| Index: source/patched-ffmpeg-mt/libavcodec/ffv1.c | 
| =================================================================== | 
| --- source/patched-ffmpeg-mt/libavcodec/ffv1.c	(revision 65184) | 
| +++ source/patched-ffmpeg-mt/libavcodec/ffv1.c	(working copy) | 
| @@ -32,10 +32,14 @@ | 
| #include "rangecoder.h" | 
| #include "golomb.h" | 
| #include "mathops.h" | 
| +#include "libavutil/avassert.h" | 
|  | 
| #define MAX_PLANES 4 | 
| #define CONTEXT_SIZE 32 | 
|  | 
| +#define MAX_QUANT_TABLES 8 | 
| +#define MAX_CONTEXT_INPUTS 5 | 
| + | 
| extern const uint8_t ff_log2_run[32]; | 
|  | 
| static const int8_t quant3[256]={ | 
| @@ -213,12 +217,15 @@ | 
| } VlcState; | 
|  | 
| typedef struct PlaneContext{ | 
| +    int16_t quant_table[MAX_CONTEXT_INPUTS][256]; | 
| int context_count; | 
| uint8_t (*state)[CONTEXT_SIZE]; | 
| VlcState *vlc_state; | 
| uint8_t interlace_bit_state[2]; | 
| } PlaneContext; | 
|  | 
| +#define MAX_SLICES 256 | 
| + | 
| typedef struct FFV1Context{ | 
| AVCodecContext *avctx; | 
| RangeCoder c; | 
| @@ -233,13 +240,26 @@ | 
| int plane_count; | 
| int ac;                              ///< 1=range coder <-> 0=golomb rice | 
| PlaneContext plane[MAX_PLANES]; | 
| -    int16_t quant_table[5][256]; | 
| +    int16_t quant_table[MAX_CONTEXT_INPUTS][256]; | 
| +    int16_t quant_tables[MAX_QUANT_TABLES][MAX_CONTEXT_INPUTS][256]; | 
| +    int context_count[MAX_QUANT_TABLES]; | 
| uint8_t state_transition[256]; | 
| int run_index; | 
| int colorspace; | 
| int_fast16_t *sample_buffer; | 
|  | 
| +    int quant_table_count; | 
| + | 
| DSPContext dsp; | 
| + | 
| +    struct FFV1Context *slice_context[MAX_SLICES]; | 
| +    int slice_count; | 
| +    int num_v_slices; | 
| +    int num_h_slices; | 
| +    int slice_width; | 
| +    int slice_height; | 
| +    int slice_x; | 
| +    int slice_y; | 
| }FFV1Context; | 
|  | 
| static av_always_inline int fold(int diff, int bits){ | 
| @@ -262,19 +282,19 @@ | 
| return mid_pred(L, L + T - LT, T); | 
| } | 
|  | 
| -static inline int get_context(FFV1Context *f, int_fast16_t *src, int_fast16_t *last, int_fast16_t *last2){ | 
| +static inline int get_context(PlaneContext *p, int_fast16_t *src, int_fast16_t *last, int_fast16_t *last2){ | 
| const int LT= last[-1]; | 
| const int  T= last[ 0]; | 
| const int RT= last[ 1]; | 
| const int L =  src[-1]; | 
|  | 
| -    if(f->quant_table[3][127]){ | 
| +    if(p->quant_table[3][127]){ | 
| const int TT= last2[0]; | 
| const int LL=  src[-2]; | 
| -        return f->quant_table[0][(L-LT) & 0xFF] + f->quant_table[1][(LT-T) & 0xFF] + f->quant_table[2][(T-RT) & 0xFF] | 
| -              +f->quant_table[3][(LL-L) & 0xFF] + f->quant_table[4][(TT-T) & 0xFF]; | 
| +        return p->quant_table[0][(L-LT) & 0xFF] + p->quant_table[1][(LT-T) & 0xFF] + p->quant_table[2][(T-RT) & 0xFF] | 
| +              +p->quant_table[3][(LL-L) & 0xFF] + p->quant_table[4][(TT-T) & 0xFF]; | 
| }else | 
| -        return f->quant_table[0][(L-LT) & 0xFF] + f->quant_table[1][(LT-T) & 0xFF] + f->quant_table[2][(T-RT) & 0xFF]; | 
| +        return p->quant_table[0][(L-LT) & 0xFF] + p->quant_table[1][(LT-T) & 0xFF] + p->quant_table[2][(T-RT) & 0xFF]; | 
| } | 
|  | 
| static inline void put_symbol_inline(RangeCoder *c, uint8_t *state, int v, int is_signed){ | 
| @@ -452,7 +472,7 @@ | 
| for(x=0; x<w; x++){ | 
| int diff, context; | 
|  | 
| -        context= get_context(s, sample[0]+x, sample[1]+x, sample[2]+x); | 
| +        context= get_context(p, sample[0]+x, sample[1]+x, sample[2]+x); | 
| diff= sample[0][x] - predict(sample[0]+x, sample[1]+x); | 
|  | 
| if(context < 0){ | 
| @@ -591,31 +611,49 @@ | 
| put_symbol(c, state, i-last-1, 0); | 
| } | 
|  | 
| +static void write_quant_tables(RangeCoder *c, int16_t quant_table[MAX_CONTEXT_INPUTS][256]){ | 
| +    int i; | 
| +    for(i=0; i<5; i++) | 
| +        write_quant_table(c, quant_table[i]); | 
| +} | 
| + | 
| static void write_header(FFV1Context *f){ | 
| uint8_t state[CONTEXT_SIZE]; | 
| -    int i; | 
| -    RangeCoder * const c= &f->c; | 
| +    int i, j; | 
| +    RangeCoder * const c= &f->slice_context[0]->c; | 
|  | 
| memset(state, 128, sizeof(state)); | 
|  | 
