diff --git a/configure b/configure
index aa45fd77ff..00f2ebe83a 100755
--- a/configure
+++ b/configure
@@ -276,6 +276,8 @@ Developer options (useful when working on FFmpeg itself):
                            Cannot be combined with --target-exec
   --samples=PATH           location of test samples for FATE, if not set use
                            \$FATE_SAMPLES at make invocation time.
+  --enable-xmm-clobber-test check XMM registers for clobbering (Win64-only;
+                           should be used only for debugging purposes)
 
 NOTE: Object files are built at the place where configure is launched.
 EOF
@@ -1085,6 +1087,7 @@ CONFIG_LIST="
     vda
     vdpau
     version3
+    xmm_clobber_test
     x11grab
     zlib
 "
@@ -1779,7 +1782,7 @@ test_deps _muxer _demuxer                                               \
     wav                                                                 \
     yuv4mpegpipe=yuv4mpeg                                               \
 
-ac3_fixed_test_deps="ac3_fixed_encoder ac3_decoder rm_muxer rm_demuxer"
+ac3_fixed_test_deps="ac3_fixed_encoder ac3_decoder"
 mpg_test_deps="mpeg1system_muxer mpegps_demuxer"
 
 # default parameters
@@ -3304,6 +3307,17 @@ check_ldflags -Wl,--warn-common
 check_ldflags -Wl,-rpath-link=libpostproc:libswresample:libswscale:libavfilter:libavdevice:libavformat:libavcodec:libavutil
 test_ldflags -Wl,-Bsymbolic && append SHFLAGS -Wl,-Bsymbolic
 
+enabled xmm_clobber_test &&                             \
+    check_ldflags -Wl,--wrap,avcodec_open2              \
+                  -Wl,--wrap,avcodec_decode_audio4      \
+                  -Wl,--wrap,avcodec_decode_video2      \
+                  -Wl,--wrap,avcodec_decode_subtitle2   \
+                  -Wl,--wrap,avcodec_encode_audio2      \
+                  -Wl,--wrap,avcodec_encode_video       \
+                  -Wl,--wrap,avcodec_encode_subtitle    \
+                  -Wl,--wrap,sws_scale ||               \
+    disable xmm_clobber_test
+
 echo "X{};" > $TMPV
 if test_ldflags -Wl,--version-script,$TMPV; then
     append SHFLAGS '-Wl,--version-script,\$(SUBDIR)lib\$(NAME).ver'
diff --git a/doc/APIchanges b/doc/APIchanges
index bc68cb0b32..d6e4decdf3 100644
--- a/doc/APIchanges
+++ b/doc/APIchanges
@@ -19,18 +19,18 @@ API changes, most recent first:
 2012-01-24 - xxxxxxx - lavfi 2.60.100
   Add avfilter_graph_dump.
 
-2012-02-01 - xxxxxxx - lavc 54.01.0
+2012-02-01 - 316fc74 - lavc 54.01.0
   Add av_fast_padded_malloc() as alternative for av_realloc() when aligned
   memory is required. The buffer will always have FF_INPUT_BUFFER_PADDING_SIZE
   zero-padded bytes at the end.
 
-2012-01-31 - xxxxxxx - lavf 54.01.0
+2012-01-31 - dd6d3b0 - lavf 54.01.0
   Add avformat_get_riff_video_tags() and avformat_get_riff_audio_tags().
 
-2012-01-31 - xxxxxxx - lavc 54.01.0
+2012-01-31 - af08d9a - lavc 54.01.0
   Add avcodec_is_open() function.
 
-2012-01-30 - xxxxxxx - lavu 51.22.0 - intfloat.h
+2012-01-30 - 8b93312 - lavu 51.22.0 - intfloat.h
   Add a new installed header libavutil/intfloat.h with int/float punning
   functions.
 
diff --git a/doc/general.texi b/doc/general.texi
index a1838296b5..dbf8775b71 100644
--- a/doc/general.texi
+++ b/doc/general.texi
@@ -497,6 +497,7 @@ following image formats are supported:
 @item Flash Screen Video v2  @tab  X  @tab  X
 @item Flash Video (FLV)      @tab  X  @tab  X
     @tab Sorenson H.263 used in Flash
+@item Forward Uncompressed   @tab     @tab  X
 @item Fraps                  @tab     @tab  X
 @item H.261                  @tab  X  @tab  X
 @item H.263 / H.263-1996     @tab  X  @tab  X
diff --git a/libavcodec/arm/ac3dsp_armv6.S b/libavcodec/arm/ac3dsp_armv6.S
index b6aee867b3..df8bfbaa03 100644
--- a/libavcodec/arm/ac3dsp_armv6.S
+++ b/libavcodec/arm/ac3dsp_armv6.S
@@ -34,24 +34,23 @@ function ff_ac3_bit_alloc_calc_bap_armv6, export=1
         add             r0,  r0,  r4,  lsl #1           @ mask + band
         add             r4,  lr,  r4
         add             r7,  r7,  r2                    @ bap + start
-        ldrb            r10, [r4], #1
 1:
         ldrsh           r9,  [r0], #2                   @ mask[band]
         mov             r8,  #0xff0
         sub             r9,  r9,  r12                   @   - snr_offset
-        mov             r11, r10
-        ldrb            r10, [r4], #1                   @ band_start_tab[band++]
+        ldrb            r10, [r4, #1]!                  @ band_start_tab[++band]
         subs            r9,  r9,  r5                    @   - floor
         it              lt
         movlt           r9,  #0
         cmp             r10, r3                         @   - end
         and             r9,  r9,  r8, lsl #1            @   & 0x1fe0
         ite             gt
-        subgt           r8,  r3,  r11
-        suble           r8,  r10, r11
+        subgt           r8,  r3,  r2
+        suble           r8,  r10, r2
+        mov             r2,  r10
         add             r9,  r9,  r5                    @   + floor => m
         tst             r8,  #1
-        add             r2,  r7,  r8
+        add             r11, r7,  r8
         bne             3f
         b               5f
 2:
@@ -65,9 +64,9 @@ function ff_ac3_bit_alloc_calc_bap_armv6, export=1
         ldrb            lr,  [r6, lr]
         strb            r8,  [r7], #1                   @ bap[bin]
         strb            lr,  [r7], #1
-5:      cmp             r7,  r2
+5:      cmp             r7,  r11
         blo             2b
-        cmp             r3,  r11
+        cmp             r3,  r10
         bgt             1b
         pop             {r4-r11,pc}
 3:
diff --git a/libavcodec/bethsoftvideo.c b/libavcodec/bethsoftvideo.c
index d80ac166a5..d85783b109 100644
--- a/libavcodec/bethsoftvideo.c
+++ b/libavcodec/bethsoftvideo.c
@@ -61,7 +61,7 @@ static int set_palette(BethsoftvidContext *ctx)
         palette[a] |= palette[a] >> 6 & 0x30303;
     }
     ctx->frame.palette_has_changed = 1;
-    return 256*3;
+    return 0;
 }
 
 static int bethsoftvid_decode_frame(AVCodecContext *avctx,
@@ -88,7 +88,13 @@ static int bethsoftvid_decode_frame(AVCodecContext *avctx,
 
     switch(block_type = bytestream2_get_byte(&vid->g)){
         case PALETTE_BLOCK: {
-            return set_palette(vid);
+            int ret;
+            *data_size = 0;
+            if ((ret = set_palette(vid)) < 0) {
+                av_log(avctx, AV_LOG_ERROR, "error reading palette\n");
+                return ret;
+            }
+            return bytestream2_tell(&vid->g);
         }
         case VIDEO_YOFF_P_FRAME:
             yoffset = bytestream2_get_le16(&vid->g);
diff --git a/libavcodec/internal.h b/libavcodec/internal.h
index 72a89441c2..b7d4a6e139 100644
--- a/libavcodec/internal.h
+++ b/libavcodec/internal.h
@@ -130,6 +130,7 @@ int avpriv_unlock_avformat(void);
  *                If avpkt->data is already set, avpkt->size is checked
  *                to ensure it is large enough.
  *                If avpkt->data is NULL, a new buffer is allocated.
+ *                avpkt->size is set to the specified size.
  *                All other AVPacket fields will be reset with av_init_packet().
  * @param size    the minimum required packet size
  * @return        0 on success, negative error code on failure
diff --git a/libavcodec/libx264.c b/libavcodec/libx264.c
index 7f817cbb4c..1380e0a438 100644
--- a/libavcodec/libx264.c
+++ b/libavcodec/libx264.c
@@ -188,12 +188,12 @@ static int X264_frame(AVCodecContext *ctx, uint8_t *buf,
 
     do {
         bufsize = orig_bufsize;
-    if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0)
-        return -1;
+        if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0)
+            return -1;
 
