X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fvmdav.c;h=a1e39c058818c7d4bbcde8d2e545a0b9d3f2ca81;hb=a1c2b48018b09d2613f075ec0748c95bd520ac00;hp=79318b027766600098cf8b73c4c7e3262ce9d8d2;hpb=3aab27b4599911aad2fd9dc14cdc66414bf27eb3;p=ffmpeg

diff --git a/libavcodec/vmdav.c b/libavcodec/vmdav.c
index 79318b02776..a1e39c05881 100644
--- a/libavcodec/vmdav.c
+++ b/libavcodec/vmdav.c
@@ -2,25 +2,25 @@
  * Sierra VMD Audio & Video Decoders
  * Copyright (C) 2004 the ffmpeg project
  *
- * This file is part of FFmpeg.
+ * This file is part of Libav.
  *
- * FFmpeg is free software; you can redistribute it and/or
+ * Libav is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
  * version 2.1 of the License, or (at your option) any later version.
  *
- * FFmpeg is distributed in the hope that it will be useful,
+ * Libav is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  * Lesser General Public License for more details.
  *
  * You should have received a copy of the GNU Lesser General Public
- * License along with FFmpeg; if not, write to the Free Software
+ * License along with Libav; if not, write to the Free Software
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
 /**
- * @file libavcodec/vmdav.c
+ * @file
  * Sierra VMD audio & video decoders
  * by Vladimir "VAG" Gneushev (vagsoft at mail.ru)
  * for more information on the Sierra VMD format, visit:
@@ -43,8 +43,12 @@
 #include <stdlib.h>
 #include <string.h>
 
+#include "libavutil/channel_layout.h"
+#include "libavutil/common.h"
 #include "libavutil/intreadwrite.h"
 #include "avcodec.h"
+#include "internal.h"
+#include "bytestream.h"
 
 #define VMD_HEADER_SIZE 0x330
 #define PALETTE_COUNT 256
@@ -56,8 +60,7 @@
 typedef struct VmdVideoContext {
 
     AVCodecContext *avctx;
-    AVFrame frame;
-    AVFrame prev_frame;
+    AVFrame *prev_frame;
 
     const unsigned char *buf;
     int size;
@@ -72,9 +75,9 @@ typedef struct VmdVideoContext {
 #define QUEUE_SIZE 0x1000
 #define QUEUE_MASK 0x0FFF
 
-static void lz_unpack(const unsigned char *src, unsigned char *dest, int dest_len)
+static void lz_unpack(const unsigned char *src, int src_len,
+                      unsigned char *dest, int dest_len)
 {
-    const unsigned char *s;
     unsigned char *d;
     unsigned char *d_end;
     unsigned char queue[QUEUE_SIZE];
@@ -85,15 +88,17 @@ static void lz_unpack(const unsigned char *src, unsigned char *dest, int dest_le
     unsigned int speclen;
     unsigned char tag;
     unsigned int i, j;
+    GetByteContext gb;
 
-    s = src;
+    bytestream2_init(&gb, src, src_len);
     d = dest;
     d_end = d + dest_len;
-    dataleft = AV_RL32(s);
-    s += 4;
+    dataleft = bytestream2_get_le32(&gb);
     memset(queue, 0x20, QUEUE_SIZE);
-    if (AV_RL32(s) == 0x56781234) {
-        s += 4;
+    if (bytestream2_get_bytes_left(&gb) < 4)
+        return;
+    if (bytestream2_peek_le32(&gb) == 0x56781234) {
+        bytestream2_get_le32(&gb);
         qpos = 0x111;
         speclen = 0xF + 3;
     } else {
@@ -101,13 +106,13 @@ static void lz_unpack(const unsigned char *src, unsigned char *dest, int dest_le
         speclen = 100;  /* no speclen */
     }
 
