Re: [PATCH v4 1/2] dt-bindings: hwmon: Add ina3221 documentation

[Guenter Roeck <linux@roeck-us.net>]

On 09/22/2018 10:31 PM, Nicolin Chen wrote:
> On Sat, Sep 22, 2018 at 10:19:42PM -0700, Guenter Roeck wrote:
>>> +2) child nodes
>>> +  Required properties:
>>> +  - input-id: Must be 1, 2 or 3
>>> +
>>> +  Optional properties:
>>> +  - input-label: Name of the input source
>>> +  - shunt-resistor: Shunt resistor value in micro-Ohm
>>> +  - status: Should be "disabled" if no input source
>>> +
>>> +  Example:
>>> +
>>> +  input1 {
>>> +          input-id = <0x1>;
>>
>> We'll have to find a better name for this. Feel free to look up examples in the
>> existing devicetree descriptions. The one that seems to be used most of the time
>> to indicate a channel index or id is "reg". It should also start with 0 - there
>> is no real reason for it to start with 1; it only makes the code more complex.
> 
> The reason is that the port start from 1 in the datasheet.
> 

Maybe, but for me I'll want to have something that we can reuse for other chips.
Having the index start with 0 for one chip and with 1 for another would be
confusing. It is bad enough that we have in[0..n] for voltages and temp[1..n]
for temperatures. I would not want to see the same in devicetree files,
and much less so on a per-device basis. It is also pretty common to start
channel numbers with 0 in devicetree files.

> I don't mind using reg or count it from 0, will look up to
> see if I can find something solid; otherwise, I'll wait for
> binding doc maintainers' opinions before sending v5.
> 
Sure, no problem.

Guenter

>>> +          status = "disabled";
>>> +  };
>>> +  input2 {
>>> +          input-id = <0x2>;
>>> +          shunt-resistor = <5000>;
>>
>> I would suggest shunt-resistor-micro-ohms as per
>> Documentation/devicetree/bindings/property-units.txt.
> 
> Will change it.
> 
> Thank you
> Nicolin
>