GE> yes, in fatal_error_signal() GE> Basically it calls system("pstack <pid>"). It is currently #if GE> 0'd -- Apparently Steve didn't like it. Steve: could you explain GE> why? G> I believe he felt it didn't work all the time or something. I've G> attached a test program that demonstrates the concept. If enough of G> you run it and don't have any problems, it's probably safe to remove G> the ifdefs (remember, though, that the pstack command did not appear G> until Solaris 2.5.1). I wrote the code that called pstack, originally. Observe that I was careful enough to check for the existence of the executable. Perhaps the problem is that XEmacs is in a possibly broken state, having just crashed, and so even things like calling access() is dangerous. It would have been nice if Steve indicated WHY this code is #if'ed out. # if 0 /* This is evil, rarely useful, and causes grief in some cases. */ /* Check for Sun-style stack printing via /proc */ { CONST char *pstack = "/usr/proc/bin/pstack"; if (access (pstack, X_OK) == 0) { char buf[100]; stderr_out ("\nC backtrace follows:\n" "(A real debugger may provide better information)\n\n"); sprintf (buf, "%s %d >&2", pstack, (int)getpid()); system (buf); } } # endif