* H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder
* Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
- * version 2 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
- * License along with this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*
*/
* @author Michael Niedermayer <michaelni@gmx.at>
*/
-#include "common.h"
#include "dsputil.h"
#include "avcodec.h"
#include "mpegvideo.h"
+#include "h264.h"
#include "h264data.h"
+#include "h264_parser.h"
#include "golomb.h"
#include "cabac.h"
//#undef NDEBUG
#include <assert.h>
-#define interlaced_dct interlaced_dct_is_a_bad_name
-#define mb_intra mb_intra_isnt_initalized_see_mb_type
-
-#define LUMA_DC_BLOCK_INDEX 25
-#define CHROMA_DC_BLOCK_INDEX 26
-
-#define CHROMA_DC_COEFF_TOKEN_VLC_BITS 8
-#define COEFF_TOKEN_VLC_BITS 8
-#define TOTAL_ZEROS_VLC_BITS 9
-#define CHROMA_DC_TOTAL_ZEROS_VLC_BITS 3
-#define RUN_VLC_BITS 3
-#define RUN7_VLC_BITS 6
-
-#define MAX_SPS_COUNT 32
-#define MAX_PPS_COUNT 256
-
-#define MAX_MMCO_COUNT 66
-
-/* Compiling in interlaced support reduces the speed
- * of progressive decoding by about 2%. */
-#define ALLOW_INTERLACE
-
-#ifdef ALLOW_INTERLACE
-#define MB_MBAFF h->mb_mbaff
-#define MB_FIELD h->mb_field_decoding_flag
-#define FRAME_MBAFF h->mb_aff_frame
-#else
-#define MB_MBAFF 0
-#define MB_FIELD 0
-#define FRAME_MBAFF 0
-#undef IS_INTERLACED
-#define IS_INTERLACED(mb_type) 0
-#endif
-
-/**
- * Sequence parameter set
- */
-typedef struct SPS{
-
- int profile_idc;
- int level_idc;
- int transform_bypass; ///< qpprime_y_zero_transform_bypass_flag
- int log2_max_frame_num; ///< log2_max_frame_num_minus4 + 4
- int poc_type; ///< pic_order_cnt_type
- int log2_max_poc_lsb; ///< log2_max_pic_order_cnt_lsb_minus4
- int delta_pic_order_always_zero_flag;
- int offset_for_non_ref_pic;
- int offset_for_top_to_bottom_field;
- int poc_cycle_length; ///< num_ref_frames_in_pic_order_cnt_cycle
- int ref_frame_count; ///< num_ref_frames
- int gaps_in_frame_num_allowed_flag;
- int mb_width; ///< frame_width_in_mbs_minus1 + 1
- int mb_height; ///< frame_height_in_mbs_minus1 + 1
- int frame_mbs_only_flag;
- int mb_aff; ///<mb_adaptive_frame_field_flag
- int direct_8x8_inference_flag;
- int crop; ///< frame_cropping_flag
- int crop_left; ///< frame_cropping_rect_left_offset
- int crop_right; ///< frame_cropping_rect_right_offset
- int crop_top; ///< frame_cropping_rect_top_offset
- int crop_bottom; ///< frame_cropping_rect_bottom_offset
- int vui_parameters_present_flag;
- AVRational sar;
- int timing_info_present_flag;
- uint32_t num_units_in_tick;
- uint32_t time_scale;
- int fixed_frame_rate_flag;
- short offset_for_ref_frame[256]; //FIXME dyn aloc?
- int bitstream_restriction_flag;
- int num_reorder_frames;
- int scaling_matrix_present;
- uint8_t scaling_matrix4[6][16];
- uint8_t scaling_matrix8[2][64];
-}SPS;
-
-/**
- * Picture parameter set
- */
-typedef struct PPS{
- int sps_id;
- int cabac; ///< entropy_coding_mode_flag
- int pic_order_present; ///< pic_order_present_flag
- int slice_group_count; ///< num_slice_groups_minus1 + 1
- int mb_slice_group_map_type;
- int ref_count[2]; ///< num_ref_idx_l0/1_active_minus1 + 1
- int weighted_pred; ///< weighted_pred_flag
- int weighted_bipred_idc;
- int init_qp; ///< pic_init_qp_minus26 + 26
- int init_qs; ///< pic_init_qs_minus26 + 26
- int chroma_qp_index_offset;
- int deblocking_filter_parameters_present; ///< deblocking_filter_parameters_present_flag
- int constrained_intra_pred; ///< constrained_intra_pred_flag
- int redundant_pic_cnt_present; ///< redundant_pic_cnt_present_flag
- int transform_8x8_mode; ///< transform_8x8_mode_flag
- uint8_t scaling_matrix4[6][16];
- uint8_t scaling_matrix8[2][64];
-}PPS;
-
-/**
- * Memory management control operation opcode.
- */
-typedef enum MMCOOpcode{
- MMCO_END=0,
- MMCO_SHORT2UNUSED,
- MMCO_LONG2UNUSED,
- MMCO_SHORT2LONG,
- MMCO_SET_MAX_LONG,
- MMCO_RESET,
- MMCO_LONG,
-} MMCOOpcode;
-
-/**
- * Memory management control operation.
- */
-typedef struct MMCO{
- MMCOOpcode opcode;
- int short_frame_num;
- int long_index;
-} MMCO;
-
-/**
- * H264Context
- */
-typedef struct H264Context{
- MpegEncContext s;
- int nal_ref_idc;
- int nal_unit_type;
-#define NAL_SLICE 1
-#define NAL_DPA 2
-#define NAL_DPB 3
-#define NAL_DPC 4
-#define NAL_IDR_SLICE 5
-#define NAL_SEI 6
-#define NAL_SPS 7
-#define NAL_PPS 8
-#define NAL_AUD 9
-#define NAL_END_SEQUENCE 10
-#define NAL_END_STREAM 11
-#define NAL_FILLER_DATA 12
-#define NAL_SPS_EXT 13
-#define NAL_AUXILIARY_SLICE 19
- uint8_t *rbsp_buffer;
- unsigned int rbsp_buffer_size;
-
- /**
- * Used to parse AVC variant of h264
- */
- int is_avc; ///< this flag is != 0 if codec is avc1
- int got_avcC; ///< flag used to parse avcC data only once
- int nal_length_size; ///< Number of bytes used for nal length (1, 2 or 4)
-
- int chroma_qp; //QPc
-
- int prev_mb_skipped;
- int next_mb_skipped;
-
- //prediction stuff
- int chroma_pred_mode;
- int intra16x16_pred_mode;
-
- int top_mb_xy;
- int left_mb_xy[2];
-
- int8_t intra4x4_pred_mode_cache[5*8];
- int8_t (*intra4x4_pred_mode)[8];
- void (*pred4x4 [9+3])(uint8_t *src, uint8_t *topright, int stride);//FIXME move to dsp?
- void (*pred8x8l [9+3])(uint8_t *src, int topleft, int topright, int stride);
- void (*pred8x8 [4+3])(uint8_t *src, int stride);
- void (*pred16x16[4+3])(uint8_t *src, int stride);
- unsigned int topleft_samples_available;
- unsigned int top_samples_available;
- unsigned int topright_samples_available;
- unsigned int left_samples_available;
- uint8_t (*top_borders[2])[16+2*8];
- uint8_t left_border[2*(17+2*9)];
-
- /**
- * non zero coeff count cache.
- * is 64 if not available.
- */
- DECLARE_ALIGNED_8(uint8_t, non_zero_count_cache[6*8]);
- uint8_t (*non_zero_count)[16];
-
- /**
- * Motion vector cache.
- */
- DECLARE_ALIGNED_8(int16_t, mv_cache[2][5*8][2]);
- DECLARE_ALIGNED_8(int8_t, ref_cache[2][5*8]);
-#define LIST_NOT_USED -1 //FIXME rename?
-#define PART_NOT_AVAILABLE -2
-
- /**
- * is 1 if the specific list MV&references are set to 0,0,-2.
- */
- int mv_cache_clean[2];
-
- /**
- * number of neighbors (top and/or left) that used 8x8 dct
- */
- int neighbor_transform_size;
-
- /**
- * block_offset[ 0..23] for frame macroblocks
- * block_offset[24..47] for field macroblocks
- */
- int block_offset[2*(16+8)];
-
- uint32_t *mb2b_xy; //FIXME are these 4 a good idea?
- uint32_t *mb2b8_xy;
- int b_stride; //FIXME use s->b4_stride
- int b8_stride;
-
- int mb_linesize; ///< may be equal to s->linesize or s->linesize*2, for mbaff
- int mb_uvlinesize;
-
- int emu_edge_width;
- int emu_edge_height;
-
- int halfpel_flag;
- int thirdpel_flag;
-
- int unknown_svq3_flag;
- int next_slice_index;
-
- SPS sps_buffer[MAX_SPS_COUNT];
- SPS sps; ///< current sps
-
- PPS pps_buffer[MAX_PPS_COUNT];
- /**
- * current pps
- */
- PPS pps; //FIXME move to Picture perhaps? (->no) do we need that?
-
- uint32_t dequant4_buffer[6][52][16];
- uint32_t dequant8_buffer[2][52][64];
- uint32_t (*dequant4_coeff[6])[16];
- uint32_t (*dequant8_coeff[2])[64];
- int dequant_coeff_pps; ///< reinit tables when pps changes
-
- int slice_num;
- uint8_t *slice_table_base;
- uint8_t *slice_table; ///< slice_table_base + 2*mb_stride + 1
- int slice_type;
- int slice_type_fixed;
-
- //interlacing specific flags
- int mb_aff_frame;
- int mb_field_decoding_flag;
- int mb_mbaff; ///< mb_aff_frame && mb_field_decoding_flag
-
- int sub_mb_type[4];
-
- //POC stuff
- int poc_lsb;
- int poc_msb;
- int delta_poc_bottom;
- int delta_poc[2];
- int frame_num;
- int prev_poc_msb; ///< poc_msb of the last reference pic for POC type 0
- int prev_poc_lsb; ///< poc_lsb of the last reference pic for POC type 0
- int frame_num_offset; ///< for POC type 2
- int prev_frame_num_offset; ///< for POC type 2
- int prev_frame_num; ///< frame_num of the last pic for POC type 1/2
-
- /**
- * frame_num for frames or 2*frame_num for field pics.
- */
- int curr_pic_num;
-
- /**
- * max_frame_num or 2*max_frame_num for field pics.
- */
- int max_pic_num;
-
- //Weighted pred stuff
- int use_weight;
- int use_weight_chroma;
- int luma_log2_weight_denom;
- int chroma_log2_weight_denom;
- int luma_weight[2][48];
- int luma_offset[2][48];
- int chroma_weight[2][48][2];
- int chroma_offset[2][48][2];
- int implicit_weight[48][48];
-
- //deblock
- int deblocking_filter; ///< disable_deblocking_filter_idc with 1<->0
- int slice_alpha_c0_offset;
- int slice_beta_offset;
-
- int redundant_pic_count;
-
- int direct_spatial_mv_pred;
- int dist_scale_factor[16];
- int dist_scale_factor_field[32];
- int map_col_to_list0[2][16];
- int map_col_to_list0_field[2][32];
-
- /**
- * num_ref_idx_l0/1_active_minus1 + 1
- */
- int ref_count[2]; ///< counts frames or fields, depending on current mb mode
- Picture *short_ref[32];
- Picture *long_ref[32];
- Picture default_ref_list[2][32];
- Picture ref_list[2][48]; ///< 0..15: frame refs, 16..47: mbaff field refs
- Picture *delayed_pic[16]; //FIXME size?
- Picture *delayed_output_pic;
-
- /**
- * memory management control operations buffer.
- */
- MMCO mmco[MAX_MMCO_COUNT];
- int mmco_index;
-
- int long_ref_count; ///< number of actual long term references
- int short_ref_count; ///< number of actual short term references
-
- //data partitioning
- GetBitContext intra_gb;
- GetBitContext inter_gb;
- GetBitContext *intra_gb_ptr;
- GetBitContext *inter_gb_ptr;
-
- DECLARE_ALIGNED_8(DCTELEM, mb[16*24]);
-
- /**
- * Cabac
- */
- CABACContext cabac;
- uint8_t cabac_state[460];
- int cabac_init_idc;
-
- /* 0x100 -> non null luma_dc, 0x80/0x40 -> non null chroma_dc (cb/cr), 0x?0 -> chroma_cbp(0,1,2), 0x0? luma_cbp */
- uint16_t *cbp_table;
- int top_cbp;
- int left_cbp;
- /* chroma_pred_mode for i4x4 or i16x16, else 0 */
- uint8_t *chroma_pred_mode_table;
- int last_qscale_diff;
- int16_t (*mvd_table[2])[2];
- DECLARE_ALIGNED_8(int16_t, mvd_cache[2][5*8][2]);
- uint8_t *direct_table;
- uint8_t direct_cache[5*8];
-
- uint8_t zigzag_scan[16];
- uint8_t zigzag_scan8x8[64];
- uint8_t zigzag_scan8x8_cavlc[64];
- uint8_t field_scan[16];
- uint8_t field_scan8x8[64];
- uint8_t field_scan8x8_cavlc[64];
- const uint8_t *zigzag_scan_q0;
- const uint8_t *zigzag_scan8x8_q0;
- const uint8_t *zigzag_scan8x8_cavlc_q0;
- const uint8_t *field_scan_q0;
- const uint8_t *field_scan8x8_q0;
- const uint8_t *field_scan8x8_cavlc_q0;
-
- int x264_build;
-}H264Context;
-
static VLC coeff_token_vlc[4];
static VLC chroma_dc_coeff_token_vlc;
static void svq3_luma_dc_dequant_idct_c(DCTELEM *block, int qp);
static void svq3_add_idct_c(uint8_t *dst, DCTELEM *block, int stride, int qp, int dc);
static void filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize);
+static void filter_mb_fast( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize);
-static always_inline uint32_t pack16to32(int a, int b){
+static av_always_inline uint32_t pack16to32(int a, int b){
#ifdef WORDS_BIGENDIAN
return (b&0xFFFF) + (a<<16);
#else
#endif
}
+const uint8_t ff_rem6[52]={
+0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3,
+};
+
+const uint8_t ff_div6[52]={
+0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8,
+};
+
+
/**
* fill a rectangle.
* @param h height of the rectangle, should be a constant
* @param w width of the rectangle, should be a constant
* @param size the size of val (1 or 4), should be a constant
*/
-static always_inline void fill_rectangle(void *vp, int w, int h, int stride, uint32_t val, int size){
+static av_always_inline void fill_rectangle(void *vp, int w, int h, int stride, uint32_t val, int size){
uint8_t *p= (uint8_t*)vp;
assert(size==1 || size==4);
assert(w<=4);
int left_block[8];
int i;
- //FIXME deblocking can skip fill_caches much of the time with multiple slices too.
- // the actual condition is whether we're on the edge of a slice,
- // and even then the intra and nnz parts are unnecessary.
- if(for_deblock && h->slice_num == 1 && !FRAME_MBAFF)
+ //FIXME deblocking could skip the intra and nnz parts.
+ if(for_deblock && (h->slice_num == 1 || h->slice_table[mb_xy] == h->slice_table[mb_xy-s->mb_stride]) && !FRAME_MBAFF)
return;
//wow what a mess, why didn't they simplify the interlacing&intra stuff, i can't imagine that these complex rules are worth it
const int left_mb_frame_flag = !IS_INTERLACED(s->current_picture.mb_type[pair_xy-1]);
const int curr_mb_frame_flag = !IS_INTERLACED(mb_type);
const int bottom = (s->mb_y & 1);
- tprintf("fill_caches: curr_mb_frame_flag:%d, left_mb_frame_flag:%d, topleft_mb_frame_flag:%d, top_mb_frame_flag:%d, topright_mb_frame_flag:%d\n", curr_mb_frame_flag, left_mb_frame_flag, topleft_mb_frame_flag, top_mb_frame_flag, topright_mb_frame_flag);
+ tprintf(s->avctx, "fill_caches: curr_mb_frame_flag:%d, left_mb_frame_flag:%d, topleft_mb_frame_flag:%d, top_mb_frame_flag:%d, topright_mb_frame_flag:%d\n", curr_mb_frame_flag, left_mb_frame_flag, topleft_mb_frame_flag, top_mb_frame_flag, topright_mb_frame_flag);
if (bottom
? !curr_mb_frame_flag // bottom macroblock
: (!curr_mb_frame_flag && !top_mb_frame_flag) // top macroblock
h->left_mb_xy[0] = left_xy[0];
h->left_mb_xy[1] = left_xy[1];
if(for_deblock){
- topleft_type = h->slice_table[topleft_xy ] < 255 ? s->current_picture.mb_type[topleft_xy] : 0;
+ topleft_type = 0;
+ topright_type = 0;
top_type = h->slice_table[top_xy ] < 255 ? s->current_picture.mb_type[top_xy] : 0;
- topright_type= h->slice_table[topright_xy] < 255 ? s->current_picture.mb_type[topright_xy]: 0;
left_type[0] = h->slice_table[left_xy[0] ] < 255 ? s->current_picture.mb_type[left_xy[0]] : 0;
left_type[1] = h->slice_table[left_xy[1] ] < 255 ? s->current_picture.mb_type[left_xy[1]] : 0;
int v = *(uint16_t*)&h->non_zero_count[mb_xy][14];
for(i=0; i<16; i++)
h->non_zero_count_cache[scan8[i]] = (v>>i)&1;
- for(list=0; list<1+(h->slice_type==B_TYPE); list++){
+ for(list=0; list<h->list_count; list++){
if(USES_LIST(mb_type,list)){
uint32_t *src = (uint32_t*)s->current_picture.motion_val[list][h->mb2b_xy[mb_xy]];
uint32_t *dst = (uint32_t*)h->mv_cache[list][scan8[0]];
- uint8_t *ref = &s->current_picture.ref_index[list][h->mb2b8_xy[mb_xy]];
+ int8_t *ref = &s->current_picture.ref_index[list][h->mb2b8_xy[mb_xy]];
for(i=0; i<4; i++, dst+=8, src+=h->b_stride){
dst[0] = src[0];
dst[1] = src[1];
}
#if 1
- //FIXME direct mb can skip much of this
if(IS_INTER(mb_type) || IS_DIRECT(mb_type)){
int list;
- for(list=0; list<1+(h->slice_type==B_TYPE); list++){
+ for(list=0; list<h->list_count; list++){
if(!USES_LIST(mb_type, list) && !IS_DIRECT(mb_type) && !h->deblocking_filter){
/*if(!h->mv_cache_clean[list]){
memset(h->mv_cache [list], 0, 8*5*2*sizeof(int16_t)); //FIXME clean only input? clean at all?
