Merge remote branch 'qatar/master'

* qatar/master: (23 commits)
  doc: Check standalone compilation before submitting new components.
  Fix standalone compilation of pipe protocol.
  Fix standalone compilation of ac3_fixed encoder.
  Fix standalone compilation of binkaudio_dct / binkaudio_rdft decoders.
  Fix standalone compilation of IMC decoder.
  Fix standalone compilation of WTV demuxer.
  Fix standalone compilation of MXPEG decoder.
  flashsv: K&R cosmetics
  matroskaenc: fix memory leak
  vc1: make overlap filter for I-frames bit-exact.
  vc1dec: use s->start/end_mb_y instead of passing them as function args.
  Revert "VC1: merge idct8x8, coeff adjustments and put_pixels."
  Replace strncpy() with av_strlcpy().
  indeo3: Eliminate use of long.
  get_bits: make cache unsigned to eliminate undefined signed overflow.
  asfdec: fix assert failure on invalid files
  avfilter: check malloc return values.
Not pulled as reason for reindent is not pulled:  mpegvideo: reindent.
  nutenc: check malloc return values.
Not pulled due to much simpler solution in ffmpeg  *: don't av_malloc(0).
  ...

Conflicts:
	doc/developer.texi
	libavcodec/Makefile
	libavcodec/get_bits.h
	libavcodec/mpegvideo.c
	libavformat/Makefile
	libavutil/log.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>

12 files changed, 398 insertions, 275 deletions

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 6c75b6510b..a4b6f9a406 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -277,7 +277,7 @@ OBJS-$(CONFIG_MSRLE_DECODER)           += msrle.o msrledec.o
 OBJS-$(CONFIG_MSVIDEO1_DECODER)        += msvideo1.o
 OBJS-$(CONFIG_MSVIDEO1_ENCODER)        += msvideo1enc.o elbg.o
 OBJS-$(CONFIG_MSZH_DECODER)            += lcldec.o
-OBJS-$(CONFIG_MXPEG_DECODER)           += mxpegdec.o
+OBJS-$(CONFIG_MXPEG_DECODER)           += mxpegdec.o mjpegdec.o mjpeg.o
 OBJS-$(CONFIG_NELLYMOSER_DECODER)      += nellymoserdec.o nellymoser.o
 OBJS-$(CONFIG_NELLYMOSER_ENCODER)      += nellymoserenc.o nellymoser.o
 OBJS-$(CONFIG_NUV_DECODER)             += nuv.o rtjpeg.o
@@ -556,6 +556,7 @@ OBJS-$(CONFIG_SPDIF_DEMUXER)           += aacadtsdec.o mpeg4audio.o
 OBJS-$(CONFIG_WEBM_MUXER)              += xiph.o mpeg4audio.o \
                                           flacdec.o flacdata.o flac.o \
                                           mpegaudiodata.o vorbis_data.o
+OBJS-$(CONFIG_WTV_DEMUXER)             += mpeg4audio.o mpegaudiodata.o
 
 # external codec libraries
 OBJS-$(CONFIG_LIBCELT_DECODER)            += libcelt_dec.o
diff --git a/libavcodec/ac3enc.c b/libavcodec/ac3enc.c
index 48df4b7e6b..90d65bf44d 100644
--- a/libavcodec/ac3enc.c
+++ b/libavcodec/ac3enc.c
@@ -33,6 +33,7 @@
 
 #include "libavutil/audioconvert.h"
 #include "libavutil/avassert.h"
+#include "libavutil/avstring.h"
 #include "libavutil/crc.h"
 #include "libavutil/opt.h"
 #include "avcodec.h"
@@ -1551,10 +1552,10 @@ static void dprint_options(AVCodecContext *avctx)
     char strbuf[32];
 
     switch (s->bitstream_id) {
-    case  6:  strncpy(strbuf, "AC-3 (alt syntax)", 32);      break;
-    case  8:  strncpy(strbuf, "AC-3 (standard)", 32);        break;
-    case  9:  strncpy(strbuf, "AC-3 (dnet half-rate)", 32);  break;
-    case 10:  strncpy(strbuf, "AC-3 (dnet quater-rate", 32); break;
+    case  6:  av_strlcpy(strbuf, "AC-3 (alt syntax)", 32);      break;
+    case  8:  av_strlcpy(strbuf, "AC-3 (standard)", 32);        break;
+    case  9:  av_strlcpy(strbuf, "AC-3 (dnet half-rate)", 32);  break;
+    case 10:  av_strlcpy(strbuf, "AC-3 (dnet quater-rate", 32); break;
     default: snprintf(strbuf, 32, "ERROR");
     }
     av_dlog(avctx, "bitstream_id: %s (%d)\n", strbuf, s->bitstream_id);
@@ -1581,9 +1582,9 @@ static void dprint_options(AVCodecContext *avctx)
     if (opt->audio_production_info) {
         av_dlog(avctx, "mixing_level: %ddB\n", opt->mixing_level);
         switch (opt->room_type) {
-        case 0:  strncpy(strbuf, "notindicated", 32); break;
-        case 1:  strncpy(strbuf, "large", 32);        break;
-        case 2:  strncpy(strbuf, "small", 32);        break;
+        case 0:  av_strlcpy(strbuf, "notindicated", 32); break;
+        case 1:  av_strlcpy(strbuf, "large", 32);        break;
+        case 2:  av_strlcpy(strbuf, "small", 32);        break;
         default: snprintf(strbuf, 32, "ERROR (%d)", opt->room_type);
         }
         av_dlog(avctx, "room_type: %s\n", strbuf);
@@ -1595,9 +1596,9 @@ static void dprint_options(AVCodecContext *avctx)
     av_dlog(avctx, "dialnorm: %ddB\n", opt->dialogue_level);
     if (s->channel_mode == AC3_CHMODE_STEREO) {
         switch (opt->dolby_surround_mode) {
-        case 0:  strncpy(strbuf, "notindicated", 32); break;
-        case 1:  strncpy(strbuf, "on", 32);           break;
-        case 2:  strncpy(strbuf, "off", 32);          break;
+        case 0:  av_strlcpy(strbuf, "notindicated", 32); break;
+        case 1:  av_strlcpy(strbuf, "on", 32);           break;
+        case 2:  av_strlcpy(strbuf, "off", 32);          break;
         default: snprintf(strbuf, 32, "ERROR (%d)", opt->dolby_surround_mode);
         }
         av_dlog(avctx, "dsur_mode: %s\n", strbuf);
@@ -1609,9 +1610,9 @@ static void dprint_options(AVCodecContext *avctx)
     if (s->bitstream_id == 6) {
         if (opt->extended_bsi_1) {
             switch (opt->preferred_stereo_downmix) {
-            case 0:  strncpy(strbuf, "notindicated", 32); break;
-            case 1:  strncpy(strbuf, "ltrt", 32);         break;
-            case 2:  strncpy(strbuf, "loro", 32);         break;
+            case 0:  av_strlcpy(strbuf, "notindicated", 32); break;
+            case 1:  av_strlcpy(strbuf, "ltrt", 32);         break;
+            case 2:  av_strlcpy(strbuf, "loro", 32);         break;
             default: snprintf(strbuf, 32, "ERROR (%d)", opt->preferred_stereo_downmix);
             }
             av_dlog(avctx, "dmix_mode: %s\n", strbuf);
@@ -1628,23 +1629,23 @@ static void dprint_options(AVCodecContext *avctx)
         }
         if (opt->extended_bsi_2) {
             switch (opt->dolby_surround_ex_mode) {
-            case 0:  strncpy(strbuf, "notindicated", 32); break;
-            case 1:  strncpy(strbuf, "on", 32);           break;
-            case 2:  strncpy(strbuf, "off", 32);          break;
+            case 0:  av_strlcpy(strbuf, "notindicated", 32); break;
+            case 1:  av_strlcpy(strbuf, "on", 32);           break;
+            case 2:  av_strlcpy(strbuf, "off", 32);          break;
             default: snprintf(strbuf, 32, "ERROR (%d)", opt->dolby_surround_ex_mode);
             }
             av_dlog(avctx, "dsurex_mode: %s\n", strbuf);
             switch (opt->dolby_headphone_mode) {
-            case 0:  strncpy(strbuf, "notindicated", 32); break;
-            case 1:  strncpy(strbuf, "on", 32);           break;
-            case 2:  strncpy(strbuf, "off", 32);          break;
+            case 0:  av_strlcpy(strbuf, "notindicated", 32); break;
+            case 1:  av_strlcpy(strbuf, "on", 32);           break;
+            case 2:  av_strlcpy(strbuf, "off", 32);          break;
             default: snprintf(strbuf, 32, "ERROR (%d)", opt->dolby_headphone_mode);
             }
             av_dlog(avctx, "dheadphone_mode: %s\n", strbuf);
 
             switch (opt->ad_converter_type) {
-            case 0:  strncpy(strbuf, "standard", 32); break;
-            case 1:  strncpy(strbuf, "hdcd", 32);     break;
+            case 0:  av_strlcpy(strbuf, "standard", 32); break;
+            case 1:  av_strlcpy(strbuf, "hdcd", 32);     break;
             default: snprintf(strbuf, 32, "ERROR (%d)", opt->ad_converter_type);
             }
             av_dlog(avctx, "ad_conv_type: %s\n", strbuf);
diff --git a/libavcodec/ass.c b/libavcodec/ass.c
index a23567cba4..cb0babf2c8 100644
--- a/libavcodec/ass.c
+++ b/libavcodec/ass.c
@@ -21,6 +21,7 @@
 
