Re: [PATCH 2/3] irq: Track the interrupt timings

From: Thomas Gleixner
Date: Thu Jun 22 2017 - 10:47:51 EST


On Tue, 20 Jun 2017, Daniel Lezcano wrote:
> +
> +struct irq_timings {
> + u64 values[IRQ_TIMINGS_SIZE]; /* our circular buffer */
> + unsigned int count; /* Number of interruptions since last inspection */

Groan. These tail comments are horrible.

Please make the struct member names tabular aligned and add proper kernel
doc comments if you want to add useful documentations for the fields.

> +};
> +
> +DECLARE_PER_CPU(struct irq_timings, irq_timings);
> +
> +static inline void remove_timings(struct irq_desc *desc)

irq_remove_timings

> +{
> + desc->istate &= ~IRQS_TIMINGS;
> +}
> +
> +static inline void setup_timings(struct irq_desc *desc, struct irqaction *act)

...

> +{
> + /*
> + * We don't need the measurement because the idle code already
> + * knows the next expiry event.
> + */
> + if (act->flags & __IRQF_TIMER)
> + return;
> +
> + desc->istate |= IRQS_TIMINGS;
> +}
> +
> +extern void irq_timings_enable(void);
> +extern void irq_timings_disable(void);
> +
> +extern struct static_key_false irq_timing_enabled;

DECLARE_STATIC_KEY_FALSE

> +/*
> + * The interrupt number and the timestamp are encoded into a single
> + * u64 variable to optimize the size.
> + * 48 bit time stamp and 16 bit IRQ number is way sufficient.
> + * Who cares an IRQ after 78 hours of idle time?
> + */
> +static inline u64 irq_timing_encode(u64 timestamp, int irq)
> +{
> + return (timestamp << 16) | irq;
> +}
> +
> +static inline void irq_timing_decode(u64 value, u64 *timestamp, int *irq)

What's wrong with using a return value instead of void?

> +{
> + *timestamp = value >> 16;
> + *irq = value & U16_MAX;
> +}

Thanks,

tglx