Lian Li - Static Data Race Detection for Pthreads Programs




Rating is available when the video has been rented.
This feature is not available right now. Please try again later.
Published on Oct 17, 2016

Abstract: Data race occurs when two threads access the same location concurrently, and at least one of the access is write. It is one of the most common bugs in multi-threaded programs and can lead to serious problem. Existing static analysis techniques for data race detection compute the correlation between locks and shared variables, and a data race is reported if a shared variable is not consistently guarded by the same lock. This method often suffers from high false positive rates. In this talk, I will introduce our recent progress on this topic: we have implemented a static race detector which computes both the lockset and the happen before order to report a race. Experimental results demonstrated that the happen before order is effective in eliminating many false positives. However, there are still many false positives being reported due to complex thread inter-leavings which are very difficult to compute statically.

Bio: Lian was the fourth member of the Parfait team at Sun Labs, he joined in 2008 after completing a PhD at UNSW. He contributed to many of analysis for Parfait, including: - symbolic analysis for detecting buffer overflows and integer overflows - pointer value flow analysis for detecting null pointer dereference. - context-sensitive points-to analysis for C/C++ - use-after-free detection using points-to Lian also worked on the Whitebox Fuzzing project for Java library code.
In the last year, Lian has been working detection of concurrent bugs, in particular data races, at the Chinese Academy of Sciences.


When autoplay is enabled, a suggested video will automatically play next.

Up next

to add this to Watch Later

Add to

Loading playlists...