HIGH Introduced in 6.8
nouveau Scheduler UAF
CVE-2026-98512
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:H
KernelScan AI7.8HIGH
01Description
In the Linux kernel, the following vulnerability has been resolved: drm/nouveau: RCU-free the scheduler-containing nouveau_sched struct nouveau_sched embeds a struct drm_gpu_scheduler (base). nouveau_sched_destroy() calls nouveau_sched_fini() (which does drm_sched_fini(&sched->base)) and then frees the object with plain kfree(sched). drm_sched_fence_get_timeline_name() returns fence->sched->name, and the scheduler fence keeps a .release callback so it is not ops-detached on signalling. A finished fence exported to userspace via drm_syncobj / sync_file therefore keeps pointing at &sched->base after nouveau_sched_destroy(), and a later get_timeline_name() -- reachable unprivileged through SYNC_IOC_FILE_INFO -- dereferences freed memory (KASAN slab-use-after-free read). Per the dma-fence lifetime contract the exporter must keep the data backing a signalled fence alive for an RCU grace period. Free the scheduler-containing object with kfree_rcu() instead of kfree().
02KernelScan AI Analysis
Risk summary
Any local user on a system with an NVIDIA GPU using the nouveau driver can trigger a use-after-free read in the GPU scheduler by destroying a channel or client (which frees the scheduler) and then querying a still-valid sync_file fence via SYNC_IOC_FILE_INFO. This leaks kernel memory, potentially defeating KASLR, and can be escalated to arbitrary kernel read/write for privilege escalation or denial of service.
Vulnerability analysis
The nouveau GPU driver embeds a drm_gpu_scheduler inside its nouveau_sched object. When a channel or client is closed, nouveau_sched_destroy() tears down the scheduler and immediately frees the containing object with kfree(). However, fences previously submitted to that scheduler remain alive in userspace via sync_file or drm_syncobj; these fences retain a pointer to the scheduler's name field. A subsequent unprivileged SYNC_IOC_FILE_INFO ioctl calls drm_sched_fence_get_timeline_name(), which dereferences the now-freed scheduler memory, causing a KASAN slab-use-after-free read. The fix adds an RCU head to nouveau_sched and switches the final free to kfree_rcu(), ensuring the scheduler memory persists until all RCU readers (including the timeline-name lookup) have completed, satisfying the dma-fence lifetime contract.
Lifecycle
03Fix Versions
| Branch | Introduced | Fixed in | Patch commit |
|---|---|---|---|
| 6.12 | 6.8 | 6.12.112 | 40edf5eefc3f |
| 6.18 | 6.8 | 6.18.55 | 35151e9a8274 |
| 7.2 | 6.8 | 7.2.9 | cd6e11b6ebe0 |
| mainline | 6.8 | 7.3 | f7eae6d8d768 |
| cip/6.12 | 6.8 | 6.12.112-cip33 | 40edf5eefc3f |
04CIP Kernels
State of the newest release of each Civil Infrastructure Platform SLTS tree, from CIP's own fix records.
| CIP tree | State |
|---|---|
| cip/6.12 | Fixed in 6.12.112-cip33 |