• 1 概述
• 2 相关类介绍
• 3 Unsafe.write
• 4 Unsafe.flush

1 概述

在Netty中，发送报文和读取报文都是通过Unsafe处理的，但是说到底发送和读取报文都是从java.nio.channels.SelectableChannel读取或者向其写报文，而Netty自定义Channel是对java.nio.channels.SelectableChannel的封装增强，所以Unsafe在发送和读取报文最终调用的还是Channel相关的方法。

我们在Handler中处理报文向Channel或者AbstractChannelHandlerContext写数据时（关于Handler、AbstractChannelHandlerContext可以参考笔者这篇文章），其实并没有实际向SelectableChannel中写，也就没有实际写入socket发送缓冲区，而是写入Unsafe中的缓存ChannelOutboundBuffer中，只有在调用flush方法或者注册了OP_WRITE事件后才会实际向socket缓冲中进行写入操作，这里要注意在NIO编程时一般不会向Channel注册OP_WRITE事件，Netty也一样，只会在因网络缓存满写半包之后剩下数据要及时写出去，才会注册OP_WRITE事件，让Channel在准备好可以写之后告诉自己继续写剩下的数据(关于OP_WRITE事件的理解可以参考笔者这篇文章)。

write和flush属于Outbound事件，所在最终调用pipeline中head节点（也就是HeadContext）的write和flush方法，HeadContext的write和flush方法如下（有关Outbound事件HeadContext可以参考笔者这篇文章）：

//HeadContext
 @Override
public void write(ChannelHandlerContext ctx, Object msg, ChannelPromise promise) throws Exception {
    unsafe.write(msg, promise);
}

@Override
public void flush(ChannelHandlerContext ctx) throws Exception {
    unsafe.flush();
}

可见最终调用的还是Unsafe的write和flush方法，所以前面的调用我们不再赘述，后面直接介绍Unsafe的write和flush方法。

2 相关类介绍

这里我们主要介绍Unsafe中的缓存ChannelOutboundBuffer类定义，ChannelOutboundBuffer使用单链表组织待写出的数据，链表节点为其内部类Entry，这里有一个知识点Entry使用了Netty的对象池技术（可参考笔者这篇文章），下面首先看ChannelOutboundBuffer的重要域：

//ChannelOutboundBuffer
//ChannelOutboundBuffer采用单链表组织要写出的ByteBuf，
//tailEntry用于指向最后一个加入到链表的节点
// The Entry which represents the tail of the buffer
private Entry tailEntry;
//每次调用addMessage方法向OutboundBuffer追加节点时
//如果unflushedEntry为空则会被置为最后追加的节点
//addMessage是需要判断unflushedEntry为空才置为最后追加的节点，
//而unflushedEntry会在flush方法调用后会置为空
//所以unflushedEntry不是始终指向尾节点，准确的说应该是指向最靠前
//的一个未flushed节点
// The Entry which is the first unflushed in the linked-list structure
private Entry unflushedEntry;
//每次调用flush方法时，都会尝试将当前所有节点写入socket缓存中，
//缓冲满时，则会注册OP_WRITE方法，在channel就绪时继续写，
//但是不管能够一次性写出去，调用flush方法都会将从头结点开始
//到unflushedEntry的所有节点表示为flush状态，flushedEntry置为
//最后未写出的unflushedEntry节点，unflushedEntry则置为空
// The Entry that is the first in the linked-list structure that was flushed
private Entry flushedEntry;
//每次调用flush方法时都可能会将多个节点尝试写入socket
//flushed则记录最后一个调用flush累计打算写的节点个数
// The number of flushed entries that are not written yet
private int flushed;

