SIGSEGV signal (no mapping at the fault address) : in _XmAddHashEntry

Question

I have ported a motif based program from Compaq Tru 64(in Tru 64 program is running without any problem) to Solaris 8. I have used Forte developer 6 C++ compiler. Program compiled & linked successsfull. But during runtime 'XmCreateMainWindow' method call generating SIGSEGV signal. Debugger says 'signal SEGV (no mapping at the fault address) in _XmAddHashEntry at 0xfee79ff0
0xfee79ff0'.
Program :
Widget children[6]; /* Children to manage */
Arg al[64]; /* Arg List */
register int ac = 0; /* Count */
XtPointer tmp_value; /* ditto */
if ( !widget_name )
widget_name = (char *)"mainWindow1";
XtSetArg(al[ac], XmNscrollingPolicy, XmAUTOMATIC);
mainWindow1 = XmCreateMainWindow ( parent, widget_name, al, ac );
Stack :
main(argc = 11, argv = 0xffbef1c4)
mMainAppShell_c::create(this = 0xef508, display =
0xeaf70, app_name = 0xeeae8 "IP AGENT- QNR002",
app_argc = 11, app_argv = 0xffbef1c4,
app_class_name = 0xc3a49 "XApplication")
mainWindow1_c::create(this = 0xef520, parent =
0xe9b08, widget_name = 0xc1d81 "mainWindow1")
XtCreateWidget(0xc1d81, 0x0, 0xe9b08, 0xffbee12c,
0x0, 0xfef0e768)
_XtCreateWidget(0xc1d81, 0xfef0e768, 0xff0f7c74,
0xffbee12c, 0x0, 0x0)
XtInitializeWidgetClass(0xfef0e768, 0x4925c,
0xff0f4000, 0x1f, 0xfef0e768, 0xfed1b86b)
XtInitializeWidgetClass(0xfef10dbc, 0x2afe,
0xff0f4000, 0x1f, 0xfef10dbc, 0xff094ae2)
XtInitializeWidgetClass(0xfef14290, 0x2afe,
0xff0f4000, 0x1f, 0xfef14290, 0x0)
ClassPartInitialize(0xfef14354, 0xfef14350,
0xfef1434c, 0xfef04000, 0xfef14290, 0xff0f787c)
XmeTraitSet(0xfef14290, 0x0, 0xe78a8, 0xfef19934,
0xfef04000, 0xfef1434c)
_XmAddHashEntry(0x0, 0xe78a8, 0xfef19934, 0xd,
0xf541c, 0xc)
Error :
t@1 (l@1) signal SEGV (no mapping at the fault address) in _XmAddHashEntry at 0xfee79ff0
0xfee79ff0: _XmAddHashEntry+0x0004:     ld      [%i0 + 0xc], %l0
Current function is mainWindow1_c::create
0xfee79ff0: _XmAddHashEntry+0x0004:     ld      [%i0 + 0xc], %l0
0xfee79ff0: _XmAddHashEntry+0x0004:     ld      [%i0 + 0xc], %l0
(dbx) (dbx)

Answer 1

Tough to debug this from the information that you posted but it appears that _XmAddHashEntry is being called with NULL (0x0) as the first parameter and the instruction that is causing problems is the load at 0xfee79ff0 that is trying to use the first parameter (%i0) with the offset of 0xc.  This address is going to be in the kernel space somewhere and the segmentation violation is pointing out that the kernel isn't happy about you peeking at it's unders.
Perhaps you can backtrack from the call to XmAddHashEntry to see what object or class is expected as the first parameter and deduce how that parameter might be getting set (or not set as it were) to NULL.
Also, you mentioned that you're moving from a 64-bit OS. Did you compile your program with the -xarch=v9 flag?
Another thing to check is your stacksize variables. There are two which control the size of various aspects of the stack.
from the C-shell you should type:
unlimit stacksize
and
setenv STACKSIZE 40000
The first sets the overall process stacksize to whatever the system will allow (unlimited). The second, controls how much any individual thread can take for its stack. The second won't effect your application unless it's multi-threaded.