Add new tests based on messages like SettleDealPaymentsExported incident that takes up lots of gas

[parthshah1]

Based on recent incidents based on legit messages taking up whole block space kicking out legit messages like WindowPoST.
Refer to https://filecoinproject.slack.com/archives/C09BS7MR7MW for more context and add such stress messages to the harness to assert basefee increase.

More context to follow once more work is done.

Tracking filecoin-project/FIPs#1224 for more updates.

[parthshah1]

this is summary of the issue (used LLM for summary):

The core issue is a divergence between Physical Resource Consumption (Block Limits) and Protocol Economic Logic (Base Fee Calculation).The Mechanism: Filecoin uses Tipsets (sets of blocks at the same height) for consensus. To accurately measure network throughput, the Base Fee algorithm ignores duplicate messages found in sibling blocks. If 5 miners include Message A, the protocol counts Message A's gas usage only once when calculating network utilization.The Incentive misalignment:Miner Strategy (Greedy): Miners run a greedy knapsack algorithm (SelectMessages). They sort the mempool by GasPremium (priority fee).The Scenario: A message appears with massive GasLimit (e.g., 7B gas, ~70% of a block) and an excessively high GasPremium.The Result: Every winning miner in the epoch independently selects this message because it is the most profitable.The Paradox:Physical Layer: 3 blocks are produced. All 3 are 70% full of the same data. The network is physically congested; other transactions are crowded out.Economic Layer: The Base Fee calculator sees 1 unique message. It calculates usage as: $\frac{7 \text{ billion gas}}{3 \text{ blocks} \times 10 \text{ billion capacity}} \approx 23% \text{ utilization}$.Outcome: Since $23% < 50%$ (the target utilization), the Base Fee decreases, despite the blocks being physically full. This creates a feedback loop where spamming the network remains cheap.

We need a message that:

Consumes ~70-90% of BlockGasLimit (standard is 10B, so target 7B-9B).

Pays a GasPremium significantly higher than the current average to guarantee selection by greedy algorithms in Lotus, Forest, and Curio.

Just an idea
Tools needed:

Contract: A simple "Gas Waster" FEVM contract (infinite loop).

Script: Python script to broadcast the transaction.

The FEVM Contract (Solidty):

// SPDX-License-Identifier: MIT
pragma solidity ^0.8.17;

contract Congestion {
    // 1. Burns CPU (Gas) without changing state (matches SettleDealPayments profile)
    function wasteGas(uint256 iterations) public {
        uint256 startGas = gasleft();
        // A simple loop to burn gas
        for (uint256 i = 0; i < iterations; i++) {
             // Doing nothing, just burning cycles
             assembly {
                 let x := mload(0x40)
             }
             // Safety break if we are about to run out (leave 1M gas)
             if (gasleft() < 1000000) {
                 break;
             }
        }
    }
}

C. The Simulation Script (Python)
Use this script to inject the message into your devnet. It pushes the message to all nodes to bypass propagation latency issues in a small devnet.

Python

import requests
import json
import time

# RPC Endpoints for your mix of nodes
NODES = [
    "http://127.0.0.1:1234/rpc/v1", # Lotus 1
    "http://127.0.0.1:2345/rpc/v1", # Lotus 2
    "http://127.0.0.1:3456/rpc/v1", # Forest
    # Curio usually sits behind a Lotus node API or has its own if configured
]

def rpc(url, method, params):
    payload = {"jsonrpc": "2.0", "method": method, "params": params, "id": 1}
    try:
        return requests.post(url, json=payload).json()['result']
    except Exception as e:
        print(f"Failed to connect to {url}: {e}")
        return None

def trigger_congestion():
    # 1. Prepare the Signed Message (Do this ONCE so the CID is identical)
    # Use 'WalletSign' on one node to get a raw signature.
    # Note: Construct the message manually or use a library to ensure strict control over GasLimit
    
    # Pseudocode for constructing the message structure
    msg = {
        "To": "t01000...", # Address of your deployed GasWaster contract
        "From": "t01234...",
        "Value": "0",
        "Method": 2, # Method ID for your contract call
        "Params": "...", # Encoded params for wasteGas()
        "GasLimit": 9000000000, # 9 Billion (Huge!)
        "GasFeeCap": "1000000000000", # Cap
        "GasPremium": "50000000000",  # VERY HIGH PREMIUM to force inclusion
        "Nonce": 10 # Ensure this is correct
    }
    
    # Sign it on Node 1
    signed_msg = rpc(NODES[0], "Filecoin.WalletSign", [msg["From"], msg])
    
    print(f"Broadcasting Malicious Message CID: {signed_msg['Cid']}")

    # 2. Broadcast to ALL nodes simultaneously (MpoolPush)
    # This ensures Forest, Curio, and Lotus all see it instantly.
    for node in NODES:
        rpc(node, "Filecoin.MpoolPush", [signed_msg])
        print(f"Pushed to {node}")

    print("Monitoring next epoch for duplicate inclusion...")


if __name__ == "__main__":
    trigger_congestion()

4. APIs to Track Base Fee & Congestion
   You need to demonstrate that ParentBaseFee is dropping (or stable) while BlockGasUsed is full.

A. Poll Chain Head (Real-time) Check the ParentBaseFee and the Blocks in the tipset.

Bash
`lotus chain head | jq '{Height: .Height, BaseFee: .ParentBaseFee, BlockCount: .Blocks|length}'