-    while (dataleft > 0) {
-        tag = *s++;
+    while (dataleft > 0 && bytestream2_get_bytes_left(&gb) > 0) {
+        tag = bytestream2_get_byteu(&gb);
         if ((tag == 0xFF) && (dataleft > 8)) {
-            if (d + 8 > d_end)
+            if (d + 8 > d_end || bytestream2_get_bytes_left(&gb) < 8)
                 return;
             for (i = 0; i < 8; i++) {
-                queue[qpos++] = *d++ = *s++;
+                queue[qpos++] = *d++ = bytestream2_get_byteu(&gb);
                 qpos &= QUEUE_MASK;
             }
             dataleft -= 8;
@@ -116,17 +121,18 @@ static void lz_unpack(const unsigned char *src, unsigned char *dest, int dest_le
                 if (dataleft == 0)
                     break;
                 if (tag & 0x01) {
-                    if (d + 1 > d_end)
+                    if (d + 1 > d_end || bytestream2_get_bytes_left(&gb) < 1)
                         return;
-                    queue[qpos++] = *d++ = *s++;
+                    queue[qpos++] = *d++ = bytestream2_get_byte(&gb);
                     qpos &= QUEUE_MASK;
                     dataleft--;
                 } else {
-                    chainofs = *s++;
-                    chainofs |= ((*s & 0xF0) << 4);
-                    chainlen = (*s++ & 0x0F) + 3;
-                    if (chainlen == speclen)
-                        chainlen = *s++ + 0xF + 3;
+                    chainofs = bytestream2_get_byte(&gb);
+                    chainofs |= ((bytestream2_peek_byte(&gb) & 0xF0) << 4);
+                    chainlen = (bytestream2_get_byte(&gb) & 0x0F) + 3;
+                    if (chainlen == speclen) {
+                        chainlen = bytestream2_get_byte(&gb) + 0xF + 3;
+                    }
                     if (d + chainlen > d_end)
                         return;
                     for (j = 0; j < chainlen; j++) {
@@ -143,54 +149,57 @@ static void lz_unpack(const unsigned char *src, unsigned char *dest, int dest_le
 }
 
 static int rle_unpack(const unsigned char *src, unsigned char *dest,
-    int src_len, int dest_len)
+    int src_count, int src_size, int dest_len)
 {
-    const unsigned char *ps;
     unsigned char *pd;
-    int i, l;
+    int i, l, used = 0;
     unsigned char *dest_end = dest + dest_len;
+    GetByteContext gb;
+    uint16_t run_val;
 
-    ps = src;
+    bytestream2_init(&gb, src, src_size);
     pd = dest;
-    if (src_len & 1)
-        *pd++ = *ps++;
+    if (src_count & 1) {
+        if (bytestream2_get_bytes_left(&gb) < 1)
+            return 0;
+        *pd++ = bytestream2_get_byteu(&gb);
+        used++;
+    }
 
-    src_len >>= 1;
-    i = 0;
     do {
-        l = *ps++;
+        if (bytestream2_get_bytes_left(&gb) < 1)
+            break;
+        l = bytestream2_get_byteu(&gb);
         if (l & 0x80) {
             l = (l & 0x7F) * 2;
-            if (pd + l > dest_end)
-                return ps - src;
-            memcpy(pd, ps, l);
-            ps += l;
+            if (pd + l > dest_end || bytestream2_get_bytes_left(&gb) < l)
+                return bytestream2_tell(&gb);
+            bytestream2_get_buffer(&gb, pd, l);
             pd += l;
         } else {
-            if (pd + i > dest_end)
-                return ps - src;
+            if (pd + l > dest_end || bytestream2_get_bytes_left(&gb) < 2)
+                return bytestream2_tell(&gb);
+            run_val = bytestream2_get_ne16(&gb);
             for (i = 0; i < l; i++) {
-                *pd++ = ps[0];
-                *pd++ = ps[1];
+                AV_WN16(pd, run_val);
+                pd += 2;
             }
-            ps += 2;
+            l *= 2;
         }
-        i += l;
-    } while (i < src_len);
+        used += l;
+    } while (used < src_count);
 
-    return ps - src;
+    return bytestream2_tell(&gb);
 }
 
-static void vmd_decode(VmdVideoContext *s)
+static int vmd_decode(VmdVideoContext *s, AVFrame *frame)
 {
     int i;
     unsigned int *palette32;
     unsigned char r, g, b;
 
-    /* point to the start of the encoded data */
-    const unsigned char *p = s->buf + 16;
+    GetByteContext gb;
 
