[RFC 3/3] ARM: dts: Don't overheat the Odroid XU3-Lite on high load
Anand Moon
linux.amoon at gmail.com
Wed Feb 17 19:17:25 PST 2016
Hi Krzysztof
On 18 February 2016 at 07:17, Krzysztof Kozlowski
<k.kozlowski at samsung.com> wrote:
> On 18.02.2016 04:53, Anand Moon wrote:
>> Hi Krzysztof,
>>
>> On 17 February 2016 at 12:25, Krzysztof Kozlowski
>> <k.kozlowski at samsung.com> wrote:
>>> After adding cpufreq-dt support to Exynos542x, the Odroid XU3-Lite can
>>> be easily overheated when launching eight CPU-intensive tasks:
>>> thermal thermal_zone3: critical temperature reached(121 C),shutting down
>>>
>>> This seems to be specific to Odroid XU3-Lite board which officially
>>> supports lower frequencies than regular XU3 or XU4. When working at
>>> maximum CPU speed (1800 MHz big and 1300 MHz LITTLE) in warmer place for
>>> longer time, the fan fails to cool down the board and it reaches
>>> critical temperature.
>>>
>>> Add CPU cooling to Exynos5422/5800 to fix this issue. When reaching 95
>>> degrees of Celsius, the board will slow down by 3 steps (around
>>> 1400/1000 MHz). When reaching 110 degrees of Celsius go to 600 MHz.
>>>
>>> Signed-off-by: Krzysztof Kozlowski <k.kozlowski at samsung.com>
>>> ---
>>> arch/arm/boot/dts/exynos5422-cpu-thermal.dtsi | 41 +++++++++++++++++++++++++++
>>> 1 file changed, 41 insertions(+)
>>>
>>> diff --git a/arch/arm/boot/dts/exynos5422-cpu-thermal.dtsi b/arch/arm/boot/dts/exynos5422-cpu-thermal.dtsi
>>> index 2b289d7c0d13..66073ce29aee 100644
>>> --- a/arch/arm/boot/dts/exynos5422-cpu-thermal.dtsi
>>> +++ b/arch/arm/boot/dts/exynos5422-cpu-thermal.dtsi
>>> @@ -34,6 +34,16 @@
>>> hysteresis = <5000>; /* millicelsius */
>>> type = "active";
>>> };
>>> + cpu_alert3: cpu-alert-3 {
>>> + temperature = <95000>; /* millicelsius */
>>> + hysteresis = <5000>; /* millicelsius */
>>> + type = "passive";
>>> + };
>>> + cpu_alert4: cpu-alert-4 {
>>> + temperature = <110000>; /* millicelsius */
>>> + hysteresis = <5000>; /* millicelsius */
>>> + type = "passive";
>>> + };
>>> cpu_crit0: cpu-crit-0 {
>>> temperature = <120000>; /* millicelsius */
>>> hysteresis = <0>; /* millicelsius */
>>> @@ -53,6 +63,37 @@
>>> trip = <&cpu_alert2>;
>>> cooling-device = <&fan0 2 3>;
>>> };
>>> +
>>> + /*
>>> + * When reaching cpu_alert3, reduce CPU
>>> + * by 3 steps. On Exynos5422/5800 that would
>>> + * be: 1400 MHz and 1000 MHz.
>>> + */
>>> + map3 {
>>> + trip = <&cpu_alert3>;
>>> + cooling-device = <&cpu0 3 3>;
>>> + };
>>> + map4 {
>>> + trip = <&cpu_alert3>;
>>> + cooling-device = <&cpu4 3 3>;
>>> + };
>>> +
>>> + /*
>>> + * When reaching cpu_alert4, reduce CPU
>>> + * to 600 MHz (11 steps for big, 7 steps for
>>> + * LITTLE).
>>> + * Exynos5420 has less OPPs and reversed
>>> + * numbering of CPUs (big/LITTLE) so this
>>> + * would not match.
>>> + */
>>> + map5 {
>>> + trip = <&cpu_alert4>;
>>> + cooling-device = <&cpu0 7 7>;
>>> + };
>>> + map6 {
>>> + trip = <&cpu_alert4>;
>>> + cooling-device = <&cpu4 11 11>;
>>> + };
>>> };
>>> };
>>> };
>>> --
>>> 2.5.0
>>>
>>
>> could you append this patch with following changes.
>
> Could you describe why?
>
>From the documentation.
Documentation/thermal/sysfs-api.txt
passive_delay: number of milliseconds to wait between polls when
performing passive cooling.
polling_delay: number of milliseconds to wait between polls when
checking whether trip points have been crossed (0 for interrupt
driven systems).
Exynos driver is interrupt driven so please ignore.
Best Regards.
