lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [day] [month] [year] [list]
Message-ID: <157429977.20050414192848@SECURITY.NNOV.RU>
Date: Thu, 14 Apr 2005 19:28:48 +0400
From: 3APA3A <3APA3A@...URITY.NNOV.RU>
To: full-disclosure@...ts.grok.org.uk, bugtraq@...urityfocus.com,
	ntbugtraq@...tserv.ntbugtraq.com
Subject: Internet Explorer wininet.dll URL parsing memory
	corruption technical details



Vendor:         Microsoft
Software:       Internet Explorer 6.0, 5.5, 5.01
Problem:        Memory corruption, code execution
Remote:         Yes
Risk Level:     Medium to low (hard to exploit)
Authors:        Axle  (ICQ  755756) bug discovery
                3APA3A, http://www.security.nnov.ru/ bug research
Original URL:   http://www.security.nnov.ru/Idocument331.html

Details:

Axle  (ICQ  755756)  found  an  interesting  and  hard  to  catch bug in
InternetCreateUrlW  function  of wininet.dll. It's hard to find, because
in  most  cases  vulnerable  application doesn't crash and it can not be
found  with  automated  code  scanner,  because  it's not typical buffer
overflow,  but  the  bug in program logic. Buffer length is checked, but
because  of  mistyping in vulnerable code it fails to handle exceptional
conditions.  Vulnerable  piece  of code is located at address 0x771E8E85
(wininet.dll  version 6.00.2900.2180).

This  part  of  code should convert hostname from Unicode. Whole process is:

1.  Calculate  size  of  target buffer (exactly twice more than actually
required to make you buy additional memory).
2. Allocate buffer with LocalAlloc()
3. Copy string to target buffer with WideCharToMultiByte()
4. NULL-terminate target string, which is already NULL terminated to add
additional CPU cycles. It's required to make you buy faster processor.

It clearly confirms Microsoft has contract with Intel. Joke. Probably.

If translate to C this code looks like

/* 771E8E85 */
if (stringlen == 0) {/* handle exceptional conditions */}
/* the bug is here. Should be stringlen <= 0 */
else {
     ...
     /* stringlen is -1 on oversized hostname */
     buflen = (stringlen)*2 + 2 /* buflen is 0 */
     buf  =  LocalAlloc(0, buflen); /* because LMEM_MOVEABLE is not set LocalAlloc returns
                                  zero size page from heap*/
     /* 771E8EC1 */
     len   =  WideCharToMultiByte
                             (0,
                              0x400,
                              pointer_to_hostname,
                              -1  /*  NULL terminated/,
                              buf,
                              buflen /* size of allocated buf */,
                              NULL,
                              0);
              /*   because   cbMultiByte  is  0  WideCharToMultiByte simply calculates
                   required length, it's equal to real length of our hostname, it doesn't
                   change content of memory pointed by buf) */

     /* 771E8EC6 */
     buf[len]  =  0;  /*  here 0 is written to some unallocated memory and we can
                     partially control address by the length of our hostname */
 ...
}


The problem is on oversized hostname stringlen is -1, not 0 as expected.
So we have:

1. Calculated buffer size is 0.
2. LocalAlloc returns valid pointer to empty chunk.
3.  Because  targed  buffer  is  empty,  WideCharToMultiByte()  does not
actually copies any data, but only calculates length of hostname string.
It prevent direct buffer overflow.
4.  As  a result we can write single '\0' byte to partially controllable
(with hostname length) location.

Because  translated  hostname  points  to empty memory chunk it contains
some garbage. In most cases it's only visible effect (you see garbage in
address  bar of Internet Explorer). In rare cases than empty chunk is at
the end of memory page on the heap Internet Explorer crashes.

Example:

<a href='http://<buffer_of_256-300_bytes>/'>TEST (CLICK)</a>

Theoretically,  this  bug is exploitable, because we can manipulate with
memory  chunks allocating, e.g. with Javascript or page redirections. In
practice it's very hard for lazy person.

References:
1. Multiple Microsoft Internet Explorer memory corruptions
http://www.security.nnov.ru/news4675.html
2. Microsoft Security Bulletin MS05-020
Cumulative Security Update for Internet Explorer (890923)
http://www.microsoft.com/technet/security/Bulletin/MS05-020.mspx
3. URL Parsing Memory Corruption Vulnerability
http://www.cve.mitre.org/cgi-bin/cvename.cgi?name=CAN-2005-0554
4. Microsoft Windows Internet Explorer Long Hostname Heap Corruption Vulnerability
http://www.idefense.com/application/poi/display?id=229&type=vulnerabilities



-- 
http://www.security.nnov.ru
         /\_/\
        { , . }     |\
+--oQQo->{ ^ }<-----+ \
|  ZARAZA  U  3APA3A   } You know my name - look up my number (The Beatles)
+-------------o66o--+ /
                    |/

_______________________________________________
Full-Disclosure - We believe in it.
Charter: http://lists.grok.org.uk/full-disclosure-charter.html
Hosted and sponsored by Secunia - http://secunia.com/


Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