to be explicitly enabled (with cflags) or it needs built into libcxx/-abi. The latter makes more sense, else we'd have to patch a lot of packages. So I tried doing that but it seems clang cannot find compiler_rt. Reason for that is twofold: it seems Clang (also Apple's host Clang) is configured with -print-path to point to a different directory (darwin20.20.0 iso darwin), but perhaps this is just a decoy since it's some gcc emulation code. Second, it appears to look for compiler_rt.builtins.<arch>, which I don't have. I only have osx.a or something.
