三、BroadcastReceiver Timeout
3.1 processNextBroadcastLocked
在学习Broadcast timeout 之前最好对广播的发生流程有一定了解,可以参考gityuan大神《Android Broadcast广播机制分析》。在广播流程有基础后,我们再来看最关键的代码块processNextBroadcastLocked,我们同样提两个问题:1.broadcast 什么情况下会ANR ? 2.ANR会有哪些log。
final void processNextBroadcastLocked(boolean fromMsg, boolean skipOomAdj) {
BroadcastRecord r;
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST, "processNextBroadcast ["
+ mQueueName + "]: "
+ mParallelBroadcasts.size() + " parallel broadcasts; "
+ mDispatcher.describeStateLocked());
mService.updateCpuStats();//更新cpu状态
//fromMsg是指processNextBroadcast()是否由BroadcastHandler所调用的. 第一次从sendBroadcast是true。
if (fromMsg) {
mBroadcastsScheduled = false;
}
// First, deliver any non-serialized broadcasts right away.
//part1: 处理并行广播,并行广播没有ANR的问题。
while (mParallelBroadcasts.size() > 0) {
r = mParallelBroadcasts.remove(0);//获取广播。
r.dispatchTime = SystemClock.uptimeMillis();
r.dispatchClockTime = System.currentTimeMillis();
if (Trace.isTagEnabled(Trace.TRACE_TAG_ACTIVITY_MANAGER)) {
Trace.asyncTraceEnd(Trace.TRACE_TAG_ACTIVITY_MANAGER,
createBroadcastTraceTitle(r, BroadcastRecord.DELIVERY_PENDING),
System.identityHashCode(r));
Trace.asyncTraceBegin(Trace.TRACE_TAG_ACTIVITY_MANAGER,
createBroadcastTraceTitle(r, BroadcastRecord.DELIVERY_DELIVERED),
System.identityHashCode(r));
}
final int N = r.receivers.size();
if (DEBUG_BROADCAST_LIGHT) Slog.v(TAG_BROADCAST, "Processing parallel broadcast ["
+ mQueueName + "] " + r);
for (int i=0; i<N; i++) {
Object target = r.receivers.get(i);
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST,
"Delivering non-ordered on [" + mQueueName + "] to registered "
+ target + ": " + r);
//分发广播给已注册的receive
deliverToRegisteredReceiverLocked(r, (BroadcastFilter)target, false, i);
}
addBroadcastToHistoryLocked(r);//将广播添加历史统计
if (DEBUG_BROADCAST_LIGHT) Slog.v(TAG_BROADCAST, "Done with parallel broadcast ["
+ mQueueName + "] " + r);
}
// Now take care of the next serialized one...
// If we are waiting for a process to come up to handle the next
// broadcast, then do nothing at this point. Just in case, we
// check that the process we're waiting for still exists.
//mPendingBroadcast 表示的是等待进程启动来处理的广播,如果进程存活则直接 return 继续等待
//如果该进程已经被杀,则重置 mPendingBroadcast 为空
if (mPendingBroadcast != null) {
if (DEBUG_BROADCAST_LIGHT) Slog.v(TAG_BROADCAST,
"processNextBroadcast [" + mQueueName + "]: waiting for "
+ mPendingBroadcast.curApp);
boolean isDead;
if (mPendingBroadcast.curApp.pid > 0) {
synchronized (mService.mPidsSelfLocked) {
ProcessRecord proc = mService.mPidsSelfLocked.get(
mPendingBroadcast.curApp.pid);
isDead = proc == null || proc.isCrashing();
}
} else {
final ProcessRecord proc = mService.mProcessList.mProcessNames.get(
mPendingBroadcast.curApp.processName, mPendingBroadcast.curApp.uid);
isDead = proc == null || !proc.pendingStart;
}
if (!isDead) {
// It's still alive, so keep waiting
return;
} else {
Slog.w(TAG, "pending app ["
+ mQueueName + "]" + mPendingBroadcast.curApp
+ " died before responding to broadcast");
mPendingBroadcast.state = BroadcastRecord.IDLE;
mPendingBroadcast.nextReceiver = mPendingBroadcastRecvIndex;
mPendingBroadcast = null;
}
}
boolean looped = false;
// 这后面的都是处理有序广播
do {
final long now = SystemClock.uptimeMillis();
r = mDispatcher.getNextBroadcastLocked(now);//从有序广播list中取出一个广播。
if (r == null) {//如果是空的话,说明没有广播需要处理,就返回。
// No more broadcasts are deliverable right now, so all done!
mDispatcher.scheduleDeferralCheckLocked(false);
mService.scheduleAppGcsLocked();//所有串行广播处理完成,则调度执行gc
if (looped) {
// If we had finished the last ordered broadcast, then
// make sure all processes have correct oom and sched
// adjustments.
mService.updateOomAdjLocked(OomAdjuster.OOM_ADJ_REASON_START_RECEIVER);
}
// when we have no more ordered broadcast on this queue, stop logging
if (mService.mUserController.mBootCompleted && mLogLatencyMetrics) {
mLogLatencyMetrics = false;
}
return;
}
boolean forceReceive = false;
// Ensure that even if something goes awry with the timeout
// detection, we catch "hung" broadcasts here, discard them,
// and continue to make progress.
//
// This is only done if the system is ready so that early-stage receivers
// don't get executed with timeouts; and of course other timeout-
// exempt broadcasts are ignored.
