Introduction

XML eXternal Entity injection (XXE), which is now part of the OWASP Top 10 via the point A4, is a type of attack against an application that parses XML input.

XXE issue is referenced under the ID 611 in the Common Weakness Enumeration referential.

This attack occurs when untrusted XML input containing a reference to an external entity is processed by a weakly configured XML parser.

This attack may lead to the disclosure of confidential data, denial of service, Server Side Request Forgery (SSRF), port scanning from the perspective of the machine where the parser is located, and other system impacts. The following guide provides concise information to prevent this vulnerability.

For more information on XXE, please visit XML External Entity (XXE).

General Guidance

The safest way to prevent XXE is always to disable DTDs (External Entities) completely. Depending on the parser, the method should be similar to the following:

  1. factory.setFeature("http://apache.org/xml/features/disallow-doctype-decl", true);

Disabling DTDs also makes the parser secure against denial of services (DOS) attacks such as Billion Laughs. If it is not possible to disable DTDs completely, then external entities and external document type declarations must be disabled in the way that's specific to each parser.

Detailed XXE Prevention guidance for a number of languages and commonly used XML parsers in those languages is provided below.

C/C++

libxml2

The Enum xmlParserOption should not have the following options defined:

  • XML_PARSE_NOENT: Expands entities and substitutes them with replacement text
  • XML_PARSE_DTDLOAD: Load the external DTD

Note:

Per: According to this post, starting with libxml2 version 2.9, XXE has been disabled by default as committed by the following patch.

Search for the usage of the following APIs to ensure there is no XML_PARSE_NOENT and XML_PARSE_DTDLOAD defined in the parameters:

  • xmlCtxtReadDoc
  • xmlCtxtReadFd
  • xmlCtxtReadFile
  • xmlCtxtReadIO
  • xmlCtxtReadMemory
  • xmlCtxtUseOptions
  • xmlParseInNodeContext
  • xmlReadDoc
  • xmlReadFd
  • xmlReadFile
  • xmlReadIO
  • xmlReadMemory

libxerces-c

Use of XercesDOMParser do this to prevent XXE:

  1. XercesDOMParser *parser = new XercesDOMParser;
  2. parser->setCreateEntityReferenceNodes(true);
  3. parser->setDisableDefaultEntityResolution(true);

Use of SAXParser, do this to prevent XXE:

  1. SAXParser* parser = new SAXParser;
  2. parser->setDisableDefaultEntityResolution(true);

Use of SAX2XMLReader, do this to prevent XXE:

  1. SAX2XMLReader* reader = XMLReaderFactory::createXMLReader();
  2. parser->setFeature(XMLUni::fgXercesDisableDefaultEntityResolution, true);

Java

Java applications using XML libraries are particularly vulnerable to XXE because the default settings for most Java XML parsers is to have XXE enabled. To use these parsers safely, you have to explicitly disable XXE in the parser you use. The following describes how to disable XXE in the most commonly used XML parsers for Java.

JAXP DocumentBuilderFactory, SAXParserFactory and DOM4J

DocumentBuilderFactory, SAXParserFactory and DOM4J XML Parsers can be configured using the same techniques to protect them against XXE.

Only the DocumentBuilderFactory example is presented here. The JAXP DocumentBuilderFactory setFeature) method allows a developer to control which implementation-specific XML processor features are enabled or disabled.

The features can either be set on the factory or the underlying XMLReader setFeature method.

Each XML processor implementation has its own features that govern how DTDs and external entities are processed.

For a syntax highlighted example code snippet using SAXParserFactory, look here.

