Re: [PATCH REBASE] dt-bindings: pinctrl: use pinctrl.yaml

[Charles Keepax <ckeepax@opensource.cirrus.com>]

On Tue, Dec 07, 2021 at 09:47:38AM +0100, Rafał Miłecki wrote:
> On 06.12.2021 23:50, Rob Herring wrote:
> >On Thu, Dec 2, 2021 at 12:32 AM Rafał Miłecki <zajec5@gmail.com> wrote:
> >>
> >>From: Rafał Miłecki <rafal@milecki.pl>
> >>
> >>Also fix some examples to avoid warnings like:
> >>brcm,ns-pinmux.example.dt.yaml: pin-controller@1800c1c0: $nodename:0: 'pin-controller@1800c1c0' does not match '^pinctrl|pinmux@[0-9a-f]+$'
> >
> >I think you missed some. linux-next now has these warnings:
> 
> Oops, sorry, I think I didn't test MFD bindings *after* modifying
> pinctrl bindings.
> 
> 
> >/builds/robherring/linux-dt/Documentation/devicetree/bindings/mfd/cirrus,madera.example.dt.yaml:
> >codec@1a: $nodename:0: 'codec@1a' does not match
> >'^(pinctrl|pinmux)(@[0-9a-f]+)?$'
> > From schema: /builds/robherring/linux-dt/Documentation/devicetree/bindings/mfd/cirrus,madera.yaml
> 
> I'm not sure how to deal with this one. If you take a look at
> cirrus,madera.yaml it seems to be some complex MFD that is a sound
> device with pin controller and regulator.
> 
> I'm not sure: is using pinctrl@ node for that hardware is a correct
> choice?

Yeah historically there has been some contention around how to
implement MFD devices, the Madera devices have a single node for
the whole chip, rather than a sub-node for each driver. In many
ways this does make sense since DT should describe the hardware
and hardware wise this is a single chip, the fact linux breaks it
down into separate drivers is really a linux specific
implementation. Although the Lochnagar stuff from your other
patch was steered in the direction of a sub-node for each driver
during review.

But either way the Madera binding is a bit old to be majorly
changing at this point, and does have quite a few out of tree
users. Not really sure how to fix this one either, the name of
codec for the node is really correct as that is what the node is.

Thanks,
Charles