diff -ruN a52dec.orig/include/a52.h a52dec/include/a52.h --- a52dec.orig/include/a52.h 2012-11-13 17:25:33.438758556 +0100 +++ a52dec/include/a52.h 2012-11-13 17:27:31.143704187 +0100 @@ -1,6 +1,6 @@ /* * a52.h - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. @@ -24,10 +24,17 @@ #ifndef A52_H #define A52_H -#ifndef LIBA52_DOUBLE -typedef float sample_t; -#else +#define LIBA52_FIXED + +#if defined(LIBA52_FIXED) +typedef int32_t sample_t; +typedef int32_t level_t; +#elif defined(LIBA52_DOUBLE) typedef double sample_t; +typedef double level_t; +#else +typedef float sample_t; +typedef float level_t; #endif typedef struct a52_state_s a52_state_t; @@ -53,9 +60,9 @@ int a52_syncinfo (uint8_t * buf, int * flags, int * sample_rate, int * bit_rate); int a52_frame (a52_state_t * state, uint8_t * buf, int * flags, - sample_t * level, sample_t bias); + level_t * level, sample_t bias); void a52_dynrng (a52_state_t * state, - sample_t (* call) (sample_t, void *), void * data); + level_t (* call) (level_t, void *), void * data); int a52_block (a52_state_t * state); void a52_free (a52_state_t * state); diff -ruN a52dec.orig/include/attributes.h a52dec/include/attributes.h --- a52dec.orig/include/attributes.h 2012-11-13 17:25:33.438758556 +0100 +++ a52dec/include/attributes.h 2012-11-13 17:27:31.147037566 +0100 @@ -1,6 +1,6 @@ /* * attributes.h - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. @@ -27,3 +27,11 @@ #else #define ATTR_ALIGN(align) #endif + +#ifdef HAVE_BUILTIN_EXPECT +#define likely(x) __builtin_expect ((x) != 0, 1) +#define unlikely(x) __builtin_expect ((x) != 0, 0) +#else +#define likely(x) (x) +#define unlikely(x) (x) +#endif diff -ruN a52dec.orig/include/mm_accel.h a52dec/include/mm_accel.h --- a52dec.orig/include/mm_accel.h 2012-11-13 17:25:33.438758556 +0100 +++ a52dec/include/mm_accel.h 2012-11-13 17:27:31.147037566 +0100 @@ -1,6 +1,6 @@ /* * mm_accel.h - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. diff -ruN a52dec.orig/include/tendra.h a52dec/include/tendra.h --- a52dec.orig/include/tendra.h 2012-11-13 17:25:33.438758556 +0100 +++ a52dec/include/tendra.h 2012-11-13 17:27:31.147037566 +0100 @@ -1,6 +1,6 @@ /* * tendra.h - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. diff -ruN a52dec.orig/liba52/a52_internal.h a52dec/liba52/a52_internal.h --- a52dec.orig/liba52/a52_internal.h 2012-11-13 17:25:33.438758556 +0100 +++ a52dec/liba52/a52_internal.h 2012-11-13 17:27:31.147037566 +0100 @@ -1,6 +1,6 @@ /* * a52_internal.h - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. @@ -37,24 +37,24 @@ uint8_t halfrate; /* halfrate factor */ uint8_t acmod; /* coded channels */ uint8_t lfeon; /* coded lfe channel */ - sample_t clev; /* centre channel mix level */ - sample_t slev; /* surround channels mix level */ + level_t clev; /* centre channel mix level */ + level_t slev; /* surround channels mix level */ int output; /* type of output */ - sample_t level; /* output level */ + level_t level; /* output level */ sample_t bias; /* output bias */ int dynrnge; /* apply dynamic range */ - sample_t dynrng; /* dynamic range */ + level_t dynrng; /* dynamic range */ void * dynrngdata; /* dynamic range callback funtion and data */ - sample_t (* dynrngcall) (sample_t range, void * dynrngdata); + level_t (* dynrngcall) (level_t range, void * dynrngdata); uint8_t chincpl; /* channel coupled */ uint8_t phsflginu; /* phase flags in use (stereo only) */ uint8_t cplstrtmant; /* coupling channel start mantissa */ uint8_t cplendmant; /* coupling channel end mantissa */ uint32_t cplbndstrc; /* coupling band structure */ - sample_t cplco[5][18]; /* coupling coordinates */ + level_t cplco[5][18]; /* coupling coordinates */ /* derived information */ uint8_t cplstrtbnd; /* coupling start band (for bit allocation) */ @@ -107,14 +107,79 @@ int start, int end, int fastleak, int slowleak, expbap_t * expbap); -int a52_downmix_init (int input, int flags, sample_t * level, - sample_t clev, sample_t slev); -int a52_downmix_coeff (sample_t * coeff, int acmod, int output, sample_t level, - sample_t clev, sample_t slev); +int a52_downmix_init (int input, int flags, level_t * level, + level_t clev, level_t slev); +int a52_downmix_coeff (level_t * coeff, int acmod, int output, level_t level, + level_t clev, level_t slev); void a52_downmix (sample_t * samples, int acmod, int output, sample_t bias, - sample_t clev, sample_t slev); + level_t clev, level_t slev); void a52_upmix (sample_t * samples, int acmod, int output); void a52_imdct_init (uint32_t mm_accel); void a52_imdct_256 (sample_t * data, sample_t * delay, sample_t bias); void a52_imdct_512 (sample_t * data, sample_t * delay, sample_t bias); + +#define ROUND(x) ((int)((x) + ((x) > 0 ? 0.5 : -0.5))) + +#ifndef LIBA52_FIXED + +typedef sample_t quantizer_t; +#define SAMPLE(x) (x) +#define LEVEL(x) (x) +#define MUL(a,b) ((a) * (b)) +#define MUL_L(a,b) ((a) * (b)) +#define MUL_C(a,b) ((a) * (b)) +#define DIV(a,b) ((a) / (b)) +#define BIAS(x) ((x) + bias) + +#else /* LIBA52_FIXED */ + +typedef int16_t quantizer_t; +#define SAMPLE(x) (sample_t)((x) * (1 << 30)) +#define LEVEL(x) (level_t)((x) * (1 << 26)) + +#if 0 +#define MUL(a,b) ((int)(((int64_t)(a) * (b) + (1 << 29)) >> 30)) +#define MUL_L(a,b) ((int)(((int64_t)(a) * (b) + (1 << 25)) >> 26)) +#elif defined(CPU_COLDFIRE) +/* loses 1 bit of accuracy */ +#define MUL(a, b) \ +({ \ + int32_t t; \ + asm volatile ( \ + "mac.l %[A], %[B], %%acc0\n\t" \ + "movclr.l %%acc0, %[t]\n\t" \ + "asl.l #1, %[t]" \ + : [t] "=d" (t) \ + : [A] "r" ((a)), [B] "r" ((b))); \ + t; \ +}) +/* loses 5 bits of accuracy */ +#define MUL_L(a, b) \ +({ \ + int32_t t; \ + asm volatile ( \ + "mac.l %[A], %[B], %%acc0\n\t" \ + "movclr.l %%acc0, %[t]\n\t" \ + "asl.l #5, %[t]" \ + : [t] "=d" (t) \ + : [A] "r" ((a)), [B] "r" ((b))); \ + t; \ +}) +#elif 1 +#define MUL(a,b) \ +({ int32_t _ta=(a), _tb=(b), _tc; \ + _tc=(_ta & 0xffff)*(_tb >> 16)+(_ta >> 16)*(_tb & 0xffff); (int32_t)(((_tc >> 14))+ (((_ta >> 16)*(_tb >> 16)) << 2 )); }) +#define MUL_L(a,b) \ +({ int32_t _ta=(a), _tb=(b), _tc; \ + _tc=(_ta & 0xffff)*(_tb >> 16)+(_ta >> 16)*(_tb & 0xffff); (int32_t)((_tc >> 10) + (((_ta >> 16)*(_tb >> 16)) << 6)); }) +#else +#define MUL(a,b) (((a) >> 15) * ((b) >> 15)) +#define MUL_L(a,b) (((a) >> 13) * ((b) >> 13)) +#endif + +#define MUL_C(a,b) MUL_L (a, LEVEL (b)) +#define DIV(a,b) ((((int64_t)LEVEL (a)) << 26) / (b)) +#define BIAS(x) ((x) + (bias*0)) + +#endif diff -ruN a52dec.orig/liba52/bit_allocate.c a52dec/liba52/bit_allocate.c --- a52dec.orig/liba52/bit_allocate.c 2012-11-13 17:25:33.438758556 +0100 +++ a52dec/liba52/bit_allocate.c 2012-11-13 17:27:31.147037566 +0100 @@ -1,6 +1,6 @@ /* * bit_allocate.c - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. @@ -21,14 +21,14 @@ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ -#include "config.h" +#include "config-a52.h" #include #include "a52.h" #include "a52_internal.h" -static int hthtab[3][50] = { +static int hthtab[3][50] IDATA_ATTR = { {0x730, 0x730, 0x7c0, 0x800, 0x820, 0x840, 0x850, 0x850, 0x860, 0x860, 0x860, 0x860, 0x860, 0x870, 0x870, 0x870, 0x880, 0x880, 0x890, 0x890, 0x8a0, 0x8a0, 0x8b0, 0x8b0, 0x8c0, 0x8c0, 0x8d0, 0x8e0, 0x8f0, 0x900, @@ -46,7 +46,7 @@ 0x8d0, 0x8b0, 0x840, 0x7f0, 0x790, 0x760, 0x7a0, 0x7c0, 0x7b0, 0x720} }; -static int8_t baptab[305] = { +static int8_t baptab[305] IDATA_ATTR = { 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, @@ -71,11 +71,11 @@ 0, 0, 0, 0 /* 148 padding elems */ }; -static int bndtab[30] = {21, 22, 23, 24, 25, 26, 27, 28, 31, 34, +static int bndtab[30] IDATA_ATTR = {21, 22, 23, 24, 25, 26, 27, 28, 31, 34, 37, 40, 43, 46, 49, 55, 61, 67, 73, 79, 85, 97, 109, 121, 133, 157, 181, 205, 229, 253}; -static int8_t latab[256] = { +static int8_t latab[256] IDATA_ATTR = { -64, -63, -62, -61, -60, -59, -58, -57, -56, -55, -54, -53, -52, -52, -51, -50, -49, -48, -47, -47, -46, -45, -44, -44, -43, -42, -41, -41, -40, -39, -38, -38, -37, -36, -36, -35, @@ -231,7 +231,7 @@ int startband, endband; startband = j; - endband = ((bndtab-20)[i] < end) ? (bndtab-20)[i] : end; + endband = (bndtab[i-20] < end) ? bndtab[i-20] : end; psd = 128 * exp[j++]; while (j < endband) { int next, delta; diff -ruN a52dec.orig/liba52/bitstream.c a52dec/liba52/bitstream.c --- a52dec.orig/liba52/bitstream.c 2012-11-13 17:25:33.442091935 +0100 +++ a52dec/liba52/bitstream.c 2012-11-13 17:27:31.147037566 +0100 @@ -1,6 +1,6 @@ /* * bitstream.c - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. @@ -21,7 +21,7 @@ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ -#include "config.h" +#include "config-a52.h" #include @@ -38,7 +38,9 @@ align = (long)buf & 3; state->buffer_start = (uint32_t *) (buf - align); state->bits_left = 0; + state->current_word = 0; bitstream_get (state, align * 8); + bitstream_get_2 (state, 0); /* pretend function is used - keep gcc happy */ } static inline void bitstream_fill_current (a52_state_t * state) diff -ruN a52dec.orig/liba52/bitstream.h a52dec/liba52/bitstream.h --- a52dec.orig/liba52/bitstream.h 2012-11-13 17:25:33.442091935 +0100 +++ a52dec/liba52/bitstream.h 2012-11-13 17:27:31.147037566 +0100 @@ -1,6 +1,6 @@ /* * bitstream.h - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. @@ -21,29 +21,29 @@ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ -/* (stolen from the kernel) */ +//#define swab32(x) (betoh32(x)) #ifdef WORDS_BIGENDIAN -# define swab32(x) (x) +# define swab32(x) (x) #else -# if 0 && defined (__i386__) +# if defined (__i386__) -# define swab32(x) __i386_swab32(x) - static inline const uint32_t __i386_swab32(uint32_t x) - { - __asm__("bswap %0" : "=r" (x) : "0" (x)); - return x; - } +# define swab32(x) __i386_swab32(x) + static inline const uint32_t __i386_swab32(uint32_t x) + { + __asm__("bswap %0" : "=r" (x) : "0" (x)); + return x; + } -# else +# else -# define swab32(x)\ +# define swab32(x)\ ((((uint8_t*)&x)[0] << 24) | (((uint8_t*)&x)[1] << 16) | \ (((uint8_t*)&x)[2] << 8) | (((uint8_t*)&x)[3])) -# endif +# endif #endif void a52_bitstream_set_ptr (a52_state_t * state, uint8_t * buf); diff -ruN a52dec.orig/liba52/config-a52.h a52dec/liba52/config-a52.h --- a52dec.orig/liba52/config-a52.h 1970-01-01 01:00:00.000000000 +0100 +++ a52dec/liba52/config-a52.h 2012-11-13 17:27:31.150370945 +0100 @@ -0,0 +1,26 @@ +#define IDATA_ATTR +#define IBSS_ATTR +/* a52dec profiling */ +/* #undef A52DEC_GPROF */ + +/* Define to 1 if you have the `memalign' function. */ +/* #undef HAVE_MEMALIGN 1 */ + +/* liba52 djbfft support */ +/* #undef LIBA52_DJBFFT */ + +/* a52 sample precision */ +/* #undef LIBA52_DOUBLE */ + +/* use fixed-point arithmetic */ +#define LIBA52_FIXED + +/* Define to 1 if your processor stores words with the most significant byte + first (like Motorola and SPARC, unlike Intel and VAX). */ + +/* Used in bitstream.h */ + +#ifdef ROCKBOX_BIG_ENDIAN +#define WORDS_BIGENDIAN 1 +#endif + diff -ruN a52dec.orig/liba52/downmix.c a52dec/liba52/downmix.c --- a52dec.orig/liba52/downmix.c 2012-11-13 17:25:33.442091935 +0100 +++ a52dec/liba52/downmix.c 2012-11-13 17:27:31.150370945 +0100 @@ -1,6 +1,6 @@ /* * downmix.c - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. @@ -21,7 +21,7 @@ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ -#include "config.h" +#include "config-a52.h" #include #include @@ -31,8 +31,8 @@ #define CONVERT(acmod,output) (((output) << 3) + (acmod)) -int a52_downmix_init (int input, int flags, sample_t * level, - sample_t clev, sample_t slev) +int a52_downmix_init (int input, int flags, level_t * level, + level_t clev, level_t slev) { static uint8_t table[11][8] = { {A52_CHANNEL, A52_DOLBY, A52_STEREO, A52_STEREO, @@ -66,94 +66,106 @@ output = table[output][input & 7]; - if ((output == A52_STEREO) && - ((input == A52_DOLBY) || ((input == A52_3F) && (clev == LEVEL_3DB)))) + if (output == A52_STEREO && + (input == A52_DOLBY || (input == A52_3F && clev == LEVEL (LEVEL_3DB)))) output = A52_DOLBY; - if (flags & A52_ADJUST_LEVEL) + if (flags & A52_ADJUST_LEVEL) { + level_t adjust; + switch (CONVERT (input & 7, output)) { case CONVERT (A52_3F, A52_MONO): - *level *= LEVEL_3DB / (1 + clev); + adjust = DIV (LEVEL_3DB, LEVEL (1) + clev); break; case CONVERT (A52_STEREO, A52_MONO): case CONVERT (A52_2F2R, A52_2F1R): case CONVERT (A52_3F2R, A52_3F1R): level_3db: - *level *= LEVEL_3DB; + adjust = LEVEL (LEVEL_3DB); break; case CONVERT (A52_3F2R, A52_2F1R): - if (clev < LEVEL_PLUS3DB - 1) + if (clev < LEVEL (LEVEL_PLUS3DB - 1)) goto level_3db; /* break thru */ case CONVERT (A52_3F, A52_STEREO): case CONVERT (A52_3F1R, A52_2F1R): case CONVERT (A52_3F1R, A52_2F2R): case CONVERT (A52_3F2R, A52_2F2R): - *level /= 1 + clev; + adjust = DIV (1, LEVEL (1) + clev); break; case CONVERT (A52_2F1R, A52_MONO): - *level *= LEVEL_PLUS3DB / (2 + slev); + adjust = DIV (LEVEL_PLUS3DB, LEVEL (2) + slev); break; case CONVERT (A52_2F1R, A52_STEREO): case CONVERT (A52_3F1R, A52_3F): - *level /= 1 + slev * LEVEL_3DB; + adjust = DIV (1, LEVEL (1) + MUL_C (slev, LEVEL_3DB)); break; case CONVERT (A52_3F1R, A52_MONO): - *level *= LEVEL_3DB / (1 + clev + 0.5 * slev); + adjust = DIV (LEVEL_3DB, LEVEL (1) + clev + MUL_C (slev, 0.5)); break; case CONVERT (A52_3F1R, A52_STEREO): - *level /= 1 + clev + slev * LEVEL_3DB; + adjust = DIV (1, LEVEL (1) + clev + MUL_C (slev, LEVEL_3DB)); break; case CONVERT (A52_2F2R, A52_MONO): - *level *= LEVEL_3DB / (1 + slev); + adjust = DIV (LEVEL_3DB, LEVEL (1) + slev); break; case CONVERT (A52_2F2R, A52_STEREO): case CONVERT (A52_3F2R, A52_3F): - *level /= 1 + slev; + adjust = DIV (1, LEVEL (1) + slev); break; case CONVERT (A52_3F2R, A52_MONO): - *level *= LEVEL_3DB / (1 + clev + slev); + adjust = DIV (LEVEL_3DB, LEVEL (1) + clev + slev); break; case CONVERT (A52_3F2R, A52_STEREO): - *level /= 1 + clev + slev; + adjust = DIV (1, LEVEL (1) + clev + slev); break; case CONVERT (A52_MONO, A52_DOLBY): - *level *= LEVEL_PLUS3DB; + adjust = LEVEL (LEVEL_PLUS3DB); break; case CONVERT (A52_3F, A52_DOLBY): case CONVERT (A52_2F1R, A52_DOLBY): - *level *= 1 / (1 + LEVEL_3DB); + adjust = LEVEL (1 / (1 + LEVEL_3DB)); break; case CONVERT (A52_3F1R, A52_DOLBY): case CONVERT (A52_2F2R, A52_DOLBY): - *level *= 1 / (1 + 2 * LEVEL_3DB); + adjust = LEVEL (1 / (1 + 2 * LEVEL_3DB)); break; case CONVERT (A52_3F2R, A52_DOLBY): - *level *= 1 / (1 + 3 * LEVEL_3DB); + adjust = LEVEL (1 / (1 + 3 * LEVEL_3DB)); break; + + default: + return output; } + *level = MUL_L (*level, adjust); + } + return output; } -int a52_downmix_coeff (sample_t * coeff, int acmod, int output, sample_t level, - sample_t clev, sample_t slev) +int a52_downmix_coeff (level_t * coeff, int acmod, int output, level_t level, + level_t clev, level_t slev) { + level_t level_3db; + + level_3db = MUL_C (level, LEVEL_3DB); + switch (CONVERT (acmod, output & A52_CHANNEL_MASK)) { case CONVERT (A52_CHANNEL, A52_CHANNEL): @@ -169,120 +181,138 @@ return 0; case CONVERT (A52_CHANNEL, A52_MONO): - coeff[0] = coeff[1] = level * LEVEL_6DB; + coeff[0] = coeff[1] = MUL_C (level, LEVEL_6DB); return 3; case CONVERT (A52_STEREO, A52_MONO): - coeff[0] = coeff[1] = level * LEVEL_3DB; + coeff[0] = coeff[1] = level_3db; return 3; case CONVERT (A52_3F, A52_MONO): - coeff[0] = coeff[2] = level * LEVEL_3DB; - coeff[1] = level * clev * LEVEL_PLUS3DB; + coeff[0] = coeff[2] = level_3db; + coeff[1] = MUL_C (MUL_L (level_3db, clev), LEVEL_PLUS6DB); return 7; case CONVERT (A52_2F1R, A52_MONO): - coeff[0] = coeff[1] = level * LEVEL_3DB; - coeff[2] = level * slev * LEVEL_3DB; + coeff[0] = coeff[1] = level_3db; + coeff[2] = MUL_L (level_3db, slev); return 7; case CONVERT (A52_2F2R, A52_MONO): - coeff[0] = coeff[1] = level * LEVEL_3DB; - coeff[2] = coeff[3] = level * slev * LEVEL_3DB; + coeff[0] = coeff[1] = level_3db; + coeff[2] = coeff[3] = MUL_L (level_3db, slev); return 15; case CONVERT (A52_3F1R, A52_MONO): - coeff[0] = coeff[2] = level * LEVEL_3DB; - coeff[1] = level * clev * LEVEL_PLUS3DB; - coeff[3] = level * slev * LEVEL_3DB; + coeff[0] = coeff[2] = level_3db; + coeff[1] = MUL_C (MUL_L (level_3db, clev), LEVEL_PLUS6DB); + coeff[3] = MUL_L (level_3db, slev); return 15; case CONVERT (A52_3F2R, A52_MONO): - coeff[0] = coeff[2] = level * LEVEL_3DB; - coeff[1] = level * clev * LEVEL_PLUS3DB; - coeff[3] = coeff[4] = level * slev * LEVEL_3DB; + coeff[0] = coeff[2] = level_3db; + coeff[1] = MUL_C (MUL_L (level_3db, clev), LEVEL_PLUS6DB); + coeff[3] = coeff[4] = MUL_L (level_3db, slev); return 31; case CONVERT (A52_MONO, A52_DOLBY): - coeff[0] = level * LEVEL_3DB; + coeff[0] = level_3db; return 0; case CONVERT (A52_3F, A52_DOLBY): - clev = LEVEL_3DB; + coeff[0] = coeff[2] = coeff[3] = coeff[4] = level; + coeff[1] = level_3db; + return 7; + case CONVERT (A52_3F, A52_STEREO): case CONVERT (A52_3F1R, A52_2F1R): case CONVERT (A52_3F2R, A52_2F2R): coeff[0] = coeff[2] = coeff[3] = coeff[4] = level; - coeff[1] = level * clev; + coeff[1] = MUL_L (level, clev); return 7; case CONVERT (A52_2F1R, A52_DOLBY): - slev = 1; + coeff[0] = coeff[1] = level; + coeff[2] = level_3db; + return 7; + case CONVERT (A52_2F1R, A52_STEREO): coeff[0] = coeff[1] = level; - coeff[2] = level * slev * LEVEL_3DB; + coeff[2] = MUL_L (level_3db, slev); return 7; case CONVERT (A52_3F1R, A52_DOLBY): - clev = LEVEL_3DB; - slev = 1; + coeff[0] = coeff[2] = level; + coeff[1] = coeff[3] = level_3db; + return 15; + case CONVERT (A52_3F1R, A52_STEREO): coeff[0] = coeff[2] = level; - coeff[1] = level * clev; - coeff[3] = level * slev * LEVEL_3DB; + coeff[1] = MUL_L (level, clev); + coeff[3] = MUL_L (level_3db, slev); return 15; case CONVERT (A52_2F2R, A52_DOLBY): - slev = LEVEL_3DB; + coeff[0] = coeff[1] = level; + coeff[2] = coeff[3] = level_3db; + return 15; + case CONVERT (A52_2F2R, A52_STEREO): coeff[0] = coeff[1] = level; - coeff[2] = coeff[3] = level * slev; + coeff[2] = coeff[3] = MUL_L (level, slev); return 15; case CONVERT (A52_3F2R, A52_DOLBY): - clev = LEVEL_3DB; + coeff[0] = coeff[2] = level; + coeff[1] = coeff[3] = coeff[4] = level_3db; + return 31; + case CONVERT (A52_3F2R, A52_2F1R): - slev = LEVEL_3DB; + coeff[0] = coeff[2] = level; + coeff[1] = MUL_L (level, clev); + coeff[3] = coeff[4] = level_3db; + return 31; + case CONVERT (A52_3F2R, A52_STEREO): coeff[0] = coeff[2] = level; - coeff[1] = level * clev; - coeff[3] = coeff[4] = level * slev; + coeff[1] = MUL_L (level, clev); + coeff[3] = coeff[4] = MUL_L (level, slev); return 31; case CONVERT (A52_3F1R, A52_3F): coeff[0] = coeff[1] = coeff[2] = level; - coeff[3] = level * slev * LEVEL_3DB; + coeff[3] = MUL_L (level_3db, slev); return 13; case CONVERT (A52_3F2R, A52_3F): coeff[0] = coeff[1] = coeff[2] = level; - coeff[3] = coeff[4] = level * slev; + coeff[3] = coeff[4] = MUL_L (level, slev); return 29; case CONVERT (A52_2F2R, A52_2F1R): coeff[0] = coeff[1] = level; - coeff[2] = coeff[3] = level * LEVEL_3DB; + coeff[2] = coeff[3] = level_3db; return 12; case CONVERT (A52_3F2R, A52_3F1R): coeff[0] = coeff[1] = coeff[2] = level; - coeff[3] = coeff[4] = level * LEVEL_3DB; + coeff[3] = coeff[4] = level_3db; return 24; case CONVERT (A52_2F1R, A52_2F2R): coeff[0] = coeff[1] = level; - coeff[2] = level * LEVEL_3DB; + coeff[2] = level_3db; return 0; case CONVERT (A52_3F1R, A52_2F2R): coeff[0] = coeff[2] = level; - coeff[1] = level * clev; - coeff[3] = level * LEVEL_3DB; + coeff[1] = MUL_L (level, clev); + coeff[3] = level_3db; return 7; case CONVERT (A52_3F1R, A52_3F2R): coeff[0] = coeff[1] = coeff[2] = level; - coeff[3] = level * LEVEL_3DB; + coeff[3] = level_3db; return 0; case CONVERT (A52_CHANNEL, A52_CHANNEL1): @@ -304,7 +334,7 @@ int i; for (i = 0; i < 256; i++) - dest[i] += src[i] + bias; + dest[i] += BIAS (src[i]); } static void mix3to1 (sample_t * samples, sample_t bias) @@ -312,7 +342,7 @@ int i; for (i = 0; i < 256; i++) - samples[i] += samples[i + 256] + samples[i + 512] + bias; + samples[i] += BIAS (samples[i + 256] + samples[i + 512]); } static void mix4to1 (sample_t * samples, sample_t bias) @@ -320,8 +350,8 @@ int i; for (i = 0; i < 256; i++) - samples[i] += (samples[i + 256] + samples[i + 512] + - samples[i + 768] + bias); + samples[i] += BIAS (samples[i + 256] + samples[i + 512] + + samples[i + 768]); } static void mix5to1 (sample_t * samples, sample_t bias) @@ -329,8 +359,8 @@ int i; for (i = 0; i < 256; i++) - samples[i] += (samples[i + 256] + samples[i + 512] + - samples[i + 768] + samples[i + 1024] + bias); + samples[i] += BIAS (samples[i + 256] + samples[i + 512] + + samples[i + 768] + samples[i + 1024]); } static void mix3to2 (sample_t * samples, sample_t bias) @@ -339,7 +369,7 @@ sample_t common; for (i = 0; i < 256; i++) { - common = samples[i + 256] + bias; + common = BIAS (samples[i + 256]); samples[i] += common; samples[i + 256] = samples[i + 512] + common; } @@ -351,7 +381,7 @@ sample_t common; for (i = 0; i < 256; i++) { - common = right[i + 256] + bias; + common = BIAS (right[i + 256]); left[i] += common; right[i] += common; } @@ -364,8 +394,8 @@ for (i = 0; i < 256; i++) { surround = samples[i + 512]; - samples[i] += bias - surround; - samples[i + 256] += bias + surround; + samples[i] += BIAS (-surround); + samples[i + 256] += BIAS (surround); } } @@ -375,7 +405,7 @@ sample_t common; for (i = 0; i < 256; i++) { - common = samples[i + 256] + samples[i + 768] + bias; + common = BIAS (samples[i + 256] + samples[i + 768]); samples[i] += common; samples[i + 256] = samples[i + 512] + common; } @@ -387,7 +417,7 @@ sample_t common, surround; for (i = 0; i < 256; i++) { - common = samples[i + 256] + bias; + common = BIAS (samples[i + 256]); surround = samples[i + 768]; samples[i] += common - surround; samples[i + 256] = samples[i + 512] + common + surround; @@ -401,8 +431,8 @@ for (i = 0; i < 256; i++) { surround = samples[i + 512] + samples[i + 768]; - samples[i] += bias - surround; - samples[i + 256] += bias + surround; + samples[i] += BIAS (-surround); + samples[i + 256] += BIAS (surround); } } @@ -412,7 +442,7 @@ sample_t common; for (i = 0; i < 256; i++) { - common = samples[i + 256] + bias; + common = BIAS (samples[i + 256]); samples[i] += common + samples[i + 768]; samples[i + 256] = common + samples[i + 512] + samples[i + 1024]; } @@ -424,7 +454,7 @@ sample_t common, surround; for (i = 0; i < 256; i++) { - common = samples[i + 256] + bias; + common = BIAS (samples[i + 256]); surround = samples[i + 768] + samples[i + 1024]; samples[i] += common - surround; samples[i + 256] = samples[i + 512] + common + surround; @@ -436,7 +466,7 @@ int i; for (i = 0; i < 256; i++) - dest[i] = src[i] + src[i + 256] + bias; + dest[i] = BIAS (src[i] + src[i + 256]); } static void zero (sample_t * samples) @@ -448,8 +478,11 @@ } void a52_downmix (sample_t * samples, int acmod, int output, sample_t bias, - sample_t clev, sample_t slev) + level_t clev, level_t slev) { + /* avoid compiler warning */ + (void)clev; + switch (CONVERT (acmod, output & A52_CHANNEL_MASK)) { case CONVERT (A52_CHANNEL, A52_CHANNEL2): @@ -583,7 +616,7 @@ break; case CONVERT (A52_3F1R, A52_3F2R): - memcpy (samples + 1027, samples + 768, 256 * sizeof (sample_t)); + memcpy (samples + 1024, samples + 768, 256 * sizeof (sample_t)); break; } } diff -ruN a52dec.orig/liba52/imdct.c a52dec/liba52/imdct.c --- a52dec.orig/liba52/imdct.c 2012-11-13 17:25:33.438758556 +0100 +++ a52dec/liba52/imdct.c 2012-11-13 17:27:31.150370945 +0100 @@ -1,6 +1,6 @@ /* * imdct.c - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * The ifft algorithms in this file have been largely inspired by Dan @@ -24,12 +24,13 @@ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ -#include "config.h" +#include "config-a52.h" #include #include #ifdef LIBA52_DJBFFT #include +#include #endif #ifndef M_PI #define M_PI 3.1415926535897932384626433832795029 @@ -57,25 +58,26 @@ }; /* Root values for IFFT */ -static sample_t roots16[3]; -static sample_t roots32[7]; -static sample_t roots64[15]; -static sample_t roots128[31]; +//static sample_t roots16[3]; +//static sample_t roots32[7]; +//static sample_t roots64[15]; +//static sample_t roots128[31]; /* Twiddle factors for IMDCT */ -static complex_t pre1[128]; -static complex_t post1[64]; -static complex_t pre2[64]; -static complex_t post2[32]; +//static complex_t pre1[128]; +//static complex_t post1[64]; +//static complex_t pre2[64]; +//static complex_t post2[32]; -static sample_t a52_imdct_window[256]; +//static sample_t a52_imdct_window[256]; +#include "imdct_lookups.h" static void (* ifft128) (complex_t * buf); static void (* ifft64) (complex_t * buf); static inline void ifft2 (complex_t * buf) { - double r, i; + sample_t r, i; r = buf[0].real; i = buf[0].imag; @@ -87,7 +89,7 @@ static inline void ifft4 (complex_t * buf) { - double tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8; + sample_t tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8; tmp1 = buf[0].real + buf[1].real; tmp2 = buf[3].real + buf[2].real; @@ -108,25 +110,37 @@ buf[3].imag = tmp6 - tmp8; } +/* basic radix-2 ifft butterfly */ + +#define BUTTERFLY_0(t0,t1,W0,W1,d0,d1) do { \ + t0 = MUL (W1, d1) + MUL (W0, d0); \ + t1 = MUL (W0, d1) - MUL (W1, d0); \ +} while (0) + +/* radix-2 ifft butterfly with bias */ + +#define BUTTERFLY_B(t0,t1,W0,W1,d0,d1) do { \ + t0 = BIAS (MUL (d1, W1) + MUL (d0, W0)); \ + t1 = BIAS (MUL (d1, W0) - MUL (d0, W1)); \ +} while (0) + /* the basic split-radix ifft butterfly */ -#define BUTTERFLY(a0,a1,a2,a3,wr,wi) do { \ - tmp5 = a2.real * wr + a2.imag * wi; \ - tmp6 = a2.imag * wr - a2.real * wi; \ - tmp7 = a3.real * wr - a3.imag * wi; \ - tmp8 = a3.imag * wr + a3.real * wi; \ - tmp1 = tmp5 + tmp7; \ - tmp2 = tmp6 + tmp8; \ - tmp3 = tmp6 - tmp8; \ - tmp4 = tmp7 - tmp5; \ - a2.real = a0.real - tmp1; \ - a2.imag = a0.imag - tmp2; \ - a3.real = a1.real - tmp3; \ - a3.imag = a1.imag - tmp4; \ - a0.real += tmp1; \ - a0.imag += tmp2; \ - a1.real += tmp3; \ - a1.imag += tmp4; \ +#define BUTTERFLY(a0,a1,a2,a3,wr,wi) do { \ + BUTTERFLY_0 (tmp5, tmp6, wr, wi, a2.real, a2.imag); \ + BUTTERFLY_0 (tmp8, tmp7, wr, wi, a3.imag, a3.real); \ + tmp1 = tmp5 + tmp7; \ + tmp2 = tmp6 + tmp8; \ + tmp3 = tmp6 - tmp8; \ + tmp4 = tmp7 - tmp5; \ + a2.real = a0.real - tmp1; \ + a2.imag = a0.imag - tmp2; \ + a3.real = a1.real - tmp3; \ + a3.imag = a1.imag - tmp4; \ + a0.real += tmp1; \ + a0.imag += tmp2; \ + a1.real += tmp3; \ + a1.imag += tmp4; \ } while (0) /* split-radix ifft butterfly, specialized for wr=1 wi=0 */ @@ -149,10 +163,10 @@ /* split-radix ifft butterfly, specialized for wr=wi */ #define BUTTERFLY_HALF(a0,a1,a2,a3,w) do { \ - tmp5 = (a2.real + a2.imag) * w; \ - tmp6 = (a2.imag - a2.real) * w; \ - tmp7 = (a3.real - a3.imag) * w; \ - tmp8 = (a3.imag + a3.real) * w; \ + tmp5 = MUL (a2.real + a2.imag, w); \ + tmp6 = MUL (a2.imag - a2.real, w); \ + tmp7 = MUL (a3.real - a3.imag, w); \ + tmp8 = MUL (a3.imag + a3.real, w); \ tmp1 = tmp5 + tmp7; \ tmp2 = tmp6 + tmp8; \ tmp3 = tmp6 - tmp8; \ @@ -169,7 +183,7 @@ static inline void ifft8 (complex_t * buf) { - double tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8; + sample_t tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8; ifft4 (buf); ifft2 (buf + 4); @@ -183,7 +197,7 @@ complex_t * buf1; complex_t * buf2; complex_t * buf3; - double tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8; + sample_t tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8; int i; buf++; @@ -196,7 +210,8 @@ i = n - 1; do { - BUTTERFLY (buf[0], buf1[0], buf2[0], buf3[0], weight[n], weight[2*i]); + BUTTERFLY (buf[0], buf1[0], buf2[0], buf3[0], + weight[0], weight[2*i-n]); buf++; buf1++; buf2++; @@ -210,7 +225,7 @@ ifft8 (buf); ifft4 (buf + 8); ifft4 (buf + 12); - ifft_pass (buf, roots16 - 4, 4); + ifft_pass (buf, roots16, 4); } static void ifft32 (complex_t * buf) @@ -218,7 +233,7 @@ ifft16 (buf); ifft8 (buf + 16); ifft8 (buf + 24); - ifft_pass (buf, roots32 - 8, 8); + ifft_pass (buf, roots32, 8); } static void ifft64_c (complex_t * buf) @@ -226,7 +241,7 @@ ifft32 (buf); ifft16 (buf + 32); ifft16 (buf + 48); - ifft_pass (buf, roots64 - 16, 16); + ifft_pass (buf, roots64, 16); } static void ifft128_c (complex_t * buf) @@ -234,11 +249,11 @@ ifft32 (buf); ifft16 (buf + 32); ifft16 (buf + 48); - ifft_pass (buf, roots64 - 16, 16); + ifft_pass (buf, roots64, 16); ifft32 (buf + 64); ifft32 (buf + 96); - ifft_pass (buf, roots128 - 32, 32); + ifft_pass (buf, roots128, 32); } void a52_imdct_512 (sample_t * data, sample_t * delay, sample_t bias) @@ -252,9 +267,7 @@ k = fftorder[i]; t_r = pre1[i].real; t_i = pre1[i].imag; - - buf[i].real = t_i * data[255-k] + t_r * data[k]; - buf[i].imag = t_r * data[255-k] - t_i * data[k]; + BUTTERFLY_0 (buf[i].real, buf[i].imag, t_r, t_i, data[k], data[255-k]); } ifft128 (buf); @@ -265,27 +278,22 @@ /* y[n] = z[n] * (xcos1[n] + j * xsin1[n]) ; */ t_r = post1[i].real; t_i = post1[i].imag; - - a_r = t_r * buf[i].real + t_i * buf[i].imag; - a_i = t_i * buf[i].real - t_r * buf[i].imag; - b_r = t_i * buf[127-i].real + t_r * buf[127-i].imag; - b_i = t_r * buf[127-i].real - t_i * buf[127-i].imag; + BUTTERFLY_0 (a_r, a_i, t_i, t_r, buf[i].imag, buf[i].real); + BUTTERFLY_0 (b_r, b_i, t_r, t_i, buf[127-i].imag, buf[127-i].real); w_1 = window[2*i]; w_2 = window[255-2*i]; - data[2*i] = delay[2*i] * w_2 - a_r * w_1 + bias; - data[255-2*i] = delay[2*i] * w_1 + a_r * w_2 + bias; + BUTTERFLY_B (data[255-2*i], data[2*i], w_2, w_1, a_r, delay[2*i]); delay[2*i] = a_i; w_1 = window[2*i+1]; w_2 = window[254-2*i]; - data[2*i+1] = delay[2*i+1] * w_2 + b_r * w_1 + bias; - data[254-2*i] = delay[2*i+1] * w_1 - b_r * w_2 + bias; + BUTTERFLY_B (data[2*i+1], data[254-2*i], w_1, w_2, b_r, delay[2*i+1]); delay[2*i+1] = b_i; } } -void a52_imdct_256(sample_t * data, sample_t * delay, sample_t bias) +void a52_imdct_256 (sample_t * data, sample_t * delay, sample_t bias) { int i, k; sample_t t_r, t_i, a_r, a_i, b_r, b_i, c_r, c_i, d_r, d_i, w_1, w_2; @@ -297,12 +305,8 @@ k = fftorder[i]; t_r = pre2[i].real; t_i = pre2[i].imag; - - buf1[i].real = t_i * data[254-k] + t_r * data[k]; - buf1[i].imag = t_r * data[254-k] - t_i * data[k]; - - buf2[i].real = t_i * data[255-k] + t_r * data[k+1]; - buf2[i].imag = t_r * data[255-k] - t_i * data[k+1]; + BUTTERFLY_0 (buf1[i].real, buf1[i].imag, t_r, t_i, data[k], data[254-k]); + BUTTERFLY_0 (buf2[i].real, buf2[i].imag, t_r, t_i, data[k+1], data[255-k]); } ifft64 (buf1); @@ -314,43 +318,34 @@ /* y1[n] = z1[n] * (xcos2[n] + j * xs in2[n]) ; */ t_r = post2[i].real; t_i = post2[i].imag; - - a_r = t_r * buf1[i].real + t_i * buf1[i].imag; - a_i = t_i * buf1[i].real - t_r * buf1[i].imag; - b_r = t_i * buf1[63-i].real + t_r * buf1[63-i].imag; - b_i = t_r * buf1[63-i].real - t_i * buf1[63-i].imag; - - c_r = t_r * buf2[i].real + t_i * buf2[i].imag; - c_i = t_i * buf2[i].real - t_r * buf2[i].imag; - d_r = t_i * buf2[63-i].real + t_r * buf2[63-i].imag; - d_i = t_r * buf2[63-i].real - t_i * buf2[63-i].imag; + BUTTERFLY_0 (a_r, a_i, t_i, t_r, buf1[i].imag, buf1[i].real); + BUTTERFLY_0 (b_r, b_i, t_r, t_i, buf1[63-i].imag, buf1[63-i].real); + BUTTERFLY_0 (c_r, c_i, t_i, t_r, buf2[i].imag, buf2[i].real); + BUTTERFLY_0 (d_r, d_i, t_r, t_i, buf2[63-i].imag, buf2[63-i].real); w_1 = window[2*i]; w_2 = window[255-2*i]; - data[2*i] = delay[2*i] * w_2 - a_r * w_1 + bias; - data[255-2*i] = delay[2*i] * w_1 + a_r * w_2 + bias; + BUTTERFLY_B (data[255-2*i], data[2*i], w_2, w_1, a_r, delay[2*i]); delay[2*i] = c_i; w_1 = window[128+2*i]; w_2 = window[127-2*i]; - data[128+2*i] = delay[127-2*i] * w_2 + a_i * w_1 + bias; - data[127-2*i] = delay[127-2*i] * w_1 - a_i * w_2 + bias; + BUTTERFLY_B (data[128+2*i], data[127-2*i], w_1, w_2, a_i, delay[127-2*i]); delay[127-2*i] = c_r; w_1 = window[2*i+1]; w_2 = window[254-2*i]; - data[2*i+1] = delay[2*i+1] * w_2 - b_i * w_1 + bias; - data[254-2*i] = delay[2*i+1] * w_1 + b_i * w_2 + bias; + BUTTERFLY_B (data[254-2*i], data[2*i+1], w_2, w_1, b_i, delay[2*i+1]); delay[2*i+1] = d_r; w_1 = window[129+2*i]; w_2 = window[126-2*i]; - data[129+2*i] = delay[126-2*i] * w_2 + b_r * w_1 + bias; - data[126-2*i] = delay[126-2*i] * w_1 - b_r * w_2 + bias; + BUTTERFLY_B (data[129+2*i], data[126-2*i], w_1, w_2, b_r, delay[126-2*i]); delay[126-2*i] = d_i; } } +/* static double besselI0 (double x) { double bessel = 1; @@ -361,66 +356,118 @@ while (--i); return bessel; } +*/ void a52_imdct_init (uint32_t mm_accel) { - int i, k; + (void)mm_accel; +/* int i, k; double sum; + double local_imdct_window[256];*/ /* compute imdct window - kaiser-bessel derived window, alpha = 5.0 */ - sum = 0; + /* sum = 0; for (i = 0; i < 256; i++) { sum += besselI0 (i * (256 - i) * (5 * M_PI / 256) * (5 * M_PI / 256)); - a52_imdct_window[i] = sum; + local_imdct_window[i] = sum; } sum++; - for (i = 0; i < 256; i++) - a52_imdct_window[i] = sqrt (a52_imdct_window[i] / sum); + */ + /* for (i = 0; i < 256; i++) + a52_imdct_window[i] = SAMPLE (sqrt (local_imdct_window[i] / sum)); + + printf("static sample_t a52_imdct_window[256]={"); + for (i=0;i<256;i++) { + if ((i % 16)==0) { printf("\n"); } + printf("%d,",a52_imdct_window[i]); + } + printf("\n}\n"); + */ + + /* for (i = 0; i < 3; i++) + roots16[i] = SAMPLE (cos ((M_PI / 8) * (i + 1))); - for (i = 0; i < 3; i++) - roots16[i] = cos ((M_PI / 8) * (i + 1)); + printf("static sample_t roots16[3]={%d,%d,%d};\n\n",roots16[0],roots16[1],roots16[2]); for (i = 0; i < 7; i++) - roots32[i] = cos ((M_PI / 16) * (i + 1)); + roots32[i] = SAMPLE (cos ((M_PI / 16) * (i + 1))); + + printf("static sample_t roots32[7]={"); + for (i=0;i<7;i++) { printf("%d%s",roots32[i],(i < 6 ? "," : "")); } + printf("};\n"); for (i = 0; i < 15; i++) - roots64[i] = cos ((M_PI / 32) * (i + 1)); + roots64[i] = SAMPLE (cos ((M_PI / 32) * (i + 1))); + + printf("static sample_t roots64[15]={"); + for (i=0;i<15;i++) { printf("%d%s",roots64[i],(i < 14 ? "," : "")); } + printf("};\n"); for (i = 0; i < 31; i++) - roots128[i] = cos ((M_PI / 64) * (i + 1)); + roots128[i] = SAMPLE (cos ((M_PI / 64) * (i + 1))); + printf("static sample_t roots128[31]={"); + for (i=0;i<31;i++) { printf("%d%s",roots128[i],(i < 30 ? "," : "")); } + printf("};\n"); + */ + /* for (i = 0; i < 64; i++) { k = fftorder[i] / 2 + 64; - pre1[i].real = cos ((M_PI / 256) * (k - 0.25)); - pre1[i].imag = sin ((M_PI / 256) * (k - 0.25)); + pre1[i].real = SAMPLE (cos ((M_PI / 256) * (k - 0.25))); + pre1[i].imag = SAMPLE (sin ((M_PI / 256) * (k - 0.25))); } for (i = 64; i < 128; i++) { k = fftorder[i] / 2 + 64; - pre1[i].real = -cos ((M_PI / 256) * (k - 0.25)); - pre1[i].imag = -sin ((M_PI / 256) * (k - 0.25)); + pre1[i].real = SAMPLE (-cos ((M_PI / 256) * (k - 0.25))); + pre1[i].imag = SAMPLE (-sin ((M_PI / 256) * (k - 0.25))); } + printf("static complex_t pre1[128]={"); + for (i=0;i<128;i++) { printf("{%d,%d}%s",pre1[i].real,pre1[i].imag,(i < 127 ? "," : "")); } + printf("};\n"); + */ + /* for (i = 0; i < 64; i++) { - post1[i].real = cos ((M_PI / 256) * (i + 0.5)); - post1[i].imag = sin ((M_PI / 256) * (i + 0.5)); + post1[i].real = SAMPLE (cos ((M_PI / 256) * (i + 0.5))); + post1[i].imag = SAMPLE (sin ((M_PI / 256) * (i + 0.5))); } + printf("static complex_t post1[64]={"); + for (i=0;i<64;i++) { printf("{%d,%d}%s",post1[i].real,post1[i].imag,(i < 63 ? "," : "")); } + printf("};\n"); + */ + + /* for (i = 0; i < 64; i++) { k = fftorder[i] / 4; - pre2[i].real = cos ((M_PI / 128) * (k - 0.25)); - pre2[i].imag = sin ((M_PI / 128) * (k - 0.25)); + pre2[i].real = SAMPLE (cos ((M_PI / 128) * (k - 0.25))); + pre2[i].imag = SAMPLE (sin ((M_PI / 128) * (k - 0.25))); } + printf("static complex_t pre2[64]={"); + for (i=0;i<64;i++) { printf("{%d,%d}%s",pre2[i].real,pre2[i].imag,(i < 63 ? "," : "")); } + printf("};\n"); + for (i = 0; i < 32; i++) { - post2[i].real = cos ((M_PI / 128) * (i + 0.5)); - post2[i].imag = sin ((M_PI / 128) * (i + 0.5)); + post2[i].real = SAMPLE (cos ((M_PI / 128) * (i + 0.5))); + post2[i].imag = SAMPLE (sin ((M_PI / 128) * (i + 0.5))); } + printf("static complex_t post2[32]={"); + for (i=0;i<32;i++) { printf("{%d,%d}%s",post2[i].real,post2[i].imag,(i < 31 ? "," : "")); } + printf("};\n"); + */ + #ifdef LIBA52_DJBFFT if (mm_accel & MM_ACCEL_DJBFFT) { +#ifndef LIBA52_DOUBLE ifft128 = (void (*) (complex_t *)) fftc4_un128; ifft64 = (void (*) (complex_t *)) fftc4_un64; +#else + ifft128 = (void (*) (complex_t *)) fftc8_un128; + ifft64 = (void (*) (complex_t *)) fftc8_un64; +#endif } else #endif { diff -ruN a52dec.orig/liba52/imdct_lookups.h a52dec/liba52/imdct_lookups.h --- a52dec.orig/liba52/imdct_lookups.h 1970-01-01 01:00:00.000000000 +0100 +++ a52dec/liba52/imdct_lookups.h 2012-11-13 17:27:31.150370945 +0100 @@ -0,0 +1,15 @@ +static sample_t a52_imdct_window[256]IDATA_ATTR={ + 146020,261886,393529,545197,719447,918478,1144416,1399394,1685589,2005234,2360623,2754115,3188134,3665170,4187773,4758556, 5380193,6055411,6786995,7577779,8430645,9348521,10334375,11391212,12522071,13730020,15018150,16389576,17847424,19394833,21034947,22770912, 24605865,26542938,28585242,30735872,32997891,35374332,37868188,40482408,43219889,46083473,49075937,52199993,55458273,58853331,62387636,66063559, 69883377,73849259,77963266,82227341,86643307,91212859,95937560,100818835,105857968,111056092,116414194,121933098,127613474,133455822,139460477,145627601, 151957182,158449029,165102772,171917855,178893540,186028900,193322822,200774000,208380940,216141958,224055176,232118527,240329753,248686407,257185854,265825270, 274601649,283511802,292552357,301719768,311010314,320420105,329945084,339581031,349323572,359168178,369110174,379144743,389266934,399471665,409753732,420107815, 430528483,441010205,451547355,462134219,472765003,483433845,494134818,504861939,515609181,526370480,537139740,547910849,558677680,569434108,580174011,590891284, 601579849,612233658,622846709,633413050,643926788,654382103,664773249,675094567,685340494,695505569,705584441,715571877,725462772,735252152,744935184,754507184, 763963620,773300119,782512477,791596659,800548807,809365245,818042484,826577226,834966364,843206992,851296404,859232096,867011771,874633340,882094922,889394844, 896531647,903504079,910311101,916951881,923425798,929732436,935871584,941843233,947647575,953284997,958756080,964061593,969202490,974179906,978995149,983649698, 988145195,992483442,996666390,1000696136,1004574919,1008305104,1011889185,1015329772,1018629583,1021791439,1024818257,1027713038,1030478862,1033118881,1035636308,1038034411, 1040316504,1042485942,1044546109,1046500412,1048352275,1050105129,1051762405,1053327531,1054803917,1056194958,1057504020,1058734435,1059889501,1060972468,1061986539,1062934861, 1063820523,1064646551,1065415903,1066131467,1066796055,1067412403,1067983168,1068510924,1068998160,1069447282,1069860607,1070240366,1070588702,1070907668,1071199230,1071465266, 1071707567,1071927836,1072127692,1072308670,1072472221,1072619716,1072752449,1072871635,1072978415,1073073858,1073158963,1073234663,1073301826,1073361257,1073413702,1073459852, 1073500344,1073535763,1073566646,1073593486,1073616731,1073636791,1073654036,1073668804,1073681398,1073692090,1073701126,1073708726,1073715084,1073720373,1073724748,1073728344, 1073731279,1073733657,1073735568,1073737090,1073738291,1073739229,1073739951,1073740500,1073740912,1073741214,1073741431,1073741582,1073741685,1073741751,1073741792,1073741814 +}; + +static sample_t roots16[3]IDATA_ATTR={992008094,759250124,410903206}; +static sample_t roots32[7]IDATA_ATTR={1053110175,992008094,892783698,759250124,596538995,410903206,209476638}; +static sample_t roots64[15]IDATA_ATTR={1068571463,1053110175,1027506861,992008094,946955747,892783698,830013654,759250124,681174602,596538995,506158392,410903206,311690798,209476638,105245103}; +static sample_t roots128[31]IDATA_ATTR={1072448454,1068571463,1062120190,1053110175,1041563127,1027506861,1010975241,992008094,970651112,946955747,920979082,892783698,862437519,830013654,795590212,759250124,721080937,681174602,639627257,596538995,552013618,506158392,459083785,410903206,361732725,311690798,260897981,209476638,157550647,105245103,52686014}; + +static