Double-Checked Locking

← Back to Concurrency Hazards & Patterns

A pattern for lazy initialization in concurrent code: check if initialized (no lock), acquire lock, check again (with lock), then initialize. Historically broken in many languages due to memory model issues (instruction reordering). Requires volatile/atomic memory ordering to be correct.

Key Properties

• Lazy Initialization Pattern
• Memory Model Pitfalls

---

concurrency patterns double-checked-locking