Re: [PATCH v6 03/12] arm64: dts: qcom: sdm845-starqltechn: fix usb regulator mistake

[Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>]

On 2.12.2024 9:10 PM, Dzmitry Sankouski wrote:
> пн, 2 дек. 2024 г. в 18:09, Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>:
>>
>> On 2.12.2024 2:41 PM, Dzmitry Sankouski wrote:
>>> сб, 26 окт. 2024 г. в 13:41, Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>:
>>>>
>>>> On 8.10.2024 6:51 PM, Dzmitry Sankouski wrote:
>>>>> Usb regulator was wrongly pointed to vreg_l1a_0p875.
>>>>> However, on starqltechn it's powered from vreg_l5a_0p8.
>>>>>
>>>>> Fixes: d711b22eee55 ("arm64: dts: qcom: starqltechn: add initial device tree for starqltechn")
>>>>> Signed-off-by: Dzmitry Sankouski <dsankouski@gmail.com>
>>>>>
>>>>> ---
>>>>
>>>> I really really doubt that the supplies for on-SoC PHYs were altered,
>>>> given these regulators are assigned based on their specific characteristics
>>>>
>>>
>>> From rooted android system from klabit87 on starqltechn:
>>>
>>> ```
>>> starqltechn:/ # cat /proc/cpuinfo | grep Hardware
>>> Hardware        : Qualcomm Technologies, Inc SDM845
>>> starqltechn:/ # uname -a
>>> Linux localhost 4.9.186-klabitV6.5 #1 SMP PREEMPT Thu Dec 10 19:42:53
>>> CST 2020 aarch64
>>> starqltechn:/ # cat
>>> /sys/kernel/debug/regulator/soc:rpmh-regulator-ldoa1-pm8998_l1/consumers
>>> Device-Supply                    EN    Min_uV   Max_uV  load_uA
>>> ae90000.qcom,dp_display-vdda-0p9 N     880000   880000        0
>>> 1d87000.ufsphy_mem-vdda-phy      Y     880000   880000    62900
>>> ae96400.qcom,mdss_dsi_phy0-vdda-0p9 N          0        0        0
>>> ae94400.qcom,mdss_dsi_phy0-vdda-0p9 Y     880000   880000    36000
>>> 1c00000.qcom,pcie-vreg-0.9       Y     880000   880000    24000
>>> pm8998_l1                        N          0        0        0
>>> starqltechn:/ # cat
>>> /sys/kernel/debug/regulator/soc:rpmh-regulator-ldoa5-pm8998_l5/consumers
>>> Device-Supply                    EN    Min_uV   Max_uV  load_uA
>>> ae90000.qcom,dp_display-vdda-usb1-ss-core N          0        0        0
>>> 88e2000.qusb-vdd                 Y     800000   800000        0
>>> 88e8000.ssphy-vdd                Y     800000   800000        0
>>> pm8998_l5                        N          0        0        0
>>> ```
>>>
>>> I also downloaded kernel source from Samsung, to check its dts,
>>> and it also powers qusb@88e2000 and ssphy@88e8000 from 'pm8998_l5' regulator.
>>
>> Interesting.. could you try to forcefully shut down L5A (I doubt you'll
>> be able to do so with L1A given it powers so much digital logic), or
>> set its regulator-min&max-microvolt to a way-too-low voltage (vmin
>> seems to be 312000)?
>>
>> You should then be able to tell fairly easily, depending on whether usb2
>> still works after a replug
>>
> 
> Usb doesn't worked in such experiment:
> Here's the log:
> ```
> [    3.541733] ldo5: Setting 312000-312000uV
> ......
> [    3.808224] dwc3 a600000.usb: Adding to iommu group 5
> [    3.823116] qcom-qmp-combo-phy 88e8000.phy: phy initialization timed-out
> [    3.824323] phy phy-88e8000.phy.0: phy init failed --> -110
> [    3.825916] dwc3 a600000.usb: error -ETIMEDOUT: failed to initialize core
> [    3.827208] dwc3 a600000.usb: probe with driver dwc3 failed with error -110
> [    3.828548] probe of a600000.usb returned 110 after 20399 usecs

Wow, I'm really surprised

That said, thank you for confirming

Konrad