diff options
author | Jens Arnold <amiconn@rockbox.org> | 2008-03-25 23:21:36 +0000 |
---|---|---|
committer | Jens Arnold <amiconn@rockbox.org> | 2008-03-25 23:21:36 +0000 |
commit | 40919d7db29d036033f57fe6fbf9e3f562e61c0d (patch) | |
tree | b04fa272f6cee51639b7307ee5fcd9b4f9737fde /firmware | |
parent | 6ceaf321f3aadc804b1daa8fb10607a7ae209351 (diff) | |
download | rockbox-40919d7db29d036033f57fe6fbf9e3f562e61c0d.tar.gz rockbox-40919d7db29d036033f57fe6fbf9e3f562e61c0d.zip |
iAudio M3: Optimised LCD driver, with more/better assembly code. Speedup is ~80% when boosted, ~15% when unboosted. Also implemented grey phase blitting. * Adapted the greyscale library, and the plugins using it. * Fixed a bug in greyscale scroll down for vertically packed pixels.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16809 a1c6a512-1295-4272-9138-f99709370657
Diffstat (limited to 'firmware')
-rw-r--r-- | firmware/SOURCES | 1 | ||||
-rw-r--r-- | firmware/target/coldfire/iaudio/m3/lcd-as-m3.S | 516 | ||||
-rw-r--r-- | firmware/target/coldfire/iaudio/m3/lcd-m3.c | 305 |
3 files changed, 548 insertions, 274 deletions
diff --git a/firmware/SOURCES b/firmware/SOURCES index 0a8ac2a8e3..1d60ea1016 100644 --- a/firmware/SOURCES +++ b/firmware/SOURCES | |||
@@ -513,6 +513,7 @@ target/coldfire/iaudio/m3/backlight-m3.c | |||
513 | target/coldfire/iaudio/m3/button-m3.c | 513 | target/coldfire/iaudio/m3/button-m3.c |
514 | target/coldfire/iaudio/m3/fmradio_i2c-m3.c | 514 | target/coldfire/iaudio/m3/fmradio_i2c-m3.c |
515 | target/coldfire/iaudio/m3/lcd-m3.c | 515 | target/coldfire/iaudio/m3/lcd-m3.c |
516 | target/coldfire/iaudio/m3/lcd-as-m3.S | ||
516 | target/coldfire/iaudio/m3/power-m3.c | 517 | target/coldfire/iaudio/m3/power-m3.c |
517 | target/coldfire/iaudio/m3/powermgmt-m3.c | 518 | target/coldfire/iaudio/m3/powermgmt-m3.c |
518 | target/coldfire/iaudio/m3/system-m3.c | 519 | target/coldfire/iaudio/m3/system-m3.c |
diff --git a/firmware/target/coldfire/iaudio/m3/lcd-as-m3.S b/firmware/target/coldfire/iaudio/m3/lcd-as-m3.S new file mode 100644 index 0000000000..5f77e01e86 --- /dev/null +++ b/firmware/target/coldfire/iaudio/m3/lcd-as-m3.S | |||
@@ -0,0 +1,516 @@ | |||
1 | /*************************************************************************** | ||
2 | * __________ __ ___. | ||
3 | * Open \______ \ ____ ____ | | _\_ |__ _______ ___ | ||
4 | * Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ / | ||
5 | * Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < < | ||
6 | * Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \ | ||
7 | * \/ \/ \/ \/ \/ | ||
8 | * $Id$ | ||
9 | * | ||
10 | * Copyright (C) 2008 by Jens Arnold | ||
11 | * | ||
12 | * All files in this archive are subject to the GNU General Public License. | ||
13 | * See the file COPYING in the source tree root for full license agreement. | ||
14 | * | ||
15 | * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY | ||
16 | * KIND, either express or implied. | ||
17 | * | ||
18 | ****************************************************************************/ | ||
19 | |||
20 | #define CLOCK_MASK 0x20000000 | ||
21 | #define DATA_MASK 0x04000000 | ||
22 | #define GPIO_OUT_ADDR 0x80000004 | ||
23 | |||
24 | #define CS_MASK 0x00010000 | ||
25 | #define RS_MASK 0x00001000 | ||
26 | #define GPIO1_OUT_ADDR 0x800000b4 | ||
27 | |||
28 | .extern cpu_frequency /* Global variable from system.c */ | ||
29 | |||
30 | .section .icode,"ax",@progbits | ||
31 | |||
32 | /* Output 8 bits to the LCD. Instruction order is devised to maximize the | ||
33 | * delay between changing the data line and the CLK L->H transition, which | ||
34 | * makes the LCD controller sample DATA. | ||
35 | * | ||
36 | * Custom calling convention: | ||
37 | * %a0 - GPIO_OUT_ADDR | ||
38 | * %d3 - data byte | ||
39 | * %d6 - DATA_MASK | ||
40 | * %d7 - CLOCK_MASK | ||
41 | * Clobbers: | ||
42 | * %d0..%d3 | ||
43 | */ | ||
44 | .write_byte: | ||
45 | move.w %sr, %d2 | ||
46 | move.w #0x2700, %sr | ||
47 | |||
48 | move.l (%a0), %d0 /* Get current state of data port */ | ||
49 | move.l %d0, %d1 | ||
50 | and.l %d6, %d1 /* Check current state of data line */ | ||
51 | beq.s 1f /* and set it as previous-state bit */ | ||
52 | bset #8, %d3 | ||
53 | 1: | ||
54 | move.l %d3, %d1 /* Compute the 'bit derivative', i.e. a value */ | ||
