It appears that kernel and kernel headers are at 6.18.3-200.fc43 for Fedora 43 KDE?? Is that correct?
Second part of the question, has anyone tried updating? I’ve got Virtualbox and Nvidia and CUDA installed. I went to the RPM Fusion site but couldn’t figure out if they have the updated to the 6.18 packages.
I’ve currently held back packages with these two files I’ve edited so it’s not currently checking or updating:
First new file: harry1@fedora:/etc/dnf/repos.override.d$ cat 99-config_manager.repo
# Generated by libdnf.
# Do not modify this file manually.
exclude=nvidia-driver,nvidia-modprobe,nvidia-persistenced,nvidia-settings,nvidia-libXNVCtrl,nvidia-xconfig
Second new file: harry1@fedora:/etc/dnf/libdnf5.conf.d$ cat 80-local.conf
Just in case anyone else is in this same configuration, I temporarily commented out my exclusions, ran Discover, updated all the 6.18 kernel-related things, and then waited 5 minutes to make sure all kernel modules had been built.
Rebooted, everything worked great…Nvidia, Virtualbox etc.
The nvidia driver from rpmfusion has no relation to the kernel version when downloaded from the repo. The driver is built locally to match each kernel as it is updated.
Stopping the updates of those packages seems likely to bite you in the future so you probably should leave the exclusions commented out permanently.
Yes, the 6.18.3 kernel was released to the updates repo (I think today).
A quick way to check if the driver is installed to match the booted kernel would be to run both uname -r to see the kernel version and dnf list --installed kmod-nvidia\* to see there is a package installed with a name that includes the kernel version. Similarly for VirtualBox.
So are you saying neither of these is necessary, I should just let the kernel update when new kernels are available?
harry1@landlord:/etc/dnf/libdnf5.conf.d$ more 80-local.conf
[main]
excludepkgs=kernel*,*nvidia*,*VirtualBox*,akmod-VirtualBox
harry1@landlord:/etc/dnf/repos.override.d$ more 99*
# Generated by dnf5 config-manager.
# Do not modify this file manually, use dnf5 config-manager instead.
[cuda-fedora42-x86_64]
exclude=nvidia-driver,nvidia-modprobe,nvidia-persistenced,nvidia-settings,nvidia-libXNVCtrl,nvidia-xconfig
What if I just don’t want kernel updates, for instance, I want to stay on the LTS 6.18 kernel until end of life?
Neither of those are necessary nor desired for the average user. Specific needs may apply.
Those exclusions of driver package names prevent upgrade of the driver packages, but do not prevent a driver being rebuilt when a kernel upgrade occurs.
IMHO the only one needed to stay with the currently installed kernel is this. excludepkgs=kernel* which will keep the existing kernel but still allow the drivers to be updated when necessary.
You may, of course, manage it as you see fit. I am just trying to avoid having others copy your setting without understanding the specific reason you have a non-standard config.
Note that a more fitting location for your edits would be to put those exclusions in /etc/dnf/dnf.conf since the ones you have edited are subject to being replaced with system updates, or to use the dnf config-manager ... command as shown in the files themselves.
Using an LTS kernel isn’t the same as not getting kernel updates. The LTS kernels continue to get security fixes. For example, release 6.12.64 recently became available.
But you can’t get 6.12.64 from the official Fedora repos. Updates for 6.12 were no longer made available in Fedora once kernel 6.13 went into the stable releases.
So, while you could set an exclusion for kernel*>=6.19, that wouldn’t really mean you were on LTS kernel 6.18. You’d just be frozen on the last version of 6.18 that Fedora released, with whatever vulnerabilities that contained never being fixed.
On the other hand, there is a COPR (unofficial build) of LTS 6.12 for Fedora, and it looks like a similar build is being prepared for 6.18. So you might want to consider that, but to stress again, that’s not an official Fedora release.