GPU VulnDB

Database/Kernel, userspace & hypervisor

Linux virtio-mmio: use-after-free of vm_dev when the platform device is unbound

CVSS 7.8CVE-2023-53515Kernel, userspace & hypervisorcurated

Impact

virtio-mmio allocated its vm_dev structure with devres, so the memory is released when the platform device goes away rather than when the embedded struct device's release callback runs, giving a use-after-free on unbind. This is the virtio transport used by lightweight VMMs; on a GPU fleet that matters where guests run under Firecracker or a QEMU configuration using MMIO virtio rather than PCI. Reaching it requires driver unbind through sysfs, which is a root operation on the host, so the realistic impact is host instability during administrative teardown rather than a tenant-reachable escalation.

Who can reach it

Local, root on the host: unbinding the virtio-mmio platform device via sysfs. Not reachable from inside a guest or a container.

What to do

Patch the host kernel to a stable release with the fix (the driver stops using devres for vm_dev) and reboot the node after draining it. Avoid sysfs unbind of virtio-mmio devices on unpatched hosts in the meantime.

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.