55 | lsr.l #1, %d1 /* with 1's where the data changes from the */ | ||
56 | eor.l %d1, %d3 /* previous state, and 0's where it doesn't */ | ||
57 | swap %d3 /* Shift data to upper byte */ | ||
58 | lsl.l #8, %d3 | ||
59 | |||
60 | eor.l %d7, %d0 /* precalculate opposite state of clock line */ | ||
61 | |||
62 | lsl.l #1,%d3 /* Shift out MSB */ | ||
63 | bcc.s 1f | ||
64 | eor.l %d6, %d0 /* 1: Flip data bit */ | ||
65 | 1: | ||
66 | move.l %d0, %d1 | ||
67 | move.l %d0, (%a0) /* Output new state and set CLK = 0*/ | ||
68 | eor.l %d7, %d1 | ||
69 | bra.w .wr_bit7 | ||
70 | |||
71 | |||
72 | /* Output 16 bits to the LCD. Instruction order is devised to maximize the | ||
73 | * delay between changing the data line and the CLK L->H transition, which | ||
74 | * makes the LCD controller sample DATA. | ||
75 | * | ||
76 | * Custom calling convention: | ||
77 | * %a0 - GPIO_OUT_ADDR | ||
78 | * %d3 - data word | ||
79 | * %d6 - DATA_MASK | ||
80 | * %d7 - CLOCK_MASK | ||
81 | * Clobbers: | ||
82 | * %d0..%d3 | ||
83 | */ | ||
84 | .write_word: | ||
85 | move.w %sr, %d2 | ||
86 | move.w #0x2700, %sr | ||
87 | |||
88 | move.l (%a0), %d0 /* Get current state of data port */ | ||
89 | move.l %d0, %d1 | ||
90 | and.l %d6, %d1 /* Check current state of data line */ | ||
91 | beq.s 1f /* and set it as previous-state bit */ | ||
92 | bset #16, %d3 | ||
93 | 1: | ||
94 | move.l %d3, %d1 /* Compute the 'bit derivative', i.e. a value */ | ||
95 | lsr.l #1, %d1 /* with 1's where the data changes from the */ | ||
96 | eor.l %d1, %d3 /* previous state, and 0's where it doesn't */ | ||
97 | swap %d3 /* Shift data to upper word */ | ||
98 | |||
99 | eor.l %d7, %d0 /* precalculate opposite state of clock line */ | ||
100 | |||
101 | lsl.l #1,%d3 /* Shift out MSB */ | ||
102 | bcc.s 1f | ||
103 | eor.l %d6, %d0 /* 1: Flip data bit */ | ||
104 | 1: | ||
105 | move.l %d0, %d1 | ||
106 | move.l %d0, (%a0) /* Output new state and set CLK = 0*/ | ||
107 | eor.l %d7, %d1 | ||
108 | nop | ||
109 | |||
110 | .macro bit_out | ||
111 | lsl.l #1,%d3 | ||
112 | bcc.s 1f | ||
113 | eor.l %d6, %d0 | ||
114 | 1: | ||
115 | move.l %d1, (%a0) /* Set CLK = 1 (delayed) */ | ||
116 | move.l %d0, %d1 | ||
117 | move.l %d0, (%a0) | ||
118 | eor.l %d7, %d1 | ||
119 | .endm | ||
120 | bit_out | ||
121 | nop | ||
122 | bit_out | ||
123 | nop | ||
124 | bit_out | ||
125 | nop | ||
126 | bit_out | ||
127 | nop | ||
128 | bit_out | ||
129 | nop | ||
130 | bit_out | ||
131 | nop | ||
132 | bit_out | ||
133 | nop | ||
134 | bit_out | ||
135 | nop | ||
136 | |||
137 | .wr_bit7: | ||
138 | bit_out | ||
139 | nop | ||
140 | bit_out | ||
141 | nop | ||
142 | bit_out | ||
143 | nop | ||
144 | bit_out | ||
145 | nop | ||
146 | bit_out | ||
147 | nop | ||
148 | bit_out | ||
149 | nop | ||
150 | bit_out | ||
151 | nop | ||
152 | |||
153 | nop | ||
154 | move.l %d1, (%a0) /* Set CLK = 1 (delayed) */ | ||
155 | move.w %d2, %sr | ||
156 | rts | ||
157 | |||
158 | |||
159 | /* Output 16 bits to the LCD as fast as possible. Use only at < 60MHz. | ||
160 | * | ||
161 | * Custom calling convention: | ||
162 | * %a0 - GPIO_OUT_ADDR | ||
163 | * %d3 - data word | ||
164 | * %d6 - DATA_MASK | ||
165 | * %d7 - CLOCK_MASK | ||
166 | * Clobbers: | ||
167 | * %d0..%d3 | ||
168 | */ | ||
169 | .write_word_fast: | ||
170 | move.w %sr, %d2 /* Get current interrupt level */ | ||
171 | move.w #0x2700, %sr /* Disable interrupts */ | ||
172 | |||
173 | move.l (%a0), %d0 /* Get current state of data port */ | ||
174 | move.l %d0, %d1 | ||
175 | and.l %d6, %d1 /* Check current state of data line */ | ||
176 | beq.s 1f /* and set it as previous-state bit */ | ||
177 | bset #16, %d3 | ||
178 | 1: | ||
179 | move.l %d3, %d1 /* Compute the 'bit derivative', i.e. a value */ | ||
180 | lsr.l #1, %d1 /* with 1's where the data changes from the */ | ||
181 | eor.l %d1, %d3 /* previous state, and 0's where it doesn't */ | ||
182 | swap %d3 /* Shift data to upper byte */ | ||
183 | |||
184 | move.l %d0, %d1 /* precalculate opposite state of clock line */ | ||
185 | eor.l %d7, %d1 | ||
186 | |||
187 | .macro bit_out_fast | ||
188 | lsl.l #1,%d3 /* Shift out MSB */ | ||
189 | bcc.s 1f | ||
190 | eor.l %d6, %d0 /* 1: Flip data bit */ | ||
191 | eor.l %d6, %d1 /* for both clock states */ | ||
192 | 1: | ||
193 | move.l %d1, (%a0) /* Output new state and set CLK = 0*/ | ||
