could not login after delete nvidia driver.

English Install/Boot/Login
1

I came across a problem in using the gnome3 in Leap 42.1, the error says
“Oh no! Something has gone wrong” and suggested me to logout. While logout
and reboot do not solve the probelm, I deleted the nvidia drivers
in icewm (at that time I could see the login window and could login with icewm),
but didnt run “mkinitrd”.

Now the system could not boot, it checked lots of service, all “ok”, but
freeze. No console.

In another disk, ubuntu is installed and I could login. I commented “blacklist nouveau”.
However, I still could not boot the system.

I think the Nouveau is still in the opensuse system, how could I re-enable it and come back
to my desktop?

Thanks!

2

How exactly did you delete them?
And why?

but didnt run “mkinitrd”.

Then the nvidia kernel module may still be in the initrd and may prevent nouveau from working properly.
Try to run it now.

Though your system should still work, just using a fallback driver.

The “Oh no! Something has gone wrong” error normally indicates a problem with OpenGL AFAIK.
Maybe there are some remnants of the nvidia driver left.

Please install “Mesa-demo-x” and run “glxinfo” and post the output.

Now the system could not boot, it checked lots of service, all “ok”, but
freeze. No console.

That’s bad. Does pressing “Ctrl+Alt+F1” get you to a console?

Try to add “nomodeset” to the boot options, does it boot then?
(press ‘e’ at the boot menu, and append “nomodeset” to the line starting with “linux” or “linuxefi”, then press F10 to boot)

Or does booting to text mode work?
Add ‘3’ or even ‘1’ to the boot options.
If all else fails, you could also try “init=/bin/sh” which boots into a minimal text mode system with no system services started at all, but at least you/we may be able to repair your system then.

I think the Nouveau is still in the opensuse system, how could I re-enable it and come back
to my desktop?

There should be no need to re-enable it, it should be used automatically.
That is, unless you created an /etc/X11/xorg.conf/ that tells Xorg to load the nvidia driver. If you have such a file, delete it.

Otherwise, please post you Xorg log.
Normally it should be located in /var/log/, but not if you use gdm. Unfortunately I don’t know exactly where gdm stores it, I think in ~/.local/share/.

3

These are the things to do:

  • Make sure nvidia’s completely gone
  • Remove the blacklisting of nouveau. Mind, when experimenting it’s possible to get a separate 50-blacklist-nouveau.conf in /etc/modprobe.d
  • Reinstall Mesa ( it contains the openGL stuff that the nvidia blob replaces with symlinks to it’s own version )
  • run ‘mkinitrd’
4

How exactly did you delete them?
And why?
I entered the icewd in the login window, start yast, and deleted the two packages related to nvidia.
Reasons: goolged the error message, some people said that it might be nvidia driver problem.
Previously the nouveau works in my computer, so I thought it is a good idea to go back to nouveau.

Then the nvidia kernel module may still be in the initrd and may prevent nouveau from working properly.
Try to run it now.
run mkinitrd as root, there are some warnings. After reboot, I still could not get the graphical session, but the font looks smaller,
it works partially.

That’s bad. Does pressing “Ctrl+Alt+F1” get you to a console?
Thanks! This help me to login in text mode, where I run mkinitrd and other commands.

Please install “Mesa-demo-x” and run “glxinfo” and post the output.
Mesa-demo-x installed, and glxinfo gave the following error:
error while loading shared library: libnvidia-tls-so.367.44, cannot open shared object file:
no such file or directory.
I guess because I deleted the nvidia driver.

Try to add “nomodeset” to the boot options, does it boot then?
Does not boot correctly. Freezed in the “started locale service”, as without the “nomodeset”.

Add ‘3’ or even ‘1’ to the boot options.
If all else fails, you could also try “init=/bin/sh” which boots into a minimal text mode system with no system services started at all, but at least you/we may be able to repair your system then.
“ctrl-atl-f1” could send me to the console and I could login, so I do not need 3 or 1 or init?

