apps/plugins/mpegplayer/idct.c

   1 /*
   2  * idct.c
   3  * Copyright (C) 2000-2003 Michel Lespinasse <walken@zoy.org>
   4  * Copyright (C) 1999-2000 Aaron Holtzman <aholtzma@ess.engr.uvic.ca>
   5  *
   6  * This file is part of mpeg2dec, a free MPEG-2 video stream decoder.
   7  * See http://libmpeg2.sourceforge.net/ for updates.
   8  *
   9  * mpeg2dec is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * mpeg2dec is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License
  20  * along with this program; if not, write to the Free Software
  21  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  22  */
  23
  24 #include "plugin.h"
  25
  26 #include "mpeg2dec_config.h"
  27
  28 #include "mpeg2.h"
  29 #include "attributes.h"
  30 #include "mpeg2_internal.h"
  31
  32 #define W1 2841 /* 2048 * sqrt (2) * cos (1 * pi / 16) */
  33 #define W2 2676 /* 2048 * sqrt (2) * cos (2 * pi / 16) */
  34 #define W3 2408 /* 2048 * sqrt (2) * cos (3 * pi / 16) */
  35 #define W5 1609 /* 2048 * sqrt (2) * cos (5 * pi / 16) */
  36 #define W6 1108 /* 2048 * sqrt (2) * cos (6 * pi / 16) */
  37 #define W7 565  /* 2048 * sqrt (2) * cos (7 * pi / 16) */
  38
  39 /* idct main entry point  */
  40 void (* mpeg2_idct_copy) (int16_t * block, uint8_t * dest, int stride);
  41 void (* mpeg2_idct_add) (int last, int16_t * block,
  42                          uint8_t * dest, int stride);
  43
  44 /*
  45  * In legal streams, the IDCT output should be between -384 and +384.
  46  * In corrupted streams, it is possible to force the IDCT output to go
  47  * to +-3826 - this is the worst case for a column IDCT where the
  48  * column inputs are 16-bit values.
  49  */
  50 #ifdef CPU_COLDFIRE
  51 static inline unsigned CLIP(int value)
  52 {
  53     asm (  /* Note: Uses knowledge that only the low byte of the result is used */
  54         "cmp.l   #255,%[v]   \n"  /* overflow? */
  55         "bls.b   1f          \n"  /* no: return value */
  56         "spl.b   %[v]        \n"  /* yes: set low byte to appropriate boundary */
  57     "1:                      \n"
  58         : /* outputs */
  59         [v]"+d"(value)
  60     );
  61     return value;
  62 }
  63 #elif defined CPU_ARM
  64 static inline unsigned CLIP(int value)
  65 {
  66     asm volatile ( /* Note: Uses knowledge that only the low byte of the result is used */
  67         "cmp     %[v], #255          \n"
  68         "mvnhi   %[v], %[v], asr #31 \n"
  69         : /* outputs */
  70         [v]"+r"(value)
  71     );
  72     return value;
  73 }
  74 #else
  75 uint8_t mpeg2_clip[3840 * 2 + 256] IBSS_ATTR;
  76 #define CLIP(i) ((mpeg2_clip + 3840)[i])
  77 #endif
  78
  79 #if 0
  80 #define BUTTERFLY(t0,t1,W0,W1,d0,d1) \
  81     do {                             \
  82         t0 = W0 * d0 + W1 * d1;      \
  83         t1 = W0 * d1 - W1 * d0;      \
  84     } while (0)
  85 #else
  86 #define BUTTERFLY(t0,t1,W0,W1,d0,d1) \
  87     do {                             \
  88         int tmp = W0 * (d0 + d1);    \
  89         t0 = tmp + (W1 - W0) * d1;   \
  90         t1 = tmp - (W1 + W0) * d0;   \
