Android 获取进程的 backtrace 信息

☆ 使用kill 发送 SIGNAL_QUIT

这种方法只能用于zygote 的子进程 (比如所有的 app 进程, 都是由zygote fork 而来).

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
# kill -3 pid

# cat /data/anr/traces.txt

...

suspend all histogram:  Sum: 290us 99% C.I. 2us-40us Avg: 14.500us Max: 40us
DALVIK THREADS (12):
"Signal Catcher" daemon prio=5 tid=2 Runnable
  | group="system" sCount=0 dsCount=0 obj=0x32c070a0 self=0xaecca000
  | sysTid=1918 nice=0 cgrp=bg_non_interactive sched=0/0 handle=0xb4406930
  | state=R schedstat=( 228351607 17443703 83 ) utm=12 stm=9 core=0 HZ=100
  | stack=0xb430a000-0xb430c000 stackSize=1014KB
  | held mutexes= "mutator lock"(shared held)
  native: #00 pc 00370e01  /system/lib/libart.so (_ZN3art15DumpNativeStackERNSt3__113basic_ostreamIcNS0_11char_traitsIcEEEEiPKcPNS_9ArtMethodEPv+160)
  native: #01 pc 0035046f  /system/lib/libart.so (_ZNK3art6Thread4DumpERNSt3__113basic_ostreamIcNS1_11char_traitsIcEEEE+150)
  native: #02 pc 0035a373  /system/lib/libart.so (_ZN3art14DumpCheckpoint3RunEPNS_6ThreadE+442)
  native: #03 pc 0035af31  /system/lib/libart.so (_ZN3art10ThreadList13RunCheckpointEPNS_7ClosureE+212)
  native: #04 pc 0035b45f  /system/lib/libart.so (_ZN3art10ThreadList4DumpERNSt3__113basic_ostreamIcNS1_11char_traitsIcEEEE+142)
  native: #05 pc 0035bb6f  /system/lib/libart.so (_ZN3art10ThreadList14DumpForSigQuitERNSt3__113basic_ostreamIcNS1_11char_traitsIcEEEE+334)
  native: #06 pc 00333cb7  /system/lib/libart.so (_ZN3art7Runtime14DumpForSigQuitERNSt3__113basic_ostreamIcNS1_11char_traitsIcEEEE+74)
  native: #07 pc 0033b01d  /system/lib/libart.so (_ZN3art13SignalCatcher13HandleSigQuitEv+928)
  native: #08 pc 0033b901  /system/lib/libart.so (_ZN3art13SignalCatcher3RunEPv+340)
  native: #09 pc 0003f45f  /system/lib/libc.so (_ZL15__pthread_startPv+30)
  native: #10 pc 00019b43  /system/lib/libc.so (__start_thread+6)
  (no managed stack frames)

"main" prio=5 tid=1 Native
  | group="main" sCount=1 dsCount=0 obj=0x74abb2a0 self=0xb4d76500
  | sysTid=1914 nice=0 cgrp=bg_non_interactive sched=0/0 handle=0xb6f3fb34
  | state=S schedstat=( 76934470 21396828 203 ) utm=3 stm=3 core=0 HZ=100
  | stack=0xbe55e000-0xbe560000 stackSize=8MB
  | held mutexes=
  native: #00 pc 00040894  /system/lib/libc.so (__epoll_pwait+20)
  native: #01 pc 00019e6f  /system/lib/libc.so (epoll_pwait+26)

  ...

结果写在 /data/anr/traces.txt 文件中, anr 是 ANR(Application Not Response)的意思.

https://android.googlesource.com/platform/art/+/android-7.1.1_r13/runtime/runtime.cc

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
void Runtime::InitNonZygoteOrPostFork(
    JNIEnv* env, bool is_system_server, NativeBridgeAction action, const char* isa) {

...

