From: Theodore Tso <tytso@mit.edu>
Subject: Re: [PATCH 1/4] ext3: Fix ext3_dx_readdir hash collision handling
Date: Wed, 1 Oct 2008 18:37:55 -0400
Message-ID: <20081001223755.GL10080@mit.edu>
References: <1221319971-29879-1-git-send-email-tytso@mit.edu> <20080917121943.904b66b6.akpm@linux-foundation.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: linux-kernel@vger.kernel.org, eugene@ibrix.com, msnitzer@ibrix.com,
	linux-ext4@vger.kernel.org
To: Andrew Morton <akpm@linux-foundation.org>
Content-Disposition: inline
In-Reply-To: <20080917121943.904b66b6.akpm@linux-foundation.org>
Sender: linux-ext4-owner@vger.kernel.org

On Wed, Sep 17, 2008 at 12:19:43PM -0700, Andrew Morton wrote:
> On Sat, 13 Sep 2008 11:32:48 -0400
> "Theodore Ts'o" <tytso@MIT.EDU> wrote:
> 
> > This fixes a bug where readdir() would return a directory entry twice
> > if there was a hash collision in an hash tree indexed directory.
> 
> That sounds like a serious problem, but given the amount of time it
> took to turn up, I guess it's pretty rare.
> 
> What are your thoughts regarding a 2.6.27 merge for this?  2.6.26.x? 
> 2.6.25.x?  ...
> 

It's not a regression, so per Linus's request that at this point we're
too late for anything other than regression fixes, I had assumed that
it should be pushed for 2.6.28, and then go into the various stable
trees.  It's true that it took quite a while for people to notice,
probably because many programs won't notice if readdir() returns a
directory entry twice.

BTW, this hasn't hit -mm yet, and I've got a number of ext3 patches
that don't appear to have hit -mm, including one that was authored by
Linus.  Should I create a git tree or a quilt series if that would
make things easier for you?

	      	   	   	      	- Ted