[PATCH v6 3/4] dt-bindings: pwm: add IPQ6018 binding

[Bjorn Andersson]

On Sun 25 Jul 21:08 PDT 2021, Baruch Siach wrote:

> Hi Bjorn,
> 
> On Sun, Jul 25 2021, Bjorn Andersson wrote:
> > On Thu 22 Jul 05:01 CDT 2021, Baruch Siach wrote:
> >> +  clocks:
> >> +    maxItems: 1
> >> +
> >> +  clock-names:
> >> +    const: core
> >
> > With a single clock, it's nice to skip the -names.
> 
> I find it nicer and better for forward compatibility with hardware
> variants the might introduce more clocks.
> 

Do you foresee any need for forward compatibility? What other clocks
would this binding have to refer to?

That said, you'd achieve the same forward compatibility by just
making sure that the current clock is the first on in the amended
binding (which you have to do with or without -names).

> Are there any downsides to -names?
> 

Look at the number of places in a typical dts that we could have added
clock-names, reg-names, interrupt-names, power-domain-names etc for a
single cell.

I do find it beneficial to keep things cleaner and sticking with the
design of "single resource has no -names".

Regards,
Bjorn