ACDT #102 · Hoodi fork postponed Revisit after audit

AllCoreDevs - Testing #102 · 2026-10-26

Testnet fork delayed by at least 2 weeks until the security audit is complete

The Hoodi testnet fork was postponed by at least 2 weeks after a security report said the gas repricing EIPs 8037 and 8038 interact with deployed contracts in unexpected ways. The schedule will be reset after the external audit results, and spec changes are kept only as a last resort.

Why it matters

After reports that already deployed contracts could break, the entire Glamsterdam testnet schedule is tied to the audit results. The impact on both users and L2s is large enough that some pointed out that if 8037 is removed from the fork, the gas limit cannot be raised safely and optimizations like BAL become meaningless.

Decisions

Discussion

EIP-8037/8038 security issue and Hoodi postponement

After a security firm reported that two gas repricing EIPs interact with deployed contracts in unexpected ways, the call discussed whether to postpone the Hoodi fork and whether to change the spec.

Outcome: Schedule decision 00:24:42

Removing the getPayload timeout

On devnets, some EL clients delayed their getPayload responses to build more expensive blocks, causing repeated timeouts. The call discussed a change that removes the timeout definition itself and requires an immediate response.

Outcome: Confirmed 00:36:39

Payload gas limit gossip check

Because ePBS makes it easier to spread invalid payloads, a proposal was made for the consensus layer (CL) to filter out payloads in the gossip stage by checking the transaction count against the gas limit.

Shorter slots spec merge strategy

With the shorter slots implementation spread across three PRs and confusion over which commit to use as the test target, a dispute broke out over whether to merge it into the Heze spec.

Outcome: Decided 01:04:15

Glossary

EIPs covered

Forkcast record · Recording · ← Core Dev Calls