Skip to contentSkip to conversation

e1000e EEPROM corruption

DISCUSSION Karsten Keil 4 participants 5 messages Oct 17, 2008 – Feb 22, 2011

lore
Share & tools
lore
#0 lore raw

Hi,

so after this ugly bug is finally fixed, here maybe left some victims of this bug.

If here is still sombody who need help to recover from this issue,

I can probably help him, I successful recovered all machines on our side, even one which do not longer show the NIC via lspci, so the only restiction is, that the machine still boots and we have a NVM image of a similar machine.
You should know the PCI ids of the NIC from the time before it crashed and the MAC address.

Signature · 3 lines
Karsten Keil
SuSE Labs
SUSE LINUX Products GmbH, Maxfeldstr.5 90409 Nuernberg, GF: Markus Rex, HRB 16746 (AG Nuernberg)
#1 lore raw
Quoted 5 lines · On Fri, 17 Oct 2008 15:59:10 +0200 Karsten Keil <[email protected]> wrote:

Hi,

so after this ugly bug is finally fixed, here maybe left some victims of this bug.

Quoted 9 lines

If here is still sombody who need help to recover from this issue,

I can probably help him, I successful recovered all machines on our side, even one which do not longer show the NIC via lspci, so the only restiction is, that the machine still boots and we have a NVM image of a similar machine.
You should know the PCI ids of the NIC from the time before it crashed and the MAC address.

I need some assistance here. Machine is a Thinkpad R61 and the device has fallen off the bus. It used to be at 00:19.0:

00:19.0 Ethernet controller: Intel Corporation 82566DC Gigabit Network Connection (rev 03)

I can get my hands on a NVM image from an identical machine during the week.

Rgds

Signature · 9 lines
     -- Pierre Ossman

  Linux kernel, MMC maintainer        http://www.kernel.org
  rdesktop, core developer          http://www.rdesktop.org

  WARNING: This correspondence is being monitored by the
  Swedish government. Make sure your server uses encryption
  for SMTP traffic and consider using PGP for end-to-end
  encryption.
#2 lore raw
In reply to Karsten Keil

Hi,

The thread on the list is quite old, but I think I have the same problem problem on Toshiba Tecra M5 laptop with 82573L Gigabit Ethernet Controller runling Linux 2.6.38-rc5. I can see this:
e1000e 0000:02:00.0: (unregistered net_device): The NVM Checksum Is Not Valid when I try to load e1000e module.
Ethernet controller is visible in lspci output.

I had such problem few times earlier but it used to disappear after few hours. Now the network card don't like to work from two days.

When I try to remove NVM check from the driver code the driver says:
(unregistered net_device): Invalid MAC Address: 00:00:00:00:00:00
So I assume that there are zeros in EPROM of my network adapter.
I can not use ethtool to read or write EPROM of my card because there is no eth0 device. I have access to the same laptop, so I can get correct EPROM contents. Still I don't know how to put this to my card.

Could you tell me how to reprogram my card?

Regards,

Maciek

W dniu 17.10.2008 15:59, Karsten Keil pisze:

Quoted 14 lines

Hi,

so after this ugly bug is finally fixed, here maybe left some victims of this bug.

If here is still sombody who need help to recover from this issue,

I can probably help him, I successful recovered all machines on our side, even one which do not longer show the NIC via lspci, so the only restiction is, that the machine still boots and we have a NVM image of a similar machine.
You should know the PCI ids of the NIC from the time before it crashed and the MAC address.

#3 lore raw
Quoted 2 lines · Maciej Kotliński

I had such problem few times earlier but it used to disappear after few hours.

Looks like a hardware issue to me. Cold/broken solders (reflowing the board could fix it), for example.

Quoted 4 lines

I have access to the same laptop, so I can get correct EPROM contents. Still I don't know how to put this to my card.

Could you tell me how to reprogram my card?

That would only help if the problem is a "weak programming" of the EEPROM, which is something I have never seen before. Writing to it with flacky circuitry is probably going to crap it for good.

Signature · 4 lines
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh
#4 lore raw

I found the way to run my network card without writing anything to EEPROM.

I removed NVM and MAC address checks from e1000e driver source. The module loads fine. Ethernet adapter have MAC 00:00:00:00:00:00. Than I change MAC to correct one by ifconfig eth0 hw ether. Network works fine. I could even read correct data by ethtool -e. It makes the problem more strange. The correct EEPROM is there but...

W dniu 22.02.2011 19:06, Henrique de Moraes Holschuh pisze:

Quoted 14 lines · Maciej Kotliński
Quoted 2 lines

I had such problem few times earlier but it used to disappear after few hours.

Looks like a hardware issue to me. Cold/broken solders (reflowing the board could fix it), for example.

Quoted 4 lines

I have access to the same laptop, so I can get correct EPROM contents. Still I don't know how to put this to my card.

Could you tell me how to reprogram my card?

That would only help if the problem is a "weak programming" of the EEPROM, which is something I have never seen before. Writing to it with flacky circuitry is probably going to crap it for good.