@@ -54,6 +54,7 @@ easier to use. The changes are:

 - boxblur filter

 - Ut Video decoder

 - Speex encoding via libspeex

+- 4:2:2 H.264 decoding support

 version 0.7:

@@ -92,7 +92,7 @@ void ff_h264_idct8_add4_neon(uint8_t *dst, const int *block_offset,

                              DCTELEM *block, int stride,

                              const uint8_t nnzc[6*8]);

-static void ff_h264dsp_init_neon(H264DSPContext *c, const int bit_depth)

+static void ff_h264dsp_init_neon(H264DSPContext *c, const int bit_depth, const int chroma_format_idc)

 {

     if (bit_depth == 8) {

     c->h264_v_loop_filter_luma   = ff_h264_v_loop_filter_luma_neon;

@@ -122,14 +122,15 @@ static void ff_h264dsp_init_neon(H264DSPContext *c, const int bit_depth)

     c->h264_idct_dc_add     = ff_h264_idct_dc_add_neon;

     c->h264_idct_add16      = ff_h264_idct_add16_neon;

     c->h264_idct_add16intra = ff_h264_idct_add16intra_neon;

-    c->h264_idct_add8       = ff_h264_idct_add8_neon;

+    if (chroma_format_idc == 1)

+        c->h264_idct_add8   = ff_h264_idct_add8_neon;

     c->h264_idct8_add       = ff_h264_idct8_add_neon;

     c->h264_idct8_dc_add    = ff_h264_idct8_dc_add_neon;

     c->h264_idct8_add4      = ff_h264_idct8_add4_neon;

     }

 }

-void ff_h264dsp_init_arm(H264DSPContext *c, const int bit_depth)

+void ff_h264dsp_init_arm(H264DSPContext *c, const int bit_depth, const int chroma_format_idc)

 {

-    if (HAVE_NEON) ff_h264dsp_init_neon(c, bit_depth);

+    if (HAVE_NEON) ff_h264dsp_init_neon(c, bit_depth, chroma_format_idc);

 }

@@ -42,7 +42,7 @@ void ff_pred8x8_0lt_dc_neon(uint8_t *src, int stride);

 void ff_pred8x8_l00_dc_neon(uint8_t *src, int stride);

 void ff_pred8x8_0l0_dc_neon(uint8_t *src, int stride);

-static void ff_h264_pred_init_neon(H264PredContext *h, int codec_id, const int bit_depth)

+static void ff_h264_pred_init_neon(H264PredContext *h, int codec_id, const int bit_depth, const int chroma_format_idc)

 {

     const int high_depth = bit_depth > 8;

@@ -74,7 +74,7 @@ static void ff_h264_pred_init_neon(H264PredContext *h, int codec_id, const int b

         h->pred16x16[PLANE_PRED8x8  ] = ff_pred16x16_plane_neon;

 }

-void ff_h264_pred_init_arm(H264PredContext *h, int codec_id, int bit_depth)

+void ff_h264_pred_init_arm(H264PredContext *h, int codec_id, int bit_depth, const int chroma_format_idc)

 {

-    if (HAVE_NEON)    ff_h264_pred_init_neon(h, codec_id, bit_depth);

+    if (HAVE_NEON)    ff_h264_pred_init_neon(h, codec_id, bit_depth, chroma_format_idc);

 }

@@ -63,8 +63,10 @@ void ff_h264_idct_dc_add_ ## depth ## _c(uint8_t *dst, DCTELEM *block, int strid

 void ff_h264_idct_add16_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\

 void ff_h264_idct_add16intra_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\

 void ff_h264_idct8_add4_ ## depth ## _c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\

+void ff_h264_idct_add8_422_ ## depth ## _c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\

 void ff_h264_idct_add8_ ## depth ## _c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);\

 void ff_h264_luma_dc_dequant_idct_ ## depth ## _c(DCTELEM *output, DCTELEM *input, int qmul);\

+void ff_h264_chroma422_dc_dequant_idct_ ## depth ## _c(DCTELEM *block, int qmul);\

 void ff_h264_chroma_dc_dequant_idct_ ## depth ## _c(DCTELEM *block, int qmul);

 H264_IDCT( 8)

@@ -942,7 +942,7 @@ static void clone_tables(H264Context *dst, H264Context *src, int i){

     dst->list_counts              = src->list_counts;

     dst->s.obmc_scratchpad = NULL;

-    ff_h264_pred_init(&dst->hpc, src->s.codec_id, src->sps.bit_depth_luma);

+    ff_h264_pred_init(&dst->hpc, src->s.codec_id, src->sps.bit_depth_luma, src->sps.chroma_format_idc);

 }

 /**

@@ -970,8 +970,8 @@ static av_cold void common_init(H264Context *h){

     s->height = s->avctx->height;

     s->codec_id= s->avctx->codec->id;

-    ff_h264dsp_init(&h->h264dsp, 8);

-    ff_h264_pred_init(&h->hpc, s->codec_id, 8);

+    ff_h264dsp_init(&h->h264dsp, 8, 1);

+    ff_h264_pred_init(&h->hpc, s->codec_id, 8, 1);

     h->dequant_coeff_pps= -1;

     s->unrestricted_mv=1;

@@ -1432,11 +1432,16 @@ static void decode_postinit(H264Context *h, int setup_finished){

         ff_thread_finish_setup(s->avctx);

 }

-static av_always_inline void backup_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize, int chroma444, int simple){

+static av_always_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;

     uint8_t *top_border;

     int top_idx = 1;

     const int pixel_shift = h->pixel_shift;

+    int chroma444 = CHROMA444;

+    int chroma422 = CHROMA422;

     src_y  -=   linesize;

     src_cb -= uvlinesize;

@@ -1460,6 +1465,14 @@ static av_always_inline void backup_mb_border(H264Context *h, uint8_t *src_y, ui

                             AV_COPY128(top_border+16, src_cb + 15*uvlinesize);

                             AV_COPY128(top_border+32, src_cr + 15*uvlinesize);

                         }

+                    } else if(chroma422) {

+                        if (pixel_shift) {

+                            AV_COPY128(top_border+32, src_cb + 15*uvlinesize);

+                            AV_COPY128(top_border+48, src_cr + 15*uvlinesize);

+                        } else {

+                            AV_COPY64(top_border+16, src_cb +  15*uvlinesize);

+                            AV_COPY64(top_border+24, src_cr +  15*uvlinesize);

+                        }

                     } else {

                         if (pixel_shift) {

                             AV_COPY128(top_border+32, src_cb+7*uvlinesize);

