Описание
In the Linux kernel, the following vulnerability has been resolved:
KVM: x86: Don't (re)check L1 intercepts when completing userspace I/O
When completing emulation of instruction that generated a userspace exit
for I/O, don't recheck L1 intercepts as KVM has already finished that
phase of instruction execution, i.e. has already committed to allowing L2
to perform I/O. If L1 (or host userspace) modifies the I/O permission
bitmaps during the exit to userspace, KVM will treat the access as being
intercepted despite already having emulated the I/O access.
Pivot on EMULTYPE_NO_DECODE to detect that KVM is completing emulation.
Of the three users of EMULTYPE_NO_DECODE, only complete_emulated_io() (the
intended "recipient") can reach the code in question. gp_interception()'s
use is mutually exclusive with is_guest_mode(), and
complete_emulated_insn_gp() unconditionally pairs EMULTYPE_NO_DECODE with
EMULTYPE_SKIP.
The bad behavior was detected by a syzkaller program that toggles port I/O
interception during the userspace I/O exit, ultimately resulting in a WARN
on vcpu->arch.pio.count being non-zero due to KVM no completing emulation
of the I/O instruction.
WARNING: CPU: 23 PID: 1083 at arch/x86/kvm/x86.c:8039 emulator_pio_in_out+0x154/0x170 [kvm]
Modules linked in: kvm_intel kvm irqbypass
CPU: 23 UID: 1000 PID: 1083 Comm: repro Not tainted 6.16.0-rc5-c1610d2d66b1-next-vm #74 NONE
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
RIP: 0010:emulator_pio_in_out+0x154/0x170 [kvm]
PKRU: 55555554
Call Trace:
kvm_fast_pio+0xd6/0x1d0 [kvm]
vmx_handle_exit+0x149/0x610 [kvm_intel]
kvm_arch_vcpu_ioctl_run+0xda8/0x1ac0 [kvm]
kvm_vcpu_ioctl+0x244/0x8c0 [kvm]
__x64_sys_ioctl+0x8a/0xd0
do_syscall_64+0x5d/0xc60
entry_SYSCALL_64_after_hwframe+0x4b/0x53
A flaw was found in the Kernel-based Virtual Machine (KVM) for x86. This vulnerability occurs when KVM incorrectly re-checks Level 1 (L1) intercepts during the completion of userspace I/O emulation for a Level 2 (L2) instruction. If the L1 or host userspace modifies I/O interception settings during this process, KVM may misinterpret already-emulated I/O as intercepted. This can lead to a kernel warning and potentially disrupt the guest virtual machine, resulting in a denial of service. Exploitation requires local access to /dev/kvm and the ability to perform nested virtualization or manipulate I/O bitmaps.
Отчет
KVM x86 was re-checking L1 instruction intercepts while completing a userspace I/O emulation for an L2 instruction. If L1 (or host userspace) modified I/O interception during the exit, KVM could wrongly treat the already-emulated I/O as intercepted and leave vcpu->arch.pio.count non-zero, triggering a WARN and potentially disrupting the guest. Practical exploitation requires local access to /dev/kvm and the ability to run nested virtualization or manipulate I/O bitmaps.
Меры по смягчению последствий
This vulnerability can be mitigated by disabling nested virtualization. Create a file in /etc/modprobe.d/ with a descriptive name, such as cve-2026-53359-mitigation.conf. In that file, disable nested virtualization for the kvm_intel and kvm_amd kernel modules:
Use lsmod to determine if either module is already loaded. If so, remove loaded modules with modprobe -r. To validate that nested virtualization is disabled, read the files /sys/module/kvm_intel/parameters/nested and /sys/module/kvm_amd/parameters/nested. If these modules are loaded, those files should read N to indicate that the feature is disabled.
