UBUNTU-CVE-2026-99147: Medium severity vulnerability
In the Linux kernel, the following vulnerability has been resolved: drm/exec: don't immediately add prelocked obj to array of locked objs When a contention is detected the drmexec object grabs a reference to the contended object, unlocks all other objects and pre-locks the contended one before anything else. The problem is that it is possible that this pre-locked object was about to be destroyed and drmexec is now holding the last reference to it. So we never actually see this pre-locked object again to be explicitly locked and so also never reserve a slot for a dmafence object on it. Should the driver now use drmexecforeachlockedobject() to add a new fence to all locked objects we eventually see a warning or even a BUGON() to prevent random memory corruption. Fix this by keeping the prelocked object out of the array of locked objs until it was explicitly mentioned by the driver again. If the prelocked object is never mentioned again by the driver just unlock and drop the reference at the end of the locking loop.
Frequently Asked Questions
What is the potential operational impact if this condition is triggered?
A driver adding fences to locked objects can hit a warning or BUG_ON(). The check is intended to prevent random memory corruption.
Which code paths are associated with the issue?
The issue involves DRM drivers that use drm_exec_for_each_locked_object() to add a new fence to locked objects after drm_exec handles lock contention and pre-locks an object.
Is a fix available?
Yes. The vulnerability is described as resolved by changing drm_exec so a prelocked object is not added to the locked-object array until the driver explicitly refers to it again.