; Chunky2Planar algorithm, originally by James McCoull
; Modified by Peter McGavin for variable size and depth
; and "dirty list" (hope I didn't slow it down too much)
;
; 	Cpu only solution
;	Originally optimised for 020+fastram
;	Tried to optimise for 68000

;void __asm c2p_8 (register __a0 UBYTE *chunky_data,
;                  register __a1 PLANEPTR raster,
;                  register __a2 UBYTE *dirty_list,
;                  register __d1 ULONG plsiz,

; a0 -> width*height chunky pixels in fastmem
; a1 -> contiguous bitplanes in chipmem
; a2 -> dirty list (1-byte flag for whether each 32 pixel "unit" needs updating)
; d1 = width*height/8   (width*height must be a multiple of 32)

	ifeq	depth-8
		xdef	_c2p_8_000
_c2p_8_000:
	else
	ifeq	depth-6
		xdef	_c2p_6_000
_c2p_6_000:
	else
	ifeq	depth-4
		xdef	_c2p_4_000
_c2p_4_000:
	else
		fail	"unsupported depth!"
	endc
	endc
	endc


wordmerge	macro		;    i1 i2 tmp
				;    \1 \2 \3
		move.l	\2,\3	;  4 \3 = CD
		move.w	\1,\2	;  4 \2 = CB
		swap	\2	;  4 \2 = BC
		move.w	\2,\1	;  4 \1 = AC
		move.w	\3,\2	;  4 \2 = BD
		endm		; 20


merge		macro		;    io  in out tmp msk sft
				;    \1  \2  \3  \4  \5  \6
				;    \1 = abqr
				;    \2 = ijyz
		move.l	\5,\3	;  4 \3 = 0x0x
		move.l	\3,\4	;  4 \4 = 0x0x
		and.l	\1,\3	;  8 \3 = 0b0r
		and.l	\2,\4	;  8 \4 = 0j0z
		eor.l	\3,\1	;  8 \1 = a0q0
		eor.l	\4,\2	;  8 \2 = i0y0
		lsr.l	#\6,\2	; 8+2*sft  \2 = 0i0y
	ifeq \6-1
		add.l	\3,\3	;  8
	else
		lsl.l	#\6,\3	; 8+2*sft  \3 = b0r0
	endc
		or.l	\2,\1	;  8 \1 = aiqy
		or.l	\4,\3	;  8 \3 = bjrz
		endm		; 72+4*sft

merge4		macro		; io  in out tmp msk
	ifgt depth-4
		merge	\1,\2,\3,\4,\5,4
	else
		move.l	\5,\3	; this version returns only 1 result
		and.l	\3,\2	; \2 = 0j0z
		and.l	\1,\3	; \3 = 0b0r
		lsl.l	#4,\3	; \3 = b0r0
		or.l	\2,\3	; \3 = bjrz
	endc
		endm


; the real c2p routine starts here

		movem.l	d2-d7/a2-a6,-(sp)

		sub.w	#24,sp		; space for temporary variables

; a0 = chunky buffer
; a1 = output area
; a2 = dirty list
; d1 = plsiz

		move.l	d1,(4,sp)	; save plsiz
		lsl.l	#3,d1		; 8*plsiz
		move.l	a0,(sp)
		add.l	d1,(sp)		; (sp) -> end of chunky buffer
		sub.l	(4,sp),d1
	ifle depth-6
		sub.l	(4,sp),d1
		sub.l	(4,sp),d1
	endc
	ifle depth-4
		sub.l	(4,sp),d1
		sub.l	(4,sp),d1
	endc
		move.l	d1,(8,sp)	; save 7*plsiz (or 5*plsiz) (or 3*plsiz)

		adda.l	d1,a1		; a1 -> plane7 (or plane5) (or plane3)

		move.l	#$00ff00ff,a3	; load byte merge mask
		move.l	#$0f0f0f0f,a4	; load nibble merge mask

; pass 1
initloop:	tst.b	(a2)+		; does next 32 pixel unit need updating?
		bne.b	c2p		; branch if yes

		adda.w	#32,a0		; skip 32 pixels on input
		addq.l	#4,a1		; skip 32 pixels on output

		cmpa.l	(sp),a0
		bne.b	initloop
		bra.w	exit		; exit if no changes

; this becomes the main loop after the first change is found
mainloop:	tst.b	(a2)+		;  8 does next 32 pixel unit need updating?
		bne.b	c2p		; 10

		adda.w	#32,a0		; skip 32 pixels on input
		addq.l	#4,a1		; skip 32 pixels on output

		cmpa.l	(sp),a0
		bne.b	mainloop
		bra.w	exit

; convert next 32 pixels and return to mainloop
c2p:		movem.l	(a0)+,d0-d7	; 76 get 32 pixels in registers

; d0 = a7a6a5a4a3a2a1a0 b7b6b5b4b3b2b1b0 c7c6c5c4c3c2c1c0 d7d6d5d4d3d2d1d0
; d1 = e7e6e5e4e3e2e1e0 f7f6f5f4f3f2f1f0 g7g6g5g4g3g2g1g0 h7h6h5h4h3h2h1h0
; d2 = i7i6i5i4i3i2i1i0 j7j6j5j4j3j2j1j0 k7k6k5k4k3k2k1k0 l7l6l5l4l3l2l1l0
; d3 = m7m6m5m4m3m2m1m0 n7n6n5n4n3n2n1n0 o7o6o5o4o3o2o1o0 p7p6p5p4p3p2p1p0
; d4 = q7q6q5q4q3q2q1q0 r7r6r5r4r3r2r1r0 s7s6s5s4s3s2s1s0 t7t6t5t4t3t2t1t0
; d5 = u7u6u5u4u3u2u1u0 v7v6v5v4v3v2v1v0 w7w6w5w4w3w2w1w0 x7x6x5x4x3x2x1x0
; d6 = y7y6y5y4y3y2y1y0 z7z6z5z4z3z2z1z0 A7A6A5A4A3A2A1A0 B7B6B5B4B3B2B1B0
; d7 = C7C6C5C4C3C2C1C0 D7D6D5D4D3D2D1D0 E7E6E5E4E3E2E1E0 F7F6F5F4F3F2F1F0

		wordmerge d0,d4,a6	; 20 d0/4 = abqr cdst
		wordmerge d1,d5,a6	; 20 d1/5 = efuv ghwx
		wordmerge d2,d6,a6	; 20 d2/6 = ijyz klAB
		wordmerge d3,d7,a6	; 20 d3/7 = mnCD opEF

; d0 = a7a6a5a4a3a2a1a0 b7b6b5b4b3b2b1b0 q7q6q5q4q3q2q1q0 r7r6r5r4r3r2r1r0
; d1 = e7e6e5e4e3e2e1e0 f7f6f5f4f3f2f1f0 u7u6u5u4u3u2u1u0 v7v6v5v4v3v2v1v0
; d2 = i7i6i5i4i3i2i1i0 j7j6j5j4j3j2j1j0 y7y6y5y4y3y2y1y0 z7z6z5z4z3z2z1z0
; d3 = m7m6m5m4m3m2m1m0 n7n6n5n4n3n2n1n0 C7C6C5C4C3C2C1C0 D7D6D5D4D3D2D1D0
; d4 = c7c6c5c4c3c2c1c0 d7d6d5d4d3d2d1d0 s7s6s5s4s3s2s1s0 t7t6t5t4t3t2t1t0
; d5 = g7g6g5g4g3g2g1g0 h7h6h5h4h3h2h1h0 w7w6w5w4w3w2w1w0 x7x6x5x4x3x2x1x0
; d6 = k7k6k5k4k3k2k1k0 l7l6l5l4l3l2l1l0 A7A6A5A4A3A2A1A0 B7B6B5B4B3B2B1B0
; d7 = o7o6o5o4o3o2o1o0 p7p6p5p4p3p2p1p0 E7E6E5E4E3E2E1E0 F7F6F5F4F3F2F1F0

; temporarily save off some registers
		movea.l	d6,a6		;  4
		movea.l	d7,a5		;  4

; set up pass 2 & 3 masks
		move.l	#$00ff00ff,a3	;  8
		move.l	#$0f0f0f0f,a4	;  8

; pass 2
		merge	d0,d2,d6,d7,a3,8 ;104 d0/d6 = aiqy bjrz
		merge	d1,d3,d7,d2,a3,8 ;104 d1/d7 = emuc fnvD

; d0 = a7a6a5a4a3a2a1a0 i7i6i5i4i3i2i1i0 q7q6q5q4q3q2q1q0 y7y6y5y4y3y2y1y0
; d6 = b7b6b5b4b3b2b1b0 j7j6j5j4j3j2j1j0 r7r6r5r4r3r2r1r0 z7z6z5z4z3z2z1z0
; d1 = e7e6e5e4e3e2e1e0 m7m6m5m4m3m2m1m0 u7u6u5u4u3u2u1u0 C7C6C5C4C3C2C1C0
; d7 = f7f6f5f4f3f2f1f0 n7n6n5n4n3n2n1n0 v7v6v5v4v3v2v1v0 D7D6D5D4D3D2D1D0

; pass 3
		merge4	d0,d1,d2,d3,a4,4 ;88 d0/d2  = ae74... ae30...
		merge4	d6,d7,d3,d1,a4,4 ;88 d6/d3  = bf74... bf30...