//如果广播处理的总时间超过2 * mConstants.TIMEOUT * numReceivers ,broadcastTimeoutLocked 也就ANR.
//所以在处理有序广播的时候不仅对单个广播时长有做限制,对于总的Receivers时间也有限制。
int numReceivers = (r.receivers != null) ? r.receivers.size() : 0;
if (mService.mProcessesReady && !r.timeoutExempt && r.dispatchTime > 0) {
if ((numReceivers > 0) &&
(now > r.dispatchTime + (2 * mConstants.TIMEOUT * numReceivers)) &&
/// M: ANR Debug Mechanism
!mService.mAnrManager.isAnrDeferrable()) {
Slog.w(TAG, "Hung broadcast ["
+ mQueueName + "] discarded after timeout failure:"
+ " now=" + now
+ " dispatchTime=" + r.dispatchTime
+ " startTime=" + r.receiverTime
+ " intent=" + r.intent
+ " numReceivers=" + numReceivers
+ " nextReceiver=" + r.nextReceiver
+ " state=" + r.state);
//发送超时广播。
broadcastTimeoutLocked(false); // forcibly finish this broadcast
forceReceive = true;
r.state = BroadcastRecord.IDLE;
}
}
if (r.state != BroadcastRecord.IDLE) {
if (DEBUG_BROADCAST) Slog.d(TAG_BROADCAST,
"processNextBroadcast("
+ mQueueName + ") called when not idle (state="
+ r.state + ")");
return;
}
// Is the current broadcast is done for any reason?
//如果广播接收者为空、下一个接收者下标越界、广播被中止等则继续分发下一个广播 ,r.receivers 什么时候会是null,finishReceiverLocked
if (r.receivers == null || r.nextReceiver >= numReceivers
|| r.resultAbort || forceReceive) {
// Send the final result if requested
if (r.resultTo != null) {
boolean sendResult = true;
// if this was part of a split/deferral complex, update the refcount and only
// send the completion when we clear all of them
if (r.splitToken != 0) {
int newCount = mSplitRefcounts.get(r.splitToken) - 1;
if (newCount == 0) {
// done! clear out this record's bookkeeping and deliver
if (DEBUG_BROADCAST_DEFERRAL) {
Slog.i(TAG_BROADCAST,
"Sending broadcast completion for split token "
+ r.splitToken + " : " + r.intent.getAction());
}
mSplitRefcounts.delete(r.splitToken);
} else {
// still have some split broadcast records in flight; update refcount
// and hold off on the callback
if (DEBUG_BROADCAST_DEFERRAL) {
Slog.i(TAG_BROADCAST,
"Result refcount now " + newCount + " for split token "
+ r.splitToken + " : " + r.intent.getAction()
+ " - not sending completion yet");
}
sendResult = false;
mSplitRefcounts.put(r.splitToken, newCount);
}
}
if (sendResult) {
try {
if (DEBUG_BROADCAST) {
Slog.i(TAG_BROADCAST, "Finishing broadcast [" + mQueueName + "] "
+ r.intent.getAction() + " app=" + r.callerApp);
}
//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);
// Set this to null so that the reference
// (local and remote) isn't kept in the mBroadcastHistory.
r.resultTo = null;
} catch (RemoteException e) {
r.resultTo = null;
Slog.w(TAG, "Failure ["
+ mQueueName + "] sending broadcast result of "
+ r.intent, e);
}
}
}
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST, "Cancelling BROADCAST_TIMEOUT_MSG");
//取消BROADCAST_TIMEOUT_MSG消息
cancelBroadcastTimeoutLocked();
if (DEBUG_BROADCAST_LIGHT) Slog.v(TAG_BROADCAST,
"Finished with ordered broadcast " + r);
// ... and on to the next...
addBroadcastToHistoryLocked(r);//已经发过的广播添加到list
if (r.intent.getComponent() == null && r.intent.getPackage() == null
&& (r.intent.getFlags()&Intent.FLAG_RECEIVER_REGISTERED_ONLY) == 0) {
// This was an implicit broadcast... let's record it for posterity.
mService.addBroadcastStatLocked(r.intent.getAction(), r.callerPackage,
r.manifestCount, r.manifestSkipCount, r.finishTime-r.dispatchTime);
}
mDispatcher.retireBroadcastLocked(r);
r = null;
looped = true;
continue;
}
// Check whether the next receiver is under deferral policy, and handle that
// accordingly. If the current broadcast was already part of deferred-delivery
// tracking, we know that it must now be deliverable as-is without re-deferral.
if (!r.deferred) {
final int receiverUid = r.getReceiverUid(r.receivers.get(r.nextReceiver));
if (mDispatcher.isDeferringLocked(receiverUid)) {
if (DEBUG_BROADCAST_DEFERRAL) {
Slog.i(TAG_BROADCAST, "Next receiver in " + r + " uid " + receiverUid
+ " at " + r.nextReceiver + " is under deferral");
}
// If this is the only (remaining) receiver in the broadcast, "splitting"
// doesn't make sense -- just defer it as-is and retire it as the
// currently active outgoing broadcast.
