From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from cuda.sgi.com (cuda1.sgi.com [192.48.157.11]) by oss.sgi.com (8.14.3/8.14.3/SuSE Linux 0.8) with ESMTP id q0OHUlmq058644 for ; Tue, 24 Jan 2012 11:30:47 -0600 Received: from b.mx.filmlight.ltd.uk (b.mx.filmlight.ltd.uk [77.107.81.251]) by cuda.sgi.com with SMTP id 2JvvZlFGhi6wsFTb for ; Tue, 24 Jan 2012 09:30:44 -0800 (PST) Subject: Re: How to fix bad superblock or xfs_repair: error - read only 0 of 512 bytes From: Roger Willcocks In-Reply-To: <4F1ED4DF.40907@sandeen.net> References: <20120123043117.GH15102@dastard> <92CDAC52-1BA1-43A8-8470-0E2DCD916610@unixhosts.org> <4F1E3BD0.7080604@sandeen.net> <1327414341.24535.582.camel@montana.filmlight.ltd.uk> <4D396236-D516-4948-8F43-1AED88539BC4@unixhosts.org> <4F1ED341.4010108@sandeen.net> <4F1ED4DF.40907@sandeen.net> Date: Tue, 24 Jan 2012 17:25:31 +0000 Message-Id: <1327425931.24535.610.camel@montana.filmlight.ltd.uk> Mime-Version: 1.0 List-Id: XFS Filesystem from SGI List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: xfs-bounces@oss.sgi.com Errors-To: xfs-bounces@oss.sgi.com To: Eric Sandeen Cc: xfs-oss , Christian Kildau On Tue, 2012-01-24 at 09:57 -0600, Eric Sandeen wrote: > On 1/24/12 9:52 AM, Christian Kildau wrote: > > On Jan 24, 2012, at 4:50 PM, Eric Sandeen wrote: > > ... > >> Can you run the hdparm command Roger suggested on your original hard drive, please? > > > > Sure, here it is: > > > > /dev/sde: > > max sectors = 2930275055/2930277168, HPA is enabled > ^^^^^^^^^^^^^^ > > ding ding ding, we have a winner. > > 2930277168-2930275055 = 2113 which is about how much xfs tried to read past the end. > > Something about the ubuntu upgrade messed with your disk. > > I'd press them very hard to investigate & resolve that. You can probably use hdparm > to remove the HPA and get your space back but this is beyond my expertise & > familiarity. It'd be interesting to know what is _in_ the HPA area first. > Kernel commit d8d9129ea28e2177749627c82962feb26e8d11e9 added code to notice if a disk partition extends into the hidden area and 'unlocks' the hidden area (uses the full disk capacity) if necessary. But if there's no partition table (the entire disk is formatted as a single filesystem) the heuristic can't work. http://www.spinics.net/lists/linux-ide/msg40244.html suggests that Ubuntu no longer always unlocks, but uses the heuristic above. In the same message Alan Cox says Ubuntu 'made a bad mistake on that one.' -- Roger Willcocks _______________________________________________ xfs mailing list xfs@oss.sgi.com http://oss.sgi.com/mailman/listinfo/xfs