md/raid1,raid10: silence warning about wait-within-wait

If you prepare_to_wait() after a previous prepare_to_wait(),
but before calling schedule(), you get warning:

  do not call blocking ops when !TASK_RUNNING; state=2

This is appropriate as it is often a bug.  The event that the
first prepare_to_wait() expects might wake up the schedule following
the second prepare_to_wait(), which could be confusing.

However if both prepare_to_wait()s are part of simple wait_event()
loops, and if the inner one is rarely called, then there is
no problem.  The inner loop is too simple to get confused by
a stray wakeup, and the outer loop won't spin unduly because the
inner doesnt affect it often.

This pattern occurs in both raid1.c and raid10.c in the use of
flush_pending_writes().

The warning can be silenced by setting current->state to TASK_RUNNING.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Shaohua Li <shli@fb.com>

1 files changed, 12 insertions, 0 deletions

diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index c131835cf008..99c9207899a7 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -900,6 +900,18 @@ static void flush_pending_writes(struct r10conf *conf)
 		bio = bio_list_get(&conf->pending_bio_list);
 		conf->pending_count = 0;
 		spin_unlock_irq(&conf->device_lock);
+
+		/*
+		 * As this is called in a wait_event() loop (see freeze_array),
+		 * current->state might be TASK_UNINTERRUPTIBLE which will
+		 * cause a warning when we prepare to wait again.  As it is
+		 * rare that this path is taken, it is perfectly safe to force
+		 * us to go around the wait_event() loop again, so the warning
+		 * is a false-positive. Silence the warning by resetting
+		 * thread state
+		 */
+		__set_current_state(TASK_RUNNING);
+
 		blk_start_plug(&plug);
 		/* flush any pending bitmap writes to disk
 		 * before proceeding w/ I/O */