| -    put_symbol(c, state, f->version, 0); | 
| -    put_symbol(c, state, f->ac, 0); | 
| -    if(f->ac>1){ | 
| -        for(i=1; i<256; i++){ | 
| -            f->state_transition[i]=ver2_state[i]; | 
| -            put_symbol(c, state, ver2_state[i] - c->one_state[i], 1); | 
| +    if(f->version < 2){ | 
| +        put_symbol(c, state, f->version, 0); | 
| +        put_symbol(c, state, f->ac, 0); | 
| +        if(f->ac>1){ | 
| +            for(i=1; i<256; i++){ | 
| +                f->state_transition[i]=ver2_state[i]; | 
| +                put_symbol(c, state, ver2_state[i] - c->one_state[i], 1); | 
| +            } | 
| } | 
| +        put_symbol(c, state, f->colorspace, 0); //YUV cs type | 
| +        if(f->version>0) | 
| +            put_symbol(c, state, f->avctx->bits_per_raw_sample, 0); | 
| +        put_rac(c, state, 1); //chroma planes | 
| +            put_symbol(c, state, f->chroma_h_shift, 0); | 
| +            put_symbol(c, state, f->chroma_v_shift, 0); | 
| +        put_rac(c, state, 0); //no transparency plane | 
| + | 
| +        write_quant_tables(c, f->quant_table); | 
| +    }else{ | 
| +        put_symbol(c, state, f->slice_count, 0); | 
| +        for(i=0; i<f->slice_count; i++){ | 
| +            FFV1Context *fs= f->slice_context[i]; | 
| +            put_symbol(c, state, (fs->slice_x     +1)*f->num_h_slices / f->width   , 0); | 
| +            put_symbol(c, state, (fs->slice_y     +1)*f->num_v_slices / f->height  , 0); | 
| +            put_symbol(c, state, (fs->slice_width +1)*f->num_h_slices / f->width -1, 0); | 
| +            put_symbol(c, state, (fs->slice_height+1)*f->num_v_slices / f->height-1, 0); | 
| +            for(j=0; j<f->plane_count; j++) | 
| +            put_symbol(c, state, f->avctx->context_model, 0); | 
| +        } | 
| } | 
| -    put_symbol(c, state, f->colorspace, 0); //YUV cs type | 
| -    if(f->version>0) | 
| -        put_symbol(c, state, f->avctx->bits_per_raw_sample, 0); | 
| -    put_rac(c, state, 1); //chroma planes | 
| -        put_symbol(c, state, f->chroma_h_shift, 0); | 
| -        put_symbol(c, state, f->chroma_v_shift, 0); | 
| -    put_rac(c, state, 0); //no transparency plane | 
| - | 
| -    for(i=0; i<5; i++) | 
| -        write_quant_table(c, f->quant_table[i]); | 
| } | 
| #endif /* CONFIG_FFV1_ENCODER */ | 
|  | 
| @@ -631,15 +669,110 @@ | 
| s->height= avctx->height; | 
|  | 
| assert(s->width && s->height); | 
| +    //defaults | 
| +    s->num_h_slices=1; | 
| +    s->num_v_slices=1; | 
|  | 
| -    s->sample_buffer = av_malloc(6 * (s->width+6) * sizeof(*s->sample_buffer)); | 
| -    if (!s->sample_buffer) | 
| -        return AVERROR(ENOMEM); | 
|  | 
| return 0; | 
| } | 
|  | 
| +static int init_slice_state(FFV1Context *f){ | 
| +    int i, j; | 
| + | 
| +    for(i=0; i<f->slice_count; i++){ | 
| +        FFV1Context *fs= f->slice_context[i]; | 
| +        for(j=0; j<f->plane_count; j++){ | 
| +            PlaneContext * const p= &fs->plane[j]; | 
| + | 
| +            if(fs->ac){ | 
| +                if(!p->    state) p->    state= av_malloc(CONTEXT_SIZE*p->context_count*sizeof(uint8_t)); | 
| +                if(!p->    state) | 
| +                    return AVERROR(ENOMEM); | 
| +            }else{ | 
| +                if(!p->vlc_state) p->vlc_state= av_malloc(p->context_count*sizeof(VlcState)); | 
| +                if(!p->vlc_state) | 
| +                    return AVERROR(ENOMEM); | 
| +            } | 
| +        } | 
| + | 
| +        if (fs->ac>1){ | 
| +            //FIXME only redo if state_transition changed | 
| +            for(j=1; j<256; j++){ | 
| +                fs->c.one_state [    j]= fs->state_transition[j]; | 
| +                fs->c.zero_state[256-j]= 256-fs->c.one_state [j]; | 
| +            } | 
| +        } | 
| +    } | 
| + | 
| +    return 0; | 
| +} | 
| + | 
| +static av_cold int init_slice_contexts(FFV1Context *f){ | 
| +    int i; | 
| + | 
| +    f->slice_count= f->num_h_slices * f->num_v_slices; | 
| + | 
| +    for(i=0; i<f->slice_count; i++){ | 
| +        FFV1Context *fs= av_mallocz(sizeof(*fs)); | 
| +        int sx= i % f->num_h_slices; | 
| +        int sy= i / f->num_h_slices; | 
