h264: factor out calculating the POC count into a separate file

This will allow decoupling the parser from the decoder.

11 files changed, 162 insertions, 152 deletions

diff --git a/libavcodec/dxva2_h264.c b/libavcodec/dxva2_h264.c
index cd13486b7d..acaacb9401 100644
--- a/libavcodec/dxva2_h264.c
+++ b/libavcodec/dxva2_h264.c
@@ -143,7 +143,7 @@ static void fill_picture_parameters(const AVCodecContext *avctx, AVDXVAContext *
     pp->num_ref_idx_l0_active_minus1  = pps->ref_count[0] - 1;
     pp->num_ref_idx_l1_active_minus1  = pps->ref_count[1] - 1;
     pp->Reserved8BitsA                = 0;
-    pp->frame_num                     = h->frame_num;
+    pp->frame_num                     = h->poc.frame_num;
     pp->log2_max_frame_num_minus4     = sps->log2_max_frame_num - 4;
     pp->pic_order_cnt_type            = sps->poc_type;
     if (sps->poc_type == 0)
diff --git a/libavcodec/h264.c b/libavcodec/h264.c
index 81c1e81832..e415103a0f 100644
--- a/libavcodec/h264.c
+++ b/libavcodec/h264.c
@@ -399,7 +399,7 @@ static int h264_init_context(AVCodecContext *avctx, H264Context *h)
     h->slice_context_count   = 1;
     h->workaround_bugs       = avctx->workaround_bugs;
     h->flags                 = avctx->flags;
-    h->prev_poc_msb          = 1 << 16;
+    h->poc.prev_poc_msb      = 1 << 16;
     h->x264_build            = -1;
     h->recovery_frame        = -1;
     h->frame_recovered       = 0;
@@ -813,10 +813,10 @@ static void decode_postinit(H264Context *h, int setup_finished)
 static void idr(H264Context *h)
 {
     ff_h264_remove_all_refs(h);
-    h->prev_frame_num        =
-    h->prev_frame_num_offset =
-    h->prev_poc_msb          =
-    h->prev_poc_lsb          = 0;
+    h->poc.prev_frame_num        =
+    h->poc.prev_frame_num_offset =
+    h->poc.prev_poc_msb          =
+    h->poc.prev_poc_lsb          = 0;
 }
 
 /* forget old pics after a seek */
@@ -857,85 +857,6 @@ static void flush_dpb(AVCodecContext *avctx)
     h->context_initialized = 0;
 }
 
