Re: loading firmware while usermodehelper disabled.

So the only thing we need is to fix the braindamaged firmware
interfaces. We need caching, and we need to replace (or at least
extend) "load_firware()" with "[get|put]_firmware()".

So at what moment after a resume does a firmware that was in use with a
device go away and what do we do about devices inserted while suspended ?

Remembering that

- there is no moment when USB says 'we have finished re-enumeration'.
  There is no point at which you can say 'we have finished
  re-enumeration, we are now starting on enumeration of new stuff'. New
  stuff wanting firmware will pop up on resume mixed in with old stuff.

- we don't seem to have any device or bus infrastructure for doing a
  'post resume' purge of such data if we could decide when that moment
  is ?

Alan