Desmond Cheong Zhi Xi ae4240d1f4 fcntl: fix potential deadlock for &fasync_struct.fa_lock
[ Upstream commit 2f488f698fda820f8e6fa0407630154eceb145d6 ]

There is an existing lock hierarchy of
&dev->event_lock --> &fasync_struct.fa_lock --> &f->f_owner.lock
from the following call chain:

  input_inject_event():
    spin_lock_irqsave(&dev->event_lock,...);
    input_handle_event():
      input_pass_values():
        input_to_handler():
          evdev_events():
            evdev_pass_values():
              spin_lock(&client->buffer_lock);
              __pass_event():
                kill_fasync():
                  kill_fasync_rcu():
                    read_lock(&fa->fa_lock);
                    send_sigio():
                      read_lock_irqsave(&fown->lock,...);

&dev->event_lock is HARDIRQ-safe, so interrupts have to be disabled
while grabbing &fasync_struct.fa_lock, otherwise we invert the lock
hierarchy. However, since kill_fasync which calls kill_fasync_rcu is
an exported symbol, it may not necessarily be called with interrupts
disabled.

As kill_fasync_rcu may be called with interrupts disabled (for
example, in the call chain above), we replace calls to
read_lock/read_unlock on &fasync_struct.fa_lock in kill_fasync_rcu
with read_lock_irqsave/read_unlock_irqrestore.

Signed-off-by: Desmond Cheong Zhi Xi <desmondcheongzx@gmail.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
2021-09-15 09:47:28 +02:00
..
2020-11-05 11:43:34 +01:00
2019-09-19 09:42:37 -07:00
2021-07-14 16:53:17 +02:00
2021-07-14 16:53:09 +02:00
2021-08-04 12:27:37 +02:00
2021-09-15 09:47:27 +02:00
2020-06-17 16:40:24 +02:00
2020-10-01 13:17:19 +02:00
2019-08-07 21:51:47 -04:00
2020-03-25 08:25:58 +01:00
2019-12-17 19:55:30 +01:00
2019-09-18 16:59:14 -07:00
2019-09-18 16:59:14 -07:00
2019-08-01 20:51:23 +02:00