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

Look up another CVE · open CVE-2026-89914 in the interactive tracker

CVE-2026-89914

KVM: arm64: Sign-extend VA for range-based TLBI invalidation

9.3 EPSS 0.2% AV:L

Published 2026-09-16 (33 days ago) by the Linux kernel CNA. Subsystem arch/arm64. Debian urgency: unimportant.

What to do, per Debian release

Debian's own status for CVE-2026-89914 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

When the decode_range_tlbi() helper was moved to be used for S1 TLBIs, the required sign extension was omitted. Add it. As a result, special care must be taken to not overflow PA bits when this is used for S2 invalidation.

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.

arch/arm64/include/asm/kvm_nested.h
arch/arm64/kvm/sys_regs.c

What this is
KERNEL VIRTUAL MACHINE FOR ARM64 (KVM/arm64)MAINTAINERS section matched on arch/arm64/include/asm/kvm* · linux-arm-kernel@lists.infradead.org · Maintained
Kernel module
kvm

Check it on the machine

grep -w CONFIG_KVM /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.

lsmod | grep -w kvm

Prints a line if that module is loaded right now, and nothing if it is not. One that is not loaded can still be loaded later by anything that needs it.

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.

  • bookwormBnot built on arm64/rt/arm64, armhf/armmp, armhf/armmp-lpae, armhf/rt/armmp
  • trixieBnot built on armel/rpi, armhf/armmp, armhf/armmp-lpae, armhf/rt/armmp
  • forky, sidBnot built on armhf/armmp, armhf/armmp-lpae
B
built in on some, a module on others, absent on the rest
Build switch
CONFIG_KVM

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
2026-09-16 (33 days ago)source
CVSS
9.3 critical
CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:Hsource recompute it
Attack vector
local (AV:L)source
EPSS
0.19%, higher than 8.4% of all CVEssource
CISA KEV
not listedthe KEV catalogue
Debian urgency
unimportantsource
Subsystem
arch/arm64

Upstream fixes

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

IntroducedFixed inCommit
6.166.18.5172bce82c4171
6.167.2.53feb83918e30
6.167.3239347008564

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
A nested arm64 L1 hypervisor reaches decode_range_tlbi() by executing a range TLBI (RVAE2/RVAE1/RIPAS2 and variants) that traps via ESR_ELx_EC_SYS64 into kvm_handle_sys_reg during KVM_RUN. This is local guest execution, not a network, adjacent-radio, or physical-device path.
AC:LAttack complexity, low
An L1 guest fully controls the TLBI Xt encoding (BASEADDR/TG/NUM/SCALE) and can deterministically hit the missing sign-extension for canonical high VAs. Nested virt and FEAT_TLBIRANGE on arm64 cloud hosts are attacker-selectable deployments, not an uncontrollable race or rare kernel config.
PR:NPrivileges required, none
Exploitation requires no host root or init-namespace capabilities beyond running hypervisor code in a tenant KVM guest on an arm64 host with nested virtualization enabled; the attacker operates entirely from within their assigned VM, consistent with other arm64 KVM nested-virt CVE scoring.
UI:NUser interaction, none
No victim user or administrator action is required beyond the attacker operating their own nested-virtualization workload; VNCR_EL2 setup, S1 page-table updates, and EL2 range TLBIs are fully attacker-driven from the guest.
S:CScope, changed
Missed high-VA VNCR invalidation leaves the host per-CPU PAGE_KERNEL fixmap covering a stale guest HPA while KVM emulates nested EL2 register state, crossing the guest-to-hypervisor security boundary (Changed per CNA KVM guidance).
C:HConfidentiality impact, high
Without sign_extend64(base, 48), range TLBIs for canonical high GVAs never match vt->gva, so the VNCR pseudo-TLB retains the old HPA; host KVM then reads that physical page through the kernel fixmap during sysreg emulation, disclosing whatever now backs that HPA.
I:HIntegrity impact, high
kvm_map_l1_vncr() installs a writable PAGE_KERNEL fixmap of the stale HPA when S1 write permissions allow, so nested EL2 register writes land in pages the guest already rebound, giving a write primitive against host-managed physical memory.
A:HAvailability impact, high
Accessing an unmapped or reused HPA through the VNCR fixmap can oops or panic the host kernel, and corrupting emulated EL2 state or leaving stale nested translations can hang or crash the hypervisor and co-resident VMs; the attacker can retrigger the range TLBI.