Intel's CPU Flaws Continue to Create Problems for the Tech Community

We can't wait out this problem and hope that it goes away. We must be proactive.

Irfan Ahmed, Assistant Professor in the Department of Computer Science at Virginia Commonwealth University

December 10, 2019

4 Min Read
Dark Reading logo in a gray background | Dark Reading

The tech community was once again blindsided with news last month of another security exploit involving Intel's processors; exploits have continued to be discovered since Meltdown and Spectre were first unveiled two years ago, causing widespread concern about the ramifications for computer systems globally. In addition to leaving sensitive data exposed, the vulnerabilities also put businesses in the difficult, but necessary, position of implementing mitigations that can seriously reduce the performance of computers and servers.

In January 2018, researchers revealed two exploits that take advantage of side-channel vulnerabilities found in computer chips manufactured since the mid-1990s. Since that time, six additional exploits — Foreshadow ZombieloadRIDLFallout, SWAPGS, and now TAA  — have been discovered that take advantage of the same vulnerabilities. While chips made by AMD and ARM are affected to a minimal degree, the vast majority of Intel's chips are affected by all of these exploits. And due to Intel's dominant market position, this vulnerability can be found in nearly every computer on the planet.

These exploits take advantage of a process called "speculative execution," a process introduced in the 1990s by Intel and other chipmakers as they sought to increase the speed of computer processors. In short, computer processors can "speculate" (or guess) what a user will run next, increasing speed by not having to wait to execute actions until they are formally received. While this process was credited with significantly improving the speed of computers, the exploits are able to give unauthorized users access to what should be confidential data, creating a vast security vulnerability. They typically leak data from different internal CPU buffers such as line-fill buffers, load ports, and store buffers.

To address this problem, Intel has provided software patches or businesses can apply other workarounds, such as disabling hyper-threading technology in vulnerable computers. However, both of these fixes can reduce the performance of CPUs. LoginVSI recently released a survey of IT professionals regarding the impact of the patches and found that approximately 20% of them experienced performance reductions of up to 10% on their systems, and another 11% said they experienced a performance hit up to 15%. Some respondents had performance impacts as high as 20%.

While addressing this problem is challenging, what is clear, as noted recently by a leading Linux developer, is that the security problems with Intel's chips "are not going away." So, we cannot simply wait this problem out and hope that it disappears. We must be proactive.

To understand the extent of the risk, the first thing any business should do is conduct an audit of the CPUs that it has in its systems. The easiest approach would then be to replace all affected CPUs with unaffected hardware. However, replacing all affected hardware may very well be cost-prohibitive.

Therefore, businesses should begin immediately diversifying and randomizing their CPUs. It can do this by purchasing unaffected chips (for example, from AMD) as it goes through its normal upgrade cycles and then randomizing affected Intel chips across its systems, strategically placing them in servers and computers where they are least vulnerable to hacks. In addition, by placing the affected CPUs in areas with lighter workloads, a business can also reduce the overall effect of the performance reductions caused by the software patches.

It would also be prudent to hire experienced IT security staff, plan for the increased energy costs of running current systems at maximum for longer periods of time to offset the performance reductions of the patches, or identify revenue streams to purchase new servers to add processing capacity.

While it is clear that being insecure is not a practical option, businesses must remember that there is not a one-size-fits-all solution. A company's remedies to this ongoing challenge must be assessed within the context of its own unique and dynamic technology environment. Undoubtedly, this challenge will be expensive, burdensome, and time consuming for businesses.

Related Content:

Check out The Edge, Dark Reading's new section for features, threat data, and in-depth perspectives. Today's top story: "Criminals Hide Fraud Behind the Green Lock Icon."

About the Author

Irfan Ahmed

Assistant Professor in the Department of Computer Science at Virginia Commonwealth University

Irfan Ahmed is an Assistant Professor in the Department of Computer Science at Virginia Commonwealth University (VCU), where he runs the Security and Forensics Engineering (SAFE) Lab. His research interests include system security, malware, digital forensics, and industrial control systems.

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