summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorThomas Martitz <kugel@rockbox.org>2010-04-05 12:02:56 +0000
committerThomas Martitz <kugel@rockbox.org>2010-04-05 12:02:56 +0000
commit20003ec1c080f683f5c3b76537a419a231f6cf14 (patch)
tree735203551d592ca8f00b7f315e75318e18fff242
parent446a67698fc1d205637f8a489cee472c611d4b5f (diff)
downloadrockbox-20003ec1c080f683f5c3b76537a419a231f6cf14.tar.gz
rockbox-20003ec1c080f683f5c3b76537a419a231f6cf14.zip
Add forgotten file (git was supposed to rename!).
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25482 a1c6a512-1295-4272-9138-f99709370657
Diffstat
-rw-r--r--firmware/target/arm/as3525/lcd-as-e200v2-fuze-fuzev2.S551
1 files changed, 551 insertions, 0 deletions
diff --git a/firmware/target/arm/as3525/lcd-as-e200v2-fuze-fuzev2.S b/firmware/target/arm/as3525/lcd-as-e200v2-fuze-fuzev2.S
new file mode 100644
index 0000000000..2a16d3a6cd
--- /dev/null
+++ b/firmware/target/arm/as3525/lcd-as-e200v2-fuze-fuzev2.S
@@ -0,0 +1,551 @@
1/***************************************************************************
2 * __________ __ ___.
3 * Open \______ \ ____ ____ | | _\_ |__ _______ ___
4 * Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
5 * Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
6 * Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
7 * \/ \/ \/ \/ \/
8 * $Id$
9 *
10 * Copyright (C) 2007 by Jens Arnold
11 * Heavily based on lcd-as-memframe.c by Michael Sevakis
12 * Adapted for Sansa Fuze/e200v2 by Rafaël Carré
13 *
14 * This program is free software; you can redistribute it and/or
15 * modify it under the terms of the GNU General Public License
16 * as published by the Free Software Foundation; either version 2
17 * of the License, or (at your option) any later version.
18 *
19 * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
20 * KIND, either express or implied.
21 *
22 ****************************************************************************/
23
24#include "config.h"
25#include "cpu.h"
26
27#define DBOP_BUSY (1<<10)
28
29/****************************************************************************
30 * void lcd_write_yuv_420_lines(unsigned char const * const src[3],
31 * int width,
32 * int stride);
33 *
34 * |R| |1.000000 -0.000001 1.402000| |Y'|
35 * |G| = |1.000000 -0.334136 -0.714136| |Pb|
36 * |B| |1.000000 1.772000 0.000000| |Pr|
37 * Scaled, normalized, rounded and tweaked to yield RGB 565:
38 * |R| |74 0 101| |Y' - 16| >> 9
39 * |G| = |74 -24 -51| |Cb - 128| >> 8
40 * |B| |74 128 0| |Cr - 128| >> 9
41 *
42 * Write four RGB565 pixels in the following order on each loop:
43 * 1 3 + > down
44 * 2 4 \/ left
45 */
46 .section .icode, "ax", %progbits
47 .align 2
48 .global lcd_write_yuv420_lines
