libvpx: Pull from upstream

source/libvpx/vp9/encoder/vp9_pickmode.c

 /*
  *  Copyright (c) 2014 The WebM project authors. All Rights Reserved.
  *
  *  Use of this source code is governed by a BSD-style license
  *  that can be found in the LICENSE file in the root of the source
  *  tree. An additional intellectual property rights grant can be found
  *  in the file PATENTS.  All contributing project authors may
  *  be found in the AUTHORS file in the root of the source tree.
  */
 
 #include <assert.h>
 #include <limits.h>
 #include <math.h>
 #include <stdio.h>
 
 #include "./vp9_rtcd.h"
 #include "./vpx_dsp_rtcd.h"
 
+#include "vpx_dsp/vpx_dsp_common.h"
 #include "vpx_mem/vpx_mem.h"
 #include "vpx_ports/mem.h"
 
 #include "vp9/common/vp9_blockd.h"
 #include "vp9/common/vp9_common.h"
 #include "vp9/common/vp9_mvref_common.h"
 #include "vp9/common/vp9_pred_common.h"
 #include "vp9/common/vp9_reconinter.h"
 #include "vp9/common/vp9_reconintra.h"
 #include "vp9/common/vp9_scan.h"
@@ skipping 651 matching lines @@
     }
   }
 
   if (*skippable == 0) {
     *rate <<= 10;
     *rate += (eob_cost << 8);
   }
 }
 #endif
 
-static void model_rd_for_sb_uv(VP9_COMP *cpi, BLOCK_SIZE bsize,
+static void model_rd_for_sb_uv(VP9_COMP *cpi, BLOCK_SIZE plane_bsize,
                                MACROBLOCK *x, MACROBLOCKD *xd,
                                int *out_rate_sum, int64_t *out_dist_sum,
-                               unsigned int *var_y, unsigned int *sse_y) {
+                               unsigned int *var_y, unsigned int *sse_y,
+                               int start_plane, int stop_plane) {
   // Note our transform coeffs are 8 times an orthogonal transform.
   // Hence quantizer step is also 8 times. To get effective quantizer
   // we need to divide by 8 before sending to modeling function.
   unsigned int sse;
   int rate;
   int64_t dist;
   int i;
 
   *out_rate_sum = 0;
   *out_dist_sum = 0;
 
-  for (i = 1; i <= 2; ++i) {
+  for (i = start_plane; i <= stop_plane; ++i) {
     struct macroblock_plane *const p = &x->plane[i];
     struct macroblockd_plane *const pd = &xd->plane[i];
     const uint32_t dc_quant = pd->dequant[0];
     const uint32_t ac_quant = pd->dequant[1];
-    const BLOCK_SIZE bs = get_plane_block_size(bsize, pd);
+    const BLOCK_SIZE bs = plane_bsize;
     unsigned int var;
 
     if (!x->color_sensitivity[i - 1])
       continue;
 
     var = cpi->fn_ptr[bs].vf(p->src.buf, p->src.stride,
                              pd->dst.buf, pd->dst.stride, &sse);
     *var_y += var;
     *sse_y += sse;
 
@@ skipping 164 matching lines @@
   struct macroblock_plane *const p = &x->plane[0];
   struct macroblockd_plane *const pd = &xd->plane[0];
   const BLOCK_SIZE bsize_tx = txsize_to_bsize[tx_size];
   uint8_t *const src_buf_base = p->src.buf;
   uint8_t *const dst_buf_base = pd->dst.buf;
   const int src_stride = p->src.stride;
   const int dst_stride = pd->dst.stride;
   int i, j;
   int rate;
   int64_t dist;
-  int64_t this_sse = INT64_MAX;
-  int is_skippable;
 
   txfrm_block_to_raster_xy(plane_bsize, tx_size, block, &i, &j);
-  assert(plane == 0);
-  (void) plane;
 
   p->src.buf = &src_buf_base[4 * (j * src_stride + i)];
   pd->dst.buf = &dst_buf_base[4 * (j * dst_stride + i)];
   // Use source buffer as an approximation for the fully reconstructed buffer.
   vp9_predict_intra_block(xd, b_width_log2_lookup[plane_bsize],
                           tx_size, args->mode,
                           x->skip_encode ? p->src.buf : pd->dst.buf,
                           x->skip_encode ? src_stride : dst_stride,
                           pd->dst.buf, dst_stride,
-                          i, j, 0);
+                          i, j, plane);
 
