On Thu, Jun 14, 2018 at 10:32:00PM +0200, Paolo Bonzini wrote:
> On 14/06/2018 17:44, Daniel P. Berrangé wrote:
> >> "-cpu" is certainly wrong for KVM_CAP_X86_DISABLE_EXITS. "-cpu" is a
> >> device option, while this is about host behavior. "-realtime"'s name is
> >> awful, but I still think it's the best place for this option. Maybe we
> >> could call it "-realtime power-mgmt={host|guest}".
> > If none of the existing ones are a suitable fit, then we should just
> > introduce a new CLI arg instead stuffing it into somewhere odd.
>
> It is related to mlock in my opinion. Both options are about applying
> settings that reduce latency but may not be applicable in general.
I would add, it's also about host resource handling.
With mlock guest memory is locked so there's no memory overcommit.
With cpu-pm guest does not give up host cpu which mostly disables
host cpu overcommit.
> Maybe "-realtime" could be renamed to "-perftune" or something like
> that, but I'd like to keep the grouping.
>
> Paolo
-overcommit ?
> > IOW, why not just "--power-mgmt host|guest"
