/*

 * Copyright (c) 2021, Alliance for Open Media. All rights reserved

 *

 * This source code is subject to the terms of the BSD 3-Clause Clear License

 * and the Alliance for Open Media Patent License 1.0. If the BSD 3-Clause Clear

 * License was not distributed with this source code in the LICENSE file, you

 * can obtain it at aomedia.org/license/software-license/bsd-3-c-c/.  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

 * aomedia.org/license/patent-license/.

 */

#include <assert.h>

#include <math.h>

#include <string.h>

#include "config/avm_scale_rtcd.h"

#include "avm/avm_integer.h"

#include "av2/common/ccso.h"

#include "av2/common/reconinter.h"

#include "av2/common/av2_common_int.h"

// Derivation of CCSO unit size, ccso unit shall not across tile boundaries

int get_ccso_unit_size_log2_adaptive_tile(const AV2_COMMON *cm,

                                          int sb_size_log2,

                                          int unit_size_log2) {

  if (cm->seq_params.ccso_unit_matches_sb_size)

    return sb_size_log2;

  else if (cm->tiles.cols == 1 && cm->tiles.rows == 1)

    return unit_size_log2;

  int unit_size = unit_size_log2;

  if (sb_size_log2 < unit_size_log2) {

    int e2 = 0, e4 = 0;

    for (int i = 0; i < cm->tiles.cols - 1; ++i) {

      const int size =

          cm->tiles.col_start_sb[i + 1] - cm->tiles.col_start_sb[i];

      e2 += size & 1;

      e4 += size & 3;

    }

    for (int i = 0; i < cm->tiles.rows - 1; ++i) {

      const int size =

          cm->tiles.row_start_sb[i + 1] - cm->tiles.row_start_sb[i];

      e2 += size & 1;

      e4 += size & 3;

    }

    if (e4 == 0)

      unit_size = AVMMIN(sb_size_log2 + 2, unit_size_log2);

    else if (e2 == 0)

      unit_size = AVMMIN(sb_size_log2 + 1, unit_size_log2);

    else

      unit_size = AVMMIN(sb_size_log2, unit_size_log2);

  } else {

    unit_size = sb_size_log2;

  }

  return unit_size;

}

/* Pad the border of a frame */

void extend_ccso_border(const YV12_BUFFER_CONFIG *frame, uint16_t *buf,

                        const int d) {

  int s = frame->y_width + (CCSO_PADDING_SIZE << 1);

  int h = frame->y_height;

  int w = frame->y_width;

  uint16_t *p = &buf[d * s + d];

  for (int y = 0; y < h; y++) {

    for (int x = 0; x < d; x++) {

      *(p - d + x) = p[0];

      p[w + x] = p[w - 1];

    }

    p += s;

  }

  p -= (s + d);

  for (int y = 0; y < d; y++) {

    memcpy(p + (y + 1) * s, p, sizeof(uint16_t) * (w + (d << 1)));

  }

  p -= ((h - 1) * s);

  for (int y = 0; y < d; y++) {

    memcpy(p - (y + 1) * s, p, sizeof(uint16_t) * (w + (d << 1)));

  }

}

void extend_ccso_tile_border(const int tile_height, const int tile_width,

                             const int tile_stride, uint16_t *buf,

                             const int d) {

  uint16_t *p = &buf[d * tile_stride + d];

  for (int y = 0; y < tile_height; y++) {

    for (int x = 0; x < d; x++) {

      *(p - d + x) = p[0];

      p[tile_width + x] = p[tile_width - 1];

    }

    p += tile_stride;

  }

  p -= (tile_stride + d);

  for (int y = 0; y < d; y++) {

    memcpy(p + (y + 1) * tile_stride, p,

           sizeof(uint16_t) * (tile_width + (d << 1)));

  }

  p -= ((tile_height - 1) * tile_stride);

  for (int y = 0; y < d; y++) {

    memcpy(p - (y + 1) * tile_stride, p,

           sizeof(uint16_t) * (tile_width + (d << 1)));

