Time-Driven Cache Attacks
Published 2015 · Computer Science
When a load instruction incurs a cache-miss, a block of memory from the lower level of the memory subsystem is loaded into a cache line. Consequently, the memory access would require considerably more time and power, and has a characteristically different electromagnetic radiation compared to when a cache-hit occurs. The loads that result in cache misses are easily distinguishable from the cache hits. These indirect manifestations of a memory access can be used by an attacker to gain considerable insight about the application currently being executed. In this chapter, we show how information about the secret key of a cipher can be gleaned from the execution time of a block cipher. We start the chapter with a simple illustration showing how information can be obtained from memory access patterns before discussing attacks on ciphers.