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: <20030120212523.GA17993@php.net>
From: s.esser at e-matters.de (Stefan Esser)
Subject: Advisory 01/2003: CVS remote vulnerability

                           e-matters GmbH
                          www.e-matters.de

                      -= Security  Advisory =-



     Advisory: CVS remote vulnerability
 Release Date: 2003/01/20
Last Modified: 2003/01/20
       Author: Stefan Esser [s.esser@...atters.de]

  Application: CVS <= 1.11.4
     Severity: A vulnerability within CVS allows remote compromise of
               CVS servers.
         Risk: Critical
Vendor Status: Vendor has released a bugfixed version.
    Reference: http://security.e-matters.de/advisories/012003.html


Overview:

   Concurrent Versions System (CVS) is the dominant open-source version 
   control software that allows developers to access the latest code using
   a network connection. CVS version 1.11.4 and below contain a flaw that
   can be used by a remote attacker to execute arbitrary code on the server.
      
   You should also note, that the CVS client/server protocol includes two 
   commands (Update-prog and Checkin-prog) that can be used by any CVS user
   with write access to the repository to execute arbitrary shell commands
   on the server. This is a questionable feature, because it is very badly
   documented, is unknown to most CVS administrators and cannot be turned
   off within the configuration files.
   
   
Details:
   
   While auditing the CVS sourcetree I found a flaw within the handling of
   the Directory request within the server code. By sending a malformed 
   directory name it is possible to trigger an error condition that will 
   make the function return at a point where a global pointer variable is 
   already freed and has not got a new value assigned yet. This will result
   in a classical double-free() when the next Directory request is handled.
   With the help of other CVS requests it is possible to either leak some
   information that could be used to determine the heap position or to
   execute arbitrary code on systems that are known to be vulnerable to
   this kind of bugs. This includes Linux, Solaris and most probably Windows
   systems. 
   
   Additionally I was able to create proof of concept code that uses this
   vulnerability to execute arbitrary shell commands on BSD servers. I was
   able to achieve this because all allocated memory is aligned on BSD 
   systems which makes it very easy to get newly allocated memory blocks 
   into the same position of already freed blocks of the same slotsize.
   In combination with some CVS requests that work on lists of pointers,
   I was able to use this bug to free arbitrary memory addresses. With the
   help of the information leak capabilities of this vulnerability it is 
   possible to guess the address of some strings that are needed for the 
   read/write access checks. Combined this allowes to bypass the write 
   access checks and to abuse the Update-prog/Checkin-prog requests to 
   execute arbitrary commands on the server with an anonymous read-only
   account.
   
   The impact of this vulnerability depends highly on the configuration of
   the server. The CVS server is by default started via inetd with root 
   privileges. If CVSROOT/passwd is left writeable to the CVS user this means
   a remote root compromise. You must also consider that chrooting the CVS
   daemon may protect the rest of your system against the intruder but will
   still leave the whole source tree vulnerable to the attacker. 

   Summarized this means that this vulnerability is a threat to most open
   source projects because nearly all of them offer anonymous CVS access to
   the source tree. Even if the attacker is not able to extend his attack
   on the developer CVS server (if it is seperated at all) he could still
   backdoor everything other people download from the anonymous server.


Proof of Concept:

   e-matters is not going to release an exploit for this vulnerability to
   the public.
   

Disclosure Timeline:

   04. January 2003 - Vendor was notified via email. Unfourtunately the
                      person that I tried to contact was on vacation, so I
                      received no answer.
   12. January 2003 - The vulnerability was disclosed to the admins of several
                      big public CVS repositories and to some distributors.
   15. January 2003 - Vendor has committed the fix to the CVS CVS repository.
   16. January 2003 - Vendor-sec was notified that a new bugfixed CVS version
                      will be released on 20th January.
   20. January 2003 - Vendor has released a new version which fixes the double
                      free problem. You can download it at:
                      http://ccvs.cvshome.org/servlets/ProjectDownloadList

   
CVE Information:

   The Common Vulnerabilities and Exposures project (cve.mitre.org) has
   assigned the name CAN-2003-0015 to this issue.


Recommendation:

   My recommendation is to immediantly update to the new version. You may also
   consider applying my patch which adds the ability to turn off Update-prog
   and Checkin-prog within your configuration files. You can download it from
   
   http://security.e-matters.de/patches/cvs_disablexprog.diff
   
   You should also consider running your CVS server chrooted over SSH instead
   of using the :pserver: method. You can find a tutorial how to setup such a
   server at
   
   http://www.netsys.com/library/papers/chrooted-ssh-cvs-server.txt
   
   
GPG-Key:

   http://security.e-matters.de/gpg_key.asc
    
   pub  1024D/75E7AAD6 2002-02-26 e-matters GmbH - Securityteam
   Key fingerprint = 43DD 843C FAB9 832A E5AB  CAEB 81F2 8110 75E7 AAD6


Copyright 2003 Stefan Esser. All rights reserved.

-- 

--------------------------------------------------------------------------
 Stefan Esser                                        s.esser@...atters.de
 e-matters Security                         http://security.e-matters.de/

 GPG-Key                gpg --keyserver pgp.mit.edu --recv-key 0xCF6CAE69 
 Key fingerprint       B418 B290 ACC0 C8E5 8292  8B72 D6B0 7704 CF6C AE69
--------------------------------------------------------------------------
 Did I help you? Consider a gift:            http://wishlist.suspekt.org/
--------------------------------------------------------------------------

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 240 bytes
Desc: not available
Url : http://lists.grok.org.uk/pipermail/full-disclosure/attachments/20030120/7703bb93/attachment.bin

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