简单易用的Android主线程耗时检测类 MainThreadMonitor

适用场景 debug 本地测试

代码类 MainThreadMonitor.java

public class MainThreadMonitor {
    private static final String TAG = "MainThreadMonitor";
    private static ScheduledExecutorService executorService = Executors.newScheduledThreadPool(10);

    public static void startMonitorMainThread(long threshold) {
        executorService.scheduleAtFixedRate(() -> {
            checkMainThreadState(threshold);
        }, 0, threshold, TimeUnit.MILLISECONDS);
    }


    private static void checkMainThreadState(long threshold) {
        AtomicBoolean exec = new AtomicBoolean(false);
        long startTime = System.currentTimeMillis();

        Handler mainHandler = new Handler(Looper.getMainLooper());
        // 预期主线程执行runnable的时候会把 exec置为true
        mainHandler.post(() -> exec.set(true));

        executorService.schedule(() -> {
            // 阈值之后发现 exec 还没有置为true，说明主线程阻塞了
            if (!exec.get()) {
                printMainThreadStackTrace(startTime);
            }
        }, threshold, TimeUnit.MILLISECONDS);
    }

    private static void printMainThreadStackTrace(long startTime) {
        StackTraceElement[] stackTrace = Looper.getMainLooper().getThread().getStackTrace();
        long consumeTime = System.currentTimeMillis() - startTime;
        Log.e(TAG, String.format("start block:(%s) ms  ====================== ", consumeTime));
        for (StackTraceElement s : stackTrace) {
            Log.d(TAG, "block: " + s);
        }
    }

}

使用方式 Application的attachBaseContext

class App : Application() {

    override fun attachBaseContext(base: Context?) {
        super.attachBaseContext(base)
        MainThreadMonitor.startMonitorMainThread(500)
    }

}

log输出示例

这里需要解释一下 start block 这行的含义，我们看到后面有一个 501ms的耗时，这里并不意味着该方法/函数执行的时间是501ms，而是表述这500ms之内都在此方法中执行，很有可能下一个500ms还是这个堆栈信息，因此 方法实际执行时间 >= 500ms