[PATCH v2] ath79: add support for Ubiquiti PowerBeam M (XW)

[Russell Senior]

On Tue, May 25, 2021 at 2:33 AM Russell Senior
<russell at personaltelco.net> wrote:
>
> On Tue, May 25, 2021 at 1:40 AM Rafał Miłecki <zajec5 at gmail.com> wrote:
> >
> > On 23.05.2021 13:59, Russell Senior wrote:
> > > +&eth0 {
> > > +     status = "okay";
> > > +
> > > +     /* default for ar934x, except for 1000M and 10M */
> > > +     pll-data = <0x02000000 0x00000101 0x00001313>;
> > > +
> > > +     mtd-mac-address = <&art 0x0>;
> >
> > I'd love to have new DTS use upstream "nvmem-cells" property for that basic case.
> >
> > Example:
> > arm64: dts: broadcom: bcm4908: add Ethernet MAC addr
> > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9f01f5cdb548352418b34ce77db02a560fe2913b
>
> Not sure this qualifies as a new DTS, it is derived almost entirely
> from other ubiquiti ar9342 hardware's DTS.  I think such a conversion
> would make more sense to do wholesale, later on. I'm not particularly
> prepared to take that on at the moment.

Looking closer, I realize that the mtd-mac-address line isn't even
needed, since it will inherit it from the included
ar9342_ubnt_xw.dtsi.

I took a swing at replacing mtd-mac-address with nvmem-cells as in
your example, but I got an incomprehensible cascade of errors from
unrelated ath79 devices dts, like this:

  "Missing #address-cells in interrupt provider"

which made no sense to me and I gave up.

>
> $ git grep mtd-mac-address | wc -l
> 1119
>
> $ git grep -l mtd-mac-address | wc -l
> 568
>
> >
> >
> > > +
> > > +     phy-mode = "rgmii-id";
> > > +     phy-handle = <&phy4>;
> > > +
> > > +     gmac-config {
> > > +             device = <&gmac>;
> > > +             rxd-delay = <3>;
> > > +             rxdv-delay = <3>;
> > > +     };
> > > +};