Skip to main content

Introduced protections against XXE attacks


ImportanceReview GuidanceRequires Scanning Tool
HIGHMerge Without ReviewNo

This change updates all instances of XMLInputFactory to prevent them from resolving external entities, which can protect you from arbitrary code execution, sensitive data exfiltration, and probably a bunch more evil things attackers are still discovering.

Without this protection, attackers can cause your XMLInputFactory parser to retrieve sensitive information with attacks like this:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE foo [ <!ENTITY xxe SYSTEM "file:///etc/passwd"> ]>

Yes, it's pretty insane that this is the default behavior. Our change hardens the factories created with the necessary security features to prevent your parser from resolving external entities.

+ import;
- XMLInputFactory xmlInputFactory = XMLInputFactory.newFactory();
+ XMLInputFactory xmlInputFactory = XMLInputFactorySecurity.hardenFactory(XMLInputFactory.newFactory());

You could take our protections one step further by changing our supplied code to prevent the user from supplying a DOCTYPE, which is more aggressive and more secure, but also more likely to affect existing code behavior:

+ import;
+ import;
XMLInputFactory xmlInputFactory = XMLInputFactorySecurity.hardenFactory(XMLInputFactory.newFactory(), XMLRestrictions.DISALLOW_DOCTYPE);


Why is this codemod marked as Merge Without Review?

We believe this change is safe and effective. The behavior of hardened XMLInputFactory instances will only be different if the XML they process uses external entities, which is exceptionally rare (and, as demonstrated, quite unsafe anyway.)