-  // TODO(jingning): This needs further refactoring.
-  block_yrd(cpi, x, &rate, &dist, &is_skippable, &this_sse, 0,
-            bsize_tx, VPXMIN(tx_size, TX_16X16));
-  x->skip_txfm[0] = is_skippable;
-  rate += vp9_cost_bit(vp9_get_skip_prob(&cpi->common, xd), is_skippable);
+  if (plane == 0) {
+    int64_t this_sse = INT64_MAX;
+    int is_skippable;
+    // TODO(jingning): This needs further refactoring.
+    block_yrd(cpi, x, &rate, &dist, &is_skippable, &this_sse, 0,
+              bsize_tx, VPXMIN(tx_size, TX_16X16));
+    x->skip_txfm[0] = is_skippable;
+    // TODO(jingning): Skip is signalled per prediciton block not per tx block.
+    rate += vp9_cost_bit(vp9_get_skip_prob(&cpi->common, xd), is_skippable);
+  } else {
+    unsigned int var, sse;
+    model_rd_for_sb_uv(cpi, plane_bsize, x, xd, &rate, &dist, &var, &sse,
+                       plane, plane);
+  }
 
   p->src.buf = src_buf_base;
   pd->dst.buf = dst_buf_base;
   args->rate += rate;
   args->dist += dist;
 }
 
 static const THR_MODES mode_idx[MAX_REF_FRAMES - 1][4] = {
   {THR_DC, THR_V_PRED, THR_H_PRED, THR_TM},
   {THR_NEARESTMV, THR_NEARMV, THR_ZEROMV, THR_NEWMV},
@@ skipping 508 matching lines @@
       }
     } else {
       this_rdc.rate += cm->interp_filter == SWITCHABLE ?
           vp9_get_switchable_rate(cpi, xd) : 0;
       this_rdc.rate += vp9_cost_bit(vp9_get_skip_prob(cm, xd), 1);
     }
 
     if (x->color_sensitivity[0] || x->color_sensitivity[1]) {
       int uv_rate = 0;
       int64_t uv_dist = 0;
+      const BLOCK_SIZE uv_bsize = get_plane_block_size(bsize, &xd->plane[1]);
       if (x->color_sensitivity[0])
         vp9_build_inter_predictors_sbp(xd, mi_row, mi_col, bsize, 1);
       if (x->color_sensitivity[1])
         vp9_build_inter_predictors_sbp(xd, mi_row, mi_col, bsize, 2);
-      model_rd_for_sb_uv(cpi, bsize, x, xd, &uv_rate, &uv_dist,
-                         &var_y, &sse_y);
+      model_rd_for_sb_uv(cpi, uv_bsize, x, xd, &uv_rate, &uv_dist,
+                         &var_y, &sse_y, 1, 2);
       this_rdc.rate += uv_rate;
       this_rdc.dist += uv_dist;
     }
 
     this_rdc.rate += rate_mv;
     this_rdc.rate +=
         cpi->inter_mode_cost[x->mbmi_ext->mode_context[ref_frame]][INTER_OFFSET(
             this_mode)];
     this_rdc.rate += ref_frame_cost[ref_frame];
     this_rdc.rdcost = RDCOST(x->rdmult, x->rddiv, this_rdc.rate, this_rdc.dist);
@@ skipping 102 matching lines @@
         continue;
 
       mbmi->mode = this_mode;
       mbmi->ref_frame[0] = INTRA_FRAME;
       args.mode = this_mode;
       args.rate = 0;
       args.dist = 0;
       mbmi->tx_size = intra_tx_size;
       vp9_foreach_transformed_block_in_plane(xd, bsize, 0,
                                              estimate_block_intra, &args);
+      // Inter and intra RD will mismatch in scale for non-screen content.
+      if (cpi->oxcf.content == VP9E_CONTENT_SCREEN) {
+        if (x->color_sensitivity[0])
+          vp9_foreach_transformed_block_in_plane(xd, bsize, 1,
+                                                 estimate_block_intra, &args);
+        if (x->color_sensitivity[1])
+          vp9_foreach_transformed_block_in_plane(xd, bsize, 2,
+                                                 estimate_block_intra, &args);
+      }
       this_rdc.rate = args.rate;
       this_rdc.dist = args.dist;
       this_rdc.rate += cpi->mbmode_cost[this_mode];
       this_rdc.rate += ref_frame_cost[INTRA_FRAME];
       this_rdc.rate += intra_cost_penalty;
       this_rdc.rdcost = RDCOST(x->rdmult, x->rddiv,
                                this_rdc.rate, this_rdc.dist);
 
       if (this_rdc.rdcost < best_rdc.rdcost) {
         best_rdc = this_rdc;
@@ skipping 325 matching lines @@
     }
   }
   mbmi->mode = xd->mi[0]->bmi[3].as_mode;
   ctx->mic = *(xd->mi[0]);
   ctx->mbmi_ext = *x->mbmi_ext;
   ctx->skip_txfm[0] = SKIP_TXFM_NONE;
   ctx->skip = 0;
   // Dummy assignment for speed -5. No effect in speed -6.
   rd_cost->rdcost = best_rd;
 }