On Thursday 18 February 2016 15:27:06 Daniel Drake wrote: > On Thu, Feb 18, 2016 at 3:04 PM, Carlo Caione <carlo@xxxxxxxxxx> wrote: > >> However, if some or all of the other devices actually are entirely > >> made up of register ranges within cbus, that would indicate that > >> cbus itself is not just a collection of random registers but something > >> that could be considered a bus of itself in hardware, and then > >> we could represent the other devices as children of this bus. > > > > I think that this is exactly the case. We are missing this cbus in the > > DTS because (for lacking of proper documentation) we are not sure > > about start / end / size. > > Here's a hint from the vendor kernel > https://github.com/endlessm/linux-meson/blob/master/arch/arm/mach-meson8b/iomapping.c#L87 > > Having a cbus bus node with child devices does sound like it would > reflect this particular view of the hardware design. How would we then > represent the hwrev registers under that? > > I am also curious if this is the common practice. We were working with > Exynos devices before, and even though many of the components are on > the AXI bus there, there is no AXI bus representation in the DT. But > now that I go digging, I see other SoCs that do have a DT bus > representation very similar to what's being described, such as the apb > and axi busses in mmp2.dtsi. Is one approach preferred over the other > for new SoC support? I would always prefer having the dts files describe the hardware as best as they can. Arnd -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html