[PATCH 5/6] thermal: Only set passive_delay for forced passive cooling

[Frans Pop <elendil@planet.nl>]

Setting polling_delay is useless as passive_delay has priority,
so the value shown in proc isn't the actual polling delay. It
also gives the impression to the user that he can change the
polling interval through proc, while in fact he can't.

Also, unset passive_delay when the forced passive trip point is
unbound to allow polling to be disabled.

Signed-off-by: Frans Pop <elendil@planet.nl>
Cc: Matthew Garrett <mjg@redhat.com>
Cc: Zhang Rui <rui.zhang@intel.com>
---

I'm not sure why polling_delay was getting set here. Possibly the
intention was to set polling_frequency instead, which is in deci-
seconds and would thus explain the factor 10 between the values.
But even for polling_frequency there is IMO no reason to set it.

diff --git a/drivers/thermal/thermal_sys.c b/drivers/thermal/thermal_sys.c
index 2d13d0d..ceda0f1 100644
--- a/drivers/thermal/thermal_sys.c
+++ b/drivers/thermal/thermal_sys.c
@@ -241,6 +241,8 @@ passive_store(struct device *dev, struct device_attribute *attr,
 								 cdev);
 		}
 		mutex_unlock(&thermal_list_lock);
+		if (!tz->passive_delay)
+			tz->passive_delay = 1000;
 	} else if (!state && tz->forced_passive) {
 		mutex_lock(&thermal_list_lock);
 		list_for_each_entry(cdev, &thermal_cdev_list, node) {
@@ -251,17 +253,12 @@ passive_store(struct device *dev, struct device_attribute *attr,
 								   cdev);
 		}
 		mutex_unlock(&thermal_list_lock);
+		tz->passive_delay = 0;
 	}
 
 	tz->tc1 = 1;
 	tz->tc2 = 1;
 
-	if (!tz->passive_delay)
-		tz->passive_delay = 1000;
-
-	if (!tz->polling_delay)
-		tz->polling_delay = 10000;
-
 	tz->forced_passive = state;
 
 	thermal_zone_device_update(tz);