[opensuse] Mismatch in nvidia drivers - opensuse 13.1
Hello, I had a working installation of opensuse 13.1 until last night's update. Since then the X server wont't start. If I run startx from the command line in superuser mode I get a message about a mismatch, where the nvidia driver component has version 304.121 while the nvidia kernel module has version 304.119. The message preceding the mismatch message indicates "Loading extension GLX". All the opensuse repos are for version 13.1. Anyone has any idea how to proceed from here? Thanks, a -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Sun, 08 Jun 2014 22:19:09 -0400 Alain Leblanc wrote:
I had a working installation of opensuse 13.1 until last night's update. Since then the X server wont't start. If I run startx from the command line in superuser mode I get a message about a mismatch, where the nvidia driver component has version 304.121 while the nvidia kernel module has version 304.119. The message preceding the mismatch message indicates "Loading extension GLX". All the opensuse repos are for version 13.1. Anyone has any idea how to proceed from here?
Hi Alain, A "working installation" can mean different things. My guess is you have been running the manually installed nVidia driver and last night's updates (applied through the package manager) have broken that installation. If you read the documentation that comes with the nVidia manual installer, you'll know that it has the ability to completely reverse the manual installation. You can then add and enable the nVidia rpm repository and use that to install the driver. The benefit of this approach is the installation will no longer be hidden from the package management system and the driver will be automatically updated during future updates if it is needed. Relevant links here: http://en.opensuse.org/SDB:NVIDIA hth & regards, Carl -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 06/08/2014 11:00 PM, Carl Hartung wrote:
On Sun, 08 Jun 2014 22:19:09 -0400 Alain Leblanc wrote:
I had a working installation of opensuse 13.1 until last night's update. Since then the X server wont't start. If I run startx from the command line in superuser mode I get a message about a mismatch, where the nvidia driver component has version 304.121 while the nvidia kernel module has version 304.119. The message preceding the mismatch message indicates "Loading extension GLX". All the opensuse repos are for version 13.1. Anyone has any idea how to proceed from here?
Hi Alain,
A "working installation" can mean different things. My guess is you have been running the manually installed nVidia driver and last night's updates (applied through the package manager) have broken that installation.
If you read the documentation that comes with the nVidia manual installer, you'll know that it has the ability to completely reverse the manual installation. You can then add and enable the nVidia rpm repository and use that to install the driver. The benefit of this approach is the installation will no longer be hidden from the package management system and the driver will be automatically updated during future updates if it is needed.
Relevant links here: http://en.opensuse.org/SDB:NVIDIA
hth & regards,
Carl Thanks Carl,
In the end all I had to do was download the file NVIDIA-Linux-x86_64-304.119.run.run from the NVIDIA web site and run it. By "working installation" I meant installed from the official DVD, only added the packman repo, did all the updates as they came. So I'm a bit surprised that this problem would show up on my machine and not others. Or maybe it's just that everyone else figured it out on their own :) Thanks again, a -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 06/09/2014 05:54 AM, Alain Leblanc wrote:
On 06/08/2014 11:00 PM, Carl Hartung wrote:
On Sun, 08 Jun 2014 22:19:09 -0400 Alain Leblanc wrote:
I had a working installation of opensuse 13.1 until last night's update. Since then the X server wont't start. If I run startx from the command line in superuser mode I get a message about a mismatch, where the nvidia driver component has version 304.121 while the nvidia kernel module has version 304.119. The message preceding the mismatch message indicates "Loading extension GLX". All the opensuse repos are for version 13.1. Anyone has any idea how to proceed from here?
Hi Alain, snip Relevant links here: http://en.opensuse.org/SDB:NVIDIA
hth & regards,
Carl Thanks Carl,
In the end all I had to do was download the file NVIDIA-Linux-x86_64-304.119.run.run from the NVIDIA web site and run it.
