Archives

The final chapter. We implement the provably correct lock-free steal, witness how it races safely against the owner, and integrate our completed deque into a working thread pool.

We build the owner's lock-free fast path: a correct push_bottom using acquire-release semantics, and a robust pop_bottom that safely resolves the critical last-item race.

A deep but gentle dive into the heart of lock-free programming: atomics, race conditions, and the strange world of memory ordering. Meet the monster that reorders your code, and learn the magic words to tame it.

Why a single shared queue is a performance trap, how cache lines become a battlefield, and how to build our first, simple deque with a lock, the correct but slow foundation for what comes next.

What if you could write to a buffer and have it loop forever, with no branches, no checks, and no wrap-around logic?

What if your allocator could recycle like a pro, pack for cache, perfect loops, and deliver unbreakable handles, without leaks or dangling pointers?

What if you could stop asking the OS for memory and just... take it? Learn how to forge an Arena: a private memory playground where allocation is instant and an entire class of bugs is simply impossible.

Ever wondered how your computer runs multiple programs without them setting each other on fire? The answer is a beautiful illusion called Virtual Memory.