On 25.10.2024 12:46 AM, Melody Olvera wrote: > > > On 10/24/2024 10:22 AM, Konrad Dybcio wrote: >> On 24.10.2024 6:33 PM, Rob Herring (Arm) wrote: >>> On Mon, 21 Oct 2024 16:21:09 -0700, Melody Olvera wrote: >>>> This series adds the initial device tree support for the SM8750 SoCs >>>> needed to boot to shell. This specifically adds support for clocks, >>>> pinctrl, rpmhpd, regulators, interconnects, and SoC and board >>>> compatibles. [...] >>> arch/arm64/boot/dts/qcom/sm8750-mtp.dtb: rsc@16500000: 'power-domains' is a required property >>> from schema $id: http://devicetree.org/schemas/soc/qcom/qcom,rpmh-rsc.yaml# >> This I'll address when reviewing the dt > > So for this, let me know if the following seems about right. > The rsc node needs a system-wide power domain defined. To accomplish this, I would define > a system_pd node and point both cluster pd nodes to it via power-domains, similar to what you did here: > https://lore.kernel.org/all/20240102-topic-x1e_fixes-v1-3-70723e08d5f6@xxxxxxxxxx/ > > Then, extrapolating from that, I can define a system-wide idle-state under the domain-idle-state node as follows: > domain_ss3: domain-sleep-0 { > compatible = "domain-idle-state"; > [...] (omitting a bunch of details, but you get the point) > }; > > And then point to that from the new system_pd node under the psci node: > system_pd: power-domain-system { > #power-domain-cells = <0>; > domain-idle-states = <&domain_ss3>; > }; > > And then of course in the rsc node, have power-domains = <&system_pd>; > > Is that correct? Yes, please go ahead with this Konrad
