Mapping CUDA memory to a GL buffer object

Hi experts

CUDA OpenGL interop support I learned so far is mapping a GL buffer to the CUDA memory space. Is there a way to perform the other way around, i.e. mapping a memory buffer allocated by cudaMalloc() to a GL buffer object?

My CUDA app pre-allocates device memory space and keeps reusing it. I prefer using the pre-allocated memory space in OpenGL to allocating additional GPU memory in GL side.