看完重要域之后，下面再看重要的方法：

//ChannelOutboundBuffer
/**
* Add given message to this {@link ChannelOutboundBuffer}. The given {@link ChannelPromise} will be notified once
* the message was written.
*/
//向链表中添加节点
public void addMessage(Object msg, int size, ChannelPromise promise) {
    //新建一个Entry节点实例
    Entry entry = Entry.newInstance(msg, size, total(msg), promise);
    //如果tailEntry为空，则表示是第一个节点
    if (tailEntry == null) {
        flushedEntry = null;
        tailEntry = entry;
    } else {
        //否则将新节点追加的链表最后，并更新tailEntry为新的尾节点
        Entry tail = tailEntry;
        tail.next = entry;
        tailEntry = entry;
    }
    //如果unflushedEntry为空，则更新其为最新的尾节点
    if (unflushedEntry == null) {
        unflushedEntry = entry;
    }

    //累积所有缓冲未写出的字节数
    // increment pending bytes after adding message to the unflushed arrays.
    // See https://github.com/netty/netty/issues/1619
    incrementPendingOutboundBytes(entry.pendingSize, false);
}

/**
* Add a flush to this {@link ChannelOutboundBuffer}. This means all previous added messages are marked as flushed
* and so you will be able to handle them.
*/
public void addFlush() {
    // There is no need to process all entries if there was already a flush before and no new messages
    // where added in the meantime.
    //
    // See https://github.com/netty/netty/issues/2577
    Entry entry = unflushedEntry;
    //判断unflushedEntry是否为空，放置节点个数没有发生变化，
    //但是多次调用addFlush方法
    if (entry != null) {
        //flushedEntry表示当前没有刷新任务，将flushedEntry节点
        //赋值为entry
        if (flushedEntry == null) {
            // there is no flushedEntry yet, so start with the entry
            flushedEntry = entry;
        }
        do {
            //递增待刷新节点个数
            flushed ++;
            //设置该节点对应的promise为不可取消的
            //这里使用AtomicReferenceFieldUpdater更新
            //如果更新失败，表示已经取消了
            if (!entry.promise.setUncancellable()) {
                // Was cancelled so make sure we free up memory and notify about the freed bytes
                //如果设置失败，则表示在刷新之前已经取消对该节点的
                //写操作，所以这里进行取消，然后从所有待写入字节数
                //中减去该节点字节数
                int pending = entry.cancel();
                decrementPendingOutboundBytes(pending, false, true);
            }
            //获取链表下一个元素
            entry = entry.next;
        } while (entry != null);

        // All flushed so reset unflushedEntry
        //因为所有节点都会被flush，所以设置unflushedEntry为空
        unflushedEntry = null;
    }
}

通过上面的ChannelOutboundBuffer.addFlush方法实现可知，在写入数据放入ChannelOutboundBuffer缓存之后如果想取消发送，则必须在addFlush方法调用之前取消，否则该消息会被设置为不可取消状态。

3 Unsafe.write

如上面的介绍Unsafe.write没有直接向socket缓冲写入数据，而是将数据写入Unsafe中的缓存ChannelOutboundBuffer，直接看AbstractUnsafe.write源码：

//AbstractUnsafe
//ChannelOutboundBuffer是AbstractUnsafe的一个域，在声明时初始化如下
private volatile ChannelOutboundBuffer outboundBuffer = new ChannelOutboundBuffer(AbstractChannel.this);

//写方法
@Override
public final void write(Object msg, ChannelPromise promise) {
    assertEventLoop();

    ChannelOutboundBuffer outboundBuffer = this.outboundBuffer;
    //outboundBuffer为空，表示初始化失败，这里抛异常
    if (outboundBuffer == null) {
        // If the outboundBuffer is null we know the channel was closed and so
        // need to fail the future right away. If it is not null the handling of the rest
        // will be done in flush0()
        // See https://github.com/netty/netty/issues/2362
        safeSetFailure(promise, WRITE_CLOSED_CHANNEL_EXCEPTION);
        // release message now to prevent resource-leak
        ReferenceCountUtil.release(msg);
        return;
    }

    int size;
    try {
        //过滤写出的报文，默认实现实际没有做什么操作，不展开介绍
        msg = filterOutboundMessage(msg);
        //得出报文的大小，这个后面介绍
        size = pipeline.estimatorHandle().size(msg);
        //如果小于0的话，赋值为0
        if (size < 0) {
            size = 0;
        }
    } catch (Throwable t) {
        safeSetFailure(promise, t);
        ReferenceCountUtil.release(msg);
        return;
    }