-    bufsize = encode_nals(ctx, buf, bufsize, nal, nnal, 0);
-    if (bufsize < 0)
-        return -1;
+        bufsize = encode_nals(ctx, buf, bufsize, nal, nnal, 0);
+        if (bufsize < 0)
+            return -1;
     } while (!bufsize && !frame && x264_encoder_delayed_frames(x4->enc));
 
     /* FIXME: libx264 now provides DTS, but AVFrame doesn't have a field for it. */
diff --git a/libavcodec/mpc.h b/libavcodec/mpc.h
index 808739fcc9..8b4deef689 100644
--- a/libavcodec/mpc.h
+++ b/libavcodec/mpc.h
@@ -66,8 +66,6 @@ typedef struct {
     int buf_size;
     AVLFG rnd;
     int frames_to_skip;
-    uint8_t *buffer;
-    int buffer_size;
     /* for synthesis */
     DECLARE_ALIGNED(16, MPA_INT, synth_buf)[MPA_MAX_CHANNELS][512*2];
     int synth_buf_offset[MPA_MAX_CHANNELS];
diff --git a/libavcodec/mpc7.c b/libavcodec/mpc7.c
index c60a621c65..5693c4ddb2 100644
--- a/libavcodec/mpc7.c
+++ b/libavcodec/mpc7.c
@@ -200,34 +200,46 @@ static int mpc7_decode_frame(AVCodecContext * avctx, void *data,
                              int *got_frame_ptr, AVPacket *avpkt)
 {
     const uint8_t *buf = avpkt->data;
-    int buf_size = avpkt->size;
+    int buf_size;
     MPCContext *c = avctx->priv_data;
     GetBitContext gb;
     int i, ch;
     int mb = -1;
     Band *bands = c->bands;
-    int off, ret;
+    int off, ret, last_frame, skip;
     int bits_used, bits_avail;
 
     memset(bands, 0, sizeof(*bands) * (c->maxbands + 1));
-    if(buf_size <= 4){
-        av_log(avctx, AV_LOG_ERROR, "Too small buffer passed (%i bytes)\n", buf_size);
-        return AVERROR(EINVAL);
+
+    buf_size = avpkt->size & ~3;
+    if (buf_size <= 0) {
+        av_log(avctx, AV_LOG_ERROR, "packet size is too small (%i bytes)\n",
+               avpkt->size);
+        return AVERROR_INVALIDDATA;
+    }
+    if (buf_size != avpkt->size) {
+        av_log(avctx, AV_LOG_WARNING, "packet size is not a multiple of 4. "
+               "extra bytes at the end will be skipped.\n");
     }
 
+    skip       = buf[0];
+    last_frame = buf[1];
+    buf       += 4;
+    buf_size  -= 4;
+
     /* get output buffer */
-    c->frame.nb_samples = buf[1] ? c->lastframelen : MPC_FRAME_SIZE;
+    c->frame.nb_samples = last_frame ? c->lastframelen : MPC_FRAME_SIZE;
     if ((ret = avctx->get_buffer(avctx, &c->frame)) < 0) {
         av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
         return ret;
     }
 
-    av_fast_padded_malloc(&c->buffer, &c->buffer_size, FFALIGN(buf_size - 1, 4));
-    if (!c->buffer)
+    av_fast_padded_malloc(&c->bits, &c->buf_size, buf_size);
+    if (!c->bits)
         return AVERROR(ENOMEM);
-    c->dsp.bswap_buf((uint32_t*)c->buffer, (const uint32_t*)(buf + 4), (buf_size - 4) >> 2);
-    init_get_bits(&gb, c->buffer, (buf_size - 4)* 8);
-    skip_bits_long(&gb, buf[0]);
+    c->dsp.bswap_buf((uint32_t *)c->bits, (const uint32_t *)buf, buf_size >> 2);
+    init_get_bits(&gb, c->bits, buf_size * 8);
+    skip_bits_long(&gb, skip);
 
     /* read subband indexes */
     for(i = 0; i <= c->maxbands; i++){
@@ -284,21 +296,21 @@ static int mpc7_decode_frame(AVCodecContext * avctx, void *data,
     ff_mpc_dequantize_and_synth(c, mb, c->frame.data[0], 2);
 
     bits_used = get_bits_count(&gb);
-    bits_avail = (buf_size - 4) * 8;
-    if(!buf[1] && ((bits_avail < bits_used) || (bits_used + 32 <= bits_avail))){
+    bits_avail = buf_size * 8;
+    if (!last_frame && ((bits_avail < bits_used) || (bits_used + 32 <= bits_avail))) {
         av_log(NULL,0, "Error decoding frame: used %i of %i bits\n", bits_used, bits_avail);
         return -1;
     }
     if(c->frames_to_skip){
         c->frames_to_skip--;
         *got_frame_ptr = 0;
-        return buf_size;
+        return avpkt->size;
     }
 
     *got_frame_ptr   = 1;
     *(AVFrame *)data = c->frame;
 
-    return buf_size;
+    return avpkt->size;
 }
 
 static void mpc7_decode_flush(AVCodecContext *avctx)
@@ -312,8 +324,8 @@ static void mpc7_decode_flush(AVCodecContext *avctx)
 static av_cold int mpc7_decode_close(AVCodecContext *avctx)
 {
     MPCContext *c = avctx->priv_data;
-    av_freep(&c->buffer);
-    c->buffer_size = 0;
+    av_freep(&c->bits);
+    c->buf_size = 0;
     return 0;
 }
 
diff --git a/libavcodec/pcm.c b/libavcodec/pcm.c
index d8b926d33e..1916c2f938 100644
--- a/libavcodec/pcm.c
+++ b/libavcodec/pcm.c
@@ -194,7 +194,6 @@ static int pcm_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
         return -1;
     }
 
-    avpkt->size = frame->nb_samples * avctx->channels * sample_size;
     *got_packet_ptr = 1;
     return 0;
 }
diff --git a/libavcodec/ra144enc.c b/libavcodec/ra144enc.c
index 725abc2f38..91bf7e174f 100644
--- a/libavcodec/ra144enc.c
+++ b/libavcodec/ra144enc.c
@@ -521,5 +521,5 @@ AVCodec ff_ra_144_encoder = {
     .close          = ra144_encode_close,
     .sample_fmts    = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S16,
                                                      AV_SAMPLE_FMT_NONE },
-    .long_name      = NULL_IF_CONFIG_SMALL("RealAudio 1.0 (14.4K) encoder"),
+    .long_name      = NULL_IF_CONFIG_SMALL("RealAudio 1.0 (14.4K)"),
 };
diff --git a/libavcodec/utils.c b/libavcodec/utils.c
index f21e36fd9c..74206720ca 100644
--- a/libavcodec/utils.c
+++ b/libavcodec/utils.c
@@ -919,16 +919,14 @@ int ff_alloc_packet(AVPacket *avpkt, int size)
 
     if (avpkt->data) {
         uint8_t *pkt_data;
-        int pkt_size;
 
         if (avpkt->size < size)
             return AVERROR(EINVAL);
 
         pkt_data = avpkt->data;
-        pkt_size = avpkt->size;
         av_init_packet(avpkt);
         avpkt->data = pkt_data;
-        avpkt->size = pkt_size;
+        avpkt->size = size;
         return 0;
     } else {
         return av_new_packet(avpkt, size);
diff --git a/libavcodec/vorbis.c b/libavcodec/vorbis.c
index 28176f3b12..fac8d0b2cd 100644
--- a/libavcodec/vorbis.c
+++ b/libavcodec/vorbis.c
@@ -156,7 +156,7 @@ void ff_vorbis_ready_floor1_list(vorbis_floor1_entry * list, int values)
     }
 }
 
-static inline void render_line_unrolled(intptr_t x, uint8_t y, int x1,
+static inline void render_line_unrolled(intptr_t x, int y, int x1,
                                         intptr_t sy, int ady, int adx,
                                         float *buf)
 {
@@ -168,30 +168,30 @@ static inline void render_line_unrolled(intptr_t x, uint8_t y, int x1,
         if (err >= 0) {
             err += ady - adx;
             y   += sy;
-            buf[x++] = ff_vorbis_floor1_inverse_db_table[y];
+            buf[x++] = ff_vorbis_floor1_inverse_db_table[av_clip_uint8(y)];
         }
-        buf[x] = ff_vorbis_floor1_inverse_db_table[y];
+        buf[x] = ff_vorbis_floor1_inverse_db_table[av_clip_uint8(y)];
     }
     if (x <= 0) {
         if (err + ady >= 0)
             y += sy;
-        buf[x] = ff_vorbis_floor1_inverse_db_table[y];
+        buf[x] = ff_vorbis_floor1_inverse_db_table[av_clip_uint8(y)];
     }
 }
 
