Re: [PATCH v2] net: usb: lan78xx: scan all MDIO addresses on LAN7801

[Martin Pålsson]

Hi Maxime,

Thanks for pointing that out. Ran a quick test with the assignment 
removed and the module behaves as desired. A third version of the patch 
coming in a few moments.

Best regards

Martin

On 2/16/26 11:34 AM, Maxime Chevallier wrote:
> Hi Martin,
>
> On 16/02/2026 06:09, Martin Pålsson wrote:
> > The LAN7801 is designed exclusively for external PHYs (unlike the
> > LAN7800/LAN7850 which have internal PHYs), but lan78xx_mdio_init()
> > restricts PHY scanning to MDIO addresses 0-7 by setting phy_mask to
> > ~(0xFF). This prevents discovery of external PHYs wired to addresses
> > outside that range.
> >
> > One such case is the DP83TC814 100BASE-T1 PHY, which is typically
> > configured at MDIO address 10 via PHYAD bootstrap pins and goes
> > undetected with the current mask.
> >
> > Set phy_mask to 0 for the LAN7801 so that all 32 MDIO addresses are
> > scanned during bus registration, allowing any external PHY to be
> > discovered regardless of its address.
> >
> > Fixes: 02dc1f3d613d ("lan78xx: add LAN7801 MAC only support")
> >
> > Signed-off-by: Martin Pålsson <martin@xxxxxxxxxxxx>
> > ---
> > Changes in v2:
> > - Added Fixes: tag
> > - Link to v1: https://lore.kernel.org/r/20260214-lan78xx-phy-mask-fix-v1-1-3baa100861e4@xxxxxxxxxxxx
> > ---
> >   drivers/net/usb/lan78xx.c | 4 ++--
> >   1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/net/usb/lan78xx.c b/drivers/net/usb/lan78xx.c
> > index 00397a807393..9d8f1ff7028d 100644
> > --- a/drivers/net/usb/lan78xx.c
> > +++ b/drivers/net/usb/lan78xx.c
> > @@ -2094,8 +2094,8 @@ static int lan78xx_mdio_init(struct lan78xx_net *dev)
> >   		dev->mdiobus->phy_mask = ~(1 << 1);
> >   		break;
> >   	case ID_REV_CHIP_ID_7801_:
> > -		/* scan thru PHYAD[2..0] */
> > -		dev->mdiobus->phy_mask = ~(0xFF);
> > +		/* scan all 32 MDIO addresses for external PHYs */
> > +		dev->mdiobus->phy_mask = 0;
> As this is the default behaviour for mdio bus controllers, and the
> default value for the mask is 0, I think you can simply discard that
> assignment.
>
> I don't even think we need a comment, as the other cases for 7800/7850
> already have a comment explaining why they are special cases.
>
> Maxime
--
Best regards / Bästa hälsningar / Mit freundlichen Grüßen

Martin Pålsson
Poleshift AB - Local Automotive Ethernet Solutions
+46 706 26 49 90
martin@xxxxxxxxxxxx