[Android6.0] 数据业务重试机制

其他 2018-07-27 05:09:54 阅读次数: 0

Android 6.0 Framework telephony中数据业务链接错误处理一般分3种情况：

1. SETUP_DATA_CALL 时返回错误

2. Modem上报DATA_CALL_LIST包含错误码或者链接中断

3. 一段时间内没有上下行数据（TX/RX）


下面具体来看每种情况的处理。

1. SETUP_DATA_CALL失败

DataConnection在收到SETUP_DATA_CALL结果后，用Message通知DcTracker处理：
  1. protected void onDataSetupComplete(AsyncResult ar) {
  2. if (ar.exception == null) {
  3. //链接成功
  4. } else{
  5. ...
  6. //标记permanent fail的次数，会影响后面onDataSetupCompleteError的判断
  7. if (isPermanentFail(cause)) apnContext.decWaitingApnsPermFailCount();
  8. apnContext.removeWaitingApn(apnContext.getApnSetting()); //从waiting列表中移除已经失败的APN
  9. onDataSetupCompleteError(ar);//继续处理错误
  10. ...
  11. }
  12. }

处理Error的逻辑：
1. 如果apnContext中的所有waiting APN都失败了，且不是每个都发生permanent fail(永久性错误)，则设置delay并重新发起这次连接
2. 如果apnContext中仍有没有尝试的waiting APN，则设置delay并尝试用下一个APN去连接
  1. /**
  2. * Error has occurred during the SETUP {aka bringUP} request and the DCT
  3. * should either try the next waiting APN or start over from the
  4. * beginning if the list is empty. Between each SETUP request there will
  5. * be a delay defined by {@link #getApnDelay()}.
  6. */
  7. @Override
  8. protected void onDataSetupCompleteError(AsyncResult ar) {
  9. String reason = "";
  10. ApnContext apnContext = getValidApnContext(ar, "onDataSetupCompleteError");
  13. if (apnContext == null) return;
  16. //已经尝试过所有APN
  17. if (apnContext.getWaitingApns().isEmpty()) {
  18. apnContext.setState(DctConstants.State.FAILED);//apnContext state设置成FAILED
  19. mPhone.notifyDataConnection(Phone.REASON_APN_FAILED, apnContext.getApnType());
  21. //清除DataConnection
  22. apnContext.setDataConnectionAc( null);
  24. //如果所有APN都发生Permanent fail，则不做重试
  25. if (apnContext.getWaitingApnsPermFailCount() == 0) {
  26. if (DBG) {
  27. log( "onDataSetupCompleteError: All APN's had permanent failures, stop retrying");
  28. }
  29. } else { //执行重试
  30. int delay = getApnDelay(Phone.REASON_APN_FAILED);
  31. if (DBG) {
  32. log( "onDataSetupCompleteError: Not all APN's had permanent failures delay="
  33. + delay);
  34. }
  35. startAlarmForRestartTrySetup(delay, apnContext);
  36. }
  37. } else { //waitingAPN中还有没有尝试的APN，继续尝试下一个
  38. if (DBG) log( "onDataSetupCompleteError: Try next APN");
  39. apnContext.setState(DctConstants.State.SCANNING);
  40. // Wait a bit before trying the next APN, so that
  41. // we're not tying up the RIL command channel
  42. startAlarmForReconnect(getApnDelay(Phone.REASON_APN_FAILED), apnContext); //试下一个APN
  44. }
  45. }

附：ApnContext的所有状态

  1. /**
  2. * IDLE: ready to start data connection setup, default state
  3. * CONNECTING: state of issued startPppd() but not finish yet
  4. * SCANNING: data connection fails with one apn but other apns are available
  5. * ready to start data connection on other apns (before INITING)
  6. * CONNECTED: IP connection is setup
  7. * DISCONNECTING: Connection.disconnect() has been called, but PDP
  8. * context is not yet deactivated
  9. * FAILED: data connection fail for all apns settings
  10. * RETRYING: data connection failed but we're going to retry.
  11. *
  12. * getDataConnectionState() maps State to DataState
  13. * FAILED or IDLE : DISCONNECTED
  14. * RETRYING or CONNECTING or SCANNING: CONNECTING
  15. * CONNECTED : CONNECTED or DISCONNECTING
  16. */
  17. public enum State {
  18. IDLE,
  19. CONNECTING,
  20. SCANNING,
  21. CONNECTED,
  22. DISCONNECTING,
  23. FAILED,
  24. RETRYING
  25. }


