After booting you may be able to use ssh and manage the system from the other one upstairs. Then you would have a working desktop that could be used to copy & paste directly from the terminal window.
It appears that you are not using uefi boot so secure boot is not a factor.
Your kernel is very old – there have been at least 4 kernel updates since the 6.9.11 kernel. Fedora is now on kernel 6.10.6.
I would suggest that you do a full update so that you may take advantage of the many fixes (some related to this exact problem) that have been done with all the updates between your last update and now. sudo dnf upgrade --refresh
Once the update completes then wait at least 5 minutes before rebooting.
Once the reboot completes let us know the result.
Thank you!! I really wish I could report back that all is well, but sadly it is still just a black screen with the pointer. I had high hopes though. Should I go through the commands again? I would assume they would be about the same, just listing a newer kernel. But maybe not. I really appreciate all the help! Thank you! Any other suggestions or ideas?
TLDR, It’s been a while since I last used RPMFusion’s repos for Nvidia, so I can’t paste here what is installed on my system to help. I think this time is not related to RPMFusion’s packages, since I don’t use it and got black screen twice with kernel 6.10.x when updating to 555.58.02 (non-beta) and now to 560.35.03 (non-beta).
Check which nvidia driver packages are installed. doas dnf history userinstalled | grep -E "nvidia"
Try the “nuclear” installation. Try installing these before uninstalling all packages you installed for nvidia and reinstalling: doas dnf install akmod-nvidia xorg-x11-drv-nvidia-cuda xorg-x11-drv-nvidia-libs.{x86_64,i686} xorg-x11-drv-nvidia-cuda-libs.{x86_64,i686} nvidia-settings nvidia-xconfig xorg-x11-drv-nvidia-power nvidia-gpu-firmware nvidia-modprobe
Monitor what’s being done via journcalctl, akmod takes some time in the background building the modules:
When using GRUB, your grub.cfg must have the needed parameters for it to work correctly. You can check if the installed kernels have this two rd.driver.blacklist=nouveau modprobe.blacklist=nouveau with: doas grubby --info ALL
If not add them manually doas grubby --update-kernel=ALL --args='rd.driver.blacklist=nouveau modprobe.blacklist=nouveau nvidia-drm.modeset=1'
and update grub.cfg file doas grub2-mkconfig -o /boot/grub2/grub.cfg
You can also run dracut to make sure. For the current kernel your booted to: doas dracut --force --printsize --parallel --verbose
Or for all installed ones. doas dracut --force --printsize --parallel --verbose --regenerate-all
After typing in the install command, I got: . is already installed
Dependencies resolved.
Nothing to do.
All kernels are confirmed with blacklist=nouveau and nvidia-drm.modeset=1.
I am sorry I am so difficult here! Thank you again for your time!
Peter
PS. Just curious, is there an easy way to build another kernel WITH nouveau without boinking the rest of the system? Creating then two Grub entries, one booting (or trying to) with the nvidia kernel and another entry booting from a nouveau kernel? Still need to get nvidia working again, but an idea to get it working again in the meantime without all the graphics support of nvidia.
With the drivers installed it is still very easy to boot using nouveau.
From the grub menu during boot press the e key to edit the grub commands.
There should be a line beginning with linux and in that line should be the 2 blacklist options already noted. rd.driver.blacklist=nouveau modprobe.blacklist=nouveau
Find those 2 options and delete them from that line then continue booting. It should now load the nouveau driver instead of the nvidia driver for testing purposes. This change only affects the current boot but is an ideal way to test.
Your list of commands does reflect how long it has been since you last installed the nvidia drivers from rpmfusion.
Not that anything is wrong per se, but definitely overkill. dnf list installed \*nvidia\* shows the installed packages. dnf install akmod-nvidia xorg-x11-drv-nvidia-cuda installs all the required packages as dependencies without needing to explicitly state each one.
Installing akmod-nvidia places the noted kernel options into the /etc/default/grub file and updates grub to include those options into the kernel command line. Thus the use of grub2-mkconfig is not required. Manual updating with dracut is not required. – In fact, the akmods process explicitly excludes the nvidia drivers from the initramfs image and they are not loaded until the system switches to the full os from the boot image.
This would be the list with the driver and cuda packages. I had a rahide installed on a external drive and checked there. I’ll install Fedora 40 there and install RPMFusion’s there to check how it’s going too.
No there is not much time since I last tried, just has been long since I last used on my main system. That extensive command is the one I prefer to use tried 1-2 weeks when installing on a externa SSD with Fedora rawhide.
