live555关于RTSP协议交互流程

RTP在和h264

RTP在和h265

RTP载荷AAC

live555闭环双向链表

概要

rtsp在交互的过程中用到很多协议:tcp,udp,rtp,rtcp,sdp等协议;该篇文章主要分析在live555中这些协议是什么时候被创建的,什么时候被使用的等协议相关流程。

TCP:服务器与客户端进行协商(OPTION DESCRIBE SETUP PLAY);

UDP/TCP:协议是rtsp服务器用来想客户端推流;当然rtsp向客户端推流也可以使用tcp协议;那么就rtsp而言使用udp推流和使用tcp推流有什么区别呢?

UDP推流

tcp连接进行rtsp信令交互;

创建新的udp套接字来发送rtp包;

创建新的udp套接字来发送rtcp包;

TCP推流

tcp连接进行rtsp信令交互;

复用rtsp的tcp连接发送rtp和rtcp包;

嵌入式开发一般使用udp推流,实时性相对较高;

RTP:对视频流(h264/h265)/音频流(AAC/MP3)裸流进行封装,用于网络传输;

RTCP:服务器和客户端用来管理流媒体协议;

TCP交互协商

在程序创建RTSPServer类对象时就会创建用于信令协商的TCP协议,见如下代码:

cpp 复制代码
//创建RTSPServer类对象
RTSPServer* rtspServer = RTSPServer::createNew(*env, 8554, authDB);
//createNew实现
RTSPServer*
RTSPServer::createNew(UsageEnvironment& env, Port ourPort,
          UserAuthenticationDatabase* authDatabase,
          unsigned reclamationSeconds) {
  int ourSocketIPv4 = setUpOurSocket(env, ourPort, AF_INET);
  int ourSocketIPv6 = setUpOurSocket(env, ourPort, AF_INET6);
  if (ourSocketIPv4 < 0 && ourSocketIPv6 < 0) return NULL;
  
  return new RTSPServer(env, ourSocketIPv4, ourSocketIPv6, ourPort, authDatabase, reclamationSeconds);
}

从源码可以看出创建RTSPServer类对象的时候会创建ipv4和ipv6两种套接字,因此理论上来说live555实现的rtsp服务器支持ipv4和ipv6两种网络传输。

cpp 复制代码
//RTSPServer构造函数
RTSPServer::RTSPServer(UsageEnvironment& env,
           int ourSocketIPv4, int ourSocketIPv6, Port ourPort,
           UserAuthenticationDatabase* authDatabase,
           unsigned reclamationSeconds)
  : GenericMediaServer(env, ourSocketIPv4, ourSocketIPv6, ourPort, reclamationSeconds),
    fHTTPServerSocketIPv4(-1), fHTTPServerSocketIPv6(-1), fHTTPServerPort(0),
    fClientConnectionsForHTTPTunneling(NULL), // will get created if needed
    fTCPStreamingDatabase(HashTable::create(ONE_WORD_HASH_KEYS)),
    fPendingRegisterOrDeregisterRequests(HashTable::create(ONE_WORD_HASH_KEYS)),
    fRegisterOrDeregisterRequestCounter(0), fAuthDB(authDatabase),
    fAllowStreamingRTPOverTCP(True),
    fOurConnectionsUseTLS(False), fWeServeSRTP(False) {
}
//GenericMediaServer构造函数
GenericMediaServer
::GenericMediaServer(UsageEnvironment& env, int ourSocketIPv4, int ourSocketIPv6, Port ourPort,
         unsigned reclamationSeconds)
  : Medium(env),
    fServerSocketIPv4(ourSocketIPv4), fServerSocketIPv6(ourSocketIPv6),
    fServerPort(ourPort), fReclamationSeconds(reclamationSeconds),
    fServerMediaSessions(HashTable::create(STRING_HASH_KEYS)),
    fClientConnections(HashTable::create(ONE_WORD_HASH_KEYS)),
    fClientSessions(HashTable::create(STRING_HASH_KEYS)),
    fPreviousClientSessionId(0),
    fTLSCertificateFileName(NULL), fTLSPrivateKeyFileName(NULL) {
  ignoreSigPipeOnSocket(fServerSocketIPv4); // so that clients on the same host that are killed don't also kill us
  ignoreSigPipeOnSocket(fServerSocketIPv6); // ditto
  
