av1/common/mvref_common.c

/*
 * Copyright (c) 2016, Alliance for Open Media. All rights reserved
 *
 * This source code is subject to the terms of the BSD 2 Clause License and
 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
 * was not distributed with this source code in the LICENSE file, you can
 * obtain it at www.aomedia.org/license/software. If the Alliance for Open
 * Media Patent License 1.0 was not distributed with this source code in the
 * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
 */

#include "av1/common/mvref_common.h"
#include "av1/common/warped_motion.h"

#define USE_CUR_GM_REFMV 1

#if CONFIG_MFMV
// Although we assign 32 bit integers, all the values are strictly under 14
// bits.
static int div_mult[64] = {
  0,    16384, 8192, 5461, 4096, 3276, 2730, 2340, 2048, 1820, 1638, 1489, 1365,
  1260, 1170,  1092, 1024, 963,  910,  862,  819,  780,  744,  712,  682,  655,
  630,  606,   585,  564,  546,  528,  512,  496,  481,  468,  455,  442,  431,
  420,  409,   399,  390,  381,  372,  364,  356,  348,  341,  334,  327,  321,
  315,  309,   303,  297,  292,  287,  282,  277,  273,  268,  264,  260,
};

// TODO(jingning): Consider the use of lookup table for (num / den)
// altogether.
static void get_mv_projection(MV *output, MV ref, int num, int den) {
  output->row =
      (int16_t)(ROUND_POWER_OF_TWO_SIGNED(ref.row * num * div_mult[den], 14));
  output->col =
      (int16_t)(ROUND_POWER_OF_TWO_SIGNED(ref.col * num * div_mult[den], 14));
}
#endif  // CONFIG_MFMV

void av1_copy_frame_mvs(const AV1_COMMON *const cm, MODE_INFO *mi, int mi_row,
                        int mi_col, int x_mis, int y_mis) {
#if CONFIG_TMV || CONFIG_MFMV
  const int frame_mvs_stride = ROUND_POWER_OF_TWO(cm->mi_cols, 1);
  MV_REF *frame_mvs =
      cm->cur_frame->mvs + (mi_row >> 1) * frame_mvs_stride + (mi_col >> 1);
  x_mis = ROUND_POWER_OF_TWO(x_mis, 1);
  y_mis = ROUND_POWER_OF_TWO(y_mis, 1);
#else
  const int frame_mvs_stride = cm->mi_cols;
  MV_REF *frame_mvs = cm->cur_frame->mvs +
                      (mi_row & 0xfffe) * frame_mvs_stride + (mi_col & 0xfffe);
  x_mis = AOMMAX(x_mis, 2);
  y_mis = AOMMAX(y_mis, 2);
#endif  // CONFIG_TMV
  int w, h;

  for (h = 0; h < y_mis; h++) {
    MV_REF *mv = frame_mvs;
    for (w = 0; w < x_mis; w++) {
#if CONFIG_MFMV
      mv->ref_frame[0] = NONE_FRAME;
      mv->ref_frame[1] = NONE_FRAME;
      mv->mv[0].as_int = 0;
      mv->mv[1].as_int = 0;

      for (int idx = 0; idx < 2; ++idx) {
        MV_REFERENCE_FRAME ref_frame = mi->mbmi.ref_frame[idx];
        if (ref_frame > INTRA_FRAME) {
          int8_t ref_idx = cm->ref_frame_side[ref_frame];
          if (ref_idx < 0) continue;
          if ((abs(mi->mbmi.mv[idx].as_mv.row) > REFMVS_LIMIT) ||
              (abs(mi->mbmi.mv[idx].as_mv.col) > REFMVS_LIMIT))
            continue;
          mv->ref_frame[ref_idx] = ref_frame;
          mv->mv[ref_idx].as_int = mi->mbmi.mv[idx].as_int;
        }
      }
#else
      mv->ref_frame[0] = mi->mbmi.ref_frame[0];
      mv->ref_frame[1] = mi->mbmi.ref_frame[1];
      mv->mv[0].as_int = mi->mbmi.mv[0].as_int;
      mv->mv[1].as_int = mi->mbmi.mv[1].as_int;
#endif
      // (TODO:yunqing) The following 2 lines won't be used and can be removed.
      mv->pred_mv[0].as_int = mi->mbmi.pred_mv[0].as_int;
      mv->pred_mv[1].as_int = mi->mbmi.pred_mv[1].as_int;
      mv++;
    }
    frame_mvs += frame_mvs_stride;
  }
}

static uint8_t add_ref_mv_candidate(
    const MODE_INFO *const candidate_mi, const MB_MODE_INFO *const candidate,
    const MV_REFERENCE_FRAME rf[2], uint8_t refmv_counts[MODE_CTX_REF_FRAMES],
    uint8_t ref_match_counts[MODE_CTX_REF_FRAMES],
    CANDIDATE_MV ref_mv_stacks[][MAX_REF_MV_STACK_SIZE], const int use_hp,
    int len,
#if USE_CUR_GM_REFMV
    int_mv *gm_mv_candidates, const WarpedMotionParams *gm_params,
#endif  // USE_CUR_GM_REFMV
    int col, int weight
#if CONFIG_AMVR
    ,
    int is_integer
#endif
) {
#if CONFIG_INTRABC
  if (!is_inter_block(candidate)) return 0;
#endif  // CONFIG_INTRABC
  int index = 0, ref;
  int newmv_count = 0;
  assert(weight % 2 == 0);

  if (rf[1] == NONE_FRAME) {
    uint8_t *refmv_count = &refmv_counts[rf[0]];
    uint8_t *ref_match_count = &ref_match_counts[rf[0]];
    CANDIDATE_MV *ref_mv_stack = ref_mv_stacks[rf[0]];
    (void)ref_match_count;

    // single reference frame
    for (ref = 0; ref < 2; ++ref) {
      if (candidate->ref_frame[ref] == rf[0]) {
        int_mv this_refmv;
#if USE_CUR_GM_REFMV
        if (is_global_mv_block(candidate_mi, gm_params[rf[0]].wmtype))
          this_refmv = gm_mv_candidates[0];
        else
#endif  // USE_CUR_GM_REFMV
          this_refmv = get_sub_block_mv(candidate_mi, ref, col);
#if CONFIG_AMVR
        lower_mv_precision(&this_refmv.as_mv, use_hp, is_integer);
#else
        lower_mv_precision(&this_refmv.as_mv, use_hp);
#endif  // CONFIG_AMVR

        for (index = 0; index < *refmv_count; ++index)
          if (ref_mv_stack[index].this_mv.as_int == this_refmv.as_int) break;

        if (index < *refmv_count) ref_mv_stack[index].weight += weight * len;

        // Add a new item to the list.
        if (index == *refmv_count && *refmv_count < MAX_REF_MV_STACK_SIZE) {
          ref_mv_stack[index].this_mv = this_refmv;
          ref_mv_stack[index].pred_diff[0] = av1_get_pred_diff_ctx(
              get_sub_block_pred_mv(candidate_mi, ref, col), this_refmv);
          ref_mv_stack[index].weight = weight * len;
          ++(*refmv_count);

#if !CONFIG_OPT_REF_MV
          if (candidate->mode == NEWMV) ++newmv_count;
#endif
        }
#if CONFIG_OPT_REF_MV
        if (have_newmv_in_inter_mode(candidate->mode)) ++newmv_count;
        ++*ref_match_count;
#endif
      }
    }
  } else {
    MV_REFERENCE_FRAME ref_frame = av1_ref_frame_type(rf);
    uint8_t *refmv_count = &refmv_counts[ref_frame];
    uint8_t *ref_match_count = &ref_match_counts[ref_frame];
    CANDIDATE_MV *ref_mv_stack = ref_mv_stacks[ref_frame];
    (void)ref_match_count;

    // compound reference frame
    if (candidate->ref_frame[0] == rf[0] && candidate->ref_frame[1] == rf[1]) {
      int_mv this_refmv[2];

      for (ref = 0; ref < 2; ++ref) {
#if USE_CUR_GM_REFMV
        if (is_global_mv_block(candidate_mi, gm_params[rf[ref]].wmtype))
          this_refmv[ref] = gm_mv_candidates[ref];
        else
#endif  // USE_CUR_GM_REFMV
          this_refmv[ref] = get_sub_block_mv(candidate_mi, ref, col);
#if CONFIG_AMVR
        lower_mv_precision(&this_refmv[ref].as_mv, use_hp, is_integer);
#else
        lower_mv_precision(&this_refmv[ref].as_mv, use_hp);
#endif
      }

      for (index = 0; index < *refmv_count; ++index)
        if ((ref_mv_stack[index].this_mv.as_int == this_refmv[0].as_int) &&
            (ref_mv_stack[index].comp_mv.as_int == this_refmv[1].as_int))
          break;

      if (index < *refmv_count) ref_mv_stack[index].weight += weight * len;

      // Add a new item to the list.
      if (index == *refmv_count && *refmv_count < MAX_REF_MV_STACK_SIZE) {
        ref_mv_stack[index].this_mv = this_refmv[0];
        ref_mv_stack[index].comp_mv = this_refmv[1];
        ref_mv_stack[index].pred_diff[0] = av1_get_pred_diff_ctx(
            get_sub_block_pred_mv(candidate_mi, 0, col), this_refmv[0]);
        ref_mv_stack[index].pred_diff[1] = av1_get_pred_diff_ctx(
            get_sub_block_pred_mv(candidate_mi, 1, col), this_refmv[1]);
        ref_mv_stack[index].weight = weight * len;
        ++(*refmv_count);

#if !CONFIG_OPT_REF_MV
        if (candidate->mode == NEW_NEWMV) ++newmv_count;
#endif
      }
#if CONFIG_OPT_REF_MV
      if (have_newmv_in_inter_mode(candidate->mode)) ++newmv_count;
      ++*ref_match_count;
#endif
    }
  }
  return newmv_count;
}

static uint8_t scan_row_mbmi(const AV1_COMMON *cm, const MACROBLOCKD *xd,
                             int mi_row, int mi_col,
                             const MV_REFERENCE_FRAME rf[2], int row_offset,
                             CANDIDATE_MV ref_mv_stack[][MAX_REF_MV_STACK_SIZE],
                             uint8_t refmv_count[MODE_CTX_REF_FRAMES],
                             uint8_t ref_match_count[MODE_CTX_REF_FRAMES],
#if USE_CUR_GM_REFMV
                             int_mv *gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
                             int max_row_offset, int *processed_rows) {
  int end_mi = AOMMIN(xd->n8_w, cm->mi_cols - mi_col);
  end_mi = AOMMIN(end_mi, mi_size_wide[BLOCK_64X64]);
  const int n8_w_8 = mi_size_wide[BLOCK_8X8];
  const int n8_w_16 = mi_size_wide[BLOCK_16X16];
  int i;
  uint8_t newmv_count = 0;
  int col_offset = 0;
  const int shift = 0;
  // TODO(jingning): Revisit this part after cb4x4 is stable.
  if (abs(row_offset) > 1) {
    col_offset = 1;
    if (mi_col & 0x01 && xd->n8_w < n8_w_8) --col_offset;
  }
  const int use_step_16 = (xd->n8_w >= 16);
  MODE_INFO **const candidate_mi0 = xd->mi + row_offset * xd->mi_stride;
  (void)mi_row;

  for (i = 0; i < end_mi;) {
    const MODE_INFO *const candidate_mi = candidate_mi0[col_offset + i];
    const MB_MODE_INFO *const candidate = &candidate_mi->mbmi;
    const int candidate_bsize = candidate->sb_type;
    const int n8_w = mi_size_wide[candidate_bsize];
    int len = AOMMIN(xd->n8_w, n8_w);
    if (use_step_16)
      len = AOMMAX(n8_w_16, len);
    else if (abs(row_offset) > 1)
      len = AOMMAX(len, n8_w_8);

    int weight = 2;
    if (xd->n8_w >= n8_w_8 && xd->n8_w <= n8_w) {
      int inc = AOMMIN(-max_row_offset + row_offset + 1,
                       mi_size_high[candidate_bsize]);
      // Obtain range used in weight calculation.
      weight = AOMMAX(weight, (inc << shift));
      // Update processed rows.
      *processed_rows = inc - row_offset - 1;
    }

