net: phy: marvell: unlock after phy_select_page() failure

We need to call phy_restore_page() even if phy_select_page() fails.
Otherwise we are holding the phy_lock_mdio_bus() lock.  This requirement
is documented at the start of the phy_select_page() function.

Fixes: e98f0bf1b985 ("net : phy: marvell: Speedup TDR data retrieval by only changing page once")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/drivers/net/phy/marvell.c b/drivers/net/phy/marvell.c
index 2c04e3b2b285fef5bce496a329d2a30b99468491..4ea226566cec94e324fa565a979c51d6a88d8b11 100644 (file)
--- a/drivers/net/phy/marvell.c
+++ b/drivers/net/phy/marvell.c
@@ -1867,7 +1867,7 @@ static int marvell_vct5_amplitude_graph(struct phy_device *phydev)
         */
        page = phy_select_page(phydev, MII_MARVELL_VCT5_PAGE);
        if (page < 0)
-               return page;
+               goto restore_page;
 
        for (distance = priv->first;
             distance <= priv->last;