summaryrefslogtreecommitdiff
path: root/firmware/target/arm/ipod/lcd-as-color-nano.S
diff options
context:
space:
mode:
Diffstat (limited to 'firmware/target/arm/ipod/lcd-as-color-nano.S')
-rwxr-xr-xfirmware/target/arm/ipod/lcd-as-color-nano.S377
1 files changed, 256 insertions, 121 deletions
diff --git a/firmware/target/arm/ipod/lcd-as-color-nano.S b/firmware/target/arm/ipod/lcd-as-color-nano.S
index d4df4d496a..f6f9cc5be3 100755
--- a/firmware/target/arm/ipod/lcd-as-color-nano.S
+++ b/firmware/target/arm/ipod/lcd-as-color-nano.S
@@ -7,7 +7,7 @@
7 * \/ \/ \/ \/ \/ 7 * \/ \/ \/ \/ \/
8 * $Id:$ 8 * $Id:$
9 * 9 *
10 * Copyright (C) 2010 by Andree Buschmann 10 * Copyright (C) 2010-2011 by Andree Buschmann
11 * 11 *
12 * Generic asm helper function used by YUV blitting. 12 * Generic asm helper function used by YUV blitting.
13 * 13 *
@@ -24,129 +24,264 @@
24#include "config.h" 24#include "config.h"
25#include "cpu.h" 25#include "cpu.h"
26 26
27 .section .icode, "ax", %progbits 27/****************************************************************************
28 * #define FORCE_FIFO_WAIT
29 *
30 * This is not needed in YUV blitting when the LCD IF is fast enough. In this
31 * case YUV-to-RGB conversion per pixel needs longer than the transfer of a
32 * pixel via the LCD IF.
33 ****************************************************************************/
34
35#include "config.h"
36
37/* Set FIFO wait for both iPod Color and iPod nano1G until we know for which
38 * devices we can switch this off. */
39#define FORCE_FIFO_WAIT
28 40
41 .section .icode, "ax", %progbits
42
29/**************************************************************************** 43/****************************************************************************
30* void lcd_yuv_write_inner_loop(unsigned char const * const ysrc, 44 * extern void lcd_write_yuv420_lines(unsigned char const * const src[3],
31* unsigned char const * const usrc, 45 * const unsigned LCD_BASE,
32* unsigned char const * const vsrc, 46 * int width,
33* int width); 47 * int stride);
34* 48 *
35* YUV- > RGB565 conversion 49 * Conversion from Motion JPEG and MPEG Y'PbPr to RGB is:
36* |R| |1.000000 -0.000001 1.402000| |Y'| 50 * |R| |1.164 0.000 1.596| |Y' - 16|
37* |G| = |1.000000 -0.334136 -0.714136| |Pb| 51 * |G| = |1.164 -0.391 -0.813| |Pb - 128|
38* |B| |1.000000 1.772000 0.000000| |Pr| 52 * |B| |1.164 2.018 0.000| |Pr - 128|
39* Scaled, normalized, rounded and tweaked to yield RGB 565: 53 *
40* |R| |74 0 101| |Y' - 16| >> 9 54 * Scaled, normalized, rounded and tweaked to yield RGB 565:
41* |G| = |74 -24 -51| |Cb - 128| >> 8 55 * |R| |74 0 101| |Y' - 16| >> 9
42* |B| |74 128 0| |Cr - 128| >> 9 56 * |G| = |74 -24 -51| |Cb - 128| >> 8
43* 57 * |B| |74 128 0| |Cr - 128| >> 9
44*/ 58 *
59 * Converts two lines from YUV to RGB565 and writes to LCD at once. First loop
60 * loads Cb/Cr, calculates the chroma offset and saves them to buffer. Within
61 * the second loop these chroma offset are reloaded from buffer. Within each
62 * loop two pixels are calculated and written to LCD.
