Xorg and kernel versions (Nvidia)

[sbratton]

Hi I've got an issue that I've had enough of.

Running LMDE for a while Kernel 3.2 and everything loads up fine.
with recent updates kernel 3.10 and 3.11 have both been added. Somewhere after 3.10 being installed I've messed up xorg or the video drivers.
Kernel 3.10 doesn't load the GUI due to a X configuration issue. (No display/screens detected).
Kernel 3.2 does load into the GUI.

I've attached xorg log file from boot attempt on 3.10 kernel.

Latest update to 3.11 with UP8 seems to have carried forward the problem from 3.10
As the xorg.conf file is fixed I'm guessing the answer lies in the kernel / drivers.

Problem most likely arose when I was "playing about" with nouveau and nvidia legacy drivers or Xrandr.

Any help to diagnose root cause and how to fix would be gratefuly received.

Snippet from the 3.10 Kernel attempt to start X

Code: Select all

[    38.644] (EE) [drm] KMS not enabled
[    38.644] (EE) No devices detected.
[    38.644] (EE) 
Fatal server error:
[    38.644] (EE) no screens found(EE) 
[    38.644] (EE) 
Please consult the The X.Org Foundation support 
	 at http://wiki.x.org
 for help. 
[    38.644] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[    38.644] (EE) 

Thank you.
Steven

[Monsta]

Check if you have the kernel headers and kbuild installed.
Run these commands:

Code: Select all

aptitude search ~ilinux-headers

Code: Select all

aptitude search ~ikbuild

and post their output here.

If you have them installed, try reconfiguring NVidia kernel modules:

Code: Select all

sudo dpkg-reconfigure nvidia-kernel-dkms

[sbratton]

Headers were installed

Code: Select all

steven@oxygen ~ $ aptitude search ~ilinux-headers
i A linux-headers-3.10-2-amd64      - Header files for Linux 3.10-2-amd64       
i A linux-headers-3.10-2-common     - Common header files for Linux 3.10-2      
i   linux-headers-3.11-2-amd64      - Header files for Linux 3.11-2-amd64       
i   linux-headers-3.11-2-common     - Common header files for Linux 3.11-2      
i   linux-headers-3.2.0-4-amd64     - Header files for Linux 3.2.0-4-amd64      
i   linux-headers-3.2.0-4-common    - Common header files for Linux 3.2.0-4     
i   linux-headers-amd64             - Header files for Linux amd64 configuration

kbuild was installed

Code: Select all

steven@oxygen ~ $ aptitude search ~ikbuild
i A linux-kbuild-3.10               - Kbuild infrastructure for Linux 3.10      
i   linux-kbuild-3.11               - Kbuild infrastructure for Linux 3.11      
i   linux-kbuild-3.2                - Kbuild infrastructure for Linux 3.2   

Interesting nvidia-kernal-dkms was NOT installed so I installed that and rebooted.
For some reason it only installed on 3.2 and 3.11 (maybe its something to do with the "A" next to the headers file.
I then ran the reconfigure command. which uninstalled and reinstalled the modules.

Unfortunately I still get the same error about screens when booting the 3.11 kernel.
3.2 is still fine.

Any more ideas ?

Steven

[Monsta]

Interesting nvidia-kernal-dkms was NOT installed so I installed that and rebooted.
For some reason it only installed on 3.2 and 3.11

That's interesting. Were there any error messages about 3.10? Run

Code: Select all

sudo dpkg-reconfigure nvidia-kernel-dkms

once more and post the output here.

[sbratton]

Output below observed from running command from within GUI of 3.2
The 3.10 kernel is ignored completely

Code: Select all

-------- Uninstall Beginning --------
Module:  nvidia-current
Version: 319.72
Kernel:  3.11-2-amd64 (x86_64)
-------------------------------------

Status: Before uninstall, this module version was ACTIVE on this kernel.

nvidia-current.ko:
 - Uninstallation
   - Deleting from: /lib/modules/3.11-2-amd64/updates/dkms/
 - Original module
   - No original module was found for this module on this kernel.
   - Use the dkms install command to reinstall any previous module version.

depmod...........

DKMS: uninstall completed.

