FFmpeg (sharpFFmpeg) 解码 - 受保护的内存错误

发布于 2024-12-01 07:29:48 字数 1120 浏览 1 评论 0原文

我正在尝试使用 FFmpeg 库的 C# 绑定调用 SharpFFmpeg 来解码我通过 RTP 接收的 H264 视频流。我认为我正确地从 RTP 数据包中解封装 NALU，但我无法解码完整的帧。函数 avcodec_decode_video 调用抛出 AccessViolationException（尝试读取或写入受保护的内存）。
以下是一些代码行：

    //buf is a byte array containing encoded frame
    int success;
    FFmpeg.avcodec_init();
    FFmpeg.avcodec_register_all();
    IntPtr codec = FFmpeg.avcodec_find_decoder(FFmpeg.CodecID.CODEC_ID_H264);
    IntPtr codecCont = FFmpeg.avcodec_alloc_context(); //AVCodecContext
    FFmpeg.avcodec_open(codecCont, codec);
    IntPtr frame = FFmpeg.avcodec_alloc_frame();  //AVFrame
    FFmpeg.avcodec_decode_video(codecCont, frame, ref success, (IntPtr)buf[0], buf.Length); //exception

函数已导入如下：

    [DllImport("avcodec.dll", CallingConvention = CallingConvention.Cdecl), SuppressUnmanagedCodeSecurity]
    public unsafe static extern int avcodec_decode_video(IntPtr pAVCodecContext, IntPtr pAVFrame, ref int got_picture_ptr, IntPtr buf, int buf_size);

不幸的是，我不知道如何处理 codecCont。有人写道，需要使用 RTSP 接收到的会话描述通过 AVCDCR 来填充此结构。但我不知道哪个字段存储此记录。
我很乐意提供任何帮助。
PS请原谅我的英语

原文

I'm trying to use C# binding of FFmpeg library calling SharpFFmpeg to decode H264 video stream that I receive by RTP. I think I correctly decapsulate NALUs from RTP-packets, but I can't decode complete frames. Function avcodec_decode_video calling throws an AccessViolationException (Attempted to read or write protected memory).
Here are some code lines:

    //buf is a byte array containing encoded frame
    int success;
    FFmpeg.avcodec_init();
    FFmpeg.avcodec_register_all();
    IntPtr codec = FFmpeg.avcodec_find_decoder(FFmpeg.CodecID.CODEC_ID_H264);
    IntPtr codecCont = FFmpeg.avcodec_alloc_context(); //AVCodecContext
    FFmpeg.avcodec_open(codecCont, codec);
    IntPtr frame = FFmpeg.avcodec_alloc_frame();  //AVFrame
    FFmpeg.avcodec_decode_video(codecCont, frame, ref success, (IntPtr)buf[0], buf.Length); //exception

Function has been imported as follows:

    [DllImport("avcodec.dll", CallingConvention = CallingConvention.Cdecl), SuppressUnmanagedCodeSecurity]
    public unsafe static extern int avcodec_decode_video(IntPtr pAVCodecContext, IntPtr pAVFrame, ref int got_picture_ptr, IntPtr buf, int buf_size);

Unfortunately, I don't know what do I have to do with the codecCont. Somebody wrote that it is needed to fill this structure by AVCDCR using session description received by RTSP. But I don't know which field(s) store this record.
I'll be glad for any help.
P.S. Excuse me for my English

分享到QQ

分享到微博

如果你对这篇内容有疑问，欢迎到本站社区发帖提问参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。

发布评论

需要登录才能够评论，你可以免费注册一个本站的账号。

掩耳倾听 2024-12-08 07:29:48

我检测到 (IntPtr)buf[0] 指向托管内存。我已经更新了我的代码如下：

  IntPtr frame = FFmpeg.avcodec_alloc_frame();
  IntPtr buffer = Marshal.AllocHGlobal(buf.Length + FFmpeg.FF_INPUT_BUFFER_PADDING_SIZE);
  for (int i = 0; i < buf.Length; i++)
      Marshal.StructureToPtr(buf[i], buffer + i, true);
  avcodec_decode_video(codecCont, frame, ref success, buffer, buf.Length + FFmpeg.FF_INPUT_BUFFER_PADDING_SIZE);

现在函数使 success 变量等于零，但不抛出异常。

I've detected that (IntPtr)buf[0] is pointing to the managed memory. And I've updated my code as follows:

  IntPtr frame = FFmpeg.avcodec_alloc_frame();
  IntPtr buffer = Marshal.AllocHGlobal(buf.Length + FFmpeg.FF_INPUT_BUFFER_PADDING_SIZE);
  for (int i = 0; i < buf.Length; i++)
      Marshal.StructureToPtr(buf[i], buffer + i, true);
  avcodec_decode_video(codecCont, frame, ref success, buffer, buf.Length + FFmpeg.FF_INPUT_BUFFER_PADDING_SIZE);

Now function makes success variable equal zero, but do not throws exception.

回复收藏 0 原文

~没有更多了~