[SOLVED] NVIDIA GeForce4 TI 4200

[MartinCz]

Hi,

I am quite new to Linux. I have installed Mandrake 10.1 Community. Then installed Kernel Source and then I installed NVIDIA NForce drivers (For motherboard).

Then I installed the NVIDIA Graphics Drivers 1.0-7174. The readme says I have to do some changes in the /etc/X11/xorg.conf file. The most important change is "nv" --> "nvidia". I have made it.

The problem is I cant run the XWindow (KDE) anymore. In the log file I can find this error output:

"(EE) NVIDIA(0): The NVIDIA kernel module does not appear to be receiving
(EE) NVIDIA(0): interrupts generated by the NVIDIA graphics device.
(EE) NVIDIA(0): Please see the FREQUENTLY ASKED QUESTIONS section in the
(EE) NVIDIA(0): README for additional information."

There were some generic drivers installed by the installer of Mandrake 10.1 nut the 3d hardware acceleration was not running (glxinfo). For example the TuxRacer was like "MS Powerpoint" :o). That is why I decided to change the drivers for the original from NVIDIA. Now I am stuck at this point and dont know what to do.

I know that you were solving a similar problem but I did not understand it much, so - please - could somebody help me - step by step.

The readme file says this error can occure because of many reasons and that I should set some boot parameters but I dont know where to make these changes.

Regards,
MartinCz.:(

[je_fro]

make sure you go into your bios and DISABLE plug-and-play OS.
Then with a text editor (as root) edit /boot/grub/grub.conf

and add "acpi=off" to your kernel line...ie, the line that says something like:
kernel (hd0,1)/boot/bzImage-gaming root=/dev/hda3 video=vesa:ywrap,mtrr vga=0x31A hdd=ide-scsi acpi=off

[JamminJoeyB]

I don't think your nvidia module is loaded. Follow the instructions below and make sure it's being loaded.

From the read me.

Q: Where should I start when diagnosing display problems?

A: One of the most useful tools for diagnosing problems is the X
log file in /var/log (the file is named: "/var/log/XFree86.<#>.log" or
"/var/log/Xorg.<#>.log" where "<#>" is the server number -- usually 0).
Lines that begin with "(II)" are information, "(WW)" are warnings, and
"(EE)" are errors. You should make sure that the correct config file
(ie the config file you are editing) is being used; look for the line
that begins with: "(==) Using config file:". Also check that the
NVIDIA driver is being used, rather than the 'nv' or 'vesa' driver;
you can look for: "(II) LoadModule: "nvidia"", and lines from the
driver should begin with: "(II) NVIDIA(0)".

[bwkaz]

Originally posted by MartinCz
"(EE) NVIDIA(0): The NVIDIA kernel module does not appear to be receiving
(EE) NVIDIA(0): interrupts generated by the NVIDIA graphics device.
(EE) NVIDIA(0): Please see the FREQUENTLY ASKED QUESTIONS section in the
(EE) NVIDIA(0): README for additional information." From the cited FREQUENTLY ASKED QUESTIONS section in the README (/usr/share/doc/NVIDIA_GLX-1.0/README):

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

"(EE) NVIDIA(0): The NVIDIA kernel module does not appear to be receiving
(EE) NVIDIA(0): interrupts generated by the NVIDIA graphics device.
(EE) NVIDIA(0): Please see the FREQUENTLY ASKED QUESTIONS section in the
(EE) NVIDIA(0): README for additional information."

A: This can be caused by a variety of problems, such as PCI IRQ routing
errors, I/O APIC problems or conflicts with other devices sharing
the IRQ (or their drivers).

If possible, configure your system such that your graphics card does
not share its IRQ with other devices (try moving the graphics card
to another slot (if applicable), unload/disable the driver(s) for the
device(s) sharing the card's IRQ, or remove/disable the device(s)).

Depending on the nature of the problem, one of (or a combination of)
these kernel parameters might also help:

pci=noacpi (don't use ACPI for PCI IRQ routing)
pci=biosirq (use PCI BIOS calls to retrieve the IRQ routing table)
noapic (don't use I/O APICs present in the system)
acpi=off (disable ACPI) First, I'd try doing a cat /proc/interrupts and see which IRQ "nvidia" is listed on. If it's sharing with anything else, I'd try moving the card that it's sharing with over to another PCI slot (if applicable).

If it isn't sharing, or if I can't move the other card (for example, because it's an onboard device), then I'd start trying those kernel parameters.

[MartinCz]

OK. I'll try these and post my results. Thank you all.

[MartinCz]

Well,

1. where I can find if the nvidia module is loaded? I tried

modeprobe -l nvidia

and it found one think.

2. I made a change according to the readme file - i added "nvidia" to the modprobe.conf or how it is called. The log file "/var/log/XFree86.<0>.log says it is using /etc/X11/xorg.conf and it loaded the driver at boot time then tried to initialize and failed. So it unloaded it back.

3. When I look to /proc/interrupts I can see no nvidia entry there :(

4. I have LILO insteda of GRUB and could not find the config files for it. In /boot/ there are configs for grub but I cant see lilo there. I will have to take a look in another folder...

5. When I put "nv" at the place of the edited "nvidia" back, the KDE starts ok after startx.

Thank you for your help.

Regards,
MartinCz.

[je_fro]

when I was using my nforce board, I quite often had to pass noapic and acpi=off at the grub kernel line or things would freeze. At least with the default kernels from Debian and Slackware. I don't know how you'd do it in lilo, but lilo.conf is the place to look.

[MartinCz]

I found the lilo.conf file. It is /etc/lilo.conf.

There are the choices which lilo shows. I created a new kernel profile in lilo.conf and i put there all these:

pci=noacpi (don't use ACPI for PCI IRQ routing)
pci=biosirq (use PCI BIOS calls to retrieve the IRQ routing table)
noapic (don't use I/O APICs present in the system)
acpi=off (disable ACPI)

then I ran lilo to apply changes. It did not help.

When I want to start windows x, still the same EError :)

Maybe this is imporatant: When i boot, it writes something like:

"Could not find rawctl somewhere. No such file or directory".

I dont know whats going on here but hope somebody of you gurus can help me. I am looking foreward to playing TuxRacer :o)

Regards,
MartinCz.

[bwkaz]

Originally posted by MartinCz
3. When I look to /proc/interrupts I can see no nvidia entry there :( Well, I would start there. If you've done an /sbin/modprobe nvidia as root and it still doesn't show up, then something isn't giving your card an IRQ line.

If that is the case, then go into your BIOS setup program and turn on "assign IRQ to VGA" or whatever it's called.

[MartinCz]

The command:

/sbin/modprobe nvidia

writes down a line, some path of a file with a .ko extension.

In BIOS, the "Assign IRG to VGA" IS ENABLED.

So, well, erm... I am not sure what the modprobe command does in fact, and I dont know if it is good or not if modprobe nvidia prints anything, so what should be the next steps?

In BIOS I found a ACPI enable / disable option. I originally had enabled, but I also tried disabled, with no success.

Please, hope I dont have to buy a new VGA card :o) In Windows everything is OK, and the original "nv" driver in Mandrake was also ok, exept there was no hardware accleration at all (using glxinfo).

Regards,
MartinCz.

PS: I think there could be some problem with the "could not load rawctl" at boot time. When I try to startx i get those errors above mentioned and when rebooting or poweroffing computer, alsa can not be stopped and I must do hard reset or poweroff.

[je_fro]

Do this:
man modprobe
man man
in a terminal.
The fact that "/sbin/modprobe nvidia" returns ANYTHING is a sign of an error.
What is the output from
modprobe nvidia
lsmod
lspci
?

[Rinias]

Hey-

if you do a modprobe nvidia you need to do it in the CLI (text part- black and white). Then you need to try your startx or display manager (gdm or kdm, according to which is installed).

If you want to see if you loaded the module, type lsmod |grep nv

lsmod lists the modules that are currently loaded.

| (a "pipe" above the enter on an american keyboard) tells the result of the command to be "piped" into another command

grep is a command that searches for a certain string- here we are searching for the string nv

normally if you installed everything correctly (nforce also) you should see nvidia, nvnet and nvsound.

Otherwise, heed the advice already given to you. I had the GF4ti4200 and MDK 10.1 RC, and I never had any problem with it...

Maybe its something else? did you try reinstalling the driver? Do you have your kernel sources etc installed?

good luck!

@+

[Rinias]

Originally posted by je_fro

The fact that "/sbin/modprobe nvidia" returns ANYTHING is a sign of an error.


I agree with the rest of what you said, but this is not always true... In the case of Nvidia, modrpobe can often give a warning about nvidia modules tainting the kernel. And in this case that's exactly what we'd like... :D

@ +

[MartinCz]

Hi. Thank you all much for your help. I'll try all of what you've written tommorow and let you know the results.

[MartinCz]

OK.

I think it should be good to sum my problem up.

