[Hardy] [regression] xorg with Intel driver crashes if Virtual > 2048

Bug #188178 reported by chantra
18
Affects Status Importance Assigned to Milestone
xf86-video-intel
Fix Released
Medium
xserver-xorg-video-intel (Debian)
Fix Released
Unknown
xserver-xorg-video-intel (Ubuntu)
Fix Released
High
Unassigned

Bug Description

Ubuntu hardy
Gnome 1:2.20.2ubuntu1
xserver-xorg-video-intel 2:2.2.0+git20080107-1ubuntu2

Using the same dual screen setup than on Gutsy, X crashes after 10 second of login.

This happens using kde or gnome
Funny enough, it does not happen when login into kde 4 !?!

Revision history for this message
chantra (chantra) wrote :
Revision history for this message
chantra (chantra) wrote :
Revision history for this message
chantra (chantra) wrote :
Changed in xserver-xorg-video-intel:
status: Unknown → Confirmed
Revision history for this message
TerryG (tgalati4) wrote :

Marking incomplete pending more info. Intel driver has known dual-screen limitation of 2048 by 2048 pixel space. If your screens (combined) exceed this limit (in either direction) you will get an xorg crash. Log file shows hitting video memory limit before lockup.

Try reducing resolution on both screens to stay within 2048 pixel limitation. KDE4 may choose safe resolutions by default.

Changed in xserver-xorg-video-intel:
status: New → Incomplete
Revision history for this message
chantra (chantra) wrote :

I am running 3200x1200 daily at work on Gutsy
xserver-xorg-video-intel 2:2.1.1-0ubuntu9.1

without any issue. It is a regression to me.
3D does not work on resolution > 2048x2048 , ok, but still I believe intel 945GM is shipped with a good bunch of laptops.
That would extremelly limit hardy in enterprise environment :(

Revision history for this message
unggnu (unggnu) wrote :

I can confirm this but it could be fixed for me with this lines under device section in xorg.conf:

       Option "AccelMethod" "EXA"
       Option "ExaNoComposite" "false"
       Option "MigrationHeuristic" "greedy"

But I guess only the last one is needed. I'll report back when I have recheck it.

Revision history for this message
chantra (chantra) wrote :

Tks Unggnu.

Option "MigrationHeuristic" "greedy"
in the device section fixed it for me too, on both hardy and debian sid

Revision history for this message
TerryG (tgalati4) wrote :

Marking as Confirmed with work arounds presented.

Changed in xserver-xorg-video-intel:
status: Incomplete → Confirmed
Revision history for this message
unggnu (unggnu) wrote :

This issue should be fixed if greedy is used per default for all Intel cards so it is connected to Bug #177492.

Changed in xserver-xorg-video-intel:
status: Unknown → Confirmed
Revision history for this message
Id2ndR (id2ndr) wrote :

The greedy workaround doesn't work for me, neither the others options.
I have errors in dmesg :"mtrr: no more MTRRs available".

Revision history for this message
chantra (chantra) wrote :

Id2ndR, can you send some more information, such as logs...., output of lspci, your xorg.conf ...

FYI, "mtrr: no more MTRRs available" is related to memory.
Chances are that you have more than 4G memory on a 32bit architecture

Revision history for this message
Id2ndR (id2ndr) wrote :

xorg.conf (virtual section was uncommented first, and then commented) :

Section "InputDevice"
 Identifier "Generic Keyboard"
 Driver "kbd"
 Option "XkbRules" "xorg"
 Option "XkbModel" "pc105"
 Option "XkbLayout" "fr"
 Option "XkbVariant" "oss"
EndSection

Section "InputDevice"
 Identifier "Configured Mouse"
 Driver "mouse"
EndSection

Section "Device"
 Identifier "Configured Video Device"
 Option "MigrationHeuristic" "greedy"
# Option "AccelMethod" "EXA"
# Option "ExaNoComposite" "false"
EndSection

Section "Monitor"
 Identifier "Configured Monitor"
EndSection

Section "Screen"
 Identifier "Default Screen"
 Monitor "Configured Monitor"
 Device "Configured Video Device"
# Virtual 2560 1024
# Virtual 2880 1200
EndSection

Revision history for this message
Id2ndR (id2ndr) wrote :

I have 2GB of memory on a 32bits system.

