解析Activity启动-生命周期角度

书接上回，在 AMS侧解析Activity启动篇章中，我们大致梳理了 Activity 启动过程中涉及到堆栈的整体流程，本篇再让我们从 Activity 生命周期的角度来梳理整个流程
顺着上一篇解析Activity启动-堆栈篇的流程，在启动Activity时，会在 resumeTopActivityInnerLocked() 函数中调用 startPausingLocked() 函数对发起者Activity触发 Pause 动作，那么我们就按照启动顺序，进行逐一分析

① onPause

在前一篇文章中，我们也有涉及 startPausingLocked() 函数的分析，但是并没有特别细致和深入，那么在此处，我们顺着该函数，较为细致的解析一下

1.1 `resumeTopActivityInnerLocked()`

frameworks\base\services\core\java\com\android\server\am\ActivityStack.java

@GuardedBy("mService")
private boolean resumeTopActivityInnerLocked(ActivityRecord prev, ActivityOptions options) {
    ...
    // 如果当前Stack中的mResumedActivity不为空，也就是当前处于 resume 状态的Activity不为空，命中if
   //去异步 pause mResumedActivity
    if (mResumedActivity != null) {
        if (DEBUG_STATES)
            Slog.d(TAG_STATES,
                   "resumeTopActivityLocked: Pausing " + mResumedActivity);
        // pausing当前Stack的mResumedActivity，此处返回值也是false，就是这里需要先pause再去resume
        pausing |= startPausingLocked(userLeaving, false, next, false);
    }
    ...
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14

首先是调用起点，剔除与Pause流程无关的代码；可以看到，只有在对应 stack中存在 mResumedActivity 时，才会去触发进行Pause流程，很好理解，如果对应stack中并不存在Resume状态的Activity，那也就不存在Activity需要进行Pause，此处也重点关注一下传入的参数

1.2 `startPausingLocked()`

frameworks\base\services\core\java\com\android\server\am\ActivityStack.java

    // pausing当前Stack中的mResumedActivity对象，只有在该mResumedActivity存在，
    // 并且异步调用finish Activity的才会返回true
    final boolean startPausingLocked(boolean userLeaving, boolean uiSleeping,
            ActivityRecord resuming, boolean pauseImmediately) {
        // 判断当前Stack的mPausingActivity是否为空,如果不为空，代表中有正在进行pause的Activity,
        //且对应display不是sleep状态，那么就完成mPausingActivity的pause流程，
        //并将 mPausingActivity置空
        if (mPausingActivity != null) {
            // 如果当前display不处于Sleeping状态，命中if
            // 完成mPausingActivity的Pause，其实就是设置相关的标志位，如state等,
            //并且将mPausingActivity置空
            if (!shouldSleepActivities()) {
                completePauseLocked(false, resuming);
            }
        }
        // 拿到当前Stack的mResumedActivity
        ActivityRecord prev = mResumedActivity;

        // mResumedActivity等于空，直接返回，就不需要pause了
        if (prev == null) {
			...
            return false;
        }

        // mResumedActivity就是需要resume的mStartActivity，直接返回，也不用pause了
        if (prev == resuming) {
            return false;
        }

        // 将当前resumed的Activity赋值给mPausingActivity和mLastPausedActivity
        //也就是移交状态，用这两个变量记录prev，代表他开始进入pause流程
        mPausingActivity = prev;
        mLastPausedActivity = prev;
        // 如果当前resumed的Activity设置了FLAG_ACTIVITY_NO_HISTORY或者FLAG_NO_HISTORY，那么就将
        // 当前resumed的Activity保存到mLastNoHistoryActivity中
        //代表该Activity不需要被保存在historyActivity列表中
        mLastNoHistoryActivity = (prev.intent.getFlags() & Intent.FLAG_ACTIVITY_NO_HISTORY) != 0
                || (prev.info.flags & ActivityInfo.FLAG_NO_HISTORY) != 0 ? prev : null;
        // 设置当前resumed的Activity的生命周期为Pausing，中间状态，正在进行pause动作
        prev.setState(PAUSING, "startPausingLocked");
        ...

        // 当前resumed的Activity进程线程都存在，命中if
        // 构建PauseActivityItem对象，去执行pause对应的Activity
        //需要特别声明，在这里最后就会调用到对应 Activity 进程中，执行 onPause() 方法
        if (prev.app != null && prev.app.thread != null) {
            try {
                ...

                //prev就是当前stack中的 mResumedActivity,也就是正处于resumed状态的Activity
                //最后是放到 prev.app.thread 去执行的
                mService.getLifecycleManager().scheduleTransaction(prev.app.thread, prev.appToken,
                        PauseActivityItem.obtain(prev.finishing, userLeaving,
                                prev.configChangeFlags, pauseImmediately));
            } catch (Exception e) {
                //捕获到异常时，将相关标识都清空
                mPausingActivity = null;
                mLastPausedActivity = null;
                mLastNoHistoryActivity = null;
            }
        //如果进程或者线程不存在，直接将mPausingActivity、mLastPausedActivity、mLastNoHistoryActivity置空
        } else {
            mPausingActivity = null;
            mLastPausedActivity = null;
            mLastNoHistoryActivity = null;
        }

        //当前不是sleep状态，去获取 wakeLock,确保不会进入sleep状态
        if (!uiSleeping && !mService.isSleepingOrShuttingDownLocked()) {
            mStackSupervisor.acquireLaunchWakelock();
        }

        // 如果此时mPausingActivity不为空，命中if
        // 根据传入的pauseImmediately决定是否要同步finish Activity
        if (mPausingActivity != null) {
            //当前流程下传入的uiSleeping是false，命中if
            //暂停mPausingActivity按键的分发
            if (!uiSleeping) {
                prev.pauseKeyDispatchingLocked();
            } else if (DEBUG_PAUSE) {
                Slog.v(TAG_PAUSE, "Key dispatch not paused for screen off");
            }

            // 是否需要立即pause，如果需要命中if
            // 将当前Stack中的mPausingActivity设置生命周期为Paused,并且将
            //mPausingActivity置空,返回false
            if (pauseImmediately) {
                completePauseLocked(false, resuming);
                return false;

            // 如果不需要立即pause，那么500ms后再去completePauseLock，这里返回 true
            //构建出的消息最后也会调用到 completePauseLock() 函数，区别在于在此处最后还会尝试去
            //resume mFocusedStack 最顶层的Activity
            } else {
                schedulePauseTimeout(prev);
                return true;
            }

        // 如果mPausingActivity为空，命中else
        // mPausingActivity为null只有两种情况：
        // 1. prev进程线程存在的情况下，scheduleTransaction()抛出异常，从而设置了scheduleTransaction为null
        // 2. prev进程线程不存在，设置mPausingActivity为null，此种情况下即已经完成了prev的pause
        } else {
            ...
        }
    }
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106

解析该函数，除去执行条件判断，主要完成了如下的动作：
1. 用 mPausingActivity和 mLastPausedActivity 保存下待Pause的Activity，也就是 mResumedActivity ；根据 mResumedActivity 中相关标志位(FLAG_ACTIVITY_NO_HISTORY\FLAG_NO_HISTORY)判断是否需要将其保存到 mLastNoHistoryActivity 中代表不需要将该Activity保存到历史堆栈中
2. 将 mResumedActivity 的生命周期设置为 PAUSING 状态
3. 构建 PauseActivityItem 对象去执行 Pause 流程
4. 根据传入的 pauseImmediately 参数，决定是否立即执行 completePauseLocked() 函数，在当前流程中是会延迟去complete，也就是在当前流程中调用的是 schedulePauseTimeout() 函数
在以上流程中，我们重点关注的是第三步和第四步，接下来就逐一分析

1.2.1 `PauseActivityItem`

ClientLifecycleManager.scheduleTransaction()

frameworks\base\services\core\java\com\android\server\am\ClientLifecycleManager.java

//解析一下入参：
//client : mResumedActivity对应的ActivityThread中的内部类 ApplicationThread
//activityToken：mResumedActivity中的 appToken 成员变量
//stateRequest：构造的PauseActivityItem对象
void scheduleTransaction(@NonNull IApplicationThread client, @NonNull IBinder activityToken,
                         @NonNull ActivityLifecycleItem stateRequest) throws RemoteException {
    //先构造出 clientTransaction 对象
    final ClientTransaction clientTransaction = transactionWithState(client, activityToken,
                                                                     stateRequest);
    //调用重载函数
    scheduleTransaction(clientTransaction);
}

void scheduleTransaction(ClientTransaction transaction) throws RemoteException {
    //获取到传入的clientTransaction对象中的mClient变量
    final IApplicationThread client = transaction.getClient();
    //调用传入的 transaction 的schedule()方法
    transaction.schedule();
    //如果获取到的client不是继承binder的，那么transaction需要回收
    if (!(client instanceof Binder)) {
        transaction.recycle();
    }
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

可以看到在 scheduleTransaction() 函数中，先根据入参构建出了 ClientTransaction 对象；然后调用重载函数，调用到 ClientTransaction.schedule() 函数

ClientTransaction.schedule()

frameworks\base\core\java\android\app\servertransaction\ClientTransaction.java
```
public void schedule() throws RemoteException {
    //调用 IApplicationThread中的方法
    mClient.scheduleTransaction(this);
}
1
2
3
4
```
- mClient 就是mResumedActivity对应的ActivityThread中的内部类 ApplicationThread

ActivityThread$ApplicationThread.scheduleTransaction()

frameworks\base\core\java\android\app\ActivityThread.java

@Override
public void scheduleTransaction(ClientTransaction transaction) throws RemoteException {
    //又走到ActivityThread中的同名方法中,这里是调用到ActivityThread的父类中去
    ActivityThread.this.scheduleTransaction(transaction);
}
1
2
3
4
5

这里就是调用到外部类，也就是 ActivityThread.scheduleTransaction() 函数中，本质是调用到 ActivityThread 的父类 ClientTransactionHandler 中

ClientTransactionHandler.scheduleTransaction()

frameworks\base\core\java\android\app\ClientTransactionHandler.java
```
void scheduleTransaction(ClientTransaction transaction) {
    //先执行 preExecute()
    transaction.preExecute(this);
    //再通过handle发送EXECUTE_TRANSACTION执行相应操作
    sendMessage(ActivityThread.H.EXECUTE_TRANSACTION, transaction);
}
1
2
3
4
5
6
```
- 可以看到先去执行了 ClientTransaction.preExecute() 函数，在当前流程中，preExecute() 没有执行任何动作，在此处先不做分析，在后续create\resume 流程中再去细说；再通过Handle，发送 EXECUTE_TRANSACTION进行下一步处理

handleMessage()

frameworks\base\core\java\android\app\ActivityThread.java

public void handleMessage(Message msg) {
    ...
    switch (msg.what) {
        ...
		case EXECUTE_TRANSACTION:
            //取出ClientTransaction对象
    		final ClientTransaction transaction = (ClientTransaction) msg.obj;
            //调用TransactionExecutor去触发执行ClientTransaction里面的流程
    		mTransactionExecutor.execute(transaction);
            //如果是系统ActivityThread，那么需要回收ClientTransaction对象
    		if (isSystem()) {
            	transaction.recycle();
        	}
    	break;
        ...
    }
    ...
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

