Open-source software projects need to improve vulnerability handling practices, researchers say
- 30 October, 2013 17:18
Many open-source software developers need to improve the way in which they handle vulnerability reports, according to researchers from security firm Rapid7, who recently found and reported vulnerabilities in seven popular open-source software applications.
There's a line of thought among some users that open-source software is more secure than commercial software because there are more people looking at the source code and providing feedback or because open-source projects can patch issues faster.
Rapid7 worked with Brandon Perry, an application security engineer and regular contributor to the Metasploit penetration testing framework, to test that theory, said Christian Kirsch, product marketing manager at Rapid7, in an interview Wednesday at the RSA Europe security conference in Amsterdam.
At the beginning of August, Perry selected seven of the most popular open-source Web applications hosted on SourceForge.net and started looking for vulnerabilities in them. Within two weeks he found security flaws in all of them, Kirsch said.
The researcher found an issue that could allow remote authenticated attackers to execute commands on the underlying operating system in six applications: Moodle, a Web-based learning/course management system that has been downloaded over 4.7 million times from SourceForge; vTiger, a Web-based customer relationship management system with over 3.6 million downloads; Zabbiz, a software product for monitoring network and application performance in enterprises with almost 3 million downloads; ISPConfig, a Web hosting control panel for Linux servers with 1.5 million downloads; OpenMediaVault, an OS distribution based on Debian Linux for network-attached storage servers with over 700,000 downloads; and NAS4Free, a network-attached storage server OS based on FreeBSD with over 600,000 downloads.
Perry also found an XXE (XML eXternal Entity) vulnerability in Openbravo ERP, an open-source enterprise resource planning (ERP) product with 2.1 million downloads on SourceForge, that could allow an attacker to read arbitrary files from the file system with the permissions of the user running the application.
The researcher and Rapid7 then alerted the developers and worked with the Computer Emergency Response Team Coordination Center (CERT/CC) to coordinate the disclosures. They also developed Metasploit exploit modules for the vulnerabilities and released them on Wednesday.
Only two of the seven software projects, Openbravo ERP and vTiger CRM, patched the issues reported to them. Four projects said that they won't fix the authenticated remote command execution issue because they believe it's by design and one project did not communicate their plans, Kirsch said.
The post-authentication command execution issue is not a vulnerability per se, but it is an exposure with security implications, Kirsch said. The developers assumed that the persons who install their applications also administer the entire servers, which is not always true, especially in shared hosting environments or in organizations where separate teams oversee the infrastructure and applications, he said.
This issue can also be used to bypass strong authentication requirements configured on some operating systems to prevent people from easily gaining root access. If the application running on such a system only requires a username and password for authentication and then allows authenticated command execution on the OS, then the stricter controls are bypassed, Kirsch said.
In the process of disclosing the identified security issues to the relevant software projects, Rapid7 found that many of them did not follow common industry practices when it came to handling vulnerability reports and working with security researchers.
"Across these seven projects, I found there were at least seven different approaches to handling incoming vulnerability reports," Tod Beardsley, the engineering manager for Metasploit, said Wednesday in a blog post.
"I won't mention which project representative asked for a password-protected zip file of the disclosure, while another filed the issue on a public bug tracker which promptly e-mailed it back in cleartext, but the level of preparedness I ran into was pretty troubling," he said.
While some of the large open-source software developers like the Apache Software Foundation or Mozilla have good processes in place for dealing with vulnerability reports, there's a lot of room for education for smaller open-source projects that don't deal with this on a daily basis, Kirsch said.
Rapid7 released a list of recommendations that includes creating a dedicated security email address -- usually security@domain -- to receive vulnerability reports, creating a PGP key and publishing it somewhere where it's easy to find, encrypting sensitive communications that deals with vulnerabilities, acknowledging the receipt of vulnerability reports, informing application users about the patched vulnerabilities, and more.