Re: [PATCH v2 2/3] ftrace: add tracepoint for hrtimer

[Thomas Gleixner <tglx@linutronix.de>]

On Mon, 6 Jul 2009, Xiao Guangrong wrote:
> +TRACE_EVENT(hrtimer_expire,
> +
> +	TP_PROTO(struct hrtimer *timer, s64 now),
> +
> +	TP_ARGS(timer, now),
> +
> +	TP_STRUCT__entry(
> +		__field( void *,	timer		)
> +		__field( void *,	function	)

  Again, function is not necessary here.

> +		__field( s64,		now		)
> +	),
> +
> +	TP_fast_assign(
> +		__entry->timer		= timer;
> +		__entry->function	= timer->function;
> +		__entry->now		= now;
> +	),
> +
> +	TP_printk("timer=%p func=%pf now=%llu", __entry->timer, __entry->function,
> +		 (unsigned long long)__entry->now)
> +);
> +
> +TRACE_EVENT(hrtimer_cancel,
> +
> +	TP_PROTO(struct hrtimer *timer),
> +	TP_ARGS(timer),
> +	TP_STRUCT__entry(
> +		__field( void *,	timer		)
> +		__field( void *,	function	)

  Ditto.

> +static inline void debug_and_trace_hrtimer_expire(struct hrtimer *timer,
> +                                                  s64 now)

  Technically not necessary.

>  
> +			debug_and_trace_hrtimer_expire(timer, basenow.tv64);
>  			__run_hrtimer(timer);

  Here I'd like to see a callback_done entry as well, so we can check
  for long lasting callbacks.

>  		}
>  		spin_unlock(&cpu_base->lock);
> @@ -1436,6 +1463,7 @@ void hrtimer_run_queues(void)
>  					hrtimer_get_expires_tv64(timer))
>  				break;
>  
> +			debug_and_trace_hrtimer_expire(timer, base->softirq_time.tv64);
>  			__run_hrtimer(timer);

  Ditto.

Thanks,

	tglx