From: Christoph Hellwig <hch@infradead.org>
Subject: Re: [PATCHSET v3.1 0/7] data integrity: Stabilize pages during
 writeback for various fses
Date: Tue, 10 May 2011 13:03:39 -0400
Message-ID: <20110510170339.GA27538@infradead.org>
References: <20110509230318.19566.66202.stgit@elm3c44.beaverton.ibm.com>
 <87tyd31fkc.fsf@devron.myhome.or.jp>
 <20110510133603.GA5823@infradead.org>
 <874o524q9h.fsf@devron.myhome.or.jp>
 <20110510144939.GI4402@quack.suse.cz>
 <87aaeur31x.fsf@devron.myhome.or.jp>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Jan Kara <jack@suse.cz>, Christoph Hellwig <hch@infradead.org>,
	"Darrick J. Wong" <djwong@us.ibm.com>,
	Theodore Tso <tytso@mit.edu>,
	Alexander Viro <viro@zeniv.linux.org.uk>,
	Jens Axboe <axboe@kernel.dk>,
	"Martin K. Petersen" <martin.petersen@oracle.com>,
	Jeff Layton <jlayton@redhat.com>,
	Dave Chinner <david@fromorbit.com>,
	linux-kernel <linux-kernel@vger.kernel.org>,
	Dave Hansen <dave@linux.vnet.ibm.com>, linux-mm@kvack.org,
	Chris Mason <chris.mason@oracle.com>,
	Joel Becker <jlbec@evilplan.org>,
	linux-scsi <linux-scsi@vger.kernel.org>,
	linux-fsdevel <linux-fsdevel@vger.kernel.org>,
	linux-ext4@vger.kernel.org, Mingming Cao <mcao@us.ibm.com>
To: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Return-path: <linux-ext4-owner@vger.kernel.org>
Received: from bombadil.infradead.org ([18.85.46.34]:40446 "EHLO
	bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1751688Ab1EJRDt (ORCPT
	<rfc822;linux-ext4@vger.kernel.org>); Tue, 10 May 2011 13:03:49 -0400
Content-Disposition: inline
In-Reply-To: <87aaeur31x.fsf@devron.myhome.or.jp>
Sender: linux-ext4-owner@vger.kernel.org
List-ID: <linux-ext4.vger.kernel.org>

On Wed, May 11, 2011 at 12:24:58AM +0900, OGAWA Hirofumi wrote:
> > under writeback and when it's freed no writeback is started.
> 
> Sure for data -> data reallocated case. metadata -> data/metadata is
> still there.

That's usually handled differently.  For XFS take a look at the
xfs_alloc_busy_* function.  For 2.6.40 they've been mostly rewritten
to rarely wait for the reuse but instead avoid busy blocks.  But that's
a real data integrity issue even without stable pages for I/O.