Revision history for this message
Id2ndR (id2ndr) wrote :
Revision history for this message
Id2ndR (id2ndr) wrote :

I've done more tests :
- Virtual size smaller than 2048 : neither 1600x800 nor 1280x1824 worked. I got the "low resolution" screen, and mtrr error in dmesg.
- Virtual size of the size of my LVDS : 1280x800, and no external screen connected. It didn't switch to gdm, and I just got the console login screen. No error on dmesg, but these errors in gdm log :

Parse error on line 48 of section Screen in file /etc/X11/xorg.conf
        "Virtual" is not a valid keyword in this section.
(EE) Problem parsing the config file
(EE) Error parsing the config file

Fatal server error:
no screens found

I haven't seen error about "Virtual" keyword when I got the external screen connected, so this may be normal.

I don't use the greed option anymore, but I use a new version of the intel driver that doesn't neet it : https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/177492/comments/84

Revision history for this message
Emil Sit (emilsit) wrote :

Not quite sure if this is related but while trying to debug poor glxgears performance on my thinkpad x40 with Hardy Beta, I got a similar crash (log attached). My normal working configuration uses Driver "i810" for an i855 1024x768 internal and 1280x1024 external lcd on the vga port. Xinerama is disabled so they are two separate screens. glxgears reports about 500 fps. I tried switching Driver to "intel" and the server would crash on startup (no login).

Ideally, I'd have two separate screens (non-Xinerama) both with acceleration enabled; purportedly this hardware can get at least 700 fps and possibly even 1270 fps (http://www.chronox.de/x40_linux/linux_on_x40.html).

I can file a separate bug if it is unrelated.

Changed in xserver-xorg-video-intel:
status: Confirmed → Fix Released
Revision history for this message
chantra (chantra) wrote :

merging patch from http://bugs.freedesktop.org/show_bug.cgi?id=13376

removing
Option "MigrationHeuristic" "greedy"
for xorg.conf

with a
Virtual 3200 1200

is not crashing anymore

Bryce Harrington (bryce)
Changed in xserver-xorg-video-intel:
importance: Undecided → High
status: Confirmed → In Progress
Revision history for this message
Clemens Hupka (clemens) wrote :

Had almost the same problem after an update from gutsy to hardy, used URandR dualscreen setup on gutsy:

Option "MigrationHeuristic" "greedy"

AND

Option "DRI" "false" # Was set true on gutsy

Prevents from crashing with a virtual of 2960x1050

Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package xserver-xorg-video-intel - 2:2.2.1-1ubuntu12

---------------
xserver-xorg-video-intel (2:2.2.1-1ubuntu12) hardy; urgency=low

  * debian/patches/17_lockup_virtual_size_2048.patch:
    - merging patch from http://bugs.freedesktop.org/show_bug.cgi?id=13376.
      Fixes crash if Virtual set to >2048 by disabling DRI in this case,
      prior to initializing the DRI subsystem. (closes LP: #188178)

 -- Emmanuel Bretelle <email address hidden> Sat, 29 Mar 2008 14:06:49 +0000

Changed in xserver-xorg-video-intel:
status: In Progress → Fix Released
Changed in xserver-xorg-video-intel:
status: Confirmed → Fix Released
Revision history for this message
Fredrik Wendt (fredrik-wendt) wrote :

@Emil Sit: According to XStrikeForce[0] xrandr1.2 can't handle good old multi display setups (non-Xinerama style), which is very unfortunate and brings us back to a Windows-alike level of usability again (where there's one desktop and it spans all outputs - if you change the workspace of the right screen then the left screen should not move).
Hopefully it will be possible to use xrandr+intel (not i810)+compiz with two DISPLAYs (not two screens that are part of one virtual display) when xrandr 1.3 is released.

[0] http://wiki.debian.org/XStrikeForce/HowToRandR12

Changed in xserver-xorg-video-intel:
importance: Unknown → Medium
Changed in xserver-xorg-video-intel:
importance: Medium → Unknown
Changed in xserver-xorg-video-intel:
importance: Unknown → Medium
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.