2 * assembly optimized versions of most funtions from draw.c
\r
3 * (C) notaz, 2006-2010,2017
\r
4 * (C) irixxxx, 2020-2024
\r
6 * This work is licensed under the terms of MAME license.
\r
7 * See COPYING file in the top-level directory.
\r
9 * this is highly specialized, be careful if changing related C code!
\r
11 * NB only does RGB565 output, BGR isn't supported
\r
14 #include "pico_int_offs.h"
\r
16 .extern DrawStripInterlace
\r
18 .equ PDRAW_WND_DIFF_PRIO, (1<<1)
\r
19 .equ PDRAW_PLANE_HI_PRIO, (1<<6)
\r
20 .equ PDRAW_SHHI_DONE, (1<<7)
\r
21 .equ PDRAW_BORDER_32, (1<<9)
\r
22 .equ PDRAW_32X_SCALE, (1<<12)
\r
23 .equ PDRAW_BGC_DMA, (1<<14)
\r
24 .equ PDRAW_SOFTSCALE, (1<<15)
\r
27 .macro add_c24 d s c
\r
28 add \d, \s, #(\c & 0x00ff00)
\r
30 add \d, \d, #(\c & 0x0000ff)
\r
33 add \d, \d, #(\c & 0xff0000)
\r
37 .macro TilePixel pat lsrr offs
\r
41 ands r4, \pat, r2, lsr #\lsrr
\r
44 strneb r4, [r1,#\offs]
\r
47 @ TileNorm (r1=pdest, r2=pixels8, r3=pal) r4: scratch, pat: register with helper pattern 0xf
\r
49 TilePixel \pat, 12, 0 @ #0x0000f000
\r
50 TilePixel \pat, 8, 1 @ #0x00000f00
\r
51 TilePixel \pat, 4, 2 @ #0x000000f0
\r
52 TilePixel \pat, 0, 3 @ #0x0000000f
\r
53 TilePixel \pat, 28, 4 @ #0xf0000000
\r
54 TilePixel \pat, 24, 5 @ #0x0f000000
\r
55 TilePixel \pat, 20, 6 @ #0x00f00000
\r
56 TilePixel \pat, 16, 7 @ #0x000f0000
\r
59 @ TileFlip (r1=pdest, r2=pixels8, r3=pal) r4: scratch, pat: register with helper pattern 0xf
\r
61 TilePixel \pat, 16, 0 @ #0x000f0000
\r
62 TilePixel \pat, 20, 1 @ #0x00f00000
\r
63 TilePixel \pat, 24, 2 @ #0x0f000000
\r
64 TilePixel \pat, 28, 3 @ #0xf0000000
\r
65 TilePixel \pat, 0, 4 @ #0x0000000f
\r
66 TilePixel \pat, 4, 5 @ #0x000000f0
\r
67 TilePixel \pat, 8, 6 @ #0x00000f00
\r
68 TilePixel \pat, 12, 7 @ #0x0000f000
\r
71 @ shadow/hilight mode
\r
74 .macro TilePixelNonSH pat lsrr offs
\r
78 ands r4, \pat, r2, lsr #\lsrr
\r
84 strb r4, [r1,#\offs]
\r
88 @ TileNormNonSH (r1=pdest, r2=pixels8, r3=pal) r4: scratch, pat: register with helper pattern 0xf
\r
89 .macro TileNormNonSH pat
\r
90 TilePixelNonSH \pat, 12, 0 @ #0x0000f000
\r
91 TilePixelNonSH \pat, 8, 1 @ #0x00000f00
\r
92 TilePixelNonSH \pat, 4, 2 @ #0x000000f0
\r
93 TilePixelNonSH \pat, 0, 3 @ #0x0000000f
\r
94 TilePixelNonSH \pat, 28, 4 @ #0xf0000000
\r
95 TilePixelNonSH \pat, 24, 5 @ #0x0f000000
\r
96 TilePixelNonSH \pat, 20, 6 @ #0x00f00000
\r
97 TilePixelNonSH \pat, 16, 7 @ #0x000f0000
\r
100 @ TileFlipNonSH (r1=pdest, r2=pixels8, r3=pal) r4: scratch, pat: register with helper pattern 0xf
\r
101 .macro TileFlipNonSH pat
\r
102 TilePixelNonSH \pat, 16, 0 @ #0x000f0000
\r
103 TilePixelNonSH \pat, 20, 1 @ #0x00f00000
\r
104 TilePixelNonSH \pat, 24, 2 @ #0x0f000000
\r
105 TilePixelNonSH \pat, 28, 3 @ #0xf0000000
\r
106 TilePixelNonSH \pat, 0, 4 @ #0x0000000f
\r
107 TilePixelNonSH \pat, 4, 5 @ #0x000000f0
\r
108 TilePixelNonSH \pat, 8, 6 @ #0x00000f00
\r
109 TilePixelNonSH \pat, 12, 7 @ #0x0000f000
\r
112 @ this one is for hi priority layer
\r
113 .macro TilePixelShHP lsrr offs
\r
117 ands r4, r12, r2, lsr #\lsrr
\r
119 ldreqb r4, [r1,#\offs]
\r
121 andeq r4, r4, #0x7f
\r
122 strb r4, [r1,#\offs]
\r
125 @ TileNormShHP (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: register with helper pattern 0xf, touches r3 high bits
\r
126 .macro TileNormShHP
\r
127 TilePixelShHP 12, 0 @ #0x0000f000
\r
128 TilePixelShHP 8, 1 @ #0x00000f00
\r
129 TilePixelShHP 4, 2 @ #0x000000f0
\r
130 TilePixelShHP 0, 3 @ #0x0000000f
\r
131 TilePixelShHP 28, 4 @ #0xf0000000
\r
132 TilePixelShHP 24, 5 @ #0x0f000000
\r
133 TilePixelShHP 20, 6 @ #0x00f00000
\r
134 TilePixelShHP 16, 7 @ #0x000f0000
\r
137 @ TileFlipShHP (r1=pdest, r2=pixels8, r3=pal) r4: scratch, pat: register with helper pattern 0xf
\r
138 .macro TileFlipShHP
\r
139 TilePixelShHP 16, 0 @ #0x000f0000
\r
140 TilePixelShHP 20, 1 @ #0x00f00000
\r
141 TilePixelShHP 24, 2 @ #0x0f000000
\r
142 TilePixelShHP 28, 3 @ #0xf0000000
\r
143 TilePixelShHP 0, 4 @ #0x0000000f
\r
144 TilePixelShHP 4, 5 @ #0x000000f0
\r
145 TilePixelShHP 8, 6 @ #0x00000f00
\r
146 TilePixelShHP 12, 7 @ #0x0000f000
\r
150 @ TileSingleSh (r1=pdest, r2=pixels8, r3=pal) r4,r7: scratch, r0=sx; r12: helper pattern 0xf
\r
151 .macro TileSingleSh
\r
152 tst r0, #1 @ not aligned?
\r
155 ldrneb r4, [r1], #1
\r
156 ldreqh r4, [r1], #2 @ 1ci
\r
159 strneb r4, [r1, #-3]
\r
160 streqh r4, [r1, #-4]
\r
163 strh r12, [r1, #-4]
\r
169 strh r12, [r1, #-2]
\r
171 strneb r4, [r1], #1
\r
175 @ TileSingleHi (r1=pdest, r2=pixels8, r3=pal) r4,r7: scratch, r0=sx, r12: register with helper pattern 0xf
\r
176 .macro TileSingleHi
\r
177 tst r1, #1 @ not aligned?
