[Core Protocol] What is the flow when deposit tx execution failed in L2

[GrapeBaBa]

Issue Description

I am still working on op-besu, want to confirm this behavior in the exec-engine.

From exec-engine spec:
Note: The state transition rules have been modified such that deposits will never fail
so if engine_forkchoiceUpdatedV2 returns STATUS_INVALID it is because a batched transaction is invalid.

1. What does it mean a batched transaction is invalid? Does it mean only non deposit tx is invalid will returning STATUS_INVALID?

2. Looked at code, the engine will rewind the state change by deposit tx but remain it in the L2 block, am I right?

3. What does user need to do when it happen?

Additional Information

No response

[ajsutton]

1. Correct, a deposit tx should never make a block invalid (because the sequencer is forced to include all deposit tx)
2. Correct, the EVM will revert and no state changes from EVM execution are peristed. However, any ETH minted by the deposit tx will remain minted and the nonce of the sender on L2 is still incremented for a failed deposit tx. https://github.com/ethereum-optimism/optimism/blob/47ba316de3e146194de05d53e5afc729a34649e0/op-e2e/deposit_test.go#L16 tests this behaviour with op-geth.
3. The user doesn't necessarily need to do anything but it depends on what the deposit tx was intended to do. There's no way to roll back the deposit transaction (L1 doesn't know about L2 so it has to assume the tx went through on L2). That's why minted ETH remains on L2 but depending on what the transaction was attempting to do it may not be possible to recover.

[github-actions[bot]]

Hey @GrapeBaBa, We hope your recent question was resolved to your satisfaction. Your feedback is invaluable and helps us improve our support services. Could you spare a moment to fill out a short feedback survey. Thank you for helping us improve our developer community.