FATAL: Error inserting nvidia (...): No such device

[paolopat]

I don't know if it's an openSUSE specific issue.

My system is a CLEVO CO. W150HRM, output device is set to DFP-0 and XV. I run openSUSE 11.4.

To enable/disable the graphic card I use the bumblebee-XXXable.template.opensuse scripts, with the following poweron/poweroff lines:

enable: echo $(acpi_call "_SB.PCI0.PEG0.PEGP._ON")
disable: echo $(acpi_call "_SB.PCI0.PEG0.PEGP._OFF")

Disable script works fine. Enable script turns on the card, but "modprobe nvidia" gives the following output.

FATAL: Error inserting nvidia (/lib/modules/2.6.37.6-0.5-desktop/weak-updates/updates/nvidia.ko): No such device

Is it something connected with the xorg.nvidia.conf file? Or a kernel issue?

Thanks.

[Ximi1970]

Can you try this for me ?
Login as root

modprobe nvidia
tail -50 /var/log/messages

Check the messages for nvidia errors. Probably you need to append the "vmalloc=512M" option to your kernel line in /boot/grub/menu.lst

[paolopat]

Here is the /var/log/mesages after each step:

After disabling card:
Jun 23 19:30:01 avernux kernel: [ 121.213747] nvidia 0000:01:00.0: PCI INT A disabled
Jun 23 19:30:01 avernux kernel: [ 121.221884] acpi_call: Module loaded successfully
Jun 23 19:30:01 avernux kernel: [ 121.225605] acpi_call: Calling _SB.PCI0.PEG0.PEGP._OFF
Jun 23 19:30:03 avernux kernel: [ 122.727683] acpi_call: Call successful: 0x0

After enabling card (without modprobe nvidia):
Jun 23 19:32:34 avernux kernel: [ 274.033208] acpi_call: Calling _SB.PCI0.PEG0.PEGP._ON
Jun 23 19:32:36 avernux kernel: [ 276.070333] acpi_call: Call successful: 0x0

After modprobe nvidia:
Jun 23 19:33:52 avernux kernel: [ 352.229110] nvidia 0000:01:00.0: power state changed by ACPI to D0
Jun 23 19:33:52 avernux kernel: [ 352.229123] nvidia 0000:01:00.0: power state changed by ACPI to D0
Jun 23 19:33:52 avernux kernel: [ 352.229136] nvidia 0000:01:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
Jun 23 19:33:52 avernux kernel: [ 352.229151] nvidia 0000:01:00.0: setting latency timer to 64
Jun 23 19:33:52 avernux kernel: [ 352.229158] vgaarb: device changed decodes: PCI:0000:01:00.0,olddecodes=none,decodes=none:owns=none
Jun 23 19:33:52 avernux kernel: [ 352.229578] NVRM: The NVIDIA GPU 0000:01:00.0 (PCI ID: 10de:0dce) installed
Jun 23 19:33:52 avernux kernel: [ 352.229580] NVRM: in this system is not supported by the 270.41.06 NVIDIA Linux
Jun 23 19:33:52 avernux kernel: [ 352.229582] NVRM: graphics driver release. Please see 'Appendix A -
Jun 23 19:33:52 avernux kernel: [ 352.229583] NVRM: Supported NVIDIA GPU Products' in this release's README,
Jun 23 19:33:52 avernux kernel: [ 352.229584] NVRM: available on the Linux graphics driver download page at
Jun 23 19:33:52 avernux kernel: [ 352.229585] NVRM: www.nvidia.com.
Jun 23 19:33:52 avernux kernel: [ 352.229596] nvidia 0000:01:00.0: PCI INT A disabled
Jun 23 19:33:52 avernux kernel: [ 352.229607] nvidia: probe of 0000:01:00.0 failed with error -1
Jun 23 19:33:52 avernux kernel: [ 352.229661] NVRM: The NVIDIA probe routine failed for 1 device(s).
Jun 23 19:33:52 avernux kernel: [ 352.229663] NVRM: None of the NVIDIA graphics adapters were initialized!

Adding "vmalloc=512M" option to grub doesn't seem to change anything.

I don't know if it's a useful information, but during the boot I get this error:
[drm:intel_dsm_platform_mux_info] ERROR MUX INFO call failed

[Ximi1970]

Seems like your card is still not supported by the 270.41.06 nVidia driver. You need the latest and the greatest from nVidia.
Uninstall bumblebee first: sudo bumblebee-uninstall.
Then you need to uninstall the nVidia repository packages with yast->Software management->View Repositories-> Delete all selected packages in this list (normally nvidia-computeG02, x11-video-nvidiaG02 and nvidia-gfxG02-kmp-desktop should be deleted). Press Accept.
Remove or disable the nVidia repository: yast->Software Repositories->Select nVidia Graphics Drivers and delete this entry. DO NOT DELETE ANYTHING ELSE !!

Download the latest driver 275.09.07 from here (nVidia Linux forum):
http://www.nvnews.net/vbulletin/showthread.php?s=1c4170cd020a1e8b461b6d168fa1cfda&t=122606

There is a GT 555M in the supported list:
ftp://download.nvidia.com/XFree86/Linux-x86/275.09.07/README/supportedchips.html

You need to switch the nVidia card on first !
You can check the ID of the chip with : lspci | grep nVidia

If you are in luck or if you just wanna try, install the latest driver with: bash NVIDIA-Linux-xxxxxxxx.run
Probably you will need to install the kernel development pattern in the Software management.

