On 04/11/2024 11:31, Marc Kleine-Budde wrote: > On 04.11.2024 10:27:04, Krzysztof Kozlowski wrote: >>> +properties: >>> + compatible: >>> + oneOf: >>> + - enum: >>> + - ti,tcan4552 >>> + - ti,tcan4553 >>> + - ti,tcan4x5x >> >> That's not really what old binding said. >> >> It said for example: >> "ti,tcan4552", "ti,tcan4x5x" >> >> Which is not allowed above. You need list. Considering there are no >> in-tree users of ti,tcan4x5x alone, I would allow only lists followed by >> ti,tcan4x5x. IOW: disallow ti,tcan4x5x alone. > > I'd like to keep the old binding. Any reason why? It's against of bindings policies. If this was a new binding, it would be a clear NAK, so here due to lack of user we can make it correct. > >> Mention this change to the binding in the commit message. > > The tcan4x5x chip family has 2 registers for automatic detection of the > chip variant. While the ID2 register of the tcan4550 is 0x0, the > registers for the tcan4552 and tcan4553 contain 4552 and 4553 > respectively in ASCII. This does not grant exception for wildcards. This only says that one model should be used as fallback. > > The driver was originally added for the tcan4550 chip, but currently > only has a compatible for "ti,tcan4x5x". That's not really related to discussion. We do not talk about changing driver. Best regards, Krzysztof
