hi, I'm installing the nvidia drivers and have already killed Xwindow to do the maintenance. However, when I rpm -Uvh with the nvidia kernel for RH 8 i386 i get the following message:

modprobe: Can't locate module nvidia

so then i tried rpm -ivh and got a message:

package NVIDIA_kernel-1.0-4191.rh80up_2.4.18_18.8.0 is already installed

so then i tried rpm -e and it gave (!):

error: package NVIDIA_kernel-1.0-4191.rh80up_2.4.18_18.8.0.i386.rpm is not installed

what is going on?!
I have tried (and failed) installing the drivers once before the RH 8 specific nvidia kernel rpm came out, so maybe thats interfering, I dont know?! Why can I never get this nvidia thing to work

you uninstall by PACKAGE name and not FILE name

rpm -e NVIDIA_kernel-1.0-4191

When you tried to uninstall it, did type:

rpm -e NVIDIA_kernel-1.0-4191.rh80up_2.4.18_18.8.0.i386.rpm

?

When you uninstall RPMs you dont use the entire version number i.e.
to uninstall the NVIDIA_kernel it would be this:

rpm -e NVIDIA_kernel

that worked to uninstall the rpm, but now when i -ivh i just get the same message i got when i -uvh'ed:

modprobe: Can't locate module nvidia

looks like a xwindow message to me, but i killed the X process.

maybe its my version of Xfree86. how should i check its version number now the X process is not running?

X -showconfig

ok its working but it says that the kernel is already installed, but when i initialise X it says that it 'failed to initialise the kernel module', so i presume it is not correctly installed. what rpm command will force it to overwrite the existing installation? (I know from the md5 that the rpm is not corrupted).

ok i think i have an idea whats going on, its saying the kernel is already installed cos its failed once, if i rpm -e the kernel, and then install it again with either -Uvh or -ivh it gives the message:

Preparing... ########################################### [100%]
1:NVIDIA_kernel ########################################### [100%]
modprobe: Can't locate module nvidia

Failed to install nvidia.o

error: %post(NVIDIA_kernel-1.0-4191.rh80up_2.4.18_18.8.0) scriptlet failed, exit status 1



and from then on it claims that the kernel is installed.
Is this a common error, what can i do to make it work, or to locate module nvidia?
Thanks in advance for any help, and for the help received so far

Did you run the chooser.sh script before you downloaded the two updates? Maybe you dont have the right packages.

yeah ive run the script and it says I should use the i686 version of the kernel, which I have done, after I typed rpm -e NVIDIA_kernel to remove the i386 ver ive been messing with. However, it still gives the same message:

rpm -ivh NVIDIA_kernel-1.0-4191.rh80up_2.4.18_18.8.0.i686.rpm
Preparing... ########################################### [100%]
1:NVIDIA_kernel ########################################### [100%]
modprobe: Can't locate module nvidia

Failed to install nvidia.o

error: %post(NVIDIA_kernel-1.0-4191.rh80up_2.4.18_18.8.0) scriptlet failed, exit status 1



Does anyone even know what this module nvidia its describing belongs to?

i think ive found a lead in an old thread entitiled:

Nvidia drivers & Red Hat 8 (2.4.18-19.8.0)


in the last post it is said that cant load module nvidia occurs when the linux src code is not present on the system. I have checked my /usr/src directory and the linux dir is labelled linux-2.4 and theres another dir callled linux-2.4.18-14. Do i need to perhaps create a /usr/src/linux directory and create a symbolic link to the linux-2.4 dir, as is described in the post, and if so how exactly do i tell bash to do that, sorry but I've never created links before .

ln -s /usr/src/linux-2.4.18-14 /usr/src/linux

Hi, thanks for that now if i check the contents of /usr/src/linux i get the following message:



[root@chameleon linux]# dir
abi CREDITS fs kernel mm Rules.make
arch crypto include lib net scripts
configs Documentation init MAINTAINERS README
COPYING drivers ipc Makefile REPORTING-BUGS



I thought from the post i found that this would cure the problem of the rpm not being able to 'locate module nvidia' but it is still giving the same message, despite having used NVChooser to ensure I should use the RH80up.i686 rpm, and following the readme. Any ideas?

Did it tell you to get the NVIDIA_kernel-1.0-4191.src.rpm? If so, you can't just install it, you have to compile the source first:


rpmbuild --rebuild NVIDIA_kernel-1.0-4191.src.rpm.

Then note where it made the rpm file and install that.

no its def not the src rpm, just the normal one: NVIDIA_kernel-1.0-4191.rh80up_2.4.18_18.8.0.i686.rpm