[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20251215194118.GN7725@frogsfrogsfrogs>
Date: Mon, 15 Dec 2025 11:41:18 -0800
From: "Darrick J. Wong" <djwong@...nel.org>
To: Christoph Hellwig <hch@....de>
Cc: Zorro Lang <zlang@...nel.org>, Anand Jain <asj@...nel.org>,
Filipe Manana <fdmanana@...e.com>, fstests@...r.kernel.org,
linux-ext4@...r.kernel.org, linux-xfs@...r.kernel.org
Subject: Re: [PATCH 01/13] dmflakey: override SCRATCH_DEV in _init_flakey
On Mon, Dec 15, 2025 at 06:29:13AM +0100, Christoph Hellwig wrote:
> On Fri, Dec 12, 2025 at 12:11:42PM -0800, Darrick J. Wong wrote:
> > Ok, so _init_flakey still sets FLAKEY_TABLE_ERROR...
> >
> > > _dmsetup_create $FLAKEY_NAME --table "$FLAKEY_TABLE" || \
> > > _fatal "failed to create flakey device"
> > >
>
> > > # no error will be injected
> > > _init_flakey
> >
> > ...but won't _init_flakey clobber the value of FLAKEY_TABLE_ERROR set
> > by make_xfs_scratch_flakey_table?
>
> I think so. But nothing really changed here except for a variable
> name, so the whatever clobber was there before is the same and apparently
> worked?
I suppose it could work (I haven't tried), but
make_xfs_scratch_flakey_table redefines FLAKEY_TABLE_ERROR so that
internal log writes continue to work but everything else doesn't,
because it's testing that write failures during AIL push operations
don't set us up to hang the system. It's not testing iclog write
failures themselves.
--D
Powered by blists - more mailing lists