| +        int sxs= f->avctx->width * sx    / f->num_h_slices; | 
| +        int sxe= f->avctx->width *(sx+1) / f->num_h_slices; | 
| +        int sys= f->avctx->height* sy    / f->num_v_slices; | 
| +        int sye= f->avctx->height*(sy+1) / f->num_v_slices; | 
| +        f->slice_context[i]= fs; | 
| +        memcpy(fs, f, sizeof(*fs)); | 
| + | 
| +        fs->slice_width = sxe - sxs; | 
| +        fs->slice_height= sye - sys; | 
| +        fs->slice_x     = sxs; | 
| +        fs->slice_y     = sys; | 
| + | 
| +        fs->sample_buffer = av_malloc(6 * (fs->width+6) * sizeof(*fs->sample_buffer)); | 
| +        if (!fs->sample_buffer) | 
| +            return AVERROR(ENOMEM); | 
| +    } | 
| +    return 0; | 
| +} | 
| + | 
| #if CONFIG_FFV1_ENCODER | 
| +static int write_extra_header(FFV1Context *f){ | 
| +    RangeCoder * const c= &f->c; | 
| +    uint8_t state[CONTEXT_SIZE]; | 
| +    int i; | 
| +    memset(state, 128, sizeof(state)); | 
| + | 
| +    f->avctx->extradata= av_malloc(f->avctx->extradata_size= 10000); | 
| +    ff_init_range_encoder(c, f->avctx->extradata, f->avctx->extradata_size); | 
| +    ff_build_rac_states(c, 0.05*(1LL<<32), 256-8); | 
| + | 
| +    put_symbol(c, state, f->version, 0); | 
| +    put_symbol(c, state, f->ac, 0); | 
| +    if(f->ac>1){ | 
| +        for(i=1; i<256; i++){ | 
| +            f->state_transition[i]=ver2_state[i]; | 
| +            put_symbol(c, state, ver2_state[i] - c->one_state[i], 1); | 
| +        } | 
| +    } | 
| +    put_symbol(c, state, f->colorspace, 0); //YUV cs type | 
| +    put_symbol(c, state, f->avctx->bits_per_raw_sample, 0); | 
| +    put_rac(c, state, 1); //chroma planes | 
| +        put_symbol(c, state, f->chroma_h_shift, 0); | 
| +        put_symbol(c, state, f->chroma_v_shift, 0); | 
| +    put_rac(c, state, 0); //no transparency plane | 
| +    put_symbol(c, state, f->num_h_slices-1, 0); | 
| +    put_symbol(c, state, f->num_v_slices-1, 0); | 
| + | 
| +    put_symbol(c, state, f->quant_table_count, 0); | 
| +    for(i=0; i<f->quant_table_count; i++) | 
| +        write_quant_tables(c, f->quant_tables[i]); | 
| + | 
| +    f->avctx->extradata_size= ff_rac_terminate(c); | 
| + | 
| +    return 0; | 
| +} | 
| + | 
| static av_cold int encode_init(AVCodecContext *avctx) | 
| { | 
| FFV1Context *s = avctx->priv_data; | 
| @@ -652,47 +785,38 @@ | 
|  | 
| s->plane_count=2; | 
| for(i=0; i<256; i++){ | 
| +        s->quant_table_count=2; | 
| if(avctx->bits_per_raw_sample <=8){ | 
| -            s->quant_table[0][i]=           quant11[i]; | 
| -            s->quant_table[1][i]=        11*quant11[i]; | 
| -            if(avctx->context_model==0){ | 
| -                s->quant_table[2][i]=     11*11*quant11[i]; | 
| -                s->quant_table[3][i]= | 
| -                s->quant_table[4][i]=0; | 
| -            }else{ | 
| -                s->quant_table[2][i]=     11*11*quant5 [i]; | 
| -                s->quant_table[3][i]=   5*11*11*quant5 [i]; | 
| -                s->quant_table[4][i]= 5*5*11*11*quant5 [i]; | 
| -            } | 
| +            s->quant_tables[0][0][i]=           quant11[i]; | 
| +            s->quant_tables[0][1][i]=        11*quant11[i]; | 
| +            s->quant_tables[0][2][i]=     11*11*quant11[i]; | 
| +            s->quant_tables[1][0][i]=           quant11[i]; | 
| +            s->quant_tables[1][1][i]=        11*quant11[i]; | 
| +            s->quant_tables[1][2][i]=     11*11*quant5 [i]; | 
| +            s->quant_tables[1][3][i]=   5*11*11*quant5 [i]; | 
| +            s->quant_tables[1][4][i]= 5*5*11*11*quant5 [i]; | 
| }else{ | 
| -            s->quant_table[0][i]=           quant9_10bit[i]; | 
| -            s->quant_table[1][i]=        11*quant9_10bit[i]; | 
| -            if(avctx->context_model==0){ | 
| -                s->quant_table[2][i]=     11*11*quant9_10bit[i]; | 
| -                s->quant_table[3][i]= | 
| -                s->quant_table[4][i]=0; | 
| -            }else{ | 
| -                s->quant_table[2][i]=     11*11*quant5_10bit[i]; | 
| -                s->quant_table[3][i]=   5*11*11*quant5_10bit[i]; | 
| -                s->quant_table[4][i]= 5*5*11*11*quant5_10bit[i]; | 
| -            } | 
