Re: [Update][PATCH 3/3] cpufreq: governor: Replace timers with utilization update callbacks

[Rafael J. Wysocki]

On Thursday, February 04, 2016 10:19:59 AM Viresh Kumar wrote:
> On 03-02-16, 02:16, Rafael J. Wysocki wrote:
> > Index: linux-pm/drivers/cpufreq/cpufreq_governor.c
> > -void gov_add_timers(struct cpufreq_policy *policy, unsigned int delay)
> > +void gov_set_update_util(struct cpu_common_dbs_info *shared,
> > +			 unsigned int delay_us)
> >  {
> > +	struct cpufreq_policy *policy = shared->policy;
> >  	struct dbs_data *dbs_data = policy->governor_data;
> > -	struct cpu_dbs_info *cdbs;
> >  	int cpu;
> >  
> > +	shared->sample_delay_ns = delay_us * NSEC_PER_USEC;
> > +	shared->time_stamp = ktime_get();
> > +
> >  	for_each_cpu(cpu, policy->cpus) {
> > -		cdbs = dbs_data->cdata->get_cpu_cdbs(cpu);
> > -		cdbs->timer.expires = jiffies + delay;
> > -		add_timer_on(&cdbs->timer, cpu);
> > +		struct cpu_dbs_info *cdbs = dbs_data->cdata->get_cpu_cdbs(cpu);
> > +
> > +		cdbs->last_sample_time = 0;
> > +		cpufreq_set_update_util_data(cpu, &cdbs->update_util);
> 
> Why no synchronize_rcu() here?

Because it is not needed.  This always changes a NULL pointer into a non-NULL.

> This can be called from ondemand governor on sampling-rate updates ..

But that calls gov_cancel_work() before, right?

> 
> >  	}
> >  }
> > -EXPORT_SYMBOL_GPL(gov_add_timers);
> > +EXPORT_SYMBOL_GPL(gov_set_update_util);
> >  
> > -static inline void gov_cancel_timers(struct cpufreq_policy *policy)
> > +static inline void gov_clear_update_util(struct cpufreq_policy *policy)
> >  {
> > -	struct dbs_data *dbs_data = policy->governor_data;
> > -	struct cpu_dbs_info *cdbs;
> >  	int i;
> >  
> > -	for_each_cpu(i, policy->cpus) {
> > -		cdbs = dbs_data->cdata->get_cpu_cdbs(i);
> > -		del_timer_sync(&cdbs->timer);
> > -	}
> > +	for_each_cpu(i, policy->cpus)
> > +		cpufreq_set_update_util_data(i, NULL);
> > +
> > +	synchronize_rcu();
> >  }
> >  
> >  void gov_cancel_work(struct cpu_common_dbs_info *shared)
> >  {
> > -	/* Tell dbs_timer_handler() to skip queuing up work items. */
> > +	/* Tell dbs_update_util_handler() to skip queuing up work items. */
> >  	atomic_inc(&shared->skip_work);
> >  	/*
> > -	 * If dbs_timer_handler() is already running, it may not notice the
> > -	 * incremented skip_work, so wait for it to complete to prevent its work
> > -	 * item from being queued up after the cancel_work_sync() below.
> > -	 */
> > -	gov_cancel_timers(shared->policy);
> > -	/*
> > -	 * In case dbs_timer_handler() managed to run and spawn a work item
> > -	 * before the timers have been canceled, wait for that work item to
> > -	 * complete and then cancel all of the timers set up by it.  If
> > -	 * dbs_timer_handler() runs again at that point, it will see the
> > -	 * positive value of skip_work and won't spawn any more work items.
> > +	 * If dbs_update_util_handler() is already running, it may not notice
> > +	 * the incremented skip_work, so wait for it to complete to prevent its
> > +	 * work item from being queued up after the cancel_work_sync() below.
> >  	 */
> > +	gov_clear_update_util(shared->policy);
> >  	cancel_work_sync(&shared->work);
> 
> How are we sure that the irq-work can't be pending at this point of
> time, which will queue the above works again ?

Good point.  The irq_work has to be waited for here too.