complex_t pre1[128]IDATA_ATTR={{761575898,756917205},{3294193,1073736770},{413944710,990742792},{-407857834,993264059},{599275209,890949340},{-206244755,1053747885},{-593797166,894609652},{212706548,1052462554},{683717842,827919933},{-101966276,1068889322},{314841678,1026545772},{-503250790,948504162},{-678624950,832099562},{108522938,1068243547},{509061229,945398418},{-308536985,1028458279},{723518379,793374223},{-49395540,1072605046},{364832651,1009860703},{-456103709,972054993},{554836544,919281193},{-257701283,1042358649},{-636978326,864395809},{160808444,1061631832},{-718636707,797798713},{55975991,1072281769},{462059540,969238095},{-358629394,1012080264},{642270168,860471112},{-154291366,1062598550},{-549185496,922668301},{264092224,1040757801},{742770847,775379244},{-23057618,1073494224},{389505993,1000603111},{-432110916,982955574},{577229727,905387953},{-232042906,1048369016},{-615573145,879767700},{186813761,1057365652},{663193747,844449855},{-128167423,1066065014},{289554159,1033963197},{-526376678,935868098},{-698841306,815194659},{82274244,1070585098},{485706670,957606670},{-333683689,1020576650},{-737999227,779922204},{29644020,1073332537},{438134083,980285687},{-383359075,1002974238},{620959710,875974053},{-180322371,1058492015},{-571663505,908912724},{238471209,1046925492},{703830091,810891303},{-75703709,1071069770},{339939548,1018509994},{-479821763,960568883},{-657999815,848503239},{134706262,1065258526},{532109148,932620694},{-283204430,1035720404},{-752230014,-766205918},{9882456,-1073696345},{-401755603,-995747929},{420016001,-988184225},{-588296766,-898236282},{219160333,-1051137599},{604730690,-887255485},{-199775197,-1054993542},{-673506508,-836247862},{115075515,-1067557553},{-302220675,-1030332066},{514852501,-942257080},{688784992,-823709134},{-95405775,-1069494853},{-497421404,-951574195},{321134518,-1024594615},{-713727978,-802193167},{62554335,-1071918121},{-352412636,-1014261720},{467997975,-966384705},{-543513771,-926020671},{270473222,-1039117770},{647537829,-856514018},{-147768480,-1063525261},{728372812,-788919863},{-42813229,-1072887940},{-450130706,-974835294},{371022172,-1007603122},{-631662502,-868287963},{167319467,-1060625145},{560466703,-915859475},{-251300639,-1043920252},{747514502,-770807091},{-16470347,-1073615495},{-426071479,-985588453},{395638246,-998194311},{-610163404,-883528225},{193298118,-1056199480},{582774217,-901829094},{-225605866,-1049773069},{-693826210,-819467323},{88841682,-1070060119},{-327415267,-1022604883},{491573291,-954608403},{668362709,-840364678},{-121623758,-1066831367},{-520624390,-939080267},{295892987,-1032167062},{-733199822,-784435800},{36229307,-1073130440},{-377197724,-1005307605},{444140755,-977578893},{-566075760,-912403275},{244890534,-1045442552},{626322896,-872147426},{-173824191,-1059578527},{708792377,-806557418},{-69130323,-1071514117},{-473918791,-963494932},{346182609,-1016404991},{-652781111,-852524677},{141240030,-1064411930},{537821584,-929338177},{-276844037,-1037438616}}; + +static complex_t post1[64]IDATA_ATTR={{1073721611,6588355},{1073559912,19764075},{1073236539,32936819},{1072751541,46104602},{1072104991,59265442},{1071296985,72417357},{1070327646,85558366},{1069197119,98686490},{1067905576,111799753},{1066453209,124896178},{1064840239,137973795},{1063066908,151030634},{1061133483,164064728},{1059040255,177074114},{1056787539,190056834},{1054375675,203010932},{1051805026,215934457},{1049075979,228825463},{1046188946,241682009},{1043144359,254502159},{1039942680,267283981},{1036584388,280025551},{1033069991,292724951},{1029400017,305380267},{1025575020,317989594},{1021595574,330551034},{1017462280,343062693},{1013175760,355522688},{1008736660,367929143},{1004145647,380280189},{999403414,392573967},{994510674,404808624},{989468165,416982318},{984276645,429093217},{978936897,441139495},{973449725,453119340},{967815955,465030947},{962036435,476872521},{956112036,488642280},{950043650,500338452},{943832191,511959274},{937478594,523502998},{930983817,534967883},{924348836,546352205},{917574653,557654248},{910662286,568872310},{903612776,580004702},{896427186,591049747},{889106597,602005783},{881652112,612871159},{874064853,623644238},{866345963,634323399},{858496605,644907034},{850517961,655393547},{842411231,665781361},{834177638,676068911},{825818420,686254647},{817334837,696337035},{808728167,706314558},{799999705,716185713},{791150766,725949012},{782182683,735602987},{773096806,745146182},{763894503,754577161}}; + +static complex_t pre2[64]IDATA_ATTR={{1073721611,-6588355},{763894503,754577161},{994510674,404808624},{416982318,989468165},{1054375675,203010932},{602005783,889106597},{215934457,1051805026},{896427186,591049747},{1069197119,98686490},{686254647,825818420},{950043650,500338452},{317989594,1025575020},{111799753,1067905576},{834177638,676068911},{1029400017,305380267},{511959274,943832191},{1072751541,46104602},{725949012,791150766},{973449725,453119340},{367929143,1008736660},{1043144359,254502159},{557654248,917574653},{164064728,1061133483},{866345963,634323399},{59265442,1072104991},{799999705,716185713},{1013175760,355522688},{465030947,967815955},{1063066908,151030634},{644907034,858496605},{267283981,1039942680},{924348836,546352205},{1073559912,19764075},{745146182,773096806},{984276645,429093217},{392573967,999403414},{1049075979,228825463},{580004702,903612776},{190056834,1056787539},{881652112,612871159},{1066453209,124896178},{665781361,842411231},{937478594,523502998},{292724951,1033069991},{85558366,1070327646},{817334837,696337035},{1021595574,330551034},{488642280,956112036},{32936819,1073236539},{782182683,735602987},{1004145647,380280189},{441139495,978936897},{1059040255,177074114},{623644238,874064853},{241682009,1046188946},{910662286,568872310},{1071296985,72417357},{706314558,808728167},{962036435,476872521},{343062693,1017462280},{137973795,1064840239},{850517961,655393547},{1036584388,280025551},{534967883,930983817}}; +static complex_t post2[32]IDATA_ATTR={{1073660973,13176463},{1073014239,39521454},{1071721163,65842639},{1069782521,92124162},{1067199482,118350193},{1063973603,144504935},{1060106825,170572632},{1055601479,196537583},{1050460278,222384146},{1044686318,248096754},{1038283079,273659918},{1031254417,299058239},{1023604566,324276418},{1015338134,349299266},{1006460100,374111709},{996975812,398698801},{986890983,423045731},{976211688,447137835},{964944359,470960600},{953095785,494499675},{940673100,517740882},{927683790,540670222},{914135677,563273882},{900036924,585538247},{885396022,607449906},{870221790,628995659},{854523369,650162530},{838310215,670937766},{821592095,691308855},{804379078,711263525},{786681534,730789756},{768510121,749875787}}; diff -ruN a52dec.orig/liba52/parse.c a52dec/liba52/parse.c --- a52dec.orig/liba52/parse.c 2012-11-13 17:25:33.438758556 +0100 +++ a52dec/liba52/parse.c 2012-11-13 17:27:31.150370945 +0100 @@ -1,6 +1,6 @@ /* * parse.c - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. @@ -21,9 +21,8 @@ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ -#include "config.h" +#include "config-a52.h" -#include #include #include @@ -32,7 +31,7 @@ #include "bitstream.h" #include "tables.h" -#ifdef HAVE_MEMALIGN +#if defined(HAVE_MEMALIGN) && !defined(__cplusplus) /* some systems have memalign() but no declaration for it */ void * memalign (size_t align, size_t size); #else @@ -41,13 +40,16 @@ #endif typedef struct { - sample_t q1[2]; - sample_t q2[2]; - sample_t q4; + quantizer_t q1[2]; + quantizer_t q2[2]; + quantizer_t q4; int q1_ptr; int q2_ptr; int q4_ptr; -} quantizer_t; +} quantizer_set_t; + +static a52_state_t istate IBSS_ATTR; +static sample_t isamples[256*12] IBSS_ATTR; static uint8_t halfrate[12] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 3}; @@ -56,16 +58,25 @@ a52_state_t * state; int i; - state = malloc (sizeof (a52_state_t)); + #if defined(CPU_COLDFIRE) + coldfire_set_macsr(EMAC_FRACTIONAL | EMAC_ROUND | EMAC_SATURATE); + #endif + /* + this needs to come back if we ever want two decoder instances + simultenously. NOTE, you also need to remove comments in a52_free. + state = (a52_state_t *) malloc (sizeof (a52_state_t)); if (state == NULL) return NULL; - state->samples = memalign (16, 256 * 12 * sizeof (sample_t)); + state->samples = (sample_t *) memalign (16, 256 * 12 * sizeof (sample_t)); if (state->samples == NULL) { free (state); return NULL; } + */ + state = &istate; + state->samples = isamples; for (i = 0; i < 256 * 12; i++) state->samples[i] = 0; @@ -129,10 +140,12 @@ } int a52_frame (a52_state_t * state, uint8_t * buf, int * flags, - sample_t * level, sample_t bias) + level_t * level, sample_t bias) { - static sample_t clev[4] = {LEVEL_3DB, LEVEL_45DB, LEVEL_6DB, LEVEL_45DB}; - static sample_t slev[4] = {LEVEL_3DB, LEVEL_6DB, 0, LEVEL_6DB}; + static level_t clev[4] = { LEVEL (LEVEL_3DB), LEVEL (LEVEL_45DB), + LEVEL (LEVEL_6DB), LEVEL (LEVEL_45DB) }; + static level_t slev[4] = { LEVEL (LEVEL_3DB), LEVEL (LEVEL_6DB), + 0, LEVEL (LEVEL_6DB) }; int chaninfo; int acmod; @@ -146,6 +159,8 @@ if ((acmod == 2) && (bitstream_get (state, 2) == 2)) /* dsurmod */ acmod = A52_DOLBY; + state->clev = state->slev = 0; + if ((acmod & 1) && (acmod != 1)) state->clev = clev[bitstream_get (state, 2)]; /* cmixlev */ @@ -162,7 +177,7 @@ state->output |= A52_LFE; *flags = state->output; /* the 2* compensates for differences in imdct */ - state->dynrng = state->level = 2 * *level; + state->dynrng = state->level = MUL_C (*level, 2); state->bias = bias; state->dynrnge = 1; state->dynrngcall = NULL; @@ -201,7 +216,7 @@ } void a52_dynrng (a52_state_t * state, - sample_t (* call) (sample_t, void *), void * data) + level_t (* call) (level_t, void *), void * data) { state->dynrnge = 0; if (call) { @@ -311,20 +326,38 @@ state->lfsr_state = (uint16_t) nstate; - return nstate; + return (3 * nstate) >> 2; } +#ifndef LIBA52_FIXED +#define COEFF(c,t,l,s,e) (c) = (t) * (s)[e] +#else +#define COEFF(c,_t,_l,s,e) do { \ + quantizer_t t = (_t); \ + level_t l = (_l); \ + int shift = e - 5; \ + sample_t tmp = t * (l >> 16) + ((t * (l & 0xffff)) >> 16); \ + if (shift >= 0) \ + (c) = tmp >> shift; \ + else \ + (c) = tmp << -shift; \ +} while (0) +#endif + static void coeff_get (a52_state_t * state, sample_t * coeff, - expbap_t * expbap, quantizer_t * quantizer, - sample_t level, int dither, int end) + expbap_t * expbap, quantizer_set_t * quant, + level_t level, int dither, int end) { int i; uint8_t * exp; int8_t * bap; + +#ifndef LIBA52_FIXED sample_t factor[25]; for (i = 0; i <= 24; i++) factor[i] = scale_factor[i] * level; +#endif exp = expbap->exp; bap = expbap->bap; @@ -336,7 +369,7 @@ switch (bapi) { case 0: if (dither) { - coeff[i] = dither_gen (state) * LEVEL_3DB * factor[exp[i]]; + COEFF (coeff[i], dither_gen (state), level, factor, exp[i]); continue; } else { coeff[i] = 0; @@ -344,76 +377,80 @@ } case -1: - if (quantizer->q1_ptr >= 0) { - coeff[i] = quantizer->q1[quantizer->q1_ptr--] * factor[exp[i]]; + if (quant->q1_ptr >= 0) { + COEFF (coeff[i], quant->q1[quant->q1_ptr--], level, + factor, exp[i]); continue; } else { int code; code = bitstream_get (state, 5); - quantizer->q1_ptr = 1; - quantizer->q1[0] = q_1_2[code]; - quantizer->q1[1] = q_1_1[code]; - coeff[i] = q_1_0[code] * factor[exp[i]]; + quant->q1_ptr = 1; + quant->q1[0] = q_1_2[code]; + quant->q1[1] = q_1_1[code]; + COEFF (coeff[i], q_1_0[code], level, factor, exp[i]); continue; } case -2: - if (quantizer->q2_ptr >= 0) { - coeff[i] = quantizer->q2[quantizer->q2_ptr--] * factor[exp[i]]; + if (quant->q2_ptr >= 0) { + COEFF (coeff[i], quant->q2[quant->q2_ptr--], level, + factor, exp[i]); continue; } else { int code; code = bitstream_get (state, 7); - quantizer->q2_ptr = 1; - quantizer->q2[0] = q_2_2[code]; - quantizer->q2[1] = q_2_1[code]; - coeff[i] = q_2_0[code] * factor[exp[i]]; + quant->q2_ptr = 1; + quant->q2[0] = q_2_2[code]; + quant->q2[1] = q_2_1[code]; + COEFF (coeff[i], q_2_0[code], level, factor, exp[i]); continue; } case 3: - coeff[i] = q_3[bitstream_get (state, 3)] * factor[exp[i]]; + COEFF (coeff[i], q_3[bitstream_get (state, 3)], level, + factor, exp[i]); continue; case -3: - if (quantizer->q4_ptr == 0) { - quantizer->q4_ptr = -1; - coeff[i] = quantizer->q4 * factor[exp[i]]; + if (quant->q4_ptr == 0) { + quant->q4_ptr = -1; + COEFF (coeff[i], quant->q4, level, factor, exp[i]); continue; } else { int code; code = bitstream_get (state, 7); - quantizer->q4_ptr = 0; - quantizer->q4 = q_4_1[code]; - coeff[i] = q_4_0[code] * factor[exp[i]]; + quant->q4_ptr = 0; + quant->q4 = q_4_1[code]; + COEFF (coeff[i], q_4_0[code], level, factor, exp[i]); continue; } case 4: - coeff[i] = q_5[bitstream_get (state, 4)] * factor[exp[i]]; + COEFF (coeff[i], q_5[bitstream_get (state, 4)], level, + factor, exp[i]); continue; default: - coeff[i] = ((bitstream_get_2 (state, bapi) << (16 - bapi)) * - factor[exp[i]]); + COEFF (coeff[i], bitstream_get_2 (state, bapi) << (16 - bapi), + level, factor, exp[i]); } } } static void coeff_get_coupling (a52_state_t * state, int nfchans, - sample_t * coeff, sample_t (* samples)[256], - quantizer_t * quantizer, uint8_t dithflag[5]) + level_t * coeff, sample_t (* samples)[256], + quantizer_set_t * quant, uint8_t dithflag[5]) { int cplbndstrc, bnd, i, i_end, ch; uint8_t * exp; int8_t * bap; - sample_t cplco[5]; + level_t cplco[5]; exp = state->cpl_expbap.exp; bap = state->cpl_expbap.bap; @@ -428,22 +465,26 @@ } cplbndstrc >>= 1; for (ch = 0; ch < nfchans; ch++) - cplco[ch] = state->cplco[ch][bnd] * coeff[ch]; + cplco[ch] = MUL_L (state->cplco[ch][bnd], coeff[ch]); bnd++; while (i < i_end) { - sample_t cplcoeff; + quantizer_t cplcoeff; int bapi; bapi = bap[i]; switch (bapi) { case 0: - cplcoeff = LEVEL_3DB * scale_factor[exp[i]]; for (ch = 0; ch < nfchans; ch++) if ((state->chincpl >> ch) & 1) { if (dithflag[ch]) - samples[ch][i] = (cplcoeff * cplco[ch] * - dither_gen (state)); +#ifndef LIBA52_FIXED + samples[ch][i] = (scale_factor[exp[i]] * + cplco[ch] * dither_gen (state)); +#else + COEFF (samples[ch][i], dither_gen (state), + cplco[ch], scale_factor, exp[i]); +#endif else samples[ch][i] = 0; } @@ -451,33 +492,33 @@ continue; case -1: - if (quantizer->q1_ptr >= 0) { - cplcoeff = quantizer->q1[quantizer->q1_ptr--]; + if (quant->q1_ptr >= 0) { + cplcoeff = quant->q1[quant->q1_ptr--]; break; } else { int code; code = bitstream_get (state, 5); - quantizer->q1_ptr = 1; - quantizer->q1[0] = q_1_2[code]; - quantizer->q1[1] = q_1_1[code]; + quant->q1_ptr = 1; + quant->q1[0] = q_1_2[code]; + quant->q1[1] = q_1_1[code]; cplcoeff = q_1_0[code]; break; } case -2: - if (quantizer->q2_ptr >= 0) { - cplcoeff = quantizer->q2[quantizer->q2_ptr--]; + if (quant->q2_ptr >= 0) { + cplcoeff = quant->q2[quant->q2_ptr--]; break; } else { int code; code = bitstream_get (state, 7); - quantizer->q2_ptr = 1; - quantizer->q2[0] = q_2_2[code]; - quantizer->q2[1] = q_2_1[code]; + quant->q2_ptr = 1; + quant->q2[0] = q_2_2[code]; + quant->q2[1] = q_2_1[code]; cplcoeff = q_2_0[code]; break; } @@ -487,17 +528,17 @@ break; case -3: - if (quantizer->q4_ptr == 0) { - quantizer->q4_ptr = -1; - cplcoeff = quantizer->q4; + if (quant->q4_ptr == 0) { + quant->q4_ptr = -1; + cplcoeff = quant->q4; break; } else { int code; code = bitstream_get (state, 7); - quantizer->q4_ptr = 0; - quantizer->q4 = q_4_1[code]; + quant->q4_ptr = 0; + quant->q4 = q_4_1[code]; cplcoeff = q_4_0[code]; break; } @@ -509,11 +550,17 @@ default: cplcoeff = bitstream_get_2 (state, bapi) << (16 - bapi); } - +#ifndef LIBA52_FIXED cplcoeff *= scale_factor[exp[i]]; +#endif for (ch = 0; ch < nfchans; ch++) - if ((state->chincpl >> ch) & 1) + if ((state->chincpl >> ch) & 1) +#ifndef LIBA52_FIXED samples[ch][i] = cplcoeff * cplco[ch]; +#else + COEFF (samples[ch][i], cplcoeff, cplco[ch], + scale_factor, exp[i]); +#endif i++; } } @@ -526,9 +573,9 @@ int i, nfchans, chaninfo; uint8_t cplexpstr, chexpstr[5], lfeexpstr, do_bit_alloc, done_cpl; uint8_t