Merge "Hybrid transform cleanups" into experimental

4 files changed, 167 insertions, 129 deletions

diff --git a/vp8/common/blockd.h b/vp8/common/blockd.h
index 96345c2bd..facbebb31 100644
--- a/vp8/common/blockd.h
+++ b/vp8/common/blockd.h
@@ -479,7 +479,35 @@ static void txfm_map(BLOCKD *b, B_PREDICTION_MODE bmode) {
       break;
   }
 }
+
+static TX_TYPE get_tx_type(MACROBLOCKD *xd, BLOCKD *b) {
+  TX_TYPE tx_type = DCT_DCT;
+#if CONFIG_HYBRIDTRANSFORM16X16
+  if (xd->mode_info_context->mbmi.txfm_size == TX_16X16) {
+    if (xd->mode_info_context->mbmi.mode < I8X8_PRED &&
+        xd->q_index < ACTIVE_HT16)
+      tx_type = b->bmi.as_mode.tx_type;
+    return tx_type;
+  }
+#endif
+#if CONFIG_HYBRIDTRANSFORM8X8
+  if (xd->mode_info_context->mbmi.txfm_size  == TX_8X8) {
+    if (xd->mode_info_context->mbmi.mode == I8X8_PRED)
+      tx_type = b->bmi.as_mode.tx_type;
+    return tx_type;
+  }
 #endif
+#if CONFIG_HYBRIDTRANSFORM
+  if (xd->mode_info_context->mbmi.txfm_size  == TX_4X4) {
+    if (xd->mode_info_context->mbmi.mode == B_PRED &&
+        xd->q_index < ACTIVE_HT)
+      tx_type = b->bmi.as_mode.tx_type;
+    return tx_type;
+  }
+#endif
+}
+#endif
+
 
 extern void vp8_build_block_doffsets(MACROBLOCKD *xd);
 extern void vp8_setup_block_dptrs(MACROBLOCKD *xd);
diff --git a/vp8/decoder/decodframe.c b/vp8/decoder/decodframe.c
index 5fb510b64..35b420e36 100644
--- a/vp8/decoder/decodframe.c
+++ b/vp8/decoder/decodframe.c
@@ -40,10 +40,6 @@
 #include <stdio.h>
 
 
-#ifdef DEC_DEBUG
-int dec_debug = 0;
-#endif
-
 #define COEFCOUNT_TESTING
 
 static int merge_index(int v, int n, int modulus) {
@@ -209,16 +205,6 @@ static void skip_recon_mb(VP8D_COMP *pbi, MACROBLOCKD *xd) {
     }
 #endif
   }
-#ifdef DEC_DEBUG
-  if (dec_debug) {
-    int i, j;
-    printf("Generating predictors\n");
-    for (i = 0; i < 16; i++) {
-      for (j = 0; j < 16; j++) printf("%3d ", xd->dst.y_buffer[i * xd->dst.y_stride + j]);
-      printf("\n");
-    }
-  }
-#endif
 
 }
 
@@ -228,7 +214,7 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
   int eobtotal = 0;
   MB_PREDICTION_MODE mode;
   int i;
-  int tx_type;
+  int tx_size;
 #if CONFIG_SUPERBLOCKS
   VP8_COMMON *pc = &pbi->common;
   int orig_skip_flag = xd->mode_info_context->mbmi.mb_skip_coeff;
@@ -255,14 +241,10 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
         xd->mode_info_context->mbmi.mode == NEARMV ||
         xd->mode_info_context->mbmi.mode == NEARESTMV)
       xd->mode_info_context->mbmi.txfm_size = TX_16X16;
-    else if (pbi->common.txfm_mode == ALLOW_8X8 &&
-        xd->mode_info_context->mbmi.mode != I8X8_PRED &&
-        xd->mode_info_context->mbmi.mode != B_PRED)
-#else
-      if (pbi->common.txfm_mode == ALLOW_8X8 &&
-          xd->mode_info_context->mbmi.mode != I8X8_PRED &&
-          xd->mode_info_context->mbmi.mode != B_PRED)
+    else
 #endif
+    if (pbi->common.txfm_mode == ALLOW_8X8 &&
+        xd->mode_info_context->mbmi.mode != B_PRED)
       xd->mode_info_context->mbmi.txfm_size = TX_8X8;
     else
       xd->mode_info_context->mbmi.txfm_size = TX_4X4;
@@ -272,37 +254,61 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
         xd->mode_info_context->mbmi.mode == NEWMV ||
         xd->mode_info_context->mbmi.mode == ZEROMV ||
         xd->mode_info_context->mbmi.mode == NEARMV ||
-        xd->mode_info_context->mbmi.mode == NEARESTMV) {
+        xd->mode_info_context->mbmi.mode == NEARESTMV)
       xd->mode_info_context->mbmi.txfm_size = TX_16X16;
-    } else if (pbi->common.txfm_mode == ALLOW_8X8 &&
-        xd->mode_info_context->mbmi.mode != I8X8_PRED &&
-        xd->mode_info_context->mbmi.mode != B_PRED &&
-        xd->mode_info_context->mbmi.mode != SPLITMV) {
-#else
+    else
+#endif
     if (pbi->common.txfm_mode == ALLOW_8X8 &&
-        xd->mode_info_context->mbmi.mode != I8X8_PRED &&
         xd->mode_info_context->mbmi.mode != B_PRED &&
         xd->mode_info_context->mbmi.mode != SPLITMV) {
-#endif
       xd->mode_info_context->mbmi.txfm_size = TX_8X8;
-    }
-    else {
+    } else {
       xd->mode_info_context->mbmi.txfm_size = TX_4X4;
     }
   }
 
-#if CONFIG_HYBRIDTRANSFORM8X8
-  if (xd->mode_info_context->mbmi.mode == I8X8_PRED) {
-    xd->mode_info_context->mbmi.txfm_size = TX_8X8;
-  }
-#endif
 #if CONFIG_SUPERBLOCKS
   if (xd->mode_info_context->mbmi.encoded_as_sb) {
     xd->mode_info_context->mbmi.txfm_size = TX_8X8;
   }
 #endif
 
-  tx_type = xd->mode_info_context->mbmi.txfm_size;
+  tx_size = xd->mode_info_context->mbmi.txfm_size;
+  mode = xd->mode_info_context->mbmi.mode;
+
+#if CONFIG_HYBRIDTRANSFORM
+  // parse transform types for intra 4x4 mode
+  QIndex = xd->q_index;
+  active_ht = (QIndex < ACTIVE_HT);
+  if (mode == B_PRED) {
+    for (i = 0; i < 16; i++) {
+      BLOCKD *b = &xd->block[i];
+      int b_mode = xd->mode_info_context->bmi[i].as_mode.first;
+      if(active_ht)
+        txfm_map(b, b_mode);
+    } // loop over 4x4 blocks
+  }
+#endif
+
+#if CONFIG_HYBRIDTRANSFORM8X8
+  if (mode == I8X8_PRED) {
+    for (i = 0; i < 4; i++) {
+      int ib = vp8_i8x8_block[i];
+      BLOCKD *b = &xd->block[ib];
+      int i8x8mode = b->bmi.as_mode.first;
+      txfm_map(b, pred_mode_conv(i8x8mode));
+    }
+  }
+#endif
+
+#if CONFIG_HYBRIDTRANSFORM16X16
+  active_ht16 = (QIndex < ACTIVE_HT16);
+  if (mode < I8X8_PRED) {
+    BLOCKD *b = &xd->block[0];
+    if(active_ht16)
+      txfm_map(b, pred_mode_conv(mode));
+  }
+#endif
 
   if (xd->mode_info_context->mbmi.mb_skip_coeff) {
     vp8_reset_mb_tokens_context(xd);
@@ -321,27 +327,16 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
       xd->eobs[i] = 0;
     }
 #if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
-    if (tx_type == TX_16X16)
+    if (tx_size == TX_16X16)
       eobtotal = vp8_decode_mb_tokens_16x16(pbi, xd);
     else
 #endif
-    if (tx_type == TX_8X8)
+    if (tx_size == TX_8X8)
       eobtotal = vp8_decode_mb_tokens_8x8(pbi, xd);
     else
       eobtotal = vp8_decode_mb_tokens(pbi, xd);
-#ifdef DEC_DEBUG
-    if (dec_debug) {
-      printf("\nTokens (%d)\n", eobtotal);
-      for (i = 0; i < 400; i++) {
-        printf("%3d ", xd->qcoeff[i]);
-        if (i % 16 == 15) printf("\n");
-      }
-      printf("\n");
-    }
-#endif
   }
 
