On Wed, 6 May 1998, Mark A. Heilpern wrote:
> if (fork()) force_warmboot();
> else my_signal(SIGSEGV, SIG_IGN);
>
> I end up with a more accurate core file -- i.e., I'm right where the crash
> was caused.
> (I know this because I've got an "admin crash" command that tries to to a
> send_to_char
> to NULL.)
In addition to this, I run a script that analyzes the core after 30
seconds and mails me the results. BTW, if you have more than SIGSEGV
handled by the functrion, you might want to ignore the actual signal, not
SIGSEGV.
void nasty_signal_handler (int no)
{
signal (no, SIG_DFL);
write_last_command();
disable_reaping();
if (!fork())
{
// Close all files!
// RACE CONDITION: if the exec completes before this part is reached,
// we'll get bind errors
for (int i = 255; i >= 0; i--)
close(i);
// Dup /dev/null to STD{IN,OUT,ERR}
open("/dev/null", O_RDWR);
dup(0);
dup(0);
if (!fork())
{
execl(CORE_EXAMINE_SCRIPT, CORE_EXAMINE_SCRIPT, (char*)NULL);
perror("execl");
_exit(EXIT_FAILURE);
}
else
return; // and dump core!
}
perform_copyover("", false);
_exit(0);
}
And the script:
#!/bin/sh
CORE=core
PEOPLE=erwin@localhost
COUNT=`cat .crashcount`
COUNT=`expr 1 + ${COUNT}`
echo ${COUNT} > .crashcount
sleep 30
gdb -batch ../src/envy ${CORE} | head -1000 | mailx -s "[AR] She's dead, Jim! (#${COUNT})" $PEOPLE
rm ${CORE}.2
mv ${CORE}.1 ${CORE}.2
mv ${CORE} ${CORE}.1
=============================================================================
Erwin Andreasen Herlev, Denmark <erwin@pip.dknet.dk> UNIX System Programmer
<URL:http://www.abandoned.org/drylock/> <*> (not speaking for) DDE
=============================================================================
+------------------------------------------------------------+
| Ensure that you have read the CircleMUD Mailing List FAQ: |
| http://democracy.queensu.ca/~fletcher/Circle/list-faq.html |
+------------------------------------------------------------+
This archive was generated by hypermail 2b30 : 12/15/00 PST