How to verify that a smart contract was deployed by us?

Question

I would like a method called from Solidity to either

1. verify the hash of an entire smart contract's code, or
2. read some immutable EVM property that contracts expose that say who deployed them, e.g. a factory contract that we deployed earlier

The use case is that I want to have an addHook(contractAddress, methodSignature) function that other contracts can call to be notified when something changes (such as new statistics or voting) occurs. But we can't allow adding JUST ANY contract and method as a hook handler, because they can simply call while(1) and destroy our contract.

However, we do want to be able to addHook() after a contract is constructed, but only call methods of contracts which we ourselves wrote, and hardcode either 1 or 2 above in our contract which exposes addHook. The contract's owner property won't do, since it's not immutable, so another contract can fake it.

Answer 1

read some immutable EVM property that contracts expose

Hmmm ... not sure what you read. There is certainly immutable blockchain info about the account that signed the transaction (or the contract) that deployed the contract but I think that is like events and not accessible from Solidity. Possibly someone else will chime in with a method.

A practical solution to your problem lies in this requirement:

of contracts which we ourselves wrote

That simplifies matters because you only need to keep track of what you do.

For example, in a hub and spoke pattern with a contract factory acting as the hub, it can keep track of deployed contracts. Supposed the hub has:

mapping(address => bool) public isTrusted;

Then it only needs to implement access control to limit access to "trusted" contracts:

modifier onlyTrusted {
  require(isTrusted[msg.sender], "I don't know you.");
  _;

Possibly the Hub isn't the only contract concerned with access control because the topology is more intricate. Still, you just need each interested contract to know the source of authority and ask:

modifier onlyTrusted {
  require(whiteListContract.isTrusted(msg.sender), "It doesn't know you.");
  _;
}

There are a few standardized contracts that implement this sort of thing. I'd suggest checking out Open Zeppelin "Whitelist" and "AccessControl." They each have methods to add and remove addresses from the list of things that are trusted. The maintenance of the lists is itself a function that deserves access control of its own (onlyOwner?) but it can work in a decentralized context, e.g. only trust contracts that this factory made itself:

function deployNewContract() public returns(address) { // anyone can
  MyThing t = new MyThing();
  // now add address(t) to the whitelist, depending on which implementation you choose
  return address(t);
}

You know t is a certain kind of contract with no unexpected behaviour because of where it came from.

Hope it helps.