blksw[5], dithflag[5]; - sample_t coeff[5]; + level_t coeff[5]; int chanbias; - quantizer_t quantizer; + quantizer_set_t quant; sample_t * samples; nfchans = nfchans_tbl[state->acmod]; @@ -546,13 +593,17 @@ dynrng = bitstream_get_2 (state, 8); if (state->dynrnge) { - sample_t range; + level_t range; +#if !defined(LIBA52_FIXED) range = ((((dynrng & 0x1f) | 0x20) << 13) * scale_factor[3 - (dynrng >> 5)]); +#else + range = ((dynrng & 0x1f) | 0x20) << (21 + (dynrng >> 5)); +#endif if (state->dynrngcall) range = state->dynrngcall (range, state->dynrngdata); - state->dynrng = state->level * range; + state->dynrng = MUL_L (state->level, range); } } } while (chaninfo--); @@ -611,8 +662,13 @@ cplcomant <<= 14; else cplcomant = (cplcomant | 0x10) << 13; +#ifndef LIBA52_FIXED state->cplco[i][j] = cplcomant * scale_factor[cplcoexp + mstrcplco]; +#else + state->cplco[i][j] = (cplcomant << 11) >> (cplcoexp + mstrcplco); +#endif + } } if ((state->acmod == 2) && state->phsflginu && cplcoe) @@ -691,11 +747,11 @@ } if (bitstream_get (state, 1)) { /* baie */ - do_bit_alloc = -1; + do_bit_alloc = 127; state->bai = bitstream_get (state, 11); } if (bitstream_get (state, 1)) { /* snroffste */ - do_bit_alloc = -1; + do_bit_alloc = 127; state->csnroffst = bitstream_get (state, 6); if (state->chincpl) /* cplinu */ state->cplba.bai = bitstream_get (state, 7); @@ -711,7 +767,7 @@ } if (bitstream_get (state, 1)) { /* deltbaie */ - do_bit_alloc = -1; + do_bit_alloc = 127; if (state->chincpl) /* cplinu */ state->cplba.deltbae = bitstream_get (state, 2); for (i = 0; i < nfchans; i++) @@ -765,20 +821,20 @@ chanbias = a52_downmix_coeff (coeff, state->acmod, state->output, state->dynrng, state->clev, state->slev); - quantizer.q1_ptr = quantizer.q2_ptr = quantizer.q4_ptr = -1; + quant.q1_ptr = quant.q2_ptr = quant.q4_ptr = -1; done_cpl = 0; for (i = 0; i < nfchans; i++) { int j; - coeff_get (state, samples + 256 * i, state->fbw_expbap +i, &quantizer, + coeff_get (state, samples + 256 * i, state->fbw_expbap +i, &quant, coeff[i], dithflag[i], state->endmant[i]); if ((state->chincpl >> i) & 1) { if (!done_cpl) { done_cpl = 1; coeff_get_coupling (state, nfchans, coeff, - (sample_t (*)[256])samples, &quantizer, + (sample_t (*)[256])samples, &quant, dithflag); } j = state->cplendmant; @@ -821,14 +877,14 @@ if (state->lfeon) { if (state->output & A52_LFE) { - coeff_get (state, samples - 256, &state->lfe_expbap, &quantizer, + coeff_get (state, samples - 256, &state->lfe_expbap, &quant, state->dynrng, 0, 7); for (i = 7; i < 256; i++) (samples-256)[i] = 0; a52_imdct_512 (samples - 256, samples + 1536 - 256, state->bias); } else { /* just skip the LFE coefficients */ - coeff_get (state, samples + 1280, &state->lfe_expbap, &quantizer, + coeff_get (state, samples + 1280, &state->lfe_expbap, &quant, 0, 0, 7); } } @@ -896,6 +952,9 @@ void a52_free (a52_state_t * state) { + (void)state; + /* free (state->samples); free (state); + */ } diff -ruN a52dec.orig/liba52/tables.h a52dec/liba52/tables.h --- a52dec.orig/liba52/tables.h 2012-11-13 17:25:33.438758556 +0100 +++ a52dec/liba52/tables.h 2012-11-13 17:27:31.150370945 +0100 @@ -1,6 +1,6 @@ /* * tables.h - * Copyright (C) 2000-2002 Michel Lespinasse + * Copyright (C) 2000-2003 Michel Lespinasse * Copyright (C) 1999-2000 Aaron Holtzman * * This file is part of a52dec, a free ATSC A-52 stream decoder. @@ -46,42 +46,44 @@ 25,25,25 }; -#define Q0 ((-2 << 15) / 3.0) -#define Q1 (0) -#define Q2 ((2 << 15) / 3.0) - -static const sample_t q_1_0[32] = { - Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0, - Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1, - Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2, - 0,0,0,0,0 -}; - -static const sample_t q_1_1[32] = { - Q0,Q0,Q0,Q1,Q1,Q1,Q2,Q2,Q2, - Q0,Q0,Q0,Q1,Q1,Q1,Q2,Q2,Q2, - Q0,Q0,Q0,Q1,Q1,Q1,Q2,Q2,Q2, - 0,0,0,0,0 -}; - -static const sample_t q_1_2[32] = { - Q0,Q1,Q2,Q0,Q1,Q2,Q0,Q1,Q2, - Q0,Q1,Q2,Q0,Q1,Q2,Q0,Q1,Q2, - Q0,Q1,Q2,Q0,Q1,Q2,Q0,Q1,Q2, - 0,0,0,0,0 +#define Q(x) ROUND (32768.0 * x) + +#define Q0 Q (-2/3) +#define Q1 Q (0) +#define Q2 Q (2/3) + +static const quantizer_t q_1_0[32] = { + Q0, Q0, Q0, Q0, Q0, Q0, Q0, Q0, Q0, + Q1, Q1, Q1, Q1, Q1, Q1, Q1, Q1, Q1, + Q2, Q2, Q2, Q2, Q2, Q2, Q2, Q2, Q2, + 0, 0, 0, 0, 0 +}; + +static const quantizer_t q_1_1[32] = { + Q0, Q0, Q0, Q1, Q1, Q1, Q2, Q2, Q2, + Q0, Q0, Q0, Q1, Q1, Q1, Q2, Q2, Q2, + Q0, Q0, Q0, Q1, Q1, Q1, Q2, Q2, Q2, + 0, 0, 0, 0, 0 +}; + +static const quantizer_t q_1_2[32] = { + Q0, Q1, Q2, Q0, Q1, Q2, Q0, Q1, Q2, + Q0, Q1, Q2, Q0, Q1, Q2, Q0, Q1, Q2, + Q0, Q1, Q2, Q0, Q1, Q2, Q0, Q1, Q2, + 0, 0, 0, 0, 0 }; #undef Q0 #undef Q1 #undef Q2 -#define Q0 ((-4 << 15) / 5.0) -#define Q1 ((-2 << 15) / 5.0) -#define Q2 (0) -#define Q3 ((2 << 15) / 5.0) -#define Q4 ((4 << 15) / 5.0) +#define Q0 Q (-4/5) +#define Q1 Q (-2/5) +#define Q2 Q (0) +#define Q3 Q (2/5) +#define Q4 Q (4/5) -static const sample_t q_2_0[128] = { +static const quantizer_t q_2_0[128] = { Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0,Q0, Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1,Q1, Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2,Q2, @@ -90,7 +92,7 @@ 0,0,0 }; -static const sample_t q_2_1[128] = { +static const quantizer_t q_2_1[128] = { Q0,Q0,Q0,Q0,Q0,Q1,Q1,Q1,Q1,Q1,Q2,Q2,Q2,Q2,Q2,Q3,Q3,Q3,Q3,Q3,Q4,Q4,Q4,Q4,Q4, Q0,Q0,Q0,Q0,Q0,Q1,Q1,Q1,Q1,Q1,Q2,Q2,Q2,Q2,Q2,Q3,Q3,Q3,Q3,Q3,Q4,Q4,Q4,Q4,Q4, Q0,Q0,Q0,Q0,Q0,Q1,Q1,Q1,Q1,Q1,Q2,Q2,Q2,Q2,Q2,Q3,Q3,Q3,Q3,Q3,Q4,Q4,Q4,Q4,Q4, @@ -99,7 +101,7 @@ 0,0,0 }; -static const sample_t q_2_2[128] = { +static const quantizer_t q_2_2[128] = { Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4, Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4, Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4,Q0,Q1,Q2,Q3,Q4, @@ -114,24 +116,23 @@ #undef Q3 #undef Q4 -static const sample_t q_3[8] = { - (-6 << 15)/7.0, (-4 << 15)/7.0, (-2 << 15)/7.0, 0, - ( 2 << 15)/7.0, ( 4 << 15)/7.0, ( 6 << 15)/7.0, 0 -}; - -#define Q0 ((-10 << 15) / 11.0) -#define Q1 ((-8 << 15) / 11.0) -#define Q2 ((-6 << 15) / 11.0) -#define Q3 ((-4 << 15) / 11.0) -#define Q4 ((-2 << 15) / 11.0) -#define Q5 (0) -#define Q6 ((2 << 15) / 11.0) -#define Q7 ((4 << 15) / 11.0) -#define Q8 ((6 << 15) / 11.0) -#define Q9 ((8 << 15) / 11.0) -#define QA ((10 << 15) / 11.0) +static const quantizer_t q_3[8] = { + Q (-6/7), Q (-4/7), Q (-2/7), Q (0), Q (2/7), Q (4/7), Q (6/7), 0 +}; + +#define Q0 Q (-10/11) +#define Q1 Q (-8/11) +#define Q2 Q (-6/11) +#define Q3 Q (-4/11) +#define Q4 Q (-2/11) +#define Q5 Q (0) +#define Q6 Q (2/11) +#define Q7 Q (4/11) +#define Q8 Q (6/11) +#define Q9 Q (8/11) +#define QA Q (10/11) -static const sample_t q_4_0[128] = { +static const quantizer_t q_4_0[128] = { Q0, Q0, Q0, Q0, Q0, Q0, Q0, Q0, Q0, Q0, Q0, Q1, Q1, Q1, Q1, Q1, Q1, Q1, Q1, Q1, Q1, Q1, Q2, Q2, Q2, Q2, Q2, Q2, Q2, Q2, Q2, Q2, Q2, @@ -146,7 +147,7 @@ 0, 0, 0, 0, 0, 0, 0 }; -static const sample_t q_4_1[128] = { +static const quantizer_t q_4_1[128] = { Q0, Q1, Q2, Q3, Q4, Q5, Q6, Q7, Q8, Q9, QA, Q0, Q1, Q2, Q3, Q4, Q5, Q6, Q7, Q8, Q9, QA, Q0, Q1, Q2, Q3, Q4, Q5, Q6, Q7, Q8, Q9, QA, @@ -173,15 +174,13 @@ #undef Q9 #undef QA -static const sample_t q_5[16] = { - (-14 << 15)/15.0,(-12 << 15)/15.0,(-10 << 15)/15.0, - ( -8 << 15)/15.0,( -6 << 15)/15.0,( -4 << 15)/15.0, - ( -2 << 15)/15.0, 0 ,( 2 << 15)/15.0, - ( 4 << 15)/15.0,( 6 << 15)/15.0,( 8 << 15)/15.0, - ( 10 << 15)/15.0,( 12 << 15)/15.0,( 14 << 15)/15.0, - 0 +static const quantizer_t q_5[16] = { + Q (-14/15), Q (-12/15), Q (-10/15), Q (-8/15), Q (-6/15), + Q (-4/15), Q (-2/15), Q (0), Q (2/15), Q (4/15), + Q (6/15), Q (8/15), Q (10/15), Q (12/15), Q (14/15), 0 }; +#ifndef LIBA52_FIXED static const sample_t scale_factor[25] = { 0.000030517578125, 0.0000152587890625, @@ -209,6 +208,7 @@ 0.00000000000363797880709171295166015625, 0.000000000001818989403545856475830078125 }; +#endif static const uint16_t dither_lut[256] = { 0x0000, 0xa011, 0xe033, 0x4022, 0x6077, 0xc066, 0x8044, 0x2055,