A potential use case is that video frames may be rendered either through a hardware overlay or through the GPU and we do not know this destination ahead of time. One solution is to have the V4L2-dss driver perform the memory allocation and use those buffers with the V4L2-GFX driver if GPU rendering is requested.
Ideally memory allocation would be separate from either driver, but the V4L2 architecture doesn't currently appear to support this.
A potential use case is that video frames may be rendered either through a hardware overlay or through the GPU and we do not know this destination ahead of time. One solution is to have the V4L2-dss driver perform the memory allocation and use those buffers with the V4L2-GFX driver if GPU rendering is requested.
Ideally memory allocation would be separate from either driver, but the V4L2 architecture doesn't currently appear to support this.