SWT 重啟案例分析（一）
和你一起終身學(xué)習(xí)，這里是程序員Android
經(jīng)典好文推薦，通過閱讀本文，您將收獲以下知識點:
一、MTK AEE Log分析工具
二、AEE Log分析流程
三、system_server 卡死案例分析及解決
本文主要針對?Exception Type: system_server_watchdog?,?system_server卡死找出的分析以及解決方案。
一、MTK AEE Log分析工具
MTK AEE Log?獲取方式：
關(guān)注微信公眾號：?程序員Android
回復(fù)?aee?即可獲取解析重啟db log的工具。
二、AEE Log分析流程
1. 使用AEE 工具解析?dbg?文件。
使用解析db.fatal.02.SWT.dbg
AEE Log 解析出來的文件
2.分析解析出來的exp_main等文件
exp_main?文件會記錄發(fā)生重啟時候的?log?打印信息。
部分重啟異常 Log信息如下：
$** *** *** *** *** *** *** *** Fatal *** *** *** *** *** *** *** **$
Build Info: 'alps-mp-o1.mp7:alps-mp-o1.mp7:mt6765:S01,ACE/AS0618/AS0618:8.1.0/O11019/1548123508:user/release-keys'
Flavor Info: 'None'
Exception Log Time:[Thu Mar 14 14:00:03 CST 2019] [38684.729626]
Exception Class: SWT
Exception Type: system_server_watchdog
Current Executing Process:
system_server
Trigger time:[2019-03-14 14:00:03.711844] pid:1029
Backtrace:
Process: system_server
Subject: Blocked in handler on ActivityManager (ActivityManager)
Build: ACE/AS0618/AS0618:8.1.0/O11019/1548123508:user/release-keys
3.exp_main 文件解析
從開頭的Log總體信息概覽，我們可以看到 發(fā)生重啟的時間、類型，觸發(fā)重啟的進程以及PID、系統(tǒng)Blocked 的地方。
結(jié)合exp_main以及?trace分析重啟 LogLog分析如下：
// 1.重啟觸發(fā)時間，以及PID
Trigger time:[2019-03-14 14:00:03.711844] pid:1029
// 2.Blocked 的進程
Backtrace:
Process: system_server
Subject: Blocked in handler on ActivityManager (ActivityManager)
// 3.根據(jù)PID 查看Trace信息
----- pid 1029 at 2019-03-14 13:59:58 -----
Cmd line: system_server
... ...
// 4.根據(jù)Backtrace 查看Blocked的信息
"ActivityManager" prio=5 tid=11 Blocked
... ...
// 5.tid=11 等待 tid=106的線程釋放資源鎖
- waiting to lock <0x090691f3> (a android.util.ArrayMap) held by thread 106
... ...
// 6.查看tid = 106 持鎖情況
"backup" prio=5 tid=106 Waiting
... ...
at java.lang.Object.wait(Native method)
- waiting on <0x06a44c62> (a com.android.server.am.ContentProviderRecord)
// 7.死鎖卡住的地方
at com.android.server.am.ActivityManagerService.getContentProviderImpl(ActivityManagerService.java:12127)
- locked <0x06a44c62> (a com.android.server.am.ContentProviderRecord)
... ...
"Binder:1029_8" prio=5 tid=107 Blocked
// 8.log中 tid=107 被 106 Blocked 進一步問題的加重
at com.android.server.notification.RankingHelper.getRecord(RankingHelper.java:258)
- waiting to lock <0x090691f3> (a android.util.ArrayMap) held by thread 106
$** *** *** *** *** *** *** *** Fatal *** *** *** *** *** *** *** **$
Log 分析大致過程截圖
完整 log 請在公眾號上獲取
三、system_server 卡死案例分析及解決
通過?Log?找到卡死原因后，解決此問題。
需要修改ActivityManagerService類。
1.修改代碼路徑如下：alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java
2.解決思路
避免?provider?長時間持鎖觸發(fā)MTK?60s?的SWT?重啟機制，設(shè)置超時時間，超過時間就要釋放資源鎖，避免發(fā)生此問題。
3.diff 修改方案如下：
--- a/[alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=d5e2e1411f3698a829e997d402c7482ec277fa8c;hb=d5e2e1411f3698a829e997d402c7482ec277fa8c)
+++ b/[alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=03208f78a2bf3167b4f0790019326e3939cc5444;hb=03208f78a2bf3167b4f0790019326e3939cc5444)
@@ [-545,7](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=d5e2e1411f3698a829e997d402c7482ec277fa8c;hb=d5e2e1411f3698a829e997d402c7482ec277fa8c#l545) [+545,9](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=03208f78a2bf3167b4f0790019326e3939cc5444;hb=03208f78a2bf3167b4f0790019326e3939cc5444#l545) @@ public class ActivityManagerService extends IActivityManager.Stub
// How long we wait for an attached process to publish its content providers
// before we decide it must be hung.
static final int CONTENT_PROVIDER_PUBLISH_TIMEOUT = 10*1000;
-
+ // How long we wait for provider to be notify before we decide it may be hung.
+ static final int CONTENT_PROVIDER_WAIT_TIMEOUT = 20*1000;
+
// How long we wait for a launched process to attach to the activity manager
// before we decide it's never going to come up for real, when the process was
// started with a wrapper for instrumentation (such as Valgrind) because it
@@ [-1745,6](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=d5e2e1411f3698a829e997d402c7482ec277fa8c;hb=d5e2e1411f3698a829e997d402c7482ec277fa8c#l1745) [+1747,7](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=03208f78a2bf3167b4f0790019326e3939cc5444;hb=03208f78a2bf3167b4f0790019326e3939cc5444#l1747) @@ public class ActivityManagerService extends IActivityManager.Stub
static final int PUSH_TEMP_WHITELIST_UI_MSG = 68;
static final int SERVICE_FOREGROUND_CRASH_MSG = 69;
static final int DISPATCH_OOM_ADJ_OBSERVER_MSG = 70;
+ static final int CONTENT_PROVIDER_WAIT_TIMEOUT_MSG = 71;
static final int START_USER_SWITCH_FG_MSG = 712;
static final int NOTIFY_VR_KEYGUARD_MSG = 74;
@@ [-2108,6](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=d5e2e1411f3698a829e997d402c7482ec277fa8c;hb=d5e2e1411f3698a829e997d402c7482ec277fa8c#l2108) [+2111,12](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=03208f78a2bf3167b4f0790019326e3939cc5444;hb=03208f78a2bf3167b4f0790019326e3939cc5444#l2111) @@ public class ActivityManagerService extends IActivityManager.Stub
synchronized (ActivityManagerService.this) {
mActivityStarter.doPendingActivityLaunchesLocked(true);
}
+ } break;
+ case CONTENT_PROVIDER_WAIT_TIMEOUT_MSG: {
+ ContentProviderRecord cpr = (ContentProviderRecord)msg.obj;
+ synchronized (ActivityManagerService.this) {
