asm_prsm.s

开放源码实时操作系统源码.

	movel	a6@(8),d2	| x1
	movel	a6@(12),d3	| y1
	movel	a6@(20),d1	| y2
	movel	a6@(16),a2	| x2

	subl	d3,d1		| (y2 - y1) -> d1
	clrl	d4		| d4 = (2 * dx) = NA
	clrl	d5		| d5 = (2 * dx) = NB
	subl	d1,d4		| (2 * dx) - dy -> d4
	subql	#1,d4		| d4 - SY -> d4
	movel	d1,d6		| dy -> d6
	movel	d5,d7		| dx -> d7
	subl	d6,d7		| dx - dy -> d7
	lsll	#1,d7		| shift left, d7 = ND

| by this point, d4 = ND, d5 = NB

	subql	#1,d6		| d6 = NC
	moveq	#6,d0		| SX, SY, XMAJ

| we are drawing a series of vertical lines, all the same length
| so no need to re-calculate all the magic numbers each time

ldraw5x_1:
	cmpl	d2,a2		| x1 <= x2
	jge	ldraw5x_2
	jra	ldraw5x_12
ldraw5x_2:
	movel	#0x0e,a0@(GCW_CMD)	| write, 4x5, 1 of 20
	tstl	d1			| dy
	bne	ldraw5x_10
	movel	d3,a0@(GCW_PY)		| pixel Y addr
	movel	d2,a0@(GCW_PX_SM)	| pixel X addr, start single mem
	bra	ldraw5x_11
ldraw5x_10:
	movel	d4,a0@(GCW_LG2)		| NA
	movel	d5,a0@(GCW_LG1)		| NB
	movel	d7,a0@(GCW_LG2)		| ND
	movel	d6,a0@(GCW_LG3)		| NC
	movel	d0,a0@(GCW_LG4)		| SX, SY, XMAJ
	movel	d3,a0@(GCW_PY)		| pixel Y
	movel	d2,a0@(GCW_PX_SL)	| pixel X, start line
ldraw5x_11:
	addql	#1,d2			| x++
	jra	ldraw5x_1
ldraw5x_12:
	moveml	sp@+,d0-d7/a0-a2
	unlk	a6
	rts

| read 1 pixel from the screen, handle the bizarre layout

	.even
.globl	r1pix3
r1pix3:
	link	a6,#0
	moveml	d1-d5/a0,sp@-
	movew	#1023,d5
	subw	a6@(14),d5	| d5 is y as 16-bit value
	movew	d5,d4
	andib	#0xfc,d4	| d4 is y0
	movew	#3,d0
	andw	d5,d0		| y & 3
	movel	#3,d3		| ensure top word is clear
	subw	d0,d3		| 3 - (y & 3)
	aslw	#3,d3		| << 3
	movel	a6@(8),d0	| x
	divuw	#20,d0		| x % 20
	clrw	d0		| clear quotient
	swap	d0		| put remainder in lower 16 bits
	divuw	#5,d0		| (x % 20) / 5
	aslw	#3,d0		| << 3
	addw	d0,d3		| ys in d3
	movew	#24,d0
	cmpw	d3,d0
	jge	r1pix3_1
	movew	#-32,d0
	addw	d0,d3
r1pix3_1:
	lea	GC_BASE,a0		| base of video control registers
	movel	a6@(16),a0@(GCW_PSEL)	| plane select mask
	movel	#0x07,a0@(GCW_LG4)
	movel	#0x66,a0@(GCW_CMD)
	movel	a6@(8),a0@(GCW_PX)	| x
	movel	d4,a0@(GCW_PY)		| y
	movel	a0@(GCR_IN_SM),d0	| read intensity, start single mem cycle
	movel	a0@(GCR_IN_SM),d0
	movel	d0,d1
	moveq	#0,d0
	notb	d0		| d0 = 0xff
	asll	d3,d0		| << fs
	andl	d1,d0		| (ic & (0xff << fs)
	asrl	d3,d0		| >> ys
	andl	#0xff,d0
	moveml	sp@+,d1-d5/a0
	unlk	a6
	rts

| read a rectangular array of pixels from the screen
| void rd_rect(int x1, int y1, int x2, int y2, char *buf)

