On Fri, Apr 01, 2022 at 09:48:28PM +0300, Laurent Pinchart wrote: > On Fri, Apr 01, 2022 at 01:33:15PM -0500, Rob Herring wrote: > > On Fri, Apr 1, 2022 at 1:25 PM Marek Vasut wrote: > > > On 4/1/22 20:21, Rob Herring wrote: > > > > On Fri, Apr 1, 2022 at 1:06 PM Marek Vasut wrote: > > > >> On 4/1/22 19:34, Rob Herring wrote: > > > >>> On Fri, Apr 01, 2022 at 03:22:19AM +0200, Marek Vasut wrote: > > > >>>> On 4/1/22 01:52, Rob Herring wrote: > > > >>>>> On Wed, 23 Mar 2022 16:48:23 +0100, Maxime Ripard wrote: > > > >>>>>> MIPI-DSI devices, if they are controlled through the bus itself, have to > > > >>>>>> be described as a child node of the controller they are attached to. > > > >>>>>> > > > >>>>>> Thus, there's no requirement on the controller having an OF-Graph output > > > >>>>>> port to model the data stream: it's assumed that it would go from the > > > >>>>>> parent to the child. > > > >>>>>> > > > >>>>>> However, some bridges controlled through the DSI bus still require an > > > >>>>>> input OF-Graph port, thus requiring a controller with an OF-Graph output > > > >>>>>> port. This prevents those bridges from being used with the controllers > > > >>>>>> that do not have one without any particular reason to. > > > >>>>>> > > > >>>>>> Let's drop that requirement. > > > >>>>>> > > > >>>>>> Signed-off-by: Maxime Ripard <maxime@xxxxxxxxxx> > > > >>>>>> --- > > > >>>>>> .../devicetree/bindings/display/bridge/chipone,icn6211.yaml | 1 - > > > >>>>>> .../devicetree/bindings/display/bridge/toshiba,tc358762.yaml | 1 - > > > >>>>>> 2 files changed, 2 deletions(-) > > > >>>>>> > > > >>>>> > > > >>>>> I tend to agree with port@0 not being needed and really like > > > >>>>> consistency. > > > >>>> > > > >>>> The consistent thing to do would be to always use port@0 and OF graph, no ? > > > >>> > > > >>> I guess it depends how wide our scope for consistency is. Just DSI bus > > > >>> controlled bridges? DSI panels? All bridges and panels? Any panel > > > >>> without a control interface has the same dilemma as those can be a child > > > >>> of the display controller (or bridge) and not even use OF graph. > > > >> > > > >> I would likely opt for the OF graph in all cases, panels, bridges, > > > >> controllers. Then it would be consistent. > > > >> > > > >>> All simple panels don't require 'port' either. That's presumably only > > > >>> consistent because we made a single schema. I'd assume 'non-simple' > > > >>> panels with their own schema are not consistent. > > > >> > > > >> Maybe we would start requiring that port even for simple panels ? > > > >> The port is physically there on that panel after all. > > > > > > > > Fix this in all the dts files and then I'll agree. Though I think this > > > > ship has already sailed. I'd like to someday get to platforms without > > > > warnings and not just keep adding new warnings. > > > > > > I doubt we can fix existing DTs, but can we at least require it for new > > > DTs ? > > > > We don't have any way to do that currently and get to warning free for > > all DTs. We'd need to be able to disable specific checks for specific > > DTs. I've thought about it, but haven't come up with a way to do it. > > I think Marek may have meant new bindings. While I do agree that > inconsistencies in sources can cause new submissions to blindly copy > mistakes, it shouldn't be a reason in itself to carry historical binding > design mistakes in new bindings. This patch and the whole thread is about consistency of existing bindings. The only way to get some consistency reliably (beyond depending on reviews which doesn't achieve it) is either have a meta-schema to check something or make it a common schema. If we have new and old bindings, then we'd need panel-simple-old.yaml and panel-simple-new.yaml. There's other options probably, but not without more work on the tools. Rob