2. 链接中断

DcController监听RIL_UNSOL_DATA_CALL_LIST_CHANGED消息，获得每一个数据连接的更新：

  1. mPhone.mCi.registerForDataNetworkStateChanged(getHandler(),
  2. DataConnection.EVENT_DATA_STATE_CHANGED, null);

RIL上报DATA_CALL_LIST_CHANGED时会带上当前的Modem中的DataCall list，DcController将此dataCall list和上层的active list做对比：

1. 已经丢失 及 断开 的连接将会重试

2. 发生变化 和 发生永久错误的链接则需要清除

  1. private void onDataStateChanged(ArrayList<DataCallResponse> dcsList) {
  3. // Create hashmap of cid to DataCallResponse
  4. HashMap<Integer, DataCallResponse> dataCallResponseListByCid =
  5. new HashMap<Integer, DataCallResponse>();
  6. for (DataCallResponse dcs : dcsList) {
  7. dataCallResponseListByCid.put(dcs.cid, dcs);
  8. }
  10. //如果上报的dcsList中并没有找到对应的active的链接，则默认连接丢失并加入重试List
  11. ArrayList<DataConnection> dcsToRetry = new ArrayList<DataConnection>();
  12. for (DataConnection dc : mDcListActiveByCid.values()) {
  13. if (dataCallResponseListByCid.get(dc.mCid) == null) {
  14. if (DBG) log( "onDataStateChanged: add to retry dc=" + dc);
  15. dcsToRetry.add(dc);
  16. }
  17. }
  18. // Find which connections have changed state and send a notification or cleanup
  19. // and any that are in active need to be retried.
  20. ArrayList<ApnContext> apnsToCleanup = new ArrayList<ApnContext>();
  22. boolean isAnyDataCallDormant = false;
  23. boolean isAnyDataCallActive = false;
  25. for (DataCallResponse newState : dcsList) {
  27. DataConnection dc = mDcListActiveByCid.get(newState.cid);
  28. //不在Active MAP中的连接，表明这个连接还没同步到上层，会有其他地方处理。
  29. if (dc == null) {
  30. // UNSOL_DATA_CALL_LIST_CHANGED arrived before SETUP_DATA_CALL completed.
  31. loge( "onDataStateChanged: no associated DC yet, ignore");
  32. continue;
  33. }
  35. if (dc.mApnContexts.size() == 0) {
  36. if (DBG) loge( "onDataStateChanged: no connected apns, ignore");
  37. } else {
  38. // Determine if the connection/apnContext should be cleaned up
  39. // or just a notification should be sent out.
  40. if (newState.active == DATA_CONNECTION_ACTIVE_PH_LINK_INACTIVE) {
  41. //连接INACTIVE，按照错误类型区分处理
  42. DcFailCause failCause = DcFailCause.fromInt(newState.status);
  43. if (failCause.isRestartRadioFail()) {
  44. //恢复需要重启radio
  45. mDct.sendRestartRadio();
  46. } else if (mDct.isPermanentFail(failCause)) {
  47. //链接发生不可恢复的错误，需要Cleanup
  48. apnsToCleanup.addAll(dc.mApnContexts.keySet());
  49. } else {
  50. for (ApnContext apnContext : dc.mApnContexts.keySet()) {
  51. if (apnContext.isEnabled()) {
  52. //apn是enabled状态，重试
  53. dcsToRetry.add(dc);
  54. break;
  55. } else {
  56. //apn已经disabled，需要cleanup
  57. apnsToCleanup.add(apnContext);
  58. }
  59. }
  60. }
  62. } else {
  63. //LinkProperty发生变化
  64. UpdateLinkPropertyResult result = dc.updateLinkProperty(newState);
  65. if (result.oldLp.equals(result.newLp)) {
  66. if (DBG) log( "onDataStateChanged: no change");
  67. } else {
  68. //判断interface是否一致
  69. if (result.oldLp.isIdenticalInterfaceName(result.newLp)) {
  70. if (! result.oldLp.isIdenticalDnses(result.newLp) ||
  71. ! result.oldLp.isIdenticalRoutes(result.newLp) ||
  72. ! result.oldLp.isIdenticalHttpProxy(result.newLp) ||
  73. ! result.oldLp.isIdenticalAddresses(result.newLp)) {
  74. // If the same address type was removed and
  75. // added we need to cleanup
  76. CompareResult<LinkAddress> car =
  77. result.oldLp.compareAddresses(result.newLp);
  78. if (DBG) {
  79. log( "onDataStateChanged: oldLp=" + result.oldLp +
  80. " newLp=" + result.newLp + " car=" + car);
  81. }
  82. boolean needToClean = false;
  83. //如果address发生变化，需要清除这个old connection
  84. for (LinkAddress added : car.added) {
  85. for (LinkAddress removed : car.removed) {
  86. if (NetworkUtils.addressTypeMatches(
  87. removed.getAddress(),
  88. added.getAddress())) {
  89. needToClean = true;
  90. break;
  91. }
  92. }
  93. }
  94. if (needToClean) {
  96. apnsToCleanup.addAll(dc.mApnContexts.keySet());
  97. } else {
  98. if (DBG) log( "onDataStateChanged: simple change");
  99. //其他的LP变化，只做notify
  100. for (ApnContext apnContext : dc.mApnContexts.keySet()) {
  101. mPhone.notifyDataConnection(
  102. PhoneConstants.REASON_LINK_PROPERTIES_CHANGED,
  103. apnContext.getApnType());
  104. }
  105. }
  106. } else {
  107. if (DBG) {
  108. log( "onDataStateChanged: no changes");
  109. }
  110. }
  111. } else {
  112. //interface发生改变，cleanUp这个old connection
  113. apnsToCleanup.addAll(dc.mApnContexts.keySet());
  114. if (DBG) {
  115. log( "onDataStateChanged: interface change, cleanup apns="
  116. + dc.mApnContexts);
  117. }
  118. }
  119. }
  120. }
  121. }
  123. ...
  124. }
  126. ...
  128. //清除链接
  129. for (ApnContext apnContext : apnsToCleanup) {
  130. mDct.sendCleanUpConnection( true, apnContext);
  131. }
  133. //通知DataConnection链接丢失，需要发起重连
  134. for (DataConnection dc : dcsToRetry) {
  135. dc.sendMessage(DataConnection.EVENT_LOST_CONNECTION, dc.mTag);
  136. }
  137. }
  138. }