@@ -1495,6 +1508,14 @@ static av_always_inline void backup_mb_border(H264Context *h, uint8_t *src_y, ui

                 AV_COPY128(top_border+16, src_cb + 16*linesize);

                 AV_COPY128(top_border+32, src_cr + 16*linesize);

             }

+        } else if(chroma422) {

+            if (pixel_shift) {

+                AV_COPY128(top_border+32, src_cb+16*uvlinesize);

+                AV_COPY128(top_border+48, src_cr+16*uvlinesize);

+            } else {

+                AV_COPY64(top_border+16, src_cb+16*uvlinesize);

+                AV_COPY64(top_border+24, src_cr+16*uvlinesize);

+            }

         } else {

             if (pixel_shift) {

                 AV_COPY128(top_border+32, src_cb+8*uvlinesize);

@@ -1773,10 +1794,11 @@ static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple, i

     /* is_h264 should always be true if SVQ3 is disabled. */

     const int is_h264 = !CONFIG_SVQ3_DECODER || simple || s->codec_id == CODEC_ID_H264;

     void (*idct_add)(uint8_t *dst, DCTELEM *block, int stride);

+    const int block_h = 16 >> s->chroma_y_shift;

     dest_y  = s->current_picture.f.data[0] + ((mb_x << pixel_shift) + mb_y * s->linesize  ) * 16;

-    dest_cb = s->current_picture.f.data[1] + ((mb_x << pixel_shift) + mb_y * s->uvlinesize) *  8;

-    dest_cr = s->current_picture.f.data[2] + ((mb_x << pixel_shift) + mb_y * s->uvlinesize) *  8;

+    dest_cb = s->current_picture.f.data[1] + (mb_x << pixel_shift)*8 + mb_y * s->uvlinesize * block_h;

+    dest_cr = s->current_picture.f.data[2] + (mb_x << pixel_shift)*8 + mb_y * s->uvlinesize * block_h;

     s->dsp.prefetch(dest_y + (s->mb_x&3)*4*s->linesize + (64 << pixel_shift), s->linesize, 4);

     s->dsp.prefetch(dest_cb + (s->mb_x&7)*s->uvlinesize + (64 << pixel_shift), dest_cr - dest_cb, 2);

@@ -1789,8 +1811,8 @@ static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple, i

         block_offset = &h->block_offset[48];

         if(mb_y&1){ //FIXME move out of this function?

             dest_y -= s->linesize*15;

-            dest_cb-= s->uvlinesize*7;

-            dest_cr-= s->uvlinesize*7;

+            dest_cb-= s->uvlinesize * (block_h - 1);

+            dest_cr-= s->uvlinesize * (block_h - 1);

         }

         if(FRAME_MBAFF) {

             int list;

@@ -1842,12 +1864,12 @@ static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple, i

                         }

                     }

                 } else {

-                    for (i = 0; i < 8; i++) {

+                    for (i = 0; i < block_h; i++) {

                         uint16_t *tmp_cb = (uint16_t*)(dest_cb + i*uvlinesize);

                         for (j = 0; j < 8; j++)

                             tmp_cb[j] = get_bits(&gb, bit_depth);

                     }

-                    for (i = 0; i < 8; i++) {

+                    for (i = 0; i < block_h; i++) {

                         uint16_t *tmp_cr = (uint16_t*)(dest_cr + i*uvlinesize);

                         for (j = 0; j < 8; j++)

                             tmp_cr[j] = get_bits(&gb, bit_depth);

@@ -1865,7 +1887,7 @@ static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple, i

                         memset(dest_cr + i*uvlinesize, 128, 8);

                     }

                 } else {

-                    for (i = 0; i < 8; i++) {

+                    for (i = 0; i < block_h; i++) {

                         memcpy(dest_cb + i*uvlinesize, h->mb + 128 + i*4,  8);

                         memcpy(dest_cr + i*uvlinesize, h->mb + 160 + i*4,  8);

                     }

@@ -1913,10 +1935,18 @@ static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple, i

                 }

             }else{

                 if(is_h264){

+                    int qp[2];

+                    if (CHROMA422) {

+                        qp[0] = h->chroma_qp[0] + 3;

+                        qp[1] = h->chroma_qp[1] + 3;

+                    } else {

+                        qp[0] = h->chroma_qp[0];

+                        qp[1] = h->chroma_qp[1];

+                    }

                     if(h->non_zero_count_cache[ scan8[CHROMA_DC_BLOCK_INDEX+0] ])

-                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16*16*1 << pixel_shift), h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);

+                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16*16*1 << pixel_shift), h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][qp[0]][0]);

                     if(h->non_zero_count_cache[ scan8[CHROMA_DC_BLOCK_INDEX+1] ])

-                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16*16*2 << pixel_shift), h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);

+                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16*16*2 << pixel_shift), h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][qp[1]][0]);

                     h->h264dsp.h264_idct_add8(dest, block_offset,

                                               h->mb, uvlinesize,

                                               h->non_zero_count_cache);

@@ -2555,11 +2585,13 @@ static int decode_slice_header(H264Context *h, H264Context *h0){

     h->b_stride=  s->mb_width*4;

+    s->chroma_y_shift = h->sps.chroma_format_idc <= 1; // 400 uses yuv420p

+

     s->width = 16*s->mb_width - (2>>CHROMA444)*FFMIN(h->sps.crop_right, (8<<CHROMA444)-1);

     if(h->sps.frame_mbs_only_flag)

-        s->height= 16*s->mb_height - (2>>CHROMA444)*FFMIN(h->sps.crop_bottom, (8<<CHROMA444)-1);

+        s->height= 16*s->mb_height - (1<<s->chroma_y_shift)*FFMIN(h->sps.crop_bottom, (16>>s->chroma_y_shift)-1);

     else

-        s->height= 16*s->mb_height - (4>>CHROMA444)*FFMIN(h->sps.crop_bottom, (8<<CHROMA444)-1);

+        s->height= 16*s->mb_height - (2<<s->chroma_y_shift)*FFMIN(h->sps.crop_bottom, (16>>s->chroma_y_shift)-1);

     if (s->context_initialized

         && (   s->width != s->avctx->width || s->height != s->avctx->height

@@ -2601,14 +2633,26 @@ static int decode_slice_header(H264Context *h, H264Context *h0){

         switch (h->sps.bit_depth_luma) {

             case 9 :

-                s->avctx->pix_fmt = CHROMA444 ? PIX_FMT_YUV444P9 : PIX_FMT_YUV420P9;

+                if (CHROMA444)

