Re: [PATCH 3/3] arm64: dts: qcom: msm8953: Add device tree for Billion Capture+

[Konrad Dybcio]

On 6/23/25 4:49 PM, cristian_ci wrote:
> On Monday, June 23rd, 2025 at 12:11, Konrad Dybcio <konrad.dybcio@xxxxxxxxxxxxxxxx> wrote:
> 
>> On 6/21/25 9:31 PM, cristian_ci wrote:
>>
>>> On Saturday, June 21st, 2025 at 12:17, Konrad Dybcio konrad.dybcio@xxxxxxxxxxxxxxxx wrote:
>>>
>>>>> +
>>>>> +&sdhc_1 {
>>>>> + vmmc-supply = <&pm8953_l8>;
>>>>> + vqmmc-supply = <&pm8953_l5>;
>>>>
>>>> you should add regulator-allow-set-load to these vregs
>>>
>>> So, do you mean I should add 'regulator-allow-set-load' property to 'pm8953_l5' and 'pm8953_l8' regulators? If so, should I do that for 'pm8953_l11' and 'pm8953_l12' regulators too (sdhc_2)?
>>
>>
>> Yes
>>
> 
> ACK, I'll do that in v2. BTW, since I've not such references in my downstream devicetree, IIUC what you mean, 'regulator-allow-set-load' property is now required in mainline for regulators supplying sdhc_{1|2}.

The sdhci driver calls regulator_set_load(), which requires that property

> Also considering I've transferred every value _as_is_ (except for max and min microvolt values ​​of one regulator - after the kernel log complained ​about ​that) for regulators/rpm_requests from downstream
> devicetree to mainline devicetree, since other msm8953 devicetrees seem to share the same situation of 'rimob' (I'm referring to
> 'potter', 'daisy', 'mido', 'tissot' and 'vince'), I wonder if it's not the case to standardize all these devicetrees in the same way. Any thoughts?

We've been there before, it's all good until one device stands out, and
then another, and then another..

Konrad