Multiple connections or I_T_x_x nexus
John_Packer at adaptec.com
Tue Feb 25 09:28:23 PST 2003
* From the T10 Reflector (t10 at t10.org), posted by:
* "Packer, John" <John_Packer at adaptec.com>
I also had an issue with this wording, and I think it needs to change. Rob
gave me his interpretation which is (hopefully I'm relaying it correctly):
1) On the wide link you can have multiple connections open for the same I_T
2) The data for an I_T_L_Q (context) can be moved on any of these
connections, following the interlock and ACK rules described.
3) If only one connection is open on PHY A of a wide link (e.g. PHYs A, B,
C, D), then data only moves on PHY A. It does not move on any of PHYs B, C,
or D. There are lots of things that make this a necessity, Rate matching
and Receiver Credits being the most glaring.
4) This discussion really only applies to wide targets attached to wide
initiators (also via wide expanders), because if any point in the connection
path is not wide only one connection can be made.
I'm satisfied with this interpretation, but I'd like to see the
specification wording clarified, and I'm glad I'm not the only one thrown
for a loop on that paragraph!
> -----Original Message-----
> From: Chris.Malakapalli at seagate.com [mailto:Chris.Malakapalli at seagate.com]
> Sent: Tuesday, February 25, 2003 8:21 AM
> To: t10 at t10.org
> Subject: Multiple connections or I_T_x_x nexus
> * From the T10 Reflector (t10 at t10.org), posted by:
> * Chris.Malakapalli at seagate.com
> I have somewhat a related question to the wide port question posted
> earlier. Under section 4.1.9 of SAS rev 3d , page 35, case b states with
> reference to Fig. 20 that:
> "the connections labeled A and C are an example of one target port with
> connections to multiple initiator ports" ...
> The question is can these connections be concurrently active to support
> separate I_T_x_x nexus? Given that there is no provision in the OPEN
> address frame to specify which phy to use for the connection, my feeling
> they cannot be active at the same time, but am hoping someone could
> this for me.
> Also, there seem to be other areas in the spec, where this subject is
> discussed. For instance, according to 7.11.1 on page 177,
> "A wide port may have separate connections on each of its phys"
> Again, can these connections be active at the same? If not, is it true
> that a wide port can support only a single connection at a time and all
> phys support that same connection, in which case, the above sentence could
> be interpreted as misleading. In rev 3c, there was some discussion of
> connection at a link layer (between the phys) and now in rev 3d, it
> to have been removed.
> Thanks for your help.
> Chris Malakapalli
> Seagate Technology
> * 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