  }

}

/* Derive the quantized index, later it can be used for retriving offset values

 * from the look-up table */

void cal_filter_support(int *rec_luma_idx, const uint16_t *rec_y,

                        const int quant_step_size, const int inv_quant_step,

                        const int *rec_idx, const int edge_clf) {

  if (edge_clf == 0) {

    for (int i = 0; i < 2; i++) {

      int d = rec_y[rec_idx[i]] - rec_y[0];

      if (d > quant_step_size)

        rec_luma_idx[i] = 2;

      else if (d < inv_quant_step)

        rec_luma_idx[i] = 0;

      else

        rec_luma_idx[i] = 1;

    }

  } else {  // if (edge_clf == 1)

    for (int i = 0; i < 2; i++) {

      int d = rec_y[rec_idx[i]] - rec_y[0];

      if (d < inv_quant_step)

        rec_luma_idx[i] = 0;

      else

        rec_luma_idx[i] = 1;

    }

  }

}

/* Derive sample locations for CCSO */

void derive_ccso_sample_pos(int *rec_idx, const int ccso_stride,

                            const uint8_t ext_filter_support) {

  // Input sample locations for CCSO

  // 4 2 0 3 5

  // 6 1 x 1 6

  // 5 3 0 2 4

  assert(ext_filter_support < 7);

  if (ext_filter_support == 0) {

    rec_idx[0] = -1 * ccso_stride;

    rec_idx[1] = 1 * ccso_stride;

  } else if (ext_filter_support == 1) {

    rec_idx[0] = -1;

    rec_idx[1] = 1;

  } else if (ext_filter_support == 4) {

    rec_idx[0] = -ccso_stride - 2;

    rec_idx[1] = ccso_stride + 2;

  } else if (ext_filter_support == 5) {

    rec_idx[0] = ccso_stride - 2;

    rec_idx[1] = -ccso_stride + 2;

  } else if (ext_filter_support == 2) {

    rec_idx[0] = -1 * ccso_stride - 1;

    rec_idx[1] = 1 * ccso_stride + 1;

  } else if (ext_filter_support == 3) {

    rec_idx[0] = -1 * ccso_stride + 1;

    rec_idx[1] = 1 * ccso_stride - 1;

  } else {  // ext_filter_support == 6

    rec_idx[0] = 2;

    rec_idx[1] = -2;

  }

}

/* Derive rect area of a tile for CCSO process */

INLINE static AV2PixelRect av2_get_tile_rect_ccso(const TileInfo *tile_info,

                                                  const AV2_COMMON *cm,

                                                  MACROBLOCKD *xd, int is_uv) {

  AV2PixelRect r;

  // Calculate position in the Y plane

  r.left = tile_info->mi_col_start * MI_SIZE;

  r.right = tile_info->mi_col_end * MI_SIZE;

  r.top = tile_info->mi_row_start * MI_SIZE;

  r.bottom = tile_info->mi_row_end * MI_SIZE;

  const int frame_w = xd->plane[0].dst.width;

  const int frame_h = xd->plane[0].dst.height;

  // Make sure we don't fall off the bottom-right of the frame.

  r.right = AVMMIN(r.right, frame_w);

  r.bottom = AVMMIN(r.bottom, frame_h);

  // Convert to coordinates in the appropriate plane

  const int ss_x = is_uv && cm->seq_params.subsampling_x;

  const int ss_y = is_uv && cm->seq_params.subsampling_y;

  r.left = ROUND_POWER_OF_TWO(r.left, ss_x);

  r.right = ROUND_POWER_OF_TWO(r.right, ss_x);

  r.top = ROUND_POWER_OF_TWO(r.top, ss_y);

  r.bottom = ROUND_POWER_OF_TWO(r.bottom, ss_y);

  return r;

}

// Cross-component Sample Offset band offset only case.