-int ff_init_poc(H264Context *h, int pic_field_poc[2], int *pic_poc)
-{
-    const SPS *sps = h->ps.sps;
-    const int max_frame_num = 1 << sps->log2_max_frame_num;
-    int field_poc[2];
-
-    h->frame_num_offset = h->prev_frame_num_offset;
-    if (h->frame_num < h->prev_frame_num)
-        h->frame_num_offset += max_frame_num;
-
-    if (sps->poc_type == 0) {
-        const int max_poc_lsb = 1 << sps->log2_max_poc_lsb;
-
-        if (h->poc_lsb < h->prev_poc_lsb &&
-            h->prev_poc_lsb - h->poc_lsb >= max_poc_lsb / 2)
-            h->poc_msb = h->prev_poc_msb + max_poc_lsb;
-        else if (h->poc_lsb > h->prev_poc_lsb &&
-                 h->prev_poc_lsb - h->poc_lsb < -max_poc_lsb / 2)
-            h->poc_msb = h->prev_poc_msb - max_poc_lsb;
-        else
-            h->poc_msb = h->prev_poc_msb;
-        field_poc[0] =
-        field_poc[1] = h->poc_msb + h->poc_lsb;
-        if (h->picture_structure == PICT_FRAME)
-            field_poc[1] += h->delta_poc_bottom;
-    } else if (sps->poc_type == 1) {
-        int abs_frame_num, expected_delta_per_poc_cycle, expectedpoc;
-        int i;
-
-        if (sps->poc_cycle_length != 0)
-            abs_frame_num = h->frame_num_offset + h->frame_num;
-        else
-            abs_frame_num = 0;
-
-        if (h->nal_ref_idc == 0 && abs_frame_num > 0)
-            abs_frame_num--;
-
-        expected_delta_per_poc_cycle = 0;
-        for (i = 0; i < sps->poc_cycle_length; i++)
-            // FIXME integrate during sps parse
-            expected_delta_per_poc_cycle += sps->offset_for_ref_frame[i];
-
-        if (abs_frame_num > 0) {
-            int poc_cycle_cnt          = (abs_frame_num - 1) / sps->poc_cycle_length;
-            int frame_num_in_poc_cycle = (abs_frame_num - 1) % sps->poc_cycle_length;
-
-            expectedpoc = poc_cycle_cnt * expected_delta_per_poc_cycle;
-            for (i = 0; i <= frame_num_in_poc_cycle; i++)
-                expectedpoc = expectedpoc + sps->offset_for_ref_frame[i];
-        } else
-            expectedpoc = 0;
-
-        if (h->nal_ref_idc == 0)
-            expectedpoc = expectedpoc + sps->offset_for_non_ref_pic;
-
-        field_poc[0] = expectedpoc + h->delta_poc[0];
-        field_poc[1] = field_poc[0] + sps->offset_for_top_to_bottom_field;
-
-        if (h->picture_structure == PICT_FRAME)
-            field_poc[1] += h->delta_poc[1];
-    } else {
-        int poc = 2 * (h->frame_num_offset + h->frame_num);
-
-        if (!h->nal_ref_idc)
-            poc--;
-
-        field_poc[0] = poc;
-        field_poc[1] = poc;
-    }
-
-    if (h->picture_structure != PICT_BOTTOM_FIELD)
-        pic_field_poc[0] = field_poc[0];
-    if (h->picture_structure != PICT_TOP_FIELD)
-        pic_field_poc[1] = field_poc[1];
-    *pic_poc = FFMIN(pic_field_poc[0], pic_field_poc[1]);
-
-    return 0;
-}
-
 /**
  * Compute profile from profile_idc and constraint_set?_flags.
  *
@@ -1064,14 +985,14 @@ again:
                 break;
 
             if (h->sei_recovery_frame_cnt >= 0 && h->recovery_frame < 0) {
-                h->recovery_frame = (h->frame_num + h->sei_recovery_frame_cnt) &
+                h->recovery_frame = (h->poc.frame_num + h->sei_recovery_frame_cnt) &
                                     ((1 << h->ps.sps->log2_max_frame_num) - 1);
             }
 
             h->cur_pic_ptr->f->key_frame |=
                 (nal->type == NAL_IDR_SLICE) || (h->sei_recovery_frame_cnt >= 0);
 
-            if (nal->type == NAL_IDR_SLICE || h->recovery_frame == h->frame_num) {
+            if (nal->type == NAL_IDR_SLICE || h->recovery_frame == h->poc.frame_num) {
                 h->recovery_frame         = -1;
                 h->cur_pic_ptr->recovered = 1;
             }
diff --git a/libavcodec/h264.h b/libavcodec/h264.h
index 6a026aa818..007ce58b5e 100644
--- a/libavcodec/h264.h
+++ b/libavcodec/h264.h
@@ -577,17 +577,7 @@ typedef struct H264Context {
 
     uint16_t *slice_table_base;
 
-    // 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
+    H264POCContext poc;
 
