Received: (from majordomo@localhost) by oss.sgi.com (8.11.3/8.11.3) id f4FKOLU20639 for linux-xfs-outgoing; Tue, 15 May 2001 13:24:21 -0700 Received: from Cantor.suse.de (ns.suse.de [213.95.15.193]) by oss.sgi.com (8.11.3/8.11.3) with ESMTP id f4FKOKF20636 for ; Tue, 15 May 2001 13:24:20 -0700 Received: from Hermes.suse.de (Hermes.suse.de [213.95.15.136]) by Cantor.suse.de (Postfix) with ESMTP id 578721E25F; Tue, 15 May 2001 22:24:19 +0200 (MEST) Date: Tue, 15 May 2001 22:23:41 +0200 From: Andi Kleen To: Tim Clymo Cc: linux-xfs@oss.sgi.com Subject: Re: Might have found a bug... Message-ID: <20010515222341.A30527@gruyere.muc.suse.de> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: ; from tim_clymo@hotmail.com on Tue, May 15, 2001 at 08:15:14PM -0000 Sender: owner-linux-xfs@oss.sgi.com Precedence: bulk On Tue, May 15, 2001 at 08:15:14PM -0000, Tim Clymo wrote: > Editing a plain text file for example with vi, saving the file and > doing an immediate reset - I could understand if the file was back to > its original state before the edit was "written", but my experience is > that instead it is useless junk. As I understand it the problem is that XFS/Linux doesn't support unwritten extents yet (unlike Irix XFS). An unwritten extent would tell XFS to zero the data on crash recovery, if it wasn't turned into an written extent by a finished flush. Without them recovery has no way to know if there is old data or new data in the extent and has to leave it alone, so you likely see old data if you hit reset fast. See the old XFS design documentation somewhere on their website for more details. -Andi