	.even
.globl	rd_rect2
rd_rect2:
	link	a6,#0
	moveml	d1-d7/a0-a1,sp@-
	movel	#1023,d7
	subl	a6@(20),d7	| d7 is y2 as 16-bit value
	movel	d7,a6@(20)
	movel	#1023,d6
	subl	a6@(12),d6	| d6 is y1 as 16-bit value
	movel	d6,a6@(12)
	movel	a6@(8),d5	| d5 is x
	lea	GC_BASE,a0	| base of video control registers
	movel	a6@(24),a1	| buf
rd_rect2_1:
	cmpw	a6@(18),d5	| x < x2
	jlt	rd_rect2_2
	jra	rd_rect2_8
rd_rect2_2:
	movel	d5,d4		| d4 is xs
	divuw	#20,d4		| x % 20
	clrw	d4		| clear quotient
	swap	d4		| remainder in lower word
	divuw	#5,d4		| (x % 20) / 5
	aslw	#3,d4		| ((x % 20) / 5) << 3
rd_rect2_3:
	cmpw	d7,d6		| y > y2
	jgt	rd_rect2_4
	jra	rd_rect2_7
rd_rect2_4:
	movel	d6,d3		| d3 is y0
	andib	#0xfc,d3	| y & 0xfffc
	moveq	#3,d0
	andw	d6,d0		| y & 3
	moveq	#3,d2		| ensure top word is clear
	subw	d0,d2		| 3 - (y & 3)
	aslw	#3,d2		| << 3
	addw	d4,d2		| ys in d2
	moveq	#24,d0
	cmpw	d2,d0
	jge	rd_rect2_5
	movew	#-32,d0
	addw	d0,d2
rd_rect2_5:
	movel	#0xff,a0@(GCW_PSEL)	| plane select mask
	movel	#0x07,a0@(GCW_LG4)
	movel	#0x66,a0@(GCW_CMD)
	movel	d5,a0@(GCW_PX)		| x
	movel	d3,a0@(GCW_PY)		| y0
	movel	a0@(GCR_IN_SM),d1	| read intensity, start single mem cycle
	movel	a0@(GCR_IN_SM),d1
	moveq	#0,d0
	notb	d0		| d0 = 0xff
	asll	d2,d0		| << fs
	andl	d1,d0		| (ic & (0xff << fs)
	asrl	d2,d0		| >> ys
	andl	#0xff,d0
	moveb	d0,a1@+		| *buf++
rd_rect2_6:
	subqw	#1,d6		| y--
	jra	rd_rect2_3
rd_rect2_7:
	addqw	#1,d5		| x++
	movel	a6@(12),d6	| y = y1
	jra	rd_rect2_1
rd_rect2_8:
	moveml	sp@+,d1-d7/a0-a1
	unlk	a6
	rts


| write a rectangular array of pixels to the screen
| void wr_rect(int x1, int y1, int x2, int y2, char *buf)

