updated patch with throttling

[Roman Kagan]

On Fri, Mar 25, 2005 at 04:05:20PM +0100, Duncan Sands wrote:
> > Here you are...
> 
> Did you have any thoughts about usb_clear_halt?

Not yet...

But it doesn't look too tough: we can add struct work_struct to
usbatm_data, which will get scheduled from usbatm_check_usb_errcode when
needed, and will usb_clear_halt() in the process context and then, if
successful, will reenable and reschedule both tasklets.  usbnet does a
similar thing.

I have to leave now, I'll code it when I'm back (Sunday evening) and
send out for review.

Cheers,
  Roman.