Frustrated: Nvidia Drives and RH8

[jjsimas]

"startx" fails saying that monitors were found but that none have a usable configuration. their configuration is exactly the same that works with vesa driver. my refresh rates are correct: HorizSync 30-110
VertRefresh 48-170
i made sure pnp os was disabled in my bios, but couldn't find a setting for assigning an IRQ to VGA... any help would be appreciated. here is my latest config file. my monitors are dual sony g410's. and my card is an MSI GF4 Ti 4200 AGP Twin View.

Section "Files"
RgbPath "/usr/X11R6/lib/X11/rgb"
FontPath "unix/:7100"
EndSection
Section "ServerFlags"
AllowMouseOpenFail
EndSection
Section "InputDevice"
Identifier "Keyboard1"
Driver "Keyboard"
Option "AutoRepeat" "250 30"
Option "XkbRules" "xfree86"
Option "XkbModel" "pc105"
Option "XkbLayout" "us"
EndSection
Section "InputDevice"
Identifier "Mouse1"
Driver "mouse"
Option "Protocol" "IMPS/2"
Option "Device" "/dev/psaux"
EndSection
Section "Module"
Load "dbe"
Load "glx"
Load "extmod"
Load "type1"
Load "freetype"
EndSection
Section "Monitor"
Identifier "MyMonitor"
VendorName "Mitsubisi"
ModelName "Diamond Plus 120u"
HorizSync 30-110
VertRefresh 48-170
EndSection
Section "Device"
Identifier "NV AGP TwinView"
VendorName "nvidia"
Driver "nvidia"
BusID "PCI:1:0:0"
Option "TwinView"
Option "SecondMonitorHorizSync" "30-110"
Option "SecondMonitorVertRefresh" "48-170"
Option "TwinViewOrientation" "RightOf"
Option "MetaModes" "1280x1024,1280x1024; 1024x768,1024x768"
Option "ConnectedMonitor" "crt,crt"
EndSection
Section "Screen"
Identifier "Screen AGP TwinView"
Device "NV AGP TwinView"
Monitor "MyMonitor"
DefaultColorDepth 24
Subsection "Display"
Depth 8
Modes "1280x1024" "1024x768" "800x600" "640x400"
EndSubsection
Subsection "Display"
Depth 16
Modes "1280x1024" "1024x768" "800x600" "640x480"
EndSubsection
Subsection "Display"
Depth 24
Modes "1280x1024" "1024x768" "800x600" "640x480"
EndSubsection
EndSection
Section "ServerLayout"
Identifier "AGPTwinView"
Screen "Screen AGP TwinView"
InputDevice "Mouse1" "CorePointer"
InputDevice "Keyboard1" "CoreKeyboard"
EndSection

[bwkaz]

Hmmm... I don't see anything immediately obvious in the config file (although it is a little hard to read without the indentation -- add [c*de] and [/c*de] (but change the * to an o) around the file next time, and that should help that).

I think the log file would give a hint for where to look, though. Can you attach it?

[jjsimas]

here is the log file.

[bwkaz]

(EE) NVIDIA(0): Failed to initialize the NVIDIA kernel module!
(EE) NVIDIA(0): *** Aborting *** Is there anything in dmesg or /var/log/messages that may be relevant?

[jjsimas]

dmesg tells me that the agp card is sharing irq 11 with the motherboards ide controller i think.

[jjsimas]

i checked out /var/log/messages and couldn't find anything unusual. only think unusual about my system is that when redhat is booting, it takes like one minute to get bast the "checking for new hardware" screen. perhaps there is a hardware issue to resolve. i have amd 266a chipset. and sometimes amd stuff doesn't work well... works fine on windows though. perhaps i should upgrade to intel this summer.

[bwkaz]

IRQ 11 shouldn't be a problem (mine's on the same IRQ), but the fact that it's sharing it with the IDE controller might be. Are you sure the IDE controller wasn't using 14 or 15? My ide0 is on IRQ 14, and ide1 is on 15, and that's the way it's been on most systems I've seen...

Try looking at /proc/interrupts to see what's sharing what. If possible, load the modules for all the hardware other than the nVidia driver first, then load the nVidia one, to hopefully make it take its own IRQ. If that doesn't work, try loading the nVidia module first, and the others later.

[jjsimas]

Here's the output from /proc/interrupts, looks like
no IRQ is assigned to my vid card?


CPU0
0: 142093 XT-PIC timer
1: 411 XT-PIC keyboard
2: 0 XT-PIC cascade
3: 141 XT-PIC ide2, usb-uhci, usb-uhci, usb-uhci
4: 402 XT-PIC eth0
8: 1 XT-PIC rtc
10: 0 XT-PIC cmpci
12: 8496 XT-PIC PS/2 Mouse
14: 8783 XT-PIC ide0
15: 230 XT-PIC ide1
NMI: 0
ERR: 0

[bwkaz]

Was the module loaded when you looked at /proc/interrupts? I should have said something about that... it has to be.

