Consider: when `stakeMore` is called on a deposit with no validator, `_changeValidatorWeight(address(0))` will be called.

[wildmolasses]

zkstaker/src/ZkStaker.sol

Lines 414 to 427 in 3228dfa

	function _stakeMore(Deposit storage deposit, DepositIdentifier _depositId, uint256 _amount)
	internal
	virtual
	override(Staker, StakerCapDeposits)
	{
	address _depositValidator = validatorForDeposit[_depositId];
	validatorForAtomicEarningPowerCalculation = _depositValidator;
	validatorStakeWeight[_depositValidator] += _amount;
	_changeValidatorWeight(_depositValidator);
	StakerCapDeposits._stakeMore(deposit, _depositId, _amount);
	validatorForAtomicEarningPowerCalculation = address(0);
	}

what are the consequences? we should probably no-op rather than increase address(0)'s stake weight...

[funkyenough]

We should consider cases for both ZkStaker and ConsensusRegistry.

ZkStaker

• There will be unintended ValidatorTotalWeightUpdated events emitted with address(0) as the validator owner. But that can also be addressed trivially by moving the early return to before the event.
• This is not limited to _stakeMore, but also applies to _stake. Current implementation of _stake considers address(0) as a valid input. Calling _stake from Staker actually assigns address(0) to the validator param. If we are to make it no-op in _stakeMore, then we should make it no-op for _stake as well.

ConsensusRegistry

ConsensusRegistry contains logic that reverts if input address is address(0).
Therefore, even if address(0) weight exceed threshold, it should never be included in the registry.

https://github.com/matter-labs/era-contracts/blob/9bfc6a65ed0dba7b90d67a3af1792e0066846e03/l2-contracts/contracts/ConsensusRegistry.sol#L582-L586a