ANR分析流程
一、ANR基本知识
1.1、发生原因
一句话总结:没有在规定的时间内,干完要干的事情,就会发生ANR。
1.2、ANR分类
从发生的场景分类:
· Input事件超过5s没有被处理完
· Service处理超时,前台20s,后台200s
· BroadcastReceiver处理超时,前台10S,后台60s
· ContentProvider执行超时,比较少见
从发生的原因分:
· 主线程有耗时操作,如有复杂的layout布局,IO操作等。
· 被Binder对端block
· 被子线程同步锁block
· Binder被占满导致主线程无法和SystemServer通信
· 得不到系统资源(CPU/RAM/IO)
从进程的角度分:
· 问题出在当前进程:
主线程本身耗时, 或则主线程的消息队列存在耗时操作;
主线程被本进程的其他子线程所blocked;
· 问题出在远端进程(一般是binder call或socket等通信方式)
二、ANR的Log分析
2.1、Log分类
发生了ANR问题,通常会抓一份bugreport
adb bugreprot xxx
最为重要的是,生成的bugreport有anr的trace,如果要单独拿出来也行
adb pull /data/anr/traces.txt xxx
一份完整的bugreport包含下面的信息,对分析ANR问题很关键
Log名称 | 作用 | 获取命令 |
---|---|---|
system.log | 包含ANR发生时间点信息、ANR发生前的CPU信息,还包含大量系统服务输出的信息 | adb logcat –b system |
main.log | 包含ANR发生前应用自身输出的信息,可供分析应用是否有异常;此外还包含输出的GC信息,可供分析内存回收的速度,判断系统是否处于低内存或内存碎片化状态 | adb logcat –b main |
event.log | 包含AMS与WMS输出的应用程序声明周期信息,可供分析窗口创建速度以及焦点转换情况 | adb logcat –b event |
kernel.log | 包含kernel打出的信息,LowMemoryKiller杀进程、内存碎片化或内存不足,mmc驱动异常都可以在这里找到。 | 无 |
那么这些Log怎么看呢?看下面案例一
2.2、案例一:sp耗时问题导致应用ANR
一般先搜索ANR in获取最直观的信息,如下:
06-16 16:16:28.590 1853 2073 E ActivityManager: ANR in com.android.camera (com.android.camera/.Camera)06-16 16:16:28.590 1853 2073 E ActivityManager: PID: 2766106-16 16:16:28.590 1853 2073 E ActivityManager: Reason: Input dispatching timed out (com.android.camera/com.android.camera.Camera, Waiting to send non-key event because the touched window has not finished processing certain input events that were delivered to it over 500.0ms ago. Wait queue length: 24. Wait queue head age: 5511.1ms.)06-16 16:16:28.590 1853 2073 E ActivityManager: Load: 16.25 / 29.48 / 38.3306-16 16:16:28.590 1853 2073 E ActivityManager: CPU usage from 0ms to 8058ms later:06-16 16:16:28.590 1853 2073 E ActivityManager: 58% 291/mediaserver: 51% user + 6.7% kernel / faults: 2457 minor 4 major06-16 16:16:28.590 1853 2073 E ActivityManager: 27% 317/mm-qcamera-daemon: 21% user + 5.8% kernel / faults: 15965 minor06-16 16:16:28.590 1853 2073 E ActivityManager: 0.4% 288/debuggerd: 0% user + 0.3% kernel / faults: 21615 minor 87 major06-16 16:16:28.590 1853 2073 E ActivityManager: 17% 27661/com.android.camera: 10% user + 6.8% kernel / faults: 2412 minor 34 major06-16 16:16:28.590 1853 2073 E ActivityManager: 16% 1853/system_server: 10% user + 6.4% kernel / faults: 1754 minor 87 major06-16 16:16:28.590 1853 2073 E ActivityManager: 10% 539/sensors.qcom: 7.8% user + 2.6% kernel / faults: 16 minor06-16 16:16:28.590 1853 2073 E ActivityManager: 4.4% 277/surfaceflinger: 1.8% user + 2.6% kernel / faults: 14 minor06-16 16:16:28.590 1853 2073 E ActivityManager: 4% 203/mmcqd/0: 0% user + 4% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 2.6% 3510/com.android.phone: 1.9% user + 0.6% kernel / faults: 1148 minor 8 major06-16 16:16:28.590 1853 2073 E ActivityManager: 2.1% 2902/com.android.systemui: 1.6% user + 0.4% kernel / faults: 1272 minor 32 major06-16 16:16:28.590 1853 2073 E ActivityManager: 1.6% 3110/com.miui.whetstone: 1.6% user + 0% kernel / faults: 2614 minor 22 major06-16 16:16:28.590 1853 2073 E ActivityManager: 0.8% 99/kswapd0: 0% user + 0.8% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 1.4% 217/jbd2/mmcblk0p25: 0% user + 1.4% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 1.4% 223/logd: 0.7% user + 0.7% kernel / faults: 4 minor06-16 16:16:28.590 1853 2073 E ActivityManager: 0.9% 12808/kworker/0:1: 0% user + 0.9% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.8% 35/kworker/u:2: 0% user + 0.8% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0% 3222/com.miui.sysbase: 0% user + 0% kernel / faults: 1314 minor 12 major06-16 16:16:28.590 1853 2073 E ActivityManager: 0.8% 3446/com.android.nfc: 0.4% user + 0.3% kernel / faults: 1223 minor 9 major06-16 16:16:28.590 1853 2073 E ActivityManager: 0.7% 10866/kworker/u:1: 0% user + 0.7% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.6% 642/mdss_fb0: 0% user + 0.6% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.6% 29336/kworker/u:7: 0% user + 0.6% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.4% 6/kworker/u:0: 0% user + 0.4% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.4% 22924/kworker/u:6: 0% user + 0.4% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.3% 4421/mpdecision: 0% user + 0.3% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.2% 276/servicemanager: 0.1% user + 0.1% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.2% 289/rild: 0.2% user + 0% kernel / faults: 20 minor06-16 16:16:28.590 1853 2073 E ActivityManager: 0.1% 4161/mcd: 0% user + 0% kernel / faults: 9 minor 1 major06-16 16:16:28.590 1853 2073 E ActivityManager: 0.1% 3/ksoftirqd/0: 0% user + 0.1% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.1% 5/kworker/0:0H: 0% user + 0.1% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.1% 7/kworker/u:0H: 0% user + 0.1% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0% 215/flush-179:0: 0% user + 0% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.1% 321/displayfeature: 0.1% user + 0% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.1% 368/irq/33-cpubw_hw: 0% user + 0.1% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 0.1% 403/qmuxd: 0% user + 0.1% kernel / faults: 60 minor06-16 16:16:28.590 1853 2073 E ActivityManager: 0% 3491/com.xiaomi.finddevice: 0% user + 0% kernel / faults: 706 minor06-16 16:16:28.590 1853 2073 E ActivityManager: 0.1% 29330/ksoftirqd/1: 0% user + 0.1% kernel06-16 16:16:28.590 1853 2073 E ActivityManager: 96% TOTAL: 56% user + 29% kernel + 6.3% iowait + 4.1% softirq
遇到ANR问题,摆在我们面前的trace是不是第一案发现场,如果ANR发生的输出的信息很多,当时的CPU和I/O资源比较紧张,那么这段日志输出的时间点可能会延迟10秒到20秒都有可能,所以我们有时候需要提高警惕,现在举例逐行解读一下:
06-16 16:16:28.590 1853 2073 E ActivityManager: ANR in com.android.camera (com.android.camera/.Camera) .
这一行得知ANR发生的时间是06-16 16:16:28.590,发生的进程是com.android.camera ,具体在com.android.camera/.Camera,其中1853是systemserver的pid,2073是ActivityManager线程的pid,ActivityManager是一个系统线程。其实在Events log中也有对应的信息,搜索关键字am_anr
06-16 16:16:20.536 1853 2073 I am_anr : [0,27661,com.android.camera,952745541,Input dispatching timed out (com.android.camera/com.android.camera.Camera, Waiting to send non-key event because the touched window has not finished processing certain input events that were delivered to it over 500.0ms ago. Wait queue length: 24. Wait queue head age: 5511.1ms.)]
由此同样可以确定ANR的时间点,类型,进程pid,进程名称等,继续看下一行
06-16 16:16:28.590 1853 2073 E ActivityManager: PID: 27661
这一行得知ANR进程的pid是27661,特殊情况,如果pid为0,说明在发生ANR之前,这个进程就被LowMemoryKiller杀死了或者出现了Crash,这种情况下,是无法接收到系统的广播或者按键消息的,故出现ANR
06-16 16:16:28.590 1853 2073 E ActivityManager: Reason: Input dispatching timed out (com.android.camera/com.android.camera.Camera, Waiting to send non-key event because the touched window has not finished processing certain input events that were delivered to it over 500.0ms ago. Wait queue length: 24. Wait queue head age: 5511.1ms.)
这一行得知ANR发生的原因是Input dispatching timed out
06-16 16:16:28.590 1853 2073 E ActivityManager: Load: 16.25 / 29.48 / 38.33
这行得知Cpu的负载,在Linux操作系统上,输入uptime也能得到一段时间的负载。
uptime 20:09:54 up 71 days, 10:48, 1 user, load average: 0.99, 0.78, 0.86
那么负载是什么意思呢?Load后面的三个数字的意思分别是1分钟、5分钟、15分钟内系统的平均负荷。当CPU完全空闲的时候,平均负荷为0;当CPU工作量饱和的时候,平均负荷为1,通过Load可以判断系统负荷是否过重。有一个形象的比喻:个CPU想象成一座大桥,桥上只有一根车道,所有车辆都必须从这根车道上通过,系统负荷为0,意味着大桥上一辆车也没有,系统负荷为0.5,意味着大桥一半的路段有车,系统负荷为1.0,意味着大桥的所有路段都有车,也就是说大桥已经"满"了,系统负荷为2.0,意味着车辆太多了,大桥已经被占满了(100%),后面等着上桥的车辆还有一倍。大桥的通行能力,就是CPU的最大工作量;桥梁上的车辆,就是一个个等待CPU处理的进程(process)。
经验法则是这样的:
当系统负荷持续大于0.7,你必须开始调查了,问题出在哪里,防止情况恶化。
当系统负荷持续大于1.0,你必须动手寻找解决办法,把这个值降下来。
当系统负荷达到5.0,就表明你的系统有很严重的问题
如果只有1分钟的系统负荷大于1.0,其他两个时间段都小于1.0,这表明只是暂时现象,问题不大。
如果15分钟内,平均系统负荷大于1.0(调整CPU核心数之后),表明问题持续存在,不是暂时现象。所以,你应该主要观察"15分钟系统负荷",将它作为电脑正常运行的指标。
而我们现在的手机是多核CPU架构,八核的多的是,意味着Cpu处理的能力就乘以了8,每个核运行的时间可以从下面的文件中得到,/sys/devices/system/cpu/cpu%d/cpufreq/stats/time_in_state 中读取的,%d代表是CPU的核。文件中记录了 CPU 从开机到读取文件时,在各个频率下的运行时间,单位:10 mS。
使用adb shell cat /sys/devices/system/cpu/cpu1/cpufreq/stats/time_in_state查看 频度
使用adb shell cat /sys/devices/system/cpu/cpu1/cpufreq/stats/time_in_state查看
频度 时间
652800 1813593
1036800 46484
1401600 521974
1689600 2956667
1843200 83065
1958400 53516
2016000 251693
关于负荷详细的可以看理解Linux系统负荷,不过多扩展。
06-16 16:16:28.590 1853 2073 E ActivityManager: CPU usage from 0ms to 8058ms later:06-16 16:16:28.590 1853 2073 E ActivityManager: 58% 291/mediaserver: 51% user + 6.7% kernel / faults: 2457 minor 4 major06-16 16:16:28.590 1853 2073 E ActivityManager: 27% 317/mm-qcamera-daemon: 21% user + 5.8% kernel / faults: 15965 minor06-16 16:16:28.590 1853 2073 E ActivityManager: 0.4% 288/debuggerd: 0% user + 0.3% kernel / faults: 21615 minor 87 major06-16 16:16:28.590 1853 2073 E ActivityManager: 17% 27661/com.android.camera: 10% user + 6.8% kernel / faults: 2412 minor 34 major....06-16 16:16:28.590 1853 2073 E ActivityManager: 96% TOTAL: 56% user + 29% kernel + 6.3% iowait + 4.1% softirq.....
这一段日志可以得到ANR发生的时候,Top进程的Cpu占用情况,user代表是用户空间,kernel是内核空间,一般的有如下的规律。
· kswapd0 cpu占用率偏高,系统整体运行会缓慢,从而引起各种ANR。把问题转给"内存优化",请他们进行优化。
· logd CPU占用率偏高,也会引起系统卡顿和ANR,因为各个进程输出LOG的操作被阻塞从而执行的极为缓慢。
· Vold占用CPU过高,会引起系统卡顿和ANR,请负责存储的同学先调查
· qcom.sensor CPU占用率过高,会引起卡顿,请系统同学调查
· 应用自身CPU占用率较高,高概率应用自身问题
· 系统CPU占用率不高,但主线程在等待一个锁,高概率应用自身问题
· 应用处于D状态,发生ANR,如果最后的操作是refriger,那么是应用被冻结了,正常情况下是功耗优化引起的。
好了,通过上面的日志我们得到了ANR的基本信息,要得到阻塞的地方,还要靠trace文件。一般都在anr目录下。在这个trace文件中搜索主线程的堆栈,如下:
----- pid 27661 at 2017-06-16 16:16:20 -----Cmd line: com.android.camera"main" prio=5 tid=1 Waiting| group="main" sCount=1 dsCount=0 obj=0x75a4b5c8 self=0xb4cf6500| sysTid=27661 nice=-10 cgrp=default sched=0/0 handle=0xb6f6cb34| state=S schedstat=( 11242036155 8689191757 38520 ) utm=895 stm=229 core=0 HZ=100| stack=0xbe4ea000-0xbe4ec000 stackSize=8MB| held mutexes=at java.lang.Object.wait!(Native method)\- waiting on <0x09e6a059> (a java.lang.Object)at java.lang.Thread.parkFor$(Thread.java:1220)\- locked <0x09e6a059> (a java.lang.Object)at sun.misc.Unsafe.park(Unsafe.java:299)at java.util.concurrent.locks.LockSupport.park(LockSupport.java:158)at java.util.concurrent.locks.AbstractQueuedSynchronizer.parkAndCheckInterrupt(AbstractQueuedSynchronizer.java:810)at java.util.concurrent.locks.AbstractQueuedSynchronizer.doAcquireSharedInterruptibly(AbstractQueuedSynchronizer.java:970)at java.util.concurrent.locks.AbstractQueuedSynchronizer.acquireSharedInterruptibly(AbstractQueuedSynchronizer.java:1278)at java.util.concurrent.CountDownLatch.await(CountDownLatch.java:203)at android.app.SharedPreferencesImpl$EditorImpl$1.run(SharedPreferencesImpl.java:366)at android.app.QueuedWork.waitToFinish(QueuedWork.java:88)at android.app.ActivityThread.handleStopActivity(ActivityThread.java:3605)at android.app.ActivityThread.access$1300(ActivityThread.java:153)at android.app.ActivityThread$H.handleMessage(ActivityThread.java:1399)at android.os.Handler.dispatchMessage(Handler.java:102)at android.os.Looper.loop(Looper.java:154)at android.app.ActivityThread.main(ActivityThread.java:5528)at java.lang.reflect.Method.invoke!(Native method)at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:740)at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:630)
解读一下部分字段的含义
字段 | 含义 |
---|---|
tid=1 | 线程号 |
