Strict enforcement of chainId

Abstract

Reject transactions that do not explicitly have the same chainId as the node's configuration.

Motivation

Per EIP-155 a transaction with a chainId = 0 is considered to be a valid transaction. This was a feature to offer developers the ability to submit replayable transactions across different chains. With the rise of evm compatible chains, many of which use forks, or packages from popular Ethereum clients, we are putting user funds at risk. This is because most wallet interfaces do not expose the chainId to the user, meaning they typically do not have insight into what chainId they are signing. Should a malicious actor (or accidental) choose to, they can easily have users submit transactions with a chainId = 0 on a non-mainnet network, allowing the malicious actor to replay the transaction on ethereum mainnet (or other networks for that matter) as a grief or sophisticated attack.

Specification

As of the fork block N, consider transactions with a chaindId = 0 to be invalid. Such that transactions are verified based on the nodes configuration. Eg:

if (node.cfg.chainId != tx.chainId) {
    // Reject transaction
}

Rationale

The configuration set by the node is the main source of truth, and thus should be explicitly used when deciding how to filter out a transaction. This check should exist in two places, as a filter on the JSON-RPC (eg: eth_sendTransaction), and strictly enforced on the EVM during transaction validation.

This ensures that users will not have transactions pending that will be guaranteed to fail, and prevents the transaction from being included in a block.

Backwards Compatibility

This breaks all applications or tooling that submit transactions with a chainId == 0 after block number N.

Test Cases

TBD

Security Considerations

It should be noted this will not prevent a malicious actor from deploying a network with chainId = 1, or copying any other networks chainId.

Copyright

Copyright and related rights waived via CC0.