reiserfs3 filesystem corruption?

reiserfs3 filesystem corruption?

[leon breedt]

hi,

not sure if i'm going mad here:

[ljb@imryrr]:/home/ljb $ ls -l /usr/bin/sdiff
-rwxr-xr-x  1 root root 24576 Apr 25  2003 /usr/bin/sdiff*

[ljb@imryrr]:/home/ljb $ cp /usr/bin/sdiff .

[ljb@imryrr]:/home/ljb $ ls -l sdiff
-rwxr-xr-x  1 ljb ljb 20480 Aug 29 01:12 sdiff*

[ljb@imryrr]:/home/ljb $ ./sdiff
zsh: bus error  ./sdiff

[ljb@imryrr]:/home/ljb $ /usr/bin/sdiff
/usr/bin/sdiff: missing operand after `/usr/bin/sdiff'
/usr/bin/sdiff: Try `/usr/bin/sdiff --help' for more information.

why would the file shrink by 4096 bytes? 

and notice, the original is still able to run. its just when i try to
copy it, that it gets corrupted somehow.

this occurs for a fair number of random files on this filesystem
(affected files all shrink by 4096 bytes). reiserfsck has no
complaints, and i've also run reiserfsck --rebuild-tree. it occurs to
the same set of files every time.

i ran across this while trying to copy my / filesystem to a new
reiser4 filesystem. first thought it was a reiser4 issue, but it
occurs when copying to an ext3 filesystem as well.

i've run badblocks on the disk, with no errors. the filesystem isn't
mounted with any special options, just "noatime".

kernel: 2.6.9-rc1-mm1
reiserfsck: 3.6.18
reiserfs format: "3.6"

its serial ATA device on ICH5-R, if that makes any difference.

the relevant dmesg output:

ReiserFS: sda7: found reiserfs format "3.6" with standard journal
ReiserFS: sda7: using ordered data mode
ReiserFS: sda7: journal params: device sda7, size 8192, journal first
block 18, max trans len 1024, max batch 900, max commit age 30, max
trans age 30
ReiserFS: sda7: checking transaction log (sda7)
ReiserFS: sda7: Using r5 hash to sort names

any ideas what to try? i'm keeping this filesystem around so i have
the test cases available.

leon

Re: reiserfs3 filesystem corruption?

[Christian Mayrhuber]

On Saturday 28 August 2004 15:27, leon breedt wrote:
> hi,
> 
> not sure if i'm going mad here:
> 
> [ljb@imryrr]:/home/ljb $ ls -l /usr/bin/sdiff
> -rwxr-xr-x  1 root root 24576 Apr 25  2003 /usr/bin/sdiff*
> 
> [ljb@imryrr]:/home/ljb $ cp /usr/bin/sdiff .
> 
> [ljb@imryrr]:/home/ljb $ ls -l sdiff
> -rwxr-xr-x  1 ljb ljb 20480 Aug 29 01:12 sdiff*
> 
> [ljb@imryrr]:/home/ljb $ ./sdiff
> zsh: bus error  ./sdiff
> 
> [ljb@imryrr]:/home/ljb $ /usr/bin/sdiff
> /usr/bin/sdiff: missing operand after `/usr/bin/sdiff'
> /usr/bin/sdiff: Try `/usr/bin/sdiff --help' for more information.
> 
> why would the file shrink by 4096 bytes? 
> 
> and notice, the original is still able to run. its just when i try to
> copy it, that it gets corrupted somehow.
> 
> this occurs for a fair number of random files on this filesystem
> (affected files all shrink by 4096 bytes). reiserfsck has no
> complaints, and i've also run reiserfsck --rebuild-tree. it occurs to
> the same set of files every time.
> 
> i ran across this while trying to copy my / filesystem to a new
> reiser4 filesystem. first thought it was a reiser4 issue, but it
> occurs when copying to an ext3 filesystem as well.
> 
> i've run badblocks on the disk, with no errors. the filesystem isn't
> mounted with any special options, just "noatime".
> 
> kernel: 2.6.9-rc1-mm1
> reiserfsck: 3.6.18
> reiserfs format: "3.6"
> 
> its serial ATA device on ICH5-R, if that makes any difference.
> 
> the relevant dmesg output:
> 
> ReiserFS: sda7: found reiserfs format "3.6" with standard journal
> ReiserFS: sda7: using ordered data mode
> ReiserFS: sda7: journal params: device sda7, size 8192, journal first
> block 18, max trans len 1024, max batch 900, max commit age 30, max
> trans age 30
> ReiserFS: sda7: checking transaction log (sda7)
> ReiserFS: sda7: Using r5 hash to sort names
> 
> any ideas what to try? i'm keeping this filesystem around so i have
> the test cases available.
> 
> leon
> 

This is a bug in 2.6.9-rc1-mm1. See:
http://marc.theaimsgroup.com/?l=linux-kernel&m=109360433916523&w=2

To fix it you have to revert these patches:

=== START Andrew Morton ===
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.9-rc1/2.6.9-rc1-mm1/ 
\
broken-out/re-fix-pagecache-reading-off-by-one-cleanup.patch

and then

ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.9-rc1/2.6.9-rc1-mm1/ 
\
broken-out/re-fix-pagecache-reading-off-by-one.patch
=== END Andrew Morton ===
Full email at: 
http://marc.theaimsgroup.com/?l=linux-kernel&m=109368723414915&w=2

-- 
lg, Chris