Re: [PATCH] PCI: qcom: add missing supplies required for msm8996

[Srinivas Kandagatla]

On 23/01/18 10:14, Stanimir Varbanov wrote:
> Hi,
>
> On 01/23/2018 11:46 AM, Srinivas Kandagatla wrote:
> > On 23/01/18 09:23, Stanimir Varbanov wrote:
> > > Hey Srini,
> > >
> > > As there are no comments I'd propose to change the endpoint supplies to
> > > more generic names.
> > Sure, I will respin this with your suggestions, except the 3v3 and 1v5
> > suffix due to the reasons below:
> > > > +- vdda-1p8-supply:
> > > > +    Usage: required for msm8996
> > > > +    Value type: <phandle>
> > > > +    Definition: A phandle to the 1.8v analog power supply
> > > > +
> > >
> > > This should be dropped, because it is part of the phy.
> > Yep.
> >
> > > >    - vdda_phy-supply:
> > > >        Usage: required for ipq/apq8064
> > > >        Value type: <phandle>
> > > > @@ -189,6 +194,15 @@
> > > >        Value type: <phandle>
> > > >        Definition: A phandle to the analog power supply for IC which
> > > > generates
> > > >                reference clock
> > > > +- vddpe-supply:
> > > > +    Usage: optional
> > > > +    Value type: <phandle>
> > > > +    Definition: A phandle to the PCIe endpoint power supply
> > >
> > > vddpe_3v3-supply
> > Why do we need suffix here? AFAIU, It does not add any value, instead it
> > would confuse the users.
>
> vddpe and vddpe1 is already confusing as well.

I partly agree with you.

How would you represent if there are two power 3v3 supplies for the 
endpoint?

> Lets imagine that powering up the endpointX needs some specific sequence
> between 3v3 and 1v5 and endpointY (which could be connected on the same
> PCIe lane) has different power sequence, how we would handle that in the
> qcom pcie host driver?

power sequencing is all together a different issue, that is not 
addressed in this patch. Am hoping that this will be fixed as part of 
making pwrseq interface more generic. Not sure where it endedup now!!

--srini


> > These are power supplies for endpoint which could be of any voltage. In
>
> I don't think that could be any values see PCIe mini card
> electromechanical specification. There on the connector are provided 3v3
> and 1v5.
>
> > this case both endpoint supplies are 3v3, these could be 1.8 or 5v or
> > 12v in some other cases.
>
> If we see hw designs with 5v and 12v we could extend the binding and the
> driver with support for them. I want to be exact in the names and
> voltages in the driver and bindings.

>