; d0 = a7a6a5a4e7e6e5e4 i7i6i5i4m7m6m5m4 q7q6q5q4u7u6u5u4 y7y6y5y4C7C6C5C4
; d2 = a3a2a1a0e3e2e1e0 i3i2i1i0m3m2m1m0 q3q2q1q0u3u2u1u0 y3y2y1y0C3C2C1C0
; d6 = b7b6b5b4f7f6f5f4 j7j6j5j4n7n6n5n4 r7r6r5r4v7v6v5v4 z7z6z5z4D7D6D5D4
; d3 = b3b2b1b0f3f2f1f0 j3j2j1j0n3n2n1n0 r3r2r1r0v3v2v1v0 z3z2z1z0D3D2D1D0

; exchange them
		exg	a6,d2		;  4
		exg	a5,d3		;  4

; d2 = k7k6k5k4k3k2k1k0 l7l6l5l4l3l2l1l0 A7A6A5A4A3A2A1A0 B7B6B5B4B3B2B1B0
; d3 = o7o6o5o4o3o2o1o0 p7p6p5p4p3p2p1p0 E7E6E5E4E3E2E1E0 F7F6F5F4F3F2F1F0

; pass 2
		merge	d4,d2,d7,d1,a3,8 ;104 d4/d7 = cksA dltB
		merge	d5,d3,d1,d2,a3,8 ;104 d5/d1 = gowE hpxF

; d4 = c7c6c5c4c3c2c1c0 k7k6k5k4k3k2k1k0 s7s6s5s4s3s2s1s0 A7A6A5A4A3A2A1A0
; d7 = d7d6d5d4d3d2d1d0 l7l6l5l4l3l2l1l0 t7t6t5t4t3t2t1t0 B7B6B5B4B3B2B1B0
; d5 = g7g6g5g4g3g2g1g0 o7o6o5o4o3o2o1o0 w7w6w5w4w3w2w1w0 E7E6E5E4E3E2E1E0
; d1 = h7h6h5h4h3h2h1h0 p7p6p5p4p3p2p1p0 x7x6x5x4x3x2x1x0 F7F6F5F4F3F2F1F0

; pass 3
		merge4	d4,d5,d3,d2,a4,4 ;88 d4/d3 = cg74.. cg30..
		merge4	d7,d1,d5,d2,a4,4 ;88 d7/d5 = dh74.. dh30..

; d4 = c7c6c5c4g7g6g5g4 k7k6k5k4o7o6o5o4 s7s6s5s4w7w6w5w4 A7A6A5A4E7E6E5E4
; d3 = c3c2c1c0g3g2g1g0 k3k2k1k0o3o2o1o0 s3s2s1s0w3w2w1w0 A3A2A1A0E3E2E1E0
; d7 = d7d6d5d4h7h6h5h4 l7l6l5l4p7p6p5p4 t7t6t5t4x7x6x5x4 B7B6B5B4F7F6F5F4
; d5 = d3d2d1d0h3h2h1h0 l3l2l1l0p3p2p1p0 t3t2t1t0x3x2x1x0 B3B2B1B0F3F2F1F0

; set up pass 4 & 5 masks
		move.l	#$33333333,a3	;  8
		move.l	#$55555555,a4	;  8

	ifgt depth-4
; pass 4
		merge	d0,d4,d1,d2,a3,2 ;80 d0/d1 = aceg76.. aceg54..
		merge	d6,d7,d4,d2,a3,2 ;80 d6/d4 = bdhf76.. bdhf54..

; d0 = a7a6c7c6e7e6g7g6 i7i6k7k6m7m6o7o6 q7q6s7s6u7u6w7w6 y7y6A7A6C7C6E7E6
; d1 = a5a4c5c4g5g4e5e4 i5i4k5k4m5m4o5o4 q5q4s5s4u5u4w5w4 y5y4A5A4C5C4E5E4
; d6 = b7b6d7d6f7f6h7h6 j7j6l7l6n7n6p7p6 r7r6t7t6v7v6x7x6 z7z6B7B6D7D6F7F6
; d4 = b5b4d5d4h5h4f5f4 j5j4l5l4n5n4p5p4 r5r4t5t4v5v4x5x4 z5z4B5B4D5D4F5F4

; pass 5
	ifgt depth-6
		merge	d0,d6,d7,d2,a4,1 ;74 d0/d7 = abcd7... abcd6...
	endc
		merge	d1,d4,d6,d2,a4,1 ;74 d1/d6 = abcd5... abcd4...

