On Wed, May 18, 2022 at 02:41:11PM +0200, Clément Léger wrote: > Le Wed, 18 May 2022 15:05:03 +0300, > Vladimir Oltean <olteanv@xxxxxxxxx> a écrit : > > > On Tue, May 17, 2022 at 08:59:24PM -0500, Rob Herring wrote: > > > On Wed, May 11, 2022 at 06:33:37PM +0300, Vladimir Oltean wrote: > > > > On Wed, May 11, 2022 at 10:22:21AM -0500, Rob Herring wrote: > > > > > > +patternProperties: > > > > > > + "^ethernet-ports$": > > > > > > > > > > Move to 'properties', not a pattern. > > > > > > > > > > With that, > > > > > > > > > > Reviewed-by: Rob Herring <robh@xxxxxxxxxx> > > > > > > > > Even if it should have been "^(ethernet-)?ports$"? > > > > > > Why? Allowing 'ports' is for existing users. New ones don't need the > > > variability and should use just 'ethernet-ports'. > > > > > > Rob > > > > Yeah, ok, somehow the memo that new DSA drivers shouldn't support "ports" > > didn't reach me. They invariably will though, since the DSA framework is > > the main parser of the property, and that is shared by both old and new > > drivers. > > Should also the subnodes of "ethernet-ports" use the > "ethernet-port@[0-9]*" naming ? Or keeping the existing pattern is ok > (ie "^(ethernet-)?port@[0-4]$") ? I prefer the former, but care less. The whole reason for 'ethernet-' prefix is to make this distinct from the graph binding that uses ports and port. Rob
