[PATCH 5.4] io_uring: prevent sq_thread from spinning when it should stop

[PATCH 5.4] io_uring: prevent sq_thread from spinning when it should stop

[Stefano Garzarella]

[ Upstream commit 7143b5ac5750f404ff3a594b34fdf3fc2f99f828 ]

This patch drops 'cur_mm' before calling cond_resched(), to prevent
the sq_thread from spinning even when the user process is finished.

Before this patch, if the user process ended without closing the
io_uring fd, the sq_thread continues to spin until the
'sq_thread_idle' timeout ends.

In the worst case where the 'sq_thread_idle' parameter is bigger than
INT_MAX, the sq_thread will spin forever.

Fixes: 6c271ce2f1d5 ("io_uring: add submission polling")
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
---
 fs/io_uring.c | 20 ++++++++++----------
 1 file changed, 10 insertions(+), 10 deletions(-)

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 709671faaed6..33a95e1ffb15 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -2740,16 +2740,6 @@ static int io_sq_thread(void *data)
 
 		to_submit = io_sqring_entries(ctx);
 		if (!to_submit) {
-			/*
-			 * We're polling. If we're within the defined idle
-			 * period, then let us spin without work before going
-			 * to sleep.
-			 */
-			if (inflight || !time_after(jiffies, timeout)) {
-				cond_resched();
-				continue;
-			}
-
 			/*
 			 * Drop cur_mm before scheduling, we can't hold it for
 			 * long periods (or over schedule()). Do this before
@@ -2762,6 +2752,16 @@ static int io_sq_thread(void *data)
 				cur_mm = NULL;
 			}
 
+			/*
+			 * We're polling. If we're within the defined idle
+			 * period, then let us spin without work before going
+			 * to sleep.
+			 */
+			if (inflight || !time_after(jiffies, timeout)) {
+				cond_resched();
+				continue;
+			}
+
 			prepare_to_wait(&ctx->sqo_wait, &wait,
 						TASK_INTERRUPTIBLE);
 
-- 
2.24.1

Re: [PATCH 5.4] io_uring: prevent sq_thread from spinning when it should stop

[Greg KH]

On Thu, Feb 27, 2020 at 11:43:11AM +0100, Stefano Garzarella wrote:
> [ Upstream commit 7143b5ac5750f404ff3a594b34fdf3fc2f99f828 ]
> 
> This patch drops 'cur_mm' before calling cond_resched(), to prevent
> the sq_thread from spinning even when the user process is finished.
> 
> Before this patch, if the user process ended without closing the
> io_uring fd, the sq_thread continues to spin until the
> 'sq_thread_idle' timeout ends.
> 
> In the worst case where the 'sq_thread_idle' parameter is bigger than
> INT_MAX, the sq_thread will spin forever.
> 
> Fixes: 6c271ce2f1d5 ("io_uring: add submission polling")
> Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
> Signed-off-by: Jens Axboe <axboe@kernel.dk>
> ---
>  fs/io_uring.c | 20 ++++++++++----------
>  1 file changed, 10 insertions(+), 10 deletions(-)

thanks for this, now queued up.

greg k-h