查看avcodec_decode_video2()函数声明发现,该函数已经过期了,取代它的是avcodec_send_packet、avcodec_receive_frame。我们发现新版的avcodec_decode_video2()最终还是调用了avcodec_send_packet、avcodec_receive_frame。如下所示。
avcodec_decode_video2->compat_decode->avcodec_send_packet、avcodec_receive_frame。

一、avcodec_send_packet()

我们先看avcodec_send_packet()函数,avcodec_send_packet()申明在文件libavcodec\avcodec.h中,如下所示。

/**
 * Supply raw packet data as input to a decoder.
 *
 * Internally, this call will copy relevant AVCodecContext fields, which can
 * influence decoding per-packet, and apply them when the packet is actually
 * decoded. (For example AVCodecContext.skip_frame, which might direct the
 * decoder to drop the frame contained by the packet sent with this function.)
 *
 * @warning The input buffer, avpkt->data must be AV_INPUT_BUFFER_PADDING_SIZE
 *          larger than the actual read bytes because some optimized bitstream
 *          readers read 32 or 64 bits at once and could read over the end.
 *
 * @warning Do not mix this API with the legacy API (like avcodec_decode_video2())
 *          on the same AVCodecContext. It will return unexpected results now
 *          or in future libavcodec versions.
 *
 * @note The AVCodecContext MUST have been opened with @ref avcodec_open2()
 *       before packets may be fed to the decoder.
 *
 * @param avctx codec context
 * @param[in] avpkt The input AVPacket. Usually, this will be a single video
 *                  frame, or several complete audio frames.
 *                  Ownership of the packet remains with the caller, and the
 *                  decoder will not write to the packet. The decoder may create
 *                  a reference to the packet data (or copy it if the packet is
 *                  not reference-counted).
 *                  Unlike with older APIs, the packet is always fully consumed,
 *                  and if it contains multiple frames (e.g. some audio codecs),
 *                  will require you to call avcodec_receive_frame() multiple
 *                  times afterwards before you can send a new packet.
 *                  It can be NULL (or an AVPacket with data set to NULL and
 *                  size set to 0); in this case, it is considered a flush
 *                  packet, which signals the end of the stream. Sending the
 *                  first flush packet will return success. Subsequent ones are
 *                  unnecessary and will return AVERROR_EOF. If the decoder
 *                  still has frames buffered, it will return them after sending
 *                  a flush packet.
 *
 * @return 0 on success, otherwise negative error code:
 *      AVERROR(EAGAIN):   input is not accepted in the current state - user
 *                         must read output with avcodec_receive_frame() (once
 *                         all output is read, the packet should be resent, and
 *                         the call will not fail with EAGAIN).
 *      AVERROR_EOF:       the decoder has been flushed, and no new packets can
 *                         be sent to it (also returned if more than 1 flush
 *                         packet is sent)
 *      AVERROR(EINVAL):   codec not opened, it is an encoder, or requires flush
 *      AVERROR(ENOMEM):   failed to add packet to internal queue, or similar
 *      other errors: legitimate decoding errors
 */
int avcodec_send_packet(AVCodecContext *avctx, const AVPacket *avpkt);

函数的英文功能介绍有很多,本文简单翻译了一下,如下。
@ 功能:支持将裸流数据输出给解码器
在函数内部,会拷贝相关的AVCodecContext结构变量,将这些结构变量应用到解码的每一个包。例如
AVCodecContext.skip_frame参数通知解码器扔掉包含该帧的包。
@ warning 输入的avpkt-data缓冲区必须大于AV_INPUT_PADDING_SIZE,因为优化的字节流读取器必须一次读取32或者64比特的数据
@ warning 不能跟之前的API(例如avcodec_decode_video2)混用,否则会返回不可预知的错误
@ note 在将包发送给解码器的时候,AVCodecContext必须已经通过avcodec_open2打开
@ 参数 avctx 解码上下文
@ 参数[输入]avpkt 输入AVPakcet.通常情况下,输入数据是一个单一的视频帧或者几个完整的音频帧。调用者保留包的原有属性,解码器不会修改包的内容。解码器可能创建对包的引用。如果包没有引用计数将拷贝一份。跟以往的API不一样,包必须完整的解码出来,如果包含有多个帧,要求多次调用avcodec_recvive_frame,在重新调用该接口之前。输入参数可以为NULL,或者AVPacket的data域设置为NULL或者size域设置为0,表示将刷新所有的包,意味着数据流已经结束了。第一次发送刷新会总会成功,第二次发送刷新包是没有必要的,并且返回AVERROR_EOF,如果解码器缓存了一些帧,返回一个刷新包,将会返回所有的解码包
@ 返回值 0 表示成功,其他的异常值说明:
AVERROR(EAGAIN):当前不接受输出,必须重新发送
AVERROR_EOF:已经刷新解码器,没有新的包可以被刷新
AVERROR(EINVAL):没有打开解码器,或者这是一个编码器,或者要求刷新
AVERRO(ENOMEN):无法添加包到内部队列
avcodec_send_packet()函数定义在文件libavcodec\decode.c中。如下所示

