Re: [PATCH net-next] net: phylink: require supported_interfaces to be filled

[Andrew Lunn <andrew@lunn.ch>]

> The 6350 looks to be similar to the 6352 in many respects, though it lacks
> a SERDES interface, but it otherwise mostly seems compatible.

Not having the SERDES is important. Without that SERDES, the bit about
Port 4 in mv88e6352_phylink_get_caps() is
incorrect. mv88e61852_phylink_get_caps() looks reasonable for this
hardware.

> Using the 6352
> phylink_get_caps function instead of the 6185 one fixes this:
> 
> --- a/drivers/net/dsa/mv88e6xxx/chip.c
> +++ b/drivers/net/dsa/mv88e6xxx/chip.c
> @@ -5418,7 +5418,7 @@ static const struct mv88e6xxx_ops mv88e6350_ops = {
>         .set_max_frame_size = mv88e6185_g1_set_max_frame_size,
>         .stu_getnext = mv88e6352_g1_stu_getnext,
>         .stu_loadpurge = mv88e6352_g1_stu_loadpurge,
> -       .phylink_get_caps = mv88e6185_phylink_get_caps,
> +       .phylink_get_caps = mv88e6352_phylink_get_caps,
>  };
> 
>  static const struct mv88e6xxx_ops mv88e6351_ops = {
> 
> 
> The story doesn't quite end here though. With this fix in place support
> for the 6350 is then again broken by commit b92143d4420f ("net: dsa:
> mv88e6xxx: add infrastructure for phylink_pcs"). This results in a dump
> on boot up:

PCS is approximately another name of a SERDES. Since there is no
SERDES, you don't don't want any of the pcs ops filled in.

Russell knows this code much better than i do. Let see what he says.

	Andrew