full-disclosure - [FD] [CVE-2014-3260] Crypto implementation flaws in Pacom GMS System

Open Source and information security mailing list archives

Message-ID: <cdc9a9b298acc969a510f0eb7c09e679@xpd.se>
Date: Wed, 09 Dec 2015 17:37:28 +0100
From: XPD Advisories Team <security-alert@....se>
To: fulldisclosure@...lists.org
Subject: [FD] [CVE-2014-3260] Crypto implementation flaws in Pacom GMS System

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

                       XPD - XPD Advisory
                         https://xpd.se

         Crypto implementation flaws in Pacom GMS System

Advisory ID: XPD-2015-001
CVE reference: CVE-2014-3260
Affected platforms: Pacom 1000 CCU ("Base Station") and Controllers 
(RTU)
Version: All versions are affected
Date: 2013-Oktober-10
Security risk: High
Vulnerability: Crypto implementation flaws in Pacom GMS System
Researcher: Joachim Strombergson, Fredrik Soderblom, Peter Norin
Vendor Status: Notified / Patch available
Vulnerability Disclosure Policy:
  https://xpd.se/advisories/xpd-disclosure-policy-01.txt
Permanent URL:
  https://xpd.se/advisories/XPD-2015-001.txt

=====================================================================

Summary:

The Pacom 1000 CCU and controllers (RTU) is used in security alarm
  installations all over the world. The flaws we have found can bypass
  the security of any unpatched installation.

"The purpose of a Pacom Controller is to control, monitor, and maintain
  security for a remote site. It is located at the site itself and 
performs
  all alarm and door control functions."
  - http://www.pacom.com/field-controllers.php

"Pacom security solutions are installed in over twenty countries on 
seven
  continents." - http://www.pacom.com/our-customers.php

Detailed description:

The Pacom 1000 implementation have several serious implementation flaws.

These vulnerabilities could in a worst case scenario lead to a full
  compromise of the protocol between the controller and the base station,
  rendering an alarm system useless.

Potentially a large number of sites could be affected by the described 
flaws.

PRNG:

The PRNG used is of a type known as a Linear Congruential Generator.
  This type of generator are known to provide random numbers with less
  than perfect uniform distribution. The PRNG is a 16-bit generator.
  This means that the generator can only generate 2**16 numbers in a
  sequence before it must be reseeded. There is no information about
  how the generator is seeded from start nor how it is reseeded.

A simulation in Dieharder shows that the used algorithm fails every
  test except for one, where it receives the result 'Weak'.

The Linear Congruential Generator can be broken by observing values
  generated by consecutive iterations of the PRNG. The system creates
  32-bit random numbers by extracting 8-bits from four consecutive
  16-bit words numbers generated by the PRNG. This means that by
  observing a single 32-bit word, an attacker has in fact half the state
  information (8 out of 16 bits) from four iterations of the generator.

MAC:

A Message Authentication Code (MAC) is generated and added to each
  message sent between CCU and Controller. The MAC generator generator
  used is not based on any well-known secure MAC functionality such as
  HMAC or OMAC. Furthermore the generated MAC is only 32 bit.

Master Code:

There is a functionality for substitution detection. According to
  Pacom the functionality is based on a proprietary Pacom encryption
  method. Key to the functionality is a a 24-bit randomly generated
  value called Master Code. The Master Code is also used to generate
  the 128-bit AES key used with the substitution detection algorithm.
  Hence the effective strength of the key is not 128 bits, nor 104 bits
  (128 - 24) but 24 bits. A very short key with low security.

Unfortunately it appears that the aforementioned (16 bit only),
  less than optimal, PRNG is used to generate the Master Code, thus
  reducing it's effective strength to 16 bit.

The Master Code is distributed from CCU to CPU-cards and other CCUs
  as well as GMS units (for logging purposes) in clear text. This means
  that the code potentially is sent unprotected over private networks,
  corporate networks, public networks etc.

Substitution detection:

According to Pacom documentation the "substitution detection involves
  appending a 128-bit check code to the controller heartbeat response
  messages. The check code is calculated from a combination of a
  hard-coded constant value, the controllers master code, and the
  message data. In essence it is another type of MAC, but one that
  employs the master code."