+                    s->avctx->pix_fmt = PIX_FMT_YUV444P9;

+                else if (CHROMA422)

+                    s->avctx->pix_fmt = PIX_FMT_YUV422P9;

+                else

+                    s->avctx->pix_fmt = PIX_FMT_YUV420P9;

                 break;

             case 10 :

-                s->avctx->pix_fmt = CHROMA444 ? PIX_FMT_YUV444P10 : PIX_FMT_YUV420P10;

+                if (CHROMA444)

+                    s->avctx->pix_fmt = PIX_FMT_YUV444P10;

+                else if (CHROMA422)

+                    s->avctx->pix_fmt = PIX_FMT_YUV422P10;

+                else

+                    s->avctx->pix_fmt = PIX_FMT_YUV420P10;

                 break;

             default:

                 if (CHROMA444){

                     s->avctx->pix_fmt = s->avctx->color_range == AVCOL_RANGE_JPEG ? PIX_FMT_YUVJ444P : PIX_FMT_YUV444P;

+                } else if (CHROMA422) {

+                    s->avctx->pix_fmt = s->avctx->color_range == AVCOL_RANGE_JPEG ? PIX_FMT_YUVJ422P : PIX_FMT_YUV422P;

                 }else{

                     s->avctx->pix_fmt = s->avctx->get_format(s->avctx,

                                                              s->avctx->codec->pix_fmts ?

@@ -3272,6 +3316,7 @@ static void loop_filter(H264Context *h, int start_x, int end_x){

     const int end_mb_y= s->mb_y + FRAME_MBAFF;

     const int old_slice_type= h->slice_type;

     const int pixel_shift = h->pixel_shift;

+    const int block_h = 16 >> s->chroma_y_shift;

     if(h->deblocking_filter) {

         for(mb_x= start_x; mb_x<end_x; mb_x++){

@@ -3288,8 +3333,8 @@ static void loop_filter(H264Context *h, int start_x, int end_x){

                 s->mb_x= mb_x;

                 s->mb_y= mb_y;

                 dest_y  = s->current_picture.f.data[0] + ((mb_x << pixel_shift) + mb_y * s->linesize  ) * 16;

-                dest_cb = s->current_picture.f.data[1] + ((mb_x << pixel_shift) + mb_y * s->uvlinesize) * (8 << CHROMA444);

-                dest_cr = s->current_picture.f.data[2] + ((mb_x << pixel_shift) + mb_y * s->uvlinesize) * (8 << CHROMA444);

+                dest_cb = s->current_picture.f.data[1] + (mb_x << pixel_shift) * (8 << CHROMA444) + mb_y * s->uvlinesize * block_h;

+                dest_cr = s->current_picture.f.data[2] + (mb_x << pixel_shift) * (8 << CHROMA444) + mb_y * s->uvlinesize * block_h;

                     //FIXME simplify above

                 if (MB_FIELD) {

@@ -3297,14 +3342,14 @@ static void loop_filter(H264Context *h, int start_x, int end_x){

                     uvlinesize = h->mb_uvlinesize = s->uvlinesize * 2;

                     if(mb_y&1){ //FIXME move out of this function?

                         dest_y -= s->linesize*15;

-                        dest_cb-= s->uvlinesize*((8 << CHROMA444)-1);

-                        dest_cr-= s->uvlinesize*((8 << CHROMA444)-1);

+                        dest_cb-= s->uvlinesize * (block_h - 1);

+                        dest_cr-= s->uvlinesize * (block_h - 1);

                     }

                 } else {

                     linesize   = h->mb_linesize   = s->linesize;

                     uvlinesize = h->mb_uvlinesize = s->uvlinesize;

                 }

-                backup_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, CHROMA444, 0);

+                backup_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 0);

                 if(fill_filter_caches(h, mb_type))

                     continue;

                 h->chroma_qp[0] = get_chroma_qp(h, 0, s->current_picture.f.qscale_table[mb_xy]);

@@ -3742,13 +3787,15 @@ static int decode_nal_units(H264Context *h, const uint8_t *buf, int buf_size){

             if(avctx->has_b_frames < 2)

                 avctx->has_b_frames= !s->low_delay;

-            if (avctx->bits_per_raw_sample != h->sps.bit_depth_luma) {

+            if (avctx->bits_per_raw_sample != h->sps.bit_depth_luma ||

+                h->cur_chroma_format_idc != h->sps.chroma_format_idc) {

                 if (h->sps.bit_depth_luma >= 8 && h->sps.bit_depth_luma <= 10) {

                     avctx->bits_per_raw_sample = h->sps.bit_depth_luma;

+                    h->cur_chroma_format_idc = h->sps.chroma_format_idc;

                     h->pixel_shift = h->sps.bit_depth_luma > 8;

-                    ff_h264dsp_init(&h->h264dsp, h->sps.bit_depth_luma);

-                    ff_h264_pred_init(&h->hpc, s->codec_id, h->sps.bit_depth_luma);

+                    ff_h264dsp_init(&h->h264dsp, h->sps.bit_depth_luma, h->sps.chroma_format_idc);

+                    ff_h264_pred_init(&h->hpc, s->codec_id, h->sps.bit_depth_luma, h->sps.chroma_format_idc);

                     s->dsp.dct_bits = h->sps.bit_depth_luma > 8 ? 32 : 16;

                     dsputil_init(&s->dsp, s->avctx);

                 } else {

@@ -39,13 +39,6 @@

 #define interlaced_dct interlaced_dct_is_a_bad_name

 #define mb_intra mb_intra_is_not_initialized_see_mb_type

-#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

@@ -92,6 +85,7 @@

 #define CABAC h->pps.cabac

 #endif

+#define CHROMA422 (h->sps.chroma_format_idc == 2)

 #define CHROMA444 (h->sps.chroma_format_idc == 3)

 #define EXTENDED_SAR          255

@@ -582,6 +576,8 @@ typedef struct H264Context{

     // Timestamp stuff

     int sei_buffering_period_present;  ///< Buffering period SEI flag

     int initial_cpb_removal_delay[32]; ///< Initial timestamps for CPBs

+

+    int cur_chroma_format_idc;

 }H264Context;

@@ -809,7 +805,7 @@ static av_always_inline void write_back_non_zero_count(H264Context *h){

     AV_COPY32(&nnz[32], &nnz_cache[4+8*11]);

     AV_COPY32(&nnz[36], &nnz_cache[4+8*12]);

-    if(CHROMA444){

+    if(!h->s.chroma_y_shift){

         AV_COPY32(&nnz[24], &nnz_cache[4+8* 8]);

         AV_COPY32(&nnz[28], &nnz_cache[4+8* 9]);

         AV_COPY32(&nnz[40], &nnz_cache[4+8*13]);

@@ -1565,7 +1565,12 @@ DECLARE_ASM_CONST(1, uint8_t, last_coeff_flag_offset_8x8)[63] = {

     5, 5, 5, 5, 6, 6, 6, 6, 7, 7, 7, 7, 8, 8, 8

 };

-static av_always_inline void decode_cabac_residual_internal( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff, int is_dc ) {

+static av_always_inline void

+decode_cabac_residual_internal(H264Context *h, DCTELEM *block,

+                               int cat, int n, const uint8_t *scantable,

+                               const uint32_t *qmul, int max_coeff,

+                               int is_dc, int chroma422)

+{

     static const int significant_coeff_flag_offset[2][14] = {

       { 105+0, 105+15, 105+29, 105+44, 105+47, 402, 484+0, 484+15, 484+29, 660, 528+0, 528+15, 528+29, 718 },

       { 277+0, 277+15, 277+29, 277+44, 277+47, 436, 776+0, 776+15, 776+29, 675, 820+0, 820+15, 820+29, 733 }

@@ -1587,12 +1592,16 @@ static av_always_inline void decode_cabac_residual_internal( H264Context *h, DCT

         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 uint8_t sig_coeff_offset_dc[7] = { 0, 0, 1, 1, 2, 2, 2 };

     /* node ctx: 0..3: abslevel1 (with abslevelgt1 == 0).

      * 4..7: abslevelgt1 + 3 (and abslevel1 doesn't matter).

      * map node ctx => cabac ctx for level=1 */

     static const uint8_t coeff_abs_level1_ctx[8] = { 1, 2, 3, 4, 0, 0, 0, 0 };

     /* map node ctx => cabac ctx for level>1 */

-    static const uint8_t coeff_abs_levelgt1_ctx[8] = { 5, 5, 5, 5, 6, 7, 8, 9 };

+    static const uint8_t coeff_abs_levelgt1_ctx[2][8] = {

+        { 5, 5, 5, 5, 6, 7, 8, 9 },

+        { 5, 5, 5, 5, 6, 7, 8, 8 }, // 422/dc case

+    };

     static const uint8_t coeff_abs_level_transition[2][8] = {

     /* update node ctx after decoding a level=1 */

         { 1, 2, 3, 3, 4, 5, 6, 7 },

@@ -1651,12 +1660,20 @@ static av_always_inline void decode_cabac_residual_internal( H264Context *h, DCT

         coeff_count= decode_significance_8x8_x86(CC, significant_coeff_ctx_base, index,

                                                  last_coeff_ctx_base, sig_off);

     } else {

-        coeff_count= decode_significance_x86(CC, max_coeff, significant_coeff_ctx_base, index,

-                                             last_coeff_ctx_base-significant_coeff_ctx_base);

+        if (is_dc && chroma422) { // dc 422

+            DECODE_SIGNIFICANCE(7, sig_coeff_offset_dc[last], sig_coeff_offset_dc[last]);

+        } else {

+            coeff_count= decode_significance_x86(CC, max_coeff, significant_coeff_ctx_base, index,

+                                                 last_coeff_ctx_base-significant_coeff_ctx_base);

+        }

 #else

         DECODE_SIGNIFICANCE( 63, sig_off[last], last_coeff_flag_offset_8x8[last] );

     } else {

-        DECODE_SIGNIFICANCE( max_coeff - 1, last, last );

+        if (is_dc && chroma422) { // dc 422

+            DECODE_SIGNIFICANCE(7, sig_coeff_offset_dc[last], sig_coeff_offset_dc[last]);

+        } else {

+            DECODE_SIGNIFICANCE(max_coeff - 1, last, last);

+        }

 #endif

     }

     assert(coeff_count > 0);

@@ -1691,7 +1708,7 @@ static av_always_inline void decode_cabac_residual_internal( H264Context *h, DCT

             } \

         } else { \

             int coeff_abs = 2; \

-            ctx = coeff_abs_levelgt1_ctx[node_ctx] + abs_level_m1_ctx_base; \

+            ctx = coeff_abs_levelgt1_ctx[is_dc && chroma422][node_ctx] + abs_level_m1_ctx_base; \

             node_ctx = coeff_abs_level_transition[1][node_ctx]; \

 \

             while( coeff_abs < 15 && get_cabac( CC, ctx ) ) { \

@@ -1733,11 +1750,18 @@ static av_always_inline void decode_cabac_residual_internal( H264Context *h, DCT

 }

 static void decode_cabac_residual_dc_internal( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, int max_coeff ) {

-    decode_cabac_residual_internal(h, block, cat, n, scantable, NULL, max_coeff, 1);

+    decode_cabac_residual_internal(h, block, cat, n, scantable, NULL, max_coeff, 1, 0);

+}

+

+static void decode_cabac_residual_dc_internal_422(H264Context *h, DCTELEM *block,

+                                                  int cat, int n, const uint8_t *scantable,

+                                                  int max_coeff)

+{

+    decode_cabac_residual_internal(h, block, cat, n, scantable, NULL, max_coeff, 1, 1);

 }

 static void decode_cabac_residual_nondc_internal( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff ) {

-    decode_cabac_residual_internal(h, block, cat, n, scantable, qmul, max_coeff, 0);

+    decode_cabac_residual_internal(h, block, cat, n, scantable, qmul, max_coeff, 0, 0);

 }

 /* cat: 0-> DC 16x16  n = 0

@@ -1761,6 +1785,19 @@ static av_always_inline void decode_cabac_residual_dc( H264Context *h, DCTELEM *

     decode_cabac_residual_dc_internal( h, block, cat, n, scantable, max_coeff );

 }

+static av_always_inline void

+decode_cabac_residual_dc_422(H264Context *h, DCTELEM *block,

+                             int cat, int n, const uint8_t *scantable,

+                             int max_coeff)

+{

+    /* read coded block flag */

+    if (get_cabac(&h->cabac, &h->cabac_state[get_cabac_cbf_ctx(h, cat, n, max_coeff, 1)]) == 0) {

+        h->non_zero_count_cache[scan8[n]] = 0;

+        return;

+    }

+    decode_cabac_residual_dc_internal_422(h, block, cat, n, scantable, max_coeff);

+}

+

 static av_always_inline void decode_cabac_residual_nondc( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff ) {

     /* read coded block flag */

     if( (cat != 5 || CHROMA444) && get_cabac( &h->cabac, &h->cabac_state[get_cabac_cbf_ctx( h, cat, n, max_coeff, 0 ) ] ) == 0 ) {

@@ -2313,7 +2350,36 @@ decode_intra_mb:

         if(CHROMA444){

             decode_cabac_luma_residual(h, scan, scan8x8, pixel_shift, mb_type, cbp, 1);

             decode_cabac_luma_residual(h, scan, scan8x8, pixel_shift, mb_type, cbp, 2);

-        } else {

+        } else if (CHROMA422) {

+            if( cbp&0x30 ){

+                int c;

+                for( c = 0; c < 2; c++ ) {

+                    //av_log( s->avctx, AV_LOG_ERROR, "INTRA C%d-DC\n",c );

+                    decode_cabac_residual_dc_422(h, h->mb + ((256 + 16*16*c) << pixel_shift), 3,

+                                                 CHROMA_DC_BLOCK_INDEX + c,

+                                                 chroma422_dc_scan, 8);

+                }

+            }

+

+            if( cbp&0x20 ) {

+                int c, i, i8x8;

+                for( c = 0; c < 2; c++ ) {

+                    DCTELEM *mb = h->mb + (16*(16 + 16*c) << pixel_shift);

+                    qmul = h->dequant4_coeff[c+1+(IS_INTRA( mb_type ) ? 0:3)][h->chroma_qp[c]];

+                    for (i8x8 = 0; i8x8 < 2; i8x8++) {

+                        for (i = 0; i < 4; i++) {

+                            const int index = 16 + 16 * c + 8*i8x8 + i;

+                            //av_log(s->avctx, AV_LOG_ERROR, "INTRA C%d-AC %d\n",c, index - 16);

+                            decode_cabac_residual_nondc(h, mb, 4, index, scan + 1, qmul, 15);

+                            mb += 16<<pixel_shift;

+                        }

+                    }

+                }

+            } else {

+                fill_rectangle(&h->non_zero_count_cache[scan8[16]], 4, 4, 8, 0, 1);

+                fill_rectangle(&h->non_zero_count_cache[scan8[32]], 4, 4, 8, 0, 1);

+            }

+        } else /* yuv420 */ {

             if( cbp&0x30 ){

                 int c;

                 for( c = 0; c < 2; c++ ) {

@@ -62,6 +62,30 @@ static const uint8_t chroma_dc_coeff_token_bits[4*5]={

  2, 3, 2, 0,

 };

+static const uint8_t chroma422_dc_coeff_token_len[4*9]={

+  1,  0,  0,  0,

+  7,  2,  0,  0,

+  7,  7,  3,  0,

+  9,  7,  7,  5,

+  9,  9,  7,  6,

+ 10, 10,  9,  7,

+ 11, 11, 10,  7,

+ 12, 12, 11, 10,

+ 13, 12, 12, 11,

+};

+

+static const uint8_t chroma422_dc_coeff_token_bits[4*9]={

+  1,   0,  0, 0,

+ 15,   1,  0, 0,

+ 14,  13,  1, 0,

+  7,  12, 11, 1,

+  6,   5, 10, 1,

+  7,   6,  4, 9,

+  7,   6,  5, 8,

+  7,   6,  5, 4,

+  7,   5,  4, 4,

+};

+

 static const uint8_t coeff_token_len[4][4*17]={

 {

      1, 0, 0, 0,

@@ -172,6 +196,26 @@ static const uint8_t chroma_dc_total_zeros_bits[3][4]= {

     { 1, 0, 0, 0,},

 };

+static const uint8_t chroma422_dc_total_zeros_len[7][8]= {

+    { 1, 3, 3, 4, 4, 4, 5, 5 },

+    { 3, 2, 3, 3, 3, 3, 3 },

+    { 3, 3, 2, 2, 3, 3 },

+    { 3, 2, 2, 2, 3 },

+    { 2, 2, 2, 2 },

+    { 2, 2, 1 },

+    { 1, 1 },

+};

+

+static const uint8_t chroma422_dc_total_zeros_bits[7][8]= {

+    { 1, 2, 3, 2, 3, 1, 1, 0 },

+    { 0, 1, 1, 4, 5, 6, 7 },

+    { 0, 1, 1, 2, 6, 7 },

+    { 6, 0, 1, 2, 7 },

+    { 0, 1, 2, 3 },

+    { 0, 1, 1 },

+    { 0, 1 },

+};

+

 static const uint8_t run_len[7][16]={

     {1,1},

     {1,2,2},

@@ -200,6 +244,10 @@ static VLC chroma_dc_coeff_token_vlc;

 static VLC_TYPE chroma_dc_coeff_token_vlc_table[256][2];

 static const int chroma_dc_coeff_token_vlc_table_size = 256;

+static VLC chroma422_dc_coeff_token_vlc;

+static VLC_TYPE chroma422_dc_coeff_token_vlc_table[8192][2];

+static const int chroma422_dc_coeff_token_vlc_table_size = 8192;

+

 static VLC total_zeros_vlc[15];

 static VLC_TYPE total_zeros_vlc_tables[15][512][2];

 static const int total_zeros_vlc_tables_size = 512;

@@ -208,6 +256,10 @@ static VLC chroma_dc_total_zeros_vlc[3];

 static VLC_TYPE chroma_dc_total_zeros_vlc_tables[3][8][2];

 static const int chroma_dc_total_zeros_vlc_tables_size = 8;

+static VLC chroma422_dc_total_zeros_vlc[7];

+static VLC_TYPE chroma422_dc_total_zeros_vlc_tables[7][32][2];

+static const int chroma422_dc_total_zeros_vlc_tables_size = 32;

+

 static VLC run_vlc[6];

 static VLC_TYPE run_vlc_tables[6][8][2];

 static const int run_vlc_tables_size = 8;

@@ -219,6 +271,14 @@ static const int run7_vlc_table_size = 96;

 #define LEVEL_TAB_BITS 8

 static int8_t cavlc_level_tab[7][1<<LEVEL_TAB_BITS][2];

+#define CHROMA_DC_COEFF_TOKEN_VLC_BITS 8

+#define CHROMA422_DC_COEFF_TOKEN_VLC_BITS 13

+#define COEFF_TOKEN_VLC_BITS           8

+#define TOTAL_ZEROS_VLC_BITS           9

+#define CHROMA_DC_TOTAL_ZEROS_VLC_BITS 3

+#define CHROMA422_DC_TOTAL_ZEROS_VLC_BITS 5

+#define RUN_VLC_BITS                   3

+#define RUN7_VLC_BITS                  6

 /**

  * gets the predicted number of non-zero coefficients.

@@ -278,6 +338,13 @@ av_cold void ff_h264_decode_init_vlc(void){

                  &chroma_dc_coeff_token_bits[0], 1, 1,

                  INIT_VLC_USE_NEW_STATIC);

+        chroma422_dc_coeff_token_vlc.table = chroma422_dc_coeff_token_vlc_table;

+        chroma422_dc_coeff_token_vlc.table_allocated = chroma422_dc_coeff_token_vlc_table_size;

+        init_vlc(&chroma422_dc_coeff_token_vlc, CHROMA422_DC_COEFF_TOKEN_VLC_BITS, 4*9,

+                 &chroma422_dc_coeff_token_len [0], 1, 1,

+                 &chroma422_dc_coeff_token_bits[0], 1, 1,

+                 INIT_VLC_USE_NEW_STATIC);

+

         offset = 0;

         for(i=0; i<4; i++){

             coeff_token_vlc[i].table = coeff_token_vlc_tables+offset;

@@ -304,6 +371,17 @@ av_cold void ff_h264_decode_init_vlc(void){

                      &chroma_dc_total_zeros_bits[i][0], 1, 1,

                      INIT_VLC_USE_NEW_STATIC);

         }

+

+        for(i=0; i<7; i++){

+            chroma422_dc_total_zeros_vlc[i].table = chroma422_dc_total_zeros_vlc_tables[i];

+            chroma422_dc_total_zeros_vlc[i].table_allocated = chroma422_dc_total_zeros_vlc_tables_size;

+            init_vlc(&chroma422_dc_total_zeros_vlc[i],

+                     CHROMA422_DC_TOTAL_ZEROS_VLC_BITS, 8,

+                     &chroma422_dc_total_zeros_len [i][0], 1, 1,

+                     &chroma422_dc_total_zeros_bits[i][0], 1, 1,

+                     INIT_VLC_USE_NEW_STATIC);

+        }

+

         for(i=0; i<15; i++){

             total_zeros_vlc[i].table = total_zeros_vlc_tables[i];

             total_zeros_vlc[i].table_allocated = total_zeros_vlc_tables_size;

@@ -373,7 +451,10 @@ static int decode_residual(H264Context *h, GetBitContext *gb, DCTELEM *block, in

     //FIXME put trailing_onex into the context

     if(max_coeff <= 8){

-        coeff_token= get_vlc2(gb, chroma_dc_coeff_token_vlc.table, CHROMA_DC_COEFF_TOKEN_VLC_BITS, 1);

+        if (max_coeff == 4)

+            coeff_token = get_vlc2(gb, chroma_dc_coeff_token_vlc.table, CHROMA_DC_COEFF_TOKEN_VLC_BITS, 1);

+        else

+            coeff_token = get_vlc2(gb, chroma422_dc_coeff_token_vlc.table, CHROMA422_DC_COEFF_TOKEN_VLC_BITS, 1);

         total_coeff= coeff_token>>2;

     }else{

         if(n >= LUMA_DC_BLOCK_INDEX){

@@ -483,11 +564,16 @@ static int decode_residual(H264Context *h, GetBitContext *gb, DCTELEM *block, in

     if(total_coeff == max_coeff)

         zeros_left=0;

     else{

-        /* FIXME: we don't actually support 4:2:2 yet. */

-        if(max_coeff <= 8)

-            zeros_left= get_vlc2(gb, (chroma_dc_total_zeros_vlc-1)[ total_coeff ].table, CHROMA_DC_TOTAL_ZEROS_VLC_BITS, 1);

-        else

+        if (max_coeff <= 8) {

+            if (max_coeff == 4)

+                zeros_left = get_vlc2(gb, (chroma_dc_total_zeros_vlc-1)[total_coeff].table,

+                                      CHROMA_DC_TOTAL_ZEROS_VLC_BITS, 1);

+            else

+                zeros_left = get_vlc2(gb, (chroma422_dc_total_zeros_vlc-1)[total_coeff].table,

+                                      CHROMA422_DC_TOTAL_ZEROS_VLC_BITS, 1);

+        } else {

             zeros_left= get_vlc2(gb, (total_zeros_vlc-1)[ total_coeff ].table, TOTAL_ZEROS_VLC_BITS, 1);

+        }

     }

 #define STORE_BLOCK(type) \

@@ -994,7 +1080,7 @@ decode_intra_mb:

     s->current_picture.f.mb_type[mb_xy] = mb_type;

     if(cbp || IS_INTRA16x16(mb_type)){

-        int i4x4, chroma_idx;

+        int i4x4, i8x8, chroma_idx;

         int dquant;

         int ret;

         GetBitContext *gb= IS_INTRA(mb_type) ? h->intra_gb_ptr : h->inter_gb_ptr;

@@ -1036,7 +1122,34 @@ decode_intra_mb:

             if( decode_luma_residual(h, gb, scan, scan8x8, pixel_shift, mb_type, cbp, 2) < 0 ){

                 return -1;

             }

-        } else {

+        } else if (CHROMA422) {

+            if(cbp&0x30){

+                for(chroma_idx=0; chroma_idx<2; chroma_idx++)

+                    if (decode_residual(h, gb, h->mb + ((256 + 16*16*chroma_idx) << pixel_shift),

+                                        CHROMA_DC_BLOCK_INDEX+chroma_idx, chroma422_dc_scan,

+                                        NULL, 8) < 0) {

+                        return -1;

+                    }

+            }

+

+            if(cbp&0x20){

+                for(chroma_idx=0; chroma_idx<2; chroma_idx++){

+                    const uint32_t *qmul = h->dequant4_coeff[chroma_idx+1+(IS_INTRA( mb_type ) ? 0:3)][h->chroma_qp[chroma_idx]];

+                    DCTELEM *mb = h->mb + (16*(16 + 16*chroma_idx) << pixel_shift);

+                    for (i8x8 = 0; i8x8 < 2; i8x8++) {

+                        for (i4x4 = 0; i4x4 < 4; i4x4++) {

+                            const int index = 16 + 16*chroma_idx + 8*i8x8 + i4x4;

+                            if (decode_residual(h, gb, mb, index, scan + 1, qmul, 15) < 0)

+                                return -1;

+                            mb += 16 << pixel_shift;

+                        }

+                    }

+                }

+            }else{

+                fill_rectangle(&h->non_zero_count_cache[scan8[16]], 4, 4, 8, 0, 1);

+                fill_rectangle(&h->non_zero_count_cache[scan8[32]], 4, 4, 8, 0, 1);

+            }

+        } else /* yuv420 */ {

             if(cbp&0x30){

                 for(chroma_idx=0; chroma_idx<2; chroma_idx++)

                     if( decode_residual(h, gb, h->mb + ((256 + 16*16*chroma_idx) << pixel_shift), CHROMA_DC_BLOCK_INDEX+chroma_idx, chroma_dc_scan, NULL, 4) < 0){

@@ -212,6 +212,7 @@ static void av_always_inline h264_filter_mb_fast_internal( H264Context *h, int m

     MpegEncContext * const s = &h->s;

     int chroma = !(CONFIG_GRAY && (s->flags&CODEC_FLAG_GRAY));

     int chroma444 = CHROMA444;

+    int chroma422 = CHROMA422;

     int mb_xy = h->mb_xy;

     int left_type= h->left_type[LTOP];

@@ -289,6 +290,23 @@ static void av_always_inline h264_filter_mb_fast_internal( H264Context *h, int m

                     filter_mb_edgeh( &img_cb[4*3*linesize], linesize, bS3, qpc, a, b, h, 0);

                     filter_mb_edgeh( &img_cr[4*3*linesize], linesize, bS3, qpc, a, b, h, 0);

                 }

+            }else if(chroma422){

+                if(left_type){

+                    filter_mb_edgecv(&img_cb[2*0<<pixel_shift], uvlinesize, bS4, qpc0, a, b, h, 1);

+                    filter_mb_edgecv(&img_cr[2*0<<pixel_shift], uvlinesize, bS4, qpc0, a, b, h, 1);

+                }

+                filter_mb_edgecv(&img_cb[2*2<<pixel_shift], uvlinesize, bS3, qpc, a, b, h, 0);

+                filter_mb_edgecv(&img_cr[2*2<<pixel_shift], uvlinesize, bS3, qpc, a, b, h, 0);

+                if(top_type){

+                    filter_mb_edgech(&img_cb[4*0*uvlinesize], uvlinesize, bSH, qpc1, a, b, h, 1);

+                    filter_mb_edgech(&img_cr[4*0*uvlinesize], uvlinesize, bSH, qpc1, a, b, h, 1);

+                }

+                filter_mb_edgech(&img_cb[4*1*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);

+                filter_mb_edgech(&img_cr[4*1*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);

+                filter_mb_edgech(&img_cb[4*2*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);

+                filter_mb_edgech(&img_cr[4*2*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);

+                filter_mb_edgech(&img_cb[4*3*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);

+                filter_mb_edgech(&img_cr[4*3*uvlinesize], uvlinesize, bS3, qpc, a, b, h, 0);

             }else{

                 if(left_type){

                     filter_mb_edgecv( &img_cb[2*0<<pixel_shift], uvlinesize, bS4, qpc0, a, b, h, 1);

@@ -411,10 +429,12 @@ static int check_mv(H264Context *h, long b_idx, long bn_idx, int mvy_limit){

     return v;

 }

-static av_always_inline void filter_mb_dir(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, int mb_xy, int mb_type, int mvy_limit, int first_vertical_edge_done, int a, int b, int chroma, int chroma444, int dir) {

+static av_always_inline void filter_mb_dir(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, int mb_xy, int mb_type, int mvy_limit, int first_vertical_edge_done, int a, int b, int chroma, int dir) {

     MpegEncContext * const s = &h->s;

     int edge;

     int chroma_qp_avg[2];

+    int chroma444 = CHROMA444;

+    int chroma422 = CHROMA422;

     const int mbm_xy = dir == 0 ? mb_xy -1 : h->top_mb_xy;

     const int mbm_type = dir == 0 ? h->left_type[LTOP] : h->top_type;

@@ -564,8 +584,9 @@ static av_always_inline void filter_mb_dir(H264Context *h, int mb_x, int mb_y, u

     for( edge = 1; edge < edges; edge++ ) {

         DECLARE_ALIGNED(8, int16_t, bS)[4];

         int qp;

+        const int deblock_edge = !IS_8x8DCT(mb_type & (edge<<24)); // (edge&1) && IS_8x8DCT(mb_type)

-        if( IS_8x8DCT(mb_type & (edge<<24)) ) // (edge&1) && IS_8x8DCT(mb_type)

+        if (!deblock_edge && (!chroma422 || dir == 0))

             continue;

         if( IS_INTRA(mb_type)) {

@@ -627,14 +648,23 @@ static av_always_inline void filter_mb_dir(H264Context *h, int mb_x, int mb_y, u

                 }

             }

         } else {

-            filter_mb_edgeh( &img_y[4*edge*linesize], linesize, bS, qp, a, b, h, 0 );

-            if (chroma) {

-                if (chroma444) {

-                    filter_mb_edgeh ( &img_cb[4*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[0], a, b, h, 0);

-                    filter_mb_edgeh ( &img_cr[4*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[1], a, b, h, 0);

-                } else if( (edge&1) == 0 ) {

-                    filter_mb_edgech( &img_cb[2*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[0], a, b, h, 0);

-                    filter_mb_edgech( &img_cr[2*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[1], a, b, h, 0);

+            if (chroma422) {

+                if (deblock_edge)

+                    filter_mb_edgeh(&img_y[4*edge*linesize], linesize, bS, qp, a, b, h, 0);

+                if (chroma) {

+                    filter_mb_edgech(&img_cb[4*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[0], a, b, h, 0);

+                    filter_mb_edgech(&img_cr[4*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[1], a, b, h, 0);

+                }

+            } else {

+                filter_mb_edgeh(&img_y[4*edge*linesize], linesize, bS, qp, a, b, h, 0);

+                if (chroma) {

+                    if (chroma444) {

+                        filter_mb_edgeh (&img_cb[4*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[0], a, b, h, 0);

+                        filter_mb_edgeh (&img_cr[4*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[1], a, b, h, 0);

+                    } else if ((edge&1) == 0) {

+                        filter_mb_edgech(&img_cb[2*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[0], a, b, h, 0);

+                        filter_mb_edgech(&img_cr[2*edge*uvlinesize], uvlinesize, bS, h->chroma_qp[1], a, b, h, 0);

+                    }

                 }

             }

         }

@@ -726,6 +756,11 @@ void ff_h264_filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint

                     filter_mb_mbaff_edgev ( h, img_cb + 8*uvlinesize, uvlinesize, bS+4, 1, bqp[1], a, b, 1 );

                     filter_mb_mbaff_edgev ( h, img_cr,                uvlinesize, bS  , 1, rqp[0], a, b, 1 );

                     filter_mb_mbaff_edgev ( h, img_cr + 8*uvlinesize, uvlinesize, bS+4, 1, rqp[1], a, b, 1 );

