jhuber6 wrote: > This breaks CUDA compilation on ARM, because `__has_builtin()` now returns > false for the host-side builtins and that causes some clang headers on ARM to > try defining their own replacement for the builtin they consider to be > missing, but which is actually still there: https://godbolt.org/z/v9qffds1Y > > https://github.com/llvm/llvm-project/blob/285009f202ca8bfcc6b607eba0e919867559e725/clang/lib/Headers/arm_acle.h#L44-L48
Right, the curse of merge compilation strikes again. Feels like damned if you do damned if you don't here. It's either broken inside of the device or outside of the device. It feels rather unfortunate to not have a mechanism to determine if we can actually *generate code* for a builtin. Perhaps we could make some offloading specific hack to work around that, otherwise it's tough to make generic headers that actually work. https://github.com/llvm/llvm-project/pull/121839 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
