Direkt zum Hauptbereich

Save Your Cloud: DoS on VMs in OpenNebula 4.6.1

This is a post about an old vulnerability that I finally found the time to blog about. It dates back to 2014, but from a technical point of view it is nevertheless interesting: An XML parser that tries to fix structural errors in a document caused a DoS problem.

All previous posts of this series focused on XSS. This time, we present a vulnerability which is connected another Cloud Management Platform: OpenNebula. This Infrastructure-as-a-Service platform started as a research project in 2005. It is used by information technology companies like IBM, Dell and Akamai as well as academic institutions and the European Space Administrations (ESA). By relying on standard Linux tools as far as possible, OpenNebula reaches a high level of customizability and flexibility in hypervisors, storage systems, and network infrastructures. OpenNebula is distributed using the Apache-2 license.


OpenNebula offers a broad variety of interfaces to control a cloud. This post focuses on Sunstone, OpenNebula's web interface (see Figure 1).

Figure 1: OpenNebula's Sunstone Interface displaying a VM's control interface

Before OpenNebula 4.6.2, Sunstone had no Cross-Site Request Forgery (CSRF) protection. This is a severe problem. Consider an attacker who lures a victim into clicking on a malicious link while being logged in at a private cloud. This enables the attacker to send arbitrary requests to the private cloud through the victims browser. However, we could find other bugs in OpenNebula that allowed us to perform much more sophisticated attacks.

Denial-of-Service on OpenNebula-VM

At its backend, OpenNebula manages VMs with XML documents. A sample for such an XML document looks like this:
<VM>
   <ID>0</ID>
   <NAME>My VM</NAME>
   <PERMISSIONS>...</PERMISSIONS>
   <MEMORY>512</MEMORY>
   <CPU>1</CPU>
   ...
</VM>
OpenNebula 4.6.1 contains a bug in the sanitization of input for these XML documents: Whenever a VM's name contains an opening XML tag (but no corresponding closing one), an XML generator at the backend automatically inserts the corresponding closing tag to ensure well-formedness of the resulting document. However, the generator outputs an XML document that does not comply with the XML schema OpenNebula expects. The listing below shows the structure that is created after renaming the VM to 'My <x> VM':
<VM>
   <ID>0</ID>
   <NAME>My <x> VM</x>
      <PERMISSIONS>...</PERMISSIONS>
      <MEMORY>512</MEMORY>
      <CPU>1</CPU>
      ...
   </NAME>
</VM>
The generator closes the <x> tag, but not the <NAME> tag. At the end of the document, the generator closes all opened tags including <NAME>.

OpenNebula saves the incorrectly generated XML document in a database. The next time the OpenNebula core retrieves information about that particular VM from the database the XML parser is mixed up and runs into an error because it only expects a string as name, not an XML tree. As a result, Sunstone cannot be used to control the VM anymore. The Denial-of-Service attack can only be reverted from the command line interface of OpenNebula.

This bug can be triggered by a CSRF-attack, which means that it is a valid attack against a private cloud: By luring a victim onto a maliciously crafted website while logged in into Sunstone, an attacker can make all the victim's VMs uncontrollable via Sunstone. A video of the attack can be seen here:


This bug has been fixed in OpenNebula 4.6.2.

This result is a collaborative work together with Mario Heiderich. It has been published at ACM CCSW 2015. The paper can be found here.

Beliebte Posts aus diesem Blog

How To Spoof PDF Signatures

One year ago, we received a contract as a PDF file. It was digitally signed. We looked at the document - ignoring the "certificate is not trusted" warning shown by the viewer - and asked ourselfs:

"How do PDF signatures exactly work?"

We are quite familiar with the security of message formats like XML and JSON. But nobody had an idea, how PDFs really work. So we started our research journey.

Today, we are happy to announce our results. In this blog post, we give an overview how PDF signatures work and on top, we reveal three novel attack classes for spoofing a digitally signed PDF document. We present our evaluation of 22 different PDF viewers and show 21 of them to be vulnerable. We additionally evaluated 8 online validation services and found 6 to be vulnerable.

In cooperation with the BSI-CERT, we contacted all vendors, provided proof-of-concept exploits, and helped them to fix the issues and three generic CVEs for each attack class were issued: CVE-2018-16042

DTD Cheat Sheet

When evaluating the security of XML based services, one should always consider DTD based attack vectors, such as XML External Entities (XXE) as,for example, our previous post XXE in SAML Interfaces demonstrates.

In this post we provide a comprehensive list of different DTD attacks.

The attacks are categorized as follows:
Denial-of-Service AttacksClassic XXEAdvanced XXEServer-Side Requst Forgery (SSRF)XIncludeXSLT

Practical Dictionary Attack on IPsec IKE

We found out that in contrast to public knowledge, the Pre-Shared Key (PSK) authentication method in main mode of IKEv1 is susceptible to offline dictionary attacks. This requires only a single active Man-in-the-Middle attack. Thus, if low entropy passwords are used as PSKs, this can easily be broken.