Submitted by EV_om
Original Issue: The owners of a rental safe can continue to use the old guard policy contract for as long as they want, regardless of a new guard policy upgrade
https://github.com/re-nft/smart-contracts/blob/97e5753e5398da65d3d26735e9d6439c757720f5/src/policies/Guard.sol#L399-L407
The original vulnerability constituted the ability of rental safe owners to indefinitely use an outdated guard policy contract, even after a newer version has been deployed. This situation arises because the protocol lacked a mechanism to enforce the update of guard policies across all safes, potentially leaving some safes operating under less secure or outdated policies.
The mitigation introduces a protocol-wide capability to deactivate guard policies, aiming to prevent the continued use of outdated or insecure policies. However, this solution inadvertently introduced a new vulnerability: the potential for bricking rental safes if their associated guard policy is deactivated without providing a viable path for these safes to update to a new guard policy.
The only way for a safe owner to update the guard, if the one they are using becomes inactive, would be via a whitelisted DelegateCall, in the same manner as they would upgrade the Stop policy as described in the audit documentation. However, the current implementation triggers a revert when a Guard policy is not active, before they can potentially execute a DelegateCall to upgrade the guard.
Allow inactive guards to call whitelisted delegates. A potential implementation could look as follows:
If this issue is addressed as suggested, it is important to keep in mind that disabled guards will also be able to call whitelisted delegates. It is crucial to ensure that this capability does not introduce any vulnerability.
While the mitigation strategy addresses the original vulnerability of outdated guard policies, it introduces a new risk of rendering rental safes inoperative.
Alec1017 (reNFT) confirmed and commented:
