Hi,
The rpmfusion nvidia driver packages install all the required dependencies needed to build and run.
If gnome-software is used it should offer to do the mok key enrolment for secure boot.
For CLI, I normally do the mok key enrolment before I install the nvidia driver.
The better solution would be Nvidia provides drivers through kernel.org. Linux developers are a scarce resource. Making it easier for Nvidia to avoid providing drivers through kernel.org is unfair to other GPU makers who do work with kernel.org.
I recently heard that there’s now an agreement and NVidia drivers would indeed be in a future kernel – was someone hallucinating?
Nvidia drivers are already in the kernel: nouveau has been in there for a long time; it just doesn’t work as well as proprietary one. What you are talking about is probably Nova.
This hasn’t been my experience. I’ve been using Fedora for almost two weeks now, and the only time I had to use the terminal was to run a couple of commands that another user suggested to help debug an issue I was having with accents. Otherwise, I haven’t felt the need to use the console at all.
It’s true that enabling this feature will require some coding. However, it’s not necessary that I do it myself (nor you, for that matter). Others might be willing to take it on once there’s consensus that enabling a graphical installation method for the drivers brings more benefits than drawbacks.
Unfortunately, we cannot control what Nvidia chooses to do. By not making it easier to install the already available drivers, we are hurting users like me—not Nvidia. It’s like taking poison and expecting the other person to suffer; it’s just not going to happen.
Hi David,
Well … welcome to Fedora 
and Linux in general …
I have been using Linux in various flavors for about 30 years, and UNIX longer than that …
The main thing about both of these OS is that the command line is pretty much king … you will find, as you gain more experience with Linux, that you can do things with/on the command line that there is no replacement for in a GUI style, point-and-click interface, like begin able to fix what broke the GUI desktop for example
Now, as for Nvidia, and frankly pretty much any graphics card mfgr, the manufacturer of the device is usually the one that decides to make their product easier for the end-user to install and use their product. Nvidia and AMD(used to be ATI) have a long history of picking and choosing what they will support, M$ OS’s being their preferred due to market share. The fact that Intel is pretty much easily supported is due to Intel being a major Linux contributor and therefore you rarely have to do anything to get Intel based display drivers to just work …
Also, the fact that there are those in the Linux/Unix community that are willing to get those drivers from Nvidia, test them, package them and find a way to distribute them to the community says a lot about the people that make up that community … one of the reasons I picked Fedora by the way …
Now as for Nvidia, IF you go to their official website, you will find that Nvidia IS slowly coming to the realization that the Linux community is getting fairly large AND wants to use their product AND would love to have a better experience for the install of their supporting drivers. So, in the mean time, we work with what we have and encourage people that want a better experience to contribute to the effort any way they can … including helping others to get up and running
So, David, what have “You” done to make it easier to get someone up and running lately?
Hi @einer,
Thank you for your welcome 
Wow, that’s a long time. It’s an honor to be talking to someone as committed as you.
I don’t doubt that the command line is more versatile. However, for a regular user like me, most of the time it is totally unnecessary. I just use the basic functions: browsing the web, chatting, video conferencing, and perhaps gaming a bit (though lately, less and less). As mentioned before, except (for) (a) (few) (hiccups), I didn’t have to deal with the underlying software structure. I don’t know if that is a good or a bad thing (perhaps worth it of another thread, as many youth today grew up with systems with almost no rabbit holes to the world of software), but the thing is that you guys did such a good job creating this system that now fedora “Just Works”
For profit corporations do whatever they believe that will bring them profit. They do not care about open source or the four freedoms, they care about profit. And yes, as you say, if an OS has a bigger market share, they will take care of that market because it is aligned with their interest.
There are some points to consider: (a) there are official Nvidia drivers for Linux (closed source, but still); (b) making it easier to install them could increase Fedora’s user base, at least of users who want a simple install process (debatable if that is the kind of users that the community wants to attract or not); (c) currently other proprietary software programs, like Slack or Discord, are available for anyone to freely and easily install from the Plasma Discover “app store”.
In light of those circumstances, allowing for an easy installation of the proprietary Nvidia drivers seems more like a political decision. Nothing against that, in the end each community has to find its ethos to maximize well-being and growth, but my user experience is being collaterally damaged in that war of principles that you guys are waging against Nvidia.
What is the end goal of this conflict besides wanting Nvidia to improve their already existing Linux drivers? That they open source their drivers? And if that is so important, why to allow proprietary software to run at all? Should each person or company have the freedom to decide how they publish and run their software on the Fedora platform?
