Re: [PATCH v5 6/7] arm_pmu: Introduce pmu_irq_ops

From: Stephen Boyd
Date: Fri Jun 19 2020 - 04:33:18 EST


Quoting Alexandru Elisei (2020-06-18 03:51:24)
> Hi,
>
> On 6/17/20 9:23 PM, Stephen Boyd wrote:
> > Quoting Alexandru Elisei (2020-06-17 04:38:50)
> >> diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
> >> index df352b334ea7..17e5952d21e4 100644
> >> --- a/drivers/perf/arm_pmu.c
> >> +++ b/drivers/perf/arm_pmu.c
> >> @@ -26,8 +26,46 @@
> >>
> >> #include <asm/irq_regs.h>
> >>
> >> +static int armpmu_count_irq_users(const int irq);
> >> +
> >> +struct pmu_irq_ops {
> >> + void (*enable_pmuirq)(unsigned int irq);
> >> + void (*disable_pmuirq)(unsigned int irq);
> >> + void (*free_pmuirq)(unsigned int irq, int cpu, void __percpu *devid);
> > Does 'cpu' need to be signed?
>
> I'm not sure what you mean. The cpu argument comes from
> drivers/perf/arm_pmu_platform.c::armpmu_free_irqs -> arpmu_free_irq, where is the
> iterator variable used by the macro for_each_cpu. The documentation for the macro
> states:
>
> /**
> * for_each_cpu - iterate over every cpu in a mask
> * @cpu: the (optionally unsigned) integer iterator ^^^^^^^^^^^^^^^^^^^
>
> I could write a patch to convert to an unsigned int, but it seems like unnecessary
> churn to me.

Ok. It would be nice to make it unsigned in the arm_pmu_platform.c file.
Not required for this patch series.