Summary

The _smoothed_price function calculates max_change using self.max_price_increment, which is initialized as 2 * 10**12 (10^12 precision) but can be set between 10**8 and 10**18 via set_max_price_increment. The formula max_change = (self.max_price_increment * (block.timestamp - self.last_update) * last_price) // 10**18

assumes last_price has 10^18 precision and divides by 10**18 to normalize, but self.max_price_increment’s variable precision (10^8 to 10^18) disrupts this alignment, leading to inconsistent max_change values that don’t match the intended smoothing behavior.

Vulnerability Details

If self.max_price_increment is set to 10**8, the calculated max_change will be much smaller than expected.

• If self.max_price_increment is set to 10**18, the division by 10**18 ensures correct scaling, but this does not hold for all values between 10**8 and 10**18. (e.g 10**12)

• This could cause unexpectedly low or high max_change values, leading to incorrect price smoothing behavior.

self.max_price_increment’s precision (10^8 to 10^18) directly affects max_change’s scale.

• Initial value (2 * 10**12) is 10^6 times smaller than 10^18, reducing max_change’s magnitude significantly compared to an intended 10^18-based rate.

• The comment suggests an exponential form (max_price_increment / 10**18) ** time, implying max_price_increment should be a rate normalized to 10^18, but the linear formula doesn’t enforce this.

Impact

• Incorrect Smoothing: When self.max_price_increment is less than 10^18 (e.g., 2 * 10**12), max_change is scaled down by a factor of self.max_price_increment / 10**18 (e.g., 2 * 10^-6), making price changes much smaller than intended, potentially freezing prices or under-smoothing.

• Over-Smoothing Risk: If set to 10**8, the effect is 10^10 times smaller, rendering smoothing ineffective.

• Behavioral Variability: Different admins setting different values (e.g., 10**12 vs. 10**18) cause unpredictable smoothing, undermining reliability.

Tools Used

Manual

Recommendations

enforce a fixed precision for self.max_price_increment (e.g., always store it as 10**18-scaled values).