Merge remote-tracking branch 'qatar/master'

* qatar/master:
  westwood_vqa: fix SND0 chunk handling
  westwood_vqa: set video stream duration
  raw: forward avpicture_fill() error code in raw_decode().
  build: Do not explicitly add the doc directory to the OBJDIRS list.
  dv: Split off DV video decoder into its own file.
  build: fix RALF decoder standalone compilation, which depends on Golomb code
  configure: Drop stray duplicate entry for --disable-fft from help output.

Conflicts:
	libavcodec/dv.c
	libavcodec/rawdec.c
	libavformat/westwood_vqa.c

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

8 files changed, 448 insertions, 401 deletions

diff --git a/configure b/configure
index 0afbf99fa1..ab242e8982 100755
--- a/configure
+++ b/configure
@@ -118,7 +118,6 @@ Component options:
   --enable-x11grab         enable X11 grabbing [no]
   --disable-network        disable network support [no]
   --disable-dct            disable DCT code
-  --disable-fft            disable FFT code
   --disable-mdct           disable MDCT code
   --disable-rdft           disable RDFT code
   --disable-fft            disable FFT code
@@ -1482,6 +1481,7 @@ png_encoder_select="zlib"
 qcelp_decoder_select="lsp"
 qdm2_decoder_select="mdct rdft mpegaudiodsp"
 ra_144_encoder_select="lpc"
+ralf_decoder_select="golomb"
 rv10_decoder_select="h263_decoder"
 rv10_encoder_select="h263_encoder"
 rv20_decoder_select="h263_decoder"
diff --git a/doc/Makefile b/doc/Makefile
index ed1425fc3e..07752c5ec7 100644
--- a/doc/Makefile
+++ b/doc/Makefile
@@ -52,7 +52,6 @@ doc/%.1: doc/%.pod $(GENTEXI)
 	$(M)pod2man --section=1 --center=" " --release=" " $< > $@
 
 $(DOCS): | doc
-OBJDIRS += doc
 
 install-progs-$(CONFIG_DOC): install-man
 
diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 6dfc9d766b..d554df5a04 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -138,7 +138,7 @@ OBJS-$(CONFIG_DVBSUB_DECODER)          += dvbsubdec.o
 OBJS-$(CONFIG_DVBSUB_ENCODER)          += dvbsub.o
 OBJS-$(CONFIG_DVDSUB_DECODER)          += dvdsubdec.o
 OBJS-$(CONFIG_DVDSUB_ENCODER)          += dvdsubenc.o
-OBJS-$(CONFIG_DVVIDEO_DECODER)         += dv.o dvdata.o
+OBJS-$(CONFIG_DVVIDEO_DECODER)         += dvdec.o dv.o dvdata.o
 OBJS-$(CONFIG_DVVIDEO_ENCODER)         += dv.o dvdata.o
 OBJS-$(CONFIG_DXA_DECODER)             += dxa.o
 OBJS-$(CONFIG_DXTORY_DECODER)          += dxtory.o
diff --git a/libavcodec/dv.c b/libavcodec/dv.c
index 309846a598..805ebe31ee 100644
--- a/libavcodec/dv.c
+++ b/libavcodec/dv.c
@@ -49,37 +49,8 @@
 #include "dvquant.h"
 #include "dv_tablegen.h"
 
-//#undef NDEBUG
-//#include <assert.h>
-
-typedef struct DVVideoContext {
-    const DVprofile *sys;
-    AVFrame          picture;
-    AVCodecContext  *avctx;
-    uint8_t         *buf;
-
-    uint8_t  dv_zigzag[2][64];
-
-    void (*get_pixels)(DCTELEM *block, const uint8_t *pixels, int line_size);
-    void (*fdct[2])(DCTELEM *block);
-    void (*idct_put[2])(uint8_t *dest, int line_size, DCTELEM *block);
-    me_cmp_func ildct_cmp;
-} DVVideoContext;
-
-#define TEX_VLC_BITS 9
-
 /* XXX: also include quantization */
-static RL_VLC_ELEM dv_rl_vlc[1184];
-
-static inline int dv_work_pool_size(const DVprofile *d)
-{
-    int size = d->n_difchan*d->difseg_size*27;
-    if (DV_PROFILE_IS_1080i50(d))
-        size -= 3*27;
-    if (DV_PROFILE_IS_720p50(d))
-        size -= 4*27;
-    return size;
-}
+RL_VLC_ELEM ff_dv_rl_vlc[1184];
 
 static inline void dv_calc_mb_coordinates(const DVprofile *d, int chan, int seq, int slot,
                                           uint16_t *tbl)
@@ -196,7 +167,7 @@ static inline void dv_calc_mb_coordinates(const DVprofile *d, int chan, int seq,
     }
 }
 
-static int dv_init_dynamic_tables(const DVprofile *d)
+int ff_dv_init_dynamic_tables(const DVprofile *d)
 {
     int j,i,c,s,p;
     uint32_t *factor1, *factor2;
@@ -257,7 +228,7 @@ static int dv_init_dynamic_tables(const DVprofile *d)
     return 0;
 }
 
-static av_cold int dvvideo_init(AVCodecContext *avctx)
+av_cold int ff_dvvideo_init(AVCodecContext *avctx)
 {
     DVVideoContext *s = avctx->priv_data;
     DSPContext dsp;
@@ -310,9 +281,9 @@ static av_cold int dvvideo_init(AVCodecContext *avctx)
                 run   = new_dv_vlc_run  [code] + 1;
                 level = new_dv_vlc_level[code];
             }
-            dv_rl_vlc[i].len   = len;
-            dv_rl_vlc[i].level = level;
-            dv_rl_vlc[i].run   = run;
+            ff_dv_rl_vlc[i].len   = len;
+            ff_dv_rl_vlc[i].level = level;
+            ff_dv_rl_vlc[i].run   = run;
         }
         ff_free_vlc(&dv_vlc);
 
@@ -357,19 +328,9 @@ static av_cold int dvvideo_init_encoder(AVCodecContext *avctx)
         return -1;
     }
 
