On 27/01/2021 07:14, Hao Xu wrote:
[...]
> This is caused by calling io_cqring_overflow_flush() which may sleep
> after calling prepare_to_wait_exclusive() which set task state to
> TASK_INTERRUPTIBLE
Looks good. The loop may use some refactoring for 5.12
Reviewed-by: Pavel Begunkov <asml.silence@xxxxxxxxx>
>
> Reported-by: Abaci <abaci@xxxxxxxxxxxxxxxxx>
> Fixes: 6c503150ae33 ("io_uring: patch up IOPOLL overflow_flush sync")
> Signed-off-by: Hao Xu <haoxu@xxxxxxxxxxxxxxxxx>
> ---
> fs/io_uring.c | 8 ++++++--
> 1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index c07913ec0cca..3ca69a425182 100644
> --- a/fs/io_uring.c
> +++ b/fs/io_uring.c
> @@ -7266,14 +7266,18 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events,
> TASK_INTERRUPTIBLE);
> /* make sure we run task_work before checking for signals */
> ret = io_run_task_work_sig();
> - if (ret > 0)
> + if (ret > 0) {
> + finish_wait(&ctx->wait, &iowq.wq);
> continue;
> + }
> else if (ret < 0)
> break;
> if (io_should_wake(&iowq))
> break;
> - if (test_bit(0, &ctx->cq_check_overflow))
> + if (test_bit(0, &ctx->cq_check_overflow)) {
> + finish_wait(&ctx->wait, &iowq.wq);
> continue;
> + }
> if (uts) {
> timeout = schedule_timeout(timeout);
> if (timeout == 0) {
>
--
Pavel Begunkov
