I conneced the barracuda (and the Archive Python) to the i486 here
which has an AHA-1542CF.  The disk worked fine.  The tape drive still
gave "incorrect lengt indicator" errors on read (kern/5647).

So the problem with sd3 on the P166 relates to the AHA-1542B
controller.  I re-enabled parity checking on aha2, but sd3 still does
not respond - despite probing ok.

Note sd3 is the only disk on aha2, the only other device is the DDS
tape drive at target 4.  There is no BIOS enabled on aha2 as it is
only a secondary controller, and the DMA speed is set to 5.7M/s which
is the factory default.

The above exercise rules out the controller hardware as an issue in
the tape problem (kern/5647), but points to the controller for the
disk hang.  I'll try setting the disk to target 0 and if that does not
help - replace the 1542B with the 1542CF.

--sjg