OkHhttp原理分析

1.OkHttp简单使用

导包：

implementation "com.squareup.okhttp3:okhttp:4.9.0"

添加网络权限：

简单实现一个网络请求：

OkHttpClient client = new OkHttpClient(); //1.创建client

Request request = new Request.Builder() //2.创建request
        .url("https:www.baidu.com")
        .build();

Call call = client.newCall(request); //3.封装call

//4.发起请求
//4.1发起同步请求
new Thread(() -> {//同步请求不能在主线程发起
    try {
        Response response = call.execute();
        android.util.Log.d("test", "onResponse : " + response.message());//5.拿到返回值并处理
    } catch (IOException e) {
        e.printStackTrace();
    }
}).start();

//4.2异步请求
call.enqueue(new Callback() {
    @Override
    public void onFailure(@NonNull Call call1, @NonNull IOException e) {
        android.util.Log.d("test", "onFailure");
    }

    @Override
    public void onResponse(@NonNull Call call1, @NonNull Response response) throws IOException {
        android.util.Log.d("test", "onResponse : " + response.message());//5.拿到返回值并处理
    }
});

OkHttp的调用过程可以简单总结为以下几步：

1.创建client
2.通过建造者模式封装url等到request实体类
3.将request封装到call实体类
4.通过调用call的execute或者enqueue方法发起同步或者异步请求
5.处理返回值

2.OkHttp异步请求过程分析

在分析之前，我们先思考一下如果自己实现一个异步请求该如何实现，就是实现一个异步请求会遇到哪些问题。

第一 如何实现一个异步请求，异步嘛那肯定是通过子线程来处理，那么子线程的实现方式有哪些，线程池和new Thread().start()，这里OkHttp使用的是线程池，那么线程池相对于Thread有哪些好处呢，值得思索。

第二 异步请求该如何处理返回值，因为是在子线程里面处理请求，那么处理完毕后如何拿到处理结果，首先想到的是回调方法。其次就是回调方法应该放到哪里去，OkHttp是通过重写Runnable，将回调方法当做一个参数封装到Runnable里面，这种做法值得思索。

第三 既然是异步请求，通过子线程来处理请求，那么我一个请求结束后怎么保证下一个请求能被执行呢，答案是可以通过队列来保证请求的先后顺序，即一个请求完毕后去判断队列里面是否还有其他请求存在，如果有那就执行，没有就说明没有需要执行的请求。

第四 请求的个数怎么控制，包括对一个host地址的请求次数怎么限制，如果疯狂请求会有什么问题。OkHttp采用两个缓存队列，一个ready一个running解决了这问题，首先一个请求过来先加到ready队列里面，执行的时候从ready转到running队列里，转之前会判断当前running队列的请求个数是否大于64，以及对同个host地址的请求个数是否大于5个，如果大于64就不执行会阻塞在ready队列里面，等待下次调用，下次调用是什么时候呢，就是第三步里面讨论的一个请求结束后再去拿另外的请求。

第五 前面都是请求前的准备工作，具体请求逻辑该如何实现呢，一个http请求包括哪些步骤呢，建立连接，缓存处理，重定向这些，okhttp采用责任链模式来处理实际的网络请求，这种处理方式的优缺点是什么，值得思索。

1.将callback封装成一个Runnable即AsyncCall

  override fun enqueue(responseCallback: Callback) {
	check(executed.compareAndSet(false, true)) { "Already Executed" }

	callStart()
	client.dispatcher.enqueue(AsyncCall(responseCallback))
  }

2.通过dispatcher的enqueue方法，将AsyncCall添加到缓存readyAsyncCalls里面，readyAsyncCalls是一个队列，用来做缓存
3.查找缓存即findExistingCallWithHost，有的话就复用

  internal fun enqueue(call: AsyncCall) {
	synchronized(this) {
	  readyAsyncCalls.add(call)

