深入理解 Android ANR 触发原理以及信息收集过程
一、概述
作为 Android 开发者，相信大家都遇到过 ANR。那么为什么会出现 ANR 呢，ANR 之后系统都做了啥。文章将对这个问题详细解说。
ANR(Application Not responding)，是指应用程序未响应，Android系统对于一些事件需要在一定的时间范围内完成，如果超过预定时间能未能得到有效响应或者响应时间过长，都会造成ANR。一般地，这时往往会弹出一个提示框，告知用户当前xxx未响应，用户可选择继续等待或者Force Close。
那么哪些场景会造成ANR呢？
-
Service Timeout:比如前台服务在20s内未执行完成；
-
BroadcastQueue Timeout：比如前台广播在10s内未执行完成
-
ContentProvider Timeout：内容提供者,在publish过超时10s;
-
InputDispatching Timeout: 输入事件分发超时5s，包括按键和触摸事件。
触发ANR的过程可分为三个步骤: 埋炸弹, 拆炸弹, 引爆炸弹。
埋炸弹可以理解为发送了一个延迟触发的消息（炸弹）；
拆炸弹可以理解为将这个延迟消息（炸弹）取消了，也就不会触发了；
引爆炸弹可以理解为延迟时间已达，开始处理延迟消息（炸弹引爆了）。
二、Service
先附上一张 service 启动流程图：
Service Timeout是位于”ActivityManager”线程中的AMS.MainHandler收到SERVICE_TIMEOUT_MSG消息时触发。
对于Service有两类:
- 对于前台服务，则超时为SERVICE_TIMEOUT = 20s；
- 对于后台服务，则超时为SERVICE_BACKGROUND_TIMEOUT = 200s
由变量ProcessRecord.execServicesFg来决定是否前台启动。
2.1 埋炸弹
其中在Service进程attach到system_server进程的过程中会调用realStartServiceLocked()方法来埋下炸弹.
首先咱们先看 service 的启动中一个方法 realStartServiceLocked：
// ActiveServices.java private final void realStartServiceLocked(ServiceRecord r, ProcessRecord app, boolean execInFg) throws RemoteException { ... //发送delay消息(SERVICE_TIMEOUT_MSG) bumpServiceExecutingLocked(r, execInFg, "create"); try { ... //最终执行服务的onCreate()方法 app.thread.scheduleCreateService(r, r.serviceInfo, mAm.compatibilityInfoForPackageLocked(r.serviceInfo.applicationInfo), app.repProcState); } catch (DeadObjectException e) { mAm.appDiedLocked(app); throw e; } finally { ... } } private final void bumpServiceExecutingLocked(ServiceRecord r, boolean fg, String why) { ... scheduleServiceTimeoutLocked(r.app); } void scheduleServiceTimeoutLocked(ProcessRecord proc) { if (proc.executingServices.size() == 0 || proc.thread == null) { return; } long now = SystemClock.uptimeMillis(); Message msg = mAm.mHandler.obtainMessage( ActivityManagerService.SERVICE_TIMEOUT_MSG); msg.obj = proc; //当超时后仍没有remove该SERVICE_TIMEOUT_MSG消息，则执行service Timeout流程 mAm.mHandler.sendMessageAtTime(msg, proc.execServicesFg ? (now+SERVICE_TIMEOUT) : (now+ SERVICE_BACKGROUND_TIMEOUT)); }
在 AS.realStartServiceLocked 启动 service 方法中，发送了了一个延时的关于超时的消息，这里又对 service 进行了前后台的区分：
// How long we wait for a service to finish executing. 20s static final int SERVICE_TIMEOUT = 20*1000; // How long we wait for a service to finish executing. 200s static final int SERVICE_BACKGROUND_TIMEOUT = SERVICE_TIMEOUT * 10;
2.2 拆炸弹
AS.realStartServiceLocked() 调用的过程会埋下一颗炸弹, 超时没有启动完成则会爆炸. 那么什么时候会拆除这颗炸弹的引线呢? 经过Binder等层层调用进入目标进程的主线程handleCreateService()的过程.
// ActivityThread，这里多说一句， ApplicationThread 是其内部类 private void handleCreateService(CreateServiceData data) { ... java.lang.ClassLoader cl = packageInfo.getClassLoader(); Service service = (Service) cl.loadClass(data.info.name).newInstance(); ... try { //创建ContextImpl对象 ContextImpl context = ContextImpl.createAppContext(this, packageInfo); context.setOuterContext(service); //创建Application对象 Application app = packageInfo.makeApplication(false, mInstrumentation); service.attach(context, this, data.info.name, data.token, app, ActivityManagerNative.getDefault()); //调用服务onCreate()方法 service.onCreate(); // ActivityManagerNative.getDefault().serviceDoneExecuting( data.token, SERVICE_DONE_EXECUTING_ANON, 0, 0); } catch (Exception e) { ... } }
在这个过程会创建目标服务对象,以及回调 onCreate() 方法, 紧接再次经过多次调用回到 system_server 来执行 serviceDoneExecuting 。
// ActiveServices private void serviceDoneExecutingLocked(ServiceRecord r, boolean inDestroying, boolean finishing) { ... if (r.executeNesting <= 0) { if (r.app != null) { r.app.execServicesFg = false; r.app.executingServices.remove(r); if (r.app.executingServices.size() == 0) { //当前服务所在进程中没有正在执行的service mAm.mHandler.removeMessages(ActivityManagerService.SERVICE_TIMEOUT_MSG, r.app); ... } ... }
// How long we wait for a service to finish executing.
