On Tue, Feb 15, 2022 at 09:56:50PM +0100, Andrew Lunn wrote:
> > > > - ethernet: usbether@2 {
> > > > - compatible = "usb0424,9730";
> > > > + ethernet: ethernet@2 {
> > > > + compatible = "usb424,9730";
> > >
> > > The change of the compatible is not mentioned in the patch description.
> > > Is this intentional?
> >
> > No, I forgot to mentione it. According to the USB schema 0 should be
> > removed. So, this compatible was incorrect as well. With leading zero
> > present yaml schema was not able to detect and validate this node.
>
> Does the current code not actually care about a leading 0? It will
> match with or without it? It would be good to mention that as well in
> the commit message, otherwise somebody like me is going to ask if this
> breaks backwards compatibility, since normally compatible is an exact
> string match.
Current kernel code do not care about exact this compatibles. There is
no driver matching against it. The USB Ethernet driver will take the
node provided by the USB core drivers without validating the compatible
against USB ID.
See:
drivers/usb/core/of.c
drivers/usb/core/message.c:2093
On other hand, DT validations tools do care about it and this nodes was
not detected automatically. I found it accidentally by grepping the
sources.
> And i actually think this is the sort of change which should be as a
> patch of its own. If this causes a regression, a git bisect would then
> tell you if it is the change of usbether -> ethernet, or 0424 to
> 424. That is part of why we ask for lots of small changes.
Sounds good, I'll update it.
Regards,
Oleksij
--
Pengutronix e.K. | |
Steuerwalder Str. 21 | http://www.pengutronix.de/ |
31137 Hildesheim, Germany | Phone: +49-5121-206917-0 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |
