On 18/10/2023 17:07, Conor Dooley wrote: >>> Then it seems like what you need is something like >>> oneOf: >>> - items: >>> - const: google,tentacruel-sku262144 >>> - const: google,tentacruel >>> - const: mediatek,mt8186 >>> - items: >>> - enum: >>> - google,tentacruel-sku262145 >>> - google,tentacruel-sku262146 >>> - google,tentacruel-sku262147 >>> - const: google,tentacruel-sku262144 >>> - const: google,tentacruel >>> - const: mediatek,mt8186 >>> >>> What you have at the moment just seems like a hack because you want to >>> stuff all of these compatible strings into a single dts. >> >> It is. And it works OK downstream. The reason we want to stuff them in >> one dts is because the firmware will not generate the fallback to >> sku262144 as the scheme above suggests. > > I'm not going to ack the hack that you have here, sorry. Maybe Rob or > Krzysztof will. The list your firmware generates above doesn't even > match the contents of this patch, with the extra "rev-4" compatibles. No Acks from me either... Chromebooks board compatibles are a long standing confusion for me and every time I forget why this is so counter-intuitively. My comments for changing this were rather receiving feedback "but our firmware likes it that way". OTOH, board compatibles just have to be unique, so we could accept some weird, counter-intuitive combinations... but it does not mean I should be happy with them. Best regards, Krzysztof
