On Friday 15 February 2002 03:13 pm, Trent Piepho wrote: > > According to what I think Alan was saying and the API said, > > VIDIOCMCAPTURE does two things: > > > > 1) Returns when a grabbed frame is available. > > Then how could you ever get a queue depth greater than 1? The way I read it, the application does not queue up frames to be captured. It just reads that frame out of the mmap ring buffer. If that frame isn't available (it has a capture pending), it blocks until it is. If you have an mmap buffer of 4 frames (the driver determines how many mmap buffers you have), the driver will capture at most 3 frames ahead of the currently processed frame before stopping. As long as your application can keep up with that no lost frames. > The purpose of VIDIOCMCAPTURE is to tell the driver that you are done with > a buffer, and that the driver should add it to the capture queue. The > purpose of VIDIOCSYNC is for the driver to tell you that it is done with > the buffer, and you should process it. That may be true. But if that is true API.html is explicitly wrong: Whenever a buffer is "used" by the program it should called VIDIOCSYNC to free this frame up and continue. Which I think is what everybody is saying... API.html is completely incorrect on the matter of reading data from mmapped buffers and my interpretation of a driver marking a buffer as "unused" meaning "available" was a side effect of the API document. It actually means the driver isn't using this buffer, the program can do with it as it pleases. The only problem I have with this is that the drivers will flag an error if you VIDIOCSYNC a frame twice, even though the second time the driver still isn't using it. The drivers I've seen have two flags: One which says "I've read the data, it is sitting in the mmap buffer, but the user hasn't SYNCed it", and another which says "I've read the data, it is sitting in the mmap buffer, and user has SYNCed it". The only purpose of this flag seems to be to allow an error to be returned if SYNC is done twice on a buffer before MCAPTURE is called. The semantic of SYNC would seem to be "block until you're done with this frame", but it only works once. Even though it is still done with the frame the second time, it returns -EINVAL. -Joe