49 .type lcd_write_yuv420_lines, %function
50lcd_write_yuv420_lines:
51 @ r0 = yuv_src
52 @ r1 = width
53 @ r2 = stride
54 stmfd sp!, { r4-r11, lr } @ save non-scratch
55
56 mov r3, #0xC8000000 @
57 orr r3, r3, #0x120000 @ r3 = DBOP_BASE
58
59 ldmia r0, { r4, r5, r6 } @ r4 = yuv_src[0] = Y'_p
60 @ r5 = yuv_src[1] = Cb_p
61 @ r6 = yuv_src[2] = Cr_p
62 @ r0 = scratch
63 ldr r12, [r3, #8] @
64 sub r2, r2, #1 @ stride -= 1
65 orr r12, r12, #3<<13 @ DBOP_CTRL |= (1<<13|1<<14) (32bit mode)
66#ifdef SANSA_FUZEV2
67 bic r12, r12, #1<<13 @ DBOP_CTRL &= ~(1<<13),still 32bit mode
68#endif
69 str r12, [r3, #8] @
7010: @ loop line @
71 ldrb r7, [r4], #1 @ r7 = *Y'_p++;
72 ldrb r8, [r5], #1 @ r8 = *Cb_p++;
73 ldrb r9, [r6], #1 @ r9 = *Cr_p++;
74 @
75 sub r7, r7, #16 @ r7 = Y = (Y' - 16)*74
76 add r12, r7, r7, asl #2 @ actually (Y' - 16)*37 and shift right
77 add r7, r12, r7, asl #5 @ by one less when adding - same for all
78 @
79 sub r8, r8, #128 @ Cb -= 128
80 sub r9, r9, #128 @ Cr -= 128
81 @
82 add r10, r9, r9, asl #1 @ r10 = Cr*51 + Cb*24
83 add r10, r10, r10, asl #4 @
84 add r10, r10, r8, asl #3 @
85 add r10, r10, r8, asl #4 @
86 @
87 add lr, r9, r9, asl #2 @ r9 = Cr*101
88 add lr, lr, r9, asl #5 @
89 add r9, lr, r9, asl #6 @
90 @
91 add r8, r8, #2 @ r8 = bu = (Cb*128 + 128) >> 8
92 mov r8, r8, asr #2 @
93 add r9, r9, #256 @ r9 = rv = (r9 + 256) >> 9
94 mov r9, r9, asr #9 @
95 rsb r10, r10, #128 @ r10 = guv = (-r10 + 128) >> 8
96 mov r10, r10, asr #8 @
97 @ compute R, G, and B
98 add r0, r8, r7, asr #8 @ r0 = b = (Y >> 9) + bu
99 add lr, r9, r7, asr #8 @ lr = r = (Y >> 9) + rv
100 add r7, r10, r7, asr #7 @ r7 = g = (Y >> 8) + guv
101 @
102 orr r12, r0, lr @ check if clamping is needed...
103 orr r12, r12, r7, asr #1 @ ...at all
104 cmp r12, #31 @
105 bls 15f @ no clamp @
106 cmp r0, #31 @ clamp b
107 mvnhi r0, r0, asr #31 @
108 andhi r0, r0, #31 @
109 cmp lr, #31 @ clamp r
110 mvnhi lr, lr, asr #31 @
111 andhi lr, lr, #31 @
112 cmp r7, #63 @ clamp g
113 mvnhi r7, r7, asr #31 @
114 andhi r7, r7, #63 @
11515: @ no clamp @
116 @
117 ldrb r12, [r4, r2] @ r12 = Y' = *(Y'_p + stride)
118 @
119 orr r0, r0, lr, lsl #11 @ r0 = (r << 11) | b
120 orr r11, r0, r7, lsl #5 @ r11 = (r << 11) | (g << 5) | b
121 orr r11, r0, r7, lsl #5 @ r11 = (r << 11) | (g << 5) | b
122#ifdef SANSA_FUZEV2
123 mov r0, r11, lsr #8 @
124 bic r11, r11, #0xff00 @
125 orr r11, r0, r11, lsl #8 @ swap bytes
126#endif
127 sub r7, r12, #16 @ r7 = Y = (Y' - 16)*74
128 add r12, r7, r7, asl #2 @
129 add r7, r12, r7, asl #5 @
130 @ compute R, G, and B
131 add r0, r8, r7, asr #8 @ r0 = b = (Y >> 9) + bu
132 add lr, r9, r7, asr #8 @ lr = r = (Y >> 9) + rv
133 add r7, r10, r7, asr #7 @ r7 = g = (Y >> 8) + guv
134 @
135 orr r12, r0, lr @ check if clamping is needed...
