Merge remote-tracking branch 'qatar/master'

* qatar/master: (23 commits)
  fix AC3ENC_OPT_MODE_ON/OFF
  h264: fix HRD parameters parsing
  prores: implement multithreading.
  prores: idct sse2/sse4 optimizations.
  swscale: use aligned move for storage into temporary buffer.
  prores: extract idct into its own dspcontext and merge with put_pixels.
  h264: fix invalid shifts in init_cavlc_level_tab()
  intfloat_readwrite: fix signed addition overflows
  mov: do not misreport empty stts
  mov: cosmetics, fix for and if spacing
  id3v2: fix NULL pointer dereference
  mov: read album_artist atom
  mov: fix disc/track numbers and totals
  doc: fix references to obsolete presets directories for avconv/ffmpeg
  flashsv: return more meaningful error value
  flashsv: fix typo in av_log() message
  smacker: validate channels and sample format.
  smacker: check buffer size before reading output size
  smacker: validate number of channels
  smacker: Separate audio flags from sample rates in smacker demuxer.
  ...

Conflicts:
	cmdutils.h
	doc/ffmpeg.texi
	libavcodec/Makefile
	libavcodec/motion_est_template.c
	libavformat/id3v2.c
	libavformat/mov.c

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

22 files changed, 788 insertions, 156 deletions

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 2978329801..78a0a300b3 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -314,7 +314,7 @@ OBJS-$(CONFIG_PNG_ENCODER)             += png.o pngenc.o
 OBJS-$(CONFIG_PPM_DECODER)             += pnmdec.o pnm.o
 OBJS-$(CONFIG_PPM_ENCODER)             += pnmenc.o pnm.o
 OBJS-$(CONFIG_PRORES_GPL_DECODER)      += proresdec_gpl.o
-OBJS-$(CONFIG_PRORES_LGPL_DECODER)     += proresdec_lgpl.o
+OBJS-$(CONFIG_PRORES_LGPL_DECODER)     += proresdec_lgpl.o proresdsp.o
 OBJS-$(CONFIG_PTX_DECODER)             += ptx.o
 OBJS-$(CONFIG_QCELP_DECODER)           += qcelpdec.o celp_math.o         \
                                           celp_filters.o acelp_vectors.o \
diff --git a/libavcodec/ac3enc.h b/libavcodec/ac3enc.h
index bf5ccea19f..4a017498f6 100644
--- a/libavcodec/ac3enc.h
+++ b/libavcodec/ac3enc.h
@@ -73,8 +73,8 @@ typedef int64_t CoefSumType;
 #define AC3ENC_OPT_OFF              0
 #define AC3ENC_OPT_ON               1
 #define AC3ENC_OPT_NOT_INDICATED    0
-#define AC3ENC_OPT_MODE_ON          1
-#define AC3ENC_OPT_MODE_OFF         2
+#define AC3ENC_OPT_MODE_ON          2
+#define AC3ENC_OPT_MODE_OFF         1
 
 /* specific option values */
 #define AC3ENC_OPT_LARGE_ROOM       1
diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index ebce93039f..c451c97155 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -144,6 +144,41 @@ void ff_init_scantable(uint8_t *permutation, ScanTable *st, const uint8_t *src_s
     }
 }
 
+void ff_init_scantable_permutation(uint8_t *idct_permutation,
+                                   int idct_permutation_type)
+{
+    int i;
+
+    switch(idct_permutation_type){
+    case FF_NO_IDCT_PERM:
+        for(i=0; i<64; i++)
+            idct_permutation[i]= i;
+        break;
+    case FF_LIBMPEG2_IDCT_PERM:
+        for(i=0; i<64; i++)
+            idct_permutation[i]= (i & 0x38) | ((i & 6) >> 1) | ((i & 1) << 2);
+        break;
+    case FF_SIMPLE_IDCT_PERM:
+        for(i=0; i<64; i++)
+            idct_permutation[i]= simple_mmx_permutation[i];
+        break;
+    case FF_TRANSPOSE_IDCT_PERM:
+        for(i=0; i<64; i++)
+            idct_permutation[i]= ((i&7)<<3) | (i>>3);
+        break;
+    case FF_PARTTRANS_IDCT_PERM:
+        for(i=0; i<64; i++)
+            idct_permutation[i]= (i&0x24) | ((i&3)<<3) | ((i>>3)&3);
+        break;
+    case FF_SSE2_IDCT_PERM:
+        for(i=0; i<64; i++)
+            idct_permutation[i]= (i&0x38) | idct_sse2_row_perm[i&7];
+        break;
+    default:
+        av_log(NULL, AV_LOG_ERROR, "Internal error, IDCT permutation not set\n");
+    }
+}
+
 static int pix_sum_c(uint8_t * pix, int line_size)
 {
     int s, i, j;
@@ -3107,32 +3142,6 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx)
             c->avg_2tap_qpel_pixels_tab[0][i]= c->avg_h264_qpel_pixels_tab[0][i];
     }
 
-    switch(c->idct_permutation_type){
-    case FF_NO_IDCT_PERM:
-        for(i=0; i<64; i++)
-            c->idct_permutation[i]= i;
-        break;
-    case FF_LIBMPEG2_IDCT_PERM:
-        for(i=0; i<64; i++)
-            c->idct_permutation[i]= (i & 0x38) | ((i & 6) >> 1) | ((i & 1) << 2);
-        break;
-    case FF_SIMPLE_IDCT_PERM:
-        for(i=0; i<64; i++)
-            c->idct_permutation[i]= simple_mmx_permutation[i];
-        break;
-    case FF_TRANSPOSE_IDCT_PERM:
-        for(i=0; i<64; i++)
-            c->idct_permutation[i]= ((i&7)<<3) | (i>>3);
-        break;
-    case FF_PARTTRANS_IDCT_PERM:
-        for(i=0; i<64; i++)
-            c->idct_permutation[i]= (i&0x24) | ((i&3)<<3) | ((i>>3)&3);
-        break;
-    case FF_SSE2_IDCT_PERM:
-        for(i=0; i<64; i++)
-            c->idct_permutation[i]= (i&0x38) | idct_sse2_row_perm[i&7];
-        break;
-    default:
-        av_log(avctx, AV_LOG_ERROR, "Internal error, IDCT permutation not set\n");
-    }
+    ff_init_scantable_permutation(c->idct_permutation,
+                                  c->idct_permutation_type);
 }
diff --git a/libavcodec/dsputil.h b/libavcodec/dsputil.h
index 057c41cca5..22c51a0962 100644
--- a/libavcodec/dsputil.h
+++ b/libavcodec/dsputil.h
@@ -204,6 +204,8 @@ typedef struct ScanTable{
 } ScanTable;
 
 void ff_init_scantable(uint8_t *, ScanTable *st, const uint8_t *src_scantable);
+void ff_init_scantable_permutation(uint8_t *idct_permutation,
+                                   int idct_permutation_type);
 
 #define EMULATED_EDGE(depth) \
 void ff_emulated_edge_mc_ ## depth (uint8_t *buf, const uint8_t *src, int linesize,\
diff --git a/libavcodec/flashsv.c b/libavcodec/flashsv.c
index 6b649bb874..3861344cb7 100644
--- a/libavcodec/flashsv.c
+++ b/libavcodec/flashsv.c
@@ -301,7 +301,7 @@ static int flashsv_decode_frame(AVCodecContext *avctx, void *data,
     /* check for changes of image width and image height */
     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");
+               "Frame width or height differs from first frame!\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);
         return AVERROR_INVALIDDATA;
@@ -367,7 +367,7 @@ static int flashsv_decode_frame(AVCodecContext *avctx, void *data,
                 if (s->color_depth != 0 && s->color_depth != 2) {
                     av_log(avctx, AV_LOG_ERROR,
                            "%dx%d invalid color depth %d\n", i, j, s->color_depth);
-                    return -1;
+                    return AVERROR_INVALIDDATA;
                 }
 
                 if (has_diff) {
diff --git a/libavcodec/golomb.h b/libavcodec/golomb.h
index 90eeb30b54..8dff0322a7 100644
--- a/libavcodec/golomb.h
+++ b/libavcodec/golomb.h
@@ -75,6 +75,20 @@ static inline int get_ue_golomb(GetBitContext *gb){
     }
 }
 