*(uint32_t*)&h->ref_cache[list][scan8[0] + 0 - 1*8]= ((top_type ? LIST_NOT_USED : PART_NOT_AVAILABLE)&0xFF)*0x01010101;
}
- //FIXME unify cleanup or sth
- if(USES_LIST(left_type[0], list)){
- const int b_xy= h->mb2b_xy[left_xy[0]] + 3;
- const int b8_xy= h->mb2b8_xy[left_xy[0]] + 1;
- *(uint32_t*)h->mv_cache[list][scan8[0] - 1 + 0*8]= *(uint32_t*)s->current_picture.motion_val[list][b_xy + h->b_stride*left_block[0]];
- *(uint32_t*)h->mv_cache[list][scan8[0] - 1 + 1*8]= *(uint32_t*)s->current_picture.motion_val[list][b_xy + h->b_stride*left_block[1]];
- h->ref_cache[list][scan8[0] - 1 + 0*8]= s->current_picture.ref_index[list][b8_xy + h->b8_stride*(left_block[0]>>1)];
- h->ref_cache[list][scan8[0] - 1 + 1*8]= s->current_picture.ref_index[list][b8_xy + h->b8_stride*(left_block[1]>>1)];
- }else{
- *(uint32_t*)h->mv_cache [list][scan8[0] - 1 + 0*8]=
- *(uint32_t*)h->mv_cache [list][scan8[0] - 1 + 1*8]= 0;
- h->ref_cache[list][scan8[0] - 1 + 0*8]=
- h->ref_cache[list][scan8[0] - 1 + 1*8]= left_type[0] ? LIST_NOT_USED : PART_NOT_AVAILABLE;
- }
-
- if(USES_LIST(left_type[1], list)){
- const int b_xy= h->mb2b_xy[left_xy[1]] + 3;
- const int b8_xy= h->mb2b8_xy[left_xy[1]] + 1;
- *(uint32_t*)h->mv_cache[list][scan8[0] - 1 + 2*8]= *(uint32_t*)s->current_picture.motion_val[list][b_xy + h->b_stride*left_block[2]];
- *(uint32_t*)h->mv_cache[list][scan8[0] - 1 + 3*8]= *(uint32_t*)s->current_picture.motion_val[list][b_xy + h->b_stride*left_block[3]];
- h->ref_cache[list][scan8[0] - 1 + 2*8]= s->current_picture.ref_index[list][b8_xy + h->b8_stride*(left_block[2]>>1)];
- h->ref_cache[list][scan8[0] - 1 + 3*8]= s->current_picture.ref_index[list][b8_xy + h->b8_stride*(left_block[3]>>1)];
- }else{
- *(uint32_t*)h->mv_cache [list][scan8[0] - 1 + 2*8]=
- *(uint32_t*)h->mv_cache [list][scan8[0] - 1 + 3*8]= 0;
- h->ref_cache[list][scan8[0] - 1 + 2*8]=
- h->ref_cache[list][scan8[0] - 1 + 3*8]= left_type[0] ? LIST_NOT_USED : PART_NOT_AVAILABLE;
- assert((!left_type[0]) == (!left_type[1]));
+ for(i=0; i<2; i++){
+ int cache_idx = scan8[0] - 1 + i*2*8;
+ if(USES_LIST(left_type[i], list)){
+ const int b_xy= h->mb2b_xy[left_xy[i]] + 3;
+ const int b8_xy= h->mb2b8_xy[left_xy[i]] + 1;
+ *(uint32_t*)h->mv_cache[list][cache_idx ]= *(uint32_t*)s->current_picture.motion_val[list][b_xy + h->b_stride*left_block[0+i*2]];
+ *(uint32_t*)h->mv_cache[list][cache_idx+8]= *(uint32_t*)s->current_picture.motion_val[list][b_xy + h->b_stride*left_block[1+i*2]];
+ h->ref_cache[list][cache_idx ]= s->current_picture.ref_index[list][b8_xy + h->b8_stride*(left_block[0+i*2]>>1)];
+ h->ref_cache[list][cache_idx+8]= s->current_picture.ref_index[list][b8_xy + h->b8_stride*(left_block[1+i*2]>>1)];
+ }else{
+ *(uint32_t*)h->mv_cache [list][cache_idx ]=
+ *(uint32_t*)h->mv_cache [list][cache_idx+8]= 0;
+ h->ref_cache[list][cache_idx ]=
+ h->ref_cache[list][cache_idx+8]= left_type[i] ? LIST_NOT_USED : PART_NOT_AVAILABLE;
+ }
}
- if(for_deblock || (IS_DIRECT(mb_type) && !h->direct_spatial_mv_pred))
+ if((for_deblock || (IS_DIRECT(mb_type) && !h->direct_spatial_mv_pred)) && !FRAME_MBAFF)
continue;
if(USES_LIST(topleft_type, list)){
h->ref_cache[list][scan8[0] + 4 - 1*8]= topright_type ? LIST_NOT_USED : PART_NOT_AVAILABLE;
}
+ if((IS_SKIP(mb_type) || IS_DIRECT(mb_type)) && !FRAME_MBAFF)
+ continue;
h->ref_cache[list][scan8[5 ]+1] =
h->ref_cache[list][scan8[7 ]+1] =
static const int8_t top [7]= {LEFT_DC_PRED8x8, 1,-1,-1};
static const int8_t left[7]= { TOP_DC_PRED8x8,-1, 2,-1,DC_128_PRED8x8};
- if(mode < 0 || mode > 6) {
+ if(mode > 6U) {
av_log(h->s.avctx, AV_LOG_ERROR, "out of range intra chroma pred mode at %d %d\n", s->mb_x, s->mb_y);
return -1;
}
const int top = h->intra4x4_pred_mode_cache[index8 - 8];
const int min= FFMIN(left, top);
- tprintf("mode:%d %d min:%d\n", left ,top, min);
+ tprintf(h->s.avctx, "mode:%d %d min:%d\n", left ,top, min);
if(min<0) return DC_PRED;
else return min;
if(i<64) i= (i+1)>>1;
- tprintf("pred_nnz L%X T%X n%d s%d P%X\n", left, top, n, scan8[n], i&31);
+ tprintf(h->s.avctx, "pred_nnz L%X T%X n%d s%d P%X\n", left, top, n, scan8[n], i&31);
return i&31;
}
static inline int fetch_diagonal_mv(H264Context *h, const int16_t **C, int i, int list, int part_width){
const int topright_ref= h->ref_cache[list][ i - 8 + part_width ];
+ MpegEncContext *s = &h->s;
/* there is no consistent mapping of mvs to neighboring locations that will
* make mbaff happy, so we can't move all this logic to fill_caches */
if(FRAME_MBAFF){
- MpegEncContext *s = &h->s;
- const int *mb_types = s->current_picture_ptr->mb_type;
+ const uint32_t *mb_types = s->current_picture_ptr->mb_type;
const int16_t *mv;
*(uint32_t*)h->mv_cache[list][scan8[0]-2] = 0;
*C = h->mv_cache[list][scan8[0]-2];
*C= h->mv_cache[list][ i - 8 + part_width ];
return topright_ref;
}else{
- tprintf("topright MV not available\n");
+ tprintf(s->avctx, "topright MV not available\n");
*C= h->mv_cache[list][ i - 8 - 1 ];
return h->ref_cache[list][ i - 8 - 1 ];
diagonal_ref= fetch_diagonal_mv(h, &C, index8, list, part_width);
match_count= (diagonal_ref==ref) + (top_ref==ref) + (left_ref==ref);
- tprintf("pred_motion match_count=%d\n", match_count);
+ tprintf(h->s.avctx, "pred_motion match_count=%d\n", match_count);
if(match_count > 1){ //most common
*mx= mid_pred(A[0], B[0], C[0]);
*my= mid_pred(A[1], B[1], C[1]);
}
}
- tprintf("pred_motion (%2d %2d %2d) (%2d %2d %2d) (%2d %2d %2d) -> (%2d %2d %2d) at %2d %2d %d list %d\n", top_ref, B[0], B[1], diagonal_ref, C[0], C[1], left_ref, A[0], A[1], ref, *mx, *my, h->s.mb_x, h->s.mb_y, n, list);
+ tprintf(h->s.avctx, "pred_motion (%2d %2d %2d) (%2d %2d %2d) (%2d %2d %2d) -> (%2d %2d %2d) at %2d %2d %d list %d\n", top_ref, B[0], B[1], diagonal_ref, C[0], C[1], left_ref, A[0], A[1], ref, *mx, *my, h->s.mb_x, h->s.mb_y, n, list);
}
/**
const int top_ref= h->ref_cache[list][ scan8[0] - 8 ];
const int16_t * const B= h->mv_cache[list][ scan8[0] - 8 ];
- tprintf("pred_16x8: (%2d %2d %2d) at %2d %2d %d list %d\n", top_ref, B[0], B[1], h->s.mb_x, h->s.mb_y, n, list);
+ tprintf(h->s.avctx, "pred_16x8: (%2d %2d %2d) at %2d %2d %d list %d\n", top_ref, B[0], B[1], h->s.mb_x, h->s.mb_y, n, list);
if(top_ref == ref){
*mx= B[0];
const int left_ref= h->ref_cache[list][ scan8[8] - 1 ];
const int16_t * const A= h->mv_cache[list][ scan8[8] - 1 ];
- tprintf("pred_16x8: (%2d %2d %2d) at %2d %2d %d list %d\n", left_ref, A[0], A[1], h->s.mb_x, h->s.mb_y, n, list);
+ tprintf(h->s.avctx, "pred_16x8: (%2d %2d %2d) at %2d %2d %d list %d\n", left_ref, A[0], A[1], h->s.mb_x, h->s.mb_y, n, list);
if(left_ref == ref){
*mx= A[0];
const int left_ref= h->ref_cache[list][ scan8[0] - 1 ];
const int16_t * const A= h->mv_cache[list][ scan8[0] - 1 ];
- tprintf("pred_8x16: (%2d %2d %2d) at %2d %2d %d list %d\n", left_ref, A[0], A[1], h->s.mb_x, h->s.mb_y, n, list);
+ tprintf(h->s.avctx, "pred_8x16: (%2d %2d %2d) at %2d %2d %d list %d\n", left_ref, A[0], A[1], h->s.mb_x, h->s.mb_y, n, list);
if(left_ref == ref){
*mx= A[0];
diagonal_ref= fetch_diagonal_mv(h, &C, scan8[4], list, 2);
- tprintf("pred_8x16: (%2d %2d %2d) at %2d %2d %d list %d\n", diagonal_ref, C[0], C[1], h->s.mb_x, h->s.mb_y, n, list);
+ tprintf(h->s.avctx, "pred_8x16: (%2d %2d %2d) at %2d %2d %d list %d\n", diagonal_ref, C[0], C[1], h->s.mb_x, h->s.mb_y, n, list);
if(diagonal_ref == ref){
*mx= C[0];
const int top_ref = h->ref_cache[0][ scan8[0] - 8 ];
const int left_ref= h->ref_cache[0][ scan8[0] - 1 ];
- tprintf("pred_pskip: (%d) (%d) at %2d %2d\n", top_ref, left_ref, h->s.mb_x, h->s.mb_y);
+ tprintf(h->s.avctx, "pred_pskip: (%d) (%d) at %2d %2d\n", top_ref, left_ref, h->s.mb_x, h->s.mb_y);
if(top_ref == PART_NOT_AVAILABLE || left_ref == PART_NOT_AVAILABLE
|| (top_ref == 0 && *(uint32_t*)h->mv_cache[0][ scan8[0] - 8 ] == 0)
int i;
for(i=0; i<h->ref_count[0]; i++){
int poc0 = h->ref_list[0][i].poc;
- int td = clip(poc1 - poc0, -128, 127);
+ int td = av_clip(poc1 - poc0, -128, 127);
if(td == 0 /* FIXME || pic0 is a long-term ref */){
h->dist_scale_factor[i] = 256;
}else{
- int tb = clip(poc - poc0, -128, 127);
- int tx = (16384 + (ABS(td) >> 1)) / td;
- h->dist_scale_factor[i] = clip((tb*tx + 32) >> 6, -1024, 1023);
+ int tb = av_clip(poc - poc0, -128, 127);
+ int tx = (16384 + (FFABS(td) >> 1)) / td;
+ h->dist_scale_factor[i] = av_clip((tb*tx + 32) >> 6, -1024, 1023);
}
}
if(FRAME_MBAFF){
const int8_t *l1ref0 = &h->ref_list[1][0].ref_index[0][b8_xy];
const int8_t *l1ref1 = &h->ref_list[1][0].ref_index[1][b8_xy];
const int is_b8x8 = IS_8X8(*mb_type);
- int sub_mb_type;
+ unsigned int sub_mb_type;
int i8, i4;
#define MB_TYPE_16x16_OR_INTRA (MB_TYPE_16x16|MB_TYPE_INTRA4x4|MB_TYPE_INTRA16x16|MB_TYPE_INTRA_PCM)
if(MB_FIELD)
*mb_type |= MB_TYPE_INTERLACED;
- tprintf("mb_type = %08x, sub_mb_type = %08x, is_b8x8 = %d, mb_type_col = %08x\n", *mb_type, sub_mb_type, is_b8x8, mb_type_col);
+ tprintf(s->avctx, "mb_type = %08x, sub_mb_type = %08x, is_b8x8 = %d, mb_type_col = %08x\n", *mb_type, sub_mb_type, is_b8x8, mb_type_col);
if(h->direct_spatial_mv_pred){
int ref[2];
}
if(IS_16X16(*mb_type)){
+ int a=0, b=0;
+
fill_rectangle(&h->ref_cache[0][scan8[0]], 4, 4, 8, (uint8_t)ref[0], 1);
fill_rectangle(&h->ref_cache[1][scan8[0]], 4, 4, 8, (uint8_t)ref[1], 1);
if(!IS_INTRA(mb_type_col)
- && ( (l1ref0[0] == 0 && ABS(l1mv0[0][0]) <= 1 && ABS(l1mv0[0][1]) <= 1)
- || (l1ref0[0] < 0 && l1ref1[0] == 0 && ABS(l1mv1[0][0]) <= 1 && ABS(l1mv1[0][1]) <= 1
+ && ( (l1ref0[0] == 0 && FFABS(l1mv0[0][0]) <= 1 && FFABS(l1mv0[0][1]) <= 1)
+ || (l1ref0[0] < 0 && l1ref1[0] == 0 && FFABS(l1mv1[0][0]) <= 1 && FFABS(l1mv1[0][1]) <= 1
&& (h->x264_build>33 || !h->x264_build)))){
if(ref[0] > 0)
- fill_rectangle(&h->mv_cache[0][scan8[0]], 4, 4, 8, pack16to32(mv[0][0],mv[0][1]), 4);
- else
- fill_rectangle(&h->mv_cache[0][scan8[0]], 4, 4, 8, 0, 4);
+ a= pack16to32(mv[0][0],mv[0][1]);
if(ref[1] > 0)
- fill_rectangle(&h->mv_cache[1][scan8[0]], 4, 4, 8, pack16to32(mv[1][0],mv[1][1]), 4);
- else
- fill_rectangle(&h->mv_cache[1][scan8[0]], 4, 4, 8, 0, 4);
+ b= pack16to32(mv[1][0],mv[1][1]);
}else{
- fill_rectangle(&h->mv_cache[0][scan8[0]], 4, 4, 8, pack16to32(mv[0][0],mv[0][1]), 4);
- fill_rectangle(&h->mv_cache[1][scan8[0]], 4, 4, 8, pack16to32(mv[1][0],mv[1][1]), 4);
+ a= pack16to32(mv[0][0],mv[0][1]);
+ b= pack16to32(mv[1][0],mv[1][1]);
}
+ fill_rectangle(&h->mv_cache[0][scan8[0]], 4, 4, 8, a, 4);
+ fill_rectangle(&h->mv_cache[1][scan8[0]], 4, 4, 8, b, 4);
}else{
for(i8=0; i8<4; i8++){
const int x8 = i8&1;
const int16_t (*l1mv)[2]= l1ref0[x8 + y8*h->b8_stride] == 0 ? l1mv0 : l1mv1;
if(IS_SUB_8X8(sub_mb_type)){
const int16_t *mv_col = l1mv[x8*3 + y8*3*h->b_stride];
- if(ABS(mv_col[0]) <= 1 && ABS(mv_col[1]) <= 1){
+ if(FFABS(mv_col[0]) <= 1 && FFABS(mv_col[1]) <= 1){
if(ref[0] == 0)
fill_rectangle(&h->mv_cache[0][scan8[i8*4]], 2, 2, 8, 0, 4);
if(ref[1] == 0)
}else
for(i4=0; i4<4; i4++){
const int16_t *mv_col = l1mv[x8*2 + (i4&1) + (y8*2 + (i4>>1))*h->b_stride];
- if(ABS(mv_col[0]) <= 1 && ABS(mv_col[1]) <= 1){
+ if(FFABS(mv_col[0]) <= 1 && FFABS(mv_col[1]) <= 1){
if(ref[0] == 0)
*(uint32_t*)h->mv_cache[0][scan8[i8*4+i4]] = 0;
if(ref[1] == 0)
/* one-to-one mv scaling */
if(IS_16X16(*mb_type)){
+ int ref, mv0, mv1;
+
fill_rectangle(&h->ref_cache[1][scan8[0]], 4, 4, 8, 0, 1);
if(IS_INTRA(mb_type_col)){
- fill_rectangle(&h->ref_cache[0][scan8[0]], 4, 4, 8, 0, 1);
- fill_rectangle(&h-> mv_cache[0][scan8[0]], 4, 4, 8, 0, 4);
- fill_rectangle(&h-> mv_cache[1][scan8[0]], 4, 4, 8, 0, 4);
+ ref=mv0=mv1=0;
}else{
const int ref0 = l1ref0[0] >= 0 ? map_col_to_list0[0][l1ref0[0]]
: map_col_to_list0[1][l1ref1[0]];
int mv_l0[2];
mv_l0[0] = (scale * mv_col[0] + 128) >> 8;
mv_l0[1] = (scale * mv_col[1] + 128) >> 8;
- fill_rectangle(&h->ref_cache[0][scan8[0]], 4, 4, 8, ref0, 1);
- fill_rectangle(&h-> mv_cache[0][scan8[0]], 4, 4, 8, pack16to32(mv_l0[0],mv_l0[1]), 4);
- fill_rectangle(&h-> mv_cache[1][scan8[0]], 4, 4, 8, pack16to32(mv_l0[0]-mv_col[0],mv_l0[1]-mv_col[1]), 4);
+ ref= ref0;
+ mv0= pack16to32(mv_l0[0],mv_l0[1]);
+ mv1= pack16to32(mv_l0[0]-mv_col[0],mv_l0[1]-mv_col[1]);
}
+ fill_rectangle(&h->ref_cache[0][scan8[0]], 4, 4, 8, ref, 1);
+ fill_rectangle(&h-> mv_cache[0][scan8[0]], 4, 4, 8, mv0, 4);
+ fill_rectangle(&h-> mv_cache[1][scan8[0]], 4, 4, 8, mv1, 4);
}else{
for(i8=0; i8<4; i8++){
const int x8 = i8&1;
if(!USES_LIST(mb_type, 0))
fill_rectangle(&s->current_picture.ref_index[0][b8_xy], 2, 2, h->b8_stride, (uint8_t)LIST_NOT_USED, 1);
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
int y;
if(!USES_LIST(mb_type, list))
continue;
*(uint64_t*)s->current_picture.motion_val[list][b_xy + 2 + y*h->b_stride]= *(uint64_t*)h->mv_cache[list][scan8[0]+2 + 8*y];
}
if( h->pps.cabac ) {
+ if(IS_SKIP(mb_type))
+ fill_rectangle(h->mvd_table[list][b_xy], 4, 4, h->b_stride, 0, 4);
+ else
for(y=0; y<4; y++){
*(uint64_t*)h->mvd_table[list][b_xy + 0 + y*h->b_stride]= *(uint64_t*)h->mvd_cache[list][scan8[0]+0 + 8*y];
*(uint64_t*)h->mvd_table[list][b_xy + 2 + y*h->b_stride]= *(uint64_t*)h->mvd_cache[list][scan8[0]+2 + 8*y];
}
{
- uint8_t *ref_index = &s->current_picture.ref_index[list][b8_xy];
+ int8_t *ref_index = &s->current_picture.ref_index[list][b8_xy];
ref_index[0+0*h->b8_stride]= h->ref_cache[list][scan8[0]];
ref_index[1+0*h->b8_stride]= h->ref_cache[list][scan8[4]];
ref_index[0+1*h->b8_stride]= h->ref_cache[list][scan8[8]];
h->rbsp_buffer= av_fast_realloc(h->rbsp_buffer, &h->rbsp_buffer_size, length);
dst= h->rbsp_buffer;
+ if (dst == NULL){
+ return NULL;
+ }
+
//printf("decoding esc\n");
si=di=0;
while(si<length){
*dst_length= di;
*consumed= si + 1;//+1 for the header
-//FIXME store exact number of bits in the getbitcontext (its needed for decoding)
+//FIXME store exact number of bits in the getbitcontext (it is needed for decoding)
return dst;
}
-#if 0
-/**
- * @param src the data which should be escaped
- * @param dst the target buffer, dst+1 == src is allowed as a special case
- * @param length the length of the src data
- * @param dst_length the length of the dst array
- * @returns length of escaped data in bytes or -1 if an error occured
- */
-static int encode_nal(H264Context *h, uint8_t *dst, uint8_t *src, int length, int dst_length){
- int i, escape_count, si, di;
- uint8_t *temp;
-
- assert(length>=0);
- assert(dst_length>0);
-
- dst[0]= (h->nal_ref_idc<<5) + h->nal_unit_type;
-
- if(length==0) return 1;
-
- escape_count= 0;
- for(i=0; i<length; i+=2){
- if(src[i]) continue;
- if(i>0 && src[i-1]==0)
- i--;
- if(i+2<length && src[i+1]==0 && src[i+2]<=3){
- escape_count++;
- i+=2;
- }
- }
-
- if(escape_count==0){
- if(dst+1 != src)
- memcpy(dst+1, src, length);
- return length + 1;
- }
-
- if(length + escape_count + 1> dst_length)
- return -1;
-
- //this should be damn rare (hopefully)
-
- h->rbsp_buffer= av_fast_realloc(h->rbsp_buffer, &h->rbsp_buffer_size, length + escape_count);
- temp= h->rbsp_buffer;
-//printf("encoding esc\n");
-
- si= 0;
- di= 0;
- while(si < length){
- if(si+2<length && src[si]==0 && src[si+1]==0 && src[si+2]<=3){
- temp[di++]= 0; si++;
- temp[di++]= 0; si++;
- temp[di++]= 3;
- temp[di++]= src[si++];
- }
- else
- temp[di++]= src[si++];
- }
- memcpy(dst+1, temp, length+escape_count);
-
- assert(di == length+escape_count);
-
- return di + 1;
-}
-
-/**
- * write 1,10,100,1000,... for alignment, yes its exactly inverse to mpeg4
- */
-static void encode_rbsp_trailing(PutBitContext *pb){
- int length;
- put_bits(pb, 1, 1);
- length= (-put_bits_count(pb))&7;
- if(length) put_bits(pb, length, 0);
-}
-#endif
-
/**
* identifies the exact end of the bitstream
* @return the length of the trailing, or 0 if damaged
*/
-static int decode_rbsp_trailing(uint8_t *src){
+static int decode_rbsp_trailing(H264Context *h, uint8_t *src){
int v= *src;
int r;
- tprintf("rbsp trailing %X\n", v);
+ tprintf(h->s.avctx, "rbsp trailing %X\n", v);
for(r=1; r<9; r++){
if(v&1) return r;
*/
static inline int get_chroma_qp(int chroma_qp_index_offset, int qscale){
- return chroma_qp[clip(qscale + chroma_qp_index_offset, 0, 51)];
-}
-
-
-#if 0
-static void h264_diff_dct_c(DCTELEM *block, uint8_t *src1, uint8_t *src2, int stride){
- int i;
- //FIXME try int temp instead of block
-
- for(i=0; i<4; i++){
- const int d0= src1[0 + i*stride] - src2[0 + i*stride];
- const int d1= src1[1 + i*stride] - src2[1 + i*stride];
- const int d2= src1[2 + i*stride] - src2[2 + i*stride];
- const int d3= src1[3 + i*stride] - src2[3 + i*stride];
- const int z0= d0 + d3;
- const int z3= d0 - d3;
- const int z1= d1 + d2;
- const int z2= d1 - d2;
-
- block[0 + 4*i]= z0 + z1;
- block[1 + 4*i]= 2*z3 + z2;
- block[2 + 4*i]= z0 - z1;
- block[3 + 4*i]= z3 - 2*z2;
- }
-
- for(i=0; i<4; i++){
- const int z0= block[0*4 + i] + block[3*4 + i];
- const int z3= block[0*4 + i] - block[3*4 + i];
- const int z1= block[1*4 + i] + block[2*4 + i];
- const int z2= block[1*4 + i] - block[2*4 + i];
-
- block[0*4 + i]= z0 + z1;
- block[1*4 + i]= 2*z3 + z2;
- block[2*4 + i]= z0 - z1;
- block[3*4 + i]= z3 - 2*z2;
- }
+ return chroma_qp[av_clip(qscale + chroma_qp_index_offset, 0, 51)];
}
-#endif
-//FIXME need to check that this doesnt overflow signed 32 bit for low qp, i am not sure, it's very close
-//FIXME check that gcc inlines this (and optimizes intra & seperate_dc stuff away)
-static inline int quantize_c(DCTELEM *block, uint8_t *scantable, int qscale, int intra, int seperate_dc){
+//FIXME need to check that this does not overflow signed 32 bit for low qp, i am not sure, it's very close
+//FIXME check that gcc inlines this (and optimizes intra & separate_dc stuff away)
+static inline int quantize_c(DCTELEM *block, uint8_t *scantable, int qscale, int intra, int separate_dc){
