[tin-bugs] tin: signal handler caught SIGSEGV signal (11)

[Urs Janßen]

In <mailman.91.1491733986.4716.tin-bugs at tin.org>, Olaf Schneider wrote:
> after updating some packages on my openSUSE tumbleweed system I get an
> error right after startup of tin:
> 
> rtin: signal handler caught SIGSEGV signal (11).
> rtin 2.4.1 20161224 ("Daill") [UNIX]: schicken Sie einen DETAILLIERTEN
> Fehlerbericht an tin-bugs at tin.org
> 
> I recompiled tin with debug option (td-conf.out and config.log attached)
> and run it via "tin -D 63" resulting in:
> 
> tin: signal handler caught SIGSEGV signal (11).
> tin 2.4.1 20161224 ("Daill") [UNIX]: schicken Sie einen DETAILLIERTEN
> Fehlerbericht an tin-bugs at tin.org Abgebrochen (Speicherabzug
> geschrieben)
> 
> The only written debug file is MALLOC (see attachment). Running gdb on
> the core dump tells:
> 
> Program terminated with signal SIGABRT, Aborted.
> #0  __GI_raise (sig=sig at entry=6)
> at ../sysdeps/unix/sysv/linux/raise.c:51
> 
> Any idea, what is going wrong there?

I recommentd to rebuild tin with CC and CFLAGS set and then run it
from inside gdb. e.g.:

~ > export CC=gcc
~ > export CFLAGS="-g -O0"
~ > ./configure [...]
~ > cd src
~ > make
~ > gdb ./tin

(gdb) run [tin cmd.-line flags if any]

and wait for the crash, you should be back in gdb then where you
could initiate a backtrace via

(gdb) bt

the output should be helpfull, esp. when you find the last frame
which belongs (instead of a library call ...), e.g.:

(gdb) bt
#0  0x00007ffff6c82ba0 in __read_nocancel ()
    at ../sysdeps/unix/syscall-template.S:81
#1  0x00007ffff7bb60e1 in read (__nbytes=1, __buf=0x7fffffffd22f, 
    __fd=<optimized out>) at /usr/include/x86_64-linux-gnu/bits/unistd.h:44
#2  fifo_push (sp=0x761210)
    at /build/ncurses-K1qcbH/ncurses-5.9+20140913/ncurses/base/lib_getch.c:308
#3  kgetch (sp=0x761210)
    at /build/ncurses-K1qcbH/ncurses-5.9+20140913/ncurses/base/lib_getch.c:709
#4  _nc_wgetch (win=win at entry=0x781d00, /result=result at entry=0x7fffffffd304, 
    use_meta=<optimized out>)
    at /build/ncurses-K1qcbH/ncurses-5.9+20140913/ncurses/base/lib_getch.c:540
#5  0x00007ffff7bb68f5 in wgetch (win=0x781d00)
    at /build/ncurses-K1qcbH/ncurses-5.9+20140913/ncurses/base/lib_getch.c:670
#6  0x000000000047d186 in ReadWch () at ./tcurses.c:579
#7  0x0000000000429a22 in handle_keypad (left_action=0x475db0 <select_left>, 
    right_action=0x475dc0 <select_right>, 
    mouse_action=0x429970 <global_mouse_action>, keys=...) at ./global.c:361
#8  0x00000000004772f0 in selection_page (
    start_groupnum=start_groupnum at entry=0, num_cmd_line_groups=0)
    at ./select.c:145
#9  0x00000000004053de in main (argc=<optimized out>, argv=<optimized out>)
    at ./main.c:424

(gdb) frame 7
#7  0x0000000000429a22 in handle_keypad (left_action=0x475db0 <select_left>, 
    right_action=0x475dc0 <select_right>, 
    mouse_action=0x429970 <global_mouse_action>, keys=...) at
./global.c:361
361             wint_t ch = ReadWch();

As MALLOC is short and only holds allocations from keymap.c the crash
seems to happen early in the startup pahse. From the logs at least
this looks a bit fishy:

checking for domain-name... /etc/NNTP_INEWS_DOMAIN
checking for NNTP default-server... news./etc/NNTP_INEWS_DOMAIN