That is, unless you created an /etc/X11/xorg.conf/ that tells Xorg to load the nvidia driver. If you have such a file, delete it.
There is no folder of xorg.conf, but xorg.conf.d. I deleted all files generated nvidia in X11 and xorg.conf.d

Otherwise, please post you Xorg log.
Normally it should be located in /var/log/, but not if you use gdm. Unfortunately I don’t know exactly where gdm stores it, I think in ~/.local/share/
/var/log/ and ~/.local/share/ are empty. journaltcl gave me a lot of logs, do you know how to specify the xorg log? google do not give a proper answer.

Summary: I could login in text mode, could install softwares.
Just got problem with the graphic card display. I have gtx950
card. What should I do next?

5

Two packages?
The driver consists of 3, 4 or 5 packages, depending on the version.

Please check that there are no nvidia packages left:

rpm -qa | grep nvidia

Normally uninstalling the packages should be enough to switch back to nouveau.

Reasons: goolged the error message, some people said that it might be nvidia driver problem.

What error message?
“Oh no something went wrong”?
This indicates a generic problem, likely related to broken OpenGL support.
So yes, it was supposedly related to the nvidia driver, but that may be fixable.

Please install “Mesa-demo-x” and run “glxinfo” and post the output.
Mesa-demo-x installed, and glxinfo gave the following error:
error while loading shared library: libnvidia-tls-so.367.44, cannot open shared object file:
no such file or directory.
I guess because I deleted the nvidia driver.

Ok, so as suspected, there still seem to be parts of the nvidia driver on your system.

Add ‘3’ or even ‘1’ to the boot options.
If all else fails, you could also try “init=/bin/sh” which boots into a minimal text mode system with no system services started at all, but at least you/we may be able to repair your system then.
“ctrl-atl-f1” could send me to the console and I could login, so I do not need 3 or 1 or init?

No.
So your system is not frozen at all.
Just Xorg fails to start.

That is, unless you created an /etc/X11/xorg.conf/ that tells Xorg to load the nvidia driver. If you have such a file, delete it.
There is no folder of xorg.conf, but xorg.conf.d. I deleted all files generated nvidia in X11 and xorg.conf.d

xorg.conf would be a file.

You should not delete anything in xorg.conf.d, nvidia doesn’t generate anything in there.

Otherwise, please post you Xorg log.
Normally it should be located in /var/log/, but not if you use gdm. Unfortunately I don’t know exactly where gdm stores it, I think in ~/.local/share/
/var/log/ and ~/.local/share/ are empty. journaltcl gave me a lot of logs, do you know how to specify the xorg log? google do not give a proper answer.

According to Google, “journalctl _COMM=Xorg” or “journalctl /usr/bin/Xorg” should work.I do remember reading of a file where gdm stores the log, but I can’t remember where that was. Maybe in /var/run/ or similar then.

Summary: I could login in text mode, could install softwares.
Just got problem with the graphic card display. I have gtx950
card. What should I do next?

As mentioned, there likely are parts of the driver still installed.
Removing that should fix your problem.

You chould be able to get to a graphical system by setting DISPLAYMANAGER=“xdm” in /etc/sysconfig/displaymanager if you feel uncomfortable in text mode. This would also give us a proper /var/log/Xorg.0.log.
(you probably should set DEFAULT_WM=“icewm” in /etc/sysconfig/windowmanager as well, to use the simple IceWM instead of GNOME, which likely won’t start)

6

Thanks to wolfi323 and Knurpht!

  • I deleted two additional nvidia packages
  • force install Mesa
  • mkinitrd

Now I could see the login windows, again I could login icewm, but not gnome.
The gnome gave the following error with white background:
“”"
Oh no! Something has gone wrong.
A problem has occurred and system cant recover.
Please logout and try again.
“”"

I found many log file under /var/log/ in the xterm in iceWM, but firefox doesnt work in iceWM.
Do you know an easier way to post the file? Is the Xorg.0.log (newest file) the right file to post?

