HTTP/2 Implementation Errors Exposing Websites to Serious Risks

Organizations that don't implement end-to-end HTTP/2 are vulnerable to attacks that redirect users to malicious sites and other threats, security researcher reveals at Black Hat USA.

4 Min Read
Gonzalo Aragon via Shutterstock

BLACK HAT USA 2021 - Implementation flaws and imperfections in the technical specifications around HTTP/2 are exposing websites using the network protocol to a brand-new set of risks, a security researcher warned in a presentation at Black Hat USA Thursday.

James Kettle - director of research at PortSwigger who at Black Hat two years demonstrated so-called Desync attacks against websites using the HTTP protocol - this week showed how similar attacks could be carried out with potentially severe consequences against websites using the HTTP/2 standard.

As proof-of-concept, Kettle described attacks he was able to execute using his techniques against websites belonging to organizations such as Netflix, those powered by Amazon's application load balancer, and websites using Imperva's cloud Web application firewall. In many instances he was able to redirect requests from Web-facing servers at these sites to his own server.

Nearly 50% of all websites currently use the HTTP/2 (H2) protocol, which was introduced in 2015 as a faster and simpler alternative to HTTP/1.1. As Google describes it, "all the core concepts, such as HTTP methods, status codes, URIs, and header fields, remain in place," with the new protocol. "Instead, HTTP/2 modifies how the data is formatted (framed) and transported between the client and server, both of which manage the entire process, and hides all the complexity from our applications within the new framing layer."

According to Kettle, a whole slew of security issues can surface when organizations fail to use HTTP/2 in an end-to-end fashion. Instead, they have a front-end server that speaks HTTP/2 with clients and then rewrites requests from those clients back to HTTP/1.1 before forwarding them to a back-end server. 

"A vast majority of the servers that speak HTTP/2 actually speak HTTP/1 to the back-end," he said during his Black Hat talk. They speak H2 to the client and H1 with the back-end, Kettle said.

"This set up is ridiculously common," he noted.  Kettle pointed to Amazon's Application Load Balancer, for example, where this communication cannot be disabled. Such HTTP/2 downgrades and protocol translations gives attackers a way to carry out Desync attacks, Kettle said.

HTTP Desync attacks basically abuse weaknesses in how back-end servers interpret and respond to consecutive requests from a front-end server, load-balancer, or proxy server. For example, front-end servers speaking HTTP/2 follow a specific format for conveying message length to the back-end server. But a back-end server that only speaks HTTP/1.1 will not recognize the data because it derives information about the length of a request via other methods. 

Attackers can take advantage of disagreements over message length between the front-end server and back-end server to essentially interfere with the way an application might handle requests.

High-Profile Targets

To show how such an attack would work, Kettle pointed to an exploit he executed against Netflix where front-end servers performed HTTP downgrading without verifying request lengths. The vulnerability allowed Kettle to develop an exploit that triggered Netflix's back-end to redirect requests from Netflix's front-end to his own server. That allowed Kettle to potentially execute malicious code to compromise Netflix accounts, steal user passwords, credit card information, and other data. Netflix patched the vulnerability and awarded Kettle its maximum bounty of $20,000 for reporting it to the company.

In another instance, Kettle discovered that Amazon's Application Load Balancer had failed to implement an HTTP/2 specification regarding certain message-header information that HTTP/1.1 uses to derive request lengths. With this vulnerability, Kettle was able to show how an attacker could exploit it to redirect requests from front-end servers to an attacker-controlled server.  He found a vulnerable law-enforcement access portal while using the Amazon load balancer.

Almost every website using the Amazon load balancer was vulnerable to exploit, Kettle said. So, too, was a CMS powering multiple news sites such as Huffington Post - and every website using an Imperva WAF, he added.

During his presentation, Kettle highlighted several other exploits he had developed to take advantage of vulnerabilities that arise when organizations downgrade HTTP/2 to HTTP. He also released an updated version of HTTP Request Smuggler, a tool that organizations can use to detect HTTP/2 specific vulnerabilities on their network. Burp Suite vulnerability scanner has also been updated to detect these vulnerabilities, Kettle said.

"Please just avoid HTTP/2 downgrading," he advised. "Just speak HTTP/2 end-to-end. If you do that, about 80% of the attacks from this presentation simply won't work."

Read more about:

Black Hat News

About the Author(s)

Jai Vijayan, Contributing Writer

Jai Vijayan is a seasoned technology reporter with over 20 years of experience in IT trade journalism. He was most recently a Senior Editor at Computerworld, where he covered information security and data privacy issues for the publication. Over the course of his 20-year career at Computerworld, Jai also covered a variety of other technology topics, including big data, Hadoop, Internet of Things, e-voting, and data analytics. Prior to Computerworld, Jai covered technology issues for The Economic Times in Bangalore, India. Jai has a Master's degree in Statistics and lives in Naperville, Ill.

Keep up with the latest cybersecurity threats, newly discovered vulnerabilities, data breach information, and emerging trends. Delivered daily or weekly right to your email inbox.

You May Also Like

More Insights