+                } else if (CHROMA422) {

+                    filter_mb_mbaff_edgecv(h, img_cb,                uvlinesize, bS  , 1, bqp[0], a, b, 1);

+                    filter_mb_mbaff_edgecv(h, img_cb + 8*uvlinesize, uvlinesize, bS+4, 1, bqp[1], a, b, 1);

+                    filter_mb_mbaff_edgecv(h, img_cr,                uvlinesize, bS  , 1, rqp[0], a, b, 1);

+                    filter_mb_mbaff_edgecv(h, img_cr + 8*uvlinesize, uvlinesize, bS+4, 1, rqp[1], a, b, 1);

                 }else{

                     filter_mb_mbaff_edgecv( h, img_cb,                uvlinesize, bS  , 1, bqp[0], a, b, 1 );

                     filter_mb_mbaff_edgecv( h, img_cb + 4*uvlinesize, uvlinesize, bS+4, 1, bqp[1], a, b, 1 );

@@ -754,9 +789,9 @@ void ff_h264_filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint

 #if CONFIG_SMALL

     for( dir = 0; dir < 2; dir++ )

-        filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, dir ? 0 : first_vertical_edge_done, a, b, chroma, CHROMA444, dir);

+        filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, dir ? 0 : first_vertical_edge_done, a, b, chroma, dir);

 #else

-    filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, first_vertical_edge_done, a, b, chroma, CHROMA444, 0);

-    filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, 0,                        a, b, chroma, CHROMA444, 1);

+    filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, first_vertical_edge_done, a, b, chroma, 0);

+    filter_mb_dir(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize, mb_xy, mb_type, mvy_limit, 0,                        a, b, chroma, 1);

 #endif

 }

@@ -510,7 +510,7 @@ static void fill_decode_caches(H264Context *h, int mb_type){

     if(top_type){

         nnz = h->non_zero_count[top_xy];

         AV_COPY32(&nnz_cache[4+8* 0], &nnz[4*3]);

-        if(CHROMA444){

+        if(!s->chroma_y_shift){

             AV_COPY32(&nnz_cache[4+8* 5], &nnz[4* 7]);

             AV_COPY32(&nnz_cache[4+8*10], &nnz[4*11]);

         }else{

@@ -534,6 +534,11 @@ static void fill_decode_caches(H264Context *h, int mb_type){

                 nnz_cache[3+8* 7 + 2*8*i]= nnz[left_block[8+1+2*i]+4*4];

                 nnz_cache[3+8*11 + 2*8*i]= nnz[left_block[8+0+2*i]+8*4];

                 nnz_cache[3+8*12 + 2*8*i]= nnz[left_block[8+1+2*i]+8*4];

+            }else if(CHROMA422) {

+                nnz_cache[3+8* 6 + 2*8*i]= nnz[left_block[8+0+2*i]-2+4*4];

+                nnz_cache[3+8* 7 + 2*8*i]= nnz[left_block[8+1+2*i]-2+4*4];

+                nnz_cache[3+8*11 + 2*8*i]= nnz[left_block[8+0+2*i]-2+8*4];

+                nnz_cache[3+8*12 + 2*8*i]= nnz[left_block[8+1+2*i]-2+8*4];

             }else{

                 nnz_cache[3+8* 6 +   8*i]= nnz[left_block[8+4+2*i]];

                 nnz_cache[3+8*11 +   8*i]= nnz[left_block[8+5+2*i]];

@@ -396,7 +396,8 @@ int ff_h264_decode_seq_parameter_set(H264Context *h){

 #endif

     sps->crop= get_bits1(&s->gb);

     if(sps->crop){

-        int crop_limit = sps->chroma_format_idc == 3 ? 16 : 8;

+        int crop_vertical_limit   = sps->chroma_format_idc  & 2 ? 16 : 8;

+        int crop_horizontal_limit = sps->chroma_format_idc == 3 ? 16 : 8;

         sps->crop_left  = get_ue_golomb(&s->gb);

         sps->crop_right = get_ue_golomb(&s->gb);

         sps->crop_top   = get_ue_golomb(&s->gb);

@@ -404,7 +405,7 @@ int ff_h264_decode_seq_parameter_set(H264Context *h){

         if(sps->crop_left || sps->crop_top){

             av_log(h->s.avctx, AV_LOG_ERROR, "insane cropping not completely supported, this could look slightly wrong ...\n");

         }

-        if(sps->crop_right >= crop_limit || sps->crop_bottom >= crop_limit){

+        if(sps->crop_right >= crop_horizontal_limit || sps->crop_bottom >= crop_vertical_limit){

             av_log(h->s.avctx, AV_LOG_ERROR, "brainfart cropping not supported, this could look slightly wrong ...\n");

         }

     }else{

@@ -80,7 +80,14 @@ static const uint8_t luma_dc_field_scan[16]={

 static const uint8_t chroma_dc_scan[4]={

  (0+0*2)*16, (1+0*2)*16,

- (0+1*2)*16, (1+1*2)*16,  //FIXME

+ (0+1*2)*16, (1+1*2)*16,

+};

+

+static const uint8_t chroma422_dc_scan[8]={

+ (0+0*2)*16, (0+1*2)*16,

+ (1+0*2)*16, (0+2*2)*16,

+ (0+3*2)*16, (1+1*2)*16,

+ (1+2*2)*16, (1+3*2)*16,

 };

 // zigzag_scan8x8_cavlc[i] = zigzag_scan8x8[(i/4) + 16*(i%4)]

@@ -41,7 +41,7 @@

 #include "h264dsp_template.c"

 #undef BIT_DEPTH

-void ff_h264dsp_init(H264DSPContext *c, const int bit_depth)

+void ff_h264dsp_init(H264DSPContext *c, const int bit_depth, const int chroma_format_idc)

 {

 #undef FUNC

 #define FUNC(a, depth) a ## _ ## depth ## _c

@@ -53,10 +53,16 @@ void ff_h264dsp_init(H264DSPContext *c, const int bit_depth)

     c->h264_idct8_dc_add= FUNC(ff_h264_idct8_dc_add, depth);\

     c->h264_idct_add16     = FUNC(ff_h264_idct_add16, depth);\

     c->h264_idct8_add4     = FUNC(ff_h264_idct8_add4, depth);\

-    c->h264_idct_add8      = FUNC(ff_h264_idct_add8, depth);\

+    if (chroma_format_idc == 1)\

+        c->h264_idct_add8  = FUNC(ff_h264_idct_add8, depth);\