Re: [Xen-devel] Re: [PATCH] x86: unconditionally mark TSC unstable under Xen

[Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>]

On Wed, 14 Jul 2010, Jeremy Fitzhardinge wrote:
> On 07/14/2010 12:24 PM, Jed Smith wrote:
> > Jeremy, Jan - what do you think?  Is this a bad move?  I feel like there
> > is a consequence to this that I am unaware of, but it fixes my issue.
> >   
> 
> Ah, well that's interesting.
> 
> There's a couple of comments:
> 
>    1. you can't do this with just a compile-time test, since the same
>       kernel can also boot native
>    2. nothing in a Xen PV domU environment should be using the tsc
>       directly, so this shouldn't have an effect.  If something is using
>       the tsc we should track it down.
> 
> I wonder, however, if you're getting the same result as Jan's suggestion
> of making sched_clock unstable by making the tsc unstable.
> 
> In that case, this patch may help:
> 
> Subject: [PATCH] xen: disable xen_sched_clock by default
> 
> xen_sched_clock only counts unstolen time.  In principle this should
> be useful to the Linux scheduler so that it knows how much time a process
> actually consumed.  But in practice this doesn't work very well as the
> scheduler expects the sched_clock time to be synchronized between
> cpus.  It also uses sched_clock to measure the time a task spends
> sleeping, in which case "unstolen time" isn't meaningful.
> 
> So just use plain xen_clocksource_read to return wallclock nanoseconds
> for sched_clock.
> 

I think that in this context is worth mentioning that
xen_clocksource_read ends up calling pvclock_get_nsec_offset that calls
native_read_tsc.


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel