nouveau kernel module makes /usr/lib/X hang

JPT48 · 2024-01-04 18:05:15

hi,
after migration to an new laptop my system wasn't able to boot into GUI.

Turned out the cause was the nouveau kernel module crashing with a null pointer.

[drm] Initialized nouveau 1.4.0 20120801 for 0000:01:00.0 on minor 0
BUG: kernel NULL pointer dereference, address: 0000000000000020

where should I report this issue?
is there any other workaround other than deleting the module file?
How do I find out which driver it uses now?
which driver should I use?

thank you very much

JPT

short logfile:
https://pastebin.com/1imDNPXP

full dmesg:
https://pastebin.com/mCBE3Rpz

Thinkpad W541
Intel(R) Core(TM) i7-4710MQ CPU @ 2.50GHz

lspci|grep -i vga
00:02.0 VGA compatible controller: Intel Corporation 4th Gen Core Processor Integrated Graphics Controller (rev 06)
01:00.0 VGA compatible controller: NVIDIA Corporation GK107GLM [Quadro K1100M] (rev a1)

wayland 1.22.0-1
plasma-desktop 5.27.10-1 
sddm 0.20.0-2
xorg-server 21.1.10-1
kernel 6.6.9.arch1-1
xf86-video-fbdev 0.5.0-3
xf86-video-vesa 2.6.0-1
xf86-video-intel 1:2.99.917+923+gb74b67f0-1

(nouveau removed because of the problem)

Last edited by JPT48 (2024-01-09 18:02:31)

seth · 2024-01-04 22:02:52

Use https://aur.archlinux.org/packages/nvidia-470xx-dkms/ ?

nouveau removed because of the problem

You cannot remove the nouveau kernel module, if you're talking about xf86-video-nouveau, it's questionable whether that'd benefit you at all, but the nullptr deref in the kernel module is a kernel bug.
https://lore.kernel.org/lkml/CACO55tvWu … gmail.com/

Also certainly remove -vesa and -fbdev, you don't want to ever use those drivers and it's better to get the problem in your face w/ a hard failure than having it hidden behind the vesa driver.
You can or can not use xf86-video-intel, but will have to use mesa-amber or explicitly select the crocus driver, or you end up w/ software rendering.

On some formal notes:
Please use [code][/code] tags. Edit your post in this regard.
Also please don't obfuscate your hostname by removing columns from the journal.
1. It's "maxi" - you'd be better off sed'ing that to something unsuspicious like eg. "pornhub"
2. You've removed the timestamps
3. It breaks my syntax highlighting

BrunoPT · 2024-01-11 18:47:46

I have the exact same laptop and I experienced the same issue. I found some more stuff on the internet https://groups.google.com/g/linux.debia … RAgo?pli=1 but I don't think any solution was found to make it work.
You can set a kernel parameter to disable nouveau:

nouveau.modeset = 0

...or you can just install the proprietary driver, version 470 works with this graphics card.

I get some weird artifacts in some apps, probably due to using Sway which is incompatible with the nvidia drivers. It sucks, but it's usable.

seth · 2024-01-11 20:57:12

https://wiki.archlinux.org/title/Sway#F … VIDIA_GPUs ?

BrunoPT · 2024-01-11 23:49:14

I tried using the sway-git, that depends on wlroots-git which apparently have the Nvidia issues fixed.
I only have one issue now, some weird lines appear on Virt-Manager.

https://imgur.com/a/tBnvbMP

Edit: Nevermind, I changed the graphics to Virtio and it started working.

Last edited by BrunoPT (2024-01-12 00:06:13)

JPT48 · 2024-01-12 09:19:18

So, to solve the issue you have to install the proprietary driver plus sway from git?
could you please add some links? I wasn't able to find the proprietary nvidia driver.

and still my question is open:
how do you see which driver is actually used?

and regarding the kernel module, it is named nouveau.ko.zst
why do you tell this was wrong?

how do you fix the kernel crashing by using a different user mode driver?

The link to the kernel bug posted above tells the bug was solved previously. but it wasn't.

thanks

Last edited by JPT48 (2024-01-12 09:27:01)

seth · 2024-01-12 09:32:54

https://wiki.archlinux.org/title/Nvidia

lspci -k

W54J04S07T · 2024-01-14 19:33:25

JPT48 wrote:

...
The link to the kernel bug posted above tells the bug was solved previously. but it wasn't.
...

Try this in GRUB;

nouveau.noaccel=1

while using nouveau driver. "Might" work.

Arch Linux

#1 2024-01-04 18:05:15

nouveau kernel module makes /usr/lib/X hang

#2 2024-01-04 22:02:52

Re: nouveau kernel module makes /usr/lib/X hang

#3 2024-01-11 18:47:46

Re: nouveau kernel module makes /usr/lib/X hang

#4 2024-01-11 20:57:12

Re: nouveau kernel module makes /usr/lib/X hang

#5 2024-01-11 23:49:14

Re: nouveau kernel module makes /usr/lib/X hang

#6 2024-01-12 09:19:18

Re: nouveau kernel module makes /usr/lib/X hang

#7 2024-01-12 09:32:54

Re: nouveau kernel module makes /usr/lib/X hang

#8 2024-01-14 19:33:25

Re: nouveau kernel module makes /usr/lib/X hang

Board footer