--- decoder.c	2004/07/26 19:32:28	1.64
+++ decoder.c	2004/08/15 11:28:38	1.66
@@ -20,7 +20,7 @@
  *  along with this program ; if not, write to the Free Software
  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
  *
- * $Id: decoder.c,v 1.64 2004/07/26 19:32:28 edgomez Exp $
+ * $Id: decoder.c,v 1.66 2004/08/15 11:28:38 syskin Exp $
  *
  ****************************************************************************/
 
@@ -62,11 +62,6 @@
 #include "image/postprocessing.h"
 #include "utils/mem_align.h"
 
-#ifdef ARCH_IS_IA32
-#define interpolate16x16_quarterpel new_interpolate16x16_quarterpel
-#define interpolate8x8_quarterpel new_interpolate8x8_quarterpel
-#endif 
-
 static int
 decoder_resize(DECODER * dec)
 {
@@ -470,6 +465,42 @@
 	}
 }
 
+static void
+validate_vector(VECTOR * mv, unsigned int x_pos, unsigned int y_pos, const DECODER * dec)
+{
+	/* clip a vector to valid range 
+	   prevents crashes if bitstream is broken 
+	*/
+	int i;
+
+	for (i = 0; i < 4; i++) {
+		
+		int border = (int)(dec->mb_width - x_pos) << (5 + dec->quarterpel);
+		if (mv[i].x > border) {
+			DPRINTF(XVID_DEBUG_MV, "mv.x > max -- %d > %d, MB %d, %d", mv[i].x, border, x_pos, y_pos);
+			mv[i].x = border;
+		} else {
+			border = (-(int)x_pos-1) << (5 + dec->quarterpel);
+			if (mv[i].x < border) {
+				DPRINTF(XVID_DEBUG_MV, "mv.x < min -- %d < %d, MB %d, %d", mv[i].x, border, x_pos, y_pos);
+				mv[i].x = border;
+			}
+		}
+
+		border = (int)(dec->mb_height - y_pos) << (5 + dec->quarterpel);
+		if (mv[i].y >  border) {
+			DPRINTF(XVID_DEBUG_MV, "mv.y > max -- %d > %d, MB %d, %d", mv[i].y, border, x_pos, y_pos);
+			mv[i].y = border;
+		} else {
+			border = (-(int)y_pos-1) << (5 + dec->quarterpel);
+			if (mv[i].y < border) {
+				DPRINTF(XVID_DEBUG_MV, "mv.y < min -- %d < %d, MB %d, %d", mv[i].y, border, x_pos, y_pos);
+				mv[i].y = border;
+			}
+		}
+	}
+}
+
 /* decode an inter macroblock */
 static void
 decoder_mbinter(DECODER * dec,
@@ -507,32 +538,7 @@
 			mv[i] = pMB->mvs[i];
 	}
 
-	for (i = 0; i < 4; i++) {
-		/* clip to valid range */
-		int border = (int)(dec->mb_width - x_pos) << (5 + dec->quarterpel);
-		if (mv[i].x > border) {
-			DPRINTF(XVID_DEBUG_MV, "mv.x > max -- %d > %d, MB %d, %d", mv[i].x, border, x_pos, y_pos);
-			mv[i].x = border;
-		} else {
-			border = (-(int)x_pos-1) << (5 + dec->quarterpel);
-			if (mv[i].x < border) {
-				DPRINTF(XVID_DEBUG_MV, "mv.x < min -- %d < %d, MB %d, %d", mv[i].x, border, x_pos, y_pos);
-				mv[i].x = border;
-			}
-		}
-
-		border = (int)(dec->mb_height - y_pos) << (5 + dec->quarterpel);
-		if (mv[i].y >  border) {
-			DPRINTF(XVID_DEBUG_MV, "mv.y > max -- %d > %d, MB %d, %d", mv[i].y, border, x_pos, y_pos);
-			mv[i].y = border;
-		} else {
-			border = (-(int)y_pos-1) << (5 + dec->quarterpel);
-			if (mv[i].y < border) {
-				DPRINTF(XVID_DEBUG_MV, "mv.y < min -- %d < %d, MB %d, %d", mv[i].y, border, x_pos, y_pos);
-				mv[i].y = border;
-			}
-		}
-	}
+	validate_vector(mv, x_pos, y_pos, dec);
 
 	start_timer();
 
