lkml.org 
[lkml]   [1999]   [Jul]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: file corruption 2.3.11 pre5

David,

It would be helpful, in my opinion, to print out some additionnal
information from the place the kernel detects the wrong access.

Basically, the problem could be due to some FS address (indirect pointer)
taken from a corrupted block or the buffer header having been corrupted.
The current messages display some arithmetic that involves both bh->b_size
and bh->b_rsector and no more.

The number of 512 bytes sectors in a block is (b_size>>9) and b_rsector is
calculated as b_blocknr*(b_size>>9).

So, the #sector beyond the last block we want is:
(b_blocknr+1)*(b_size>>9).

The message display half this value (1K block scaled value), that, by the
way, appears to me not being what we 'want' but just beyond what we
want. ;-)

It would interesting, in my opinion, to also print out the following
fields:

- bh->b_size
- bh->b_blocknr
- bh->b_rsector

Gérard.


On Sun, 18 Jul 1999, David Fries wrote:

> I'm having file system corruption on 2.3.11 pre5, don't ask me to try to
> reproduce, I'm just trying to salvage what I have and figure out the
> damange.
>
> Somehow the system logs seemed to be ok, compared to the other stuff
> they are worthless to me.
>
> It almost seemed like files that were modified over nfs were corrupted, but
> I'm not positive about that. The files look to be about the correct size,
> they are just filled with nulls. I had rebooted the nfs server to the
> 2.3.11 pre5 kernel without re-booting the clients.
>
> I know there are three scripts I had modified about that time were nuked,
> they weren't big, I just didn't want to re-write them. One of my fvwm2
> config files that is under cvs so I just lost the changes, netscape's
> preferences, netscape's bookmarks (had a 2 day old backup) it had changed,
> some gimp configuration files
>
> pages and pages of ...
> Jul 17 23:11:07 AeroSpace kernel: attempt to access beyond end of device
> Jul 17 23:11:07 AeroSpace kernel: 08:02: rw=0, want=740456819,
> limit=8827904
> Jul 17 23:11:07 AeroSpace kernel: attempt to access beyond end of device
> Jul 17 23:11:07 AeroSpace kernel: 08:02: rw=0, want=976232993,
> limit=8827904
> Jul 17 23:11:07 AeroSpace kernel: attempt to access beyond end of device
> Jul 17 23:11:07 AeroSpace kernel: 08:02: rw=0, want=2016489525,
> limit=8827904
>
> one of,
> Jul 17 23:11:27 AeroSpace kernel: e_blocks: Freeing blocks not in datazone
> - block = 740455284, count = 1
>
> there were about 156 messages like this these,
> Jul 17 23:11:27 AeroSpace kernel: EXT2-fs error (device sd(8,2)):
> ext2_free_blocks: Freeing blocks not in datazone - block = 573072997, count
> = 1
> Jul 17 23:11:27 AeroSpace kernel: EXT2-fs error (device sd(8,2)):
> ext2_free_blocks: Freeing blocks not in datazone - block = 1802723700,
> count = 1
> Jul 17 23:11:27 AeroSpace kernel: EXT2-fs error (device sd(8,2)):
> ext2_free_blocks: Freeing blocks not in datazone - block = 779247970, count
> = 1
>
> Server is a Dual Pentium was running 2.3.11 pre5 ncr53c876 ultrawide IBM
> disk,
> two nfs clients, the one I work from I think was running 2.3.5 at the
> time, a 486 running 2.3.11 pre 4 I think,
>
> no kernel patches, I am running the kernel nfs with the version 1.4.2 from
> H.J. Lu. It might not have anything to do with nfs, just my setup relies
> heavily on nfs. I did a fine on files newer than when I figured there
> would have been the problem and all the files that were corrupted (mostly
> listed above, if that is it I concider myself luckey), all of the corrupted
> files I think would have been modified under nfs.
>
> When e2fsck ran it did not seem to have hardly any errors, I think it went
> through all five levels without errors and had some wrong number of
> something at the last stage.
>
> Time to start looking into someway of backup, any suggestions? I've been
> thinking of CD-R/RW.
>
> --
> +---------------------------------+
> | David Fries |
> | dfries@mail.win.org |
> +---------------------------------+
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/
>
>


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:53    [W:0.068 / U:0.260 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site