于android上的bug定位的文档很少,因为应用程序千差万别的,出现的问题也不尽相同,不过也是有规律可循,大的方向定位是可以做到的,我们对应用程上的问题可以得到相关的信息。
首先,要对Java的Throwable比较熟悉,因为Android上的应用和服务都是Java的代码,它的Error和Exception都是沿用Java的,比如Error有AssertionError,VirtualMachineError,OutOfMemoryError和其他的Error类。Exception有RuntimeException和IOException,请参考相应的文档查询,Adb logcat里面会把出现错误的Error或Exception打印出来。
分类1, 应用程序错误,什么样子的?大家如果用过android手机会碰到过xxxx process意外停止,Force close的对话框弹出来。这一般都是应用程序错误。这个过程一般有uncaughtException,crash(TAG,e),handleApplicationError,sendSingal(SIGQUIT),logThreadStacks然后会在/data/anr/traces.txt追加process crash信息。
举例:
11-04 08:55:37.114 W/AudioFlinger( 1032): write blocked for 55 msecs
11-04 08:55:37.334 W/dalvikvm( 1103): threadid=35: thread exiting with uncaught exception (group=0x2aadda08)
11-04 08:55:37.354 E/AndroidRuntime( 1103): Uncaught handler: thread WindowManagerPolicy exiting due to uncaug
ht exception
11-04 08:55:37.374 E/AndroidRuntime( 1103): *** EXCEPTION IN SYSTEM PROCESS. System will crash.
11-04 08:55:37.394 I/global ( 1566): Default buffer size used in BufferedReader constructor. It would be bett
er to be explicit if an 8k-char buffer is required.
11-04 08:55:37.464 E/AndroidRuntime( 1103): java.lang.NullPointerException
11-04 08:55:37.464 E/AndroidRuntime( 1103): at android.graphics.Canvas.throwIfRecycled(Canvas.java:954)
11-04 08:55:37.464 E/AndroidRuntime( 1103): at android.graphics.Canvas.drawBitmap(Canvas.java:980)
11-04 08:55:37.464 E/AndroidRuntime( 1103): at com.android.internal.policy.impl.UnlockSliderView$AbstractView
State.drawCanalBmp(UnlockSliderView.java:851)
11-04 08:55:37.464 E/AndroidRuntime( 1103): at com.android.internal.policy.impl.UnlockSliderView$AbstractView
State.drawSlideCanalBmp(UnlockSliderView.java:822)
11-04 08:55:37.464 E/AndroidRuntime( 1103): at com.android.internal.policy.impl.UnlockSliderView$ViewStateSli
ding.drawSlideImage(UnlockSliderView.java:1798)
11-04 08:55:37.464 E/AndroidRuntime( 1103): at com.android.internal.policy.impl.UnlockSliderView$ViewStateSli
ding.onDraw(UnlockSliderView.java:1766)
11-04 08:55:37.464 E/AndroidRuntime( 1103): at com.android.internal.policy.impl.UnlockSliderView.onDraw(Unloc
kSliderView.java:507)
分类2,Java application Hang,当应用程序停止响应事件比如按键/Touch的时候,会诸如如下流程,broadcastTimeout,appNotRespondingLocked(frameworks/base/services/java/com/android/server/am/ActivityManagerService.java),sendSingal(SIGQUIT)(frameworks/base/core/java/android/os/Process.java),SingalCatcherThreadStart(dalvik/vm/SignalCatcher.c),logThreadStacks() (dalvik/vm/SignalCatcher.c)然后会在/data/anr/traces.txt追加process crash信息。
举例:
11-04 15:02:00.795 I/dalvikvm( 1270): processname:com.android.phone
11-04 15:02:00.795 I/dalvikvm( 1270): crashstring:Java Crash/Hang (SIGQUIT)
11-04 15:02:00.795 I/dalvikvm( 1270): crashlog:
其他的android错误,通常开发人员会用Log.e来甄别这些信息。
Dalvik/ Core Libraries的错误
类似于01-06 17:27:24.526: INFO/DEBUG(963): crashlog:
