Intel CPUs vulnerable to new LVI attacks
Today, a team of academics from universities across the world, along with vulnerability researchers from Bitdefender, have disclosed a new security flaw in Intel processors.
Named Load Value Injection, or LVI for short, this is a new class of theoretical attacks against Intel CPUs.
While the attack has been deemed only a theoretical threat, Intel has released firmware patches to mitigate attacks against current CPUs, and fixes will be deployed at the hardware (silicon design) level in future generations.
A reverse Meltdown attack
To understand what an LVI attack is, users must first be aware of the Meltdown and Spectre attacks, and more particularly Meltdown.
Disclosed in January 2018, the Meltdown attack allowed an attacker running code on a CPU to read data from the CPU's memory, while the CPU was processing "speculative" operations.
Speculative execution is a feature of all modern CPUs, one in which the CPU computes information in advance in an attempt to guess future results. The entire idea of speculative execution is to have the data ready for the CPU, if it ever needs it, and help improve the CPU's speed and performance. Once data is not needed, it's discarded. Meltdown and Spectre attacks target data while in this "transient" state, while waiting to be dismissed.
The Meltdown and Spectre attacks were groundbreaking when they were first revealed in 2018, showing a major flaw in the designs of modern CPUs.
Based on the original attacks, academics around the world later expanded the original research and discovered an entire class of so-called "transient attacks" that also leaked data from CPUs in their "transient" speculative execution states.
Besides Meltdown and Spectre, other transient attacks were eventually discovered during the past two years, including the likes of Foreshadow, Zombieload, RIDL, Fallout, and LazyFP.
LVI's position in all these attacks is, technically, of a reverse-Meltdown. While the original Meltdown bug allowed attackers to read an app's data from inside a CPU's memory while in a transient state, LVI allows the attacker to inject code inside the CPU and have it executed as a transient "temporary" operation, giving attackers more control over what happens.
Tests performed by the two research teams -- who found the LVI attack independently from one another -- have been successful at proving the attack's broad impact.
For example, the academic research team focused on leaking data (an encryption key) from an Intel SGX enclave, a secure area of Intel processors, while Bitdefender focused on proving the attack's impact on cloud environments.
"This type of attack is particularly devastating in multi-tenant environments such as enterprise workstations or servers in the data center, where one less-privileged tenant would be able to leak sensitive information from a more privileged user or from a different
virtualised environment on top of the hypervisor," Bitdefender said.
Meltdown also needs a hardware fix
But the biggest finding related to this research paper is about how the Meltdown & LVI will need to be addressed.
When Meltdown was fist disclosed in January 2018, Intel said that a firmware patch was all that was needed, while a change of the CPU's silicon design was only needed for the class of Spectre attacks.
Now, researchers say this is not true anymore. Both the academic research team and the Bitdefender team say that the class of Meltdown and LVI attacks also now needs a hardware fix.
LVI bypasses some Meltdown fixes
"We exploit the same hardware operations as Meltdown," Daniel Gruss, an assistant professor at the Graz University of Technology, and a member of the academic research team told ZDNet.
"Therefore, if Meltdown works, LVI works as well."
Furthermore, Gruss says his team was also able to successfully use LVI attacks on systems that received software fixes against Meltdown-type attacks, suggesting that existing fixes may not be always successful at blocking new versions of Meltdown-like attacks.
Only Intel CPUs confirmed to be affected -- for now
Currently, only Intel CPUs have been confirmed to be impacted by the new LVI attacks in real-world tests. However, researchers don't rule out that CPUs from AMD and ARM could also be affected.
"In principle, any processor that is vulnerable to Meltdown-type data leakage would also be vulnerable to LVI-style data injection," researchers wrote on a website dedicated to the LVI attacks.
"Some non-Intel processors have been shown to be affected by some variants of Meltdown and Foreshadow," they added.
"We maintain an up-to-date overview on the website Transient.fail website. Select Meltdown + vendor ARM or AMD."
Researchers suggest that the Meltdown variations listed on the website, of which there are few, could be used for theoretical injection points for an LVI attack on other vendors' CPUs; although they have not verified any such claims in practice, so far.
JavaScript delivery vector
Current LVI attack demos rely on running malicious code on a computer, suggesting that local access is needed -- such as delivering malicious code to the target via malware.
However, a remote attack is also possible via JavaScript, by tricking users into accessing a malicious site -- similar to the original Meltdown attack, which could also be carried out via JavaScript.
This JavaScript attack vector has not been proven in a test; however, both Gruss and Bitdefender have told ZDNet that a JavaScript delivery method should also work, at least theoretically.
A general view, patches, performance impact
But while the two research teams discovered on their own, they also came at another shared conclusion -- namely that an LVI attack would be hard to pull off in practice, compared to many other side-channel attacks disclosed in the past.
Currently, LVI is viewed as more of a theoretical attack that opens the door for future discussions. Similar to how each cryptographic attack has built on previous research to the point where an encryption algorithm was eventually deemed broken, the LVI attack is just another brick in the wall of CPU vulnerabilities that have been disclosed in the past two years.
An LVI attack may not be a direct danger to users right now, but as academics disclose more attacks and learn more about how CPUs actually work, the current CPU design will eventually be proven to be insecure.
Both research teams hope their work would push CPU design towards a secure model before any attacks become reliable for use in a real-world scenario.
While a change in the silicon design will eventually come with future CPUs, currently, Intel has prepared software-based mitigations, in the form updates to the SGX Platform Software and its SDK (software development kit).
However, according to preliminary tests, these mitigations come with a severe performance impacted that may slow down computations from 2 to 19 times, depending on the number of mitigations system administrators decide to apply to their CPUs.
Currently, many administrators are expected to skip these patches, primarily because of the severe performance impact.
For good reasons, Intel has downplayed the severity of the LVI attack, and, for once, researchers have agreed.
"Due to the numerous complex requirements that must be satisfied to successfully carry out, Intel does not believe LVI is a practical method in real-world environments where the OS and VMM are trusted," an Intel spokesperson told ZDNet in an email last week.
"Agree with Intel," Bogdan Botezatu, Director of Threat Research and Reporting, told ZDNet yesterday. "This type of attack is much harder to pull off in practice, compared with other side-channel attacks such as MDS, L1TF, SWAPGS."
Below is a long list of resources that ZDNet readers can consult to learn more about the new LVI attacks disclosed today:
LVI website
LVI academic research paper [PDF]
Bitdefender LVI paper [PDF]
Bitdefender blog post on LVI
Intel LVI technical paper (includes list of impacted CPUs) [PDF]
List of impacted CPUs [spreadsheet]
Intel one-page summary on LVI [PDF]
Intel blog post on LVI
Academic team's proof-of-concept LVI demo code
Bitdefender's proof-of-concept LVI demo code