linux-kernel - Re: [PATCH] mq-deadline: the dd->dispatch queue follows a FIFO policy

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <33dcc737-1419-4f3e-8952-2f34db47a087@acm.org>
Date: Mon, 8 Dec 2025 08:24:16 -0800
From: Bart Van Assche <bvanassche@....org>
To: chengkaitao <pilgrimtao@...il.com>, axboe@...nel.dk
Cc: linux-block@...r.kernel.org, linux-kernel@...r.kernel.org,
 Chengkaitao <chengkaitao@...inos.cn>
Subject: Re: [PATCH] mq-deadline: the dd->dispatch queue follows a FIFO policy

On 12/8/25 4:02 AM, chengkaitao wrote:
> From: Chengkaitao <chengkaitao@...inos.cn>
> 
> In the initial implementation, the 'list_add(&rq->queuelist, ...' statement
> added to the dd_insert_request function was designed to differentiate
> priorities among various IO-requests within the same linked list. For
> example, 'Commit 945ffb60c11d ("mq-deadline: add blk-mq adaptation of the
> deadline IO scheduler")', introduced this 'list_add' operation to ensure
> that requests with the at_head flag would always be dispatched before
> requests without the REQ_TYPE_FS flag.
> 
> Since 'Commit 7687b38ae470 ("bfq/mq-deadline: remove redundant check for
> passthrough request")', removed blk_rq_is_passthrough, the dd->dispatch
> list now contains only requests with the at_head flag. In this context,
> all at_head requests should be treated as having equal priority, and a
> first-in-first-out (FIFO) policy better aligns with the current situation.
> Therefore, replacing list_add with list_add_tail is more appropriate.
> 
> Signed-off-by: Chengkaitao <chengkaitao@...inos.cn>
> ---
>   block/mq-deadline.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/block/mq-deadline.c b/block/mq-deadline.c
> index 3e3719093aec..dcd7f4f1ecd2 100644
> --- a/block/mq-deadline.c
> +++ b/block/mq-deadline.c
> @@ -661,7 +661,7 @@ static void dd_insert_request(struct blk_mq_hw_ctx *hctx, struct request *rq,
>   	trace_block_rq_insert(rq);
>   
>   	if (flags & BLK_MQ_INSERT_AT_HEAD) {
> -		list_add(&rq->queuelist, &dd->dispatch);
> +		list_add_tail(&rq->queuelist, &dd->dispatch);
>   		rq->fifo_time = jiffies;
>   	} else {
>   		deadline_add_rq_rb(per_prio, rq);

I think the current behavior (LIFO) is on purpose and also that it only
should be changed if there is a strong reason to change it. I don't see
a strong reason being mentioned in the patch description.

Bart.