void ccso_filter_block_hbd_wo_buf_bo_only_c(

    const uint16_t *src_y, uint16_t *dst_yuv, const int x, const int y,

    const int pic_width, const int pic_height, const int8_t *offset_buf,

    const int src_y_stride, const int dst_stride, const int y_uv_hscale,

    const int y_uv_vscale, const int max_val, const int blk_size_x,

    const int blk_size_y, const bool isSingleBand, const uint8_t shift_bits) {

  const int y_end = AVMMIN(pic_height - y, blk_size_y);

  const int x_end = AVMMIN(pic_width - x, blk_size_x);

  for (int y_start = 0; y_start < y_end; y_start++) {

    const int y_pos = y_start;

    for (int x_start = 0; x_start < x_end; x_start++) {

      const int x_pos = x + x_start;

      const int band_num = isSingleBand

                               ? 0

                               : src_y[(y_pos << y_uv_vscale) * src_y_stride +

                                       (x_pos << y_uv_hscale)] >>

                                     shift_bits;

      const int lut_idx_ext = (band_num << 4);

      const int offset_val = offset_buf[lut_idx_ext];

      dst_yuv[y_pos * dst_stride + x_pos] =

          clamp(offset_val + dst_yuv[y_pos * dst_stride + x_pos], 0, max_val);

    }

  }

}

void ccso_filter_block_hbd_wo_buf_c(

    const uint16_t *src_y, uint16_t *dst_yuv, const int x, const int y,

    const int pic_width, const int pic_height, int *src_cls,

    const int8_t *offset_buf, const int src_y_stride, const int dst_stride,

    const int y_uv_hscale, const int y_uv_vscale, const int thr,

    const int neg_thr, const int *src_loc, const int max_val,

    const int blk_size_x, const int blk_size_y, const bool isSingleBand,

    const uint8_t shift_bits, const int edge_clf, const uint8_t ccso_bo_only) {

  const int y_end = AVMMIN(pic_height - y, blk_size_y);

  const int x_end = AVMMIN(pic_width - x, blk_size_x);

  for (int y_start = 0; y_start < y_end; y_start++) {

    const int y_pos = y_start;

    for (int x_start = 0; x_start < x_end; x_start++) {

      const int x_pos = x + x_start;

      if (!ccso_bo_only) {

        cal_filter_support(src_cls,

                           &src_y[(y_pos << y_uv_vscale) * src_y_stride +

                                  (x_pos << y_uv_hscale)],

                           thr, neg_thr, src_loc, edge_clf);

      } else {

        src_cls[0] = 0;

        src_cls[1] = 0;

      }

      const int band_num = isSingleBand

                               ? 0

                               : src_y[(y_pos << y_uv_vscale) * src_y_stride +

                                       (x_pos << y_uv_hscale)] >>

                                     shift_bits;

      const int lut_idx_ext = (band_num << 4) + (src_cls[0] << 2) + src_cls[1];

      const int offset_val = offset_buf[lut_idx_ext];

      dst_yuv[y_pos * dst_stride + x_pos] =

          clamp(offset_val + dst_yuv[y_pos * dst_stride + x_pos], 0, max_val);

    }

  }

}

// If there is at-least 1 segment is lossless in a frame, we have

// to do 4x4 processing, because minimum lossless block can be 4x4

// size. Although, regardless the value of

// cm->features.has_lossless_segment, we can always do 4x4

// processing, however, for software optimization purpose we have

// used  full block processing for whole lossy frame.

void ccso_filter_block_hbd_wo_buf_4x4_c(

    AV2_COMMON *cm, const uint16_t *src_y, uint16_t *dst_yuv,

    int tile_col_start, int tile_row_start, const int x, const int y,

    const int pic_width, const int pic_height, int *src_cls,

    const int8_t *offset_buf, const int src_y_stride, const int dst_stride,

    const int y_uv_hscale, const int y_uv_vscale, const int thr,

    const int neg_thr, const int *src_loc, const int max_val,

    const int blk_size_x, const int blk_size_y, const bool isSingleBand,

    const uint8_t shift_bits, const int edge_clf, const uint8_t ccso_bo_only,

    int plane) {

  const CommonModeInfoParams *const mi_params = &cm->mi_params;

  const int y_end = AVMMIN(pic_height - y, blk_size_y);

  const int x_end = AVMMIN(pic_width - x, blk_size_x);

  int min_b_size_x = (1 << MI_SIZE_LOG2) >> y_uv_hscale;

  int min_b_size_y = (1 << MI_SIZE_LOG2) >> y_uv_vscale;

  for (int y_start = 0; y_start < y_end; y_start += min_b_size_y) {

    const int y_pos = y_start;

    for (int x_start = 0; x_start < x_end; x_start += min_b_size_x) {

      const int x_pos = x + x_start;

      const int this_mi_row =

          ((tile_row_start + y_pos + y) << y_uv_vscale) >> MI_SIZE_LOG2;

      const int this_mi_col =

          ((tile_col_start + x_pos) << y_uv_hscale) >> MI_SIZE_LOG2;

