Received-SPF: pass (google.com: domain of linux-kernel+bounces-36723-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Precedence: bulk
MIME-Version: 1.0
References: <20240124085334.3060748-1-zhaoyang.huang@unisoc.com> <ZbDWSUkT/OjHTe0t@infradead.org>
In-Reply-To: <ZbDWSUkT/OjHTe0t@infradead.org>
From: Zhaoyang Huang <huangzhaoyang@gmail.com>
Date: Wed, 24 Jan 2024 17:35:47 +0800
Message-ID: <CAGWkznHAi_U5erM0s8vFWwwucRKPoSzimNr9tdsPvs_sApGxgQ@mail.gmail.com>
Subject: Re: [PATCHv2 1/1] block: introduce content activity based ioprio
To: Christoph Hellwig <hch@infradead.org>
Cc: "zhaoyang.huang" <zhaoyang.huang@unisoc.com>, Jens Axboe <axboe@kernel.dk>, 
	linux-block@vger.kernel.org, linux-kernel@vger.kernel.org, 
	steve.kang@unisoc.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Wed, Jan 24, 2024 at 5:20=E2=80=AFPM Christoph Hellwig <hch@infradead.or=
g> wrote:
>
> On Wed, Jan 24, 2024 at 04:53:34PM +0800, zhaoyang.huang wrote:
> >  void __bio_add_page(struct bio *bio, struct page *page,
> >               unsigned int len, unsigned int off)
> >  {
> > +     int class, level, hint, activity;
> > +
> > +     class =3D IOPRIO_PRIO_CLASS(bio->bi_ioprio);
> > +     level =3D IOPRIO_PRIO_LEVEL(bio->bi_ioprio);
> > +     hint =3D IOPRIO_PRIO_HINT(bio->bi_ioprio);
> > +     activity =3D IOPRIO_PRIO_ACTIVITY(bio->bi_ioprio);
> > +
> >       WARN_ON_ONCE(bio_flagged(bio, BIO_CLONED));
> >       WARN_ON_ONCE(bio_full(bio, len));
> >
> >       bvec_set_page(&bio->bi_io_vec[bio->bi_vcnt], page, len, off);
> >       bio->bi_iter.bi_size +=3D len;
> >       bio->bi_vcnt++;
> > +     activity +=3D bio_page_if_active(bio, page, IOPRIO_NR_ACTIVITY);
> > +     bio->bi_ioprio =3D IOPRIO_PRIO_VALUE_ACTIVITY(class, level, hint,=
 activity);
>
> If you need to touch anything in the block layer I/O path
> you're doign this wrong.  The file system that is submitting the
> I/O needs to be in control of the priorities.
> must not hack the assignment behind the
Please correct me if I am wrong. According to my understanding,
bio(request)'s ioprio is set via either task's scheduler priority or
blkcg's priority during submit_bio, that is, there is no explicit
operation over ioprio from the file system so far. Furthermore, this
commit is actually an add-on feature which is not against the current
way of setting ioprio.