[PATCH 3/4] net: mdio: teach dm_eth_phy_connect to connect to fixed PHY

[Bin Meng]

Hi Vladimir,

On Wed, Feb 17, 2021 at 6:48 AM Vladimir Oltean <olteanv at gmail.com> wrote:
>
> From: Vladimir Oltean <vladimir.oltean at nxp.com>
>
> It would be desirable for top-level callers of PHYLIB to deal with as
> little complexity as possible, and when they call dm_eth_phy_connect,
> they get a struct phy_device that "just works".
>
> There is a phy_connect_fixed interception put in phy_connect, however
> dm_eth_phy_connect will not reach there: if will search for a phy-handle
> all by itself, and error out if there isn't one. So we can make callers
> of dm_eth_phy_connect suffer by having them call:
>
>         err = dm_eth_phy_connect();
>         if (err)
>                 err = dm_eth_phy_connect_fixed();
>
> or we can just add the logic in dm_eth_phy_connect() that searches for a
> fixed-link before searching for a phy-handle.

This logic already exists since:

commit bdf319273b4a752664c089fbffee5bb2024c8586
Author: Rasmus Villemoes <rasmus.villemoes at prevas.dk>
Date:   Mon Oct 5 15:15:16 2020 +0200

    mdio-uclass.c: support fixed-link subnodes

> In fact we already have an in-tree driver that can make use of this
> refactoring: the Freescale TSEC driver.

I will drop the dm_eth_phy_connect() changes and only keep the TSEC
changes in v2.

>
> Signed-off-by: Vladimir Oltean <vladimir.oltean at nxp.com>
> ---
>  drivers/net/tsec.c |  6 +-----
>  net/mdio-uclass.c  | 12 ++++++++++--
>  2 files changed, 11 insertions(+), 7 deletions(-)
>

Regards,
Bin