Re: [RFC PATCH 2/4] thermal: introduce device tree parser

[Eduardo Valentin]

On 15-07-2013 13:03, R, Durgadoss wrote:
>> -----Original Message-----
>> From: linux-pm-owner@xxxxxxxxxxxxxxx [mailto:linux-pm-
>> owner@xxxxxxxxxxxxxxx] On Behalf Of Eduardo Valentin
>> Sent: Monday, July 15, 2013 5:25 PM
>> To: Wei Ni
>> Cc: Eduardo Valentin; linux-pm@xxxxxxxxxxxxxxx; R, Durgadoss;
>> amit.daniel@xxxxxxxxxxx; Zhang, Rui; linux-kernel@xxxxxxxxxxxxxxx
>> Subject: Re: [RFC PATCH 2/4] thermal: introduce device tree parser
>>
>> On 10-07-2013 02:48, Wei Ni wrote:
>>> On 07/09/2013 10:00 PM, Eduardo Valentin wrote:
>>>> In order to be able to build thermal policies
>>>> based on generic sensors, like I2C device, that
>>>> can be places in different points on different boards,
>>>> there is a need to have a way to feed board dependent
>>>> data into the thermal framework.
>>>>
>>>> This patch introduces a thermal data parser for device
>>>> tree. The parsed data is used to build thermal zones
>>>> and thermal binding parameters. The output data
>>>> can then be used to deploy thermal policies.
>>>>
>>>> This patch adds also documentation regarding this
>>>> API and how to define define tree nodes to use
>>>> this infrastructure.
>>>
>>> It looks good, with this infrastructure, we can add generic sensor
>>> driver into the thermal fw easily.
>>>
>>>
>>>> +
>>>> +Below is an example:
>>>> +thermal_zone {
>>>> +            type = "CPU";
>>>> +            mask = <0x03>; /* trips writability */
>>>> +            passive_delay = <250>; /* milliseconds */
>>>> +            polling_delay = <1000>; /* milliseconds */
>>>> +            governor = "step_wise";
>>>> +            trips {
>>>> +                    alert@100000{
>>>> +                            temperature = <100000>; /* milliCelsius */
>>>> +                            hysteresis = <0>; /* milliCelsius */
>>>> +                            type = <1>;
>>>
>>> how about to use the trip type name directly, such as named as
>>> "passive-trip;", I think it's more readable. for example:
>>> trip0 {
>>> ....
>>> passive-trip;
>>> }
>>> trip1 {
>>> ....
>>> active-trip;
>>> }
>>>
>>>> +                    };
>>>> +                    crit@125000{
>>>> +                            temperature = <125000>; /* milliCelsius */
>>>> +                            hysteresis = <0>; /* milliCelsius */
>>>> +                            type = <3>;
>>>> +                    };
>>>> +            };
>>>> +            bind_params {
>>>> +                    action@0{
>>>> +                            cooling_device = "thermal-cpufreq";
>>>> +                            weight = <100>; /* percentage */
>>>> +                            mask = <0x01>;
>>>> +                    };
>>>> +            };
>>>> +};
>>>
>>> as we know, thermal_zone_bind_cooling_device() will set the upper/lower
>>> in the thermal_instance. In the default .bind function, it just set to
>>> THERMAL_NO_LIMIT, but for some platform, it need to set these
>>> upper/lower values for different cooling device and trips, how to pass
>>> these values in DT? how about to set:
>>> action@0 {
>>> ...
>>> mask = <0x03>; //or you can remove this property;
>>
>> Well, this has been added accordingly to current API needs.
>>
>>> trip0 = <&alert 1 2>; //1 is lower value, 2 is upper value;
>>> trip1 = <&crit 3 4>;
>>
>> I suppose the first item in you 3-tuple is the trip point.
>>
>>> }
>>
>> Yeah, I also noticed that I was missing the upper and lower limits. But
>> unfortunately this is a limitation on the thermal FW API too!
>>
>> If one passes a bind params, the structure which represents platform
>> info, then it won't be able to pass the upper and lower limits. But by
>> passing a .bind callback, then you have the opportunity to match it
>> using these two values.
>>
>> I believe we would need to change the data structures and the API to
>> support upper and lower limits via platform representation. We could
>> simply use the .bind callback of the dt thermal zone, but I believe that
>> would abusing the API, assuming that .match is for platform binding.
>> Durga, what do you think?
> 
> okay, I see.. Two approaches I could think of:
> 1. Introduce two arrays (size = number of trips in the tz) named
> upper/lower_limits[size] in the 'thermal_bind_params' structure.
> This way we don't need any API change. We can slightly change the
> implementation inside '__bind' function in thermal_core.c to get this
> working.
> 
> 2. Pass 3 more parameters in the .match function:
> .match(tz, cdev, trip, &lower, &upper). The platform layer
> then determines whether there is a match; and if so,
> provides sane values for lower and upper variables.
> 
> At this point of time, I think I prefer method 1 ;)
> Let me know your thoughts.
> 

Yeah, I agree that (1) is likely to scale. I will cook something with it
for next version.

> Thanks,
> Durga
>>
>>
>>>
>>>
>>> Thanks.
>>> Wei.
>>>
>>>
>>>
>>
>>
>> --
>> You have got to be excited about what you are doing. (L. Lamport)
>>
>> Eduardo Valentin
> 
> 
> 


-- 
You have got to be excited about what you are doing. (L. Lamport)

Eduardo Valentin

Attachment: signature.asc
Description: OpenPGP digital signature