Screen freeze and high CPU for xfwm4 (Qubes certified laptop)
Qubes OS release
- Qubes 4.2.4
- Qubes Certified Nova Custom V56 laptop
Brief summary & steps to reproduce
About every day, my screen freezes. Only the pointer of the mouse can move on the screen.
Difficult to predict when it will happen… Among others, it happened when:
- I was watching a film or
- when I was using Qubes Updates or
- several times during teleconferences, making Qubes impossible to use professionally for me.
Sometimes, a workaround (kind of) fixes the situation in 30 seconds. Sometimes, a full reboot is necessary, which obviously needs much more time.
Workaround:
- CTRL + ALT + F2
- login
- top
- Notice that the process xfwm4 is at the top of the processes for CPU
- pkill xfwm4 and exit
- CTRL + ALT + F1
Consequences after CTRL + ALT + F1:
- Most of the time: in the system tray, the Network Manager icon from disp-sys-net and the VPN icons from my VPN VMs have disappeared, even if I am successfully able to go on the internet. Killing & restarting those VMs bring back their icons.
- Sometimes: The system remains frozen and rebooting the laptop seems the only option.
Expected behavior
No screen freeze
Actual behavior
Screen freeze every day.
Most of the time: in the system tray, the Network Manager icon from disp-sys-net and the VPN icons from my VPN VMs have disappeared, even if I am successfully able to go on the internet. Killing & restarting those VMs bring back their icons.
I also have multiple issues with XFCE, and XFCE system tray in particular. Some of those: #9806, #9805, #10319 and etc.
I think, there is no way to restart XFCE sys-tray properly, because it gets no icons for network manager and other apps after restart. So, it's probably inevitable problem if you restart it. I hope migration to KDE would finally fix all these issues with XFCE.
because it gets no icons for network manager and other apps after restart.
This is going to be fixed with https://github.com/QubesOS/qubes-gui-daemon/pull/173
I have the same issue on the V54 model running kernel-latest. It has only been happening for the past few day, probably since the last round of dom0 updates:
Install kernel-1000:6.12.59-1.qubes.fc37.x86_64 @qubes-dom0-cached
Install kernel-devel-1000:6.12.59-1.qubes.fc37.x86_64 @qubes-dom0-cached
Install kernel-latest-1000:6.17.9-1.qubes.fc37.x86_64 @qubes-dom0-cached
Install kernel-latest-devel-1000:6.17.9-1.qubes.fc37.x86_64 @qubes-dom0-cached
Install kernel-latest-modules-1000:6.17.9-1.qubes.fc37.x86_64 @qubes-dom0-cached
Install kernel-latest-qubes-vm-1000:6.17.9-1.qubes.fc37.x86_64 @qubes-dom0-cached
Install kernel-modules-1000:6.12.59-1.qubes.fc37.x86_64 @qubes-dom0-cached
Install kernel-qubes-vm-1000:6.12.59-1.qubes.fc37.x86_64 @qubes-dom0-cached
Upgrade amd-gpu-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade amd-ucode-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade atheros-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade brcmfmac-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade cirrus-audio-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade intel-audio-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade intel-gpu-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade linux-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade linux-firmware-whence-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade mt7xxx-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade nvidia-gpu-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Upgrade qubes-input-proxy-1.0.45-1.fc37.x86_64 @qubes-dom0-cached
Upgrade qubes-input-proxy-receiver-1.0.45-1.fc37.x86_64 @qubes-dom0-cached
Upgrade qubes-input-proxy-sender-1.0.45-1.fc37.x86_64 @qubes-dom0-cached
Upgrade realtek-firmware-1:20251125-1.fc37.noarch @qubes-dom0-cached
Reason Change GraphicsMagick-1.3.38-3.fc37.x86_64
Will try the suggested workaround next time. edit: Workaround works for me too
Is there anything suggestive in the journal after this happens? In dom0: sudo journalctl -b. How does free -m look while xfwm4 is going rogue? What other processes are high-ish CPU in top, or is it only the window manager process?
Forgot to check the free -m now, but journalctl show this at about the right time:
Dez` 23 18:51:00 dom0 qui-updates[8056]: gdk_window_thaw_toplevel_updates: assertion 'window->update_and_descendants_freeze_count > 0' failed