Re: [PATCH v5 1/4] dt-bindings: usb: add documentation for typec switch simple driver

[Rob Herring <robh@kernel.org>]

On Mon, Nov 9, 2020 at 6:24 AM Jun Li <jun.li@nxp.com> wrote:
> > From: Rob Herring <robh@kernel.org>
> > On Fri, Nov 6, 2020 at 5:07 AM Jun Li <jun.li@nxp.com> wrote:
> > > > From: Rob Herring <robh@kernel.org>

> > > > > +properties:
> > > > > +  compatible:
> > > > > +    const: typec-orientation-switch
> > > > > +
> > > > > +  switch-gpios:
> > > > > +    description: |
> > > > > +      gpio specifier to switch the super speed active channel,
> > > > > +      GPIO_ACTIVE_HIGH: GPIO state high for cc1;
> > > > > +      GPIO_ACTIVE_LOW:  GPIO state low for cc1.
> > > >
> > > > What does active mean? There isn't really an active and inactive state,
> > right?
> > > > It's more a mux selecting 0 or 1 input?
> > >
> > > Yes, I will change the description:
> > > gpio specifier to select the target channel of mux.
> >
> > I wonder if the existing mux bindings should be used here.
>
> If only consider typec switch via "gpio", existing "mux-gpio"
> binding may be used with property "mux-control-names" to be
> "typec-xxx" for match, but we still need create typec stuff at
> mux driver to hook to typec system, so little benefit, considering
> this binding is going to be for a generic typec orientation switch
> simple driver, I think a new typec binding make sense.

You can instantiate drivers without a compatible. That's just the easy
way. However, using the mux binding doesn't necessarily mean you have
to use 'mux-gpio'. Consider if you need to do more control than just
the GPIO line. For example, the chips you mentioned may have a s/w
controlled power supply or reset.

Also, consider what the mux would look like with different control
interfaces. That could be I2C or some sub-block in a PMIC or ??? I'm
sure we already have some examples. I'm not happy with these piecemeal
additions to TypeC related bindings that don't consider more than 1
h/w possibility.

> > > > I think you want flags 0 (aka GPIO_ACTIVE_HIGH) unless there's an
> > > > inverter in the middle.
> > >
> > > This depends on the switch IC design and board design, leave 2 flags
> > > (GPIO_ACTIVE_HIGH and GPIO_ACTIVE_LOW) can cover all possible cases.
> > >
> > > NXP has 2 diff IC parts for this:
> > > 1. PTN36043(used on iMX8MQ)
> > > Output selection control
> > > When SEL=0, RX_AP_*/TX_AP_* are connected to RX_CON_2*/TX_CON_2*, and
> > > RX_CON_1*/TX_CON_1* are connected to VDD thru low ohmic resistor.
> > > When SEL=1, RX_AP_*/TX_AP_* are connected to RX_CON_1*/TX_CON_1*, and
> > > RX_CON_2*/TX_CON_2* are connected to VDD thru low ohmic resistor.
> > >
> > > Board design connects RX_CON_1*/TX_CON_1* to typec connector CC1, so
> > > GPIO_ACTIVE_HIGH
> > >
> > > 2. CBTU02043(used on iMX8MP)
> > > SEL        Function
> > > --------------------------------------
> > > Low        A to B ports and vice versa
> > > High       A to C ports and vice versa
> > >
> > > Board design connects B to typec connector CC1, so GPIO_ACTIVE_LOW
> > >
> > > Therefore, we need 2 flags.
> >
> > I'm not saying you don't. Just that the description is a bit odd.
> > Please expand the description for how one decides how to set the flags.
>
> Misunderstood your point, OK, I thought the "how to set the flags" was
> simple and clear enough:
> Use GPIO_ACTIVE_HIGH if GPIO physical state high is for cc1; or
> Use GPIO_ACTIVE_LOW if GPIO physical state low is for cc1.

Okay.

> > > > > +examples:
> > > > > +  - |
> > > > > +    #include <dt-bindings/gpio/gpio.h>
> > > > > +    ptn36043 {
> > > > > +        compatible = "typec-orientation-switch";
> > > > > +        pinctrl-names = "default";
> > > > > +        pinctrl-0 = <&pinctrl_ss_sel>;
> > > > > +        switch-gpios = <&gpio3 15 GPIO_ACTIVE_HIGH>;
> > > > > +
> > > > > +        port {
> > > > > +                usb3_data_ss: endpoint {
> > > > > +                        remote-endpoint = <&typec_con_ss>;
> > > >
> > > > The data goes from the connector to here and then where? You need a
> > > > connection to the USB host controller.
> > >
> > > The orientation switch only need interact with type-c, no any
> > > interaction with USB controller, do we still need a connection to it?
> >
> > If you have 2 USB hosts and 2 connectors (and 2 muxes), how would you describe
> > which connector goes with which host?
>
> One instance of typec orientation switch defined by this binding only for
> One typec connector. With that, my understanding is
> Whether a connection need be described depends on if the connector
> (typec driver) need notify the host controller driver to do something
> (e.g. role switch need a connection between controller node and connector
> node for controller driver to swap usb role). If the mux/switch control is
> transparent to usb host controller (e.g. my case, usb controller drivers
> normally don't need do anything for orientation change), there is no need
> to describe connection between orientation switch node and host controller
> node.

There can be several reasons you need to know the association. When
writing the DT you can't assume what information is or isn't needed.
That may vary by h/w or can evolve in an OS and the DT shouldn't
change.

Rob