[置顶] 深入分析UI 上层事件处理核心机制 Choreographer

深入分析UI 上层事件处理核心机制 Choreographer

结论写在前面：Choreographer就是一个消息处理器，根据vsync 信号 来计算frame，而计算frame的方式就是处理三种回调，包括事件回调、动画回调、绘制回调。这三种事件在消息输入、加入动画、准备绘图layout 等动作时均会发给Choreographer。

下面来看分析过程

看过一些源码后，发现ui 绘制的管理，调度都是通过Choreographer这个类。

1 Choreographer 是什么？有什么？

Choreographer 是个普通类，final 表示不能被继承修改其行为。

public final class Choreographer

单例模式持有一个本地进程的单例对象，

    private static final ThreadLocal<Choreographer> sThreadInstance =
            new ThreadLocal<Choreographer>()

该对象必须持有looper，意味着将使用消息队列

        protected Choreographer initialValue() {
            Looper looper = Looper.myLooper();
            if (looper == null) {
                throw new IllegalStateException("The current thread must have a looper!");
            }
            return new Choreographer(looper);
        }

持有一个handler 对象

    private final FrameHandler mHandler;

这个handler对象仅处理3种事件：

    private final class FrameHandler extends Handler {
        public FrameHandler(Looper looper) {
            super(looper);
        }

        @Override
        public void handleMessage(Message msg) {
            switch (msg.what) {
                case MSG_DO_FRAME:
                    doFrame(System.nanoTime(), 0);
                    break;
                case MSG_DO_SCHEDULE_VSYNC:
                    doScheduleVsync();
                    break;
                case MSG_DO_SCHEDULE_CALLBACK:
                    doScheduleCallback(msg.arg1);
                    break;
            }
        }
    }

在scheduleFrameLocked中事件被静态变量USE_VSYNC分开，也就是系统仅使用MSG_DO_SCHEDULE_VSYNC或MSG_DO_FRAME。

查一下概念，的确符合，DO_FRAME 是帧刷新，SCHEDULE_VSYNC是 垂直同步刷新。

在android4.1上加入的VSYNC 特性，并使用三重缓冲大幅改善了android 图像方面的性能。

这里有篇文章讲的很清楚：http://www.androidpolice.com/2012/07/12/getting-to-know-android-4-1-part-3-project-butter-how-it-works-and-what-it-added/

    private void scheduleFrameLocked(long now) {
            if (USE_VSYNC) {
                if (isRunningOnLooperThreadLocked()) {
                    scheduleVsyncLocked();
                } else {
                    Message msg = mHandler.obtainMessage(MSG_DO_SCHEDULE_VSYNC);
                    msg.setAsynchronous(true);
                    mHandler.sendMessageAtFrontOfQueue(msg);
                }
            } else {
                Message msg = mHandler.obtainMessage(MSG_DO_FRAME);
                msg.setAsynchronous(true);
                mHandler.sendMessageAtTime(msg, nextFrameTime);
            }

2 从VSYNC 开始case MSG_DO_SCHEDULE_VSYNC:

    void doScheduleVsync() {
        synchronized (mLock) {
            if (mFrameScheduled) {
                scheduleVsyncLocked();
            }
        }
    }

    private void scheduleVsyncLocked() {
        mDisplayEventReceiver.scheduleVsync();
    }

    public void scheduleVsync() {
        if (mReceiverPtr == 0) {
            Log.w(TAG, "Attempted to schedule a vertical sync pulse but the display event "
                    + "receiver has already been disposed.");
        } else {
            nativeScheduleVsync(mReceiverPtr);
        }
    }

一路调用到nativeScheduleVsync();

native的东西先不往下看，回头来看mDisplayEventReceiver

    private final class FrameDisplayEventReceiver extends DisplayEventReceiver
            implements Runnable 

观察一下方法名都是onXX ，一看就是回调的节奏

    // Called from native code.
    @SuppressWarnings("unused")
    private void dispatchVsync(long timestampNanos, int builtInDisplayId, int frame) {
        onVsync(timestampNanos, builtInDisplayId, frame);
    }

看到这里的注释，就明确了这个类就是被native回调了。回调时调用onVsync();而这个方法时在FrameDisplayEventReceiver中重写的。

        @Override
        public void onVsync(long timestampNanos, int builtInDisplayId, int frame) {
            mTimestampNanos = timestampNanos;
            mFrame = frame;
            Message msg = Message.obtain(mHandler, this);
            msg.setAsynchronous(true);
            mHandler.sendMessageAtTime(msg, timestampNanos / NANOS_PER_MS);

        @Override
        public void run() {
            mHavePendingVsync = false;
            doFrame(mTimestampNanos, mFrame);
        }

