Open moralismercatus opened 7 years ago
Remember that while this may be useful as a starting place for BB coverage, the fact that CoverageSanitizer requires compile-time instrumentation, makes it of little deployable use.
For research purposes, I believe it can be useful - or at least the idea. For example, say we have two trace selection strategies we're testing, and we want to compare their rates of coverage dynamically. This may be a way to calculate that.
In particular, how can the basic-block coverage be leveraged?