Re: [RFC] remote: semantics of 'k' (kill) message

[Michael Snyder <msnyder@redhat.com>]

Andrew Cagney wrote:
> 
> > Andrew, you recently added this comment:
> >
> > ! FIXME: @emph{There is no description of how to operate when a specific
> > ! thread context has been selected (ie.@: does 'k' kill only that thread?)}.
> >
> > Maybe with a little discussion we can resolve this?
> > I believe the 'k' message is only sent in one context:
> > when the user asks gdb to kill the inferior process.
> > On a native system, that is clearly interpreted as meaning
> > to kill all of the threads.  Is there any reason why we
> > should not agree that it means the same thing on an
> > embedded target?
> 
> Hmm, yes.  You're right.  I shouldn't be trying to specify ``future
> behavour'' in the protocol.  Rather it should just be specifying things
> based on GDB's existing behavour on a well implemented native system.


Well, we might conceivably want to be able to kill 
a specified thread or process on an embedded system --
but at present we can't do that on a native system either.