Applications that build a Content-Disposition header value from untrusted input may be vulnerable to HTTP response splitting when the input is a malicious file name.

Specifically, an application can be vulnerable in either of the following cases:

• The application sets a response's Content-Disposition header from untrusted input, and the underlying HTTP connector does not itself reject malicious characters in header values. The servlet and reactive connectors supported by Spring Framework (Tomcat, Reactor Netty, Jetty, Undertow) already reject such characters, so this case applies only to non-default or legacy connectors.
• The application uses untrusted input to set the Content-Disposition of a part when building an outbound multipart request body. This case is independent of the underlying connector.

When either condition is met, an attacker who controls content disposition values can break the HTTP header, or corrupt the framing of an outbound multipart body.