Date: Thu, 14 Aug 2008 12:30:54 +1000
From: Dave Chinner <david@fromorbit.com>
To: Daniel Walker <dwalker@mvista.com>
Cc: xfs@oss.sgi.com, linux-kernel@vger.kernel.org, matthew@wil.cx
Subject: Re: [PATCH 4/6] Replace inode flush semaphore with a completion
Message-ID: <20080814023054.GH6119@disturbed>
Mail-Followup-To: Daniel Walker <dwalker@mvista.com>, xfs@oss.sgi.com,
	linux-kernel@vger.kernel.org, matthew@wil.cx
References: <1214556284-4160-1-git-send-email-david@fromorbit.com> <1214556284-4160-5-git-send-email-david@fromorbit.com> <1218597077.6166.15.camel@dhcp32.mvista.com> <20080813075057.GZ6119@disturbed> <1218641641.6166.32.camel@dhcp32.mvista.com> <20080814001938.GC6119@disturbed> <1218677690.6166.51.camel@dhcp32.mvista.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1218677690.6166.51.camel@dhcp32.mvista.com>
User-Agent: Mutt/1.5.18 (2008-05-17)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1509
Lines: 36

On Wed, Aug 13, 2008 at 06:34:49PM -0700, Daniel Walker wrote:
> On Thu, 2008-08-14 at 10:19 +1000, Dave Chinner wrote:
> > *However*, given that we already have this exact state in the
> > completion itself, I see little reason for adding the additional
> > locking overhead and the complexity of race conditions of keeping
> > this state coherent with the completion. Modifying the completion
> > API slightly to export this state is the simplest, easiest solution
> > to the problem....
> > 
> 
> I'm not suggesting anything concrete at this point, I'm just thinking
> about it.
> 
> If you assume that most of the time your doing async flushing, you
> wouldn't often need to do blocking on the completion .. Another way of
> doing it would be drop the completion most of the time, and just use the
> flag.

I don't think that will work, either - the internal completion
counter must stay in sync so there must be a wait_for_completion()
call for every complete() call. If we make the wait_for_completion()
conditional, then we have to do the same thing for the complete()
call, and that introduces complexity and potential races because the
external flag and the completion cannot be updated atomically....

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/