194 | move.l %d0, (%a0) /* set CLK = 1 */ | ||
195 | .endm | ||
196 | bit_out_fast | ||
197 | bit_out_fast | ||
198 | bit_out_fast | ||
199 | bit_out_fast | ||
200 | bit_out_fast | ||
201 | bit_out_fast | ||
202 | bit_out_fast | ||
203 | bit_out_fast | ||
204 | bit_out_fast | ||
205 | bit_out_fast | ||
206 | bit_out_fast | ||
207 | bit_out_fast | ||
208 | bit_out_fast | ||
209 | bit_out_fast | ||
210 | bit_out_fast | ||
211 | bit_out_fast | ||
212 | |||
213 | move.w %d2, %sr /* Restore interrupt level */ | ||
214 | rts | ||
215 | |||
216 | |||
217 | .global lcd_write_command | ||
218 | .type lcd_write_command, @function | ||
219 | |||
220 | lcd_write_command: | ||
221 | lea.l (-4*4, %sp), %sp | ||
222 | movem.l %d2-%d3/%d6-%d7, (%sp) | ||
223 | |||
224 | move.l (4*4+4, %sp), %d3 /* cmd */ | ||
225 | |||
226 | lea.l GPIO_OUT_ADDR, %a0 | ||
227 | lea.l GPIO1_OUT_ADDR, %a1 | ||
228 | move.l #DATA_MASK, %d6 | ||
229 | move.l #CLOCK_MASK, %d7 | ||
230 | |||
231 | move.l #~(RS_MASK+CS_MASK), %d0 | ||
232 | and.l %d0, (%a1) | ||
233 | |||
234 | bsr.w .write_byte | ||
235 | |||
236 | move.l #CS_MASK, %d0 | ||
237 | or.l %d0, (%a1) | ||
238 | |||
239 | movem.l (%sp), %d2-%d3/%d6-%d7 | ||
240 | lea.l (4*4, %sp), %sp | ||
241 | rts | ||
242 | |||
243 | |||
244 | .global lcd_write_command_e | ||
245 | .type lcd_write_command_e, @function | ||
246 | |||
247 | lcd_write_command_e: | ||
248 | lea.l (-4*4, %sp), %sp | ||
249 | movem.l %d2-%d3/%d6-%d7, (%sp) | ||
250 | |||
251 | movem.l (4*4+4, %sp), %d2-%d3 /* cmd, data */ | ||
252 | |||
253 | lea.l GPIO_OUT_ADDR, %a0 | ||
254 | lea.l GPIO1_OUT_ADDR, %a1 | ||
255 | move.l #DATA_MASK, %d6 | ||
256 | move.l #CLOCK_MASK, %d7 | ||
257 | |||
258 | move.l #~(RS_MASK+CS_MASK), %d0 | ||
259 | and.l %d0, (%a1) | ||
260 | |||
261 | lsl.l #8, %d2 | ||
262 | or.l %d2, %d3 | ||
263 | bsr.w .write_word | ||
264 | |||
265 | move.l #CS_MASK, %d0 | ||
266 | or.l %d0, (%a1) | ||
267 | |||
268 | movem.l (%sp), %d2-%d3/%d6-%d7 | ||
269 | lea.l (4*4, %sp), %sp | ||
270 | rts | ||
271 | |||
272 | |||
273 | .global lcd_write_data | ||
274 | .type lcd_write_data, @function | ||
275 | |||
276 | lcd_write_data: | ||
277 | lea.l (-7*4, %sp), %sp | ||
278 | movem.l %d2-%d4/%d6-%d7/%a2-%a3, (%sp) | ||
279 | |||
280 | move.l (7*4+4, %sp), %a2 /* p_words */ | ||
281 | move.l (7*4+8, %sp), %d4 /* count */ | ||
282 | |||
283 | lea.l GPIO_OUT_ADDR, %a0 | ||
284 | lea.l GPIO1_OUT_ADDR, %a1 | ||
285 | move.l #DATA_MASK, %d6 | ||
286 | move.l #CLOCK_MASK, %d7 | ||
287 | |||
288 | lea.l .write_word, %a3 | ||
289 | move.l cpu_frequency, %d0 | ||
290 | cmp.l #60000000, %d0 | ||
291 | bhi.b 1f | ||
292 | lea.l .write_word_fast, %a3 | ||
293 | 1: | ||
294 | |||
295 | move.l #RS_MASK, %d0 | ||
296 | or.l %d0, (%a1) | ||
297 | move.l #~CS_MASK, %d0 | ||
298 | and.l %d0, (%a1) | ||
299 | |||
300 | .wd_loop: | ||
301 | clr.l %d3 | ||
302 | move.w (%a2)+, %d3 | ||
303 | jsr (%a3) | ||
304 | subq.l #1, %d4 | ||
305 | bne.s .wd_loop | ||
306 | |||
307 | move.l #CS_MASK, %d0 | ||
308 | or.l %d0, (%a1) | ||
309 | |||
310 | movem.l (%sp), %d2-%d4/%d6-%d7/%a2-%a3 | ||
311 | lea.l (7*4, %sp), %sp | ||
312 | rts | ||
313 | |||
314 | |||
315 | /*** The following functions are only needed for main LCDs ***/ | ||
316 | |||
317 | |||
318 | .global lcd_mono_data | ||
319 | .type lcd_mono_data, @function | ||
320 | |||
321 | lcd_mono_data: | ||
322 | lea.l (-7*4, %sp), %sp | ||
323 | movem.l %d2-%d4/%d6-%d7/%a2-%a3, (%sp) | ||
324 | |||
325 | move.l (7*4+4, %sp), %a2 /* p_bytes */ | ||
326 | move.l (7*4+8, %sp), %d4 /* count */ | ||
327 | |||
328 | lea.l GPIO_OUT_ADDR, %a0 | ||
329 | lea.l GPIO1_OUT_ADDR, %a1 | ||
330 | move.l #DATA_MASK, %d6 | ||
331 | move.l #CLOCK_MASK, %d7 | ||
332 | |||
333 | lea.l .write_word, %a3 | ||
334 | move.l cpu_frequency, %d0 | ||
335 | cmp.l #60000000, %d0 | ||
336 | bhi.b 1f | ||
337 | lea.l .write_word_fast, %a3 | ||
338 | 1: | ||
339 | |||
340 | move.l #RS_MASK, %d0 | ||
341 | or.l %d0, (%a1) | ||
342 | move.l #~CS_MASK, %d0 | ||
343 | and.l %d0, (%a1) | ||
344 | |||
345 | .md_loop: | ||
346 | clr.l %d3 | ||
347 | move.b (%a2)+, %d3 | ||
348 | move.l %d3, %d2 | ||
349 | lsl.l #8, %d2 | ||
350 | or.l %d2, %d3 | ||
351 | jsr (%a3) | ||
352 | subq.l #1, %d4 | ||
353 | bne.s .md_loop | ||
354 | |||
355 | move.l #CS_MASK, %d0 | ||
356 | or.l %d0, (%a1) | ||
357 | |||
358 | movem.l (%sp), %d2-%d4/%d6-%d7/%a2-%a3 | ||
359 | lea.l (7*4, %sp), %sp | ||
360 | rts | ||
361 | |||
362 | |||
363 | .global lcd_grey_data | ||
364 | .type lcd_grey_data,@function | ||
365 | |||
366 | lcd_grey_data: | ||
367 | lea.l (-9*4, %sp), %sp | ||
368 | movem.l %d2-%d7/%a2-%a4, (%sp) | ||
369 | |||
370 | movem.l (9*4+4, %sp), %a2-%a4 /* values, phases, length */ | ||
371 | add.l %a4, %a4 | ||
372 | lea.l (%a3, %a4.l*4), %a4 /* end address */ | ||
373 | |||
374 | lea.l GPIO_OUT_ADDR, %a0 | ||
375 | lea.l GPIO1_OUT_ADDR, %a1 | ||
376 | move.l #DATA_MASK, %d6 | ||
377 | move.l #CLOCK_MASK, %d7 | ||
378 | |||
379 | move.l #RS_MASK, %d0 | ||
380 | or.l %d0, (%a1) | ||
381 | move.l #~CS_MASK, %d0 | ||
382 | and.l %d0, (%a1) | ||
383 | |||
384 | clr.l %d5 | ||
385 | move.l (%a3), %d4 /* fetch 4 pixel phases */ | ||
386 | bclr.l #31, %d4 /* Z = !(p0 & 0x80); p0 &= ~0x80; */ | ||
387 | seq.b %d5 /* %d5 = ........................00000000 */ | ||
388 | lsl.l #1, %d5 /* %d5 = .......................00000000. */ | ||
389 | bclr.l #23, %d4 /* Z = !(p1 & 0x80); p1 &= ~0x80; */ | ||
390 | seq.b %d5 /* %d5 = .......................011111111 */ | ||
391 | lsl.l #1, %d5 /* %d5 = ......................011111111. */ | ||
392 | bclr.l #15, %d4 /* Z = !(p2 & 0x80); p2 &= ~0x80; */ | ||
393 | seq.b %d5 /* %d5 = ......................0122222222 */ | ||
394 | lsl.l #1, %d5 /* %d5 = .....................0122222222. */ | ||
395 | bclr.l #7, %d4 /* Z = !(p3 & 0x80); p3 &= ~0x80; */ | ||
396 | seq.b %d5 /* %d5 = .....................01233333333 */ | ||
397 | lsl.l #1, %d5 /* %d5 = ....................01233333333. */ | ||
398 | add.l (%a2)+, %d4 /* add 4 pixel values to the phases */ | ||
399 | move.l %d4, (%a3)+ /* store new phases, advance pointer */ | ||
400 | |||
401 | move.l (%a3), %d4 /* fetch 4 pixel phases */ | ||
402 | bclr.l #31, %d4 /* Z = !(p0 & 0x80); p0 &= ~0x80; */ | ||
403 | seq.b %d5 /* %d5 = ....................012344444444 */ | ||
404 | lsl.l #1, %d5 /* %d5 = ...................012344444444. */ | ||
405 | bclr.l #23, %d4 /* Z = !(p1 & 0x80); p1 &= ~0x80; */ | ||
406 | seq.b %d5 /* %d5 = ...................0123455555555 */ | ||
407 | lsl.l #1, %d5 /* %d5 = ..................0123455555555. */ | ||
408 | bclr.l #15, %d4 /* Z = !(p2 & 0x80); p2 &= ~0x80; */ | ||
409 | seq.b %d5 /* %d5 = ..................01234566666666 */ | ||
410 | lsl.l #1, %d5 /* %d5 = .................01234566666666. */ | ||
411 | bclr.l #7, %d4 /* Z = !(p3 & 0x80); p3 &= ~0x80; */ | ||
412 | seq.b %d5 /* %d5 = .................012345677777777 */ | ||
413 | lsr.l #7, %d5 /* %d5 = ........................01234567 */ | ||
414 | add.l (%a2)+, %d4 /* add 4 pixel values to the phases */ | ||
415 | move.l %d4, (%a3)+ /* store new phases, advance pointer */ | ||
416 | |||
417 | move.l %d5, %d3 | ||
418 | lsl.l #8, %d3 | ||
419 | or.l %d5, %d3 | ||
420 | |||
421 | cmp.l %a3, %a4 | ||
422 | bls.w .gd_last | ||
423 | |||
424 | .gd_loop: | ||
425 | move.w %sr, %d2 | ||
426 | move.w #0x2700, %sr | ||
427 | |||
428 | move.l (%a0), %d0 /* Get current state of data port */ | ||
429 | move.l %d0, %d1 | ||
430 | and.l %d6, %d1 /* Check current state of data line */ | ||
431 | beq.s 1f /* and set it as previous-state bit */ | ||
432 | bset #16, %d3 | ||
433 | 1: | ||
434 | move.l %d3, %d1 /* Compute the 'bit derivative', i.e. a value */ | ||
435 | lsr.l #1, %d1 /* with 1's where the data changes from the */ | ||
436 | eor.l %d1, %d3 /* previous state, and 0's where it doesn't */ | ||
437 | swap %d3 /* Shift data to upper word */ | ||
438 | |||
439 | eor.l %d7, %d0 /* precalculate opposite state of clock line */ | ||
440 | |||
441 | lsl.l #1,%d3 /* Shift out MSB */ | ||
442 | bcc.s 1f | ||
443 | eor.l %d6, %d0 /* 1: Flip data bit */ | ||
444 | 1: | ||
445 | move.l %d0, %d1 | ||
446 | move.l %d0, (%a0) /* Output new state and set CLK = 0*/ | ||
447 | eor.l %d7, %d1 | ||
448 | |||
449 | move.l (%a3), %d4 /* fetch 4 pixel phases */ | ||
450 | bit_out | ||
451 | bclr.l #31, %d4 /* Z = !(p0 & 0x80); p0 &= ~0x80; */ | ||
452 | seq.b %d5 /* %d5 = ........................00000000 */ | ||
453 | lsl.l #1, %d5 /* %d5 = .......................00000000. */ | ||
454 | bit_out | ||
455 | bclr.l #23, %d4 /* Z = !(p1 & 0x80); p1 &= ~0x80; */ | ||