-    return dvvideo_init(avctx);
+    return ff_dvvideo_init(avctx);
 }
 
-typedef struct BlockInfo {
-    const uint32_t *factor_table;
-    const uint8_t *scan_table;
-    uint8_t pos; /* position in block */
-    void (*idct_put)(uint8_t *dest, int line_size, DCTELEM *block);
-    uint8_t partial_bit_count;
-    uint32_t partial_bit_buffer;
-    int shift_offset;
-} BlockInfo;
-
 /* bit budget for AC only in 5 MBs */
 static const int vs_total_ac_bits = (100 * 4 + 68*2) * 5;
 static const int mb_area_start[5] = { 1, 6, 21, 43, 64 };
@@ -379,268 +340,6 @@ static inline int put_bits_left(PutBitContext* s)
     return (s->buf_end - s->buf) * 8 - put_bits_count(s);
 }
 
-/* decode AC coefficients */
-static void dv_decode_ac(GetBitContext *gb, BlockInfo *mb, DCTELEM *block)
-{
-    int last_index = gb->size_in_bits;
-    const uint8_t  *scan_table   = mb->scan_table;
-    const uint32_t *factor_table = mb->factor_table;
-    int pos               = mb->pos;
-    int partial_bit_count = mb->partial_bit_count;
-    int level, run, vlc_len, index;
-
-    OPEN_READER(re, gb);
-    UPDATE_CACHE(re, gb);
-
-    /* if we must parse a partial VLC, we do it here */
-    if (partial_bit_count > 0) {
-        re_cache = re_cache >> partial_bit_count | mb->partial_bit_buffer;
-        re_index -= partial_bit_count;
-        mb->partial_bit_count = 0;
-    }
-
-    /* get the AC coefficients until last_index is reached */
-    for (;;) {
-        av_dlog(NULL, "%2d: bits=%04x index=%d\n", pos, SHOW_UBITS(re, gb, 16),
-                re_index);
-        /* our own optimized GET_RL_VLC */
-        index   = NEG_USR32(re_cache, TEX_VLC_BITS);
-        vlc_len = dv_rl_vlc[index].len;
-        if (vlc_len < 0) {
-            index = NEG_USR32((unsigned)re_cache << TEX_VLC_BITS, -vlc_len) + dv_rl_vlc[index].level;
-            vlc_len = TEX_VLC_BITS - vlc_len;
-        }
-        level = dv_rl_vlc[index].level;
-        run   = dv_rl_vlc[index].run;
-
-        /* gotta check if we're still within gb boundaries */
-        if (re_index + vlc_len > last_index) {
-            /* should be < 16 bits otherwise a codeword could have been parsed */
-            mb->partial_bit_count = last_index - re_index;
-            mb->partial_bit_buffer = re_cache & ~(-1u >> mb->partial_bit_count);
-            re_index = last_index;
-            break;
-        }
-        re_index += vlc_len;
-
-        av_dlog(NULL, "run=%d level=%d\n", run, level);
-        pos += run;
-        if (pos >= 64)
-            break;
-
-        level = (level * factor_table[pos] + (1 << (dv_iweight_bits - 1))) >> dv_iweight_bits;
-        block[scan_table[pos]] = level;
-
-        UPDATE_CACHE(re, gb);
-    }
-    CLOSE_READER(re, gb);
-    mb->pos = pos;
-}
-
-static inline void bit_copy(PutBitContext *pb, GetBitContext *gb)
-{
-    int bits_left = get_bits_left(gb);
-    while (bits_left >= MIN_CACHE_BITS) {
-        put_bits(pb, MIN_CACHE_BITS, get_bits(gb, MIN_CACHE_BITS));
-        bits_left -= MIN_CACHE_BITS;
-    }
-    if (bits_left > 0) {
-        put_bits(pb, bits_left, get_bits(gb, bits_left));
-    }
-}
-
-static inline void dv_calculate_mb_xy(DVVideoContext *s, DVwork_chunk *work_chunk, int m, int *mb_x, int *mb_y)
-{
-     *mb_x = work_chunk->mb_coordinates[m] & 0xff;
-     *mb_y = work_chunk->mb_coordinates[m] >> 8;
-
-     /* We work with 720p frames split in half. The odd half-frame (chan==2,3) is displaced :-( */
-     if (s->sys->height == 720 && !(s->buf[1]&0x0C)) {
-         *mb_y -= (*mb_y>17)?18:-72; /* shifting the Y coordinate down by 72/2 macro blocks */
-     }
-}
-
-/* mb_x and mb_y are in units of 8 pixels */
-static int dv_decode_video_segment(AVCodecContext *avctx, void *arg)
-{
-    DVVideoContext *s = avctx->priv_data;
-    DVwork_chunk *work_chunk = arg;
-    int quant, dc, dct_mode, class1, j;
-    int mb_index, mb_x, mb_y, last_index;
-    int y_stride, linesize;
-    DCTELEM *block, *block1;
-    int c_offset;
-    uint8_t *y_ptr;
-    const uint8_t *buf_ptr;
-    PutBitContext pb, vs_pb;
-    GetBitContext gb;
-    BlockInfo mb_data[5 * DV_MAX_BPM], *mb, *mb1;
-    LOCAL_ALIGNED_16(DCTELEM, sblock, [5*DV_MAX_BPM], [64]);
-    LOCAL_ALIGNED_16(uint8_t, mb_bit_buffer, [  80 + FF_INPUT_BUFFER_PADDING_SIZE]); /* allow some slack */
-    LOCAL_ALIGNED_16(uint8_t, vs_bit_buffer, [5*80 + FF_INPUT_BUFFER_PADDING_SIZE]); /* allow some slack */
-    const int log2_blocksize = 3-s->avctx->lowres;
-    int is_field_mode[5];
-
-    assert((((int)mb_bit_buffer) & 7) == 0);
-    assert((((int)vs_bit_buffer) & 7) == 0);
-
-    memset(sblock, 0, 5*DV_MAX_BPM*sizeof(*sblock));
-
-    /* pass 1: read DC and AC coefficients in blocks */
-    buf_ptr = &s->buf[work_chunk->buf_offset*80];
-    block1  = &sblock[0][0];
-    mb1     = mb_data;
-    init_put_bits(&vs_pb, vs_bit_buffer, 5 * 80);
-    for (mb_index = 0; mb_index < 5; mb_index++, mb1 += s->sys->bpm, block1 += s->sys->bpm * 64) {
-        /* skip header */
-        quant = buf_ptr[3] & 0x0f;
-        buf_ptr += 4;
-        init_put_bits(&pb, mb_bit_buffer, 80);
-        mb    = mb1;
-        block = block1;
-        is_field_mode[mb_index] = 0;
-        for (j = 0; j < s->sys->bpm; j++) {
-            last_index = s->sys->block_sizes[j];
-            init_get_bits(&gb, buf_ptr, last_index);
-
-            /* get the DC */
-            dc       = get_sbits(&gb, 9);
-            dct_mode = get_bits1(&gb);
-            class1   = get_bits(&gb, 2);
-            if (DV_PROFILE_IS_HD(s->sys)) {
-                mb->idct_put     = s->idct_put[0];
-                mb->scan_table   = s->dv_zigzag[0];
-                mb->factor_table = &s->sys->idct_factor[(j >= 4)*4*16*64 + class1*16*64 + quant*64];
-                is_field_mode[mb_index] |= !j && dct_mode;
-            } else {
-                mb->idct_put     = s->idct_put[dct_mode && log2_blocksize == 3];
-                mb->scan_table   = s->dv_zigzag[dct_mode];
