CWE-226: Sensitive Information in Resource Not Removed… | Glexia

Causes

• This example shows how an attacker can take advantage of an incorrect state transition. Suppose a device is transitioning from state A to state B. During state A, it can read certain private keys from the hidden fuses that are only accessible in state A but not in state B. The device reads the keys, performs operations using those keys, then transitions to state B, where those private keys should no longer be accessible. After the transition to state B, even though the private keys are no longer accessible directly from the fuses in state B, they can be accessed indirectly by reading the memory that contains the private keys.
• The following code calls realloc() on a buffer containing sensitive data: There is an attempt to scrub the sensitive data from memory, but realloc() is used, so it could return a pointer to a different part of memory. The memory that was originally allocated for cleartext_buffer could still contain an uncleared copy of the data.
• The following example code is excerpted from the AES wrapper/interface, aes0_wrapper, module of one of the AES engines (AES0) in the Hack@DAC'21 buggy OpenPiton System-on-Chip (SoC). Note that this SoC contains three distinct AES engines. Within this wrapper module, four 32-bit registers are utilized to store the message intended for encryption, referred to as p_c[i]. Using the AXI Lite interface, these registers are filled with the 128-bit message to be encrypted. The above code snippet [REF-1402] illustrates an instance of a vulnerable implementation of the AES wrapper module, where p_c[i] registers are cleared at reset. Otherwise, p_c[i]registers either maintain their old values (if reglk_ctrl_i[3]is true) or get filled through the AXI signal wdata. Note that p_c[i]registers can be read through the AXI Lite interface (not shown in snippet). However, p_c[i] registers are never cleared after their usage once the AES engine has completed the encryption process of the message. In a multi-user or multi-process environment, not clearing registers may result in the attacker process accessing data left by the victim, leading to data leakage or unintentional information disclosure. To fix this issue, it is essential to ensure that these internal registers are cleared in a timely manner after their usage, i.e., the encryption process is complete. This is illustrated below by monitoring the assertion of the cipher text valid signal, ct_valid [REF-1403].

Remediation

• Architecture and Design Implementation: During critical state transitions, information not needed in the next state should be removed or overwritten with fixed patterns (such as all 0's) or random data, before the transition to the next state.
• Architecture and Design Implementation: When releasing, de-allocating, or deleting a resource, overwrite its data and relevant metadata with fixed patterns or random data. Be cautious about complex resource types whose underlying representation might be non-contiguous or change at a low level, such as how a file might be split into different chunks on a file system, even though "logical" file positions are contiguous at the application layer. Such resource types might require invocation of special modes or APIs to tell the underlying operating system to perform the necessary clearing, such as SDelete (Secure Delete) on Windows, although the appropriate functionality might not be available at the application layer.

Detection

• Manual Analysis: Write a known pattern into each sensitive location. Trigger the release of the resource or cause the desired state transition to occur. Read data back from the sensitive locations. If the reads are successful, and the data is the same as the pattern that was originally written, the test fails and the product needs to be fixed. Note that this test can likely be automated.
• Automated Static Analysis: Automated static analysis, commonly referred to as Static Application Security Testing (SAST), can find some instances of this weakness by analyzing source code (or binary/compiled code) without having to execute it. Typically, this is done by building a model of data flow and control flow, then searching for potentially-vulnerable patterns that connect "sources" (origins of input) with "sinks" (destinations where the data interacts with external components, a lower layer such as the OS, etc.)

Related CWEs

No related CWE relationships are published yet.

Related CVEs

Related CVE mappings appear after CVE records are cross-indexed.

ATT&CK Relevance

ATT&CK relevance is shown only when reviewed or responsibly inferred.