Hi Sergey, On Wed, 2018-08-29 at 13:34 +0900, Sergey Senozhatsky wrote: > Hi, > > Cc-ing Benjamin on this. Thanks! > On (08/29/18 03:23), Dmitry Safonov wrote: > > BUG: unable to handle kernel paging request at 0000000000002260 > > IP: [..] n_tty_receive_buf_common+0x5f/0x86d > > Workqueue: events_unbound flush_to_ldisc > > Call Trace: > > [..] n_tty_receive_buf2 > > [..] tty_ldisc_receive_buf > > [..] flush_to_ldisc > > [..] process_one_work > > [..] worker_thread > > [..] kthread > > [..] ret_from_fork > > Seems that you are not the first one to hit this NULL deref. > > > I think, tty_ldisc_reinit() should be called with ldisc_sem hold > > for > > writing, which will protect any reader against line discipline > > changes. > > Per https://lore.kernel.org/patchwork/patch/777220/ > > : Note that we noticed one path that called reinit without the ldisc > lock > : held for writing, we added that, but it didn't fix the problem. Probably, it's worth to know what exactly has he tried and what was the backtrace he got in the result.. Hopefully, we'll hear more. It might be also worth to review tty_ldisc_deinit(), I thought it's safe to destroy ldisc there under tty lock during tty release, but may be that is another non-safe place. -- Thanks again, Dmitry