int attribute_align_arg avcodec_send_packet(AVCodecContext *avctx, const AVPacket *avpkt)
{
    AVCodecInternal *avci = avctx->internal;
    int ret;

    if (!avcodec_is_open(avctx) || !av_codec_is_decoder(avctx->codec))
        return AVERROR(EINVAL);

    if (avctx->internal->draining)
        return AVERROR_EOF;

    if (avpkt && !avpkt->size && avpkt->data)
        return AVERROR(EINVAL);

    av_packet_unref(avci->buffer_pkt);
    if (avpkt && (avpkt->data || avpkt->side_data_elems)) {
        ret = av_packet_ref(avci->buffer_pkt, avpkt);
        if (ret < 0)
            return ret;
    }

    ret = av_bsf_send_packet(avci->filter.bsfs[0], avci->buffer_pkt);
    if (ret < 0) {
        av_packet_unref(avci->buffer_pkt);
        return ret;
    }

    if (!avci->buffer_frame->buf[0]) {
        ret = decode_receive_frame_internal(avctx, avci->buffer_frame);
        if (ret < 0 && ret != AVERROR(EAGAIN) && ret != AVERROR_EOF)
            return ret;
    }

    return 0;
}

      可以看到,avcodec_send_packet 的关键是 decode_receive_frame_internal 这个函数。av_bsf_send_packet等结构体或函数主要用于存储 AVPacket。

1.1 av_bsf_send_packet()

int av_bsf_send_packet(AVBSFContext *ctx, AVPacket *pkt)
{
    int ret;

    if (!pkt || (!pkt->data && !pkt->side_data_elems)) {
        ctx->internal->eof = 1;
        return 0;
    }

    if (ctx->internal->eof) {
        av_log(ctx, AV_LOG_ERROR, "A non-NULL packet sent after an EOF.\n");
        return AVERROR(EINVAL);
    }

    if (ctx->internal->buffer_pkt->data ||
        ctx->internal->buffer_pkt->side_data_elems)
        return AVERROR(EAGAIN);

    ret = av_packet_make_refcounted(pkt);
    if (ret < 0)
        return ret;
    av_packet_move_ref(ctx->internal->buffer_pkt, pkt);

    return 0;
}
void av_packet_move_ref(AVPacket *dst, AVPacket *src)
{
    *dst = *src;
    av_init_packet(src);
    src->data = NULL;
    src->size = 0;
}

     可以看到,av_bsf_send_packet的作用是把传过来的AVPacket放到指定的 AVBSFContext内部,并将该AVPacket置为空。

1.2 decode_receive_frame_internal()

decode_receive_frame_internal 是解码的关键函数,定义如下:

static int decode_receive_frame_internal(AVCodecContext *avctx, AVFrame *frame)
{
    AVCodecInternal *avci = avctx->internal;
    int ret;

    av_assert0(!frame->buf[0]);

    if (avctx->codec->receive_frame)
        ret = avctx->codec->receive_frame(avctx, frame);
    else
        ret = decode_simple_receive_frame(avctx, frame);

    if (ret == AVERROR_EOF)
        avci->draining_done = 1;

    if (!ret) {
        /* the only case where decode data is not set should be decoders
         * that do not call ff_get_buffer() */
        av_assert0((frame->private_ref && frame->private_ref->size == sizeof(FrameDecodeData)) ||
                   !(avctx->codec->capabilities & AV_CODEC_CAP_DR1));

        if (frame->private_ref) {
            FrameDecodeData *fdd = (FrameDecodeData*)frame->private_ref->data;

            if (fdd->post_process) {
                ret = fdd->post_process(avctx, frame);
                if (ret < 0) {
                    av_frame_unref(frame);
                    return ret;
                }
            }
        }
    }

