You are not logged in.
Pages: 1
Hello!
Before posting, I prefer to say that my system is really clean and recent. Everything has been upgraded, updated and the installation is only two days old. The SSD is a brand new Samsung 980 (Evo or Pro), bought especially for Arch.
My issue is that I am getting, fairly often and seemingly randomly, a pretty debilitating system hang. I have been looking every nooks and cranny on there and the wiki but could not find something solving my problem, so here I am.
Symptoms of the problem:
Every once in a while, sometimes just as soon as I boot the computer, sometimes when I resize a window or just move my mouse, I get a freeze. Mouse unresponsive, keyboard unresponsive, date stuck, I did not check if the sounds was stuck, looping or stopping as I did not have sound playing when it happened. Nothing I can do except REISUB and only the "b" key (reboot), "e" key (kill) does not provide any result. I am either forced to press the button or do a REISUB.
My Setup:
I run a NVIDIA RTX 3060, latest drivers, tested both open / proprietary to no avail. I have a Ryzen 7 5700X and 16gb of RAM. Running on that same Samsung 980. As for the system, well :
Kernel: 6.13.2-arch1-1
DE: GNOME 47.4 (Wayland)
WM: Mutter
What I have done:
Basically everything from that page (except "pci=nomsi" as I was unable to boot into the system with it).
https://wiki.archlinux.org/title/Ryzen
I even tried to disable tcl, I do not have it. I tried to tinker with the SSD, changed the SSD slot, to no avail. I did not remove SWAP though.
To be more precise, here are all the GRUB options that I currently have:
GRUB_CMDLINE_LINUX_DEFAULT="loglevel=3 quiet processor.max_cstate=1 intel_idle.max_cstate=1 nvme_core.default_ps_max_latency_us=0 initrd=/boot/amd-ucode.img rcu_nocbs=0-15 idle=nomwait sysrq_always_enabled=1"
I am really bothered as I want to daily drive this system and this has been the only issue I have been encountering so far on that system.
Thank you in advance and have a great day!
The Fix
It was fixed (tried during two weeks on both 565 and 570 dkms, without a single hang). Here is the fix in the topic:
https://bbs.archlinux.org/viewtopic.php … 5#p2226575
The kernel parameter in question:
nvidia.NVreg_EnableGpuFirmware=0 Last edited by Gallimantis (2025-02-27 10:43:43)
Offline
After a REISUB and the subsequent reboot, post the link you get from
sudo journalctl -b | curl -F 'file=@-' 0x0.stthere was a new nvidia driver released very recently that could be contributing and you might have to test a downgrade via ALA or so.
Last edited by V1del (2025-02-11 18:44:56)
Offline
Thank you! I managed to crash after a few hours, almost twice in a row (while I was typing this message the first time). Here is the output: https://0x0.st/8Z_U.txt
As for the drivers, I went from free to proprietary to do something unrelated. Both were yielding the same results. I might try to downgrade however!
Offline
Hello,
I would rather come back to give more insights. Downgrading the drivers to 565.77 worked just fine (560 would cause issues with the kernel) but the problem still occurs. We talk about once after sleep (usually, I lock the session, log-in and it hangs). But it is almost 90% less prevalent than the 570.86.16 version and the system is actually usable now.
Find attached kernel logs related to NVIDIA if it is ever useful:
570.86.16 : https://0x0.st/8ZOM.txt
565.77 : https://0x0.st/8ZWQ.txt
I will wait for a bit before trying out newer drivers, I could not find any other post here or on nvidia forum talking about this specific issue unfortunately.
Offline
There are some GSP crashes in the 565 dump. You could try to switch to nvidia rather than nvidia-open and setting
nvidia.NVreg_EnableGpuFirmware=0 on your kernel parameters.
Offline
Pages: 1