Implement block cache and buffer management system #63
Description
Goal
Build a buffer cache/pager system to smooth repeated reads, coalesce writes, and provide centralized eviction/flush policy.
Context
Block caching sits between the block device driver and filesystem layer, providing performance optimization and write coordination.
Definition of Done
- Buffer cache: Cache frequently accessed disk blocks in memory
- Read caching: Eliminate repeated reads of the same blocks
- Write coalescing: Combine multiple writes to the same block
- Eviction policy: LRU or similar algorithm for cache management
- Flush policy: Coordinated writeback with configurable timing
- Dirty tracking: Track modified blocks that need writeback
- Synchronization: Thread-safe access to cached blocks
Implementation Details
- Implement hash table for fast block lookup by device and sector
- Add LRU chain for cache eviction policy
- Create dirty block tracking and periodic flush
- Implement read-ahead for sequential access patterns
- Add write-behind with configurable delays
- Support both sync and async I/O modes
- Integrate with block device layer
Cache Features
- Read-ahead: Prefetch sequential blocks
- Write coalescing: Merge adjacent or overlapping writes
- Sync/async modes: Support both blocking and non-blocking I/O
- Cache pressure: Handle memory pressure by evicting clean blocks
- Integrity: Ensure data consistency during crashes
Performance Goals
- Cache hit rate greater than 90% for typical workloads
- Minimize memory footprint while maximizing effectiveness
- Reduce I/O latency through read-ahead
- Optimize write patterns through coalescing
Files to Create
- block/buffer_cache.h - Buffer cache interface
- block/buffer_cache.c - Cache implementation
- block/writeback.c - Dirty block writeback system
- block/readahead.c - Sequential read optimization
Testing Strategy
- Measure cache hit rates with various workloads
- Test data integrity under concurrent access
- Verify proper writeback under memory pressure
- Stress test with large file operations
Dependencies
- Block device driver implementation
- Memory management for cache allocation
- Timer system for periodic flush operations
Related Issues
Part of filesystem support stack - builds on block device driver.