Intel TDX MigTD CVE-2026-49036

Exploitation requires the attacker to already be privileged system software - a hypervisor or Ring-0 operator (PR:H) - with LOCAL access to the host running the Intel TDX Migration TD (AV:L); there is no remote or network path and no user interaction (UI:N). The specific prerequisite is the MigTD shared-memory interface being in use (i.e. a TDX live-migration deployment relying on MigTD), and the attacker must win the time-of-check-to-time-of-use race window (AT:P, attack requirements present) to corrupt the shared data between validation and use. Limiting factors: the attacker must hold privileged local control of the untrusted host, the flaw yields no confidentiality disclosure and does not propagate to subsequent systems (SC/SI/SA:N), and it is only relevant where MigTD is actually deployed to mediate TD migration against an untrusted hypervisor.

This is a genuine but moderate-priority issue rather than a critical emergency. The CVSS 4.0 vector (AV:L/AC:L/AT:P/PR:H/UI:N, base 5.7) establishes that exploitation requires LOCAL access, an already-PRIVILEGED actor (PR:H), and the presence of specific attack requirements (AT:P) - the input explicitly describes a 'System software adversary with a privileged user,' i.e. a malicious or compromised hypervisor/Ring-0 operator. Impact is scoped to integrity (high) and availability (high) of the vulnerable MigTD with NO confidentiality impact and NO subsequent-system impact (SC/SI/SA:N), so this does not break out of the migration domain into tenant secrets. No EPSS score, no CISA KEV listing, and no PoC reference are present in the supplied data, so there is no signal of active or probable exploitation - consistent with the vuln.today decision label of 'this_month' (attend to it in the normal patch cycle, not as an emergency). The one tension worth noting: the attacker is already a privileged Ring-0/hypervisor-level adversary, which in a confidential-computing threat model is exactly the party TDX is designed to defend against, so the real-world severity depends heavily on whether the deployment relies on MigTD to protect live TD migration against an untrusted host. EPSS and PoC status are MISSING rather than confirmed-negative.

A malicious or compromised hypervisor (Ring-0 system software) with privileged access targets the shared-memory region used by the Intel TDX Migration TD, and races its own access against MigTD's check-then-use sequence (a TOCTOU window) to alter the shared data between validation and consumption. Winning this race corrupts MigTD state, degrading or denying the migration service and undermining the integrity of the migration process. No public exploit code is identified in the supplied data; exploitation requires the attacker to already hold privileged local control of the host.