We currently have to use polling / timeout in order to check when the GPU is done with a V4L2 buffer. This introduces a potential latency in the media pipeline which could effect performance.
Ideally we should get a callback from the SGX kernel services to help us.
One approach is to use the SGX general event queue to issue a callback to check the sync object status.
We currently have to use polling / timeout in order to check when the GPU is done with a V4L2 buffer. This introduces a potential latency in the media pipeline which could effect performance.
Ideally we should get a callback from the SGX kernel services to help us.
One approach is to use the SGX general event queue to issue a callback to check the sync object status.