Basically: The person who designed the crossbar solution was not having a good day. Or maybe was at his/her peak performance but would have been better placed at a different position in the staff hierarchy. The concept might have sounded slick when originally thought about, but having the enabling/disabling of one peripherial move the pin allocations for multiple other peripherials is a very hurtful design choice. Real life products have a tendancy to change during their life time.

A very, very good processor design would have had a full crossbar switch, allow new peripherials to be used without affecting any other assignments.

A "high up" Cygnal (it was in those days) person told me: "A full, "desirable" crossbar was in the first SILabs chip prototype. It turned out to take 7/8 of a too large die. Thus the "hurtful" crossbar was implemented."

the crossbar as is is not that big a deal if you know what you are doing i.e. not your first experience with it. I, today, if there is space, will allocate a few unused devices in the prototype crossbar to have some "shuffle room"

Erik