\r
180 ldrneb r4, [r1], #1
\r
181 ldreqh r4, [r1], #2 @ 1ci
\r
184 strneb r4, [r1, #-3]
\r
185 streqh r4, [r1, #-4]
\r
188 strh r12, [r1, #-4]
\r
194 strh r12, [r1, #-2]
\r
196 strneb r4, [r1], #1
\r
200 .macro TileDoShGenPixel shift ofs
\r
202 ands r4, r12, r2, lsr #\shift
\r
208 ldrgeb r7, [r1,#\ofs]
\r
209 orrlt r7, r3, r4 @ normal
\r
212 orrge r7, r7, r4, lsl #6
\r
213 strb r7, [r1,#\ofs]
\r
217 @ TileFlipSh (r1=pdest, r2=pixels8, r3=pal) r4,r7: scratch, r0=sx, r12: register with helper pattern 0xf
\r
219 TileDoShGenPixel 16, 0 @ #0x000f0000
\r
220 TileDoShGenPixel 20, 1 @ #0x00f00000
\r
221 TileDoShGenPixel 24, 2 @ #0x0f000000
\r
222 TileDoShGenPixel 28, 3 @ #0xf0000000
\r
223 TileDoShGenPixel 0, 4 @ #0x0000000f
\r
224 TileDoShGenPixel 4, 5 @ #0x000000f0
\r
225 TileDoShGenPixel 8, 6 @ #0x00000f00
\r
226 TileDoShGenPixel 12, 7 @ #0x0000f000
\r
229 @ TileNormSh (r1=pdest, r2=pixels8, r3=pal) r4,r7: scratch, r0=sx, r12: register with helper pattern 0xf
\r
231 TileDoShGenPixel 12, 0 @ #0x0000f000
\r
232 TileDoShGenPixel 8, 1 @ #0x00000f00
\r
233 TileDoShGenPixel 4, 2 @ #0x000000f0
\r
234 TileDoShGenPixel 0, 3 @ #0x0000000f
\r
235 TileDoShGenPixel 28, 4 @ #0xf0000000
\r
236 TileDoShGenPixel 24, 5 @ #0x0f000000
\r
237 TileDoShGenPixel 20, 6 @ #0x00f00000
\r
238 TileDoShGenPixel 16, 7 @ #0x000f0000
\r
241 .macro TileDoShGenPixel_markop shift ofs
\r
243 ands r4, r12, r2, lsr #\shift
\r
249 ldrgeb r4, [r1,#\ofs]
\r
251 orrge r4, r4, #0x40
\r
252 strb r4, [r1,#\ofs]
\r
256 .macro TileFlipSh_markop
\r
257 TileDoShGenPixel_markop 16, 0 @ #0x000f0000
\r
258 TileDoShGenPixel_markop 20, 1 @ #0x00f00000
\r
259 TileDoShGenPixel_markop 24, 2 @ #0x0f000000
\r
260 TileDoShGenPixel_markop 28, 3 @ #0xf0000000
\r
261 TileDoShGenPixel_markop 0, 4 @ #0x0000000f
\r
262 TileDoShGenPixel_markop 4, 5 @ #0x000000f0
\r
263 TileDoShGenPixel_markop 8, 6 @ #0x00000f00
\r
264 TileDoShGenPixel_markop 12, 7 @ #0x0000f000
\r
267 .macro TileNormSh_markop
\r
268 TileDoShGenPixel_markop 12, 0 @ #0x0000f000
\r
269 TileDoShGenPixel_markop 8, 1 @ #0x00000f00
\r
270 TileDoShGenPixel_markop 4, 2 @ #0x000000f0
\r
271 TileDoShGenPixel_markop 0, 3 @ #0x0000000f
\r
272 TileDoShGenPixel_markop 28, 4 @ #0xf0000000
\r
273 TileDoShGenPixel_markop 24, 5 @ #0x0f000000
\r
274 TileDoShGenPixel_markop 20, 6 @ #0x00f00000
\r
275 TileDoShGenPixel_markop 16, 7 @ #0x000f0000
\r
278 .macro TileDoShGenPixel_onlyop_lp shift ofs
\r
280 ands r7, r12, r2, lsr #\shift
\r
284 ldrneb r4, [r1,#\ofs]
\r
289 bicne r4, r4, #0x40
\r
291 orrne r4, r4, r7, lsl #6
\r
292 strneb r4, [r1,#\ofs]
\r
296 .macro TileFlipSh_onlyop_lp
\r
297 TileDoShGenPixel_onlyop_lp 16, 0 @ #0x000f0000
\r
298 TileDoShGenPixel_onlyop_lp 20, 1 @ #0x00f00000
\r
299 TileDoShGenPixel_onlyop_lp 24, 2 @ #0x0f000000
\r
300 TileDoShGenPixel_onlyop_lp 28, 3 @ #0xf0000000
\r
301 TileDoShGenPixel_onlyop_lp 0, 4 @ #0x0000000f
\r
302 TileDoShGenPixel_onlyop_lp 4, 5 @ #0x000000f0
\r
303 TileDoShGenPixel_onlyop_lp 8, 6 @ #0x00000f00
\r
304 TileDoShGenPixel_onlyop_lp 12, 7 @ #0x0000f000
\r
307 .macro TileNormSh_onlyop_lp
\r
308 TileDoShGenPixel_onlyop_lp 12, 0 @ #0x0000f000
\r
309 TileDoShGenPixel_onlyop_lp 8, 1 @ #0x00000f00
\r
310 TileDoShGenPixel_onlyop_lp 4, 2 @ #0x000000f0
\r
311 TileDoShGenPixel_onlyop_lp 0, 3 @ #0x0000000f
\r
312 TileDoShGenPixel_onlyop_lp 28, 4 @ #0xf0000000
\r
313 TileDoShGenPixel_onlyop_lp 24, 5 @ #0x0f000000
\r
314 TileDoShGenPixel_onlyop_lp 20, 6 @ #0x00f00000
\r
315 TileDoShGenPixel_onlyop_lp 16, 7 @ #0x000f0000
\r
319 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
323 @ int nametab; // 0x00
\r
324 @ int line; // 0x04
\r
325 @ int hscroll; // 0x08
\r
326 @ int xmask; // 0x0C
\r
327 @ int *hc; // 0x10 (pointer to cache buffer)
\r
328 @ int cells; // 0x14
\r
331 @ void DrawLayer(int lflags, int *hcache, int cellskip, int maxcells,
\r
332 @ struct PicoEState *est)
\r
337 ldr r12, [sp] @ est
\r
338 stmfd sp!, {r4-r11,lr}
\r
340 ldr r11, [r12, #OFS_EST_Pico]
\r
343 ldrb r7, [r11, #OFS_Pico_video_reg+16] @ ??vv??hh
\r
345 mov r6, r1 @ hcache
\r
346 orr r9, r3, r0, lsl #29 @ r9=force[31]|sh[30]|plane[29]
\r
347 orr r9, r9, r2, lsl #8 @ |cellskip[15:8]|maxcells[7:0] (tmp)
\r
350 orr r1, r1, #0x00ff
\r
354 biclt r1, r1, #0xfc00
\r
355 biceq r1, r1, #0xfe00
\r
358 movgt r1, #0x00ff @ r1=ymask=(height<<8)|0xff; ...; // Y Mask in pixels
\r
363 movgt r10, #7 @ r10=shift[width] (5,6,5,7)
\r
365 ldr r2, [r12, #OFS_EST_DrawScanline]
\r
366 ldr lr, [r12, #OFS_EST_PicoMem_vram]
\r
370 ldreqb r12, [r11, #OFS_Pico_video_reg+2]
\r
371 ldrneb r12, [r11, #OFS_Pico_video_reg+4]
\r
374 mov r5, r8, lsl r10
\r
375 sub r5, r5, #1 @ r5=xmask
\r
377 moveq r12, r12, lsl #10
\r
378 movne r12, r12, lsl #13
\r
379 and r12, r12, #(7<<13) @ r12=(ts->nametab<<1) (halfword compliant)
\r
381 ldrh r8, [r11, #OFS_Pico_video_reg+12]
\r
382 ldrb r7, [r11, #OFS_Pico_video_reg+11]
\r
384 mov r4, r8, lsr #8 @ pvid->reg[13]
\r
385 mov r4, r4, lsl #10 @ htab=pvid->reg[13]<<9; (halfwords)
\r
390 mov r3, r2, lsl #2 @ htab+=DrawScanline<<1; // Offset by line
\r
391 biceq r3, r3, #0x1f @ htab&=~0xf; // Offset by tile
\r
392 andlt r3, r3, #0x1f
\r
394 0: add r4, r4, r0, lsl #1 @ htab+=plane
\r
395 bic r4, r4, #0x00ff0000 @ just in case
\r
396 ldrh r3, [lr, r4] @ r3=hscroll
\r
399 bne .DrawStrip_vsscroll
\r
401 @ Get vertical scroll value:
\r
402 add_c24 r7, lr, (OFS_PMEM_vsram-OFS_PMEM_vram)
\r
407 bne .DrawStrip_interlace
\r
410 moveq r7, r7, lsl #16
\r
411 mov r7, r7, lsr #16
\r
413 @ Find the line in the name table
\r
417 add r10, r10, #1 @ shift[width]++
\r
418 add r12, r12, r4, lsl r10 @ nametab+=(ts.line>>3)<<shift[width];
\r
420 @ ldmia r0, {r1,r2,r3,r5,r6,r9} @ r2=line, r3=ts->hscroll, r5=ts->xmask, r6=ts->hc, r9=ts->cells
\r
423 mov r10,r10, lsl #1 @ r10=ty=(ts->line&7)<<1;
\r
424 orr r10,r10, r9, lsl #24
\r
427 mov r8, r8, lsr #3 @ r8=tilex=(-ts->hscroll)>>3
\r
431 add r7, r1, #1 @ r7=dx=((ts->hscroll-1)&7)+1
\r
433 movs r3, r9, lsl #1 @ (force[31]|sh[30]) << 1
\r
435 orrmi r10,r10, #1<<23 @ r10=cells[31:24]|sh[23]|hi_not_empty[22]
\r
436 @ orrcc r10,r10, #1<<20 @ |had_output[21]|!force[20]|hscroll[18:16]|ty[15:0]
\r
437 movmi r3, #0x80 @ default to shadowed pal on sh mode
\r
440 orr r10,r10, r4, lsl #16 @ we will process cells+1 if there is scroll
\r
442 and r9, r9, #0xff00
\r
443 add r8, r8, r9, lsr #8 @ tilex+=cellskip
\r
444 add r7, r7, r9, lsr #5 @ dx+=cellskip<<3;
\r
445 sub r10,r10,r9, lsl #16 @ cells-=cellskip
\r
447 @ cache some stuff to avoid mem access
\r
448 ldr r11,[sp, #9*4] @ est
\r
450 ldr r11,[r11, #OFS_EST_HighCol]
\r
452 mvn r9, #0 @ r9=prevcode=-1
\r
453 add r1, r11, r7 @ r1=pdest
\r
455 @ r10=cells[31:24]|sh[23]|hi_not_empty[22]|had_output[21]|!force[20]|hscroll[18:16]|ty[15:0]
\r
456 @ r1=pd+dx r2=pack r3=pal r5=xmask r6=hc r8=tilex r9=prevcode r11=HighCol r12=nametab lr=vram
\r
457 @ r4 & r7 are scratch in this loop
\r
459 ands r4, r10, #7<<16 @ hscroll?