	  // Mutate the AsyncCall so that it shares the AtomicInteger of an existing running call to
	  // the same host.
	  if (!call.call.forWebSocket) {
		val existingCall = findExistingCallWithHost(call.host)
		if (existingCall != null) call.reuseCallsPerHostFrom(existingCall)
	  }
	}
	promoteAndExecute()
  }

4.简单说就是缓存转移，即从readyAsyncCalls转移到runningAsyncCalls，并且中间插入一些判断条件，比如正在运行的最大容量，还有同一个host的请求个数不要超过5个等等。

5.缓存转移完毕之后调用AsyncCall的executeOn方法，同时传一个线程池过去
这个线程池，核心线程数为0，最大线程数是Integer的最大值，存活时间是60s

  private fun promoteAndExecute(): Boolean {
	this.assertThreadDoesntHoldLock()

	val executableCalls = mutableListOf()
	val isRunning: Boolean
	synchronized(this) {
	  val i = readyAsyncCalls.iterator()
	  while (i.hasNext()) {
		val asyncCall = i.next()

		if (runningAsyncCalls.size >= this.maxRequests) break // Max capacity.
		if (asyncCall.callsPerHost.get() >= this.maxRequestsPerHost) continue // Host max capacity.

		i.remove()
		asyncCall.callsPerHost.incrementAndGet()
		executableCalls.add(asyncCall)
		runningAsyncCalls.add(asyncCall)
	  }
	  isRunning = runningCallsCount() > 0
	}

	for (i in 0 until executableCalls.size) {
	  val asyncCall = executableCalls[i]
	  asyncCall.executeOn(executorService)
	}

	return isRunning
  }

线程池的实现，此处的线程池的配置是没有核心线程，总的线程个数是没有限制的，也就是说都是非核心线程，并且个数没有限制，非核心线程等待的时间是60秒，并且使用的任务队列是SynchronousQueue，它是一个没有容量的阻塞队列，只会当里面没有任务的时候，才能往里面放任务，当放完之后，只能等它的任务被取走才能放，这不就是jdk里面提供的Executors.newCachedThreadPool线程池吗，可能是okhttp想自己定义线程工厂的参数吧，定义线程的名字。

  @get:Synchronized
  @get:JvmName("executorService") val executorService: ExecutorService
	get() {
	  if (executorServiceOrNull == null) {
		executorServiceOrNull = ThreadPoolExecutor(0, Int.MAX_VALUE, 60, TimeUnit.SECONDS,
			SynchronousQueue(), threadFactory("$okHttpName Dispatcher", false))
	  }
	  return executorServiceOrNull!!
	}

6.AsyncCall的executeOn方法里面简单说就是一个线程池的执行过程，因为AsyncCall实际是一个Runnable，所以线程池运行后要看run方法

fun executeOn(executorService: ExecutorService) {
  client.dispatcher.assertThreadDoesntHoldLock()

  var success = false
  try {
    executorService.execute(this)
    success = true
  } catch (e: RejectedExecutionException) {
    val ioException = InterruptedIOException("executor rejected")
    ioException.initCause(e)
    noMoreExchanges(ioException)
    responseCallback.onFailure(this@RealCall, ioException)
  } finally {
    if (!success) {
      client.dispatcher.finished(this) // This call is no longer running!
    }
  }
}

override fun run() {
  threadName("OkHttp ${redactedUrl()}") {
    var signalledCallback = false
    timeout.enter()
    try {
      val response = getResponseWithInterceptorChain()
      signalledCallback = true
      responseCallback.onResponse(this@RealCall, response)
    } catch (e: IOException) {
      if (signalledCallback) {
        // Do not signal the callback twice!
        Platform.get().log("Callback failure for ${toLoggableString()}", Platform.INFO, e)
      } else {
        responseCallback.onFailure(this@RealCall, e)
      }
    } catch (t: Throwable) {
      cancel()
      if (!signalledCallback) {
        val canceledException = IOException("canceled due to $t")
        canceledException.addSuppressed(t)
        responseCallback.onFailure(this@RealCall, canceledException)
      }
      throw t
    } finally {
      client.dispatcher.finished(this)
    }
  }
}

