Advisory: IceWarp WebMail Server: Cross Site Scripting in Email View During a penetration test, RedTeam Pentesting discovered that the IceWarp WebMail Server is prone to Cross Site Scripting attacks in its email view. This enables attackers to send emails with embedded JavaScript code, for example, to steal users' session IDs. Details ======= Product: IceWarp eMail Server / WebMail Server Affected Versions: 9.4.1 Fixed Versions: 9.4.2 Vulnerability Type: Cross Site Scripting Security Risk: high Vendor URL: http://www.icewarp.com/ Vendor Status: notified, fixed version released Advisory URL: http://www.redteam-pentesting.de/advisories/rt-sa-2009-001 Advisory Status: published CVE: CVE-2009-1467 CVE URL: https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2009-1467 Introduction ============ "Feature complete yet easy to use, WebMail Server Pro provides feature rich Web 2.0 web-based access to email, calendars, contacts, files and shared data from any computer with browser and internet connection, without the usual configuration hassle. Thanks to advanced technologies and application-like look and feel, Pro suggests it was born to become the ultimate replacement of Outlook and similar desktop mail clients." (from the vendor's homepage) More Details ============ To prevent the execution of JavaScript and VBScript code in HTML emails and to remove unwanted HTML tags, the IceWarp WebMail Server filters HTML emails with the function cleanHTML() that is defined in the PHP file html/webmail/server/inc/tools.php This filtering function can be circumvented in various ways, to still allow XSS to happen. Tag Removal ----------- Beginning in line 462 down to line 482, the cleanHTML() function removes or changes a variety of keywords which are considered malicious. This includes the removal of all attributes starting with "on" (e.g. onmouseover, onload etc.) and the rewriting of the words "javascript" and "vbscript" to "noscript". Later, in line 485, the cleanHTML() function completely removes various HTML tags from the email: $string = preg_replace('#]*>#i',"",$string); By inserting one of these HTML tags, which gets removed by the filtering function, between the keywords which get filtered before, the cleanHTML() function will not recognize them anymore. Later, it will remove the HTML tag and thereby make the keyword valid again. Proof of Concept Tag Removal ---------------------------- An HTML email with the following content will open an alert box in victims' browsers when they move the mouse over the "XSS" text of the
tag:
nmouseover="alert('XSS')";>XSS
This circumvention should work in all recent browsers with JavaScript support, during the penetration test Firefox 3.x was used. Hex Encoding ------------ Another way to circumvent the filtering is to use hex-encoded HTML entities. The character "j" can e.g. be encoded as "j". The cleanHTML() function decodes the given string once in line 459: $string = html_entity_decode($string, ENT_COMPAT, "UTF-8"); By double encoding the input, all regular expressions used for filtering will fail to recognize malicious content even after the first decoding. The effect of this vulnerability is limited by the fact that HTML encoded entities are recognized by browsers only if they are part of a text node or attribute values. Proof of Concept Hex Encoding ----------------------------- The "src" attribute value of the HTML tag can be double encoded as This will be decoded by the html_entity_decode() function in line 459 to the string The value of the "src" attribute will not be recognized by the filtering attempts following the decoding in the cleanHTML() function. A message with such an image tag in the body will trigger the JavaScript and open a popup box if the browser supports "javascript:" attribute values in image elements' "src" attributes. This particular proof of concept works with Internet Explorer 6, the newest Firefox will not execute the JavaScript. Workaround ========== Users can disable the HTML message view in their configuration, so emails are only displayed as plaintext. Fix === Upgrade to version 9.4.2. Security Risk ============= The risk of this vulnerability is estimated as high. Attackers have the possibility to execute arbitrary JavaScript code in users' browsers and are thereby able to steal, for example, the users' session IDs. Users also cannot disallow JavaScript, as it is needed by the application itself to work properly. History ======= 2009-03-23 Vulnerabilities identified during a penetration test 2009-04-01 Meeting with customer and vendor 2009-04-28 CVE number assigned 2009-05-05 Vendor publishes fixed version 2009-05-05 Advisory released RedTeam Pentesting GmbH ======================= RedTeam Pentesting offers individual penetration tests, short pentests, performed by a team of specialised IT-security experts. Hereby, security weaknesses in company networks or products are uncovered and can be fixed immediately. As there are only few experts in this field, RedTeam Pentesting wants to share its knowledge and enhance the public knowledge with research in security-related areas. The results are made available as public security advisories. More information about RedTeam Pentesting can be found at http://www.redteam-pentesting.de.