;
;   Life in 68020!  Three passes.
;
	public	_pass1
	public	_pass2
	public	_pass3
	public	_pass3t
;
;   First pass:  Calculates t+(t<<1)+(t>>1) for a huge bit string.
;
;   Arguments:  long *srca, *dsta, *dstb, len
;
;      srca is the source.
;      dsta is where the low-order destination is written
;      dstb is where the high-order destination is written
;      len is the total number of longwords needed
;
_pass1:
	movem.l	d1/d2/d3/d4/d5/d7/a0/a1/a2,-(sp)
	move.l	40(a7),a0
	move.l	44(a7),a1
	move.l	48(a7),a2
	move.l	52(a7),d7
	move.l	#0,d5
	bra	pass1in
in2:
	swap	d7
inagain:
	move.l	(a0)+,d0
	bne	inover
	lsr.b	#1,d5
	roxr.l	#1,d0
	move.l	#0,d5
	move.b	(a0),d3
	add.b	d3,d3
	addx.w	d5,d0
	move.l	d0,(a1)+
	move.l	d5,(a2)+
	dbra	d7,inagain
	swap	d7
	dbra	d7,in2
	bra	inov2
inover:
	move.l	d0,d1
	move.l	d0,d2
	lsr.b	#1,d5
	roxr.l	#1,d1
	move.l	d0,d5
	move.b	(a0),d3
	add.b	d3,d3
	addx.l	d2,d2
	move.l	d0,d3
	eor.l	d1,d0
	and.l	d1,d3
	move.l	d2,d4
	and.l	d0,d4
	eor.l	d2,d0
	eor.l	d4,d3
	move.l	d0,(a1)+
	move.l	d3,(a2)+
pass1in:
	dbra	d7,inagain
	swap	d7
	dbra	d7,in2
inov2:
	movem.l	(sp)+,d1/d2/d3/d4/d5/d7/a0/a1/a2
	move.l	#0,d0
	rts
;
;   This second pass does the main work.  It's also the main time sink.
;
;   Arguments:  long *srca, *srcb, *srcc, *dsta, mod, len
;
;      srca = main source.
;      srcb = low order source from above
;      srcc = high order source from above
;      mod = byte offset to next/previous row
;      len = number of longwords to do (again)
;
_pass2:
	movem.l	d1/d2/d3/d4/d5/d6/d7/a0/a1/a2/a3/a4/a5/a6,-(sp)
	move.l	60(a7),a0
	move.l	64(a7),a2
	move.l	68(a7),a3
	move.l	72(a7),d0
	move.l	76(a7),d7
	move.l	d0,d6
	lea	(a2,d6.l),a4
	lea	(a3,d6.l),a1
	neg.l	d0
	lea	(a2,d0.l),a5
	lea	(a3,d0.l),a6
	bra	pass2in
in3:
	swap	d7
in4:
	move.l	(a2)+,d0
	move.l	d0,d1
	move.l	(a5)+,d2
	eor.l	d2,d0
	and.l	d2,d1
	move.l	(a4)+,d2
	move.l	d2,d3
	and.l	d0,d3
	eor.l	d2,d0
	or.l	d3,d1
	move.l	(a1)+,d2
	move.l	d2,d3
	and.l	d1,d2
	eor.l	d3,d1
	move.l	(a6)+,d3
	move.l	d3,d4
	and.l	d1,d3
	eor.l	d4,d1
	eor.l	d3,d2
	move.l	(a0),d5
	beq	in4q
	move.l	(a3)+,d3
	move.l	d3,d4
	and.l	d1,d3
	eor.l	d4,d1
	eor.l	d3,d2
	not.l	d0
	eor.l	d1,d0
	eor.l	d1,d2
	or.l	d5,d1
	and.l	d1,d0
	and.l	d2,d0
	move.l	d0,(a0)+
pass2in:
	dbra	d7,in4
	swap	d7
	dbra	d7,in3
out4q:
	movem.l	(sp)+,d1/d2/d3/d4/d5/d6/d7/a0/a1/a2/a3/a4/a5/a6
	move.l	#0,d0
	rts
;
;   A minor optimization.  If the current long-word is zero, we can make some
;   simplifications.
;
in4q:
	not.l	d2
	and.l	d2,d0
	and.l	d1,d0
	move.l	d0,(a0)+
	add.w	#4,a3
	dbra	d7,in4
	swap	d7
	dbra	d7,in3
	bra	out4q
;
;   Third pass.  Clears the perimeter.
;
;   Arguments:  long *dst, width, height
;
;      dst = destination (and source) to `clear'
;      width = width of screen in bytes (must be a multiple of 4)
;      height = height of screen in pixels
;
_pass3:
	movem.l	d1/d2/d7/a0,-(sp)
	move.l	20(a7),a0
	move.l	#0,d0
	move.l	24(a7),d7
	lsr.w	#2,d7
	bra	loop5
in5:
	move.l	d0,(a0)+
loop5:
	dbra	d7,in5
	sub.w	#2,a0
	move.l	28(a7),d7
	sub.l	#1,d7
	move.l	24(a7),d2
	move.l	#$fffe7fff,d1
	bra	loop6
in6:
	and.l	d1,(a0)
	add.w	d2,a0
loop6:
	dbra	d7,in6
	sub.w	d2,a0
	add.w	#2,a0
	move.l	24(a7),d7
	lsr.w	#2,d7
	bra	loop7
in7:
	move.l	d0,(a0)+
loop7:
	dbra	d7,in7
	movem.l	(sp)+,d1/d2/d7/a0
	move.l	#0,d0
	rts
;
;   Third pass.  Clears the perimeter, doing a torus wrap.
;
;   Arguments:  long *dst, width, height
;
;      dst = destination (and source) to `clear'
;      width = width of screen in bytes (must be a multiple of 4)
;      height = height of screen in pixels
;
_pass3t:
	movem.l	d1/d2/d7/a0/a1,-(sp)
	move.l	24(a7),a0
	move.l	#0,d0
	move.l	32(a7),d2
	sub.l	#2,d2
	mulu.w	30(a7),d2
	lea	(a0,d2.l),a1
	move.l	28(a7),d7
	lsr.w	#2,d7
	bra	loop5t
in5t:
	move.l	(a1)+,(a0)+
loop5t:
	dbra	d7,in5t
	move.l	28(a7),d7
	lsr.w	#2,d7
	bra	loop5u
in5u:
	move.l	(a0)+,(a1)+
loop5u:
	dbra	d7,in5u
	move.l	24(a7),a0
	move.l	28(a7),d0
	lea	-1(a0,d0.l),a1
	move.l	32(a7),d7
	bra	int7t
int6t:
	btst	#6,(a0)
	beq	fb1
	or.b	#1,(a1)
	btst	#1,(a1)
	beq	fb2
fb4:
	or.b	#$80,(a0)
	bra	fb3
fb1:
	and.b	#$fe,(a1)
	btst	#1,(a1)
	bne	fb4
fb2:
	and.b	#$7f,(a0)
fb3:
	add.w	d0,a0
	add.w	d0,a1
int7t:
	dbra	d7,int6t
	movem.l	(sp)+,d1/d2/d7/a0/a1
	move.l	#0,d0
	rts