fix(gossip): bind message ids to topics

[niran]

Summary

• bind gossipsub message IDs to the topic using the OP Stack preimage shape: domain || topic_len_le || topic || payload
• preserve the existing Snappy decoded-size guard and valid/invalid Snappy domains
• add a regression test that byte-identical payloads on V3 and V4 block topics produce different message IDs

Context

The relevant op-node implementation is op-node/p2p/gossip.go::BuildMsgIdFn, which includes the topic length and topic before hashing the payload. Without topic binding, libp2p-gossipsub duplicate-cache entries can collide across Base block-version topics.

Test Plan

• cargo test --package base-consensus-gossip --lib
• cargo clippy --package base-consensus-gossip --all-targets -- -D warnings

[cb-heimdall]

🟡 Heimdall Review Status

Requirement	Status	More Info
Reviews	🟡 0/1	Denominator calculation Show calculation 1 if user is bot 0 1 if user is external 0 2 if repo is sensitive 0 From .codeflow.yml 1 Additional review requirements Show calculation Max 0 0 From CODEOWNERS 0 Global minimum 0 Max 1 1 1 if commit is unverified 0 Sum 1

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
base	Ignored		Apr 29, 2026 9:57pm

[github-actions]

Review Summary

No blocking issues found. The change is correct and well-motivated.

What was reviewed

Single file change to crates/consensus/gossip/src/config.rs — binds gossipsub message IDs to the topic string, matching the upstream OP Stack BuildMsgIdFn implementation.

Key observations

• Preimage layout is correct: domain (4B) || topic_len_le (8B) || topic || data. The fixed-width u64 length prefix before the topic eliminates any preimage ambiguity between topic and data fields.
• Domain constants: Promoted from inline Vec allocations to const [u8; 4] — eliminates per-call heap allocation in a hot path. Good improvement.
• Both snappy paths are consistent: Valid and invalid snappy paths both include the topic in the hash preimage via the shared message_id_hash helper.
• Regression test (test_compute_message_id_includes_topic): Directly verifies the behavioral property — byte-identical payloads on /optimism/8453/2/blocks vs /optimism/8453/3/blocks produce different message IDs.

Note

The local spec (docs/specs/pages/protocol/consensus/p2p.md, lines 159–168) documents message ID computation without topic binding. If the spec is intended to track the actual implementation, it should be updated to reflect this change. This is non-blocking — the PR description makes clear this matches the upstream Go implementation.