-  mode = xd->mode_info_context->mbmi.mode;
 #if CONFIG_SWITCHABLE_INTERP
   if (pbi->common.frame_type != KEY_FRAME)
     vp8_setup_interp_filters(xd, xd->mode_info_context->mbmi.interp_filter,
@@ -366,39 +361,10 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
     }
   }
 
-#ifdef DEC_DEBUG
-  if (dec_debug) {
-    int i, j;
-    printf("Generating predictors\n");
-    for (i = 0; i < 16; i++) {
-      for (j = 0; j < 16; j++) printf("%3d ", xd->dst.y_buffer[i * xd->dst.y_stride + j]);
-      printf("\n");
-    }
-  }
-#endif
-
   // moved to be performed before detokenization
 //  if (xd->segmentation_enabled)
 //    mb_init_dequantizer(pbi, xd);
 
-#if CONFIG_HYBRIDTRANSFORM || CONFIG_HYBRIDTRANSFORM16X16
-  // parse transform types for intra 4x4 mode
-  QIndex = xd->q_index;
-  active_ht = (QIndex < ACTIVE_HT);
-  if (mode == B_PRED) {
-    for (i = 0; i < 16; i++) {
-      BLOCKD *b = &xd->block[i];
-      int b_mode = xd->mode_info_context->bmi[i].as_mode.first;
-      if(active_ht)
-        txfm_map(b, b_mode);
-    } // loop over 4x4 blocks
-  }
-#endif
-
-#if CONFIG_HYBRIDTRANSFORM16X16
-  active_ht16 = (QIndex < ACTIVE_HT16);
-#endif
-
   /* do prediction */
   if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) {
 #if CONFIG_SUPERBLOCKS
@@ -451,9 +417,6 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
       unsigned char *pre = xd->block[ib].predictor;
       unsigned char *dst = *(xd->block[ib].base_dst) + xd->block[ib].dst;
       int stride = xd->dst.y_stride;
-
-      tx_type = TX_4X4;
-      xd->mode_info_context->mbmi.txfm_size = TX_4X4;
 #endif
 
       b = &xd->block[ib];
@@ -462,7 +425,6 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
       (b, i8x8mode, b->predictor);
 
 #if CONFIG_HYBRIDTRANSFORM8X8
-      txfm_map(b, pred_mode_conv(i8x8mode));
       vp8_ht_dequant_idct_add_8x8_c(b->bmi.as_mode.tx_type,
                                     q, dq, pre, dst, 16, stride);
       q += 64;
@@ -546,7 +508,7 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
     BLOCKD *b = &xd->block[24];
 
