On Mon Sep 12, 2022 at 4:17 PM IST, Krzysztof Kozlowski wrote: > On 12/09/2022 05:33, Sireesh Kodali wrote: > > On Fri Sep 9, 2022 at 1:30 PM IST, Krzysztof Kozlowski wrote: > >> On 08/09/2022 20:49, Sireesh Kodali wrote: > >>> The qcom,pronto compatible is used in the wcn36xx driver to determine > >>> which register to access. However, this compatible was not documented. > >>> This patch documents the existing compatible as is, since it isn't > >>> immediately clear why the wcn36xx driver uses this extra compatible, > >>> rather than relying directly on the regular compatible string. > >> > >> The patch does much more - messes entirely all compatibles... > > > > Is there another preferred way to handle this? > > The one which does not introduces any other changes than what you wrote > here. You wrote here, that qcom,pronto is being added, so why some > things are changed to oneOf or to enums? I think I didn't explain what the patch is doing properly.. Right now, the remoteproc driver expects "qcom,pronto-v2/v3" for pronto devices, and "qcom,riva" for riva. This has been already documented properly. The wcn36xx driver expects "qcom,pronto" for pronto devices. I am not sure why wcn36xx was written like this. But it is the current state of the driver that I am documenting. So the device tree will have compatible strings like "qcom,pronto-v2", "qcom,pronto"; Both need to be present. For Riva it would just be compatible = "qcom,riva"; Hence the oneOf. I will add a comment explaining this in v3 > > Best regards, > Krzysztof
