It's linux-firmware.
Added it to my worryingly growing IgnorePkg list:
IgnorePkg = linux-firmware linux-firmware-amdgpu lib32-libpcap lib32-libnl libpcap libcupsfilters

It's working again, thank you very much!

export DESKTOP_SESSION=plasma
exec startplasma-x11

This was run when I started my plasma session via sddm, so it started the x11 session. After removing these lines the wayland session would not start, but I have now switched to Plasma Login Manager and everything seems to work now.

The thing is, I'm working away quietly on my computer (an AMD) when suddenly I lose my network connection—nothing responds at all. About a minute after it happens, if I go to the journal, I see this appear:

sep 12 12:06:46 hell kernel: r8169 0000:0e:00.0 enp14s0: NETDEV WATCHDOG: CPU: 14: transmit queue 0 timed out 5440 ms
sep 12 12:06:46 hell kernel: r8169 0000:0e:00.0 enp14s0: ASPM disabled on Tx timeout
sep 12 12:06:46 hell kernel: r8169 0000:0e:00.0 enp14s0: rtl_rxtx_empty_cond == 0 (loop: 42, delay: 100).
sep 12 12:06:52 hell kernel: r8169 0000:0e:00.0 enp14s0: NETDEV WATCHDOG: CPU: 14: transmit queue 0 timed out 5777 ms
sep 12 12:07:08 hell kernel: r8169 0000:0e:00.0 enp14s0: NETDEV WATCHDOG: CPU: 3: transmit queue 0 timed out 5681 ms
sep 12 12:07:14 hell kernel: r8169 0000:0e:00.0 enp14s0: NETDEV WATCHDOG: CPU: 14: transmit queue 0 timed out 5399 ms
sep 12 12:07:20 hell kernel: r8169 0000:0e:00.0 enp14s0: NETDEV WATCHDOG: CPU: 11: transmit queue 0 timed out 5693 ms
sep 12 12:07:26 hell kernel: r8169 0000:0e:00.0 enp14s0: NETDEV WATCHDOG: CPU: 1: transmit queue 0 timed out 5492 ms
sep 12 12:07:26 hell kernel: r8169 0000:0e:00.0 enp14s0: rtl_rxtx_empty_cond == 0 (loop: 42, delay: 100).
sep 12 12:07:58 hell kernel: r8169 0000:0e:00.0 enp14s0: NETDEV WATCHDOG: CPU: 9: transmit queue 0 timed out 5784 ms
sep 12 12:07:58 hell kernel: r8169 0000:0e:00.0 enp14s0: rtl_rxtx_empty_cond == 0 (loop: 42, delay: 100).

At that point, the only thing left for me to do is restart. There is (as far as I know) no way to recover the network.

This doesn't happen to me all the time; it's completely random. It might happen shortly after I start up the computer, or I might go days without it crashing. The only thing I know for sure is that this is a problem with Kernel version 7.

Anyway, I'm just leaving this here to let you know that if this happens to anyone else, you're not alone, my friend.

Moderator Note
Closing this old thread.

looks sketchy (see the format value matching the width, might be bug or stack corruption)

This seems to be debugger noise.

I've updated the issue info after some new test: https://gitlab.gnome.org/GNOME/gtk/-/work_items/8333

Pointing to software decoded video between gsk and vulkan.

ERROR: GPU:1: nvHandleCpEventDeferredWork: DisplayId 0x10000 Invalid

I rolled back to version 610.57.04 to get a functioning desktop environment.

strongswan  6.1.0 disables IKE1 by default. https://gitlab.archlinux.org/archlinux/ … ote_553014
Does you company provide any alternative protocols you could connect with instead?

Welp, since I did clean pacman's cache right after that update (disk was almost full), I guess I have to download previous strongswan version then.
Would libreswan package from AUR fix my problem? And if yes, will that move from AUR to core or extra repo?

Unfortunately, this IKEv1 protocol is the only way they provide for me to connect to that VPN network which is frustrating since it also doesn't work on my cellphone (iirc, since Android v11 l2tp is disabled).
Thank you @-thc and @loqs for your feedback.

EDIT: after downgrading from strongswan-6.1.0-2 to strongswan-6.0.7-1, now everything is back in order.

I did some research and apparently booting from a snapshot only boots in read only mode but it allows you to find the right snapshot before the system broke. There's a "rollback" command on the openSUSE systems but apparently in Arch you have to do it manually. Mount your desired snapshot temporarily and overwrite the root subvolume but like you, I'm going to wait until it happens to play around with this.

I have snapper, snap-pac, and grub-btrfs (available from btrfs-progs). Snapper does have a rollback command.

You boot into the snapshot from GRUB

Confirm it is the correct snapshot

Rollback with sudo snapper -c root rollback <snapshot number>

That replaces your 0 snapshot (current) with the desired snapshot.

Then you can reboot the system and it will resume operations from the restored snapshot.

If you mean a literal "rollback" command, one could create an alias of the snapper command.

After a bit of searching I have also found another fix, but I would like to avoid recompiling the kernel if I can.

Try and get the patch accepted upstream so you can avoid the recompile?

Does this repo contain patches for both cameras or one camera and another fix?

I have a system-wide configuration file in place that clearly tells OpenAL to use PulseAudio, but for some reason it fails to load libpulse.so.0, even though it definitely exists in my /usr/lib. Strangely, older versions such as 1.8.9 are not affected. I don't know why this happens, but preloading the library fixes the problem.

I created a small wrapper script "/usr/local/bin/minecraft-launcher":

#!/usr/bin/sh -eu
export LD_PRELOAD=/usr/lib/libpulse.so.0
exec /usr/bin/minecraft-launcher

Now the launcher and everything run by it has that thing already loaded and ready for use.