lkml.org 
[lkml]   [2013]   [Jul]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH v5 5/7] thermal:boost: Automatic enable/disable of BOOST feature
On Thu, 4 Jul 2013 17:19:04 +0000
"R, Durgadoss" <durgadoss.r@intel.com> wrote:
Hi,

> Hi Lukasz,
>
> > -----Original Message-----
> > From: linux-pm-owner@vger.kernel.org [mailto:linux-pm-
> > owner@vger.kernel.org] On Behalf Of Lukasz Majewski
> > Sent: Thursday, July 04, 2013 2:20 PM
> > To: Viresh Kumar; Rafael J. Wysocki; Zhang, Rui; Eduardo Valentin
> > Cc: cpufreq@vger.kernel.org; Linux PM list; Jonghwa Lee; Lukasz
> > Majewski; l.majewski@majess.pl; linux-kernel; Andre Przywara;
> > Daniel Lezcano; Kukjin Kim; Myungjoo Ham
> > Subject: [PATCH v5 5/7] thermal:boost: Automatic enable/disable of
> > BOOST feature
> >
> > This patch provides auto disable/enable operation for boost. When
> > any defined trip point is passed, the boost is disabled.
> > In that moment thermal monitor workqueue is woken up and it monitors
> > if the device temperature drops below 75% of the smallest trip
> > point. When device cools down, the boost is enabled again.
> >
> > Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
> > Signed-off-by: Myungjoo Ham <myungjoo.ham@samsung.com>
> >
> > ---
> > Changes for v5:
> > - Move boost disable code from cpu_cooling.c to thermal_core.c
> > (to handle_non_critical_trips)
> > - Extent struct thermal_zone_device by adding overheated bool flag
> > - Implement auto enable of boost after device cools down
> > - Introduce boost_polling flag, which indicates if thermal uses
> > it's predefined pool delay or has woken up thermal workqueue only
> > to wait until device cools down.
> >
> > Changes for v4:
> > - New patch
> >
> > drivers/thermal/thermal_core.c | 31
> > +++++++++++++++++++++++++++++++ include/linux/thermal.h |
> > 2 ++ 2 files changed, 33 insertions(+)
> >
> > diff --git a/drivers/thermal/thermal_core.c
> > b/drivers/thermal/thermal_core.c index d755440..12adbad 100644
> > --- a/drivers/thermal/thermal_core.c
> > +++ b/drivers/thermal/thermal_core.c
> > @@ -33,6 +33,7 @@
> > #include <linux/idr.h>
> > #include <linux/thermal.h>
> > #include <linux/reboot.h>
> > +#include <linux/cpufreq.h>
> > #include <net/netlink.h>
> > #include <net/genetlink.h>
> >
> > @@ -326,6 +327,15 @@ static void monitor_thermal_zone(struct
> > thermal_zone_device *tz)
> > static void handle_non_critical_trips(struct thermal_zone_device
> > *tz, int trip, enum thermal_trip_type trip_type)
> > {
> > + if (cpufreq_boost_supported()) {
> > + tz->overheated = true;
> > + cpufreq_boost_trigger_state(0);
> > + if (!tz->polling_delay) {
> > + tz->boost_polling = true;
> > + tz->polling_delay = 1000;
> > + }
> > + }
> > +
> > if (tz->governor)
> > tz->governor->throttle(tz, trip);
> > }
> > @@ -453,6 +463,27 @@ static void thermal_zone_device_check(struct
> > work_struct *work)
> > struct thermal_zone_device *tz = container_of(work, struct
> > thermal_zone_device,
> > poll_queue.work);
> > + long trip_temp;
> > +
> > + if (cpufreq_boost_supported() && tz->overheated) {
>
> Not all thermal drivers support trip points. So, we first need a
> if (tz->ops->get_trip_temp) check here.

Ok, thanks for tip. Bluntly speaking, I thought, that all SoCs supported
by thermal set trip points.

>
> > + tz->ops->get_trip_temp(tz, 0, &trip_temp);
> > + /*
> > + * Enable boost again only when current
> > temperature is less
> > + * than 75% of trip_temp[0]
> > + */
> > + if ((tz->temperature + (trip_temp >> 2)) <
> > trip_temp) {
>
> Another way would be to use the get_trend APIs for this thermal zone.
> If the trend is cooling we can re-enable boost otherwise not.

Trend evaluation seems like a good complementary idea.

However, I would also like to have the relative temperature drop
measurement (if possible) like above (to 75% of the first trip point).

Then I would be more confident, that everything cooled down and that I
can start boost again.

> > + tz->overheated = false;
> > + if (tz->boost_polling) {
> > + tz->boost_polling = false;
> > + tz->polling_delay = 0;
> > + monitor_thermal_zone(tz);
> > + }
>
> Overall, I believe this will work well only if the thermal zone is
> CPU.

My assumption:

When I enable boost at CPU, then I also shall cool down the CPU. And
the CPU zone seemed a natural choice.

However I might be missing something, so hints are welcome.

>
> Another suggestion is: We tried hard to remove all throttling logic
> from thermal_core.c.

By throttling logic you mean:
if ((tz->temperature + (trip_temp >> 2)) and other conditions (like
trend measurement)?
> May be we should include this kind of logic in
> step_wise.c ?

It sounds interesting (since ->throttle at thermal_core.c is called
always when needed), but I'm afraid of a code duplication when one
use Boost with fair_share or other thermal governor.

> Rui/Eduardo: Any thoughts on this ?
>
> Thanks,
> Durga
> > +
> > + cpufreq_boost_trigger_state(1);
> > + return;
> > + }
> > + }
> > +
> > thermal_zone_device_update(tz);
> > }
> >
> > diff --git a/include/linux/thermal.h b/include/linux/thermal.h
> > index a386a1c..f1aa3c2 100644
> > --- a/include/linux/thermal.h
> > +++ b/include/linux/thermal.h
> > @@ -172,6 +172,8 @@ struct thermal_zone_device {
> > int emul_temperature;
> > int passive;
> > unsigned int forced_passive;
> > + bool overheated;
> > + bool boost_polling;
> > const struct thermal_zone_device_ops *ops;
> > const struct thermal_zone_params *tzp;
> > struct thermal_governor *governor;
> > --
> > 1.7.10.4
> >
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-pm"
> > in the body of a message to majordomo@vger.kernel.org
> > More majordomo info at http://vger.kernel.org/majordomo-info.html



\
 
 \ /
  Last update: 2013-07-05 00:41    [W:0.301 / U:4.436 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site