Merge remote-tracking branch 'qatar/master'

* qatar/master: (46 commits)
  mtv: Make sure audio_subsegments is not 0
  v4l2: use V4L2_FMT_FLAG_EMULATED only if it is defined
  avconv: add symbolic names for -vsync parameters
  flvdec: Fix compiler warning for uninitialized variables
  rtsp: Fix compiler warning for uninitialized variable
  ulti: convert to new bytestream API.
  swscale: Use standard multiple inclusion guards in ppc/ header files.
  Place some START_TIMER invocations in separate blocks.
  v4l2: list available formats
  v4l2: set the proper codec_tag
  v4l2: refactor device_open
  v4l2: simplify away io_method
  v4l2: cosmetics
  v4l2: uniform and format options
  v4l2: do not force interlaced mode
  avio: exit early in fill_buffer without read_packet
  vc1dec: fix invalid memory access for small video dimensions
  rv34: fix invalid memory access for small video dimensions
  rv34: joint coefficient decoding and dequantization
  avplay: Don't call avio_set_interrupt_cb(NULL)
  ...

Conflicts:
	Changelog
	avconv.c
	doc/APIchanges
	doc/indevs.texi
	libavcodec/adxenc.c
	libavcodec/dnxhdenc.c
	libavcodec/h264.c
	libavdevice/v4l2.c
	libavformat/flvdec.c
	libavformat/mtv.c
	libswscale/utils.c

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

17 files changed, 190 insertions, 352 deletions

diff --git a/libavcodec/adxdec.c b/libavcodec/adxdec.c
index fdff6875e1..ec4b1041af 100644
--- a/libavcodec/adxdec.c
+++ b/libavcodec/adxdec.c
@@ -165,6 +165,13 @@ static int adx_decode_frame(AVCodecContext *avctx, void *data,
     return buf - avpkt->data;
 }
 
+static void adx_decode_flush(AVCodecContext *avctx)
+{
+    ADXContext *c = avctx->priv_data;
+    memset(c->prev, 0, sizeof(c->prev));
+    c->eof = 0;
+}
+
 AVCodec ff_adpcm_adx_decoder = {
     .name           = "adpcm_adx",
     .type           = AVMEDIA_TYPE_AUDIO,
@@ -172,6 +179,7 @@ AVCodec ff_adpcm_adx_decoder = {
     .priv_data_size = sizeof(ADXContext),
     .init           = adx_decode_init,
     .decode         = adx_decode_frame,
+    .flush          = adx_decode_flush,
     .capabilities   = CODEC_CAP_DR1,
     .long_name      = NULL_IF_CONFIG_SMALL("SEGA CRI ADX ADPCM"),
 };
diff --git a/libavcodec/adxenc.c b/libavcodec/adxenc.c
index 51545fa06c..b029e2eb78 100644
--- a/libavcodec/adxenc.c
+++ b/libavcodec/adxenc.c
@@ -19,9 +19,9 @@
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "libavutil/intreadwrite.h"
 #include "avcodec.h"
 #include "adx.h"
+#include "bytestream.h"
 #include "put_bits.h"
 
 /**
@@ -33,167 +33,135 @@
  * adx2wav & wav2adx http://www.geocities.co.jp/Playtown/2004/
  */
 
-/* 18 bytes <-> 32 samples */
-
-static void adx_encode(ADXContext *c, unsigned char *adx, const short *wav,
-                       ADXChannelState *prev)
+static void adx_encode(ADXContext *c, uint8_t *adx, const int16_t *wav,
+                       ADXChannelState *prev, int channels)
 {
     PutBitContext pb;
     int scale;
-    int i;
-    int s0,s1,s2,d;
-    int max=0;
-    int min=0;
-    int data[32];
+    int i, j;
+    int s0, s1, s2, d;
+    int max = 0;
+    int min = 0;
+    int data[BLOCK_SAMPLES];
 
     s1 = prev->s1;
     s2 = prev->s2;
-    for(i=0;i<32;i++) {
+    for (i = 0, j = 0; j < 32; i += channels, j++) {
         s0 = wav[i];
         d = ((s0 << COEFF_BITS) - c->coeff[0] * s1 - c->coeff[1] * s2) >> COEFF_BITS;
-        data[i]=d;
-        if (max<d) max=d;
-        if (min>d) min=d;
+        data[j] = d;
+        if (max < d)
+            max = d;
+        if (min > d)
+            min = d;
         s2 = s1;
         s1 = s0;
     }
     prev->s1 = s1;
     prev->s2 = s2;
 
-    /* -8..+7 */
-
-    if (max==0 && min==0) {
-        memset(adx,0,18);
+    if (max == 0 && min == 0) {
+        memset(adx, 0, BLOCK_SIZE);
         return;
     }
 
-    if (max/7>-min/8) scale = max/7;
-    else scale = -min/8;
+    if (max / 7 > -min / 8)
+        scale = max / 7;
+    else
+        scale = -min / 8;
 
-    if (scale==0) scale=1;
+    if (scale == 0)
+        scale = 1;
 
     AV_WB16(adx, scale);
 
     init_put_bits(&pb, adx + 2, 16);
-    for (i = 0; i < 32; i++)
-        put_sbits(&pb, 4, av_clip(data[i]/scale, -8, 7));
+    for (i = 0; i < BLOCK_SAMPLES; i++)
+        put_sbits(&pb, 4, av_clip(data[i] / scale, -8, 7));
     flush_put_bits(&pb);
 }
 
-static int adx_encode_header(AVCodecContext *avctx,unsigned char *buf,size_t bufsize)
+#define HEADER_SIZE 36
+
+static int adx_encode_header(AVCodecContext *avctx, uint8_t *buf, int bufsize)
 {
-#if 0
-    struct {
-        uint32_t offset; /* 0x80000000 + sample start - 4 */
-        unsigned char unknown1[3]; /* 03 12 04 */
-        unsigned char channel; /* 1 or 2 */
-        uint32_t freq;
-        uint32_t size;
-        uint32_t unknown2; /* 01 f4 03 00 */
-        uint32_t unknown3; /* 00 00 00 00 */
-        uint32_t unknown4; /* 00 00 00 00 */
-
-    /* if loop
-        unknown3 00 15 00 01
-        unknown4 00 00 00 01
-        long loop_start_sample;
-        long loop_start_byte;
-        long loop_end_sample;
-        long loop_end_byte;
-        long
-    */
-    } adxhdr; /* big endian */
-    /* offset-6 "(c)CRI" */
-#endif
     ADXContext *c = avctx->priv_data;
 