If the module was in fact loaded, then the problem is the "assign irq to vga" setting that you said you couldn't find. Talk to your motherboard manufacturer (or system mfr. if you bought this all as a package) and get them to tell you how to get the BIOS to assign an IRQ to the video card. If they say you don't need to do that, tell them they're wrong. ;)

[jjsimas]

sorry i don't know how to determine wether the module was loaded / or in what order it was loaded.

[SubTexel]

insmod nvidia

Should tell you if its loaded or not, (if it's not it will load it, if it is loaded it'll tell you its been installed already..

[jjsimas]

insmod says no module by that name found. so it doesn't load it.

[bwkaz]

insmod nvidia will only load the driver if you have installed driver version 4191. If the last driver that you installed was <= 3123, then you have to insmod NVdriver instead.

But the real way to tell what's loaded is to lsmod. Then, if you see either NVdriver or nvidia in the output, you can conclude that the kernel module is loaded, regardless of what it's called. Much easier. ;)

If lsmod doesn't show either nvidia or NVdriver, then you have to load it, by modprobe'ing either NVdriver or nvidia. Since insmod nvidia didn't work, it seems that you either:

1) Don't have any NVIDIA_kernel package installed
2) Have the wrong NVIDIA_kernel package installed (i.e. you installed one that was looking for the wrong Linux kernel version)
Or
3) Have NVIDIA_kernel.1.0-3123 or before installed.

Do an rpm -qa | grep -i nvidia to show which packages you have, and also a uname -r to show which kernel you run. Post the results of both of these.

[jjsimas]

ok i did lsmod:
no nvidia and no NVdriver

then i did the rpm -qa | grep -i nvidia and got:

[root@root]# rpm -qa | grep -i nvidia
NVIDIA_kernel-1.0-4191.rh80up_2.4.18_18.8.0
NVIDIA_GLX-1.0-4191
[root@root]# uname -r
2.4.18-24.8.0

even when i was running kernel 2.4.18-18.8.0 i was
having the same problem.

so what does it mean when the rpm command
says i have the write things, but the lsmod
command says i don't?

what can i do to get the nvidia module loaded?

[Stratman]

You will need to remove the current Nvidia rpm's. They were compiled for your 2.4.18_18.8.0 kernel.

So start with this

rpm -e NVIDIA_GLX
rpm -e NVIDIA_kernel

Once those are removed, make new ones for your 2.4.18-24.8.0 kernel. I would suggest grabbing the kernel (http://download.nvidia.com/XFree86_40/1.0-4191/NVIDIA_kernel-1.0-4191.src.rpm) drivers as well as the GLX (http://download.nvidia.com/XFree86_40/1.0-4191/NVIDIA_GLX-1.0-4191.src.rpm).

Then do the following

rpmbuild --rebuild NVIDIA_kernel-1.0-4191.src.rpm
rpmbuild --rebuild NVIDIA_GLX-1.0-4191.src.rpm
rpm -ivh /usr/src/redhat/RPMS/i386/NVIDIA_kernel-1.0-4191.i386.rpm
rpm -ivh /usr/src/redhat/RPMS/i386/NVIDIA_GLX-1.0-4191.i386.rpm

Then edit /etc/X11/XF86Config(-4). Making sure the driver is nvidia, "Load" "glx" is there. Once that is done all that's left is init 3 then init 5. You should be up and going then.

Yes you need to assign an IRQ to your video, other than that you should be seeing the Nvidia Logo as X starts up.

[jjsimas]

you are awesome.

got paypal? i can send you some $.

or if you are nvidia technician, then i can donate to fav. cause...

jason

thanks again.

[Stratman]

Not a technician....... AT Nvidia anyways :)

No problem....

You can send a hit or two my way (http://www.linuxgaming.net ) if you want :afro: That is the p1mp icon right LOL

[zatnktel]

try the tar ball for the kernel driver i had a problem like that the kernel rpm does not install the nvidia.o but the tar ball does

[bwkaz]

OK, this guy's problem is solved too... but anyway,

Originally posted by zatnktel
the kernel rpm does not install the nvidia.o but the tar ball does Yes, the kernel RPM does install nvidia.o (assuming you're using 4191). The thing is, though, if the version of your running kernel doesn't match the version kernel that the rpm was built for (which is right in the name of the rpm file, btw), it will install it into the wrong place.

If you download NVIDIA_kernel-xxxx-2.4.18-18.8.0, for example, then it is hardcoded to put the nvidia.o file into /lib/modules/2.4.18-18.8.0/kernel/drivers/video. If you are not running kernel 2.4.18-18.8.0 (if you've upgraded to -24.8.0 for example), then it will appear the nvidia.o never got installed, but it did. It's just not living where modprobe is looking for it.