     /**
      * frame_num for frames or 2 * frame_num + 1 for field pics.
@@ -1059,7 +1049,6 @@ void ff_h264_unref_picture(H264Context *h, H264Picture *pic);
 int ff_h264_slice_context_init(H264Context *h, H264SliceContext *sl);
 
 void ff_h264_draw_horiz_band(const H264Context *h, H264SliceContext *sl, int y, int height);
-int ff_init_poc(H264Context *h, int pic_field_poc[2], int *pic_poc);
 
 int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl);
 int ff_h264_execute_decode_slices(H264Context *h, unsigned context_count);
diff --git a/libavcodec/h264_parse.c b/libavcodec/h264_parse.c
index bd1a50ec17..4ab0fde5f7 100644
--- a/libavcodec/h264_parse.c
+++ b/libavcodec/h264_parse.c
@@ -225,3 +225,83 @@ fail:
     ref_count[1] = 0;
     return AVERROR_INVALIDDATA;
 }
+
+int ff_h264_init_poc(int pic_field_poc[2], int *pic_poc,
+                     const SPS *sps, H264POCContext *pc,
+                     int picture_structure, int nal_ref_idc)
+{
+    const int max_frame_num = 1 << sps->log2_max_frame_num;
+    int field_poc[2];
+
+    pc->frame_num_offset = pc->prev_frame_num_offset;
+    if (pc->frame_num < pc->prev_frame_num)
+        pc->frame_num_offset += max_frame_num;
+
+    if (sps->poc_type == 0) {
+        const int max_poc_lsb = 1 << sps->log2_max_poc_lsb;
+
+        if (pc->poc_lsb < pc->prev_poc_lsb &&
+            pc->prev_poc_lsb - pc->poc_lsb >= max_poc_lsb / 2)
+            pc->poc_msb = pc->prev_poc_msb + max_poc_lsb;
+        else if (pc->poc_lsb > pc->prev_poc_lsb &&
+                 pc->prev_poc_lsb - pc->poc_lsb < -max_poc_lsb / 2)
+            pc->poc_msb = pc->prev_poc_msb - max_poc_lsb;
+        else
+            pc->poc_msb = pc->prev_poc_msb;
+        field_poc[0] =
+        field_poc[1] = pc->poc_msb + pc->poc_lsb;
+        if (picture_structure == PICT_FRAME)
+            field_poc[1] += pc->delta_poc_bottom;
+    } else if (sps->poc_type == 1) {
+        int abs_frame_num, expected_delta_per_poc_cycle, expectedpoc;
+        int i;
+
+        if (sps->poc_cycle_length != 0)
+            abs_frame_num = pc->frame_num_offset + pc->frame_num;
+        else
+            abs_frame_num = 0;
+
+        if (nal_ref_idc == 0 && abs_frame_num > 0)
+            abs_frame_num--;
+
+        expected_delta_per_poc_cycle = 0;
+        for (i = 0; i < sps->poc_cycle_length; i++)
+            // FIXME integrate during sps parse
+            expected_delta_per_poc_cycle += sps->offset_for_ref_frame[i];
+
+        if (abs_frame_num > 0) {
+            int poc_cycle_cnt          = (abs_frame_num - 1) / sps->poc_cycle_length;
+            int frame_num_in_poc_cycle = (abs_frame_num - 1) % sps->poc_cycle_length;
+
+            expectedpoc = poc_cycle_cnt * expected_delta_per_poc_cycle;
+            for (i = 0; i <= frame_num_in_poc_cycle; i++)
+                expectedpoc = expectedpoc + sps->offset_for_ref_frame[i];
+        } else
+            expectedpoc = 0;
+
+        if (nal_ref_idc == 0)
+            expectedpoc = expectedpoc + sps->offset_for_non_ref_pic;
+
+        field_poc[0] = expectedpoc + pc->delta_poc[0];
+        field_poc[1] = field_poc[0] + sps->offset_for_top_to_bottom_field;
+
+        if (picture_structure == PICT_FRAME)
+            field_poc[1] += pc->delta_poc[1];
+    } else {
+        int poc = 2 * (pc->frame_num_offset + pc->frame_num);
+
+        if (!nal_ref_idc)
+            poc--;
+
+        field_poc[0] = poc;
+        field_poc[1] = poc;
+    }
+
+    if (picture_structure != PICT_BOTTOM_FIELD)
+        pic_field_poc[0] = field_poc[0];
+    if (picture_structure != PICT_TOP_FIELD)
+        pic_field_poc[1] = field_poc[1];
+    *pic_poc = FFMIN(pic_field_poc[0], pic_field_poc[1]);
+
+    return 0;
+}
diff --git a/libavcodec/h264_parse.h b/libavcodec/h264_parse.h
index 23676b36c7..c47b420e49 100644
--- a/libavcodec/h264_parse.h
+++ b/libavcodec/h264_parse.h
@@ -39,6 +39,19 @@ typedef struct H264PredWeightTable {
     int implicit_weight[48][48][2];
 } H264PredWeightTable;
 
