Difficulties with activation of another current Nvidia graphic driver

English Hardware
#1

I have dared to try the following packages out according to the version 352.63_k3.16.6_2-18.1.

  • nvidia-computeG04
  • nvidia-gfxG04-kmp-default
  • nvidia-gfxG04-kmp-desktop
  • nvidia-glG04
  • nvidia-uvm-gfxG04-kmp-default
  • nvidia-uvm-gfxG04-kmp-desktop
  • x11-video-nvidiaG04

Unfortunately, I get the error message “Failed to initialize the NVIDIA kernel module.” during the start of the X server.

Linux 4.1.1-1-desktop:

…
    44.699] (EE) [drm] Failed to open DRM device for (null): -22
    44.699] (EE) [drm] Failed to open DRM device for pci:0000:04:00.0: -22
…

Linux 4.4.0-1-default:

…
    47.708] (EE) [drm] Failed to open DRM device for pci:0000:04:00.0: -19
…

A fallback driver is then selected which provides me a lower screen resolution than what my primary monitor is capable of.
Which fine-tuning should I reconsider so that the full graphic resolution will be displayed again?

How are the chances to restore any previous software versions which worked for a while?

#2

So what hardware do you have?? Is it a Notebook??

Uninstall the GO4 flavour packages and reinstall what you used before.

#3

Would you like to look at the software configuration challenges around my graphic card “GT 730” once more?

Is it a Notebook?

No. - Can surprises from Optimus stuff be excluded for this help request?

Uninstall the GO4 flavour packages and reinstall what you used before.

Can it be that I will need to return to a graphic driver version (like 352.30) before the X.Org xserver ABI 20 (xorg-server 1.18) was introduced?

It seems that the ABI version is not explicitly expressed as an important package property. Will dependency resolution become interesting then if I would attempt a downgrade of corresponding software components anyhow?

#4

Uninstall GO4 Install GO3 or what ever worked before. Uninstall to be sure to remove all parts then install to replace with older driver

Though GO4 should work with your card.

It is important to uninstall before installing. If you don’t some parts of the previous driver could be left.

#5

Yes. In particular, the kernel module might just get removed otherwise.

The same applies when switching from G03 to G04 though.
If the “old” driver is not uninstalled before, there might be problems. A “side-grade” is not supported.

This is explicitly mentioned here since quite some time:
https://en.opensuse.org/SDB:NVIDIA_drivers

And it seems to have been the problem here.

#6

It seems that there are still some software development surprises to consider.

  • The previous graphic driver is not stored in the local directory “/var/cache/zypp/packages/nVidia” any more. Would you like to suggest any other installation source from which I could retrieve an appropriate RPM file that should work for my system configuration?
  • Do I need to reuse the script “NVIDIA-Linux-x86_64-352.30.run” eventually?
#7

On the first: of course it’s not stored anymore. The folder is used for caching packages before install. You should use the software manager or zypper.

Please post output of


rpm -qa | grep nvidia
zypper lr

On the second: No. Not unless you know exactly what you’re doing. You first need to get rid of all the nvidia stuff.

#8

And which ones?
Could you provide more details about what you mean with that?

The previous graphic driver is not stored in the local directory “/var/cache/zypp/packages/nVidia” any more. Would you like to suggest any other installation source from which I could retrieve an appropriate RPM file that should work for my system configuration?

No.
Both the G03 and the G04 packages are available in the online repo.

Do I need to reuse the script “NVIDIA-Linux-x86_64-352.30.run” eventually?

Well, how did you install the driver in the first place?

If you used the NVIDIA-Linux-xxx.run (i.e. nvidia’s own installer), you have to reuse it, yes.
If you used the packages from the nvidia repo, just uninstall them via YaST or zypper and install the new ones. But do so in separate steps, i.e. completely remove them and then afterwards install the other ones. An “update”/“downgrade” (or as I called it “side-grade”) is not supported.

A side-note to you, personally, based on my previous experiences: you seem to run heavily modified systems, but always fail to provide that information.
Your note about the Xorg ABI versions seem to underline that.
It’s impossible (or quite difficult) to help you, considering this.

But, the latest G04 version should work with the latest Xorg version, I think. For older nvidia drivers (G03 e.g.), that’s not always the case.
NVidia seems to need longer to adapt the legacy drivers to the latest Xorg ABI, if they do at all.

PS: as this is about Tumbleweed, you might want to try the latest Beta version of the nvidia driver. That should work fine AFAIK.
But again, please uninstall the previous one first, to prevent problems. (although updating from G04 to a later G04 should work fine)

#9

Do the displayed return codes like “-22” or “-19” indicate software development surprises which might become interesting to some degree?

Both the G03 and the G04 packages are available in the online repo.

Yes. - Unfortunately, I do not see that the version to which I would return to eventually is provided by this RPM interface at the moment.

Well, how did you install the driver in the first place?

Is an usual installation source for a package like “x11-video-nvidiaG04” connected with the information “Vendor: obs://build.suse.de/home:sndirsch:drivers”?

If you used the packages from the nvidia repo, just uninstall them via YaST or zypper and install the new ones.

