Unrecoverable Kernel Oops when launching X on 9400M (ION)

[antjoh]

Hello,

I have a Zotac IONITX-B mother board with the Nvidia GeForce 9400M chipset. Whenever I try to launch X with the nvidia module it results in a kernel Oops and X is unusable.

As soon as I try to load X (xorg-server 1.6.2) the pc crashes with a kernel Oops. The pc is equipped with 4Gb of DDR2 800Mhz RAM. No other perhiperals are used except for one wifi usb dongle and a usb keyboard. The pc is hooked up to a 40" lcd tv via HDMI. The distro of choice is Arch Linux.

This is what I have done so far to try pin pointing the problem:

* Installed the nvidia package from the ArchLinux repositories = results in kernel Oops when trying to run X.
* unistalled the ArchLinux nvidia package and installed the driver available on the nvidia website (the .run file) = results in kernel Oops when trying to run X.
* Removed one stick of ram and tested each stick individually on both slots on the motherboard = results in kernel Oops on all tries when trying to run X.
* Installed xf86-video-vesa from the ArchLinux repositories and used that module in xorg.conf = results in no kernel Oops and a well working X environment.

My xorg.conf is very minimal as it depends on HAL to detect what needs to be loaded.

Do you have any idea what might be causing this behaviour?

I tried to run the nvidia-bug-report script but it hangs and probably doesn't provide a full report.
Hopefully you can make something out of the logs I have attached and provide me with a solution to this problem.

Many thanks!
Best regards
Anton Johanson

[zander]

The Oops reads:
BUG: unable to handle kernel paging request at 000d8a0b
IP: [<c016c19c>] m_show+0x9c/0x1a0
*pde = 00000000
Oops: 0000 [#1] PREEMPT SMP
last sysfs file: /sys/devices/pci0000:00/0000:00:10.0/0000:03:00.0/resource
Modules linked in: ipt_REJECT xt_tcpudp nf_conntrack_ipv4 nf_defrag_ipv4 xt_state \
nf_conntrack iptable_filter ip_tables x_tables snd_hda_codec_nvhdmi arc4 ecb \
snd_hda_codec_realtek rt73usb crc_itu_t rt2x00usb rt2x00lib led_class \
input_polldev joydev snd_seq_dummy snd_hda_intel mac80211 snd_seq_oss usbhid \
snd_seq_midi_event hid snd_hda_codec snd_seq snd_seq_device cfg80211 nvidia(P) \
snd_pcm_oss snd_mixer_oss snd_hwdep snd_pcm snd_timer snd agpgart ohci_hcd \
soundcore psmouse i2c_nforce2 shpchp ehci_hcd pcspkr snd_page_alloc serio_raw sg \
pci_hotplug usbcore wmi i2c_core forcedeth evdev thermal processor fan button \
battery ac rtc_cmos rtc_core rtc_lib jfs sd_mod ahci ata_generic pata_acpi libata \
scsi_mod
Pid: 1971, comm: X Tainted: P (2.6.30-ARCH #1) To Be Filled By O.E.M.
EIP: 0060:[<c016c19c>] EFLAGS: 00210282 CPU: 0
EIP is at m_show+0x9c/0x1a0
EAX: 00000000 EBX: f6745a80 ECX: 00000f59 EDX: 00000000
ESI: f9dc8740 EDI: 000d8a0b EBP: f9dc8870 ESP: f5e25eb4
DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
Process X (pid: 1971, ti=f5e24000 task=f5e20400 task.ti=f5e24000)
Stack:
f6745a80 c0458a98 fa04058c 00011884 f9dc8744 f5e25f30 c03d3398 647d5c20
c03e1768 f6745a80 f9dc8744 f5e25f30 c01eea13 f5d9c9f8 c01b3e41 0000007c
00000000 b7f8a022 f6745aa8 000003de 00000022 f5ca9d80 f5e25f90 0000001b
Call Trace:
[<c03d3398>] ? mutex_lock+0x18/0x40
[<c01eea13>] ? seq_read+0x263/0x470
[<c01b3e41>] ? __do_fault+0x351/0x430
[<c01ee7b0>] ? seq_read+0x0/0x470
[<c0219e89>] ? proc_reg_read+0x79/0xc0
[<c01d1b43>] ? vfs_read+0xc3/0x1a0
[<c0219e10>] ? proc_reg_read+0x0/0xc0
[<c01d1d28>] ? sys_read+0x58/0xb0
[<c0103c73>] ? sysenter_do_call+0x12/0x28
Code: 10 31 c0 81 c5 2c 01 00 00 39 ef 74 2d 66 90 8b 47 08 89 1c 24 c7 44 24 \
04 98 8a 45 c0 83 c0 0c 89 44 24 08 e8 06 24 08 00 8b 3f <8b> 07 0f 18 00 90 \
39 ef 75 da b8 01 00 00 00 8b 96 d4 00 00 00
EIP: [<c016c19c>] m_show+0x9c/0x1a0 SS:ESP 0068:f5e25eb4
CR2: 00000000000d8a0b
---[ end trace 9da229b657f2dd53 ]---
There's no indication that the NVIDIA kernel module is involved. I briefly looked around for similar traces and found a variety of reports with different modules. This looks like a kernel bug to me.

[antjoh]

I don't know much about these kind of errors, but how do you determine that the nvidia module isn't involved? I can see it listed as a linked in module, but maybe that is not enough?

The thing I find odd is that X starts ok without the nvidia module but crashes as soon as it is loaded.

Edit: Thanks for a really fast reply btw... (y)

[zander]

nvidia.ko is already linked and from the look of it, the kernel crashes when X reads /proc/modules; there are no NVIDIA symbols in the backtrace.

[antjoh]

Could a moderator please move this post to the general linux forum?
I really don't know how to diagnose this problem and would appreciate some help. Maybe someone over there knows what to do.

[gammalam]

Follow the instruction to install the driver if u are using fedora ^^
http://forums.fedoraforum.org/showthread.php?t=204752&highlight=nvidia

I think is the Selinux problem. I had this problem as well