Analysis of Android ANR (Application Not Responding) Causes and Diagnosis

ANR (Application Not Responding) occurs when the UI thread of an Android app is blocked for too long, triggering a system dialog that allows the user to force‑quit the app.

ANR Classification

• Input dispatching timed out – >5 s for key or touch events. • Broadcast of Intent – 10 s for foreground, 60 s for background. • Executing service – 20 s for foreground, 200 s for background. • ContentProvider – rarely, publish not completed within 10 s. • startForegroundService without startForeground within 5 s.

Root Cause Analysis

1. Common patterns when diagnosing ANR: a) Slow I/O operations on the main thread (complex layouts, frequent I/O). b) Long‑running computations on the main thread. c) Synchronous binder calls to another process that take a long time. d) Main thread blocked waiting for a long operation in another thread. e) Deadlock involving the main thread.

2. Specific causes identified in logs: • Blocking operations, suspends, deadlocks, infinite loops on the UI thread. • CPU resource contention – high CPU usage by other processes prevents the app from obtaining time slices. • Binder resource exhaustion – delayed inter‑process communication. • Memory pressure – low‑memory killer activity, frequent GC, or memory leaks. • GPU rendering stalls, Block I/O delays, native crashes, watchdog triggers, etc.

Log Analysis Procedure

Search Android system logs for the keyword “ANR in” to obtain the affected process, PID, and ANR type. Then examine:

CPU load and memory pressure.

Binder transaction logs.

Top‑process CPU usage to see if the ANR app has insufficient CPU time.

Systrace or additional trace logs for deeper insight.

Typical examples from a banking app illustrate issues such as Block I/O, GPU rendering delays, SDK adaptation problems, camera module failures, audio module crashes, APK signature verification, permission interception, provider timeouts, park lock contention, CPU time‑slice shortage, empty processes, and chipset performance limits.

Conclusion

Beyond simple timeout detection, ANR can be caused by system‑level factors like high kswapd0 activity, CPU saturation, or binder exhaustion. Comprehensive analysis requires correlating ANR traces with system logs, searching for relevant keywords, and reviewing systrace or auxiliary data to pinpoint the root cause.