Re: PROBLEM: (regression) thermal zone - fan is turned on at higher temperature than it should

[yakui_zhao <yakui.zhao@intel.com>]

On Wed, 2009-05-06 at 03:05 +0800, Vladimir Zajac wrote:
> [1.] One line summary of the problem:
> (regression) thermal zone - fan is turned on at higher temperature than it should
> 
> [2.] Full description of the problem/report:
> After upgrade to 2.6.30-rc4 I noticed that my laptop's fan sometimes runs at lower speed than it used to.
> 2.6.29 does not have the bug.
> 
> The laptop (HP Compaq 6710b) has 5 ACPI cooling devices which correspond to different fan speeds. They are not turned on when I expect them to be.
> 
> Expected behavior: 
> Each active cooling device should be turned on when current temperature is equal or higher than its trip point.
> (This is how it worked in 2.6.29 and earlier versions and also how it is defined by ACPI specification.)
> 
> Actual behavior in 2.6.30-rc4: 
> Each cooling device is turned on only when current temperature is strictly greater than the corresponding trip point but not when current temperature is equal to the trip point.
> 
> It is caused by commit b1569e99c795bf83b4ddf41c4f1c42761ab7f75e:
> 
> author	Matthew Garrett <mjg59@srcf.ucam.org>
> 	Wed, 3 Dec 2008 17:55:32 +0000 (17:55 +0000)
> committer	Len Brown <len.brown@intel.com>
> 	Fri, 20 Feb 2009 23:41:56 +0000 (18:41 -0500)
> commit	b1569e99c795bf83b4ddf41c4f1c42761ab7f75e
> ACPI: move thermal trip handling to generic thermal layer
> 
> 
> This commit (probably unintentionally) also changed how kernel determines if a cooling device has to be activated:
>  before the commit: cooling device is activated if current temperature >= trip point temperature
>   after the commit: cooling device is activated if current temperature > trip point temperature
Do you mean that the issue can be fixed by changing the "temperature >
trip point" to "temperature >= trip point"?

If so, what you have done is reasonable.

Thanks.
> 
> I tried to change the trip point test back to "temp >= trip_temp" in drivers/thermal/termal_sys.c (attached patch). It fixes the issue on my laptop.
> 
> [3.] Keywords (i.e., modules, networking, kernel):
> acpi, thermal zone
> 
> [4.] Kernel information
> [4.1.] Kernel version (from /proc/version):
> Linux version 2.6.30-rc4-64v01kms (root@V2G64) (gcc version 4.3.2 (Gentoo 4.3.2-r3 p1.6, pie-10.1.5) ) #2 SMP PREEMPT Fri May 1 00:11:41 CEST 2009
> 
> [5.] Most recent kernel version which did not have the bug:
> 2.6.29
> 
> [X.] Other notes, patches, fixes, workarounds:
> 
> patch which should fix the issue(for 2.6.30-rc4):
> 
> --- linux-2.6.30-rc4/drivers/thermal/thermal_sys.c	2009-04-30 23:52:59.000000000 +0200
> +++ linux-2.6.30-rc4-p1/drivers/thermal/thermal_sys.c	2009-05-04 19:58:30.000000000 +0200
> @@ -961,7 +961,7 @@ void thermal_zone_device_update(struct t
>  
>  		switch (trip_type) {
>  		case THERMAL_TRIP_CRITICAL:
> -			if (temp > trip_temp) {
> +			if (temp >= trip_temp) {
>  				if (tz->ops->notify)
>  					ret = tz->ops->notify(tz, count,
>  							      trip_type);
> @@ -974,7 +974,7 @@ void thermal_zone_device_update(struct t
>  			}
>  			break;
>  		case THERMAL_TRIP_HOT:
> -			if (temp > trip_temp)
> +			if (temp >= trip_temp)
>  				if (tz->ops->notify)
>  					tz->ops->notify(tz, count, trip_type);
>  			break;
> @@ -986,14 +986,14 @@ void thermal_zone_device_update(struct t
>  
>  				cdev = instance->cdev;
>  
> -				if (temp > trip_temp)
> +				if (temp >= trip_temp)
>  					cdev->ops->set_cur_state(cdev, 1);
>  				else
>  					cdev->ops->set_cur_state(cdev, 0);
>  			}
>  			break;
>  		case THERMAL_TRIP_PASSIVE:
> -			if (temp > trip_temp || tz->passive)
> +			if (temp >= trip_temp || tz->passive)
>  				thermal_zone_device_passive(tz, temp,
>  							    trip_temp, count);
>  			break;
> --
> To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html