7.run方法里面主要是通过getResponseWithInterceptorChain()，通过责任链模式拿到response，然后调用封装的responseCallback实现异步请求的onResponse和onFailure方法调用

8.最后通过dispatcher的finished方法，清除runningAsyncCalls里的缓存数据

  internal fun finished(call: AsyncCall) {
	call.callsPerHost.decrementAndGet()
	finished(runningAsyncCalls, call)
  }

  /** Used by [Call.execute] to signal completion. */
  internal fun finished(call: RealCall) {
	finished(runningSyncCalls, call)
  }

  private fun  finished(calls: Deque, call: T) {
	val idleCallback: Runnable?
	synchronized(this) {
	  if (!calls.remove(call)) throw AssertionError("Call wasn't in-flight!")
	  idleCallback = this.idleCallback
	}

	val isRunning = promoteAndExecute()

	if (!isRunning && idleCallback != null) {
	  idleCallback.run()
	}
  }

3.OkHttp中的拦截器

在getResponseWithInterceptorChain()方法里，创建一个list集合，然后将拦截器依次添加到集合中，通过RealInterceptorChain的proceed发起调用，依次返回reponse，最终给到AsyncCall的run方法里回调函数去操作reponse。

  @Throws(IOException::class)
  internal fun getResponseWithInterceptorChain(): Response {
	// Build a full stack of interceptors.
	val interceptors = mutableListOf()
	interceptors += client.interceptors
	interceptors += RetryAndFollowUpInterceptor(client)
	interceptors += BridgeInterceptor(client.cookieJar)
	interceptors += CacheInterceptor(client.cache)
	interceptors += ConnectInterceptor
	if (!forWebSocket) {
	  interceptors += client.networkInterceptors
	}
	interceptors += CallServerInterceptor(forWebSocket)

	val chain = RealInterceptorChain(
		call = this,
		interceptors = interceptors,
		index = 0,
		exchange = null,
		request = originalRequest,
		connectTimeoutMillis = client.connectTimeoutMillis,
		readTimeoutMillis = client.readTimeoutMillis,
		writeTimeoutMillis = client.writeTimeoutMillis
	)

	var calledNoMoreExchanges = false
	try {
	  val response = chain.proceed(originalRequest)
	  if (isCanceled()) {
		response.closeQuietly()
		throw IOException("Canceled")
	  }
	  return response
	} catch (e: IOException) {
	  calledNoMoreExchanges = true
	  throw noMoreExchanges(e) as Throwable
	} finally {
	  if (!calledNoMoreExchanges) {
		noMoreExchanges(null)
	  }
	}
  }

根据源码可知，一共七个拦截器：

• addInterceptor(Interceptor)，这是由开发者设置的，会按照开发者的要求，在所有的拦截器处理之前进行最早的拦截处理，比如一些公共参数，Header都可以在这里添加。

• RetryAndFollowUpInterceptor，这里会对连接做一些初始化工作，以及请求失败的充实工作，重定向的后续请求工作。跟他的名字一样，就是做重试工作还有一些连接跟踪工作。

• BridgeInterceptor，这里会为用户构建一个能够进行网络访问的请求，同时后续工作将网络请求回来的响应Response转化为用户可用的Response，比如添加文件类型，content-length计算添加，gzip解包。

• CacheInterceptor，这里主要是处理cache相关处理，会根据OkHttpClient对象的配置以及缓存策略对请求值进行缓存，而且如果本地有了可?的Cache，就可以在没有网络交互的情况下就返回缓存结果。

• ConnectInterceptor，这里主要就是负责建立连接了，会建立TCP连接或者TLS连接，以及负责编码解码的HttpCodec

• networkInterceptors，这里也是开发者自己设置的，所以本质上和第一个拦截器差不多，但是由于位置不同，所以用处也不同。这个位置添加的拦截器可以看到请求和响应的数据了，所以可以做一些网络调试。