int i;
const int * const quant_table= quant_coeff[qscale];
const int bias= intra ? (1<<QUANT_SHIFT)/3 : (1<<QUANT_SHIFT)/6;
const unsigned int threshold2= (threshold1<<1);
int last_non_zero;
- if(seperate_dc){
+ if(separate_dc){
if(qscale<=18){
//avoid overflows
const int dc_bias= intra ? (1<<(QUANT_SHIFT-2))/3 : (1<<(QUANT_SHIFT-2))/6;
#define LOAD_TOP_RIGHT_EDGE\
- const int t4= topright[0];\
- const int t5= topright[1];\
- const int t6= topright[2];\
- const int t7= topright[3];\
+ const int av_unused t4= topright[0];\
+ const int av_unused t5= topright[1];\
+ const int av_unused t6= topright[2];\
+ const int av_unused t7= topright[3];\
#define LOAD_LEFT_EDGE\
- const int l0= src[-1+0*stride];\
- const int l1= src[-1+1*stride];\
- const int l2= src[-1+2*stride];\
- const int l3= src[-1+3*stride];\
+ const int av_unused l0= src[-1+0*stride];\
+ const int av_unused l1= src[-1+1*stride];\
+ const int av_unused l2= src[-1+2*stride];\
+ const int av_unused l3= src[-1+3*stride];\
#define LOAD_TOP_EDGE\
- const int t0= src[ 0-1*stride];\
- const int t1= src[ 1-1*stride];\
- const int t2= src[ 2-1*stride];\
- const int t3= src[ 3-1*stride];\
+ const int av_unused t0= src[ 0-1*stride];\
+ const int av_unused t1= src[ 1-1*stride];\
+ const int av_unused t2= src[ 2-1*stride];\
+ const int av_unused t3= src[ 3-1*stride];\
static void pred4x4_down_right_c(uint8_t *src, uint8_t *topright, int stride){
const int lt= src[-1-1*stride];
const int lt= src[-1-1*stride];
LOAD_TOP_EDGE
LOAD_LEFT_EDGE
- const __attribute__((unused)) int unu= l3;
src[0+0*stride]=
src[1+2*stride]=(lt + t0 + 1)>>1;
static void pred4x4_vertical_left_c(uint8_t *src, uint8_t *topright, int stride){
LOAD_TOP_EDGE
LOAD_TOP_RIGHT_EDGE
- const __attribute__((unused)) int unu= t7;
src[0+0*stride]=(t0 + t1 + 1)>>1;
src[1+0*stride]=
const int lt= src[-1-1*stride];
LOAD_TOP_EDGE
LOAD_LEFT_EDGE
- const __attribute__((unused)) int unu= t3;
src[0+0*stride]=
src[2+1*stride]=(lt + l0 + 1)>>1;
src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2;
}
-static void pred16x16_vertical_c(uint8_t *src, int stride){
+void ff_pred16x16_vertical_c(uint8_t *src, int stride){
int i;
const uint32_t a= ((uint32_t*)(src-stride))[0];
const uint32_t b= ((uint32_t*)(src-stride))[1];
}
}
-static void pred16x16_horizontal_c(uint8_t *src, int stride){
+void ff_pred16x16_horizontal_c(uint8_t *src, int stride){
int i;
for(i=0; i<16; i++){
}
}
-static void pred16x16_dc_c(uint8_t *src, int stride){
+void ff_pred16x16_dc_c(uint8_t *src, int stride){
int i, dc=0;
for(i=0;i<16; i++){
}
}
-static void pred16x16_left_dc_c(uint8_t *src, int stride){
+void ff_pred16x16_left_dc_c(uint8_t *src, int stride){
int i, dc=0;
for(i=0;i<16; i++){
}
}
-static void pred16x16_top_dc_c(uint8_t *src, int stride){
+void ff_pred16x16_top_dc_c(uint8_t *src, int stride){
int i, dc=0;
for(i=0;i<16; i++){
}
}
-static void pred16x16_128_dc_c(uint8_t *src, int stride){
+void ff_pred16x16_128_dc_c(uint8_t *src, int stride){
int i;
for(i=0; i<16; i++){
static inline void pred16x16_plane_compat_c(uint8_t *src, int stride, const int svq3){
int i, j, k;
int a;
- uint8_t *cm = cropTbl + MAX_NEG_CROP;
+ uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
const uint8_t * const src0 = src+7-stride;
const uint8_t *src1 = src+8*stride-1;
const uint8_t *src2 = src1-2*stride; // == src+6*stride-1;
}
}
-static void pred16x16_plane_c(uint8_t *src, int stride){
+void ff_pred16x16_plane_c(uint8_t *src, int stride){
pred16x16_plane_compat_c(src, stride, 0);
}
-static void pred8x8_vertical_c(uint8_t *src, int stride){
+void ff_pred8x8_vertical_c(uint8_t *src, int stride){
int i;
const uint32_t a= ((uint32_t*)(src-stride))[0];
const uint32_t b= ((uint32_t*)(src-stride))[1];
}
}
-static void pred8x8_horizontal_c(uint8_t *src, int stride){
+void ff_pred8x8_horizontal_c(uint8_t *src, int stride){
int i;
for(i=0; i<8; i++){
}
}
-static void pred8x8_128_dc_c(uint8_t *src, int stride){
+void ff_pred8x8_128_dc_c(uint8_t *src, int stride){
int i;
for(i=0; i<8; i++){
}
}
-static void pred8x8_left_dc_c(uint8_t *src, int stride){
+void ff_pred8x8_left_dc_c(uint8_t *src, int stride){
int i;
int dc0, dc2;
}
}
-static void pred8x8_top_dc_c(uint8_t *src, int stride){
+void ff_pred8x8_top_dc_c(uint8_t *src, int stride){
int i;
int dc0, dc1;
}
-static void pred8x8_dc_c(uint8_t *src, int stride){
+void ff_pred8x8_dc_c(uint8_t *src, int stride){
int i;
int dc0, dc1, dc2, dc3;
}
}
-static void pred8x8_plane_c(uint8_t *src, int stride){
+void ff_pred8x8_plane_c(uint8_t *src, int stride){
int j, k;
int a;
- uint8_t *cm = cropTbl + MAX_NEG_CROP;
+ uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
const uint8_t * const src0 = src+3-stride;
const uint8_t *src1 = src+4*stride-1;
const uint8_t *src2 = src1-2*stride; // == src+2*stride-1;
const int l0 = ((has_topleft ? SRC(-1,-1) : SRC(-1,0)) \
+ 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2; \
PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) \
- const int l7 attribute_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2
+ const int l7 av_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2
#define PT(x) \
const int t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
const int t0 = ((has_topleft ? SRC(-1,-1) : SRC(0,-1)) \
+ 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; \
PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) \
- const int t7 attribute_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \
+ const int t7 av_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \
+ 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2
#define PTR(x) \
const int pic_width = 16*s->mb_width;
const int pic_height = 16*s->mb_height >> MB_MBAFF;
- if(!pic->data[0])
+ if(!pic->data[0]) //FIXME this is unacceptable, some senseable error concealment must be done for missing reference frames
return;
if(mx&7) extra_width -= 3;
prefetch_motion(h, 1);
}
-static void decode_init_vlc(H264Context *h){
+static void decode_init_vlc(void){
static int done = 0;
if (!done) {
h->pred8x8l[TOP_DC_PRED ]= pred8x8l_top_dc_c;
h->pred8x8l[DC_128_PRED ]= pred8x8l_128_dc_c;
- h->pred8x8[DC_PRED8x8 ]= pred8x8_dc_c;
- h->pred8x8[VERT_PRED8x8 ]= pred8x8_vertical_c;
- h->pred8x8[HOR_PRED8x8 ]= pred8x8_horizontal_c;
- h->pred8x8[PLANE_PRED8x8 ]= pred8x8_plane_c;
- h->pred8x8[LEFT_DC_PRED8x8]= pred8x8_left_dc_c;
- h->pred8x8[TOP_DC_PRED8x8 ]= pred8x8_top_dc_c;
- h->pred8x8[DC_128_PRED8x8 ]= pred8x8_128_dc_c;
+ h->pred8x8[DC_PRED8x8 ]= ff_pred8x8_dc_c;
+ h->pred8x8[VERT_PRED8x8 ]= ff_pred8x8_vertical_c;
+ h->pred8x8[HOR_PRED8x8 ]= ff_pred8x8_horizontal_c;
+ h->pred8x8[PLANE_PRED8x8 ]= ff_pred8x8_plane_c;
+ h->pred8x8[LEFT_DC_PRED8x8]= ff_pred8x8_left_dc_c;
+ h->pred8x8[TOP_DC_PRED8x8 ]= ff_pred8x8_top_dc_c;
+ h->pred8x8[DC_128_PRED8x8 ]= ff_pred8x8_128_dc_c;
- h->pred16x16[DC_PRED8x8 ]= pred16x16_dc_c;
- h->pred16x16[VERT_PRED8x8 ]= pred16x16_vertical_c;
- h->pred16x16[HOR_PRED8x8 ]= pred16x16_horizontal_c;
- h->pred16x16[PLANE_PRED8x8 ]= pred16x16_plane_c;
- h->pred16x16[LEFT_DC_PRED8x8]= pred16x16_left_dc_c;
- h->pred16x16[TOP_DC_PRED8x8 ]= pred16x16_top_dc_c;
- h->pred16x16[DC_128_PRED8x8 ]= pred16x16_128_dc_c;
+ h->pred16x16[DC_PRED8x8 ]= ff_pred16x16_dc_c;
+ h->pred16x16[VERT_PRED8x8 ]= ff_pred16x16_vertical_c;
+ h->pred16x16[HOR_PRED8x8 ]= ff_pred16x16_horizontal_c;
+ h->pred16x16[PLANE_PRED8x8 ]= ff_pred16x16_plane_c;
+ h->pred16x16[LEFT_DC_PRED8x8]= ff_pred16x16_left_dc_c;
+ h->pred16x16[TOP_DC_PRED8x8 ]= ff_pred16x16_top_dc_c;
+ h->pred16x16[DC_128_PRED8x8 ]= ff_pred16x16_128_dc_c;
}
static void free_tables(H264Context *h){
+ int i;
av_freep(&h->intra4x4_pred_mode);
av_freep(&h->chroma_pred_mode_table);
av_freep(&h->cbp_table);
av_freep(&h->mb2b8_xy);
av_freep(&h->s.obmc_scratchpad);
+
+ for(i = 0; i < MAX_SPS_COUNT; i++)
+ av_freep(h->sps_buffers + i);
+
+ for(i = 0; i < MAX_PPS_COUNT; i++)
+ av_freep(h->pps_buffers + i);
}
static void init_dequant8_coeff_table(H264Context *h){
}
for(q=0; q<52; q++){
- int shift = div6[q];
- int idx = rem6[q];
+ int shift = ff_div6[q];
+ int idx = ff_rem6[q];
for(x=0; x<64; x++)
h->dequant8_coeff[i][q][transpose ? (x>>3)|((x&7)<<3) : x] =
((uint32_t)dequant8_coeff_init[idx][ dequant8_coeff_init_scan[((x>>1)&12) | (x&3)] ] *
continue;
for(q=0; q<52; q++){
- int shift = div6[q] + 2;
- int idx = rem6[q];
+ int shift = ff_div6[q] + 2;
+ int idx = ff_rem6[q];
for(x=0; x<16; x++)
h->dequant4_coeff[i][q][transpose ? (x>>2)|((x<<2)&0xF) : x] =
((uint32_t)dequant4_coeff_init[idx][(x&1) + ((x>>2)&1)] *
s->low_delay= 1;
avctx->pix_fmt= PIX_FMT_YUV420P;
- decode_init_vlc(h);
+ decode_init_vlc();
if(avctx->extradata_size > 0 && avctx->extradata &&
*(char *)avctx->extradata == 1){
return 0;
}
-static inline void backup_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize){
+static inline void backup_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize, int simple){
MpegEncContext * const s = &h->s;
int i;
*(uint64_t*)(h->top_borders[0][s->mb_x]+0)= *(uint64_t*)(src_y + 16*linesize);
*(uint64_t*)(h->top_borders[0][s->mb_x]+8)= *(uint64_t*)(src_y +8+16*linesize);
- if(!(s->flags&CODEC_FLAG_GRAY)){
+ if(simple || !(s->flags&CODEC_FLAG_GRAY)){
h->left_border[17 ]= h->top_borders[0][s->mb_x][16+7];
h->left_border[17+9]= h->top_borders[0][s->mb_x][24+7];
for(i=1; i<9; i++){
}
}
-static inline void xchg_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize, int xchg){
+static inline void xchg_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize, int xchg, int simple){
MpegEncContext * const s = &h->s;
int temp8, i;
uint64_t temp64;
}
}
- if(!(s->flags&CODEC_FLAG_GRAY)){
+ if(simple || !(s->flags&CODEC_FLAG_GRAY)){
if(deblock_left){
for(i = !deblock_top; i<9; i++){
XCHG(h->left_border[i+17 ], src_cb[i*uvlinesize], temp8, xchg);
int deblock_left = (s->mb_x > 0);
int deblock_top = (s->mb_y > 1);
- tprintf("xchg_pair_border: src_y:%p src_cb:%p src_cr:%p ls:%d uvls:%d\n", src_y, src_cb, src_cr, linesize, uvlinesize);
+ tprintf(s->avctx, "xchg_pair_border: src_y:%p src_cb:%p src_cr:%p ls:%d uvls:%d\n", src_y, src_cb, src_cr, linesize, uvlinesize);
src_y -= 2 * linesize + 1;
src_cb -= 2 * uvlinesize + 1;
}
}
-static void hl_decode_mb(H264Context *h){
+static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple){
MpegEncContext * const s = &h->s;
const int mb_x= s->mb_x;
const int mb_y= s->mb_y;
int i;
int *block_offset = &h->block_offset[0];
const unsigned int bottom = mb_y & 1;
- const int transform_bypass = (s->qscale == 0 && h->sps.transform_bypass);
+ const int transform_bypass = (s->qscale == 0 && h->sps.transform_bypass), is_h264 = (simple || s->codec_id == CODEC_ID_H264);
void (*idct_add)(uint8_t *dst, DCTELEM *block, int stride);
void (*idct_dc_add)(uint8_t *dst, DCTELEM *block, int stride);
- if(!s->decode)
- return;
-
dest_y = s->current_picture.data[0] + (mb_y * 16* s->linesize ) + mb_x * 16;
dest_cb = s->current_picture.data[1] + (mb_y * 8 * s->uvlinesize) + mb_x * 8;
dest_cr = s->current_picture.data[2] + (mb_y * 8 * s->uvlinesize) + mb_x * 8;
- if (MB_FIELD) {
+ s->dsp.prefetch(dest_y + (s->mb_x&3)*4*s->linesize + 64, s->linesize, 4);
+ s->dsp.prefetch(dest_cb + (s->mb_x&7)*s->uvlinesize + 64, dest_cr - dest_cb, 2);
+
+ if (!simple && MB_FIELD) {
linesize = h->mb_linesize = s->linesize * 2;
uvlinesize = h->mb_uvlinesize = s->uvlinesize * 2;
block_offset = &h->block_offset[24];
}
if(FRAME_MBAFF) {
int list;
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
if(!USES_LIST(mb_type, list))
continue;
if(IS_16X16(mb_type)){
idct_add = s->dsp.h264_idct_add;
}
- if(FRAME_MBAFF && h->deblocking_filter && IS_INTRA(mb_type)
+ if(!simple && FRAME_MBAFF && h->deblocking_filter && IS_INTRA(mb_type)
&& (!bottom || !IS_INTRA(s->current_picture.mb_type[mb_xy-s->mb_stride]))){
int mbt_y = mb_y&~1;
uint8_t *top_y = s->current_picture.data[0] + (mbt_y * 16* s->linesize ) + mb_x * 16;
xchg_pair_border(h, top_y, top_cb, top_cr, s->linesize, s->uvlinesize, 1);
}
- if (IS_INTRA_PCM(mb_type)) {
+ if (!simple && IS_INTRA_PCM(mb_type)) {
unsigned int x, y;
// The pixels are stored in h->mb array in the same order as levels,
}
} else {
if(IS_INTRA(mb_type)){
- if(h->deblocking_filter && !FRAME_MBAFF)
- xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 1);
+ if(h->deblocking_filter && (simple || !FRAME_MBAFF))
+ xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 1, simple);
- if(!(s->flags&CODEC_FLAG_GRAY)){
+ if(simple || !(s->flags&CODEC_FLAG_GRAY)){
h->pred8x8[ h->chroma_pred_mode ](dest_cb, uvlinesize);
h->pred8x8[ h->chroma_pred_mode ](dest_cr, uvlinesize);
}
if(IS_INTRA4x4(mb_type)){
- if(!s->encoding){
+ if(simple || !s->encoding){
if(IS_8x8DCT(mb_type)){
for(i=0; i<16; i+=4){
uint8_t * const ptr= dest_y + block_offset[i];
const int dir= h->intra4x4_pred_mode_cache[ scan8[i] ];
const int nnz = h->non_zero_count_cache[ scan8[i] ];
h->pred8x8l[ dir ](ptr, (h->topleft_samples_available<<i)&0x8000,
- (h->topright_samples_available<<(i+1))&0x8000, linesize);
+ (h->topright_samples_available<<i)&0x4000, linesize);
if(nnz){
if(nnz == 1 && h->mb[i*16])
idct_dc_add(ptr, h->mb + i*16, linesize);
h->pred4x4[ dir ](ptr, topright, linesize);
nnz = h->non_zero_count_cache[ scan8[i] ];
if(nnz){
- if(s->codec_id == CODEC_ID_H264){
+ if(is_h264){
if(nnz == 1 && h->mb[i*16])
idct_dc_add(ptr, h->mb + i*16, linesize);
else
}
}else{
h->pred16x16[ h->intra16x16_pred_mode ](dest_y , linesize);
- if(s->codec_id == CODEC_ID_H264){
+ if(is_h264){
if(!transform_bypass)
h264_luma_dc_dequant_idct_c(h->mb, s->qscale, h->dequant4_coeff[IS_INTRA(mb_type) ? 0:3][s->qscale][0]);
}else
svq3_luma_dc_dequant_idct_c(h->mb, s->qscale);
}
- if(h->deblocking_filter && !FRAME_MBAFF)
- xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 0);
- }else if(s->codec_id == CODEC_ID_H264){
+ if(h->deblocking_filter && (simple || !FRAME_MBAFF))
+ xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 0, simple);
+ }else if(is_h264){
hl_motion(h, dest_y, dest_cb, dest_cr,
- s->dsp.put_h264_qpel_pixels_tab, s->dsp.put_h264_chroma_pixels_tab,
- s->dsp.avg_h264_qpel_pixels_tab, s->dsp.avg_h264_chroma_pixels_tab,
+ s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab,
+ s->me.qpel_avg, s->dsp.avg_h264_chroma_pixels_tab,
s->dsp.weight_h264_pixels_tab, s->dsp.biweight_h264_pixels_tab);
}
if(!IS_INTRA4x4(mb_type)){
- if(s->codec_id == CODEC_ID_H264){
+ if(is_h264){
if(IS_INTRA16x16(mb_type)){
for(i=0; i<16; i++){
if(h->non_zero_count_cache[ scan8[i] ])
}
}
- if(!(s->flags&CODEC_FLAG_GRAY)){
+ if(simple || !(s->flags&CODEC_FLAG_GRAY)){
uint8_t *dest[2] = {dest_cb, dest_cr};
if(transform_bypass){
idct_add = idct_dc_add = s->dsp.add_pixels4;
chroma_dc_dequant_idct_c(h->mb + 16*16, h->chroma_qp, h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp][0]);
chroma_dc_dequant_idct_c(h->mb + 16*16+4*16, h->chroma_qp, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp][0]);
}
- if(s->codec_id == CODEC_ID_H264){
+ if(is_h264){
for(i=16; i<16+8; i++){
if(h->non_zero_count_cache[ scan8[i] ])
idct_add(dest[(i&4)>>2] + block_offset[i], h->mb + i*16, uvlinesize);
}
}
if(h->deblocking_filter) {
- if (FRAME_MBAFF) {
+ if (!simple && FRAME_MBAFF) {
//FIXME try deblocking one mb at a time?
// the reduction in load/storing mvs and such might outweigh the extra backup/xchg_border
const int mb_y = s->mb_y - 1;
// deblock a pair
// top
s->mb_y--;
- tprintf("call mbaff filter_mb mb_x:%d mb_y:%d pair_dest_y = %p, dest_y = %p\n", mb_x, mb_y, pair_dest_y, dest_y);
+ tprintf(h->s.avctx, "call mbaff filter_mb mb_x:%d mb_y:%d pair_dest_y = %p, dest_y = %p\n", mb_x, mb_y, pair_dest_y, dest_y);
fill_caches(h, mb_type_top, 1); //FIXME don't fill stuff which isn't used by filter_mb
h->chroma_qp = get_chroma_qp(h->pps.chroma_qp_index_offset, s->current_picture.qscale_table[mb_xy]);
filter_mb(h, mb_x, mb_y, pair_dest_y, pair_dest_cb, pair_dest_cr, linesize, uvlinesize);
// bottom
s->mb_y++;
- tprintf("call mbaff filter_mb\n");
+ tprintf(h->s.avctx, "call mbaff filter_mb\n");
fill_caches(h, mb_type_bottom, 1); //FIXME don't fill stuff which isn't used by filter_mb
h->chroma_qp = get_chroma_qp(h->pps.chroma_qp_index_offset, s->current_picture.qscale_table[mb_xy+s->mb_stride]);
filter_mb(h, mb_x, mb_y+1, dest_y, dest_cb, dest_cr, linesize, uvlinesize);
} else {
- tprintf("call filter_mb\n");
- backup_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize);
+ tprintf(h->s.avctx, "call filter_mb\n");
+ backup_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, simple);
fill_caches(h, mb_type, 1); //FIXME don't fill stuff which isn't used by filter_mb
- filter_mb(h, mb_x, mb_y, dest_y, dest_cb, dest_cr, linesize, uvlinesize);
+ filter_mb_fast(h, mb_x, mb_y, dest_y, dest_cb, dest_cr, linesize, uvlinesize);
}
}
}
+/**
+ * Process a macroblock; this case avoids checks for expensive uncommon cases.
+ */
+static void hl_decode_mb_simple(H264Context *h){
+ hl_decode_mb_internal(h, 1);
+}
+
+/**
+ * Process a macroblock; this handles edge cases, such as interlacing.
+ */
+static void av_noinline hl_decode_mb_complex(H264Context *h){
+ hl_decode_mb_internal(h, 0);
+}
+
+static void hl_decode_mb(H264Context *h){
+ MpegEncContext * const s = &h->s;
+ const int mb_x= s->mb_x;
+ const int mb_y= s->mb_y;
+ const int mb_xy= mb_x + mb_y*s->mb_stride;
+ const int mb_type= s->current_picture.mb_type[mb_xy];
+ int is_complex = FRAME_MBAFF || MB_FIELD || IS_INTRA_PCM(mb_type) || s->codec_id != CODEC_ID_H264 || (s->flags&CODEC_FLAG_GRAY) || s->encoding;
+
+ if(!s->decode)
+ return;
+
+ if (is_complex)
+ hl_decode_mb_complex(h);
+ else hl_decode_mb_simple(h);
+}
+
/**
* fills the default_ref_list.
