Merge remote-tracking branch 'qatar/master'

* qatar/master:
  swscale: make yuv2yuv1 use named registers.
  h264: mark h264_idct_add8_10 with number of XMM registers.
  swscale: fix V plane memory location in bilinear/unscaled RGB/YUYV case.
  vp8: always update next_framep[] before returning from decode_frame().
  avconv: estimate next_dts from framerate if it is set.
  avconv: better next_dts usage.
  avconv: rename InputStream.pts to last_dts.
  avconv: reduce overloading for InputStream.pts.
  avconv: rename InputStream.next_pts to next_dts.
  avconv: rework -t handling for encoding.
  avconv: set encoder timebase for subtitles.
  pva-demux test: add -vn
  swscale: K&R formatting cosmetics for SPARC code
  apedec: allow the user to set the maximum number of output samples per call
  apedec: do not unnecessarily zero output samples for mono frames
  apedec: allocate a single flat buffer for decoded samples
  apedec: use sizeof(field) instead of sizeof(type)
  swscale: split C output functions into separate file.
  swscale: Split C input functions into separate file.
  bytestream: Add bytestream2 writing API.

The avconv changes are due to massive regressions and bugs not merged yet.

Conflicts:
	ffmpeg.c
	libavcodec/vp8.c
	libswscale/swscale.c
	libswscale/x86/swscale_template.c
	tests/fate/demux.mak
	tests/ref/lavf/asf
	tests/ref/lavf/avi
	tests/ref/lavf/mkv
	tests/ref/lavf/mpg
	tests/ref/lavf/nut
	tests/ref/lavf/ogg
	tests/ref/lavf/rm
	tests/ref/lavf/ts
	tests/ref/seek/lavf_avi
	tests/ref/seek/lavf_mkv
	tests/ref/seek/lavf_rm

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

4 files changed, 206 insertions, 90 deletions

diff --git a/libavcodec/apedec.c b/libavcodec/apedec.c
index 35c9a23dd6..1cd7d1e66d 100644
--- a/libavcodec/apedec.c
+++ b/libavcodec/apedec.c
@@ -25,13 +25,13 @@
 #include "bytestream.h"
 #include "libavutil/audioconvert.h"
 #include "libavutil/avassert.h"
+#include "libavutil/opt.h"
 
 /**
  * @file
  * Monkey's Audio lossless audio decoder
  */
 
-#define BLOCKS_PER_LOOP     4608
 #define MAX_CHANNELS        2
 #define MAX_BYTESPERSAMPLE  3
 
@@ -126,6 +126,7 @@ typedef struct APEPredictor {
 
 /** Decoder context */
 typedef struct APEContext {
+    AVClass *class;                          ///< class for AVOptions
     AVCodecContext *avctx;
     AVFrame frame;
     DSPContext dsp;
@@ -142,8 +143,10 @@ typedef struct APEContext {
     int frameflags;                          ///< frame flags
     APEPredictor predictor;                  ///< predictor used for final reconstruction
 
-    int32_t decoded0[BLOCKS_PER_LOOP];       ///< decoded data for the first channel
-    int32_t decoded1[BLOCKS_PER_LOOP];       ///< decoded data for the second channel
+    int32_t *decoded_buffer;
+    int decoded_size;
+    int32_t *decoded[MAX_CHANNELS];          ///< decoded data for each channel
+    int blocks_per_loop;                     ///< maximum number of samples to decode for each call
 
     int16_t* filterbuf[APE_FILTER_LEVELS];   ///< filter memory
 
@@ -170,8 +173,9 @@ static av_cold int ape_decode_close(AVCodecContext *avctx)
     for (i = 0; i < APE_FILTER_LEVELS; i++)
         av_freep(&s->filterbuf[i]);
 