-                mb->factor_table = &s->sys->idct_factor[(class1 == 3)*2*22*64 + dct_mode*22*64 +
-                                                        (quant + dv_quant_offset[class1])*64];
-            }
-            dc = dc << 2;
-            /* convert to unsigned because 128 is not added in the
-               standard IDCT */
-            dc += 1024;
-            block[0] = dc;
-            buf_ptr += last_index >> 3;
-            mb->pos               = 0;
-            mb->partial_bit_count = 0;
-
-            av_dlog(avctx, "MB block: %d, %d ", mb_index, j);
-            dv_decode_ac(&gb, mb, block);
-
-            /* write the remaining bits in a new buffer only if the
-               block is finished */
-            if (mb->pos >= 64)
-                bit_copy(&pb, &gb);
-
-            block += 64;
-            mb++;
-        }
-
-        /* pass 2: we can do it just after */
-        av_dlog(avctx, "***pass 2 size=%d MB#=%d\n", put_bits_count(&pb), mb_index);
-        block = block1;
-        mb    = mb1;
-        init_get_bits(&gb, mb_bit_buffer, put_bits_count(&pb));
-        put_bits32(&pb, 0); // padding must be zeroed
-        flush_put_bits(&pb);
-        for (j = 0; j < s->sys->bpm; j++, block += 64, mb++) {
-            if (mb->pos < 64 && get_bits_left(&gb) > 0) {
-                dv_decode_ac(&gb, mb, block);
-                /* if still not finished, no need to parse other blocks */
-                if (mb->pos < 64)
-                    break;
-            }
-        }
-        /* all blocks are finished, so the extra bytes can be used at
-           the video segment level */
-        if (j >= s->sys->bpm)
-            bit_copy(&vs_pb, &gb);
-    }
-
-    /* we need a pass over the whole video segment */
-    av_dlog(avctx, "***pass 3 size=%d\n", put_bits_count(&vs_pb));
-    block = &sblock[0][0];
-    mb    = mb_data;
-    init_get_bits(&gb, vs_bit_buffer, put_bits_count(&vs_pb));
-    put_bits32(&vs_pb, 0); // padding must be zeroed
-    flush_put_bits(&vs_pb);
-    for (mb_index = 0; mb_index < 5; mb_index++) {
-        for (j = 0; j < s->sys->bpm; j++) {
-            if (mb->pos < 64) {
-                av_dlog(avctx, "start %d:%d\n", mb_index, j);
-                dv_decode_ac(&gb, mb, block);
-            }
-            if (mb->pos >= 64 && mb->pos < 127)
-                av_log(avctx, AV_LOG_ERROR, "AC EOB marker is absent pos=%d\n", mb->pos);
-            block += 64;
-            mb++;
-        }
-    }
-
-    /* compute idct and place blocks */
-    block = &sblock[0][0];
-    mb    = mb_data;
-    for (mb_index = 0; mb_index < 5; mb_index++) {
-        dv_calculate_mb_xy(s, work_chunk, mb_index, &mb_x, &mb_y);
-
-        /* idct_put'ting luminance */
-        if ((s->sys->pix_fmt == PIX_FMT_YUV420P) ||
-            (s->sys->pix_fmt == PIX_FMT_YUV411P && mb_x >= (704 / 8)) ||
-            (s->sys->height >= 720 && mb_y != 134)) {
-            y_stride = (s->picture.linesize[0] << ((!is_field_mode[mb_index]) * log2_blocksize));
-        } else {
-            y_stride = (2 << log2_blocksize);
-        }
-        y_ptr = s->picture.data[0] + ((mb_y * s->picture.linesize[0] + mb_x) << log2_blocksize);
-        linesize = s->picture.linesize[0] << is_field_mode[mb_index];
-        mb[0]    .idct_put(y_ptr                                   , linesize, block + 0*64);
-        if (s->sys->video_stype == 4) { /* SD 422 */
-            mb[2].idct_put(y_ptr + (1 << log2_blocksize)           , linesize, block + 2*64);
-        } else {
-            mb[1].idct_put(y_ptr + (1 << log2_blocksize)           , linesize, block + 1*64);
-            mb[2].idct_put(y_ptr                         + y_stride, linesize, block + 2*64);
-            mb[3].idct_put(y_ptr + (1 << log2_blocksize) + y_stride, linesize, block + 3*64);
-        }
-        mb += 4;
-        block += 4*64;
-
-        /* idct_put'ting chrominance */
-        c_offset = (((mb_y >>  (s->sys->pix_fmt == PIX_FMT_YUV420P)) * s->picture.linesize[1] +
-                     (mb_x >> ((s->sys->pix_fmt == PIX_FMT_YUV411P) ? 2 : 1))) << log2_blocksize);
-        for (j = 2; j; j--) {
-            uint8_t *c_ptr = s->picture.data[j] + c_offset;
-            if (s->sys->pix_fmt == PIX_FMT_YUV411P && mb_x >= (704 / 8)) {
-                  uint64_t aligned_pixels[64/8];
-                  uint8_t *pixels = (uint8_t*)aligned_pixels;
-                  uint8_t *c_ptr1, *ptr1;
-                  int x, y;
-                  mb->idct_put(pixels, 8, block);
-                  for (y = 0; y < (1 << log2_blocksize); y++, c_ptr += s->picture.linesize[j], pixels += 8) {
-                      ptr1   = pixels + (1 << (log2_blocksize - 1));
-                      c_ptr1 = c_ptr + (s->picture.linesize[j] << log2_blocksize);
-                      for (x = 0; x < (1 << (log2_blocksize - 1)); x++) {
-                          c_ptr[x]  = pixels[x];
-                          c_ptr1[x] = ptr1[x];
-                      }
-                  }
-                  block += 64; mb++;
-            } else {
-                  y_stride = (mb_y == 134) ? (1 << log2_blocksize) :
-                                             s->picture.linesize[j] << ((!is_field_mode[mb_index]) * log2_blocksize);
-                  linesize = s->picture.linesize[j] << is_field_mode[mb_index];
-                  (mb++)->    idct_put(c_ptr           , linesize, block); block += 64;
-                  if (s->sys->bpm == 8) {
-                      (mb++)->idct_put(c_ptr + y_stride, linesize, block); block += 64;
-                  }
-            }
-        }
-    }
-    return 0;
-}
-
 #if CONFIG_SMALL
 /* Converts run and level (where level != 0) pair into VLC, returning bit size */
 static av_always_inline int dv_rl2vlc(int run, int level, int sign, uint32_t* vlc)
@@ -1060,65 +759,6 @@ static int dv_encode_video_segment(AVCodecContext *avctx, void *arg)
     return 0;
 }
 
-#if CONFIG_DVVIDEO_DECODER
-/* NOTE: exactly one frame must be given (120000 bytes for NTSC,
-   144000 bytes for PAL - or twice those for 50Mbps) */
-static int dvvideo_decode_frame(AVCodecContext *avctx,
-                                 void *data, int *data_size,
-                                 AVPacket *avpkt)
-{
-    const uint8_t *buf = avpkt->data;
-    int buf_size = avpkt->size;
-    DVVideoContext *s = avctx->priv_data;
-    const uint8_t* vsc_pack;
-    int apt, is16_9;
-
-    s->sys = avpriv_dv_frame_profile2(avctx, s->sys, buf, buf_size);
-    if (!s->sys || buf_size < s->sys->frame_size || dv_init_dynamic_tables(s->sys)) {
-        av_log(avctx, AV_LOG_ERROR, "could not find dv frame profile\n");
-        return -1; /* NOTE: we only accept several full frames */
-    }
-
-    if (s->picture.data[0])
-        avctx->release_buffer(avctx, &s->picture);
-
-    avcodec_get_frame_defaults(&s->picture);
-    s->picture.reference = 0;
-    s->picture.key_frame = 1;
-    s->picture.pict_type = AV_PICTURE_TYPE_I;
-    avctx->pix_fmt   = s->sys->pix_fmt;
-    avctx->time_base = s->sys->time_base;
-    avcodec_set_dimensions(avctx, s->sys->width, s->sys->height);
-    if (avctx->get_buffer(avctx, &s->picture) < 0) {
-        av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
-        return -1;
-    }
-    s->picture.interlaced_frame = 1;
-    s->picture.top_field_first  = 0;
-
-    s->buf = buf;
-    avctx->execute(avctx, dv_decode_video_segment, s->sys->work_chunks, NULL,
-                   dv_work_pool_size(s->sys), sizeof(DVwork_chunk));
-
-    emms_c();
-
-    /* return image */
-    *data_size = sizeof(AVFrame);
-    *(AVFrame*)data = s->picture;
-
-    /* Determine the codec's sample_aspect ratio from the packet */
-    vsc_pack = buf + 80*5 + 48 + 5;
-    if ( *vsc_pack == dv_video_control ) {
-        apt = buf[4] & 0x07;
-        is16_9 = (vsc_pack[2] & 0x07) == 0x02 || (!apt && (vsc_pack[2] & 0x07) == 0x07);
-        avctx->sample_aspect_ratio = s->sys->sar[is16_9];
-    }
-
-    return s->sys->frame_size;
-}
-#endif /* CONFIG_DVVIDEO_DECODER */
-
-
 static inline int dv_write_pack(enum dv_pack_type pack_id, DVVideoContext *c,
                                 uint8_t* buf)
 {
@@ -1284,7 +924,7 @@ static int dvvideo_encode_frame(AVCodecContext *c, AVPacket *pkt,
     int ret;
 
     s->sys = avpriv_dv_codec_profile(c);
-    if (!s->sys || dv_init_dynamic_tables(s->sys))
+    if (!s->sys || ff_dv_init_dynamic_tables(s->sys))
         return -1;
     if ((ret = ff_alloc_packet2(c, pkt, s->sys->frame_size)) < 0)
         return ret;
@@ -1307,20 +947,7 @@ static int dvvideo_encode_frame(AVCodecContext *c, AVPacket *pkt,
 
     return 0;
 }
-#endif
-
-static int dvvideo_close(AVCodecContext *c)
-{
-    DVVideoContext *s = c->priv_data;
-
-    if (s->picture.data[0])
-        c->release_buffer(c, &s->picture);
 
