1 @ vim:filetype=armasm
\r
2 @ some color conversion and blitting routines
\r
4 @ (c) Copyright 2006, notaz
\r
5 @ All Rights Reserved
\r
8 @ Convert 0000bbb0 ggg0rrr0 0000bbb0 ggg0rrr0
\r
9 @ to 00000000 rrr00000 ggg00000 bbb00000 ...
\r
11 @ lr = 0x00e000e0, out: r3=lower_pix, r2=higher_pix; trashes rin
\r
12 @ if sh==2, r8=0x00404040 (sh!=0 destroys flags!)
\r
13 .macro convRGB32_2 rin sh=0
\r
14 and r2, lr, \rin, lsr #4 @ blue
\r
16 orr r2, r2, r3, lsl #8 @ g0b0g0b0
\r
18 mov r3, r2, lsl #16 @ g0b00000
\r
19 and \rin,lr, \rin, ror #12 @ 00r000r0 (reversed)
\r
20 orr r3, r3, \rin, lsr #16 @ g0b000r0
\r
22 mov r3, r3, ror #17 @ shadow mode
\r
24 adds r3, r3, #0x40000000 @ green
\r
25 orrcs r3, r3, #0xe0000000
\r
27 adds r3, r3, #0x40000000
\r
28 orrcs r3, r3, #0xe0000000
\r
30 adds r3, r3, #0x40000000
\r
31 orrcs r3, r3, #0xe0000000
\r
33 orr r3, r3, r3, lsr #3
\r
35 mov r3, r3, ror #16 @ r3=low
\r
36 orr r3, r3, r3, lsr #3
\r
42 orr r2, r2, \rin, lsl #16
\r
47 adds r2, r2, #0x40000000 @ blue
\r
48 orrcs r2, r2, #0xe0000000
\r
50 adds r2, r2, #0x40000000
\r
51 orrcs r2, r2, #0xe0000000
\r
53 adds r2, r2, #0x40000000
\r
54 orrcs r2, r2, #0xe0000000
\r
56 orr r2, r2, r2, lsr #3
\r
58 orr r2, r2, r2, lsr #3
\r
65 .global vidConvCpyRGB32 @ void *to, void *from, int pixels
\r
68 stmfd sp!, {r4-r7,lr}
\r
70 mov r12, r2, lsr #3 @ repeats
\r
85 ldmfd sp!, {r4-r7,lr}
\r
89 .global vidConvCpyRGB32sh @ void *to, void *from, int pixels
\r
92 stmfd sp!, {r4-r7,lr}
\r
94 mov r12, r2, lsr #3 @ repeats
\r
109 ldmfd sp!, {r4-r7,lr}
\r
113 .global vidConvCpyRGB32hi @ void *to, void *from, int pixels
\r
116 stmfd sp!, {r4-r7,lr}
\r
118 mov r12, r2, lsr #3 @ repeats
\r
119 mov lr, #0x00e00000
\r
120 orr lr, lr, #0x00e0
\r
132 ldmfd sp!, {r4-r7,lr}
\r
136 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
138 @ -------- M2 stuff ---------
\r
140 .global vidConvCpy_90 @ void *to, void *from, int width
\r
143 stmfd sp!, {r4-r10,lr}
\r
145 mov lr, #0x00F00000
\r
146 orr lr, lr, #0x00F0
\r
148 mov r12, #224/4 @ row counter
\r
149 mov r10, r2, lsl #2 @ we do 2 pixel wide copies
\r
151 add r8, r0, #256*4 @ parallel line
\r
152 add r1, r1, #0x23000
\r
153 add r1, r1, #0x00B80 @ r1+=328*223*2+8*2
\r
156 mov r4, #0 @ fill bottom border
\r
168 @ at first this loop was written differently: src pixels were fetched with ldm's and
\r
169 @ dest was not sequential. It ran nearly 2 times slower. It seems it is very important
\r
170 @ to do sequential memory access on those items, which we have more (to offload addressing bus?).