#if CONFIG_AMVR
    newmv_count += add_ref_mv_candidate(
        candidate_mi, candidate, rf, refmv_count, ref_match_count, ref_mv_stack,
        cm->allow_high_precision_mv, len,
#if USE_CUR_GM_REFMV
        gm_mv_candidates, cm->global_motion,
#endif  // USE_CUR_GM_REFMV
        col_offset + i, weight, cm->cur_frame_force_integer_mv);
#else
    newmv_count += add_ref_mv_candidate(
        candidate_mi, candidate, rf, refmv_count, ref_match_count, ref_mv_stack,
        cm->allow_high_precision_mv, len,
#if USE_CUR_GM_REFMV
        gm_mv_candidates, cm->global_motion,
#endif  // USE_CUR_GM_REFMV
        col_offset + i, weight);
#endif

    i += len;
  }

  return newmv_count;
}

static uint8_t scan_col_mbmi(const AV1_COMMON *cm, const MACROBLOCKD *xd,
                             int mi_row, int mi_col,
                             const MV_REFERENCE_FRAME rf[2], int col_offset,
                             CANDIDATE_MV ref_mv_stack[][MAX_REF_MV_STACK_SIZE],
                             uint8_t refmv_count[MODE_CTX_REF_FRAMES],
                             uint8_t ref_match_count[MODE_CTX_REF_FRAMES],
#if USE_CUR_GM_REFMV
                             int_mv *gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
                             int max_col_offset, int *processed_cols) {
  int end_mi = AOMMIN(xd->n8_h, cm->mi_rows - mi_row);
  end_mi = AOMMIN(end_mi, mi_size_high[BLOCK_64X64]);
  const int n8_h_8 = mi_size_high[BLOCK_8X8];
  const int n8_h_16 = mi_size_high[BLOCK_16X16];
  int i;
  uint8_t newmv_count = 0;
  int row_offset = 0;
  const int shift = 0;
  if (abs(col_offset) > 1) {
    row_offset = 1;
    if (mi_row & 0x01 && xd->n8_h < n8_h_8) --row_offset;
  }
  const int use_step_16 = (xd->n8_h >= 16);
  (void)mi_col;

  for (i = 0; i < end_mi;) {
    const MODE_INFO *const candidate_mi =
        xd->mi[(row_offset + i) * xd->mi_stride + col_offset];
    const MB_MODE_INFO *const candidate = &candidate_mi->mbmi;
    const int candidate_bsize = candidate->sb_type;
    const int n8_h = mi_size_high[candidate_bsize];
    int len = AOMMIN(xd->n8_h, n8_h);
    if (use_step_16)
      len = AOMMAX(n8_h_16, len);
    else if (abs(col_offset) > 1)
      len = AOMMAX(len, n8_h_8);

    int weight = 2;
    if (xd->n8_h >= n8_h_8 && xd->n8_h <= n8_h) {
      int inc = AOMMIN(-max_col_offset + col_offset + 1,
                       mi_size_wide[candidate_bsize]);
      // Obtain range used in weight calculation.
      weight = AOMMAX(weight, (inc << shift));
      // Update processed cols.
      *processed_cols = inc - col_offset - 1;
    }

#if CONFIG_AMVR
    newmv_count += add_ref_mv_candidate(
        candidate_mi, candidate, rf, refmv_count, ref_match_count, ref_mv_stack,
        cm->allow_high_precision_mv, len,
#if USE_CUR_GM_REFMV
        gm_mv_candidates, cm->global_motion,
#endif  // USE_CUR_GM_REFMV
        col_offset, weight, cm->cur_frame_force_integer_mv);
#else
    newmv_count += add_ref_mv_candidate(
        candidate_mi, candidate, rf, refmv_count, ref_match_count, ref_mv_stack,
        cm->allow_high_precision_mv, len,
#if USE_CUR_GM_REFMV
        gm_mv_candidates, cm->global_motion,
#endif  // USE_CUR_GM_REFMV
        col_offset, weight);
#endif
    i += len;
  }

  return newmv_count;
}

static uint8_t scan_blk_mbmi(const AV1_COMMON *cm, const MACROBLOCKD *xd,
                             const int mi_row, const int mi_col,
                             const MV_REFERENCE_FRAME rf[2], int row_offset,
                             int col_offset,
                             CANDIDATE_MV ref_mv_stack[][MAX_REF_MV_STACK_SIZE],
                             uint8_t ref_match_count[MODE_CTX_REF_FRAMES],
#if USE_CUR_GM_REFMV
                             int_mv *gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
                             uint8_t refmv_count[MODE_CTX_REF_FRAMES]) {
  const TileInfo *const tile = &xd->tile;
  POSITION mi_pos;
  uint8_t newmv_count = 0;

  mi_pos.row = row_offset;
  mi_pos.col = col_offset;

  if (is_inside(tile, mi_col, mi_row, cm->mi_rows, cm, &mi_pos)) {
    const MODE_INFO *const candidate_mi =
        xd->mi[mi_pos.row * xd->mi_stride + mi_pos.col];
    const MB_MODE_INFO *const candidate = &candidate_mi->mbmi;
    const int len = mi_size_wide[BLOCK_8X8];

#if CONFIG_AMVR
    newmv_count += add_ref_mv_candidate(
        candidate_mi, candidate, rf, refmv_count, ref_match_count, ref_mv_stack,
        cm->allow_high_precision_mv, len,
#if USE_CUR_GM_REFMV
        gm_mv_candidates, cm->global_motion,
#endif  // USE_CUR_GM_REFMV
        mi_pos.col, 2, cm->cur_frame_force_integer_mv);
#else
    newmv_count += add_ref_mv_candidate(
        candidate_mi, candidate, rf, refmv_count, ref_match_count, ref_mv_stack,
        cm->allow_high_precision_mv, len,
#if USE_CUR_GM_REFMV
        gm_mv_candidates, cm->global_motion,
#endif  // USE_CUR_GM_REFMV
        mi_pos.col, 2);
#endif
  }  // Analyze a single 8x8 block motion information.

  return newmv_count;
}

static int has_top_right(const AV1_COMMON *cm, const MACROBLOCKD *xd,
                         int mi_row, int mi_col, int bs) {
  const int sb_mi_size = mi_size_wide[cm->seq_params.sb_size];
  const int mask_row = mi_row & (sb_mi_size - 1);
  const int mask_col = mi_col & (sb_mi_size - 1);

  if (bs > mi_size_wide[BLOCK_64X64]) return 0;

  // In a split partition all apart from the bottom right has a top right
  int has_tr = !((mask_row & bs) && (mask_col & bs));

  // bs > 0 and bs is a power of 2
  assert(bs > 0 && !(bs & (bs - 1)));

  // For each 4x4 group of blocks, when the bottom right is decoded the blocks
  // to the right have not been decoded therefore the bottom right does
  // not have a top right
  while (bs < sb_mi_size) {
    if (mask_col & bs) {
      if ((mask_col & (2 * bs)) && (mask_row & (2 * bs))) {
        has_tr = 0;
        break;
      }
    } else {
      break;
    }
    bs <<= 1;
  }

  // The left hand of two vertical rectangles always has a top right (as the
  // block above will have been decoded)
  if (xd->n8_w < xd->n8_h)
    if (!xd->is_sec_rect) has_tr = 1;

  // The bottom of two horizontal rectangles never has a top right (as the block
  // to the right won't have been decoded)
  if (xd->n8_w > xd->n8_h)
    if (xd->is_sec_rect) has_tr = 0;

#if CONFIG_EXT_PARTITION_TYPES
  // The bottom left square of a Vertical A (in the old format) does
  // not have a top right as it is decoded before the right hand
  // rectangle of the partition
  if (xd->mi[0]->mbmi.partition == PARTITION_VERT_A)
    if ((mask_row & bs) && !(mask_col & bs)) has_tr = 0;
#endif  // CONFIG_EXT_PARTITION_TYPES

  return has_tr;
}

#if CONFIG_MFMV
static int check_sb_border(const int mi_row, const int mi_col,
                           const int row_offset, const int col_offset) {
  const int sb_mi_size = mi_size_wide[BLOCK_64X64];
  const int row = mi_row & (sb_mi_size - 1);
  const int col = mi_col & (sb_mi_size - 1);

  if (row + row_offset < 0 || row + row_offset >= sb_mi_size ||
      col + col_offset < 0 || col + col_offset >= sb_mi_size)
    return 0;

  return 1;
}

static int add_tpl_ref_mv(const AV1_COMMON *cm,
                          const MV_REF *prev_frame_mvs_base,
                          const MACROBLOCKD *xd, int mi_row, int mi_col,
                          MV_REFERENCE_FRAME ref_frame, int blk_row,
                          int blk_col, int_mv *gm_mv_candidates,
                          uint8_t refmv_count[MODE_CTX_REF_FRAMES],
                          CANDIDATE_MV ref_mv_stacks[][MAX_REF_MV_STACK_SIZE],
                          int16_t *mode_context) {
  (void)prev_frame_mvs_base;
  POSITION mi_pos;
  int idx;
  int coll_blk_count = 0;
  const int weight_unit = 1;  // mi_size_wide[BLOCK_8X8];

  (void)gm_mv_candidates;

  mi_pos.row = (mi_row & 0x01) ? blk_row : blk_row + 1;
  mi_pos.col = (mi_col & 0x01) ? blk_col : blk_col + 1;

  if (!is_inside(&xd->tile, mi_col, mi_row, cm->mi_rows, cm, &mi_pos))
    return coll_blk_count;

  const TPL_MV_REF *prev_frame_mvs =
      cm->tpl_mvs + ((mi_row + mi_pos.row) >> 1) * (cm->mi_stride >> 1) +
      ((mi_col + mi_pos.col) >> 1);

  MV_REFERENCE_FRAME rf[2];
  av1_set_ref_frame(rf, ref_frame);

  if (rf[1] == NONE_FRAME) {
    int cur_frame_index = cm->cur_frame->cur_frame_offset;
    int buf_idx_0 = cm->frame_refs[FWD_RF_OFFSET(rf[0])].idx;
    int frame0_index = cm->buffer_pool->frame_bufs[buf_idx_0].cur_frame_offset;
    int cur_offset_0 = cur_frame_index - frame0_index;
    CANDIDATE_MV *ref_mv_stack = ref_mv_stacks[rf[0]];

    for (int i = 0; i < MFMV_STACK_SIZE; ++i) {
      if (prev_frame_mvs->mfmv0[i].as_int != INVALID_MV) {
        int_mv this_refmv;

        get_mv_projection(&this_refmv.as_mv, prev_frame_mvs->mfmv0[i].as_mv,
                          cur_offset_0, prev_frame_mvs->ref_frame_offset[i]);
#if CONFIG_AMVR
        lower_mv_precision(&this_refmv.as_mv, cm->allow_high_precision_mv,
                           cm->cur_frame_force_integer_mv);
#else
        lower_mv_precision(&this_refmv.as_mv, cm->allow_high_precision_mv);
#endif

