fsdb&smartctl&/var/log/messages

[Miroslav Lachman]

Dmitry Lunts wrote:
> Hello,all!
> The question is as follows.
> 1).When I try to upgrade some port or merely execute the command pkgdb -uF
> I get the error:
> Input/output error - /var/db/pkg/kdeutils-3.5.10_5/+CONTENTS
>
> 2)$sudo cat /var/log/messages|grep DMA|tail -2
> gives:
> Jul 12 03:07:06 dim007 kernel: ad6: FAILURE - READ_DMA
> status=51<READY,DSC,ERROR>  error=40<UNCORRECTABLE>  LBA=4007967
> Jul 12 03:07:09 dim007 kernel: ad6: FAILURE - READ_DMA
> status=51<READY,DSC,ERROR>  error=40<UNCORRECTABLE>  LBA=4007967
>
> 3)From the output of
> $cat /etc/fstab and
> $sudo bsdlabel ad6s1:
> #        size   offset    fstype   [fsize bsize bps/cpg]
>    a:  1048576        0    4.2BSD     2048 16384     8
>    b:  1048576  1048576      swap
>    c: 100663227        0    unused        0     0         # "raw" part, don't
> edit
>    d:  2097152  2097152    4.2BSD     2048 16384 28552
>    e:   655360  4194304    4.2BSD     2048 16384 40968
>    f: 37748736  4849664    4.2BSD     2048 16384 28552
>    g: 58064827 42598400    4.2BSD     2048 16384 28552
> I can conclude that LBA=4007967 falls into /dev/ad6s1d partition mounted as
> /var
>
> Next:
>   4) $sudo fsdb -r /dev/ad6s1d
> ** /dev/ad6s1d (NO WRITE)
> Examining file system `/dev/ad6s1d'
> Last Mounted on /var
> current inode: directory
> I=2 MODE=40755 SIZE=512
>          BTIME=Jan  1 15:07:34 2009 [0 nsec]
>          MTIME=Jul 12 03:03:19 2010 [0 nsec]
>          CTIME=Jul 12 03:03:19 2010 [0 nsec]
>          ATIME=Jul 10 01:09:22 2010 [0 nsec]
> OWNER=root GRP=wheel LINKCNT=27 FLAGS=0 BLKCNT=4 GEN=5e655284
>
> Offset of bad LBA sector within /dev/ad6s1d (i.e., /var) partition is:
> (bad LBA sector-63-offset of /dev/ad6s1d=4007967-63-2097152=1910752 (see the
> output of bsdlabel above)
>
> 5) Searching for inode:
> fsdb (inum: 2)>findblk 1910752
> 1910752: data block of inode 117934
>
> 6)Searching for file:
> $sudo find /var -inum 117934
> /var/db/pkg/kdeutils-3.5.10_5/+CONTENTS
> which exactly corresponds to error message from pkgdb -uF (see item 1)
> above)

[...]

> 10) And what's more:
> offset of LBA_of_first_error within /dev/ad6s1d is
> 4007996-63-2097152=1910781
> fsdb->findblk returns nothing:
> fsdb (inum: 2)>  findblk 1910781
> fsdb (inum: 2)>
>
> So, the puzzle is:
> from one hand, pkgdb -uF, /var/log/messages, fsdb point to the same bad
> sector (4007967) and filename
> this sector belongs to
> (and even cat /var/db/pkg/kdeutils-3.5.10_5/+CONTENTS returns Input/Output
> error),
> but low level reading of bad sector returns without a sign of error.
>> From the other hand, smartctl long test performed immediately after low
> level read test with dd
> gives quite another number of bad sector (4007996) which in turn doesn't
> belong to any file.
>
> And what makes me completely lost is that
> the tests in items 1)-10) were repeated two times and gave  the same
> results!
> So where are 29 (4007996-4007967) sectors lost?
> Could anyone give me a hint where I'm wrong?

I had similar problem in the past. You can find more about it here 
http://lists.freebsd.org/pipermail/freebsd-current/2010-March/015987.html

Miroslav Lachman