[Xen-users] Re: OpenSUSE 11.2 Xen HVM guest crash causing BUG() in dom0

[Pasi Kärkkäinen <pasik@xxxxxx>]

On Mon, Jan 25, 2010 at 08:30:33AM +0000, Jan Beulich wrote:
> >>> Pasi Kärkkäinen<pasik@xxxxxx> 23.01.10 20:14 >>>
> >Just to be more clear, Mike used 2.6.32-41.1 x86_64 kernel from 
> >http://download.opensuse.org/repositories/Kernel:/HEAD/openSUSE_11.2/x86_64/ 
> >
> >and it seems he doesn't get BUG() and traceback in dom0 anymore when 
> >HVM guest crashes:
> >
> >http://pastebin.com/m5aff9158 
> >
> >(XEN) Domain 3 reported crashed by domain 0 on cpu#0:
> >
> >lots of these still though:
> >
> >(XEN) domain_crash called from p2m.c:1091
> >(XEN) p2m_pod_demand_populate: Out of populate-on-demand memory!
> >
> >and then
> >
> >[  233.171451] BUG: soft lockup - CPU#0 stuck for 61s! [qemu-dm:3722]
> >
> >So I guess that's OK now..
> 
> Hmm, OK is probably not the right statement with there still being
> lockups. However, I take your wording as the lockups not being
> connected to the domain crashes anymore. 
>

Sorry, I wrote it a bit wrong. The lockup still happens when the HVM guest 
crashes,
but now there was not traceback anymore.

> Disabling C-state management (cpuidle=0 or max_cstate=0 on the Xen command 
> line)
> might be worth trying (based on recent observations reported on
> xen-devel).
> 

Hopefully Mike can try it..

> Of course, fixing the underlying config problem might help the user
> even more: Apparently they configure the guest with mem!=maxmem,
> and either the balloon driver is not being loaded at all or isn't getting
> its target set low enough (we're apparently having the latter problem
> in SLE11SP1, so it's not impossible for it to also exist in 11.2).
>

Ok. The HVM guest was Freenas (FreeBSD), so I don't know if it even has 
balloon driver available..

-- Pasi


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users