Re: [PATCH 8/8] watchdog: Remove hack to make full dynticks working

[Peter Zijlstra]

On Tue, Jul 23, 2013 at 08:33:31AM -0400, Don Zickus wrote:
> On Tue, Jul 23, 2013 at 02:31:06AM +0200, Frederic Weisbecker wrote:
> > A perf event can be used without forcing the tick to
> > stay alive if it doesn't use a frequency but a sample
> > period and if it doesn't throttle (raise storm of events).
> > 
> > Since the lockup detector neither use a perf event frequency
> > nor should ever throttle due to its high period, it can now
> > run concurrently with the full dynticks feature.
> 
> Thanks.  Dumb question, I keep wondering if the lockup detector would be
> better or worse off if it used the perf event frequency as opposed to
> using a sample period?  The idea is it could follow the varying cpu
> frequencies better (and probably simplify some of the code too).

Right, trouble is that someone didn't consider fractional frequencies
when writing the interface :/ Lowest we can go is 1 Hz while we'd want
something like 0.1 Hz or smaller.

Also, like the above says, that would interfere with the nohz efforts as
perf needs the tick to re-compute those frequency thingies.


--
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/