CWE-649: Reliance on Obfuscation or Encryption of Security-Relevant Inputs without Integrity Checking
Reliance on Obfuscation or Encryption of Security-Relevant Inputs without Integrity Checking
Weakness ID: 649 (Weakness Base)
Status: Incomplete
Description
Description Summary
The software uses obfuscation or encryption of inputs that should not be mutable by an external actor, but the software does not use integrity checks to detect if those inputs have been modified.
Extended Description
When an application relies on obfuscation or incorrectly applied / weak encryption to protect client-controllable tokens or parameters, that may have an effect on the user state, system state, or some decision made on the server. Without protecting the tokens/parameters for integrity, the application is vulnerable to an attack where an adversary blindly traverses the space of possible values of the said token/parameter in order to attempt to gain an advantage. The goal of the attacker is to find another admissible value that will somehow elevate his or her privileges in the system, disclose information or change the behavior of the system in some way beneficial to the attacker. If the application does not protect these critical tokens/parameters for integrity, it will not be able to determine that these values have been tampered with. Measures that are used to protect data for confidentiality should not be relied upon to provide the integrity service.
Time of Introduction
Architecture and Design
Implementation
Applicable Platforms
Languages
All
Common Consequences
Scope
Effect
Integrity
Technical Impact: Unexpected state
The inputs could be modified without detection, causing the software
to have unexpected system state or make incorrect security
decisions.
Likelihood of Exploit
High
Enabling Factors for Exploitation
The application uses client controllable tokens/parameters in order to
make decisions on the server side about user state, system state or other
decisions related to the functionality of the application.
The application does not protect client controllable tokens/parameters for
integrity and thus not able to catch tampering.
An IPSec configuration does not perform integrity
checking of the IPSec packet as the result of either not configuring ESP
properly to support the integrity service or using AH improperly. In either
case, the security gateway receiving the IPSec packet would not validate the
integrity of the packet to ensure that it was not changed. Thus if the
packets were intercepted the attacker could undetectably change some of the
bits in the packets. The meaningful bit flipping was possible due to the
known weaknesses in the CBC encryption mode. Since the attacker knew the
structure of the packet, he or she was able (in one variation of the attack)
to use bit flipping to change the destination IP of the packet to the
destination machine controlled by the attacker. And so the destination
security gateway would decrypt the packet and then forward the plaintext to
the machine controlled by the attacker. The attacker could then read the
original message. For instance if VPN was used with the vulnerable IPSec
configuration the attacker could read the victim's e-mail. This
vulnerability demonstrates the need to enforce the integrity service
properly when critical data could be modified by an attacker. This problem
might have also been mitigated by using an encryption mode that is not
susceptible to bit flipping attacks, but the preferred mechanism to address
this problem still remains message verification for integrity. While this
attack focuses on the network layer and requires a man in the middle
scenario, the situation is not much different at the software level where an
attacker can modify tokens/parameters used by the
application.
Potential Mitigations
Phase: Architecture and Design
Protect important client controllable tokens/parameters for integrity
using PKI methods (i.e. digital signatures) or other means, and checks
for integrity on the server side.
Phase: Architecture and Design
Repeated requests from a particular user that include invalid values
of tokens/parameters (those that should not be changed manually by
users) should result in the user account lockout.
Phase: Architecture and Design
Client side tokens/parameters should not be such that it would be
easy/predictable to guess another valid state.
Phase: Architecture and Design
Obfuscation should not be relied upon. If encryption is used, it needs
to be properly applied (i.e. proven algorithm and implementation, use
padding, use random initialization vector, user proper encryption mode).
Even with proper encryption where the ciphertext does not leak
information about the plaintext or reveal its structure, compromising
integrity is possible (although less likely) without the provision of
the integrity service.
Relying on Obfuscation or
Encryption with no Integrity Checking to Protect User Controllable
Parameters that are Used to Determine User or System
State
Description