-    const unsigned char *pb;
     unsigned char meth;
     unsigned char *dp;   /* pointer to current frame */
     unsigned char *pp;   /* pointer to previous frame */
@@ -199,12 +208,29 @@ static void vmd_decode(VmdVideoContext *s)
 
     int frame_x, frame_y;
     int frame_width, frame_height;
-    int dp_size;
 
     frame_x = AV_RL16(&s->buf[6]);
     frame_y = AV_RL16(&s->buf[8]);
     frame_width = AV_RL16(&s->buf[10]) - frame_x + 1;
     frame_height = AV_RL16(&s->buf[12]) - frame_y + 1;
+    if (frame_x < 0 || frame_width < 0 ||
+        frame_x >= s->avctx->width ||
+        frame_width > s->avctx->width ||
+        frame_x + frame_width > s->avctx->width) {
+        av_log(s->avctx, AV_LOG_ERROR,
+               "Invalid horizontal range %d-%d\n",
+               frame_x, frame_width);
+        return AVERROR_INVALIDDATA;
+    }
+    if (frame_y < 0 || frame_height < 0 ||
+        frame_y >= s->avctx->height ||
+        frame_height > s->avctx->height ||
+        frame_y + frame_height > s->avctx->height) {
+        av_log(s->avctx, AV_LOG_ERROR,
+               "Invalid vertical range %d-%d\n",
+               frame_x, frame_width);
+        return AVERROR_INVALIDDATA;
+    }
 
     if ((frame_width == s->avctx->width && frame_height == s->avctx->height) &&
         (frame_x || frame_y)) {
@@ -217,109 +243,143 @@ static void vmd_decode(VmdVideoContext *s)
 
     /* if only a certain region will be updated, copy the entire previous
      * frame before the decode */
-    if (frame_x || frame_y || (frame_width != s->avctx->width) ||
-        (frame_height != s->avctx->height)) {
+    if (s->prev_frame->data[0] &&
+        (frame_x || frame_y || (frame_width != s->avctx->width) ||
+        (frame_height != s->avctx->height))) {
 
-        memcpy(s->frame.data[0], s->prev_frame.data[0],
-            s->avctx->height * s->frame.linesize[0]);
+        memcpy(frame->data[0], s->prev_frame->data[0],
+            s->avctx->height * frame->linesize[0]);
     }
 
     /* check if there is a new palette */
+    bytestream2_init(&gb, s->buf + 16, s->size - 16);
     if (s->buf[15] & 0x02) {
-        p += 2;
+        bytestream2_skip(&gb, 2);
         palette32 = (unsigned int *)s->palette;
-        for (i = 0; i < PALETTE_COUNT; i++) {
-            r = *p++ * 4;
-            g = *p++ * 4;
-            b = *p++ * 4;
-            palette32[i] = (r << 16) | (g << 8) | (b);
+        if (bytestream2_get_bytes_left(&gb) >= PALETTE_COUNT * 3) {
+            for (i = 0; i < PALETTE_COUNT; i++) {
+                r = bytestream2_get_byteu(&gb) * 4;
+                g = bytestream2_get_byteu(&gb) * 4;
+                b = bytestream2_get_byteu(&gb) * 4;
+                palette32[i] = (r << 16) | (g << 8) | (b);
+            }
+        } else {
+            av_log(s->avctx, AV_LOG_ERROR, "Incomplete palette\n");
+            return AVERROR_INVALIDDATA;
         }
-        s->size -= (256 * 3 + 2);
+        s->size -= PALETTE_COUNT * 3 + 2;
     }
-    if (s->size >= 0) {
-        /* originally UnpackFrame in VAG's code */
-        pb = p;
-        meth = *pb++;
-        if (meth & 0x80) {
-            lz_unpack(pb, s->unpack_buffer, s->unpack_buffer_size);
-            meth &= 0x7F;
-            pb = s->unpack_buffer;
+
+    if (!s->size)
+        return 0;
+
+    /* originally UnpackFrame in VAG's code */
+    if (bytestream2_get_bytes_left(&gb) < 1)
+        return AVERROR_INVALIDDATA;
+    meth = bytestream2_get_byteu(&gb);
+    if (meth & 0x80) {
+        if (!s->unpack_buffer_size) {
+            av_log(s->avctx, AV_LOG_ERROR,
+                   "Trying to unpack LZ-compressed frame with no LZ buffer\n");
+            return AVERROR_INVALIDDATA;
         }
+        lz_unpack(gb.buffer, bytestream2_get_bytes_left(&gb),
+                  s->unpack_buffer, s->unpack_buffer_size);
+        meth &= 0x7F;
+        bytestream2_init(&gb, s->unpack_buffer, s->unpack_buffer_size);
+    }
 