#if CONFIG_OPT_REF_MV
        if (blk_row == 0 && blk_col == 0)
          if (abs(this_refmv.as_mv.row - gm_mv_candidates[0].as_mv.row) >= 16 ||
              abs(this_refmv.as_mv.col - gm_mv_candidates[0].as_mv.col) >= 16)
            mode_context[ref_frame] |= (1 << GLOBALMV_OFFSET);
#else
        if (blk_row == 0 && blk_col == 0)
          if (abs(this_refmv.as_mv.row) >= 16 ||
              abs(this_refmv.as_mv.col) >= 16)
            mode_context[ref_frame] |= (1 << GLOBALMV_OFFSET);
#endif

        for (idx = 0; idx < refmv_count[rf[0]]; ++idx)
          if (this_refmv.as_int == ref_mv_stack[idx].this_mv.as_int) break;

        if (idx < refmv_count[rf[0]])
          ref_mv_stack[idx].weight += 2 * weight_unit;

        if (idx == refmv_count[rf[0]] &&
            refmv_count[rf[0]] < MAX_REF_MV_STACK_SIZE) {
          ref_mv_stack[idx].this_mv.as_int = this_refmv.as_int;
          // TODO(jingning): Hard coded context number. Need to make it better
          // sense.
          ref_mv_stack[idx].pred_diff[0] = 1;
          ref_mv_stack[idx].weight = 2 * weight_unit;
          ++(refmv_count[rf[0]]);
        }

        ++coll_blk_count;
        return coll_blk_count;
      }
    }
  } else {
    // Process compound inter mode
    int cur_frame_index = cm->cur_frame->cur_frame_offset;
    int buf_idx_0 = cm->frame_refs[FWD_RF_OFFSET(rf[0])].idx;
    int frame0_index = cm->buffer_pool->frame_bufs[buf_idx_0].cur_frame_offset;

    int cur_offset_0 = cur_frame_index - frame0_index;
    int buf_idx_1 = cm->frame_refs[FWD_RF_OFFSET(rf[1])].idx;
    int frame1_index = cm->buffer_pool->frame_bufs[buf_idx_1].cur_frame_offset;
    int cur_offset_1 = cur_frame_index - frame1_index;
    CANDIDATE_MV *ref_mv_stack = ref_mv_stacks[ref_frame];

    for (int i = 0; i < MFMV_STACK_SIZE; ++i) {
      if (prev_frame_mvs->mfmv0[i].as_int != INVALID_MV) {
        int_mv this_refmv;
        int_mv comp_refmv;
        get_mv_projection(&this_refmv.as_mv, prev_frame_mvs->mfmv0[i].as_mv,
                          cur_offset_0, prev_frame_mvs->ref_frame_offset[i]);
        get_mv_projection(&comp_refmv.as_mv, prev_frame_mvs->mfmv0[i].as_mv,
                          cur_offset_1, prev_frame_mvs->ref_frame_offset[i]);

#if CONFIG_AMVR
        lower_mv_precision(&this_refmv.as_mv, cm->allow_high_precision_mv,
                           cm->cur_frame_force_integer_mv);
        lower_mv_precision(&comp_refmv.as_mv, cm->allow_high_precision_mv,
                           cm->cur_frame_force_integer_mv);
#else
        lower_mv_precision(&this_refmv.as_mv, cm->allow_high_precision_mv);
        lower_mv_precision(&comp_refmv.as_mv, cm->allow_high_precision_mv);
#endif

#if CONFIG_OPT_REF_MV
        if (blk_row == 0 && blk_col == 0)
          if (abs(this_refmv.as_mv.row - gm_mv_candidates[0].as_mv.row) >= 16 ||
              abs(this_refmv.as_mv.col - gm_mv_candidates[0].as_mv.col) >= 16 ||
              abs(comp_refmv.as_mv.row - gm_mv_candidates[1].as_mv.row) >= 16 ||
              abs(comp_refmv.as_mv.col - gm_mv_candidates[1].as_mv.col) >= 16)
            mode_context[ref_frame] |= (1 << GLOBALMV_OFFSET);
#else
        if (blk_row == 0 && blk_col == 0)
          if (abs(this_refmv.as_mv.row) >= 16 ||
              abs(this_refmv.as_mv.col) >= 16 ||
              abs(comp_refmv.as_mv.row) >= 16 ||
              abs(comp_refmv.as_mv.col) >= 16)
            mode_context[ref_frame] |= (1 << GLOBALMV_OFFSET);
#endif

        for (idx = 0; idx < refmv_count[ref_frame]; ++idx)
          if (this_refmv.as_int == ref_mv_stack[idx].this_mv.as_int &&
              comp_refmv.as_int == ref_mv_stack[idx].comp_mv.as_int)
            break;

        if (idx < refmv_count[ref_frame])
          ref_mv_stack[idx].weight += 2 * weight_unit;

        if (idx == refmv_count[ref_frame] &&
            refmv_count[ref_frame] < MAX_REF_MV_STACK_SIZE) {
          ref_mv_stack[idx].this_mv.as_int = this_refmv.as_int;
          ref_mv_stack[idx].comp_mv.as_int = comp_refmv.as_int;
          // TODO(jingning): Hard coded context number. Need to make it better
          // sense.
          ref_mv_stack[idx].pred_diff[0] = 1;
          ref_mv_stack[idx].pred_diff[1] = 1;
          ref_mv_stack[idx].weight = 2 * weight_unit;
          ++(refmv_count[ref_frame]);
        }

        ++coll_blk_count;
        return coll_blk_count;
      }
    }
  }

  return coll_blk_count;
}
#else
static int add_col_ref_mv(const AV1_COMMON *cm,
                          const MV_REF *prev_frame_mvs_base,
                          int prev_frame_mvs_stride, const MACROBLOCKD *xd,
                          int mi_row, int mi_col, MV_REFERENCE_FRAME ref_frame,
                          int blk_row, int blk_col,
                          uint8_t refmv_count[MODE_CTX_REF_FRAMES],
                          CANDIDATE_MV ref_mv_stacks[][MAX_REF_MV_STACK_SIZE],
                          int16_t *mode_context) {
#if CONFIG_TMV
  const MV_REF *prev_frame_mvs = prev_frame_mvs_base +
                                 (blk_row >> 1) * prev_frame_mvs_stride +
                                 (blk_col >> 1);
#else
  const MV_REF *prev_frame_mvs =
      prev_frame_mvs_base + blk_row * prev_frame_mvs_stride + blk_col;
#endif
  POSITION mi_pos;
  int ref, idx;
  int coll_blk_count = 0;
  const int weight_unit = mi_size_wide[BLOCK_8X8];
  CANDIDATE_MV *ref_mv_stack = ref_mv_stacks[ref_frame];

#if CONFIG_TMV
  mi_pos.row = blk_row;
  mi_pos.col = blk_col;
#else
  mi_pos.row = (mi_row & 0x01) ? blk_row : blk_row + 1;
  mi_pos.col = (mi_col & 0x01) ? blk_col : blk_col + 1;
#endif  // CONFIG_TMV

  if (!is_inside(&xd->tile, mi_col, mi_row, cm->mi_rows, cm, &mi_pos))
    return coll_blk_count;
  for (ref = 0; ref < 2; ++ref) {
    if (prev_frame_mvs->ref_frame[ref] == ref_frame) {
      int_mv this_refmv = prev_frame_mvs->mv[ref];
#if CONFIG_AMVR
      lower_mv_precision(&this_refmv.as_mv, cm->allow_high_precision_mv,
                         cm->cur_frame_force_integer_mv);
#else
      lower_mv_precision(&this_refmv.as_mv, cm->allow_high_precision_mv);
#endif

#if CONFIG_OPT_REF_MV
      if (blk_row == 0 && blk_col == 0)
#endif
      {
        if (abs(this_refmv.as_mv.row) >= 16 || abs(this_refmv.as_mv.col) >= 16)
          mode_context[ref_frame] |= (1 << GLOBALMV_OFFSET);
      }

      for (idx = 0; idx < refmv_count[ref_frame]; ++idx)
        if (this_refmv.as_int == ref_mv_stack[idx].this_mv.as_int) break;

      if (idx < refmv_count[ref_frame])
        ref_mv_stack[idx].weight += 2 * weight_unit;

      if (idx == refmv_count[ref_frame] &&
          refmv_count[ref_frame] < MAX_REF_MV_STACK_SIZE) {
        ref_mv_stack[idx].this_mv.as_int = this_refmv.as_int;
        ref_mv_stack[idx].pred_diff[0] =
            av1_get_pred_diff_ctx(prev_frame_mvs->pred_mv[ref], this_refmv);
        ref_mv_stack[idx].weight = 2 * weight_unit;
        ++(refmv_count[ref_frame]);
      }

      ++coll_blk_count;
    }
  }

  return coll_blk_count;
}
#endif  // CONFIG_MFMV

static void setup_ref_mv_list(
    const AV1_COMMON *cm, const MACROBLOCKD *xd, MV_REFERENCE_FRAME ref_frame,
    uint8_t refmv_count[MODE_CTX_REF_FRAMES],
    CANDIDATE_MV ref_mv_stack[][MAX_REF_MV_STACK_SIZE],
    int_mv mv_ref_list[][MAX_MV_REF_CANDIDATES],
#if USE_CUR_GM_REFMV
    int_mv *gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
    int mi_row, int mi_col, int16_t *mode_context, int compound_search) {
#if CONFIG_TMV
  const int prev_frame_mvs_stride = ROUND_POWER_OF_TWO(cm->mi_cols, 1);
  const int tmi_row = mi_row & 0xfffe;
  const int tmi_col = mi_col & 0xfffe;
  const MV_REF *const prev_frame_mvs_base =
      cm->use_prev_frame_mvs
          ? cm->prev_frame->mvs + (tmi_row >> 1) * prev_frame_mvs_stride +
                (tmi_col >> 1)
          : NULL;
#else
  const int prev_frame_mvs_stride = cm->mi_cols;
  const MV_REF *const prev_frame_mvs_base =
      cm->use_prev_frame_mvs
          ? cm->prev_frame->mvs +
                (((mi_row >> 1) << 1) + 1) * prev_frame_mvs_stride +
                ((mi_col >> 1) << 1) + 1
          : NULL;
#endif  // CONFIG_TMV

  const int bs = AOMMAX(xd->n8_w, xd->n8_h);
  const int has_tr = has_top_right(cm, xd, mi_row, mi_col, bs);
  MV_REFERENCE_FRAME rf[2];

  const TileInfo *const tile = &xd->tile;
  int max_row_offset = 0, max_col_offset = 0;
  const int row_adj = (xd->n8_h < mi_size_high[BLOCK_8X8]) && (mi_row & 0x01);
  const int col_adj = (xd->n8_w < mi_size_wide[BLOCK_8X8]) && (mi_col & 0x01);
  int processed_rows = 0;
  int processed_cols = 0;

  av1_set_ref_frame(rf, ref_frame);
  mode_context[ref_frame] = 0;
  refmv_count[ref_frame] = 0;
  (void)compound_search;

  // Find valid maximum row/col offset.
  if (xd->up_available) {
    max_row_offset = -(MVREF_ROWS << 1) + row_adj;
#if CONFIG_OPT_REF_MV
    if (xd->n8_h < mi_size_high[BLOCK_8X8])
      max_row_offset = -(2 << 1) + row_adj;
#endif
    max_row_offset =
        find_valid_row_offset(tile, mi_row, cm->mi_rows, cm, max_row_offset);
  }

  if (xd->left_available) {
    max_col_offset = -(MVREF_COLS << 1) + col_adj;
#if CONFIG_OPT_REF_MV
    if (xd->n8_w < mi_size_wide[BLOCK_8X8])
      max_col_offset = -(2 << 1) + col_adj;
#endif
    max_col_offset = find_valid_col_offset(tile, mi_col, max_col_offset);
  }

  uint8_t ref_match_count[MODE_CTX_REF_FRAMES] = { 0 };
  uint8_t newmv_count = 0;

