r8169 driver in leap 15.3 (5.3.18-59.19 kernel) only gives 100Mb/sec and not gigabit speed

anandrajan · August 26, 2021, 4:37pm

I did a fresh install of LEAP 15.3 (on a machine running LEAP 15.1). After successfully installing everything (with very few hiccups I must say), I noticed my ethernet driver (r8169) is now capped at 100Mb/sec and not gigabit. I was getting gigabit speeds in LEAP 15.1 (don’t know the kernel range sorry).

Here’s the relevant output. As you can see, the speed is 100Mb/sec instead of gigabit. I’ve also attached relevant information from dmesg. Is this related to a linux kernel bug?

agape-anand:/home/anand # ethtool eth0
Settings for eth0:
Supported ports: TP MII ]
Supported link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
Supported pause frame use: Symmetric Receive-only
Supports auto-negotiation: Yes
Supported FEC modes: Not reported
Advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
1000baseT/Full
Advertised pause frame use: Symmetric Receive-only
Advertised auto-negotiation: Yes
Advertised FEC modes: Not reported
Link partner advertised link modes: 10baseT/Half 10baseT/Full
100baseT/Half 100baseT/Full
Link partner advertised pause frame use: Symmetric
Link partner advertised auto-negotiation: Yes
Link partner advertised FEC modes: Not reported
Speed: 100Mb/s
Duplex: Full
Auto-negotiation: on
master-slave cfg: preferred slave
master-slave status: slave
Port: MII
PHYAD: 0
Transceiver: external
Supports Wake-on: pumbg
Wake-on: d
Link detected: yes

agape-anand:/home/anand # dmesg |grep r8169
4.491170] r8169 0000:03:00.0: can’t disable ASPM; OS doesn’t have ASPM control
4.549083] libphy: r8169: probed
4.549870] r8169 0000:03:00.0 eth0: RTL8168g/8111g, 6c:4b:90:4c:a5:c3, XID 4c0, IRQ 137
4.549871] r8169 0000:03:00.0 eth0: jumbo features [frames: 9194 bytes, tx checksumming: ko]
5.204480] Generic FE-GE Realtek PHY r8169-300:00: attached PHY driver [Generic FE-GE Realtek PHY] (mii_bus:phy_addr=r8169-300:00, irq=IGNORE)
5.457850] r8169 0000:03:00.0 eth0: Link is Down
17.448425] r8169 0000:03:00.0 eth0: Link is Up - 100Mbps/Full - flow control rx/tx

peteh100 · August 26, 2021, 6:45pm

Try installing the r8168 from packman. The r8168 works much better the r8169. Or you could download the r8168 from git and set it up with dkms.

Svyatko · August 27, 2021, 9:39am

Set link speed to “auto negotiate” or needed parameters.
Default is 100Mbit, half duplex.

anandrajan · August 27, 2021, 4:52pm

@Svyatko

When I do

agape-anand:/home/anand # ethtool -s eth0 speed 1000 autoneg on duplex full

I get in dmesg

[350200.177550] r8169 0000:03:00.0 eth0: Link is Down

I had to switch back to

agape-anand:/home/anand # ethtool -s eth0 speed 100 autoneg on duplex full

upon which dmesg reports

[350304.465573] r8169 0000:03:00.0 eth0: Link is Down
[350430.020047] r8169 0000:03:00.0 eth0: Link is Up - 100Mbps/Full - flow control rx/tx

And I get a viable ethernet connection.

@peteh100

I’ll try the r8168 driver from packman and will blacklist r8169 next. Thanks!

Sauerland · August 27, 2021, 4:59pm

Please use Code-Tags when you post something from Konsole…

Svyatko · August 29, 2021, 11:52am

Try to disable ASPM (power management) for card in BIOS.
Test with another cable.

gogalthorp · August 29, 2021, 5:48pm

Note network speed will not be greater then the slowest card on the network.

brunomcl · August 29, 2021, 9:02pm

Not sure I understand you.

I have half a dozen oS boxes in my LAN, some Gigabit and others 100 Mbit, connected to a Gigabit switch. All Gigabit talk to each other at full speed, regardless of how many 100 Mbit card there are in the LAN.

dcurtisfra · August 30, 2021, 3:20pm

@anandrajan:

Please post the content of ‘/etc/sysconfig/network/ifcfg-eth0’.

