Qubes os freezes and reboot while playing video

My qubes os version 4.2 has started to freeze and reboot while playing video. This did not happen before or at least not this often. It has happend 3 times in the last 2-3 weeks. I am using debian 13 in place updated templates but this should crash the qube and not freeze and reboot the entire machine if it is linked. It only happens when playing videos, not anything else. I am using chrome to view the video and the qube that displays the video has plenty of ram available and enough vcpu. I am playing the video on an second monitor via hdmi as this is a two moniutor setup.

I am trying to debug it but I don’t know why but entries are missing in the journalctl of the qube. There are a lot of gaps in the data logged by journalctl and I do not see anything leading to the crash. I do suspend my machine everyday instead of shutting it down if this is relevent to the journalctl missing entries. As an example, if I open a terminal and run journalctl -f, those entries will go missing if the machine reboots unexpectantly

dom0 also nothing of interest with journalctl which is odd since the reboot / crash should be noted in the dom0

How would one go about debugging something like this when we are missing key logs?

EDIT: more info on the crash. The entire system freezes and the sound loops over the last second or two until the machine reboots

What gpu you have?

AMD radeon 680m. Cpu is amd ryzen 9 6900hx

Can you post your xorg configuration?
It’s in /etc/X11/xorg.conf.d/20-amdgpu.conf

no such file exists in dom0. I only have 00-keyboard.conf in the directory and can’t find any amdgpu.conf anywhere

this is interesting, seems that dom0 does not have the amdgpu modules / drivers ?

[ 35.548] Kernel command line: placeholder root=/dev/mapper/qubes_dom0-root ro rd.lvm.lv=qubes_dom0/root rd.lvm.lv=qubes_dom0/swap plymouth.ignore-serial-consoles 6.6.48-1.qubes.fc37.x86_64 x86_64 rhgb quiet usbcore.authorized_default=0 amdgpu.dcdebugmask=0x10
[ 35.574] (II) LoadModule: “amdgpu”
[ 35.575] (WW) Warning, couldn’t open module amdgpu
[ 35.575] (EE) Failed to load module “amdgpu” (module does not exist, 0)

I have Intel iGPU and I have 20-intel.conf.
Is there anyone who could help with AMD iGPU config?

and I can’t install xorg-x11-drv-amdgpu becaus dnf does not find it in the qubes os dom0 repo. Do I need to add additional repos in dom0?

You don’t use dnf in dom0 but qubes-dom0-update

1 Like

I just about to write that, thanks! Installing the drivers right now

just installed the drivers, rebooted and confirmed that the drivers are indeed used, hopefully this solves the issue

1 Like

This looks like a hardware issue. Maybe related to overheating.
Try to run CPU/GPU stress tests and see if it’ll crash your system.

That is also a possiblity and I will try ro test that but it is harder to do in qubes os. The timing of the issue leads me to more of a software / kernel issue (and there are many with amd) but will definately look for hardware issues

happened again this morning after about 4 hours of playing video. I had journalctl -a -f running on both the qube and dom0 and none of them logged anything warnings or errors before the freeze and subsequent reboot. Anything else I can enable for debugging before I try to stress test the hardware?

Try it with Fedora Workstation 37 live usb and then with 41

I have some time this week so I’ll definately do this. Might as well use the live usb to stress test the machine as well

So, thanks to @MellowPoison for pointing out potential thermal issues and @KitsuneNoBaka for telling me to try it out from fedora 37 on a usb key. When I booted fedora 37 from a usb key, I was able to do a stress test on cpu and gpu and was able to trigger the reboot within 10-15 minutes. So, this clearly removed qubes os from the culprit list but could be a hardware issue. This is where the fun begins. I’m using a mini pc and it’s poorly designed. See, the air intake is on the bottom of the pc (I didn’t notice that until today…). This would not necessarily be a deal breaker but my dumb mistake made it worst. The mini pc was sitting atop another mini pc which was smaller. Small enough that the feet of the bigger one cleared the smaller one and made it so that the air intake was blocked. After putting my qubes os mini pc on the desk directly, I was able to run the stress test for over 2h without any issues. Hopefully this was the issue and the end of those reboots. Sorry for raising this issue unecessarily.

I used this website to stress test the CPU and GPU from a web browser which was perfect for testing via a live usb distro. It won’t be as thorough as a dedicated software but it was enough for my testing

1 Like

I used to have this problem too. Video playback would freeze my computer, only power button after holding down would shut down and work. I realized it was a bios setting issue. Intel memory optimization, power saving mode set to performance. Setting it backt to factory defaults and the problem went away and I haven’t had video playback issues for months now. Crossing my fingers that was indeed the solution and this isn’t going to jinx it now. Bios settings aren’t mentioned nearly enough when others aren’t having the same problem as you, no logs, hardware issue suspected.