Atomic Cross-L2 Transactions: Coordination Theory and Interlock Protocol

As Layer 2 rollups proliferate to address Ethereum's scaling challenges, enabling atomic transactions across independent rollups has become critical for cross-chain composability. This work extends the Interlock protocol, which coordinates atomic cross-rollup transactions through a stateless shared sequencer. Interlock resolves atomic transactions sequentially during execution via lightweight messaging, while optimistically processing non-dependent transactions in parallel. This approach avoids computational overhead at the sequencer layer and eliminates the need for transaction pre-computation or state simulation. We provide correctness proofs and convergence stability analysis, establishing the conditions under which coordinated resolution outperforms blind optimistic execution. We compare three coordination architectures and analyze their trade-offs in sequencer complexity, execution layer requirements, and coordination overhead. Interlock is compatible with parallel execution models, ensuring atomicity while allowing continuous optimistic processing of transactions.