ums fails to initialise correctly

[Chris Hedley]

On Mon, 22 Feb 2010, Hans Petter Selasky wrote:

>> But nothing at all when it hasn't initialised properly.
>
> Hi,
>
> If you get no "ums_intr_callback()" printouts at all in the failing case, and
> you are running 9-current, I guess that your KWM switch has some special
> requirements on the USB programming which I am not aware about.

It is all a bit confusing: I tried the other day to boot with the mouse 
receiver plugged directly into the computer so I could remove the KVM from 
the equation, and that didn't work either, until I unplugged it and 
plugged it back in after it had booted.  But today it was fine.  Then 
again, I had problems getting the ehci driver to see it when plugging it 
into the computer after it had booted, though the ohci driver was more 
accommodating.  It does have the hallmarks of one of those pesky 
intermittent problems...

I tried the usbconfig ... reset command, but other than producing 
convincing-looking messages, that didn't do the trick: it looked like it 
was there, but absolutely nothing when moving it about.  I even tried the 
power_off/power_on combo in the hope that would be enough, but that didn't 
work either.  It's being extremely stubborn.

One possibly significant element is that usbconfig lists the mouse as 
being full speed (12 MB/s) and the KVM as being low speed (1.5 MB/s) which 
I understand could potentially cause timing problems or some such; though 
that doesn't explain why it seems happy enough once I've physically 
detached and reattached the mouse receiver, since the two devices still 
show the same respective speeds afterwards.

If this really is unworkable, I've been toying with the idea of running my 
workstation purely in legacy mode (assuming I can get it working...) but 
I'm not sure if that's really the best way ahead, either for me or the 
FreeBSD project!  But as long as you or anybody else is happy to help look 
into this odd little problem I've dug up, I'll provide whatever 
information I can.