CWE-1257: Improper Access Control Applied to Mirrored or Aliased Memory Regions

How it's found

Improper Access Control Applied to Mirrored or Aliased Memory Regions is typically found by tracing untrusted input from where it enters the system to the point where it is used without the check or neutralization this weakness describes, combining manual code review with dynamic testing.

Hardware product designs often need to implement memory protection features that enable privileged software to define isolated memory regions and access control (read/write) policies. Isolated memory regions can be defined on different memory spaces in a design (e.g. system physical address, virtual address, memory mapped IO). Each memory cell should be mapped and assigned a system address that the core software can use to read/write to that memory. It is possible to map the same memory cell to multiple system addresses such that read/write to any of the aliased system addresses would be decoded to the same memory cell. This is commonly done in hardware designs for redundancy and simplifying address decoding logic. If one of the memory regions is corrupted or faulty, then that hardware can switch to using the data in the mirrored memory region. Memory aliases can also be created in the system address map if the address decoder unit ignores higher order address bits when mapping a smaller address region into the full system address. A common security weakness that can exist in such memory mapping is that aliased memory regions could have different read/write access protections enforced by the hardware such that an untrusted agent is blocked from accessing a memory address but is not blocked from accessing the corresponding aliased memory address. Such inconsistency can then be used to bypass the access protection of the primary memory block and read or modify the protected memory. An untrusted agent could also possibly create memory aliases in the system address map for malicious purposes if it is able to change the mapping of an address region or modify memory region sizes.

Vulnerable vs. safe

In this design the aliased memory address ranges are these:

0x0000 - 0x0FFF

0x1000 - 0x1FFF

0x2000 - 0x2FFF

0x3000 - 0x3FFF

The same register can be accessed using four different addresses: 0x0000, 0x1000, 0x2000, 0x3000.

The system address filter only blocks access to range 0x0000 - 0x0FFF and does not block access to the aliased addresses in 0x1000 - 0x3FFF range. Thus, untrusted software can leverage the aliased memory addresses to bypass the memory protection.

In this design the aliased memory addresses (0x1000 - 0x3FFF) could be blocked from all system software access since they are not used by software.

Alternately, the MPU logic can be changed to apply the memory protection policies to the full address range mapped to Unit_A (0x0000 - 0x3FFF).

Consequences

• Read Memory
• Modify Memory
• DoS: Instability

Mitigations

• Architecture and Design/Implementation: The checks should be applied for consistency access rights between primary memory regions and any mirrored or aliased memory regions. If different memory protection units (MPU) are protecting the aliased regions, their protected range definitions and policies should be synchronized.
• Architecture and Design/Implementation: The controls that allow enabling memory aliases or changing the size of mapped memory regions should only be programmable by trusted software components.

Where this fits in a TurboPentest engagement

This weakness is not covered by the automated black-box pentest. IntegSec pentesters cover it in a manual engagement.

Frequently asked questions

Related CWEs

• Pillar weakness
  CWE-284: Improper Access Control