+/**
+ * Read an unsigned Exp-Golomb code in the range 0 to UINT32_MAX-1.
+ */
+static inline unsigned get_ue_golomb_long(GetBitContext *gb)
+{
+    unsigned buf, log;
+
+    buf = show_bits_long(gb, 32);
+    log = 31 - av_log2(buf);
+    skip_bits_long(gb, log);
+
+    return get_bits_long(gb, log + 1) - 1;
+}
+
  /**
  * read unsigned exp golomb code, constraint to a max of 31.
  * the return value is undefined if the stored value exceeds 31.
diff --git a/libavcodec/h264_cavlc.c b/libavcodec/h264_cavlc.c
index d2b979fc2c..b0dc999132 100644
--- a/libavcodec/h264_cavlc.c
+++ b/libavcodec/h264_cavlc.c
@@ -298,17 +298,18 @@ static inline int pred_non_zero_count(H264Context *h, int n){
 }
 
 static av_cold void init_cavlc_level_tab(void){
-    int suffix_length, mask;
+    int suffix_length;
     unsigned int i;
 
     for(suffix_length=0; suffix_length<7; suffix_length++){
         for(i=0; i<(1<<LEVEL_TAB_BITS); i++){
             int prefix= LEVEL_TAB_BITS - av_log2(2*i);
-            int level_code= (prefix<<suffix_length) + (i>>(LEVEL_TAB_BITS-prefix-1-suffix_length)) - (1<<suffix_length);
 
-            mask= -(level_code&1);
-            level_code= (((2+level_code)>>1) ^ mask) - mask;
             if(prefix + 1 + suffix_length <= LEVEL_TAB_BITS){
+                int level_code = (prefix << suffix_length) +
+                    (i >> (av_log2(i) - suffix_length)) - (1 << suffix_length);
+                int mask = -(level_code&1);
+                level_code = (((2 + level_code) >> 1) ^ mask) - mask;
                 cavlc_level_tab[suffix_length][i][0]= level_code;
                 cavlc_level_tab[suffix_length][i][1]= prefix + 1 + suffix_length;
             }else if(prefix + 1 <= LEVEL_TAB_BITS){
diff --git a/libavcodec/h264_ps.c b/libavcodec/h264_ps.c
index f0ec0ff9a9..158ca3205a 100644
--- a/libavcodec/h264_ps.c
+++ b/libavcodec/h264_ps.c
@@ -143,8 +143,8 @@ static inline int decode_hrd_parameters(H264Context *h, SPS *sps){
     get_bits(&s->gb, 4); /* bit_rate_scale */
     get_bits(&s->gb, 4); /* cpb_size_scale */
     for(i=0; i<cpb_count; i++){
-        get_ue_golomb(&s->gb); /* bit_rate_value_minus1 */
-        get_ue_golomb(&s->gb); /* cpb_size_value_minus1 */
+        get_ue_golomb_long(&s->gb); /* bit_rate_value_minus1 */
+        get_ue_golomb_long(&s->gb); /* cpb_size_value_minus1 */
         get_bits1(&s->gb);     /* cbr_flag */
     }
     sps->initial_cpb_removal_delay_length = get_bits(&s->gb, 5) + 1;
@@ -494,6 +494,7 @@ int ff_h264_decode_picture_parameter_set(H264Context *h, int bit_length){
     unsigned int pps_id= get_ue_golomb(&s->gb);
     PPS *pps;
     const int qp_bd_offset = 6*(h->sps.bit_depth_luma-8);
+    int bits_left;
 
     if(pps_id >= MAX_PPS_COUNT) {
         av_log(h->s.avctx, AV_LOG_ERROR, "pps_id (%d) out of range\n", pps_id);
@@ -570,6 +571,7 @@ int ff_h264_decode_picture_parameter_set(H264Context *h, int bit_length){
     memcpy(pps->scaling_matrix4, h->sps_buffers[pps->sps_id]->scaling_matrix4, sizeof(pps->scaling_matrix4));
     memcpy(pps->scaling_matrix8, h->sps_buffers[pps->sps_id]->scaling_matrix8, sizeof(pps->scaling_matrix8));
 
+    bits_left = bit_length - get_bits_count(&s->gb);
     if(get_bits_count(&s->gb) < bit_length){
         pps->transform_8x8_mode= get_bits1(&s->gb);
         decode_scaling_matrices(h, h->sps_buffers[pps->sps_id], pps, 0, pps->scaling_matrix4, pps->scaling_matrix8);
diff --git a/libavcodec/motion_est.c b/libavcodec/motion_est.c
index c8af093653..319aaf97ca 100644
--- a/libavcodec/motion_est.c
+++ b/libavcodec/motion_est.c
@@ -52,7 +52,7 @@ static inline int sad_hpel_motion_search(MpegEncContext * s,
                                   int src_index, int ref_index,
                                   int size, int h);
 
-static inline int update_map_generation(MotionEstContext *c)
+static inline unsigned update_map_generation(MotionEstContext *c)
 {
     c->map_generation+= 1<<(ME_MAP_MV_BITS*2);
     if(c->map_generation==0){
diff --git a/libavcodec/motion_est_template.c b/libavcodec/motion_est_template.c
index 72150b4092..b7b7b6d521 100644
--- a/libavcodec/motion_est_template.c
+++ b/libavcodec/motion_est_template.c
@@ -90,8 +90,8 @@ static int hpel_motion_search(MpegEncContext * s,
                      + (mv_penalty[bx   - pred_x] + mv_penalty[by+2 - pred_y])*c->penalty_factor;
 
 #if 1
-        int key;
-        int map_generation= c->map_generation;
+        unsigned key;
+        unsigned map_generation= c->map_generation;
 #ifndef NDEBUG
         uint32_t *map= c->map;
 #endif
@@ -210,7 +210,7 @@ static int qpel_motion_search(MpegEncContext * s,
     const int mx = *mx_ptr;
     const int my = *my_ptr;
     const int penalty_factor= c->sub_penalty_factor;
-    const int map_generation= c->map_generation;
+    const unsigned map_generation = c->map_generation;
     const int subpel_quality= c->avctx->me_subpel_quality;
     uint32_t *map= c->map;
     me_cmp_func cmpf, chroma_cmpf;
@@ -356,7 +356,7 @@ static int qpel_motion_search(MpegEncContext * s,
 
 #define CHECK_MV(x,y)\
 {\
-    const int key= ((y)<<ME_MAP_MV_BITS) + (x) + map_generation;\
+    const unsigned key = ((y)<<ME_MAP_MV_BITS) + (x) + map_generation;\
     const int index= (((y)<<ME_MAP_SHIFT) + (x))&(ME_MAP_SIZE-1);\
     assert((x) >= xmin);\
     assert((x) <= xmax);\
@@ -384,7 +384,7 @@ static int qpel_motion_search(MpegEncContext * s,
 
 #define CHECK_MV_DIR(x,y,new_dir)\
 {\
-    const int key= ((y)<<ME_MAP_MV_BITS) + (x) + map_generation;\
+    const unsigned key = ((y)<<ME_MAP_MV_BITS) + (x) + map_generation;\
     const int index= (((y)<<ME_MAP_SHIFT) + (x))&(ME_MAP_SIZE-1);\
 /*printf("check_mv_dir %d %d %d\n", x, y, new_dir);*/\
     if(map[index]!=key){\
@@ -422,13 +422,13 @@ static av_always_inline int small_diamond_search(MpegEncContext * s, int *best,
     int next_dir=-1;
     LOAD_COMMON
     LOAD_COMMON2
-    int map_generation= c->map_generation;
+    unsigned map_generation = c->map_generation;
 
     cmpf= s->dsp.me_cmp[size];
     chroma_cmpf= s->dsp.me_cmp[size+1];
 
     { /* ensure that the best point is in the MAP as h/qpel refinement needs it */
-        const int key= (best[1]<<ME_MAP_MV_BITS) + best[0] + map_generation;
+        const unsigned key = (best[1]<<ME_MAP_MV_BITS) + best[0] + map_generation;
         const int index= ((best[1]<<ME_MAP_SHIFT) + best[0])&(ME_MAP_SIZE-1);
         if(map[index]!=key){ //this will be executed only very rarey
             score_map[index]= cmp(s, best[0], best[1], 0, 0, size, h, ref_index, src_index, cmpf, chroma_cmpf, flags);
@@ -464,7 +464,7 @@ static int funny_diamond_search(MpegEncContext * s, int *best, int dmin,
     int dia_size;
     LOAD_COMMON
     LOAD_COMMON2
-    int map_generation= c->map_generation;
+    unsigned map_generation = c->map_generation;
 
     cmpf= s->dsp.me_cmp[size];
     chroma_cmpf= s->dsp.me_cmp[size+1];
@@ -505,7 +505,7 @@ static int hex_search(MpegEncContext * s, int *best, int dmin,
     me_cmp_func cmpf, chroma_cmpf;
     LOAD_COMMON
     LOAD_COMMON2
-    int map_generation= c->map_generation;
+    unsigned map_generation = c->map_generation;
     int x,y,d;
     const int dec= dia_size & (dia_size-1);
 
