Robust and Efficient Elimination of Cache and Timing Side Channels (2015)

Re:Robust and Efficient Elimination of Cache and Timing Side... by tptacek on Friday January 05, 2018 @10:07PM (#16083623)

Practical secret extraction from caches --- practical as in "papers documented C programs the authors wrote to verify their claims" --- go back to at least 2005.

Here's a very well-known paper: https://www.cs.tau.ac.il/~tromer/papers/cache.pdf

It's the speculative execution side channel that's new. What we didn't fully grok was the idea that code that never really runs could leave predictable cache footprints.

[ Parent | on HN ]

Re:Robust and Efficient Elimination of Cache and Timing Side... by cryptonector on Saturday January 06, 2018 @12:40AM (#16084030)

It's so obvious now, in retrospect :(

[ Parent | on HN ]

Re:Robust and Efficient Elimination of Cache and Timing Side... by dgoldstein0 on Sunday January 07, 2018 @02:15AM (#16089721)

and that you can poison prediction state (branch predictor / branch target buffer etc) and cache state to force longer periods of processor speculation, leaving more room for unintended side effects.

[ Parent | on HN ]