• CallServerInterceptor，这里就是进行网络数据的请求和响应了，也就是实际的网络I/O操作，通过socket读写数据。

现在让我们来debug跑一遍请求的拦截过程。

RealInterceptorChain创建完毕，可以看到里面有五个拦截器，以及request信息

RetryAndFollowUpInterceptor执行到proceed时的request

BridgeInterceptor执行到proceed时的request，可以看到请求头多了很多东西，也就是BridgeInterceptor会添加一些请求头信息

CacheInterceptor执行到proceed时的request，可以看到多了一个lazyCacheControl

ConnectInterceptor执行到proceed时的request，这个request没有什么变化

CallServerInterceptor就没有proceed方法了，而是直接进行网络数据请求和响应了，最终会依次向上传递response，可以看到CallSercaerInterceptor返回的reponse为

CallServerInterceptor之后回调到ConnectInterceptor，ConnectInterceptor的reponse也没加什么新东西。

ConnectInterceptor之后回调到CacheInterceptor，CacheInterceptor又加了两个缓存，一个是cacheResponse为null，一个是networkResponse.

CacheInterceptor之后回调到BridgeInterceptor,BridgeInterceptor有一个gzip的解压操作，在这里

然后就是body的GZipSource

BridgeInterceptor之后回调到RetryAndFollowUpInterceptor,这个倒是没有什么操作了，然后就返回到getResponseWithInterceptorChain()方法里，交给callback处理了。

可以看到在RealCall的getResponseWithInterceptorChain()方法里，最终拿到的reponse如图

4.OkHttp怎么实现连接池

分析连接池之前让我们先思考一下几个问题：

第一点：如何实现一个http请求，okhttp通过socket实现http请求。TCP报文包括请求头和TCP数据，只需要将http的报文填充到TCP数据中。这一步谁来实现，答案是BridgeInterceptor

第二点：如何建立socket连接，这一步谁来实现，答案是ConnectInterceptor，socket连接就是TCP连接，需要三次握手，因此建立连接是一个耗时过程。
那么优化点就是建立连接池，复用socket连接，怎么去复用socket连接，okhttp4.9版本使用的是ConcurrentLinkedQueue，老版本用的是ArrayDeque一个双端队列，

ConcurrentLinkedQueue是基于链接节点的、线程安全的队列。并发访问不需要同步。因为它在队列的尾部添加元素并从头部删除它们，所以只要不需要知道队列的大小， ConcurrentLinkedQueue 对公共集合的共享访问就可以工作得很好。

第三点：如何去复用一个socket连接，每次建立socket连接之前要去队列里面查看有没有可用连接，有的话就复用，没有的话就创建一个新链接。那么问题来了，怎么判断一个连接是不是可以被复用的链接呢。okhttp是怎么做的，可以看下面的注释，8个判断条件。

  internal fun isEligible(address: Address, routes: List?): Boolean {
    assertThreadHoldsLock()

    // If this connection is not accepting new exchanges, we're done.
    if (calls.size >= allocationLimit || noNewExchanges) return false

    // If the non-host fields of the address don't overlap, we're done.
    if (!this.route.address.equalsNonHost(address)) return false

    // If the host exactly matches, we're done: this connection can carry the address.
    if (address.url.host == this.route().address.url.host) {
      return true // This connection is a perfect match.
    }

    // At this point we don't have a hostname match. But we still be able to carry the request if
    // our connection coalescing requirements are met. See also:
    // https://hpbn.co/optimizing-application-delivery/#eliminate-domain-sharding
    // https://daniel.haxx.se/blog/2016/08/18/http2-connection-coalescing/

    // 1. This connection must be HTTP/2.
    if (http2Connection == null) return false

    // 2. The routes must share an IP address.
    if (routes == null || !routeMatchesAny(routes)) return false

    // 3. This connection's server certificate's must cover the new host.
    if (address.hostnameVerifier !== OkHostnameVerifier) return false
    if (!supportsUrl(address.url)) return false

