Re: [RFC] sched_clock: Track monotonic raw clock

[Pawel Moll]

On Tue, 2014-07-22 at 17:34 +0100, Peter Zijlstra wrote:
> On Tue, Jul 22, 2014 at 05:17:29PM +0100, Pawel Moll wrote:
> > With both Ingo and John showing preference towards the clock alignment,
> > so that's where I looked this time (I've already done custom perf
> > ioctls, posix clocks... don't really know how many different ways I've
> > tried).
> 
> So we should probably also talk about which clock to track, MONO has the
> advantage of making it far easier to trace clusters but has the
> disadvantage of stacked control loops with NTP adjusting MONO and us
> adjusting sched_clock.

John suggested (and I fully agree with him) MONO_RAW, as it is not
getting NTP-d. 

> And I would really prefer to pick 1 and not make it configurable.

Same here. One thing I keep in mind is the fact that userspace must be
able to say whether it can expect the correlation or not. "Not" being
either an architecture which sched_clock is not using the generic
solution (I'm sure there will be some) or "not" because of the
synchronisation failure. My idea so far was a debugfs file saying this
(or missing, which is a message on its own).

Pawel


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/