Nouveau module is not loading after upgrading to kernel 6.8.10

After upgrading to the latest kernel 6.8.10, the nouveau module is not loading; I see no output of the command lsmod | grep nouveau.

I am running Fedora 40 KDE plasma. Since I have an NVIDIA card on my system, without the module, I cannot use my external monitor. I know that for the time being, I will have to use the older kernel versions. But when will a fix arrive for this problem?

Added f40

All I can say for now is the safe thing is to load an older kernel.

there are post here already looking into the issue with 6.8.10. . .

https://discussion.fedoraproject.org/t/not-booting-after-update-to-kernel-6-8-10/117769/18

I don’t know how the Nouveau driver was blacklisted during the update process, especially since I don’t have proprietary drivers installed on my system. I booted into an older kernel and noticed that the Nouveau module was not loading.

After some web searches, I pressed ‘e’ during boot and discovered that the Nouveau driver had been blacklisted. I edited the boot options, removed the blacklist commands, and added nomodset. After booting successfully, I updated the GRUB configuration file in the /etc/default/grub directory to make these changes permanent. As a result, I can now use my external monitor again.

1 Like

I am on Workstation and not KDE, but I upgraded to the 6.8.10 kernel with the nvidia 550.78 driver and the boot worked perfectly.

$ modinfo nvidia
filename:       /lib/modules/6.8.10-300.fc40.x86_64/extra/nvidia/nvidia.ko.xz
alias:          char-major-195-*
version:        550.78
supported:      external
license:        NVIDIA
firmware:       nvidia/550.78/gsp_tu10x.bin
firmware:       nvidia/550.78/gsp_ga10x.bin
1 Like

If you didn’t have it blacklisted, then how could it have happened? :thinking:

I’m still wondering. I don’t see any process that would try to blacklist the Nouveau driver.

1 Like

6.8.10 is strange. . . Someone had a UUID disappear as well.

1 Like

Maybe that’s why some people are skeptical about using Fedora. I believe the developer team should have been choosy regarding going with a kernel update. I know there is an option to use 2 or 3 of the older kernels; however, that’s still a hassle.

1 Like

Developers can’t test even a small fraction of array of hardware people want to use (and many go to linux because vendor has ended support for recent non-free OS’s on their hardware). I used to conduct “practical sessions” in courses where students (most with PhD’s) from developing countries were expected to use a specialized linux package, so had to consider how to run it when they returned home. I suggested they install linux on an older unloved and unwanted PC that was going unused.

Fedora plays an important role by exposing recent kernels to a wide variety of hardware. There are important security benefits from recent kernels, but it is inevitable that some configurations will have problems. In my opinion, having alternate ways (e.g., Fedora on an external USB3 drive) to boot and alternate (e.g., USB WiFi or access to ethernet) are just as important as backups (backups are useless if you can’t boot the system).

I think that was just confusion over the format of entries in /var/log/boot.log-YYYYDDMM that used a hexadecimal representation (\x2d) for dashes.

It does appear that 6.8.10 is more “careful” (or “fussy”) about some details. With external USB drive used to test-drive 6.8.10, it reported an issue with the ESP system partition:

 FAT-fs (sdb1): invalid media value (0xe2)
1 Like

I see one issue that may be important when using an external drive such as this.
The initramfs image needs to contain the drivers for all the systems it will be used on, and if they are not all similar with needing the same drivers then the user needs to identify the specific devices that are missing on the default installation where the usb device was installed. Once the devices that exist on other systems have been identified then the user must modify the initramfs image and add the drivers needed to support the other systems.

Without this care it seems likely that a usb device that boots well on one system may not boot properly on a different system.

As you note above, the partition id on the device also must be interpreted correctly on the system being booted from that usb.

The USB drive should get a rescue kernel when it is created. If the installer boots, so should the rescue image. In practice, I update and test the rescue image every few months.

I boot the USB drive by adding the following to /etc/grub.d/40_custom.conf:

menuentry "Boot from USB Drive" {
    set root=UUID=<UUID>
    linux /vmlinuz root=<UUID> ro
    initrd /initrd.img
}

Booting the install USB or external USB drive hasn’t been reliable on my desktop – I think something happens before my monitor wakes up and I end up in the memory test. Legacy Apple devices have always been really easy to boot from external drives, so I’ve been booting linux on Apple systems for many years. That is how I discovered that my USB SSD is faster than either of the internal SATA drives on my old iMac.

There are lots of 128GB SSD’s being pulled from Windows PC’s advertised on eBay for under US$20, and Fedora runs well on a 128GB drive, so having a second boot drive is cheap insurance.

1 Like