Robert_Pluim(a)BayNetworks.COM (Robert Pluim) writes:
Just hit C-h d in a gnus article buffer, and got a core dump.
Anyone have any suggestions for how to debug this?
Hi Robert,
first of all the most important question of debugging: Can you
reproduce it, i.e. can you give a sequence of actions which always
produces this crash?
I had a look at the mailing-list archive and found four crash-reports
involving skip-syntax-backward all on Solaris with gcc. I have not yet
managed to crash my SUNcc-compiled xemacs but I will recompile with
gcc.
In one of the messages (ID: 9806/msg00909) there were instructions for
crashing:
----------------------------------------------------------------------
Hi,
xemacs b46 was core-dumped when doing following operations:
% xemacs -q
C-x C-f ~/test.c RET (open new file in cc-mode)
M-; M-; (press indent-for-comment twice)
----------------------------------------------------------------------
Does this work (i.e. crash) for you? The proposed solution was:
----------------------------------------------------------------------
I found that
this is compiler optimization problem (egcs-1.0.2 on sparc solaris-2.5.1).
gcc -g -O search.c ... ---> no problem
gcc -g -O2 search.c ... ---> core-dumped by above operations.
S.Senda
----------------------------------------------------------------------
This might solve the problem for now, but I still think we should
track it down a bit more.
If you still have the original coredump, please go up to frame #3, do
'print *buffer' and mail the result.
(gdb) bt
#0 0xef1f471c in kill ()
#1 0x70ac8 in fatal_error_signal (sig=11) at emacs.c:262
#2 <signal handler called>
#3 skip_chars (buf=0x69b200, forwardp=0, syntaxp=1, string=2062122, lim=1)
at buffer.h:643
#4 0x11483c in Fskip_syntax_backward (syntax=1612674844, lim=-1608028156,
buffer=-1608028156) at search.c:941
#5 0x755d8 in funcall_recording_as (recorded_as=-1607919668, nargs=1,
args=0xefffe8c8) at eval.c:3193
I also had a look at the diffs against previous versions (as Steve
suggested) and to me they all looked kind of innocent but if you have
the time you might still try an older version...
Gunnar
--
Gunnar Evermann
Natural Language Systems Group, University of Hamburg
EMail: Gunnar.Evermann(a)nats.informatik.uni-hamburg.de