  91     } while (0)
  92 #endif
  93
  94 static inline void idct_row (int16_t * const block)
  95 {
  96     int d0, d1, d2, d3;
  97     int a0, a1, a2, a3, b0, b1, b2, b3;
  98     int t0, t1, t2, t3;
  99
 100     /* shortcut */
 101     if (likely (!(block[1] | ((int32_t *)block)[1] | ((int32_t *)block)[2] |
 102                   ((int32_t *)block)[3]))) {
 103         uint32_t tmp = (uint16_t) (block[0] >> 1);
 104         tmp |= tmp << 16;
 105         ((int32_t *)block)[0] = tmp;
 106         ((int32_t *)block)[1] = tmp;
 107         ((int32_t *)block)[2] = tmp;
 108         ((int32_t *)block)[3] = tmp;
 109         return;
 110     }
 111
 112     d0 = (block[0] << 11) + 2048;
 113     d1 = block[1];
 114     d2 = block[2] << 11;
 115     d3 = block[3];
 116     t0 = d0 + d2;
 117     t1 = d0 - d2;
 118     BUTTERFLY (t2, t3, W6, W2, d3, d1);
 119     a0 = t0 + t2;
 120     a1 = t1 + t3;
 121     a2 = t1 - t3;
 122     a3 = t0 - t2;
 123
 124     d0 = block[4];
 125     d1 = block[5];
 126     d2 = block[6];
 127     d3 = block[7];
 128     BUTTERFLY (t0, t1, W7, W1, d3, d0);
 129     BUTTERFLY (t2, t3, W3, W5, d1, d2);
 130     b0 = t0 + t2;
 131     b3 = t1 + t3;
 132     t0 -= t2;
 133     t1 -= t3;
 134     b1 = ((t0 + t1) >> 8) * 181;
 135     b2 = ((t0 - t1) >> 8) * 181;
 136
 137     block[0] = (a0 + b0) >> 12;
 138     block[1] = (a1 + b1) >> 12;
 139     block[2] = (a2 + b2) >> 12;
 140     block[3] = (a3 + b3) >> 12;
 141     block[4] = (a3 - b3) >> 12;
 142     block[5] = (a2 - b2) >> 12;
 143     block[6] = (a1 - b1) >> 12;
 144     block[7] = (a0 - b0) >> 12;
 145 }
 146
 147 static inline void idct_col (int16_t * const block)
 148 {
 149     int d0, d1, d2, d3;
 150     int a0, a1, a2, a3, b0, b1, b2, b3;
 151     int t0, t1, t2, t3;
 152
 153     d0 = (block[8*0] << 11) + 65536;
 154     d1 = block[8*1];
 155     d2 = block[8*2] << 11;
 156     d3 = block[8*3];
 157     t0 = d0 + d2;
 158     t1 = d0 - d2;
 159     BUTTERFLY (t2, t3, W6, W2, d3, d1);
 160     a0 = t0 + t2;
 161     a1 = t1 + t3;
 162     a2 = t1 - t3;
 163     a3 = t0 - t2;
 164
 165     d0 = block[8*4];
 166     d1 = block[8*5];
 167     d2 = block[8*6];
 168     d3 = block[8*7];
 169     BUTTERFLY (t0, t1, W7, W1, d3, d0);
 170     BUTTERFLY (t2, t3, W3, W5, d1, d2);
 171     b0 = t0 + t2;
 172     b3 = t1 + t3;
 173     t0 -= t2;
 174     t1 -= t3;
 175     b1 = ((t0 + t1) >> 8) * 181;
 176     b2 = ((t0 - t1) >> 8) * 181;
 177
 178     block[8*0] = (a0 + b0) >> 17;
 179     block[8*1] = (a1 + b1) >> 17;
 180     block[8*2] = (a2 + b2) >> 17;
 181     block[8*3] = (a3 + b3) >> 17;
 182     block[8*4] = (a3 - b3) >> 17;
 183     block[8*5] = (a2 - b2) >> 17;
 184     block[8*6] = (a1 - b1) >> 17;
 185     block[8*7] = (a0 - b0) >> 17;
 186 }
 187
 188 static void mpeg2_idct_copy_c (int16_t * block, uint8_t * dest,
 189                                const int stride)
 190 {
 191     int i;
 192
 193     for (i = 0; i < 8; i++)
 194         idct_row (block + 8 * i);
 195     for (i = 0; i < 8; i++)
 196         idct_col (block + i);
 197     do {
