Xorg crashed with SIGSEGV in i830_emit_state

Xorg.0.log

I find "(II) intel(0): Setting screen physical size to 270 x 203" especially strange.

The picture (as long as it lasts) looks fine.

Xorg.0.log.old - don't know whether it's needed, though.

Both theses files were from booting from the Live CD.

Xorg.0.log - this one is the most recent file from my current file system - thus it hopefully shows the last boot, which was successful and from which I'm currently reporting this bug.

This strange physical size thing appears here, too.

Thank you for taking the time to report this bug and help make Ubuntu better. Your log shows Xorg crashing. Please try to obtain a backtrace by following the steps at https://wiki.ubuntu.com/X/Backtracing.

Could you also please also run "sudo lspci -vvnn > lspci.log" and attach the resulting file "lspci.log" to this bug report.

Thanks in advance.

lspci.log

I will try the backtrace tomorrow, since I first have to get hold of a second computer.

I'm not really sure whether I have to "Backtrace with gdb" or "Debugging Error Exits". I have done the later and will attach the logfile. Please tell me in case I should have done the first.

Again, this time with dbg packages. . .

kern.log

syslog

I can confirm this bug for a Kubuntu Hardy 8.04.1 upgrade to a Kubuntu 8.10 Beta 1 upgrade.

I can also confirm this bug for the live cd. This is my hardware Dell Latitude X300 laptop, Intel 855GM graphics card, Broadcom 4309 wifi card.

Thanks Salze. Unfortunately, it was the backtrace with GDB that I was asking for. You will need to follow the instructions in the 'Untrap Signals' section, and then I suggest trying to obtain a post-mortem backtrace, as described in 'Post-mortem backtrace' section.

Sorry for having to ask back again, but...do I have to do just 'Untrap Signals' and 'Post-mortem backtrace' - or everything in between, too?

You don't need to worry about anything in between. The 'Untrap Signals' bit disables Xorgs own signal handler, and then the 'Post-mortem backtrace' section tells you how to get Xorg to do a core dump when it crashes, which you run gdb on afterwards.

I forgot to mention, you will need to install libgl1-mesa-dri-dbg, xserver-xorg-core-dbg, xserver-xorg-dbgsym and libc6-dbg. To get all of these, you should add the following line to your /etc/apt/sources.list:

deb http://ddebs.ubuntu.com hardy main universe

Then run:

sudo apt-get update
sudo apt-get install libgl1-mesa-dri-dbg xserver-xorg-core-dbg xserver-xorg-dbgsym libc6-dbg

I exchanged hardy for intrepid in the sources.list line, since that's what I'm running. If that's not correct then please tell me and I will repeat the steps.

--- cut here ---
GNU gdb 6.8-debian
Copyright (C) 2008 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "i486-linux-gnu"...

