- Ethereum postponed Glamsterdam’s October 26 Hoodi activation and canceled the testnet’s planned 200 million gas limit increase.
- EIP-8037 separates execution gas from state-creation gas, potentially changing how deployed contracts handle failed calls.
- ChainSecurity identified 13 real-world examples requiring investigation, with responsible disclosure still underway.
- Developers should review gas-dependent failure checks, fixed gas allowances and external calls that permanently update contract records.
Ethereum’s postponed Glamsterdam upgrade has exposed a compatibility problem in the Ethereum Virtual Machine (EVM): smart contracts that rely on gas exhaustion to reject unsafe transactions may behave differently under proposed changes to state-creation costs.
The Ethereum Foundation announced on October 22 that Glamsterdam would no longer activate on the Hoodi testnet on October 26. Developers also withdrew the planned increase in Hoodi’s gas limit to 200 million.
The decision followed concerns raised during All Core Developers Execution meeting #248 about interactions between EIP-8037 and existing mainnet contracts.
ChainSecurity identified 13 real-world examples of potentially affected contracts. The projects have not been publicly named while researchers coordinate responsible disclosure.
The compatibility issue concerns contracts that interpret gas availability as evidence that an external operation succeeded, failed safely or exhausted its allocated resources.
EIP-8037 Introduces Two Separate Gas Accounting Pools
EIP-8037 changes how Ethereum accounts for the cost of creating persistent blockchain state.
Under the proposal, transactions have two accounting components: execution gas and a state-gas reservoir.
Execution gas pays for EVM computation and other execution-related operations. State gas covers additional charges associated with creating persistent blockchain data, including new storage allocations.
The state-gas reservoir initially holds the portion of a transaction’s gas budget above the execution-gas cap. State-creation charges draw from this reservoir first and use execution gas when the reservoir cannot cover the required amount.
The distinction affects Solidity’s gasleft() function.
The underlying GAS opcode reports the remaining execution gas, not the separate state-gas reservoir.
Under EIP-150, external calls generally cannot forward more than approximately 63/64 of the available execution gas. EIP-8037 applies this restriction to execution gas, while the state-gas reservoir is passed to the child call separately.
A contract can consequently have limited execution gas remaining while retaining substantial capacity for state creation.
Glamsterdam separates ordinary execution costs from charges for creating persistent blockchain state.
Why Failed Calls Could Produce Different Contract Records
A contract that attempts an external operation may continue executing when that operation fails.
Consider a batch-processing application that records whether an attempted payment or other operation was unsuccessful.
The following simplified Solidity-style example illustrates the relevant pattern:
The external call can fail while the surrounding transaction continues.
If the contract permanently marks the operation as failed, a separate retry mechanism might subsequently reject a legitimate attempt.
Some implementations have historically relied on insufficient remaining gas to prevent the failure handler from completing an expensive storage operation. In those cases, the entire transaction would revert instead.
EIP-8037, together with EIP-8038, changes the cost distribution of those storage operations.
According to ChainSecurity’s technical analysis, creating a fresh storage slot becomes substantially less expensive in execution-gas terms when the state-gas reservoir covers the separate state-creation charge.
The result can be a different transaction outcome: an external operation fails, but the caller retains enough execution gas to record that failure permanently.
Whether this is exploitable depends on the surrounding contract logic, including retry rules, accounting checks and who can influence the gas available to the external call.
The Storage Cost Change Behind the Compatibility Problem
Under the existing Osaka gas schedule, writing to a fresh, warm storage slot costs approximately 20,000 gas.
The proposed Glamsterdam rules split the cost into an execution component and a separate state-creation charge.
The execution portion falls to 10,100 gas, while state creation incurs an additional 97,920 state gas.
These amounts cannot simply be added and compared with the previous execution-gas cost because the two pools have different accounting rules.
Glamsterdam separates execution charges from state creation, changing the gas available for subsequent contract instructions.
Source: ChainSecurity’s EIP-8037 and EIP-8038 analysis. Figures represent the specified storage operation and exclude incidental instruction costs.
The proposed rules also change how gas is restored when state creation is reversed.
EIP-8037 uses a last-in, first-out refill mechanism. When eligible state-creation charges are reversed, the accounting system can replenish execution gas previously used for those charges, with any remaining amount returning to the reservoir.
Under qualifying conditions, the amount reported by gasleft() can therefore increase during execution.
A normal REVERT and an exceptional out-of-gas halt do not have identical consequences. Reverted state changes can restore applicable gas accounting, while an exceptional halt consumes the affected call frame’s execution gas.
Developers must examine the actual call path rather than assuming that every failed operation returns gas to its caller.
Three Contract Patterns Developers Should Review
The compatibility risk is concentrated in contracts that use gas measurements to make decisions or continue executing after an external call fails.
Three patterns deserve particular attention.
Illustrative Solidity patterns for code review, not confirmed vulnerabilities in every implementation.
Developers should review these patterns alongside proxies, intermediary contracts and any retry or withdrawal logic.
Tests should cover insufficient-gas failures, new storage allocation, storage clearing, intermediary reverts and transactions with different state-gas reservoir balances.
Comparing the same scenarios under Osaka and proposed Glamsterdam rules can identify execution outcomes that change unexpectedly.
Glamsterdam’s Scaling Benefits Face a Compatibility Test
Glamsterdam is designed to support higher Ethereum layer-1 capacity through coordinated execution and consensus changes.
Its broader roadmap includes EIP-7732, which introduces enshrined proposer-builder separation, and EIP-7928, which introduces block-level access lists.
EIP-8037 addresses a separate scaling constraint: persistent blockchain state imposes long-term storage costs on network participants.
By accounting for state creation independently, Ethereum aims to support greater execution capacity without allowing persistent state growth to become disproportionately cheap.
The difficulty is maintaining compatibility with contracts that already depend on established gas behavior.
A change to execution costs can affect financial application logic even when contract code and interfaces remain unchanged.
The Hoodi postponement gives core developers time to investigate the affected patterns before deciding whether to retain, modify or delay the proposed gas rules.
The Ethereum Foundation’s revised announcement instructs Hoodi node operators to update their execution and consensus clients to prevent the canceled activation.
No replacement Hoodi activation date or Glamsterdam mainnet schedule has been announced. Ethereum mainnet users do not need to take action because of the testnet postponement.
Ethereum Must Resolve the Failure-Handling Risk
The findings do not establish a widespread exploit affecting Ethereum today. They identify a compatibility risk that could become relevant if the proposed rules activate without adequate mitigation.
Changing gas thresholds alone may not be sufficient. Nested calls, reverted state creation and separate gas accounting can alter how much execution gas remains after an operation fails.
For application developers, the safer approach is to avoid treating gasleft() as definitive evidence of why an external call failed and to preserve recovery paths when financial operations cannot complete.
Glamsterdam’s next technical milestone is proving that its new gas model preserves the safety of existing contract execution paths, not simply setting another activation date.

