Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 6b2eef8f authored by Roland Dreier's avatar Roland Dreier
Browse files

IB/mad: Fix possible lock-lock-timer deadlock 

Lockdep reported a possible deadlock with cm_id_priv->lock,
mad_agent_priv->lock and mad_agent_priv->timed_work.timer; this
happens because the mad module does

	cancel_delayed_work(&mad_agent_priv->timed_work);

while holding mad_agent_priv->lock.  cancel_delayed_work() internally
does del_timer_sync(&mad_agent_priv->timed_work.timer).

This can turn into a deadlock because mad_agent_priv->lock is taken
inside cm_id_priv->lock, so we can get the following set of contexts
that deadlock each other:

 A: holding cm_id_priv->lock, waiting for mad_agent_priv->lock
 B: holding mad_agent_priv->lock, waiting for del_timer_sync()
 C: interrupt during mad_agent_priv->timed_work.timer that takes
    cm_id_priv->lock

Fix this by using the new __cancel_delayed_work() interface (which
internally does del_timer() instead of del_timer_sync()) in all the
places where we are holding a lock.

Addresses: http://bugzilla.kernel.org/show_bug.cgi?id=13757


Reported-by: default avatarBart Van Assche <bart.vanassche@gmail.com>
Signed-off-by: default avatarRoland Dreier <rolandd@cisco.com>
parent 60f2b652

drivers/infiniband/core/mad.c

+3 −3
Original line number Diff line number Diff line
@@ -1974,7 +1974,7 @@ static void adjust_timeout(struct ib_mad_agent_private *mad_agent_priv) 
	unsigned long delay;



	if (list_empty(&mad_agent_priv->wait_list)) {

		cancel_delayed_work(&mad_agent_priv->timed_work);

		__cancel_delayed_work(&mad_agent_priv->timed_work);

	} else {

		mad_send_wr = list_entry(mad_agent_priv->wait_list.next,

					 struct ib_mad_send_wr_private,
@@ -1983,7 +1983,7 @@ static void adjust_timeout(struct ib_mad_agent_private *mad_agent_priv) 
		if (time_after(mad_agent_priv->timeout,

			       mad_send_wr->timeout)) {

			mad_agent_priv->timeout = mad_send_wr->timeout;

			cancel_delayed_work(&mad_agent_priv->timed_work);

			__cancel_delayed_work(&mad_agent_priv->timed_work);

			delay = mad_send_wr->timeout - jiffies;

			if ((long)delay <= 0)

				delay = 1;
@@ -2023,7 +2023,7 @@ static void wait_for_response(struct ib_mad_send_wr_private *mad_send_wr) 


	/* Reschedule a work item if we have a shorter timeout */

	if (mad_agent_priv->wait_list.next == &mad_send_wr->agent_list) {

		cancel_delayed_work(&mad_agent_priv->timed_work);

		__cancel_delayed_work(&mad_agent_priv->timed_work);

		queue_delayed_work(mad_agent_priv->qp_info->port_priv->wq,

				   &mad_agent_priv->timed_work, delay);

	}