Received-SPF: pass (google.com: domain of linux-kernel+bounces-39929-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Date: Fri, 26 Jan 2024 09:36:36 +0000
From: Matthew Wilcox <willy@infradead.org>
To: Zhaoyang Huang <huangzhaoyang@gmail.com>
Cc: "zhaoyang.huang" <zhaoyang.huang@unisoc.com>,
	Alexander Viro <viro@zeniv.linux.org.uk>,
	Christian Brauner <brauner@kernel.org>,
	linux-fsdevel@vger.kernel.org,
	Andrew Morton <akpm@linux-foundation.org>,
	Jens Axboe <axboe@kernel.dk>, Yu Zhao <yuzhao@google.com>,
	Damien Le Moal <dlemoal@kernel.org>,
	Niklas Cassel <niklas.cassel@wdc.com>,
	"Martin K . Petersen" <martin.petersen@oracle.com>,
	Hannes Reinecke <hare@suse.de>,
	Linus Walleij <linus.walleij@linaro.org>, linux-mm@kvack.org,
	linux-block@vger.kernel.org, linux-kernel@vger.kernel.org,
	steve.kang@unisoc.com
Subject: Re: [PATCHv3 1/1] block: introduce content activity based ioprio
Message-ID: <ZbN9JDE50Th-dT3Y@casper.infradead.org>
References: <20240125071901.3223188-1-zhaoyang.huang@unisoc.com>
 <CAGWkznGpW=bUxET8yZGu4dNTBfsj7n79yXsTD23fE5-SWkdjfA@mail.gmail.com>
 <ZbNziLeet7TbDKEl@casper.infradead.org>
 <CAGWkznGG1xLcPMsWbbXqO5iUWqC2UmyWwcJaFd4WBQ-aFE=-jA@mail.gmail.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <CAGWkznGG1xLcPMsWbbXqO5iUWqC2UmyWwcJaFd4WBQ-aFE=-jA@mail.gmail.com>

On Fri, Jan 26, 2024 at 05:28:58PM +0800, Zhaoyang Huang wrote:
> On Fri, Jan 26, 2024 at 4:55 PM Matthew Wilcox <willy@infradead.org> wrote:
> >
> > On Fri, Jan 26, 2024 at 03:59:48PM +0800, Zhaoyang Huang wrote:
> > > loop more mm and fs guys for more comments
> >
> > I agree with everything Damien said.  But also ...
> ok, I will find a way to solve this problem.
> >
> > > > +bool BIO_ADD_FOLIO(struct bio *bio, struct folio *folio, size_t len,
> > > > +               size_t off)
> >
> > You don't add any users of these functions.  It's hard to assess whether
> > this is the right API when there are no example users.
> Actually, the code has been tested on ext4 and f2fs by patchv2 on a
> v6.6 6GB android system where I get the test result posted on the
> commit message. These APIs is to keep block layer clean and wrap
> things up for fs.

well, where's patch v2?  i don't see it in my inbox.  i'm not going
to go hunting around the email lists for it.  this is not good enough.

> > why are BIO_ADD_PAGE and BIO_ADD_FOLIO so very different from each
> > other?
> These two API just repeat the same thing that bio_add_page and
> bio_add_folio do.

what?

here's the patch you sent.  these two functions do wildly different
things:

+bool BIO_ADD_FOLIO(struct bio *bio, struct folio *folio, size_t len,
+		size_t off)
+{
+	int class, level, hint, activity;
+
+	if (len > UINT_MAX || off > UINT_MAX)
+		return false;
+
+	class = IOPRIO_PRIO_CLASS(bio->bi_ioprio);
+	level = IOPRIO_PRIO_LEVEL(bio->bi_ioprio);
+	hint = IOPRIO_PRIO_HINT(bio->bi_ioprio);
+	activity = IOPRIO_PRIO_ACTIVITY(bio->bi_ioprio);
+
+	activity += (bio->bi_vcnt + 1 <= IOPRIO_NR_ACTIVITY &&
+			PageWorkingset(&folio->page)) ? 1 : 0;
+	if (activity >= bio->bi_vcnt / 2)
+		class = IOPRIO_CLASS_RT;
+	else if (activity >= bio->bi_vcnt / 4)
+		class = max(IOPRIO_PRIO_CLASS(get_current_ioprio()), IOPRIO_CLASS_BE);
+
+	bio->bi_ioprio = IOPRIO_PRIO_VALUE_ACTIVITY(class, level, hint, activity);
+
+	return bio_add_page(bio, &folio->page, len, off) > 0;
+}
+
+int BIO_ADD_PAGE(struct bio *bio, struct page *page,
+		unsigned int len, unsigned int offset)
+{
+	int class, level, hint, activity;
+
+	if (bio_add_page(bio, page, len, offset) > 0) {
+		class = IOPRIO_PRIO_CLASS(bio->bi_ioprio);
+		level = IOPRIO_PRIO_LEVEL(bio->bi_ioprio);
+		hint = IOPRIO_PRIO_HINT(bio->bi_ioprio);
+		activity = IOPRIO_PRIO_ACTIVITY(bio->bi_ioprio);
+		activity += (bio->bi_vcnt <= IOPRIO_NR_ACTIVITY && PageWorkingset(page)) ? 1 : 0;
+		bio->bi_ioprio = IOPRIO_PRIO_VALUE_ACTIVITY(class, level, hint, activity);
+	}
+
+	return len;
+}

did you change one and forget to change the other?

> These white spaces are trimmed by vim, I will change them back in next version.

vim doesn't do that by default.