@@ -539,7 +539,7 @@ static int l2s_dia_search(MpegEncContext * s, int *best, int dmin,
     me_cmp_func cmpf, chroma_cmpf;
     LOAD_COMMON
     LOAD_COMMON2
-    int map_generation= c->map_generation;
+    unsigned map_generation = c->map_generation;
     int x,y,i,d;
     int dia_size= c->dia_size&0xFF;
     const int dec= dia_size & (dia_size-1);
@@ -577,7 +577,7 @@ static int umh_search(MpegEncContext * s, int *best, int dmin,
     me_cmp_func cmpf, chroma_cmpf;
     LOAD_COMMON
     LOAD_COMMON2
-    int map_generation= c->map_generation;
+    unsigned map_generation = c->map_generation;
     int x,y,x2,y2, i, j, d;
     const int dia_size= c->dia_size&0xFE;
     static const int hex[16][2]={{-4,-2}, {-4,-1}, {-4, 0}, {-4, 1}, {-4, 2},
@@ -624,7 +624,7 @@ static int full_search(MpegEncContext * s, int *best, int dmin,
     me_cmp_func cmpf, chroma_cmpf;
     LOAD_COMMON
     LOAD_COMMON2
-    int map_generation= c->map_generation;
+    unsigned map_generation = c->map_generation;
     int x,y, d;
     const int dia_size= c->dia_size&0xFF;
 
@@ -653,7 +653,7 @@ static int full_search(MpegEncContext * s, int *best, int dmin,
 
 #define SAB_CHECK_MV(ax,ay)\
 {\
-    const int key= ((ay)<<ME_MAP_MV_BITS) + (ax) + map_generation;\
+    const unsigned key = ((ay)<<ME_MAP_MV_BITS) + (ax) + map_generation;\
     const int index= (((ay)<<ME_MAP_SHIFT) + (ax))&(ME_MAP_SIZE-1);\
 /*printf("sab check %d %d\n", ax, ay);*/\
     if(map[index]!=key){\
@@ -692,7 +692,7 @@ static int sab_diamond_search(MpegEncContext * s, int *best, int dmin,
     int i, j;
     LOAD_COMMON
     LOAD_COMMON2
-    int map_generation= c->map_generation;
+    unsigned map_generation = c->map_generation;
 
     cmpf= s->dsp.me_cmp[size];
     chroma_cmpf= s->dsp.me_cmp[size+1];
@@ -777,7 +777,7 @@ static int var_diamond_search(MpegEncContext * s, int *best, int dmin,
     int dia_size;
     LOAD_COMMON
     LOAD_COMMON2
-    int map_generation= c->map_generation;
+    unsigned map_generation = c->map_generation;
 
     cmpf= s->dsp.me_cmp[size];
     chroma_cmpf= s->dsp.me_cmp[size+1];
@@ -869,7 +869,7 @@ static av_always_inline int epzs_motion_search_internal(MpegEncContext * s, int
     int d;                   ///< the score (cmp + penalty) of any given mv
     int dmin;                /**< the best value of d, i.e. the score
                                corresponding to the mv stored in best[]. */
-    int map_generation;
+    unsigned map_generation;
     int penalty_factor;
     const int ref_mv_stride= s->mb_stride; //pass as arg  FIXME
     const int ref_mv_xy= s->mb_x + s->mb_y*ref_mv_stride; //add to last_mv beforepassing FIXME
@@ -997,7 +997,7 @@ static int epzs_motion_search4(MpegEncContext * s,
     MotionEstContext * const c= &s->me;
     int best[2]={0, 0};
     int d, dmin;
-    int map_generation;
+    unsigned map_generation;
     const int penalty_factor= c->penalty_factor;
     const int size=1;
     const int h=8;
@@ -1057,7 +1057,7 @@ static int epzs_motion_search2(MpegEncContext * s,
     MotionEstContext * const c= &s->me;
     int best[2]={0, 0};
     int d, dmin;
-    int map_generation;
+    unsigned map_generation;
     const int penalty_factor= c->penalty_factor;
     const int size=0; //FIXME pass as arg
     const int h=8;
diff --git a/libavcodec/mpegvideo.h b/libavcodec/mpegvideo.h
index 7d6c8d24b8..f640c8e315 100644
--- a/libavcodec/mpegvideo.h
+++ b/libavcodec/mpegvideo.h
@@ -156,7 +156,7 @@ typedef struct MotionEstContext{
     int best_bits;
     uint32_t *map;                     ///< map to avoid duplicate evaluations
     uint32_t *score_map;               ///< map to store the scores
-    int map_generation;
+    unsigned map_generation;
     int pre_penalty_factor;
     int penalty_factor;                /**< an estimate of the bits required to
                                         code a given mv value, e.g. (1,0) takes
diff --git a/libavcodec/proresdec_lgpl.c b/libavcodec/proresdec_lgpl.c
index 89e5582a37..156f87530a 100644
--- a/libavcodec/proresdec_lgpl.c
+++ b/libavcodec/proresdec_lgpl.c
@@ -34,17 +34,19 @@
 
 #include "libavutil/intmath.h"
 #include "avcodec.h"
-#include "dsputil.h"
+#include "proresdsp.h"
 #include "get_bits.h"
 
-#define BITS_PER_SAMPLE 10                              ///< output precision of that decoder
-#define BIAS     (1 << (BITS_PER_SAMPLE - 1))           ///< bias value for converting signed pixels into unsigned ones
-#define CLIP_MIN (1 << (BITS_PER_SAMPLE - 8))           ///< minimum value for clipping resulting pixels
-#define CLIP_MAX (1 << BITS_PER_SAMPLE) - CLIP_MIN - 1  ///< maximum value for clipping resulting pixels
-
+typedef struct {
+    const uint8_t *index;            ///< pointers to the data of this slice
+    int slice_num;
+    int x_pos, y_pos;
+    int slice_width;
+    DECLARE_ALIGNED(16, DCTELEM, blocks[8 * 4 * 64]);
+} ProresThreadData;
 
 typedef struct {
-    DSPContext dsp;
+    ProresDSPContext dsp;
     AVFrame    picture;
     ScanTable  scantable;
     int        scantable_type;           ///< -1 = uninitialized, 0 = progressive, 1/2 = interlaced
@@ -57,9 +59,9 @@ typedef struct {
     int        prev_slice_sf;            ///< scalefactor of the previous decoded slice
     DECLARE_ALIGNED(16, int16_t, qmat_luma_scaled[64]);
     DECLARE_ALIGNED(16, int16_t, qmat_chroma_scaled[64]);
-    DECLARE_ALIGNED(16, DCTELEM, blocks[8 * 4 * 64]);
     int        total_slices;            ///< total number of slices in a picture
-    const uint8_t **slice_data_index;   ///< array of pointers to the data of each slice
+    ProresThreadData *slice_data;
+    int        pic_num;
     int        chroma_factor;
     int        mb_chroma_factor;
     int        num_chroma_blocks;       ///< number of chrominance blocks in a macroblock
@@ -100,12 +102,12 @@ static av_cold int decode_init(AVCodecContext *avctx)
     ProresContext *ctx = avctx->priv_data;
 
     ctx->total_slices     = 0;
-    ctx->slice_data_index = 0;
+    ctx->slice_data       = NULL;
 
     avctx->pix_fmt = PIX_FMT_YUV422P10; // set default pixel format
 
-    avctx->bits_per_raw_sample = BITS_PER_SAMPLE;
-    dsputil_init(&ctx->dsp, avctx);
+    avctx->bits_per_raw_sample = PRORES_BITS_PER_SAMPLE;
+    ff_proresdsp_init(&ctx->dsp);
 
     avctx->coded_frame = &ctx->picture;
     avcodec_get_frame_defaults(&ctx->picture);
@@ -271,9 +273,9 @@ static int decode_picture_header(ProresContext *ctx, const uint8_t *buf,
     }
 
     if (ctx->total_slices != num_slices) {
-        av_freep(&ctx->slice_data_index);
-        ctx->slice_data_index = av_malloc((num_slices + 1) * sizeof(uint8_t*));
-        if (!ctx->slice_data_index)
+        av_freep(&ctx->slice_data);
+        ctx->slice_data = av_malloc((num_slices + 1) * sizeof(ctx->slice_data[0]));
+        if (!ctx->slice_data)
             return AVERROR(ENOMEM);
         ctx->total_slices = num_slices;
     }
@@ -288,10 +290,10 @@ static int decode_picture_header(ProresContext *ctx, const uint8_t *buf,
     data_ptr = index_ptr + num_slices * 2;
 
     for (i = 0; i < num_slices; i++) {
-        ctx->slice_data_index[i] = data_ptr;
+        ctx->slice_data[i].index = data_ptr;
         data_ptr += AV_RB16(index_ptr + i * 2);
     }
-    ctx->slice_data_index[i] = data_ptr;
+    ctx->slice_data[i].index = data_ptr;
 
     if (data_ptr > buf + data_size) {
         av_log(avctx, AV_LOG_ERROR, "out of slice data\n");
@@ -449,52 +451,11 @@ static inline void decode_ac_coeffs(GetBitContext *gb, DCTELEM *out,
 }
 
 
-#define CLIP_AND_BIAS(x) (av_clip((x) + BIAS, CLIP_MIN, CLIP_MAX))
-
-/**
- * Add bias value, clamp and output pixels of a slice
- */
-static void put_pixels(const DCTELEM *in, uint16_t *out, int stride,
-                       int mbs_per_slice, int blocks_per_mb)
-{
-    int mb, x, y, src_offset, dst_offset;
-    const DCTELEM *src1, *src2;
-    uint16_t *dst1, *dst2;
-
-    src1 = in;
-    src2 = in + (blocks_per_mb << 5);
-    dst1 = out;
-    dst2 = out + (stride << 3);
-
-    for (mb = 0; mb < mbs_per_slice; mb++) {
-        for (y = 0, dst_offset = 0; y < 8; y++, dst_offset += stride) {
-            for (x = 0; x < 8; x++) {
-                src_offset = (y << 3) + x;
-
-                dst1[dst_offset + x] = CLIP_AND_BIAS(src1[src_offset]);
-                dst2[dst_offset + x] = CLIP_AND_BIAS(src2[src_offset]);
-
-                if (blocks_per_mb > 2) {
-                    dst1[dst_offset + x + 8] =
-                        CLIP_AND_BIAS(src1[src_offset + 64]);
-                    dst2[dst_offset + x + 8] =
-                        CLIP_AND_BIAS(src2[src_offset + 64]);
-                }
-            }
-        }
-
-        src1 += blocks_per_mb << 6;
-        src2 += blocks_per_mb << 6;
-        dst1 += blocks_per_mb << 2;
-        dst2 += blocks_per_mb << 2;
-    }
-}
-
-
 /**
  * Decode a slice plane (luma or chroma).
  */
-static void decode_slice_plane(ProresContext *ctx, const uint8_t *buf,
+static void decode_slice_plane(ProresContext *ctx, ProresThreadData *td,
+                               const uint8_t *buf,
                                int data_size, uint16_t *out_ptr,
                                int linesize, int mbs_per_slice,
                                int blocks_per_mb, int plane_size_factor,
@@ -502,43 +463,47 @@ static void decode_slice_plane(ProresContext *ctx, const uint8_t *buf,
 {
     GetBitContext gb;
     DCTELEM *block_ptr;
-    int i, blk_num, blocks_per_slice;
+    int mb_num, blocks_per_slice;
 
     blocks_per_slice = mbs_per_slice * blocks_per_mb;
 
-    memset(ctx->blocks, 0, 8 * 4 * 64 * sizeof(*ctx->blocks));
+    memset(td->blocks, 0, 8 * 4 * 64 * sizeof(*td->blocks));
 
     init_get_bits(&gb, buf, data_size << 3);
 
-    decode_dc_coeffs(&gb, ctx->blocks, blocks_per_slice);
+    decode_dc_coeffs(&gb, td->blocks, blocks_per_slice);
 
-    decode_ac_coeffs(&gb, ctx->blocks, blocks_per_slice,
+    decode_ac_coeffs(&gb, td->blocks, blocks_per_slice,
                      plane_size_factor, ctx->scantable.permutated);
 
     /* inverse quantization, inverse transform and output */
-    block_ptr = ctx->blocks;
-
-    for (blk_num = 0; blk_num < blocks_per_slice; blk_num++, block_ptr += 64) {
-        /* TODO: the correct solution shoud be (block_ptr[i] * qmat[i]) >> 1
-         * and the input of the inverse transform should be scaled by 2
-         * in order to avoid rounding errors.
-         * Due to the fact the existing Libav transforms are incompatible with
-         * that input I temporally introduced the coarse solution below... */
-        for (i = 0; i < 64; i++)
-            block_ptr[i] = (block_ptr[i] * qmat[i]) >> 2;
-
-        ctx->dsp.idct(block_ptr);
+    block_ptr = td->blocks;
+
+    for (mb_num = 0; mb_num < mbs_per_slice; mb_num++, out_ptr += blocks_per_mb * 4) {
+        ctx->dsp.idct_put(out_ptr,                    linesize, block_ptr, qmat);
+        block_ptr += 64;
+        if (blocks_per_mb > 2) {
+            ctx->dsp.idct_put(out_ptr + 8,            linesize, block_ptr, qmat);
+            block_ptr += 64;
+        }
+        ctx->dsp.idct_put(out_ptr + linesize * 4,     linesize, block_ptr, qmat);
+        block_ptr += 64;
+        if (blocks_per_mb > 2) {
+            ctx->dsp.idct_put(out_ptr + linesize * 4 + 8, linesize, block_ptr, qmat);
+            block_ptr += 64;
+        }
     }
-
-    put_pixels(ctx->blocks, out_ptr, linesize >> 1, mbs_per_slice,
-               blocks_per_mb);
 }
 
 
-static int decode_slice(ProresContext *ctx, int pic_num, int slice_num,
-                        int mb_x_pos, int mb_y_pos, int mbs_per_slice,
-                        AVCodecContext *avctx)
+static int decode_slice(AVCodecContext *avctx, ProresThreadData *td)
 {
+    ProresContext *ctx = avctx->priv_data;
+    int mb_x_pos  = td->x_pos;
+    int mb_y_pos  = td->y_pos;
+    int pic_num   = ctx->pic_num;
+    int slice_num = td->slice_num;
+    int mbs_per_slice = td->slice_width;
     const uint8_t *buf;
     uint8_t *y_data, *u_data, *v_data;
     AVFrame *pic = avctx->coded_frame;
@@ -546,8 +511,8 @@ static int decode_slice(ProresContext *ctx, int pic_num, int slice_num,
     int slice_data_size, hdr_size, y_data_size, u_data_size, v_data_size;
     int y_linesize, u_linesize, v_linesize;
 
-    buf             = ctx->slice_data_index[slice_num];
-    slice_data_size = ctx->slice_data_index[slice_num + 1] - buf;
+    buf             = ctx->slice_data[slice_num].index;
+    slice_data_size = ctx->slice_data[slice_num + 1].index - buf;
 
     slice_width_factor = av_log2(mbs_per_slice);
 
@@ -593,20 +558,20 @@ static int decode_slice(ProresContext *ctx, int pic_num, int slice_num,
     if (ctx->qmat_changed || sf != ctx->prev_slice_sf) {
         ctx->prev_slice_sf = sf;
         for (i = 0; i < 64; i++) {
-            ctx->qmat_luma_scaled[i]   = ctx->qmat_luma[i]   * sf;
-            ctx->qmat_chroma_scaled[i] = ctx->qmat_chroma[i] * sf;
+            ctx->qmat_luma_scaled[ctx->dsp.idct_permutation[i]]   = ctx->qmat_luma[i]   * sf;
+            ctx->qmat_chroma_scaled[ctx->dsp.idct_permutation[i]] = ctx->qmat_chroma[i] * sf;
         }
     }
 
     /* decode luma plane */
-    decode_slice_plane(ctx, buf + hdr_size, y_data_size,
+    decode_slice_plane(ctx, td, buf + hdr_size, y_data_size,
                        (uint16_t*) (y_data + (mb_y_pos << 4) * y_linesize +
                                     (mb_x_pos << 5)), y_linesize,
                        mbs_per_slice, 4, slice_width_factor + 2,
                        ctx->qmat_luma_scaled);
 
     /* decode U chroma plane */
-    decode_slice_plane(ctx, buf + hdr_size + y_data_size, u_data_size,
+    decode_slice_plane(ctx, td, buf + hdr_size + y_data_size, u_data_size,
                        (uint16_t*) (u_data + (mb_y_pos << 4) * u_linesize +
                                     (mb_x_pos << ctx->mb_chroma_factor)),
                        u_linesize, mbs_per_slice, ctx->num_chroma_blocks,
@@ -614,7 +579,7 @@ static int decode_slice(ProresContext *ctx, int pic_num, int slice_num,
                        ctx->qmat_chroma_scaled);
 
