That’s to begin with. It would be okay, were it possible to keep display 2 to the
right of display 1 from session to session. Unfortunately, that is not possible
with NVIDIA and the NVIDIA driver won’t build any more. So, I am completely
screwed and there is no way out, except back to 11.3.
Here is what I have done:
figure out where multiple displays get configured. (They are hardware monitors
I am configuring, but multi-displays is a perssonal desktop setting, as everyone
knows)
Change “2 is clone of 1” to “2 is right of 1”.
Click “apply”. Nothing happens.
Figure out that I have to change, “primary output” must be changed from “none”
to “1”.
Click “apply”
Click “really apply” (“yes, it is okay”)
Click “really, really apply” (aka, “save as default”)
Click “really, really, really apply” (aka “yes, I want it set as default”)
Click “thank you SO much” (aka, “yes, I know it is now default now”)
log in again. We’re back to square 1.
This ought to be a bug report, but bugzilla is timing out.
That's to begin with. It would be okay, were it possible to keep display 2 to the
right of display 1 from session to session. Unfortunately, that is not possible
with NVIDIA and the NVIDIA driver won’t build any more. So, I am completely
screwed and there is no way out, except back to 11.3.
Here is what I have done:
figure out where multiple displays get configured. (They are hardware monitors
I am configuring, but multi-displays is a perssonal desktop setting, as everyone
knows)
Change “2 is clone of 1” to “2 is right of 1”.
Click “apply”. Nothing happens.
Figure out that I have to change, “primary output” must be changed from “none”
to “1”.
Click “apply”
Click “really apply” (“yes, it is okay”)
Click “really, really apply” (aka, “save as default”)
Click “really, really, really apply” (aka “yes, I want it set as default”)
Click “thank you SO much” (aka, “yes, I know it is now default now”)
log in again. We’re back to square 1.
This ought to be a bug report, but bugzilla is timing out.
So bkorb, tell us more about what you have done? Did you do a clean install of openSUSE 11.4 or an upgrade using 32 or 64 bits? What desktop are you using KDE or GNOME? What nVIDIA hardware are you using? How did you try to install the nVIDIA driver? It still loads just fine for me. Did you add in the nomodeset kernel load option? Did you think to set the NO_KMS_IN_INITRD=yes in your /etc/sysconfig? There are lots of reasons why you might have trouble. I suggest you look at this info to see if it might be helpful to you:
I did do a clean install. On first boot, I booted into init 3 and tried to run the
latest NVIDIA installation script (“The Hard Way”), but it would not compile.
I see there is now a new version. I got my current one only a few months
ago, let’s see if that works.
Still and all:
It turns out that there have been a lot of fixes to the current open source
driver and it is now capable of detecting a dual head configuration and actually
supports it, too. Really Cool!!! It just forgets the config from login to login,
meaning I either have to reconfigure my monitor every single time, or I have
to drop back to 11.3 (with the official NVIDIA driver) or I have to manually change
some config file so that the right thing happens on login. I’d as soon not be
tied to “The Hard Way”, if there is an effective alternative.
The latest NVIDIA package doesn’t recognize the GCC compiler and when I tell it
to go ahead anyway, it doesn’t recognize the kernel sources either. No way out.
Open YaST / Software / Software Management - Select the View Button on the top left and pick Patterns. Now, you will see several Patterns listed and you want to select:
Development
[X] Base Development
[X] Linux Kernel Development
[X] C/C++ Development
Then Press the Accept button on the bottom right and allow these applications to install. This should allow you to install the nVIDIA driver among other things.
$ sr make cloneconfig
HOSTCC scripts/basic/fixdep
/usr/lib64/gcc/x86_64-suse-linux/4.5/cc1: error while loading shared libraries: libmpfr.so.1: cannot open shared object file: No such file or directory
make[1]: *** [scripts/basic/fixdep] Error 1
make: *** [scripts_basic] Error 2
I guess I am at a loss here. I have never used the make config command to install the nVIDIA driver, but perhaps to compile a Linux Kernel.
Is it possible to outline just what actions you are trying to take here to help clarify the help that is required? Forgive me if I have missed some detail. It has been kind of crazy in the forums today with the recent release of openSUSE 11.4. I do know how to install the nVIDIA driver using a repository or the hard way and I know how to compile a Linux kernel if you wanted to do that, but I am not very good I guess at reading between the lines. Heck, I have trouble often reading the lines them selfs. As always, I do want to help.