I imagine that this approach will still not produce the desired result occasionally.

A side-note to you, personally, based on my previous experiences: you seem to run heavily modified systems,

I am using an openSUSE Tumbleweed system together with a bit of fine-tuning for my needs.

but always fail to provide that information.

I am increasing my feedback incrementally.

Your note about the Xorg ABI versions seem to underline that.

Is this another aspect which belongs to the usual concerns around the discussed software evolution?

But, the latest G04 version should work with the latest Xorg version, I think.

This would be nice usually. - How do you think about to clarify the really usable version ranges together with the mentioned Linux variants a bit more?

#10

Not to me, or probably most users here.

Better ask the kernel or nvidia developers, or look at the source code.

Yes. - Unfortunately, I do not see that the version to which I would return to eventually is provided by this RPM interface at the moment.

There’s G02, G03, and G04 in there.

But it’s for Leap 42.1.
You are using Tumbleweed, aren’t you?

Those packages won’t necessarily work on Tumbleweed.
They do compile the kernel module on installation, but might not be compatible with the kernel or Xorg version in Tumbleweed.

Is an usual installation source for a package like “x11-video-nvidiaG04” connected with the information “Vendor: obs://build.suse.de/home:sndirsch:drivers”?

No, that’s ok.
Those packages are built by someone at (open)SUSE (in /home:sndirsch:drivers) and then copied over to nvidia’s server.

I imagine that this approach will still not produce the desired result occasionally.

On Tumbleweed, probably not.
It is not recommended to use Tumbleweed if you rely on third-party drivers like nvidia.
They might work (especially in the case of nvidia), but might also break at times because of updates.

I am using an openSUSE Tumbleweed system together with a bit of fine-tuning for my needs.

Yeah, but those “fine-tuning for my needs” are a possible point of problem.
Not necessarily in this case, but I know they have lead to (spurious) bug reports by you in the past.

I am increasing my feedback incrementally.

Great. Thanks.

But your “fine-tuning” still might make it difficult to help you.
Especially if you have self-compiled libraries in /usr/local/ nobody knows of… :wink:

Is this another aspect which belongs to the usual concerns around the discussed software evolution?

I don’t know what “software evolution” you speak of especially.
But yes, you get always the latest Xorg in Tumbleweed.
NVidia may or may not support that in time.

This would be nice usually. - How do you think about to clarify the really usable version ranges together with the mentioned Linux variants a bit more?

Well, that’s something you’d have to ask NVidia I suppose.
It’s their (proprietary, closed-source, binary) driver.

Although, NVidia isn’t necessarily aware of upcomig changes in the Kernel or Xorg either, I suppose. They just have to keep up.

#11

Can any repositories be configured in the way that a few of them should keep the loaded RPM files longer than others?

#12

It is nice when this approach works, isn’t it?

They might work (especially in the case of nvidia),

Often enough?

but might also break at times because of updates.

I was affected once more by such an occasion.

Not necessarily in this case, but I know they have lead to (spurious) bug reports by you in the past.

I am working with various software so that it will happen occasionally to stumble on unpleasant issues. Are you interested to improve situations around such reports any further?

I don’t know what “software evolution” you speak of especially.
But yes, you get always the latest Xorg in Tumbleweed.
NVidia may or may not support that in time.

There are a lot of versions floating around for the involved software components over time, aren’t there?

#13

Yes, it is I suppose. It’s always nice when someone’s approach works, no?

But I’m not sure I fully understand your statement/question here.

Often enough?

Most of the time, yes.
But at least if a new major kernel or Xorg version is released and published in Tumbleweed, there might be problems.

And you have to reinstall the driver after certain updates anyway, if you use nvidia’s installer.
In particular after updates to the kernel, Xorg, and Mesa.

The rpm packages would help you there, but as mentioned they are not available for Tumbleweed, and the ones for different openSUSE versions are not guaranteed to work on Tumbleweed.

I was affected once more by such an occasion.

That’s why it is not recommended to use Tumbleweed if you rely on 3rd party drivers.
If you do anyway, you should at least know what you are doing.

I am working with various software so that it will happen occasionally to stumble on unpleasant issues. Are you interested to improve situations around such reports any further?

Again, I don’t really understand your question.

I am not interested to improve the situation, because I am not involved in that part.
And if you modify/replace openSUSE’s packages or the binaries they install, you cannot really expect support or improvement from openSUSE.

The distribution provides things that fit together.
If the packages provided by the distribution don’t fit together or don’t work, it is a bug in the distribution and you should report it to openSUSE’s bugzilla.
If you replace random things by installing (incompatible) stuff to /usr/local/, you are on your own, sorry.

You can ask for help here, but you cannot request or expect help in openSUSE’s bugzilla.

And as said, if you do that (installing own versions of software to /usr/local/) complicates helping as well, especially if you don’t even mention that in advance.
If you do that, you should know what you are doing yourself.

There are a lot of versions floating around for the involved software components over time, aren’t there?

Of course.
Another reason to not use Tumbleweed if you don’t know what you are doing, or how to fix your 3rd party drivers at least.