sysTid=27661 | 主线程的线程号和进程号相同 |
Waiting | 线程状态,其中state也是线程状态,如果state=D代表底层被blocked了。 |
nice | nice值越小,则优先级越高。因为是主线程此处nice=-10, 可以看到优先级很高了 |
schedstat | 括号中的3个数字,依次是Running, Runable, Switch时间。Running时间:CPU运行的时间,单位ns。 Runable时间:RQ队列的等待时间,单位ns。 Switch次数:CPU调度切换次数 |
utm | 该线程在用户态所执行的时间,单位是jiffies |
stm | 该线程在内核态所执行的时间,单位是jiffies |
sCount | 此线程被挂起的次数 |
dsCount | 线程被调试器挂起的次数,当一个进程被调试后,sCount会重置为0,调试完毕后sCount会根据是否被正常挂起增长,但是dsCount不会被重置为0,所以dsCount也可以用来判断这个线程是否被调试过 |
self | 线程本身的地址 |
在说一下线程的状态
状态 | 值 | 说明 |
---|---|---|
THREAD_ZOMBIE | 0 | TERMINATED |
THREAD_RUNNING | 1 | RUNNABLE or running now |
THREAD_TIMED_WAIT | 2 | TIMED_WAITING in Object.wait() |
THREAD_MONITOR | 3 | BLOCKED on a monitor |
THREAD_INITIALIZING | 5 | allocated not yet running |
THREAD_STARTING | 6 | started not yet on thread list |
THREAD_NATIVE | 7 | off in a JNI native method |
THREAD_VMWAIT | 8 | waiting on a VM resource |
THREAD_SUSPENDED | 9 | suspended usually by GC or debugger |
那么这个问题怎么搞呢,通过上面的基础介绍与trace文件,我们知道,blocked点是
at java.util.concurrent.CountDownLatch.await(CountDownLatch.java:203)at android.app.SharedPreferencesImpl$EditorImpl$1.run(SharedPreferencesImpl.java:366)at android.app.QueuedWork.waitToFinish(QueuedWork.java:88)at android.app.ActivityThread.handleStopActivity(ActivityThread.java:3605)at android.app.ActivityThread.access$1300(ActivityThread.java:153)先来看QueuedWork.waitToFinish77 /**78 * Finishes or waits for async operations to complete.79 * (e.g. SharedPreferences$Editor#startCommit writes)80 *81 * Is called from the Activity base class's onPause(), after82 * BroadcastReceiver's onReceive, after Service command handling,83 * etc. (so async work is never lost)84 */85 public static void waitToFinish() {86 Runnable toFinish;//等待所有等待完成的任务完成87 while ((toFinish = sPendingWorkFinishers.poll()) != null) {88 toFinish.run();89 }90 }
QueuedWork.waitToFinish会在Activity的onPause或者BroadcastReceiver的onReceive之后被调用,目的是确保异步任务执行完成.在waitToFinish中遍历sPendingWorkFinishers所有等待完成的任务,并等待他们的完成。在来看SharedPreferencesImpl.apply,这个方法里面会将等待写入到文件系统的任务放到QueuedWork的等待完成队列里
361 public void apply() {362 final MemoryCommitResult mcr = commitToMemory();363 final Runnable awaitCommit = new Runnable() {364 public void run() {365 try {366 mcr.writtenToDiskLatch.await();367 } catch (InterruptedException ignored) {368 }369 }370 };371//将等待写入到文件系统的任务放到QueuedWork的等待队列中372 QueuedWork.add(awaitCommit);373 ... ... ... ... ... ...388 }
虽然apply方法本身可以很快返回,但是当Activity的onPause被调用时,会等待写入到文件系统的任务完成.也就是说,虽然apply本身不会阻塞调用线程,但是会将等待时间转嫁到主线程.因此,如果写入任务执行比较慢,activity, service, broadcast在生命周期结束时, sp操作没有完成,就会阻塞主线程造成ANR。分析到这里,明显是个系统问题了,而App也是无能为力,好在小米手机上已经缓解了这个问题,方案不透露了。一般的看trace有如下规律:
· 发生ANR时,trace中找不到相应进程,检查一下Android Runtime是否因为应用的崩溃给ShutDown了,如果ShutDown了,此时要去查ShutDown的原因。
· 应用发生ANR,如果主线程正在执行getContentProvider,那么它正在请求另一个应用的ContentProvider,此时要查一下目标ContentProvider的宿主进程,看看正在做什么
· 主线程执行数据库操作或网络请求,应该是应用自身问题
· 主线程等待其他线程持有的锁,而目标线程执行数据库操作或网络请求,那么是应用自身问题。
这里只是先搞一个案例熟悉一下ANR分析基本流程。到这里总结一下上面的套路:
· 抓取bugreport,搜索ANR in,查看发生的时间和进程
· 根据进程寻找主线程的trace,发现被blocked的地方
· 结合源码进行分析解决
当然通过这两个步骤就定位到ANR发生的原因,说明我们的运气比较好,然而大多数时候不是这个样子的。
上面我们分析了一个系统问题导致的ANR,这里你可能会想,我的app什么活都没有干,竟然发生ANR了,以后可以把锅直接甩给系统了,非也,具体问题还是要具体分析,怀疑系统,我们要有证据,证据从哪里来,还是从Log里面来,继续看第三节,系统耗时分析方案
三、系统耗时分析方案
系统做了一些耗时分析的操作,在一些手机厂商中,Log里面还有其他的加强,这里列举比较通用的一些
3.1、binder_sample
· A.功能说明: 监控每个进程的主线程的binder transaction的耗时情况, 当超过阈值时,则输出相应的目标调用信息,默认1000ms打开。
· B.log格式: 52004 binder_sample (descriptor|3),(method_num|1|5),(time|1|3),(blocking_package|3),(sample_percent|1|6)
· C.log实例:
2754 2754 I binder_sample: [android.app.IActivityManager,35,2900,android.process.media,5]
从上面的log中可以得出
1.主线程2754;
2.执行android.app.IActivityManager接口
\1. 所对应方法code =35(即STOP_SERVICE_TRANSACTION),
\2. 所花费时间为2900ms.