 #if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
-    if (tx_type == TX_16X16) {
+    if (tx_size == TX_16X16) {
 #if CONFIG_HYBRIDTRANSFORM16X16
       if (mode < I8X8_PRED && active_ht16) {
         BLOCKD *bd = &xd->block[0];
@@ -570,7 +532,7 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
     }
     else
 #endif
-    if (tx_type == TX_8X8) {
+    if (tx_size == TX_8X8) {
 #if CONFIG_SUPERBLOCKS
       void *orig = xd->mode_info_context;
       int n, num = xd->mode_info_context->mbmi.encoded_as_sb ? 4 : 1;
@@ -598,16 +560,6 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
           continue; // only happens for SBs, which are already in dest buffer
 #endif
       DEQUANT_INVOKE(&pbi->dequant, block_2x2)(b);
-#ifdef DEC_DEBUG
-      if (dec_debug) {
-        int j;
-        printf("DQcoeff Haar\n");
-        for (j = 0; j < 16; j++) {
-          printf("%d ", b->dqcoeff[j]);
-        }
-        printf("\n");
-      }
-#endif
       IDCT_INVOKE(RTCD_VTABLE(idct), ihaar2)(&b->dqcoeff[0], b->diff, 8);
       ((int *)b->qcoeff)[0] = 0;// 2nd order block are set to 0 after inverse transform
       ((int *)b->qcoeff)[1] = 0;
@@ -665,20 +617,22 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
 #if CONFIG_SUPERBLOCKS
     if (!xd->mode_info_context->mbmi.encoded_as_sb) {
 #endif
-      if (tx_type == TX_8X8
+      if (xd->mode_info_context->mbmi.mode != I8X8_PRED) {
+        if (tx_size == TX_8X8
 #if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
-      || tx_type == TX_16X16
+            || tx_size == TX_16X16
 #endif
-      )
-    DEQUANT_INVOKE(&pbi->dequant, idct_add_uv_block_8x8) //
-    (xd->qcoeff + 16 * 16, xd->block[16].dequant,
-     xd->predictor + 16 * 16, xd->dst.u_buffer, xd->dst.v_buffer,
-     xd->dst.uv_stride, xd->eobs + 16, xd); //
-  else if (xd->mode_info_context->mbmi.mode != I8X8_PRED)
-    DEQUANT_INVOKE(&pbi->dequant, idct_add_uv_block)
-    (xd->qcoeff + 16 * 16, xd->block[16].dequant,
-     xd->predictor + 16 * 16, xd->dst.u_buffer, xd->dst.v_buffer,
-     xd->dst.uv_stride, xd->eobs + 16);
+           )
+          DEQUANT_INVOKE(&pbi->dequant, idct_add_uv_block_8x8) //
+              (xd->qcoeff + 16 * 16, xd->block[16].dequant,
+               xd->predictor + 16 * 16, xd->dst.u_buffer, xd->dst.v_buffer,
+               xd->dst.uv_stride, xd->eobs + 16, xd); //
+        else
+          DEQUANT_INVOKE(&pbi->dequant, idct_add_uv_block)
+              (xd->qcoeff + 16 * 16, xd->block[16].dequant,
+               xd->predictor + 16 * 16, xd->dst.u_buffer, xd->dst.v_buffer,
+               xd->dst.uv_stride, xd->eobs + 16);
+      }
 #if CONFIG_SUPERBLOCKS
   }
 #endif
@@ -753,10 +707,6 @@ decode_sb_row(VP8D_COMP *pbi, VP8_COMMON *pc, int mbrow, MACROBLOCKD *xd) {
         continue;
       }
 
-#ifdef DEC_DEBUG
-      dec_debug = (pc->current_video_frame == 0 && mb_row == 0 && mb_col == 0);
-#endif
-
       // Set above context pointer
       xd->above_context = pc->above_context + mb_col;
       xd->left_context = pc->left_context + (i >> 1);
diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c
index bbbbc7870..e25b64305 100644
--- a/vp8/encoder/encodeframe.c
+++ b/vp8/encoder/encodeframe.c
@@ -1833,7 +1833,6 @@ void vp8cx_encode_intra_macro_block(VP8_COMP *cpi,
   else
 #endif
   if (cpi->common.txfm_mode == ALLOW_8X8
-      && mbmi->mode != I8X8_PRED
       && mbmi->mode != B_PRED) {
     mbmi->txfm_size = TX_8X8;
     cpi->t8x8_count++;
@@ -1930,7 +1929,6 @@ void vp8cx_encode_inter_macroblock (VP8_COMP *cpi, MACROBLOCK *x,
   } else
 #endif
   if (cpi->common.txfm_mode == ALLOW_8X8
-      && mbmi->mode != I8X8_PRED
       && mbmi->mode != B_PRED
       && mbmi->mode != SPLITMV) {
     mbmi->txfm_size = TX_8X8;
@@ -2111,7 +2109,6 @@ void vp8cx_encode_inter_superblock(VP8_COMP *cpi, MACROBLOCK *x, TOKENEXTRA **t,
 