-    return 0;
-}
-
-
-#if CONFIG_DVVIDEO_ENCODER
 AVCodec ff_dvvideo_encoder = {
     .name           = "dvvideo",
     .type           = AVMEDIA_TYPE_VIDEO,
@@ -1333,18 +960,3 @@ AVCodec ff_dvvideo_encoder = {
     .long_name = NULL_IF_CONFIG_SMALL("DV (Digital Video)"),
 };
 #endif // CONFIG_DVVIDEO_ENCODER
-
-#if CONFIG_DVVIDEO_DECODER
-AVCodec ff_dvvideo_decoder = {
-    .name           = "dvvideo",
-    .type           = AVMEDIA_TYPE_VIDEO,
-    .id             = CODEC_ID_DVVIDEO,
-    .priv_data_size = sizeof(DVVideoContext),
-    .init           = dvvideo_init,
-    .close          = dvvideo_close,
-    .decode         = dvvideo_decode_frame,
-    .capabilities   = CODEC_CAP_DR1 | CODEC_CAP_SLICE_THREADS,
-    .max_lowres = 3,
-    .long_name = NULL_IF_CONFIG_SMALL("DV (Digital Video)"),
-};
-#endif
diff --git a/libavcodec/dv_tablegen.h b/libavcodec/dv_tablegen.h
index 678be73940..9b3e406f6a 100644
--- a/libavcodec/dv_tablegen.h
+++ b/libavcodec/dv_tablegen.h
@@ -24,6 +24,8 @@
 #define AVCODEC_DV_TABLEGEN_H
 
 #include <stdint.h>
+
+#include "libavutil/attributes.h"
 #include "dv_vlc_data.h"
 
 #if CONFIG_SMALL
@@ -46,7 +48,7 @@ typedef struct dv_vlc_pair {
 #else
 static struct dv_vlc_pair dv_vlc_map[DV_VLC_MAP_RUN_SIZE][DV_VLC_MAP_LEV_SIZE];
 
-static void dv_vlc_map_tableinit(void)
+static void av_unused dv_vlc_map_tableinit(void)
 {
     int i, j;
     for (i = 0; i < NB_DV_VLC - 1; i++) {
diff --git a/libavcodec/dvdata.h b/libavcodec/dvdata.h
index 8c1be96183..e3ce98ad12 100644
--- a/libavcodec/dvdata.h
+++ b/libavcodec/dvdata.h
@@ -29,6 +29,8 @@
 
 #include "libavutil/rational.h"
 #include "avcodec.h"
+#include "dsputil.h"
+#include "get_bits.h"
 
 typedef struct DVwork_chunk {
     uint16_t  buf_offset;
@@ -65,6 +67,20 @@ typedef struct DVprofile {
     const uint8_t  (*audio_shuffle)[9];     /* PCM shuffling table */
 } DVprofile;
 
+typedef struct DVVideoContext {
+    const DVprofile *sys;
+    AVFrame          picture;
+    AVCodecContext  *avctx;
+    uint8_t         *buf;
+
+    uint8_t  dv_zigzag[2][64];
+
+    void (*get_pixels)(DCTELEM *block, const uint8_t *pixels, int line_size);
+    void (*fdct[2])(DCTELEM *block);
+    void (*idct_put[2])(uint8_t *dest, int line_size, DCTELEM *block);
+    me_cmp_func ildct_cmp;
+} DVVideoContext;
+
 enum dv_section_type {
      dv_sect_header  = 0x1f,
      dv_sect_subcode = 0x3f,
@@ -106,10 +122,38 @@ enum dv_pack_type {
  */
 #define DV_MAX_BPM 8
 
+#define TEX_VLC_BITS 9
+
+extern RL_VLC_ELEM ff_dv_rl_vlc[1184];
+
 const DVprofile* avpriv_dv_frame_profile(const DVprofile *sys,
                                   const uint8_t* frame, unsigned buf_size);
 const DVprofile* avpriv_dv_frame_profile2(AVCodecContext* codec, const DVprofile *sys,
                                   const uint8_t* frame, unsigned buf_size);
 const DVprofile* avpriv_dv_codec_profile(AVCodecContext* codec);
 
