linux-kernel - Re: [linus:master] [vfs] 681ce86235: filebench.sum

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date: Thu, 30 May 2024 15:37:13 +0800
From: Yafang Shao <laoar.shao@...il.com>
To: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: kernel test robot <oliver.sang@...el.com>, oe-lkp@...ts.linux.dev, lkp@...el.com, 
	linux-kernel@...r.kernel.org, Al Viro <viro@...iv.linux.org.uk>, 
	Christian Brauner <brauner@...nel.org>, Jan Kara <jack@...e.cz>, Waiman Long <longman@...hat.com>, 
	Matthew Wilcox <willy@...radead.org>, Wangkai <wangkai86@...wei.com>, 
	Colin Walters <walters@...bum.org>, linux-fsdevel@...r.kernel.org, ying.huang@...el.com, 
	feng.tang@...el.com, fengwei.yin@...el.com
Subject: Re: [linus:master] [vfs] 681ce86235: filebench.sum_operations/s -7.4% regression

On Thu, May 30, 2024 at 12:38 AM Linus Torvalds
<torvalds@...ux-foundation.org> wrote:
>
> On Tue, 28 May 2024 at 22:52, kernel test robot <oliver.sang@...el.com> wrote:
> >
> > kernel test robot noticed a -7.4% regression of filebench.sum_operations/s on:
> >
> > commit: 681ce8623567 ("vfs: Delete the associated dentry when deleting a file")
>
> Well, there we are. I guess I'm reverting this, and we're back to the
> drawing board for some of the other alternatives to fixing Yafang's
> issue.

Hi Linus,

I just checked the test case webproxy.f[0], which triggered the regression.

This test case follows a deletefile-createfile pattern, as shown below:

  - flowop deletefile name=deletefile1, filesetname=bigfileset
  - flowop createfile name=createfile1, filesetname=bigfileset, fd=1

It seems that this pattern is causing the regression. As we discussed
earlier, my patch might negatively impact this delete-create pattern.
The question is whether this scenario is something we need to address.
Perhaps it only occurs in this specific benchmark and doesn't
represent a real-world workload.

[0] https://github.com/filebench/filebench/blob/master/workloads/webproxy.f

>
> Al, did you decide on what approach you'd prefer?
>
>                      Linus



-- 
Regards
Yafang