BroadcastRecord defer;
if (r.nextReceiver + 1 == numReceivers) {
if (DEBUG_BROADCAST_DEFERRAL) {
Slog.i(TAG_BROADCAST, "Sole receiver of " + r
+ " is under deferral; setting aside and proceeding");
}
defer = r;
mDispatcher.retireBroadcastLocked(r);
} else {
// Nontrivial case; split out 'uid's receivers to a new broadcast record
// and defer that, then loop and pick up continuing delivery of the current
// record (now absent those receivers).
// The split operation is guaranteed to match at least at 'nextReceiver'
defer = r.splitRecipientsLocked(receiverUid, r.nextReceiver);
if (DEBUG_BROADCAST_DEFERRAL) {
Slog.i(TAG_BROADCAST, "Post split:");
Slog.i(TAG_BROADCAST, "Original broadcast receivers:");
for (int i = 0; i < r.receivers.size(); i++) {
Slog.i(TAG_BROADCAST, " " + r.receivers.get(i));
}
Slog.i(TAG_BROADCAST, "Split receivers:");
for (int i = 0; i < defer.receivers.size(); i++) {
Slog.i(TAG_BROADCAST, " " + defer.receivers.get(i));
}
}
// Track completion refcount as well if relevant
if (r.resultTo != null) {
int token = r.splitToken;
if (token == 0) {
// first split of this record; refcount for 'r' and 'deferred'
r.splitToken = defer.splitToken = nextSplitTokenLocked();
mSplitRefcounts.put(r.splitToken, 2);
if (DEBUG_BROADCAST_DEFERRAL) {
Slog.i(TAG_BROADCAST,
"Broadcast needs split refcount; using new token "
+ r.splitToken);
}
} else {
// new split from an already-refcounted situation; increment count
final int curCount = mSplitRefcounts.get(token);
if (DEBUG_BROADCAST_DEFERRAL) {
if (curCount == 0) {
Slog.wtf(TAG_BROADCAST,
"Split refcount is zero with token for " + r);
}
}
mSplitRefcounts.put(token, curCount + 1);
if (DEBUG_BROADCAST_DEFERRAL) {
Slog.i(TAG_BROADCAST, "New split count for token " + token
+ " is " + (curCount + 1));
}
}
}
}
mDispatcher.addDeferredBroadcast(receiverUid, defer);
r = null;
looped = true;
continue;
}
}
} while (r == null);
// Get the next receiver...
// recIdx 表示当前接收者的下标,r.nextReceiver 为下一个接收者的下标,默认为 0
int recIdx = r.nextReceiver++;
// Keep track of when this receiver started, and make sure there
// is a timeout message pending to kill it if need be.
r.receiverTime = SystemClock.uptimeMillis();
if (recIdx == 0) {
r.dispatchTime = r.receiverTime;//设置分发时间,这个时间对timeout比较关键。
r.dispatchClockTime = System.currentTimeMillis();
if (mLogLatencyMetrics) {
StatsLog.write(
StatsLog.BROADCAST_DISPATCH_LATENCY_REPORTED,
r.dispatchClockTime - r.enqueueClockTime);
}
打印一些trace 信息。一般情况是不开启打印的。eng版本才有开启。
if (Trace.isTagEnabled(Trace.TRACE_TAG_ACTIVITY_MANAGER)) {
Trace.asyncTraceEnd(Trace.TRACE_TAG_ACTIVITY_MANAGER,
createBroadcastTraceTitle(r, BroadcastRecord.DELIVERY_PENDING),
System.identityHashCode(r));
Trace.asyncTraceBegin(Trace.TRACE_TAG_ACTIVITY_MANAGER,
createBroadcastTraceTitle(r, BroadcastRecord.DELIVERY_DELIVERED),
System.identityHashCode(r));
}
if (DEBUG_BROADCAST_LIGHT) Slog.v(TAG_BROADCAST, "Processing ordered broadcast ["
+ mQueueName + "] " + r);
}
//mPendingBroadcastTimeoutMessage 这个值在刚才执行cancelBroadcastTimeoutLocked 设置false。
if (! mPendingBroadcastTimeoutMessage) {
long timeoutTime = r.receiverTime + mConstants.TIMEOUT;
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST,
"Submitting BROADCAST_TIMEOUT_MSG ["
+ mQueueName + "] for " + r + " at " + timeoutTime);
setBroadcastTimeoutLocked(timeoutTime);//在10s 后发送BROADCAST_TIMEOUT_MSG 广播,最终会执行broadcastTimeoutLocked。后面有对这个方法说明。
}
final BroadcastOptions brOptions = r.options;
final Object nextReceiver = r.receivers.get(recIdx);
//如果是动态广播
if (nextReceiver instanceof BroadcastFilter) {
// Simple case: this is a registered receiver who gets
// a direct call.
BroadcastFilter filter = (BroadcastFilter)nextReceiver;
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST,
"Delivering ordered ["
+ mQueueName + "] to registered "
+ filter + ": " + r);
deliverToRegisteredReceiverLocked(r, filter, r.ordered, recIdx);//直接执行
//如果广播已经执行完的情况
if (r.receiver == null || !r.ordered) {
// The receiver has already finished, so schedule to
// process the next one.
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST, "Quick finishing ["
+ mQueueName + "]: ordered="
+ r.ordered + " receiver=" + r.receiver);
r.state = BroadcastRecord.IDLE;
scheduleBroadcastsLocked();
} else {
if (filter.receiverList != null) {
maybeAddAllowBackgroundActivityStartsToken(filter.receiverList.app, r);
// r is guaranteed ordered at this point, so we know finishReceiverLocked()
// will get a callback and handle the activity start token lifecycle.
