Offchain Arbitrum Challenge v2

Findings · 34

1. 1 Go Root function does not check for an empty Merkle expansion Medium
2. 2 Go Root function does not accept Merkle expansion of MAX_LEVEL length High
3. 3 NewHistoryCommitment does not validate height Undetermined
4. 4 Unused errors Informational
5. 5 GeneratePrefixProof does not work in some cases High
6. 6 Divergence in VerifyPrefixProof error handling Informational
7. 7 Missing validation in Golang’s GeneratePrefixProof function Low
8. 8 Substantial amount of code duplication Informational
9. 9 Consider implementing “sanity checks” as assertions Informational
10. 10 Allow one-step proofs for length one SmallStep-type unrivaled edges Informational
11. 11 Incorrect state transition in edgeAtOneStepProof Medium
12. 12 Lack of a terminal state Informational
13. 13 Possibly unnecessary state transition Informational
14. 14 Possible state transitions never happen Informational
15. 15 Consider failing early to minimize the impact of griefing attacks Undetermined
16. 16 Presumptive edge tracker never reaches confirmation High
17. 17 Front-running a validator can trigger a denial of service High
18. 18 *LevelZeroEdge snapshots are not updated High
19. 19 Claimed edge’s timer of a BigStep edge is counted twice Medium
20. 20 Top level assertion timer not included in honest path timer calculation High
21. 21 Incorrect input parameter used to get the unrivaled time of the honest top level assertion High
22. 22 The earliestCreatedRivalBlockNumber function can be optimized to reduce looping Informational
23. 23 The localTimer function can be optimized to reduce computation Informational
24. 24 Remove honest nodes from the mutual ids map Informational
25. 25 Unsafe Uint64 operation for block number Low
26. 26 Watcher could miss edges validated by time Medium
27. 27 Possible nil deref when getting a top level assertion Low
28. 28 Discrepancy between on and off-chain confirmation timers Medium
29. 29 Front-running certain validator operations leads to honest edges not being tracked High
30. 30 Consider adding an EdgeAwaitingConfirmation state to avoid unnecessary computation Medium
31. 31 Unclear code comment regarding the ability to disable and enable staking Informational
32. 32 Validate the withdrawn amount by a staker is greater than zero Low
33. 33 Consider deleting the staker when their stake is reduced to zero Low
34. 34 Initial assertion’s status is not confirmed Informational

Findings extracted from the published report PDF. See the full report below for details and remediation.