static final int SERVICE_TIMEOUT = 20*1000;
该方法会在 service 启动完成后移除服务超时消息 SERVICE_TIMEOUT_MSG，时间是 20s。
2.3 引爆炸弹
前面介绍了埋炸弹和拆炸弹的过程, 如果在炸弹倒计时结束之前成功拆卸炸弹,那么就没有爆炸的机会, 但是世事难料. 总有些极端情况下无法即时拆除炸弹,导致炸弹爆炸, 其结果就是 App 发生 ANR. 接下来,带大家来看看炸弹爆炸的现场:
在 system_server 进程中有一个Handler线程，当倒计时结束便会向该 Handler 线程发送一条信息SERVICE_TIMEOUT_MSG,
// ActivityManagerService.java ::MainHandler final class MainHandler extends Handler { public MainHandler(Looper looper) { super(looper, null, true); } @Override public void handleMessage(Message msg) { switch (msg.what) { ......case SERVICE_TIMEOUT_MSG: { mServices.serviceTimeout((ProcessRecord)msg.obj); } break; } }
当延时时间到了之后，就会对消息进行处理，下面看下具体处理逻辑：
oid serviceTimeout(ProcessRecord proc) { String anrMessage = null; synchronized(mAm) { if (proc.executingServices.size() == 0 || proc.thread == null) { return; } final long now = SystemClock.uptimeMillis(); final long maxTime = now - (proc.execServicesFg ? SERVICE_TIMEOUT : SERVICE_BACKGROUND_TIMEOUT); ServiceRecord timeout = null; long nextTime = 0; for (int i=proc.executingServices.size()-1; i>=0; i--) {
// 从进程里面获取正在运行的 service ServiceRecord sr = proc.executingServices.valueAt(i); if (sr.executingStart < maxTime) { timeout = sr; break; } if (sr.executingStart > nextTime) { nextTime = sr.executingStart; } } if (timeout != null && mAm.mLruProcesses.contains(proc)) { Slog.w(TAG, "Timeout executing service: " + timeout); StringWriter sw = new StringWriter(); PrintWriter pw = new FastPrintWriter(sw, false, 1024); pw.println(timeout); timeout.dump(pw, " "); pw.close(); mLastAnrDump = sw.toString(); mAm.mHandler.removeCallbacks(mLastAnrDumpClearer); mAm.mHandler.postDelayed(mLastAnrDumpClearer, LAST_ANR_LIFETIME_DURATION_MSECS); anrMessage = "executing service " + timeout.shortName; } } if (anrMessage != null) { //当存在timeout的service，则执行appNotResponding mAm.appNotResponding(proc, null, null, false, anrMessage); } }
其中anrMessage的内容为”executing service [发送超时serviceRecord信息]”;
2.4 前台与后台服务的区别
系统对前台服务启动的超时为20s，而后台服务超时为200s，那么系统是如何区别前台还是后台服务呢？来看看ActiveServices的核心逻辑：
ComponentName startServiceLocked(...) { final boolean callerFg; if (caller != null) { final ProcessRecord callerApp = mAm.getRecordForAppLocked(caller); callerFg = callerApp.setSchedGroup != ProcessList.SCHED_GROUP_BACKGROUND; } else { callerFg = true; } ... ComponentName cmp = startServiceInnerLocked(smap, service, r, callerFg, addToStarting); return cmp; }
在startService过程根据发起方进程 callerApp 所属的进程调度组来决定被启动的服务是属于前台还是后台。当发起方进程不等于ProcessList.SCHED_GROUP_BACKGROUND (后台进程组) 则认为是前台服务，否则为后台服务，并标记在ServiceRecord的成员变量createdFromFg。
什么进程属于SCHED_GROUP_BACKGROUND调度组呢？进程调度组大体可分为TOP、前台、后台，进程优先级（Adj）和进程调度组（SCHED_GROUP）算法较为复杂，其对应关系可粗略理解为Adj等于0的进程属于Top进程组，Adj等于100或者200的进程属于前台进程组，Adj大于200的进程属于后台进程组。关于Adj的含义见下表，简单来说就是Adj>200的进程对用户来说基本是无感知，主要是做一些后台工作，故后台服务拥有更长的超时阈值，同时后台服务属于后台进程调度组，相比前台服务属于前台进程调度组，分配更少的CPU时间片。
前台服务准确来说，是指由处于前台进程调度组的进程发起的服务。这跟常说的fg-service服务有所不同，fg-service是指挂有前台通知的服务。
需要注意的问题，如果日志中出现 Reason: executing service com.example.baidu/.AnrService 也不一定是因为服务本身耗时导致，比如启动服务后，执行了耗时的操作，启动服务时onCreate函数或者 onStartCommand函数不能执行，超时后，仍然会造成anr
三、BroadcastReceiver
BroadcastReceiver Timeout 是位于”ActivityManager”线程中的BroadcastQueue.BroadcastHandler收到BROADCAST_TIMEOUT_MSG消息时触发。
对于广播队列有两个: foreground 队列和 background 队列:
- 对于前台广播，则超时为 BROADCAST_FG_TIMEOUT = 10s；
- 对于后台广播，则超时为 BROADCAST_BG_TIMEOUT = 60s
3.1 埋炸弹
先看发送广播的逻辑：
// ActivityManagerService.java] public final int broadcastIntent(IApplicationThread caller, Intent intent, String resolvedType, IIntentReceiver resultTo, int resultCode, String resultData, Bundle resultExtras, String[] requiredPermissions, int appOp, Bundle bOptions, boolean serialized, boolean sticky, int userId) { enforceNotIsolatedCaller("broadcastIntent"); synchronized(this) {
// 验证广播的有效性 intent = verifyBroadcastLocked(intent); // 获取发送广播的进程信息 final ProcessRecord callerApp = getRecordForAppLocked(caller); final int callingPid = Binder.getCallingPid(); final int callingUid = Binder.getCallingUid(); final long origId = Binder.clearCallingIdentity(); try { return broadcastIntentLocked(callerApp, callerApp != null ? callerApp.info.packageName : null, intent, resolvedType, resultTo, resultCode, resultData, resultExtras, requiredPermissions, appOp, bOptions, serialized, sticky, callingPid, callingUid, callingUid, callingPid, userId); } finally { Binder.restoreCallingIdentity(origId); } } }
broadcastIntent()方法有两个布尔参数 serialized 和 sticky 来共同决定是普通广播，有序广播，还是 Sticky 广播，参数如下：
| 类型 | serialized | sticky |
|---|---|---|
| sendBroadcast | false | false |
