Re: [PATCH v4 2/2] arm64: dts: qcom: Add Xiaomi 12 Lite 5G (taoyao) DTS

[Stanislav Zaikin <zstaseg@gmail.com>]

On 3/16/26 9:47 PM, Val Packett wrote:
>
> On 3/16/26 8:57 AM, Stanislav Zaikin wrote:
>> On 3/16/26 5:05 AM, Val Packett wrote:
>>> On 3/5/26 6:39 AM, Stanislav Zaikin wrote:
>>>
>>>> Xiaomi 12 Lite 5G is a handset released in 2022
>>>>
>>>> This commit has the following features working:
>>>> - Display (with simple fb)
>>>> - Touchscreen
>>>> - UFS
>>>> - Power and volume buttons
>>>> - Pinctrl
>>>> - RPM Regulators
>>>> - Remoteprocs - wifi, bluetooth
>>>> - USB (Device Mode)
>>>> […]
>>>> +        vreg_l7b_2p96: ldo7 {
>>>> +            regulator-name = "vreg_l7b_2p96";
>>>> +            /* Constrained for UFS VCC, at least until UFS driver 
>>>> scales voltage */
>>>> +            regulator-min-microvolt = <2952000>;
>>>> +            regulator-max-microvolt = <2952000>;
>>>> +            regulator-initial-mode = <RPMH_REGULATOR_MODE_HPM>;
>>>> +        };
>>>> […]
>>>> +&ufs_mem_hc {
>>>> +    reset-gpios = <&tlmm 175 GPIO_ACTIVE_LOW>;
>>>> +
>>>> +    vcc-supply = <&vreg_l7b_2p96>;
>>>> +    vcc-max-microamp = <800000>;
>>>> +    /*
>>>> +     * Technically l9b enables an eLDO (supplied by s1b) which 
>>>> then powers
>>>> +     * VCCQ2 of the UFS.
>>>> +     */
>>>> +    vccq-supply = <&vreg_l9b_1p2>;
>>>> +    vccq-max-microamp = <900000>;
>>>
>>> If this device has UFS 3.x flash, which it most likely does, the 
>>> copy-pasted eLDO comment does not apply.. and VCC range is 
>>> 2.4V-2.7V, 2.95 is overvoltage! 
>>
>> Hello Val,
>>
>> Thank you for the review.
>>
>> Though you're right that this comment is a copy-paste, taoyao has 
>> ufs2. This is the explaining comment from the downstream sources:
>>
>>> /*
>>> * Yupik target supports both UFS2.2 & UFS3.1, here
>>> * vccq2 is supplied via eLDO, [..]
>>
>> So I thought it's better to keep a small comment regarding eLDO.
>
> That's a comment from the qcom reference sources! I've literally seen 
> it already in the downstream DTS for the device I'm currently working 
> on (motorola-dubai) which definitely has UFS 3, hehe.
>
> You can read 
> /sys/bus/platform/devices/1d84000.ufshc/device_descriptor/specification_version 
> to see which UFS spec version your flash is (0x310 == 3.1 etc.)
/home/user # cat 
/sys/bus/platform/devices/1d84000.ufshc/device_descriptor/specification_version 


0x0220

> (Lots of interesting stuff around there, there's also 
> health_descriptor that shows your flash's wear estimation..)
>
> Also if you have the schematics, you can check for whether an LDO is 
> actually present there and whether VCCQ or VCCQ2 is connected :)
Unfortunately, I wasn't able to find schematics. It also a bummer that I 
don't know panel's name because I have a (semi-generated) driver for it.
>>> If it's actually UFS 2 after all, it would be better to rename the 
>>> property to vccq2-supply.
>> Could you elaborate please? Do you mean l9b? 
>
> No, the "vccq-supply". If you look at 
> Documentation/devicetree/bindings/ufs/ufs-common.yaml you can see it 
> supports both `vccq-supply` and `vccq2-supply`, the former is used for 
> UFS3 and the latter for UFS2.

Will fix it in v5. Thanks a lot for so many inputs!

Best regards,
Stanislav