\r
172 ldr r4, [r1], #-328*2
\r
173 ldr r5, [r1], #-328*2
\r
174 ldr r6, [r1], #-328*2
\r
175 ldr r7, [r1], #-328*2
\r
187 bne .loopM2RGB32_90
\r
189 mov r4, #0 @ top border
\r
192 stmia r0!, {r4-r6,r12}
\r
193 stmia r0!, {r4-r6,r12}
\r
194 stmia r8!, {r4-r6,r12}
\r
195 stmia r8!, {r4-r6,r12}
\r
198 ldmeqfd sp!, {r4-r10,pc} @ return
\r
200 add r0, r8, #16*4 @ set new dst pointer
\r
202 add r9, r9, #2*2 @ fix src pointer
\r
205 stmia r0!, {r4-r6,r12} @ bottom border
\r
206 stmia r0!, {r4-r6,r12}
\r
207 stmia r8!, {r4-r6,r12}
\r
208 stmia r8!, {r4-r6,r12}
\r
210 mov r12, #224/4 @ restore row counter
\r
215 @ converter for vidConvCpy_270
\r
216 @ lr = 0x00F000F0, out: r3=lower_pix, r2=higher_pix; trashes rin
\r
217 .macro convRGB32_3 rin
\r
218 and r2, lr, \rin, lsr #4 @ blue
\r
220 orr r2, r2, r3, lsl #8 @ g0b0g0b0
\r
222 mov r3, r2, lsl #16 @ g0b00000
\r
223 and \rin,lr, \rin, ror #12 @ 00r000r0 (reversed)
\r
224 orr r3, r3, \rin, lsr #16 @ g0b000r0
\r
226 mov r2, r2, lsr #16
\r
227 orr r2, r2, \rin, lsl #16
\r
230 mov \rin,r3, ror #16 @ r3=low
\r
233 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
236 @ takes byte-sized pixels from r3-r6, fetches from pal and stores to r7,r8,r10,lr
\r
238 .macro mode2_4pix shift
\r
239 and r7, r11, r3, lsr #\shift
\r
240 ldr r7, [r2, r7, lsl #2]
\r
242 and r8, r11, r4, lsr #\shift
\r
243 ldr r8, [r2, r8, lsl #2]
\r
245 and r10,r11, r5, lsr #\shift
\r
246 ldr r10,[r2, r10,lsl #2]
\r
248 and lr, r11, r6, lsr #\shift
\r
249 ldr lr, [r2, lr, lsl #2]
\r
253 .macro mode2_4pix_getpix0 dreg sreg
\r
254 and \dreg, r11, \sreg
\r
255 ldr \dreg, [r2, \dreg, lsl #2]
\r
258 .macro mode2_4pix_getpix1 dreg sreg
\r
259 and \dreg, r11, \sreg, lsr #8
\r
260 ldr \dreg, [r2, \dreg, lsl #2]
\r
263 .macro mode2_4pix_getpix2 dreg sreg
\r
264 and \dreg, r11, \sreg, lsr #16
\r
265 ldr \dreg, [r2, \dreg, lsl #2]
\r
268 .macro mode2_4pix_getpix3 dreg sreg
\r
269 and \dreg, r11, \sreg, lsr #24
\r
270 ldr \dreg, [r2, \dreg, lsl #2]
\r
273 @ takes byte-sized pixels from reg, fetches from pal and stores to r3-r6
\r
275 .macro mode2_4pix2_0 reg
\r
276 mode2_4pix_getpix0 r3, \reg
\r
277 mode2_4pix_getpix1 r4, \reg
\r
278 mode2_4pix_getpix2 r5, \reg
\r
279 mode2_4pix_getpix3 r6, \reg
\r
283 .macro mode2_4pix2_180 reg
\r
284 mode2_4pix_getpix3 r3, \reg
\r
285 mode2_4pix_getpix2 r4, \reg
\r
286 mode2_4pix_getpix1 r5, \reg
\r
287 mode2_4pix_getpix0 r6, \reg
\r
290 @ takes byte-sized pixels from reg, fetches from pal and stores to r3-r5
\r
291 @ r11=0xFF, r2=pal, r10=0xfcfcfc, r6=tmp
\r
292 .macro mode2_4pix_to3 reg is180
\r
294 mode2_4pix_getpix3 r3, \reg
\r
295 mode2_4pix_getpix2 r4, \reg
\r
297 mode2_4pix_getpix0 r3, \reg @ gathering loads cause a weird-hang
