Nvidia Linux Driver Problem Reedition For Andy

[tsubasa]

Hello Andy,
This is the famous Log files :)

[tsubasa]

This is the second famous Log files :

[tsubasa]

My Xfree Config

[bwkaz]

From your X log:

(EE) NVIDIA(0): Failed to initialize the NVIDIA kernel module!
(EE) NVIDIA(0): *** Aborting *** And from the README:

Q: My X server fails to start, and my XFree86 log file contains the error:

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

A: Nothing will work if the NVIDIA kernel module doesn't function
properly. If you see anything in the X log file like "(EE)
NVIDIA(0): Failed to initialize the NVIDIA kernel module!" then
there is most likely a problem with the NVIDIA kernel module.
First, you should verify that if you installed from rpm that
the rpm was built specifically for the kernel you are using.
You should also check that the module is loaded ('/sbin/lsmod');
if it is not loaded try loading it explicitly with 'insmod' or
'modprobe' (be sure to exit the X server before installing a new
kernel module). If you receive errors about unresolved symbols,
then the kernel module has most likely been built using header files
for a different kernel revision than what you are running. You can
explicitly control what kernel header files are used when building
the NVIDIA kernel module with the --kernel-include-dir option (see
`sh NVIDIA-Linux-x86-1.0-4363.run --advanced-options`
for details).

Please note that the convention for the location of kernel header
files changed approximately at the time of the 2.4.0 kernel release,
as did the location of kernel modules. If the kernel module fails to
load properly, modprobe/insmod may be trying to load an older kernel
module (assuming you've upgraded). cd'ing into the directory with
the new kernel module and doing 'insmod ./nvidia.o' may help.

Another cause may be that the /dev/nvidia* device files may be missing.

Finally, the NVIDIA kernel module may print error messages indicating
a problem -- to view these messages please check /var/log/messages, or
wherever syslog is directed to place kernel messages. These messages
are prepended with "NVRM".

[tsubasa]

if it is not loaded try loading it explicitly with 'insmod' or
'modprobe' (be sure to exit the X server before installing a new
kernel module).

I write insmod ? or insmod nvidia ?
i write modprobe ? or modprobe nvidia ? At Every Start ???

I am not a professionnal of linux :)
be sure to exit the X server before installing a new
kernel module). <- It Say I Do Reinstall Something ???

I dont about the kernel but i use a mandrake 9.1 u will know better me what kernel is use i think its 2.4 but i dont understand the lines :

If the kernel module fails to
load properly, modprobe/insmod may be trying to load an older kernel
module (assuming you've upgraded). cd'ing into the directory with
the new kernel module and doing 'insmod ./nvidia.o' may help.

I a little bit lost :'(

[bwkaz]

Originally posted by tsubasa
if it is not loaded try loading it explicitly with 'insmod' or
'modprobe' (be sure to exit the X server before installing a new
kernel module).

I write insmod ? or insmod nvidia ?
i write modprobe ? or modprobe nvidia ? At Every Start ??? For the moment, yes, on every reboot, do an /sbin/modprobe nvidia when su'ed to root, then exit the root shell and startx as your user. If X works that way, then we can try to get it to load automatically, either on boot or the first time it's needed. One step at a time. ;)

be sure to exit the X server before installing a new kernel module). <- It Say I Do Reinstall Something ??? Well, I think I'd reword it if I could. It means "don't run the modprobe when X is running", basically. However, given the problem that the modprobe is fixing, that shouldn't even be a problem -- X is failing to start. ;)

When it says "install", it should really say "load". Linux lets you overwrite (AKA delete and recreate) any kind of file that's in use; what happens is, both copies of the file stay on disk until all processes close the file handle that they had on the old version. This is why it never takes a full reboot to upgrade anything but the kernel -- the files are just overwritten, and reloaded from disk next time they're needed. Much better. :)

But yeah, you have to not be running X to load the kernel module. You also should be outside of X when upgrading the driver, just so that the installer knows which module is loaded.

