IPoIB: Fix deadlock on RTNL in ipoib_stop()
Commit c8c2afe3 ("IPoIB: Use rtnl lock/unlock when changing device
flags") added a call to rtnl_lock() in ipoib_mcast_join_task(), which
is run from the ipoib_workqueue. However, ipoib_stop() (which is run
inside rtnl_lock()) flushes this workqueue, which leads to a deadlock
if the join task is pending.
Fix this by simply not flushing the workqueue from ipoib_stop(). It
turns out that we really don't care about workqueue tasks running
during or after ipoib_stop(), as long as we make sure to flush the
workqueue before unregistering a netdev.
This fixes <https://bugs.openfabrics.org/show_bug.cgi?id=1114>.
Signed-off-by: Roland Dreier <rolandd@cisco.com>
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c
index f51201b..7e9e218 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
@@ -156,14 +156,8 @@
netif_stop_queue(dev);
- /*
- * Now flush workqueue to make sure a scheduled task doesn't
- * bring our internal state back up.
- */
- flush_workqueue(ipoib_workqueue);
-
- ipoib_ib_dev_down(dev, 1);
- ipoib_ib_dev_stop(dev, 1);
+ ipoib_ib_dev_down(dev, 0);
+ ipoib_ib_dev_stop(dev, 0);
if (!test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags)) {
struct ipoib_dev_priv *cpriv;
@@ -1314,7 +1308,7 @@
register_failed:
ib_unregister_event_handler(&priv->event_handler);
- flush_scheduled_work();
+ flush_workqueue(ipoib_workqueue);
event_failed:
ipoib_dev_cleanup(priv->dev);
@@ -1373,7 +1367,12 @@
list_for_each_entry_safe(priv, tmp, dev_list, list) {
ib_unregister_event_handler(&priv->event_handler);
- flush_scheduled_work();
+
+ rtnl_lock();
+ dev_change_flags(priv->dev, priv->dev->flags & ~IFF_UP);
+ rtnl_unlock();
+
+ flush_workqueue(ipoib_workqueue);
unregister_netdev(priv->dev);
ipoib_dev_cleanup(priv->dev);