Incorrect block hash when forking other EVM chains

[leduythuccs]

When forking other EVM chains (like Avax, Avax testnet, Aurora, Polygon, ...), the blockhash getting from eth_getBlockByNumber is not match with the RPC, the reason is:

• In this: ForkBlockchain.ts#L272, hardhat use the blockdata to create a new Block instance. Block instance will use Ethereum formula to calculate the block hash.
• On Avax (and many other chains), they don't follow Ethereum's block hash formula so the calculated hash result will differ from the actual hash from the RPC.

---

Reproduce this:

• yarn hardhat node --fork https://api.avax-test.network/ext/bc/C/rpc
• Fetch a random-block information:

curl -X GET \
  'localhost:8545' \
  --data-raw '{
  "method": "eth_getBlockByNumber",
  "params": [
    "0xC5CFC8",
    false
  ],
  "jsonrpc": "2.0",
  "id": 44
}'

• That block is here: https://testnet.snowtrace.io/block/12963784, its hash is 0x454706c3214e4c4c1336dce67b57d608d6feca590a882922b2e0ca440e075bbd while hardhat is returning 0xbd00f86284b05bf7b69f04a0c3186fc548692d3664f900389487d9199957afbb

---

How to fix

• I found a hacky fix is that we can override the .hash() function of the block instance, like what i did in this commit

[github-actions[bot]]

This issue is also being tracked on Linear.

We use Linear to manage our development process, but we keep the conversations on Github.

LINEAR-ID: 16b29507-4f6d-4de8-8804-d6ea49d712f4

[leduythuccs]

This issue is not a problem for most of the users tho. I only realized this when trying to run a subgraph with a local hardhat fork. Subgraph verifies the block hash in order to detect re-org.

[fvictorio]

Hi @leduythuccs, thanks for reporting this.

This is a hard one because ethereumjs doesn't provide any way to set the hash. Monkey-patching is always an option, of course, but I don't think the risk is justified in this case. In any case, we really don't want to have that unnecessary hash computation (we should just use whatever the remote node returned).