{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,26]],"date-time":"2026-09-26T11:59:23Z","timestamp":1790423963540,"version":"4.1.0"},"reference-count":48,"publisher":"Association for Computing Machinery (ACM)","issue":"OOPSLA1","license":[{"start":{"date-parts":[[2026,4,10]],"date-time":"2026-04-10T00:00:00Z","timestamp":1775779200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/legalcode"}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":["Proc. ACM Program. Lang."],"published-print":{"date-parts":[[2026,4,10]]},"abstract":"<jats:p>GPUs play an increasingly important role in modern software. However, the heterogeneous host-device execution model and expanding software stacks make GPU programs prone to memory-safety and concurrency bugs that evade static analysis. While fuzz-testing, combined with dynamic error checking tools, offers a plausible solution, it remains underutilized for GPUs. In this work, we identify three main obstacles limiting prior GPU fuzzing efforts: (1) kernel-level fuzzing leading to false positives, (2) lack of device-side coverage-guided feedback, and (3) incompatibility between coverage and sanitization tools. We present cuFuzz, the first CUDA-oriented fuzzer that makes GPU fuzzing practical by addressing these obstacles.<\/jats:p>\n                  <jats:p>\n                    cuFuzz uses whole program fuzzing to avoid false positives from independently fuzzing device-side kernels. It leverages NVBit to instrument device-side instructions and merges the resultant coverage with compiler-based host coverage. Finally, cuFuzz decouples sanitization from coverage collection by executing host- and device-side sanitizers in separate processes. cuFuzz uncovers 43 previously unknown bugs (19 in commercial libraries) across 14 CUDA programs, including illegal memory accesses, uninitialized reads, and data races. cuFuzz achieves significantly more discovered edges and unique inputs compared to baseline approaches, especially on closed-source targets. Moreover, we quantify the execution time overheads of the different cuFuzz components and add persistent-mode support to improve the overall fuzzing throughput. Our results demonstrate that cuFuzz is an effective and deployable addition to the GPU testing toolbox. cuFuzz is publicly available at\n                    <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/github.com\/NVlabs\/cuFuzz\/\">https:\/\/github.com\/NVlabs\/cuFuzz\/<\/jats:ext-link>\n                    .\n                  <\/jats:p>","DOI":"10.1145\/3798231","type":"journal-article","created":{"date-parts":[[2026,4,10]],"date-time":"2026-04-10T14:33:43Z","timestamp":1775831623000},"page":"877-904","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Hunting CUDA Bugs at Scale with cuFuzz"],"prefix":"10.1145","volume":"10","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-6971-6996","authenticated-orcid":false,"given":"Mohamed","family":"Tarek Ibn Ziad","sequence":"first","affiliation":[{"name":"NVIDIA, Westford, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3154-7530","authenticated-orcid":false,"given":"Christos","family":"Kozyrakis","sequence":"additional","affiliation":[{"name":"NVIDIA, Santa Clara, USA"},{"name":"Stanford University, Stanford, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2026,4,10]]},"reference":[{"key":"e_1_3_1_2_2","first-page":"265","volume-title":"OSDI\u201916: Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation","author":"Abadi Mart\u00edn","year":"2016","unstructured":"Mart\u00edn Abadi, Paul Barham, Jianmin Chen, Zhifeng Chen, Andy Davis, Jeffrey Dean, Matthieu Devin, Sanjay Ghemawat, Geoffrey Irving, Michael Isard, Manjunath Kudlur, Josh Levenberg, Rajat Monga, Sherry Moore, Derek G. Murray, Benoit Steiner, Paul Tucker, Vijay Vasudevan, Pete Warden, Martin Wicke, Yuan Yu, and Xiaoqiang Zheng. 