1. I am using LILO. I have edited /etc/lilo.conf so it now contains this start item which is defined as follows:

image=/boot/vmlinuz
label="linux-noacpi"
root=/dev/hda3
initrd=/boot/initrd.img
append="noapic acpi=off splash=silent pci=noacpi"
vga=788
read-only

The bold text is what I have edited, the rest is a copy of an item which already was in the .conf file.

2. My distro is Mandrake 10.1 community version. There is no gdm or kdm. I am not sure what this is, but mandrake provides mcc (media control center).

3. I have installed drivers for motherboard (nforce) and vga (nvidia). Before this i had to install kernel-source which was without any problem. I have tried to reinstall the drivers (both nforce and nvidia) with no success.

4. When linux is waking up, it says it cant find some master raw device in /dev/ravctl or something like that. I cant find it in /var/boot.log so I am not sure, it is quickly out of scope.

5. modprobe nvidia says nothing

6. the nvidia module is probably loaded at the linux-load time, i have added line:

nvidia

to the /etc/modprobe.preload file and it makes no problem.

This is, when I run

lsmod | grep nvidia

it says:

nvidia-agp 5788 1
nvidia 3915452 0
agpgart 27752 nvidia-agp, nvidia

Which means that the module is loaded ok.

7. During the installation of Mandrake on my system, I selected Geforce 4 (generic) driver, which was functioning ok but without the 3d acceleration support. In /etc/X11/xorg.conf i have made one change:

in Section "Device" i have changed

Driver "nv" to Driver "nvidia"
This make me not able to startx. When i put back nv, it is ok.

8. When I was in one of my previous posts saying that modprobe says someting with .ko extantion I was not right. I meant
modprobe -l nvidia, which seams to be quite similar to lsmod nvidia.

9. When I try to startx, the screen is black for about 8 seconds, then it comes back to text mode and writes the error I have written in my first post. It also writes some message coming from syslogd. It says disabling IRQ #11.

10. I have tried to switch off ACPI (or APIC?? not sure) in BIOS, then PnP in BIOS... no success.

11. The nvnet and nvsound are not printed out by lsmod | grep nv, because I dont use them, I have SB Audigy 2 for sound and Realtec NIC. But hope that the rest of the motherboard drivers is alive and kicking :o) (like fast IDE and whatever it should be there).

12. now to lspci:

00:00.0 Host bridge: nVidia Corporation nForce2 AGP (different version?) (revc1)
00:00.1 RAM memory: nVidia Corporation nForce2 Memory Controller 1 (rev c1)
00:00.2 RAM memory: nVidia Corporation nForce2 Memory Controller 4 (rev c1)
00:00.3 RAM memory: nVidia Corporation nForce2 Memory Controller 3 (rev c1)
00:00.4 RAM memory: nVidia Corporation nForce2 Memory Controller 2 (rev c1)
00:00.5 RAM memory: nVidia Corporation nForce2 Memory Controller 5 (rev c1)
00:01.0 ISA bridge: nVidia Corporation nForce 2 ISA Bridge (rev a4)
00:01.1 SMBus: nVidia Corporation nForce2 SMBus (MCP) (rev a2)
00:02.0 USB Controller: nVidia Corporation nForce2 USB Controller (rev a4)
00:02.1 USB Controller: nVidia Corporation nForce2 USB Controller (rev a4)
00:02.2 USB Controller: nVidia Corporation nForce2 USB Controller (rev a4)
00:04.0 Ethernet Controller: nVidia Corporation nForce 2 Ethernet Controller (rev a1)
00:08.0 PCI Bridge: nVidia Corporation nForce2 External PCI Bridge (rev a3)
00:09.0 IDE interface: nVidia Corporation nForce2 IDE (rev a2)
00:1e.0 PCI Bridge: nVidia Corporation nForce2 AGP (rev c1)
01:06.0 Multimedia audio controller: Creative Labs SB Audigi (rev 04)
01:06.1 Input device controller: Creative Labs SB Audigi MIDI/Game port (rev 04)
01:06.2 (IEEE 1394): Creative Labs SB Audigi FireWire Port (rev 04)
01:ob.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8139/8139C/8139C+ (rev 10)
01:0d.0 FireWire (IEEE 1394): Lucent Microelectronics FW323 (rev 61)
02:00.0 VGA compatible: nVidia Corporation NV25 [GeForce Ti 4200] (rev a3)

Uffff. I had to rewrite lspci from my big computer's crt to my notebook.

I think - the last line of this pintout (lspci) looks good - it knows that I have GF TI 4200 :o)

SO, gurus, please ... hope this is not the case of wrong problem :)

Thank you for your ideas and suggestions.

Regards,
MartinCz.

