forge script nonce issue related to selectFork

[sakulstra]

Component

Forge

Have you ensured that all of these are up to date?

• Foundry
• Foundryup

What version of Foundry are you on?

forge 0.2.0 (deae4f1 2023-10-11T00:27:48.322843000Z)

What command(s) is the bug in?

forge script

Operating System

macOS (Apple Silicon)

Describe the bug

Hello, at aave governance v3 we have a script that:

1. iterates trough multiple chains to check for some configuration correctness
2. submits a single txn on mainnet

While when testing everything works/worked fine, when actually submitting via ledger we're getting nonce issues like:

nonce changed unexpectedly while sending transactions. Expected 2007 got 2006 from provider.

The issue seems to happen whenever we do selectFork within a script.

Here i created a minimal reproduction:

// SPDX-License-Identifier: MIT
pragma solidity ^0.8.0;
import {Script} from 'forge-std/Script.sol';

contract Reproduction is Script {
  function run() public {
    uint256 previousFork = vm.activeFork();
    vm.selectFork(previousFork);
    vm.startBroadcast();
    address(0x25F2226B597E8F9514B3F68F00f494cF4f286491).call{value: 1}(new bytes(0));
  }
}

The script is started with a mainnet rpc, calls selectFork witht he activeForkId which is 0 and i think should essentially just be a noop in that case?.

The trace looks as follows:

Traces:
  [41465] Reproduction::run() 
    ├─ [0] VM::activeFork() [staticcall]
    │   └─ ← 0
    ├─ [0] VM::selectFork(0) 
    │   └─ ← ()
    ├─ [0] VM::startBroadcast() 
    │   └─ ← ()
    ├─ [7180] 0x25F2226B597E8F9514B3F68F00f494cF4f286491::fallback{value: 1}() 
    │   ├─ [55] 0x10c74b37Ad4541E394c607d78062e6d22D9ad632::fallback() [delegatecall]
    │   │   └─ ← ()
    │   └─ ← ()
    └─ ← ()
    
Script ran successfully.

Instead of being able to submit the txn though, i'm getting:

###
Finding wallets for all the necessary addresses...
##
Sending transactions [0 - 0].

Error: 
EOA nonce changed unexpectedly while sending transactions. Expected 46 got 45 from provider.

Seems related:

• Multichain deployment gives EOA nonce changed unexpectedly while sending transactions #4701

[sakulstra]

edit: updated the example to be even more minimal (initially though was related to multichain, but just calling selectFork(activeForkId) will reproduce the same issue.)

In practice what we do is:

• store the current forkId
• switch to another fork
• test some stuff
• switch back to the previous fork
  The example is simplified, as it seems like most of this is unrelated to the actual error.

---

Also, for anyone facing this, one workaround which seems to work for us is to simply not switch back to the fork but always create a new one:

- uint256 previousFork = vm.activeFork();
- vm.selectFork(previousFork);
+ vm.createSelectFork(vm.rpcUrl('mainnet'));

[ArshanKhanifar]

I've proposed a fix here. #6271

[RedaOps]

@sakulstra @Evalir can close this as completed since #6271 is merged?

[sakulstra]

@RedaOps i think it's working fine now thx.