I just committed 2 fixes, which hopefully should resolve the bluetooth
problems. The first changes an option to the hciattach program, since
it seems the PXA BTUART doesn't actually do hardware flow control,
which is not entirely obvious from the PXA docs. The 2nd patch fixes a
bug in the linux kernel which returns the wrong UART status flags, so
that the serial driver thinks that CTS is always set, and will send
even if the receiver's not ready for it.
If you'd like to try the new code, please go ahead and try. I've done
some little testing, and it seems to be working for me; I'll be doing
more in the morning too.