63 */
45 .align 2 64 .align 2
46 .global lcd_yuv_write_inner_loop 65 .global lcd_write_yuv420_lines
47 .type lcd_yuv_write_inner_loop, %function 66 .type lcd_write_yuv420_lines, %function
48 67lcd_write_yuv420_lines:
49lcd_yuv_write_inner_loop: 68 /* r0 = src = yuv_src */
50 @ r0 = ysrc 69 /* r1 = dst = LCD_BASE */
51 @ r1 = usrc 70 /* r2 = width */
52 @ r2 = vsrc 71 /* r3 = stride */
53 @ r3 = width 72 stmfd sp!, { r4-r10, lr } /* save non-scratch */
54 stmfd sp!, { r4-r11, lr } @ save regs 73 ldmia r0, { r9, r10, r12 } /* r9 = yuv_src[0] = Y'_p */
55 mov r4, #0x70000000 @ r4 = LCD2_BLOCK_CTRL - 0x20 74 /* r10 = yuv_src[1] = Cb_p */
56 add r4, r4, #0x8a00 @ 75 /* r12 = yuv_src[2] = Cr_p */
57 add r5, r4, #0x100 @ r5 = LCD2_BLOCK_DATA 76 add r3, r9, r3 /* r3 = &ysrc[stride] */
5810: @ loop 77 add r4, r2, r2, asr #1 /* chroma buffer lenght = width/2 *3 */
59 78 mov r4, r4, asl #2 /* use words for str/ldm possibility */
60 ldrb r7, [r1], #1 @ *usrc++ 79 add r4, r4, #19 /* plus room for 4 additional words, */
61 ldrb r8, [r2], #1 @ *vsrc++ 80 bic r4, r4, #3 /* rounded up to multiples of 4 byte */
62 81 sub sp, sp, r4 /* and allocate on stack */
63 sub r7, r7, #128 @ Cb -= 128 82 stmia sp, {r1-r4} /* LCD_BASE, width, &ysrc[stride], stack_alloc */
64 sub r8, r8, #128 @ Cr -= 128 83
65 84 mov r7, r2 /* r7 = loop count */
66 add r10, r8, r8, asl #2 @ Cr*101 85 add r8, sp, #16 /* chroma buffer */
67 add r10, r10, r8, asl #5 86 add lr, r1, #0x100 /* LCD data port = LCD2_BASE + 0x100 */
68 add r10, r10, r8, asl #6 87
69 88 /* 1st loop start */
70 add r11, r8, r8, asl #1 @ Cr*51 + Cb*24 8910: /* loop start */
71 add r11, r11, r11, asl #4 90
72 add r11, r11, r7, asl #3 91 ldrb r0, [r10], #1 /* r0 = *usrc++ = *Cb_p++ */
73 add r11, r11, r7, asl #4 92 ldrb r1, [r12], #1 /* r1 = *vsrc++ = *Cr_p++ */
74 93
75 add r12, r7, #2 @ r12 = bu = (Cb*128 + 256) >> 9 94 sub r0, r0, #128 /* r0 = Cb-128 */
76 mov r12, r12, asr #2 95 sub r1, r1, #128 /* r1 = Cr-128 */
77 add r10, r10, #256 @ r10 = rv = (Cr*101 + 256) >> 9 96
78 mov r10, r10, asr #9 97 add r2, r1, r1, asl #1 /* r2 = Cr*51 + Cb*24 */
79 rsb r11, r11, #128 @ r11 = guv = (-r11 + 128) >> 8 98 add r2, r2, r2, asl #4
80 mov r11, r11, asr #8 99 add r2, r2, r0, asl #3
81 100 add r2, r2, r0, asl #4
82@ pixel_1 101
83 ldrb r7, [r0], #1 @ *ysrc++ 102 add r4, r1, r1, asl #2 /* r1 = Cr*101 */
84 sub r7, r7, #16 @ Y = (Y' - 16) * 37 103 add r4, r4, r1, asl #5
85 add r8, r7, r7, asl #2 104 add r1, r4, r1, asl #6