}
if (brOptions != null && brOptions.getTemporaryAppWhitelistDuration() > 0) {
scheduleTempWhitelistLocked(filter.owningUid,
brOptions.getTemporaryAppWhitelistDuration(), r);
}
}
return;
}
// Hard case: need to instantiate the receiver, possibly
// starting its application process to host it.
//对于静态注册的广播接收者 对各种权限进行判断处理,跟之前的并行广播类似。
ResolveInfo info =
(ResolveInfo)nextReceiver;
ComponentName component = new ComponentName(
info.activityInfo.applicationInfo.packageName,
info.activityInfo.name);
boolean skip = false;
if (brOptions != null &&
(info.activityInfo.applicationInfo.targetSdkVersion
< brOptions.getMinManifestReceiverApiLevel() ||
info.activityInfo.applicationInfo.targetSdkVersion
> brOptions.getMaxManifestReceiverApiLevel())) {
skip = true;
}
if (!skip && !mService.validateAssociationAllowedLocked(r.callerPackage, r.callingUid,
component.getPackageName(), info.activityInfo.applicationInfo.uid)) {
Slog.w(TAG, "Association not allowed: broadcasting "
+ r.intent.toString()
+ " from " + r.callerPackage + " (pid=" + r.callingPid
+ ", uid=" + r.callingUid + ") to " + component.flattenToShortString());
skip = true;
}
if (!skip) {
skip = !mService.mIntentFirewall.checkBroadcast(r.intent, r.callingUid,
r.callingPid, r.resolvedType, info.activityInfo.applicationInfo.uid);
if (skip) {
Slog.w(TAG, "Firewall blocked: broadcasting "
+ r.intent.toString()
+ " from " + r.callerPackage + " (pid=" + r.callingPid
+ ", uid=" + r.callingUid + ") to " + component.flattenToShortString());
}
}
int perm = mService.checkComponentPermission(info.activityInfo.permission,
r.callingPid, r.callingUid, info.activityInfo.applicationInfo.uid,
info.activityInfo.exported);
if (!skip && perm != PackageManager.PERMISSION_GRANTED) {
if (!info.activityInfo.exported) {
Slog.w(TAG, "Permission Denial: broadcasting "
+ r.intent.toString()
+ " from " + r.callerPackage + " (pid=" + r.callingPid
+ ", uid=" + r.callingUid + ")"
+ " is not exported from uid " + info.activityInfo.applicationInfo.uid
+ " due to receiver " + component.flattenToShortString());
} else {
Slog.w(TAG, "Permission Denial: broadcasting "
+ r.intent.toString()
+ " from " + r.callerPackage + " (pid=" + r.callingPid
+ ", uid=" + r.callingUid + ")"
+ " requires " + info.activityInfo.permission
+ " due to receiver " + component.flattenToShortString());
}
skip = true;
} else if (!skip && info.activityInfo.permission != null) {
final int opCode = AppOpsManager.permissionToOpCode(info.activityInfo.permission);
if (opCode != AppOpsManager.OP_NONE
&& mService.mAppOpsService.noteOperation(opCode, r.callingUid,
r.callerPackage) != AppOpsManager.MODE_ALLOWED) {
Slog.w(TAG, "Appop Denial: broadcasting "
+ r.intent.toString()
+ " from " + r.callerPackage + " (pid="
+ r.callingPid + ", uid=" + r.callingUid + ")"
+ " requires appop " + AppOpsManager.permissionToOp(
info.activityInfo.permission)
+ " due to registered receiver "
+ component.flattenToShortString());
skip = true;
}
}
if (!skip && info.activityInfo.applicationInfo.uid != Process.SYSTEM_UID &&
r.requiredPermissions != null && r.requiredPermissions.length > 0) {
for (int i = 0; i < r.requiredPermissions.length; i++) {
String requiredPermission = r.requiredPermissions[i];
try {
perm = AppGlobals.getPackageManager().
