On Wed, 2018-08-29 at 16:40 +0200, Jiri Slaby wrote:

On 08/29/2018, 04:23 AM, Dmitry Safonov wrote:

quoted

tty_ldisc_reinit() doesn't race with neither tty_ldisc_hangup()
nor set_ldisc() nor tty_ldisc_release() as they use tty lock.
But it races with anyone who expects line discipline to be the same
after hoding read semaphore in tty_ldisc_ref().

We've seen the following crash on v4.9.108 stable:

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

I think, tty_ldisc_reinit() should be called with ldisc_sem hold
for
writing, which will protect any reader against line discipline
changes.

Note: I failed to reproduce the described crash, so obiviously
can't
guarantee that this is the place where line discipline was
switched.

Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jiri Slaby <redacted>
Cc: stable@vger.kernel.org
Signed-off-by: Dmitry Safonov <redacted>
---
 drivers/tty/tty_io.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/drivers/tty/tty_io.c b/drivers/tty/tty_io.c
index 5e5da9acaf0a..3ef8b977b167 100644
--- a/drivers/tty/tty_io.c
+++ b/drivers/tty/tty_io.c

@@ -1267,15 +1267,20 @@ static int tty_reopen(struct tty_struct

*tty)
 	if (test_bit(TTY_EXCLUSIVE, &tty->flags) &&
!capable(CAP_SYS_ADMIN))
 		return -EBUSY;
 
-	tty->count++;
+	retval = tty_ldisc_lock(tty, 5 * HZ);

Why 5 secs? This would cause random errors on machines under heavy
load.

Yeah, I think MAX_SCHEDULE_TIMEOUT will make more sense here..
Not sure, why I decided to go with 5*HZ instead.
Will resend with new timeout, if everything else looks good to you.
(having in mind my argument for count++ in 1/4)

quoted

+	if (retval)
+		return retval;
 
+	tty->count++;
 	if (tty->ldisc)
-		return 0;
+		goto out_unlock;
 
 	retval = tty_ldisc_reinit(tty, tty->termios.c_line);
 	if (retval)
 		tty->count--;
 
+out_unlock:
+	tty_ldisc_unlock(tty);
 	return retval;

So what about:
        tty_ldisc_lock(tty, MAX_SCHEDULE_TIMEOUT);
        if (!tty->ldisc)
                ret = tty_ldisc_reinit(tty, tty->termios.c_line);
        tty_ldisc_unlock(tty);

        if (!ret)
                tty->count++;

        return ret;

-- 
Thanks,
             Dmitry