NVIDIA Resiliency Extension: Predictable log-file names in the log-aggregation path let an attacker pre-create
Impact
Predictable log-file names in the log-aggregation path let an attacker pre-create or hijack the target file, reaching privilege escalation and code execution. Scored 8.4 with no privileges required. The Resiliency Extension is what restarts failed large training jobs, so it runs with broad access across the job's nodes.
Who can reach it
Local, no privileges required, no user interaction. Any account on a node participating in a resilient training job.
What to do
Update the Resiliency Extension per bulletin 5746 and rebuild training images. Cost: image rebuild and job restart; no host driver or firmware change.
References
Related entries
- NVIDIA Resiliency Extension: A race condition in the checkpointing core reaches information disclosure, data tamperingCVE-2025-33235 · NVIDIA Resiliency ExtensionHigh
- TensorRT-LLM: RCE via unsafe deserializationCVE-2026-24233 · TensorRT-LLMHigh
- DGX H100 BMC: Privesc + code execution (web UI input validation)CVE-2023-25533 · DGX H100 BMCHigh
- DGX H100 BMC (REST): Code execution + privescCVE-2023-31009 · DGX H100 BMC (REST)High
- Container Toolkit: Container escape to host filesystem (bypass of the CVE-2024-0132 fix)CVE-2025-23359 · Container ToolkitHigh
- Jetson Xavier / Orin: Authentication bypass in a network serviceCVE-2026-24148 · Jetson Xavier / OrinHigh
This entry is curated: imported from vendor advisories with machine assistance, not yet individually verified. Confirm against your vendor's advisory before acting, and report anything wrong.