@inproceedings{2027cxleak,title={CXLeak: Characterizing and Exploiting Contention in Compute Express Link (CXL) Memory},author={Li, Luyi and Park, Jangseon and Jamadagni, Nagendra Kumar and Wang, Zixuan and Zhao, Jishen and Rosing, Tajana and Tullsen, Dean},booktitle={IEEE International Symposium on High-Performance Computer Architecture (to appear)},year={2027},note={* Co-first authors: Luyi Li and Jangseon Park.},}
HPCA 2027
CXLens: From System Characterization to Validated Simulation of CXL-based Disaggregated Memory
Jangseon Park, Luyi Li, Zixuan Wang, Junhee Park, Seunghak Lee, Jinin So, Jongryool Kim, Younghoon Min, Jishen Zhao, Dean Tullsen, and Tajana Rosing
In IEEE International Symposium on High-Performance Computer Architecture (to appear), 2027
@inproceedings{2027cxlens,title={CXLens: From System Characterization to Validated Simulation of CXL-based Disaggregated Memory},author={Park, Jangseon and Li, Luyi and Wang, Zixuan and Park, Junhee and Lee, Seunghak and So, Jinin and Kim, Jongryool and Min, Younghoon and Zhao, Jishen and Tullsen, Dean and Rosing, Tajana},booktitle={IEEE International Symposium on High-Performance Computer Architecture (to appear)},year={2027},}
2026
In submission
DPUKV: DPU-Managed KV Cache Offloading for LLM Serving
Luyi Li, Nachuan Wang, Zhiye Song, Misun Park, Peng Xu, Wei Tang, Liguang Xie, Wu Xiang, Nam Sung Kim, and Dean Tullsen
@unpublished{2026dpukv,title={DPUKV: DPU-Managed KV Cache Offloading for LLM Serving},author={Li, Luyi and Wang, Nachuan and Song, Zhiye and Park, Misun and Xu, Peng and Tang, Wei and Xie, Liguang and Xiang, Wu and Kim, Nam Sung and Tullsen, Dean},year={2026},note={Submitted to MLSys 2027.},}
2024
arXiv
The Hitchhiker’s Guide to Programming and Optimizing Cache Coherent Heterogeneous Systems: CXL, NVLink-C2C, and AMD Infinity Fabric
Zixuan Wang, Suyash Mahar, Luyi Li, Jangseon Park, Jinpyo Kim, Theodore Michailidis, Yue Pan, Mingyao Shen, Tajana Rosing, Dean Tullsen, Steven Swanson, and Jishen Zhao
@article{2024hitchhiker,title={The Hitchhiker's Guide to Programming and Optimizing Cache Coherent Heterogeneous Systems: CXL, NVLink-C2C, and AMD Infinity Fabric},author={Wang, Zixuan and Mahar, Suyash and Li, Luyi and Park, Jangseon and Kim, Jinpyo and Michailidis, Theodore and Pan, Yue and Shen, Mingyao and Rosing, Tajana and Tullsen, Dean and Swanson, Steven and Zhao, Jishen},journal={arXiv:2411.02814},year={2024},}
USENIX Security 2024 — Distinguished Paper Award. IEEE Micro Top Picks 2025 — Honorable Mention.
This paper introduces novel high-precision Branch Target Injection (BTI) attacks, leveraging the intricate structures of the Indirect Branch Predictor (IBP) and the Branch Target Buffer (BTB) in high-end Intel CPUs. It presents, for the first time, a comprehensive picture of the IBP and the BTB within the most recent Intel processors, revealing their size, structure, and the precise functions governing index and tag hashing. Additionally, this study reveals new details into the inner workings of Intel’s hardware defenses, such as IBPB, IBRS, and STIBP, including previously unknown holes in their coverage. Leveraging insights from reverse engineering efforts, this research develops highly precise Branch Target Injection (BTI) attacks to breach security boundaries across diverse scenarios, including cross-process and cross-privilege scenarios and uses the IBP and the BTB to break Address Space Layout Randomization (ASLR).
@inproceedings{2024usenix,title={Indirector: High-Precision Branch Target Injection Attacks Exploiting the Indirect Branch Predictor},author={Li, Luyi and Yavarzadeh, Hosein and Tullsen, Dean},booktitle={USENIX Security Symposium},year={2024},note={* Co-first authors: Luyi Li and Hosein Yavarzadeh.},}
TC (Journal)
PREFENDER: A Prefetching Defender against Cache Side Channel Attacks as A Pretender (Extended Version)
Luyi Li, Jiayi Huang, Lang Feng, and Zhongfeng Wang
Cache side channel attacks are increasingly alarming in modern processors due to the recent emergence of Spectre and Meltdown attacks. A typical attack performs intentional cache access and manipulates cache states to leak secrets by observing the victim’s cache access patterns. Different countermeasures have been proposed to defend against both general and transient execution based attacks. Despite their effectiveness, they mostly trade some level of performance for security, or have restricted security scope. In this paper, we seek an approach to enforcing security while maintaining performance. We leverage the insight that attackers need to access cache in order to manipulate and observe cache state changes for information leakage. Specifically, we propose Prefender, a secure prefetcher that learns and predicts attack-related accesses for prefetching the cachelines to simultaneously help security and performance. Our results show that Prefender is effective against several cache side channel attacks while maintaining or even improving performance for SPEC CPU 2006 and 2017 benchmarks.
