Many programs accept untrusted data originating from unvalidated users, network connections, and other untrusted sources and then pass the (modified or unmodified) data across a trust boundary to a different trusted domain. Frequently the data is in the form of a string with some internal syntactic structure, which the subsystem must parse. Such data must be sanitized, both because the subsystem may be unprepared to handle the malformed input and because unsanitized input may include an injection attack.

In particular, programs must sanitize all string data that is passed to command interpreters or parsers so that the resulting string is innocuous in the context in which it will be parsed or interpreted.

Many command interpreters and parsers provide their own sanitization and validation methods. When available, their use is preferred over custom sanitization techniques because homegrown sanitization can often neglect special cases or hidden complexities in the parser. Another problem with custom sanitization code is that it may not be adequately maintained when new capabilities are added to the command interpreter or parser software.

SQL Injection

An SQL injection vulnerability arises when the original SQL query can be altered to form an altogether different query. Execution of this altered query may result in information leaks or data modification. The primary means of preventing SQL injection are sanitizing and validating untrusted input and parameterizing the query.

Suppose a database contains user names and passwords used to authenticate users of the system. The user names have a string size limit of 8. The passwords have a size limit of 20.

An SQL command to authenticate a user might take the form:

SELECT * FROM db_user WHERE username=<USERNAME> AND password=<PASSWORD>

If it returns any records, the user name and password are valid.

However, if an attacker can substitute any strings for <USERNAME> and <PASSWORD>, they can perform an SQL injection by using the following string for <USERNAME>:

validuser' OR '1'='1

When injected into the command, the command becomes:

SELECT * FROM db_user WHERE username='validuser' OR '1'='1' AND password=<PASSWORD>

If validuser is actually a valid user name, this SELECT statement will select the validuser record in the table. The hashed password is never checked because the expression '1'='1' is always true. Consequently the attacker is granted the access of validuser.

To comply with rule MSC05-J. Store passwords using a hash function, the passwords must be hashed. Unfortunately, many small systems fail to comply; consequently, the password text added in the query string would precisely match what the user enters. An attacker could supply a string for <PASSWORD> such as:

' or '1'='1

This would yield the following command:

SELECT * FROM db_user WHERE username='anything' AND password='' OR '1'='1'

This time, the '1'='1' tautology disables both user name and password validation, and the attacker is falsely logged in without knowing a login ID or password.

Noncompliant Code Example

This noncompliant code example shows JDBC code to authenticate a user to a system. The password is passed as a char array, the database connection is created, and then the passwords are hashed to comply with MSC05-J. Store passwords using a hash function and MSC10-J. Limit the lifetime of sensitive data.

Unfortunately, this code example permits an SQL injection attack because the SQL statement sqlString accepts unsanitized input arguments. The attack scenario outlined above would work as described.

class Login {
  public Connection getConnection() throws SQLException {
    DriverManager.registerDriver(new com.microsoft.jdbc.sqlserver.SQLServerDriver());
    String dbConnection = PropertyManager.getProperrty("db.connection");
    // can hold some value like "jdbc:microsoft:sqlserver://<HOST>:1433,<UID>,<PWD>"
    return DriverManager.getConnection(dbConnection);
  }

  String hashPassword(char[] password) {
    // create hash of password
  }

  public void doPrivilegedAction(String username, char[] password) throws SQLException {
    Connection connection = getConnection();
    if (connection == null) {
      // handle error
    }
    String pwd = hashPassword(password);

    String sqlString = "SELECT * FROM db_user WHERE username = '" + username +
                       "' AND password = '" + pwd + "'";
    Statement stmt = connection.createStatement();
    ResultSet rs = stmt.executeQuery(sqlString);

    if (!rs.next()) {
    	 throw new SecurityException("User name or password incorrect");
    }

    // Authenticated; proceed
  }
}

Compliant Solution (PreparedStatement)

Fortunately, the JDBC library provides an API for building SQL commands that sanitize untrusted data. The java.sql.PreparedStatement class properly escapes input strings, preventing SQL injection when used properly. This is an example of component-based sanitization.

