On Monday 07 September 2020 16:42:28 Andrew Lunn wrote:
> On Mon, Sep 07, 2020 at 01:27:17PM +0200, Pali Rohár wrote:
> > Espressobin boards have 3 ethernet ports and some of them got assigned more
> > then one MAC address. MAC addresses are stored in U-Boot environment.
> >
> > Since commit a2c7023f7075c ("net: dsa: read mac address from DT for slave
> > device") kernel can use MAC addresses from DT for particular DSA port.
> >
> > Currently Espressobin DTS file contains alias just for ethernet0.
> >
> > This patch defines additional ethernet aliases in Espressobin DTS files, so
> > bootloader can fill correct MAC address for DSA switch ports if more MAC
> > addresses were specified.
> >
> > DT alias ethernet1 is used for wan port, DT aliases ethernet2 and ethernet3
> > are used for lan ports for both Espressobin revisions (V5 and V7).
> >
> > Fixes: 5253cb8c00a6f ("arm64: dts: marvell: espressobin: add ethernet alias")
> > Signed-off-by: Pali Rohár <pali@xxxxxxxxxx>
>
> I'm not sure a Fixes: is appropriate here. What is actually broken?
> This just seems like a new feature.
Hello Andrew! With "fixes" I mean that this patch fixes aliases list in
that mentioned commit as it was incomplete. Probably better would be
"related" or "extended" keyword in this case. But I do not know.
I would not say it is a "new feature". But rather that patch in this
email fixes issue that Linux kernel did not set correct MAC address for
DSA slave ports. I think it is something which could be backported also
to stable releases as "ignoring" vendor/factory MAC address is not
correct behavior.
If you have an idea how to reformulate commit description I will do it.
