Re: [PATCH 2/2] dt-bindings: usb: ti,usb8041: Add binding for TI USB8044 hub controller

[Conor Dooley <conor@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 2067 bytes --]

On Fri, May 09, 2025 at 07:56:35AM +0200, Mike Looijmans wrote:
> On 08-05-2025 17:53, Conor Dooley wrote:
> > On Thu, May 08, 2025 at 05:19:03PM +0200, Mike Looijmans wrote:
> > > On 08-05-2025 16:58, Conor Dooley wrote:
> > > > On Wed, May 07, 2025 at 03:11:43PM +0200, Mike Looijmans wrote:
> > > > > The TI USB8044 is similar to the USB8041.
> > > > Similar how? Why's a fallback not suitable?
> > > I don't quite understand what is meant by "fallback" here?
> > A fallback compatible, since you;re using the same match data as the
> > 8041.
> 
> I think it would work. It would look strange though, having to put an
> additional vid/pid in the devicetree to make it work.

That's how a huge number of other devices work in devicetree when
handling differs between devices.

> > > It's similar in that the USB8044 provides the same functionality and can use
> > > the same driver as the USB8041, all that is needed is to add the PID/VID
> > > values.
> > Is this onboard_dev_id_table table with the vid/pid used in combination
> > with dt, or in-place of dt when device detection is dynamic? If the
> > latter, why can't dt use a fallback compatible since the handling is
> > identical to the 8041?
> 
> My basic understanding is:
> 
> The devicetree match creates a platform device that controls the reset pin
> of the hub. It's basic task is to de-assert the reset, so the hub starts
> negotiating. This part also works with the 8041 devicetree entry (which is
> how I first tried to get it up and running).
> 
> The VID/PID table then matches the hub on the USB bus, which can then be
> associated with its platform device. Since the 8044 reports a different
> VID/PID, this part only worked when I added the entries to the tables.

Right, so you do actually need to use the dt entry /and/ the dynamic
data? In that case, there's little value in a fallback, since you need
non-fallback information in the driver for things to work. If that's a
correct understanding,
Acked-by: Conor Dooley <conor.dooley@microchip.com>

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