usb: chipidea: fix deadlock in ci_otg_del_timer
authorDuoming Zhou <duoming@zju.edu.cn>
Sun, 18 Sep 2022 03:33:12 +0000 (11:33 +0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 26 Nov 2022 08:24:45 +0000 (09:24 +0100)
commitead83b0db81f4ceeee7866820c1f3249b2d94db8
tree8f91eff3a1bb6e02af45d7afb8b341de0f979e26
parentcc9e6d8c55c91b642688e5cf0a181f7c095d7cda
usb: chipidea: fix deadlock in ci_otg_del_timer

commit 7a58b8d6021426b796eebfae80983374d9a80a75 upstream.

There is a deadlock in ci_otg_del_timer(), the process is
shown below:

    (thread 1)                  |        (thread 2)
ci_otg_del_timer()              | ci_otg_hrtimer_func()
  ...                           |
  spin_lock_irqsave() //(1)     |  ...
  ...                           |
  hrtimer_cancel()              |  spin_lock_irqsave() //(2)
  (block forever)

We hold ci->lock in position (1) and use hrtimer_cancel() to
wait ci_otg_hrtimer_func() to stop, but ci_otg_hrtimer_func()
also need ci->lock in position (2). As a result, the
hrtimer_cancel() in ci_otg_del_timer() will be blocked forever.

This patch extracts hrtimer_cancel() from the protection of
spin_lock_irqsave() in order that the ci_otg_hrtimer_func()
could obtain the ci->lock.

What`s more, there will be no race happen. Because the
"next_timer" is always under the protection of
spin_lock_irqsave() and we only check whether "next_timer"
equals to NUM_OTG_FSM_TIMERS in the following code.

Fixes: 3a316ec4c91c ("usb: chipidea: use hrtimer for otg fsm timers")
Cc: stable <stable@kernel.org>
Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Link: https://lore.kernel.org/r/20220918033312.94348-1-duoming@zju.edu.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/usb/chipidea/otg_fsm.c