Projects per year
Memory encryption is an effective security building block broadly available on commodity systems from Intel® and AMD. Schemes, such as Intel® TME-MK and AMD SEV, help provide data confidentiality and integrity, enabling cryptographic isolation of workloads on shared platforms. However, due to their coarse encryption granularity (i.e., pages or entire virtual machines), these hardware-enabled primitives cannot unleash their full potential to provide protection for other security applications, such as memory safety. To this end, we present a novel approach to achieving sub-page-granular memory encryption without hardware modifications on off-the-shelf systems featuring Intel®’s TME-MK. We showcase how to utilize our fine-grained memory encryption approach for memory safety by introducing MEMES. MEMES is capable of mitigating both spatial and temporal heap memory vulnerabilities by encrypting individual memory objects with different encryption keys. Compared to other hardware-based memory safety schemes, our approach works on existing commodity hardware, which allows easier adoption. Our extensive analysis attests to the strong security benefits which are provided at a geometric mean runtime overhead of just 16–27%.
|Title of host publication||SECRYPT 2023: 20th International Conference on Security and Cryptography|
|Publication status||Accepted/In press - 23 Apr 2023|
|Event||20th International Conference on Security and Cryptography: SECRYPT 2023 - Rome, Italy|
Duration: 10 Jul 2023 → 12 Jul 2023
|Conference||20th International Conference on Security and Cryptography: SECRYPT 2023|
|Period||10/07/23 → 12/07/23|
FingerprintDive into the research topics of 'MEMES: Memory Encryption-based Memory Safety on Commodity Hardware'. Together they form a unique fingerprint.
- 1 Active