On 08/11/2025 08:58, Russell King (Oracle) wrote:

On Fri, Nov 07, 2025 at 07:23:26PM +0100, Maxime Chevallier wrote:

quoted

Hi Russell,

On 07/11/2025 15:29, Russell King (Oracle) wrote:

quoted

-	switch (plat_dat->phy_interface) {
-	case PHY_INTERFACE_MODE_RGMII:
-	case PHY_INTERFACE_MODE_RGMII_ID:
-	case PHY_INTERFACE_MODE_RGMII_RXID:
-	case PHY_INTERFACE_MODE_RGMII_TXID:
-		phy_intf_sel = ETHER_CONFIG_INTF_RGMII;
-		break;
-	case PHY_INTERFACE_MODE_MII:
-		phy_intf_sel = ETHER_CONFIG_INTF_MII;
-		break;
-	case PHY_INTERFACE_MODE_RMII:
-		phy_intf_sel = ETHER_CONFIG_INTF_RMII;
-		break;
-	default:
+	int phy_intf_sel;
+
+	phy_intf_sel = stmmac_get_phy_intf_sel(plat_dat->phy_interface);
+	if (phy_intf_sel != PHY_INTF_SEL_GMII_MII &&
+	    phy_intf_sel != PHY_INTF_SEL_RGMII &&
+	    phy_intf_sel != PHY_INTF_SEL_RMII) {
 		dev_err(&pdev->dev, "Unsupported phy-mode (%d)\n", plat_dat->phy_interface);
 		return -EOPNOTSUPP;
 	}

Probably not too big of a deal, but don't we now incorrectly accept the
"gmii" mode ?

We will accept GMII mode, but (a) does that matter, and (b) shouldn't
the DT binding be checking the phy-mode (we have some bindings that do.)

Thanks for the clarification, that's fine by me then :)

Maxime