Video-nvidia-470xx fails on kernel 6.0

Optimus Manager has to load the nvidia module anyway in order to make Hybrid Mode work, but in my case that fails. What driver are you using?

Oh, shoot… Didn’t realized this is about nvidia 470. I’m on the newest version on testing branch, so I should have not spoke up here… :frowning:

Just updated linux60-nvidia-470xx to 470.141.03-6. Same error.

Can you, please, share with us the contents of the make.log file (it’s in a subfolder in the /var/lib/dkms/nvidia/470.141.03 folder).
That would be the most interesting bit of information - it would allow us to see what happens during build.

@SoftExpert I use the pre-compiled module included in Manjaro.

Here’s a couple of related threads from the Nvidia forum. Apparently someone managed to make it work by applying a few more patches.

In the meantime version 470.141.10 is out.

@BunchOfAtoms : thank you for keeping us up to date - not always simple to have the latest information available.
My system is quite old (about 10 years) but it does not have an iGPU (or it is switched off in hardware); I have managed to have it working with the whole 6.0.x series of kernels and once I identified the required switches to enable in the xorg.conf file, I don’t even have an issue with the backlight (likely to become an issue again with the upcoming 6.1.x series of kernel).
Once the patch discussed here 470xx drivers and linux 6.0 kernel was included in the AUR package, DKMS was able to build the nvidia module every time I updated the kernel.
Obviously, the issue lies with a set of switches you would need to set in order to have the correct module loaded when booting. There is also a matter of having the appropriate kernel headers installed when you want to build the nvidia module through dkms.
I would say also that, in my experience, the pre-compiled modules might not always match your specific configuration - they are more of a convenience.
Exchanging lots of outputs here allows us all to better understand how your specific context reacts to the approaches we try - in the end we will finally determine which is the best set of parameters for your system and the overall knowledge about configuring this cheeky nvidia driver will increase.

1 Like

I uninstalled the Manjaro pre-compiled module and compiled the driver from the AUR package including these patches (0009, 0020, 0021, 0022, 0023), but I still have the same error. I guess I’ll wait for a new driver.

@BunchOfAtoms does the nvidia module gets built, at least ? it’s easy to check, just look into /var/lib/dkms/nvidia - do you see a directory bearing the name of your kernel version ?
On my machine I have something like:

|-470.141.03
   |---5.19.17-1-MANJARO
   |-----x86_64
   |-------log
   |-------module
   |---6.0.5-x64v2-xanmod1
   |-----x86_64
   |-------log
   |-------module
   |---6.0.6-2-MANJARO
   |-----x86_64
   |-------log
   |-------module
   |---6.0.6-x64v1-xanmod1
   |-----x86_64
   |-------log
   |-------module

I also have signaled the dkms driver as out-of-date, probably in the next few days the maintainer will find the time to take a look …

I managed to make it work by removing the “kernel-6.0.patch”, which is applied by the default PKGBUILD of the AUR package, while applying the “Debian patches” (0009, 0020, 0021, 0022, 0023). Now the problem is that Manjaro wants me to update the AUR package from 470.141.03-2 to a 470.141.03-3 version from the “Official Repositories (extra)”. This is probably an easy issue to solve, but I am not an expert when it comes to package management. Should I add the package to some ignore list?