FCP Linked Cmds and Xchng ID

Joseph C. Nemeth jnemeth at concentric.net
Fri Feb 6 09:54:45 PST 1998


* From the T10 (formerly SCSI) Reflector (t10 at symbios.com), posted by:
* "Joseph C. Nemeth" <jnemeth at concentric.net>
*
The FCP standard makes it pretty clear that linked commands are all part of
a single FC exchange, and therefore presumably should have the same exchange
identifier.

However, the command linking is a high-level concept (FC-4), and so far as I
can tell, affects in a practical way only high-level protocol operations
(command queuing, status response), and can be determined only by parsing of
the high-level protocol (the Link bit of the Control byte of the CDB, or the
Intermediate/IntermediateCM status). The exchange identifier, by contrast,
is a lower-level concept (FC-2).

In many (most? all?) implementations, with highly automated FCP chipsets,
these two levels don't even know about each other, as it should be. The
chipset certainly should not be required to parse the CDB looking for the
link bit or the status value, while the chipset software drivers have no
control whatsoever over the exchange identifier. Therefore, in practice,
linked commands are going to end up having different exchange identifiers,
and thus will technically be different exchanges. In FCP, I don't see that
this makes a whole lot of difference.

It makes sense to say that linked commands are all part of the same task --
that's the whole point of linked commands. On a parallel bus, where bus
management is part of the higher-level protocol, it also makes sense to say
that linked commands have an effect on the bus management, specifically
avoiding the bus free phase and arbitration/selection process between the
end of one command and the beginning of the next linked command. But with
FCP, where arbitrary multiplexing of multiple exchanges is already built
into the lower levels of the protocol, making the linked commands part of
the same exchange seems superfluous, and almost certainly isn't going to
happen in practice anyway.

Or did I miss something?

Joseph C. Nemeth, Precision Algorithms

*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com




More information about the T10 mailing list