> On Thu, Feb 13, 2003 at 08:29:31AM -0700, Justin T. Gibbs wrote:
>> > 
>> > Note that the processor thinks it's been asked to do an *overlapped*
>> > command (you know, that abortive attempt at TQ before the SCSI folks
>> > invented TQ?).  Something is deeply, deeply wrong here; data corruption 
>> > by script bugs or some kind of DMA mistake, perhaps?
>> 
>> More likely, NetBSD is still using a very ancient port of the aic7xxx
>> driver.
> 
> Well, yes, obviously.  However, did your code ever _intentionally_ do
> overlapped commands?  If not, something is deeply wrong somewhere, likely
> as I suggested, n'est-ce pas?

No.  Overlapped commands were never an intential feature of the driver.
BTW, your definition of overlapped commands, at least as meant by a
target returning ASC/ASCQ 0x4E,0x00, is a little off.  It means we
tried to use the same tag identifier as that used for an already outstanding
command.  Assuming the target is correct, it is a very serious problem.

--
Justin