AWS re:Invent 2019: Speculation & leakage: Timing side channels & multi-tenant computing
> In January 2018, the world learned about Spectre and Meltdown, a new class of issues that affects virtually all modern CPUs via nearly imperceptible changes to their micro-architectural states and can result in full access to physical RAM or leaking of state between threads, processes, or guests. In this session, we examine one of these side-channel attacks in detail and explore the implications for multi-tenant computing. We discuss AWS design decisions and what AWS does to protect your instances, containers, and function invocations. Finally, we discuss what the future looks like in the presence of this new class of issue.
This is a good recap. Specific defenses starts around 42:00.
TRRespass: Exploiting the Many Sides of Target Row Refresh
> Well, after two years of rigorous research, looking inside what is implemented inside CPUs and DDR4 chips using novel reverse engineering techniques, we can tell you that we do not live in a Rowhammer-free world. And we will not for the better part of this decade. Turns out while the old hammering techniques no longer work, once we understand the exact nature of these mitigations inside modern DDR4 chips, using new hammering patterns it is trivial to again trigger plenty of new bit flips. Yet again, these results show the perils of lack of transparency and security-by-obscurity. This is especially problematic since unlike software vulnerabilities, we cannot fix these hardware bit flips post-production.
LVI - Hijacking Transient Execution with Load Value Injection
> LVI is a new class of transient-execution attacks exploiting microarchitectural flaws in modern processors to inject attacker data into a victim program and steal sensitive data and keys from Intel SGX, a secure vault in Intel processors for your personal data.
> LVI turns previous data extraction attacks around, like Meltdown, Foreshadow, ZombieLoad, RIDL and Fallout, and defeats all existing mitigations. Instead of directly leaking data from the victim to the attacker, we proceed in the opposite direction: we smuggle — “inject” — the attacker’s data through hidden processor buffers into a victim program and hijack transient execution to acquire sensitive information, such as the victim’s fingerprints or passwords.
Take A Way: Exploring the Security Implications of AMD’s Cache Way Predictors
> In this paper, we are the first to exploit the cache way predictor. We reverse-engineered AMD’s L1D cache way predictor in microarchitectures from 2011 to 2019, resulting in two new attack techniques. With Collide+Probe, an attacker can monitor a victim’s memory accesses without knowledge of physical addresses or shared memory when time-sharing a logical core. With Load+ Reload, we exploit the way predictor to obtain highly-accurate memory-access traces of victims on the same physical core. While Load+Reload relies on shared memory, it does not invalidate the cache line, allowing stealthier attacks that do not induce any last-level-cache evictions.
KASLR: Break It, Fix It, Repeat
Escaping the Chrome Sandbox with RIDL
> Vulnerabilities that leak cross process memory can be exploited to escape the Chrome sandbox. An attacker is still required to compromise the renderer prior to mounting this attack. To protect against attacks on affected CPUs make sure your microcode is up to date and disable hyper-threading (HT).
This is a pretty clear write-up and comes with a nice footnote:
> When I started working on this I was surprised that it’s still exploitable even though the vulnerabilities have been public for a while. If you read guidance on the topic, they will usually talk about how these vulnerabilities have been mitigated if your OS is up to date with a note that you should disable hyper threading to protect yourself fully. The focus on mitigations certainly gave me a false sense that the vulnerabilities have been addressed and I think these articles could be more clear on the impact of leaving hyper threading enabled.
Speculative Load Hardening
> While several approaches are being actively pursued to mitigate specific branches and/or loads inside especially risky software (most notably various OS kernels), these approaches require manual and/or static analysis aided auditing of code and explicit source changes to apply the mitigation. They are unlikely to scale well to large applications. We are proposing a comprehensive mitigation approach that would apply automatically across an entire program rather than through manual changes to the code. While this is likely to have a high performance cost, some applications may be in a good position to take this performance / security tradeoff.
> SafeSide is a project to understand and mitigate software-observable side-channels: information leaks between software domains caused by implementation details outside the software abstraction.
CPU Introspection: Intel Load Port Snooping
> We’re going to go into a unique technique for observing and sequencing all load port traffic on Intel processors. By using a CPU vulnerability from the MDS set of vulnerabilities, specifically multi-architectural load port data sampling (MLPDS, CVE-2018-12127), we are able to observe values which fly by on the load ports. Since (to my knowledge) all loads must end up going through load ports, regardless of requestor, origin, or caching, this means in theory, all contents of loads ever performed can be observed. By using a creative scanning technique we’re able to not only view “random” loads as they go by, but sequence loads to determine the ordering and timing of them.
> We’ll go through some examples demonstrating that this technique can be used to view all loads as they are performed on a cycle-by-cycle basis. We’ll look into an interesting case of the micro-architecture updating accessed and dirty bits using a microcode assist. These are invisible loads dispatched on the CPU on behalf of the user when a page is accessed for the first time.
OpenSSH Key Shielding
> On June 21, 2019, support for SSH key shielding was introduced into the OpenBSD tree, from which the OpenSSH releases are derived. SSH key shielding is a measure intended to protect private keys in RAM against attacks that abuse bugs in speculative execution that current CPUs exhibit. This functionality has been part of OpenSSH since the 8.1 release. SSH private keys are now being held in memory in a shielded form; keys are only unshielded when they are used and re‐shielded as soon as they are no longer in active use. When a key is shielded, it is encrypted in memory with AES‐256‐CTR; this is how it works:
> Modern processors are being pushed to perform faster than ever before - and with this comes increases in heat and power consumption. To manage this, many chip manufacturers allow frequency and voltage to be adjusted as and when needed. But more than that, they offer the user the opportunity to modify the frequency and voltage through priviledged software interfaces. With Plundervolt we showed that these software interfaces can be exploited to undermine the system’s security. We were able to corrupt the integrity of Intel SGX on Intel Core processors by controling the voltage when executing enclave computations. This means that even Intel SGX’s memory encryption/authentication technology cannot protect against Plundervolt.
