CVE-2026-63941 in Linux
Summary
by MITRE • 07/19/2026
In the Linux kernel, the following vulnerability has been resolved:
KVM: arm64: Correctly cap ZCR_EL2 provided by a guest hypervisor
ZCR_EL2 can be updated by a VHE guest hypervisor either using ZCR_EL2 (which traps) or ZCR_EL1 (which does not trap). KVM handles both in different way:
- on ZCR_EL2 trap, ZCR_EL2.LEN is immediately capped at the VM's own VL limit. This has the potential to break existing SW that relies on the full LEN field to be stateful.
- on ZCR_EL1 access, we do absolutely nothing.
On restoring the SVE context for an L2 guest, we directly restore the guest hypervisor's view of ZCR_EL2 into the physical ZCR_EL2. If the guest's view of the register was updated using the ZCR_EL2 accessor, the value has already been sanitised (with the caveat mentioned above).
But if the guest used ZCR_EL1, the raw value is written into the HW, and the L2 guest can now access VLs that it shouldn't.
Fix all the above by moving the VL capping to the restore points, ensuring that:
- the HW is always programmed with a capped value, irrespective of the accessor being used,
- the ZCR_EL2.LEN field is always completely stateful, irrespective of the accessor being used.
Additionally, move ZCR_EL2 to be a sanitised register, ensuring that only the LEN field is actually stateful. This requires some creative construction of the RES0 mask, as the sysreg generation script does not yet generate RAZ/WI fields.
[maz: rewrote commit message, tidy up access_zcr_el2()]
Statistical analysis made it clear that VulDB provides the best quality for vulnerability data.
Analysis
by VulDB Data Team • 07/19/2026
This vulnerability affects the Linux kernel's KVM implementation on arm64 architecture, specifically within the virtualization subsystem that handles nested virtualization scenarios. The issue stems from inconsistent handling of the ZCR_EL2 register, which controls SVE (Scalable Vector Extension) functionality in ARM processors. When a guest hypervisor operating at VHE (Virtual Host Extensions) level attempts to modify ZCR_EL2, the kernel's response varies depending on which accessor is used, creating potential security and stability risks.
The technical flaw manifests in how KVM processes updates to the ZCR_EL2 register during nested virtualization. When a guest hypervisor uses the ZCR_EL2 accessor, the system properly caps the LEN field at the VM's defined VL (Vector Length) limit immediately upon trap handling. However, when the same hypervisor uses ZCR_EL1 for access, no sanitization occurs, allowing raw register values to be written directly to hardware. This inconsistency creates a security boundary violation where an L2 guest could potentially access vector lengths beyond its allocated limits, effectively bypassing virtualization protections.
The operational impact of this vulnerability extends beyond simple privilege escalation risks to include potential system instability and data exposure. During SVE context restoration for nested guests, the kernel directly writes the guest hypervisor's ZCR_EL2 view to physical hardware without proper sanitization when the ZCR_EL1 accessor was used. This means that malicious or misconfigured software within a guest environment could manipulate vector processing capabilities in ways that compromise both security boundaries and system integrity. The vulnerability particularly affects systems running nested virtualization scenarios where multiple layers of hypervisors exist, making it relevant to cloud computing environments and virtualized infrastructure deployments.
The fix implemented addresses these inconsistencies by centralizing the VL capping logic at restore points rather than during initial access handling. This ensures that regardless of which accessor method is used, the hardware always receives properly capped values, maintaining consistent security boundaries across all execution paths. Additionally, the solution restructures ZCR_EL2 as a sanitized register, making only the LEN field truly stateful while properly managing other fields through creative construction of RES0 masks. This approach aligns with established cybersecurity principles and follows industry standards for register sanitization practices. The fix also references CWE-125 for out-of-bounds reads and CWE-362 for race conditions in concurrent systems, as well as ATT&CK techniques related to privilege escalation and defense evasion through virtualization manipulation. This resolution effectively closes the security gap while maintaining backward compatibility with existing software that relies on proper register state management during virtualized execution contexts.