Created attachment 282941 [details] concatenation of full dmesg and post-boot logs of kernel+NM+wpa_supplicant Hardware: Thinkpad P50 with Intel Corporation Wireless 8260 [8086:24f3] Software: Gentoo Linux with vanilla kernel, NetworkManager using wpa_supplicant In Linux 5.0 the WiFi connection to my home network is rather stable. In Linux 5.1 (5.1.5 at the moment) it's unstable with the following symptoms. Ping check to the router looks like this 64 bytes from 192.168.183.1: icmp_seq=54 ttl=64 time=5.81 ms 64 bytes from 192.168.183.1: icmp_seq=55 ttl=64 time=1.00 ms 64 bytes from 192.168.183.1: icmp_seq=56 ttl=64 time=1.17 ms 64 bytes from 192.168.183.1: icmp_seq=57 ttl=64 time=3.31 ms From 192.168.183.5 icmp_seq=58 Destination Host Unreachable From 192.168.183.5 icmp_seq=59 Destination Host Unreachable From 192.168.183.5 icmp_seq=60 Destination Host Unreachable From 192.168.183.5 icmp_seq=65 Destination Host Unreachable From 192.168.183.5 icmp_seq=66 Destination Host Unreachable From 192.168.183.5 icmp_seq=67 Destination Host Unreachable From 192.168.183.5 icmp_seq=68 Destination Host Unreachable From 192.168.183.5 icmp_seq=69 Destination Host Unreachable From 192.168.183.5 icmp_seq=70 Destination Host Unreachable 64 bytes from 192.168.183.1: icmp_seq=78 ttl=64 time=1.07 ms 64 bytes from 192.168.183.1: icmp_seq=79 ttl=64 time=0.940 ms 64 bytes from 192.168.183.1: icmp_seq=80 ttl=64 time=0.951 ms 64 bytes from 192.168.183.1: icmp_seq=81 ttl=64 time=0.950 ms 64 bytes from 192.168.183.1: icmp_seq=82 ttl=64 time=1.17 ms and so on, changing from ok to host unreachable a few times per minute. The dmesg output looks like this (full log in attachment): [ 408.580520] wlp4s0: associated [ 420.822273] wlp4s0: Connection to AP ec:43:f6:07:90:84 lost [ 423.420415] wlp4s0: authenticate with ec:43:f6:07:90:84 [ 423.431798] wlp4s0: send auth to ec:43:f6:07:90:84 (try 1/3) [ 423.436016] wlp4s0: authenticated [ 423.438864] wlp4s0: associate with ec:43:f6:07:90:84 (try 1/3) [ 423.542285] wlp4s0: associate with ec:43:f6:07:90:84 (try 2/3) [ 423.550083] wlp4s0: RX AssocResp from ec:43:f6:07:90:84 (capab=0xc11 status=0 aid=3) [ 423.552862] wlp4s0: associated [ 424.045586] iwlwifi 0000:04:00.0: No beacon heard and the time event is over already... [ 424.045622] wlp4s0: Connection to AP ec:43:f6:07:90:84 lost [ 425.178388] wlp4s0: authenticate with ec:43:f6:07:90:84 [ 425.188576] wlp4s0: send auth to ec:43:f6:07:90:84 (try 1/3) [ 425.191274] wlp4s0: authenticated [ 425.192183] wlp4s0: associate with ec:43:f6:07:90:84 (try 1/3) [ 425.202483] wlp4s0: RX AssocResp from ec:43:f6:07:90:84 (capab=0xc11 status=0 aid=3) [ 425.206217] wlp4s0: associated [ 425.802779] iwlwifi 0000:04:00.0: No beacon heard and the time event is over already... [ 425.802814] wlp4s0: Connection to AP ec:43:f6:07:90:84 lost [ 427.060487] wlp4s0: authenticate with ec:43:f6:07:90:84 [ 427.069870] wlp4s0: send auth to ec:43:f6:07:90:84 (try 1/3) [ 427.072586] wlp4s0: authenticated [ 427.078966] wlp4s0: associate with ec:43:f6:07:90:84 (try 1/3) [ 427.185781] wlp4s0: associate with ec:43:f6:07:90:84 (try 2/3) [ 427.222630] wlp4s0: RX AssocResp from ec:43:f6:07:90:84 (capab=0xc11 status=0 aid=3) [ 427.225860] wlp4s0: associated [ 427.683876] iwlwifi 0000:04:00.0: No beacon heard and the time event is over already... [ 427.683911] wlp4s0: Connection to AP ec:43:f6:07:90:84 lost [ 440.104459] wlp4s0: authenticate with ec:43:f6:07:90:84 [ 440.112495] wlp4s0: send auth to ec:43:f6:07:90:84 (try 1/3) [ 440.116197] wlp4s0: authenticated [ 440.119343] wlp4s0: associate with ec:43:f6:07:90:84 (try 1/3) [ 440.124865] wlp4s0: RX AssocResp from ec:43:f6:07:90:84 (capab=0xc11 status=0 aid=3) [ 440.129543] wlp4s0: associated [ 440.750534] IPv6: ADDRCONF(NETDEV_CHANGE): wlp4s0: link becomes ready [ 452.055008] wlp4s0: Connection to AP ec:43:f6:07:90:84 lost The corresponding part of NetworkManager logs is attached too.