I dont about the kernel but i use a mandrake 9.1 u will know better me what kernel is use i think its 2.4 but i dont understand the lines :

If the kernel module fails to
load properly, modprobe/insmod may be trying to load an older kernel
module (assuming you've upgraded). cd'ing into the directory with
the new kernel module and doing 'insmod ./nvidia.o' may help. This very likely isn't even an issue anymore. I'd just not worry about it. :)

[tsubasa]

For information, i never try to install the nvidia driver with X start !
I always start Linux in Failsafe for install !

[tsubasa]

I have format my linux
-I reinstall a clean mandrake 9.1
-I boot in failsafe
-I install nvidia drivers 4363
-I VI the XFconf86-4 fot make the change
-I starts X, always the same errors
- I try your solutions, i write in the console /sbin/modprobe nvidia
"That says Nvidia kernel loading sucessfully with warning"
-I starts X, Black screen and stop my pc
-I reset my pc, and restart linux in konsole.
-I try another one /sbin/insmode nvidia, i have the same message "nvidia kernel loading sucessfully with warning".
-I startx, black screen and stop my pc
I look in the XFconf.log i have always the same error :'(.
Look at that :

[tsubasa]

This is another log system if it can help you

Thanks for your help ! I am very disapointed

Best regards

[bwkaz]

Is this error the one you get after modprobe'ing nvidia? Or before?

I have a feeling the issue that's causing the need to modprobe nvidia manually is that you booted to failsafe when you installed the drivers. In Mandrake, failsafe doesn't mount devfs (the device filesystem -- devfs is a completely in-memory filesystem; failsafe mode uses the on-disk device files instead), so the installer doesn't put the proper alias in /etc/modules.conf (the alias depends on whether your distro uses devfs or not; if it does, the alias should read alias /dev/nvidia* nvidia, and if not, it should read alias char-major-195 nvidia instead). So the kernel module doesn't load when you're back in real (non failsafe) mode.

Think of it this way -- you wouldn't install a video driver while booting Windows to safe mode, would you? I know I wouldn't... though I don't know exactly what all the differences are between Win normal and Win safe, I do know one very important one in Mandrake.

But if you're still getting that error after you modprobe nvidia, then you need to look in your kernel logs after an X attempt (use dmesg to see the kernel logs). It should have a reason in there that it failed.

[tsubasa]

I probe nvidia after installing driver, i reset my computer i choose
Linux at liloo and my screen start flashing when mandrake is loading xinetd services after i have the error (ee) screen(s) found,but none have usable configuration.
At root i do a /sbin/modprobe nvidia, and i have this message "nvidia kernel loaded with warning". And i do a startx but the computer crashes too black screen.

but how can i do for start my mandrake for install driver ? I dont know how i can do that, the only thing i know to boot my linux without startx is the failsafe. Say me the another method please..

Thanx a lot a lot

[tsubasa]

I have format and reinstall mandrake 9.1.

In KDE i open a console and in SU i write init 3.
I have try to install nvidia driver, and after i also modify the XFconf86-4.
At reboot, my pc crashes after load xinetd services and i have a black screen.
i have tried to modprobe nvidia and startx but it failed too, black screen.
I have copy the xfconf.log but its the same than the other one,

Kernel load failed
(EE)Screen(s) found, but non have a usabe confi....

:'(

[bwkaz]

OK, the problem is a blank screen then.

After you've modprobe'ed nvidia, and run startx (and you get your blank screen), then try ctrl-alt-backspace to see if you can kill X. If that doesn't help, maybe ctrl-alt-F2 will (swich to virtual console 2).

You need to get to a working shell, BEFORE you reboot. Because when you reboot, X starts up again (since you told Mandrake to do that... not always a good idea, like now), and your log from the blank screen run gets overwritten.

If none of that helps, you can edit /etc/inittab and change the id:5:initdefault: line to read id:3:initdefault: instead. Then reboot, and you'll be looking at a textmode login screen. Do the modprobe and startx, then when it fails, reboot but do NOT startx until you've copied the /var/log/XFree86.0.log file to somewhere else (to later attach here).

You can revert to a graphical login later, by changing the 3 back to a 5, but I'd be wary of that. If something else ever gets misconfigured with X, you'll have to do the same thing. Personally, I find it's not hard to startx on every login, but of course, you may not agree with me. Whatever.

