[T10] REVIEW T10/22-096r0 For our upcoming T10 meeting week - RE: Write and Verify command operation ambiguity

Bill Martin bill.martin at samsung.com
Wed Sep 7 17:28:35 PDT 2022


I have uploaded a proposal for this ambiguity that I would like review of at the next T10 plenary meeting. I am calling it out to you now so that you can review prior to the meeting for a vote to approve for inclusion in the next revision of SBC-5.

Thank you

Bill Martin

Chair INCITS T10
Co-Chair SNIA Technical Council

Chair SNIA CMSI

NVMe Board of Directors
SSD I/O Standards
Samsung Semiconductor, Inc.
Cell (408) 499-1839

From: Rich Deglin
Sent: Thursday, June 16, 2022 12:39 PM
To: Elliott, Robert (Servers) <elliott at hpe.com>; 'John Geldman' <John.Geldman at kioxia.com>; Bill Martin <bill.martin at samsung.com>; T10 Reflector <t10 at t10.org>
Subject: RE: Write and Verify command operation ambiguity

That's interesting. Regardless of the intention, obvious questions arise:

1. why was this not stated explicitly in the standard?
2. what is the behavior of older and current block devices?

It's quite possible that this subtle requirement has been missed and the standard misinterpreted. If indeed many devices have used WRPROTECT as if it were also VRPROTECT, then are they misbehaving, or is the standard "wrong" ? Does industry practice take precedence?

From: Elliott, Robert (Servers) [mailto:elliott at hpe.com]
Sent: Thursday, June 16, 2022 12:04 PM
To: 'John Geldman' <John.Geldman at kioxia.com<mailto:John.Geldman at kioxia.com>>; Bill Martin <bill.martin at samsung.com<mailto:bill.martin at samsung.com>>; T10 Reflector <t10 at t10.org<mailto:t10 at t10.org>>
Cc: Rich Deglin <r.deglin at samsung.com<mailto:r.deglin at samsung.com>>
Subject: RE: Write and Verify command operation ambiguity

The protection information proposal purposely excluded the verify operation in this command from checking the protection information with this reasoning:
"As a result of the way the write part of the write and verify command works it is treated just like any other write. And since the data that is verified is the same data that was just written it was felt there is no need to do anything except the byte check (if the bit is set that is). However, the byte check should be for everything that was written not just the user data."

The COMPARE AND WRITE command from a few years later also took that approach, not doing any checking on its initial read operation.


From: t10-bounces at t10.org<mailto:t10-bounces at t10.org> <t10-bounces at t10.org<mailto:t10-bounces at t10.org>> On Behalf Of John Geldman
Sent: Sunday, June 5, 2022 12:48 AM
To: Bill Martin <bill.martin at samsung.com<mailto:bill.martin at samsung.com>>; T10 Reflector <t10 at t10.org<mailto:t10 at t10.org>>
Cc: Rich Deglin <r.deglin at samsung.com<mailto:r.deglin at samsung.com>>
Subject: Re: [T10] Write and Verify command operation ambiguity

Bill,

The VRPROTECT does need to get referenced, but I'd suggest:
a)      renaming the field in the Write and Verify commands "wvprotect";
b)      use the value of wvprotect as wrprotect in the write operation of this command; and
c)       use the value of wvprotect as vrprotect in the verify operation of this command.

Or something like that.

Thank you,
John

From: t10-bounces at t10.org<mailto:t10-bounces at t10.org> <t10-bounces at t10.org<mailto:t10-bounces at t10.org>> On Behalf Of Bill Martin
Sent: Friday, June 3, 2022 11:23 AM
To: T10 Reflector <t10 at t10.org<mailto:t10 at t10.org>>
Cc: Rich Deglin <r.deglin at samsung.com<mailto:r.deglin at samsung.com>>
Subject: [T10] Write and Verify command operation ambiguity

In the WRITE AND VERIFY command it specifies that the WRPROTECT field is as defined in the WRITE(10) command. This command specifies how data is to be written to the media. It does not specify anything about the verify operation. I believe that there should be text that indicates: For the verify operation the WRPROTECT field shall be interpreted as the VRPROTECT field in the VERIFY(10) command.

If there is consensus on this, I will generate a proposal for the next T10 CAP AdHoc to review.

Bill Martin
Chair INCITS T10
Co-Chair SNIA Technical Council
Chair SNIA CMSI
NVMe Board of Directors
SSD I/O Standards
Samsung Semiconductor, Inc.
Cell (408) 499-1839

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.t10.org/pipermail/t10/attachments/20220908/0e0a38ee/attachment.html>


More information about the T10 mailing list