Behavior of SCSI device for START STOP UNIT command
munjal_mistry at mentor.com
Thu Nov 20 21:41:37 PST 2014
Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1411202_f.htm">HTML-formatted message</a>
Thank you for your reply. We are referring JEDEC UFS 2.0 specification
which supports SAM-5, SBC-3 and SPC-4 specification. UFS stands for
"Universal Flash Storage" device which can be embedded into the device
or removable device.
Please find my reply embedded with your queries.
On 11/20/2014 08:41 PM, Gerry Houlder wrote:
> There are a lot of configuration questions that affect the response in
> your situation.
> (1) Are Logical unit 0 and logical Unit 1 two separate disk drives, or
> are they two virtual logical units that are on the same physical
> drive? This determines whether the commands for each LUN are treated
> separately or whether the commands are treated as if they are all for
> the same device.
[MM]: It is a single device and two virtual logical units are there.
> (2) The standard terms for power down options are "idle" or "standby".
> Which of these do you mean by sleep/power down? Recovery behavior is
> usually different for idle versus standby.
[MM]: In terms of UFS, there are four power modes: Active, Idle, Sleep
and Power Down. As per UFS specification, in Sleep and Power Down mode,
only START STOP UNIT and REQUEST SENSE commands are supported.
In sleep mode, power consumption is in between the Active
mode and Power Down mode. Idle mode indicates, there are no commands are
processed by the device and power down mode is the maximum power saving
> (3) What are the setting in the CCF_IDLE and CCF_STANDBY fields of the
> power Condition mode page? These settings determine the recovery
> behavior from idle and standby conditions, respectively, and may
> result in either transitioning to active and processing an incoming
> command or responding with CHECK CONDITION status to an incoming command.
[MM]: In UFS, Power condition mode page is not supported. Power mode can
be changed only with the START STOP UNIT command.
> (4) Disk drives can respond to some commands while in an idle or
> standby condition. The exact list of such commands is vendor specific,
> but the idea is that any command that doesn't require access to the
> disk might be processed properly without the drive having to leave the
> idle or standby condition. READ CAPACITY is in the gray area (might
> not require medium access) where some vendors might treat this as a
> command that is serviced while in a low power condition.
> From you list of choices, either response 1) or 2) is possible. Choice
> 3) should never be possible. Also choice 4) all of the commands
> complete with GOOD status is possible. Commands to a drive that is in
> a low power condition may take longer to complete then if the drive
> were in the active condition.
> On Thu, Nov 20, 2014 at 4:06 AM, Munjal Mistry
> <munjal_mistry at mentor.com <mailto:munjal_mistry at mentor.com>> wrote:
> Hello All,
> I have a query regarding the device behavior for changing the
> power condition during on-going read/write operation.
> Suppose, the device is in active power mode and logical unit 0 is
> processing write command and logical unit 1 is processing read
> command. In between this process, host sends START STOP UNIT
> command which requests to change power mode to sleep/power down
> mode, followed by READ CAPACITY (10) command.
> In this case, which behavior of device is correct?
> (1) Device will wait for the READ and WRITE commands to completion
> and then move to pre-sleep/pre-power down mode and sends CHECK
> CONDITION status for READ CAPACITY (10) command.
> (2) Device will send response with GOOD status for READ CAPACITY
> (10) command and wait for completion of READ-WRITE command and
> then move to pre-sleep/pre-power down mode
> (3) Device stops processing READ-WRITE command and sends CHECK
> CONDITION status for READ-WRITE command and move to
> pre-sleep/pre-power down mode. Then device sends CHECK CONDITION
> status for READ CAPACITY (10) command.
> In-case any of the above mentioned behavior is not correct then it
> would be great if you could let me know the expected behavior of
> the device.
> Thanks in advance.
> Warm Regards,
> Munjal Mistry
More information about the T10