자료유형 | 학위논문 |
---|---|
서명/저자사항 | Enabling Program Analysis Through Deterministic Replay and Optimistic Hybrid Analysis. |
개인저자 | Devecsery, David. |
단체저자명 | University of Michigan. Computer Science & Engineering. |
발행사항 | [S.l.]: University of Michigan., 2018. |
발행사항 | Ann Arbor: ProQuest Dissertations & Theses, 2018. |
형태사항 | 115 p. |
기본자료 저록 | Dissertation Abstracts International 79-12B(E). Dissertation Abstract International |
ISBN | 9780438126992 |
학위논문주기 | Thesis (Ph.D.)--University of Michigan, 2018. |
일반주기 |
Source: Dissertation Abstracts International, Volume: 79-12(E), Section: B.
Adviser: Peter M. Chen. |
요약 | As software continues to evolve, software systems increase in complexity. With software systems composed of many distinct but interacting components, today's system programmers, users, and administrators find themselves requiring automated ways |
요약 | In this thesis I focus on enabling practical entire-system retroactive analysis, allowing programmers, users, and system administrators to diagnose and understand the impact of these devastating mishaps. I focus primarly on two techniques. First |
요약 | Record and replay systems greatly aid in solving a variety of problems, such as fault tolerance, forensic analysis, and information providence. These solutions, however, assume ubiquitous recording of any application which may have a problem. Cu |
요약 | Dynamic analysis is used to retroactively identify and address many forms of system mis-behaviors including: programming errors, data-races, private information leakage, and memory errors. Unfortunately, the runtime overhead of dynamic analysis |
요약 | In this thesis I demonstrate that Arnold's ability to record and replay entire computer systems, combined with optimistic hybrid analysis's ability to quickly analyze prior computation, enable a practical and useful entire system retroactive ana |
일반주제명 | Computer science. |
언어 | 영어 |
바로가기 | ![]() |