Bug 3289 - mgp (magicpoint) crashes Xorg with Composite
Summary: mgp (magicpoint) crashes Xorg with Composite
Status: RESOLVED FIXED
Alias: None
Product: xorg
Classification: Unclassified
Component: Server/General (show other bugs)
Version: git
Hardware: Other Linux (All)
: medium normal
Assignee: Xorg Project Team
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-05-13 08:12 UTC by Bernhard Reiter
Modified: 2007-02-22 14:27 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
gdb Debugging session (5.81 KB, text/plain)
2005-05-17 13:01 UTC, Bernhard Reiter
no flags Details
Typical log file when crashing. (34.46 KB, text/plain)
2005-05-24 05:17 UTC, Bernhard Reiter
no flags Details
One xorg.conf that I can see the crash with. (2.42 KB, text/plain)
2005-05-24 05:19 UTC, Bernhard Reiter
no flags Details

Description Bernhard Reiter 2005-05-13 08:12:23 UTC
Xorg CVS HEAD from 20050429 (6.8.99.4?)and 6.8.2 (with and without some fedora
patches):

Plattform: Debian GNU/Linux ppc (G4 Powerbook with Radeon 9600 M10)
Running mgp crashes Xorg, signal 11 I could not get further information
out of the server yet. Here is an attempt to gdb it:

Program received signal SIGSEGV, Segmentation fault.
0x0fe036d0 in mallopt () from /lib/libc.so.6
(gdb) Undefined command: "module".  Try "help".
(gdb) #0  0x0fe036d0 in mallopt () from /lib/libc.so.6
#1  0x0fed1a1c in ?? () from /lib/libc.so.6
#2  0x0fed1a1c in ?? () from /lib/libc.so.6
Previous frame identical to this frame (corrupt stack?)
(gdb) Continuing.

   *** If unresolved symbols were reported above, they might not
   *** be the reason for the server aborting.

Fatal server error:
Caught signal 11.  Server aborting

I know a second user that can confirm the problem.

Here is an strace from mgp:

http://intevation.de/Intevation/~bernhard/ppc/powerbook5,6/mgp-strace-20050503.txt

and an strace from one server run:
http://intevation.de/~bernhard/ppc/powerbook5,6/xorg-strace-20050504.txt
Comment 1 Bernhard Reiter 2005-05-13 08:13:11 UTC
Evne when building with -g and Dll modules, I cannot get more information 
from gdb.
Comment 2 Bernhard Reiter 2005-05-17 07:32:31 UTC
The stack corrupt message could be due to a gdb problem with -f PIC and -O2 on
powerpc. I wonder if other people have problems with magicpoint on xorg.
Comment 3 Georg C. F. Greve 2005-05-17 08:03:30 UTC
I also experience problems with Magicpoint and X.Org, but am not 100% sure where
they originate. It _used_ to work just fine, but after doing a dist-upgrade to
sarge last weekend, magicpoint now dies with the message:

 X Error of failed request:  BadMatch (invalid parameter attributes)
   Major opcode of failed request:  75 (X_PolyText16)
   Serial number of failed request:  604
   Current serial number in output stream:  605

FYI -- installed versions:
ii  xorg-common              6.8.1-0.4                X Window System (X.Org)
infrastructure
ii  xorg-driver-synaptics    0.13.6-2                 Synaptics TouchPad driver
for X.Org server
ii  xserver-xorg             6.8.1-0.4                the X.Org X server
ii  mgp                      1.11b-2                  MagicPoint- an X11 based
presentation tool
Comment 4 Michel Dänzer 2005-05-17 08:11:51 UTC
(In reply to comment #2)
> The stack corrupt message could be due to a gdb problem with -f PIC and -O2 on
> powerpc. I wonder if other people have problems with magicpoint on xorg.

