Tongping Liu
Assistant Professor
Department of Electrical and Computer Engineering
University of Massachusetts Amherst

Email: tongping AT umass DOT edu. Phone: 413-545-0239
Office: KEB 309J

[Home]  [Publications & SourceCode]  [Teaching]  [Misc

Note: students under my supervision are underlined.


  • GraphZero: A High-Performance Subgraph Matching System
    ACM SIGOPS Operating Systems Review
    Daniel Mawhirter, Sam Reinehr, Connor Holmes, Tongping Liu, Bo Wu. [pdf]

  • NumaPerf: Predictive NUMA Profiling
    Proceedings of International Conference on Supercomputing (ICS'21)
    Xin Zhao, Jin Zhou, Hui Guan, Wei Wang, Xu Liu, Tongping Liu. [pdf] Acceptance rate: 24.2% (38/157)

    NumaPerf could identify NUMA performance issues even when the application is running on a Non-NUMA machine, which is the first tool of such kind.


  • Watcher: In-Situ Failure Diagnosis for In-Production Software
    Proceedings of the ACM on Programming Languages (OOPSLA'20)
    Hongyu Liu, Sam Silvestro, Xiangyu Zhang, Jian Huang, Tongping Liu. [pdf] [Youtube]

    Watcher is the first ins-situ failure diagnosis tool that could identify root causes of program crashes automatically within seconds or minutes, which is built on top of iReplayer.

    Media Report (English): UMass News
    Media Report (Chinese): Jiqizhixin, sina, sohu,, kknews, thepaper, matpool, 51cto, linkerresearcher ...... (more)

  • Prober: Practically Defending Overflows with Page Protection
    Proceedings of the 35th IEEE/ACM International Conference on Automated Software Engineering (ASE'20)
    Hongyu Liu*, Ruiqin Tian*, Bin Ren, Tongping Liu. [*: equally-contributed] [pdf]
    Acceptance rate: 22.5% (93/408)

    Prober is based on a key observation: all heap overflows are related to arrays. Therefore, it utilizes static analysis to mark all susceptible call-stacks during the compilation, and then employs the page protection to detect any invalid accesses during the runtime. It imposes negligible overhead, with 1.1% on average.


  • CSOD: Context-Sensitive Overflow Detection
    Proceedings of The 2019 International Symposium on Code Generation and Optimization (CGO'19).
    Hongyu Liu, Sam Silvestro Xiaoyin Wang, Lide Duan, Tongping Liu. [pdf]

    CSOD proposes a new overflow detection technique based on calling contexts of memory allocations.


  • Sampler: PMU-based Sampling to Detect Memory Errors Latent in Production Software
    Proceedings of The 51th International Symposium on Microarchitecture (Micro'18)
    Sam Silvestro, Hongyu Liu, Tong Zhang, Changhee Jung, Dongyoon Lee, Tongping Liu. [pdf] [presentation]
    Acceptance rate: 21% (74/351)

    Sampler is a novel sampling-based technique to detect memory vulnerabilities latent in production software, with negligible performance overhead (around 2%).

  • Guarder: A Tunable Secure Allocator
    Proceedings of The 27th USENIX Security Symposium (Security'18)
    Sam Silvestro, Hongyu Liu, Tianyi Liu, Zhiqiang Lin, Tongping Liu. [pdf] [presentation]
    Acceptance Rate: 19.1% (100/524)

    Guarder allows users to specify their security guarantee, which is also the most secure but still efficient allocator (only 3% slower than the Linux allocator).

  • iReplayer: In-situ and Identical Record-and-Replay for Multithreaded Applications
    Proceedings of The 37th annual ACM SIGPLAN conference on Programming Language Design and Implementation (PLDI'18)
    Hongyu Liu, Sam Silvestro, Wei Wang, Chen Tian, Tongping Liu. [pdf] [presentation]
    Acceptance Rate: 19.8% (55/277)

    iReplayer is the first in-situ and identical record-and-replay system, with only 3% recording overhead. iReplayer unlocks numerous possibilities in failure diagnosis and failure remediation......

  • A User Space-based Project for Practicing Core Memory Management Concepts
    Proceedings of The 49th ACM Technical Symposium on Computer Science Education (SIGCSE'18)
    Sam Silvestro, Timothy T. Yuen, Corey Crosser, Dakai Zhu, Turgay Korkmaz, Tongping Liu. [pdf] [ProjectWebsite]

    Describes one OS project that allows users to connect different concepts of memory management together. Good for both undergraduate and graduate students.


  • FreeGuard: A Faster Secure Heap Allocator
    Proceedings of ACM Conference on Computer and Communications Security (CCS'17). Citation:
    Sam Silvestro, Hongyu Liu, Corey Crosser, Zhiqiang Lin, Tongping Liu. [pdf] [presentation] [SourceCode]
    Acceptance Rate:18% (151/836)

    FreeGuard is a secure allocator that balances security and performance, with 1% overhead.

  • UnDead: Detecting and Tolerating Deadlocks in Production Software
    Proceedings of 32nd IEEE/ACM International Conference on Automated Software Engineering (ASE'17). Citation: N/A.
    Jinpeng Zhou, Sam Silvestro, Hongyu Liu, Yan Cai, Tongping Liu. [pdf] [presentation] [SourceCode]
    Acceptance Rate:21% (65/314)

    UnDead proposes a light-weight tool for detecting deadlocks, with 2% performance overhead. UnDead also only imposes bounded memory overhead. After the detection, it could also tolerate found deadlocks in future executions. To my opinion, this is the last work in deadlock detection, due to its effectiveness and low overhead.