    // 4. Certificate pinning must match the host.
    try {
      address.certificatePinner!!.check(address.url.host, handshake()!!.peerCertificates)
    } catch (_: SSLPeerUnverifiedException) {
      return false
    }

    return true // The caller's address can be carried by this connection.
  }

第四点：无用的连接如何清理掉，什么时候触发清理操作，采用什么策略清理，如何判定一个连接是无用连接。

• 什么时候触发清理操作：
  添加一个新连接到队列时，可以去触发清理操作
• 采用什么策略清理
  这个和如何判定连接是无效的放在一起的。
• 如何判定一个连接是无用连接
  就是java在进行垃圾回收的时候有一种方法叫标记清除法，就是这个对象被引用就加个标记，没有被引用就把这个标记清理掉，最后垃圾回收的时候通过判断是否有标记来回收对象。这里连接也可以套用，就是查看这个连接有没有被引用，被谁引用呢，封装的Call对象，就是有多少Call引用了这个connection,这样连接又可以分为正在使用和idle空闲连接，然后通过连接空闲的时间和设置的最大空闲连接数来判定当前这个连接要不要被清理掉。

到这里整个okhttp的请求，以及连接复用可能出现的问题已经清楚了。具体再来分析一下代码实现：

1. okhttp如何通过socket连接发送http请求：

http是应用层的协议，socket是TCP的协议，TCP协议通过什么样的组装可以组成Http协议

IP 作为以太网的直接底层，IP 的头部和数据合起来作为以太网的数据，同样的 TCP/UDP 的头部和数据合起来作为 IP 的数据，HTTP 的头部和数据合起来作为 TCP/UDP 的数据。

我们来看OkHttp如何通过Interceptor添加请求头部的。要看这个BridgeInterceptor

@Throws(IOException::class)
override fun intercept(chain: Interceptor.Chain): Response {
    val userRequest = chain.request()
    val requestBuilder = userRequest.newBuilder()

    val body = userRequest.body
    if (body != null) {
      val contentType = body.contentType()
      if (contentType != null) {
        requestBuilder.header("Content-Type", contentType.toString())
      }

      val contentLength = body.contentLength()
      if (contentLength != -1L) {
        requestBuilder.header("Content-Length", contentLength.toString())
        requestBuilder.removeHeader("Transfer-Encoding")
      } else {
        requestBuilder.header("Transfer-Encoding", "chunked")
        requestBuilder.removeHeader("Content-Length")
      }
    }

    if (userRequest.header("Host") == null) {
      requestBuilder.header("Host", userRequest.url.toHostHeader())
    }

    if (userRequest.header("Connection") == null) {
      requestBuilder.header("Connection", "Keep-Alive")
    }

    // If we add an "Accept-Encoding: gzip" header field we're responsible for also decompressing
    // the transfer stream.
    var transparentGzip = false
    if (userRequest.header("Accept-Encoding") == null && userRequest.header("Range") == null) {
      transparentGzip = true
      requestBuilder.header("Accept-Encoding", "gzip")
    }

    val cookies = cookieJar.loadForRequest(userRequest.url)
    if (cookies.isNotEmpty()) {
      requestBuilder.header("Cookie", cookieHeader(cookies))
    }

    if (userRequest.header("User-Agent") == null) {
      requestBuilder.header("User-Agent", userAgent)
    }

    val networkResponse = chain.proceed(requestBuilder.build())

    cookieJar.receiveHeaders(userRequest.url, networkResponse.headers)

    val responseBuilder = networkResponse.newBuilder()
        .request(userRequest)

    if (transparentGzip &&
        "gzip".equals(networkResponse.header("Content-Encoding"), ignoreCase = true) &&
        networkResponse.promisesBody()) {
      val responseBody = networkResponse.body
      if (responseBody != null) {
        val gzipSource = GzipSource(responseBody.source())
        val strippedHeaders = networkResponse.headers.newBuilder()
            .removeAll("Content-Encoding")
            .removeAll("Content-Length")
            .build()
        responseBuilder.headers(strippedHeaders)
        val contentType = networkResponse.header("Content-Type")
        responseBuilder.body(RealResponseBody(contentType, -1L, gzipSource.buffer()))
      }
    }

    return responseBuilder.build()
}

2. socket连接建立的具体过程：

object ConnectInterceptor : Interceptor {
  @Throws(IOException::class)
  override fun intercept(chain: Interceptor.Chain): Response {
    val realChain = chain as RealInterceptorChain
    val exchange = realChain.call.initExchange(chain) //1.ConnectInterceptor创建exchange
    val connectedChain = realChain.copy(exchange = exchange)
    return connectedChain.proceed(realChain.request)
  }
}