+    av_freep(&s->decoded_buffer);
     av_freep(&s->data);
-    s->data_size = 0;
+    s->decoded_size = s->data_size = 0;
 
     return 0;
 }
@@ -469,19 +473,13 @@ static inline int ape_decode_value(APEContext *ctx, APERice *rice)
 
 static void entropy_decode(APEContext *ctx, int blockstodecode, int stereo)
 {
-    int32_t *decoded0 = ctx->decoded0;
-    int32_t *decoded1 = ctx->decoded1;
+    int32_t *decoded0 = ctx->decoded[0];
+    int32_t *decoded1 = ctx->decoded[1];
 
-    if (ctx->frameflags & APE_FRAMECODE_STEREO_SILENCE) {
-        /* We are pure silence, just memset the output buffer. */
-        memset(decoded0, 0, blockstodecode * sizeof(int32_t));
-        memset(decoded1, 0, blockstodecode * sizeof(int32_t));
-    } else {
-        while (blockstodecode--) {
-            *decoded0++ = ape_decode_value(ctx, &ctx->riceY);
-            if (stereo)
-                *decoded1++ = ape_decode_value(ctx, &ctx->riceX);
-        }
+    while (blockstodecode--) {
+        *decoded0++ = ape_decode_value(ctx, &ctx->riceY);
+        if (stereo)
+            *decoded1++ = ape_decode_value(ctx, &ctx->riceX);
     }
 }
 
@@ -525,7 +523,7 @@ static void init_predictor_decoder(APEContext *ctx)
     APEPredictor *p = &ctx->predictor;
 
     /* Zero the history buffers */
-    memset(p->historybuffer, 0, PREDICTOR_SIZE * sizeof(int32_t));
+    memset(p->historybuffer, 0, PREDICTOR_SIZE * sizeof(*p->historybuffer));
     p->buf = p->historybuffer;
 
     /* Initialize and zero the coefficients */
@@ -593,8 +591,8 @@ static av_always_inline int predictor_update_filter(APEPredictor *p,
 static void predictor_decode_stereo(APEContext *ctx, int count)
 {
     APEPredictor *p = &ctx->predictor;
-    int32_t *decoded0 = ctx->decoded0;
-    int32_t *decoded1 = ctx->decoded1;
+    int32_t *decoded0 = ctx->decoded[0];
+    int32_t *decoded1 = ctx->decoded[1];
 
     while (count--) {
         /* Predictor Y */
@@ -610,7 +608,8 @@ static void predictor_decode_stereo(APEContext *ctx, int count)
 
         /* Have we filled the history buffer? */
         if (p->buf == p->historybuffer + HISTORY_SIZE) {
-            memmove(p->historybuffer, p->buf, PREDICTOR_SIZE * sizeof(int32_t));
+            memmove(p->historybuffer, p->buf,
+                    PREDICTOR_SIZE * sizeof(*p->historybuffer));
             p->buf = p->historybuffer;
         }
     }
@@ -619,7 +618,7 @@ static void predictor_decode_stereo(APEContext *ctx, int count)
 static void predictor_decode_mono(APEContext *ctx, int count)
 {
     APEPredictor *p = &ctx->predictor;
-    int32_t *decoded0 = ctx->decoded0;
+    int32_t *decoded0 = ctx->decoded[0];
     int32_t predictionA, currentA, A, sign;
 
     currentA = p->lastA[0];
@@ -650,7 +649,8 @@ static void predictor_decode_mono(APEContext *ctx, int count)
 
         /* Have we filled the history buffer? */
         if (p->buf == p->historybuffer + HISTORY_SIZE) {
-            memmove(p->historybuffer, p->buf, PREDICTOR_SIZE * sizeof(int32_t));
+            memmove(p->historybuffer, p->buf,
+                    PREDICTOR_SIZE * sizeof(*p->historybuffer));
             p->buf = p->historybuffer;
         }
 
@@ -668,8 +668,8 @@ static void do_init_filter(APEFilter *f, int16_t *buf, int order)
     f->delay       = f->historybuffer + order * 2;
     f->adaptcoeffs = f->historybuffer + order;
 