-    AV_WB32(buf+0x00,0x80000000|0x20);
-    AV_WB32(buf+0x04,0x03120400|avctx->channels);
-    AV_WB32(buf+0x08,avctx->sample_rate);
-    AV_WB32(buf+0x0c,0); /* FIXME: set after */
-    AV_WB16(buf + 0x10, c->cutoff);
-    AV_WB32(buf + 0x12, 0x03000000);
-    AV_WB32(buf + 0x16, 0x00000000);
-    AV_WB32(buf + 0x1a, 0x00000000);
-    memcpy (buf + 0x1e, "(c)CRI", 6);
-    return 0x20+4;
+    if (bufsize < HEADER_SIZE)
+        return AVERROR(EINVAL);
+
+    bytestream_put_be16(&buf, 0x8000);              /* header signature */
+    bytestream_put_be16(&buf, HEADER_SIZE - 4);     /* copyright offset */
+    bytestream_put_byte(&buf, 3);                   /* encoding */
+    bytestream_put_byte(&buf, BLOCK_SIZE);          /* block size */
+    bytestream_put_byte(&buf, 4);                   /* sample size */
+    bytestream_put_byte(&buf, avctx->channels);     /* channels */
+    bytestream_put_be32(&buf, avctx->sample_rate);  /* sample rate */
+    bytestream_put_be32(&buf, 0);                   /* total sample count */
+    bytestream_put_be16(&buf, c->cutoff);           /* cutoff frequency */
+    bytestream_put_byte(&buf, 3);                   /* version */
+    bytestream_put_byte(&buf, 0);                   /* flags */
+    bytestream_put_be32(&buf, 0);                   /* unknown */
+    bytestream_put_be32(&buf, 0);                   /* loop enabled */
+    bytestream_put_be16(&buf, 0);                   /* padding */
+    bytestream_put_buffer(&buf, "(c)CRI", 6);       /* copyright signature */
+
+    return HEADER_SIZE;
 }
 
 static av_cold int adx_encode_init(AVCodecContext *avctx)
 {
     ADXContext *c = avctx->priv_data;
 
-    if (avctx->channels > 2)
-        return -1; /* only stereo or mono =) */
-    avctx->frame_size = 32;
-
-    avctx->coded_frame= avcodec_alloc_frame();
-    avctx->coded_frame->key_frame= 1;
+    if (avctx->channels > 2) {
+        av_log(avctx, AV_LOG_ERROR, "Invalid number of channels\n");
+        return AVERROR(EINVAL);
+    }
+    avctx->frame_size = BLOCK_SAMPLES;
 
-//    avctx->bit_rate = avctx->sample_rate*avctx->channels*18*8/32;
+    avctx->coded_frame = avcodec_alloc_frame();
 
     /* the cutoff can be adjusted, but this seems to work pretty well */
     c->cutoff = 500;
     ff_adx_calculate_coeffs(c->cutoff, avctx->sample_rate, COEFF_BITS, c->coeff);
 
-    av_log(avctx, AV_LOG_DEBUG, "adx encode init\n");
-
     return 0;
 }
 
 static av_cold int adx_encode_close(AVCodecContext *avctx)
 {
     av_freep(&avctx->coded_frame);
-
     return 0;
 }
 
-static int adx_encode_frame(AVCodecContext *avctx,
-                uint8_t *frame, int buf_size, void *data)
+static int adx_encode_frame(AVCodecContext *avctx, uint8_t *frame,
+                            int buf_size, void *data)
 {
-    ADXContext *c = avctx->priv_data;
-    const short *samples = data;
-    unsigned char *dst = frame;
-    int rest = avctx->frame_size;
-
-/*
-    input data size =
-    ffmpeg.c: do_audio_out()
-    frame_bytes = enc->frame_size * 2 * enc->channels;
-*/
+    ADXContext *c          = avctx->priv_data;
+    const int16_t *samples = data;
+    uint8_t *dst           = frame;
+    int ch;
 
-//    printf("sz=%d ",buf_size); fflush(stdout);
     if (!c->header_parsed) {
-        int hdrsize = adx_encode_header(avctx,dst,buf_size);
-        dst+=hdrsize;
+        int hdrsize;
+        if ((hdrsize = adx_encode_header(avctx, dst, buf_size)) < 0) {
+            av_log(avctx, AV_LOG_ERROR, "output buffer is too small\n");
+            return AVERROR(EINVAL);
+        }
+        dst      += hdrsize;
+        buf_size -= hdrsize;
         c->header_parsed = 1;
     }
+    if (buf_size < BLOCK_SIZE * avctx->channels) {
+        av_log(avctx, AV_LOG_ERROR, "output buffer is too small\n");
+        return AVERROR(EINVAL);
+    }
 
-    if (avctx->channels==1) {
-        while(rest>=32) {
-            adx_encode(c, dst, samples, c->prev);
-            dst+=18;
-            samples+=32;
-            rest-=32;
-        }
-    } else {
-        while(rest>=32*2) {
-            short tmpbuf[32*2];
-            int i;
-
-            for(i=0;i<32;i++) {
-                tmpbuf[i] = samples[i*2];
-                tmpbuf[i+32] = samples[i*2+1];
-            }
-
-            adx_encode(c, dst,      tmpbuf,      c->prev);
-            adx_encode(c, dst + 18, tmpbuf + 32, c->prev + 1);
-            dst+=18*2;
-            samples+=32*2;
-            rest-=32*2;
-        }
+    for (ch = 0; ch < avctx->channels; ch++) {
+        adx_encode(c, dst, samples + ch, &c->prev[ch], avctx->channels);
+        dst += BLOCK_SIZE;
     }
-    return dst-frame;
+    return dst - frame;
 }
 
 AVCodec ff_adpcm_adx_encoder = {
@@ -204,6 +172,7 @@ AVCodec ff_adpcm_adx_encoder = {
     .init           = adx_encode_init,
     .encode         = adx_encode_frame,
     .close          = adx_encode_close,
-    .sample_fmts = (const enum AVSampleFormat[]){AV_SAMPLE_FMT_S16,AV_SAMPLE_FMT_NONE},
-    .long_name = NULL_IF_CONFIG_SMALL("SEGA CRI ADX ADPCM"),
+    .sample_fmts    = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S16,
+                                                      AV_SAMPLE_FMT_NONE },
+    .long_name      = NULL_IF_CONFIG_SMALL("SEGA CRI ADX ADPCM"),
 };
diff --git a/libavcodec/arm/rv34dsp_init_neon.c b/libavcodec/arm/rv34dsp_init_neon.c
index acf2a7dcd3..9a09fde7a9 100644
--- a/libavcodec/arm/rv34dsp_init_neon.c
+++ b/libavcodec/arm/rv34dsp_init_neon.c
@@ -25,12 +25,9 @@
 
 void ff_rv34_inv_transform_neon(DCTELEM *block);
 void ff_rv34_inv_transform_noround_neon(DCTELEM *block);
-void ff_rv34_dequant4x4_neon(DCTELEM *block, int Qdc, int Q);
 
