T10/04-262r0

Curtis Stevens curtis.stevens at wdc.com
Thu Aug 12 13:56:21 PDT 2004


* From the T10 Reflector (t10 at t10.org), posted by:
* "Curtis Stevens" <curtis.stevens at wdc.com>
*
I agree...

I think we are getting to the purpose of the CDB transport mechanism at this
point.  My current vision is manufacturing and maintenance.  This would mean
that SCSI would be the normal mode of operation for reading and writing the
media.  However, in the case of a bridged (P/S)ATA device (USB, 1394, SAS,
etc) you may want to tech support the device by issuing a SMART command.
Using CDB pass-through, I think this can be done using existing standard
driver pass-throughs.  It would also allow device failure diagnosis without
opening the box and removing the ATA device.  Our major use will be for
vendor specific commands...

At this point, I do not really see this being used in a performance
environment.  So, the devices will already have to deal with the SCSI
presentation issue.  The one case I can see is if there are many drives
presented as one device by the SCSI-ATA bridge.  Then there would need to be
a mechanism for addressing the individual ATA drives.  I think in this
environment  (which is starting to look like a RAID), there is a bigger
problem to solve.


------------------------------------------------
Curtis E. Stevens
20511 Lake Forest Dr.  #C 214-D
Lake Forest, Ca. 92630
 
Phone: 949-672-7933
Cell: 949-307-5050
E-Mail: Curtis.Stevens at wdc.com

 -----Original Message-----
From: 	Jeff Garzik [mailto:jgarzik at pobox.com] 
Sent:	Wednesday, August 11, 2004 7:28 PM
To:	Curtis Stevens
Cc:	Sheffield, Robert L; t10 at t10.org; Forum at t13.org
Subject:	Re: [t13] SAT: Reminder - Teleconference Thur 8/12 9:00 AM
PDT

Curtis Stevens wrote:
>             Just a heads up, I posted a proposal (04-260.PDF) for 
> documenting how a SCSI command maps to an ATA command.

Another quick comment.  Sorry I didn't include this in the first email.

Quoting...
> The host shall set the DEV bit to 0 and the device shall ignore the DEV 
> bit. This is required because the presence of multiple devices cannot be 
> detected. If multiple devices are present on a single cable, the host 
> may indicate this to the host software by reporting multiple LUNs. If 
> there are multiple LUNs, the bridge shall set the DEV bit to the 
> appropriate value before setting the ATA command register.


I agree with this, but for a reason not stated:

If the SAT translator software exports ATA devices as individual SCSI 
targets (rather than LUNs), then the DEV bit is irrelevant at the ATA 
passthru level.  The low-level ATA driver _must_ set or clear the DEV 
bit as appropriate, and the SCSI I/T nexus provides all the information 
necessary to do so.


And a parting shot:  existing SATA controllers _do_ occasionally present 
SATA devices as master/slave.  Several (Intel, SiS, VIA, others?) have a 
compatibility mode that in current hardware is often enabled by default, 
presenting two SATA devices as a master/slave pair.

	Jeff




*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org




More information about the T10 mailing list