 198         dest[0] = CLIP (block[0]);
 199         dest[1] = CLIP (block[1]);
 200         dest[2] = CLIP (block[2]);
 201         dest[3] = CLIP (block[3]);
 202         dest[4] = CLIP (block[4]);
 203         dest[5] = CLIP (block[5]);
 204         dest[6] = CLIP (block[6]);
 205         dest[7] = CLIP (block[7]);
 206
 207         ((int32_t *)block)[0] = 0;        ((int32_t *)block)[1] = 0;
 208         ((int32_t *)block)[2] = 0;        ((int32_t *)block)[3] = 0;
 209
 210         dest += stride;
 211         block += 8;
 212     } while (--i);
 213 }
 214
 215 static void mpeg2_idct_add_c (const int last, int16_t * block,
 216                               uint8_t * dest, const int stride)
 217 {
 218     int i;
 219
 220     if (last != 129 || (block[0] & (7 << 4)) == (4 << 4)) {
 221         for (i = 0; i < 8; i++)
 222             idct_row (block + 8 * i);
 223         for (i = 0; i < 8; i++)
 224             idct_col (block + i);
 225         do {
 226             dest[0] = CLIP (block[0] + dest[0]);
 227             dest[1] = CLIP (block[1] + dest[1]);
 228             dest[2] = CLIP (block[2] + dest[2]);
 229             dest[3] = CLIP (block[3] + dest[3]);
 230             dest[4] = CLIP (block[4] + dest[4]);
 231             dest[5] = CLIP (block[5] + dest[5]);
 232             dest[6] = CLIP (block[6] + dest[6]);
 233             dest[7] = CLIP (block[7] + dest[7]);
 234
 235             ((int32_t *)block)[0] = 0;        ((int32_t *)block)[1] = 0;
 236             ((int32_t *)block)[2] = 0;        ((int32_t *)block)[3] = 0;
 237
 238             dest += stride;
 239             block += 8;
 240         } while (--i);
 241     } else {
 242         int DC;
 243
 244         DC = (block[0] + 64) >> 7;
 245         block[0] = block[63] = 0;
 246         i = 8;
 247         do {
 248             dest[0] = CLIP (DC + dest[0]);
 249             dest[1] = CLIP (DC + dest[1]);
 250             dest[2] = CLIP (DC + dest[2]);
 251             dest[3] = CLIP (DC + dest[3]);
 252             dest[4] = CLIP (DC + dest[4]);
 253             dest[5] = CLIP (DC + dest[5]);
 254             dest[6] = CLIP (DC + dest[6]);
 255             dest[7] = CLIP (DC + dest[7]);
 256             dest += stride;
 257         } while (--i);
 258     }
 259 }
 260
 261 void mpeg2_idct_init (void)
 262 {
 263     extern uint8_t default_mpeg2_scan_norm[64];
 264     extern uint8_t default_mpeg2_scan_alt[64];
 265     extern uint8_t mpeg2_scan_norm[64];
 266     extern uint8_t mpeg2_scan_alt[64];
 267     int i, j;
 268
 269     mpeg2_idct_copy = mpeg2_idct_copy_c;
 270     mpeg2_idct_add = mpeg2_idct_add_c;
 271
 272 #if !defined(CPU_COLDFIRE) && !defined(CPU_ARM)
 273     for (i = -3840; i < 3840 + 256; i++)
 274         CLIP(i) = (i < 0) ? 0 : ((i > 255) ? 255 : i);
 275 #endif
 276
 277     for (i = 0; i < 64; i++)
 278     {
 279         j = default_mpeg2_scan_norm[i];
 280         mpeg2_scan_norm[i] = ((j & 0x36) >> 1) | ((j & 0x09) << 2);
 281
 282         j = default_mpeg2_scan_alt[i];
 283         mpeg2_scan_alt[i] = ((j & 0x36) >> 1) | ((j & 0x09) << 2);
 284     }
 285 }