      MB_MODE_INFO **this_mbmi_ptr = mi_params->mi_grid_base +

                                     this_mi_row * mi_params->mi_stride +

                                     this_mi_col;

      MB_MODE_INFO **this_mbmi =

          get_mi_location_from_collocated_mi(cm, this_mbmi_ptr, plane);

      const int is_lossless =

          cm->features.lossless_segment[this_mbmi[0]->segment_id];

      if (!is_lossless) {

        int j_max = AVMMIN(x_pos + min_b_size_x, x + x_start + x_end);

        int i_max = AVMMIN(y_pos + min_b_size_y, y_end);

        for (int i_pos_4x4 = y_pos; i_pos_4x4 < i_max; i_pos_4x4++) {

          for (int j_pos_4x4 = x_pos; j_pos_4x4 < j_max; j_pos_4x4++) {

            if (!ccso_bo_only) {

              cal_filter_support(

                  src_cls,

                  &src_y[(i_pos_4x4 << y_uv_vscale) * src_y_stride +

                         (j_pos_4x4 << y_uv_hscale)],

                  thr, neg_thr, src_loc, edge_clf);

            } else {

              src_cls[0] = 0;

              src_cls[1] = 0;

            }

            const int band_num =

                isSingleBand ? 0

                             : src_y[(i_pos_4x4 << y_uv_vscale) * src_y_stride +

                                     (j_pos_4x4 << y_uv_hscale)] >>

                                   shift_bits;

            const int lut_idx_ext =

                (band_num << 4) + (src_cls[0] << 2) + src_cls[1];

            const int offset_val = offset_buf[lut_idx_ext];

            dst_yuv[i_pos_4x4 * dst_stride + j_pos_4x4] =

                clamp(offset_val + dst_yuv[i_pos_4x4 * dst_stride + j_pos_4x4],

                      0, max_val);

          }

        }

      }

    }

  }

}

// Apply CCSO for each process block row

void av2_apply_ccso_filter_for_row(AV2_COMMON *cm, MACROBLOCKD *xd,

                                   const uint16_t *src_y, uint16_t *dst_yuv,

                                   int *src_loc, int *src_cls, int blk_row,

                                   int thr, int blk_size, int blk_size_proc,

                                   int blk_log2_x, int blk_log2_y,

                                   int unit_log2_x, int unit_log2_y,

                                   int plane) {

  const CommonModeInfoParams *const mi_params = &cm->mi_params;

  const int src_y_stride = xd->plane[0].dst.width + (CCSO_PADDING_SIZE << 1);

  const uint8_t max_band_log2 = cm->ccso_info.max_band_log2[plane];

  const int dst_stride = xd->plane[plane].dst.stride;

  const int pic_height = xd->plane[plane].dst.height;

  const int pic_width = xd->plane[plane].dst.width;

  const int neg_thr = thr * -1;

  const bool is_single_band = !max_band_log2;

  const uint8_t shift_bits = cm->seq_params.bit_depth - max_band_log2;

  const int max_val = (1 << cm->seq_params.bit_depth) - 1;

  const int edge_clf = cm->ccso_info.edge_clf[plane];

  const int y_uv_hscale = xd->plane[plane].subsampling_x;

  const int y_uv_vscale = xd->plane[plane].subsampling_y;

  for (int blk_col = 0; blk_col < pic_width; blk_col += blk_size_proc) {

    const int ccso_blk_idx =

        (blk_size >> MI_SIZE_LOG2) * (blk_row >> blk_log2_y) *

            mi_params->mi_stride +

        (blk_size >> MI_SIZE_LOG2) * (blk_col >> blk_log2_x);

    const bool use_ccso =

        (plane == 0)   ? mi_params->mi_grid_base[ccso_blk_idx]->ccso_blk_y

        : (plane == 1) ? mi_params->mi_grid_base[ccso_blk_idx]->ccso_blk_u

                       : mi_params->mi_grid_base[ccso_blk_idx]->ccso_blk_v;

    if (!use_ccso) continue;