 void ff_rv34dsp_init_neon(RV34DSPContext *c, DSPContext* dsp)
 {
     c->rv34_inv_transform_tab[0] = ff_rv34_inv_transform_neon;
     c->rv34_inv_transform_tab[1] = ff_rv34_inv_transform_noround_neon;
-
-    c->rv34_dequant4x4 = ff_rv34_dequant4x4_neon;
 }
diff --git a/libavcodec/arm/rv34dsp_neon.S b/libavcodec/arm/rv34dsp_neon.S
index 423b537fb9..f700f5c321 100644
--- a/libavcodec/arm/rv34dsp_neon.S
+++ b/libavcodec/arm/rv34dsp_neon.S
@@ -107,27 +107,3 @@ function ff_rv34_inv_transform_noround_neon, export=1
         vst4.16         {d0[3], d1[3], d2[3], d3[3]}, [r2,:64], r1
         bx              lr
 endfunc
-
-function ff_rv34_dequant4x4_neon, export=1
-        mov             r3,  r0
-        mov             r12, #16
-        vdup.16         q0,  r2
-        vmov.16         d0[0], r1
-        vld1.16         {d2},     [r0,:64], r12
-        vld1.16         {d4},     [r0,:64], r12
-        vld1.16         {d6},     [r0,:64], r12
-        vld1.16         {d16},    [r0,:64], r12
-        vmull.s16       q1,  d2,  d0
-        vmull.s16       q2,  d4,  d1
-        vmull.s16       q3,  d6,  d1
-        vmull.s16       q8,  d16, d1
-        vqrshrn.s32     d2,  q1,  #4
-        vqrshrn.s32     d4,  q2,  #4
-        vqrshrn.s32     d6,  q3,  #4
-        vqrshrn.s32     d16, q8,  #4
-        vst1.16         {d2},     [r3,:64], r12
-        vst1.16         {d4},     [r3,:64], r12
-        vst1.16         {d6},     [r3,:64], r12
-        vst1.16         {d16},    [r3,:64], r12
-        bx              lr
-endfunc
diff --git a/libavcodec/bytestream.h b/libavcodec/bytestream.h
index 7ca36f8ad3..73ea0c880f 100644
--- a/libavcodec/bytestream.h
+++ b/libavcodec/bytestream.h
@@ -39,11 +39,15 @@ static av_always_inline void bytestream_put_ ##name(uint8_t **b, const type valu
     write(*b, value);\
     (*b) += bytes;\
 }\
+static av_always_inline type bytestream2_get_ ## name ## u(GetByteContext *g)\
+{\
+    return bytestream_get_ ## name(&g->buffer);\
+}\
 static av_always_inline type bytestream2_get_ ## name(GetByteContext *g)\
 {\
     if (g->buffer_end - g->buffer < bytes)\
         return 0;\
-    return bytestream_get_ ## name(&g->buffer);\
+    return bytestream2_get_ ## name ## u(g);\
 }\
 static av_always_inline type bytestream2_peek_ ## name(GetByteContext *g)\
 {\
diff --git a/libavcodec/cabac.c b/libavcodec/cabac.c
index e03043f91f..c603dafddd 100644
--- a/libavcodec/cabac.c
+++ b/libavcodec/cabac.c
@@ -109,10 +109,6 @@ void ff_init_cabac_encoder(CABACContext *c, uint8_t *buf, int buf_size){
     c->low= 0;
     c->range= 0x1FE;
     c->outstanding_count= 0;
-#ifdef STRICT_LIMITS
-    c->sym_count =0;
-#endif
-
     c->pb.bit_left++; //avoids firstBitFlag
 }
 
@@ -183,10 +179,6 @@ static void put_cabac(CABACContext *c, uint8_t * const state, int bit){
     }
 
     renorm_cabac_encoder(c);
-
-#ifdef STRICT_LIMITS
-    c->symCount++;
-#endif
 }
 
 /**
@@ -208,10 +200,6 @@ static void put_cabac_bypass(CABACContext *c, int bit){
         put_cabac_bit(c, 1);
         c->low -= 0x400;
     }
-
-#ifdef STRICT_LIMITS
-    c->symCount++;
-#endif
 }
 
 /**
@@ -236,10 +224,6 @@ static int put_cabac_terminate(CABACContext *c, int bit){
         flush_put_bits(&c->pb); //FIXME FIXME FIXME XXX wrong
     }
 
-#ifdef STRICT_LIMITS
-    c->symCount++;
-#endif
-
     return (put_bits_count(&c->pb)+7)>>3;
 }
 
@@ -365,21 +349,6 @@ START_TIMER
             av_log(NULL, AV_LOG_ERROR, "CABAC failure at %d\n", i);
 STOP_TIMER("get_cabac")
     }
-#if 0
-    for(i=0; i<SIZE; i++){
-START_TIMER
-        if( r[i] != get_cabac_u(&c, state, (i&1) ? 6 : 7, 3, i&1) )
-            av_log(NULL, AV_LOG_ERROR, "CABAC unary (truncated) binarization failure at %d\n", i);
-STOP_TIMER("get_cabac_u")
-    }
-
-    for(i=0; i<SIZE; i++){
-START_TIMER
-        if( r[i] != get_cabac_ueg(&c, state, 3, 0, 1, 2))
-            av_log(NULL, AV_LOG_ERROR, "CABAC unary (truncated) binarization failure at %d\n", i);
-STOP_TIMER("get_cabac_ueg")
-    }
-#endif
     if(!get_cabac_terminate(&c))
         av_log(NULL, AV_LOG_ERROR, "where's the Terminator?\n");
 
diff --git a/libavcodec/cabac.h b/libavcodec/cabac.h
index ed156e6fca..d31d75b030 100644
--- a/libavcodec/cabac.h
+++ b/libavcodec/cabac.h
@@ -41,9 +41,6 @@ typedef struct CABACContext{
     int low;
     int range;
     int outstanding_count;
-#ifdef STRICT_LIMITS
-    int symCount;
-#endif
     const uint8_t *bytestream_start;
     const uint8_t *bytestream;
     const uint8_t *bytestream_end;
@@ -216,62 +213,4 @@ static int av_unused get_cabac_terminate(CABACContext *c){
     }
 }
 
