Actually, I just looked at the code again, and it always does a wakeup in
free_xs unless no one was waiting and there is a driver specific start
routine.  There is, however, a bug which prevents free_xs from doing the
wakeup if you increase the number of openings when someone is waiting.
Fortunately, since the aic7xxx driver increments the opening count only
during the probe for devices (when we're doing polled SCSI I/O), this never
happens.  Just another bug that needs to be squashed in the SCSI code
revamp.

--
Justin T. Gibbs
===========================================
  FreeBSD: Turning PCs into workstations
===========================================