Read the docs
TI-2026-81890 Aliases CVE-2026-99183GHSA-cgc9-44mx-vhwg

NULL pointer dereference in Linux kernel rocket NPU driver

Priority LOW 22Published · Updated

Key facts

Attack vector
—
Privileges required
—
User interaction
—
Attack complexity
—
Impact type
—
Confidentiality · integrity · availability
—
Fix status
Fixed

No CVSS vector from any source. Fix status from CNA.

Summary

The Linux kernel accel/rocket driver from 6.18 searches only part of its core slots when looking up a scheduler after a core is unbound. Unbinding a core that is not the last one makes the lookup return NULL for a core that is still bound and running jobs, so the driver dereferences a NULL pointer. Clients submitting jobs to the remaining cores can trigger the kernel NULL pointer dereference.

Written with Claude Haiku 5.5 from the sources below · 2026-11-06

Description from the CNA

In the Linux kernel, the following vulnerability has been resolved: accel/rocket: search every core slot when looking up a scheduler sched_to_core() walks rdev->cores[] up to rdev->num_cores, and rocket_remove() decrements num_cores for every core it removes. Unbind a core that is not the last one and the cores behind it fall outside the search, so sched_to_core() returns NULL for a core that is still bound and still running jobs. Neither caller checks the result: rocket_job_run(): rocket_fence_create(core), core->dev rocket_job_timedout(): dev_err(core->dev, "NPU job timed out") Unbinding the middle core of the three on an RK3588 while three clients are submitting to all of them faults twice, once from the surviving core's job queue and once from its reset work: KASAN: null-ptr-deref in range [0x0000000000000220-0x0000000000000227] Workqueue: fdad0000.npu drm_sched_run_job_work [gpu_sched] pc : rocket_job_run+0x234/0x838 [rocket] Call trace: rocket_job_run+0x234/0x838 [rocket] drm_sched_run_job_work+0x2cc/0xad8 [gpu_sched] process_one_work+0x640/0x14f0 KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007] Workqueue: rocket-reset-2 drm_sched_job_timedout [gpu_sched] pc : rocket_job_timedout+0xf0/0x1e0 [rocket] Call trace: rocket_job_timedout+0xf0/0x1e0 [rocket] drm_sched_job_timedout+0x188/0x6a0 [gpu_sched] Both are the third core: the workqueue names are its device and its core->index, and it was left at slot 2 while num_cores had dropped to 2. Search all the slots that were allocated, the way find_core_for_dev() now does. A core that is still bound is then found, and the two callers get the pointer they already assume they have. This does not make unbinding one core out of several safe. An open client keeps an entity pointing at the scheduler of the core that went away: drm_sched reports it as not ready for every job that lands on it, and the client waits in dma_fence_default_wait for a fence that will never signal. Stopping the NULL dereference is what belongs in a fix; the rest wants more thought.

Why it matters now

EPSS 0.00167 5th percentileScored 2026-11-07
Fix Fixed Linux (Linux): 6.18.61, 7.2.13, 7.3

SSVC decision points

Not available: CISA has not published SSVC decision points for this record.

Affected products

VendorProductAffected versionsFixed inStatusSource
LinuxLinux6.18 and later6.18.61, 7.2.13, 7.3Patch availableCNA
  • Linux VendorLinux Affected 6.18 and later Fixed in 6.18.61, 7.2.13, 7.3 Patch available CNA

CPE names Threatist (inferred) · 1

Inferred from how NVD named this product before; not given by a source.

CPEVersionsRoleVendor
cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*——Linux
  • cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:* Versions — — ·Linux

Remediation

Upgrade the Linux kernel to 6.18.61, 7.2.13 or 7.3.

Written with Claude Haiku 5.5 from the sources below · 2026-11-06

  • Fix · CNA Linux (Linux) Fixed in 6.18.61, 7.2.13, 7.3

Vendor assessments

Red Hat

Red Hat rating: Moderate. Red Hat's page for this CVE →

Not affected: 7 products
  • Red Hat Enterprise Linux 10
  • Red Hat Enterprise Linux 6
  • Red Hat Enterprise Linux 7
  • Red Hat Enterprise Linux 8
  • Red Hat Enterprise Linux 9
  • Red Hat OpenShift Container Platform 4
  • Red Hat OpenShift Container Platform 5

Timeline

  1. 2026-10-28 Threatist record created from CVE-2026-99183
  2. 2026-10-28 Alias added: GHSA-cgc9-44mx-vhwg
  3. 2026-11-02 Affected versions changed · Red Hat
  4. 2026-11-02 CVSS score added · Red Hat
  5. 2026-11-03 Affected versions changed · Red Hat
  6. 2026-11-06 Fix available: Linux/Linux@16c95870c95f2a3d78c81b9fbc0645f4b7178c11, Linux/[email protected] · CNA
  7. 2026-11-06 1 new reference · CNA

References

  • TI-2025-26358 tls: fix handling of zero-length records on the rx_list KEVCVSS 3.1CRITICAL 9.8
  • TI-2026-38352 netfilter: bridge: make ebt_snat ARP rewrite writable KEVCVSS 3.1HIGH 8.8
  • TI-2021-02452 Incorrect authorization in Ubuntu linux kernel KEVCVSS 3.1HIGH 8.8
  • TI-2013-05264 The (1) get_user and (2) put_user API functions in the Linux kernel before 3.5.5 on the… KEVCVSS 3.1HIGH 8.8
  • TI-2022-00125 Integer overflow in kernel KEVCVSS 3.1HIGH 8.4
  • TI-2013-01864 The perf_swevent_init function in kernel/events/core.c in the Linux kernel before 3.8.9… KEVCVSS 3.1HIGH 8.4
See all 21741 →

Sources