> > It's a further oddity that you can single-step (or not) "into" the > > system call and then get a ptrace stop "inside" it, that being for > > PTRACE_EVENT_FORK et al. > > And utrace-ptrace should be compatible here, yes?
As far as the sequence of stops that a ptracer observes, the general answer is always yes. If something is strange enough we can discuss it and might decide it's a bug in the vanilla ptrace semantics. In that event, we would either change it upstream first, or else be very clear about the intended change without bug-compatibility when we submit utrace-ptrace. Thanks, Roland
