Contents and sources 3 sections

Key points

  • The November 6 development is a call for checks; the Hoodi delay was announced on October 22.
  • With separate state-gas accounting, remaining execution gas can increase under certain conditions.
  • Raising a gas limit does not complete a compatibility review. Failure accounting and retries still need testing.

A call for checks, following an earlier delay

On November 6, the Ethereum Foundation asked developers and auditors to check the impact of EIP-8037 and EIP-8038, providing eight potentially broken gas assumptions for projects to examine.[1][6]

The timeline matters: the Hoodi testnet upgrade delay was announced on October 22, without a replacement activation date. Today's call for checks continues the work on that compatibility issue.[1][2]

Why remaining gas can increase

EIP-8037 separates execution-gas and state-gas metering. The former pays for work such as computation; the latter accounts for lasting state growth, such as new accounts and stored data. Users still pay the sum. The design aims to keep the data nodes must retain from growing too quickly as block capacity increases.[3]

EIP-8038 reprices state access, including higher cold-account access costs and separate write charges. It does not raise every operation's cost by the same percentage. Whether an existing gas budget can still run the same code needs checking.[4]

A less intuitive change comes from EIP-8037's reservoir mechanism. The GAS opcode returns only remaining execution gas, excluding the state-gas reservoir. When newly created state is undone, state gas previously paid from the execution balance can be credited back under the specified rules. A gasleft() reading can therefore rise after some code has run.[3]

That affects code which measures consumption by subtracting the final balance from the initial one. If the final reading is higher, the difference is negative; checked unsigned arithmetic can then revert an otherwise valid flow. This is a balance refill during execution, distinct from a fee refund after the transaction.[3]

DeFi teams need to test what happens after failure

Fixed budgets can also sit in off-chain relayers or transaction settings, beyond the contracts themselves. ChainSecurity's analysis identifies another conditional risk: designs relying on insufficient gas after a failed external call to revert the entire transaction may instead finish writing a failure state under the new accounting. The actual flow determines exposure.[5][6]

Our assessment is that teams should test fixed budgets, failure accounting and retry behavior together. Simply raising a transaction's gas limit cannot establish that its logic remains safe. Readers assessing protocol readiness can look for the version tested, failure paths covered and fixes made. No reported problem is not evidence of a completed check.[3][4][5][6]

We have not run contract tests or confirmed vulnerabilities in individual protocols. This analysis uses proposals and announcements read on November 6, 2026. Specifications and upgrade arrangements may change; the checklist does not establish that every DeFi contract is affected.[1][3][4]

AI assisted with research, writing and translation. The cover is an AI-generated conceptual illustration. This article is not sponsored and is not investment advice.

Sources & further reading

(6)

When checking a claim, compare the source date, measurement period and conditions. Some linked pages change over time.

  1. EIP-8037/EIP-8038 Glamsterdam (opens in a new tab)
  2. Glamsterdam Hoodi delay (opens in a new tab)
  3. EIP-8037: State Creation Gas Cost Increase (opens in a new tab)
  4. EIP-8038: State-Access Gas Cost Update (opens in a new tab)
  5. Glamsterdam Hard Fork: When Spent Gas Comes Back (opens in a new tab)
  6. Glamsterdam Gas Invariants — pinned README (opens in a new tab)

For information and education, not investment advice. Crypto assets are volatile. Verify independently.

Author and editorial process ↗Found an error? Let us know ↗
Explore this section ↗