protection questions

Gerry.Houlder at seagate.com Gerry.Houlder at seagate.com
Mon Dec 4 16:26:35 PST 2006


* From the T10 Reflector (t10 at t10.org), posted by:
* Gerry.Houlder at seagate.com
*
GAH: I have added comments in line.
	     Gary.Franco at Emule						   
	     x.Com							   
	     Sent by:							To 
	     owner-t10 at t10.org	       <t10 at t10.org>			   
	     No Phone Info						cc 
	     Available							   
								   Subject 
				       protection questions		   
	     12/04/2006 12:08						   
	     PM 							   
Since the following is described in SBC3r07 I have a couple stupid
questions
4.17.2.3 Type 1 protection
a) defines the content of the LOGICAL BLOCK GUARD field;
b) does not define the content of the LOGICAL BLOCK APPLICATION TAG field;
and
c) defines the content the LOGICAL BLOCK REFERENCE TAG field.
4.17.2.4 Type 2 protection
a) defines the content of the LOGICAL BLOCK GUARD field;
b) does not define the content of the LOGICAL BLOCK APPLICATION TAG field;
and
>GAH: The application tag field is defined in the two related fields of the
32 byte CDB commands. All subsequent blocks of that command shall contain
the same value. "Does not define the content.." means the 32 byte CDB
defines the content, not some other content prediction algorithm.
c) defines, except for the first logical block addressed by the command,
the content of the LOGICAL BLOCK REFERENCE TAG field.
>GAH: The first block is defined in related field of 32 byte CDB commands
and this initial value is incremented for subsequent blocks transferred in
that command. This "incrementing behavior" is the defined content
prediction algorithm for the reference tag.
4.17.2.5 Type 3 protection
a) defines the content of the LOGICAL BLOCK GUARD field within the logical
blocks of the data-in buffer and/or data-out buffer;
b) does not define the content of the LOGICAL BLOCK APPLICATION TAG field;
and
c) does not define the content of the LOGICAL BLOCK REFERENCE TAG field.
My confusion is as follows:
1)	when the protection word says to check the LOGICAL BLOCK
APPLICATION TAG, is that imply that it is only checked for Type 2
protection when the 32-byte CDBs are used? And for Type 1 / Type 3
protection it should be ignored altogether even though the WRPROTECT, etc.
says to check it.
>GAH: If you assume that you only have knowledge of things from definitions
in the standard, this is a true statement. Type 3 was created specifically
to allow vendor specific definitions of the APPLICATION TAG and REFERENCE
TAG fields, meaning both ends need to know the definition in order for them
to do any checking. If you don't have this built-in vendor specific
knowledge, no checking is possible and type 3 is not very useful.
2)	Does this logic hold as well when the protection type specifies
that these fields are not defined?
>GAH: Yes, you can't be expected to check something if you don't know the
definition.
3)	Or only when Type 2 protection is defined and the APPLICATION TAG
OWNERSHIP is set to one?
>GAH: When Tag Ownership is set to 1, the application client owns the tag
and the other devices can only check the value if the application client
has somehow told them what to expect. Type 2 with 32 byte CDBs is the only
case (within the standard) where the application client tells the other
devices what to expect.
>GAH: Are you concerned that even the HBA can't predict or check the
returned information during Read opearations? This knowledge must be gained
|from the application client itself and stored in local registers or
something like that.
This information changes the BlockGuard profile definitions somewhat.
Please respond ASAP,
Thanks in advance.
__________________________________
Gary Franco Consultant Engineer
Emulex Network Systems
972-671-7433 Dallas Office
972-671-7435 Dallas Office Fax
720-652-6387 Longmont Colorado Office
720-494-1817 Longmont Colorado Office Fax
972-839-5694 Cell Phone
Today's Quote
The mind is like a parachute. It doesn't work unless it's open.
-Unknown
*
* 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