opensuse 13.1 x86_64
kde 4.14.3
my graphical adapter is an asus silent nvidia Geforce 210
02:00.0 VGA compatible controller: NVIDIA Corporation GT218 [GeForce 210] (rev a2)
yesterday there was an update of nvidia driver to 340.65
i check that “no_kms_in_inird” = yes and nouveau is blacklisted
then
i restart the pc
then
we have a fist opensuse graphical panel
then
a black screen with a text line at the top
and that’s all.
no graphical panel to open a session !
i tried to restart several times and
i tried to install again the new driver then same bad result.
Since you uninstalled the driver manually, YaST/zypper probably installed the wrong packages.
And without a log file it’s not possible at all to know or even guess what was the problem anyway.
So please, try again to install the driver.
Check which kernel you use with “uname -a”, then install exactly those packages (select them manually in YaST, and replace -desktop with -default if you are using kernel-default):
nvidia-gfxG03-kmp-desktop, nvidia-uvm-gfxG03-kmp-desktop, x11-video-nvidiaG03, nvidia-glG03, nvidia-computeG03
If it doesn’t work then, try to reboot once again.
If it still doesn’t work, boot to recovery mode and post the following:
it’s strange that there are some kernel packages with version 3.11.10-17.1 , 3.11.10-17.2 .
if i search them with yast software manager they are not installed .
i will try soon to install again the nvidia driver
Since that’s the driver I use, I know a little more about that.
There was an update in the nvidia repo. However, the updated (newer) driver has a different vendor than the older driver. And the vendor change was inhibiting the update. I forced it, and all is well.
My guess is that something similar happened to the OP of this thread. He got a partial update, but not all of the Nvidia packages were updated, leaving inconsistencies.
The listed vendor seems to be the name of an OBS project. My guess is that the maintainer has changed so is using a different project name.
For folk who have installed Nvidia drivers from the repo: Check. And either update everything that you have installed from that repo, or update nothing. Don’t mix and match.
To check, I used Yast Software Management, and selected the “Repositories” view. Then I clicked on the Nvidia repository. Then I used the versions tab to see what I have and what is available. I had three packages. They were now listed in the “Versions” tab as “System” packages. Right clicking on each of those packages, I forced an update to the newer version in the Nvidia repo. After reboot, all was fine.
it’s strange in the xorg log there are references to nvidia 331.49
for example
3662.909] (II) LoadModule: “glx”
3662.909] (II) Loading /usr/lib64/xorg/modules/extensions/libglx.so
3662.938] (II) Module glx: vendor=“NVIDIA Corporation”
3662.938] compiled for 4.0.2, module version = 1.0.0
3662.938] Module class: X.Org Server Extension
3662.938] (II) NVIDIA GLX Module 331.49 Wed Feb 12 20:17:10 PST 2014
>>There was an update in the nvidia repo. However, the updated (newer) driver has a different vendor than the older driver. And the >>vendor change was inhibiting the update. I forced it, and all is well.
yes i encountered this and i forced the update then i discovered it does not work.
then
i uninstalled all nvidia packages and i deleted all about nvidia in /usr/src/kernel-modules/
then
with one click i installed again the nvidia driver then it does not work
i uninstalled all nvidia packages
then
i deleted all nvidia* folders and files remaining in /etc/ and /usr/
then
with one click i installed again the nvidia driver
then
same bad result but this time i have a good Xorg.0.log.old file here https://bugzilla.opensuse.org/attachment.cgi?id=616995
you can see interesting error messages
178.171] (EE) open /dev/dri/card0: No such file or directory
178.171] (WW) Falling back to old probe method for modesetting
178.171] (EE) open /dev/dri/card0: No such file or directory
…
178.174] (==) RandR enabled
178.180] (EE) Failed to initialize GLX extension (Compatible NVIDIA X driver not found)
178.257] (II) config/udev: Adding input device Power Button (/dev/input/event3)
I had the same problem with the nvidia driver update, and I found the cause in my case.
