Re: [PATCH v2] dt-bindings: iio: adc: stm32-adc: convert bindings to json-schema

[Rob Herring <robh@kernel.org>]

On Tue, Jan 14, 2020 at 10:02 AM Fabrice Gasnier <fabrice.gasnier@st.com> wrote:
>
> On 12/18/19 12:43 AM, Rob Herring wrote:
> > On Fri, Dec 06, 2019 at 05:17:08PM +0100, Fabrice Gasnier wrote:
> >> Convert the STM32 ADC binding to DT schema format using json-schema
> >>
> >> Signed-off-by: Fabrice Gasnier <fabrice.gasnier@st.com>
> >> ---
> >> Note: this applies on top of IIO tree currently (iio-for-5.5c).
> >>
> >> Changes in V2:
> >> - Take almost all of Rob suggestions (removed reg generic description,
> >>   added minItems, maxItems, st,max-clk-rate-hz range, drop some pipes,
> >>   simplify clock-names, remove unneeded allOfs)
> >> - For now, keep all in one file despite there are lots of if/thens in the
> >>   bindings
> >> ---
> >>  .../devicetree/bindings/iio/adc/st,stm32-adc.txt   | 149 -------
> >>  .../devicetree/bindings/iio/adc/st,stm32-adc.yaml  | 454 +++++++++++++++++++++
> >>  2 files changed, 454 insertions(+), 149 deletions(-)
> >>  delete mode 100644 Documentation/devicetree/bindings/iio/adc/st,stm32-adc.txt
> >>  create mode 100644 Documentation/devicetree/bindings/iio/adc/st,stm32-adc.yaml
> >
> >
>
>
> [snip]
>
> >> +
> >> +      st,adc-channels:
> >> +        description: |
> >> +          List of single-ended channels muxed for this ADC. It can have up to:
> >> +            - 16 channels, numbered from 0 to 15 (for in0..in15) on stm32f4
> >> +            - 20 channels, numbered from 0 to 19 (for in0..in19) on stm32h7 and
> >> +              stm32mp1.
> >> +        allOf:
> >> +          - $ref: /schemas/types.yaml#/definitions/uint32-array
> >> +
>
> [snip]
>
> >> +
> >> +    allOf:
> >> +      - if:
> >> +          properties:
> >> +            compatible:
> >> +              contains:
> >> +                const: st,stm32f4-adc
> >> +
> >> +        then:
> >> +          properties:
> >> +            reg:
> >> +              enum:
> >> +                - 0x0
> >> +                - 0x100
> >> +                - 0x200
> >> +
> >> +            interrupts:
> >> +              minimum: 0
> >> +              maximum: 2
> >> +
> >> +            assigned-resolution-bits:
> >> +              enum: [6, 8, 10, 12]
> >> +              default: 12
> >> +
> >> +            st,adc-channels:
> >> +              minItems: 1
> >> +              maxItems: 16
> >> +              minimum: 0
> >> +              maximum: 15
> >
> > You are mixing array and scalar constraints here. You need:
> >
> > minItems: 1
> > maxItems:16
> > items:
> >   minimum: 0
> >   maximum: 15
> >
> > Update dtschema. It will now catch this. There's a few others too.
>
> Hi Rob,
>
> Sorry for the late reply. I updated dtschema. Now it catches it.
>
> I've tried your suggestion, but when I test it, I don't get any error on
> maxItems.
>
> In the example: "st,adc-channels = <0>, <1>, ... more than 16 items;"
>
> Is it possible I face some other issue with dtschema ?

The problem is how "<0>, <1>" vs. "<0 1>" gets encoded. While those
are the same in the dtb, in yaml we have "[[0], [1]]" vs. "[[0, 1]]".
Making the brackets significant is helpful for some things like
phandle+args and 'reg' where we have a matrix of values, but for
arrays it just gets in the way. I think as I suggested is the right
form for the binding schema, and we need to either decide what's the
correct way for brackets or improve the tool to accept both ways.

Rob