Hi,
On Wed, Aug 15, 2018 at 3:35 AM Andrew Lunn <andrew@xxxxxxx> wrote:
>
> On Tue, Aug 14, 2018 at 05:58:12PM +0200, Uwe Kleine-König wrote:
> > Hello Ahmad,
> >
> >
> > On Tue, Aug 14, 2018 at 04:12:40PM +0200, Ahmad Fatoum wrote:
> > > The referenced commit broke initializing macb on the EVB-KSZ9477 eval board.
> > > There, of_mdiobus_register was called even for the fixed-link representing
> > > the SPI-connected switch PHY, with the result that the driver attempts to
> > > enumerate PHYs on a non-existent MDIO bus:
> > >
I ran into a similar problem on v14.4 for davinci_mdio I had to patch
it with [1].
The cpsw has 2 phys one phy is connected to KSZ9031 and other to
ksz9897 Ethernet switch
which is treated as a fixed phy with no mdio lines because of which
mdio_read/write failed.
This didn’t happen in v4.9.x something in core has changed ?
[1]
diff --git a/drivers/net/ethernet/ti/davinci_mdio.c
b/drivers/net/ethernet/ti/davinci_mdio.c
index 3e84107..197baa6 100644
--- a/drivers/net/ethernet/ti/davinci_mdio.c
+++ b/drivers/net/ethernet/ti/davinci_mdio.c
@@ -245,6 +245,13 @@ static int davinci_mdio_read(struct mii_bus *bus,
int phy_id, int phy_reg)
u32 reg;
int ret;
+ if (phy_id == 2)
+ return 0;
+
if (phy_reg & ~PHY_REG_MASK || phy_id & ~PHY_ID_MASK)
return -EINVAL;
@@ -289,6 +296,13 @@ static int davinci_mdio_write(struct mii_bus
*bus, int phy_id,
u32 reg;
int ret;
+ if (phy_id == 2)
+ return 0;
+
if (phy_reg & ~PHY_REG_MASK || phy_id & ~PHY_ID_MASK)
return -EINVAL;
Cheers,
--Prabhakar Lad
