|Summary:||Kernel 5.19 with NetworkManager and iwd: Wifi reconnect on resume is broken|
|Component:||network-wireless-intel||Assignee:||Default virtual assignee for network-wireless-intel (drivers_network-wireless-intel)|
|Severity:||normal||CC:||andyrtr, dereklh24, jlinenkohl, kernel+bugzilla, plumerlis|
Description Vorpal 2022-08-22 16:35:21 UTC
Created attachment 301630 [details] Relevant journalctl.txt After upgrading to kernel 5.19 (Arch Linux kernel, which is minimally modified, but I tested a few alternative kernels too, such as zen), wifi reconnect on resume is broken when using NetworkManager + iwd. When downgrading to 5.18 or linux-lts (5.15.62), auto reconnect starts working again. When switching NetworkManager to use wpa_supplicant auto reconnect starts working on newer kernel again. Manual reconnection works using iwd+new kernel. Hardware: * Lenovo Thinkpad T480 with "Intel Corporation Wireless 8265 / 8275 (rev 78)" (according to lspci). * Access point is a Asus AC-68U running the Asus Merlin Firmware. I have another laptop, with Atheros wifi and the same Arch Linux with NM + iwd setup. It does not suffer from this problem. This appears to be specific to this hardware. I have attached the relevant fragment of dmesg/journalctl (MAC addresses and SSID changed for privacy). Additional info: * Arch Linux package version(s): - linux-5.19.3.arch1-1 (also affects 5.19.2 at least) - networkmanager-1.38.4-1 - iwd-1.29-1 (broken) - wpa_supplicant-2:2.10-5 (works) Steps to reproduce: 1. Suspend computer with iwlwifi using NetworkManager + iwd 2. After computer has gone to sleep, wake it up 3. Observe failure to reconnect to wifi network that was connected before. I do not know if this is a kernel bug, IWD bug or Network Manager bug. Perhaps it is just how they interact. Distro bug report: https://bugs.archlinux.org/task/75670
Comment 1 Emmanuel Grumbach 2022-08-26 14:34:05 UTC
*** This bug has been marked as a duplicate of bug 215937 ***
Comment 2 Blanche Schaefer 2022-08-30 15:20:40 UTC
I have similar hardware and system setup and faced same issue since linux 5.19. Neither logs posted above or my own mention iwlmei, it's just the: iwlwifi 0000:04:00.0: Not associated and the time event is over already... I also rebuild kernel with CONFIG_IWLMEI=n and it didn't make any difference so I don't think this bug should be duplicate of https://bugzilla.kernel.org/show_bug.cgi?id=215937 . Please re-open it.
Comment 3 Blanche Schaefer 2022-08-30 19:16:33 UTC
Also the aforementioned bug was originally reported for kernel 5.17 while this one didn't occur before 5.19
Comment 4 plum 2022-09-05 14:15:00 UTC
I have same issue https://bugzilla.redhat.com/show_bug.cgi?id=2118137
Comment 5 Emmanuel Grumbach 2022-09-08 19:23:47 UTC
Vorpal, can you tell us if compiling IWLMEI out helps. If yes, I'll ask Blanche to open a new bug.
Comment 6 Vorpal 2022-09-08 22:19:44 UTC
(In reply to Emmanuel Grumbach from comment #5) > Vorpal, can you tell us if compiling IWLMEI out helps. > If yes, I'll ask Blanche to open a new bug. I'm currently way too busy to set up and recompile a kernel, but I should have some time to try this during next week.
Comment 7 Vorpal 2022-09-09 08:29:22 UTC
I ended up getting time to do this much sooner than I expected. I can confirm that disabling iwlmei does not affect this.
Comment 8 Justin Linenkohl 2022-11-15 09:34:28 UTC
Any chance that this is related to my remarks re: SSID not being returned by nl80211 CMD_GET_INTERFACE? I have not tried to reproduce the symptoms described here in 216397, but think perhaps NetworkManager and similar "network config daemons" might also be affected as it's not an issue _in_ them but rather a symptom of the missing netlink return data. Like I mention in "iw --debug" and systemd-251-3 atop Linux 5.19.2 and beyond. # details https://github.com/systemd/systemd/issues/24585 # other reports https://github.com/systemd/systemd/issues/25384 https://github.com/systemd/systemd/issues/24411 # Yu Watanabe (yuwata) - systemd - also reported to Red Hat back in Sept 2022. https://bugzilla.redhat.com/show_bug.cgi?id=2124670 I have not reported to jmberg-intel as I know he was quite busy Sept/Oct but was curious if a bug report exists or anyone else is tracking/testing this as I spent a good chunk of time on it and hope to spare others the discovery time... Thanks. -Justin