1/* 2 * Copyright (c) 2012 3 * MIPS Technologies, Inc., California. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its 14 * contributors may be used to endorse or promote products derived from 15 * this software without specific prior written permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * 29 * Authors: Branimir Vasic (bvasic@mips.com) 30 * Nedeljko Babic (nbabic@mips.com) 31 * 32 * Various AC-3 DSP Utils optimized for MIPS 33 * 34 * This file is part of FFmpeg. 35 * 36 * FFmpeg is free software; you can redistribute it and/or 37 * modify it under the terms of the GNU Lesser General Public 38 * License as published by the Free Software Foundation; either 39 * version 2.1 of the License, or (at your option) any later version. 40 * 41 * FFmpeg is distributed in the hope that it will be useful, 42 * but WITHOUT ANY WARRANTY; without even the implied warranty of 43 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 44 * Lesser General Public License for more details. 45 * 46 * You should have received a copy of the GNU Lesser General Public 47 * License along with FFmpeg; if not, write to the Free Software 48 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 49 */ 50 51/** 52 * @file 53 * Reference: libavcodec/ac3dsp.c 54 */ 55 56#include "config.h" 57#include "libavcodec/ac3dsp.h" 58#include "libavcodec/ac3.h" 59#include "libavcodec/ac3tab.h" 60#include "libavutil/mips/asmdefs.h" 61 62#if HAVE_INLINE_ASM 63#if HAVE_MIPSDSP 64static void ac3_bit_alloc_calc_bap_mips(int16_t *mask, int16_t *psd, 65 int start, int end, 66 int snr_offset, int floor, 67 const uint8_t *bap_tab, uint8_t *bap) 68{ 69 int band, band_end, cond; 70 int m, address1, address2; 71 int16_t *psd1, *psd_end; 72 uint8_t *bap1; 73 74 if (snr_offset == -960) { 75 memset(bap, 0, AC3_MAX_COEFS); 76 return; 77 } 78 79 psd1 = &psd[start]; 80 bap1 = &bap[start]; 81 band = ff_ac3_bin_to_band_tab[start]; 82 83 do { 84 m = (FFMAX(mask[band] - snr_offset - floor, 0) & 0x1FE0) + floor; 85 band_end = ff_ac3_band_start_tab[++band]; 86 band_end = FFMIN(band_end, end); 87 psd_end = psd + band_end - 1; 88 89 __asm__ volatile ( 90 "slt %[cond], %[psd1], %[psd_end] \n\t" 91 "beqz %[cond], 1f \n\t" 92 "2: \n\t" 93 "lh %[address1], 0(%[psd1]) \n\t" 94 "lh %[address2], 2(%[psd1]) \n\t" 95 PTR_ADDIU " %[psd1], %[psd1], 4 \n\t" 96 "subu %[address1], %[address1], %[m] \n\t" 97 "sra %[address1], %[address1], 5 \n\t" 98 "addiu %[address1], %[address1], -32 \n\t" 99 "shll_s.w %[address1], %[address1], 26 \n\t" 100 "subu %[address2], %[address2], %[m] \n\t" 101 "sra %[address2], %[address2], 5 \n\t" 102 "sra %[address1], %[address1], 26 \n\t" 103 "addiu %[address1], %[address1], 32 \n\t" 104 "lbux %[address1], %[address1](%[bap_tab]) \n\t" 105 "addiu %[address2], %[address2], -32 \n\t" 106 "shll_s.w %[address2], %[address2], 