This compliant solution uses a PreparedStatement instead of java.sql.Statement. This code also validates the length of the username argument, preventing an attacker from submitting an arbitrarily long user name.

class Login {
  public void doPrivilegedAction(String username, char[] password) throws SQLException {
    Connection connection = getConnection();
    if (connection == null) {
      // handle error
    }
    String pwd = hashPassword(password);

    // Ensure that the length of user name is legitimate
    if ((username.length() >= 8) {
      // Handle error
    }

    String sqlString = "select * from db_user where username=? and password=?";
    PreparedStatement stmt = connection.prepareStatement(sqlString);
    stmt.setString(1, username);
    stmt.setString(2, pwd);
    ResultSet rs = stmt.executeQuery();
    if (!rs.next()) {
      throw new SecurityException("User name or password incorrect");
    }

    // Authenticated; proceed
  }
}

Use the set*() methods of the PreparedStatement class to enforce strong type checking. This mitigates the SQL injection vulnerability because the input is properly escaped by automatic entrapment within double quotes. Note that prepared statements must be used even with queries that insert data into the database.

XML Injection

Because of its platform independence, flexibility, and relative simplicity, the extensible markup language (XML) has found use in applications ranging from remote procedure calls to systematic storage, exchange, and retrieval of data. However, because of its versatility, XML is vulnerable to a wide spectrum of attacks. One such attack is called XML Injection.

A user who has the ability to add structured XML as input can override the contents of an XML document by injecting XML tags in data fields. These tags are interpreted and classified by an XML parser as executable content and as a result may cause certain data members to be overridden.

Consider the following XML code snippet from an online store application, designed primarily to query a back-end database. The user has the ability to specify the quantity of an item available for purchase.

<item>
  <description>Widget</description>
  <price>500.0</price>
  <quantity>1</quantity>
</item>

A malicious user might input the following string instead of a simple number in the quantity field.

1</quantity><price>1</price><quantity>1

Consequently, the XML resolves to the following block:

<item>
  <description>Widget</description>
  <price>500.0</price>
  <quantity>1</quantity><price>1.0</price><quantity>1</quantity>
</item>

A SAX parser (org.xml.sax and javax.xml.parsers.SAXParser) interprets the above XML such that the second price field overrides the first, leaving the price of the item as $1. Even when it is not possible to perform such an attack, the attacker may be able to inject special characters, such as comment blocks and CDATA delimiters, which distort the meaning of the XML.

Noncompliant Code Example

In this noncompliant code example, a client method uses simple string concatenation to build an XML query to send to a server. XML injection is possible because the method performs no input validation.

private void createXMLStream(BufferedOutputStream outStream, String quantity) throws IOException {
  String xmlString;
  xmlString = "<item>\n<description>Widget</description>\n<price>500.0</price>\n" +
              "<quantity>" + quantity + "</quantity></item>";
  outStream.write(xmlString.getBytes());
  outStream.flush();
}

Compliant Solution (Whitelisting)

Depending on the specific data and command interpreter or parser to which data is being sent, different methods must be used to sanitize untrusted user data. This compliant solution uses whitelisting to sanitize the input. In this compliant solution, the method requires that the quantity field must be a number between 0 and 9.

private void createXMLStream(BufferedOutputStream outStream, String quantity) {
  // Write XML string if quantity contains numbers only (whitelisting)
  // Blacklisting of invalid characters can also be done in conjunction

  if (!Pattern.matches("[0-9]+", quantity)) {
    // Format violation
  }

  String xmlString = "<item>\n<description>Widget</description>\n<price>500</price>\n" +
                     "<quantity>" + quantity + "</quantity></item>";
  outStream.write(xmlString.getBytes());
  outStream.flush();
}

XML External Entity Attacks

An XML document can be dynamically constructed from smaller logical blocks called entities. Entities can be either internal, external, or parameter-based. External entities allow the inclusion of XML data from external files.

According to XML W3C Recommendation [[W3C 2008]], Section 4.4.3, "Included If Validating,"

When an XML processor recognizes a reference to a parsed entity, to validate the document, the processor MUST include its replacement text. If the entity is external, and the processor is not attempting to validate the XML document, the processor MAY, but need not, include the entity's replacement text.

An attacker may attempt to cause denial of service or program crashes by manipulating the URI of the entity to refer to special files existing on the local file system, for example, by specifying /dev/random or /dev/tty as input URIs. This may crash or block the program indefinitely. This is called an XML External Entity (XXE) attack. Because inclusion of replacement text from an external entity is optional, not all XML processors are vulnerable to external entity attacks.

Noncompliant Code Example

This noncompliant code example attempts to parse the file evil.xml, reports any errors, and exits. However, a SAX or a DOM parser will attempt to access the URL specified by the SYSTEM attribute, which means it will attempt to read the contents of the local /dev/tty file. On POSIX system, reading this file causes the program to block until input data is supplied to the machine's console. Consequently, an attacker can use this malicious XML file to cause the program to hang.

class XXE {
  private static void receiveXMLStream(InputStream inStream,
                                       DefaultHandler defaultHandler)
      throws ParserConfigurationException, SAXException, IOException {
    SAXParserFactory factory = SAXParserFactory.newInstance();
    SAXParser saxParser = factory.newSAXParser();
    saxParser.parse(inStream, defaultHandler);
  }

  public static void main(String[] args)
      throws ParserConfigurationException, SAXException, IOException {
    receiveXMLStream(new FileInputStream("evil.xml"), new DefaultHandler());
  }
}

Unfortunately, this program is subject to a remote XXE attack if the evil.xml file contains the following:

<?xml version="1.0"?>
<!DOCTYPE foo SYSTEM "file:/dev/tty">
<foo>bar</foo>

This noncompliant code example may also violate rule ERR06-J. Do not allow exceptions to expose sensitive information if the information contained in the exceptions is considered sensitive.

Compliant Solution (EntityResolver)

This compliant solution defines a CustomResolver class that implements the interface org.xml.sax.EntityResolver. This enables a SAX application to implement customized handling of external entities. The setEntityResolver() method registers the implementation with the corresponding SAX driver. The customized handler uses a simple whitelist for external entities. The resolveEntity method returns an empty InputSource when an input fails to resolve to any of the specified, safe entity source paths. Consequently, when parsing malicious input, the empty InputSource returned by the custom resolver causes a java.net.MalformedURLException to be thrown. Note that you must create an XMLReader object on which to set the custom entity resolver.

This is an example of component-based sanitization.

class CustomResolver implements EntityResolver {
  public InputSource resolveEntity(String publicId, String systemId)
    throws SAXException, IOException {

    // check for known good entities
    String entityPath = "/home/username/java/xxe/file";
    if (systemId.equals(entityPath)) {
      System.out.println("Resolving entity: " + publicId + " " + systemId);
      return new InputSource(entityPath);
    } else {
      return new InputSource(); // Disallow unknown entities, by returning blank path
    }
  }
}

class XXE {
  private static void receiveXMLStream(InputStream inStream,
                                       DefaultHandler defaultHandler)
      throws ParserConfigurationException, SAXException, IOException {
    SAXParserFactory factory = SAXParserFactory.newInstance();
    SAXParser saxParser = factory.newSAXParser();

    // To set the Entity Resolver, an XML reader needs to be created
    XMLReader reader = saxParser.getXMLReader();
    reader.setEntityResolver(new CustomResolver());
    reader.setErrorHandler(defaultHandler);

    InputSource is = new InputSource(inStream);
    reader.parse(is);
  }

  public static void main(String[] args)
      throws ParserConfigurationException, SAXException, IOException {
    receiveXMLStream(new FileInputStream("evil.xml"), new DefaultHandler());
  }
}

Risk Assessment

Failure to sanitize user input before processing or storing it can lead to injection attacks.

Related Vulnerabilities

CVE-2008-2370 describes a vulnerability in Apache Tomcat 4.1.0 through 4.1.37, 5.5.0 through 5.5.26, and 6.0.0 through 6.0.16. When a RequestDispatcher is used, Tomcat performs path normalization before removing the query string from the URI, which allows remote attackers to conduct directory traversal attacks and read arbitrary files via a .. (dot dot) in a request parameter.

Search for other vulnerabilities resulting from the violation of this rule on the CERT website.

Bibliography

[[OWASP 2005]]
[[OWASP 2007]]
[[OWASP 2008]] Testing for XML Injection (OWASP-DV-008)
[[W3C 2008]] 4.4.3 Included If Validating

            IDS02-J. Normalize strings before validating them