There were still versions of the old driver and links to it in the following directories that were NOT removed after installation or de-installation of the nvidia drivers.
Have a look at the content in the following directories:
They should be empty after installation of the latest nvidia driver.
I removed everything in those directories, and all was well after a restart!
Maybe it works for you too.
the packager/developer confuses between 13.1 and 13.2 nvidia installation . the installation perhaps is not the same thing.
this is not the first time xorg does not find any module to load because in the wrong place. i encountered the case of with nvidia in a past version (-2 or -3 i don’t remember).
Maybe something went wrong with the 13.1 packages. The way libglx is installed was completely changed in 13.2, if the 13.1 packages do it the same it won’t work.
But I just had a look and the 13.1 nvidia packages do still have libglx in /usr/lib64/xorg/modules/updates/extensions/, so this directory definitely should not be empty on 13.1!
@promeneur:
I saw this in your Xorg.log (the first one, when the wrong libglx was loaded):
3662.909] (II) Loading /usr/lib64/xorg/modules/extensions/libglx.so
But the 13.1 packages did and do put libglx.so into /usr/lib64/xorg/modules/updates/extensions/.
So where did this come from? Did you ever install the driver by hand via the .run installer and did not remove it completely?
Seems to be the case, as you mention you removed all nvidia* files and folders in /etc/ and /usr (there should not be any left after uninstalling the rpm packages).
Do you have nvidia-glG03 installed now? As I wrote, the 1-click install will not install it automatically if you ever uninstalled it.
Try to reinstall it, as libglx is not found:
sudo zypper in -f nvidia-glG03
And also check again that you have the correct packages installed now, after removing .
The packages/developer does not confuse anything.
This is done automatically by the spec file/OBS.
this is not the first time xorg does not find any module to load because in the wrong place. i encountered the case of with nvidia in a past version (-2 or -3 i don’t remember).
There was a one-time problem on 13.1 when this change has been made for 13.2. But this has been fixed on the next day.
And to me everything looks fine, but I haven’t installed the new packages (neither the 13.1 nor the 13.2 ones) myself.
I too had problems with upgrading the nvidia drivers from the repository. Since I’m new to openSUSE (and Nvidia ), it took me some time to investigate.
After I read my Xorg.0.log, it came down to libglx.so not being version 340.65, but 340.58 instead. I removed all nvidia packages, checked that all the files been removed too and reinstalled the packages from the download.nvidia.com repository. I made sure that the repository is 13.2 and reinstalled the packages, with no luck. The libglx.so had the wrong version, even though, the one referenced in /etc/alternatives had been removed and installed with the packages.
I’m sorry I can’t provide my Xorg.0.log. I went away from the Nvidia repository and installed the driver manually. That worked flawless (init 3, ./NVIDIA-Linux-x86_64-343.36.run, init 5), but I can’t provide my old Xorg.0.log.
Maybe my investigations point someone to a right location or maybe I had a completely different problem, but I’d like any feedback!
Hm, this does sound like a different problem.
First, in 13.2 the way libglx is installed has completely changed.
Second, epromeneur has libglx 340.49 in the Xorg.0.log he posted, and (as I now see) 340.65 (i.e. the correct one) in his second log.
And I do see 340.65 in the repo for 13.2.
Maybe you installed at a time when the repo was in an inconsistent state?
You’re booting “recovery mode”, which means the nvidia driver is not even used (and that’s on purpose)!
According to the log you’re using fbdev, but Mesa’s software renderer is broken by the nvidia installation (nvidia’s libglx and libGL are incompatible to Mesa’s).
That explains the “Failed to initialize GLX extension (Compatible NVIDIA X driver not found)” error of course.
Do a normal boot.
If you still have problems then, reboot to recovery mode again and post Xorg.0.old.
