Hi Rui,
[cut.]
> > +static void notify_user_space(struct thermal_zone_device *tz, int trip)
> > +{
> > + mutex_lock(&tz->lock);
> > +
> > + kobject_uevent(&tz->device.kobj, KOBJ_CHANGE);
> > +
> > + mutex_unlock(&tz->lock);
> > +}
> > +
> > +static void handle_non_critical_trips(struct thermal_zone_device *tz,
> > + int trip, enum thermal_trip_type trip_type)
> > +{
> > + int throttle_policy = THERMAL_STEP_WISE;
> > +
> > + if (tz->tzp)
> > + throttle_policy = tz->tzp->throttle_policy;
> > +
> > + switch (throttle_policy) {
> > + case THERMAL_FAIR_SHARE:
> > + fair_share_throttle(tz, trip);
> > + break;
> > + case THERMAL_STEP_WISE:
> > + step_wise_throttle(tz, trip);
> > + break;
> > + case THERMAL_USER_SPACE:
> > + notify_user_space(tz, trip);
> > + break;
> > + }
>
> this is a little different from what I thought.
> IMO, each policy should register its pointer to the thermal framework.
> and tz->tzp points to the pointer of the policy using.
> and then, in handle_non_critical_trips(), the code would be like this:
> if(tz->tzp)
> tz->tzp->throttle(tz, trip);
>
> But this is also okay for now, because we have only one callback for
> each policy.
I completely agree with you here. When I was doing my initial development,
I found that some policies need two arguments/ some need three etc..
That’s why could not do a function pointer implementation.
Will try to fix it up in the next version of the patches.
Thank you,
Durga
��.n��������+%������w��{.n�����{�����ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f
