>One control block per target is insufficient if you have a full complement
>of targets attached and access those simultaneously (like in a ccd(4) array).
>We (now) allocate (somewhat arbitrarily) three per target.
>Noticed by Marshall Midden.
* Handle message retransmissions and partially sent messages correctly.
* Make sure we clear ATN after the last message is sent.
* Do the right thing if the target initiates negotiation for async mode
after having negotiated sync mode.
* Fix some cases where we set ATN with no message queued, or schedule a
message without setting ATN.
* Issue a REQUEST SENSE after an unexpected disconnect, per SCSI spec.
* Fix abort handling in a number of cases.
* Recognize selection timeouts better (to speed up probing).