  // Create the thread pools.
  heap_->CreateThreadPool();
  // Reset the gc performance data at zygote fork so that the GCs
  // before fork aren't attributed to an app.
  heap_->ResetGcPerformanceInfo();
  if (!is_system_server &&
      !safe_mode_ &&
      (jit_options_->UseJitCompilation() || jit_options_->GetSaveProfilingInfo()) &&
      jit_.get() == nullptr) {
    // Note that when running ART standalone (not zygote, nor zygote fork),
    // the jit may have already been created.
    CreateJit();
  }
  StartSignalCatcher();
  // Start the JDWP thread. If the command-line debugger flags specified "suspend=y",
  // this will pause the runtime, so we probably want this to come last.
  Dbg::StartJdwp();
}

可以看出非zygote的进程都会启动 Signal Catcher的 线程.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
root@shamu:/ # ps -t 2089
USER      PID   PPID  VSIZE  RSS   WCHAN            PC  NAME
u0_a58    2089  386   1565280 53732 SyS_epoll_ b6cb0894 S com.hujiang.dict:pushservice
u0_a58    2092  2089  1565280 53732 do_sigtime b6cb0b68 S Signal Catcher
u0_a58    2095  2089  1565280 53732 unix_strea b6cb194c S JDWP
u0_a58    2096  2089  1565280 53732 futex_wait b6c875e8 S ReferenceQueueD
u0_a58    2097  2089  1565280 53732 futex_wait b6c875e8 S FinalizerDaemon
u0_a58    2099  2089  1565280 53732 futex_wait b6c875e8 S FinalizerWatchd
u0_a58    2100  2089  1565280 53732 futex_wait b6c875e8 S HeapTaskDaemon
u0_a58    2101  2089  1565280 53732 binder_thr b6cb09c0 S Binder_1
u0_a58    2102  2089  1565280 53732 binder_thr b6cb09c0 S Binder_2
u0_a58    2107  2089  1565280 53732 SyS_epoll_ b6cb0894 S Thread-123
u0_a58    2108  2089  1565280 53732 futex_wait b6c875e8 S taskService-pro

zygote 自己没有该线程.

root@shamu:/ # ps | grep -i zy
root      386   1     1528448 67416 poll_sched b6cb0a5c S zygote

127|root@shamu:/ # ps -t 386
USER      PID   PPID  VSIZE  RSS   WCHAN            PC  NAME
root      386   1     1528448 67416 poll_sched b6cb0a5c S zygote
root      2203  386   1528448 67416 futex_wait b6c875e8 S ReferenceQueueD
root      2204  386   1528448 67416 futex_wait b6c875e8 S FinalizerDaemon
root      2205  386   1528448 67416 futex_wait b6c875e8 S FinalizerWatchd
root      2206  386   1528448 67416 futex_wait b6c875e8 S HeapTaskDaemon

https://android.googlesource.com/platform/art/+/android-7.1.1_r13/runtime/signal_catcher.cc

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
...

void SignalCatcher::Output(const std::string& s) {
  if (stack_trace_file_.empty()) {
    LOG(INFO) << s;
    return;
  }
...

void SignalCatcher::HandleSigQuit() {
  Runtime* runtime = Runtime::Current();
  std::ostringstream os;
  os << "\n"
      << "----- pid " << getpid() << " at " << GetIsoDate() << " -----\n";
  DumpCmdLine(os);
  // Note: The strings "Build fingerprint:" and "ABI:" are chosen to match the format used by
  // debuggerd. This allows, for example, the stack tool to work.
  std::string fingerprint = runtime->GetFingerprint();
  os << "Build fingerprint: '" << (fingerprint.empty() ? "unknown" : fingerprint) << "'\n";
  os << "ABI: '" << GetInstructionSetString(runtime->GetInstructionSet()) << "'\n";
  os << "Build type: " << (kIsDebugBuild ? "debug" : "optimized") << "\n";
  runtime->DumpForSigQuit(os);
  if ((false)) {
    std::string maps;
    if (ReadFileToString("/proc/self/maps", &maps)) {
      os << "/proc/self/maps:\n" << maps;
    }
  }
  os << "----- end " << getpid() << " -----\n";
  Output(os.str());
}

...

while (true) {
    int signal_number = signal_catcher->WaitForSignal(self, signals);
    if (signal_catcher->ShouldHalt()) {
      runtime->DetachCurrentThread();
      return nullptr;
    }
    switch (signal_number) {
    case SIGQUIT:
      signal_catcher->HandleSigQuit();
      break;
    case SIGUSR1:
      signal_catcher->HandleSigUsr1();
      break;
    default:
      LOG(ERROR) << "Unexpected signal %d" << signal_number;
      break;
    }
  }
...

