block: Clear BLK_ZONE_WPLUG_PLUGGED when aborting plugged BIOs

Commit fe0418eb9bd6 ("block: Prevent potential deadlocks in zone write
plug error recovery") added a WARN check in disk_put_zone_wplug() to
verify that when the last reference to a zone write plug is dropped,
this zone write plug does not have the BLK_ZONE_WPLUG_PLUGGED flag set,
that is, that it is not plugged.

However, the function disk_zone_wplug_abort(), which is called for zone
reset and zone finish operations, does not clear this flag after
emptying a zone write plug BIO list. This can result in the
disk_put_zone_wplug() warning to trigger if the user (erroneously as
that is bad pratcice) issues zone reset or zone finish operations while
the target zone still has plugged BIOs.

Modify disk_put_zone_wplug() to clear the BLK_ZONE_WPLUG_PLUGGED flag.
And while at it, also add a lockdep annotation to ensure that this
function is called with the zone write plug spinlock held.

Fixes: fe0418eb9bd6 ("block: Prevent potential deadlocks in zone write plug error recovery")
Cc: stable@vger.kernel.org
Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Niklas Cassel <cassel@kernel.org>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

diff --git a/block/blk-zoned.c b/block/blk-zoned.c
index dcc295721c2c08426dd20e259d467d9ab24ed1aa..394d8d74bba9b5357e7e600c337eca25d088e147 100644 (file)
--- a/block/blk-zoned.c
+++ b/block/blk-zoned.c
@@ -741,6 +741,8 @@ static void disk_zone_wplug_abort(struct blk_zone_wplug *zwplug)
 {
        struct bio *bio;
 
+       lockdep_assert_held(&zwplug->lock);
+
        if (bio_list_empty(&zwplug->bio_list))
                return;
 
@@ -748,6 +750,8 @@ static void disk_zone_wplug_abort(struct blk_zone_wplug *zwplug)
                            zwplug->disk->disk_name, zwplug->zone_no);
        while ((bio = bio_list_pop(&zwplug->bio_list)))
                blk_zone_wplug_bio_io_error(zwplug, bio);
+
+       zwplug->flags &= ~BLK_ZONE_WPLUG_PLUGGED;
 }
 
 /*