On Mon, 2020-10-19 at 23:00 +0200, Andrew Lunn wrote:
> > +static int m88e1111_finisar_config_init(struct phy_device *phydev)
> > +{
> > + int err;
> > + int extsr = phy_read(phydev, MII_M1111_PHY_EXT_SR);
> > +
> > + if (extsr < 0)
> > + return extsr;
> > +
> > + /* If using 1000BaseX and 1000BaseX auto-negotiation is
> > disabled, enable it */
> > + if (phydev->interface == PHY_INTERFACE_MODE_1000BASEX &&
> > + (extsr & MII_M1111_HWCFG_MODE_MASK) ==
> > + MII_M1111_HWCFG_MODE_COPPER_1000BX_NOAN) {
> > + err = phy_modify(phydev, MII_M1111_PHY_EXT_SR,
> > + MII_M1111_HWCFG_MODE_MASK |
> > + MII_M1111_HWCFG_SERIAL_AN_BYPASS,
> > + MII_M1111_HWCFG_MODE_COPPER_1000BX_AN
> > |
> > + MII_M1111_HWCFG_SERIAL_AN_BYPASS);
> > + if (err < 0)
> > + return err;
> > + }
> > +
> > + return m88e1111_config_init(phydev);
> > +}
>
> Hi Robert
>
> Is this really specific to the Finisar? It seems like any application
> of the m88e1111 in 1000BaseX would benefit from this?
I suppose that part would be pretty harmless, as you would likely want
that behavior whenever that if condition was triggered. So
m88e1111_finisar_config_init could likely be merged into
m88e1111_config_init.
Mainly what stopped me from making all of these changes generic to all
88E1111 is that I'm a bit less confident in the different config_aneg
behavior, it might be more specific to this SFP copper module case?
--
Robert Hancock
Senior Hardware Designer, Advanced Technologies
www.calian.com