   /* test code: set transform size based on mode selection */
   if (cpi->common.txfm_mode == ALLOW_8X8
-      && x->e_mbd.mode_info_context->mbmi.mode != I8X8_PRED
       && x->e_mbd.mode_info_context->mbmi.mode != B_PRED
       && x->e_mbd.mode_info_context->mbmi.mode != SPLITMV) {
     x->e_mbd.mode_info_context->mbmi.txfm_size = TX_8X8;
diff --git a/vp8/encoder/tokenize.c b/vp8/encoder/tokenize.c
index d0c315283..5b1f8ffb4 100644
--- a/vp8/encoder/tokenize.c
+++ b/vp8/encoder/tokenize.c
@@ -42,6 +42,8 @@ void vp8_stuff_mb(VP8_COMP *cpi,
                   MACROBLOCKD *xd, TOKENEXTRA **t, int dry_run);
 void vp8_stuff_mb_8x8(VP8_COMP *cpi,
                       MACROBLOCKD *xd, TOKENEXTRA **t, int dry_run);
+void vp8_stuff_mb_8x8_4x4uv(VP8_COMP *cpi,
+                            MACROBLOCKD *xd, TOKENEXTRA **t, int dry_run);
 #if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
 void vp8_stuff_mb_16x16(VP8_COMP *cpi, MACROBLOCKD *xd,
                         TOKENEXTRA **t, int dry_run);
@@ -764,6 +766,10 @@ int mb_is_skippable_8x8(MACROBLOCKD *xd) {
   return (mby_is_skippable_8x8(xd) & mbuv_is_skippable_8x8(xd));
 }
 
+int mb_is_skippable_8x8_4x4uv(MACROBLOCKD *xd) {
+  return (mby_is_skippable_8x8(xd) & mbuv_is_skippable(xd));
+}
+
 #if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
 int mby_is_skippable_16x16(MACROBLOCKD *xd) {
   int skip = 1;
@@ -822,7 +828,10 @@ void vp8_tokenize_mb(VP8_COMP *cpi,
       break;
 #endif
     case TX_8X8:
-      xd->mode_info_context->mbmi.mb_skip_coeff = mb_is_skippable_8x8(xd);
+      if (xd->mode_info_context->mbmi.mode == I8X8_PRED)
+        xd->mode_info_context->mbmi.mb_skip_coeff = mb_is_skippable_8x8_4x4uv(xd);
+      else
+        xd->mode_info_context->mbmi.mb_skip_coeff = mb_is_skippable_8x8(xd);
       break;
     default:
       xd->mode_info_context->mbmi.mb_skip_coeff = mb_is_skippable(xd, has_y2_block);
@@ -838,9 +847,12 @@ void vp8_tokenize_mb(VP8_COMP *cpi,
         vp8_stuff_mb_16x16(cpi, xd, t, dry_run);
       else
 #endif
-      if (tx_type == TX_8X8)
-        vp8_stuff_mb_8x8(cpi, xd, t, dry_run);
-      else
+      if (tx_type == TX_8X8) {
+        if (xd->mode_info_context->mbmi.mode == I8X8_PRED)
+          vp8_stuff_mb_8x8_4x4uv(cpi, xd, t, dry_run);
+        else
+          vp8_stuff_mb_8x8(cpi, xd, t, dry_run);
+      } else
         vp8_stuff_mb(cpi, xd, t, dry_run);
     } else {
       vp8_fix_contexts(xd);
@@ -895,6 +907,10 @@ void vp8_tokenize_mb(VP8_COMP *cpi,
   if (tx_type == TX_8X8) {
     ENTROPY_CONTEXT *A = (ENTROPY_CONTEXT *)xd->above_context;
     ENTROPY_CONTEXT *L = (ENTROPY_CONTEXT *)xd->left_context;
+#if CONFIG_HYBRIDTRANSFORM8X8
+    if (xd->mode_info_context->mbmi.mode == I8X8_PRED)
+      plane_type = PLANE_TYPE_Y_WITH_DC;
+#endif
     for (b = 0; b < 16; b += 4) {
       tokenize1st_order_b_8x8(xd,
                               xd->block + b, t, plane_type, xd->frame_type,
@@ -904,21 +920,34 @@ void vp8_tokenize_mb(VP8_COMP *cpi,
       *(A + vp8_block2above_8x8[b] + 1) = *(A + vp8_block2above_8x8[b]);
       *(L + vp8_block2left_8x8[b] + 1)  = *(L + vp8_block2left_8x8[b]);
     }
-    for (b = 16; b < 24; b += 4) {
-      tokenize1st_order_b_8x8(xd,
-                              xd->block + b, t, 2, xd->frame_type,
-                              A + vp8_block2above_8x8[b],
-                              L + vp8_block2left_8x8[b],
-                              cpi, dry_run);
-      *(A + vp8_block2above_8x8[b] + 1) = *(A + vp8_block2above_8x8[b]);
-      *(L + vp8_block2left_8x8[b] + 1) = *(L + vp8_block2left_8x8[b]);
+#if CONFIG_HYBRIDTRANSFORM8X8
+    if (xd->mode_info_context->mbmi.mode == I8X8_PRED) {
+        tokenize1st_order_chroma(xd, t, PLANE_TYPE_UV, cpi, dry_run);
+    } else
+#endif
+    {
+      for (b = 16; b < 24; b += 4) {
+        tokenize1st_order_b_8x8(xd,
+                                xd->block + b, t, 2, xd->frame_type,
+                                A + vp8_block2above_8x8[b],
+                                L + vp8_block2left_8x8[b],
+                                cpi, dry_run);
+        *(A + vp8_block2above_8x8[b] + 1) = *(A + vp8_block2above_8x8[b]);
+        *(L + vp8_block2left_8x8[b] + 1) = *(L + vp8_block2left_8x8[b]);
+      }
     }
   } else {
 #if CONFIG_HYBRIDTRANSFORM
     if(active_ht) {
       tokenize1st_order_ht(xd, t, plane_type, cpi, dry_run);
-    } else {
+    } else
+#endif
+    {
+      tokenize1st_order_b(xd, t, plane_type, cpi, dry_run);
+    }
+  }
 
