git.sesse.net Git - x264/blob - common/bitstream.c

   1 /*****************************************************************************
   2  * bitstream.c: bitstream writing
   3  *****************************************************************************
   4  * Copyright (C) 2003-2015 x264 project
   5  *
   6  * Authors: Laurent Aimar <fenrir@via.ecp.fr>
   7  *          Fiona Glaser <fiona@x264.com>
   8  *
   9  * This program is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * This program is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License
  20  * along with this program; if not, write to the Free Software
  21  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
  22  *
  23  * This program is also available under a commercial proprietary license.
  24  * For more information, contact us at licensing@x264.com.
  25  *****************************************************************************/
  26
  27 #include "common.h"
  28
  29 static uint8_t *x264_nal_escape_c( uint8_t *dst, uint8_t *src, uint8_t *end )
  30 {
  31     if( src < end ) *dst++ = *src++;
  32     if( src < end ) *dst++ = *src++;
  33     while( src < end )
  34     {
  35         if( src[0] <= 0x03 && !dst[-2] && !dst[-1] )
  36             *dst++ = 0x03;
  37         *dst++ = *src++;
  38     }
  39     return dst;
  40 }
  41
  42 uint8_t *x264_nal_escape_mmx2( uint8_t *dst, uint8_t *src, uint8_t *end );
  43 uint8_t *x264_nal_escape_sse2( uint8_t *dst, uint8_t *src, uint8_t *end );
  44 uint8_t *x264_nal_escape_avx2( uint8_t *dst, uint8_t *src, uint8_t *end );
  45 void x264_cabac_block_residual_rd_internal_sse2       ( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  46 void x264_cabac_block_residual_rd_internal_sse2_lzcnt ( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  47 void x264_cabac_block_residual_rd_internal_ssse3      ( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  48 void x264_cabac_block_residual_rd_internal_ssse3_lzcnt( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  49 void x264_cabac_block_residual_8x8_rd_internal_sse2       ( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  50 void x264_cabac_block_residual_8x8_rd_internal_sse2_lzcnt ( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  51 void x264_cabac_block_residual_8x8_rd_internal_ssse3      ( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  52 void x264_cabac_block_residual_8x8_rd_internal_ssse3_lzcnt( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  53 void x264_cabac_block_residual_internal_sse2       ( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  54 void x264_cabac_block_residual_internal_sse2_lzcnt ( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  55 void x264_cabac_block_residual_internal_avx2_bmi2 ( dctcoef *l, int b_interlaced, intptr_t ctx_block_cat, x264_cabac_t *cb );
  56
  57 uint8_t *x264_nal_escape_neon( uint8_t *dst, uint8_t *src, uint8_t *end );
  58
  59 /****************************************************************************
  60  * x264_nal_encode:
  61  ****************************************************************************/
  62 void x264_nal_encode( x264_t *h, uint8_t *dst, x264_nal_t *nal )
  63 {
  64     uint8_t *src = nal->p_payload;
  65     uint8_t *end = nal->p_payload + nal->i_payload;
  66     uint8_t *orig_dst = dst;
  67
  68     if( h->param.b_annexb )
  69     {
  70         if( nal->b_long_startcode )
  71             *dst++ = 0x00;
  72         *dst++ = 0x00;
  73         *dst++ = 0x00;
  74         *dst++ = 0x01;
  75     }
  76     else /* save room for size later */
  77         dst += 4;
  78
  79     /* nal header */
  80     *dst++ = ( 0x00 << 7 ) | ( nal->i_ref_idc << 5 ) | nal->i_type;
  81
  82     dst = h->bsf.nal_escape( dst, src, end );
  83     int size = (dst - orig_dst) - 4;
  84
  85     /* Write the size header for mp4/etc */
  86     if( !h->param.b_annexb )
  87     {
  88         /* Size doesn't include the size of the header we're writing now. */
  89         orig_dst[0] = size>>24;
  90         orig_dst[1] = size>>16;
  91         orig_dst[2] = size>> 8;
  92         orig_dst[3] = size>> 0;
  93     }
  94
  95     nal->i_payload = size+4;
  96     nal->p_payload = orig_dst;
  97     x264_emms();
  98 }
  99
 100 void x264_bitstream_init( int cpu, x264_bitstream_function_t *pf )
 101 {
 102     memset( pf, 0, sizeof(*pf) );
 103
 104     pf->nal_escape = x264_nal_escape_c;
 105 #if HAVE_MMX
 106 #if ARCH_X86_64
 107     pf->cabac_block_residual_internal = x264_cabac_block_residual_internal_sse2;
 108     pf->cabac_block_residual_rd_internal = x264_cabac_block_residual_rd_internal_sse2;
 109     pf->cabac_block_residual_8x8_rd_internal = x264_cabac_block_residual_8x8_rd_internal_sse2;
 110 #endif
 111
 112     if( cpu&X264_CPU_MMX2 )
 113         pf->nal_escape = x264_nal_escape_mmx2;
 114     if( cpu&X264_CPU_SSE2 )
 115     {
 116 #if ARCH_X86_64
 117         if( cpu&X264_CPU_LZCNT )
 118         {
 119             pf->cabac_block_residual_internal = x264_cabac_block_residual_internal_sse2_lzcnt;
 120             pf->cabac_block_residual_rd_internal = x264_cabac_block_residual_rd_internal_sse2_lzcnt;
 121             pf->cabac_block_residual_8x8_rd_internal = x264_cabac_block_residual_8x8_rd_internal_sse2_lzcnt;
 122         }
 123 #endif
 124         if( cpu&X264_CPU_SSE2_IS_FAST )
 125             pf->nal_escape = x264_nal_escape_sse2;
 126     }
 127 #if ARCH_X86_64
 128     if( cpu&X264_CPU_SSSE3 )
 129     {
 130         pf->cabac_block_residual_rd_internal = x264_cabac_block_residual_rd_internal_ssse3;
 131         pf->cabac_block_residual_8x8_rd_internal = x264_cabac_block_residual_8x8_rd_internal_ssse3;
 132         if( cpu&X264_CPU_LZCNT )
 133         {
 134             pf->cabac_block_residual_rd_internal = x264_cabac_block_residual_rd_internal_ssse3_lzcnt;
 135             pf->cabac_block_residual_8x8_rd_internal = x264_cabac_block_residual_8x8_rd_internal_ssse3_lzcnt;
 136         }
 137     }
 138
 139     if( cpu&X264_CPU_AVX2 )
 140     {
 141         pf->nal_escape = x264_nal_escape_avx2;
 142         if( cpu&X264_CPU_BMI2 )
 143             pf->cabac_block_residual_internal = x264_cabac_block_residual_internal_avx2_bmi2;
 144     }
 145 #endif
 146 #endif
 147 #if HAVE_ARMV6
 148     if( cpu&X264_CPU_NEON )
 149         pf->nal_escape = x264_nal_escape_neon;
 150 #endif
 151 #if ARCH_AARCH64
 152     if( cpu&X264_CPU_NEON )
 153         pf->nal_escape = x264_nal_escape_neon;
 154 #endif
 155 }