Re: [PATCH v2] ARM: dts: dove.dtsi: Move ethphy to fix schema error

[Andrew Lunn <andrew@lunn.ch>]

On Thu, Dec 29, 2022 at 03:14:47PM +0100, Michał Grzelak wrote:
> Hi Andrew,
> 
> Thanks for quick reply.
> 
> On Thu, Dec 29, 2022 at 2:32 AM Andrew Lunn <andrew@lunn.ch> wrote:
> >
> > On Wed, Dec 28, 2022 at 09:02:34PM +0100, Michał Grzelak wrote:
> > > Running 'make dtbs_check' with schema in net/marvell,orion-mdio.yaml
> > > gives the following warnings:
> > > mdio-bus@72004: Unevaluated properties are not allowed
> > > ('ethernet-phy' was unexpected)
> > >       arch/arm/boot/dts/dove-cubox.dtb
> > >       arch/arm/boot/dts/dove-cubox-es.dtb
> > >       arch/arm/boot/dts/dove-d2plug.dtb
> > >       arch/arm/boot/dts/dove-d2plug.dtb
> > >       arch/arm/boot/dts/dove-dove-db.dtb
> > >       arch/arm/boot/dts/dove-d3plug.dtb
> > >       arch/arm/boot/dts/dove-sbc-a510.dtb
> > > Fix them by removing empty ethphy subnode from dove.dtsi and describe
> > > PHY on the relevant board .dts files level.
> >
> > I don't think your description is correct. What i think is causing the
> > problem is that the ethphy subnode in dove.dtsi does not have an @X.
> >
> 
> Yes, it is exactly the case. The problem is after adding dummy address
> e.g. ethernet-phy@0 in dove.dtsi 'make dtbs_check' shows warnings
> about missing required 'reg' property, that's why ethernet-phy is
> moved to .dts files.
> 
> > By moving it into the .dts file, you can then give it the correct @1,
> > or @3, which makes the linter happy. The kernel itself does not care
> > about this, it is an example of the linter being much more strict than
> > the kernel.
> >
> > If you agree with me, please update the description.
> 
> This patch exactly fixes this issue by setting proper
> ethernet-phy@<reg value> in board files.
> In such a case would you like me to update the commit message to
> better describe the change or do you have other remarks to the diff?

I would like the description to actually describe the problem, that
the linter allows ethernet-phy@X but not ethernet-phy. You description
does not make any sense to me, it is not an empty node which is
causing the problem, but that node is expected to have an @X.

	Andrew