Epoch Island: withdrawForfeit recomputes the forfeit on an un-scaled reward base, so partial forfeits over-repay

Source & credit. Reproduction of a public audit finding curated by AuditVault — the original finding: 59895-calling-withdrawforfeit-multiple-times-for-a-single-deposit. The historical source/toolchain is unavailable; this entry is documentation only and claims no executable Forge PoC.

Vulnerability classes: vuln/reward-accounting · vuln/missing-state-scaling

Reproduction: a faithful minimal reproduction of the vulnerable finding — the vulnerable function is reproduced verbatim (marked @>) with faithful minimal doubles; local deploy, no fork.

Root cause#

forfeitReward is computed as (earned + rewardTokensClaimed) * percentage / 1e18, but rewardTokensClaimed is never reduced as the deposit is partially forfeited. Forfeiting in multiple partial steps recomputes the forfeit against the un-scaled base each time, so the user repays more rewardToken (150 vs the fair 100) than they ever received.

SOLIDITY

    // VERBATIM vulnerable path from Vepoch.sol::withdrawForfeit().
    function withdrawForfeit(uint256 _depositId, uint256 percentage) external {
        uint256 forfeitReward = ((earned[_depositId] + rewardTokensClaimed[_depositId]) * percentage) / 1e18; // @>
        rewardToken.transferFrom(msg.sender, address(this), forfeitReward);

Why it's exploitable here#

Each partial forfeit recomputes against the full, un-reduced reward base.
rewardTokensClaimed is never decreased as the deposit is partially forfeited.
Summed over multiple partial steps, the user repays strictly more than a single full forfeit would cost.

Attack path#

flowchart TD A["User forfeits a deposit in partial steps"] --> B["forfeit = (earned + rewardTokensClaimed) * pct"] B --> C["rewardTokensClaimed never scaled down"] C --> D["Each step recomputes on the full base"] D --> E["User over-repays 50 reward tokens (150 vs 100)"]

Marked-line walkthrough (Playground)#

The EVM Playground pins each step to the exact executed source line in Vepoch:

Line 90 — VULN. the forfeit uses (earned + rewardTokensClaimed) * percentage, but rewardTokensClaimed is not scaled down per partial forfeit.
Line 91 — the computed forfeitReward is pulled from the user via transferFrom.
Line 92 — totalForfeitPaid accumulates; across multiple partial steps the user over-repays by the un-scaled excess.

PoC#

Registry (Foundry, local deploy — exploit path + a fixed-variant control):

BASH

cd 59895-calling-withdrawforfeit-multiple-times-for-a-single-deposi_exp
forge test -vv

Expected: both tests PASS — the exploit test forfeits in partial steps and asserts 50 tokens of over-repayment; the fixed version scales the base so total repayment equals the fair 100. The browser EVM Playground is served at /hacks/59895-calling-withdrawforfeit-multiple-times-for-a-single-deposi/.

Remediation#

Scale rewardTokensClaimed (and the reward base) down by the forfeited percentage so partial forfeits sum to a single full forfeit.

References#

AuditVault finding: https://github.com/Auditware/AuditVault/blob/main/findings/59895-calling-withdrawforfeit-multiple-times-for-a-single-deposit.md

Sources & further analysis#

Reproductions & code

No executable Forge reproduction is claimed; the historical source/toolchain was unavailable for this finding.
AuditVault finding: 59895-calling-withdrawforfeit-multiple-times-for-a-single-deposit.
Upstream DeFiHackLabs PoC directory: src/test.

Alerts & third-party analyses

DeFiHackLabs incident explorer: search "Epoch Island: withdrawForfeit recomputes the forfeit on an un-scaled reward base, so partial forfeits over-repay".
Web3Sec X hacked database: search.
Rekt leaderboard: search.
Solodit incident search: search.

These dashboards index community alerts tweets, post-mortems, and independent write-ups. Reach them through the protocol name above to cross-check this reproduction against other analyses.

Source-level debugger — step opcodes and Solidity in sync