Re: [RESEND PATCH V3 3/8] perf/x86/intel: Support hardware TopDown metrics

[Andi Kleen <ak@linux.intel.com>]

On Wed, Aug 28, 2019 at 05:02:38PM +0200, Peter Zijlstra wrote:
> > 
> > To avoid reading the METRICS register multiple times, the metrics and
> > slots value can only be updated by the first slots/metrics event in a
> > group. All active slots and metrics events will be updated one time.
> 
> Can't we require SLOTS to be the group leader for any Metric group?

Metrics are really useful to collect with any other sampling event
to give you more context in the program behavior.

> Is there ever a case where we want to add other events to a metric
> group?

Yes. Any normal leader sampling case. You just collect metrics too.

> it also musn't be too cold. But that leaves it unspecified what exactly
> is the right range.
> 
> IOW, you want a Goldilocks number of SLOTS.

That's too strong. You probably don't want minutes, but seconds
worth of measurement should be ok.

> > NMI
> > ======
> > 
> > The METRICS register may be overflow. The bit 48 of STATUS register
> > will be set. If so, update all active slots and metrics events.
> 
> that happen? It would be useful to get that METRIC_OVF (can we please

This happens when the internal counters that feed the metrics
overflow.

> If this is so; then we can use this to update/reset PERF_METRICS and
> nothing else.

It has to be handled in the PMI.

> Then there is no mucking about with that odd counter/metrics msr pair
> reset nonsense. Becuase that really stinks.

You have to write them to reset the internal counters.

-Andi