	.even
.globl	wr_rect2
wr_rect2:
	link	a6,#0
	moveml	d1-d7/a0-a1,sp@-
	movel	#1023,d7
	subl	a6@(20),d7	| d7 is y2 as 16-bit value
	movel	d7,a6@(20)
	movel	#1023,d6
	subl	a6@(12),d6	| d6 is y1 as 16-bit value
	movel	d6,a6@(12)
	movel	a6@(8),d5	| d5 is x
	lea	GC_BASE,a0	| base of video control registers
	movel	a6@(24),a1	| buf
wr_rect2_1:
	cmpw	a6@(18),d5	| x < x2
	jlt	wr_rect2_2
	jra	wr_rect2_8
wr_rect2_2:
	movel	d5,d4		| d4 is xs
	divuw	#20,d4		| x % 20
	clrw	d4		| clear quotient
	swap	d4		| remainder in lower word
	divuw	#5,d4		| (x % 20) / 5
	aslw	#3,d4		| ((x % 20) / 5) << 3
wr_rect2_3:
	cmpw	d7,d6		| y > y2
	jgt	wr_rect2_4
	jra	wr_rect2_7
wr_rect2_4:
	movel	d6,d3		| d3 is y0
	andib	#0xfc,d3	| y & 0xfffc
	movel	#0xff,a0@(GCW_PSEL)	| plane select mask
	movel	#0x07,a0@(GCW_LG4)
	movel	#0x66,a0@(GCW_CMD)
	movel	d5,a0@(GCW_PX)		| x
	movel	d3,a0@(GCW_PY)		| y0
	movel	a0@(GCR_IN_SM),d1	| read intensity, start single mem cycle
	movel	a0@(GCR_IN_SM),d1	| d1 is ic
	moveq	#3,d0
	andw	d6,d0		| y & 3
	moveq	#3,d2		| ensure top word is clear
	subw	d0,d2		| 3 - (y & 3)
	aslw	#3,d2		| << 3
	addw	d4,d2		| ys in d2
	moveq	#24,d0
	cmpw	d2,d0
	jge	wr_rect2_5
	movew	#-32,d0
	addw	d0,d2
wr_rect2_5:
	moveq	#0,d0
	notb	d0		| d0 = 0xff
	asll	d2,d0		| <<fs
	notl	d0		| invert
	andl	d1,d0		| (ic & (0cff << ys))
	clrl	d1
	moveb	a1@+,d1		| *buf++
	asll	d2,d1		| << ys
	orl	d1,d0
	movel	#0xff,a0@(GCW_PSEL)	| plane select mask
	movel	#0x07,a0@(GCW_LG4)
	movel	#0x6e,a0@(GCW_CMD)
	movel	d5,a0@(GCW_PX)		| x
	movel	d3,a0@(GCW_PY)		| y0
	movel	d0,a0@(GCW_IN_SM)	| write inten, start single mem cycle
wr_rect2_6:
	subqw	#1,d6		| y--
	jra	wr_rect2_3
wr_rect2_7:
	addqw	#1,d5		| x++
	movel	a6@(12),d6	| y = y1
	jra	wr_rect2_1
wr_rect2_8:
	moveml	sp@+,d1-d7/a0-a1
	unlk	a6
	rts

| write a rectangular array of pixels to the screen
| void wr_rect4(int x1, int y1, int x2, int y2, char *buf)
|
| This version writes the block with the x axis as the inner loop,
| this is slower than wr_rect2() but is necessary for the VNC
| CopyDataToScreen() function.

	.even
.globl	wr_rect4
wr_rect4:
	link	a6,#0
	moveml	d1-d7/a0-a1,sp@-
	movel	#1023,d7
	subl	a6@(20),d7	| d7 is y2 as 16-bit value
	movel	d7,a6@(20)
	movel	#1023,d6
	subl	a6@(12),d6	| d6 is y1 as 16-bit value
	movel	d6,a6@(12)
	movel	a6@(8),d5	| d5 is x
	lea	GC_BASE,a0	| base of video control registers
	movel	a6@(24),a1	| buf
wr_rect4_1:
	cmpw	d7,d6		| y > y2
	jgt	wr_rect4_2
	jra	wr_rect4_8
wr_rect4_2:
	movel	d6,d3		| d3 is y0
	andib	#0xfc,d3	| y & 0xfffc
	moveq	#3,d0
	andw	d6,d0		| y & 3
	moveq	#3,d2		| ensure top word is clear
	subw	d0,d2		| 3 - (y & 3)
	aslw	#3,d2		| << 3
wr_rect4_3:
	cmpw	a6@(18),d5	| x < x2
	jlt	wr_rect4_4
	jra	wr_rect4_7
wr_rect4_4:
	movel	#0xff,a0@(GCW_PSEL)	| plane select mask
	movel	#0x07,a0@(GCW_LG4)
	movel	#0x66,a0@(GCW_CMD)
	movel	d5,a0@(GCW_PX)		| x
	movel	d3,a0@(GCW_PY)		| y0
	movel	a0@(GCR_IN_SM),d1	| read intensity, start single mem cycle
	movel	a0@(GCR_IN_SM),d1	| d1 is ic
	movel	d5,d4		| d4 is xs
	divuw	#20,d4		| x % 20
	clrw	d4		| clear quotient
	swap	d4		| remainder in lower word
	divuw	#5,d4		| (x % 20) / 5
	aslw	#3,d4		| ((x % 20) / 5) << 3
	addw	d2,d4		| ys in d2, xs in d4
	moveq	#24,d0
	cmpw	d4,d0
	jge	wr_rect4_5
	movew	#-32,d0
	addw	d0,d4
wr_rect4_5:
	moveq	#0,d0
	notb	d0		| d0 = 0xff
	asll	d4,d0		| <<fs
	notl	d0		| invert
	andl	d1,d0		| (ic & (0cff << ys))
	clrl	d1
	moveb	a1@+,d1		| *buf++
	asll	d4,d1		| << ys
	orl	d1,d0
	movel	#0xff,a0@(GCW_PSEL)	| plane select mask
	movel	#0x07,a0@(GCW_LG4)
	movel	#0x6e,a0@(GCW_CMD)
	movel	d5,a0@(GCW_PX)		| x
	movel	d3,a0@(GCW_PY)		| y0
	movel	d0,a0@(GCW_IN_SM)	| write inten, start single mem cycle
wr_rect4_6:
	addqw	#1,d5		| x++
	jra	wr_rect4_3
wr_rect4_7:
	subqw	#1,d6		| y--
	movel	a6@(8),d5	| x = x1
	jra	wr_rect4_1
wr_rect4_8:
	moveml	sp@+,d1-d7/a0-a1
	unlk	a6
	rts