26 \n\t" 107 "sb %[address1], 0(%[bap1]) \n\t" 108 "slt %[cond], %[psd1], %[psd_end] \n\t" 109 "sra %[address2], %[address2], 26 \n\t" 110 "addiu %[address2], %[address2], 32 \n\t" 111 "lbux %[address2], %[address2](%[bap_tab]) \n\t" 112 "sb %[address2], 1(%[bap1]) \n\t" 113 PTR_ADDIU " %[bap1], %[bap1], 2 \n\t" 114 "bnez %[cond], 2b \n\t" 115 PTR_ADDIU " %[psd_end], %[psd_end], 2 \n\t" 116 "slt %[cond], %[psd1], %[psd_end] \n\t" 117 "beqz %[cond], 3f \n\t" 118 "1: \n\t" 119 "lh %[address1], 0(%[psd1]) \n\t" 120 PTR_ADDIU " %[psd1], %[psd1], 2 \n\t" 121 "subu %[address1], %[address1], %[m] \n\t" 122 "sra %[address1], %[address1], 5 \n\t" 123 "addiu %[address1], %[address1], -32 \n\t" 124 "shll_s.w %[address1], %[address1], 26 \n\t" 125 "sra %[address1], %[address1], 26 \n\t" 126 "addiu %[address1], %[address1], 32 \n\t" 127 "lbux %[address1], %[address1](%[bap_tab]) \n\t" 128 "sb %[address1], 0(%[bap1]) \n\t" 129 PTR_ADDIU " %[bap1], %[bap1], 1 \n\t" 130 "3: \n\t" 131 132 : [address1]"=&r"(address1), [address2]"=&r"(address2), 133 [cond]"=&r"(cond), [bap1]"+r"(bap1), 134 [psd1]"+r"(psd1), [psd_end]"+r"(psd_end) 135 : [m]"r"(m), [bap_tab]"r"(bap_tab) 136 : "memory" 137 ); 138 } while (end > band_end); 139} 140 141static void ac3_update_bap_counts_mips(uint16_t mant_cnt[16], uint8_t *bap, 142 int len) 143{ 144 void *temp0, *temp2, *temp4, *temp5, *temp6, *temp7; 145 int temp1, temp3; 146 147 __asm__ volatile ( 148 "andi %[temp3], %[len], 3 \n\t" 149 PTR_ADDU "%[temp2], %[bap], %[len] \n\t" 150 PTR_ADDU "%[temp4], %[bap], %[temp3] \n\t" 151 "beq %[temp2], %[temp4], 4f \n\t" 152 "1: \n\t" 153 "lbu %[temp0], -1(%[temp2]) \n\t" 154 "lbu %[temp5], -2(%[temp2]) \n\t" 155 "lbu %[temp6], -3(%[temp2]) \n\t" 156 "sll %[temp0], %[temp0], 1 \n\t" 157 PTR_ADDU "%[temp0], %[mant_cnt], %[temp0] \n\t" 158 "sll %[temp5], %[temp5], 1 \n\t" 159 PTR_ADDU "%[temp5], %[mant_cnt], %[temp5] \n\t" 160 "lhu %[temp1], 0(%[temp0]) \n\t" 161 "sll %[temp6], %[temp6], 1 \n\t" 162 PTR_ADDU "%[temp6], %[mant_cnt], %[temp6] \n\t" 163 "addiu %[temp1], %[temp1], 1 \n\t" 164 "sh %[temp1], 0(%[temp0]) \n\t" 165 "lhu %[temp1], 0(%[temp5]) \n\t" 166 "lbu %[temp7], -4(%[temp2]) \n\t" 167 PTR_ADDIU "%[temp2],%[temp2], -4 \n\t" 168 "addiu %[temp1], %[temp1], 1 \n\t" 169 "sh %[temp1], 0(%[temp5]) \n\t" 170 "lhu %[temp1], 0(%[temp6]) \n\t" 171 "sll %[temp7], %[temp7], 1 \n\t" 172 PTR_ADDU "%[temp7], %[mant_cnt], %[temp7] \n\t" 173 "addiu %[temp1], %[temp1],1 \n\t" 174 "sh %[temp1], 0(%[temp6]) \n\t" 175 "lhu %[temp1], 0(%[temp7]) \n\t" 176 "addiu %[temp1], %[temp1], 1 \n\t" 177 "sh %[temp1], 0(%[temp7]) \n\t" 178 "bne %[temp2], %[temp4], 1b \n\t" 179 "4: \n\t" 180 "beqz %[temp3], 2f \n\t" 181 "3: \n\t" 182 "addiu %[temp3], %[temp3], -1 \n\t" 183 "lbu %[temp0], -1(%[temp2]) \n\t" 184 PTR_ADDIU "%[temp2],%[temp2], -1 \n\t" 185 "sll %[temp0], %[temp0], 1 \n\t" 186 PTR_ADDU "%[temp0], %[mant_cnt], %[temp0] \n\t" 187 "lhu %[temp1], 0(%[temp0]) \n\t" 188 "addiu %[temp1], %[temp1], 1 \n\t" 189 "sh %[temp1], 0(%[temp0]) \n\t" 190 "bgtz %[temp3], 3b \n\t" 191 "2: \n\t" 192 193 : [temp0] "=&r" (temp0), [temp1] "=&r" (temp1), 194 [temp2] "=&r" (temp2), [temp3] "=&r" (temp3), 195 [temp4] "=&r" (temp4), [temp5] "=&r" (temp5), 196 [temp6] "=&r" (temp6), [temp7] "=&r" (temp7) 197 : [len] "r" (len), [bap] "r" (bap), 198 [mant_cnt] "r" (mant_cnt) 199 : "memory" 200 ); 201} 202#endif 203 204#if HAVE_MIPSFPU 205#if !HAVE_MIPS32R6 && !HAVE_MIPS64R6 206static void float_to_fixed24_mips(int32_t *dst, const float *src, unsigned int len) 207{ 208 const float scale = 1 << 24; 209 float src0, src1, src2, src3, src4, src5, src6, src7; 210 int temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7; 211 212 do { 213 __asm__ volatile ( 214 "lwc1 %[src0], 0(%[src]) \n\t" 215 "lwc1 %[src1], 4(%[src]) \n\t" 216 "lwc1 %[src2], 8(%[src]) \n\t" 217 "lwc1 %[src3], 12(%[src]) \n\t" 218 "lwc1 %[src4], 16(%[src]) \n\t" 219 "lwc1 %[src5], 20(%[src]) \n\t" 220 "lwc1 %[src6], 24(%[src]) \n\t" 221 "lwc1 %[src7], 28(%[src]) \n\t" 222 "mul.s %[src0], %[src0], %[scale] \n\t" 223 "mul.s %[src1], %[src1], %[scale] \n\t" 224 "mul.s %[src2], %[src2], %[scale] \n\t" 225 "mul.s %[src3], %[src3], %[scale] \n\t" 226 "mul.s %[src4], %[src4], %[scale] \n\t" 227 "mul.s %[src5], %[src5], %[scale] \n\t" 228 "mul.s %[src6], %[src6], %[scale] \n\t" 229 "mul.s %[src7], %[src7], %[scale] \n\t" 230 "cvt.w.s %[src0], %[src0] \n\t" 231 "cvt.w.s %[src1], %[src1] \n\t" 232 "cvt.w.s %[src2], %[src2] \n\t" 233 "cvt.w.s %[src3], %[src3] \n\t" 234 "cvt.w.s %[src4], %[src4] \n\t" 235 "cvt.w.s %[src5], %[src5] \n\t" 236 "cvt.w.s %[src6], %[src6] \n\t" 237 "cvt.w.s %[src7], %[src7] \n\t" 238 "mfc1 %[temp0], %[src0] \n\t" 239 "mfc1 %[temp1], %[src1] \n\t" 240 "mfc1 %[temp2], %[src2] \n\t" 241 "mfc1 %[temp3], %[src3] \n\t" 242 "mfc1 %[temp4], %[src4] \n\t" 243 "mfc1 %[temp5], %[src5] \n\t" 244 "mfc1 %[temp6], %[src6] \n\t" 245 "mfc1 %[temp7], %[src7] \n\t" 246 "sw %[temp0], 0(%[dst]) \n\t" 247 "sw %[temp1], 4(%[dst]) \n\t" 248 "sw %[temp2], 8(%[dst]) \n\t" 249 "sw %[temp3], 12(%[dst]) \n\t" 250 "sw %[temp4], 16(%[dst]) \n\t" 251 "sw %[temp5], 20(%[dst]) \n\t" 252 "sw %[temp6], 24(%[dst]) \n\t" 253 "sw %[temp7], 28(%[dst]) \n\t" 254 255 : [dst] "+r" (dst), [src] "+r" (src), 256 [src0] "=&f" (src0), [src1] "=&f" (src1), 257 [src2] "=&f" (src2), [src3] "=&f" (src3), 258 [src4] "=&f" (src4), [src5] "=&f" (src5), 259 [src6] "=&f" (src6), [src7] "=&f" (src7), 260 [temp0] "=r" (temp0), [temp1] "=r" (temp1), 261 [temp2] "=r" (temp2), [temp3] "=r" (temp3), 262 [temp4] "=r" (temp4), [temp5] "=r" (temp5), 263 [temp6] "=r" (temp6), [temp7] "=r" (temp7) 264 : [scale] "f" (scale) 265 : "memory" 266 ); 267 src = src + 8; 268 dst = dst + 8; 269 len -= 8; 270 } while (len > 0); 271} 272 273static void ac3_downmix_mips(float **samples, float (*matrix)[2], 274 int out_ch, int in_ch, int len) 275{ 276 int i, j, i1, i2, i3; 277 float v0, v1, v2, v3; 278 float v4, v5, v6, v7; 279 float samples0, samples1, samples2, samples3, matrix_j, matrix_j2; 280 float *samples_p, *samples_sw, *matrix_p, **samples_x, **samples_end; 281 282 __asm__ volatile( 283 ".set push \n\t" 284 ".set noreorder \n\t" 285 286 "li %[i1], 2 \n\t" 287 "sll %[len], 2 \n\t" 288 "move %[i], $zero \n\t" 289 "sll %[j], %[in_ch], " PTRLOG " \n\t" 290 291 "bne %[out_ch], %[i1], 3f \n\t" // if (out_ch == 2) 292 " li %[i2], 1 \n\t" 293 294 "2: \n\t" // start of the for loop (for (i = 0; i < len; i+=4)) 295 "move %[matrix_p], %[matrix] \n\t" 296 "move %[samples_x], %[samples] \n\t" 297 "mtc1 $zero, %[v0] \n\t" 298 "mtc1 $zero, %[v1] \n\t" 299 "mtc1 $zero, %[v2] \n\t" 300 "mtc1 $zero, %[v3] \n\t" 301 "mtc1 $zero, %[v4] \n\t" 302 "mtc1 $zero, %[v5] \n\t" 303 "mtc1 $zero, %[v6] \n\t" 304 "mtc1 $zero, %[v7] \n\t" 305 "addiu %[i1], %[i], 4 \n\t" 306 "addiu %[i2], %[i], 8 \n\t" 307 PTR_L " %[samples_p], 0(%[samples_x]) \n\t" 308 "addiu %[i3], %[i], 12 \n\t" 309 PTR_ADDU "%[samples_end],%[samples_x], %[j] \n\t" 310 "move %[samples_sw], %[samples_p] \n\t" 311 312 "1: \n\t" // start of the inner for loop (for (j = 0; j < in_ch; j++)) 313 "lwc1 %[matrix_j], 0(%[matrix_p]) \n\t" 314 "lwc1 %[matrix_j2], 4(%[matrix_p]) \n\t" 315 "lwxc1 %[samples0], %[i](%[samples_p]) \n\t" 316 "lwxc1 %[samples1], %[i1](%[samples_p]) \n\t" 317 "lwxc1 %[samples2], %[i2](%[samples_p]) \n\t" 318 "lwxc1 %[samples3], %[i3](%[samples_p]) \n\t" 319 PTR_ADDIU "%[matrix_p], 8 \n\t" 320 PTR_ADDIU "%[samples_x]," PTRSIZE " \n\t" 321 "madd.s %[v0], %[v0], %[samples0], %[matrix_j] \n\t" 322 "madd.s %[v1], %[v1], %[samples1], %[matrix_j] \n\t" 323 "madd.s %[v2], %[v2], %[samples2], %[matrix_j] \n\t" 324 "madd.s %[v3], %[v3], %[samples3], %[matrix_j] \n\t" 325 "madd.s %[v4], %[v4], %[samples0], %[matrix_j2]\n\t" 326 "madd.s %[v5], %[v5], %[samples1], %[matrix_j2]\n\t" 327 "madd.s %[v6], %[v6], %[samples2], %[matrix_j2]\n\t" 328 "madd.s %[v7], %[v7], %[samples3], %[matrix_j2]\n\t" 329 "bne %[samples_x], %[samples_end], 1b \n\t" 330 PTR_L " %[samples_p], 0(%[samples_x]) \n\t" 331 332 PTR_L " %[samples_p], " PTRSIZE "(%[samples]) \n\t" 333 "swxc1 %[v0], %[i](%[samples_sw]) \n\t" 334 "swxc1 %[v1], %[i1](%[samples_sw]) \n\t" 335 "swxc1 %[v2], %[i2](%[samples_sw]) \n\t" 336 "swxc1 %[v3], %[i3](%[samples_sw]) \n\t" 337 "swxc1 %[v4], %[i](%[samples_p]) \n\t" 338 "addiu %[i], 16 \n\t" 339 "swxc1 %[v5], %[i1](%[samples_p]) \n\t" 340 "swxc1 %[v6], %[i2](%[samples_p]) \n\t" 341 "bne %[i], %[len], 2b \n\t" 342 " swxc1 %[v7], %[i3](%[samples_p]) \n\t" 