This flaw in the Linux kernel allows a Xen paravirtualized guest to present guest‑physical addresses that map to memory belonging to other domains. When virtio‑gpu reports framebuffer backing pages using sg_phys() without validating the address translation, the host can read or write those unrelated pages. The result is exposure of confidential data from other guests or, in the worst case, corruption of host memory. The weakness is rooted in improper isolation of memory addresses and failure to enforce access control on the guest‑host boundary.

The issue affects the Linux kernel in Xen paravirtualized domains that use virtio‑gpu without the iommu_platform flag, as demonstrated with Xen 4.21 and QEMU 8.2 on kernel releases 6.8 and 6.18 LTS. PVH (paravirtualized host) dom0s are not impacted because they use identity mapping; only PXV dom0s with the affected kernel are vulnerable.

Exploitation requires a guest that can influence virtio‑gpu configuration in a PV domain and can control whether QEMU enables the iommu_platform option. The likely attack vector is a compromised guest environment presenting crafted virtio‑gpu descriptors, and is inferred from the description. The lack of a CVSS score and an unavailable EPSS score indicate that the general exploitation probability has not been quantified, and the vulnerability is not currently listed in the CISA KEV catalog. Nonetheless, the potential for cross‑domain information disclosure or host memory corruption makes this a high‑severity issue for affected Xen setups.