456 | seq.b %d5 /* %d5 = .......................011111111 */ | ||
457 | lsl.l #1, %d5 /* %d5 = ......................011111111. */ | ||
458 | bit_out | ||
459 | bclr.l #15, %d4 /* Z = !(p2 & 0x80); p2 &= ~0x80; */ | ||
460 | seq.b %d5 /* %d5 = ......................0122222222 */ | ||
461 | lsl.l #1, %d5 /* %d5 = .....................0122222222. */ | ||
462 | bit_out | ||
463 | bclr.l #7, %d4 /* Z = !(p3 & 0x80); p3 &= ~0x80; */ | ||
464 | seq.b %d5 /* %d5 = .....................01233333333 */ | ||
465 | lsl.l #1, %d5 /* %d5 = ....................01233333333. */ | ||
466 | bit_out | ||
467 | add.l (%a2)+, %d4 /* add 4 pixel values to the phases */ | ||
468 | bit_out | ||
469 | move.l %d4, (%a3)+ /* store new phases, advance pointer */ | ||
470 | |||
471 | bit_out | ||
472 | move.l (%a3), %d4 /* fetch 4 pixel phases */ | ||
473 | bit_out | ||
474 | bclr.l #31, %d4 /* Z = !(p0 & 0x80); p0 &= ~0x80; */ | ||
475 | seq.b %d5 /* %d5 = ....................012344444444 */ | ||
476 | lsl.l #1, %d5 /* %d5 = ...................012344444444. */ | ||
477 | bit_out | ||
478 | bclr.l #23, %d4 /* Z = !(p1 & 0x80); p1 &= ~0x80; */ | ||
479 | seq.b %d5 /* %d5 = ...................0123455555555 */ | ||
480 | lsl.l #1, %d5 /* %d5 = ..................0123455555555. */ | ||
481 | bit_out | ||
482 | bclr.l #15, %d4 /* Z = !(p2 & 0x80); p2 &= ~0x80; */ | ||
483 | seq.b %d5 /* %d5 = ..................01234566666666 */ | ||
484 | lsl.l #1, %d5 /* %d5 = .................01234566666666. */ | ||
485 | bit_out | ||
486 | bclr.l #7, %d4 /* Z = !(p3 & 0x80); p3 &= ~0x80; */ | ||
487 | seq.b %d5 /* %d5 = .................012345677777777 */ | ||
488 | lsr.l #7, %d5 /* %d5 = ........................01234567 */ | ||
489 | bit_out | ||
490 | add.l (%a2)+, %d4 /* add 4 pixel values to the phases */ | ||
491 | bit_out | ||
492 | move.l %d4, (%a3)+ /* store new phases, advance pointer */ | ||
493 | |||
494 | bit_out | ||
495 | nop | ||
496 | bit_out | ||
497 | move.l %d5, %d3 | ||
498 | lsl.l #8, %d3 | ||
499 | or.l %d5, %d3 | ||
500 | |||
501 | nop | ||
502 | move.l %d1, (%a0) /* Set CLK = 1 (delayed) */ | ||
503 | move.w %d2, %sr | ||
504 | |||
505 | cmp.l %a3, %a4 | ||
506 | bhi.w .gd_loop | ||
507 | |||
508 | .gd_last: | ||
509 | bsr.w .write_word | ||
510 | |||
511 | move.l #CS_MASK, %d0 | ||
512 | or.l %d0, (%a1) | ||
513 | |||
514 | movem.l (%sp), %d2-%d7/%a2-%a4 | ||
515 | lea.l (9*4, %sp), %sp | ||
516 | rts | ||
diff --git a/firmware/target/coldfire/iaudio/m3/lcd-m3.c b/firmware/target/coldfire/iaudio/m3/lcd-m3.c index ae72832a82..81f19bd73f 100644 --- a/firmware/target/coldfire/iaudio/m3/lcd-m3.c +++ b/firmware/target/coldfire/iaudio/m3/lcd-m3.c | |||
@@ -61,263 +61,6 @@ static int cached_contrast = DEFAULT_CONTRAST_SETTING; | |||
61 | bool initialized = false; | 61 | bool initialized = false; |
62 | 62 | ||
63 | 63 | ||
64 | /* Standard low-level byte writer. Requires CLK high on entry */ | ||
65 | static inline void _write_byte(unsigned data) | ||
66 | { | ||
67 | asm volatile ( | ||
68 | "move.l (%[gpo0]), %%d0 \n" /* Get current state of data line */ | ||
69 | "and.l %[dbit], %%d0 \n" | ||
70 | "beq.s 1f \n" /* and set it as previous-state bit */ | ||
71 | "bset #8, %[data] \n" | ||
72 | "1: \n" | ||
73 | "move.l %[data], %%d0 \n" /* Compute the 'bit derivative', i.e. a value */ | ||
74 | "lsr.l #1, %%d0 \n" /* with 1's where the data changes from the */ | ||
75 | "eor.l %%d0, %[data] \n" /* previous state, and 0's where it doesn't */ | ||
76 | "swap %[data] \n" /* Shift data to upper byte */ | ||
77 | "lsl.l #8, %[data] \n" | ||
78 | |||
79 | "move.l %[cbit], %%d1 \n" /* Prepare mask for flipping CLK */ | ||
80 | "or.l %[dbit], %%d1 \n" /* and DATA at once */ | ||
81 | |||
82 | "lsl.l #1,%[data] \n" /* Shift out MSB */ | ||
83 | "bcc.s 1f \n" | ||
84 | "eor.l %%d1, (%[gpo0]) \n" /* 1: Flip both CLK and DATA */ | ||
85 | ".word 0x51fa \n" /* (trapf.w - shadow next insn) */ | ||
86 | "1: \n" | ||
87 | "eor.l %[cbit], (%[gpo0]) \n" /* else flip CLK only */ | ||
88 | "eor.l %[cbit], (%[gpo0]) \n" /* Flip CLK again */ | ||
89 | |||
90 | "lsl.l #1,%[data] \n" /* ..unrolled.. */ | ||
91 | "bcc.s 1f \n" | ||
92 | "eor.l %%d1, (%[gpo0]) \n" | ||
93 | ".word 0x51fa \n" | ||
94 | "1: \n" | ||
95 | "eor.l %[cbit], (%[gpo0]) \n" | ||
96 | "eor.l %[cbit], (%[gpo0]) \n" | ||
97 | |||
98 | "lsl.l #1,%[data] \n" | ||