| sendOrderedBroadcast | true | false |
| sendStickyBroadcast | false | true |
说完发送广播，接下去就要讲讲讲收广播的操作了。
首先广播发出去之后，肯定会存在一个队列里面来进行处理。
// ActivityManagerService public ActivityManagerService(Context systemContext, ActivityTaskManagerService atm) { // ...... 创建了三个队列来保存不同的广播类型 mFgBroadcastQueue = new BroadcastQueue(this, mHandler, "foreground", foreConstants, false); mBgBroadcastQueue = new BroadcastQueue(this, mHandler, "background", backConstants, true); mOffloadBroadcastQueue = new BroadcastQueue(this, mHandler, "offload", offloadConstants, true); mBroadcastQueues[0] = mFgBroadcastQueue; mBroadcastQueues[1] = mBgBroadcastQueue; mBroadcastQueues[2] = mOffloadBroadcastQueue; }
在 ams 的构造函数里面，可以发现这里对广播进行了分类，分别有前台广播，后台广播，Offload 广播，并用一个新的数组将这三个队列放在一起。这里的 handler 是 MainHandler，也就是主线程的。传入是为了获取其 looper 。
BroadcastQueue(ActivityManagerService service, Handler handler, String name, BroadcastConstants constants, boolean allowDelayBehindServices) { mService = service;
// 广播的 handler 主要是获取到 ams 中 handler looper 来创建的 mHandler = new BroadcastHandler(handler.getLooper()); mQueueName = name; mDelayBehindServices = allowDelayBehindServices; mConstants = constants; mDispatcher = new BroadcastDispatcher(this, mConstants, mHandler, mService); }
下面就说下处理广播的逻辑：
private final class BroadcastHandler extends Handler { public BroadcastHandler(Looper looper) { super(looper, null, true); } @Override public void handleMessage(Message msg) { switch (msg.what) { case BROADCAST_INTENT_MSG: { if (DEBUG_BROADCAST) Slog.v( TAG_BROADCAST, "Received BROADCAST_INTENT_MSG [" + mQueueName + "]");
// 开始处理广播 processNextBroadcast(true); } break; case BROADCAST_TIMEOUT_MSG: { synchronized (mService) { broadcastTimeoutLocked(true); } } break; } } }
可以发现这里调用的是 processNextBroadcast 方法来处理广播。
final void processNextBroadcast(boolean fromMsg) { synchronized(mService) { //part1: 处理并行广播 while (mParallelBroadcasts.size() > 0) { r = mParallelBroadcasts.remove(0); r.dispatchTime = SystemClock.uptimeMillis(); r.dispatchClockTime = System.currentTimeMillis(); final int N = r.receivers.size(); for (int i=0; i<N; i++) { Object target = r.receivers.get(i); //分发广播给已注册的receiver deliverToRegisteredReceiverLocked(r, (BroadcastFilter)target, false); } addBroadcastToHistoryLocked(r);//将广播添加历史统计 } //part2: 处理当前有序广播 do { if (mOrderedBroadcasts.size() == 0) { mService.scheduleAppGcsLocked(); //没有更多的广播等待处理 if (looped) { mService.updateOomAdjLocked(); } return; } r = mOrderedBroadcasts.get(0); //获取串行广播的第一个广播 boolean forceReceive = false; int numReceivers = (r.receivers != null) ? r.receivers.size() : 0; if (mService.mProcessesReady && r.dispatchTime > 0) { long now = SystemClock.uptimeMillis(); if ((numReceivers > 0) && (now > r.dispatchTime + (2*mTimeoutPeriod*numReceivers))) { broadcastTimeoutLocked(false); //当广播处理时间超时，则强制结束这条广播 } } ... if (r.receivers == null || r.nextReceiver >= numReceivers || r.resultAbort || forceReceive) { if (r.resultTo != null) { //处理广播消息消息，调用到onReceive() performReceiveLocked(r.callerApp, r.resultTo, new Intent(r.intent), r.resultCode, r.resultData, r.resultExtras, false, false, r.userId); } cancelBroadcastTimeoutLocked(); //取消BROADCAST_TIMEOUT_MSG消息 addBroadcastToHistoryLocked(r); mOrderedBroadcasts.remove(0); continue; } } while (r == null); //part3: 获取下一个receiver r.receiverTime = SystemClock.uptimeMillis(); if (recIdx == 0) { r.dispatchTime = r.receiverTime; r.dispatchClockTime = System.currentTimeMillis(); } if (!mPendingBroadcastTimeoutMessage) { long timeoutTime = r.receiverTime + mTimeoutPeriod; setBroadcastTimeoutLocked(timeoutTime); //设置广播超时延时消息 } //part4: 处理下条有序广播 ProcessRecord app = mService.getProcessRecordLocked(targetProcess, info.activityInfo.applicationInfo.uid, false); if (app != null && app.thread != null) { app.addPackage(info.activityInfo.packageName, info.activityInfo.applicationInfo.versionCode, mService.mProcessStats); processCurBroadcastLocked(r, app); //[处理串行广播] return; ... } //该receiver所对应的进程尚未启动，则创建该进程 if ((r.curApp=mService.startProcessLocked(targetProcess, info.activityInfo.applicationInfo, true, r.intent.getFlags() | Intent.FLAG_FROM_BACKGROUND, "broadcast", r.curComponent, (r.intent.getFlags()&Intent.FLAG_RECEIVER_BOOT_UPGRADE) != 0, false, false)) == null) { ... return; } } }
对于广播超时处理时机：
-
首先在part3的过程中setBroadcastTimeoutLocked(timeoutTime) 设置超时广播消息；
-
然后在part2根据广播处理情况来处理：
-
当广播接收者等待时间过长，则调用 broadcastTimeoutLocked(false)；也就是引爆炸弹
-
当执行完广播,则调用 cancelBroadcastTimeoutLocked; 也就是拆除炸弹
-