2016. TensorFlow: A System for Large-Scale Machine Learning. In OSDI\u201916: Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation. Savannah, GA, USA, 265\u2013283. https:\/\/www.usenix.org\/conference\/osdi16\/technical-sessions\/presentation\/abadi"},{"key":"e_1_3_1_3_2","doi-asserted-by":"publisher","DOI":"10.1145\/2384616.2384625"},{"key":"e_1_3_1_4_2","unstructured":"Black Duck. 2025. Coverity Documentation 2025.6.2. https:\/\/documentation.blackduck.com\/bundle\/coverity-docs\/page\/webhelp-files\/help_center_start.html."},{"key":"e_1_3_1_5_2","volume-title":"USENIX Security\u201923: Proceedings of the 32nd USENIX Conference on Security Symposium","author":"Christou Neophytos","year":"2023","unstructured":"Neophytos Christou, Di Jin, Vaggelis Atlidakis, Baishakhi Ray, and Vasileios P. Kemerlis. 2023. IvySyn: automated vulnerability discovery in deep learning frameworks. In USENIX Security\u201923: Proceedings of the 32nd USENIX Conference on Security Symposium. USENIX Association, Anaheim, CA, USA, Article 134, 18 pages. https:\/\/www.usenix.org\/conference\/usenixsecurity23\/presentation\/christou"},{"key":"e_1_3_1_6_2","doi-asserted-by":"publisher","DOI":"10.1145\/3597503.3623343"},{"key":"e_1_3_1_7_2","volume-title":"WOOT\u201920: Proceedings of the 14th USENIX Conference on Offensive Technologies","author":"Fioraldi Andrea","year":"2020","unstructured":"Andrea Fioraldi, Dominik Maier, Heiko Ei\u00dffeldt, and Marc Heuse. 2020. AFL++: combining incremental steps of fuzzing research. In WOOT\u201920: Proceedings of the 14th USENIX Conference on Offensive Technologies. USENIX Association, USA, Article 10, 1 pages. https:\/\/www.usenix.org\/conference\/woot20\/presentation\/fioraldi"},{"key":"e_1_3_1_8_2","first-page":"2577","volume-title":"USENIX Security\u201920: Proceedings of the 29th USENIX Security Symposium","author":"Gan Shuitao","year":"2020","unstructured":"Shuitao Gan, Chao Zhang, Peng Chen, Bodong Zhao, Xiaojun Qin, Dong Wu, and Zuoning Chen. 2020. GREYONE: Data Flow Sensitive Fuzzing. In USENIX Security\u201920: Proceedings of the 29th USENIX Security Symposium. San Antonio, TX, USA, 2577\u20132594. https:\/\/www.usenix.org\/conference\/usenixsecurity20\/presentation\/gan"},{"key":"e_1_3_1_9_2","unstructured":"Google. 2016. OSS-Fuzz: Continuous Fuzzing for Open Source Software. https:\/\/google.github.io\/oss-fuzz\/."},{"key":"e_1_3_1_10_2","unstructured":"Google. 2019. american fuzzy lop: a security-oriented fuzzer. https:\/\/github.com\/google\/AFL"},{"key":"e_1_3_1_11_2","doi-asserted-by":"publisher","DOI":"10.1145\/3324884.3416571"},{"key":"e_1_3_1_12_2","first-page":"4033","volume-title":"USENIX Security\u201924: Proceedings of the 33rd USENIX Security Symposium","author":"Guo Yanan","year":"2024","unstructured":"Yanan Guo, Zhenkai Zhang, and Jun Yang. 2024. GPU Memory Exploitation for Fun and Profit. In USENIX Security\u201924: Proceedings of the 33rd USENIX Security Symposium. USENIX Association, Philadelphia, PA, USA, 4033\u20134050. https:\/\/www.usenix.org\/conference\/usenixsecurity24\/presentation\/guo-yanan"},{"key":"e_1_3_1_13_2","doi-asserted-by":"publisher","DOI":"10.1145\/3587156"},{"key":"e_1_3_1_14_2","doi-asserted-by":"publisher","DOI":"10.1109\/SC41406.2024.00042"},{"key":"e_1_3_1_15_2","doi-asserted-by":"publisher","DOI":"10.1109\/COMPSAC48688.2020.0-156"},{"key":"e_1_3_1_16_2","doi-asserted-by":"publisher","DOI":"10.1109\/ISPASS57527.2023.00041"},{"key":"e_1_3_1_17_2","unstructured":"Zheming Jin and Jeffrey S. Vetter. 