Hello folks, here's another attempt of polishing the serial drivers: * lots of minor cleanups to make checkpatch happier (eg. formatting, includes, inttypes, ...) * use appropriate logging helpers instead of printk() * consequent use of mapsize/mapbase fields: the basic idea is, all drivers should fill mapbase/mapbase fields at init time and later only use those fields, instead of hardcoded values (later on, we can add generic helpers for the map/unmap stuff, etc) * untwisting serial8250_port_size() at all: move the iomem size probing to initialization time, move out some platform specific magic to corresponding platform code, etc. Unfortunately, I don't have the actual hardware to really test all the code, so please let me know if there's something broken in here. have fun, --mtx
