Re: [PATCH 11/17] Fix a possible backwards warp of kvmclock

[Glauber Costa <glommer@redhat.com>]

On Wed, Jun 16, 2010 at 05:13:02PM +0300, Avi Kivity wrote:
> On 06/16/2010 04:58 PM, Glauber Costa wrote:
> >On Wed, Jun 16, 2010 at 04:11:26PM +0800, Jason Wang wrote:
> >>Zachary Amsden wrote:
> >>>Kernel time, which advances in discrete steps may progress much slower
> >>>than TSC.  As a result, when kvmclock is adjusted to a new base, the
> >>>apparent time to the guest, which runs at a much higher, nsec scaled
> >>>rate based on the current TSC, may have already been observed to have
> >>>a larger value (kernel_ns + scaled tsc) than the value to which we are
> >>>setting it (kernel_ns + 0).
> >>>
> >>This is one issue of kvmclock which tries to supply a clocksource whose
> >>precision may even higher than host.
> >What if we export to the guest the current clock resolution, and when doing guest
> >reads, simply chop whatever value we got to the lowest acceptable value?
> 
> The clock resolution can change, and while we can expose it reliably
> through pvclock, do we need a notification so that the guest can
> update other internal structures?
I believe the only thing we need is to warn the guest whenever this changes.
We can probably fit it in one of the paddings we have, and add a flag to say
it is valid (now that we have the infrastructure).