POODLE attack forces the Internet to move away from SSL 3.0

Posted by   Virus Bulletin on   Oct 15, 2014

Users and administrators urged to stop supporting the protocol, or at least to prevent downgrade attacks.

After Heartbleed and Shellshock, or the SSL/TLS attacks CRIME and BEAST, 'POODLE' does sound rather cute. Yet the vulnerability in version 3.0 of the SSL protocol that was disclosed by Google researchers yesterday is fairly serious and shouldn't be ignored.

Details on POODLE, which stands for 'Padding Oracle On Downgraded Legacy Encryption', were shared in a document on OpenSSL's website (PDF), but it is important to note that the vulnerability is in the protocol rather than in the implementations, and other implementations of SSL 3.0 are just as vulnerable. Microsoft, for instance, published an advisory here.

The vulnerability has to do with the way SSL 3.0 handles padding: adding characters to a plaintext to make its length an integral number of blocks, which is necessary in order to make cipher-block chaining (CBC) work. Technical details on the vulnerability and how it can be attacked - both of which require some basic understanding of cryptography - have been posted by, among others, Adam Langley, Matthew Green and Daniel Fox Franke.

  RFC 6101, which defines SSL 3.0.

The good news is that there is a replacement for SSL 3.0. It is called TLS and has been available since 1999. There are very few clients and servers that do not support it (Internet Explorer 6 is a notable exception).

The bad news is that an attacker who finds himself in a man-in-the-middle position may be able to 'downgrade' the connection, making both ends use the vulnerable protocol, even though they both support TLS 1.0 or later.

The easiest way to fix this vulnerability is simply to disable SSL 3.0 support in all clients and servers you manage. If for some reason that is not possible, Google recommends supporting TLS_FALLBACK_SCSV, which prevents downgrade attacks. Yesterday, OpenSSL released a patch that adds support for TLS_FALLBACK_SCSV.

Of course, SSL/TLS isn't only used for securing HTTP connections. However, as POODLE requires an attacker to force the target to make a very large number of specific connections, the ability for the attacker to use it on other protocols seems all but impossible.

Still, one never knows, and there is always the possibility for the attack to be improved now that it is public. Johannes Ullrich explains how to turn off SSL 3.0 in various commonly used clients and servers.

Backwards compatibility is a good thing and makes the Internet work even for those who, for various reasons, aren't able to use the latest software and protocols. But eventually, we will have to stop supporting outdated protocols. For SSL 3.0, that time has come.

Posted on 15 October 2014 by Martijn Grooten



Latest posts:

Firefox 59 to make it a lot harder to use data URIs in phishing attacks

Firefox developer Mozilla has announced that, as of version 59 of the browser, many kinds of data URIs, which provide a way to create "domainless web content", will not be rendered in the browser, thus making this trick - used in various phishing…

Standalone product test: FireEye Endpoint

Virus Bulletin ran a standalone test on FireEye's Endpoint Security solution.

VB2017 video: Consequences of bad security in health care

Jelena Milosevic, a nurse with a passion for IT security, is uniquely placed to witness poor security practices in the health care sector, and to fully understand the consequences. Today, we publish the recording of a presentation given by Jelena at…

Vulnerabilities play only a tiny role in the security risks that come with mobile phones

Both bad news (all devices were pwnd) and good news (pwning is increasingly difficult) came from the most recent mobile Pwn2Own competition. But the practical security risks that come with using mobile phones have little to do with vulnerabilities.

VB2017 paper: The (testing) world turned upside down

At VB2017 in Madrid, industry veteran and ESET Senior Research Fellow David Harley presented a paper on the state of security software testing. Today we publish David's paper in both HTML and PDF format.