\3. 该block所在package为 android.process.media,最后一个参数是sample比例(没有太大价值)
3.2、dvm_lock_sample
· A.功能说明: 当某个线程等待lock的时间blocked超过阈值,则输出当前的持锁状态 ;
· B.log格式: 20003 dvm_lock_sample (process|3),(main|1|5),(thread|3),(time|1|3),(file|3),(line|1|5),(ownerfile|3),(ownerline|1|5),(sample_percent|1|6)
· C.log实例:
dvm_lock_sample: [system_server,1,Binder_9,1500,ActivityManagerService.java,6403,-,1448,0]
意思是system_server: Binder_9,执行到ActivityManagerService.java的6403行代码,一直在等待AMS锁, 而该锁所同一文件的1448行代码所持有, 从而导致Binder_9线程被阻塞1500ms.
3.3、 binder starved
· A.功能说明: 当system_server等进程的线程池使用完, 无空闲线程时, 则binder通信都处于饥饿状态, 则饥饿状态超过一定阈值则输出信息;
· B.云控参数: persist.sys.binder.starvation (默认值16ms)
· C.log实例:
1232 1232 “binder thread pool (16 threads) starved for 100 ms”
· D.log解析: system_server进程的 线程池已满的持续长达100ms
一般有了这些信息之后,可以辅助我们确定问题原因归属是系统原因还是App原因,看下面的案例二:
3.4、案例二:疯狂Binder Call导致应用ANR
搜索ANR in
08-28 18:54:00.110 1000 1825 1848 E ActivityManager: ANR in com.jeejen.family (com.jeejen.family/com.jeejen.home.launcher.ShoppingActivity)08-28 18:54:00.110 1000 1825 1848 E ActivityManager: PID: 2057608-28 18:54:00.110 1000 1825 1848 E ActivityManager: Reason: Input dispatching timed out (com.jeejen.family/com.jeejen.home.launcher.WelcomeActivity, Waiting to send non-key event because the touched window has not finished processing certain input events that were delivered to it over 500.0ms ago. Wait queue length: 2. Wait queue head age: 10064.4ms.)08-28 18:54:00.110 1000 1825 1848 E ActivityManager: Parent: com.jeejen.family/com.jeejen.home.launcher.WelcomeActivity08-28 18:54:00.110 1000 1825 1848 E ActivityManager: Load: 1.25 / 1.1 / 1.3708-28 18:54:00.110 1000 1825 1848 E ActivityManager: CPU usage from 5166ms to 0ms ago (2018-08-28 18:53:51.270 to 2018-08-28 18:53:56.436):08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 7.7% 1825/system_server: 5.6% user + 2.1% kernel / faults: 1329 minor08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 3.6% 20683/com.jeejen.family:pushcenter_pushservice: 3% user + 0.5% kernel / faults: 542 minor08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 2.7% 4114/cnss_diag: 1.9% user + 0.7% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 2.1% 422/kworker/u16:7: 0% user + 2.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 1.9% 20830/com.jeejen.family:store: 1.3% user + 0.5% kernel / faults: 199 minor08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 1.7% 20608/com.jeejen.family:pushcenter: 1.1% user + 0.5% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 1.5% 725/android.hardware.sensors@1.0-service: 0.7% user + 0.7% kernel / faults: 1 minor08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.9% 3538/com.android.systemui: 0.7% user + 0.1% kernel / faults: 11 minor08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.5% 241/crtc_commit:111: 0% user + 0.5% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.5% 419/kworker/u16:4: 0% user + 0.5% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.5% 786/surfaceflinger: 0.5% user + 0% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.3% 185/IPCRTR_dsps_sme: 0% user + 0.3% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.3% 730/android.hardware.wifi@1.0-service: 0.1% user + 0.1% kernel / faults: 28 minor08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.3% 820/dsps_IPCRTR: 0% user + 0.3% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.3% 1147/msm_irqbalance: 0.1% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.3% 4113/sugov:0: 0% user + 0.3% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 10/rcuop/0: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 18/ksoftirqd/1: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0% 34/ksoftirqd/3: 0% user + 0% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0% 53/rcuop/5: 0% user + 0% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0% 61/rcuop/6: 0% user + 0% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 242/crtc_event:111: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 538/ueventd: 0.1% user + 0% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 577/jbd2/sda22-8: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 591/logd: 0.1% user + 0% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 719/android.hardware.graphics.composer@2.1-service: 0.1% user + 0% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 928/thermal-engine: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 3490/cds_mc_thread: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 