Category Archives: Security engineering

Bad security, good security, case studies, lessons learned

Heartbleed and RSA private keys

UPDATE:

Matthew Arcus pointed out that a freed buffer will always have the first 16 bytes (x86_64) corrupted by two free list pointers, according to glibc’s malloc implementation. This means the leak mechanism I described below will only explain the partial leak I exploited, but not complete prime number leaks others have seen. In fact, the uncorrupted numbers come from the long-lived copy of p and q in the Montgomery context.

——–

By now everyone knows about the OpenSSL Heartbleed vulnerability: a missing bounds check in one of the most popular TLS implementations has made millions of web servers (and others) leak all sorts of sensitive information from memory. This could leak login credentials, authentication cookies and web traffic to attackers. But could it be used to recover the site’s TLS private key? This would enable complete decryption of previously-recorded traffic if perfect forward secrecy was not negotiated at the time and otherwise Man-in-The-Middle attacks to all future TLS sessions.

Since this would be a much more serious consequence of Heartbleed, I decided to investigate. The results were positive: I was able to extract private keys from a test Nginx server after a few day’s work. Later I applied my techniques to solve the CloudFlare Challenge. Along with a few other security researchers, we independently demonstrated that RSA private keys are indeed at risk. In this blog post, I’ll provide some detail on how to extract the private key and why the attack is possible.

How to extract the private key

Readers not familiar with RSA can read about it here. To simplify things a bit, a large (2048 bits) number N is constructed by multiplying together two large randomly generated prime numbers p and q. N is made public while p and q are kept secret. Finding p or q allows recovery of the private key. A generic attack is just factorising N, but this is believed to be difficult. However, with a vulnerability like Heartbleed, the attack is much simpler: because the web server needs the private key in memory to sign the TLS handshake, p and q must live in memory and we can try to obtain them with Heartbleed packets. Then the problem simply becomes how to identify them in the returned data. This is easy, as we know p and q are 1024 bits (128 bytes) long, and OpenSSL represents big numbers little-endian in memory, so a brute-force approach of treating every 128 consecutive bytes in the heartbleed packets as little-endian numbers and testing if it divides N is sufficient to spot potential leaks. This is how most people solved the CloudFlare challenge.

Coppersmith improvement

But wait, isn’t our scenario just like a cold boot attack? And there has been a lot of research on recovering RSA private keys with partial information. One of the most famous papers from Coppersmith presents message recovery attacks on related messages or insufficient padding, as well as factorisation with partial knowledge with the help of lattice basis reduction. With the Coppersmith attack, N can be efficiently factorised if either the top or bottom half bits of p is known. Put it in context, we only need the top/bottom 64 bytes of p in order to compute the private keys, compared to the naive brute-force which requires all 128 bytes. In practice, reaching Coppersmith’s limit is computational expensive (although still much better than factorisation), but assuming 77 bytes (60%) known we can comb the heartbleed packets for potential private key segments very quickly.

In retrospect, there were 242 instances of private key remnants suitable for the Coppersmith attack, out of 10,000 packets (64 KB each) that I had collected. Implementation of the Coppersmith attack was made easy thanks to the comprehensive computer algebra building blocks from Sage (although I later found out that Sage already has Coppersmith implemented).

Can we do even better? If you have ever viewed a RSA private key using openssl rsa -text -in server.key, you will notice that there are quite a few numbers other than the two prime factors p and q. In fact, they are precomputed values for RSA’s Chinese Remainder Theorem optimisation. If some of them are leaked they can also be used to deduce p. And what about the Montgomery representations of p and q that OpenSSL use for fast multiplication? They also admit a variant of Coppersmith so partial bits are useful as well. With this in mind I set out to search for them in the collected packets from my test server where these values are all known. But not even single one of them appears partially (> 16 bytes) in the dataset. How is this possible?

Note: all my experiments and the CloudFlare challenge are targeting Nginx which is single-threaded. It is possible that for a multi-threaded web server, more leakage can be observed.

Why p and only p leaks

When heartbleed first came out, people argued that RSA private keys should not be leaked because they are loaded when the web server starts so they occupy a lower memory address, and because the heap grows upwards, a later-allocated buffer leaked by Heartbleed shouldn’t reach them. This argument is consistent with my inability to find any CRT precomputed values, but with one exception: p is definitely leaked somehow. If we assume this argument is correct, the question becomes: why is p leaked?

To add more to the mystery, OpenSSL apparently scrubs all temporary BigNums it used! To reduce the overhead of dynamically allocating temporary values OpenSSL provides a BN_CTX which is a pool of BigNums operating in stack (LIFO) fashion. Upon finishing the context is destroyed and all allocated buffers are scrubbed. This would mean that when the heartbleed packet is constructed there shouldn’t be any temporary values left in memory (again assuming single thread) because the BN_CTX has long been released.

