Categories: News

AMD VM security tools can be bypassed, letting hackers infilitrate your devices, experts warn

University researchers find way to trick AMD’s secure VM tech into sharing sensitive information

AMD released a security advisory with recommended mitigations

The bug requires physical access, but in some scenarios could be abused through software alone

AMD Virtual Machine (VM) security tools can be bypassed, allowing cybercriminals to infiltrate devices and steal sensitive data, experts have warned.

Security researchers from multiple universities in the EU and the UK outlined how AMD chips come with Secure Encrypted Virtualization (SEV), a security feature designed to encrypt virtual machine (VM) memory to protect it from unauthorized access, including by the hypervisor or other VMs. SEV uses hardware-based encryption keys unique to each VM, ensuring data remains confidential and tamper-proof even in shared environments.

Recently, AMD upgraded SEV with Secure Nested Paging (SNP), an enhancement that should guarantee the integrity of a virtual machine’s memory by validating memory page tables and preventing unauthorized modifications by the hypervisor. However, the researchers claim SNP can be bypassed, if the attacker has physical access to the target device. They used a Raspberry Pi Pico to “fake” additional RAM, and exfiltrated secrets from it. They call the attack BadRAM.

Attacks with and without hardware

By abusing the Serial Presence Detect (SPD) chip on the memory’s module, they created aliases for physical memory, which they could later access.

“In our attacks, we double the apparent size of the Dual Inline Memory Module (DIMM) installed in the system to trick the CPU’s memory controller into using additional ‘ghost’ addressing bits,” the researchers said in a research paper titled “BadRAM: Practical Memory Aliasing Attacks on Trusted Execution Environments”.

“These addressing bits will be unused within the virtually enlarged DIMM, creating an interesting aliasing effect where two different physical addresses now refer to the same DRAM location.”

The technique works on DDR4 and DDR5 memory, and there’s even a possibility to use it without hardware, if the SPD chip is left unlocked (which, according to The Register, is sometimes the case).

Are you a pro? Subscribe to our newsletter Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed! Contact me with news and offers from other Future brands Receive email from us on behalf of our trusted partners or sponsors

In response, AMD released a security advisory, and is now tracking the flaw as CVE-2024-21944. It has gotten a severity score of 5.4 (medium). It recommends utilizing memory modules that lock SPD, as well as following physical security best practices. “In addition, the AGESA and SEV FW versions listed below have been released to the Original Equipment Manufacturers (OEM) to mitigate this issue. Please refer to your OEM for the BIOS update specific to your product.”

Further details can be found here.

Via The Register

Original Author: Sead Fadilpašić | Source: TechRadar

Akshit Behera

Share
Published by
Akshit Behera

Recent Posts

Trump administration’s deal is structured to prevent Intel from selling foundry unit | TechCrunch

The deal allows the U.S. to take more equity in Intel if the company doesn't…

5 months ago

3 Apple Watches are rumored to arrive on September 9 – these are the models to expect

We're expecting two new models alongside the all-new Apple Watch Series 11. | Original Author:…

5 months ago

Fujitsu is teaming with Nvidia to build probably the world’s fastest AI supercomputer ever at 600,000 FP8 Petaflops – so Feyman GPU could well feature

Japan’s FugakuNEXT supercomputer will combine Fujitsu CPUs and Nvidia GPUs to deliver 600EFLOPS AI performance…

5 months ago

Microsoft fires two more employees for participating in Palestine protests on campus

Microsoft has fired two more employees who participated in recent protests against the company’s contracts…

5 months ago

Microsoft launches its first in-house AI models

Microsoft announced its first homegrown AI models on Thursday: MAI-Voice-1 AI and MAI-1-preview. The company…

5 months ago

Life 3.0 – Being Human in the Age of Artificial Intelligence by Max Tegmark

A comprehensive review of Max Tegmark's Life 3.0, exploring the future of artificial intelligence and…

5 months ago