-        dp = &s->frame.data[0][frame_y * s->frame.linesize[0] + frame_x];
-        dp_size = s->frame.linesize[0] * s->avctx->height;
-        pp = &s->prev_frame.data[0][frame_y * s->prev_frame.linesize[0] + frame_x];
-        switch (meth) {
-        case 1:
-            for (i = 0; i < frame_height; i++) {
-                ofs = 0;
-                do {
-                    len = *pb++;
-                    if (len & 0x80) {
-                        len = (len & 0x7F) + 1;
-                        if (ofs + len > frame_width)
-                            return;
-                        memcpy(&dp[ofs], pb, len);
-                        pb += len;
-                        ofs += len;
-                    } else {
-                        /* interframe pixel copy */
-                        if (ofs + len + 1 > frame_width)
-                            return;
-                        memcpy(&dp[ofs], &pp[ofs], len + 1);
-                        ofs += len + 1;
-                    }
-                } while (ofs < frame_width);
-                if (ofs > frame_width) {
-                    av_log(s->avctx, AV_LOG_ERROR, "VMD video: offset > width (%d > %d)\n",
-                        ofs, frame_width);
-                    break;
+    dp = &frame->data[0][frame_y * frame->linesize[0] + frame_x];
+    pp = &s->prev_frame->data[0][frame_y * s->prev_frame->linesize[0] + frame_x];
+    switch (meth) {
+    case 1:
+        for (i = 0; i < frame_height; i++) {
+            ofs = 0;
+            do {
+                len = bytestream2_get_byte(&gb);
+                if (len & 0x80) {
+                    len = (len & 0x7F) + 1;
+                    if (ofs + len > frame_width ||
+                        bytestream2_get_bytes_left(&gb) < len)
+                        return AVERROR_INVALIDDATA;
+                    bytestream2_get_buffer(&gb, &dp[ofs], len);
+                    ofs += len;
+                } else {
+                    /* interframe pixel copy */
+                    if (ofs + len + 1 > frame_width || !s->prev_frame->data[0])
+                        return AVERROR_INVALIDDATA;
+                    memcpy(&dp[ofs], &pp[ofs], len + 1);
+                    ofs += len + 1;
                 }
-                dp += s->frame.linesize[0];
-                pp += s->prev_frame.linesize[0];
-            }
-            break;
-
-        case 2:
-            for (i = 0; i < frame_height; i++) {
-                memcpy(dp, pb, frame_width);
-                pb += frame_width;
-                dp += s->frame.linesize[0];
-                pp += s->prev_frame.linesize[0];
+            } while (ofs < frame_width);
+            if (ofs > frame_width) {
+                av_log(s->avctx, AV_LOG_ERROR,
+                       "VMD video: offset > width (%d > %d)\n",
+                       ofs, frame_width);
+                return AVERROR_INVALIDDATA;
             }
-            break;
+            dp += frame->linesize[0];
+            pp += s->prev_frame->linesize[0];
+        }
+        break;
 
-        case 3:
-            for (i = 0; i < frame_height; i++) {
-                ofs = 0;
-                do {
-                    len = *pb++;
-                    if (len & 0x80) {
-                        len = (len & 0x7F) + 1;
-                        if (*pb++ == 0xFF)
-                            len = rle_unpack(pb, &dp[ofs], len, frame_width - ofs);
-                        else
-                            memcpy(&dp[ofs], pb, len);
-                        pb += len;
-                        ofs += len;
+    case 2:
+        for (i = 0; i < frame_height; i++) {
+            bytestream2_get_buffer(&gb, dp, frame_width);
+            dp += frame->linesize[0];
+            pp += s->prev_frame->linesize[0];
+        }
+        break;
+
+    case 3:
+        for (i = 0; i < frame_height; i++) {
+            ofs = 0;
+            do {
+                len = bytestream2_get_byte(&gb);
+                if (len & 0x80) {
+                    len = (len & 0x7F) + 1;
+                    if (bytestream2_peek_byte(&gb) == 0xFF) {
+                        int slen = len;
+                        bytestream2_get_byte(&gb);
+                        len = rle_unpack(gb.buffer, &dp[ofs],
+                                         len, bytestream2_get_bytes_left(&gb),
+                                         frame_width - ofs);
+                        ofs += slen;
+                        bytestream2_skip(&gb, len);
                     } else {
-                        /* interframe pixel copy */
-                        if (ofs + len + 1 > frame_width)
-                            return;
-                        memcpy(&dp[ofs], &pp[ofs], len + 1);
-                        ofs += len + 1;
+                        bytestream2_get_buffer(&gb, &dp[ofs], len);
+                        ofs += len;
                     }
-                } while (ofs < frame_width);
-                if (ofs > frame_width) {
-                    av_log(s->avctx, AV_LOG_ERROR, "VMD video: offset > width (%d > %d)\n",
-                        ofs, frame_width);
+                } else {
+                    /* interframe pixel copy */
+                    if (ofs + len + 1 > frame_width || !s->prev_frame->data[0])
+                        return AVERROR_INVALIDDATA;
+                    memcpy(&dp[ofs], &pp[ofs], len + 1);
+                    ofs += len + 1;
                 }
-                dp += s->frame.linesize[0];
-                pp += s->prev_frame.linesize[0];
+            } while (ofs < frame_width);
+            if (ofs > frame_width) {
+                av_log(s->avctx, AV_LOG_ERROR,
+                       "VMD video: offset > width (%d > %d)\n",
+                       ofs, frame_width);
+                return AVERROR_INVALIDDATA;
             }
-            break;
+            dp += frame->linesize[0];
+            pp += s->prev_frame->linesize[0];
         }
+        break;
     }
+    return 0;
+}
+
+static av_cold int vmdvideo_decode_end(AVCodecContext *avctx)
+{
+    VmdVideoContext *s = avctx->priv_data;
+
+    av_frame_free(&s->prev_frame);
+    av_free(s->unpack_buffer);
+
+    return 0;
 }
 
 static av_cold int vmdvideo_decode_init(AVCodecContext *avctx)
@@ -333,7 +393,7 @@ static av_cold int vmdvideo_decode_init(AVCodecContext *avctx)
     unsigned char *raw_palette;
 
     s->avctx = avctx;
-    avctx->pix_fmt = PIX_FMT_PAL8;
+    avctx->pix_fmt = AV_PIX_FMT_PAL8;
 
     /* make sure the VMD header made it */
     if (s->avctx->extradata_size != VMD_HEADER_SIZE) {
@@ -344,9 +404,11 @@ static av_cold int vmdvideo_decode_init(AVCodecContext *avctx)
     vmd_header = (unsigned char *)avctx->extradata;
 
     s->unpack_buffer_size = AV_RL32(&vmd_header[800]);
-    s->unpack_buffer = av_malloc(s->unpack_buffer_size);
-    if (!s->unpack_buffer)
-        return -1;
+    if (s->unpack_buffer_size) {
+        s->unpack_buffer = av_malloc(s->unpack_buffer_size);
+        if (!s->unpack_buffer)
+            return AVERROR(ENOMEM);
+    }
 
     /* load up the initial palette */
     raw_palette = &vmd_header[28];
@@ -358,68 +420,65 @@ static av_cold int vmdvideo_decode_init(AVCodecContext *avctx)
         palette32[i] = (r << 16) | (g << 8) | (b);
     }
 
+    s->prev_frame = av_frame_alloc();
+    if (!s->prev_frame) {
+        vmdvideo_decode_end(avctx);
+        return AVERROR(ENOMEM);
+    }
+
     return 0;
 }
 
 static int vmdvideo_decode_frame(AVCodecContext *avctx,
-                                 void *data, int *data_size,
+                                 void *data, int *got_frame,
                                  AVPacket *avpkt)
 {
     const uint8_t *buf = avpkt->data;
     int buf_size = avpkt->size;
     VmdVideoContext *s = avctx->priv_data;
+    AVFrame *frame = data;
+    int ret;
 
     s->buf = buf;
     s->size = buf_size;
 
     if (buf_size < 16)
-        return buf_size;
+        return AVERROR_INVALIDDATA;
 
-    s->frame.reference = 1;
-    if (avctx->get_buffer(avctx, &s->frame)) {
+    if ((ret = ff_get_buffer(avctx, frame, AV_GET_BUFFER_FLAG_REF)) < 0) {
         av_log(s->avctx, AV_LOG_ERROR, "VMD Video: get_buffer() failed\n");
-        return -1;
+        return ret;
     }
 
-    vmd_decode(s);
+    if ((ret = vmd_decode(s, frame)) < 0)
+        return ret;
 
     /* make the palette available on the way out */
-    memcpy(s->frame.data[1], s->palette, PALETTE_COUNT * 4);
+    memcpy(frame->data[1], s->palette, PALETTE_COUNT * 4);
 
     /* shuffle frames */
-    FFSWAP(AVFrame, s->frame, s->prev_frame);
-    if (s->frame.data[0])
-        avctx->release_buffer(avctx, &s->frame);
+    av_frame_unref(s->prev_frame);
+    if ((ret = av_frame_ref(s->prev_frame, frame)) < 0)
+        return ret;
 
-    *data_size = sizeof(AVFrame);
-    *(AVFrame*)data = s->prev_frame;
+    *got_frame      = 1;
 
     /* report that the buffer was completely consumed */
     return buf_size;
 }
 
-static av_cold int vmdvideo_decode_end(AVCodecContext *avctx)
-{
-    VmdVideoContext *s = avctx->priv_data;
-
-    if (s->prev_frame.data[0])
-        avctx->release_buffer(avctx, &s->prev_frame);
-    av_free(s->unpack_buffer);
-
-    return 0;
-}
-
 
 /*
  * Audio Decoder
  */
 
+#define BLOCK_TYPE_AUDIO    1
+#define BLOCK_TYPE_INITIAL  2
+#define BLOCK_TYPE_SILENCE  3
+
 typedef struct VmdAudioContext {
-    AVCodecContext *avctx;
-    int channels;
-    int bits;
-    int block_align;
-    int predictors[2];
+    int out_bps;
+    int chunk_size;
 } VmdAudioContext;
 
 static const uint16_t vmdaudio_table[128] = {
@@ -442,122 +501,155 @@ static av_cold int vmdaudio_decode_init(AVCodecContext *avctx)
 {
     VmdAudioContext *s = avctx->priv_data;
 
-    s->avctx = avctx;
-    s->channels = avctx->channels;
-    s->bits = avctx->bits_per_coded_sample;
-    s->block_align = avctx->block_align;
-    avctx->sample_fmt = SAMPLE_FMT_S16;
+    if (avctx->channels < 1 || avctx->channels > 2) {
+        av_log(avctx, AV_LOG_ERROR, "invalid number of channels\n");
+        return AVERROR(EINVAL);
+    }
+    if (avctx->block_align < 1) {
+        av_log(avctx, AV_LOG_ERROR, "invalid block align\n");
+        return AVERROR(EINVAL);
+    }
+
+    avctx->channel_layout = avctx->channels == 1 ? AV_CH_LAYOUT_MONO :
+                                                   AV_CH_LAYOUT_STEREO;
 
-    av_log(s->avctx, AV_LOG_DEBUG, "%d channels, %d bits/sample, block align = %d, sample rate = %d\n",
-            s->channels, s->bits, s->block_align, avctx->sample_rate);
+    if (avctx->bits_per_coded_sample == 16)
+        avctx->sample_fmt = AV_SAMPLE_FMT_S16;
+    else
+        avctx->sample_fmt = AV_SAMPLE_FMT_U8;
+    s->out_bps = av_get_bytes_per_sample(avctx->sample_fmt);
+
+    s->chunk_size = avctx->block_align + avctx->channels * (s->out_bps == 2);
+
+    av_log(avctx, AV_LOG_DEBUG, "%d channels, %d bits/sample, "
+           "block align = %d, sample rate = %d\n",
+           avctx->channels, avctx->bits_per_coded_sample, avctx->block_align,
+           avctx->sample_rate);
 
     return 0;
 }
 
-static void vmdaudio_decode_audio(VmdAudioContext *s, unsigned char *data,
-    const uint8_t *buf, int buf_size, int stereo)
+static void decode_audio_s16(int16_t *out, const uint8_t *buf, int buf_size,
+                             int channels)
 {
-    int i;
-    int chan = 0;
-    int16_t *out = (int16_t*)data;
+    int ch;
+    const uint8_t *buf_end = buf + buf_size;
+    int predictor[2];
+    int st = channels - 1;
+
+    /* decode initial raw sample */
+    for (ch = 0; ch < channels; ch++) {
+        predictor[ch] = (int16_t)AV_RL16(buf);
+        buf += 2;
+        *out++ = predictor[ch];
+    }
 
-    for(i = 0; i < buf_size; i++) {
-        if(buf[i] & 0x80)
-            s->predictors[chan] -= vmdaudio_table[buf[i] & 0x7F];
+    /* decode DPCM samples */
+    ch = 0;
+    while (buf < buf_end) {
+        uint8_t b = *buf++;
+        if (b & 0x80)
+            predictor[ch] -= vmdaudio_table[b & 0x7F];
         else
-            s->predictors[chan] += vmdaudio_table[buf[i]];
-        s->predictors[chan] = av_clip_int16(s->predictors[chan]);
-        out[i] = s->predictors[chan];
-        chan ^= stereo;
+            predictor[ch] += vmdaudio_table[b];
+        predictor[ch] = av_clip_int16(predictor[ch]);
+        *out++ = predictor[ch];
+        ch ^= st;
     }
 }
 
