0xDjango - First depositor can break minting of shares

[sherlock-admin]

0xDjango

high

First depositor can break minting of shares

Summary

The common "first depositor" vulnerability is found in InsuranceFund.depositFor(). The first account to deposit into the insurance fund can steal value from subsequent depositors by:

• Minting 1 wei shares
• Directly transferring assets into the contract to inflate the poolValue
• Subsequent depositors deposit assets but are minted 0 shares due to precision loss
• First depositor steals the assets

Vulnerability Detail

The depositor's shares are calculated via:

        if (_pool == 0) {
            shares = amount;
        } else {
            shares = amount * _totalSupply / _pool;
        }

Upon first deposit, the _pool value will be 0. The attacker will transact with an amount = 1 wei to mint 1 wei of shares. Then the attacker will transfer some value of asset directly to the contract. For this example, the attacker transfers 10,000 USDC.

Next, a subsequent depositor attempts to mint shares with 5,000 VUSD.

shares = 5000 ether * 1 wei / 10,000 ether = 0 due to precision loss.

The attacker can now withdraw the second depositor's assets.

Impact

• Theft of deposits

Code Snippet

https://github.com/sherlock-audit/2023-04-hubble-exchange/blob/main/hubble-protocol/contracts/InsuranceFund.sol#L104-L108

Tool used

Manual Review

Recommendation

Mint a certain number of shares and transfer them to address(0) within the initialize() function.

[0xshinobii]

This issue was already mentioned in code arena audit. As done in v1, this will be fixed by adding the initial amount to the insurance fund from Hubble at the time of deployment.

[hrishibhat]

@ctf-sec this issue was already reported in the previous contest. Shouldn't this be low?

[hrishibhat]

Considering this a low issue since it was already found in the previous audit.
https://github.com/sherlock-audit/2023-04-hubble-exchange#q-please-provide-links-to-previous-audits-if-any
And this is acknowledged to by the protocol.
#140 (comment)