On Wed, 2020-08-05 at 10:49 +0200, Frank Wunderlich wrote:
> > Gesendet: Mittwoch, 05. August 2020 um 10:36 Uhr
> > Von: "David Woodhouse" <dwmw2@infradead.org>
> > > mt7623.dtsi => mt7623n.dtsi => mt7623n-bananapi-bpi-r2.dts
> > > mt7623.dtsi => mt7623a.dtsi => mt7623a-unielec-u7623.dts (not
> > > existing yet,
> > > openwrt seems to use a board-specific dtsi)
> > 
> > Yes, I think we should.
> 
> i want to see what MTK/DT owner says to this...
> my current way will be still adding the nodes to existing mt7623.dtsi
> (like ryder lee did it in original patch)
> but disabling them to not break mt7623a and splitting it afterwards.
> 
> > I'll create mt7623a.dtsi and upstream the U7623 support; I think that
> > can happen without conflicting with anything you do.
> > 
> > I note that the GPU node has been added to mt7623.dtsi in 5.8 too;
> > that'll want to move to the new mt7623n.dtsi that you create, along
> > with your other new additions.
> 
> i guess mali-node also needs to be moved to mt7623n.dtsi, so my
> current way seems right...
> but it's decision of MTK/DT owner. if they make a note i squash the
> disabling-commit into this and post v5

Yes, the mali node needs moving too. I've pushed an untested series to
http://git.infradead.org/users/dwmw2/linux.git/shortlog/refs/heads/mt7623
which does that and adds the UniElec board.