View Full Version : Nvidia and Debian
Hi, I have trouble to run driver installer (.run) on my Knoppix (Debian) release.
I had no problem with Mandrake 9.1, in Knoppix the Kernel does not support the Nvidia card. What NVIDIA_kernel I can install that is compatible with my RivaTnt2 64 ?
Can I still use the NVIDIA-Linux-x86-1.0-4349.run after the kernel installer?
Thanks
Originally posted by syd
Can I still use the NVIDIA-Linux-x86-1.0-4349.run after the kernel installer? Huh? The nVidia installer (the .run file) will install the kernel module for you...
What are the problems you're having with it? Oh, and BTW -- try 4363, too. 4349 had a few bugs in it; most of them have been fixed in 4363 (AFAIK anyway).
Post /var/log/nvidia-installer.log, too.
The installer cannot proceed with the installation of the driver because the kernel not support the card nvidia:( . ..and give me an error to find the kernel to install in automatic (on the site nvidia). Now I am using Mandrake, as soon as possible I will put the log on this forum. L' X server works well however, KDE is visualized correctly but I have not all the quality of my card. Thanks :p
Oh -- you're getting the "could not find sources for your kernel" generic-type message? It's not that the kernel doesn't support your card, it's that there is no precompiled kernel interface for Knoppix's kernel. No big deal, you just have to build it yourself.
To do that, you need to install Knoppix's kernel-source package. Since Knoppix is based on Debian, it might be named kernel-headers instead. Make sure you install the right version (the version of kernel-headers has to match what uname -r prints).
Thank you very much :p ;)
...therefore I use only the Installer .run and not also NVIDIA_GLX and NVIDIA_kernel?
Sorry, but I'm newbie in linux :angel:
Yes, you only need the .run file. In the future, this is all that will (supposedly) be provided.
Ok, this is my installer.log nvidia-installer log file '/var/log/nvidia-installer.log' creation time: Sat May 3 16:24:19 2003 option status: license pre-accepted : false update : false force update : false expert : false uninstall : false driver info : false no precompiled interface: false no ncurses color : false query latest driver ver : false no OpenGL header files : false no questions : false silent : false XFree86 install prefix : /usr/X11R6 OpenGL install prefix : /usr Installer install prefix: /usr kernel include path : (not specified) kernel install path : (not specified) proc mount point : /proc ui : (not specified) tmpdir : /tmp ftp site : ftp://download.nvidia.com Using: nvidia-installer ncurses user interface -> License accepted. -> No precompiled kernel interface was found to match your kernel; would you li ke the installer to attempt to download a kernel interface for your kernel f rom the NVIDIA ftp site (ftp://download.nvidia.com)? (Answer: Yes) -> No matching precompiled kernel interface was found on the NVIDIA ftp site; this means that the installer will need to compile a kernel interface for your kernel. -> Kernel include path: '/lib/modules/2.4.20-xfs/build/include' -> Cleaning kernel module build directory. executing: 'cd ./usr/src/nv; make clean'... rm -f nv.o os-agp.o os-interface.o os-registry.o nv-linux.o nv_compiler.h * .d NVdriver nvidia.o -> Building kernel module: executing: 'cd ./usr/src/nv; make nvidia.o SYSINCLUDE=/lib/modules/2.4.20-xf s/build/include'... You appear to be compiling the NVIDIA kernel module with a compiler different from the one that was used to compile the running kernel. This may be perfectly fine, but there are cases where this can lead to unexpected behaviour and system crashes. If you know what you are doing and want to override this check, you can do so by setting IGNORE_CC_MISMATCH. In any other case, set the CC environment variable to the name of the compiler that was used to compile the kernel. [1;31m*** Failed cc sanity check. Bailing out! *** [0mmake: *** [gcc-check] Error 1 -> Error. ERROR: Unable to build the NVIDIA kernel module. ERROR: Installation has failed. Please see the file '/var/log/nvidia-installer.log' for details. You may find suggestions on fixing installation problems in the README available on the Linux driver download page at www.nvidia.com.
My uname -r is:
2.4.20-xfs
I get :
apt-get install kernel-headers-2.4.20-xfs kernel-source-2.4.20
but this is the massage:
couldn't find package kernel-headers-2.4.20-xfs
:eek: :confused:
Thanks :angel:
Doesn't that log file have returns in it?!?! :eek:
Anyway, it looks like the problem is a compiler version mismatch. What do cat /proc/version and cc -dumpversion tell you?
You do seem to have some kernel source package installed, though.
For any Debian based distribution what you need to do is:
apt-get install kernel-headers-'uname -r'
sh NVIDIA-Linux-x86-1.0-4363.run --kernel-include-path=/usr/src/kernel-headers-2.4.18-bf2.4/include/
But I think that Knoppix kernel headers may not be available...
The thing is that Debian store kernel headers at /usr/src/kernel-headers-VERSION
Hi, this is cat /proc/version :
Linux version 2.4.20-xfs (root@Knoppix) (gcc version 2.95.4 20011002 (Debian prerelease)) #1 SMP Mit Mär 26 15:37:36 CET 2003
and this cc -dumpversion
3.2.3
I try apt-get install 2.4.20 without -xfs and it runs !! But it is correct? I do not use the Filesystem Xfs ...
However later having downloaded and installed the Headers for the Kernel 4.2.20, Nvidia.run gives me the same mistakes.
:mad: :(
Read the README. Using gcc 2 on your kernel (in your case, 2.95.3) and gcc 3 on your nVidia driver WILL fail (even if you force it). That is why it complains and quits -- because if it didn't, there'd be ten thousand people trying to do exactly what you're doing, and they'd all have problems with ioremap() failing (along with countless other things) inside the driver.
You need to recompile your kernel using your system compiler.
I have resolved... withexport CC=/usr/bin/gcc-2.95. .
Then I have
installed NVIDIA.run and has compiled the kernel alone. I do not know
the meant one of export... you can say it to me? Thanks for the
patience :angel:
"export" tells the shell to set that variable for the current shell, and all subshells. If you don't use export (if you just say CC=/usr/bin/gcc-2.95), then the variable only gets set for the current shell.
You need it set in child shells too, because the run file either spawns a shell itself, or you spawn one, when you do sh filename. The "sh" is a new shell.
Thanks for everything ... you have been very kind ;)
Instead of telling where the CC is, you could simply do the following (as root)
$ env IGNORE_CC_MISMATCH=1 ./nvidia-installer
(that is when you unpacked the package, otherwise, use sh NVIDIA-... instead of ./nvidia-...
It worked for me (Knoppix 3.3)
Unless you're very sure of what you're doing, you should always compile kernel modules with the same version of compiler as you did the rest of the kernel.
The kernel is very sensitive to which version of gcc is used to compile it. For 2.4.x the only safe version is gcc 2.95.x (x>= 3).
For 2.6.x, gcc 3.3.1 has given me no problems, but few people are running that kernel yet :)
vBulletin® v3.7.1, Copyright ©2000-2012, Jelsoft Enterprises Ltd.