99 | "bcc.s 1f \n" | ||
100 | "eor.l %%d1, (%[gpo0]) \n" | ||
101 | ".word 0x51fa \n" | ||
102 | "1: \n" | ||
103 | "eor.l %[cbit], (%[gpo0]) \n" | ||
104 | "eor.l %[cbit], (%[gpo0]) \n" | ||
105 | |||
106 | "lsl.l #1,%[data] \n" | ||
107 | "bcc.s 1f \n" | ||
108 | "eor.l %%d1, (%[gpo0]) \n" | ||
109 | ".word 0x51fa \n" | ||
110 | "1: \n" | ||
111 | "eor.l %[cbit], (%[gpo0]) \n" | ||
112 | "eor.l %[cbit], (%[gpo0]) \n" | ||
113 | |||
114 | "lsl.l #1,%[data] \n" | ||
115 | "bcc.s 1f \n" | ||
116 | "eor.l %%d1, (%[gpo0]) \n" | ||
117 | ".word 0x51fa \n" | ||
118 | "1: \n" | ||
119 | "eor.l %[cbit], (%[gpo0]) \n" | ||
120 | "eor.l %[cbit], (%[gpo0]) \n" | ||
121 | |||
122 | "lsl.l #1,%[data] \n" | ||
123 | "bcc.s 1f \n" | ||
124 | "eor.l %%d1, (%[gpo0]) \n" | ||
125 | ".word 0x51fa \n" | ||
126 | "1: \n" | ||
127 | "eor.l %[cbit], (%[gpo0]) \n" | ||
128 | "eor.l %[cbit], (%[gpo0]) \n" | ||
129 | |||
130 | "lsl.l #1,%[data] \n" | ||
131 | "bcc.s 1f \n" | ||
132 | "eor.l %%d1, (%[gpo0]) \n" | ||
133 | ".word 0x51fa \n" | ||
134 | "1: \n" | ||
135 | "eor.l %[cbit], (%[gpo0]) \n" | ||
136 | "eor.l %[cbit], (%[gpo0]) \n" | ||
137 | |||
138 | "lsl.l #1,%[data] \n" | ||
139 | "bcc.s 1f \n" | ||
140 | "eor.l %%d1, (%[gpo0]) \n" | ||
141 | ".word 0x51fa \n" | ||
142 | "1: \n" | ||
143 | "eor.l %[cbit], (%[gpo0]) \n" | ||
144 | "eor.l %[cbit], (%[gpo0]) \n" | ||
145 | : /* outputs */ | ||
146 | [data]"+d"(data) | ||
147 | : /* inputs */ | ||
148 | [gpo0]"a"(&GPIO_OUT), | ||
149 | [cbit]"d"(0x20000000), | ||
150 | [dbit]"d"(0x04000000) | ||
151 | : /* clobbers */ | ||
152 | "d0", "d1" | ||
153 | ); | ||
154 | } | ||
155 | |||
156 | /* Fast low-level byte writer. Don't use with high CPU clock. | ||
157 | * Requires CLK high on entry */ | ||
158 | static inline void _write_fast(unsigned data) | ||
159 | { | ||
160 | asm volatile ( | ||
161 | "move.w %%sr,%%d3 \n" /* Get current interrupt level */ | ||
162 | "move.w #0x2700,%%sr \n" /* Disable interrupts */ | ||
163 | |||
164 | "move.l (%[gpo0]), %%d0 \n" /* Get current state of data port */ | ||
165 | "move.l %%d0, %%d1 \n" | ||
166 | "and.l %[dbit], %%d1 \n" /* Check current state of data line */ | ||
167 | "beq.s 1f \n" /* and set it as previous-state bit */ | ||
168 | "bset #8, %[data] \n" | ||
169 | "1: \n" | ||
170 | "move.l %[data], %%d1 \n" /* Compute the 'bit derivative', i.e. a value */ | ||
171 | "lsr.l #1, %%d1 \n" /* with 1's where the data changes from the */ | ||
172 | "eor.l %%d1, %[data] \n" /* previous state, and 0's where it doesn't */ | ||
173 | "swap %[data] \n" /* Shift data to upper byte */ | ||
174 | "lsl.l #8, %[data] \n" | ||
175 | |||
176 | "move.l %%d0, %%d1 \n" /* precalculate opposite state of clock line */ | ||
177 | "eor.l %[cbit], %%d1 \n" | ||
178 | |||
179 | "lsl.l #1,%[data] \n" /* Shift out MSB */ | ||
180 | "bcc.s 1f \n" | ||
181 | "eor.l %[dbit], %%d0 \n" /* 1: Flip data bit */ | ||
182 | "eor.l %[dbit], %%d1 \n" /* for both clock states */ | ||
183 | "1: \n" | ||
184 | "move.l %%d1, (%[gpo0]) \n" /* Output new state and set CLK */ | ||
185 | "move.l %%d0, (%[gpo0]) \n" /* reset CLK */ | ||
186 | |||
187 | "lsl.l #1,%[data] \n" /* ..unrolled.. */ | ||
188 | "bcc.s 1f \n" | ||
189 | "eor.l %[dbit], %%d0 \n" | ||
190 | "eor.l %[dbit], %%d1 \n" | ||
191 | "1: \n" | ||
192 | "move.l %%d1, (%[gpo0]) \n" | ||
193 | "move.l %%d0, (%[gpo0]) \n" | ||
194 | |||
195 | "lsl.l #1,%[data] \n" | ||
196 | "bcc.s 1f \n" | ||
197 | "eor.l %[dbit], %%d0 \n" | ||
198 | "eor.l %[dbit], %%d1 \n" | ||
199 | "1: \n" | ||
200 | "move.l %%d1, (%[gpo0]) \n" | ||
201 | "move.l %%d0, (%[gpo0]) \n" | ||
202 | |||
203 | "lsl.l #1,%[data] \n" | ||
204 | "bcc.s 1f \n" | ||
205 | "eor.l %[dbit], %%d0 \n" | ||
206 | "eor.l %[dbit], %%d1 \n" | ||
207 | "1: \n" | ||
208 | "move.l %%d1, (%[gpo0]) \n" | ||
209 | "move.l %%d0, (%[gpo0]) \n" | ||
210 | |||
211 | "lsl.l #1,%[data] \n" | ||
212 | "bcc.s 1f \n" | ||
213 | "eor.l %[dbit], %%d0 \n" | ||
214 | "eor.l %[dbit], %%d1 \n" | ||
215 | "1: \n" | ||
216 | "move.l %%d1, (%[gpo0]) \n" | ||
217 | "move.l %%d0, (%[gpo0]) \n" | ||
218 | |||
219 | "lsl.l #1,%[data] \n" | ||
220 | "bcc.s 1f \n" | ||
221 | "eor.l %[dbit], %%d0 \n" | ||
222 | "eor.l %[dbit], %%d1 \n" | ||
223 | "1: \n" | ||
224 | "move.l %%d1, (%[gpo0]) \n" | ||
225 | "move.l %%d0, (%[gpo0]) \n" | ||
