Chink in encryption armour discovered

An underlying flaw in the widely used encryption protocol Open Secure Shell (OpenSSH) has been made public by researchers from the Royal Holloway, University of London.

The flaw, which lies in version 4.7 of OpenSSH on Debian/GNU Linux, allows 32 bits of encrypted text to be rendered in plaintext, according to a research team from the Royal Holloway Information Security Group (ISG).

An attacker has a 2^{-18} (that is, one in 262,144) chance of success. ISG lead professor Kenny Paterson told ZDNet UK last Monday that the flaw was more significant than previous vulnerabilities in OpenSSH.

"This is a design flaw in OpenSSH," said Paterson. "The other vulnerabilities have been more about coding errors."

According to Paterson, a man-in-the-middle attacker could sit on a network and grab blocks of encrypted text as they are sent from client to server. By re-transmitting the blocks to the server, an attacker can work out the first four bytes of corresponding plaintext. The attacker can do this by counting how many bytes the attacker sends until the server generates an error message and tears down the connection, then working backwards to deduce what was in the OpenSSH encryption field before encryption.

The attack relies on flaws in the RFC (Request for Comments) internet standards that define SSH, said Paterson.

Paterson gave a talk on Monday at the IEEE Symposium on Security and Privacy in California to explain his group's research findings. The three ISG academics involved in the research were Paterson, Martin Albrecht and Gaven Watson.

This vulnerability was first made public in November 2008 by the UK Centre for the Protection of National Infrastructure (CPNI), although full details of the flaw were not then given. According to the CPNI advisory, the OpenSSH flaw could be mitigated by IT professionals using AES in counter mode (CTR) to encrypt, instead of cipher-block chaining mode (CBC).

Paterson said his group had worked with OpenSSH developers to mitigate the flaw, and that OpenSSH version 5.2 contained countermeasures.

"They've fixed [OpenSSH]; they've put countermeasures in place to stop our attack," said Paterson. "But the standard has not changed."

Paterson said that he did not believe this flaw had been exploited in the wild, and that to deduce a message of appreciable length could take days. In addition, proprietary SSH vendors had been informed of the issue in advance, and had put countermeasures in their code. However, Paterson added that it always takes time for sysadmins to apply patches to servers and clients, no matter whether the software is open source or proprietary.