Re: [PATCH v2 1/3] devicetree: mfd: Add binding for the TI LM3533

From: Rob Herring
Date: Fri Oct 30 2015 - 16:18:46 EST


On Fri, Oct 30, 2015 at 2:41 PM, Bjorn Andersson
<bjorn.andersson@xxxxxxxxxxxxxx> wrote:
> On Fri 30 Oct 11:42 PDT 2015, Lee Jones wrote:
>
> Rob, please see the discussion regarding ti,boost-freq-khz below. Should
> we both specify unit at the same time as we use standard units? (This is
> not the first time I have to change this back and forth)
>
>> On Tue, 27 Oct 2015, Bjorn Andersson wrote:
>>
> [..]
>> > +- ti,hwen-gpios:
>> > + Usage: required
>> > + Value type: <prop-encoded-array>
>> > + Definition: reference to gpio pin connected to the HWEN input; as
>> > + specified in "gpio/gpio.txt"
>>
>> Why have you made this a vendor binding?
>>
>> *-gpios is a generic property.
>>
>
> Because the hwen gpio is a ti lm3533 specific thing, but I get what
> you're saying. Will drop the prefix.

Actually, that is fine. -gpios is common, but the rest is specific to
a binding. But if it is a common function, then a common name would be
fine. Enable gpios are common for example.

>> > +- ti,als-supply:
>> > + Usage: optional
>> > + Value type: <prop-encoded-array>
>> > + Definition: reference to regulator powering the V_als input; as
>> > + specified in "regulator/regulator.txt"
>>
>> Same goes for *-supply.
>>
>
> Same here
>
>> > +- ti,boost-freq-khz:
>> > + Usage: required
>> > + Value type: <u32>
>> > + Definition: switch-frequency of the boost converter, must be either:
>> > + 500 or 1000
>>
>> Quite a few vendors are using 'boost' now.
>>
>
> The ti,boost-low-freq from the bq25890 binding is the only other
> property I can find that describes the same thing. So I'm not sure I
> follow you here.
>
>> Perhaps we need to create a set of generic bindings.
>>
>> Also, we usually measure DT bindings in HZ, not kHz.

Surprisingly, there are not enough examples to draw much conclusion.

> I thought we had defined frequencies to be in HZ and HZ only, but then
> Rob's comment that I need to actually specify the unit doesn't make any
> sense.

I don't think we decided, but let's decide now. Go with Hz.

Really, I first prefer the property name has units and second having
standardized units. But if there is a common property without units, I
prefer that even more.

> Do we want these properties in a standard unit or do we want them
> specifying the unit? Having both seems excessive.

You mean "freq" would imply the units? No, we want the actual units in
the property.

Rob
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/