+typedef struct H264POCContext {
+    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
+} H264POCContext;
+
 struct SPS;
 struct PPS;
 
@@ -65,4 +78,8 @@ int ff_h264_parse_ref_count(int *plist_count, int ref_count[2],
                             GetBitContext *gb, const struct PPS *pps,
                             int slice_type_nos, int picture_structure);
 
+int ff_h264_init_poc(int pic_field_poc[2], int *pic_poc,
+                     const struct SPS *sps, H264POCContext *poc,
+                     int picture_structure, int nal_ref_idc);
+
 #endif /* AVCODEC_H264_PARSE_H */
diff --git a/libavcodec/h264_parser.c b/libavcodec/h264_parser.c
index d62eb73515..991a84128a 100644
--- a/libavcodec/h264_parser.c
+++ b/libavcodec/h264_parser.c
@@ -47,6 +47,7 @@ typedef struct H264ParseContext {
     ParseContext pc;
     H264ParamSets ps;
     H264DSPContext h264dsp;
+    H264POCContext poc;
     int got_first;
 } H264ParseContext;
 
@@ -274,10 +275,10 @@ static inline int parse_nal_units(AVCodecParserContext *s,
         case NAL_IDR_SLICE:
             s->key_frame = 1;
 
-            h->prev_frame_num        = 0;
-            h->prev_frame_num_offset = 0;
-            h->prev_poc_msb          =
-            h->prev_poc_lsb          = 0;
+            p->poc.prev_frame_num        = 0;
+            p->poc.prev_frame_num_offset = 0;
+            p->poc.prev_poc_msb          =
+            p->poc.prev_poc_lsb          = 0;
         /* fall through */
         case NAL_SLICE:
             get_ue_golomb(&nal.gb);  // skip first_mb_in_slice
@@ -310,7 +311,7 @@ static inline int parse_nal_units(AVCodecParserContext *s,
             h->ps.pps = p->ps.pps;
             sps = p->ps.sps;
 
-            h->frame_num = get_bits(&nal.gb, sps->log2_max_frame_num);
+            p->poc.frame_num = get_bits(&nal.gb, sps->log2_max_frame_num);
 
             s->coded_width  = 16 * sps->mb_width;
             s->coded_height = 16 * sps->mb_height;
@@ -357,26 +358,27 @@ static inline int parse_nal_units(AVCodecParserContext *s,
             if (h->nal_unit_type == NAL_IDR_SLICE)
                 get_ue_golomb(&nal.gb); /* idr_pic_id */
             if (sps->poc_type == 0) {
-                h->poc_lsb = get_bits(&nal.gb, sps->log2_max_poc_lsb);
+                p->poc.poc_lsb = get_bits(&nal.gb, sps->log2_max_poc_lsb);
 
                 if (p->ps.pps->pic_order_present == 1 &&
                     h->picture_structure == PICT_FRAME)
-                    h->delta_poc_bottom = get_se_golomb(&nal.gb);
+                    p->poc.delta_poc_bottom = get_se_golomb(&nal.gb);
             }
 
             if (sps->poc_type == 1 &&
                 !sps->delta_pic_order_always_zero_flag) {
-                h->delta_poc[0] = get_se_golomb(&nal.gb);
+                p->poc.delta_poc[0] = get_se_golomb(&nal.gb);
 
                 if (p->ps.pps->pic_order_present == 1 &&
                     h->picture_structure == PICT_FRAME)
-                    h->delta_poc[1] = get_se_golomb(&nal.gb);
+                    p->poc.delta_poc[1] = get_se_golomb(&nal.gb);
             }
 
