10 likes | 124 Vues
This work explores a novel approach to detecting sharing errors, data races, and atomicity violations at runtime in multi-threaded applications. The system offers support for locks, joins, events, and condition variables, providing a comprehensive framework for ensuring memory safety. Its lightweight design makes it suitable for inclusion in all debug builds, thus facilitating extensive error detection during development. Researchers from Microsoft Research and the University of Maryland, including Jean-Philippe Martin, Michael Hicks, and others, have collaborated on this innovative solution.
E N D
Thread Ownership of Memory Finds sharing errors at runtime Data races Atomicity Violations Memory Errors Supports locks but also joins, events, condition variables, etc. Cheap enough to keep in all debug builds Ask me how! Jean-Philippe Martin Microsoft Research Michael Hicks University of Maryland, College Park Manuel Costa Microsoft Research Periklis Akritidis University of Cambridge Miguel Castro Microsoft Research by Jean-Philippe Martin, Michael Hicks, Manuel Costa, Periklis Akritidis, and Miguel Castro (in collaboration between Microsoft Research and the University of Maryland, College Park) contact: jpmartin@microsoft.com or more details at: http://research.microsoft.com/apps/pubs/default.aspx?id=117864