    /* free the per-frame decode data */
    av_buffer_unref(&frame->private_ref);

    return ret;
}

      可以看到,decode_receive_frame_internal首先会判断AVCodec是否存在 receive_frame 这个函数指针,如果存在,就使用该指针对应的函数实现进行解码,否则调用decode_simple_receive_frame解码。
以 H264 格式为例,对应的解码器为 ff_h264_decoder:

libavcodec\h264dec.c
AVCodec ff_h264_decoder = {
    .name                  = "h264",
    .long_name             = NULL_IF_CONFIG_SMALL("H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
    .type                  = AVMEDIA_TYPE_VIDEO,
    .id                    = AV_CODEC_ID_H264,
    .priv_data_size        = sizeof(H264Context),
    .init                  = h264_decode_init,
    .close                 = h264_decode_end,
    .decode                = h264_decode_frame,
    .capabilities          = /*AV_CODEC_CAP_DRAW_HORIZ_BAND |*/ AV_CODEC_CAP_DR1 |
                             AV_CODEC_CAP_DELAY | AV_CODEC_CAP_SLICE_THREADS |
                             AV_CODEC_CAP_FRAME_THREADS,
    .hw_configs            = (const AVCodecHWConfigInternal*[]) {
#if CONFIG_H264_DXVA2_HWACCEL
                               HWACCEL_DXVA2(h264),
#endif
#if CONFIG_H264_D3D11VA_HWACCEL
                               HWACCEL_D3D11VA(h264),
#endif
#if CONFIG_H264_D3D11VA2_HWACCEL
                               HWACCEL_D3D11VA2(h264),
#endif
#if CONFIG_H264_NVDEC_HWACCEL
                               HWACCEL_NVDEC(h264),
#endif
#if CONFIG_H264_VAAPI_HWACCEL
                               HWACCEL_VAAPI(h264),
#endif
#if CONFIG_H264_VDPAU_HWACCEL
                               HWACCEL_VDPAU(h264),
#endif
#if CONFIG_H264_VIDEOTOOLBOX_HWACCEL
                               HWACCEL_VIDEOTOOLBOX(h264),
#endif
                               NULL
                           },
    .caps_internal         = FF_CODEC_CAP_INIT_THREADSAFE | FF_CODEC_CAP_EXPORTS_CROPPING,
    .flush                 = flush_dpb,
    .init_thread_copy      = ONLY_IF_THREADS_ENABLED(decode_init_thread_copy),
    .update_thread_context = ONLY_IF_THREADS_ENABLED(ff_h264_update_thread_context),
    .profiles              = NULL_IF_CONFIG_SMALL(ff_h264_profiles),
    .priv_class            = &h264_class,
};

     没有发现函数指针receive_frame,因此这里直接看decode_simple_receive_frame,如下。

static int decode_simple_receive_frame(AVCodecContext *avctx, AVFrame *frame)
{
    int ret;

    while (!frame->buf[0]) {
        ret = decode_simple_internal(avctx, frame);
        if (ret < 0)
            return ret;
    }

    return 0;
}

可以看到,它直接调用decode_simple_internal这个函数:

/*
 * The core of the receive_frame_wrapper for the decoders implementing
 * the simple API. Certain decoders might consume partial packets without
 * returning any output, so this function needs to be called in a loop until it
 * returns EAGAIN.
 **/
 /*
 *某些解码器可能会消耗部分数据包而不返回任何输出,因此需要在循环中调用此函** 数,直到它返回EAGAIN
 */
static int decode_simple_internal(AVCodecContext *avctx, AVFrame *frame)
{
    AVCodecInternal   *avci = avctx->internal;
    DecodeSimpleContext *ds = &avci->ds;
    AVPacket           *pkt = ds->in_pkt;
    // copy to ensure we do not change pkt
    int got_frame, actual_got_frame;
    int ret;

    if (!pkt->data && !avci->draining) {
        av_packet_unref(pkt);
        //获取在执行av_bsf_send_packet时缓存的AVPacket
        ret = ff_decode_get_packet(avctx, pkt);
        if (ret < 0 && ret != AVERROR_EOF)
            return ret;
    }

