Re: mv88e6240 configuration broken for B850v3

[Vladimir Oltean <olteanv@gmail.com>]

On Mon, Dec 06, 2021 at 07:24:56PM +0000, Martyn Welch wrote:
> On Mon, 2021-12-06 at 20:50 +0200, Vladimir Oltean wrote:
> > On Mon, Dec 06, 2021 at 06:37:44PM +0000, Martyn Welch wrote:
> > > On Mon, 2021-12-06 at 20:31 +0200, Vladimir Oltean wrote:
> > > > On Mon, Dec 06, 2021 at 06:26:25PM +0000, Martyn Welch wrote:
> > > > > On Mon, 2021-12-06 at 17:44 +0000, Martyn Welch wrote:
> > > > > > On Fri, 2021-12-03 at 17:25 +0100, Andrew Lunn wrote:
> > > > > > > > Hi Andrew,
> > > > > > > 
> > > > > > > Adding Russell to Cc:
> > > > > > > 
> > > > > > > > I'm currently in the process of updating the GE B850v3
> > > > > > > > [1] to
> > > > > > > > run
> > > > > > > > a
> > > > > > > > newer kernel than the one it's currently running. 
> > > > > > > 
> > > > > > > Which kernel exactly. We like bug reports against net-next,
> > > > > > > or
> > > > > > > at
> > > > > > > least the last -rc.
> > > > > > > 
> > > > > > 
> > > > > > I tested using v5.15-rc3 and that was also affected.
> > > > > > 
> > > > > 
> > > > > I've just tested v5.16-rc4 (sorry - just realised I previously
> > > > > wrote
> > > > > v5.15-rc3, it was v5.16-rc3...) and that was exactly the same.
> > > > 
> > > > Just to clarify: you're saying that you're on v5.16-rc4 and that
> > > > if
> > > > you
> > > > revert commit 3be98b2d5fbc ("net: dsa: Down cpu/dsa ports phylink
> > > > will
> > > > control"), the link works again?
> > > > 
> > > 
> > > Correct
> > > 
> > > > It is a bit strange that the external ports negotiate at
> > > > 10Mbps/Full,
> > > > is that the link speed you intend the ports to work at?
> > > 
> > > Yes, that's 100% intentional due to what's connected to to those
> > > ports
> > > and the environment it works in.
> > > 
> > > Martyn
> > 
> > Just out of curiosity, can you try this change? It looks like a
> > simple
> > case of mismatched conditions inside the mv88e6xxx driver between
> > what
> > is supposed to force the link down and what is supposed to force it
> > up:
> > 
> > diff --git a/net/dsa/port.c b/net/dsa/port.c
> > index 20f183213cbc..d235270babf7 100644
> > --- a/net/dsa/port.c
> > +++ b/net/dsa/port.c
> > @@ -1221,7 +1221,7 @@ int dsa_port_link_register_of(struct dsa_port
> > *dp)
> >                 if (of_phy_is_fixed_link(dp->dn) || phy_np) {
> >                         if (ds->ops->phylink_mac_link_down)
> >                                 ds->ops->phylink_mac_link_down(ds, port,
> > -                                       MLO_AN_FIXED, PHY_INTERFACE_MODE_NA);
> > +                                       MLO_AN_PHY, PHY_INTERFACE_MODE_NA);
> >                         return dsa_port_phylink_register(dp);
> >                 }
> >                 return 0;
> 
> Yes, that appears to also make it work.
> 
> Martyn

Well, I just pointed out what the problem is, I don't know how to solve
it, honest! :)

It's clear that the code is wrong, because it's in an "if" block that
checks for "of_phy_is_fixed_link(dp->dn) || phy_np" but then it omits
the "phy_np" part of it. On the other hand we can't just go ahead and
say "if (phy_np) mode = MLO_AN_PHY; else mode = MLO_AN_FIXED;" because
MLO_AN_INBAND is also a valid option that we may be omitting. So we'd
have to duplicate part of the logic from phylink_parse_mode(), which
does not appear ideal at all. What would be ideal is if this fabricated
phylink call would not be done at all, but I don't know enough about the
systems that need it, I expect Andrew knows more.