[OGo-Developer] gsmake2 branch segmentation fault

Will Hawkins developer@opengroupware.org
Fri, 15 Feb 2008 16:11:51 -0500


Good afternoon!

I have been working to get an installation of Ogo using the gsmake2
versions of SOPE and OGo. I am able to get the software installed and
running. I can log in to the Web Interface. However, the action taken
immediately after log in causes the ogo-webui to restart. I have the
following backtrace from the process in gdb:

#0  0x00000000 in ?? ()
#1  0x01088b35 in -[NSAutoreleasePool emptyPool] (self=0xb93a780,
     _cmd=0x1395cf8) at NSAutoreleasePool.m:405
#2  0x0108895a in -[NSAutoreleasePool dealloc] (self=0xb93a780,
_cmd=0x1395cf0)
     at NSAutoreleasePool.m:324
#3  0x0108890f in -[NSAutoreleasePool release] (self=0xb93a780,
_cmd=0x13cab78)
     at NSAutoreleasePool.m:317
#4  0x0116966e in -[NSRunLoop acceptInputForMode:beforeDate:]
(self=0x9409f70,
     _cmd=0x13cac58, mode=0x13ca418, limit_date=0x940f5d0) at
NSRunLoop.m:1137
#5  0x011698f8 in -[NSRunLoop runMode:beforeDate:] (self=0x9409f70,
     _cmd=0x36a648, mode=0x13ca418, date=0x9409838) at NSRunLoop.m:1186
#6  0x002422f0 in -[WOCoreApplication run] (self=0x93cf328, _cmd=0x380e50)
     at WOCoreApplication.m:541
#7  0x00273b5c in WOApplicationMain (_appClassName=0x80564dc, argc=1,
     argv=0xbffecca4) at WOApplicationMain.m:42
#8  0x002962a0 in WOWatchDogApplicationMain (appName=0x80564dc, argc=1,
     argv=0xbffecca4) at WOWatchDogApplicationMain.m:316
#9  0x08049a0c in main ()

I have attempted to debug this but after almost two days I am no closer
to the source. In the emptyPool function he segmentation fault occurs
because imps[hash] is zero (NSAutoreleasePool.m:405) for the object
being deallocated from the pool in that iteration.

If I put a conditional around that line of (ie, if (imps[hash])) the
segmentation fault occurs further down in the objc runtime. I realize
this is probably *not* a problem with the objc runtime, but I was unable
to see any other cause for this further up in the OGo code.

I have seen this bug after having installed gnustep-startup-0.18.4 and
with gnustep-startup-0.19.2. The line number reference above is for
version 0.19.2.

If there is anything else I can do to debug this problem please let me
know. I am very interested in getting to the bottom of this but I seem
to have hit a roadblock. Any information you can share would be very
helpful!

Thanks in advance,
Will