[SOLVED] Network unusable on kernels 5.13 and above

Alucryd · 2021-09-04 10:06:11

Hi all, I've been having issues with my AMD based system since kernel 5.13 came out, 5.14 didn't fix the issue. The LTS kernel does work fine however.

It seems my network stack can't be brought up, without any obvious error in journalctl, no igb loading errors, only NetworkManager complains:

Sep 04 11:41:09 desktop kernel: igb 0000:04:00.0 enp4s0: renamed from eth0
Sep 04 11:41:19 desktop NetworkManager[1997]: <info>  [1630748479.0058] device (enp4s0): driver '(null)' does not support carrier detection.
Sep 04 11:41:19 desktop NetworkManager[1997]: <info>  [1630748479.0058] device (enp4s0): driver 'igb' does not support carrier detection.
Sep 04 11:41:19 desktop NetworkManager[1997]: <info>  [1630748479.0060] manager: (enp4s0): new Ethernet device (/org/freedesktop/NetworkManager/Devices/2)
Sep 04 11:41:19 desktop NetworkManager[1997]: <info>  [1630748479.0063] device (enp4s0): state change: unmanaged -> unavailable (reason 'managed', sys-iface-state: 'external')

The following is from LTS, where it works:

Sep 04 11:46:15 desktop kernel: igb 0000:04:00.0 enp4s0: renamed from eth0
Sep 04 11:46:24 desktop NetworkManager[1756]: <info>  [1630748784.0493] manager: (enp4s0): new Ethernet device (/org/freedesktop/NetworkManager/Devices/2)
Sep 04 11:46:24 desktop NetworkManager[1756]: <info>  [1630748784.0496] device (enp4s0): state change: unmanaged -> unavailable (reason 'managed', sys-iface-state: 'external')
Sep 04 11:46:27 desktop kernel: igb 0000:04:00.0 enp4s0: igb: enp4s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.7155] device (enp4s0): carrier: link connected
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.7156] device (enp4s0): state change: unavailable -> disconnected (reason 'carrier-changed', sys-iface-state: 'managed')
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.7168] device (enp4s0): Activation: starting connection 'Wired connection 3' (6b069e0d-1502-375d-a7dd-2ddc415a9b1f)
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.7168] device (enp4s0): state change: disconnected -> prepare (reason 'none', sys-iface-state: 'managed')
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.7171] device (enp4s0): state change: prepare -> config (reason 'none', sys-iface-state: 'managed')
Sep 04 11:46:27 desktop kernel: IPv6: ADDRCONF(NETDEV_CHANGE): enp4s0: link becomes ready
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.9040] device (enp4s0): state change: config -> ip-config (reason 'none', sys-iface-state: 'managed')
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.9042] dhcp4 (enp4s0): activation: beginning transaction (timeout in 45 seconds)
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.9103] dhcp4 (enp4s0): state changed unknown -> bound, address=192.168.1.11
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.9109] device (enp4s0): state change: ip-config -> ip-check (reason 'none', sys-iface-state: 'managed')
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.9117] device (enp4s0): state change: ip-check -> secondaries (reason 'none', sys-iface-state: 'managed')
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.9118] device (enp4s0): state change: secondaries -> activated (reason 'none', sys-iface-state: 'managed')
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.9122] policy: set 'Wired connection 3' (enp4s0) as default for IPv4 routing and DNS
Sep 04 11:46:27 desktop NetworkManager[1756]: <info>  [1630748787.9124] device (enp4s0): Activation: successful, device activated.

Not sure what this means, but commands like `ip link` or `lspci` get stuck and never return anything.

Would someone have any idea on how to debug the issue?

I'm using an Asrock X570m Pro 4 which has an Intel I211 NIC, with a Ryzen 7 5800X:

04:00.0 Ethernet controller: Intel Corporation I211 Gigabit Network Connection (rev 03)

The motherboard is fully updated, and I'm also using the latest amd-ucode.