| write a rectangular array of pixels to the screen
| void wr_rect5(int x1, int y1, int x2, int y2, char *buf)
|
| This version writes the block with the x axis as the inner loop,
| this is slower than wr_rect2() but is necessary for the VNC
| CopyDataToScreen() function.
|
| This version uses a lookup table for the pixel mask and shift count.
| This removes one of the divide instructions (divides are slow).

	.even
.globl	wr_rect5
wr_rect5:
	link	a6,#0
	moveml	d1-d7/a0-a2,sp@-
	movel	#1023,d7
	subl	a6@(20),d7	| d7 is y2
	movel	#1023,d6
	subl	a6@(12),d6	| d6 is y1
	movel	a6@(8),d5	| d5 is x
	lea	GC_BASE,a0	| base of video control registers
	movel	a6@(24),a1	| buf
	lea	pm_tab,a2	| base of lookup table
	movel	#0xff,a0@(GCW_PSEL)	| plane select mask
wr_rect5_1:
	cmpl	d7,d6		| y > y2
	jle	wr_rect5_8
	movel	d6,d3		| d3 is y0
	andib	#0xfc,d3	| y & 0xfffc
	moveq	#3,d2
	andl	d6,d2		| y & 3
	asll	#3,d2		| (y & 3) * 8
wr_rect5_3:
	cmpl	a6@(16),d5	| x < x2
	jge	wr_rect5_7
	movel	#0x07,a0@(GCW_LG4)
	movel	#0x66,a0@(GCW_CMD)
	movel	d5,a0@(GCW_PX)		| x
	movel	d3,a0@(GCW_PY)		| y0
	movel	a0@(GCR_IN_SM),d1	| read intensity, start single mem cycle
	movel	a0@(GCR_IN_SM),d1	| d1 is ic
	movel	d5,d4		| d4 is xs
	divuw	#20,d4		| x % 20
	clrw	d4		| clear quotient
	swap	d4		| remainder in lower word
	asll	#5,d4		| *32
	addl	d2,d4		| ((x % 20) * 32) + ((y % 4) * 8)
	movel	a2@(d4:w:1),d0	| fetch mask
	andl	d1,d0		| (ic & mask)
|	andl	a2@(d4:w:1),d0	| (ic & mask)
	clrl	d1
	moveb	a1@+,d1		| *buf++
	addql	#4,d4
	movel	a2@(d4:w:1),d4	| fetch shift count
	asll	d4,d1		| << ys
	orl	d1,d0
	movel	#0x07,a0@(GCW_LG4)
	movel	#0x6e,a0@(GCW_CMD)
	movel	d5,a0@(GCW_PX)		| x
	movel	d3,a0@(GCW_PY)		| y0
	movel	d0,a0@(GCW_IN_SM)	| write inten, start single mem cycle
wr_rect5_6:
	addql	#1,d5		| x++
	jra	wr_rect5_3
wr_rect5_7:
	subql	#1,d6		| y--
	movel	a6@(8),d5	| x = x1
	jra	wr_rect5_1
wr_rect5_8:
	moveml	sp@+,d1-d7/a0-a2
	unlk	a6
	rts

