International Association
for Cryptologic Research

Speculative out-of-order execution is a strategy of masking execution latency by allowing younger instructions to execute before older instructions. While originally considered to be innocuous, speculative out-of-order execution was brought into the spotlight with the 2018 publication of the Spectre and Meltdown attacks. These attacks demonstrated that microarchitectural side channels can leak sensitive data accessed by speculatively executed instructions that are not part of the normal program execution. Since then, a significant effort has been vested in investigating how microarchitectural side channels can leak data from speculatively executed instructions and how to control this leakage. However, much less is known about how speculative out-of-order execution affects microarchitectural side-channel attacks.In this paper, we investigate how speculative out-of-order execution affects the Evict+Time cache attack. Evict+Time is based on the observation that cache misses are slower than cache hits, hence by measuring the execution time of code, an attacker can determine if a cache miss occurred during the execution. We demonstrate that, due to limited resources for tracking out-of-order execution, under certain conditions an attacker can gain more fine-grained information and determine whether a cache miss occurred in part of the executed code.Based on the observation, we design the Evict+Spec+Time attack, a variant of Evict+Time that can learn not only whether a cache miss occurred, but also in which part of the victim code it occurred. We demonstrate that Evict+Spec+Time is an order of magnitude more efficient than Evict+Time when attacking a T-tables-based implementation of AES. We further show an Evict+Spec+Time attack on an S-boxbased implementation of AES, recovering the key with as little as 14 815 decryptions. To the best of our knowledge, ours is the first successful Evict+Time attack on such a victim.

Protocols for secure multiparty computation enable a set of parties to compute a function of their inputs without revealing anything but the output. The security properties of the protocol must be preserved in the presence of adversarial behavior. The two classic adversary models considered are semi-honest (where the adversary follows the protocol specification but tries to learn more than allowed by examining the protocol transcript) and malicious (where the adversary may follow any arbitrary attack strategy). Protocols for semi-honest adversaries are often far more efficient, but in many cases the security guarantees are not strong enough. In this paper, we present new protocols for securely computing any functionality represented by an arithmetic circuit, assuming an honest majority exists . We utilize a new method for verifying that the adversary does not cheat, that yields a cost of just twice that of semi-honest protocols in some settings. Our protocols are information-theoretically secure in the presence of malicious adversaries. We present protocol variants for small and large fields, and show how to efficiently instantiate them based on replicated secret sharing and Shamir secret sharing. In particular, for large fields, our protocol requires each party to send just  2 field elements per multiplication gate in the three-party setting, and just  12 field elements per multiplication gate for any number of parties. As with previous works in this area aiming to achieve high efficiency, our protocol is secure with abort and does not achieve fairness, meaning that the adversary may receive output while the honest parties do not. We implemented our protocol and ran experiments for different numbers of parties, different network configurations and different circuit depths. Our protocol significantly outperforms the previous best for this setting (Lindell and Nof, CCS 2017); for a large number of parties (e.g., 100 parties), our implementation runs almost an order of magnitude faster than theirs.

Oblivious RAM (ORAM) is a powerful technique to prevent harmful data breaches. Despite tremendous progress in improving the concrete performance of ORAM, it remains too slow for use in many practical settings; recent breakthroughs in lower bounds indicate this inefficiency is inherent for ORAM and even some natural relaxations. This work introduces snapshot-oblivious RAMs, a new secure memory access primitive. Snapshot-oblivious RAMs bypass lower bounds by providing security only for transcripts whose length (call it c) is fixed and known ahead of time. Intuitively, snapshot-oblivious RAMs provide strong security for attacks of short duration, such as the snapshot attacks targeted by many encrypted databases. We give an ORAM-style definition of this new primitive, and present several constructions. The underlying design principle of our constructions is to store the history of recent operations in a data structure that can be accessed obliviously. We instantiate this paradigm with data structures that remain on the client, giving a snapshot-oblivious RAM with constant bandwidth overhead. We also show how these data structures can be stored on the server and accessed using oblivious memory primitives. Our most efficient instantiation achieves O(log c) bandwidth overhead. By extending recent ORAM lower bounds, we show this performance is asymptotically optimal. Along the way, we define a new hash queue data structure—essentially, a dictionary whose elements can be modified in a first-in-first-out fashion—which may be of independent interest.

