Database/Kernel, userspace & hypervisor

Linux kernel xen-netback: failed ring mapping leads to double unmap of guest data rings (XSA-532)
Impact
When xen-netback fails to map a guest's frontend data rings after one ring is already mapped, it unmaps them but leaves the ring pointers in place, so the cleanup path then unmaps the same rings a second time. The netback backend runs in the host or driver domain and serves every guest's virtual NIC, so trouble in that path affects the host and the guests that share it, not just one VM. The record gives no exploitability assessment beyond the double unmap, and the Xen project tracks it as XSA-532. It only matters on Xen hosts that use the in-kernel netback driver. KVM, ESXi and bare-metal GPU nodes are not affected.
Who can reach it
A guest VM on a Xen host that controls its own netfront ring setup during connection, which can make the backend's ring mapping fail. Needs guest kernel control, not host credentials. Only Xen dom0 or driver domains running xen-netback are exposed.
What to do
Install a kernel with the stable fix in the dom0 or driver domain and reboot the host. Guests have to be migrated off or shut down first. Check XSA-532 on oss-security for distribution and version details. The record names no workaround.
References
Related entries
- Linux kernel FUSE over io_uring: unprivileged user can trigger BUG_ON and panic the hostCVE-2026-99293 · Linux kernel FUSE over io_uring (fuse_copy_fill)Unscored
- Linux kernel libceph: short monitor command ack reads unreceived bytes and can crash rbd mapCVE-2026-99296 · Linux kernel libceph (mon_command ack decode)Unscored
- Linux kernel nvmet-rdma: short RDMA-CM connection request causes out-of-bounds read on the NVMe-oF targetCVE-2026-99302 · Linux kernel NVMe-oF RDMA target (nvmet-rdma connect request parsing)Unscored
- Linux kernel CephFS client: integer overflow in snap trace bounds check lets an MDS trigger out-of-bounds readCVE-2026-99304 · Linux kernel CephFS client (ceph_update_snap_trace bounds check)Unscored
- Linux kernel CephFS client: mdsmap claiming more than CEPH_MAX_MDS ranks causes stack out-of-bounds readCVE-2026-99306 · Linux kernel CephFS client (mdsmap rank count, check_new_map)Unscored
- Linux kernel CephFS client: undersized dentry lease record from MDS causes out-of-bounds read on lookupCVE-2026-99307 · Linux kernel CephFS client (MDS reply dentry lease parsing)Unscored
This entry is curated: imported from vendor advisories with machine assistance, not yet individually verified. Confirm against your vendor's advisory before acting, and report anything wrong.