Screen setup broken after upgrade in KDE_Current Repo (OpenSuse 12.3)

English Hardware
#21

I have a related question.

SInce this problem is obviously linked to the update I made (it is highly unlikely that there was a hardware problem on 2 screens at the very same moment, plus today I tried with a beamer and same behavior), it would be important for me to know which packages were updated. I do not remember (did not even read the list). Is it possible to have a list of packages ordered by the date of installation / last update?

P

#22

In yast - software management there is a history. check the menus

Be sure that all NVIDIA packages match ie all that have a kernel flavor are the same as the kernel flavor you are running. Also that all are G03 versions. Remove any that don’t match, add any that are missing (there are 5 packages making up the NVIDIA constellation). I have seen NVIDA updates do odd things so I always keep a close eye on what is planned to happen when NVIDIA drivers are involved. Mixing verious NVIDIA package flavers and versions can cause problems.

#23

thanks. The problem as I see it is that in the same update I updated kscreen (and most of kde) some python-kde packages, the nvidia drivers; I had also installed vagrant, ansible and a virtual machine in virtualbox. I reverted all changes but the kde ones.

Do you think that problems with python-kde and python-sip might affect the issue? I had some dependency issue about those, that I solved. I assumed that it would not be the cause of the problem. Though probably kscreen uses python for visualisation?

Be sure that all NVIDIA packages match ie all that have a kernel flavor are the same as the kernel flavor you are running. Also that all are G03 versions. Remove any that don’t match, add any that are missing (there are 5 packages making up the NVIDIA constellation). I have seen NVIDA updates do odd things so I always keep a close eye on what is planned to happen when NVIDIA drivers are involved. Mixing verious NVIDIA package flavers and versions can cause problems.

I have checked and it was indeed the case that the kernel modules had a sub-subversion that was different (31 instead of 30) from the driver. But correcting that did not solve the issue.

#24

You already got an answer to that, but this is another possibility:

rpm -qa --last

No.
Kscreen does not use python at all.
And you have the same problem in nvidia-settings anyway. Even your Xorg.0.log shows three monitors as connected.

I have checked and it was indeed the case that the kernel modules had a sub-subversion that was different (31 instead of 30) from the driver.

Hm?
That’s no “sub-subversion”. That’s the build number.
And there is no 31, are you sure it was the same version? More likely it’s a completely different version, the older 340.32 maybe.
It is perfectly valid and normal to have more than one version of the kmp packages installed, just as you have more than one kernel normally.

I cannot imagine an installation problem with the nvidia packages cause such a problem anyway (detection of monitors that do not exist).
I was more thinking about a bug in the latest driver or something, that’s why I suggested to downgrade to the previous version.

#25

Update after one week of doing real work (on one screen).

  1. I tried to use the nouveau driver.
    Result: now it correctly displays one screen when one screen is plugged, and three when three are plugged.
    Problem 1: the two extra screens still wouldn’t switch on, because I cannot set their resolution to the correct one in kscreen: it is locked to 1600x900, no way to change it. See screenshot.
    Problem 2: the performance is bad and the image is not stable -> so I went back to Nvidia
    https://dl.dropboxusercontent.com/u/1118370/kscreen_resolution_locked.png
  2. I tried to check if my Nvidia install had some problems in it. These are my nvidia-related installed packages
libdrm_nouveau2                | Userspace interface for Kernel DRM se... | 2.4.42-1.1.1         
libdrm_nouveau2-32bit          | Userspace interface for Kernel DRM se... | 2.4.42-1.1.1       
nvidia-computeG03              | NVIDIA driver for computing with GPGPU   | 340.46-30.1           
nvidia-gfxG03-kmp-desktop      | NVIDIA graphics driver kernel module ... | 340.46_k3.7.10_1.1-30.1   
nvidia-glG03                   | NVIDIA GL libraries for OpenGL accele... | 340.46-30.1            
nvidia-uvm-gfxG03-kmp-desktop  | NVIDIA Unified Memory kernel module      | 340.46_k3.7.10_1.1-30.1  
x11-video-nvidiaG03            | NVIDIA graphics driver for GeForce 8x... | 340.46-30.1             
xf86-video-nv                  | NVIDIA video driver for the Xorg X se... | 2.1.20-3.1.1           
xorg-x11-driver-video-nouveau  | Accelerated Open Source driver for nV... | 1.0.6-2.1.1

I do not see anything wrong here. I have the right kernel for these drivers. ?oreovr, it all worked as a charm for one year prior to update…

…which left me with

  1. install an older version of the Nvidia driver.
    Problem: I went to the nvidia website and all I could find was the current driver. So not much luck there… any pointer to where I can find Nvidia legacy (ie, last week’s) drivers?

Thanks!

P

#26

SOLVED.

DOwngrading the nvidia driver ‘the hard way’ solved the issue.

Thanks wolfi and everyone who helped!

P

#27

So it is caused by a change in the nvidia driver.

Maybe you should report this to nvidia then…