在handle中又会借助 TransactionExecutor 对象去触发执行 ClientTransaction

TransactionExecutor.execute()

frameworks\base\core\java\android\app\servertransaction\TransactionExecutor.java

public void execute(ClientTransaction transaction) {

    //先执行 callback 中的execute
    executeCallbacks(transaction);

    //再执行 lifecycleItem的execute
    executeLifecycleState(transaction);
    //清空 pendingActions 列表
    mPendingActions.clear();
}
1
2
3
4
5
6
7
8
9
10

可以看到在 TransactionExecutor.execute() 中，会先去触发执行 callback 中的流程，而因为当前流程下，没有callback对象，所以跳过callback步骤，直接看 executeLifecycleState() 函数

TransactionExecutor.executeLifecycleState()

frameworks\base\core\java\android\app\servertransaction\TransactionExecutor.java

在这之前先说明一下 ClientTransaction 中保存的相关成员变量，具体的流程可以自行分析 ClientTransaction 的构建
- mClient：mResumedActivity对应的ActivityThread中的内部类 ApplicationThread
- mActivityToken：mResumedActivity中的 appToken 成员变量
- mLifecycleStateRequest：构建的 PauseActivityItem 对象

private void executeLifecycleState(ClientTransaction transaction) {
    //拿到ClientTransaction中的ActivityLifecycleItem对象，确保不为空
    //当前lifecycleItem就是PauseActivityItem对象
    final ActivityLifecycleItem lifecycleItem = transaction.getLifecycleStateRequest();
    ...

    //拿到ClientTransaction中的token,也就是 mResumedActivity
    //中的 appToken 成员变量
    final IBinder token = transaction.getActivityToken();
    //通过appToken拿到ActivityClientRecord对象，在mResumedActivity 的
    //ActivityThread中获取到
    final ActivityClientRecord r = mTransactionHandler.getActivityClient(token);

    ...

    //调用cycleToPath(),去补齐需要执行的生命周期，此时 r 的生命周期为 ON_RESUME,
    //PauseActivityItem的目标生命周期为ON_PAUSE,这之间没有中间状态
    cycleToPath(r, lifecycleItem.getTargetState(), true /* excludeLastState */);

    //开始执行，当前流程下，就是执行 PauseActivityItem.execute()
    lifecycleItem.execute(mTransactionHandler, token, mPendingActions);
    lifecycleItem.postExecute(mTransactionHandler, token, mPendingActions);
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

抛开对应变量的获取，在该函数中就是在去执行 PauseActivityItem.execute() 和 PauseActivityItem.postExecute()；其中的 cycleToPath() 函数因为当前流程中并没有涉及，我们放到后面 ON_START 流程中进行解析

PauseActivityItem.execute()

frameworks\base\core\java\android\app\servertransaction\PauseActivityItem.java

public void execute(ClientTransactionHandler client, IBinder token,
                    PendingTransactionActions pendingActions) {
    Trace.traceBegin(TRACE_TAG_ACTIVITY_MANAGER, "activityPause");
    //调用到ActivityThread中的handlePauseActivity() 函数
    client.handlePauseActivity(token, mFinished, mUserLeaving, mConfigChanges, pendingActions,
                               "PAUSE_ACTIVITY_ITEM");
    Trace.traceEnd(TRACE_TAG_ACTIVITY_MANAGER);
}
1
2
3
4
5
6
7
8

这里也就又走到了 mResumedActivity 的 ActivityThread 中了

ActivityThread.handlePauseActivity()

frameworks\base\core\java\android\app\ActivityThread.java

@Override
public void handlePauseActivity(IBinder token, boolean finished, boolean userLeaving,
                                int configChanges, PendingTransactionActions pendingActions, String reason) {
    //根据token(也就是ActivityRecord中的appToken对象)拿到对应的 ActivityClientRecord 对象
    ActivityClientRecord r = mActivities.get(token);
    //首先保证能够拿到对应的 ActivityClientRecord 对象
    if (r != null) {
        //如果userLeaving为true,那么最后就会去调用到Activity中的 onUserInteraction()和
        //onUserLeaveHint() 函数
        if (userLeaving) {
            performUserLeavingActivity(r);
        }

        //保存下configChanges的值
        r.activity.mConfigChangeFlags |= configChanges;
        //进行paused
        performPauseActivity(r, finished, reason, pendingActions);

        //如果当前是 Android 3.0 或者之前的版本，命中if
        if (r.isPreHoneycomb()) {
            QueuedWork.waitToFinish();
        }
        //mSomeActivitiesChanged置true，代表当前有Activity状态变化
        mSomeActivitiesChanged = true;
    }
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

该函数中完成的操作也比较一目了然，顺着流程，调用到 performPauseActivity()

ActivityThread.performPauseActivity()

frameworks\base\core\java\android\app\ActivityThread.java

private Bundle performPauseActivity(ActivityClientRecord r, boolean finished, String reason,
                                    PendingTransactionActions pendingActions) {
    //如果传入的 ActivityClientRecord 状态已经是 paused，那么就不需要
    //再进行pause
    if (r.paused) {
        ...
    }
    //如果传入的Activity是finish状态的，那么设置activity.mFinished为true
    if (finished) {
        r.activity.mFinished = true;
    }

    ...

    //继续调用去pause
    performPauseActivityIfNeeded(r, reason);

    //拿到该Activity的所有 onPause 监听
    ArrayList<OnActivityPausedListener> listeners;
    synchronized (mOnPauseListeners) {
        listeners = mOnPauseListeners.remove(r.activity);
    }
    //调用所有监听的 onPaused() 方法
    int size = (listeners != null ? listeners.size() : 0);
    for (int i = 0; i < size; i++) {
        listeners.get(i).onPaused(r.activity);
    }
    ...

    return shouldSaveState ? r.state : null;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31

同样继续调用 performPauseActivityIfNeeded() 完成Pause操作

ActivityThread.performPauseActivityIfNeeded()

frameworks\base\core\java\android\app\ActivityThread.java

private void performPauseActivityIfNeeded(ActivityClientRecord r, String reason) {
    //如果已经是 paused 状态，那么直接返回
    if (r.paused) {
        return;
    }

    try {
        //设置mCalled初始值为false
        r.activity.mCalled = false;
        //借助Instrumentation去调用onPause
        mInstrumentation.callActivityOnPause(r.activity);
        //mCalled为false是会命中if，代表没有调用基类 Activity 中的 onPause() 方法,
        //当 mCalled 为false时，会抛出异常，也就是说必须调用基类Activity的onPause()
        if (!r.activity.mCalled) {
            throw new SuperNotCalledException("Activity " + safeToComponentShortString(r.intent)
                                              + " did not call through to super.onPause()");
        }
    } catch (SuperNotCalledException e) {
        throw e;
    } catch (Exception e) {
        ...
    }
    //将该Activity设置状态为 pause
    r.setState(ON_PAUSE);
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

接下来就是借助 Instrumentation 再调用到 Activity 中的 onPause() 方法中，后续流程比较简单，就不做展开分析

ActivityStack.activityPausedLocked()

frameworks\base\services\core\java\com\android\server\am\ActivityStack.java

在前面我们还留了个小尾巴，就是 PauseActivityItem.postExecute() 方法，在该方法中最后会调用到 AMS.activityPaused() 中，然后又会调用到 mResumedActivity 所属的Stack中的 activityPausedLocked()方法中，接下来再看一下该函数

final void activityPausedLocked(IBinder token, boolean timeout) {

    //如果对应stack和task中存在token所对应的Activity，那么此时r就不为空
    final ActivityRecord r = isInStackLocked(token);
    //能够正确获取到r时，命中if
    if (r != null) {
        //移除 PAUSE_TIMEOUT_MSG 消息
        mHandler.removeMessages(PAUSE_TIMEOUT_MSG, r);
        //如果此时mPausingActivity就是传入的Activity，那么直接去complete pause mPausingActivity
        if (mPausingActivity == r) {
            mService.mWindowManager.deferSurfaceLayout();
            try {
                //进行complete Pause，此处传入的resumeNext为true，代表需要进行resume Activity动作
                completePauseLocked(true /* resumeNext */, null /* resumingActivity */);
            } finally {
                mService.mWindowManager.continueSurfaceLayout();
            }
            return;
        //如果mPausingActivity不是传入的Activity，命中else
        } else {
            //如果当前传入的Activity生命周期是PAUSING，那么需要设置其生命周期为PAUSED
            if (r.isState(PAUSING)) {
                r.setState(PAUSED, "activityPausedLocked");
                if (r.finishing) {
                    finishCurrentActivityLocked(r, FINISH_AFTER_VISIBLE, false,
                                                "activityPausedLocked");
                }
            }
        }
    }
    //校正系统中所有Activity的可见性
    mStackSupervisor.ensureActivitiesVisibleLocked(null, 0, !PRESERVE_WINDOWS);
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

在该函数中，首先会判断能够根据传入的 ActivityRecord.appToken 在对应的 stack 和 Task 中获取到对应的 ActivityRecord 对象，在可以获取到并且系统中的 mPausingActivity 就是当前获取到的 ActivityRecord 时，会接着去调用 completePauseLocked() 去完成 Pause 操作，并且根据此处传参 resumeNext = true ，还会去 Resume 对应 stack 最顶部的Activity

ActivityStack.completePauseLocked()

frameworks\base\services\core\java\com\android\server\am\ActivityStack.java

// 将当前Stack中的mPausingActivity给设置生命周期为Paused,并且将mPausingActivity置空，并且由
//resumeNext 参数决定，是否需要去 resume Activity
private void completePauseLocked(boolean resumeNext, ActivityRecord resuming) {
    // 拿到当前Stack中的mPausingActivity
    ActivityRecord prev = mPausingActivity;
    if (DEBUG_PAUSE)
        Slog.v(TAG_PAUSE, "Complete pause: " + prev);