  /** Finds a new or pooled connection to carry a forthcoming request and response. */
  internal fun initExchange(chain: RealInterceptorChain): Exchange {
    synchronized(this) {
      check(expectMoreExchanges) { "released" }
      check(!responseBodyOpen)
      check(!requestBodyOpen)
    }

    val exchangeFinder = this.exchangeFinder!!
    val codec = exchangeFinder.find(client, chain)//2.RealCall#initExchange()方法通过exchangeFinder.find()方法创建codec
    val result = Exchange(this, eventListener, exchangeFinder, codec)
    this.interceptorScopedExchange = result
    this.exchange = result
    synchronized(this) {
      this.requestBodyOpen = true
      this.responseBodyOpen = true
    }

    if (canceled) throw IOException("Canceled")
    return result
  }

  fun find(
    client: OkHttpClient,
    chain: RealInterceptorChain
  ): ExchangeCodec {
    try {
      val resultConnection = findHealthyConnection(//3.ExchangeFinder#find()方法通过findHealthyConnection方法创建resultConnection
          connectTimeout = chain.connectTimeoutMillis,
          readTimeout = chain.readTimeoutMillis,
          writeTimeout = chain.writeTimeoutMillis,
          pingIntervalMillis = client.pingIntervalMillis,
          connectionRetryEnabled = client.retryOnConnectionFailure,
          doExtensiveHealthChecks = chain.request.method != "GET"
      )
      return resultConnection.newCodec(client, chain)
    } catch (e: RouteException) {
      trackFailure(e.lastConnectException)
      throw e
    } catch (e: IOException) {
      trackFailure(e)
      throw RouteException(e)
    }
  }

  @Throws(IOException::class)
  private fun findHealthyConnection(
    connectTimeout: Int,
    readTimeout: Int,
    writeTimeout: Int,
    pingIntervalMillis: Int,
    connectionRetryEnabled: Boolean,
    doExtensiveHealthChecks: Boolean
  ): RealConnection {
    while (true) {
      val candidate = findConnection(//4.ExchangeFinder#find()方法通过findConnection拿到一个可用的connection
          connectTimeout = connectTimeout,
          readTimeout = readTimeout,
          writeTimeout = writeTimeout,
          pingIntervalMillis = pingIntervalMillis,
          connectionRetryEnabled = connectionRetryEnabled
      )

      // Confirm that the connection is good.
      if (candidate.isHealthy(doExtensiveHealthChecks)) {
        return candidate
      }

      // If it isn't, take it out of the pool.
      candidate.noNewExchanges()

      // Make sure we have some routes left to try. One example where we may exhaust all the routes
      // would happen if we made a new connection and it immediately is detected as unhealthy.
      if (nextRouteToTry != null) continue

      val routesLeft = routeSelection?.hasNext() ?: true
      if (routesLeft) continue

      val routesSelectionLeft = routeSelector?.hasNext() ?: true
      if (routesSelectionLeft) continue

      throw IOException("exhausted all routes")
    }
}

  @Throws(IOException::class)
  private fun findConnection(
    connectTimeout: Int,
    readTimeout: Int,
    writeTimeout: Int,
    pingIntervalMillis: Int,
    connectionRetryEnabled: Boolean
  ): RealConnection {
    if (call.isCanceled()) throw IOException("Canceled")

    // Attempt to reuse the connection from the call.
    val callConnection = call.connection // This may be mutated by releaseConnectionNoEvents()!
    if (callConnection != null) {
      var toClose: Socket? = null
      synchronized(callConnection) {
        if (callConnection.noNewExchanges || !sameHostAndPort(callConnection.route().address.url)) {
          toClose = call.releaseConnectionNoEvents()
        }
      }