-#if 0
-/**
- * Get (truncated) unary binarization.
- */
-static int get_cabac_u(CABACContext *c, uint8_t * state, int max, int max_index, int truncated){
-    int i;
-
-    for(i=0; i<max; i++){
-        if(get_cabac(c, state)==0)
-            return i;
-
-        if(i< max_index) state++;
-    }
-
-    return truncated ? max : -1;
-}
-
-/**
- * get unary exp golomb k-th order binarization.
- */
-static int get_cabac_ueg(CABACContext *c, uint8_t * state, int max, int is_signed, int k, int max_index){
-    int i, v;
-    int m= 1<<k;
-
-    if(get_cabac(c, state)==0)
-        return 0;
-
-    if(0 < max_index) state++;
-
-    for(i=1; i<max; i++){
-        if(get_cabac(c, state)==0){
-            if(is_signed && get_cabac_bypass(c)){
-                return -i;
-            }else
-                return i;
-        }
-
-        if(i < max_index) state++;
-    }
-
-    while(get_cabac_bypass(c)){
-        i+= m;
-        m+= m;
-    }
-
-    v=0;
-    while(m>>=1){
-        v+= v + get_cabac_bypass(c);
-    }
-    i += v;
-
-    if(is_signed && get_cabac_bypass(c)){
-        return -i;
-    }else
-        return i;
-}
-#endif /* 0 */
-
 #endif /* AVCODEC_CABAC_H */
diff --git a/libavcodec/dnxhdenc.c b/libavcodec/dnxhdenc.c
index 7021a70650..e4c6274ae4 100644
--- a/libavcodec/dnxhdenc.c
+++ b/libavcodec/dnxhdenc.c
@@ -578,9 +578,8 @@ static int dnxhd_encode_thread(AVCodecContext *avctx, void *arg, int jobnr, int
 
         for (i = 0; i < 8; i++) {
             DCTELEM *block = ctx->blocks[i];
-            int last_index, overflow;
-            int n = dnxhd_switch_matrix(ctx, i);
-            last_index = ctx->m.dct_quantize(&ctx->m, block, 4&(2*i), qscale, &overflow);
+            int overflow, n = dnxhd_switch_matrix(ctx, i);
+            int last_index = ctx->m.dct_quantize(&ctx->m, block, 4&(2*i), qscale, &overflow);
             //START_TIMER;
             dnxhd_encode_block(ctx, block, last_index, n);
             //STOP_TIMER("encode_block");
diff --git a/libavcodec/h264.c b/libavcodec/h264.c
index 8a79311f7c..8cd9fe72ef 100644
--- a/libavcodec/h264.c
+++ b/libavcodec/h264.c
@@ -4051,7 +4051,7 @@ static int decode_frame(AVCodecContext *avctx,
     H264Context *h = avctx->priv_data;
     MpegEncContext *s = &h->s;
     AVFrame *pict = data;
-    int buf_index;
+    int buf_index = 0;
     Picture *out;
     int i, out_idx;
 
@@ -4081,7 +4081,7 @@ static int decode_frame(AVCodecContext *avctx,
             *pict= *(AVFrame*)out;
         }
 
-        return buf_size;
+        return buf_index;
     }
     if(h->is_avc && buf_size >= 9 && buf[0]==1 && buf[2]==0 && (buf[4]&0xFC)==0xFC && (buf[5]&0x1F) && buf[8]==0x67){
         int cnt= buf[5]&0x1f;
@@ -4112,7 +4112,6 @@ not_extra:
 
     if (!s->current_picture_ptr && h->nal_unit_type == NAL_END_SEQUENCE) {
         av_assert0(buf_index <= buf_size);
-        buf_size = buf_index;
         goto out;
     }
 
@@ -4193,9 +4192,7 @@ int main(void){
 
     init_get_bits(&gb, temp, 8*SIZE);
     for(i=0; i<COUNT; i++){
-        int j, s;
-
-        s= show_bits(&gb, 24);
+        int j, s = show_bits(&gb, 24);
 
         {START_TIMER
         j= get_ue_golomb(&gb);
@@ -4218,9 +4215,7 @@ int main(void){
 
     init_get_bits(&gb, temp, 8*SIZE);
     for(i=0; i<COUNT; i++){
-        int j, s;
-
-        s= show_bits(&gb, 24);
+        int j, s = show_bits(&gb, 24);
 
         {START_TIMER
         j= get_se_golomb(&gb);
diff --git a/libavcodec/indeo5.c b/libavcodec/indeo5.c
index fc534dd4e4..b1c50c2adb 100644
--- a/libavcodec/indeo5.c
+++ b/libavcodec/indeo5.c
@@ -760,7 +760,7 @@ static int decode_frame(AVCodecContext *avctx, void *data, int *data_size,
 
     switch_buffers(ctx);
 
-    //START_TIMER;
+    //{ START_TIMER;
 
     if (ctx->frame_type != FRAMETYPE_NULL) {
         for (p = 0; p < 3; p++) {
@@ -775,7 +775,7 @@ static int decode_frame(AVCodecContext *avctx, void *data, int *data_size,
         }
     }
 
-    //STOP_TIMER("decode_planes");
+    //STOP_TIMER("decode_planes"); }
 
     if (ctx->frame.data[0])
         avctx->release_buffer(avctx, &ctx->frame);
diff --git a/libavcodec/libspeexenc.c b/libavcodec/libspeexenc.c
index a3176d1388..c89f0748a1 100644
--- a/libavcodec/libspeexenc.c
+++ b/libavcodec/libspeexenc.c
@@ -83,7 +83,8 @@ typedef struct {
     int abr;                    ///< flag to enable ABR
     int pkt_frame_count;        ///< frame count for the current packet
     int lookahead;              ///< encoder delay
-    int sample_count;           ///< total sample count (used for pts)
+    int64_t next_pts;           ///< next pts, in sample_rate time base
+    int pkt_sample_count;       ///< sample count in the current packet
 } LibSpeexEncContext;
 
 static av_cold void print_enc_params(AVCodecContext *avctx,
@@ -201,7 +202,7 @@ static av_cold int encode_init(AVCodecContext *avctx)
 
     /* set encoding delay */
     speex_encoder_ctl(s->enc_state, SPEEX_GET_LOOKAHEAD, &s->lookahead);
-    s->sample_count = -s->lookahead;
+    s->next_pts = -s->lookahead;
 