  // Scan the first above row mode info. row_offset = -1;
  if (abs(max_row_offset) >= 1)
    newmv_count += scan_row_mbmi(cm, xd, mi_row, mi_col, rf, -1, ref_mv_stack,
                                 refmv_count, ref_match_count,
#if USE_CUR_GM_REFMV
                                 gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
                                 max_row_offset, &processed_rows);
  // Scan the first left column mode info. col_offset = -1;
  if (abs(max_col_offset) >= 1)
    newmv_count += scan_col_mbmi(cm, xd, mi_row, mi_col, rf, -1, ref_mv_stack,
                                 refmv_count, ref_match_count,
#if USE_CUR_GM_REFMV
                                 gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
                                 max_col_offset, &processed_cols);
  // Check top-right boundary
  if (has_tr)
    newmv_count += scan_blk_mbmi(cm, xd, mi_row, mi_col, rf, -1, xd->n8_w,
                                 ref_mv_stack, ref_match_count,
#if USE_CUR_GM_REFMV
                                 gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
                                 refmv_count);

  uint8_t nearest_match[MODE_CTX_REF_FRAMES];
  uint8_t nearest_refmv_count[MODE_CTX_REF_FRAMES];

  nearest_match[ref_frame] = ref_match_count[ref_frame];
  nearest_refmv_count[ref_frame] = refmv_count[ref_frame];

  // TODO(yunqing): for comp_search, do it for all 3 cases.
  for (int idx = 0; idx < nearest_refmv_count[ref_frame]; ++idx)
    ref_mv_stack[ref_frame][idx].weight += REF_CAT_LEVEL;

#if CONFIG_MFMV
  if (cm->use_ref_frame_mvs) {
    int coll_blk_count[MODE_CTX_REF_FRAMES] = { 0 };
    const int voffset = AOMMAX(mi_size_high[BLOCK_8X8], xd->n8_h);
    const int hoffset = AOMMAX(mi_size_wide[BLOCK_8X8], xd->n8_w);
    const int blk_row_end = AOMMIN(xd->n8_h, mi_size_high[BLOCK_64X64]);
    const int blk_col_end = AOMMIN(xd->n8_w, mi_size_wide[BLOCK_64X64]);

    const int tpl_sample_pos[3][2] = {
      { voffset, -2 },
      { voffset, hoffset },
      { voffset - 2, hoffset },
    };
    const int allow_extension = (xd->n8_h >= mi_size_high[BLOCK_8X8]) &&
                                (xd->n8_h < mi_size_high[BLOCK_64X64]) &&
                                (xd->n8_w >= mi_size_wide[BLOCK_8X8]) &&
                                (xd->n8_w < mi_size_wide[BLOCK_64X64]);

    for (int blk_row = 0; blk_row < blk_row_end;
         blk_row += mi_size_high[BLOCK_8X8]) {
      for (int blk_col = 0; blk_col < blk_col_end;
           blk_col += mi_size_wide[BLOCK_8X8]) {
        // (TODO: yunqing) prev_frame_mvs_base is not used here, tpl_mvs is
        // used.
        // Can be modified the same way.
        int is_available = add_tpl_ref_mv(
            cm, prev_frame_mvs_base, xd, mi_row, mi_col, ref_frame, blk_row,
            blk_col, gm_mv_candidates, refmv_count, ref_mv_stack, mode_context);
        if (blk_row == 0 && blk_col == 0)
          coll_blk_count[ref_frame] = is_available;
      }
    }

    if (coll_blk_count[ref_frame] == 0)
      mode_context[ref_frame] |= (1 << GLOBALMV_OFFSET);

    for (int i = 0; i < 3 && allow_extension; ++i) {
      const int blk_row = tpl_sample_pos[i][0];
      const int blk_col = tpl_sample_pos[i][1];

      if (!check_sb_border(mi_row, mi_col, blk_row, blk_col)) continue;
      // (TODO: yunqing) prev_frame_mvs_base is not used here, tpl_mvs is used.
      // Can be modified the same way.
      coll_blk_count[ref_frame] += add_tpl_ref_mv(
          cm, prev_frame_mvs_base, xd, mi_row, mi_col, ref_frame, blk_row,
          blk_col, gm_mv_candidates, refmv_count, ref_mv_stack, mode_context);
    }
  }
#else
  if (cm->use_prev_frame_mvs && rf[1] == NONE_FRAME) {
    int coll_blk_count[MODE_CTX_REF_FRAMES] = { 0 };
    const int mi_step = (xd->n8_w == 1 || xd->n8_h == 1)
                            ? mi_size_wide[BLOCK_8X8]
                            : mi_size_wide[BLOCK_16X16];

    for (int blk_row = 0; blk_row < xd->n8_h; blk_row += mi_step) {
      for (int blk_col = 0; blk_col < xd->n8_w; blk_col += mi_step) {
#if CONFIG_TMV
        int is_available =
            add_col_ref_mv(cm, prev_frame_mvs_base, prev_frame_mvs_stride, xd,
                           tmi_row, tmi_col, ref_frame, blk_row, blk_col,
                           refmv_count, ref_mv_stack, mode_context);
#else
        int is_available =
            add_col_ref_mv(cm, prev_frame_mvs_base, prev_frame_mvs_stride, xd,
                           mi_row, mi_col, ref_frame, blk_row, blk_col,
                           refmv_count, ref_mv_stack, mode_context);
#endif  // CONFIG_TMV
#if CONFIG_OPT_REF_MV
        if (blk_row == 0 && blk_col == 0)
          coll_blk_count[ref_frame] = is_available;
#else
        coll_blk_count[ref_frame] += is_available;
#endif
      }
    }

    if (coll_blk_count[ref_frame] == 0)
      mode_context[ref_frame] |= (1 << GLOBALMV_OFFSET);
  } else {
    mode_context[ref_frame] |= (1 << GLOBALMV_OFFSET);
  }
#endif  // CONFIG_MFMV

  // Scan the second outer area.
  scan_blk_mbmi(cm, xd, mi_row, mi_col, rf, -1, -1, ref_mv_stack,
                ref_match_count,
#if USE_CUR_GM_REFMV
                gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
                refmv_count);

  for (int idx = 2; idx <= MVREF_ROWS; ++idx) {
    const int row_offset = -(idx << 1) + 1 + row_adj;
    const int col_offset = -(idx << 1) + 1 + col_adj;

    if (abs(row_offset) <= abs(max_row_offset) &&
        abs(row_offset) > processed_rows)
      scan_row_mbmi(cm, xd, mi_row, mi_col, rf, row_offset, ref_mv_stack,
                    refmv_count, ref_match_count,
#if USE_CUR_GM_REFMV
                    gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
                    max_row_offset, &processed_rows);

    if (abs(col_offset) <= abs(max_col_offset) &&
        abs(col_offset) > processed_cols)
      scan_col_mbmi(cm, xd, mi_row, mi_col, rf, col_offset, ref_mv_stack,
                    refmv_count, ref_match_count,
#if USE_CUR_GM_REFMV
                    gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
                    max_col_offset, &processed_cols);
  }

  const int col_offset = -(MVREF_COLS << 1) + 1 + col_adj;
  if (abs(col_offset) <= abs(max_col_offset) &&
      abs(col_offset) > processed_cols)
    scan_col_mbmi(cm, xd, mi_row, mi_col, rf, col_offset, ref_mv_stack,
                  refmv_count, ref_match_count,
#if USE_CUR_GM_REFMV
                  gm_mv_candidates,
#endif  // USE_CUR_GM_REFMV
                  max_col_offset, &processed_cols);

#if CONFIG_OPT_REF_MV
  switch (nearest_match[ref_frame])
#else
  switch (nearest_refmv_count[ref_frame])
#endif
  {
    case 0: mode_context[ref_frame] |= 0;
#if CONFIG_OPT_REF_MV
      if (ref_match_count[ref_frame] >= 1) mode_context[ref_frame] |= 1;
      if (ref_match_count[ref_frame] == 1)
        mode_context[ref_frame] |= (1 << REFMV_OFFSET);
      else if (ref_match_count[ref_frame] >= 2)
        mode_context[ref_frame] |= (2 << REFMV_OFFSET);
#else
      if (refmv_count[ref_frame] >= 1) mode_context[ref_frame] |= 1;
      if (refmv_count[ref_frame] == 1)
        mode_context[ref_frame] |= (1 << REFMV_OFFSET);
      else if (refmv_count[ref_frame] >= 2)
        mode_context[ref_frame] |= (2 << REFMV_OFFSET);
#endif
      break;
    case 1: mode_context[ref_frame] |= (newmv_count > 0) ? 2 : 3;
#if CONFIG_OPT_REF_MV
      if (ref_match_count[ref_frame] == 1)
        mode_context[ref_frame] |= (3 << REFMV_OFFSET);
      else if (ref_match_count[ref_frame] >= 2)
        mode_context[ref_frame] |= (4 << REFMV_OFFSET);
#else
      if (refmv_count[ref_frame] == 1)
        mode_context[ref_frame] |= (3 << REFMV_OFFSET);
      else if (refmv_count[ref_frame] >= 2)
        mode_context[ref_frame] |= (4 << REFMV_OFFSET);
#endif
      break;

    case 2:
    default:
      if (newmv_count >= 2)
        mode_context[ref_frame] |= 4;
      else if (newmv_count == 1)
        mode_context[ref_frame] |= 5;
      else
        mode_context[ref_frame] |= 6;

      mode_context[ref_frame] |= (5 << REFMV_OFFSET);
      break;
  }

  // Rank the likelihood and assign nearest and near mvs.
  int len = nearest_refmv_count[ref_frame];
  while (len > 0) {
    int nr_len = 0;
    for (int idx = 1; idx < len; ++idx) {
      if (ref_mv_stack[ref_frame][idx - 1].weight <
          ref_mv_stack[ref_frame][idx].weight) {
        CANDIDATE_MV tmp_mv = ref_mv_stack[ref_frame][idx - 1];
        ref_mv_stack[ref_frame][idx - 1] = ref_mv_stack[ref_frame][idx];
        ref_mv_stack[ref_frame][idx] = tmp_mv;
        nr_len = idx;
      }
    }
    len = nr_len;
  }

  len = refmv_count[ref_frame];
  while (len > nearest_refmv_count[ref_frame]) {
    int nr_len = nearest_refmv_count[ref_frame];
    for (int idx = nearest_refmv_count[ref_frame] + 1; idx < len; ++idx) {
      if (ref_mv_stack[ref_frame][idx - 1].weight <
          ref_mv_stack[ref_frame][idx].weight) {
        CANDIDATE_MV tmp_mv = ref_mv_stack[ref_frame][idx - 1];
        ref_mv_stack[ref_frame][idx - 1] = ref_mv_stack[ref_frame][idx];
        ref_mv_stack[ref_frame][idx] = tmp_mv;
        nr_len = idx;
      }
    }
    len = nr_len;
  }

  if (rf[1] > NONE_FRAME) {
    for (int idx = 0; idx < refmv_count[ref_frame]; ++idx) {
      clamp_mv_ref(&ref_mv_stack[ref_frame][idx].this_mv.as_mv,
                   xd->n8_w << MI_SIZE_LOG2, xd->n8_h << MI_SIZE_LOG2, xd);
      clamp_mv_ref(&ref_mv_stack[ref_frame][idx].comp_mv.as_mv,
                   xd->n8_w << MI_SIZE_LOG2, xd->n8_h << MI_SIZE_LOG2, xd);
    }
  } else {
#if CONFIG_OPT_REF_MV
    // Handle single reference frame extension
    int mi_width = AOMMIN(mi_size_wide[BLOCK_64X64], xd->n8_w);
    mi_width = AOMMIN(mi_width, cm->mi_cols - mi_col);
    int mi_height = AOMMIN(mi_size_high[BLOCK_64X64], xd->n8_h);
    mi_height = AOMMIN(mi_height, cm->mi_rows - mi_row);
    int mi_size = AOMMIN(mi_width, mi_height);

    for (int idx = 0; abs(max_row_offset) >= 1 && idx < mi_size &&
                      refmv_count[ref_frame] < MAX_MV_REF_CANDIDATES;) {
      const MODE_INFO *const candidate_mi = xd->mi[-xd->mi_stride + idx];
      const MB_MODE_INFO *const candidate = &candidate_mi->mbmi;
      const int candidate_bsize = candidate->sb_type;

