| Here's a message from FreeBSD's SCSI mailing list that looks relevant:

Thanks.  I did upgrade to LYK8 (which was fun, since I had no
MSDOS/drivers to run the Quantum installation stuff with, initially),
but the problem has recurred. :(

| In CAM, the recovery code was able to bring the drive back
| to life by issuing a BUS DEVICE RESET, but I don't know that the older
| aic7xxx and NCR drivers would fair as well.

At a guess, hacking in a BUS DEVICE RESET after noticing the
queue full would be an interim work around?   For now, though,
tagged queueing stays off. 


May 27 20:14:05 sean /netbsd: ahc0 at pci0 dev 8 function 0
May 27 20:14:05 sean /netbsd: ahc0: interrupting at irq 11
May 27 20:14:05 sean /netbsd: ahc0: aic7880 Wide Channel, SCSI Id=7, 16 SCBs
May 27 20:14:05 sean /netbsd: scsibus0 at ahc0 channel 0: 16 targets
May 27 20:14:05 sean /netbsd: ahc0: target 0 using 16Bit transfers
May 27 20:14:05 sean /netbsd: ahc0: target 0 synchronous at 20.0MHz, offset = 0x
8
May 27 20:14:05 sean /netbsd: ahc0: target 0 Tagged Queuing Device
May 27 20:14:05 sean /netbsd: sd0 at scsibus0 targ 0 lun 0: <QUANTUM, XP34550W, 
LYK8> SCSI2 0/direct fixed
May 27 20:14:05 sean /netbsd: sd0: 4341MB, 5899 cyl, 10 head, 150 sec, 512 bytes
/sect x 8890760 sectors

...

May 27 20:51:08 sean /netbsd: sd0(ahc0:0:0): Queue Full
May 27 20:51:09 sean /netbsd: ahc0: WARNING no command for scb 15 (cmdcmplt)
May 27 20:51:09 sean /netbsd: QOUTCNT == 0
May 27 20:58:16 sean syslogd: restart

	Sean.