Comment # 27 on bug 1202541 from Dario Faggioli

(In reply to Bernhard Wiedemann from comment #16)
> Another interesting find:
> 
> atreju6-suse:~ # cat /proc/cpuinfo |grep lwp|wc -l
>      16
> atreju6-suse:~ # cat /proc/cpuinfo |grep nodeid_msr|wc -l
>      32
> 
So, about this LWP thing, I've found this:

https://bugzilla.kernel.org/show_bug.cgi?id=216211

Apparently, it's a known issue on some (Bulldozer/Fam15h) AMD CPUs, with some
microcodes. Boris (Cc-ing him) is on the case, AFAICT. If/when a fix comes out,
we probably should consider backporting it, but I guess this will be taken care
of by the kernel team, as it's not at all a virt-related issue, and as we see
in that bug, it causes issues on baremetal as well.

Basically, it's possible that this could be the cause of, or that it plays a
role in, this crash, and we probably want it fixed anyway, but the key point
here is to stop using host-passthrough for VMs that needs to be migrated.