Last edited by Alucryd (2021-09-13 14:19:14)

solskog · 2021-09-04 13:43:56

I have allmost identical hardware as you are and I don't experience any issue with the new kernel nor network module. But I use systemd-networkd instead NetworkManager. Have you tried to restart NetworkManager.

Alucryd · 2021-09-04 14:02:17

I did, but NetworkManager isn't at fault here. systemd-networkd doesn't work either, even a manual dhcpcd just freezes the network card

Alucryd · 2021-09-04 14:21:17

Some more information, calling dhcpcd pegs one core to 100%, and I can see the following message in dmesg:

igb 04:00.0 enp4s0: PCIe link lost, device now detached

loqs · 2021-09-04 16:28:43

Have you tried the igb driver from https://sourceforge.net/projects/e1000/ … ble/5.7.2/ ?
If that also has the issue I would report it upstream either on that site or kernel.org.

seth · 2021-09-04 18:53:20

Context? (Full dmesg)
This might rather be the bus, did you try eg. "pcie_aspm=off", "pcie_port_pm=off", "pcie_pme=nomsi"...?

solskog · 2021-09-05 02:48:50

I agree with @seth that the CONTEXT is important here. a full output from journalctl for the new kernel could be of more help.

Alucryd · 2021-09-05 11:55:30

loqs wrote:

Have you tried the igb driver from https://sourceforge.net/projects/e1000/ … ble/5.7.2/ ?
If that also has the issue I would report it upstream either on that site or kernel.org.

I have and it didn't help unfortunately. Will see to reporting it upstream.

In the meantime, here is a full journalctl: https://paste.xinu.at/YdLb36lNGWV4Hh1K/

seth · 2021-09-05 12:19:16

Please omit "-x" (it's spam) and post a comparative journal from the lts kernel.
Wild guess, try "igb.IntMode=0" (There's not much to explain the carrier detection being unsupported and the log doesn't have the link loss)

solskog · 2021-09-05 15:53:14

And also, here is an easy way to test it:
1, Boot with archlinux-2021.09.01-x86_64.iso and save journalctl -b >/tmp/log to an external usb drive.
Since the iso use systemd-networkd, you are going to see if the error is also "carrier detection" or something else.
2. Choose an older iso that you don't get any error for network device and save the log to an external usb drive.
Paste both logs in paste.xinu.at please.

Alucryd · 2021-09-13 08:24:47

Interestingly the latest archiso works, so I ended up doing a fresh install, which solved the issue. I couldn't find what was wrong with my install, and why only LTS would work, but at least it's working fine now. Thanks for the help guys.

solskog · 2021-09-13 13:00:18

Great it worked! please close your thread by adding [SOLVED] to the tittle.

Last edited by solskog (2021-09-13 13:00:44)

Arch Linux

#1 2021-09-04 10:06:11

[SOLVED] Network unusable on kernels 5.13 and above

#2 2021-09-04 13:43:56

Re: [SOLVED] Network unusable on kernels 5.13 and above

#3 2021-09-04 14:02:17

Re: [SOLVED] Network unusable on kernels 5.13 and above

#4 2021-09-04 14:21:17

Re: [SOLVED] Network unusable on kernels 5.13 and above

#5 2021-09-04 16:28:43

Re: [SOLVED] Network unusable on kernels 5.13 and above

#6 2021-09-04 18:53:20

Re: [SOLVED] Network unusable on kernels 5.13 and above

#7 2021-09-05 02:48:50

Re: [SOLVED] Network unusable on kernels 5.13 and above

#8 2021-09-05 11:55:30

Re: [SOLVED] Network unusable on kernels 5.13 and above

#9 2021-09-05 12:19:16

Re: [SOLVED] Network unusable on kernels 5.13 and above

#10 2021-09-05 15:53:14

Re: [SOLVED] Network unusable on kernels 5.13 and above

#11 2021-09-13 08:24:47

Re: [SOLVED] Network unusable on kernels 5.13 and above

#12 2021-09-13 13:00:18

Re: [SOLVED] Network unusable on kernels 5.13 and above

Board footer