“STARTMODE” should be set to “auto”
“MTU” should be “1500” for most Ethernet hardware.

anandrajan · August 30, 2021, 4:14pm

@dcurtisfra, here you go.

cat /etc/sysconfig/network/ifcfg-eth0

NAME=’’
BOOTPROTO=‘dhcp’
STARTMODE=‘hotplug’
ZONE=’’

Since this didn’t have the MTU Information, here’s

ip a

2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 6c:4b:90:4c:a5:c3 brd ff:ff:ff:ff:ff:ff
altname enp3s0
inet 192.168.50.171/24 brd 192.168.50.255 scope global dynamic noprefixroute eth0
valid_lft 61459sec preferred_lft 61459sec
inet6 fe80::6e4b:90ff:fe4c:a5c3/64 scope link noprefixroute
valid_lft forever preferred_lft forever

@svyatko

There was no option in the Lenovo Ideacentre 720-IKL BIOS to disable ASPM for the onboard Realtek ethernet

Sauerland · August 30, 2021, 5:00pm

Please use Code-Tags:
https://forums.opensuse.org/showthread.php/536143-Using-Code-Tags-Around-Your-Paste

You have not to quote the command, paste the whole output in a Code-Tag incl. the wole input line:

linux64:~ # cat /etc/sysconfig/network/ifcfg-eth0
NETWORK=''
DHCLIENT_SET_DEFAULT_ROUTE='yes'
IPADDR='192.168.0.2/24'
MTU='1500'
BOOTPROTO='static'
STARTMODE='auto'
ZONE='public'
linux64:~ #

It all now a little bit better to read…

dcurtisfra · August 31, 2021, 3:16pm

@anandrajan:

The mileage you’re experiencing may be due to the following ‘ifcfg-eth0’ issues –

The variable “STARTMODE” is set to “hotplug
” – this needs the “wickedd-nanny” service … - If a physical Ethernet interface, what is the reason for using the value “hotplug
”?*=2]Normally the value “auto” is more than sufficient for physical Ethernet interfaces …
The “MTU” variable isn’t set.

Svyatko · August 31, 2021, 3:43pm

Default setting for notebooks is “hotplug” - for power saving IMHO.
Default setting for notebooks is NetworkManager, not wicked.
Works OK with MTU isn’t set there - because NetworkManager can set it (MTU = auto works OK).

To OP: try to use another card - USB plugged or another.

gogalthorp · August 31, 2021, 3:51pm

When card first connects there are handshaking protocols that determine the speed of the network connection the card adjusts speed to match. A switch is an active device thus can maintain multiple speeds. Maybe a port on the switch is broken. Try plugging into another port

anandrajan · September 1, 2021, 12:33am

I connected the ethernet cable to USB (front panel USB 3.1 Gen 2) just to check. I get gigabit ethernet this way. Thanks! I’ll stick with this.

Download speeds from speedtest.net go up to 920Mb/sec. Output from ethtool also shows gigabit.

agape-anand:/home/anand # ethtool eth1
Settings for eth1:
        Supported ports:  TP    MII ]
        Supported link modes:   10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
                                1000baseT/Half 1000baseT/Full
        Supported pause frame use: No
        Supports auto-negotiation: Yes
        Supported FEC modes: Not reported
        Advertised link modes:  10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
                                1000baseT/Full
        Advertised pause frame use: No
        Advertised auto-negotiation: Yes
        Advertised FEC modes: Not reported
        Link partner advertised link modes:  10baseT/Half 10baseT/Full
                                             100baseT/Half 100baseT/Full
                                             1000baseT/Full
        Link partner advertised pause frame use: Symmetric
        Link partner advertised auto-negotiation: Yes
        Link partner advertised FEC modes: Not reported
        Speed: 1000Mb/s
        Duplex: Full
        Auto-negotiation: on
        Port: MII
        PHYAD: 32
        Transceiver: internal
        Supports Wake-on: pumbg
        Wake-on: g
        Current message level: 0x00007fff (32767)
                               drv probe link timer ifdown ifup rx_err tx_err tx_queued intr tx_done rx_status pktdata hw wol
        Link detected: yes

Svyatko · September 1, 2021, 11:24am

What about bug report?

anandrajan · September 1, 2021, 2:49pm

What should I do?

There’s bug 202275 in kernel bugzilla which is still open and may be related to my problem.

Should I look for a corresponding opensuse bugzilla entry? And report this if I can’t find one.

Thank you so much for the ethernet to USB 3.1 suggestion. It has already made a tangible difference in handling big data transfers from the server.

Svyatko · September 3, 2021, 2:03pm

When (and if) kernel bug will be solved, we may wait for a patch in Leap for 1-3 years. That’s why it is better to report it now.
To accelerate Ethernet connection you may use 2.5G/5G cards: https://en.wikipedia.org/wiki/2.5GBASE-T_and_5GBASE-T
(Chip RTL8156 & others).

For materialised thanks use 6-rayed star button in the left bottom corner of a post.

anandrajan · September 4, 2021, 6:55pm

@Svyatko

For materialised thanks use 6-rayed star button in the left bottom corner of a post.

Done.

What about bug report?

Filed bug 1190184.

Svyatko · September 5, 2021, 4:13pm

Correct me, if I am wrong but I think there is a bug in a new kernel 5.8.0-34.