    // 如果该Stack中的mPausingActivity不为空,命中if
    // 结束该mPausingActivity，并将mPausingActivity置为空
    if (prev != null) {
        // 设置mPausingActivity
        prev.setWillCloseOrEnterPip(false);
        // 判断mPausingActivity生命周期是否是STOPPING
        final boolean wasStopping = prev.isState(STOPPING);
        // 设置mPausingActivity的生命周期为PAUSED
        prev.setState(PAUSED, "completePausedLocked");
        // 如果mPausingActivity是finishing状态,那么就去finish他
        if (prev.finishing) {
            if (DEBUG_PAUSE)
                Slog.v(TAG_PAUSE, "Executing finish of activity: " + prev);
            // 结束 mPausingActivity，如果该mPausingActivity已从history中移除，那么此时拿到的返回值prev为空
            // 否则还为mPausingActivity
            prev = finishCurrentActivityLocked(prev, FINISH_AFTER_VISIBLE, false,
                                               "completedPausedLocked");
        //如果mPausingActivity所属的app还在，命中else if
        //根据相关标志执行对应动作
        } else if (prev.app != null) {
            if (DEBUG_PAUSE)
                Slog.v(TAG_PAUSE, "Enqueue pending stop if needed: " + prev
                       + " wasStopping=" + wasStopping + " visible=" + prev.visible);
            //将 mPausingActivity 从 mActivitiesWaitingForVisibleActivity列表中移除
            if (mStackSupervisor.mActivitiesWaitingForVisibleActivity.remove(prev)) {
                if (DEBUG_SWITCH || DEBUG_PAUSE)
                    Slog.v(TAG_PAUSE,
                           "Complete pause, no longer waiting: " + prev);
            }
            //判断是否在paused完成后，进行relaunch操作
            if (prev.deferRelaunchUntilPaused) {
                prev.relaunchActivityLocked(false /* andResume */,
                                            prev.preserveWindowOnDeferredRelaunch);
                //判断 mPausingActivity 是否是处在 Stopping 状态，如果是，则重新将生命周期设置为 Stopping
            } else if (wasStopping) {
                prev.setState(STOPPING, "completePausedLocked");
                //如果 mPausingActivity 不可见或者 需要进入sleep/shut down，命中else if
                //将 mPausingActivity 添加到 mStoppingActivities 列表中，并且直接触发该Activity的stop操作
            } else if (!prev.visible || shouldSleepOrShutDownActivities()) {
                prev.setDeferHidingClient(false);

                addToStopping(prev, true /* scheduleIdle */, false /* idleDelayed */);
            }
            //除上面的情况下，都直接将prev置空
        } else {
           
            prev = null;
        }
		...
        //将 mPausingActivity 置空
        mPausingActivity = null;
    }

    // 如果传入的resumeNext为true，命中if，会去resume当前mFocusedStack上最顶层的Activity
    if (resumeNext) {
        // 拿到 mFocusedStack
        final ActivityStack topStack = mStackSupervisor.getFocusedStack();
        //不是sleep的，去resume mFocusedStack最顶层的Activity
        if (!topStack.shouldSleepOrShutDownActivities()) {
            mStackSupervisor.resumeFocusedStackTopActivityLocked(topStack, prev, null);
        } else {
            checkReadyForSleep();
            ActivityRecord top = topStack.topRunningActivityLocked();
            if (top == null || (prev != null && top != prev)) {
                mStackSupervisor.resumeFocusedStackTopActivityLocked();
            }
        }
    }

    // 如果mPausingActivity不为空，命中if
    // 检测mPausingActivity所在Process的情况
    if (prev != null) {
        ...
    }
    
    //如果 存在自由窗口Stack 或者 可见性发生变化时，命中if
    //需要通知相关的Task Stack改变，并将 mAppVisibilitiesChangedSinceLastPause 置回false
    if (mStackSupervisor.mAppVisibilitiesChangedSinceLastPause
        || getDisplay().hasPinnedStack()) {
        mService.mTaskChangeNotificationController.notifyTaskStackChanged();
        mStackSupervisor.mAppVisibilitiesChangedSinceLastPause = false;
    }

    // 确保传入的resuming Activity可见
    mStackSupervisor.ensureActivitiesVisibleLocked(resuming, 0, !PRESERVE_WINDOWS);
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94

在该函数中，就会去将 mPausingActivity 的生命周期设置为 PAUSED，并且将 mPausingActivity 置空，代表 Pause 完成，然后根据传入的 resumeNext 参数决定是否需要在此处触发 resume Activity 的动作，最后会去校正系统中Activity的可见性

至此 PauseActivityItem 对象下的 Pause 流程就全部梳理完毕了

1.2.2 `schedulePauseTimeout()`

schedulePauseTimeout() 流程较为简单，其实就是通过 Handle 延迟500ms后再去调用 activityPausedLocked()，之后和上面的流程基本相同，此处就不展开分析了
从目前来看，此时的 mPausingActivity.visible 还是等于 true 的
ActivityStack.resumeTopActivityInnerLocked() 中的 startPausingLocked() 中调用 onPaused()
- mService.getLifecycleManager().scheduleTransaction(prev.app.thread, prev.appToken, PauseActivityItem.obtain(prev.finishing, userLeaving, prev.configChangeFlags, pauseImmediately));
- PauseActivityItem.execute()
  - client.handlePauseActivity(token, mFinished, mUserLeaving, mConfigChanges, pendingActions, "PAUSE_ACTIVITY_ITEM");
  1. 这个client就是 scheduleTransaction(prev.app.thread…）中的 prev.app.thread 也就是prev的线程中去调用的
  2. token 就是 prev.appToken
  3. mFinished 就是 prev.finishing
  4. mUserLeaving 就是 userLeaving
  5. mConfigChanges 就是 prev.configChangeFlags
    - performPauseActivity()
      - performPauseActivityIfNeeded()
        mInstrumentation.callActivityOnPause()
        activity.performPause()
        onPause()

② oncreate

根据上篇文章的分析，可以知道 onCreate() 流程的起点是在 realStartActivityLocked() 中进入的，我们就从该函数关键代码处开始解析

2.1 `realStartActivityLocked()`

frameworks\base\services\core\java\com\android\server\am\ActivityStackSupervisor.java

final boolean realStartActivityLocked(ActivityRecord r, ProcessRecord app,
                                      boolean andResume, boolean checkConfig) throws RemoteException {
    ...
    final ClientTransaction clientTransaction = ClientTransaction.obtain(app.thread,
                                                                         r.appToken);
    //构建出LaunchActivityItem作为callback添加到ClientTransaction对象中
    clientTransaction.addCallback(LaunchActivityItem.obtain(new Intent(r.intent),
                                                            System.identityHashCode(r), r.info,
                                                            // TODO: Have this take the merged configuration instead of separate global
                                                            // and override configs.
                                                            mergedConfiguration.getGlobalConfiguration(),
                                                            mergedConfiguration.getOverrideConfiguration(), r.compat,
                                                            r.launchedFromPackage, task.voiceInteractor, app.repProcState, r.icicle,
                                                            r.persistentState, results, newIntents, mService.isNextTransitionForward(),
                                                            profilerInfo));
	//设置Activity所需的最终状态
    final ActivityLifecycleItem lifecycleItem;
    //如果Activity最后是需要resume的，那么构建出ResumeActivityItem作为lifecycleItem
    if (andResume) {
        lifecycleItem = ResumeActivityItem.obtain(mService.isNextTransitionForward());
    } else {
        lifecycleItem = PauseActivityItem.obtain();
    }
    //将lifecycleItem保存到clientTransaction中
    clientTransaction.setLifecycleStateRequest(lifecycleItem);

    //开始执行clientTransaction，这里最后就会调用TransactionExecutor.execute()方法
    //而在这个方法中，会先去执行callback，在此处即LaunchActivityItem.execute()，完成后
    //再去执行lifecycleItem.execute()
    mService.getLifecycleManager().scheduleTransaction(clientTransaction);
    ...
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32

排除掉与当前主题无关的代码，可以看到在该函数中，同样构建了 clientTransaction 对象和 LaunchActivityItem 对象，不过和 [onPause](① onPause) 流程不同的是在当前流程中，LaunchActivityItem 对象是作为 callback 被添加到 clientTransaction 对象中，最后同样调用 scheduleTransaction() 触发，这部分流程和上面并无差异，也不再详细解析，直接看最后的执行流程

2.2 `TransactionExecutor.executeCallbacks()`

frameworks\base\core\java\android\app\servertransaction\TransactionExecutor.java

@VisibleForTesting
public void executeCallbacks(ClientTransaction transaction) {
    //拿到ClientTransaction中的callback对象
    final List<ClientTransactionItem> callbacks = transaction.getCallbacks();
    //如果没有callback，直接返回
    if (callbacks == null) {
        return;
    }
    log("Resolving callbacks");

    //通过 ActivityRecord.appToken变量拿到对应的 ActivityClientRecord 对象
    final IBinder token = transaction.getActivityToken();
    ActivityClientRecord r = mTransactionHandler.getActivityClient(token);

    ...

    //取出每个callback去触发
    final int size = callbacks.size();
    for (int i = 0; i < size; ++i) {
        ...
        //执行execute()和postExecute()
        item.execute(mTransactionHandler, token, mPendingActions);
        item.postExecute(mTransactionHandler, token, mPendingActions);
        ...
    }
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

其实和上面的流程很类似，同样是调用到callback.execute() 和 callback.postExecute()，而当前流程下就是调用 LaunchActivityItem.execute() 和 LaunchActivityItem.postExecute()，而针对 LaunchActivityItem 对象来说， postExecute() 函数是未实现的，所以无需关心，只关注 execute() 函数即可

2.3 `LaunchActivityItem.execute()`

frameworks\base\core\java\android\app\servertransaction\LaunchActivityItem.java

@Override
public void execute(ClientTransactionHandler client, IBinder token,
                    PendingTransactionActions pendingActions) {
    Trace.traceBegin(TRACE_TAG_ACTIVITY_MANAGER, "activityStart");
    //根据传入的数据构建出 ActivityClientRecord 对象
    ActivityClientRecord r = new ActivityClientRecord(token, mIntent, mIdent, mInfo,
                                                      mOverrideConfig, mCompatInfo, mReferrer, mVoiceInteractor, mState, mPersistentState,
                                                      mPendingResults, mPendingNewIntents, mIsForward,
                                                      mProfilerInfo, client);
    //调用到待启动的Activity的 ActivityThread.handleLaunchActivity() 函数
    client.handleLaunchActivity(r, pendingActions, null /* customIntent */);
    Trace.traceEnd(TRACE_TAG_ACTIVITY_MANAGER);
}
1
2
3
4
5
6
7
8
9
10
11
12
13

可以看到，针对经历完整生命周期启动的Activity，其对应的 ActivityClientRecord 对象是在 LaunchActivityItem 对象中进行创建的，后续继续调用其 ActivityThread.handleLaunchActivity() 函数去继续启动

2.4 `ActivityThread.handleLaunchActivity()`

frameworks\base\core\java\android\app\ActivityThread.java

@Override
public Activity handleLaunchActivity(ActivityClientRecord r,
                                     PendingTransactionActions pendingActions, Intent customIntent) {
    //移除GC_WHEN_IDLE消息，存在GC任务的前提下，移除mGcIdler
    unscheduleGcIdler();
    //标识activity状态变化，这个只有在初始化之前和GC之后才会为false
    mSomeActivitiesChanged = true;

    //如果存在profilerInfo，那么设置profilerInfo
    if (r.profilerInfo != null) {
        mProfiler.setProfiler(r.profilerInfo);
        mProfiler.startProfiling();
    }

    //处理 config 的变化
    handleConfigurationChanged(null, null);

    //如果没有禁用 Render,那么调用earlyInitEGL()
    if (!ThreadedRenderer.sRendererDisabled) {
        GraphicsEnvironment.earlyInitEGL();
    }
    //在WindowManagerGlobal中获取到 WMS 的IBinder，用于和WMS进行通讯
    WindowManagerGlobal.initialize();

