Skip to content

Merge upstream v2.8.1 - #14

Merged
sduchesneau merged 12 commits into
release/2.xfrom
bump/world-chain-2.8.1
Nov 7, 2026
Merged

sduchesneau merged 12 commits into
release/2.xfrom
bump/world-chain-2.8.1

Conversation

@dfuse-bot

@dfuse-bot dfuse-bot commented Nov 7, 2026 •

Copy link
Copy Markdown

Merges worldcoin/world-chain v2.8.1 (5 commits since v2.6.0), which World Chain asks every node operator to run to avoid falling out of sync. Upstream moved reth to ethereum-optimism/reth op-v2.6.0-2 (reth v2.6.0) and the optimism monorepo to op-reth/v2.8.1, so both Firehose forks follow on their world-chain lines. alloy-evm moves to streamingfast/evm v0.39.0-sf; firehose-tracer stays at 5.5.0, so the block model does not change and the tag stays on fh3.1.

Release: once merged, tag the merge commit v2.8.1-fh3.1 (newest tag on release/2.x is v2.6.0-fh3.1; the prompt's tag matches the naming rule).

Merge order. This PR pins tags that do not exist yet and its build fails until they do:

  1. Move the world-chain line to ethereum-optimism/reth op-v2.6.0-2 reth#91, merge and tag world-chain-v2.8.1-fh3.1
  2. Move the world-chain line to op-reth v2.8.1 optimism#29, merge and tag world-chain-v2.8.1-fh3.1
  3. this PR, then tag v2.8.1-fh3.1

Cargo.lock records the heads of those two PRs (0d253a1b, 889f75b6) for the tags. If either tag lands on another commit, refresh the lock with cargo update -p reth-firehose -p reth-optimism-firehose before merging.

Merge

  • bin/world-chain/src/main.rs: keeps the Firehose init_tracer call, takes upstream's parse_args().
  • Cargo.toml: upstream keeps the kona crates (proof services and devnet only) on optimism rev da6d3252, with the op-alloy, op-revm and alloy-op-evm crates of that rev, while the node uses the same names at op-reth/v2.8.1. A [patch] covers every rev of its URL, so patching those names would move the proof client onto the new rev, where it does not compile. The 8 root dependencies on names that exist at both revs (op-alloy-*, alloy-op-hardforks, op-revm, alloy-op-evm) therefore point at the fork tag directly, and the optimism [patch] keeps only the crates that exist only at the new rev (reth-optimism-*, op-alloy-rpc-jsonrpsee). The reth patch is unchanged in shape.
  • Cargo.lock: upstream's v2.8.1 lock with the node's sources swapped to the forks. The da6d3252 set is identical to upstream's (31 crates, same commit), and proofs/measured/ is untouched. firehose-tracer is locked back to 5.5.0 (a fresh resolve picks 5.7.0, which reth-firehose does not compile against). No unused patch remains.
  • Dockerfile.sf: base image Rust 1.99, as upstream's Dockerfile (the pinned nightly still compiles the node).
  • ACTIVATIONS.md reviewed: no World Chain fork or timestamp changed (details below).
  • Plan and conflict notes: .dev/update-to-v2.8.1-fh.md.

Operator-visible upstream changes

  • --proofs-history.storage-version v1 is now rejected, and --engine.num-state-masking-blocks must be 0.
  • --engine.persistence-threshold defaults to 50: after an unclean shutdown the node re-executes, and re-emits, up to 50 blocks.

Firehose bug patterns

  1. Serial order: none found. Live blocks go through OpFirehoseEngineValidatorBuilder with parallel_bal_execution = false; payload building and flashblock validation use the untraced EVM config. The upstream commits do not touch block execution.
  2. Emission after acceptance: none found. Same validator as op-reth v2.8.1 (Merge op-reth v2.8.1 optimism#28); its new extraData and BAL gas-limit checks run before mark_verified().
  3. Activations: WorldChainHardfork unchanged, and the World Chain mainnet and Sepolia superchain-registry files are identical at the new gitlink (13907b9b). Tropo and Strato still have no execution code. Lagoon still maps to Never; the only fork-gated op-revm change since da6d3252 is the Lagoon BLS precompile input cap, and handler.rs changed only in comments, so the fee vault credits still match.
  4. Fields and enums: OpTxEnvelope, OpTxType, OpReceipt and the header unchanged; World Chain adds no transaction type.
  5. Transactions without a hash: not affected.

Test Execution Summary

worldcoin/world-chain v2.8.1 merged into streamingfast/world-chain release/2.x, with streamingfast/reth and streamingfast/optimism moved on their release/world-chain-2.x lines (branch bump/world-chain-2.8.1 in all three). The forks were built through local world-chain-v2.8.1-fh3.1 tags on the PR heads (never pushed), so the lock files already resolve the tag names.