And modprobe'ing nvidia then rebooting will accomplish exactly nothing. When you modprobe (or insmod, it's the same thing) a module, you're loading it from disk into memory. When you reboot, you're clearing all memory, so the module effectively gets unloaded.

[tsubasa]

No I dont reboot after I tape modprobe nvidia lol, I tape startx after . And The blank screen arrive.
( I have installed the drivers in init 3 mode)

I restart in Failsafe for copy the log
and XFconf it is same file you have, the same error.
But i am disapointted you dont say to me to try some "command" for help me ...
I am waiting for your instruction, my linux partition is on suspend with this error. You have the Xfconf.log and I have a black screen for symptom. I really would like to try some command for resolve the problem.

Thanx a lot a lot a lot a lot...
I really sorry to boring you with that, but i really would like to see my nvidia working and after play Counter strike on my mandrake :) thanks thanksssss man

[bwkaz]

The log file that I have is from a run where the kernel module wasn't loaded.

Load the kernel module, rerun startx, and post the new log.

[tsubasa]

Hello,
I have set my etc/inittab at 3.
I reboot my pc, and boot to linux in normal mode.
And I arrive at init 3 in konsole mode.
In Root I do Modprobe nividia, i have a message like that nvidia module loaded with warning.
i do Startx and 1 second after blank screen.

I try Ctrl+Alt+Backspace, nothing ...
I try Ctrl+Alt+F2, nothing. Another Ctrl+alt+F3 , nothing too. And I finally try ctrl+alt+suppr, nothing too.
My pc is totally blocked at blank screen.
I am oblige to reboot, and I do.
My computer boot at init 3 and I copy the /var/log/XFree86.0.log to my home partition.

This is the files but... it is blank ...

:'( i don't understand

[bwkaz]

What motherboard chipset do you have? Is it a KT400 by any chance?

Take a look in /proc/driver/nvidia/agp/host-bridge.

[etr]

Originally posted by tsubasa
I try Ctrl+Alt+Backspace, nothing ...
I try Ctrl+Alt+F2, nothing. Another Ctrl+alt+F3 , nothing too. And I finally try ctrl+alt+suppr, nothing too.
My pc is totally blocked at blank screen.
I am oblige to reboot, and I do.
My computer boot at init 3 and I copy the /var/log/XFree86.0.log to my home partition.

Just a sick thought real quick...

Do you have another computer handy? If so, I would probably consider getting the telnet daemon up so you can try to log in and poke around before re-booting. If you're lucky, you might even be able to kill X (if it's still running) and regain some control.

Originally posted by bwkaz
You can revert to a graphical login later, by changing the 3 back to a 5, but I'd be wary of that. If something else ever gets misconfigured with X, you'll have to do the same thing. Personally, I find it's not hard to startx on every login, but of course, you may not agree with me. Whatever.

I'm in complete agreement here, but I'd like to note (for tsubasa's benefit) that you one can always add a line to .bashrc to run startx--so long as there is a back-up account you can fall back to when X isn't available.

[tsubasa]

Hello,

My motherboard Is an MSI ULTRA KT333 with an athlon XP2000+.

My Graphic Card is this one :
3D Prophet III Titanium 500 from http://www.hercules.com

Thanks a lot all to help me .

[bwkaz]

Try to poke around, like etr is saying, but if that doesn't work, then the linux-bugs@nvidia.com email address will probably want to know. Give them all your hardware (as in chipset, video card, video BIOS (which you can get at the early, early screen when you first turn on your computer -- when your video card does its power-on self test, before the rest of your computer does its POST), perhaps motherboard BIOS version (accessible from the screen where the rest of your computer is doing its POST), and stuff like that), plus what it's doing -- the blank screen on startx with the module loaded.

They probably don't need to know about the "could not initialize NVIDIA driver" issue, since that's a separate one.

Might also be good to tell them whether you see an nVidia logo or not when you startx -- it sounds like you don't.