DataConnection ActiveState在收到LOST_CONNECTION消息后：

1. 如果重试次数没有达到上限，则设置定时重试，并切换到RetryingState

2. 如果不需要重试，则切换到Inactive状态，并可能通知DcTracker处理（onDataSetupCompleteError，可看第一种情况）

  1. case EVENT_LOST_CONNECTION: {
  2. if (DBG) {
  3. log( "DcActiveState EVENT_LOST_CONNECTION dc=" + DataConnection. this);
  4. }
  5. if (mRetryManager.isRetryNeeded()) {
  6. // We're going to retry
  7. int delayMillis = mRetryManager.getRetryTimer();
  8. //重试
  9. mDcRetryAlarmController.startRetryAlarm(EVENT_RETRY_CONNECTION, mTag,
  10. delayMillis);
  11. transitionTo(mRetryingState);
  12. } else {
  13. mInactiveState.setEnterNotificationParams(DcFailCause.LOST_CONNECTION);
  14. transitionTo(mInactiveState);
  15. }
  16. retVal = HANDLED;
  17. break;
  18. }


RetryingState 收到RETRY消息后，发起连接并切换到ActivatingState

  1. case EVENT_RETRY_CONNECTION: {
  2. if (msg.arg1 == mTag) {
  3. mRetryManager.increaseRetryCount(); //计数
  5. onConnect(mConnectionParams); //开始连接
  6. transitionTo(mActivatingState); //切换到Activating State
  7. } else {
  8. if (DBG) {
  9. log( "DcRetryingState stale EVENT_RETRY_CONNECTION"
  10. + " tag:" + msg.arg1 + " != mTag:" + mTag);
  11. }
  12. }
  13. retVal = HANDLED;
  14. break;
  15. }