      // TODO(jingning): Refactor the following code.
      for (int rf_idx = 0; rf_idx < 2; ++rf_idx) {
        if (candidate->ref_frame[rf_idx] > INTRA_FRAME) {
          int_mv this_mv = candidate->mv[rf_idx];
          if (cm->ref_frame_sign_bias[candidate->ref_frame[rf_idx]] !=
              cm->ref_frame_sign_bias[ref_frame]) {
            this_mv.as_mv.row = -this_mv.as_mv.row;
            this_mv.as_mv.col = -this_mv.as_mv.col;
          }
          int stack_idx;
          for (stack_idx = 0; stack_idx < refmv_count[ref_frame]; ++stack_idx) {
            int_mv stack_mv = ref_mv_stack[ref_frame][stack_idx].this_mv;
            if (this_mv.as_int == stack_mv.as_int) break;
          }

          if (stack_idx == refmv_count[ref_frame]) {
            ref_mv_stack[ref_frame][stack_idx].this_mv = this_mv;

            // TODO(jingning): Set an arbitrary small number here. The weight
            // doesn't matter as long as it is properly initialized.
            ref_mv_stack[ref_frame][stack_idx].weight = 2;
            ++refmv_count[ref_frame];
          }
        }
      }
      idx += mi_size_wide[candidate_bsize];
    }

    for (int idx = 0; abs(max_col_offset) >= 1 && idx < mi_size &&
                      refmv_count[ref_frame] < MAX_MV_REF_CANDIDATES;) {
      const MODE_INFO *const candidate_mi = xd->mi[idx * xd->mi_stride - 1];
      const MB_MODE_INFO *const candidate = &candidate_mi->mbmi;
      const int candidate_bsize = candidate->sb_type;

      // TODO(jingning): Refactor the following code.
      for (int rf_idx = 0; rf_idx < 2; ++rf_idx) {
        if (candidate->ref_frame[rf_idx] > INTRA_FRAME) {
          int_mv this_mv = candidate->mv[rf_idx];
          if (cm->ref_frame_sign_bias[candidate->ref_frame[rf_idx]] !=
              cm->ref_frame_sign_bias[ref_frame]) {
            this_mv.as_mv.row = -this_mv.as_mv.row;
            this_mv.as_mv.col = -this_mv.as_mv.col;
          }
          int stack_idx;
          for (stack_idx = 0; stack_idx < refmv_count[ref_frame]; ++stack_idx) {
            int_mv stack_mv = ref_mv_stack[ref_frame][stack_idx].this_mv;
            if (this_mv.as_int == stack_mv.as_int) break;
          }

          if (stack_idx == refmv_count[ref_frame]) {
            ref_mv_stack[ref_frame][stack_idx].this_mv = this_mv;

            // TODO(jingning): Set an arbitrary small number here. The weight
            // doesn't matter as long as it is properly initialized.
            ref_mv_stack[ref_frame][stack_idx].weight = 2;
            ++refmv_count[ref_frame];
          }
        }
      }
      idx += mi_size_high[candidate_bsize];
    }

    for (int idx = refmv_count[ref_frame]; idx < MAX_MV_REF_CANDIDATES; ++idx)
      mv_ref_list[rf[0]][idx].as_int = gm_mv_candidates[0].as_int;
#endif
    for (int idx = 0;
         idx < AOMMIN(MAX_MV_REF_CANDIDATES, refmv_count[ref_frame]); ++idx) {
      mv_ref_list[rf[0]][idx].as_int =
          ref_mv_stack[ref_frame][idx].this_mv.as_int;
      clamp_mv_ref(&mv_ref_list[rf[0]][idx].as_mv, xd->n8_w << MI_SIZE_LOG2,
                   xd->n8_h << MI_SIZE_LOG2, xd);
    }
  }
  (void)nearest_match;
}

// This function searches the neighbourhood of a given MB/SB
// to try and find candidate reference vectors.
static void find_mv_refs_idx(const AV1_COMMON *cm, const MACROBLOCKD *xd,
                             MODE_INFO *mi, MV_REFERENCE_FRAME ref_frame,
                             int_mv *mv_ref_list, int mi_row, int mi_col,
                             find_mv_refs_sync sync, void *const data,
                             int16_t *mode_context, int_mv zeromv,
                             uint8_t refmv_count) {
  const int *ref_sign_bias = cm->ref_frame_sign_bias;
  const int sb_mi_size = mi_size_wide[cm->seq_params.sb_size];
  int i;
  int context_counter = 0;

#if CONFIG_MFMV
  (void)sync;
  (void)data;
#else
#if CONFIG_TMV
  int tmi_row = mi_row & 0xfffe;
  int tmi_col = mi_col & 0xfffe;
  POSITION mi_pos = { 0, 0 };
  int inside = is_inside(&xd->tile, tmi_col, tmi_row, cm->mi_rows, cm, &mi_pos);
  const MV_REF *const prev_frame_mvs =
      cm->use_prev_frame_mvs && inside
          ? cm->prev_frame->mvs + (tmi_row >> 1) * ((cm->mi_cols + 1) >> 1) +
                (tmi_col >> 1)
          : NULL;
#else
  const TileInfo *const tile_ = &xd->tile;
  int mi_row_end = tile_->mi_row_end;
  int mi_col_end = tile_->mi_col_end;
  const MV_REF *const prev_frame_mvs =
      cm->use_prev_frame_mvs
          ? cm->prev_frame->mvs +
                AOMMIN(((mi_row >> 1) << 1) + 1 + (((xd->n8_h - 1) >> 1) << 1),
                       mi_row_end - 1) *
                    cm->mi_cols +
                AOMMIN(((mi_col >> 1) << 1) + 1 + (((xd->n8_w - 1) >> 1) << 1),
                       mi_col_end - 1)
          : NULL;
#endif  // CONFIG_TMV
#endif  // CONFIG_MFMV

#if CONFIG_INTRABC
  assert(IMPLIES(ref_frame == INTRA_FRAME, cm->use_prev_frame_mvs == 0));
#endif
  const TileInfo *const tile = &xd->tile;
  const BLOCK_SIZE bsize = mi->mbmi.sb_type;
  const int bw = block_size_wide[AOMMAX(bsize, BLOCK_8X8)];
  const int bh = block_size_high[AOMMAX(bsize, BLOCK_8X8)];
  POSITION mv_ref_search[MVREF_NEIGHBOURS];
  const int num_8x8_blocks_wide = num_8x8_blocks_wide_lookup[bsize];
  const int num_8x8_blocks_high = num_8x8_blocks_high_lookup[bsize];
  mv_ref_search[0].row = num_8x8_blocks_high - 1;
  mv_ref_search[0].col = -1;
  mv_ref_search[1].row = -1;
  mv_ref_search[1].col = num_8x8_blocks_wide - 1;
  mv_ref_search[2].row = -1;
  mv_ref_search[2].col = (num_8x8_blocks_wide - 1) >> 1;
  mv_ref_search[3].row = (num_8x8_blocks_high - 1) >> 1;
  mv_ref_search[3].col = -1;
  mv_ref_search[4].row = -1;
  mv_ref_search[4].col = -1;
#if CONFIG_EXT_PARTITION_TYPES
  if (num_8x8_blocks_wide == num_8x8_blocks_high) {
    mv_ref_search[5].row = -1;
    mv_ref_search[5].col = 0;
    mv_ref_search[6].row = 0;
    mv_ref_search[6].col = -1;
  } else {
    mv_ref_search[5].row = -1;
    mv_ref_search[5].col = num_8x8_blocks_wide;
    mv_ref_search[6].row = num_8x8_blocks_high;
    mv_ref_search[6].col = -1;
  }
#else
  mv_ref_search[5].row = -1;
  mv_ref_search[5].col = num_8x8_blocks_wide;
  mv_ref_search[6].row = num_8x8_blocks_high;
  mv_ref_search[6].col = -1;
#endif  // CONFIG_EXT_PARTITION_TYPES
  mv_ref_search[7].row = -1;
  mv_ref_search[7].col = -3;
  mv_ref_search[8].row = num_8x8_blocks_high - 1;
  mv_ref_search[8].col = -3;

  for (i = 0; i < MVREF_NEIGHBOURS; ++i) {
    mv_ref_search[i].row *= 2;
    mv_ref_search[i].col *= 2;
  }

  // The nearest 2 blocks are treated differently
  // if the size < 8x8 we get the mv from the bmi substructure,
  // and we also need to keep a mode count.
  for (i = 0; i < 2; ++i) {
    const POSITION *const mv_ref = &mv_ref_search[i];
    if (is_inside(tile, mi_col, mi_row, cm->mi_rows, cm, mv_ref)) {
      const MODE_INFO *const candidate_mi =
          xd->mi[mv_ref->col + mv_ref->row * xd->mi_stride];
      const MB_MODE_INFO *const candidate = &candidate_mi->mbmi;
#if CONFIG_INTRABC
      if (ref_frame == INTRA_FRAME && !is_intrabc_block(candidate)) continue;
#endif  // CONFIG_INTRABC
      // Keep counts for entropy encoding.
      context_counter += mode_2_counter[candidate->mode];
    }
  }

  if (refmv_count >= MAX_MV_REF_CANDIDATES) goto Done;

// TODO(hkuang): Remove this sync after fixing pthread_cond_broadcast
// on windows platform. The sync here is unncessary if use_perv_frame_mvs
// is 0. But after removing it, there will be hang in the unit test on windows
// due to several threads waiting for a thread's signal.
#if defined(_WIN32) && !HAVE_PTHREAD_H
  if (cm->frame_parallel_decode && sync != NULL) {
    sync(data, mi_row);
  }
#endif

#if !CONFIG_MFMV
  // Check the last frame's mode and mv info.
  if (cm->use_prev_frame_mvs) {
    // Synchronize here for frame parallel decode if sync function is provided.
    if (cm->frame_parallel_decode && sync != NULL) {
      sync(data, mi_row);
    }

    if (prev_frame_mvs->ref_frame[0] == ref_frame) {
      ADD_MV_REF_LIST(prev_frame_mvs->mv[0], refmv_count, mv_ref_list, bw, bh,
                      xd, Done);
    } else if (prev_frame_mvs->ref_frame[1] == ref_frame) {
      ADD_MV_REF_LIST(prev_frame_mvs->mv[1], refmv_count, mv_ref_list, bw, bh,
                      xd, Done);
    }
  }
#endif  // !CONFIG_MFMV