7

The latter two shouldn’t be necessary when installing/uninstalling the nvidia rpm packages from the repo.
But then it doesn’t hurt either.

I found many log file under /var/log/ in the xterm in iceWM, but firefox doesnt work in iceWM.

Hm, it should work.
But maybe it is also affected by the remaining graphics driver problem like GNOME.

Do you know an easier way to post the file?

Install the package “susepaste” and run:

susepaste /var/log/Xorg.0.log

Then please post the link it outputs.

Is the Xorg.0.log (newest file) the right file to post?

Yes, that’s the one from the current boot.

8

Here are links:
http://susepaste.org/20931562
http://paste.opensuse.org/20931562

with errors/warnings:

  1. 879.470] (WW) Warning, couldn’t open module nvidia

  2. 879.470] (EE) Failed to load module “nvidia” (module does not exist, 0)

879.487] (EE) Unknown chipset: NV126

These are interesting, because I got the same white screen with the “Oh no” message when I used
the nvidia driver.

9

These are normal when the nvidia driver is not installed. (Xorg tried to load nvidia first, and then falls back to another driver if it isn’t installed)

But this seems to be the problem:

879.487] (EE) Unknown chipset: NV126

Apparently your chipset isn’t supported by nouveau, and this also breaks OpenGL with the “modesetting” driver that’s being used instead.
See https://bugs.freedesktop.org/show_bug.cgi?id=94728

If you add “nomodeset” now, GNOME should work at least, using software OpenGL rendering (which will be slow though).
Though it should work without it too (but also with software OpenGL only).
So the glxinfo output would be interesting here.

Your best bet is getting the nvidia driver to work though, I suppose.
The G04 driver should support your card.
I would recommend trying to install it again, by explicitly specifying the packages:

sudo zypper in x11-video-nvidiaG04 nvidia-gfxG04-kmp-default nvidia-glG04 nvidia-computeG04

At least xdm/IceWM should work even if GNOME doesn’t, in the latter case it would be nice to see the Xorg.0.log and glxinfo then to see what’s wrong.

10

New Xorg.0.log
http://paste.opensuse.org/43055894

and glxinfo
http://paste.opensuse.org/35675307

No Error, but I still get the “Oh no” in gnome.

I didnt run mkinitrd after installing nvidia driver, but I
saw the installation do it for me.

11

Right, the nvidia driver is used and fully working, including OpenGL.

IMHO, this is rather a problem of your GNOME installation, unrelated to the graphics driver.
I don’t have any experience with GNOME though.

Can you run Firefox in IceWM now?
If not, it may point to the problem (Firefox is a GTK application).
Run it in xterm, and post the output.

Also try to create a new user account and login there.
That would show if it is a system-wide problem, or something user-specific (wrong settings, permissions, …)

12

OK, try it tomorrow in the office.

BTW, I have btrfs format for the “/”, could I rollback to a previous snapshot?
How to do it exactly?

13

Yes, if you didn’t disable snapshots.

How to do it exactly?

You should be able to boot a previous snapshot by choosing it in the boot menu.
If it boots successfully, you’d just need to run “snapper rollback” to revert your system to that state.

See SUSE Documentation

14

===
firefox woks in icewm,
messages from xterm.
1476345901552 addons.webextension.gmailnoads@mywebber.com WARN Loading extension ‘gmailnoads@mywebber.com’: Reading manifest: Error processing background.persistent: Event pages are not currently supported. This will run as a persistent background page.
1476345901556 addons.webextension.gmailnoads@mywebber.com WARN Loading extension ‘gmailnoads@mywebber.com’: Reading manifest: Error processing options_page: An unexpected property was found in the WebExtension manifest.
1476345901617 addons.webextension.Clipper@OneNote.com WARN Please specify whether you want browser_style or not in your browser_action options.
console.warn: nsLoginManager: searchLogins: A hostname is recommended
console.warn: nsLoginManager: searchLogins: formSubmitURL or httpRealm is recommended

===
Is the roll back options in advanced setting? I didnt see any snapshots.