+int ff_dv_init_dynamic_tables(const DVprofile *d);
+int ff_dvvideo_init(AVCodecContext *avctx);
+
+static inline int dv_work_pool_size(const DVprofile *d)
+{
+    int size = d->n_difchan*d->difseg_size*27;
+    if (DV_PROFILE_IS_1080i50(d))
+        size -= 3*27;
+    if (DV_PROFILE_IS_720p50(d))
+        size -= 4*27;
+    return size;
+}
+
+static inline void dv_calculate_mb_xy(DVVideoContext *s, DVwork_chunk *work_chunk, int m, int *mb_x, int *mb_y)
+{
+     *mb_x = work_chunk->mb_coordinates[m] & 0xff;
+     *mb_y = work_chunk->mb_coordinates[m] >> 8;
+
+     /* We work with 720p frames split in half. The odd half-frame (chan==2,3) is displaced :-( */
+     if (s->sys->height == 720 && !(s->buf[1]&0x0C)) {
+         *mb_y -= (*mb_y>17)?18:-72; /* shifting the Y coordinate down by 72/2 macro blocks */
+     }
+}
+
 #endif /* AVCODEC_DVDATA_H */
diff --git a/libavcodec/dvdec.c b/libavcodec/dvdec.c
new file mode 100644
index 0000000000..8e0ee65bef
--- /dev/null
+++ b/libavcodec/dvdec.c
@@ -0,0 +1,387 @@
+/*
+ * DV decoder
+ * Copyright (c) 2002 Fabrice Bellard
+ * Copyright (c) 2004 Roman Shaposhnik
+ *
+ * 50 Mbps (DVCPRO50) support
+ * Copyright (c) 2006 Daniel Maas <dmaas@maasdigital.com>
+ *
+ * 100 Mbps (DVCPRO HD) support
+ * Initial code by Daniel Maas <dmaas@maasdigital.com> (funded by BBC R&D)
+ * Final code by Roman Shaposhnik
+ *
+ * Many thanks to Dan Dennedy <dan@dennedy.org> for providing wealth
+ * of DV technical info.
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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.
+ *
+ * FFmpeg 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 FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/**
+ * @file
+ * DV decoder
+ */
+
+#include "libavutil/pixdesc.h"
+#include "avcodec.h"
+#include "dsputil.h"
+#include "get_bits.h"
+#include "put_bits.h"
+#include "simple_idct.h"
+#include "dvdata.h"
+#include "dvquant.h"
+#include "dv_tablegen.h"
+
+typedef struct BlockInfo {
+    const uint32_t *factor_table;
+    const uint8_t *scan_table;
+    uint8_t pos; /* position in block */
+    void (*idct_put)(uint8_t *dest, int line_size, DCTELEM *block);
+    uint8_t partial_bit_count;
+    uint32_t partial_bit_buffer;
+    int shift_offset;
+} BlockInfo;
+
+/* decode AC coefficients */
+static void dv_decode_ac(GetBitContext *gb, BlockInfo *mb, DCTELEM *block)
+{
+    int last_index = gb->size_in_bits;
+    const uint8_t  *scan_table   = mb->scan_table;
+    const uint32_t *factor_table = mb->factor_table;
+    int pos               = mb->pos;
+    int partial_bit_count = mb->partial_bit_count;
+    int level, run, vlc_len, index;
+
+    OPEN_READER(re, gb);
+    UPDATE_CACHE(re, gb);
+
+    /* if we must parse a partial VLC, we do it here */
+    if (partial_bit_count > 0) {
+        re_cache = re_cache >> partial_bit_count | mb->partial_bit_buffer;
+        re_index -= partial_bit_count;
+        mb->partial_bit_count = 0;
+    }
+
+    /* get the AC coefficients until last_index is reached */
+    for (;;) {
+        av_dlog(NULL, "%2d: bits=%04x index=%d\n", pos, SHOW_UBITS(re, gb, 16),
+                re_index);
+        /* our own optimized GET_RL_VLC */
+        index   = NEG_USR32(re_cache, TEX_VLC_BITS);
+        vlc_len = ff_dv_rl_vlc[index].len;
+        if (vlc_len < 0) {
+            index = NEG_USR32((unsigned)re_cache << TEX_VLC_BITS, -vlc_len) +
+                    ff_dv_rl_vlc[index].level;
+            vlc_len = TEX_VLC_BITS - vlc_len;
+        }
+        level = ff_dv_rl_vlc[index].level;
+        run   = ff_dv_rl_vlc[index].run;
+
+        /* gotta check if we're still within gb boundaries */
+        if (re_index + vlc_len > last_index) {
+            /* should be < 16 bits otherwise a codeword could have been parsed */
+            mb->partial_bit_count = last_index - re_index;
+            mb->partial_bit_buffer = re_cache & ~(-1u >> mb->partial_bit_count);
+            re_index = last_index;
+            break;
+        }
+        re_index += vlc_len;
+
+        av_dlog(NULL, "run=%d level=%d\n", run, level);
+        pos += run;
+        if (pos >= 64)
+            break;
+
+        level = (level * factor_table[pos] + (1 << (dv_iweight_bits - 1))) >> dv_iweight_bits;
+        block[scan_table[pos]] = level;
+
+        UPDATE_CACHE(re, gb);
+    }
+    CLOSE_READER(re, gb);
+    mb->pos = pos;
+}
+
+static inline void bit_copy(PutBitContext *pb, GetBitContext *gb)
+{
+    int bits_left = get_bits_left(gb);
+    while (bits_left >= MIN_CACHE_BITS) {
+        put_bits(pb, MIN_CACHE_BITS, get_bits(gb, MIN_CACHE_BITS));
+        bits_left -= MIN_CACHE_BITS;
+    }
+    if (bits_left > 0) {
+        put_bits(pb, bits_left, get_bits(gb, bits_left));
+    }
+}
+
+/* mb_x and mb_y are in units of 8 pixels */
+static int dv_decode_video_segment(AVCodecContext *avctx, void *arg)
+{
+    DVVideoContext *s = avctx->priv_data;
+    DVwork_chunk *work_chunk = arg;
+    int quant, dc, dct_mode, class1, j;
+    int mb_index, mb_x, mb_y, last_index;
+    int y_stride, linesize;
+    DCTELEM *block, *block1;
+    int c_offset;
+    uint8_t *y_ptr;
+    const uint8_t *buf_ptr;
+    PutBitContext pb, vs_pb;
+    GetBitContext gb;
+    BlockInfo mb_data[5 * DV_MAX_BPM], *mb, *mb1;
+    LOCAL_ALIGNED_16(DCTELEM, sblock, [5*DV_MAX_BPM], [64]);
+    LOCAL_ALIGNED_16(uint8_t, mb_bit_buffer, [  80 + FF_INPUT_BUFFER_PADDING_SIZE]); /* allow some slack */
