[Gusty] Broken Nvidia driver after upgrade to Gusty (7.10)

Bug #138933 reported by jdm64
6
Affects Status Importance Assigned to Milestone
linux-restricted-modules-2.6.22 (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

Binary package hint: nvidia-glx

I just upgraded to Gusty from Fiesty a few days ago. Now, the Nvidia driver doesn't work (it worked in Fiesty)! In the restricted-manager I selected "nvidia-glx" (I have a Nvidia GeForce4 MX 440) and restarted, but the X server couldn't load and instead dropped me into fail-safe mode.

I've also tried nvidia-glx-new and that doesn't work.

Here's the relevant part of Xorg.log:
(II) Setting vga for screen 0.
(**) NVIDIA(0): Depth 24, (--) framebuffer bpp 32
(==) NVIDIA(0): RGB weight 888
(==) NVIDIA(0): Default visual is TrueColor
(==) NVIDIA(0): Using gamma correction (1.0, 1.0, 1.0)
(**) NVIDIA(0): Option "NoLogo" "True"
(**) NVIDIA(0): Option "AddARGBGLXVisuals" "True"
(**) NVIDIA(0): Enabling RENDER acceleration
(II) NVIDIA(0): Support for GLX with the Damage and Composite X extensions is
(II) NVIDIA(0): enabled.
(EE) NVIDIA(0): Failed to load the NVIDIA kernel module!
(EE) NVIDIA(0): *** Aborting ***
(II) UnloadModule: "nvidia"
(II) UnloadModule: "fb"
(EE) Screen(s) found, but none have a usable configuration.

Fatal server error:
no screens found

Revision history for this message
jdm64 (jdm64) wrote :
Revision history for this message
jdm64 (jdm64) wrote :
Revision history for this message
jdm64 (jdm64) wrote :
Revision history for this message
jdm64 (jdm64) wrote :
Revision history for this message
jdm64 (jdm64) wrote :
Revision history for this message
jdm64 (jdm64) wrote :
Revision history for this message
jdm64 (jdm64) wrote :
Revision history for this message
jdm64 (jdm64) wrote :

Xorg still works with the nv driver.

I did notice _very_ interesting lines in dmesg when using the nvidia driver:

[ 43.196323] nvidia: module license 'NVIDIA' taints kernel.
[ 43.270935] NVRM: The NVIDIA GeForce4 MX 440 GPU installed in this system is
[ 43.270941] NVRM: supported through the NVIDIA 1.0-96xx Legacy drivers. Please
[ 43.270944] NVRM: visit http://www.nvidia.com/object/unix.html for more
[ 43.270946] NVRM: information. The 100.14.11 NVIDIA driver will ignore
[ 43.270949] NVRM: this GPU. Continuing probe...
[ 43.270959] NVRM: No NVIDIA graphics adapter found!
[ 43.473389] input: PC Speaker as /class/input/input3
[ 43.769181] ACPI: PCI Interrupt 0000:01:00.0[A] -> GSI 16 (level, low) -> IRQ 18
[ 43.769905] NVRM: loading NVIDIA Linux x86 Kernel Module 1.0-7185 Mon Apr 2 18:29:54 PDT 2007
...
[ 83.517925] NVRM: RM/client version mismatch!!
[ 83.517933] NVRM: aborting to avoid catastrophe!
[ 85.057583] Failure registering capabilities with primary security module.
[ 87.641704] NVRM: RM/client version mismatch!!
[ 87.641712] NVRM: aborting to avoid catastrophe!
[ 91.758352] NVRM: RM/client version mismatch!!
[ 91.758361] NVRM: aborting to avoid catastrophe!

It seems like nvidia-glx-new is installed, but it isn't! Did somebody miss-package nvidia-glx with the nvidia-glx-new drivers? nvidia-glx should have the 1.0-96xx drivers not the 100.14.11 drivers!

justin@ubuntu:~$ sudo apt-get install nvidia-glx
Reading package lists... Done
Building dependency tree
Reading state information... Done
nvidia-glx is already the newest version.
0 upgraded, 0 newly installed, 0 to remove and 1 not upgraded.
justin@ubuntu:~$ sudo apt-get remove nvidia-glx-new
Reading package lists... Done
Building dependency tree
Reading state information... Done
Package nvidia-glx-new is not installed, so not removed
0 upgraded, 0 newly installed, 0 to remove and 1 not upgraded.
justin@ubuntu:~$ sudo apt-get remove nvidia-glx-legacy
Reading package lists... Done
Building dependency tree
Reading state information... Done
Package nvidia-glx-legacy is not installed, so not removed
0 upgraded, 0 newly installed, 0 to remove and 1 not upgraded.

jdm64 (jdm64)
Changed in linux-restricted-modules-2.6.22:
status: New → Confirmed
Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :
Revision history for this message
jdm64 (jdm64) wrote :

It seems the problem was two fold:

There was a hidden in /lib/linux-restricted-modules/ that confused Xorg byt thinking it had the wrong driver version.
I ran: sudo rm /lib/linux-restricted-modules/.nvidia* and it fixed that problem.

The second problem was that GLX and Direct Rendering wouldn't load. That was caused by the XGL package still being installed.
I removed all XGL packages and the problem was solved.

I suggest that the update script should check for both of these problems, as it will probably be very common for people upgrading from Feisty to Gusty!

Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

jdm64:
Since bugs can only be about one single issue which one do you want this bug to be about? The hidden dotfile issue is Bug #106217.

When you are using XGL direct rendering will not be available (which pretty much makes sense given the way that XGL works). Just because you direct rendering is off doesn't mean your GL won't be accelerated though (it often means this but the XGL case is different). It is not clear exactly what GL problems you were having under XGL from your comment so you will have elaborate a bit more if you feel the bug is about that issue.

Revision history for this message
jdm64 (jdm64) wrote :

I guess I wasn't clear enough in my last message. I already solved my problem - sorry about that.

The problem was three fold:
1. The hidden .nvidia* file: causing the nvidia driver not to load because it thought it had the wrong version. Removing the file fixed the problem.
2. XGL: was conflicting with AIGLX and direct rendering. Removing all XGL packages fix the problem.
3. Xorg.conf: was improperly configured (most likely configured for XGL), and was probably causing GL problems.

What I was trying to say is that many people will probably have the same or similar problems upgrading from Fiesty to Gusty. If they had installed XGL (like I did to try out Beryl), then it will conflict with Gusty's AIGLX and there will be no 3D acceleration. Also if they have a Nvidia card, then the hidden .nvidia file could cause problems with loading the correct driver.

The update-manager or the nvidia-glx* packages or some other script should fix the hidden .nvidia problem.
Also a message should be shown if the user had installed XGL; informing them that it will conflict with AIGLX and is no longer necessary and should be removed.

Other than that you can close this bug. It's more turned into a suggestion of how to stop this problem from happening to others that upgrade.

unggnu (unggnu)
Changed in linux-restricted-modules-2.6.22:
status: Confirmed → Invalid
Revision history for this message
Sitsofe Wheeler (sitsofe) wrote :

(It's actually possible to run XGL on AIGLX - they don't actually conflict. However there are longish reasons why you might prefer one to the other. A common XGL problem on Gutsy for people who are upgrading with XGL already installed is Bug #136598 where XGL is started twice thus breaking it)

Revision history for this message
James Cole (intangi) wrote :

Just want to add to this:

I triggered this bug by installing nvidia-glx-new, which didn't work for a specific card (too old), and then removing that and installing nvidia-glx. It looks like the hidden file /lib/linux-restricted-modules/.nvidia* caused the old 7XXX series kernel driver to be used instead of the 9XXX series after nvidia-glx-new was removed.

The fix is to just make the removal scripts for nvidia-glx-new make sure to remove the hidden file @ /lib/linux-restricted-modules/.nvidia*

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.