136 orr r12, r12, r7, asr #1 @ ...at all
137 cmp r12, #31 @
138 bls 15f @ no clamp @
139 cmp r0, #31 @ clamp b
140 mvnhi r0, r0, asr #31 @
141 andhi r0, r0, #31 @
142 cmp lr, #31 @ clamp r
143 mvnhi lr, lr, asr #31 @
144 andhi lr, lr, #31 @
145 cmp r7, #63 @ clamp g
146 mvnhi r7, r7, asr #31 @
147 andhi r7, r7, #63 @
14815: @ no clamp @
149 @
150 ldrb r12, [r4], #1 @ r12 = Y' = *(Y'_p++)
151 @
152 orr r0, r0, lr, lsl #11 @ r0 = (r << 11) | b
153 orr r0, r0, r7, lsl #5 @ r0 = (r << 11) | (g << 5) | b
154
155#ifdef SANSA_FUZEV2
156 mov r7, r0, lsr #8 @
157 bic r7, r7, #0xff00 @
158 orr r0, r7, r0, lsl #8 @ swap bytes
159#endif
160
161 orr r0, r11, r0, lsl#16 @ pack with 2nd pixel
162 str r0, [r3, #0x10] @ write pixel
163 @
164 sub r7, r12, #16 @ r7 = Y = (Y' - 16)*74
165 add r12, r7, r7, asl #2 @
166 add r7, r12, r7, asl #5 @
167 @ compute R, G, and B
168 add r0, r8, r7, asr #8 @ r0 = b = (Y >> 9) + bu
169 add lr, r9, r7, asr #8 @ lr = r = (Y >> 9) + rv
170 add r7, r10, r7, asr #7 @ r7 = g = (Y >> 8) + guv
171 @
172 orr r12, r0, lr @ check if clamping is needed...
173 orr r12, r12, r7, asr #1 @ ...at all
174 cmp r12, #31 @
175 bls 15f @ no clamp @
176 cmp r0, #31 @ clamp b
177 mvnhi r0, r0, asr #31 @
178 andhi r0, r0, #31 @
179 cmp lr, #31 @ clamp r
180 mvnhi lr, lr, asr #31 @
181 andhi lr, lr, #31 @
182 cmp r7, #63 @ clamp g
183 mvnhi r7, r7, asr #31 @
184 andhi r7, r7, #63 @
18515: @ no clamp @
186 @
187 ldrb r12, [r4, r2] @ r12 = Y' = *(Y'_p + stride)
188 @
189 @
190 orr r0, r0, lr, lsl #11 @ r0 = (r << 11) | b
191 orr r11, r0, r7, lsl #5 @ r0 = (r << 11) | (g << 5) | b
192
193#ifdef SANSA_FUZEV2
194 mov r0, r11, lsr #8 @
195 bic r11, r11, #0xff00 @
196 orr r11, r0, r11, lsl #8 @ swap byte
197#endif
198
199 sub r7, r12, #16 @ r7 = Y = (Y' - 16)*74
200 add r12, r7, r7, asl #2 @
201 add r7, r12, r7, asl #5 @
202 @ compute R, G, and B
203 add r0, r8, r7, asr #8 @ r0 = b = (Y >> 9) + bu
204 add lr, r9, r7, asr #8 @ lr = r = (Y >> 9) + rv
205 add r7, r10, r7, asr #7 @ r7 = g = (Y >> 8) + guv
206 @
207 orr r12, r0, lr @ check if clamping is needed...
208 orr r12, r12, r7, asr #1 @ ...at all
209 cmp r12, #31 @
210 bls 15f @ no clamp @
211 cmp r0, #31 @ clamp b
212 mvnhi r0, r0, asr #31 @
213 andhi r0, r0, #31 @
214 cmp lr, #31 @ clamp r
215 mvnhi lr, lr, asr #31 @
216 andhi lr, lr, #31 @
217 cmp r7, #63 @ clamp g
218 mvnhi r7, r7, asr #31 @
219 andhi r7, r7, #63 @
22015: @ no clamp @
221 @
222 orr r0, r0, lr, lsl #11 @ r0 = (r << 11) | b
223 orr r0, r0, r7, lsl #5 @ r0 = (r << 11) | (g << 5) | b
224
225#ifdef SANSA_FUZEV2
226 mov r7, r0, lsr #8 @
227 bic r7, r7, #0xff00 @
228 orr r0, r7, r0, lsl #8 @ swap bytes
229#endif
230
231 orr r0, r11, r0, lsl#16 @ pack with 2nd pixel
232 str r0, [r3, #0x10] @ write pixel
233 @
234 subs r1, r1, #2 @ subtract block from width
235 bgt 10b @ loop line @
236 @
2371: @ busy
238 @ writing at max 110*32 (LCD_WIDTH/2), the fifo is bigger
239 @ so polling fifo empty only after each line is save
240 ldr r7, [r3,#0xc] @ r7 = DBOP_STATUS
241 tst r7, #DBOP_BUSY @ fifo not empty?
