how do I get back to 13.2 from tumbleweed

English Install/Boot/Login
1

I used zypper as described to move to tumbleweed. X fails to start “xf86OpenConsole VT_ACTIVATE failed operation not permitted” in the Xorg log (nvidia driver used).
Also get startup message “systemmd-fstab-generator[108] failed to create mount unit file /run/systemd/generator/sysroot.mount as it already exists”. There are no such files and /etc/fstab has no duplicates.

I had not noticed the fine print on tumbleweed - that it was high maintenance. I just want to use the computer, not play with the OS. Is there any simple way back to 13.2, without loosing all settings and files etc.?

I have burned a 13.2 install disc, but it cannot ‘upgrade’ to its old versions.

Thanks

2

The 2nd problem you should be able to ignore. It is your nvidia driver that is the problem as I suspect as this does not work with the newest kernel (yet) as included with the last tumbleweed issue. Tumbelweed it pretty stable at the moment - although by the very nature of it can go wobbly in the future.

As a temporary solution it might be best to remove the nvidia driver and stick with tumbleweed until the new version - leap - is issued tomorrow. It will be an easier system change as the software is broadly similiar. I would suggest that you make a backup of your data in case it goes wrong.

3

I would suggest to switch back all your repos to the 13.2 ones and then run “zypper dup” (either in YaST->Software Repositories, or by editing the files in /etc/zypp/repos.d/ by hand)
Even if not supported it should work. You might have a few problems afterwards, like (mainly GTK) applications not showing correct fonts, or awk not being available, but just ask and I will tell you how to rectify this.

Or I can tell you already here anyway, run this afterwards:

zypper in -f gawk
/usr/bin/pango-querymodules --update-cache
/usr/bin/pango-querymodules-64 --update-cache

The second line might give an error message on 64bit system, the third one on a 32bit system; please run them anyway and ignore those messages… :wink:

At least that are the two problems I am aware of when downgrading from Tumbleweed from 13.2.

There might be more meanwhile, if you encounter some, please ask. It should be solveable… :wink:

4

the same way you went to tumbleweed, change your repo list and do a dup
but Leap is out you can do a switch to leap
basically you need to disable or remove the tw repo’s with zypper or yast add the Leap or 13.2 repo’s then do a zypper dup, I’d suggest doing it from run level 3

5

Thanks for your suggestions folks.

I installed leap from DVD and added nvidia rep 42.1 and did zypper inr among a lot of other things, to no avail.

The various log messages say X fails to start. I tried everything (except experts only bit) in the SDB configuring graphics cards and the SDB on NVIDIA drivers, but X reliably fails to start.

Xorg -configure seg faults in the module intel_drv.so - is that relevant?

help

6

Yes, that is relevant insofar, as you seem to have an Optimus system (hybrid intel and nvidia graphics), and therefore cannot (and must not) use the standard nvidia driver.

So remove the nvidia repo and all nvidia packages, and X should start again.

If you want to use the nvidia card, you have to install Bumblebee:
https://en.opensuse.org/SDB:NVIDIA_Bumblebee

That’s not specific to Leap or Tumbleweed though, you would have problems on 13.2 as well.

If it is possible, you could disable the intel chip in the BIOS settings though. Then the nvidia driver should work.

7

I did all the stuff in SDB::NVIDIA - X still dead. I used the nvidia option as I do some CUDA work.

Are the CUDA toolkit 7.5 drivers OK? Currently just letting bumblebee do its own thing.

New message on start up: failed to start load kernel modules.

Xorg -config seg faults in modprobe in lib64/libc.so

I don’t know what bbswitch is or if I need it, but it was among things which messages said could not be loaded.

I have a Ge760 NVIDIA card and the motherboard seems to come with Intel HD graphics hardware.

SUSE 13.1 and 13.2 worked without any trouble.

Nothing done in x11 config file seems to make any difference.

Is the problem elsewhere? It all started with tumbleweed…

Thanks for replying

8

Bumblebee uses nouveau for the nvidia card if you didn’t install nvidia-bumblebee.

