libavcodec/armv4l/dsputil_arm.c

   1 /*
   2  * ARMv4L optimized DSP utils
   3  * Copyright (c) 2001 Lionel Ulmer.
   4  *
   5  * This file is part of FFmpeg.
   6  *
   7  * FFmpeg is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2.1 of the License, or (at your option) any later version.
  11  *
  12  * FFmpeg is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with FFmpeg; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21
  22 #include "../dsputil.h"
  23 #ifdef HAVE_IPP
  24 #include "ipp.h"
  25 #endif
  26
  27 extern void dsputil_init_iwmmxt(DSPContext* c, AVCodecContext *avctx);
  28
  29 extern void j_rev_dct_ARM(DCTELEM *data);
  30 extern void simple_idct_ARM(DCTELEM *data);
  31
  32 extern void simple_idct_armv5te(DCTELEM *data);
  33 extern void simple_idct_put_armv5te(uint8_t *dest, int line_size,
  34                                     DCTELEM *data);
  35 extern void simple_idct_add_armv5te(uint8_t *dest, int line_size,
  36                                     DCTELEM *data);
  37
  38 /* XXX: local hack */
  39 static void (*ff_put_pixels_clamped)(const DCTELEM *block, uint8_t *pixels, int line_size);
  40 static void (*ff_add_pixels_clamped)(const DCTELEM *block, uint8_t *pixels, int line_size);
  41
  42 void put_pixels8_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
  43 void put_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
  44 void put_pixels8_y2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
  45 void put_pixels8_xy2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
  46
  47 void put_no_rnd_pixels8_x2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
  48 void put_no_rnd_pixels8_y2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
  49 void put_no_rnd_pixels8_xy2_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
  50
  51 void put_pixels16_arm(uint8_t *block, const uint8_t *pixels, int line_size, int h);
  52
  53 CALL_2X_PIXELS(put_pixels16_x2_arm , put_pixels8_x2_arm , 8)
  54 CALL_2X_PIXELS(put_pixels16_y2_arm , put_pixels8_y2_arm , 8)
  55 CALL_2X_PIXELS(put_pixels16_xy2_arm, put_pixels8_xy2_arm, 8)
  56 CALL_2X_PIXELS(put_no_rnd_pixels16_x2_arm , put_no_rnd_pixels8_x2_arm , 8)
  57 CALL_2X_PIXELS(put_no_rnd_pixels16_y2_arm , put_no_rnd_pixels8_y2_arm , 8)
  58 CALL_2X_PIXELS(put_no_rnd_pixels16_xy2_arm, put_no_rnd_pixels8_xy2_arm, 8)
  59
  60 static void add_pixels_clamped_ARM(short *block, unsigned char *dest, int line_size)
  61 {
  62     asm volatile (
  63                   "mov r10, #8 \n\t"
  64
  65                   "1: \n\t"
  66
  67                   /* load dest */
  68                   "ldr r4, [%1] \n\t"
  69                   /* block[0] and block[1]*/
  70                   "ldrsh r5, [%0] \n\t"
  71                   "ldrsh r7, [%0, #2] \n\t"
  72                   "and r6, r4, #0xFF \n\t"
  73                   "and r8, r4, #0xFF00 \n\t"
  74                   "add r6, r5, r6 \n\t"
  75                   "add r8, r7, r8, lsr #8 \n\t"
  76                   "mvn r5, r5 \n\t"
  77                   "mvn r7, r7 \n\t"
  78                   "tst r6, #0x100 \n\t"
  79                   "movne r6, r5, lsr #24 \n\t"
  80                   "tst r8, #0x100 \n\t"
  81                   "movne r8, r7, lsr #24 \n\t"
  82                   "mov r9, r6 \n\t"
  83                   "ldrsh r5, [%0, #4] \n\t" /* moved form [A] */
  84                   "orr r9, r9, r8, lsl #8 \n\t"
  85                   /* block[2] and block[3] */
  86                   /* [A] */
  87                   "ldrsh r7, [%0, #6] \n\t"
  88                   "and r6, r4, #0xFF0000 \n\t"
