quoted
Ok, what I did was grepping for all .map_io-s. Then I sorted the list
and had a look at first 100 and about 50% of them were doing more than
just creating mappings.
The answer to that is: they shouldn't be now that we have the init_early
hook. The only remainder for .map_io is where platforms make run-time
decisions about what to map based on some register value somewhere
(eg, Assabet vs Assabet+Neponset).
I do have a large patch series floating around in my git tree which tries
to clean up to all those map_io functions - the biggest stumbling block
to them is the Samsung stuff being indirected through its own tables.
Awesome. I'll work with an assumption that future map_io-s will only
create static mappings and nothing more then.
Cheers!
Pawe?