On Tue, Dec 12 2023 at 8:13P -0500,
Christoph Hellwig <hch@xxxxxxxxxxxxx> wrote:
> On Tue, Dec 12, 2023 at 07:11:46PM +0800, Hongyu Jin wrote:
> > From: Hongyu Jin <hongyu.jin@xxxxxxxxxx>
> >
> > Move bio_set_ioprio() into submit_bio():
> > 1. Only call bio_set_ioprio() once to set the priority of original bio,
> > the bio that cloned and splited from original bio will auto inherit
> > the priority of original bio in clone process.
> >
> > 2. The IO priority can be passed to module that implement
> > struct gendisk::fops::submit_bio, help resolve some
> > of the IO priority loss issues.
>
> Can we reword this a bit. AFAICS what this primarily does it to ensure
> the priority is set before dispatching to submit_bio based drivers or
> blk-mq instead of just blk-mq, and the rest follows from that.
Yeah, I agree.. something like:
Move bio_set_ioprio() and caller up from blk_mq_submit_bio() to
submit_bio(). This ensures all block drivers call bio_set_ioprio()
during initial bio submission.
> > +static void bio_set_ioprio(struct bio *bio)
> > +{
> > + /* Nobody set ioprio so far? Initialize it based on task's nice value */
> > + if (IOPRIO_PRIO_CLASS(bio->bi_ioprio) == IOPRIO_CLASS_NONE)
> > + bio->bi_ioprio = get_current_ioprio();
> > + blkcg_set_ioprio(bio);
> > +}
>
> I don't think we need the check here as anyone resubmitting a bio should
> be using submit_bio_noact.
This patch moves the caller from blk_mq_submit_bio() to submit_bio().
So I'm not sure why you're seizing on the "resubmitting a bio" usecase
as reason for dropping this check (which occurs in submit_bio).
The original justification for the check is detailed in commit
a78418e6a04c93b ("block: Always initialize bio IO priority on submit").
Mike
