Fedora 39 Post Installation Steps

Steps to perform after a fresh Fedora 39 installation.

This page covers the necessary steps after installing a fresh Fedora 39 distribution. In particular, it deals with the cannot-wake-from-suspension issue caused by NVIDIA drivers.

In the following steps, we assume an initial system update has been performed, either via Terminal or GNOME Software.

DNF Configuration  

Edit the file /etc/dnf/dnf.conf and append the following contents to it.

## Added for speed
fastestmirror=True
max_parallel_downloads=10
keepcache=True

Enable RPM Fusion  

Run the following commands.

sudo dnf install https://mirrors.rpmfusion.org/free/fedora/rpmfusion-free-release-$(rpm -E %fedora).noarch.rpm https://mirrors.rpmfusion.org/nonfree/fedora/rpmfusion-nonfree-release-$(rpm -E %fedora).noarch.rpm
sudo dnf groupupdate core

Add Flathub  

flatpak remote-add --if-not-exists flathub https://dl.flathub.org/repo/flathub.flatpakrepo

Change Hostname  

sudo hostnamectl set-hostname fedora-desktop

Install NVIDIA Proprietary Drivers  

For a fresh installation, if we go to Settings --> About --> System Details, we can see that the “Graphics” field shows “NV137”, which is the open-source nouveau driver for NVIDIA cards. This driver doesn’t work well and we need to replace it with NVIDIA proprietary drivers.

The following commands assume that RPM Fusion has been enabled.

sudo dnf update
sudo dnf install kernel-devel kernel-headers gcc make dkms acpid libglvnd-glx libglvnd-opengl libglvnd-devel pkgconfig
sudo dnf install akmod-nvidia xorg-x11-drv-nvidia-cuda

The above installs the NVIDIA proprietary driver.

The following solves the cannot-wake-from-suspension issue (see RPM Fusion HOWTO NVIDIA#Suspend). Use them only if the problem exists.

sudo dnf install xorg-x11-drv-nvidia-power
sudo systemctl enable nvidia-{suspend,resume,hibernate}

Use the following command to check if the driver has finished building. When installing for the first time, the command should output the version of the driver such as 545.29.06 if yes and modinfo: ERROR: Module nvidia not found otherwise (see RPM Fusion HOWTO NVIDIA#Install). When upgrading, it should output the upgraded version number instead of the existing one. The build process usually takes about five minutes.

modinfo -F version nvidia

After reboot, Settings --> About --> System Details should show something like “NVIDIA GeForce GTX 1050”, which means proprietary driver is in use. Waking up from suspension also works, for both Wayland and X11.

NVIDIA driver is a custom built kernel module. Custom built kernel modules only work with a specific kernel version, so they have to be built again every time the kernel is upgraded. Maybe this rebuilt is done properly by a distro via its package manager after kernel upgrade, or maybe not. Since we have the fallback NVIDIA driver nouveau blacklisted on both Debian (automatically) and Fedora (manually), it might happen that display is not working after a kernel upgrade. In this case, we should try Ctrl + Alt + F1/F2 to enter a terminal session with the otherwise functioning OS and simply rebuilding the kernel module via CLI.

There are tools that automatically rebuild kernel modules after kernel upgrade, such as Dynamic Kernel Module Support. RPM Fusion has a similar tool Kmods2, which enables shipping precompiled kernel modules for the latest kernels released by Fedora as RPM packages. There’s also the Akmods tool that automate the whole process. Therefore, if NVIDIA proprietary drivers are installed on Fedora via RPM Fusion (i.e. using the steps above), it should continue to work after kernel upgrade without users’ manual intervention. This is a good reason NOT to install drivers by downloading binaries from vendor website directly.

Install Software  

VS Code  

We install an RPM repository and install code from the repo.

Install a Microsoft RPM repo.

sudo rpm --import https://packages.microsoft.com/keys/microsoft.asc
sudo sh -c 'echo -e "[code]\nname=Visual Studio Code\nbaseurl=https://packages.microsoft.com/yumrepos/vscode\nenabled=1\ngpgcheck=1\ngpgkey=https://packages.microsoft.com/keys/microsoft.asc" > /etc/yum.repos.d/vscode.repo'

Install the code package.

dnf check-update
sudo dnf install code # or code-insiders

Dash-to-Dock  

In GNOME Software, install the RPM version of “Tweaks” and the Flathub version of “Extensions”. Install “Dash-to-Dock” by:

sudo dnf install gnome-shell-extension-dash-to-dock

Open “Extensions” app and configure Dash-to-Dock feature.

DejaVu Sans Mono Font  

In GNOME Software, search “DejaVu Sans Mono” and install the regular and bold fonts. Then run:

sudo fc-cache -v

Configure Dual Boot  

If Fedora 39 was installed after Debian 12, Fedora’s bootloader lists Debian, but not the other way around. To add Fedora to Debian’s bootloader, do the following on Debian.

• Edit /etc/default/grub. Uncomment the line with the following content.

  #GRUB_DISABLE_OS_PROBER=false

• Run the following command.

  sudo update-grub

  Output should include a line like Found Fedora Linux 39 (Workstation Edition) on /dev/sda4.

• Change boot sequence in BIOS setting. Move Debian in front of Fedora. Or sudo dnf install efibootmgr and use efibootmgr to do the same thing ( reference).

Blacklist Kernel Modules  

When dual booting Fedora 39 from the Debian 12 GRUB, the Wireless adapter driver ath9k reports error and NVIDIA drivers somehow couldn’t be detected. Fedora is still able to boot, but reverts to using nouveau as the graphics card driver. We can blacklist the ath9k and nouveau kernel modules to deal with this.

  Tip

nouveau is automatically blacklisted after installing NVIDIA proprietary drivers on Debian.

To find the kernel module corresponding to the Wireless adapter:

lspci -k

To blacklist the ath9k kernel module, create a file /etc/modprobe.d/ath9k-blacklist.conf as root and put the following contents in it:

## There is something wrong with this driver for Wireless adapter.
## It causes problem when Fedora is booted from Debian's GRUB.
blacklist ath9k

To blacklist nouveau, create the file /etc/modprobe.d/nouveau-blacklist.conf with the following content:

## Try blacklisting nouveau to see if it solves the NVIDIA issue.
blacklist nouveau

Then run the following to regenerate initramfs.

sudo dracut -f

  Warning

dracut is NOT the default tool to create initramfs on Debian. Checkout Debian blacklisting kernel modules for how to do that on Debian.

After reboot, use the following command to verify that the two modules have been blacklisted.

modprobe --showconfig | grep blacklist

Comments on Kernel Taints  

After the operations above, when booting Fedora 39, we still get the following messages on the start-up screen:

kernel: nvidia: loading out-of-tree module taints kernel.
kernel: nvidia: module license 'NVIDIA' taints kernel.
kernel: nvidia: module verification failed: signature and/or required key missing - tainting kernel
kernel: nvidia: module license taints kernel.
kernel: nvidia-nvlink: Nvlink Core is being initialized, major device number 235
kernel: nvidia 0000:01:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=io+mem
kernel: nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint.
kernel: nvidia-uvm: Loaded the UVM driver, major device number 511.
kernel: nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms  545.29.06  Thu Nov 16 01:47:29 UTC 2023
kernel: [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
kernel: [drm] Initialized nvidia-drm 0.0.0 20160202 for 0000:01:00.0 on minor 2
kernel: nvidia 0000:01:00.0: vgaarb: deactivate vga console
kernel: fbcon: nvidia-drmdrmfb (fb0) is primary device
kernel: nvidia 0000:01:00.0: [drm] fb0: nvidia-drmdrmfb frame buffer device

These are the logs form systemd-journald.service. They can be viewed by the command:

sudo journalctl -k | grep nvidia

The -k option shows kernel messages only. We can grep by any regex.

These are just warning messages. The lines starting with “nvidia-modeset” indicates that the NVIDIA driver is working properly. Fedora boots more slowly than Debian because they use different initrd schemas.

Messages that contain “taints kernel” means NVIDIA drivers’ license and “closed-sourceness” makes it impossible to properly troubleshoot some kernel problems, hence “taint”. “Out-of-tree” means NVIDIA driver source code is not part of Linux kernel source code. “License” is self-explanatory. “Required key missing” is due to not signing the NVIDIA driver, a custom build kernel module, but if Secure Boot is disabled the driver still works. See tainted kernel for more information.

  Note

If we didn’t blacklist nouveau, the kernel taints will cause Fedora to fall back to using it.

Comments on Using Dual Monitor  

On the back of my PC there are four video output connectors: one HDMI from the CPU integrated graphics card; one HDMI, one DVI and one DisplayPort from the NVIDIA GeForce GTX 1050 graphics card.

When using two monitors connected to the two HDMI ports, both Debian 12 and Fedora 39 show that the integrated and the NVIDIA graphics cards are in use together. To make both monitors use the NVIDIA card only, the solution is to plug both to the NVIDIA card — one via HDMI and one via DVI is OK. Sometimes the solution is not in software configurations but in hardware arrangements.

Interestingly, when only one monitor was plugged to the NVIDIA card’s HDMI port, both distros were able to display with either graphics card.