From: Jan Kara Subject: Re: ext2/3/4 online defrag Date: Mon, 21 May 2007 12:33:19 +0200 Message-ID: <20070521103319.GA29416@atrey.karlin.mff.cuni.cz> References: <1178770799.26695.11.camel@eric-laptop> <20070517161119.GB3230@atrey.karlin.mff.cuni.cz> <1179465114.7891.17.camel@eric-laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-ext4 To: Eric Return-path: Received: from atrey.karlin.mff.cuni.cz ([195.113.31.123]:39586 "EHLO atrey.karlin.mff.cuni.cz" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756569AbXEUKdU (ORCPT ); Mon, 21 May 2007 06:33:20 -0400 Content-Disposition: inline In-Reply-To: <1179465114.7891.17.camel@eric-laptop> Sender: linux-ext4-owner@vger.kernel.org List-Id: linux-ext4.vger.kernel.org > On Thu, 2007-05-17 at 18:11 +0200, Jan Kara wrote: > > But me (and several other people > > independently as I've learnt recently) have written some tools which > > should result in something useful. If you're interested, you can join > > prefetch-devel@googlegroups.com - it's led by one guy who is doing > > defrag and stuff as his google summer of code project. > > Is this different from the ext4/extent-based defrag patch that's been > mentioned on this list? Yes, it is different. In particular, it's offline only tool so far... > > > *An implementation of an ext* filesystem driver can work with any > > > ext2/3/4 filesystem as long as it supports the necessary revision > > > (GOOD_OLD_REV or DYNAMIC_REV) and feature flags set in the filesystem. > > Not sure what you mean here... > > The "ext2 filesystem"/"ext3 filesystem"/"ext4 filesystem" terminology > was confusing to me when I first started reading about them. In my mind, > it implied that those three filesystems were more different than they > actually are. > > I think it would be more accurate to say that they are all essentially > the same filesystem, and that any filesystem driver that can mount a > given filesystem can mount any other ext2/3/4 filesystem of the same > revision with the same feature flags set. > > I was asking for confirmation of this assumption, but I've since found a > lot of really good documentation that has cleared up a lot of things. Yes, basically it's just a question of a feature set. But for example current online defrag from Takashi requires extents, which are not available for ext2 or ext3. Honza -- Jan Kara SuSE CR Labs