On 03/09/17 14:21, Jorge Ramirez wrote: > On 09/03/2017 02:27 AM, Mark Thompson wrote: >>> + /* 0. subscribe to source change event */ >>> + memset(&sub, 0, sizeof(sub)); >>> + sub.type = V4L2_EVENT_SOURCE_CHANGE; >>> + ret = ioctl(s->fd, VIDIOC_SUBSCRIBE_EVENT, &sub); >>> + if ( ret < 0) >>> + av_log(avctx, AV_LOG_WARNING, "decoding does not support >>> resolution change\n"); >> How fatal should that be? What happens if it changes anyway? >> > the buffers are allocated and queued in the driver with a size for a > resolution and format: if the resolution changes down-up the image will not > fit in the previously allocated buffers - not sure if that will cause some > drivers to generate bus errors....I guess it will depend on each driver; also > the sizes reported back with the frame after decoding to ffmpeg will not have > been fixed properly since we couldnt receive the event so I doubt the images > can be recomposed. > > Basically the v4l2 kernel driver will have no way to report changes back to > ffmpeg...
Well, if this is likely to be entirely broken then maybe it should be disabled if the event isn't available? _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
