DU today with kernel upgrade brakes Nvidia-driver

[charlyheinz]

The DU today with kernel-upgrade to 5.19.3-1-siduction brakes the Nvidia-driver 510.73.08-01 from exterimental. Is anyone having the same issue or Is there a new driver coming?

[der_bud]

... Nvidia-driver 510.73.08-01 from experimental...

As of https://packages.debian.org/search?keywords=nvidia%2Ddriver , nvidia-driver in debian experimental is 510.85.02-1, try that one

[dibl]

Not only Nvidia -- it failed to start sddm on my Intel graphics as well.  Here is the relevant hardware:

Code Auswählen Erweitern

Graphics:
  Device-1: Intel TigerLake-H GT1 [UHD Graphics] driver: i915 v: kernel
  Device-2: NVIDIA GA104M [GeForce RTX 3070 Mobile / Max-Q] driver: nvidia
    v: 510.85.02
  Device-3: Acer BisonCam NB Pro type: USB driver: uvcvideo
  Display: x11 server: X.Org v: 1.21.1.4 with: Xwayland v: 22.1.3 driver:
    X: loaded: modesetting,nvidia unloaded: fbdev,nouveau,vesa gpu: i915
    resolution: 1920x1080~144Hz
  OpenGL: renderer: Mesa Intel UHD Graphics (TGL GT1) v: 4.6 Mesa
    22.2.0-rc3

It boots with the Intel GPU.  I spent two days trying to figure out why boot ended with these lines:

Code Auswählen Erweitern

...
sddm Failed to read display number from pipe
sddm Could not start Display server on vt 7

Google hits suggested the kernel was trying to launch sddm before the CPU had settled. More google pointed me to /etc/systemd/system/display-manager.service.

To insert a little delay in the launch of sddm, I changed this default section:

Code Auswählen Erweitern

[Service]
# temporary safety check until all DMs are converted to correct
# display-manager.service symlink handling
ExecStart=/usr/bin/sddm
Restart=always
RestartSec=1s
EnvironmentFile=-/etc/default/locale

to:

Code Auswählen Erweitern

[Service]
# temporary safety check until all DMs are converted to correct
# display-manager.service symlink handling
ExecStartPre=/bin/sleep 5
ExecStart=/usr/bin/sddm
Restart=always
RestartSec=1s
EnvironmentFile=-/etc/default/locale

Now it automatically starts sddm once again.

I hope this helps someone. 

[michaa7]

Here this kernel +sddm + nvidia 470 boots and works as it should:

$ uname -r
5.19.3-1-siduction-amd64

~$ apt policy nvidia-driver
nvidia-driver:
  Installiert:           470.141.03-1
  Installationskandidat: 470.141.03-1
  Versionstabelle:
     510.85.02-1 1
          1 http://neutower:3142/deb.debian.org/debian experimental/non-free amd64 Packages
*** 470.141.03-1 500
        500 http://neutower:3142/deb.debian.org/debian unstable/non-free amd64 Packages
        100 /var/lib/dpkg/status
     470.129.06-6~deb11u1 500
        500 http://neutower:3142/deb.debian.org/debian stable/non-free amd64 Packages

~$ apt policy sddm
sddm:
  Installiert:           0.19.0-3
  Installationskandidat: 0.19.0-3
  Versionstabelle:
*** 0.19.0-3 500
        500 http://neutower:3142/deb.debian.org/debian unstable/main amd64 Packages
        500 http://neutower:3142/deb.debian.org/debian stable/main amd64 Packages
        100 /var/lib/dpkg/status

YMMV

[dibl]

Of my four siduction installations, the problem only exists on the one with NVMe SSD.

[charlyheinz]

After editing the sourcelist entry for the experimental Repo as "der_Bud" has told to I could install the new Nvidia-driver. This driver is working as expected. Thx to "der_Bud"

uname -r
5.19.3-1-siduction-amd64

apt policy nvidia-driver
nvidia-driver:
  Installiert:           510.85.02-1
  Installationskandidat: 510.85.02-1

[terroreek]

Ok, I am having this issue as well, using the experimental driver doesn't help.  I will try dibl's fix, as this has an NVME drive as well.


No that didn't fix the problem for me either.

[terroreek]

hmm this is what I see in my xorg log, I might be missing the obvious but I dont see an error thats causing it not to load.

[terroreek]

Ok I figured it out, I had to blacklist the nouveau driver.  After that xorg was loading correctly but the system was still broken, as it was clear things were not rendering correct.  I removed all the nvidia packages, then re-installed the nvidia drivers from experimental and am working correctly now. 

[meander]

On my system, the nvidia-driver update installed and set the nvidia tesla drivers. Since I don't own a tesla card, my screen became very low resolution

For me, running and setting the non tesla driver worked.

update-glx --config nvidia