solid state mode/ vpd/ event work

Ralph O. Weber roweber at IEEE.org
Tue Dec 21 18:31:02 PST 2004


* From the T10 Reflector (t10 at t10.org), posted by:
* "Ralph O. Weber" <roweber at ieee.org>
*
Dear Mr. LaVarre,

The brief list quoted in the most recent message is
(as best as I understand it) the advice given to Mr.
Hellmold by the CAP working group in Austin.

Specifically:

1) Data that the host wants to set belongs in a Mode page.

2) Unchangeable data belongs in a VPD page.

3) Data that needs to be polled (i.e., the Health Indicator)
   belongs in a Log page (with threshold enabled), or polled
   with a Request Sense command, or polled with a TBD
   specialized command.

My recollection further suggests that the CAP group has yet
to come to grips with the best way to handle the Health
Indicator. Further wrestling with specifics appears to be
necessary.

However, like Michael Banther, I am totally flummoxed by
statements in the original posting.

I have no idea what "Massively distributed hosts" has to
do with the proposal. Is the goal that only Distributed
Cluster systems with large numbers of processing nodes
are the only initiators that care about the solid state
disk information?

What is "...from any PDT x 0E 07 05 04 00 device"? What is
a PDT? Sorry, but PDT is not in my acronyms glossary.

I fail to follow the logic that puts testing the mode page
first and retrieving the VPD page second. Why not perform
the "test" by retrieving the VPD page? Most of the really
interesting data will be in the VPD page (at least that was
the impression I got from Mr. Hellmold's presentation). So,
why not go for the gold up front?

Finally, I side 100% with Michael regarding the following
statement "...inventing a new use for EVPD and yet another
form of polling, ..." To me, this looks like a proposal
to use a VPD page to contain data that is polled. Since
such a plan would clearly violate the "Unchangeable data
belongs in in a VPD page" guideline, I flat out cannot
follow the proposal.

All the best,

.Ralph

Pat LaVarre wrote:

> I thought that's what the committee told me to do?  On review now, I 
> see the meeting notes only say "can be represented" or "needs to be 
> represented".  I took that English as a less confrontational way of 
> saying "should be represented".  More specifically, I thought I was 
> being told to:
>
> 1) Represent "Data that the host wants to set" "in a Mode page".
>
> 2) Represent "unchangeable data" "in a VPD page".
>
> 3) Represent "data that needs to be polled (e.g., the health 
> indicator)" "by a Log page (with unit attention)" or "polling with 
> Request Sense", or "polling with a specialized command".
>
> All that quoted text above comes from what "George Penokie noted" in 
> the CAP WG Nov minutes:
> http://www.t10.org/ftp/t10/document.04/04-367r1.htm
>
> The parts not quoted came from me think I was being shown the light of 
> the correct way to proceed.  Have I misunderstood?  Help?
>
> Thanks again in advance, Pat LaVarre
>
> -----Original Message-----
> From: owner-t10 at t10.org on behalf of Banther, Michael
> Sent: Mon 12/20/2004 3:28 AM
> To: t10 at t10.org
> Subject: RE: solid state mode/ vpd/ event work
>
> I do not understand your comment below, 'By inventing a new use for EVPD
> ....'  Do you intend to define a new VPD page to hold the 'read-only
> descriptive data?'
>
> Regards,
> Michael Banther
> Hewlett-Packard Ltd.
> Telephone +44 (117) 312-9503
>
>
> -----Original Message-----
> From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of Pat
> LaVarre
> Sent: 17 December 2004 17:36
> To: t10 at t10.org
> Subject: solid state mode/ vpd/ event work
>
> * From the T10 Reflector (t10 at t10.org), posted by:
> * "Pat LaVarre" <plavarre at lexarmedia.com>
> *
> Hello again, remember Me?  We've now volunteered me to redraft the
> (04-362r0.pdf from Lexar) proposal for vpd/ mode/ etc. definitions that
> help the host tune its use of solid-state storage.  I hope to see ya'll
> in person at the next CAP WG.
>
> I'm working face to face with people who attended the November CAP WG,
> also from the written minutes that I quote below.  For example, I plan
> to begin by answering the request to divide the solid-state data into
> three categories: a) rewritable settings data, b) read-only descriptive
> data, and c) pollable event data.
>
> But if I may, I'd first like to take a moment to confirm I understand
> the overall approach.  I hear we the committee saying Four things:
>
> ---
>
> 1) Massively distributed hosts may choose to adopt this new standard by
> beginning with the negligible risk of trying to fetch the newly standard
> solid-state mode page, from any PDT x 0E 07 05 04 00 device.
>
> 2) With devices that do not make the new page available, the system will
> survive as it already does when the host looks for other mode pages with
> less than universal support, such as the code x05 HDD C:H:S page or the
> x2A C/DVD media compatibility page.
>
> 3) After confirming a particular device does make the new page
> available, the host may proceed to fetch the read-only descriptive data
> by trying op x12 INQUIRY with the byte 1 mask x01 EVPD bit of the CDB
> set.  The host may alter the rewritable settings via Mode Select, in
> particular choosing to enable the new unit attention or not.  Thereafter
> the host may poll for events.
>
> 4) By inventing a new use for EVPD and yet another form of polling, we
> might break (intrinsically fragile) opaque bridge chips, but that's a
> cost the device folk who wish to establish this new standard can agree
> to pay before shipping a device that does make the new page available.
>
> ---
>
> Is that our thinking?
>
> Thanks in advance, hope this helps, Pat LaVarre
> http://www.t10.org/ftp/t10/members.txt
>
> --- Background ---
>
> http://t10.org
>
> http://t10.org/t10_mins.htm
>
> 04-367r1 Minutes of CAP Working Group - Nov 9-10, 2004
> http://www.t10.org/ftp/t10/document.04/04-367r1.htm
>
> 5.3.3  SBC-2: Proposal for USB Solid State Drive Mode Sense
> specification (04-362r0) [Furuhjelm]
> ftp://ftp.t10.org/t10/document.04/04-362r0.pdf
>
> Steffen Hellmold presented a proposal for defining a mode page for USB
> solid-state drives (04-362r0). The group recommended using a VPD page
> instead of a mode page for most of the proposed data, as well as other
> changes in the proposal.
>
> George Penokie noted that the proposal contains three types data:
>
>
> o      Unchangeable data that can be represented in a VPD page
>
> o      Data that the host wants to set that can be represented in a Mode
> page
>
> o      Data that needs to be polled (e.g., the health indicator) which
> needs to be represented by a Log page (with unit attention), polling
> with Request Sense, or polling with a specialized command
>
> Steffen agreed to prepare a revised proposal for consideration at the
> next meeting.
>
> ...
>
> ---
>
> *
> * For T10 Reflector information, send a message with
> * 'info t10' (no quotes) in the message body to majordomo at t10.org
>



*
* 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