  • SyncPerf: Categorizing, Detecting, and Diagnosing Synchronization Performance Bugs
    Proceedings of 2017 European Conference on Computer Systems (EuroSys'17). Citation: 1.
    Mohammad Mejbah ul Alam*, Tongping Liu*, Guangming Zeng, Abdullah Muzahid. [*: Equally Contributed] [pdf] [presentation] [SourceCode]
    Acceptance Rate:20.5% (41/200)

    SyncPerf identifies performance issues of using mutex locks and conditional variables, with the light-weight performance overhead. It discovers many unknown performance issues. The results also show that busy-waiting may achieve much better performance than traditional mutex-based implementation.


  • DoubleTake: Fast and Precise Error Detection via Evidence-Based Dynamic Analysis.
    Proceedings of the 38th International Conference on Software Engineering (ICSE'16). Citation: 4.
    Tongping Liu, Charlie Curtsinger, Emery D. Berger. [pdf] [SourceCode]
    Acceptance Rate:19% (101/530)

    DoubleTake is the single-threaded version of iReplayer. It proposes the evidence-based approach to detect memory vulnerabilities.

  • Cheetah: Detecting False Sharing Efficiently and Effectively
    Proceedings of the 2016 International Symposium on Code Generation and Optimization (CGO 2016). Citation: 1.
    Tongping Liu*, Xu Liu*. [*: Equally Contributed] [pdf] [SourceCode]
    Acceptance Rate:23% (25/108)

    Cheetah employs PMUs to detect false sharing issues. It could quantitatively predict the performance impact of false sharing issues.


  • FORESEER: Workload-Aware Data Storage for MapReduce
    International Conference on Distributed Computing Systems (ICDCS 2015), poster
    Jia Zou, Juwei Shi, Tongping Liu, Zhao Cao, and Chen Wang. [pdf]

    Before 2014

  • Predator: Predictive False Sharing Detection
    Proceedings of the 19th ACM SIGPLAN Symposium on Principles and Practice of Par- allel Programming (PPOPP'14). Citation: 12 (Updated)
    Tongping Liu, Chen Tian, Ziang Hu, Emery D. Berger. [Abstract] [pdf] [SourceCode] [BibTeX]
    Acceptance Rate:16% (28/179)

    Predator employs compiler-based instrumentation to detect false sharing issues. It is the first work that could predict the existence of false sharing, even if they do not appear in the execution.

  • Dthreads: Efficient Deterministic Multithreading
    Proceedings of the Twenty-Third ACM Symposium on Operating Systems Principles (SOSP'11)). Citation 240 (Updated).
    Tongping Liu, Charlie Curtsinger, Emery D. Berger. [Abstract] [pdf] [BibTeX] [SourceCode] [Technical report version, including input parameters of benchmarks]
    Acceptance Rate: 18% (28/153)

    Dthreads proposes the process-based isolation to achieve the deterministic multithreading, which is one piorneer research in this direction.

  • Sheriff: Precise Detection and Automatic Mitigation of False Sharing
    Proceedings of the 26th Annual ACM SIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications (OOPSLA'11). Citation: 65 (Updated).
    Tongping Liu, Emery D. Berger. [Abstract] [pdf] [SourceCode] [BibTeX]
    Acceptance Rate: 37% (61/166)

    Sheriff is the first work that could pinpoint the statements or variables with false sharing issue.

  • Grace: Safe Multithreaded Programming for C/C++
    Proceedings of the 24th Annual ACM SIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications(OOPSLA 2009). Citation: 308 (Updated).
    Emery D. Berger, Ting Yang, Tongping Liu, Gene Novark. [Abstract] [pdf] [BibTeX]
    Acceptance Rate: 17% (25/144)

    Grace is one pioneer work in deterministic multithreading.

  • Redline: First Class Support for Interactivity in Commodity Operating Systems
    Proceedings of the 8th USENIX Symposium on Operating Systems Design and Implementation (OSDI 2008). Citation: 84 (Updated).
    Ting Yang, Tongping Liu, Emery D. Berger, Scott F. Kaplan and J. Eliot B. Moss. [Abstract] [pdf] [BibTeX]
    Acceptance Rate: 13% (26/193)

    Redline proposes the collaboration of multiple OS components, such as scheduler, memory manager, and IO scheduler, to achieve the interactivity of applications.

  • Dynamic Detection of Stack Overflow
    Tongping Liu. Journal of Computer Science (China), No.9, 2007.


  • System and Method for Detecting False Sharing. Tongping Liu, Chen Tian, Ziang Hu. No. 61/858,824.

  • System and Method for False Sharing Prediction. Chen Tian, Tongping Liu, Ziang Hu. No. 61/858,857.

  • Protection of Complex Shared Data through Memory Paging Mechanisms. Daniel G. Waddington, Chen Tian, Tongping Liu. No. 61/523,231.

  • Coupled Lock Allocation and Lookup for Shared Data Synchronization in Symmetric Multithreading Environments. Daniel G. Waddington, Tongping Liu, Chen Tian. No. 61/524,677.

  • Mapping Guest Pages to Disk Blocks to Improve Virtual Machine Management Process. Kiran Tati, Rajesh Venkatasubramanian, Carl A. Waldspurger, Alexander Thomas Garthwaite, Tongping Liu. No. 61/595,522.