add cuMemcpy

Signed-off-by: leozhang <nowerzt@gmail.com>

include/ffnvcodec/dynlink_cuda.h

@@ -365,6 +365,8 @@ typedef CUresult CUDAAPI tcuMemAlloc_v2(CUdeviceptr *dptr, size_t bytesize);
 typedef CUresult CUDAAPI tcuMemAllocPitch_v2(CUdeviceptr *dptr, size_t *pPitch, size_t WidthInBytes, size_t Height, unsigned int ElementSizeBytes);
 typedef CUresult CUDAAPI tcuMemsetD8Async(CUdeviceptr dstDevice, unsigned char uc, size_t N, CUstream hStream);
 typedef CUresult CUDAAPI tcuMemFree_v2(CUdeviceptr dptr);
+typedef CUresult CUDAAPI tcuMemcpy(CUdeviceptr dst, CUdeviceptr src, size_t bytesize);
+typedef CUresult CUDAAPI tcuMemcpyAsync(CUdeviceptr dst, CUdeviceptr src, size_t bytesize, CUstream hStream);
 typedef CUresult CUDAAPI tcuMemcpy2D_v2(const CUDA_MEMCPY2D *pcopy);
 typedef CUresult CUDAAPI tcuMemcpy2DAsync_v2(const CUDA_MEMCPY2D *pcopy, CUstream hStream);
 typedef CUresult CUDAAPI tcuGetErrorName(CUresult error, const char** pstr);

include/ffnvcodec/dynlink_loader.h

@@ -152,6 +152,8 @@ typedef struct CudaFunctions {
     tcuMemAllocPitch_v2 *cuMemAllocPitch;
     tcuMemsetD8Async *cuMemsetD8Async;
     tcuMemFree_v2 *cuMemFree;
+    tcuMemcpy *cuMemcpy;
+    tcuMemcpyAsync *cuMemcpyAsync;
     tcuMemcpy2D_v2 *cuMemcpy2D;
     tcuMemcpy2DAsync_v2 *cuMemcpy2DAsync;
     tcuGetErrorName *cuGetErrorName;
@@ -290,6 +292,8 @@ static inline int cuda_load_functions(CudaFunctions **functions, void *logctx)
     LOAD_SYMBOL(cuMemAllocPitch, tcuMemAllocPitch_v2, "cuMemAllocPitch_v2");
     LOAD_SYMBOL(cuMemsetD8Async, tcuMemsetD8Async, "cuMemsetD8Async");
     LOAD_SYMBOL(cuMemFree, tcuMemFree_v2, "cuMemFree_v2");
+    LOAD_SYMBOL(cuMemcpy, tcuMemcpy, "cuMemcpy");
+    LOAD_SYMBOL(cuMemcpyAsync, tcuMemcpyAsync, "cuMemcpyAsync");
     LOAD_SYMBOL(cuMemcpy2D, tcuMemcpy2D_v2, "cuMemcpy2D_v2");
     LOAD_SYMBOL(cuMemcpy2DAsync, tcuMemcpy2DAsync_v2, "cuMemcpy2DAsync_v2");
     LOAD_SYMBOL(cuGetErrorName, tcuGetErrorName, "cuGetErrorName");