01-06 17:27:24.526: INFO/DEBUG(963): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
01-06 17:27:24.526: INFO/DEBUG(963): Build fingerprint: 'xxxxbuild/test-
keys'
01-06 17:27:24.526: INFO/DEBUG(963): pid: 2539, tid: 2539 >>> /system/bin/vold <<<
01-06 17:27:24.526: INFO/DEBUG(963): signal 11 (SIGSEGV), fault addr 00000000
01-06 17:27:24.526: INFO/DEBUG(963): r0 00000000 r1 00020000 r2 00000000 r3 80808080
01-06 17:27:24.526: INFO/DEBUG(963): r4 00000000 r5 00000000 r6 beff7490 r7 000111cc
01-06 17:27:24.526: INFO/DEBUG(963): r8 00000000 r9 00000000 10 00000000 fp 00000000
01-06 17:27:24.526: INFO/DEBUG(963): ip 00000000 sp beff7468 lr 0000a6c1 pc afe0e1b8 cpsr 40000010
01-06 17:27:24.576: INFO/DEBUG(963): #00 pc 0000e1b8 /system/lib/libc.so
01-06 17:27:24.576: INFO/DEBUG(963): #01 pc 0000a6be /system/bin/vold
01-06 17:27:24.576: INFO/DEBUG(963): #02 pc 0000a858 /system/bin/vold
01-06 17:27:24.576: INFO/DEBUG(963): #03 pc 0000a918 /system/bin/vold
01-06 17:27:24.586: INFO/DEBUG(963): #04 pc 000097b2 /system/bin/vold
01-06 17:27:24.586: INFO/DEBUG(963): #05 pc 0001fd7a /system/lib/libc.so
01-06 17:27:24.586: INFO/DEBUG(963): #06 pc 0000bcd2 /system/lib/libc.so
01-06 17:27:24.586: INFO/DEBUG(963): #07 pc b000157e /system/bin/linker
01-06 17:27:24.586: INFO/DEBUG(963): code:
01-06 17:27:24.586: INFO/DEBUG(963): afe0e1a8
01-06 17:27:24.586: INFO/DEBUG(963): e3120003
01-06 17:27:24.586: INFO/DEBUG(963): e28cc001
。。。
具体调用过程__linker_init() (bionic/linker/linker.c) -> debugger_init() (bionic/linker/debugger.c)-> debugger_signal_handler()Called when a signal is received from Kernel, uses socket() to connect to The “android:debuggerd” socket, and write()s to the socket -> main()(system/core/debuggerd/debuggerd.c)The debuggerd daemon creates a socket server android:debuggerd and loops forever,waiting for some client to write into the socket -> handle_crashing_process()->engrave_tombstone() (system/core/debuggerd/debuggerd.c)Dump stack trace and registers in /data/tombstone/ folder在/data/tombstone/会看到这些东西
kernel错误:比较难定位,如果系统发生crash现象,可以开机之后取得最后一次的kernel日志来定位,可以在/proc/last_kmsg得到相关的东西来定位系统是不是kernel问题。在日志里面通常最后会有Kernel panic 。。。。
Modem的问题,这个跟平台有关系,比如G1是高通平台,如果发生在Modem的问题crash了,你可以到/proc/last_amsslog 找到,发给高通解决。比如从/proc/last_kmsg得到
05-29 09:09:02.409 <0>[54026.002654] Kernel panic - not syncing: Modem has crashed... 这样可以定位到modem的问题,找到/proc/last_amsslog 相关信息,不过这个是二进制文件的。
总结:如果是用户程序Exception或者Dalvik Error Invoking Runtime都会通过AcitivityMangerService发出SIGQUIT信号给process,再调用SignalCatcher.c再把crashed process信息放在/data/anr/traces.txt里面。
如果是SYSTEM process????或者通过Log.e(TAG,str,trowable)会直接报告exception到logcat里面。
如果是Dalvik Error, Invoking debuggerd/C code/LibC Error, 会调用Tombstone,然后把信息打印到/data/tombstones目录里面。
如果是kernel错误,会直接放到/proc/last_kmsg文件(下次重启后会有)
如果是Modem错误,会直接有/proc/last_amass文件出现(高通平台)。