checkPermission(requiredPermission,
info.activityInfo.applicationInfo.packageName,
UserHandle
.getUserId(info.activityInfo.applicationInfo.uid));
} catch (RemoteException e) {
perm = PackageManager.PERMISSION_DENIED;
}
if (perm != PackageManager.PERMISSION_GRANTED) {
Slog.w(TAG, "Permission Denial: receiving "
+ r.intent + " to "
+ component.flattenToShortString()
+ " requires " + requiredPermission
+ " due to sender " + r.callerPackage
+ " (uid " + r.callingUid + ")");
skip = true;
break;
}
int appOp = AppOpsManager.permissionToOpCode(requiredPermission);
if (appOp != AppOpsManager.OP_NONE && appOp != r.appOp
&& mService.mAppOpsService.noteOperation(appOp,
info.activityInfo.applicationInfo.uid, info.activityInfo.packageName)
!= AppOpsManager.MODE_ALLOWED) {
Slog.w(TAG, "Appop Denial: receiving "
+ r.intent + " to "
+ component.flattenToShortString()
+ " requires appop " + AppOpsManager.permissionToOp(
requiredPermission)
+ " due to sender " + r.callerPackage
+ " (uid " + r.callingUid + ")");
skip = true;
break;
}
}
}
if (!skip && r.appOp != AppOpsManager.OP_NONE
&& mService.mAppOpsService.noteOperation(r.appOp,
info.activityInfo.applicationInfo.uid, info.activityInfo.packageName)
!= AppOpsManager.MODE_ALLOWED) {
Slog.w(TAG, "Appop Denial: receiving "
+ r.intent + " to "
+ component.flattenToShortString()
+ " requires appop " + AppOpsManager.opToName(r.appOp)
+ " due to sender " + r.callerPackage
+ " (uid " + r.callingUid + ")");
skip = true;
}
boolean isSingleton = false;
try {
isSingleton = mService.isSingleton(info.activityInfo.processName,
info.activityInfo.applicationInfo,
info.activityInfo.name, info.activityInfo.flags);
} catch (SecurityException e) {
Slog.w(TAG, e.getMessage());
skip = true;
}
if ((info.activityInfo.flags&ActivityInfo.FLAG_SINGLE_USER) != 0) {
if (ActivityManager.checkUidPermission(
android.Manifest.permission.INTERACT_ACROSS_USERS,
info.activityInfo.applicationInfo.uid)
!= PackageManager.PERMISSION_GRANTED) {
Slog.w(TAG, "Permission Denial: Receiver " + component.flattenToShortString()
+ " requests FLAG_SINGLE_USER, but app does not hold "
+ android.Manifest.permission.INTERACT_ACROSS_USERS);
skip = true;
}
}
if (!skip && info.activityInfo.applicationInfo.isInstantApp()
&& r.callingUid != info.activityInfo.applicationInfo.uid) {
Slog.w(TAG, "Instant App Denial: receiving "
+ r.intent
+ " to " + component.flattenToShortString()
+ " due to sender " + r.callerPackage
+ " (uid " + r.callingUid + ")"
+ " Instant Apps do not support manifest receivers");
skip = true;
}
if (!skip && r.callerInstantApp
&& (info.activityInfo.flags & ActivityInfo.FLAG_VISIBLE_TO_INSTANT_APP) == 0
&& r.callingUid != info.activityInfo.applicationInfo.uid) {
Slog.w(TAG, "Instant App Denial: receiving "
+ r.intent
+ " to " + component.flattenToShortString()
+ " requires receiver have visibleToInstantApps set"
+ " due to sender " + r.callerPackage
+ " (uid " + r.callingUid + ")");
skip = true;
}
if (r.curApp != null && r.curApp.isCrashing()) {
// If the target process is crashing, just skip it.
Slog.w(TAG, "Skipping deliver ordered [" + mQueueName + "] " + r
+ " to " + r.curApp + ": process crashing");
skip = true;
}
if (!skip) {
boolean isAvailable = false;
try {
isAvailable = AppGlobals.getPackageManager().isPackageAvailable(
info.activityInfo.packageName,
UserHandle.getUserId(info.activityInfo.applicationInfo.uid));
} catch (Exception e) {
// all such failures mean we skip this receiver
Slog.w(TAG, "Exception getting recipient info for "
+ info.activityInfo.packageName, e);
}
if (!isAvailable) {
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST,
"Skipping delivery to " + info.activityInfo.packageName + " / "
+ info.activityInfo.applicationInfo.uid
+ " : package no longer available");
skip = true;
}
}
// If permissions need a review before any of the app components can run, we drop
// the broadcast and if the calling app is in the foreground and the broadcast is
// explicit we launch the review UI passing it a pending intent to send the skipped
// broadcast.
if (!skip) {
if (!requestStartTargetPermissionsReviewIfNeededLocked(r,
info.activityInfo.packageName, UserHandle.getUserId(
info.activityInfo.applicationInfo.uid))) {
skip = true;
}
}
// This is safe to do even if we are skipping the broadcast, and we need
// this information now to evaluate whether it is going to be allowed to run.
final int receiverUid = info.activityInfo.applicationInfo.uid;
// If it's a singleton, it needs to be the same app or a special app
if (r.callingUid != Process.SYSTEM_UID && isSingleton
&& mService.isValidSingletonCall(r.callingUid, receiverUid)) {
info.activityInfo = mService.getActivityInfoForUser(info.activityInfo, 0);
}
//该receiver所对应的进程已经运行,则直接处理
String targetProcess = info.activityInfo.processName;
ProcessRecord app = mService.getProcessRecordLocked(targetProcess,
info.activityInfo.applicationInfo.uid, false);
if (!skip) {
final int allowed = mService.getAppStartModeLocked(
info.activityInfo.applicationInfo.uid, info.activityInfo.packageName,
info.activityInfo.applicationInfo.targetSdkVersion, -1, true, false, false);
if (allowed != ActivityManager.APP_START_MODE_NORMAL) {
// We won't allow this receiver to be launched if the app has been
// completely disabled from launches, or it was not explicitly sent
// to it and the app is in a state that should not receive it
// (depending on how getAppStartModeLocked has determined that).
