On Tue, Sep 23, 2025 at 03:17:20PM +0800, Chaoyi Chen wrote:
> On 9/23/2025 12:51 PM, Dmitry Baryshkov wrote:
>
> > On Tue, Sep 23, 2025 at 11:40:33AM +0800, Chaoyi Chen wrote:
> > > On 9/23/2025 11:17 AM, Dmitry Baryshkov wrote:
> > >
> > > > On Tue, Sep 23, 2025 at 09:53:06AM +0800, Chaoyi Chen wrote:
> > > > > Hi Dmitry,
> > > > >
> > > > > On 9/23/2025 9:12 AM, Dmitry Baryshkov wrote:
> > > > > > On Mon, Sep 22, 2025 at 09:20:34AM +0800, Chaoyi Chen wrote:
> > > > > > > From: Chaoyi Chen <[email protected]>
> > > > > > >
> > > > > > > The RK3399 SoC integrates two USB/DP combo PHYs, each of which
> > > > > > > supports software-configurable pin mapping and DisplayPort lane
> > > > > > > assignment. These capabilities enable the PHY itself to handle
> > > > > > > both
> > > > > > > mode switching and orientation switching, based on the Type-C plug
> > > > > > > orientation and USB PD negotiation results.
> > > > > > >
> > > > > > > While an external Type-C controller is still required to detect
> > > > > > > cable
> > > > > > > attachment and report USB PD events, the actual mode and
> > > > > > > orientation
> > > > > > > switching is performed internally by the PHY through software
> > > > > > > configuration. This allows the PHY to act as a Type-C multiplexer
> > > > > > > for
> > > > > > > both data role and DP altmode configuration.
> > > > > > >
> > > > > > > To reflect this hardware design, this patch introduces a new
> > > > > > > "mode-switch" property for the dp-port node in the device tree
> > > > > > > bindings.
> > > > > > > This property indicates that the connected PHY is capable of
> > > > > > > handling
> > > > > > > Type-C mode switching itself.
> > > > > > >
> > > > > > > Signed-off-by: Chaoyi Chen <[email protected]>
> > > > > > >
> > > > > > > Acked-by: Krzysztof Kozlowski <[email protected]>
> > > > > > > ---
> > > > > > >
> > > > > > > Changes in v4:
> > > > > > > - Remove "|" in description.
> > > > > > >
> > > > > > > Changes in v3:
> > > > > > > - Add more descriptions to clarify the role of the PHY in
> > > > > > > switching.
> > > > > > >
> > > > > > > Changes in v2:
> > > > > > > - Reuse dp-port/usb3-port in rk3399-typec-phy binding.
> > > > > > >
> > > > > > > .../devicetree/bindings/phy/rockchip,rk3399-typec-phy.yaml |
> > > > > > > 6 ++++++
> > > > > > > 1 file changed, 6 insertions(+)
> > > > > > >
> > > > > > > diff --git
> > > > > > > a/Documentation/devicetree/bindings/phy/rockchip,rk3399-typec-phy.yaml
> > > > > > >
> > > > > > > b/Documentation/devicetree/bindings/phy/rockchip,rk3399-typec-phy.yaml
> > > > > > > index 91c011f68cd0..83ebcde096ea 100644
> > > > > > > ---
> > > > > > > a/Documentation/devicetree/bindings/phy/rockchip,rk3399-typec-phy.yaml
> > > > > > > +++
> > > > > > > b/Documentation/devicetree/bindings/phy/rockchip,rk3399-typec-phy.yaml
> > > > > > > @@ -51,6 +51,12 @@ properties:
> > > > > > > '#phy-cells':
> > > > > > > const: 0
> > > > > > > + mode-switch:
> > > > > > Having the mode-switch here is a bit strange. I think the whole PHY
> > > > > > device should be an orientation-switch and mode-switch. Otherwise it
> > > > > > feels weird to me.
> > > > > I think this is a difference in practice. In the previous binding,
> > > > > there was already an orientation-switch under the usb-port. I trying
> > > > > to add both an orientation-switch and a mode-switch to the top-level
> > > > > device in v2. And Krzysztof reminded me that adding a mode-switch
> > > > > under the dp-port would be better, so I changed it to the current
> > > > > form :)
> > > > I couldn't find the comment on lore. Could you please point it out?
> > > Sorry, it is v1. I added an orientation-switch and a mode-switch in the
> > > top-level PHY [0]. Comment is here: [1]
> > My interpretation of [1] doesn't quite match yours. It doesn't say
> > anything about moving mode-switch to dp-port. It basically pointed out
> > that you already have two ports.
>
> Yes, I think this can be easily changed, as long as the issue you mentioned
> below is resolved.
>
>
> >
> > Also, I'm not sure how the current construction works: you register
> > switch and mux for the dev_fwnode(tcphy->dev), however the lookfup
> > functions should be looking for a device corresponding to the port OF
> > node (which doesn't exist).
>
> In v1, that is fwnode = dev_fwnode(tcphy->dev) . And dt like is:
>
>
> &tcphy {
> port {
> tcphy0_orientation_switch: endpoint@0 { ... };
> tcphy_dp_altmode_switch: endpoint@1 { ... };
> };
> };
>
>
> Since the binding already includes a "usb3-port" and a "dp-port", it can not
> add another new port.
>
> So after v1, that is fwnode = device_get_named_child_node(tcphy->dev,
> "usb3-port") . And dt like this:
I see it now, I was probably looking at the different revision, sorry
about it.
>
>
> &tcphy0_dp {
> port { ... };
> };
>
> &tcphy0_usb3 {
> port { ... };
> };
>
>
> Sorry, this looks a bit hacky. Do you have a better idea? Thank you.
No, it's fine from my POV now.
>
>
> >
> > >
> > > [0] https://lore.kernel.org/all/[email protected]/
> > >
> > > [1]
> > > https://lore.kernel.org/all/[email protected]/
> > >
> > >
> > >
> > > > >
> > > > > > > + description:
> > > > > > > + Indicates the PHY can handle altmode switching. In
> > > > > > > this case,
> > > > > > > + requires an external USB Type-C controller to report
> > > > > > > USB PD message.
> > > > > > > + type: boolean
> > > > > > > +
> > > > > > > port:
> > > > > > > $ref: /schemas/graph.yaml#/properties/port
> > > > > > > description: Connection to USB Type-C connector
> > > > > > > --
> > > > > > > 2.49.0
> > > > > > >
> > > > > --
> > > > > Best,
> > > > > Chaoyi
> > > > >
> > > --
> > > Best,
> > > Chaoyi
> > >
> > >
> > > --
> > > linux-phy mailing list
> > > [email protected]
> > > https://lists.infradead.org/mailman/listinfo/linux-phy
>
> --
> Best,
> Chaoyi
>
--
With best wishes
Dmitry