I saw this in your Xorg.log (the first one, when the wrong libglx was loaded):
Code:
3662.909] (II) Loading /usr/lib64/xorg/modules/extensions/libglx.so
But the 13.1 packages did and do put libglx.so into /usr/lib64/xorg/modules/updates/extensions/.
Did you ever install the driver by hand via the .run installer and did not remove it completely?
Seems to be the case, as you mention you removed all nvidia* files and folders in /etc/ and /usr (there should not be any left after uninstalling the rpm packages).
in the past i used a one click dkms installation from user “Knuth” then i
switched to standard one click nvidia installation. this is a part of
the pb . i uninstalled not completely the dkms installation.
but as
i said i uninstalled the nvidia driver then i deleted all remaining
nvidia folders and files then installed again with one click the nvidia
driver. So now i hope we are in the standard case when we try to install
(not to update) the nvidia driver.
So where did this come from?
it comes from /usr/lib64/xorg/modules/extensions/
/usr/lib64/xorg/modules/extensions/
/usr/lib64/xorg/modules/extensions/libglx.so
/usr/lib64/xorg/modules/extensions/libglx.so.331.49
/usr/lib64/xorg/modules/extensions/libvnc.so
i see now i did not clean completly all files about previous installations.
perhaps there is some files , links somewhere which break any new installation.
linux-b4lz:~ # zypper in -f nvidia-glG03
Retrieving repository 'FAH' metadata ....................................................................[done]
Retrieving repository 'Free guide' metadata .............................................................[done]
Retrieving repository 'Java Platform' metadata ..........................................................[done]
Retrieving repository 'VirtualBox' metadata .............................................................[done]
Loading repository data...
Reading installed packages...
Forcing installation of 'nvidia-glG03-340.65-36.1.x86_64' from repository 'nVidia Graphics Drivers'.
Resolving package dependencies...
The following package is going to be reinstalled:
nvidia-glG03
1 package to reinstall.
Overall download size: 20.9 MiB. No additional space will be used or freed after the operation.
Continue? [y/n/? shows all options] (y): y
Retrieving package nvidia-glG03-340.65-36.1.x86_64 (1/1), 20.9 MiB (129.3 MiB unpacked)
Retrieving: nvidia-glG03-340.65-36.1.x86_64.rpm ...........................................[done (498.2 KiB/s)]
(1/1) Installing: nvidia-glG03-340.65-36.1 ..............................................................[done]
Additional rpm output:
warning: /var/cache/zypp/packages/download.nvidia.com-opensuse/x86_64/nvidia-glG03-340.65-36.1.x86_64.rpm: Header V4 DSA/SHA1 Signature, key ID c66b6eae: NOKEY
:~ #
but as
i said i uninstalled the nvidia driver then i deleted all remaining
nvidia folders and files then installed again with one click the nvidia
driver. So now i hope we are in the standard case when we try to install
(not to update) the nvidia driver.
No. The packages you removed are soft-locked by YaST/zypper, so the 1-click install will most likely install not all necessary packages, and the wrong kmps.
So again, select them manually in YaST, all 5 I mentioned!
it comes from /usr/lib64/xorg/modules/extensions/
/usr/lib64/xorg/modules/extensions/
/usr/lib64/xorg/modules/extensions/libglx.so
/usr/lib64/xorg/modules/extensions/libglx.so.331.49
/usr/lib64/xorg/modules/extensions/libvnc.so
i see now i did not clean completly all files about previous installations.
perhaps there is some files , links somewhere which break any new installation.
Maybe, but those should not be used, if there’s a libglx.so in the “updates” folder (that one has precedence).
But better to remove those anyway, and maybe reinstall xorg-x11-server, as its files seem to have been replaced. (and maybe Mesa-* as well to be on the safe side)
[/QUOTE]
You use kernel-desktop, but have the kmps for kernel-default installed. This cannot work.
Install nvidia-gfxG03-kmp-desktop and nvidia-uvm-gfxG03-desktop as I told you already yesterday. And remove those two *-default packages.
), it took me some time to investigate.