| write 1 pixel to the screen, handle the bizarre layout

	.even
.globl	w1pix3
w1pix3:
	link	a6,#0
	moveml	d1-d5/a0,sp@-
	movew	#1023,d5
	subw	a6@(14),d5	| d5 is y as 16-bit value
	movew	d5,d4
	andib	#0xfc,d4	| d4 is y0
	lea	GC_BASE,a0		| base of video control registers
	movel	a6@(16),a0@(GCW_PSEL)	| plane select mask
	movel	#0x07,a0@(GCW_LG4)
	movel	#0x66,a0@(GCW_CMD)
	movel	a6@(8),a0@(GCW_PX)	| x
	movel	d4,a0@(GCW_PY)		| y0
	movel	a0@(GCR_IN_SM),d0	| read intensity, start single mem cycle
	movel	a0@(GCR_IN_SM),d0	| 4 pixels in d0
	movel	d0,d3		| ic in d3
	movew	#3,d0
	andw	d5,d0		| y & 3
	movel	#3,d2		| ensure top word is clear
	subw	d0,d2		| 3 - (y & 3)
	aslw	#3,d2		| << 3
	movel	a6@(8),d0	| x
	divuw	#20,d0		| x % 20
	clrw	d0		| clear quotient
	swap	d0		| put remainder in lower 16 bits
	divuw	#5,d0		| (x % 20) / 5
	aslw	#3,d0		| << 3
	addw	d0,d2		| ys in d2
	movew	#24,d0
	cmpw	d2,d0
	jge	w1pix3_1
	movew	#-32,d0
	addw	d0,d2
w1pix3_1:
	moveq	#0,d0
	notb	d0		| d0 = 0xff
	asll	d2,d0		| << ys
	notl	d0		| invert
	andl	d3,d0		| (ic & (0xff << ys)
	movel	a6@(20),d1	| i
	asll	d2,d1		| i << ys
	orl	d1,d0
	movel	a6@(16),a0@(GCW_PSEL)	| plane select mask
	movel	#0x07,a0@(GCW_LG4)
	movel	#0x6e,a0@(GCW_CMD)
	movel	a6@(8),a0@(GCW_PX)	| x
	movel	d4,a0@(GCW_PY)		| y0
	movel	d0,a0@(GCW_IN_SM)	| write inten, start single mem cycle
	moveml	sp@+,d1-d5/a0
	unlk	a6
	rts

| data

	.data

| lookup table for wr_rect5()
| this gives the pixel mask and shift count to locate each pixel in 
| the 4 x 20 array pattern used by the Prisma video board