More futzing around yielded the real problem: cc1 won’t work without LD_LIBRARY_PATH,
despite the fact I verified the correct directory list in /etc/ld.so.conf and then ran ldconfig.
Needless to say, I don’t go setting LD_LIBRARY_PATH willy-nilly or fiddle with ld.so.conf
unless I have to. It ought to “just work”, but it doesn’t. Anyway, that library issue yields
messages about the linux kernel being misconfigured, whereas the problem is with ld.so.
Good error messages are, in fact, important.
So, I finally slogged past that one yielding this in /var/log/nvidia*.log:
-> Kernel module compilation complete.
ERROR: Unable to load the kernel module 'nvidia.ko'. This happens most
frequently when this kernel module was built against the wrong or
improperly configured kernel sources, with a version of gcc that differs
from the one used to build the target kernel, or if a driver such as
rivafb, nvidiafb, or nouveau is present and prevents the NVIDIA kernel
module from obtaining ownership of the NVIDIA graphics device(s), or
NVIDIA GPU installed in this system is not supported by this NVIDIA
Linux graphics driver release.
Please see the log entries 'Kernel module load error' and 'Kernel
messages' at the end of the file '/var/log/nvidia-installer.log' for
more information.
Blech. I’ve tried the automated stuff before, but it has failed.
I went back and tried again and it seems to be working.
Won’t know until the next reboot and that will wipe this session.
Back in a few minutes…
“nomodeset” no clue set, either. What would that be? The “clonconfig” was a response to the nvidia configure script
claiming that it could not find the generated version header. “make cloneconfig” should clone the default configuration
from /boot and regenerate that header. However, the message was wrong. The correct message should have been
that “gcc” was failing to work correctly because it requires LD_LIBRARY_PATH=/usr/lib64 in the environment.
Also wrong, but it got me to the point where the driver was built, but would just not run.
Now I’ve added the nvidia repo to my software repository list and pulled the driver and rebooted.
Much, much, much, much, much worse off now. The new driver will not recognize the dual
head at all, so it is no longer possible to reconfigure it after each reboot. I cannot reconfigure
it at all, period.
Seriously it looks like you are going about this the hard way. I have never seen anyone approach this this way. Have you actually tried to do it the proscribed way???
I am certainly doing it the proscribed way. I’d like to be doing it the prescribed way, though. I started by doing a straight install and doing nothing at all special. I set the monitor with “configure desktop/display and monitor” stuff. That really ought to be the prescribed way to do it. But the settings didn’t stick. Next, I tried the tried and true build and install of the downloaded nvidia driver. It didn’t work claiming it was because the kernel sources were not configured correctly, but really because ld.so isn’t linking cc1 to a library that lives in /usr/lib64 – unless I use LD_LIBRARY_PATH. I finally get that problem straightened out and the resulting driver won’t load. So, I finally use the software repo+software management route and I now have a driver more useless than what I started with. Where did I jump the track? Perhaps I ought to have retried the automated route before the “hard way” route earlier, but past experience had shown me that it did not work, but the download, build and install by hand did. Why do I now have a driver that won’t see the dual head connection?
I’d not heard of that. I missed the reference in the “how to load the nVIDIA driver” stuff. Still, the online man page says:
8. TODO
There are many things still to be added to nvidia-settings, some of which include:
- configurability of TwinView (NVIDIA is planning to implement this)
- configurability of multiple X screens (NVIDIA is planning to implement this)
which sorta says to me, “Not ready for prime time.” It also looks like I need to re-install
the repo-drivers from the nvidia repository. (I uninstalled them so I could get both of
my monitors back.)
It (the nVIDIA configurator) is not ready for prime time. And the whole reason I’ve spent the last 6 hours futzing with this is because the oss drivers do not work correctly. It does the work just fine, but it somehow forgets what its configuration was, so you have to go through the process of reconfiguring your monitor with every login. User unfriendly enough to reasonably be called, “broken”. I’m on an x86-64 platform, so the 32 bit helps don’t really help. Thank you! I do appreciate your assistance in getting to the bottom of this! Regards, Bruce