===
The grub fonts are ugly huge and unclear. I tried to set the resolution in bootloader of yast, via the vga mode, but has no effect.
Are there other places I should modify?

===
Here is the snapshot I got, could I roll back?
The grub fonts is Type | # | Pre # | Date | User | Cleanup | Description | Userdata
-------±----±------±---------------------------------±-----±--------±----------------------±-------------
single | 0 | | | root | | current |
single | 1 | | Thu 29 Sep 2016 11:40:02 AM CEST | root | | first root filesystem |
single | 2 | | Thu 29 Sep 2016 11:54:42 AM CEST | root | number | after installation | important=yes
pre | 7 | | Thu 29 Sep 2016 12:12:54 PM CEST | root | number | zypp(zypper) | important=yes
post | 8 | 7 | Thu 29 Sep 2016 12:16:38 PM CEST | root | number | | important=yes
pre | 118 | | Tue 04 Oct 2016 03:04:03 PM CEST | root | number | zypp(y2base) | important=yes
post | 119 | 118 | Tue 04 Oct 2016 03:04:14 PM CEST | root | number | | important=yes
pre | 270 | | Wed 12 Oct 2016 06:03:50 PM CEST | root | number | zypp(zypper) | important=no
post | 271 | 270 | Wed 12 Oct 2016 06:03:53 PM CEST | root | number | | important=no
pre | 274 | | Wed 12 Oct 2016 06:38:47 PM CEST | root | number | zypp(zypper) | important=no
post | 275 | 274 | Wed 12 Oct 2016 06:42:09 PM CEST | root | number | | important=no
pre | 276 | | Thu 13 Oct 2016 09:34:55 AM CEST | root | number | yast bootloader |
post | 277 | 276 | Thu 13 Oct 2016 09:36:32 AM CEST | root | number | |
pre | 278 | | Thu 13 Oct 2016 09:43:05 AM CEST | root | number | yast bootloader |
post | 279 | 278 | Thu 13 Oct 2016 09:44:55 AM CEST | root | number | |
pre | 280 | | Thu 13 Oct 2016 09:52:16 AM CEST | root | number | zypp(zypper) | important=no
post | 281 | 280 | Thu 13 Oct 2016 09:52:17 AM CEST | root | number | | important=no
pre | 282 | | Thu 13 Oct 2016 09:54:54 AM CEST | root | number | zypp(zypper) | important=no
post | 283 | 282 | Thu 13 Oct 2016 09:54:57 AM CEST | root | number | | important=no
pre | 284 | | Thu 13 Oct 2016 10:00:51 AM CEST | root | number | yast bootloader |
post | 285 | 284 | Thu 13 Oct 2016 10:02:46 AM CEST | root | number |

===
I created a new new user, showing the same error…

15

Ok, then maybe try to run “gnome” instead, maybe this will give a clue.

Is the roll back options in advanced setting? I didnt see any snapshots.

No, you should have an entry “Boot a previous snapshot” (or similar) in the main menu.
But this only works if /boot is on your btrfs partition and not a separate one.

The grub fonts are ugly huge and unclear. I tried to set the resolution in bootloader of yast, via the vga mode, but has no effect.
Are there other places I should modify?

The “vga mode” only applies to the text-mode console, not grub.
Further down the page should be an option to change grub’s resolution.
It should be “Autodetect” by default, but in my experience this doesn’t work with nvidia. Set it to some explicit resolution.

Here is the snapshot I got, could I roll back?

Yes, should be possible too, by specifying the snapshot number in the “snapper rollback” command.
Something like:

sudo snapper rollback 2

But to be honest, I never did this myself.

I created a new new user, showing the same error…

Ok, then it seems to be an installation problem.
Can you post your repo list, please?

zypper lr -d
16

@wolfi323
The grub setting is correctly done.

