strange badblocks problem

marnixk lfs at mkaart.net
Thu Jan 15 12:52:52 PST 2004


Bill's LFS Login wrote:

> Leaving only the commonality of the OS and utilities. That's good, I
> think.

I think we can narrow this down to the utilities (?). I have found a debian
(2.4.24 kernel) machine somewhere at work and i ran badblocks
-sv /dev/hdaxx on each partition (mounted FS, production server) followed
by dd if=/dev/hdaxx of=/dev/null. And guess what... Same behaviour on one
of the partitions. 

badblocks output:
tux:~# badblocks -v /dev/hda6
Checking for bad blocks in read-only mode
>From block 0 to 746991
746988
746989
746990
Pass completed, 3 bad blocks found

and the logs:
Jan 15 18:37:37 tux kernel: attempt to access beyond end of device
Jan 15 18:37:37 tux kernel: 03:06: rw=0, want=746992, limit=746991
Jan 15 18:37:37 tux kernel: attempt to access beyond end of device
Jan 15 18:37:37 tux kernel: 03:06: rw=0, want=746992, limit=746991
Jan 15 18:37:37 tux kernel: attempt to access beyond end of device
Jan 15 18:37:37 tux kernel: 03:06: rw=0, want=746992, limit=746991
Jan 15 18:37:37 tux kernel: attempt to access beyond end of device
Jan 15 18:37:37 tux kernel: 03:06: rw=0, want=746992, limit=746991

dd output:
tux:~# dd if=/dev/hda6 of=/dev/null       
dd: reading `/dev/hda6': Input/output error
1493976+0 records in
1493976+0 records out
764915712 bytes transferred in 93.036158 seconds (8221704 bytes/sec)

and similar logs...

This Debian machine has never even been near a Gentoo CD, so the problem is
not there. It is my guess that many more people actually have this
"problem", but just haven't noticed it yet. I think that this only happens
on partitions where "the math" doesn't work, although I haven't checked
them all. So all your explanations about blocksizes and sectors in your
previous posts are probably right on target after all!

> That is an option. If there is *nothing* critical, it could be
> reasonable. But my presumption would be that whatever is causing the
> *visible* problem may also cause some invisible (for *now*) ones. I
> would think it risky *unless* the systemand it's data could be
> considered expendable.

I think eventually (after all this troubleshooting) i will try to carefully
adjust partitions to make the "math" work and then i will probably not have
this problem anymore. Yet, the problem still remains in the tools or the
kernel somewhere, so the hunt is still open...

>> I still have to do the "boot with Knoppix" test in my previous post and i
>> will post my findings here.

Boot with knoppix did not provide different results, same errors. On one
partition the errors even occured before i mounted it. I'm beginning to
think this mount/unmount stuff really doesn't matter much after all.
 
I will post my other findings as a reply to your other post...

Marnix



More information about the lfs-support mailing list