Am 2016-12-16 um 19:31 schrieb Franz Sirl:
> Am 2016-12-16 um 15:31 schrieb Simon Flood:
>> On 16/12/16 13:48, Franz Sirl wrote:
>>
>>> the complete message is:
>>>
>>> [ 3411.029025] Uhhuh. NMI received for unknown reason 30 on CPU 0.
>>> [ 3411.029028] Do you have a strange power saving mode enabled?
>>> [ 3411.029028] Dazed and confused, but trying to continue
>>> [ 3441.028990] Uhhuh. NMI received for unknown reason 20 on CPU 0.
>>> [ 3441.028996] Do you have a strange power saving mode enabled?
>>> [ 3441.028996] Dazed and confused, but trying to continue
>>>
>>> When it happens, it seems to happen very often (every 30 seconds) and
>>> the guest typically locks up totally after 1-2 days.
>>>
>>> Host and guest are both SLES12SP2 with kernel-4.4.21-84/88. There is no
>>> problem when the guest is freshly started (virsh start guest), but as
>>> soon as I issue a 'reboot' within the guest, the messages start after a
>>> few seconds.
>>>
>>> Anyone seen this and knows a workaround?

>>
>> Can you try updating a guest (and possibly the host) to latest kernel
>> available for SLES12 SP2 - 4.4.21-90.1 ?

>
> Hi Simon,
>
> I misremembered, it was with kernel-4.4.21-84/90, the host is still
> kernel-4.4.21-84. qemu and xen-libs on the host have the latest updates
> though. I just retried the reboot on the guest again, but after ~10min
> the messages appeared again.
>
> I'll try to fully update the host too over the weekend, but I don't have
> much hope because only 2 CVEs are listed for the -90 kernel update.


I was right, the -90 kernel update doesn't change anything. But I think
I have found a hint now, it only happens with the Q35 emulation, another
VM with the I440FX emulation reboots fine. So there is likely a problem
in qemu with Q35 emulation. I'll try to find more when there is time.

Franz