             /* Decode POC of this picture.
              * The prev_ values needed for decoding POC of the next picture are not set here. */
             field_poc[0] = field_poc[1] = INT_MAX;
-            ff_init_poc(h, field_poc, &s->output_picture_number);
+            ff_h264_init_poc(field_poc, &s->output_picture_number, sps,
+                             &p->poc, h->picture_structure, nal.ref_idc);
 
             /* Continue parsing to check if MMCO_RESET is present.
              * FIXME: MMCO_RESET could appear in non-first slice.
@@ -389,15 +391,15 @@ static inline int parse_nal_units(AVCodecParserContext *s,
             }
 
             /* Set up the prev_ values for decoding POC of the next picture. */
-            h->prev_frame_num        = got_reset ? 0 : h->frame_num;
-            h->prev_frame_num_offset = got_reset ? 0 : h->frame_num_offset;
+            p->poc.prev_frame_num        = got_reset ? 0 : p->poc.frame_num;
+            p->poc.prev_frame_num_offset = got_reset ? 0 : p->poc.frame_num_offset;
             if (h->nal_ref_idc != 0) {
                 if (!got_reset) {
-                    h->prev_poc_msb = h->poc_msb;
-                    h->prev_poc_lsb = h->poc_lsb;
+                    p->poc.prev_poc_msb = p->poc.poc_msb;
+                    p->poc.prev_poc_lsb = p->poc.poc_lsb;
                 } else {
-                    h->prev_poc_msb = 0;
-                    h->prev_poc_lsb =
+                    p->poc.prev_poc_msb = 0;
+                    p->poc.prev_poc_lsb =
                         h->picture_structure == PICT_BOTTOM_FIELD ? 0 : field_poc[0];
                 }
             }
diff --git a/libavcodec/h264_picture.c b/libavcodec/h264_picture.c
index 3e2c84e566..4777fd6f34 100644
--- a/libavcodec/h264_picture.c
+++ b/libavcodec/h264_picture.c
@@ -156,11 +156,11 @@ int ff_h264_field_end(H264Context *h, H264SliceContext *sl, int in_setup)
     if (in_setup || !(avctx->active_thread_type & FF_THREAD_FRAME)) {
         if (!h->droppable) {
             err = ff_h264_execute_ref_pic_marking(h, h->mmco, h->mmco_index);
-            h->prev_poc_msb = h->poc_msb;
-            h->prev_poc_lsb = h->poc_lsb;
+            h->poc.prev_poc_msb = h->poc.poc_msb;
+            h->poc.prev_poc_lsb = h->poc.poc_lsb;
         }
-        h->prev_frame_num_offset = h->frame_num_offset;
-        h->prev_frame_num        = h->frame_num;
+        h->poc.prev_frame_num_offset = h->poc.frame_num_offset;
+        h->poc.prev_frame_num        = h->poc.frame_num;
     }
 