\r
461 subs r10,r10, #0x01000000
\r
464 and r7, r5, r8 @ do first cut tile
\r
465 add r7, lr, r7, lsl #1 @ Pico.vram+((tilex&ts->xmask) as halfwords)
\r
466 ldrh r9, [r7, r12] @ r7=code (int, but from unsigned, no sign extend)
\r
470 tst r9, #0x1000 @ if (code&0x1000)
\r
471 mov r2, r9, lsl #21
\r
472 add r2, r2, r10, lsl #17
\r
473 eorne r2, r2, #0xe<<17 @ if (code&0x1000) addr^=0xe;
\r
475 ldr r2, [lr, r2, lsr #16] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
\r
478 mov r4, r4, lsr #16-2 @ (dx&7)*4
\r
480 moveq r7, r7, lsl r4 @ mask = ~0 [shift] (dx&7)*4
\r
481 movne r7, r7, lsr r4
\r
482 bic r2, r2, r7, ror #16 @ pack&~mask
\r
484 orr r9, r9, #0x80000000 @ invalidate oldcode since pack is masked
\r
485 b .DrawStrip_samecode
\r
489 .dsloop: @ 40-41 times
\r
490 subs r10,r10, #0x01000000
\r
494 add r7, lr, r7, lsl #1 @ Pico.vram+((tilex&ts->xmask) as halfwords)
\r
495 ldrh r7, [r7, r12] @ r7=code (int, but from unsigned, no sign extend)
\r
501 beq .DrawStrip_samecode @ we know stuff about this tile already
\r
503 mov r9, r7 @ remember code
\r
505 tst r9, #0x1000 @ if (code&0x1000)
\r
506 mov r2, r9, lsl #21
\r
507 add r2, r2, r10, lsl #17
\r
508 eorne r2, r2, #0x0e<<17 @ if (code&0x1000) addr^=0xe;
\r
510 ldr r2, [lr, r2, lsr #16] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
\r
511 .DrawStrip_samecode:
\r
513 @ tstne r10, #1<<20 @ !force[20]
\r
514 bne .DrawStrip_hiprio
\r
516 orr r10, r10, #1<<21 @ seen non hi-prio tile
\r
518 beq .dsloop @ tileline blank
\r
521 and r3, r9, #0x6000
\r
522 add r3, r7, r3, lsr #9 @ r3=pal=((code&0x6000)>>9);
\r
525 beq .DrawStrip_SingleColor @ tileline singlecolor
\r
528 bne .DrawStrip_TileFlip
\r
530 @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r0: helper pattern
\r
531 .DrawStrip_TileNorm:
\r
535 .DrawStrip_TileFlip:
\r
539 .DrawStrip_SingleColor:
\r
542 orr r4, r4, r4, lsl #8
\r
543 tst r1, #1 @ not aligned?
\r
544 strneb r4, [r1], #1
\r
545 streqh r4, [r1], #2
\r
549 strneb r4, [r1], #1 @ have a remaining unaligned pixel?
\r
553 tst r10, #(1<<23) @ sh[23]
\r
554 tsteq r2, r2 @ if (!sh[23] && code==blank) continue
\r
557 @ orr r10, r10, #1<<22 @ hi_not_empty[22]
\r
559 orr r7, r9, r7, lsl #16
\r
560 orr r7, r7, r10, lsl #25 @ (ty<<25)
\r
562 eorne r7, r7, #0xe<<25 @ if(code&0x1000) cval^=0xe<<25;
\r
563 str r7, [r6], #4 @ cache hi priority tile code
\r
564 str r2, [r6], #4 @ cache hi priority tile data
\r
568 ands r4,r10, #7<<16 @ hscroll?
\r
569 beq .DrawStrip_noscroll
\r
571 and r7, r5, r8 @ do one more cut tile
\r
572 add r7, lr, r7, lsl #1 @ Pico.vram+((tilex&ts->xmask) as halfwords)
\r
573 ldrh r9, [r7, r12] @ r7=code (int, but from unsigned, no sign extend)
\r
577 tst r9, #0x1000 @ if (code&0x1000)
\r
578 mov r2, r9, lsl #21
\r
579 add r2, r2, r10, lsl #17
\r
580 eorne r2, r2, #0x0e<<17 @ if (code&0x1000) addr^=0xe;
\r
582 ldr r2, [lr, r2, lsr #16] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
\r
585 mov r4, r4, lsr #16-2 @ (dx&7)*4
\r
587 moveq r7, r7, lsl r4 @ mask = ~0 [shift] (dx&7)*4
\r
588 movne r7, r7, lsr r4
\r
589 and r2, r2, r7, ror #16 @ pack&mask
\r
591 bic r10,r10, #7<<16
\r
592 b .DrawStrip_samecode @ one last time, with last tile now masked
\r
594 .DrawStrip_noscroll:
\r
595 tst r10, #1<<21 @ seen non hi-prio tile
\r
596 ldr r1, [sp, #9*4] @ est
\r
598 ldreq r2, [r1, #OFS_EST_rendstatus]
\r
599 str r0, [r6] @ terminate the cache list
\r
600 orreq r2, r2, #PDRAW_PLANE_HI_PRIO @ had a layer with all hi-prio tiles
\r
601 streq r2, [r1, #OFS_EST_rendstatus]
\r
603 ldmfd sp!, {r4-r11,pc}
\r
605 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
607 .DrawStrip_vsscroll:
\r
608 tst r8, #1 @ if h40: lflags |= 0x10000
\r
609 orrne r0, r0, #0x10000
\r
612 mov r8, r8, lsr #3 @ r8=tilex=(-ts->hscroll)>>3
\r
613 bic r8, r8, #0x3fc00000
\r
614 orr r8, r8, r5, lsl #25 @ r8=(xmask[31:25]|had_output[24]|!force[23]|tilex[21:0])
\r
616 ldr r11, [sp, #9*4] @ est
\r
617 orr r5, r1, r10, lsl #24
\r
618 ldr r4, [r11, #OFS_EST_DrawScanline]
\r
620 orr r5, r5, r4, lsl #16 @ r5=(shift_width[31:24]|scanline[23:16]|ymask[15:0])
\r
622 add r7, r1, #1 @ r7=dx=((ts->hscroll-1)&7)+1
\r
624 mov r10,r9, lsl #16
\r
625 orr r10,r10, #0xff000000 @ will be adjusted on entering loop
\r
627 orrne r10,r10, #0x8000
\r
628 tst r3, #0x0f @ hscroll & 0x0f?
\r
631 sub r10,r10, #1<<24 @ cell-- // start from negative for hscroll
\r
633 subne r10,r10, #1<<16 @ cells--
\r
634 subne r10,r10, #1<<24 @ cell-- // even more negative
\r
636 add_c24 r1, lr, (OFS_PMEM_vsram-OFS_PMEM_vram)
\r
637 tst r0, #0x10000 @ h40?
\r
638 ldrne r3, [r1, #0x00] @ r3=vsram[0x00..0x01]
\r
639 ldreq r3, [r1, #0x40] @ r3=vsram[0x20..0x21]
\r
640 str r3, [r1, #0x7c] @ vsram[0x3e..0x3f]=r3
\r
644 orrne r10,r10, #1<<23 @ r10=(cell[31:24]|sh[23]|hi_not_empty[22]|cells_max[21:16]|plane[15]|ty[14:0])
\r
645 movne r3, #0x80 @ default to shadowed pal on sh mode
\r
647 @ orreq r8, r8, #1<<23
\r
649 and r9, r9, #0xff00
\r
650 add r8, r8, r9, lsr #8 @ tilex+=cellskip
\r
651 add r7, r7, r9, lsr #5 @ dx+=cellskip<<3;
\r
652 add r10,r10,r9, lsl #16 @ cell+=cellskip
\r
654 @ cache some stuff to avoid mem access
\r
655 ldr r11,[sp, #9*4] @ est
\r
657 ldr r11,[r11, #OFS_EST_HighCol]
\r
659 mvn r9, #0 @ r9=prevcode=-1
\r
660 add r1, r11, r7 @ r1=pdest
\r
662 @ r10=cells[31:24]|sh[23]|hi_not_empty[22]|cells_max[21:16]|plane[15]|ty[14:0]
\r
663 @ r8=xmask[31:25]|had_output[24]|!force[23]|tilex[21:0]
\r
664 @ r5=shift_width[31:24]|scanline[23:16]|ymask[15:0]
\r
665 @ r3=nametabadd[31:16]|must_be_0[15:8]|pal[7:0]
\r
666 @ r1=pd+dx r2=pack r6=hc r9=prevcode r11=HighCol r12=nametab lr=vram
\r
667 @ r4 & r7 are scratch in this loop
\r
669 @ need to calc new ty?
\r
670 movs r7, r10, lsl #7 @ (cell&1)?
\r
671 bmi .dsloop_vs_subr1
\r
673 @ calc offset and read tileline code to r7, also calc ty
\r
674 add_c24 r7, lr, (OFS_PMEM_vsram-OFS_PMEM_vram)
\r
675 and r4, r10, #0x3e000000
\r
676 add r7, r7, r4, asr #23 @ vsram + ((cell&0x3e)<<1)
\r
677 tst r10,#0x8000 @ plane1?
\r
679 ldrh r7, [r7] @ r7=vscroll
\r
681 bic r10,r10,#0xff @ clear old ty
\r
682 and r4, r5, #0xff0000 @ scanline
\r
683 add r4, r4, r7, lsl #16 @ ... += vscroll
\r
684 and r4, r4, r5, lsl #16 @ ... &= ymask
\r
685 and r7, r4, #0x70000
\r
686 orr r10,r10,r7, lsr #15 @ new ty
\r
688 mov r4, r4, lsr #19
\r
689 mov r7, r5, lsr #24
\r
690 mov r4, r4, lsl r7 @ nametabadd
\r
692 orr r3, r3, r4, lsl #16 @ r3=(nametabadd[31:16],pal[15:0])
\r
696 .dsloop_vs: @ 40-41 times
\r
697 add r10,r10, #0x01000000
\r
698 and r4, r10, #0x003f0000
\r
699 cmp r4, r10, asr #8
\r
700 ble .dsloop_vs_exit
\r
702 @ need to calc new ty?
\r
703 movs r7, r10, lsl #7 @ (cell&1)?
\r
706 @ calc offset and read tileline code to r7, also calc ty
\r
707 add_c24 r7, lr, (OFS_PMEM_vsram-OFS_PMEM_vram)
\r
708 and r4, r10, #0x3e000000
\r
709 add r7, r7, r4, asr #23 @ vsram + ((cell&0x3e)<<1)
\r
710 tst r10,#0x8000 @ plane1?