warning: Can't read pathname for load map: Input/output error.
Reading symbols from /usr/lib/libpciaccess.so.0...done.
Loaded symbols for /usr/lib/libpciaccess.so.0
Reading symbols from /lib/tls/i686/cmov/libdl.so.2...Reading symbols from /usr/lib/debug/lib/tls/i686/cmov/libdl-2.8.90.so...done.
done.
Loaded symbols for /lib/tls/i686/cmov/libdl.so.2
Reading symbols from /lib/tls/i686/cmov/libpthread.so.0...Reading symbols from /usr/lib/debug/lib/tls/i686/cmov/libpthread-2.8.90.so...done.
done.
Loaded symbols for /lib/tls/i686/cmov/libpthread.so.0
Reading symbols from /usr/lib/libXfont.so.1...done.
Loaded symbols for /usr/lib/libXfont.so.1
Reading symbols from /usr/lib/libXau.so.6...done.
Loaded symbols for /usr/lib/libXau.so.6
Reading symbols from /usr/lib/libfontenc.so.1...done.
Loaded symbols for /usr/lib/libfontenc.so.1
Reading symbols from /usr/lib/libpixman-1.so.0...done.
Loaded symbols for /usr/lib/libpixman-1.so.0
Reading symbols from /usr/lib/libhal.so.1...done.
Loaded symbols for /usr/lib/libhal.so.1
Reading symbols from /lib/libdbus-1.so.3...done.
Loaded symbols for /lib/libdbus-1.so.3
Reading symbols from /usr/lib/libXdmcp.so.6...done.
Loaded symbols for /usr/lib/libXdmcp.so.6
Reading symbols from /usr/lib/i686/cmov/libcrypto.so.0.9.8...done.
Loaded symbols for /usr/lib/i686/cmov/libcrypto.so.0.9.8
Reading symbols from /lib/tls/i686/cmov/libm.so.6...Reading symbols from /usr/lib/debug/lib/tls/i686/cmov/libm-2.8.90.so...done.
done.
Loaded symbols for /lib/tls/i686/cmov/libm.so.6
Reading symbols from /lib/tls/i686/cmov/librt.so.1...Reading symbols from /usr/lib/debug/lib/tls/i686/cmov/librt-2.8.90.so...done.
done.
Loaded symbols for /lib/tls/i686/cmov/librt.so.1
Reading symbols from /lib/libgcc_s.so.1...done.
Loaded symbols for /lib/libgcc_s.so.1
Reading symbols from /lib/tls/i686/cmov/libc.so.6...Reading symbols from /usr/lib/debug/lib/tls/i686/cmov/libc-2.8.90.so...done.
done.
Loaded symbols for /lib/tls/i686/cmov/libc.so.6
Reading symbols from /lib/ld-linux.so.2...Reading symbols from /usr/lib/debug/lib/ld-2.8.90.so...done.
done.
Loaded symbols for /lib/ld-linux.so.2
Reading symbols from /usr/lib/libfreetype.so.6...done.
Loaded symbols for /usr/lib/libfreetype.so.6
Reading symbols from /usr/lib/libz.so.1...done.
Loaded symbols for /usr/lib/libz.so.1
Reading symbols from /lib/tls/i686/cmov/libnss_compat.so.2...Reading symbols from /usr/lib/debug/lib/tls/i686/cmov/libnss_compat-2.8.90.so...done.
done.
Loaded symbols for /lib/tls/i686/cmov/libnss_compat.so.2
Reading symbols from /lib/tls/i686/cmov/libnsl.so.1...Reading symbols from /usr/lib/debug/lib/tls/i686/cmov/libnsl-2.8.90.so...done.
done.
Loa...

>- it crashed at about the third or forth icon of the KDE splash (instead of the first without those debug packages)
>- I got some "fancy colored dots and lines" at crashing time (sorry, don't know how to describe it better) which were then replaced by the correct displayed kdm
>- the consoles (Alt-F1 ...) hung, the screen showed those same colored pattern - without the debug packages the consoles were usable

After upgrading from 8.04 to 8.10 beta, I can confirm a crash after login with these symptoms (except GNOME instead of KDE) on a GMA-950, external monitor @ 1920x1200 hooked up.

00:02.0 VGA compatible controller: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller (rev 03)

Salze - Thanks, but your gdb output actually appears to be missing the backtrace. Did you run 'backtrace full'? And it would be better for you to attach the gdb output as a text file to the bug report, as opposed to posting it inline.

If you invoke gdb like the following, then it will log all terminal output to a file, which you can attach to this report.

sudo su
gdb /usr/bin/Xorg /etc/X11/core 2>&1 | tee /tmp/gdb-Xorg.txt

Ok, sorry. I hope it's better now.

I just upgraded to the -5 kernel plus serveral other updates, of course. The problem still exists, but the handling has worsend (for me). Now x direct restarts with login in after crashing. Before it fell back to kdm.

I upgraded to the -6 kernel - plus several other updates. It still crashes. With my xorg.conf I get two "flashes" of a completely white screen (first longer than second) before the crash. Without a xorg.conf it just crashes. In both cases it crashes+restarts in a loop.

I have the same problem on a HP desktop computer with integrated graphics and DVI add-on card. NoDRI works.

00:02.0 VGA compatible controller: Intel Corporation 82865G Integrated Graphics Controller (rev 02)
        Subsystem: Hewlett-Packard Company Device 12bc
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
        Status: Cap+ 66MHz- UDF- FastB2B+ ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
        Latency: 0
        Interrupt: pin A routed to IRQ 16
        Region 0: Memory at f0000000 (32-bit, prefetchable) [size=128M]
        Region 1: Memory at f8400000 (32-bit, non-prefetchable) [size=512K]
        Region 2: I/O ports at 14e0 [size=8]
        Capabilities: [d0] Power Management version 1
                Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
                Status: D0 PME-Enable- DSel=0 DScale=0 PME-
        Kernel modules: intelfb

Thanks Salze. DId you install the libgl1-mesa-dri-dbg package though? It seems that the symbols are missing from your backtrace.