+  /*
 #if CONFIG_HYBRIDTRANSFORM8X8
       if (xd->mode_info_context->mbmi.mode == I8X8_PRED) {
         ENTROPY_CONTEXT *A = (ENTROPY_CONTEXT *)xd->above_context;
@@ -946,6 +975,7 @@ void vp8_tokenize_mb(VP8_COMP *cpi,
     tokenize1st_order_b(xd, t, plane_type, cpi, dry_run);
 #endif
   }
+  */
   if (dry_run)
     *t = t_backup;
 }
@@ -1341,7 +1371,6 @@ void vp8_stuff_mb_8x8(VP8_COMP *cpi,
     *t = t_backup;
 }
 
-
 #if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
 static __inline
 void stuff1st_order_b_16x16(const BLOCKD *const b,
@@ -1493,6 +1522,40 @@ void vp8_stuff_mb(VP8_COMP *cpi, MACROBLOCKD *xd,
   if (dry_run)
     *t = t_backup;
 }
+
+void vp8_stuff_mb_8x8_4x4uv(VP8_COMP *cpi,
+                            MACROBLOCKD *xd,
+                            TOKENEXTRA **t,
+                            int dry_run) {
+  ENTROPY_CONTEXT *A = (ENTROPY_CONTEXT *)xd->above_context;
+  ENTROPY_CONTEXT *L = (ENTROPY_CONTEXT *)xd->left_context;
+  int plane_type;
+  int b;
+  TOKENEXTRA *t_backup = *t;
+
+  stuff2nd_order_b_8x8(xd->block + 24, t, 1, xd->frame_type,
+                       A + vp8_block2above_8x8[24],
+                       L + vp8_block2left_8x8[24], cpi, dry_run);
+  plane_type = 0;
+
+  for (b = 0; b < 16; b += 4) {
+    stuff1st_order_b_8x8(xd->block + b, t, plane_type, xd->frame_type,
+                         A + vp8_block2above_8x8[b],
+                         L + vp8_block2left_8x8[b],
+                         cpi, dry_run);
+    *(A + vp8_block2above_8x8[b] + 1) = *(A + vp8_block2above_8x8[b]);
+    *(L + vp8_block2left_8x8[b] + 1)  = *(L + vp8_block2left_8x8[b]);
+  }
+  for (b = 16; b < 24; b++)
+    stuff1st_order_buv(t,
+                       A + vp8_block2above[b],
+                       L + vp8_block2left[b],
+                       cpi, dry_run);
+
+  if (dry_run)
+    *t = t_backup;
+}
+
 void vp8_fix_contexts(MACROBLOCKD *xd) {
   /* Clear entropy contexts for Y2 blocks */
   if ((xd->mode_info_context->mbmi.mode != B_PRED