RetryManager负责重试相关的计数：

  1. public boolean isRetryNeeded() {
  2. boolean retVal = mRetryForever || (mRetryCount < mCurMaxRetryCount);
  3. if (DBG) log( "isRetryNeeded: " + retVal);
  4. return retVal;
  5. }

3. 一段时间内持续没有接收到新的数据包

在Data完成连接后，DcTracker会定时检查TX/RX的更新，如果RX的值持续没有更新并超过设置的上限值，就会触发Recovery动作。



首先来看方法onDataStallAlarm，它由Alarm定时触发，执行这些操作：

更新TX/RX数据 -> 判断是否需要Recover并执行 -> 重新设置Alarm来触发下一次检查。

  1. protected void onDataStallAlarm(int tag) {
  2. if (mDataStallAlarmTag != tag) {
  3. if (DBG) {
  4. log( "onDataStallAlarm: ignore, tag=" + tag + " expecting " + mDataStallAlarmTag);
  5. }
  6. return;
  7. }
  8. //更新mSentSinceLastRecv
  9. updateDataStallInfo();
  11. //默认值是10
  12. int hangWatchdogTrigger = Settings.Global.getInt(mResolver,
  13. Settings.Global.PDP_WATCHDOG_TRIGGER_PACKET_COUNT,
  14. NUMBER_SENT_PACKETS_OF_HANG);
  16. boolean suspectedStall = DATA_STALL_NOT_SUSPECTED;
  17. if (mSentSinceLastRecv >= hangWatchdogTrigger) {
  18. //一段时间没有RX，且超过watchdog的值，需要recover
  19. suspectedStall = DATA_STALL_SUSPECTED;
  20. sendMessage(obtainMessage(DctConstants.EVENT_DO_RECOVERY));
  21. } else {
  22. if (VDBG_STALL) {
  23. log( "onDataStallAlarm: tag=" + tag + " Sent " + String.valueOf(mSentSinceLastRecv) +
  24. " pkts since last received, < watchdogTrigger=" + hangWatchdogTrigger);
  25. }
  26. }
  27. //重新设置Alarm任务，一段时间后再次执行本方法（onDataStallAlarm）
  28. startDataStallAlarm(suspectedStall);
  29. }


updateDataStallInfo()负责记数，处理分3种情况：

1. 有TX 也有RX  -> 正常，重置计数和Recovery action（Recovery action后面会写到）

2. 有TX没有RX -> 异常，累计TX数据

3. 没有TX 只有RX  -> 正常，重置计数和Recovery action

  1. private void updateDataStallInfo() {
  2. long sent, received;
  4. TxRxSum preTxRxSum = new TxRxSum(mDataStallTxRxSum);
  5. mDataStallTxRxSum.updateTxRxSum();
  7. sent = mDataStallTxRxSum.txPkts - preTxRxSum.txPkts;
  8. received = mDataStallTxRxSum.rxPkts - preTxRxSum.rxPkts;
  10. //收发正常，RecoveryAction重置
  11. if ( sent > 0 && received > 0 ) {
  12. if (VDBG_STALL) log( "updateDataStallInfo: IN/OUT");
  13. mSentSinceLastRecv = 0;
  14. putRecoveryAction(RecoveryAction.GET_DATA_CALL_LIST);
  15. } else if (sent > 0 && received == 0) {
  17. //没有RX；若不在通话状态则需要累计本次发送量
  18. if (isPhoneStateIdle()) {
  19. mSentSinceLastRecv += sent;
  20. } else {
  21. mSentSinceLastRecv = 0;
  22. }
  24. //没有发数据，RecoveryAction重置
  25. } else if (sent == 0 && received > 0) {
  26. if (VDBG_STALL) log( "updateDataStallInfo: IN");
  27. mSentSinceLastRecv = 0;
  28. putRecoveryAction(RecoveryAction.GET_DATA_CALL_LIST);
  29. } else {
  30. if (VDBG_STALL) log( "updateDataStallInfo: NONE");
  31. }
  32. }

TX/RX数据由TrafficStats提供的静态方法获得，是native层方法统计所有Mobile的iface后返回的数据：

  1. public void updateTxRxSum() {
  2. this.txPkts = TrafficStats.getMobileTcpTxPackets();
  3. this.rxPkts = TrafficStats.getMobileTcpRxPackets();
  4. }