 #include "avcodec.h"
 #include "ass.h"
+#include "libavutil/avstring.h"
 
 int ff_ass_subtitle_header(AVCodecContext *avctx,
                            const char *font, int font_size,
@@ -97,8 +98,7 @@ int ff_ass_add_rect(AVSubtitle *sub, const char *dialog,
     rects[sub->num_rects]->type = SUBTITLE_ASS;
     rects[sub->num_rects]->ass  = av_malloc(len + dlen + 1);
     strcpy (rects[sub->num_rects]->ass      , header);
-    strncpy(rects[sub->num_rects]->ass + len, dialog, dlen);
-    rects[sub->num_rects]->ass[len+dlen] = 0;
+    av_strlcpy(rects[sub->num_rects]->ass + len, dialog, dlen + 1);
     sub->num_rects++;
     return dlen;
 }
diff --git a/libavcodec/flashsv.c b/libavcodec/flashsv.c
index 12aabe6004..18327ad462 100644
--- a/libavcodec/flashsv.c
+++ b/libavcodec/flashsv.c
@@ -148,7 +148,7 @@ static int flashsv_decode_frame(AVCodecContext *avctx, void *data,
     if ((avctx->width != s->image_width) || (avctx->height != s->image_height)) {
         av_log(avctx, AV_LOG_ERROR, "Frame width or height differs from first frames!\n");
         av_log(avctx, AV_LOG_ERROR, "fh = %d, fv %d  vs  ch = %d, cv = %d\n", avctx->height,
-        avctx->width,s->image_height, s->image_width);
+        avctx->width, s->image_height, s->image_width);
         return -1;
     }
 
@@ -195,7 +195,7 @@ static int flashsv_decode_frame(AVCodecContext *avctx, void *data,
                 s->zstream.next_in   = buf + (get_bits_count(&gb) / 8);
                 s->zstream.avail_in  = size;
                 s->zstream.next_out  = s->tmpblock;
-                s->zstream.avail_out = s->block_size*3;
+                s->zstream.avail_out = s->block_size * 3;
                 ret = inflate(&(s->zstream), Z_FINISH);
                 if (ret == Z_DATA_ERROR) {
                     av_log(avctx, AV_LOG_ERROR, "Zlib resync occurred\n");
diff --git a/libavcodec/get_bits.h b/libavcodec/get_bits.h
index 2f1a851853..8579c87cd1 100644
--- a/libavcodec/get_bits.h
+++ b/libavcodec/get_bits.h
@@ -127,7 +127,7 @@ for examples see get_bits, show_bits, skip_bits, get_vlc
 
 #   define OPEN_READER(name, gb)                \
     unsigned int name##_index = (gb)->index;    \
-    av_unused int name##_cache
+    av_unused unsigned int name##_cache
 
 #   define CLOSE_READER(name, gb) (gb)->index = name##_index
 
diff --git a/libavcodec/indeo3.c b/libavcodec/indeo3.c
index c9d8573692..14526a239c 100644
--- a/libavcodec/indeo3.c
+++ b/libavcodec/indeo3.c
@@ -149,13 +149,13 @@ static av_cold void iv_free_func(Indeo3DecodeContext *s)
 }
 
 struct ustr {
-    long xpos;
-    long ypos;
-    long width;
-    long height;
-    long split_flag;
-    long split_direction;
-    long usl7;
+    int xpos;
+    int ypos;
+    int width;
+    int height;
+    int split_flag;
+    int split_direction;
+    int usl7;
 };
 
 
@@ -203,12 +203,12 @@ struct ustr {
 
 static void iv_Decode_Chunk(Indeo3DecodeContext *s,
         uint8_t *cur, uint8_t *ref, int width, int height,
-        const uint8_t *buf1, long cb_offset, const uint8_t *hdr,
+        const uint8_t *buf1, int cb_offset, const uint8_t *hdr,
         const uint8_t *buf2, int min_width_160)
 {
     uint8_t bit_buf;
-    unsigned long bit_pos, lv, lv1, lv2;
-    long *width_tbl, width_tbl_arr[10];
+    unsigned int bit_pos, lv, lv1, lv2;
+    int *width_tbl, width_tbl_arr[10];
     const signed char *ref_vectors;
     uint8_t *cur_frm_pos, *ref_frm_pos, *cp, *cp2;
     uint32_t *cur_lp, *ref_lp;
@@ -982,7 +982,7 @@ static int iv_decode_frame(AVCodecContext *avctx,
     Indeo3DecodeContext *s = avctx->priv_data;
     unsigned int image_width, image_height,
                  chroma_width, chroma_height;
-    unsigned long flags, cb_offset, data_size,
+    unsigned int flags, cb_offset, data_size,
                   y_offset, v_offset, u_offset, mc_vector_count;
     const uint8_t *hdr_pos, *buf_pos;
 
diff --git a/libavcodec/ppc/vc1dsp_altivec.c b/libavcodec/ppc/vc1dsp_altivec.c
index 05edb53b7c..69670619da 100644
--- a/libavcodec/ppc/vc1dsp_altivec.c
+++ b/libavcodec/ppc/vc1dsp_altivec.c
@@ -130,8 +130,7 @@ do { \
 
 /** Do inverse transform on 8x8 block
 */
-static void vc1_inv_trans_8x8_altivec(DCTELEM block[64],
-                                      int sign, int rangered)
+static void vc1_inv_trans_8x8_altivec(DCTELEM block[64])
 {
     vector signed short src0, src1, src2, src3, src4, src5, src6, src7;
     vector signed int s0, s1, s2, s3, s4, s5, s6, s7;
@@ -145,9 +144,6 @@ static void vc1_inv_trans_8x8_altivec(DCTELEM block[64],
     const vector unsigned int vec_2 = vec_splat_u32(2);
     const vector  signed int vec_1s = vec_splat_s32(1);
     const vector unsigned int vec_1 = vec_splat_u32(1);
-    const vector unsigned short rangered_shift = vec_splat_u16(1);
-    const vector   signed short signed_bias = vec_sl(vec_splat_s16(4),
-                                                     vec_splat_u16(4));
 
     src0 = vec_ld(  0, block);
     src1 = vec_ld( 16, block);
@@ -217,27 +213,6 @@ static void vc1_inv_trans_8x8_altivec(DCTELEM block[64],
     src6 = vec_pack(sE, s6);
     src7 = vec_pack(sF, s7);
 
-    if (rangered) {
-        if (!sign) {
-            src0 = vec_sub(src0, signed_bias);
-            src1 = vec_sub(src1, signed_bias);
-            src2 = vec_sub(src2, signed_bias);
-            src3 = vec_sub(src3, signed_bias);
-            src4 = vec_sub(src4, signed_bias);
-            src5 = vec_sub(src5, signed_bias);
-            src6 = vec_sub(src6, signed_bias);
-            src7 = vec_sub(src7, signed_bias);
-        }
-        src0 = vec_sl(src0, rangered_shift);
-        src1 = vec_sl(src1, rangered_shift);
-        src2 = vec_sl(src2, rangered_shift);
-        src3 = vec_sl(src3, rangered_shift);
-        src4 = vec_sl(src4, rangered_shift);
-        src5 = vec_sl(src5, rangered_shift);
-        src6 = vec_sl(src6, rangered_shift);
-        src7 = vec_sl(src7, rangered_shift);
-    }
-
     vec_st(src0,  0, block);
     vec_st(src1, 16, block);
     vec_st(src2, 32, block);
@@ -248,36 +223,6 @@ static void vc1_inv_trans_8x8_altivec(DCTELEM block[64],
     vec_st(src7,112, block);
 }
 