3491/cds_ol_rx_threa: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 3680/com.android.phone: 0% user + 0.1% kernel / faults: 16 minor08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 4248/com.miui.daemon: 0.1% user + 0% kernel / faults: 4 minor08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 4488/com.miui.powerkeeper: 0.1% user + 0% kernel / faults: 10 minor08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 5545/com.lbe.security.miui: 0% user + 0.1% kernel / faults: 6 minor08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 6490/kworker/u17:2: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 7535/kworker/u16:15: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0% 7723/kworker/3:5: 0% user + 0% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 15111/kworker/1:0: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 15138/kworker/3:0: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0% 19857/kworker/0:3: 0% user + 0% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 0.1% 20492/kworker/5:3: 0% user + 0.1% kernel08-28 18:54:00.110 1000 1825 1848 E ActivityManager: 3.8% TOTAL: 2% user + 1.1% kernel + 0% iowait + 0.3% irq + 0.1% softirq
按照上面的套路看看,各方面比较正常,发生时间大概是08-28 18:54:00.110 ,在看主线程的trace。
----- pid 20576 at 2018-08-28 18:53:56 -----Cmd line: com.jeejen.family"main" prio=5 tid=1 Native| group="main" sCount=1 dsCount=0 flags=1 obj=0x77ffca18 self=0xecfce000| sysTid=20576 nice=-10 cgrp=default sched=0/0 handle=0xf0bf2494| state=S schedstat=( 628294395 402363898 957 ) utm=42 stm=20 core=4 HZ=100| stack=0xff5fe000-0xff600000 stackSize=8MB| held mutexes=kernel: (couldn't read /proc/self/task/20576/stack)native: #00 pc 00053cfc /system/lib/libc.so (__ioctl+8)native: #01 pc 00021cd3 /system/lib/libc.so (ioctl+30)native: #02 pc 0003d3f5 /system/lib/libbinder.so (android::IPCThreadState::talkWithDriver(bool)+204)native: #03 pc 0003dde3 /system/lib/libbinder.so (android::IPCThreadState::waitForResponse(android::Parcel*, int*)+26)native: #04 pc 0003713d /system/lib/libbinder.so (android::BpBinder::transact(unsigned int, android::Parcel const&, android::Parcel*, unsigned int)+36)native: #05 pc 000c3cf1 /system/lib/libandroid_runtime.so (android_os_BinderProxy_transact(_JNIEnv*, _jobject*, int, _jobject*, _jobject*, int)+200)at android.os.BinderProxy.transactNative(Native method)at android.os.BinderProxy.transact(Binder.java:1127)at android.net.wifi.IWifiManager$Stub$Proxy.getConnectionInfo(IWifiManager.java:1441)at android.net.wifi.WifiManager.getConnectionInfo(WifiManager.java:1778)at org.chromium.net.NetworkChangeNotifierAutoDetect$WifiManagerDelegate.getWifiInfoLocked(NetworkChangeNotifierAutoDetect.java:28)at org.chromium.net.NetworkChangeNotifierAutoDetect$WifiManagerDelegate.getWifiSsid(NetworkChangeNotifierAutoDetect.java:22)\- locked <0x0f4edae7> (a java.lang.Object)at org.chromium.net.NetworkChangeNotifierAutoDetect.getCurrentNetworkState(NetworkChangeNotifierAutoDetect.java:67)at org.chromium.net.NetworkChangeNotifierAutoDetect.<init>(NetworkChangeNotifierAutoDetect.java:21)at org.chromium.net.NetworkChangeNotifier.setAutoDetectConnectivityStateInternal(NetworkChangeNotifier.java:61)
看样子是binder call阻塞了,调用的接口是IWifiManager.getConnectionInfo()。因为是binder调用,查看一下binder_sample。
08-28 18:54:01.384 10171 20576 20576 I binder_sample: [android.net.wifi.IWifiManager,24,16004,com.jeejen.family,100]08-28 18:54:04.868 10171 20576 20576 I binder_sample: [android.net.wifi.IWifiManager,24,3479,com.jeejen.family,100]08-28 18:56:12.712 10171 21885 21885 I binder_sample: [android.net.wifi.IWifiManager,24,8963,com.jeejen.family,100]
可以看到在ANR附近的时间,使用IWifiManager接口的binder调用确实耗时比较长,那么这个是不是系统原因呢?那么去看他的对端Sysytem的代码。
1763 /**1764 * See {@link android.net.wifi.WifiManager#getConnectionInfo()}1765 * @return the Wi-Fi information, contained in {@link WifiInfo}.1766 */1767 @Override1768 public WifiInfo getConnectionInfo() {1769 enforceAccessPermission();1770 mLog.trace("getConnectionInfo uid=%").c(Binder.getCallingUid()).flush();1771 /*1772 * Make sure we have the latest information, by sending1773 * a status request to the supplicant.1774 */1775 return mWifiStateMachine.syncRequestConnectionInfo();1776 }1521 public WifiInfo syncRequestConnectionInfo() {1522 WifiInfo result = new WifiInfo(mWifiInfo);1523 return result;1524 }
getConnectionInfo直接通过wifiService调用wifiStateMachine中syncRequestConnectionInfo,这部分的实现不会阻塞住,难道Binder被沾满了?从trace中并没有看出这一点,那么怎么回事呢?我们尝试复现这个问题,好在比较容易复现。
