Speakers: Ben Gras, Kaveh Razavi, brainsmoke, Antonio Barresi We are 4 security researchers who have collectively worked on 3 different attack techniques that all (ab)use memory deduplication in one way or another. There is a cross-vm data leak attack, a cross-vm data write attack, and an in-sandbox (MS Edge) Javascript data leak + full memory read/write attack based in MS Edge. In this talk we detail how memory deduplication works and the many different ways it is exploited in our attacks. Memory deduplication is a widely applied technique to reduce memory consumption in servers, VM hosts, desktop systems and even mobile devices. Deduplication maps multiple identical copies of a physical page onto a single shared copy with copy-on-write semantics. As a result, a write to such a shared page triggers a page fault and is thus measurably slower than a write to a unshared page. Prior work has shown that an attacker able to craft pages on the target system can use this timing difference as a simple single-bit side channel to discover that certain pages exist in the system. In this talk, we show that the security implications of using memory deduplication are much more severe than initially assumed. We show that by maliciously programming memory deduplication, an attacker can build primitives to read arbitrary data from memory and even write to memory in a limited but powerful way. We exemplify these primitives using three attacks that we have recently developed. The first attack, CAIN, uses memory deduplication to brute-force ASLR’s entropy bits from a co-hosted victim VM. The second attack, Dedup Est Machina, extends CAIN in order to leak arbitrary data such as ASLR heap/code pointers and password hashes in a victim’s browser from JavaScript. Using the leaked pointers, Dedup Est Machina uses a Rowhammer exploit to own Microsoft Edge without relying on a single software vulnerability. The third attack, Flip Feng Shui, uses memory deduplication to control the p