linux-ext4 - Re: [PATCH v3 05/13] generic/1226: Add atomic write test using fio crc check verifier

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

Message-ID: <aI205U8Afi7tALyr@li-dc0c254c-257c-11b2-a85c-98b6c1322444.ibm.com>
Date: Sat, 2 Aug 2025 12:19:09 +0530
From: Ojaswin Mujoo <ojaswin@...ux.ibm.com>
To: John Garry <john.g.garry@...cle.com>
Cc: "Darrick J. Wong" <djwong@...nel.org>, Zorro Lang <zlang@...hat.com>,
        fstests@...r.kernel.org, Ritesh Harjani <ritesh.list@...il.com>,
        tytso@....edu, linux-xfs@...r.kernel.org, linux-kernel@...r.kernel.org,
        linux-ext4@...r.kernel.org
Subject: Re: [PATCH v3 05/13] generic/1226: Add atomic write test using fio
 crc check verifier

On Fri, Aug 01, 2025 at 09:23:46AM +0100, John Garry wrote:
> On 01/08/2025 07:41, Ojaswin Mujoo wrote:
> > Got it, I think I can make this test work for ext4 only but then it might
> > be more appropriate to run the fio tests directly on atomic blkdev and
> > skip the FS, since we anyways want to focus on the storage stack.
> > 
> 
> testing on ext4 will prove also that the FS and iomap behave correctly in
> that they generate a single bio per atomic write (as well as testing the
> block stack and below).

Okay, I think we are already testing those in the ext4/061 ext4/062
tests of this patchset. Just thought blkdev test might be useful to keep
in generic. Do you see a value in that or shall I just drop the generic
overlapping write tests?

Also, just for the records, ext4 passes the fio tests ONLY because we use
the same io size for all threads. If we happen to start overlapping
RWF_ATOMIC writes with different sizes that can get torn due to racing
unwritten conversion. 

> 
> > > > I'll try to check if we can modify the tests to write on non-overlapping
> > > > ranges in a file.
> > > JFYI, for testing SW-based atomic writes on XFS, I do something like this. I
> > > have multiple threads each writing to separate regions of a file or writing
> > > to separate files. I use this for power-fail testing with my RPI. Indeed, I
> > > have also being using this sort of test in qemu for shutting down the VM
> > > when fio is running - I would like to automate this, but I am not sure how
> > > yet.
> > > 
> > > Please let me know if you want further info on the fio script.
> > Got it, thanks for the insights. I was thinking of something similar now
> > where I can modify the fio files of this test to write on non
> > overlapping ranges in the same file. The only doubt i have right now is
> > that when I have eg, numjobs=10 filesize=1G, how do i ensure each job
> > writes to its own separate range and not overlap with each other.
> > 
> > I saw the offset_increment= fio options which might help, yet to try it
> > out though. If you know any better way please do share.
> 
> Yeah, so I use something like:
> --numjobs=2 --offset_align=0 --offset_increment=1M --size=1M

Got it, thanks!
ojaswin

> 
> Thanks,
> John
>