> Well actually the PHY need to be initialized (at least 1 mII reg > written), which from marvel LSP driver always occurs, while it > doesn't with mainline PHY driver (drivers/net/phy/marvell.c), so the > only simple way I found to have at least one PHY reg on both > interface written is to have both eth up at OS config level. Thanks for the information. This sounds like a wake on LAN feature. I've seen other Marvell hardware connect a PHY LED output pin to the circuit controlling the main power supply. When the PHY detects the magic wake-up packet, it 'blinks' the LED so turning the power back on. My guess is, the register write to the PHY is configuring the LED. Do you have the datasheet for the PHY? Can you check this? > Probably the best option would be to have a reg-init = <reg offset > value> on both phy dts definition but the current armada mii doesn't > support this dts config... Once we understand what is going on here, we can consider adding support for this. Andrew -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html