Bug 10586

Created attachment 9535 [details]
X log file

Could you try again with the released server/driver?

Still crashes with the latest debian unstable i.e. xorg-core 1.3.0, intel 2.0.0, mesa 6.5.2.

Also briefly tried VT switching while running xcompmgr. That seemed to work fine.

I found a workaround!
It seems the bug can only be reproduced with vga (as opposed to framebuffer) terminals. With the kernel using vesafb, I can't make it crash anymore.

I'm now using xorg-core 1.3.0, intel 2.1.0, mesa 6.5.3, kernel 2.6.22-rc7.
With the regular vga terminals, the same crash occured every once in a while, even if any GL apps hadn't been used.

Martin,

I've reported a similar bug at launchpad: https://bugs.launchpad.net/bugs/127101
There's even a video there showing how the screen looks when my laptop hangs, do you think it's the same bug?
Thanks.

Yes, that's exactly what it looks like!

Also, I said earlier that vesafb seemed to help. Actually it does still hang occasionally. With vesafb the screen just stops (no blinking).

I'm not seeing this issue on:
-- Ubuntu Feisty 7.04 (xserver 1.2.0, intel 1.9.94, Mesa 6.5.2) on 965GM
-- Fedora 7 (xserver 1.3.0, intel 2.0.0, Mesa 6.5.2) on 965G

Gordon: 3/3 reporters whose hardware I checked were using a 945GM, so I wouldn't expect testing on a G/GM965 to show the issue.

I'm seeing this too. It happens randomly and when the hang happens the blocks on the screen are pretty much as in the video on the launchpad bug. The laptop is a Lenovo T60 running at 1024x768.

lspci output is:
00:02.0 VGA compatible controller [0300]: Intel Corporation Mobile 945GM/GMS/940GML Express Integrated Graphics Controller [8086:27a2] (rev 03)

It seems to happen more often when a composite manager is being used but I have also had this happen when logging out of sessions that have only been using metacity. It does not seem to happen every time. I am currently trailing whether this happens when DRI is disabled.

Version information:
xorg 1:7.2-5ubuntu7
xserver-xorg-video-intel 2:2.1.1-0ubuntu2

I am experiencing the same issue with Intel driver 2:2.1.1-0ubuntu3 and i915GM on Ubuntu Gutsy Gibbon.
This doesn't happen with i810 driver.

Hardware:
Sony Vaio TX2
00:02.0 VGA compatible controller: Intel Corporation Mobile 915GM/GMS/910GML Express Graphics Controller (rev 03)

This seems to be fairly reproducible using steps mentioned by Brian inhttps://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/127101 - basically start/stop video and switch back and forth to VTs. 

It seems that this issue has something to do with bug report http://bugs.freedesktop.org/show_bug.cgi?id=10809 . The patch seems to fix the issue at least for me after some tests.

Btw, I've uploaded the patch to fix this in Ubuntu today:

https://bugs.launchpad.net/ubuntu/+bug/127101

Created attachment 11945 [details]
gdb output

Unfortunately, the patch in bug #10809 doesn't fix this bug.
I'm attaching gdb output when trying to reproduce this bug. I was able to reproduce it after my second attempt.

Steps I did to reproduce it:
1. Disable compiz and enable compiz again.
2. Switch to VT1 and switch back to VT7.
3. Repeat steps 1 and 2 several times.

You can replace step 1 with playing some video file using xv, but sometimes you get a SIGABRT and end up with bug #12437.

After the last "step" in the output attached, you can see the gray blocks and the whole laptop hangs (no networking, sysrq, etc.).

I was using:
xserver-xorg-video-intel                   2:2.1.1-0ubuntu5
xserver-xorg-core                          2:1.3.0.0.dfsg-12ubuntu8

Created attachment 11964 [details]
gdb output with packages from debian unstable

I've installed latest -core and -intel packages from Debian unstable to test this again. gdb output attached. Package versions are:

xserver-xorg-core                          2:1.4-3
xserver-xorg-video-intel                   2:2.1.1-4

The laptop hangs in the following loop at lines 1959-1961 (file i830_driver.c):

   for(i = 0; i < 256; i++) {
      OUTREG(PALETTE_A + (i << 2), pI830->savePaletteA[i]);
   }

*** Bug 10768 has been marked as a duplicate of this bug. ***

could you please try the patch in bug# 10768 comment# 23?

There have been some VT fixes committed in the git recently, which do fix some VT bug reports.
Could anyone retest this bug with the git tip or 2.2 release?

Feedback timed out. I'm assuming it's fixed by the recent VT-related patches. Please reopen if you still see it with 2.2.x driver or git tip.