I won’t bother you with all the pain I went through to identify the cause, so here is the spoiler: when a BigNum is being extended to a bigger buffer size, its original buffer is not zeroised before being freed. The chain of the control flow path that leads to the leak of p is even more subtle. During intial TLS handshake the server key exchange is signed with the private key. The CRT signing performs a modulo p operation, which caused p<<BN_BITS2 to be stored in a temporary variable allocated from the BN_CTX pool. Later in the CRT fault-injection check, that temporary variable is reused (remember BN_CTX operates like a stack) as val[0]. An interesting fact is that a reallocated temporary variable only gets its lowest nibble zeroised, and in the case of p<<BN_BITS2 nothing is destroyed. val[0] immediately receives a Montgomery-reduced value, but since the original buffer is unable to accommodate the new value, it gets extended and p gets released to free heap space, waiting to be grabbed. And because this happens every time a TLS handshake occurs, it can be spilled everywhere.

Since it is difficult to find which BigNums may be extended and leaked statically, I instrumented OpenSSL and experimented a bit. It turned out that a shifted version of the Montgomery representation of p will also be freed at the leaky point, but that only happened at the first RSA exponentiation when the Montgomery context is initialised, so it will live in a low memory address and I was unable to locate it in any collected packets.

The OpenSSL team has been notified about the above leak. Although to be fair this is not exactly a security bug by itself as OpenSSL is never explicitly designed to prevent sensitive data leakage to heap.

I’d like to thank Joseph Bonneau for the useful comments and proof reading.

Current state of anonymous email usability

As part of another project, I needed to demonstrate how the various user-interface options for sending anonymous email through Mixmaster appeared to the email sender. This is very difficult to explain in words, so I recorded some screencasts. The tools I used were the Mixmaster command line tool, the Mutt email client with Mixmaster plugin, QuickSilver Lite, and finally a web-based interface.

The project is now over, but in case these screencasts are of wider interest, I’ve put them on YouTube.

Overall, the usability of Mixmaster is not great. All of the secure options are difficult to configure and use (QuickSilver Lite is probably the best), emails take a long time to be sent, recipients of anonymous email can’t send replies, and there is a high chance that the email will be dropped en-route.

Continue reading Current state of anonymous email usability

Financial cryptography 2014

I will be trying to liveblog Financial Cryptography 2014. I just gave a keynote talk entitled “EMV – Why Payment Systems Fail” summarising our last decade’s research on what goes wrong with Chip and PIN. There will be a paper on this out in a few months; meanwhile here’s the slides and here’s our page of papers on bank security.

The sessions of refereed papers will be blogged in comments to this post.

Research Assistants and Associates in OS, Compiler and CPU Security

We are pleased to announce a job ad for two new research assistants or post-doctoral research associates working on our CTSRD Project, whose target research areas include OS, compiler, and CPU security. This is a joint project between the University of Cambridge’s Security, NetOS, and Computer Architecture research groups, as well as the Computer Science Laboratory at SRI International.

Research Assistants and Associates in OS, Compiler and CPU Security
Fixed-term: The funds for this post are available for 18 months in the first instance.

We are seeking multiple Research Assistants and Post-Doctoral Research Associates to join the CTSRD Project, which is investigating fundamental improvements to CPU-architecture, operating-system (OS), program-analysis, and programming-language structure in support of computer security. The CTSRD Project is a collaboration between the University of Cambridge and SRI International, and part of the DARPA CRASH research programme on clean-slate computer system design for security. More information may be found at:

This position will be an integral part of an international team of researchers spanning multiple institutions in academia and industry. Successful candidates will contribute to the larger research effort by performing system-software, compiler, and hardware implementation and experimentation, developing and evaluating novel hypotheses about refinements to the vertical hardware-software stack. Possible areas of responsibility include: modifying OS kernels (e.g., FreeBSD), adapting compiler suites (e.g., Clang/LLVM); extending an open-source Bluespec-based research-processor design (CHERI); supporting an early-adopter user community for open-source hardware and software; and improving the quality and performance of hardware-software prototypes. The successful candidate must be willing to travel in the UK and abroad engaging with downstream user communities.
Continue reading Research Assistants and Associates in OS, Compiler and CPU Security

Why dispute resolution is hard

Today we release a paper on security protocols and evidence which analyses why dispute resolution mechanisms in electronic systems often don’t work very well. On this blog we’ve noted many many problems with EMV (Chip and PIN), as well as other systems from curfew tags to digital tachographs. Time and again we find that electronic systems are truly awful for courts to deal with. Why?

The main reason, we observed, is that their dispute resolution aspects were never properly designed, built and tested. The firms that delivered the main production systems assumed, or hoped, that because some audit data were available, lawyers would be able to use them somehow.

As you’d expect, all sorts of things go wrong. We derive some principles, and show how these are also violated by new systems ranging from phone banking through overlay payments to Bitcoin. We also propose some enhancements to the EMV protocol which would make it easier to resolve disputes over Chip and PIN transactions.

