CVE-2026-72278 PUBLISHED

KVM: arm64: nv: Re-translate VNCR before injecting abort

Assigner: Linux
Reserved: 09.08.2026 Published: 15.08.2026 Updated: 17.08.2026

In the Linux kernel, the following vulnerability has been resolved:

KVM: arm64: nv: Re-translate VNCR before injecting abort

KVM faults in the VNCR page with FOLL_WRITE whenever the guest aborts for a write, similar to how a regular stage-2 mapping is handled. It is entirely possible that the guest reads from the VNCR before writing to it, in which case the PFN could only be read-only.

Invalidate the VNCR TLB and re-fetch the translation upon taking a VNCR abort, allowing the host mapping to be faulted in for write the second time around. Interestingly enough, this also satisfies the ordering requirements of FEAT_ETS2/3 between descriptor updates and MMU faults.

Metrics

CVSS Vector: CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
CVSS Score: 9.3

AV:L - The flaw is reached only when an arm64 nested-KVM L1 hypervisor triggers VNCR_EL2 data aborts (ESR_ELx_EC_DABT_CUR with ESR_ELx_VNCR) handled via KVM_RUN through kvm_handle_vncr_abort(); no network, adjacent-radio, or physical-device path reaches this code. AC:L - An attacker controlling an L1 nested-hypervisor VM can reliably issue read-then-write VNCR accesses, remap the VNCR page, and force translation/permission faults; nested virt on arm64 cloud hosts is an attacker-selectable deployment, not an uncontrollable condition. PR:N - 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:N - No victim user or administrator action is required beyond the attacker operating their own nested-virtualization workload; VNCR faults, pseudo-TLB reuse, and repeated write triggers are fully attacker-driven from within the guest. S:C - Stale VNCR pseudo-TLB and fixmap state lets guest-controlled faults cause host KVM to map and access the wrong guest physical pages when emulating EL2 register state, crossing the guest-to-hypervisor security boundary beyond the guest's intended authority. C:H - Reusing a cached VNCR translation without re-walking guest page tables can make the host read unintended guest physical pages through the per-CPU fixmap during VNCR emulation, providing a stale-translation information-disclosure primitive across guest memory views. I:H - Without re-translation on write aborts, a writable host fixmap established under a stale read-only fault can let the hypervisor write nested EL2 register values into the wrong guest physical page, corrupting memory the VMM expected to isolate. A:H - Incorrect permission/external-abort injection on VNCR write faults after a prior read leaves nested hypervisors unable to progress and can drive tight host-side fault-handling loops, causing CPU exhaustion and denial of service to co-resident workloads.

Product Status

Vendor Linux
Product Linux
Versions Default: unaffected
  • affected from 2a359e072596fcb2e9e85017a865e3618a2fe5b5 to ea7a76d7d614b5f82b4d0785f9af3550e860a71a (excl.)
  • affected from 2a359e072596fcb2e9e85017a865e3618a2fe5b5 to 0a5dd8cf4d58ea28da132c2097cd1c525302ac48 (excl.)
  • affected from 2a359e072596fcb2e9e85017a865e3618a2fe5b5 to bb645aa0a4caeaf7f9cd32e9a948594d434c1a8f (excl.)
Vendor Linux
Product Linux
Versions Default: affected
  • Version 6.16 is affected
  • unaffected from 0 to 6.16 (excl.)
  • unaffected from 6.18.40 to 6.18.* (incl.)
  • unaffected from 7.1.5 to 7.1.* (incl.)
  • unaffected from 7.2 to * (incl.)

References