242 beq 1b @
243
244 ldmfd sp!, { r4-r11, pc } @ restore registers and return
245 bx lr @
246 .ltorg @ dump constant pool
247 .size lcd_write_yuv420_lines, .-lcd_write_yuv420_lines
248
249/****************************************************************************
250 * void lcd_write_yuv_420_lines_odither(unsigned char const * const src[3],
251 * int width,
252 * int stride,
253 * int x_screen,
254 * int y_screen);
255 *
256 * |R| |1.000000 -0.000001 1.402000| |Y'|
257 * |G| = |1.000000 -0.334136 -0.714136| |Pb|
258 * |B| |1.000000 1.772000 0.000000| |Pr|
259 * Red scaled at twice g & b but at same precision to place it in correct
260 * bit position after multiply and leave instruction count lower.
261 * |R| |258 0 408| |Y' - 16|
262 * |G| = |149 -49 -104| |Cb - 128|
263 * |B| |149 258 0| |Cr - 128|
264 *
265 * Write four RGB565 pixels in the following order on each loop:
266 * 1 3 + > down
267 * 2 4 \/ left
268 *
269 * Kernel pattern (raw|rotated|use order):
270 * 5 3 4 2 2 6 3 7 row0 row2 > down
271 * 1 7 0 6 | 4 0 5 1 | 2 4 6 0 3 5 7 1 col0 left
272 * 4 2 5 3 | 3 7 2 6 | 3 5 7 1 2 4 6 0 col2 \/
273 * 0 6 1 7 5 1 4 0
274 */
275 .section .icode, "ax", %progbits
276 .align 2
277 .global lcd_write_yuv420_lines_odither
278 .type lcd_write_yuv420_lines_odither, %function
279lcd_write_yuv420_lines_odither:
280 @ r0 = yuv_src
281 @ r1 = width
282 @ r2 = stride
283 @ r3 = x_screen
284 @ [sp] = y_screen
285 stmfd sp!, { r4-r11, lr } @ save non-scratch
286 ldmia r0, { r4, r5, r6 } @ r4 = yuv_src[0] = Y'_p
287 @ r5 = yuv_src[1] = Cb_p
288 @ r6 = yuv_src[2] = Cr_p
289 @
290 ldr r14, [sp, #40] @ Line up pattern and kernel quadrant
291 sub r2, r2, #1 @ stride =- 1
292 eor r14, r14, r3 @
293 and r14, r14, #0x2 @
294 mov r14, r14, lsl #6 @ 0x00 or 0x80
295
296 mov r3, #0xC8000000 @
297 orr r3, r3, #0x120000 @ r3 = DBOP_BASE, need to be redone
298 @ due to lack of registers
299 ldr r12, [r3, #8] @
300 orr r12, r12, #3<<13 @ DBOP_CTRL |= (1<<13|1<<14)
301#ifdef SANSA_FUZEV2
302 bic r12, r12, #1<<13 @ DBOP_CTRL &= ~(1<<13), still 32bit mode
303#endif
304 str r12, [r3, #8] @ (32bit mode)
30510: @ loop line @
306 @
307 ldrb r7, [r4], #1 @ r7 = *Y'_p++;
308 ldrb r8, [r5], #1 @ r8 = *Cb_p++;
309 ldrb r9, [r6], #1 @ r9 = *Cr_p++;
310 @
311 eor r14, r14, #0x80 @ flip pattern quadrant
312 @
313 sub r7, r7, #16 @ r7 = Y = (Y' - 16)*149
314 add r12, r7, r7, asl #2 @
315 add r12, r12, r12, asl #4 @
316 add r7, r12, r7, asl #6 @
317 @
318 sub r8, r8, #128 @ Cb -= 128
319 sub r9, r9, #128 @ Cr -= 128
320 @
321 add r10, r8, r8, asl #4 @ r10 = guv = Cr*104 + Cb*49
