[PATCH 0/2] Demangler crash handler

[Tom Tromey]

>>>>> "Gary" == Gary Benson <gbenson@redhat.com> writes:

Gary>  char *
Gary>  gdb_demangle (const char *name, int options)
Gary>  {
[...]
Gary> +  sigaction (SIGSEGV, &sa, &old_sa);
[...]
Gary> +  crash_signal = SIGSETJMP (gdb_demangle_jmp_buf);
[...]
Gary> +  sigaction (SIGSEGV, &old_sa, NULL);

This adds two calls to sigaction and a call to sigsetjmp to every
demangling invocation.

I wonder whether the performance cost of this is noticeable; and if so,
how large the effect is.

If it is too large we could perhaps arrange to do the sigaction calls
just once and see if that helps.

Tom