[nvidia-glx-legacy] nvidia module won't load with 2.6.22-10 kernel (gutsy)
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
linux-restricted-modules-2.6.22 (Ubuntu) |
Invalid
|
Undecided
|
Unassigned |
Bug Description
Binary package hint: nvidia-glx-legacy
The nvidia module won't load with the 2.6.22-10 kernel. I just get an error message like "error inserting module" when typing 'modprobe nvidia'. The kernel version I'm using is linux-image-
Sitsofe Wheeler (sitsofe) wrote : | #1 |
Changed in linux-restricted-modules-2.6.22: | |
status: | New → Incomplete |
Stephan Bourduas (stephan-bourduas) wrote : | #2 |
$ dpkg -l *nvidia* | grep ii
ii nvidia-glx-legacy 1.0.7185+
ii nvidia-
I'm logged in right now without the nvidia driver. If I type 'modprobe nvidia', I get the following:
$ sudo modprobe -v nvidia
install /sbin/lrm-video nvidia
FATAL: Error running install command for nvidia
Stephan Bourduas (stephan-bourduas) wrote : | #3 |
Sitsofe Wheeler (sitsofe) wrote : | #4 |
Stephan:
Interestingly enough one of your graphics card appears to be a GeForce4 MX 440. Consequently you will be better off using the nvidia-glx (96xx) driver (NVIDIA won't support older drivers on that hardware). Does switching to nvidia-glx make things any better?
Additionally can you indicate whether you have ever manually installed binary drivers in the past or followed any guides on installing drivers?
Stephan Bourduas (stephan-bourduas) wrote : | #5 |
> Interestingly enough one of your graphics card appears to be a GeForce4 MX 440.
> Consequently you will be better off using the nvidia-glx (96xx) driver (NVIDIA won't support older drivers on that hardware).
The last time I checked the nvidia site, I read that the GeForce4 MX 440 needed the legacy drivers. Check out this page:
http://
However, I just went through the NVidia "wizard" to select my driver, and it took me here:
http://
Strangely, the supported products list (http://
> Does switching to nvidia-glx make things any better?
I'm not in a position to try it right now, but will do it later and get back to you.
> Additionally can you indicate whether you have ever manually installed binary drivers in the past or followed any guides on installing drivers?
My system is a brand new install from about 2 weeks ago. I haven't compiled any custom kernels or modules. I had originally installed Feisty and then upgraded to Gutsy, using the stock kernels and drivers.
Stephan Bourduas (stephan-bourduas) wrote : | #6 |
I installed nvidia-glx for the 2.6.22 kernel and I got the same error. The 'modprobe nvidia' just refuses to work. I downgraded it to 2.6.20.5-16, and it works properly. So as it stands, nvidia-glx and nvidia-glx-legacy don't seem to work for the 2.6.22 kernel and my GeForce 4 MX 440 card.
Sitsofe Wheeler (sitsofe) wrote : | #7 |
Stephan:
With regard to the phrase "legacy" there's "new legacy" (cards recently made legacy and roughly everything less than a Geforce 5000 but newer than a Geforce 2), and there's "original" legacy (cards that were made legacy a few years ago, roughly everything before a Geforce 2). "Helpfull"y NVIDIA call both sets of cards legacy despite the fact the last usable driver varies depending on whether it's "original legacy" or "new legacy". See http://
Thanks for the update. Can you post the output of:
sudo sh -x /sbin/lrm-video nvidia
?
Sitsofe Wheeler (sitsofe) wrote : | #8 |
Additionally can you post the output of
uname -a
dpkg -l linux\* | grep ii | tr -s " " | cut -d" " -f2-3
?
Stephan Bourduas (stephan-bourduas) wrote : | #9 |
> Thanks for the update. Can you post the output of:
> sudo sh -x /sbin/lrm-video nvidia
$ sudo sh -x /sbin/lrm-video nvidia
+ PATH=/sbin:/bin
+ MODULE=nvidia
+ shift
+ [ nvidia = nvidia ]
+ [ -e /lib/linux-
+ [ -e /lib/linux-
+ XORG=nvidia
+ cat /etc/X11/xorg.conf
+ sed -n -e /^[ \t]*section[ \\t]*"device"/I,/^[ \t]*endsection/
+ grep -q -w nvidia
+ modprobe --ignore-install -Qb nvidia
> Additionally can you post the output of
> uname -a
> dpkg -l linux\* | grep ii | tr -s " " | cut -d" " -f2-3
$ uname -a
Linux atlas 2.6.22-10-386 #1 Wed Aug 22 07:43:24 GMT 2007 i686 GNU/Linux
$ dpkg -l linux\* | grep ii | tr -s " " | cut -d" " -f2-3
linux-generic 2.6.22.10.11
linux-headers-
linux-headers-
linux-headers-
linux-headers-
linux-headers-
linux-image-
linux-image-
linux-image-
linux-image-
linux-image-
linux-image-
linux-image-386 2.6.22.10.11
linux-image-generic 2.6.22.10.11
linux-kernel-devel 2.6.22-10.30
linux-libc-dev 2.6.22-10.30
linux-restricte
linux-restricte
linux-restricte
linux-restricte
linux-restricte
linux-restricte
linux-sound-base 1.0.14-1ubuntu1
linux-source 2.6.22.10.11
linux-source-2.6.22 2.6.22-10.30
linux-ubuntu-
linux-ubuntu-
Sitsofe Wheeler (sitsofe) wrote : | #10 |
Stephan:
Can you boot the 2.6.22-generic kernel and check whether
sudo modprobe nvidia
complains. If it does can you run
sudo depmod
sudo sh -x /sbin/lrm-video nvidia
dmesg | tail
?
Stephan Bourduas (stephan-bourduas) wrote : | #11 |
$ uname -a
Linux atlas 2.6.22-10-generic #1 SMP Wed Aug 22 08:11:52 GMT 2007 i686 GNU/Linux
$ sudo modprobe nvidia
FATAL: Error running install command for nvidia
$ sudo depmod
$ sudo sh -x /sbin/lrm-video nvidia
+ PATH=/sbin:/bin
+ MODULE=nvidia
+ shift
+ [ nvidia = nvidia ]
+ [ -e /lib/linux-
+ [ -e /lib/linux-
+ XORG=nvidia
+ cat /etc/X11/xorg.conf
+ sed -n -e /^[ \t]*section[ \\t]*"device"/I,/^[ \t]*endsection/
+ grep -q -w nvidia
+ modprobe --ignore-install -Qb nvidia
$ dmesg | tail
[ 126.405411] Bluetooth: L2CAP ver 2.8
[ 126.405423] Bluetooth: L2CAP socket layer initialized
[ 126.419087] Bluetooth: RFCOMM socket layer initialized
[ 126.419289] Bluetooth: RFCOMM TTY layer initialized
[ 126.419296] Bluetooth: RFCOMM ver 1.8
[ 316.515968] nvidia: module license 'NVIDIA' taints kernel.
[ 316.790795] PCI: Enabling device 0000:01:00.0 (0004 -> 0006)
[ 316.791659] ACPI: PCI Interrupt Link [LNKA] enabled at IRQ 11
[ 316.791672] ACPI: PCI Interrupt 0000:01:00.0[A] -> Link [LNKA] -> GSI 11 (level, low) -> IRQ 11
[ 316.792956] NVRM: loading NVIDIA Linux x86 Kernel Module 1.0-9639 Mon Apr 16 20:20:06 PDT 2007
So it seems to be loading the module properly for the generic kernel.
Stephan Bourduas (stephan-bourduas) wrote : | #12 |
I just repeated the above with the 386 kernel, and got the same results:
$ uname -a
Linux atlas 2.6.22-10-386 #1 Wed Aug 22 07:43:24 GMT 2007 i686 GNU/Linux
$ sudo modprobe nvidia
FATAL: Error running install command for nvidia
$ sudo depmod
$ sudo sh -x /sbin/lrm-video nvidia
+ PATH=/sbin:/bin
+ MODULE=nvidia
+ shift
+ [ nvidia = nvidia ]
+ [ -e /lib/linux-
+ [ -e /lib/linux-
+ XORG=nvidia
+ cat /etc/X11/xorg.conf
+ sed -n -e /^[ \t]*section[ \\t]*"device"/I,/^[ \t]*endsection/
+ grep -q -w nvidia
+ modprobe --ignore-install -Qb nvidia
$ dmesg | tail
[ 114.725260] Bluetooth: L2CAP ver 2.8
[ 114.725274] Bluetooth: L2CAP socket layer initialized
[ 114.751551] Bluetooth: RFCOMM socket layer initialized
[ 114.751988] Bluetooth: RFCOMM TTY layer initialized
[ 114.751996] Bluetooth: RFCOMM ver 1.8
[ 300.255920] nvidia: module license 'NVIDIA' taints kernel.
[ 300.529987] PCI: Enabling device 0000:01:00.0 (0004 -> 0006)
[ 300.530806] ACPI: PCI Interrupt Link [LNKA] enabled at IRQ 11
[ 300.530817] ACPI: PCI Interrupt 0000:01:00.0[A] -> Link [LNKA] -> GSI 11 (level, low) -> IRQ 11
[ 300.531991] NVRM: loading NVIDIA Linux x86 Kernel Module 1.0-9639 Mon Apr 16 20:20:06 PDT 2007
Is this a bug with the driver installation procedure?
Sitsofe Wheeler (sitsofe) wrote : | #13 |
Stephen:
Can you indicate whether
sudo modprobe nvidia
just starts complaining in the generic kernel when you reboot? (don't run depmod first)
Stephan Bourduas (stephan-bourduas) wrote : | #14 |
When I first booted into the 2.6.22 generic kernel, I tried "modprobe nvidia" and had the same error. Once I ran depmod, I was able to load the nvidia module and restart X properly.
Sitsofe Wheeler (sitsofe) wrote : | #15 |
Stephen:
Just so I'm clear, you are saying every single time you boot your computer into the same kernel you will be unable to do
sudo modprobe nvidia
without it failing until you use depmod first?
Stephan Bourduas (stephan-bourduas) wrote : | #16 |
I just finished upgrading my kernel to 2.6.22.11.12 and I could not get nvidia to load no matter what I did. I tried depmod like before, and it didn't work. I then did:
$ sudo apt-get install --reinstall nvidia-glx
$ sudo modprobe nvidia
to no avail. Just to be sure, I tried depmod again, which resulted in the same error as before.
I then booted into the 2.6.22.10 kernel and the nvidia module loaded fine, without the need for depmod:
$ uname -a
Linux atlas 2.6.22-10-386 #1 Wed Aug 22 07:43:24 GMT 2007 i686 GNU/Linux
$ lsmod | grep nv
nvidia 4714036 22
i2c_core 25104 1 nvidia
agpgart 33584 2 nvidia,intel_agp
$ dmesg | grep -i nvidia
[ 53.241245] nvidia: module license 'NVIDIA' taints kernel.
[ 53.524608] NVRM: loading NVIDIA Linux x86 Kernel Module 1.0-9639 Mon Apr 16 20:20:06 PDT 2007
$ dpkg -l nvidia-glx | grep ^ii
ii nvidia-glx 1:1.0.9639+
So it seems that the nvidia-glx works properly for the -10 kernel, but not the -11.
Sitsofe Wheeler (sitsofe) wrote : | #17 |
Can you post the output of
dpkg -l linux-restricte
dpkg -l linux-image\* | grep ii | tr -s " " | cut -d" " -f2-3
sudo sh -x /sbin/lrm-video nvidia
sudo modprobe --ignore-install nvidia
grep linux-restricted /var/log/dpkg.log*
?
jdm64 (jdm64) wrote : | #18 |
I seem to be having the same problem. I reported this as:
bug #138933: [Gusty] Broken Nvidia driver after upgrade to Gusty (7.10)
What might be the problem is a hidden file in /lib/linux-
sudo rm /lib/linux-
Also see these bug reports for more information:
Bug #106217: Hidden file does not get removed when switching from nvidia-
Bug #130799: Gutsy nvidia-glx no gpu or API mismatch
Stephan Bourduas (stephan-bourduas) wrote : | #19 |
- linux-restricted.log Edit (18.9 KiB, text/plain)
Sorry about the late reply. I don't have the time at the moment to mess around with my machine. I'm currently running the working 2.6.22.10 kernel:
$ uname -a
Linux atlas 2.6.22-10-386 #1 Wed Aug 22 07:43:24 GMT 2007 i686 GNU/Linux
$ dpkg -l linux-restricte
linux-restricte
linux-restricte
linux-restricte
linux-restricte
linux-restricte
$ dpkg -l linux-image\* | grep ii | tr -s " " | cut -d" " -f2-3
linux-image-
linux-image-
linux-image-
linux-image-
linux-image-
linux-image-
linux-image-
linux-image-386 2.6.22.11.12
$ grep linux-restricted /var/log/dpkg.log* > /tmp/linux-
I'll post the modprobe stuff when I reboot into the 2.6.22-11.12 kernel. Maybe the problem will be fixed by now.
> What might be the problem is a hidden file in /lib/linux-
> sudo rm /lib/linux-
jmd64: I have no such files on my system:
$ ls -a1 /lib/linux-
.
..
2.6.20-15-386
2.6.20-16-386
2.6.22-10-386
2.6.22-10-generic
Timo Aaltonen (tjaalton) wrote : | #20 |
Does it work if you purge the -386 kernel and only leave -generic of the latest version?
Michael Lustfield (michaellustfield) wrote : | #21 |
We are closing this bug report because it lacks the information we need to investigate the problem, as described in the previous comments. Please reopen it if you can give us the missing information, and don't hesitate to submit bug reports in the future. To reopen the bug report you can click on the current status, under the Status column, and change the Status back to "New".
If you can confirm that bug bug has been resolved in Intrepid, please let us know if we can set the status to "Fix Released"
Thanks again!
Changed in linux-restricted-modules-2.6.22: | |
status: | Incomplete → Invalid |
Thank you for your bug report.
Stephan:
Could you attach your
/var/log/Xorg.0.log
file from a failed run along with the output of
dmesg > dmesg.txt
(which will make a file called dmesg.txt) and
dpkg -l *nvidia* | grep ii
?