SPI-4 rev. 9 table 85 incorrect

Elliott, Robert Robert.Elliott at COMPAQ.com
Mon Feb 4 16:54:07 PST 2002


* From the T10 Reflector (t10 at t10.org), posted by:
* "Elliott, Robert" <Robert.Elliott at COMPAQ.com>
*
Good catch.

Short page
==========
SPI-4 Table 85, since revision 8, has the PROTOCOL IDENTIFIER field in
byte 2 of the protocol-specific target port control mode page.  This
matches SPC-3 and FCP-2, which have used byte 2 for a long time.

SPI-3, and SPI-4 prior to revision 8, had it in byte 3.

00-396r2, the proposal to add the long pages, was written before SPI-4
revision 8.  It recommended SPI-4 use byte 3, not changing its location.
It also recommended SPC-3 also use byte 3 when describing the generic
short format (copying the table from SPI-4).  This would change SPC-3,
which currently uses byte 2.  00-396r2 has not yet been incorporated in
SPC-3.

In summary:
SPI-3: byte 3
SPI-4 revision 0 through 7: byte 3
SPI-4 revision 8 and 9: byte 2 (byte 3 is reserved)
SPC-2/3: byte 2 (bytes 3..n are protocol specific)
FCP-2: byte 2

Did any SPI-3 devices implement this page?

Assuming they did, I suggest we change SPI-4 back to using byte 3. Add a
note to SPI-4 warning that it does not comply with the suggested
standard layout in SPC-3.  SPC-3 could carry a similar warning.

If they did not, we can go ahead and change SPI-4 to match the others.

I'm sure we cannot change FCP-2 - it was the first protocol to use this
page and I assume there are many target ports implementing it.

It's nice but probably not crucial that the mode page layout match
across protocols. Applications should have some other way to tell what
the target port protocol is without depending on the PROTOCOL IDENTIFIER
field in this mode page (or the LUN control mode page) - that's a job
for a command like INQUIRY.  FCP-2 and SPI-4 are the only protocols with
this mode page so far, and the whole page is optional anyway, so it's
not a reliable indicator of the target port protocol.

Long page
=========
Less important is the long page layout. Any SPI-4 implementations using
this are not shipping yet, FCP-2 doesn't use it, and SPC-3 hasn't
incorporated it yet.  

It would be nice if the format is common across protocols. Using byte 4
makes it like SPC-2/FCP-2's short page; byte 5 is like SPI-3's short
page.

Long format:
SPI-4: byte 5
SPC-3 proposed: byte 4

If SPI-4's short format changes back to SPI-3, then the long format will
line up.
If SPI-4's short format remains, I suggest leaving its long format
unchanged too (avoid another rev of SPI-4).  

In either case, change SPC-3 to match the final SPI-4.
---
Rob Elliott, Compaq Server Storage
Robert.Elliott at compaq.com



> -----Original Message-----
> From: Gerry.Houlder at seagate.com [mailto:Gerry.Houlder at seagate.com]
> Sent: Monday, February 04, 2002 5:04 PM
> To: t10 at t10.org
> Subject: SPI-4 rev. 9 table 85 incorrect
> 
> 
> * From the T10 Reflector (t10 at t10.org), posted by:
> * Gerry.Houlder at seagate.com
> *
> When I look at SPI-4 rev. 9:
> 
> Table 85 is the port control page short format. I see that 
> bytes 2 & 3 are
> defined as protocol identifier byte then reserved byte, 
> respectively. I
> believe the original proposal that was approved had these bytes in the
> order reserved byte then protocol identifier field, 
> respectively. Table 85
> has swapped them for no apparent reason.
> 
> Table 86 is the port control page long format. In this table 
> bytes 4 & 5
> are equivalent to bytes 2 & 3 in table 85. These have the bytes in the
> order reserved byte then protocol identifier byte, respectively.
> 
> I strongly believe Table 85 and Table 86 should line up with 
> each other.
> They should both use the order reserved byte then protocol 
> identifier byte
> or vice versa, but not one of each. Since one table must 
> change to make the
> tables agree, I propose changing table 85. Then the standard 
> will agree
> with Seagate's implementation. Do any other implementators 
> have an opinion
> on this? Please speak up!
> 
> *
> * 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