Martin Buchholz <martin(a)xemacs.org&gt; writes: In that case wouldn't it be possible to do that check (and maybe even the sprintf) unconditionally early on in main() and set some variable like pstack_available_p. That would only leave the system() in the signal handler. Of course this variable could potentially be overwritten by the crash but that's probably rather unlikely to happen. What happens if pstack finds an utterly corrupted stack. Can it crash or (worse) hang? yes, indeed. FWIW the ChangeLog entry says: Wed Mar 12 14:29:40 1997 Steven L Baur <steve(a)altair.xemacs.org&gt; * emacs.c (fatal_error_signal): (mostly) useless check on /usr/proc/bin/pstack removed. Not much better, is it? :-) The only problem with these pstack traces is that they might discourage people from obtaining a real backtrace with their debugger from the core, but that's a risk I'm willing to take. I see comments like 'I couldn't find the core file' far too often and many people don't even bother reading our crash message but just send it verbatim to us. The more useful info we put in there the better. Gunnar -- Gunnar Evermann Speech, Vision & Robotics Group Engineering Department Cambridge University