// BroadcastQueue final void setBroadcastTimeoutLocked(long timeoutTime) { if (! mPendingBroadcastTimeoutMessage) { Message msg = mHandler.obtainMessage(BROADCAST_TIMEOUT_MSG, this); mHandler.sendMessageAtTime(msg, timeoutTime); mPendingBroadcastTimeoutMessage = true; } }
设置定时广播 BROADCAST_TIMEOUT_MSG，即当前往后推 mTimeoutPeriod 时间广播还没处理完毕，则进入广播超时流程。
// BroadcastConstants.java
private static final long DEFAULT_TIMEOUT = 10_000; // Timeout period for this broadcast queue public long TIMEOUT = DEFAULT_TIMEOUT; // Unspecified fields retain their current value rather than revert to default 超时时间还是可以设置的 TIMEOUT = mParser.getLong(KEY_TIMEOUT, TIMEOUT);
来看下具体时间的设置，超时设置的是 10 s。
3.2 拆炸弹
broadcast跟service超时机制大抵相同：
// 取消超时 final void cancelBroadcastTimeoutLocked() { if (mPendingBroadcastTimeoutMessage) { // 移除消息 mHandler.removeMessages(BROADCAST_TIMEOUT_MSG, this); mPendingBroadcastTimeoutMessage = false; } }
移除广播超时消息 BROADCAST_TIMEOUT_MSG，这样就把诈弹拆除了。
3.3 引爆炸弹
下面看下引爆炸弹的逻辑，前面我们已经介绍了 BroadcastQueue 中的 handler 的实现了，下面直接看下超时的处理逻辑：
//fromMsg = true final void broadcastTimeoutLocked(boolean fromMsg) { if (fromMsg) { mPendingBroadcastTimeoutMessage = false; } if (mOrderedBroadcasts.size() == 0) { return; } long now = SystemClock.uptimeMillis(); BroadcastRecord r = mOrderedBroadcasts.get(0); if (fromMsg) { if (mService.mDidDexOpt) { mService.mDidDexOpt = false; long timeoutTime = SystemClock.uptimeMillis() + mTimeoutPeriod; setBroadcastTimeoutLocked(timeoutTime); return; } if (!mService.mProcessesReady) { return; //当系统还没有准备就绪时，广播处理流程中不存在广播超时 } long timeoutTime = r.receiverTime + mTimeoutPeriod; if (timeoutTime > now) { //如果当前正在执行的receiver没有超时，则重新设置广播超时 setBroadcastTimeoutLocked(timeoutTime); return; } } BroadcastRecord br = mOrderedBroadcasts.get(0); if (br.state == BroadcastRecord.WAITING_SERVICES) { //广播已经处理完成，但需要等待已启动service执行完成。当等待足够时间，则处理下一条广播。 br.curComponent = null; br.state = BroadcastRecord.IDLE; processNextBroadcast(false); return; } r.receiverTime = now; //当前BroadcastRecord的anr次数执行加1操作 r.anrCount++; if (r.nextReceiver <= 0) { return; } ... Object curReceiver = r.receivers.get(r.nextReceiver-1); //查询App进程 if (curReceiver instanceof BroadcastFilter) { BroadcastFilter bf = (BroadcastFilter)curReceiver; if (bf.receiverList.pid != 0 && bf.receiverList.pid != ActivityManagerService.MY_PID) { synchronized (mService.mPidsSelfLocked) { app = mService.mPidsSelfLocked.get( bf.receiverList.pid); } } } else { app = r.curApp; } if (app != null) { anrMessage = "Broadcast of " + r.intent.toString(); } if (mPendingBroadcast == r) { mPendingBroadcast = null; } //继续移动到下一个广播接收者 finishReceiverLocked(r, r.resultCode, r.resultData, r.resultExtras, r.resultAbort, false); scheduleBroadcastsLocked(); if (anrMessage != null) { // 发送 anr 消息，带上了 anr 进程信息和 anr 消息 mHandler.post(new AppNotResponding(app, anrMessage)); } }
-
mOrderedBroadcasts已处理完成，则不会anr;
-
正在执行dexopt，则不会anr;
-
系统还没有进入ready状态(mProcessesReady=false)，则不会anr;
-
如果当前正在执行的receiver没有超时，则重新设置广播超时，不会anr;
来看下 AppNotResponding 实现：
private final class AppNotResponding implements Runnable { private final ProcessRecord mApp; private final String mAnnotation; public AppNotResponding(ProcessRecord app, String annotation) { mApp = app; mAnnotation = annotation; } @Override public void run() { mApp.appNotResponding(null, null, null, null, false, mAnnotation); } }
最终会让 ProcessRecord 来处理 anr，并且其内部持有 ActivityManagerService 实例。
3.4 前台与后台广播超时
前台广播超时为10s，后台广播超时为60s，那么如何区分前台和后台广播呢？来看看AMS的核心逻辑：
BroadcastQueue broadcastQueueForIntent(Intent intent) { final boolean isFg = (intent.getFlags() & Intent.FLAG_RECEIVER_FOREGROUND) != 0; return (isFg) ? mFgBroadcastQueue : mBgBroadcastQueue; } mFgBroadcastQueue = new BroadcastQueue(this, mHandler, "foreground", BROADCAST_FG_TIMEOUT, false); mBgBroadcastQueue = new BroadcastQueue(this, mHandler, "background", BROADCAST_BG_TIMEOUT, true);
根据发送广播sendBroadcast(Intent intent)中的intent的flags是否包含 FLAG_RECEIVER_FOREGROUND 来决定把该广播是放入前台广播队列或者后台广播队列，前台广播队列的超时为10s，后台广播队列的超时为60s，默认情况下广播是放入后台广播队列，除非指明加上 FLAG_RECEIVER_FOREGROUND 标识。
后台广播比前台广播拥有更长的超时阈值，同时在广播分发过程遇到后台service的启动(mDelayBehindServices)会延迟分发广播，等待service的完成，因为等待service而导致的广播ANR会被忽略掉；后台广播属于后台进程调度组，而前台广播属于前台进程调度组。简而言之，后台广播更不容易发生ANR，同时执行的速度也会更慢。
另外，只有串行处理的广播才有超时机制，因为接收者是串行处理的，前一个receiver处理慢，会影响后一个receiver；并行广播通过一个循环一次性向所有的receiver分发广播事件，所以不存在彼此影响的问题，则没有广播超时。
前台广播准确来说，是指位于前台广播队列的广播。
四 ContentProvider
ContentProvider Timeout是位于”ActivityManager”线程中的AMS.MainHandler收到CONTENT_PROVIDER_PUBLISH_TIMEOUT_MSG消息时触发。
ContentProvider 超时为CONTENT_PROVIDER_PUBLISH_TIMEOUT = 10s. 这个跟前面的Service和BroadcastQueue完全不同, 由 Provider 进程启动过程相关.
