Skip to content

ENCLAVE-03: Multi-enclave co-tenancy

Property Value
Severity medium
Platform AWS Nitro Enclaves
Lifecycle stage runtime
Mode Active: runs fixed read-only commands on the host. Skipped with --read-only.

What it checks

It passes when:

  • … enclave running. No co-tenancy risk.

Example failures:

  • … enclaves running on the same host: …. Shared hugepage pool and CPU contention create side-channel and resource starvation risks.
  • 2 enclaves running on the same host: i-0abc123-enc-0001, i-0abc123-enc-0002. Shared hugepage pool and CPU contention create side-channel and resource starvation risks.

How to fix

  • Run exactly one enclave per host. Separate workloads onto dedicated EC2 instances to eliminate resource contention and co-tenancy side-channels.

After the fix, scan again: the check should report PASS.

When it is not assessed

SKIP:

  • No instance ID specified; skipping check.

ERROR:

  • SSM command did not complete within timeout.
  • nitro-cli not available or no enclaves found.
  • Could not send SSM command: …
  • Could not parse nitro-cli output.

Scan inputs

The untrust scan options this check reads:

Option Description
--instance-id EC2 instance ID running the Nitro Enclave.
--region AWS region.

Permissions

The cloud API calls this check makes, read from the scanner's source:

Cloud API call Permission Access
AWS ssm.get_command_invocation ssm:GetCommandInvocation read
AWS ssm.send_command ssm:SendCommand run command

Every AWS scan also calls sts:GetCallerIdentity. See Required permissions for a complete policy.

Evidence recorded

Keys in the finding's evidence: running_enclaves, total_count

Generated from untrust 1.3.0 (commit 7ef462d9a0c3).