xfs
[Top] [All Lists]

Re: [PATCH, -stable RFC] xfs: fix periodic log flushing

To: Ben Myers <bpm@xxxxxxx>
Subject: Re: [PATCH, -stable RFC] xfs: fix periodic log flushing
From: Dave Chinner <david@xxxxxxxxxxxxx>
Date: Wed, 19 Dec 2012 18:04:07 +1100
Cc: Andre Noll <maan@xxxxxxxxxxxxxxx>, xfs@xxxxxxxxxxx
In-reply-to: <20121219033013.GH27055@xxxxxxx>
References: <20121218221616.GH15182@dastard> <20121218224022.GJ31273@xxxxxxxxxxxxxxx> <20121218232533.GJ15182@dastard> <20121219002145.GK31273@xxxxxxxxxxxxxxx> <20121219033013.GH27055@xxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Tue, Dec 18, 2012 at 09:30:13PM -0600, Ben Myers wrote:
> Looks to me like 3.4 doesn't have the problem that Dave is trying to address
> here because it doesn't check for MS_ACTIVE in xfs_sync_worker.  You're 
> already
> good to go.

Right.

> Dave, what you've done makes sense b/c MS_ACTIVE is set after mount time and
> cleared at unmount.  This is the time during which we want the sync worker to
> be running.  I do think that the check is racy:  The sync worker can check the
> flag and continue at snail's pace, and there is nothing to prevent unmount
> clearing the flag and wiping out the structures used by the sync worker.

I'm fully aware of that possibility. IMO, the only thing that matters
here is fixing the bug that causes data loss. Trading guaranteed
data loss on a crash for a hard-to-hit unmount race condition
that's never been seen outside a test lab is a no-brainer, IMO.

Cheers,

Dave.
-- 
Dave Chinner
david@xxxxxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>