Build Status

  • streamingfast/reth: cargo test -p reth-firehose -p reth-firehose-tests (builds the crates): PASSED
  • streamingfast/optimism (rust/): cargo check --locked -p op-reth -p reth-optimism-firehose -p alloy-op-evm --tests: PASSED
  • streamingfast/world-chain: cargo build --locked --bin world-chain: PASSED
  • streamingfast/world-chain: cargo check --locked --workspace --all-targets: PASSED
  • streamingfast/world-chain: cargo clippy --locked --workspace --all-targets -- -D warnings: PASSED (only cargo's manifest notices about unused workspace dependencies and [lints], as upstream)
  • Docker image (Dockerfile.sf): NOT RUN. It fetches the fork tags, which do not exist until the reth and optimism PRs are merged and tagged; the sf-release.yml PR build fails for the same reason.

Unit Tests

Repo Package / command Status Notes
streamingfast/reth cargo test -p reth-firehose -p reth-firehose-tests PASSED 33 unit tests, 2 prestate tests
streamingfast/optimism cargo test --locked -p reth-optimism-firehose -p alloy-op-evm -p reth-optimism-evm -p reth-optimism-node -p reth-optimism-payload-builder --no-fail-fast PASSED alloy-op-evm 114, reth-optimism-evm 58, reth-optimism-node 51 + 25 + 2 (1 ignored), reth-optimism-payload-builder 41 + 1, doc tests 3
streamingfast/world-chain cargo test --locked -p world-chain-evm -p world-chain-node -p world-chain-cli --no-fail-fast PASSED world-chain-cli 36 (includes upstream's new state masking test), world-chain-evm 15, world-chain-node 6 + 2

The world-chain tests ran with CARGO_PROFILE_DEV_DEBUG=0 CARGO_PROFILE_TEST_DEBUG=0, as upstream's rust-ci.yml sets: with full debug info, linking the world-chain-node integration test fails (rust-lld: relocation R_X86_64_32 out of range in .debug_macro, over 4 GiB of debug info).

Integration Tests

  • Status: NOT RUN (the workspace's own Docker devnet tests; Battlefield below drives that devnet).

Battlefield Tests

  • Status: PASSED, with one expected false diff in the block comparison (same as the v2.6.0 run)
  • Pass/Fail: 78 passing / 1 failing, 25 pending (skipped by the suite on this network: Amsterdam and BAL, blobs, slot number, Arc assets and other chains' features)
  • Setup: streamingfast/battlefield-ethereum master (879ea9a), target world-chain-devnet, fireeth v2.24.1, world-chain built from this branch (dev profile). Devnet hardforks: Bedrock through Karst at genesis.
  • Failures:
    1. Compare blocks: Firehose blocks match RPC blocks (fireeth tools compare-blocks-rpc, blocks Fix sf-release CI branch refs after release/2.x rename #1–chore(builder): add ability to limit uncompressed block size worldcoin/world-chain#441): all 441 blocks differ, and only in transaction_traces[N].blob_gas and transaction_traces[N].receipt.blob_gas_used, which the RPC side has and the Firehose block does not. Expected diff, not a node change: under Jovian, op-reth's receipt puts the transaction's DA footprint in receipt blobGasUsed (block 100: a deposit 0x7e transaction with blobGasUsed = 0xa5a0 while the header's blobGasUsed is 0x0), and compare-blocks-rpc copies it into blob_gas. The fix belongs in the compare tool.
  • No tracked snapshot changed. The suite wrote test/snapshots/genesis/fh3.0/world-chain-devnet/, which the repository does not have yet.
  • Reader: FIRE INIT read with fh_version 3.1, no panic or error in the reader or node logs, and the node kept up with the sequencer at 2 s blocks (to block 653, LIB 453).
  • Commands, from battlefield-ethereum:
    1. ./scripts/world_chain/run_world_chain_devnet.sh (with WORLD_CHAIN_REPO = this branch)
    2. WORLD_CHAIN_BINARY=<world-chain>/target/debug/world-chain ./scripts/run_firehose_world_chain_devnet.sh
    3. pnpm test:fh3.0:world-chain-devnet, once fireeth served on :8089
  • Workarounds for this Linux environment, none committed:
    • The devnet's kona-node, op-conductor and service containers advertise host.docker.internal, which only resolves on Docker Desktop. A local change to crates/devnet/src/full_stack.rs added with_host("host.docker.internal", Host::HostGateway) to them.
    • The devnet ran as root, because it rewrites files its containers create as root in bind mounts.
    • find_op_workdir in the Battlefield script tries macOS stat -f %m first; on Linux that succeeds with filesystem output, so a shim made it fall back to stat -c %Y.
    • The devnet's first start (pulling a 7.4 GB SP1 image) takes longer than the script's 600 s wait, so the devnet was started and awaited before the script's wait began.