-static void render_line(int x0, uint8_t y0, int x1, int y1, float *buf)
+static void render_line(int x0, int y0, int x1, int y1, float *buf)
 {
     int dy  = y1 - y0;
     int adx = x1 - x0;
     int ady = FFABS(dy);
     int sy  = dy < 0 ? -1 : 1;
-    buf[x0] = ff_vorbis_floor1_inverse_db_table[y0];
+    buf[x0] = ff_vorbis_floor1_inverse_db_table[av_clip_uint8(y0)];
     if (ady*2 <= adx) { // optimized common case
         render_line_unrolled(x0, y0, x1, sy, ady, adx, buf);
     } else {
         int base  = dy / adx;
         int x     = x0;
-        uint8_t y = y0;
+        int y     = y0;
         int err   = -adx;
         ady -= FFABS(base) * adx;
         while (++x < x1) {
@@ -201,7 +201,7 @@ static void render_line(int x0, uint8_t y0, int x1, int y1, float *buf)
                 err -= adx;
                 y   += sy;
             }
-            buf[x] = ff_vorbis_floor1_inverse_db_table[y];
+            buf[x] = ff_vorbis_floor1_inverse_db_table[av_clip_uint8(y)];
         }
     }
 }
@@ -210,8 +210,7 @@ void ff_vorbis_floor1_render_list(vorbis_floor1_entry * list, int values,
                                   uint16_t *y_list, int *flag,
                                   int multiplier, float *out, int samples)
 {
-    int lx, i;
-    uint8_t ly;
+    int lx, ly, i;
     lx = 0;
     ly = y_list[0] * multiplier;
     for (i = 1; i < values; i++) {
diff --git a/libavcodec/vorbisdec.c b/libavcodec/vorbisdec.c
index f1d9a79297..f71d606dad 100644
--- a/libavcodec/vorbisdec.c
+++ b/libavcodec/vorbisdec.c
@@ -1256,20 +1256,20 @@ static int vorbis_floor1_decode(vorbis_context *vc,
             floor1_flag[i]               = 1;
             if (val >= room) {
                 if (highroom > lowroom) {
-                    floor1_Y_final[i] = val - lowroom + predicted;
+                    floor1_Y_final[i] = av_clip_uint16(val - lowroom + predicted);
                 } else {
-                    floor1_Y_final[i] = predicted - val + highroom - 1;
+                    floor1_Y_final[i] = av_clip_uint16(predicted - val + highroom - 1);
                 }
             } else {
                 if (val & 1) {
-                    floor1_Y_final[i] = predicted - (val + 1) / 2;
+                    floor1_Y_final[i] = av_clip_uint16(predicted - (val + 1) / 2);
                 } else {
-                    floor1_Y_final[i] = predicted + val / 2;
+                    floor1_Y_final[i] = av_clip_uint16(predicted + val / 2);
                 }
             }
         } else {
             floor1_flag[i]    = 0;
-            floor1_Y_final[i] = predicted;
+            floor1_Y_final[i] = av_clip_uint16(predicted);
         }
 
         av_dlog(NULL, " Decoded floor(%d) = %u / val %u\n",
diff --git a/libavcodec/x86/Makefile b/libavcodec/x86/Makefile
index 3b8ee56a49..dc8c66afde 100644
--- a/libavcodec/x86/Makefile
+++ b/libavcodec/x86/Makefile
@@ -83,3 +83,4 @@ OBJS-$(HAVE_MMX)                       += x86/dsputil_mmx.o             \
                                           x86/mpegvideo_mmx.o           \
                                           x86/simple_idct_mmx.o         \
 
+OBJS-$(CONFIG_XMM_CLOBBER_TEST)        += x86/w64xmmtest.o
diff --git a/libavcodec/x86/dsputil_yasm.asm b/libavcodec/x86/dsputil_yasm.asm
index ad9ec2c339..966344f7c7 100644
--- a/libavcodec/x86/dsputil_yasm.asm
+++ b/libavcodec/x86/dsputil_yasm.asm
@@ -1063,7 +1063,7 @@ emu_edge mmx
 ; %4 = CLIPD function takes min/max as float instead of int (CLIPD_SSE2)
 ; %5 = suffix
 %macro VECTOR_CLIP_INT32 4-5
-cglobal vector_clip_int32%5, 5,5,%2, dst, src, min, max, len
+cglobal vector_clip_int32%5, 5,5,%1, dst, src, min, max, len
 %if %4
     cvtsi2ss  m4, minm
     cvtsi2ss  m5, maxm
diff --git a/libavcodec/x86/w64xmmtest.c b/libavcodec/x86/w64xmmtest.c
new file mode 100644
index 0000000000..f6e3de9496
--- /dev/null
+++ b/libavcodec/x86/w64xmmtest.c
@@ -0,0 +1,80 @@
+/*
+ * check XMM registers for clobbers on Win64
+ * Copyright (c) 2012 Ronald S. Bultje <rsbultje@gmail.com>
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavcodec/avcodec.h"
+#include "libavutil/x86/w64xmmtest.h"
+
+wrap(avcodec_open2(AVCodecContext *avctx,
+                   AVCodec *codec,
+                   AVDictionary **options))
+{
+    testxmmclobbers(avcodec_open2, avctx, codec, options);
+}
+
+wrap(avcodec_decode_audio4(AVCodecContext *avctx,
+                           AVFrame *frame,
+                           int *got_frame_ptr,
+                           AVPacket *avpkt))
+{
+    testxmmclobbers(avcodec_decode_audio4, avctx, frame,
+                    got_frame_ptr, avpkt);
+}
+
+wrap(avcodec_decode_video2(AVCodecContext *avctx,
+                           AVFrame *picture,
+                           int *got_picture_ptr,
+                           AVPacket *avpkt))
+{
+    testxmmclobbers(avcodec_decode_video2, avctx, picture,
+                    got_picture_ptr, avpkt);
+}
+
+wrap(avcodec_decode_subtitle2(AVCodecContext *avctx,
+                              AVSubtitle *sub,
+                              int *got_sub_ptr,
+                              AVPacket *avpkt))
+{
+    testxmmclobbers(avcodec_decode_subtitle2, avctx, sub,
+                    got_sub_ptr, avpkt);
+}
+
+wrap(avcodec_encode_audio2(AVCodecContext *avctx,
+                           AVPacket *avpkt,
+                           const AVFrame *frame,
+                           int *got_packet_ptr))
+{
+    testxmmclobbers(avcodec_encode_audio2, avctx, avpkt, frame,
+                    got_packet_ptr);
+}
+
+wrap(avcodec_encode_video(AVCodecContext *avctx,
+                          uint8_t *buf, int buf_size,
+                          const AVFrame *pict))
+{
+    testxmmclobbers(avcodec_encode_video, avctx, buf, buf_size, pict);
+}
+
+wrap(avcodec_encode_subtitle(AVCodecContext *avctx,
+                             uint8_t *buf, int buf_size,
+                             const AVSubtitle *sub))
+{
+    testxmmclobbers(avcodec_encode_subtitle, avctx, buf, buf_size, sub);
+}
diff --git a/libavformat/isom.h b/libavformat/isom.h
index 40455e5d75..eb9218ef45 100644
--- a/libavformat/isom.h
+++ b/libavformat/isom.h
@@ -129,6 +129,7 @@ typedef struct MOVStreamContext {
     int has_palette;
     int64_t data_size;
     uint32_t tmcd_flags;  ///< tmcd track flags
+    int64_t track_end;    ///< used for dts generation in fragmented movie files
 } MOVStreamContext;
 
 typedef struct MOVContext {
diff --git a/libavformat/mov.c b/libavformat/mov.c
index e345355d95..ddbe49083a 100644
--- a/libavformat/mov.c
+++ b/libavformat/mov.c
@@ -1012,6 +1012,32 @@ static int mov_read_glbl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
     return 0;
 }
 
