data built 2026-10-19T13:16:53Z

Look up another CVE · open CVE-2026-89570 in the interactive tracker

CVE-2026-89570

cxl/mce: Make the MCE notifier per-region

7.8 EPSS 0.2% AV:L

Published 2026-09-11 (37 days ago) by the Linux kernel CNA. Subsystem drivers/cxl. Debian urgency: unimportant.

What to do, per Debian release

Debian's own status for CVE-2026-89570 in every release this site tracks, and what each one means in practice. Every status below is the one the Debian Security Team publishes for this CVE. source

ReleaseStatus VersionWhat this means
Debian 12bookwormnot affected6.1.190-1Nothing to do. bookworm never shipped the vulnerable code.
Debian 13trixienot affected6.12.113-1Nothing to do. trixie never shipped the vulnerable code.
Debian 14forkyfixed7.2.6-1Update the kernel and reboot. Fixed in 7.2.6-1. A running kernel keeps the old code until the machine restarts.
Debian unstablesidfixed7.2.6-1Update the kernel and reboot. Fixed in 7.2.6-1. A running kernel keeps the old code until the machine restarts.

The vulnerability

What the bug is

Flavien Solt reported lifetime issues with the CXL MCE notifier, which can lead to NULL dereferences and use-after-free in the MCE handler. The notifier was registered per memory device and stored in 'struct cxl_memdev_state', even though it only needs the region state (the region's SPA range and its extended linear cache size). Instead of keeping the memory device and endpoint alive, the correct fix is to move the notifier into 'struct cxl_region' and register it from cxl_region_probe() as it should be a per-region notifier. Setup the registration to only happen for regions that have an extended linear cache as that is the only current usage. Remove cxl_port_get_spa_cache_alias() as it is now dead code. [ dj: Update dev_warn() when notifier fails due to kconfig. (Ben) ]

Where it lives

The fix touches these files. If the subsystem is one you do not use, the practical exposure is lower, though the package is still the vulnerable one.

drivers/cxl/core/mbox.c
drivers/cxl/core/mce.c
drivers/cxl/core/region.c
drivers/cxl/cxl.h
drivers/cxl/cxlmem.h

What this is
COMPUTE EXPRESS LINK (CXL)MAINTAINERS section matched on drivers/cxl/ · linux-cxl@vger.kernel.org · Maintained

Check it on the machine

grep -w CONFIG_CXL_BUS /boot/config-$(uname -r)

Reads the booted kernel's own configuration, on the machine and nowhere else. =y is built in, =m is a module, and an is not set line means the code is not there.

Show whether Debian builds this, per release

Debian builds this, but not on every flavour. At least one release below builds the code on some of its kernel flavours and not on others, so the answer depends on the architecture and flavour installed.

  • bookwormYnot built on armhf/armmp, armhf/armmp-lpae, armhf/rt/armmp, i386/686, i386/686-pae, i386/rt/686-pae
  • trixieYnot built on armel/rpi, armhf/armmp, armhf/armmp-lpae, armhf/rt/armmp, s390x/s390x
  • forky, sidYnot built on armhf/armmp, armhf/armmp-lpae, loong64/loong64, s390x/s390x
Y
built in where it is built at all; some flavours do not build it
Build switch
CONFIG_CXL_BUS

From the configuration Debian publishes for every kernel flavour it builds. It covers only the architectures Debian ships a linux-config package for, and nothing about a kernel you built yourself. Debian's own status above stays authoritative about the source package.MAINTAINERS kbuild Makefiles Debian linux-config packages

Signals

Published
2026-09-11 (37 days ago)source
CVSS
7.8 high
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:Hsource recompute it
Attack vector
local (AV:L)source
EPSS
0.17%, higher than 5.8% of all CVEssource
CISA KEV
not listedthe KEV catalogue
Debian urgency
unimportantsource
Subsystem
drivers/cxl

Upstream fixes

The versions the CNA records as introducing and fixing this, and the commit for each. source

IntroducedFixed inCommit
6.156.18.515563db13c952
6.157.2.4491d8c9ac98d
6.157.3775d0f4558f4

Why the CNA scored it this way

Quoted from the kernel CNA's own scoring file, word for word. A highlighted metric carries the most severe value CVSS v3.1 defines for it, which is what pushes the score up. source

AV:LAttack vector, local
cxl_handle_mce() is invoked from the host x86 MCE decode chain (mce_gen_pool_process() workqueue) after a local machine check; it is not reached from a network protocol, Bluetooth/WiFi, or USB handler. Local access to a CXL Type-3 host is the applicable vector (higher severity than Physical).
AC:LAttack complexity, low
The notifier is registered on the PCI device before cxlds->cxlmd is set and while cxlmd->endpoint is ERR_PTR(-ENXIO); cxl_memdev_unregister() never NULLs cxlmd, so probe/AER/unbind leaves a dangling pointer while the notifier stays registered. Any later MCE with a usable address hits the UAF/NULL deref deterministically; EINJ/mce-inject also make the trigger attacker-controlled.
PR:LPrivileges required, low
cxl_handle_mce() has no capability check; cxl_pci probe installs the notifier automatically. CXL Type-3 capacity is ordinary System RAM any unprivileged local user can allocate, and async probe, cxl_mem not-yet-bound (ERR_PTR endpoint), or AER-driven detach create the bad lifetime state without init-namespace root, matching CVE-2024-26762.
UI:NUser interaction, none
MCE decode and CXL probe/unbind/AER recovery run in kernel workqueues with no victim mount, file open, or other interactive action; the attacker (or automatic error handling) drives the path themselves.
S:UScope, unchanged
The NULL dereference and use-after-free corrupt host kernel CXL/MCE state in the same OS security authority. This is standard kernel memory corruption/privilege escalation, not a KVM/IOMMU/sandbox boundary crossing.
C:HConfidentiality impact, high
The handler dereferences mds->cxlds.cxlmd and cxlmd->endpoint (and walks endpoint->regions) after those objects can be freed, a use-after-free that can be reclaimed to expose kernel memory. CVSS kernel guidance treats UAF as High confidentiality impact.
I:HIntegrity impact, high
The same UAF is groomable for an arbitrary write. The callback also computes spa_alias from the dangling objects and calls memory_failure()/set_mce_nospec() on that PFN, so a sprayed cxl_memdev/region can offline attacker-chosen pages and hijack control flow.
A:HAvailability impact, high
NULL deref of cxlmd, walking an ERR_PTR endpoint, and UAF of freed memdev/port/region objects cause a kernel oops/panic in the MCE workqueue; UAFs also crash even when not fully exploited.