+    LOCAL_ALIGNED_16(uint8_t, vs_bit_buffer, [5*80 + FF_INPUT_BUFFER_PADDING_SIZE]); /* allow some slack */
+    const int log2_blocksize = 3-s->avctx->lowres;
+    int is_field_mode[5];
+
+    assert((((int)mb_bit_buffer) & 7) == 0);
+    assert((((int)vs_bit_buffer) & 7) == 0);
+
+    memset(sblock, 0, 5*DV_MAX_BPM*sizeof(*sblock));
+
+    /* pass 1: read DC and AC coefficients in blocks */
+    buf_ptr = &s->buf[work_chunk->buf_offset*80];
+    block1  = &sblock[0][0];
+    mb1     = mb_data;
+    init_put_bits(&vs_pb, vs_bit_buffer, 5 * 80);
+    for (mb_index = 0; mb_index < 5; mb_index++, mb1 += s->sys->bpm, block1 += s->sys->bpm * 64) {
+        /* skip header */
+        quant = buf_ptr[3] & 0x0f;
+        buf_ptr += 4;
+        init_put_bits(&pb, mb_bit_buffer, 80);
+        mb    = mb1;
+        block = block1;
+        is_field_mode[mb_index] = 0;
+        for (j = 0; j < s->sys->bpm; j++) {
+            last_index = s->sys->block_sizes[j];
+            init_get_bits(&gb, buf_ptr, last_index);
+
+            /* get the DC */
+            dc       = get_sbits(&gb, 9);
+            dct_mode = get_bits1(&gb);
+            class1   = get_bits(&gb, 2);
+            if (DV_PROFILE_IS_HD(s->sys)) {
+                mb->idct_put     = s->idct_put[0];
+                mb->scan_table   = s->dv_zigzag[0];
+                mb->factor_table = &s->sys->idct_factor[(j >= 4)*4*16*64 + class1*16*64 + quant*64];
+                is_field_mode[mb_index] |= !j && dct_mode;
+            } else {
+                mb->idct_put     = s->idct_put[dct_mode && log2_blocksize == 3];
+                mb->scan_table   = s->dv_zigzag[dct_mode];
+                mb->factor_table = &s->sys->idct_factor[(class1 == 3)*2*22*64 + dct_mode*22*64 +
+                                                        (quant + dv_quant_offset[class1])*64];
+            }
+            dc = dc << 2;
+            /* convert to unsigned because 128 is not added in the
+               standard IDCT */
+            dc += 1024;
+            block[0] = dc;
+            buf_ptr += last_index >> 3;
+            mb->pos               = 0;
+            mb->partial_bit_count = 0;
+
+            av_dlog(avctx, "MB block: %d, %d ", mb_index, j);
+            dv_decode_ac(&gb, mb, block);
+
+            /* write the remaining bits in a new buffer only if the
+               block is finished */
+            if (mb->pos >= 64)
+                bit_copy(&pb, &gb);
+
+            block += 64;
+            mb++;
+        }
+
+        /* pass 2: we can do it just after */
+        av_dlog(avctx, "***pass 2 size=%d MB#=%d\n", put_bits_count(&pb), mb_index);
+        block = block1;
+        mb    = mb1;
+        init_get_bits(&gb, mb_bit_buffer, put_bits_count(&pb));
+        put_bits32(&pb, 0); // padding must be zeroed
+        flush_put_bits(&pb);
+        for (j = 0; j < s->sys->bpm; j++, block += 64, mb++) {
+            if (mb->pos < 64 && get_bits_left(&gb) > 0) {
+                dv_decode_ac(&gb, mb, block);
+                /* if still not finished, no need to parse other blocks */
+                if (mb->pos < 64)
+                    break;
+            }
+        }
+        /* all blocks are finished, so the extra bytes can be used at
+           the video segment level */
+        if (j >= s->sys->bpm)
+            bit_copy(&vs_pb, &gb);
+    }
+
+    /* we need a pass over the whole video segment */
+    av_dlog(avctx, "***pass 3 size=%d\n", put_bits_count(&vs_pb));
+    block = &sblock[0][0];
+    mb    = mb_data;
+    init_get_bits(&gb, vs_bit_buffer, put_bits_count(&vs_pb));
+    put_bits32(&vs_pb, 0); // padding must be zeroed
+    flush_put_bits(&vs_pb);
+    for (mb_index = 0; mb_index < 5; mb_index++) {
+        for (j = 0; j < s->sys->bpm; j++) {
+            if (mb->pos < 64) {
+                av_dlog(avctx, "start %d:%d\n", mb_index, j);
+                dv_decode_ac(&gb, mb, block);
+            }
+            if (mb->pos >= 64 && mb->pos < 127)
+                av_log(avctx, AV_LOG_ERROR, "AC EOB marker is absent pos=%d\n", mb->pos);
+            block += 64;
+            mb++;
+        }
+    }
+
+    /* compute idct and place blocks */
+    block = &sblock[0][0];
+    mb    = mb_data;
+    for (mb_index = 0; mb_index < 5; mb_index++) {
+        dv_calculate_mb_xy(s, work_chunk, mb_index, &mb_x, &mb_y);
+
+        /* idct_put'ting luminance */
+        if ((s->sys->pix_fmt == PIX_FMT_YUV420P) ||
+            (s->sys->pix_fmt == PIX_FMT_YUV411P && mb_x >= (704 / 8)) ||
+            (s->sys->height >= 720 && mb_y != 134)) {
+            y_stride = (s->picture.linesize[0] << ((!is_field_mode[mb_index]) * log2_blocksize));
+        } else {
+            y_stride = (2 << log2_blocksize);
+        }
+        y_ptr = s->picture.data[0] + ((mb_y * s->picture.linesize[0] + mb_x) << log2_blocksize);
+        linesize = s->picture.linesize[0] << is_field_mode[mb_index];
+        mb[0]    .idct_put(y_ptr                                   , linesize, block + 0*64);
+        if (s->sys->video_stype == 4) { /* SD 422 */
+            mb[2].idct_put(y_ptr + (1 << log2_blocksize)           , linesize, block + 2*64);
+        } else {
+            mb[1].idct_put(y_ptr + (1 << log2_blocksize)           , linesize, block + 1*64);
+            mb[2].idct_put(y_ptr                         + y_stride, linesize, block + 2*64);
+            mb[3].idct_put(y_ptr + (1 << log2_blocksize) + y_stride, linesize, block + 3*64);
+        }
+        mb += 4;
