Re: [PATCH] dt-bindings: bus: renesas-bsc: allow additional properties

[Rob Herring <robh@kernel.org>]

On Wed, Oct 08, 2025 at 12:02:29PM +0200, Wolfram Sang wrote:
> 
> > > As said, I copied this from 'bus/allwinner,sun50i-a64-de2.yaml', so this
> > > got reviewed already. I have no strong opinions on your suggestions
> > > above. But whatever we agree on, it should be reused for all busses, I'd
> > > say. So, we should put it where it can be referenced?
> > 
> > In my defense, this is a DT bindings file for Renesas BSC, so we can
> > (and IMHO should) add restrictions.
> > 
> > If you want to go the fully generic way: as per the device tree
> > specification, a unit-address can take all characters from the same
> > set as a node name...
> 
> Well, I think consistency makes sense, but I don't even have a strong
> opinion on that. If DT maintainers are fine with having multiple
> patternProperties for busses, I will include the restriction you
> suggested.
> 

There's always examples not to follow...

Adding back context:

> +  "^[a-zA-Z][a-zA-Z0-9,+\\-._]{0,63}@[0-9a-fA-F]+$":                                                

The node name pattern before the '@' is just what's a valid node name. 
That is checked by the core schema on *every* node already. Even ',', 
'.', '+', and '_' are allowed which have long been deprecated. (Do you 
know how long it took to get QCom to stop with the 'qcom,' node name 
prefixes...).

The unit-address is the only part a bus schema should define. And there, 
lowercase hex is standard unless you have distinct fields (e.g. chip 
select and offset).

Rob