Anyway, yeah, this is an "I'm out of ideas" post... :(

[tsubasa]

In all of my test, i have never seen a Nvidia logo.

:confused:
Not lucky is my second name in linux hardware

But i dont understand, i have no tricks to try. Why ?
U say :
"What motherboard chipset do you have? Is it a KT400 by any chance?

Take a look in /proc/driver/nvidia/agp/host-bridge."
I say i have a KT333 u dont think the origine of problem is the same on the 2 mainboard ?


Why u dont demand to me to try some command for KT400 like that :

"I've got a GeForce3 sitting in VIA KT400 chipset.
First I've tried the 2.4.21-rc1-ac4 kernel which supposedly has good kt400-agpgart (only 2x) support. Didn't work.
Then I've set all necessary options in the XF86Config to use NvAgp, set AGP to 4x (hey, if agp then let's do it right), installed the latest driver 43.63 (without the installer) - all options were right...but all I got was a lousy black screen with nothing else to do then reboot.
I've fought with this error for hours...to find a solution in a completely different corner. It has turned out that IRQ handling must be managed by the BIOS and not ACPI (now go figure why).
All you need to do is insert an append string like that into your lilo.conf:

append = " pci=noacpi pci=biosirq"
"
Or why u dont demand to me to replace my agp drivers ?
you sure my probleme dont come from that ?
How i can for install NVagp ?

Please do u have some ideas of commands or tricks sensible to get my system work ?

I have seen on this forum a lot of error related to KT400 mainboard, for all this people its impossible and insoluble like me ?
No i think some people have get her nvidia running well on KT400 and KT333 certainly, how this people do ? i want to know :'(

I dont understand why u say i must post my problem at nvidia support, i work hard to retrieve all this log, its hopeless to think andy is an official Support for us? if is true, andy have my log and my description probleme here, if is false i dont understand why andy have in her signature Nvidia Coporation.


Thanks for all :confused: :confused: :confused: :confused:

Sorry for my little brutality but iam choked.

[tsubasa]

I have the /dev/nvidia0 device it is good ?

[tsubasa]

Any no news on how to install Nvidia Graphics on Via KT333 or KT400 Mainboard ?

Thanks a lot ! :)

[Gyokuran]

OK, I'll give some details on my successfull installation hoping that it will help. The following settings enable AGP 4x with the nvidia agp driver.

Hardware:
Gainward GeForce3 with VIA KT400

Linux kernel: 2.4.20, patched with newest acpi and bttv drivers. IMPORTANT: disable agpgart support in kernel (it only supports AGP x2 for KT400 and not good anyway)

NVIDIA driver version: 43.63

/etc/X11/XF86Config: (relevant parts)
....
Section "Module"
Load "dbe"
Load "type1"
Load "speedo"
Load "extmod"
Load "freetype"
Load "v4l"
Load "glx"
EndSection
....
Section "Device"
Identifier "Device[0]"
Driver "nvidia"
BoardName "Gainward"
Option "dpms"
Screen 0
Option "NoLogo" "true"
Option "NvAgp" "1"
Option "PageFlip" "true"
Option "DigitalVibrance" "2"
Option "HWCursor" "true"
Option "SWCursor" "false"
EndSection

/etc/modules.conf:
...
alias char-major-195 nvidia
options nvidia NVreb_EnableVia4x=1
options nvidia NVreg_ReqAGPRate=0x4

/etc/lilo.conf
...
image = /boot/kernel2420
label = linux-noacpi
root = /dev/hda7
append = " pci=noacpi pci=biosirq"
...
(do not forget to run lilo after changing lilo.conf)

Good luck.

[tsubasa]

Hello !

I always have a append line "but different" in my lilo.conf.
Can i add a second append line ?
append = " pci=noacpi pci=biosirq"

Linux kernel: 2.4.20, patched with newest acpi and bttv drivers.
How i can DO that ?
I think in my mandrake 9.1 the kernel is 2.4.

IMPORTANT: disable agpgart support in kernel (it only supports AGP x2 for KT400 and not good anyway)

How i can DO that ?

I think it maybe a good way but i need more step by step please :) Thank for your solutions mann !! :)