WARNING!. Problems with the incoming update - kernel 6.1.0-18 & nvidia driver

[ListLess]

I have discovered that you must have the 'contrib' and 'non-free' repositories enabled in sources.list.

[otello.an]

Only regarding the 470 tesla driver:
Same problem with gnome online accounts (already well known and fixed in previous versions)
https://gitlab.gnome.org/GNOME/gnome-on ... issues/252
is proposed again in this version ... (once a serious bug is closed, another less serious one opens)

[qbwdp]

When this issue started i went to backports, in my sources i added:
deb http://deb.debian.org/debian bookworm-backports main

Then i installed:
6.5.0-0.deb12.4-amd64

Rebooted and nvidia driver worked fine. Now that i read things are fixed, I took out the backports and added:
deb http://ftp.us.debian.org/debian bookworm-proposed-updates main contrib non-free non-free-firmware
I did:
apt update; apt upgrade; apt dist-upgrade;

It installed 6.1 but as secondary and 6.5 is still default. When i choose 6.1.0-18-amd64 at boot up, i don't get nvidia to start up.
in dpkg i see a lot of 525.147.05-7 stuff installed.
ii nvidia-driver 525.147.05-7~deb12u1 amd64 NVIDIA metapackage
ii firmware-nvidia-gsp 525.147.05-7~deb12u1 amd64 NVIDIA GSP firmware

What do i need to do to get back to 6.1* kernel and get nvidia driver to work?

[porygon-beam10]

Thank you! proposed updates saved my headaches..
Since last June I had been holding out on upgrading because of my nvidia card 980Ti, then Oct came and went... Then I figured everything's got to be stable upon stable since debian is one of thee most stable and reliable upgrades. SO, thank you for the people or person who fixed this in the proposed updates and for the comments in this thread.

--Debian user since Sarge on ppc

[appye]

Hi All,

If you simply enabled the 'proposed-updates' repository and ran your upgrade to fix the dkms issue, you didn't just receive the updated Nvidia driver; you also upgraded to other packages within that repository as well. While most were likely stable, without APT pinning to selectively upgrade only the nvidia packages, you risked encountering unforeseen issues due to upgrading to packages not yet ready for general use. This is basically what proposed-updates is for. If all proposed packages were stable at the time of upgrade, you're likely safe. However, if any pose unexpected consequences overlooked by maintainers, you may have to address them.

Disabling proposed updates after applying the fix would have been wise, and unless you are interested in testing packages, you should certainly do so now. Once disabled, it is PROBABLY safe to resume stable channel upgrades until your system aligns with the state of proposed-updates at the time of your upgrade, but there is no way of knowing if it is DEFINITELY safe without reviewing what else was done. Food for thought.

On another note, I have been using snapshots for this sort of thing, timeshift in my case, and only manual, but there are other options. When I want to make a permanent change (install a new program, edit or add a config file, etc) to a machine, I revert to a previous snapshot just to make sure things are "clean," make my change, do a full-upgrade and some basic cleanup and create another snapshot. When dealing with this issue, this workflow allowed me to play around with the various fixes I have seen on this forum in environments that were functionally isolated from one another. I bring this up because it allowed me to be a bit more carefree with running the 'proposed-updates' fix (without pinning, but I did disable immediately) with the understanding that the fix should make it into stable. On another machine, I went with the backported 6.5 kernel for no reason in particular. Today, due to snapshotting, both machines are 100% back on the stable branch and functionally never experienced this issue.

[mlauronen]

When this issue started i went to backports, in my sources i added:
deb http://deb.debian.org/debian bookworm-backports main

Then i installed:
6.5.0-0.deb12.4-amd64

I also tried this, but for some reason the wine program I am using started to work clearly slower with it. I am using Wine version 9.2.

In version 6.1.17 CPU consumption was 0.3%, but in version 6.5 it jumped to 0.7%. The download speed from the internet also slowed down significantly. So the version of Wine I'm using doesn't like the backport kernel. This is not a backport issue, it's the kernel itself, because kernel version 6.7 did the same problem.

So those who use wine programs should check the performance.