  // Arrange to handle connections from others:
  env.taskScheduler().turnOnBackgroundReadHandling(fServerSocketIPv4, incomingConnectionHandlerIPv4, this);
  env.taskScheduler().turnOnBackgroundReadHandling(fServerSocketIPv6, incomingConnectionHandlerIPv6, this);
}

在GenericMediaServer构造函数中会把创建的fServerSocketIPv4和fServerSocketIPv6这两个套接字插入到双向闭环链表中等待doEventLoop循环处理,对应的处理函数分别为:incomingConnectionHandlerIPv4, incomingConnectionHandlerIPv6;最终都会调用incomingConnectionHandlerOnSocket函数;

cpp 复制代码
void GenericMediaServer::incomingConnectionHandlerOnSocket(int serverSocket) {
  struct sockaddr_storage clientAddr;
  SOCKLEN_T clientAddrLen = sizeof clientAddr;
  int clientSocket = accept(serverSocket, (struct sockaddr*)&clientAddr, &clientAddrLen);
  if (clientSocket < 0) {
    int err = envir().getErrno();
    if (err != EWOULDBLOCK) {
      envir().setResultErrMsg("accept() failed: ");
    }
    return;
  }
  ignoreSigPipeOnSocket(clientSocket); // so that clients on the same host that are killed don't also kill us
  makeSocketNonBlocking(clientSocket);
  increaseSendBufferTo(envir(), clientSocket, 50*1024);
  
#ifdef DEBUG
  envir() << "accept()ed connection from " << AddressString(clientAddr).val() << "\n";
#endif
  
  // Create a new object for handling this connection:
  (void)createNewClientConnection(clientSocket, clientAddr);
}
//createNewClientConnection函数实现
GenericMediaServer::ClientConnection*
RTSPServer::createNewClientConnection(int clientSocket, struct sockaddr_storage const& clientAddr) {
  return new RTSPClientConnection(*this, clientSocket, clientAddr, fOurConnectionsUseTLS);
}

在doEventLoop循环中会议中accept监视tcp连接,如果有客户端连接就会创建客户端连接类RTSPClientConnection;最终会把客户端套接字clientSocket传递给ClientConnection构造函数;

cpp 复制代码
GenericMediaServer::ClientConnection
::ClientConnection(GenericMediaServer& ourServer,
       int clientSocket, struct sockaddr_storage const& clientAddr,
       Boolean useTLS)
  : fOurServer(ourServer), fOurSocket(clientSocket), fClientAddr(clientAddr), fTLS(envir()) {
  fInputTLS = fOutputTLS = &fTLS;

  // Add ourself to our 'client connections' table:
  fOurServer.fClientConnections->Add((char const*)this, this);
  
  if (useTLS) {
    // Perform extra processing to handle a TLS connection:
    fTLS.setCertificateAndPrivateKeyFileNames(ourServer.fTLSCertificateFileName,
                ourServer.fTLSPrivateKeyFileName);
    fTLS.isNeeded = True;

    fTLS.tlsAcceptIsNeeded = True; // call fTLS.accept() the next time the socket is readable
  }

  // Arrange to handle incoming requests:
  resetRequestBuffer();
  envir().taskScheduler()
    .setBackgroundHandling(fOurSocket, SOCKET_READABLE|SOCKET_EXCEPTION, incomingRequestHandler, this);
}
//incomingRequestHandler函数最终调用
void GenericMediaServer::ClientConnection::incomingRequestHandler() {
  if (fInputTLS->tlsAcceptIsNeeded) { // we need to successfully call fInputTLS->accept() first:
    if (fInputTLS->accept(fOurSocket) <= 0) return; // either an error, or we need to try again later

    fInputTLS->tlsAcceptIsNeeded = False;
    // We can now read data, as usual:
  }

  int bytesRead;
  if (fInputTLS->isNeeded) {
    bytesRead = fInputTLS->read(&fRequestBuffer[fRequestBytesAlreadySeen], fRequestBufferBytesLeft);
  } else {
    struct sockaddr_storage dummy; // 'from' address, meaningless in this case
  
    bytesRead = readSocket(envir(), fOurSocket, &fRequestBuffer[fRequestBytesAlreadySeen], fRequestBufferBytesLeft, dummy);
  }
  handleRequestBytes(bytesRead);//该函数实现了对 OPTION DESCRIBE SETUP等各种信令的处理逻辑
}