if (allowed == ActivityManager.APP_START_MODE_DISABLED) {
Slog.w(TAG, "Background execution disabled: receiving "
+ r.intent + " to "
+ component.flattenToShortString());
skip = true;
} else if (((r.intent.getFlags()&Intent.FLAG_RECEIVER_EXCLUDE_BACKGROUND) != 0)
|| (r.intent.getComponent() == null
&& r.intent.getPackage() == null
&& ((r.intent.getFlags()
& Intent.FLAG_RECEIVER_INCLUDE_BACKGROUND) == 0)
&& !isSignaturePerm(r.requiredPermissions))) {
mService.addBackgroundCheckViolationLocked(r.intent.getAction(),
component.getPackageName());
Slog.w(TAG, "Background execution not allowed: receiving "
+ r.intent + " to "
+ component.flattenToShortString());
skip = true;
}
}
}
if (!skip && !Intent.ACTION_SHUTDOWN.equals(r.intent.getAction())
&& !mService.mUserController
.isUserRunning(UserHandle.getUserId(info.activityInfo.applicationInfo.uid),
0 /* flags */)) {
skip = true;
Slog.w(TAG,
"Skipping delivery to " + info.activityInfo.packageName + " / "
+ info.activityInfo.applicationInfo.uid + " : user is not running");
}
/// M: DuraSpeed @{
boolean isSuppress = false;
isSuppress = mService.mAmsExt.onBeforeStartProcessForStaticReceiver(
info.activityInfo.packageName);
if (isSuppress) {
skip = true;
}
/// @}
/// M: For ago performance
if (!mService.mAmsExt.isComponentNeedsStart(info.activityInfo.packageName,
"static_broadcast")) {
skip = true;
}
if (skip) {
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST,
"Skipping delivery of ordered [" + mQueueName + "] "
+ r + " for reason described above");
r.delivery[recIdx] = BroadcastRecord.DELIVERY_SKIPPED;
r.receiver = null;
r.curFilter = null;
r.state = BroadcastRecord.IDLE;
r.manifestSkipCount++;
scheduleBroadcastsLocked();
return;
}
r.manifestCount++;
r.delivery[recIdx] = BroadcastRecord.DELIVERY_DELIVERED;
r.state = BroadcastRecord.APP_RECEIVE;
r.curComponent = component;
r.curReceiver = info.activityInfo;
if (DEBUG_MU && r.callingUid > UserHandle.PER_USER_RANGE) {
Slog.v(TAG_MU, "Updated broadcast record activity info for secondary user, "
+ info.activityInfo + ", callingUid = " + r.callingUid + ", uid = "
+ receiverUid);
}
if (brOptions != null && brOptions.getTemporaryAppWhitelistDuration() > 0) {
scheduleTempWhitelistLocked(receiverUid,
brOptions.getTemporaryAppWhitelistDuration(), r);
}
// Broadcast is being executed, its package can't be stopped.
try {
AppGlobals.getPackageManager().setPackageStoppedState(
r.curComponent.getPackageName(), false, r.userId);
} catch (RemoteException e) {
} catch (IllegalArgumentException e) {
Slog.w(TAG, "Failed trying to unstop package "
+ r.curComponent.getPackageName() + ": " + e);
}
// Is this receiver's application already running?
if (app != null && app.thread != null && !app.killed) {
try {
app.addPackage(info.activityInfo.packageName,
info.activityInfo.applicationInfo.longVersionCode, mService.mProcessStats);
maybeAddAllowBackgroundActivityStartsToken(app, r);
processCurBroadcastLocked(r, app, skipOomAdj);
return;
} catch (RemoteException e) {
Slog.w(TAG, "Exception when sending broadcast to "
+ r.curComponent, e);
} catch (RuntimeException e) {
Slog.wtf(TAG, "Failed sending broadcast to "
+ r.curComponent + " with " + r.intent, e);
// If some unexpected exception happened, just skip
// this broadcast. At this point we are not in the call
// from a client, so throwing an exception out from here
// will crash the entire system instead of just whoever
// sent the broadcast.
logBroadcastReceiverDiscardLocked(r);
finishReceiverLocked(r, r.resultCode, r.resultData,
r.resultExtras, r.resultAbort, false);
scheduleBroadcastsLocked();
// We need to reset the state if we failed to start the receiver.
r.state = BroadcastRecord.IDLE;
return;
}
// If a dead object exception was thrown -- fall through to
// restart the application.
}
// Not running -- get it started, to be executed when the app comes up.
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST,
"Need to start app ["
+ mQueueName + "] " + targetProcess + " for broadcast " + r);
//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,
new HostingRecord("broadcast", r.curComponent),
(r.intent.getFlags()&Intent.FLAG_RECEIVER_BOOT_UPGRADE) != 0, false, false))
== null) {
// Ah, this recipient is unavailable. Finish it if necessary,
// and mark the broadcast record as ready for the next.
