[PATCHv2 9/9] arm: perf: fold hotplug notifier into arm_pmu

[Mark Rutland]

On Mon, Oct 27, 2014 at 08:37:24PM +0000, Stephen Boyd wrote:
> On 10/27/2014 05:06 AM, Mark Rutland wrote:
> > Handling multiple PMUs using a single hotplug notifier requires a list
> > of PMUs to be maintained, with synchronisation in the probe, remove, and
> > notify paths. This is error-prone and makes the code much harder to
> > maintain.
> >
> > Instead of using a single notifier, we can dynamically allocate a
> > notifier block per-PMU. The end result is the same, but the list of PMUs
> > is implicit in the hotplug notifier list rather than within a perf-local
> > data structure, which makes the code far easier to handle.
> >
> > Signed-off-by: Mark Rutland <mark.rutland at arm.com>
> 
> Reviewed-by: Stephen Boyd <sboyd at codeaurora.org>
> 
> One nit below.
> 
> > @@ -169,6 +192,11 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
> >  	if (!cpu_hw_events)
> >  		return -ENOMEM;
> >  
> > +	cpu_pmu->hotplug_nb.notifier_call = cpu_pmu_notify;
> > +	err = register_cpu_notifier(&cpu_pmu->hotplug_nb);
> > +	if (err)
> > +		goto out_hw_events;
> > +
> >  	for_each_possible_cpu(cpu) {
> >  		struct pmu_hw_events *events = per_cpu_ptr(cpu_hw_events, cpu);
> >  		raw_spin_lock_init(&events->pmu_lock);
> > @@ -188,38 +216,19 @@ static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
> >  		cpu_pmu->pmu.capabilities |= PERF_PMU_CAP_NO_INTERRUPT;
> >  
> >  	return 0;
> > +
> > +out_hw_events:
> > +	free_percpu(cpu_hw_events);
> > +	return err;
> >  }
> >  
> >  static void cpu_pmu_destroy(struct arm_pmu *cpu_pmu)
> >  {
> >  	free_percpu(cpu_pmu->hw_events);
> > +	unregister_cpu_notifier(&cpu_pmu->hotplug_nb);
> >  }
> >
> 
> I would expect the order to be the other way, but it probably doesn't
> matter all that much on the registration error path.

I've flipped this around locally, I agree it's nicer the other way
around.

Thanks for the review, it's much appreciated.

Thanks,
Mark.