QlikView '.qvw' File Remote Integer Overflow Vulnerability

ID EDB-ID:38388
Type exploitdb
Reporter A. Antukh
Modified 2013-03-13T00:00:00


QlikView '.qvw' File Remote Integer Overflow Vulnerability. Remote exploit for windows platform

                                            source: http://www.securityfocus.com/bid/58463/info

QlikView is prone to a remote integer-overflow vulnerability.

Successful attacks will allow attackers to execute arbitrary code within the context of the application. Failed exploit attempts will result in a denial-of-service condition.

QlikView 11.00 SR2 is vulnerable; other versions may also be affected. 

Vulnerability details:
The .qvw file is divided into several sections with a specified delimiter.
Among others, there is a parameter which is responsible for defining the
section length. On the hex listing below it's the DWORD A4 00 00 00 (address

000315B0:  00 00 01 00-00 00 0E 23-23 23 23 23-23 23 23 23
000315C0:  23 23 23 23-23 01 2E 00-00 00 00 00-00 00 00 00
000315D0:  00 00 00 00-00 00 00 00-00 00 00 00-00 00 00 03
000315E0:  00 00 00 00-00 00 00 90-02 00 00 00-00 04 00 A4
000315F0:  00 00 00 78-9C 3D CC CB-4A 02 50 14-86 D1 1F 47

If by any reasons the value is bigger than the actual size of the section,
an error is handled by a C++ EH and a message "Document failed to load" is
shown. The check condition can be seen here:

.text:00D6BD66                 mov     eax, [edi+28h]
.text:00D6BD69                 mov     ebx, [eax]      ; here is the length parameter
.text:00D6BD6B                 add     eax, 4
.text:00D6BD6E                 mov     [edi+28h], eax
.text:00D6BD71                 cmp     ebx, [ebp+var_14]
.text:00D6BD74                 jg      loc_D6BBAC      ; check if the parameter value 
                                                         is bigger than actual length

However, the comparison operates with a signed number and doesn't check if it's
less than zero. In other words, if an attacker supplies a DWORD bigger than
0x80000000, the jump will not be taken (as the number will be considered as 
negative), causing an integer overflow. After that, the length parameter is used 
as the DstSize argument to the CArchive::Read function:

.text:00D6BD7A                 mov     eax, [ebp+Dst]
.text:00D6BD7D                 push    ebx             ; DstSize
.text:00D6BD7E                 push    eax             ; Dst
.text:00D6BD7F                 mov     ecx, edi
.text:00D6BD81                 call    ?Read () CArchive@@QAEIPAXI () Z ; CArchive::Read(void *,uint)

A large amount of data is read. It is used later to fill the created Archive
whose size is 0x8000:
.text:00B26207                 push    0
.text:00B26209                 push    8000h
.text:00B2620E                 push    1
.text:00B26210                 lea     eax, [ebp+var_60]
.text:00B26213                 push    eax
.text:00B26214                 lea     ecx, [ebp+var_A8]
.text:00B2621A                 call    ??0CArchive@@QAE () PAVCFile@@IHPAX () Z ; 
                                                     CArchive::CArchive(CFile *,uint,int,void *)

This results in the controlled address being overwritten with the controlled

.text:009F3092                 mov     ecx, [esi]
.text:009F3094                 mov     edx, [esi+4]
.text:009F3097                 mov     [ecx+4], edx    ; here the error occurs;
.text:009F3097                                         ; trying to write at non-existing address

An extract from a debugger with the occurence of the error is presented below.

eax=04735f14 ebx=00000000 ecx=bbbbbbb7 edx=aaaaaaa6 esi=04b2fbc0 edi=04735f10
eip=01723097 esp=003527f8 ebp=00352818 iopl=0         nv up ei pl nz ac pe nc
cs=0023  ss=002b  ds=002b  es=002b  fs=0053  gs=002b             efl=00010216
01723097 895104          mov     dword ptr [ecx+4],edx ds:002b:bbbbbbbb=????????