From: Xing Zhengjun <zhengjun.xing@linux.intel.com>
To: lkp@lists.01.org
Subject: Re: [ext4] 05c2c00f37: aim7.jobs-per-min -11.8% regression
Date: Fri, 03 Sep 2021 13:28:25 +0800 [thread overview]
Message-ID: <8ac28fce-0cfb-cc4d-1a3d-2f104ff46767@linux.intel.com> (raw)
In-Reply-To: <20210603161004.GL23647@quack2.suse.cz>
[-- Attachment #1: Type: text/plain, Size: 1617 bytes --]
Hi Jan,
Do you have time to look at this? I re-test it in v5.13 and v5.14,
the regression still existed. Thanks.
On 6/4/2021 12:10 AM, Jan Kara wrote:
> Similarly to previous test, 'Orig' is the original state before 05c2c00f37,
> 'Patched' is a state after commit 05c2c00f37, 'Hack1' is 05c2c00f37 but with
> lock_buffer() calls removed from orphan handling, 'Hack2' is 05c2c00f37 with
> lock_buffer() calls removed and checksumming moved from under orphan_lock,
> 'BH orphan lock' is 05c2c00f37 with orphan_lock replaced with sb buffer
> lock.
>
> As we can see with fixed filesystem size, the regression isn't actually
> that big anymore but it about matches what 0-day reported. Replacing orphan
> lock with superblock buffer_head lock makes things even much worse - not
> really surprising given we are replacing optimized mutex implementation
> with a bitlock. Just removing buffer lock (Hack1 test) doesn't seem to
> improve the results noticeably so that is not a problem. Moving
> checksumming out from under the orphan_lock would probably help noticeably
> (Hack2 test) but there's the problem when to compute checksums for
> nojournal mode and also we'd need to be very careful with all the other
> places updating superblock so that they serialize against orphan operations
> so that they cannot invalidate the checksum - IMO not very compelling.
>
> So as we chatted on today's call probably the best option is to leave the
> code as is for now and instead work on moving away from orphan list
> altogether. I'll revive my patches to do that.
--
Zhengjun Xing
WARNING: multiple messages have this Message-ID (diff)
From: Xing Zhengjun <zhengjun.xing@linux.intel.com>
To: Jan Kara <jack@suse.cz>, "Theodore Y. Ts'o" <tytso@mit.edu>
Cc: kernel test robot <oliver.sang@intel.com>,
LKML <linux-kernel@vger.kernel.org>,
lkp@lists.01.org, lkp@intel.com
Subject: Re: [LKP] [ext4] 05c2c00f37: aim7.jobs-per-min -11.8% regression
Date: Fri, 3 Sep 2021 13:28:25 +0800 [thread overview]
Message-ID: <8ac28fce-0cfb-cc4d-1a3d-2f104ff46767@linux.intel.com> (raw)
In-Reply-To: <20210603161004.GL23647@quack2.suse.cz>
Hi Jan,
Do you have time to look at this? I re-test it in v5.13 and v5.14,
the regression still existed. Thanks.
On 6/4/2021 12:10 AM, Jan Kara wrote:
> Similarly to previous test, 'Orig' is the original state before 05c2c00f37,
> 'Patched' is a state after commit 05c2c00f37, 'Hack1' is 05c2c00f37 but with
> lock_buffer() calls removed from orphan handling, 'Hack2' is 05c2c00f37 with
> lock_buffer() calls removed and checksumming moved from under orphan_lock,
> 'BH orphan lock' is 05c2c00f37 with orphan_lock replaced with sb buffer
> lock.
>
> As we can see with fixed filesystem size, the regression isn't actually
> that big anymore but it about matches what 0-day reported. Replacing orphan
> lock with superblock buffer_head lock makes things even much worse - not
> really surprising given we are replacing optimized mutex implementation
> with a bitlock. Just removing buffer lock (Hack1 test) doesn't seem to
> improve the results noticeably so that is not a problem. Moving
> checksumming out from under the orphan_lock would probably help noticeably
> (Hack2 test) but there's the problem when to compute checksums for
> nojournal mode and also we'd need to be very careful with all the other
> places updating superblock so that they serialize against orphan operations
> so that they cannot invalidate the checksum - IMO not very compelling.
>
> So as we chatted on today's call probably the best option is to leave the
> code as is for now and instead work on moving away from orphan list
> altogether. I'll revive my patches to do that.
--
Zhengjun Xing
next prev parent reply other threads:[~2021-09-03 5:28 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-27 12:08 [ext4] 05c2c00f37: aim7.jobs-per-min -11.8% regression kernel test robot
2021-02-27 12:08 ` kernel test robot
2021-05-20 7:13 ` Xing Zhengjun
2021-05-20 9:51 ` Jan Kara
2021-05-20 9:51 ` [LKP] " Jan Kara
2021-05-21 1:16 ` Xing Zhengjun
2021-05-21 1:16 ` [LKP] " Xing Zhengjun
2021-05-21 9:27 ` Jan Kara
2021-05-21 9:27 ` [LKP] " Jan Kara
2021-05-21 16:42 ` Theodore Y. Ts'o
2021-05-21 16:42 ` [LKP] " Theodore Y. Ts'o
2021-05-25 9:22 ` Jan Kara
2021-05-25 9:22 ` [LKP] " Jan Kara
2021-05-25 17:15 ` Theodore Ts'o
2021-05-25 17:15 ` [LKP] " Theodore Ts'o
2021-05-31 16:57 ` Jan Kara
2021-05-31 16:57 ` [LKP] " Jan Kara
2021-06-03 16:10 ` Jan Kara
2021-06-03 16:10 ` [LKP] " Jan Kara
2021-09-03 5:28 ` Xing Zhengjun [this message]
2021-09-03 5:28 ` Xing Zhengjun
2021-09-03 12:32 ` Theodore Ts'o
2021-09-03 12:32 ` [LKP] " Theodore Ts'o
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=8ac28fce-0cfb-cc4d-1a3d-2f104ff46767@linux.intel.com \
--to=zhengjun.xing@linux.intel.com \
--cc=lkp@lists.01.org \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.