    // Some codecs (at least wma lossless) will crash when feeding drain packets
    // after EOF was signaled.
    if (avci->draining_done)
        return AVERROR_EOF;

    if (!pkt->data &&
        !(avctx->codec->capabilities & AV_CODEC_CAP_DELAY ||
          avctx->active_thread_type & FF_THREAD_FRAME))
        return AVERROR_EOF;

    got_frame = 0;

    if (HAVE_THREADS && avctx->active_thread_type & FF_THREAD_FRAME) {
        //获取异步解码缓存的AVFrame
        ret = ff_thread_decode_frame(avctx, frame, &got_frame, pkt);
    } else {
        //解码的关键函数
        ret = avctx->codec->decode(avctx, frame, &got_frame, pkt);

        if (!(avctx->codec->caps_internal & FF_CODEC_CAP_SETS_PKT_DTS))
            frame->pkt_dts = pkt->dts;
        if (avctx->codec->type == AVMEDIA_TYPE_VIDEO) {
            if(!avctx->has_b_frames)
                frame->pkt_pos = pkt->pos;
            //FIXME these should be under if(!avctx->has_b_frames)
            /* get_buffer is supposed to set frame parameters */
            if (!(avctx->codec->capabilities & AV_CODEC_CAP_DR1)) {
                if (!frame->sample_aspect_ratio.num)  frame->sample_aspect_ratio = avctx->sample_aspect_ratio;
                if (!frame->width)                    frame->width               = avctx->width;
                if (!frame->height)                   frame->height              = avctx->height;
                if (frame->format == AV_PIX_FMT_NONE) frame->format              = avctx->pix_fmt;
            }
        }
    }
    emms_c();
    actual_got_frame = got_frame;

    if (avctx->codec->type == AVMEDIA_TYPE_VIDEO) {
        if (frame->flags & AV_FRAME_FLAG_DISCARD)
            got_frame = 0;
        if (got_frame)
            frame->best_effort_timestamp = guess_correct_pts(avctx,
                                                             frame->pts,
                                                             frame->pkt_dts);
    } else if (avctx->codec->type == AVMEDIA_TYPE_AUDIO) {
        uint8_t *side;
        int side_size;
        uint32_t discard_padding = 0;
        uint8_t skip_reason = 0;
        uint8_t discard_reason = 0;

        if (ret >= 0 && got_frame) {
            frame->best_effort_timestamp = guess_correct_pts(avctx,
                                                             frame->pts,
                                                             frame->pkt_dts);
            if (frame->format == AV_SAMPLE_FMT_NONE)
                frame->format = avctx->sample_fmt;
            if (!frame->channel_layout)
                frame->channel_layout = avctx->channel_layout;
            if (!frame->channels)
                frame->channels = avctx->channels;
            if (!frame->sample_rate)
                frame->sample_rate = avctx->sample_rate;
        }

        side= av_packet_get_side_data(avci->last_pkt_props, AV_PKT_DATA_SKIP_SAMPLES, &side_size);
        if(side && side_size>=10) {
            avctx->internal->skip_samples = AV_RL32(side) * avctx->internal->skip_samples_multiplier;
            discard_padding = AV_RL32(side + 4);
            av_log(avctx, AV_LOG_DEBUG, "skip %d / discard %d samples due to side data\n",
                   avctx->internal->skip_samples, (int)discard_padding);
            skip_reason = AV_RL8(side + 8);
            discard_reason = AV_RL8(side + 9);
        }

        if ((frame->flags & AV_FRAME_FLAG_DISCARD) && got_frame &&
            !(avctx->flags2 & AV_CODEC_FLAG2_SKIP_MANUAL)) {
            avctx->internal->skip_samples = FFMAX(0, avctx->internal->skip_samples - frame->nb_samples);
            got_frame = 0;
        }

