Re: [PATCH v4 1/3] dt-bindings: soc: sophgo: add TOP syscon for CV18XX/SG200X series SoC

[Conor Dooley <conor@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 2251 bytes --]

On Sat, Oct 25, 2025 at 10:27:13AM +0800, Longbin Li wrote:
> On Fri, Oct 24, 2025 at 05:46:03PM +0100, Conor Dooley wrote:
> 
> > > ...
> > > +
> > > +properties:
> > > +  compatible:
> > > +    oneOf:
> > > +      - items:
> > > +          - const: sophgo,cv1800b-top-syscon
> > > +          - const: syscon
> > > +          - const: simple-mfd
> > > +
> > > +  reg:
> > > +    maxItems: 1
> > > +
> > > +  "#address-cells":
> > > +    const: 1
> > > +
> > > +  "#size-cells":
> > > +    const: 1
> > > +
> > > +  ranges: true
> > 
> > Do you actually use ranges?
> > 
> > > +patternProperties:
> > > +  "dma-router@[0-9a-f]+$":
> > 
> > Do these actually appear at variable addresses, or is it always 48 for
> > the phy and 53 for the dma router?
> > 
> > > +    $ref: /schemas/dma/sophgo,cv1800b-dmamux.yaml#
> > > +    unevaluatedProperties: false
> > > +
> > > +  "phy@[0-9a-f]+$":
> > > +    $ref: /schemas/phy/sophgo,cv1800b-usb2-phy.yaml#
> > > +    unevaluatedProperties: false
> > > +
> > > +required:
> > > +  - compatible
> > > +  - reg
> > > +  - "#address-cells"
> > > +  - "#size-cells"
> > > +
> > > +additionalProperties: false
> > > +
> > > +examples:
> > > +  - |
> > > +    #include <dt-bindings/clock/sophgo,cv1800.h>
> > > +
> > > +    syscon@3000000 {
> > > +      compatible = "sophgo,cv1800b-top-syscon", "syscon", "simple-mfd";
> > > +      reg = <0x03000000 0x1000>;
> > > +      #address-cells = <1>;
> > > +      #size-cells = <1>;
> > > +
> > > +      usbphy: phy@48 {
> > 
> > Drop the labels off of these two, since they're unused in the example.
> > 
> 
> Hi,
> 
> Thanks for the review.
> 
> There are a couple of different opinions here.
> Could you please help confirm which approach is preferred here?
> 
> This is Rob's reply:
> https://lore.kernel.org/all/20251015134144.GA3265377-robh@kernel.org/

I don't think that Rob and I disgree about the example, having child
nodes is good, having the label on them is what I don't want.

Re: the child node patterns, I was not looking for a return to what Rob
didn't like, but rather a move to just making them normal properties
rather than a pattern, if there's only one possible address.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]