{"dataType":"CVE_RECORD","dataVersion":"5.2","cveMetadata":{"cveId":"CVE-2026-63941","assignerOrgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","state":"PUBLISHED","assignerShortName":"Linux","dateReserved":"2026-07-19T07:54:57.022Z","datePublished":"2026-07-19T14:55:37.047Z","dateUpdated":"2026-08-05T12:37:28.804Z"},"containers":{"cna":{"providerMetadata":{"orgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","shortName":"Linux","dateUpdated":"2026-08-05T12:37:28.804Z"},"descriptions":[{"lang":"en","value":"In the Linux kernel, the following vulnerability has been resolved:\n\nKVM: arm64: Correctly cap ZCR_EL2 provided by a guest hypervisor\n\nZCR_EL2 can be updated by a VHE guest hypervisor either using ZCR_EL2\n(which traps) or ZCR_EL1 (which does not trap). KVM handles both in\ndifferent way:\n\n- on ZCR_EL2 trap, ZCR_EL2.LEN is immediately capped at the VM's own\n  VL limit. This has the potential to break existing SW that relies\n  on the full LEN field to be stateful.\n\n- on ZCR_EL1 access, we do absolutely nothing.\n\nOn restoring the SVE context for an L2 guest, we directly restore the\nguest hypervisor's view of ZCR_EL2 into the physical ZCR_EL2. If the\nguest's view of the register was updated using the ZCR_EL2 accessor,\nthe value has already been sanitised (with the caveat mentioned above).\n\nBut if the guest used ZCR_EL1, the raw value is written into the HW,\nand the L2 guest can now access VLs that it shouldn't.\n\nFix all the above by moving the VL capping to the restore points,\nensuring that:\n\n- the HW is always programmed with a capped value, irrespective of\n  the accessor being used,\n\n- the ZCR_EL2.LEN field is always completely stateful, irrespective\n  of the accessor being used.\n\nAdditionally, move ZCR_EL2 to be a sanitised register, ensuring that\nonly the LEN field is actually stateful. This requires some creative\nconstruction of the RES0 mask, as the sysreg generation script does\nnot yet generate RAZ/WI fields.\n\n[maz: rewrote commit message, tidy up access_zcr_el2()]"}],"metrics":[{"cvssV3_1":{"version":"3.1","vectorString":"CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H","baseScore":8.8,"baseSeverity":"HIGH"},"scenarios":[{"lang":"en","value":"AV:L - The bug is reached only through KVM nested-virtualization and SVE context-restore paths (e.g., KVM_RUN → guest entry → __hyp_sve_restore_guest / fpsimd_lazy_switch_to_guest), not via any network protocol or remote packet handler.\nAC:L - A malicious L1 guest hypervisor can reliably disable SVE traps, write ZCR_EL1 directly, and run an L2 guest that uses SVE; the attacker controls both the hypervisor configuration and the nested guest workload that triggers restore.\nPR:L - Exploitation requires code execution inside a VM with nested virtualization and SVE enabled (KVM_ARM_VCPU_HAS_EL2 + SVE); it is not triggerable by an unprivileged host user or an unauthenticated remote attacker without guest access.\nUI:N - No end-user or administrator action beyond the attacker operating their own nested guest environment is required to trigger the flawed ZCR_EL2 programming.\nS:C - Incorrect programming of physical ZCR_EL2 for an L2 guest breaks the KVM hypervisor isolation boundary and can corrupt host kernel memory (the vcpu SVE state buffer in host heap), enabling a nested-guest-to-host escape.\nC:H - Programming a ZCR_EL2 LEN above the vcpu's allocated sve_max_vl lets SVE save/restore operate beyond the kmalloc'd sve_state buffer, yielding out-of-bounds access that can be leveraged for arbitrary kernel memory disclosure.\nI:H - The same uncapped ZCR_EL2 value enables out-of-bounds SVE register save/restore into adjacent host kernel objects, providing a memory-corruption primitive that can be developed into arbitrary write or host code execution.\nA:H - The mismatch between hardware vector length and the allocated SVE state buffer can cause kernel oops/panic during SVE context save/restore, and repeated triggering can deny service to other VMs on the host."}]}],"affected":[{"product":"Linux","vendor":"Linux","defaultStatus":"unaffected","repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["arch/arm64/include/asm/kvm_host.h","arch/arm64/kvm/hyp/include/hyp/switch.h","arch/arm64/kvm/nested.c","arch/arm64/kvm/sys_regs.c"],"versions":[{"version":"b3d29a8230998b36afecf494b199211d26052785","lessThan":"742a9b5ccd8c46caf983ca90c94f855466968e34","status":"affected","versionType":"git"},{"version":"b3d29a8230998b36afecf494b199211d26052785","lessThan":"83726330748981372bde86ed5411d7b306612991","status":"affected","versionType":"git"}]},{"product":"Linux","vendor":"Linux","defaultStatus":"affected","repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["arch/arm64/include/asm/kvm_host.h","arch/arm64/kvm/hyp/include/hyp/switch.h","arch/arm64/kvm/nested.c","arch/arm64/kvm/sys_regs.c"],"versions":[{"version":"6.11","status":"affected"},{"version":"0","lessThan":"6.11","status":"unaffected","versionType":"semver"},{"version":"7.0.12","lessThanOrEqual":"7.0.*","status":"unaffected","versionType":"semver"},{"version":"7.1","lessThanOrEqual":"*","status":"unaffected","versionType":"original_commit_for_fix"}]}],"cpeApplicability":[{"nodes":[{"operator":"OR","negate":false,"cpeMatch":[{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.11","versionEndExcluding":"7.0.12"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.11","versionEndExcluding":"7.1"}]}]}],"references":[{"url":"https://git.kernel.org/stable/c/742a9b5ccd8c46caf983ca90c94f855466968e34"},{"url":"https://git.kernel.org/stable/c/83726330748981372bde86ed5411d7b306612991"}],"title":"KVM: arm64: Correctly cap ZCR_EL2 provided by a guest hypervisor","x_generator":{"engine":"bippy-1.2.0"}}}}