I followed your instructions above:

"I forgot to mention, you will need to install libgl1-mesa-dri-dbg, xserver-xorg-core-dbg, xserver-xorg-dbgsym and libc6-dbg. To get all of these, you should add the following line to your /etc/apt/sources.list:
deb http://ddebs.ubuntu.com hardy main universe
Then run:
sudo apt-get update
 sudo apt-get install libgl1-mesa-dri-dbg xserver-xorg-core-dbg xserver-xorg-dbgsym libc6-dbg"

The only thing I did different was changing hardy to intrepid. May that be the cause?

Or may it be that I have to remove the packages without -dbg at the end?

--- cut here ---
Salze@lllu0143:~$ apt-cache policy libgl1-mesa-dri-dbg
libgl1-mesa-dri-dbg:
  Installiert: 7.2-1ubuntu1
  Kandidat: 7.2-1ubuntu1
  Versions-Tabelle:
 *** 7.2-1ubuntu1 0
        500 http://de.archive.ubuntu.com intrepid/main Packages
        100 /var/lib/dpkg/status
Salze@lllu0143:~$ apt-cache policy libgl1-mesa-dri
libgl1-mesa-dri:
  Installiert: 7.2-1ubuntu1
  Kandidat: 7.2-1ubuntu1
  Versions-Tabelle:
 *** 7.2-1ubuntu1 0
        500 http://de.archive.ubuntu.com intrepid/main Packages
        100 /var/lib/dpkg/status
--- cut here ---

I'm not sure why it didn't work. Did you install the -dbg packages and then run gdb on an old coredump? If so, it could be that the version of libgl1-mesa-dri now, is newer than when you reproduced the crash and obtained the coredump. If that is the case, it would be worth trying to obtain a new coredump to run GDB on. If that doesn't work, try installing the libgl1-mesa-dri-dbgsym package instead.

Thanks

I'm quite sure that I have first installed the packages and then created the debug file. Are the debug symbols of the other packages missing, too? Or just those of libg1-mesa-dri?

What about my questions from the last posting?

--- cut here ---
The only thing I did different was changing hardy to intrepid. May that be the cause?
Or may it be that I have to remove the packages without -dbg at the end?
--- cut here ---

Concerning new available versions: yesterday (12.10.08) at about this time I ran apt-get upgrade && apt-get update. Afterwards it doesn't even work with the NoDRI-Option. So the situation has become worse.

Because of that I had to restore a backup, because I was totally limited to textmode only.

With the NoDRI-Option it did not crash, but I had either (at first) an all-white screen with just the mouse-pointer, or (after some time) I had a nearly-all-black screen with some very, very faint shadows of some windows.

Unfortunately the whole backup-restore-process takes more time than I currently have. So I cannot say at which time I will be able to re-try the debugging-process with the new packages. I hope you can still answer my questions from above, so that I know exactly how to proceed when I find the necessary time.

An additional question: in aptitude, when selecting the libc6-dbg package, I read:

--- cut here ---
The libraries are installed in /usr/lib/debug and can be used by placing that directory in LD_LIBRARY_PATH.
--- cut here ---

Is that of any concern for me? Either for this package or for the others? Do I have to set any variables?

Just a strange question, if you select GNOME instead of KDE in the menu (if you have that available), do you also have this problem? Currently, I am not able to work in KDE, with what appears as the same issues you have, but GNOME is working for me.

I don't have Gnome installed. Does Gnome use dri? Do the newest packages work for you with NoDRI and KDE?

Can you run the debug thing Chris needs?

> I don't have Gnome installed. Does Gnome use dri? Do the newest packages
> work for you with NoDRI and KDE?

I have not looked into DRI and NoDRI, so I don't even know what that is,
I just noticed that my problems resemble yours. I don't know if Gnome
uses it.

> Can you run the debug thing Chris needs?

I will have a look at this tonight, but the NoDRI didn't solve my KDE
problem. I'll report back.

Paul

Can somebody please post how to get the backtraces step by step. I can do a fresh install just for testing purposes if necessary. I really want to see this fixed by the final release. Also I think that the importance level should be raised. This graphics card is fairly common and thus this bug could render the installation of thousands of people useless if it stays up to the final release. Any news from upstream on this issue? Thanks to everyone. I know you are working really hard to make this a really good release.

I also have this same problem on an IBM Thinkpad X40. Installed fresh, first Kubuntu hardy, then upgraded to intrepid. Now KDE won't start unless I have the NoDRI option in my xorg.conf. The "failsafe" session option in kdm doesn't work either. No problems on hardy previously.
I kinda need the DRI, so as a workaround I installed IceWM and that loads up just fine. I can even use KDE apps. Konsole, KNetworkManager and Konqueror all work just fine. Accelerated OpenGL also works fine.
I'd be happy to provide any additional information you need.