-Anand Moon
>> diff --git a/arch/arm/boot/dts/exynos5422-cpu-thermal.dtsi
>> b/arch/arm/boot/dts/exynos5422-cpu-thermal.dtsi
>> index 66073ce..4e72637 100644
>> --- a/arch/arm/boot/dts/exynos5422-cpu-thermal.dtsi
>> +++ b/arch/arm/boot/dts/exynos5422-cpu-thermal.dtsi
>> @@ -16,8 +16,8 @@
>> thermal-zones {
>> cpu0_thermal: cpu0-thermal {
>> thermal-sensors = <&tmu_cpu0 0>;
>> - polling-delay-passive = <0>;
>> - polling-delay = <0>;
>> + polling-delay-passive = <250>; /* milliseconds */
>> + polling-delay = <500>; /* milliseconds */
>> trips {
>> cpu_alert0: cpu-alert-0 {
>> temperature = <50000>; /*
>> millicelsius */
>> ---
>> On running linaro pm-qa diagnostic tool
>> ----------------------------------------------------------
>>
>> thermal_01.28: checking 'thermal_zone2'/'trip_point_2_temp' ='110000'... Ok
>> thermal_01.29: checking 'cdev0_trip_point' exists in
>> '/sys/devices/virtual/thermal/thermal_zone0'... Ok
>> thermal_01.30: checking 'thermal_zone0/cdev0_trip_point' valid binding... Ok
>> thermal_01.31: checking 'cdev4_trip_point' exists in
>> '/sys/devices/virtual/thermal/thermal_zone0'... Ok
>> thermal_01.32: checking 'thermal_zone0/cdev4_trip_point' valid binding... Err
>> thermal_01.33: checking 'cdev4_trip_point' exists in
>> '/sys/devices/virtual/thermal/thermal_zone0'... Ok
>> thermal_01.34: checking 'thermal_zone0/cdev4_trip_point' valid binding... Err
>> thermal_01.35: checking 'cdev4_trip_point' exists in
>> '/sys/devices/virtual/thermal/thermal_zone0'... Ok
>> thermal_01.36: checking 'thermal_zone0/cdev4_trip_point' valid binding... Err
>> thermal_01.37: checking 'cdev4_trip_point' exists in
>> '/sys/devices/virtual/thermal/thermal_zone0'... Ok
>> thermal_01.38: checking 'thermal_zone0/cdev4_trip_point' valid binding... Err
>>
>> thermal_01: fail
>> -------------------------------------------------------
>> I also got lot's of error.
>>
>> root at odroidxu4l:~# cpu[ 3050.847663] cpu cpu4: Failed to find dev_opp: -19
>> [ 3171.640836] cpu cpu4: device_opp_debug_create_link: Failed to create link
>> [ 3171.646197] cpu cpu4: _add_list_dev: Failed to register opp debugfs (-12)
>> [ 3171.653574] cpu cpu7: device_opp_debug_create_link: Failed to create link
>> [ 3171.659752] cpu cpu7: _add_list_dev: Failed to register opp debugfs (-12)
>> [ 3171.697011] cpu cpu5: cpufreq_init: failed to get clk: -2
>> [ 3171.732505] cpu cpu6: cpufreq_init: failed to get clk: -2
>> [ 3171.768160] cpu cpu7: cpufreq_init: failed to get clk: -2
>>
>> Tested on Odroid-XU4
>>
>> Reviewed-by: Anand Moon <linux.amoon at gmail.com>
>> Tested-by: Anand Moon <linux.amoon at gmail.com>
>
> The patch is not sufficient. It does not work the way it should...
>
> BTW, I found the issue. The order of trip points in DT:
> thermal_zone0/trip_point_0_hyst:5000
> thermal_zone0/trip_point_0_temp:50000
> thermal_zone0/trip_point_0_type:active
> thermal_zone0/trip_point_1_hyst:5000
> thermal_zone0/trip_point_1_temp:60000
> thermal_zone0/trip_point_1_type:active
> thermal_zone0/trip_point_2_hyst:5000
> thermal_zone0/trip_point_2_temp:70000
> thermal_zone0/trip_point_2_type:active
> thermal_zone0/trip_point_3_hyst:0
> thermal_zone0/trip_point_3_temp:120000 <---- this should be last one!
> thermal_zone0/trip_point_3_type:critical
> thermal_zone0/trip_point_4_hyst:5000
> thermal_zone0/trip_point_4_temp:90000
> thermal_zone0/trip_point_4_type:passive
> thermal_zone0/trip_point_5_hyst:5000
> thermal_zone0/trip_point_5_temp:110000
> thermal_zone0/trip_point_5_type:passive
>
> After fixing the order in DT, the cpu cooler starts working.
>
> Best regards,
> Krzysztof
>
More information about the linux-arm-kernel
mailing list