\r
298 mode2_4pix_getpix1 r4, \reg
\r
301 sub r3, r3, r3, lsr #2 @ r3 *= 0.75
\r
302 add r3, r3, r4, lsr #2 @ r3 += r4 * 0.25
\r
306 mode2_4pix_getpix1 r5, \reg
\r
307 mode2_4pix_getpix0 r6, \reg
\r
309 mode2_4pix_getpix2 r5, \reg
\r
310 mode2_4pix_getpix3 r6, \reg
\r
314 add r4, r4, r5, lsr #1 @ r4 = (r4 + r5) / 2;
\r
316 sub r6, r6, r6, lsr #2 @ r6 *= 0.75
\r
317 add r5, r6, r5, lsr #2 @ r5 = r6 + r5 * 0.25
\r
322 @ void *to, void *from, void *pal, int width
\r
323 .macro vidConvCpyM2_landscape is270
\r
324 stmfd sp!, {r4-r11,lr}
\r
328 mov r12, #(224/4-1)<<16 @ row counter
\r
329 orr r12, r12, r3, lsl #1 @ we do 4 pixel wide copies (right to left)
\r
334 add r1, r1, #0x11c00
\r
335 add r1, r1, #0x00308 @ 328*224+8
\r
339 mov r3, #0 @ fill top border
\r
345 add r7, r0, #256*4-8*4
\r
348 add r7, r7, #256*4-8*4
\r
351 add r7, r7, #256*4-8*4
\r
355 0: @ .loopM2RGB32_270:
\r
356 subs r12, r12, #1<<16
\r
364 ldr r3, [r1, #-328]!
\r
365 ldr r4, [r1, #-328]!
\r
366 ldr r5, [r1, #-328]!
\r
367 ldr r6, [r1, #-328]!
\r
375 stmia r0, {r7,r8,r10,lr}
\r
383 stmia r0, {r7,r8,r10,lr}
\r
391 stmia r0, {r7,r8,r10,lr}
\r
399 stmia r0!,{r7,r8,r10,lr}
\r
400 sub r0, r0, #256*4*3
\r
402 bpl 0b @ .loopM2RGB32_270
\r
404 mov r3, #0 @ bottom border
\r
410 add r0, r0, #256*4-8*4
\r
413 add r0, r0, #256*4-8*4
\r
416 add r0, r0, #256*4-8*4
\r
418 nop @ phone crashes if this is commented out. Do I stress it too much?
\r
421 add r12, r12, #1<<16
\r
423 ldmeqfd sp!, {r4-r11,pc} @ return
\r
427 sub r9, r9, #4 @ fix src pointer
\r
433 stmia r0!, {r3-r6} @ top border
\r
435 add r7, r0, #256*4-8*4
\r
438 add r7, r7, #256*4-8*4
\r
441 add r7, r7, #256*4-8*4
\r
445 orr r12, r12, #(224/4-1)<<16 @ restore row counter
\r
446 b 0b @ .loopM2RGB32_270
\r
450 .global vidConvCpy_90 @ void *to, void *from, void *pal, int width
\r
453 vidConvCpyM2_landscape 0
\r
456 .global vidConvCpy_270 @ void *to, void *from, void *pal, int width
\r
459 vidConvCpyM2_landscape 1
\r
462 .global vidConvCpy_center_0 @ void *to, void *from, void *pal
\r
464 vidConvCpy_center_0:
\r
465 stmfd sp!, {r4-r6,r11,lr}
\r
468 add r1, r1, #8 @ not border (centering 32col here)
\r
470 mov r12, #(240/4-1)<<16
\r
475 subs r12, r12, #1<<16
\r
482 adds r12, r12, #1<<16
\r
483 ldmeqfd sp!, {r4-r6,r11,pc} @ return
\r
486 orr r12, #(240/4-1)<<16
\r
490 .global vidConvCpy_center_180 @ void *to, void *from, void *pal
\r
492 vidConvCpy_center_180:
\r
493 stmfd sp!, {r4-r6,r11,lr}
\r
496 add r1, r1, #0x11c00
\r
497 add r1, r1, #0x002B8 @ #328*224-72
\r
499 mov r12, #(240/4-1)<<16
\r
504 subs r12, r12, #1<<16
\r
508 bpl .loopRGB32_c180
\r
511 adds r12, r12, #1<<16
\r
512 ldmeqfd sp!, {r4-r6,r11,pc} @ return
\r