     /* create header packet bytes from header struct */
     /* note: libspeex allocates the memory for header_data, which is freed
@@ -235,7 +236,6 @@ static int encode_frame(AVCodecContext *avctx, uint8_t *frame, int buf_size,
 {
     LibSpeexEncContext *s = avctx->priv_data;
     int16_t *samples      = data;
-    int sample_count      = s->sample_count;
 
     if (data) {
         /* encode Speex frame */
@@ -243,7 +243,7 @@ static int encode_frame(AVCodecContext *avctx, uint8_t *frame, int buf_size,
             speex_encode_stereo_int(samples, s->header.frame_size, &s->bits);
         speex_encode_int(s->enc_state, samples, &s->bits);
         s->pkt_frame_count++;
-        s->sample_count += avctx->frame_size;
+        s->pkt_sample_count += avctx->frame_size;
     } else {
         /* handle end-of-stream */
         if (!s->pkt_frame_count)
@@ -259,8 +259,10 @@ static int encode_frame(AVCodecContext *avctx, uint8_t *frame, int buf_size,
     if (s->pkt_frame_count == s->frames_per_packet) {
         s->pkt_frame_count = 0;
         avctx->coded_frame->pts =
-            av_rescale_q(sample_count, (AVRational){ 1, avctx->sample_rate },
+            av_rescale_q(s->next_pts, (AVRational){ 1, avctx->sample_rate },
                          avctx->time_base);
+        s->next_pts += s->pkt_sample_count;
+        s->pkt_sample_count = 0;
         if (buf_size > speex_bits_nbytes(&s->bits)) {
             int ret = speex_bits_write(&s->bits, frame, buf_size);
             speex_bits_reset(&s->bits);
diff --git a/libavcodec/rv34.c b/libavcodec/rv34.c
index afe8c1f874..12a539e89a 100644
--- a/libavcodec/rv34.c
+++ b/libavcodec/rv34.c
@@ -212,7 +212,7 @@ static int rv34_decode_cbp(GetBitContext *gb, RV34VLC *vlc, int table)
 /**
  * Get one coefficient value from the bistream and store it.
  */
-static inline void decode_coeff(DCTELEM *dst, int coef, int esc, GetBitContext *gb, VLC* vlc)
+static inline void decode_coeff(DCTELEM *dst, int coef, int esc, GetBitContext *gb, VLC* vlc, int q)
 {
     if(coef){
         if(coef == esc){
@@ -225,14 +225,14 @@ static inline void decode_coeff(DCTELEM *dst, int coef, int esc, GetBitContext *
         }
         if(get_bits1(gb))
             coef = -coef;
-        *dst = coef;
+        *dst = (coef*q + 8) >> 4;
     }
 }
 
 /**
  * Decode 2x2 subblock of coefficients.
  */
-static inline void decode_subblock(DCTELEM *dst, int code, const int is_block2, GetBitContext *gb, VLC *vlc)
+static inline void decode_subblock(DCTELEM *dst, int code, const int is_block2, GetBitContext *gb, VLC *vlc, int q)
 {
     int coeffs[4];
 
@@ -240,15 +240,35 @@ static inline void decode_subblock(DCTELEM *dst, int code, const int is_block2,
     coeffs[1] = modulo_three_table[code][1];
     coeffs[2] = modulo_three_table[code][2];
     coeffs[3] = modulo_three_table[code][3];
-    decode_coeff(dst  , coeffs[0], 3, gb, vlc);
+    decode_coeff(dst  , coeffs[0], 3, gb, vlc, q);
     if(is_block2){
-        decode_coeff(dst+8, coeffs[1], 2, gb, vlc);
-        decode_coeff(dst+1, coeffs[2], 2, gb, vlc);
+        decode_coeff(dst+8, coeffs[1], 2, gb, vlc, q);
+        decode_coeff(dst+1, coeffs[2], 2, gb, vlc, q);
     }else{
-        decode_coeff(dst+1, coeffs[1], 2, gb, vlc);
-        decode_coeff(dst+8, coeffs[2], 2, gb, vlc);
+        decode_coeff(dst+1, coeffs[1], 2, gb, vlc, q);
+        decode_coeff(dst+8, coeffs[2], 2, gb, vlc, q);
     }
-    decode_coeff(dst+9, coeffs[3], 2, gb, vlc);
+    decode_coeff(dst+9, coeffs[3], 2, gb, vlc, q);
+}
+
+static inline void decode_subblock3(DCTELEM *dst, int code, const int is_block2, GetBitContext *gb, VLC *vlc,
+                                    int q_dc, int q_ac1, int q_ac2)
+{
+    int coeffs[4];
+
+    coeffs[0] = modulo_three_table[code][0];
+    coeffs[1] = modulo_three_table[code][1];
+    coeffs[2] = modulo_three_table[code][2];
+    coeffs[3] = modulo_three_table[code][3];
+    decode_coeff(dst  , coeffs[0], 3, gb, vlc, q_dc);
+    if(is_block2){
+        decode_coeff(dst+8, coeffs[1], 2, gb, vlc, q_ac1);
+        decode_coeff(dst+1, coeffs[2], 2, gb, vlc, q_ac1);
+    }else{
+        decode_coeff(dst+1, coeffs[1], 2, gb, vlc, q_ac1);
+        decode_coeff(dst+8, coeffs[2], 2, gb, vlc, q_ac1);
+    }
+    decode_coeff(dst+9, coeffs[3], 2, gb, vlc, q_ac2);
 }
 
 /**
@@ -262,7 +282,7 @@ static inline void decode_subblock(DCTELEM *dst, int code, const int is_block2,
  *  o--o
  */
 
-static inline void rv34_decode_block(DCTELEM *dst, GetBitContext *gb, RV34VLC *rvlc, int fc, int sc)
+static inline void rv34_decode_block(DCTELEM *dst, GetBitContext *gb, RV34VLC *rvlc, int fc, int sc, int q_dc, int q_ac1, int q_ac2)
 {
     int code, pattern;
 
@@ -271,40 +291,24 @@ static inline void rv34_decode_block(DCTELEM *dst, GetBitContext *gb, RV34VLC *r
     pattern = code & 0x7;
 
     code >>= 3;
-    decode_subblock(dst, code, 0, gb, &rvlc->coefficient);
+    decode_subblock3(dst, code, 0, gb, &rvlc->coefficient, q_dc, q_ac1, q_ac2);
 
     if(pattern & 4){
         code = get_vlc2(gb, rvlc->second_pattern[sc].table, 9, 2);
-        decode_subblock(dst + 2, code, 0, gb, &rvlc->coefficient);
+        decode_subblock(dst + 2, code, 0, gb, &rvlc->coefficient, q_ac2);
     }
     if(pattern & 2){ // Looks like coefficients 1 and 2 are swapped for this block
         code = get_vlc2(gb, rvlc->second_pattern[sc].table, 9, 2);
-        decode_subblock(dst + 8*2, code, 1, gb, &rvlc->coefficient);
+        decode_subblock(dst + 8*2, code, 1, gb, &rvlc->coefficient, q_ac2);
     }
     if(pattern & 1){
         code = get_vlc2(gb, rvlc->third_pattern[sc].table, 9, 2);
-        decode_subblock(dst + 8*2+2, code, 0, gb, &rvlc->coefficient);
+        decode_subblock(dst + 8*2+2, code, 0, gb, &rvlc->coefficient, q_ac2);
     }
 
