Re: [PATCH 2/9] dt-bindings: phy: qcom,qmp-usb: Add Glymur USB UNI PHY compatible

[Wesley Cheng <wesley.cheng@oss.qualcomm.com>]

On 9/23/2025 12:27 AM, Johan Hovold wrote:
> On Mon, Sep 22, 2025 at 06:00:04PM -0700, Wesley Cheng wrote:
>>
>>
>> On 9/20/2025 8:22 AM, Dmitry Baryshkov wrote:
>>> On Fri, Sep 19, 2025 at 08:21:01PM -0700, Wesley Cheng wrote:
>>>> The Glymur USB subsystem contains a multiport controller, which utilizes
>>
>>> two QMP UNI PHYs.  Add the proper compatible string for the Glymur SoC.
>>>> @@ -16,6 +16,7 @@ description:
>>>>    properties:
>>>>      compatible:
>>>>        enum:
>>>> +      - qcom,glymur-qmp-usb3-uni-phy
> 
> Odd indentation?
> 
>>>>          - qcom,ipq5424-qmp-usb3-phy
>>>>          - qcom,ipq6018-qmp-usb3-phy
>>>>          - qcom,ipq8074-qmp-usb3-phy
>>>> @@ -62,6 +63,8 @@ properties:
>>>>    
>>>>      vdda-pll-supply: true
>>>>    
>>>> +  refgen-supply: true
>>>
>>> You've added it, but it's not referenced as required. Why is it so?
> 
>> The refgen clock isn't always required on each and every platform unlike
>> the .9v and 1.2v rail/supply, which directly power the QMP PHY.  It only
>> really depends on how the refclk/CXO network is built for that
>> particular chipset.  The refgen ensures that we're properly voting for
>> the supply that is powering our CXO buffer.
> 
> I thought we discussed this before and concluded that this is not an
> accurate description of the hardware (even if you now call this supply
> refgen instead of qref):
Hi Johan,

refgen and qrefs are different things.  I will try to clarify as much as 
I can from the discussion you linked below.  (based on my understanding 
of the reference clock network)  The refgen is the main supply that 
controls the reference clock (CXO) into a specific branch.  Within each 
of these branches there are clock repeaters that are supplied by QREFs, 
and is basically the supply to the clkref switch controlled by the TCSR 
registers.

The way some of the tech blocks are connected, the QREFs/refgen may 
share the same regulator as some of the PHY's core supply.  Some may not 
even have QREFs at all.  One example is the QMP PHY that is associated 
to the primary controller on Glymur.  It has a refgen regulator, but no 
QREFs, hence we only need to vote the refgen accordingly.

I don't know if that helps you understand it a bit more to convince you 
of the new regulator addition.  If anything we may need to add an 
explicit QREF supply also :).

Thanks
Wesley Cheng
> 
> 	https://lore.kernel.org/lkml/aEBfV2M-ZqDF7aRz@hovoldconsulting.com/
> 
> Given your description above this still looks wrong (at least after a
> quick look).
> 
> Johan