GPU VulnDB

Database/Kernel, userspace & hypervisor

Linux kernel KVM SEV: userspace can trivially trigger a kernel WARN via an oversized encrypt-region ioctl

CVSS 5.5CVE-2026-31590Kernel, userspace & hypervisorcurated

Impact

A process able to issue KVM_MEMORY_ENCRYPT_REG_REGION with an absurd size hits a WARN in sev_pin_memory() when the page count overflows an int. The upstream note is blunt that the WARN is trivial to trigger from userspace; the fix simply drops it. The practical effect is log spam and, on hosts configured with panic_on_warn, a host panic - which on a confidential-computing GPU node means every guest on that box goes down. There is no memory corruption or escape described here. Exposure is limited to AMD SEV hosts where something untrusted can reach the KVM device.

Who can reach it

Local process with access to /dev/kvm on an AMD SEV host - typically the VMM, so in most fleets this is the hypervisor stack rather than a tenant.

What to do

Pick up a stable kernel with the referenced commits and reboot the affected SEV hosts, which means draining guests first; there is no module-level or runtime mitigation. If panic_on_warn is set on hosts where untrusted code can reach /dev/kvm, clearing it removes the only outcome worse than a log line while you wait for the reboot window.

References

Related entries

All Kernel, userspace & hypervisor entries

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.