snip Thanks again,
a I had problems with x due to the most recent updates. I did "zypper up" and it complained of problems with the nvidia rpm's (from the suse repo). After the update, my desktop's resolution was completely messed up and I had huge icons. The system was no longer using the nvidia drivers. The yast package manager showed six nvidia rpms (some of which were not correct). I removed ALL of the nvidia rpms, rebooted the system and then the desktop resolution returned to normal. I then selected only the two appropriate drivers (g02) for my card, rebooted. After the second reboot the system was now using the drivers and the desktop was back to normal. First time anything like this happened to me. I only use the suse nvidia rpm's and never install with packages from the nvidia site. Gustav
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 09/06/14 07:43, Gustav Degreef wrote:
I had problems with x due to the most recent updates. I did "zypper up" and it complained of problems with the nvidia rpm's (from the suse repo). After the update, my desktop's resolution was completely messed up and I had huge icons. The system was no longer using the nvidia drivers. The yast package manager showed six nvidia rpms (some of which were not correct). I removed ALL of the nvidia rpms, rebooted the system and then the desktop resolution returned to normal. I then selected only the two appropriate drivers (g02) for my card, rebooted. After the second reboot the system was now using the drivers and the desktop was back to normal. First time anything like this happened to me. I only use the suse nvidia rpm's and never install with packages from the nvidia site. Gustav
I had this problem with one machine, and completely non-functional graphics on another. There seems to be an additional package in the latest update: nvidia-uvm-gfxG0x-kmp-<kernel-flavour>. On the machine with incorrect resolution, I did basically the same (remove all except the latest G03 packages) and had to install the uvm package. This resulted in the kernel being *downgraded* to 3.11.6. On the machine with no graphics at all, I had to taboo the uvm package to get a graphical display - package nvidia-gfxG03-kmp-desktop (version 331.79_k3.11.6_4-26.1) installs fine without it and the kernel has stayed at version 3.11.10-11.1 - I get a boot warning that the uvm module is not found, but the nvidia driver loads nonetheless. The machines have different graphics adaptors, which presumably accounts for the differing behaviour/requirements, and both _require_ the proprietary drivers (rather than nouveau) because we run graphics software which uses CUDA. Dx -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 06/09/2014 09:55 AM, Dylan wrote:
On 09/06/14 07:43, Gustav Degreef wrote:
I had problems with x due to the most recent updates. I did "zypper up" and it complained of problems with the nvidia rpm's (from the suse repo). After the update, my desktop's resolution was completely messed up and I had huge icons. The system was no longer using the nvidia drivers. The yast package manager showed six nvidia rpms (some of which were not correct). I removed ALL of the nvidia rpms, rebooted the system and then the desktop resolution returned to normal. I then selected only the two appropriate drivers (g02) for my card, rebooted. After the second reboot the system was now using the drivers and the desktop was back to normal. First time anything like this happened to me. I only use the suse nvidia rpm's and never install with packages from the nvidia site. Gustav
I had this problem with one machine, and completely non-functional graphics on another. There seems to be an additional package in the latest update: nvidia-uvm-gfxG0x-kmp-<kernel-flavour>.
On the machine with incorrect resolution, I did basically the same (remove all except the latest G03 packages) and had to install the uvm package. This resulted in the kernel being *downgraded* to 3.11.6.
On the machine with no graphics at all, I had to taboo the uvm package to get a graphical display - package nvidia-gfxG03-kmp-desktop (version 331.79_k3.11.6_4-26.1) installs fine without it and the kernel has stayed at version 3.11.10-11.1 - I get a boot warning that the uvm module is not found, but the nvidia driver loads nonetheless.
The machines have different graphics adaptors, which presumably accounts for the differing behaviour/requirements, and both _require_ the proprietary drivers (rather than nouveau) because we run graphics software which uses CUDA.
I am on a different machine using the nouveau driver. The affected machine has the 13.1 installation (with the nvidia drivers) . But it is now running 12.3 (off a different partition) so I can not give precise info. But perhaps this additional info is of use. I only needed nvidia-gfxG02-kmp-desktop and the x11-nvidia rpm to get the desktop to work properly. The update that messed up x installed nvidia-gfxG02-kmp-desktop, nvidia-gfxG02-kmp-default, the x11 rpm plus another gfxg02 rpm (don't remember the name). It also installed the same four gfxG03 rpm's - a total of 8 nvidia rpm's (not six as I had previously said) - very odd indeed. After I removed ALL of the nvidia rpms and then I re-installed the correct gfxg02-desktop and the x11nvidia packages the kernel was not downgraded. I did not look very carefully about the error messages after the update, but it said something to the effect that correct symlinks had not been properly set. I think there was a bug somewhere in this last update. Gustav.
Dx
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 09/06/14 09:17, Gustav Degreef wrote: [SNIP]
The update that messed up x installed nvidia-gfxG02-kmp-desktop, nvidia-gfxG02-kmp-default, the x11 rpm plus another gfxg02 rpm (don't remember the name). It also installed the same four gfxG03 rpm's - a total of 8 nvidia rpm's (not six as I had previously said) - very odd indeed.
After I removed ALL of the nvidia rpms and then I re-installed the correct gfxg02-desktop and the x11nvidia packages the kernel was not downgraded.
I did not look very carefully about the error messages after the update, but it said something to the effect that correct symlinks had not been properly set. I think there was a bug somewhere in this last update.
I'm satisfied that both machines are functional for my needs right now, and the older kernel isn't a particular issue, but I agree there was something awry with the latest nvidia update Dx -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
participants (4)
-
Alain Leblanc -
Carl Hartung -
Dylan -
Gustav Degreef