I'm pretty sure Fedora kernel folks are familiar with this:
https://bugs.edge.launchpad.net/ubuntu/+source/linux/+bug/317781
And the associated fixes.
Are Ted's patches, that are queued up for 2.6.30, going to be applied to F-11 (and F-10) 2.6.29 kernels or not? Asking just out of curiosity.
Bojan Smojver wrote:
I'm pretty sure Fedora kernel folks are familiar with this:
https://bugs.edge.launchpad.net/ubuntu/+source/linux/+bug/317781
And the associated fixes.
Are Ted's patches, that are queued up for 2.6.30, going to be applied to F-11 (and F-10) 2.6.29 kernels or not? Asking just out of curiosity.
I sure hope they will. This is pretty much a blocker for the KDE spin.
Kevin Kofler
Kevin Kofler wrote:
Bojan Smojver wrote:
I'm pretty sure Fedora kernel folks are familiar with this:
https://bugs.edge.launchpad.net/ubuntu/+source/linux/+bug/317781
And the associated fixes.
Are Ted's patches, that are queued up for 2.6.30, going to be applied to F-11 (and F-10) 2.6.29 kernels or not? Asking just out of curiosity.
I sure hope they will. This is pretty much a blocker for the KDE spin.
Kevin Kofler
They are in F11, yes.
Chuck said he'd merge them to F10 as well.
Kevin, now that ext4 has introduced the Linux world to the concept of buffered data and fsync .... ;)
Do you know people who know people who can maybe see about making KDE more robust without requiring the filesystem to play nanny for the config files? :)
Thanks,
-Eric
p.s. everyone please go read http://flamingspork.com/talks/2007/06/eat_my_data.odp
On Thu, Mar 12, 2009 at 10:03 PM, Kevin Kofler kevin.kofler@chello.at wrote:
Are Ted's patches, that are queued up for 2.6.30, going to be applied to F-11 (and F-10) 2.6.29 kernels or not? Asking just out of curiosity.
I sure hope they will. This is pretty much a blocker for the KDE spin.
Is there any design document about how KDE persists configuration files? I have noticed application crashes affecting recently-applied KDE settings in the past, even on ext3, while GConf settings seem to be persisted to disk almost immediately.
Thanks,
On Thu, 2009-03-12 at 22:41 -0400, Michel Salim wrote:
On Thu, Mar 12, 2009 at 10:03 PM, Kevin Kofler kevin.kofler@chello.at wrote:
Are Ted's patches, that are queued up for 2.6.30, going to be applied to F-11 (and F-10) 2.6.29 kernels or not? Asking just out of curiosity.
I sure hope they will. This is pretty much a blocker for the KDE spin.
Is there any design document about how KDE persists configuration files? I have noticed application crashes affecting recently-applied KDE settings in the past, even on ext3, while GConf settings seem to be persisted to disk almost immediately.
Is KDE Doing It Wrong?
https://bugs.edge.launchpad.net/ubuntu/+source/linux/+bug/317781/comments/54
rename() is what I mean by atomic filesystem operations.
Callum Lerwick wrote:
On Thu, 2009-03-12 at 22:41 -0400, Michel Salim wrote:
On Thu, Mar 12, 2009 at 10:03 PM, Kevin Kofler kevin.kofler@chello.at wrote:
Are Ted's patches, that are queued up for 2.6.30, going to be applied to F-11 (and F-10) 2.6.29 kernels or not? Asking just out of curiosity.
I sure hope they will. This is pretty much a blocker for the KDE spin.
Is there any design document about how KDE persists configuration files? I have noticed application crashes affecting recently-applied KDE settings in the past, even on ext3, while GConf settings seem to be persisted to disk almost immediately.
Is KDE Doing It Wrong?
https://bugs.edge.launchpad.net/ubuntu/+source/linux/+bug/317781/comments/54
rename() is what I mean by atomic filesystem operations.
I would add that you need to fsync() the directories as well to make sure that the rename survives a power failure. Note that in the worst case without the directory renames, you would still have your old file intact (and a temp file to clean up) so it is not horrible.
Ric