issues installing on fedora core 1

[rum beverage]

i can't seem to get the nvidia installer to generate the kernel interface for my kernel

the following error log is created

nvidia-installer log file '/var/log/nvidia-installer.log'
creation time: Wed Dec 10 22:50:31 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
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/2.4.22-1.2129.nptl/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/2.4.22-1.
2129.nptl/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.

*** Failed cc sanity check. Bailing out! ***
make: *** [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.

====================================

i think i have gcc 3.2 set as my default
any help would be appreciated

[rum beverage]

bump

[Papol]

I've only used their installer without patching it one time(I'm using 2.6.0-test11-bk?" kernel sources), but there seems to be a few things going on.

- The fc1 default kernel version isn't one that a precompiled module has been built against and stuff in the ./nv*.run package.
- Sometimes, the /lib/modules/{kernelver#}/build link isn't correct, but this *seems* ok in your case.
- You definitely have kernel src installed if that link works.

So you can try at least a couple of things:

$ export IGNORE_CC_MISMATCH=1
$ rerun your nvi*{ver}.run installer again.

or (and if the above steps fail)

build your kernel from source. The reason it tells you there's a gcc mismatch is because it checks the gcc version of the running kernel against the version of the gcc compiler that's installed. If you haven't built your own kernel, odds are against it being the same.

On vanilla installs, its always hard to figure out where they stuffed the actual kernel .config file for the running kernel. If it's not in /boot/config-2.4.x.x.x.x.x. (whatever it is), then it may be about as easy (or hard) to just take one of the config-{athlon}{i386}{and others} from /usr/src/linux-2.4/config* and copy it to .config and start from there until you get a running kernel from it. Then there's no conflict possible between compilers since you'll have compiled both kernel and nvidia.

Gotta do some work, so I can't explain further now and that's kind of cryptic stuff I suppose, but just ask questions about it for more.

g/l

ps:: If you do want to know which verson of gcc (probably 3.3.2-1) you've got, you can always just do a "gcc --version" cmd. Knowing that won't do you a lot of good if there's no precompiled module, however.

Originally posted by rum beverage
i can't seem to get the nvidia installer to generate the kernel interface for my kernel

nvidia-installer log file '/var/log/nvidia-installer.log'
creation time: Wed Dec 10 22:50:31 2003


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/2.4.22-1.2129.nptl/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/2.4.22-1.
2129.nptl/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.

*** Failed cc sanity check. Bailing out! ***
make: *** [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.

====================================

i think i have gcc 3.2 set as my default
any help would be appreciated

[rum beverage]

ok using the IGNORE_CC_MISMATCH helped get the kernel interface compiled but now i'm getting an error stating that the nvidia.o module could not be loaded
log file:

it's stating a kernel mismatch but i pointed the kernel-include-path to /usr/src/linux-2.4.22-1.2129.nptl/include and still got this error

if anybody can give me some advice it would be appreciated
==========================================
nvidia-installer log file '/var/log/nvidia-installer.log'
creation time: Thu Dec 11 19:22:51 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
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 : /usr/src/linux-2.4.22-1.2129.nptl/include
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.
-> Using the kernel include path '/usr/src/linux-2.4.22-1.2129.nptl/include' as
specified by the '--kernel-include-dir' commandline option.
-> Kernel include path: '/usr/src/linux-2.4.22-1.2129.nptl/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=/usr/src/linux-2.4.22-
1.2129.nptl/include'...
echo \#define NV_COMPILER \"`gcc32 -v 2>&1 | tail -n 1`\" > nv_compiler.h
gcc32 -c -Wall -Wimplicit -Wreturn-type -Wswitch -Wformat -Wchar-subscripts
-Wparentheses -Wpointer-arith -Wcast-qual -Wno-multichar -O -MD -D__KERNEL_
_ -DMODULE -D_LOOSE_KERNEL_NAMES -DNTRM -D_GNU_SOURCE -D_LOOSE_KERNEL_NAMES
-D__KERNEL__ -DMODULE -DNV_MAJOR_VERSION=1 -DNV_MINOR_VERSION=0 -DNV_PATCHL
EVEL=4620 -DNV_UNIX -DNV_LINUX -DNV_INT64_OK -DNVCPU_X86 -DREMAP
_PAGE_RANGE_4 -I. -I/usr/src/linux-2.4.22-1.2129.nptl/include -Wno-cast-qua
l nv.c
In file included from /usr/src/linux-2.4.22-1.2129.nptl/include/linux/vmallo
c.h:8,
from nv-linux.h:72,
from nv.c:14:
/usr/src/linux-2.4.22-1.2129.nptl/include/linux/highmem.h: In function `bh_k
map':
/usr/src/linux-2.4.22-1.2129.nptl/include/linux/highmem.h:20: warning: point
er of type `void *' used in arithmetic
gcc32 -c -Wall -Wimplicit -Wreturn-type -Wswitch -Wformat -Wchar-subscripts
-Wparentheses -Wpointer-arith -Wcast-qual -Wno-multichar -O -MD -D__KERNEL_
_ -DMODULE -D_LOOSE_KERNEL_NAMES -DNTRM -D_GNU_SOURCE -D_LOOSE_KERNEL_NAMES
-D__KERNEL__ -DMODULE -DNV_MAJOR_VERSION=1 -DNV_MINOR_VERSION=0 -DNV_PATCHL
EVEL=4620 -DNV_UNIX -DNV_LINUX -DNV_INT64_OK -DNVCPU_X86 -DREMAP
_PAGE_RANGE_4 -I. -I/usr/src/linux-2.4.22-1.2129.nptl/include -Wno-cast-qua
l os-agp.c
In file included from /usr/src/linux-2.4.22-1.2129.nptl/include/linux/vmallo
c.h:8,
from nv-linux.h:72,
from os-agp.c:24:
/usr/src/linux-2.4.22-1.2129.nptl/include/linux/highmem.h: In function `bh_k
map':
/usr/src/linux-2.4.22-1.2129.nptl/include/linux/highmem.h:20: warning: point
er of type `void *' used in arithmetic
gcc32 -c -Wall -Wimplicit -Wreturn-type -Wswitch -Wformat -Wchar-subscripts
-Wparentheses -Wpointer-arith -Wcast-qual -Wno-multichar -O -MD -D__KERNEL_
_ -DMODULE -D_LOOSE_KERNEL_NAMES -DNTRM -D_GNU_SOURCE -D_LOOSE_KERNEL_NAMES
-D__KERNEL__ -DMODULE -DNV_MAJOR_VERSION=1 -DNV_MINOR_VERSION=0 -DNV_PATCHL
EVEL=4620 -DNV_UNIX -DNV_LINUX -DN
V_INT64_OK -DNVCPU_X86 -DREMAP_PAGE_RANGE_4 -I. -I/usr/src/linux-2.4
.22-1.2129.nptl/include -Wno-cast-qual os-interface.c
In file included from /usr/src/linux-2.4.22-1.2129.nptl/include/linux/vmallo
c.h:8,
from nv-linux.h:72,
from os-interface.c:26:
/usr/src/linux-2.4.22-1.2129.nptl/include/linux/highmem.h: In function `bh_k
map':
/usr/src/linux-2.4.22-1.2129.nptl/include/linux/highmem.h:20: warning: point
er of type `void *' used in arithmetic
gcc32 -c -Wall -Wimplicit -Wreturn-type -Wswitch -Wformat -Wchar-subscripts
-Wparentheses -Wpointer-arith -Wcast-qual -Wno-multichar -O -MD -D__KERNEL_
_ -DMODULE -D_LOOSE_KERNEL_NAMES -DNTRM -D_GNU_SOURCE -D_LOOSE_KERNEL_NAMES
-D__KERNEL__ -DMODULE -DNV_MAJOR_VERSION=1 -DNV_MINOR_VERSION=0 -DNV_PATCHL
EVEL=4620 -DNV_UNIX -DNV_LINUX -DNV_INT64_OK -DNVCPU_X86 -DREMAP
_PAGE_RANGE_4 -I. -I/usr/src/linux-2.4.22-1.2129.nptl/include -Wno-cast-qua
l os-registry.c
In file included from /usr/src/linux-2.4.22-1.2129.nptl/include/linux/vmallo
c.h:8,
from nv-linux.h:72,
from os-registry.c:14:
/usr/src/linux-2.4.22-1.2129.nptl/include/linux/highmem.h: In function `bh_k
map':
/usr/src/linux-2.4.22-1.2129.nptl/include/linux/highmem.h:20: warning: point
er of type `void *' used in arithmetic
ld -r -o nv-linux.o nv.o os-agp.o os-interface.o os-registry.o
ld -r -o nvidia.o nv-linux.o nv-kernel.o
-> done.
-> Kernel module compilation complete.
ERROR: Unable to load the kernel module 'nvidia.o'. This is most likely
because the kernel module was built using the wrong kernel header files.
Please make sure you have installed the kernel header files for your
kernel; on Red Hat Linux systems, for example, be sure you have the
'kernel-source' rpm installed. If you know the correct kernel header
files are installed, you may specify the kernel include path with the
'--kernel-include-path' commandline option.
-> Kernel module load error: ./usr/src/nv/nvidia.o: kernel-module version
mismatch
./usr/src/nv/nvidia.o was compiled for kernel version 2.4.22
while this kernel is version 2.4.22-1.2129.nptl.
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.

