SCSI/ATA Translation of Command Duration Limit (T10/15-093)

Paul Suhler Paul.Suhler at hgst.com
Sun Jun 14 08:02:03 PDT 2015


Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1506141_f.htm">HTML-formatted message</a>

Thanks to Gerry, Fred, and Hatless Ralph for all the suggestions.  To see
what I made of this, please have a look at:
   http://www.t10.org/cgi-bin/ac.pl?t=d&f=15-093r1.pdf
It appears that I will not be at the July meeting, and I would not ask Joe
Breher to do anything more than show the document for a vote on approving the
document as is.  So, if you have any changes to suggest then please let me
know and I may revise the document before the meeting.
Thanks,
Paul
Paul A. Suhler, PhD
Research Staff Member
HGST Research
paul.suhler at hgst.com
o: 949-757-3432
m: 949-241-6443
3001 Daimler St.
Santa Ana, CA 92705-5812
www.hgst.com<<a href="http://www.hgst.com/&gt">http://www.hgst.com/&gt;
From: owner-t10 at t10.org<mailto:owner-t10 at t10.org> [mailto:owner-t10 at t10.org]
On Behalf Of Knight, Frederick
Sent: Friday, June 12, 2015 9:13 AM
To: Gerry Houlder
Cc: T10 E-mail Reflector (t10 at t10.org<mailto:t10 at t10.org>)
Subject: RE: SCSI/ATA Translation of Command Duration Limit (T10/15-093)
We had some similar issues with the power timers.  We used rounding there
too.
See SAT 4r01 - 10.1.11.2.2.5 MODE SELECT command timer field translation for
EPC and table 102.  It specifies the formula for the translation and how
rounding is used and reported.
		Fred
From: owner-t10 at t10.org<mailto:owner-t10 at t10.org> [mailto:owner-t10 at t10.org]
On Behalf Of Gerry Houlder
Sent: Thursday, June 11, 2015 10:22 AM
Cc: T10 E-mail Reflector (t10 at t10.org<mailto:t10 at t10.org>)
Subject: Re: SCSI/ATA Translation of Command Duration Limit (T10/15-093)
Please remember that a true SCSI device probably doesn't support the entire
minimum to maximum range that is possible, and may not support resolution
down to 1 usec either. I don't remember what the T10 proposal specifies
(i.e., rounding or rejecting) but the SAT-3 behavior ought to be the same.
Also, are ATA devices required to handle the entire expressable range of
time, or is there an ATA mechanism for rejecting/ rounding out of range
values?
In the interest of friendliness to the host (i.e., making it as easy as
possible for a host to determine minimum and maximum acceptable times),
rounding is best. A CHECK CONDITION occurs when rounding happens, so the host
will be aware that the requested time was not acceptable and the device has
made best effort to implement its closest option.
On Wed, Jun 10, 2015 at 7:42 PM, Ralph Weber
<Ralph.Weber at wdc.com> wrote:
As David Black would say at a time like this ... "With my chair's hat totally
removed ..."
Option b for times larger than 90 seconds and smaller than 1 millisecond
Option a for all other conversions
All the best,
.Ralph
________________________________
From: owner-t10 at t10.org<mailto:owner-t10 at t10.org>
[owner-t10 at t10.org<mailto:owner-t10 at t10.org>] on behalf of Paul Suhler
[Paul.Suhler at hgst.com]
Sent: Wednesday, June 10, 2015 7:23 PM
To: T10 E-mail Reflector (t10 at t10.org<mailto:t10 at t10.org>)
Subject: SCSI/ATA Translation of Command Duration Limit (T10/15-093)
Hi, everyone.
 I would like to solicit opinions on how to do convert the command duration
limit times expressible in SCSI into the isochronous command completion (ICC)
times expressible in ATA.  The problems are:
 a)  SCSI can express times as low as 1 microsecond, while ATA only goes down
to ten milliseconds.
b)  SCSI can express times as large as 32k – 1 seconds, while ATA only goes
up to 64 seconds.
c)  In some ranges of times, SCSI can express values with a finer granularity
that ATA can.
 There are two possible solutions for the SATL:
a)  Reject any MODE SELECT command that specifies a value that does not map
precisely to an ATA value; or
b)  Round the SCSI value to the nearest ATA value.
Rounding (b) will allow the ATA device to make a best effort at setting a
command duration limit.  On the other hand, the value used may be as much as
a factor of 10,000 larger than that specified (1 usec), or as little as
1/500th of that specified (32k-1).
I’ll also note that neither ATA nor SCSI has a mechanism to report the
actual range of values supported by a particular device, and neither ACS-4
nor SPC-5 specifies what to do in case of an unsupported value.
Straw poll time.  Vote early and vote often!
Thanks,
Paul
Paul A. Suhler, PhD
Research Staff Member
HGST Research
paul.suhler at hgst.com
o: 949-757-3432<tel:949-757-3432>
m: 949-241-6443<tel:949-241-6443>
3001 Daimler St.
Santa Ana, CA 92705-5812
www.hgst.com<<a href="http://www.hgst.com/&gt">http://www.hgst.com/&gt;



More information about the T10 mailing list