-    memset(f->historybuffer, 0, (order * 2) * sizeof(int16_t));
-    memset(f->coeffs, 0, order * sizeof(int16_t));
+    memset(f->historybuffer, 0, (order * 2) * sizeof(*f->historybuffer));
+    memset(f->coeffs, 0, order * sizeof(*f->coeffs));
     f->avg = 0;
 }
 
@@ -725,7 +725,7 @@ static void do_apply_filter(APEContext *ctx, int version, APEFilter *f,
         /* Have we filled the history buffer? */
         if (f->delay == f->historybuffer + HISTORY_SIZE + (order * 2)) {
             memmove(f->historybuffer, f->delay - (order * 2),
-                    (order * 2) * sizeof(int16_t));
+                    (order * 2) * sizeof(*f->historybuffer));
             f->delay = f->historybuffer + order * 2;
             f->adaptcoeffs = f->historybuffer + order;
         }
@@ -773,33 +773,29 @@ static int init_frame_decoder(APEContext *ctx)
 
 static void ape_unpack_mono(APEContext *ctx, int count)
 {
-    int32_t *decoded0 = ctx->decoded0;
-    int32_t *decoded1 = ctx->decoded1;
-
     if (ctx->frameflags & APE_FRAMECODE_STEREO_SILENCE) {
-        entropy_decode(ctx, count, 0);
         /* We are pure silence, so we're done. */
         av_log(ctx->avctx, AV_LOG_DEBUG, "pure silence mono\n");
         return;
     }
 
     entropy_decode(ctx, count, 0);
-    ape_apply_filters(ctx, decoded0, NULL, count);
+    ape_apply_filters(ctx, ctx->decoded[0], NULL, count);
 
     /* Now apply the predictor decoding */
     predictor_decode_mono(ctx, count);
 
     /* Pseudo-stereo - just copy left channel to right channel */
     if (ctx->channels == 2) {
-        memcpy(decoded1, decoded0, count * sizeof(*decoded1));
+        memcpy(ctx->decoded[1], ctx->decoded[0], count * sizeof(*ctx->decoded[1]));
     }
 }
 
 static void ape_unpack_stereo(APEContext *ctx, int count)
 {
     int32_t left, right;
-    int32_t *decoded0 = ctx->decoded0;
-    int32_t *decoded1 = ctx->decoded1;
+    int32_t *decoded0 = ctx->decoded[0];
+    int32_t *decoded1 = ctx->decoded[1];
 
     if (ctx->frameflags & APE_FRAMECODE_STEREO_SILENCE) {
         /* We are pure silence, so we're done. */
@@ -883,9 +879,6 @@ static int ape_decode_frame(AVCodecContext *avctx, void *data,
         }
         s->samples = nblocks;
 
-        memset(s->decoded0,  0, sizeof(s->decoded0));
-        memset(s->decoded1,  0, sizeof(s->decoded1));
-
         /* Initialize the frame decoder */
         if (init_frame_decoder(s) < 0) {
             av_log(avctx, AV_LOG_ERROR, "Error reading frame header\n");
@@ -900,7 +893,16 @@ static int ape_decode_frame(AVCodecContext *avctx, void *data,
         return avpkt->size;
     }
 
-    blockstodecode = FFMIN(BLOCKS_PER_LOOP, s->samples);
+    blockstodecode = FFMIN(s->blocks_per_loop, s->samples);
+
+    /* reallocate decoded sample buffer if needed */
+    av_fast_malloc(&s->decoded_buffer, &s->decoded_size,
+                   2 * FFALIGN(blockstodecode, 8) * sizeof(*s->decoded_buffer));
+    if (!s->decoded_buffer)
+        return AVERROR(ENOMEM);
+    memset(s->decoded_buffer, 0, s->decoded_size);
+    s->decoded[0] = s->decoded_buffer;
+    s->decoded[1] = s->decoded_buffer + FFALIGN(blockstodecode, 8);
 
