[Openmp-commits] [PATCH] D75788: [WIP][OpenMP] Reuse CUDA wrappers in `nvptx` target regions.

[Artem Belevich via Phabricator via Openmp-commits]

tra added a comment.

Couple of nits below. LGTM for CUDA headers otherwise.



================
Comment at: clang/lib/Headers/__clang_cuda_math_forward_declares.h:42-43
 __DEVICE__ long long abs(long long);
-__DEVICE__ double abs(double);
-__DEVICE__ float abs(float);
-#endif
----------------
Shouldn't float and double abs variants be preserved?


================
Comment at: clang/lib/Headers/cuda_wrappers/new:107
+#pragma omp end declare variant
+#endif
+
----------------
You may want to push/pop `__DEVICE__` macro here and in other headers where it's not done.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D75788/new/

https://reviews.llvm.org/D75788