Skip to contentSkip to conversation

aio: fix kioctx not being freed after cancellation at exit time

PATCH In review Benjamin LaHaise 2 participants 2 messages Feb 13, 2013 – Feb 14, 2013

Subsystems (2)
lore
Share & tools
lore

Benjamin LaHaise

author maintainer
#0 lore raw

PATCH 2/2 aio: fix kioctx not being freed after cancellation at exit time

The recent changes overhauling fs/aio.c introduced a bug that results in the kioctx not being freed when outstanding kiocbs are cancelled at exit_aio() time. Specifically, a kiocb that is cancelled has its completion events discarded by batch_complete_aio(), which then fails to wake up the process stuck in free_ioctx(). Fix this by adding a wake_up() in batch_complete_aio() and modifying the wait_event() condition in free_ioctx() appropriately.

---
 fs/aio.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)
fs/aio.c +4 -1
index dc52b0c..46f9dd0 100644
@@ -335,7 +335,9 @@ static void free_ioctx(struct kioctx *ctx)
 	kunmap_atomic(ring);
 
 	while (atomic_read(&ctx->reqs_available) < ctx->nr) {
-		wait_event(ctx->wait, head != ctx->shadow_tail);
+		wait_event(ctx->wait,
+			   (head != ctx->shadow_tail) ||
+			   (atomic_read(&ctx->reqs_available) != ctx->nr));
 
 		avail = (head <= ctx->shadow_tail ?
 			 ctx->shadow_tail : ctx->nr) - head;
@@ -754,6 +756,7 @@ void batch_complete_aio(struct batch_complete *batch)
 			 * with free_ioctx()
 			 */
 			atomic_inc(&req->ki_ctx->reqs_available);
+			wake_up(&req->ki_ctx->wait);
 			aio_put_req(req);
 			continue;
 		}
Signature · 11 lines
1.7.4.1


-- 
"Thought is the essence of where you are now."

--
To unsubscribe, send a message with 'unsubscribe linux-aio' in
the body to [email protected].  For more info on Linux AIO,
see: http://www.kvack.org/aio/
Don't email: <a href=mailto:"[email protected]">[email protected]</a>
#1 lore raw
Quoted 17 lines · Benjamin LaHaise

The recent changes overhauling fs/aio.c introduced a bug that results in the kioctx not being freed when outstanding kiocbs are cancelled at exit_aio() time. Specifically, a kiocb that is cancelled has its completion events discarded by batch_complete_aio(), which then fails to wake up the process stuck in free_ioctx(). Fix this by adding a wake_up() in batch_complete_aio() and modifying the wait_event() condition in free_ioctx() appropriately.

---
 fs/aio.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)
fs/aio.c +0 -0
index dc52b0c..46f9dd0 100644

fs/aio.c static void free_ioctx(struct kioctx *ctx) · line 340

Quoted 7 lines
fs/aio.c +3 -1
@@ -335,7 +335,9 @@ static void free_ioctx(struct kioctx *ctx)
 	kunmap_atomic(ring);
 
 	while (atomic_read(&ctx->reqs_available) < ctx->nr) {
-		wait_event(ctx->wait, head != ctx->shadow_tail);
+		wait_event(ctx->wait,
+			   (head != ctx->shadow_tail) ||
+			   (atomic_read(&ctx->reqs_available) != ctx->nr));

That test looks backwards - I think we want to wait until reqs_available == ctx->nr

Quoted 17 lines
fs/aio.c +1 -0
 		avail = (head <= ctx->shadow_tail ?
 			 ctx->shadow_tail : ctx->nr) - head;
@@ -754,6 +756,7 @@ void batch_complete_aio(struct batch_complete *batch)
 			 * with free_ioctx()
 			 */
 			atomic_inc(&req->ki_ctx->reqs_available);
+			wake_up(&req->ki_ctx->wait);
 			aio_put_req(req);
 			continue;
 		}
Signature · 5 lines
1.7.4.1


-- 
"Thought is the essence of where you are now."