--- decoder.c	2002/10/30 18:01:48	1.37.2.6
+++ decoder.c	2002/09/21 03:07:56	1.38
@@ -53,9 +53,9 @@
  *  26.03.2002  interlacing support - moved transfers outside decode loop
  *  26.12.2001  decoder_mbinter: dequant/idct moved within if(coded) block
  *  22.12.2001  lock based interpolation
- *  01.12.2001  inital version; (c)2001 peter ross <pross@cs.rmit.edu.au>
+ *  01.12.2001  inital version; (c)2001 peter ross <pross@xvid.org>
  *
- *  $Id: decoder.c,v 1.37.2.6 2002/10/30 18:01:48 Isibaar Exp $
+ *  $Id: decoder.c,v 1.38 2002/09/21 03:07:56 suxen_drol Exp $
  *
  *************************************************************************/
 
@@ -84,7 +84,6 @@
 #include "prediction/mbprediction.h"
 #include "utils/timer.h"
 #include "utils/emms.h"
-#include "motion/motion.h"
 
 #include "image/image.h"
 #include "image/colorspace.h"
@@ -137,15 +136,6 @@
 		return XVID_ERR_MEMORY;
 	}
 
-	if (image_create(&dec->refh, dec->edged_width, dec->edged_height)) {
-		image_destroy(&dec->cur, dec->edged_width, dec->edged_height);
-		image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height);
-		image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height);
-		image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height);
-		xvid_free(dec);
-		return XVID_ERR_MEMORY;
-	}
-
 	dec->mbs =
 		xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height,
 					CACHE_LINE);
@@ -154,7 +144,6 @@
 		image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height);
 		image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height);
 		image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height);
-		image_destroy(&dec->refh, dec->edged_width, dec->edged_height);
 		xvid_free(dec);
 		return XVID_ERR_MEMORY;
 	}
@@ -172,7 +161,6 @@
 		image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height);
 		image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height);
 		image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height);
-		image_destroy(&dec->refh, dec->edged_width, dec->edged_height);
 		xvid_free(dec);
 		return XVID_ERR_MEMORY;
 	}
@@ -198,7 +186,6 @@
 	image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height);
 	image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height);
 	image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height);
-	image_destroy(&dec->refh, dec->edged_width, dec->edged_height);
 	image_destroy(&dec->cur, dec->edged_width, dec->edged_height);
 	xvid_free(dec);
 
@@ -281,10 +268,8 @@
 		start_timer();
 		if (cbp & (1 << (5 - i)))	// coded
 		{
-			int direction = dec->alternate_vertical_scan ?
-				2 : pMB->acpred_directions[i];
-
-			get_intra_block(bs, &block[i * 64], direction, start_coeff);
+			get_intra_block(bs, &block[i * 64], pMB->acpred_directions[i],
+							start_coeff);
 		}
 		stop_coding_timer();
 
@@ -326,6 +311,9 @@
 
 #define SIGN(X) (((X)>0)?1:-1)
 #define ABS(X) (((X)>0)?(X):-(X))
+static const uint32_t roundtab[16] =
+	{ 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2 };
+
 
 // decode an inter macroblock
 
@@ -362,96 +350,69 @@
 
 		if (dec->quarterpel)
 		{
-			uv_dx = (uv_dx >> 2) + roundtab_78[uv_dx & 0x7];
-			uv_dy = (uv_dy >> 2) + roundtab_78[uv_dy & 0x7];
-		}
-		else {
-			uv_dx = (uv_dx >> 1) + roundtab_79[uv_dx & 0x3];
-			uv_dy = (uv_dy >> 1) + roundtab_79[uv_dy & 0x3];
-		}
-
-		start_timer();
-		if(dec->quarterpel) {
-			interpolate16x16_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64,
-	 								    dec->refh.y + 128, 16*x_pos, 16*y_pos,
-									    pMB->mvs[0].x, pMB->mvs[0].y, stride,  rounding);
-		}
-		else {
-			interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos,
-								  pMB->mvs[0].x, pMB->mvs[0].y, stride,  rounding);
-			interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos,
-							      pMB->mvs[1].x, pMB->mvs[1].y, stride,  rounding);
-			interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos + 8,
-								  pMB->mvs[2].x, pMB->mvs[2].y, stride,  rounding);
-			interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos + 8, 
-								  pMB->mvs[3].x, pMB->mvs[3].y, stride,  rounding);
+			uv_dx = (uv_dx >> 1) | (uv_dx & 1);
+			uv_dy = (uv_dy >> 1) | (uv_dy & 1);
 		}
 
-		interpolate8x8_switch(dec->cur.u, dec->refn[0].u, 8 * x_pos, 8 * y_pos,
-							  uv_dx, uv_dy, stride2, rounding);
-		interpolate8x8_switch(dec->cur.v, dec->refn[0].v, 8 * x_pos, 8 * y_pos,
-							  uv_dx, uv_dy, stride2, rounding);
-		stop_comp_timer();
-
+		uv_dx = (uv_dx & 3) ? (uv_dx >> 1) | 1 : uv_dx / 2;
+		uv_dy = (uv_dy & 3) ? (uv_dy >> 1) | 1 : uv_dy / 2;
 	} else {
 		int sum;
-		
-		if(dec->quarterpel)
-			sum = (pMB->mvs[0].x / 2) + (pMB->mvs[1].x / 2) + (pMB->mvs[2].x / 2) + (pMB->mvs[3].x / 2);
-		else
-			sum = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x;
-
-		uv_dx = (sum >> 3) + roundtab_76[sum & 0xf];
-
-		if(dec->quarterpel)
-			sum = (pMB->mvs[0].y / 2) + (pMB->mvs[1].y / 2) + (pMB->mvs[2].y / 2) + (pMB->mvs[3].y / 2);
-		else
-			sum = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y;
+		sum = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x;
+
+		if (dec->quarterpel)
+		{
+			sum /= 2;
+		}
 
-		uv_dy = (sum >> 3) + roundtab_76[sum & 0xf];
+		uv_dx = (sum == 0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + (ABS(sum) / 16) * 2));
 
-		start_timer();
-		if(dec->quarterpel) {
-			interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64,
-									  dec->refh.y + 128, 16*x_pos, 16*y_pos,
-									  pMB->mvs[0].x, pMB->mvs[0].y, stride,  rounding);
-			interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64,
-									  dec->refh.y + 128, 16*x_pos + 8, 16*y_pos,
-									  pMB->mvs[1].x, pMB->mvs[1].y, stride,  rounding);
-			interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64,
-									  dec->refh.y + 128, 16*x_pos, 16*y_pos + 8,
-									  pMB->mvs[2].x, pMB->mvs[2].y, stride,  rounding);
-			interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64,
-									  dec->refh.y + 128, 16*x_pos + 8, 16*y_pos + 8,
-									  pMB->mvs[3].x, pMB->mvs[3].y, stride,  rounding);
+		sum = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y;
+
+		if (dec->quarterpel)
+		{
+			sum /= 2;
 		}
-		else {
-			interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos,
+
+		uv_dy = (sum == 0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + (ABS(sum) / 16) * 2));
+	}
+
+	start_timer();
+	if(dec->quarterpel) {
+		DEBUG("QUARTERPEL");
+		interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos,
 								  pMB->mvs[0].x, pMB->mvs[0].y, stride,  rounding);
-			interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos,
+		interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos,
 								  pMB->mvs[1].x, pMB->mvs[1].y, stride,  rounding);
-			interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos + 8,
+		interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos + 8,
 								  pMB->mvs[2].x, pMB->mvs[2].y, stride,  rounding);
-			interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos + 8, 
+		interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos + 8,
 								  pMB->mvs[3].x, pMB->mvs[3].y, stride,  rounding);
-		}
-
-		interpolate8x8_switch(dec->cur.u, dec->refn[0].u, 8 * x_pos, 8 * y_pos,
-							  uv_dx, uv_dy, stride2, rounding);
-		interpolate8x8_switch(dec->cur.v, dec->refn[0].v, 8 * x_pos, 8 * y_pos,
-							  uv_dx, uv_dy, stride2, rounding);
-		stop_comp_timer();
+	}
+	else {
+		interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos,
+							  pMB->mvs[0].x, pMB->mvs[0].y, stride,  rounding);
+		interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos,
+							  pMB->mvs[1].x, pMB->mvs[1].y, stride,  rounding);
+		interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos + 8,
+							  pMB->mvs[2].x, pMB->mvs[2].y, stride,  rounding);
+		interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos + 8, 
+							  pMB->mvs[3].x, pMB->mvs[3].y, stride,  rounding);
 	}
 
-	for (i = 0; i < 6; i++) {
-		int direction = dec->alternate_vertical_scan ? 2 : 0;
+	interpolate8x8_switch(dec->cur.u, dec->refn[0].u, 8 * x_pos, 8 * y_pos,
+						  uv_dx, uv_dy, stride2, rounding);
+	interpolate8x8_switch(dec->cur.v, dec->refn[0].v, 8 * x_pos, 8 * y_pos,
+						  uv_dx, uv_dy, stride2, rounding);
+	stop_comp_timer();
 
