Re: Crashes with latest gtk+2.0-2.2.2 package
On Tue, Jul 22, 2003 at 01:59:36AM +0900, Akira TAGOH wrote:
>
>
> AH> $ gdb /opt/gnome/bin/eog core
> AH> [ ... gdb starts ... ]
> AH> (gdb) bt 8
> AH> #0 0x4088ace1 in kill () from /lib/libc.so.6
> AH> #1 0x4066ccc1 in pthread_kill () from /lib/libpthread.so.0
> AH> #2 0x4066cf95 in raise () from /lib/libpthread.so.0
> AH> #3 0x4088aac6 in raise () from /lib/libc.so.6
> AH> #4 0x4088bced in abort () from /lib/libc.so.6
> AH> #5 0x408cccef in _IO_file_xsputn () from /lib/libc.so.6
> AH> #6 0x408cddf1 in free () from /lib/libc.so.6
> AH> #7 0x40822ed2 in g_free (mem=0x0) at gmem.c:186
> AH> #8 0x4083ad04 in g_utf8_collate_key (str=0x0, len=-1) at gunicollate.c:211
> AH> (gdb) frame 8
> AH> #8 0x4083ad04 in g_utf8_collate_key (str=0x0, len=-1) at gunicollate.c:211
> AH> 211 g_free (result_wc);
>
> Well, if that backtrace is correct, the actual problem is
> why g_utf8_collate_key() is given NULL, I think. perhaps it
> might be already broken due to the memory corruption, though.
>
I've added some printf() statements and this is what I see ...
$ eog
Entering g_utf8_collate_key
str_norm: '.'
xrfm_len: 1
sizeof(gunichar): 4
sizeof(wchar_t): 4
result_wc: 0x81669b0
result_len: 0
i: 0
result_len: 1
result: 0x81669d8
result_len: 0
i: 0
result_len: 1
Aborted (core dumped)
$ gdb /opt/gnome/bin/eog core
(gdb) bt 10
#0 0x4088ace1 in kill () from /lib/libc.so.6
#1 0x4066ccc1 in pthread_kill () from /lib/libpthread.so.0
#2 0x4066cf95 in raise () from /lib/libpthread.so.0
#3 0x4088aac6 in raise () from /lib/libc.so.6
#4 0x4088bced in abort () from /lib/libc.so.6
#5 0x408cccef in _IO_file_xsputn () from /lib/libc.so.6
#6 0x408cddf1 in free () from /lib/libc.so.6
#7 0x40822f02 in g_free (mem=0x0) at gmem.c:186
#8 0x4083ae22 in g_utf8_collate_key (str=0x81669d8 "", len=-1) at gunicollate.c:231
#9 0x404021d6 in gtk_file_selection_get_selections () from /usr/lib/libgtk-x11-2.0.so.0
(More stack frames follow...)
(gdb)
Notice this time that the g_utf8_collate_key() has a non-null string. It
seems like this bug isn't always producing identical core dumps ...
>
> Hmm, but it's a workaround, and it happened glibc 2.3.2
> might be buggy, doesn't it?
>
Possibly. Maybe once more libc-2.3.2 packages are in use more people
will see this and pinpoint it.
Art Haas
--
Man once surrendering his reason, has no remaining guard against absurdities
the most monstrous, and like a ship without rudder, is the sport of every wind.
-Thomas Jefferson to James Smith, 1822
Reply to: