This flaw involves incorrect handling of DMA API usage in the virtio‑gpu subsystem when running on Xen PV guests. Due to a missing Xen domain check, guest‑physical addresses are conveyed to the host via sg_phys(), which the host interprets as addresses belonging to other domains. Consequently, the host can read or overwrite memory that belongs to unrelated guests or the host itself, leading to confidentiality loss and potential host memory corruption. The weakness manifests as an improper access control and information disclosure, involving CWE‑497.

Affected systems are Linux kernel versions that support the DRM/virtio‑gpu driver, running on Xen PV dom0 hosts where the virtio‑gpu device is instantiated without the iommu_platform flag. The vulnerability has been reproduced on Xen 4.21 with QEMU 8.2 on kernel releases 6.8 and 6.18 LTS. PVH dom0 hosts that use identity mapping are not impacted, but standard PV dom0 installations are exposed.

The CVSS score of 7.8 indicates moderate‑to‑high severity, while the EPSS score of < 1 % suggests a low current exploitation probability. The vulnerability is not listed in CISA’s KEV catalog. The likely attack vector is a compromised guest that can influence virtio‑gpu configuration, as inferred from the requirement that the guest must supply a guest‑physical framebuffer to the host. Based on the description, it is inferred that an attacker would need a guest capable of controlling virtio‑gpu settings to exploit the flaw. A successful exploit could allow a malicious domain to leak or corrupt host memory or data from other domains, posing a serious risk to multi‑tenant Xen deployments.