Re: [PATCH v4 1/2] dt-bindings: usb: Add the binding example for the Genesys Logic GL3523 hub

[Conor Dooley <conor@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 5216 bytes --]

On Fri, Nov 24, 2023 at 04:18:23PM +0530, Anand Moon wrote:
> Hi Conor
> 
> On Thu, 23 Nov 2023 at 23:26, Conor Dooley <conor@kernel.org> wrote:
> >
> > On Wed, Nov 22, 2023 at 11:53:46PM +0530, Anand Moon wrote:
> > > Add the binding example for the USB3.1 Genesys Logic GL3523
> > > integrates with USB 3.1 Gen 1 Super Speed and USB 2.0 High-Speed
> > > hub.
> > >
> > > Onboard USB hub supports USB 3.x and USB 2.0 peer controllers.
> > > which has a common reset pin and power supply.
> > > peer-hub phandle each peer controller with proper gpio reset
> > > and help each peer power on during initialization
> > > and power off during suspend.
> > >
> > > Signed-off-by: Anand Moon <linux.amoon@gmail.com>
> > > ---
> > > v4: Fix the description of peer-hub and update the commit message.
> > > Schematics of the Odroid N2+
> > > https://dn.odroid.com/S922X/ODROID-N2/Schematic/odroid-n2_rev0.6_20210121.pdf
> > > V3: fix the dt_binding_check error, added new example for Genesys GL3523
> > > v2: added Genesys GL3523 binding
> > > v1: none
> > > ---
> > >  .../bindings/usb/genesys,gl850g.yaml          | 67 +++++++++++++++++--
> > >  1 file changed, 63 insertions(+), 4 deletions(-)
> > >
> > > diff --git a/Documentation/devicetree/bindings/usb/genesys,gl850g.yaml b/Documentation/devicetree/bindings/usb/genesys,gl850g.yaml
> > > index ee08b9c3721f..bc3b3f4c8473 100644
> > > --- a/Documentation/devicetree/bindings/usb/genesys,gl850g.yaml
> > > +++ b/Documentation/devicetree/bindings/usb/genesys,gl850g.yaml
> > > @@ -9,9 +9,6 @@ title: Genesys Logic USB hub controller
> > >  maintainers:
> > >    - Icenowy Zheng <uwu@icenowy.me>
> > >
> > > -allOf:
> > > -  - $ref: usb-device.yaml#
> > > -
> > >  properties:
> > >    compatible:
> > >      enum:
> > > @@ -27,12 +24,48 @@ properties:
> > >
> > >    vdd-supply:
> > >      description:
> > > -      the regulator that provides 3.3V core power to the hub.
> > > +      phandle to the regulator that provides power to the hub.
> > > +
> > > +  peer-hub:
> >
> > Should the property not be "peer-controller"? Your description refers to
> > them as such.
> 
> No, as per my understanding, peer-hub represents a complete USB hub.
> See the lock diagram in the below link.
> 
> >
> > > +    $ref: /schemas/types.yaml#/definitions/phandle
> > > +    description:
> > > +      onboard USB hub supports USB 3.x and USB 2.0 peer controllers.
> >
> >
> > > +      which has a common reset pin and power supply.
> > > +      peer-hub phandle each peer controller with proper gpio reset

This is what I don't get. You say "peer-hub phandle each peer
controller..". It is hard for me to understand that portion of the
sentence, but the interchanging of "hub" and "controller" is
confusing. The title of the binding says "hub controller", so maybe it
is better to use that here.

> > > +      and help each peer power on during initialization
> > > +      and power off during suspend.
> >
> > I generally hate to talk about non-native speakers grammar etc, but what
> > you have here is in need of a lot of improvement. The below is my
> > attempt to understand what you are trying to say:
> >
> > "For onboard hubs that support USB 3.x and USB 2.0 controllers with
> > shared resets and power supplies, this property is used to identify
> > the controllers with which these are shared."

"For onboard hub controllers that support USB 3.x and USB 2.0 hubs
with shared resets and power supplies, this property is used to identify
the hubs with which these are shared."

I re-worded this again to try and remove the use of "controller".
Do you think that this still makes sense?

> Sorry for the poor grammar, I will update this in the next v5.
> 
> > Also - this is one particular system, what prevents there being a hub
> > that has more than 2 controllers? Also, as you insist that this is
> > generic, and not just for genesys, should this not be defined in a
> > common location?
> 
> Here is the block diagram of the Genesys GL3523 hub.
> [0] https://www.genesyslogic.com.tw/en/product_view.php?show=67 [Block Diagram]
> 
> It has two USB 2.0 and USB 3.1 controllers, so using peer-hub node
> the onboard hub module will bring up this hub.
> 
> There are many examples that use similar properties hence it is generic.
> 
> # Documentation/devicetree/bindings/usb/cypress,hx3.yaml
> # Documentation/devicetree/bindings/usb/microchip,usb5744.yaml
> # Documentation/devicetree/bindings/usb/realtek,rts5411.yaml
> # Documentation/devicetree/bindings/usb/ti,usb8041.yaml
> # Documentation/devicetree/bindings/usb/vialab,vl817.yaml

Which brings me back to the unanswered question, should this not be
defined in a common location given there are several devices using it?
I assume because it only applies to hub controllers and not other types
of devices.

Also, the descriptions that I saw when looking at some of those other
bindings are similarly poor. I can't bring myself to care any more,
just clean up the ambiguous wording here and I'll ack the next version,
I don't expect you to sort out the wording in other bindings.

Cheers,
Conor.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]