343 344 "3: \n\t" 345 "bne %[out_ch], %[i2], 6f \n\t" // if (out_ch == 1) 346 " nop \n\t" 347 348 "5: \n\t" // start of the outer for loop (for (i = 0; i < len; i+=4)) 349 "move %[matrix_p], %[matrix] \n\t" 350 "move %[samples_x], %[samples] \n\t" 351 "mtc1 $zero, %[v0] \n\t" 352 "mtc1 $zero, %[v1] \n\t" 353 "mtc1 $zero, %[v2] \n\t" 354 "mtc1 $zero, %[v3] \n\t" 355 "addiu %[i1], %[i], 4 \n\t" 356 "addiu %[i2], %[i], 8 \n\t" 357 PTR_L " %[samples_p], 0(%[samples_x]) \n\t" 358 "addiu %[i3], %[i], 12 \n\t" 359 PTR_ADDU "%[samples_end],%[samples_x], %[j] \n\t" 360 "move %[samples_sw], %[samples_p] \n\t" 361 362 "4: \n\t" // start of the inner for loop (for (j = 0; j < in_ch; j++)) 363 "lwc1 %[matrix_j], 0(%[matrix_p]) \n\t" 364 "lwxc1 %[samples0], %[i](%[samples_p]) \n\t" 365 "lwxc1 %[samples1], %[i1](%[samples_p]) \n\t" 366 "lwxc1 %[samples2], %[i2](%[samples_p]) \n\t" 367 "lwxc1 %[samples3], %[i3](%[samples_p]) \n\t" 368 PTR_ADDIU "%[matrix_p], 8 \n\t" 369 PTR_ADDIU "%[samples_x]," PTRSIZE " \n\t" 370 "madd.s %[v0], %[v0], %[samples0], %[matrix_j] \n\t" 371 "madd.s %[v1], %[v1], %[samples1], %[matrix_j] \n\t" 372 "madd.s %[v2], %[v2], %[samples2], %[matrix_j] \n\t" 373 "madd.s %[v3], %[v3], %[samples3], %[matrix_j] \n\t" 374 "bne %[samples_x], %[samples_end], 4b \n\t" 375 PTR_L " %[samples_p], 0(%[samples_x]) \n\t" 376 377 "swxc1 %[v0], %[i](%[samples_sw]) \n\t" 378 "addiu %[i], 16 \n\t" 379 "swxc1 %[v1], %[i1](%[samples_sw]) \n\t" 380 "swxc1 %[v2], %[i2](%[samples_sw]) \n\t" 381 "bne %[i], %[len], 5b \n\t" 382 " swxc1 %[v3], %[i3](%[samples_sw]) \n\t" 383 "6: \n\t" 384 385 ".set pop" 386 :[samples_p]"=&r"(samples_p), [matrix_j]"=&f"(matrix_j), [matrix_j2]"=&f"(matrix_j2), 387 [samples0]"=&f"(samples0), [samples1]"=&f"(samples1), 388 [samples2]"=&f"(samples2), [samples3]"=&f"(samples3), 389 [v0]"=&f"(v0), [v1]"=&f"(v1), [v2]"=&f"(v2), [v3]"=&f"(v3), 390 [v4]"=&f"(v4), [v5]"=&f"(v5), [v6]"=&f"(v6), [v7]"=&f"(v7), 391 [samples_x]"=&r"(samples_x), [matrix_p]"=&r"(matrix_p), 392 [samples_end]"=&r"(samples_end), [samples_sw]"=&r"(samples_sw), 393 [i1]"=&r"(i1), [i2]"=&r"(i2), [i3]"=&r"(i3), [i]"=&r"(i), 394 [j]"=&r"(j), [len]"+r"(len) 395 :[samples]"r"(samples), [matrix]"r"(matrix), 396 [in_ch]"r"(in_ch), [out_ch]"r"(out_ch) 397 :"memory" 398 ); 399} 400#endif /* !HAVE_MIPS32R6 && !HAVE_MIPS64R6 */ 401#endif /* HAVE_MIPSFPU */ 402#endif /* HAVE_INLINE_ASM */ 403 404void ff_ac3dsp_init_mips(AC3DSPContext *c, int bit_exact) { 405#if HAVE_INLINE_ASM 406#if HAVE_MIPSDSP 407 c->bit_alloc_calc_bap = ac3_bit_alloc_calc_bap_mips; 408 c->update_bap_counts = ac3_update_bap_counts_mips; 409#endif 410#if HAVE_MIPSFPU 411#if !HAVE_MIPS32R6 && !HAVE_MIPS64R6 412 c->float_to_fixed24 = float_to_fixed24_mips; 413 //c->downmix = ac3_downmix_mips; 414#endif 415#endif 416 417#endif 418} 419