Bug#798207: nvidia-legacy-340xx-driver: Wrong module loaded, X fails to start when legacy installed alongside mainline driver

Luca Boccassi luca.boccassi at gmail.com
Fri Sep 18 22:05:28 UTC 2015


On Fri, 2015-09-18 at 01:55 +0200, Andreas Beckmann wrote:
> On 2015-09-18 00:54, Luca Boccassi wrote:
> > I am really out of ideas! :-(
> 
> so just that I got it right, you configured the nvidia alternative to
> use nvidia-current and not nvidia-legacy-340xx
> 
> upon boot the 340.76 (nvidia-legacy-340xx) gets loaded by unknown
means
> 
> 
> let's start with manual (un-)loading
> 
> rmmod nvidia-uvm nvidia
> modprobe -v nvidia

insmod /lib/modules/4.1.0-2-amd64/updates/dkms/nvidia-current.ko NVreg_DeviceFileUID=0 NVreg_DeviceFileGID=44 NVreg_DeviceFileMode=0660
install modprobe -i nvidia-current $CMDLINE_OPTS NVreg_DeviceFileUID=0 NVreg_DeviceFileGID=44 NVreg_DeviceFileMode=0660 

> rmmod nvidia-uvm nvidia
> modprobe -v nvidia-uvm

modprobe: FATAL: Module nvidia-current-uvm not found.
modprobe: ERROR: ../libkmod/libkmod-module.c:960 command_do() Error running install command for nvidia_uvm
modprobe: ERROR: could not insert 'nvidia_uvm': Operation not permitted
install modprobe nvidia ; modprobe -i nvidia-current-uvm $CMDLINE_OPTS 

This is very weird, the module is there. I see a permission error, but
it was all ran as root.

> rmmod nvidia-uvm nvidia
> modprobe -v char-major-195-0

insmod /lib/modules/4.1.0-2-amd64/updates/dkms/nvidia-current.ko NVreg_DeviceFileUID=0 NVreg_DeviceFileGID=44 NVreg_DeviceFileMode=0660
install modprobe -i nvidia-current $CMDLINE_OPTS NVreg_DeviceFileUID=0 NVreg_DeviceFileGID=44 NVreg_DeviceFileMode=0660 

> rmmod nvidia-uvm nvidia
> 
> nvidia-modprobe
> cat /proc/driver/nvidia/version

NVRM version: NVIDIA UNIX x86_64 Kernel Module  340.93  Wed Aug 19 16:49:15 PDT 2015
GCC version:  gcc version 4.9.2 (Debian 4.9.2-10) 

> rmmod nvidia-uvm nvidia
> 
> nvidia-modprobe -u

modprobe: FATAL: Module nvidia-current-uvm not found.
modprobe: ERROR: ../libkmod/libkmod-module.c:960 command_do() Error running install command for nvidia_uvm
modprobe: ERROR: could not insert 'nvidia_uvm': Operation not permitted

> cat /proc/driver/nvidia/version

NVRM version: NVIDIA UNIX x86_64 Kernel Module  340.93  Wed Aug 19
16:49:15 PDT 2015
GCC version:  gcc version 4.9.2 (Debian 4.9.2-10) 

> rmmod nvidia-uvm nvidia
> 
> 
> If no module is loaded and you have a manual xorg configuration that
> enables nvidia, what happens if you start X? The module should get
> loaded - the correct one? uvm, too ?

The correct module is loaded. UVM is not, as expected.

> In the next round you could try to add this to
> /etc/nvidia/nvidia-modprobe.conf:
> 
> alias	pci:v000010DEd00000E00sv*sd*bc04sc80i00*	nvidia
> alias	pci:v000010DEd00000AA3sv*sd*bc0Bsc40i00*	nvidia
> alias	pci:v000010DEd*sv*sd*bc03sc02i00*		nvidia
> alias	pci:v000010DEd*sv*sd*bc03sc00i00*		nvidia

This works and the right module is loaded, but then Gnome crashes
because of the error below :-( Still, progress is progress.

If I don't have the legagy-340xx the 340.93 driver works fine of course.
From the reportbug --template (same as my previous email), and from
looking around the filesystem, all the symlinks point to the right
libraries as far as I can see. This is the error in the journal:

Sep 18 22:13:45 luca-desktop /usr/bin/dbus-launch[1882]: gnome-session-check-accelerated: Helper exited with code 256
Sep 18 22:13:45 luca-desktop /usr/bin/dbus-launch[1882]: Current serial number in output stream:  26
Sep 18 22:13:45 luca-desktop /usr/bin/dbus-launch[1882]: Serial number of failed request:  25
Sep 18 22:13:45 luca-desktop /usr/bin/dbus-launch[1882]: Value in failed request:  0x0
Sep 18 22:13:45 luca-desktop /usr/bin/dbus-launch[1882]: Minor opcode of failed request:  3 (X_GLXCreateContext)
Sep 18 22:13:45 luca-desktop /usr/bin/dbus-launch[1882]: Major opcode of failed request:  155 (GLX)
Sep 18 22:13:45 luca-desktop /usr/bin/dbus-launch[1882]: X Error of failed request:  BadValue (integer parameter out of range for operation)

Weird thing is that then killing gdm and doing startx manually works
fine.

Kind regards,
Luca Boccassi

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: This is a digitally signed message part
URL: <http://lists.alioth.debian.org/pipermail/pkg-nvidia-devel/attachments/20150918/0962713a/attachment.sig>


More information about the pkg-nvidia-devel mailing list