// Since we couldn't find 2 mvs from the same reference frame
// go back through the neighbors and find motion vectors from
// different reference frames.
#if CONFIG_INTRABC
  if (ref_frame != INTRA_FRAME) {
#endif  // CONFIG_INTRABC
    for (i = 0; i < MVREF_NEIGHBOURS; ++i) {
      const POSITION *mv_ref = &mv_ref_search[i];
      if (is_inside(tile, mi_col, mi_row, cm->mi_rows, cm, mv_ref)) {
        const MB_MODE_INFO *const candidate =
            !xd->mi[mv_ref->col + mv_ref->row * xd->mi_stride]
                ? NULL
                : &xd->mi[mv_ref->col + mv_ref->row * xd->mi_stride]->mbmi;
        if (candidate == NULL) continue;
        if ((mi_row & (sb_mi_size - 1)) + mv_ref->row >= sb_mi_size ||
            (mi_col & (sb_mi_size - 1)) + mv_ref->col >= sb_mi_size)
          continue;

        // If the candidate is INTRA we don't want to consider its mv.
        IF_DIFF_REF_FRAME_ADD_MV(candidate, ref_frame, ref_sign_bias,
                                 refmv_count, mv_ref_list, bw, bh, xd, Done);
      }
    }
#if CONFIG_INTRABC
  }
#endif  // CONFIG_INTRABC

#if !CONFIG_MFMV
  // Since we still don't have a candidate we'll try the last frame.
  if (cm->use_prev_frame_mvs) {
    if (prev_frame_mvs->ref_frame[0] != ref_frame &&
        prev_frame_mvs->ref_frame[0] > INTRA_FRAME) {
      int_mv mv = prev_frame_mvs->mv[0];
      if (ref_sign_bias[prev_frame_mvs->ref_frame[0]] !=
          ref_sign_bias[ref_frame]) {
        mv.as_mv.row *= -1;
        mv.as_mv.col *= -1;
      }
      ADD_MV_REF_LIST(mv, refmv_count, mv_ref_list, bw, bh, xd, Done);
    }

    if (prev_frame_mvs->ref_frame[1] > INTRA_FRAME &&
        prev_frame_mvs->ref_frame[1] != ref_frame) {
      int_mv mv = prev_frame_mvs->mv[1];
      if (ref_sign_bias[prev_frame_mvs->ref_frame[1]] !=
          ref_sign_bias[ref_frame]) {
        mv.as_mv.row *= -1;
        mv.as_mv.col *= -1;
      }
      ADD_MV_REF_LIST(mv, refmv_count, mv_ref_list, bw, bh, xd, Done);
    }
  }
#endif  // !CONFIG_MFMV

Done:
  if (mode_context)
    mode_context[ref_frame] = counter_to_context[context_counter];
  for (i = refmv_count; i < MAX_MV_REF_CANDIDATES; ++i)
    mv_ref_list[i].as_int = zeromv.as_int;
}

void av1_find_mv_refs(const AV1_COMMON *cm, const MACROBLOCKD *xd,
                      MODE_INFO *mi, MV_REFERENCE_FRAME ref_frame,
                      uint8_t ref_mv_count[MODE_CTX_REF_FRAMES],
                      CANDIDATE_MV ref_mv_stack[][MAX_REF_MV_STACK_SIZE],
                      int16_t *compound_mode_context,
                      int_mv mv_ref_list[][MAX_MV_REF_CANDIDATES], int mi_row,
                      int mi_col, find_mv_refs_sync sync, void *const data,
                      int16_t *mode_context, int compound_search) {
  int_mv zeromv[2];
  BLOCK_SIZE bsize = mi->mbmi.sb_type;
  MV_REFERENCE_FRAME rf[2];
  av1_set_ref_frame(rf, ref_frame);
  if (!CONFIG_INTRABC || ref_frame != INTRA_FRAME) {
    zeromv[0].as_int =
        gm_get_motion_vector(&cm->global_motion[rf[0]],
                             cm->allow_high_precision_mv, bsize, mi_col, mi_row
#if CONFIG_AMVR
                             ,
                             cm->cur_frame_force_integer_mv
#endif
                             )
            .as_int;
    zeromv[1].as_int = (rf[1] != NONE_FRAME)
                           ? gm_get_motion_vector(&cm->global_motion[rf[1]],
                                                  cm->allow_high_precision_mv,
                                                  bsize, mi_col, mi_row
#if CONFIG_AMVR
                                                  ,
                                                  cm->cur_frame_force_integer_mv
#endif
                                                  )
                                 .as_int
                           : 0;
  } else {
    zeromv[0].as_int = zeromv[1].as_int = 0;
  }

  setup_ref_mv_list(cm, xd, ref_frame, ref_mv_count, ref_mv_stack, mv_ref_list,
#if USE_CUR_GM_REFMV
                    zeromv,
#endif  // USE_CUR_GM_REFMV
                    mi_row, mi_col, mode_context, compound_search);

#if !CONFIG_OPT_REF_MV
  if (ref_frame <= ALTREF_FRAME)
    find_mv_refs_idx(cm, xd, mi, ref_frame, mv_ref_list[rf[0]], mi_row, mi_col,
                     sync, data, compound_mode_context, zeromv[0],
                     ref_mv_count[ref_frame]);
#else
  (void)compound_mode_context;
  (void)data;
  (void)sync;
#endif
}

void av1_find_best_ref_mvs(int allow_hp, int_mv *mvlist, int_mv *nearest_mv,
                           int_mv *near_mv
#if CONFIG_AMVR
                           ,
                           int is_integer
#endif
) {
  int i;
  // Make sure all the candidates are properly clamped etc
  for (i = 0; i < MAX_MV_REF_CANDIDATES; ++i) {
#if CONFIG_AMVR
    lower_mv_precision(&mvlist[i].as_mv, allow_hp, is_integer);
#else
    lower_mv_precision(&mvlist[i].as_mv, allow_hp);
#endif
  }
  *nearest_mv = mvlist[0];
  *near_mv = mvlist[1];
}

void av1_setup_frame_buf_refs(AV1_COMMON *cm) {
  cm->cur_frame->cur_frame_offset = cm->frame_offset;
  int alt_buf_idx = cm->frame_refs[ALTREF_FRAME - LAST_FRAME].idx;
  int lst_buf_idx = cm->frame_refs[LAST_FRAME - LAST_FRAME].idx;
  int gld_buf_idx = cm->frame_refs[GOLDEN_FRAME - LAST_FRAME].idx;

  int lst2_buf_idx = cm->frame_refs[LAST2_FRAME - LAST_FRAME].idx;
  int lst3_buf_idx = cm->frame_refs[LAST3_FRAME - LAST_FRAME].idx;
  int bwd_buf_idx = cm->frame_refs[BWDREF_FRAME - LAST_FRAME].idx;
  int alt2_buf_idx = cm->frame_refs[ALTREF2_FRAME - LAST_FRAME].idx;

  if (alt_buf_idx >= 0)
    cm->cur_frame->alt_frame_offset =
        cm->buffer_pool->frame_bufs[alt_buf_idx].cur_frame_offset;

  if (lst_buf_idx >= 0)
    cm->cur_frame->lst_frame_offset =
        cm->buffer_pool->frame_bufs[lst_buf_idx].cur_frame_offset;

  if (gld_buf_idx >= 0)
    cm->cur_frame->gld_frame_offset =
        cm->buffer_pool->frame_bufs[gld_buf_idx].cur_frame_offset;

  if (lst2_buf_idx >= 0)
    cm->cur_frame->lst2_frame_offset =
        cm->buffer_pool->frame_bufs[lst2_buf_idx].cur_frame_offset;

  if (lst3_buf_idx >= 0)
    cm->cur_frame->lst3_frame_offset =
        cm->buffer_pool->frame_bufs[lst3_buf_idx].cur_frame_offset;

  if (bwd_buf_idx >= 0)
    cm->cur_frame->bwd_frame_offset =
        cm->buffer_pool->frame_bufs[bwd_buf_idx].cur_frame_offset;

  if (alt2_buf_idx >= 0)
    cm->cur_frame->alt2_frame_offset =
        cm->buffer_pool->frame_bufs[alt2_buf_idx].cur_frame_offset;
}

void av1_setup_frame_sign_bias(AV1_COMMON *cm) {
  MV_REFERENCE_FRAME ref_frame;
  for (ref_frame = LAST_FRAME; ref_frame <= ALTREF_FRAME; ++ref_frame) {
    const int buf_idx = cm->frame_refs[ref_frame - LAST_FRAME].idx;
    if (buf_idx != INVALID_IDX) {
      const int ref_frame_offset =
          cm->buffer_pool->frame_bufs[buf_idx].cur_frame_offset;
      cm->ref_frame_sign_bias[ref_frame] =
          (ref_frame_offset <= (int)cm->frame_offset) ? 0 : 1;
    } else {
      cm->ref_frame_sign_bias[ref_frame] = 0;
    }
  }
}

#if CONFIG_MFMV
#define MAX_OFFSET_WIDTH 64
#define MAX_OFFSET_HEIGHT 0

static int get_block_position(AV1_COMMON *cm, int *mi_r, int *mi_c, int blk_row,
                              int blk_col, MV mv, int sign_bias) {
  const int base_blk_row = (blk_row >> 3) << 3;
  const int base_blk_col = (blk_col >> 3) << 3;

  const int row_offset = (mv.row >= 0) ? (mv.row >> (4 + MI_SIZE_LOG2))
                                       : -((-mv.row) >> (4 + MI_SIZE_LOG2));

  const int col_offset = (mv.col >= 0) ? (mv.col >> (4 + MI_SIZE_LOG2))
                                       : -((-mv.col) >> (4 + MI_SIZE_LOG2));

  int row = (sign_bias == 1) ? blk_row - row_offset : blk_row + row_offset;
  int col = (sign_bias == 1) ? blk_col - col_offset : blk_col + col_offset;

  if (row < 0 || row >= (cm->mi_rows >> 1) || col < 0 ||
      col >= (cm->mi_cols >> 1))
    return 0;

  if (row <= base_blk_row - (MAX_OFFSET_HEIGHT >> 3) ||
      row >= base_blk_row + 8 + (MAX_OFFSET_HEIGHT >> 3) ||
      col <= base_blk_col - (MAX_OFFSET_WIDTH >> 3) ||
      col >= base_blk_col + 8 + (MAX_OFFSET_WIDTH >> 3))
    return 0;

  *mi_r = row;
  *mi_c = col;

  return 1;
}

static int motion_field_projection(AV1_COMMON *cm, MV_REFERENCE_FRAME ref_frame,
                                   int ref_stamp, int dir) {
  TPL_MV_REF *tpl_mvs_base = cm->tpl_mvs;
  int cur_rf_index[TOTAL_REFS_PER_FRAME] = { 0 };
  int ref_rf_idx[TOTAL_REFS_PER_FRAME] = { 0 };
  int cur_offset[TOTAL_REFS_PER_FRAME] = { 0 };
  int ref_offset[TOTAL_REFS_PER_FRAME] = { 0 };

