[PVE-User] kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0xc1 data 0xffffffffff1b1e41

[Laurent CARON]

On Tue, Mar 06, 2012 at 11:18:02AM +0100, Laurent CARON wrote:
> On Sun, Feb 19, 2012 at 01:27:57PM +0100, Laurent CARON wrote:
> > Hi,
> > 
> > I'm experiencing this kind of messages:
> > __ratelimit: 59 callbacks suppressed
> > kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0xc1 data 0xffffffffff1b1e41
> > kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0x186 data 0x5100c0
> > kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0xc1 data 0x0
> > kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0xc1 data 0x0
> > kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0x186 data 0x5100c0
> > kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0xc1 data 0x0
> > kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0xc1 data 0xffffffffff1b1e41
> > kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0x186 data 0x5100c0
> > kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0xc1 data 0x0
> > kvm: 4288: cpu2 unimplemented perfctr wrmsr: 0xc1 data 0x0
> > 
> > with one of my VMs (windows 2003 64bit):
> > 
> > the $MACHINE_ID.conf is:
> > name: kav
> > ide2: none,media=cdrom
> > net0: e1000=8A:26:7C:8A:C3:02,bridge=vmbr0
> > bootdisk: ide0
> > ostype: w2k8
> > ide0: VG_SSD_pm-siege-001:vm-502-disk-1
> > memory: 6144
> > sockets: 2
> > onboot: 0
> > cores: 2
> > 
> > Do you guys have a clue about it ?
> > 
> > Might it be because I put ostype: w2k8 instead of wxp ?
> 
> Hi,
> 
> After migrating this machine to another host (different CPU), I don't
> experience this issue anymore.
> 
> Shall I report this bug to kvm bugzilla ?

Hi,

Update on this: Dell published a BIOS update for servers using E7 CPUs.

"This BIOS release resolves the potential for unpredictable system
behavior in Dell systems running Intel(R) Xeon(R) Processor
E7-2800/4800/8800 Series. For systems running Intel(R) Xeon(R) Processor
E7-2800/4800/8800 Series, Dell strongly recommends applying this
critical update as soon as possible. Systems updated with this version
of BIOS or newer and running Intel(R) Xeon(R) Processor"

The microcode has apparently been upgraded following to a bug.

Everything is now normal and no strange error messages are showing using
x64 windows VMs.

Thanks

Laurent