I found the snapshot, after select one, it said,“if ok run snapper rollback and reboot”.
Nothing happened.

repo list

| Alias | Name | Enabled | GPG Check | Refresh | Priority | Type | URI | Service

—±------------------------------------±----------------------------------------±--------±----------±--------±---------±-------±-----------------------------------------------------------------------------------------------±-------
1 | download.nvidia.com-leap | nVidia Graphics Drivers | Yes | (r ) Yes | Yes | 99 | rpm-md | http://download.nvidia.com/opensuse/leap/42.1 |
2 | download.opensuse.org-non-oss | Main Repository (NON-OSS) | Yes | (r ) Yes | Yes | 99 | yast2 | http://download.opensuse.org/distribution/leap/42.1/repo/non-oss/ |
3 | download.opensuse.org-non-oss_1 | Update Repository (Non-Oss) | Yes | (r ) Yes | Yes | 99 | rpm-md | http://download.opensuse.org/update/leap/42.1/non-oss/ |
4 | download.opensuse.org-oss | Main Repository (OSS) | Yes | (r ) Yes | Yes | 99 | yast2 | http://download.opensuse.org/distribution/leap/42.1/repo/oss/ |
5 | download.opensuse.org-oss_1 | Main Update Repository | Yes | (r ) Yes | Yes | 99 | rpm-md | http://download.opensuse.org/update/leap/42.1/oss |
6 | http-download.opensuse.org-59bde11d | home:snwh:moka | Yes | (r ) Yes | Yes | 99 | rpm-md | http://download.opensuse.org/repositories/home:/snwh:/moka/openSUSE_Leap_42.1/ |
7 | http-download.opensuse.org-5b0e9cbd | home:Lazy_Kent | Yes | (r ) Yes | Yes | 99 | rpm-md | http://download.opensuse.org/repositories/home:/Lazy_Kent/openSUSE_Leap_42.1/ |
8 | http-download.opensuse.org-d7176856 | home:Horst3180 | Yes | (r ) Yes | Yes | 99 | rpm-md | http://download.opensuse.org/repositories/home:/Horst3180/openSUSE_Leap_42.1/ |
9 | http-download.opensuse.org-f12175f5 | filesystems | Yes | (r ) Yes | Yes | 99 | rpm-md | http://download.opensuse.org/repositories/filesystems/openSUSE_Leap_42.1/ |
10 | openSUSE-42.1-0 | openSUSE-42.1-0 | No | ---- | Yes | 99 | yast2 | hd:///?device=/dev/disk/by-id/usb-Kingston_DataTraveler_SE9_001731897888BB917000007D-0:0-part2 |
11 | repo-debug | openSUSE-Leap-42.1-Debug | No | ---- | Yes | 99 | NONE | http://download.opensuse.org/debug/distribution/leap/42.1/repo/oss/ |
12 | repo-debug-non-oss | openSUSE-Leap-42.1-Debug-Non-Oss | No | ---- | Yes | 99 | NONE | http://download.opensuse.org/debug/distribution/leap/42.1/repo/non-oss/ |
13 | repo-debug-update | openSUSE-Leap-42.1-Update-Debug | No | ---- | Yes | 99 | NONE | http://download.opensuse.org/debug/update/leap/42.1/oss |
14 | repo-debug-update-non-oss | openSUSE-Leap-42.1-Update-Debug-Non-Oss | No | ---- | Yes | 99 | NONE | http://download.opensuse.org/debug/update/leap/42.1/non-oss/ |
15 | repo-source | openSUSE-Leap-42.1-Source | No | ---- | Yes | 99 | NONE | http://download.opensuse.org/source/distribution/leap/42.1/repo/oss/ |
16 | repo-update | openSUSE-Leap-42.1-Update | Yes | (r ) Yes | Yes | 99 | rpm-md | http://download.opensuse.org/update/leap/42.1/oss/ |
17 | vivaldi | vivaldi | Yes | ( p) Yes | Yes | 99 | rpm-md | http://repo.vivaldi.com/archive/rpm/x86_64 |
~
~
~

