Re: [PATCH v2 1/2] dt-bindings: usb: misc: Add binding for Microchip usb5744 hub

[Rob Herring <robh@kernel.org>]

On Thu, Feb 11, 2021 at 3:35 AM Michal Simek <michal.simek@xilinx.com> wrote:
>
> Hi Rob,
>
> On 2/10/21 11:22 PM, Rob Herring wrote:
> > On Tue, Feb 09, 2021 at 11:48:09AM +0100, Michal Simek wrote:
> >> From: Piyush Mehta <piyush.mehta@xilinx.com>
> >>
> >> Added dt binding for usb5744 driver.
> >>
> >> Signed-off-by: Piyush Mehta <piyush.mehta@xilinx.com>
> >> Signed-off-by: Michal Simek <michal.simek@xilinx.com>
> >> ---
> >>
> >> Changes in v2: None
> >>
> >>  .../bindings/usb/microchip,usb5744.yaml       | 56 +++++++++++++++++++
> >>  MAINTAINERS                                   |  1 +
> >>  2 files changed, 57 insertions(+)
> >>  create mode 100644 Documentation/devicetree/bindings/usb/microchip,usb5744.yaml
> >>
> >> diff --git a/Documentation/devicetree/bindings/usb/microchip,usb5744.yaml b/Documentation/devicetree/bindings/usb/microchip,usb5744.yaml
> >> new file mode 100644
> >> index 000000000000..fe222f6db81d
> >> --- /dev/null
> >> +++ b/Documentation/devicetree/bindings/usb/microchip,usb5744.yaml
> >> @@ -0,0 +1,56 @@
> >> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> >> +%YAML 1.2
> >> +---
> >> +$id: "http://devicetree.org/schemas/usb/microchip,usb5744.yaml#"
> >> +$schema: "http://devicetree.org/meta-schemas/core.yaml#"
> >> +
> >> +title: Bindings for the Microchip USB5744 4-port Hub Controller
> >> +
> >> +description:
> >> +  Microchip’s USB5744 SmartHub™ IC is a 4 port, SuperSpeed (SS)/Hi-Speed (HS),
> >> +  low power, low pin count configurable and fully compliant with the USB 3.1
> >> +  Gen 1 specification. The USB5744 also supports Full Speed (FS) and Low Speed
> >> +  (LS) USB signaling, offering complete coverage of all defined USB operating
> >> +  speeds. The new SuperSpeed hubs operate in parallel with the USB 2.0
> >> +  controller, so 5 Gbps SuperSpeed data transfers are not affected by slower
> >> +  USB 2.0 traffic.
> >> +
> >> +maintainers:
> >> +  - Piyush Mehta <piyush.mehta@xilinx.com>
> >> +  - Michal Simek <michal.simek@xilinx.com>
> >> +
> >> +properties:
> >> +  compatible:
> >> +    const: microchip,usb5744
> >> +
> >> +  reg:
> >> +    maxItems: 1
> >> +    description: |
> >> +      Specifies the i2c slave address, it is required and should be 0x2d
> >> +      if I2C is used.
> >
> > If I2C is not used, then this should be underneath the USB host as a USB
> > device. That also implies a different compatible string. I'd suggest you
> > just say I2C is required if that's your use.
>
> We can't say that i2c is required because we have both cases. One is
> really usb hub connected over i2c which at least requires to send one
> smbus command to start operate. But it can be extended to add more
> features - limit speeds, disable ports, etc.
>
> And the second is really the same usb hub without i2c connected which
> runs in default mode. But reset is required to ensure proper reset
> sequence.
> Hub also have external clock chip which is not handled now because it is
> just crystal on the board but if you want I can also model it via fixed
> clock and call clock enable for it.
>
> It is the same use case as is with
> Documentation/devicetree/bindings/usb/usb3503.txt

Yes, there are examples of how we don't want to do it.

> Can you please elaborate why different compatible string should be used?
> It is still the same device and not quite sure why different compatible
> string should be used.
>
> Do you also want to example where this node is the part of usb node?

See usb/usb-device.txt. And there is this[1] under review.

For these cases with I2C, I'd really rather see the hub always under
the USB bus with a link to the I2C bus when connected.

Rob

[1] https://lore.kernel.org/linux-devicetree/20210210091015.v5.1.I248292623d3d0f6a4f0c5bc58478ca3c0062b49a@changeid/