On 01/12/2025 13:55, Konrad Dybcio wrote:
> On 12/1/25 1:50 PM, Casey Connolly wrote:
>>
>>
>> On 01/12/2025 13:48, Konrad Dybcio wrote:
>>> On 11/30/25 1:08 AM, David Heidelberg via B4 Relay wrote:
>>>> From: David Heidelberg <[email protected]>
>>>>
>>>> The lab and ibb regulators aren't used here. Disable them.
>>>>
>>>> Removes following warnings:
>>>> qcom-lab-ibb-regulator c440000.spmi:pmic@3:labibb: Failed to create device
>>>> link (0x180) with supplier c440000.spmi for
>>>> /soc@0/spmi@c440000/pmic@3/labibb/lab
>>>> qcom-lab-ibb-regulator c440000.spmi:pmic@3:labibb: Failed to create device
>>>> link (0x180) with supplier c440000.spmi for
>>>> /soc@0/spmi@c440000/pmic@3/labibb/ibb
>>>
>>> These are only vaguely related, as there's nothing to be wary about that's
>>> specific to these devices - it's just devlink being grumpy
>>>
>>>> Fixes: 288ef8a42612 ("arm64: dts: sdm845: add oneplus6/6t devices")
>>>> Signed-off-by: David Heidelberg <[email protected]>
>>>> ---
>>>> I assume this is right approach, as OLEDs on both devices are driven by
>>>> different regulators.
>>>>
>>>> Question is, if should be labibb nodes enabled by default?
>>>
>>> They're onboard. I'd rather keep them predictably parked than left in
>>> whatever (potentially ON) state the bootloader may leave them at
>>
>> Shouldn't they be default disabled in the pmic dtsi and only enabled on
>> the devices that actually use them? Many SDM845 devices with OLED panels
>> don't use these regulators.
>
> As I said, I wouldn't be surprised if they were enabled by the bootloader
> as part of some reference/common routine and left hanging. Linux will
> switch them off if they're never used and I'm fairly sure the users won't
> mind the odd couple dozen bytes of runtime kernel memory usage (which if
> we go that route probably balance out with the added couple characters for
> status=disabled in the resulting DTB)
Ahh yeah I understand, the DT node has to be enabled for the driver to
load and actually turn off the regulators if they're unused. Makes sense.
Thanks,>
> Konrad
--
// Casey (she/her)