  1. import javax.xml.parsers.DocumentBuilderFactory;
  2. import javax.xml.parsers.ParserConfigurationException; // catching unsupported features
  3. ...
  4. DocumentBuilderFactory dbf = DocumentBuilderFactory.newInstance();
  5. String FEATURE = null;
  6. try {
  7. // This is the PRIMARY defense. If DTDs (doctypes) are disallowed, almost all
  8. // XML entity attacks are prevented
  9. // Xerces 2 only - http://xerces.apache.org/xerces2-j/features.html#disallow-doctype-decl
  10. FEATURE = "http://apache.org/xml/features/disallow-doctype-decl";
  11. dbf.setFeature(FEATURE, true);
  12. // If you can't completely disable DTDs, then at least do the following:
  13. // Xerces 1 - http://xerces.apache.org/xerces-j/features.html#external-general-entities
  14. // Xerces 2 - http://xerces.apache.org/xerces2-j/features.html#external-general-entities
  15. // JDK7+ - http://xml.org/sax/features/external-general-entities
  16. FEATURE = "http://xml.org/sax/features/external-general-entities";
  17. dbf.setFeature(FEATURE, false);
  18. // Xerces 1 - http://xerces.apache.org/xerces-j/features.html#external-parameter-entities
  19. // Xerces 2 - http://xerces.apache.org/xerces2-j/features.html#external-parameter-entities
  20. // JDK7+ - http://xml.org/sax/features/external-parameter-entities
  21. FEATURE = "http://xml.org/sax/features/external-parameter-entities";
  22. dbf.setFeature(FEATURE, false);
  23. // Disable external DTDs as well
  24. FEATURE = "http://apache.org/xml/features/nonvalidating/load-external-dtd";
  25. dbf.setFeature(FEATURE, false);
  26. // and these as well, per Timothy Morgan's 2014 paper: "XML Schema, DTD, and Entity Attacks"
  27. dbf.setXIncludeAware(false);
  28. dbf.setExpandEntityReferences(false);
  29. // And, per Timothy Morgan: "If for some reason support for inline DOCTYPEs are a requirement, then
  30. // ensure the entity settings are disabled (as shown above) and beware that SSRF attacks
  31. // (http://cwe.mitre.org/data/definitions/918.html) and denial
  32. // of service attacks (such as billion laughs or decompression bombs via "jar:") are a risk."
  33. // remaining parser logic
  34. ...
  35. } catch (ParserConfigurationException e) {
  36. // This should catch a failed setFeature feature
  37. logger.info("ParserConfigurationException was thrown. The feature '" + FEATURE
  38. + "' is probably not supported by your XML processor.");
  39. ...
  40. } catch (SAXException e) {
  41. // On Apache, this should be thrown when disallowing DOCTYPE
  42. logger.warning("A DOCTYPE was passed into the XML document");
  43. ...
  44. } catch (IOException e) {
  45. // XXE that points to a file that doesn't exist
  46. logger.error("IOException occurred, XXE may still possible: " + e.getMessage());
  47. ...
  48. }
  49. // Load XML file or stream using a XXE agnostic configured parser...
  50. DocumentBuilder safebuilder = dbf.newDocumentBuilder();

Xerces 1Features:

  • Do not include external entities by setting this feature to false.
  • Do not include parameter entities by setting this feature to false.
  • Do not include external DTDs by setting this feature to false.

Xerces 2Features:

  • Disallow an inline DTD by setting this feature to true.
  • Do not include external entities by setting this feature to false.
  • Do not include parameter entities by setting this feature to false.
  • Do not include external DTDs by setting this feature to false.

Note: The above defenses require Java 7 update 67, Java 8 update 20, or above, because the above countermeasures for DocumentBuilderFactory and SAXParserFactory are broken in earlier Java versions, per: CVE-2014-6517.

XMLInputFactory (a StAX parser)

StAX parsers such as XMLInputFactory allow various properties and features to be set.

To protect a Java XMLInputFactory from XXE, do this:

  1. // This disables DTDs entirely for that factory
  2. xmlInputFactory.setProperty(XMLInputFactory.SUPPORT_DTD, false);
  3. // disable external entities
  4. xmlInputFactory.setProperty("javax.xml.stream.isSupportingExternalEntities", false);

TransformerFactory

To protect a javax.xml.transform.TransformerFactory from XXE, do this:

  1. TransformerFactory tf = TransformerFactory.newInstance();
  2. tf.setAttribute(XMLConstants.ACCESS_EXTERNAL_DTD, "");
  3. tf.setAttribute(XMLConstants.ACCESS_EXTERNAL_STYLESHEET, "");

Validator

To protect a javax.xml.validation.Validator from XXE, do this:

  1. SchemaFactory factory = SchemaFactory.newInstance("http://www.w3.org/2001/XMLSchema");
  2. Schema schema = factory.newSchema();
  3. Validator validator = schema.newValidator();
  4. validator.setProperty(XMLConstants.ACCESS_EXTERNAL_DTD, "");
  5. validator.setProperty(XMLConstants.ACCESS_EXTERNAL_SCHEMA, "");

SchemaFactory

To protect a javax.xml.validation.SchemaFactory from XXE, do this:

  1. SchemaFactory factory = SchemaFactory.newInstance("http://www.w3.org/2001/XMLSchema");
  2. factory.setProperty(XMLConstants.ACCESS_EXTERNAL_DTD, "");
  3. factory.setProperty(XMLConstants.ACCESS_EXTERNAL_SCHEMA, "");
  4. Schema schema = factory.newSchema(Source);

SAXTransformerFactory

To protect a javax.xml.transform.sax.SAXTransformerFactory from XXE, do this:

  1. SAXTransformerFactory sf = SAXTransformerFactory.newInstance();
  2. sf.setAttribute(XMLConstants.ACCESS_EXTERNAL_DTD, "");
  3. sf.setAttribute(XMLConstants.ACCESS_EXTERNAL_STYLESHEET, "");
  4. sf.newXMLFilter(Source);

Note: Use of the following XMLConstants requires JAXP 1.5, which was added to Java in 7u40 and Java 8:

  • javax.xml.XMLConstants.ACCESS_EXTERNAL_DTD
  • javax.xml.XMLConstants.ACCESS_EXTERNAL_SCHEMA
  • javax.xml.XMLConstants.ACCESS_EXTERNAL_STYLESHEET

XMLReader

To protect a Java org.xml.sax.XMLReader from XXE, do this:

  1. XMLReader reader = XMLReaderFactory.createXMLReader();
  2. reader.setFeature("http://apache.org/xml/features/disallow-doctype-decl", true);
  3. // This may not be strictly required as DTDs shouldn't be allowed at all, per previous line.
  4. reader.setFeature("http://apache.org/xml/features/nonvalidating/load-external-dtd", false);
  5. reader.setFeature("http://xml.org/sax/features/external-general-entities", false);
  6. reader.setFeature("http://xml.org/sax/features/external-parameter-entities", false);

SAXReader

To protect a Java org.dom4j.io.SAXReader from XXE, do this:

  1. saxReader.setFeature("http://apache.org/xml/features/disallow-doctype-decl", true);
  2. saxReader.setFeature("http://xml.org/sax/features/external-general-entities", false);
  3. saxReader.setFeature("http://xml.org/sax/features/external-parameter-entities", false);

Based on testing, if you are missing one of these, you can still be vulnerable to an XXE attack.

SAXBuilder

To protect a Java org.jdom2.input.SAXBuilder from XXE, do this:

  1. SAXBuilder builder = new SAXBuilder();
  2. builder.setFeature("http://apache.org/xml/features/disallow-doctype-decl",true);
  3. builder.setFeature("http://xml.org/sax/features/external-general-entities", false);
  4. builder.setFeature("http://xml.org/sax/features/external-parameter-entities", false);
  5. Document doc = builder.build(new File(fileName));

No-op EntityResolver

For APIs that take an EntityResolver, you can neutralize an XML parser's ability to resolve entities by supplying a no-op implementation:

  1. public final class NoOpEntityResolver implements EntityResolver {
  2. public InputSource resolveEntity(String publicId, String systemId) {
  3. return new InputSource(new StringReader(""));
  4. }
  5. }
  6. // ...
  7. xmlReader.setEntityResolver(new NoOpEntityResolver());
  8. documentBuilder.setEntityResolver(new NoOpEntityResolver());

or more simply:

  1. EntityResolver noop = (publicId, systemId) -> new InputSource(new StringReader(""));
  2. xmlReader.setEntityResolver(noop);
  3. documentBuilder.setEntityResolver(noop);

JAXB Unmarshaller

Since a javax.xml.bind.Unmarshaller parses XML and does not support any flags for disabling XXE, it's imperative to parse the untrusted XML through a configurable secure parser first, generate a source object as a result, and pass the source object to the Unmarshaller. For example:

