r314334 - [CUDA] Work around conflicting function definitions in CUDA-9 headers.

Artem Belevich via cfe-commits cfe-commits at lists.llvm.org
Wed Sep 27 12:07:16 PDT 2017


Author: tra
Date: Wed Sep 27 12:07:15 2017
New Revision: 314334

URL: http://llvm.org/viewvc/llvm-project?rev=314334&view=rev
Log:
[CUDA] Work around conflicting function definitions in CUDA-9 headers.

Differential Revision: https://reviews.llvm.org/D38326

Modified:
    cfe/trunk/lib/Headers/__clang_cuda_runtime_wrapper.h

Modified: cfe/trunk/lib/Headers/__clang_cuda_runtime_wrapper.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/__clang_cuda_runtime_wrapper.h?rev=314334&r1=314333&r2=314334&view=diff
==============================================================================
--- cfe/trunk/lib/Headers/__clang_cuda_runtime_wrapper.h (original)
+++ cfe/trunk/lib/Headers/__clang_cuda_runtime_wrapper.h Wed Sep 27 12:07:15 2017
@@ -173,7 +173,18 @@ inline __host__ double __signbitd(double
 // __device__.
 #pragma push_macro("__forceinline__")
 #define __forceinline__ __device__ __inline__ __attribute__((always_inline))
+
+#pragma push_macro("__float2half_rn")
+#if CUDA_VERSION >= 9000
+// CUDA-9 has conflicting prototypes for __float2half_rn(float f) in
+// cuda_fp16.h[pp] and device_functions.hpp. We need to get the one in
+// device_functions.hpp out of the way.
+#define __float2half_rn  __float2half_rn_disabled
+#endif
+
 #include "device_functions.hpp"
+#pragma pop_macro("__float2half_rn")
+
 
 // math_function.hpp uses the __USE_FAST_MATH__ macro to determine whether we
 // get the slow-but-accurate or fast-but-inaccurate versions of functions like




More information about the cfe-commits mailing list