[drewm]

in the shell before you compile, do an export CC=/usr/bin/gcc32 (or setenv CC /usr/bin/gcc32 for non-bash) and it will compile and load just fine. Hmmm, or was it CC=gcc32. It's one of those two.

[Papol]

/usr/src/linux-2.4.22-1.2129.nptl/include/linux/highmem.h:20: warning: point
er of type `void *' used in arithmetic
ld -r -o nv-linux.o nv.o os-agp.o os-interface.o os-registry.o
ld -r -o nvidia.o nv-linux.o nv-kernel.o
-> done.
-> Kernel module compilation complete.
ERROR: Unable to load the kernel module 'nvidia.o'. This is most likely
because the kernel module was built using the wrong kernel header files.
Please make sure you have installed the kernel header files for your
kernel; on Red Hat Linux systems, for example, be sure you have the
'kernel-source' rpm installed. If you know the correct kernel header
files are installed, you may specify the kernel include path with the
'--kernel-include-path' commandline option.
-> Kernel module load error: ./usr/src/nv/nvidia.o: kernel-module version
mismatch
./usr/src/nv/nvidia.o was compiled for kernel version 2.4.22
while this kernel is version 2.4.22-1.2129.nptl.

The plot thickens.. First thing is: Have you tried what drewm said to do? I don't know why that'd help in this case, but I'd hate to throw out any possible answer.

From reading your error messages, it really seems as if you did everything right. The only way I can see for the nvidia-installer to think that your module was compiled for 2.4.22, rather than 2.4.22-1.2129.nptl, is if after all of that, it somehow yanked a regular 2.4.22 module out of its precompiled stuff, or if it just tried to load the *wrong* module.

Could you do this command and show its output?

$ find /lib/modules/ -iname 'nv*' -print -exec ls -al {} \;

it should show at least one nvidia.o module, depending on how many /lib/module directories there are.

Then before giving up on what you've got there already, just try manually:

$ modprobe nvidia

This will try to load from whatever is in /lib/modules/`uname -r`, rather than from /usr/src/nv which is the tmp workspace the nvidia installer uses,

g/l

[rum beverage]

one thing i should say is since the last time i had that error i tried installing via one of the rpms out there and X wouldn't load so it went back to default config... so if it seems like some modules are where they should be it might just be from that rpm install

here is the output from that command:

=================================
/lib/modules/2.4.22-1.2115.nptl/kernel/drivers/char/nvram.o
-rwxr--r-- 1 root root 9056 Oct 29 16:39 /lib/modules/2.4.22-1.2115.nptl/kernel/drivers/char/nvram.o
/lib/modules/2.4.22-1.2129.nptl/kernel/drivers/char/nvram.o
-rwxr--r-- 1 root root 9056 Dec 1 09:58 /lib/modules/2.4.22-1.2129.nptl/kernel/drivers/char/nvram.o
/lib/modules/2.4.22-1.2129.nptl/kernel/drivers/video/nvidia
total 1868
drwxr-xr-x 2 root root 4096 Dec 11 19:50 .
drwxr-xr-x 7 root root 4096 Dec 11 19:50 ..
-rw-r--r-- 1 root root 1899812 Dec 3 05:58 nvidia.o
/lib/modules/2.4.22-1.2129.nptl/kernel/drivers/video/nvidia/nvidia.o
-rw-r--r-- 1 root root 1899812 Dec 3 05:58 /lib/modules/2.4.22-1.2129.nptl/kernel/drivers/video/nvidia/nvidia.o
======================================

[Papol]

Originally posted by rum beverage
one thing i should say is since the last time i had that error i tried installing via one of the rpms out there and X wouldn't load so it went back to default config... so if it seems like some modules are where they should be it might just be from that rpm install

here is the output from that command:

