RE: Marvell Phy (1510) issue since v4.7 kernel

-----Original Message-----
From: Andrew Lunn [mailto:andrew@lunn.ch]
Sent: Monday, January 09, 2017 6:56 PM
To: Kwok, WingMan
Cc: Karicheri, Muralidharan; netdev@vger.kernel.org
Subject: Re: Marvell Phy (1510) issue since v4.7 kernel

quoted

But when kernel is reboot into u-boot, "mii read 0 0" shows
0x1000, ie. bit 11 is 0, and still the phy auto-nego times out.

O.K, not so simple then.

I suggest you dump all the registers, in both the good and bad state,
and see how they differ.

    Andrew

From Marvell's brief description http://www.marvell.com/transceivers/alaska-gbe/,
it seems that 88E1510/1518 don't support fiber. Only 88E1512 does. In
that case, the fiber support patch is not applicable to 88E1510/1518.

I investigate a little more of the u-boot auto-nego time out
problem after reboot and found that the problem is caused by
marvell_read_status() which leaves the fiber page out because the
phydev->link is TRUE. But if fiber is not applicable to 88E1510/1518,
this is an invalid check.

Any suggestion?

WingMan