-static void vc1_inv_trans_8x8_add_altivec(uint8_t *dest, int stride, DCTELEM *b)
-{
-    vc1_inv_trans_8x8_altivec(b, 0, 0);
-    ff_add_pixels_clamped_c(b, dest, stride);
-}
-
-static void vc1_inv_trans_8x8_put_signed_altivec(uint8_t *dest, int stride, DCTELEM *b)
-{
-    vc1_inv_trans_8x8_altivec(b, 1, 0);
-    ff_put_signed_pixels_clamped_c(b, dest, stride);
-}
-
-static void vc1_inv_trans_8x8_put_signed_rangered_altivec(uint8_t *dest, int stride, DCTELEM *b)
-{
-    vc1_inv_trans_8x8_altivec(b, 1, 1);
-    ff_put_signed_pixels_clamped_c(b, dest, stride);
-}
-
-static void vc1_inv_trans_8x8_put_altivec(uint8_t *dest, int stride, DCTELEM *b)
-{
-    vc1_inv_trans_8x8_altivec(b, 0, 0);
-    ff_put_pixels_clamped_c(b, dest, stride);
-}
-
-static void vc1_inv_trans_8x8_put_rangered_altivec(uint8_t *dest, int stride, DCTELEM *b)
-{
-    vc1_inv_trans_8x8_altivec(b, 0, 1);
-    ff_put_pixels_clamped_c(b, dest, stride);
-}
-
 /** Do inverse transform on 8x4 part of block
 */
 static void vc1_inv_trans_8x4_altivec(uint8_t *dest, int stride, DCTELEM *block)
@@ -396,11 +341,7 @@ void ff_vc1dsp_init_altivec(VC1DSPContext* dsp)
     if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC))
         return;
 
-    dsp->vc1_inv_trans_8x8_add = vc1_inv_trans_8x8_add_altivec;
-    dsp->vc1_inv_trans_8x8_put_signed[0] = vc1_inv_trans_8x8_put_signed_altivec;
-    dsp->vc1_inv_trans_8x8_put_signed[1] = vc1_inv_trans_8x8_put_signed_rangered_altivec;
-    dsp->vc1_inv_trans_8x8_put[0] = vc1_inv_trans_8x8_put_altivec;
-    dsp->vc1_inv_trans_8x8_put[1] = vc1_inv_trans_8x8_put_rangered_altivec;
+    dsp->vc1_inv_trans_8x8 = vc1_inv_trans_8x8_altivec;
     dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_altivec;
     dsp->put_no_rnd_vc1_chroma_pixels_tab[0] = put_no_rnd_vc1_chroma_mc8_altivec;
     dsp->avg_no_rnd_vc1_chroma_pixels_tab[0] = avg_no_rnd_vc1_chroma_mc8_altivec;
diff --git a/libavcodec/vc1.c b/libavcodec/vc1.c
index ef44a34f50..e062a35cc1 100644
--- a/libavcodec/vc1.c
+++ b/libavcodec/vc1.c
@@ -280,28 +280,6 @@ static int vop_dquant_decoding(VC1Context *v)
 
 static int decode_sequence_header_adv(VC1Context *v, GetBitContext *gb);
 
-static void simple_idct_put_rangered(uint8_t *dest, int line_size, DCTELEM *block)
-{
-    int i;
-    ff_simple_idct(block);
-    for (i = 0; i < 64; i++) block[i] = (block[i] - 64) << 1;
-    ff_put_pixels_clamped_c(block, dest, line_size);
-}
-
-static void simple_idct_put_signed(uint8_t *dest, int line_size, DCTELEM *block)
-{
-    ff_simple_idct(block);
-    ff_put_signed_pixels_clamped_c(block, dest, line_size);
-}
-
-static void simple_idct_put_signed_rangered(uint8_t *dest, int line_size, DCTELEM *block)
-{
-    int i;
-    ff_simple_idct(block);
-    for (i = 0; i < 64; i++) block[i] <<= 1;
-    ff_put_signed_pixels_clamped_c(block, dest, line_size);
-}
-
 /**
  * Decode Simple/Main Profiles sequence header
  * @see Figure 7-8, p16-17
@@ -359,11 +337,7 @@ int vc1_decode_sequence_header(AVCodecContext *avctx, VC1Context *v, GetBitConte
     v->res_fasttx = get_bits1(gb);
     if (!v->res_fasttx)
     {
-        v->vc1dsp.vc1_inv_trans_8x8_add = ff_simple_idct_add;
-        v->vc1dsp.vc1_inv_trans_8x8_put[0] = ff_simple_idct_put;
-        v->vc1dsp.vc1_inv_trans_8x8_put[1] = simple_idct_put_rangered;
-        v->vc1dsp.vc1_inv_trans_8x8_put_signed[0] = simple_idct_put_signed;
-        v->vc1dsp.vc1_inv_trans_8x8_put_signed[1] = simple_idct_put_signed_rangered;
+        v->vc1dsp.vc1_inv_trans_8x8 = ff_simple_idct;
         v->vc1dsp.vc1_inv_trans_8x4 = ff_simple_idct84_add;
         v->vc1dsp.vc1_inv_trans_4x8 = ff_simple_idct48_add;
         v->vc1dsp.vc1_inv_trans_4x4 = ff_simple_idct44_add;
diff --git a/libavcodec/vc1.h b/libavcodec/vc1.h
index 19be3c3452..6d4c0aa7a3 100644
--- a/libavcodec/vc1.h
+++ b/libavcodec/vc1.h
@@ -317,6 +317,8 @@ typedef struct VC1Context{
     int bi_type;
     int x8_type;
 
+    DCTELEM (*block)[6][64];
+    int n_allocated_blks, cur_blk_idx, left_blk_idx, topleft_blk_idx, top_blk_idx;
     uint32_t *cbp_base, *cbp;
     uint8_t *is_intra_base, *is_intra;
     int16_t (*luma_mv_base)[2], (*luma_mv)[2];
diff --git a/libavcodec/vc1dec.c b/libavcodec/vc1dec.c
index 125994d8b9..d35e664aca 100644
--- a/libavcodec/vc1dec.c
+++ b/libavcodec/vc1dec.c
@@ -160,6 +160,72 @@ enum Imode {
 
 /** @} */ //Bitplane group
 
+static void vc1_put_signed_blocks_clamped(VC1Context *v)
+{
+    MpegEncContext *s = &v->s;
+
+    /* The put pixels loop is always one MB row behind the decoding loop,
+     * because we can only put pixels when overlap filtering is done, and
+     * for filtering of the bottom edge of a MB, we need the next MB row
+     * present as well.
+     * Within the row, the put pixels loop is also one MB col behind the
+     * decoding loop. The reason for this is again, because for filtering
+     * of the right MB edge, we need the next MB present. */
+    if (!s->first_slice_line) {
+        if (s->mb_x) {
+            s->dsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][0],
+                                             s->dest[0] - 16 * s->linesize - 16,
+                                             s->linesize);
+            s->dsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][1],
+                                             s->dest[0] - 16 * s->linesize - 8,
+                                             s->linesize);
+            s->dsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][2],
+                                             s->dest[0] - 8 * s->linesize - 16,
+                                             s->linesize);
+            s->dsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][3],
+                                             s->dest[0] - 8 * s->linesize - 8,
+                                             s->linesize);
+            s->dsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][4],
+                                             s->dest[1] - 8 * s->uvlinesize - 8,
+                                             s->uvlinesize);
+            s->dsp.put_signed_pixels_clamped(v->block[v->topleft_blk_idx][5],
+                                             s->dest[2] - 8 * s->uvlinesize - 8,
+                                             s->uvlinesize);
+        }
+        if (s->mb_x == s->mb_width - 1) {
+            s->dsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][0],
+                                             s->dest[0] - 16 * s->linesize,
+                                             s->linesize);
+            s->dsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][1],
+                                             s->dest[0] - 16 * s->linesize + 8,
+                                             s->linesize);
+            s->dsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][2],
+                                             s->dest[0] - 8 * s->linesize,
+                                             s->linesize);
+            s->dsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][3],
+                                             s->dest[0] - 8 * s->linesize + 8,
+                                             s->linesize);
+            s->dsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][4],
+                                             s->dest[1] - 8 * s->uvlinesize,
+                                             s->uvlinesize);
+            s->dsp.put_signed_pixels_clamped(v->block[v->top_blk_idx][5],
+                                             s->dest[2] - 8 * s->uvlinesize,
+                                             s->uvlinesize);
+        }
+    }
+
+#define inc_blk_idx(idx) do { \
+        idx++; \
+        if (idx >= v->n_allocated_blks) \
+            idx = 0; \
+    } while (0)
+
+    inc_blk_idx(v->topleft_blk_idx);
+    inc_blk_idx(v->top_blk_idx);
+    inc_blk_idx(v->left_blk_idx);
+    inc_blk_idx(v->cur_blk_idx);
+}
+
 static void vc1_loop_filter_iblk(VC1Context *v, int pq)
 {
     MpegEncContext *s = &v->s;
@@ -187,6 +253,151 @@ static void vc1_loop_filter_iblk(VC1Context *v, int pq)
     }
 }
 