Now you can pull the latest tree and install bumblebee again but without installing the nvidia repository or drivers !

Small warning:
Please be careful when switching between nVidia repo and the nVidia install script. Make sure you uninstall everything when you switch.
Use yast to uninstall the rpm's from the repo and
bash NVIDIA-Linux-xxxx -uninstall to remove the nVidia script.

Hopefully this will work,

Regards,

Maxime

[paolopat]

Nope... It has the same behavior for what concerns the enable/disable problem, but it's worse because I couldn't start glxgears with both cards.

[Ximi1970]

If you get the same error from the nvidia card:
NVRM: The NVIDIA GPU 0000:01:00.0 (PCI ID: 10de:0dce) installed
NVRM: in this system is not supported by the 275.09.07 NVIDIA Linux
NVRM: graphics driver release. Please see 'Appendix A -

Then your card is not supported yet by nVidia. The ID of your card (0x0DCE) is not in the list. So you have to wait for a driver update :-( Keep in mind that the list is NOT complete, so keep trying newer drivers.

You can uninstall the nvidia script with bash NVIDIA-Linux-275.xxxxxx.run --uninstall
The Intel should be running fine after that.

[paolopat]

Ok.
It's strange, because it's a GT 555M... Could you tell me where you checked the support for 0x0DCE? I'll keep an eye on it.
In any case, all the rest works, so thank you!

[Ximi1970]

You can find the latest drivers here:
http://www.nvnews.net/vbulletin/showthread.php?s=1c4170cd020a1e8b461b6d168fa1cfda&t=122606
Sometimes you can also download special beta drivers.

If you click on the drivers version number, a new page will open. Here you can find a link to the README file. Just click on the x86 or x86_64 number. Goto section Appendix A, Supported products
If you search this file for the 555M you will see that only a version with ID 0x0DD6 is supported.
You got another maybe newer version of the chip.

This list is not complete, for example the 540M is missing. But the driver works fine in the ASUS N53SV.

[paolopat]

Hi!
I have news for you.
Even if in that list my specific device code is not present, at boot the nvidia module is loaded, as I can find it with the lsmod command.
Commenting out the "modprobe -r nvidia" section in the disable script, I succeeded in disabling/enabling card, as the nvidia module is never unloaded (and so no error is possible when trying to load it again).
I have to check if the battery stat changes but, on a first sight, after disabling the card bumblebee is not able to run optirun64 glxgears.

EDIT: the problem now is that a second run of optirun64 makes my screen black and seems to hang the system...

[Ximi1970]

When I disable the loading and unloading of the nvidia module, I get the same results. But I did another experiment.
I unloaded the nvidia module and disabled the card. As soon as I tried to load the nvidia module again without enable the card, I got the same errors as you had "Chip is not supported".
It looks like you cannot enable your card again. How did you find your enable / disable commands ?
Did you follow the instructions on:

http://linux-hybrid-graphics.blogspot.com/2010/07/using-acpicall-module-to-switch-onoff.html

Search for: Using acpi_call module to switch on/off discrete graphics card in Linux

[Ximi1970]

My card also stays visible when disabled.
I will dump my lspci, please wait

[paolopat]

Yes, I used that script.
when I disable nvidia, "lspci -vnnn | grep VGA" gives:
00:02.0 VGA compatible controller [0300]: Intel Corporation 2nd Generation Core Processor Family Integrated Graphics Controller 8086:0126 (prog-if 00 [VGA controller])
01:00.0 VGA compatible controller [0300]: nVidia Corporation Device 10de:0dce (prog-if ff)
when I enable it, the result is:
00:02.0 VGA compatible controller [0300]: Intel Corporation 2nd Generation Core Processor Family Integrated Graphics Controller 8086:0126 (prog-if 00 [VGA controller])
01:00.0 VGA compatible controller [0300]: nVidia Corporation Device 10de:0dce (prog-if 00 [VGA controller])

[Ximi1970]

Got the same results. A friend has simular problems enabling the card again. He has to go into the BIOS setup and do a restore to defaults to get the card working again.

disabled:

00:02.0 VGA compatible controller [0300]: Intel Corporation 2nd Generation Core Processor Family Integrated Graphics Controller [8086:0116](rev 09) (prog-if 00 [VGA controller])
01:00.0 VGA compatible controller [0300]: nVidia Corporation Device [10de:0df4](rev ff) (prog-if ff)

enabled:
00:02.0 VGA compatible controller [0300]: Intel Corporation 2nd Generation Core Processor Family Integrated Graphics Controller [8086:0116](rev 09) (prog-if 00 [VGA controller])
01:00.0 VGA compatible controller [0300]: nVidia Corporation Device [10de:0df4](rev a1) (prog-if 00 [VGA controller])

[paolopat]

Ok, so the problem is that the nvidia should be completely turned off and should not appear in the lspci, right?
I'll try to look around to find a solution.