515 orr r12, #(240/4-1)<<16
\r
519 @ note: the following code assumes that (pal[x] & 0x030303) == 0
\r
521 .global vidConvCpy_center2_40c_0 @ void *to, void *from, void *pal, int lines
\r
523 vidConvCpy_center2_40c_0:
\r
524 stmfd sp!, {r4-r6,r10,r11,lr}
\r
528 orr r10, r10, lsl #8
\r
529 orr r10, r10, lsl #8
\r
530 add r1, r1, #8 @ border
\r
532 mov r12, #(240/3-1)<<16
\r
535 .loopRGB32_c2_40c_0:
\r
537 subs r12, r12, #1<<16
\r
539 mode2_4pix_to3 lr, 0
\r
542 bpl .loopRGB32_c2_40c_0
\r
545 adds r12, r12, #1<<16
\r
546 ldmeqfd sp!, {r4-r6,r10,r11,pc} @ return
\r
549 orr r12, #(240/3-1)<<16
\r
550 b .loopRGB32_c2_40c_0
\r
553 .global vidConvCpy_center2_40c_180 @ void *to, void *from, void *pal, int lines
\r
555 vidConvCpy_center2_40c_180:
\r
556 stmfd sp!, {r4-r6,r10,r11,lr}
\r
560 orr r10, r10, lsl #8
\r
561 orr r10, r10, lsl #8
\r
565 @ add r1, r1, #0x11000
\r
566 @ add r1, r1, #0x00f00 @ #328*224
\r
568 mov r12, #(240/3-1)<<16
\r
573 subs r12, r12, #1<<16
\r
575 mode2_4pix_to3 lr, 1
\r
578 bpl .loop_c2_40c_180
\r
581 adds r12, r12, #1<<16
\r
582 ldmeqfd sp!, {r4-r6,r10,r11,pc} @ return
\r
585 orr r12, #(240/3-1)<<16
\r
589 .global vidConvCpy_center2_32c_0 @ void *to, void *from, void *pal, int lines
\r
591 vidConvCpy_center2_32c_0:
\r
592 stmfd sp!, {r4-r11,lr}
\r
595 orr r10, r10, lsl #8
\r
596 orr r10, r10, lsl #8
\r
598 add r1, r1, #8 @ border
\r
600 mov r12, #(240/15-1)<<16
\r
604 ldmia r1!, {r7-r9,lr}
\r
605 subs r12, r12, #1<<16
\r
613 mode2_4pix_to3 lr, 0
\r
618 adds r12, r12, #1<<16
\r
619 ldmeqfd sp!, {r4-r11,pc} @ return
\r
622 orr r12, #(240/15-1)<<16
\r
626 .global vidConvCpy_center2_32c_180 @ void *to, void *from, void *pal, int lines
\r
628 vidConvCpy_center2_32c_180:
\r
629 stmfd sp!, {r4-r11,lr}
\r
632 orr r10, r10, lsl #8
\r
633 orr r10, r10, lsl #8
\r
638 @ add r1, r1, #0x11000
\r
639 @ add r1, r1, #0x00f00 @ #328*224
\r
641 mov r12, #(240/15-1)<<16
\r
645 ldmdb r1!, {r7-r9,lr}
\r
646 subs r12, r12, #1<<16
\r
654 mode2_4pix_to3 r7, 1
\r
656 bpl .loop_c2_32c_180
\r
659 adds r12, r12, #1<<16
\r
660 ldmeqfd sp!, {r4-r11,pc} @ return
\r
663 orr r12, #(240/15-1)<<16
\r
667 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
670 .global vidClear @ void *to, int lines
\r
675 orr r12, r1, r12, lsl #16
\r
682 subs r12, r12, #1<<16
\r
684 stmia r0!, {r1-r3,lr}
\r
685 stmia r0!, {r1-r3,lr}
\r
686 stmia r0!, {r1-r3,lr}
\r
687 stmia r0!, {r1-r3,lr}
\r
691 adds r12, r12, #1<<16
\r
692 ldmeqfd sp!, {pc} @ return
\r
694 orr r12, #(240/16-1)<<16
\r
697 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
699 .equ EExecSetExceptionHandler, (90)
\r
701 .global my_SetExceptionHandler
\r
703 my_SetExceptionHandler:
\r
705 swi EExecSetExceptionHandler
\r