Problem with wording in 10-162r2

Gerry Houlder gerry.houlder at seagate.com
Thu Oct 7 16:01:41 PDT 2010


Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1010070_f.htm">HTML-formatted message</a>

I have an issue with new wording added in the model clause by proposal
10-162r2. I apologize for not having read this more carefully before the
last T10 meeting. These words are added in the description of the
Application Tag interpretation:
if the ATMPE bit in the Control mode page (see SPC-4) is:
a) set to one, then the contents of logical block application tags are
defined by the Application Tag mode
page (see 6.x.x); or
b) set to zero, then the contents of logical block application tags are not
defined by this standard.
First problem: When ATMPE bit is set to zero, Type 2 PI still defines a
method for defining the contents of the Application Tag field (i.e., 32 byte
CDB field). It is wrong to state that the contents are not defined by this
standard. I see that this mis-statement is a carry over from existing words,
but since this proposal is modifying those existing words we should try
harder to get it right.
Second problem: When Type 2 PI is in use and ATMPE is set to one, there is a
conflict between using the APPLICATION TAG field of 32 byte CDBs and using
the mode page to get the values. Use of the mode page has to be prohibited
for type 2 PI (for sure) and maybe type 3 PI even if the ATMPE bit is set to
one. I am still considering whether the note added to the PI checking tables
resolves this conflict or not.



More information about the T10 mailing list