Distinguish target and "fake" PID values in extended-remote's "run", and when connecting in non-stop mode too.

[Ulrich Weigand]

Pedro Alves wrote:
> Here's the promised fix.  Non-stop's thread/inferior discovery on connection also
> needed to be handled.  I've made sure "info proc" would complain when the multiprocess
> extensions were off, with extended-remote/run, and also with plain remote + non-stop, and
> would operate on the currect process otherwise.  I've ran the testsuite against
> local gdbserver, and no regressions showed up.  I also ran the testsuite with
> multiprocess extensions off, and the only regressions were caused by core file
> generation no longer working, which was expected.
> 
> I've simplified the code that was previously put in a bit while factoring it
> out to the new add_current_inferior_and_thread function, avoiding the
> ptid_build ugliness.
> 
> Comments?

Looks good to me.

Thanks,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com