     /* get output buffer */
     s->frame.nb_samples = blockstodecode;
@@ -927,25 +929,25 @@ static int ape_decode_frame(AVCodecContext *avctx, void *data,
     case 8:
         sample8 = (uint8_t *)s->frame.data[0];
         for (i = 0; i < blockstodecode; i++) {
-            *sample8++ = (s->decoded0[i] + 0x80) & 0xff;
+            *sample8++ = (s->decoded[0][i] + 0x80) & 0xff;
             if (s->channels == 2)
-                *sample8++ = (s->decoded1[i] + 0x80) & 0xff;
+                *sample8++ = (s->decoded[1][i] + 0x80) & 0xff;
         }
         break;
     case 16:
         sample16 = (int16_t *)s->frame.data[0];
         for (i = 0; i < blockstodecode; i++) {
-            *sample16++ = s->decoded0[i];
+            *sample16++ = s->decoded[0][i];
             if (s->channels == 2)
-                *sample16++ = s->decoded1[i];
+                *sample16++ = s->decoded[1][i];
         }
         break;
     case 24:
         sample24 = (int32_t *)s->frame.data[0];
         for (i = 0; i < blockstodecode; i++) {
-            *sample24++ = s->decoded0[i] << 8;
+            *sample24++ = s->decoded[0][i] << 8;
             if (s->channels == 2)
-                *sample24++ = s->decoded1[i] << 8;
+                *sample24++ = s->decoded[1][i] << 8;
         }
         break;
     }
@@ -964,6 +966,21 @@ static void ape_flush(AVCodecContext *avctx)
     s->samples= 0;
 }
 
+#define OFFSET(x) offsetof(APEContext, x)
+#define PAR (AV_OPT_FLAG_DECODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM)
+static const AVOption options[] = {
+    { "max_samples", "maximum number of samples decoded per call",             OFFSET(blocks_per_loop), AV_OPT_TYPE_INT,   { 4608 },    1,       INT_MAX, PAR, "max_samples" },
+    { "all",         "no maximum. decode all samples for each packet at once", 0,                       AV_OPT_TYPE_CONST, { INT_MAX }, INT_MIN, INT_MAX, PAR, "max_samples" },
+    { NULL},
+};
+
+static const AVClass ape_decoder_class = {
+    .class_name = "APE decoder",
+    .item_name  = av_default_item_name,
+    .option     = options,
+    .version    = LIBAVUTIL_VERSION_INT,
+};
+
 AVCodec ff_ape_decoder = {
     .name           = "ape",
     .type           = AVMEDIA_TYPE_AUDIO,
@@ -975,4 +992,5 @@ AVCodec ff_ape_decoder = {
     .capabilities   = CODEC_CAP_SUBFRAMES | CODEC_CAP_DELAY | CODEC_CAP_DR1,
     .flush = ape_flush,
     .long_name = NULL_IF_CONFIG_SMALL("Monkey's Audio"),
+    .priv_class     = &ape_decoder_class,
 };
diff --git a/libavcodec/bytestream.h b/libavcodec/bytestream.h
index 71c70aac84..d315e3f818 100644
--- a/libavcodec/bytestream.h
+++ b/libavcodec/bytestream.h
@@ -1,6 +1,7 @@
 /*
  * Bytestream functions
  * copyright (c) 2006 Baptiste Coudurier <baptiste.coudurier@free.fr>
+ * Copyright (c) 2012 Aneesh Dogra (lionaneesh) <lionaneesh@gmail.com>
  *
  * This file is part of FFmpeg.
  *
@@ -30,6 +31,11 @@ typedef struct {
     const uint8_t *buffer, *buffer_end, *buffer_start;
 } GetByteContext;
 
