Explores the dangerous consequences of mixing Rust and C memory allocators through hands-on experiments. The author builds a comprehensive testing framework to demonstrate how allocator mixing leads to silent memory corruption (exit code 0) rather than immediate crashes, making it far more dangerous than expected. Key findings include glibc's metadata structure, size class optimizations, and the shocking discovery that 75% of freed memory data persists after deallocation. The experiments reveal that different allocator combinations fail in distinct ways, with debug allocators catching errors fastest while arena allocators silently leak memory.
Table of contents
Prerequisites #Table of Contents #The Interview Question That Started Everything #Why Memory Allocators Don’t Mix #Memory Fundamentals: Building Our Mental Model #Building a Memory Testing Laboratory #First Experiments: Surprising Results #Key Takeaways and What’s Next #Sort: