Using: Intel DG41 RQ board
Intel Core2Duo E8400 processor
NVIDIA 9800GT
Dual boot- Windows 7 and Linux Mint 9
Kernel Linux 2.6.32-21-generic
GNOME 2.30.0

Problem: Installed Linux Mint 9. Installed nvidia driver from Linux Hardware detection utility. Then, manually installed NVIDIA 195.36.24 driver. Worked OK until I updated certain things. I'm not sure what. Now when I boot, an UBUNTU error crops up and says that working in low graphics mode as NVIDIA kernel could not be loaded. I've tried everything that came to my noob mind (this is the first time I've divorced Mr. Gates). Reinstalled the driver. The resolution came up fine. When I reboot....the error comes up again.

To properly install the official nvidia driver you need to be root and working in run level 3. You also need the kernel development packages installed. The nvidia driver will then get compiled against the particular kernel you are using and will adjust your xorg.conf to suit.

Anytime you update the linux kernel you will have to repeat the process as the nvidia driver will not be compiled against the new kernel.

There may be other methods available for Mint which is basically Ubuntu AFAIK.

So the nvidia driver installs OK, and works, but then fails at the next boot.

Not sure about mint, but ubuntu 10.04 tries to load the nouveau driver for my nvidia card. If that happens, nouveau takes over the video hardware, and the nvidia driver cannot use it.
The solution is to blacklist the nouveau driver:

In /etc/modprobe.d/blacklist.conf add the line
blacklist nouveau

Then reboot.

It is possible that mint is using some other driver (not nouveau). Find out with lsmod in a terminal when you are running in "low graphics" and then you'll need to blacklist that one.

1 members found this post helpful.

@smoker
Working in run level 3. As root. Installed the drivers. As root ran nvidia-settings (gksu nvidia-settings). Checked and saved xorg.conf. Rebooted. And there it is again!!!
NVIDIA(0): Failed to initialize the NVIDIA kernel module.........................Screen(s) found, but none have a usable configuration.
By the way how can I find out if I've kernel development packages installed?

@tradegar
Didn't find nouveau in lsmod running low graphics. Found 'nvidia size 9932176 used by 0'. Apart from that couln't make head or tail of the output data. Shall try to post the complete result if you want.

Boo....hoo...hoo will somebody help out a poor neo-convert????

I take it this means the settings were lost ?

If so, how did you install Mint ? Was it from a LiveCD ?
You may have something that automatically sets up the graphics at boot, which you will need to disable.

sudo apt-get install build-essential
Should install all you need.

Mine says "nvidia size 9961216 used by 34"
So your module might be loaded, but it's not being used.

Ubuntu have changed the way the computer boots. Lots of things happen at once, rather than sequentially, so sometimes the nvidia driver gets loaded before the opensource one (and nvidia works), sometimes the opensource one gets loaded first and nvidia does not work.

The solution is to blacklist the offending module as in my post at #3

If it is not the nouveau driver it is probably nv which needs to be blacklisted. might give us a clue (when running in low graphics mode)

Be warned that the concept of "runlevel"s is gone from the more recent ubuntu (on which mint is based) releases. /etc/inittab does not even exist in ubuntu 10.04.

1 members found this post helpful.

Output of lsmod:


Module Size Used by
nls_utf8 1069 1
isofs 29250 1
binfmt_misc 6587 1
dm_crypt 11331 0
snd_hda_codec_realtek 203168 1
snd_hda_intel 21877 2
snd_hda_codec 74201 2 snd_hda_codec_realtek,snd_hda_intel
snd_hwdep 5412 1 snd_hda_codec
snd_pcm_oss 35308 0
snd_mixer_oss 13746 1 snd_pcm_oss
snd_pcm 70662 3 snd_hda_intel,snd_hda_codec,snd_pcm_oss
snd_seq_dummy 1338 0
snd_seq_oss 26726 0
snd_seq_midi 4557 0
snd_rawmidi 19056 1 snd_seq_midi
snd_seq_midi_event 6003 2 snd_seq_oss,snd_seq_midi
snd_seq 47263 6 snd_seq_dummy,snd_seq_oss,snd_seq_midi,snd_seq_midi_event
nvidia 9932176 0
vga16fb 11385 0
snd_timer 19098 2 snd_pcm,snd_seq
snd_seq_device 5700 5 snd_seq_dummy,snd_seq_oss,snd_seq_midi,snd_rawmidi,snd_seq
ppdev 5259 0
parport_pc 25962 1
vgastate 8961 1 vga16fb
snd 54148 16 snd_hda_codec_realtek,snd_hda_intel,snd_hda_codec,snd_hwdep,snd_pcm_oss,snd_mixer_oss,snd_pcm,snd_se q_oss,snd_rawmidi,snd_seq,snd_timer,snd_seq_device
psmouse 63245 0
serio_raw 3978 0
lp 7028 0
parport 32635 3 ppdev,parport_pc,lp
soundcore 6620 1 snd
snd_page_alloc 7076 2 snd_hda_intel,snd_pcm
dm_raid45 81647 0
xor 15028 1 dm_raid45
fbcon 35102 73
tileblit 2031 1 fbcon
font 7557 1 fbcon
bitblit 4707 1 fbcon
softcursor 1189 1 bitblit
vesafb 3542 1
usbhid 36110 0
hid 67032 1 usbhid
floppy 53016 0
intel_agp 24177 0
8139too 18545 0
8139cp 16186 0
video 17375 0
output 1871 1 video
r8169 33884 0
mii 4381 3 8139too,8139cp,r8169
agpgart 31724 2 nvidia,intel_agp



Can you please smoke out the culprit?

Can it be agpgart?

Here is a snippet of dmesg output just after starting in low graphics mode. I think it identifies the problem. But how to rectify it? A step by step advise shall be appreciated, if the community stalwarts please....

m,decodes=nonewns=io+mem
[ 7.555269] NVRM: loading NVIDIA UNIX x86 Kernel Module 195.36.15 Thu Mar 11 21:41:46 PST 2010
[ 14.804024] ata4.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 frozen
[ 14.804029] ata4.00: failed command: IDENTIFY DEVICE
[ 14.804035] ata4.00: cmd ec/00:01:00:00:00/00:00:00:00:00/00 tag 0 pio 512 in
[ 14.804036] res 40/00:01:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
[ 14.804039] ata4.00: status: { DRDY }
[ 19.844008] ata4: link is slow to respond, please be patient (ready=0)
[ 24.828009] ata4: device not ready (errno=-16), forcing hardreset
[ 24.828015] ata4: soft resetting link
[ 25.048645] ata4.00: configured for UDMA/133
[ 25.048656] ata4: EH complete
[ 26.064100] r8169: eth1: link down
[ 26.064232] ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 26.080824] eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
[ 26.214246] NVRM: API mismatch: the client has the version 195.36.24, but
[ 26.214247] NVRM: this kernel module has the version 195.36.15. Please
[ 26.214248] NVRM: make sure that this kernel module and all NVIDIA driver
[ 26.214248] NVRM: components have the same version.
[ 26.376502] NVRM: API mismatch: the client has the version 195.36.24, but
[ 26.376503] NVRM: this kernel module has the version 195.36.15. Please
[ 26.376504] NVRM: make sure that this kernel module and all NVIDIA driver
[ 26.376504] NVRM: components have the same version.
[ 26.534345] NVRM: API mismatch: the client has the version 195.36.24, but
[ 26.534346] NVRM: this kernel module has the version 195.36.15. Please
[ 26.534347] NVRM: make sure that this kernel module and all NVIDIA driver
[ 26.534348] NVRM: components have the same version.
[ 26.692921] NVRM: API mismatch: the client has the version 195.36.24, but
[ 26.692922] NVRM: this kernel module has the version 195.36.15. Please
[ 26.692923] NVRM: make sure that this kernel module and all NVIDIA driver
[ 26.692924] NVRM: components have the same version.
[ 26.871925] NVRM: API mismatch: the client has the version 195.36.24, but
[ 26.871927] NVRM: this kernel module has the version 195.36.15. Please
[ 26.871928] NVRM: make sure that this kernel module and all NVIDIA driver
[ 26.871928] NVRM: components have the same version.
[ 27.030984] NVRM: API mismatch: the client has the version 195.36.24, but
[ 27.030985] NVRM: this kernel module has the version 195.36.15. Please
[ 27.030986] NVRM: make sure that this kernel module and all NVIDIA driver
[ 27.030987] NVRM: components have the same version.
[ 36.512004] eth0: no IPv6 routers present
[ 81.452644] ISO 9660 Extensions: Microsoft Joliet Level 3
[ 81.482301] ISOFS: changing to secondary root

You may find this link useful :
http://ubuntuforums.org/showthread.php?t=650161

1 members found this post helpful.

No, I can't
When posting blocks of code,please use code tags (In this simple editor window, click "Go Advanced") because it makes it easier to read.
No.

The dmesg output is clear though. You have the wrong nvidia module version for your kernel, and it needs to be recompiled from the binary on the nvidia site, or re-installed with your package manager.

This mismatch probably happened when your kernel was last updated.
I think mint allows the nvidia module to be downloaded from its repositories, if so, uninstall the current one, then reinstall it, and you should get the correct version next time.

If you need to get it from the nvidia site, there are many HOWTOs on the net.

Let us know how you get on.

Thanks for that link smoker.

Thanks tradegar and thanks for the link smoker. It worked, though it was messy. First, when I rebooted, ubuntu again threatened me with low graphics mode but surprisingly the display came up with full 1440X900. It ran dmesg and found that some rivafb or nvidiafb or rivatv is trying to gatecrash. I found rivafb and deleted it. Then nouveau tried to butt in. I had to blacklist it. Rebooted and eureka!!!

Thanks for the tip, sir. I'll keep it in mind in future.

Thanks again folks. You guys rock

PS: By the way, the developers need to sort such issues out if Linux has to come out from geekdom to mass consumption

That's all very easy to say, but the issue is with a proprietary closed source driver. There's not a lot the developers can do without proper access to the driver code. I'm sure the nv driver and the new nouveau driver model will only improve over time, but they will always be playing catchup while the graphics card manufacturers keep their APIs secret.

Pleased you got it working. But I'd be interested to know if your VT's are working ( ctrl-alt-F1 to F6 ).

And, yes, it's more NVIDIA's fault than the kernel's. If they were more open, we could write better drivers.