  1. //Disable XXE
  2. SAXParserFactory spf = SAXParserFactory.newInstance();
  3. spf.setFeature("http://xml.org/sax/features/external-general-entities", false);
  4. spf.setFeature("http://xml.org/sax/features/external-parameter-entities", false);
  5. spf.setFeature("http://apache.org/xml/features/nonvalidating/load-external-dtd", false);
  6. //Do unmarshall operation
  7. Source xmlSource = new SAXSource(spf.newSAXParser().getXMLReader(),
  8. new InputSource(new StringReader(xml)));
  9. JAXBContext jc = JAXBContext.newInstance(Object.class);
  10. Unmarshaller um = jc.createUnmarshaller();
  11. um.unmarshal(xmlSource);

Java 8 and up

Since JDK-8010393, which is in OpenJDK 8 beta 86, javax.xml.bind.Unmarshaller instances are safe by default. The other classes mentioned here are still unsafe by default in Java 8.

XPathExpression

A javax.xml.xpath.XPathExpression can not be configured securely by itself, so the untrusted data must be parsed through another securable XML parser first.

For example:

  1. DocumentBuilderFactory df = DocumentBuilderFactory.newInstance();
  2. df.setAttribute(XMLConstants.ACCESS_EXTERNAL_DTD, "");
  3. df.setAttribute(XMLConstants.ACCESS_EXTERNAL_SCHEMA, "");
  4. DocumentBuilder builder = df.newDocumentBuilder();
  5. String result = new XPathExpression().evaluate( builder.parse(
  6. new ByteArrayInputStream(xml.getBytes())) );

java.beans.XMLDecoder

The readObject() method in this class is fundamentally unsafe.

Not only is the XML it parses subject to XXE, but the method can be used to construct any Java object, and execute arbitrary code as described here.

And there is no way to make use of this class safe except to trust or properly validate the input being passed into it.

As such, we'd strongly recommend completely avoiding the use of this class and replacing it with a safe or properly configured XML parser as described elsewhere in this cheat sheet.

Other XML Parsers

There are many 3rd party libraries that parse XML either directly or through their use of other libraries. Please test and verify their XML parser is secure against XXE by default. If the parser is not secure by default, look for flags supported by the parser to disable all possible external resource inclusions like the examples given above. If there's no control exposed to the outside, make sure the untrusted content is passed through a secure parser first and then passed to insecure 3rd party parser similar to how the Unmarshaller is secured.

Spring Framework MVC/OXM XXE Vulnerabilities

For example, some XXE vulnerabilities were found in Spring OXM and Spring MVC. The following versions of the Spring Framework are vulnerable to XXE:

  • 3.0.0 to 3.2.3 (Spring OXM & Spring MVC)
  • 4.0.0.M1 (Spring OXM)
  • 4.0.0.M1-4.0.0.M2 (Spring MVC)

There were other issues as well that were fixed later, so to fully address these issues, Spring recommends you upgrade to Spring Framework 3.2.8+ or 4.0.2+.

For Spring OXM, this is referring to the use of org.springframework.oxm.jaxb.Jaxb2Marshaller. Note that the CVE for Spring OXM specifically indicates that 2 XML parsing situations are up to the developer to get right, and 2 are the responsibility of Spring and were fixed to address this CVE.

Here's what they say:

Two situations developers must handle:

  • For a DOMSource, the XML has already been parsed by user code and that code is responsible for protecting against XXE.
  • For a StAXSource, the XMLStreamReader has already been created by user code and that code is responsible for protecting against XXE.

The issue Spring fixed:

For SAXSource and StreamSource instances, Spring processed external entities by default thereby creating this vulnerability.

Here's an example of using a StreamSource that was vulnerable, but is now safe, if you are using a fixed version of Spring OXM or Spring MVC:

  1. import org.springframework.oxm.Jaxb2Marshaller;
  2. import org.springframework.oxm.jaxb.Jaxb2Marshaller;
  3. Jaxb2Marshaller marshaller = new Jaxb2Marshaller();
  4. // Must cast return Object to whatever type you are unmarshalling
  5. marshaller.unmarshal(new StreamSource(new StringReader(some_string_containing_XML));

So, per the Spring OXM CVE writeup, the above is now safe. But if you were to use a DOMSource or StAXSource instead, it would be up to you to configure those sources to be safe from XXE.