[XiaoKJ]

I have never seen the module nvidia-agp in my life.

modprobe -r nvidia-agp will remove it. you can try startx after that.

if it works, then that is your problem and you can remove that entry from the module loader in mdk.

if it doesn't then continue --- I am only trying my best to help.

[MartinCz]

nvidia-agp is in use. how can i remove it from being loaded at boot time?

[MartinCz]

I meant it cant be removed cause it is in use...

[je_fro]

That (I think) is the kernel agp mocule for nvidia, which I NEVER use. I've removed any hint of agp out of my kernel .config and everything works fine. Try editing /etc/modprobe.preload and get rid of the nvidia-agp line by commenting it out.
like
# nvidia-agp
^ that comments it out.

[XiaoKJ]

anyone know what will use nvidia-agp in mdk?

because it can be serious matter.

I always resort to rmmod nvidia-agp if I get this kind of errors, because it can be devastating to remove it from the modules.preload esp if something important is using it.

Finally, did you use rivafb as your framebuffer of choice? it may cause problems too.

[MartinCz]

OK,

I have removed nvidia-agp from /etc/modprobe.conf.

Now it is not loaded at boot-time, it is not listed by lsmod | grep nvidia,

but I cant still startx.

Any other ideas? Or should I remove nvidia-agp entries from other configuration files?

Thank you.

[je_fro]

is nvidia loaded?

[MartinCz]

lsmod | grep nv

nvidia 3915452 0
agpgart 27752 1 nvidia

thats all

[je_fro]

take agpgart out also
then try startx
if it fails give me the output from
grep EE /var/log/Xorg.0.log

(or maybe it's /var/log/XFree86.log)

[MartinCz]

There is no agpgart in modprobe.preload ....

Where should I look for it?

I appreciate your help much.

[je_fro]

maybe it's built in...
try startx and what I said above...

[MartinCz]

OK. Just a moment... I must startx then go to win, run PartitionMagic, get the ext3 log file with it and I will send it. So, please, stand by...

[MartinCz]

OK. I made a selection of important files. I made restart, then immediatelly startx, then restart to win, then partitionmagic and copied the files. Here they go.

[je_fro]

We're back to this:
Q: My X server fails to start, and my X log file contains the error:

"(EE) NVIDIA(0): The NVIDIA kernel module does not appear to be receiving
(EE) NVIDIA(0): interrupts generated by the NVIDIA graphics device.
(EE) NVIDIA(0): Please see the FREQUENTLY ASKED QUESTIONS section in the
(EE) NVIDIA(0): README for additional information."

A: This can be caused by a variety of problems, such as PCI IRQ routing
errors, I/O APIC problems or conflicts with other devices sharing
the IRQ (or their drivers).

If possible, configure your system such that your graphics card does
not share its IRQ with other devices (try moving the graphics card
to another slot (if applicable), unload/disable the driver(s) for the
device(s) sharing the card's IRQ, or remove/disable the device(s)).

Depending on the nature of the problem, one of (or a combination of)
these kernel parameters might also help:

pci=noacpi (don't use ACPI for PCI IRQ routing)
pci=biosirq (use PCI BIOS calls to retrieve the IRQ routing table)
noapic (don't use I/O APICs present in the system)
acpi=off (disable ACPI)

what have you tried from all this?

[je_fro]

what is the output from
cat /proc/interrupts
?

[MartinCz]

I have put all four parameters to lilo.conf. Hope the syntax is ok. I have posted a snippet of the file in my todays first post. There I have put all of these as you can see:

pci=noacpi
pci=biosirq
noapic
acpi=off

But there must be a lot of combinations of these parameters and BIOS settings. Nothing worked, though.

The cat /proc/interrupts is still the same I hope (again my first or second todays post).

Thank you.

[je_fro]

start by taking out all those lilo options
pci=noacpi
pci=biosirq
noapic
acpi=off

then if that fails, just use acpi=off
if that fails use just noapic
carefully go through your bios
let me know what happens

[MartinCz]

ok

[MartinCz]

What bios settings should I use for this testing? PnP off and ACPI off?

[je_fro]

PnP off for sure.
allocate irq's should be ON.

[MartinCz]

In BIOS, there is APIC Mode: enabled/disabled.

Should I change to disabled for this testing?

In bios I can find these:

PnP on/off - it is now off
Assign IRG to VGA - it is on
and finaly APIC Mode

Thats all important and relevant.

[bwkaz]

Originally posted by je_fro
take agpgart out also This won't help. (And actually, I highly doubt that taking nvidia-agp out would help either, see below.)

