On 31/01/2021 14:39, Hao Xu wrote:
[...]
> So check if sqthread gets park flag right before schedule().
> since ctx_list is always empty when this problem happens, here I put
> kthread_should_park() before setting the wakeup flag(ctx_list is empty
> so this for loop is fast), where is close enough to schedule(). The
> problem doesn't show again in my repro testing after this fix.
>
Reviewed-by: Pavel Begunkov <asml.silence@xxxxxxxxx>
> Reported-by: Abaci <abaci@xxxxxxxxxxxxxxxxx>
> Signed-off-by: Hao Xu <haoxu@xxxxxxxxxxxxxxxxx>
> ---
>
> v1-->v2
> - tweak the commit message
>
> v2-->v3
> - remove duplicate kthread_should_park() since thread parking is rare
> operation
> - put kthread_should_park() in if (needs_sched)
>
>
> fs/io_uring.c | 5 +----
> 1 file changed, 1 insertion(+), 4 deletions(-)
>
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index c07913ec0cca..d9019ce2bda0 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -7115,9 +7115,6 @@ static int io_sq_thread(void *data)
> continue;
> }
>
> - if (kthread_should_park())
> - continue;
> -
> needs_sched = true;
> prepare_to_wait(&sqd->wait, &wait, TASK_INTERRUPTIBLE);
> list_for_each_entry(ctx, &sqd->ctx_list, sqd_list) {
> @@ -7132,7 +7129,7 @@ static int io_sq_thread(void *data)
> }
> }
>
> - if (needs_sched) {
> + if (needs_sched && !kthread_should_park()) {
> list_for_each_entry(ctx, &sqd->ctx_list, sqd_list)
> io_ring_set_wakeup_flag(ctx);
>
>
--
Pavel Begunkov