在onsync中把自己的runnable 加到消息队列中执行，这里使用异步消息，调用绘制do_frame()方法。

    void doFrame(long frameTimeNanos, int frame) {

        synchronized (mLock) {
            //省略一些赋值
            //可能时跳帧的情况，直接调用vsync，并return
            if (frameTimeNanos < mLastFrameTimeNanos) {
                if (DEBUG) {
                    Log.d(TAG, "Frame time appears to be going backwards.  May be due to a "
                            + "previously skipped frame.  Waiting for next vsync.");
                }
                scheduleVsyncLocked();
                return;
            }

        }
        //先后处理事件回调、动画回调、绘制回调
        doCallbacks(Choreographer.CALLBACK_INPUT, frameTimeNanos);
        doCallbacks(Choreographer.CALLBACK_ANIMATION, frameTimeNanos);
        doCallbacks(Choreographer.CALLBACK_TRAVERSAL, frameTimeNanos);
    }

在do_frame()中判断有跳帧可能性，直接继续vsync。

否则开始新的一帧的绘制，依次处理事件回调、动画回调、绘制回调

在doCallbacks() 中先根据当前时间去除队列中第一个有效的回调，然后依次处理这些回调。

    void doCallbacks(int callbackType, long frameTimeNanos) {
        CallbackRecord callbacks;
        synchronized (mLock) {
            // We use "now" to determine when callbacks become due because it's possible
            // for earlier processing phases in a frame to post callbacks that should run
            // in a following phase, such as an input event that causes an animation to start.
            final long now = SystemClock.uptimeMillis();
            //按时间取队列头
            callbacks = mCallbackQueues[callbackType].extractDueCallbacksLocked(now);
            if (callbacks == null) {
                return;
            }
            mCallbacksRunning = true;
        }
        try {
            for (CallbackRecord c = callbacks; c != null; c = c.next) {
                if (DEBUG) {
                    Log.d(TAG, "RunCallback: type=" + callbackType
                            + ", action=" + c.action + ", token=" + c.token
                            + ", latencyMillis=" + (SystemClock.uptimeMillis() - c.dueTime));
                }
                //回调事件
                c.run(frameTimeNanos);
            }
        } finally {
            synchronized (mLock) {
                mCallbacksRunning = false;
                do {
                    final CallbackRecord next = callbacks.next;
                    recycleCallbackLocked(callbacks);
                    callbacks = next;
                } while (callbacks != null);
            }
        }
    }

回调的数据结构是CallbackRecord，一个单向链表。根据token 的标志，来回调doframe 或者runnable

    private static final class CallbackRecord {
        public CallbackRecord next;
        public long dueTime;
        public Object action; // Runnable or FrameCallback
        public Object token;

        public void run(long frameTimeNanos) {
            if (token == FRAME_CALLBACK_TOKEN) {
                ((FrameCallback)action).doFrame(frameTimeNanos);
            } else {
                ((Runnable)action).run();
            }
        }
    }

从这里可以看到，处理的事件分为两类：一类是doframe 事件，另一类是runnable 方法。接着，就来看一下到底是加的这些事件。

3. 哪里会使用Choreographer？

在《Android 动画animation 深入分析》http://blog.csdn.net/farmer_cc/article/details/18259117中分析到scheduleAnimation 的时候就是调用的android.view.Choreographer.postCallback(int, Runnable, Object) 方法。查看该方法的调用，在UI 绘制、和其他和动画相关的类中均有调用。这里并没有全部列出来。

scheduleTraversals() : void - android.view.ViewRootImpl

scheduleAnimation() : void - android.animation.ValueAnimator.AnimationHandler

scheduleAnimationLocked() : void - com.android.server.wm.WindowManagerService

postOnAnimation(Runnable) : void - android.view.View

在postCallbackDelayedInternal()中 把新来的事件加入到队列中，并根据时间来判断是去frame 还是直接回调

    private void postCallbackDelayedInternal(int callbackType,
            Object action, Object token, long delayMillis) {
        if (DEBUG) {
            Log.d(TAG, "PostCallback: type=" + callbackType
                    + ", action=" + action + ", token=" + token
                    + ", delayMillis=" + delayMillis);
        }

        synchronized (mLock) {
            final long now = SystemClock.uptimeMillis();
            final long dueTime = now + delayMillis;
            // 入队列
            mCallbackQueues[callbackType].addCallbackLocked(dueTime, action, token);

            if (dueTime <= now) {
                scheduleFrameLocked(now);
            } else {
                // 直接回调
                Message msg = mHandler.obtainMessage(MSG_DO_SCHEDULE_CALLBACK, action);
                msg.arg1 = callbackType;
                msg.setAsynchronous(true);
                mHandler.sendMessageAtTime(msg, dueTime);
            }
        }

至此，事件的来龙去脉就明确了。Choreographer就是一个消息处理器，根据vsync 信号 来计算frame，而计算frame的方式就是处理三种回调，包括事件回调、动画回调、绘制回调。这三种事件在消息输入、加入动画、准备绘图layout 等动作时均会发给Choreographer。

写在后面：结合前文《Android 动画animation 深入分析》http://blog.csdn.net/farmer_cc/article/details/18259117中可以更多了解动画相关内容