| +            s->quant_tables[0][0][i]=           quant9_10bit[i]; | 
| +            s->quant_tables[0][1][i]=        11*quant9_10bit[i]; | 
| +            s->quant_tables[0][2][i]=     11*11*quant9_10bit[i]; | 
| +            s->quant_tables[1][0][i]=           quant9_10bit[i]; | 
| +            s->quant_tables[1][1][i]=        11*quant9_10bit[i]; | 
| +            s->quant_tables[1][2][i]=     11*11*quant5_10bit[i]; | 
| +            s->quant_tables[1][3][i]=   5*11*11*quant5_10bit[i]; | 
| +            s->quant_tables[1][4][i]= 5*5*11*11*quant5_10bit[i]; | 
| } | 
| } | 
| +    memcpy(s->quant_table, s->quant_tables[avctx->context_model], sizeof(s->quant_table)); | 
|  | 
| for(i=0; i<s->plane_count; i++){ | 
| PlaneContext * const p= &s->plane[i]; | 
|  | 
| +        memcpy(p->quant_table, s->quant_table, sizeof(p->quant_table)); | 
| if(avctx->context_model==0){ | 
| p->context_count= (11*11*11+1)/2; | 
| }else{ | 
| p->context_count= (11*11*5*5*5+1)/2; | 
| } | 
| - | 
| -        if(s->ac){ | 
| -            if(!p->state) p->state= av_malloc(CONTEXT_SIZE*p->context_count*sizeof(uint8_t)); | 
| -        }else{ | 
| -            if(!p->vlc_state) p->vlc_state= av_malloc(p->context_count*sizeof(VlcState)); | 
| -        } | 
| } | 
|  | 
| avctx->coded_frame= &s->picture; | 
| @@ -708,7 +832,7 @@ | 
| av_log(avctx, AV_LOG_ERROR, "bits_per_raw_sample of more than 8 needs -coder 1 currently\n"); | 
| return -1; | 
| } | 
| -        s->version= 1; | 
| +        s->version= FFMAX(s->version, 1); | 
| case PIX_FMT_YUV444P: | 
| case PIX_FMT_YUV422P: | 
| case PIX_FMT_YUV420P: | 
| @@ -727,43 +851,84 @@ | 
|  | 
| s->picture_number=0; | 
|  | 
| +    if(s->version>1){ | 
| +        s->num_h_slices=2; | 
| +        s->num_v_slices=2; | 
| +        write_extra_header(s); | 
| +    } | 
| + | 
| +    if(init_slice_contexts(s) < 0) | 
| +        return -1; | 
| +    if(init_slice_state(s) < 0) | 
| +        return -1; | 
| + | 
| return 0; | 
| } | 
| #endif /* CONFIG_FFV1_ENCODER */ | 
|  | 
|  | 
| static void clear_state(FFV1Context *f){ | 
| -    int i, j; | 
| +    int i, si, j; | 
|  | 
| -    for(i=0; i<f->plane_count; i++){ | 
| -        PlaneContext *p= &f->plane[i]; | 
| +    for(si=0; si<f->slice_count; si++){ | 
| +        FFV1Context *fs= f->slice_context[si]; | 
| +        for(i=0; i<f->plane_count; i++){ | 
| +            PlaneContext *p= &fs->plane[i]; | 
|  | 
| -        p->interlace_bit_state[0]= 128; | 
| -        p->interlace_bit_state[1]= 128; | 
| +            p->interlace_bit_state[0]= 128; | 
| +            p->interlace_bit_state[1]= 128; | 
|  | 
| -        for(j=0; j<p->context_count; j++){ | 
| -            if(f->ac){ | 
| -                memset(p->state[j], 128, sizeof(uint8_t)*CONTEXT_SIZE); | 
| -            }else{ | 
| -                p->vlc_state[j].drift= 0; | 
| -                p->vlc_state[j].error_sum= 4; //FFMAX((RANGE + 32)/64, 2); | 
| -                p->vlc_state[j].bias= 0; | 
| -                p->vlc_state[j].count= 1; | 
| +            for(j=0; j<p->context_count; j++){ | 
| +                if(fs->ac){ | 
| +                    memset(p->state[j], 128, sizeof(uint8_t)*CONTEXT_SIZE); | 
| +                }else{ | 
| +                    p->vlc_state[j].drift= 0; | 
| +                    p->vlc_state[j].error_sum= 4; //FFMAX((RANGE + 32)/64, 2); | 
| +                    p->vlc_state[j].bias= 0; | 
| +                    p->vlc_state[j].count= 1; | 
| +                } | 
| } | 
| } | 
| } | 
| } | 
|  | 
| #if CONFIG_FFV1_ENCODER | 
| +static int encode_slice(AVCodecContext *c, void *arg){ | 
| +    FFV1Context *fs= *(void**)arg; | 
| +    FFV1Context *f= fs->avctx->priv_data; | 
| +    int width = fs->slice_width; | 
| +    int height= fs->slice_height; | 
| +    int x= fs->slice_x; | 
| +    int y= fs->slice_y; | 
| +    AVFrame * const p= &f->picture; | 
| + | 
| +    if(f->colorspace==0){ | 
| +        const int chroma_width = -((-width )>>f->chroma_h_shift); | 
| +        const int chroma_height= -((-height)>>f->chroma_v_shift); | 
| +        const int cx= x>>f->chroma_h_shift; | 
| +        const int cy= y>>f->chroma_v_shift; | 
| + | 
| +        encode_plane(fs, p->data[0] + x + y*p->linesize[0], width, height, p->linesize[0], 0); | 
| + | 
| +        encode_plane(fs, p->data[1] + cx+cy*p->linesize[1], chroma_width, chroma_height, p->linesize[1], 1); | 
| +        encode_plane(fs, p->data[2] + cx+cy*p->linesize[2], chroma_width, chroma_height, p->linesize[2], 1); | 
| +    }else{ | 
| +        encode_rgb_frame(fs, (uint32_t*)(p->data[0]) + x + y*(p->linesize[0]/4), width, height, p->linesize[0]/4); | 
| +    } | 
| +    emms_c(); | 
| + | 
| +    return 0; | 
| +} | 
| + | 
| static int encode_frame(AVCodecContext *avctx, unsigned char *buf, int buf_size, void *data){ | 
| FFV1Context *f = avctx->priv_data; | 
| -    RangeCoder * const c= &f->c; | 
| +    RangeCoder * const c= &f->slice_context[0]->c; | 
| AVFrame *pict = data; | 
| -    const int width= f->width; | 
| -    const int height= f->height; | 
| AVFrame * const p= &f->picture; | 
| int used_count= 0; | 
| uint8_t keystate=128; | 
| +    uint8_t *buf_p; | 
| +    int i; | 
|  | 
| ff_init_range_encoder(c, buf, buf_size); | 
| ff_build_rac_states(c, 0.05*(1LL<<32), 256-8); | 
| @@ -784,7 +949,7 @@ | 
| if(!f->ac){ | 
| used_count += ff_rac_terminate(c); | 
| //printf("pos=%d\n", used_count); | 
| -        init_put_bits(&f->pb, buf + used_count, buf_size - used_count); | 
| +        init_put_bits(&f->slice_context[0]->pb, buf + used_count, buf_size - used_count); | 
| }else if (f->ac>1){ | 
| int i; | 
| for(i=1; i<256; i++){ | 
| @@ -793,43 +958,63 @@ | 
| } | 
| } | 
|  | 
| -    if(f->colorspace==0){ | 
| -        const int chroma_width = -((-width )>>f->chroma_h_shift); | 
| -        const int chroma_height= -((-height)>>f->chroma_v_shift); | 
| +    for(i=1; i<f->slice_count; i++){ | 
| +        FFV1Context *fs= f->slice_context[i]; | 
| +        uint8_t *start= buf + (buf_size-used_count)*i/f->slice_count; | 
| +        int len= buf_size/f->slice_count; | 
|  | 
| -        encode_plane(f, p->data[0], width, height, p->linesize[0], 0); | 
| +        if(fs->ac){ | 
| +            ff_init_range_encoder(&fs->c, start, len); | 
| +        }else{ | 
| +            init_put_bits(&fs->pb, start, len); | 
| +        } | 
| +    } | 
| +    avctx->execute(avctx, encode_slice, &f->slice_context[0], NULL, f->slice_count, sizeof(void*)); | 
|  | 
| -        encode_plane(f, p->data[1], chroma_width, chroma_height, p->linesize[1], 1); | 
| -        encode_plane(f, p->data[2], chroma_width, chroma_height, p->linesize[2], 1); | 
| -    }else{ | 
| -        encode_rgb_frame(f, (uint32_t*)(p->data[0]), width, height, p->linesize[0]/4); | 
| +    buf_p=buf; | 
| +    for(i=0; i<f->slice_count; i++){ | 
| +        FFV1Context *fs= f->slice_context[i]; | 
| +        int bytes; | 
| + | 
| +        if(fs->ac){ | 
| +            uint8_t state=128; | 
| +            put_rac(&fs->c, &state, 0); | 
| +            bytes= ff_rac_terminate(&fs->c); | 
| +        }else{ | 
| +            flush_put_bits(&fs->pb); //nicer padding FIXME | 
| +            bytes= used_count + (put_bits_count(&fs->pb)+7)/8; | 
| +            used_count= 0; | 
| +        } | 
| +        if(i>0){ | 
| +            av_assert0(bytes < buf_size/f->slice_count); | 
| +            memmove(buf_p, fs->ac ? fs->c.bytestream_start : fs->pb.buf, bytes); | 
| +            av_assert0(bytes < (1<<24)); | 
| +            AV_WB24(buf_p+bytes, bytes); | 
| +            bytes+=3; | 
| +        } | 
| +        buf_p += bytes; | 
| } | 
| -    emms_c(); | 
|  | 
| f->picture_number++; | 
| - | 
| -    if(f->ac){ | 
| -        return ff_rac_terminate(c); | 
| -    }else{ | 
| -        flush_put_bits(&f->pb); //nicer padding FIXME | 
| -        return used_count + (put_bits_count(&f->pb)+7)/8; | 
| -    } | 
| +    return buf_p-buf; | 
| } | 
| #endif /* CONFIG_FFV1_ENCODER */ | 
|  | 
| static av_cold int common_end(AVCodecContext *avctx){ | 
| FFV1Context *s = avctx->priv_data; | 
| -    int i; | 
| +    int i, j; | 
|  | 
| -    for(i=0; i<s->plane_count; i++){ | 
| -        PlaneContext *p= &s->plane[i]; | 
| +    for(j=0; j<s->slice_count; j++){ | 
| +        FFV1Context *fs= s->slice_context[j]; | 
| +        for(i=0; i<s->plane_count; i++){ | 
| +            PlaneContext *p= &fs->plane[i]; | 
|  | 
| -        av_freep(&p->state); | 
| -        av_freep(&p->vlc_state); | 
| +            av_freep(&p->state); | 
| +            av_freep(&p->vlc_state); | 
| +        } | 
| +        av_freep(&fs->sample_buffer); | 
| } | 
|  | 
| -    av_freep(&s->sample_buffer); | 
| - | 
| return 0; | 
| } | 
|  | 
| @@ -844,13 +1029,14 @@ | 
| for(x=0; x<w; x++){ | 
| int diff, context, sign; | 
|  | 
| -        context= get_context(s, sample[1] + x, sample[0] + x, sample[1] + x); | 
| +        context= get_context(p, sample[1] + x, sample[0] + x, sample[1] + x); | 
| if(context < 0){ | 
| context= -context; | 
| sign=1; | 
| }else | 
| sign=0; | 
|  | 
| +        av_assert2(context < p->context_count); | 
|  | 
| if(s->ac){ | 
| diff= get_symbol_inline(c, p->state[context], 1); | 
| @@ -967,6 +1153,34 @@ | 
| } | 
| } | 
|  | 
| +static int decode_slice(AVCodecContext *c, void *arg){ | 
| +    FFV1Context *fs= *(void**)arg; | 
| +    FFV1Context *f= fs->avctx->priv_data; | 
| +    int width = fs->slice_width; | 
| +    int height= fs->slice_height; | 
| +    int x= fs->slice_x; | 
| +    int y= fs->slice_y; | 
| +    AVFrame * const p= &f->picture; | 
| + | 
| +    av_assert1(width && height); | 
| +    if(f->colorspace==0){ | 
| +        const int chroma_width = -((-width )>>f->chroma_h_shift); | 
| +        const int chroma_height= -((-height)>>f->chroma_v_shift); | 
| +        const int cx= x>>f->chroma_h_shift; | 
| +        const int cy= y>>f->chroma_v_shift; | 
| +        decode_plane(fs, p->data[0] + x + y*p->linesize[0], width, height, p->linesize[0], 0); | 
| + | 
| +        decode_plane(fs, p->data[1] + cx+cy*p->linesize[1], chroma_width, chroma_height, p->linesize[1], 1); | 
| +        decode_plane(fs, p->data[2] + cx+cy*p->linesize[1], chroma_width, chroma_height, p->linesize[2], 1); | 
| +    }else{ | 
| +        decode_rgb_frame(fs, (uint32_t*)p->data[0] + x + y*(p->linesize[0]/4), width, height, p->linesize[0]/4); | 
| +    } | 
| + | 
| +    emms_c(); | 
| + | 
| +    return 0; | 
| +} | 
| + | 
| static int read_quant_table(RangeCoder *c, int16_t *quant_table, int scale){ | 
| int v; | 
| int i=0; | 
| @@ -995,13 +1209,29 @@ | 
| return 2*v - 1; | 
| } | 
|  | 
| -static int read_header(FFV1Context *f){ | 
| +static int read_quant_tables(RangeCoder *c, int16_t quant_table[MAX_CONTEXT_INPUTS][256]){ | 
| +    int i; | 
| +    int context_count=1; | 
| + | 
| +    for(i=0; i<5; i++){ | 
| +        context_count*= read_quant_table(c, quant_table[i], context_count); | 
| +        if(context_count > 32768U){ | 
| +            return -1; | 
| +        } | 
| +    } | 
| +    return (context_count+1)/2; | 
| +} | 
| + | 
| +static int read_extra_header(FFV1Context *f){ | 
| +    RangeCoder * const c= &f->c; | 
| uint8_t state[CONTEXT_SIZE]; | 
| -    int i, context_count; | 
| -    RangeCoder * const c= &f->c; | 
| +    int i; | 
|  | 
| memset(state, 128, sizeof(state)); | 
|  | 
| +    ff_init_range_decoder(c, f->avctx->extradata, f->avctx->extradata_size); | 
| +    ff_build_rac_states(c, 0.05*(1LL<<32), 256-8); | 
| + | 
| f->version= get_symbol(c, state, 0); | 
| f->ac= f->avctx->coder_type= get_symbol(c, state, 0); | 
| if(f->ac>1){ | 
| @@ -1010,14 +1240,57 @@ | 
| } | 
| } | 
| f->colorspace= get_symbol(c, state, 0); //YUV cs type | 
| -    if(f->version>0) | 
| -        f->avctx->bits_per_raw_sample= get_symbol(c, state, 0); | 
| +    f->avctx->bits_per_raw_sample= get_symbol(c, state, 0); | 
| get_rac(c, state); //no chroma = false | 
| f->chroma_h_shift= get_symbol(c, state, 0); | 
| f->chroma_v_shift= get_symbol(c, state, 0); | 
| get_rac(c, state); //transparency plane | 
| f->plane_count= 2; | 
| +    f->num_h_slices= 1 + get_symbol(c, state, 0); | 
| +    f->num_v_slices= 1 + get_symbol(c, state, 0); | 
| +    if(f->num_h_slices > (unsigned)f->width || f->num_v_slices > (unsigned)f->height){ | 
| +        av_log(f->avctx, AV_LOG_ERROR, "too many slices\n"); | 
| +        return -1; | 
| +    } | 
|  | 
| +    f->quant_table_count= get_symbol(c, state, 0); | 
| +    if(f->quant_table_count > (unsigned)MAX_QUANT_TABLES) | 
| +        return -1; | 
| +    for(i=0; i<f->quant_table_count; i++){ | 
| +        if((f->context_count[i]= read_quant_tables(c, f->quant_tables[i])) < 0){ | 
| +            av_log(f->avctx, AV_LOG_ERROR, "read_quant_table error\n"); | 
| +            return -1; | 
| +        } | 
| +    } | 
| + | 
| +    return 0; | 
| +} | 
| + | 
| +static int read_header(FFV1Context *f){ | 
| +    uint8_t state[CONTEXT_SIZE]; | 
| +    int i, j, context_count; | 
| +    RangeCoder * const c= &f->slice_context[0]->c; | 
| + | 
| +    memset(state, 128, sizeof(state)); | 
| + | 
| +    if(f->version < 2){ | 
| +        f->version= get_symbol(c, state, 0); | 
| +        f->ac= f->avctx->coder_type= get_symbol(c, state, 0); | 
| +        if(f->ac>1){ | 
| +            for(i=1; i<256; i++){ | 
| +                f->state_transition[i]= get_symbol(c, state, 1) + c->one_state[i]; | 
| +            } | 
| +        } | 
| +        f->colorspace= get_symbol(c, state, 0); //YUV cs type | 
| +        if(f->version>0) | 
| +            f->avctx->bits_per_raw_sample= get_symbol(c, state, 0); | 
| +        get_rac(c, state); //no chroma = false | 
| +        f->chroma_h_shift= get_symbol(c, state, 0); | 
| +        f->chroma_v_shift= get_symbol(c, state, 0); | 
| +        get_rac(c, state); //transparency plane | 
| +        f->plane_count= 2; | 
| +    } | 
| + | 
| if(f->colorspace==0){ | 
| if(f->avctx->bits_per_raw_sample<=8){ | 
| switch(16*f->chroma_h_shift + f->chroma_v_shift){ | 
| @@ -1052,27 +1325,60 @@ | 
| } | 
|  | 
| //printf("%d %d %d\n", f->chroma_h_shift, f->chroma_v_shift,f->avctx->pix_fmt); | 
| - | 
| -    context_count=1; | 
| -    for(i=0; i<5; i++){ | 
| -        context_count*= read_quant_table(c, f->quant_table[i], context_count); | 
| -        if(context_count < 0 || context_count > 32768){ | 
| -            av_log(f->avctx, AV_LOG_ERROR, "read_quant_table error\n"); | 
| +    if(f->version < 2){ | 
| +        context_count= read_quant_tables(c, f->quant_table); | 
| +        if(context_count < 0){ | 
| +                av_log(f->avctx, AV_LOG_ERROR, "read_quant_table error\n"); | 
| +                return -1; | 
| +        } | 
| +    }else{ | 
| +        f->slice_count= get_symbol(c, state, 0); | 
| +        if(f->slice_count > (unsigned)MAX_SLICES) | 
| return -1; | 
| -        } | 
| } | 
| -    context_count= (context_count+1)/2; | 
|  | 
| -    for(i=0; i<f->plane_count; i++){ | 
| -        PlaneContext * const p= &f->plane[i]; | 
| +    for(j=0; j<f->slice_count; j++){ | 
| +        FFV1Context *fs= f->slice_context[j]; | 
| +        fs->ac= f->ac; | 
|  | 
| -        p->context_count= context_count; | 
| +        if(f->version >= 2){ | 
| +            fs->slice_x     = get_symbol(c, state, 0)   *f->width ; | 
| +            fs->slice_y     = get_symbol(c, state, 0)   *f->height; | 
| +            fs->slice_width =(get_symbol(c, state, 0)+1)*f->width  + fs->slice_x; | 
| +            fs->slice_height=(get_symbol(c, state, 0)+1)*f->height + fs->slice_y; | 
|  | 
| -        if(f->ac){ | 
| -            if(!p->state) p->state= av_malloc(CONTEXT_SIZE*p->context_count*sizeof(uint8_t)); | 
| -        }else{ | 
| -            if(!p->vlc_state) p->vlc_state= av_malloc(p->context_count*sizeof(VlcState)); | 
| +            fs->slice_x /= f->num_h_slices; | 
| +            fs->slice_y /= f->num_v_slices; | 
| +            fs->slice_width  = fs->slice_width /f->num_h_slices - fs->slice_x; | 
| +            fs->slice_height = fs->slice_height/f->num_v_slices - fs->slice_y; | 
| +            if((unsigned)fs->slice_width > f->width || (unsigned)fs->slice_height > f->height) | 
| +                return -1; | 
| +            if(    (unsigned)fs->slice_x + (uint64_t)fs->slice_width  > f->width | 
| +                || (unsigned)fs->slice_y + (uint64_t)fs->slice_height > f->height) | 
| +                return -1; | 
| } | 
| + | 
| +        for(i=0; i<f->plane_count; i++){ | 
| +            PlaneContext * const p= &fs->plane[i]; | 
| + | 
| +            if(f->version >= 2){ | 
| +                int idx=get_symbol(c, state, 0); | 
| +                if(idx > (unsigned)f->quant_table_count){ | 
| +                    av_log(f->avctx, AV_LOG_ERROR, "quant_table_index out of range\n"); | 
| +                    return -1; | 
