[PATCH v2] io_uring: fix skipping disabling sqo on exec

[PATCH v2] io_uring: fix skipping disabling sqo on exec

[Pavel Begunkov]

If there are no requests at the time __io_uring_task_cancel() is called,
tctx_inflight() returns zero and and it terminates not getting a chance
to go through __io_uring_files_cancel() and do
io_disable_sqo_submit(). And we absolutely want them disabled by the
time cancellation ends.

Reported-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---

v2: drop sq_data stuff, can't happen

 fs/io_uring.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index d494c4269fc5..383ff6ed3734 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -9085,6 +9085,10 @@ void __io_uring_task_cancel(void)
 	/* make sure overflow events are dropped */
 	atomic_inc(&tctx->in_idle);
 
+	/* trigger io_disable_sqo_submit() */
+	if (tctx->sqpoll)
+		__io_uring_files_cancel(NULL);
+
 	do {
 		/* read completions before cancelations */
 		inflight = tctx_inflight(tctx);
-- 
2.24.0

Re: [PATCH v2] io_uring: fix skipping disabling sqo on exec

[Jens Axboe]

On 1/16/21 7:29 PM, Pavel Begunkov wrote:
> If there are no requests at the time __io_uring_task_cancel() is called,
> tctx_inflight() returns zero and and it terminates not getting a chance
> to go through __io_uring_files_cancel() and do
> io_disable_sqo_submit(). And we absolutely want them disabled by the
> time cancellation ends.

Thanks, that looks much better! Applied.

-- 
Jens Axboe