In Linux kernels, a race condition in the IOMMU vt-d driver causes a freshly allocated PASID table to be written to a directory entry before a CPU cache flush completes. Because many IOMMU hardware units are non‑coherent, the device could observe stale, non‑zero contents in the PASID table while the kernel memory still holds uninitialized data. If a device accesses the table during this narrow window, it may be shown incorrect page table entries, which could allow the device to read or write memory locations it should not see. This could lead to memory corruption or escalation of the device’s privileges beyond its intended permissions—an outcome that is inferred from the description but not explicitly stated in the CVE notice. The CVSS score of 7.8 signals high severity and indicates that the vulnerability could have significant impact if exploited.

All Linux kernels derived from upstream before the commit that reordered the cache‑flush and PASID directory write remain vulnerable. The CPE details indicate the entire Linux kernel family. No specific version range is listed, so any distribution using a kernel built prior to the fix is at risk.

The CVSS score of 7.8 signals high severity, while the EPSS score of < 1% indicates a low exploitation probability. The flaw is not listed in the CISA KEV catalog. Exploitation would require privileged control over a device or hypervisor that can trigger IOMMU passthrough and time the access against the race. Because the attack vector involves a hardware timing race, successful exploitation is non‑trivial and would typically be limited to environments with direct device access, resulting in an overall risk that is low to moderate.