On Fri, Jan 16, 2026 at 1:00 PM Ivan Vecera <[email protected]> wrote:
>
> On 1/16/26 4:23 PM, Rob Herring wrote:
> > On Thu, Jan 15, 2026 at 6:02 AM Ivan Vecera <[email protected]> wrote:
> >>
> >> On 1/8/26 7:23 PM, Ivan Vecera wrote:
> >>> Introduce a common schema for DPLL pin consumers. Devices such as Ethernet
> >>> controllers and PHYs may require connections to DPLL pins for Synchronous
> >>> Ethernet (SyncE) or other frequency synchronization tasks.
> >>>
> >>> Defining these properties in a shared schema ensures consistency across
> >>> different device types that consume DPLL resources.
> >>>
> >>> Signed-off-by: Ivan Vecera <[email protected]>
> >>> ---
> >>>    .../bindings/dpll/dpll-pin-consumer.yaml      | 30 +++++++++++++++++++
> >>>    MAINTAINERS                                   |  1 +
> >>>    2 files changed, 31 insertions(+)
> >>>    create mode 100644 
> >>> Documentation/devicetree/bindings/dpll/dpll-pin-consumer.yaml
> >>>
> >>> diff --git 
> >>> a/Documentation/devicetree/bindings/dpll/dpll-pin-consumer.yaml 
> >>> b/Documentation/devicetree/bindings/dpll/dpll-pin-consumer.yaml
> >>> new file mode 100644
> >>> index 0000000000000..60c184c18318a
> >>> --- /dev/null
> >>> +++ b/Documentation/devicetree/bindings/dpll/dpll-pin-consumer.yaml
> >>> @@ -0,0 +1,30 @@
> >>> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> >>> +%YAML 1.2
> >>> +---
> >>> +$id: http://devicetree.org/schemas/dpll/dpll-pin-consumer.yaml#
> >>> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> >>> +
> >>> +title: DPLL Pin Consumer
> >>> +
> >>> +maintainers:
> >>> +  - Ivan Vecera <[email protected]>
> >>> +
> >>> +description: |
> >>> +  Common properties for devices that require connection to DPLL (Digital 
> >>> Phase
> >>> +  Locked Loop) pins for frequency synchronization (e.g. SyncE).
> >>> +
> >>> +properties:
> >>> +  dpll-pins:
> >>> +    $ref: /schemas/types.yaml#/definitions/phandle-array
> >>> +    description:
> >>> +      List of phandles to the DPLL pin nodes connected to this device.
> >>> +
> >>> +  dpll-pin-names:
> >>> +    $ref: /schemas/types.yaml#/definitions/string-array
> >>> +    description:
> >>> +      Names for the DPLL pins defined in 'dpll-pins', in the same order.
> >>> +
> >>> +dependencies:
> >>> +  dpll-pin-names: [ dpll-pins ]
> >>> +
> >>> +additionalProperties: true
> >>> diff --git a/MAINTAINERS b/MAINTAINERS
> >>> index 765ad2daa2183..f6f58dfb20931 100644
> >>> --- a/MAINTAINERS
> >>> +++ b/MAINTAINERS
> >>> @@ -7648,6 +7648,7 @@ M:      Jiri Pirko <[email protected]>
> >>>    L:  [email protected]
> >>>    S:  Supported
> >>>    F:  Documentation/devicetree/bindings/dpll/dpll-device.yaml
> >>> +F:   Documentation/devicetree/bindings/dpll/dpll-pin-consumer.yaml
> >>>    F:  Documentation/devicetree/bindings/dpll/dpll-pin.yaml
> >>>    F:  Documentation/driver-api/dpll.rst
> >>>    F:  drivers/dpll/
> >>
> >> Based on private discussion with Andrew Lunn (thanks a lot), this is
> >> wrong approach. Referencing directly dpll-pin nodes and using their
> >> phandles in consumers is at least unusual.
> >>
> >> The right approach should be referencing dpll-device and use cells
> >> to specify the dpll pin that is used.
> >
> > You only need a cells property if you expect the number of cells to
> > vary by provider.
> >
> > However, the DPLL device just appears to be a clock provider and
> > consumer, so why not just use the clock binding here? Also, there is
> > no rule that using foo binding means you have to use foo subsystem in
> > the kernel.
>
> Hmm, do you mean something like this example?
>
> &dpll0 {
>      ...
>      #clock-cells = <2>; /* 1st pin index, 2nd pin type (input/output) */
>
>      input-pins {
>          pin@2 {
>              reg = <2>;
>              ...
>          };
>          pin@4 {
>              reg = <4>;
>              ...
>          };
>      };
>      output-pins {
>          pin@3 {
>              reg = <3>;
>          };
>      };
> };
> &phy0 {
>      ...
>      clock-names = "rclk0", "rclk1", "synce_ref";
>      clocks = <&dpll0 2 DPLL_INPUT>,
>               <&dpll0 4 DPLL_INPUT>,
>               <&dpll0 3 DPLL_OUTPUT>;
>      ...
> };

No, clock providers are always the clock outputs, and clock consumers
are the clock inputs. So something like this:

&dpll0 {
     ...
     #clock-cells = <1>; /* 1st pin index */

     // clocks index corresponds to input pins on dpll0 */
     clocks = <&phy0 0>, <&phy0 1>, <&phy1 0>, <&phy1 1>
};
&phy0 {
     ...
     #clock-cells = <1>;
     clocks = <&dpll0 3>;
     ...
};

Rob

Reply via email to