*/
limit= best_poc;
sorted_short_ref[out_i]= *h->short_ref[best_i];
- tprintf("sorted poc: %d->%d poc:%d fn:%d\n", best_i, out_i, sorted_short_ref[out_i].poc, sorted_short_ref[out_i].frame_num);
+ tprintf(h->s.avctx, "sorted poc: %d->%d poc:%d fn:%d\n", best_i, out_i, sorted_short_ref[out_i].poc, sorted_short_ref[out_i].frame_num);
if (-1 == smallest_poc_greater_than_current) {
if (h->short_ref[best_i]->poc >= s->current_picture_ptr->poc) {
smallest_poc_greater_than_current = out_i;
if(s->picture_structure == PICT_FRAME){
if(h->slice_type==B_TYPE){
int list;
- tprintf("current poc: %d, smallest_poc_greater_than_current: %d\n", s->current_picture_ptr->poc, smallest_poc_greater_than_current);
+ tprintf(h->s.avctx, "current poc: %d, smallest_poc_greater_than_current: %d\n", s->current_picture_ptr->poc, smallest_poc_greater_than_current);
// find the largest poc
for(list=0; list<2; list++){
}
#ifdef TRACE
for (i=0; i<h->ref_count[0]; i++) {
- tprintf("List0: %s fn:%d 0x%p\n", (h->default_ref_list[0][i].long_ref ? "LT" : "ST"), h->default_ref_list[0][i].pic_id, h->default_ref_list[0][i].data[0]);
+ tprintf(h->s.avctx, "List0: %s fn:%d 0x%p\n", (h->default_ref_list[0][i].long_ref ? "LT" : "ST"), h->default_ref_list[0][i].pic_id, h->default_ref_list[0][i].data[0]);
}
if(h->slice_type==B_TYPE){
for (i=0; i<h->ref_count[1]; i++) {
- tprintf("List1: %s fn:%d 0x%p\n", (h->default_ref_list[1][i].long_ref ? "LT" : "ST"), h->default_ref_list[1][i].pic_id, h->default_ref_list[0][i].data[0]);
+ tprintf(h->s.avctx, "List1: %s fn:%d 0x%p\n", (h->default_ref_list[1][i].long_ref ? "LT" : "ST"), h->default_ref_list[1][i].pic_id, h->default_ref_list[0][i].data[0]);
}
}
#endif
print_long_term(h);
if(h->slice_type==I_TYPE || h->slice_type==SI_TYPE) return 0; //FIXME move before func
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
memcpy(h->ref_list[list], h->default_ref_list[list], sizeof(Picture)*h->ref_count[list]);
if(get_bits1(&s->gb)){
int pred= h->curr_pic_num;
for(index=0; ; index++){
- int reordering_of_pic_nums_idc= get_ue_golomb(&s->gb);
- int pic_id;
+ unsigned int reordering_of_pic_nums_idc= get_ue_golomb(&s->gb);
+ unsigned int pic_id;
int i;
Picture *ref = NULL;
if(reordering_of_pic_nums_idc<3){
if(reordering_of_pic_nums_idc<2){
- const int abs_diff_pic_num= get_ue_golomb(&s->gb) + 1;
+ const unsigned int abs_diff_pic_num= get_ue_golomb(&s->gb) + 1;
if(abs_diff_pic_num >= h->max_pic_num){
av_log(h->s.avctx, AV_LOG_ERROR, "abs_diff_pic_num overflow\n");
ref->pic_id= ref->frame_num;
}else{
pic_id= get_ue_golomb(&s->gb); //long_term_pic_idx
+ if(pic_id>31){
+ av_log(h->s.avctx, AV_LOG_ERROR, "long_term_pic_idx overflow\n");
+ return -1;
+ }
ref = h->long_ref[pic_id];
- ref->pic_id= pic_id;
- assert(ref->reference == 3);
- assert(ref->long_ref);
- i=0;
+ if(ref){
+ ref->pic_id= pic_id;
+ assert(ref->reference == 3);
+ assert(ref->long_ref);
+ i=0;
+ }else{
+ i=-1;
+ }
}
if (i < 0) {
}
}
}
-
- if(h->slice_type!=B_TYPE) break;
}
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
for(index= 0; index < h->ref_count[list]; index++){
if(!h->ref_list[list][index].data[0])
h->ref_list[list][index]= s->current_picture;
}
- if(h->slice_type!=B_TYPE) break;
}
if(h->slice_type==B_TYPE && !h->direct_spatial_mv_pred)
return 0;
}
-static int fill_mbaff_ref_list(H264Context *h){
+static void fill_mbaff_ref_list(H264Context *h){
int list, i, j;
- for(list=0; list<2; list++){
+ for(list=0; list<2; list++){ //FIXME try list_count
for(i=0; i<h->ref_count[list]; i++){
Picture *frame = &h->ref_list[list][i];
Picture *field = &h->ref_list[list][16+2*i];
int poc0 = h->ref_list[0][ref0].poc;
for(ref1=0; ref1 < h->ref_count[1]; ref1++){
int poc1 = h->ref_list[1][ref1].poc;
- int td = clip(poc1 - poc0, -128, 127);
+ int td = av_clip(poc1 - poc0, -128, 127);
if(td){
- int tb = clip(cur_poc - poc0, -128, 127);
- int tx = (16384 + (ABS(td) >> 1)) / td;
- int dist_scale_factor = clip((tb*tx + 32) >> 6, -1024, 1023) >> 2;
+ int tb = av_clip(cur_poc - poc0, -128, 127);
+ int tx = (16384 + (FFABS(td) >> 1)) / td;
+ int dist_scale_factor = av_clip((tb*tx + 32) >> 6, -1024, 1023) >> 2;
if(dist_scale_factor < -64 || dist_scale_factor > 128)
h->implicit_weight[ref0][ref1] = 32;
else
if(pic) unreference_pic(h, pic);
h->long_ref[ mmco[i].long_index ]= remove_short(h, mmco[i].short_frame_num);
- h->long_ref[ mmco[i].long_index ]->long_ref=1;
- h->long_ref_count++;
+ if (h->long_ref[ mmco[i].long_index ]){
+ h->long_ref[ mmco[i].long_index ]->long_ref=1;
+ h->long_ref_count++;
+ }
break;
case MMCO_LONG2UNUSED:
pic= remove_long(h, mmco[i].long_index);
case MMCO_RESET:
while(h->short_ref_count){
pic= remove_short(h, h->short_ref[0]->frame_num);
- unreference_pic(h, pic);
+ if(pic) unreference_pic(h, pic);
}
for(j = 0; j < 16; j++) {
pic= remove_long(h, j);
}*/
}
if(opcode==MMCO_SHORT2LONG || opcode==MMCO_LONG2UNUSED || opcode==MMCO_LONG || opcode==MMCO_SET_MAX_LONG){
- h->mmco[i].long_index= get_ue_golomb(&s->gb);
- if(/*h->mmco[i].long_index >= h->long_ref_count || h->long_ref[ h->mmco[i].long_index ] == NULL*/ h->mmco[i].long_index >= 16){
+ unsigned int long_index= get_ue_golomb(&s->gb);
+ if(/*h->mmco[i].long_index >= h->long_ref_count || h->long_ref[ h->mmco[i].long_index ] == NULL*/ long_index >= 16){
av_log(h->s.avctx, AV_LOG_ERROR, "illegal long ref in memory management control operation %d\n", opcode);
return -1;
}
+ h->mmco[i].long_index= long_index;
}
- if(opcode > MMCO_LONG){
+ if(opcode > (unsigned)MMCO_LONG){
av_log(h->s.avctx, AV_LOG_ERROR, "illegal memory management control operation %d\n", opcode);
return -1;
}
*/
static int decode_slice_header(H264Context *h){
MpegEncContext * const s = &h->s;
- int first_mb_in_slice, pps_id;
+ unsigned int first_mb_in_slice;
+ unsigned int pps_id;
int num_ref_idx_active_override_flag;
static const uint8_t slice_type_map[5]= {P_TYPE, B_TYPE, I_TYPE, SP_TYPE, SI_TYPE};
- int slice_type;
+ unsigned int slice_type, tmp;
int default_ref_list_done = 0;
s->current_picture.reference= h->nal_ref_idc != 0;
first_mb_in_slice= get_ue_golomb(&s->gb);
+ if((s->flags2 & CODEC_FLAG2_CHUNKS) && first_mb_in_slice == 0){
+ h->slice_num = 0;
+ s->current_picture_ptr= NULL;
+ }
+
slice_type= get_ue_golomb(&s->gb);
if(slice_type > 9){
av_log(h->s.avctx, AV_LOG_ERROR, "slice type too large (%d) at %d %d\n", h->slice_type, s->mb_x, s->mb_y);
s->pict_type= h->slice_type; // to make a few old func happy, it's wrong though
pps_id= get_ue_golomb(&s->gb);
- if(pps_id>255){
+ if(pps_id>=MAX_PPS_COUNT){
av_log(h->s.avctx, AV_LOG_ERROR, "pps_id out of range\n");
return -1;
}
- h->pps= h->pps_buffer[pps_id];
- if(h->pps.slice_group_count == 0){
+ if(!h->pps_buffers[pps_id]) {
av_log(h->s.avctx, AV_LOG_ERROR, "non existing PPS referenced\n");
return -1;
}
+ h->pps= *h->pps_buffers[pps_id];
- h->sps= h->sps_buffer[ h->pps.sps_id ];
- if(h->sps.log2_max_frame_num == 0){
+ if(!h->sps_buffers[h->pps.sps_id]) {
av_log(h->s.avctx, AV_LOG_ERROR, "non existing SPS referenced\n");
return -1;
}
+ h->sps = *h->sps_buffers[h->pps.sps_id];
if(h->dequant_coeff_pps != pps_id){
h->dequant_coeff_pps = pps_id;
h->mb_aff_frame = h->sps.mb_aff;
}
}
-
- s->resync_mb_x = s->mb_x = first_mb_in_slice % s->mb_width;
- s->resync_mb_y = s->mb_y = (first_mb_in_slice / s->mb_width) << h->mb_aff_frame;
- if(s->mb_y >= s->mb_height){
+ assert(s->mb_num == s->mb_width * s->mb_height);
+ if(first_mb_in_slice << h->mb_aff_frame >= s->mb_num ||
+ first_mb_in_slice >= s->mb_num){
+ av_log(h->s.avctx, AV_LOG_ERROR, "first_mb_in_slice overflow\n");
return -1;
}
+ s->resync_mb_x = s->mb_x = first_mb_in_slice % s->mb_width;
+ s->resync_mb_y = s->mb_y = (first_mb_in_slice / s->mb_width) << h->mb_aff_frame;
+ assert(s->mb_y < s->mb_height);
if(s->picture_structure==PICT_FRAME){
h->curr_pic_num= h->frame_num;
if(h->slice_type==B_TYPE)
h->ref_count[1]= get_ue_golomb(&s->gb) + 1;
- if(h->ref_count[0] > 32 || h->ref_count[1] > 32){
+ if(h->ref_count[0]-1 > 32-1 || h->ref_count[1]-1 > 32-1){
av_log(h->s.avctx, AV_LOG_ERROR, "reference overflow\n");
+ h->ref_count[0]= h->ref_count[1]= 1;
return -1;
}
}
- }
+ if(h->slice_type == B_TYPE)
+ h->list_count= 2;
+ else
+ h->list_count= 1;
+ }else
+ h->list_count= 0;
if(!default_ref_list_done){
fill_default_ref_list(h);
if(FRAME_MBAFF)
fill_mbaff_ref_list(h);
- if( h->slice_type != I_TYPE && h->slice_type != SI_TYPE && h->pps.cabac )
- h->cabac_init_idc = get_ue_golomb(&s->gb);
+ if( h->slice_type != I_TYPE && h->slice_type != SI_TYPE && h->pps.cabac ){
+ tmp = get_ue_golomb(&s->gb);
+ if(tmp > 2){
+ av_log(s->avctx, AV_LOG_ERROR, "cabac_init_idc overflow\n");
+ return -1;
+ }
+ h->cabac_init_idc= tmp;
+ }
h->last_qscale_diff = 0;
- s->qscale = h->pps.init_qp + get_se_golomb(&s->gb);
- if(s->qscale<0 || s->qscale>51){
- av_log(s->avctx, AV_LOG_ERROR, "QP %d out of range\n", s->qscale);
+ tmp = h->pps.init_qp + get_se_golomb(&s->gb);
+ if(tmp>51){
+ av_log(s->avctx, AV_LOG_ERROR, "QP %u out of range\n", tmp);
return -1;
}
+ s->qscale= tmp;
h->chroma_qp = get_chroma_qp(h->pps.chroma_qp_index_offset, s->qscale);
//FIXME qscale / qp ... stuff
if(h->slice_type == SP_TYPE){
h->slice_alpha_c0_offset = 0;
h->slice_beta_offset = 0;
if( h->pps.deblocking_filter_parameters_present ) {
- h->deblocking_filter= get_ue_golomb(&s->gb);
+ tmp= get_ue_golomb(&s->gb);
+ if(tmp > 2){
+ av_log(s->avctx, AV_LOG_ERROR, "deblocking_filter_idc %u out of range\n", tmp);
+ return -1;
+ }
+ h->deblocking_filter= tmp;
if(h->deblocking_filter < 2)
h->deblocking_filter^= 1; // 1<->0
h->emu_edge_height= FRAME_MBAFF ? 0 : h->emu_edge_width;
if(s->avctx->debug&FF_DEBUG_PICT_INFO){
- av_log(h->s.avctx, AV_LOG_DEBUG, "slice:%d %s mb:%d %c pps:%d frame:%d poc:%d/%d ref:%d/%d qp:%d loop:%d:%d:%d weight:%d%s\n",
+ av_log(h->s.avctx, AV_LOG_DEBUG, "slice:%d %s mb:%d %c pps:%u frame:%d poc:%d/%d ref:%d/%d qp:%d loop:%d:%d:%d weight:%d%s\n",
h->slice_num,
(s->picture_structure==PICT_FRAME ? "F" : s->picture_structure==PICT_TOP_FIELD ? "T" : "B"),
first_mb_in_slice,
);
}
+ if((s->avctx->flags2 & CODEC_FLAG2_FAST) && !s->current_picture.reference){
+ s->me.qpel_put= s->dsp.put_2tap_qpel_pixels_tab;
+ s->me.qpel_avg= s->dsp.avg_2tap_qpel_pixels_tab;
+ }else{
+ s->me.qpel_put= s->dsp.put_h264_qpel_pixels_tab;
+ s->me.qpel_avg= s->dsp.avg_h264_qpel_pixels_tab;
+ }
+
return 0;
}
if(total_coeff==0)
return 0;
+ if(total_coeff > (unsigned)max_coeff) {
+ av_log(h->s.avctx, AV_LOG_ERROR, "corrupted macroblock %d %d (total_coeff=%d)\n", s->mb_x, s->mb_y, total_coeff);
+ return -1;
+ }
trailing_ones= coeff_token&3;
- tprintf("trailing:%d, total:%d\n", trailing_ones, total_coeff);
+ tprintf(h->s.avctx, "trailing:%d, total:%d\n", trailing_ones, total_coeff);
assert(total_coeff<=16);
for(i=0; i<trailing_ones; i++){
fill_caches(h, mb_type, 0); //FIXME check what is needed and what not ...
pred_direct_motion(h, &mb_type);
- if(h->pps.cabac){
- fill_rectangle(h->mvd_cache[0][scan8[0]], 4, 4, 8, 0, 4);
- fill_rectangle(h->mvd_cache[1][scan8[0]], 4, 4, 8, 0, 4);
- }
+ mb_type|= MB_TYPE_SKIP;
}
else
{
pred_pskip_motion(h, &mx, &my);
fill_rectangle(&h->ref_cache[0][scan8[0]], 4, 4, 8, 0, 1);
fill_rectangle( h->mv_cache[0][scan8[0]], 4, 4, 8, pack16to32(mx,my), 4);
- if(h->pps.cabac)
- fill_rectangle(h->mvd_cache[0][scan8[0]], 4, 4, 8, 0, 4);
}
write_back_motion(h, mb_type);
- s->current_picture.mb_type[mb_xy]= mb_type|MB_TYPE_SKIP;
+ s->current_picture.mb_type[mb_xy]= mb_type;
s->current_picture.qscale_table[mb_xy]= s->qscale;
h->slice_table[ mb_xy ]= h->slice_num;
h->prev_mb_skipped= 1;
static int decode_mb_cavlc(H264Context *h){
MpegEncContext * const s = &h->s;
const int mb_xy= s->mb_x + s->mb_y*s->mb_stride;
- int mb_type, partition_count, cbp;
+ int partition_count;
+ unsigned int mb_type, cbp;
int dct8x8_allowed= h->pps.transform_8x8_mode;
s->dsp.clear_blocks(h->mb); //FIXME avoid if already clear (move after skip handlong?
- tprintf("pic:%d mb:%d/%d\n", h->frame_num, s->mb_x, s->mb_y);
+ tprintf(s->avctx, "pic:%d mb:%d/%d\n", h->frame_num, s->mb_x, s->mb_y);
cbp = 0; /* avoid warning. FIXME: find a solution without slowing
down the code */
if(h->slice_type != I_TYPE && h->slice_type != SI_TYPE){
assert(h->slice_type == I_TYPE);
decode_intra_mb:
if(mb_type > 25){
- av_log(h->s.avctx, AV_LOG_ERROR, "mb_type %d in %c slice to large at %d %d\n", mb_type, av_get_pict_type_char(h->slice_type), s->mb_x, s->mb_y);
+ av_log(h->s.avctx, AV_LOG_ERROR, "mb_type %d in %c slice too large at %d %d\n", mb_type, av_get_pict_type_char(h->slice_type), s->mb_x, s->mb_y);
return -1;
}
partition_count=0;
if(IS_INTRA_PCM(mb_type)){
unsigned int x, y;
- // we assume these blocks are very rare so we dont optimize it
+ // We assume these blocks are very rare so we do not optimize it.
align_get_bits(&s->gb);
// The pixels are stored in the same order as levels in h->mb array.
for(y=0; y<16; y++){
const int index= 4*(y&3) + 32*((y>>2)&1) + 128*(y>>3);
for(x=0; x<16; x++){
- tprintf("LUMA ICPM LEVEL (%3d)\n", show_bits(&s->gb, 8));
+ tprintf(s->avctx, "LUMA ICPM LEVEL (%3d)\n", show_bits(&s->gb, 8));
h->mb[index + (x&3) + 16*((x>>2)&1) + 64*(x>>3)]= get_bits(&s->gb, 8);
}
}
for(y=0; y<8; y++){
const int index= 256 + 4*(y&3) + 32*(y>>2);
for(x=0; x<8; x++){
- tprintf("CHROMA U ICPM LEVEL (%3d)\n", show_bits(&s->gb, 8));
+ tprintf(s->avctx, "CHROMA U ICPM LEVEL (%3d)\n", show_bits(&s->gb, 8));
h->mb[index + (x&3) + 16*(x>>2)]= get_bits(&s->gb, 8);
}
}
for(y=0; y<8; y++){
const int index= 256 + 64 + 4*(y&3) + 32*(y>>2);
for(x=0; x<8; x++){
- tprintf("CHROMA V ICPM LEVEL (%3d)\n", show_bits(&s->gb, 8));
+ tprintf(s->avctx, "CHROMA V ICPM LEVEL (%3d)\n", show_bits(&s->gb, 8));
h->mb[index + (x&3) + 16*(x>>2)]= get_bits(&s->gb, 8);
}
}
//mb_pred
if(IS_INTRA(mb_type)){
+ int pred_mode;
// init_top_left_availability(h);
if(IS_INTRA4x4(mb_type)){
int i;
// fill_intra4x4_pred_table(h);
for(i=0; i<16; i+=di){
- const int mode_coded= !get_bits1(&s->gb);
- const int predicted_mode= pred_intra_mode(h, i);
- int mode;
+ int mode= pred_intra_mode(h, i);
- if(mode_coded){
+ if(!get_bits1(&s->gb)){
const int rem_mode= get_bits(&s->gb, 3);
- if(rem_mode<predicted_mode)
- mode= rem_mode;
- else
- mode= rem_mode + 1;
- }else{
- mode= predicted_mode;
+ mode = rem_mode + (rem_mode >= mode);
}
if(di==4)
if(h->intra16x16_pred_mode < 0)
return -1;
}
- h->chroma_pred_mode= get_ue_golomb(&s->gb);
- h->chroma_pred_mode= check_intra_pred_mode(h, h->chroma_pred_mode);
- if(h->chroma_pred_mode < 0)
+ pred_mode= check_intra_pred_mode(h, get_ue_golomb(&s->gb));
+ if(pred_mode < 0)
return -1;
+ h->chroma_pred_mode= pred_mode;
}else if(partition_count==4){
int i, j, sub_partition_count[4], list, ref[2][4];
for(i=0; i<4; i++){
h->sub_mb_type[i]= get_ue_golomb(&s->gb);
if(h->sub_mb_type[i] >=13){
- av_log(h->s.avctx, AV_LOG_ERROR, "B sub_mb_type %d out of range at %d %d\n", h->sub_mb_type[i], s->mb_x, s->mb_y);
+ av_log(h->s.avctx, AV_LOG_ERROR, "B sub_mb_type %u out of range at %d %d\n", h->sub_mb_type[i], s->mb_x, s->mb_y);
return -1;
}
sub_partition_count[i]= b_sub_mb_type_info[ h->sub_mb_type[i] ].partition_count;
for(i=0; i<4; i++){
h->sub_mb_type[i]= get_ue_golomb(&s->gb);
if(h->sub_mb_type[i] >=4){
- av_log(h->s.avctx, AV_LOG_ERROR, "P sub_mb_type %d out of range at %d %d\n", h->sub_mb_type[i], s->mb_x, s->mb_y);
+ av_log(h->s.avctx, AV_LOG_ERROR, "P sub_mb_type %u out of range at %d %d\n", h->sub_mb_type[i], s->mb_x, s->mb_y);
return -1;
}
sub_partition_count[i]= p_sub_mb_type_info[ h->sub_mb_type[i] ].partition_count;
}
}
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
int ref_count= IS_REF0(mb_type) ? 1 : h->ref_count[list];
- if(ref_count == 0) continue;
for(i=0; i<4; i++){
if(IS_DIRECT(h->sub_mb_type[i])) continue;
if(IS_DIR(h->sub_mb_type[i], 0, list)){
- ref[list][i] = get_te0_golomb(&s->gb, ref_count); //FIXME init to 0 before and skip?
+ unsigned int tmp = get_te0_golomb(&s->gb, ref_count); //FIXME init to 0 before and skip?
+ if(tmp>=ref_count){
+ av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", tmp);
+ return -1;
+ }
+ ref[list][i]= tmp;
}else{
//FIXME
ref[list][i] = -1;
if(dct8x8_allowed)
dct8x8_allowed = get_dct8x8_allowed(h);
- for(list=0; list<2; list++){
- const int ref_count= IS_REF0(mb_type) ? 1 : h->ref_count[list];
- if(ref_count == 0) continue;
-
+ for(list=0; list<h->list_count; list++){
for(i=0; i<4; i++){
if(IS_DIRECT(h->sub_mb_type[i])) {
h->ref_cache[list][ scan8[4*i] ] = h->ref_cache[list][ scan8[4*i]+1 ];
pred_motion(h, index, block_width, list, h->ref_cache[list][ scan8[index] ], &mx, &my);
mx += get_se_golomb(&s->gb);
my += get_se_golomb(&s->gb);
- tprintf("final mv:%d %d\n", mx, my);
+ tprintf(s->avctx, "final mv:%d %d\n", mx, my);
if(IS_SUB_8X8(sub_mb_type)){
- mv_cache[ 0 ][0]= mv_cache[ 1 ][0]=
+ mv_cache[ 1 ][0]=
mv_cache[ 8 ][0]= mv_cache[ 9 ][0]= mx;
- mv_cache[ 0 ][1]= mv_cache[ 1 ][1]=
+ mv_cache[ 1 ][1]=
mv_cache[ 8 ][1]= mv_cache[ 9 ][1]= my;
}else if(IS_SUB_8X4(sub_mb_type)){
- mv_cache[ 0 ][0]= mv_cache[ 1 ][0]= mx;
- mv_cache[ 0 ][1]= mv_cache[ 1 ][1]= my;
+ mv_cache[ 1 ][0]= mx;
+ mv_cache[ 1 ][1]= my;
}else if(IS_SUB_4X8(sub_mb_type)){
- mv_cache[ 0 ][0]= mv_cache[ 8 ][0]= mx;
- mv_cache[ 0 ][1]= mv_cache[ 8 ][1]= my;
- }else{
- assert(IS_SUB_4X4(sub_mb_type));
- mv_cache[ 0 ][0]= mx;
- mv_cache[ 0 ][1]= my;
+ mv_cache[ 8 ][0]= mx;
+ mv_cache[ 8 ][1]= my;
}
+ mv_cache[ 0 ][0]= mx;
+ mv_cache[ 0 ][1]= my;
}
}else{
uint32_t *p= (uint32_t *)&h->mv_cache[list][ scan8[4*i] ][0];
int list, mx, my, i;
//FIXME we should set ref_idx_l? to 0 if we use that later ...
if(IS_16X16(mb_type)){
- for(list=0; list<2; list++){
- if(h->ref_count[list]>0){
+ for(list=0; list<h->list_count; list++){
+ unsigned int val;
if(IS_DIR(mb_type, 0, list)){
- const int val= get_te0_golomb(&s->gb, h->ref_count[list]);
- fill_rectangle(&h->ref_cache[list][ scan8[0] ], 4, 4, 8, val, 1);
+ val= get_te0_golomb(&s->gb, h->ref_count[list]);
+ if(val >= h->ref_count[list]){
+ av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", val);
+ return -1;
+ }
}else
- fill_rectangle(&h->ref_cache[list][ scan8[0] ], 4, 4, 8, (LIST_NOT_USED&0xFF), 1);
- }
+ val= LIST_NOT_USED&0xFF;
+ fill_rectangle(&h->ref_cache[list][ scan8[0] ], 4, 4, 8, val, 1);
}
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
+ unsigned int val;
if(IS_DIR(mb_type, 0, list)){
pred_motion(h, 0, 4, list, h->ref_cache[list][ scan8[0] ], &mx, &my);
mx += get_se_golomb(&s->gb);
my += get_se_golomb(&s->gb);
- tprintf("final mv:%d %d\n", mx, my);
+ tprintf(s->avctx, "final mv:%d %d\n", mx, my);
- fill_rectangle(h->mv_cache[list][ scan8[0] ], 4, 4, 8, pack16to32(mx,my), 4);
+ val= pack16to32(mx,my);
}else
- fill_rectangle(h->mv_cache[list][ scan8[0] ], 4, 4, 8, 0, 4);
+ val=0;
+ fill_rectangle(h->mv_cache[list][ scan8[0] ], 4, 4, 8, val, 4);
}
}
else if(IS_16X8(mb_type)){
- for(list=0; list<2; list++){
- if(h->ref_count[list]>0){
+ for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
+ unsigned int val;
if(IS_DIR(mb_type, i, list)){
- const int val= get_te0_golomb(&s->gb, h->ref_count[list]);
- fill_rectangle(&h->ref_cache[list][ scan8[0] + 16*i ], 4, 2, 8, val, 1);
+ val= get_te0_golomb(&s->gb, h->ref_count[list]);
+ if(val >= h->ref_count[list]){
+ av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", val);
+ return -1;
+ }
}else
- fill_rectangle(&h->ref_cache[list][ scan8[0] + 16*i ], 4, 2, 8, (LIST_NOT_USED&0xFF), 1);
+ val= LIST_NOT_USED&0xFF;
+ fill_rectangle(&h->ref_cache[list][ scan8[0] + 16*i ], 4, 2, 8, val, 1);
}
- }
}
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
+ unsigned int val;
if(IS_DIR(mb_type, i, list)){
pred_16x8_motion(h, 8*i, list, h->ref_cache[list][scan8[0] + 16*i], &mx, &my);
mx += get_se_golomb(&s->gb);
my += get_se_golomb(&s->gb);
- tprintf("final mv:%d %d\n", mx, my);
+ tprintf(s->avctx, "final mv:%d %d\n", mx, my);
- fill_rectangle(h->mv_cache[list][ scan8[0] + 16*i ], 4, 2, 8, pack16to32(mx,my), 4);
+ val= pack16to32(mx,my);
}else
- fill_rectangle(h->mv_cache[list][ scan8[0] + 16*i ], 4, 2, 8, 0, 4);
+ val=0;
+ fill_rectangle(h->mv_cache[list][ scan8[0] + 16*i ], 4, 2, 8, val, 4);
}
}
}else{
assert(IS_8X16(mb_type));
- for(list=0; list<2; list++){
- if(h->ref_count[list]>0){
+ for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
+ unsigned int val;
if(IS_DIR(mb_type, i, list)){ //FIXME optimize
- const int val= get_te0_golomb(&s->gb, h->ref_count[list]);
- fill_rectangle(&h->ref_cache[list][ scan8[0] + 2*i ], 2, 4, 8, val, 1);
+ val= get_te0_golomb(&s->gb, h->ref_count[list]);
+ if(val >= h->ref_count[list]){
+ av_log(h->s.avctx, AV_LOG_ERROR, "ref %u overflow\n", val);
+ return -1;
+ }
}else
- fill_rectangle(&h->ref_cache[list][ scan8[0] + 2*i ], 2, 4, 8, (LIST_NOT_USED&0xFF), 1);
+ val= LIST_NOT_USED&0xFF;
+ fill_rectangle(&h->ref_cache[list][ scan8[0] + 2*i ], 2, 4, 8, val, 1);
}
- }
}
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
+ unsigned int val;
if(IS_DIR(mb_type, i, list)){
pred_8x16_motion(h, i*4, list, h->ref_cache[list][ scan8[0] + 2*i ], &mx, &my);
mx += get_se_golomb(&s->gb);
my += get_se_golomb(&s->gb);
- tprintf("final mv:%d %d\n", mx, my);
+ tprintf(s->avctx, "final mv:%d %d\n", mx, my);
- fill_rectangle(h->mv_cache[list][ scan8[0] + 2*i ], 2, 4, 8, pack16to32(mx,my), 4);
+ val= pack16to32(mx,my);
}else
- fill_rectangle(h->mv_cache[list][ scan8[0] + 2*i ], 2, 4, 8, 0, 4);
+ val=0;
+ fill_rectangle(h->mv_cache[list][ scan8[0] + 2*i ], 2, 4, 8, val, 4);
}
}
}
if(!IS_INTRA16x16(mb_type)){
cbp= get_ue_golomb(&s->gb);
if(cbp > 47){
- av_log(h->s.avctx, AV_LOG_ERROR, "cbp too large (%d) at %d %d\n", cbp, s->mb_x, s->mb_y);
+ av_log(h->s.avctx, AV_LOG_ERROR, "cbp too large (%u) at %d %d\n", cbp, s->mb_x, s->mb_y);
return -1;
}
else
cbp= golomb_to_inter_cbp[cbp];
}
+ h->cbp = cbp;
if(dct8x8_allowed && (cbp&15) && !IS_INTRA(mb_type)){
if(get_bits1(&s->gb))
ctx += 1;
}
- return get_cabac( &h->cabac, &h->cabac_state[70 + ctx] );
+ return get_cabac_noinline( &h->cabac, &h->cabac_state[70 + ctx] );
}
static int decode_cabac_intra_mb_type(H264Context *h, int ctx_base, int intra_slice) {
ctx++;
if( h->slice_table[mbb_xy] == h->slice_num && !IS_INTRA4x4( s->current_picture.mb_type[mbb_xy] ) )
ctx++;
- if( get_cabac( &h->cabac, &state[ctx] ) == 0 )
+ if( get_cabac_noinline( &h->cabac, &state[ctx] ) == 0 )
return 0; /* I4x4 */
state += 2;
}else{
- if( get_cabac( &h->cabac, &state[0] ) == 0 )
+ if( get_cabac_noinline( &h->cabac, &state[0] ) == 0 )
return 0; /* I4x4 */
}
return 25; /* PCM */
mb_type = 1; /* I16x16 */
- mb_type += 12 * get_cabac( &h->cabac, &state[1] ); /* cbp_luma != 0 */
- if( get_cabac( &h->cabac, &state[2] ) ) /* cbp_chroma */
- mb_type += 4 + 4 * get_cabac( &h->cabac, &state[2+intra_slice] );
- mb_type += 2 * get_cabac( &h->cabac, &state[3+intra_slice] );
- mb_type += 1 * get_cabac( &h->cabac, &state[3+2*intra_slice] );
+ mb_type += 12 * get_cabac_noinline( &h->cabac, &state[1] ); /* cbp_luma != 0 */
+ if( get_cabac_noinline( &h->cabac, &state[2] ) ) /* cbp_chroma */
+ mb_type += 4 + 4 * get_cabac_noinline( &h->cabac, &state[2+intra_slice] );
+ mb_type += 2 * get_cabac_noinline( &h->cabac, &state[3+intra_slice] );
+ mb_type += 1 * get_cabac_noinline( &h->cabac, &state[3+2*intra_slice] );
return mb_type;
}
if( h->slice_type == I_TYPE ) {
return decode_cabac_intra_mb_type(h, 3, 1);
} else if( h->slice_type == P_TYPE ) {
- if( get_cabac( &h->cabac, &h->cabac_state[14] ) == 0 ) {
+ if( get_cabac_noinline( &h->cabac, &h->cabac_state[14] ) == 0 ) {
/* P-type */
- if( get_cabac( &h->cabac, &h->cabac_state[15] ) == 0 ) {
+ if( get_cabac_noinline( &h->cabac, &h->cabac_state[15] ) == 0 ) {
/* P_L0_D16x16, P_8x8 */
- return 3 * get_cabac( &h->cabac, &h->cabac_state[16] );
+ return 3 * get_cabac_noinline( &h->cabac, &h->cabac_state[16] );
} else {
/* P_L0_D8x16, P_L0_D16x8 */
- return 2 - get_cabac( &h->cabac, &h->cabac_state[17] );
+ return 2 - get_cabac_noinline( &h->cabac, &h->cabac_state[17] );
}
} else {
return decode_cabac_intra_mb_type(h, 17, 0) + 5;
if( h->slice_table[mbb_xy] == h->slice_num && !IS_DIRECT( s->current_picture.mb_type[mbb_xy] ) )
ctx++;
- if( !get_cabac( &h->cabac, &h->cabac_state[27+ctx] ) )
+ if( !get_cabac_noinline( &h->cabac, &h->cabac_state[27+ctx] ) )
return 0; /* B_Direct_16x16 */
- if( !get_cabac( &h->cabac, &h->cabac_state[27+3] ) ) {
- return 1 + get_cabac( &h->cabac, &h->cabac_state[27+5] ); /* B_L[01]_16x16 */
+ if( !get_cabac_noinline( &h->cabac, &h->cabac_state[27+3] ) ) {
+ return 1 + get_cabac_noinline( &h->cabac, &h->cabac_state[27+5] ); /* B_L[01]_16x16 */
}
- bits = get_cabac( &h->cabac, &h->cabac_state[27+4] ) << 3;
- bits|= get_cabac( &h->cabac, &h->cabac_state[27+5] ) << 2;
- bits|= get_cabac( &h->cabac, &h->cabac_state[27+5] ) << 1;
- bits|= get_cabac( &h->cabac, &h->cabac_state[27+5] );
+ bits = get_cabac_noinline( &h->cabac, &h->cabac_state[27+4] ) << 3;
+ bits|= get_cabac_noinline( &h->cabac, &h->cabac_state[27+5] ) << 2;
+ bits|= get_cabac_noinline( &h->cabac, &h->cabac_state[27+5] ) << 1;
+ bits|= get_cabac_noinline( &h->cabac, &h->cabac_state[27+5] );
if( bits < 8 )
return bits + 3; /* B_Bi_16x16 through B_L1_L0_16x8 */
else if( bits == 13 ) {
else if( bits == 15 )
return 22; /* B_8x8 */
- bits= ( bits<<1 ) | get_cabac( &h->cabac, &h->cabac_state[27+5] );
+ bits= ( bits<<1 ) | get_cabac_noinline( &h->cabac, &h->cabac_state[27+5] );
return bits - 4; /* B_L0_Bi_* through B_Bi_Bi_* */
} else {
/* TODO SI/SP frames? */
if( h->slice_type == B_TYPE )
ctx += 13;
- return get_cabac( &h->cabac, &h->cabac_state[11+ctx] );
+ return get_cabac_noinline( &h->cabac, &h->cabac_state[11+ctx] );
}
static int decode_cabac_mb_intra4x4_pred_mode( H264Context *h, int pred_mode ) {
if( h->slice_table[mbb_xy] == h->slice_num && h->chroma_pred_mode_table[mbb_xy] != 0 )
ctx++;
- if( get_cabac( &h->cabac, &h->cabac_state[64+ctx] ) == 0 )
+ if( get_cabac_noinline( &h->cabac, &h->cabac_state[64+ctx] ) == 0 )
return 0;
- if( get_cabac( &h->cabac, &h->cabac_state[64+3] ) == 0 )
+ if( get_cabac_noinline( &h->cabac, &h->cabac_state[64+3] ) == 0 )
return 1;
- if( get_cabac( &h->cabac, &h->cabac_state[64+3] ) == 0 )
+ if( get_cabac_noinline( &h->cabac, &h->cabac_state[64+3] ) == 0 )
return 2;
else
return 3;
if( h->slice_table[h->top_mb_xy] == h->slice_num ) {
cbp_b = h->top_cbp;
- tprintf("cbp_b = top_cbp = %x\n", cbp_b);
+ tprintf(h->s.avctx, "cbp_b = top_cbp = %x\n", cbp_b);
}
for( i8x8 = 0; i8x8 < 4; i8x8++ ) {
cbp_a = cbp;
else if( h->slice_table[h->left_mb_xy[0]] == h->slice_num ) {
cbp_a = h->left_cbp;
- tprintf("cbp_a = left_cbp = %x\n", cbp_a);
+ tprintf(h->s.avctx, "cbp_a = left_cbp = %x\n", cbp_a);
}
if( y > 0 )
ctx = 0;
if( cbp_a > 0 ) ctx++;
if( cbp_b > 0 ) ctx += 2;
- if( get_cabac( &h->cabac, &h->cabac_state[77 + ctx] ) == 0 )
+ if( get_cabac_noinline( &h->cabac, &h->cabac_state[77 + ctx] ) == 0 )
return 0;
ctx = 4;
if( cbp_a == 2 ) ctx++;
if( cbp_b == 2 ) ctx += 2;
- return 1 + get_cabac( &h->cabac, &h->cabac_state[77 + ctx] );
+ return 1 + get_cabac_noinline( &h->cabac, &h->cabac_state[77 + ctx] );
}
static int decode_cabac_mb_dqp( H264Context *h) {
MpegEncContext * const s = &h->s;
if( h->last_qscale_diff != 0 )
ctx++;
- while( get_cabac( &h->cabac, &h->cabac_state[60 + ctx] ) ) {
+ while( get_cabac_noinline( &h->cabac, &h->cabac_state[60 + ctx] ) ) {
if( ctx < 2 )
ctx = 2;
else
}
static inline int decode_cabac_mb_transform_size( H264Context *h ) {
- return get_cabac( &h->cabac, &h->cabac_state[399 + h->neighbor_transform_size] );
+ return get_cabac_noinline( &h->cabac, &h->cabac_state[399 + h->neighbor_transform_size] );
}
static int decode_cabac_mb_ref( H264Context *h, int list, int n ) {
ctx = 4;
else
ctx = 5;
+ if(ref >= 32 /*h->ref_list[list]*/){
+ av_log(h->s.avctx, AV_LOG_ERROR, "overflow in decode_cabac_mb_ref\n");
+ return 0; //FIXME we should return -1 and check the return everywhere
+ }
}
return ref;
}
while( get_cabac_bypass( &h->cabac ) ) {
mvd += 1 << k;
k++;
+ if(k>24){
+ av_log(h->s.avctx, AV_LOG_ERROR, "overflow in decode_cabac_mb_mvd\n");
+ return INT_MIN;
+ }
}
while( k-- ) {
if( get_cabac_bypass( &h->cabac ) )
mvd += 1 << k;
}
}
- if( get_cabac_bypass( &h->cabac ) ) return -mvd;
- else return mvd;
+ return get_cabac_bypass_sign( &h->cabac, -mvd );
}
-static int inline get_cabac_cbf_ctx( H264Context *h, int cat, int idx ) {
+static inline int get_cabac_cbf_ctx( H264Context *h, int cat, int idx ) {
int nza, nzb;
int ctx = 0;
return ctx + 4 * cat;
}
+static const attribute_used uint8_t last_coeff_flag_offset_8x8[63] = {
+ 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
+ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
+ 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4,
+ 5, 5, 5, 5, 6, 6, 6, 6, 7, 7, 7, 7, 8, 8, 8
+};
+
static int decode_cabac_residual( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff) {
const int mb_xy = h->s.mb_x + h->s.mb_y*h->s.mb_stride;
static const int significant_coeff_flag_offset[2][6] = {
static const int coeff_abs_level_m1_offset[6] = {
227+0, 227+10, 227+20, 227+30, 227+39, 426
};
- static const int significant_coeff_flag_offset_8x8[2][63] = {
+ static const uint8_t significant_coeff_flag_offset_8x8[2][63] = {
{ 0, 1, 2, 3, 4, 5, 5, 4, 4, 3, 3, 4, 4, 4, 5, 5,
4, 4, 4, 4, 3, 3, 6, 7, 7, 7, 8, 9,10, 9, 8, 7,
7, 6,11,12,13,11, 6, 7, 8, 9,14,10, 9, 8, 6,11,
9, 9,10,10, 8,11,12,11, 9, 9,10,10, 8,13,13, 9,
9,10,10, 8,13,13, 9, 9,10,10,14,14,14,14,14 }
};
- static const int last_coeff_flag_offset_8x8[63] = {
- 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
- 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
- 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4,
- 5, 5, 5, 5, 6, 6, 6, 6, 7, 7, 7, 7, 8, 8, 8
- };
int index[64];
- int i, last;
+ int last;
int coeff_count = 0;
int abslevel1 = 1;
uint8_t *last_coeff_ctx_base;
uint8_t *abs_level_m1_ctx_base;
+#ifndef ARCH_X86
+#define CABAC_ON_STACK
+#endif
+#ifdef CABAC_ON_STACK
+#define CC &cc
+ CABACContext cc;
+ cc.range = h->cabac.range;
+ cc.low = h->cabac.low;
+ cc.bytestream= h->cabac.bytestream;
+#else
+#define CC &h->cabac
+#endif
+
+
/* cat: 0-> DC 16x16 n = 0
* 1-> AC 16x16 n = luma4x4idx
* 2-> Luma4x4 n = luma4x4idx
/* read coded block flag */
if( cat != 5 ) {
- if( get_cabac( &h->cabac, &h->cabac_state[85 + get_cabac_cbf_ctx( h, cat, n ) ] ) == 0 ) {
+ if( get_cabac( CC, &h->cabac_state[85 + get_cabac_cbf_ctx( h, cat, n ) ] ) == 0 ) {
if( cat == 1 || cat == 2 )
h->non_zero_count_cache[scan8[n]] = 0;
else if( cat == 4 )
h->non_zero_count_cache[scan8[16+n]] = 0;
-
+#ifdef CABAC_ON_STACK
+ h->cabac.range = cc.range ;
+ h->cabac.low = cc.low ;
+ h->cabac.bytestream= cc.bytestream;
+#endif
return 0;
}
}
#define DECODE_SIGNIFICANCE( coefs, sig_off, last_off ) \
for(last= 0; last < coefs; last++) { \
uint8_t *sig_ctx = significant_coeff_ctx_base + sig_off; \
- if( get_cabac( &h->cabac, sig_ctx )) { \
+ if( get_cabac( CC, sig_ctx )) { \
uint8_t *last_ctx = last_coeff_ctx_base + last_off; \
index[coeff_count++] = last; \
- if( get_cabac( &h->cabac, last_ctx ) ) { \
+ if( get_cabac( CC, last_ctx ) ) { \
last= max_coeff; \
break; \
} \
} \
+ }\
+ if( last == max_coeff -1 ) {\
+ index[coeff_count++] = last;\
}
- const int *sig_off = significant_coeff_flag_offset_8x8[MB_FIELD];
+ const uint8_t *sig_off = significant_coeff_flag_offset_8x8[MB_FIELD];
+#if defined(ARCH_X86) && defined(CONFIG_7REGS) && defined(HAVE_EBX_AVAILABLE) && !defined(BROKEN_RELOCATIONS)
+ coeff_count= decode_significance_8x8_x86(CC, significant_coeff_ctx_base, index, sig_off);
+ } else {
+ coeff_count= decode_significance_x86(CC, max_coeff, significant_coeff_ctx_base, index);
+#else
DECODE_SIGNIFICANCE( 63, sig_off[last], last_coeff_flag_offset_8x8[last] );
} else {
DECODE_SIGNIFICANCE( max_coeff - 1, last, last );
- }
- if( last == max_coeff -1 ) {
- index[coeff_count++] = last;
+#endif
}
assert(coeff_count > 0);
fill_rectangle(&h->non_zero_count_cache[scan8[n]], 2, 2, 8, coeff_count, 1);
}
- for( i = coeff_count - 1; i >= 0; i-- ) {
+ for( coeff_count--; coeff_count >= 0; coeff_count-- ) {
uint8_t *ctx = (abslevelgt1 != 0 ? 0 : FFMIN( 4, abslevel1 )) + abs_level_m1_ctx_base;
- int j= scantable[index[i]];
+ int j= scantable[index[coeff_count]];
- if( get_cabac( &h->cabac, ctx ) == 0 ) {
+ if( get_cabac( CC, ctx ) == 0 ) {
if( !qmul ) {
- if( get_cabac_bypass( &h->cabac ) ) block[j] = -1;
- else block[j] = 1;
+ block[j] = get_cabac_bypass_sign( CC, -1);
}else{
- if( get_cabac_bypass( &h->cabac ) ) block[j] = (-qmul[j] + 32) >> 6;
- else block[j] = ( qmul[j] + 32) >> 6;
+ block[j] = (get_cabac_bypass_sign( CC, -qmul[j]) + 32) >> 6;;
}
abslevel1++;
} else {
int coeff_abs = 2;
ctx = 5 + FFMIN( 4, abslevelgt1 ) + abs_level_m1_ctx_base;
- while( coeff_abs < 15 && get_cabac( &h->cabac, ctx ) ) {
+ while( coeff_abs < 15 && get_cabac( CC, ctx ) ) {
coeff_abs++;
}
if( coeff_abs >= 15 ) {
int j = 0;
- while( get_cabac_bypass( &h->cabac ) ) {
- coeff_abs += 1 << j;
+ while( get_cabac_bypass( CC ) ) {
j++;
}
+ coeff_abs=1;
while( j-- ) {
- if( get_cabac_bypass( &h->cabac ) )
- coeff_abs += 1 << j ;
+ coeff_abs += coeff_abs + get_cabac_bypass( CC );
}
+ coeff_abs+= 14;
}
if( !qmul ) {
- if( get_cabac_bypass( &h->cabac ) ) block[j] = -coeff_abs;
+ if( get_cabac_bypass( CC ) ) block[j] = -coeff_abs;
else block[j] = coeff_abs;
}else{
- if( get_cabac_bypass( &h->cabac ) ) block[j] = (-coeff_abs * qmul[j] + 32) >> 6;
+ if( get_cabac_bypass( CC ) ) block[j] = (-coeff_abs * qmul[j] + 32) >> 6;
else block[j] = ( coeff_abs * qmul[j] + 32) >> 6;
}
abslevelgt1++;
}
}
+#ifdef CABAC_ON_STACK
+ h->cabac.range = cc.range ;
+ h->cabac.low = cc.low ;
+ h->cabac.bytestream= cc.bytestream;
+#endif
return 0;
}
-static void inline compute_mb_neighbors(H264Context *h)
+static inline void compute_mb_neighbors(H264Context *h)
{
MpegEncContext * const s = &h->s;
const int mb_xy = s->mb_x + s->mb_y*s->mb_stride;
s->dsp.clear_blocks(h->mb); //FIXME avoid if already clear (move after skip handlong?)
