On Fri, Apr 14, 2023 at 9:14 PM David Wei <davidhwei@xxxxxxxx> wrote:
>
> A multishot timeout submission will repeatedly generate completions with
> the IORING_CQE_F_MORE cflag set. Depending on the value of the `off'
> field in the submission, these timeouts can either repeat indefinitely
> until cancelled (`off' = 0) or for a fixed number of times (`off' > 0).
>
> Only noseq timeouts (i.e. not dependent on the number of I/O
> completions) are supported.
>
> An indefinite timer will be cancelled with EOVERFLOW if the CQ ever
> overflows.
>
> Signed-off-by: David Wei <davidhwei@xxxxxxxx>
...
> +static void io_timeout_complete(struct io_kiocb *req, struct io_tw_state *ts)
> +{
> + struct io_timeout *timeout = io_kiocb_to_cmd(req, struct io_timeout);
> + struct io_timeout_data *data = req->async_data;
> + struct io_ring_ctx *ctx = req->ctx;
> +
> + if (!io_timeout_finish(timeout, data)) {
> + bool filled;
> + filled = io_aux_cqe(ctx, false, req->cqe.user_data, -ETIME,
> + IORING_CQE_F_MORE, false);
I think that the defer parameter (the first false) might want to be ts->locked.