To come back to your original problem in this thread:
Did you remove the nvidia driver and reinstall it?

You didn’t answer any of the questions yet, like where/how you installed it in the first place.

Please, don’t let people guess…

#14

I find this feedback a bit strange. - You seem to be interested enough to continue the dialogue with me. Do you try also to adjust the situation with advices?

And if you modify/replace openSUSE’s packages or the binaries they install, you cannot really expect support or improvement from openSUSE.

How do you think about the possibility that any more users from “the community” will share useful ideas here?

If the packages provided by the distribution don’t fit together or don’t work, it is a bug in the distribution and you should report it to openSUSE’s bugzilla.

I did this occasionally. - Would you like to add anything to the remaining open issues?

You can ask for help here, but you cannot request or expect help in openSUSE’s bugzilla.

I got support by both communication channels in different degree.

Did you remove the nvidia driver and reinstall it?

Not yet. - I would like to clarify the involved software dependencies a bit more this time.

  1. What is the last Linux version with which the Nvidia graphic software 352.30 (or 352.41) will work?
  2. Will a downgrade be possible for components around the software “xorg-x11-server” from the version “7.6_1.18.0-428.1 (ABI 20)” to “7.6_1.16.1-28.1 (ABI ??)” for example?
  3. Can the graphic module “nouveau”, “nv” or “vesa” display also a higher screen resolution than “1280 x 1024” (what I see at the moment)?

You didn’t answer any of the questions yet, like where/how you installed it in the first place.

Is the following information display clearer than my previous answers?

Name        : nvidia-gfxG04-kmp-desktop
Version     : 352.63_k3.16.6_2
Release     : 18.1
…
Install Date: So 17 Jan 2016 13:18:59 CET
…
Build Date  : Mo 16 Nov 2015 16:53:25 CET
Build Host  : kemter
…
Vendor      : obs://build.suse.de/home:sndirsch:drivers
Summary     : NVIDIA graphics driver kernel module for GeForce 400 series and newer
…
Distribution: home:sndirsch:drivers / openSUSE_13.2
#15

I try to help people with problems.
And I already gave you advices.

But I’m not involved in the nvidia driver development or packaging, nor in Xorg.

How do you think about the possibility that any more users from “the community” will share useful ideas here?

Well, without further information from you, I doubt there will be much more useful ideas here.

I did this occasionally. - Would you like to add anything to the remaining open issues?

What open issues?
If you find a bug in the distribution pacakges, report it.
The nvidia driver is not part of the distribution though.
If there’s a bug in it, or the installation doesn’t work, report it to nvidia. Or wait until it is fixed/updated.

I got support by both communication channels in different degree.

Fine.
But bugzilla is a bug tracking system, not a user support forum.

What is the last Linux version with which the Nvidia graphic software 352.30 (or 352.41) will work?

I don’t know. Neither the release notes nor nvidia’s homepage tell that.
But 352.30 and 352.41 are old. And likely too old for Tumbleweed.
Use the latest versions. They have the best chance to work.

The current version is

For G03 the latest version 340.96, for G04 it’s 352.63.
And the latest Beta version is

Will a downgrade be possible for components around the software “xorg-x11-server” from the version “7.6_1.18.0-428.1 (ABI 20)” to “7.6_1.16.1-28.1 (ABI ??)” for example?

No, unless you compile it yourself.
Or switch back to Leap or 13.2.

If you use btrfs, you might be able to switch back to an earlier snapshot, but then you miss all updates since then.

Can the graphic module “nouveau”, “nv” or “vesa” display also a higher screen resolution than “1280 x 1024” (what I see at the moment)?

Yes. And nouveau supports OpenGL/GLX as well.
[li]The other two are probably not a good idea to use. vesa is a generic fallback driver, nv is a very old driver (originally by nvidia itself) that probably doesn’t support the latest cards and only provides some 2D acceleration, no 3D/OpenGL.[/li]
But you have to remove all nvidia packages for those other drivers to work fully.

Is the following information display clearer than my previous answers?

Name        : nvidia-gfxG04-kmp-desktop
Version     : 352.63_k3.16.6_2
...

Yes, a bit.
But why do you have nvidia-gfxG04-kmp-desktop?
That only works with kernel-desktop obviously, which doesn’t exist any more in Tumbleweed.
Actually this package is from 13.2, not Leap.
Remove it.

And please post your repo list:

zypper lr -d

What other nvidia packages have you installed now exactly?

rpm -qa | grep nvidia

As already said, try to remove all of them completely, and install them fresh.

zypper rm *nvidia*
zypper in x11-video-nvidiaG04

As already indicated, two things are important when using the packages from the repo:

  • you cannot directly upgrade G03 to G04, this doesn’t work, you have to remove the old version completely before installing the new one.
  • the kernel module is compiled on package installation, it might not work after a kernel update (especially after a major one like 4.3.x to 4.4.0).

Reinstalling the driver packages (after removing them) should fix both cases. (actually it should suffice if you only reinstall the 2 kmp packages)

That’s the latest available version. If it doesn’t work, you’re out of luck.

