Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1691329 > unrolled thread

[PATCH V3 2/9] cpufreq: Use transition_delay_us for legacy governors as well

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  [PATCH V3 2/9] cpufreq: Use transition_delay_us for legacy governors as well Viresh Kumar <viresh.kumar@linaro.org> - 2017-07-19 12:20 +0200
    Re: [PATCH V3 2/9] cpufreq: Use transition_delay_us for legacy  governors as well Peter Zijlstra <peterz@infradead.org> - 2017-07-24 18:20 +0200
      Re: [PATCH V3 2/9] cpufreq: Use transition_delay_us for legacy  governors as well Viresh Kumar <viresh.kumar@linaro.org> - 2017-07-28 06:50 +0200

#1691329 — [PATCH V3 2/9] cpufreq: Use transition_delay_us for legacy governors as well

The policy->transition_delay_us field is used only by the schedutil
governor currently, and this field describes how fast the driver wants
the cpufreq governor to change CPUs frequency. It should rather be a
common thing across all governors, as it doesn't have any schedutil
dependency here.

Create a new helper cpufreq_policy_transition_delay_us() to get the
transition delay across all governors.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
---
 drivers/cpufreq/cpufreq.c          | 15 +++++++++++++++
 drivers/cpufreq/cpufreq_governor.c |  9 +--------
 include/linux/cpufreq.h            |  1 +
 kernel/sched/cpufreq_schedutil.c   | 11 +----------
 4 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index 9bf97a366029..c426d21822f7 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -524,6 +524,21 @@ unsigned int cpufreq_driver_resolve_freq(struct cpufreq_policy *policy,
 }
 EXPORT_SYMBOL_GPL(cpufreq_driver_resolve_freq);
 
+unsigned int cpufreq_policy_transition_delay_us(struct cpufreq_policy *policy)
+{
+	unsigned int latency;
+
+	if (policy->transition_delay_us)
+		return policy->transition_delay_us;
+
+	latency = policy->cpuinfo.transition_latency / NSEC_PER_USEC;
+	if (latency)
+		return latency * LATENCY_MULTIPLIER;
+
+	return LATENCY_MULTIPLIER;
+}
+EXPORT_SYMBOL_GPL(cpufreq_policy_transition_delay_us);
+
 /*********************************************************************
  *                          SYSFS INTERFACE                          *
  *********************************************************************/
diff --git a/drivers/cpufreq/cpufreq_governor.c b/drivers/cpufreq/cpufreq_governor.c
index 858081f9c3d7..eed069ecfd5e 100644
--- a/drivers/cpufreq/cpufreq_governor.c
+++ b/drivers/cpufreq/cpufreq_governor.c
@@ -389,7 +389,6 @@ int cpufreq_dbs_governor_init(struct cpufreq_policy *policy)
 	struct dbs_governor *gov = dbs_governor_of(policy);
 	struct dbs_data *dbs_data;
 	struct policy_dbs_info *policy_dbs;
-	unsigned int latency;
 	int ret = 0;
 
 	/* State should be equivalent to EXIT */
@@ -428,13 +427,7 @@ int cpufreq_dbs_governor_init(struct cpufreq_policy *policy)
 	if (ret)
 		goto free_policy_dbs_info;
 
-	/* policy latency is in ns. Convert it to us first */
-	latency = policy->cpuinfo.transition_latency / 1000;
-	if (latency == 0)
-		latency = 1;
-
-	/* Bring kernel and HW constraints together */
-	dbs_data->sampling_rate = LATENCY_MULTIPLIER * latency;
+	dbs_data->sampling_rate = cpufreq_policy_transition_delay_us(policy);
 
 	if (!have_governor_per_policy())
 		gov->gdbs_data = dbs_data;
diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h
index 02aec384cab9..aaadfc543f63 100644
--- a/include/linux/cpufreq.h
+++ b/include/linux/cpufreq.h
@@ -523,6 +523,7 @@ int __cpufreq_driver_target(struct cpufreq_policy *policy,
 				   unsigned int relation);
 unsigned int cpufreq_driver_resolve_freq(struct cpufreq_policy *policy,
 					 unsigned int target_freq);
+unsigned int cpufreq_policy_transition_delay_us(struct cpufreq_policy *policy);
 int cpufreq_register_governor(struct cpufreq_governor *governor);
 void cpufreq_unregister_governor(struct cpufreq_governor *governor);
 
diff --git a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c
index 29a397067ffa..89c4dd9777bb 100644
--- a/kernel/sched/cpufreq_schedutil.c
+++ b/kernel/sched/cpufreq_schedutil.c
@@ -528,16 +528,7 @@ static int sugov_init(struct cpufreq_policy *policy)
 		goto stop_kthread;
 	}
 