-------- Uninstall Beginning --------
Module:  nvidia-current
Version: 319.72
Kernel:  3.2.0-4-amd64 (x86_64)
-------------------------------------

Status: Before uninstall, this module version was ACTIVE on this kernel.

nvidia-current.ko:
 - Uninstallation
   - Deleting from: /lib/modules/3.2.0-4-amd64/updates/dkms/
 - Original module
   - No original module was found for this module on this kernel.
   - Use the dkms install command to reinstall any previous module version.

depmod......

DKMS: uninstall completed.

------------------------------
Deleting module version: 319.72
completely from the DKMS tree.
------------------------------
Done.
Loading new nvidia-current-319.72 DKMS files...
Building for 3.2.0-4-amd64 and 3.11-2-amd64
Building initial module for 3.2.0-4-amd64
Done.

nvidia-current:
Running module version sanity check.
 - Original module
   - No original module exists within this kernel
 - Installation
   - Installing to /lib/modules/3.2.0-4-amd64/updates/dkms/

depmod....

DKMS: install completed.
Building initial module for 3.11-2-amd64
Done.

nvidia-current:
Running module version sanity check.
 - Original module
   - No original module exists within this kernel
 - Installation
   - Installing to /lib/modules/3.11-2-amd64/updates/dkms/

depmod....

DKMS: install completed.

Thank you

[sbratton]

Am I using the wrong version of the Nvidia driver ?
My card is a FX5200 should I be using 173.xx or is 319.xx the latest / correct version ?
Nvidia website has 325.15 listed as the latest version (aug 2013) for my card.

Do I need to ensure Nouveau is disabled first ?
How do I check ?
How do I ensure it is not being used ?


Thanks
Steven

[Monsta]

Am I using the wrong version of the Nvidia driver ?
My card is a FX5200 should I be using 173.xx or is 319.xx the latest / correct version ?
Nvidia website has 325.15 listed as the latest version (aug 2013) for my card.

That makes sense. The description of nvidia-legacy-173xx-driver package states that it's the last version that supports your card. Maybe you should try installing these drivers.

The only thing I can't understand is how your kernel 3.2 runs fine with 319.xx series. Or maybe nouveau is loaded somehow?
Boot that kernel and post the output of:

Code: Select all

inxi -Gx

[sbratton]

Yes the 3.2 Kernel is loading nouveau

Code: Select all

Graphics:  Card: NVIDIA NV34 [GeForce FX 5200] bus-ID: 01:00.0 X.Org: 1.14.3 driver: nouveau Resolution: 1400x1050@60.0hz 
           GLX Renderer: N/A GLX Version: N/A Direct Rendering: N/A

I'm guessing the 3.10 & 3.11 are trying to load NVIDIA 319xx and for some reason it's not working.

I see my options as
1. change 3.11 Kernel to use nouveau drivers.
2. change xorg.conf to use NVIDIA 319xx or legacy-173xx, this may stop 3.2 from working
3. something else

I see option 1 as the safest way to use Kernel 3.11
and option 2 as the more complicated way but gives me nvidia drivers.

What are your thoughts on the best way forward ?

Thanks
Steven

[sbratton]

I've got somewhere.

First using the 3.2 kernel GUI package manager I removed the version 304 nvidia drivers

Then ran

Code: Select all

sudo update-alterntives --config nvidia

to select the 173 kernel (was already installed via package manager)

created a file called blacklist.conf to /etc/modprobe.d/
with one line "blacklist nouveau=1" (without quotes)

Code: Select all

nvidia-xconfig 

to configure xorg.conf ready for the nvidia driver not nouveau

Then I ran the command

Code: Select all

sudo update-initramfs -u 

The new problems I have now is that mdm doesn't start and X requires a manual start. I have a temporary fix.

Code: Select all

 
Alt+F1
login with username and password
"startx"
Then Alt+F7 to get back to the GUI

Other problems
1. Not all programmes show on the display when I start them, but this could be the last place I left my desktop before I messed it up with razor-QT.
2. Without mdm I can't get out of Razor-qt desktop to Mate.

That's for me to try an fix myself another time.

Thank you very much for you help in diagnosing the problems and getting a solution in place. It is much appreciated.

[sbratton]

I was wasting too much time faffing with display drivers so I performed a back up and system re-install this evening.

All working now with 3.2 and 3.11 kernels using the stock drivers.
As I had a separate /home partition the install went very smoothly and I didn't need to restore files/settings from the back up.

Thankyou Monsta for your help. I now know a lot more about the internal workings of driver selection and xorg.conf and some of the tools to configure driver packages.

Steven