从源码中发现除了SIGQUIT 还可以发送 SIGUSR1 , 这个信号可以使进程java 虚拟机执行GC
操作 kill -10 pid

1
2
3
4
void SignalCatcher::HandleSigUsr1() {
  LOG(INFO) << "SIGUSR1 forcing GC (no HPROF)";
  Runtime::Current()->GetHeap()->CollectGarbage(false);
}

☆ 使用 debugged 命令行

这种方法是全系统通用的, 可以用于非zygote的进程.

1
2
Usage: -b [<tid>]
  -b dump backtrace to console, otherwise dump full tombstone file

通过 -b 参数指定进程pid, 即可.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
# debuggerd -b 23850

Sending request to dump task 23850.


----- pid 23850 at 1970-01-27 03:57:11 -----
Cmd line: /sbin/adbd
ABI: 'arm'

"adbd" sysTid=23850
  #00 pc 0002b158  /sbin/adbd
  #01 pc 0002467f  /sbin/adbd
  #02 pc 00020854  [stack]

"adbd" sysTid=23851
  #00 pc 0002fd38  /sbin/adbd
  #01 pc 0002a501  /sbin/adbd
  #02 pc 0000000b  <unknown>

"adbd" sysTid=23852
  #00 pc 0002b624  /sbin/adbd
  #01 pc 000106cf  /sbin/adbd
  #02 pc 00010301  /sbin/adbd
  #03 pc 0002a613  /sbin/adbd
  #04 pc 00030283  /sbin/adbd

"adbd" sysTid=23853
  #00 pc 0002b628  /sbin/adbd
  #01 pc 00013999  /sbin/adbd
  #02 pc 000112ed  /sbin/adbd
  #03 pc 000104e1  /sbin/adbd
  #04 pc 0002a613  /sbin/adbd
  #05 pc 00030283  /sbin/adbd

"adbd" sysTid=23862
  #00 pc 0002b888  /sbin/adbd
  #01 pc 0000a503  /sbin/adbd
  #02 pc 00009527  /sbin/adbd
  #03 pc 0002a613  /sbin/adbd
  #04 pc 00030283  /sbin/adbd

----- end 23850 -----
  ...

☆ java 代码中打印调用栈

1
2
3
4
5
6
try {  
 ...  
} catch (RemoteException e) {  
  e.printStackTrace();  
  ...  
}

☆ C++代码中打印调用栈

CallStack.cpp

1
2
3
4
5
6
7
8
9
#include <utils/CallStack.h>

int main() {
    android::CallStack stack;
    stack.update(); 
    stack.dump(1);

    return 0;
}

Android.mk

1
2
3
4
5
6
7
8
9
10
11
12
LOCAL_PATH:= $(call my-dir)


include $(CLEAR_VARS)
LOCAL_SRC_FILES:= CallStack.cpp
LOCAL_SHARED_LIBRARIES += libutils
LOCAL_LDLIBS += -ldl -lutils

LOCAL_CFLAGS := $(common_CFLAGS)
LOCAL_MODULE := CallStack
include $(BUILD_EXECUTABLE)

执行后显示类似下面的结果

1
2
3
4
5

root@shamu:/data/local/tmp # ./CallStack
#00 pc 000006d1  /data/local/tmp/CallStack
#01 pc 00017359  /system/lib/libc.so (__libc_init+44)
#02 pc 0000074c  /data/local/tmp/CallStack
#Android
Android 获取进程的 backtrace 信息
http://usmacd.com/cn/backtrace_in_android/
Author
henices
Posted on
September 6, 2023
Licensed under