The function clCreateFromGLBuffer() takes as its second argument CL_MEM_READ_ONLY, CL_MEM_WRITE_ONLY, or CL_MEM_READ_WRITE.
The descriptions of these values refers to how the memory object will be used in a kernel.
But what about using clEnqueueWriteBuffer() when I use CL_MEM_READ_ONLY flag? The enqueue buffer command is coming from the host, not a kernel. Does this distinction matter? The specification doesn’t say.