Arch Linux

Please read this before reporting a bug:
https://wiki.archlinux.org/title/Bug_reporting_guidelines

Do NOT report bugs when a package is just outdated, or it is in the AUR. Use the 'flag out of date' link on the package page, or the Mailing List.

REPEAT: Do NOT report bugs for outdated packages!
Tasklist

FS#77631 - Wifi rtw88_8723de breaks boot and suspension on 6.1 kernels

Attached to Project: Arch Linux
Opened by Fabrizio Pomata (pomatius) - Thursday, 23 February 2023, 17:35 GMT
Last edited by Toolybird (Toolybird) - Saturday, 15 April 2023, 22:01 GMT
Task Type Bug Report
Category Kernel
Status Closed
Assigned To No-one
Architecture All
Severity Medium
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 2
Private No

Details

Description:

If rtw88_8723de module is loaded on a 6.1.x kernel, the machine gets stuck on boot and on resume from suspension.

Additional info:
* Any 6.1.x kernel (including lts) with latest Lightdm and i3 binaries from the official repos and the rtw88_8723de module
* journalctl log from a defective boot sequence
* There aren't any reports on upstream

Steps to reproduce:

1. Boot up your computer. It will get stuck on boot.
2. If you arch-chroot and blacklist the module by creating a .conf file on /etc/modprobe.d with the line blacklist rtw88_8723de, you will be able to boot up correctly. Once booted up, you can load the module using sudo modprobe rtw88_8723de and it works just fine.
3. If you managed to complete step 2., if the machine enters suspension (by, for example, closing your lid), when you want to resume from suspension, it gets stuck again. If you unloaded the module before going to suspension, everything works fine.
4. If you boot up a 5.x.x kernel, everything works perfectly (I had to install the 5.x.x. rt kernel from the Arch repo as a workaround in order to be able to use my laptop). I didn't tested now, but I remember it also worked on 6.0.x kernels as well. So, the problem is clearly on 6.1.x.
   rtw_bug_6.1.log (80.6 KiB)
This task depends upon

Closed by  Toolybird (Toolybird)
Saturday, 15 April 2023, 22:01 GMT
Reason for closing:  Fixed
Additional comments about closing:  See comments
Comment by Fabrizio Pomata (pomatius) - Thursday, 23 February 2023, 17:46 GMT
lshw output:

*-network
description: Wireless interface
product: RTL8723DE 802.11b/g/n PCIe Adapter
vendor: Realtek Semiconductor Co., Ltd.
physical id: 0
bus info: pci@0000:03:00.0
logical name: wlo1
version: 00
serial: 80:91:33:c5:b3:71
width: 64 bits
clock: 33MHz
capabilities: pm msi pciexpress bus_master cap_list ethernet physical wireless
configuration: broadcast=yes driver=rtw_8723de driverversion=6.1.12.arch1-1 firmware=N/A ip=192.168.100.45 latency=0 link=yes multicast=yes wireless=IEEE 802.11
resources: irq:131 ioport:3000(size=256) memory:a4000000-a400ffff

Reports from other users:

https://ask.fedoraproject.org/t/kernel-6-1-6-does-not-boot/30947/6
https://forum.endeavouros.com/t/kernel-6-1-stuck-in-boot-before-login-window/35127
https://www.reddit.com/r/linux4noobs/comments/10dsru2/i_can_only_boot_to_graphical_interface_in_the/
Comment by Toolybird (Toolybird) - Friday, 24 February 2023, 04:11 GMT
Could you please test the latest 6.2.arch1-1 kernel in [testing]? It appears the rtw88 driver has received some patches.
Comment by Fabrizio Pomata (pomatius) - Saturday, 25 February 2023, 06:20 GMT
Hello @Toolybird

I followed these steps:
1. Enabled test and community-test on pacman.conf
2. sudo pacman -Syu && sudo pacman -S linux
3. Created a new entry on /boot/loader/entries (systemd-boot)
4. Rebooted
5. Selected the test kernel to boot into the latest 6.2.arch1-1 kernel

Unfortunately same results, it got stuck again. I also tried blacklisting the module again with this kernel, and it booted successfully. In other words, the buggy behaviour is still replicable on this kernel as well.I run uname -a and lshw just to make sure the 6.2 kernel was actually being used during my tests.
Comment by Toolybird (Toolybird) - Monday, 27 February 2023, 05:39 GMT
Ok, then it seems like a kernel regression. General debugging advice here [1]. You could also try the linux-wireless mailing list or kernel bugzilla to see if anyone can help, or find some other upstream resource that might be able to help. Please let us know if you make any progress.

[1] https://wiki.archlinux.org/title/Kernel#Debugging_regressions
Comment by francis (yukido) - Wednesday, 08 March 2023, 17:23 GMT
I can confirm that this is still the case on 6.2.2.arch1-1. I'd like to add that disabling network services that use the module may also be an option to boot. But in particular for me, if NetworkManager had Wi-Fi enabled the kernel would crash (either on tty login at boot or as soon as you enable it from the applet or command-line). Also iwd or wpa_supplicant seem to connect to Wi-Fi just fine but if you were to restart the daemons (e.g. systemctl restart iwd) the kernel would crash. Downgrading to 5.15.94-1 was my workaround.
Comment by Toolybird (Toolybird) - Wednesday, 08 March 2023, 21:41 GMT
Dupe  FS#77651 

Someone affected *really* needs to report this upstream...
Comment by Toolybird (Toolybird) - Thursday, 13 April 2023, 08:09 GMT
Ping? Still happening with latest kernels? Nothing will happen unless someone reports it to the kernel folks.
Comment by Fabrizio Pomata (pomatius) - Saturday, 15 April 2023, 03:49 GMT
Hello:

Someone else reported the bug, here: https://bugzilla.kernel.org/show_bug.cgi?id=217016

The fix is set to ship with kernel 6.3, apparently: https://lore.kernel.org/linux-wireless/20230216053633 . 20366-1-pkshih@realtek.com/T/#u

When it's out, I will test and report back. Hopefully, this solves the issue.

Cheers.
Comment by loqs (loqs) - Saturday, 15 April 2023, 07:27 GMT
@pomatius the fix you reference is as you say due to be part of 6.3 [1], it has already been back-ported to stable since 6.2.3 [2]. Have you not tested kernels newer than 6.2.2?

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=4a267bc5ea8f159b614d0549030216d0434eccca
[2] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?id=976fe246a57db503d2a0841c8cb813f2f66b78af
Comment by Fabrizio Pomata (pomatius) - Saturday, 15 April 2023, 15:23 GMT
I am happy to report that, indeed, everything works both on linux and linux-lts (here you can see that the fix was backported to stable as well: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v6.1.24&id=19fcf5d50c5a16c6de3dc18279ee33f062c7d9d2)
Since I didn't know about this, I was still holding back on to a 5.x.x kernel.So, I think we can close this up.

Thank you very much, @loqs

Loading...