    //将消息添加到缓冲中
    outboundBuffer.addMessage(msg, size, promise);
}

上面pipeline.estimatorHandle().size(msg)默认实现如下源码所示：

//DefaultMessageSizeEstimator.HandleImpl
private static final class HandleImpl implements Handle {
    private final int unknownSize;

    private HandleImpl(int unknownSize) {
        this.unknownSize = unknownSize;
    }

    //size方法返回的就是ByteBuf的可读字节数
    @Override
    public int size(Object msg) {
        if (msg instanceof ByteBuf) {
            return ((ByteBuf) msg).readableBytes();
        }
        if (msg instanceof ByteBufHolder) {
            return ((ByteBufHolder) msg).content().readableBytes();
        }
        if (msg instanceof FileRegion) {
            return 0;
        }
        return unknownSize;
    }
}

4 Unsafe.flush

Unsafe.flush在AbstractUnsafe中实现如下：

//AbstractUnsafe
@Override
public final void flush() {
    assertEventLoop();

    ChannelOutboundBuffer outboundBuffer = this.outboundBuffer;
    if (outboundBuffer == null) {
        return;
    }
    //这个会将outboundBuffer所有待写入节点修改为flushing状态
    //具体实现在上面第二节已经介绍过
    outboundBuffer.addFlush();

    flush0();
}

@SuppressWarnings("deprecation")
protected void flush0() {
    //inFlush0在进入时被置为true，结束一次flush
    //后变为false，这里判断为true直接返回，避免
    //正在刷新中又触发调用
    if (inFlush0) {
        // Avoid re-entrance
        return;
    }

    final ChannelOutboundBuffer outboundBuffer = this.outboundBuffer;
    if (outboundBuffer == null || outboundBuffer.isEmpty()) {
        return;
    }

    inFlush0 = true;
    //AbstractUnsafe是AbstractChannel的内部类，这个方法是
    //AbstractChannel的方法，保证刷新时通道还是有效状态
    // Mark all pending write requests as failure if the channel is inactive.
    if (!isActive()) {
        try {
            //同样判断通道状态是否为打开状态
            if (isOpen()) {
                outboundBuffer.failFlushed(FLUSH0_NOT_YET_CONNECTED_EXCEPTION, true);
            } else {
                // Do not trigger channelWritabilityChanged because the channel is closed already.
                outboundBuffer.failFlushed(FLUSH0_CLOSED_CHANNEL_EXCEPTION, false);
            }
        } finally {
            inFlush0 = false;
        }
        return;
    }

    try {
        //进行向通道的实际写操作，这里的doWrite也是Channel的方法
        doWrite(outboundBuffer);
    } catch (Throwable t) {
        //一些异常处理
        if (t instanceof IOException && config().isAutoClose()) {
            /**
                * Just call {@link #close(ChannelPromise, Throwable, boolean)} here which will take care of
                * failing all flushed messages and also ensure the actual close of the underlying transport
                * will happen before the promises are notified.
                *
                * This is needed as otherwise {@link #isActive()} , {@link #isOpen()} and {@link #isWritable()}
                * may still return {@code true} even if the channel should be closed as result of the exception.
                */
            close(voidPromise(), t, FLUSH0_CLOSED_CHANNEL_EXCEPTION, false);
        } else {
            try {
                shutdownOutput(voidPromise(), t);
            } catch (Throwable t2) {
                close(voidPromise(), t2, FLUSH0_CLOSED_CHANNEL_EXCEPTION, false);
            }
        }
    } finally {
        inFlush0 = false;
    }
}

