Received a notification this morning that an important automatic update was available. The files appeared to all be Nvidia driver related.
I allowed the install, rebooted and received an error message saying that, “Oh no! Something has gone wrong. A problem has occurred and the system can’t recover. Please log out and try again”. When I press the “Log Out” button on the screen (by clicking enter) I get a blank screen. I have tried rebooting into recovery for both the current kernel (3.11.10-21-desktop) and the previous kernel(3.11.10-17-desktop) and in both cases ended up at the original error message.
I am running 13.1 on a Thinkpad T61 that has an Nvidia NVS Quadro 140M graphics chip.
Any suggestions for how to get my system back?
Thanks.
On 2014-08-17 02:26, portsample wrote:
>
> Received a notification this morning that an important automatic update
> was available. The files appeared to all be Nvidia driver related.
> I allowed the install, rebooted and received an error message saying
> that, “Oh no! Something has gone wrong. A problem has occurred and the
> system can’t recover. Please log out and try again”.
That message is from Gnome, and it typically indicates a problem with
the graphics driver, that gnome is not happy with it.
The way out is to go to text mode by pressing ctrl-alt-f1, and login
there as root.
Once there, first run:
init 3
If that does not work, you have to boot in text mode instead:
Main grub 2 display
Adding 3 to linux line
Then you have to edit the file “/etc/sysconfig/displaymanager”; I
suggest you use “joe”, or mcedit (if installed). Alternatives are “vi”.
(mcedit comes with “mc”, which I recommend you install in advance for
times such as these. Simply run in the terminal:
zypper in mc
mcedit is a very simple text editor with menu, and ‘mc’ is a very
powerful file browser, which can also edit files at the cursor).
On that file, you will see something like this:
## Path: Desktop/Display manager
## Description: settings to generate a proper displaymanager config
## Type: string(kdm,xdm,gdm,wdm,console)
## Default: ""
## Config: xdm,kdm,gdm
#
# Here you can set the default Display manager (kdm/xdm/gdm/wdm/console).
#
DISPLAYMANAGER="lightdm"
You probably have there “gdm”, which will not work, being part of gnome.
Use any other you have installed… lightdm is a good choice. Maybe you
have to install it.
Once installed, run:
init 5
which should give you back a graphical login screen. The next problem is
that you can not login on gnome, you have to use any other desktop you
have. There is usually a very spartan desktop installed for these
ocassions, I forget the name. Ugly, but it will let you work on repair
the video driver “problem”.
(or you can install XFCE, which is reasonably good)
On that part (nvidia), let somebody else turn in and help - or browse
these forums for that “Oh no! Something has gone wrong.” message, you
will see it more than once.
–
Cheers / Saludos,
Carlos E. R.
(from 13.1 x86_64 “Bottle” at Telcontar)
Thanks, however I am slightly confused here:
"If that does not work, you have to boot in text mode instead:
Main grub 2 display
Adding 3 to linux line"
…entering “Main grub 2 display” causes an error message on “Main”, entering “grub 2 display” gets me to a “grub>” prompt…what am I missing?
Thanks for your patience.
On 2014-08-17 03:26, portsample wrote:
>
> Thanks, however I am slightly confused here:
>
> “If that does not work, you have to boot in text mode instead:
>
> ’ Main grub 2 display’ (http://susepaste.org/57493157)
> ’ Adding 3 to linux line’ (http://susepaste.org/90114364)”
>
> …entering “Main grub 2 display” causes an error message on “Main”,
> entering “grub 2 display” gets me to a “grub>” prompt…what am I
> missing?
Those are just the tittles of the photos… Just look at the photos and
do the same as you see there.
–
Cheers / Saludos,
Carlos E. R.
(from 13.1 x86_64 “Bottle” at Telcontar)
On 2014-08-17 03:26, portsample wrote:
>
> Thanks, however I am slightly confused here:
>
> “If that does not work, you have to boot in text mode instead:
>
> ’ Main grub 2 display’ (http://susepaste.org/57493157)
> ’ Adding 3 to linux line’ (http://susepaste.org/90114364)”
Photo of main grub 2 display
Photo of adding 3 to Linux line
Just link to photos, not something for you to type.
–
Cheers / Saludos,
Carlos E. R.
(from 13.1 x86_64 “Bottle” at Telcontar)
Are you typing “Main grub 2 display”
Easy pezee at boot press e that gets to edit mode find the line that starts linux… Be aware that this is a long line and thus covers several on screen lines. go to the true end of the line (note press End key is best way) enter a space and then a 3 press F10 to boot this will boot you to a terminal. You can probably fix things here but first need to understand what went wrong. What I would do is log in as root then run yast - software management search for nvidia see what is actually installed. Latley we have seen problems here with kernel updates and NVIDIA drivers. You should only have 5 packages with nvidia in the name. (note other things will show up we only have interest in nvidia named packages. I beleive that you would be using the 03 packages so all the packages should have an 03 in the name. Also we need be sure that the right packages for your kernel are being used. If you have made no manual changes in the kernel that comes with 13.1 then you have the desktop kernel. If you see packages with default/xen/ec2 then remove that nvidia package.
Then search for kernel make sure you have only the desktop flavor of kernel stuff installed. Uninstall any kernel that has other then desktop in the name. removing a kernel will also remove any dependencies
For me a recent kernel update dragged in some incorrect files for my system luckely I pay attention and noticed that odd packages were being installed so I fixed things before reboot. In my case I got the default kernel and some default nvidia files that were not needed or wanted.
Thanks.
Using vi I have changed “gdm” to “xfce” in /etc/sysconfig/displaymanager
However I am still unclear regarding how to edit the grub2 file shown in the link.
EDIT: GOT IT…editing Grub2 now…standby…
Thanks again for your patience.
I am pretty sure xfce is not a display manager, sure you aren’t looking for lightdm?
Roger that- I’ve switched to lightdm. Thanks.
I’m still slightly hung up w/editing the grub file…
Okay, I’ve changed “gdm” to “lightdm” in /etc/sysconfig/displaymanager, and added a “3” at the end of the “linux line” in grub edit (pressing “e” during boot, the adding the “3” then pressing F10). I then get the openSuse “chameleon and beanstalk” on the right with no list of users, then the screen goes black and I’m looking at a terminal with,“Welcome to openSuse 13.1 “Bottle” - Kernel 3.11.10-21-desktop (tty1)
ajaxlinux-user login:”
However, if I login as root and enter “startx” I get infinitely scrolling error messages that say,“No changes for /etc/X11/xdm/Xservers
No changes for /etc/X11xdm/xdm-config
Using MD5DIR=”/var/adm/SuSEconfig/md5"…"
Thanks again for your collective wisdom and patience.
EDIT: …inserting “init 5” rather than startx in the above example produces a black screen with a very dark and greyed out chameleon and beanstalk on the right. Thanks again.
Got it.
zypper install -f xorg-x11-server
New Nvidia files are now happy and productive. Thanks.
There was a bug in the updated nvidia packages that broke GLX, which GNOME absolutely requires as mentioned.
https://bugzilla.novell.com/show_bug.cgi?id=892251
This should be fix in the mantime already (they fixed it on a Sunday this time, yay! 
)
So install the updated nvidia driver packages and GNOME should work again.
On 2014-08-17 05:06, portsample wrote:
>
> Okay, I’ve changed “gdm” to “lightdm” in /etc/sysconfig/displaymanager,
> and added a “3” at the end of the “linux line” in grub edit (pressing
> “e” during boot, the adding the “3” then pressing F10).
Wait, that was not what I said, or what I wanted to say. You misunderstood.
The idea of changing the grub line was to reach a working text mode
system, where you can edit “/etc/sysconfig/displaymanager”.
As you already edited “/etc/sysconfig/displaymanager”, why do you want
to change grub now? It is pointless, you do not need it.
You just needed:
init 3
edit file, install components missing
init 5, or startx to test.
> I then get the
> openSuse “chameleon and beanstalk” on the right with no list of users,
> then the screen goes black and I’m looking at a terminal
> with,
> “Welcome to openSuse 13.1 “Bottle” - Kernel
> 3.11.10-21-desktop (tty1)
> ajaxlinux-user login:”
> However, if I login as root and enter “startx” I get infinitely
> scrolling error messages that say,
>
> “No changes for /etc/X11/xdm/Xservers
> No changes for /etc/X11xdm/xdm-config
> Using MD5DIR=”/var/adm/SuSEconfig/md5"…"
Mmm. Something got more broken than usual in the NVidia update, it seems.
On 2014-08-17 06:16, portsample wrote:
> Got it.
>
> Code:
> --------------------
> zypper install -f xorg-x11-server
> --------------------
>
> New Nvidia files are now happy and productive. Thanks.
IIRC, NVidia replaces some files in the xorg-x11-server package, and
viceversa. So installing, as you did, the xorg-x11-server package again
overwrites some nvidia components - but it is necessary as a temporary
measure, apparently - till they repair the NVidia package, which
wolfi323 says has been done already.
I’d suggest, as a precaution, to keep safe a copy of the working NVidia
rpms, because if after an update things break, you can not revert to an
older driver by normal means, because the old versions are sometimes
removed from the repository.
