Android ANR机制

大致有四种类型，下面逐一介绍：

一.输入事件无响应ANR

调用过程：

(InputDispatcher#) dispatchMotionLocked --> findFocusedWindowTargetsLocked--> onANRLocked --> doNotifyANRLockedInterruptible -->
(InputDispatcherPolicyInterface#)mPolicy->notifyANR --> (NativeInputManager(:public InputDispatcherPolicyInterface)#)notifyANR -->
(InputManagerService#) notifyANR --> (InputManagerCallback: WindowManagerCallbacks#)mWindowManagerCallbacks.notifyANR( token, reason) -->
(AMS#)mService.mAmInternal.inputDispatchingTimedOut(windowState.mSession.mPid, aboveSystem, reason) --> (ProcessRecord#)proc.appNotResponding(activityShortComponentName, aInfo,parentShortComponentName, parentProcess, aboveSystem, annotation);

部分源码：

int32_t InputDispatcher::findFocusedWindowTargetsLocked(...) {
    ...
    if(focusedWindowHandle==null){
        // If there is no currently then drop the event. 
        handleTargetsNotReadyLocked{
            if(currentTime >= currentTime +DEFAULT_INPUT_DISPATCHING_TIMEOUT){
                //如果超过DEFAULT_INPUT_DISPATCHING_TIMEOUT也就是5s    
                onANRLocked();
            }
        }
    } 
    ...
}
(ProcessRecord#) void appNotResponding(String activityShortComponentName, ApplicationInfo aInfo,String parentShortComponentName, WindowProcessController parentProcess,boolean aboveSystem, String annotation) {
    ...
    //部分情况只标记不处理，比如开关机过程中的输入事件
    // 记录anr日志
    ActivityManagerService.dumpStackTraces:{...
        final File tracesDir = new File(“/data/anr/anr_xx”);
    }
    //是否是静默ANR(一般情况都会弹出ANR对话框，除非已经配置了不让弹出对话框，这种情况直接在后台进行杀死进程的操作)
    //如果是静默ANR：直接杀掉进程
    kill();
    //否则,弹出ANR对话框
    (AppErrors#) mAppErrors.handleShowAnrUi(msg){
        ...
        anrDialog = new AppNotRespondingDialog;anrDialog.show();
        //选择强制关闭杀掉进程
        mService.killAppAtUsersRequest(mProc, AppNotRespondingDialog.this);
        //选择等待，会等待APP相应并上报给AMS更新lastANR等相关信息
        mService.mServices.scheduleServiceTimeoutLocked(app);
    }
}

小结：

1. 当输入事件派发时超过5S没有获取到目标（当前焦点）窗口，且当前应用进程未结束，(比如输入事件时应用进程主线程阻塞时间超过5S);
2. 输入事件的派发顺序:(SS#)InputDispatcher线程 --> InputMangerService --> WindowMangerService --> (APP#)PhoneWindow -->
   ViewRootImpl --> View;

二.广播处理时间过久

调用过程：

(BroadcastQueue#) processNextBroadcastLocked() -->
-->1)if ((numReceivers > 0) && (now > r.dispatchTime + (2 * mConstants.TIMEOUT * numReceivers)) --> broadcastTimeoutLocked(false) --> (AppNotResponding#)run() --> (ProcessRecord#)appNotResponding();
-->2)setBroadcastTimeoutLocked --> mHandler.sendMessageAtTime(mHandler.obtainMessage(BROADCAST_TIMEOUT_MSG, this), timeoutTime) --> broadcastTimeoutLocked(true) ...

部分源码:

BroadcastQueue#

static final int BROADCAST_FG_TIMEOUT = 10*1000;
static final int BROADCAST_BG_TIMEOUT = 60*1000;


