Bug 194569

This harmless warning should be fixed backporting [0] to 4.9 series.

[0] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=0b9aefea860063bb39e36bd7fe6c7087fed0ba87

I went over iwlwifi commits since then and I couldn't find any issue regarding TCP/GSO. That said, I agree that it seems to be harmless in that setup and that backporting 0b9aefea860063bb39e36bd7fe6c7087fed0ba87 should help on identifying if it's a false-positive or not.

Someone needs to send an email to netdev@ asking for it, though. kernel bugzilla is not how it's done for networking subsystem. More info in ./Documentation/networking/netdev-FAQ.txt . Thanks.

BTW, I had this same warning with vmxnet3 driver with 4.9. I tried a newer kernel and the message didn't appear. I posted my findings here: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=864642#95 - for reference:

===
It looks like the warning appeared with this first commit:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=dcb17d22e1c2cd72e72190c736349a675362b3bc
but was later corrected with this second one:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=0b9aefea860063bb39e36bd7fe6c7087fed0ba87

The first commit was merged in the 4.9 tree, but not the second one (at least not up to 4.9.106), and was only merged in 4.11.

So I would conclude the warning may be harmless, unless you got it on a >= 4.11 kernel. And I also verified I don't get it on such a newer kernel.

I would close this bug, but it may eventually be nice to backport the second commit in 4.9 to avoid the harmless warning.
===

Searching on the net, there are other reports with the same issue and with various other drivers.

I added in CC: here the net/ipv4 maintainers I found in MAINTAINERS file:

NETWORKING [IPv4/IPv6]
M:	"David S. Miller" <davem@davemloft.net>
M:	Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
M:	Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
L:	netdev@vger.kernel.org
T:	git git://git.kernel.org/pub/scm/linux/kernel/git/davem/net.git
S:	Maintained
F:	net/ipv4/
F:	net/ipv6/
F:	include/net/ip*
F:	arch/x86/net/*

I see it Linux 5.5.4 with a OpenVPN(UDP)/tun/VPN tunnel.

TCP: tun0: Driver has suspect GRO implementation, TCP performance may be compromised.

Is this the same bug? Something different?

I see it

TCP: wlan0: Driver has suspect GRO implementation, TCP performance may be compromised.


[    4.287544] iwlwifi 0000:01:00.0: firmware: direct-loading firmware iwlwifi-8000C-36.ucode
[    4.289638] iwlwifi 0000:01:00.0: loaded firmware version 36.8fd77bb3.0 op_mode iwlmvm
[    4.484488] iwlwifi 0000:01:00.0: Detected Intel(R) Dual Band Wireless AC 8260, REV=0x208
[    4.562788] iwlwifi 0000:01:00.0: base HW address: 28:16:ad:81:b2:8a
[    4.653536] ieee80211 phy0: Selected rate control algorithm 'iwl-mvm-rs'



root@debian:~# uname -a
Linux debian 5.4.0-4-amd64 #1 SMP Debian 5.4.19-1 (2020-02-13) x86_64 GNU/Linux

In a machine here it happens in a Gigabit Ethernet controller:

[81788.983716] TCP: enp4s0: Driver has suspect GRO implementation, TCP performance may be compromised.

uname -a
     Linux omitted 5.3.0-45-generic #37~18.04.1-Ubuntu SMP Fri Mar 27 15:58:10 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux


This enp4s0 is:
         description: Ethernet interface
         product: RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller
         vendor: Realtek Semiconductor Co., Ltd.
         physical id: 0
         bus info: pci@0000:04:00.0
         logical name: enp4s0
         version: 10
         serial: <omitted>
         size: 1Gbit/s
         capacity: 1Gbit/s
         width: 64 bits
         clock: 33MHz
         capabilities: bus_master cap_list ethernet physical tp 10bt 10bt-fd 100bt 100bt-fd 1000bt-fd autonegotiation
         configuration: autonegotiation=on broadcast=yes driver=r8168 driverversion=8.048.00-NAPI duplex=full ip=<omitted> latency=0 link=yes multicast=yes port=twisted pair speed=1Gbit/s
         resources: irq:126 ioport:d000(size=256) memory:df104000-df104fff memory:df100000-df103fff


modinfo r8168
     filename:       /lib/modules/5.3.0-45-generic/kernel/drivers/net/ethernet/realtek/r8168.ko
     version:        8.048.00-NAPI
     license:        GPL
     description:    RealTek RTL-8168 Gigabit Ethernet driver
     srcversion:     88B6F0A5DB60825A3057505


ethtool -i enp4s0
     driver: r8168
     version: 8.048.00-NAPI
     firmware-version: 
     expansion-rom-version: 
     bus-info: 0000:04:00.0
     supports-statistics: yes
     supports-test: no
     supports-eeprom-access: no
     supports-register-dump: yes
     supports-priv-flags: no

( The directory /lib/firmware/rtl_nic/ has the firmware files: rtl8168d-1.fw rtl8168d-2.fw rtl8168e-1.fw rtl8168e-2.fw rtl8168e-3.fw rtl8168f-1.fw rtl8168f-2.fw rtl8168g-1.fw rtl8168g-2.fw rtl8168g-3.fw rtl8168h-1.fw rtl8168h-2.fw )

In a machine here it happens in a RTL8188CE 802.11b/g/n WiFi Adapter (rev 01):

[20965.687617] wlp1s0: associated
[20965.695282] IPv6: ADDRCONF(NETDEV_CHANGE): wlp1s0: link becomes ready
[33544.393882] TCP: wlp1s0: Driver has suspect GRO implementation, TCP performance may be compromised.

$ uname -a
Linux omitted 5.7.0-1.vanilla.1.fc31.x86_64 #1 SMP Mon Jun 1 20:23:36 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux

in old kernels sometimes occur a error:
https://bugzilla.kernel.org/show_bug.cgi?id=208001

this msg first time i see  in this kernel, is not show in olds...