Attention Nvidia GPU / Driver users! update to latest kernel and drivers could cause issue on plasma wayland [resolved]

Thanks its been great so far!

So, I had the problem as well. Thanks to @joekamprad 's instructions, I was able to boot into my Plasma desktop again. Then I checked eos-update and saw another update to tnvidia:

extra/nvidia-dkms              560.35.03-4  560.35.03-5    0.00 MiB      49.07 MiB
extra/nvidia-utils             560.35.03-4  560.35.03-5    0.00 MiB     233.79 MiB

after installing that, I undid the changes just to test and everything worked normally. I guess this was a fix?

yea can be the case i added //News: in initial post :wink:

1 Like

thanks @joekamprad for all the updates and pinning all the easy instructions up top for everyone! Saved me a whole lot of hassle - I went from broken desktop to opening this forum on my cell phone and back to desktop working again in just 5 minutes because of you :slight_smile:

Good to get such feedback, as we often have the feeling no one gets the info, but it seems that persistence and always repeating that you get the information here when something important happens bears fruit!

And i try my best to catch issues and report solutions as fast as possibleā€¦ still have the grub apocalypse in my neck :beers:

2 Likes

I also just got the updates for nvidia-dkms and nvidia-utils. I left the kernel parameter intact and it boots and loads to the desktop with Wayland. No issues. This is using grub also.

1 Like

Thanks for the instructions. I opened /etc/kernel/cmdline in nano and didnā€™t know which or where to add the nvidia_drm.fbdev=1.

If you have installed EOS recently they have ditched Grub2 in favor of systemd-boot
and that is what I have with 3 PCā€™s so itā€™s not Grub related but nvidia related.
My most used PC I initially thought my video card had taken a crap but when I switched to LTS it worked fine.

THIS IS A GREAT EXAMPLE TO USE THE MAIN KERNEL BUT ALSO INSTALL LTS!!!

(( I never edited my kernel thingy I just waited for the nvidia update that happened today. )))

AGAIN WITHOUT INSTALLING THE LTS I WOULD HAVE BEEN VERY FRUSTRATED.

Thanks Joekamprad and Stibbons for your tutorials.

Cool beans yo, saved me heaps of time.

Is that so? I thought you can choose during installation.

I guess you can never really paid attention. I like new and cutting edge and thatā€™s what EOS is aiming for so I ditched Grub2

Thanks

1 Like

Yeah so you dropped it not EOS. That is what i asked. I guess you can never really paid attention to what i wrote.

1 Like

An update was released. I had this problem yesterday. The update seemed to have fixed this.

It was an Nvidia DKMS update

Fix for systemd-boot bootloader:
You would need to fire up a terminal (e.g. Konsole) and edit the file /etc/kernel/cmdline ā€¦

Interesting. Is this the preferred place? The system I am typing this from has it configured (manually) in /boot/loader/entries/archlinux-core-main.conf.

What are the advantages of putting it in /etc/kernel/cmdline? Split between optional and mandatory parts?

Off-topic: Is nvme_load=YES advantageous if the system boots ithout it? Seems to be a relic from 10 years ago when SSDs were not that widespread as today.

I read that the kernel renamed one API call in 6.11, and Nvidia had this fixed already, but not backported to the current driver. Risks of being bleeding edgeā€¦

nvidia_drm.fbdev=1 maybe labeled as experimental, but it seems to be beneficial to have it. Gut feeling after reading about this option, even before this happened.

I also had this problem earlier this week.
Today after udpating my system simply doesnt boot anymore.

Error Message: Timed out waiting for device /dev/gpt-auto-root

Before that: dev-gpt/x2dauto/x2droot.device/start timed out.

Could this be linked to the nvidia problem?

The last udpate contained a nvidia update

/etc/kernel/cmdline is from kernel-install afaik but there are many waysā€¦ this is only the most simple.

The nvme option will only effect systems needing itā€¦ so it does not hurtā€¦ you can remove if it hurts you indeed. :wink:

ā€¦
EndeavourOS does never ditched grubā€¦ we added systemd-boot and setting it as default selected only
ā€¦

Not 200% sure but from what I read yesterday it seems fbdev and dem modeset options are now build in the Nvidia modules directlyā€¦ would be great.

I have the opposite. But not a black screen and I can go to tty.

Nvidia GTX 1070/PCIe/SSE2

On the LTS kernel 6.6.52-1-lts it stopped loading.
Error during loading, hang on ā€˜reached target Graphical Interface OKā€™.
Error in the SDDM Log start error ā€˜Failed to read display number from pipeā€™
At the same time, the Kernel is 6.11.1-arch1-1
Starts normally.

It seems that the default Kernel version is the one that broke. And it doesnā€™t matter what version it is.

Iā€™m not sure that the error is with Kernel 6.11
My system was on 6.6.52-1-lts and the error I described above.
And on 6.11.1-arch1-1 the start is normal.
So, it is not a fact that the patch 6.11.1 here https://gitlab.archlinux.org/archlinux/packaging/packages/nvidia-utils/-/blob/main/make-modeset-fbdev-default.patch?ref_type=heads will solve the problem

What do you mean you canā€™t do above? Is your bootloader grub or systemd-boot?

Oh no, I thought it was a mistake. Because nothing changed.
As a result, the team updated GRUB

I updated via tty 6.6.52-1-lts and looked via the ā€˜Eā€™ key, it changed these Nvidia settings.

But it still didnā€™t help. It also doesnā€™t start on 6.6.52-1-lts

only on 6.11.1-arch1-1

But you have to select its start via an additional menu.

Use ā€œnvidia-instā€ for the driver installer right?

If you have ran nvidia-inst once, just run sudo pacman -Syu or yay.