4.1 埋炸弹
埋炸弹的过程其实是在进程创建的过程，进程创建后会调用attachApplicationLocked() 进入system_server进程。
// ActivityManagerService private final boolean attachApplicationLocked(IApplicationThread thread, int pid) { ProcessRecord app; if (pid != MY_PID && pid >= 0) { synchronized (mPidsSelfLocked) { app = mPidsSelfLocked.get(pid); // 根据pid获取ProcessRecord } } ... //系统处于ready状态或者该app为FLAG_PERSISTENT进程则为true boolean normalMode = mProcessesReady || isAllowedWhileBooting(app.info); List<ProviderInfo> providers = normalMode ? generateApplicationProvidersLocked(app) : null; //app进程存在正在启动中的provider,则超时10s后发送CONTENT_PROVIDER_PUBLISH_TIMEOUT_MSG消息 if (providers != null && checkAppInLaunchingProvidersLocked(app)) { Message msg = mHandler.obtainMessage(CONTENT_PROVIDER_PUBLISH_TIMEOUT_MSG); msg.obj = app; mHandler.sendMessageDelayed(msg, CONTENT_PROVIDER_PUBLISH_TIMEOUT); } thread.bindApplication(...); ... }
// 10s
static final int CONTENT_PROVIDER_PUBLISH_TIMEOUT = 10*1000;
10s 之后引爆该炸弹.
4.2 拆炸弹
当 provider 成功 publish 之后,便会拆除该炸弹.
public final void publishContentProviders(IApplicationThread caller, List<ContentProviderHolder> providers) { ... synchronized (this) { final ProcessRecord r = getRecordForAppLocked(caller); final int N = providers.size(); for (int i = 0; i < N; i++) { ContentProviderHolder src = providers.get(i); ... ContentProviderRecord dst = r.pubProviders.get(src.info.name); if (dst != null) { ComponentName comp = new ComponentName(dst.info.packageName, dst.info.name); mProviderMap.putProviderByClass(comp, dst); //将该provider添加到mProviderMap String names[] = dst.info.authority.split(";"); for (int j = 0; j < names.length; j++) { mProviderMap.putProviderByName(names[j], dst); } int launchingCount = mLaunchingProviders.size(); int j; boolean wasInLaunchingProviders = false; for (j = 0; j < launchingCount; j++) { if (mLaunchingProviders.get(j) == dst) { //将该provider移除mLaunchingProviders队列 mLaunchingProviders.remove(j); wasInLaunchingProviders = true; j--; launchingCount--; } } //成功pubish则移除该消息 if (wasInLaunchingProviders) { mHandler.removeMessages(CONTENT_PROVIDER_PUBLISH_TIMEOUT_MSG, r); } synchronized (dst) { dst.provider = src.provider; dst.proc = r; //唤醒客户端的wait等待方法 dst.notifyAll(); } ... } } } }
4.3 引爆炸弹
在system_server进程中有一个Handler线程, 名叫”ActivityManager”.当倒计时结束便会向该Handler线程发送 一条信息CONTENT_PROVIDER_PUBLISH_TIMEOUT_MSG. MainHandler 是 AMS 的内部类。
final class MainHandler extends Handler { public void handleMessage(Message msg) { switch (msg.what) { case CONTENT_PROVIDER_PUBLISH_TIMEOUT_MSG: { ... ProcessRecord app = (ProcessRecord)msg.obj; synchronized (ActivityManagerService.this) { //【见小节4.3.2】 processContentProviderPublishTimedOutLocked(app); } } break; ... } ... } } private final void processContentProviderPublishTimedOutLocked(ProcessRecord app) { //[见4.3.3] cleanupAppInLaunchingProvidersLocked(app, true); //[见小节4.3.4] removeProcessLocked(app, false, true, "timeout publishing content providers"); } boolean cleanupAppInLaunchingProvidersLocked(ProcessRecord app, boolean alwaysBad) { boolean restart = false; for (int i = mLaunchingProviders.size() - 1; i >= 0; i--) { ContentProviderRecord cpr = mLaunchingProviders.get(i); if (cpr.launchingApp == app) { if (!alwaysBad && !app.bad && cpr.hasConnectionOrHandle()) { restart = true; } else { //移除死亡的provider removeDyingProviderLocked(app, cpr, true); } } } return restart; }
removeDyingProviderLocked()的功能跟进程的存活息息相关：详见ContentProvider引用计数 []小节4.5]
-
对于stable类型的provider(即conn.stableCount > 0),则会杀掉所有跟该provider建立stable连接的非persistent进程.
-
对于unstable类的provider(即conn.unstableCount > 0),并不会导致client进程被级联所杀.
