After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 456391 - All GTK+ builds newer than 2.10.9 cause crash on Win32 !!!
All GTK+ builds newer than 2.10.9 cause crash on Win32 !!!
Status: RESOLVED DUPLICATE of bug 316645
Product: gtk+
Classification: Platform
Component: Backend: Win32
2.10.x
Other Windows
: Normal critical
: ---
Assigned To: gtk-win32 maintainers
gtk-bugs
Depends on:
Blocks:
 
 
Reported: 2007-07-12 20:57 UTC by MuldeR
Modified: 2007-07-13 00:06 UTC
See Also:
GNOME target: ---
GNOME version: ---



Description MuldeR 2007-07-12 20:57:39 UTC
Hi!

I'm not sure if this is the right place to post this bug report, but here we go: It turns out that all GTK+ builds after v2.10.9 crash on some Win32 machines - including mine. The application I am referring to is Avidemux (http://www.avidemux.org), an OpenSource cross-platform video editor and transcoder. The user interface of Avidemux is based on GTK+. There also is a QT4 version of Avidemux though. I'm very interested in the development of Avidemux for Win32, since I'm on WindowsXP. The application is working great so far. But at a certain point I encountered mysterious crashes: The application was disappearing suddenly without any notice. This was happening with features that normally work stable. Also we were not able to track down the reason for the crash - no useful information could be obtained from the Log or the Debugger. Furthermore only a few people encountered the problem, others were unable to reproduce them. So I began to check various dependencies, including t
he GTK+ libraries. Finally it turned out that it was GTK+ which caused those mysterious crashes! As soon as I downgraded to GTK+ v2.10.9 everything worked 100% fine again. All newer builds I tested (v2.10.11, v2.10.12 and v2.10.13) cause crashes! Are there any known problems like this? For information: My system is WindowsXP 32-Bit SP-2 running on an AMD AthlonXP 2800+.

The discussion about the GTK+ crash is here:
http://www.avidemux.org/admForum/viewtopic.php?pid=19685#p19685

The Win32 builds are hosted here:
* http://www.razorbyte.com.au/avidemux/
* http://mulder.dummwiedeutsch.de/home/?page=projects#avidemux


----------------------------------------------

Here is some information from the Windows Debugger:

(c34.578): Break instruction exception - code 80000003 (first chance)
eax=7ffdf000 ebx=00000001 ecx=00000002 edx=00000003 esi=00000004 edi=00000005
eip=7c911230 esp=0390ffcc ebp=0390fff4 iopl=0         nv up ei pl zr na pe nc
cs=001b  ss=0023  ds=0023  es=0023  fs=0038  gs=0000             efl=00000246
*** ERROR: Symbol file could not be found.  Defaulted to export symbols for C:\WINDOWS\system32\ntdll.dll - ntdll!DbgBreakPoint:
7c911230 cc              int     3
0:003> g
(c34.ca0): Unknown exception - code c000008e (first chance)
(c34.ca0): Unknown exception - code c000008e (!!! second chance !!!)
eax=0022ce10 ebx=00000176 ecx=0022cd90 edx=2cd30000 esi=0022ce4c edi=0027b574
eip=58d9433e esp=0022cd60 ebp=0022cdc8 iopl=0         nv up ei pl nz na pe nc
cs=001b  ss=0023  ds=0023  es=0023  fs=003b  gs=0000             efl=00000206


----------------------------------------------

This is what gdb said:

Program received signal SIGFPE, Arithmetic exception.
0x58d9433e in ICM32!CMCheckColorsInGamut ()
from /cygdrive/c/WINDOWS/system32/icm32.dll
(gdb) bt
  • #0 ICM32!CMCheckColorsInGamut
    from /cygdrive/c/WINDOWS/system32/icm32.dll
  • #1 ??


----------------------------------------------

Cheeeeeeeers
MuldeR
Comment 1 Tor Lillqvist 2007-07-12 22:57:08 UTC
This is most probably caused by a problem in some third-party DLL that attaches itself to all programs and that has been built with a Borland compiler that generates code that mishandles the floating-point control word in the CPU. One well-known culprit is something called "AllChars". Do you run that? Also some HP printer drivers have been mentioned.

See bug #316645 and the other bugs that have been marked as a duplicate of that.

*** This bug has been marked as a duplicate of 316645 ***
Comment 2 MuldeR 2007-07-12 23:06:18 UTC
Nope, I don't run something like "AllChars" on my system.
I have Borldand Delphi 7 Professional installed, but it is not running, when the error occurs...
Comment 3 Tor Lillqvist 2007-07-12 23:14:11 UTC
The crash in icm32.dll looks *very* much like the ones in the other bug reports.

Can you check if some Borland-supplied DLL is attached to the process when it crashes? Some good debugger should be able to tell you that. Then stop that from happening (temporarily rename that DLL, turn off whatever software it is that installs some hook that uses that DLL, or whatever) and try the problematic GTK+  app again.
Comment 4 MuldeR 2007-07-12 23:53:08 UTC
*sigh* I'm suddenly unable to reproduce the crash! In the last days I used a recent build of Avidemux, which uses GTK+ v2.10.9. Now I reverted back to an older build of Avidemux, which used the newer (problematic) GTK+ libs. Although this older build used to crash on me (reproduceable!), right now I cannot reproduce the crash. I also tried to use the latest build and replaced the GTK+ libs with v2.10.13 and I still couldn't reproduce the crash. I really have no idea what has changed and could have fixed the problem. Sorry...

Nevertheless I used ProcessExplorer to seach for injected DLLs in Avideumx:
* C:\Programme\Gemeinsame Dateien\Logitech\Scrolling\LGMSGHK.DLL
* C:\Programme\Logitech\MouseWare\system\LgWndHk.dll
* C:\Programme\RocketDock\RocketDock.dll
* C:\WINDOWS\BricoPacks\Vista Inspirat 2\UberIcon\UberIcon.dll
* D:\ATI Tray Tools\raphook.dll

Are any of these suspicious ???
Comment 5 Tor Lillqvist 2007-07-13 00:06:20 UTC
> Are any of these suspicious ???

I wouldn't know. At least the *names* RocketDock and UberIcon don't exactly sound reliable;)

Anyway, you seem to know your way fine around debuggers and other tools, so I am sure that when you manage to reproduce the problem again, you will be able to track the culprit down.