ums fails to initialise correctly

I wasn't sure whether this question was better asked on this list or the 
freebsd-usb one; but since I'm already subscribed here, I figured it was a 
good place to start.

I'm having a lot of problems getting the ums (USB mouse) driver to 
initialise the mouse correctly.  The only time it'll do so is if I 
physically unplug the mouse and reattach it; otherwise, it seems there's 
no way of getting it to initialise correctly.  The mouse in question is a 
Logitech MX Revolution wireless model, though I'm not sure how relevant 
that is.

Confusingly, the kernel messages are identical when it comes to seeing the 
mouse initially and reattaching it; perhaps significantly, if I plug it 
into my KVM switch (as opposed to one of the computer's own ports), it 
still fails to initialise if I switch it away from the FreeBSD system and 
back again: I have to do the physical detach/reattach process to get it to 
reinitialise.  One possible lead is that once the mouse is working, it's 
fine when it's switched between the two FreeBSD systems, but switching to 
Windows and back renders it unusable again until I physically reattach it.

I had wondered if it might be some hardware weirdness, but Windows (XP and 
7) and Linux (on the same computer as FreeBSD) have no problems with it.

The symptoms are just a lack of any data; otherwise everything seems 
identical between it working and not: the messages are the same, the /dev 
entry is being created correctly, devd starts up the moused process for 
me, but there's nothing.  It's not a moused problem either, cat -v 
/dev/ums1 also reveals no data.

I've tried pretty much every BIOS configuration I can think of, I've tried 
FreeBSD with both the ehci and ohci drivers, both compiled into the kernel 
and as modules; I've used usbconfig to see if there's any difference in 
configuration between the mouse working and not working, but nothing at 
all; variously loading and unloading ums, attempting to reset it or do a 
software power off and back on with usbconfig also won't bring it back to 
life.  Meanwhile, the USB keyboard "just works", whatever I do.

I spent much of the past couple of days Googling for suggestions and 
discovered it wasn't an uncommon problem, but could find no definite 
solution (other than suggestions to disable legacy support if it was 
enabled and vice versa - which didn't work).  I did find similar queries 
going back to 2003, which makes me wonder if it's a problem without any 
obvious solution, but I thought I'd ask anyway.

I'm using -current in its amd64 form on an old-ish dual-core AMD with an 
Asus motherboard, if that gives anybody any clues.  But I figure I should 
get back to whether or not this is the right place to ask before I start 
posting reams of configuration and debugging information!