[Path_unwalked]

So, I have the same update problem that is described here, the error messages are like in the first reply by roseway. My knowledge of the package manager is unfortunately not advanced enough to know this. Do I still have to wait until the update is out in my region? Or is there something I have to do so apt-get update realises the packages changed? When I call update, it only seems to pull from the repository I added for the sake of Lutris:

Code: Select all

Holen:1 http://download.opensuse.org/repositories/home:/strycore/Debian_Testing  InRelease [1.512 B]
OK:2 http://deb.debian.org/debian bookworm InRelease                           
OK:3 http://security.debian.org/debian-security bookworm-security InRelease    
OK:4 https://dl.winehq.org/wine-builds/debian bookworm InRelease

Then, when I upgrade, I get
The following packages were automatically installed and are no longer needed:

Code: Select all

  linux-headers-6.1.0-11-amd64 linux-headers-6.1.0-11-common
  linux-image-6.1.0-11-amd64

Then I get these errors

Code: Select all

/etc/kernel/postinst.d/dkms:
dkms: running auto installation service for kernel 6.1.0-18-amd64.
Sign command: /usr/lib/linux-kbuild-6.1/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub

Building module:
Cleaning build area...
env NV_VERBOSE=1 make -j32 modules KERNEL_UNAME=6.1.0-18-amd64......(bad exit status: 2)
Error! Bad return status for module build on kernel: 6.1.0-18-amd64 (x86_64)
Consult /var/lib/dkms/nvidia-current/525.147.05/build/make.log for more information.
Error! One or more modules failed to install during autoinstall.
Refer to previous errors for more information.
dkms: autoinstall for kernel: 6.1.0-18-amd64 failed!
run-parts: /etc/kernel/postinst.d/dkms exited with return code 11
dpkg: Fehler beim Bearbeiten des Paketes linux-image-6.1.0-18-amd64 (--configure):
 »installiertes post-installation-Skript des Paketes linux-image-6.1.0-18-amd64«-Unterprozess gab den Fehlerwert 1 zurück
linux-headers-6.1.0-18-amd64 (6.1.76-1) wird eingerichtet ...
/etc/kernel/header_postinst.d/dkms:
dkms: running auto installation service for kernel 6.1.0-18-amd64.
Sign command: /usr/lib/linux-kbuild-6.1/scripts/sign-file
Signing key: /var/lib/dkms/mok.key
Public certificate (MOK): /var/lib/dkms/mok.pub

Building module:
Cleaning build area...
env NV_VERBOSE=1 make -j32 modules KERNEL_UNAME=6.1.0-18-amd64......(bad exit status: 2)
Error! Bad return status for module build on kernel: 6.1.0-18-amd64 (x86_64)
Consult /var/lib/dkms/nvidia-current/525.147.05/build/make.log for more information.
Error! One or more modules failed to install during autoinstall.
Refer to previous errors for more information.
dkms: autoinstall for kernel: 6.1.0-18-amd64 failed!
run-parts: /etc/kernel/header_postinst.d/dkms exited with return code 11
Failed to process /etc/kernel/header_postinst.d at /var/lib/dpkg/info/linux-headers-6.1.0-18-amd64.postinst line 11.
dpkg: Fehler beim Bearbeiten des Paketes linux-headers-6.1.0-18-amd64 (--configure):
 »installiertes post-installation-Skript des Paketes linux-headers-6.1.0-18-amd64«-Unterprozess gab den Fehlerwert 1 zurück
dpkg: Abhängigkeitsprobleme verhindern Konfiguration von linux-image-amd64:
 linux-image-amd64 hängt ab von linux-image-6.1.0-18-amd64 (= 6.1.76-1); aber:
  Paket linux-image-6.1.0-18-amd64 ist noch nicht konfiguriert.

dpkg: Fehler beim Bearbeiten des Paketes linux-image-amd64 (--configure):
 Abhängigkeitsprobleme - verbleibt unkonfiguriert
dpkg: Abhängigkeitsprobleme verhindern Konfiguration von linux-headers-amd64:
 linux-headers-amd64 hängt ab von linux-headers-6.1.0-18-amd64 (= 6.1.76-1); aber:
  Paket linux-headers-6.1.0-18-amd64 ist noch nicht konfiguriert.

dpkg: Fehler beim Bearbeiten des Paketes linux-headers-amd64 (--configure):
 Abhängigkeitsprobleme - verbleibt unkonfiguriert
Fehler traten auf beim Bearbeiten von:
 linux-image-6.1.0-18-amd64
 linux-headers-6.1.0-18-amd64
 linux-image-amd64
 linux-headers-amd64
E: Sub-process /usr/bin/dpkg returned an error code (1)

Is there anything I can do?

[sunrat]

Is there anything I can do?

The nvidia-driver problem is fixed with a new version in the bookworm-updates repository. Make sure you have added it to sources:

Code: Select all

deb https://deb.debian.org/debian/ bookworm-updates non-free contrib main

and do

Code: Select all

apt update && apt upgrade

IIRC, bookworm-updates should be included in a default install but for some reason it is not in your sources you posted above.

[qbwdp]

Code: Select all

apt update && apt upgrade

I do have bookwork-updates already there from the install. and i did update and upgrade.

