On 2023/4/19 2:46, Rob Herring wrote: > On Tue, Apr 18, 2023 at 1:42 PM Rob Herring <robh@xxxxxxxxxx> wrote: >> >> On Thu, Apr 13, 2023 at 10:41:23AM +0200, Krzysztof Kozlowski wrote: >>> On 13/04/2023 04:34, Changhuang Liang wrote: >>>>>>>> + lane_maps: >>>>>>> >>>>>>> Why did this appear? Underscores are not allowed. It looks like you >>>>>>> re-implement some standard property. >>>>>>> >>>>>> >>>>>> Will change to lane-maps. >>>>>> Yes, according to Vinod advice, lane mapping table use device tree >>>>>> to parse makes sense. >>>>> >>>>> Hm, I have a feeling that I saw such property, so you should dig into >>>>> existing and in-flight bindings. >>>>> >>>>> Best regards, >>>>> Krzysztof >>>>> >>>> >>>> A standard property? Like "clocks" or "resets"? >>> >>> Like lane-polarities now submitted to one MIPI. >>> >> >> data-lanes perhaps? > > Except that is for the controller's endpoint rather than the phy. > Presumably if the controller knows the mapping, then it can tell the > phy if it needs the information. IOW, don't just copy 'data-lanes' to > the phy. Follow the normal patterns. > > Rob I am not sure if phy can fetch from the other controller's endpoint. In addition, like our JH7110 SoC, it have data-lanes configure (data-lanes = <1 2>) in csi2rx controller, but this data-lanes configure is not appropriate to phy, maybe they are independent.