You could try to install the latest Beta in this case, this should work on Tumbleweed AFAIK. But you’d have to download the installer from nvidia and install it “the hard way” (which isn’t really hard, but you need to reinstall it after every update to the Kernel, Mesa, or Xorg).
But remove all installed nvidia packages first, and the repo as well.

If that still doesn’t help, you probably should post your /var/log/Xorg.0.log so that we can see what’s actually going wrong.

But I repeat: using Tumbleweed with 3rd party drivers like nvidia is not supported (by openSUSE).
And you are using the driver packages for Leap, they are created for Leap, they are tested on Leap, they should work on Leap.
Not Tumbleweed!

Tumbleweed is constantly changing, that’s its very nature and purpose as a rolling distribution. 3rd party drivers might not always work with the latest versions immediately (kernel and Xorg in this case).
If you want a stable experience with the nvidia driver, use a stable distribution like 13.2 or Leap, not Tumbleweed.

#16

Hi,
If you are using tumbleweed, you should use the .run installer from the nvidia site.
You can’t rely on the rpm as it will not be released same time with
a new kernel update.
You will need the gcc, make, kernel source and kernel devel packages to build
the nvidia driver using the .run installer.

#17

Thanks. - How many improvements do you achieve on this way occasionally?

But I’m not involved in the nvidia driver development or packaging, nor in Xorg.

Did your support efforts result in interesting contacts from this application domain?

Well, without further information from you, I doubt there will be much more useful ideas here.

Does our constructive dialogue contain enough information for further inspirations?

What open issues?

Examples for further clarification requests:
[ul]
[li]Difficulties with update for “Git” because of a special dependency on “Perl” [/li][li]missing icons for buttons in Linphone [/li][/ul]

The nvidia driver is not part of the distribution though.

I know …

Do a lot of users try to find solutions for their difficulties by alternative means because of this fact?

Neither the release notes nor nvidia’s homepage tell that.

I hope that more knowledge about usable version ranges will become available.

But 352.30 and 352.41 are old. And likely too old for Tumbleweed.

One of these versions worked on my system for a while before I dared to try another update out.

Use the latest versions. They have the best chance to work.

I have got an alternative view on the effects from corresponding software evolution.

No, unless you compile it yourself.
Or switch back to Leap or 13.2.

I hoped to avoid this.

And nouveau supports OpenGL/GLX as well.
[li]The other two are probably not a good idea to use. vesa is a generic fallback driver, nv is a very old driver (originally by nvidia itself) that probably doesn’t support the latest cards and only provides some 2D acceleration, no 3D/OpenGL.[/li]
How should be determined which of these modules is responsible for the display of the screen resolution “1280 x 1024” at the moment?

But why do you have nvidia-gfxG04-kmp-desktop?

I want to keep the software “Linux 4.1.1-1-desktop” running on my system until I know that alternative versions will also work together with other components as expected for my needs.

That only works with kernel-desktop obviously, which doesn’t exist any more in Tumbleweed.

It was also provided by this installation source for a while.

And please post your repo list:

I would like to omit the repository list here at the moment because this information did not really help in similar clarification tries so far.

What other nvidia packages have you installed now exactly?

[ol]
[li]nvidia-computeG04-352.63-18.1.x86_64 [/li][li]nvidia-gfxG04-kmp-desktop-352.63_k3.16.6_2-18.1.x86_64 [/li][li]nvidia-glG04-352.63-18.1.x86_64 [/li][li]nvidia-uvm-gfxG04-kmp-desktop-352.63_k3.16.6_2-18.1.x86_64 [/li][li]x11-video-nvidiaG04-352.63-18.1.x86_64 [/li][/ol]

As already said, try to remove all of them completely, and install them fresh.

I have tried this after noon.

If that still doesn’t help, you probably should post your /var/log/Xorg.0.log so that we can see what’s actually going wrong.