+static void vc1_loop_filter_iblk_delayed(VC1Context *v, int pq)
+{
+    MpegEncContext *s = &v->s;
+    int j;
+
+    /* The loopfilter runs 1 row and 1 column behind the overlap filter, which
+     * means it runs two rows/cols behind the decoding loop. */
+    if (!s->first_slice_line) {
+        if (s->mb_x) {
+            if (s->mb_y >= s->start_mb_y + 2) {
+                v->vc1dsp.vc1_v_loop_filter16(s->dest[0] - 16 * s->linesize - 16, s->linesize, pq);
+
+                if (s->mb_x >= 2)
+                    v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize - 16, s->linesize, pq);
+                v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize - 8, s->linesize, pq);
+                for(j = 0; j < 2; j++) {
+                    v->vc1dsp.vc1_v_loop_filter8(s->dest[j+1] - 8 * s->uvlinesize - 8, s->uvlinesize, pq);
+                    if (s->mb_x >= 2) {
+                        v->vc1dsp.vc1_h_loop_filter8(s->dest[j+1] - 16 * s->uvlinesize - 8, s->uvlinesize, pq);
+                    }
+                }
+            }
+            v->vc1dsp.vc1_v_loop_filter16(s->dest[0] - 8 * s->linesize - 16, s->linesize, pq);
+        }
+
+        if (s->mb_x == s->mb_width - 1) {
+            if (s->mb_y >= s->start_mb_y + 2) {
+                v->vc1dsp.vc1_v_loop_filter16(s->dest[0] - 16 * s->linesize, s->linesize, pq);
+
+                if (s->mb_x)
+                    v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize, s->linesize, pq);
+                v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 32 * s->linesize + 8, s->linesize, pq);
+                for(j = 0; j < 2; j++) {
+                    v->vc1dsp.vc1_v_loop_filter8(s->dest[j+1] - 8 * s->uvlinesize, s->uvlinesize, pq);
+                    if (s->mb_x >= 2) {
+                        v->vc1dsp.vc1_h_loop_filter8(s->dest[j+1] - 16 * s->uvlinesize, s->uvlinesize, pq);
+                    }
+                }
+            }
+            v->vc1dsp.vc1_v_loop_filter16(s->dest[0] - 8 * s->linesize, s->linesize, pq);
+        }
+
+        if (s->mb_y == s->mb_height) {
+            if (s->mb_x) {
+                if (s->mb_x >= 2)
+                    v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize - 16, s->linesize, pq);
+                v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize - 8, s->linesize, pq);
+                if (s->mb_x >= 2) {
+                    for(j = 0; j < 2; j++) {
+                        v->vc1dsp.vc1_h_loop_filter8(s->dest[j+1] - 8 * s->uvlinesize - 8, s->uvlinesize, pq);
+                    }
+                }
+            }
+
+            if (s->mb_x == s->mb_width - 1) {
+                if (s->mb_x)
+                    v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize, s->linesize, pq);
+                v->vc1dsp.vc1_h_loop_filter16(s->dest[0] - 16 * s->linesize + 8, s->linesize, pq);
+                if (s->mb_x) {
+                    for(j = 0; j < 2; j++) {
+                        v->vc1dsp.vc1_h_loop_filter8(s->dest[j+1] - 8 * s->uvlinesize, s->uvlinesize, pq);
+                    }
+                }
+            }
+        }
+    }
+}
+
+static void vc1_smooth_overlap_filter_iblk(VC1Context *v)
+{
+    MpegEncContext *s = &v->s;
+    int mb_pos;
+
+    if (v->condover == CONDOVER_NONE)
+        return;
+
+    mb_pos = s->mb_x + s->mb_y * s->mb_stride;
+
+    /* Within a MB, the horizontal overlap always runs before the vertical.
+     * To accomplish that, we run the H on left and internal borders of the
+     * currently decoded MB. Then, we wait for the next overlap iteration
+     * to do H overlap on the right edge of this MB, before moving over and
+     * running the V overlap. Therefore, the V overlap makes us trail by one
+     * MB col and the H overlap filter makes us trail by one MB row. This
+     * is reflected in the time at which we run the put_pixels loop. */
+    if(v->condover == CONDOVER_ALL || v->pq >= 9 || v->over_flags_plane[mb_pos]) {
+        if(s->mb_x && (v->condover == CONDOVER_ALL || v->pq >= 9 ||
+                       v->over_flags_plane[mb_pos - 1])) {
+            v->vc1dsp.vc1_h_s_overlap(v->block[v->left_blk_idx][1],
+                                      v->block[v->cur_blk_idx][0]);
+            v->vc1dsp.vc1_h_s_overlap(v->block[v->left_blk_idx][3],
+                                      v->block[v->cur_blk_idx][2]);
+            if(!(s->flags & CODEC_FLAG_GRAY)) {
+                v->vc1dsp.vc1_h_s_overlap(v->block[v->left_blk_idx][4],
+                                          v->block[v->cur_blk_idx][4]);
+                v->vc1dsp.vc1_h_s_overlap(v->block[v->left_blk_idx][5],
+                                          v->block[v->cur_blk_idx][5]);
+            }
+        }
+        v->vc1dsp.vc1_h_s_overlap(v->block[v->cur_blk_idx][0],
+                                  v->block[v->cur_blk_idx][1]);
+        v->vc1dsp.vc1_h_s_overlap(v->block[v->cur_blk_idx][2],
+                                  v->block[v->cur_blk_idx][3]);
+
+        if (s->mb_x == s->mb_width - 1) {
+            if(!s->first_slice_line && (v->condover == CONDOVER_ALL || v->pq >= 9 ||
+                                        v->over_flags_plane[mb_pos - s->mb_stride])) {
+                v->vc1dsp.vc1_v_s_overlap(v->block[v->top_blk_idx][2],
+                                          v->block[v->cur_blk_idx][0]);
+                v->vc1dsp.vc1_v_s_overlap(v->block[v->top_blk_idx][3],
+                                          v->block[v->cur_blk_idx][1]);
+                if(!(s->flags & CODEC_FLAG_GRAY)) {
+                    v->vc1dsp.vc1_v_s_overlap(v->block[v->top_blk_idx][4],
+                                              v->block[v->cur_blk_idx][4]);
+                    v->vc1dsp.vc1_v_s_overlap(v->block[v->top_blk_idx][5],
+                                              v->block[v->cur_blk_idx][5]);
+                }
+            }
+            v->vc1dsp.vc1_v_s_overlap(v->block[v->cur_blk_idx][0],
+                                      v->block[v->cur_blk_idx][2]);
+            v->vc1dsp.vc1_v_s_overlap(v->block[v->cur_blk_idx][1],
+                                      v->block[v->cur_blk_idx][3]);
+        }
+    }
+    if (s->mb_x && (v->condover == CONDOVER_ALL || v->over_flags_plane[mb_pos - 1])) {
+        if(!s->first_slice_line && (v->condover == CONDOVER_ALL || v->pq >= 9 ||
+                                    v->over_flags_plane[mb_pos - s->mb_stride - 1])) {
+            v->vc1dsp.vc1_v_s_overlap(v->block[v->topleft_blk_idx][2],
+                                      v->block[v->left_blk_idx][0]);
+            v->vc1dsp.vc1_v_s_overlap(v->block[v->topleft_blk_idx][3],
+                                      v->block[v->left_blk_idx][1]);
+            if(!(s->flags & CODEC_FLAG_GRAY)) {
+                v->vc1dsp.vc1_v_s_overlap(v->block[v->topleft_blk_idx][4],
+                                          v->block[v->left_blk_idx][4]);
+                v->vc1dsp.vc1_v_s_overlap(v->block[v->topleft_blk_idx][5],
+                                          v->block[v->left_blk_idx][5]);
+            }
+        }
+        v->vc1dsp.vc1_v_s_overlap(v->block[v->left_blk_idx][0],
+                                  v->block[v->left_blk_idx][2]);
+        v->vc1dsp.vc1_v_s_overlap(v->block[v->left_blk_idx][1],
+                                  v->block[v->left_blk_idx][3]);
+    }
+}
+
 /** Do motion compensation over 1 macroblock
  * Mostly adapted hpel_motion and qpel_motion from mpegvideo.c
  */
@@ -2016,7 +2227,8 @@ static int vc1_decode_p_block(VC1Context *v, DCTELEM block[64], int n, int mquan
             if(i==1)
                 v->vc1dsp.vc1_inv_trans_8x8_dc(dst, linesize, block);
             else{
-                v->vc1dsp.vc1_inv_trans_8x8_add(dst, linesize, block);
+                v->vc1dsp.vc1_inv_trans_8x8(block);
+                s->dsp.add_pixels_clamped(block, dst, linesize);
             }
         }
         break;
@@ -2258,7 +2470,7 @@ static int vc1_decode_p_mb(VC1Context *v)
 {
     MpegEncContext *s = &v->s;
     GetBitContext *gb = &s->gb;
-    int i;
+    int i, j;
     int mb_pos = s->mb_x + s->mb_y * s->mb_stride;
     int cbp; /* cbp decoding stuff */
     int mqdiff, mquant; /* MB quantization */
@@ -2288,8 +2500,6 @@ static int vc1_decode_p_mb(VC1Context *v)
     {
         if (!skipped)
         {
-            vc1_idct_func idct8x8_fn;
-
             GET_MVDATA(dmv_x, dmv_y);
 