CHANGELOG

## v2.8.1-fh3.1

### Changed

* Merged upstream `worldcoin/world-chain` `v2.8.1` (5 commits since `v2.6.0`, through
  `v2.8.1-rc.2`), which World Chain requires to keep nodes from falling out of sync. Upstream moved
  reth from `ethereum-optimism/reth` `0fbe4285` to `op-v2.6.0-2` (reth v2.6.0), the optimism
  monorepo from `da6d3252` to `op-reth/v2.8.1`, revm to 43 and alloy-evm to 0.39.
* Both Firehose forks moved to the new revs on their `release/world-chain-2.x` branches:
  `streamingfast/reth` and `streamingfast/optimism` `world-chain-v2.8.1-fh3.1`.
* `alloy-evm` now comes from `streamingfast/evm` `v0.39.0-sf`, so block traces keep their
  `systemCalls`.
* `firehose-tracer` stays locked at 5.5.0, the version the forks are built against. The Firehose
  block model does not change.
* Upstream now rejects `--proofs-history.storage-version v1` and any nonzero
  `--engine.num-state-masking-blocks`.

Open questions

  • The kona fix changes 8 upstream dependency lines in Cargo.toml (op-alloy-*, alloy-op-hardforks, op-revm, alloy-op-evm point at our fork directly), so they will conflict on future upstream bumps. A [patch] cannot cover only one rev of a URL, and the alternative, editing proofs/measured/kona-client/Cargo.toml, would change measured inputs. Once upstream moves kona to the same rev as op-reth, these lines can go back to upstream's and into the patch.
  • The Battlefield block comparison fails on the same blob_gas false diff as v2.6.0. It needs a fix in fireeth tools compare-blocks-rpc (skip receipt blobGasUsed on OP chains after Jovian), not in this repository.
  • Dockerfile.sf was not built: it needs the fork tags. Please check the sf-release.yml build once the tags exist.

alessandromazza98 and others added 10 commits October 22, 2026 10:33
* fix(proofs): upgrade SP1 guests and tooling to 6.10.0

* fix(proofs): bump enclave tokio to 1.53.3 for RUSTSEC-2026-0386

Amp-Thread-ID: https://ampcode.com/threads/T-01a181bc-23bf-77cb-aaa9-57a24e1657f8
Co-authored-by: Amp <amp@ampcode.com>

* chore(proofs): refresh measurements from CI run 40676382882

Amp-Thread-ID: https://ampcode.com/threads/T-01a181bc-23bf-77cb-aaa9-57a24e1657f8
Co-authored-by: Amp <amp@ampcode.com>

* ci(proofs): install SP1 6.10.0 toolchain

* chore(proofs): ignore RUSTSEC-2026-0387 until upstream crypto-bigint upgrade

Amp-Thread-ID: https://ampcode.com/threads/T-01a181bc-23bf-77cb-aaa9-57a24e1657f8
Co-authored-by: Amp <amp@ampcode.com>

---------

Co-authored-by: Phillip Kemper <phillip.kemper@toolsforhumanity.com>
Co-authored-by: Amp <amp@ampcode.com>
* chore(deps): upgrade op-reth to v2.8.1-rc.2

Amp-Thread-ID: https://ampcode.com/threads/T-01a1aae7-1d46-73ab-8531-2bdb9d43ec63
Co-authored-by: Amp <amp@ampcode.com>

* docs(cli): regenerate reference for op-reth upgrade

Amp-Thread-ID: https://ampcode.com/threads/T-01a1aae7-1d46-73ab-8531-2bdb9d43ec63
Co-authored-by: Amp <amp@ampcode.com>

* fix(cli): disable state masking to match op-reth

Amp-Thread-ID: https://ampcode.com/threads/T-01a1aae7-1d46-73ab-8531-2bdb9d43ec63
Co-authored-by: Amp <amp@ampcode.com>

* fix(node): import state-masking guard from color-eyre

Amp-Thread-ID: https://ampcode.com/threads/T-01a1aae7-1d46-73ab-8531-2bdb9d43ec63
Co-authored-by: Amp <amp@ampcode.com>

---------

Co-authored-by: Amp <amp@ampcode.com>
Both Firehose forks move to world-chain-v2.8.1-fh3.1 and alloy-evm to v0.39.0-sf.
firehose-tracer stays locked at 5.5.0.
@dfuse-bot
dfuse-bot requested a review from sduchesneau November 7, 2026 15:12
StreamingFast Chain Update Bot added 2 commits November 7, 2026 15:30
A git patch covers every rev of its URL, which moved the proof client onto
op-reth v2.8.1. Pin the node's op crates to the fork directly instead.
@sduchesneau
sduchesneau merged commit 4ccb230 into release/2.x Nov 7, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants