On Thu, Oct 27, 2016 at 10:23:24PM +0200, Vincent Stehlé wrote: > video_device_release() takes a pointer to struct video_device as argument. > Fix two call sites where the address of the pointer is passed instead.
Sorry, I messed up: please ignore that "fix". The 0day robot made me realize this is indeed not a proper fix. The issue remains, though: we cannot call video_device_release() on the vdev structure member, as this will in turn call kfree(). Most probably, vdev needs to be dynamically allocated, or the call to video_device_release() dropped completely. Sorry for the bad patch. Best regards, Vincent.
