CVE-2026-72296

The CVE-2026-72296 entry describes a Linux kernel vulnerability in the net/ife path where ife_decode() could return after pulling only the outer IFE header and metadata, causing eth_type_trans() to see an inner Ethernet header that may be absent in the linear data area. This could lead to a crash...

CVE-2026-72297

CVE-2026-72297 : Linux kernel vulnerability in ATM QoS validation. The issue arises when QoS is accepted and larger traffic_class values are not rejected, leaving vcc->qos with invalid values. This can lead to an out-of-bounds read in /proc/net/atm/pvc when indexing class_name[] with vcc->q...

CVE-2026-72295

Technical details (affected product/version, root cause, impact, and remediation) are not publicly available in the provided documents. Monitor for updates from official advisories.

CVE-2026-72294

The CVE-2026-72294 case affects the Linux kernel on LoongArch under KVM. The vulnerable component is kvm_vcpu_ioctl_interrupt() where irq validity checking was previously missing, allowing potential misvalidation when called from userspace. The root cause is insufficient validation of IRQs in the...

CVE-2026-72293

In the Linux kernel CVE-2026-72293, the KVM s390 vsie path is affected: missing radix_tree_preload() in _gaccess_shadow_fault() could stall forward progress. The core of _gaccess_shadow_fault() has been split into ___gaccess_shadow_fault() to simplify locking. The fix adds the missing radix_tree_...

CVE-2026-72292

The CVE-2026-72292 issue in the Linux kernel affects KVM on s390 where kvm_s390_get_cmma_bits() allocates its output with vmalloc(), which does not zero pages. In the non-peek migration path, dat_get_cmma() reports a byte span from the first to the last dirty page, but __dat_get_cmma_pte() only w...

CVE-2026-72291

CVE-2026-72291 : In the Linux kernel (KVM for s390), a fix was applied for an unlikely race in try_get_locked_pte() that could occur if puds or pmds are unmapped between p?dp_get() and p?d_offset() calls. The issue affects the page-table handling path and is mitigated by the update; affected comp...

CVE-2026-72290

CVE-2026-72290 affects the Linux kernel KVM when running on s390. The issue is a GISC (guest ISC) refcount leak that occurs because kvm_s390_gisc_register() registers the GISC before pinning guest interrupt forwarding pages and allocating the AISB bit; on error, the unwind path did not unregister...

CVE-2026-72288

The CVE-2026-72288 entry concerns the Linux kernel KVM arm64 vgic race between interrupt affinity changes and LPI disabling. Under a specific sequence—LPI-I pending in vcpu-B’s AP list; vcpu-A disables LPIs via vcpu-B’s RD; vcpu-C moves I from B to C—the vgic_prune_ap_list() path can drop irq and...

CVE-2026-72289

CVE-2026-72289 affects the Linux kernel (KVM on arm64) where vgic interrupt migration logic can mis-handle pending/active interrupts during lock-protected migrations. Specifically, vgic_prune_ap_list() drops ap_list_lock and irq_lock and rechecks only affinity, which can pass for an interrupt tha...

CVE-2026-72287

The CVE-2026-72287 entry covers a Linux kernel KVM nVMX issue where the vTPR vs. TPR Threshold consistency check was moved into normal checks. Affected component: KVM’s nested VMX path in the Linux kernel (vmcs12.tpr_threshold vs vTPR) and the handling of GUEST_CR3 when EPT is disabled. Reported ...

CVE-2026-72286

The CVE concerns the Linux kernel KVM SEV SNP intra-host migration/mirroring. The vulnerability arises because the intra-host migration/mirroring feature is not fully implemented for SEV-SNP VMs. The migration requires SNP-specific state (guest_req_mutex, guest_req_buf, guest_resp_buf) to be tran...

CVE-2026-72285

CVE-2026-72285 – Linux kernel KVM/TDX CPUID count race : The vulnerability occurs when userspace can change cpuid.nent between the initial read and the allocation of the kvm_tdx_init_vm copy, causing a mismatch in the number of CPUID entries that are copied and later processed. The fix enforces t...

CVE-2026-72284

CVE-2026-72284 affects the Linux kernel KVM on x86, where a logic error in arch/x86/kvm/lapic.c (line ~3338) caused PV EOIs to be ignored if the vCPU later disables PV EOIs, potentially triggering a kernel oops (invalid opcode) during KVM vCPU ioctls. The vulnerability is locally exploitable as p...

CVE-2026-72283

CVE-2026-72283 concerns the Linux kernel's KVM/x86 path. The issue: when updating the IRTE for bypass fails, irqfd->producer is not cleared, creating a potential use-after-free if the irqfd is reachable via KVM routing while the producer is freed separately (e.g., VFIO PCI, where the producer ...

CVE-2026-72282

The CVE-2026-72282 entry concerns the Linux kernel KVM path: kvm_io_bus_get_dev() now requires the caller to hold the srcu lock, moving locking responsibilities to the caller. The function returns a device matched only by address, which can lead to a lifetime issue if the device’s type isn’t as e...

CVE-2026-72281

The CVE-2026-72281 entry concerns the Linux kernel KVM arm64/pKVM reclaim against the VM memory manager. The issue arose because protected guest faults could pin long-term resources to the VM’s mm (memory descriptor), and teardown could execute later from file release with current->mm potentia...

CVE-2026-72280

CVE-2026-72280 concerns the Linux kernel KVM on arm64. The vulnerability arises from a logic issue where a guest could write to the ZCR_EL2 sysreg alias in a nested context if FEAT_NV2 is advertised to the L1 hypervisor. The fix is to drop the bogus WARN that could lead to a hyp_panic once hyp ve...

CVE-2026-72279

The CVE-2026-72279 entry concerns the Linux kernel KVM arm64 implementation for the L1 VNCR mapping. The fix ensures that the host stage-1 mapping respects the read-only status of the guest PFN by caching PFN writability in the VNCR TLB and using this to constrain fixmap permissions. If a guest t...

CVE-2026-72278

In CVE-2026-72278, the Linux kernel KVM for arm64 mitigates a VNCR abort handling issue by invalidating the VNCR TLB and re-fetching the translation when a VNCR abort occurs. This allows the host mapping to be faulted in for write on the second attempt, addressing a scenario where a guest could r...