Bug 36314 - i915 dual-monitor config causes kernel oops
Summary: i915 dual-monitor config causes kernel oops
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/intel (show other bugs)
Version: 7.6 (2010.12)
Hardware: x86-64 (AMD64) Linux (All)
: medium critical
Assignee: Chris Wilson
QA Contact: Xorg Project Team
Depends on:
Reported: 2011-04-16 19:47 UTC by Bryan Christ
Modified: 2011-06-27 10:50 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:

Defend against a NULL crtc (2.57 KB, patch)
2011-04-16 22:45 UTC, Chris Wilson
no flags Details | Splinter Review

Description Bryan Christ 2011-04-16 19:47:19 UTC
When KMS is enable on a dual-monitor configuration (either DVI or DP) plymouthd aborts after kernel oops in drm module.

Platform is a Dell Latitude E6500 laptop with Ubuntu 11.04 alpha1, alpha2, beta1, and beta2 builds.

If I disable KMS at boot time with nomodeset, the problem goes away but then again so does xrandr support for the dual-monitors.
Comment 1 Bryan Christ 2011-04-16 19:48:05 UTC
Screenshots of kernel oops:

Comment 2 Chris Wilson 2011-04-16 22:38:30 UTC
The usual question is which kernel is that? Can you produce with the kernel from the mainline repository (or even one from xorg-edgers)?
Comment 3 Chris Wilson 2011-04-16 22:45:50 UTC
Created attachment 45723 [details] [review]
Defend against a NULL crtc

So this looks to be the most likely explanation. But first you need to get a mainline (or equivalent) so that you can patch and test. :)
Comment 4 Bryan Christ 2011-04-17 20:13:13 UTC

This is a difficult proposition.  My laptop is currently running Ubuntu 10.10--which is working mostly well for my daily work.  I found this bug by using the Ubuntu 11.04 Live alpha and beta discs to see if 11.04 is viable (which at this point it is not because of this bug).  I cannot simply wipe out my laptop to install 11.04 with the prospect of it not working as it would affect my ability to work.  I am open to ideas though.  Please let me know what you think.
Comment 5 Chris Wilson 2011-04-17 22:25:06 UTC
Build the kernel for your normal installation and remove it afterwards. It should hit the same bug as it looks to be before userspace starts.
Comment 6 Bryan Christ 2011-04-19 15:35:20 UTC
I compiled the vanilla kernel and ran into this problem on my existing system.  I will apply the patch an compile again tonight or tomorrow and see if that remedies it.
Comment 7 Bryan Christ 2011-04-19 18:25:40 UTC

I am able to use the vanilla kernel on my dual DVI configuration at home while I get the kernel oops at work (with the two DP monitors).  When I boot from the Live CDs I have the problem whether it is DVI or DP.  Does that make sense to you?

Regardless, I will try patching the kernel, recompiling, and testing on my DP configuration tomorrow.
Comment 8 Chris Wilson 2011-04-19 23:20:29 UTC
Yeah, AIUI, the OOPS requires the DP to be activated by the BIOS on the second pipe and then we try to disable it without setting it up (and so deref a NULL pointer).
Comment 9 Bryan Christ 2011-04-20 08:38:58 UTC
Booted with no oops on DP this morning with the patched kernel.  Looks like this solves the problem.  Need to get the Ubuntu team to pick up this patch before they GM.  Do you have contacts to do that?
Comment 10 Chris Wilson 2011-04-20 08:43:49 UTC
With your positive feedback from testing, I've earmarked the patch for stable, so it should feed quickly into the distributions.
Comment 11 Bryan Christ 2011-04-20 09:25:04 UTC

On Wed, Apr 20, 2011 at 10:43 AM, <bugzilla-daemon@freedesktop.org> wrote:

> https://bugs.freedesktop.org/show_bug.cgi?id=36314
> --- Comment #10 from Chris Wilson <chris@chris-wilson.co.uk> 2011-04-20
> 08:43:49 PDT ---
> With your positive feedback from testing, I've earmarked the patch for
> stable,
> so it should feed quickly into the distributions.
> --
> Configure bugmail: https://bugs.freedesktop.org/userprefs.cgi?tab=email
> ------- You are receiving this mail because: -------
> You are on the CC list for the bug.
> You reported the bug.
Comment 12 Bryan Christ 2011-04-20 09:25:33 UTC
Comment 13 Chris Wilson 2011-06-27 10:50:47 UTC
The patch (rather a variant thereof) is now upstream, closing.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.