2023. HeCBench: heterogeneous computing benchmark suite. https:\/\/github.com\/zjin-lcf\/HeCBench."},{"key":"e_1_3_1_18_2","doi-asserted-by":"publisher","DOI":"10.14722\/ndss.2025.230218"},{"key":"e_1_3_1_19_2","doi-asserted-by":"publisher","DOI":"10.1145\/3477132.3483545"},{"key":"e_1_3_1_20_2","first-page":"4931","volume-title":"USENIX Security\u201923: Proceedings of the 32nd USENIX Security Symposium","author":"Kim Tae Eun","year":"2023","unstructured":"Tae Eun Kim, Jaeseung Choi, Kihong Heo, and Sang Kil Cha. 2023. DAFL: Directed Grey-box Fuzzing guided by Data Dependency. In USENIX Security\u201923: Proceedings of the 32nd USENIX Security Symposium. Anaheim, CA, USA, 4931\u20134948. https:\/\/www.usenix.org\/conference\/usenixsecurity23\/presentation\/kim-tae-eun"},{"key":"e_1_3_1_21_2","doi-asserted-by":"publisher","DOI":"10.1145\/2830772.2830818"},{"key":"e_1_3_1_22_2","doi-asserted-by":"publisher","DOI":"10.1145\/3520313.3534655"},{"key":"e_1_3_1_23_2","doi-asserted-by":"publisher","unstructured":"Wentao Li Zhiwen Chen Xin He Guoyun Duan Jianhua Sun and Hao Chen. 2022. CVFuzz: Detecting complexity vulnerabilities in OpenCL kernels via automated pathological input generation. Future Generation Computer Systems 127 (2022) 384\u2013395. doi:10.1016\/j.future.2021.09.006","DOI":"10.1016\/j.future.2021.09.006"},{"key":"e_1_3_1_24_2","doi-asserted-by":"publisher","DOI":"10.1145\/3588195.3592991"},{"key":"e_1_3_1_25_2","doi-asserted-by":"publisher","DOI":"10.1109\/SP46214.2022.9833594"},{"key":"e_1_3_1_26_2","doi-asserted-by":"publisher","unstructured":"Geert Litjens Thijs Kooi Babak Ehteshami Bejnordi Arnaud Arindra Adiyoso Setio Francesco Ciompi Mohsen Ghafoorian Jeroen A.W.M. van der Laak Bram van Ginneken and Clara I. S\u00e1nchez. 2017. A survey on deep learning in medical image analysis. Medical Image Analysis 42 (2017) 60\u201388. doi:10.1016\/j.media.2017.07.005","DOI":"10.1016\/j.media.2017.07.005"},{"key":"e_1_3_1_27_2","unstructured":"LLVM. 2016. LibFuzzer: a library for coverage-guided fuzz testing. https:\/\/llvm.org\/docs\/LibFuzzer.html."},{"key":"e_1_3_1_28_2","doi-asserted-by":"publisher","DOI":"10.1145\/96267.96279"},{"key":"e_1_3_1_29_2","unstructured":"NVIDIA. 2020. Multi-Process Service. https:\/\/docs.nvidia.com\/deploy\/mps\/index.html."},{"key":"e_1_3_1_30_2","unstructured":"NVIDIA. 2025. Compute Sanitizer v2025.2.0. https:\/\/docs.nvidia.com\/compute-sanitizer\/ComputeSanitizer\/index.html."},{"key":"e_1_3_1_31_2","unstructured":"NVIDIA. 2025. CUDA Library Samples. https:\/\/github.com\/NVIDIA\/CUDALibrarySamples\/tree\/master."},{"key":"e_1_3_1_32_2","unstructured":"NVIDIA. 2025. NVIDIA CUDA-X Libraries. https:\/\/developer.nvidia.com\/gpu-accelerated-libraries."},{"key":"e_1_3_1_33_2","doi-asserted-by":"publisher","unstructured":"Sang-Ok Park Ohmin Kwon Yonggon Kim Sang Kil Cha and Hyunsoo Yoon. 2021. Mind control attack: Undermining deep learning with GPU memory exploitation. Computers & Security 102 (2021) 102\u2013115. doi:10.1016\/j.cose.2020.102115","DOI":"10.1016\/j.cose.2020.102115"},{"key":"e_1_3_1_34_2","doi-asserted-by":"publisher","DOI":"10.5555\/3454287.3455008"},{"key":"e_1_3_1_35_2","doi-asserted-by":"publisher","DOI":"10.1145\/3366428.3380768"},{"key":"e_1_3_1_36_2","unstructured":"David Peter. 