pm_tab:		.long	0x00ffffff, 24	|  0,0
		.long	0xff00ffff, 16	|  0,1
		.long	0xffff00ff,  8	|  0,2
		.long	0xffffff00,  0	|  0,3
		.long	0x00ffffff, 24	|  1,0
		.long	0xff00ffff, 16	|  1,1
		.long	0xffff00ff,  8	|  1,2
		.long	0xffffff00,  0	|  1,3
		.long	0x00ffffff, 24	|  2,0
		.long	0xff00ffff, 16	|  2,1
		.long	0xffff00ff,  8	|  2,2
		.long	0xffffff00,  0	|  2,3
		.long	0x00ffffff, 24	|  3,0
		.long	0xff00ffff, 16	|  3,1
		.long	0xffff00ff,  8	|  3,2
		.long	0xffffff00,  0	|  3,3
		.long	0x00ffffff, 24	|  4,0
		.long	0xff00ffff, 16	|  4,1
		.long	0xffff00ff,  8	|  4,2
		.long	0xffffff00,  0	|  4,3
		.long	0xffffff00,  0	|  5,0
		.long	0x00ffffff, 24	|  5,1
		.long	0xff00ffff, 16	|  5,2
		.long	0xffff00ff,  8	|  5,3
		.long	0xffffff00,  0	|  6,0
		.long	0x00ffffff, 24	|  6,1
		.long	0xff00ffff, 16	|  6,2
		.long	0xffff00ff,  8	|  6,3
		.long	0xffffff00,  0	|  7,0
		.long	0x00ffffff, 24	|  7,1
		.long	0xff00ffff, 16	|  7,2
		.long	0xffff00ff,  8	|  7,3
		.long	0xffffff00,  0	|  8,0
		.long	0x00ffffff, 24	|  8,1
		.long	0xff00ffff, 16	|  8,2
		.long	0xffff00ff,  8	|  8,3
		.long	0xffffff00,  0	|  9,0
		.long	0x00ffffff, 24	|  9,1
		.long	0xff00ffff, 16	|  9,2
		.long	0xffff00ff,  8	|  9,3
		.long	0xffff00ff,  8	| 10,0
		.long	0xffffff00,  0	| 10,1
		.long	0x00ffffff, 24	| 10,2
		.long	0xff00ffff, 16	| 10,3
		.long	0xffff00ff,  8	| 11,0
		.long	0xffffff00,  0	| 11,1
		.long	0x00ffffff, 24	| 11,2
		.long	0xff00ffff, 16	| 11,3
		.long	0xffff00ff,  8	| 12,0
		.long	0xffffff00,  0	| 12,1
		.long	0x00ffffff, 24	| 12,2
		.long	0xff00ffff, 16	| 12,3
		.long	0xffff00ff,  8	| 13,0
		.long	0xffffff00,  0	| 13,1
		.long	0x00ffffff, 24	| 13,2
		.long	0xff00ffff, 16	| 13,3
		.long	0xffff00ff,  8	| 14,0
		.long	0xffffff00,  0	| 14,1
		.long	0x00ffffff, 24	| 14,2
		.long	0xff00ffff, 16	| 14,3
		.long	0xff00ffff, 16	| 15,0
		.long	0xffff00ff,  8	| 15,1
		.long	0xffffff00,  0	| 15,2
		.long	0x00ffffff, 24	| 15,3
		.long	0xff00ffff, 16	| 16,0
		.long	0xffff00ff,  8	| 16,1
		.long	0xffffff00,  0	| 16,2
		.long	0x00ffffff, 24	| 16,3
		.long	0xff00ffff, 16	| 17,0
		.long	0xffff00ff,  8	| 17,1
		.long	0xffffff00,  0	| 17,2
		.long	0x00ffffff, 24	| 17,3
		.long	0xff00ffff, 16	| 18,0
		.long	0xffff00ff,  8	| 18,1
		.long	0xffffff00,  0	| 18,2
		.long	0x00ffffff, 24	| 18,3
		.long	0xff00ffff, 16	| 19,0
		.long	0xffff00ff,  8	| 19,1
		.long	0xffffff00,  0	| 19,2
		.long	0x00ffffff, 24	| 19,3


| temporary storage for line drawing routine

gr_data:
line_x1:	.long	0	|  0 (668)
line_y1:	.long	0	|  4 (66c)
line_x2:	.long	0	|  8 (670)
line_y2:	.long	0	| 12 (674)
		.long	0	| 16 (678)
		.long	0	| 20 (67c)
		.long	0	| 24 (680)
		.long	0	| 28 (684)
		.long	0	| 32 (688)
		.long	0	| 36 (68c)
line_xdir:	.long	0	| 40 (690)
line_ydir:	.long	0	| 44 (694)
		.long	0	| 48 (698)
line_selmask:	.long	0	| 52 (69c)
line_inten:	.long	0	| 56 (6A0)
line_flag:	.word	0	| 60 (6a4)
init_flag1:	.word	0	| 62 (6a6)
		.long	0	| 64 (6a8)
		.long	0	| 64 (6ac)
init_flag4:	.long	0	| 64 (6b0)
		.long	0	| 64 (6b4)
		.long	0	| 64 (6b8)
		.long	0	| 64 (6bc)
		.long	0	| 64 (6c0)
		.long	0	| 64 (6c4)
		.long	0	| 64 (6c8)
		.long	0	| 64 (6cc)
		.long	0	| 64 (6d0)
		.long	0	| 64 (6d4)
		.long	0	| 64 (6d8)
		.long	0	| 64 (6dc)
		.long	0	| 64 (6e0)
		.long	0	| 64 (6e4)
		.long	0	| 64 (6e8)
		.long	0	| 64 (6ec)
init_flag6:	.long	0	| 64 (6f0)
init_flag5:	.word	0	| 64 (6f4)
init_flag2:	.word	0	| 64 (6f6)
init_flag3:	.word	0	| 64 (6f8)