+typedef struct {
+    uint8_t *buffer, *buffer_end, *buffer_start;
+    int eof;
+} PutByteContext;
+
 #define DEF_T(type, name, bytes, read, write)                             \
 static av_always_inline type bytestream_get_ ## name(const uint8_t **b){\
     (*b) += bytes;\
@@ -39,6 +45,17 @@ static av_always_inline void bytestream_put_ ##name(uint8_t **b, const type valu
     write(*b, value);\
     (*b) += bytes;\
 }\
+static av_always_inline void bytestream2_put_ ## name ## u(PutByteContext *p, const type value)\
+{\
+    bytestream_put_ ## name(&p->buffer, value);\
+}\
+static av_always_inline void bytestream2_put_ ## name(PutByteContext *p, const type value){\
+    if (!p->eof && (p->buffer_end - p->buffer >= bytes)) {\
+        write(p->buffer, value);\
+        p->buffer += bytes;\
+    } else\
+        p->eof = 1;\
+}\
 static av_always_inline type bytestream2_get_ ## name ## u(GetByteContext *g)\
 {\
     return bytestream_get_ ## name(&g->buffer);\
@@ -119,22 +136,53 @@ static av_always_inline void bytestream2_init(GetByteContext *g,
     g->buffer_end = buf + buf_size;
 }
 
+static av_always_inline void bytestream2_init_writer(PutByteContext *p,
+                                                     uint8_t *buf, int buf_size)
+{
+    p->buffer       = buf;
+    p->buffer_start = buf;
+    p->buffer_end   = buf + buf_size;
+    p->eof          = 0;
+}
+
 static av_always_inline unsigned int bytestream2_get_bytes_left(GetByteContext *g)
 {
     return g->buffer_end - g->buffer;
 }
 
+static av_always_inline unsigned int bytestream2_get_bytes_left_p(PutByteContext *p)
+{
+    return p->buffer_end - p->buffer;
+}
+
 static av_always_inline void bytestream2_skip(GetByteContext *g,
                                               unsigned int size)
 {
     g->buffer += FFMIN(g->buffer_end - g->buffer, size);
 }
 
+static av_always_inline void bytestream2_skip_p(PutByteContext *p,
+                                                unsigned int size)
+{
+    int size2;
+    if (p->eof)
+        return;
+    size2 = FFMIN(p->buffer_end - p->buffer, size);
+    if (size2 != size)
+        p->eof = 1;
+    p->buffer += size2;
+}
+
 static av_always_inline int bytestream2_tell(GetByteContext *g)
 {
     return (int)(g->buffer - g->buffer_start);
 }
 
+static av_always_inline int bytestream2_tell_p(PutByteContext *p)
+{
+    return (int)(p->buffer - p->buffer_start);
+}
+
 static av_always_inline int bytestream2_seek(GetByteContext *g, int offset,
                                              int whence)
 {
@@ -158,6 +206,36 @@ static av_always_inline int bytestream2_seek(GetByteContext *g, int offset,
     return bytestream2_tell(g);
 }
 
+static av_always_inline int bytestream2_seek_p(PutByteContext *p, int offset,
+                                               int whence)
+{
+    p->eof = 0;
+    switch (whence) {
+    case SEEK_CUR:
+        if (p->buffer_end - p->buffer < offset)
+            p->eof = 1;
+        offset = av_clip(offset, -(p->buffer - p->buffer_start),
+                         p->buffer_end - p->buffer);
+        p->buffer += offset;
+        break;
+    case SEEK_END:
+        if (offset > 0)
+            p->eof = 1;
+        offset = av_clip(offset, -(p->buffer_end - p->buffer_start), 0);
+        p->buffer = p->buffer_end + offset;
+        break;
+    case SEEK_SET:
+        if (p->buffer_end - p->buffer_start < offset)
+            p->eof = 1;
+        offset = av_clip(offset, 0, p->buffer_end - p->buffer_start);
+        p->buffer = p->buffer_start + offset;
+        break;
+    default:
+        return AVERROR(EINVAL);
+    }
+    return bytestream2_tell_p(p);
+}
+
 static av_always_inline unsigned int bytestream2_get_buffer(GetByteContext *g,
                                                             uint8_t *dst,
                                                             unsigned int size)
@@ -168,6 +246,40 @@ static av_always_inline unsigned int bytestream2_get_buffer(GetByteContext *g,
     return size2;
 }
 