; d0 = a7b7c7d7e7f7g7h7 i7j7k7l7m7n7o7p7 q7r7s7t7u7v7w7x7 y7z7A7B7C7D7E7F7
; d7 = a6b6c6d6e6f6g6h6 i6j6k6l6m6n6o6p6 q6r6s6t6u6v6w6x6 y6z6A6B6C6D6E6F6
; d1 = a5b5c5d5g5h5e5f5 i5j5k5l5m5n5o5p5 q5r5s5t5u5v5w5x5 y5z5A5B5C5D5E5F5
; d6 = a4b4c4d4g4h4e4f4 i4j4k4l4m4n4o4p4 q4r4s4t4u4v4w4x4 y4z4A4B4C4D4E4F4

	endc

		move.l	(4,sp),d2	; 12 d2 = plsiz

	ifgt depth-4
	ifgt depth-6
		move.l	d0,(a1)		;  8 plane7
		suba.l	d2,a1		;  8 -plsiz
		move.l	d7,(a1)		;  8 plane6
		suba.l	d2,a1		;  8 -plsiz
	endc
		move.l	d1,(a1)		;  8 plane5
		suba.l	d2,a1		;  8 -plsiz
		move.l	d6,(a1)		;  8 plane4
		suba.l	d2,a1		;  8 -plsiz
	endc

; restore them
		move.l	a6,d1		;  4
		move.l	a5,d7		;  4

; d1 = a3a2a1a0e3e2e1e0 i3i2i1i0m3m2m1m0 q3q2q1q0u3u2u1u0 y3y2y1y0C3C2C1C0
; d7 = b3b2b1b0f3f2f1f0 j3j2j1j0n3n2n1n0 r3r2r1r0v3v2v1v0 z3z2z1z0D3D2D1D0
; d3 = c3c2c1c0g3g2g1g0 k3k2k1k0o3o2o1o0 s3s2s1s0w3w2w1w0 A3A2A1A0E3E2E1E0
; d5 = d3d2d1d0h3h2h1h0 l3l2l1l0p3p2p1p0 t3t2t1t0x3x2x1x0 B3B2B1B0F3F2F1F0

; pass 4
		merge	d1,d3,d6,d0,a3,2 ;80 d1/d6 = aceg32.. aceg10..
		merge	d7,d5,d0,d3,a3,2 ;80 d7/d0 = bdhf32.. bdhf10..

; d1 = a3a2c3c2e3e2g3g2 i3i2k3k2m3m2o3o2 q3q2s3s2u3u2w3w2 y3y2A3A2C3C2E3E2
; d6 = a1a0c1c0g1g0e1e0 i1i0k1k0m1m0o1o0 q1q0s1s0u1u0w1w0 y1y0A1A0C1C0E1E0
; d7 = b3b2d3d2f3f2h3h2 j3j2l3l2n3n2p3p2 r3r2t3t2v3v2x3x2 z3z2B3B2D3D2F3F2
; d0 = b1b0d1d0h1h0f1f0 j1j0l1l0n1n0p1p0 r1r0t1t0v1v0x1x0 z1z0B1B0D1D0F1F0

; pass 5
		merge	d1,d7,d5,d3,a4,1 ;74 d1/d5 = abcd3... abcd2...
		merge	d6,d0,d7,d3,a4,1 ;74 d6/d7 = abcd1... abcd0...

; d1 = a3b3c3d3e3f3g3h3 i3j3k3l3m3n3o3p3 q3r3s3t3u3v3w3x3 y3z3A3B3C3D3E3F3
; d5 = a2b2c2d2e2f2g2h2 i2j2k2l2m2n2o2p2 q2r2s2t2u2v2w2x2 y2z2A2B2C2D2E2F2
; d6 = a1b1c1d1g1h1e1f1 i1j1k1l1m1n1o1p1 q1r1s1t1u1v1w1x1 y1z1A1B1C1D1E1F1
; d7 = a0b0c0d0g0h0e0f0 i0j0k0l0m0n0o0p0 q0r0s0t0u0v0w0x0 y0z0A0B0C0D0E0F0

		move.l	d1,(a1)		;  8 plane3
		suba.l	d2,a1		;  8 -plsiz
		move.l	d5,(a1)		;  8 plane2
		suba.l	d2,a1		;  8 -plsiz
		move.l	d6,(a1)		;  8 plane1
		suba.l	d2,a1		;  8 -plsiz
		move.l	d7,(a1)+	;  8 plane 0
		adda.l	(8,sp),a1	; 18 +7*plsiz (or 5*plsiz) (or 3*plsiz)

		cmp.l	(sp),a0		; 14
		bne.w	mainloop	; 10 end of main loop

					;1788 cycles = 55.9 cycles/pixel
exit:
		add.w	#24,sp
		movem.l	(sp)+,d2-d7/a2-a6
		rts
end:
		end