322 add r10, r10, r8, asl #5 @
323 add r10, r10, r9, asl #3 @
324 add r10, r10, r9, asl #5 @
325 add r10, r10, r9, asl #6 @
326 @
327 mov r8, r8, asl #1 @ r8 = bu = Cb*258
328 add r8, r8, r8, asl #7 @
329 @
330 add r9, r9, r9, asl #1 @ r9 = rv = Cr*408
331 add r9, r9, r9, asl #4 @
332 mov r9, r9, asl #3 @
333 @
334 @ compute R, G, and B
335 add r0, r8, r7 @ r0 = b' = Y + bu
336 add r11, r9, r7, asl #1 @ r11 = r' = Y*2 + rv
337 rsb r7, r10, r7 @ r7 = g' = Y + guv
338 @
339 @ r8 = bu, r9 = rv, r10 = guv
340 @
341 sub r12, r0, r0, lsr #5 @ r0 = 31/32*b + b/256
342 add r0, r12, r0, lsr #8 @
343 @
344 sub r12, r11, r11, lsr #5 @ r11 = 31/32*r + r/256
345 add r11, r12, r11, lsr #8 @
346 @
347 sub r12, r7, r7, lsr #6 @ r7 = 63/64*g + g/256
348 add r7, r12, r7, lsr #8 @
349 @
350 add r12, r14, #0x100 @
351 @
352 add r0, r0, r12 @ b = r0 + delta
353 add r11, r11, r12, lsl #1 @ r = r11 + delta*2
354 add r7, r7, r12, lsr #1 @ g = r7 + delta/2
355 @
356 orr r12, r0, r11, asr #1 @ check if clamping is needed...
357 orr r12, r12, r7 @ ...at all
358 movs r12, r12, asr #15 @
359 beq 15f @ no clamp @
360 movs r12, r0, asr #15 @ clamp b
361 mvnne r0, r12, lsr #15 @
362 andne r0, r0, #0x7c00 @ mask b only if clamped
363 movs r12, r11, asr #16 @ clamp r
364 mvnne r11, r12, lsr #16 @
365 movs r12, r7, asr #15 @ clamp g
366 mvnne r7, r12, lsr #15 @
36715: @ no clamp @
368 @
369 ldrb r12, [r4, r2] @ r12 = Y' = *(Y'_p + stride)
370 @
371 and r11, r11, #0xf800 @ pack pixel
372 and r7, r7, #0x7e00 @ r0 = pixel = (r & 0xf800) |
373 orr r11, r11, r7, lsr #4 @ ((g & 0x7e00) >> 4) |
374 orr r3, r11, r0, lsr #10 @ (b >> 10)
375#ifdef SANSA_FUZEV2
376 mov r7, r3, lsr #8 @
377 bic r3, r3, #0xff00 @
378 orr r3, r7, r3, lsl #8 @ swap pixel
379#endif
380 @ save pixel
381 sub r7, r12, #16 @ r7 = Y = (Y' - 16)*149
382 add r12, r7, r7, asl #2 @
383 add r12, r12, r12, asl #4 @
384 add r7, r12, r7, asl #6 @
385 @ compute R, G, and B
386 add r0, r8, r7 @ r0 = b' = Y + bu
387 add r11, r9, r7, asl #1 @ r11 = r' = Y*2 + rv
388 rsb r7, r10, r7 @ r7 = g' = Y + guv
389 @
390 sub r12, r0, r0, lsr #5 @ r0 = 31/32*b' + b'/256
391 add r0, r12, r0, lsr #8 @
392 @
393 sub r12, r11, r11, lsr #5 @ r11 = 31/32*r' + r'/256
394 add r11, r12, r11, lsr #8 @
395 @
396 sub r12, r7, r7, lsr #6 @ r7 = 63/64*g' + g'/256
397 add r7, r12, r7, lsr #8 @
398 @
399 add r12, r14, #0x200 @
400 @
401 add r0, r0, r12 @ b = r0 + delta
402 add r11, r11, r12, lsl #1 @ r = r11 + delta*2
403 add r7, r7, r12, lsr #1 @ g = r7 + delta/2
404 @
405 orr r12, r0, r11, asr #1 @ check if clamping is needed...