\r
712 ldrh r7, [r7] @ r7=vscroll
\r
714 bic r10,r10,#0xff @ clear old ty
\r
715 and r4, r5, #0xff0000 @ scanline
\r
716 add r4, r4, r7, lsl #16 @ ... += vscroll
\r
717 and r4, r4, r5, lsl #16 @ ... &= ymask
\r
718 and r7, r4, #0x70000
\r
719 orr r10,r10,r7, lsr #15 @ new ty
\r
721 mov r4, r4, lsr #19
\r
722 mov r7, r5, lsr #24
\r
723 mov r4, r4, lsl r7 @ nametabadd
\r
725 orr r3, r3, r4, lsl #16 @ r3=(nametabadd[31:16],pal[15:0])
\r
727 and r7, r8, r8, lsr #25
\r
728 add r7, lr, r7, lsl #1 @ PicoMem.vram+((tilex&ts->xmask) as halfwords)
\r
729 add r7, r7, r3, lsr #15
\r
730 ldrh r7, [r7, r12] @ r7=code (int, but from unsigned, no sign extend)
\r
735 orr r7, r7, r10, lsl #24 @ code | (ty << 24)
\r
737 beq .DrawStrip_vs_samecode @ we know stuff about this tile already
\r
739 mov r9, r7 @ remember code
\r
741 tst r9, #0x1000 @ if (code&0x1000)
\r
742 mov r2, r9, lsl #21
\r
743 add r2, r2, r10, lsl #17
\r
744 eorne r2, r2, #0x0e<<17 @ if (code&0x1000) addr^=0xe;
\r
746 ldr r2, [lr, r2, lsr #16] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
\r
748 .DrawStrip_vs_samecode:
\r
750 @ tstne r8, #1<<23 @ !force[23]
\r
751 bne .DrawStrip_vs_hiprio
\r
753 orr r8, r8, #(1<<24)@ seen non hi-prio tile
\r
755 beq .dsloop_vs @ tileline blank
\r
758 and r3, r9, #0x6000
\r
759 add r3, r7, r3, lsr #9 @ r3=pal=((code&0x6000)>>9);
\r
762 beq .DrawStrip_vs_SingleColor @ tileline singlecolor
\r
765 bne .DrawStrip_vs_TileFlip
\r
767 @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r0: helper pattern
\r
768 .DrawStrip_vs_TileNorm:
\r
772 .DrawStrip_vs_TileFlip:
\r
776 .DrawStrip_vs_SingleColor:
\r
779 orr r4, r4, r4, lsl #8
\r
780 tst r1, #1 @ not aligned?
\r
781 strneb r4, [r1], #1
\r
782 streqh r4, [r1], #2
\r
786 strneb r4, [r1], #1 @ have a remaining unaligned pixel?
\r
789 .DrawStrip_vs_hiprio:
\r
790 tst r10, #(1<<23) @ sh[23]
\r
791 tsteq r2, r2 @ if (!sh[23] && code==blank) continue
\r
794 @ orr r10, r10, #1<<22 @ hi_not_empty[22]
\r
796 orr r7, r9, r7, lsl #16
\r
797 orr r7, r7, r10, lsl #25 @ (ty<<25)
\r
799 eorne r7, r7, #7<<26 @ if(code&0x1000) cval^=7<<26;
\r
800 str r7, [r6], #4 @ cache hi priority tile code
\r
801 str r2, [r6], #4 @ cache hi priority tile data
\r
805 tst r8, #(1<<24) @ seen non hi-prio tile
\r
806 ldr r1, [sp, #9*4] @ est
\r
808 ldreq r2, [r1, #OFS_EST_rendstatus]
\r
809 str r0, [r6] @ terminate the cache list
\r
810 orreq r2, r2, #PDRAW_PLANE_HI_PRIO @ had a layer with all hi-prio tiles
\r
811 streq r2, [r1, #OFS_EST_rendstatus]
\r
813 ldmfd sp!, {r4-r11,pc}
\r
816 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
818 @ interlace mode 2? Sonic 2?
\r
819 .DrawStrip_interlace:
\r
821 movne r7, r7, lsr #16
\r
822 mov r7, r7, lsl #21
\r
824 @ Find the line in the name table
\r
825 add r2, r7, r2, lsl #22 @ r2=(vscroll+(DrawScanline<<1))<<21 (11 bits);
\r
826 orr r1, r1, #0x80000000
\r
827 and r2, r2, r1, ror #10 @ &((ymask<<1)|1)<<21;
\r
828 mov r2, r2, lsr #21
\r
830 mov r12, r12, lsr #1 @ halfwords
\r
831 add r0, r12, r4, lsl r10 @ nametab+=(ts.line>>4)<<shift[width];
\r
835 stmia sp, {r0,r2,r3,r5,r6,r9}
\r
838 mov r1, r9, lsr #29
\r
839 bl DrawStripInterlace @ struct TileStrip *ts, int plane_sh
\r
842 ldmfd sp!, {r4-r11,pc}
\r
846 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
848 @ void BackFill(int reg7, int sh, struct PicoEState *est)
\r
853 stmfd sp!, {r4-r9,lr}
\r
855 ldr lr, [r2, #OFS_EST_HighCol]
\r
856 orr r0, r0, r1, lsl #7
\r
857 orr r0, r0, r0, lsl #8
\r
858 orr r0, r0, r0, lsl #16
\r
870 stmia lr!, {r0-r7} @ 10*8*4
\r
881 ldmfd sp!, {r4-r9,pc}
\r
884 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
886 @ void DrawTilesFromCache(int *hc, int sh, int rlim, struct PicoEState *est)
\r
888 .global DrawTilesFromCache
\r
890 DrawTilesFromCache:
\r
891 stmfd sp!, {r4-r9,r11,lr}
\r
893 @ cache some stuff to avoid mem access
\r
894 ldr r11,[r3, #OFS_EST_HighCol]
\r
896 ldr lr, [r3, #OFS_EST_PicoMem_vram]
\r
900 orr r8, r8, r2, lsl #1
\r
901 bne .dtfc_check_rendflags
\r
905 ldr r6, [r0], #4 @ read code
\r
906 movs r1, r6, lsr #16 @ r1=dx;
\r
907 ldmeqfd sp!, {r4-r9,r11,pc} @ dx is never zero, this must be a terminator, return
\r
908 bic r4, r1, #0xfe00
\r
909 add r1, r11, r4 @ r1=pdest
\r
911 ldr r2, [r0], #4 @ read pixel data
\r
913 and r3, r6, #0x6000
\r
914 mov r3, r3, lsr #9 @ r3=pal=((code&0x6000)>>9);
\r
916 rsbs r4, r4, r8, lsr #1
\r
926 beq .dtfc_SingleColor @ tileline singlecolor
\r
931 @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern
\r
943 orr r4, r4, r4, lsl #8
\r
944 tst r1, #1 @ not aligned?
\r
945 strneb r4, [r1], #1
\r
946 streqh r4, [r1], #2
\r
950 strneb r4, [r1], #1 @ have a remaining unaligned pixel?
\r
955 beq .dtfc_shadow_blank
\r
958 beq .dtfc_SingleColor @ tileline singlecolor
\r
961 bne .dtfc_TileFlipShHP
\r
963 @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern
\r
964 .dtfc_TileNormShHP:
\r
968 .dtfc_TileFlipShHP:
\r
972 .dtfc_shadow_blank:
\r
977 strneb r4, [r1], #1
\r
979 orr r6, r6, r6, lsl #8
\r
996 ble .dtfc_loop @ off limits
\r
998 rsb r4, r4, #0 @ 1-7
\r
1001 tst r6, #0x0800 @ flipped?
\r
1002 moveq r12,r12, lsl r4
\r
1003 movne r12,r12, lsr r4
\r
1004 and r2, r2, r12, ror #16
\r
1011 beq .dtfc_TileNorm
\r
1014 @ check if we have detected layer covered with hi-prio tiles:
\r
1015 .dtfc_check_rendflags:
\r
1016 ldr r2, [r9, #OFS_EST_rendstatus]
\r
1017 tst r2, #(PDRAW_PLANE_HI_PRIO|PDRAW_SHHI_DONE)
\r
1019 bic r8, r8, #1 @ sh/hi mode off
\r
1020 tst r2, #PDRAW_SHHI_DONE
\r
1021 bne .dtfc_loop @ already processed
\r
1022 orr r2, r2, #PDRAW_SHHI_DONE
\r
1023 str r2, [r9, #OFS_EST_rendstatus]
\r
1028 orr r6, r6, r6, lsl #8
\r
1029 orr r6, r6, r6, lsl #16
\r
1030 .dtfc_loop_shprep:
\r
1031 ldmia r1, {r2,r4,r5,r7}
\r
1037 stmia r1!,{r2,r4,r5,r7}
\r
1038 bne .dtfc_loop_shprep
\r
1044 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
1047 @ void DrawSpritesSHi(unsigned char *sprited, struct PicoEState *est)
\r
1049 .global DrawSpritesSHi
\r
1054 ands r3, r3, #0x7f
\r
1057 stmfd sp!, {r1,r3-r11,lr} @ +est
\r
1058 strb r12,[r0,#3] @ set end marker
\r
1060 add r10,r0, #4 @ r10=HighLnSpr end
\r
1062 tst r12,#0x6 @ masking in slot 1 and tile ovfl?
\r
1063 ldmeqfd sp!, {r1,r3-r11,pc}
\r
1064 add r10,r10,r3 @ r10=HighLnSpr end
\r
1066 ldrb r12,[r10,#0] @ width of last sprite
\r
1067 ldr r11,[r1, #OFS_EST_HighCol]
\r
1070 ldr lr, [r1, #OFS_EST_PicoMem_vram]
\r
1074 @ draw next sprite
\r
1075 ldr r7, [sp] @ est
\r
1076 ldrb r0, [r10,#-1]!
\r
1077 ldr r1, [r7, #OFS_EST_HighPreSpr]
\r
1079 ldmeqfd sp!, {r1,r3-r11,pc} @ end of list
\r
1081 add r0, r1, r0, lsl #3
\r
1083 ldr r9, [r0, #4] @ sprite[1]
\r
1084 mov r2, r9, asr #16 @ r2=sx
\r
1086 mov r9, r9, lsl #16
\r
1087 mov r3, r9, lsr #31 @ priority
\r
1088 mov r9, r9, lsr #16
\r
1089 @ orr r9, r9, r8, lsl #31 @ r9=code|sh[31] @@ sh is always on here now
\r
1090 and r4, r9, #0x6000
\r
1091 orr r9, r9, r4, lsl #16
\r
1092 orr r9, r9, #0x90000000 @ r9=scc1 ???? ... <code> (s=shadow/hilight, cc=pal)
\r
1093 cmp r12,r9, lsr #28 @ sh/hi with pal3?