 }
 
 /**
- * Dequantize 4x4 block of DC values for 16x16 macroblock.
- * @todo optimize
- */
-static inline void rv34_dequant4x4_16x16(DCTELEM *block, int Qdc, int Q)
-{
-    int i;
-
-    for(i = 0; i < 3; i++)
-         block[rv34_dezigzag[i]] = (block[rv34_dezigzag[i]] * Qdc + 8) >> 4;
-    for(; i < 16; i++)
-         block[rv34_dezigzag[i]] = (block[rv34_dezigzag[i]] * Q + 8) >> 4;
-}
-/** @} */ //block functions
-
-
-/**
  * @name RV30/40 bitstream parsing
  * @{
  */
@@ -676,8 +680,9 @@ static inline void rv34_mc(RV34DecContext *r, const int block_type,
     srcY += src_y * s->linesize + src_x;
     srcU += uvsrc_y * s->uvlinesize + uvsrc_x;
     srcV += uvsrc_y * s->uvlinesize + uvsrc_x;
-    if(   (unsigned)(src_x - !!lx*2) > s->h_edge_pos - !!lx*2 - (width <<3) - 4
-       || (unsigned)(src_y - !!ly*2) > s->v_edge_pos - !!ly*2 - (height<<3) - 4){
+    if(s->h_edge_pos - (width << 3) < 6 || s->v_edge_pos - (height << 3) < 6 ||
+       (unsigned)(src_x - !!lx*2) > s->h_edge_pos - !!lx*2 - (width <<3) - 4 ||
+       (unsigned)(src_y - !!ly*2) > s->v_edge_pos - !!ly*2 - (height<<3) - 4) {
         uint8_t *uvbuf = s->edge_emu_buffer + 22 * s->linesize;
 
         srcY -= 2 + 2*s->linesize;
@@ -1097,6 +1102,7 @@ static int rv34_decode_macroblock(RV34DecContext *r, int8_t *intra_types)
     MpegEncContext *s = &r->s;
     GetBitContext *gb = &s->gb;
     int cbp, cbp2;
+    int q_dc, q_ac;
     int i, blknum, blkoff;
     LOCAL_ALIGNED_16(DCTELEM, block16, [64]);
     int luma_dc_quant;
@@ -1133,31 +1139,34 @@ static int rv34_decode_macroblock(RV34DecContext *r, int8_t *intra_types)
 
     luma_dc_quant = r->block_type == RV34_MB_P_MIX16x16 ? r->luma_dc_quant_p[s->qscale] : r->luma_dc_quant_i[s->qscale];
     if(r->is16){
+        q_dc = rv34_qscale_tab[luma_dc_quant];
+        q_ac = rv34_qscale_tab[s->qscale];
         memset(block16, 0, 64 * sizeof(*block16));
-        rv34_decode_block(block16, gb, r->cur_vlcs, 3, 0);
-        rv34_dequant4x4_16x16(block16, rv34_qscale_tab[luma_dc_quant],rv34_qscale_tab[s->qscale]);
+        rv34_decode_block(block16, gb, r->cur_vlcs, 3, 0, q_dc, q_dc, q_ac);
         r->rdsp.rv34_inv_transform_tab[1](block16);
     }
 
+    q_ac = rv34_qscale_tab[s->qscale];
     for(i = 0; i < 16; i++, cbp >>= 1){
         if(!r->is16 && !(cbp & 1)) continue;
         blknum = ((i & 2) >> 1) + ((i & 8) >> 2);
         blkoff = ((i & 1) << 2) + ((i & 4) << 3);
         if(cbp & 1)
-            rv34_decode_block(s->block[blknum] + blkoff, gb, r->cur_vlcs, r->luma_vlc, 0);
-        r->rdsp.rv34_dequant4x4(s->block[blknum] + blkoff, rv34_qscale_tab[s->qscale],rv34_qscale_tab[s->qscale]);
+            rv34_decode_block(s->block[blknum] + blkoff, gb,
+                              r->cur_vlcs, r->luma_vlc, 0, q_ac, q_ac, q_ac);
         if(r->is16) //FIXME: optimize
             s->block[blknum][blkoff] = block16[(i & 3) | ((i & 0xC) << 1)];
         r->rdsp.rv34_inv_transform_tab[0](s->block[blknum] + blkoff);
     }
     if(r->block_type == RV34_MB_P_MIX16x16)
         r->cur_vlcs = choose_vlc_set(r->si.quant, r->si.vlc_set, 1);
+    q_dc = rv34_qscale_tab[rv34_chroma_quant[1][s->qscale]];
+    q_ac = rv34_qscale_tab[rv34_chroma_quant[0][s->qscale]];
     for(; i < 24; i++, cbp >>= 1){
         if(!(cbp & 1)) continue;
         blknum = ((i & 4) >> 2) + 4;
         blkoff = ((i & 1) << 2) + ((i & 2) << 4);
-        rv34_decode_block(s->block[blknum] + blkoff, gb, r->cur_vlcs, r->chroma_vlc, 1);
-        r->rdsp.rv34_dequant4x4(s->block[blknum] + blkoff, rv34_qscale_tab[rv34_chroma_quant[1][s->qscale]],rv34_qscale_tab[rv34_chroma_quant[0][s->qscale]]);
+        rv34_decode_block(s->block[blknum] + blkoff, gb, r->cur_vlcs, r->chroma_vlc, 1, q_dc, q_ac, q_ac);
         r->rdsp.rv34_inv_transform_tab[0](s->block[blknum] + blkoff);
     }
     if (IS_INTRA(s->current_picture_ptr->f.mb_type[mb_pos]))
diff --git a/libavcodec/rv34data.h b/libavcodec/rv34data.h
index 2155084d09..1b608e7e35 100644
--- a/libavcodec/rv34data.h
+++ b/libavcodec/rv34data.h
@@ -101,16 +101,6 @@ static const uint16_t rv34_qscale_tab[32] = {
 };
 
 /**
- * 4x4 dezigzag pattern
- */
-static const uint8_t rv34_dezigzag[16] = {
-  0,  1,  8, 16,
-  9,  2,  3, 10,
- 17, 24, 25, 18,
- 11, 19, 26, 27
-};
-
-/**
  * tables used to translate a quantizer value into a VLC set for decoding
  * The first table is used for intraframes.
  */
diff --git a/libavcodec/rv34dsp.c b/libavcodec/rv34dsp.c
index 974bf9ec16..1f4cea8544 100644
--- a/libavcodec/rv34dsp.c
+++ b/libavcodec/rv34dsp.c
@@ -100,26 +100,10 @@ static void rv34_inv_transform_noround_c(DCTELEM *block){
 /** @} */ // transform
 
 
-/**
- * Dequantize ordinary 4x4 block.
- */
-void ff_rv34_dequant4x4_neon(DCTELEM *block, int Qdc, int Q);
-static void rv34_dequant4x4_c(DCTELEM *block, int Qdc, int Q)
-{
-    int i, j;
-
-    block[0] = (block[0] * Qdc + 8) >> 4;
-    for (i = 0; i < 4; i++)
-        for (j = !i; j < 4; j++)
-            block[j + i*8] = (block[j + i*8] * Q + 8) >> 4;
-}
-
 av_cold void ff_rv34dsp_init(RV34DSPContext *c, DSPContext* dsp) {
     c->rv34_inv_transform_tab[0] = rv34_inv_transform_c;
     c->rv34_inv_transform_tab[1] = rv34_inv_transform_noround_c;
 
