Hi,
sorry for dropping the ball on this one in the past ...
The main problem isn't that there are two different firmware versions (the differences are small enough), but rather the fact that they changed the task structures and the scheduler quite a bit. I kinda gave up when debugging turned unreasonably painful because of the inability to debug that easily using the qemu port. That situation might have improved noticeably.
I'm not sure I'll find time to hack away at this anytime soon, but the differing firmware are less of an issue than the necessity of a lot of try & error to find out how scheduling works now; and what structures need to be access / piggy-backed to make ML work as expected.
count