[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [Xen-devel] tsc_mode == 1 and hvm guests

On Mon, 10 May 2010, Dan Magenheimer wrote:
> > However I can tell you that without the patch when vtsc == 1
> > hvm_gtsc_need_scale returns 0 therefore scaling is never done.
> > For this reason I cannot see how the tsc frequency would remain the
> > same when vtsc == 1 and the underling hardware has a different tsc
> > rate.
> > The offset would be OK though.
> I think when vtsc==1 (e.g. tsc_mode==1), no scaling is
> done on HVM as the underlying machine is presented
> to be a 1GHz machine (regardless of the true TSC rate
> or whether migration had occurred or not).  PV also
> presented only a 1GHz clock for awhile and I changed
> it (with input from Jeremy as I recall) so that it
> would be less confusing to end users who might be
> confused by seeing Linux boot output claiming the
> underlying machine "is only running at 1GHz!!".
> There's a large number of different combinations to be
> tested (and possibly broken) here.  I'd prefer to see
> "I tested this feature on 4.0.0 and it doesn't work" before
> putting a patch into the stable stream.  For xen-unstable
> I have no problem with the patch of course.

I agree on this.
Since you definitely have the right infrastructure to test this, maybe it
is better to hold on until you can spare some time to do some more tests
on 4.0 with the patch applied.

At the moment I can only report "I tested pvlock in an HVM guest on
4.0.0 and it doesn't work".

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.