Dual Intel 82574L NIC's onboard my Supermicro x9scm-iif - packet of death Syndrome

MooCow

MooCow

[H]F Junkie
Joined
Apr 13, 2000
Messages
8,216
So I was completely unaware about a specific batch of Intel 82574L NIC's with shitty EEPROM's that cause the NIC's to completely die out. I was using ESXi v5.1 hardware pass-through to a Linux router/firewall distro. One NIC conencts to the Cable modem and the other connects to the switch. In less than two weeks, my internet connection completely died out twice on two separate occasions. I looked it up and found out that I got hit with the bad batch of 82574 NIC's that are affected by the "Packet of Death". Intel blamed a specific motherboard manufacturer for the problem and wouldn't name who. Then another Intel community user called them out on it since they use an Intel made motherboard with the onboard 82574 and their NIC goes offline all the time.

I've read a bunch about the Intel Packet of Death specifically only affecting the 82574 chipset. Some people offered a EEPROM fix to disable ASPM on the NIC, others said to modify their buggy linux e1000e drivers. I've also read that other's tried every fix available on the internet and their NIC's still shut off randomly.

Has anyone actually had a bad batch of Intel 82574L's on a Supermicro motherboard and fixed them with whatever means necessary? Right now my only work around is to take up a PCI-e slot for a dual LAN port card based on the Intel 82571EB hardware. The 82571 has been very solid for me when I used it on an older motherboard in the past, but I wanted to go without a PCI-e NIC by switching to a Supermicro board with integrated LAN. I can run my setup like this, but I'm really pissed that my motherboard has a useless set of NIC's.

I know Linux has buggy e1000e drivers, but I don't think that is the root cause of the problem.

There's a lot of half ass solutions for this problem from what I'm reading through. Has someone with a Supermicro board with onboard 82574L's able to confirm a real working fix?

If anyone wants to know what I'm talking about, go here: https://communities.intel.com/commu...-82574l-gigabit-ethernet-controller-statement

http://blog.krisk.org/2013/02/packets-of-death.html
http://blog.krisk.org/2013/02/packets-of-death-update.html
 
I had some on a Supermicro ITX Atom D525 board. A new EEPROM resolved my issue. I believe the issue was high CPU usage and they maxed out at 50MB/s before the fix.
 
Can you host the file somewhere for me to see? I obviously won't run it, but I just want to see who published the file, readme's, etc.
 
You must log in or register to reply here.
Back
Top