Re: [PATCH 3/8] dt-bindings: serial: fsl-linflexuart: Add compatible for S32G2

[Rob Herring <robh@kernel.org>]

On Thu, Aug 12, 2021 at 06:27:57PM +0200, Andreas Färber wrote:
> On 05.08.21 08:54, Chester Lin wrote:
> > Add a compatible string for the uart binding of NXP S32G2 platforms. Here
> > we use "s32v234-linflexuart" as fallback since the current linflexuart
> > driver still works on S32G2.
> > 
> > Signed-off-by: Chester Lin <clin@suse.com>
> > ---
> >  .../bindings/serial/fsl,s32-linflexuart.yaml  | 26 ++++++++++++++++---
> >  1 file changed, 22 insertions(+), 4 deletions(-)
> > 
> > diff --git a/Documentation/devicetree/bindings/serial/fsl,s32-linflexuart.yaml b/Documentation/devicetree/bindings/serial/fsl,s32-linflexuart.yaml
> > index acfe34706ccb..e731f3f6cea4 100644
> > --- a/Documentation/devicetree/bindings/serial/fsl,s32-linflexuart.yaml
> > +++ b/Documentation/devicetree/bindings/serial/fsl,s32-linflexuart.yaml
> > @@ -21,10 +21,20 @@ allOf:
> >  
> >  properties:
> >    compatible:
> > -    description: The LINFlexD controller on S32V234 SoC, which can be
> > -      configured in UART mode.
> > -    items:
> > -      - const: fsl,s32v234-linflexuart
> > +    minItems: 1
> > +    maxItems: 2
> 
> Are these necessary for oneOf?
> 
> > +    oneOf:
> > +      - description: The LINFlexD controller on S32G2 SoC, which can be
> > +          configured in UART mode.
> > +        items:
> > +          - enum:
> > +              - fsl,s32g2-linflexuart
> > +          - const: fsl,s32v234-linflexuart
> 
> This reads inconsistent to me: Either this oneOf is for S32G2 only, then
> please turn the enum into a const. Or change the description to "on SoCs
> compatible with S32V234" if we expect the enum list to grow.
> 
> I believe the idea here was to avoid unnecessary driver compatible and
> earlycon compatible additions, while preparing for eventual quirks
> specific to S32G2.
> 
> @NXP: Should this be s32g2- like above or s32g274a- specifically? Do you
> agree this is a useful thing to prepare here, as opposed to using only
> s32v234- in the s32g2* DT?
> 
> I assume the ordering is done alphabetically as S32G < S32V;
> alternatively we might order S32V234 first and then the compatible ones.
> 
> > +
> > +      - description: The LINFlexD controller on S32V234 SoC, which can be
> > +          configured in UART mode.
> > +        items:
> > +          - const: fsl,s32v234-linflexuart
> 
> To minimize this S32G2 patch, would it be valid to do oneOf for the
> single S32V in the preceding patch already? Then we would avoid the text
> movement and re-indentation above and more easily see the lines newly
> getting added for S32G2.
> 
> >  
> >    reg:
> >      maxItems: 1
> > @@ -41,8 +51,16 @@ unevaluatedProperties: false
> >  
> >  examples:
> >    - |
> > +    /* S32V234 */
> 
> Could this be:
>   - description: S32V234
>     |
> ?

No, that would not be valid json-schema.

Rob