Hello,

I'm using GnuStep for a couple of months now but I
encoutered from the beginning this problem:

font_cacher gives segv

backtrace:
(gdb) run
Starting program: 
/home/build/gnustep/gnustep-xgps-0.7.0/Tools/shared_debug_obj/ix86/linux-gnu/gnu-gnu-gnu/font_cacher
[New Thread 1024 (LWP 31693)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1024 (LWP 31693)]
0x406e8853 in hash_string () from 
/usr/GNUstep/System/Libraries/ix86/linux-gnu/libobjc.so.1
(gdb) backtrace
#0  0x406e8853 in hash_string () from 
/usr/GNUstep/System/Libraries/ix86/linux-gnu/libobjc.so.1
#1  0x0804f868 in _OBJC_SELECTOR_TABLE ()
#2  0x406e625b in hash_value_for_key ()
    from /usr/GNUstep/System/Libraries/ix86/linux-gnu/libobjc.so.1
#3  0x406e8440 in __sel_register_typed_name ()
    from /usr/GNUstep/System/Libraries/ix86/linux-gnu/libobjc.so.1
#4  0x406e6a7d in __objc_exec_class ()
    from /usr/GNUstep/System/Libraries/ix86/linux-gnu/libobjc.so.1
#5  0x0804b9ec in global constructors keyed to XGFontPropULong () at 
XGCommonFont.m:284
#6  0x0804beb5 in __do_global_ctors_aux ()
#7  0x0804916a in _init ()
#8  0x407260c7 in __libc_start_main () from /lib/libc.so.6

I'm using gcc 2.96 on mandrake 8.0 linux
but I found this problem on every machine I touched
(various linuces gcc 2.95.2 intel, gcc 2.96 alpha (ev6), gcc 3.0 alpha 
(ev6))

I use the gnustep-objc-1.0.1  patches (exept for gcc 3.0 an alpha)


I don't see any static initializers in the XGCommonFont.m
(the line 284 is at EOF) and expecially in XGFontPropULong.
Are "+initialize" methods called before main, though 
__do_global_ctors_aux by the runtime ?

I can't breakpoint inside __do_global_ctors_aux
(I've also compiled all packages with debug infos but it was worst)

It seems to be a runtime problem not bound with X
(btw I'm using XFree86 4.0.3 and 4.0.1)

Does anyone have the same problem ?

I would like to try fixing it but I need some insight on how
the objc runtime works (looking in the sources is insightful but
tedious).

Thanks


Marko Mikulicic


_______________________________________________
Bug-gnustep mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/bug-gnustep

Reply via email to