-    c->rv34_dequant4x4 = rv34_dequant4x4_c;
-
     if (HAVE_NEON)
         ff_rv34dsp_init_neon(c, dsp);
 }
diff --git a/libavcodec/rv34dsp.h b/libavcodec/rv34dsp.h
index 01352ea793..f2bc20e911 100644
--- a/libavcodec/rv34dsp.h
+++ b/libavcodec/rv34dsp.h
@@ -56,7 +56,6 @@ typedef struct RV34DSPContext {
     h264_chroma_mc_func avg_chroma_pixels_tab[3];
     rv40_weight_func rv40_weight_pixels_tab[2];
     rv34_inv_transform_func rv34_inv_transform_tab[2];
-    void (*rv34_dequant4x4)(DCTELEM *block, int Qdc, int Q);
     rv40_weak_loop_filter_func rv40_weak_loop_filter[2];
     rv40_strong_loop_filter_func rv40_strong_loop_filter[2];
     rv40_loop_filter_strength_func rv40_loop_filter_strength[2];
diff --git a/libavcodec/ulti.c b/libavcodec/ulti.c
index 4f0d90f665..6e7ba35f9d 100644
--- a/libavcodec/ulti.c
+++ b/libavcodec/ulti.c
@@ -38,16 +38,9 @@ typedef struct UltimotionDecodeContext {
     int width, height, blocks;
     AVFrame frame;
     const uint8_t *ulti_codebook;
+    GetByteContext gb;
 } UltimotionDecodeContext;
 
-#define CHECK_OVERREAD_SIZE(size) \
-    do { \
-        if (buf_end - buf < (size)) { \
-            av_log(avctx, AV_LOG_ERROR, "Insufficient data\n"); \
-            return AVERROR_INVALIDDATA; \
-        } \
-    } while(0)
-
 static av_cold int ulti_decode_init(AVCodecContext *avctx)
 {
     UltimotionDecodeContext *s = avctx->priv_data;
@@ -232,7 +225,6 @@ static int ulti_decode_frame(AVCodecContext *avctx,
     int i;
     int skip;
     int tmp;
-    const uint8_t *buf_end = buf + buf_size;
 
     s->frame.reference = 3;
     s->frame.buffer_hints = FF_BUFFER_HINTS_VALID | FF_BUFFER_HINTS_PRESERVE | FF_BUFFER_HINTS_REUSABLE;
@@ -241,18 +233,20 @@ static int ulti_decode_frame(AVCodecContext *avctx,
         return -1;
     }
 
+    bytestream2_init(&s->gb, buf, buf_size);
+
     while(!done) {
         int idx;
         if(blocks >= s->blocks || y >= s->height)
             break;//all blocks decoded
 
-        CHECK_OVERREAD_SIZE(1);
-        idx = *buf++;
+        if (bytestream2_get_bytes_left(&s->gb) < 1)
+            goto err;
+        idx = bytestream2_get_byteu(&s->gb);
         if((idx & 0xF8) == 0x70) {
             switch(idx) {
             case 0x70: //change modifier
-                CHECK_OVERREAD_SIZE(1);
-                modifier = *buf++;
+                modifier = bytestream2_get_byte(&s->gb);
                 if(modifier>1)
                     av_log(avctx, AV_LOG_INFO, "warning: modifier must be 0 or 1, got %i\n", modifier);
                 break;
@@ -266,8 +260,7 @@ static int ulti_decode_frame(AVCodecContext *avctx,
                 done = 1;
                 break;
             case 0x74: //skip some blocks
-                CHECK_OVERREAD_SIZE(1);
-                skip = *buf++;
+                skip = bytestream2_get_byte(&s->gb);
                 if ((blocks + skip) >= s->blocks)
                     break;
                 blocks += skip;
@@ -294,8 +287,7 @@ static int ulti_decode_frame(AVCodecContext *avctx,
             } else {
                 cf = 0;
                 if (idx) {
-                    CHECK_OVERREAD_SIZE(1);
-                    chroma = *buf++;
+                    chroma = bytestream2_get_byte(&s->gb);
                 }
             }
             for (i = 0; i < 4; i++) { // for every subblock
@@ -303,15 +295,13 @@ static int ulti_decode_frame(AVCodecContext *avctx,
                 if(!code) //skip subblock
                     continue;
                 if(cf) {
-                    CHECK_OVERREAD_SIZE(1);
-                    chroma = *buf++;
+                    chroma = bytestream2_get_byte(&s->gb);
                 }
                 tx = x + block_coords[i * 2];
                 ty = y + block_coords[(i * 2) + 1];
                 switch(code) {
                 case 1:
-                    CHECK_OVERREAD_SIZE(1);
-                    tmp = *buf++;
+                    tmp = bytestream2_get_byte(&s->gb);
 
                     angle = angle_by_index[(tmp >> 6) & 0x3];
 
@@ -331,8 +321,7 @@ static int ulti_decode_frame(AVCodecContext *avctx,
 
                 case 2:
                     if (modifier) { // unpack four luma samples
-                        CHECK_OVERREAD_SIZE(3);
-                        tmp = bytestream_get_be24(&buf);
+                        tmp = bytestream2_get_be24(&s->gb);
 
                         Y[0] = (tmp >> 18) & 0x3F;
                         Y[1] = (tmp >> 12) & 0x3F;
@@ -340,8 +329,7 @@ static int ulti_decode_frame(AVCodecContext *avctx,
                         Y[3] = tmp & 0x3F;
                         angle = 16;
                     } else { // retrieve luma samples from codebook
-                        CHECK_OVERREAD_SIZE(2);
-                        tmp = bytestream_get_be16(&buf);
+                        tmp = bytestream2_get_be16(&s->gb);
 
                         angle = (tmp >> 12) & 0xF;
                         tmp &= 0xFFF;
@@ -357,27 +345,27 @@ static int ulti_decode_frame(AVCodecContext *avctx,
                     if (modifier) { // all 16 luma samples
                         uint8_t Luma[16];
 
-                        CHECK_OVERREAD_SIZE(12);
-
-                        tmp = bytestream_get_be24(&buf);
+                        if (bytestream2_get_bytes_left(&s->gb) < 12)
+                            goto err;
+                        tmp = bytestream2_get_be24u(&s->gb);
                         Luma[0] = (tmp >> 18) & 0x3F;
                         Luma[1] = (tmp >> 12) & 0x3F;
                         Luma[2] = (tmp >> 6) & 0x3F;
                         Luma[3] = tmp & 0x3F;
 