+static int mov_read_dvc1(MOVContext *c, AVIOContext *pb, MOVAtom atom)
+{
+    AVStream *st;
+    uint8_t profile_level;
+
+    if (c->fc->nb_streams < 1)
+        return 0;
+    st = c->fc->streams[c->fc->nb_streams-1];
+
+    if (atom.size >= (1<<28) || atom.size < 7)
+        return AVERROR_INVALIDDATA;
+
+    profile_level = avio_r8(pb);
+    if (profile_level & 0xf0 != 0xc0)
+        return 0;
+
+    av_free(st->codec->extradata);
+    st->codec->extradata = av_mallocz(atom.size - 7 + FF_INPUT_BUFFER_PADDING_SIZE);
+    if (!st->codec->extradata)
+        return AVERROR(ENOMEM);
+    st->codec->extradata_size = atom.size - 7;
+    avio_seek(pb, 6, SEEK_CUR);
+    avio_read(pb, st->codec->extradata, st->codec->extradata_size);
+    return 0;
+}
+
 /**
  * An strf atom is a BITMAPINFOHEADER struct. This struct is 40 bytes itself,
  * but can have extradata appended at the end after the 40 bytes belonging
@@ -1706,6 +1732,7 @@ static int mov_read_stts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
     st->nb_frames= total_sample_count;
     if (duration)
         st->duration= duration;
+    sc->track_end = duration;
     return 0;
 }
 
@@ -2326,7 +2353,7 @@ static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
 
     if (flags & 0x001) data_offset        = avio_rb32(pb);
     if (flags & 0x004) first_sample_flags = avio_rb32(pb);
-    dts = st->duration - sc->time_offset;
+    dts    = sc->track_end - sc->time_offset;
     offset = frag->base_data_offset + data_offset;
     distance = 0;
     av_dlog(c->fc, "first sample flags 0x%x\n", first_sample_flags);
@@ -2356,7 +2383,7 @@ static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
         sc->data_size += sample_size;
     }
     frag->moof_offset = offset;
-    st->duration = dts + sc->time_offset;
+    st->duration = sc->track_end = dts + sc->time_offset;
     return 0;
 }
 
@@ -2538,6 +2565,7 @@ static const MOVParseTableEntry mov_default_parse_table[] = {
 { MKTAG('w','f','e','x'), mov_read_wfex },
 { MKTAG('c','m','o','v'), mov_read_cmov },
 { MKTAG('c','h','a','n'), mov_read_chan }, /* channel layout */
+{ MKTAG('d','v','c','1'), mov_read_dvc1 },
 { 0, NULL }
 };
 
diff --git a/libavutil/x86/w64xmmtest.h b/libavutil/x86/w64xmmtest.h
new file mode 100644
index 0000000000..1c1ded86ec
--- /dev/null
+++ b/libavutil/x86/w64xmmtest.h
@@ -0,0 +1,71 @@
+/*
+ * check XMM registers for clobbers on Win64
+ * Copyright (c) 2008 Ramiro Polla <ramiro.polla@gmail.com>
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <stdint.h>
+#include <stdlib.h>
+#include <stdarg.h>
+
+#include "libavutil/bswap.h"
+
+#define storexmmregs(mem)               \
+    __asm__ volatile(                   \
+        "movups %%xmm6 , 0x00(%0)\n\t"  \
+        "movups %%xmm7 , 0x10(%0)\n\t"  \
+        "movups %%xmm8 , 0x20(%0)\n\t"  \
+        "movups %%xmm9 , 0x30(%0)\n\t"  \
+        "movups %%xmm10, 0x40(%0)\n\t"  \
+        "movups %%xmm11, 0x50(%0)\n\t"  \
+        "movups %%xmm12, 0x60(%0)\n\t"  \
+        "movups %%xmm13, 0x70(%0)\n\t"  \
+        "movups %%xmm14, 0x80(%0)\n\t"  \
+        "movups %%xmm15, 0x90(%0)\n\t"  \
+        :: "r"(mem) : "memory")
+
+#define testxmmclobbers(func, ctx, ...)                         \
+    uint64_t xmm[2][10][2];                                     \
+    int ret;                                                    \
+    storexmmregs(xmm[0]);                                       \
+    ret = __real_ ## func(ctx, __VA_ARGS__);                    \
+    storexmmregs(xmm[1]);                                       \
+    if (memcmp(xmm[0], xmm[1], sizeof(xmm[0]))) {               \
+        int i;                                                  \
+        av_log(ctx, AV_LOG_ERROR,                               \
+               "XMM REGS CLOBBERED IN %s!\n", #func);           \
+        for (i = 0; i < 10; i ++)                               \
+            if (xmm[0][i][0] != xmm[1][i][0] ||                 \
+                xmm[0][i][1] != xmm[1][i][1]) {                 \
+                av_log(ctx, AV_LOG_ERROR,                       \
+                       "xmm%-2d = %016"PRIx64"%016"PRIx64"\n",  \
+                       6 + i, av_bswap64(xmm[0][i][0]),         \
+                       av_bswap64(xmm[0][i][1]));               \
+                av_log(ctx, AV_LOG_ERROR,                       \
+                         "     -> %016"PRIx64"%016"PRIx64"\n",  \
+                       av_bswap64(xmm[1][i][0]),                \
+                       av_bswap64(xmm[1][i][1]));               \
+            }                                                   \
+        abort();                                                \
+    }                                                           \
+    return ret
+
+#define wrap(func)      \
+int __real_ ## func;    \
+int __wrap_ ## func;    \
+int __wrap_ ## func
diff --git a/libswscale/Makefile b/libswscale/Makefile
index 77d896a76b..b761470fd1 100644
--- a/libswscale/Makefile
+++ b/libswscale/Makefile
@@ -25,6 +25,8 @@ MMX-OBJS-$(HAVE_YASM)      +=  x86/input.o              \
 
 $(SUBDIR)x86/swscale_mmx.o: CFLAGS += $(NOREDZONE_FLAGS)
 
+OBJS-$(CONFIG_XMM_CLOBBER_TEST) += x86/w64xmmtest.o
+
 TESTPROGS = colorspace swscale
 
 DIRS = bfin mlib ppc sparc x86
diff --git a/libswscale/bfin/internal_bfin.S b/libswscale/bfin/internal_bfin.S
index cb8d71253c..eab30aa6ce 100644
--- a/libswscale/bfin/internal_bfin.S
+++ b/libswscale/bfin/internal_bfin.S
@@ -30,11 +30,11 @@ and converts it to RGB565. R:5 bits, G:6 bits, B:5 bits.. packed into shorts.
 
 The following calculation is used for the conversion:
 
-  r = clipz((y-oy)*cy  + crv*(v-128))
-  g = clipz((y-oy)*cy  + cgv*(v-128) + cgu*(u-128))
-  b = clipz((y-oy)*cy  + cbu*(u-128))
+  r = clipz((y - oy) * cy  + crv * (v - 128))
+  g = clipz((y - oy) * cy  + cgv * (v - 128) + cgu * (u - 128))
+  b = clipz((y - oy) * cy  + cbu * (u - 128))
 
-y,u,v are prescaled by a factor of 4 i.e. left-shifted to gain precision.
+y, u, v are prescaled by a factor of 4 i.e. left-shifted to gain precision.
 
 
 New factorization to eliminate the truncation error which was
@@ -47,7 +47,7 @@ occurring due to the byteop3p.
 2) Scale operands up by a factor of 4 not 8 because Blackfin
    multiplies include a shift.
 
-3) Compute into the accumulators cy*yx0, cy*yx1.
+3) Compute into the accumulators cy * yx0, cy * yx1.
 
 4) Compute each of the linear equations:
      r = clipz((y - oy) * cy  + crv * (v - 128))
@@ -73,7 +73,7 @@ occurring due to the byteop3p.
 
 Where coeffs have the following layout in memory.
 
-uint32_t oy,oc,zero,cy,crv,rmask,cbu,bmask,cgu,cgv;
+uint32_t oy, oc, zero, cy, crv, rmask, cbu, bmask, cgu, cgv;
 
 coeffs is a pointer to oy.
 
diff --git a/libswscale/bfin/swscale_bfin.c b/libswscale/bfin/swscale_bfin.c
index 870636ea05..3cd4f28387 100644
--- a/libswscale/bfin/swscale_bfin.c
+++ b/libswscale/bfin/swscale_bfin.c
@@ -27,32 +27,34 @@
 #include <assert.h>
 #include "config.h"
 #include <unistd.h>
+
 #include "libswscale/rgb2rgb.h"
 #include "libswscale/swscale.h"
 #include "libswscale/swscale_internal.h"
 
 #if defined (__FDPIC__) && CONFIG_SRAM
-#define L1CODE __attribute__ ((l1_text))
+#define L1CODE __attribute__((l1_text))
 #else
 #define L1CODE
 #endif
 
-int ff_bfin_uyvytoyv12(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
-                       int width, int height,
+int ff_bfin_uyvytoyv12(const uint8_t *src, uint8_t *ydst, uint8_t *udst,
+                       uint8_t *vdst, int width, int height,
                        int lumStride, int chromStride, int srcStride) L1CODE;
 