I have absolutely no experience with Bumblebee, but I don’t think the driver from the CUDA repo will work either.

I have a Ge760 NVIDIA card and the motherboard seems to come with Intel HD graphics hardware.

So, is this a Laptop, or a desktop machine with just an intel chip integrated in the CPU (or on the mainboard)?
In the latter case it’s not Optimus, and you should remove bumblebee. But even then it might help to disable the intel onboard chip if possible.

Please post your /var/log/Xorg.0.log for being able to tell more.

To get a useable system, you might uninstall nvidia and bumblebee first and maybe login to IceWM.
If necessary, you could also try to switch to xdm as display manager in /etc/sysconfig/displaymanager.

SUSE 13.1 and 13.2 worked without any trouble.

Well, I have no idea what you did install (or not) in 13.1 or 13.2.
An Optimus system does run without Bumblebee, but you will only be able to use the intel chip then.

9

It is a desktop. I have removed bumblebee archive and package and copied the vanilla .install to the xorg config file. Also removed entry Driver nvidia from the config.d/ thing.

Then did a zypper dup and mkinitrd. (I don’t know what I am doing).

The reboot still fails with its new message - failed to start load kernel modules.

Xorg log: -

40.072]

X.Org X Server 1.17.2
Release Date: 2015-06-16
40.072] X Protocol Version 11, Revision 0
40.072] Build Operating System: openSUSE SUSE LINUX
40.072] Current Operating System: Linux linux-rtmg 4.2.3-1-default #1 SMP PREEMPT Mon Oct 12 11:32:29 UTC 2015 (202286f) x86_64
40.072] Kernel command line: BOOT_IMAGE=/boot/vmlinuz-4.2.3-1-default root=UUID=7e2948f8-9055-4ec5-b4d3-6c1289057714 ro resume=/dev/disk/by-id/ata-Crucial_CT120M500SSD1_14150C145A92-part3 splash=silent quiet showopts nouveau.modset=0
40.072] Build Date: 25 October 2015 12:36:08PM
40.072]
40.072] Current version of pixman: 0.32.6
40.072] Before reporting problems, check http://wiki.x.org
to make sure that you have the latest version.
40.072] Markers: (–) probed, (**) from config file, (==) default setting,
(++) from command line, (!!) notice, (II) informational,
(WW) warning, (EE) error, (NI) not implemented, (??) unknown.
40.072] (==) Log file: “/var/log/Xorg.0.log”, Time: Sun Nov 15 18:00:16 2015
40.072] (==) Using config file: “/etc/X11/xorg.conf”
40.072] (==) Using config directory: “/etc/X11/xorg.conf.d”
40.072] (==) Using system config directory “/usr/share/X11/xorg.conf.d”
40.072] Parse error on line 10 of section InputClass in file /etc/X11/xorg.conf.d/50-device.conf
" m" is not a valid keyword in this section.
40.072] (EE) Problem parsing the config file
40.072] (EE) Error parsing the config file
40.072] (EE)
Fatal server error:
40.072] (EE) no screens found(EE)
40.072] (EE)
Please consult the The X.Org Foundation support
at http://wiki.x.org
for help.
40.072] (EE) Please also check the log file at “/var/log/Xorg.0.log” for additional information.
40.072] (EE)
40.072] (EE) Server terminated with error (1). Closing log file.

10

Doing random things will not fix stuff.

TO start remove the /etc/X11/xorg.conf file in general it is not needed unless you need to do some special configuration, which you don’t.

Removing an archive does not remove any packages from it. So unless you did something you are not telling us Bumblebee is still installed you must remove the Bumblebee package to uninstall it.

zypper dup is used to move to a new OS version it should only be used for that unless you fully understand the consequences

mkinitrd is used to set up the boot sequences for you hardware and drivers (more or less). Should not hurt or help to run it in this case

You probably still have part of bumblebee and maybe NVIDIA driver hanging around since it does not appear that you actually removed their packages.

