核心概念
BackCache aims to prevent contention-based cache timing attacks by always achieving cache hits instead of misses and hiding cache line evictions.
摘要
The article discusses the vulnerabilities of caches to timing attacks, proposing BackCache as a hardware-software co-design to mitigate these attacks. It introduces a fully associative backup cache, RURP replacement policy, and dynamic resizing mechanism for enhanced security. Evaluation on gem5 simulator shows performance degradation but effective attack prevention.
- Caches reduce CPU-memory speed gap for efficiency.
- Attackers exploit cache timing differences for data theft.
- Existing countermeasures have limitations or overheads.
- BackCache proposes innovative design for attack mitigation.
- Security analysis and evaluation demonstrate effectiveness.
統計資料
BackCache can degrade performance by 1.33%, 7.34%, and 7.59% For OS kernel, single-thread, and multi-thread benchmarks.