    // FPU level skip

    const int x_mbmi = (blk_col >> unit_log2_x) << unit_log2_x;

    const int y_mbmi = (blk_row >> unit_log2_y) << unit_log2_y;

    const int mbmi_idx =

        get_mi_grid_idx(mi_params, y_mbmi >> (MI_SIZE_LOG2 - y_uv_vscale),

                        x_mbmi >> (MI_SIZE_LOG2 - y_uv_hscale));

    const int use_ccso_local =

        mi_params->mi_grid_base[mbmi_idx]->local_ccso_blk_flag;

    if (!use_ccso_local) continue;

    if (cm->bru.enabled &&

        mi_params->mi_grid_base[mbmi_idx]->sb_active_mode != BRU_ACTIVE_SB) {

      avm_internal_error(

          &cm->error, AVM_CODEC_ERROR,

          "Invalid BRU activity in CCSO: only active SB can be filtered");

      return;

    }

    if (cm->bridge_frame_info.is_bridge_frame) {

      avm_internal_error(

          &cm->error, AVM_CODEC_ERROR,

          "Invalid Bridge frame activity in CCSO: can not be filtered");

      return;

    }

    if (cm->features.has_lossless_segment) {

      ccso_filter_block_hbd_wo_buf_4x4_c(

          cm, src_y, dst_yuv, 0, 0, blk_col, blk_row, pic_width, pic_height,

          src_cls, cm->ccso_info.filter_offset[plane], src_y_stride, dst_stride,

          y_uv_hscale, y_uv_vscale, thr, neg_thr, src_loc, max_val,

          blk_size_proc, blk_size_proc, is_single_band, shift_bits, edge_clf,

          cm->ccso_info.ccso_bo_only[plane], plane);

    } else {

      if (cm->ccso_info.ccso_bo_only[plane]) {

        ccso_filter_block_hbd_wo_buf_bo_only(

            src_y, dst_yuv, blk_col, blk_row, pic_width, pic_height,

            cm->ccso_info.filter_offset[plane], src_y_stride, dst_stride,

            y_uv_hscale, y_uv_vscale, max_val, blk_size_proc, blk_size_proc,

            is_single_band, shift_bits);

      } else {

        ccso_filter_block_hbd_wo_buf(

            src_y, dst_yuv, blk_col, blk_row, pic_width, pic_height, src_cls,

            cm->ccso_info.filter_offset[plane], src_y_stride, dst_stride,

            y_uv_hscale, y_uv_vscale, thr, neg_thr, src_loc, max_val,

            blk_size_proc, blk_size_proc, is_single_band, shift_bits, edge_clf,

            0);

      }

    }

  }

}

/* Apply CCSO on luma or chroma component when single or multiple bands are

 * applied */

void apply_ccso_filter(AV2_COMMON *cm, MACROBLOCKD *xd, int plane,

                       const uint16_t *src_y, uint16_t *dst_yuv, int dst_stride,

                       int proc_unit_log2, uint16_t thr, uint8_t filter_sup,

                       uint8_t max_band_log2, int edge_clf) {

  const int ccso_ext_stride = xd->plane[0].dst.width + (CCSO_PADDING_SIZE << 1);

  const int pic_height = xd->plane[plane].dst.height;

  int src_cls[2];

  int src_loc[2];

  const int y_uv_hscale = xd->plane[plane].subsampling_x;

  const int y_uv_vscale = xd->plane[plane].subsampling_y;

  derive_ccso_sample_pos(src_loc, ccso_ext_stride, filter_sup);