             if (s->mb_intra) {
@@ -2324,7 +2534,6 @@ static int vc1_decode_p_mb(VC1Context *v)
                                 VC1_TTMB_VLC_BITS, 2);
             if(!s->mb_intra) vc1_mc_1mv(v, 0);
             dst_idx = 0;
-            idct8x8_fn = v->vc1dsp.vc1_inv_trans_8x8_put_signed[!!v->rangeredfrm];
             for (i=0; i<6; i++)
             {
                 s->dc_val[0][s->block_index[i]] = 0;
@@ -2342,9 +2551,9 @@ static int vc1_decode_p_mb(VC1Context *v)
 
                     vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i&4)?v->codingset2:v->codingset);
                     if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue;
-                    idct8x8_fn(s->dest[dst_idx] + off,
-                               i & 4 ? s->uvlinesize : s->linesize,
-                               s->block[i]);
+                    v->vc1dsp.vc1_inv_trans_8x8(s->block[i]);
+                    if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1;
+                    s->dsp.put_signed_pixels_clamped(s->block[i], s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
                     if(v->pq >= 9 && v->overlap) {
                         if(v->c_avail)
                             v->vc1dsp.vc1_h_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
@@ -2380,7 +2589,6 @@ static int vc1_decode_p_mb(VC1Context *v)
         {
             int intra_count = 0, coded_inter = 0;
             int is_intra[6], is_coded[6];
-            vc1_idct_func idct8x8_fn;
             /* Get CBPCY */
             cbp = get_vlc2(&v->s.gb, v->cbpcy_vlc->table, VC1_CBPCY_P_VLC_BITS, 2);
             for (i=0; i<6; i++)
@@ -2431,7 +2639,6 @@ static int vc1_decode_p_mb(VC1Context *v)
             }
             if (!v->ttmbf && coded_inter)
                 ttmb = get_vlc2(gb, ff_vc1_ttmb_vlc[v->tt_index].table, VC1_TTMB_VLC_BITS, 2);
-            idct8x8_fn = v->vc1dsp.vc1_inv_trans_8x8_put_signed[!!v->rangeredfrm];
             for (i=0; i<6; i++)
             {
                 dst_idx += i >> 2;
@@ -2447,9 +2654,9 @@ static int vc1_decode_p_mb(VC1Context *v)
 
                     vc1_decode_intra_block(v, s->block[i], i, is_coded[i], mquant, (i&4)?v->codingset2:v->codingset);
                     if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue;
-                    idct8x8_fn(s->dest[dst_idx] + off,
-                               (i&4)?s->uvlinesize:s->linesize,
-                               s->block[i]);
+                    v->vc1dsp.vc1_inv_trans_8x8(s->block[i]);
+                    if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1;
+                    s->dsp.put_signed_pixels_clamped(s->block[i], s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize);
                     if(v->pq >= 9 && v->overlap) {
                         if(v->c_avail)
                             v->vc1dsp.vc1_h_overlap(s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
@@ -2497,7 +2704,7 @@ static void vc1_decode_b_mb(VC1Context *v)
 {
     MpegEncContext *s = &v->s;
     GetBitContext *gb = &s->gb;
-    int i;
+    int i, j;
     int mb_pos = s->mb_x + s->mb_y * s->mb_stride;
     int cbp = 0; /* cbp decoding stuff */
     int mqdiff, mquant; /* MB quantization */
@@ -2510,7 +2717,6 @@ static void vc1_decode_b_mb(VC1Context *v)
     int skipped, direct;
     int dmv_x[2], dmv_y[2];
     int bmvtype = BMV_TYPE_BACKWARD;
-    vc1_idct_func idct8x8_fn;
 
     mquant = v->pq; /* Loosy initialization */
     s->mb_intra = 0;
@@ -2608,7 +2814,6 @@ static void vc1_decode_b_mb(VC1Context *v)
         }
     }
     dst_idx = 0;
-    idct8x8_fn = v->vc1dsp.vc1_inv_trans_8x8_put_signed[!!v->rangeredfrm];
     for (i=0; i<6; i++)
     {
         s->dc_val[0][s->block_index[i]] = 0;
@@ -2626,9 +2831,9 @@ static void vc1_decode_b_mb(VC1Context *v)
 
             vc1_decode_intra_block(v, s->block[i], i, val, mquant, (i&4)?v->codingset2:v->codingset);
             if((i>3) && (s->flags & CODEC_FLAG_GRAY)) continue;
-            idct8x8_fn(s->dest[dst_idx] + off,
-                       i & 4 ? s->uvlinesize : s->linesize,
-                       s->block[i]);
+            v->vc1dsp.vc1_inv_trans_8x8(s->block[i]);
+            if(v->rangeredfrm) for(j = 0; j < 64; j++) s->block[i][j] <<= 1;
+            s->dsp.put_signed_pixels_clamped(s->block[i], s->dest[dst_idx] + off, i & 4 ? s->uvlinesize : s->linesize);
         } else if(val) {
             vc1_decode_p_block(v, s->block[i], i, mquant, ttmb, first_block, s->dest[dst_idx] + off, (i&4)?s->uvlinesize:s->linesize, (i&4) && (s->flags & CODEC_FLAG_GRAY), NULL);
             if(!v->ttmbf && ttmb < 8) ttmb = -1;
@@ -2641,12 +2846,11 @@ static void vc1_decode_b_mb(VC1Context *v)
  */
 static void vc1_decode_i_blocks(VC1Context *v)
 {
-    int k;
+    int k, j;
     MpegEncContext *s = &v->s;
     int cbp, val;
     uint8_t *coded_val;
     int mb_pos;
-    vc1_idct_func idct8x8_fn;
 
     /* select codingmode used for VLC tables selection */
     switch(v->y_ac_table_index){
@@ -2681,10 +2885,6 @@ static void vc1_decode_i_blocks(VC1Context *v)
     s->mb_x = s->mb_y = 0;
     s->mb_intra = 1;
     s->first_slice_line = 1;
-    if(v->pq >= 9 && v->overlap) {
-        idct8x8_fn = v->vc1dsp.vc1_inv_trans_8x8_put_signed[!!v->rangeredfrm];
-    } else
-        idct8x8_fn = v->vc1dsp.vc1_inv_trans_8x8_put[!!v->rangeredfrm];
     for(s->mb_y = 0; s->mb_y < s->mb_height; s->mb_y++) {
         s->mb_x = 0;
         ff_init_block_index(s);
@@ -2721,9 +2921,14 @@ static void vc1_decode_i_blocks(VC1Context *v)
                 vc1_decode_i_block(v, s->block[k], k, val, (k<4)? v->codingset : v->codingset2);
 
                 if (k > 3 && (s->flags & CODEC_FLAG_GRAY)) continue;
-                idct8x8_fn(dst[k],
-                           k & 4 ? s->uvlinesize : s->linesize,
-                           s->block[k]);
+                v->vc1dsp.vc1_inv_trans_8x8(s->block[k]);
+                if(v->pq >= 9 && v->overlap) {
+                    if (v->rangeredfrm) for(j = 0; j < 64; j++) s->block[k][j] <<= 1;
+                    s->dsp.put_signed_pixels_clamped(s->block[k], dst[k], k & 4 ? s->uvlinesize : s->linesize);
+                } else {
+                    if (v->rangeredfrm) for(j = 0; j < 64; j++) s->block[k][j] = (s->block[k][j] - 64) << 1;
+                    s->dsp.put_pixels_clamped(s->block[k], dst[k], k & 4 ? s->uvlinesize : s->linesize);
+                }
             }
 
             if(v->pq >= 9 && v->overlap) {
@@ -2770,7 +2975,7 @@ static void vc1_decode_i_blocks(VC1Context *v)
 