     if (avctx->hwaccel) {
diff --git a/libavcodec/h264_refs.c b/libavcodec/h264_refs.c
index e0a9334806..08c3bffaf1 100644
--- a/libavcodec/h264_refs.c
+++ b/libavcodec/h264_refs.c
@@ -656,7 +656,7 @@ int ff_h264_execute_ref_pic_marking(H264Context *h, MMCO *mmco, int mmco_count)
             for (j = 0; j < 16; j++) {
                 remove_long(h, j, 0);
             }
-            h->frame_num  = h->cur_pic_ptr->frame_num = 0;
+            h->poc.frame_num = h->cur_pic_ptr->frame_num = 0;
             h->mmco_reset = 1;
             h->cur_pic_ptr->mmco_reset = 1;
             break;
diff --git a/libavcodec/h264_slice.c b/libavcodec/h264_slice.c
index e2617e2ad4..82a5a82851 100644
--- a/libavcodec/h264_slice.c
+++ b/libavcodec/h264_slice.c
@@ -409,7 +409,7 @@ int ff_h264_update_thread_context(AVCodecContext *dst,
     h->nal_length_size = h1->nal_length_size;
 
     // POC timing
-    copy_fields(h, h1, poc_lsb, current_slice);
+    copy_fields(h, h1, poc, current_slice);
 
     copy_picture_range(h->short_ref, h1->short_ref, 32, h, h1);
     copy_picture_range(h->long_ref, h1->long_ref, 32, h, h1);
@@ -421,11 +421,11 @@ int ff_h264_update_thread_context(AVCodecContext *dst,
 
     if (!h->droppable) {
         err = ff_h264_execute_ref_pic_marking(h, h->mmco, h->mmco_index);
-        h->prev_poc_msb = h->poc_msb;
-        h->prev_poc_lsb = h->poc_lsb;
+        h->poc.prev_poc_msb = h->poc.poc_msb;
+        h->poc.prev_poc_lsb = h->poc.poc_lsb;
     }
-    h->prev_frame_num_offset = h->frame_num_offset;
-    h->prev_frame_num        = h->frame_num;
+    h->poc.prev_frame_num_offset = h->poc.frame_num_offset;
+    h->poc.prev_frame_num        = h->poc.frame_num;
 
     h->recovery_frame        = h1->recovery_frame;
     h->frame_recovered       = h1->frame_recovered;
@@ -447,7 +447,7 @@ static int h264_frame_start(H264Context *h)
     pic->reference              = h->droppable ? 0 : h->picture_structure;
     pic->f->coded_picture_number = h->coded_picture_number++;
     pic->field_picture          = h->picture_structure != PICT_FRAME;
-    pic->frame_num               = h->frame_num;
+    pic->frame_num               = h->poc.frame_num;
     /*
      * Zero key_frame here; IDR markings per slice in frame or fields are ORed
      * in later.
@@ -1106,7 +1106,7 @@ int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl)
 
     frame_num = get_bits(&sl->gb, sps->log2_max_frame_num);
     if (!h->setup_finished)
-        h->frame_num = frame_num;
+        h->poc.frame_num = frame_num;
 
     sl->mb_mbaff       = 0;
 
@@ -1149,19 +1149,19 @@ int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl)
     } else {
         /* Shorten frame num gaps so we don't have to allocate reference
          * frames just to throw them away */
-        if (h->frame_num != h->prev_frame_num) {
-            int unwrap_prev_frame_num = h->prev_frame_num;
+        if (h->poc.frame_num != h->poc.prev_frame_num) {
+            int unwrap_prev_frame_num = h->poc.prev_frame_num;
             int max_frame_num         = 1 << sps->log2_max_frame_num;
 
-            if (unwrap_prev_frame_num > h->frame_num)
+            if (unwrap_prev_frame_num > h->poc.frame_num)
                 unwrap_prev_frame_num -= max_frame_num;
 
-            if ((h->frame_num - unwrap_prev_frame_num) > sps->ref_frame_count) {
-                unwrap_prev_frame_num = (h->frame_num - sps->ref_frame_count) - 1;
+            if ((h->poc.frame_num - unwrap_prev_frame_num) > sps->ref_frame_count) {
+                unwrap_prev_frame_num = (h->poc.frame_num - sps->ref_frame_count) - 1;
                 if (unwrap_prev_frame_num < 0)
                     unwrap_prev_frame_num += max_frame_num;
 
