Re: [PATCH net-next v2 7/9] net: phy: introduce ethtool_phy_ops to get and set phy configuration

From: Russell King (Oracle)
Date: Fri Oct 04 2024 - 15:59:48 EST


On Fri, Oct 04, 2024 at 08:42:42PM +0200, Andrew Lunn wrote:
> > +int phy_set_config(struct phy_device *phydev,
> > + const struct phy_device_config *phy_cfg,
> > + struct netlink_ext_ack *extack)
> > +{
> > + bool isolate_change;
> > + int ret;
> > +
> > + mutex_lock(&phydev->lock);
> > + isolate_change = (phy_cfg->isolate != phydev->isolated);
> > + mutex_unlock(&phydev->lock);
> > +
> > + if (!isolate_change)
> > + return 0;
> > +
> > + ret = phy_isolate(phydev, phy_cfg->isolate);
> > + if (ret)
> > + NL_SET_ERR_MSG(extack, "Error while configuring PHY isolation");
>
> This seems overly simplistic to me. Don't you need to iterate over all
> the other PHYs attached to this MAC and ensure they are isolated? Only
> one can be unisolated at once.
>
> It is also not clear to me how this is going to work from a MAC
> perspective. Does the MAC call phy_connect() multiple times? How does
> ndev->phydev work? Who is responsible for the initial configuration,
> such that all but one PHY is isolated?
>
> I assume you have a real board that needs this. So i think we need to
> see a bit more of the complete solution, including the MAC changes and
> the device tree for the board, so we can see the big picture.

Also what the ethernet driver looks like too!

One way around the ndev->phydev problem, assuming that we decide that
isolate is a good idea, would be to isolate the current PHY, disconnect
it from the net_device, connect the new PHY, and then clear the isolate
on the new PHY. Essentially, ndev->phydev becomes the currently-active
PHY.

However, I still want to hear whether multiple PHYs can be on the same
MII bus from a functional electrical perspective.

I know that on the Macchiatobin, where the 10G serdes signals go to the
88X3310 on doubleshot boards and to the SFP cage on singleshot boards,
this is controlled by the placement of zero ohm resistors to route the
serdes signals to the appropriate device, thus minimising the unused
stub lengths.

--
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!