From: Keith Busch <kbusch@xxxxxxxxxx>
The bitmap wait state may have set an active count, but doesn't have an
active waiter due to racing with adding it. If that happens, the state's
wait_cnt will be set at the wrong value, and could prevent future wakes
until the atomic_dec wraps back to 0.
Check the waitqueue_active before decrementing so that we don't need to
account for fixing it up after.
Fixes: 040b83fcecfb8 ("sbitmap: fix possible io hung due to lost wakeup")
Signed-off-by: Keith Busch <kbusch@xxxxxxxxxx>
---
lib/sbitmap.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/lib/sbitmap.c b/lib/sbitmap.c
index a39b1a877366..ad0670b580d3 100644
--- a/lib/sbitmap.c
+++ b/lib/sbitmap.c
@@ -609,12 +609,16 @@ static bool __sbq_wake_up(struct sbitmap_queue *sbq)
if (!ws)
return false;
+ if (!waitqueue_active(&ws->wait))
+ return true;
+
wait_cnt = atomic_dec_return(&ws->wait_cnt);
+
/*
* For concurrent callers of this, callers should call this function
* again to wakeup a new batch on a different 'ws'.
*/
- if (wait_cnt < 0 || !waitqueue_active(&ws->wait))
+ if (wait_cnt < 0)
return true;
if (wait_cnt > 0)
--
2.30.2