  89                   "and r8, r4, #0xFF000000 \n\t"
  90                   "add r6, r5, r6, lsr #16 \n\t"
  91                   "add r8, r7, r8, lsr #24 \n\t"
  92                   "mvn r5, r5 \n\t"
  93                   "mvn r7, r7 \n\t"
  94                   "tst r6, #0x100 \n\t"
  95                   "movne r6, r5, lsr #24 \n\t"
  96                   "tst r8, #0x100 \n\t"
  97                   "movne r8, r7, lsr #24 \n\t"
  98                   "orr r9, r9, r6, lsl #16 \n\t"
  99                   "ldr r4, [%1, #4] \n\t"       /* moved form [B] */
 100                   "orr r9, r9, r8, lsl #24 \n\t"
 101                   /* store dest */
 102                   "ldrsh r5, [%0, #8] \n\t" /* moved form [C] */
 103                   "str r9, [%1] \n\t"
 104
 105                   /* load dest */
 106                   /* [B] */
 107                   /* block[4] and block[5] */
 108                   /* [C] */
 109                   "ldrsh r7, [%0, #10] \n\t"
 110                   "and r6, r4, #0xFF \n\t"
 111                   "and r8, r4, #0xFF00 \n\t"
 112                   "add r6, r5, r6 \n\t"
 113                   "add r8, r7, r8, lsr #8 \n\t"
 114                   "mvn r5, r5 \n\t"
 115                   "mvn r7, r7 \n\t"
 116                   "tst r6, #0x100 \n\t"
 117                   "movne r6, r5, lsr #24 \n\t"
 118                   "tst r8, #0x100 \n\t"
 119                   "movne r8, r7, lsr #24 \n\t"
 120                   "mov r9, r6 \n\t"
 121                   "ldrsh r5, [%0, #12] \n\t" /* moved from [D] */
 122                   "orr r9, r9, r8, lsl #8 \n\t"
 123                   /* block[6] and block[7] */
 124                   /* [D] */
 125                   "ldrsh r7, [%0, #14] \n\t"
 126                   "and r6, r4, #0xFF0000 \n\t"
 127                   "and r8, r4, #0xFF000000 \n\t"
 128                   "add r6, r5, r6, lsr #16 \n\t"
 129                   "add r8, r7, r8, lsr #24 \n\t"
 130                   "mvn r5, r5 \n\t"
 131                   "mvn r7, r7 \n\t"
 132                   "tst r6, #0x100 \n\t"
 133                   "movne r6, r5, lsr #24 \n\t"
 134                   "tst r8, #0x100 \n\t"
 135                   "movne r8, r7, lsr #24 \n\t"
 136                   "orr r9, r9, r6, lsl #16 \n\t"
 137                   "add %0, %0, #16 \n\t" /* moved from [E] */
 138                   "orr r9, r9, r8, lsl #24 \n\t"
 139                   "subs r10, r10, #1 \n\t" /* moved from [F] */
 140                   /* store dest */
 141                   "str r9, [%1, #4] \n\t"
 142
 143                   /* [E] */
 144                   /* [F] */
 145                   "add %1, %1, %2 \n\t"
 146                   "bne 1b \n\t"
 147                   : "+r"(block),
 148                     "+r"(dest)
 149                   : "r"(line_size)
 150                   : "r4", "r5", "r6", "r7", "r8", "r9", "r10", "cc", "memory" );
 151 }
 152
 153 /* XXX: those functions should be suppressed ASAP when all IDCTs are
 154    converted */
 155 static void j_rev_dct_ARM_put(uint8_t *dest, int line_size, DCTELEM *block)
 156 {
 157     j_rev_dct_ARM (block);
 158     ff_put_pixels_clamped(block, dest, line_size);
 159 }
 160 static void j_rev_dct_ARM_add(uint8_t *dest, int line_size, DCTELEM *block)
 161 {
 162     j_rev_dct_ARM (block);
 163     ff_add_pixels_clamped(block, dest, line_size);
 164 }
 165 static void simple_idct_ARM_put(uint8_t *dest, int line_size, DCTELEM *block)
 166 {
 167     simple_idct_ARM (block);
 168     ff_put_pixels_clamped(block, dest, line_size);
 169 }
 170 static void simple_idct_ARM_add(uint8_t *dest, int line_size, DCTELEM *block)