-                h->prev_frame_num = unwrap_prev_frame_num;
+                h->poc.prev_frame_num = unwrap_prev_frame_num;
             }
         }
 
@@ -1184,7 +1184,7 @@ int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl)
                                               last_pic_structure == PICT_TOP_FIELD);
                 }
             } else {
-                if (h->cur_pic_ptr->frame_num != h->frame_num) {
+                if (h->cur_pic_ptr->frame_num != h->poc.frame_num) {
                     /* This and previous field were reference, but had
                      * different frame_nums. Consider this field first in
                      * pair. Throw away previous field except for reference
@@ -1216,20 +1216,20 @@ int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl)
             }
         }
 
-        while (h->frame_num != h->prev_frame_num &&
-               h->frame_num != (h->prev_frame_num + 1) % (1 << sps->log2_max_frame_num)) {
+        while (h->poc.frame_num != h->poc.prev_frame_num &&
+               h->poc.frame_num != (h->poc.prev_frame_num + 1) % (1 << sps->log2_max_frame_num)) {
             H264Picture *prev = h->short_ref_count ? h->short_ref[0] : NULL;
             av_log(h->avctx, AV_LOG_DEBUG, "Frame num gap %d %d\n",
-                   h->frame_num, h->prev_frame_num);
+                   h->poc.frame_num, h->poc.prev_frame_num);
             ret = initialize_cur_frame(h);
             if (ret < 0) {
                 h->first_field = 0;
                 return ret;
             }
 
-            h->prev_frame_num++;
-            h->prev_frame_num        %= 1 << sps->log2_max_frame_num;
-            h->cur_pic_ptr->frame_num = h->prev_frame_num;
+            h->poc.prev_frame_num++;
+            h->poc.prev_frame_num        %= 1 << sps->log2_max_frame_num;
+            h->cur_pic_ptr->frame_num = h->poc.prev_frame_num;
             ff_thread_report_progress(&h->cur_pic_ptr->tf, INT_MAX, 0);
             ff_thread_report_progress(&h->cur_pic_ptr->tf, INT_MAX, 1);
             ret = ff_generate_sliding_window_mmcos(h, 1);
@@ -1259,7 +1259,7 @@ int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl)
                                   h->mb_height * 16);
                     h->short_ref[0]->poc = prev->poc + 2;
                 }
-                h->short_ref[0]->frame_num = h->prev_frame_num;
+                h->short_ref[0]->frame_num = h->poc.prev_frame_num;
             }
         }
 
@@ -1278,7 +1278,7 @@ int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl)
                 h->cur_pic_ptr = NULL;
                 h->first_field = FIELD_PICTURE(h);
             } else {
-                if (h->cur_pic_ptr->frame_num != h->frame_num) {
+                if (h->cur_pic_ptr->frame_num != h->poc.frame_num) {
                     /* This and the previous field had different frame_nums.
                      * Consider this field first in pair. Throw away previous
                      * one except for reference purposes. */
@@ -1318,10 +1318,10 @@ int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl)
     assert(sl->mb_y < h->mb_height);
 
     if (h->picture_structure == PICT_FRAME) {
-        h->curr_pic_num = h->frame_num;
+        h->curr_pic_num = h->poc.frame_num;
         h->max_pic_num  = 1 << sps->log2_max_frame_num;
     } else {
-        h->curr_pic_num = 2 * h->frame_num + 1;
+        h->curr_pic_num = 2 * h->poc.frame_num + 1;
         h->max_pic_num  = 1 << (sps->log2_max_frame_num + 1);
     }
 
@@ -1332,12 +1332,12 @@ int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl)
         int poc_lsb = get_bits(&sl->gb, sps->log2_max_poc_lsb);
 
         if (!h->setup_finished)
-            h->poc_lsb = poc_lsb;
+            h->poc.poc_lsb = poc_lsb;
 
         if (pps->pic_order_present == 1 && h->picture_structure == PICT_FRAME) {
             int delta_poc_bottom = get_se_golomb(&sl->gb);
             if (!h->setup_finished)
-                h->delta_poc_bottom = delta_poc_bottom;
+                h->poc.delta_poc_bottom = delta_poc_bottom;
         }
     }
 
