An external PHY needs settling time after power up or reser.
In the bind() function an mdio bus is registered. If at this point
the external PHY is still initialising, no valid PHY ID will be
read and on phy_find_first() the bind() function will fail.

If an external PHY is present, wait the maximum time specified
in 802.3 45.2.7.1.1.

Fixes: 05b35e7eb9a1 ("smsc95xx: add phylib support")
Signed-off-by: Alexandru Tachici <alexandru.tachici@analog.com>
---
 drivers/net/usb/smsc95xx.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/net/usb/smsc95xx.c b/drivers/net/usb/smsc95xx.c
index bfb58c91db04..5ed001c0cd56 100644
--- a/drivers/net/usb/smsc95xx.c
+++ b/drivers/net/usb/smsc95xx.c
@@ -1134,8 +1134,15 @@ static int smsc95xx_bind(struct usbnet *dev, struct usb_interface *intf)
 		goto free_mdio;
 
 	is_internal_phy = !(val & HW_CFG_PSEL_);
-	if (is_internal_phy)
+	if (is_internal_phy) {
 		pdata->mdiobus->phy_mask = ~(1u << SMSC95XX_INTERNAL_PHY_ID);
+	} else {
+		/* Driver has no knowledge at this point about the external PHY.
+		 * The 802.3 specifies that the reset process shall
+		 * be completed within 0.5 s.
+		 */
+		fsleep(500000);
+	}
 
 	pdata->mdiobus->priv = dev;
 	pdata->mdiobus->read = smsc95xx_mdiobus_read;
-- 
2.34.1

Hello!

On 11/14/22 4:16 PM, Alexandru Tachici wrote:

> An external PHY needs settling time after power up or reser.

   Reset? :-)

> In the bind() function an mdio bus is registered. If at this point
> the external PHY is still initialising, no valid PHY ID will be
> read and on phy_find_first() the bind() function will fail.
> 
> If an external PHY is present, wait the maximum time specified
> in 802.3 45.2.7.1.1.
> 
> Fixes: 05b35e7eb9a1 ("smsc95xx: add phylib support")
> Signed-off-by: Alexandru Tachici <alexandru.tachici@analog.com>

[...]

MBR, Sergey

On Mon, Nov 14, 2022 at 03:16:43PM +0200, Alexandru Tachici wrote:
> +	} else {
> +		/* Driver has no knowledge at this point about the external PHY.
> +		 * The 802.3 specifies that the reset process shall
> +		 * be completed within 0.5 s.
> +		 */
> +		fsleep(500000);
> +	}

It's slightly more code, but would it be better to do this in the
mdiobus reset() method?

Thanks.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!