Re: [PATCH 1/2] KVM: Expose MCE control MSRs to userspace

[Huang Ying <ying.huang@intel.com>]

Sorry for late.

On Thu, 2010-07-08 at 17:12 +0800, Avi Kivity wrote:
[..snip..]
> >>> And if we clear MSR_IA32_MCG_CTL, the machine check reporting is
> >>> disabled according to SDM Vol 3A, section 15.3.1.3
> >>>
> >>>        
> >> Won't the kernel reenable MCE?  In my testing, the sequence
> >> MCE-reset-MCE worked after the patch (whereas it would fail without it).
> >>      
> > Yes. Because kernel will reenable it. But if we only clear
> > MSR_IA32_MCG_STATUS only, MCE-reset-MCE should work too.
> >    
> 
> What happens if we reboot into a kernel that doesn't enable MCE?
> 
> I guess it doesn't matter: the new kernel will keep cr4.mce cleared and 
> thus MCE will be blocked.
> 
> I'd like to keep the patch as is, so live migration works for MCE (we'll 
> need to add bank support).  I think there's no problem clearing _CTL on 
> reset.  If there is, we can patch qemu not to clear the MSR.  Is that 
> acceptable?

Yes. At least for Linux that is OK. Don't know whether Windows will set
_CTL during boot.

Best Regards,
Huang Ying