        if (avctx->internal->skip_samples > 0 && got_frame &&
            !(avctx->flags2 & AV_CODEC_FLAG2_SKIP_MANUAL)) {
            if(frame->nb_samples <= avctx->internal->skip_samples){
                got_frame = 0;
                avctx->internal->skip_samples -= frame->nb_samples;
                av_log(avctx, AV_LOG_DEBUG, "skip whole frame, skip left: %d\n",
                       avctx->internal->skip_samples);
            } else {
                av_samples_copy(frame->extended_data, frame->extended_data, 0, avctx->internal->skip_samples,
                                frame->nb_samples - avctx->internal->skip_samples, avctx->channels, frame->format);
                if(avctx->pkt_timebase.num && avctx->sample_rate) {
                    int64_t diff_ts = av_rescale_q(avctx->internal->skip_samples,
                                                   (AVRational){1, avctx->sample_rate},
                                                   avctx->pkt_timebase);
                    if(frame->pts!=AV_NOPTS_VALUE)
                        frame->pts += diff_ts;
#if FF_API_PKT_PTS
FF_DISABLE_DEPRECATION_WARNINGS
                    if(frame->pkt_pts!=AV_NOPTS_VALUE)
                        frame->pkt_pts += diff_ts;
FF_ENABLE_DEPRECATION_WARNINGS
#endif
                    if(frame->pkt_dts!=AV_NOPTS_VALUE)
                        frame->pkt_dts += diff_ts;
                    if (frame->pkt_duration >= diff_ts)
                        frame->pkt_duration -= diff_ts;
                } else {
                    av_log(avctx, AV_LOG_WARNING, "Could not update timestamps for skipped samples.\n");
                }
                av_log(avctx, AV_LOG_DEBUG, "skip %d/%d samples\n",
                       avctx->internal->skip_samples, frame->nb_samples);
                frame->nb_samples -= avctx->internal->skip_samples;
                avctx->internal->skip_samples = 0;
            }
        }

        if (discard_padding > 0 && discard_padding <= frame->nb_samples && got_frame &&
            !(avctx->flags2 & AV_CODEC_FLAG2_SKIP_MANUAL)) {
            if (discard_padding == frame->nb_samples) {
                got_frame = 0;
            } else {
                if(avctx->pkt_timebase.num && avctx->sample_rate) {
                    int64_t diff_ts = av_rescale_q(frame->nb_samples - discard_padding,
                                                   (AVRational){1, avctx->sample_rate},
                                                   avctx->pkt_timebase);
                    frame->pkt_duration = diff_ts;
                } else {
                    av_log(avctx, AV_LOG_WARNING, "Could not update timestamps for discarded samples.\n");
                }
                av_log(avctx, AV_LOG_DEBUG, "discard %d/%d samples\n",
                       (int)discard_padding, frame->nb_samples);
                frame->nb_samples -= discard_padding;
            }
        }

        if ((avctx->flags2 & AV_CODEC_FLAG2_SKIP_MANUAL) && got_frame) {
            AVFrameSideData *fside = av_frame_new_side_data(frame, AV_FRAME_DATA_SKIP_SAMPLES, 10);
            if (fside) {
                AV_WL32(fside->data, avctx->internal->skip_samples);
                AV_WL32(fside->data + 4, discard_padding);
                AV_WL8(fside->data + 8, skip_reason);
                AV_WL8(fside->data + 9, discard_reason);
                avctx->internal->skip_samples = 0;
            }
        }
    }

    if (avctx->codec->type == AVMEDIA_TYPE_AUDIO &&
        !avci->showed_multi_packet_warning &&
        ret >= 0 && ret != pkt->size && !(avctx->codec->capabilities & AV_CODEC_CAP_SUBFRAMES)) {
        av_log(avctx, AV_LOG_WARNING, "Multiple frames in a packet.\n");
        avci->showed_multi_packet_warning = 1;
    }

    if (!got_frame)
        av_frame_unref(frame);

    if (ret >= 0 && avctx->codec->type == AVMEDIA_TYPE_VIDEO && !(avctx->flags & AV_CODEC_FLAG_TRUNCATED))
        ret = pkt->size;

#if FF_API_AVCTX_TIMEBASE
    if (avctx->framerate.num > 0 && avctx->framerate.den > 0)
        avctx->time_base = av_inv_q(av_mul_q(avctx->framerate, (AVRational){avctx->ticks_per_frame, 1}));
#endif

