Nvidia fail with DKMS on linux-generic-pae - needs depends on -pae kernel header file package

[Angel Guzman Maeso]

Binary package hint: xorg

I try to install linux-generic-pae for my 32-bit system that it have 4 GB of ram, but only use 3 GB with the current kernel.

When I start with the new kernel, the screen flicker quickly forever and I cannot start Ubuntu. If I try with recovery mode I get start ubuntu, but changing the Drivet to vesa on xorg.conf file.

When I install linux-generic-pae, the terminal show this fail message:
Examining /etc/kernel/postinst.d.
run-parts: executing /etc/kernel/postinst.d/dkms
 * Running DKMS auto installation service for kernel 2.6.31-14-generic-pae
 * nvidia (185.18.36)... nvidia (185.18.36): Installing module.
  Kernel headers for 2.6.31-14-generic-pae are not installed. Cannot install this module.
  Try installing linux-headers-2.6.31-14-generic-pae or equivalent.
                                                                                               [fail]
run-parts: executing /etc/kernel/postinst.d/nvidia-common

I attach the previous Xorg.0.log before change to vesa.

ProblemType: Bug
Architecture: i386
CurrentDmesg: [ 208.185070] mtrr: base(0xe5000000) is not aligned on a size(0xe00000) boundary
Date: Sat Oct 17 22:51:37 2009
DistroRelease: Ubuntu 9.10
MachineType: Gigabyte Technology Co., Ltd. P35-DS3L
Package: xorg 1:7.4+3ubuntu5
ProcCmdLine: root=UUID=db9f0ad8-2017-4e50-8c67-2d85dcf9eb0f ro single
ProcEnviron:
 LANG=es_ES.UTF-8
 SHELL=/bin/bash
ProcVersionSignature: Ubuntu 2.6.31-14.48-generic-pae
RelatedPackageVersions:
 xserver-xorg 1:7.4+3ubuntu5
 libgl1-mesa-glx 7.6.0-1ubuntu4
 libdrm2 2.4.14-1ubuntu1
 xserver-xorg-video-intel 2:2.9.0-1ubuntu2
 xserver-xorg-video-ati 1:6.12.99+git20090929.7968e1fb-0ubuntu1
SourcePackage: xorg
Uname: Linux 2.6.31-14-generic-pae i686
dmi.bios.date: 08/13/2007
dmi.bios.vendor: Award Software International, Inc.
dmi.bios.version: F4
dmi.board.name: P35-DS3L
dmi.board.vendor: Gigabyte Technology Co., Ltd.
dmi.board.version: x.x
dmi.chassis.type: 3
dmi.chassis.vendor: Gigabyte Technology Co., Ltd.
dmi.modalias: dmi:bvnAwardSoftwareInternational,Inc.:bvrF4:bd08/13/2007:svnGigabyteTechnologyCo.,Ltd.:pnP35-DS3L:pvr:rvnGigabyteTechnologyCo.,Ltd.:rnP35-DS3L:rvrx.x:cvnGigabyteTechnologyCo.,Ltd.:ct3:cvr:
dmi.product.name: P35-DS3L
dmi.sys.vendor: Gigabyte Technology Co., Ltd.
fglrx: Not loaded
system:
 distro: Ubuntu
 architecture: i686kernel: 2.6.31-14-generic-pae

[Angel Guzman Maeso (shakaran)]

Xorg.0.log before to vesa (with nvidia Driver on xorg.conf file)

[Ralf Müller (avbelow)]

i have exactly the same problem. pae-kernel is not working

[Adam Plumb (adamplumb)]

I'm running Ubuntu 9.10 RC 32bit with the PAE kernel and I have the same issue when I try to use the 185.x nvidia drivers from the repository.

[neuromancer (neuromancer)]

Have you tried to install driver with envyng rather than ubuntu hardware driver (install it from repo)?

Try to remove nvidia driver and reboot, install pae, and then install new nvidia driver with envyng.

[Patrick L Archibald (patrick-archibald)]

Using envyng to install the drivers worked for me. I did the following:

sudo apt-get install envyng-core
Ctrl-Alt-F1
sudo /etc/init.d/gdm stop
envyng -t
Took option 1 - Install the NVIDIA driver
Wait for completion and took the reboot option

[Patrick L Archibald (patrick-archibald)]

envyng installation worked but now I don't have glx support. Tried "sudo nvidia-xconfig" with a reboot but it didn't enable glx support.

