Check for stale stakes

[taco-paco]

eigenlayer-middleware/src/BLSSignatureChecker.sol

Line 190 in 3d70abd

registryCoordinator.quorumUpdateBlockNumber(uint8(quorumNumbers[i])) + withdrawalDelayBlocks >= referenceBlockNumber,

Hi,
Here the check assumes that update was triggered within RegistryCoordinator::updateOperatorsForQuorum but stakes could also be update from RegistryCoordinator::updateOperators. Shouldn't this be checked against StakeRegistry::operatorStakeHistory?

[samlaf]

Don't have full context here but will give you my partial answer to move this along, and leave those with more context (@8sunyuan ?) to give you a full answer.

I think the goal here is just to make sure that the FULL operator set (for a quorum) has been updated within a given period (which is what updateOperatorsForQuorum does). The goal is not to make sure no one has called the update function in the given period, but oppositely to make sure someone did call it during some bounded period of time.

updateOperators is meant to be called by operators themselves, who are willing to pay the gas costs to update their position in the AVS (perhaps because they received a lot of deposits and don't want to wait for the next full quorum update). It's fine if it happens, and doesn't affect the checkSignature logic because we are referring to a snapshop of operatorState at a given block in the past (before the update).