Re: [PATCH net-next] net: phy: marvell10g: enable WoL for mv2110

Hi Russell,

On Wed, Jun 23, 2021 at 09:06:18PM +0100, Russell King (Oracle) wrote:

On Wed, Jun 23, 2021 at 09:09:29PM +0800, Ling Pei Lee wrote:

quoted

+static void mv2110_get_wol(struct phy_device *phydev, struct ethtool_wolinfo *wol)
+{
+	int ret = 0;

This initialiser doesn't do anything.

quoted

+
+	wol->supported = WAKE_MAGIC;
+	wol->wolopts = 0;
+
+	ret = phy_read_mmd(phydev, MDIO_MMD_VEND2, MV_V2_WOL_CTRL);
+
+	if (ret & MV_V2_WOL_MAGIC_PKT_EN)
+		wol->wolopts |= WAKE_MAGIC;

You need to check whether "ret" is a negative number - if phy_read_mmd()
returns an error, this test could be true or false. It would be better
to have well defined behaviour (e.g. reporting that WOL is disabled?)

quoted

+		/* Reset the clear WOL status bit as it does not self-clear */
+		ret = phy_clear_bits_mmd(phydev, MDIO_MMD_VEND2,
+					 MV_V2_WOL_CTRL,
+					 MV_V2_WOL_CLEAR_STS);
+
+		if (ret < 0)
+			return ret;
+	} else {
+		/* Disable magic packet matching & reset WOL status bit */
+		ret = phy_modify_mmd(phydev, MDIO_MMD_VEND2,
+				     MV_V2_WOL_CTRL,
+				     MV_V2_WOL_MAGIC_PKT_EN,
+				     MV_V2_WOL_CLEAR_STS);
+
+		if (ret < 0)
+			return ret;
+
+		ret = phy_clear_bits_mmd(phydev, MDIO_MMD_VEND2,
+					 MV_V2_WOL_CTRL,
+					 MV_V2_WOL_CLEAR_STS);
+
+		if (ret < 0)
+			return ret;

This phy_clear_bits_mmd() is the same as the tail end of the other part
of the if() clause. Consider moving it after the if () { } else { }
statement...

quoted

+	}
+
+	return ret;

and as all paths return "ret" just do:

	return phy_clear_bits_mmd(...

I will also need to check whether this is the same as the 88x3310, but
I'm afraid I don't have the energy this evening - please email me a
remind to look at this tomorrow. Thanks.

Shall we add this for the 88x3310 as well?

BR,
 VK