Improper Prevention of Lock Bit Modification

Stable Base
Structure: Simple
Description

The product uses a trusted lock bit for restricting access to registers, address regions, or other resources, but the product does not prevent the value of the lock bit from being modified after it has been set.

Extended Description

In integrated circuits and hardware intellectual property (IP) cores, device configuration controls are commonly programmed after a device power reset by a trusted firmware or software module (e.g., BIOS/bootloader) and then locked from any further modification. This behavior is commonly implemented using a trusted lock bit. When set, the lock bit disables writes to a protected set of registers or address regions. Design or coding errors in the implementation of the lock bit protection feature may allow the lock bit to be modified or cleared by software after it has been set. Attackers might be able to unlock the system and features that the bit is intended to protect.

Common Consequences 1
Scope: Access Control

Impact: Modify Memory

Registers protected by lock bit can be modified even when lock is set.

Detection Methods 1
Manual AnalysisHigh
Set the lock bit. Power cycle the device. Attempt to clear the lock bit. If the information is changed, implement a design fix. Retest. Also, attempt to indirectly clear the lock bit or bypass it.
Potential Mitigations 1
Phase: Architecture and DesignImplementationTesting
- Security lock bit protections must be reviewed for design inconsistency and common weaknesses. - Security lock programming flow and lock properties must be tested in pre-silicon and post-silicon testing.

Effectiveness: High

Demonstrative Examples 2
Consider the example design below for a digital thermal sensor that detects overheating of the silicon and triggers system shutdown. The system critical temperature limit (CRITICAL_TEMP_LIMIT) and thermal sensor calibration (TEMP_SENSOR_CALIB) data have to be programmed by firmware, and then the register needs to be locked (TEMP_SENSOR_LOCK).

Code Example:

Bad
Other
RegisterField description
CRITICAL_TEMP_LIMIT[31:8] Reserved field; Read only; Default 0 [7:0] Critical temp 0-255 Centigrade; Read-write-lock; Default 125
TEMP_SENSOR_CALIB[31:0] Thermal sensor calibration data. Slope value used to map sensor reading to degrees Centigrade.
TEMP_SENSOR_LOCK[31:1] Reserved field; Read only; Default 0 [0] Lock bit, locks CRITICAL_TEMP_LIMIT and TEMP_SENSOR_CALIB registers; Write-1-once; Default 0
TEMP_HW_SHUTDOWN[31:2] Reserved field; Read only; Default 0 [1] Enable hardware shutdown on critical temperature detection; Read-write; Default 0
CURRENT_TEMP[31:8] Reserved field; Read only; Default 0 [7:0] Current Temp 0-255 Centigrade; Read-only; Default 0
In this example, note that if the system heats to critical temperature, the response of the system is controlled by the TEMP_HW_SHUTDOWN bit [1], which is not lockable. Thus, the intended security property of the critical temperature sensor cannot be fully protected, since software can misconfigure the TEMP_HW_SHUTDOWN register even after the lock bit is set to disable the shutdown response.

Code Example:

Good
Other

To fix this weakness, one could change the TEMP_HW_SHUTDOWN field to be locked by TEMP_SENSOR_LOCK.

| | | | TEMP_HW_SHUTDOWN | [31:2] Reserved field; Read only; Default 0 [1] Enable hardware shutdown on critical temperature detection; Read-write-Lock; Default 0 [0] Locked by TEMP_SENSOR_LOCK |

The following example code is a snippet from the register locks inside the buggy OpenPiton SoC of HACK@DAC'21 [REF-1350]. Register locks help prevent SoC peripherals' registers from malicious use of resources. The registers that can potentially leak secret data are locked by register locks.
In the vulnerable code, the reglk_mem is used for locking information. If one of its bits toggle to 1, the corresponding peripheral's registers will be locked. In the context of the HACK@DAC System-on-Chip (SoC), it is pertinent to note the existence of two distinct categories of reset signals.
First, there is a global reset signal denoted as "rst_ni," which possesses the capability to simultaneously reset all peripherals to their respective initial states.
Second, we have peripheral-specific reset signals, such as "rst_9," which exclusively reset individual peripherals back to their initial states. The administration of these reset signals is the responsibility of the reset controller module.

Code Example:

Bad
Verilog
verilog

if(~(rst_ni && ~jtag_unlock && ~rst_9))**

verilog
In the buggy SoC architecture during HACK@DAC'21, a critical issue arises within the reset controller module. Specifically, the reset controller can inadvertently transmit a peripheral reset signal to the register lock within the user privilege domain.
This unintentional action can result in the reset of the register locks, potentially exposing private data from all other peripherals, rendering them accessible and readable.
To mitigate the issue, remove the extra reset signal rst_9 from the register lock if condition. [REF-1351]

Code Example:

Good
Verilog
verilog

if(~(rst_ni && ~jtag_unlock))**

verilog
Observed Examples 1
CVE-2017-6283chip reset clears critical read/write lock permissions for RSA function
Applicable Platforms
Languages:
Not Language-Specific : Undetermined
Technologies:
Not Technology-Specific : Undetermined
Modes of Introduction
Architecture and Design
Implementation
Related Attack Patterns
Related Weaknesses