[Bug 216486] [xfstests generic/447] xfs_scrub always complains fs corruption

[bugzilla-daemon@kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=216486

--- Comment #2 from Zorro Lang (zlang@redhat.com) ---
(In reply to Darrick J. Wong from comment #1)
> On Wed, Sep 14, 2022 at 08:12:56AM +0000, bugzilla-daemon@kernel.org wrote:
> > https://bugzilla.kernel.org/show_bug.cgi?id=216486
> > 
> >             Bug ID: 216486
> >            Summary: [xfstests generic/447] xfs_scrub always complains  fs
> >                     corruption
> >            Product: File System
> >            Version: 2.5
> >     Kernel Version: 6.0.0-rc4+
> >           Hardware: All
> >                 OS: Linux
> >               Tree: Mainline
> >             Status: NEW
> >           Severity: normal
> >           Priority: P1
> >          Component: XFS
> >           Assignee: filesystem_xfs@kernel-bugs.kernel.org
> >           Reporter: zlang@redhat.com
> >         Regression: No
> > 
> > Recently xfstests generic/447 always fails[1][2][3] on latest xfs kernel
> with
> > xfsprogs. It's reproducible on 1k blocksize and rmapbt enabled XFS (-b
> > size=1024 -m rmapbt=1). Not sure if it's a kernel bug or a xfsprogs issue,
> or
> > an expected failure.
> 
> It's an expected failure that is one of the many things fixed by the
> online fsck patchset.  The solution I came up with is described here:
> https://djwong.org/docs/xfs-online-fsck-design/#eventual-consistency-vs-
> online-fsck
> 
> The TLDR is that scrub is probably racing with a thread that's in the
> middle of doing a file mapping change that involves both an rmap and a
> refcount update.  This is possible because we don't hold the AGF buffer
> between work items in a defer ops chain.

Thanks Darrick, for your reply and confirmation!

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.