Jos Wetzels

In this talk we will take a look at the 'Vault 7' Protego documents, which have received very little attention so far, and challenge the assertion that Protego was a 'suspected assassination module for [a] GPS guided missile system ... used on-board Pratt & Whitney aircraft' based on system block diagrams, build instructions and a few interesting news items. In addition, we will discuss hypothetical weaknesses in systems like it.

In March 2017, WikiLeaks published the 'Vault 7' series of documents detailing 'cyber' activities and capabilities of the United States' Central Intelligence Agency (CIA). Among a wide variety of implant & exploit frameworks the final documents released as part of the dump, related to a project code-named 'Protego', stood out as unusual due to describing a piece of missile control technology rather than CNO capabilities. As a result, these documents have received comparatively little attention from the media and security researchers. While initially described by WikiLeaks as a 'suspected assassination module for [a] GPS guided missile system ... used on-board Pratt & Whitney aircraft', a closer look at the documents sheds significant doubt on that assertion. Instead, it seems more likely that Protego was part of an arms control solution used in covert CIA supply programs delivering various kinds of weapons to proxy forces while attempting to counteract undesired proliferation. In this talk we will take a look at the Protego documents and show how we can piece quite a bit of information together from a handful of block diagrams, some build instructions and a few news articles. Finally, we will discuss the potential weaknesses of such 'lockdown' systems which have been proposed for and are deployed in everything from theft prevention solutions and livestock management to firearms control and consumer UAVs.

In this talk we will present a deep-dive analysis of the anatomy of QNX: a proprietary, real-time operating system aimed at the embedded market used in many sensitive and critical systems, particularly within the automotive industry. We will present the first reverse-engineering and analysis of the exploit mitigations, secure random number generators and memory management internals of QNX versions up to and including 6.6 and the brand new 64-bit QNX 7.0 (released in March 2017) and uncover a variety of design issues and vulnerabilities.

QNX is a proprietary, closed-source, Unix-like real-time operating system aimed at the embedded market. It is found in everything from BlackBerry products, carrier-grade routers and medical devices to military radios, UAVs and nuclear powerplants. On top of that, it dominates the automotive market and is found in millions of cars. While some prior security research has discussed QNX, mainly as a byproduct of BlackBerry mobile research, there is no prior work on QNX exploit mitigations or its secure random number generators. This talk seeks to close that gap by presenting the first reverse-engineering and analysis of the exploit mitigations, secure random number generators and memory management internals of QNX. We dissect the NX / DEP, ASLR, Stack Cookies and RELRO mitigations as well as the /dev/random and kernel PRNGs. We subsequently uncover a variety of design issues and vulnerabilities in these mitigations and PRNGs, which have significant implications for the exploitability of memory corruption vulnerabilities on QNX as well as the strength of its cryptographic ecosystem. Finally, we provide information on available patches and hardening measures available to defenders seeking to harden their QNX-based systems against the discussed issues.

Other people: Ali Abbasi

Secure random number generators play a crucial role in the wider security ecosystem. In the absence of a dedicated hardware True Random Number Generator (TRNG), computer systems have to resort to a software (cryptographically secure) Pseudo-Random Number Generator (CSPRNG). Since the (secure) design of a CSPRNG is an involved and complicated effort and since randomness is such a security-critical resource, many operating systems provide a CSPRNG as a core system service and many popular security software products assume their presence. The constraints imposed by the embedded world, however, pose a variety of unique challenges to proper OS (CS)PRNG design and implementation which have historically resulted in security failures. In this talk we will discuss these challenges, how they affect the quality of (CS)PRNGs in embedded operating systems and illustrate our arguments by means of the first public analysis of the OS random number generators of several popular embedded operating systems.

Randomness is a fundamental, security-critical resource in the wider security ecosystem utilized by everything from cryptographic software (eg. key and nonce generation) to exploit mitigations (eg. ASLR and stack canary generation). Ideally secure random number generation is done using a dedicated hardware True Random Number Generator (TRNG) collecting entropy from physical processes such as radioactive decay or shot noise. TRNGs, however, are both relatively slow in their provision of random data and often too expensive to integrate in a system which means computer systems have to resort to a software (cryptographically secure) Pseudo-Random Number Generator (CSPRNG). Such a CSPRNG is seeded (both initially and continuously) from a variety of sources of 'true' entropy which are effectively stretched into additional pseudo-random data using cryptographic methods. Since the design and implementation of such CSPRNGs is a complicated and involved effort, many operating systems provide one as a system service (eg. /dev/(u)random on UNIX-like systems) and as a result many security software suites assume their existence. The embedded world, however, poses a variety of unique challenges (resulting from constraints and deployment scenarios, which differ significantly from the general-purpose world) when designing and implementing (CS)PRNGs. Resulting inadequacies in embedded OS random number generators have led to various security failures in the past (from weak cryptographic keys in network devices to broken exploit mitigations in smartphones) emphasizing the need for public scrutiny of their security, especially considering the nature of embedded system deployments (in everything from vehicles and critical infrastructure to networking equipment) and the sheer variety of ebmedded operating systems compared to the general-purpose world. In this talk we will discuss various challenges posed by the embedded world to (CS)PRNG design and implementation and illustrate our arguments by means of the first public analysis of the OS random number generators of several popular embedded operating systems and a discussion of how their flaws related to these previously identified challenges.

Other people: Ali Abbasi