-int ff_bfin_yuyvtoyv12(const uint8_t *src, uint8_t *ydst, uint8_t *udst, uint8_t *vdst,
-                       int width, int height,
+int ff_bfin_yuyvtoyv12(const uint8_t *src, uint8_t *ydst, uint8_t *udst,
+                       uint8_t *vdst, int width, int height,
                        int lumStride, int chromStride, int srcStride) L1CODE;
 
-static int uyvytoyv12_unscaled(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
-                               int srcSliceH, uint8_t* dst[], int dstStride[])
+static int uyvytoyv12_unscaled(SwsContext *c, uint8_t *src[], int srcStride[],
+                               int srcSliceY, int srcSliceH, uint8_t *dst[],
+                               int dstStride[])
 {
-    uint8_t *dsty = dst[0] + dstStride[0]*srcSliceY;
-    uint8_t *dstu = dst[1] + dstStride[1]*srcSliceY/2;
-    uint8_t *dstv = dst[2] + dstStride[2]*srcSliceY/2;
-    uint8_t *ip   = src[0] + srcStride[0]*srcSliceY;
-    int w         = dstStride[0];
+    uint8_t *dsty = dst[0] + dstStride[0] * srcSliceY;
+    uint8_t *dstu = dst[1] + dstStride[1] * srcSliceY / 2;
+    uint8_t *dstv = dst[2] + dstStride[2] * srcSliceY / 2;
+    uint8_t *ip   = src[0] + srcStride[0] * srcSliceY;
+    int w = dstStride[0];
 
     ff_bfin_uyvytoyv12(ip, dsty, dstu, dstv, w, srcSliceH,
                        dstStride[0], dstStride[1], srcStride[0]);
@@ -60,14 +62,15 @@ static int uyvytoyv12_unscaled(SwsContext *c, uint8_t* src[], int srcStride[], i
     return srcSliceH;
 }
 
-static int yuyvtoyv12_unscaled(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
-                               int srcSliceH, uint8_t* dst[], int dstStride[])
+static int yuyvtoyv12_unscaled(SwsContext *c, uint8_t *src[], int srcStride[],
+                               int srcSliceY, int srcSliceH, uint8_t *dst[],
+                               int dstStride[])
 {
-    uint8_t *dsty = dst[0] + dstStride[0]*srcSliceY;
-    uint8_t *dstu = dst[1] + dstStride[1]*srcSliceY/2;
-    uint8_t *dstv = dst[2] + dstStride[2]*srcSliceY/2;
-    uint8_t *ip   = src[0] + srcStride[0]*srcSliceY;
-    int w         = dstStride[0];
+    uint8_t *dsty = dst[0] + dstStride[0] * srcSliceY;
+    uint8_t *dstu = dst[1] + dstStride[1] * srcSliceY / 2;
+    uint8_t *dstv = dst[2] + dstStride[2] * srcSliceY / 2;
+    uint8_t *ip   = src[0] + srcStride[0] * srcSliceY;
+    int w = dstStride[0];
 
     ff_bfin_yuyvtoyv12(ip, dsty, dstu, dstv, w, srcSliceH,
                        dstStride[0], dstStride[1], srcStride[0]);
@@ -75,15 +78,16 @@ static int yuyvtoyv12_unscaled(SwsContext *c, uint8_t* src[], int srcStride[], i
     return srcSliceH;
 }
 
-
 void ff_bfin_get_unscaled_swscale(SwsContext *c)
 {
     if (c->dstFormat == PIX_FMT_YUV420P && c->srcFormat == PIX_FMT_UYVY422) {
-        av_log (NULL, AV_LOG_VERBOSE, "selecting Blackfin optimized uyvytoyv12_unscaled\n");
+        av_log(NULL, AV_LOG_VERBOSE,
+               "selecting Blackfin optimized uyvytoyv12_unscaled\n");
         c->swScale = uyvytoyv12_unscaled;
     }
     if (c->dstFormat == PIX_FMT_YUV420P && c->srcFormat == PIX_FMT_YUYV422) {
-        av_log (NULL, AV_LOG_VERBOSE, "selecting Blackfin optimized yuyvtoyv12_unscaled\n");
+        av_log(NULL, AV_LOG_VERBOSE,
+               "selecting Blackfin optimized yuyvtoyv12_unscaled\n");
         c->swScale = yuyvtoyv12_unscaled;
     }
 }
diff --git a/libswscale/bfin/yuv2rgb_bfin.c b/libswscale/bfin/yuv2rgb_bfin.c
index 7a7dc7f0e6..e7f657fe00 100644
--- a/libswscale/bfin/yuv2rgb_bfin.c
+++ b/libswscale/bfin/yuv2rgb_bfin.c
@@ -26,15 +26,16 @@
 #include <string.h>
 #include <inttypes.h>
 #include <assert.h>
-#include "config.h"
 #include <unistd.h>
 #include "libavutil/pixdesc.h"
+
+#include "config.h"
 #include "libswscale/rgb2rgb.h"
 #include "libswscale/swscale.h"
 #include "libswscale/swscale_internal.h"
 
 #if defined(__FDPIC__) && CONFIG_SRAM
-#define L1CODE __attribute__ ((l1_text))
+#define L1CODE __attribute__((l1_text))
 #else
 #define L1CODE
 #endif
@@ -48,21 +49,20 @@ void ff_bfin_yuv2rgb565_line(uint8_t *Y, uint8_t *U, uint8_t *V, uint8_t *out,
 void ff_bfin_yuv2rgb24_line(uint8_t *Y, uint8_t *U, uint8_t *V, uint8_t *out,
                             int w, uint32_t *coeffs) L1CODE;
 
-typedef void (* ltransform)(uint8_t *Y, uint8_t *U, uint8_t *V, uint8_t *out,
-                            int w, uint32_t *coeffs);
-
+typedef void (*ltransform)(uint8_t *Y, uint8_t *U, uint8_t *V, uint8_t *out,
+                           int w, uint32_t *coeffs);
 
 static void bfin_prepare_coefficients(SwsContext *c, int rgb, int masks)
 {
     int oy;
-    oy      = c->yOffset&0xffff;
-    oy      = oy >> 3; // keep everything U8.0 for offset calculation
+    oy = c->yOffset & 0xffff;
+    oy = oy >> 3;      // keep everything U8.0 for offset calculation
 
-    c->oc   = 128*0x01010101U;
-    c->oy   =  oy*0x01010101U;
+    c->oc = 128 * 0x01010101U;
+    c->oy = oy * 0x01010101U;
 
     /* copy 64bit vector coeffs down to 32bit vector coeffs */
-    c->cy  = c->yCoeff;
+    c->cy   = c->yCoeff;
     c->zero = 0;
 
     if (rgb) {
@@ -77,7 +77,6 @@ static void bfin_prepare_coefficients(SwsContext *c, int rgb, int masks)
         c->cgv = c->ugCoeff;
     }
 
-
     if (masks == 555) {
         c->rmask = 0x001f * 0x00010001U;
         c->gmask = 0x03e0 * 0x00010001U;
@@ -89,27 +88,25 @@ static void bfin_prepare_coefficients(SwsContext *c, int rgb, int masks)
     }
 }
 
-static int core_yuv420_rgb(SwsContext *c,
-                           uint8_t **in, int *instrides,
-                           int srcSliceY, int srcSliceH,
-                           uint8_t **oplanes, int *outstrides,
-                           ltransform lcscf, int rgb, int masks)
+static int core_yuv420_rgb(SwsContext *c, uint8_t **in, int *instrides,
+                           int srcSliceY, int srcSliceH, uint8_t **oplanes,
+                           int *outstrides, ltransform lcscf,
+                           int rgb, int masks)
 {
-    uint8_t *py,*pu,*pv,*op;
+    uint8_t *py, *pu, *pv, *op;
     int w  = instrides[0];
-    int h2 = srcSliceH>>1;
+    int h2 = srcSliceH >> 1;
     int i;
 
     bfin_prepare_coefficients(c, rgb, masks);
 
     py = in[0];
-    pu = in[1+(1^rgb)];
-    pv = in[1+(0^rgb)];
+    pu = in[1 + (1 ^ rgb)];
+    pv = in[1 + (0 ^ rgb)];
 
-    op = oplanes[0] + srcSliceY*outstrides[0];
-
-    for (i=0;i<h2;i++) {
+    op = oplanes[0] + srcSliceY * outstrides[0];
 
+    for (i = 0; i < h2; i++) {
         lcscf(py, pu, pv, op, w, &c->oy);
 
         py += instrides[0];
@@ -126,9 +123,7 @@ static int core_yuv420_rgb(SwsContext *c,
     return srcSliceH;
 }
 