Greetings again and THANK YOU again for all your time and help!!! So, no I have not yet tried uninstalling all the nvidia stuff, is it save or even possible to use an * like uninstall nvidia or wouldn’t that work so well? I did run through the list to install and everything said it was already installed, nothing to do. So not sure about the dev version of the xorg driver, but on the list it said everything was already installed so … a bit confused by this. Also it was working fine until a recent update - not sure if it was actually a Fedora update though, could have been a certain package update.
To that end, some more news that may help. Removing the blacklist=nouveau lines STILL results in the black screen with the pointer after login. The login is still graphical, but once logged in it goes to the black screen. So it’s obviously something graphical, but perhaps that may help narrow a focus or something. On that computer I also have a Debian install and the GRUB in use kinda goes back and forth depending on … not sure but from time to time it changes. I always keep both up to date. Currently it is using the GRUB from Debian, the entry is:
insmod part_msdos
insmod btrfs
insmod ext2
search --no-floppy --fs-uuid --set=root e09f7118-34ed-49ba-82f4-74b481b8a674
linux /boot/vmlinuz-6.10.6-200.fc40.x86_64 root=/dev/nvme0n1p1 nvidia-drm.modeset=1 rd.driver.blacklist=nouveau modprobe.blacklist=nouveau
initrd /boot/initramfs-6.10.6-200.fc40.x86_64.img
If I switch to the Fedora GRUB, the entry is:
insmod part_msdos
insmod btrfs
insmod ext2
search --no-floppy --fs-uuid --set=root e09f7118-34ed-49ba-82f4-74b481b8a674
linux /boot/vmlinuz-6.10.6-200.fc40.x86_64 root=UUID=e09f7118-34ed-49ba-82f4-74b481b8a674 ro rd.driver.blacklist=nouveau modprobe.blacklist=nouveau resume=UUID=51d3ae35-28fe-4185-b0b7-db9f5f15d7e0 rhgb quiet nvidia-drm.modeset=1
initrd /boot/initramfs-6.10.6-200.fc40.x86_64.img
Interesting, if I boot from the Fedora GRUB the boot just seems to hang indefinitely on a screen with the 3 large “dots” - probably a better name for it but I just don’t know it, sorry about that!
So, is there a quick way to uninstall ALL the nvidia packages to reinstall, does the info about the black screen after removing the blacklist=nouveau options change anything, do the actual GRUB entries tell anyone anything, if I am missing a xorg-x11 package how should I install it as all previous install commands returned that they were already installed, and if it is installed but not showing up in the list could that be part of the problem?
Whew, okay, wow, thank you guys for sticking with me and all your help!
If uninstalling everything NVIDIA and reinstalling doesn’t get it working, there is the option I use that is the Negativo17’s repos. Two options will be available there with akmods and with dkms. Just remember to remove the ones installed first and disable RPMFusion.
sudo dnf config-manager --add-repo=https://negativo17.org/repos/fedora-nvidia.repo
# once again a nuclear one liner
# You can choose dkms-nvidia or akmod-nvidia
sudo dnf install dkms-nvidia nvidia-driver nvidia-modprobe nvidia-driver-libs.{i686,x86_64} libnvidia-ml.{x86_64,i686} nvidia-libXNVCtrl nvidia-xconfig nvidia-settings nvidia-driver-NvFBCOpenGL.{i686,x86_64} nvidia-driver-cuda nvidia-driver-cuda-libs.{x86_64,i686} nvidia-gpu-firmware nvidia-settings
Combining two commands sudo dnf remove $(doas dnf history userinstalled | grep nvidia)
will uninstall everything. BUT, take your time to observe/check what is being listed to be removed before confirming. Sometimes DNF likes to remove kernel packages together with NVIDIA drivers.
you should remove rd.driver.blacklist=nouveau, modprobe.blacklist=nouveau and nvidia-drm.modeset=1
The ones provided here on the Current GeForce/Quadro/Tesla in the RPMFusions HowTos should install the required ones, which I, personally, have intermittent success.
If the sudo dnf history userinstalled | grep nvidia did not show the files, they, somehow, were not installed. To be sure, adding “kmod” to the filter to check if akmod files are there and if the kmod files are genereted for each of the kernel versions installed (not necessary for all, but at least for the ones booted after installing the drivers). sudo dnf history userinstalled | grep -E "nvidia|kmod"
Are you on Fedora Workstation or another spin?
Make sure to be booting from Fedora’s one by entering your motherborad’s UEFI’s Boot Selection Menu, where should show all bootable entries. They should be named.
Just a brief semi-pointless update. We are now completing our move and I will be away from my computer until October, at which time i will get back to it and try the most recent suggestions. I THANK YOU all for all your help and time!!! I am sorry for this interruption here but all will be back to normal in October, or shortly thereafter as I am sure there will be a lot of work waiting for me, hahaha!! Thank you again, I will post back again when I can try things again. I hope you all have a fantastic September!