Peter Oberparleiter wrote:
> The source of this problem is most likely due to creating a
> filesystem on a CDL formatted device instead of a partition.

A quick update to my statement above: trying to mount a filesystem
created on a CDL formatted device will immediately fail because the first
records of track 0 and all records of track 1 are formatted with a
different blocksize than the rest of the device and the ext2/3 filesystem
tries to store a super block at position 0.

What is more likely to have happened is that the label information
from a previously CDL formatted disk was erased (e.g. by a direct write
to the first records of the device). In that case, during the next
IPL, the partition detection code will assume a LDL formatted device
due to the missing label and create one partition from record 4 to the
device end. This partition includes the different-sized records on
track 1. Once the partition is sufficiently populated with files and
directories, either file or filesystem data is stored at this location,
therefore becoming corrupted.

Steps to reproduce the error on /dev/dasdb:

dasdfmt -f /dev/dasdb -b 4096
dd if=/dev/zero of=/dev/dasdb bs=4096 count=3
blockdev --rereadpt /dev/dasdb
mkfs.ext2 /dev/dasdb1
mount /dev/dasdb1 /mnt

then create enough files and directories on the device and filesystem
corruption will occur (more precisely, patterns of 0xe5 will show up).

Suggested solution: reformat disks prior to installing production
systems.


Regards,
  Peter Oberparleiter

--
Peter Oberparleiter
Linux on zSeries Development
IBM Development Lab, Boeblingen/Germany

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to