Re: [PATCH] dt-bindings: pinctrl: ti,dra7-iodelay: Convert to DT schema

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Medium] The child node regex `^mmc.*_conf$` arbitrarily restricts I/O delay configuration nodes to those starting with 'mmc', even though the controller is generic.
- [Medium] The `pinctrl-pin-array` property is not marked as required within the child node schema.
--

--- Patch [1]: [PATCH] dt-bindings: pinctrl: ti,dra7-iodelay: Convert to DT schema ---
commit 59d5959fa2ce86d89ce637daf5f846ce2bedcb0d
Author: Bhargav Joshi <j.bhargav.u@gmail.com>

dt-bindings: pinctrl: ti,dra7-iodelay: Convert to DT schema

This commit converts the TI IODELAY controller bindings from plain text
to DT schema. It also adds documentation for child properties missing
from the original text bindings.

> diff --git a/Documentation/devicetree/bindings/pinctrl/ti,dra7-iodelay.yaml b/Documentation/devicetree/bindings/pinctrl/ti,dra7-iodelay.yaml
> new file mode 100644
> index 0000000000000..a399946ae8172
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/pinctrl/ti,dra7-iodelay.yaml

[ ... ]

> +patternProperties:
> +  ^mmc.*_conf$:

[Severity: Medium]
Looking at the patternProperties definition, does this regression arbitrarily
restrict delay configuration nodes to only those starting with mmc?

The commit message mentions that mmc pins are just an example of pins
needing dynamic configuration. If a device tree dynamically configures delay
for other pins, will this reject them as invalid?

Would it be better to use a more generic pattern?

> +    type: object
> +    additionalProperties: false
> +    properties:
> +      pinctrl-pin-array:
> +        $ref: /schemas/types.yaml#/definitions/uint32-array
> +        description: |
> +          An array of 32-bit integers defining the pin delay configuration.

[Severity: Medium]
Could this regression allow incomplete configuration nodes to pass schema
validation?

Since pinctrl-pin-array is not marked as required, an empty child node would
pass dt_binding_check, but the driver expects this property to be present
to initialize the mapping. Should it be added to a required list within the
patternProperties block?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260713-ti-iodelay-v1-1-5f06d36c4aac@gmail.com?part=1