+	for (i = 0; i < 6; i++) {
 		if (cbp & (1 << (5 - i)))	// coded
 		{
 			memset(&block[i * 64], 0, 64 * sizeof(int16_t));	// clear
 
 			start_timer();
-			get_inter_block(bs, &block[i * 64], direction);
+			get_inter_block(bs, &block[i * 64]);
 			stop_coding_timer();
 
 			start_timer();
@@ -624,7 +585,7 @@
 
 	start_timer();
 	image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height,
-				   dec->width, dec->height);
+				   dec->width, dec->height, dec->interlacing);
 	stop_edges_timer();
 
 	bound = 0;
@@ -745,6 +706,7 @@
 								rounding);
 			} else				// not coded
 			{
+				DEBUG2("P-frame MB at (X,Y)=",x,y);
 				mb->mode = MODE_NOT_CODED;
 				mb->mvs[0].x = mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = 0;
 				mb->mvs[0].y = mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = 0;
@@ -914,14 +876,12 @@
 	stop_comp_timer();
 
 	for (i = 0; i < 6; i++) {
-		int direction = dec->alternate_vertical_scan ? 2 : 0;
-
 		if (cbp & (1 << (5 - i)))	// coded
 		{
 			memset(&block[i * 64], 0, 64 * sizeof(int16_t));	// clear
 
 			start_timer();
-			get_inter_block(bs, &block[i * 64], direction);
+			get_inter_block(bs, &block[i * 64]);
 			stop_coding_timer();
 
 			start_timer();
@@ -959,6 +919,7 @@
 	stop_transfer_timer();
 }
 
+
 // add by MinChen <chenm001@163.com>
 // decode an B-frame direct &  inter macroblock
 void
@@ -1066,47 +1027,27 @@
 	interpolate8x8_switch(dec->refn[2].v, backward.v, 8 * x_pos, 8 * y_pos,
 						  b_uv_dx, b_uv_dy, stride2, 0);
 
-	interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos,
-						dec->cur.y + (16 * y_pos * stride) + 16 * x_pos,
-						dec->refn[2].y + (16 * y_pos * stride) + 16 * x_pos,
-						stride, 0);
-
-	interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8,
-						dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8,
-						dec->refn[2].y + (16 * y_pos * stride) + 16 * x_pos + 8,
-						stride, 0);
-
-	interpolate8x8_avg2(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos,
-						dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos,
-						dec->refn[2].y + ((16 * y_pos + 8) * stride) + 16 * x_pos,
-						stride, 0);
-
-	interpolate8x8_avg2(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8,
-						dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8,
-						dec->refn[2].y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8,
-						stride, 0);
-
-	interpolate8x8_avg2(dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos,
-						dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos,
-						dec->refn[2].u + (8 * y_pos * stride2) + 8 * x_pos,
-						stride2, 0);
-
-	interpolate8x8_avg2(dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos,
-						dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos,
-						dec->refn[2].v + (8 * y_pos * stride2) + 8 * x_pos,
-						stride2, 0);
-
+	interpolate8x8_c(dec->cur.y, dec->refn[2].y, 16 * x_pos, 16 * y_pos,
+					 stride);
+	interpolate8x8_c(dec->cur.y, dec->refn[2].y, 16 * x_pos + 8, 16 * y_pos,
+					 stride);
+	interpolate8x8_c(dec->cur.y, dec->refn[2].y, 16 * x_pos, 16 * y_pos + 8,
+					 stride);
+	interpolate8x8_c(dec->cur.y, dec->refn[2].y, 16 * x_pos + 8,
+					 16 * y_pos + 8, stride);
+	interpolate8x8_c(dec->cur.u, dec->refn[2].u, 8 * x_pos, 8 * y_pos,
+					 stride2);
+	interpolate8x8_c(dec->cur.v, dec->refn[2].v, 8 * x_pos, 8 * y_pos,
+					 stride2);
 	stop_comp_timer();
 
 	for (i = 0; i < 6; i++) {
-		int direction = dec->alternate_vertical_scan ? 2 : 0;
-
 		if (cbp & (1 << (5 - i)))	// coded
 		{
 			memset(&block[i * 64], 0, 64 * sizeof(int16_t));	// clear
 
 			start_timer();
-			get_inter_block(bs, &block[i * 64], direction);
+			get_inter_block(bs, &block[i * 64]);
 			stop_coding_timer();
 
 			start_timer();
@@ -1201,9 +1142,9 @@
 
 	start_timer();
 	image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height,
-				   dec->width, dec->height);
+				   dec->width, dec->height, dec->interlacing);
 	image_setedges(&dec->refn[1], dec->edged_width, dec->edged_height,
-				   dec->width, dec->height);
+				   dec->width, dec->height, dec->interlacing);
 	stop_edges_timer();
 
 #ifdef BFRAMES_DEC_DEBUG
@@ -1381,9 +1322,6 @@
 
 	BitstreamInit(&bs, frame->bitstream, frame->length);
 
-	if(BitstreamShowBits(&bs, 8) == 0x7f)
-		return XVID_ERR_OK;
-
 	// add by chenm001 <chenm001@163.com>
 	// for support B-frame to reference last 2 frame
 	dec->frames++;
@@ -1425,9 +1363,6 @@
 	case N_VOP:				// vop not coded
 		// when low_delay==0, N_VOP's should interpolate between the past and future frames
 		image_copy(&dec->cur, &dec->refn[0], dec->edged_width, dec->height);
-#ifdef BFRAMES_DEC
-		DEBUG1("N_VOP  Time=", dec->time);
-#endif
 		break;
 
 	default: