On Mon, Dec 05 2005, Eli Zaretskii wrote:
>> From: Reiner Steib <[EMAIL PROTECTED]>
>> Date: Mon, 05 Dec 2005 14:35:08 +0100
>> Cc: [email protected]
>> On Sun, Dec 04 2005, Richard M. Stallman wrote:
[...]
>> > Can you please examine the arguments that were passed to
>> > XSetWMIconName? Are they all valid data?
[...]
> In what frame out of the 6 (#0 to #5) listed in the C backtrace did
> you issue these print commands?
I didn't issue any frame command, so it was frame #0, I think:
(gdb) frame
#0 0x0000002a97462fe2 in memcpy () from /lib64/tls/libc.so.6
> You need to issue them in the frame #5, the one that calls
> XSetWMIconName; the GDB command to get there from the place where
> the program stops due to the fatal signal is "frame 5" or "up 5".
(gdb) frame 5
#5 0x00000000004c874e in x_set_name_internal (f=0x3563680, name=Variable
"name" is not available.
)
at [...]/emacs/src/xfns.c:1657
1657 XSetWMIconName (FRAME_X_DISPLAY (f), FRAME_OUTER_WINDOW (f),
&icon);
(gdb) p f
$1 = 0x3563680
(gdb) p icon
$2 = {
value = 0x71ab178 <Address 0x71ab178 out of bounds>,
encoding = 31,
format = 8,
nitems = 9
}
(gdb) p &icon
$3 = (XTextProperty *) 0x7fbfffc570
> The backtrace seems to indicate that some of the arguments were
> optimized away by GCC and not saved anywhere where GDB can find them.
> So perhaps you will need to examine some other variables whose values
> can give evidence about the arguments. Alternatively (assuming you
> can reproduce the crash), modify xfns.c to introduce temporary global
> variables, declared `volatile' to avoid optimizations, and put the
> argument values into them just before XSetWMIconName is called. Then
> you can examine those temporary variables in GDB (after getting to the
> correct frame as explained above).
I can't reproduce the crash upto now. But I'll keep your suggestions
in mind.
Bye, Reiner.
--
,,,
(o o)
---ooO-(_)-Ooo--- | PGP key available | http://rsteib.home.pages.de/
_______________________________________________
emacs-pretest-bug mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/emacs-pretest-bug
