Researchers from a joint team involving multiple universities and Google have discovered two new cybersecurity attacks that exploit a specific feature in modern Intel processors.
These findings, which highlight significant vulnerabilities affecting billions of processors in use today, will be presented at the upcoming 2024 ACM ASPLOS Conference.
The study, titled “Pathfinder: High-Resolution Control-Flow Attacks Exploiting the Conditional Branch Predictor,” focuses on a part of the processor called the conditional branch predictor.
This feature helps improve computer efficiency by predicting the future paths a program might take.
However, the researchers have shown that this capability can be manipulated to gain unauthorized access to private data.
The team, led by the University of California San Diego, includes experts from Purdue University, Georgia Tech, the University of North Carolina Chapel Hill, and Google.
They have identified vulnerabilities in the Path History Register of the branch predictor, which records the order and addresses of branches a program takes. By exploiting this register, attackers can observe and manipulate program execution more precisely than ever before.
One of the attacks developed by the researchers allows them to reconstruct the sequence of branches a program takes, capturing detailed information about program execution.
This method was even used to extract images processed by a popular software library, demonstrating the potential for significant data breaches.
Another attack introduced by the researchers is a refined version of the known Spectre attack, which involves inducing errors in the branch predictor to force a program to execute unintended code. This can lead to the exposure of sensitive information such as encryption keys.
The researchers showcased this by manipulating an encryption algorithm to reveal its secret key.
The implications of these findings are vast, as the vulnerabilities could affect any device using the compromised processors, from personal computers to large data centers.
Both Intel and Advanced Micro Devices (AMD) have acknowledged the issues raised by the researchers and are set to release security updates to address these vulnerabilities.
Intel is preparing a Security Announcement, while AMD plans to issue a Security Bulletin, AMD-SB-7015, to inform users about the necessary precautions and updates.
The research team has also informed other hardware and software vendors affected by the discoveries, ensuring a coordinated response to this significant security threat.
This research not only exposes new vulnerabilities but also underscores the ongoing challenges in cybersecurity for modern computing infrastructures.
As processors become more advanced, so too do the techniques for exploiting them, highlighting the need for continuous advancements in security measures.
Source: UC San Diego.
