Thor still freezes with R38.2.2 and does not accept any input from keyboard or mouse for 2 to 3 seconds, several times a minute. If there was meant to be a fix for the problem in 38.2.2 it did not work.
Running “nvidia-smi dmon” still stops the freezing.
I’ve attached some logs:
Xorg.0.log (22.5 KB)
journal_2025-10-09_19-40-41.txt (320.4 KB)
dmesg_r38.2.2.txt (101.2 KB)
Sorry for the late response, we’re investigating this issue, will update once have progress. Thanks
I want to add, that I experience exactly the same issue as @whitesscott with a complete new Thor developer board. Freezes of about 2 seconds happen regularly, several times a minute, mostly when typing something in the shell. When running “nvidia-smi dmon” and then opening a new shell, this freezing doesn’t happen at all. Also running “watch nvidia-smi” does seem to have the same effect. So somehow this seems to be related to keeping the device “busy” somehow …
I am copying the response of @RGobbel from the previous thread which seems to be relevant
I’m connected through the debug port, and I noticed that the freezing seems to be correlated with these messages coming out:
INFO: enter idle task. INFO: END TASK:MB�� ��ph =��ter idle ta��GPC ��sk. ��> logic map: 0=>0 1=>2 2=>1 GPC ph => logic map: 0=>0 1=>2 2=>1 ��INFO: END TASK:MB��No idea what’s generating these, but their appearance does seem to match the timing of the freezes. Hope this can be of some help.
@gordon.boeer To stop freeze for now, here’s an, I think, less resource consuming method. At each boot I run freeze.sh once.
echo "echo on |sudo tee /sys/bus/pci/devices/0000:01:00.0/power/control" >~/.local/bin/freeze.sh
chmod +x ~/.local/bin/freeze.sh
Thanks, I’ll try disabling power saving mode as you’ve suggested.
Edit:
Indeed, changing power control mode to “on” does resolve the issue.
Setting it back to “auto” as before, brings back the freezes.