LinuxQuestions.org - GLX problem

- Slackware (https://www.linuxquestions.org/questions/slackware-14/)

- - GLX problem (https://www.linuxquestions.org/questions/slackware-14/glx-problem-366468/)

Quote:

Originally posted by ilredil
you can just copy the old config file to ./.config but then the command would be `make oldconfig`

and then you'd get prompted for everything you're missing... i mean, on slackware the 2.6 config weighs 53K, while the 2.4 one weighs 41K, so i assume there's gonna be a lot of new stuff... really your best bet is to just start-off with a known-good 2.6 config file... then do a "make xconfig" instead of an "make oldconfig" and strip it down to what you need... IMHO using a 2.4 config to build a 2.6 kernel is a bad idea unless you're just trying to experiment with your box, which might be the case...

ftp://ftp.oregonstate.edu/pub/slackw.../config-2.6.13

ftp://ftp.oregonstate.edu/pub/slackw.../bare.i/config

I finally took the time to rebuild my kernel properly. I like keeping only the options that I need, and if I add things on later I can always just recompile and make modules_install.

I was able to get the NVIDIA 1.0-7676 driver to compile after uninstalling the driver, then using the -k $(uname -r) flag. But when I updated xorg.conf and tried starting X, I got a familiar error message:

Code:

FATAL: Error inserting nvidia (/lib/modules/2.6.13.4/kernel/drivers/video/nvidia.ko): No such device

(EE) NVIDIA(0): Failed to load the NVIDIA kernel module!

(EE) NVIDIA(0):  *** Aborting ***

(EE) Screen(s) found, but none have a usable configuration.

I tried modprobe to see what error that would give me. The results of sudo /sbin/modprobe nvidia were:

Code:

FATAL: Error inserting nvidia (/lib/modules/2.6.13.4/kernel/drivers/video/nvidia.ko): No such device

So I double checked to make sure my video card didn't fall out... it's still there. I'm using an AGP GeForce 4 TI 4400 128MB. Here's the relevant output of lspci:

Code:

01:00.0 VGA compatible controller: nVidia Corporation NV25 [GeForce4 Ti 4400] (rev a3)

So my card works, lspci knows it's there, the kernel is a fresh compile / install and there aren't any other problems with the kernel, is this a problem with the driver from nVidia? I've compiled nVidia drivers before and never have any trouble until now. Any suggestions? Time to try old builds of nvidia's driver? contact nvidia? give up on their closed source driver (at least until they release a newer one)?

Quote:

Originally posted by hanzerik
Try this as root:
mv /usr/lib/tls /usr/lib/tls-backup

Then start X

I wanted to confirm this worked for me with 7676 drivers.

ilredil, i'd check the /etc/X11/xorg.conf file cuz the error does sound like it could be caused by a setting from there...

Quote:

(EE) Screen(s) found, but none have a usable configuration.

BTW, does it work fine if you use the non-3d open source "nv" driver??

yep, the "nv" driver works fine. Just changing to the "nvidia" driver and I get that module uploadable error. I recompiled the kernel a few times, made sure that I had all the options I needed and everything seems to be working flawlessly, except nVidia's driver. :confused:

I wanted to get some games to work on Linux, but without the advanced 3D support that will not be feasible. And when I say games, I'm talking about id software, which runs on Linux without an emulation. I have more important things to set up so I'm not exactly losing any sleep over it, but now I want to do it just because it's being so difficult.

I'd like to take my kernel to another computer (adding the IDE controller chipset driver and other required components that differ) and see if I can get it working but my laptop has an Intel video card, the two servers I have don't have X installed, and the media center is still under construction.

Is there anything that's not obvious that I need in my kernel? For example, SCSI disk support is needed for a USB or firewire hard drive which threw me off at first.

[edit]/usr/lib/tls doesn't exist on my system[/edit]

I took someone's advice and decided what nvidia.com had to tell me. I found a very relivant link: http://nvidia.custhelp.com/cgi-bin/n...i=&p_topview=1

It didn't help, but it's good to see that I'm not the only one who's had this problem. Besides, that link might help other people who have this problem and happen to come across this thread.

Suprisingly, I do have /dev/nvidia*

Code:

ls /dev/nvidia*

/dev/nvidia0  /dev/nvidia1  /dev/nvidiactl

Also, specifying the actual .o file didn't help much either...

Code:

root@the-beast:/usr/src/linux-2.6.13.4/drivers/video/nvidia# insmod ./nvidia.o

insmod: error inserting './nvidia.o': -1 Invalid module format

root@the-beast:/usr/src/linux-2.6.13.4/drivers/video/nvidia# modprobe ./nvidia.o

FATAL: Module ./nvidia.o not found.

I suspect that this issue might explain why my system has sometimes been crashing on logging out from kde.

Anyway, I have done

mv /usr/lib/tls /usr/lib/tls-backup

And I haven't had any more crashes ......yet

I have a Slackware 10.2 installation and I am obviously experiencing the problem referred to in this thread after installing NVIDIA-Linux-x86-1.0-8174-pkg1.run.

I have read about the solution using the following switch:

--force-tls=classic

My question is: do I have a solution to this without completely reinstalling the drivers?. For example, can I apply some commandline switch form the NVIDIA ----------.run to allow me to only rebuild the Kernel?

Thank you.

Quote:

Originally Posted by Toods

renaming the /usr/lib/tls directory should do the trick... i mean, it's worked for several people using 7676... i'm not sure why you'd want to rebuild your kernel for this, though...

Quote:

Originally Posted by win32sux

renaming the /usr/lib/tls directory should do the trick... i mean, it's worked for several people using 7676... i'm not sure why you'd want to rebuild your kernel for this, though...

Thanks I've done this and it worked. Now I've just got to deal with the shutdown anomaly that I put in another thread