+ processContentProviderWaitTimedOutLocked(cpr);
+ }
} break;
case KILL_APPLICATION_MSG: {
synchronized (ActivityManagerService.this) {
@@ [-7029,7](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=d5e2e1411f3698a829e997d402c7482ec277fa8c;hb=d5e2e1411f3698a829e997d402c7482ec277fa8c#l7029) [+7038,31](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=03208f78a2bf3167b4f0790019326e3939cc5444;hb=03208f78a2bf3167b4f0790019326e3939cc5444#l7038) @@ public class ActivityManagerService extends IActivityManager.Stub
cleanupAppInLaunchingProvidersLocked(app, true);
removeProcessLocked(app, false, true, "timeout publishing content providers");
}
+
+ @GuardedBy("this")
+ private final void processContentProviderWaitTimedOutLocked(ContentProviderRecord cpr) {
+ try {
+ if (mLaunchingProviders.contains(cpr)) {
+ if (DEBUG_MU) Slog.v(TAG_MU,
+ "Remove from mLaunchingProviders, " + cpr
+ + " launchingApp=" + cpr.launchingApp);
+ mLaunchingProviders.remove(cpr);
+ }
+ if (DEBUG_MU) Slog.v(TAG_MU,
+ "RemoveMessages CONTENT_PROVIDER_WAIT_TIMEOUT_MSG, " + cpr
+ + " launchingApp=" + cpr.launchingApp);
+ mHandler.removeMessages(CONTENT_PROVIDER_WAIT_TIMEOUT_MSG, cpr);
+ synchronized (cpr) {
+ cpr.notifyAll();
+ cpr.launchingApp = null;
+ }
+ } catch (Exception e) {
+ if (DEBUG_MU) Slog.v(TAG_MU,
+ "processContentProviderWaitTimedOutLocked exception, " + e);
+ }
+ }
+
private final void processStartTimedOutLocked(ProcessRecord app) {
final int pid = app.pid;
boolean gone = false;
@@ [-12124,11](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=d5e2e1411f3698a829e997d402c7482ec277fa8c;hb=d5e2e1411f3698a829e997d402c7482ec277fa8c#l12124) [+12157,33](http://192.168.11.104/gitweb/?p=alps-mp-o1.mp1-V1.git;a=blob;f=alps/frameworks/base/services/core/java/com/android/server/am/ActivityManagerService.java;h=03208f78a2bf3167b4f0790019326e3939cc5444;hb=03208f78a2bf3167b4f0790019326e3939cc5444#l12157) @@ public class ActivityManagerService extends IActivityManager.Stub
if (conn != null) {
conn.waiting = true;
}
+ // add 20s wait timeout，avoid
+ if (!mHandler.hasMessages(CONTENT_PROVIDER_WAIT_TIMEOUT_MSG, cpr)) {
+ if (DEBUG_MU) Slog.v(TAG_MU,
+ "SendMessageDelayed CONTENT_PROVIDER_WAIT_TIMEOUT_MSG, " + cpr
+ + " launchingApp=" + cpr.launchingApp);
+ Message msg = mHandler.obtainMessage(CONTENT_PROVIDER_WAIT_TIMEOUT_MSG);
+ msg.obj = cpr;
+ mHandler.sendMessageDelayed(msg, CONTENT_PROVIDER_WAIT_TIMEOUT);
+ } else {
+ if (DEBUG_MU) Slog.v(TAG_MU,
+ "There is another waiting to start provider " + cpr
+ + " launchingApp=" + cpr.launchingApp
+ + ", not send CONTENT_PROVIDER_WAIT_TIMEOUT_MSG again");
+ }
+
cpr.wait();
} catch (InterruptedException ex) {
} finally {
if (conn != null) {
conn.waiting = false;
+ }
+ // remove wait time out message
+ if (mHandler.hasMessages(CONTENT_PROVIDER_WAIT_TIMEOUT_MSG, cpr)) {
+ if (DEBUG_MU) Slog.v(TAG_MU,
+ "After wait removeMessages CONTENT_PROVIDER_WAIT_TIMEOUT_MSG, "
+ + cpr + " launchingApp=" + cpr.launchingApp);
+ mHandler.removeMessages(CONTENT_PROVIDER_WAIT_TIMEOUT_MSG, cpr);
}
}
}
至此，本篇已結(jié)束。轉(zhuǎn)載網(wǎng)絡(luò)的文章，小編覺得很優(yōu)秀，歡迎點擊閱讀原文，支持原創(chuàng)作者，如有侵權(quán)，懇請聯(lián)系小編刪除。同時感謝您的閱讀，期待您的關(guān)注。
評論
圖片
表情