On 14/12/2023 17:20, Heiko Stübner wrote: > Am Donnerstag, 14. Dezember 2023, 17:07:27 CET schrieb Krzysztof Kozlowski: >> On 14/12/2023 16:22, Alex Bee wrote: >>> >>> Am 14.12.23 um 08:53 schrieb Krzysztof Kozlowski: >>>> On 13/12/2023 20:51, Alex Bee wrote: >>>>> Document the compatible for RK3128's HDMI controller block. >>>>> The integration for this SoC is somewhat different here: It needs the PHY's >>>> Please wrap commit message according to Linux coding style / submission >>>> process (neither too early nor over the limit): >>>> https://elixir.bootlin.com/linux/v6.4-rc1/source/Documentation/process/submitting-patches.rst#L597 >>> OK. Not sure why checkpatch --strict didn't tell me that I'm over the >>> limit here. >>>> >>>>> reference clock rate to calculate the ddc bus frequency correctly. This >>>>> clock is part of a power-domain (PD_VIO), so this gets added as an optional >>>>> property too. >>>> If clock is part of power domain, then the power domain must be in the >>>> clock controller, not here. So either you put power domain in wrong >>>> place or you used incorrect reason for a change. >>> Rockchip defines it's powerdomains per clock and I was little to much >>> in that world when writing this. Actually the controller itself is part >>> of the powerdomain. Will rephrase. >> >> Does it mean you have like 200 different power domains in one SoC? Then >> how are they different than clock if there is one-to-one mapping? > > It's more like the other way around. Controllers and their clocks belong > to specific power-domains. So there are of course more clocks than domains. That's fine and expected. Here the comment was suggested that you need to add power-domain because clock is in power-domain. That would be clearly wrong and instead the clock controller should model the power domain relationship. > > > > Best regards, Krzysztof