lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [day] [month] [year] [list]
Date: Wed, 25 Jan 2006 08:49:56 +0200
From: "Amit Klein (AKsecurity)" <aksecurity@...pop.com>
To: bugtraq@...urityfocus.com
Subject: Technical Note by Amit Klein: "XST Strikes Back"


                          Technical note

                         XST Strikes Back
              (or perhaps "Return from the Proxy"...)

                     Amit Klein, January 2006


Introduction
============

About three years ago, the concept of "Cross Site Tracing" [1] 
was introduced to the web application security community. In 
essence, the classic XST is about amplifying an existing XSS 
vulnerability such that HttpOnly cookies and HTTP authentication 
credentials can be compromised. This is done using a client side 
XmlHttpRequest object that sends a TRACE request back to the 
server, receives the request echoed back by the server's TRACE 
function, and extracts the information from the echoed back 
request.
The recommendation in [1] is to turn off TRACE support in the web 
server, which indeed takes care of the attack as described.

However, let us now consider a situation wherein there is a proxy 
server somewhere between the client (browser) and the server. In 
such case, it is possible to force the proxy server (at least, in 
theory) to respond to the TRACE request, rather than the origin 
server itself. Thus, HTTP TRACE can still be used to compromise 
the credentials of the user, even if the server does not support 
the TRACE request.


The technique
=============

Forcing the first proxy server in the chain to respond to the 
TRACE request (rather than forward it) is as simple as including 
an HTTP request header "Max-Forwards: 0" ([2], section 14.31). 

So, for IE (up to and including 6.0 SP1) and for Mozilla/Firefox 
(up to and including Firefox 1.0.6), the XSS payload should be 
(IE code, Mozilla/Firefox modifications commented):

   var x = new ActiveXObject("Microsoft.XMLHTTP"); 
       // var x = new XMLHttpRequest();
   x.open("TRACE","/",false);
   x.setRequestHeader("Max-Forwards","0");
   x.send(); 
       // x.send("");
   alert(x.responseText);

In IE 6.0 SP2, it seems that Microsoft silently removed support 
for TRACE in the XmlHttpRequest object. That is, no method 
starting with "TRACE" is allowed. However, a simple trick, 
involving a technique similar to the one used in [3] and [4] can 
be used to bypass this protection. Instead of using "TRACE" for 
the method, one can simply use "\r\nTRACE". To quote from [2] 
(section 4.1):
   
   "In the interest of robustness, servers SHOULD ignore any 
   empty line(s) received where a Request-Line is expected. In 
   other words, if the server is reading the protocol stream 
   at the beginning of a message and receives a CRLF first, it 
   should ignore the CRLF."

So the XSS payload for IE 6.0 SP2 would be:

   var x = new ActiveXObject("Microsoft.XMLHTTP");
   x.open("\r\nTRACE","/",false);
   x.setRequestHeader("Max-Forwards","0");
   x.send();
   alert(x.responseText);


Squid (2.5stable10/NT) ,Apache (2.0.54 mod_proxy) and other 
popular proxy servers were found to support TRACE and Max-
Forwards. 


Recommendations
===============

Proxy server vendors
--------------------

1. Ship proxy servers with default secure configuration, namely 

   no TRACE support disabled.

2. In the least, enable turning off support for TRACE via a 
   configuration option.

Proxy server owners/maintainers
-------------------------------

Disable support for TRACE.

1. For Squid, add the following to the Squid configuration file 
   (squid.conf):

   acl TRACE method TRACE
   ...
   http_access deny TRACE

2. For Apache, use mod_rewrite to prevent support for TRACE (see 
   [1]). Make sure to place the directive in the <proxy> section of 
   the httpd.conf file. Also, It would be a good idea to append the 
   "[nocase]" flag to the RewriteCond directive, to ensure case 
   insensitive comparison (though it seems that Apache will only 
   serve fully uppercase HTTP methods).

Browser vendors
--------------

Disable support for TRACE in the XmlHttpRequest object. Make sure 
you do it right though.

Web site owners
---------------

As a workaround (perhaps not too practical), enable SSL traffic 
only to your site. 


Summary
=======

This is yet another example of peripheral web security issue, 
such as the ones discussed in [5]. A web application may be 
compromised through issues that are beyond the control of the web 
site owner - in this case, support for TRACE in browsers and 
proxy servers. In fact, in many cases the site owner has no way 
of even knowing that the attack took place, because the TRACE 
request is answered at the proxy server, and never arrives at the 
web server (of course, if the first proxy server is the site's 
reverse proxy server, or if no proxy server at all is present, 
then the site owner may find out). 

It seems that the TRACE method should be disabled across the 
board - not just in web servers, but also in proxy servers and in 
browsers (and possibly in other web devices).


References
==========

[1] "Cross-Site Tracing (XST)", Jeremiah Grossman, January 20th, 
2003
http://www.cgisecurity.com/whitehat-mirror/WhitePaper_screen.pdf

[2] "Hypertext Transfer Protocol -- HTTP/1.1" RFC 2616
http://www.ietf.org/rfc/rfc2616.txt

[3] "XS(T) attack variants which can, in some cases, eliminate 
the need for TRACE", Amit Klein, WebAppSec mailing list submission, 
January 26th, 2003
http://www.securityfocus.com/archive/107/308433

[4] "Exploiting the XmlHttpRequest object in IE - Referrer 
spoofing, and a lot more...", Amit Klein, BugTraq mailing list 
submission, September 24th, 2005 
http://www.securityfocus.com/archive/1/411585

[5] "Meanwhile, at the other side of the web server", Amit Klein, 
BugTraq mailing list submission, June 9th, 2005
http://www.securityfocus.com/archive/1/401866




Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