09-04 18:24:29.182 D/WifiStateMachine( 1312): syncRequestConnectionInfo/in SSID: MIOffice-5G, BSSID: 70:3a:0e:2c:bb:f1, MAC: 80:ad:16:4c:0b:fe, Supplicant state: COMPLETED, RSSI: -44, Link speed: 400Mbps, Frequency: 5180MHz, Net ID: 0, Metered hint: false, score: 60 09-04 18:24:29.182 D/WifiStateMachine( 1312): syncRequestConnectionInfo/out SSID: MIOffice-5G, BSSID: 70:3a:0e:2c:bb:f1, MAC: 80:ad:16:4c:0b:fe, Supplicant state: COMPLETED, RSSI: -44, Link speed: 400Mbps, Frequency: 5180MHz, Net ID: 0, Metered hint: false, score: 60
发现在主线程进行大量输出上面的Log,极简桌面在1分钟内调用此接口160次,导致了SystemServer不能及时响应这个App,造成了App自己的ANR。像Binder call导致ANR的问题很常见,存在被blocked的风险,此刻可以尝试放到异步里面执行,其次不要短时间的大量Binder调用,这样行为轻则App自己有问题,重则系统发生Watchdog死机重启。
3.5、案例三:广播超时导致App的ANR
继续看一个案例三,按照上面的套路,首先在event log中查看发生ANR的时间。
12-17 06:02:14.463 1566 1583 I am_anr : [0,8769,com.android.updater,952680005,Broadcast of Intent { act=android.intent.action.BOOT_COMPLETED flg=0x9000010 cmp=com.android.updater/.BootCompletedReceiver (has extras) }]
发生ANR的时间是am_anr 时间点是12-17 06:02:14.463,继续看Log
12-17 06:02:00.370 1566 1583 W BroadcastQueue: Timeout of broadcast BroadcastRecord{21ef8c2 u0 android.intent.action.BOOT_COMPLETED} - receiver=android.os.BinderProxy@2a6c365, started 60006ms ago12-17 06:02:00.370 1566 1583 W BroadcastQueue: Receiver during timeout: ResolveInfo{5a8283a com.android.updater/.BootCompletedReceiver m=0x108000}12-17 06:02:00.370 1566 1583 I am_broadcast_discard_app: [0,35584194,android.intent.action.BOOT_COMPLETED,49,ResolveInfo{5a8283a com.android.updater/.BootCompletedReceiver m=0x108000}]
但是我们发现在12-17 06:02:00.370 已经发生ANR,说明event log中的时间是个大概值,可能由于系统资源比较紧张造成一定程度的滞后。由于是android.intent.action.BOOT_COMPLETED这个广播接收出现的ANR,那么我们顺藤摸瓜。
12-17 06:01:00.383 1566 3524 I ActivityManager: Start proc 8769:com.android.updater/9802 for broadcast com.android.updater/.BootCompletedReceiver caller=null
在12-17 06:01:00.383的时候启动了广播进程
12-17 06:01:36.721 8769 8769 D BootCompletedReceiver: onReceive android.intent.action.BOOT_COMPLETED12-17 06:02:14.725 8769 8769 D UpdateService: onCreate
在12-17 06:01:36.721的时候,客户端BootCompletedReceiver onReceiver方法 开始回调,然后onReceive 启动UpdateService,调用UpdateService.onCreate时间是12-17 06:02:14.725。根据上面的分析有两个初步的疑问。
启动广播是在12-17 06:01:00.383,广播的onReceiver方法开始回调是在12-17 06:01:36.721,ANR的时间是在12-17 06:02:00.370 ,那么为什么启动了36s后我才收到bootcompleted广播,这本身就不正常吧,其次为什么广播启动UpdateService耗时将近化了一分多钟?分析到这里App同学认为无法分析下去,八成是系统原因,从cpu的统计上来看,认为是某些程序占用过高导致的,贴出下面的Log。
12-17 06:02:19.286 1566 1583 E ActivityManager: ANR in com.android.updater12-17 06:02:19.286 1566 1583 E ActivityManager: PID: 876912-17 06:02:19.286 1566 1583 E ActivityManager: Reason: Broadcast of Intent { act=android.intent.action.BOOT_COMPLETED flg=0x9000010 cmp=com.android.updater/.BootCompletedReceiver (has extras) }12-17 06:02:19.286 1566 1583 E ActivityManager: Load: 0.0 / 0.0 / 0.012-17 06:02:19.286 1566 1583 E ActivityManager: CPU usage from 0ms to 18846ms later (2017-12-17 06:02:00.379 to 2017-12-17 06:02:19.224):12-17 06:02:19.286 1566 1583 E ActivityManager: 195% 6142/com.immomo.momo: 195% user + 0% kernel12-17 06:02:19.286 1566 1583 E ActivityManager: 2.3% 8170/com.tencent.mm: 2.3% user + 0% kernel / faults: 448 minor12-17 06:02:19.286 1566 1583 E ActivityManager: 0.7% 1566/system_server: 0.4% user + 0.3% kernel / faults: 150 minor 1 major12-17 06:02:19.286 1566 1583 E ActivityManager: 0.4% 90/kworker/u16:3: 0% user + 0.4% kernel12-17 06:02:19.286 1566 1583 E ActivityManager: 0.3% 4704/com.tencent.mm:push: 0.1% user + 0.2% kernel / faults: 116 minor12-17 06:02:19.286 1566 1583 E ActivityManager: 0.3% 8769/com.android.updater: 0.2% user + 0.1% kernel / faults: 1600 minor 2 major12-17 06:02:19.286 1566 1583 E ActivityManager: 0.2% 4790/com.tencent.mm:patch: 0.2% user + 0% kernel / faults: 748 minor12-17 06:02:19.286 1566 1583 E ActivityManager: 0.2% 329/mmc-cmdqd/0: 0% user + 0.2% kernel12-17 06:02:19.286 1566 1583 E ActivityManager: 0.2% 5429/com.tencent.mm:push: 0% user + 0.1% kernel / faults: 17 minor12-17 06:02:19.286 1566 1583 E ActivityManager: 0.2% 5435/com.tencent.mm:patch: 0.2% user + 0% kernel / faults: 82 minor12-17 06:02:19.286 1566 1583 E ActivityManager: 0.2% 8712/com.tencent.mm:exdevice: 0.1% user + 0% kernel12-17 06:02:19.286 1566 1583 E ActivityManager: 0.1% 432/logd: 0.1% user + 0% kernel / faults: 4 minor12-17 06:02:19.286 1566 1583 E ActivityManager: 0.1% 844/msm_irqbalance: 0% user + 0.1% kernel / faults: 4 minor12-17 06:02:19.286 1566 1583 E ActivityManager: 0.1% 7580/kworker/u16:2: 0% user + 0.1% kernel12-17 06:02:19.286 1566 1583 E ActivityManager: 0.1% 7/rcu_preempt: 0% user + 0.1% kernel12-17 06:02:19.286 1566 1583 E ActivityManager: 0.1% 1240/zygote: 0% user + 0.1% kernel / faults: 84 minor12-17 06:02:19.286 1566 1583 E ActivityManager: 0% 3216/com.xiaomi.simactivate.service: 0% user + 0% kernel / faults: 5 minor12-17 06:02:19.286 1566 1583 E ActivityManager: 0.1% 8645/kworker/7:0: 0% user + 0.1% kernel12-17 06:02:19.286 1566 1583 E ActivityManager: 0.1% 8730/kworker/4:2: 0% user + 0.1% kernel12-17 06:02:19.286 1566 1583 E ActivityManager: 0% 45/rcuop/4: 0% user + 0% kernel
然而cpu占用195%并不算高,在多核中每个核最大占用率都是100%(八核占用率是800%),其次Load: 0.0 / 0.0 / 0.0,Load在15分钟这段时间都是0,停止运转了?貌似这个Log也不太正确。然而在小米手机上,ANR的监控会有加强的,输出了下面的Log。
12-17 06:02:14.693 8769 8769 W MIUI-BLOCK-MONITOR: The msg { when=-36s107ms what=113 obj=ReceiverData{intent=Intent { act=android.intent.action.BOOT_COMPLETED flg=0x9000010 cmp=com.android.updater/.BootCompletedReceiver (has extras) } packageName=com.android.updater resultCode=0 resultData=null resultExtras=null} target=android.app.ActivityThread$H planTime=1513461660613 dispatchTime=1513461696720 finishTime=0 } took 74080ms and took 37973ms after dispatch.
我们对每一个Message都额外记录了它各个状态的时间点,方便我们进行分析
· when:消息从应该被执行到发生anr的时间
· planTime:消息计划被执行的时间点
· dispatchTime:消息真正被执行的时间点
· finishTime:消息完成时的时间点
计算消息执行的时间为:-when-(dispatchTime-planTime)=0,那么这意味着什么呢?意味着113这个Message正要开始执行,还没有开始执行就发生了ANR,在主线程的Looper消息队列中,等待了36秒。这份日志中没有主线程的trace,有也没有作用,因为可以看到这个消息还没有执行呢?那么这个36秒期间在做什么呢?有比较多下面的Log。
12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: The binder call took 3973ms.12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: java.lang.Throwable12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.os.AnrMonitor.checkBinderCallTime(AnrMonitor.java:591)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.os.BinderProxy.transact(Binder.java:623)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.content.pm.IPackageManager$Stub$Proxy.getApplicationInfo(IPackageManager.java:2658)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.app.ApplicationPackageManager.getApplicationInfoAsUser(ApplicationPackageManager.java:340)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.app.ApplicationPackageManager.getApplicationInfo(ApplicationPackageManager.java:333)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at miui.core.ManifestParser.create(SourceFile:64)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at miui.core.SdkManager.start(SourceFile:186)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at java.lang.reflect.Method.invoke(Native Method)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at miui.external.a.abx()12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at miui.external.a.attachBaseContext()12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.app.Application.attach(Application.java:193)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.app.Instrumentation.newApplication(Instrumentation.java:1009)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.app.Instrumentation.newApplication(Instrumentation.java:993)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.app.LoadedApk.makeApplication(LoadedApk.java:800)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.app.ActivityThread.handleBindApplication(ActivityThread.java:5471)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.app.ActivityThread.-wrap2(ActivityThread.java)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.app.ActivityThread$H.handleMessage(ActivityThread.java:1584)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.os.Handler.dispatchMessage(Handler.java:102)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.os.Looper.loop(Looper.java:163)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at android.app.ActivityThread.main(ActivityThread.java:6221)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at java.lang.reflect.Method.invoke(Native Method)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:904)12-17 06:01:29.334 8769 8769 W MIUI-BLOCK-MONITOR: at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:794)
36秒期间都在做bindApplication操作,这种情况下极有可能是系统状态此时并不乐观的原因,其次通过上面的分析也可以看到BootCompletedReceiver的onReceiver是在主线程处理的,启动Service也花费了很多的时间,我们也可以考虑在注册Receiver的时候,可以指定Handler,让onReceiver运行在子线程中(怎么做,可以看源码哦)
四、ANR问题的分析套路
· 抓取bugreport,搜索ANR in,查看发生的时间和进程,Cpu的负载有没有问题
· 根据进程寻找主线程的trace,发现被blocked的地方,如果是Binder call则,进一步确认下对端的情况;如果是耗时操作,直接修改成异步,怀疑系统执行慢可以看看binder_sample,dvm_lock等信息,其次gc多不多,lmk杀进程是不是很频繁,都可以看出系统的健康状态。
· 结合源码进行分析解决
关于负荷详细的可以看理解Linux系统负荷,不过多扩展。
·