在构造函数中setBackgroundHandling会把客户端套接字fOurSocket和对应的处理函数incomingRequestHandler添加到闭环双链表中,在doEventLoop中循环遍历,客户端有信令交互就调用相关的处理函数;至此用于协商的TCP协议处理流程就结束了。

关于live555的闭环双向链表参考我的另一篇文章:live555的核心数据结构值之闭环双向链表-CSDN博客

UDP流媒体传输

UDP流媒体传输服务器需要创建两个四个UDP套接字,用于传输音频RTP,音频RTCP,视频RTP,视频RTCP;该文档是以H264的传输为例所以只介绍视频RTP端口,视频RTCP端口的创建过程,音频类似;

RTP,RTCP端口是在SETUP信令处理函数handleCmd_SETUP中被创建,该函数最终调用了getStreamParameters函数:

cpp 复制代码
subsession->getStreamParameters(fOurSessionId, fOurClientConnection->fClientAddr,
            clientRTPPort, clientRTCPPort,
            fStreamStates[trackNum].tcpSocketNum, rtpChannelId, rtcpChannelId,
                                    &fOurClientConnection->fTLS,
            destinationAddress, destinationTTL, fIsMulticast,
            serverRTPPort, serverRTCPPort,
            fStreamStates[trackNum].streamToken);

该函数将客户端的RTP端口:clientRTPPort和RTCP端口:clientRTCPPort都进行了处理;这两个端口是客户端发送SETUP信令时携带的消息;告诉服务器RTP RTCP包改往哪里发;getStreamParameters也创建了服务器的RTP RTCP端口:serverRTPPort, serverRTCPPort;

getStreamParameters内部调用了createGroupsock函数:

cpp 复制代码
void OnDemandServerMediaSubsession ::getStreamParameters(...)
{
    .
    .
    .
    if (clientRTPPort.num() != 0 || tcpSocketNum >= 0)
    { // Normal case: Create destinations
      portNumBits serverPortNum;
      if (clientRTCPPort.num() == 0)
      {
        // We're streaming raw UDP (not RTP). Create a single groupsock:
        NoReuse dummy(envir()); // ensures that we skip over ports that are already in use
        for (serverPortNum = fInitialPortNum;; ++serverPortNum)
        {
          serverRTPPort = serverPortNum;
          rtpGroupsock = createGroupsock(nullAddress(destinationAddress.ss_family), serverRTPPort);
          if (rtpGroupsock->socketNum() >= 0)
            break; // success
        }

        udpSink = BasicUDPSink::createNew(envir(), rtpGroupsock);
      }
      else
      {
        // Normal case: We're streaming RTP (over UDP or TCP).  Create a pair of
        // groupsocks (RTP and RTCP), with adjacent port numbers (RTP port number even).
        // (If we're multiplexing RTCP and RTP over the same port number, it can be odd or even.)
        NoReuse dummy(envir()); // ensures that we skip over ports that are already in use
        for (portNumBits serverPortNum = fInitialPortNum;; ++serverPortNum)
        {
          serverRTPPort = serverPortNum;
          //创建RTP端口(rtp的UDP套接字)
          rtpGroupsock = createGroupsock(nullAddress(destinationAddress.ss_family), serverRTPPort);
          if (rtpGroupsock->socketNum() < 0)
          {
            delete rtpGroupsock;
            continue; // try again
          }

          if (fMultiplexRTCPWithRTP)
          {
            // Use the RTP 'groupsock' object for RTCP as well:
            serverRTCPPort = serverRTPPort;
            rtcpGroupsock = rtpGroupsock;
          }
          else
          {
            // Create a separate 'groupsock' object (with the next (odd) port number) for RTCP:
            //RTCP端口号在RTP端口号的基础上加1
            serverRTCPPort = ++serverPortNum;
            //创建RTCP端口(rtcp的UDP套接字)
            rtcpGroupsock = createGroupsock(nullAddress(destinationAddress.ss_family), serverRTCPPort);
            if (rtcpGroupsock->socketNum() < 0)
            {
              delete rtpGroupsock;
              delete rtcpGroupsock;
              continue; // try again
            }
          }