2023. hyperfine. https:\/\/github.com\/sharkdp\/hyperfine"},{"key":"e_1_3_1_37_2","doi-asserted-by":"publisher","DOI":"10.1145\/3722041.3723099"},{"key":"e_1_3_1_38_2","doi-asserted-by":"publisher","DOI":"10.1109\/SP63933.2026.00047"},{"key":"e_1_3_1_39_2","volume-title":"ATC \u201912: Proceedings of the 2012 USENIX Annual Technical Conference","author":"Serebryany Konstantin","year":"2012","unstructured":"Konstantin Serebryany, Derek Bruening, Alexander Potapenko, and Dmitry Vyukov. 2012. AddressSanitizer: a fast address sanity checker. In ATC \u201912: Proceedings of the 2012 USENIX Annual Technical Conference. Boston, MA, USA, 10 pages."},{"key":"e_1_3_1_40_2","doi-asserted-by":"publisher","unstructured":"Mohammad Shoeybi Mostofa Patwary Raul Puri Patrick LeGresley Jared Casper and Bryan Catanzaro. 2019. Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism. arXiv preprint arXiv:1909.08053 (2019). https:\/\/doi.org\/10.48550\/arXiv.1909.08053 10.48550\/arXiv.1909.08053","DOI":"10.48550\/arXiv.1909.08053"},{"key":"e_1_3_1_41_2","unstructured":"Saurabh Singh Ruobing Han Jaewon Lee Seonjin Na Yonghae Kim Taesoo Kim and Hyesoon Kim. 2026. CuFuzz: Hardening CUDA Programs through Transformation and Fuzzing. arXiv preprint arXiv:2601.01048 (2026). https:\/\/arxiv.org\/abs\/2601.01048"},{"key":"e_1_3_1_42_2","doi-asserted-by":"crossref","unstructured":"Yasunari Suzuki Yoshiaki Kawase Yuya Masumura Yuria Hiraga Masahiro Nakadai Jiabao Chen Ken M. Nakanishi Kosuke Mitarai Ryosuke Imai Shiro Tamiya Takahiro Yamamoto Tennin Yan Toru Kawakubo Yuya O. Nakagawa Yohei Ibe Youyuan Zhang Hirotsugu Yamashita Hikaru Yoshimura Akihiro Hayashi and Keisuke Fujii. 2020. Qulacs: a fast and versatile quantum circuit simulator for research purpose. https:\/\/arxiv.org\/abs\/2011.13524.","DOI":"10.22331\/q-2021-10-06-559"},{"key":"e_1_3_1_43_2","doi-asserted-by":"publisher","DOI":"10.1109\/SP.2013.13"},{"key":"e_1_3_1_44_2","doi-asserted-by":"publisher","DOI":"10.1145\/3591225"},{"key":"e_1_3_1_45_2","doi-asserted-by":"publisher","DOI":"10.5281\/zenodo.18727045"},{"key":"e_1_3_1_46_2","doi-asserted-by":"publisher","DOI":"10.1145\/3352460.3358307"},{"key":"e_1_3_1_47_2","doi-asserted-by":"publisher","DOI":"10.1145\/3510003.3510041"},{"key":"e_1_3_1_48_2","doi-asserted-by":"publisher","DOI":"10.1145\/3533767.3534220"},{"key":"e_1_3_1_49_2","doi-asserted-by":"publisher","unstructured":"Yuhao Zhou Peng Jia Jiayong Liu and Ximing Fan. 2026. Fuzz4Cuda: A fuzzing framework for GPU library vulnerability detection. Computers & Security 161 (2026). doi:10.1016\/j.cose.2025.104754","DOI":"10.1016\/j.cose.2025.104754"}],"container-title":["Proceedings of the ACM on Programming Languages"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3798231","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,26]],"date-time":"2026-09-26T11:23:29Z","timestamp":1790421809000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3798231"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,10]]},"references-count":48,"journal-issue":{"issue":"OOPSLA1","published-print":{"date-parts":[[2026,4,10]]}},"alternative-id":["10.1145\/3798231"],"URL":"https:\/\/doi.org\/10.1145\/3798231","relation":{},"ISSN":["2475-1421"],"issn-type":[{"value":"2475-1421","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,10]]},"assertion":[{"value":"2025-10-10","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-02-17","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-04-10","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}