Problems with FTDI Vinculum protocol. Anyone used it here?

I am trying to use a Vinculum VNC1L device to access USB Flash but am having 
some problems which look to me like the protocol is broken. Yes, I have 
contacted FTDI/Vinculum support but recent questions have gone unaswered 
(and unacknowledged). So I am asking here if anyone has experience of the 
device and its protocol?

The story so far.
* I have a board designed that has an NXP ARM, RAM, ROM, LCD and the 
Vinculum thingy.
* I have flashed the chip with the VDAP firmware version 3.66 and verified 
it.
* I'm currently big-banging the low level 'SPI' signals even though the CPU 
can do 'standard' SPI on these pins. This is to minimise the issues at this 
stage. The Vinculum SPI is an abuse of the term since the interface is only 
superficially like SPI.
* I can talk to the chip and have read/written data successfully, but 
unreliably.

My problem, though, is this. Most of the command protocol consists of 
sending commands and receiving status and data. Thats all well and good. 
However, when a USB device is plugged into or unplugged from either USB host 
port I get an asynchronous notification, by which I mean bytes are received 
via the interface without having sent a request. The problem a see with this 
is that there is a race condition where receive data is flushed just before 
sending a command then an asynchronous status measage is mixed up with then 
command response because it was generated between the flush and receipt of 
command. I can't see how its possible to get round this. My concern is when 
data is being written to a USB Flash device while a mouse/keyboard/printer 
or whatever is being connected on the other host port.

Has anyone seen this problem? Any pointer to resolving it?

Thanks in advance,
Peter