AbstractUnsafe.flush0方法中调用的doWrite其实是外部类AbstractChannel定义的方法，我们直接看其子类AbstractNioByteChannel中的实现：

//AbstractNioByteChannel
 @Override
protected void doWrite(ChannelOutboundBuffer in) throws Exception {
    //可以看下getWriteSpinCount方法的注释
    //因为每次写入可能会因为socket缓冲满而写不了数据
    //导致写入字节数为0，writeSpinCount就是控制如果
    //写入字节数为0，尝试进行多少次写入
    //也就是注释所说的：向channel写入返回直到返回非0值
    //尝试次数，默认为16，可用使用WRITE_SPIN_COUNT进行配置

    //上面的getWriteSpinCount方法注释的含义，但是看代码，每次写成功
    //一些数据（写入字节数大于0）节点会递减writeSpinCount的值
    //所以writeSpinCount也表示一次flush操作尝试写的次数
    int writeSpinCount = config().getWriteSpinCount();
    do {
        //获取outboundBuffer当前的头结点，进行写入
        Object msg = in.current();
        //如果返回为空，表示outboundBuffer所有节点
        //的数据已经全部发送
        if (msg == null) {
            //所有数据都已发送，所以取消OP_WRITE事件的注册，
            //避免select因write事件返回，但是程序实际上无
            //数据要写
            // Wrote all messages.
            clearOpWrite();
            // Directly return here so incompleteWrite(...) is not called.
            return;
        }
        //这里进行实际写操作，如果当前节点in.current写成功
        //会从outboundBuffer中删除该节点，并将in.current
        //置为后面一个节点
        writeSpinCount -= doWriteInternal(in, msg);
        //while如果writeSpinCount小于0，则不再进行写入
    } while (writeSpinCount > 0);

    //如果writeSpinCount<0，则表示没有写入完成，还有数据未写，
    //所以会注册OP_WRITE，让channel在可写时主动告诉程序，
    //应用会再次进行写尝试
    incompleteWrite(writeSpinCount < 0);
}

private int doWriteInternal(ChannelOutboundBuffer in, Object msg) throws Exception {
    //如果消息类型为ByteBuf，进入如下分支
    if (msg instanceof ByteBuf) {
        ByteBuf buf = (ByteBuf) msg;
        //该ByteBuf已经没有数据可读，直接返回
        //没有进行实际的写操作，所以writeSpinCount不递减
        if (!buf.isReadable()) {
            in.remove();
            return 0;
        }
        //进行实际的写操作，这里我们后面会列一下NioSocketChannel
        //的实现
        final int localFlushedAmount = doWriteBytes(buf);
        //如果返回值大于0，表示已经写出一部分字节
        if (localFlushedAmount > 0) {
            //更新写出进度
            in.progress(localFlushedAmount);
            //如果该ByteBuf没有可读数据，表示该ByteBuf已经全部写完
            //把它从outboundBuffer中移除，
        //这会将outboundBuffer.current节点指向当前节点后面一个节点
            if (!buf.isReadable()) {
                in.remove();
            }
            //进行了实际写操作，所以writeSpinCount递减一
            return 1;
        }
    } else if (msg instanceof FileRegion) {
        //和写ByteBuf一样的逻辑，不再介绍
        FileRegion region = (FileRegion) msg;
        if (region.transferred() >= region.count()) {
            in.remove();
            return 0;
        }

        long localFlushedAmount = doWriteFileRegion(region);
        if (localFlushedAmount > 0) {
            in.progress(localFlushedAmount);
            if (region.transferred() >= region.count()) {
                in.remove();
            }
            return 1;
        }
    } else {
        // Should not reach here.
        throw new Error();
    }
    //到这里表示实际写操作返回值小于等于0，可能socket缓冲区已满
    //所以这里返回一个特别大的数（2147483647），令doWrite直接
    //跳出while循环
    return WRITE_STATUS_SNDBUF_FULL;
}

