Hi Dan, On Tue, Apr 05, 2022 at 10:01:34AM -0500, Dan Vacura wrote: >Looks like we found the same issue, I submitted the same change the >other week here: >https://lore.kernel.org/all/20220331184024.23918-1-w36195@motorola.com/ > >One difference is you had the reset outside of the queue lock. I figured >to keep it within the lock since we can get a cancel while the pump >worker is running and this can negate the reset. Do you agree? Yes! Your patch is to favour and mine can be dropped from this series. Thanks, Michael > >On Tue, Apr 05, 2022 at 11:43:16AM +0300, Sergey Shtylyov wrote: >> Hello! >> >> On 4/3/22 2:39 AM, Michael Grzeschik wrote: >> >> > On uvcg_queue_cancel the buf_used counter has to be reset. Since the >> > encode function uses the variable to decide if the encoded data has >> > reached the end of frame. Intermediate calls of uvcg_queue_cancel can >> > therefor lead to miscalculations in the encode functions, if buf_used >> >> Therefore? >> >> > was not properly reset. >> > >> > Signed-off-by: Michael Grzeschik >> [...] >> >> MBR, Sergey > -- Pengutronix e.K. | | Steuerwalder Str. 21 | http://www.pengutronix.de/ | 31137 Hildesheim, Germany | Phone: +49-5121-206917-0 | Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |