umangm / rapid Goto Github PK

Hello, while studying Rapid, I noticed that it doesn't handle wait/notify events. It only deals with "Acquire", "Release", "[A]?Rd", "[A]?Wr", "Start", "Join", "Enter", "Exit", "Dummy", and "Br" events. However, in reality, there is a sequential relationship between the two. If not handled, could this lead to false positives?
Looking forward to your response, I'm fortunate to be able to communicate with you!

Hello, after receiving your source code, I feel a bit lost and don't know how to run it. Do you have any basic usage tutorials?

I'm confused with how to generate a trace with STD/CSV format.
Could you please tell me what command I should use?
Thanks.

Recently, I've been using Rapid to detect data races and employing RoadRunner to record and generate traces(rrrun -noFP -tool=P -values -noxml -noEnter -quiet ). However, I quickly noticed that RoadRunner's recording of traces doesn't seem to stop at all.In many benchmarks, this seems to be a common issue, such as in benchmarks like Moldyn and Raytracer.In contrast, running the tested program directly concludes within a few seconds. While recording traces may naturally slow down the execution speed, the pace of RoadRunner seems excessively abnormal. Additionally, the generated number of traces is exceptionally high, reaching hundreds of millions of events. This is significantly different from the event counts reported in the literature for other data race detection tools.

Initially, I thought about filtering out some Java packages, such as org.apache.*, but I learned from the literature that RoadRunner only instruments application code, not Java library code. Therefore, that reasoning doesn't hold. I would like to know if there is something I overlooked that might be causing this issue.