The most recent nvidia kernel modules explicitly depend on agpgart. (If you use the built-in nvidia AGP driver, it still uses the agpgart module as the AGP frontend, with the nvidia module as the backend. If you use the default setting (kernel's AGP driver), then nvidia uses the preexisting AGP support that's either built into the kernel or loaded at boot time.

The nvidia-agp driver is the kernel's AGP bridge driver for nForce motherboards. If you don't plan on using the nvidia driver's AGP support, you will NEED this module installed (along with agpgart, which you will need in any case, unless you don't want any AGP support at all).

Originally posted by MartinCz
There is no agpgart in modprobe.preload .... That's because it's loaded when you modprobe nvidia.

Originally posted by je_fro
maybe it's built in...
try startx and what I said above... Well, no, because then it wouldn't show when you do an lsmod. ;)

It's an explicit dependency of nvidia.ko, that's why it's loaded. This is not a problem.

Back to previous posts...

Originally posted by MartinCz
9. When I try to startx, the screen is black for about 8 seconds, then it comes back to text mode and writes the error I have written in my first post. It also writes some message coming from syslogd. It says disabling IRQ #11. What's the exact message, in its entirety?

If the kernel is disabling IRQ 11 for some reason, that's why the nvidia driver is complaining that it can't receive interrupts from the card. We need to find the reason the kernel is disabling that IRQ, and fix it. But we can't without knowing the exact message (and possibly a few others that get logged around the same time).

Other comments:

We still haven't seen the output from cat /proc/interrupts, and (just to double-check) are you rerunning /sbin/lilo as root every time you update lilo.conf?

[MartinCz]

HI FRIENDS

I FINALLY GOT IT WORKING !

It was not dependant on the settings of my BIOS. The problem was really in the kenrel parameters passed by LILO. The original parameters added by Mandrake 10.1 installer did not work, and adding all the parameters listed in README of nvidia at once was not working either.

Now I have as parameters only: splashscreen=silent and that's it! Everything is ok. I also can add one of the parameters suggested in the readme file.

I will make this more concrete as soon as I get my brother out of the computer.

PS: The parameter which was making mess in the original state of LILO was probably acpi=ht or something like this. And my version of set of parameters, it was wrong because they where all added at once.

[je_fro]

Originally posted by bwkaz

The most recent nvidia kernel modules explicitly depend on agpgart. (If you use the built-in nvidia AGP driver, it still uses the agpgart module as the AGP frontend, with the nvidia module as the backend. If you use the default setting (kernel's AGP driver), then nvidia uses the preexisting AGP support that's either built into the kernel or loaded at boot time.

So why is it that my stuff works with no agpgart _at_all_ in the kernel? And (for older releases) it works far better this way using nvidias agp driver.

< > /dev/agpgart (AGP Support)
< > Direct Rendering Manager (XFree86 4.1.0 and higher DRI support)

using nvidia-kernel-1.0.7174

[bwkaz]

Originally posted by je_fro
So why is it that my stuff works with no agpgart _at_all_ in the kernel? I don't know. Maybe it determines what drivers were built into the kernel (either built-in, or as modules) at driver compile time?

What does your /proc/driver/nvidia/agp/status file say about which AGP support it's using? I'm guessing it's probably "Disabled", but it might be "NVIDIA".

I pass NVreg_NvAGP=1 to the nvidia module at load time (using an "options" line in modprobe.conf), in order to disable use of the kernel via-agp driver in favor of nVidia's. But if I didn't do that (and didn't add via-agp to /etc/hotplug/blacklist), the driver would use via-agp -- which also works fine for me. It used to be slightly slower, but not much (and I haven't checked in a while).

(Side note: Of course DRM should be off. That won't ever be used by the nvidia driver. ;))

[je_fro]

jeff@speedy jeff $ cat /proc/driver/nvidia/agp/status
Status: Enabled
Driver: NVIDIA
AGP Rate: 8x
Fast Writes: Disabled
SBA: Enabled

[bwkaz]

Well then, I have to assume that it detects at driver-compile time whether you have the agpgart.ko module turned on (or configured in), and adjusts itself accordingly. I know mine needs it -- or at least, it uses it:

$ cat /proc/driver/nvidia/agp/status
Status: Enabled
Driver: NVIDIA
AGP Rate: 8x
Fast Writes: Enabled
SBA: Enabled
$ lsmod | grep 'nvidia\|agp'
nvidia 3920764 12
agpgart 34600 1 nvidia Notice, no via-agp (which is why Driver: NVIDIA is above), but nvidia still uses agpgart.