MorpheusAI

Summary

Vulnerability Details

L2MessageReceiver.sol::retryMessage() checks the transaction has a failed message as follow:

Then, goes ahead to call the _nonblockingLzReceive before deleting the payload from the failed message:

However, in the _nonblockingLzReceive() there is an external call to the token contract to mint to the user

If this token is changed in the future to a token that uses a _safeMint or implements its before/afterTransfer callbacks, then an attacker would be able to receive an infinite amount of reward by reentering the retryMessage before the failedMessages[senderChainId_][senderAndReceiverAddresses_][nonce_].

Proof Of Concept:

Attack Scenario:

• Users deposit into the protocol and are now eligible for rewards

• Time passes and protocol makes changes to the rewardToken address to a contract that uses _safeMint in the mint function call

• An attacker notices this and deploys a contract that deposits into the protocol and later calls claim() which sends the mint message to the l2

• lzReceive gets triggered on l2 and call to IL2MessageReceiver(address(this)).nonblockingLzReceive() fails

• The attacker then calls retryMessage which mints to the attacker contract, however, the payload isn't yet before making the external call.

• So the attacker contract makes call into the retryMessage and performs the same thing again.

Impact

If this token is changed in the future to a token that uses a _safeMint or implements its before/afterTransfer callbacks, then an attacker would be able to receive an infinite amount of reward by reentering the retryMessage before the failedMessages[senderChainId_][senderAndReceiverAddresses_][nonce_].

Recommendations

Make this change