    /* do not stop draining when actual_got_frame != 0 or ret < 0 */
    /* got_frame == 0 but actual_got_frame != 0 when frame is discarded */
    if (avctx->internal->draining && !actual_got_frame) {
        if (ret < 0) {
            /* prevent infinite loop if a decoder wrongly always return error on draining */
            /* reasonable nb_errors_max = maximum b frames + thread count */
            int nb_errors_max = 20 + (HAVE_THREADS && avctx->active_thread_type & FF_THREAD_FRAME ?
                                avctx->thread_count : 1);

            if (avci->nb_draining_errors++ >= nb_errors_max) {
                av_log(avctx, AV_LOG_ERROR, "Too many errors when draining, this is a bug. "
                       "Stop draining and force EOF.\n");
                avci->draining_done = 1;
                ret = AVERROR_BUG;
            }
        } else {
            avci->draining_done = 1;
        }
    }

    avci->compat_decode_consumed += ret;

    //如果这个 AVPacket 已被消耗完,则释放内存,否则调整指针、成员变量
    if (ret >= pkt->size || ret < 0) {
        av_packet_unref(pkt);
    } else {
        int consumed = ret;

        pkt->data                += consumed;
        pkt->size                -= consumed;
        avci->last_pkt_props->size -= consumed; // See extract_packet_props() comment.
        pkt->pts                  = AV_NOPTS_VALUE;
        pkt->dts                  = AV_NOPTS_VALUE;
        avci->last_pkt_props->pts = AV_NOPTS_VALUE;
        avci->last_pkt_props->dts = AV_NOPTS_VALUE;
    }

    if (got_frame)
        av_assert0(frame->buf[0]);

    return ret < 0 ? ret : 0;
}

     这个函数其实很长,但最关键的只有 avctx->codec->decode 这一句代码,decode 是 结构体 AVCodec 的函数指针,不同的编码格式对应不同的函数实现,以 H264 为例,对应的解码器为 ff_h264_decoder (上面把它的声明贴出来了),函数实现是 h264_decode_frame。
这里简单看一下 h264_decode_frame:

static int h264_decode_frame(AVCodecContext *avctx, void *data,
                             int *got_frame, AVPacket *avpkt)
{
    const uint8_t *buf = avpkt->data;
    int buf_size       = avpkt->size;
    H264Context *h     = avctx->priv_data;
    AVFrame *pict      = data;
    int buf_index;
    int ret;

    h->flags = avctx->flags;
    h->setup_finished = 0;
    h->nb_slice_ctx_queued = 0;

    ff_h264_unref_picture(h, &h->last_pic_for_ec);

    /* end of stream, output what is still in the buffers */
    if (buf_size == 0)
        return send_next_delayed_frame(h, pict, got_frame, 0);

    if (h->is_avc && av_packet_get_side_data(avpkt, AV_PKT_DATA_NEW_EXTRADATA, NULL)) {
        int side_size;
        uint8_t *side = av_packet_get_side_data(avpkt, AV_PKT_DATA_NEW_EXTRADATA, &side_size);
        if (is_extra(side, side_size))
            ff_h264_decode_extradata(side, side_size,
                                     &h->ps, &h->is_avc, &h->nal_length_size,
                                     avctx->err_recognition, avctx);
    }
    if (h->is_avc && buf_size >= 9 && buf[0]==1 && buf[2]==0 && (buf[4]&0xFC)==0xFC) {
        if (is_extra(buf, buf_size))
            return ff_h264_decode_extradata(buf, buf_size,
                                            &h->ps, &h->is_avc, &h->nal_length_size,
                                            avctx->err_recognition, avctx);
    }

    buf_index = decode_nal_units(h, buf, buf_size);
    if (buf_index < 0)
        return AVERROR_INVALIDDATA;

    if (!h->cur_pic_ptr && h->nal_unit_type == H264_NAL_END_SEQUENCE) {
        av_assert0(buf_index <= buf_size);
        return send_next_delayed_frame(h, pict, got_frame, buf_index);
    }

    if (!(avctx->flags2 & AV_CODEC_FLAG2_CHUNKS) && (!h->cur_pic_ptr || !h->has_slice)) {
        if (avctx->skip_frame >= AVDISCARD_NONREF ||
            buf_size >= 4 && !memcmp("Q264", buf, 4))
            return buf_size;
        av_log(avctx, AV_LOG_ERROR, "no frame!\n");
        return AVERROR_INVALIDDATA;
    }

    if (!(avctx->flags2 & AV_CODEC_FLAG2_CHUNKS) ||
        (h->mb_y >= h->mb_height && h->mb_height)) {
        if ((ret = ff_h264_field_end(h, &h->slice_ctx[0], 0)) < 0)
            return ret;

