Precision Loss

id: LS21M
title: Precision Loss 
baseSeverity: M
category: arithmetic
language: solidity
blockchain: [ethereum, optimism, arbitrum, polygon, bsc]
impact: Inaccurate payouts, rounding errors, and fund misallocation
status: draft
complexity: low
attack_vector: internal
mitigation_difficulty: easy
versions: [">=0.4.0", "<=0.8.25"]
cwe: CWE-1339
swc: SWC-131

📝 Description

• Precision loss arises when smart contracts perform arithmetic operations with inadequate decimal scaling, especially during:
• Token exchange rate calculations
• Fee or interest rate computation
• Proportional reward or dividend splits
• Since Solidity does not support floating point arithmetic, all calculations must be done using scaled integers (e.g., amount * 1e18 / rate). Omitting proper scaling or truncating during intermediate steps can lead to:
• Incorrect fund distribution
• Rounding errors that accumulate over time
• Unfair outcomes in staking, yield farming, bonding, or liquidity incentives

🚨 Vulnerable Code

pragma solidity ^0.8.0;

contract RewardVault {
    mapping(address => uint256) public shares;
    uint256 public totalShares;
    uint256 public rewardPool;

    function claimReward(address user) external {
        uint256 userShare = shares[user];
        uint256 reward = rewardPool * userShare / totalShares; // ❌ precision loss for small shares
        // Transfer reward (omitted)
    }
}

🧪 Exploit Scenario

Step-by-step exploit process:

1. A protocol has 1,000,000 total shares.
2. A new user has 1 share and is eligible for a portion of the reward.
3. The user’s share becomes permanently unrecoverable due to rounding truncation.

Assumptions:

• Dividing contributions across multiple addresses
• Timing exit/redemption when rounding errors benefit them

✅ Fixed Code

pragma solidity ^0.8.0;

contract PreciseRewardVault {
    uint256 public constant PRECISION = 1e18;

    mapping(address => uint256) public shares;
    uint256 public totalShares;
    uint256 public rewardPool;

    function claimReward(address user) external {
        uint256 userShare = shares[user];
        uint256 scaledReward = rewardPool * userShare * PRECISION / totalShares;
        uint256 reward = scaledReward / PRECISION;
        // Transfer reward (omitted)
    }
}

🧭 Contextual Severity

- context: "Default"
  severity: M
  reasoning: "Precision issues may cause unfairness or small loss without full breakage."
- context: "Public DeFi protocol with rewards, fees, or voting"
  severity: H
  reasoning: "Critical in staking, DAO voting, and fee splits—may lead to measurable financial loss."
- context: "Private utility contract with limited scope"
  severity: L
  reasoning: "Impact is minor, affecting only non-critical outcomes or simulations."

🛡️ Prevention

Primary Defenses

• Use consistent PRECISION constants for fractional logic
• Always multiply before divide when calculating ratios
• Unit test edge cases (very small / very large ratios)

Additional Safeguards

• Consider libraries like PRBMath or FixedPoint
• Emit warnings or cap minimum values when precision loss is unavoidable

Detection Methods

• Static analysis for integer division without prior scaling
• Tools: Slither (divide-before-multiply, truncation), custom lint rules

🕰️ Historical Exploits

• Name: Synthetix SNX Rewards Distribution Drift
• Date: 2019-10
• Loss: ~$7,000
• Post-mortem: Link to post-mortem

📚 Further Reading

• SWC-131: Incorrect Calculation
• CWE-1339: Precision Errors
• Solidity Docs – Integer Division

---

✅ Vulnerability Report

id: LS21M
title: Precision Loss 
severity: M
score:
impact: 3      
exploitability: 3 
reachability: 5   
complexity: 2    
detectability: 5  
finalScore: 3.4

---

📄 Justifications & Analysis

• Impact: Users may be underpaid or overpaid due to rounding artifacts
• Exploitability: Possible if attacker times edge cases or fragments positions
• Reachability: Common across staking, bonding, dividend, and token exchange logic
• Complexity: Simple math mistake, often in low-attention areas
• Detectability: Clear in unit tests or visual audits of amount * X / Y patterns