  (void)dir;

  int ref_frame_idx = cm->frame_refs[FWD_RF_OFFSET(ref_frame)].idx;
  if (ref_frame_idx < 0) return 0;

  if (cm->buffer_pool->frame_bufs[ref_frame_idx].mi_rows != cm->mi_rows ||
      cm->buffer_pool->frame_bufs[ref_frame_idx].mi_cols != cm->mi_cols)
    return 0;

  int ref_frame_index =
      cm->buffer_pool->frame_bufs[ref_frame_idx].cur_frame_offset;
  int cur_frame_index = cm->cur_frame->cur_frame_offset;
  int ref_to_cur = ref_frame_index - cur_frame_index;

  ref_rf_idx[LAST_FRAME] =
      cm->buffer_pool->frame_bufs[ref_frame_idx].lst_frame_offset;
  ref_rf_idx[GOLDEN_FRAME] =
      cm->buffer_pool->frame_bufs[ref_frame_idx].gld_frame_offset;
  ref_rf_idx[LAST2_FRAME] =
      cm->buffer_pool->frame_bufs[ref_frame_idx].lst2_frame_offset;
  ref_rf_idx[LAST3_FRAME] =
      cm->buffer_pool->frame_bufs[ref_frame_idx].lst3_frame_offset;
  ref_rf_idx[BWDREF_FRAME] =
      cm->buffer_pool->frame_bufs[ref_frame_idx].bwd_frame_offset;
  ref_rf_idx[ALTREF2_FRAME] =
      cm->buffer_pool->frame_bufs[ref_frame_idx].alt2_frame_offset;
  ref_rf_idx[ALTREF_FRAME] =
      cm->buffer_pool->frame_bufs[ref_frame_idx].alt_frame_offset;

  for (MV_REFERENCE_FRAME rf = LAST_FRAME; rf <= INTER_REFS_PER_FRAME; ++rf) {
    int buf_idx = cm->frame_refs[FWD_RF_OFFSET(rf)].idx;
    if (buf_idx >= 0)
      cur_rf_index[rf] = cm->buffer_pool->frame_bufs[buf_idx].cur_frame_offset;
    cur_offset[rf] = cur_frame_index - cur_rf_index[rf];
    ref_offset[rf] = ref_frame_index - ref_rf_idx[rf];
  }

  if (dir == 1) {
    ref_to_cur = -ref_to_cur;
    for (MV_REFERENCE_FRAME rf = LAST_FRAME; rf <= INTER_REFS_PER_FRAME; ++rf) {
      cur_offset[rf] = -cur_offset[rf];
      ref_offset[rf] = -ref_offset[rf];
    }
  }

  if (dir == 2) ref_to_cur = -ref_to_cur;

  MV_REF *mv_ref_base = cm->buffer_pool->frame_bufs[ref_frame_idx].mvs;
  const int mvs_rows = (cm->mi_rows + 1) >> 1;
  const int mvs_cols = (cm->mi_cols + 1) >> 1;

  for (int blk_row = 0; blk_row < mvs_rows; ++blk_row) {
    for (int blk_col = 0; blk_col < mvs_cols; ++blk_col) {
      MV_REF *mv_ref = &mv_ref_base[blk_row * mvs_cols + blk_col];
      MV fwd_mv = mv_ref->mv[dir & 0x01].as_mv;

      if (mv_ref->ref_frame[dir & 0x01] > INTRA_FRAME) {
        int_mv this_mv;
        int mi_r, mi_c;
        const int ref_frame_offset = ref_offset[mv_ref->ref_frame[dir & 0x01]];

        get_mv_projection(&this_mv.as_mv, fwd_mv, ref_to_cur, ref_frame_offset);
        int pos_valid = get_block_position(cm, &mi_r, &mi_c, blk_row, blk_col,
                                           this_mv.as_mv, dir >> 1);
        if (pos_valid) {
          int mi_offset = mi_r * (cm->mi_stride >> 1) + mi_c;

          tpl_mvs_base[mi_offset].mfmv0[ref_stamp].as_mv.row =
              (dir == 1) ? -fwd_mv.row : fwd_mv.row;
          tpl_mvs_base[mi_offset].mfmv0[ref_stamp].as_mv.col =
              (dir == 1) ? -fwd_mv.col : fwd_mv.col;
          tpl_mvs_base[mi_offset].ref_frame_offset[ref_stamp] =
              ref_frame_offset;
        }
      }
    }
  }

  return 1;
}

void av1_setup_motion_field(AV1_COMMON *cm) {
  int cur_frame_index = cm->cur_frame->cur_frame_offset;
  int alt_frame_index = 0, gld_frame_index = 0;
  int bwd_frame_index = 0, alt2_frame_index = 0;

  TPL_MV_REF *tpl_mvs_base = cm->tpl_mvs;
  int size = ((cm->mi_rows + MAX_MIB_SIZE) >> 1) * (cm->mi_stride >> 1);
  for (int idx = 0; idx < size; ++idx) {
    for (int i = 0; i < MFMV_STACK_SIZE; ++i) {
      tpl_mvs_base[idx].mfmv0[i].as_int = INVALID_MV;
      tpl_mvs_base[idx].ref_frame_offset[i] = 0;
    }
  }

  int gld_buf_idx = cm->frame_refs[GOLDEN_FRAME - LAST_FRAME].idx;
  int alt_buf_idx = cm->frame_refs[ALTREF_FRAME - LAST_FRAME].idx;
  int lst_buf_idx = cm->frame_refs[LAST_FRAME - LAST_FRAME].idx;
  int bwd_buf_idx = cm->frame_refs[BWDREF_FRAME - LAST_FRAME].idx;
  int alt2_buf_idx = cm->frame_refs[ALTREF2_FRAME - LAST_FRAME].idx;

  if (alt_buf_idx >= 0)
    alt_frame_index = cm->buffer_pool->frame_bufs[alt_buf_idx].cur_frame_offset;

  if (gld_buf_idx >= 0)
    gld_frame_index = cm->buffer_pool->frame_bufs[gld_buf_idx].cur_frame_offset;

  if (bwd_buf_idx >= 0)
    bwd_frame_index = cm->buffer_pool->frame_bufs[bwd_buf_idx].cur_frame_offset;

  if (alt2_buf_idx >= 0)
    alt2_frame_index =
        cm->buffer_pool->frame_bufs[alt2_buf_idx].cur_frame_offset;

  memset(cm->ref_frame_side, 0, sizeof(cm->ref_frame_side));
  for (int ref_frame = LAST_FRAME; ref_frame <= INTER_REFS_PER_FRAME;
       ++ref_frame) {
    int buf_idx = cm->frame_refs[ref_frame - LAST_FRAME].idx;
    int frame_index = -1;
    if (buf_idx >= 0)
      frame_index = cm->buffer_pool->frame_bufs[buf_idx].cur_frame_offset;
    if (frame_index > cur_frame_index)
      cm->ref_frame_side[ref_frame] = 1;
    else if (frame_index == cur_frame_index)
      cm->ref_frame_side[ref_frame] = -1;
  }

  int ref_stamp = MFMV_STACK_SIZE - 1;

  if (lst_buf_idx >= 0) {
    const int alt_frame_idx =
        cm->buffer_pool->frame_bufs[lst_buf_idx].alt_frame_offset;

    const int is_lst_overlay = (alt_frame_idx == gld_frame_index);
    if (!is_lst_overlay) motion_field_projection(cm, LAST_FRAME, ref_stamp, 1);

    --ref_stamp;
  }

  if (bwd_frame_index > cur_frame_index) {
    if (motion_field_projection(cm, BWDREF_FRAME, ref_stamp, 0)) --ref_stamp;
  }

  if (alt2_frame_index > cur_frame_index) {
    if (motion_field_projection(cm, ALTREF2_FRAME, ref_stamp, 0)) --ref_stamp;
  }

  if (alt_frame_index > cur_frame_index && ref_stamp >= 0)
    if (motion_field_projection(cm, ALTREF_FRAME, ref_stamp, 0)) --ref_stamp;
}
#endif  // CONFIG_MFMV

#if CONFIG_EXT_WARPED_MOTION
static INLINE void record_samples(MB_MODE_INFO *mbmi, int *pts, int *pts_inref,
                                  int row_offset, int sign_r, int col_offset,
                                  int sign_c) {
  int bw = block_size_wide[mbmi->sb_type];
  int bh = block_size_high[mbmi->sb_type];
  int x = col_offset * MI_SIZE + sign_c * AOMMAX(bw, MI_SIZE) / 2 - 1;
  int y = row_offset * MI_SIZE + sign_r * AOMMAX(bh, MI_SIZE) / 2 - 1;

  pts[0] = (x * 8);
  pts[1] = (y * 8);
  pts_inref[0] = (x * 8) + mbmi->mv[0].as_mv.col;
  pts_inref[1] = (y * 8) + mbmi->mv[0].as_mv.row;
}

// Select samples according to the motion vector difference.
int selectSamples(MV *mv, int *pts, int *pts_inref, int len, BLOCK_SIZE bsize) {
  const int bw = block_size_wide[bsize];
  const int bh = block_size_high[bsize];
  const int thresh = clamp(AOMMAX(bw, bh), 16, 112);
  int pts_mvd[SAMPLES_ARRAY_SIZE] = { 0 };
  int i, j, k, l = len;
  int ret = 0;
  assert(len <= LEAST_SQUARES_SAMPLES_MAX);

  // Obtain the motion vector difference.
  for (i = 0; i < len; ++i) {
    pts_mvd[i] = abs(pts_inref[2 * i] - pts[2 * i] - mv->col) +
                 abs(pts_inref[2 * i + 1] - pts[2 * i + 1] - mv->row);

    if (pts_mvd[i] > thresh)
      pts_mvd[i] = -1;
    else
      ret++;
  }

  // Keep at least 1 sample.
  if (!ret) return 1;

  i = 0;
  j = l - 1;
  for (k = 0; k < l - ret; k++) {
    while (pts_mvd[i] != -1) i++;
    while (pts_mvd[j] == -1) j--;
    assert(i != j);
    if (i > j) break;

    // Replace the discarded samples;
    pts_mvd[i] = pts_mvd[j];
    pts[2 * i] = pts[2 * j];
    pts[2 * i + 1] = pts[2 * j + 1];
    pts_inref[2 * i] = pts_inref[2 * j];
    pts_inref[2 * i + 1] = pts_inref[2 * j + 1];
    i++;
    j--;
  }

  return ret;
}

// Note: Samples returned are at 1/8-pel precision
// Sample are the neighbor block center point's coordinates relative to the
// left-top pixel of current block.
int findSamples(const AV1_COMMON *cm, MACROBLOCKD *xd, int mi_row, int mi_col,
                int *pts, int *pts_inref) {
  MB_MODE_INFO *const mbmi0 = &(xd->mi[0]->mbmi);
  int ref_frame = mbmi0->ref_frame[0];
  int up_available = xd->up_available;
  int left_available = xd->left_available;
  int i, mi_step = 1, np = 0;

  const TileInfo *const tile = &xd->tile;
  int do_tl = 1;
  int do_tr = 1;

  // scan the nearest above rows
  if (up_available) {
    int mi_row_offset = -1;
    MODE_INFO *mi = xd->mi[mi_row_offset * xd->mi_stride];
    MB_MODE_INFO *mbmi = &mi->mbmi;
    uint8_t n8_w = mi_size_wide[mbmi->sb_type];

    if (xd->n8_w <= n8_w) {
      // Handle "current block width <= above block width" case.
      int col_offset = -mi_col % n8_w;

      if (col_offset < 0) do_tl = 0;
      if (col_offset + n8_w > xd->n8_w) do_tr = 0;

      if (mbmi->ref_frame[0] == ref_frame && mbmi->ref_frame[1] == NONE_FRAME) {
        record_samples(mbmi, pts, pts_inref, 0, -1, col_offset, 1);
        pts += 2;
        pts_inref += 2;
        np++;
        if (np >= LEAST_SQUARES_SAMPLES_MAX) return LEAST_SQUARES_SAMPLES_MAX;
      }
    } else {
      // Handle "current block width > above block width" case.
      for (i = 0; i < AOMMIN(xd->n8_w, cm->mi_cols - mi_col); i += mi_step) {
        int mi_col_offset = i;
        mi = xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride];
        mbmi = &mi->mbmi;
        n8_w = mi_size_wide[mbmi->sb_type];
        mi_step = AOMMIN(xd->n8_w, n8_w);

        if (mbmi->ref_frame[0] == ref_frame &&
            mbmi->ref_frame[1] == NONE_FRAME) {
          record_samples(mbmi, pts, pts_inref, 0, -1, i, 1);
          pts += 2;
          pts_inref += 2;
          np++;
          if (np >= LEAST_SQUARES_SAMPLES_MAX) return LEAST_SQUARES_SAMPLES_MAX;
        }
      }
    }
  }
  assert(np <= LEAST_SQUARES_SAMPLES_MAX);