\r
1094 cmpne r3, #1 @ if not, is it hi prio?
\r
1095 strne r3, [sp, #4] @ reset last sprite width
\r
1096 bne DrawSpriteSHi @ non-operator low sprite, already drawn
\r
1098 ldr r3, [r0] @ sprite[0]
\r
1099 mov r6, r3, lsr #28
\r
1100 sub r6, r6, #1 @ r6=width-1 (inc later)
\r
1101 mov r5, r3, lsr #24
\r
1102 and r5, r5, #7 @ r5=height
\r
1104 ldr r7, [r7, #OFS_EST_DrawScanline]
\r
1105 mov r0, r3, lsl #16 @ r4=sy<<16 (tmp)
\r
1107 sub r7, r7, r0, asr #16 @ r7=row=DrawScanline-sy
\r
1110 movne r0, r5, lsl #3
\r
1112 subne r7, r0, r7 @ if (code&0x1000) row=(height<<3)-1-row; // Flip Y
\r
1114 add r8, r9, r7, lsr #3 @ tile+=row>>3; // Tile number increases going down
\r
1116 mlane r8, r5, r6, r8 @ if (code&0x0800) { tile+=delta*(width-1);
\r
1117 rsbne r5, r5, #0 @ delta=-delta; } // r5=delta now
\r
1119 mov r8, r8, lsl #21
\r
1120 mov r8, r8, lsr #17
\r
1122 add r8, r8, r7, lsl #1 @ tile+=(row&7)<<1; // Tile address
\r
1125 add r6, r6, #1 @ inc now
\r
1126 cmp r0, #0 @ check width of last sprite
\r
1129 strne r0, [sp, #4]
\r
1131 mov r5, r5, lsl #4 @ delta<<=4; // Delta of address
\r
1132 mov r3, r4, lsr #9 @ r3=pal=((code>>9)&0x30);
\r
1134 adds r0, r2, #0 @ mov sx to r0 and set ZV flags
\r
1135 b .dsprShi_loop_enter
\r
1138 subs r6, r6, #1 @ width--
\r
1140 adds r0, r0, #8 @ sx+=8
\r
1141 add r8, r8, r5 @ tile+=delta
\r
1143 .dsprShi_loop_enter:
\r
1144 ble .dsprShi_loop @ sx <= 0
\r
1148 bic r8, r8, #0xf8000 @ tile&=0x7fff; // Clip tile address
\r
1149 ldr r2, [lr, r8, lsl #1] @ pack=*(unsigned int *)(PicoMem.vram+addr); // Get 8 pixels
\r
1150 add r1, r11, r0 @ r1=pdest
\r
1154 cmp r12, r9, lsr #28
\r
1155 beq .dsprShi_shadow
\r
1157 cmp r2, r2, ror #4
\r
1158 beq .dsprShi_SingleColor @ tileline singlecolor
\r
1161 bne .dsprShi_TileFlip
\r
1163 @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern
\r
1165 .dsprShi_TileNorm:
\r
1169 .dsprShi_TileFlip:
\r
1173 .dsprShi_SingleColor:
\r
1176 orr r4, r4, r4, lsl #8
\r
1177 tst r0, #1 @ not aligned?
\r
1178 strneb r4, [r1], #1
\r
1179 streqh r4, [r1], #2
\r
1183 strneb r4, [r1], #1
\r
1188 beq .dsprShi_shadow_lowpri
\r
1190 cmp r2, r2, ror #4
\r
1191 beq .dsprShi_singlec_sh
\r
1194 bne .dsprShi_TileFlip_sh
\r
1196 @ (r1=pdest, r2=pixels8, r3=pal) r4, r7: scratch, r12: helper pattern
\r
1197 .dsprShi_TileNorm_sh:
\r
1201 .dsprShi_TileFlip_sh:
\r
1205 .dsprShi_singlec_sh:
\r
1206 cmp r2, #0xe0000000
\r
1207 bcc .dsprShi_SingleColor @ normal singlecolor tileline (carry inverted in ARM)
\r
1208 tst r2, #0x10000000
\r
1209 bne .dsprShi_sh_sh
\r
1217 .dsprShi_shadow_lowpri:
\r
1219 bne .dsprShi_TileFlip_sh_lp
\r
1221 .dsprShi_TileNorm_sh_lp:
\r
1222 TileNormSh_onlyop_lp
\r
1225 .dsprShi_TileFlip_sh_lp:
\r
1226 TileFlipSh_onlyop_lp
\r
1231 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
1233 @ void DrawAllSprites(unsigned char *sprited, int prio, int sh,
\r
1234 @ struct PicoEState *est)
\r
1236 .global DrawAllSprites
\r
1239 orr r1, r2, r1, lsl #1
\r
1241 ands r2, r2, #0x7f
\r
1244 @ time to do some real work
\r
1245 stmfd sp!, {r1,r3-r11,lr} @ +sh|prio<<1 +est
\r
1247 strb r12,[r0,#3] @ set end marker
\r
1251 tst r12,#0x6 @ masking in slot 1 and tile ovfl?
\r
1252 ldmeqfd sp!, {r1,r3-r11,pc}
\r
1253 add r10,r10,r2 @ r10=HighLnSpr end
\r
1255 ldrb r12,[r10,#0] @ width of last sprite
\r
1256 ldr r11,[r3, #OFS_EST_HighCol]
\r
1257 orr r1 ,r1 ,r12,lsl #24
\r
1260 ldr lr, [r3, #OFS_EST_PicoMem_vram]
\r
1262 @ + 0 : hhhhvvvv ----hhvv yyyyyyyy yyyyyyyy // v, h: horiz. size
\r
1263 @ + 4 : xxxxxxxx xxxxxxxx pccvhnnn nnnnnnnn // x: x coord + 8
\r
1266 @ draw next sprite
\r
1267 ldrb r0, [r10,#-1]!
\r
1268 ldr r4, [sp] @ sh|prio<<1|lastw<<24
\r
1269 ldr r7, [sp, #4] @ est
\r
1270 mov r2, r0, lsl #24
\r
1272 ldmeqfd sp!, {r1,r3-r11,pc} @ end of list
\r
1273 eors r2, r2, r4, lsl #30
\r
1274 bic r2, r4, #0xff000000
\r
1276 bmi DrawSprite @ wrong priority
\r
1277 ldr r1, [r7, #OFS_EST_HighPreSpr]
\r
1279 add r0, r1, r0, lsl #3
\r
1281 ldr r3, [r0] @ sprite[0]
\r
1282 ldr r7, [r7, #OFS_EST_DrawScanline]
\r
1283 mov r6, r3, lsr #28
\r
1284 sub r6, r6, #1 @ r6=width-1 (inc later)
\r
1285 mov r5, r3, lsr #24
\r
1286 and r5, r5, #7 @ r5=height
\r
1288 mov r8, r3, lsl #16 @ r8=sy<<16 (tmp)
\r
1291 sub r7, r7, r8, asr #16 @ r7=row=DrawScanline-sy
\r
1293 mov r2, r9, asr #16 @ r2=sx
\r
1294 mov r9, r9, lsl #16
\r
1295 mov r9, r9, lsr #16
\r
1296 orr r9, r9, r4, lsl #31 @ r9=code|sh[31]
\r
1299 movne r8, r5, lsl #3
\r
1301 subne r7, r8, r7 @ if (code&0x1000) row=(height<<3)-1-row; // Flip Y
\r
1303 add r8, r9, r7, lsr #3 @ tile+=row>>3; // Tile number increases going down
\r
1305 mlane r8, r5, r6, r8 @ if (code&0x0800) { tile+=delta*(width-1);
\r
1306 rsbne r5, r5, #0 @ delta=-delta; } // r5=delta now
\r
1308 mov r8, r8, lsl #21
\r
1309 mov r8, r8, lsr #17
\r
1311 add r8, r8, r7, lsl #1 @ tile+=(row&7)<<1; // Tile address
\r
1313 add r6, r6, #1 @ inc now
\r
1314 cmp r4, #0x1000000 @ check width of last sprite
\r
1315 movhs r6, r4, lsr #24
\r
1317 @ cache some stuff to avoid mem access
\r
1318 mov r5, r5, lsl #4 @ delta<<=4; // Delta of address
\r
1319 and r4, r9, #0x6000
\r
1320 orr r9, r9, r4, lsl #16
\r
1321 orrs r9, r9, #0x10000000 @ r9=scc1 ???? ... <code> (s=shadow/hilight, cc=pal)
\r
1323 mov r3, r4, lsr #9 @ r3=pal=((code>>9)&0x30);
\r
1324 orrmi r3, r3, #0x80 @ for sh/hi
\r
1326 adds r0, r2, #0 @ mov sx to r0 and set ZV flags
\r
1327 b .dspr_loop_enter
\r
1330 subs r6, r6, #1 @ width--
\r
1332 adds r0, r0, #8 @ sx+=8
\r
1333 add r8, r8, r5 @ tile+=delta
\r
1336 ble .dspr_loop @ sx <= 0
\r
1340 bic r8, r8, #0xf8000 @ tile&=0x7fff; // Clip tile address
\r
1341 ldr r2, [lr, r8, lsl #1] @ pack=*(unsigned int *)(PicoMem.vram+addr); // Get 8 pixels
\r
1342 add r1, r11, r0 @ r1=pdest
\r
1346 cmp r12, r9, lsr #28
\r
1349 tst r9, #0x80000000
\r
1352 cmp r2, r2, ror #4
\r
1353 beq .dspr_SingleColor @ tileline singlecolor
\r
1356 bne .dspr_TileFlip
\r
1358 @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern
\r
1369 cmp r2, #0xe0000000
\r
1370 bcs .dspr_TileNorm_sh @ op. tileline, markop. XXX: maybe add a spec. handler?
\r
1372 .dspr_SingleColor:
\r
1375 orr r4, r4, r4, lsl #8
\r
1376 tst r0, #1 @ not aligned?