//doWrite中如果writeSpinCount<0，则表示没有写入完成，还有数据未写，
//这里传入true
protected final void incompleteWrite(boolean setOpWrite) {
    // Did not write completely.
    //如果传入ture，表示上次写入没有完成，还有数据待写，所以注册
    //OP_WRITE事件，让通道在准备好写入时告诉自己进行写入
    if (setOpWrite) {
        setOpWrite();
    } else {
        // It is possible that we have set the write OP, woken up by NIO because the socket is writable, and then
        // use our write quantum. In this case we no longer want to set the write OP because the socket is still
        // writable (as far as we know). We will find out next time we attempt to write if the socket is writable
        // and set the write OP if necessary.
        //表示数据写入完毕，所以取消对OP_WRITE的注册
        clearOpWrite();

        // Schedule flush again later so other tasks can be picked up in the meantime
        //向channel加入一个flushTask，该任务会调用unsafe.flush0方法
        eventLoop().execute(flushTask);
    }
}
//注册OP_WRITE事件
protected final void setOpWrite() {
    final SelectionKey key = selectionKey();
    // Check first if the key is still valid as it may be canceled as part of the deregistration
    // from the EventLoop
    // See https://github.com/netty/netty/issues/2104
    if (!key.isValid()) {
        return;
    }
    final int interestOps = key.interestOps();
    if ((interestOps & SelectionKey.OP_WRITE) == 0) {
        key.interestOps(interestOps | SelectionKey.OP_WRITE);
    }
}
//取消注册的OP_WRITE事件
protected final void clearOpWrite() {
    final SelectionKey key = selectionKey();
    // Check first if the key is still valid as it may be canceled as part of the deregistration
    // from the EventLoop
    // See https://github.com/netty/netty/issues/2104
    if (!key.isValid()) {
        return;
    }
    final int interestOps = key.interestOps();
    if ((interestOps & SelectionKey.OP_WRITE) != 0) {
        key.interestOps(interestOps & ~SelectionKey.OP_WRITE);
    }
}

private final Runnable flushTask = new Runnable() {
    @Override
    public void run() {
        // Calling flush0 directly to ensure we not try to flush messages that were added via write(...) in the
        // meantime.
        ((AbstractNioUnsafe) unsafe()).flush0();
    }
};

下面AbstractNioByteChannel.doWriteBytes(ByteBuf buf)我们看下NioSocketChannel.doWriteBytes(ByteBuf buf)`的实现：

//NioSocketChannel
//可见就是向java.nio.channels.SocketChannel写数据
@Override
protected int doWriteBytes(ByteBuf buf) throws Exception {
    final int expectedWrittenBytes = buf.readableBytes();
    return buf.readBytes(javaChannel(), expectedWrittenBytes);
}

如果向通道注册了OP_WRITE事件可以在NioEventLoop.run方法select返回后，在processSelectedKey处理返回的SelectionKey的代码中会调用Unsafe.forceFlush

private void processSelectedKey(SelectionKey k, AbstractNioChannel ch) {
    ...

    try {
        ...
        // Process OP_WRITE first as we may be able to write some queued buffers and so free memory.
        if ((readyOps & SelectionKey.OP_WRITE) != 0) {
            // Call forceFlush which will also take care of clear the OP_WRITE once there is nothing left to write
            //forceFlush最终会调用Unsafe.flush0方法
            ch.unsafe().forceFlush();
        }

        ...
    } catch (CancelledKeyException ignored) {
        unsafe.close(unsafe.voidPromise());
    }
}

最后总结一下，我们在向通道进行写入时，调用write方法只是将待写入的数据放入ChannelOutboundBuffer缓冲中，实际向socket进行写出有两个地方触发，第一个就是调用Unsafe.flush方法；另一个就是发生了部分写，注册OP_WRITE之后，Selector.select返回写就绪，也会触发实际socket写操作。