Code: Select all

 
deb http://deb.debian.org/debian/ bookworm-updates main non-free-firmware contrib non-free
deb-src http://deb.debian.org/debian/ bookworm-updates main non-free-firmware contrib non-free
 

I booted into 6.1.0-18-amd64 and was dropped at console. here are the errors i collected with journalctl:

Code: Select all

eb 21 06:53:38 hdpdts01 systemd-modules-load[587]: Error running install command 'modprobe nvidia-modeset ; modprobe -i nvidia-current-drm ' for module nvidia_drm: retcode 1
Feb 21 06:53:38 hdpdts01 systemd-modules-load[587]: Failed to insert module 'nvidia_drm': Invalid argument
Feb 21 06:53:38 hdpdts01 (udev-worker)[653]: Error running install command 'modprobe -i nvidia-current ' for module nvidia: retcode 1
Feb 21 06:53:39 hdpdts01 systemd-modules-load[1088]: Error running install command 'modprobe nvidia-modeset ; modprobe -i nvidia-current-drm ' for module nvidia_drm: retcode 1
Feb 21 06:53:39 hdpdts01 systemd-modules-load[1088]: Failed to insert module 'nvidia_drm': Invalid argument
Feb 21 06:53:40 hdpdts01 nvidia-persistenced[1216]: Failed to query NVIDIA devices. Please ensure that the NVIDIA device files (/dev/nvidia*) exist, and that user 118 has read and write permissions for those files.
Feb 21 06:53:40 hdpdts01 systemd[1]: Failed to start nvidia-persistenced.service - NVIDIA Persistence Daemon.
Subject: A start job for unit nvidia-persistenced.service has failed
 A start job for unit nvidia-persistenced.service has finished with a failure.

After this i did:

Code: Select all

apt remove nvidia-driver
apt purge nvidia-driver
apt install nvidia-driver

I did not see initramfs being updated during the install of the drivers. So i did:

Code: Select all

update-initramfs -k 6.1.0-18-amd64 -u

Rebooted into 6.1.0-18 and same, it drops me to console with same message that it failed to start nvidia-persistenced.service. I had to go back to 6.5.0-0. I have not installed the driver straight from nvidia. Should i do that as the next step?

[None1975]

Should i do that as the next step?

You can remove all packages on your system with nvidia in the name by running:

Code: Select all

apt purge "*nvidia*"

And then reboot the system with:

Code: Select all

systemctl reboot

If it seems to still be having issues, you may also try running:

Code: Select all

apt install --reinstall xserver-xorg-core xserver-xorg-video-nouveau

or

Code: Select all

X -configure

[randomguy32]

I have an old nvidia gpu (only using it for games as I have an on board intel card for the rest) and I am using the nvidia 470 tesla driver for it.

I am unable to upgrade to the 6.1.0-18 kernel. Has this been already fixed or do I need to do something special for the upgrade to run sucessfully?

[qbwdp]

If it seems to still be having issues, you may also try running:

Code: Select all

apt install --reinstall xserver-xorg-core xserver-xorg-video-nouveau

or

Code: Select all

X -configure

Thank you for the information. I did the following to solve the issue:

Code: Select all

# dkms status
nvidia-current/525.147.05, 6.5.0-0.deb12.4-amd64, x86_64: installed

So after booting into 6.1.0.18 kernel, i did dkms autoinstall and then got:

Code: Select all

# dkms status
nvidia-current/525.147.05, 6.1.0-18-amd64, x86_64: installed
nvidia-current/525.147.05, 6.5.0-0.deb12.4-amd64, x86_64: installed
# uname -r
6.1.0-18-amd64

I am back to a stable kernel and X is working. Thank you for all the help.

[randomguy32]

I have an old nvidia gpu (only using it for games as I have an on board intel card for the rest) and I am using the nvidia 470 tesla driver for it.

I am unable to upgrade to the 6.1.0-18 kernel. Has this been already fixed or do I need to do something special for the upgrade to run sucessfully?

My update on this issue. Initially I used the proposed-updates update channel to get the nvidia to work on 6.1.0-18. Though afterwards I found out that the kernel 6.1.0-18 also breaks my cheap chinese ps2 dualshock adapter (detected as ps3 controller).

What I did was to completely uninstall the nvidia drivers, install the kernel from backports (6.5) and its headers and then install the nvidia-tesla-470-driver again...

Now everything is working, the nvidia drivers and the cheap chinese ps2 dualshock adapter...

Though this is pretty frustrating as both the nvidia card and the ps2 controller were working properly before 6.1.0-18. I liked the debian 12 on launch but the latest updates have been disappointed, as they have not been stable for me.