\r
1377 strneb r4, [r1], #1
\r
1378 streqh r4, [r1], #2
\r
1382 strneb r4, [r1], #1
\r
1387 bne .dspr_TileFlipNonSH
\r
1389 @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern
\r
1391 .dspr_TileNormNonSH:
\r
1395 .dspr_TileFlipNonSH:
\r
1400 cmp r2, r2, ror #4
\r
1401 beq .dspr_singlec_sh
\r
1404 bne .dspr_TileFlip_sh
\r
1406 @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r12: helper pattern
\r
1407 .dspr_TileNorm_sh:
\r
1411 .dspr_TileFlip_sh:
\r
1416 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
1418 @ void DrawWindow(int tstart, int tend, int prio, int sh
\r
1419 @ struct PicoEState *est)
\r
1421 .global DrawWindow
\r
1424 ldr r12, [sp] @ est
\r
1425 stmfd sp!, {r4-r11,lr}
\r
1427 ldr r6, [r12, #OFS_EST_Pico]
\r
1428 ldr r10, [r12, #OFS_EST_DrawScanline]
\r
1429 mov r11, r12 @ est
\r
1430 ldrb r12, [r6, #OFS_Pico_video_reg+3] @ pvid->reg[3]
\r
1432 ldr r4, [r6, #OFS_Pico_video_reg+12]
\r
1433 mov r5, r10, lsr #3
\r
1435 mov r10, r10, lsl #1 @ r10=ty
\r
1437 ldr r6, [r11, #OFS_EST_rendstatus]
\r
1438 ldr lr, [r11, #OFS_EST_PicoMem_vram]
\r
1440 mov r12, r12, lsl #10
\r
1442 tst r4, #1 @ 40 cell mode?
\r
1443 andne r12, r12, #0xf000 @ 0x3c<<10
\r
1444 andeq r12, r12, #0xf800
\r
1445 addne r12, r12, r5, lsl #7
\r
1446 addeq r12, r12, r5, lsl #6 @ nametab
\r
1447 add r12, r12, r0, lsl #2 @ +starttile
\r
1449 ands r6, r6, #PDRAW_WND_DIFF_PRIO
\r
1450 cmpeq r2, #1 @ prio && !(rendstatus & WND_DIFF_PRIO)?
\r
1451 ldmeqfd sp!, {r4-r11,pc} @ yes, assume that whole window uses same priority
\r
1454 orr r6, r6, r3, lsl #8 @ shadow mode
\r
1458 @ cache some stuff to avoid mem access
\r
1459 ldr r11, [r11, #OFS_EST_HighCol]
\r
1460 mov r8, r8, lsl #1 @ cells
\r
1461 mvn r9, #0 @ r9=prevcode=-1
\r
1462 add r1, r11, r0, lsl #4 @ r1=pdest=HighCol+starttile (+8 added in loop)
\r
1465 @ r4,r5 are scratch in this loop
\r
1469 ldrh r7, [lr, r12] @ r7=code (int, but from unsigned, no sign extend)
\r
1470 add r12, r12, #2 @ halfwords
\r
1472 bmi .dwloop_end @ done
\r
1474 eor r5, r6, r7, lsr #15
\r
1476 orrne r6, r6, #PDRAW_WND_DIFF_PRIO @ wrong pri
\r
1480 beq .dw_samecode @ we know stuff about this tile already
\r
1482 mov r9, r7 @ remember code
\r
1484 tst r9, #0x1000 @ if (code&0x1000)
\r
1485 mov r2, r9, lsl #21
\r
1486 add r2, r2, r10, lsl #17
\r
1487 eorne r2, r2, #0xe<<17 @ if (code&0x1000) addr^=0xe;
\r
1489 ldr r2, [lr, r2, lsr #16] @ pack=*(unsigned int *)(Pico.vram+addr); // Get 8 pixels
\r
1491 and r3, r9, #0x6000
\r
1492 mov r3, r3, lsr #9 @ r3=pal=((code&0x6000)>>9);
\r
1499 beq .dwloop @ tileline blank
\r
1501 cmp r2, r2, ror #4
\r
1502 beq .dw_SingleColor @ tileline singlecolor
\r
1507 @ (r1=pdest, r2=pixels8, r3=pal) r4: scratch, r0: helper pattern
\r
1517 and r4, r0, r2 @ #0x0000000f
\r
1519 orr r4, r4, r4, lsl #8
\r
1520 orr r4, r4, r4, lsl #16
\r
1522 stmia r1!, {r4,r5}
\r
1523 b .dwloop_nor1 @ we incremeted r1 ourselves
\r
1526 tst r6, #1 @ hi pri?
\r
1527 orreq r3, r3, #0x80
\r
1528 beq .dw_shadow_done
\r
1531 orr r5, r5, r5, lsl #8
\r
1532 orr r5, r5, r5, lsl #16
\r
1541 and r2, r6, #PDRAW_WND_DIFF_PRIO
\r
1542 ldmfd sp!, {r4-r11,lr}
\r
1544 ldr r1, [r0, #OFS_EST_rendstatus]
\r
1546 str r1, [r0, #OFS_EST_rendstatus]
\r
1551 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
1554 @ Convert 0000bbb0 ggg0rrr0
\r
1555 @ to rrrrrggg gggbbbbb
\r
1557 @ r2,r3 - scratch, lr = 0x001c001c, r8 = 0x08610861
\r
1558 .macro convRGB565 reg
\r
1559 and r2, lr, \reg,lsr #7 @ b
\r
1560 and r3, lr, \reg,lsr #3 @ g
\r
1561 and \reg, lr, \reg,lsl #1 @ r
\r
1562 orr r2, r2, r3, lsl #6
\r
1563 orr \reg, r2, \reg,lsl #11
\r
1565 and r2, r8, \reg,lsr #4
\r
1566 orr \reg, \reg, r2
\r
1569 @ trashes: r2-r8,r12,lr; r8 = 0x08610861; r0,r1 are advanced
\r
1570 .macro vidConvCpyRGB565_local
\r
1571 mov r12, r2, lsr #3 @ repeats
\r
1572 mov lr, #0x001c0000
\r
1573 orr lr, lr, #0x01c @ lr == pattern 0x001c001c
\r
1576 ldmia r1!, {r4-r7}
\r
1591 .global vidConvCpyRGB565
\r
1593 vidConvCpyRGB565: @ void *to, void *from, int pixels
\r
1594 stmfd sp!, {r4-r9,lr}
\r
1596 orr r8, r8, #0x0800
\r
1597 orr r8, r8, r8, lsl #16
\r
1598 vidConvCpyRGB565_local
\r
1599 ldmfd sp!, {r4-r9,pc}
\r
1602 @ void PicoDoHighPal555(int sh, int line, struct PicoEState *est)
\r
1604 .global PicoDoHighPal555
\r
1607 stmfd sp!, {r4-r10,lr}
\r
1609 ldr r8, [r10, #OFS_EST_Pico]
\r
1613 add r0, r10, #OFS_EST_HighPal
\r
1616 strb r1, [r8, #OFS_Pico_m_dirtyPal]
\r
1618 ldr r1, [r10, #OFS_EST_PicoMem_cram]
\r
1621 orr r8, r8, #0x0800
\r
1622 orr r8, r8, r8, lsl #16
\r
1624 vidConvCpyRGB565_local
\r
1627 beq PicoDoHighPal555_end
\r
1629 add r3, r10, #OFS_EST_HighPal
\r
1630 add r4, r3, #0x40*2
\r
1632 @ hilighted pixels (0x40-0x7f):
\r
1633 @ t = ((dpal[i] >> 1) & 0x738e738e) + 0x738e738e;
\r
1634 @ t |= (t >> 4) & 0x08610861;
\r
1637 orr r12,r12,#0x7300
\r
1638 orr r12,r12,r12,lsl #16
\r
1640 .fl_loopcpRGB555_hi:
\r
1641 ldmia r3!, {r1,r6}
\r
1642 and r1, r12, r1, lsr #1
\r
1643 and r6, r12, r6, lsr #1
\r
1646 and r5, r8, r1, lsr #4
\r
1647 and r7, r8, r6, lsr #4
\r
1650 stmia r4!, {r1,r6}
\r
1652 bne .fl_loopcpRGB555_hi
\r
1654 sub r3, r3, #0x40*2
\r
1655 @ shadowed (0x80-0xbf), shadow|hilight (aka normal, 0xc0-0xff) pixels:
\r
1656 add r5, r3, #0xc0*2
\r
1658 .fl_loopcpRGB555_sh:
\r
1659 ldmia r3!, {r1,r6}
\r
1661 stmia r5!, {r1,r6} @ 0xc0, normal
\r
1662 and r1, r12, r1, lsr #1
\r
1663 and r6, r12, r6, lsr #1
\r
1664 stmia r4!, {r1,r6}
\r
1665 bne .fl_loopcpRGB555_sh
\r
1668 PicoDoHighPal555_end:
\r
1669 ldmfd sp!, {r4-r10,pc}
\r
1672 @ void FinalizeLine555(int sh, int line, struct PicoEState *est)
\r
1674 .global FinalizeLine555
\r
1677 ldr r3, [r2, #OFS_EST_rendstatus]
\r
1679 tst r3, #PDRAW_BGC_DMA
\r
1682 stmfd sp!, {r4-r11,lr}
\r
1686 bl PicoDrawUpdateHighPal
\r
1688 ldr r8, [r11, #OFS_EST_Pico]
\r
1689 add r3, r11, #OFS_EST_HighPal
\r
1692 mov lr, lr, lsl #1
\r
1694 ldr r5, [r11, #OFS_EST_PicoOpt]
\r
1695 ldr r1, [r11, #OFS_EST_HighCol]
\r
1696 ldr r0, [r11, #OFS_EST_DrawLineDest]
\r
1697 ldr r7, [r5, #OFS_PicoIn_AHW-OFS_PicoIn_opt]
\r
1698 ldrb r12,[r8, #OFS_Pico_video_reg+12]
\r
1699 ldrb r6, [r8, #OFS_Pico_video_reg+0]
\r
1700 ldr r2, [r8, #OFS_Pico_m_hardware]
\r
1703 tst r7, #0x20 @ GG ?
\r
1704 tstne r2, #0x2 @ LCD ?
\r
1707 tst r7, #0x10 @ SMS ?
