2003-06-04 16:01:30

by Jeremy Salch

[permalink] [raw]
Subject: Access past end of device

I'm using a dell powerall web 120 with scsi drives installed
Using redhat 7.2 with the 2.4.18-19.7.x kernel installed



And when using badblocks on the swap partition I get







Attempt to access beyond end of device

08:06: rw=0, want=1044196, limit=1044193

1044192

done

Pass completed, 1 bad blocks found.





And fdisk reports there to be 1044193+ blocks in the partition ?





I wish to be personally CC'ed any answer / comments..



Thanks a million




2003-06-04 16:50:14

by Stewart Smith

[permalink] [raw]
Subject: Re: Access past end of device

On Wed, Jun 04, 2003 at 11:13:02AM -0500, Jeremy Salch wrote:
> I'm using a dell powerall web 120 with scsi drives installed
> Using redhat 7.2 with the 2.4.18-19.7.x kernel installed
>
> Attempt to access beyond end of device
> 08:06: rw=0, want=1044196, limit=1044193
> 1044192
> Pass completed, 1 bad blocks found.
> And fdisk reports there to be 1044193+ blocks in the partition ?

Sounds like the partition map is a bit incorrect.

Try running badblocks on the drive itself (/dev/hda, not hda1).
If the problem disappears then it's with the partition map (i'd guess).

- stew

2003-06-04 17:03:00

by Jeremy Salch

[permalink] [raw]
Subject: RE: Access past end of device

Ok I'll give that a try. One other question. If the partitiontable was bad
wouldn't fdisk report a incorrect size for the partitions that are having
the problems ?

Also the system freezes when under a heavy load without giving any error
output. What it does do is report something like 600000% idle time in top


-----Original Message-----
From: Stewart Smith [mailto:[email protected]]
Sent: Wednesday, June 04, 2003 12:02 PM
To: Jeremy Salch
Cc: [email protected]
Subject: Re: Access past end of device


On Wed, Jun 04, 2003 at 11:13:02AM -0500, Jeremy Salch wrote:
> I'm using a dell powerall web 120 with scsi drives installed Using
> redhat 7.2 with the 2.4.18-19.7.x kernel installed
>
> Attempt to access beyond end of device
> 08:06: rw=0, want=1044196, limit=1044193
> 1044192
> Pass completed, 1 bad blocks found.
> And fdisk reports there to be 1044193+ blocks in the partition ?

Sounds like the partition map is a bit incorrect.

Try running badblocks on the drive itself (/dev/hda, not hda1). If the
problem disappears then it's with the partition map (i'd guess).

- stew

2003-06-04 17:12:20

by Jeremy Salch

[permalink] [raw]
Subject: RE: Access past end of device

I did a fdisk on the device /dev/sda and it found no bad blocks.

I was given the suggestion from one person it sounds like a kernel bug since
redhat patches their kernels soo much.


-----Original Message-----
From: Stewart Smith [mailto:[email protected]]
Sent: Wednesday, June 04, 2003 12:02 PM
To: Jeremy Salch
Cc: [email protected]
Subject: Re: Access past end of device


On Wed, Jun 04, 2003 at 11:13:02AM -0500, Jeremy Salch wrote:
> I'm using a dell powerall web 120 with scsi drives installed Using
> redhat 7.2 with the 2.4.18-19.7.x kernel installed
>
> Attempt to access beyond end of device
> 08:06: rw=0, want=1044196, limit=1044193
> 1044192
> Pass completed, 1 bad blocks found.
> And fdisk reports there to be 1044193+ blocks in the partition ?

Sounds like the partition map is a bit incorrect.

Try running badblocks on the drive itself (/dev/hda, not hda1). If the
problem disappears then it's with the partition map (i'd guess).

- stew

2003-06-04 18:28:04

by Stewart Smith

[permalink] [raw]
Subject: Re: Access past end of device

On Wed, Jun 04, 2003 at 12:23:50PM -0500, Jeremy Salch wrote:
> I did a fdisk on the device /dev/sda and it found no bad blocks.

That's a good thing - be greatful :)

> I was given the suggestion from one person it sounds like a kernel bug since
> redhat patches their kernels soo much.

Could be, try a kernel.org kernel and see if the problem persists. If it goes away, file a bug report with redhat. I would think fdisk would come up with an error on a corrupt partition map, so if it doesn't then ur probably looking at some oddity... try the kernel.org kernel :)

- stew