On Sun, 12 Mar 2023 11:47:19 +0100 Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx> wrote: > On 11/03/2023 20:34, Jonathan Cameron wrote: > > >> + > >> +additionalProperties: false > >> + > >> +required: > >> + - compatible > >> + - reg > >> + - interrupt > > It would nice to relax the need for an interrupt if the device is still useable > > with timeouts etc. Board folk have a habit of deciding they don't need to wire > > up interrupts. We can relax that a later date though if you prefer not to do > > it now. > >> + - vdd-supply > >> + - vled-supply > > > > Whilst true that the supplies need to be connected, that doesn't > > mean they need to provided in the device tree binding. If they are > > always powered up I think we can fallback to stub regulators. > > We can, but others might not. The binding should still require them if > they are required for device to work. Mark also made it clear recently: > > https://lore.kernel.org/all/31ca0ede-012c-4849-bf25-d0492b116681@xxxxxxxxxxxxx/ > https://lore.kernel.org/all/5cd6764c-9b04-42ea-932d-9f14aa465605@xxxxxxxxxxxxx/ > https://lore.kernel.org/all/f6f02138-8ef9-4a33-9b51-0b7cd371230f@xxxxxxxxxxxxx/ OK. Then there are a lot of bindings to fix. Seems odd to me but meh it's not something I care about. Note this means that we can't have trivial-device.yaml for instance. Ah well, I guess views change or crystallise over time or just differed in the first place. Jonathan > > Best regards, > Krzysztof >