-	if (policy->transition_delay_us) {
-		tunables->rate_limit_us = policy->transition_delay_us;
-	} else {
-		unsigned int lat;
-
-		tunables->rate_limit_us = LATENCY_MULTIPLIER;
-		lat = policy->cpuinfo.transition_latency / NSEC_PER_USEC;
-		if (lat)
-			tunables->rate_limit_us *= lat;
-	}
+	tunables->rate_limit_us = cpufreq_policy_transition_delay_us(policy);
 
 	policy->governor_data = sg_policy;
 	sg_policy->tunables = tunables;
-- 
2.13.0.71.gd7076ec9c9cb

[toc] | [next] | [standalone]

#1694870 — Re: [PATCH V3 2/9] cpufreq: Use transition_delay_us for legacy governors as well

On Wed, Jul 19, 2017 at 03:42:42PM +0530, Viresh Kumar wrote:
> The policy->transition_delay_us field is used only by the schedutil
> governor currently, and this field describes how fast the driver wants
> the cpufreq governor to change CPUs frequency. It should rather be a
> common thing across all governors, as it doesn't have any schedutil
> dependency here.
> 
> Create a new helper cpufreq_policy_transition_delay_us() to get the
> transition delay across all governors.
> 
> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
> ---
>  drivers/cpufreq/cpufreq.c          | 15 +++++++++++++++
>  drivers/cpufreq/cpufreq_governor.c |  9 +--------
>  include/linux/cpufreq.h            |  1 +
>  kernel/sched/cpufreq_schedutil.c   | 11 +----------
>  4 files changed, 18 insertions(+), 18 deletions(-)
> 
> diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
> index 9bf97a366029..c426d21822f7 100644
> --- a/drivers/cpufreq/cpufreq.c
> +++ b/drivers/cpufreq/cpufreq.c
> @@ -524,6 +524,21 @@ unsigned int cpufreq_driver_resolve_freq(struct cpufreq_policy *policy,
>  }
>  EXPORT_SYMBOL_GPL(cpufreq_driver_resolve_freq);
>  
> +unsigned int cpufreq_policy_transition_delay_us(struct cpufreq_policy *policy)
> +{
> +	unsigned int latency;
> +
> +	if (policy->transition_delay_us)
> +		return policy->transition_delay_us;
> +
> +	latency = policy->cpuinfo.transition_latency / NSEC_PER_USEC;
> +	if (latency)
> +		return latency * LATENCY_MULTIPLIER;
> +
> +	return LATENCY_MULTIPLIER;
> +}
> +EXPORT_SYMBOL_GPL(cpufreq_policy_transition_delay_us);

I realize you're just moving code about, but _why_ are we doing that
division?

[toc] | [prev] | [next] | [standalone]

#1698466 — Re: [PATCH V3 2/9] cpufreq: Use transition_delay_us for legacy governors as well

On 24-07-17, 18:17, Peter Zijlstra wrote:
> On Wed, Jul 19, 2017 at 03:42:42PM +0530, Viresh Kumar wrote:
> > The policy->transition_delay_us field is used only by the schedutil
> > governor currently, and this field describes how fast the driver wants
> > the cpufreq governor to change CPUs frequency. It should rather be a
> > common thing across all governors, as it doesn't have any schedutil
> > dependency here.
> > 
> > Create a new helper cpufreq_policy_transition_delay_us() to get the
> > transition delay across all governors.
> > 
> > Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
> > ---
> >  drivers/cpufreq/cpufreq.c          | 15 +++++++++++++++
> >  drivers/cpufreq/cpufreq_governor.c |  9 +--------
> >  include/linux/cpufreq.h            |  1 +
> >  kernel/sched/cpufreq_schedutil.c   | 11 +----------
> >  4 files changed, 18 insertions(+), 18 deletions(-)
> > 
> > diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
> > index 9bf97a366029..c426d21822f7 100644
> > --- a/drivers/cpufreq/cpufreq.c
> > +++ b/drivers/cpufreq/cpufreq.c
> > @@ -524,6 +524,21 @@ unsigned int cpufreq_driver_resolve_freq(struct cpufreq_policy *policy,
> >  }
> >  EXPORT_SYMBOL_GPL(cpufreq_driver_resolve_freq);
> >  
> > +unsigned int cpufreq_policy_transition_delay_us(struct cpufreq_policy *policy)
> > +{
> > +	unsigned int latency;
> > +
> > +	if (policy->transition_delay_us)
> > +		return policy->transition_delay_us;
> > +
> > +	latency = policy->cpuinfo.transition_latency / NSEC_PER_USEC;
> > +	if (latency)
> > +		return latency * LATENCY_MULTIPLIER;
> > +
> > +	return LATENCY_MULTIPLIER;
> > +}
> > +EXPORT_SYMBOL_GPL(cpufreq_policy_transition_delay_us);
> 
> I realize you're just moving code about, but _why_ are we doing that
> division?

We are doing division by NSEC_PER_USEC because the values of
sampling_rate and rate_limit_us are in us, while transition_latency is
in ns.

We shouldn't be breaking the userspace ABI and so if we want, we can
actually make transition_latency be stored in us instead.

Though I am not sure why are we multiplying with LATENCY_MULTIPLIER
(1000) as that seems to be a huge value.

-- 
viresh

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web