 171 {
 172     simple_idct_ARM (block);
 173     ff_add_pixels_clamped(block, dest, line_size);
 174 }
 175
 176 #ifdef HAVE_IPP
 177 static void simple_idct_ipp(DCTELEM *block)
 178 {
 179     ippiDCT8x8Inv_Video_16s_C1I(block);
 180 }
 181 static void simple_idct_ipp_put(uint8_t *dest, int line_size, DCTELEM *block)
 182 {
 183     ippiDCT8x8Inv_Video_16s8u_C1R(block, dest, line_size);
 184 }
 185
 186 void add_pixels_clamped_iwmmxt(const DCTELEM *block, uint8_t *pixels, int line_size);
 187
 188 static void simple_idct_ipp_add(uint8_t *dest, int line_size, DCTELEM *block)
 189 {
 190     ippiDCT8x8Inv_Video_16s_C1I(block);
 191 #ifdef HAVE_IWMMXT
 192     add_pixels_clamped_iwmmxt(block, dest, line_size);
 193 #else
 194     add_pixels_clamped_ARM(block, dest, line_size);
 195 #endif
 196 }
 197 #endif
 198
 199 void dsputil_init_armv4l(DSPContext* c, AVCodecContext *avctx)
 200 {
 201     int idct_algo= avctx->idct_algo;
 202
 203     ff_put_pixels_clamped = c->put_pixels_clamped;
 204     ff_add_pixels_clamped = c->add_pixels_clamped;
 205
 206     if(idct_algo == FF_IDCT_AUTO){
 207 #if defined(HAVE_IPP)
 208         idct_algo = FF_IDCT_IPP;
 209 #elif defined(HAVE_ARMV5TE)
 210         idct_algo = FF_IDCT_SIMPLEARMV5TE;
 211 #else
 212         idct_algo = FF_IDCT_ARM;
 213 #endif
 214     }
 215
 216     if(idct_algo==FF_IDCT_ARM){
 217         c->idct_put= j_rev_dct_ARM_put;
 218         c->idct_add= j_rev_dct_ARM_add;
 219         c->idct    = j_rev_dct_ARM;
 220         c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;/* FF_NO_IDCT_PERM */
 221     } else if (idct_algo==FF_IDCT_SIMPLEARM){
 222         c->idct_put= simple_idct_ARM_put;
 223         c->idct_add= simple_idct_ARM_add;
 224         c->idct    = simple_idct_ARM;
 225         c->idct_permutation_type= FF_NO_IDCT_PERM;
 226 #ifdef HAVE_ARMV5TE
 227     } else if (idct_algo==FF_IDCT_SIMPLEARMV5TE){
 228         c->idct_put= simple_idct_put_armv5te;
 229         c->idct_add= simple_idct_add_armv5te;
 230         c->idct    = simple_idct_armv5te;
 231         c->idct_permutation_type = FF_NO_IDCT_PERM;
 232 #endif
 233 #ifdef HAVE_IPP
 234     } else if (idct_algo==FF_IDCT_IPP){
 235         c->idct_put= simple_idct_ipp_put;
 236         c->idct_add= simple_idct_ipp_add;
 237         c->idct    = simple_idct_ipp;
 238         c->idct_permutation_type= FF_NO_IDCT_PERM;
 239 #endif
 240     }
 241
 242 /*     c->put_pixels_tab[0][0] = put_pixels16_arm; */ // NG!
 243     c->put_pixels_tab[0][1] = put_pixels16_x2_arm; //OK!
 244     c->put_pixels_tab[0][2] = put_pixels16_y2_arm; //OK!
 245 /*     c->put_pixels_tab[0][3] = put_pixels16_xy2_arm; /\* NG *\/ */
 246 /*     c->put_no_rnd_pixels_tab[0][0] = put_pixels16_arm; */
 247     c->put_no_rnd_pixels_tab[0][1] = put_no_rnd_pixels16_x2_arm; // OK
 248     c->put_no_rnd_pixels_tab[0][2] = put_no_rnd_pixels16_y2_arm; //OK
 249 /*     c->put_no_rnd_pixels_tab[0][3] = put_no_rnd_pixels16_xy2_arm; //NG */
 250     c->put_pixels_tab[1][0] = put_pixels8_arm; //OK
 251     c->put_pixels_tab[1][1] = put_pixels8_x2_arm; //OK
 252 /*     c->put_pixels_tab[1][2] = put_pixels8_y2_arm; //NG */
 253 /*     c->put_pixels_tab[1][3] = put_pixels8_xy2_arm; //NG */
 254     c->put_no_rnd_pixels_tab[1][0] = put_pixels8_arm;//OK
 255     c->put_no_rnd_pixels_tab[1][1] = put_no_rnd_pixels8_x2_arm; //OK
 256     c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_arm; //OK
 257 /*     c->put_no_rnd_pixels_tab[1][3] = put_no_rnd_pixels8_xy2_arm;//NG */
 258
 259 #ifdef HAVE_IWMMXT
 260     dsputil_init_iwmmxt(c, avctx);
 261 #endif
 262 }