Can't boot (again)

This seems to happen every time there is a kernel and Nvidia drivers update.

I am running (and liking) EOS on my Nvidia/Intel hybrid graphics laptop. Last time it happened I was advised to choose X11 instead of Wayland at login. That advice worked and it has been fine since (actually the error then was freezing, but it also would only boot from the LTS kernel).

There was an update to both kernel and Nvidia graphics today, and now it won’t boot.

At startup, it hangs on 2 errors: A start job is running for dracut pre-udev hook, and A start job to mount the system disk (which is a NVME).

This happens with main and LTS kernels.

Any help much appreciated :+1:

There’s a couple guides you can reference to help restore your boot.

Repair a non-booting GRUB

Once you have access to your system via arch-chroot, I’d be interested to see what the output of this is:

yay -Q | grep -E 'linux|nvidia'

Hi Bink, thanks a lot. I’ll certainly do that a bit later.

At the Grub screen, I chose the ‘intramfs’ option and booted. To my surprise, it worked. I am now running on kernel 6.19.11-arch1-1, apparently OK (it’s only been a few minutes). I don’t know what I have done bu choosing that option, but I’m happyit works, at least for now so I am reluctant to reboot for a while. I sure you understand :wink: .

Once I do reboot, I’ll try & remember to get the information you asked about.

Best,

E-G

It’s good to hear you’re back on your feet for now.

You can run the below command quite safely though, it’ll simply list what packages you have installed that include the phrases linux or nvidia. It may help us spot an issue… or not, but it’s something to check.

yay -Q | grep -E 'linux|nvidia'

Fair enough. Here it is my friend:

yay -Q | grep -E ‘linux|nvidia’
archlinux-keyring 20260323-1
libva-nvidia-driver 0.0.16-1
linux 6.19.11.arch1-1
linux-api-headers 6.19-1
linux-firmware 20260309-1
linux-firmware-amdgpu 20260309-1
linux-firmware-atheros 20260309-1
linux-firmware-broadcom 20260309-1
linux-firmware-cirrus 20260309-1
linux-firmware-intel 20260309-1
linux-firmware-mediatek 20260309-1
linux-firmware-nvidia 20260309-1
linux-firmware-other 20260309-1
linux-firmware-radeon 20260309-1
linux-firmware-realtek 20260309-1
linux-firmware-whence 20260309-1
linux-headers 6.19.11.arch1-1
linux-lts 6.18.21-1
linux-lts-headers 6.18.21-1
nvidia-open 595.58.03-3
nvidia-open-lts 1:595.58.03-3
nvidia-settings 595.58.03-1
nvidia-utils 595.58.03-1
opencl-nvidia 595.58.03-1
util-linux 2.42-1
util-linux-libs 2.42-1

I hope you know what to look for, because I don’t.

E-G

Nothing stands out as particularly unusual.

I’d only note that sometimes using the nvidia-open-dkms package instead of the nvidia-open and nvidia-open-lts can fix some issues.

The -dkms package will integrate itself into any installed kernel, that also has its respective headers installed. You already have the the -headers installed for both the linux and linux-lts kernels.

If you’d like to give that a try, now or at a later date, this should do the trick:

yay -Syu nvidia-open-dkms

It will ask you if you’d like to remove the nvidia-open and nvidia-open-lts drivers which it will replace, and you should say yes to that.

OK. But I would appreciate knowing how to reverse that if it turns out to result in not being able to boot again. Is that a possible scenario?

No problem. You’d re-install the non-dkms drivers (per kernel), and allow them to replace nvidia-open-dkms.

yay -Syu nvidia-open nvidia-open-lts

I actually think the -dkms driver generally, has less issues because it’s able to cater to any kernel update. It does however have the downside that the integration step takes a bit of time during the update, and that will be necessary any time either the Nvidia driver, or the kernel, is updated.

Thanks for all the help. I’ve installed the dkms modules and it took a little while, as expected. In the morning, when I boot, I’ll see where we are.

E-G

Hi again, I installed the dkms versions with your command, thanks, and when I booted this morning went back to the normal boot. I tried Wayland at login, but was immediately freezing again. So, another reboot, back to X11 and it’s just a normal system, running nicely.

SO, thanks for you help once again, and I’ll make his as fixed. I’ll update again if there are any developments.

I’m glad to hear you’re back in.

With respect to Wayland, did you per chance, enable HDR at some point?

With respect to Wayland, did you per chance, enable HDR at some point?

Not intentionally. I’ll look in the settings. Should I leave it off?

I only ask because when I enabled it on Nvidia hardware, it broke Wayland in a similar way to what you’ve described. That was many months ago though, and I’m not sure if HDR is still causing that issue. I’ve simply left it alone since.

Have a look here at @joekamprad 's solution, for how to disable Wayland’s HDR from your X11 session.

In kwinoutputconfig.json, I have “highDynamicRange”: false,

I assume this disables it?

Mike

Yeah I’d assume so. It’s defined per display, so you could check it with this command just to make sure it doesn’t show up again differently later in the file (the numbers represent line-number).

cat -n ~/.config/kwinoutputconfig.json | grep highDynamicRange

I’m not sure what you might look for in ~/.config/kwinrc, but it was a factor in fixing the issue.