[    47.001] 
X.Org X Server 1.18.0
Release Date: 2015-11-09
[    47.001] X Protocol Version 11, Revision 0
[    47.002] Build Operating System: openSUSE SUSE LINUX
[    47.002] Current Operating System: Linux Sonne 4.1.1-1-desktop #1 SMP PREEMPT Wed Jul 8 14:23:40 UTC 2015 (cac28b3) x86_64
[    47.002] Kernel command line: root=/dev/duda/root resume=LABEL=SSHD-temp splash=silent quiet vga=0x31b nomodeset
[    47.002] Build Date: 18 January 2016  07:18:00PM
[    47.002]  
[    47.002] Current version of pixman: 0.33.6
…
[    47.002] (==) Log file: "/var/log/Xorg.0.log", Time: Tue Jan 19 14:58:14 2016
[    47.049] (==) Using config directory: "/etc/X11/xorg.conf.d"
[    47.049] (==) Using system config directory "/usr/share/X11/xorg.conf.d"
[    47.086] (==) No Layout section.  Using the first Screen section.
[    47.086] (==) No screen section available. Using defaults.
[    47.086] (**) |-->Screen "Default Screen Section" (0)
[    47.086] (**) |   |-->Monitor "<default monitor>"
[    47.087] (==) No monitor specified for screen "Default Screen Section".
    Using a default monitor configuration.
[    47.087] (==) Automatically adding devices
[    47.087] (==) Automatically enabling devices
[    47.087] (==) Automatically adding GPU devices
[    47.087] (==) Max clients allowed: 256, resource mask: 0x1fffff
[    47.174] (WW) The directory "/usr/share/fonts/misc/sgi" does not exist.
[    47.175]     Entry deleted from font path.
[    47.176] (==) FontPath set to:
    /usr/share/fonts/misc:unscaled,
    /usr/share/fonts/Type1/,
    /usr/share/fonts/100dpi:unscaled,
    /usr/share/fonts/75dpi:unscaled,
    /usr/share/fonts/ghostscript/,
    /usr/share/fonts/cyrillic:unscaled,
    /usr/share/fonts/truetype/,
    built-ins
[    47.176] (==) ModulePath set to "/usr/lib64/xorg/modules"
[    47.176] (II) The server relies on udev to provide the list of input devices.
    If no devices become available, reconfigure udev or disable AutoAddDevices.
[    47.176] (II) Loader magic: 0x80fd00
[    47.176] (II) Module ABI versions:
[    47.176]     X.Org ANSI C Emulation: 0.4
[    47.176]     X.Org Video Driver: 20.0
[    47.176]     X.Org XInput driver : 22.1
[    47.176]     X.Org Server Extension : 9.0
[    47.178] (++) using VT number 7

