Google Cloud Flags Three Side-Channel Flaws in AMD Turin CPUs Powering Its VMs
Google Cloud has disclosed three side-channel vulnerabilities affecting AMD's Turin generation of CPUs, which underpin a portion of its virtual machine fleet. According to the company, exploitation of these flaws could allow data to leak across processes running on the same hardware. Google states it has already rolled out infrastructure-level mitigations to block attacks that cross virtual machine boundaries, but customers still need to apply a guest operating system patch and perform a stop-and-start of their instances to be fully protected.
Side-channel attacks exploit the physical characteristics of how a processor executes instructions — such as timing, cache behavior, or power consumption — rather than a software logic bug. In cloud and virtualization contexts, these flaws are particularly concerning because multiple customers' workloads often share the same physical CPU cores. A successful side-channel exploit can potentially allow one tenant's process, or a malicious VM, to infer or extract sensitive data (such as cryptographic keys or memory contents) belonging to another tenant running on the same hardware, even though the workloads are supposed to be isolated from one another.
This class of vulnerability has a long history in cloud computing, tracing back to issues like Spectre and Meltdown, and recurring disclosures tied to speculative execution and microarchitectural design choices in modern CPUs. Because these flaws live at the hardware level, fixing them typically requires a combination of microcode or firmware updates from the chip vendor, hypervisor-level mitigations from the cloud provider, and guest OS patches — making full remediation a multi-party effort rather than a single patch.
For defenders and cloud administrators, the practical takeaway is to treat vendor advisories about CPU and hypervisor-level flaws with the same urgency as application vulnerabilities, especially when they specify manual follow-up actions like OS patching and instance restarts. Organizations running workloads on shared cloud infrastructure should verify whether their instances rely on AMD Turin hardware, apply the required guest OS patches promptly, and restart affected VMs to ensure mitigations take effect, rather than assuming cloud-provider-side fixes alone are sufficient.
Source: The Hacker News
Read the original report ↗This is an original summary written by CyberSkii for reference. All rights to the underlying reporting remain with the publisher linked above.