-
-static int bfin_yuv420_rgb555(SwsContext *c,
-                              uint8_t **in, int *instrides,
+static int bfin_yuv420_rgb555(SwsContext *c, uint8_t **in, int *instrides,
                               int srcSliceY, int srcSliceH,
                               uint8_t **oplanes, int *outstrides)
 {
@@ -136,8 +131,7 @@ static int bfin_yuv420_rgb555(SwsContext *c,
                            outstrides, ff_bfin_yuv2rgb555_line, 1, 555);
 }
 
-static int bfin_yuv420_bgr555(SwsContext *c,
-                              uint8_t **in, int *instrides,
+static int bfin_yuv420_bgr555(SwsContext *c, uint8_t **in, int *instrides,
                               int srcSliceY, int srcSliceH,
                               uint8_t **oplanes, int *outstrides)
 {
@@ -145,8 +139,7 @@ static int bfin_yuv420_bgr555(SwsContext *c,
                            outstrides, ff_bfin_yuv2rgb555_line, 0, 555);
 }
 
-static int bfin_yuv420_rgb24(SwsContext *c,
-                             uint8_t **in, int *instrides,
+static int bfin_yuv420_rgb24(SwsContext *c, uint8_t **in, int *instrides,
                              int srcSliceY, int srcSliceH,
                              uint8_t **oplanes, int *outstrides)
 {
@@ -154,8 +147,7 @@ static int bfin_yuv420_rgb24(SwsContext *c,
                            outstrides, ff_bfin_yuv2rgb24_line, 1, 888);
 }
 
-static int bfin_yuv420_bgr24(SwsContext *c,
-                             uint8_t **in, int *instrides,
+static int bfin_yuv420_bgr24(SwsContext *c, uint8_t **in, int *instrides,
                              int srcSliceY, int srcSliceH,
                              uint8_t **oplanes, int *outstrides)
 {
@@ -163,8 +155,7 @@ static int bfin_yuv420_bgr24(SwsContext *c,
                            outstrides, ff_bfin_yuv2rgb24_line, 0, 888);
 }
 
-static int bfin_yuv420_rgb565(SwsContext *c,
-                              uint8_t **in, int *instrides,
+static int bfin_yuv420_rgb565(SwsContext *c, uint8_t **in, int *instrides,
                               int srcSliceY, int srcSliceH,
                               uint8_t **oplanes, int *outstrides)
 {
@@ -172,8 +163,7 @@ static int bfin_yuv420_rgb565(SwsContext *c,
                            outstrides, ff_bfin_yuv2rgb565_line, 1, 565);
 }
 
-static int bfin_yuv420_bgr565(SwsContext *c,
-                              uint8_t **in, int *instrides,
+static int bfin_yuv420_bgr565(SwsContext *c, uint8_t **in, int *instrides,
                               int srcSliceY, int srcSliceH,
                               uint8_t **oplanes, int *outstrides)
 {
@@ -181,18 +171,29 @@ static int bfin_yuv420_bgr565(SwsContext *c,
                            outstrides, ff_bfin_yuv2rgb565_line, 0, 565);
 }
 
-
 SwsFunc ff_yuv2rgb_get_func_ptr_bfin(SwsContext *c)
 {
     SwsFunc f;
 
-    switch(c->dstFormat) {
-    case PIX_FMT_RGB555: f = bfin_yuv420_rgb555; break;
-    case PIX_FMT_BGR555: f = bfin_yuv420_bgr555; break;
-    case PIX_FMT_RGB565: f = bfin_yuv420_rgb565; break;
-    case PIX_FMT_BGR565: f = bfin_yuv420_bgr565; break;
-    case PIX_FMT_RGB24:  f = bfin_yuv420_rgb24;  break;
-    case PIX_FMT_BGR24:  f = bfin_yuv420_bgr24;  break;
+    switch (c->dstFormat) {
+    case PIX_FMT_RGB555:
+        f = bfin_yuv420_rgb555;
+        break;
+    case PIX_FMT_BGR555:
+        f = bfin_yuv420_bgr555;
+        break;
+    case PIX_FMT_RGB565:
+        f = bfin_yuv420_rgb565;
+        break;
+    case PIX_FMT_BGR565:
+        f = bfin_yuv420_bgr565;
+        break;
+    case PIX_FMT_RGB24:
+        f = bfin_yuv420_rgb24;
+        break;
+    case PIX_FMT_BGR24:
+        f = bfin_yuv420_bgr24;
+        break;
     default:
         return 0;
     }
diff --git a/libswscale/x86/input.asm b/libswscale/x86/input.asm
index 9a8e24f0b0..50e071a89a 100644
--- a/libswscale/x86/input.asm
+++ b/libswscale/x86/input.asm
@@ -51,6 +51,19 @@ bgr_Vcoeff_3x56: times 2 dw RV, 0, GV, RV
 rgb_Vcoeff_12x4: times 2 dw RV, GV, 0, RV
 rgb_Vcoeff_3x56: times 2 dw BV, 0, GV, BV
 
+rgba_Ycoeff_rb:  times 4 dw RY, BY
+rgba_Ycoeff_br:  times 4 dw BY, RY
+rgba_Ycoeff_ga:  times 4 dw GY, 0
+rgba_Ycoeff_ag:  times 4 dw 0,  GY
+rgba_Ucoeff_rb:  times 4 dw RU, BU
+rgba_Ucoeff_br:  times 4 dw BU, RU
+rgba_Ucoeff_ga:  times 4 dw GU, 0
+rgba_Ucoeff_ag:  times 4 dw 0,  GU
+rgba_Vcoeff_rb:  times 4 dw RV, BV
+rgba_Vcoeff_br:  times 4 dw BV, RV
+rgba_Vcoeff_ga:  times 4 dw GV, 0
+rgba_Vcoeff_ag:  times 4 dw 0,  GV
+
 shuf_rgb_12x4:   db 0, 0x80, 1, 0x80,  2, 0x80,  3, 0x80, \
                     6, 0x80, 7, 0x80,  8, 0x80,  9, 0x80
 shuf_rgb_3x56:   db 2, 0x80, 3, 0x80,  4, 0x80,  5, 0x80, \
@@ -294,6 +307,150 @@ RGB24_FUNCS 11, 13
 INIT_XMM avx
 RGB24_FUNCS 11, 13
 
