Hi, There,
We are using Texas - DP83867IRRGZR in one of our design.
We build around, 1K units in production and most of them are working fine at 1GBPS.
But our production has come across around 4 units (out of around 1K units) which has issue to communicate at 1GBPS but working fine at 100MBPS.
I verified few, registers on SMI bus.
Non working unit Registers value.
=> mdio read FEC0 0x3 0x01
1 - 0x7949
=> mdio read FEC0 0x3 0x08
8 - 0x6801
=> mdio read FEC0 0x3 0x0A
10 - 0x0800
=> mdio read FEC0 0x3 0x11
17 – 0x7f02
=> mdio read FEC0 0x3 0x11
17 - 0x6f02
=> mdio read FEC0 0x3 0x13
19 – 0x9c40
Working unit Registers value.
=> mdio read FEC0 0x3 0x01
1 - 0x796d
=> mdio read FEC0 0x3 0x08
8 - 0x4000
=> mdio read FEC0 0x3 0x0A
10 - 0x3800
=> mdio read FEC0 0x3 0x11
17 – 0xbf02
=> mdio read FEC0 0x3 0x11
17 - 0xaf02
=> mdio read FEC0 0x3 0x13
19 – 0x1c40
Initial thought drive me that, Auto negotiation on Non working unit is failing at 1GBPS. Due to with ssh and ping was not responding on these units.
I tried setting this Non Working unit to manually operate on 100MBPS using command,
`sudo ethtool -s eth0 autoneg on speed 100 duplex full`
With this change unit started responding at 100MBPS for ssh and ping.
But if I set Non working unit to operate at 1 GBPS, I again see same issue.
We checked components and possible dry solder on non working units but we found things same between Non working and Working units.
This is initial information from my side to start this thread,
I need support from you,
Where should I look for root cause. ? I mean, DP83867IRRGZR H/W register configuration (Schematic) , Clock , PCB Design ….....
Let me know what more information you need from me.
And sorry for such a long email, 
. But I spend more time in investigation so though I should start thread with as much information as I can.
Thanks,
Suhant Raval