      // If the call's connection wasn't released, reuse it. We don't call connectionAcquired() here
      // because we already acquired it.
      if (call.connection != null) {
        check(toClose == null)
        return callConnection
      }

      // The call's connection was released.
      toClose?.closeQuietly()
      eventListener.connectionReleased(call, callConnection)
    }

    // We need a new connection. Give it fresh stats.
    refusedStreamCount = 0
    connectionShutdownCount = 0
    otherFailureCount = 0

    // Attempt to get a connection from the pool.
    if (connectionPool.callAcquirePooledConnection(address, call, null, false)) {//5.通过RealConnectionPool#callAcquirePooledConnection()方法callAcquirePooledConnection复用连接池
      val result = call.connection!!
      eventListener.connectionAcquired(call, result)
      return result
    }

    // Nothing in the pool. Figure out what route we'll try next.
    val routes: List?
    val route: Route
    if (nextRouteToTry != null) {
      // Use a route from a preceding coalesced connection.
      routes = null
      route = nextRouteToTry!!
      nextRouteToTry = null
    } else if (routeSelection != null && routeSelection!!.hasNext()) {
      // Use a route from an existing route selection.
      routes = null
      route = routeSelection!!.next()
    } else {
      // Compute a new route selection. This is a blocking operation!
      var localRouteSelector = routeSelector
      if (localRouteSelector == null) {
        localRouteSelector = RouteSelector(address, call.client.routeDatabase, call, eventListener)
        this.routeSelector = localRouteSelector
      }
      val localRouteSelection = localRouteSelector.next()
      routeSelection = localRouteSelection
      routes = localRouteSelection.routes

      if (call.isCanceled()) throw IOException("Canceled")

      // Now that we have a set of IP addresses, make another attempt at getting a connection from
      // the pool. We have a better chance of matching thanks to connection coalescing.
      if (connectionPool.callAcquirePooledConnection(address, call, routes, false)) {
        val result = call.connection!!
        eventListener.connectionAcquired(call, result)
        return result
      }

      route = localRouteSelection.next()
    }

    // Connect. Tell the call about the connecting call so async cancels work.
    val newConnection = RealConnection(connectionPool, route)
    call.connectionToCancel = newConnection
    try {
      newConnection.connect(//6.连接池里没有可用连接，新建一个连接
          connectTimeout,
          readTimeout,
          writeTimeout,
          pingIntervalMillis,
          connectionRetryEnabled,
          call,
          eventListener
      )
    } finally {
      call.connectionToCancel = null
    }
    call.client.routeDatabase.connected(newConnection.route())

    // If we raced another call connecting to this host, coalesce the connections. This makes for 3
    // different lookups in the connection pool!
    if (connectionPool.callAcquirePooledConnection(address, call, routes, true)) {
      val result = call.connection!!
      nextRouteToTry = route
      newConnection.socket().closeQuietly()
      eventListener.connectionAcquired(call, result)
      return result
    }

    synchronized(newConnection) {
      connectionPool.put(newConnection) //7.将连接添加到连接池
      call.acquireConnectionNoEvents(newConnection)
    }

    eventListener.connectionAcquired(call, newConnection)
    return newConnection //8.返回可用连接
  }

3. 判断一个连接是否可以被复用：

通过RealConnectionPool#callAcquirePooledConnection()方法，判断一个连接是否可以被复用

  /**
   * Attempts to acquire a recycled connection to [address] for [call]. Returns true if a connection
   * was acquired.
   *
   * If [routes] is non-null these are the resolved routes (ie. IP addresses) for the connection.
   * This is used to coalesce related domains to the same HTTP/2 connection, such as `square.com`
   * and `square.ca`.
   */
  fun callAcquirePooledConnection(
    address: Address,
    call: RealCall,
    routes: List?,
    requireMultiplexed: Boolean
  ): Boolean {
    for (connection in connections) {
      synchronized(connection) {
        if (requireMultiplexed && !connection.isMultiplexed) return@synchronized
        if (!connection.isEligible(address, routes)) return@synchronized
        call.acquireConnectionNoEvents(connection)
        return true
      }
    }
    return false
  }