  const int ccso_blk_size = get_ccso_unit_size_log2_adaptive_tile(

      cm, cm->mib_size_log2 + MI_SIZE_LOG2, CCSO_BLK_SIZE);

  const int blk_log2 = ccso_blk_size;

  const int blk_size = 1 << blk_log2;

  const int blk_log2_x = blk_log2 - y_uv_hscale;

  const int blk_log2_y = blk_log2 - y_uv_vscale;

  src_y += CCSO_PADDING_SIZE * ccso_ext_stride + CCSO_PADDING_SIZE;

  const int unit_log2_x = AVMMIN(proc_unit_log2, blk_log2_x);

  const int unit_log2_y = AVMMIN(proc_unit_log2, blk_log2_y);

  const CommonModeInfoParams *const mi_params = &cm->mi_params;

  const uint8_t shift_bits = cm->seq_params.bit_depth - max_band_log2;

  const bool is_single_band = !max_band_log2;

  const int max_val = (1 << cm->seq_params.bit_depth) - 1;

  const int neg_thr = thr * -1;

  const int unit_size_x = 1 << unit_log2_x;

  const int unit_size_y = 1 << unit_log2_y;

  const int blk_size_x = 1 << blk_log2_x;

  const int blk_size_y = 1 << blk_log2_y;

  if (cm->seq_params.disable_loopfilters_across_tiles) {

    int tile_rows = cm->tiles.rows;

    int tile_cols = cm->tiles.cols;

    TileInfo tile_info_y;

    AV2PixelRect tile_rect_y;

    for (int tile_row = 0; tile_row < tile_rows; ++tile_row) {

      int tile_height = 0;

      for (int tile_col = 0; tile_col < tile_cols; ++tile_col) {

        av2_tile_init(&tile_info_y, cm, tile_row, tile_col);

        tile_rect_y = av2_get_tile_rect_ccso(&tile_info_y, cm, xd, 0);

        int tile_row_start = tile_rect_y.top;

        int tile_col_start = tile_rect_y.left;

        int tile_row_end = tile_rect_y.bottom;

        int tile_col_end = tile_rect_y.right;

        int tile_width = tile_col_end - tile_col_start;

        tile_height = tile_row_end - tile_row_start;

        const int ccso_ext_tile_stride = tile_width + (CCSO_PADDING_SIZE << 1);

        derive_ccso_sample_pos(src_loc, ccso_ext_tile_stride, filter_sup);

        uint16_t *ext_rec_tile_y = NULL;

        // const uint16_t *rec_y = cm->cur_frame->buf.y_buffer;

        ext_rec_tile_y = avm_malloc(sizeof(*ext_rec_tile_y) *

                                    (tile_height + (CCSO_PADDING_SIZE << 1)) *

                                    (tile_width + (CCSO_PADDING_SIZE << 1)));

        for (int r = 0; r < tile_height; ++r) {

          for (int c = 0; c < tile_width; ++c) {

            ext_rec_tile_y[(r + CCSO_PADDING_SIZE) * ccso_ext_tile_stride + c +

                           CCSO_PADDING_SIZE] =

                src_y[(r + tile_row_start) * ccso_ext_stride +

                      (c + tile_col_start)];

          }

        }

        extend_ccso_tile_border(tile_height, tile_width, ccso_ext_tile_stride,

                                ext_rec_tile_y, CCSO_PADDING_SIZE);

        uint16_t *src_tile_y = ext_rec_tile_y;

        src_tile_y +=

            CCSO_PADDING_SIZE * ccso_ext_tile_stride + CCSO_PADDING_SIZE;

        if (plane != 0) {

          TileInfo tile_info_uv;

          AV2PixelRect tile_rect_uv;

          av2_tile_init(&tile_info_uv, cm, tile_row, tile_col);

          tile_rect_uv = av2_get_tile_rect_ccso(&tile_info_uv, cm, xd, 1);

          tile_row_start = tile_rect_uv.top;

          tile_col_start = tile_rect_uv.left;

          tile_row_end = tile_rect_uv.bottom;

          tile_col_end = tile_rect_uv.right;

          tile_width = tile_col_end - tile_col_start;

          tile_height = tile_row_end - tile_row_start;