> > -	gov_cancel_timers(shared->policy);
> >  	atomic_set(&shared->skip_work, 0);
> >  }
> >  EXPORT_SYMBOL_GPL(gov_cancel_work);
> >  
> > -/* Will return if we need to evaluate cpu load again or not */
> > -static bool need_load_eval(struct cpu_common_dbs_info *shared,
> > -			   unsigned int sampling_rate)
> > -{
> > -	if (policy_is_shared(shared->policy)) {
> > -		ktime_t time_now = ktime_get();
> > -		s64 delta_us = ktime_us_delta(time_now, shared->time_stamp);
> > -
> > -		/* Do nothing if we recently have sampled */
> > -		if (delta_us < (s64)(sampling_rate / 2))
> > -			return false;
> > -		else
> > -			shared->time_stamp = time_now;
> > -	}
> > -
> > -	return true;
> > -}
> > -
> >  static void dbs_work_handler(struct work_struct *work)
> >  {
> >  	struct cpu_common_dbs_info *shared = container_of(work, struct
> > @@ -235,14 +212,10 @@ static void dbs_work_handler(struct work
> >  	struct cpufreq_policy *policy;
> >  	struct dbs_data *dbs_data;
> >  	unsigned int sampling_rate, delay;
> > -	bool eval_load;
> >  
> >  	policy = shared->policy;
> >  	dbs_data = policy->governor_data;
> >  
> > -	/* Kill all timers */
> > -	gov_cancel_timers(policy);
> > -
> >  	if (dbs_data->cdata->governor == GOV_CONSERVATIVE) {
> >  		struct cs_dbs_tuners *cs_tuners = dbs_data->tuners;
> >  
> > @@ -253,37 +226,53 @@ static void dbs_work_handler(struct work
> >  		sampling_rate = od_tuners->sampling_rate;
> >  	}
> >  
> > -	eval_load = need_load_eval(shared, sampling_rate);
> > -
> >  	/*
> > -	 * Make sure cpufreq_governor_limits() isn't evaluating load in
> > +	 * Make sure cpufreq_governor_limits() isn't evaluating load or the
> > +	 * ondemand governor isn't reading the time stamp and sampling rate in
> >  	 * parallel.
> >  	 */
> >  	mutex_lock(&shared->timer_mutex);
> > -	delay = dbs_data->cdata->gov_dbs_timer(policy, eval_load);
> > +	delay = dbs_data->cdata->gov_dbs_timer(policy);
> > +	shared->sample_delay_ns = jiffies_to_nsecs(delay);
> > +	shared->time_stamp = ktime_get();
> >  	mutex_unlock(&shared->timer_mutex);
> >  
> > +	smp_mb__before_atomic();
> 
> And why is this required exactly ? Maybe a comment as well to clarify
> this as this isn't obvious ?

OK, you have a point.

This relies on the atomic_dec() below to happen after sample_delay_ns has
been updated, to prevent dbs_update_util_handler() from using a stale
value.

> >  	atomic_dec(&shared->skip_work);
> > +}
> >  
> > -	gov_add_timers(policy, delay);
> > +static void dbs_irq_work(struct irq_work *irq_work)
> > +{
> > +	struct cpu_common_dbs_info *shared;
> > +
> > +	shared = container_of(irq_work, struct cpu_common_dbs_info, irq_work);
> > +	schedule_work(&shared->work);
> >  }
> >  
> > -static void dbs_timer_handler(unsigned long data)
> > +static void dbs_update_util_handler(struct update_util_data *data, u64 time,
> > +				    unsigned long util, unsigned long max)
> >  {
> > -	struct cpu_dbs_info *cdbs = (struct cpu_dbs_info *)data;
> > +	struct cpu_dbs_info *cdbs = container_of(data, struct cpu_dbs_info, update_util);
> >  	struct cpu_common_dbs_info *shared = cdbs->shared;
> >  
> >  	/*
> > -	 * Timer handler may not be allowed to queue the work at the moment,
> > -	 * because:
> > -	 * - Another timer handler has done that
> > -	 * - We are stopping the governor
> > -	 * - Or we are updating the sampling rate of the ondemand governor
> > +	 * The work may not be allowed to be queued up right now.
> > +	 * Possible reasons:
> > +	 * - Work has already been queued up or is in progress.
> > +	 * - The governor is being stopped.
> > +	 * - It is too early (too little time from the previous sample).
> >  	 */
> > -	if (atomic_inc_return(&shared->skip_work) > 1)
> > -		atomic_dec(&shared->skip_work);
> > -	else
> > -		queue_work(system_wq, &shared->work);
> > +	if (atomic_inc_return(&shared->skip_work) == 1) {
> > +		u64 delta_ns;
> > +
> > +		delta_ns = time - cdbs->last_sample_time;
> > +		if ((s64)delta_ns >= shared->sample_delay_ns) {
> > +			cdbs->last_sample_time = time;
> > +			irq_work_queue_on(&shared->irq_work, smp_processor_id());
> > +			return;
> > +		}
> > +	}
> > +	atomic_dec(&shared->skip_work);
> >  }
> >  
> >  static void set_sampling_rate(struct dbs_data *dbs_data,
> > @@ -467,9 +456,6 @@ static int cpufreq_governor_start(struct
> >  		io_busy = od_tuners->io_is_busy;
> >  	}
> >  
> > -	shared->policy = policy;
> > -	shared->time_stamp = ktime_get();
> > -
> >  	for_each_cpu(j, policy->cpus) {
> >  		struct cpu_dbs_info *j_cdbs = cdata->get_cpu_cdbs(j);
> >  		unsigned int prev_load;
> > @@ -485,10 +471,10 @@ static int cpufreq_governor_start(struct
> >  		if (ignore_nice)
> >  			j_cdbs->prev_cpu_nice = kcpustat_cpu(j).cpustat[CPUTIME_NICE];
> >  
> > -		__setup_timer(&j_cdbs->timer, dbs_timer_handler,
> > -			      (unsigned long)j_cdbs,
> > -			      TIMER_DEFERRABLE | TIMER_IRQSAFE);
> > +		j_cdbs->update_util.func = dbs_update_util_handler;
> >  	}
> > +	shared->policy = policy;
> > +	init_irq_work(&shared->irq_work, dbs_irq_work);
> >  
> >  	if (cdata->governor == GOV_CONSERVATIVE) {
> >  		struct cs_cpu_dbs_info_s *cs_dbs_info =
> > @@ -505,7 +491,7 @@ static int cpufreq_governor_start(struct
> >  		od_ops->powersave_bias_init_cpu(cpu);
> >  	}
> >  
> > -	gov_add_timers(policy, delay_for_sampling_rate(sampling_rate));
> > +	gov_set_update_util(shared, sampling_rate);
> >  	return 0;
> >  }
> >  
> > Index: linux-pm/drivers/cpufreq/cpufreq_ondemand.c
> > ===================================================================
> > --- linux-pm.orig/drivers/cpufreq/cpufreq_ondemand.c
> > +++ linux-pm/drivers/cpufreq/cpufreq_ondemand.c
> > @@ -191,7 +191,7 @@ static void od_check_cpu(int cpu, unsign
> >  	}
> >  }
> >  
> > -static unsigned int od_dbs_timer(struct cpufreq_policy *policy, bool modify_all)
> > +static unsigned int od_dbs_timer(struct cpufreq_policy *policy)
> >  {
> >  	struct dbs_data *dbs_data = policy->governor_data;
> >  	unsigned int cpu = policy->cpu;
> > @@ -200,9 +200,6 @@ static unsigned int od_dbs_timer(struct
> >  	struct od_dbs_tuners *od_tuners = dbs_data->tuners;
> >  	int delay = 0, sample_type = dbs_info->sample_type;
> 
> Perhaps, the delay = 0 can be dropped now and ...
> 
> >  
> > -	if (!modify_all)
> > -		goto max_delay;
> > -
> >  	/* Common NORMAL_SAMPLE setup */
> >  	dbs_info->sample_type = OD_NORMAL_SAMPLE;
> >  	if (sample_type == OD_SUB_SAMPLE) {
> > @@ -218,7 +215,6 @@ static unsigned int od_dbs_timer(struct
> >  		}
> >  	}
> >  
> > -max_delay:
> >  	if (!delay)
> >  		delay = delay_for_sampling_rate(od_tuners->sampling_rate
> >  				* dbs_info->rate_mult);
> 
> ^^ can be moved to the else part of above block ..

