Re: [PATCH v2 10/11] dt-bindings: arm: sunxi: add PocketBook Verse

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 4 potential issue(s) to consider:
- [Medium] rtc: sun6i: missing clock provider initialization for sun8i-b288-rtc
- [Low] dt-bindings: pinctrl: missing interrupts array constraint for sun8i-b288-pinctrl
- [Low] dt-bindings: pinctrl: allwinner,pinmux property allowed across all sunxi SoCs
- [Low] dt-bindings: rtc: missing clock-output-names constraint for sun8i-b288-rtc
--

commit 52bfb11b9f38d6aaa40748a8fbc5e4f1f1b2ec66
Author: Pablo Mazzini <pmazzini@gmail.com>

dt-bindings: arm: sunxi: add PocketBook Verse

This commit adds a new compatible string for the PocketBook Verse (PB629),
a 6" e-book reader built around the Allwinner B288 SoC.

> diff --git a/Documentation/devicetree/bindings/arm/sunxi.yaml b/Documentation/devicetree/bindings/arm/sunxi.yaml
> index f04db28c630ac..f66c681d3788a 100644
> --- a/Documentation/devicetree/bindings/arm/sunxi.yaml
> +++ b/Documentation/devicetree/bindings/arm/sunxi.yaml
> @@ -811,6 +811,11 @@ properties:
>            - const: pocketbook,614-plus
>            - const: allwinner,sun5i-a13
>  
> +      - description: PocketBook Verse (PB629)
> +        items:
> +          - const: pocketbook,verse
> +          - const: allwinner,sun8i-b288

[Severity: Medium]
Is there a missing clock provider initialization for the sun8i-b288-rtc
companion driver?

Looking at sun6i_rtc_dt_ids() in drivers/rtc/rtc-sun6i.c, the compatible
string is registered:

    { .compatible = "allwinner,sun8i-b288-rtc" },

but it lacks the necessary CLK_OF_DECLARE_DRIVER() block to actually
initialize early clock structures. This omission means the RTC driver fails
to register as a clock provider. Any driver attempting to acquire <&rtc 0>
via of_clk_get() or similar will trigger an EPROBE_DEFER.

[Severity: Low]
Does the sun8i-b288-pinctrl compatible also need to be added to the
interrupts constraint block in the pinctrl schema?

In Documentation/devicetree/bindings/pinctrl/allwinner,sun4i-a10-pinctrl.yaml:

  - if:
      properties:
        compatible:
          enum:
            - allwinner,sun6i-a31-pinctrl
            - allwinner,sun6i-a31s-pinctrl
            - allwinner,sun50i-h6-pinctrl

    then:
      properties:
        interrupts:
          minItems: 4
          maxItems: 4

This conditional block restricts the interrupts array size to exactly 4.
Without including the new sun8i-b288-pinctrl compatible, the schema permits
incorrect interrupt arrays (defaulting to 1 to 8 items) for the b288 pinctrl,
bypassing strict validation.

[Severity: Low]
Does this allow the b288-specific allwinner,pinmux property to leak into the
global sunxi pinctrl namespace?

In the same allwinner,sun4i-a10-pinctrl.yaml file, the allOf block includes:

    then:
      patternProperties:
        "^([rs]-)?(([a-z0-9]{3,}|[a-oq-z0-9][a-z0-9]*?)?-)+?(p[a-ilm][0-9]*?-)??pins?$":
          required:
            - allwinner,pinmux

This makes the property required for b288 without rejecting it for all other
compatibles. Adding the property to any non-b288 board device tree would
incorrectly pass dt_binding_check. An additional conditional block rejecting
this property for non-b288 devices must be added after this block to prevent
schema bypass.

[Severity: Low]
Are we missing a clock-output-names constraint for the sun8i-b288-rtc?

In Documentation/devicetree/bindings/rtc/allwinner,sun6i-a31-rtc.yaml,
the allOf block contains constraints like:

  - if:
      properties:
        compatible:
          contains:
            const: allwinner,sun6i-a31-rtc

The sun8i-b288-rtc compatible must be added to one of these conditional
blocks to restrict its clock-output-names length. Because it is omitted,
the schema incorrectly allows 1 to 3 clock output names for the b288 RTC
instead of strictly enforcing 1.

> +
>        - description: Point of View Protab2-IPS9
>          items:
>            - const: pov,protab2-ips9

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260927151016.186493-1-pmazzini@gmail.com?part=10