        }

        uint16_t *dst_tile_yuv = dst_yuv + tile_col_start;

        for (int frame_pxl_y = tile_row_start; frame_pxl_y < tile_row_end;

             frame_pxl_y += blk_size_y) {

          for (int frame_pxl_x = tile_col_start; frame_pxl_x < tile_col_end;

               frame_pxl_x += blk_size_x) {

            // int x = frame_pxl_x;

            // int y = frame_pxl_y;

            int tile_pxl_x = frame_pxl_x - tile_col_start;

            int tile_pxl_y = frame_pxl_y - tile_row_start;

            const int ccso_blk_idx =

                (blk_size >> MI_SIZE_LOG2) * (frame_pxl_y >> blk_log2_y) *

                    mi_params->mi_stride +

                (blk_size >> MI_SIZE_LOG2) * (frame_pxl_x >> blk_log2_x);

            const bool use_ccso =

                (plane == 0) ? mi_params->mi_grid_base[ccso_blk_idx]->ccso_blk_y

                : (plane == 1)

                    ? mi_params->mi_grid_base[ccso_blk_idx]->ccso_blk_u

                    : mi_params->mi_grid_base[ccso_blk_idx]->ccso_blk_v;

            if (!use_ccso) continue;

            const uint16_t *src_unit_y = src_tile_y;

            uint16_t *dst_unit_yuv = dst_tile_yuv;

            const int y_end = AVMMIN(tile_row_end - frame_pxl_y, blk_size_y);

            const int x_end = AVMMIN(tile_col_end - frame_pxl_x, blk_size_x);

            for (int unit_y = 0; unit_y < y_end; unit_y += unit_size_y) {

              for (int unit_x = 0; unit_x < x_end; unit_x += unit_size_x) {

                // FPU level skip

                const int mbmi_idx = get_mi_grid_idx(

                    mi_params,

                    (frame_pxl_y + unit_y) >> (MI_SIZE_LOG2 - y_uv_vscale),

                    (frame_pxl_x + unit_x) >> (MI_SIZE_LOG2 - y_uv_hscale));

                const int use_ccso_local =

                    mi_params->mi_grid_base[mbmi_idx]->local_ccso_blk_flag;

                if (!use_ccso_local) {

                  continue;

                }

                if (cm->bru.enabled &&

                    mi_params->mi_grid_base[mbmi_idx]->sb_active_mode !=

                        BRU_ACTIVE_SB) {

                  avm_internal_error(&cm->error, AVM_CODEC_ERROR,

                                     "Invalid BRU activity in CCSO: only "

                                     "active SB can be filtered");

                  return;

                }

                if (cm->features.has_lossless_segment) {

                  ccso_filter_block_hbd_wo_buf_4x4_c(

                      cm, src_unit_y, dst_unit_yuv, tile_col_start,

                      tile_row_start, tile_pxl_x + unit_x, tile_pxl_y + unit_y,

                      tile_width, tile_height, src_cls,

                      cm->ccso_info.filter_offset[plane], ccso_ext_tile_stride,

                      dst_stride, y_uv_hscale, y_uv_vscale, thr, neg_thr,

                      src_loc, max_val, unit_size_x, unit_size_y,

                      is_single_band, shift_bits, edge_clf,

                      cm->ccso_info.ccso_bo_only[plane], plane);

                } else {

                  if (cm->ccso_info.ccso_bo_only[plane]) {

                    ccso_filter_block_hbd_wo_buf_c(

                        src_unit_y, dst_unit_yuv, tile_pxl_x + unit_x,

                        tile_pxl_y + unit_y, tile_width, tile_height, src_cls,

                        cm->ccso_info.filter_offset[plane],

                        ccso_ext_tile_stride, dst_stride, y_uv_hscale,

                        y_uv_vscale, thr, neg_thr, src_loc, max_val,

                        unit_size_x, unit_size_y, is_single_band, shift_bits,

                        edge_clf, cm->ccso_info.ccso_bo_only[plane]);

