Software Engineering Institute

An intruder can send certain kinds of data to services that he is not ordinarily able to reach. By crafting the data such that it is redirected through any program the victim uses to render the malicious HTML, the intruder is able send that data to any services that the victim can send data to. If the victim is either tricked into clicking on a form submission button or a JavaScript program submits the form on behalf of the victim, the intruder's data may be sent to the service specified. Since the connection originates from the victim, any access control lists or restrictions designed to protect the server (such as a firewall) may not be effective. The data that the intruder is able to send is usually encoded as "multipart/form-data" by the browser, which necessarily inserts some header and encoding metadata, and is subject to any limitations of the protocol it attempts to attack.

This vulnerability has been called "cross-protocol scripting."

An intruder may be able to use this vulnerability to send mail (Spam), post News, get or send files from or to an FTP server, or send data to an HTTP server. It may even be possible to exploit a vulnerability in one of these services through this problem, though we are not certain of that at this time. For example, an intruder may be able to exploit this problem as a means of attacking a vulnerable web server that would ordinarily be protected by a firewall. Additionally, it may be possible for an intruder to cause denial-of-service conditions within the network by sending unexpected data to network services. This unexpected data may crash or hang the services receiving the data.

Upgrade your application according to your manufacturer's recommendations, if any. Additionally, do not rely solely on firewalls to provide a guarantee that an intruder can not reach a service. Keep internal systems up to date with respect to patches and workarounds.