     /* decode V chroma plane */
-    decode_slice_plane(ctx, buf + hdr_size + y_data_size + u_data_size,
+    decode_slice_plane(ctx, td, buf + hdr_size + y_data_size + u_data_size,
                        v_data_size,
                        (uint16_t*) (v_data + (mb_y_pos << 4) * v_linesize +
                                     (mb_x_pos << ctx->mb_chroma_factor)),
@@ -633,6 +598,7 @@ static int decode_picture(ProresContext *ctx, int pic_num,
 
     slice_num = 0;
 
+    ctx->pic_num = pic_num;
     for (y_pos = 0; y_pos < ctx->num_y_mbs; y_pos++) {
         slice_width = 1 << ctx->slice_width_factor;
 
@@ -641,15 +607,18 @@ static int decode_picture(ProresContext *ctx, int pic_num,
             while (ctx->num_x_mbs - x_pos < slice_width)
                 slice_width >>= 1;
 
-            if (decode_slice(ctx, pic_num, slice_num, x_pos, y_pos,
-                             slice_width, avctx) < 0)
-                return -1;
+            ctx->slice_data[slice_num].slice_num   = slice_num;
+            ctx->slice_data[slice_num].x_pos       = x_pos;
+            ctx->slice_data[slice_num].y_pos       = y_pos;
+            ctx->slice_data[slice_num].slice_width = slice_width;
 
             slice_num++;
         }
     }
 
-    return 0;
+    return avctx->execute(avctx, (void *) decode_slice,
+                          ctx->slice_data, NULL, slice_num,
+                          sizeof(ctx->slice_data[0]));
 }
 
 
@@ -712,7 +681,7 @@ static av_cold int decode_close(AVCodecContext *avctx)
     if (ctx->picture.data[0])
         avctx->release_buffer(avctx, &ctx->picture);
 
-    av_freep(&ctx->slice_data_index);
+    av_freep(&ctx->slice_data);
 
     return 0;
 }
@@ -726,6 +695,6 @@ AVCodec ff_prores_lgpl_decoder = {
     .init           = decode_init,
     .close          = decode_close,
     .decode         = decode_frame,
-    .capabilities   = CODEC_CAP_DR1,
+    .capabilities   = CODEC_CAP_DR1 | CODEC_CAP_SLICE_THREADS,
     .long_name      = NULL_IF_CONFIG_SMALL("Apple ProRes (iCodec Pro)")
 };
diff --git a/libavcodec/proresdsp.c b/libavcodec/proresdsp.c
new file mode 100644
index 0000000000..7e753e9dc7
--- /dev/null
+++ b/libavcodec/proresdsp.c
@@ -0,0 +1,63 @@
+/*
+ * Apple ProRes compatible decoder
+ *
+ * Copyright (c) 2010-2011 Maxim Poliakovski
+ *
+ * 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 "proresdsp.h"
+#include "simple_idct.h"
+
+#define BIAS     (1 << (PRORES_BITS_PER_SAMPLE - 1))           ///< bias value for converting signed pixels into unsigned ones
+#define CLIP_MIN (1 << (PRORES_BITS_PER_SAMPLE - 8))           ///< minimum value for clipping resulting pixels
+#define CLIP_MAX (1 << PRORES_BITS_PER_SAMPLE) - CLIP_MIN - 1  ///< maximum value for clipping resulting pixels
+
+#define CLIP_AND_BIAS(x) (av_clip((x) + BIAS, CLIP_MIN, CLIP_MAX))
+
+/**
+ * Add bias value, clamp and output pixels of a slice
+ */
+static void put_pixels(uint16_t *dst, int stride, const DCTELEM *in)
+{
+    int x, y, src_offset, dst_offset;
+
+    for (y = 0, dst_offset = 0; y < 8; y++, dst_offset += stride) {
+        for (x = 0; x < 8; x++) {
+            src_offset = (y << 3) + x;
+
+            dst[dst_offset + x] = CLIP_AND_BIAS(in[src_offset]);
+        }
+    }
+}
+
+static void prores_idct_put_c(uint16_t *out, int linesize, DCTELEM *block, const int16_t *qmat)
+{
+    ff_prores_idct(block, qmat);
+    put_pixels(out, linesize >> 1, block);
+}
+
+void ff_proresdsp_init(ProresDSPContext *dsp)
+{
+    dsp->idct_put = prores_idct_put_c;
+    dsp->idct_permutation_type = FF_NO_IDCT_PERM;
+
+    if (HAVE_MMX) ff_proresdsp_x86_init(dsp);
+
+    ff_init_scantable_permutation(dsp->idct_permutation,
+                                  dsp->idct_permutation_type);
+}
diff --git a/libavcodec/proresdsp.h b/libavcodec/proresdsp.h
new file mode 100644
index 0000000000..8b864faabd
--- /dev/null
+++ b/libavcodec/proresdsp.h
@@ -0,0 +1,40 @@
+/*
+ * Apple ProRes compatible decoder
+ *
+ * Copyright (c) 2010-2011 Maxim Poliakovski
+ *
+ * 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
+ */
+
+#ifndef AVCODEC_PRORESDSP_H
+#define AVCODEC_PRORESDSP_H
+
+#include "dsputil.h"
+
+#define PRORES_BITS_PER_SAMPLE 10 ///< output precision of prores decoder
+
+typedef struct {
+    int idct_permutation_type;
+    uint8_t idct_permutation[64];
+    void (* idct_put) (uint16_t *out, int linesize, DCTELEM *block, const int16_t *qmat);
+} ProresDSPContext;
+
+void ff_proresdsp_init(ProresDSPContext *dsp);
+
+void ff_proresdsp_x86_init(ProresDSPContext *dsp);
+
+#endif /* AVCODEC_PRORESDSP_H */
diff --git a/libavcodec/simple_idct.c b/libavcodec/simple_idct.c
index ffe5a0b070..b4e763a704 100644
--- a/libavcodec/simple_idct.c
+++ b/libavcodec/simple_idct.c
@@ -221,3 +221,20 @@ void ff_simple_idct44_add(uint8_t *dest, int line_size, DCTELEM *block)
         idct4col_add(dest + i, line_size, block + i);
     }
 }
+
+void ff_prores_idct(DCTELEM *block, const int16_t *qmat)
+{
+    int i;
+
+    for (i = 0; i < 64; i++)
+        block[i] *= qmat[i];
+
+    for (i = 0; i < 8; i++)
+        idctRowCondDC_10(block + i*8);
+
+    for (i = 0; i < 64; i++)
+        block[i] >>= 2;
+
+    for (i = 0; i < 8; i++)
+        idctSparseCol_10(block + i);
+}
diff --git a/libavcodec/simple_idct.h b/libavcodec/simple_idct.h
index 10ac7da5e6..64d3c2ac31 100644
--- a/libavcodec/simple_idct.h
+++ b/libavcodec/simple_idct.h
@@ -38,6 +38,12 @@ void ff_simple_idct_8(DCTELEM *block);
 void ff_simple_idct_put_10(uint8_t *dest, int line_size, DCTELEM *block);
 void ff_simple_idct_add_10(uint8_t *dest, int line_size, DCTELEM *block);
 void ff_simple_idct_10(DCTELEM *block);
+/**
+ * Special version of ff_simple_idct_10() which does dequantization
+ * and scales by a factor of 2 more between the two IDCTs to account
+ * for larger scale of input coefficients.
+ */
+void ff_prores_idct(DCTELEM *block, const int16_t *qmat);
 
 void ff_simple_idct_mmx(int16_t *block);
 void ff_simple_idct_add_mmx(uint8_t *dest, int line_size, int16_t *block);
diff --git a/libavcodec/sipr.c b/libavcodec/sipr.c
index a2e2fe4267..098d1074a7 100644
--- a/libavcodec/sipr.c
+++ b/libavcodec/sipr.c
@@ -194,14 +194,16 @@ static void decode_parameters(SiprParameters* parms, GetBitContext *pgb,
 {
     int i, j;
 