406 orr r12, r12, r7 @ ...at all
407 movs r12, r12, asr #15 @
408 beq 15f @ no clamp @
409 movs r12, r0, asr #15 @ clamp b
410 mvnne r0, r12, lsr #15 @
411 andne r0, r0, #0x7c00 @ mask b only if clamped
412 movs r12, r11, asr #16 @ clamp r
413 mvnne r11, r12, lsr #16 @
414 movs r12, r7, asr #15 @ clamp g
415 mvnne r7, r12, lsr #15 @
41615: @ no clamp @
417 @
418 ldrb r12, [r4], #1 @ r12 = Y' = *(Y'_p++)
419
420 and r11, r11, #0xf800 @ pack pixel
421 and r7, r7, #0x7e00 @ r0 = pixel = (r & 0xf800) |
422 orr r11, r11, r7, lsr #4 @ ((g & 0x7e00) >> 4) |
423 orr r0, r11, r0, lsr #10 @ (b >> 10)
424#ifdef SANSA_FUZEV2
425 mov r7, r0, lsr #8 @
426 bic r0, r0, #0xff00 @
427 orr r0, r7, r0, lsl #8 @ swap pixel
428#endif
429 orr r3, r3, r0, lsl#16 @ pack with 2nd pixel
430 mov r0, #0xC8000000 @
431 orr r0, r0, #0x120000 @ r3 = DBOP_BASE
432
433 str r3, [r0, #0x10] @ write pixel
434 @
435 sub r7, r12, #16 @ r7 = Y = (Y' - 16)*149
436 add r12, r7, r7, asl #2 @
437 add r12, r12, r12, asl #4 @
438 add r7, r12, r7, asl #6 @
439 @ compute R, G, and B
440 add r0, r8, r7 @ r0 = b' = Y + bu
441 add r11, r9, r7, asl #1 @ r11 = r' = Y*2 + rv
442 rsb r7, r10, r7 @ r7 = g' = Y + guv
443 @
444 @ r8 = bu, r9 = rv, r10 = guv
445 @
446 sub r12, r0, r0, lsr #5 @ r0 = 31/32*b' + b'/256
447 add r0, r12, r0, lsr #8 @
448 @
449 sub r12, r11, r11, lsr #5 @ r11 = 31/32*r' + r'/256
450 add r11, r12, r11, lsr #8 @
451 @
452 sub r12, r7, r7, lsr #6 @ r7 = 63/64*g' + g'/256
453 add r7, r12, r7, lsr #8 @
454 @
455 add r12, r14, #0x300 @
456 @
457 add r0, r0, r12 @ b = r0 + delta
458 add r11, r11, r12, lsl #1 @ r = r11 + delta*2
459 add r7, r7, r12, lsr #1 @ g = r7 + delta/2
460 @
461 orr r12, r0, r11, asr #1 @ check if clamping is needed...
