Skip to main content

Technical Diff Overview

Many modifications were made to successfully support fraud proofs and secure batch posting. The following is an in-depth summary which describes the key changes made to core Arbitrum to enable compatibility with EigenDA.

Batch Submission & Derivation​

Nitro​

  • Extended batch poster to take in an eigenDAWriter struct that writes blobs to DA via eigenda-proxy
    • Embed ABI calldata for tx submissions to SequencerInbox
  • Extended inbox message derivation to support type processing for an eigenda batch type (i.e, 0xed prefix)
  • Compute batchHeaderHash locally using batchHeader fields when querying blobs from eigenda proxy
  • Updated batch poster config to support 2MB batches when using eigenda

Nitro Contracts​

  • Extended SequencerInbox.sol to support new entry-point function for processing eigenda batch types (i.e, addSequencerL2BatchFromEigenDA)
    • Verifies certificates against stateful dependency RollupManager.sol contract which handles communication with EigenDAServiceManager.sol
    • Updated data hash computation where hash = keccak256(msgHeader, bytePrefixFlag, abi.Pack(commitment.X, commitment.Y, blob.len()))
  • Updated forge tests to verify inbox submission flow
  • Updated deployment scripts to deploy a RollupManager contract which lives as part of the RollupDeployer contract parameters and is set to the SequencerInbox storage after deployment

Nitro TestNode​

  • Updated config.ts to enable eigenda system flow
  • Updated docker-compose.yml to use eigenda-proxy dependency with mem-store
  • Updated core bash script to deploy and teardown eigenda-proxy resource

Nitro Go-Ethereum​

  • Updated system configs to use eigenda field

Fraud Proofs & Stateless Block Execution​

Nitro​

  • Default encode blobs (i.e, modulo encode, length prefix encoding, pad to nearest of 2) before pre-image injection to ensure data is in proper format for generating kzg commitments and witness proofs
  • Decode blobs to raw binary or nitro compressed batch representation when reading
  • Generate pre-image hashes using the length and commitment fields provided by the eigenda certificate which is persisted into the sequencer inbox

Arbitrator

  • Extended arbitrator to use an eigenda preimage type which is targeted during transpilation from host go code (i.e, WavmReadEigenDAHashPreimage)
  • Embed mainnet SRS values into test-files subdirectory (i.e, g1.point, g2.point, g2.point.powerOf2)
  • Updated machine proof serialization logic to target prove_kzg_preimage_bn254 when preimage.type() == PreimageType::EigenDAHash
  • Add custom proof generation logic for READPREIMAGE opcode with EigenDAHash type which computes a machine state proof containing a kzg proof using a point opening at the 32 byte offset. The machine state proof buffer format is as follows:
  • Extended E2E proof equivalence tests to serialize machine state proofs using EigenDA preimage types and ensure that post-states when one step proven on-chain match the post-state machine hashes generated by the off-chain arbitrator opcode test
  • Built kzg-bn254 library for performing kzg operations over the bn254 curve in rust

Validator

Updated replay script (replay/main.go) to use eigenDAReader when populating pre-image oracle for stateless block execution. EigenDA preimage hashes are computed as:

keccak256(commitment.X, commitment.Y, preimage.len())

Computing the length as part of the preimage hash is necessary for removing a trust assumption on the one-step-proof challenger. Unlike 4844, EigenDA preimages are variadic in size.

Nitro Contracts​

  • Extended OneStepProverHostIO.sol to perform bn254 kzg verification for an opening proof provided by a challenger
  • Added forge tests to assert the correctness of proof verification logic