Re: [PATCH v2] net: mtk_sgmii: implement mtk_pcs_ops

[Russell King (Oracle)]

On Mon, Jan 16, 2023 at 03:45:24PM +0100, Bjørn Mork wrote:
> "Russell King (Oracle)" <linux@xxxxxxxxxxxxxxx> writes:
> > On Mon, Jan 16, 2023 at 02:08:30PM +0100, Bjørn Mork wrote:
> >> Frank Wunderlich <frank-w@xxxxxxxxxxxxxxx> writes:
> >> 
> >> > apart from this little problem it works much better than it actually is so imho more
> >> > people should test it on different platforms.
> >> 
> >> Hello!  I've been banging my head against an MT7986 board with two
> >> Maxlinear GPY211C phys for a while. One of those phys is connected to
> >> port 5 of the MT7531 switch.  This is working perfectly.
> >> 
> >> The other GPY211C is connected to the second MT7986 mac.  This one is
> >> giving me a headache...
> >> 
> >> I can only get the port to work at 2500Mb/s.  Changing the speed to
> >> anything lower looks fine in ethtool etc, but traffic is blocked.
> >
> > My guess would be that the GPY PHY is using in-band SGMII negotiation
> > (it sets VSPEC1_SGMII_ANEN_ANRS when entering SGMII mode and clears
> > it in 2500base-X), but as the link is not using in-band mode on the
> > PCS side, the PHY never sees its in-band negotiation complete, so the
> > link between PCS and PHY never comes up.
> >
> > Both sides need to agree on that detail.
> 
> Any hints on how I would go about doing that?  I am a little lost here,
> changing arbitrary bits I don't understand the meaning of.

Hi,

To prove the point, in mtk_pcs_config():

        if (interface == PHY_INTERFACE_MODE_SGMII) {
                sgm_mode = SGMII_IF_MODE_SGMII;
                if (phylink_autoneg_inband(mode)) {

Force the second if() to always be true, and see whether that allows
traffic to pass.

Thanks.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!