+; %1 = nr. of XMM registers
+; %2-5 = rgba, bgra, argb or abgr (in individual characters)
+%macro RGB32_TO_Y_FN 5-6
+cglobal %2%3%4%5 %+ ToY, 6, 6, %1, dst, src, u1, u2, w, u3
+    mova           m5, [rgba_Ycoeff_%2%4]
+    mova           m6, [rgba_Ycoeff_%3%5]
+%if %0 == 6
+    jmp mangle(program_name %+ _ %+ %6 %+ ToY %+ SUFFIX).body
+%else ; %0 == 6
+.body:
+%if ARCH_X86_64
+    movsxd         wq, wd
+%endif
+    lea          srcq, [srcq+wq*4]
+    add            wq, wq
+    add          dstq, wq
+    neg            wq
+    mova           m4, [rgb_Yrnd]
+    pcmpeqb        m7, m7
+    psrlw          m7, 8                  ; (word) { 0x00ff } x4
+.loop:
+    ; FIXME check alignment and use mova
+    movu           m0, [srcq+wq*2+0]      ; (byte) { Bx, Gx, Rx, xx }[0-3]
+    movu           m2, [srcq+wq*2+mmsize] ; (byte) { Bx, Gx, Rx, xx }[4-7]
+    DEINTB          1,  0,  3,  2,  7     ; (word) { Gx, xx (m0/m2) or Bx, Rx (m1/m3) }[0-3]/[4-7]
+    pmaddwd        m1, m5                 ; (dword) { Bx*BY + Rx*RY }[0-3]
+    pmaddwd        m0, m6                 ; (dword) { Gx*GY }[0-3]
+    pmaddwd        m3, m5                 ; (dword) { Bx*BY + Rx*RY }[4-7]
+    pmaddwd        m2, m6                 ; (dword) { Gx*GY }[4-7]
+    paddd          m0, m4                 ; += rgb_Yrnd
+    paddd          m2, m4                 ; += rgb_Yrnd
+    paddd          m0, m1                 ; (dword) { Y[0-3] }
+    paddd          m2, m3                 ; (dword) { Y[4-7] }
+    psrad          m0, 9
+    psrad          m2, 9
+    packssdw       m0, m2                 ; (word) { Y[0-7] }
+    mova    [dstq+wq], m0
+    add            wq, mmsize
+    jl .loop
+    REP_RET
+%endif ; %0 == 3
+%endmacro
+
+; %1 = nr. of XMM registers
+; %2-5 = rgba, bgra, argb or abgr (in individual characters)
+%macro RGB32_TO_UV_FN 5-6
+cglobal %2%3%4%5 %+ ToUV, 7, 7, %1, dstU, dstV, u1, src, u2, w, u3
+%if ARCH_X86_64
+    mova           m8, [rgba_Ucoeff_%2%4]
+    mova           m9, [rgba_Ucoeff_%3%5]
+    mova          m10, [rgba_Vcoeff_%2%4]
+    mova          m11, [rgba_Vcoeff_%3%5]
+%define coeffU1 m8
+%define coeffU2 m9
+%define coeffV1 m10
+%define coeffV2 m11
+%else ; x86-32
+%define coeffU1 [rgba_Ucoeff_%2%4]
+%define coeffU2 [rgba_Ucoeff_%3%5]
+%define coeffV1 [rgba_Vcoeff_%2%4]
+%define coeffV2 [rgba_Vcoeff_%3%5]
+%endif ; x86-64/32
+%if ARCH_X86_64 && %0 == 6
+    jmp mangle(program_name %+ _ %+ %6 %+ ToUV %+ SUFFIX).body
+%else ; ARCH_X86_64 && %0 == 6
+.body:
+%if ARCH_X86_64
+    movsxd         wq, dword r5m
+%else ; x86-32
+    mov            wq, r5m
+%endif
+    add            wq, wq
+    add         dstUq, wq
+    add         dstVq, wq
+    lea          srcq, [srcq+wq*2]
+    neg            wq
+    pcmpeqb        m7, m7
+    psrlw          m7, 8                  ; (word) { 0x00ff } x4
+    mova           m6, [rgb_UVrnd]
+.loop:
+    ; FIXME check alignment and use mova
+    movu           m0, [srcq+wq*2+0]      ; (byte) { Bx, Gx, Rx, xx }[0-3]
+    movu           m4, [srcq+wq*2+mmsize] ; (byte) { Bx, Gx, Rx, xx }[4-7]
+    DEINTB          1,  0,  5,  4,  7     ; (word) { Gx, xx (m0/m4) or Bx, Rx (m1/m5) }[0-3]/[4-7]
+    pmaddwd        m3, m1, coeffV1        ; (dword) { Bx*BV + Rx*RV }[0-3]
+    pmaddwd        m2, m0, coeffV2        ; (dword) { Gx*GV }[0-3]
+    pmaddwd        m1, coeffU1            ; (dword) { Bx*BU + Rx*RU }[0-3]
+    pmaddwd        m0, coeffU2            ; (dword) { Gx*GU }[0-3]
+    paddd          m3, m6                 ; += rgb_UVrnd
+    paddd          m1, m6                 ; += rgb_UVrnd
+    paddd          m2, m3                 ; (dword) { V[0-3] }
+    paddd          m0, m1                 ; (dword) { U[0-3] }
+    pmaddwd        m3, m5, coeffV1        ; (dword) { Bx*BV + Rx*RV }[4-7]
+    pmaddwd        m1, m4, coeffV2        ; (dword) { Gx*GV }[4-7]
+    pmaddwd        m5, coeffU1            ; (dword) { Bx*BU + Rx*RU }[4-7]
+    pmaddwd        m4, coeffU2            ; (dword) { Gx*GU }[4-7]
+    paddd          m3, m6                 ; += rgb_UVrnd
+    paddd          m5, m6                 ; += rgb_UVrnd
+    psrad          m0, 9
+    paddd          m1, m3                 ; (dword) { V[4-7] }
+    paddd          m4, m5                 ; (dword) { U[4-7] }
+    psrad          m2, 9
+    psrad          m4, 9
+    psrad          m1, 9
+    packssdw       m0, m4                 ; (word) { U[0-7] }
+    packssdw       m2, m1                 ; (word) { V[0-7] }
+%if mmsize == 8
+    mova   [dstUq+wq], m0
+    mova   [dstVq+wq], m2
+%else ; mmsize == 16
+    mova   [dstUq+wq], m0
+    mova   [dstVq+wq], m2
+%endif ; mmsize == 8/16
+    add            wq, mmsize
+    jl .loop
+    REP_RET
+%endif ; ARCH_X86_64 && %0 == 3
+%endmacro
+
+; %1 = nr. of XMM registers for rgb-to-Y func
+; %2 = nr. of XMM registers for rgb-to-UV func
+%macro RGB32_FUNCS 2
+RGB32_TO_Y_FN %1, r, g, b, a
+RGB32_TO_Y_FN %1, b, g, r, a, rgba
+RGB32_TO_Y_FN %1, a, r, g, b, rgba
+RGB32_TO_Y_FN %1, a, b, g, r, rgba
+
+RGB32_TO_UV_FN %2, r, g, b, a
+RGB32_TO_UV_FN %2, b, g, r, a, rgba
+RGB32_TO_UV_FN %2, a, r, g, b, rgba
+RGB32_TO_UV_FN %2, a, b, g, r, rgba
+%endmacro
+
+%if ARCH_X86_32
+INIT_MMX mmx
+RGB32_FUNCS 0, 0
+%endif
+
+INIT_XMM sse2
+RGB32_FUNCS 8, 12
+
+INIT_XMM avx
+RGB32_FUNCS 8, 12
+
 ;-----------------------------------------------------------------------------
 ; YUYV/UYVY/NV12/NV21 packed pixel shuffling.
 ;
diff --git a/libswscale/x86/swscale_mmx.c b/libswscale/x86/swscale_mmx.c
index ab5b68fb0b..1118515164 100644
--- a/libswscale/x86/swscale_mmx.c
+++ b/libswscale/x86/swscale_mmx.c
@@ -308,6 +308,10 @@ extern void ff_ ## fmt ## ToUV_ ## opt(uint8_t *dstU, uint8_t *dstV, \
     INPUT_FUNC(yuyv, opt); \
     INPUT_UV_FUNC(nv12, opt); \
     INPUT_UV_FUNC(nv21, opt); \
+    INPUT_FUNC(rgba, opt); \
+    INPUT_FUNC(bgra, opt); \
+    INPUT_FUNC(argb, opt); \
+    INPUT_FUNC(abgr, opt); \
     INPUT_FUNC(rgb24, opt); \
     INPUT_FUNC(bgr24, opt)
 
@@ -406,6 +410,10 @@ switch(c->dstBpc){ \
             break;
         case_rgb(rgb24, RGB24, mmx);
         case_rgb(bgr24, BGR24, mmx);
+        case_rgb(bgra,  BGRA,  mmx);
+        case_rgb(rgba,  RGBA,  mmx);
+        case_rgb(abgr,  ABGR,  mmx);
+        case_rgb(argb,  ARGB,  mmx);
         default:
             break;
         }
@@ -450,6 +458,10 @@ switch(c->dstBpc){ \
             break;
         case_rgb(rgb24, RGB24, sse2);
         case_rgb(bgr24, BGR24, sse2);
+        case_rgb(bgra,  BGRA,  sse2);
+        case_rgb(rgba,  RGBA,  sse2);
+        case_rgb(abgr,  ABGR,  sse2);
+        case_rgb(argb,  ARGB,  sse2);
         default:
             break;
         }
@@ -493,6 +505,10 @@ switch(c->dstBpc){ \
             break;
         case_rgb(rgb24, RGB24, avx);
         case_rgb(bgr24, BGR24, avx);
+        case_rgb(bgra,  BGRA,  avx);
+        case_rgb(rgba,  RGBA,  avx);
+        case_rgb(abgr,  ABGR,  avx);
+        case_rgb(argb,  ARGB,  avx);
         default:
             break;
         }
diff --git a/libswscale/x86/w64xmmtest.c b/libswscale/x86/w64xmmtest.c
new file mode 100644
index 0000000000..dd9a2a4378
--- /dev/null
+++ b/libswscale/x86/w64xmmtest.c
@@ -0,0 +1,31 @@
+/*
+ * check XMM registers for clobbers on Win64
+ * Copyright (c) 2012 Ronald S. Bultje <rsbultje@gmail.com>
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/x86/w64xmmtest.h"
+#include "libswscale/swscale.h"
+
+wrap(sws_scale(struct SwsContext *c, const uint8_t *const srcSlice[],
+               const int srcStride[], int srcSliceY, int srcSliceH,
+               uint8_t *const dst[], const int dstStride[]))
+{
+    testxmmclobbers(sws_scale, c, srcSlice, srcStride, srcSliceY,
+                    srcSliceH, dst, dstStride);
+}
diff --git a/tests/codec-regression.sh b/tests/codec-regression.sh
index 5da37b9b9a..b8d27cb7b0 100755
--- a/tests/codec-regression.sh
+++ b/tests/codec-regression.sh
@@ -368,7 +368,7 @@ $tiny_psnr $pcm_dst $pcm_ref 2 1924
 fi
 
