[sduchesneau]

Merges upstream reth v2.8.0. revm and alloy-evm are unchanged, so the v0.39.0-sf evm fork still applies.

Firehose review of the new execution features:

• --engine.prewarm-reuse commits a prewarm result instead of executing the transaction, which would hide it from the tracer. The traced path (execute_and_trace_block) now passes no prewarm results, and FirehoseEvmConfig keeps the default reuses_prewarm_results() == false, so reuse is off on our binary. The new firehose_live_tracing_prewarm_reuse test enables reuse on a stock EthEvmConfig node: with reuse wired into the traced path, 5 of 16 transfers went missing from the Firehose block.
• Prewarm and txpool prewarm workers use their own BalanceTracker inspector, never the Firehose one.
• BAL parallel execution stays off while tracing, and the sequential path still delivers transactions in block order.
• --snap.v2 (opt-in snap sync) skips executing blocks below its pivot, so no Firehose blocks exist below it. Documented in the changelog.
• No new post-execution check runs after the Firehose block is emitted.
• Hoodi activates Amsterdam on 2026-10-26 17:42:48 UTC (alloy-hardforks 0.4.10). Mainnet and Sepolia schedules are unchanged.

Tests: reth-firehose, reth-firehose-tests, and the three firehose_live_tracing* engine-tree tests pass. Battlefield reth-dev (Amsterdam, battlefield-ethereum ffd9260): 94 passing, 8 pending, compare-blocks against RPC matches. Battlefield reth-devnet (builder-playground, pre-Amsterdam): 84 passing, 19 pending (Amsterdam and BNB-only), blocks 1–131 identical to RPC.

[sduchesneau]

commits description

1. 61ee17d Merge reth v2.8.0
   • Conflict resolution:
     • Version bumps in Cargo.toml and docs/vocs/vocs.config.ts.
     • Both new dev-dependencies kept in crates/engine/tree/Cargo.toml.
     • In bin/reth/src/main.rs, our Firehose node builder now launches through upstream's new launch_with_debug_capabilities_and_backfill. That keeps the opt-in --snap.v2 working.
     • Cargo.lock taken from upstream, with firehose-tracer pinned back to 5.5.0 (cargo had moved it to 5.7.0).
   • The one real change, in our traced twin execute_and_trace_block (payload_validator.rs): upstream added a prewarm-results argument to execute_transactions, and our copy no longer compiled. I pass None there, with a comment saying why. When upstream gets prewarm results, it saves them without running the transaction, so the tracer would miss it. Passing None means the traced path always runs every transaction.
2. 0f73e2e Test live tracing with prewarm reuse enabled
   • New test crates/engine/tree/tests/firehose_live_tracing_prewarm_reuse.rs.
   • It turns on prewarm reuse on a node and checks that all 16 transfers in a block reach the Firehose block.
   • With reuse wired into the traced path, 5 of the 16 went missing, so the test catches the problem.
3. 0dec4e1 Add reth-v2.8.0-fh3.1 changelog section
   • The upstream bump.
   • The Hoodi Amsterdam date (2026-10-26).
   • That --engine.prewarm-reuse has no effect while tracing.
   • That --snap.v2 produces no Firehose blocks below its starting block.

Nothing else Firehose-specific changed, because no other upstream change touched the tracing path.