SOP Device model

Joe Breher joe at lingua-data.com
Thu Apr 19 12:36:32 PDT 2012


Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1204191_f.htm">HTML-formatted message</a>

Further fun with sop-r02d:
Item 1)
4.1.2.5.4.1 appears to have become the victim of a cut-n-paste drive-by
shooting:
If the bridge is contained in an SOP device that contains one or more SOP
ports that are SCSI initiator ports, then the Bridging Device Server class:
Editor’s Note 7: These 2 xrefs do not look right...
a)  received from the SCSI initiator port (see 4.1.2.9xxx);
b)  invokes the Request Data-Out operation (see 6.4.9)xxx? in the SOP target
port for each data-out request received from the SCSI initiator port; and
c)  invokes the Terminate Data Transfer operation (see 6.4.11) in the SOP
target port for each terminate data transfer request received from the SCSI
initiator port. 
Item 2)
Do we need to define the term 'far side'? The term is used in the definition
of nexus identifier, but is not itself defined outside of this reference in
4.3.2: "The bridge shall maintain a non-zero 16-bit nexus identifier for
every I_T nexus in which it is able to participate on the far side of the
bridge (i.e., involving a local port and a remote port)." When a bridge's
local port is acting as a target port, is it on the 'far side' of the bridge?
How about in the case where all of a bridge's ports are SOP, and are capable
of operating as either a target port or an initiator port? 
Item 3) 
Table 24 contains a field "TARGET INITIATOR BRIDGE PRESENT". This is
undefined in the text, but the text does define a field "SOP INITIATOR DEVICE
BRIDGE PRESENT", which does not appear in the table. Shall I assume the table
is in error and that both should be "SOP INITIATOR DEVICE BRIDGE PRESENT"?
Item 4)
|From whence are the SCSI protocol services operations inherited?
- 4.1.2.4.4.1 claims that "This operation is inherited from the SCSI Target
Port class (see SAM-5)." 
- 4.1.2.4.4.2 claims that "This operation is inherited from the SCSI Target
Port class (see SAM-5)."
It appears to me that these are actually inherited from the SCSI Initiator
Port, correct?-
Joe Breher
President,
lingua data (a dba of q music inc)
(478) 2-Breher
(478) 227-3437
On Apr 13, 2012, at 9:21 AM, Elliott, Robert (Server Storage) wrote:
> The capitalization convention is:
> - UML class names are Mixed Case (e.g., SCSI Target Port class)
> - names of objects are lower case (e.g., SCSI target port)
>  
> I agree with your new wording suggestion, except both the old and new
wording mix up classes and objects.  These sentences should just discuss the
objects:
>  
> If one or more SOP ports in a SOP device is a SCSI initiator port, then the
SOP device is a SOP initiator device.
> If one or more SOP ports in a SOP device is a SCSI target port, then the
SOP device is a SOP target device. 
>  
>  
> From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf Of Joe Breher
> Sent: Thursday, 12 April, 2012 1:07 PM
> To: T10 Reflector
> Subject: SOP Device model
>  
> Reference sop-r02d section 4.1.2.3
>  
> The grammar employed in describing Figure 9 is misleading - specifically,
it implies that there is a single SOP Port associated with any given SOP
Device. This is in clear disagreement with the multiplicity upon the UML
diagram itself, so I believe this is a mere grammatical issue.
>  
> The text currently reads:
> If the SOP port in a SOP Device class is a SCSI initiator port, then the
SOP Device class is a SOP initiator device.
> 
> If the SOP port in a SOP Device class is a SCSI target port, then the SOP
Device class is a SOP target device. 
> 
> I believe it should read something like:
> If one or more SOP ports in a SOP Device class is a SCSI initiator port,
then the SOP Device class is a SOP initiator device.
> 
> If one or more SOP ports in a SOP Device class is a SCSI target port, then
the SOP Device class is a SOP target device. 
> 
>  
> In tracing this, I discover also that sam5r10 uses multiple forms of
capitalization for 'SCSI Target Port', 'SCSI target port', 'SCSI Initiator
Port', and 'SCSI initiator port'. I am left with the queasy feeling of being
unsure as to whether or not a 'SCSI Target Port' is the same thing as a 'SCSI
target port', and whether a 'SCSI Initiator Port' is the same thing as a
'SCSI initiator port'. Anyone care to speak to this?
>  
> Joe Breher
> President,
> lingua data (a dba of q music inc)
> (478) 2-Breher
> (478) 227-3437
>  
>  



More information about the T10 mailing list