(BroadcastQueue#) void processNextBroadcastLocked(){
    // First, deliver any non-serialized broadcasts right away. 首先处理并行广播，并行广播没有ANR的问题
    while (mParallelBroadcasts.size() > 0) {
        for (int i=0; i<N; i++) {
                    Object target = r.receivers.get(i);
            deliverToRegisteredReceiverLocked(r, (BroadcastFilter)target, false, i);
        }
    }
    int numReceivers = (r.receivers != null) ? r.receivers.size() : 0;
    if ((numReceivers > 0) &&  (now > r.dispatchTime + (2 * mConstants.TIMEOUT * numReceivers)){
        //如果广播处理的总时间大于2 * mConstants.TIMEOUT * numReceivers, 则直接作为广播处理超时ANR处理       
        broadcastTimeoutLocked(false);  
    }
    if (! mPendingBroadcastTimeoutMessage) {
        long timeoutTime = r.receiverTime + mConstants.TIMEOUT; 
        setBroadcastTimeoutLocked (timeoutTime);
    }
}

(BroadcastQueue#)void broadcastTimeoutLocked(boolean fromMsg){
     mHandler.post(new AppNotResponding(app, anrMessage));
}

(BroadcastQueue$AppNotResponding:Thread#)void run(){
    mApp.appNotResponding(null, null, null, null, false, mAnnotation);//ProcessRecord#appNotResponding 同type1
}

(BroadcastQueue#)void setBroadcastTimeoutLocked(long timeoutTime){
    //预备一个广播处理超时ANR消息处理，延迟timeoutTime发送，在receiver处理完会调用cancelBroadcastTimeoutLocked方法，remove预备的广播处理超时ANR处理。     
    //什么时候执行cancelBroadcastTimeoutLocked？当广播结束会(BroadcastReceiver#).sendFinished ---> (AMS#) finishReceiver -->  r.queue.finishReceiverLocked  --> (BroadcastQueue#) r.queue.processNextBroadcastLocked --> (BroadcastQueue#) cancelBroadcastTimeoutLocked()取消预备处理ANR的消息
    Message msg = mHandler.obtainMessage(BROADCAST_TIMEOUT_MSG, this);
    mHandler.sendMessageAtTime(msg, timeoutTime);
    mPendingBroadcastTimeoutMessage = true;
}

小结：

1. 并行无序广播处理时间不受ANR管制;
2. 广播处理总时间大于2 * mConstants.TIMEOUT * numReceivers 会产生ANR;
3. 单个广播处理时间大于（前台10S，后台60S）产生ANR；
   PS:BroadcastQueue负责广播分发管理。

三.Service ANR：

调用过程：

(AMS#)startService -->

1. (ActiveServices#)realStartServiceLocked --> bumpServiceExecutingLocked -->(ActivityThread#)app.thread.scheduleCreateService(r, r.serviceInfo,mAm.compatibilityInfoForPackage(r.serviceInfo.applicationInfo) --> service.onCreate();
2. (ActiveServices#)requestServiceBindingLocked --> bumpServiceExecutingLocked --> (ActivityThread#)r.app.thread.scheduleBindService(r, i.intent.getIntent(), rebind,r.app.getReportedProcState()) --> (s.onBind(data.intent); or s.onRebind(data.intent);
3. (ActiveServices#)sendServiceArgsLocked --> bumpServiceExecutingLocked --> (ActivityThread#)r.app.thread.scheduleServiceArgs(r, slice); --> s.onStartCommand()
4. (ActiveServices#)bringDownServiceLocked --> bumpServiceExecutingLocked --> (ActivityThread#)r.app.thread.scheduleUnbindService(r,ibr.intent.getIntent()); --> s.onUnbind(data.intent);
5. (ActiveServices#)removeConnectionLocked --> bumpServiceExecutingLocked --> (ActivityThread#)s.app.thread.scheduleUnbindService(s, b.intent.intent.getIntent()); --> s.onUnbind(data.intent);
   //bumpServiceExecutingLocked启动ANR机制
   bumpServiceExecutingLocked --> scheduleServiceTimeoutLocked --> mAm.mHandler.sendMessageDelayed(mAm.mHandler.obtainMessage(ActivityManagerService.SERVICE_TIMEOUT_MSG),proc.execServicesFg ? SERVICE_TIMEOUT : SERVICE_BACKGROUND_TIMEOUT)
   --> (AMS#) handle --> (ActiveServices#)serviceTimeout(proc) --> proc.appNotResponding();...(同type1)
   //以上调用表明在Service的onCreate、onStartCommand、onStart、onBind、onUnbind，这几个方法都有ANR处理超时机制
   //当Service的onCreate、onStartCommand、onStart等方法执行完成后，通过 (ActiveServices#)serviceDoneExecutingLocked方法取消之前预备ANR处理消息。

