The get_trip_hyst() ops is optional. It results all around the thermal
framework code a check against the ops pointer and different action
adding more complexity and making the code less readable and
understandable.

A zero hysteresis value is perfectly valid, so instead of adding more
circumvolutions in the code, create unconditionnaly the hysteresis and
use the thermal_zone_get_trip() function which returns a zero
hysteresis if the get_trip_hyst() is not defined.

Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
---
 drivers/thermal/thermal_sysfs.c | 25 +++++++++----------------
 1 file changed, 9 insertions(+), 16 deletions(-)

diff --git a/drivers/thermal/thermal_sysfs.c b/drivers/thermal/thermal_sysfs.c
index 75f6b151a58b..18cdd7cd0008 100644
--- a/drivers/thermal/thermal_sysfs.c
+++ b/drivers/thermal/thermal_sysfs.c
@@ -424,23 +424,20 @@ static int create_trip_attrs(struct thermal_zone_device *tz, int mask)
 		return -ENOMEM;
 	}
 
-	if (tz->ops->get_trip_hyst) {
-		tz->trip_hyst_attrs = kcalloc(tz->num_trips,
-					      sizeof(*tz->trip_hyst_attrs),
-					      GFP_KERNEL);
-		if (!tz->trip_hyst_attrs) {
-			kfree(tz->trip_type_attrs);
-			kfree(tz->trip_temp_attrs);
-			return -ENOMEM;
-		}
+	tz->trip_hyst_attrs = kcalloc(tz->num_trips,
+				      sizeof(*tz->trip_hyst_attrs),
+				      GFP_KERNEL);
+	if (!tz->trip_hyst_attrs) {
+		kfree(tz->trip_type_attrs);
+		kfree(tz->trip_temp_attrs);
+		return -ENOMEM;
 	}
 
 	attrs = kcalloc(tz->num_trips * 3 + 1, sizeof(*attrs), GFP_KERNEL);
 	if (!attrs) {
 		kfree(tz->trip_type_attrs);
 		kfree(tz->trip_temp_attrs);
-		if (tz->ops->get_trip_hyst)
-			kfree(tz->trip_hyst_attrs);
+		kfree(tz->trip_hyst_attrs);
 		return -ENOMEM;
 	}
 
@@ -473,9 +470,6 @@ static int create_trip_attrs(struct thermal_zone_device *tz, int mask)
 		}
 		attrs[indx + tz->num_trips] = &tz->trip_temp_attrs[indx].attr.attr;
 
-		/* create Optional trip hyst attribute */
-		if (!tz->ops->get_trip_hyst)
-			continue;
 		snprintf(tz->trip_hyst_attrs[indx].name, THERMAL_NAME_LENGTH,
 			 "trip_point_%d_hyst", indx);
 
@@ -512,8 +506,7 @@ static void destroy_trip_attrs(struct thermal_zone_device *tz)
 
 	kfree(tz->trip_type_attrs);
 	kfree(tz->trip_temp_attrs);
-	if (tz->ops->get_trip_hyst)
-		kfree(tz->trip_hyst_attrs);
+	kfree(tz->trip_hyst_attrs);
 	kfree(tz->trips_attribute_group.attrs);
 }
 
-- 
2.34.1

On Wed, 2022-09-21 at 11:42 +0200, Daniel Lezcano wrote:
> The get_trip_hyst() ops is optional. It results all around the
> thermal
> framework code a check against the ops pointer and different action
> adding more complexity and making the code less readable and
> understandable.
> 
> A zero hysteresis value is perfectly valid, so instead of adding more
> circumvolutions in the code, create unconditionnaly the hysteresis
> and
> use the thermal_zone_get_trip() function which returns a zero
> hysteresis if the get_trip_hyst() is not defined.

I think an alternative way is to use the .is_visible() callback to make
this cleanup transparent to userspace.

With this, we can remove the
	if (tz->ops->set_trip_hyst)check in create_trip_attrs() as well.

But you make the cal. :)

thanks,
rui

On 23/09/2022 16:19, Zhang Rui wrote:
> On Wed, 2022-09-21 at 11:42 +0200, Daniel Lezcano wrote:
>> The get_trip_hyst() ops is optional. It results all around the
>> thermal
>> framework code a check against the ops pointer and different action
>> adding more complexity and making the code less readable and
>> understandable.
>>
>> A zero hysteresis value is perfectly valid, so instead of adding more
>> circumvolutions in the code, create unconditionnaly the hysteresis
>> and
>> use the thermal_zone_get_trip() function which returns a zero
>> hysteresis if the get_trip_hyst() is not defined.
> 
> I think an alternative way is to use the .is_visible() callback to make
> this cleanup transparent to userspace.

This would apply for the group not per trip point file, no?


> With this, we can remove the
> 	if (tz->ops->set_trip_hyst)check in create_trip_attrs() as well.
> 
> But you make the cal. :)
> 
> thanks,
> rui
> 
> 


-- 
<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