• News
• Infocus
• Foundations
• Microsoft
• Unix
• IDS
• Incidents
• Virus
• Pen-Test
• Firewalls
• Focus On: Vista
• Columnists
• Mailing Lists
• Newsletters
• Bugtraq
• Focus on IDS
• Focus on Linux
• Focus on Microsoft
• Forensics
• Pen-test
• Security Basics
• Vuln Dev
• Vulnerabilities
• Jobs
• Job Opportunities
• Resumes
• Job Seekers
• Employers
• Tools
• RSS
• News
• Vulns

• info
• discussion
• exploit
• solution
• references

Multiple Vendor HTML Form Protocol Vulnerability

Malicious HTML forms can submit data to any port on an arbitrary machine. This opens the potential for remote command-execution attacks originating from an unsuspecting web user who submits the form. This attack may be initiated with JavaScript when a victim views a malicious page or HTML email.

To exploit this issue, an attacker must send the malicious form data unencoded to a server that uses an ASCII-based protocol. Possible targets include SMTP, NNTP, POP3, IMAP, and IRC. Remote commands may be passed to a pertinent service by including them as form content.

This issue may allow the attacker to take advantage of a trust relationship that exists between the victim and a third party (e.g. if both the victim and the third party are located behind a firewall).

Some servers may return user-supplied data, often as part of an error message. If the user-supplied data includes JavaScript, it will be executed in the context of the server. This is a type of cross-site scripting attack and may result in the disclosure of sensitive information such as cookie data.

This issue is known to be a problem with browsers and HTML-enabled email clients, but some server implementations and proxies will accept data sent in this manner.

NOTE: An attacker may be able to circumvent browsers that prevent access to certain ports by adding 65536 to the number of the port that the attacker is sending data to.