[    47.178] (II) systemd-logind: logind integration requires -keeptty and -keeptty was not provided, disabling logind integration
[    47.181] (--) PCI:*(0:4:0:0) 10de:1287:1043:84d6 rev 161, Mem @ 0xfa000000/16777216, 0xd8000000/134217728, 0xd6000000/33554432, I/O @ 0x0000ec00/128, BIOS @ 0x????????/524288
[    47.182] (II) LoadModule: "glx"
[    47.209] (II) Loading /usr/lib64/xorg/modules/extensions/libglx.so
[    47.837] (II) Module glx: vendor="NVIDIA Corporation"
[    47.837]     compiled for 4.0.2, module version = 1.0.0
[    47.837]     Module class: X.Org Server Extension
[    47.860] (II) NVIDIA GLX Module  352.63  Sat Nov  7 20:52:00 PST 2015
[    47.870] (==) Matched nvidia as autoconfigured driver 0
[    47.870] (==) Matched nouveau as autoconfigured driver 1
[    47.870] (==) Matched nv as autoconfigured driver 2
[    47.870] (==) Matched modesetting as autoconfigured driver 3
[    47.870] (==) Matched fbdev as autoconfigured driver 4
[    47.870] (==) Matched vesa as autoconfigured driver 5
[    47.870] (==) Assigned the driver to the xf86ConfigLayout
[    47.870] (II) LoadModule: "nvidia"
[    47.923] (II) Loading /usr/lib64/xorg/modules/drivers/nvidia_drv.so
[    47.985] (II) Module nvidia: vendor="NVIDIA Corporation"
[    47.985]     compiled for 4.0.2, module version = 1.0.0
[    47.985]     Module class: X.Org Video Driver
[    47.995] (II) LoadModule: "nouveau"
[    47.996] (II) Loading /usr/lib64/xorg/modules/drivers/nouveau_drv.so
[    48.000] (II) Module nouveau: vendor="X.Org Foundation"
[    48.000]     compiled for 1.18.0, module version = 1.0.12
[    48.000]     Module class: X.Org Video Driver
[    48.000]     ABI class: X.Org Video Driver, version 20.0
[    48.000] (II) LoadModule: "nv"
[    48.001] (II) Loading /usr/lib64/xorg/modules/drivers/nv_drv.so
[    48.003] (II) Module nv: vendor="X.Org Foundation"
[    48.003]     compiled for 1.18.0, module version = 2.1.20
[    48.003]     Module class: X.Org Video Driver
[    48.003]     ABI class: X.Org Video Driver, version 20.0
[    48.003] (II) LoadModule: "modesetting"
[    48.004] (II) Loading /usr/lib64/xorg/modules/drivers/modesetting_drv.so
[    48.024] (II) Module modesetting: vendor="X.Org Foundation"
[    48.024]     compiled for 1.18.0, module version = 1.18.0
[    48.024]     Module class: X.Org Video Driver
[    48.024]     ABI class: X.Org Video Driver, version 20.0
[    48.024] (II) LoadModule: "fbdev"
[    48.024] (II) Loading /usr/lib64/xorg/modules/drivers/fbdev_drv.so
[    48.026] (II) Module fbdev: vendor="X.Org Foundation"
[    48.026]     compiled for 1.18.0, module version = 0.4.4
[    48.026]     Module class: X.Org Video Driver
[    48.026]     ABI class: X.Org Video Driver, version 20.0
[    48.026] (II) LoadModule: "vesa"
[    48.026] (II) Loading /usr/lib64/xorg/modules/drivers/vesa_drv.so
[    48.027] (II) Module vesa: vendor="X.Org Foundation"
[    48.028]     compiled for 1.18.0, module version = 2.3.4
[    48.028]     Module class: X.Org Video Driver
[    48.028]     ABI class: X.Org Video Driver, version 20.0
[    48.028] (II) NVIDIA dlloader X Driver  352.63  Sat Nov  7 20:29:25 PST 2015
[    48.028] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
[    48.029] (II) NOUVEAU driver 
…
[    48.030] (II) modesetting: Driver for Modesetting Kernel Drivers: kms
[    48.030] (II) FBDEV: driver for framebuffer: fbdev
[    48.030] (II) VESA: driver for VESA chipsets: vesa
[    48.044] (II) Loading sub module "fb"
[    48.044] (II) LoadModule: "fb"
[    48.044] (II) Loading /usr/lib64/xorg/modules/libfb.so
[    48.062] (II) Module fb: vendor="X.Org Foundation"
[    48.062]     compiled for 1.18.0, module version = 1.0.0
[    48.062]     ABI class: X.Org ANSI C Emulation, version 0.4
[    48.062] (II) Loading sub module "wfb"
[    48.062] (II) LoadModule: "wfb"
[    48.062] (II) Loading /usr/lib64/xorg/modules/libwfb.so
[    48.071] (II) Module wfb: vendor="X.Org Foundation"
[    48.071]     compiled for 1.18.0, module version = 1.0.0
[    48.071]     ABI class: X.Org ANSI C Emulation, version 0.4
[    48.071] (II) Loading sub module "ramdac"
[    48.071] (II) LoadModule: "ramdac"
[    48.071] (II) Module "ramdac" already built-in
[    48.263] (EE) NVIDIA: Failed to initialize the NVIDIA kernel module. Please see the
[    48.263] (EE) NVIDIA:     system's kernel log for additional error messages and
[    48.263] (EE) NVIDIA:     consult the NVIDIA README for details.
[    48.480] (EE) [drm] Failed to open DRM device for pci:0000:04:00.0: -19
[    48.694] (EE) [drm] Failed to open DRM device for pci:0000:04:00.0: -19
[    48.694] (EE) open /dev/dri/card0: No such file or directory
[    48.694] (WW) Falling back to old probe method for modesetting
[    48.694] (EE) open /dev/dri/card0: No such file or directory
[    48.694] (II) Loading sub module "fbdevhw"
[    48.694] (II) LoadModule: "fbdevhw"
[    48.694] (II) Loading /usr/lib64/xorg/modules/libfbdevhw.so
[    48.715] (II) Module fbdevhw: vendor="X.Org Foundation"
[    48.715]     compiled for 1.18.0, module version = 0.0.2
[    48.715]     ABI class: X.Org Video Driver, version 20.0
[    48.716] (**) FBDEV(1): claimed PCI slot 4@0:0:0
[    48.716] (II) FBDEV(1): using default device
[    48.716] (WW) Falling back to old probe method for vesa
[    48.716] (EE) Screen 0 deleted because of no matching config section.
[    48.716] (II) UnloadModule: "modesetting"
[    48.716] (II) FBDEV(0): Creating default Display subsection in Screen section
    "Default Screen Section" for depth/fbbpp 24/32
[    48.716] (==) FBDEV(0): Depth 24, (==) framebuffer bpp 32
[    48.716] (==) FBDEV(0): RGB weight 888
[    48.716] (==) FBDEV(0): Default visual is TrueColor
[    48.716] (==) FBDEV(0): Using gamma correction (1.0, 1.0, 1.0)
[    48.716] (II) FBDEV(0): hardware: VESA VGA (video memory: 10240kB)
[    48.716] (II) FBDEV(0): checking modes against framebuffer device...
[    48.716] (II) FBDEV(0): checking modes against monitor...
[    48.716] (--) FBDEV(0): Virtual size is 1280x1024 (pitch 1280)
[    48.716] (**) FBDEV(0):  Built-in mode "current": 131.1 MHz, 80.3 kHz, 76.6 Hz
[    48.716] (II) FBDEV(0): Modeline "current"x0.0  131.09  1280 1312 1472 1632  1024 1028 1032 1048 -hsync -vsync -csync (80.3 kHz b)
[    48.716] (==) FBDEV(0): DPI set to (96, 96)
[    48.716] (II) Loading sub module "fb"
[    48.716] (II) LoadModule: "fb"
[    48.716] (II) Loading /usr/lib64/xorg/modules/libfb.so
[    48.716] (II) Module fb: vendor="X.Org Foundation"
[    48.716]     compiled for 1.18.0, module version = 1.0.0
[    48.716]     ABI class: X.Org ANSI C Emulation, version 0.4
[    48.716] (**) FBDEV(0): using shadow framebuffer
[    48.716] (II) Loading sub module "shadow"
[    48.716] (II) LoadModule: "shadow"
[    48.717] (II) Loading /usr/lib64/xorg/modules/libshadow.so
[    48.729] (II) Module shadow: vendor="X.Org Foundation"
[    48.729]     compiled for 1.18.0, module version = 1.1.0
[    48.729]     ABI class: X.Org ANSI C Emulation, version 0.4
[    48.729] (II) UnloadModule: "vesa"
[    48.729] (II) Unloading vesa
[    48.729] (==) Depth 24 pixmap format is 32 bpp
[    48.729] (II) FBDEV(0): FBIOBLANK: Invalid argument (Screen blanking not supported by kernel - disabling)
[    48.746] (==) FBDEV(0): Backing store enabled
[    48.748] (==) FBDEV(0): DPMS enabled
[    48.748] (==) RandR enabled
[    48.772] (EE) Failed to initialize GLX extension (Compatible NVIDIA X driver not found)
…