=================================
/lib/modules/2.4.22-1.2115.nptl/kernel/drivers/char/nvram.o
-rwxr--r-- 1 root root 9056 Oct 29 16:39 /lib/modules/2.4.22-1.2115.nptl/kernel/drivers/char/nvram.o
/lib/modules/2.4.22-1.2129.nptl/kernel/drivers/char/nvram.o
-rwxr--r-- 1 root root 9056 Dec 1 09:58 /lib/modules/2.4.22-1.2129.nptl/kernel/drivers/char/nvram.o
/lib/modules/2.4.22-1.2129.nptl/kernel/drivers/video/nvidia
total 1868
drwxr-xr-x 2 root root 4096 Dec 11 19:50 .
drwxr-xr-x 7 root root 4096 Dec 11 19:50 ..
-rw-r--r-- 1 root root 1899812 Dec 3 05:58 nvidia.o
/lib/modules/2.4.22-1.2129.nptl/kernel/drivers/video/nvidia/nvidia.o
-rw-r--r-- 1 root root 1899812 Dec 3 05:58 /lib/modules/2.4.22-1.2129.nptl/kernel/drivers/video/nvidia/nvidia.o
======================================

Cool, thanks. So it looks like you've had kernel 2.4.22-1.2115.nptl once, and now are upgraded to 2.4.22-2129.nptl, and the nvidia.o module in *2129.nptl is from 6am on 3-Dec. If you just did a build, that makes no sense at all. The module you just made should be dated today (or whevever you made it). Can you also do an "ls -alR /usr/src/nv"? If the nvidia.o in that path is dated 10 or 11Dec, then it may be the most correct of the two. The installer is *supposed* to replace /lib/modules/2.4.22-1.2129.nptl/kernel/drivers/video/nvidia/nvidia.o with one made and dated by it. I wonder if that's the one that came from your .rpm install. It would explain a modprobe nvidia failing with a wrong-kernel error.

So how to fix this mess.. Maybe it'd be best to clean out everything other than the installer*.run file and start this over. Get rid of /usr/src/nv. (rm -r -f /usr/src/nv) so that there's no way in hell it can use something in there that was maybe from the .rpm install and doesn't cleanup correctly on the next make. The other thing would be to rm /lib/modules/2.4.22-1.2129.nptl/kernel/drivers/video/nvidia/nvidia.o. It's not doing you any good anyway.

Then run the *.run nvidia file again and specific the correct path (like you did, I think) and run that find command again to make sure it gets where it belongs, in the /lib/modules/*2129.nptl path.

Something tells me this ain't gonna be easy (duh) but as long as you have

a) kernel source that you made your running kernel from
b) /lib/modules/*2129.nptl/build -> as a symlink to /usr/src/linux-*2129.nptl,
c) no bull in the /usr/src/nv dir,

it should work.


***edit***

Check Lordmorgul's reply out and the 2 links to this spot on his sig in this thread:

http://www.nvnews.net/vbulletin/show...463#post245463

He may actually have the right rpm for you, which would simplify things if all of these things I'm having you try fail.

[rum beverage]

that's where i got the rpm in the first place

[rum beverage]

Ok I finally fixed my issue. What was happening is somewhere in my source package for kernel-2.4.22-1.2129.nptl it was identifying the kernel as 2.4.22 instead of 2.4.22-1.2129 so that's why I was getting the kernel mismatch error. What I did was uninstalled the source rpm and then re-installed it. It may have happened because after I installed the source for 2129 I installed the source from 2115 and then the updater was picking up an upgrade for source to 2129 and it must have over written somethin incorrectly. Anyway it works now and I can finally play tux racer. Boo yah! Now to add NTFS support.

PS thanks for all the help Papol

[Papol]

Cool. Glad to hear you got it working. If you notice any log messages that have to do with 'Pci_badness in module ***' followed by a bunch of __nvidia symbols (a stack trace), while using the new driver, I might have the beginnings of a solution for that.

Its a small kernel patch I wrote against the 2.6.0-test11-bk* tree but which might be backported to 2.4.x (I'm not sure if it'll slip into 2.4.x, but I'll take a look if someone wants to try it. In my case it got rid of the fairly harmless but time-wasting log messages, and exposed an Alsa problem with my sound that I didn't see because of all the damn nvidia messages lol..