 /** Decode blocks of I-frame for advanced profile
  */
-static void vc1_decode_i_blocks_adv(VC1Context *v, int mby_start, int mby_end)
+static void vc1_decode_i_blocks_adv(VC1Context *v)
 {
     int k;
     MpegEncContext *s = &v->s;
@@ -2779,9 +2984,7 @@ static void vc1_decode_i_blocks_adv(VC1Context *v, int mby_start, int mby_end)
     int mb_pos;
     int mquant = v->pq;
     int mqdiff;
-    int overlap;
     GetBitContext *gb = &s->gb;
-    vc1_idct_func idct8x8_fn;
 
     /* select codingmode used for VLC tables selection */
     switch(v->y_ac_table_index){
@@ -2812,27 +3015,20 @@ static void vc1_decode_i_blocks_adv(VC1Context *v, int mby_start, int mby_end)
     s->mb_x = s->mb_y = 0;
     s->mb_intra = 1;
     s->first_slice_line = 1;
-    s->mb_y = mby_start;
-    if (mby_start) {
+    s->mb_y = s->start_mb_y;
+    if (s->start_mb_y) {
         s->mb_x = 0;
         ff_init_block_index(s);
         memset(&s->coded_block[s->block_index[0]-s->b8_stride], 0,
                s->b8_stride * sizeof(*s->coded_block));
     }
-    idct8x8_fn = v->vc1dsp.vc1_inv_trans_8x8_put_signed[0];
-    for(; s->mb_y < mby_end; s->mb_y++) {
+    for(; s->mb_y < s->end_mb_y; s->mb_y++) {
         s->mb_x = 0;
         ff_init_block_index(s);
         for(;s->mb_x < s->mb_width; s->mb_x++) {
-            uint8_t *dst[6];
+            DCTELEM (*block)[64] = v->block[v->cur_blk_idx];
             ff_update_block_index(s);
-            dst[0] = s->dest[0];
-            dst[1] = dst[0] + 8;
-            dst[2] = s->dest[0] + s->linesize * 8;
-            dst[3] = dst[2] + 8;
-            dst[4] = s->dest[1];
-            dst[5] = s->dest[2];
-            s->dsp.clear_blocks(s->block[0]);
+            s->dsp.clear_blocks(block[0]);
             mb_pos = s->mb_x + s->mb_y * s->mb_stride;
             s->current_picture.mb_type[mb_pos] = MB_TYPE_INTRA;
             s->current_picture.motion_val[1][s->block_index[0]][0] = 0;
@@ -2845,13 +3041,8 @@ static void vc1_decode_i_blocks_adv(VC1Context *v, int mby_start, int mby_end)
             else
                 v->s.ac_pred = v->acpred_plane[mb_pos];
 
-            if(v->condover == CONDOVER_SELECT) {
-                if(v->overflg_is_raw)
-                    overlap = get_bits1(&v->s.gb);
-                else
-                    overlap = v->over_flags_plane[mb_pos];
-            } else
-                overlap = (v->condover == CONDOVER_ALL);
+            if (v->condover == CONDOVER_SELECT && v->overflg_is_raw)
+                v->over_flags_plane[mb_pos] = get_bits1(&v->s.gb);
 
             GET_MQUANT();
 
@@ -2873,40 +3064,18 @@ static void vc1_decode_i_blocks_adv(VC1Context *v, int mby_start, int mby_end)
                 v->a_avail = !s->first_slice_line || (k==2 || k==3);
                 v->c_avail = !!s->mb_x || (k==1 || k==3);
 
-                vc1_decode_i_block_adv(v, s->block[k], k, val, (k<4)? v->codingset : v->codingset2, mquant);
+                vc1_decode_i_block_adv(v, block[k], k, val, (k<4)? v->codingset : v->codingset2, mquant);
 
                 if (k > 3 && (s->flags & CODEC_FLAG_GRAY)) continue;
-                idct8x8_fn(dst[k],
-                           k & 4 ? s->uvlinesize : s->linesize,
-                           s->block[k]);
+                v->vc1dsp.vc1_inv_trans_8x8(block[k]);
             }
 
-            if(overlap) {
-                if(s->mb_x) {
-                    v->vc1dsp.vc1_h_overlap(s->dest[0], s->linesize);
-                    v->vc1dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize, s->linesize);
-                    if(!(s->flags & CODEC_FLAG_GRAY)) {
-                        v->vc1dsp.vc1_h_overlap(s->dest[1], s->uvlinesize);
-                        v->vc1dsp.vc1_h_overlap(s->dest[2], s->uvlinesize);
-                    }
-                }
-                v->vc1dsp.vc1_h_overlap(s->dest[0] + 8, s->linesize);
-                v->vc1dsp.vc1_h_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize);
-                if(!s->first_slice_line) {
-                    v->vc1dsp.vc1_v_overlap(s->dest[0], s->linesize);
-                    v->vc1dsp.vc1_v_overlap(s->dest[0] + 8, s->linesize);
-                    if(!(s->flags & CODEC_FLAG_GRAY)) {
-                        v->vc1dsp.vc1_v_overlap(s->dest[1], s->uvlinesize);
-                        v->vc1dsp.vc1_v_overlap(s->dest[2], s->uvlinesize);
-                    }
-                }
-                v->vc1dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize, s->linesize);
-                v->vc1dsp.vc1_v_overlap(s->dest[0] + 8 * s->linesize + 8, s->linesize);
-            }
-            if(v->s.loop_filter) vc1_loop_filter_iblk(v, v->pq);
+            vc1_smooth_overlap_filter_iblk(v);
+            vc1_put_signed_blocks_clamped(v);
+            if(v->s.loop_filter) vc1_loop_filter_iblk_delayed(v, v->pq);
 
             if(get_bits_count(&s->gb) > v->bits) {
-                ff_er_add_slice(s, 0, mby_start, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END));
+                ff_er_add_slice(s, 0, s->start_mb_y, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END));
                 av_log(s->avctx, AV_LOG_ERROR, "Bits overconsumption: %i > %i\n", get_bits_count(&s->gb), v->bits);
                 return;
             }
@@ -2917,12 +3086,21 @@ static void vc1_decode_i_blocks_adv(VC1Context *v, int mby_start, int mby_end)
             ff_draw_horiz_band(s, (s->mb_y-1) * 16, 16);
         s->first_slice_line = 0;
     }
+
+    /* raw bottom MB row */
+    s->mb_x = 0;
+    ff_init_block_index(s);
+    for(;s->mb_x < s->mb_width; s->mb_x++) {
+        ff_update_block_index(s);
+        vc1_put_signed_blocks_clamped(v);
+        if(v->s.loop_filter) vc1_loop_filter_iblk_delayed(v, v->pq);
+    }
     if (v->s.loop_filter)
         ff_draw_horiz_band(s, (s->mb_height-1)*16, 16);
-    ff_er_add_slice(s, 0, mby_start, s->mb_width - 1, mby_end - 1, (AC_END|DC_END|MV_END));
+    ff_er_add_slice(s, 0, s->start_mb_y, s->mb_width - 1, s->end_mb_y - 1, (AC_END|DC_END|MV_END));
 }
 
-static void vc1_decode_p_blocks(VC1Context *v, int mby_start, int mby_end)
+static void vc1_decode_p_blocks(VC1Context *v)
 {
     MpegEncContext *s = &v->s;
     int apply_loop_filter;
@@ -2955,17 +3133,17 @@ static void vc1_decode_p_blocks(VC1Context *v, int mby_start, int mby_end)
     apply_loop_filter = s->loop_filter && !(s->avctx->skip_loop_filter >= AVDISCARD_NONKEY);
     s->first_slice_line = 1;
     memset(v->cbp_base, 0, sizeof(v->cbp_base[0])*2*s->mb_stride);
-    for(s->mb_y = mby_start; s->mb_y < mby_end; s->mb_y++) {
+    for(s->mb_y = s->start_mb_y; s->mb_y < s->end_mb_y; s->mb_y++) {
         s->mb_x = 0;
         ff_init_block_index(s);
         for(; s->mb_x < s->mb_width; s->mb_x++) {
             ff_update_block_index(s);
 
             vc1_decode_p_mb(v);
-            if (s->mb_y != mby_start && apply_loop_filter)
+            if (s->mb_y != s->start_mb_y && apply_loop_filter)
                 vc1_apply_p_loop_filter(v);
             if(get_bits_count(&s->gb) > v->bits || get_bits_count(&s->gb) < 0) {
-                ff_er_add_slice(s, 0, mby_start, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END));
+                ff_er_add_slice(s, 0, s->start_mb_y, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END));
                 av_log(s->avctx, AV_LOG_ERROR, "Bits overconsumption: %i > %i at %ix%i\n", get_bits_count(&s->gb), v->bits,s->mb_x,s->mb_y);
                 return;
             }
@@ -2974,7 +3152,7 @@ static void vc1_decode_p_blocks(VC1Context *v, int mby_start, int mby_end)
         memmove(v->ttblk_base, v->ttblk, sizeof(v->ttblk_base[0])*s->mb_stride);
         memmove(v->is_intra_base, v->is_intra, sizeof(v->is_intra_base[0])*s->mb_stride);
         memmove(v->luma_mv_base, v->luma_mv, sizeof(v->luma_mv_base[0])*s->mb_stride);
-        if (s->mb_y != mby_start) ff_draw_horiz_band(s, (s->mb_y-1) * 16, 16);
+        if (s->mb_y != s->start_mb_y) ff_draw_horiz_band(s, (s->mb_y-1) * 16, 16);
         s->first_slice_line = 0;
     }
     if (apply_loop_filter) {
@@ -2985,12 +3163,12 @@ static void vc1_decode_p_blocks(VC1Context *v, int mby_start, int mby_end)
             vc1_apply_p_loop_filter(v);
         }
     }
-    if (mby_end >= mby_start)
-        ff_draw_horiz_band(s, (mby_end-1) * 16, 16);
-    ff_er_add_slice(s, 0, mby_start, s->mb_width - 1, mby_end - 1, (AC_END|DC_END|MV_END));
+    if (s->end_mb_y >= s->start_mb_y)
+        ff_draw_horiz_band(s, (s->end_mb_y-1) * 16, 16);
+    ff_er_add_slice(s, 0, s->start_mb_y, s->mb_width - 1, s->end_mb_y - 1, (AC_END|DC_END|MV_END));
 }
 
-static void vc1_decode_b_blocks(VC1Context *v, int mby_start, int mby_end)
+static void vc1_decode_b_blocks(VC1Context *v)
 {
     MpegEncContext *s = &v->s;
 
@@ -3020,7 +3198,7 @@ static void vc1_decode_b_blocks(VC1Context *v, int mby_start, int mby_end)
     }
 
     s->first_slice_line = 1;
-    for(s->mb_y = mby_start; s->mb_y < mby_end; s->mb_y++) {
+    for(s->mb_y = s->start_mb_y; s->mb_y < s->end_mb_y; s->mb_y++) {
         s->mb_x = 0;
         ff_init_block_index(s);
         for(; s->mb_x < s->mb_width; s->mb_x++) {
@@ -3028,7 +3206,7 @@ static void vc1_decode_b_blocks(VC1Context *v, int mby_start, int mby_end)
 
             vc1_decode_b_mb(v);
             if(get_bits_count(&s->gb) > v->bits || get_bits_count(&s->gb) < 0) {
-                ff_er_add_slice(s, 0, mby_start, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END));
+                ff_er_add_slice(s, 0, s->start_mb_y, s->mb_x, s->mb_y, (AC_END|DC_END|MV_END));
                 av_log(s->avctx, AV_LOG_ERROR, "Bits overconsumption: %i > %i at %ix%i\n", get_bits_count(&s->gb), v->bits,s->mb_x,s->mb_y);
                 return;
             }
@@ -3042,7 +3220,7 @@ static void vc1_decode_b_blocks(VC1Context *v, int mby_start, int mby_end)
     }
     if (v->s.loop_filter)
         ff_draw_horiz_band(s, (s->mb_height-1)*16, 16);
-    ff_er_add_slice(s, 0, mby_start, s->mb_width - 1, mby_end - 1, (AC_END|DC_END|MV_END));
+    ff_er_add_slice(s, 0, s->start_mb_y, s->mb_width - 1, s->end_mb_y - 1, (AC_END|DC_END|MV_END));
 }
 
 static void vc1_decode_skip_blocks(VC1Context *v)
@@ -3064,17 +3242,21 @@ static void vc1_decode_skip_blocks(VC1Context *v)
     s->pict_type = AV_PICTURE_TYPE_P;
 }
 
-static void vc1_decode_blocks(VC1Context *v, int mby_start, int mby_end)
+static void vc1_decode_blocks(VC1Context *v)
 {
 
     v->s.esc3_level_length = 0;
     if(v->x8_type){
         ff_intrax8_decode_picture(&v->x8, 2*v->pq+v->halfpq, v->pq*(!v->pquantizer) );
     }else{
+        v->cur_blk_idx = 0;
+        v->left_blk_idx = -1;
+        v->topleft_blk_idx = 1;
+        v->top_blk_idx = 2;
         switch(v->s.pict_type) {
         case AV_PICTURE_TYPE_I:
             if(v->profile == PROFILE_ADVANCED)
-                vc1_decode_i_blocks_adv(v, mby_start, mby_end);
+                vc1_decode_i_blocks_adv(v);
             else
                 vc1_decode_i_blocks(v);
             break;
@@ -3082,16 +3264,16 @@ static void vc1_decode_blocks(VC1Context *v, int mby_start, int mby_end)
             if(v->p_frame_skipped)
                 vc1_decode_skip_blocks(v);
             else
-                vc1_decode_p_blocks(v, mby_start, mby_end);
+                vc1_decode_p_blocks(v);
             break;
         case AV_PICTURE_TYPE_B:
             if(v->bi_type){
                 if(v->profile == PROFILE_ADVANCED)
-                    vc1_decode_i_blocks_adv(v, mby_start, mby_end);
+                    vc1_decode_i_blocks_adv(v);
                 else
                     vc1_decode_i_blocks(v);
             }else
-                vc1_decode_b_blocks(v, mby_start, mby_end);
+                vc1_decode_b_blocks(v);
             break;
         }
     }
@@ -3341,6 +3523,8 @@ static av_cold int vc1_decode_init(AVCodecContext *avctx)
     v->acpred_plane = av_malloc(s->mb_stride * s->mb_height);
     v->over_flags_plane = av_malloc(s->mb_stride * s->mb_height);
 
+    v->n_allocated_blks = s->mb_width + 2;
+    v->block = av_malloc(sizeof(*v->block) * v->n_allocated_blks);
     v->cbp_base = av_malloc(sizeof(v->cbp_base[0]) * 2 * s->mb_stride);
     v->cbp = v->cbp_base + s->mb_stride;
     v->ttblk_base = av_malloc(sizeof(v->ttblk_base[0]) * 2 * s->mb_stride);
@@ -3556,8 +3740,9 @@ static int vc1_decode_frame(AVCodecContext *avctx,
         for (i = 0; i <= n_slices; i++) {
             if (i && get_bits1(&s->gb))
                 vc1_parse_frame_header_adv(v, &s->gb);
-            vc1_decode_blocks(v, i == 0 ? 0 : FFMAX(0, slices[i-1].mby_start),
-                i == n_slices ? s->mb_height : FFMIN(s->mb_height, slices[i].mby_start));
+            s->start_mb_y = (i == 0)        ? 0 : FFMAX(0, slices[i-1].mby_start);
+            s->end_mb_y   = (i == n_slices) ? s->mb_height : FFMIN(s->mb_height, slices[i].mby_start);
+            vc1_decode_blocks(v);
             if (i != n_slices) s->gb = slices[i].gb;
         }
 //av_log(s->avctx, AV_LOG_INFO, "Consumed %i/%i bits\n", get_bits_count(&s->gb), s->gb.size_in_bits);
@@ -3614,6 +3799,7 @@ static av_cold int vc1_decode_end(AVCodecContext *avctx)
     av_freep(&v->acpred_plane);
     av_freep(&v->over_flags_plane);
     av_freep(&v->mb_type_base);
+    av_freep(&v->block);
     av_freep(&v->cbp_base);
     av_freep(&v->ttblk_base);
     av_freep(&v->is_intra_base); // FIXME use v->mb_type[]
diff --git a/libavcodec/vc1dsp.c b/libavcodec/vc1dsp.c
index dbe2120829..2eaa47a05b 100644
--- a/libavcodec/vc1dsp.c
+++ b/libavcodec/vc1dsp.c
@@ -78,6 +78,58 @@ static void vc1_h_overlap_c(uint8_t* src, int stride)
     }
 }
 