I've managed to get the newest packages running and still being able to use the computer. I have the following results:

A: NoDRI set in /etc/X11/xorg.conf
B: NoDRI _not_ set in /etc/X11/xorg.conf
1: [Compositing] -> Enabled=true in .kde/share/config/kwinrc
2: 1: [Compositing] -> Enabled=false in .kde/share/config/kwinrc

A1: black windows on black background, some veeeeeeery faint shadows, mouse cursor is ok
A2: ok
B1: crash, see attached file - libgl1-mesa-dri-dbgsym instead of libgl1-mesa-dri-dbg
B2: ok

Procedure for attached file:

1.) made sure that /etc/X11/core does _not_ exist
2.) NoDRI _not_ set
3.) Compositing enabled
4.) made sure
--- cut here ---
Section "ServerFlags"
        Option "NoTrapSignals" "true"
EndSection
--- cut here ---
is in /etc/X11/xorg.conf
5.) made sure
--- cut here ---
enabled=0
--- cut here ---
is in /etc/default/apport
6.) made sure
--- cut here ---
deb http://ddebs.ubuntu.com hardy main universe
--- cut here ---
is in /etc/apt/sources.list
7.) sudo apt-get install xserver-xorg-core-dbg xserver-xorg-video-intel-dbg xserver-xorg-dbgsym libc6-dbg libgl1-mesa-dri-dbgsym
8.) rebooted
9.) Alt-F1, login, sudo su -, ulimit -c unlimited, /etc/init.d/kdm restart
10.) Login as user -> crash
11.) Reboot, sudo su, gdb /usr/bin/Xorg /etc/X11/core 2>&1 | tee /tmp/gdb-Xorg.txt
12.) Reversed all above changes

Hope it helps.

More testing: switching to XRender instead of OpenGL helps. But then some things are quite slow, wobbly windows don't wobble at all and I sometimes get display errors like strange shadows or some regions on the display show what should have been shown on other parts of the screen.

http://groups.google.com/group/linux.debian.bugs.dist/browse_thread/thread/0f60030cdaadec73?pli=1

<- does this have anything to do with the problem?

Hi Salze, no, I think that problem is the cause of bug 256142. Your stack trace is still missing debug symbols, which has me a bit confused. I'll ask someone to take a quick look at this.

Salze - On the advice of Emmet Hikory, I would like you to try and get Apport to catch the crash, and then submit that as a new bug report.

To do this, you may need to enable Apport. To enable Apport, edit /etc/default/apport and change "enabled=0" to "enabled=1". Once you have done this, you will need to restart Apport by typing "sudo /etc/init.d/apport restart".

Once you have done this, try to recreate the Xorg crash. You will still need NoTrapSignals enabled in your Xorg.conf, as if you were generating a coredump (but you don't need to use ulimit like you did before).

Once you have recreated the crash with Apport enabled, you may get a crash notification next time you log in. If not, navigate to /var/crash where there will hopefully be a crash file. Double clicking on this will allow you to submit the crash file as a new bug report.

Hopefully the retracer will generate a more useful stacktrace.

Thanks

Shall the missing debug symbols be reported as a bug on their own?

1.) Made sure apport is set to enabled
2.) Made sure NoTrapSignals is in xorg.conf
3.) Rebooted
4.) Reproduced crash
5.) No new file in /var/crash

I follow climatewarriors position that this bug should have a higher importance set - and I think it maybe should even be set to complete, since everything possible under the given circumstances has been done in terms of providing information/logfiles.

Many users have this graphics card - and apparently it's not only one model that is affected. The only chance of getting the install to run is to use textmode. The only chance to log in after installation as a new user is by editing a textfile two levels inside the hidden .kde directory. I think this is not quite the Ubuntu way.

This is a big regression from having everything running flawless in Hardy.

If it's not possible to get the bug fixed upstream then at least a workaround for new user-installations/upgrades is needed, IMHO.

The update to -1ubuntu8 didn't help.

I'm unable to login to a KDE session as well, also using a rather old intel chip (855GM).
I really think the importance of this bug should be bumped, preferably to critical. As things stand now, people with this rather common chipset which supposedly runs on good open source drivers cannot even log in to a kubuntu session.

