xfs
[Top] [All Lists]

Re: Unable to handle kernel NULL pointer dereference at virtual address

To: Denis Nikiforov <denis@xxxxxxxx>
Subject: Re: Unable to handle kernel NULL pointer dereference at virtual address 00000000
From: Timothy Shimmin <tes@xxxxxxx>
Date: Wed, 21 Jun 2006 11:08:24 +1000
Cc: xfs@xxxxxxxxxxx
In-reply-to: <87hd2fsvsl.fsf@xxxxxxxxxxxxxxx>
Organization: SGI
References: <87hd2fsvsl.fsf@xxxxxxxxxxxxxxx>
Sender: xfs-bounce@xxxxxxxxxxx
User-agent: KMail/1.8.2
Hi Denis,

On Wednesday 21 June 2006 1:15 am, Denis Nikiforov wrote:
>
> I'd upgraded my computer from Celeron P4 to Athlon 64. And xfs partition
> couldn't be mounted now ;( How can I fix it?
>
> .............
>
> | XFS mounting filesystem hdb3
> | Starting XFS recovery on filesystem: hdb3 (logdev: internal)
> | Unable to handle kernel NULL pointer dereference at virtual address
> ...............
> |  [<f8c5290e>] xfs_buf_offset+0x35/0x3a [xfs]
> |  [<f8c3eb19>] xlog_recover_do_inode_trans+0x14a/0x718 [xfs]

Your problem is due to a bug in xfs.
A kernel from 2006/05/24 onwards will be able to support this filesystem.
The problem is that the ondisk format for a few log items in the ondisk log is 
different on 32 bits versus 64 bit systems. With the newer XFS it can decode 
both forms.
What can you do?
You can either mount and unmount the FS on a 32 bit os so that the log is 
clean and no recovery is needed.
Or you can get a new version of XFS.
(In the future, xfs_repair will probably be able to replay the log but not 
yet).

--Tim


<Prev in Thread] Current Thread [Next in Thread>