Update (2013-03-07): This post was mentioned on Bruce Schneier’s blog, and this is some good discussion there.

Update (2014-03-03): The slides for the presentation at Financial Cryptography are now online.

Why bouncing droplets are a pretty good model of quantum mechanics

Today Robert Brady and I publish a paper that solves an outstanding problem in physics. We explain the beautiful bouncing droplet experiments of Yves Couder, Emmanuel Fort and their colleagues.

For years now, people interested in the foundations of physics have been intrigued by the fact that droplets bouncing on a vibrating tray of fluid can behave in many ways like quantum mechanical particles, with single-slit and double-slit diffraction, tunneling, Anderson localisation and quantised orbits.

In our new paper, Robert Brady and I explain why. The wave field surrounding the droplet is, to a good approximation, Lorentz covariant with the constant c being the speed of surface waves. This plus the inverse square force between bouncing droplets (which acts like the Coulomb force) gives rise to an analogue of the magnetic force, which can be observed clearly in the droplet data. There is also an analogue of the Schrödinger equation, and even of the Pauli exclusion principle.

These results not only solve a fascinating puzzle, but might perhaps nudge more people to think about novel models of quantum foundations, about which we’ve written three previous papers.

"Perfectly" Encrypt 50 Letters By Hand

When I read about cryptography before computers, I sometimes wonder why people did this and that instead of something a bit more secure. We may ridicule portable encryption systems based on monoalphabetic or even simple polyalphabetic ciphers but we may also change our opinion after actually trying it for real.
Continue reading "Perfectly" Encrypt 50 Letters By Hand

Anatomy of Passwords

Passwords have not really changed since they were first used. Let’s go down the memory lane a bit and then analyse how password systems work and how they could be improved. You may say – forget passwords, OTP is the way forward. My next question would then be: So why do we use OTP in combination with passwords when they are so good?
Continue reading Anatomy of Passwords

2013 Capsicum year in review

It’s been a busy year for Capsicum, practical capabilities for UNIX, so a year-end update seemed in order:

The FreeBSD Foundation and Google jointly funded a Capsicum Integration Project that took place throughout 2013 — described by Foundation project technical director Ed Maste in a recent blog article. Pawel Jakub Dawidek refined several Capsicum APIs, improving support for ioctls and increasing the number of supported capability rights for FreeBSD 10. He also developed Casper, a helper daemon that provides services (such as DNS, access to random numbers) to sandboxes — and can, itself, sandbox services. Casper is now in the FreeBSD 11.x development branch, enabled by default, and should appear in FreeBSD 10.1. The Google Open Source Program Office (OSPO) blog also carried a September 2013 article on their support for open-source security, featuring Capsicum.

Capsicum is enabled by default in the forthcoming FreeBSD 10.0 release — capability mode, capabilities, and process descriptors are available in the out-of-the-box GENERIC kernel. A number of system services use Capsicum to sandbox themselves — such as the DHCP client, high-availability storage daemon, audit log distribution daemon, but also command-line tools like kdump and tcpdump that handle risky data. Even more will appear in FreeBSD 10.1 next year, now that Casper is available.

David Drysdale at Google announced Capsicum for Linux, an adaptation of Linux to provide Capsicum’s capability mode and capabilities, in November 2013. David and Ben Laurie visited us in Cambridge multiple times this year to discuss the design and implementation, review newer Capsicum APIs, and talk about future directions. They hope to upstream this work to the Linux community. Joris Giovannangeli also announced an adaptation of Capsicum to DragonFlyBSD in October 2013.

Over the summer, Mariusz Zaborski and Daniel Peryolon were funded by Google Summer of Code to work on a variety of new Capsicum features and services, adapting core UNIX components and third-party applications to support sandboxing. For example, Mariusz looked at sandboxing BSD grep: if a vulnerability arises in grep’s regular-expression matching, why should processing a file of malicious origin yield full rights to your UNIX account?

In May 2013, our colleagues at the University of Wisconsin, Madison, led by Bill Harris, published a paper at the IEEE Symposium on Security and Privacy (“Oakland”) on “Declarative, Temporal, and Practical Programming with Capabilities” — how to model program behaviour, and automatically transform some classes of applications to use Capsicum sandboxing. We were very pleased to lend a hand with this work, and feel the art of programming for compartmentalisation is a key research challenge. We also collaborated with folk at SRI and Google on a a workshop paper developing our ideas about application compartmentalisation, which appeared at the Security Protocols Workshop here in Cambridge in March 2013.

Google and the FreeBSD Foundation are committed to further work on Capsicum and its integration with applications, and research continues on how to apply Capsicum at several institutions including here at Cambridge. We hope to kick off a new batch of application adaptation in coming months — as well as integration with features such as DNSSEC. However, we also need your help in adapting applications to use Capsicum on systems that support it!