+static void vc1_v_s_overlap_c(DCTELEM *top,  DCTELEM *bottom)
+{
+    int i;
+    int a, b, c, d;
+    int d1, d2;
+    int rnd1 = 4, rnd2 = 3;
+    for(i = 0; i < 8; i++) {
+        a = top[48];
+        b = top[56];
+        c = bottom[0];
+        d = bottom[8];
+        d1 = a - d;
+        d2 = a - d + b - c;
+
+        top[48]   = ((a << 3) - d1 + rnd1) >> 3;
+        top[56]   = ((b << 3) - d2 + rnd2) >> 3;
+        bottom[0] = ((c << 3) + d2 + rnd1) >> 3;
+        bottom[8] = ((d << 3) + d1 + rnd2) >> 3;
+
+        bottom++;
+        top++;
+        rnd2 = 7 - rnd2;
+        rnd1 = 7 - rnd1;
+    }
+}
+
+static void vc1_h_s_overlap_c(DCTELEM *left, DCTELEM *right)
+{
+    int i;
+    int a, b, c, d;
+    int d1, d2;
+    int rnd1 = 4, rnd2 = 3;
+    for(i = 0; i < 8; i++) {
+        a = left[6];
+        b = left[7];
+        c = right[0];
+        d = right[1];
+        d1 = a - d;
+        d2 = a - d + b - c;
+
+        left[6]  = ((a << 3) - d1 + rnd1) >> 3;
+        left[7]  = ((b << 3) - d2 + rnd2) >> 3;
+        right[0] = ((c << 3) + d2 + rnd1) >> 3;
+        right[1] = ((d << 3) + d1 + rnd2) >> 3;
+
+        right += 8;
+        left += 8;
+        rnd2 = 7 - rnd2;
+        rnd1 = 7 - rnd1;
+    }
+}
+
 /**
  * VC-1 in-loop deblocking filter for one line
  * @param src source block type
@@ -199,7 +251,7 @@ static void vc1_inv_trans_8x8_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
     }
 }
 
-static av_always_inline void vc1_inv_trans_8x8_c(DCTELEM block[64], int shl, int sub)
+static void vc1_inv_trans_8x8_c(DCTELEM block[64])
 {
     int i;
     register int t1,t2,t3,t4,t5,t6,t7,t8;
@@ -254,50 +306,20 @@ static av_always_inline void vc1_inv_trans_8x8_c(DCTELEM block[64], int shl, int
         t3 =  9 * src[ 8] - 16 * src[24] +  4 * src[40] + 15 * src[56];
         t4 =  4 * src[ 8] -  9 * src[24] + 15 * src[40] - 16 * src[56];
 
-        dst[ 0] = (((t5 + t1    ) >> 7) - sub) << shl;
-        dst[ 8] = (((t6 + t2    ) >> 7) - sub) << shl;
-        dst[16] = (((t7 + t3    ) >> 7) - sub) << shl;
-        dst[24] = (((t8 + t4    ) >> 7) - sub) << shl;
-        dst[32] = (((t8 - t4 + 1) >> 7) - sub) << shl;
-        dst[40] = (((t7 - t3 + 1) >> 7) - sub) << shl;
-        dst[48] = (((t6 - t2 + 1) >> 7) - sub) << shl;
-        dst[56] = (((t5 - t1 + 1) >> 7) - sub) << shl;
+        dst[ 0] = (t5 + t1) >> 7;
+        dst[ 8] = (t6 + t2) >> 7;
+        dst[16] = (t7 + t3) >> 7;
+        dst[24] = (t8 + t4) >> 7;
+        dst[32] = (t8 - t4 + 1) >> 7;
+        dst[40] = (t7 - t3 + 1) >> 7;
+        dst[48] = (t6 - t2 + 1) >> 7;
+        dst[56] = (t5 - t1 + 1) >> 7;
 
         src++;
         dst++;
     }
 }
 
-static void vc1_inv_trans_8x8_add_c(uint8_t *dest, int linesize, DCTELEM *block)
-{
-    vc1_inv_trans_8x8_c(block, 0, 0);
-    ff_add_pixels_clamped_c(block, dest, linesize);
-}
-
-static void vc1_inv_trans_8x8_put_signed_c(uint8_t *dest, int linesize, DCTELEM *block)
-{
-    vc1_inv_trans_8x8_c(block, 0, 0);
-    ff_put_signed_pixels_clamped_c(block, dest, linesize);
-}
-
-static void vc1_inv_trans_8x8_put_signed_rangered_c(uint8_t *dest, int linesize, DCTELEM *block)
-{
-    vc1_inv_trans_8x8_c(block, 1, 0);
-    ff_put_signed_pixels_clamped_c(block, dest, linesize);
-}
-
-static void vc1_inv_trans_8x8_put_c(uint8_t *dest, int linesize, DCTELEM *block)
-{
-    vc1_inv_trans_8x8_c(block, 0, 0);
-    ff_put_pixels_clamped_c(block, dest, linesize);
-}
-
-static void vc1_inv_trans_8x8_put_rangered_c(uint8_t *dest, int linesize, DCTELEM *block)
-{
-    vc1_inv_trans_8x8_c(block, 1, 64);
-    ff_put_pixels_clamped_c(block, dest, linesize);
-}
-
 /** Do inverse transform on 8x4 part of block
 */
 static void vc1_inv_trans_8x4_dc_c(uint8_t *dest, int linesize, DCTELEM *block)
@@ -692,11 +714,7 @@ static void avg_no_rnd_vc1_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*a
 }
 
 av_cold void ff_vc1dsp_init(VC1DSPContext* dsp) {
-    dsp->vc1_inv_trans_8x8_add = vc1_inv_trans_8x8_add_c;
-    dsp->vc1_inv_trans_8x8_put_signed[0] = vc1_inv_trans_8x8_put_signed_c;
-    dsp->vc1_inv_trans_8x8_put_signed[1] = vc1_inv_trans_8x8_put_signed_rangered_c;
-    dsp->vc1_inv_trans_8x8_put[0] = vc1_inv_trans_8x8_put_c;
-    dsp->vc1_inv_trans_8x8_put[1] = vc1_inv_trans_8x8_put_rangered_c;
+    dsp->vc1_inv_trans_8x8 = vc1_inv_trans_8x8_c;
     dsp->vc1_inv_trans_4x8 = vc1_inv_trans_4x8_c;
     dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_c;
     dsp->vc1_inv_trans_4x4 = vc1_inv_trans_4x4_c;
@@ -706,6 +724,8 @@ av_cold void ff_vc1dsp_init(VC1DSPContext* dsp) {
     dsp->vc1_inv_trans_4x4_dc = vc1_inv_trans_4x4_dc_c;
     dsp->vc1_h_overlap = vc1_h_overlap_c;
     dsp->vc1_v_overlap = vc1_v_overlap_c;
+    dsp->vc1_h_s_overlap = vc1_h_s_overlap_c;
+    dsp->vc1_v_s_overlap = vc1_v_s_overlap_c;
     dsp->vc1_v_loop_filter4 = vc1_v_loop_filter4_c;
     dsp->vc1_h_loop_filter4 = vc1_h_loop_filter4_c;
     dsp->vc1_v_loop_filter8 = vc1_v_loop_filter8_c;
diff --git a/libavcodec/vc1dsp.h b/libavcodec/vc1dsp.h
index db9d892a23..93a9ea3858 100644
--- a/libavcodec/vc1dsp.h
+++ b/libavcodec/vc1dsp.h
@@ -30,13 +30,9 @@
 
 #include "dsputil.h"
 
-typedef void (*vc1_idct_func)(uint8_t *dest, int line_size, DCTELEM *block);
-
 typedef struct VC1DSPContext {
     /* vc1 functions */
-    vc1_idct_func vc1_inv_trans_8x8_add;
-    vc1_idct_func vc1_inv_trans_8x8_put_signed[2];
-    vc1_idct_func vc1_inv_trans_8x8_put[2];
+    void (*vc1_inv_trans_8x8)(DCTELEM *b);
     void (*vc1_inv_trans_8x4)(uint8_t *dest, int line_size, DCTELEM *block);
     void (*vc1_inv_trans_4x8)(uint8_t *dest, int line_size, DCTELEM *block);
     void (*vc1_inv_trans_4x4)(uint8_t *dest, int line_size, DCTELEM *block);
@@ -44,8 +40,10 @@ typedef struct VC1DSPContext {
     void (*vc1_inv_trans_8x4_dc)(uint8_t *dest, int line_size, DCTELEM *block);
     void (*vc1_inv_trans_4x8_dc)(uint8_t *dest, int line_size, DCTELEM *block);
     void (*vc1_inv_trans_4x4_dc)(uint8_t *dest, int line_size, DCTELEM *block);
-    void (*vc1_v_overlap)(uint8_t* src, int stride);
-    void (*vc1_h_overlap)(uint8_t* src, int stride);
+    void (*vc1_v_overlap)(uint8_t *src, int stride);
+    void (*vc1_h_overlap)(uint8_t *src, int stride);
+    void (*vc1_v_s_overlap)(DCTELEM *top,  DCTELEM *bottom);
+    void (*vc1_h_s_overlap)(DCTELEM *left, DCTELEM *right);
     void (*vc1_v_loop_filter4)(uint8_t *src, int stride, int pq);
     void (*vc1_h_loop_filter4)(uint8_t *src, int stride, int pq);
     void (*vc1_v_loop_filter8)(uint8_t *src, int stride, int pq);