An integer underflow (or numeric underflow in general) happens when an arithmetic operation results in a numeric value that is outside of the range for the represented datatype (integer, long, double, etc). The arithmetic output wraps around to a wrong value. This causes a non-deterministic program behaviour such as program crash, validation bypass or denial of service.

Remediation

• Use the arithmetic properties to check for overflow condition. For example, given positive x and y, x + y is always equal or bigger than x or y. When overflow happens, the result of addition does not follow this math property.
• Use a dynamic data type (e.g. BigInteger) for boundary check before assigning the value to the fixed data type (e.g. Integer).
• Refer to SecDim's Numeric Overflow short course for more information.