Originally Posted by Sir Ementaler

We don't yet know the details, but I believe that rewriting some of mask checking functions in an earlier version caused problems in what must have been fragile and technically incorrect code of JJ2 at assembly level. The issue has been addressed in 5.0 and my modifications to the functions seemed to grant much more appropriate results (even better than in vanilla JJ2 or previous versions of JJ2+ for that matter).