You are not logged in.

#1 2009-11-20 23:23:47

kleptophobiac
Member
From: Sunnyvale, CA
Registered: 2004-04-25
Posts: 488

[SOLVED] Loading module i915 causes kernel OOPS

Update:
Fix this problem by removing the Nouveau packages nouevea-drm and xf86-video-nouveau. Thanks to rdmelin for the solution!

Original post:

I'm trying to get KMS/Xorg working on my Intel Q35 based computer, but am not having the best of luck. I can modprobe intel_agp and drm just fine, but when it comes to i915, the kernel isn't terribly happy to oblige.

Does anybody know how to even begin trying to figure out why this happens? I'm assuming that the i915 module is sufficiently popular that if there were something terribly wrong with it that people would be crying all over the forum. Has anybody used it with the Q35 chipset before? Maybe it's a bug limited to a thin slice of hardware? Maybe it's a BIOS setting? I honestly have no idea.

Any direction would be very much appreciated!

Here's the output in the system log file after I load the module:

Nov 20 18:13:59 hydrogen kernel: Modules linked in: i915(+) drm_kms_helper video output drm ipv6 ipt_MASQUERADE iptable_nat nf_nat xt_tcpudp nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack iptable_filter ip_tables x_tables ext2 mbcache dm_mod raid456 raid6_pq async_xor async_memcpy async_tx xor md_mod msp3400 saa7134_alsa saa7115 saa7134_dvb lnbp21 wm8775 tda826x mt352 tuner tea5767 tda8290 tda18271 xc5000 tda9887 mt20xx tea5761 tuner_xc2028 zl10036 tda10086 dvb_pll videobuf_dvb cx25840 tda10048 lgdt3305 dvb_core nxt200x isl6421 tuner_simple tuner_types isl6405 mt312 zl10353 tda1004x tda827x ivtv snd_hda_codec_realtek snd_seq_dummy snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device saa7134 i2c_algo_bit cx2341x ir_common snd_hda_intel snd_hda_codec v4l2_common videobuf_dma_sg videodev videobuf_core snd_pcm_oss snd_hwdep snd_mixer_oss snd_pcm snd_timer intel_agp snd uhci_hcd e1000e i2c_i801 v4l1_compat tveeprom ehci_hcd agpgart soundcore i2c_core usbcore fan thermal iTCO_wdt iTCO_vendor_support snd_page_alloc sg psmouse evdev battery serio_raw ac pcspkr container processor button rtc_cmos rtc_core rtc_lib reiserfs sr_mod cdrom sd_mod pata_it8213 pata_acpi ata_generic ata_piix libata scsi_mod
Nov 20 18:13:59 hydrogen kernel: 
Nov 20 18:13:59 hydrogen kernel: Pid: 1808, comm: modprobe Not tainted (2.6.31-ARCH #1) empty
Nov 20 18:13:59 hydrogen kernel: EIP: 0060:[<fa0b4276>] EFLAGS: 00010286 CPU: 3
Nov 20 18:13:59 hydrogen kernel: EIP is at i915_driver_load+0x1b6/0x1050 [i915]
Nov 20 18:13:59 hydrogen kernel: EAX: 00000000 EBX: f6847800 ECX: f6612be8 EDX: 00000000
Nov 20 18:13:59 hydrogen kernel: ESI: f4337000 EDI: 00001c50 EBP: 00000001 ESP: f6485ddc
Nov 20 18:13:59 hydrogen kernel: DS: 007b ES: 007b FS: 00d8 GS: 00e0 SS: 0068
Nov 20 18:13:59 hydrogen kernel: fa026f5d fa026654 fa026f70 00000000 00000001 0000003f f6847800 f6847a7c
Nov 20 18:13:59 hydrogen kernel: <0> 00000000 f2a636ad f6847800 f7092800 00000000 fa0ea260 fa01a2bb 00000001
Nov 20 18:13:59 hydrogen kernel: <0> fa026f5d fa026638 fa026f61 f70deb70 f64500c0 00000000 c1151e1c 00000001
Nov 20 18:13:59 hydrogen kernel: [<fa01a2bb>] ? drm_get_dev+0x2cb/0x500 [drm]
Nov 20 18:13:59 hydrogen kernel: [<c1151e1c>] ? sysfs_do_create_link+0xbc/0x160
Nov 20 18:13:59 hydrogen kernel: [<c11b138a>] ? local_pci_probe+0x1a/0x40
Nov 20 18:13:59 hydrogen kernel: [<c11b2641>] ? pci_device_probe+0x81/0xb0
Nov 20 18:13:59 hydrogen kernel: [<c1234869>] ? driver_probe_device+0x89/0x170
Nov 20 18:13:59 hydrogen kernel: [<c12349e1>] ? __driver_attach+0x91/0xa0
Nov 20 18:13:59 hydrogen kernel: [<c1234950>] ? __driver_attach+0x0/0xa0
Nov 20 18:13:59 hydrogen kernel: [<c1233ee2>] ? bus_for_each_dev+0x62/0xa0
Nov 20 18:13:59 hydrogen kernel: [<c1234692>] ? driver_attach+0x22/0x40
Nov 20 18:13:59 hydrogen kernel: [<c1234950>] ? __driver_attach+0x0/0xa0
Nov 20 18:13:59 hydrogen kernel: [<c123366e>] ? bus_add_driver+0xce/0x2b0
Nov 20 18:13:59 hydrogen kernel: [<c11b2520>] ? pci_device_remove+0x0/0x60
Nov 20 18:13:59 hydrogen kernel: [<c1234d5f>] ? driver_register+0x6f/0x130
Nov 20 18:13:59 hydrogen kernel: [<c1068706>] ? notifier_call_chain+0x46/0x80
Nov 20 18:13:59 hydrogen kernel: [<fa0ef000>] ? i915_init+0x0/0x63 [i915]
Nov 20 18:13:59 hydrogen kernel: [<c11b2ad9>] ? __pci_register_driver+0x49/0xd0
Nov 20 18:13:59 hydrogen kernel: [<c100115b>] ? do_one_initcall+0x3b/0x1b0
Nov 20 18:13:59 hydrogen kernel: [<c107e675>] ? sys_init_module+0xe5/0x230
Nov 20 18:13:59 hydrogen kernel: [<c1003cb3>] ? sysenter_do_call+0x12/0x28
Nov 20 18:13:59 hydrogen kernel: ---[ end trace ebc278f591e86437 ]---