 if [ -n "$do_ac3_fixed" ] ; then
-do_audio_encoding ac3.rm "-vn -acodec ac3_fixed"
+do_audio_encoding ac3.ac3 "-vn -acodec ac3_fixed"
 # binaries configured with --disable-sse decode ac3 differently
 #do_audio_decoding
 #$tiny_psnr $pcm_dst $pcm_ref 2 1024
diff --git a/tests/ref/acodec/ac3_fixed b/tests/ref/acodec/ac3_fixed
index dba2dfc5e7..0c2f9b7214 100644
--- a/tests/ref/acodec/ac3_fixed
+++ b/tests/ref/acodec/ac3_fixed
@@ -1,2 +1,2 @@
-e7fa185030a56d9db8663ad9e38c6c94 *./tests/data/acodec/ac3.rm
-98751 ./tests/data/acodec/ac3.rm
+a1d1fc116463b771abf5aef7ed37d7b1 *./tests/data/acodec/ac3.ac3
+96408 ./tests/data/acodec/ac3.ac3
diff --git a/tests/ref/fate/vc1-ism b/tests/ref/fate/vc1-ism
index e2fda62d18..0e8520605f 100644
--- a/tests/ref/fate/vc1-ism
+++ b/tests/ref/fate/vc1-ism
@@ -117,4 +117,3 @@
 0, 438750, 37440, 0xf0fe8c1c
 0, 442500, 37440, 0xc0036222
 0, 446250, 37440, 0x3058385c
-0, 450000, 37440, 0x68141016
diff --git a/tests/ref/seek/ac3_ac3 b/tests/ref/seek/ac3_ac3
new file mode 100644
index 0000000000..167dc8d716
--- /dev/null
+++ b/tests/ref/seek/ac3_ac3
@@ -0,0 +1,49 @@
+ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:      0 size:   556
+ret: 0         st:-1 flags:0  ts:-1.000000
+ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:      0 size:   556
+ret: 0         st:-1 flags:1  ts: 1.894167
+ret: 0         st: 0 flags:1 dts: 1.880400 pts: 1.880400 pos:  30092 size:   558
+ret: 0         st: 0 flags:0  ts: 0.788333
+ret: 0         st: 0 flags:1 dts: 0.800911 pts: 0.800911 pos:  12818 size:   556
+ret:-1         st: 0 flags:1  ts:-0.317500
+ret: 0         st:-1 flags:0  ts: 2.576668
+ret: 0         st: 0 flags:1 dts: 2.576844 pts: 2.576844 pos:  41238 size:   558
+ret: 0         st:-1 flags:1  ts: 1.470835
+ret: 0         st: 0 flags:1 dts: 1.462533 pts: 1.462533 pos:  23406 size:   556
+ret: 0         st: 0 flags:0  ts: 0.365000
+ret: 0         st: 0 flags:1 dts: 0.383044 pts: 0.383044 pos:   6130 size:   558
+ret:-1         st: 0 flags:1  ts:-0.740833
+ret: 0         st:-1 flags:0  ts: 2.153336
+ret: 0         st: 0 flags:1 dts: 2.158978 pts: 2.158978 pos:  34552 size:   556
+ret: 0         st:-1 flags:1  ts: 1.047503
+ret: 0         st: 0 flags:1 dts: 1.044667 pts: 1.044667 pos:  16718 size:   558
+ret: 0         st: 0 flags:0  ts:-0.058333
+ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:      0 size:   556
+ret: 0         st: 0 flags:1  ts: 2.835833
+ret: 0         st: 0 flags:1 dts: 2.820600 pts: 2.820600 pos:  45140 size:   556
+ret: 0         st:-1 flags:0  ts: 1.730004
+ret: 0         st: 0 flags:1 dts: 1.741111 pts: 1.741111 pos:  27864 size:   556
+ret: 0         st:-1 flags:1  ts: 0.624171
+ret: 0         st: 0 flags:1 dts: 0.591978 pts: 0.591978 pos:   9474 size:   556
+ret: 0         st: 0 flags:0  ts:-0.481667
+ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:      0 size:   556
+ret: 0         st: 0 flags:1  ts: 2.412500
+ret: 0         st: 0 flags:1 dts: 2.402733 pts: 2.402733 pos:  38452 size:   558
+ret: 0         st:-1 flags:0  ts: 1.306672
+ret: 0         st: 0 flags:1 dts: 1.323244 pts: 1.323244 pos:  21176 size:   558
+ret: 0         st:-1 flags:1  ts: 0.200839
+ret: 0         st: 0 flags:1 dts: 0.174111 pts: 0.174111 pos:   2786 size:   558
+ret: 0         st: 0 flags:0  ts:-0.904989
+ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:      0 size:   556
+ret: 0         st: 0 flags:1  ts: 1.989178
+ret: 0         st: 0 flags:1 dts: 1.984867 pts: 1.984867 pos:  31764 size:   558
+ret: 0         st:-1 flags:0  ts: 0.883340
+ret: 0         st: 0 flags:1 dts: 0.905378 pts: 0.905378 pos:  14488 size:   558
+ret:-1         st:-1 flags:1  ts:-0.222493
+ret: 0         st: 0 flags:0  ts: 2.671678
+ret: 0         st: 0 flags:1 dts: 2.681311 pts: 2.681311 pos:  42910 size:   558
+ret: 0         st: 0 flags:1  ts: 1.565844
+ret: 0         st: 0 flags:1 dts: 1.532178 pts: 1.532178 pos:  24520 size:   558
+ret: 0         st:-1 flags:0  ts: 0.460008
+ret: 0         st: 0 flags:1 dts: 0.487511 pts: 0.487511 pos:   7802 size:   556
+ret:-1         st:-1 flags:1  ts:-0.645825
diff --git a/tests/ref/seek/ac3_rm b/tests/ref/seek/ac3_rm
deleted file mode 100644
index ecf03c29a1..0000000000
--- a/tests/ref/seek/ac3_rm
+++ /dev/null
@@ -1,41 +0,0 @@
-ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:    271 size:   556
-ret: 0         st:-1 flags:0  ts:-1.000000
-ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:    271 size:   556
-ret:-1         st:-1 flags:1  ts: 1.894167
-ret:-1         st: 0 flags:0  ts: 0.788000
-ret: 0         st: 0 flags:1  ts:-0.317000
-ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:    271 size:   556
-ret: 0         st:-1 flags:0  ts: 2.576668
-ret: 0         st: 0 flags:1 dts: 2.124000 pts: 2.124000 pos:  34997 size:   558
-ret:-1         st:-1 flags:1  ts: 1.470835
-ret:-1         st: 0 flags:0  ts: 0.365000
-ret: 0         st: 0 flags:1  ts:-0.741000
-ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:    271 size:   556
-ret: 0         st:-1 flags:0  ts: 2.153336
-ret: 0         st: 0 flags:1 dts: 2.124000 pts: 2.124000 pos:  34997 size:   558
-ret:-1         st:-1 flags:1  ts: 1.047503
-ret: 0         st: 0 flags:0  ts:-0.058000
-ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:    271 size:   556
-ret: 0         st: 0 flags:1  ts: 2.836000
-ret: 0         st: 0 flags:1 dts: 2.124000 pts: 2.124000 pos:  34997 size:   558
-ret:-1         st:-1 flags:0  ts: 1.730004
-ret:-1         st:-1 flags:1  ts: 0.624171
-ret: 0         st: 0 flags:0  ts:-0.482000
-ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:    271 size:   556
-ret: 0         st: 0 flags:1  ts: 2.413000
-ret: 0         st: 0 flags:1 dts: 2.124000 pts: 2.124000 pos:  34997 size:   558
-ret: 0         st:-1 flags:0  ts: 1.306672
-ret: 0         st: 0 flags:1 dts:65.537000 pts:65.537000 pos:  87488 size:  6132
-ret:-1         st:-1 flags:1  ts: 0.200839
-ret: 0         st: 0 flags:0  ts:-0.905000
-ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:    271 size:   556
-ret:-1         st: 0 flags:1  ts: 1.989000
-ret:-1         st:-1 flags:0  ts: 0.883340
-ret: 0         st:-1 flags:1  ts:-0.222493
-ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:    271 size:   556
-ret:-1         st: 0 flags:0  ts: 2.672000
-ret:-1         st: 0 flags:1  ts: 1.566000
-ret: 0         st:-1 flags:0  ts: 0.460008
-ret: 0         st: 0 flags:1 dts: 1.567000 pts: 1.567000 pos:  25889 size:   556
-ret: 0         st:-1 flags:1  ts:-0.645825
-ret: 0         st: 0 flags:1 dts: 0.000000 pts: 0.000000 pos:    271 size:   556