Dapp开发区块链教程

兄弟连区块链入门教程以太坊源码分析p2p-peer.go源码分析

2018-10-24  本文已影响9人  ab6973df9221

  兄弟连区块链入门教程以太坊源码分析p2p-peer.go源码分析,2018年下半年,区块链行业正逐渐褪去发展之初的浮躁、回归理性,表面上看相关人才需求与身价似乎正在回落。但事实上,正是初期泡沫的渐退,让人们更多的关注点放在了区块链真正的技术之上。

nat是网络地址转换的意思。  这部分的源码比较独立而且单一,这里就暂时不分析了。 大家了解基本的功能就行了。

nat下面有upnp和pmp两种网络协议。

### upnp的应用场景(pmp是和upnp类似的协议)

如果用户是通过NAT接入Internet的,同时需要使用BC、电骡eMule等P2P这样的软件,这时UPnP功能就会带来很大的便利。利用UPnP能自动的把BC、电骡eMule等侦听的端口号映射到公网上,以便公网上的用户也能对NAT私网侧发起连接。

主要功能就是提供接口可以把内网的IP+端口 映射为  路由器的IP+端口。 这样就等于内网的程序有了外网的IP地址, 这样公网的用户就可以直接对你进行访问了。 不然就需要通过UDP打洞这种方式来进行访问。

### p2p中的UDP协议

现在大部分用户运行的环境都是内网环境。 内网环境下监听的端口,其他公网的程序是无法直接访问的。需要经过一个打洞的过程。 双方才能联通。这就是所谓的UDP打洞。

在p2p代码里面。 peer代表了一条创建好的网络链路。在一条链路上可能运行着多个协议。比如以太坊的协议(eth)。 Swarm的协议。 或者是Whisper的协议。

peer的结构

    type protoRW struct {

        Protocol

        in     chan Msg        // receices read messages

        closed <-chan struct{} // receives when peer is shutting down

        wstart <-chan struct{} // receives when write may start

        werr   chan<- error    // for write results

        offset uint64

        w      MsgWriter

    }

    // Protocol represents a P2P subprotocol implementation.

    type Protocol struct {

        // Name should contain the official protocol name,

        // often a three-letter word.

        Name string

        // Version should contain the version number of the protocol.

        Version uint

        // Length should contain the number of message codes used

        // by the protocol.

        Length uint64

        // Run is called in a new groutine when the protocol has been

        // negotiated with a peer. It should read and write messages from

        // rw. The Payload for each message must be fully consumed.

        //

        // The peer connection is closed when Start returns. It should return

        // any protocol-level error (such as an I/O error) that is

        // encountered.

        Run func(peer *Peer, rw MsgReadWriter) error

        // NodeInfo is an optional helper method to retrieve protocol specific metadata

        // about the host node.

        NodeInfo func() interface{}

        // PeerInfo is an optional helper method to retrieve protocol specific metadata

        // about a certain peer in the network. If an info retrieval function is set,

        // but returns nil, it is assumed that the protocol handshake is still running.

        PeerInfo func(id discover.NodeID) interface{}

    }

    // Peer represents a connected remote node.

    type Peer struct {

        rw      *conn

running map[string]*protoRW   //运行的协议

        log     log.Logger

        created mclock.AbsTime

        wg       sync.WaitGroup

        protoErr chan error

        closed   chan struct{}

        disc     chan DiscReason

        // events receives message send / receive events if set

        events *event.Feed

    }

peer的创建,根据匹配找到当前Peer支持的protomap

    func newPeer(conn *conn, protocols []Protocol) *Peer {

        protomap := matchProtocols(protocols, conn.caps, conn)

        p := &Peer{

            rw:       conn,

            running:  protomap,

            created:  mclock.Now(),

            disc:     make(chan DiscReason),

            protoErr: make(chan error, len(protomap)+1), // protocols + pingLoop

            closed:   make(chan struct{}),

            log:      log.New("id", conn.id, "conn", conn.flags),

        }

        return p

    }

peer的启动, 启动了两个goroutine线程。 一个是读取。一个是执行ping操作。

    func (p *Peer) run() (remoteRequested bool, err error) {

        var (

writeStart = make(chan struct{}, 1)  //用来控制什么时候可以写入的管道。

            writeErr   = make(chan error, 1)

            readErr    = make(chan error, 1)

            reason     DiscReason // sent to the peer

        )

        p.wg.Add(2)

        go p.readLoop(readErr)

        go p.pingLoop()

        // Start all protocol handlers.

        writeStart <- struct{}{}

//启动所有的协议。

        p.startProtocols(writeStart, writeErr)

        // Wait for an error or disconnect.

    loop:

        for {

            select {

            case err = <-writeErr:

                // A write finished. Allow the next write to start if

                // there was no error.

