[PATCH v3 09/10] dt-bindings: pinctrl: stm32: Support I/O synchronization parameters

Antonio Borneo antonio.borneo at foss.st.com
Mon Oct 20 08:09:14 PDT 2025


On Wed, 2025-10-15 at 15:35 +0100, Conor Dooley wrote:
> On Wed, Oct 15, 2025 at 02:56:56PM +0200, Antonio Borneo wrote:
> > On Tue, 2025-10-14 at 19:10 +0100, Conor Dooley wrote:
> > > On Tue, Oct 14, 2025 at 04:04:50PM +0200, Antonio Borneo wrote:
> 
> > > 
> > > > +
> > > > +          st,io-sync:
> > > > +            description: |
> > > > +              IO synchronization through re-sampling or inversion
> > > > +              0: data or clock GPIO pass-through
> > > > +              1: clock GPIO inverted
> > > > +              2: data GPIO re-sampled on clock rising edge
> > > > +              3: data GPIO re-sampled on clock falling edge
> > > > +              4: data GPIO re-sampled on both clock edges
> > > > +            $ref: /schemas/types.yaml#/definitions/uint32
> > > > +            enum: [0, 1, 2, 3, 4]
> > > 
> > > I really don't like this kinds of properties that lead to "random"
> > > numbers in devicetree. I'd much rather see a string list here.
> > 
> > Agree!
> > I just need to figure out some reasonably short but still meaningful
> > string for them.
> 
> pass-through
> inverted
> rising-edge
> falling-edge
> both-edges
> 
> perhaps?

Since these are strings in a custom property, I think I will use something longer and more explicit for them.

Linus,

pinconf-generic only accepts positive numeric values for both generic and custom properties in struct pinconf_generic_params.
Plus, I haven't found any existing driver that mixes pinconf-generic with custom string values.
I'm going to extend the current pinconf-generic to handle such case.
Or did I missed something?

Regards,
Antonio





More information about the linux-arm-kernel mailing list