Intel e1000e Ethernet not working

Question

I have the problem that the ethernet device on the Intel NUC is not found. ifconfig -a only shows lo and wlan0 but not eth0.

sudo lshw -C net gives the following output:

*-network UNCLAIMED
description: Ethernet controller  
product: Ethernet Connection (3) I218-V        
vendor: Intel Corporation       
physical id: 19             
bus info: pci@0000:00:19.0       
version: 03 
width: 32 bits   
clock: 33MHz   
capabilities: cap_list
configuration: latency=0   
memory:f7100000-f711ffff memory:f713b000-f713bfff 
ioport: f080(size=32)   

After running sudo lshw -C net and dmesg | grep -e eth -e e1000, I get this output:

[    0.847256] e1000e: Intel(R) PRO/1000 Network Driver - 2.3.2-k
[    0.847260] e1000e: Copyright(c) 1999 - 2014 Intel Corporation.
[    0.863337] e1000e 0000:00:19.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
[    1.673743] e1000e: probe of 0000:00:19.0 failed with error -2
[ 4315.170940] e1000e: module verification failed: signature and/or  required key missing - tainting kernel
[ 4315.171534] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.4.2-NAPI
[ 4315.171536] e1000e: Copyright(c) 1999 - 2015 Intel Corporation.
[ 4315.171711] e1000e 0000:00:19.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
[ 4315.994262] e1000e: probe of 0000:00:19.0 failed with error -2

Product information (lspci -nnk) is the following:

00:19.0 Ethernet controller [0200]: Intel Corporation Ethernet Connection (3) I218-V [8086:15a3] (rev 03)
        Subsystem: Intel Corporation Device [8086:2057]

Kernel version is 3.19.0-22. I am booting in UEFI mode. The BIOS version is the latest version.

As suggested, I updated the BIOS to the newest version, Left the device without power for a while and changed the hardware configuration. So far without luck!

I just installed Windows, and the device also isn't working there. This probably means that it's either a BIOS problem, as suggested, or the hardware is actually damaged.

Any further thoughts on this?

---

For others with similar problems:

In the beginning I thought this was a driver issue. I wanted to install the newest version of the e1000e driver. This requires version 8.x.x of the e1000 driver. When I tried to install this driver, I got the following error message (even though I run kernel version 3.19):

 *** Aborting the build. *** This driver is not supported on kernel versions older than 2.4.0.  Stop.

However, it turns out that the old driver on the system was working fine. The real problem has to lay somewhere else.

Solution

From the logs it seems to me you have managed to update the driver from version 2.3.2-k to 3.2.4.2-NAPI.

I fixed my Intel NUC non working ethernet by:

• download the driver from https://downloadcenter.intel.com/download/15817, currently 3.2.4.2 (as shown in lshw -C above)
• make install in the src folder
• rmmod e1000e
• modprobe e1000e
• and to make the new driver survive a reboot update-initramfs -u

This I have to repeat at every kernel update, since kernel updates still (3.13.0-63) contain the old driver version 2.3.2-k, which does not work with my Intel NUC.

Comments

[gertvdijk]

Do you have the numeric vendor/product ID of the device (12ab:34cd)? (lspci -nnk | grep -iEA3 "(network|ethernet)"

[Daniel Eckert]

Added both outputs!

[gertvdijk]

The device by the ID 8086:15a3 should be perfectly supported by any recent kernel. What's the exact kernel version you're running? (uname -rv) I am guessing we're looking at a hardware initialisation error here. Is there a BIOS update available for your system? And are you booting in legacy/BIOS mode or UEFI?

[Daniel Eckert]

@gertvdijk Post is updated!

[gertvdijk]

Got a similar issue with an HP Elitebook integrated Intel network card. Hardware does not initialize sometimes after booting, and sometimes even for 20 times rebooting. Not sure what's going on. I sporadically find some other people having this issue too. Usually changing hardware configuration in some way helps for me, or resetting the BIOS to factory defaults. Happens also for PXE (reports link down) and Windows, so seems unrelated to the OS. One of the random google results: forums.gentoo.org/viewtopic-p-7775942.html

[gertvdijk]

So, try to cut the power completely for a few minutes and turn it back on. If that works, it definitely a hardware init error (BIOS bug). Some registers aren't properly cleared and stay in a state in which it can't get out of it seems.

[Daniel Eckert]

I tried all these things, but so far without luck.

[Greg]

Did you build the newest driver from source or get a binary from Intel to install? Not sure what is "aborting the build".

[Daniel Eckert]

I built the driver from source with 'sudo make install'... when I build the e1000e driver, everything worked fine this way. Its only the e1000 driver that gives the error.

[Greg]

Weird. It seems like the Makefile is handling the kernel version wrong, or is expecting to find the development kernel headers in a different place. Surely you must have the kernel headers if you could build the e1000e driver properly. You might want to compare their Makefiles. Before you break out the solder iron for the BIOS, you should get those drivers up to date, especially with warnings of adverse interactions between them.

[Cec]

For anyone using a Z97 Chipset motherboard, I suggest the firmware to the latest build of 2016, as the driver is not to blame: the motherboard is not properly initializing the network card. I'm now running hassle-free with Debian Stretch for more than a month.

[Natim]

It wasn't compiling due to code model kernel does not support PIC mode I changed the line EXTRA_CFLAGS += $(CFLAGS_EXTRA) into EXTRA_CFLAGS += $(CFLAGS_EXTRA) -fno-pie to fix it.

[Natim]

I then had to use iwconfig to grab the interface name and then sudo ifconfig enp0s31f6 up to enable it (because ifconfig wasn't showing it)