On Tuesday 20 February 2007 22:05, Martin Funk wrote: > Hi Thomas, > > verry same output of v4l-video here, just the bus_info differs. > It also makes no difference if the given uvc_ids is added to the code or > not.
Here's what bothers me in your kernel log: [ 284.165685] uvcvideo: uvc_v4l2_ioctl [ 284.165691] v4l2 ioctl VIDIOC_STREAMON, dir=-w (0x40045612) [ 284.165967] uvcvideo: uvc_v4l2_ioctl [ 284.165970] v4l2 ioctl VIDIOC_DQBUF, dir=rw (0xc0585611) [ 284.166061] uvcvideo: Dequeuing buffer 0. [ 284.171785] uvcvideo: Dropping packet (out of sync). [ 284.171788] uvcvideo: Dropping packet (out of sync). [ 284.171790] uvcvideo: Dropping packet (out of sync). [ 284.171791] uvcvideo: Dropping packet (out of sync). Right after started streaming, the application tries to dequeue the first buffer using VIDIOC_DQBUF. The ioctl checks the first buffer in the queue (which is buffer 0) and blocks until the buffer is filled with image data. It seems from what you reported that this never happens. The driver keeps printing out-of-sync messages, and that shouldn't happen. An out-of-sync message can happen from times to times, but definitely not at the very beginning. Could you hack uvc_video.c to print more information in the out of sync message, such as the buffer number (buff->buf.index) and state (buff->buf.index) ? Best regards, Laurent Pinchart _______________________________________________ Linux-uvc-devel mailing list [email protected] https://lists.berlios.de/mailman/listinfo/linux-uvc-devel