最后看下doRecovery方法如何执行恢复数据。

doRecovery方法中有5种不同的Recovery action对应着各自的处理：
1. 向Modem主动查询DATA CALL LIST
2. 清除现有的数据链接
3. 重新驻网
4. 重启Radio
5. 深度重启Radio（根据高通的注释，这个操作涉及到RIL的设计）

如果一种方法执行之后，连接依然有问题，则执行下一种恢复方法，顺序类似于循环链表，直到恢复正常后updateDataStallInfo()将Action重置：

  1. protected void doRecovery() {
  2. if (getOverallState() == DctConstants.State.CONNECTED) {
  3. // Go through a series of recovery steps, each action transitions to the next action
  4. int recoveryAction = getRecoveryAction();
  5. switch (recoveryAction) {
  6. case RecoveryAction.GET_DATA_CALL_LIST:
  7. mPhone.mCi.getDataCallList(obtainMessage(DctConstants.EVENT_DATA_STATE_CHANGED));
  8. putRecoveryAction(RecoveryAction.CLEANUP);
  9. break;
  10. case RecoveryAction.CLEANUP:
  11. cleanUpAllConnections(Phone.REASON_PDP_RESET);
  12. putRecoveryAction(RecoveryAction.REREGISTER);
  13. break;
  14. case RecoveryAction.REREGISTER:
  15. mPhone.getServiceStateTracker().reRegisterNetwork( null);
  16. putRecoveryAction(RecoveryAction.RADIO_RESTART);
  17. break;
  18. case RecoveryAction.RADIO_RESTART:
  19. putRecoveryAction(RecoveryAction.RADIO_RESTART_WITH_PROP);
  20. restartRadio();
  21. break;
  22. case RecoveryAction.RADIO_RESTART_WITH_PROP:
  23. // This is in case radio restart has not recovered the data.
  24. // It will set an additional "gsm.radioreset" property to tell
  25. // RIL or system to take further action.
  26. // The implementation of hard reset recovery action is up to OEM product.
  27. // Once RADIO_RESET property is consumed, it is expected to set back
  28. // to false by RIL.
  29. EventLog.writeEvent(EventLogTags.DATA_STALL_RECOVERY_RADIO_RESTART_WITH_PROP, - 1);
  30. if (DBG) log( "restarting radio with gsm.radioreset to true");
  31. SystemProperties.set(RADIO_RESET_PROPERTY, "true");
  32. // give 1 sec so property change can be notified.
  33. try {
  34. Thread.sleep( 1000);
  35. } catch (InterruptedException e) {}
  36. restartRadio();
  37. putRecoveryAction(RecoveryAction.GET_DATA_CALL_LIST);
  38. break;
  39. default:
  40. throw new RuntimeException( "doRecovery: Invalid recoveryAction=" +
  41. recoveryAction);
  42. }
  43. mSentSinceLastRecv = 0;
  44. }
  45. }

猜你喜欢

转载自blog.csdn.net/zhgeliang/article/details/81034684
今日推荐
TIOBE 5 月榜单：Fortran “复活”进入 Top 10
GCC 14.1 发布
面壁智能发布 Eurux-8x22B 开源大模型 —— 堪称「理科状元」
开源日报 | 谷歌扶持鸿蒙上位；开源Rabbit R1；Docker加持的安卓手机；微软的焦虑和野心；海尔电器把开放平台关了
中国码农的“35岁魔咒”
蘭雅 CorelDRAW 插件 2024.5.1 国际劳动节版，免费下载
Arc Browser for Windows 1.0 正式 GA
90后程序员开发视频搬运软件、不到一年获利超 700 万，结局很刑！
周排行
基本数据类型封装类比较 Java源码解读(一) 8种基本类型对应的封装类型
JS实现无缝滚动上
深入解析HashMap原理（基于JDK1.8）
mysql的连接池
关于.htc
linux下的ubuntu12.04图形界面
【数论】好推不好记的扩展欧几里德
设备树详解
cscope + tags 简单设置
xml学习
每日归档
更多
2024-05-09(35)
2024-05-08(42)
2024-05-07(14)
2024-05-06(40)
2024-05-05(0)
2024-05-04(7)
2024-05-03(19)
2024-05-02(0)
2024-05-01(4)
2024-04-30(1)

代码天地 © 2018 codetd.com

境外服务器   最新电视剧2022