Ethereum core developers are actively seeking urgent assistance from application teams to help determine whether crucial proposed gas modifications within the upcoming Glamsterdam upgrade require substantial revision. The outreach stems from the sheer scale of the ecosystem, as internal testing frameworks and developer audits cannot possibly cover every single smart contract, decentralized application, and off-chain service deployed across the network.
The appeal was formally launched through a post on the Ethereum Foundation blog by core developer Marius van der Wijden. The communication urges developers, protocol engineers, and independent security auditors to meticulously analyze eight distinct gas assumptions that will be directly impacted by the proposed network upgrade. Prominent security auditing firms ChainSecurity and Dedaub have been actively investigating these alterations alongside internal development teams. However, core developers have emphasized that a comprehensive risk assessment requires individual projects to step forward and explain precisely how they would identify, isolate, and repair affected systems if the upgrade goes live in its current form.
This collaborative push for external feedback follows a recent, precautionary network delay. The planned activation of the Hoodi testnet, which was scheduled to include a significant 200 million gas-limit increase, was officially postponed. The decision to halt the testnet deployment was made to address mounting technical concerns regarding how the upgrade handles resource accounting and fee management. Van der Wijden’s follow-up appeal makes it clear that this delay will remain in effect until these complex issues are fully resolved. Consequently, any concrete timeline for mainnet activation has been left entirely unset while developers gather data and evaluate potential remedies.
The potential consequences of the proposed gas alterations are far-reaching and touch multiple layers of the Ethereum stack. They extend from automated transaction relayers recovering far too little of their underlying execution costs to wallet authorizations inadvertently becoming reusable after specific types of transaction failures. Because these technical nuances can threaten user funds and disrupt relayer economics, the practical repairability of decentralized applications has become a core piece of evidence that developers must weigh when deciding whether to adjust, delay, or rewrite parts of the upgrade specifications.
At the center of the Glamsterdam upgrade are two major Ethereum Improvement Proposals. EIP-8037 is designed to fundamentally alter how the network accounts for persistent storage by raising the cost of creating new blockchain data. It introduces a mechanism to meter state growth completely separately from standard execution costs. Meanwhile, EIP-8038 is structured to reprice access and write operations directed at existing blockchain state. Together, these proposals aim to support higher transaction throughput across the network while exercising tighter control over the total volume of data that Ethereum validator nodes must permanently retain.
To systematically evaluate the security implications of these changes, developers established an eight-invariant framework. This analytical structure neatly separates three price assumptions, which primarily cover fixed budgets and structural reserves, from five accounting assumptions that govern gas returns, failure handling, billing mechanics, and code delegation.
Gas accounting changes payments and wallet safety
The complexities of the new gas accounting rules introduce significant challenges for automated financial systems, particularly regarding transaction relayers. A relayer typically pays upfront to execute a transaction on behalf of an end user and subsequently recovers that exact expense by measuring the remaining gas before and after the computational work is completed. However, the framework’s fifth entry, known as INV-5, highlights how the proposed accounting model can severely disrupt that vital calculation.
Under the rules outlined in EIP-8037, a transaction automatically funds a separate state-gas reservoir whenever its initial gas limit exceeds 16,777,216 units. When state creation operations occur, they draw resources directly from that dedicated state pool first before touching any remaining execution gas. A critical complication arises because Solidity’s standard gasleft() function reading deliberately excludes this underlying reservoir. As a result, simply subtracting two gas readings taken before and after a transaction can completely miss substantial costs that the relayer actually paid out of pocket.
If a later gas reading happens to exceed an earlier one due to these structural shifts, standard checked subtraction routines will automatically revert the transaction, completely blocking the operation. Conversely, using unchecked subtraction can cause the calculation to wrap around into an enormous number, potentially producing excessive fee reimbursement depending on the surrounding arithmetic and payment validation logic. At the same time, invisible reservoir spending can easily leave end users undercharged while protocol operators and relayers are left quietly covering the financial shortfall.
To mitigate these risks, the current billing guidance suggests that applications should agree upon a user-signed fee derived from an accurate estimate under the newly established rules. Where dynamic metering remains strictly necessary, clamping the subtraction result and capping the final payment at a signed maximum can help limit potential overcharges, even though reservoir-funded costs will continue to sit entirely outside standard gas measurements.
Meanwhile, an open specification proposal designated as pull request #12581 has been floated to prevent gas readings from artificially increasing. While that specific modification would effectively address the hazardous subtraction problem, it would leave the fundamental billing blind spot of the state reservoir unaddressed.
A closely related vulnerability identified within the framework is INV-8, which examines the mechanics of EIP-7702 delegation. This proposal allows a standard user account to temporarily authorize delegated contract code to execute actions on its behalf. Under the newly proposed rules, if a transaction encounters insufficient gas during the pre-execution phase, the network may still include the transaction within a block while simultaneously rolling back its delegation and authorization-related updates to the account’s sequence number, or nonce.
This scenario creates a dangerous window of vulnerability. An exposed authorization that continues to satisfy its basic validity checks—including the authorizing account’s unchanged nonce—can potentially be captured and reused by malicious actors. If the delegated smart contract also features an initialization function that permits the very first caller to designate themselves as the owner, an attacker could exploit this loophole to apply the authorization and seize ownership of the account within a single transaction.
While authorization front-running is already a known phenomenon within public mempools, this particular issue introduces a novel risk where a failed, mined transaction can expose a reusable authorization even when the transaction was submitted privately. Furthermore, delegations that were successfully applied prior to the top-level execution frame remain active even when that later frame ultimately fails, requiring developers to implement rigorous secondary checks on an account’s resulting code and nonce.
Proposed wallet safeguards to counter this risk include binding contract initialization routines directly to the account’s cryptographic signature or mandatory self-calls, alongside mechanisms to actively invalidate exposed authorizations. While charging the absolute worst-case authorization cost upfront has been suggested as a potential protocol-level change, that specific remedy remains unadopted.
Repair options enter the upgrade decision
The economic implications of these changes are stark when evaluated against previous network upgrades. Under the proposed schedule, creating a cold storage slot will cost 110,020 total gas, a sharp increase from the 22,100 gas required under the preceding Osaka upgrade. Out of that proposed total, 12,100 represents standard execution gas, while the remaining 97,920 is classified as state gas, with the state reservoir dictating which balance ultimately funds that massive portion.
Research published by security firm ChainSecurity illustrates how these altered cost structures and modified gas returns can allow a smart contract to permanently record an underfunded operation as definitively unsuccessful. If that permanent record accidentally removes the application’s internal retry path, a completely valid user operation can become permanently impossible to repeat or recover.
To combat this, protocol engineers are urging projects to evaluate whether revalidated budgets, improved failure propagation, and robust retry logic can successfully neutralize these risks. The core development repository is actively asking affected projects whether a simple configuration setting, a smart contract upgrade, or a comprehensive data migration can resolve their specific compatibility issues, and approximately how long such fixes would take to implement.
These practical answers will help core developers accurately gauge how much compatibility work decentralized applications can reasonably absorb, as well as pinpointing exact areas where further specification changes could prevent widespread damage. Concrete reports detailing what breaks in real-world environments and how those systems can be repaired will directly inform the next rounds of technical specification discussions for the Glamsterdam upgrade.