[neuromancer (neuromancer)]

great!
Meanwhile is a good graphical solution :)

[Bryce Harrington (bryce)]

Btw, in the description, the error says you did not have the -pae headers installed. It is good to hear that envyng sorted the issue out, but I'd be interested in hearing if installing the proper kernel header package would have resolved it as well?

[neuromancer (neuromancer)]

Obviously in karmic 9.10 64bit version, wich doesn't need pae extension, all work right.
In a new recent installation, all ram memory was recognized (3x2Gb ram) and when I've installed also the nvidia driver (185) no problems occur.

[Bryce Harrington (bryce)]

Okay, thanks... guess we can close this bug then?

I think ultimately the issue is that specialty kernels like -pae do not get as heavily tested as the standard Ubuntu kernel, so plain old build issues on -nvidia don't always get noticed quickly. So the long term solution really is that we need to encourage -pae users to test early and often during the development cycle. I'm going to try to work out how to get the make.log included in our xorg apport script to at least simplify the reporting of these issues.

[manroh (agenthanu)]

I had the same issue. Installing the PAE kernel image with the according header files will fix this issue.
I'm running the PAE kernel now with the NVIDIA driver set.

[lunkwill (launchpad-lunkwill)]

Note related bug in jockey: https://bugs.launchpad.net/ubuntu/karmic/+source/jockey/+bug/451305

I installed linux-headers-generic-pae and did a

dpkg-reconfigure nvidia-185-kernel-source

But I got the "Module build for the currently running kernel was skipped since the kernel source for this kernel does not seem to be installed" error (even though the headers were there). I had to

apt-get remove --purge linux-headers-2.6.31-15

And then, when I did an

apt-get install linux-headers-2.6.31-15-pae

it automatically rebuilt (successfully) the nvidia.ko.

(Also, I note that when xorg fails, it restarts very very quickly and doesn't ever seem to give up, making it impossible to login to a VT, as was possible with the earlier init scheme. This should probably also be considered a significant bug.)

[lunkwill (launchpad-lunkwill)]

Update: I think that's a false alarm. I installed the 2.6.31-15 headers but the -14 kernel.

[Rich Wales (richw)]

I, too, had a similar problem (with nvidia-glx-195 on a Karmic box), and the problem cleared up after I installed linux-headers-generic-pae. I'm now successfully running 2.6.31-16-generic-pae and nvidia-glx-195 and seeing all 4 GB of RAM in my system.

[bugbot (bugbot)]

This bug report was filed against an old version of Ubuntu.
Can you confirm whether this is still an issue in natty?

If you don't mind, it would be very helpful if you could update the bug
report in launchpad to 'Fix Released' if it is no longer an issue for
you, or if it is still occurring under natty, please tag the bug 'natty'
so it's easier for us to track.

[Roland Meier (rm369)]

This is still an issue.
Since the default kernel was automatically changed to a -pae variant some time this year, every time the kernel is updated the matching -pae header is not updated along with it even though the installed proprietary nvidia driver's kernel module depends on it.
After the next reboot the X server fails to start (which was a major nuisance for me at the beginning since the text mode couldn't be displayed on the attached TV and I had to attach a monitor for debugging) failing to load nvidia.ko.
So now I have to check after every update whether the kernel image was updated and in this case to manually install the matching -pae header package, otherwise I would get a black screen after reboot again (at least now I have a textmode my TV can handle...).

Today the kernel image was updated from 3.0.0-12-generic-pae to 3.0.0-13-generic-pae, linux-headers-3.0.0-13-generic-pae was again not installed automatically. After installing it manually run-parts automatically created
/lib/modules/3.0.0-13-generic-pae/kernel/drivers/video/nvidia
/lib/modules/3.0.0-13-generic-pae/kernel/drivers/video/nvidia/nvidiafb.ko
/lib/modules/3.0.0-13-generic-pae/updates/dkms/nvidia_current_updates.ko
/lib/modules/3.0.0-13-generic-pae/updates/dkms/nvidia_current.ko
/lib/modules/3.0.0-13-generic-pae/updates/dkms/nvidia_173.ko

So yes, this is still a VERY annoying issue until today in Oneiric.
There should be a dependency making sure that the matching kernel header is installed after an update of the kernel image if the nvidia X-server is installed.