Woof! On Mon, Jan 24, 2011 at 9:46 PM, Douglas Hubler <[email protected]> wrote:
> >So maybe this is better: > > struct sigaction sig_actions; > sig_actions.sa_handler = shutdown_handler; > sigemptyset(&sig_actions.sa_mask); > sig_actions.sa_flags = 0; > int register_sig_action = sigaction(SIGTERM, &sig_actions, NULL); > I would really like to use a debugger but also understand why sipXecs > daemons are so different than any others. It is a conundrum. It's been that way for a long time, and I never did figure it out (the debug issues preceded the signal handling thread design, BTW). For non IMDB using code, this is probably fine. But the IMDB code isn't re-entrant, and so if SIGTERM is caught while in IMDB, one cannot then call the IMDB cleanup code without possible locking issues. That was the impetus for the new signal handling thread model, to make sure IMDB threads could NOT be the one that ended up calling the shutdown_handler. I believe one can tell gdb to ignore certain signals, and perhaps that is something to explore. I used to have a gcc 'rc' file set stuff up and it was usable. I don't have them anymore and don't recall exactly, but it was something like "handle <signal> ignore" (or was it nostop?) --Woof!
_______________________________________________ sipx-dev mailing list [email protected] List Archive: http://list.sipfoundry.org/archive/sipx-dev/
