IDS06-J. Exclude user input from format strings

Interpretation of Java format strings is stricter than in languages such as C [Seacord 2005]. The standard library implementations throw appropriate exceptions when any conversion argument fails to match the corresponding format specifier. This approach reduces opportunities for malicious exploits. Nevertheless, malicious user input can exploit format strings and can cause information leaks or denial of service. As a result, strings from an untrusted source should not be incorporated into format strings.

Noncompliant Code Example

This noncompliant code example demonstrates an information leak issue. It accepts a credit card expiration date as an input argument and uses it within the format string.

class Format {
  static Calendar c = 
   new GregorianCalendar(1995, GregorianCalendar.MAY, 23);
  public static void main(String[] args) {  
    // args[0] is the credit card expiration date
    // args[0] can contain either %1$tm, %1$te or %1$tY as malicious
    // arguments
    // First argument prints 05 (May), second prints 23 (day) 
    // and third prints 1995 (year)
    // Perform comparison with c, if it doesn't match print the 
    // following line
    System.out.printf(args[0] + 
    " did not match! HINT: It was issued on %1$terd of some month", c);
  }
}

In the absence of proper input validation, an attacker can determine the date against which the input is being verified by supplying an input that includes one of the format string arguments %1$tm, %1$te, or %1$tY.

Compliant Solution

This compliant solution ensures that user-generated input is excluded from format strings.

class Format {
  static Calendar c = 
    new GregorianCalendar(1995, GregorianCalendar.MAY, 23);
  public static void main(String[] args) {  
    // args[0] is the credit card expiration date
    // Perform comparison with c, 
    // if it doesn't match print the following line
    System.out.printf("%s did not match! "
        + " HINT: It was issued on %1$terd of some month", args[0],c);
  }
}

Risk Assessment

Allowing user input to taint a format string may cause information leaks or denial of service.

Automated Detection

Static analysis tools that perform taint analysis can diagnose some violations of this rule.

Related Guidelines

Bibliography

IDS05-J. Use a subset of ASCII for file and path names            IDS07-J. Do not pass untrusted, unsanitized data to the Runtime.exec() method