    //开始启动 Activity,启动成功，则返回对应的Activity
    final Activity a = performLaunchActivity(r, customIntent);

    //如果启动成功，则会命中if
    if (a != null) {
        r.createdConfig = new Configuration(mConfiguration);
        reportSizeConfigurations(r);
        //如果当前的Activity不是finish状态，且pendingActions不为空,
        //那么需要将当前Activity的状态存到 pendingActions 中，代表后续可以进行恢复
        if (!r.activity.mFinished && pendingActions != null) {
            pendingActions.setOldState(r.state);
            pendingActions.setRestoreInstanceState(true);
            pendingActions.setCallOnPostCreate(true);
        }
    //如果启动不成功，命中else，结束该Activity
    } else {
        try {
            ActivityManager.getService()
                .finishActivity(r.token, Activity.RESULT_CANCELED, null,
                                Activity.DONT_FINISH_TASK_WITH_ACTIVITY);
        } catch (RemoteException ex) {
            throw ex.rethrowFromSystemServer();
        }
    }

    //返回activity对象
    return a;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52

在该函数中，会进行新启动Activity的一些初始环境准备，然后就会调用 performLaunchActivity() 函数根据传入的 ActivityClientRecord 对象去构建出 Activity 对象

2.5 `ActivityThread.performLaunchActivity()`

frameworks\base\core\java\android\app\ActivityThread.java

private Activity performLaunchActivity(ActivityClientRecord r, Intent customIntent) {
    //根据传入的ActivityClientRecord获取到activityInfo
    ActivityInfo aInfo = r.activityInfo;
    //如果传入的clientRecord没有packageInfo，那么尝试获取到对应的packageInfo
    if (r.packageInfo == null) {
        r.packageInfo = getPackageInfo(aInfo.applicationInfo, r.compatInfo,
                                       Context.CONTEXT_INCLUDE_CODE);
    }

    //拿到 component
    ComponentName component = r.intent.getComponent();
    if (component == null) {
        component = r.intent.resolveActivity(
            mInitialApplication.getPackageManager());
        r.intent.setComponent(component);
    }

    if (r.activityInfo.targetActivity != null) {
        component = new ComponentName(r.activityInfo.packageName,
                                      r.activityInfo.targetActivity);
    }

    //创建出 ContextImpl 对象
    ContextImpl appContext = createBaseContextForActivity(r);
    Activity activity = null;
    try {
        //通过 ContextImpl 获取到 classLoader
        java.lang.ClassLoader cl = appContext.getClassLoader();
        //在Instrumentation中借助ClassLoader反射创建出 Activity 对象
        activity = mInstrumentation.newActivity(
            cl, component.getClassName(), r.intent);
        //严格模式下，需要对activity进行计数
        StrictMode.incrementExpectedActivityCount(activity.getClass());
        r.intent.setExtrasClassLoader(cl);
        r.intent.prepareToEnterProcess();
        if (r.state != null) {
            r.state.setClassLoader(cl);
        }
    } catch (Exception e) {
        ...
    }

    try {
        Application app = r.packageInfo.makeApplication(false, mInstrumentation);
        
        //如果activity创建成功，命中if
        if (activity != null) {
            CharSequence title = r.activityInfo.loadLabel(appContext.getPackageManager());
            Configuration config = new Configuration(mCompatConfiguration);
            if (r.overrideConfig != null) {
                config.updateFrom(r.overrideConfig);
            }
            if (DEBUG_CONFIGURATION) Slog.v(TAG, "Launching activity "
                                            + r.activityInfo.name + " with config " + config);
            //构建出window对象,如果当前的 Activity 需要保留窗口，且待删除的窗口不为空
            //那么就将待删除的窗口赋值给window，然后将待删除窗口置空
            Window window = null;
            if (r.mPendingRemoveWindow != null && r.mPreserveWindow) {
                window = r.mPendingRemoveWindow;
                r.mPendingRemoveWindow = null;
                r.mPendingRemoveWindowManager = null;
            }
            appContext.setOuterContext(activity);
            //activity启动前 window 相关的变量构建，构建了 PhoneWindow 和 WindowManagerImpl
            //其中 r.token 是个 IBinder，对应的就是 AMS 中的ActivityRecord，此处存在待删除窗口时
            //传入的window才不为空
            activity.attach(appContext, this, getInstrumentation(), r.token,
                            r.ident, app, r.intent, r.activityInfo, title, r.parent,
                            r.embeddedID, r.lastNonConfigurationInstances, config,
                            r.referrer, r.voiceInteractor, window, r.configCallback);

            if (customIntent != null) {
                activity.mIntent = customIntent;
            }
            r.lastNonConfigurationInstances = null;
            //检查网络规则
            checkAndBlockForNetworkAccess();
            //设置activity的mStartedActivity初始值为false
            activity.mStartedActivity = false;
            //如果有theme资源，那么设置theme到该activity中
            int theme = r.activityInfo.getThemeResource();
            if (theme != 0) {
                activity.setTheme(theme);
            }

            //同样设置activity的mCalled初始值为false
            activity.mCalled = false;
            //调用 onCreate() 方法
            if (r.isPersistable()) {
                mInstrumentation.callActivityOnCreate(activity, r.state, r.persistentState);
            } else {
                mInstrumentation.callActivityOnCreate(activity, r.state);
            }
            //同pause一样，如果没有调用基类Activity的onCreate()方法，mCalled为false,命中if
            //抛出异常，所以必须调用基类Activity的onCreate()方法
            if (!activity.mCalled) {
                throw new SuperNotCalledException(
                    "Activity " + r.intent.getComponent().toShortString() +
                    " did not call through to super.onCreate()");
            }
            //将创建出来的Activity对象同样记录到ActivtyClientRecord中
            r.activity = activity;
        }
        //设置状态为 ON_CREATE
        r.setState(ON_CREATE);

        //将当前的 ActivityClientRecord 以 token(其实就是ActivityRecord中的appToken对象)
        //为key存到ActivityThread 的 mActivities 中
        mActivities.put(r.token, r);

    } catch (SuperNotCalledException e) {
        throw e;

    } catch (Exception e) {
        ...
    }

    return activity;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119

在该函数中，可以看到，借助 Instrumentation 对象创建出了 Activity 对象，最后同样是借助 Instrumentation.callActivityOnCreate() 调用到 Activity.onCreate() 方法；
那么至此，onCreate()流程也解析完毕

③ onstart

onStart流程作为一个中间过渡状态，并不如上面几个流程一样，有对应的 ActivityLifecycleItem 对象进行触发，他是在执行完 onCreate() 之后，嵌入在 TransactionExecutor.execute() 函数中的，接下来让我们继续解析

3.1 `TransactionExecutor.execute()`

frameworks\base\core\java\android\app\servertransaction\TransactionExecutor.java

public void execute(ClientTransaction transaction) {
    final IBinder token = transaction.getActivityToken();
    log("Start resolving transaction for client: " + mTransactionHandler + ", token: " + token);

    //先执行 callback 中的execute
    executeCallbacks(transaction);

    //再执行 lifecycleItem的execute
    executeLifecycleState(transaction);
    mPendingActions.clear();
    log("End resolving transaction");
}
1
2
3
4
5
6
7
8
9
10
11
12

通过上面的分析，我们知道在 executeCallbacks() 函数中完成了 onCreate() 流程，接下来就是 executeLifecycleState() 函数

3.2 `TransactionExecutor.executeLifecycleState()`

frameworks\base\core\java\android\app\servertransaction\TransactionExecutor.java

private void executeLifecycleState(ClientTransaction transaction) {
    //拿到ClientTransaction中的ActivityLifecycleItem对象，确保不为空
    //当前lifecycleItem就是ResumeActivityItem对象
    final ActivityLifecycleItem lifecycleItem = transaction.getLifecycleStateRequest();
    if (lifecycleItem == null) {
        // No lifecycle request, return early.
        return;
    }

    //拿到ClientTransaction中的token和ActivityClientRecord
    final IBinder token = transaction.getActivityToken();
    final ActivityClientRecord r = mTransactionHandler.getActivityClient(token);

    if (r == null) {
        // Ignore requests for non-existent client records for now.
        return;
    }

    //调用cycleToPath(),当前lifecycleItem.getTargetState()拿的就是ON_RESUME
    cycleToPath(r, lifecycleItem.getTargetState(), true /* excludeLastState */);

    //开始执行，当前流程下，就是执行 ResumeActivityItem.execute()
    lifecycleItem.execute(mTransactionHandler, token, mPendingActions);
    lifecycleItem.postExecute(mTransactionHandler, token, mPendingActions);
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

在该函数中，会先去获取 ClientTransaction 对象中的 ActivityLifecycleItem对象，而该对象，根据 onCreate 流程中的分析，我们可以知道就是 ResumeActivityItem 对象，那么接下去就是去执行 cycleToPath() 函数

3.3 `TransactionExecutor.cycleToPath()`

frameworks\base\core\java\android\app\servertransaction\TransactionExecutor.java

private void cycleToPath(ActivityClientRecord r, int finish,
                         boolean excludeLastState) {
    //拿到ActivityClientRecord当前的状态，当前流程下是ON_CREATE
    final int start = r.getLifecycleState();
    log("Cycle from: " + start + " to: " + finish + " excludeLastState:" + excludeLastState);
    //调用getLifecyclePath()，此时start=ON_CREATE，finish=ON_RESUME，excludeLastState=true
    //这个就是去拿中间状态，当前流程下拿到的是 ON_START
    final IntArray path = mHelper.getLifecyclePath(start, finish, excludeLastState);
    //处理中间状态
    performLifecycleSequence(r, path);
}
1
2
3
4
5
6
7
8
9
10
11

该函数中会去拿当前还需要执行的中间状态，该函数也比较简单

frameworks\base\core\java\android\app\servertransaction\TransactionExecutorHelper.java

@VisibleForTesting
public IntArray getLifecyclePath(int start, int finish, boolean excludeLastState) {
    //首先要确保 start 和 finish 都是合法符合要求的
    if (start == UNDEFINED || finish == UNDEFINED) {
        throw new IllegalArgumentException("Can't resolve lifecycle path for undefined state");
    }
    if (start == ON_RESTART || finish == ON_RESTART) {
        throw new IllegalArgumentException(
            "Can't start or finish in intermittent RESTART state");
    }
    if (finish == PRE_ON_CREATE && start != finish) {
        throw new IllegalArgumentException("Can only start in pre-onCreate state");
    }

    //做一次mLifecycleSequence清空
    mLifecycleSequence.clear();
    //如果finish是大于等于start，那么就将中间状态都加到mLifecycleSequence中
    //当前的流程中，就会加上 ON_START
    if (finish >= start) {
        // just go there
        for (int i = start + 1; i <= finish; i++) {
            mLifecycleSequence.add(i);
        }
    //而如果finish小于start，则根据情况添加
    } else { // finish < start, can't just cycle down
        if (start == ON_PAUSE && finish == ON_RESUME) {
            // Special case when we can just directly go to resumed state.
            mLifecycleSequence.add(ON_RESUME);
        } else if (start <= ON_STOP && finish >= ON_START) {
            // Restart and go to required state.

