Re: [PATCH net v2] phy: mscc: Fix PTP for vsc8574 and VSC8572

[Vladimir Oltean <vladimir.oltean@nxp.com>]

On Fri, Sep 26, 2025 at 09:11:11AM +0200, Horatiu Vultur wrote:
> I have been asking around about these revisions of the PHYs and what is
> available:
> vsc856x - only rev B exists
> vsc8575 - only rev B exists
> vsc8582 - only rev B exists
> vsc8584 - only rev B exists
> vsc8574 - rev A,B,C,D,E exists
> vsc8572 - rev A,B,C,D,E exists
> 
> For vsc856x, vsc8575, vsc8582, vsc8584 the lower 4 bits in register 3
> will have a value of 1.
> For vsc8574 and vsc8572 the lower 4 bits in register 3 will have a value
> of 0 for rev A, 1 for rev B and C, 2 for D and E.
> 
> Based on this information, I think both commits a5afc1678044 and
> 75a1ccfe6c72 are correct regarding the revision check.
> 
> So, now to be able to fix the PTP for vsc8574 and vsc8572, I can do the
> following:
> - start to use PHY_ID_MATCH_MODEL for vsc856x, vsc8575, vsc8582, vsc8584
> - because of this change I will need to remove also the WARN_ON() in the
>   function vsc8584_config_init()
> - then I can drop the check for revision in vsc8584_probe()
> - then I can make vsc8574 and vsc8572 to use vsc8584_probe()
> 
> What do you think about this?

This sounds good, however I don't exactly understand how it fits in with
your response to Russell to replace phydev->phy_id with phydev->drv->phy_id
in the next revision. If the revision check in vsc8584_probe() goes away,
where will you use phydev->drv->phy_id?