Thousands of RX errors caused by overruns in eth0 after upgrading from Ubuntu 24.10 oracular to Ubuntu 25.05 plucky

[fprietog]

Device: Raspberry Pi 5 Model B Rev 1.1 (16 GB)
System: Ubuntu 25.05 plucky
Kernel: 6.14.0-1005-raspi #5-Ubuntu

After upgrading from Ubuntu 24.10 oracular to Ubuntu 25.05 plucky I'm getting thousands of RX errors caused by overruns in the eth0 network device. These error are silent (no journal related entries detected) but can be seen using ifconfig or ethtool commands.

I've tested a fresh install with updated packages of both Ubuntu versions involved (oracular & plucky) to make sure that this isn't a HW problem.

An example:

Using Ubuntu 24.10 oracular I don't get any RX error caused by overruns in the longrun:

# uname -a
Linux fpgrpi 6.11.0-1012-raspi #12-Ubuntu SMP PREEMPT_DYNAMIC Thu Apr 17 03:03:47 UTC 2025 aarch64 aarch64 aarch64 GNU/Linux

# ifconfig
eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500
...
        RX packets 56634495 bytes 83340934416 (83.3 GB)
        RX errors 0 dropped 0 overruns 0 frame 0
...

Using Ubuntu 25.05 plucky I get lots of RX errors caused by overruns:

# uname -a
Linux fpgrpi 6.14.0-1005-raspi #5-Ubuntu SMP PREEMPT_DYNAMIC Sat Apr 5 14:33:30 UTC 2025 aarch64 aarch64 aarch64 GNU/Linux
# ifconfig
eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500
...
        RX packets 20545305 bytes 29656990611 (29.6 GB)
        RX errors 3733 dropped 0 overruns 3733 frame 0
...

At this point I think that something has changed at kernel level from 6.11 to 6.14 that is causing these errors.

Thanks and best regards.
---
ProblemType: Bug
ApportVersion: 2.32.0-0ubuntu5
Architecture: arm64
AudioDevicesInUse:
 USER PID ACCESS COMMAND
 /dev/snd/controlC0: fprietog 2289 F.... wireplumber
 /dev/snd/seq: fprietog 2284 F.... pipewire
CasperMD5CheckResult: unknown
CloudArchitecture: aarch64
CloudID: none
CloudName: none
CloudPlatform: none
CloudSubPlatform: config
CurrentDesktop: ubuntu:GNOME
DistroRelease: Ubuntu 25.04
IwDevWlan0Link: Not connected.
Lspci-vt: -[0002:00]---00.0-[01]----00.0 Raspberry Pi Ltd RP1 PCIe 2.0 South Bridge
Package: linux-raspi (not installed)
ProcFB: 0 vc4drmfb
ProcKernelCmdLine: reboot=w coherent_pool=1M 8250.nr_uarts=1 pci=pcie_bus_safe cgroup_disable=memory numa_policy=interleave numa=fake=8 system_heap.max_order=0 smsc95xx.macaddr=2C:CF:67:C6:EA:83 vc_mem.mem_base=0x3fc00000 vc_mem.mem_size=0x40000000 zswap.enabled=1 zswap.zpool=z3fold zswap.compressor=zstd multipath=off dwc_otg.lpm_enable=0 console=tty1 root=PARTUUID=29cbc784-0b rootfstype=ext4 rootwait fixrtc rootdelay=5
ProcVersionSignature: Ubuntu 6.14.0-1005.5-raspi 6.14.0
RelatedPackageVersions:
 linux-restricted-modules-6.14.0-1005-raspi N/A
 linux-backports-modules-6.14.0-1005-raspi N/A
 linux-firmware 20250317.git1d4c88ee-0ubuntu1
StagingDrivers: rpivid_hevc
Tags: plucky staging
Uname: Linux 6.14.0-1005-raspi aarch64
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: N/A
_MarkForUpload: True
acpidump:

[Edoardo Canepa (ecanepa)]

This bug is missing log files that will aid in diagnosing the problem. From a terminal window please run:

apport-collect 2109967

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

apport information

[fprietog (fprietog)]

> This bug is missing log files that will aid in diagnosing the problem. From a terminal window please run:
>
> apport-collect 2109967

Done in a practically fresh install of Raspberry Pi Ubuntu plucky. The errors shown in ifconfig before sending apport data were:

eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500
...
        RX packets 6564437 bytes 9743289008 (9.7 GB)
        RX errors 168 dropped 4 overruns 168 frame 0
...

I thought it may be CIFS related as I use a NAS but no, errors happens even with no CIFS connection done.

[fprietog (fprietog)]

Installing and booting latest oracular 6.11 kernel in a plucky installation made the overruns errors gone.

Commands used to install latest oracular kernel in plucky:

# wget https://launchpad.net/ubuntu/+archive/primary/+files/linux-image-6.11.0-1013-raspi_6.11.0-1013.13_arm64.deb

# wget https://launchpad.net/ubuntu/+archive/primary/+files/linux-modules-6.11.0-1013-raspi_6.11.0-1013.13_arm64.deb

# dpkg -i linux-image-6.11.0-1013-raspi_6.11.0-1013.13_arm64.deb linux-modules-6.11.0-1013-raspi_6.11.0-1013.13_arm64.deb

# flash-kernel --force 6.11.0-1013-raspi

[fprietog (fprietog)]

Updated to Ubuntu 25.10 Questing, kernel 6.17.0-1003-raspi: problem gone!

So, summarizing:
- kernel 6.11: OK
- kernel 6.14: lot's of overrun errors.
- kernel 6.17: OK

There are lot's of changes between 6.14 and 6.17 regarding the RP1 and the Cadence GEM Ethernet controller. Also I don't know if these patches https://www.phoronix.com/news/Raspberry-Pi-5-Ethernet-Linux are included in 6.17 or will be included in a future version.

Anyways it works now so just closed this.