-static int vmdaudio_loadsound(VmdAudioContext *s, unsigned char *data,
-    const uint8_t *buf, int silence, int data_size)
+static int vmdaudio_decode_frame(AVCodecContext *avctx, void *data,
+                                 int *got_frame_ptr, AVPacket *avpkt)
 {
-    int bytes_decoded = 0;
-    int i;
+    AVFrame *frame     = data;
+    const uint8_t *buf = avpkt->data;
+    const uint8_t *buf_end;
+    int buf_size = avpkt->size;
+    VmdAudioContext *s = avctx->priv_data;
+    int block_type, silent_chunks, audio_chunks;
+    int ret;
+    uint8_t *output_samples_u8;
+    int16_t *output_samples_s16;
+
+    if (buf_size < 16) {
+        av_log(avctx, AV_LOG_WARNING, "skipping small junk packet\n");
+        *got_frame_ptr = 0;
+        return buf_size;
+    }
+
+    block_type = buf[6];
+    if (block_type < BLOCK_TYPE_AUDIO || block_type > BLOCK_TYPE_SILENCE) {
+        av_log(avctx, AV_LOG_ERROR, "unknown block type: %d\n", block_type);
+        return AVERROR(EINVAL);
+    }
+    buf      += 16;
+    buf_size -= 16;
+
+    /* get number of silent chunks */
+    silent_chunks = 0;
+    if (block_type == BLOCK_TYPE_INITIAL) {
+        uint32_t flags;
+        if (buf_size < 4) {
+            av_log(avctx, AV_LOG_ERROR, "packet is too small\n");
+            return AVERROR(EINVAL);
+        }
+        flags         = AV_RB32(buf);
+        silent_chunks = av_popcount(flags);
+        buf      += 4;
+        buf_size -= 4;
+    } else if (block_type == BLOCK_TYPE_SILENCE) {
+        silent_chunks = 1;
+        buf_size = 0; // should already be zero but set it just to be sure
+    }
 
-//    if (silence)
-//        av_log(s->avctx, AV_LOG_INFO, "silent block!\n");
-    if (s->channels == 2) {
+    /* ensure output buffer is large enough */
+    audio_chunks = buf_size / s->chunk_size;
 
-        /* stereo handling */
-        if (silence) {
-            memset(data, 0, data_size * 2);
+    /* drop incomplete chunks */
+    buf_size     = audio_chunks * s->chunk_size;
+
+    /* get output buffer */
+    frame->nb_samples = ((silent_chunks + audio_chunks) * avctx->block_align) /
+                        avctx->channels;
+    if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) {
+        av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
+        return ret;
+    }
+    output_samples_u8  =            frame->data[0];
+    output_samples_s16 = (int16_t *)frame->data[0];
+
+    /* decode silent chunks */
+    if (silent_chunks > 0) {
+        int silent_size = FFMIN(avctx->block_align * silent_chunks,
+                                frame->nb_samples * avctx->channels);
+        if (s->out_bps == 2) {
+            memset(output_samples_s16, 0x00, silent_size * 2);
+            output_samples_s16 += silent_size;
         } else {
-            if (s->bits == 16)
-                vmdaudio_decode_audio(s, data, buf, data_size, 1);
-            else {
-                /* copy the data but convert it to signed */
-                for (i = 0; i < data_size; i++){
-                    *data++ = buf[i] + 0x80;
-                    *data++ = buf[i] + 0x80;
-                }
-            }
+            memset(output_samples_u8,  0x80, silent_size);
+            output_samples_u8 += silent_size;
         }
-    } else {
-        bytes_decoded = data_size * 2;
+    }
 
-        /* mono handling */
-        if (silence) {
-            memset(data, 0, data_size * 2);
-        } else {
-            if (s->bits == 16) {
-                vmdaudio_decode_audio(s, data, buf, data_size, 0);
+    /* decode audio chunks */
+    if (audio_chunks > 0) {
+        buf_end = buf + (buf_size & ~(avctx->channels > 1));
+        while (buf + s->chunk_size <= buf_end) {
+            if (s->out_bps == 2) {
+                decode_audio_s16(output_samples_s16, buf, s->chunk_size,
+                                 avctx->channels);
+                output_samples_s16 += avctx->block_align;
             } else {
-                /* copy the data but convert it to signed */
-                for (i = 0; i < data_size; i++){
-                    *data++ = buf[i] + 0x80;
-                    *data++ = buf[i] + 0x80;
-                }
+                memcpy(output_samples_u8, buf, s->chunk_size);
+                output_samples_u8  += avctx->block_align;
             }
+            buf += s->chunk_size;
         }
     }
 