| +                } | 
| +                memcpy(p->quant_table, f->quant_tables[idx], sizeof(p->quant_table)); | 
| +                context_count= f->context_count[idx]; | 
| +            }else{ | 
| +                memcpy(p->quant_table, f->quant_table, sizeof(p->quant_table)); | 
| +            } | 
| + | 
| +            if(p->context_count < context_count){ | 
| +                av_freep(&p->state); | 
| +                av_freep(&p->vlc_state); | 
| +            } | 
| +            p->context_count= context_count; | 
| +        } | 
| } | 
|  | 
| return 0; | 
| @@ -1080,10 +1386,16 @@ | 
|  | 
| static av_cold int decode_init(AVCodecContext *avctx) | 
| { | 
| -//    FFV1Context *s = avctx->priv_data; | 
| +    FFV1Context *f = avctx->priv_data; | 
|  | 
| common_init(avctx); | 
|  | 
| +    if(avctx->extradata && read_extra_header(f) < 0) | 
| +        return -1; | 
| + | 
| +    if(init_slice_contexts(f) < 0) | 
| +        return -1; | 
| + | 
| return 0; | 
| } | 
|  | 
| @@ -1091,12 +1403,11 @@ | 
| const uint8_t *buf = avpkt->data; | 
| int buf_size = avpkt->size; | 
| FFV1Context *f = avctx->priv_data; | 
| -    RangeCoder * const c= &f->c; | 
| -    const int width= f->width; | 
| -    const int height= f->height; | 
| +    RangeCoder * const c= &f->slice_context[0]->c; | 
| AVFrame * const p= &f->picture; | 
| -    int bytes_read; | 
| +    int bytes_read, i; | 
| uint8_t keystate= 128; | 
| +    const uint8_t *buf_p; | 
|  | 
| AVFrame *picture = data; | 
|  | 
| @@ -1109,6 +1420,9 @@ | 
| p->key_frame= 1; | 
| if(read_header(f) < 0) | 
| return -1; | 
| +        if(init_slice_state(f) < 0) | 
| +            return -1; | 
| + | 
| clear_state(f); | 
| }else{ | 
| p->key_frame= 0; | 
| @@ -1121,9 +1435,6 @@ | 
| } | 
| } | 
|  | 
| -    if(!f->plane[0].state && !f->plane[0].vlc_state) | 
| -        return -1; | 
| - | 
| p->reference= 0; | 
| if(avctx->get_buffer(avctx, p) < 0){ | 
| av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n"); | 
| @@ -1137,24 +1448,28 @@ | 
| bytes_read = c->bytestream - c->bytestream_start - 1; | 
| if(bytes_read ==0) av_log(avctx, AV_LOG_ERROR, "error at end of AC stream\n"); //FIXME | 
| //printf("pos=%d\n", bytes_read); | 
| -        init_get_bits(&f->gb, buf + bytes_read, buf_size - bytes_read); | 
| +        init_get_bits(&f->slice_context[0]->gb, buf + bytes_read, buf_size - bytes_read); | 
| } else { | 
| bytes_read = 0; /* avoid warning */ | 
| } | 
|  | 
| -    if(f->colorspace==0){ | 
| -        const int chroma_width = -((-width )>>f->chroma_h_shift); | 
| -        const int chroma_height= -((-height)>>f->chroma_v_shift); | 
| -        decode_plane(f, p->data[0], width, height, p->linesize[0], 0); | 
| - | 
| -        decode_plane(f, p->data[1], chroma_width, chroma_height, p->linesize[1], 1); | 
| -        decode_plane(f, p->data[2], chroma_width, chroma_height, p->linesize[2], 1); | 
| -    }else{ | 
| -        decode_rgb_frame(f, (uint32_t*)p->data[0], width, height, p->linesize[0]/4); | 
| +    buf_p= buf + buf_size; | 
| +    for(i=f->slice_count-1; i>0; i--){ | 
| +        FFV1Context *fs= f->slice_context[i]; | 
| +        int v= AV_RB24(buf_p-3)+3; | 
| +        if(buf_p - buf <= v){ | 
| +            av_log(avctx, AV_LOG_ERROR, "Slice pointer chain broken\n"); | 
| +            return -1; | 
| +        } | 
| +        buf_p -= v; | 
| +        if(fs->ac){ | 
| +            ff_init_range_decoder(&fs->c, buf_p, v); | 
| +        }else{ | 
| +            init_get_bits(&fs->gb, buf_p, v); | 
| +        } | 
| } | 
|  | 
| -    emms_c(); | 
| - | 
| +    avctx->execute(avctx, decode_slice, &f->slice_context[0], NULL, f->slice_count, sizeof(void*)); | 
| f->picture_number++; | 
|  | 
| *picture= *p; | 
| @@ -1163,14 +1478,7 @@ | 
|  | 
| *data_size = sizeof(AVFrame); | 
|  | 
| -    if(f->ac){ | 
| -        bytes_read= c->bytestream - c->bytestream_start - 1; | 
| -        if(bytes_read ==0) av_log(f->avctx, AV_LOG_ERROR, "error at end of frame\n"); | 
| -    }else{ | 
| -        bytes_read+= (get_bits_count(&f->gb)+7)/8; | 
| -    } | 
| - | 
| -    return bytes_read; | 
| +    return buf_size; | 
| } | 
|  | 
| AVCodec ffv1_decoder = { | 
|  |