五、input超时机制
input的超时检测机制跟service、broadcast、provider截然不同，为了更好的理解input过程先来介绍两个重要线程的相关工作：
-
InputReader线程负责通过EventHub(监听目录/dev/input)读取输入事件，一旦监听到输入事件则放入到InputDispatcher的mInBoundQueue队列，并通知其处理该事件；
-
InputDispatcher线程负责将接收到的输入事件分发给目标应用窗口，分发过程使用到3个事件队列：
-
mInBoundQueue用于记录InputReader发送过来的输入事件；
-
outBoundQueue用于记录即将分发给目标应用窗口的输入事件；
-
waitQueue用于记录已分发给目标应用，且应用尚未处理完成的输入事件；
-
input的超时机制并非时间到了一定就会爆炸，而是处理后续上报事件的过程才会去检测是否该爆炸，所以更像是扫雷的过程，具体如下图所示。
-
InputReader线程通过EventHub监听底层上报的输入事件，一旦收到输入事件则将其放至mInBoundQueue队列，并唤醒InputDispatcher线程
-
InputDispatcher开始分发输入事件，设置埋雷的起点时间。先检测是否有正在处理的事件(mPendingEvent)，如果没有则取出mInBoundQueue队头的事件，并将其赋值给mPendingEvent，且重置ANR的timeout；否则不会从mInBoundQueue中取出事件，也不会重置timeout。然后检查窗口是否就绪(checkWindowReadyForMoreInputLocked)，满足以下任一情况，则会进入扫雷状态(检测前一个正在处理的事件是否超时)，终止本轮事件分发，否则继续执行步骤3。当应用窗口准备就绪，则将mPendingEvent转移到outBoundQueue队列
-
对于按键类型的输入事件，则outboundQueue或者waitQueue不为空，
-
对于非按键的输入事件，则waitQueue不为空，且等待队头时间超时500ms
-
-
当outBoundQueue不为空，且应用管道对端连接状态正常，则将数据从outboundQueue中取出事件，放入waitQueue队列
-
InputDispatcher通过socket告知目标应用所在进程可以准备开始干活
-
App在初始化时默认已创建跟中控系统双向通信的socketpair，此时App的包工头(main线程)收到输入事件后，会层层转发到目标窗口来处理
-
包工头完成工作后，会通过socket向中控系统汇报工作完成，则中控系统会将该事件从waitQueue队列中移除。
input超时机制为什么是扫雷，而非定时爆炸呢？是由于对于input来说即便某次事件执行时间超过timeout时长，只要用户后续在没有再生成输入事件，则不会触发ANR。 这里的扫雷是指当前输入系统中正在处理着某个耗时事件的前提下，后续的每一次input事件都会检测前一个正在处理的事件是否超时（进入扫雷状态），检测当前的时间距离上次输入事件分发时间点是否超过timeout时长。如果前一个输入事件，则会重置ANR的timeout，从而不会爆炸。
到这里，关于 service ，广播，provider 的 anr 原因都讲清楚了。下面就看看是如何对 anr 信息进行收集的。
六、appNotResponding处理流程
不管是啥 anr ，最终都会调用到 ProcessRecord 的 appNotResponding 方法，下面来看看这个方法里面具体都做了啥：
// ProcessRecord.java void appNotResponding(String activityShortComponentName, ApplicationInfo aInfo, String parentShortComponentName, WindowProcessController parentProcess, boolean aboveSystem, String annotation) { ArrayList<Integer> firstPids = new ArrayList<>(5); SparseArray<Boolean> lastPids = new SparseArray<>(20); mWindowProcessController.appEarlyNotResponding(annotation, () -> kill("anr", true)); // anr 时间，实际上发生 anr 的时候，此时收集的运行堆栈有可能并不是引起 anr 的堆栈 long anrTime = SystemClock.uptimeMillis(); if (isMonitorCpuUsage()) { mService.updateCpuStatsNow(); } synchronized (mService) { // PowerManager.reboot() can block for a long time, so ignore ANRs while shutting down. 关机时发生 anr 会被忽略，因为可能会引起长时间阻塞 if (mService.mAtmInternal.isShuttingDown()) { Slog.i(TAG, "During shutdown skipping ANR: " + this + " " + annotation); return; } else if (isNotResponding()) { Slog.i(TAG, "Skipping duplicate ANR: " + this + " " + annotation); return; } else if (isCrashing()) { Slog.i(TAG, "Crashing app skipping ANR: " + this + " " + annotation); return; } else if (killedByAm) { Slog.i(TAG, "App already killed by AM skipping ANR: " + this + " " + annotation); return; } else if (killed) { Slog.i(TAG, "Skipping died app ANR: " + this + " " + annotation); return; } // In case we come through here for the same app before completing // this one, mark as anring now so we will bail out. 这样可以避免重复进入 setNotResponding(true); // Log the ANR to the event log. 记录 anr 到 eventlog EventLog.writeEvent(EventLogTags.AM_ANR, userId, pid, processName, info.flags, annotation); // Dump thread traces as quickly as we can, starting with "interesting" processes. 将当前进程添加到 firstPids 中 firstPids.add(pid); // Don't dump other PIDs if it's a background ANR if (!isSilentAnr()) { int parentPid = pid; if (parentProcess != null && parentProcess.getPid() > 0) { parentPid = parentProcess.getPid(); } if (parentPid != pid) firstPids.add(parentPid); // 将system_server进程添加到firstPids if (MY_PID != pid && MY_PID != parentPid) firstPids.add(MY_PID); for (int i = getLruProcessList().size() - 1; i >= 0; i--) { ProcessRecord r = getLruProcessList().get(i); if (r != null && r.thread != null) { int myPid = r.pid; if (myPid > 0 && myPid != pid && myPid != parentPid && myPid != MY_PID) { if (r.isPersistent()) { firstPids.add(myPid); // 将persistent进程添加到firstPids if (DEBUG_ANR) Slog.i(TAG, "Adding persistent proc: " + r); } else if (r.treatLikeActivity) { firstPids.add(myPid); // 使用了 BIND_TREAT_LIKE_ACTIVITY if (DEBUG_ANR) Slog.i(TAG, "Adding likely IME: " + r); } else { lastPids.put(myPid, Boolean.TRUE); // 其他进程添加到lastPids if (DEBUG_ANR) Slog.i(TAG, "Adding ANR proc: " + r); } } } } } } // Log the ANR to the main log. 