The implementation of the substitution detection uses a "check code"
  which is said to be 128-bits long and is appended to response messages.
  However due to a design flaw, the code is only 64 bit.

In total the heartbeat response message is 5 bytes (40 bits) long:

  Byte 1: The message type (e.g. heartbeat response)
  Byte 2: A value based on random numbers sent in the heartbeat command 
from the CCU
  Byte 3: The controller summary status
  Byte 4: The heartbeat sequence number (zero or one)
  Byte 5: Always zero

Of the five bytes in the heartbeat response message, two bytes (4 and 5)
  are either one or zero, or always zero. Byte 3 is a simple status. So,
  of 40 bits, 32 bits are most likely predictable and the remaining 8 
bits
  is probably choosen based on the weak PRNG. This means that a big part
  of the response  message can be guessed.

The so called "128-bit check code" is then calculated over these 5 bytes
  using the aforementioned flawed Master Code and a 2 byte address of the
  controller, forming a 40 bit key, which is used with a hard coded 
constant
  to form an AES-128 key.

The resulting "128-bit check code" from the AES encryption is XOR:ed 
with
  its own cleartext. This means that there is a direct path from 
cleartext to
  ciphertext bypassing the AES encryption. This leaks information about
  the cleartext as well as opening up for chosen plaintext attacks.

Hard coded constants:

The security functionality uses several hard coded, secret constants for
  random number generation, MAC calculation, Substitution detection
  algorithm etc. Unfortunately, the way these constants are used,
  information about them are leaked through the very messages, which
  opens up for recovery of the constants. If the constants are recovered
  and thus system security is lost, the units must be reprogrammed in
  the field or even replaced.

=====================================================================
Conclusion:

We do not recommend relying on the security features in the system and
  the system should be viewed as an unprotected system.

If the system is to be used, separate communication security mechanism
  should be added.

However, the usage of hard coded constants in the units and the 
associated
  need for field service or replacement if a breach occurs, makes us
  recommend that the system needs severe redesign before it is ready for
  production use.

We questions if the system has been designed with any knowledge of what
  has been known good security practices since at least 30 years, nor 
good
  engineering practice.

=====================================================================
Versions affected:

All versions of Pacom 1000 (CCU & RTU) - According to Pacom, this 
firmware
  will not be fixed. Customers are advised to switch to the EMCS
  platform instead.

All versions of EMCS (Pacom .is) prior to 1.3

The vendor reports that the following versions are patched:
  EMCS (Pacom .is) version 1.3 and above

=====================================================================
Credits

This vulnerability was discovered and researched by Joachim Strombergson
  from Assured AB, Fredrik Soderblom and Peter Norin from XPD AB.

=====================================================================
References

https://en.wikipedia.org/wiki/Linear_congruential_generator
https://en.wikipedia.org/wiki/Diehard_tests

=====================================================================
History

2013-10-10 Initial Discovery
2013-10-22 Initial attempt to contact the vendor
2013-11-12 Reply from Niscayah, case is assigned to internal resource
2014-05-07 CVE-2014-3260 is assigned
2014-06-05 Draft of the advisory sent to the vendor
2014-09-01 Pacom notifies us that fixed firmware (EMCS only) is ready
2015-12-08 Public disclosure

=====================================================================
About Assured

Assured AB is a privately held company with headquarters in Gothenburg,
  Sweden. Established in 2015, Assured is an independant security
  consultancy firm that provides expert knowledge, advisory and
  design of IT- security solutions.

http://assured.se

About XPD

XPD AB is a privately held company with Headquarters in Stockholm, 
Sweden.
  Established in 2002, XPD AB is an independent security consulting and
  research firm, with a focus on security and perimeter security 
solutions.

https://xpd.se

=====================================================================
Disclaimer and Copyright

The information provided in this advisory is provided "as is" without
  warranty of any kind. XPD AB and Assured AB disclaims all warranties,
  either express or implied, including the warranties of merchantability 
and
  fitness for a particular purpose. In no event shall XPD AB or Assured 
AB,
  or its suppliers be liable for any damages whatsoever including direct,
  indirect, incidental, consequential, loss of business profits or
  special damages, even if XPD AB or Assured AB, or its suppliers have 
been
  advised of the possibility of such damages.

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