Scanclave: Verifying Application Runtime Integrity in Untrusted Environments

Abstract: Data hosted in a cloud environment can be subject to attacks from a higher privileged adversary, such as a malicious or compromised cloud provider. To provide confidentiality and integrity even in the presence of such an adversary, a number of Trusted Execution Environments (TEEs) have been developed. A TEE aims to protect data and code within its environment against high privileged adversaries, such as a malicious operating system or hypervisor. While mechanisms exist to attest a TEE's integrity at load time, there are no mechanisms to attest its integrity at runtime. Additionally, work also exists that discusses mechanisms to verify the runtime integrity of programs and system components. However, those verification mechanisms are themselves not protected against attacks from a high privileged adversary. It is therefore desirable to combine the protection mechanisms of TEEs with the ability of application runtime integrity verification. In this paper, we present Scanclave, a lightweight design which achieves three design goals: Trustworthiness of the verifier, a minimal trusted software stack and the possibility to access an application's memory from a TEE. Having achieved our goals, we are able to verify the runtime integrity of applications even in the presence of a high privileged adversary. We refrain from discussing which properties define the runtime integrity of an application, as different applications will require different verification methods. Instead, we show how Scanclave enables a remote verifier to determine the runtime integrity of an application. Afterwards, we perform a security analysis for the different steps of our design. Additionally, we discuss different enclave implementations that might be used for the implementation of Scanclave.