Red Hat Bugzilla – Bug 54348
Kernel BUG at inode.c 1039!
Last modified: 2005-10-31 17:00:50 EST
From Bugzilla Helper:
User-Agent: Mozilla/4.7 [en] (Win98; U)
Description of problem:
Please fill in each of the sections below.
NetVista PC (model type 6578-D5U) with Hitachi DVD-Ram Drive
Linux RedHat 7.1
kernel level 2.4.3-12
kernel level 2.4.7-xx
Steps to Reproduce:
1.We were able to use our DVD-Drive in previous 2.4.2 kernel. When we upgrade
to 2.4.3-12 kernel from RedHat, we were unable to mount the DVD drive with the
actual DVD Ram media. Mounting a CDROM is OK. After talking to RedHat, we were
asked to try out kernel level 2.4.7-xx from the RedHat website. This is not yet
released by RedHat and is still a Gold/Beta version. Using this kernel, we were
successfully mounting the DVD-Driver to do I/O to the media. However when using
vi or rm on a file on the DVD-Media, we hit a kernel BUG at inode.c: 1039!
A person in our team in Endicott dowloaded kernel 2.4.9-ac7 from kernel.org and
was able to run ok, however due to legal requirements we must use a kernel from
We have verified that this problem was fixed in 2.4.10.
Requesting this fix be ported to version 7.1
Version-Release number of selected component (if applicable):
Steps to Reproduce:
From the originator:
Below is the status of kernels that work/fail when attempting to utilize the DVD-RAM capabilities of the HMC
Kernel version Status
2.4.2-2 works, old RedHat 7.1 kernel
2.4.7-2 fails, RedHat kernel for 7.2, release candidate 2
2.4.7-6 fails, new RedHat kernel from rawhide
2.4.9-ac7 works, from kernel.org
2.4.10 works, from kernel.org
Again, be careful distinguishing from a DVD-RAM failure and a UDF (filesystem) error. I believe we have the latter.
Using diff, we attempted to fix kernel 2.4.3-12 by replacing the udf.c(??) source module with the back level one that worked (2.4.2-2). I do not
remember the exact module name, but we can provide it if required (John). We we not successful. Only 1 source code line was changed - from my
partially erased white board:
if ( !(.... = get_HardBlockSize( ) ) ) was changed to if ( !(.... = get_HardSectorSize( ) )) or something like that.
Whatever library the get_HardBlock/SectorSize( ) calls reside in that the udf.c (??) module links with, apparently changed as well.
Closing as we received the required support in the errata: RHSA-2001:142-15. Thanks for all the work.