14-176r0 - Including SIZE OF UNMAP REGION in UNMAP CDB

Dave Landsman Dave.Landsman at sandisk.com
Thu Jul 3 10:10:37 PDT 2014


Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1407032_f.htm">HTML-formatted message</a>

Gerry,
This is about command execution efficiency.
Hosts send UNMAP commands fairly frequently with all sizes ranging from
smallest to the Maximum UNMAP LBA count advertised by the drive.  Device
servers usually optimize the handling of Immediate write commands with
smaller payloads by avoiding the data pull-out handshake using Xfer Ready
mechanism.
In much the same way, the ability to decode the size of the total UNMAP
region from the CDB will help Device server to optimize the UNMAP processing
if they don’t have to pull-in the parameter list first. Dave
From: Gerry Houlder [mailto:gerry.houlder at seagate.com]
Sent: Thursday, July 03, 2014 07:30
To: Dave Landsman
Cc: t10 at t10.org
Subject: Re: 14-176r0 - Including SIZE OF UNMAP REGION in UNMAP CDB
I am wondering what problem this will solve.
The drive can currently report, in VPD page B0h (Block Limits), the Maximum
Unmap LBA Count and Maximum Unmap Descriptor Count. The drive should be able
to accept any count that is less than these values or else the drive is
reporting a count that is too high. If a host takes notice of the VPD page
values and stays within the limits, then there should be no need for this
additional check (which the drive has to do by transferring all of the
descriptors and calculating whether the lengths of all of the descriptors
matches the new parameter).
The drive can do this instead -- calculate the length of all descriptors and
compare it to the maximum Unmap LBA count in the VPD page instead. This
doesn't require any new parameters.
Are you concerned about catching a rogue host that isn't paying attention to
the VPD page values and sends unmap lengths that are too large anyway? Those
hosts will just put zero in the size of unmap region field anyway, so adding
this check won't help with that.
On Wed, Jul 2, 2014 at 4:33 PM, Dave Landsman
<Dave.Landsman at sandisk.com> wrote:
It is advantageous in certain situations if a device knows the total length
of a region being requested to be unmapped, before having to process the
parameter data which describes the UNMAP data in detail.
An enhancement to the UNMAP command is proposed.  Please see   
http://www.t10.org/cgi-bin/ac.pl?t=d&f=14-176r0.pdf.
dave
________________________________
PLEASE NOTE: The information contained in this electronic mail message is
intended only for the use of the designated recipient(s) named above. If the
reader of this message is not the intended recipient, you are hereby notified
that you have received this message in error and that any review,
dissemination, distribution, or copying of this message is strictly
prohibited. If you have received this communication in error, please notify
the sender by telephone or e-mail (as shown above) immediately and destroy
any and all copies of this message in your possession (whether hard copies or
electronically stored copies).



More information about the T10 mailing list