Not sure anyone should care about SGX anymore, all things considered, but for completeness, here’s another one.
TAA and other RIDL issues
> On Nov 12, 2019, we (VUSec) disclose TSX Asynchronous Abort (TAA), a new speculation-based vulnerability in Intel CPUs as well as other MDS-related issues, as described in our new RIDL addendum. In reality, this is no new vulnerability. We disclosed TAA (and other issues) as part of our original RIDL submission to Intel in Sep 2018. Unfortunately, the Intel PSIRT team missed our submitted proof-of-concept exploits (PoCs), and as a result, the original MDS mitigations released in May 2019 only partially addressed RIDL. You can read the full story below.
> On July 3, 2019, we finally learned that, to our surprise, the Intel PSIRT team had missed the PoCs from our Sep 29 submission, despite having awarded a bounty for it, explaining why Intel had failed to address - or even publicly acknowledge - many RIDL-class vulnerabilities on May 14, 2019.
When you have so many problems you’re paying out bounties without knowing what for...
TPM—Fail TPM meets Timing and Lattice Attacks
> We discovered timing leakage on Intel firmware-based TPM (fTPM) as well as in STMicroelectronics’ TPM chip. Both exhibit secret-dependent execution times during cryptographic signature generation. While the key should remain safely inside the TPM hardware, we show how this information allows an attacker to recover 256-bit private keys from digital signature schemes based on elliptic curves.
> This research shows that even rigorous testing as required by Common Criteria certification is not flawless and may miss attacks that have explicitly been checked for. The STMicroelectronics TPM chip is Common Criteria certified at EAL4+ for the TPM protection profiles and FIPS 140-2 certified at level 2, while the Intel TPM is certified according to FIPS 140-2. However, the certification has failed to protect the product against an attack that is considered by the protection profile.
> Light Commands is a vulnerability of MEMS microphones that allows attackers to remotely inject inaudible and invisible commands into voice assistants, such as Google assistant, Amazon Alexa, Facebook Portal, and Apple Siri using light.
> In our paper we demonstrate this effect, successfully using light to inject malicious commands into several voice controlled devices such as smart speakers, tablets, and phones across large distances and through glass windows.
Minerva: Lattice attacks strike again
> This page describes our discovery of a group of side-channel vulnerabilities in implementations of ECDSA/EdDSA in programmable smart cards and cryptographic software libraries. Our attack allows for practical recovery of the long-term private key. We have found implementations which leak the bit-length of the scalar during scalar multiplication on an elliptic curve. This leakage might seem minuscule as the bit-length presents a very small amount of information present in the scalar. However, in the case of ECDSA/EdDSA signature generation, the leaked bit-length of the random nonce is enough for full recovery of the private key used after observing a few hundreds to a few thousands of signatures on known messages, due to the application of lattice techniques.
50 ways to leak your data: an exploration of apps’ circumvention of the Android permissions system
> This paper is a study of Android apps in the wild that leak permission protected data (identifiers which can be used for tracking, and location information), where those apps should not have been able to see such data due to a lack of granted permissions. By detecting such leakage and analysing the responsible apps, the authors uncover a number of covert and side channels in real-world use.
Dragonblood new results
> August 2019 — During our initial disclosure, the Wi-Fi Alliance privately created security recommendations to mitigate our attacks. In these recommendations, they claim that Brainpool curves are safe to use, at least if products securely implement Dragonfly’s quadratic residue test (i.e. it must be implemented without side-channel leaks). However, we found that using Brainpool curves introduces a second class of side-channel leaks in the Dragonfly handshake of WPA3. In other words, even if the advice of the Wi-Fi Alliance is followed, implementations remain at risk of attacks. This demonstrates that implementing Dragonfly and WPA3 without side-channel leaks is surprisingly hard. It also, once again, shows that privately creating security recommendations and standards is at best irresponsible and at worst inept.
That last line.
> RSA is an intrinsically fragile cryptosystem containing countless foot-guns which the average software engineer cannot be expected to avoid. Weak parameters can be difficult, if not impossible, to check, and its poor performance compels developers to take risky shortcuts. Even worse, padding oracle attacks remain rampant 20 years after they were discovered. While it may be theoretically possible to implement RSA correctly, decades of devastating attacks have proven that such a feat may be unachievable in practice.
Defending against transient execution attacks
> It is important to build up a systematic understanding of these attacks and possible defenses
RAMBleed Reading Bits in Memory Without Accessing Them
> RAMBleed is based on a previous side channel called Rowhammer, which enables an attacker to flip bits in the memory space of other processes. We show in our paper that an attacker, by observing Rowhammer-induced bit flips in her own memory, can deduce the values in nearby DRAM rows. Thus, RAMBleed shifts Rowhammer from being a threat not only to integrity, but confidentiality as well. Furthermore, unlike Rowhammer, RAMBleed does not require persistent bit flips, and is thus effective against ECC memory commonly used by server computers.