Folks:
	I have a custom (x86) motherboard that won't reboot unless a magic
	value is written into an on-board FPGA.  I've hacked cpu_reboot to
	DTRT on this machine, but was thinking about a more general way to
	solve this problem (preferrably one that didn't involve my having to
	hack in my patch every time I update my tree).

	It came to me that I could use the shutdown hooks, but only if I could
	pass 'howto' down to the shutdown routine (and possibly invoke the
	shutdown hooks more than once in the case of a howto == RB_HALT and
	the user then hitting 'enter' to reboot).

	So I came up with this little scheme, and would like some feedback
	on it:

		(1) When registering shutdown hooks, one could specify a
		    'when' parameter, which would just be a bitwise OR of
		    all the RB_xxx conditions we'd like to be called in.

		(2) doshutdownhooks() would be passed 'howto' as a parameter,
		    and only hooks where (when & howto) != 0 would be called.
		    It may also be good for howto to be passed through to the
		    hook function itself, for reasons given below.  

		(3) each shutdown hook would return a value to indicate 
		    whether it had completed its' work and should be removed
		    from the list, or whether it has done nothing/only 
		    partially done its' work/... and thus should be kept in
		    the list.

	Then, cpu_reboot would call 'doshutdownhooks(howto)' after it had
	completed dumps/sync/... and then call 'doshutdownhooks(RB_REBOOT)'
	[I'm not sure how RB_AUTOBOOT is used, but it wouldn't work here 
	 as (RB_AUTOBOOT & anything) == 0] if it fell into the reboot case.

Comments, ideas?
--rafal

----
Rafal Boni                                                   rafal@mediaone.net