rough-draft VKERNEL host-initiated shutdown patch

[Matthew Dillon]

:OK.. Updated / cleaned this against -HEAD (as of this morning) and
:things appear to be working.
:
:(vk still too fast to catch the 'no init yet' case)

    This looks quite nice.  I just realized though that it isn't safe
    to run reboot() (or probably ksignal either) from the idle halt
    code, so scratch that part.

    We have to find a good place to put this.  I think what we may have
    to do is create a kernel thread whos only job is to shut the kernel
    down, and then check the mailbox from inside signalmailbox() and have
    it wake the special kernel thread up.

    Would you like to have a go at that or would you like me to do it?

						-Matt