        /* Wait for second field. */
        if (h->next_output_pic) {
            ret = finalize_frame(h, pict, h->next_output_pic, got_frame);
            if (ret < 0)
                return ret;
        }
    }

    av_assert0(pict->buf[0] || !*got_frame);

    ff_h264_unref_picture(h, &h->last_pic_for_ec);

    return get_consumed_bytes(buf_index, buf_size);
}

可以看到,这个函数大致可以分为 2 个步骤:
    1) 判断当前是否已到达文件的末尾,如果是,则返回解码后依然存在于缓存中的数据
    2) 否则根据 NAL 的类型进行解码,如果是 SPS、PPS 数据,则调用 ff_h264_decode_extradata,否则调用 decode_nal_units
其中函数 is_extra 用于判断数据类型是否为 SPS、PPS:

static int is_extra(const uint8_t *buf, int buf_size)
{
    int cnt= buf[5]&0x1f;
    const uint8_t *p= buf+6;
    if (!cnt)
        return 0;
    while(cnt--){
        int nalsize= AV_RB16(p) + 2;
        if(nalsize > buf_size - (p-buf) || (p[2] & 0x9F) != 7)
            return 0;
        p += nalsize;
    }
    cnt = *(p++);
    if(!cnt)
        return 0;
    while(cnt--){
        int nalsize= AV_RB16(p) + 2;
        if(nalsize > buf_size - (p-buf) || (p[2] & 0x9F) != 8)
            return 0;
        p += nalsize;
    }
    return 1;
}

二、avcodec_receive_frame

int attribute_align_arg avcodec_receive_frame(AVCodecContext *avctx, AVFrame *frame)
{
    AVCodecInternal *avci = avctx->internal;
    int ret, changed;

    av_frame_unref(frame);

    if (!avcodec_is_open(avctx) || !av_codec_is_decoder(avctx->codec))
        return AVERROR(EINVAL);

    if (avci->buffer_frame->buf[0]) {
        av_frame_move_ref(frame, avci->buffer_frame);
    } else {
        ret = decode_receive_frame_internal(avctx, frame);
        if (ret < 0)
            return ret;
    }

    if (avctx->codec_type == AVMEDIA_TYPE_VIDEO) {
        ret = apply_cropping(avctx, frame);
        if (ret < 0) {
            av_frame_unref(frame);
            return ret;
        }
    }

    avctx->frame_number++;

    if (avctx->flags & AV_CODEC_FLAG_DROPCHANGED) {

        if (avctx->frame_number == 1) {
            avci->initial_format = frame->format;
            switch(avctx->codec_type) {
            case AVMEDIA_TYPE_VIDEO:
                avci->initial_width  = frame->width;
                avci->initial_height = frame->height;
                break;
            case AVMEDIA_TYPE_AUDIO:
                avci->initial_sample_rate = frame->sample_rate ? frame->sample_rate :
                                                                 avctx->sample_rate;
                avci->initial_channels       = frame->channels;
                avci->initial_channel_layout = frame->channel_layout;
                break;
            }
        }

        if (avctx->frame_number > 1) {
            changed = avci->initial_format != frame->format;

           switch(avctx->codec_type) {
            case AVMEDIA_TYPE_VIDEO:
                changed |= avci->initial_width  != frame->width ||
                           avci->initial_height != frame->height;
                break;
            case AVMEDIA_TYPE_AUDIO:
                changed |= avci->initial_sample_rate    != frame->sample_rate ||
                           avci->initial_sample_rate    != avctx->sample_rate ||
                           avci->initial_channels       != frame->channels ||
                           avci->initial_channel_layout != frame->channel_layout;
                break;
            }

            if (changed) {
                avci->changed_frames_dropped++;
                av_log(avctx, AV_LOG_INFO, "dropped changed frame #%d pts %"PRId64
                                            " drop count: %d \n",
                                            avctx->frame_number, frame->pts,
                                            avci->changed_frames_dropped);
                av_frame_unref(frame);
                return AVERROR_INPUT_CHANGED;
            }
        }
    }
    return 0;
}

     可以看到,这个函数的逻辑很简单,它首先会尝试从缓存中直接获取 AVFrame,如果不存在,则调用decode_receive_frame_internal进行解码。