On Tue, Feb 18, 2025 at 02:40:06PM -0600, Rob Herring (Arm) wrote:
>
> From: Anshuman Khandual <anshuman.khandual@xxxxxxx>
[...]
> BRBE records are invalidated whenever events are reconfigured, a new
> task is scheduled in, or after recording is paused (and the records
> have been recorded for the event). The architecture allows branch
> records to be invalidated by the PE under implementation defined
> conditions. It is expected that these conditions are rare.
[...]
> +static void armv8pmu_sched_task(struct perf_event_pmu_context *pmu_ctx, bool sched_in)
> +{
> + struct arm_pmu *armpmu = *this_cpu_ptr(&cpu_armpmu);
> + struct pmu_hw_events *hw_events = this_cpu_ptr(armpmu->hw_events);
> +
> + if (!hw_events->branch_users)
> + return;
> +
> + if (sched_in)
> + brbe_invalidate();
> +}
Just a minor concern. I don't see any handling for task migration.
E.g., for a task is migrated from one CPU to another CPU, I expect we
need to save and restore branch records based on BRBE injection. So
far, the driver simply invalidates all records.
I think this topic is very likely discussed before. If this is the
case, please ignore my comment. Except this, the code looks good
to me.
Thanks,
Leo
