>  I thought the problem below was long gone -- yet it just
>happened between a 1.5mumble and -current.

	this happens in some cases for me, in most cases not.  i guess
	there should be insufficient memory initialization (like missing bzero)
	or something.

itojun