What if you build with 'makeg' instead of 'make'? That should add -g and remove -O2.
Comment 5 Bernhard Reiter 2005-05-17 09:59:20 UTC
Michel: Thanks for the hint, it seems that
http://wiki.x.org/wiki/DebuggingTheXserver has mislet me into believing
that seeing the option in host.def would be fine, but I just rechecked
and found out that I actually did not use "-g" when building the version,
so now I try makeg and the build roles. Can you update the wiki?
Comment 6 Bernhard Reiter 2005-05-17 11:18:57 UTC
Now really rebuild xorg cvs from today (20050517) and starting mgp
gives me the following message now:

X Error of failed request:  BadAlloc (insufficient resources for operation)
Major opcode of failed request:  78 (X_CreateColormap)
Serial number of failed request:  9
Current serial number in output stream:  10
Comment 7 Bernhard Reiter 2005-05-17 11:23:44 UTC
In my next attempts I get the crash again, so maybe xorg is corrupting its own
stack?
Comment 8 Bernhard Reiter 2005-05-17 13:01:56 UTC
Created attachment 2694 [details]
gdb Debugging session

Here is a gdb debugging session. gdb itself seems to be able to produce
backtraces
to a certain extend.
Comment 9 Bernhard Reiter 2005-05-17 13:02:49 UTC
My tests are on cvs_head now.
Comment 10 Michel Dänzer 2005-05-23 13:06:17 UTC
Please also attach the X server configuration and log files. In particular, I'm
curious whether the Composite extension is enabled?
Comment 11 Bernhard Reiter 2005-05-24 05:17:18 UTC
Created attachment 2762 [details]
Typical log file when crashing.
Comment 12 Bernhard Reiter 2005-05-24 05:19:28 UTC
Created attachment 2763 [details]
One xorg.conf that I can see the crash with.

Note that I want to have an external display working,
thus the larger virtual screen. I did see the crash without external display,
too.
Comment 13 Bernhard Reiter 2005-05-24 05:21:09 UTC
Composite currently is enabled.
I will try without next.

Note that I do start this server as extra server on a free screen terminal
so I can keep my production XFree 4.3. running in parallel. Could this make 
a difference? (I believe I tested it without Xfree running and saw no change in
the behaviour.)
Comment 14 Bernhard Reiter 2005-05-24 05:32:56 UTC
Heureka!
Disabling Composite was the right thing to do!
I can use mgp now!  :)
Michel: Thanks again for the hint!

Thus I am changing the title and the severity of the bug.
Is that fine now?
Comment 15 Eric Anholt 2005-10-01 12:26:53 UTC
Just a note that I failed to reproduce this issue with "mgp -g 400x300
/usr/X11R6/share/doc/mgp/sample/sample.mgp".  CVS head as of around 2005-09-28
and magicpoint 1.11b, composite enabled, XAA.
Comment 16 Eric Anholt 2006-06-16 16:07:20 UTC
Need more info on how to reproduce, or a good stack trace from gdb.  Also, does
the problem still exist with current xserver from git?  I just fixed a crash
that may occur with XAA and Composite.
Comment 17 Bernhard Reiter 2006-06-26 09:47:16 UTC
(In reply to comment #16) 
> Need more info on how to reproduce, or a good stack trace from gdb.   
 
My machine is a powerpc (big endian) this could be a major difference. 
The problem with gdb is, that the version I got with Debian Sarge 
does not work proberly regarding stack traces, so it is a bit away for me 
to get a better backtrace. 
 
> Also, does 
> the problem still exist with current xserver from git?  I just fixed a crash 
> that may occur with XAA and Composite. 
 
Unfortunatley it might take me a while until I can do 
a new test compile and find out if the bug is still there. 
I am currently running without composite which makes it non-critical for me. 
Thanks a lot for the response, the hint about the potential fix and for 
contributing to Free Software in general! 
 
 
Comment 18 Timo Jyrinki 2007-02-22 14:27:55 UTC
Marking broken (status null/blank) bugs in xorg with no activity in a long time as fixed. Please reopen if you think it's necessary, but first do a search if a similar bug report is already filed and in a NEW/ASSIGNED state. These bugs do not currently show in most search results as they do not have any status.

Sorry for this janitorial spam, you know where to send hate mails to when your inbox gets full of bugs you're subscribed to.


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.