I’ve had this happen about 4 times now: Fedora hangs hard and requires a hard reboot. Just prior to the hang I was usually selecting something with the mouse; once I was running a VM (QEMU-KVM).
These lines appear in the logs (journalctl) adjacent to the hang and reboot every time. I don’t know if this is the problem or a symptom:
Jan 25 15:29:32 fedora kwin_wayland: kwin_libinput: Libinput: client bug: timer event8 tap: scheduled expiry is in the past (-98ms), your system is too slow
Jan 25 15:29:32 fedora kwin_wayland: kwin_libinput: Libinput: client bug: timer event8 hold: scheduled expiry is in the past (-517ms), your system is too slow
Dell 9520 laptop, i7-12700H, 32GB RAM, ~750GB SSD (the rest is Windows 11). The 3050 GPU is not operational (some driver issue that I can’t resolve) and the i7-12700 GPU is doing all the graphics.
How do I troubleshoot this more, or file a bug on it? (I looked at bugzilla via google earlier; there’s a very similar sounding issue that was raised in 2020 but is still being kicked around it looks like).
I don’t even understand how Fedora is hanging this hard. To me it points at a driver bug/issue, possibly a deadlock.
These libinput messages are a red herring and are fairly common. Next time this happens, it would be useful to review the tail end output of journal -b -1to see if there are any messages there that might be helpful and report them back here.
I noticed you’re using kwin with Wayland. Does this also happen if you run KDE with Xorg?
You said the 3050 is not working. That fact may be related, and should be resolved. We probably can help.
If the nvidia drivers are not loaded and working properly the nouveau driver is loaded instead. The nouveau driver does not support hardware acceleration on the GPU, which tasks the CPU with software rendering of the graphics and drastically interferes with performance, potentially delaying other inputs as noted with the mouse.
For that issue, please post the output of inxi -Fzxx and dnf list installed '*nvidia*'
See https://ask.fedoraproject.org/t/nvidia-driver-being-unloaded-on-boot-dell-9520-laptop/30568: it appears the rpmfusion drivers are loaded but the 3050 is not being used by anything: Blender can’t find a CUDA-capable GPU; Firefox has hardware acceleration enabled. I’ve tried setting a performance counter on the GPU but it never moves from 0%. nvidia-settings apparently shows that the 3050 is identified and the drivers (525.78.01) are loaded; nvidia-smi shows that it’s Off.
it’s blacklisted and not loaded (we addressed this confusing error message previously):
Jan 28 15:03:53 fedora systemd: nvidia-fallback.service - Fallback to nouveau as nvidia did not load was skipped because of a failed condition check (ConditionPathExists=!/sys/module/nvidia).
I have noticed I’m running a little low on memory:
ID-1: swap-1 type: zram size: 8 GiB used: 7.24 GiB (90.5%) priority: 100
Use of uninitialized value $unit in concatenation (.) or string at /usr/bin/inxi line 24273.
Use of uninitialized value $value in concatenation (.) or string at /usr/bin/inxi line 24273.
Src: /sys System Temperatures: cpu: 68.0 C mobo: N/A
Fan Speeds (RPM): N/A
Power: 12v: N/A 5v: 5 3.3v: N/A vbat: N/A
Processes: 524 Uptime: 1d 7h 7m Memory: 31.02 GiB used: 27.71 GiB (89.3%)
It’s not exactly critical but the amount of swap space is a little low for my liking. The only time I crashed Debian was when I ran it out of physical and virtual memory. It did a kernel panic and dumped the core. I’m not seeing anything like that here, if indeed I am managing to run the machine out of memory: it simply hangs hard like it’s deadlocked at kernel level.