-                        tmp = bytestream_get_be24(&buf);
+                        tmp = bytestream2_get_be24u(&s->gb);
                         Luma[4] = (tmp >> 18) & 0x3F;
                         Luma[5] = (tmp >> 12) & 0x3F;
                         Luma[6] = (tmp >> 6) & 0x3F;
                         Luma[7] = tmp & 0x3F;
 
-                        tmp = bytestream_get_be24(&buf);
+                        tmp = bytestream2_get_be24u(&s->gb);
                         Luma[8] = (tmp >> 18) & 0x3F;
                         Luma[9] = (tmp >> 12) & 0x3F;
                         Luma[10] = (tmp >> 6) & 0x3F;
                         Luma[11] = tmp & 0x3F;
 
-                        tmp = bytestream_get_be24(&buf);
+                        tmp = bytestream2_get_be24u(&s->gb);
                         Luma[12] = (tmp >> 18) & 0x3F;
                         Luma[13] = (tmp >> 12) & 0x3F;
                         Luma[14] = (tmp >> 6) & 0x3F;
@@ -385,22 +373,23 @@ static int ulti_decode_frame(AVCodecContext *avctx,
 
                         ulti_convert_yuv(&s->frame, tx, ty, Luma, chroma);
                     } else {
-                        CHECK_OVERREAD_SIZE(4);
-                        tmp = *buf++;
+                        if (bytestream2_get_bytes_left(&s->gb) < 4)
+                            goto err;
+                        tmp = bytestream2_get_byteu(&s->gb);
                         if(tmp & 0x80) {
                             angle = (tmp >> 4) & 0x7;
-                            tmp = (tmp << 8) + *buf++;
+                            tmp = (tmp << 8) + bytestream2_get_byteu(&s->gb);
                             Y[0] = (tmp >> 6) & 0x3F;
                             Y[1] = tmp & 0x3F;
-                            Y[2] = (*buf++) & 0x3F;
-                            Y[3] = (*buf++) & 0x3F;
+                            Y[2] = bytestream2_get_byteu(&s->gb) & 0x3F;
+                            Y[3] = bytestream2_get_byteu(&s->gb) & 0x3F;
                             ulti_grad(&s->frame, tx, ty, Y, chroma, angle); //draw block
                         } else { // some patterns
                             int f0, f1;
-                            f0 = *buf++;
+                            f0 = bytestream2_get_byteu(&s->gb);
                             f1 = tmp;
-                            Y[0] = (*buf++) & 0x3F;
-                            Y[1] = (*buf++) & 0x3F;
+                            Y[0] = bytestream2_get_byteu(&s->gb) & 0x3F;
+                            Y[1] = bytestream2_get_byteu(&s->gb) & 0x3F;
                             ulti_pattern(&s->frame, tx, ty, f1, f0, Y[0], Y[1], chroma);
                         }
                     }
@@ -422,6 +411,11 @@ static int ulti_decode_frame(AVCodecContext *avctx,
     *(AVFrame*)data= s->frame;
 
     return buf_size;
+
+err:
+    av_log(avctx, AV_LOG_ERROR,
+           "Insufficient data\n");
+    return AVERROR_INVALIDDATA;
 }
 
 AVCodec ff_ulti_decoder = {
diff --git a/libavcodec/vc1dec.c b/libavcodec/vc1dec.c
index e4ecf14b5b..1469d815ce 100644
--- a/libavcodec/vc1dec.c
+++ b/libavcodec/vc1dec.c
@@ -568,6 +568,7 @@ static void vc1_mc_1mv(VC1Context *v, int dir)
     }
 
     if (v->rangeredfrm || (v->mv_mode == MV_PMODE_INTENSITY_COMP)
+        || s->h_edge_pos < 22 || v_edge_pos < 22
         || (unsigned)(src_x - s->mspel) > s->h_edge_pos - (mx&3) - 16 - s->mspel * 3
         || (unsigned)(src_y - s->mspel) > v_edge_pos    - (my&3) - 16 - s->mspel * 3) {
         uint8_t *uvbuf = s->edge_emu_buffer + 19 * s->linesize;
@@ -799,6 +800,7 @@ static void vc1_mc_4mv_luma(VC1Context *v, int n, int dir)
     if (fieldmv && (src_y & 1) && src_y < 4)
         src_y--;
     if (v->rangeredfrm || (v->mv_mode == MV_PMODE_INTENSITY_COMP)
+        || s->h_edge_pos < 13 || v_edge_pos < 23
         || (unsigned)(src_x - s->mspel) > s->h_edge_pos - (mx & 3) - 8 - s->mspel * 2
         || (unsigned)(src_y - (s->mspel << fieldmv)) > v_edge_pos - (my & 3) - ((8 + s->mspel * 2) << fieldmv)) {
         srcY -= s->mspel * (1 + (s->linesize << fieldmv));
@@ -998,6 +1000,7 @@ static void vc1_mc_4mv_chroma(VC1Context *v, int dir)
     }
 
     if (v->rangeredfrm || (v->mv_mode == MV_PMODE_INTENSITY_COMP)
+        || s->h_edge_pos < 18 || v_edge_pos < 18
         || (unsigned)uvsrc_x > (s->h_edge_pos >> 1) - 9
         || (unsigned)uvsrc_y > (v_edge_pos    >> 1) - 9) {
         s->dsp.emulated_edge_mc(s->edge_emu_buffer     , srcU, s->uvlinesize,
@@ -1102,6 +1105,7 @@ static void vc1_mc_4mv_chroma4(VC1Context *v)
         if (fieldmv && (uvsrc_y & 1) && uvsrc_y < 2)
             uvsrc_y--;
         if ((v->mv_mode == MV_PMODE_INTENSITY_COMP)
+            || s->h_edge_pos < 10 || v_edge_pos < (5 << fieldmv)
             || (unsigned)uvsrc_x > (s->h_edge_pos >> 1) - 5
             || (unsigned)uvsrc_y > v_edge_pos - (5 << fieldmv)) {
             s->dsp.emulated_edge_mc(s->edge_emu_buffer, srcU, s->uvlinesize,
@@ -2006,7 +2010,7 @@ static void vc1_interp_mc(VC1Context *v)
         srcV = s->edge_emu_buffer + 18 * s->linesize;
     }
 
-    if (v->rangeredfrm
+    if (v->rangeredfrm || s->h_edge_pos < 22 || v_edge_pos < 22
         || (unsigned)(src_x - s->mspel) > s->h_edge_pos - (mx & 3) - 16 - s->mspel * 3
         || (unsigned)(src_y - s->mspel) > v_edge_pos    - (my & 3) - 16 - s->mspel * 3) {
         uint8_t *uvbuf = s->edge_emu_buffer + 19 * s->linesize;