-    parms->ma_pred_switch           = get_bits(pgb, p->ma_predictor_bits);
+    if (p->ma_predictor_bits)
+        parms->ma_pred_switch       = get_bits(pgb, p->ma_predictor_bits);
 
     for (i = 0; i < 5; i++)
         parms->vq_indexes[i]        = get_bits(pgb, p->vq_indexes_bits[i]);
 
     for (i = 0; i < p->subframe_count; i++) {
         parms->pitch_delay[i]       = get_bits(pgb, p->pitch_delay_bits[i]);
-        parms->gp_index[i]          = get_bits(pgb, p->gp_index_bits);
+        if (p->gp_index_bits)
+            parms->gp_index[i]      = get_bits(pgb, p->gp_index_bits);
 
         for (j = 0; j < p->number_of_fc_indexes; j++)
             parms->fc_indexes[i][j] = get_bits(pgb, p->fc_index_bits[j]);
diff --git a/libavcodec/smacker.c b/libavcodec/smacker.c
index 3a4ba1a114..c7eafbc261 100644
--- a/libavcodec/smacker.c
+++ b/libavcodec/smacker.c
@@ -560,6 +560,10 @@ static av_cold int decode_end(AVCodecContext *avctx)
 
 static av_cold int smka_decode_init(AVCodecContext *avctx)
 {
+    if (avctx->channels < 1 || avctx->channels > 2) {
+        av_log(avctx, AV_LOG_ERROR, "invalid number of channels\n");
+        return AVERROR(EINVAL);
+    }
     avctx->channel_layout = (avctx->channels==2) ? AV_CH_LAYOUT_STEREO : AV_CH_LAYOUT_MONO;
     avctx->sample_fmt = avctx->bits_per_coded_sample == 8 ? AV_SAMPLE_FMT_U8 : AV_SAMPLE_FMT_S16;
     return 0;
@@ -583,6 +587,11 @@ static int smka_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
     int bits, stereo;
     int pred[2] = {0, 0};
 
+    if (buf_size <= 4) {
+        av_log(avctx, AV_LOG_ERROR, "packet is too small\n");
+        return AVERROR(EINVAL);
+    }
+
     unp_size = AV_RL32(buf);
 
     init_get_bits(&gb, buf + 4, (buf_size - 4) * 8);
@@ -598,6 +607,14 @@ static int smka_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
         av_log(avctx, AV_LOG_ERROR, "Frame is too large to fit in buffer\n");
         return -1;
     }
+    if (stereo ^ (avctx->channels != 1)) {
+        av_log(avctx, AV_LOG_ERROR, "channels mismatch\n");
+        return AVERROR(EINVAL);
+    }
+    if (bits && avctx->sample_fmt == AV_SAMPLE_FMT_U8) {
+        av_log(avctx, AV_LOG_ERROR, "sample format mismatch\n");
+        return AVERROR(EINVAL);
+    }
 
     memset(vlc, 0, sizeof(VLC) * 4);
     memset(h, 0, sizeof(HuffContext) * 4);
diff --git a/libavcodec/x86/Makefile b/libavcodec/x86/Makefile
index bce2c3ecaf..d031f6505b 100644
--- a/libavcodec/x86/Makefile
+++ b/libavcodec/x86/Makefile
@@ -34,6 +34,8 @@ MMX-OBJS-$(CONFIG_ENCODERS)            += x86/dsputilenc_mmx.o
 YASM-OBJS-$(CONFIG_ENCODERS)           += x86/dsputilenc_yasm.o
 MMX-OBJS-$(CONFIG_GPL)                 += x86/idct_mmx.o
 MMX-OBJS-$(CONFIG_LPC)                 += x86/lpc_mmx.o
+YASM-OBJS-$(CONFIG_PRORES_LGPL_DECODER)     += x86/proresdsp.o
+MMX-OBJS-$(CONFIG_PRORES_LGPL_DECODER)      += x86/proresdsp-init.o
 MMX-OBJS-$(CONFIG_DWT)                 += x86/snowdsp_mmx.o
 MMX-OBJS-$(CONFIG_VC1_DECODER)         += x86/vc1dsp_mmx.o
 YASM-OBJS-$(CONFIG_VP3_DECODER)        += x86/vp3dsp.o
diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index 96983773da..2322a29fb7 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -64,6 +64,8 @@ DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_64 ) = {0x0040004000400040ULL, 0x00400
 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_96 ) = 0x0060006000600060ULL;
 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_128) = 0x0080008000800080ULL;
 DECLARE_ALIGNED(8,  const uint64_t, ff_pw_255) = 0x00ff00ff00ff00ffULL;
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_512) = {0x0200020002000200ULL, 0x0200020002000200ULL};
+DECLARE_ALIGNED(16, const xmm_reg,  ff_pw_1019)= {0x03FB03FB03FB03FBULL, 0x03FB03FB03FB03FBULL};
 
 DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_0  ) = {0x0000000000000000ULL, 0x0000000000000000ULL};
 DECLARE_ALIGNED(16, const xmm_reg,  ff_pb_1  ) = {0x0101010101010101ULL, 0x0101010101010101ULL};
diff --git a/libavcodec/x86/proresdsp-init.c b/libavcodec/x86/proresdsp-init.c
new file mode 100644
index 0000000000..9760105ab6
--- /dev/null
+++ b/libavcodec/x86/proresdsp-init.c
@@ -0,0 +1,54 @@
+/*
+ * Apple ProRes compatible decoder
+ *
+ * Copyright (c) 2010-2011 Maxim Poliakovski
+ *
+ * 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/proresdsp.h"
+
+void ff_prores_idct_put_10_sse2(uint16_t *dst, int linesize,
+                                DCTELEM *block);
+void ff_prores_idct_put_10_sse4(uint16_t *dst, int linesize,
+                                DCTELEM *block);
+void ff_prores_idct_put_10_avx (uint16_t *dst, int linesize,
+                                DCTELEM *block);
+
+void ff_proresdsp_x86_init(ProresDSPContext *dsp)
+{
+#if ARCH_X86_64
+    int flags = av_get_cpu_flags();
+
+    if (flags & AV_CPU_FLAG_SSE2) {
+        dsp->idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
+        dsp->idct_put = ff_prores_idct_put_10_sse2;
+    }
+
+    if (flags & AV_CPU_FLAG_SSE4) {
+        dsp->idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
+        dsp->idct_put = ff_prores_idct_put_10_sse4;
+    }
+
+#if HAVE_AVX
+    if (flags & AV_CPU_FLAG_AVX) {
+        dsp->idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
+        dsp->idct_put = ff_prores_idct_put_10_avx;
+    }
+#endif
+#endif
+}
diff --git a/libavcodec/x86/proresdsp.asm b/libavcodec/x86/proresdsp.asm
new file mode 100644
index 0000000000..9365bf1ea4
--- /dev/null
+++ b/libavcodec/x86/proresdsp.asm
@@ -0,0 +1,432 @@
+;******************************************************************************
+;* x86-SIMD-optimized IDCT for prores
+;* this is identical to "simple" IDCT except for the clip range
+;*
+;* Copyright (c) 2011 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
+;* 51, Inc., Foundation Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+;******************************************************************************
+
+%include "x86inc.asm"
+%include "x86util.asm"
+
+%define W1sh2 22725 ; W1 = 90901 = 22725<<2 + 1
+%define W2sh2 21407 ; W2 = 85627 = 21407<<2 - 1
+%define W3sh2 19265 ; W3 = 77062 = 19265<<2 + 2
+%define W4sh2 16384 ; W4 = 65535 = 16384<<2 - 1
+%define W5sh2 12873 ; W5 = 51491 = 12873<<2 - 1
+%define W6sh2  8867 ; W6 = 35468 =  8867<<2
+%define W7sh2  4520 ; W7 = 18081 =  4520<<2 + 1
+
+%ifdef ARCH_X86_64
+
+SECTION_RODATA
+
+w4_plus_w2: times 4 dw W4sh2, +W2sh2
+w4_min_w2:  times 4 dw W4sh2, -W2sh2
+w4_plus_w6: times 4 dw W4sh2, +W6sh2
+w4_min_w6:  times 4 dw W4sh2, -W6sh2
+w1_plus_w3: times 4 dw W1sh2, +W3sh2
+w3_min_w1:  times 4 dw W3sh2, -W1sh2
+w7_plus_w3: times 4 dw W7sh2, +W3sh2
+w3_min_w7:  times 4 dw W3sh2, -W7sh2
+w1_plus_w5: times 4 dw W1sh2, +W5sh2
+w5_min_w1:  times 4 dw W5sh2, -W1sh2
+w5_plus_w7: times 4 dw W5sh2, +W7sh2
+w7_min_w5:  times 4 dw W7sh2, -W5sh2
+row_round:  times 8 dw (1<<14)
+
+cextern pw_4
+cextern pw_8
+cextern pw_512
+cextern pw_1019
+
+section .text align=16
+
+; interleave data while maintaining source
+; %1=type, %2=dstlo, %3=dsthi, %4=src, %5=interleave
+%macro SBUTTERFLY3 5
+    punpckl%1   m%2, m%4, m%5
+    punpckh%1   m%3, m%4, m%5
+%endmacro
+
+; %1/%2=src1/dst1, %3/%4=dst2, %5/%6=src2, %7=shift
+; action: %3/%4 = %1/%2 - %5/%6; %1/%2 += %5/%6
+;         %1/%2/%3/%4 >>= %7; dword -> word (in %1/%3)
+%macro SUMSUB_SHPK 7
+    psubd       %3,  %1,  %5       ; { a0 - b0 }[0-3]
+    psubd       %4,  %2,  %6       ; { a0 - b0 }[4-7]
+    paddd       %1,  %5            ; { a0 + b0 }[0-3]
+    paddd       %2,  %6            ; { a0 + b0 }[4-7]
+    psrad       %1,  %7
+    psrad       %2,  %7
+    psrad       %3,  %7
+    psrad       %4,  %7
+    packssdw    %1,  %2            ; row[0]
+    packssdw    %3,  %4            ; row[7]
+%endmacro
+
+; %1 = row or col (for rounding variable)
+; %2 = number of bits to shift at the end
+; %3 = optimization
+%macro IDCT_1D 3
+    ; a0 = (W4 * row[0]) + (1 << (15 - 1));
+    ; a1 = a0;
+    ; a2 = a0;
+    ; a3 = a0;
+    ; a0 += W2 * row[2];
+    ; a1 += W6 * row[2];
+    ; a2 -= W6 * row[2];
+    ; a3 -= W2 * row[2];
+%ifidn %1, col
+    paddw       m10,[pw_8]
+%endif
+    SBUTTERFLY3 wd,  0,  1, 10,  8 ; { row[0], row[2] }[0-3]/[4-7]
+%ifidn %1, row
+    psubw       m10,[row_round]
+%endif
+    SIGNEXTEND  m8,  m9,  m14      ; { row[2] }[0-3] / [4-7]
+    SIGNEXTEND  m10, m11, m14      ; { row[0] }[0-3] / [4-7]
+    pmaddwd     m2,  m0, [w4_plus_w6]
+    pmaddwd     m3,  m1, [w4_plus_w6]
+    pmaddwd     m4,  m0, [w4_min_w6]
+    pmaddwd     m5,  m1, [w4_min_w6]
+    pmaddwd     m6,  m0, [w4_min_w2]
+    pmaddwd     m7,  m1, [w4_min_w2]
+    pmaddwd     m0, [w4_plus_w2]
+    pmaddwd     m1, [w4_plus_w2]
+    pslld       m2,  2
+    pslld       m3,  2
+    pslld       m4,  2
+    pslld       m5,  2
+    pslld       m6,  2
+    pslld       m7,  2
+    pslld       m0,  2
+    pslld       m1,  2
+
+    ; a0: -1*row[0]-1*row[2]
+    ; a1: -1*row[0]
+    ; a2: -1*row[0]
+    ; a3: -1*row[0]+1*row[2]
+    psubd       m2,  m10           ; a1[0-3]
+    psubd       m3,  m11           ; a1[4-7]
+    psubd       m4,  m10           ; a2[0-3]
+    psubd       m5,  m11           ; a2[4-7]
+    psubd       m0,  m10
+    psubd       m1,  m11
+    psubd       m6,  m10
+    psubd       m7,  m11
+    psubd       m0,  m8            ; a0[0-3]
+    psubd       m1,  m9            ; a0[4-7]
+    paddd       m6,  m8            ; a3[0-3]
+    paddd       m7,  m9            ; a3[4-7]
+
+    ; a0 +=   W4*row[4] + W6*row[6]; i.e. -1*row[4]
+    ; a1 -=   W4*row[4] + W2*row[6]; i.e. -1*row[4]-1*row[6]
+    ; a2 -=   W4*row[4] - W2*row[6]; i.e. -1*row[4]+1*row[6]
+    ; a3 +=   W4*row[4] - W6*row[6]; i.e. -1*row[4]
+    SBUTTERFLY3 wd,  8,  9, 13, 12 ; { row[4], row[6] }[0-3]/[4-7]
+    SIGNEXTEND  m13, m14, m10      ; { row[4] }[0-3] / [4-7]
+    pmaddwd     m10, m8, [w4_plus_w6]
+    pmaddwd     m11, m9, [w4_plus_w6]
+    pslld       m10, 2
+    pslld       m11, 2
+    psubd       m10,  m13
+    psubd       m11,  m14
+    paddd       m0,  m10            ; a0[0-3]
+    paddd       m1,  m11            ; a0[4-7]
+    pmaddwd     m10, m8, [w4_min_w6]
+    pmaddwd     m11, m9, [w4_min_w6]
+    pslld       m10, 2
+    pslld       m11, 2
+    psubd       m10, m13
+    psubd       m11, m14
+    paddd       m6,  m10           ; a3[0-3]
+    paddd       m7,  m11           ; a3[4-7]
+    pmaddwd     m10, m8, [w4_min_w2]
+    pmaddwd     m11, m9, [w4_min_w2]
+    pmaddwd     m8, [w4_plus_w2]
+    pmaddwd     m9, [w4_plus_w2]
+    pslld       m10, 2
+    pslld       m11, 2
+    pslld       m8,  2
+    pslld       m9,  2
+    psubd       m10, m13
+    psubd       m11, m14
+    psubd       m8,  m13
+    psubd       m9,  m14
+    psubd       m4,  m10           ; a2[0-3] intermediate
+    psubd       m5,  m11           ; a2[4-7] intermediate
+    psubd       m2,  m8            ; a1[0-3] intermediate
+    psubd       m3,  m9            ; a1[4-7] intermediate
+    SIGNEXTEND  m12, m13, m10      ; { row[6] }[0-3] / [4-7]
+    psubd       m4,  m12           ; a2[0-3]
+    psubd       m5,  m13           ; a2[4-7]
+    paddd       m2,  m12           ; a1[0-3]
+    paddd       m3,  m13           ; a1[4-7]
+
+    ; load/store
+    mova   [r2+  0], m0
+    mova   [r2+ 32], m2
+    mova   [r2+ 64], m4
+    mova   [r2+ 96], m6