226 | |||
227 | "lsl.l #1,%[data] \n" | ||
228 | "bcc.s 1f \n" | ||
229 | "eor.l %[dbit], %%d0 \n" | ||
230 | "eor.l %[dbit], %%d1 \n" | ||
231 | "1: \n" | ||
232 | "move.l %%d1, (%[gpo0]) \n" | ||
233 | "move.l %%d0, (%[gpo0]) \n" | ||
234 | |||
235 | "lsl.l #1,%[data] \n" | ||
236 | "bcc.s 1f \n" | ||
237 | "eor.l %[dbit], %%d0 \n" | ||
238 | "eor.l %[dbit], %%d1 \n" | ||
239 | "1: \n" | ||
240 | "move.l %%d1, (%[gpo0]) \n" | ||
241 | "move.l %%d0, (%[gpo0]) \n" | ||
242 | |||
243 | "move.w %%d3, %%sr \n" /* Restore interrupt level */ | ||
244 | : /* outputs */ | ||
245 | [data]"+d"(data) | ||
246 | : /* inputs */ | ||
247 | [gpo0]"a"(&GPIO_OUT), | ||
248 | [cbit]"d"(0x20000000), | ||
249 | [dbit]"d"(0x04000000) | ||
250 | : /* clobbers */ | ||
251 | "d0", "d1", "d2", "d3" | ||
252 | ); | ||
253 | } | ||
254 | |||
255 | void lcd_write_command(int cmd) | ||
256 | { | ||
257 | RS_LO; | ||
258 | CS_LO; | ||
259 | _write_byte(cmd); | ||
260 | CS_HI; | ||
261 | } | ||
262 | |||
263 | void lcd_write_command_e(int cmd, int data) | ||
264 | { | ||
265 | RS_LO; | ||
266 | CS_LO; | ||
267 | _write_byte(cmd); | ||
268 | _write_byte(data); | ||
269 | CS_HI; | ||
270 | } | ||
271 | |||
272 | void lcd_write_data(const fb_data *p_words, int count) | ||
273 | { | ||
274 | const unsigned char *p_bytes = (const unsigned char *)p_words; | ||
275 | const unsigned char *p_end = (const unsigned char *)(p_words + count); | ||
276 | |||
277 | RS_HI; | ||
278 | CS_LO; | ||
279 | if (cpu_frequency < 50000000) | ||
280 | { | ||
281 | while (p_bytes < p_end) | ||
282 | _write_fast(*p_bytes++); | ||
283 | } | ||
284 | else | ||
285 | { | ||
286 | while (p_bytes < p_end) | ||
287 | _write_byte(*p_bytes++); | ||
288 | } | ||
289 | CS_HI; | ||
290 | } | ||
291 | |||
292 | static void lcd_mono_data(const unsigned char *p_words, int count) | ||
293 | { | ||
294 | unsigned data; | ||
295 | const unsigned char *p_bytes = p_words; | ||
296 | const unsigned char *p_end = p_words + count; | ||
297 | |||
298 | RS_HI; | ||
299 | CS_LO; | ||
300 | if (cpu_frequency < 50000000) | ||
301 | { | ||
302 | while (p_bytes < p_end) | ||
303 | { | ||
304 | data = *p_bytes++; | ||
305 | _write_fast(data); | ||
306 | _write_fast(data); | ||
307 | } | ||
308 | } | ||
309 | else | ||
310 | { | ||
311 | while (p_bytes < p_end) | ||
312 | { | ||
313 | data = *p_bytes++; | ||
314 | _write_byte(data); | ||
315 | _write_byte(data); | ||
316 | } | ||
317 | } | ||
318 | CS_HI; | ||
319 | } | ||
320 | |||
321 | int lcd_default_contrast(void) | 64 | int lcd_default_contrast(void) |
322 | { | 65 | { |
323 | return DEFAULT_CONTRAST_SETTING; | 66 | return DEFAULT_CONTRAST_SETTING; |
@@ -457,6 +200,9 @@ void lcd_init_device(void) | |||
457 | #endif | 200 | #endif |
458 | } | 201 | } |
459 | 202 | ||
203 | /* Helper function. */ | ||
204 | void lcd_mono_data(const unsigned char *data, int count); | ||
205 | |||
460 | /* Performance function that works with an external buffer | 206 | /* Performance function that works with an external buffer |
461 | note that by and bheight are in 8-pixel units! */ | 207 | note that by and bheight are in 8-pixel units! */ |
462 | void lcd_blit_mono(const unsigned char *data, int x, int by, int width, | 208 | void lcd_blit_mono(const unsigned char *data, int x, int by, int width, |
@@ -477,36 +223,44 @@ void lcd_blit_mono(const unsigned char *data, int x, int by, int width, | |||
477 | } | 223 | } |
478 | } | 224 | } |
479 | 225 | ||
480 | /* TODO: implement grey blit function */ | 226 | /* Helper function for lcd_grey_phase_blit(). */ |
227 | void lcd_grey_data(unsigned char *values, unsigned char *phases, int count); | ||
481 | 228 | ||
482 | /* Performance function that works with an external buffer | 229 | /* Performance function that works with an external buffer |
483 | note that by and bheight are in 8-pixel units! */ | 230 | note that by and bheight are in 8-pixel units! */ |
484 | void lcd_blit_grey_phase(unsigned char *values, unsigned char *phases, | 231 | void lcd_blit_grey_phase(unsigned char *values, unsigned char *phases, |
485 | int x, int by, int width, int bheight, int stride) | 232 | int x, int by, int width, int bheight, int stride) |
486 | { | 233 | { |
487 | (void)values; | 234 | if (initialized) |
488 | (void)phases; | 235 | { |
