I think port is doable.

Packages of Maemo5 can be divided in three category:
*firmware
*driver
*software

For porting:
Maemo5 firmware-files is not needed, but you need firmware of target device. Typically firmware is not compiled for targeting specific system, so it would be possible to use stock firmware of target device.

Maemo5 drivers is not needed, but you need drivers for target device. Drivers are compiled against specific system, so you really need open source drivers for target system.

Software: this contains libraries, services and applications. If some of them are missing, you don't have that functionality.

----
This is (full) list of closed source packages: http://wiki.maemo.org/Fremantle_closed_packages Many package is already rewrote for open source use. Some packages are considerer useless and can be dropped without loss of functionality. There are still work to do. Inspecting what can drop, and coding alternatives.


First target to "port" maemo5 should be N900. Seems majority of users/devs nowadays understand technical and practical reasons to have as much as possible packages open source. For porting, percentage should be full 100% (maybe excluding firmware).

Second target could be N8x0, because it has (almost) every driver as open source (what is missing?). Of course this is more of proof of concept, because N8x0 is not phone.

Need an X86 port, soon there will be a ton of Windows 8 32bit tablets coming out, Maemo 5 on them would be way better.