- tprintf("pic:%d mb:%d/%d\n", h->frame_num, s->mb_x, s->mb_y);
+ tprintf(s->avctx, "pic:%d mb:%d/%d\n", h->frame_num, s->mb_x, s->mb_y);
if( h->slice_type != I_TYPE && h->slice_type != SI_TYPE ) {
int skip;
/* a skipped mb needs the aff flag from the following mb */
const uint8_t *ptr;
unsigned int x, y;
- // We assume these blocks are very rare so we dont optimize it.
+ // We assume these blocks are very rare so we do not optimize it.
// FIXME The two following lines get the bitstream position in the cabac
// decode, I think it should be done by a function in cabac.h (or cabac.c).
ptr= h->cabac.bytestream;
- if (h->cabac.low&0x1) ptr-=CABAC_BITS/8;
+ if(h->cabac.low&0x1) ptr--;
+ if(CABAC_BITS==16){
+ if(h->cabac.low&0x1FF) ptr--;
+ }
// The pixels are stored in the same order as levels in h->mb array.
for(y=0; y<16; y++){
const int index= 4*(y&3) + 32*((y>>2)&1) + 128*(y>>3);
for(x=0; x<16; x++){
- tprintf("LUMA ICPM LEVEL (%3d)\n", *ptr);
+ tprintf(s->avctx, "LUMA ICPM LEVEL (%3d)\n", *ptr);
h->mb[index + (x&3) + 16*((x>>2)&1) + 64*(x>>3)]= *ptr++;
}
}
for(y=0; y<8; y++){
const int index= 256 + 4*(y&3) + 32*(y>>2);
for(x=0; x<8; x++){
- tprintf("CHROMA U ICPM LEVEL (%3d)\n", *ptr);
+ tprintf(s->avctx, "CHROMA U ICPM LEVEL (%3d)\n", *ptr);
h->mb[index + (x&3) + 16*(x>>2)]= *ptr++;
}
}
for(y=0; y<8; y++){
const int index= 256 + 64 + 4*(y&3) + 32*(y>>2);
for(x=0; x<8; x++){
- tprintf("CHROMA V ICPM LEVEL (%3d)\n", *ptr);
+ tprintf(s->avctx, "CHROMA V ICPM LEVEL (%3d)\n", *ptr);
h->mb[index + (x&3) + 16*(x>>2)]= *ptr++;
}
}
fill_caches(h, mb_type, 0);
if( IS_INTRA( mb_type ) ) {
- int i;
+ int i, pred_mode;
if( IS_INTRA4x4( mb_type ) ) {
if( dct8x8_allowed && decode_cabac_mb_transform_size( h ) ) {
mb_type |= MB_TYPE_8x8DCT;
if( h->intra16x16_pred_mode < 0 ) return -1;
}
h->chroma_pred_mode_table[mb_xy] =
- h->chroma_pred_mode = decode_cabac_mb_chroma_pre_mode( h );
+ pred_mode = decode_cabac_mb_chroma_pre_mode( h );
- h->chroma_pred_mode= check_intra_pred_mode( h, h->chroma_pred_mode );
- if( h->chroma_pred_mode < 0 ) return -1;
+ pred_mode= check_intra_pred_mode( h, pred_mode );
+ if( pred_mode < 0 ) return -1;
+ h->chroma_pred_mode= pred_mode;
} else if( partition_count == 4 ) {
int i, j, sub_partition_count[4], list, ref[2][4];
sub_partition_count[i]= b_sub_mb_type_info[ h->sub_mb_type[i] ].partition_count;
h->sub_mb_type[i]= b_sub_mb_type_info[ h->sub_mb_type[i] ].type;
}
- if( IS_DIRECT(h->sub_mb_type[0]) || IS_DIRECT(h->sub_mb_type[1])
- || IS_DIRECT(h->sub_mb_type[2]) || IS_DIRECT(h->sub_mb_type[3])) {
+ if( IS_DIRECT(h->sub_mb_type[0] | h->sub_mb_type[1] |
+ h->sub_mb_type[2] | h->sub_mb_type[3]) ) {
pred_direct_motion(h, &mb_type);
if( h->ref_count[0] > 1 || h->ref_count[1] > 1 ) {
for( i = 0; i < 4; i++ )
}
}
- for( list = 0; list < 2; list++ ) {
- if( h->ref_count[list] > 0 ) {
+ for( list = 0; list < h->list_count; list++ ) {
for( i = 0; i < 4; i++ ) {
if(IS_DIRECT(h->sub_mb_type[i])) continue;
if(IS_DIR(h->sub_mb_type[i], 0, list)){
h->ref_cache[list][ scan8[4*i]+1 ]=
h->ref_cache[list][ scan8[4*i]+8 ]=h->ref_cache[list][ scan8[4*i]+9 ]= ref[list][i];
}
- }
}
if(dct8x8_allowed)
dct8x8_allowed = get_dct8x8_allowed(h);
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
for(i=0; i<4; i++){
if(IS_DIRECT(h->sub_mb_type[i])){
fill_rectangle(h->mvd_cache[list][scan8[4*i]], 2, 2, 8, 0, 4);
mx = mpx + decode_cabac_mb_mvd( h, list, index, 0 );
my = mpy + decode_cabac_mb_mvd( h, list, index, 1 );
- tprintf("final mv:%d %d\n", mx, my);
+ tprintf(s->avctx, "final mv:%d %d\n", mx, my);
if(IS_SUB_8X8(sub_mb_type)){
- mv_cache[ 0 ][0]= mv_cache[ 1 ][0]=
+ mv_cache[ 1 ][0]=
mv_cache[ 8 ][0]= mv_cache[ 9 ][0]= mx;
- mv_cache[ 0 ][1]= mv_cache[ 1 ][1]=
+ mv_cache[ 1 ][1]=
mv_cache[ 8 ][1]= mv_cache[ 9 ][1]= my;
- mvd_cache[ 0 ][0]= mvd_cache[ 1 ][0]=
+ mvd_cache[ 1 ][0]=
mvd_cache[ 8 ][0]= mvd_cache[ 9 ][0]= mx - mpx;
- mvd_cache[ 0 ][1]= mvd_cache[ 1 ][1]=
+ mvd_cache[ 1 ][1]=
mvd_cache[ 8 ][1]= mvd_cache[ 9 ][1]= my - mpy;
}else if(IS_SUB_8X4(sub_mb_type)){
- mv_cache[ 0 ][0]= mv_cache[ 1 ][0]= mx;
- mv_cache[ 0 ][1]= mv_cache[ 1 ][1]= my;
+ mv_cache[ 1 ][0]= mx;
+ mv_cache[ 1 ][1]= my;
- mvd_cache[ 0 ][0]= mvd_cache[ 1 ][0]= mx- mpx;
- mvd_cache[ 0 ][1]= mvd_cache[ 1 ][1]= my - mpy;
+ mvd_cache[ 1 ][0]= mx - mpx;
+ mvd_cache[ 1 ][1]= my - mpy;
}else if(IS_SUB_4X8(sub_mb_type)){
- mv_cache[ 0 ][0]= mv_cache[ 8 ][0]= mx;
- mv_cache[ 0 ][1]= mv_cache[ 8 ][1]= my;
-
- mvd_cache[ 0 ][0]= mvd_cache[ 8 ][0]= mx - mpx;
- mvd_cache[ 0 ][1]= mvd_cache[ 8 ][1]= my - mpy;
- }else{
- assert(IS_SUB_4X4(sub_mb_type));
- mv_cache[ 0 ][0]= mx;
- mv_cache[ 0 ][1]= my;
+ mv_cache[ 8 ][0]= mx;
+ mv_cache[ 8 ][1]= my;
- mvd_cache[ 0 ][0]= mx - mpx;
- mvd_cache[ 0 ][1]= my - mpy;
+ mvd_cache[ 8 ][0]= mx - mpx;
+ mvd_cache[ 8 ][1]= my - mpy;
}
+ mv_cache[ 0 ][0]= mx;
+ mv_cache[ 0 ][1]= my;
+
+ mvd_cache[ 0 ][0]= mx - mpx;
+ mvd_cache[ 0 ][1]= my - mpy;
}
}else{
uint32_t *p= (uint32_t *)&h->mv_cache[list][ scan8[4*i] ][0];
} else {
int list, mx, my, i, mpx, mpy;
if(IS_16X16(mb_type)){
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
if(IS_DIR(mb_type, 0, list)){
- if(h->ref_count[list] > 0 ){
const int ref = h->ref_count[list] > 1 ? decode_cabac_mb_ref( h, list, 0 ) : 0;
fill_rectangle(&h->ref_cache[list][ scan8[0] ], 4, 4, 8, ref, 1);
- }
}else
- fill_rectangle(&h->ref_cache[list][ scan8[0] ], 4, 4, 8, (uint8_t)LIST_NOT_USED, 1);
+ fill_rectangle(&h->ref_cache[list][ scan8[0] ], 4, 4, 8, (uint8_t)LIST_NOT_USED, 1); //FIXME factorize and the other fill_rect below too
}
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
if(IS_DIR(mb_type, 0, list)){
pred_motion(h, 0, 4, list, h->ref_cache[list][ scan8[0] ], &mpx, &mpy);
mx = mpx + decode_cabac_mb_mvd( h, list, 0, 0 );
my = mpy + decode_cabac_mb_mvd( h, list, 0, 1 );
- tprintf("final mv:%d %d\n", mx, my);
+ tprintf(s->avctx, "final mv:%d %d\n", mx, my);
fill_rectangle(h->mvd_cache[list][ scan8[0] ], 4, 4, 8, pack16to32(mx-mpx,my-mpy), 4);
fill_rectangle(h->mv_cache[list][ scan8[0] ], 4, 4, 8, pack16to32(mx,my), 4);
}
}
else if(IS_16X8(mb_type)){
- for(list=0; list<2; list++){
- if(h->ref_count[list]>0){
+ for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
if(IS_DIR(mb_type, i, list)){
const int ref= h->ref_count[list] > 1 ? decode_cabac_mb_ref( h, list, 8*i ) : 0;
}else
fill_rectangle(&h->ref_cache[list][ scan8[0] + 16*i ], 4, 2, 8, (LIST_NOT_USED&0xFF), 1);
}
- }
}
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
if(IS_DIR(mb_type, i, list)){
pred_16x8_motion(h, 8*i, list, h->ref_cache[list][scan8[0] + 16*i], &mpx, &mpy);
mx = mpx + decode_cabac_mb_mvd( h, list, 8*i, 0 );
my = mpy + decode_cabac_mb_mvd( h, list, 8*i, 1 );
- tprintf("final mv:%d %d\n", mx, my);
+ tprintf(s->avctx, "final mv:%d %d\n", mx, my);
fill_rectangle(h->mvd_cache[list][ scan8[0] + 16*i ], 4, 2, 8, pack16to32(mx-mpx,my-mpy), 4);
fill_rectangle(h->mv_cache[list][ scan8[0] + 16*i ], 4, 2, 8, pack16to32(mx,my), 4);
}
}else{
assert(IS_8X16(mb_type));
- for(list=0; list<2; list++){
- if(h->ref_count[list]>0){
+ for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
if(IS_DIR(mb_type, i, list)){ //FIXME optimize
const int ref= h->ref_count[list] > 1 ? decode_cabac_mb_ref( h, list, 4*i ) : 0;
}else
fill_rectangle(&h->ref_cache[list][ scan8[0] + 2*i ], 2, 4, 8, (LIST_NOT_USED&0xFF), 1);
}
- }
}
- for(list=0; list<2; list++){
+ for(list=0; list<h->list_count; list++){
for(i=0; i<2; i++){
if(IS_DIR(mb_type, i, list)){
pred_8x16_motion(h, i*4, list, h->ref_cache[list][ scan8[0] + 2*i ], &mpx, &mpy);
mx = mpx + decode_cabac_mb_mvd( h, list, 4*i, 0 );
my = mpy + decode_cabac_mb_mvd( h, list, 4*i, 1 );
- tprintf("final mv:%d %d\n", mx, my);
+ tprintf(s->avctx, "final mv:%d %d\n", mx, my);
fill_rectangle(h->mvd_cache[list][ scan8[0] + 2*i ], 2, 4, 8, pack16to32(mx-mpx,my-mpy), 4);
fill_rectangle(h->mv_cache[list][ scan8[0] + 2*i ], 2, 4, 8, pack16to32(mx,my), 4);
}else{
cbp |= decode_cabac_mb_cbp_chroma( h ) << 4;
}
- h->cbp_table[mb_xy] = cbp;
+ h->cbp_table[mb_xy] = h->cbp = cbp;
if( dct8x8_allowed && (cbp&15) && !IS_INTRA( mb_type ) ) {
if( decode_cabac_mb_transform_size( h ) )
for( i4x4 = 0; i4x4 < 4; i4x4++ ) {
const int index = 4*i8x8 + i4x4;
//av_log( s->avctx, AV_LOG_ERROR, "Luma4x4: %d\n", index );
+//START_TIMER
if( decode_cabac_residual(h, h->mb + 16*index, 2, index, scan, h->dequant4_coeff[IS_INTRA( mb_type ) ? 0:3][s->qscale], 16) < 0 )
return -1;
+//STOP_TIMER("decode_residual")
}
} else {
uint8_t * const nnz= &h->non_zero_count_cache[ scan8[4*i8x8] ];
}
-static void filter_mb_edgev( H264Context *h, uint8_t *pix, int stride, int bS[4], int qp ) {
+static void filter_mb_edgev( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
int i, d;
- const int index_a = clip( qp + h->slice_alpha_c0_offset, 0, 51 );
- const int alpha = alpha_table[index_a];
- const int beta = beta_table[clip( qp + h->slice_beta_offset, 0, 51 )];
+ const int index_a = qp + h->slice_alpha_c0_offset;
+ const int alpha = (alpha_table+52)[index_a];
+ const int beta = (beta_table+52)[qp + h->slice_beta_offset];
if( bS[0] < 4 ) {
int8_t tc[4];
for(i=0; i<4; i++)
- tc[i] = bS[i] ? tc0_table[index_a][bS[i] - 1] : -1;
+ tc[i] = bS[i] ? (tc0_table+52)[index_a][bS[i] - 1] : -1;
h->s.dsp.h264_h_loop_filter_luma(pix, stride, alpha, beta, tc);
} else {
/* 16px edge length, because bS=4 is triggered by being at
const int q1 = pix[1];
const int q2 = pix[2];
- if( ABS( p0 - q0 ) < alpha &&
- ABS( p1 - p0 ) < beta &&
- ABS( q1 - q0 ) < beta ) {
+ if( FFABS( p0 - q0 ) < alpha &&
+ FFABS( p1 - p0 ) < beta &&
+ FFABS( q1 - q0 ) < beta ) {
- if(ABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
- if( ABS( p2 - p0 ) < beta)
+ if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
+ if( FFABS( p2 - p0 ) < beta)
{
const int p3 = pix[-4];
/* p0', p1', p2' */
/* p0' */
pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
}
- if( ABS( q2 - q0 ) < beta)
+ if( FFABS( q2 - q0 ) < beta)
{
const int q3 = pix[3];
/* q0', q1', q2' */
pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
pix[ 0] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
}
- tprintf("filter_mb_edgev i:%d d:%d\n# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, d, p2, p1, p0, q0, q1, q2, pix[-2], pix[-1], pix[0], pix[1]);
+ tprintf(h->s.avctx, "filter_mb_edgev i:%d d:%d\n# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, d, p2, p1, p0, q0, q1, q2, pix[-2], pix[-1], pix[0], pix[1]);
}
pix += stride;
}
}
}
-static void filter_mb_edgecv( H264Context *h, uint8_t *pix, int stride, int bS[4], int qp ) {
+static void filter_mb_edgecv( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
int i;
- const int index_a = clip( qp + h->slice_alpha_c0_offset, 0, 51 );
- const int alpha = alpha_table[index_a];
- const int beta = beta_table[clip( qp + h->slice_beta_offset, 0, 51 )];
+ const int index_a = qp + h->slice_alpha_c0_offset;
+ const int alpha = (alpha_table+52)[index_a];
+ const int beta = (beta_table+52)[qp + h->slice_beta_offset];
if( bS[0] < 4 ) {
int8_t tc[4];
for(i=0; i<4; i++)
- tc[i] = bS[i] ? tc0_table[index_a][bS[i] - 1] + 1 : 0;
+ tc[i] = bS[i] ? (tc0_table+52)[index_a][bS[i] - 1] + 1 : 0;
h->s.dsp.h264_h_loop_filter_chroma(pix, stride, alpha, beta, tc);
} else {
h->s.dsp.h264_h_loop_filter_chroma_intra(pix, stride, alpha, beta);
}
}
-static void filter_mb_mbaff_edgev( H264Context *h, uint8_t *pix, int stride, int bS[8], int qp[2] ) {
+static void filter_mb_mbaff_edgev( H264Context *h, uint8_t *pix, int stride, int16_t bS[8], int qp[2] ) {
int i;
for( i = 0; i < 16; i++, pix += stride) {
int index_a;
}
qp_index = MB_FIELD ? (i >> 3) : (i & 1);
- index_a = clip( qp[qp_index] + h->slice_alpha_c0_offset, 0, 51 );
- alpha = alpha_table[index_a];
- beta = beta_table[clip( qp[qp_index] + h->slice_beta_offset, 0, 51 )];
+ index_a = qp[qp_index] + h->slice_alpha_c0_offset;
+ alpha = (alpha_table+52)[index_a];
+ beta = (beta_table+52)[qp[qp_index] + h->slice_beta_offset];
if( bS[bS_index] < 4 ) {
- const int tc0 = tc0_table[index_a][bS[bS_index] - 1];
+ const int tc0 = (tc0_table+52)[index_a][bS[bS_index] - 1];
const int p0 = pix[-1];
const int p1 = pix[-2];
const int p2 = pix[-3];
const int q1 = pix[1];
const int q2 = pix[2];
- if( ABS( p0 - q0 ) < alpha &&
- ABS( p1 - p0 ) < beta &&
- ABS( q1 - q0 ) < beta ) {
+ if( FFABS( p0 - q0 ) < alpha &&
+ FFABS( p1 - p0 ) < beta &&
+ FFABS( q1 - q0 ) < beta ) {
int tc = tc0;
int i_delta;
- if( ABS( p2 - p0 ) < beta ) {
- pix[-2] = p1 + clip( ( p2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( p1 << 1 ) ) >> 1, -tc0, tc0 );
+ if( FFABS( p2 - p0 ) < beta ) {
+ pix[-2] = p1 + av_clip( ( p2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( p1 << 1 ) ) >> 1, -tc0, tc0 );
tc++;
}
- if( ABS( q2 - q0 ) < beta ) {
- pix[1] = q1 + clip( ( q2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( q1 << 1 ) ) >> 1, -tc0, tc0 );
+ if( FFABS( q2 - q0 ) < beta ) {
+ pix[1] = q1 + av_clip( ( q2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( q1 << 1 ) ) >> 1, -tc0, tc0 );
tc++;
}
- i_delta = clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
- pix[-1] = clip_uint8( p0 + i_delta ); /* p0' */
- pix[0] = clip_uint8( q0 - i_delta ); /* q0' */
- tprintf("filter_mb_mbaff_edgev i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d, tc:%d\n# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, qp[qp_index], index_a, alpha, beta, tc, bS[bS_index], pix[-3], p1, p0, q0, q1, pix[2], p1, pix[-1], pix[0], q1);
+ i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
+ pix[-1] = av_clip_uint8( p0 + i_delta ); /* p0' */
+ pix[0] = av_clip_uint8( q0 - i_delta ); /* q0' */
+ tprintf(h->s.avctx, "filter_mb_mbaff_edgev i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d, tc:%d\n# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, qp[qp_index], index_a, alpha, beta, tc, bS[bS_index], pix[-3], p1, p0, q0, q1, pix[2], p1, pix[-1], pix[0], q1);
}
}else{
const int p0 = pix[-1];
const int q1 = pix[1];
const int q2 = pix[2];
- if( ABS( p0 - q0 ) < alpha &&
- ABS( p1 - p0 ) < beta &&
- ABS( q1 - q0 ) < beta ) {
+ if( FFABS( p0 - q0 ) < alpha &&
+ FFABS( p1 - p0 ) < beta &&
+ FFABS( q1 - q0 ) < beta ) {
- if(ABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
- if( ABS( p2 - p0 ) < beta)
+ if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
+ if( FFABS( p2 - p0 ) < beta)
{
const int p3 = pix[-4];
/* p0', p1', p2' */
/* p0' */
pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
}
- if( ABS( q2 - q0 ) < beta)
+ if( FFABS( q2 - q0 ) < beta)
{
const int q3 = pix[3];
