Re: [PATCH net] net: dsa: mv88e6xx: fix supported_interfaces setup in mv88e6250_phylink_get_caps()

[Matthias Schiffer]

On Tue, 2024-04-16 at 17:56 +0100, Russell King (Oracle) wrote:
> On Tue, Apr 16, 2024 at 05:50:54PM +0200, Matthias Schiffer wrote:
> > +int mv88e6250_port_get_mode(struct mv88e6xxx_chip *chip, int port,
> > +			    phy_interface_t *mode)
> > +{
> > +	int err;
> > +	u16 reg;
> > +
> > +	if (port < 5) {
> > +		*mode = PHY_INTERFACE_MODE_INTERNAL;
> > +		return 0;
> > +	}
> 
> Note that if mv88e6xxx_phy_is_internal() returns TRUE for the port,
> then this will be handled automatically.
> 
> > +
> > +	err = mv88e6xxx_port_read(chip, port, MV88E6XXX_PORT_STS, &reg);
> > +	if (err)
> > +		return err;
> > +
> > +	switch (reg & MV88E6250_PORT_STS_PORTMODE_MASK) {
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_10_HALF_PHY:
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_100_HALF_PHY:
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_10_FULL_PHY:
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_100_FULL_PHY:
> > +		*mode = PHY_INTERFACE_MODE_REVMII;
> > +		break;
> > +
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_HALF:
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_FULL:
> > +		*mode = PHY_INTERFACE_MODE_MII;
> > +		break;
> > +
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_DUAL_100_RMII_FULL_PHY:
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_200_RMII_FULL_PHY:
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_10_100_RMII_HALF_PHY:
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_10_100_RMII_FULL_PHY:
> > +		*mode = PHY_INTERFACE_MODE_REVRMII;
> > +		break;
> > +
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_DUAL_100_RMII_FULL:
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_10_100_RMII_FULL:
> > +		*mode = PHY_INTERFACE_MODE_RMII;
> > +		break;
> > +
> > +	case MV88E6250_PORT_STS_PORTMODE_MII_100_RGMII:
> > +		*mode = PHY_INTERFACE_MODE_RGMII;
> > +		break;
> > +
> > +	default:
> > +		*mode = PHY_INTERFACE_MODE_NA;
> 
> What does this mean? I don't allow PHY_INTERFACE_MODE_NA to be set in
> the list of supported interfaces because it isn't an interface mode.
> If it's invalid, then it's probably best to return an error.
> 
> I wonder whether it would just be better to pass the
> supported_interfaces bitmap into this function and have it set the
> appropriate bit itself, renaming the function to something more
> better suited to that purpose.
> 
> Thanks.

I'm explicitly checking for PHY_INTERFACE_MODE_NA in the caller to handle the "this interface isn't
useable" case. Passing supported_interfaces into the function handling the port modes is fine with
me, too - will send a v2 later.

Best regards,
Matthias


> 

-- 
TQ-Systems GmbH | Mühlstraße 2, Gut Delling | 82229 Seefeld, Germany
Amtsgericht München, HRB 105018
Geschäftsführer: Detlef Schneider, Rüdiger Stahl, Stefan Schneider
https://www.tq-group.com/