            // Go to stopped state first.
            for (int i = start + 1; i <= ON_STOP; i++) {
                mLifecycleSequence.add(i);
            }
            // Restart
            mLifecycleSequence.add(ON_RESTART);
            // Go to required state
            for (int i = ON_START; i <= finish; i++) {
                mLifecycleSequence.add(i);
            }
        } else {
            // Relaunch and go to required state

            // Go to destroyed state first.
            for (int i = start + 1; i <= ON_DESTROY; i++) {
                mLifecycleSequence.add(i);
            }
            // Go to required state
            for (int i = ON_CREATE; i <= finish; i++) {
                mLifecycleSequence.add(i);
            }
        }
    }
    //如果传入的excludeLastState为true，且mLifecycleSequence不为空
    //那么删除mLifecycleSequence中最后一个状态数据
    if (excludeLastState && mLifecycleSequence.size() != 0) {
        mLifecycleSequence.remove(mLifecycleSequence.size() - 1);
    }

    //返回mLifecycleSequence
    return mLifecycleSequence;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63

其实就是根据Activity当前的状态和最终的状态进行判断，取出两者之间的中间状态，记录到 mLifecycleSequence 返回

3.4 `TransactionExecutor.performLifecycleSequence()`

frameworks\base\core\java\android\app\servertransaction\TransactionExecutorHelper.java

private void performLifecycleSequence(ActivityClientRecord r, IntArray path) {
    final int size = path.size();
    //处理path中所有的中间状态
    for (int i = 0, state; i < size; i++) {
        state = path.get(i);
        log("Transitioning to state: " + state);
        switch (state) {
            ...
            //当前流程只有一个ON_START，命中case，调用handleStartActivity()
            case ON_START:
                mTransactionHandler.handleStartActivity(r, mPendingActions);
                break;
            ...
        }
    }
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

根据上一步拿到的中间状态，匹配对应的流程，当前流程下只有一个 ON_START 状态，所以最后就是去执行正启动Activity的 ActivityThread.handleStartActivity() 函数

3.5 `ActivityThread.handleStartActivity()`

frameworks\base\core\java\android\app\ActivityThread.java

@Override
public void handleStartActivity(ActivityClientRecord r,
                                PendingTransactionActions pendingActions) {
    //拿到ActivityClientRecord对应的Activity对象，不能为空,该对像在onCreate流程中已完成创建
    final Activity activity = r.activity;
    if (r.activity == null) {
        return;
    }
    //如果传入的Activity，已经start过了，那么就直接抛出异常返回
    if (!r.stopped) {
        throw new IllegalStateException("Can't start activity that is not stopped.");
    }
    //如果对应的Activity是finish状态的，也直接返回，不进行start
    if (r.activity.mFinished) {
        // TODO(lifecycler): How can this happen?
        return;
    }

    // Start
    //调用performStart()
    activity.performStart("handleStartActivity");
    //设置状态为 ON_START
    r.setState(ON_START);

    ...
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

后面的流程就相对比较简单，调用链就是： Activity.performStart() --> mInstrumentation.callActivityOnStart() --> activity.onStart() ，这些流程也不再赘述了

④ onresume

继续顺着流程往下梳理，我们已经完成了中间状态 onStart() ，接下来就是该走到正在启动Activity的最终状态，也就是 TransactionExecutor.executeLifecycleState() 中的 lifecycleItem.execute() 和 lifecycleItem.postExecute()，在当前流程下，就是 ResumeActivityItem.execute() 和 ResumeActivityItem.postExecute()，该段代码可见 [onstart](3.2 TransactionExecutor.executeLifecycleState()) 中的解析

4.1.1 `ResumeActivityItem.execute()`

frameworks\base\core\java\android\app\servertransaction\ResumeActivityItem.java

@Override
public void execute(ClientTransactionHandler client, IBinder token,
                    PendingTransactionActions pendingActions) {
    Trace.traceBegin(TRACE_TAG_ACTIVITY_MANAGER, "activityResume");
    //处理resume相关事宜
    client.handleResumeActivity(token, true /* finalStateRequest */, mIsForward,
                                "RESUME_ACTIVITY");
    Trace.traceEnd(TRACE_TAG_ACTIVITY_MANAGER);
}
1
2
3
4
5
6
7
8
9

调用流程都类似，也是走到了正在启动的Activity的ActivityThread中去

4.1.2 `ActivityThread.handleResumeActivity()`

1

@Override
public void handleResumeActivity(IBinder token, boolean finalStateRequest, boolean isForward,
                                 String reason) {
    //移除GC_WHEN_IDLE消息，如果有GC任务，移除 mGcIdler 任务
    unscheduleGcIdler();
    mSomeActivitiesChanged = true;
    
    //传入的token标识着对应的Activity，在这里执行 onResume()
    final ActivityClientRecord r = performResumeActivity(token, finalStateRequest, reason);
    ...

}
1
2
3
4
5
6
7
8
9
10
11
12

移除与当前流程无关的代码，我们可以看到其实很相似的，调用链是： ActivityThread.handleResumeActivity() --> ActivityThread.performResumeActivity() --> Activity.performResume() --> Instrumetation.callActivityOnResume() --> Activity.onResume() 后续流程也比较简单，就不做展开分析了

4.2.1 `ResumeActivityItem.postExecute()`

frameworks\base\core\java\android\app\servertransaction\ResumeActivityItem.java

@Override
public void postExecute(ClientTransactionHandler client, IBinder token,
                        PendingTransactionActions pendingActions) {
    try {
        //调用 AMS.activityResumed() 方法
        ActivityManager.getService().activityResumed(token);
    } catch (RemoteException ex) {
        throw ex.rethrowFromSystemServer();
    }
}
1
2
3
4
5
6
7
8
9
10

4.2.2 `ActivityManagerService.activityResumed()`

frameworks\base\services\core\java\com\android\server\am\ActivityManagerService.java

@Override
public final void activityResumed(IBinder token) {
    final long origId = Binder.clearCallingIdentity();
    synchronized(this) {
        //拿到token对应的ActivityRecord对象，复位其中的icicle和haveState
        ActivityRecord.activityResumedLocked(token);
        mWindowManager.notifyAppResumedFinished(token);
    }
    Binder.restoreCallingIdentity(origId);
}
1
2
3
4
5
6
7
8
9
10

至此 onResume 的流程也分析完成。

⑤ onstop

接下来让我们分析一下 onStop 流程，onStop 的流程入口是在 onResume 之后，具体是在 ActivityThread.handleResumeActivity() 函数中

5.1 `ActivityThread.handleResumeActivity()`

frameworks\base\core\java\android\app\ActivityThread.java

@Override
public void handleResumeActivity(IBinder token, boolean finalStateRequest, boolean isForward,
                                 String reason) {
	...
    //赋值mNewActivities为当前Activity
    r.nextIdle = mNewActivities;
    mNewActivities = r;
    if (localLOGV) Slog.v(TAG, "Scheduling idle handler for " + r);
    //添加 Idler 任务，在该任务中就会进行 stop Activity
    Looper.myQueue().addIdleHandler(new Idler());
}
1
2
3
4
5
6
7
8
9
10
11

不关注和当前 stop 流程无关的代码，我们可以看到在resume的最后，添加了一个 Idler 对象到当前 thread 的MessageQueue中
MessageQueue.addIdleHandler()

public void addIdleHandler(@NonNull IdleHandler handler) {
    if (handler == null) {
        throw new NullPointerException("Can't add a null IdleHandler");
    }
    synchronized (this) {
        mIdleHandlers.add(handler);
    }
}
1
2
3
4
5
6
7
8

可以看到是将 Idler 对象添加到了 MessageQueue.mIdleHandlers 中，而在当前ActivityThread正式开始运行后，会在looper中进行message的处理

5.2 `MessageQueue.next()`

frameworks\base\core\java\android\os\MessageQueue.java

Message next() {
    ...
    for (;;) {
        if (nextPollTimeoutMillis != 0) {
            Binder.flushPendingCommands();
        }

        nativePollOnce(ptr, nextPollTimeoutMillis);

        ...
            //如果当前 idleHandler 的计数是小于0的，并且当前没有待处理的message或者还没到处理下一个message的时间
            //命中if，拿到 mIdleHandlers的大小，存在 pendingIdleHandlerCount 变量中
            if (pendingIdleHandlerCount < 0
                && (mMessages == null || now < mMessages.when)) {
                pendingIdleHandlerCount = mIdleHandlers.size();
            }
            //如果 mIdleHandlers 中没有数据，命中if
            //继续去等待message消息，继续阻塞
            if (pendingIdleHandlerCount <= 0) {
                // No idle handlers to run.  Loop and wait some more.
                mBlocked = true;
                continue;
            }

            //如果走到这里，说明 mIdleHandlers 中存在数据，那么这里就需要将其拿出来存到 mPendingIdleHandlers 中
            if (mPendingIdleHandlers == null) {
                mPendingIdleHandlers = new IdleHandler[Math.max(pendingIdleHandlerCount, 4)];
            }
            mPendingIdleHandlers = mIdleHandlers.toArray(mPendingIdleHandlers);
        }

        //开始去处理从 mIdleHandlers 列表中拿出来的数据
        for (int i = 0; i < pendingIdleHandlerCount; i++) {
            final IdleHandler idler = mPendingIdleHandlers[i];
            mPendingIdleHandlers[i] = null; // release the reference to the handler

            boolean keep = false;
            try {
                //逐一调用 mIdleHandlers 中的 IdleHandler 元素的 queueIdle() 方法
                // queueIdle() 方法返回的就是 false
                keep = idler.queueIdle();
            } catch (Throwable t) {
                Log.wtf(TAG, "IdleHandler threw exception", t);
            }

            //肯定会命中该if，从 mIdleHandlers 中移除已经处理的 idler
            if (!keep) {
                synchronized (this) {
                    mIdleHandlers.remove(idler);
                }
            }
        }

        //处理完毕 IdleHandler,将数量重新置0
        pendingIdleHandlerCount = 0;
		...
    }
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58

可以看到，当 MessageQueue中没有待处理消息后，会去尝试获取mIdleHandlers中的对象，进行处理，而此时很显然 mIdleHandlers 中是存在数据的，即会调用到 idler.queueIdle() 函数

