From: Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH] improve jbd fsync batching
Date: Tue, 4 Nov 2008 01:12:42 -0800
Message-ID: <20081104011242.cc767dba.akpm@linux-foundation.org>
References: <20081028201614.GA21600@unused.rdu.redhat.com>
	<20081103122729.60582692.akpm@linux-foundation.org>
	<20081104052428.GP3184@webber.adilger.int>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Cc: Josef Bacik <jbacik@redhat.com>, linux-kernel@vger.kernel.org,
	linux-fsdevel@vger.kernel.org, linux-ext4@vger.kernel.org,
	rwheeler@redhat.com
To: Andreas Dilger <adilger@sun.com>
Return-path: <linux-fsdevel-owner@vger.kernel.org>
In-Reply-To: <20081104052428.GP3184@webber.adilger.int>
Sender: linux-fsdevel-owner@vger.kernel.org
List-Id: linux-ext4.vger.kernel.org

On Mon, 03 Nov 2008 22:24:28 -0700 Andreas Dilger <adilger@sun.com> wrote:

> On Nov 03, 2008  12:27 -0800, Andrew Morton wrote:
> > On Tue, 28 Oct 2008 16:16:15 -0400
> > Josef Bacik <jbacik@redhat.com> wrote:
> > > +		spin_lock(&journal->j_state_lock);
> > > +		commit_time = journal->j_average_commit_time;
> > > +		spin_unlock(&journal->j_state_lock);
> > 
> > OK, the lock is needed on 32-bit machines, I guess.
> 
> Should we pessimize the 64-bit performance in that case, for 32-bit
> increasingly rare 32-bit platforms?

In general no.

But spinlocks also do memory ordering stuff on both 32- and 64-bit
machines.  Introducing differences there needs thinking about.

In this case it's fsync which is going to be monster slow anyway.