Castor

Castor is a data binding framework for Java. It allows conversion between Java objects, XML, and relational tables. The XML features in Castor prior to version 1.3.3 are vulnerable to XXE, and should be upgraded to the latest version. For additional information, check the official XML configuration file

.NET

The following information for XXE injection in .NET is directly from this web application of unit tests by Dean Fleming.

This web application covers all currently supported .NET XML parsers, and has test cases for each demonstrating when they are safe from XXE injection and when they are not.

Previously, this information was based on James Jardine's excellent .NET XXE article.

It originally provided more recent and more detailed information than the older article from Microsoft on how to prevent XXE and XML Denial of Service in .NET, however, it has some inaccuracies that the web application covers.

The following table lists all supported .NET XML parsers and their default safety levels:

XML ParserSafe by default?
LINQ to XMLYes
XmlDictionaryReaderYes
XmlDocument
…prior to 4.5.2No
…in versions 4.5.2+Yes
XmlNodeReaderYes
XmlReaderYes
XmlTextReader
…prior to 4.5.2No
…in versions 4.5.2+Yes
XPathNavigator
…prior to 4.5.2No
…in versions 4.5.2+Yes
XslCompiledTransformYes

LINQ to XML

Both the XElement and XDocument objects in the System.Xml.Linq library are safe from XXE injection by default. XElement parses only the elements within the XML file, so DTDs are ignored altogether. XDocument has DTDs disabled by default, and is only unsafe if constructed with a different unsafe XML parser.

XmlDictionaryReader

System.Xml.XmlDictionaryReader is safe by default, as when it attempts to parse the DTD, the compiler throws an exception saying that "CData elements not valid at top level of an XML document". It becomes unsafe if constructed with a different unsafe XML parser.

XmlDocument

Prior to .NET Framework version 4.5.2, System.Xml.XmlDocument is unsafe by default. The XmlDocument object has an XmlResolver object within it that needs to be set to null in versions prior to 4.5.2. In versions 4.5.2 and up, this XmlResolver is set to null by default.

The following example shows how it is made safe:

  1. static void LoadXML()
  2. {
  3. string xxePayload = "<!DOCTYPE doc [<!ENTITY win SYSTEM 'file:///C:/Users/testdata2.txt'>]>"
  4. + "<doc>&win;</doc>";
  5. string xml = "<?xml version='1.0' ?>" + xxePayload;
  6. XmlDocument xmlDoc = new XmlDocument();
  7. // Setting this to NULL disables DTDs - Its NOT null by default.
  8. xmlDoc.XmlResolver = null;
  9. xmlDoc.LoadXml(xml);
  10. Console.WriteLine(xmlDoc.InnerText);
  11. Console.ReadLine();
  12. }

XmlDocument can become unsafe if you create your own nonnull XmlResolver with default or unsafe settings. If you need to enable DTD processing, instructions on how to do so safely are described in detail in the referenced MSDN article.

XmlNodeReader

System.Xml.XmlNodeReader objects are safe by default and will ignore DTDs even when constructed with an unsafe parser or wrapped in another unsafe parser.

XmlReader

System.Xml.XmlReader objects are safe by default.

They are set by default to have their ProhibitDtd property set to false in .NET Framework versions 4.0 and earlier, or their DtdProcessing property set to Prohibit in .NET versions 4.0 and later.

Additionally, in .NET versions 4.5.2 and later, the XmlReaderSettings belonging to the XmlReader has its XmlResolver set to null by default, which provides an additional layer of safety.

Therefore, XmlReader objects will only become unsafe in version 4.5.2 and up if both the DtdProcessing property is set to Parse and the XmlReaderSetting's XmlResolver is set to a nonnull XmlResolver with default or unsafe settings. If you need to enable DTD processing, instructions on how to do so safely are described in detail in the referenced MSDN article.

XmlTextReader

System.Xml.XmlTextReader is unsafe by default in .NET Framework versions prior to 4.5.2. Here is how to make it safe in various .NET versions:

Prior to .NET 4.0

In .NET Framework versions prior to 4.0, DTD parsing behavior for XmlReader objects like XmlTextReader are controlled by the Boolean ProhibitDtd property found in the System.Xml.XmlReaderSettings and System.Xml.XmlTextReader classes.

Set these values to true to disable inline DTDs completely.

