Device Unlock Credential Sharing
The credentials necessary for unlocking a device are shared across multiple parties and may expose sensitive information.
"Unlocking a device" often means activating certain unadvertised debug and manufacturer-specific capabilities of a device using sensitive credentials. Unlocking a device might be necessary for the purpose of troubleshooting device problems. For example, suppose a device contains the ability to dump the content of the full system memory by disabling the memory-protection mechanisms. Since this is a highly security-sensitive capability, this capability is "locked" in the production part. Unless the device gets unlocked by supplying the proper credentials, the debug capabilities are not available. For cases where the chip designer, chip manufacturer (fabricator), and manufacturing and assembly testers are all employed by the same company, the risk of compromise of the credentials is greatly reduced. However, the risk is greater when the chip designer is employed by one company, the chip manufacturer is employed by another company (a foundry), and the assemblers and testers are employed by yet a third company. Since these different companies will need to perform various tests on the device to verify correct device function, they all need to share the unlock key. Unfortunately, the level of secrecy and policy might be quite different at each company, greatly increasing the risk of sensitive credentials being compromised.
Impact: Modify MemoryRead MemoryModify Files or DirectoriesRead Files or DirectoriesModify Application DataExecute Unauthorized Code or CommandsGain Privileges or Assume IdentityBypass Protection Mechanism
Once unlock credentials are compromised, an attacker can use the credentials to unlock the device and gain unauthorized access to the hidden functionalities protected by those credentials.
Code Example:
Suppose a semiconductor chipmaker, "C", uses the foundry "F" for fabricating its chips. Now, F has many other customers in addition to C, and some of the other customers are much smaller companies. F has dedicated teams for each of its customers, but somehow it mixes up the unlock credentials and sends the unlock credentials of C to the wrong team. This other team does not take adequate precautions to protect the credentials that have nothing to do with them, and eventually the unlock credentials of C get leaked.
Code Example:
Vertical integration of a production company is one effective method of protecting sensitive credentials. Where vertical integration is not possible, strict access control and need-to-know are methods which can be implemented to reduce these risks.