On 2018/08/29 22:53, Jiri Slaby wrote: > On 07/24/2018, 05:22 PM, Tetsuo Handa wrote: >> From 118c64e86641a97d44dec39e313a95b12d9bc3b2 Mon Sep 17 00:00:00 2001 >> From: Tetsuo Handa <penguin-kernel@xxxxxxxxxxxxxxxxxxx> >> Date: Wed, 25 Jul 2018 00:15:18 +0900 >> Subject: [PATCH v2] n_tty: Protect tty->disc_data using refcount. >> >> syzbot is reporting NULL pointer dereference at n_tty_set_termios() [1]. >> This is because ioctl(TIOCVHANGUP) versus ioctl(TCSETS) can race. >> >> Since we don't want to introduce new locking dependency, this patch >> converts "struct n_tty_data *ldata = tty->disc_data;" in individual >> function into a function argument which follows "struct tty *", and >> holds tty->disc_data at each "struct tty_ldisc_ops" hook using refcount >> in order to ensure that memory which contains "struct n_tty_data" will >> not be released while processing individual function. > > This does not look correct and is way too complicated. ioctls should not > be called while changing/killing/hanging/whatever a ldisc. But there is > one missing lock in tty_reopen. > > So does the attached patch helps instead? > > thanks, > That patch seems to help avoiding crashes. (You can use #syz test: command.) But I think you need to check tty_ldisc_lock() return value...
