linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Christoph Hellwig <hch@infradead.org>
To: Qu Wenruo <quwenruo.btrfs@gmx.com>
Cc: Christoph Hellwig <hch@infradead.org>, Qu Wenruo <wqu@suse.com>,
	linux-btrfs@vger.kernel.org, djwong@kernel.org,
	linux-xfs@vger.kernel.org, linux-fsdevel@vger.kernel.org,
	linux-block@vger.kernel.org, linux-mm@kvack.org,
	martin.petersen@oracle.com, jack@suse.com
Subject: Re: O_DIRECT vs BLK_FEAT_STABLE_WRITES, was Re: [PATCH] btrfs: never trust the bio from direct IO
Date: Tue, 21 Oct 2025 00:48:08 -0700	[thread overview]
Message-ID: <aPc6uLKJkavZ_SkM@infradead.org> (raw)
In-Reply-To: <b91eb17a-71ce-422c-99a1-c2970a015666@gmx.com>

On Tue, Oct 21, 2025 at 01:47:03PM +1030, Qu Wenruo wrote:
> Off-topic a little, mind to share the performance drop with PI enabled on
> XFS?

If the bandwith of the SSDs get close or exceeds the DRAM bandwith
buffered I/O can be 50% or less of the direct I/O performance.

> With this patch I'm able to enable direct IO for inodes with checksums.
> I thought it would easily improve the performance, but the truth is, it's
> not that different from buffered IO fall back.

That's because you still copy data.

> So I start wondering if it's the checksum itself causing the miserable
> performance numbers.

Only indirectly by touching all the cachelines.  But once you copy you
touch them again.  Especially if not done in small chunks.



  reply	other threads:[~2025-10-21  7:48 UTC|newest]

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <1ee861df6fbd8bf45ab42154f429a31819294352.1760951886.git.wqu@suse.com>
2025-10-20 10:00 ` O_DIRECT vs BLK_FEAT_STABLE_WRITES, was Re: [PATCH] btrfs: never trust the bio from direct IO Christoph Hellwig
2025-10-20 10:24   ` Qu Wenruo
2025-10-20 11:45     ` Christoph Hellwig
2025-10-20 11:16   ` Jan Kara
2025-10-20 11:44     ` Christoph Hellwig
2025-10-20 13:59       ` Jan Kara
2025-10-20 14:59         ` Matthew Wilcox
2025-10-20 15:58           ` Jan Kara
2025-10-20 17:55             ` John Hubbard
2025-10-21  8:27               ` Jan Kara
2025-10-21 16:56                 ` John Hubbard
2025-10-20 19:00             ` David Hildenbrand
2025-10-21  7:49               ` Christoph Hellwig
2025-10-21  7:57                 ` David Hildenbrand
2025-10-21  9:33                   ` Jan Kara
2025-10-21  9:43                     ` David Hildenbrand
2025-10-21  9:22                 ` Jan Kara
2025-10-21  9:37                   ` David Hildenbrand
2025-10-21  9:52                     ` Jan Kara
2025-10-21  3:17   ` Qu Wenruo
2025-10-21  7:48     ` Christoph Hellwig [this message]
2025-10-21  8:15       ` Qu Wenruo
2025-10-21 11:30         ` Johannes Thumshirn
2025-10-22  2:27           ` Qu Wenruo
2025-10-22  5:04             ` hch
2025-10-22  6:17               ` Qu Wenruo
2025-10-22  6:24                 ` hch

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=aPc6uLKJkavZ_SkM@infradead.org \
    --to=hch@infradead.org \
    --cc=djwong@kernel.org \
    --cc=jack@suse.com \
    --cc=linux-block@vger.kernel.org \
    --cc=linux-btrfs@vger.kernel.org \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linux-xfs@vger.kernel.org \
    --cc=martin.petersen@oracle.com \
    --cc=quwenruo.btrfs@gmx.com \
    --cc=wqu@suse.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).