Investigate growing set of protected mutexes for cluster-analysis #1936
Description
With the update rules, clusters of protecting mutexes can now grow again when some code is found to be unreachable.
Currently, we do not exploit this in mutex-meet and the other analyses in this style, because our clusters there only ever shrink.
May be interesting to check if a set of protecting mutexes growing ever happens in practice. If it does, one could:
- See if the update rules can be modified so our analyses can benefit (e.g., after cluster has grown destructively update per origin)
- Or dream up some dedicated mechanism for this.
Might make for an interesting BSc or MSc thesis.