17

On the journalctl log, I found something interesting:

Oct 13 12:35:58 a240-pc100 gnome-session[2205]: WARNING: Unable to find required component 'gnome-shell’
Oct 13 12:35:58 a240-pc100 gnome-session[2205]: gnome-session[2205]: WARNING: Unable to find required component ‘gnome-shell’
Oct 13 12:35:58 a240-pc100 gnome-session[2205]: Entering running state
Oct 13 12:36:01 a240-pc100 org.a11y.atspi.Registry[2254]: XIO: fatal IO error 11 (Resource temporarily unavailable) on X server “:0”
Oct 13 12:36:01 a240-pc100 org.a11y.atspi.Registry[2254]: after 25 requests (25 known processed) with 0 events remaining.
Oct 13 12:36:01 a240-pc100 org.gtk.vfs.Daemon[2227]: A connection to the bus can’t be made
Oct 13 12:36:14 a240-pc100 org.a11y.Bus[2503]: Activating service name=‘org.a11y.atspi.Registry’
Oct 13 12:36:14 a240-pc100 org.a11y.Bus[2503]: Successfully activated service ‘org.a11y.atspi.Registry’
Oct 13 12:36:14 a240-pc100 org.a11y.atspi.Registry[2512]: SpiRegistry daemon is running with well-known name - org.a11y.atspi.Registry
Oct 13 12:36:14 a240-pc100 org.a11y.atspi.Registry[2512]: Xlib: extension “XEVIE” missing on display “:0”.

However, I have gnome-shell installed

rpm -qa | grep “gnome-shell”
gnome-shell-calendar-3.16.4-1.2.x86_64
gnome-shell-lang-3.16.4-1.2.noarch
gnome-shell-search-provider-gnome-terminal-3.16.2-2.4.x86_64
gnome-shell-search-provider-seahorse-3.16.0-2.6.x86_64
gnome-shell-search-provider-gnome-calculator-3.16.2-2.4.x86_64
gnome-shell-browser-plugin-3.16.4-1.2.x86_64
gnome-shell-search-provider-nautilus-3.16.3-5.1.x86_64
gnome-shell-search-provider-contacts-3.16.2-2.4.x86_64
gnome-shell-3.16.4-1.2.x86_64
gnome-shell-search-provider-documents-3.16.5-7.2.x86_64

18

Just to be clear: you need to press RETURN to boot the selected snapshot.

repo list

Seems to be ok.
The problem may have been caused by mixing (GNOME) components from different repos, but I don’t see anything obvious in the repos you have.

Yes, this may point to something, though I have no idea to what (I have basically no experience with GNOME).
The “XEVIE missing” error does ring a bell, but I don’t think that’s a problem.

What does this command say?

rpm -V gnome-shell

Maybe some necessary file got deleted somehow.
If the command doesn’t output anything, the package is fine though.

If it does report missing files, reinstall the package:

sudo zypper in -f gnome-shell
19

I do press RETURN, and nothing happened

Seems to be ok.
The problem may have been caused by mixing (GNOME) components from different repos, but I don’t see anything obvious in the repos you have.

Yes, this may point to something, though I have no idea to what (I have basically no experience with GNOME).
The “XEVIE missing” error does ring a bell, but I don’t think that’s a problem.

What does this command say?

rpm -V gnome-shell

Maybe some necessary file got deleted somehow.
If the command doesn’t output anything, the package is fine though.

If it does report missing files, reinstall the package:

sudo zypper in -f gnome-shell

I force re-installed, still white background with the error message.

rpm -V gnome-shell

show nothing with sudo.

I am planning a reinstall…

20

Well, as already suggested, I would try to run “gnome” inside IceWM (in a terminal window), maybe there’s some more obvious error message then.
Or maybe “gnome-shell” itself, you may need to use “gnome-shell --replace” though.