It depends on your actual hardware. I got this fixed by installing
r8158-dkms, turned out the kernel's r8169 driver was not quite right for
my ethernet hardware.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.ne
[Expired for linux (Ubuntu) because there has been no activity for 60
days.]
** Changed in: linux (Ubuntu)
Status: Incomplete => Expired
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1880076
T
tried with newer kernel on r2 (was 5.4.21 before) and now the 5.4.51
like my custom build and now i get my ~940Mbit without retransmitts...so
custom kernel on client fixes my problem...maybe because it is newer
than ubuntus 5.4 (which seems to be based on 5.4.0) or maybe because i
have no additiona
I tried:
echo 'performance' > /sys/module/pcie_aspm/parameters/policy
This did not improve performance or reduce packet loss.
I don't think it's right to give up and say "Well, this is a BIOS bug."
The 18.04 kernel does not exhibit this problem, nor does Windows, but
the 20.04 kernel does. It's
frank@frank-G5-U20:~
$ iperf3 -c 192.168.0.11 -R
Connecting to host 192.168.0.11, port 5201
Reverse mode, remote host 192.168.0.11 is sending
[ 5] local 192.168.0.21 port 43304 connected to 192.168.0.11 port 5201
[ ID] Interval Transfer Bitrate
[ 5] 0.00-1.00 sec 112 MBytes
In addition you can try to set kernel command line parameter pcie_aspm=off.
Or set pcie_aspm.policy=performance.
In general it's a tradeoff: Older kernels didn't allow to enable ASPM at
all, resulting in less battery life on notebooks. Newer kernels disable
ASPM per default, but allow to re-enable
And another data point: I wiped the computer clean and installed Ubuntu
18.04.4; the problem disappeared. TCP performance was near wire speed.
Kernel:
4.15.0-111-generic #112-Ubuntu SMP Thu Jul 9 20:32:34 UTC 2020 x86_64
x86_64 x86_64 GNU/Linux
This clearly points to a regression in software, not
One more data point. Using the ethtool command described in comment 16
(rx-usecs 200, tx-usecs 200) I see about a 10% improvement in bandwidth
but still a large number of dropped incoming packets resulting in small
TCP congestion windows (and thus a small fraction of the overall TCP
goodput achieve
Two additional data points:
1) When I boot the "broken" computer -- i.e. the one that gets rx misses
under Linux -- into Windows, Windows can drive the Ethernet chip at full
wire speed; I get 500mbps to a WAN host over TCP. I believe this
suggests it is not a problem with the BIOS.
2) I actually
In my case the bug is in the receiving direction. I see no problems with
transmitting.
Here is the output of lspci -vv (for only the ethernet controller in
question) on the broken system. Below, I have the output from the
working system.
Broken system:
01:00.0 Ethernet controller: Realtek Semico
L1SubCtl1: PCI-PM_L1.2+ PCI-PM_L1.1+ ASPM_L1.2+ ASPM_L1.1-
This means ASPM sub-state 1.2 is enabled. This should not be the case
because the driver disables ASPM at probe time. Also there's no message
in dmesg that ASPM can't be controlled by OS. So this might be a BIOS
bug. This could also explai
frank@frank-G5-U20:~
$ uname -a
Linux frank-G5-U20 5.4.52-custom #1 SMP Thu Jul 16 19:16:51 CEST 2020 x86_64
x86_64 x86_64 GNU/Linux
frank@frank-G5-U20:~
$ iperf3 -c 192.168.0.11
Connecting to host 192.168.0.11, port 5201
[ 5] local 192.168.0.21 port 49482 connected to 192.168.0.11 port 5201
[
sorry
03:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168/8411
PCI Express Gigabit Ethernet Controller (rev 15)
Subsystem: Dell RTL8111/8168/8411 PCI Express Gigabit Ethernet
Controller
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr-
Ste
The "lspci -vv" output misses the relevant part. Please execute the command as
root.
The firmware hasn't changed for years, so this should not be the reason.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/b
btw. it looks like i have a rtl8168h as lshw report this firmware to be
loaded
firmware=rtl8168h-2_0.0.2
that matches with jeremy's test...there was also 8168h affected
can be the firmware also be the case? how can i load firmware from 18.4?
--
You received this bug notification because you ar
i did not find any bios-setting regarding pcie/aspm and no "expert-mode"
to get more settings
also the "powercfg /energy" command in windows does not print any
helpful information
if i can builtin all needed drivers (localmodconfig),maybe i can use
tftp with grub anyhow (if i know how to configur
lspci -vv (20.4)
03:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168/8411
PCI Express Gigabit Ethernet Controller (rev 15)
Subsystem: Dell RTL8111/8168/8411 PCI Express Gigabit Ethernet
Controller
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- P
Main reason of missed rx packets is ASPM issues. Helpful would be the
output of "lspci -vv" to see whether the ASPM L1 sub-states are enabled.
What you can also try:
- Change ASPM settings in BIOS
- Comment out the following in rtl_hw_start_8168h_1:
rtl_hw_aspm_clkreq_enable(tp, true);
- Play wi
between 5.3 and 5.4 there are many changes in
drivers/net/ethernet/realtek
$ git log --format="%h %cd %s" --date=short 5.3-main..5.4-main --
drivers/net/ethernet/realtek/
44ee727013d5 2020-05-20 r8169: re-establish support for RTL8401 chip version
74107d56d1e8 2020-04-13 r8169: change back SG and
@jeremy
Thanks for confirming.
Is it possible to install 18.4 kernel (afair i used 5.3) in 20.4? If the
problem is gone (i guess so) we have only compare 2 major versions. Are
there any Patches from ubuntu on top of 5.3 (18.4) and 5.4 (20.4)? If
not problem is in mainline source...is the above mai
Re comment #4, I installed the latest kernel and am still having the
same issue. Kernel used:
5.8.0-050800rc5-generic #202007140715 SMP Tue Jul 14 11:18:29 UTC 2020
x86_64 x86_64 x86_64 GNU/Linux
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed
I am having an almost identical issue with Ubuntu 20.04, kernel
5.4.0-40-generic. I believe it is related to the ethernet driver. I have
two almost identical computers, both running 5.4.0-40. The bad one has
the following driver:
eth0: RTL8168h/8111h, 00:e4:4f:68:0e:77, XID 541, IRQ 126
On this c
Please test latest mainline kernel:
https://kernel.ubuntu.com/~kernel-ppa/mainline/v5.8-rc5/
** Changed in: linux (Ubuntu)
Status: Confirmed => Incomplete
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.n
Have anyone reproduced the bug? Any idea how to fix it or which change
introduces the issue?
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1880076
Title:
ubuntu 20.4 - retransmitts with r8169
To ma
** Tags added: hwe-networking-ethernet
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1880076
Title:
ubuntu 20.4 - retransmitts with r8169
To manage notifications about this bug go to:
https://bugs.
25 matches
Mail list logo