//启动失败的情况
Slog.w(TAG, "Unable to launch app "
+ info.activityInfo.applicationInfo.packageName + "/"
+ receiverUid + " for broadcast "
+ r.intent + ": process is bad");
logBroadcastReceiverDiscardLocked(r);
finishReceiverLocked(r, r.resultCode, r.resultData,
r.resultExtras, r.resultAbort, false);
scheduleBroadcastsLocked();
r.state = BroadcastRecord.IDLE;
return;
}
maybeAddAllowBackgroundActivityStartsToken(r.curApp, r);
mPendingBroadcast = r;
mPendingBroadcastRecvIndex = recIdx;
}
上面的代码很长总的可以分为两部分:1.处理并行广播的情况。2 .处理串行广播的情况,也就我们说的有序广播。 而ANR只有是有序广播的时候才会发生。
归纳是:1.while 循环先把并行广播处理完成。
2. 从mDispatcher.getNextBroadcastLocked 获取 BroadcastRecord实例,如果是空的话,说明没有广播需要处理,就返回。
3.对总的 numReceivers时间进行判断,判断依据是:2 * mConstants.TIMEOUT * numReceivers 。如果超过这个时间就broadcastTimeoutLocked,产生ANR。系统不仅对单个receiver的时间有管控,总的receivers时间也有管控。
4 . cancelBroadcastTimeoutLocked,取消超时广播,这个对广播第一个receiver来说没什么意义,主要是对第二个之后的receiver的影响。
5. 如果是第一个receiver的话就把分发时间付给它。 if (recIdx == 0) {r.dispatchTime = r.receiverTime}。
6 .发生一个延迟广播 setBroadcastTimeoutLocked(timeoutTime),时间是10s ,如果超过这个时间还没有执行cancelBroadcastTimeoutLocked 。就会引发ANR。那么提个问题,什么情况下会执行cancelBroadcastTimeoutLocked 。答案是,当广播结束完会调用AMS finishReceiver方法,nextreceiver ,执行 r.queue.processNextBroadcastLocked。由回到processNextBroadcastLocked方法中,然后就会调用cancelBroadcastTimeoutLocked
7.如果是动态广播,就比较简单,执行deliverToRegisteredReceiverLocked,静态广播 要判断各种权限问题,还需要判断广播的进程是否存在,如果不存在需要创建新的进程。
public void finishReceiver(IBinder who, int resultCode, String resultData,
Bundle resultExtras, boolean resultAbort, int flags) {
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST, "Finish receiver: " + who);
// Refuse possible leaked file descriptors
if (resultExtras != null && resultExtras.hasFileDescriptors()) {
throw new IllegalArgumentException("File descriptors passed in Bundle");
}
final long origId = Binder.clearCallingIdentity();
try {
boolean doNext = false;
BroadcastRecord r;
BroadcastQueue queue;
synchronized(this) {
if (isOnOffloadQueue(flags)) {
queue = mOffloadBroadcastQueue;
} else {
queue = (flags & Intent.FLAG_RECEIVER_FOREGROUND) != 0
? mFgBroadcastQueue : mBgBroadcastQueue;
}
r = queue.getMatchingOrderedReceiver(who);
if (r != null) {
doNext = r.queue.finishReceiverLocked(r, resultCode,
resultData, resultExtras, resultAbort, true);
}
if (doNext) {
r.queue.processNextBroadcastLocked(/*fromMsg=*/ false, /*skipOomAdj=*/ true);
}
// updateOomAdjLocked() will be done here
trimApplicationsLocked(OomAdjuster.OOM_ADJ_REASON_FINISH_RECEIVER);
}
} finally {
Binder.restoreCallingIdentity(origId);
}
}
3.2 broadcastTimeoutLocked
final void broadcastTimeoutLocked(boolean fromMsg) {
if (fromMsg) {
mPendingBroadcastTimeoutMessage = false;
/// M: ANR Debug Mechanism
mService.mAnrManager.removeBroadcastMonitorMessage();
}
//有序广播已经执行结束,或者广播接受者已经finish了。
if (mDispatcher.isEmpty() || mDispatcher.getActiveBroadcastLocked() == null) {
return;
}
long now = SystemClock.uptimeMillis();
BroadcastRecord r = mDispatcher.getActiveBroadcastLocked();
if (fromMsg) {
/// M: ANR Debug Mechanism @{
if (mService.mAnrManager.isAnrDeferrable()) {
setBroadcastTimeoutLocked(SystemClock.uptimeMillis() + mConstants.TIMEOUT);
return;
} /// @}
if (!mService.mProcessesReady) {
// Only process broadcast timeouts if the system is ready; some early
// broadcasts do heavy work setting up system facilities
return;
}
// If the broadcast is generally exempt from timeout tracking, we're done
if (r.timeoutExempt) {
if (DEBUG_BROADCAST) {
Slog.i(TAG_BROADCAST, "Broadcast timeout but it's exempt: "
+ r.intent.getAction());
}
return;
}
long timeoutTime = r.receiverTime + mConstants.TIMEOUT;
//如果现在的时间小于超时时间,说明还没有超时,就再执行setBroadcastTimeoutLocked,TIMEOUT=10s
if (timeoutTime > now) {
// We can observe premature timeouts because we do not cancel and reset the
// broadcast timeout message after each receiver finishes. Instead, we set up
// an initial timeout then kick it down the road a little further as needed
// when it expires.
if (DEBUG_BROADCAST) Slog.v(TAG_BROADCAST,
"Premature timeout ["
+ mQueueName + "] @ " + now + ": resetting BROADCAST_TIMEOUT_MSG for "
+ timeoutTime);
setBroadcastTimeoutLocked(timeoutTime);
return;
}
}
//已经finish receiver ,也就是在执行 finishReceiverLocked方法时
if (r.state == BroadcastRecord.WAITING_SERVICES) {
// In this case the broadcast had already finished, but we had decided to wait
// for started services to finish as well before going on. So if we have actually
// waited long enough time timeout the broadcast, let's give up on the whole thing
// and just move on to the next.