+    mova        m10,[r2+ 16]       ; { row[1] }[0-7]
+    mova        m8, [r2+ 48]       ; { row[3] }[0-7]
+    mova        m13,[r2+ 80]       ; { row[5] }[0-7]
+    mova        m14,[r2+112]       ; { row[7] }[0-7]
+    mova   [r2+ 16], m1
+    mova   [r2+ 48], m3
+    mova   [r2+ 80], m5
+    mova   [r2+112], m7
+%ifidn %1, row
+    pmullw      m10,[r3+ 16]
+    pmullw      m8, [r3+ 48]
+    pmullw      m13,[r3+ 80]
+    pmullw      m14,[r3+112]
+%endif
+
+    ; b0 = MUL(W1, row[1]);
+    ; MAC(b0, W3, row[3]);
+    ; b1 = MUL(W3, row[1]);
+    ; MAC(b1, -W7, row[3]);
+    ; b2 = MUL(W5, row[1]);
+    ; MAC(b2, -W1, row[3]);
+    ; b3 = MUL(W7, row[1]);
+    ; MAC(b3, -W5, row[3]);
+    SBUTTERFLY3 wd,  0,  1, 10, 8  ; { row[1], row[3] }[0-3]/[4-7]
+    SIGNEXTEND  m10, m11, m12      ; { row[1] }[0-3] / [4-7]
+    SIGNEXTEND  m8,  m9,  m12      ; { row[3] }[0-3] / [4-7]
+    pmaddwd     m2,  m0, [w3_min_w7]
+    pmaddwd     m3,  m1, [w3_min_w7]
+    pmaddwd     m4,  m0, [w5_min_w1]
+    pmaddwd     m5,  m1, [w5_min_w1]
+    pmaddwd     m6,  m0, [w7_min_w5]
+    pmaddwd     m7,  m1, [w7_min_w5]
+    pmaddwd     m0, [w1_plus_w3]
+    pmaddwd     m1, [w1_plus_w3]
+    pslld       m2,  2
+    pslld       m3,  2
+    pslld       m4,  2
+    pslld       m5,  2
+    pslld       m6,  2
+    pslld       m7,  2
+    pslld       m0,  2
+    pslld       m1,  2
+
+    ; b0: +1*row[1]+2*row[3]
+    ; b1: +2*row[1]-1*row[3]
+    ; b2: -1*row[1]-1*row[3]
+    ; b3: +1*row[1]+1*row[3]
+    psubd       m2,  m8
+    psubd       m3,  m9
+    paddd       m0,  m8
+    paddd       m1,  m9
+    paddd       m8,  m10           ; { row[1] + row[3] }[0-3]
+    paddd       m9,  m11           ; { row[1] + row[3] }[4-7]
+    paddd       m10, m10
+    paddd       m11, m11
+    paddd       m0,  m8            ; b0[0-3]
+    paddd       m1,  m9            ; b0[4-7]
+    paddd       m2,  m10           ; b1[0-3]
+    paddd       m3,  m11           ; b2[4-7]
+    psubd       m4,  m8            ; b2[0-3]
+    psubd       m5,  m9            ; b2[4-7]
+    paddd       m6,  m8            ; b3[0-3]
+    paddd       m7,  m9            ; b3[4-7]
+
+    ; MAC(b0,  W5, row[5]);
+    ; MAC(b0,  W7, row[7]);
+    ; MAC(b1, -W1, row[5]);
+    ; MAC(b1, -W5, row[7]);
+    ; MAC(b2,  W7, row[5]);
+    ; MAC(b2,  W3, row[7]);
+    ; MAC(b3,  W3, row[5]);
+    ; MAC(b3, -W1, row[7]);
+    SBUTTERFLY3 wd,  8,  9, 13, 14 ; { row[5], row[7] }[0-3]/[4-7]
+    SIGNEXTEND  m13, m12, m11      ; { row[5] }[0-3] / [4-7]
+    SIGNEXTEND  m14, m11, m10      ; { row[7] }[0-3] / [4-7]
+
+    ; b0: -1*row[5]+1*row[7]
+    ; b1: -1*row[5]+1*row[7]
+    ; b2: +1*row[5]+2*row[7]
+    ; b3: +2*row[5]-1*row[7]
+    paddd       m4,  m13
+    paddd       m5,  m12
+    paddd       m6,  m13
+    paddd       m7,  m12
+    psubd       m13, m14           ; { row[5] - row[7] }[0-3]
+    psubd       m12, m11           ; { row[5] - row[7] }[4-7]
+    paddd       m14, m14
+    paddd       m11, m11
+    psubd       m0,  m13
+    psubd       m1,  m12
+    psubd       m2,  m13
+    psubd       m3,  m12
+    paddd       m4,  m14
+    paddd       m5,  m11
+    paddd       m6,  m13
+    paddd       m7,  m12
+
+    pmaddwd     m10, m8, [w1_plus_w5]
+    pmaddwd     m11, m9, [w1_plus_w5]
+    pmaddwd     m12, m8, [w5_plus_w7]
+    pmaddwd     m13, m9, [w5_plus_w7]
+    pslld       m10, 2
+    pslld       m11, 2
+    pslld       m12,  2
+    pslld       m13,  2
+    psubd       m2,  m10           ; b1[0-3]
+    psubd       m3,  m11           ; b1[4-7]
+    paddd       m0,  m12            ; b0[0-3]
+    paddd       m1,  m13            ; b0[4-7]
+    pmaddwd     m12, m8, [w7_plus_w3]
+    pmaddwd     m13, m9, [w7_plus_w3]
+    pmaddwd     m8, [w3_min_w1]
+    pmaddwd     m9, [w3_min_w1]
+    pslld       m12, 2
+    pslld       m13, 2
+    pslld       m8,  2
+    pslld       m9,  2
+    paddd       m4,  m12           ; b2[0-3]
+    paddd       m5,  m13           ; b2[4-7]
+    paddd       m6,  m8            ; b3[0-3]
+    paddd       m7,  m9            ; b3[4-7]
+
+    ; row[0] = (a0 + b0) >> 15;
+    ; row[7] = (a0 - b0) >> 15;
+    ; row[1] = (a1 + b1) >> 15;
+    ; row[6] = (a1 - b1) >> 15;
+    ; row[2] = (a2 + b2) >> 15;
+    ; row[5] = (a2 - b2) >> 15;
+    ; row[3] = (a3 + b3) >> 15;
+    ; row[4] = (a3 - b3) >> 15;
+    mova        m8, [r2+ 0]        ; a0[0-3]
+    mova        m9, [r2+16]        ; a0[4-7]
+    SUMSUB_SHPK m8,  m9,  m10, m11, m0,  m1,  %2
+    mova        m0, [r2+32]        ; a1[0-3]
+    mova        m1, [r2+48]        ; a1[4-7]
+    SUMSUB_SHPK m0,  m1,  m9,  m11, m2,  m3,  %2
+    mova        m1, [r2+64]        ; a2[0-3]
+    mova        m2, [r2+80]        ; a2[4-7]
+    SUMSUB_SHPK m1,  m2,  m11, m3,  m4,  m5,  %2
+    mova        m2, [r2+96]        ; a3[0-3]
+    mova        m3, [r2+112]       ; a3[4-7]
+    SUMSUB_SHPK m2,  m3,  m4,  m5,  m6,  m7,  %2
+%endmacro
+
+; void prores_idct_put_10_<opt>(uint8_t *pixels, int stride,
+;                               DCTELEM *block, const int16_t *qmat);
+%macro idct_put_fn 2
+cglobal prores_idct_put_10_%1, 4, 4, %2
+    movsxd      r1,  r1d
+    pxor        m15, m15           ; zero
+
+    ; for (i = 0; i < 8; i++)
+    ;     idctRowCondDC(block + i*8);
+    mova        m10,[r2+ 0]        ; { row[0] }[0-7]
+    mova        m8, [r2+32]        ; { row[2] }[0-7]
+    mova        m13,[r2+64]        ; { row[4] }[0-7]
+    mova        m12,[r2+96]        ; { row[6] }[0-7]
+
+    pmullw      m10,[r3+ 0]
+    pmullw      m8, [r3+32]
+    pmullw      m13,[r3+64]
+    pmullw      m12,[r3+96]
+
+    IDCT_1D     row, 17,  %1
+
+    ; transpose for second part of IDCT
+    TRANSPOSE8x8W 8, 0, 1, 2, 4, 11, 9, 10, 3
+    mova   [r2+ 16], m0
+    mova   [r2+ 48], m2
+    mova   [r2+ 80], m11
+    mova   [r2+112], m10
+    SWAP         8,  10
+    SWAP         1,   8
+    SWAP         4,  13
+    SWAP         9,  12
+
+    ; for (i = 0; i < 8; i++)
+    ;     idctSparseColAdd(dest + i, line_size, block + i);
+    IDCT_1D     col, 20,  %1
+
+    ; clip/store
+    mova        m6, [pw_512]
+    mova        m3, [pw_4]
+    mova        m5, [pw_1019]
+    paddw       m8,  m6
+    paddw       m0,  m6
+    paddw       m1,  m6
+    paddw       m2,  m6
+    paddw       m4,  m6
+    paddw       m11, m6
+    paddw       m9,  m6
+    paddw       m10, m6
+    pmaxsw      m8,  m3
+    pmaxsw      m0,  m3
+    pmaxsw      m1,  m3
+    pmaxsw      m2,  m3
+    pmaxsw      m4,  m3
+    pmaxsw      m11, m3
+    pmaxsw      m9,  m3
+    pmaxsw      m10, m3
+    pminsw      m8,  m5
+    pminsw      m0,  m5
+    pminsw      m1,  m5
+    pminsw      m2,  m5
+    pminsw      m4,  m5
+    pminsw      m11, m5
+    pminsw      m9,  m5
+    pminsw      m10, m5
+
+    lea         r2, [r1*3]
+    mova  [r0     ], m8
+    mova  [r0+r1  ], m0
+    mova  [r0+r1*2], m1
+    mova  [r0+r2  ], m2
+    lea         r0, [r0+r1*4]
+    mova  [r0     ], m4
+    mova  [r0+r1  ], m11
+    mova  [r0+r1*2], m9
+    mova  [r0+r2  ], m10
+    RET
+%endmacro
+
+%macro signextend_sse2 3 ; dstlow, dsthigh, tmp
+    pxor        %3,  %3
+    pcmpgtw     %3,  %1
+    mova        %2,  %1
+    punpcklwd   %1,  %3
+    punpckhwd   %2,  %3
+%endmacro
+
+%macro signextend_sse4 2-3 ; dstlow, dsthigh
+    movhlps     %2,  %1
+    pmovsxwd    %1,  %1
+    pmovsxwd    %2,  %2
+%endmacro
+
+INIT_XMM
+%define SIGNEXTEND signextend_sse2
+idct_put_fn sse2, 16
+INIT_XMM
+%define SIGNEXTEND signextend_sse4
+idct_put_fn sse4, 16
+INIT_AVX
+idct_put_fn avx,  16
+
+%endif