5.3 `Idler.queueIdle()`

frameworks\base\core\java\android\app\ActivityThread.java

public final boolean queueIdle() {
    //拿到当前需要处理的Activity
    ActivityClientRecord a = mNewActivities;
    boolean stopProfiling = false;
    //判断是否需要stop profile
    if (mBoundApplication != null && mProfiler.profileFd != null
        && mProfiler.autoStopProfiler) {
        stopProfiling = true;
    }
    //如果需要处理的Activity不为空，命中if
    if (a != null) {
        //将待处理的Activity置空，代表已进行处理
        mNewActivities = null;
        //这里就是拿到了AMS的Ibinder
        IActivityManager am = ActivityManager.getService();
        ActivityClientRecord prev;
        //处理所有的需要处理的Activity
        do {
            if (localLOGV) Slog.v(
                TAG, "Reporting idle of " + a +
                " finished=" +
                (a.activity != null && a.activity.mFinished));
            //如果拿到的Activity不为空且不为finish状态，那么就去stop该Activity
            if (a.activity != null && !a.activity.mFinished) {
                try {
                    //调用AMS中的activityIdle()方法
                    am.activityIdle(a.token, a.createdConfig, stopProfiling);
                    a.createdConfig = null;
                } catch (RemoteException ex) {
                    throw ex.rethrowFromSystemServer();
                }
            }
            prev = a;
            //拿到下一个需要处理的Activity
            a = a.nextIdle;
            prev.nextIdle = null;
        } while (a != null);
    }
    if (stopProfiling) {
        mProfiler.stopProfiling();
    }
    ensureJitEnabled();
    return false;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44

在该函数中，就会去遍历取出存入的需要处理的Activity，借助 AMS进行处理

5.4 `ActivityManagerService.activityIdle()`

frameworks\base\services\core\java\com\android\server\am\ActivityManagerService.java

@Override
public final void activityIdle(IBinder token, Configuration config, boolean stopProfiling) {
    final long origId = Binder.clearCallingIdentity();
    synchronized (this) {
        //拿到传入的token(ActivityRecord中的appToken)所在的Stack
        ActivityStack stack = ActivityRecord.getStackLocked(token);
        //在stack不为空的情况下，处理对应的Activity
        if (stack != null) {
            ActivityRecord r =
                mStackSupervisor.activityIdleInternalLocked(token, false /* fromTimeout */,
                                                            false /* processPausingActivities */, config);
            if (stopProfiling) {
                if ((mProfileProc == r.app) && mProfilerInfo != null) {
                    clearProfilerLocked();
                }
            }
        }
    }
    Binder.restoreCallingIdentity(origId);
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

继续调用 activityIdleInternalLocked() 进行处理

5.5 `ActivityStackSupervisor.activityIdleInternalLocked()`

frameworks\base\services\core\java\com\android\server\am\ActivityStackSupervisor.java

@GuardedBy("mService")
final ActivityRecord activityIdleInternalLocked(final IBinder token, boolean fromTimeout,
                                                boolean processPausingActivities, Configuration config) {
    if (DEBUG_ALL) Slog.v(TAG, "Activity idle: " + token);

    //需要设置为finish状态的activity列表
    ArrayList<ActivityRecord> finishes = null;
    ArrayList<UserState> startingUsers = null;
    int NS = 0;
    int NF = 0;
    boolean booting = false;
    boolean activityRemoved = false;

    //拿到Ibinder对应的Activity对象
    ActivityRecord r = ActivityRecord.forTokenLocked(token);
    //在能正确获取到Activity的情况下，命中if
    if (r != null) {
        if (DEBUG_IDLE) Slog.d(TAG_IDLE, "activityIdleInternalLocked: Callers="
                               + Debug.getCallers(4));
        //移除IDLE_TIMEOUT_MSG消息
        mHandler.removeMessages(IDLE_TIMEOUT_MSG, r);
        //把对应Activity中的 launchTickTime 置为0，且移除LAUNCH_TICK_MSG消息
        r.finishLaunchTickingLocked();
        //如果是来自 IDLE_TIMEOUT_MSG 消息命中的调用，那么会命中if
        if (fromTimeout) {
            reportActivityLaunchedLocked(fromTimeout, r, -1, -1);
        }

        //如果传入的config不为空，命中if
        if (config != null) {
            r.setLastReportedGlobalConfiguration(config);
        }

        //设置activityRecord.idle标志为true
        r.idle = true;

        //Slog.i(TAG, "IDLE: mBooted=" + mBooted + ", fromTimeout=" + fromTimeout);
        //如果该Activity所处的Stack就是 mFocusedStack 或者函数的执行源自 IDLE_TIMEOUT_MSG消息，命中if
        //拿到系统是否正在启动
        if (isFocusedStack(r.getStack()) || fromTimeout) {
            booting = checkFinishBootingLocked();
        }
    }

    //如果 mFocuseStack.mResumedActivity.idle 为true，命中if
    //代表获取焦点的Activity正处在空闲状态
    if (allResumedActivitiesIdle()) {
        //如果待stop的Activity不为空，命中if
        //触发对应thread的GC操作
        if (r != null) {
            mService.scheduleAppGcsLocked();
        }

        //如果当前的wakelock是持有状态，命中if
        //移除LAUNCH_TIMEOUT_MSG消息，并且release wakelock
        if (mLaunchingActivity.isHeld()) {
            mHandler.removeMessages(LAUNCH_TIMEOUT_MSG);
            if (VALIDATE_WAKE_LOCK_CALLER &&
                Binder.getCallingUid() != Process.myUid()) {
                throw new IllegalStateException("Calling must be system uid");
            }
            mLaunchingActivity.release();
        }
        //设置activity的可见性等
        ensureActivitiesVisibleLocked(null, 0, !PRESERVE_WINDOWS);
    }

    //拿到所有需要操作stop的Activity
    final ArrayList<ActivityRecord> stops = processStoppingActivitiesLocked(r,
                                                                            true /* remove */, processPausingActivities);
    NS = stops != null ? stops.size() : 0;
    //如果mFinishingActivities不为空，将其设置到finishes中，然后清空
    if ((NF = mFinishingActivities.size()) > 0) {
        finishes = new ArrayList<>(mFinishingActivities);
        mFinishingActivities.clear();
    }

    if (mStartingUsers.size() > 0) {
        startingUsers = new ArrayList<>(mStartingUsers);
        mStartingUsers.clear();
    }

    //遍历stops,判断activity是否是finish状态，如果是则finish对应的Activity，否则stop
    for (int i = 0; i < NS; i++) {
        r = stops.get(i);
        final ActivityStack stack = r.getStack();
        if (stack != null) {
            if (r.finishing) {
                stack.finishCurrentActivityLocked(r, ActivityStack.FINISH_IMMEDIATELY, false,
                                                  "activityIdleInternalLocked");
            } else {
                //将相应Activity执行stop操作
                stack.stopActivityLocked(r);
            }
        }
    }

    //把本来就是finish状态的Activity，去destroy掉
    for (int i = 0; i < NF; i++) {
        r = finishes.get(i);
        final ActivityStack stack = r.getStack();
        if (stack != null) {
            activityRemoved |= stack.destroyActivityLocked(r, true, "finish-idle");
        }
    }

    if (!booting) {
        if (startingUsers != null) {
            for (int i = 0; i < startingUsers.size(); i++) {
                mService.mUserController.finishUserSwitch(startingUsers.get(i));
            }
        }
    }

    mService.trimApplications();
    //dump();
    //mWindowManager.dump();

    if (activityRemoved) {
        resumeFocusedStackTopActivityLocked();
    }

    return r;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124

首先该函数的调用有两处来源
1. 如当前流程通过 MessageQueue 进行触发
2. 通过 handle 进行触发，而消息发送的调用就在 completeResumeLocked() 中的 mStackSupervisor.scheduleIdleTimeoutLocked(this);
在调用到该函数后，会首先获取到需要进行stop操作的Activity，通过函数 processStoppingActivitiesLocked() 进行获取，获取到需要 stop 的Activity列表之后，在调用 stopActivityLocked() 去进行 stop 操作，逐一查阅

5.5.1 `ActivityStackSupervisor.processStoppingActivitiesLocked()`

frameworks\base\services\core\java\com\android\server\am\ActivityStackSupervisor.java

//返回需要进行stop的Activity列表
final ArrayList<ActivityRecord> processStoppingActivitiesLocked(ActivityRecord idleActivity,
                                                                boolean remove, boolean processPausingActivities) {
    ArrayList<ActivityRecord> stops = null;

    //如果当前系统中存在任一Stack中的mResumedActivity不可见，那么nowVisible就是false
    final boolean nowVisible = allResumedActivitiesVisible();
    //遍历mStoppingActivities列表
    for (int activityNdx = mStoppingActivities.size() - 1; activityNdx >= 0; --activityNdx) {
        ActivityRecord s = mStoppingActivities.get(activityNdx);
        //等待变可见列表中存在从 stopping列表中取出的Activity时，返回为true
        boolean waitingVisible = mActivitiesWaitingForVisibleActivity.contains(s);
        if (DEBUG_STATES) Slog.v(TAG, "Stopping " + s + ": nowVisible=" + nowVisible
                                 + " waitingVisible=" + waitingVisible + " finishing=" + s.finishing);
        //当前系统中每一个mResumedActivity都可见，且当前取出的Activity包含在mActivitiesWaitingForVisibleActivity列表中
        //命中if，将相应activity从mActivitiesWaitingForVisibleActivity移除
        if (waitingVisible && nowVisible) {
            //将该Activity从mActivitiesWaitingForVisibleActivity列表中移除
            mActivitiesWaitingForVisibleActivity.remove(s);
            //将waitingVisible复位为false
            waitingVisible = false;
            //如果取出的Activity是finish状态的，将其设置为不可见
            if (s.finishing) {
                if (DEBUG_STATES) Slog.v(TAG, "Before stopping, can hide: " + s);
                s.setVisibility(false);
            }
        }
        //如果传入的remove为true，命中if
        //整理mStoppingActivities列表，从中筛选出真正需要去stop的Activity，放到 stops 列表中返回
        if (remove) {
            //如果取出的Activity所在的Stack不为空，判断当前stack所处的display是否是sleep状态的
            final ActivityStack stack = s.getStack();
            final boolean shouldSleepOrShutDown = stack != null
                ? stack.shouldSleepOrShutDownActivities()
                : mService.isSleepingOrShuttingDownLocked();
            //如果waitingVisible为false或者对应display是sleep状态的，命中if
            if (!waitingVisible || shouldSleepOrShutDown) {
                //如果传入的processPausingActivities为false，并且当前判断的Activity是处于 Pausing状态
                //命中if，延迟10s后再处理stop,也就是先让他pause完，直接判断下一个activity
                if (!processPausingActivities && s.isState(PAUSING)) {
                    // Defer processing pausing activities in this iteration and reschedule
                    // a delayed idle to reprocess it again
                    //移除IDLE_TIMEOUT_MSG消息
                    removeTimeoutsForActivityLocked(idleActivity);
                    //延迟10s处理IDLE_TIMEOUT_MSG消息
                    scheduleIdleTimeoutLocked(idleActivity);
                    continue;
                }

                if (DEBUG_STATES) Slog.v(TAG, "Ready to stop: " + s);
                if (stops == null) {
                    stops = new ArrayList<>();
                }
                //如果满足了条件，将该Activity添加到stops列表中
                stops.add(s);

