4

I followed the following guide looking for a way to solve at last an issue I have been having in 17.10 (Worked in 17.04 but I can't confirm if it did all the time).

The issue can be found in https://ubuntuforums.org/showthread.php?t=2356217 but basically, the Intel ethernet card, suddenly stopped connecting. It simply does not detect an ethernet connection established.

I have physically tried connecting several LAN RJ45 cables to it (Changing between 4 that I know work because they work in other computers at home). I also tried testing it between my computer and an Asus RT5300 and an Asus AC-68U. Currently it is connected directly to a Netgear Orbi (The router) and it does not work (I tried another computer to the same port in the Orbi and it works perfect).

Dmesg mentions the following about the card:

enter image description here

And the Gnome Network GUI shows this:

enter image description here

LSHW shows the following:

enter image description here

ModInfo gave me version 3.2.6-K

enter image description here

Trying to compile the same drivers, following the guide above gave me this (after editing the nvm.c file as mentioned in the guide):

enter image description here

My computer is the following:

enter image description here

The motherboard is the Asus ROG-Maximus VIII Hero Alpha with the latest 3703 firmware update.

I have literally tried a lot of things over the past months with no luck, including the recommendations mentioned here Ethernet adapter was disable on Ubuntu 17.04

Luis Alvarado
  • 211,503
  • I would try booting with a Ubuntu 16.04 Live USB flash drive and seeing if the Ethernet adapter works ok. Then I would work up from there. – WinEunuuchs2Unix Jan 14 '18 at 02:03
  • I'm not sure we know what to suggest as you are running a 4.15-rc7 kernel. By definition, things may or may not work as expected in an rc kernel. Can you boot into a default Ubuntu kernel? Is the symptom the same? Is there no other message in the log? dmesg | grep e100 – chili555 Jan 14 '18 at 02:48
  • This applies to the original kernel also. I loaded the 4.15 at the moment of the photo but I have 4.13 by default there also. – Luis Alvarado Jan 14 '18 at 02:52
  • I see no checksum error as in the ubuntuforums link. Does that appear somewhere in dmesg? – chili555 Jan 14 '18 at 03:09
  • Nowhere in dmesg. Also thought about doing that and got nothing. What other ideas do you have buddy? – Luis Alvarado Jan 14 '18 at 03:48
  • @LuisAlvarado - did you ever find a solution to this? The ethernet port on my external USB-3 hub identifies as "enp0s31f6" and also shows as "cable unplugged". – BaronGrivet Aug 27 '18 at 23:16
  • Sadly I did. I checked by installing windows 10 on the computer thinking it was Linux. Wrong idea. Ubuntu was at least detecting the device but as it looks the motherboard is failing. It could be either a firmware update with them that Asus never cared to fix (which am not the only one based on asus forum) or the motherboard got fried and that was the only issue that suffered. – Luis Alvarado Aug 28 '18 at 00:55

2 Answers2

1

The ubuntuforums link that you followed is a technique to patch the driver to ignore an NVM Checksum error. This error prevents the driver from loading at all. The first symptom is that the device appears as unclaimed in lshw. As well, the error is quite apparent in dmesg; it usually manifests like:

[1.049935] e1000e 0000:00:1f.6: The NVM Checksum Is Not Valid
[1.084914] e1000e: probe of 0000:00:1f.6 failed with error -5

Since the driver won’t load, no interface such as your enp0s31f6 is created.

Incidentally, for the benefit of the searchers, there are much easier ways than hacking the driver to solve the NVM Checksum error. Please see: https://ubuntuforums.org/showthread.php?t=2351572 Essentially, it involves downloading a bit of code from Intel and running it to tell the hardware that the checksum it has is correct. Afterwards, the driver sees no checksum error, loads and runs.

It is not productive, in my opinion, to apply the checksum fix in a setting where no checksum error is evident.

In your case, I suggest that you run:

sudo ethtool enp0s31f6

I suspect you will find, in part:

Speed: Unknown!
Duplex: Unknown!

You could try telling the interface what speed and duplex to assume to see if it is helpful:

sudo ethtool -s enp0s31f6 speed 1000 duplex full autoneg on

Does the interface spring to life? If so, we can make the parameters persistent.

If none of this helps, as I strongly suspect, then I believe that the ethernet jack on the motherboard or one or more of the PCB traces to the actual chip is defective.

Intel gigabit ethernet cards are inexpensive and readily available. I suggest that you abandon the defective on-board chip.

chili555
  • 60,188
  • That is correct, found Unknown for both, Speed and Duplex. I ran the ethtool line below but the interface is still dead. I am one step ahead in regards of literally abandoning the motherboard, just saving money to get a new one if it comes to that. If no answer is found over this week then I will follow your advice, and as always, excellent work chili555, you surprise me as always. – Luis Alvarado Jan 14 '18 at 14:51
  • I hope a better solution is found, Luis. I appreciate your very kind comments. – chili555 Jan 14 '18 at 21:10
0

Had the same problem. Your chip is fine most probably.

try this:

sudo apt install ifplugd

Then reboot with cable plugged in.

Should do the trick.

Ifly777
  • 123