On Fri 17 Apr 07:26 PDT 2020, Sibi Sankar wrote: > Add TCSR node and update MSS node to support MSA based Modem boot on > SC7180 SoCs. > > Signed-off-by: Sibi Sankar <sibis@xxxxxxxxxxxxxx> > --- > > Depends on the following bindings: > iommus: https://patchwork.kernel.org/patch/11443101/ > spare-regs: https://patchwork.kernel.org/patch/11491425/ > > arch/arm64/boot/dts/qcom/sc7180-idp.dts | 42 +++++++++++++++++++++++++ > arch/arm64/boot/dts/qcom/sc7180.dtsi | 5 +++ > 2 files changed, 47 insertions(+) > > diff --git a/arch/arm64/boot/dts/qcom/sc7180-idp.dts b/arch/arm64/boot/dts/qcom/sc7180-idp.dts > index e613d70cc0198..6f472872be1a3 100644 > --- a/arch/arm64/boot/dts/qcom/sc7180-idp.dts > +++ b/arch/arm64/boot/dts/qcom/sc7180-idp.dts > @@ -319,6 +319,48 @@ &qupv3_id_1 { > status = "okay"; > }; > > +&remoteproc_mpss { > + compatible = "qcom,sc7180-mss-pil"; > + reg = <0 0x04080000 0 0x410>, <0 0x04180000 0 0x48>; I think we should overspecify the properties in the platform dtsi, whenever possible - it shouldn't be a problem that the pas driver doesn't use all the properties provided by the binding. As such I think you should move the reg, clocks, resets, halt regs and power-domains to the platform. > + reg-names = "qdsp6", "rmb"; > + > + clocks = <&gcc GCC_MSS_CFG_AHB_CLK>, > + <&gcc GCC_MSS_Q6_MEMNOC_AXI_CLK>, > + <&gcc GCC_MSS_NAV_AXI_CLK>, > + <&gcc GCC_MSS_SNOC_AXI_CLK>, > + <&gcc GCC_MSS_MFAB_AXIS_CLK>, > + <&rpmhcc RPMH_CXO_CLK>; > + clock-names = "iface", "bus", "nav", "snoc_axi", > + "mnoc_axi", "xo"; > + > + iommus = <&apps_smmu 0x460 0x1>, <&apps_smmu 0x444 0x2>; > + > + resets = <&aoss_reset AOSS_CC_MSS_RESTART>, > + <&pdc_reset PDC_MODEM_SYNC_RESET>; > + reset-names = "mss_restart", "pdc_reset"; > + > + qcom,halt-regs = <&tcsr_mutex_regs 0x23000 0x25000 0x24000>; > + qcom,spare-regs = <&tcsr_regs 0xb3e4>; > + > + power-domains = <&aoss_qmp AOSS_QMP_LS_MODEM>, > + <&rpmhpd SC7180_CX>, > + <&rpmhpd SC7180_MX>, > + <&rpmhpd SC7180_MSS>; > + power-domain-names = "load_state", "cx", "mx", "mss"; > + > + /delete-property/memory-region; > + > + status = "okay"; > + > + mba { > + memory-region = <&mba_mem>; When I wrote this I was under the impression that memory-region wasn't allowed to take an array of regions, perhaps we can make the mss binding and driver support a multi-cell memory-region in the of_node directly and drop these sub children. Then it would be a cleaner update of the pas' memory-region. But I'm fine with us putting this part on the todo list for the time being... Regards, Bjorn > + }; > + > + mpss { > + memory-region = <&mpss_mem>; > + }; > +}; > + > &uart3 { > status = "okay"; > > diff --git a/arch/arm64/boot/dts/qcom/sc7180.dtsi b/arch/arm64/boot/dts/qcom/sc7180.dtsi > index e319762a0bffc..c49801ddb9d70 100644 > --- a/arch/arm64/boot/dts/qcom/sc7180.dtsi > +++ b/arch/arm64/boot/dts/qcom/sc7180.dtsi > @@ -981,6 +981,11 @@ tcsr_mutex_regs: syscon@1f40000 { > reg = <0 0x01f40000 0 0x40000>; > }; > > + tcsr_regs: syscon@1fc0000 { > + compatible = "syscon"; > + reg = <0 0x01fc0000 0 0x40000>; > + }; > + > tlmm: pinctrl@3500000 { > compatible = "qcom,sc7180-pinctrl"; > reg = <0 0x03500000 0 0x300000>, > -- > The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, > a Linux Foundation Collaborative Project