462 orr r12, r12, r7 @ ...at all
463 movs r12, r12, asr #15 @
464 beq 15f @ no clamp @
465 movs r12, r0, asr #15 @ clamp b
466 mvnne r0, r12, lsr #15 @
467 andne r0, r0, #0x7c00 @ mask b only if clamped
468 movs r12, r11, asr #16 @ clamp r
469 mvnne r11, r12, lsr #16 @
470 movs r12, r7, asr #15 @ clamp g
471 mvnne r7, r12, lsr #15 @
47215: @ no clamp @
473 @
474 ldrb r12, [r4, r2] @ r12 = Y' = *(Y'_p + stride)
475 @
476 and r11, r11, #0xf800 @ pack pixel
477 and r7, r7, #0x7e00 @ r0 = pixel = (r & 0xf800) |
478 orr r11, r11, r7, lsr #4 @ ((g & 0x7e00) >> 4) |
479 orr r3, r11, r0, lsr #10 @ (b >> 10)
480#ifdef SANSA_FUZEV2
481 mov r7, r3, lsr #8 @
482 bic r3, r3, #0xff00 @
483 orr r3, r7, r3, lsl #8 @ swap pixel
484#endif
485 @ save pixel
486 @
487 sub r7, r12, #16 @ r7 = Y = (Y' - 16)*149
488 add r12, r7, r7, asl #2 @
489 add r12, r12, r12, asl #4 @
490 add r7, r12, r7, asl #6 @
491 @ compute R, G, and B
492 add r0, r8, r7 @ r0 = b' = Y + bu
493 add r11, r9, r7, asl #1 @ r11 = r' = Y*2 + rv
494 rsb r7, r10, r7 @ r7 = g' = Y + guv
495 @
496 sub r12, r0, r0, lsr #5 @ r0 = 31/32*b + b/256
497 add r0, r12, r0, lsr #8 @
498 @
499 sub r12, r11, r11, lsr #5 @ r11 = 31/32*r + r/256
500 add r11, r12, r11, lsr #8 @
501 @
502 sub r12, r7, r7, lsr #6 @ r7 = 63/64*g + g/256
503 add r7, r12, r7, lsr #8 @
504 @
505 @ This element is zero - use r14 @
506 @
507 add r0, r0, r14 @ b = r0 + delta
508 add r11, r11, r14, lsl #1 @ r = r11 + delta*2
509 add r7, r7, r14, lsr #1 @ g = r7 + delta/2
510 @
511 orr r12, r0, r11, asr #1 @ check if clamping is needed...
512 orr r12, r12, r7 @ ...at all
513 movs r12, r12, asr #15 @
514 beq 15f @ no clamp @
515 movs r12, r0, asr #15 @ clamp b
516 mvnne r0, r12, lsr #15 @
517 andne r0, r0, #0x7c00 @ mask b only if clamped
518 movs r12, r11, asr #16 @ clamp r
519 mvnne r11, r12, lsr #16 @
520 movs r12, r7, asr #15 @ clamp g
521 mvnne r7, r12, lsr #15 @
52215: @ no clamp @
523 @
524 and r11, r11, #0xf800 @ pack pixel
525 and r7, r7, #0x7e00 @ r0 = pixel = (r & 0xf800) |
526 orr r11, r11, r7, lsr #4 @ ((g & 0x7e00) >> 4) |
527 orr r0, r11, r0, lsr #10 @ (b >> 10)
528#ifdef SANSA_FUZEV2
529 mov r7, r0, lsr #8 @
530 bic r0, r0, #0xff00 @
531 orr r0, r7, r0, lsl #8 @ swap pixel
532#endif
533 orr r3, r3, r0, lsl#16 @ pack with 2nd pixel
534 mov r0, #0xC8000000 @
535 orr r0, r0, #0x120000 @ r3 = DBOP_BASE
536
537 str r3, [r0, #0x10] @ write pixel
538 @
539 subs r1, r1, #2 @ subtract block from width
540 bgt 10b @ loop line @
541 @
5421: @ busy @
543 @ writing at max 110*32 (LCD_WIDTH/2), the fifo is bigger (128*32)
544 @ so polling fifo empty only after each line is save
545 ldr r7, [r0,#0xc] @ r7 = DBOP_STATUS
546 tst r7, #DBOP_BUSY @ fifo not empty?
547 beq 1b @
548
549 ldmfd sp!, { r4-r11, pc } @ restore registers and return
550 .ltorg @ dump constant pool
551 .size lcd_write_yuv420_lines_odither, .-lcd_write_yuv420_lines_odither
generated by cgit v1.2.3 (git 2.39.1) at 2024-08-23 03:28:41 +0000