Re: [PATCH net-next v11 06/15] ax88179_178a: Add HW support for AX179A-based chips

On 9/15/26 14:07, Andrew Lunn wrote:

On Tue, Sep 15, 2026 at 02:01:48AM +0200, Birger Koblitz wrote:

quoted

On 9/14/26 18:53, Andrew Lunn wrote:

quoted

quoted

quoted

+static const struct net_device_ops ax88179a_netdev_ops = {
+	.ndo_open		= usbnet_open,
+	.ndo_stop		= usbnet_stop,
+	.ndo_start_xmit		= usbnet_start_xmit,
+	.ndo_tx_timeout		= usbnet_tx_timeout,
+	.ndo_get_stats64	= dev_get_tstats64,
+	.ndo_change_mtu		= ax88179_change_mtu,
+	.ndo_set_mac_address	= ax88179_set_mac_addr,
+	.ndo_validate_addr	= eth_validate_addr,
+	.ndo_eth_ioctl		= usbnet_mii_ioctl,

dev->mii.mdio_read is never initialized for ax88179a, so
dev->mii.mdio_read()
through usbnet_mii_ioctl() is a NULL function pointer.

Something like:

    static int ax88179a_mii_ioctl(...)
    {
    	struct ax88179_data *data = netdev2data(net);
    	return phylink_mii_ioctl(data->phylink, ifr, cmd);
    }

Is this transitory? Does a later patch fix this up?
The aim is to remove all dev->mii accesses.

No, unfortunately, this is an mii-leftover. I thought that
usbnet was emulating the mii ioctl somehow, not that it was expecting
to be able to use a mii_mdio_read. I will remove the

quoted

quoted

quoted

+	.ndo_eth_ioctl		= usbnet_mii_ioctl,

line, entirely.

It would be nice to use what is proposed, so the ioctl keeps working.

You can test it with mii-tool.

Thanks for the help, Andrew. I now implemented it as proposed by Nicolai 
and will send it shortly, after a bit more testing.

Birger