@@ -1022,7 +1028,7 @@
 decoder_bf_interpolate_mbinter(DECODER * dec,
 								IMAGE forward,
 								IMAGE backward,
-								const MACROBLOCK * pMB,
+								MACROBLOCK * pMB,
 								const uint32_t x_pos,
 								const uint32_t y_pos,
 								Bitstream * bs,
@@ -1039,6 +1045,9 @@
 	pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3);
 	pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3);
 
+	validate_vector(pMB->mvs, x_pos, y_pos, dec);
+	validate_vector(pMB->b_mvs, x_pos, y_pos, dec);
+
 	if (!direct) {
 		uv_dx = pMB->mvs[0].x;
 		uv_dy = pMB->mvs[0].y;
@@ -1115,68 +1124,38 @@
 
 	if(dec->quarterpel) {
 		if(!direct) {
-			interpolate16x16_quarterpel(dec->tmp.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
- 										dec->qtmp.y + 128, 16*x_pos, 16*y_pos,
-										pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0);
+			interpolate16x16_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
+					dec->qtmp.y + 128, 16*x_pos, 16*y_pos,
+					pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0);
 		} else {
-			interpolate8x8_quarterpel(dec->tmp.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
- 										dec->qtmp.y + 128, 16*x_pos, 16*y_pos,
-										pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0);
-			interpolate8x8_quarterpel(dec->tmp.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
- 										dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos,
-										pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0);
-			interpolate8x8_quarterpel(dec->tmp.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
- 										dec->qtmp.y + 128, 16*x_pos, 16*y_pos + 8,
-										pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0);
-			interpolate8x8_quarterpel(dec->tmp.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
- 										dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos + 8,
-										pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0);
+			interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
+					dec->qtmp.y + 128, 16*x_pos, 16*y_pos,
+					pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0);
+			interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
+					dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos,
+					pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0);
+			interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
+					dec->qtmp.y + 128, 16*x_pos, 16*y_pos + 8,
+					pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0);
+			interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64,
+					dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos + 8,
+					pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0);
 		}
 	} else {
-		interpolate8x8_switch(dec->tmp.y, backward.y, 16 * x_pos, 16 * y_pos,
-							pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0);
-		interpolate8x8_switch(dec->tmp.y, backward.y, 16 * x_pos + 8,
-							16 * y_pos, pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0);
-		interpolate8x8_switch(dec->tmp.y, backward.y, 16 * x_pos,
-							16 * y_pos + 8, pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0);
-		interpolate8x8_switch(dec->tmp.y, backward.y, 16 * x_pos + 8,
-							16 * y_pos + 8, pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0);
-	}
-
-	interpolate8x8_switch(dec->tmp.u, backward.u, 8 * x_pos, 8 * y_pos,
-						b_uv_dx, b_uv_dy, stride2, 0);
-	interpolate8x8_switch(dec->tmp.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->tmp.y + (16 * y_pos * stride) + 16 * x_pos,
-						stride, 0, 8);
-
-	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->tmp.y + (16 * y_pos * stride) + 16 * x_pos + 8,
-						stride, 0, 8);
-
-	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->tmp.y + ((16 * y_pos + 8) * stride) + 16 * x_pos,
-						stride, 0, 8);
-
-	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->tmp.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8,
-						stride, 0, 8);
-
-	interpolate8x8_avg2(dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos,
-						dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos,
-						dec->tmp.u + (8 * y_pos * stride2) + 8 * x_pos,
-						stride2, 0, 8);
-
-	interpolate8x8_avg2(dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos,
-						dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos,
-						dec->tmp.v + (8 * y_pos * stride2) + 8 * x_pos,
-						stride2, 0, 8);
+		interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos, 16 * y_pos,
+				pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0);
+		interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos + 8,
+				16 * y_pos, pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0);
+		interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos,
+				16 * y_pos + 8, pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0);
+		interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos + 8,
+				16 * y_pos + 8, pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0);
+	}
+
+	interpolate8x8_add_switch(dec->cur.u, backward.u, 8 * x_pos, 8 * y_pos,
+			b_uv_dx, b_uv_dy, stride2, 0);
+	interpolate8x8_add_switch(dec->cur.v, backward.v, 8 * x_pos, 8 * y_pos,
+			b_uv_dx, b_uv_dy, stride2, 0);
 
 	stop_comp_timer();
 
@@ -1416,7 +1395,6 @@
 	}
 }
 
-
 int
 decoder_decode(DECODER * dec,
 				xvid_dec_frame_t * frame, xvid_dec_stats_t * stats)