I discovered the following error in my cron output recently: mcelog: warning: record length longer than expected. Consider update. This problem appears to only occur when the running kernel is >=2.6.31. This is because 2.6.31 included changes to the data structure which mcelog reads. More information is here: https://bugzilla.redhat.com/show_bug.cgi?id=507026 A patch to fix the issue is also included in the redhat bug report.
they are readying 0.9. Seems there is a source RPM but no official source tar archive.
Not much progress from upstream. Any chance of getting the patch added in a rev bump pending the new release?
I am also interested in this fix. May I contribute or help in any way?
Created attachment 212621 [details, diff] Patch from https://bugzilla.redhat.com/show_bug.cgi?id=507026
Created attachment 212622 [details, diff] Patch for mcelog ebuild
FYI, new mcelog pre-releases are at http://www.kernel.org/pub/linux/utils/cpu/mce/
Created attachment 216280 [details] mcelog-1.0_pre2.ebuild This takes the latest 1.0_pre2 version (only change is the patch version and dbquery no longer exists).
pre3 is out: http://www.kernel.org/pub/linux/utils/cpu/mce/mcelog-1.0pre3.tar.bz2 only changes from pre2 are: > Add Xeon75xx support > Add Intel architectural mca fallback for new family 6 Intel CPUs
(In reply to comment #8) > pre3 is out: > http://www.kernel.org/pub/linux/utils/cpu/mce/mcelog-1.0pre3.tar.bz2 Is the epatch command on line 27 wrong? After emerge failed when it couldn't find the patch, I commented out the line and the program runs totally fine without it.
(In reply to comment #9) > (In reply to comment #8) > > pre3 is out: > > http://www.kernel.org/pub/linux/utils/cpu/mce/mcelog-1.0pre3.tar.bz2 > > Is the epatch command on line 27 wrong? After emerge failed when it couldn't > find the patch, I commented out the line and the program runs totally fine > without it. > You also need to add 'eutils' to the inherit on line 5.
Created attachment 219061 [details, diff] add eutils to inherit line.
I've committed now 1.0_pre3.
This has been in here for a while. I think it should probably be taken care of. I know that the _pre builds aren't ready to be marked stable, but I see no reason whatever that the 0.7-r1 patch can't be added to a new stable mcelog-0.7_r1.ebuild. Is there a problem with not knowing which kernel the users have? If so, I would think that the ebuild could easily check the kernel version and apply the epatch or not based on >= 2.6.31?