Re: KVM: x86: fix tsc catchup issue with tsc scaling

[Marcelo Tosatti <mtosatti@redhat.com>]

On Wed, Jan 15, 2014 at 05:53:22PM +0100, Paolo Bonzini wrote:
> Il 15/01/2014 17:37, Marcelo Tosatti ha scritto:
> > > Right.  Another question, what about this comment:
> > > 
> > >         /* Reset of TSC must disable overshoot protection below */
> > >         vcpu->arch.hv_clock.tsc_timestamp = 0;
> > >         vcpu->arch.last_guest_tsc = data;
> > > 
> > > Should it be instead like this:
> > > 
> > > 	vcpu->arch.hv_clock.tsc_timestamp += data - vcpu->arch.last_guest_tsc;
> > > 	vcpu->arch.last_guest_tsc = data;
> > > 
> > > ?
> > 
> > Don't see why it should?
> 
> Setting tsc_timestamp to 0 makes no sense with the current code, I'm
> trying to understand if the right fix is to delete that line or
> something else.  What I proposed (plus a version increase) would make a
> pvclock read continuous before and after the change.
> 
> But looking more at the surrounding code, the calls to kvm_write_tsc
> ultimately result in a master clock update as soon as all TSCs agree and
> the master clock is re-enabled.  This master clock update rewrites
> tsc_timestamp.  Then I think that the line that sets tsc_timestamp to 0
> can be deleted.

ACK.