  // scan the nearest left columns
  if (left_available) {
    int mi_col_offset = -1;

    MODE_INFO *mi = xd->mi[mi_col_offset];
    MB_MODE_INFO *mbmi = &mi->mbmi;
    uint8_t n8_h = mi_size_high[mbmi->sb_type];

    if (xd->n8_h <= n8_h) {
      // Handle "current block height <= above block height" case.
      int row_offset = -mi_row % n8_h;

      if (row_offset < 0) do_tl = 0;

      if (mbmi->ref_frame[0] == ref_frame && mbmi->ref_frame[1] == NONE_FRAME) {
        record_samples(mbmi, pts, pts_inref, row_offset, 1, 0, -1);
        pts += 2;
        pts_inref += 2;
        np++;
        if (np >= LEAST_SQUARES_SAMPLES_MAX) return LEAST_SQUARES_SAMPLES_MAX;
      }
    } else {
      // Handle "current block height > above block height" case.
      for (i = 0; i < AOMMIN(xd->n8_h, cm->mi_rows - mi_row); i += mi_step) {
        int mi_row_offset = i;
        mi = xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride];
        mbmi = &mi->mbmi;
        n8_h = mi_size_high[mbmi->sb_type];
        mi_step = AOMMIN(xd->n8_h, n8_h);

        if (mbmi->ref_frame[0] == ref_frame &&
            mbmi->ref_frame[1] == NONE_FRAME) {
          record_samples(mbmi, pts, pts_inref, i, 1, 0, -1);
          pts += 2;
          pts_inref += 2;
          np++;
          if (np >= LEAST_SQUARES_SAMPLES_MAX) return LEAST_SQUARES_SAMPLES_MAX;
        }
      }
    }
  }
  assert(np <= LEAST_SQUARES_SAMPLES_MAX);

  // Top-left block
  if (do_tl && left_available && up_available) {
    int mi_row_offset = -1;
    int mi_col_offset = -1;

    MODE_INFO *mi = xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride];
    MB_MODE_INFO *mbmi = &mi->mbmi;

    if (mbmi->ref_frame[0] == ref_frame && mbmi->ref_frame[1] == NONE_FRAME) {
      record_samples(mbmi, pts, pts_inref, 0, -1, 0, -1);
      pts += 2;
      pts_inref += 2;
      np++;
      if (np >= LEAST_SQUARES_SAMPLES_MAX) return LEAST_SQUARES_SAMPLES_MAX;
    }
  }
  assert(np <= LEAST_SQUARES_SAMPLES_MAX);

  // Top-right block
  if (do_tr &&
      has_top_right(cm, xd, mi_row, mi_col, AOMMAX(xd->n8_w, xd->n8_h))) {
    POSITION trb_pos = { -1, xd->n8_w };

    if (is_inside(tile, mi_col, mi_row, cm->mi_rows, cm, &trb_pos)) {
      int mi_row_offset = -1;
      int mi_col_offset = xd->n8_w;

      MODE_INFO *mi = xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride];
      MB_MODE_INFO *mbmi = &mi->mbmi;

      if (mbmi->ref_frame[0] == ref_frame && mbmi->ref_frame[1] == NONE_FRAME) {
        record_samples(mbmi, pts, pts_inref, 0, -1, xd->n8_w, 1);
        np++;
        if (np >= LEAST_SQUARES_SAMPLES_MAX) return LEAST_SQUARES_SAMPLES_MAX;
      }
    }
  }
  assert(np <= LEAST_SQUARES_SAMPLES_MAX);

  return np;
}
#else
void calc_projection_samples(MB_MODE_INFO *const mbmi, int x, int y,
                             int *pts_inref) {
  pts_inref[0] = (x * 8) + mbmi->mv[0].as_mv.col;
  pts_inref[1] = (y * 8) + mbmi->mv[0].as_mv.row;
}

// Note: Samples returned are at 1/8-pel precision
// Sample are the neighbor block center point's coordinates relative to the
// left-top pixel of current block.
int findSamples(const AV1_COMMON *cm, MACROBLOCKD *xd, int mi_row, int mi_col,
                int *pts, int *pts_inref) {
  MB_MODE_INFO *const mbmi0 = &(xd->mi[0]->mbmi);
  int ref_frame = mbmi0->ref_frame[0];
  int up_available = xd->up_available;
  int left_available = xd->left_available;
  int i, mi_step, np = 0;

  // scan the above row
  if (up_available) {
    for (i = 0; i < AOMMIN(xd->n8_w, cm->mi_cols - mi_col); i += mi_step) {
      int mi_row_offset = -1;
      int mi_col_offset = i;

      MODE_INFO *mi = xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride];
      MB_MODE_INFO *mbmi = &mi->mbmi;

      mi_step = AOMMIN(xd->n8_w, mi_size_wide[mbmi->sb_type]);

      if (mbmi->ref_frame[0] == ref_frame && mbmi->ref_frame[1] == NONE_FRAME) {
        int bw = block_size_wide[mbmi->sb_type];
        int bh = block_size_high[mbmi->sb_type];
        int x = i * MI_SIZE + AOMMAX(bw, MI_SIZE) / 2 - 1;
        int y = -AOMMAX(bh, MI_SIZE) / 2 - 1;

        pts[0] = (x * 8);
        pts[1] = (y * 8);
        calc_projection_samples(mbmi, x, y, pts_inref);
        pts += 2;
        pts_inref += 2;
        np++;
        if (np >= LEAST_SQUARES_SAMPLES_MAX) return LEAST_SQUARES_SAMPLES_MAX;
      }
    }
  }
  assert(2 * np <= SAMPLES_ARRAY_SIZE);

  // scan the left column
  if (left_available) {
    for (i = 0; i < AOMMIN(xd->n8_h, cm->mi_rows - mi_row); i += mi_step) {
      int mi_row_offset = i;
      int mi_col_offset = -1;

      MODE_INFO *mi = xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride];
      MB_MODE_INFO *mbmi = &mi->mbmi;

      mi_step = AOMMIN(xd->n8_h, mi_size_high[mbmi->sb_type]);

      if (mbmi->ref_frame[0] == ref_frame && mbmi->ref_frame[1] == NONE_FRAME) {
        int bw = block_size_wide[mbmi->sb_type];
        int bh = block_size_high[mbmi->sb_type];
        int x = -AOMMAX(bw, MI_SIZE) / 2 - 1;
        int y = i * MI_SIZE + AOMMAX(bh, MI_SIZE) / 2 - 1;

        pts[0] = (x * 8);
        pts[1] = (y * 8);
        calc_projection_samples(mbmi, x, y, pts_inref);
        pts += 2;
        pts_inref += 2;
        np++;
        if (np >= LEAST_SQUARES_SAMPLES_MAX) return LEAST_SQUARES_SAMPLES_MAX;
      }
    }
  }
  assert(2 * np <= SAMPLES_ARRAY_SIZE);

  if (left_available && up_available) {
    int mi_row_offset = -1;
    int mi_col_offset = -1;

    MODE_INFO *mi = xd->mi[mi_col_offset + mi_row_offset * xd->mi_stride];
    MB_MODE_INFO *mbmi = &mi->mbmi;

    if (mbmi->ref_frame[0] == ref_frame && mbmi->ref_frame[1] == NONE_FRAME) {
      int bw = block_size_wide[mbmi->sb_type];
      int bh = block_size_high[mbmi->sb_type];
      int x = -AOMMAX(bw, MI_SIZE) / 2 - 1;
      int y = -AOMMAX(bh, MI_SIZE) / 2 - 1;

      pts[0] = (x * 8);
      pts[1] = (y * 8);
      calc_projection_samples(mbmi, x, y, pts_inref);
      np++;
    }
  }
  assert(2 * np <= SAMPLES_ARRAY_SIZE);

  return np;
}
#endif  // CONFIG_EXT_WARPED_MOTION

#if CONFIG_EXT_SKIP
void av1_setup_skip_mode_allowed(AV1_COMMON *cm) {
  cm->is_skip_mode_allowed = 0;
  cm->ref_frame_idx_0 = cm->ref_frame_idx_1 = INVALID_IDX;

  if (frame_is_intra_only(cm) || cm->reference_mode == SINGLE_REFERENCE) return;

  RefCntBuffer *const frame_bufs = cm->buffer_pool->frame_bufs;
  const int cur_frame_offset = cm->frame_offset;
  int ref_frame_offset[2] = { -1, INT_MAX };
  int ref_idx[2] = { INVALID_IDX, INVALID_IDX };

  // Identify the nearest forward and backward references.
  for (int i = 0; i < INTER_REFS_PER_FRAME; ++i) {
    const int buf_idx = cm->frame_refs[i].idx;
    if (buf_idx == INVALID_IDX) continue;

    const int ref_offset = frame_bufs[buf_idx].cur_frame_offset;
    if (ref_offset < cur_frame_offset) {
      // Forward reference
      if (ref_offset > ref_frame_offset[0]) {
        ref_frame_offset[0] = ref_offset;
        ref_idx[0] = i;
      }
    } else if (ref_offset > cur_frame_offset) {
      // Backward reference
      if (ref_offset < ref_frame_offset[1]) {
        ref_frame_offset[1] = ref_offset;
        ref_idx[1] = i;
      }
    }
  }

  if (ref_idx[0] != INVALID_IDX && ref_idx[1] != INVALID_IDX) {
    // == Bi-directional prediction ==
    cm->is_skip_mode_allowed = 1;
    cm->ref_frame_idx_0 = AOMMIN(ref_idx[0], ref_idx[1]);
    cm->ref_frame_idx_1 = AOMMAX(ref_idx[0], ref_idx[1]);
  } else if (ref_idx[0] != INVALID_IDX && ref_idx[1] == INVALID_IDX) {
    // == Forward prediction only ==
    // Identify the second nearest forward reference.
    ref_frame_offset[1] = -1;
    for (int i = 0; i < INTER_REFS_PER_FRAME; ++i) {
      const int buf_idx = cm->frame_refs[i].idx;
      if (buf_idx == INVALID_IDX) continue;

      const int ref_offset = frame_bufs[buf_idx].cur_frame_offset;
      if (ref_offset < ref_frame_offset[0] &&
          ref_offset > ref_frame_offset[1]) {
        // Second closest forward reference
        ref_frame_offset[1] = ref_offset;
        ref_idx[1] = i;
      }
    }
    if (ref_frame_offset[1] >= 0) {
      cm->is_skip_mode_allowed = 1;
      cm->ref_frame_idx_0 = AOMMIN(ref_idx[0], ref_idx[1]);
      cm->ref_frame_idx_1 = AOMMAX(ref_idx[0], ref_idx[1]);
    }
  }
}
#endif  // CONFIG_EXT_SKIP