Does this detailed information help really more for the desired problem resolution than my initial request?

Not Tumbleweed!

To which release name would a software like “Linux 4.1.1-1-desktop” belong in your view?

Tumbleweed is constantly changing, that’s its very nature and purpose as a rolling distribution.

I like this way for big parts of the provided software collection in principle. - I would appreciate if it will become possible to consider some special maintenance rules for my aging hardware.

Can the Open Build Service help me to resolve special requirements for this use case?

If you want a stable experience with the nvidia driver, use a stable distribution like 13.2 or Leap, not Tumbleweed.

How “stable” can you really keep this component if you might notice that you will need the provided software corrections?

#18

Depends on what you mean with “improvements” here.

Did your support efforts result in interesting contacts from this application domain?

Not really. Why should it?
I just try to help users with problems here in the forum.

Does our constructive dialogue contain enough information for further inspirations?

Well, maybe. At least it made me aware that Xorg has been updated recently in Tumbleweed, with a new ABI version.

Examples for further clarification requests:

Well, git works fine on my Tumbleweed installation, so it seems to be a problem on your end. You probably tried to install a package not intended for Tumbleweed, or have some incompatible repos configured. git is included in the distribution, and should require the perl included in the distribution.
Never used Linphone though.

Do a lot of users try to find solutions for their difficulties by alternative means because of this fact?

Most that want to use the nvidia driver use a stable openSUSE release I suppose, as recommended.

I hope that more knowledge about usable version ranges will become available.

Ask nvidia to provide it. But they probably don’t know either in advance, when a new kernel or Xorg is being released that causes the driver to not work.

One of these versions worked on my system for a while before I dared to try another update out.

Well, yes. That’s the point. You will always get updates on Tumbleweed that might break the nvidia driver.
You have to update the driver then to the latest version, or wait until a new version (that works) is released by nvidia.

G03 is not expected to work at all at the moment in Tumbleweed, btw:
http://bugzilla.opensuse.org/show_bug.cgi?id=954725#c9
No idea about G04.

I have got an alternative view on the effects from corresponding software evolution.

Well.
Older versions are definitely not going to work on Tumbleweed.

I hoped to avoid this.

As you prefer.
If you don’t like to have hassles with nvidia driver, better switch to nouveau then.
But that has its problems as well, and might not be an option for you anyway.

How should be determined which of these modules is responsible for the display of the screen resolution “1280 x 1024” at the moment?

Look into /var/log/Xorg.0.log which graphics driver is used.
And/or post it so that we can see what’s going on, as I asked already.

I want to keep the software “Linux 4.1.1-1-desktop” running on my system until I know that alternative versions will also work together with other components as expected for my needs.

You didn’t tell that you have that installed.

That kernel is old, but ok. The nvidia driver should work with it.

It was also provided by this installation source for a while.

Yes, but kernel-desktop does not exist any more, neither in Tumbleweed nor in Leap.
This is the 13.2 version, as I wrote.

I would like to omit the repository list here at the moment because this information did not really help in similar clarification tries so far.

Do what you prefer.
But don’t expect proper help if you don’t even provide the requested information.
Can you at least post the URL of the used nvidia repo then?

  1. nvidia-computeG04-352.63-18.1.x86_64
  2. nvidia-gfxG04-kmp-desktop-352.63_k3.16.6_2-18.1.x86_64
  3. nvidia-glG04-352.63-18.1.x86_64
  4. nvidia-uvm-gfxG04-kmp-desktop-352.63_k3.16.6_2-18.1.x86_64
  5. x11-video-nvidiaG04-352.63-18.1.x86_64

Ok, at least there are no version mismatches.
And it is the latest version.

Are those all packages? Did you remove nvidia-gfxG04-kmp-default meanwhile?

I have tried this after noon.