[appye]

... Initially I used the proposed-updates update channel to get the nvidia to work on 6.1.0-18. Though afterwards I found out that the kernel 6.1.0-18 also breaks my cheap chinese ps2 dualshock adapter (detected as ps3 controller).

... pretty frustrating as both the nvidia card and the ps2 controller were working properly before 6.1.0-18. I liked the debian 12 on launch but the latest updates have been disappointed, as they have not been stable for me.

I worry others who just added proposed-updates and went on with their day may be in the same boat, even if they disabled it immediately afterwards:

If you simply enabled the 'proposed-updates' repository and ran your upgrade to fix the dkms issue, you didn't just receive the updated Nvidia driver; you also upgraded to other packages within that repository as well. While most were likely stable, without APT pinning to selectively upgrade only the nvidia packages, you risked encountering unforeseen issues due to upgrading to packages not yet ready for general use. This is basically what proposed-updates is for. If all proposed packages were stable at the time of upgrade, you're likely safe. However, if any pose unexpected consequences overlooked by maintainers, you may have to address them.

While you may be correct in that your issue is kernel related, you may want to see if you have other packages installed from proposed updates and see if any of these are related to your issue. That repo is for developers and testers and comes with a risk of unexpected breakage. Also, I don't know how updates to backports packages are handled, but you may not get timely security updates with your 6.5 kernel.

[Hugo Pavh]

Regarding to this error.

After doing a clean installation of Debian 12.5 / kernel 6.1.0-18 (Linux 6.1.76) and waiting for the bug fix to appear in bookworm-updates, I installed the nvidia driver (specifically nvidia-tesla-470) and no problem and everything worked well!.

However, yesterday I downloaded the Linux kernel 6.1.79 from 'kernel.org' to compile it specifically for my machine; When I tried to install the newly compiled vanilla kernel, it gave me the same error that the kernel 6.1.0-18 threw when trying to update it from 6.1.0-17/Debian 12.4, and because of this I couldn't install that recently compiled kernel.

This makes me think that this error isn't really corrected, and that in the next Debian's kernel update, the same problem seen in Debian 12.4 with the nvidia driver will occur again.

Has anyone else facing the same problem or, is it just an isolated problem with my machine?


Regards!

[debian1]

I am using nvidia-current/530.30.02, 6.1.0-15-amd64 due to requiring CUDA. Any time frame when it will be fixed for this version?

I noticed Package nvidia-kernel-545.23.08 is not installed. cannot be installed as well.

Guessing I need to wait for the fix. I can boot using older kernal 6.1.0.15 without any issues.

I installed 545 and it broke nvidia. If you are on 530. Don't try to update yet.

I purged nvidia and reboot, and trying to install it from the normal repo and it won't install.

Error! Bad return status for module build on kernel: 6.1.0-18-amd64 (x86_64)

If you did what I did and messed it up badly, you can still remove the kernel + nvidia and then install the nvidia + CUDA and it works with the old kernel still. Guide on how to remove kernel viewtopic.php?p=792736#p792736

[donald]

Just popping in to note that this thread has over 15,000 views. 3 pages, all tech, no arguing.

So proud of you all.

[debian1]

Same issue here. Probably need to wait for an update for the nvidia driver which seems incompatible. For now, if you have upgraded to Debian 12.5, you can purge the kernel 6.1.0-18 and continue to use the 6.1.0-17 kernel if you use the nvidia driver. Do not run autoremove!

Code: Select all

apt purge linux-image-amd64 linux-headers-amd64 linux-image-6.1.0-18-amd64 linux-headers-6.1.0-18-amd64

Reinstall linux-image-amd64 and linux-headers-amd64 when a fix is available.

Or you can just boot 6.1.0-17 kernel from GRUB until the fix is in the repo. I only did the purge as it was blocking install of other packages.

I edited the title to show the issue is with nvidia driver.

A fix is on the way. See viewtopic.php?p=792730#p792730

Thank you this. I used it to remove 6.1.0.18 so I can reinstall nvidia + CUDA and everything worked fine.

[appye]

Just popping in to note that this thread has over 15,000 views. 3 pages, all tech, no arguing.

So proud of you all.

Ha! Such a good poppa. This thread has had a bit of snark though ...

[donald]

Just popping in to note that this thread has over 15,000 views. 3 pages, all tech, no arguing.

So proud of you all.

Ha! Such a good poppa. This thread has had a bit of snark though ...

haha

You missed the other thread that got locked by another admin, it looked like the original Thunderdome movie. I actually don't mind people being passionate about their views and defense of FLOSS and Debian and to be honest one can actually learn a lot from those threads from the old heads popping in and correcting nearly everyone. But it got to be ... a bit much.