                  } else {

                    ccso_filter_block_hbd_wo_buf(

                        src_unit_y, dst_unit_yuv, tile_pxl_x + unit_x,

                        tile_pxl_y + unit_y, tile_width, tile_height, src_cls,

                        cm->ccso_info.filter_offset[plane],

                        ccso_ext_tile_stride, dst_stride, y_uv_hscale,

                        y_uv_vscale, thr, neg_thr, src_loc, max_val,

                        unit_size_x, unit_size_y, is_single_band, shift_bits,

                        edge_clf, 0);

                  }

                }

              }

              dst_unit_yuv += (dst_stride << unit_log2_y);

              src_unit_y +=

                  (ccso_ext_tile_stride << (unit_log2_y + y_uv_vscale));

            }

          }

          dst_tile_yuv += (dst_stride << blk_log2_y);

          src_tile_y += (ccso_ext_tile_stride << (blk_log2_y + y_uv_vscale));

        }

        avm_free(ext_rec_tile_y);

      }

      dst_yuv += (dst_stride * tile_height);

      // src_y += (ccso_ext_stride * (cm->tiles.height << (MI_SIZE_LOG2)));

    }

    return;

  }

  const int blk_log2_proc = CCSO_PROC_BLK_LOG2;

  const int blk_size_proc = 1 << blk_log2_proc;

  for (int blk_row = 0; blk_row < pic_height; blk_row += blk_size_proc) {

    av2_apply_ccso_filter_for_row(

        cm, xd, src_y, dst_yuv, src_loc, src_cls, blk_row, thr, blk_size,

        blk_size_proc, blk_log2_x, blk_log2_y, unit_log2_x, unit_log2_y, plane);

    dst_yuv += dst_stride << blk_log2_proc;

    src_y += ccso_ext_stride << (blk_log2_proc + y_uv_vscale);

  }

}

/* Apply CCSO for one frame */

void ccso_frame(YV12_BUFFER_CONFIG *frame, AV2_COMMON *cm, MACROBLOCKD *xd,

                uint16_t *ext_rec_y) {

  const int num_planes = av2_num_planes(cm);

  av2_setup_dst_planes(xd->plane, frame, 0, 0, 0, num_planes, NULL);

  for (int plane = 0; plane < num_planes; plane++) {

    const int dst_stride = xd->plane[plane].dst.stride;

    const uint16_t quant_step_size = quant_sz[cm->ccso_info.scale_idx[plane]]

                                             [cm->ccso_info.quant_idx[plane]];

    if (cm->ccso_info.ccso_enable[plane]) {

      apply_ccso_filter(

          cm, xd, plane, ext_rec_y, &(xd->plane[plane].dst.buf)[0], dst_stride,

          cm->mib_size_log2 -

              AVMMAX(xd->plane[plane].subsampling_x,

                     xd->plane[plane].subsampling_y) +

              MI_SIZE_LOG2,

          quant_step_size, cm->ccso_info.ext_filter_support[plane],

          cm->ccso_info.max_band_log2[plane], cm->ccso_info.edge_clf[plane]);

    }

  }

}

// This function is to copy ccso filter parameters between frames when

// ccso_reuse is true.

void av2_copy_ccso_filters(CcsoInfo *to, CcsoInfo *from, int plane,

                           bool frame_level, bool block_level, int sb_count) {

  if (frame_level) {

    memcpy(to->filter_offset[plane], from->filter_offset[plane],

           sizeof(to->filter_offset[plane]));

    to->quant_idx[plane] = from->quant_idx[plane];

    to->ext_filter_support[plane] = from->ext_filter_support[plane];

    to->edge_clf[plane] = from->edge_clf[plane];

    to->ccso_bo_only[plane] = from->ccso_bo_only[plane];

    to->max_band_log2[plane] = from->max_band_log2[plane];

    to->scale_idx[plane] = from->scale_idx[plane];

  }

  if (block_level) {

    if (to->sb_filter_control[plane]) {

      memcpy(to->sb_filter_control[plane], from->sb_filter_control[plane],

             sizeof(*from->sb_filter_control[plane]) * sb_count);

    }

  }

  to->ccso_enable[plane] = from->ccso_enable[plane];

}