Created attachment 283993 [details] journal excerpt for intel 8265 NIC on iwd/networkd, iwlwifi debug=0x20
I am experiencing what seems to be the same issue on the intel 8265. ``` $ lspci -kvnn | sed -n '/Network/,/^$/ p' 02:00.0 Network controller [0280]: Intel Corporation Wireless 8265 / 8275 [8086:24fd] (rev 78) Subsystem: Intel Corporation Dual Band Wireless-AC 8265 [8086:1010] Flags: bus master, fast devsel, latency 0, IRQ 141 Memory at edc00000 (64-bit, non-prefetchable) [size=8K] Capabilities: <access denied> Kernel driver in use: iwlwifi Kernel modules: iwlwifi ``` In my experience the interface can remain stable for a long time, but occasionally will disconnect and then struggles to reassociate/authenticate for a while before becoming stable again. It sometimes records the 'no beacon heard' message and always disconnects with reason 4, being inactivity. I'm using a different network stack, so I'll leave journal excerpts in case the iwd logging is more helpful. I've got some debugging set for iwlwifi as I was trying to debug my issue. $ printf %#x\n $(cat /sys/module/iwlwifi/parameters/debug) 0x20 $ journalctl --boot=517d350fbd224b28a65e491ff95036db _KERNEL_DEVICE=+pci:0000:02:00.0 + _SYSTEMD_UNIT={iwd,systemd-network}.service (attached)
forgot to mention -- in my case, $ uname -r 5.2.2-arch1-1-ARCH
Can see this occuring Intel Corporation Wireless 7260 as well on `5.1.15-arch1-1-ARCH`.
It's happening also on my Intel 3165, even on Linux 5.0.
This could be related to TX-AMSDU, which we unfortunately had to disable for older devices. Can you try this and see if it helps you with stability? https://patchwork.kernel.org/patch/11029027/
Already used that patch compiling the kernel, but I had same issues.
If I'm not mistaken, patch 11029027 has been backported to the Arch core/linux since the release of version 5.2.arch2. See here for the first appearance of the patch in Arch linux.git (v5.2, since 07/09): https://git.archlinux.org/linux.git/commit/?h=v5.2-arch2&id=f6f6b798c7330d7851fee4bbe835c3954886fc3f Additionally see here for the inclusion of the patch in Arch linux.git (v5.2.2): https://git.archlinux.org/linux.git/commit/?h=v5.2.2-arch1&id=d426d34276c9f17081179be6497cddd74154556a Which means that the TX-AMSDU patch had already been applied when I encountered the error as I described above, and applied for the boot where my journal excerpt was taken and attached above. Additionally, I have examples in my journal of what appears to be the same error (cycling disconnect with iwlwifi time event over already error) in 5.1.15-arch1, which predates the application of that patch, and another isolated example all the way back in 5.1.4-arch1 (a different AP as well), which predates the authorship of that patch. I can attach more logs if you would like. So I have examples of the error with and without the patch. For better or for worse, this error has somehow become much less common for me. The most recent example of the iwlwifi error for me was one spurious disconnect 1 week ago, in 5.2.7-arch1, even though I use this laptop every day.
Thanks kurmikon@libero.it! But your device is very different from Denis' and Ronan's... I think it would be best to create a new bug report here and provide all the logs we recommend in our wiki[1] [1] https://wireless.wiki.kernel.org/en/users/drivers/iwlwifi/debugging
Thanks for all the info, Ronan! Can you also provide trace-cmd logs as explained in the wiki I linked in comment 9?
Okay, If I get the error again, I'll trace it. I don't have a reliable way to reproduce atm, so if anyone experiencing this issue knows of a method that may trigger it, please let me know.
I've tried the patch suggested in comment 6 on top of vanilla 5.2.9 and the bug seems to be still there. Anything else to try? Should I provide the trace-cmd trace too?
Thanks for testing, Denis. So, if the issue is not related, we could use some traces as mentioned in comment #9.
Created attachment 284575 [details] iwlwifi trace (5.2.9 w/patch from comment 6) I've generated a short trace during which there were a number of lost ICMP packets and messages from wpa_supplicant in the log. The kernel was 5.2.9 + patch from comment 6. The command used was trace-cmd record -e iwlwifi -e mac80211 -e cfg80211 -e iwlwifi_msg
Thanks again. We will assign someone internally to look into this.
Created attachment 284643 [details] trace-cmd: Intel Corporation Wireless 7260 (rev 83) Trace from Intel 7260 in ThinkPad T440s attached!
Created attachment 284645 [details] Intel 7260 syslog accompanying the trace This may help correlate timestamps in the trace.
Intel 7260 seems to be affected. Linux version 5.2.9-arch1-1-ARCH (builduser@heftig-119803) (gcc version 9.1.0 (GCC)) #1 SMP PREEMPT Fri Aug 16 11:29:43 UTC 2019
Confirmed likely a software issue. Just installed another 7260 in same machine. Same issue.
The problem is still present in 5.3.0. Maybe they're independent, but I've also seen multiple firmware errors in GEO_TX_POWER_LIMIT (looking like https://bugzilla.kernel.org/show_bug.cgi?id=204151) on 5.3 while checking for this bug.