Re: [PATCH 1/2] dt-bindings: usb: qcom,pmic-typec: drop port description

From: Dmitry Baryshkov
Date: Fri Mar 22 2024 - 09:41:39 EST


On Fri, 22 Mar 2024 at 14:35, Bryan O'Donoghue
<bryan.odonoghue@xxxxxxxxxx> wrote:
>
> On 22/03/2024 11:52, Dmitry Baryshkov wrote:
> > The PMIC Type-C controller doesn't have separate role-switching signal.
> > Instead it has an HS signal connection between embedded USB-C connector
> > node and the HS port of the USB controller.
>
> I take your point on port as a signal but the way type-c determines
> data-role is via the DR_Swap message.
>
> https://www.embedded.com/usb-type-c-and-power-delivery-101-power-delivery-protocol/
>
> We receive an IRQ which is a packet containing DR_Swap - TCPM consumes
> that data and does a data-role switch.
>
> The port then establishes the link between typec-port and redriver or PHY.
>
> So, I think HS should be dropped from the commit logs and names in both
> series.

Then the actual usage doesn't match the schema. usb-c-connector
clearly defines HS, SS and SBU ports
The snps,dwc3.yaml describes ports as ones handling usb-role-switch,
but then clearly writes that port@0 is HS and port@1 is SS. As such, I
think, the correct name for the ports is to have _hs_ in the name

We have pmic-typec/port, separate graph port for role-switching
(supported by TCPM code), but we didn't use it at all on our platforms
(nor do we need it, as we use the HS port).

>
> BTW for the GLINK devices I think the adsp firmware just notifies the
> APSS of the data-role switch so, these types of devices probably should
> have an epdoint with "usb_role_switch" in the name.
>
> ---
> bod
>


--
With best wishes
Dmitry