-    return data_size * 2;
-}
+    *got_frame_ptr = 1;
 
-static int vmdaudio_decode_frame(AVCodecContext *avctx,
-                                 void *data, int *data_size,
-                                 AVPacket *avpkt)
-{
-    const uint8_t *buf = avpkt->data;
-    int buf_size = avpkt->size;
-    VmdAudioContext *s = avctx->priv_data;
-    unsigned char *output_samples = (unsigned char *)data;
-
-    /* point to the start of the encoded data */
-    const unsigned char *p = buf + 16;
-
-    if (buf_size < 16)
-        return buf_size;
-
-    if (buf[6] == 1) {
-        /* the chunk contains audio */
-        *data_size = vmdaudio_loadsound(s, output_samples, p, 0, buf_size - 16);
-    } else if (buf[6] == 2) {
-        /* initial chunk, may contain audio and silence */
-        uint32_t flags = AV_RB32(p);
-        int raw_block_size = s->block_align * s->bits / 8;
-        int silent_chunks;
-        if(flags == 0xFFFFFFFF)
-            silent_chunks = 32;
-        else
-            silent_chunks = av_log2(flags + 1);
-        if(*data_size < (s->block_align*silent_chunks + buf_size - 20) * 2)
-            return -1;
-        *data_size = 0;
-        memset(output_samples, 0, raw_block_size * silent_chunks);
-        output_samples += raw_block_size * silent_chunks;
-        *data_size = raw_block_size * silent_chunks;
-        *data_size += vmdaudio_loadsound(s, output_samples, p + 4, 0, buf_size - 20);
-    } else if (buf[6] == 3) {
-        /* silent chunk */
-        *data_size = vmdaudio_loadsound(s, output_samples, p, 1, 0);
-    }
-
-    return buf_size;
+    return avpkt->size;
 }
 
 
@@ -565,27 +657,25 @@ static int vmdaudio_decode_frame(AVCodecContext *avctx,
  * Public Data Structures
  */
 
-AVCodec vmdvideo_decoder = {
-    "vmdvideo",
-    CODEC_TYPE_VIDEO,
-    CODEC_ID_VMDVIDEO,
-    sizeof(VmdVideoContext),
-    vmdvideo_decode_init,
-    NULL,
-    vmdvideo_decode_end,
-    vmdvideo_decode_frame,
-    CODEC_CAP_DR1,
-    .long_name = NULL_IF_CONFIG_SMALL("Sierra VMD video"),
+AVCodec ff_vmdvideo_decoder = {
+    .name           = "vmdvideo",
+    .long_name      = NULL_IF_CONFIG_SMALL("Sierra VMD video"),
+    .type           = AVMEDIA_TYPE_VIDEO,
+    .id             = AV_CODEC_ID_VMDVIDEO,
+    .priv_data_size = sizeof(VmdVideoContext),
+    .init           = vmdvideo_decode_init,
+    .close          = vmdvideo_decode_end,
+    .decode         = vmdvideo_decode_frame,
+    .capabilities   = CODEC_CAP_DR1,
 };
 
-AVCodec vmdaudio_decoder = {
-    "vmdaudio",
-    CODEC_TYPE_AUDIO,
-    CODEC_ID_VMDAUDIO,
-    sizeof(VmdAudioContext),
-    vmdaudio_decode_init,
-    NULL,
-    NULL,
-    vmdaudio_decode_frame,
-    .long_name = NULL_IF_CONFIG_SMALL("Sierra VMD audio"),
+AVCodec ff_vmdaudio_decoder = {
+    .name           = "vmdaudio",
+    .long_name      = NULL_IF_CONFIG_SMALL("Sierra VMD audio"),
+    .type           = AVMEDIA_TYPE_AUDIO,
+    .id             = AV_CODEC_ID_VMDAUDIO,
+    .priv_data_size = sizeof(VmdAudioContext),
+    .init           = vmdaudio_decode_init,
+    .decode         = vmdaudio_decode_frame,
+    .capabilities   = CODEC_CAP_DR1,
 };