–
Cheers / Saludos,
Carlos E. R.
(from 13.1 x86_64 “Bottle” at Telcontar)
The “Sunday” Nvidia patch is broken as well.
I’ve just installed it and ended up with the same situation as yesterday. Boot to “chameleon and beanstalk” with nothing after this…no window manager box. Did a power-off reboot, then added “3” to the end of the linux line in the grub file, logged in as root and enter “startx”, got the infinitely scrolling error messages that say,“No changes for /etc/X11/xdm/Xservers
No changes for /etc/X11xdm/xdm-config
Using MD5DIR=”/var/adm/SuSEconfig/md5"…"
Reboot again, login as root then as before reinstalled x-server files and mappings via,
zypper install -f xorg-x11-server
and then everything is fine again.
Whew baby.
Well, according to the bug report it should be fixed. And the files in the repo do have a changedate of today, and the changelog entry:
* Sat Aug 16 2014 sndirsch@suse.com
- fixed installation of libglx on %suse_version < 1315 (regression)
And if there’s a problem with the nvidia driver packages, re-installing xorg-x11-server shouldn’t actually fix it anyway.
One thing I am aware of though, is that the nvidia driver packages don’t work at first boot after installation/update (on some systems at least), but they work fine afterwards.
So maybe the reboot actually “fixed” your problem now, and not the re-installation of xorg-x11-server?
Could you please post which driver you are actually using now?
You didn’t uninstall the nvidia driver IIUYC, right?
Please install the package “Mesa-demo-x” if it’s not installed, and post the output of:
glxinfo | grep render
That happened to me today on my factory system. Everything was fine, until I tried to login to Gnome. Then I got the “Oh no!” screen. And the logout from there did not work, so I had to hit CTRL-BACKSPACE twice to exit the hung Gnome session.
Looking at the error indicators, there seemed to be a problem with GLX.
There are actually a couple of bug reports on this:
Bug 892251 (opensuse 12.3)
Bug 892253 (opensuse 13.1)
The bug reports give some hints at workarounds.
In my case, I just had to setup a suitable symlink, and all was fine once again. I have not reported a bug for factory on this.
Yeah, right. I did post a link to the latter one.
But according to the reports (both) the problem should be fixed.
(I do not know for sure, because I’m not near any nvidia systems right now)
I have not reported a bug for factory on this.
Well, there are no packages for Factory anyway, so it makes no sense to report it for Factory.
Are you using the 13.1 packages on Factory, or what?
I installed Nvidia 304.123 the hard way. When I checked the glx library, it looks is if it is now trying to use “update-alternatives” instead of a direct symlink to the nvidia library. But the update-alternatives implementation seems half-baked.
But then you cannot be affected by an openSUSE packaging bug (which is the content of those bug reports).
Why would you even report a bug to openSUSE then? 
OTOH, why would you be affected by an update to openSUSE’s driver packages then?
When I checked the glx library, it looks is if it is now trying to use “update-alternatives” instead of a direct symlink to the nvidia library. But the update-alternatives implementation seems half-baked.
Yeah, right. There seems to have been a similar problem with the openSUSE packages as well.
But there it was caused by a bug in the spec file apparently.
If the nvidia installer does something wrong, you should report it to nvidia. (the openSUSE packages don’t use the nvidia installer, they extract the files and install them manually)
I wasn’t even aware that the nvidia installer now uses update-alternatives though. Must be a recent change…
Okay, let me give a little more detail to clarify.
This is what I initially saw:
% cd /usr/lib64/xorg/modules/extensions
% ls -l
total 10968
lrwxrwxrwx 1 croot root 17 Aug 17 06:42 libglx.so -> /etc/alternatives/libglx.so
-rwxr-xr-x 1 croot root 10527048 Aug 16 17:54 libglx.so.304.123
-rwxr-xr-x 1 croot root 693528 Aug 6 05:32 libvnc.so
drwxr-xr-x 2 croot root 4096 Aug 17 06:42 xorg
and this is what I now see after my work-around:
% ls -l
total 10968
lrwxrwxrwx 1 croot root 17 Aug 17 12:13 libglx.so -> libglx.so.304.123
-rwxr-xr-x 1 croot root 10527048 Aug 16 17:54 libglx.so.304.123
lrwxrwxrwx 1 croot root 27 Aug 17 06:42 libglx.so.orig -> /etc/alternatives/libglx.so
-rwxr-xr-x 1 croot root 693528 Aug 6 05:32 libvnc.so
drwxr-xr-x 2 croot root 4096 Aug 17 06:42 xorg
Before I made those changes, I did an “update-alternatives --query libglx.so”, but it only showed the one option (no alternative).