记录 anr 到 mainlog StringBuilder info = new StringBuilder(); info.setLength(0); info.append("ANR in ").append(processName); if (activityShortComponentName != null) { info.append(" (").append(activityShortComponentName).append(")"); } info.append("\n"); info.append("PID: ").append(pid).append("\n"); if (annotation != null) { info.append("Reason: ").append(annotation).append("\n"); } if (parentShortComponentName != null && parentShortComponentName.equals(activityShortComponentName)) { info.append("Parent: ").append(parentShortComponentName).append("\n"); } // 创建 cpu tracker 对象 ProcessCpuTracker processCpuTracker = new ProcessCpuTracker(true); // don't dump native PIDs for background ANRs unless it is the process of interest String[] nativeProcs = null; if (isSilentAnr()) { for (int i = 0; i < NATIVE_STACKS_OF_INTEREST.length; i++) { if (NATIVE_STACKS_OF_INTEREST[i].equals(processName)) { nativeProcs = new String[] { processName }; break; } } } else { nativeProcs = NATIVE_STACKS_OF_INTEREST; } // 获取 native 进程 int[] pids = nativeProcs == null ? null : Process.getPidsForCommands(nativeProcs); ArrayList<Integer> nativePids = null; if (pids != null) { nativePids = new ArrayList<>(pids.length); for (int i : pids) { nativePids.add(i); } } // For background ANRs, don't pass the ProcessCpuTracker to // avoid spending 1/2 second collecting stats to rank lastPids. 收集堆栈信息 File tracesFile = ActivityManagerService.dumpStackTraces(firstPids, (isSilentAnr()) ? null : processCpuTracker, (isSilentAnr()) ? null : lastPids, nativePids); String cpuInfo = null;
// 添加 cpu 信息 if (isMonitorCpuUsage()) { mService.updateCpuStatsNow(); synchronized (mService.mProcessCpuTracker) { cpuInfo = mService.mProcessCpuTracker.printCurrentState(anrTime); } info.append(processCpuTracker.printCurrentLoad()); info.append(cpuInfo); } info.append(processCpuTracker.printCurrentState(anrTime)); Slog.e(TAG, info.toString()); if (tracesFile == null) { // There is no trace file, so dump (only) the alleged culprit's threads to the log Process.sendSignal(pid, Process.SIGNAL_QUIT); } StatsLog.write(StatsLog.ANR_OCCURRED, uid, processName, activityShortComponentName == null ? "unknown": activityShortComponentName, annotation, (this.info != null) ? (this.info.isInstantApp() ? StatsLog.ANROCCURRED__IS_INSTANT_APP__TRUE : StatsLog.ANROCCURRED__IS_INSTANT_APP__FALSE) : StatsLog.ANROCCURRED__IS_INSTANT_APP__UNAVAILABLE, isInterestingToUserLocked() ? StatsLog.ANROCCURRED__FOREGROUND_STATE__FOREGROUND : StatsLog.ANROCCURRED__FOREGROUND_STATE__BACKGROUND, getProcessClassEnum(), (this.info != null) ? this.info.packageName : ""); final ProcessRecord parentPr = parentProcess != null ? (ProcessRecord) parentProcess.mOwner : null;
// 将traces文件 和 CPU使用率信息保存到dropbox，即data/system/dropbox目录 mService.addErrorToDropBox("anr", this, processName, activityShortComponentName, parentShortComponentName, parentPr, annotation, cpuInfo, tracesFile, null); if (mWindowProcessController.appNotResponding(info.toString(), () -> kill("anr", true), () -> { synchronized (mService) { mService.mServices.scheduleServiceTimeoutLocked(this); } })) { return; } synchronized (mService) { // mBatteryStatsService can be null if the AMS is constructed with injector only. This // will only happen in tests. if (mService.mBatteryStatsService != null) { mService.mBatteryStatsService.noteProcessAnr(processName, uid); } // 杀死后台 anr 的进程 if (isSilentAnr() && !isDebugging()) {
kill("bg anr", true); return; } // Set the app's notResponding state, and look up the errorReportReceiver makeAppNotRespondingLocked(activityShortComponentName, annotation != null ? "ANR " + annotation : "ANR", info.toString()); // mUiHandler can be null if the AMS is constructed with injector only. This will only // happen in tests. if (mService.mUiHandler != null) { // Bring up the infamous App Not Responding dialog Message msg = Message.obtain(); msg.what = ActivityManagerService.SHOW_NOT_RESPONDING_UI_MSG; msg.obj = new AppNotRespondingDialog.Data(this, aInfo, aboveSystem); // 发送 anr 弹窗信息 mService.mUiHandler.sendMessage(msg); } } }
/**
* Unless configured otherwise, swallow ANRs in background processes & kill the process.
* Non-private access is for tests only. 如果是后台 ANR 会被吞噬，不会提示 anr，
*/
@VisibleForTesting
boolean isSilentAnr() {
return !getShowBackground() && !isInterestingForBackgroundTraces();
}
当发生ANR时, 会按顺序依次执行:
-
输出ANR Reason信息到EventLog. 也就是说ANR触发的时间点最接近的就是EventLog中输出的am_anr信息;
-
收集并输出重要进程列表中的各个线程的traces信息，该方法较耗时; 【见小节2】
-
输出当前各个进程的CPU使用情况以及CPU负载情况;
-
将traces文件和 CPU使用情况信息保存到dropbox，即data/system/dropbox目录
-
根据进程类型,来决定直接后台杀掉,还是弹框告知用户.
