What causes deadlocks and how can you prevent them?

Deep dive

Strategies:

• Lock ordering to prevent cycles.
• Try-lock + backoff to avoid hold-and-wait.
• Timeouts and deadlock detection for recovery.
• Reduce shared resources to minimize contention.

Examples

Lock ordering rule:

Always acquire locks in order: A -> B -> C

Common pitfalls

• Inconsistent lock ordering across code paths.
• Long critical sections increasing contention.
• Ignoring deadlocks in rare error paths.

Interview follow-ups

• When would you prefer detection vs prevention?
• How do you debug a deadlock in production?
• What is livelock and how is it different?