Here you will need to help me understand because I’m a bit lost. What does it say about the Fedora community to not provide a graphical installation method of the proprietary Nvidia drivers?
Well, I started this thread which it might help other users like me understand why they are forced by the community to have a worse user experience than they potentially could. And that, if you ask me, is not an easy task
Here you will need to help me understand because I’m a bit lost. What does it say about the Fedora community to not provide a graphical installation method of the proprietary Nvidia drivers?
Fedora does as much as possible via the nouveau drivers where Nvidia is concerned.
Also, Fedora avoids patent and copyright infringement like the plague therefore any proprietary driver/software that is not explicitly granted/public domain is intentionally excluded from Fedora’s distribution media.
Well, I started this thread which it might help other users like me understand why they are forced by the community to have a worse user experience than they potentially could.
That’s a question for the providers/mfgrs of the specific device/software
Coming back to this point, it sounds like as of Fedora 41, GUI installation of the nVidia drivers from the rpmfusion repos (including support for Secure Boot) should work in GNOME.
Fedora change proposal (accepted in 41): Changes/NvidiaInstallationWithSecureboot - Fedora Project Wiki
Does this actually work? Usually on the forum I see people recommending the command-line install even to GNOME users.
You have to install it first
After that you need to enable it
which starts the key enrolment process
And you can succesfully do all that through the GUI with no command line?
(I don’t use GNOME, so never tried it myself.)
FTR I haven’t used gnome since Fedora-15.
The F42 cinnamon spin release has switched from dnfdragora to gnome-software.
That’s wonderful news! I hope they make it available to KDE as well.
As you saw, it is possible to do it as they have done it in Gnome. The drivers will not be distributed in the base installation, however it should be possible to install them easily, just as other proprietary software like Zoom, Discord, Slack, etc. There is no reason to discriminate against Nvidia in that regard.
I tried this procedure on a temporary Gnome Workstation, and as far as I can see, the mok keys are not created until after the akmod modules has been built for the nvidia kernel modules. Therefore, these modules will not be signed. On the next kernel update, the kernel modules will be signed. Or you can force rebuilding the akmods modules, which requires a terminal interaction.
I don’t doubt that the concensus and desire to have such a tool exists.
It does seem to be a legal quagmire since it is not even at present considered appropriate to have the rpmfusion-nonfree-nvidia-drivers repo automatically enabled at installation time. Enabling that repo requires an explicit action by the user. Once that repo is enabled (often done during the first boot setup) then installing the driver is also a one step user action.
It does not seem a matter of lack of interest nor willingness to do so but rather the legal distribution rights for the nvidia driver itself. Those legal restrictions are put into place by nvidia and the linux community must comply with the restrictions in place at the point of distribution.
Fedora is based in the US and must meet all legal restrictions here. Other distros are based in other jurisdictions and often have differing rights or limitations.
The mok keys are automatically created by
If not already installed before the nvidia drivers are installed from rpmfusion then the akmods package is installed as a dependency at that time and additional steps are required to recreate the drivers with the signature included.
It is, however, always mandatory that the user follow the instructions in the /usr/share/doc/akmods/README.secureboot file to import that key into the bios before the keys can be used to verify the modules
Not quite.
When upgrading the kernel, the mok key will be generated before building the module, unless the key already exists.
When installing akmod-nvidia, the mok key isn’t generated before building the module. The difference is whether the build process is started by systemd or if it is started using the nohup command.
This is probably very rarely tested, as once the key is generated, you can’t repeat the test without carefully removing the generated keys.
I’m confused about this. You say that there are restrictions to the distribution of the driver, however here we are talking about making the driver available through the GUI installers. From a previous message by @pg-tips it seems that the proprietary Nvidia drivers are already available through the Gnome software installer:
How To Test
1. Open GNOME Software.
2. Search for "nvidia".
3. Choose the Nvidia driver, click Install and follow the prompts.
4. Reboot and enroll the self-signing key in the mokutil tool following <<the documentation will be added>>
5. The OS should boot up with the Nvidia driver enabled.
As far as I understand, the only thing missing is to implement the same method on KDE Fedora’s Discover. Did I get this right?
There are developer needed to implement, manpower to maintain this whole procedure inkl. testing the wide spectrum of hardware available.
On the other hand there are users missing which want to take a financial responsibility for that. Just demanding as you do!
As you can see Nvidia since long makes trouble, even Torwalds told them what he thinks of them