> ... even if you do this, unless you renumber all device tables (an

Talk about a flag day for compatibility, eeew. But it would be a VERY nice
thing to have one set of major numbers across all ports, just to eliminate
the problem of "this became an MI driver so now it needs a new number on
every port that might want to start using it".

> ports to have the same one, you still end up needing multiple "one
> place"s.

Next best thing: a special purpose "lint" that verifies the multiple "one
place"s are at least consistent with each other? Is there a well-defined
test that could be implemented and added to kernel builds?

Todd Whitesel
toddpw @ best.com