Wide Links and Wide Ports

Packer, John John_Packer at adaptec.com
Tue Feb 25 11:05:01 PST 2003


* From the T10 Reflector (t10 at t10.org), posted by:
* "Packer, John" <John_Packer at adaptec.com>
*
However, the spec implies and can be interpreted that the L_Q nexus can
wander across the various I_T nexus connections.  For example if I had two
connections, A and B, open for an I_T Nexus and three L_Q Nexus X and Y and
Z active I could see any of X, Y and Z move on A or B (but any one I_T nexus
cannot move on A and B simultaneously).

John Packer
Adaptec

> -----Original Message-----
> From: Sheffield, Robert L [mailto:robert.l.sheffield at intel.com]
> Sent: Tuesday, February 25, 2003 9:55 AM
> To: david.freeman at datatransit.com; t10 at t10.org
> Subject: Re: Wide Links and Wide Ports
> 
> * From the T10 Reflector (t10 at t10.org), posted by:
> * "Sheffield, Robert L" <robert.l.sheffield at intel.com>
> *
> David,
> 
> You are correct that only one data transfer can be in progress for a given
> I_T_L_Q nexus at a time (special exception for bidirectional cmds), but
> many I_T_L_Q nexuses can be active at the same time and use connections in
> a wide link as available. So a given I_T can have traffic in parallel on
> parallel connections, only each connection on the wide link carries
> traffic for a different L_Q nexus.
> 
> Bob
> -----------------
> Sent from my BlackBerry Handheld.
> 
> -----Original Message-----
> From: David Freeman <david.freeman at datatransit.com>
> To: t10 at t10.org <t10 at t10.org>
> Sent: Mon Feb 24 18:27:24 2003
> Subject: Re: Wide Links and Wide Ports
> 
> I guess that confusing part would be "A wide port shall only transmit DATA
> frames for a given I_T_L_Q nexus on one phy at a time"
> This seems to be saying that within a given I_T_L_Q nexus only one phy is
> in use at any given time.
> 
> Does the wording of this section need to be clearer or am I
> misinterpreting this sentence?
> 
> David
> 
> ----- Original Message -----
> From: Seto,  <mailto:pak-lung.seto at intel.com> Pak-lung
> To: 'David Freeman'   ; t10 at t10.org
> Sent: Monday, February 24, 2003 5:03 PM
> Subject: RE: Wide Links and Wide Ports
> 
> No, it explained very clear in the standard as you quoted in your email.
> It can even use different PHYS.  The only requirement is that you can not
> switch link until all the data frame has been acknowledged.  You can even
> transmit and receive on two different links within the wideport.
> 
> Pak
> 
> -----Original Message-----
> From: David Freeman [mailto:david.freeman at datatransit.com]
> Sent: Monday, February 24, 2003 4:42 PM
> To: t10 at t10.org
> Subject: Wide Links and Wide Ports
> 
> 
> I seem to be confused about how wide ports and wide links are handled.
> Section 8.3.3.1.4 of the SAS spec. (rev. D) says...
> 
> "  A wide port shall only transmit DATA frames for a given I_T_L_Q nexus
> on one phy at a time. The wide port may switch phys for DATA frames once
> all of the previous DATA frames have been acknowledged by the link layer.
> Read DATA frames and write DATA frames for the same I_T_L_Q nexus may be
> transmitted simultaneously and may be on the same or different phys."
> 
> Does this mean that inside a wide link all of the frames between and
> including the "Command" frame and the "Response" frame will only occur on
> one physical link?[Seto]  No
> 
> This seems to defeat the purpose of having a wide link.
> 
> If there is any one out there that could help clear this confusion, I
> would appreciate it.
> 
> Thank You
> 
> David R. Freeman
> Data Transit
> 
> 
> *
> * 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