部分源码：

// How long we wait for a service to finish executing.
 static final int SERVICE_TIMEOUT = 20*1000;
// How long we wait for a service to finish executing.
 static final int SERVICE_BACKGROUND_TIMEOUT = SERVICE_TIMEOUT * 10;

(ActiveServices#)void bumpServiceExecutingLocked(){
    // For b/34123235: Services within the system server won't start until SystemServer does Looper.loop(), so we shouldn't try to start/bind to them too early in the boot process. However, since there's a little point of showing the ANR dialog in that case, let's suppress the timeout until PHASE_THIRD_PARTY_APPS_CAN_START. (Note there are multiple services start at PHASE_THIRD_PARTY_APPS_CAN_START too, which technically could also trigger this timeout if there's a system server that takes a long time to handle PHASE_THIRD_PARTY_APPS_CAN_START, but that shouldn't happen.)if(mAm.mBootPhase < SystemService.PHASE_THIRD_PARTY_APPS_CAN_START){timeoutNeeded = false} else{timeoutNeeded = true}//在SS的启动阶段不用ANR机制
    if(timeoutNeeded){
        scheduleServiceTimeoutLocked(r.app);
    }
}
    
(ActiveServices#)void scheduleServiceTimeoutLocked(){
    //无论是startService还是bindService都会调用scheduleServiceTimeoutLocked预备一个ANR的处理超时消息，延时发送 
    // 前台服务20s，后台服务200s，service相应方法调用完成后通过调用serviceDoneExecutingLocked取消预备的ANR消息；
    Message msg = mAm.mHandler.obtainMessage(
    ActivityManagerService.SERVICE_TIMEOUT_MSG);
    msg.obj = proc;
    mAm.mHandler.sendMessageDelayed(msg,proc.execServicesFg ? SERVICE_TIMEOUT : SERVICE_BACKGROUND_TIMEOUT);
} 

//若在ANR超时时间内完成业务取消，则不会产生ANR
(ActiveServices#)void serviceDoneExecutingLocked(){
    mAm.mHandler.removeMessages(ActivityManagerService.SERVICE_TIMEOUT_MSG, r.app);
}

小结：

1. Service的onCreate、onStartCommand、onStart、onBind、onRebind方法都有ANR机制，一旦处理超时就会抛出ANR异常。

四.ContentProvider Timeout

调用过程：

(ContentProviderClient#) query/insert/delete/update等函数中会在业务处理前调用beforeRemote函数：延时mAnrTimeout后发出ANR处理超时的消息，在这些query等函数业务处理后调用afterRemote取消预备发送的ANR消息。
mAnrTimeout值可通过mProviderClient.setDetectNotResponding(long timeout)函数进行设置：0表示无限期，不采取ANR行动。
(ContentProviderClient#)beforeRemote() --> (ContentResolver#) mContentResolver.appNotRespondingViaProvider(mContentProvider); -->
(ActivityThread#)mMainThread.appNotRespondingViaProvider(icp.asBinder()) --> (AMS#)ActivityManager.getService().appNotRespondingViaProvider(prc.holder.connection) -->
(ProcessRecord)conn.provider.proc.appNotResponding(null, null, null, null, false, "ContentProvider not responding");

部分源码：

(ContentProviderClient#)void  beforeRemote(){
    sAnrHandler.postDelayed(mAnrRunnable, mAnrTimeout)；
}
(ContentProviderClient#)void afterRemote() {
    if (mAnrRunnable != null) {
        ///insert/delete/update等函数中会在处理后调用afterRemote取消预备ANR处理超时的消息
            sAnrHandler.removeCallbacks(mAnrRunnable);
    }
}

小结：

1. 通过ContentProviderClient#setDetectNotResponding(long timeout)函数可以自主设置触发ANR的时间，timeout表示允许挂起调用的持续时间，在远程提供者被认为是无响应的时间，也就是ANR的超时时间, 0表示无限期，不采取ANR行动。
   2.ContentProvider默认是无ANR超时期限的，不采取ANR处理;