  1. XmlTextReader reader = new XmlTextReader(stream);
  2. // NEEDED because the default is FALSE!!
  3. reader.ProhibitDtd = true;

.NET 4.0 - .NET 4.5.2

In .NET Framework version 4.0, DTD parsing behavior has been changed. The ProhibitDtd property has been deprecated in favor of the new DtdProcessing property.

However, they didn't change the default settings so XmlTextReader is still vulnerable to XXE by default.

Setting DtdProcessing to Prohibit causes the runtime to throw an exception if a <!DOCTYPE> element is present in the XML.

To set this value yourself, it looks like this:

  1. XmlTextReader reader = new XmlTextReader(stream);
  2. // NEEDED because the default is Parse!!
  3. reader.DtdProcessing = DtdProcessing.Prohibit;

Alternatively, you can set the DtdProcessing property to Ignore, which will not throw an exception on encountering a <!DOCTYPE> element but will simply skip over it and not process it. Finally, you can set DtdProcessing to Parse if you do want to allow and process inline DTDs.

.NET 4.5.2 and later

In .NET Framework versions 4.5.2 and up, XmlTextReader's internal XmlResolver is set to null by default, making the XmlTextReader ignore DTDs by default. The XmlTextReader can become unsafe if if you create your own nonnull XmlResolver with default or unsafe settings.

XPathNavigator

System.Xml.XPath.XPathNavigator is unsafe by default in .NET Framework versions prior to 4.5.2.

This is due to the fact that it implements IXPathNavigable objects like XmlDocument, which are also unsafe by default in versions prior to 4.5.2.

You can make XPathNavigator safe by giving it a safe parser like XmlReader (which is safe by default) in the XPathDocument's constructor.

Here is an example:

  1. XmlReader reader = XmlReader.Create("example.xml");
  2. XPathDocument doc = new XPathDocument(reader);
  3. XPathNavigator nav = doc.CreateNavigator();
  4. string xml = nav.InnerXml.ToString();

XslCompiledTransform

System.Xml.Xsl.XslCompiledTransform (an XML transformer) is safe by default as long as the parser it's given is safe.

It is safe by default because the default parser of the Transform() methods is an XmlReader, which is safe by default (per above).

The source code for this method is here.

Some of the Transform() methods accept an XmlReader or IXPathNavigable (e.g., XmlDocument) as an input, and if you pass in an unsafe XML Parser then the Transform will also be unsafe.

iOS

libxml2

iOS includes the C/C++ libxml2 library described above, so that guidance applies if you are using libxml2 directly.

However, the version of libxml2 provided up through iOS6 is prior to version 2.9 of libxml2 (which protects against XXE by default).

NSXMLDocument

iOS also provides an NSXMLDocument type, which is built on top of libxml2.

However, NSXMLDocument provides some additional protections against XXE that aren't available in libxml2 directly.

Per the 'NSXMLDocument External Entity Restriction API' section of this page:

  • iOS4 and earlier: All external entities are loaded by default.
  • iOS5 and later: Only entities that don't require network access are loaded. (which is safer)

However, to completely disable XXE in an NSXMLDocument in any version of iOS you simply specify NSXMLNodeLoadExternalEntitiesNever when creating the NSXMLDocument.

PHP

Per the PHP documentation, the following should be set when using the default PHP XML parser in order to prevent XXE:

  1. libxml_disable_entity_loader(true);

A description of how to abuse this in PHP is presented in a good SensePost article describing a cool PHP based XXE vulnerability that was fixed in Facebook.

Python

The Python 3 official documentation contains a section on xml vulnerabilities. For Python 2, you can refer to this page.


The end of life for Python 2 is expected to be January 1st, 2020.


The following table gives an overview of various modules in Python 3 used for XML parsing and whether or not they are vulnerable.

Attack Typesaxetreeminidompulldomxmlrpc
Billion LaughsVulnerableVulnerableVulnerableVulnerableVulnerable
Quadratic BlowupVulnerableVulnerableVulnerableVulnerableVulnerable
External Entity ExpansionSafeSafeSafeSafeSafe
DTD RetrievalSafeSafeSafeSafeSafe
Decompression BombSafeSafeSafeSafeVulnerable

To protect your application from the applicable attacks, two packages exist to help you sanitize your input and protect your application against DDoS and remote attacks.

References