Fix initialization failure when phy_mdio_port is 0

[torlye]

This fixes an issue I reproduced with my TRENDnet TEG-10GECTX v2.0R and the driver from branch release/linux-6.7.y-1 running on Linux 6.8.

This card has Device ID 4027, Subsystem Device ID 3015.

With my card; the for loop in bdx_mdio_scan_phy_id completes on the very first iteration, which means the resulting value of i is 0, which is then assigned to priv->phy_mdio_port.

When the port is set to 0, the if condition !priv->phy_mdio_port is true in bdx_phy_init, which results in an immediate return and failed detection of the PHY.

If I understand correctly, there may be up to 32 MDIO ports, but these are represented as values 0 to 31 in phy_mdio_port. Thus 0 is a valid port and should not result in a failure.

I changed the if condition in question to use the return value of bdx_mdio_scan_phy_id instead. This value is 0 when the for loop has completed 32 iterations without finding a PHY.

I also found that the call to priv->phy_ops.mdio_reset on line 463 was hard-coded to use a port number of 1 rather than the detected port number stored in phy_mdio_port. Adjusted the same.

The driver initialization completes successfully with these changes in my own testing.

[kalvdans]

Looks correct to me! (I have no merge rights)

[DatPat]

I discovered this issue a while ago and did the very same thing. This definitely fixed the driver on my card.

[NF117]

I tested the pull request with my network adapter. It fix the problem. Please merge to master
@torlye thank you very much for the fix