489 | (void)x; | 236 | stride <<= 3; /* 8 pixels per block */ |
490 | (void)by; | 237 | while (bheight--) |
491 | (void)width; | 238 | { |
492 | (void)bheight; | 239 | lcd_write_command(LCD_SET_PAGE | ((by > 5 ? by + 2 : by) & 0xf)); |
493 | (void)stride; | 240 | lcd_write_command_e(LCD_SET_COLUMN | ((x >> 4) & 0xf), x & 0xf); |
241 | |||
242 | lcd_grey_data(values, phases, width); | ||
243 | values += stride; | ||
244 | phases += stride; | ||
245 | by++; | ||
246 | } | ||
247 | } | ||
494 | } | 248 | } |
495 | 249 | ||
496 | /* Update the display. | 250 | /* Update the display. |
497 | This must be called after all other LCD functions that change the display. */ | 251 | This must be called after all other LCD functions that change the display. */ |
498 | void lcd_update(void) ICODE_ATTR; | ||
499 | void lcd_update(void) | 252 | void lcd_update(void) |
500 | { | 253 | { |
501 | int y; | 254 | int y; |
502 | if (initialized) | 255 | if (initialized) |
503 | { | 256 | { |
504 | for(y = 0;y < LCD_FBHEIGHT;y++) { | 257 | for(y = 0;y < LCD_FBHEIGHT;y++) |
258 | { | ||
505 | /* Copy display bitmap to hardware. | 259 | /* Copy display bitmap to hardware. |
506 | The COM48-COM63 lines are not connected so we have to skip | 260 | The COM48-COM63 lines are not connected so we have to skip |
507 | them. Further, the column address doesn't wrap, so we | 261 | them. Further, the column address doesn't wrap, so we |
508 | have to update one page at a time. */ | 262 | have to update one page at a time. */ |
509 | lcd_write_command(LCD_SET_PAGE | (y>5?y+2:y)); | 263 | lcd_write_command(LCD_SET_PAGE | (y > 5 ? y + 2 : y)); |
510 | lcd_write_command_e(LCD_SET_COLUMN | 0, 0); | 264 | lcd_write_command_e(LCD_SET_COLUMN | 0, 0); |
511 | lcd_write_data(lcd_framebuffer[y], LCD_WIDTH); | 265 | lcd_write_data(lcd_framebuffer[y], LCD_WIDTH); |
512 | } | 266 | } |
@@ -514,7 +268,6 @@ void lcd_update(void) | |||
514 | } | 268 | } |
515 | 269 | ||
516 | /* Update a fraction of the display. */ | 270 | /* Update a fraction of the display. */ |
517 | void lcd_update_rect(int, int, int, int) ICODE_ATTR; | ||
518 | void lcd_update_rect(int x, int y, int width, int height) | 271 | void lcd_update_rect(int x, int y, int width, int height) |
519 | { | 272 | { |
520 | if (initialized) | 273 | if (initialized) |
@@ -536,7 +289,7 @@ void lcd_update_rect(int x, int y, int width, int height) | |||
536 | COM48-COM63 are not connected, so we need to skip those */ | 289 | COM48-COM63 are not connected, so we need to skip those */ |
537 | for (; y <= ymax; y++) | 290 | for (; y <= ymax; y++) |
538 | { | 291 | { |
539 | lcd_write_command(LCD_SET_PAGE | ((y > 5?y + 2:y) & 0xf)); | 292 | lcd_write_command(LCD_SET_PAGE | ((y > 5 ? y + 2 : y) & 0xf)); |
540 | lcd_write_command_e(LCD_SET_COLUMN | ((x >> 4) & 0xf), x & 0xf); | 293 | lcd_write_command_e(LCD_SET_COLUMN | ((x >> 4) & 0xf), x & 0xf); |
541 | 294 | ||
542 | lcd_write_data(&lcd_framebuffer[y][x], width); | 295 | lcd_write_data(&lcd_framebuffer[y][x], width); |
@@ -547,19 +300,23 @@ void lcd_update_rect(int x, int y, int width, int height) | |||
547 | void lcd_set_invert_display(bool yesno) | 300 | void lcd_set_invert_display(bool yesno) |
548 | { | 301 | { |
549 | cached_invert = yesno; | 302 | cached_invert = yesno; |
550 | if(initialized) | 303 | if (initialized) |
551 | lcd_write_command(LCD_REVERSE | yesno); | 304 | lcd_write_command(LCD_REVERSE | yesno); |
552 | } | 305 | } |
553 | 306 | ||
554 | void lcd_set_flip(bool yesno) | 307 | void lcd_set_flip(bool yesno) |
555 | { | 308 | { |
556 | cached_flip = yesno; | 309 | cached_flip = yesno; |
557 | if(initialized) { | 310 | if (initialized) |
558 | if(yesno) { | 311 | { |
312 | if(yesno) | ||
313 | { | ||
559 | lcd_write_command(LCD_SELECT_ADC | 0); | 314 | lcd_write_command(LCD_SELECT_ADC | 0); |
560 | lcd_write_command(LCD_SELECT_SHL | 0); | 315 | lcd_write_command(LCD_SELECT_SHL | 0); |
561 | lcd_write_command_e(LCD_SET_COM0, 16); | 316 | lcd_write_command_e(LCD_SET_COM0, 16); |
562 | } else { | 317 | } |
318 | else | ||
319 | { | ||
563 | lcd_write_command(LCD_SELECT_ADC | 1); | 320 | lcd_write_command(LCD_SELECT_ADC | 1); |
564 | lcd_write_command(LCD_SELECT_SHL | 8); | 321 | lcd_write_command(LCD_SELECT_SHL | 8); |
565 | lcd_write_command_e(LCD_SET_COM0, 0); | 322 | lcd_write_command_e(LCD_SET_COM0, 0); |