Andrew Lunn <and...@lunn.ch> writes: >> diff --git a/drivers/net/phy/marvell.c b/drivers/net/phy/marvell.c index >> 0e0978d8a0eb..f03a510f1247 100644 >> --- a/drivers/net/phy/marvell.c >> +++ b/drivers/net/phy/marvell.c >> @@ -457,6 +457,21 @@ static int marvell_of_reg_init(struct phy_device >> *phydev) } #endif /* CONFIG_OF_MDIO */ >> >> +static int m88e1318_config_intr(struct phy_device *phydev) { >> + int err; >> + >> + err = marvell_config_intr(phydev); >> + if (err) >> + return err; >> + >> + /* Setup LED[2] as interrupt pin (active low) */ >> + return phy_modify(phydev, MII_88E1318S_PHY_LED_TCR, >> + MII_88E1318S_PHY_LED_TCR_FORCE_INT, >> + MII_88E1318S_PHY_LED_TCR_INTn_ENABLE | >> + MII_88E1318S_PHY_LED_TCR_INT_ACTIVE_LOW); >> >> Can we move this part of the code to m88e1121_config_init() ? >> >> Every time whether we disable or enable the interrupts this part of code >> will execute. > > Yes, doing this once would be better. But please allow the LED pin to > be used as an LED when not using interrupts. phy_interrupt_is_valid() > should be involved somehow.
This should be addressed in v2 of the patch which I have already posted. /Esben