                if err != nil {

                    reason = DiscNetworkError

                    break loop

                }

                writeStart <- struct{}{}

            case err = <-readErr:

                if r, ok := err.(DiscReason); ok {

                    remoteRequested = true

                    reason = r

                } else {

                    reason = DiscNetworkError

                }

                break loop

            case err = <-p.protoErr:

                reason = discReasonForError(err)

                break loop

            case err = <-p.disc:

                break loop

            }

        }

        close(p.closed)

        p.rw.close(reason)

        p.wg.Wait()

        return remoteRequested, err

    }

startProtocols方法,这个方法遍历所有的协议。

    func (p *Peer) startProtocols(writeStart <-chan struct{}, writeErr chan<- error) {

        p.wg.Add(len(p.running))

        for _, proto := range p.running {

            proto := proto

            proto.closed = p.closed

            proto.wstart = writeStart

            proto.werr = writeErr

            var rw MsgReadWriter = proto

            if p.events != nil {

                rw = newMsgEventer(rw, p.events, p.ID(), proto.Name)

            }

            p.log.Trace(fmt.Sprintf("Starting protocol %s/%d", proto.Name, proto.Version))

// 等于这里为每一个协议都开启了一个goroutine。 调用其Run方法。

            go func() {

// proto.Run(p, rw)这个方法应该是一个死循环。 如果返回就说明遇到了错误。

                err := proto.Run(p, rw)

                if err == nil {

                    p.log.Trace(fmt.Sprintf("Protocol %s/%d returned", proto.Name, proto.Version))

                    err = errProtocolReturned

                } else if err != io.EOF {

                    p.log.Trace(fmt.Sprintf("Protocol %s/%d failed", proto.Name, proto.Version), "err", err)

                }

                p.protoErr <- err

                p.wg.Done()

            }()

        }

    }

回过头来再看看readLoop方法。 这个方法也是一个死循环。 调用p.rw来读取一个Msg(这个rw实际是之前提到的frameRLPx的对象,也就是分帧之后的对象。然后根据Msg的类型进行对应的处理,如果Msg的类型是内部运行的协议的类型。那么发送到对应协议的proto.in队列上面。

    func (p *Peer) readLoop(errc chan<- error) {

        defer p.wg.Done()

        for {

            msg, err := p.rw.ReadMsg()

            if err != nil {

                errc <- err

                return

            }

            msg.ReceivedAt = time.Now()

            if err = p.handle(msg); err != nil {

                errc <- err

                return

            }

        }

    }

    func (p *Peer) handle(msg Msg) error {

        switch {

        case msg.Code == pingMsg:

            msg.Discard()

            go SendItems(p.rw, pongMsg)

        case msg.Code == discMsg:

            var reason [1]DiscReason

            // This is the last message. We don't need to discard or

            // check errors because, the connection will be closed after it.

            rlp.Decode(msg.Payload, &reason)

            return reason[0]

        case msg.Code < baseProtocolLength:

            // ignore other base protocol messages

            return msg.Discard()

        default:

            // it's a subprotocol message

            proto, err := p.getProto(msg.Code)

            if err != nil {

                return fmt.Errorf("msg code out of range: %v", msg.Code)

            }

            select {

            case proto.in <- msg:

                return nil

            case <-p.closed:

                return io.EOF

            }

        }

        return nil

    }

在看看pingLoop。这个方法很简单。就是定时的发送pingMsg消息到对端。

    func (p *Peer) pingLoop() {

        ping := time.NewTimer(pingInterval)

        defer p.wg.Done()

        defer ping.Stop()

        for {

            select {

            case <-ping.C:

                if err := SendItems(p.rw, pingMsg); err != nil {

                    p.protoErr <- err

                    return

                }

                ping.Reset(pingInterval)

            case <-p.closed:

                return

            }

        }

    }

最后再看看protoRW的read和write方法。 可以看到读取和写入都是阻塞式的。

    func (rw *protoRW) WriteMsg(msg Msg) (err error) {

        if msg.Code >= rw.Length {

            return newPeerError(errInvalidMsgCode, "not handled")

        }

        msg.Code += rw.offset

        select {

case <-rw.wstart:  //等到可以写入的受在执行写入。 这难道是为了多线程控制么。

            err = rw.w.WriteMsg(msg)

            // Report write status back to Peer.run. It will initiate

            // shutdown if the error is non-nil and unblock the next write

            // otherwise. The calling protocol code should exit for errors

            // as well but we don't want to rely on that.

            rw.werr <- err

        case <-rw.closed:

            err = fmt.Errorf("shutting down")

        }

        return err

    }

    func (rw *protoRW) ReadMsg() (Msg, error) {

        select {

        case msg := <-rw.in:

            msg.Code -= rw.offset

            return msg, nil

        case <-rw.closed:

            return Msg{}, io.EOF

        }

    }

上一篇 下一篇

猜你喜欢

热点阅读