+static av_always_inline unsigned int bytestream2_put_buffer(PutByteContext *p,
+                                                            const uint8_t *src,
+                                                            unsigned int size)
+{
+    int size2;
+    if (p->eof)
+        return 0;
+    size2 = FFMIN(p->buffer_end - p->buffer, size);
+    if (size2 != size)
+        p->eof = 1;
+    memcpy(p->buffer, src, size2);
+    p->buffer += size2;
+    return size2;
+}
+
+static av_always_inline void bytestream2_set_buffer(PutByteContext *p,
+                                                    const uint8_t c,
+                                                    unsigned int size)
+{
+    int size2;
+    if (p->eof)
+        return;
+    size2 = FFMIN(p->buffer_end - p->buffer, size);
+    if (size2 != size)
+        p->eof = 1;
+    memset(p->buffer, c, size2);
+    p->buffer += size2;
+}
+
+static av_always_inline unsigned int bytestream2_get_eof(PutByteContext *p)
+{
+    return p->eof;
+}
+
 static av_always_inline unsigned int bytestream_get_buffer(const uint8_t **b, uint8_t *dst, unsigned int size)
 {
     memcpy(dst, *b, size);
diff --git a/libavcodec/vp8.c b/libavcodec/vp8.c
index 370fb0070c..4728393d10 100644
--- a/libavcodec/vp8.c
+++ b/libavcodec/vp8.c
@@ -1561,18 +1561,6 @@ static void release_queued_segmaps(VP8Context *s, int is_close)
     s->maps_are_invalid = 0;
 }
 
-/**
- * Sets things up for skipping the current frame.
- * In particular, removes it from the reference buffers.
- */
-static void skipframe_clear(VP8Context *s)
-{
-    s->invisible = 1;
-    s->next_framep[VP56_FRAME_CURRENT] = NULL;
-    if (s->update_last)
-        s->next_framep[VP56_FRAME_PREVIOUS] = NULL;
-}
-
 static int vp8_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
                             AVPacket *avpkt)
 {
@@ -1584,7 +1572,7 @@ static int vp8_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
     release_queued_segmaps(s, 0);
 
     if ((ret = decode_frame_header(s, avpkt->data, avpkt->size)) < 0)
-        return ret;
+        goto err;
 
     prev_frame = s->framep[VP56_FRAME_CURRENT];
 
@@ -1594,6 +1582,11 @@ static int vp8_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
     skip_thresh = !referenced ? AVDISCARD_NONREF :
                     !s->keyframe ? AVDISCARD_NONKEY : AVDISCARD_ALL;
 
+    if (avctx->skip_frame >= skip_thresh) {
+        s->invisible = 1;
+        memcpy(&s->next_framep[0], &s->framep[0], sizeof(s->framep[0]) * 4);
+        goto skip_decode;
+    }
     s->deblock_filter = s->filter.level && avctx->skip_loop_filter < skip_thresh;
 
     // release no longer referenced frames
@@ -1618,6 +1611,27 @@ static int vp8_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
         av_log(avctx, AV_LOG_FATAL, "Ran out of free frames!\n");
         abort();
     }
+    if (curframe->data[0])
+        vp8_release_frame(s, curframe, 1, 0);
+
+    // Given that arithmetic probabilities are updated every frame, it's quite likely
+    // that the values we have on a random interframe are complete junk if we didn't
+    // start decode on a keyframe. So just don't display anything rather than junk.
+    if (!s->keyframe && (!s->framep[VP56_FRAME_PREVIOUS] ||
+                         !s->framep[VP56_FRAME_GOLDEN] ||
+                         !s->framep[VP56_FRAME_GOLDEN2])) {
+        av_log(avctx, AV_LOG_WARNING, "Discarding interframe without a prior keyframe!\n");
+        ret = AVERROR_INVALIDDATA;
+        goto err;
+    }
+
+    curframe->key_frame = s->keyframe;
+    curframe->pict_type = s->keyframe ? AV_PICTURE_TYPE_I : AV_PICTURE_TYPE_P;
+    curframe->reference = referenced ? 3 : 0;
+    if ((ret = vp8_alloc_frame(s, curframe))) {
+        av_log(avctx, AV_LOG_ERROR, "get_buffer() failed!\n");
+        goto err;
+    }
 