\r
1711 movne r2, #248/8 @ len = 248
\r
1712 addne r1, r1, #8 @ ps += 8
\r
1713 moveq r2, #256/8 @ len = 256
\r
1714 b .fl_check32scaling
\r
1717 mov r2, #160/8 @ len = 160
\r
1718 tst r4, #PDRAW_SOFTSCALE
\r
1719 bne .fl_20scale_RGB555 @ scale 160->320
\r
1723 tst r12, #1 @ h32?
\r
1724 movne r2, #320/8 @ len = 320
\r
1725 bne .fl_40colRGB555
\r
1726 mov r2, #256/8 @ len = 256
\r
1728 .fl_check32scaling:
\r
1729 tst r4, #PDRAW_SOFTSCALE
\r
1730 rsbne r7, r2, #256/8
\r
1731 addne r0, r0, r7, lsl #3 @ pd += (256-len)>>1
\r
1732 bne .fl_32scale_RGB555 @ scale 256->320
\r
1735 tst r4, #PDRAW_BORDER_32
\r
1736 rsbne r7, r2, #320/8 @ pd += (320-len)/2
\r
1737 addne r0, r0, r7, lsl #3
\r
1740 #ifdef UNALIGNED_DRAWLINEDEST
\r
1741 @ this is basically for Gizmondo, which has unaligned odd lines in the framebuffer
\r
1750 and r4, lr, r12, lsl #1
\r
1752 and r5, lr, r12, lsr #7
\r
1754 and r6, lr, r12, lsr #15
\r
1756 orr r4, r4, r5, lsl #16
\r
1758 and r5, lr, r12, lsr #23
\r
1760 and r8, lr, r7, lsl #1
\r
1762 orr r5, r6, r5, lsl #16
\r
1764 and r6, lr, r7, lsr #7
\r
1766 and r12,lr, r7, lsr #15
\r
1767 ldrh r12,[r3, r12]
\r
1768 and r7, lr, r7, lsr #23
\r
1770 orr r8, r8, r6, lsl #16
\r
1773 orr r12,r12, r7, lsl #16
\r
1775 stmia r0!, {r4,r5,r8,r12}
\r
1776 bne .fl_loopRGB555
\r
1778 ldmfd sp!, {r4-r11,pc}
\r
1781 .fl_32scale_RGB555:
\r
1782 ldr r5, [r5, #OFS_PicoIn_filter-OFS_PicoIn_opt]
\r
1784 mov r9, #0xf700 @ f800 07e0 001f | e000 0780 001c | 3800 01e0 0007
\r
1785 orr r9, r9, #0x00de
\r
1787 #ifdef UNALIGNED_DRAWLINEDEST
\r
1789 bne .fl_32scale_RGB555u
\r
1793 add pc, pc, r5, lsl #2
\r
1804 and r4, lr, r12, lsl #1
\r
1806 and r5, lr, r12, lsr #7
\r
1808 and r6, lr, r12, lsr #15
\r
1810 and r10,lr, r12, lsr #23
\r
1811 ldrh r10,[r3, r10]
\r
1813 orr r4, r4, r5, lsl #16
\r
1814 orr r5, r6, r6, lsl #16
\r
1816 and r6, lr, r7, lsl #1
\r
1818 and r8, lr, r7, lsr #7
\r
1820 and r12,lr, r7, lsr #15
\r
1821 ldrh r12,[r3, r12]
\r
1822 and r7, lr, r7, lsr #23
\r
1825 orr r6, r10,r6, lsl #16
\r
1826 orr r8, r8,r12, lsl #16
\r
1830 orr r10,r12,r7, lsl #16
\r
1832 stmia r0!, {r4,r5,r6,r8,r10}
\r
1833 bne .fl_32scale_nn
\r
1835 b .fl_32scale_8bit
\r
1841 and r4, lr, r12, lsl #1
\r
1843 and r5, lr, r12, lsr #7
\r
1845 and r6, lr, r12, lsr #15
\r
1847 and r10,lr, r12, lsr #23
\r
1848 ldrh r10,[r3, r10]
\r
1852 orr r4, r4, r5, lsl #16
\r
1855 mov r5, r5, lsr #1
\r
1856 orr r5, r5, r6, lsl #16
\r
1858 and r6, lr, r7, lsl #1
\r
1860 and r8, lr, r7, lsr #7
\r
1862 and r12,lr, r7, lsr #15
\r
1863 ldrh r12,[r3, r12]
\r
1864 and r7, lr, r7, lsr #23
\r
1869 orr r6, r10,r6, lsl #16
\r
1872 orr r10,r12,r7, lsl #16
\r
1876 mov r12,r12,lsr #1
\r
1877 orr r8, r8,r12, lsl #16
\r
1881 stmia r0!, {r4,r5,r6,r8,r10}
\r
1882 bne .fl_32scale_snn
\r
1884 b .fl_32scale_8bit
\r
1890 and r4, lr, r12, lsl #1
\r
1892 and r5, lr, r12, lsr #7
\r
1894 and r6, lr, r12, lsr #15
\r
1900 mov r10,r10,lsr #1
\r
1901 orr r4, r4, r10,lsl #16 @ px0 | (px0+px1)/2
\r
1905 mov r5, r5, lsr #1
\r
1906 orr r5, r5, r6, lsl #16 @ (px1+px2)/2 | px2
\r
1908 and r10,lr, r12, lsr #23
\r
1909 ldrh r10,[r3, r10]
\r
1910 and r8, lr, r7, lsl #1
\r
1915 orr r6, r10,r8, lsl #16 @ px3 | px4
\r
1917 and r12,lr, r7, lsr #15
\r
1918 ldrh r12,[r3, r12]
\r
1919 and r10, lr, r7, lsr #23
\r
1920 ldrh r10, [r3, r10]
\r
1921 and r7, lr, r7, lsr #7
\r
1926 orr r10,r12,r10, lsl #16 @ px6 | px7
\r
1931 mov r8, r8, lsr #1
\r
1932 mov r12,r12,lsr #1
\r
1933 orr r8, r8,r12, lsl #16 @ (px4+px5)/2 | (px5+px6)/2
\r
1937 stmia r0!, {r4,r5,r6,r8,r10}
\r
1938 bne .fl_32scale_bl2
\r
1940 b .fl_32scale_8bit
\r
1943 // TODO this should reflect the bl4 C algorithm, but it doesn't, it's bln.
\r
1944 and r9, r9, r9, lsl #1 @ nuke 2 LSBs to avoid spilling for n/4
\r
1949 and r4, lr, r12,lsl #1
\r
1951 and r5, lr, r12,lsr #7
\r
1954 @ r4 = 1/4px0+3/4px1 : px0
\r
1956 orr r4, r4, r4, lsl #14 @ r4[31:16] = 1/4 pix_s 0
\r
1958 sub r6, r5, r5, lsr #2 @ r6 = 3/4 pix_s 1
\r
1959 add r4, r4, r6, lsl #16 @ pix_d 0, 1
\r
1961 and r6, lr, r12,lsr #15
\r
1963 and r12,lr, r12,lsr #23
\r
1964 ldrh r12,[r3, r12]
\r
1966 @ r5 = 3/4px2+1/4px3 : (px1+px2)/2
\r
1969 mov r5, r5, lsr #1
\r
1970 sub r6, r6, r6, lsr #2 @ r6 = 3/4 pix_s 2
\r
1971 orr r5, r5, r6, lsl #16
\r
1973 and r6, lr, r7, lsl #1
\r
1976 add r5, r5, r12,lsl #14 @ pix_d 2, 3
\r
1980 orr r6, r12,r6, lsl #16 @ pix_d 4, 5
\r
1982 @ r8 = (px5+px6)/2 : 1/4px4+3/4px5
\r
1983 and r12,lr, r7, lsr #7
\r
1984 ldrh r12,[r3, r12]
\r
1985 and r10,lr, r7, lsr #15
\r
1986 ldrh r10,[r3, r10]
\r
1988 sub r8, r12,r12,lsr #2 @ r8 = 3/4 pix_s 1
\r
1989 add r8, r8, r6, lsr #18
\r
1991 and r7, lr, r7, lsr #23
\r
1994 orr r8, r8, r10,lsl #15
\r
1995 add r8, r8, r12,lsl #15 @ pix_d 6, 7
\r
1997 @ r10 = px7 : 3/4px6+1/4px7
\r
1998 sub r10,r10,r10,lsr #2 @ r10= 3/4 pix_s 2
\r
2000 add r10,r10,r7, lsr #2 @ += 1/4 pix_s 3
\r
2001 orr r10,r10,r7, lsl #16 @ pix_d 8, 9
\r
2005 stmia r0!, {r4,r5,r6,r8,r10}
\r
2006 bne .fl_32loop_bl4
\r
2009 ldr r4, [r11, #OFS_EST_rendstatus]
\r
2010 add r0, r1, #320-256
\r
2012 tst r4, #PDRAW_32X_SCALE
\r
2013 ldmeqfd sp!, {r4-r11,pc}
\r
2016 .fl_32scale_8bit_nn:
\r
2017 ldr r7, [r1, #-4]!
\r
2018 ldr r12, [r1, #-4]!
\r
2020 and r4, lr, r12, lsl #0
\r
2021 and r5, lr, r12, lsr #8
\r
2022 and r6, lr, r12, lsr #16
\r
2023 and r10,lr, r12, lsr #24
\r
2025 orr r4, r4, r5, lsl #8
\r
2026 orr r5, r6, r6, lsl #8
\r
2028 and r6, lr, r7, lsl #0
\r
2029 and r8, lr, r7, lsr #8
\r
2030 and r12,lr, r7, lsr #16
\r
2031 and r7, lr, r7, lsr #24
\r
2033 orr r6, r10,r6, lsl #8
\r
2034 orr r8, r8,r12, lsl #8
\r
2038 orr r10,r12,r7, lsl #8
\r
2040 strh r10, [r0, #-2]!
\r
2041 strh r8, [r0, #-2]!
\r
2042 strh r6, [r0, #-2]!
\r
2043 strh r5, [r0, #-2]!
\r
2044 strh r4, [r0, #-2]!