86 add r7, r8, r7, asl #5 105
87 106 add r1, r1, #256 /* r1 = rv = (r1 + 256) >> 9 */
88 add r9, r10, r7, asr #8 @ R = (Y >> 8) + rv 107 mov r1, r1, asr #9
89 add r8, r11, r7, asr #7 @ G = (Y >> 7) + guv 108 rsb r2, r2, #128 /* r2 = guv = (-r2 + 128) >> 8 */
90 add r7, r12, r7, asr #8 @ B = (Y >> 8) + bu 109 mov r2, r2, asr #8
91 110 add r0, r0, #2 /* r0 = bu = (Cb*128 + 256) >> 9 */
92 cmp r9, #31 @ clamp R 111 mov r0, r0, asr #2
93 mvnhi r9, r9, asr #31 112 stmia r8!, {r0-r2} /* store r0, r1 and r2 to chroma buffer */
94 andhi r9, r9, #31 113
95 114 /* 1st loop, first pixel */
96 cmp r8, #63 @ clamp G 115 ldrb r5, [r9], #1 /* r5 = *ysrc++ = *Y'_p++ */
97 mvnhi r8, r8, asr #31 116 sub r5, r5, #16 /* r5 = (Y'-16) * 74 */
98 andhi r8, r8, #63 117 add r3, r5, r5, asl #2
99 118 add r5, r3, r5, asl #5
100 cmp r7, #31 @ clamp B 119
101 mvnhi r7, r7, asr #31 120 add r6, r1, r5, asr #8 /* r6 = r = (Y >> 9) + rv */
102 andhi r7, r7, #31 121 add r3, r2, r5, asr #7 /* r3 = g = (Y >> 8) + guv */
103 122 add r4, r0, r5, asr #8 /* r4 = b = (Y >> 9) + bu */
104 orr r6, r7, r8, lsl #5 @ pack pixel 123
105 orr r6, r6, r9, lsl #11 124 orr r5, r6, r4 /* check if clamping is needed... */
106 125 orr r5, r5, r3, asr #1 /* ...at all */
107 mov r7, r6, lsl #8 @ swap bytes 126 cmp r5, #31
108 and r7, r7, #0xff00 127 bls 15f /* -> no clamp */
109 add r6, r7, r6, lsr #8 128 cmp r6, #31 /* clamp r */
110 129 mvnhi r6, r6, asr #31
111@ pixel_2 130 andhi r6, r6, #31
112 ldrb r7, [r0], #1 @ *ysrc++ 131 cmp r3, #63 /* clamp g */
113 sub r7, r7, #16 @ Y = (Y' - 16) * 37 132 mvnhi r3, r3, asr #31
114 add r8, r7, r7, asl #2 133 andhi r3, r3, #63
115 add r7, r8, r7, asl #5 134 cmp r4, #31 /* clamp b */
116 135 mvnhi r4, r4, asr #31
117 add r9, r10, r7, asr #8 @ R = (Y >> 8) + rv 136 andhi r4, r4, #31
118 add r8, r11, r7, asr #7 @ G = (Y >> 7) + guv 13715: /* no clamp */
119 add r7, r12, r7, asr #8 @ B = (Y >> 8) + bu 138
120 139 /* calculate pixel_1 and save to r4 for later pixel packing */
121 cmp r9, #31 @ clamp R 140 orr r4, r4, r3, lsl #5 /* pixel_1 = r<<11 | g<<5 | b */
122 mvnhi r9, r9, asr #31 141 orr r4, r4, r6, lsl #11 /* r4 = pixel_1 */
123 andhi r9, r9, #31 142
124 143 /* 1st loop, second pixel */
125 cmp r8, #63 @ clamp G 144 ldrb r5, [r9], #1 /* r5 = *ysrc++ = *Y'_p++ */
126 mvnhi r8, r8, asr #31 145 sub r5, r5, #16 /* r5 = (Y'-16) * 74 */
127 andhi r8, r8, #63 146 add r3, r5, r5, asl #2
128 147 add r5, r3, r5, asl #5
129 cmp r7, #31 @ clamp B 148