                // Make sure to remove it in all cases in case we entered this block with
                // shouldSleepOrShutDown
                //将该activity从 mActivitiesWaitingForVisibleActivity 和 mStoppingActivities 中移除
                mActivitiesWaitingForVisibleActivity.remove(s);
                mStoppingActivities.remove(activityNdx);
            }
        }
    }

    //返回stops列表
    return stops;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68

可以看到，该函数中其中就是对 mStoppingActivities 列表中的Activity对象做了二次判断，符合要求的才会添加到返回列表中进行后续操作，那么 mStoppingActivities 列表又是在何处进行添加的呢？回头查阅 [Pause](① onPause) 流程可以看到在 completePauseLocked() 函数中，有这么一段：

private void completePauseLocked(boolean resumeNext, ActivityRecord resuming) {
    // 拿到当前Stack中的mPausingActivity
    ActivityRecord prev = mPausingActivity;
    if (DEBUG_PAUSE)
        Slog.v(TAG_PAUSE, "Complete pause: " + prev);

    // 如果该Stack中的mPausingActivity不为空,命中if
    // 结束该mPausingActivity，并将mPausingActivity置为空
    if (prev != null) {
        ...
        if (prev.finishing) {
            ...
        } else if (prev.app != null) {
            ...
            if (prev.deferRelaunchUntilPaused) {
                ...
            } else if (wasStopping) {
             	...
            //如果 mPausingActivity 不可见或者 需要进入sleep/shut down，命中else if
            //将 mPausingActivity 添加到 mStoppingActivities 列表中，并且直接触发该Activity的stop操作
            } else if (!prev.visible || shouldSleepOrShutDownActivities()) {
                // Clear out any deferred client hide we might currently have.
                prev.setDeferHidingClient(false);
                addToStopping(prev, true /* scheduleIdle */, false /* idleDelayed */);
            }
        }
        ...
    }
	...
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

可以看到，当prev.visible 为 false的时候，即prev不可见的时候，会将其添加到 mStoppingActivities列表中

5.5.2 `ActivityStack.stopActivityLocked()`

frameworks\base\services\core\java\com\android\server\am\ActivityStack.java

final void stopActivityLocked(ActivityRecord r) {
    ...

    //当对应Activity进程线程都还存在，命中if
    if (r.app != null && r.app.thread != null) {
        adjustFocusedActivityStack(r, "stopActivity");
        r.resumeKeyDispatchingLocked();
        try {
            r.stopped = false;
            if (DEBUG_STATES)
                Slog.v(TAG_STATES,
                       "Moving to STOPPING: " + r + " (stop requested)");
            //设置当前的Activity为 STOPPING 状态
            r.setState(STOPPING, "stopActivityLocked");
            if (DEBUG_VISIBILITY)
                Slog.v(TAG_VISIBILITY,
                       "Stopping visible=" + r.visible + " for " + r);
            if (!r.visible) {
                r.setVisible(false);
            }
            EventLogTags.writeAmStopActivity(
                r.userId, System.identityHashCode(r), r.shortComponentName);
            //构建StopActivityItem对象并触发执行
            mService.getLifecycleManager().scheduleTransaction(r.app.thread, r.appToken,
                                                               StopActivityItem.obtain(r.visible, r.configChangeFlags));
            if (shouldSleepOrShutDownActivities()) {
                r.setSleeping(true);
            }
            //延迟11s后发送STOP_TIMEOUT_MSG消息
            Message msg = mHandler.obtainMessage(STOP_TIMEOUT_MSG, r);
            mHandler.sendMessageDelayed(msg, STOP_TIMEOUT);
        } catch (Exception e) {
            ...
        }
    }
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36

在该函数中就会去构建出 StopActivityItem 对象，并触发执行，并且延迟 STOP_TIMEOUT 时间后发送 STOP_TIMEOUT_MSG 消息触发 ActivityRecordactivityStoppedLocked() 函数的执行
在 StopActivityItem 对象中，和之前几个生命周期中执行相似，流程就是： ActivityThread.handleStopActivity() --> ActivityThread.performStopActivityInner() -(ActivityRecord.visible == false)-> ActivityThread.callActivityOnStop() --> Activity.performStop() --> Instrumentation.callActivityOnStop() --> Activity.onStop()

疑问：

梳理完这五个生命周期，可以发现，只有在 ActivityRecord.visible 值为false，即Activity不可见的时候，才会进行 stop，那么这个visible的值究竟是在何处进行变化的呢？
- 其实在解析整体流程时，我们能反复看到 ensureActivitiesVisibleLocked() 函数，从函数名称其实也可以看得出来，该函数中会进行Activity可见性的判断，那么接下来我们深入解析一下该函数，直接看 ActivityStack.ensureActivitiesVisibleLocked() 函数

ActivityStack.ensureActivitiesVisibleLocked()

frameworks\base\services\core\java\com\android\server\am\ActivityStack.java

final void ensureActivitiesVisibleLocked(ActivityRecord starting, int configChanges,
                                         boolean preserveWindows, boolean notifyClients) {
    ...
    try {
        // 拿到当前Stack最顶部的非finish状态的Activity
        ActivityRecord top = topRunningActivityLocked();
        ...

        //在top不为空的时候，aboveTop为true
        boolean aboveTop = top != null;
        // 判断当前的堆栈是否可见，如果可见则返回true
        final boolean stackShouldBeVisible = shouldBeVisible(starting);
        // 当前stack是否是被遮盖住的，和 stackShouldBeVisible 是相反的
        boolean behindFullscreenActivity = !stackShouldBeVisible;
        // 当 当前stack就是 mFocusedStack 并且 传入的 starting Activity 不在当前stack中，那么resumeNextActivity为true
        boolean resumeNextActivity = mStackSupervisor.isFocusedStack(this)
            && (isInStackLocked(starting) == null);
        // 如果当前Stack是有所属display的，并且对应display最上层的非pinnedWindow的stack就是当前的stack，那么isTopNotPinnedStack为true
        //代表当前的stack就是对应display上最顶层的非自由窗口的stack
        final boolean isTopNotPinnedStack = isAttached() && getDisplay().isTopNotPinnedStack(this);
        //开始从上向下遍历当前stack中的Activity
        for (int taskNdx = mTaskHistory.size() - 1; taskNdx >= 0; --taskNdx) {
            final TaskRecord task = mTaskHistory.get(taskNdx);
            final ArrayList<ActivityRecord> activities = task.mActivities;
            for (int activityNdx = activities.size() - 1; activityNdx >= 0; --activityNdx) {
                final ActivityRecord r = activities.get(activityNdx);
                //如果拿到的Activity是finish状态的，直接判断下一个
                if (r.finishing) {
                    continue;
                }
                //判断当前获取到的Activity是否是topActivity，如果是则为 true
                final boolean isTop = r == top;
                //如果aboveTop是true 且 当前判断的Activity不是topActivity,直接判断下一个
                if (aboveTop && !isTop) {
                    continue;
                }
                //将aboveTop置为false，代表不再受topActivity约束
                aboveTop = false;

                //忽略 keyguard 的情况下，是否可见
                final boolean visibleIgnoringKeyguard = r.shouldBeVisibleIgnoringKeyguard(
                    behindFullscreenActivity);
                r.visibleIgnoringKeyguard = visibleIgnoringKeyguard;

                //判断当前遍历的Activity是否可见
                final boolean reallyVisible = checkKeyguardVisibility(r,
                                                                      visibleIgnoringKeyguard, isTop && isTopNotPinnedStack);
                //在判断上一个Activity忽略Keyguard时可见的情况下，需要再重新针对
                //behindFullscreenActivity进行判断，关键就在此处
                if (visibleIgnoringKeyguard) {
                    behindFullscreenActivity = updateBehindFullscreen(!stackShouldBeVisible,
                                                                      behindFullscreenActivity, r);
                }
                //如果是可见的
                if (reallyVisible) {
                    ...
                } else {
                    makeInvisible(r);
                }
            }
            ...
        }

        ...
    }
    ...
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67

略去相关性不大的代码，可以看到，与可见性判断呈直接关系的是如下的几个函数：
- shouldBeVisible()
- shouldBeVisibleIgnoringKeyguard()
- checkKeyguardVisibility()
- updateBehindFullscreen()
我们假定一个场景，当前就是在判断 mFocusedStack ，且当前待启动的Activity是不透明的，且假定对应stack中只存在两个Activity对象，那么判断顺序如下：

ActivityStack.shouldBeVisible()

// 判断该堆栈是否可见，如果可见则返回true
boolean shouldBeVisible(ActivityRecord starting) {
    //如果当前的stack没有所属的display 或者是 设置了强制隐藏，命中if
    //直接返回false，代表不可见
    if (!isAttached() || mForceHidden) {
        return false;
    }
    //如果判断的当前stack就是mFocusedStack，那么直接返回true，代表可见
    if (mStackSupervisor.isFocusedStack(this)) {
        return true;
    }

    //拿到当前stack最顶部的不为finish状态的Activity
    final ActivityRecord top = topRunningActivityLocked();
    //如果拿不到top,并且传入的Activity不在当前Stack中 并且 当前的stack不是所属的display最顶层的stack，命中if
    //直接返回false，代表不可见
    if (top == null && isInStackLocked(starting) == null && !isTopStackOnDisplay()) {
        return false;
    }

    //拿到所属的display
    final ActivityDisplay display = getDisplay();
    //是否有不透明的分屏stack
    boolean gotSplitScreenStack = false;
    //是否是不透明的主分屏stack
    boolean gotOpaqueSplitScreenPrimary = false;
    //是否是不透明的副分屏stack
    boolean gotOpaqueSplitScreenSecondary = false;
    //拿到当前的stack的windowMode
    final int windowingMode = getWindowingMode();
    //判断当前的Stack是否是ACTIVITY_TYPE_ASSISTANT类型
    final boolean isAssistantType = isActivityTypeAssistant();
    //从上往下遍历对应display中的所有ActivityStack
    for (int i = display.getChildCount() - 1; i >= 0; --i) {
        final ActivityStack other = display.getChildAt(i);
        //如果该display中存在当前的stack，那么直接返回true，代表可见
        if (other == this) {
            return true;
        }

        //拿到当前遍历的stack的windowModevisible
        final int otherWindowingMode = other.getWindowingMode();