Both this and the above are valid observation, but those changes should be
made in a follow-up patch IMO.

> > @@ -264,7 +260,7 @@ static void update_sampling_rate(struct
> >  		struct od_cpu_dbs_info_s *dbs_info;
> >  		struct cpu_dbs_info *cdbs;
> >  		struct cpu_common_dbs_info *shared;
> > -		unsigned long next_sampling, appointed_at;
> > +		ktime_t next_sampling, appointed_at;
> >  
> >  		dbs_info = &per_cpu(od_cpu_dbs_info, cpu);
> >  		cdbs = &dbs_info->cdbs;
> > @@ -292,16 +288,19 @@ static void update_sampling_rate(struct
> >  			continue;
> >  
> >  		/*
> > -		 * Checking this for any CPU should be fine, timers for all of
> > -		 * them are scheduled together.
> > +		 * Checking this for any CPU sharing the policy should be fine,
> > +		 * they are all scheduled to sample at the same time.
> >  		 */
> > -		next_sampling = jiffies + usecs_to_jiffies(new_rate);
> > -		appointed_at = dbs_info->cdbs.timer.expires;
> > +		next_sampling = ktime_add_us(ktime_get(), new_rate);
> >  
> > -		if (time_before(next_sampling, appointed_at)) {
> > -			gov_cancel_work(shared);
> > -			gov_add_timers(policy, usecs_to_jiffies(new_rate));
> > +		mutex_lock(&shared->timer_mutex);
> 
> Why is taking this lock important here ?

Because this reads both time_stamp and sample_delay_ns and uses them in
a computation.  If they happen to be out of sync, this surely isn't right. 

> > +		appointed_at = ktime_add_ns(shared->time_stamp,
> 
> Also I failed to understand why we need time_stamp variable at all?
> Why can't we use last_sample_time ?

Because the time base for last_sample_time may be different, so comparing it
to the return value of ktime_get() may not lead to correct decisions, so to
speak.

> > +					    shared->sample_delay_ns);
> > +		mutex_unlock(&shared->timer_mutex);
> >  
> > +		if (ktime_before(next_sampling, appointed_at)) {
> > +			gov_cancel_work(shared);
> > +			gov_set_update_util(shared, new_rate);
> 
> You don't need to a complete update here, the pointers are all fine.

I do, but that's not because of the pointers.

Effectively, I need to change sample_delay_ns and that's the most startghtforward
way to do that safely.

It may not be the most efficient, but this is not a fast path anyway.

> >  		}
> >  	}
> 
> 

Thanks,
Rafael