Arch Linux

Ocye

Member

Registered: 2011-12-07

Posts: 29

Issue with r8169

Resurrecting an older PC I run into trouble with the ethernet. First couple of reboots it worked well. But later the ethernet wasn't to convince to serve data. The dmesg tells me 'link is down', and the wisdom of the web points to autonegotiation. Sounds easy to disable and

ethtool -s enp6s0 speed 100 duplex full autoneg off

works (ethtool tells off) but after 'netctl restart' the negotiation is up again. Neither r8168 from AUR or adding stuff to the modprobe.conf helped. For those who also experience this issue my solution was to disable netctl and up the device manually per

ip link set enp6s0 up

Eventually using KDE's Networkmanager worked also. And surprise surprise, the GUI as well as ethtool says autoneg is on!
My suggestion is to add a short advice to the networking wiki. But experts should confirm the issue/solution first.

notbuu

Member

From: AUT

Registered: 2014-02-14

Posts: 7

Re: Issue with r8169

Hello!

I've bought a new mainboard (ASRock) which has an onboard Realtek ethernet adapter:

04:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller (rev 06)

After the base installation of the system I recognized random internet disconnects. I investigated and found out that the package loss during ping sessions is very huge. I've 3 computers connected to my router, one of them is a Windows PC the other one is an Arch Laptop. Neither of them have the same problem as this mainboard has. On this PC with the problematic mainboard I've dual Boot installed, and under Windows this package loss does not occur. Therefore this is not an router or a physical network problem.

64 bytes from 188.21.9.119: icmp_seq=60 ttl=58 time=24.5 ms
64 bytes from 188.21.9.119: icmp_seq=93 ttl=58 time=24.3 ms
64 bytes from 188.21.9.119: icmp_seq=94 ttl=58 time=25.7 ms
64 bytes from 188.21.9.119: icmp_seq=95 ttl=58 time=24.4 ms
64 bytes from 188.21.9.119: icmp_seq=96 ttl=58 time=27.4 ms
64 bytes from 188.21.9.119: icmp_seq=97 ttl=58 time=29.6 ms
64 bytes from 188.21.9.119: icmp_seq=98 ttl=58 time=25.0 ms
64 bytes from 188.21.9.119: icmp_seq=99 ttl=58 time=24.7 ms
64 bytes from 188.21.9.119: icmp_seq=100 ttl=58 time=25.4 ms
^C
--- www.google.com ping statistics ---
100 packets transmitted, 34 received, 66% packet loss, time 99068ms
rtt min/avg/max/mdev = 23.970/34.360/123.340/24.999 ms

After some research I've installed the AUR r8168 package with the same result - massive package loss.

I've also tried to set autoneg off, but it was not possible even with the information from the previous post.

Also the logs (dmesg, journalctl, ...) aren't showing any problem related information.

Afterwards i plugged in a quite old (realy old) Intel Pro 100 Ethernet PCI adpter to prove the evidence if the onboard nic is really causing the troubles.

Result: With the old Intel NIC, there is no packet loss anymore!

Reflecting the aforesaid and also the previous post, I would guess that this could be a kernel<->driver related problem.

If anyone needs more detailed information, let me know.

Mario

Last edited by notbuu (2016-02-28 08:15:06)

krabat

Member

Registered: 2014-07-03

Posts: 40

Re: Issue with r8169

Exactly the problems you described have been an issue with module r8169 for a very long time, searching the web should yield a plethora of results. But these are actually fixed in more recent kernels.
So if you say you resurrected an elderly device are you by any chance still running a not so up to date kernel? If so upgrading may solve your issue.

notbuu

Member

From: AUT

Registered: 2014-02-14

Posts: 7

Re: Issue with r8169

Thank you for your reply! Currently running:

Linux autonoe 4.4.1-2-ARCH #1 SMP PREEMPT Wed Feb 3 13:12:33 UTC 2016 x86_64 GNU/Linux

Just for information: I installed this PC on March 18th, and at this time I haven't any issues. I am not quite sure but maybe it came with the latest update. I will give the kernel-lts a try.

Mario

Last edited by notbuu (2016-03-01 19:56:28)