Occlum: Secure and Efficient Multitasking Inside a Single Enclave of Intel SGX
Not on the map. Why?
Abstract (arXiv)
Intel Software Guard Extensions (SGX) enables user-level code to create private memory regions called enclaves, whose code and data are protected by the CPU from software and hardware attacks outside the enclaves. Recent work introduces library operating systems (LibOSes) to SGX so that legacy applications can run inside enclaves with few or even no modifications. As virtually any non-trivial application demands multiple processes, it is essential for LibOSes to support multitasking. However, none of the existing SGX LibOSes support multitasking both securely and efficiently. This paper presents Occlum, a system that enables secure and efficient multitasking on SGX. We implement the LibOS processes as SFI-Isolated Processes (SIPs). SFI is a software instrumentation technique for sandboxing untrusted modules (called domains). We design a novel SFI scheme named MPX-based, Multi-Domain SFI (MMDSFI) and leverage MMDSFI to enforce the isolation of SIPs. We also design an independent verifier to ensure the security guarantees of MMDSFI. With SIPs safely sharing the single address space of an enclave, the LibOS can implement multitasking efficiently. The Occlum LibOS outperforms the state-of-the-art SGX LibOS on multitasking-heavy workloads by up to 6,600X on micro-benchmarks and up to 500X on application benchmarks.
Key facts. Cited by 11 papers in the map (data through 7 Oct 2026). First posted on arXiv 21 Jan 2020.
Builds on
No reference list yet.
Built on by
Citing papers per year: 2021: 2; 2022: 1; 2023: 1; 2024: 2; 2025: 2; 2026: 3.
11 papers.
- 2023No Privacy Left Outside: On the (In-)Security of TEE-Shielded DNN Partition for On-Device MLcited by 21
- 2021Citadel: Protecting Data Privacy and Model Confidentiality for Collaborative Learning with SGXcited by 14
- 2022Machine Learning with Confidential Computing: A Systematization of Knowledgecited by 13
- 2024TransLinkGuard: Safeguarding Transformer Models Against Model Stealing in Edge Deploymentcited by 11
- 2025Confidential LLM Inference: Performance and Cost Across CPU and GPU TEEscited by 5
- 2024TEESlice: Protecting Sensitive Neural Network Models in Trusted Execution Environments When Attackers have Pre-Trained Modelscited by 3
- 2021Plinius: Secure and Persistent Machine Learning Model Trainingcited by 2
- 2025Towards Confidential and Efficient LLM Inference with Dual Privacy Protectioncited by 1
- 2026Opal: Private Memory for Personal AIcited by 1
- 2026Implement Kubernetes Pod-Level Remote Attestation for Confidential Workloads on dstackcited by 0
- 2026TEE-X: TEE-aware Acceleration Framework for Large Vision Models at the Edgecited by 0
- 2026TEE-X: TEE-aware Acceleration Framework for Large Vision Models at the Edge24 Aug 2026
- 2026Implement Kubernetes Pod-Level Remote Attestation for Confidential Workloads on dstack2 Jun 2026
- 2026Opal: Private Memory for Personal AI2 Apr 2026
- 2025Confidential LLM Inference: Performance and Cost Across CPU and GPU TEEs23 Sep 2025
- 2025Towards Confidential and Efficient LLM Inference with Dual Privacy Protection11 Sep 2025
- 2024TEESlice: Protecting Sensitive Neural Network Models in Trusted Execution Environments When Attackers have Pre-Trained Models15 Nov 2024
- 2024TransLinkGuard: Safeguarding Transformer Models Against Model Stealing in Edge Deployment17 Apr 2024
- 2023No Privacy Left Outside: On the (In-)Security of TEE-Shielded DNN Partition for On-Device ML11 Oct 2023
- 2022Machine Learning with Confidential Computing: A Systematization of Knowledge22 Aug 2022
- 2021Citadel: Protecting Data Privacy and Model Confidentiality for Collaborative Learning with SGX4 May 2021
- 2021Plinius: Secure and Persistent Machine Learning Model Training7 Apr 2021