Re: [PATCH v3 02/10] dt-bindings: pincfg-node: Add properties 'skew-delay-{in,out}put'

[Conor Dooley <conor@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 3160 bytes --]

On Tue, Oct 14, 2025 at 09:33:14PM +0200, Linus Walleij wrote:
> On Tue, Oct 14, 2025 at 8:04 PM Conor Dooley <conor@kernel.org> wrote:
> > On Tue, Oct 14, 2025 at 04:04:43PM +0200, Antonio Borneo wrote:
> 
> > > +  skew-delay-input:
> > > +    $ref: /schemas/types.yaml#/definitions/uint32
> > > +    description:
> > > +      this affects the expected clock skew on input pins.
> > > +      Typically indicates how many double-inverters are used to
> > > +      delay the signal.
> >
> > This property seems to be temporal, I would expect to see a unit of time
> > mentioned here, otherwise it'll totally inconsistent in use between
> > devices, and also a standard unit suffix in the property name.
> > pw-bot: changes-requested
> 
> Don't blame the messenger, the existing property skew-delay
> says:
> 
>   skew-delay:
>     $ref: /schemas/types.yaml#/definitions/uint32
>     description:
>       this affects the expected clock skew on input pins
>       and the delay before latching a value to an output
>       pin. Typically indicates how many double-inverters are
>       used to delay the signal.
> 
> This in turn comes from the original
> Documentation/devicetree/bindings/pinctrl/pinctrl-bindings.txt
> document, which in turn comes from this commit:
> 
> commit e0e1e39de490a2d9b8a173363ccf2415ddada871
> Author: Linus Walleij <linus.walleij@linaro.org>
> Date:   Sat Oct 28 15:37:17 2017 +0200
> 
>     pinctrl: Add skew-delay pin config and bindings
> 
>     Some pin controllers (such as the Gemini) can control the
>     expected clock skew and output delay on certain pins with a
>     sub-nanosecond granularity. This is typically done by shunting
>     in a number of double inverters in front of or behind the pin.
>     Make it possible to configure this with a generic binding.
> 
>     Cc: devicetree@vger.kernel.org
>     Acked-by: Rob Herring <robh@kernel.org>
>     Acked-by: Hans Ulli Kroll <ulli.kroll@googlemail.com>
>     Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
> 
> So by legacy skew-delay is a custom format, usually the number of
> (double) inverters.

Yeah, I actually noticed this after sending the mail. But as you say
below, the new properties can be done with a unit etc

> 
> I don't recall the reason for this way of defining things, but one reason
> could be that the skew-delay incurred by two inverters is very
> dependent on the production node of the silicon, and can be
> nanoseconds or picoseconds, these days mostly picoseconds.
> Example: Documentation/devicetree/bindings/net/adi,adin.yaml
> 
> Antonio, what do you say? Do you have the actual skew picosecond
> values for the different settings so we could define this as
> 
> skew-delay-input-ps:
> skew-delay-output-ps:
> 
> and translate it to the right register values in the driver?

The patch for the specific binding does have values in units of seconds
assigned to each register value, so I think this should be doable.

> 
> If we have the proper data this could be a good time to add this
> ISO unit to these two props.
> 
> Yours,
> Linus Walleij

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]