    47.001] 
X.Org X Server 1.18.0
Release Date: 2015-11-09
    47.001] X Protocol Version 11, Revision 0
    47.002] Build Operating System: openSUSE SUSE LINUX
    47.002] Current Operating System: Linux Sonne 4.1.1-1-desktop #1 SMP PREEMPT Wed Jul 8 14:23:40 UTC 2015 (cac28b3) x86_64
    47.002] Kernel command line: root=/dev/duda/root resume=LABEL=SSHD-temp splash=silent quiet vga=0x31b nomodeset
    47.002] Build Date: 18 January 2016  07:18:00PM
    47.002]  
    47.002] Current version of pixman: 0.33.6
…
    47.182] (II) LoadModule: "glx"
    47.209] (II) Loading /usr/lib64/xorg/modules/extensions/libglx.so
    47.837] (II) Module glx: vendor="NVIDIA Corporation"
    47.837]     compiled for 4.0.2, module version = 1.0.0
    47.837]     Module class: X.Org Server Extension
    47.860] (II) NVIDIA GLX Module  352.63  Sat Nov  7 20:52:00 PST 2015
...
    48.263] (EE) NVIDIA: Failed to initialize the NVIDIA kernel module. Please see the
    48.263] (EE) NVIDIA:     system's kernel log for additional error messages and
    48.263] (EE) NVIDIA:     consult the NVIDIA README for details.
    48.480] (EE) [drm] Failed to open DRM device for pci:0000:04:00.0: -19
    48.694] (EE) [drm] Failed to open DRM device for pci:0000:04:00.0: -19
    48.694] (EE) open /dev/dri/card0: No such file or directory
...
    48.716] (II) FBDEV(0): Creating default Display subsection in Screen section
    "Default Screen Section" for depth/fbbpp 24/32
    48.716] (==) FBDEV(0): Depth 24, (==) framebuffer bpp 32
    48.716] (==) FBDEV(0): RGB weight 888
    48.716] (==) FBDEV(0): Default visual is TrueColor
    48.716] (==) FBDEV(0): Using gamma correction (1.0, 1.0, 1.0)
    48.716] (II) FBDEV(0): hardware: VESA VGA (video memory: 10240kB)
    48.716] (II) FBDEV(0): checking modes against framebuffer device...
    48.716] (II) FBDEV(0): checking modes against monitor...
    48.716] (--) FBDEV(0): Virtual size is 1280x1024 (pitch 1280)
    48.716] (**) FBDEV(0):  Built-in mode "current": 131.1 MHz, 80.3 kHz, 76.6 Hz
    48.716] (II) FBDEV(0): Modeline "current"x0.0  131.09  1280 1312 1472 1632  1024 1028 1032 1048 -hsync -vsync -csync (80.3 kHz b)
    48.716] (==) FBDEV(0): DPI set to (96, 96)
...
    48.772] (EE) Failed to initialize GLX extension (Compatible NVIDIA X driver not found)
…

Does this detailed information help really more for the desired problem resolution than my initial request?

Yes. It shows that the nvidia kernel module cannot be found, and the generic fbdev driver is being used instead of nvidia.

So either you didn’t correctly reinstall the driver, or something went wrong.
That version should definitely work with Kernel 4.1.1, so run this and post the output please:

sudo zypper in -f nvidia-gfxG04-kmp-desktop

As those packages are intended for 13.2, I’m not completely sure that they would work at all though.

Better try to install nvidia-gfxG04-kmp-default from the Leap repo, and try with kernel-default.

To which release name would a software like “Linux 4.1.1-1-desktop” belong in your view?

To none.

That kernel was once part of Tumbleweed, but is old and not available any more. It isn’t part of any openSUSE release.
Leap does come with 4.1.x, but originally had 4.1.12, which has been updated to 4.1.13 meanwhile (and there’s only kernel-default as I already mentioned, kernel-desktop does not exist any more since > 4.1.1).

I like this way for big parts of the provided software collection in principle. - I would appreciate if it will become possible to consider some special maintenance rules for my aging hardware.

It’s not about the age of your hardware, it’s about nvidia needing some time to update their driver to the latest Xorg/Kernel versions.

Can the Open Build Service help me to resolve special requirements for this use case?

Well, you can use it to build your own packages, that are updated automatically when a new kernel is released.
But it won’t magically “fix” the binary, closed-source nvidia driver.

You could of course also build an older Xorg there, but at the moment not even the kernel module can’t be loaded, so this won’t help either.

How “stable” can you really keep this component if you might notice that you will need the provided software corrections?

It is “stable” in that sense as that it is frozen at the release and doesn’t change any more. You get only bug fix and security updates, that are supposed to be compatible, and not cause problems with 3rd party applications and drivers like nvidia.

#19

Thanks for your general advice.

I find that it is helpful only in a limited way.

You can’t rely on the rpm as it will not be released same time with a new kernel update.

I know this detail. I am used to this to some degree.

You will need the gcc, make, kernel source and kernel devel packages to build the nvidia driver using the .run installer.

I select the Linux versions with which I fiddle around for better applications and software development occasionally by my special requirements.

#20

You should have those things installed anyway, the rpm packages require it too.

But, one thing regarding this: as you are using kernel-desktop-4.1.1 now, do you have the corresponding kernel-devel-4.1.1 and kernel-desktop-devel-4.1.1 installed as well?
If not, building the kernel module for 4.1.1 will not work obviously.

So please maybe also post the output of this to verify:

rpm -qa | grep kernel

If those are missing, you’d probably better switch to the current kernel-default again, as it won’t be easy to find the necessaryn 4.1.1 versions anywhere.