GPU VulnDB

Database/NVIDIA / GPU stack

Linux kernel drm/xe: double-free of a managed buffer object when GSC initialization fails at probe

CVSS 7.8CVE-2026-64517NVIDIA / GPU stackcurated

Impact

If xe_gsc_init_post_hwconfig() fails, the Intel Xe GPU driver frees a buffer object explicitly. That object already has a devm cleanup action, so it is freed a second time when probe unwinds. The result is a host kernel double-free during driver load, which can mean memory corruption or a crash on nodes with Intel datacenter GPUs. In practice the exposure is narrow: the bug is only reached when GSC initialization fails during probe, which happens at boot or driver load, not through anything a tenant does. The record names no unprivileged trigger.

Who can reach it

Only reached on the driver probe failure path on hosts with Intel Xe GPUs. That is in effect operator or hardware initiated. The record does not identify any tenant or userspace trigger, even though it is scored AV:L/PR:L.

What to do

Install a stable kernel that includes the fix commits in the record and reboot, on hosts that use the xe driver. Nodes without Intel Xe GPUs are not affected. The record lists commits, not a fixed release number. Since the only path is probe failure, this can wait for the normal kernel update cycle.

References

Related entries

All NVIDIA / GPU stack 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.