Re: [PATCH v8 3/4] arm64: dts: qcom: sc7280: Add PCIe nodes for IDP board

[Prasad Malisetty <pmaliset@codeaurora.org>]

On 2021-09-21 01:21, Stephen Boyd wrote:
> Quoting Prasad Malisetty (2021-09-17 10:15:46)
>> diff --git a/arch/arm64/boot/dts/qcom/sc7280-idp.dtsi 
>> b/arch/arm64/boot/dts/qcom/sc7280-idp.dtsi
>> index 99f9ee5..ee00df0 100644
>> --- a/arch/arm64/boot/dts/qcom/sc7280-idp.dtsi
>> +++ b/arch/arm64/boot/dts/qcom/sc7280-idp.dtsi
>> @@ -199,6 +199,39 @@
>>         modem-init;
>>  };
>> 
>> +&pcie1 {
>> +       status = "okay";
>> +
>> +       perst-gpio = <&tlmm 2 GPIO_ACTIVE_LOW>;
>> +       pinctrl-0 = <&pcie1_default_state &nvme_ldo_enable_pin>;
>> +};
>> +
>> +&pcie1_phy {
>> +       status = "okay";
>> +
>> +       vdda-phy-supply = <&vreg_l10c_0p8>;
>> +       vdda-pll-supply = <&vreg_l6b_1p2>;
>> +};
>> +
>> +&pcie1_default_state {
>> +       reset-n {
>> +               pins = "gpio2";
>> +               function = "gpio";
>> +
>> +               drive-strength = <16>;
>> +               output-low;
>> +               bias-disable;
>> +       };
>> +
>> +       wake-n {
>> +               pins = "gpio3";
>> +               function = "gpio";
>> +
>> +               drive-strength = <2>;
>> +               bias-pull-up;
>> +       };
> 
> I think the previous round of this series Bjorn was saying that these
> should be different nodes and tacked onto the pinctrl-0 list for the
> pcie1 device instead of adding them as subnodes of the "default state".
> 

Hi Stephen,

Here NVMe gpio entry is endpoint related where as wake-n and reset-n are 
PCIe controller gpio's. I think Bjorn was saying keep endpoint related 
gpio (NVMe) in separate state entry in pinctrl-0 list.

Thanks
-Prasad.

>> +};
>> +
>>  &pmk8350_vadc {
>>         pmk8350_die_temp {
>>                 reg = <PMK8350_ADC7_DIE_TEMP>;
>> @@ -343,3 +376,10 @@
>>                 bias-pull-up;
>>         };
>>  };
>> +
>> +&tlmm {
>> +       nvme_ldo_enable_pin: nvme_ldo_enable_pin {
> 
> Please use dashes where you use underscores in node names
> 
>        nvme_ldo_enable_pin: nvme-ldo-enable-pin {
> 
>> +               function = "gpio";
>> +               bias-pull-up;
> 
> Of course with that said, the name of this node makes it sound like 
> this
> is a gpio controlled regulator. Why not use that binding then and 
> enable
> the regulator either by default with regulator properties like
> regulator-always-on and regulator-boot-enable and/or reference it from
> the pcie device somehow so that it can be turned off during suspend?
> 
Agree, I will add in next patch series.

>> +       };
>> +};