On Thu, Mar 21, 2019 at 08:40:03PM -0400, Liang, Kan wrote:
>
>
> On 3/21/2019 5:20 PM, Peter Zijlstra wrote:
> > On Thu, Mar 21, 2019 at 01:56:44PM -0700, [email protected] wrote:
> > > @@ -933,6 +1001,34 @@ pebs_update_state(bool needed_cb, struct
> > > cpu_hw_events *cpuc, struct pmu *pmu)
> > > update = true;
> > > }
> > > + /*
> > > + * The PEBS record doesn't shrink on the del. Because to get
> > > + * an accurate config needs to go through all the existing pebs events.
> > > + * It's not necessary.
> > > + * There is no harmful for a bigger PEBS record, except little
> > > + * performance impacts.
> > > + * Also, for most cases, the same pebs config is applied for all
> > > + * pebs events.
> > > + */
> > > + if (x86_pmu.intel_cap.pebs_baseline && add) {
> > > + u64 pebs_data_cfg;
> > > +
> > > + /* Clear pebs_data_cfg and pebs_record_size for first PEBS. */
> > > + if (cpuc->n_pebs == 1) {
> > > + cpuc->pebs_data_cfg = 0;
> > > + cpuc->pebs_record_size = sizeof(struct pebs_basic);
> > > + }
> >
> > Argh, no. This is daft. The previous site was fine, it was just the
> > pebs_record_size assignment I'm confused about.
> >
> > Note how by setting ->pebs_data_cfs to 0, you force the below branch to
> > true and call adaptive_pebs_record_size_update() ? So _why_ do you have
> > to set pebs_record_size()?
> >
>
> I think we have to reset both cpuc->pebs_data_cfg and
> cpuc->pebs_record_size. Because pebs_update_adaptive_cfg() can return 0.
> If so, adaptive_pebs_record_size_update() will not be called. The
> cpuc->pebs_record_size still use the stale data, which may be wrong.
Oh, bugger.. I see.
> > > +
> > > + pebs_data_cfg = pebs_update_adaptive_cfg(event);
> > > +
> > > + /* Update pebs_record_size if new event requires more data. */
> > > + if (pebs_data_cfg & ~cpuc->pebs_data_cfg) {
> > > + cpuc->pebs_data_cfg |= pebs_data_cfg;
> > > + adaptive_pebs_record_size_update();
> > > + update = true;
> > > + }
> > > + }
> > > +
> > > if (update)
> > > pebs_update_threshold(cpuc);
> > > }