        //如果当前遍历的stack是 WINDOWING_MODE_FULLSCREEN 类型，那么只有在当前Stack是 WINDOWING_MODE_SPLIT_SCREEN_PRIMARY
        //类型并且 遍历的stack是ACTIVITY_TYPE_HOME 或者 是 ACTIVITY_TYPE_ASSISTANT类型且mRecentsAnimationController不为空
        //时才可见
        if (otherWindowingMode == WINDOWING_MODE_FULLSCREEN) {
            final int activityType = other.getActivityType();
            if (windowingMode == WINDOWING_MODE_SPLIT_SCREEN_PRIMARY) {
                if (activityType == ACTIVITY_TYPE_HOME
                    || (activityType == ACTIVITY_TYPE_ASSISTANT
                        && mWindowManager.getRecentsAnimationController() != null)) {
                    return true;
                }
            }
            //如果判断的stack是透明的，那么继续判断下一个
            if (other.isStackTranslucent(starting)) {
                // Can be visible behind a translucent fullscreen stack.
                continue;
            }
            return false;
            //如果遍历的stack是主分屏stack，且尚不存在不透明的分屏主窗口，命中if
            //那么如果判断other Stack是不透明的stack，且当前的stack是分屏主窗口类型,直接返回false，不可见
        } else if (otherWindowingMode == WINDOWING_MODE_SPLIT_SCREEN_PRIMARY
                   && !gotOpaqueSplitScreenPrimary) {
            gotSplitScreenStack = true;
            gotOpaqueSplitScreenPrimary = !other.isStackTranslucent(starting);
            if (windowingMode == WINDOWING_MODE_SPLIT_SCREEN_PRIMARY
                && gotOpaqueSplitScreenPrimary) {
                return false;
            }
            //与主分屏同理，判断副分屏的情况
        } else if (otherWindowingMode == WINDOWING_MODE_SPLIT_SCREEN_SECONDARY
                   && !gotOpaqueSplitScreenSecondary) {
            gotSplitScreenStack = true;
            gotOpaqueSplitScreenSecondary = !other.isStackTranslucent(starting);
            if (windowingMode == WINDOWING_MODE_SPLIT_SCREEN_SECONDARY
                && gotOpaqueSplitScreenSecondary) {
                return false;
            }
        }
        //当主副分屏窗口都不透明时，直接返回false，不可见
        if (gotOpaqueSplitScreenPrimary && gotOpaqueSplitScreenSecondary) {
            return false;
        }
        //如果当前的Stack是ACTIVITY_TYPE_ASSISTANT类型 并且 display中存在不透明的stack，直接返回false，不可见
        if (isAssistantType && gotSplitScreenStack) {
            return false;
        }
    }

    //默认就是返回true，可见
    return true;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94

首先判断当前stack是否是可见，根据我们假设的前提，当前stack就是 mFocusedStack ，此处判断出来是可见的，即当前 stackShouldBeVisible = true，而 behindFullscreenActivity = false

ActivityRecord.shouldBeVisibleIgnoringKeyguard(false)

boolean shouldBeVisibleIgnoringKeyguard(boolean behindFullscreenActivity) {
    //判断当前Activity是否是可展示的
    if (!okToShowLocked()) {
        return false;
    }

    //根据传入的behindFullscreenActivity值或者mLaunchTaskBehind判断是否可见
    return !behindFullscreenActivity || mLaunchTaskBehind;
}
1
2
3
4
5
6
7
8
9

可以看到，在 shouldBeVisibleIgnoringKeyguard() 中，确保Activity可展示的前提下，会根据传入的 behindFullscreenActivity 或者 mLaunchTaskBehind 来确定是否可见，而我们当下传入的 behindFullscreenActivity = false，所以此时肯定返回的是true，即 visibleIgnoringKeyguard = true

ActivityStack.checkKeyguardVisibility(r, true, true)

// 结合当前是否是锁屏状态，判断Activity可见性
boolean checkKeyguardVisibility(ActivityRecord r, boolean shouldBeVisible, boolean isTop) {
    final int displayId = mDisplayId != INVALID_DISPLAY ? mDisplayId : DEFAULT_DISPLAY;
    //判断熄屏显示或者锁屏显示的状态
    final boolean keyguardOrAodShowing = mStackSupervisor.getKeyguardController()
        .isKeyguardOrAodShowing(displayId);
    //判断现在是不是锁屏显示的状态
    final boolean keyguardLocked = mStackSupervisor.getKeyguardController().isKeyguardLocked();
    //判断传入的Activity是否可以在锁屏状态进行显示
    final boolean showWhenLocked = r.canShowWhenLocked();
    //判断传入的Activity是否存在FLAG_DISMISS_KEYGUARD的window
    final boolean dismissKeyguard = r.hasDismissKeyguardWindows();
    //如果传入的shouldBeVisible为true，代表不考虑keyguard的时候可以显示，命中if
    //那么就去判断当前display是否有FLAG_CAN_SHOW_WITH_INSECURE_KEYGUARD标识，并且当前的锁屏可以不用凭证直接解除，
    //满足上述条件直接返回true，代表可见
    if (shouldBeVisible) {
        if (dismissKeyguard && mTopDismissingKeyguardActivity == null) {
            mTopDismissingKeyguardActivity = r;
        }

        if (isTop) {
            mTopActivityOccludesKeyguard |= showWhenLocked;
        }

        //判断当前stack所属的display是否有FLAG_CAN_SHOW_WITH_INSECURE_KEYGUARD标识
        final boolean canShowWithKeyguard = canShowWithInsecureKeyguard()
            && mStackSupervisor.getKeyguardController().canDismissKeyguard(); //是否可以不输入凭证的情况下接触锁屏
        //如果所属display有FLAG_CAN_SHOW_WITH_INSECURE_KEYGUARD标识，并且不用密码就可以解除锁屏，直接返回true，代表可以显示
        if (canShowWithKeyguard) {
            return true;
        }
    }
    //显示锁屏或者AOD时，命中if
    //只有在当前Activity可以解除锁屏，并且AOD不可见的时候才可见，即返回true
    if (keyguardOrAodShowing) {
        return shouldBeVisible && mStackSupervisor.getKeyguardController()
            .canShowActivityWhileKeyguardShowing(r, dismissKeyguard);
     //显示锁屏时，命中else if
    //判断当前是否可以覆盖keyguard进行显示，是则返回true
    } else if (keyguardLocked) {
        return shouldBeVisible && mStackSupervisor.getKeyguardController().canShowWhileOccluded(
            dismissKeyguard, showWhenLocked);
    //既没有锁屏也没有AOD，直接返回shouldBeVisible，即不考虑keyguard时的可见性即可
    } else {
        return shouldBeVisible;
    }
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47

在该函数中，会根据是否显示 keyguard 或者 AOD 来判断该Activity是否真的可见，如果当前没有显示 keyguard 或者 AOD，那么就直接以传入的 shouldBeVisible 为准，此时就是直接返回 shouldBeVisible ，也就是 reallyVisible = true

ActivityStack.updateBehindFullscreen(false, false, r)
```
private boolean updateBehindFullscreen(boolean stackInvisible, boolean behindFullscreenActivity,
                                       ActivityRecord r) {
    //根据传入的Activity是否全屏显示，来决定behindFullscreenActivity的值
    if (r.fullscreen) {
        behindFullscreenActivity = true;
    }
    return behindFullscreenActivity;
}
1
2
3
4
5
6
7
8
```
- 可以看到，如果前一个传入的Activity是fullscreen的，即全屏显示的，那么就会设置behindFullscreenActivity = true，即代表在该Activity下面的Activity都是处于被盖住显示的状态
- 针对当前的Activity，reallyVisible = true也就是可见的，那么在 ensureActivitiesVisibleLocked()函数中，会命中 if (reallyVisible){...} ，接下来会循环判断下一个 Activity 的可见性，此时 behindFullscreenActivity 是为 true
ActivityRecord.shouldBeVisibleIgnoringKeyguard(true)
- 由于此时传入的 behindFullscreenActivity 是为 true，进入判断后返回则为false，即 visibleIgnoringKeyguard = false
ActivityStack.checkKeyguardVisibility(r, false, false)
- 由于此时传入的 visibleIgnoringKeyguard = false，所以此时判断返回的 reallyVisible = false，即当前的Activity是不可见的，那么接下来就会命中 makeInvisible(r) ，也就会设置当前的 Activity 为 Invisible，并且在这之中，还会根据当前Activity的状态，决定是否将其添加到 mStoppingActivities 列表中，方便后续进行stop该Activity

相关阅读:
C#三层架构
 演练成员内部类
 C盘清理攻略
 echarts制作关系图谱
 html 中 title与h1，b与strong,i与em区别？
【C语言】学生宿舍信息管理系统
 控制I/O设备四种方式
 沉睡者IT - 10个问题说清楚：什么是元宇宙？
ElasticSearch学习篇6_ES实践与Lucene对比及原理分析技术分享小记
 无网络maven私服添加jar和pom
原文地址：https://blog.csdn.net/super_marie/article/details/125470835

解析Activity启动-生命周期角度

解析Activity启动-生命周期角度

① onPause

1.1 resumeTopActivityInnerLocked()

1.2 startPausingLocked()

1.2.1 PauseActivityItem

1.2.2 schedulePauseTimeout()

② oncreate

2.1 realStartActivityLocked()

2.2 TransactionExecutor.executeCallbacks()

2.3 LaunchActivityItem.execute()

2.4 ActivityThread.handleLaunchActivity()

2.5 ActivityThread.performLaunchActivity()

③ onstart

3.1 TransactionExecutor.execute()

3.2 TransactionExecutor.executeLifecycleState()

3.3 TransactionExecutor.cycleToPath()

3.4 TransactionExecutor.performLifecycleSequence()

3.5 ActivityThread.handleStartActivity()

④ onresume

4.1.1 ResumeActivityItem.execute()

4.1.2 ActivityThread.handleResumeActivity()

4.2.1 ResumeActivityItem.postExecute()

4.2.2 ActivityManagerService.activityResumed()

⑤ onstop

5.1 ActivityThread.handleResumeActivity()

5.2 MessageQueue.next()

5.3 Idler.queueIdle()

5.4 ActivityManagerService.activityIdle()

5.5 ActivityStackSupervisor.activityIdleInternalLocked()

5.5.1 ActivityStackSupervisor.processStoppingActivitiesLocked()

5.5.2 ActivityStack.stopActivityLocked()

疑问：

1.1 `resumeTopActivityInnerLocked()`

1.2 `startPausingLocked()`

1.2.1 `PauseActivityItem`

1.2.2 `schedulePauseTimeout()`

2.1 `realStartActivityLocked()`

2.2 `TransactionExecutor.executeCallbacks()`

2.3 `LaunchActivityItem.execute()`

2.4 `ActivityThread.handleLaunchActivity()`

2.5 `ActivityThread.performLaunchActivity()`

3.1 `TransactionExecutor.execute()`

3.2 `TransactionExecutor.executeLifecycleState()`

3.3 `TransactionExecutor.cycleToPath()`

3.4 `TransactionExecutor.performLifecycleSequence()`

3.5 `ActivityThread.handleStartActivity()`

4.1.1 `ResumeActivityItem.execute()`

4.1.2 `ActivityThread.handleResumeActivity()`

4.2.1 `ResumeActivityItem.postExecute()`

4.2.2 `ActivityManagerService.activityResumed()`

5.1 `ActivityThread.handleResumeActivity()`

5.2 `MessageQueue.next()`

5.3 `Idler.queueIdle()`

5.4 `ActivityManagerService.activityIdle()`

5.5 `ActivityStackSupervisor.activityIdleInternalLocked()`

5.5.1 `ActivityStackSupervisor.processStoppingActivitiesLocked()`

5.5.2 `ActivityStack.stopActivityLocked()`