Maybe this bug should be sent to someone from the release team to see if they think it is a release critical bug. Just my 0.02$.

To the person who marked bug 282855 as a duplicate of this one - it is generally a good idea to add any information which appears in the duplicate report and which might be important in to the master bug report, when adding the duplicate link. Generally, people probably won't go trawling through duplicates to piece together missing bits of information. There was some useful information in the duplicate report (particularly the link to the upstream bug report, which I have now added to this master).

Paul - in response to your earlier comment regarding this being an issue for you in KDE and not GNOME (https://bugs.edge.launchpad.net/ubuntu/+source/mesa/+bug/277709/comments/27) - do you use Compiz when you log in to GNOME? If not, then you probably won't experience this crash (my understanding is that this only affects compositing window managers). KWin does some compositing which is why KDE uses are seeing it.

I'm going to change this to confirmed for now, but I really would appreciate it if somebody could try to get a good backtrace of this crash. Not even the upstream report contains one, which will make it very difficult to fix.

Please see my instructions earlier in this report.

Chris, I'm not seeing this crash on Gnome with compiz enabled. In fact, if I login to a gnome session (with compiz enabled, but that doesn't matter for this example), and then choose to switch users, and login another user to a KDE session, I don't see this crash. In a case like that, however, kwin's compositing is naturally disabled, and that's, I presume, the reason I can complete a flawless login.

About the backtrace, if I find the time, I will definitely try to capture it for you. But real life is craving these days...

@Chris: Sorry, the person who marked this bug duplicated was me.

But i get the full backtrace by compiling mesa on my own with ./configure --with-driver=dri --with-demos=xdemos --libdir=/usr/lib/glx --with-dri-driverdir=/usr/lib/dri --prefix=/usr --mandir=/usr/share/man --infodir=/usr/share/info --sysconfdir=/etc --localstatedir=/var --enable-debug .

Hope the config flags are right?

tschenser - Thanks for the backtrace, that is very useful. I will send the backtrace upstream shortly.

I applied the patch from upstream. Everything looks fine now, i also didnt see any black screens as haihao descibed.

Subscribing Ubuntu Release Team.

I've created a debdiff with the upstream patch incorporated. I've also uploaded it to my PPA so people can test it (https://launchpad.net/~chrisccoulson/+archive). I would appreciate some feedback.

Thanks

D'oh! Would help if I attached the debdiff.

Test successful. Thank you!

Fixed for me as well. Thanks a lot!

Bryce, could you review the above patch?

no need to subscribe ubuntu-release for a bugfix-only change - please get this uploaded to the queue ASAP instead if it's confirmed good, and the release team will evaluate it for post-RC inclusion in 8.10.

@Scott: Patch looks good, it merely adds a NULL ptr check before dereferencing so is safe and correct.

@Steve: I will upload after verifying it builds.

This bug was fixed in the package mesa - 7.2-1ubuntu2

---------------
mesa (7.2-1ubuntu2) intrepid; urgency=low

  * debian/patches/103_fix-crash-in-i830_emit_state.dpatch:
    - Apply upstream commit to fix a crash in i830_emit_state
      (LP: #277709)

 -- Chris Coulson <email address hidden> Tue, 21 Oct 2008 17:55:10 +0100

this has been bugging me for months, but I just found this report. After reading through all the comments and dups, I found that I just needed to
  aptitude update && aptitude upgrade
Thanks everyone involved in this fix!

I'm still experiencing the crash with kwin in opengl composite mode after installing the update (tried Chris' ppa Mesa too). I have the same chipset several others have mentioned, intel 855GM. Trying to generate a backtrace with gdb.

there is no core dump in /etc/X11/core, perhaps I misunderstood. Should I build mesa from scratch as tschenser did to get a useful backtrace?

Eric - This bug is fixed now so if you're still experiencing a crash, then it is likely to be a separate issue. As a starting point, please open a new bug report and attach your /var/log/Xorg.0.log and /var/log/Xorg.0.log.old after the crash to the new report, as this may offer an insight as to where the crash is happening (there is nothing at this stage to say it is happening in mesa). All further dialog should go in to the new bug report.

Thanks

Here are the backtraces from X and gdm with UnTrap Signals. I'm running latest offical mesa packages.

Ok Chris, sorry it just seems so similar since it only effects kwin, same chipset. Reported as
https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-intel/+bug/285250

Eric - That is an unrelated crash. Please open a separate report.

Thanks

oops, bad link
https://bugs.launchpad.net/ubuntu/+source/xorg/+bug/287485