The approve
function in the CapitalPool
contract is designed to accept only one argument—the token address. This address is then used to approve the transfer of tokens from the CapitalPool
to the TokenManager
:
The approve
function in the CapitalPool
contract is designed to accept only one argument—the token address. This address is then used to approve the transfer of tokens from the CapitalPool
to the TokenManager
:
However, the TokenManager::_transfer
function mistakenly calls this approve
function using the TokenManager
address instead of the token address. Since the TokenManager
does not have an approve
function, the call will revert.
Any withdrawal involving a wrapped native token will fail due to the incorrect parameter being passed, leading to a transaction revert.
Manual review.
Ensure the correct argument—the token address—is passed to the approve
function:
If we consider the correct permissioned implementation for the `approve()` function within `CapitalPool.sol`, this would be a critical severity issue, because the withdrawal of funds will be permanently blocked and must be rescued by the admin via the `Rescuable.sol` contract, given it will always revert [here](https://github.com/Cyfrin/2024-08-tadle/blob/04fd8634701697184a3f3a5558b41c109866e5f8/src/core/CapitalPool.sol#L36-L38) when attempting to call a non-existent function selector `approve` within the TokenManager contract. The argument up in the air is since the approval function `approve` was made permisionless, the `if` block within the internal `_transfer()` function will never be invoked if somebody beforehand calls approval for the TokenManager for the required token, so the transfer will infact not revert when a withdrawal is invoked. I will leave open for escalation discussions, but based on my first point, I believe high severity is appropriate.
The contest is live. Earn rewards by submitting a finding.
This is your time to appeal against judgements on your submissions.
Appeals are being carefully reviewed by our judges.