Yes, I know it's lame. Yes, I name my computers after the elements in order of performance. Yes, this is my fastest computer. Yes, it's silly because I would have to rename them all if I get a faster one. Don't think so hard!

Last edited by kleptophobiac (2009-11-22 07:51:05)

Offline

#2 2009-11-21 22:14:13

grey
Member
From: Europe
Registered: 2007-08-23
Posts: 679

Re: [SOLVED] Loading module i915 causes kernel OOPS

I thought the i915 module needs to be part of the initrd if you want to use KMS, which is why it needs to be added to mkinitcpio.conf and then you need to run mkinitcpio.
The old intel boards - 82845, 82865 - don't work without KMS according to several forum threads. If that is true for the Q35 as well, then the i915 has to be loaded already by the time to get to a prompt.


Good ideas do not need lots of lies told about them in order to gain public acceptance.

Offline

#3 2009-11-22 01:50:36

kleptophobiac
Member
From: Sunnyvale, CA
Registered: 2004-04-25
Posts: 488

Re: [SOLVED] Loading module i915 causes kernel OOPS

Well, I was trying to figure out why it was crashing, so in order to capture the error message I loaded it by hand after boot. It wasn't working from initrd either. I think there's a configuration setting somewhere that's making the module or kernel unhappy... it's just a matter of trying to figure out what that is.

Thanks!

Offline

#4 2009-11-22 05:56:18

rdmelin
Member
From: Ellensburg, WA, USA
Registered: 2007-08-27
Posts: 45

Re: [SOLVED] Loading module i915 causes kernel OOPS

The nouveau-drm package installs a second drm.ko that gets loaded instead of the one that's included in kernel26. See if you have nouveau-drm and xf86-video-nouveau installed. You probably don't need them.
If you remove them you will need to "mkinitcpio -p kernel26" to install the correct drm.ko in the initrd.

Offline

#5 2009-11-22 07:48:45

kleptophobiac
Member
From: Sunnyvale, CA
Registered: 2004-04-25
Posts: 488

Re: [SOLVED] Loading module i915 causes kernel OOPS

rdmelin wrote:

The nouveau-drm package installs a second drm.ko that gets loaded instead of the one that's included in kernel26.

Bingo. That indeed was the problem, and removing the Nouveau packages fixed it right quick. Curse my laziness when I chose to install all of the xf86-video-drivers group packages. But... how on earth did you know!?

Offline

#6 2009-11-22 15:23:41

rdmelin
Member
From: Ellensburg, WA, USA
Registered: 2007-08-27
Posts: 45

Re: [SOLVED] Loading module i915 causes kernel OOPS

But... how on earth did you know!?

I spent a few hours sorting this one out on a larch livecd development system. I had all the drivers installed to support as wide a variety of hardware as possible. But my laptop didn't work with it.
Saw your post while searching for the cause.
I think it may be a bug that xf86-video-nouveau and nouveau-drm are now pulled in by the xorg-video-drivers group since xf86-video-intel will be broken.

Offline

Board footer

Powered by FluxBB