On Thu, Jan 28, 2021 at 1:45 PM Nicolas Dufresne <nicolas@xxxxxxxxxxxx> wrote: > > Le vendredi 15 janvier 2021 à 20:33 -0800, Fritz Koenig a écrit : > > Hi, > > > > I'm trying to sort some issues with regards to decoding multiple > > streams at once. Logging can be helpful with messages like this: > > > > [105188.500450] video2: VIDIOC_STREAMOFF: type=vid-out-mplane > > > > but when I have: > > > > [105188.498697] video2: VIDIOC_STREAMOFF: type=vid-out-mplane > > [105188.498715] video2: VIDIOC_STREAMOFF: type=vid-cap-mplane > > [105188.499116] video2: VIDIOC_UNSUBSCRIBE_EVENT: type=0x5, id=0x0, flags=0x0 > > [105188.500450] video2: VIDIOC_STREAMOFF: type=vid-out-mplane > > [105188.500467] video2: VIDIOC_STREAMOFF: type=vid-cap-mplane > > [105188.500611] video2: VIDIOC_UNSUBSCRIBE_EVENT: type=0x5, id=0x0, flags=0x0 > > [105188.501879] video2: VIDIOC_UNSUBSCRIBE_EVENT: type=0x5, id=0x0, flags=0x0 > > [105188.502276] video2: VIDIOC_STREAMOFF: type=vid-out-mplane > > [105188.502331] video2: VIDIOC_STREAMOFF: type=vid-cap-mplane > > > > I can't tell which stream VIDIOC_STREAMOFF is associated with. Is > > there a logging option that I'm missing that would help indicate which > > stream that ioctl is associated with. > > I suppose this is m2m driver ? I think it would be nice to trace some instance > ID, otherwise indeed, if you have two stream on the same m2m device node, the > traces get confused. > Yes, m2m. Instance doesn't seem to be a concept at that level. The driver itself holds an idea of instances (at least on venus). I was able to add debugging to trace instances in vdec.c, but I don't see a generic way to add it. -Fritz > > > > > Thanks. > > > > -Fritz > >