+        block += 4*64;
+
+        /* idct_put'ting chrominance */
+        c_offset = (((mb_y >>  (s->sys->pix_fmt == PIX_FMT_YUV420P)) * s->picture.linesize[1] +
+                     (mb_x >> ((s->sys->pix_fmt == PIX_FMT_YUV411P) ? 2 : 1))) << log2_blocksize);
+        for (j = 2; j; j--) {
+            uint8_t *c_ptr = s->picture.data[j] + c_offset;
+            if (s->sys->pix_fmt == PIX_FMT_YUV411P && mb_x >= (704 / 8)) {
+                  uint64_t aligned_pixels[64/8];
+                  uint8_t *pixels = (uint8_t*)aligned_pixels;
+                  uint8_t *c_ptr1, *ptr1;
+                  int x, y;
+                  mb->idct_put(pixels, 8, block);
+                  for (y = 0; y < (1 << log2_blocksize); y++, c_ptr += s->picture.linesize[j], pixels += 8) {
+                      ptr1   = pixels + (1 << (log2_blocksize - 1));
+                      c_ptr1 = c_ptr + (s->picture.linesize[j] << log2_blocksize);
+                      for (x = 0; x < (1 << (log2_blocksize - 1)); x++) {
+                          c_ptr[x]  = pixels[x];
+                          c_ptr1[x] = ptr1[x];
+                      }
+                  }
+                  block += 64; mb++;
+            } else {
+                  y_stride = (mb_y == 134) ? (1 << log2_blocksize) :
+                                             s->picture.linesize[j] << ((!is_field_mode[mb_index]) * log2_blocksize);
+                  linesize = s->picture.linesize[j] << is_field_mode[mb_index];
+                  (mb++)->    idct_put(c_ptr           , linesize, block); block += 64;
+                  if (s->sys->bpm == 8) {
+                      (mb++)->idct_put(c_ptr + y_stride, linesize, block); block += 64;
+                  }
+            }
+        }
+    }
+    return 0;
+}
+
+/* NOTE: exactly one frame must be given (120000 bytes for NTSC,
+   144000 bytes for PAL - or twice those for 50Mbps) */
+static int dvvideo_decode_frame(AVCodecContext *avctx,
+                                 void *data, int *data_size,
+                                 AVPacket *avpkt)
+{
+    const uint8_t *buf = avpkt->data;
+    int buf_size = avpkt->size;
+    DVVideoContext *s = avctx->priv_data;
+    const uint8_t* vsc_pack;
+    int apt, is16_9;
+
+    s->sys = avpriv_dv_frame_profile2(avctx, s->sys, buf, buf_size);
+    if (!s->sys || buf_size < s->sys->frame_size || ff_dv_init_dynamic_tables(s->sys)) {
+        av_log(avctx, AV_LOG_ERROR, "could not find dv frame profile\n");
+        return -1; /* NOTE: we only accept several full frames */
+    }
+
+    if (s->picture.data[0])
+        avctx->release_buffer(avctx, &s->picture);
+
+    avcodec_get_frame_defaults(&s->picture);
+    s->picture.reference = 0;
+    s->picture.key_frame = 1;
+    s->picture.pict_type = AV_PICTURE_TYPE_I;
+    avctx->pix_fmt   = s->sys->pix_fmt;
+    avctx->time_base = s->sys->time_base;
+    avcodec_set_dimensions(avctx, s->sys->width, s->sys->height);
+    if (avctx->get_buffer(avctx, &s->picture) < 0) {
+        av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
+        return -1;
+    }
+    s->picture.interlaced_frame = 1;
+    s->picture.top_field_first  = 0;
+
+    s->buf = buf;
+    avctx->execute(avctx, dv_decode_video_segment, s->sys->work_chunks, NULL,
+                   dv_work_pool_size(s->sys), sizeof(DVwork_chunk));
+
+    emms_c();
+
+    /* return image */
+    *data_size = sizeof(AVFrame);
+    *(AVFrame*)data = s->picture;
+
+    /* Determine the codec's sample_aspect ratio from the packet */
+    vsc_pack = buf + 80*5 + 48 + 5;
+    if ( *vsc_pack == dv_video_control ) {
+        apt = buf[4] & 0x07;
+        is16_9 = (vsc_pack[2] & 0x07) == 0x02 || (!apt && (vsc_pack[2] & 0x07) == 0x07);
+        avctx->sample_aspect_ratio = s->sys->sar[is16_9];
+    }
+
+    return s->sys->frame_size;
+}
+
+static int dvvideo_close(AVCodecContext *c)
+{
+    DVVideoContext *s = c->priv_data;
+
+    if (s->picture.data[0])
+        c->release_buffer(c, &s->picture);
+
+    return 0;
+}
+
+AVCodec ff_dvvideo_decoder = {
+    .name           = "dvvideo",
+    .type           = AVMEDIA_TYPE_VIDEO,
+    .id             = CODEC_ID_DVVIDEO,
+    .priv_data_size = sizeof(DVVideoContext),
+    .init           = ff_dvvideo_init,
+    .close          = dvvideo_close,
+    .decode         = dvvideo_decode_frame,
+    .capabilities   = CODEC_CAP_DR1 | CODEC_CAP_SLICE_THREADS,
+    .max_lowres = 3,
+    .long_name = NULL_IF_CONFIG_SMALL("DV (Digital Video)"),
+};
diff --git a/libavcodec/rawdec.c b/libavcodec/rawdec.c
index f27df2ad2d..65593e809f 100644
--- a/libavcodec/rawdec.c
+++ b/libavcodec/rawdec.c
@@ -139,6 +139,7 @@ static int raw_decode(AVCodecContext *avctx,
     int buf_size = avpkt->size;
     int linesize_align = 4;
     RawVideoContext *context = avctx->priv_data;
+    int res;
 
     AVFrame   *frame   = data;
     AVPicture *picture = data;
@@ -190,7 +191,9 @@ static int raw_decode(AVCodecContext *avctx,
     if(buf_size < context->length - (avctx->pix_fmt==PIX_FMT_PAL8 ? 256*4 : 0))
         return -1;
 
-    avpicture_fill(picture, buf, avctx->pix_fmt, avctx->width, avctx->height);
+    if ((res = avpicture_fill(picture, buf, avctx->pix_fmt,
+                              avctx->width, avctx->height)) < 0)
+        return res;
     if((avctx->pix_fmt==PIX_FMT_PAL8 && buf_size < context->length) ||
        (av_pix_fmt_descriptors[avctx->pix_fmt].flags & PIX_FMT_PSEUDOPAL)) {
         frame->data[1]= context->palette;