On 26/07/2023 13:55, Mark Brown wrote: > On Wed, Jul 26, 2023 at 09:32:17AM +0200, Krzysztof Kozlowski wrote: >> On 25/07/2023 12:43, Mark Brown wrote: > >>> This is just adding an OS specific note in the description, it's not >>> actually marking the properties as required. > >> They cannot be required, because it depends whether the regulator is >> used or not. IOW, they are not required for unused regulators, which is >> not possible to encode in the schema. > > Oh, you mean if the regulator is in use in the system rather than if > it's enabled! Enabled as "always-on" or "boot-on" could be encoded in the schema with multiple if::then:. But it is not enough, because regulators can be enabled on demand by drivers. So that's what I meant by "used". > I suspect that there's a requirement that either at least > one of the supplies be provided so that the chip I/O works, or there's > some other currently undocumented supply that is required for that > reason. I can add requirement of at least one supply. I don't think it changes much, but sure. > > BTW there's also a formatting error: > > + vin3-supply: > + description: > + Input supply for ldo3, required if regulatoris enabled > > missing space before is. I'll fix with above at least one supply required. Best regards, Krzysztof
