Fedora failing to boot after attempting to update to from 42 to 43

Ask Fedora
, , ,
1

Video

I was attempting to update from Fedora 42 to Fedora 43, and during the update process, my laptop restarted automatically. When I booted back into Fedora to continue the installation, I was met with a fast moving wall of text, and then a black screen with a blinking underscore in the top left corner. I am unable to proceed past this screen. The last error message that flashes on screen is [FAILED] Failed to start system-update-utmp-runlevel.service

Here’s what I’ve already tried:

  • Ran sudo akmods --rebuild --force in the TTY console and rebooted.
  • Made sure that the kernel parameter modprobe.blacklist=nouveau,nova_core was present in the GRUB menu.
  • Tried booting into an older kernel.
  • Tried removing rghb quiet from the kernel parameters, then booting.

I am using an Asus Zephyrus G16, Intel CPU, Nvidia GPU. Any help at all is appreciated!

3

That error message about utmp is pretty harmless. The issue is that either your graphics driver fails to start or that your system greeter fails to start.

Can you switch to a “normal” TTY when this happens via Ctrl-Alt-F4? You should see a Login prompt at which point you can enter your username and password, and we can have a look at the journalctl output to see what happened AFTER the message about utmp and which is resulting in your desktop going AWOL.

If you can get into a TTY, can you issue systemctl status gdm (I think it’s gdm for Gnome, it’s sddm if you’re on KDE…)

1 Like
4

The journalctl output filled the console with the same message over and over
Feb 04 02:13:46 fedora kernel: NVRM: gpuHandleSanityCheckRegReadError_GM107: Possible bad register read: addr: 0xb830b0, regvalue: 0xbadf1201, error code: NV_PPRIV_SYS_PRI_ERROR_CODE_FECS_PRI_RESET

20260207_233225
20260207_2332251920×1440 659 KB

and here’s the output for system status gdm:

20260207_233658
20260207_2336581920×2560 630 KB

Its worth noting that I have both KDE and GNOME installed but I was using gnome prior to the update. When I ran system status ssdm , it gave an error saying that the command wasn’t recognised.

5

I did find this gpuHandleSanityCheckRegReadError_GM107: Possible bad register read · Issue #688 · NVIDIA/open-gpu-kernel-modules · GitHub with the only useful suggestion that overclocking caused it for one user.
Are you overclocking?

6

I’ll guess that was a typo with the “system status ssdm” and you really did use systemctl for that command.

The black screen and no greeter is because your graphics card is not initialising correctly - greeter can’t display the login page as it had no graphics framebuffer to use as the card isn’t providing one. I also assume you were running quite happily on 42 with either secure boot disabled, or enabled and with the keys installed and everyone was happy. Nvidia drivers loaded cleanly and so on.

Check mokutil --sb-state if you are using secure boot - those keys need to be installed or we’ll be reubuilding stuff that will never get loaded.

I know you state you have already rebuilt the modules after the installation, but lets hit it with a largish hammer.

sudo dnf reinstall akmod-nvidia
sudo akmods --force --rebuild
sudo dracut --regenerate-all --force

Give it all a few minutes to run, compile everything, flush caches, then reboot.

If you get any issues or error messages, shout up.

7

20260208_002257
20260208_0022571920×1440 219 KB

mokutil --sb-state says that secure boot is enabled

sudo akmods --force --rebuild ran without any issues, but the other two didn’t. The errors are above in the picture.

Also yeah, everything was perfectly fine in Fedora 42. I’ve had issues with the nvidia drivers in the past but they were fixed!

8

Nope, I’m not overclocking. Thanks for the suggestion though.

9

Typo in my “nividia” command which you corrected! Good spot.

You stated in your origjnal post that you were in the middle of upgrading to F43but you are currently booted into an F42 kernel - did the upgrade actually take place or is there not an F43 kernel to select from the grub menu?

In the meantime, lets just rebuild dracut for this kernel you’re currently running now (6.18.8-100).

sudo dracut --kver 6.18.8-100.fc42.x86_64 --force

10

In the grub menu, the latest boot option is

Fedora Linux (6.18.8-100.fc42.x86_64) 42 (Workstation Edition)

But it says Fedora 43 at the top of the TTY console??

I ran sudo dracut --kver 6.18.8-100.fc42.x86_64 --force and it seemed to work just fine. There was no error message, but no output either.

20260208_004745
20260208_0047453000×1300 955 KB

11

OK. No output is good as it means no errors.

Try a reboot and see what we get.

Steve

12

No luck, same blank screen as before unfortunately.

13

For new installs, F43 increased the size of the boot partition to accommodate firmware “bloat”. Many Nvidia users have had to either remove the rescue kernel, expand the partition, or do a full install.

Have you checked that the boot partition isn’t full? Are you able boot an older kernel?

If your issue isn’t related to the increased size needed for the boot partition, you need to make it easy for members of the linux community using similar hardware to find this topic. https://linux-hardware.org/?computer=c04f3432712d has just one (duplicated) probe, so not widely used for Linux.

  1. post the output from running inxi -Fzxx as pre-formatted web searchable text, and
  2. ensure that vendor firmware is fully updated so a) you aren’t chasing a solved problem and b) to make it easier for others to test the same configuration. If you can boot to a terminal session or connect using ssh try to update Fedora packages as well.
14

Ok so I used a software on windows to check the partitions on my hard drive and according to that, my Linux partition is 100% full…

I did try booting an older kernel but it gave the same result.

Would it be a good idea to remove the linux swap partition in order to expand the boot partition?

image
image468×174 1.95 KB

15

In the terminal window when booted please run lsblk -f and df -h so we can see the actual used & free space in each partition. Your statement my Linux partition is 100% full is rather ambiguous since there are usually at least 3 linux partitions and we do not know which was referenced.

Windows itself does not know anything about the actual file system structure or usage of linux partitions/file systems.

16

Repos seem to all be out of date after upgrading to 43 - #4 by vgaetera

3 Likes
17

Here’s the output of lsblk -f and df -h

20260208_165323
20260208_1653231920×914 259 KB

From what I can see, it looks like the partition software I used was misleading, because nothing looks full.

18

Hey there, thanks for the suggestion. Here’s a picture of the output. Safe to say it didn’t work for me.

20260208_170645
20260208_1706451920×1440 879 KB

19

Here’s a workaround for your resolver:

sudo mv -f /etc/resolv.conf{,.bak}
sudo systemctl restart NetworkManager.service
2 Likes
21

That worked! After a reboot everything booted as normal. One strange thing I noticed is that it still says I am on Fedora 42 on the grub menu, but it says Im on Fedora 43 in the gnome settings, and there are no pending updates.

So is there anything else I should do or am I good to go?

1 Like
22

One of your screenshot shows that dracut tried wrongly to write in /boot/efi/MACHINE_ID
where MACHINE_ID is the identifier of your machine. See cat /etc/machine-id

This has been reported during some updates to 43. Those MACHINE_ID directories should be
suppressed and the last kernel re-installed.

If /boot contains the vmlinuz fc42 files (that you see in GRUB), but no .fc43 ones, try this:

sudo mv /boot/$(cat /etc/machine-id){,-OFF} # Ignore if /boot/MACHINE_ID doesn't exist
sudo mv /boot/efi/$(cat /etc/machine-id){,-OFF}
sudo dnf reinstall kernel-core-6.18.8-200.fc43
sudo akmods # By security

If no errors, reboot in 6.18.8

1 Like