Over the past two decades, cache attacks have been identified as a threat to the security of cipher implementations. These attacks recover secret information by combining observations of the victim cache accesses with the knowledge of the internal structure of the cipher. So far, cache attacks have been applied to ciphers that have fixed state transformations, leaving open the question of whether using secret, key-dependent transformations enhances the security against such attacks. In this paper we investigate this question. We look at an implementation of the North Korean cipher Pilsung, as reverse-engineered by Kryptos Logic. Like AES, Pilsung is a permutation-substitution cipher, but unlike AES, both the substitution and the permutation steps in Pilsung depend on the key, and are not known to the attacker. We analyze Pilsung and design a cache-based attack. We improve the state of the art by developing techniques for reversing secret-dependent transformations. Our attack, which requires an average of eight minutes on a typical laptop computer, demonstrates that secret transformations do not necessarily protect ciphers against side channel attacks.

Protocols for secure multiparty computation enable a set of parties to compute a function of their inputs without revealing anything but the output. The security properties of the protocol must be preserved in the presence of adversarial behavior. The two classic adversary models considered are semi-honest (where the adversary follows the protocol specification but tries to learn more than allowed by examining the protocol transcript) and malicious (where the adversary may follow any arbitrary attack strategy). Protocols for semi-honest adversaries are often far more efficient, but in many cases the security guarantees are not strong enough.In this paper, we present new protocols for securely computing any functionality represented by an arithmetic circuit. We utilize a new method for verifying that the adversary does not cheat, that yields a cost of just twice that of semi-honest protocols in some settings. Our protocols are information-theoretically secure in the presence of a malicious adversaries, assuming an honest majority. We present protocol variants for small and large fields, and show how to efficiently instantiate them based on replicated secret sharing and Shamir sharing. As with previous works in this area aiming to achieve high efficiency, our protocol is secure with abort and does not achieve fairness, meaning that the adversary may receive output while the honest parties do not.We implemented our protocol and ran experiments for different numbers of parties, different network configurations and different circuit depths. Our protocol significantly outperforms the previous best for this setting (Lindell and Nof, CCS 2017); for a large number of parties, our implementation runs almost an order of magnitude faster than theirs.

Intel Software Guard Extensions (SGX) allows users to perform secure computation on platforms that run untrusted software. To validate that the computation is correctly initialized and that it executes on trusted hardware, SGX supports attestation providers that can vouch for the user’s computation. Communication with these attestation providers is based on the Extended Privacy ID (EPID) protocol, which not only validates the computation but is also designed to maintain the user’s privacy. In particular, EPID is designed to ensure that the attestation provider is unable to identify the host on which the computation executes. In this work we investigate the security of the Intel implementation of the EPID protocol. We identify an implementation weakness that leaks information via a cache side channel. We show that a malicious attestation provider can use the leaked information to break the unlinkability guarantees of EPID. We analyze the leaked information using a lattice-based approach for solving the hidden number problem, which we adapt to the zero-knowledge proof in the EPID scheme, extending prior attacks on signature schemes.

It is well known that constant-time implementations of modular exponentiation cannot use sliding windows. However, software libraries such as Libgcrypt, used by GnuPG, continue to use sliding windows. It is widely believed that, even if the complete pattern of squarings and multiplications is observed through a side-channel attack, the number of exponent bits leaked is not sufficient to carry out a full key-recovery attack against RSA. Specifically, 4-bit sliding windows leak only 40% of the bits, and 5-bit sliding windows leak only 33% of the bits.In this paper we demonstrate a complete break of RSA-1024 as implemented in Libgcrypt. Our attack makes essential use of the fact that Libgcrypt uses the left-to-right method for computing the sliding-window expansion. We show for the first time that the direction of the encoding matters: the pattern of squarings and multiplications in left-to-right sliding windows leaks significantly more information about the exponent than right-to-left. We show how to extend the Heninger-Shacham algorithm for partial key reconstruction to make use of this information and obtain a very efficient full key recovery for RSA-1024. For RSA-2048 our attack is efficient for 13% of keys.