bme is not any problem, we know how to deal with that bugger quite well.

We have a few handles (force_host bit for musb core in SoC et al) to poke hardware, but results are random. So what's needed is a really good kernel hacker, preferably familiar with USB drivers, to code a stable driver alternative to musb_core.c and relatives, that will take care of all the hardware handles wiggling and dials wheeling.
Alas musb_core.c plus related files is really complex and so are the changes that need to be applied and any debugging as that's kernel code you can't strace or run under gdb.
For now we are in a limbo state :-S I probably should clearly state I'm *not* the project head of the sw part of h-e-n aka usb hostmode. I am just the hardware dude to help when it comes to reading datasheets. we are missing developers

cheers
jOERG