linux-kernel - Re: [PATCH] loop: inherit the ioprio in loop woker thread

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <CA+3AYtTbkG_8KWNWJ8rZ-z=v-V+A9CqKCUUsXLPJyHZgL-FjwQ@mail.gmail.com>
Date: Thu, 23 May 2024 22:52:58 +0800
From: yunlong xing <yunlongxing23@...il.com>
To: Jens Axboe <axboe@...nel.dk>
Cc: Bart Van Assche <bvanassche@....org>, Yunlong Xing <yunlong.xing@...soc.com>, niuzhiguo84@...il.com, 
	Hao_hao.Wang@...soc.com, linux-block@...r.kernel.org, 
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] loop: inherit the ioprio in loop woker thread

Jens Axboe <axboe@...nel.dk> 于2024年5月23日周四 21:04写道：
>
> On 5/23/24 12:04 AM, yunlong xing wrote:
> > Bart Van Assche <bvanassche@....org> ?2024?5?23??? 02:12???
> >>
> >> On 5/22/24 10:57, Jens Axboe wrote:
> >>> On 5/22/24 11:38 AM, Bart Van Assche wrote:
> >>>> On 5/22/24 00:48, Yunlong Xing wrote:
> >>>>> @@ -1913,6 +1921,10 @@ static void loop_handle_cmd(struct loop_cmd *cmd)
> >>>>>            set_active_memcg(old_memcg);
> >>>>>            css_put(cmd_memcg_css);
> >>>>>        }
> >>>>> +
> >>>>> +    if (ori_ioprio != cmd_ioprio)
> >>>>> +        set_task_ioprio(current, ori_ioprio);
> >>>>> +
> >>>>>     failed:
> >>>>>        /* complete non-aio request */
> >>>>>        if (!use_aio || ret) {
> >>>>
> >>>> Does adding this call in the hot path have a measurable performance impact?
> >>>
> >>> It's loop, I would not be concerned with overhead. But it does look pretty
> >>> bogus to modify the task ioprio from here.
> >>
> >> Hi Jens,
> >>
> >> Maybe Yunlong uses that call to pass the I/O priority to the I/O submitter?
> >>
> >> I think that it is easy to pass the I/O priority to the kiocb submitted by
> >> lo_rw_aio() without calling set_task_ioprio().
> >>
> >> lo_read_simple() and lo_write_simple() however call vfs_iter_read() /
> >> vfs_iter_write(). This results in a call of do_iter_readv_writev() and
> >> init_sync_kiocb(). The latter function calls get_current_ioprio(). This is
> >> probably why the set_task_ioprio() call has been added?
> >
> > Yeah that's why I call set_task_ioprio.  I want to the loop kwoker
> > task?submit I/O to the real disk device?can pass the iopriority of the
> > loop device request? both lo_rw_aio() and
> > lo_read_simple()/lo_write_simple().
>
> And that's a totally backwards and suboptimal way to do it. The task
> priority is only used as a last resort lower down, if the IO itself
> hasn't been appropriately marked.
>
> Like I said, it's back to the drawing board on this patch, there's no
> way it's acceptable in its current form.
>
> --
> Jens Axboe
>
Thanks for your advice. So, you can't accept pass the ioprio by set_task_ioprio？
If only the method of lo_rw_aio() counld you accept? I don't want to submit this
part of the modifications separately. I just want to know, this is ok
to you or not？

@@ -442,7 +442,6 @@ static int lo_rw_aio(struct loop_device *lo,
struct loop_cmd *cmd,
        cmd->iocb.ki_filp = file;
        cmd->iocb.ki_complete = lo_rw_aio_complete;
        cmd->iocb.ki_flags = IOCB_DIRECT;
-       cmd->iocb.ki_ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE, 0);

        if (rw == ITER_SOURCE)
                ret = call_write_iter(file, &cmd->iocb, &iter);
@@ -1856,6 +1855,9 @@ static blk_status_t loop_queue_rq(struct
blk_mq_hw_ctx *hctx,
                break;
        }

+       /* get request's ioprio */
+       cmd->iocb.ki_ioprio = rq->ioprio;
+
        /* always use the first bio's css */
        cmd->blkcg_css = NULL;
        cmd->memcg_css = NULL;