Brits pound OpenSSL bugs

Brits pound OpenSSL bugs
Kevin Poulsen, SecurityFocus 2003-09-30

Research by the U.K. government into a once-overlooked class of software vulnerability has surfaced three new security holes in the ubiquitous OpenSSL software package, according to advisories released Tuesday.

All versions of OpenSSL up to and including 0.9.6j and 0.9.7b, and all versions of SSLeay, are open to the attacks: two of the holes can crash the software; a third could lead to an attacker gaining control over vulnerable machines, although the latter scenario remains undemonstrated, according to an advisory from the OpenSSL Project, the collaborative effort that maintains the open-source package.

The vulnerabilities are in the way OpenSSL processes a data format called Abstract Syntax Notation One (ASN.1) -- an internationally recognized standard for coding and transmitting complex data structures, and a building block of the digital certificates that make SSL work.

ASN.1 security commanded high-level U.S. government interest last year after researchers at the Oulu University Secure Programming Group in Finland exploited the format in a novel attack technique that proved effective against dozens of implementations of the Simple Network Management Protocol (SNMP) -- the Internet's standard language for monitoring and controlling routers, switches and other devices. The research forced nearly two hundred companies to evaluate, and in some cases patch, products that used SNMP.

The Oulu research involved systematically flinging a wide variety of intentionally-malformed data at the servers, deliberately violating the rules of ASN.1 in a number of ways that programmers hadn't anticipated: lying about the amount of data being transmitted in a particular field, for example. Where coders didn't plan for illegally-formatted messages, the vulnerable system would crash, or in some cases allow an attacker to overflow an internal buffer and execute malicious instructions on the target machine.

"It's sort of the Monte Carlo approach, because that's the easiest way to deal with ASN.1," says Bruce Schneier, CTO at Counterpane Internet Security Inc. "You really can't look at it systematically because it's so opaque."

U.S. government and industry officials became gravely concerned that the same attack method might be equally effective against other networks and protocols relying on ASN.1, a long list of "critical infrastructures" that includes telephone switching networks, parcel delivery tracking systems, credit card verification networks, and electric utility SCADA systems.

A government working group was formed to tackle the problem, and then-cybersecurity czar Richard Clarke, and his deputy Howard Schmidt, personally briefed President Bush on the issue, Clarke said at the time.

But Tuesday it was a parallel effort by the U.K. government's National Infrastructure Security Coordination Center (NISCC) that bore fruit. Building on the Oulu University's work, the center's been performing and funding its own research into ASN.1 vulnerabilities that could affect critical infrastructures. In the process, NISCC researchers developed a customized test suite that they provided to the OpenSSL Project -- primarily based in the U.K. and Germany -- which used it to find the three holes, according to advisories from both groups.

The only thing surprising is that there haven't been more ASN.1 implementation holes made public in other programs, says Schneier. "I've always thought there should be more, but my guess is that they're just hard to find."

The OpenSSL holes are triggered by delivering malformed ASN.1 data to the vulnerable server through a client certificate-- but another bug in the package makes it effective even against SSL/TLS servers that haven't enabled client authentication. New versions of the software are available that close the holes.