@@ -1345,18 +1345,19 @@ int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl)
         int delta_poc = get_se_golomb(&sl->gb);
 
         if (!h->setup_finished)
-            h->delta_poc[0] = delta_poc;
+            h->poc.delta_poc[0] = delta_poc;
 
         if (pps->pic_order_present == 1 && h->picture_structure == PICT_FRAME) {
             delta_poc = get_se_golomb(&sl->gb);
 
             if (!h->setup_finished)
-                h->delta_poc[1] = delta_poc;
+                h->poc.delta_poc[1] = delta_poc;
         }
     }
 
     if (!h->setup_finished)
-        ff_init_poc(h, h->cur_pic_ptr->field_poc, &h->cur_pic_ptr->poc);
+        ff_h264_init_poc(h->cur_pic_ptr->field_poc, &h->cur_pic_ptr->poc,
+                         sps, &h->poc, h->picture_structure, h->nal_ref_idc);
 
     if (pps->redundant_pic_cnt_present)
         sl->redundant_pic_count = get_ue_golomb(&sl->gb);
@@ -1557,7 +1558,7 @@ int ff_h264_decode_slice_header(H264Context *h, H264SliceContext *sl)
                av_get_picture_type_char(sl->slice_type),
                sl->slice_type_fixed ? " fix" : "",
                h->nal_unit_type == NAL_IDR_SLICE ? " IDR" : "",
-               pps_id, h->frame_num,
+               pps_id, h->poc.frame_num,
                h->cur_pic_ptr->field_poc[0],
                h->cur_pic_ptr->field_poc[1],
                sl->ref_count[0], sl->ref_count[1],
diff --git a/libavcodec/vaapi_h264.c b/libavcodec/vaapi_h264.c
index 9c17ac1374..5dc90adc52 100644
--- a/libavcodec/vaapi_h264.c
+++ b/libavcodec/vaapi_h264.c
@@ -278,7 +278,7 @@ static int vaapi_h264_start_frame(AVCodecContext          *avctx,
     pic_param->pic_fields.bits.deblocking_filter_control_present_flag = pps->deblocking_filter_parameters_present;
     pic_param->pic_fields.bits.redundant_pic_cnt_present_flag   = pps->redundant_pic_cnt_present;
     pic_param->pic_fields.bits.reference_pic_flag               = h->nal_ref_idc != 0;
-    pic_param->frame_num                                        = h->frame_num;
+    pic_param->frame_num                                        = h->poc.frame_num;
 
     /* Fill in VAIQMatrixBufferH264. */
     iq_matrix = ff_vaapi_alloc_iq_matrix(vactx, sizeof(VAIQMatrixBufferH264));
diff --git a/libavcodec/vdpau_h264.c b/libavcodec/vdpau_h264.c
index 877e4e6435..5aeca67ccb 100644
--- a/libavcodec/vdpau_h264.c
+++ b/libavcodec/vdpau_h264.c
@@ -134,7 +134,7 @@ static int vdpau_h264_start_frame(AVCodecContext *avctx,
     info->field_order_cnt[0]                     = h264_foc(pic->field_poc[0]);
     info->field_order_cnt[1]                     = h264_foc(pic->field_poc[1]);
     info->is_reference                           = h->nal_ref_idc != 0;
-    info->frame_num                              = h->frame_num;
+    info->frame_num                              = h->poc.frame_num;
     info->field_pic_flag                         = h->picture_structure != PICT_FRAME;
     info->bottom_field_flag                      = h->picture_structure == PICT_BOTTOM_FIELD;
     info->num_ref_frames                         = sps->ref_frame_count;