It maybe just the bogus xorg.conf file you copied in . I have no idea where that came from or what’s in it

11

That’s ok, but some thing is wrong:

Xorg log: -

40.072]

X.Org X Server 1.17.2
Release Date: 2015-06-16
40.072] X Protocol Version 11, Revision 0
40.072] Build Operating System: openSUSE SUSE LINUX
40.072] Current Operating System: Linux linux-rtmg 4.2.3-1-default #1 SMP PREEMPT Mon Oct 12 11:32:29 UTC 2015 (202286f) x86_64
40.072] Kernel command line: BOOT_IMAGE=/boot/vmlinuz-4.2.3-1-default root=UUID=7e2948f8-9055-4ec5-b4d3-6c1289057714 ro resume=/dev/disk/by-id/ata-Crucial_CT120M500SSD1_14150C145A92-part3 splash=silent quiet showopts nouveau.modset=0
40.072] Build Date: 25 October 2015 12:36:08PM
40.072]
40.072] Current version of pixman: 0.32.6
40.072] Before reporting problems, check http://wiki.x.org
to make sure that you have the latest version.
40.072] Markers: (–) probed, (**) from config file, (==) default setting,
(++) from command line, (!!) notice, (II) informational,
(WW) warning, (EE) error, (NI) not implemented, (??) unknown.
40.072] (==) Log file: “/var/log/Xorg.0.log”, Time: Sun Nov 15 18:00:16 2015
40.072] (==) Using config file: “/etc/X11/xorg.conf”
40.072] (==) Using config directory: “/etc/X11/xorg.conf.d”
40.072] (==) Using system config directory “/usr/share/X11/xorg.conf.d”
40.072] Parse error on line 10 of section InputClass in file /etc/X11/xorg.conf.d/50-device.conf
" m" is not a valid keyword in this section.
40.072] (EE) Problem parsing the config file
40.072] (EE) Error parsing the config file
40.072] (EE)
Fatal server error:
40.072] (EE) no screens found(EE)
40.072] (EE)
Please consult the The X.Org Foundation support
at http://wiki.x.org
for help.
40.072] (EE) Please also check the log file at “/var/log/Xorg.0.log” for additional information.
40.072] (EE)
40.072] (EE) Server terminated with error (1). Closing log file.

So, fix /etc/X11/xorg.conf.d/50-device.conf.
If in doubt, post the file.

Or delete it completely, it should not matter anyway, normally all things are commented out and not used.

12

I did zypper rm bumblebee, after the zypper rr … for the repo, does that do it all?

The xorg.conf.install came with the Leap installation I thought - the xorg.config that was there before presumably came with previous installations - I have not needed to tweak these things for years - but the nvidia setup tool did adjust them.

I will remove the 2 config files as you suggest and see what happens - and try to find any nvidia remnants.

Thanks for replying.

I am using the bios to dual boot to windows to access the net - all a bit messy.

13

Deleting these config files did the trick :):).

Strange that the file in config.d/ had all lines commented out (leading ‘#’) so I don’t understand what was going on.

I had ceased to believe it would ever happen, and was about to tar up my home dir, save it and do a clean ubuntu install over the suse partitions.

Is it safe to add the NVIDIA Leap 42.1 repo now?

Anyway thanks.

14

Should be…

15

Yes, that should get rid of bumblebee.
Although you might need to remove the bbswitch and nvidia-bumblebee packages too if you installed them.

The xorg.conf.install came with the Leap installation I thought

The xorg.conf.install is actually created by the installer, and is only used for “Recovery Mode” (which doesn’t exist in Leap any more).
It only loads some generic driver.

  • the xorg.config that was there before presumably came with previous installations - I have not needed to tweak these things for years - but the nvidia setup tool did adjust them.

Actually you should not need any xorg.conf at all (and the nvidia packages do not create/modify it), the nvidia driver is loaded automatically by Xorg if it is installed (and it’s preferred over nouveau).
But don’t delete the other files in /etc/X11/xorg.conf.d/, they are required for certain things. E.g. 00-keyboard.conf sets the keyboard layout.

I will remove the 2 config files as you suggest and see what happens - and try to find any nvidia remnants.

Why 2?
You don’t need to remove xorg.conf.install, if that’s what you mean.
But then, you don’t exactly need it either.