ANR输出重要进程的traces信息，这些进程包含:
-
firstPids队列：第一个是ANR进程，第二个是system_server，剩余是所有persistent进程；
-
Native队列：是指/system/bin/目录的mediaserver,sdcard 以及surfaceflinger进程；
-
lastPids队列: 是指mLruProcesses中的不属于firstPids的所有进程。
下面看下收集各进程堆栈信息逻辑：
// AMS /** * If a stack trace dump file is configured, dump process stack traces. * @param firstPids of dalvik VM processes to dump stack traces for first * @param lastPids of dalvik VM processes to dump stack traces for last * @param nativePids optional list of native pids to dump stack crawls */ public static File dumpStackTraces(ArrayList<Integer> firstPids, ProcessCpuTracker processCpuTracker, SparseArray<Boolean> lastPids, ArrayList<Integer> nativePids) { ArrayList<Integer> extraPids = null; Slog.i(TAG, "dumpStackTraces pids=" + lastPids + " nativepids=" + nativePids); // Measure CPU usage as soon as we're called in order to get a realistic sampling // of the top users at the time of the request. if (processCpuTracker != null) { processCpuTracker.init(); try { Thread.sleep(200); // 等待 200ms } catch (InterruptedException ignored) { } // 测量CPU使用情况 processCpuTracker.update(); // We'll take the stack crawls of just the top apps using CPU. 收集 5 个最高使用 cpu 的 进程 final int N = processCpuTracker.countWorkingStats(); extraPids = new ArrayList<>(); for (int i = 0; i < N && extraPids.size() < 5; i++) { ProcessCpuTracker.Stats stats = processCpuTracker.getWorkingStats(i); if (lastPids.indexOfKey(stats.pid) >= 0) { if (DEBUG_ANR) Slog.d(TAG, "Collecting stacks for extra pid " + stats.pid); extraPids.add(stats.pid); } else { Slog.i(TAG, "Skipping next CPU consuming process, not a java proc: " + stats.pid); } } } final File tracesDir = new File(ANR_TRACE_DIR); // Each set of ANR traces is written to a separate file and dumpstate will process // all such files and add them to a captured bug report if they're recent enough. 每一个 anr 都保存在单独的文件中的 maybePruneOldTraces(tracesDir); // NOTE: We should consider creating the file in native code atomically once we've // gotten rid of the old scheme of dumping and lot of the code that deals with paths // can be removed. 创建 anr 文件 File tracesFile = createAnrDumpFile(tracesDir); if (tracesFile == null) { return null; } // 收集 anr 堆栈 dumpStackTraces(tracesFile.getAbsolutePath(), firstPids, nativePids, extraPids); return tracesFile; } // 创建 anr 文件 private static synchronized File createAnrDumpFile(File tracesDir) { if (sAnrFileDateFormat == null) { sAnrFileDateFormat = new SimpleDateFormat("yyyy-MM-dd-HH-mm-ss-SSS"); } final String formattedDate = sAnrFileDateFormat.format(new Date());
// anr 文件名是 anr_加上时间 final File anrFile = new File(tracesDir, "anr_" + formattedDate); ...return anrFile; }
// 收集堆栈逻辑 public static void dumpStackTraces(String tracesFile, ArrayList<Integer> firstPids, ArrayList<Integer> nativePids, ArrayList<Integer> extraPids) { Slog.i(TAG, "Dumping to " + tracesFile); // We don't need any sort of inotify based monitoring when we're dumping traces via // tombstoned. Data is piped to an "intercept" FD installed in tombstoned so we're in full // control of all writes to the file in question. // We must complete all stack dumps within 20 seconds. 在 20s 里面完成堆栈收集工作，未完成也会直接退出 long remainingTime = 20 * 1000; // First collect all of the stacks of the most important pids. 收集最重要的几个进程的信息 if (firstPids != null) { int num = firstPids.size(); for (int i = 0; i < num; i++) { Slog.i(TAG, "Collecting stacks for pid " + firstPids.get(i)); final long timeTaken = dumpJavaTracesTombstoned(firstPids.get(i), tracesFile, remainingTime); remainingTime -= timeTaken; if (remainingTime <= 0) { Slog.e(TAG, "Aborting stack trace dump (current firstPid=" + firstPids.get(i) + "); deadline exceeded."); return; } } } // Next collect the stacks of the native pids 收集 native 堆栈 if (nativePids != null) { for (int pid : nativePids) { Slog.i(TAG, "Collecting stacks for native pid " + pid); final long nativeDumpTimeoutMs = Math.min(NATIVE_DUMP_TIMEOUT_MS, remainingTime); final long start = SystemClock.elapsedRealtime(); Debug.dumpNativeBacktraceToFileTimeout( pid, tracesFile, (int) (nativeDumpTimeoutMs / 1000)); final long timeTaken = SystemClock.elapsedRealtime() - start; remainingTime -= timeTaken;
... 超时则停止收集
} } // Lastly, dump stacks for all extra PIDs from the CPU tracker. 最后是前面最高的 5 个 if (extraPids != null) { for (int pid : extraPids) { Slog.i(TAG, "Collecting stacks for extra pid " + pid); final long timeTaken = dumpJavaTracesTombstoned(pid, tracesFile, remainingTime); remainingTime -= timeTaken; ... } } Slog.i(TAG, "Done dumping"); }
- 收集firstPids进程的stacks；
-
第一个是发生ANR进程；
-
第二个是system_server；
-
mLruProcesses中所有的persistent进程；
-
-
收集Native进程的stacks；(dumpNativeBacktraceToFile)
-
依次是mediaserver,sdcard,surfaceflinger进程；
-
-
收集lastPids进程的stacks;；
-
依次输出CPU使用率top 5的进程；
-
七、总结
当出现ANR时，都是调用到AMS.appNotResponding()方法，当然这里介绍的 provider 例外.
Timeout时长
-
对于前台服务，则超时为SERVICE_TIMEOUT = 20s；
-
对于后台服务，则超时为SERVICE_BACKGROUND_TIMEOUT = 200s
-
对于前台广播，则超时为BROADCAST_FG_TIMEOUT = 10s；
-
对于后台广播，则超时为BROADCAST_BG_TIMEOUT = 60s;
-
ContentProvider超时为CONTENT_PROVIDER_PUBLISH_TIMEOUT = 10s;
超时检测
Service超时检测机制：
- 超过一定时间没有执行完相应操作来触发移除延时消息，则会触发anr;
BroadcastReceiver超时检测机制：
-
有序广播的总执行时间超过 2* receiver个数 * timeout时长，则会触发anr;
-
有序广播的某一个receiver执行过程超过 timeout时长，则会触发anr;
另外:
-
对于Service, Broadcast, Input发生ANR之后,最终都会调用AMS.appNotResponding;
-
对于provider,在其进程启动时publish过程可能会出现ANR, 则会直接杀进程以及清理相应信息，而不会弹出ANR的对话框. appNotRespondingViaProvider()过程会走appNotResponding(), 这个就不介绍了，很少使用，由用户自定义超时时间.
最后，真诚感谢 gityuan 的博客。
参考文章
http://gityuan.com/2016/12/02/app-not-response/
http://gityuan.com/2016/07/02/android-anr/