@article{2024tc,title={PREFENDER: A Prefetching Defender against Cache Side Channel Attacks as A Pretender (Extended Version)},author={Li, Luyi and Huang, Jiayi and Feng, Lang and Wang, Zhongfeng},journal={IEEE Transactions on Computers},year={2024},doi={10.1109/TC.2024.3377891},}
2022
DATE 2022
PREFENDER: A Prefetching Defender against Cache Side Channel Attacks as A Pretender
Luyi Li, Jiayi Huang, Lang Feng, and Zhongfeng Wang
Cache side channel attacks are increasingly alarming in modern processors due to the recent emergence of Spectre and Meltdown attacks. A typical attack performs intentional cache access and manipulates cache states to leak secrets by observing the victim’s cache access patterns. Different countermeasures have been proposed to defend against both general and transient execution based attacks. Despite their effectiveness, they all trade some level of performance for security. In this paper, we seek an approach to enforcing security while maintaining performance. We leverage the insight that attackers need to access cache in order to manipulate and observe cache state changes for information leakage. Specifically, we propose PREFENDER, a secure prefetcher that learns and predicts attack-related accesses for prefetching the cachelines to simultaneously help security and performance. Our results show that PREFENDER is effective against several cache side channel attacks while maintaining or even improving performance for SPEC CPU2006 benchmarks.
@inproceedings{2022date,title={PREFENDER: A Prefetching Defender against Cache Side Channel Attacks as A Pretender},author={Li, Luyi and Huang, Jiayi and Feng, Lang and Wang, Zhongfeng},booktitle={Design, Automation and Test in Europe},year={2022},doi={10.23919/DATE54114.2022.9774658},}
2021
TC (Journal)
RVDFI: A RISC-V Architecture with Security Enforcement by High Performance Complete Data-Flow Integrity
Lang Feng*, Jiayi Huang*, Luyi Li, Haochen Zhang, and Zhongfeng Wang
With the rapid revolution of open-source hardware, RISC-V architecture has been prevalent in both academic research and industrial developments. Due to the increasing threats of information leakage, it is imperative to provide a secure RISC-V ecosystem to defend against malicious software exploits. Toward this goal, data-flow integrity (DFI) is employed as a strict security policy for enforcing the legitimacy of each data access, thereby filtering out most of the attack exploits. However, due to the intensive computations needed by DFI, there are only limited proposals successfully implementing partial DFI with low performance overhead. Moreover, all the previous studies failed to enforce the complete DFI policy in a real hardware platform, while trading off security strength for performance efficiency. To provide RISC-V architecture with high security enforcement and low performance overhead, we leverage the open-source Rocket Chip and propose RVDFI, the first complete DFI implementation based on RISC-V architecture with only 17.8% performance overhead on average and 3.9% in minimum, incurring much less performance loss compared to the 166.3% overhead caused by previous complete DFI implementation.
@article{2021tc,title={RVDFI: A RISC-V Architecture with Security Enforcement by High Performance Complete Data-Flow Integrity},author={Feng, Lang and Huang, Jiayi and Li, Luyi and Zhang, Haochen and Wang, Zhongfeng},journal={IEEE Transactions on Computers},year={2021},note={* Co-first authors: Lang Feng and Jiayi Huang.},doi={10.1109/TC.2021.3133701},}
ISVLSI 2021
PipeBSW: A Two-Stage Pipeline Structure for Banded Smith-Waterman Algorithm on FPGA
Luyi Li, Jun Lin, and Zhongfeng Wang
In IEEE Computer Society Annual Symposium on VLSI, 2021
With the development of bioinformatics, there is unprecedented progress and wide applications of DNA sequencing technology, especially in helping biological and medical researchers obtain and analyze complete DNA sequences. Smith-Waterman (S-W) algorithm is one of the critical algorithms used in this technology. However, the tremendous amount of DNA sequence data makes it very inefficient to implement the algorithm only with software, so many recent works have tried to accelerate it on hardware. Unfortunately, many of them implemented on the CPU+FPGA heterogeneous platform are likely to require high memory bandwidth and large hardware resource. To mitigate these drawbacks, in this paper we propose PipeBSW, a hardware acceleration implementation on FPGA based on the concept of banded S-W. Using the lookahead calculation technique in cells and a two-stage pipeline structure for processing elements and the backtracking module, the experimental results show that the proposed system can achieve a 720.9Mbps throughput with a 58.4% LUT consumption reduction, compared with the prior work without the pipeline structure.
@inproceedings{2021isvlsi,title={PipeBSW: A Two-Stage Pipeline Structure for Banded Smith-Waterman Algorithm on FPGA},author={Li, Luyi and Lin, Jun and Wang, Zhongfeng},booktitle={IEEE Computer Society Annual Symposium on VLSI},year={2021},doi={10.1109/ISVLSI51109.2021.00042},}