I've been posting recently. This one adds:
   - slave_attach() and slave_detach()
   - code clean up looking for a problem **
   - more debug code allowing scanning cmd sequence to
     be seen in the log (when opts=1)

** after several (never the first) sequences of modprobe/rmmod
on scsi_debug there is either:
   - an oops during modprobe when driverfs tries to create a
     directory
   - or a WARN_ON() at drivers/base/bus.c:277 during rmmod
[examples attached]

I'm not sure whether the problem is in scsi_debug, the
scsi mid level or in the driverfs code. Grepping indicates
that not many people currently utilize per driver parameters
with driverfs (i.e. driver_create_file() and
driver_remove_file()).