Check Point Research analysts who discovered two vulnerabilities in the Microsoft Azure cloud infrastructure have published the details of how these flaws were found and how attackers could potentially use them.
The research team began exploring Azure infrastructure in an effort “to disprove the assumption that cloud infrastructures are secure,” says security researcher Ronen Shustin in a blog post. Check Point informed Microsoft of the vulnerabilities as they were discovered throughout 2019; security patches were deployed for both flaws by the end of the year.
CVE-2019-1234 is a server-side request forgery bug in an on-prem Azure environment called Azure Stack, a hybrid cloud tool for enterprise use. A spoofing flaw exists when Azure Stack fails to validate certain requests. Attackers could exploit this by sending a crafted request to the Azure Stack portal; if successful, they could make requests to internal Azure Stack resources.
Researchers conducted this investigation by first installing the Azure Stack Development Kit (ASDK) on their own servers and mapping areas they thought they might find vulnerabilities. ASDK comes with a set of core components that can be extended via features like App Service and SQL Providers, among others. It has a limited number of features compared to the Azure cloud and usually runs software that’s a couple of versions behind, researchers explain. Azure Stack shares similar features with the Azure public cloud, so they focused on those vectors.
One Azure Stack service they investigated was called DataService, which didn’t require any authentication. This flaw could enable an attacker to obtain sensitive data belonging to any business with its machine running on Azure, whether it’s on a shared or isolated machine. They would first have to gain access to Azure Stack Portal and then send unauthenticated HTTP requests, which could provide screenshots and data about the tenants and infrastructure machines.
This vulnerability is only valid to the Azure Stack, which is also “a very valid attack vector,” said Yaniv Balmas, head of security research, in a meeting with reporters at last week’s CPX 360 event. “Practically, if someone has a big Azure Stack with a lot of tenants, I can take screenshots of other machines. It could be dangerous; it could not be. It depends.”
Remote code execution vulnerability CVE-2019-1372 is in Azure App Service, which lets users build and host Web apps, mobile backends, and restful APIs in a programming language they choose without managing infrastructure. The bug exists when Azure Stack fails to check the length of a buffer before copying memory to it. An attacker could let an unprivileged function, run by the user, execute code in the context of NT AUTHORITY/system and escape the sandbox.
Attackers could use this vulnerability to compromise tenant applications, data, and accounts by creating a free user in the Azure Cloud and running malicious Azure functions. Alternatively, they could send unauthenticated HTTP requests to the Azure Stack user portal. If successful, they could potentially take control over the entire Azure server and the business code it holds.
“[This] can give complete visibility into every workload running on the same server,” said Balmas. “[I] can modify them, I can delete them, I can do whatever I want.”
In its research, Check Point only used this vulnerability to demonstrate how it could be exploited to crash the Dynamic WAS Service (DWASSVC). However, it could also be used for privilege escalation, they explain in a blog post about CVE-2019-1372.
Kelly Sheridan is the Staff Editor at Dark Reading, where she focuses on cybersecurity news and analysis. She is a business technology journalist who previously reported for InformationWeek, where she covered Microsoft, and Insurance & Technology, where she covered financial … View Full Bio