Xorg segmentation fault with Nvidia (ver. 96) driver

[ker]

Binary package hint: xorg

Hi!
I can't use Nvidia (ver. 96) drivers anymore after a reboot. Choosing an old kernel from the grub list has no effect.
I've attached my Xorg.0.log.
Thanks

ProblemType: Bug
DistroRelease: Ubuntu 10.04
Package: xorg 1:7.5+5ubuntu1
ProcVersionSignature: Ubuntu 2.6.32-24.38-generic 2.6.32.15+drm33.5
Uname: Linux 2.6.32-24-generic i686
NonfreeKernelModules: nvidia
Architecture: i386
Date: Tue Jul 27 15:18:27 2010
DkmsStatus:
 virtualbox-ose, 3.1.6, 2.6.32-24-generic, i686: installed
 virtualbox-ose, 3.1.6, 2.6.32-23-generic, i686: installed
 nvidia-96, 96.43.17, 2.6.32-21-generic, i686: installed
 nvidia-96, 96.43.17, 2.6.32-24-generic, i686: installed
InstallationMedia: Ubuntu 10.04 LTS "Lucid Lynx" - Release i386 (20100429)
Lsusb:
 Bus 005 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 004 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 003 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 002 Device 001: ID 1d6b:0001 Linux Foundation 1.1 root hub
 Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
MachineType: MICRO-STAR INTERNATIONAL CO., LTD MS-7211
ProcCmdLine: BOOT_IMAGE=/boot/vmlinuz-2.6.32-24-generic root=UUID=e0808299-ff45-4a95-a4bc-e9ad06ba1bba ro quiet splash
ProcEnviron:
 LANG=it_IT.utf8
 SHELL=/bin/bash
SourcePackage: xorg
dmi.bios.date: 01/13/2006
dmi.bios.vendor: Phoenix Technologies, LTD
dmi.bios.version: 6.00 PG
dmi.board.name: MS-7211
dmi.board.vendor: MICRO-STAR INTERNATIONAL CO., LTD
dmi.board.version: 1.00
dmi.chassis.type: 3
dmi.modalias: dmi:bvnPhoenixTechnologies,LTD:bvr6.00PG:bd01/13/2006:svnMICRO-STARINTERNATIONALCO.,LTD:pnMS-7211:pvr1.00:rvnMICRO-STARINTERNATIONALCO.,LTD:rnMS-7211:rvr1.00:cvn:ct3:cvr:
dmi.product.name: MS-7211
dmi.product.version: 1.00
dmi.sys.vendor: MICRO-STAR INTERNATIONAL CO., LTD
system:
 distro: Ubuntu
 codename: lucid
 architecture: i686
 kernel: 2.6.32-24-generic

[james (james-nurealm)]

I'm also seeing this seg fault, particularly annoying because I would otherwise run Twinview. I've also tried the Nvidia packages

 NVIDIA-Linux-x86-96.43.16-pkg1.run
and
 NVIDIA-Linux-x86-96.43.18-pkg1.run

and also tried with the older kernel

 2.6.32-21-generic

 - no joy, except that, with the Nvidia 96.43.18 package, there is no seg fault, but instead, the log file ends with:

 ...
 (II) NVIDIA(0): Initialized GART.
 (II) NVIDIA(0): kernel module enabled successfully
 (II) NVIDIA(0): Memory mapped
 (EE) NVIDIA(0): Failed to allocate/map the primary surface!

That sort-of sounds like another driver is loaded, but no, there is no other driver loaded. Still, Xorg dmesg has this to say:

 [ 511.284620] Xorg:3820 conflicting memory types e8000000-e8d80000 write-combining<->uncached-minus
 [ 511.284626] reserve_memtype failed 0xe8000000-0xe8d80000, track write-combining, req write-combining

With 2.6.32-21 and 96.43.16, the log ends like this - with similarity to other combinations, "Segmentation fault at address 0xc":

...
(II) NVIDIA(0): Memory mapped
(II) NVIDIA(0): Using built-in logo image.

Backtrace:
0: /usr/bin/X (xorg_backtrace+0x3b) [0x80e937b]
1: /usr/bin/X (0x8048000+0x61c7d) [0x80a9c7d]
2: (vdso) (__kernel_rt_sigreturn+0x0) [0x7ad410]
Segmentation fault at address 0xc

Caught signal 11 (Segmentation fault). Server aborting

This is on an NV11 [GeForce2 MX/MX 400] dual display, with an AMD Sempron processor. I've also tried this same card and software on an older Athlon machine, but same results.

[james (james-nurealm)]

Uhm - well, there is a work-around for this problem. As has been pointed-out elsewhere, effectively, if /etc/default/grub is modified so that

 GRUB_TERMINAL=console

is un-commented, or instead, the kernel command line is modified to include "vga=normal", as in

 GRUB_CMDLINE_LINUX_DEFAULT=" <blah> vga=normal <blah> "

or, generally,

 GRUB_CMDLINE_LINUX_DEFAULT=" <blah> vga=[normal|ext|0x<some text mode>] <blah> "

then the framebuffer console is not used, and the nvidia module will run without a seg fault.

Something is still broken, with either Xorg or the nvidia module, but you can get on with life, just without running high-res on the framebuffer.

BTW, you cannot actually set the text-mode resolution with "vga=...". I seem to get 80x25 text mode regardless of what is set at "vga=0x???". I don't know if this is because of grub2 faults, or if intstead, these text modes have been disabled in the newer kernels. grub2 itself will fail for "vga=ask".

[dino99 (9d9)]

That version is no more supported; please open a new bug report if the actual archive found version also has the same issue.