In OpenShift 4, a MachineConfig can be utilized to create a modprobe configuration allowing control of the module on applicable nodes. See the following article for an example.
https://access.redhat.com/solutions/6979679
Затронутые пакеты
| Платформа | Пакет | Состояние | Рекомендация | Релиз |
|---|---|---|---|---|
| Red Hat Enterprise Linux 10 | kernel | Affected | ||
| Red Hat Enterprise Linux 6 | kernel | Not affected | ||
| Red Hat Enterprise Linux 7 | kernel | Affected | ||
| Red Hat Enterprise Linux 7 | kernel-rt | Affected | ||
| Red Hat Enterprise Linux 9 | kernel-rt | Affected | ||
| Red Hat Enterprise Linux for NVIDIA 26 | kernel | Not affected | ||
| Red Hat OpenShift Container Platform 4 | openshift/ose-rhel-coreos-9 | Affected | ||
| Red Hat Enterprise Linux 10.0 Extended Update Support | kernel | Fixed | RHSA-2026:49030 | 31.07.2026 |
| Red Hat Enterprise Linux 8 | kernel-rt | Fixed | RHSA-2026:45116 | 24.07.2026 |
| Red Hat Enterprise Linux 8 | kernel | Fixed | RHSA-2026:45115 | 24.07.2026 |
Показывать по
Дополнительная информация
Статус:
EPSS
5.5 Medium
CVSS3
Связанные уязвимости
In the Linux kernel, the following vulnerability has been resolved: KVM: x86: Don't (re)check L1 intercepts when completing userspace I/O When completing emulation of instruction that generated a userspace exit for I/O, don't recheck L1 intercepts as KVM has already finished that phase of instruction execution, i.e. has already committed to allowing L2 to perform I/O. If L1 (or host userspace) modifies the I/O permission bitmaps during the exit to userspace, KVM will treat the access as being intercepted despite already having emulated the I/O access. Pivot on EMULTYPE_NO_DECODE to detect that KVM is completing emulation. Of the three users of EMULTYPE_NO_DECODE, only complete_emulated_io() (the intended "recipient") can reach the code in question. gp_interception()'s use is mutually exclusive with is_guest_mode(), and complete_emulated_insn_gp() unconditionally pairs EMULTYPE_NO_DECODE with EMULTYPE_SKIP. The bad behavior was detected by a syzkaller program that toggles port I/O...
In the Linux kernel, the following vulnerability has been resolved: KVM: x86: Don't (re)check L1 intercepts when completing userspace I/O When completing emulation of instruction that generated a userspace exit for I/O, don't recheck L1 intercepts as KVM has already finished that phase of instruction execution, i.e. has already committed to allowing L2 to perform I/O. If L1 (or host userspace) modifies the I/O permission bitmaps during the exit to userspace, KVM will treat the access as being intercepted despite already having emulated the I/O access. Pivot on EMULTYPE_NO_DECODE to detect that KVM is completing emulation. Of the three users of EMULTYPE_NO_DECODE, only complete_emulated_io() (the intended "recipient") can reach the code in question. gp_interception()'s use is mutually exclusive with is_guest_mode(), and complete_emulated_insn_gp() unconditionally pairs EMULTYPE_NO_DECODE with EMULTYPE_SKIP. The bad behavior was detected by a syzkaller program that toggles port I/
KVM: x86: Don't (re)check L1 intercepts when completing userspace I/O
In the Linux kernel, the following vulnerability has been resolved: K ...
In the Linux kernel, the following vulnerability has been resolved: KVM: x86: Don't (re)check L1 intercepts when completing userspace I/O When completing emulation of instruction that generated a userspace exit for I/O, don't recheck L1 intercepts as KVM has already finished that phase of instruction execution, i.e. has already committed to allowing L2 to perform I/O. If L1 (or host userspace) modifies the I/O permission bitmaps during the exit to userspace, KVM will treat the access as being intercepted despite already having emulated the I/O access. Pivot on EMULTYPE_NO_DECODE to detect that KVM is completing emulation. Of the three users of EMULTYPE_NO_DECODE, only complete_emulated_io() (the intended "recipient") can reach the code in question. gp_interception()'s use is mutually exclusive with is_guest_mode(), and complete_emulated_insn_gp() unconditionally pairs EMULTYPE_NO_DECODE with EMULTYPE_SKIP. The bad behavior was detected by a syzkaller program that toggles port...
EPSS
5.5 Medium
CVSS3