Re: OAF-2 problem

[Bill Haneman <bill haneman sun com>]

I am replying to myself...

Bill Haneman wrote:
[...]
> oaf-query seems OK but running oaf_activate (or activate_from_id)
> seems to make oafd SEGV.
> 
> make check seems to fail in the same way.  I attach what gdb had to
> say about the matter, looks like a null-valued property is being
> stored which seems suspicious, nothing in the app code that looks odd.
etc.

after some time with gdb, I see that by this call both val and newval
are pointing to null.

it seems that the alignment macros in my build of ORBit were set for
'zero' byte alignment, which was causing the pointers to get mangled. 
I haven't found the root cause in the config, but hacking the
alignment macros solved the problem.

Just FYI

regards,

-Bill

> ----------------------------------------------------------------------
> Received signal SIGSEGV, Segmentation fault.
> 0x4006bb87 in _ORBit_copy_value (val=0xbffff500, newval=0xbffff4f0, tc=0x40075620)
>     at corba_any.c:843
> 843             *(CORBA_char **)*newval = CORBA_string_dup(*(CORBA_char **)*val);

--------------
Bill Haneman
Gnome Accessibility / Batik SVG Toolkit
Sun Microsystems Ireland