\r
2046 bne .fl_32scale_8bit_nn
\r
2048 ldmfd sp!, {r4-r11,pc}
\r
2051 .fl_20scale_RGB555:
\r
2052 ldr r5, [r5, #OFS_PicoIn_filter-OFS_PicoIn_opt]
\r
2054 mov r9, #0xf700 @ f800 07e0 001f | e000 0780 001c | 3800 01e0 0007
\r
2055 orr r9, r9, #0x00de
\r
2057 #ifdef UNALIGNED_DRAWLINEDEST
\r
2059 bne .fl_20scale_RGB555u
\r
2063 add pc, pc, r5, lsl #1
\r
2072 and r4, lr, r12, lsl #1
\r
2074 and r5, lr, r12, lsr #7
\r
2076 and r6, lr, r12, lsr #15
\r
2078 and r8 ,lr, r12, lsr #23
\r
2081 orr r4, r4, r4, lsl #16
\r
2082 orr r5, r5, r5, lsl #16
\r
2083 orr r6, r6, r6, lsl #16
\r
2084 orr r8, r8, r8, lsl #16
\r
2085 stmia r0!, {r4,r5,r6,r8}
\r
2087 and r4, lr, r7, lsl #1
\r
2089 and r5, lr, r7, lsr #7
\r
2091 and r6 ,lr, r7, lsr #15
\r
2093 and r8, lr, r7, lsr #23
\r
2096 orr r4, r4, r4, lsl #16
\r
2097 orr r5, r5, r5, lsl #16
\r
2098 orr r6, r6, r6, lsl #16
\r
2099 orr r8, r8, r8, lsl #16
\r
2100 stmia r0!, {r4,r5,r6,r8}
\r
2103 bne .fl_20scale_nn
\r
2105 ldmfd sp!, {r4-r11,pc}
\r
2110 and r8, lr, r8, lsl #1
\r
2113 mov r8, r8, lsl #16
\r
2119 and r4, lr, r12, lsl #1
\r
2121 and r5, lr, r12, lsr #7
\r
2123 and r6, lr, r12, lsr #15
\r
2127 add r10,r4, r8, lsr #16
\r
2128 mov r10,r10,lsr #1
\r
2129 orr r4, r10,r4, lsl #16 @ (px-1+px0)/2 | px0
\r
2131 and r8 ,lr, r12, lsr #23
\r
2135 add r10,r5, r4, lsr #16
\r
2136 mov r10,r10,lsr #1
\r
2137 orr r5, r10,r5, lsl #16 @ (px0 +px1)/2 | px1
\r
2140 add r10,r6, r5, lsr #16
\r
2141 mov r10,r10,lsr #1
\r
2142 orr r6, r10,r6, lsl #16 @ (px1 +px2)/2 | px2
\r
2145 add r10,r8, r6, lsr #16
\r
2146 mov r10,r10,lsr #1
\r
2147 orr r8, r10,r8, lsl #16 @ (px2 +px3)/2 | px3
\r
2149 stmia r0!, {r4,r5,r6,r8}
\r
2151 and r4, lr, r7, lsl #1
\r
2153 and r5, lr, r7, lsr #7
\r
2155 and r6, lr, r7, lsr #15
\r
2159 add r10,r4, r8, lsr #16
\r
2160 mov r10,r10,lsr #1
\r
2161 orr r4, r10,r4, lsl #16 @ (px-1+px0)/2 | px0
\r
2163 and r8 ,lr, r7, lsr #23
\r
2167 add r10,r5, r4, lsr #16
\r
2168 mov r10,r10,lsr #1
\r
2169 orr r5, r10,r5, lsl #16 @ (px0 +px1)/2 | px1
\r
2172 add r10,r6, r5, lsr #16
\r
2173 mov r10,r10,lsr #1
\r
2174 orr r6, r10,r6, lsl #16 @ (px1 +px2)/2 | px2
\r
2177 add r10,r8, r6, lsr #16
\r
2178 mov r10,r10,lsr #1
\r
2179 orr r8, r10,r8, lsl #16 @ (px2 +px3)/2 | px3
\r
2182 stmia r0!, {r4,r5,r6,r8}
\r
2183 bne .fl_20loop_bl2
\r
2185 ldmfd sp!, {r4-r11,pc}
\r
2188 #ifdef UNALIGNED_DRAWLINEDEST
\r
2189 @ unaligned versions of loops
\r
2190 @ warning: starts drawing 2bytes before dst
\r
2193 sub r0, r0, #2 @ initial adjustment
\r
2200 and r6, lr, r12,lsl #1
\r
2202 and r5, lr, r12,lsr #7
\r
2204 orr r4, r8, r6, lsl #16
\r
2206 and r6, lr, r12,lsr #15
\r
2208 and r8, lr, r12,lsr #23
\r
2210 orr r5, r5, r6, lsl #16
\r
2212 and r6, lr, r7, lsl #1
\r
2214 and r12,lr, r7, lsr #7
\r
2215 ldrh r12,[r3, r12]
\r
2216 orr r6, r8, r6, lsl #16
\r
2218 and r8, lr, r7, lsr #15
\r
2220 and r7, lr, r7, lsr #23
\r
2223 orr r12,r12,r8, lsl #16
\r
2226 stmia r0!, {r4,r5,r6,r12}
\r
2227 bne .fl_loopRGB555u
\r
2231 ldmfd sp!, {r4-r11,pc}
\r
2234 .fl_32scale_RGB555u:
\r
2235 sub r0, r0, #2 @ initial adjustment
\r
2238 @ r9 f800 07e0 001f | e000 0780 001c | 3800 01e0 0007
\r
2239 .fl_loop32scale_RGB555u:
\r
2243 and r6, lr, r12,lsl #1
\r
2245 and r5, lr, r12,lsr #7
\r
2248 orr r4, r4, r6, lsl #16 @ r4 = pix_d -1, 0
\r
2251 sub r8, r5, r5, lsr #2 @ r8 = 3/4 pix_s 1
\r
2252 add r6, r8, r6, lsr #2 @ r6 = (1/4 pix_s 0) + (3/4 pix_s 1)
\r
2253 orr r5, r6, r5, lsl #15
\r
2255 and r6, lr, r12,lsr #15
\r
2257 and r12,lr, r12,lsr #23
\r
2258 ldrh r12,[r3, r12]
\r
2260 add r5, r5, r6, lsl #15 @ r5 = pix_d 1, 2
\r
2262 and r8, lr, r7, lsl #1
\r
2264 and r10,lr, r7, lsr #7
\r
2265 ldrh r10,[r3, r10]
\r
2267 sub r6, r6, r6, lsr #2 @ r6 = 3/4 pix_s 2
\r
2268 add r6, r6, r12,lsr #2
\r
2269 orr r6, r6, r12,lsl #16 @ r6 = pix_d 3, 4
\r
2273 sub r12,r10,r10,lsr #2 @ r12 = 3/4 pix_s 5
\r
2274 orr r8, r8, r8, lsl #14
\r
2275 add r8, r8, r12,lsl #16 @ r8 = pix_d 5, 6
\r
2276 and r12,lr, r7, lsr #15
\r
2277 ldrh r12,[r3, r12]
\r
2278 and r7, lr, r7, lsr #23
\r
2282 mov r10,r10, lsr #1
\r
2283 sub r12,r12,r12,lsr #2 @ r12 = 3/4 pix_s 6
\r
2284 orr r10,r10,r12,lsl #16
\r
2286 add r10,r10,r7, lsl #14 @ r10 = pix_d 7, 8
\r
2290 stmia r0!, {r4,r5,r6,r8,r10}
\r
2292 bne .fl_loop32scale_RGB555u
\r
2296 ldmfd sp!, {r4-r11,pc}
\r
2298 #endif /* UNALIGNED_DRAWLINEDEST */
\r
2301 @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
\r
2304 .global blockcpy @ void *dst, void *src, size_t n
\r
2307 stmfd sp!, {r4,r5}
\r
2314 ldmia r1!, {r3-r5,r12}
\r
2316 stmia r0!, {r3-r5,r12}
\r
2320 adds r2, r2, #16-4
\r
2321 ldmltfd sp!, {r4,r5}
\r
2328 bge blockcpy_loop2
\r
2330 ldmfd sp!, {r4,r5}
\r
2340 ldmdb r1!, {r3-r5,r12}
\r
2342 stmdb r0!, {r3-r5,r12}
\r
2343 bge blockcpyhi_loop
\r
2346 adds r2, r2, #16-4
\r
2347 ldmltfd sp!, {r4,r5}
\r
2351 ldr r3, [r1, #-4]!
\r
2353 str r3, [r0, #-4]!
\r
2354 bge blockcpyhi_loop2
\r
2356 ldmfd sp!, {r4,r5}
\r
2360 .global blockcpy_or @ void *dst, void *src, size_t n, int pat
\r
2363 stmfd sp!, {r4-r6}
\r
2364 orr r3, r3, r3, lsl #8
\r
2365 orr r3, r3, r3, lsl #16
\r
2372 ldmia r1!, {r4-r6,r12}
\r
2378 stmia r0!, {r4-r6,r12}
\r
2379 bge blockcpy_loop_or
\r
2382 adds r2, r2, #16-4
\r
2383 ldmltfd sp!, {r4-r6}
\r
2386 blockcpy_loop_or2:
\r
2391 bge blockcpy_loop_or2
\r
2393 ldmfd sp!, {r4-r6}
\r
2401 blt blockcpyhi_or2
\r
2402 blockcpyhi_loop_or:
\r
2403 ldmdb r1!, {r4-r6,r12}
\r
2409 stmdb r0!, {r4-r6,r12}
\r
2410 bge blockcpyhi_loop_or
\r
2413 adds r2, r2, #16-4
\r
2414 ldmltfd sp!, {r4-r6}
\r
2417 blockcpyhi_loop_or2:
\r
2418 ldr r4, [r1, #-4]!
\r
2421 str r4, [r0, #-4]!
\r
2422 bge blockcpyhi_loop_or2
\r
2424 ldmfd sp!, {r4-r6}
\r
2427 @ vim:filetype=armasm
\r