On Mon, Nov 6, 2017 at 4:20 PM, Oliver Neukum <oneu...@suse.com> wrote: > Am Montag, den 06.11.2017, 13:30 +0100 schrieb Andrey Konovalov: >> On Mon, Nov 6, 2017 at 10:49 AM, Oliver Neukum <oneu...@suse.com> wrote: >> > >> > >> > 2. Will a device work after that? The appropriate fix may be to wait >> > until the device is properly initialized. >> >> This shouldn't affect real devices as far as I understand. The crash >> can be caused by a crafted malicious device. > > Hi! > > Hm. That seems strange as driver_priv is kmalloced. Do you > still have a descriptor that causes this? > Shouldn't we rather reject such a broken device?
I do have a way to reproduce this. As far as I understand, for this particular device ax88172_bind() is called, which doesn't assign anything to dev->driver_priv, so that's why it is NULL in suspend() and resume(). > > Regards > Oliver >
