Re: GEOM error

On Wed, Nov 23, 2005 at 10:26:01PM +0100, Bernd Walter wrote:
> On Wed, Nov 23, 2005 at 10:01:56PM +0100, Christian Brueffer wrote:
> > On Tue, Nov 22, 2005 at 02:11:23PM -0800, Doug White wrote:
> > > On Sun, 20 Nov 2005, Christian Brueffer wrote:
> > > 
> > > > On Sat, Nov 19, 2005 at 04:22:11PM -0800, Doug White wrote:
> > > > > On Tue, 15 Nov 2005, Ulf Kieber wrote:
> > > > >
> > > > > > Re,
> > > > > >
> > > > > > on a 6.0-RELEASE I receive the following error since I tried restoring
> > > > > > a large dump
> > > > > >
> > > > > > Nov 14 12:30:11 nexus kernel: g_vfs_done():da1s1d.bde[WRITE(offset=72350695424, length=131072)]error = 1
> > > > > >
> > > > > > Besides that, no other errors are logged, especially no SCSI errors.
> > > > > > The problem persists even after the restore has completed.
> > > > >
> > > > > errno 1 is EPERM ("Operation not permitted") and is generally returned if
> > > > > you attempt to write somewhere you're not allowed to. Considering the
> > > > > offset is near the end of the disk, GBDE may be trying to prevent you from
> > > > > overwriting metadata blocks at the end of the partition. How or why
> > > > > restore(8) would be writing there I'm not sure.
> > > > >
> > > > > A SCSI error would return as errno 5 (EIO, "Input/output error").
> > > > >
> > > >
> > > > I get the same messages with my external USB drive from time to time
> > > > (interestingly also GBDE encrypted).
> > > >
> > > > Nov 20 02:03:30 haakonia kernel: g_vfs_done():da3s2c.bde[WRITE(offset=383341297664, length=65536)]error = 1
> > > >
> > > > The message repeats every 30 seconds and trying to unmount the file
> > > > system fails.  When I try to shut the system down, the message appears
> > > > n > 50 times followed by a panic.
> > > >
> > > > Is it possible that the system tries to write on a bad sector and
> > > > consequently fails (provided that the on-disk sector remapping also
> > > > fails)?
> > > 
> > > You would get a SCSI error in that case since usb storage is attached
> > > through CAM.
> > > 
> > 
> > Ok.  Any suggestion on where to go from here?  I can trigger this pretty
> > reliably on the drive.  Just have to copy enough data around.
> 
> I don't know if the offset is in Bytes or sector, but even in case of
> Bytes the second example is at 714G, which is obviously out of disk
> size unless you are using multiple drives, which is not likely for an
> USB drive.
> The first one is at 134G, which is also very high, the owner should
> compare it with the physical disk size.
> So Doug is absolutely correct, the access isn't allowed, because out
> of range.
> I would say a corrupted FS, partition table or something like that.
> Don't know why - maybe GBDE's fault, but could very well any other
> reason for data corruption, faulty RAM, etc...
> 

In my case it is a 400GB PATA drive in an external enclosure.  It only
happens with this drive, tried it on two boxes (connection via USB or
FireWire doesn't make a difference either).

- Christian

-- 
Christian Brueffer	chris_at_unixpages.org	brueffer_at_FreeBSD.org
GPG Key:	 http://people.freebsd.org/~brueffer/brueffer.key.asc
GPG Fingerprint: A5C8 2099 19FF AACA F41B  B29B 6C76 178C A0ED 982D