[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <b29cff36-b6c0-e9df-475b-f5138cff06c4@roeck-us.net>
Date: Sat, 22 Sep 2018 16:56:55 -0700
From: Guenter Roeck <linux@...ck-us.net>
To: Nicolin Chen <nicoleotsuka@...il.com>
Cc: jdelvare@...e.com, robh+dt@...nel.org, mark.rutland@....com,
corbet@....net, afd@...com, linux-hwmon@...r.kernel.org,
devicetree@...r.kernel.org, linux-kernel@...r.kernel.org,
linux-doc@...r.kernel.org
Subject: Re: [PATCH v3 1/2] dt-bindings: hwmon: Add ina3221 documentation
On 09/22/2018 11:03 AM, Nicolin Chen wrote:
>>> +2) child nodes
>>> + The names of child nodes should indicate input source names
>>> +
>>> + Required properties:
>>> + - input-id: Must be 1, 2 or 3
>>> +
>>> + Optional properties:
>>> + - shunt-resistor: Shunt resistor value in micro-Ohm
>>> + - status: Should be "disabled" if no input source
>>> +
>>> + Example:
>>> +
>>> + input1 {
>>> + input-id = <0x1>;
>>> + status = "disabled";
>>> + };
>>> + VDD_GPU {
>>> + input-id = <0x2>;
>>> + shunt-resistor = <5000>;
>>> + };
>>>
>>
>> Using child nodes is a good idea. However, you are converting the node name into
>> the hwmon 'label' attribute which I can not accept. First, it is undocumented,
>> second, it effectively creates an undocumented property (if one wants to configure
>> the shunt resistor value, one has to configure a child node which is converted
>> into a label), and third, it violates the hwmon ABI ('input1' is not a "hint
>> about what this voltage channel is being used for").
>
> Oh. I see the point here now. Then a child name could be just input[123],
> and I will add a separate optional child property to indicate the label.
>
Exactly. "label" is quite widely used as property name, so that should be acceptable.
I am not sure about index; we'll see if Rob has any comments.
Thanks,
Guenter
Powered by blists - more mailing lists