Re: [ISSUE] The time cost of IOSQE_IO_LINK

[Pavel Begunkov <asml.silence@gmail.com>]

[-- Attachment #1.1: Type: text/plain, Size: 1474 bytes --]

On 15/02/2020 09:01, Jens Axboe wrote:
> diff --git a/fs/io_uring.c b/fs/io_uring.c
> index fb94b8bac638..530dcd91fa53 100644
> @@ -4630,6 +4753,14 @@ static void __io_queue_sqe(struct io_kiocb *req, const struct io_uring_sqe *sqe)
>  	 */
>  	if (ret == -EAGAIN && (!(req->flags & REQ_F_NOWAIT) ||
>  	    (req->flags & REQ_F_MUST_PUNT))) {
> +
> +		if (io_arm_poll_handler(req, &retry_count)) {
> +			if (retry_count == 1)
> +				goto issue;

Better to sqe=NULL before retrying, so it won't re-read sqe and try to init the
req twice.

Also, the second sync-issue may -EAGAIN again, and as I remember, read/write/etc
will try to copy iovec into req->io. But iovec is already in req->io, so it will
self memcpy(). Not a good thing.

> +			else if (!retry_count)
> +				goto done_req;
> +			INIT_IO_WORK(&req->work, io_wq_submit_work);

It's not nice to reset it as this:
- prep() could set some work.flags
- custom work.func is more performant (adds extra switch)
- some may rely on specified work.func to be called. e.g. close(), even though
it doesn't participate in the scheme

> +		}
>  punt:
>  		if (io_op_defs[req->opcode].file_table) {
>  			ret = io_grab_files(req);
> @@ -5154,26 +5285,40 @@ void io_uring_task_handler(struct task_struct *tsk)
>  {
>  	LIST_HEAD(local_list);
>  	struct io_kiocb *req;
> +	long state;
>  
>  	spin_lock_irq(&tsk->uring_lock);
>  	if (!list_empty(&tsk->uring_work))

-- 
Pavel Begunkov


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]