[PATCH v3 1/6] net: introduce DSA class for Ethernet switches

[Vladimir Oltean]

On Tue, 17 Dec 2019 at 12:36, Alexandru Marginean
<alexm.osslist at gmail.com> wrote:
> It wouldn't go through mdio_phy_connect, dm_eth_phy_connect would need
> to figure out if it's a fixed link or PHY and then call the proper
> function, like in your snippet.
> The MDIO NULL topic is a bit more complicated I think.  Some of the Eth
> drivers just deal with fixed link internally and don't create a phy
> device at all.  If there is a PHY device I think we need to have a MDIO
> bus for it, and we shouldn't piggy-back on real MDIO buses.  I'm
> thinking we could have a dummy MDIO bus created automatically by
> ETH/MDIO uclass code for fixed link PHY devices.

Well, either one or the other, really. The dummy MDIO bus is not
really needed, but on the other hand, if that's what it takes to make
DM_MDIO support fixed-link with the least amount of special cases,
then ok, fine with me.

>
> Thoughts?
> Alex
>

Regards,
-Vladimir