Project Zero Goes To War | TechSNAP 177

chris

8 years ago

Pre-crime is here, with technology that lets you predicting a hack before it happens. We’ll tell you how. Google’s project zero goes to war, we get real about virtualization.

And then its a great batch of your questions, our answers & much more!

Thanks to:

Direct Download:

RSS Feeds:

Become a supporter on Patreon:

— Show Notes: —

Predicting which sites will get hacked, before it happens

Researchers from Carnegie Mellon University have developed a tool that can help predict if a website is likely to become compromised or malicious in the future
Using the Archive.org “Wayback Machine” they looked at websites before they were hacked, and tried to identify trends and other information that may be predictors
“The classifier correctly predicted 66 percent of future hacks in a one-year period with a false positive rate of 17 percent”
“The classifier is focused on Web server malware or, put more simply, the hacking and hijacking of a website that is then used to attack all its visitors”
The tool looks at the server software, outdated versions of Apache and PHP can be good indicators of future vulnerabilities
It also looks at how the website is laid out, how often it is updated, what applications it runs (outdated wordpress is a good hacking target)
It also compares the sites to sites that have been compromised. If a site is very like another, and that other was compromised, there is an increased probability that the first site will also be compromised
The classifier looks at many other factors as well: “For instance, if a certain website suddenly sees a change in popularity, it could mean that it became used as part of a [malicious] redirection campaign,”
The most common marker for a hackable website: The presence of the ‘generator’ meta tag with a value of ‘Wordpress 3.2.1’ or ‘Wordpress 3.3.1’
Research PDF from USENIX
There are tools like those from Norse, that analyze network traffic and attempt to detect new 0-day exploits before they are known

Google’s Project Zero exploits the unexploitable bug

Well over a month ago Google’s Project Zero reported a bug in glibc, however there was much skepticism about the exploitability of the bug, so it was not fixed
However, this week the Google researchers were able to create a working exploit for the bug, including an ASLR bypass for 32bit OSs
The blog post details the process the Project Zero team went through to develop the exploit and gain root privileges
The blog post also details an interesting (accidental) mitigation found in Ubuntu, they caused the researchers to target Fedora to more easily develop the exploit
The blog also discusses a workaround for other issues they ran into. Once they had exploited the set-uid binary, they found that running: system(“/bin/bash”) started the shell with their original privileges, rather than as root. Instead, they called chroot() on a directory they had setup to contain their own /bin/sh that calls setuid(0) and then executes a real shell as the system root user.
The path they used to get a root shell relies on a memory leak in the setuid binary pkexec, which they recommend be fixed as well as the original glibc bug
“The ability to lower ASLR strength by running setuid binaries with carefully chosen ulimits is unwanted behavior. Ideally, setuid programs would not be subject to attacker-chosen ulimit values”
“The exploit would have been complicated significantly if the malloc main linked listed hardening was also applied to the secondary linked list for large chunks”
The glibc bug has since been fixed

Secret Service warns over 1000 businesses hit by Backoff Point-of-Sales terminal malware

The Secret Service and DHS have released an advisory warning businesses about the POS (Point-of-Sales terminal) malware that has been going around for a while
Advisory
“The Department of Homeland Security (DHS) encourages organizations, regardless of size, to proactively check for possible Point of Sale (PoS) malware infections. One particular family of malware, which was detected in October 2013 and was not recognized by antivirus software solutions until August 2014, has likely infected many victims who are unaware that they have been compromised”
“Seven PoS system providers/vendors have confirmed that they have had multiple clients affected“
“Backoff has experts concerned because it’s effective in swiping customer credit card data from businesses using a variety of exfiltration tools, including memory, or RAM scraping, techniques, keyloggers and injections into running processes”
“A report from US-CERT said attackers use Backoff to steal payment card information once they’ve breached a remote desktop or administration application, especially ones that are using weak or default credentials”
“Backoff is then installed on a point-of-sale device and injects code into the explorer.exe process that scrapes memory from running processes in order to steal credit card numbers before they’re encrypted on the device and sent to a payment processor. “
“Keylogging functionality is also present in most recent variants of ‘Backoff’. Additionally, the malware has a C2 component that is responsible for uploading discovered data, updating the malware, downloading/executing further malware, and uninstalling the malware,”
US-CERT Advisory
Krebs reports that Dairy Queen may also be a victim of this attack
“Dairy Queen says it has no indication of a card breach at any of its thousands of locations, but the company also acknowledges that nearly all stores are franchises and that there is no established company process or requirement that franchisees communicate security issues or card breaches to Dairy Queen headquarters”