     // check if golden and altref are swapped
     if (s->update_altref != VP56_FRAME_NONE) {
@@ -1637,36 +1651,6 @@ static int vp8_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
     }
     s->next_framep[VP56_FRAME_CURRENT]      = curframe;
 
-    if (avctx->skip_frame >= skip_thresh) {
-        skipframe_clear(s);
-        ret = avpkt->size;
-        goto skip_decode;
-    }
-
-    // Given that arithmetic probabilities are updated every frame, it's quite likely
-    // that the values we have on a random interframe are complete junk if we didn't
-    // start decode on a keyframe. So just don't display anything rather than junk.
-    if (!s->keyframe && (!s->framep[VP56_FRAME_PREVIOUS] ||
-                         !s->framep[VP56_FRAME_GOLDEN] ||
-                         !s->framep[VP56_FRAME_GOLDEN2])) {
-        av_log(avctx, AV_LOG_WARNING, "Discarding interframe without a prior keyframe!\n");
-        skipframe_clear(s);
-        ret = AVERROR_INVALIDDATA;
-        goto skip_decode;
-    }
-
-    if (curframe->data[0])
-        vp8_release_frame(s, curframe, 1, 0);
-
-    curframe->key_frame = s->keyframe;
-    curframe->pict_type = s->keyframe ? AV_PICTURE_TYPE_I : AV_PICTURE_TYPE_P;
-    curframe->reference = referenced ? 3 : 0;
-    if ((ret = vp8_alloc_frame(s, curframe))) {
-        av_log(avctx, AV_LOG_ERROR, "get_buffer() failed!\n");
-        skipframe_clear(s);
-        goto skip_decode;
-    }
-
     ff_thread_finish_setup(avctx);
 
     s->linesize   = curframe->linesize[0];
@@ -1778,20 +1762,22 @@ static int vp8_decode_frame(AVCodecContext *avctx, void *data, int *data_size,
     }
 
     ff_thread_report_progress(curframe, INT_MAX, 0);
-    ret = avpkt->size;
+    memcpy(&s->framep[0], &s->next_framep[0], sizeof(s->framep[0]) * 4);
+
 skip_decode:
     // if future frames don't use the updated probabilities,
     // reset them to the values we saved
     if (!s->update_probabilities)
         s->prob[0] = s->prob[1];
 
-    memcpy(&s->framep[0], &s->next_framep[0], sizeof(s->framep[0]) * 4);
-
     if (!s->invisible) {
         *(AVFrame*)data = *curframe;
         *data_size = sizeof(AVFrame);
     }
 
+    return avpkt->size;
+err:
+    memcpy(&s->next_framep[0], &s->framep[0], sizeof(s->framep[0]) * 4);
     return ret;
 }
 
diff --git a/libavcodec/x86/h264_idct_10bit.asm b/libavcodec/x86/h264_idct_10bit.asm
index f94207bb11..27a18f47da 100644
--- a/libavcodec/x86/h264_idct_10bit.asm
+++ b/libavcodec/x86/h264_idct_10bit.asm
@@ -315,7 +315,7 @@ IDCT_ADD16INTRA_10 avx
 ; h264_idct_add8(pixel **dst, const int *block_offset, dctcoef *block, int stride, const uint8_t nnzc[6*8])
 ;-----------------------------------------------------------------------------
 %macro IDCT_ADD8 1
-cglobal h264_idct_add8_10_%1,5,7
+cglobal h264_idct_add8_10_%1,5,7,7
 %if ARCH_X86_64
     mov r10, r0
 %endif