The Linux kernel KVM nested virtualization code did not properly release vmcs12 pages when a nested VM-Enter operation failed due to an invalid guest state. Specifically, when KVM synthesizes a nested VM-Exit during emulation of VMLAUNCH or VMRESUME, the code path that handles invalid guest states bypassed the nested_vmx_vmexit() API, which is responsible for releasing vmcs12 pages. This omission caused pinned pages and/or mappings to leak if the L1 hypervisor retried the VM-Enter operation. The issue was fixed by ensuring all vmcs12 pages are put in this failure path. The problem does not occur in non-VM-Enter scenarios because vmcs12 pages are only pinned if L2 is active, guaranteeing a full VM-Exit before retrying.

The vulnerability could cause resource leaks of pinned memory pages or mappings in the KVM nested virtualization environment when nested VM-Enter fails due to invalid guest state. This may lead to increased memory usage and potential stability or performance degradation on the host running nested virtual machines. There is no indication of direct code execution or privilege escalation from this issue.

A fix has been implemented in the Linux kernel to ensure vmcs12 pages are properly released when nested VM-Enter fails due to invalid guest state. Users should update to the fixed kernel version once available. Patch status is not explicitly confirmed in the provided data; therefore, check the official Linux kernel advisories or vendor updates for the exact fixed versions and apply them accordingly.