Recently I saw some reports about an oops with kvm, for example http://bugs.launchpad.net/bugs/729085
Some people are hitting them, as can be seen by already marked duplicates on above ticket. The problem was also reported/discussed at kvm mailing list: http://thread.gmane.org/gmane.comp.emulators.kvm.devel/68736 >From where a fix then came, and from what I saw affects only 38.x and later, but the fix wasn't included in 38.x series. A quick check to git history shows that this regression probably came from commit dacccfd in 2.6.38 One of the reporters at http://bugs.launchpad.net/bugs/729085 confirmed that commit 831ca60 "KVM: SVM: Load %gs earlier if CONFIG_X86_32_LAZY_GS=n" fixed the issue for him on a test kernel I provided with this commit applied. So this is a request to include this commit on 38.x stable series, on a reply to this I include the cherry-picked patch. -- []'s Herton _______________________________________________ stable mailing list [email protected] http://linux.kernel.org/mailman/listinfo/stable