/* q0', q1', q2' */
pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
pix[ 0] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
}
- tprintf("filter_mb_mbaff_edgev i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d\n# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x, %02x, %02x]\n", i, qp[qp_index], index_a, alpha, beta, p2, p1, p0, q0, q1, q2, pix[-3], pix[-2], pix[-1], pix[0], pix[1], pix[2]);
+ tprintf(h->s.avctx, "filter_mb_mbaff_edgev i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d\n# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x, %02x, %02x]\n", i, qp[qp_index], index_a, alpha, beta, p2, p1, p0, q0, q1, q2, pix[-3], pix[-2], pix[-1], pix[0], pix[1], pix[2]);
}
}
}
}
-static void filter_mb_mbaff_edgecv( H264Context *h, uint8_t *pix, int stride, int bS[8], int qp[2] ) {
+static void filter_mb_mbaff_edgecv( H264Context *h, uint8_t *pix, int stride, int16_t bS[8], int qp[2] ) {
int i;
for( i = 0; i < 8; i++, pix += stride) {
int index_a;
}
qp_index = MB_FIELD ? (i >> 2) : (i & 1);
- index_a = clip( qp[qp_index] + h->slice_alpha_c0_offset, 0, 51 );
- alpha = alpha_table[index_a];
- beta = beta_table[clip( qp[qp_index] + h->slice_beta_offset, 0, 51 )];
+ index_a = qp[qp_index] + h->slice_alpha_c0_offset;
+ alpha = (alpha_table+52)[index_a];
+ beta = (beta_table+52)[qp[qp_index] + h->slice_beta_offset];
if( bS[bS_index] < 4 ) {
- const int tc = tc0_table[index_a][bS[bS_index] - 1] + 1;
+ const int tc = (tc0_table+52)[index_a][bS[bS_index] - 1] + 1;
const int p0 = pix[-1];
const int p1 = pix[-2];
const int q0 = pix[0];
const int q1 = pix[1];
- if( ABS( p0 - q0 ) < alpha &&
- ABS( p1 - p0 ) < beta &&
- ABS( q1 - q0 ) < beta ) {
- const int i_delta = clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
+ if( FFABS( p0 - q0 ) < alpha &&
+ FFABS( p1 - p0 ) < beta &&
+ FFABS( q1 - q0 ) < beta ) {
+ const int i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
- pix[-1] = clip_uint8( p0 + i_delta ); /* p0' */
- pix[0] = clip_uint8( q0 - i_delta ); /* q0' */
- tprintf("filter_mb_mbaff_edgecv i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d, tc:%d\n# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, qp[qp_index], index_a, alpha, beta, tc, bS[bS_index], pix[-3], p1, p0, q0, q1, pix[2], p1, pix[-1], pix[0], q1);
+ pix[-1] = av_clip_uint8( p0 + i_delta ); /* p0' */
+ pix[0] = av_clip_uint8( q0 - i_delta ); /* q0' */
+ tprintf(h->s.avctx, "filter_mb_mbaff_edgecv i:%d, qp:%d, indexA:%d, alpha:%d, beta:%d, tc:%d\n# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, qp[qp_index], index_a, alpha, beta, tc, bS[bS_index], pix[-3], p1, p0, q0, q1, pix[2], p1, pix[-1], pix[0], q1);
}
}else{
const int p0 = pix[-1];
const int q0 = pix[0];
const int q1 = pix[1];
- if( ABS( p0 - q0 ) < alpha &&
- ABS( p1 - p0 ) < beta &&
- ABS( q1 - q0 ) < beta ) {
+ if( FFABS( p0 - q0 ) < alpha &&
+ FFABS( p1 - p0 ) < beta &&
+ FFABS( q1 - q0 ) < beta ) {
pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2; /* p0' */
pix[0] = ( 2*q1 + q0 + p1 + 2 ) >> 2; /* q0' */
- tprintf("filter_mb_mbaff_edgecv i:%d\n# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x, %02x, %02x]\n", i, pix[-3], p1, p0, q0, q1, pix[2], pix[-3], pix[-2], pix[-1], pix[0], pix[1], pix[2]);
+ tprintf(h->s.avctx, "filter_mb_mbaff_edgecv i:%d\n# bS:4 -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x, %02x, %02x]\n", i, pix[-3], p1, p0, q0, q1, pix[2], pix[-3], pix[-2], pix[-1], pix[0], pix[1], pix[2]);
}
}
}
}
-static void filter_mb_edgeh( H264Context *h, uint8_t *pix, int stride, int bS[4], int qp ) {
+static void filter_mb_edgeh( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
int i, d;
- const int index_a = clip( qp + h->slice_alpha_c0_offset, 0, 51 );
- const int alpha = alpha_table[index_a];
- const int beta = beta_table[clip( qp + h->slice_beta_offset, 0, 51 )];
+ const int index_a = qp + h->slice_alpha_c0_offset;
+ const int alpha = (alpha_table+52)[index_a];
+ const int beta = (beta_table+52)[qp + h->slice_beta_offset];
const int pix_next = stride;
if( bS[0] < 4 ) {
int8_t tc[4];
for(i=0; i<4; i++)
- tc[i] = bS[i] ? tc0_table[index_a][bS[i] - 1] : -1;
+ tc[i] = bS[i] ? (tc0_table+52)[index_a][bS[i] - 1] : -1;
h->s.dsp.h264_v_loop_filter_luma(pix, stride, alpha, beta, tc);
} else {
/* 16px edge length, see filter_mb_edgev */
const int q1 = pix[1*pix_next];
const int q2 = pix[2*pix_next];
- if( ABS( p0 - q0 ) < alpha &&
- ABS( p1 - p0 ) < beta &&
- ABS( q1 - q0 ) < beta ) {
+ if( FFABS( p0 - q0 ) < alpha &&
+ FFABS( p1 - p0 ) < beta &&
+ FFABS( q1 - q0 ) < beta ) {
const int p3 = pix[-4*pix_next];
const int q3 = pix[ 3*pix_next];
- if(ABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
- if( ABS( p2 - p0 ) < beta) {
+ if(FFABS( p0 - q0 ) < (( alpha >> 2 ) + 2 )){
+ if( FFABS( p2 - p0 ) < beta) {
/* p0', p1', p2' */
pix[-1*pix_next] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
pix[-2*pix_next] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
/* p0' */
pix[-1*pix_next] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
}
- if( ABS( q2 - q0 ) < beta) {
+ if( FFABS( q2 - q0 ) < beta) {
/* q0', q1', q2' */
pix[0*pix_next] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
pix[1*pix_next] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
pix[-1*pix_next] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
pix[ 0*pix_next] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
}
- tprintf("filter_mb_edgeh i:%d d:%d, qp:%d, indexA:%d, alpha:%d, beta:%d\n# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, d, qp, index_a, alpha, beta, bS[i], p2, p1, p0, q0, q1, q2, pix[-2*pix_next], pix[-pix_next], pix[0], pix[pix_next]);
+ tprintf(h->s.avctx, "filter_mb_edgeh i:%d d:%d, qp:%d, indexA:%d, alpha:%d, beta:%d\n# bS:%d -> [%02x, %02x, %02x, %02x, %02x, %02x] =>[%02x, %02x, %02x, %02x]\n", i, d, qp, index_a, alpha, beta, bS[i], p2, p1, p0, q0, q1, q2, pix[-2*pix_next], pix[-pix_next], pix[0], pix[pix_next]);
}
pix++;
}
}
}
-static void filter_mb_edgech( H264Context *h, uint8_t *pix, int stride, int bS[4], int qp ) {
+static void filter_mb_edgech( H264Context *h, uint8_t *pix, int stride, int16_t bS[4], int qp ) {
int i;
- const int index_a = clip( qp + h->slice_alpha_c0_offset, 0, 51 );
- const int alpha = alpha_table[index_a];
- const int beta = beta_table[clip( qp + h->slice_beta_offset, 0, 51 )];
+ const int index_a = qp + h->slice_alpha_c0_offset;
+ const int alpha = (alpha_table+52)[index_a];
+ const int beta = (beta_table+52)[qp + h->slice_beta_offset];
if( bS[0] < 4 ) {
int8_t tc[4];
for(i=0; i<4; i++)
- tc[i] = bS[i] ? tc0_table[index_a][bS[i] - 1] + 1 : 0;
+ tc[i] = bS[i] ? (tc0_table+52)[index_a][bS[i] - 1] + 1 : 0;
h->s.dsp.h264_v_loop_filter_chroma(pix, stride, alpha, beta, tc);
} else {
h->s.dsp.h264_v_loop_filter_chroma_intra(pix, stride, alpha, beta);
}
}
+static void filter_mb_fast( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize) {
+ MpegEncContext * const s = &h->s;
+ int mb_xy, mb_type;
+ int qp, qp0, qp1, qpc, qpc0, qpc1, qp_thresh;
+
+ mb_xy = mb_x + mb_y*s->mb_stride;
+
+ if(mb_x==0 || mb_y==0 || !s->dsp.h264_loop_filter_strength ||
+ (h->deblocking_filter == 2 && (h->slice_table[mb_xy] != h->slice_table[h->top_mb_xy] ||
+ h->slice_table[mb_xy] != h->slice_table[mb_xy - 1]))) {
+ filter_mb(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize);
+ return;
+ }
+ assert(!FRAME_MBAFF);
+
+ mb_type = s->current_picture.mb_type[mb_xy];
+ qp = s->current_picture.qscale_table[mb_xy];
+ qp0 = s->current_picture.qscale_table[mb_xy-1];
+ qp1 = s->current_picture.qscale_table[h->top_mb_xy];
+ qpc = get_chroma_qp( h->pps.chroma_qp_index_offset, qp );
+ qpc0 = get_chroma_qp( h->pps.chroma_qp_index_offset, qp0 );
+ qpc1 = get_chroma_qp( h->pps.chroma_qp_index_offset, qp1 );
+ qp0 = (qp + qp0 + 1) >> 1;
+ qp1 = (qp + qp1 + 1) >> 1;
+ qpc0 = (qpc + qpc0 + 1) >> 1;
+ qpc1 = (qpc + qpc1 + 1) >> 1;
+ qp_thresh = 15 - h->slice_alpha_c0_offset;
+ if(qp <= qp_thresh && qp0 <= qp_thresh && qp1 <= qp_thresh &&
+ qpc <= qp_thresh && qpc0 <= qp_thresh && qpc1 <= qp_thresh)
+ return;
+
+ if( IS_INTRA(mb_type) ) {
+ int16_t bS4[4] = {4,4,4,4};
+ int16_t bS3[4] = {3,3,3,3};
+ if( IS_8x8DCT(mb_type) ) {
+ filter_mb_edgev( h, &img_y[4*0], linesize, bS4, qp0 );
+ filter_mb_edgev( h, &img_y[4*2], linesize, bS3, qp );
+ filter_mb_edgeh( h, &img_y[4*0*linesize], linesize, bS4, qp1 );
+ filter_mb_edgeh( h, &img_y[4*2*linesize], linesize, bS3, qp );
+ } else {
+ filter_mb_edgev( h, &img_y[4*0], linesize, bS4, qp0 );
+ filter_mb_edgev( h, &img_y[4*1], linesize, bS3, qp );
+ filter_mb_edgev( h, &img_y[4*2], linesize, bS3, qp );
+ filter_mb_edgev( h, &img_y[4*3], linesize, bS3, qp );
+ filter_mb_edgeh( h, &img_y[4*0*linesize], linesize, bS4, qp1 );
+ filter_mb_edgeh( h, &img_y[4*1*linesize], linesize, bS3, qp );
+ filter_mb_edgeh( h, &img_y[4*2*linesize], linesize, bS3, qp );
+ filter_mb_edgeh( h, &img_y[4*3*linesize], linesize, bS3, qp );
+ }
+ filter_mb_edgecv( h, &img_cb[2*0], uvlinesize, bS4, qpc0 );
+ filter_mb_edgecv( h, &img_cb[2*2], uvlinesize, bS3, qpc );
+ filter_mb_edgecv( h, &img_cr[2*0], uvlinesize, bS4, qpc0 );
+ filter_mb_edgecv( h, &img_cr[2*2], uvlinesize, bS3, qpc );
+ filter_mb_edgech( h, &img_cb[2*0*uvlinesize], uvlinesize, bS4, qpc1 );
+ filter_mb_edgech( h, &img_cb[2*2*uvlinesize], uvlinesize, bS3, qpc );
+ filter_mb_edgech( h, &img_cr[2*0*uvlinesize], uvlinesize, bS4, qpc1 );
+ filter_mb_edgech( h, &img_cr[2*2*uvlinesize], uvlinesize, bS3, qpc );
+ return;
+ } else {
+ DECLARE_ALIGNED_8(int16_t, bS[2][4][4]);
+ uint64_t (*bSv)[4] = (uint64_t(*)[4])bS;
+ int edges;
+ if( IS_8x8DCT(mb_type) && (h->cbp&7) == 7 ) {
+ edges = 4;
+ bSv[0][0] = bSv[0][2] = bSv[1][0] = bSv[1][2] = 0x0002000200020002ULL;
+ } else {
+ int mask_edge1 = (mb_type & (MB_TYPE_16x16 | MB_TYPE_8x16)) ? 3 :
+ (mb_type & MB_TYPE_16x8) ? 1 : 0;
+ int mask_edge0 = (mb_type & (MB_TYPE_16x16 | MB_TYPE_8x16))
+ && (s->current_picture.mb_type[mb_xy-1] & (MB_TYPE_16x16 | MB_TYPE_8x16))
+ ? 3 : 0;
+ int step = IS_8x8DCT(mb_type) ? 2 : 1;
+ edges = (mb_type & MB_TYPE_16x16) && !(h->cbp & 15) ? 1 : 4;
+ s->dsp.h264_loop_filter_strength( bS, h->non_zero_count_cache, h->ref_cache, h->mv_cache,
+ (h->slice_type == B_TYPE), edges, step, mask_edge0, mask_edge1 );
+ }
+ if( IS_INTRA(s->current_picture.mb_type[mb_xy-1]) )
+ bSv[0][0] = 0x0004000400040004ULL;
+ if( IS_INTRA(s->current_picture.mb_type[h->top_mb_xy]) )
+ bSv[1][0] = 0x0004000400040004ULL;
+
+#define FILTER(hv,dir,edge)\
+ if(bSv[dir][edge]) {\
+ filter_mb_edge##hv( h, &img_y[4*edge*(dir?linesize:1)], linesize, bS[dir][edge], edge ? qp : qp##dir );\
+ if(!(edge&1)) {\
+ filter_mb_edgec##hv( h, &img_cb[2*edge*(dir?uvlinesize:1)], uvlinesize, bS[dir][edge], edge ? qpc : qpc##dir );\
+ filter_mb_edgec##hv( h, &img_cr[2*edge*(dir?uvlinesize:1)], uvlinesize, bS[dir][edge], edge ? qpc : qpc##dir );\
+ }\
+ }
+ if( edges == 1 ) {
+ FILTER(v,0,0);
+ FILTER(h,1,0);
+ } else if( IS_8x8DCT(mb_type) ) {
+ FILTER(v,0,0);
+ FILTER(v,0,2);
+ FILTER(h,1,0);
+ FILTER(h,1,2);
+ } else {
+ FILTER(v,0,0);
+ FILTER(v,0,1);
+ FILTER(v,0,2);
+ FILTER(v,0,3);
+ FILTER(h,1,0);
+ FILTER(h,1,1);
+ FILTER(h,1,2);
+ FILTER(h,1,3);
+ }
+#undef FILTER
+ }
+}
+
static void filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize) {
MpegEncContext * const s = &h->s;
const int mb_xy= mb_x + mb_y*s->mb_stride;
*/
const int pair_xy = mb_x + (mb_y&~1)*s->mb_stride;
const int left_mb_xy[2] = { pair_xy-1, pair_xy-1+s->mb_stride };
- int bS[8];
+ int16_t bS[8];
int qp[2];
int chroma_qp[2];
int mb_qp, mbn0_qp, mbn1_qp;
get_chroma_qp( h->pps.chroma_qp_index_offset, mbn1_qp ) + 1 ) >> 1;
/* Filter edge */
- tprintf("filter mb:%d/%d MBAFF, QPy:%d/%d, QPc:%d/%d ls:%d uvls:%d", mb_x, mb_y, qp[0], qp[1], chroma_qp[0], chroma_qp[1], linesize, uvlinesize);
- { int i; for (i = 0; i < 8; i++) tprintf(" bS[%d]:%d", i, bS[i]); tprintf("\n"); }
+ tprintf(s->avctx, "filter mb:%d/%d MBAFF, QPy:%d/%d, QPc:%d/%d ls:%d uvls:%d", mb_x, mb_y, qp[0], qp[1], chroma_qp[0], chroma_qp[1], linesize, uvlinesize);
+ { int i; for (i = 0; i < 8; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
filter_mb_mbaff_edgev ( h, &img_y [0], linesize, bS, qp );
filter_mb_mbaff_edgecv( h, &img_cb[0], uvlinesize, bS, chroma_qp );
filter_mb_mbaff_edgecv( h, &img_cr[0], uvlinesize, bS, chroma_qp );
int mbn_xy = mb_xy - 2 * s->mb_stride;
int qp, chroma_qp;
int i, j;
- int bS[4];
+ int16_t bS[4];
for(j=0; j<2; j++, mbn_xy += s->mb_stride){
if( IS_INTRA(mb_type) ||
// Do not use s->qscale as luma quantizer because it has not the same
// value in IPCM macroblocks.
qp = ( s->current_picture.qscale_table[mb_xy] + s->current_picture.qscale_table[mbn_xy] + 1 ) >> 1;
- tprintf("filter mb:%d/%d dir:%d edge:%d, QPy:%d ls:%d uvls:%d", mb_x, mb_y, dir, edge, qp, tmp_linesize, tmp_uvlinesize);
- { int i; for (i = 0; i < 4; i++) tprintf(" bS[%d]:%d", i, bS[i]); tprintf("\n"); }
+ tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d ls:%d uvls:%d", mb_x, mb_y, dir, edge, qp, tmp_linesize, tmp_uvlinesize);
+ { int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
filter_mb_edgeh( h, &img_y[j*linesize], tmp_linesize, bS, qp );
chroma_qp = ( h->chroma_qp +
get_chroma_qp( h->pps.chroma_qp_index_offset, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1;
/* mbn_xy: neighbor macroblock */
const int mbn_xy = edge > 0 ? mb_xy : mbm_xy;
const int mbn_type = s->current_picture.mb_type[mbn_xy];
- int bS[4];
+ int16_t bS[4];
int qp;
if( (edge&1) && IS_8x8DCT(mb_type) )
int v = 0;
for( l = 0; !v && l < 1 + (h->slice_type == B_TYPE); l++ ) {
v |= ref2frm[h->ref_cache[l][b_idx]+2] != ref2frm[h->ref_cache[l][bn_idx]+2] ||
- ABS( h->mv_cache[l][b_idx][0] - h->mv_cache[l][bn_idx][0] ) >= 4 ||
- ABS( h->mv_cache[l][b_idx][1] - h->mv_cache[l][bn_idx][1] ) >= mvy_limit;
+ FFABS( h->mv_cache[l][b_idx][0] - h->mv_cache[l][bn_idx][0] ) >= 4 ||
+ FFABS( h->mv_cache[l][b_idx][1] - h->mv_cache[l][bn_idx][1] ) >= mvy_limit;
}
bS[0] = bS[1] = bS[2] = bS[3] = v;
mv_done = 1;
bS[i] = 0;
for( l = 0; l < 1 + (h->slice_type == B_TYPE); l++ ) {
if( ref2frm[h->ref_cache[l][b_idx]+2] != ref2frm[h->ref_cache[l][bn_idx]+2] ||
- ABS( h->mv_cache[l][b_idx][0] - h->mv_cache[l][bn_idx][0] ) >= 4 ||
- ABS( h->mv_cache[l][b_idx][1] - h->mv_cache[l][bn_idx][1] ) >= mvy_limit ) {
+ FFABS( h->mv_cache[l][b_idx][0] - h->mv_cache[l][bn_idx][0] ) >= 4 ||
+ FFABS( h->mv_cache[l][b_idx][1] - h->mv_cache[l][bn_idx][1] ) >= mvy_limit ) {
bS[i] = 1;
break;
}
// Do not use s->qscale as luma quantizer because it has not the same
// value in IPCM macroblocks.
qp = ( s->current_picture.qscale_table[mb_xy] + s->current_picture.qscale_table[mbn_xy] + 1 ) >> 1;
- //tprintf("filter mb:%d/%d dir:%d edge:%d, QPy:%d, QPc:%d, QPcn:%d\n", mb_x, mb_y, dir, edge, qp, h->chroma_qp, s->current_picture.qscale_table[mbn_xy]);
- tprintf("filter mb:%d/%d dir:%d edge:%d, QPy:%d ls:%d uvls:%d", mb_x, mb_y, dir, edge, qp, linesize, uvlinesize);
- { int i; for (i = 0; i < 4; i++) tprintf(" bS[%d]:%d", i, bS[i]); tprintf("\n"); }
+ //tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d, QPc:%d, QPcn:%d\n", mb_x, mb_y, dir, edge, qp, h->chroma_qp, s->current_picture.qscale_table[mbn_xy]);
+ tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d ls:%d uvls:%d", mb_x, mb_y, dir, edge, qp, linesize, uvlinesize);
+ { int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); }
if( dir == 0 ) {
filter_mb_edgev( h, &img_y[4*edge], linesize, bS, qp );
if( (edge&1) == 0 ) {
align_get_bits( &s->gb );
/* init cabac */
- ff_init_cabac_states( &h->cabac, ff_h264_lps_range, ff_h264_mps_state, ff_h264_lps_state, 64 );
+ ff_init_cabac_states( &h->cabac);
ff_init_cabac_decoder( &h->cabac,
s->gb.buffer + get_bits_count(&s->gb)/8,
( s->gb.size_in_bits - get_bits_count(&s->gb) + 7)/8);
for( i= 0; i < 460; i++ ) {
int pre;
if( h->slice_type == I_TYPE )
- pre = clip( ((cabac_context_init_I[i][0] * s->qscale) >>4 ) + cabac_context_init_I[i][1], 1, 126 );
+ pre = av_clip( ((cabac_context_init_I[i][0] * s->qscale) >>4 ) + cabac_context_init_I[i][1], 1, 126 );
else
- pre = clip( ((cabac_context_init_PB[h->cabac_init_idc][i][0] * s->qscale) >>4 ) + cabac_context_init_PB[h->cabac_init_idc][i][1], 1, 126 );
+ pre = av_clip( ((cabac_context_init_PB[h->cabac_init_idc][i][0] * s->qscale) >>4 ) + cabac_context_init_PB[h->cabac_init_idc][i][1], 1, 126 );
if( pre <= 63 )
h->cabac_state[i] = 2 * ( 63 - pre ) + 0;
}
for(;;){
+//START_TIMER
int ret = decode_mb_cabac(h);
int eos;
+//STOP_TIMER("decode_mb_cabac")
if(ret>=0) hl_decode_mb(h);
}
eos = get_cabac_terminate( &h->cabac );
- if( ret < 0 || h->cabac.bytestream > h->cabac.bytestream_end + 1) {
+ if( ret < 0 || h->cabac.bytestream > h->cabac.bytestream_end + 2) {
av_log(h->s.avctx, AV_LOG_ERROR, "error while decoding MB %d %d, bytestream (%d)\n", s->mb_x, s->mb_y, h->cabac.bytestream_end - h->cabac.bytestream);
ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x, s->mb_y, (AC_ERROR|DC_ERROR|MV_ERROR)&part_mask);
return -1;
}
if( eos || s->mb_y >= s->mb_height ) {
- tprintf("slice end %d %d\n", get_bits_count(&s->gb), s->gb.size_in_bits);
+ tprintf(s->avctx, "slice end %d %d\n", get_bits_count(&s->gb), s->gb.size_in_bits);
ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x-1, s->mb_y, (AC_END|DC_END|MV_END)&part_mask);
return 0;
}
++s->mb_y;
}
if(s->mb_y >= s->mb_height){
- tprintf("slice end %d %d\n", get_bits_count(&s->gb), s->gb.size_in_bits);
+ tprintf(s->avctx, "slice end %d %d\n", get_bits_count(&s->gb), s->gb.size_in_bits);
if(get_bits_count(&s->gb) == s->gb.size_in_bits ) {
ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x-1, s->mb_y, (AC_END|DC_END|MV_END)&part_mask);
}
if(get_bits_count(&s->gb) >= s->gb.size_in_bits && s->mb_skip_run<=0){
- tprintf("slice end %d %d\n", get_bits_count(&s->gb), s->gb.size_in_bits);
+ tprintf(s->avctx, "slice end %d %d\n", get_bits_count(&s->gb), s->gb.size_in_bits);
if(get_bits_count(&s->gb) == s->gb.size_in_bits ){
ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x-1, s->mb_y, (AC_END|DC_END|MV_END)&part_mask);
switch(type){
case 5:
- if(decode_unregistered_user_data(h, size) < 0);
+ if(decode_unregistered_user_data(h, size) < 0)
return -1;
break;
default:
static inline int decode_vui_parameters(H264Context *h, SPS *sps){
MpegEncContext * const s = &h->s;
- int aspect_ratio_info_present_flag, aspect_ratio_idc;
+ int aspect_ratio_info_present_flag;
+ unsigned int aspect_ratio_idc;
int nal_hrd_parameters_present_flag, vcl_hrd_parameters_present_flag;
aspect_ratio_info_present_flag= get_bits1(&s->gb);
sps->bitstream_restriction_flag = get_bits1(&s->gb);
if(sps->bitstream_restriction_flag){
+ unsigned int num_reorder_frames;
get_bits1(&s->gb); /* motion_vectors_over_pic_boundaries_flag */
get_ue_golomb(&s->gb); /* max_bytes_per_pic_denom */
get_ue_golomb(&s->gb); /* max_bits_per_mb_denom */
get_ue_golomb(&s->gb); /* log2_max_mv_length_horizontal */
get_ue_golomb(&s->gb); /* log2_max_mv_length_vertical */
- sps->num_reorder_frames = get_ue_golomb(&s->gb);
- get_ue_golomb(&s->gb); /* max_dec_frame_buffering */
+ num_reorder_frames= get_ue_golomb(&s->gb);
+ get_ue_golomb(&s->gb); /*max_dec_frame_buffering*/
+
+ if(num_reorder_frames > 16 /*max_dec_frame_buffering || max_dec_frame_buffering > 16*/){
+ av_log(h->s.avctx, AV_LOG_ERROR, "illegal num_reorder_frames %d\n", num_reorder_frames);
+ return -1;
+ }
+
+ sps->num_reorder_frames= num_reorder_frames;
}
return 0;
}
}
+/**
+ * Returns and optionally allocates SPS / PPS structures in the supplied array 'vec'
+ */
+static void *
+alloc_parameter_set(H264Context *h, void **vec, const unsigned int id, const unsigned int max,
+ const size_t size, const char *name)
+{
+ if(id>=max) {
+ av_log(h->s.avctx, AV_LOG_ERROR, "%s_id (%d) out of range\n", name, id);
+ return NULL;
+ }
+
+ if(!vec[id]) {
+ vec[id] = av_mallocz(size);
+ if(vec[id] == NULL)
+ av_log(h->s.avctx, AV_LOG_ERROR, "cannot allocate memory for %s\n", name);
+ }
+ return vec[id];
+}
+
static inline int decode_seq_parameter_set(H264Context *h){
MpegEncContext * const s = &h->s;
int profile_idc, level_idc;
- int sps_id, i;
+ unsigned int sps_id, tmp, mb_width, mb_height;
+ int i;
SPS *sps;
profile_idc= get_bits(&s->gb, 8);
level_idc= get_bits(&s->gb, 8);
sps_id= get_ue_golomb(&s->gb);
- sps= &h->sps_buffer[ sps_id ];
+ sps = alloc_parameter_set(h, (void **)h->sps_buffers, sps_id, MAX_SPS_COUNT, sizeof(SPS), "sps");
+ if(sps == NULL)
+ return -1;
+
sps->profile_idc= profile_idc;
sps->level_idc= level_idc;
sps->delta_pic_order_always_zero_flag= get_bits1(&s->gb);
sps->offset_for_non_ref_pic= get_se_golomb(&s->gb);
sps->offset_for_top_to_bottom_field= get_se_golomb(&s->gb);
- sps->poc_cycle_length= get_ue_golomb(&s->gb);
+ tmp= get_ue_golomb(&s->gb);
+
+ if(tmp >= sizeof(sps->offset_for_ref_frame) / sizeof(sps->offset_for_ref_frame[0])){
+ av_log(h->s.avctx, AV_LOG_ERROR, "poc_cycle_length overflow %u\n", tmp);
+ return -1;
+ }
+ sps->poc_cycle_length= tmp;
for(i=0; i<sps->poc_cycle_length; i++)
sps->offset_for_ref_frame[i]= get_se_golomb(&s->gb);
- }
- if(sps->poc_type > 2){
+ }else if(sps->poc_type != 2){
av_log(h->s.avctx, AV_LOG_ERROR, "illegal POC type %d\n", sps->poc_type);
return -1;
}
- sps->ref_frame_count= get_ue_golomb(&s->gb);
- if(sps->ref_frame_count > MAX_PICTURE_COUNT-2){
+ tmp= get_ue_golomb(&s->gb);
+ if(tmp > MAX_PICTURE_COUNT-2){
av_log(h->s.avctx, AV_LOG_ERROR, "too many reference frames\n");
}
+ sps->ref_frame_count= tmp;
sps->gaps_in_frame_num_allowed_flag= get_bits1(&s->gb);
- sps->mb_width= get_ue_golomb(&s->gb) + 1;
- sps->mb_height= get_ue_golomb(&s->gb) + 1;
- if((unsigned)sps->mb_width >= INT_MAX/16 || (unsigned)sps->mb_height >= INT_MAX/16 ||
- avcodec_check_dimensions(NULL, 16*sps->mb_width, 16*sps->mb_height))
+ mb_width= get_ue_golomb(&s->gb) + 1;
+ mb_height= get_ue_golomb(&s->gb) + 1;
+ if(mb_width >= INT_MAX/16 || mb_height >= INT_MAX/16 ||
+ avcodec_check_dimensions(NULL, 16*mb_width, 16*mb_height)){
+ av_log(h->s.avctx, AV_LOG_ERROR, "mb_width/height overflow\n");
return -1;
+ }
+ sps->mb_width = mb_width;
+ sps->mb_height= mb_height;
sps->frame_mbs_only_flag= get_bits1(&s->gb);
if(!sps->frame_mbs_only_flag)
#ifndef ALLOW_INTERLACE
if(sps->mb_aff)
- av_log(h->s.avctx, AV_LOG_ERROR, "MBAFF support not included; enable it compilation time\n");
+ av_log(h->s.avctx, AV_LOG_ERROR, "MBAFF support not included; enable it at compile-time.\n");
#endif
if(!sps->direct_8x8_inference_flag && sps->mb_aff)
av_log(h->s.avctx, AV_LOG_ERROR, "MBAFF + !direct_8x8_inference is not implemented\n");
decode_vui_parameters(h, sps);
if(s->avctx->debug&FF_DEBUG_PICT_INFO){
- av_log(h->s.avctx, AV_LOG_DEBUG, "sps:%d profile:%d/%d poc:%d ref:%d %dx%d %s %s crop:%d/%d/%d/%d %s\n",
+ av_log(h->s.avctx, AV_LOG_DEBUG, "sps:%u profile:%d/%d poc:%d ref:%d %dx%d %s %s crop:%d/%d/%d/%d %s\n",
sps_id, sps->profile_idc, sps->level_idc,
sps->poc_type,
sps->ref_frame_count,
static inline int decode_picture_parameter_set(H264Context *h, int bit_length){
MpegEncContext * const s = &h->s;
- int pps_id= get_ue_golomb(&s->gb);
- PPS *pps= &h->pps_buffer[pps_id];
+ unsigned int tmp, pps_id= get_ue_golomb(&s->gb);
+ PPS *pps;
+
+ pps = alloc_parameter_set(h, (void **)h->pps_buffers, pps_id, MAX_PPS_COUNT, sizeof(PPS), "pps");
+ if(pps == NULL)
+ return -1;
+
+ tmp= get_ue_golomb(&s->gb);
+ if(tmp>=MAX_SPS_COUNT || h->sps_buffers[tmp] == NULL){
+ av_log(h->s.avctx, AV_LOG_ERROR, "sps_id out of range\n");
+ return -1;
+ }
+ pps->sps_id= tmp;
- pps->sps_id= get_ue_golomb(&s->gb);
pps->cabac= get_bits1(&s->gb);
pps->pic_order_present= get_bits1(&s->gb);
pps->slice_group_count= get_ue_golomb(&s->gb) + 1;
}
pps->ref_count[0]= get_ue_golomb(&s->gb) + 1;
pps->ref_count[1]= get_ue_golomb(&s->gb) + 1;
- if(pps->ref_count[0] > 32 || pps->ref_count[1] > 32){
+ if(pps->ref_count[0]-1 > 32-1 || pps->ref_count[1]-1 > 32-1){
av_log(h->s.avctx, AV_LOG_ERROR, "reference overflow (pps)\n");
+ pps->ref_count[0]= pps->ref_count[1]= 1;
return -1;
}
if(get_bits_count(&s->gb) < bit_length){
pps->transform_8x8_mode= get_bits1(&s->gb);
- decode_scaling_matrices(h, &h->sps_buffer[pps->sps_id], pps, 0, pps->scaling_matrix4, pps->scaling_matrix8);
+ decode_scaling_matrices(h, h->sps_buffers[pps->sps_id], pps, 0, pps->scaling_matrix4, pps->scaling_matrix8);
get_se_golomb(&s->gb); //second_chroma_qp_index_offset
}
if(s->avctx->debug&FF_DEBUG_PICT_INFO){
- av_log(h->s.avctx, AV_LOG_DEBUG, "pps:%d sps:%d %s slice_groups:%d ref:%d/%d %s qp:%d/%d/%d %s %s %s %s\n",
+ av_log(h->s.avctx, AV_LOG_DEBUG, "pps:%u sps:%u %s slice_groups:%d ref:%d/%d %s qp:%d/%d/%d %s %s %s %s\n",
pps_id, pps->sps_id,
pps->cabac ? "CABAC" : "CAVLC",
pps->slice_group_count,
return 0;
}
-/**
- * finds the end of the current frame in the bitstream.
- * @return the position of the first byte of the next frame, or -1
- */
-static int find_frame_end(H264Context *h, const uint8_t *buf, int buf_size){
- int i;
- uint32_t state;
- ParseContext *pc = &(h->s.parse_context);
-//printf("first %02X%02X%02X%02X\n", buf[0], buf[1],buf[2],buf[3]);
-// mb_addr= pc->mb_addr - 1;
- state= pc->state;
- for(i=0; i<=buf_size; i++){
- if((state&0xFFFFFF1F) == 0x101 || (state&0xFFFFFF1F) == 0x102 || (state&0xFFFFFF1F) == 0x105){
- tprintf("find_frame_end new startcode = %08x, frame_start_found = %d, pos = %d\n", state, pc->frame_start_found, i);
- if(pc->frame_start_found){
- // If there isn't one more byte in the buffer
- // the test on first_mb_in_slice cannot be done yet
- // do it at next call.
- if (i >= buf_size) break;
- if (buf[i] & 0x80) {
- // first_mb_in_slice is 0, probably the first nal of a new
- // slice
- tprintf("find_frame_end frame_end_found, state = %08x, pos = %d\n", state, i);
- pc->state=-1;
- pc->frame_start_found= 0;
- return i-4;
- }
- }
- pc->frame_start_found = 1;
- }
- if((state&0xFFFFFF1F) == 0x107 || (state&0xFFFFFF1F) == 0x108 || (state&0xFFFFFF1F) == 0x109){
- if(pc->frame_start_found){
- pc->state=-1;
- pc->frame_start_found= 0;
- return i-4;
- }
- }
- if (i<buf_size)
- state= (state<<8) | buf[i];
- }
-
- pc->state= state;
- return END_NOT_FOUND;
-}
-
-static int h264_parse(AVCodecParserContext *s,
- AVCodecContext *avctx,
- uint8_t **poutbuf, int *poutbuf_size,
- const uint8_t *buf, int buf_size)
-{
- H264Context *h = s->priv_data;
- ParseContext *pc = &h->s.parse_context;
- int next;
-
- next= find_frame_end(h, buf, buf_size);
-
- if (ff_combine_frame(pc, next, (uint8_t **)&buf, &buf_size) < 0) {
- *poutbuf = NULL;
- *poutbuf_size = 0;
- return buf_size;
- }
-
- *poutbuf = (uint8_t *)buf;
- *poutbuf_size = buf_size;
- return next;
-}
-
-static int h264_split(AVCodecContext *avctx,
- const uint8_t *buf, int buf_size)
-{
- int i;
- uint32_t state = -1;
- int has_sps= 0;
-
- for(i=0; i<=buf_size; i++){
- if((state&0xFFFFFF1F) == 0x107)
- has_sps=1;
-/* if((state&0xFFFFFF1F) == 0x101 || (state&0xFFFFFF1F) == 0x102 || (state&0xFFFFFF1F) == 0x105){
- }*/
- if((state&0xFFFFFF00) == 0x100 && (state&0xFFFFFF1F) != 0x107 && (state&0xFFFFFF1F) != 0x108 && (state&0xFFFFFF1F) != 0x109){
- if(has_sps){
- while(i>4 && buf[i-5]==0) i--;
- return i-4;
- }
- }
- if (i<buf_size)
- state= (state<<8) | buf[i];
- }
- return 0;
-}
-
-
static int decode_nal_units(H264Context *h, uint8_t *buf, int buf_size){
MpegEncContext * const s = &h->s;
AVCodecContext * const avctx= s->avctx;
av_log(NULL, AV_LOG_ERROR,"%02X ", buf[i]);
}
#endif
- h->slice_num = 0;
- s->current_picture_ptr= NULL;
+ if(!(s->flags2 & CODEC_FLAG2_CHUNKS)){
+ h->slice_num = 0;
+ s->current_picture_ptr= NULL;
+ }
+
for(;;){
int consumed;
int dst_length;
nalsize = 0;
for(i = 0; i < h->nal_length_size; i++)
nalsize = (nalsize << 8) | buf[buf_index++];
- if(nalsize <= 1){
+ if(nalsize <= 1 || (nalsize+buf_index > buf_size)){
if(nalsize == 1){
buf_index++;
continue;
} else {
// start code prefix search
for(; buf_index + 3 < buf_size; buf_index++){
- // this should allways succeed in the first iteration
+ // This should always succeed in the first iteration.
if(buf[buf_index] == 0 && buf[buf_index+1] == 0 && buf[buf_index+2] == 1)
break;
}
}
ptr= decode_nal(h, buf + buf_index, &dst_length, &consumed, h->is_avc ? nalsize : buf_size - buf_index);
- if(ptr[dst_length - 1] == 0) dst_length--;
- bit_length= 8*dst_length - decode_rbsp_trailing(ptr + dst_length - 1);
+ if (ptr==NULL || dst_length < 0){
+ return -1;
+ }
+ while(ptr[dst_length - 1] == 0 && dst_length > 0)
+ dst_length--;
+ bit_length= !dst_length ? 0 : (8*dst_length - decode_rbsp_trailing(h, ptr + dst_length - 1));
if(s->avctx->debug&FF_DEBUG_STARTCODE){
av_log(h->s.avctx, AV_LOG_DEBUG, "NAL %d at %d/%d length %d\n", h->nal_unit_type, buf_index, buf_size, dst_length);
buf_index += consumed;
- if( (s->hurry_up == 1 && h->nal_ref_idc == 0) //FIXME dont discard SEI id
+ if( (s->hurry_up == 1 && h->nal_ref_idc == 0) //FIXME do not discard SEI id
||(avctx->skip_frame >= AVDISCARD_NONREF && h->nal_ref_idc == 0))
continue;
h->inter_gb_ptr= &h->inter_gb;
if(h->redundant_pic_count==0 && h->intra_gb_ptr && s->data_partitioning
+ && s->context_initialized
&& s->hurry_up < 5
&& (avctx->skip_frame < AVDISCARD_NONREF || h->nal_ref_idc)
&& (avctx->skip_frame < AVDISCARD_BIDIR || h->slice_type!=B_TYPE)
}
}
- if(!s->current_picture_ptr) return buf_index; //no frame
-
- s->current_picture_ptr->qscale_type= FF_QSCALE_TYPE_H264;
- s->current_picture_ptr->pict_type= s->pict_type;
-
- h->prev_frame_num_offset= h->frame_num_offset;
- h->prev_frame_num= h->frame_num;
- if(s->current_picture_ptr->reference){
- h->prev_poc_msb= h->poc_msb;
- h->prev_poc_lsb= h->poc_lsb;
- }
- if(s->current_picture_ptr->reference)
- execute_ref_pic_marking(h, h->mmco, h->mmco_index);
-
- ff_er_frame_end(s);
-
- MPV_frame_end(s);
-
return buf_index;
}
return pos;
}else{
- if(pos==0) pos=1; //avoid infinite loops (i doubt thats needed but ...)
+ if(pos==0) pos=1; //avoid infinite loops (i doubt that is needed but ...)
if(pos+10>buf_size) pos=buf_size; // oops ;)
return pos;
/* no supplementary picture */
if (buf_size == 0) {
+ Picture *out;
+ int i, out_idx;
+
+//FIXME factorize this with the output code below
+ out = h->delayed_pic[0];
+ out_idx = 0;
+ for(i=1; h->delayed_pic[i] && !h->delayed_pic[i]->key_frame; i++)
+ if(h->delayed_pic[i]->poc < out->poc){
+ out = h->delayed_pic[i];
+ out_idx = i;
+ }
+
+ for(i=out_idx; h->delayed_pic[i]; i++)
+ h->delayed_pic[i] = h->delayed_pic[i+1];
+
+ if(out){
+ *data_size = sizeof(AVFrame);
+ *pict= *(AVFrame*)out;
+ }
+
return 0;
}
if(s->flags&CODEC_FLAG_TRUNCATED){
- int next= find_frame_end(h, buf, buf_size);
+ int next= ff_h264_find_frame_end(h, buf, buf_size);
- if( ff_combine_frame(&s->parse_context, next, &buf, &buf_size) < 0 )
+ if( ff_combine_frame(&s->parse_context, next, (const uint8_t **)&buf, &buf_size) < 0 )
return buf_size;
//printf("next:%d buf_size:%d last_index:%d\n", next, buf_size, s->parse_context.last_index);
}
cnt = *(p+5) & 0x1f; // Number of sps
p += 6;
for (i = 0; i < cnt; i++) {
- nalsize = BE_16(p) + 2;
+ nalsize = AV_RB16(p) + 2;
if(decode_nal_units(h, p, nalsize) < 0) {
av_log(avctx, AV_LOG_ERROR, "Decoding sps %d from avcC failed\n", i);
return -1;
// Decode pps from avcC
cnt = *(p++); // Number of pps
for (i = 0; i < cnt; i++) {
- nalsize = BE_16(p) + 2;
+ nalsize = AV_RB16(p) + 2;
if(decode_nal_units(h, p, nalsize) != nalsize) {
av_log(avctx, AV_LOG_ERROR, "Decoding pps %d from avcC failed\n", i);
return -1;
h->got_avcC = 1;
}
- if(!h->is_avc && s->avctx->extradata_size && s->picture_number==0){
+ if(avctx->frame_number==0 && !h->is_avc && s->avctx->extradata_size){
if(decode_nal_units(h, s->avctx->extradata, s->avctx->extradata_size) < 0)
return -1;
}
if(buf_index < 0)
return -1;
- //FIXME do something with unavailable reference frames
-
-// if(ret==FRAME_SKIPPED) return get_consumed_bytes(s, buf_index, buf_size);
- if(!s->current_picture_ptr){
- av_log(h->s.avctx, AV_LOG_DEBUG, "error, NO frame\n");
+ if(!(s->flags2 & CODEC_FLAG2_CHUNKS) && !s->current_picture_ptr){
+ av_log(avctx, AV_LOG_ERROR, "no frame!\n");
return -1;
}
- {
+ if(!(s->flags2 & CODEC_FLAG2_CHUNKS) || (s->mb_y >= s->mb_height && s->mb_height)){
Picture *out = s->current_picture_ptr;
+ Picture *cur = s->current_picture_ptr;
+ Picture *prev = h->delayed_output_pic;
+ int i, pics, cross_idr, out_of_order, out_idx;
+
+ s->mb_y= 0;
+
+ s->current_picture_ptr->qscale_type= FF_QSCALE_TYPE_H264;
+ s->current_picture_ptr->pict_type= s->pict_type;
+
+ h->prev_frame_num_offset= h->frame_num_offset;
+ h->prev_frame_num= h->frame_num;
+ if(s->current_picture_ptr->reference){
+ h->prev_poc_msb= h->poc_msb;
+ h->prev_poc_lsb= h->poc_lsb;
+ }
+ if(s->current_picture_ptr->reference)
+ execute_ref_pic_marking(h, h->mmco, h->mmco_index);
+
+ ff_er_frame_end(s);
+
+ MPV_frame_end(s);
+
+ //FIXME do something with unavailable reference frames
+
#if 0 //decode order
*data_size = sizeof(AVFrame);
#else
/* Sort B-frames into display order */
- Picture *cur = s->current_picture_ptr;
- Picture *prev = h->delayed_output_pic;
- int i, pics, cross_idr, out_of_order, out_idx;
if(h->sps.bitstream_restriction_flag
&& s->avctx->has_b_frames < h->sps.num_reorder_frames){
pics = 0;
while(h->delayed_pic[pics]) pics++;
+
+ assert(pics+1 < sizeof(h->delayed_pic) / sizeof(h->delayed_pic[0]));
+
h->delayed_pic[pics++] = cur;
if(cur->reference == 0)
cur->reference = 1;
START_TIMER
j= get_ue_golomb(&gb);
if(j != i){
- printf("missmatch! at %d (%d should be %d) bits:%6X\n", i, j, i, s);
+ printf("mismatch! at %d (%d should be %d) bits:%6X\n", i, j, i, s);
// return -1;
}
STOP_TIMER("get_ue_golomb");
START_TIMER
j= get_se_golomb(&gb);
if(j != i - COUNT/2){
- printf("missmatch! at %d (%d should be %d) bits:%6X\n", i, j, i, s);
+ printf("mismatch! at %d (%d should be %d) bits:%6X\n", i, j, i, s);
// return -1;
}
STOP_TIMER("get_se_golomb");
printf("\n");*/
for(j=0; j<16; j++){
- int diff= ABS(src[j] - ref[j]);
+ int diff= FFABS(src[j] - ref[j]);
error+= diff*diff;
max_error= FFMAX(max_error, diff);
}
if(memcmp(bitstream, out, COUNT)){
- printf("missmatch\n");
+ printf("mismatch\n");
return -1;
}
}
.flush= flush_dpb,
};
-AVCodecParser h264_parser = {
- { CODEC_ID_H264 },
- sizeof(H264Context),
- NULL,
- h264_parse,
- ff_parse_close,
- h264_split,
-};
-
#include "svq3.c"