GPUCC-REATTEST-01: Attestation is continuous, not one-shot (post-attest drift is caught)¶
| Property | Value |
|---|---|
| Severity | high |
| Platform | NVIDIA GPU Confidential Computing |
| Lifecycle stage | attest |
| Trust boundary | attestation |
| Assurance | operator-declared |
| Mode | Passive: reads only the files and values you pass to the scan. |
What a failure enables¶
- An unverified GPU trusted for the life of the workload
What it checks¶
It passes when:
- Attestation is continuous and re-runs on GPU state change.
How to fix¶
- Re-attest the GPU periodically and on state-change events (CC-mode change, GPU reset, relaunch); gate continued key/data use on a fresh attestation, not a boot-time one.
After the fix, scan again: the check should report PASS.
When it is not assessed¶
SKIP:
- No --gpu-verifier-policy specified; skipping check.
ERROR:
- Could not read/parse verifier policy: …
Scan inputs¶
The untrust scan options this check reads:
| Option | Description |
|---|---|
--gpu-verifier-policy |
Verifier policy JSON: pinned RIM measurements, cert chain, revocation. |
Permissions¶
None. The check reads only the files or values you pass to the scan.
Related¶
- Runtime detection
DET-G7(NVIDIA GPU CC): GPU attestation went stale
Generated from untrust 1.3.0 (commit 7ef462d9a0c3).