130 mvnhi r7, r7, asr #31 149 add r6, r1, r5, asr #8 /* r6 = r = (Y >> 9) + rv */
131 andhi r7, r7, #31 150 add r3, r2, r5, asr #7 /* r3 = g = (Y >> 8) + guv */
132 151 add r5, r0, r5, asr #8 /* r5 = b = (Y >> 9) + bu */
133 orr r7, r7, r8, lsl #5 @ pack pixel 152
134 orr r7, r7, r9, lsl #11 153 orr r0, r6, r5 /* check if clamping is needed... */
135 154 orr r0, r0, r3, asr #1 /* ...at all */
136 orr r6, r6, r7, lsl #24 @ swap bytes and add pixels simultaneously 155 cmp r0, #31
137 mov r7, r7, lsr #8 156 bls 15f /* -> no clamp */
138 orr r6, r6, r7, lsl #16 157 cmp r6, #31 /* clamp r */
139#if 1 158 mvnhi r6, r6, asr #31
14011: @ while (!(LCD2_BLOCK_CTRL & LCD2_BLOCK_TXOK)); 159 andhi r6, r6, #31
141 ldr r11, [r4, #0x20] @ 160 cmp r3, #63 /* clamp g */
142 tst r11, #0x1000000 @ 161 mvnhi r3, r3, asr #31
143 beq 11b @ 162 andhi r3, r3, #63
163 cmp r5, #31 /* clamp b */
164 mvnhi r5, r5, asr #31
165 andhi r5, r5, #31
16615: /* no clamp */
167
168 /* calculate pixel_2 and pack with pixel_1 before writing */
169 orr r5, r5, r3, lsl #5 /* pixel_2 = r<<11 | g<<5 | b */
170 orr r5, r5, r6, lsl #11 /* r5 = pixel_2 */
171#ifdef FORCE_FIFO_WAIT
172 /* wait for FIFO half full */
173.fifo_wait1:
174 ldr r3, [lr, #-0xE0] /* while !(LCD2_BLOCK_CTRL & 0x1000000); */
175 tst r3, #0x1000000
176 beq .fifo_wait1
144#endif 177#endif
145 str r6, [r5] @ send two pixels
146 178
147 subs r3, r3, #2 @ decrease width 179 mov r3, r4, lsl #8 /* swap pixel_1 */
148 bgt 10b @ loop 180 and r3, r3, #0xff00
181 add r4, r3, r4, lsr #8
182
183 orr r4, r4, r5, lsl #24 /* swap pixel_2 and pack with pixel_1 */
184 mov r5, r5, lsr #8
185 orr r4, r4, r5, lsl #16
186
187 str r4, [lr] /* write pixel_1 and pixel_2 */
188
189 subs r7, r7, #2 /* check for loop end */
190 bgt 10b /* back to beginning */
191 /* 1st loop end */
192
193 /* Reload several registers for pointer rewinding for next loop */
194 add r8, sp, #16 /* chroma buffer */
195 ldmia sp, { r1, r7, r9} /* r1 = LCD_BASE */
196 /* r7 = loop count */
197 /* r9 = &ysrc[stride] */
198
199 /* 2nd loop start */
20020: /* loop start */
201 /* restore r0 (bu), r1 (rv) and r2 (guv) from chroma buffer */
202 ldmia r8!, {r0-r2}
203
204 /* 2nd loop, first pixel */
205 ldrb r5, [r9], #1 /* r5 = *ysrc++ = *Y'_p++ */
206 sub r5, r5, #16 /* r5 = (Y'-16) * 74 */
207 add r3, r5, r5, asl #2
208 add r5, r3, r5, asl #5
209
210 add r6, r1, r5, asr #8 /* r6 = r = (Y >> 9) + rv */
211 add r3, r2, r5, asr #7 /* r3 = g = (Y >> 8) + guv */
212 add r4, r0, r5, asr #8 /* r4 = b = (Y >> 9) + bu */