4. 清理无用连接：

触发时机:

RealConnectionPool#put()方法，添加新的连接到连接池中，同时触发清理空闲连接操作

  fun put(connection: RealConnection) {
    connection.assertThreadHoldsLock()

    connections.add(connection)
    cleanupQueue.schedule(cleanupTask) //添加新connection时触发清理操作
  }

判定一个connection是否要被清理：

RealConnectionPool#cleanup()方法，先通过pruneAndGetAllocationCount()方法拿到当前连接是工作连接还是空闲连接，然后针对空闲连接结合空闲时间以及最大空闲连接数，判断是否要被清理。

 fun cleanup(now: Long): Long {
    var inUseConnectionCount = 0
    var idleConnectionCount = 0
    var longestIdleConnection: RealConnection? = null
    var longestIdleDurationNs = Long.MIN_VALUE

    // Find either a connection to evict, or the time that the next eviction is due.
    for (connection in connections) {
      synchronized(connection) {
        // If the connection is in use, keep searching.
        if (pruneAndGetAllocationCount(connection, now) > 0) {
          inUseConnectionCount++
        } else {
          idleConnectionCount++

          // If the connection is ready to be evicted, we're done.
          val idleDurationNs = now - connection.idleAtNs
          if (idleDurationNs > longestIdleDurationNs) {
            longestIdleDurationNs = idleDurationNs
            longestIdleConnection = connection
          } else {
            Unit
          }
        }
      }
    }

    when {
      longestIdleDurationNs >= this.keepAliveDurationNs
          || idleConnectionCount > this.maxIdleConnections -> {
        // We've chosen a connection to evict. Confirm it's still okay to be evict, then close it.
        val connection = longestIdleConnection!!
        synchronized(connection) {
          if (connection.calls.isNotEmpty()) return 0L // No longer idle.
          if (connection.idleAtNs + longestIdleDurationNs != now) return 0L // No longer oldest.
          connection.noNewExchanges = true
          connections.remove(longestIdleConnection)
        }

        connection.socket().closeQuietly()
        if (connections.isEmpty()) cleanupQueue.cancelAll()

        // Clean up again immediately.
        return 0L
      }

      idleConnectionCount > 0 -> {
        // A connection will be ready to evict soon.
        return keepAliveDurationNs - longestIdleDurationNs
      }

      inUseConnectionCount > 0 -> {
        // All connections are in use. It'll be at least the keep alive duration 'til we run
        // again.
        return keepAliveDurationNs
      }

      else -> {
        // No connections, idle or in use.
        return -1
      }
    }
  }

  /**
   * Prunes any leaked calls and then returns the number of remaining live calls on [connection].
   * Calls are leaked if the connection is tracking them but the application code has abandoned
   * them. Leak detection is imprecise and relies on garbage collection.
   */
  private fun pruneAndGetAllocationCount(connection: RealConnection, now: Long): Int {
    connection.assertThreadHoldsLock()

    val references = connection.calls
    var i = 0
    while (i < references.size) {
      val reference = references[i]

      if (reference.get() != null) {
        i++
        continue
      }

      // We've discovered a leaked call. This is an application bug.
      val callReference = reference as CallReference
      val message = "A connection to ${connection.route().address.url} was leaked. " +
          "Did you forget to close a response body?"
      Platform.get().logCloseableLeak(message, callReference.callStackTrace)

      references.removeAt(i)
      connection.noNewExchanges = true

      // If this was the last allocation, the connection is eligible for immediate eviction.
      if (references.isEmpty()) {
        connection.idleAtNs = now - keepAliveDurationNs
        return 0
      }
    }

    return references.size
}

参考文档

谈谈OKHttp的几道面试题

面试官:Okhttp连接池是咋实现的?

OkHttp中的Socket连接

Java并发编程之ConcurrentLinkedQueue详解