Re: [PATCH net-next 3/4] drivers/net/phy: Add driver for the onsemi NCN26000 10BASE-T1S PHY

[Piergiorgio Beruto]

On Sun, Dec 04, 2022 at 07:58:07PM +0100, Andrew Lunn wrote:
> > > > +static int ncn26000_enable(struct phy_device *phydev)
> > > > +{
> > > 
> > > This is actually the config_aneg() implementation, it should be named
> > > as such.
> > I can certainly rename it, however I did this for a reason. The NCN26000
> > only supports P2MP mode. Therefore, it does not support AN (this is
> > clearly indicated in the IEEE specifications as well).
> > 
> > However, it is my understanding that the config_aneg() callback is
> > invoked also for PHYs that do not support AN, and this is actually the
> > only way to set a link_control bit to have the PHY enable the PMA/PCS
> > functions. So I thought to call this function "enable" to make it clear
> > we're not really implementing autoneg, but link_control.
> 
> Anybody familiar with PHY drivers knows the name is not ideal, but
> when they see config_aneg() they have a good idea what it does without
> having to look at the code. All PHY drivers should have the same basic
> structure, naming etc, just to make knowledge transfer between drivers
> easy, maintenance easy, etc.
Fair enough, I'll change the name in the next patchset version