          break; // success
        }

        unsigned char rtpPayloadType = 96 + trackNumber() - 1; // if dynamic
        rtpSink = mediaSource == NULL ? NULL
                                      : createNewRTPSink(rtpGroupsock, rtpPayloadType, mediaSource);
        if (rtpSink != NULL)
        {
          if (fParentSession->streamingUsesSRTP)
          {
            rtpSink->setupForSRTP(fMIKEYStateMessage, fMIKEYStateMessageSize);
          }
          if (rtpSink->estimatedBitrate() > 0)
            streamBitrate = rtpSink->estimatedBitrate();
        }
      }
    .
    .
    .
}

由代码可以看出serverRTPPort的初始值是fInitialPortNum;而fInitialPortNum在创建OnDemandServerMediaSubsession对象时有个默认值6970;如果没有设置端口号则使用默认端口号;

上面代码可以看出而RTCP端口号是在RTP的端口号的基础上加1

cpp 复制代码
OnDemandServerMediaSubsession(UsageEnvironment& env, Boolean reuseFirstSource,
        portNumBits initialPortNum = 6970,
        Boolean multiplexRTCPWithRTP = False);

当第二个客户端连接时,依然是从6970开始创建所需的RTP RTCP端口号,但是createGroupsock会发现6970 6971端口号被占用,于是返回-1;继续for循环将端口号累加;

cpp 复制代码
 for (portNumBits serverPortNum = fInitialPortNum;; ++serverPortNum)
        {
          serverRTPPort = serverPortNum;
          rtpGroupsock = createGroupsock(nullAddress(destinationAddress.ss_family), serverRTPPort);
          if (rtpGroupsock->socketNum() < 0)
          {
            delete rtpGroupsock;
            continue; // try again
          }
          .
          .
          .
        }

//fInitialPortNum为基数6970;

第一个客户端:rtp:6970 rtcp:6971

第二个客户端:6970 6971 被占用createGroupsock返回-1;因此for循环continue继续累加++serverPortNum; rtp:6972 rtcp:6973

......

那么怎么自定义端口号呢?

我们在做rtsp服务器的时候都会创建一个类用于实现createNewStreamSource虚函数该类继承于OnDemandServerMediaSubsession;而类的构造函数里会执行OnDemandServerMediaSubsession的构造函数;所以如果你想要自己定义服务器的RTP端口号只需在执行OnDemandServerMediaSubsession构造函数是传入参数即可:

cpp 复制代码
H264LiveVideoServerMediaSubssion::H264LiveVideoServerMediaSubssion(
    UsageEnvironment &env, Boolean reuseFirstSource)
    : OnDemandServerMediaSubsession(env, reuseFirstSource, 1234) {}

TCP流媒体传输使用的时信令交互的套接字,这里不做解释;关于流媒体裸流怎么打包成RTP的参考上面的文章;

该文章在持续更新,望持续关注;

相关推荐
Lary_Rock1 小时前
RK3576 LINUX RKNN SDK 测试
linux·运维·服务器
云飞云共享云桌面3 小时前
8位机械工程师如何共享一台图形工作站算力?
linux·服务器·网络
Peter_chq3 小时前
【操作系统】基于环形队列的生产消费模型
linux·c语言·开发语言·c++·后端
一坨阿亮4 小时前
Linux 使用中的问题
linux·运维
dsywws5 小时前
Linux学习笔记之vim入门
linux·笔记·学习
幺零九零零6 小时前
【C++】socket套接字编程
linux·服务器·网络·c++
小林熬夜学编程7 小时前
【Linux系统编程】第四十一弹---线程深度解析:从地址空间到多线程实践
linux·c语言·开发语言·c++·算法
程思扬8 小时前
为什么Uptime+Kuma本地部署与远程使用是网站监控新选择?
linux·服务器·网络·经验分享·后端·网络协议·1024程序员节
sun0077008 小时前
拷贝 cp -rdp 和 cp -a
linux·运维·服务器
我喜欢就喜欢8 小时前
基于qt vs下的视频播放
开发语言·qt·音视频