Bug #507

update kernel to 2.6.30

Added by graysky over 11 years ago. Updated almost 11 years ago.

Status:Closed Start date:
Priority:Normal Due date:12/31/1969
Assignee:cecil % Done:

0%

Category:- Spent time: -
Target version:7.3

Description

I used a 'bug report' rather than a feature request on this because the 2.6.28 kernel has a known bug with ext4 filesystems around data loss in the event of an ungraceful umount (i.e. power loss) when data is in the buffer, but not written to the disk. I felt this first hand on my Arch box before the .30 update to Arch. Anyway, kernel 2.6.30 has fixed this. You may have this fix backported to 2.6.28? I can dig up the references if you wish.

History

Updated by Human over 11 years ago

I bumped up the Severity because of the consequences of the bug, and I upped the Priority to emphasize it being in the final release.

However, if time is an issue, another option is to drop ext4 from this release and wait for the next one.

Updated by cecil over 11 years ago

Yes, please do dig up the details. If it can be backported to 2.6.28 that would be great. Else, this will have to wait post 6.01.00.

Updated by graysky over 11 years ago

Sorry guys, I dumped most of my bookmarks relating to this since 2.6.30 fixed it. Basically, there have been some modifications to 2.6.30 that prevent/minimize data loss on ext4 partition. From memory, the problem was a function of long wait times to write data from HDD cache to the FS. The modifications shortened this to allow for greater reliability at the cost of some FS performance. Again, I deleted most of my bookmarks detailing this once Arch when to 2.6.30. I do remember seeing something in the AUR about the 2.6.30-git stuff around this backported to at the time 2.6.28 or 2.6.29 - I just used some modifications to my /etc/fstab to minimize the errors. See [[http://bbs.archlinux.org/viewtopic.php?id=73300|this post]] for more details. These can be consider temp fixes until you guys can get 2.6.30 into the repos...

Here are a few I was able to find just through simple google searches, but again, I can't find the main ones. Perhaps in kernel.org in the changelogs...

http://www.h-online.com/open/Ext4-data-loss-explanations-and-workarounds--/news/112892

http://lwn.net/Articles/326342/

http://www.linuxquestions.org/questions/linux-news-59/ext4-data-loss-explanations-and-workarounds-712905/

Also available in: Atom PDF