CVE-2024-26912

What it actually does

The vulnerability is a set of DMA buffer leaks in the Nouveau driver where nvkm_gsp_mem objects are not deallocated after GSP-RM initialization or on driver unload; the fix also adds poisoning and clears objects on free, and corrects the size field type.

Why it matters to the business

On systems using Nouveau graphics, repeated driver load/unload cycles can exhaust kernel memory, leading to degraded performance, instability, or crashes that disrupt workloads relying on those GPUs.

How an attack unfolds

A local attacker with access to trigger Nouveau initialization repeatedly causes the driver to allocate DMA buffers that are never freed, gradually consuming kernel memory until the system becomes unresponsive or crashes.

How to fix it

Upgrade the Linux kernel to a fixed version such as 6.7.6 or apply the commit 6190d4c08897d748dd25f0b78267a90aa1694e15; ensure all downstream stable trees incorporate the same changes.

How to confirm the fix worked

Confirm the running kernel contains the referenced commit or is at least version 6.7.6, then verify Nouveau initialization completes without memory growth attributable to nvkm_gsp_mem objects.

Typical attack path

1. Obtain local access to a Linux system using the Nouveau driver
2. Trigger repeated initialization and teardown of the graphics driver or GSP-RM subsystem
3. Accumulate leaked DMA buffers until kernel memory is exhausted
4. Cause denial of service through resource starvation or forced crashes

How to detect it

Review kernel logs and memory accounting for repeated allocations of nvkm_gsp_mem objects that are not freed on driver unload; correlate with Nouveau initialization events.

What your security team should do

• Prioritize patching kernel packages on all systems using Nouveau graphics
• Monitor memory usage on affected hosts during driver load/unload cycles
• Consider disabling Nouveau on servers where the GPU is not required until patches are applied