[1671] in linux-scsi channel archive
Re: ATN Initiator Detected Error
daemon@ATHENA.MIT.EDU (Gerard Roudier)
Wed Apr 2 14:12:44 1997
Date: Wed, 2 Apr 1997 20:02:37 +0000 (GMT)
From: Gerard Roudier <groudier@club-internet.fr>
To: Marc SCHAEFER <schaefer@alphanet.ch>
cc: linux-scsi@vger.rutgers.edu
In-Reply-To: <199704020640.IAA20343@vulcan.alphanet.ch>
On Wed, 2 Apr 1997, Marc SCHAEFER wrote:
> > With ATN, the controller ask the drive to switch to message out phase.
> > When that happens in DATA IN phase, that looks like a nice scsi parity
> > error detected by the controller in my opinion.
>
> There is a specific message for this. I don't understand the
> ``initiator detected error''.
It is this one. The message MESSAGE PARITY ERROR means that a scsi parity
error has been detected while receiving a message. It can be send by any
scsi device, initiator or target, but data is not message in scsi semantic.
So, when a scsi parity error is detected in DATA IN phase, sending a
message MESSAGE PARITY ERROR is a protocol error and the target generally
indicates this by releasing the bus immediately (unexpected disconnect)
in order to inform the initiator of this protocol error.
AFAIK, SCSI standards recommendation is the initiator send a INITIATOR
DETECTED ERROR message in order to notify a target of scsi parity error in
DATA IN phase. So, in my opinion, the behaviour of the 154X controller is
quite correct (Could it be different?).
Regards, Gerard.