Slog.i(TAG, "Waited long enough for: " + (r.curComponent != null
? r.curComponent.flattenToShortString() : "(null)"));
r.curComponent = null;
r.state = BroadcastRecord.IDLE;
processNextBroadcast(false);
return;
}
// If the receiver app is being debugged we quietly ignore unresponsiveness, just
// tidying up and moving on to the next broadcast without crashing or ANRing this
// app just because it's stopped at a breakpoint.
final boolean debugging = (r.curApp != null && r.curApp.isDebugging());
Slog.w(TAG, "Timeout of broadcast " + r + " - receiver=" + r.receiver
+ ", started " + (now - r.receiverTime) + "ms ago");
r.receiverTime = now;
if (!debugging) {
r.anrCount++;
}
ProcessRecord app = null;
String anrMessage = null;
Object curReceiver;
if (r.nextReceiver > 0) {
curReceiver = r.receivers.get(r.nextReceiver-1);
r.delivery[r.nextReceiver-1] = BroadcastRecord.DELIVERY_TIMEOUT;
} else {
curReceiver = r.curReceiver;
}
Slog.w(TAG, "Receiver during timeout of " + r + " : " + curReceiver);
logBroadcastReceiverDiscardLocked(r);
//查询App进程
if (curReceiver != null && 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) {//如果进程还存在没有被杀死的话创建anrMessage
anrMessage = "Broadcast of " + r.intent.toString();
}
if (mPendingBroadcast == r) {
mPendingBroadcast = null;
}
// Move on to the next receiver.
//关闭receiver 启动下一个receiver
finishReceiverLocked(r, r.resultCode, r.resultData,
r.resultExtras, r.resultAbort, false);
scheduleBroadcastsLocked();
if (!debugging && anrMessage != null) {
// Post the ANR to the handler since we do not want to process ANRs while
// potentially holding our lock.
mHandler.post(new AppNotResponding(app, anrMessage));
}
}
//向event log 里面写log,关键字是:am_broadcast_discard_app
final void logBroadcastReceiverDiscardLocked(BroadcastRecord r) {
final int logIndex = r.nextReceiver - 1;
if (logIndex >= 0 && logIndex < r.receivers.size()) {
Object curReceiver = r.receivers.get(logIndex);
if (curReceiver instanceof BroadcastFilter) {
BroadcastFilter bf = (BroadcastFilter) curReceiver;
EventLog.writeEvent(EventLogTags.AM_BROADCAST_DISCARD_FILTER,
bf.owningUserId, System.identityHashCode(r),
r.intent.getAction(), logIndex, System.identityHashCode(bf));
} else {
ResolveInfo ri = (ResolveInfo) curReceiver;
EventLog.writeEvent(EventLogTags.AM_BROADCAST_DISCARD_APP,
UserHandle.getUserId(ri.activityInfo.applicationInfo.uid),
System.identityHashCode(r), r.intent.getAction(), logIndex, ri.toString());
}
} else {
if (logIndex < 0) Slog.w(TAG,
"Discarding broadcast before first receiver is invoked: " + r);
EventLog.writeEvent(EventLogTags.AM_BROADCAST_DISCARD_APP,
-1, System.identityHashCode(r),
r.intent.getAction(),
r.nextReceiver,
"NONE");
}
}
3.3 mHandler.post(new AppNotResponding(app, anrMessage))
public void run() {
mApp.appNotResponding(null, null, null, null, false, mAnnotation);
}
mApp是ProcessRecord ,所有会调用ProcessRecord 的appNotResponding,这部分上一章有说过:深入理解 ANR 触发原理:一
所以对于Broadcast ANR问题算研究完毕。那么对于开始提的两个问题我们就回答了
1.broadcast 什么情况下会ANR ?
答:1.所属的广播必须是有序广播,并行广播不会产生ANR问题。
2.发生有序广播,receivers>=2。如果只有一个那么也是不会产生ANR.
3 前台超时10s,后台超时60s,如果是多个receivers 那么总的时间不能超过 2 * mConstants.TIMEOUT * numReceivers
4.如果 有序广播是最后一个执行者,或者进程已经被杀掉。也不会产生ANR. if (mDispatcher.isEmpty() || mDispatcher.getActiveBroadcastLocked() == null) {
return;
}
5.或者在执行broadcastTimeoutLocked 方法是,receiver在执行finishReceiverLocked的方法。
2.ANR会有哪些log
答:1. Slog.w(TAG, "Hung broadcast ["
+ mQueueName + "] discarded after timeout failure:"
+ " now=" + now
+ " dispatchTime=" + r.dispatchTime
+ " startTime=" + r.receiverTime
+ " intent=" + r.intent
+ " numReceivers=" + numReceivers
+ " nextReceiver=" + r.nextReceiver
+ " state=" + r.state);
2. 在event log 里面能搜索到 am_broadcast_discard_app ,是由logBroadcastReceiverDiscardLocked 方法得来。
3 .Slog.w(TAG, "Timeout of broadcast " + r + " - receiver=" + r.receiver
+ ", started " + (now - r.receiverTime) + "ms ago");
4. Slog.w(TAG, "Receiver during timeout of " + r + " : " + curReceiver);
5. EventLog.writeEvent(EventLogTags.AM_ANR, userId, pid, processName, info.flags,
annotation); 这个log的输出在某些情况是不会输出的,用户定义了ANR的处理。还有就是另外几种情况,isShuttingDown ,isNotResponding ,isCrashing ,killedByAm,killed
6 .生成 data/anr/文件,里面保存了一些pid的trace 信息以及cpu信息。
7.DropBoxManager也会输出一些anr的trace信息。
8 .如果是前台进程还会有一个AppNotRespondingDialog。