Optimize database logging and network reload handling

[kasnder]

Summary

This PR implements performance optimizations to reduce database I/O overhead and improve responsiveness by batching log writes and debouncing network state change handlers.

Key Changes

Database Logging Optimization (DatabaseHelper.java)

• Implemented log batching to reduce per-packet database I/O operations
  • Logs are accumulated in memory and flushed in batches of 50 or every 5 seconds
  • Moved ContentValues construction outside the database transaction
  • Batch flush is triggered before log queries and clears to ensure consistency

Network Reload Debouncing (ServiceSinkhole.java)

• Added debounced reload mechanism to prevent excessive reloads from rapid network state changes
  • Network callbacks (onAvailable, onLinkPropertiesChanged, onCapabilitiesChanged, onLost) now use debouncedReload() instead of direct reload() calls
  • Enforces minimum 1-second interval between consecutive reloads
  • Pending reloads are coalesced to avoid redundant operations

Battery Optimization (ServiceSinkhole.java)

• Skip stats updates when screen is off to reduce battery drain
  • Stats update frequency increased from 1000ms to 2000ms default

ICMP Response for Blocked UDP (udp.c)

• Send ICMP Port Unreachable responses for blocked UDP packets
  • IPv4: ICMP Destination Unreachable (Type 3, Code 3)
  • IPv6: ICMPv6 Destination Unreachable (Type 1, Code 4)
  • Includes original packet data per RFC 792/4443 for proper error handling
  • Prevents long timeouts and improves app responsiveness

Implementation Details

• Log batching uses synchronized list to safely accumulate entries across threads
• Debouncing uses Handler with delayed execution to coalesce rapid network changes
• ICMP responses properly construct headers with correct checksums and include original packet data
• All optimizations maintain backward compatibility and data consistency

https://claude.ai/code/session_01EAkj12Bt6ywz1bQxNJEcKt