213
214 orr r5, r6, r4 /* check if clamping is needed... */
215 orr r5, r5, r3, asr #1 /* ...at all */
216 cmp r5, #31
217 bls 15f /* -> no clamp */
218 cmp r6, #31 /* clamp r */
219 mvnhi r6, r6, asr #31
220 andhi r6, r6, #31
221 cmp r3, #63 /* clamp g */
222 mvnhi r3, r3, asr #31
223 andhi r3, r3, #63
224 cmp r4, #31 /* clamp b */
225 mvnhi r4, r4, asr #31
226 andhi r4, r4, #31
22715: /* no clamp */
228 /* calculate pixel_1 and save to r4 for later pixel packing */
229 orr r4, r4, r3, lsl #5 /* pixel_1 = r<<11 | g<<5 | b */
230 orr r4, r4, r6, lsl #11 /* r4 = pixel_1 */
231
232 /* 2nd loop, second pixel */
233 ldrb r5, [r9], #1 /* r5 = *ysrc++ = *Y'_p++ */
234 sub r5, r5, #16 /* r5 = (Y'-16) * 74 */
235 add r3, r5, r5, asl #2
236 add r5, r3, r5, asl #5
237
238 add r6, r1, r5, asr #8 /* r6 = r = (Y >> 9) + rv */
239 add r3, r2, r5, asr #7 /* r3 = g = (Y >> 8) + guv */
240 add r5, r0, r5, asr #8 /* r5 = b = (Y >> 9) + bu */
241
242 orr r0, r6, r5 /* check if clamping is needed... */
243 orr r0, r0, r3, asr #1 /* ...at all */
244 cmp r0, #31
245 bls 15f /* -> no clamp */
246 cmp r6, #31 /* clamp r */
247 mvnhi r6, r6, asr #31
248 andhi r6, r6, #31
249 cmp r3, #63 /* clamp g */
250 mvnhi r3, r3, asr #31
251 andhi r3, r3, #63
252 cmp r5, #31 /* clamp b */
253 mvnhi r5, r5, asr #31
254 andhi r5, r5, #31
25515: /* no clamp */
256
257 /* calculate pixel_2 and pack with pixel_1 before writing */
258 orr r5, r5, r3, lsl #5 /* pixel_2 = r<<11 | g<<5 | b */
259 orr r5, r5, r6, lsl #11 /* r5 = pixel_2 */
260#ifdef FORCE_FIFO_WAIT
261 /* wait for FIFO half full */
262.fifo_wait2:
263 ldr r3, [lr, #-0xE0] /* while !(LCD2_BLOCK_CTRL & 0x1000000); */
264 tst r3, #0x1000000
265 beq .fifo_wait2
266#endif
267
268 mov r3, r4, lsl #8 /* swap pixel_1 */
269 and r3, r3, #0xff00
270 add r4, r3, r4, lsr #8
271
272 orr r4, r4, r5, lsl #24 /* swap pixel_2 and pack with pixel_1 */
273 mov r5, r5, lsr #8
274 orr r4, r4, r5, lsl #16
275
276 str r4, [lr] /* write pixel_1 and pixel_2 */
277
278 subs r7, r7, #2 /* check for loop end */
279 bgt 20b /* back to beginning */
280 /* 2nd loop end */
281
282 ldr r3, [sp, #12]
283 add sp, sp, r3 /* deallocate buffer */
284 ldmpc regs=r4-r10 /* restore registers */
149 285
150 ldmpc regs=r4-r11 @ restore regs 286 .ltorg
151 .ltorg @ dump constant pool 287 .size lcd_write_yuv420_lines, .-lcd_write_yuv420_lines
152 .size lcd_yuv_write_inner_loop, .-lcd_yuv_write_inner_loop
generated by cgit v1.2.3 (git 2.39.1) at 2024-09-12 17:51:13 +0000