bugtraq - Re: MAGIC XSS INTO THE DNS: coelacanth

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]

Message-ID: <20040617204937.14079.qmail@www.securityfocus.com>
Date: 17 Jun 2004 20:49:37 -0000
From: <qazxdrgb@...mail.com>
To: bugtraq@...urityfocus.com
Subject: Re: MAGIC XSS INTO THE DNS: coelacanth

In-Reply-To: <200406151517.i5FFH8pC029012@...179.megawebservers.com>

This just plain simple XSS attacks, and additionally it relies on a (long since?) patched vulnerability in IIS.

>Still unclear how or why this can be interpreted into the site 
>or through the browser.

What is unclear?
1. they allow (whatever).(domainanme) hostnames into  site. That is not very uncommon.
2. they generate absolute paths by concatenating "http://"+hostname+"/URI"
3. webserver does not abort with HTTP/1.1 400 Bad Request as it should.

This is not that uncommon, looking for this we will most likely find it in a lot of CGI/PHP/JSP/ASP code. Luckily, the attack requires the host to accept silly hostnames. The problem with e-gold.com is that they use an old webserver with an already fixed IIS vulnerability I think;

bash-2.02$ cat test.txt
GET /hello/just/a/test/please/forgive/me HTTP/1.1
Host: ">&lt;script&gt;alert()&lt;/script&gt;

bash-2.02$ nc www.microsoft.com 80 < test.txt
HTTP/1.1 400 Bad Request
Content-Type: text/html
Date: Thu, 17 Jun 2004 20:15:07 GMT
Connection: close
Content-Length: 20

<h1>Bad Request</h1>bash-2.02$ nc www.e-gold.com 80  < test.txt
HTTP/1.1 404 Object Not Found
Server: Microsoft-IIS/4.0
Date: Thu, 17 Jun 2004 20:15:56 GMT
Connection: close
Content-Length: 930
Content-Type: text/html
<cut junk>

To extend the attack to more systems, one need to find dangerous meta characters which are not filtered by normal Bad Request / Bad Address filters.

I did a very hasty search for webservers which would output unformated hostnames or URI's in error messages, without any luck. But I am certain someone more tenacious will succeed. The net is vast.

Basically, searches for potential vulnerable sites can be automated by testing the pattern such as:

GET / HTTP/1.1
Host: XXXXXXXXXXXXXXXXX

GET /some_script HTTP/1.1
Host: XXXXXXXXXXXXXXXXX

GET /GIVE-ME-NOT-FOUND HTTP/1.1
Host: XXXXXXXXXXXXXXXXX

GET GIVE-ME-BAD-URI HTTP/1.1
Host: XXXXXXXXXXXXXXXXX

Do we get XXXXXXXXXXXXXXXXX back in HTML?

Would be pretty easy to add the most basic searches to vulnerability scanners I think.

Sincerly yours,
Peter, 11a nu