Need help debuggung kvm issue on F19

Chris Murphy lists at colorremedies.com
Fri May 17 19:48:24 UTC 2013


On May 17, 2013, at 1:37 PM, Chris Murphy <lists at colorremedies.com> wrote:

> 
> On May 17, 2013, at 1:20 PM, Matthias Runge <mrunge at matthias-runge.de> wrote:
> 
>> On 05/17/2013 08:23 PM, Chris Murphy wrote:
>>> There are a number of bugs on this. I'm hitting these two (maybe the
>>> same bug) consistently: 954181 963966
>>> 
>>> But the interesting part is that I only get the kernel oops on an F19
>>> host (consistently upon remote restart or poweroff of the VM via
>>> ssh). I don't get the oops if the same host is running F18, with the
>>> same kernel.
>>> 
>>> 3.9.2-200.fc18 3.9.2-301.fc19
>>> 
>>> So I'm thinking the newer versions of qemu-kvm on F19 are tickling
>>> the kernel in a way it doesn't like, but for now I've reverted to F18
>>> on baremetal and test F19 in VM.
>> Good to know, thank you.
>> 
>> Interesting part is a kernel from f18 works, and the 3.10-rc?? from rawhide-nodebug works as well.
> 
> Host: F18 up to date with 3.9.2-200.fc18, and an F19 guest in qemu-kvm with virt-manager. This one is stable.
> 
> Host: F19 up to date with any of the fc19 kernels including 3.9.2-301.fc19 or 3.10.0-0.rc1.git5.1.fc20, with the same F19 guest, is not stable. It's wonky on the host through the GUI (really really sluggish even with the nodebug kernels) but doesn't crash. Only if I ssh into the VM, and issue a reboot or poweroff does the VM hang and immediately the host kernel oopses.
> 
> Are kernels 3.9.2-200.fc18 and 3.9.2-301.fc19 binary identical? Or is there some small meaningful difference?

And FWIW the two hosts above are the same physical machine.


Chris Murphy


More information about the test mailing list