Re: [PATCH V3 3/6] sched: idle: Get the next timer event and pass it the cpuidle framework

[Daniel Lezcano <daniel.lezcano@linaro.org>]

On 11/10/2014 01:43 PM, Peter Zijlstra wrote:
> On Fri, Nov 07, 2014 at 03:31:24PM +0100, Daniel Lezcano wrote:
>>   static void cpu_idle_loop(void)
>>   {
>> -	unsigned int latency_req;
>> +	unsigned int latency_req, next_timer_event;
>>
>>   	while (1) {
>>   		/*
>> @@ -221,6 +222,9 @@ static void cpu_idle_loop(void)
>>
>>   			latency_req = pm_qos_request(PM_QOS_CPU_DMA_LATENCY);
>>
>> +			next_timer_event =
>> +				ktime_to_us(tick_nohz_get_sleep_length());
>> +
>>   			/*
>>   			 * In poll mode we reenable interrupts and spin.
>>   			 *
>> @@ -238,7 +242,8 @@ static void cpu_idle_loop(void)
>>   			    tick_check_broadcast_expired())
>>   				cpu_idle_poll();
>>   			else
>> -				cpuidle_idle_call(latency_req);
>> +				cpuidle_idle_call(latency_req,
>> +						  next_timer_event);
>>
>>   			arch_cpu_idle_exit();
>>   		}
>
> Why do we want to query the next timer in the poll case? Afaict the
> other patches don't make use of this either.

Well, the direction I am taking when writing those cleanups is to have 
something like:

"I will sleep X usec, I have Y usec latency constraints". Grouping the 
latency req and the next timer allows to stick to the next changes.


-- 
  <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog