SAS Wide link question

Wed Dec 18 09:15:01 PST 2002

* From the T10 Reflector (t10 at t10.org), posted by:
* "Elliott, Robert (Server Storage)" <Elliott at hp.com>
*
This is a multi-part message in MIME format.

------_=_NextPart_001_01C2A6B8.FF83CA44
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

Whether the SAS address is a device address or port address changed =
from
sas-r00 to sas-r03. You've found an instance of the old definition =
still
lingering.
=20
The SAS address reported in the IDENTIFY address frame sent by an end
device is a port address.
The SAS address reported in the IDENTIFY address frame sent by an
expander device is a device address.
=20
In that example, the initiator device has chosen to use different SAS
addresses for phys A and B (one port) than it uses for phys C, D, E, =
and
F (another port).
=20
The expander "ports" are just conceptual.  The expander still sends the
same SAS address in IDENTIFY out each phy.  When it receives different
addresses back from the attached (end) device, it knows that its phys
have been split into multiple ports. The expander ports all have the
same SAS address.
=20
We should correct the definition of SAS ADDRESS field in the IDENTIFY
address frame (in 7.7.2):
"The SAS ADDRESS field indicates the SAS address of the device
transmitting the IDENTIFY address frame."

should be:

"The SAS ADDRESS field indicates the SAS address of the port
transmitting the IDENTIFY address frame."

or if the concept of expander ports having SAS addresses is not desired
(it's not listed in 4.4.2 now):

"In end devices, the SAS ADDRESS field indicates the SAS address of the
port transmitting the IDENTIFY address frame. In expander devices, the
SAS ADDRESS field indicates the SAS address of the expander device
transmitting the IDENTIFY address frame."

--=20
Rob Elliott, elliott at hp.com=20
Hewlett-Packard Industry Standard Server Storage Advanced Technology=20
https://ecardfile.com/id/RobElliott
<https://ecardfile.com/id/RobElliott> =20

-----Original Message-----
From: Sheffield, Robert L [mailto:robert.l.sheffield at intel.com]=20
Sent: Wednesday, December 18, 2002 9:56 AM
To: Elliott, Robert (Server Storage); t10 at t10.org
Subject: RE: SAS Wide link question

Rob,
=20
Figure 17 in SAS-r03 shows two expander ports attached to two initiator
ports (A,B) and (C,D,E,F). I'm wondering how the expander knows to
partition the ports that way. The Identify address frame reports the =
SAS
address of the "device". If this is taken literally, the same SAS
address would be reported by the initiator on all 6 phys. I assume if
the initiator wants the expander to recognize distinct ports it would
report the SAS addresses for the distinct 'ports' in the Identify
Address Frame, but I can't find anywhere in the standard where it says
so. As far as I can see, Expander Devices don't have SAS addresses for
their "ports", so the mapping of ports across phys in the expander
device must be based on information gathered from the attached
initiators and targets. Can you clarify this for me, or point me to a
place in the document that explains it?
=20
Thanks,
Bob
=20
Bob Sheffield
Intel Corporation - CH6-333
Storage Components Division (SCD)
5000 W. Chandler Blvd
Chandler, AZ 85226-3699
Phone: 480-554-8597
Fax: 480-554-6617

-----Original Message-----
From: Elliott, Robert (Server Storage) [mailto:Elliott at hp.com]
Sent: Wednesday, December 11, 2002 5:19 PM
To: t10 at t10.org
Subject: RE: SAS Wide link question

If the target port is wide (i.e. it has more than one phy, advertising
the same SAS address), it may request a connection using any phy in the
wide port.  It doesn't have to remember the phy through which the
command arrived.
=20
Similarly, a wide initiator port may request a connection to a target
using a different phy than it has used previously.
=20
Expanders are also free to choose any link in a wide link as they route
connection requests through the domain.
=20
A wide initiator port should not ask for more than one connection at a
time to the same target port unless it knows the target port is also
wide (otherwise, all but one of its requests will lose arbitration to
the winner).
--=20
Rob Elliott, elliott at hp.com=20
Hewlett-Packard Industry Standard Server Storage Advanced Technology=20
https://ecardfile.com/id/RobElliott
<https://ecardfile.com/id/RobElliott> =20

-----Original Message-----
From: Kevin_Marks at Dell.com [mailto:Kevin_Marks at Dell.com]=20
Sent: Tuesday, December 10, 2002 1:20 PM
To: t10 at t10.org
Subject: SAS Wide link question

With an initiator and target in a SAS domain having a wide link
configuration, on a reconnect does the target for example have to
maintain allegiance to the original path (phy) when reconnecting or can
arbitrate for any path back to the initiator.

=20

Thanks

=20

Kevin

=20

Kevin Marks ( Kevin_Marks at Dell.com <mailto:Kevin_Marks at Dell.com>  )
=20

=20

------_=_NextPart_001_01C2A6B8.FF83CA44
Content-Type: text/html;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">

Message P.MsoNormal { 	FONT-SIZE: 12pt; MARGIN: 0in 0in 0pt; FONT-FAMILY: "Times New Roman" } LI.MsoNormal { 	FONT-SIZE: 12pt; MARGIN: 0in 0in 0pt; FONT-FAMILY: "Times New Roman" } DIV.MsoNormal { 	FONT-SIZE: 12pt; MARGIN: 0in 0in 0pt; FONT-FAMILY: "Times New Roman" } A:link { 	COLOR: blue; TEXT-DECORATION: underline } SPAN.MsoHyperlink { 	COLOR: blue; TEXT-DECORATION: underline } A:visited { 	COLOR: purple; TEXT-DECORATION: underline } SPAN.MsoHyperlinkFollowed { 	COLOR: purple; TEXT-DECORATION: underline } SPAN.EmailStyle17 { 	COLOR: windowtext; FONT-FAMILY: Arial } DIV.Section1 { 	page: Section1 } Whether the SAS=20 address is a device address or port address changed from sas-r00 to=20 sas-r03. You've found an instance of the old definition still=20 lingering.

 The = SAS address=20 reported in the IDENTIFY address frame sent by an end = device is a port=20 address.
 The SAS address = reported in the=20 IDENTIFY address frame sent by an expander device is a device = address.

 In = that example, the=20 initiator device has chosen to use different SAS addresses for phys A = and B (one=20 port) than it uses for phys C, D, E, and F (another = port).

 The = expander "ports"=20 are just conceptual.  The expander still sends the same = SAS=20 address in IDENTIFY out each phy.  When it receives different = addresses=20 back from the attached (end) device, it knows that its phys have been = split into=20 multiple ports. The expander ports all have the same SAS=20 address.

 We = should correct=20 the definition of SAS ADDRESS field in the IDENTIFY address frame (in=20 7.7.2):
 "The SAS = ADDRESS=20 field indicates the SAS address of the device = transmitting=20 the IDENTIFY address frame." should=20 be: "The SAS ADDRESS = field indicates the SAS address of the port transmitting the = IDENTIFY=20 address frame." or if the concept of expander = ports having=20 SAS addresses is not desired (it's not listed in 4.4.2=20 now): "In end devices, the SAS ADDRESS=20 field indicates the SAS address of the port = transmitting the=20 IDENTIFY address frame. In expander devices, the SAS = ADDRESS=20 field indicates the SAS address of the expander = device=20 transmitting the IDENTIFY address frame."
 -- = 
Rob Elliott, = elliott at hp.com=20 
Hewlett-Packard = Industry Standard=20 Server Storage Advanced Technology 
https://ecardfile.com/id/Ro= bElliott=20 

-----Original Message-----
From: = Sheffield,=20 Robert L [mailto:robert.l.sheffield at intel.com] 
Sent: = Wednesday,=20 December 18, 2002 9:56 AM
To: Elliott, Robert (Server = Storage);=20 t10 at t10.org
Subject: RE: SAS Wide link = question

 Rob,

 Figure 17 in SAS-r03 shows two expander = ports=20 attached to two initiator ports (A,B) and (C,D,E,F). I'm wondering = how the=20 expander knows to partition the ports that way. The Identify address = frame=20 reports the SAS address of the "device". If this is taken literally, = the same=20 SAS address would be reported by the initiator on all 6 phys. I = assume if the=20 initiator wants the expander to recognize distinct ports it would = report the=20 SAS addresses for the distinct 'ports' in the Identify Address Frame, = but I=20 can't find anywhere in the standard where it says so. As far as I can = see,=20 Expander Devices don't have SAS addresses for their "ports", so the = mapping of=20 ports across phys in the expander device must be based on information = gathered=20 from the attached initiators and targets. Can you clarify this for = me, or=20 point me to a place in the document that explains = it?

 Thanks,
 Bob

 Bob Sheffield
Intel Corporation -=20 CH6-333
Storage Components Division (SCD)
5000 W. Chandler=20 Blvd
Chandler, AZ 85226-3699
Phone: 480-554-8597
Fax:=20 480-554-6617

 -----Original Message-----
From: Elliott, Robert = (Server=20 Storage) [mailto:Elliott at hp.com]
Sent: Wednesday, = December 11,=20 2002 5:19 PM
To: t10 at t10.org
Subject: RE: SAS = Wide link=20 question

 If=20 the target port is wide (i.e. it has more than one phy, advertising = the same=20 SAS address), it may request a connection using any phy in the wide = port.  It doesn't have to remember the phy through which the = command=20 arrived.

 Similarly, a wide initiator port may request a connection = to a target=20 using a different phy than it has used = previously.

 Expanders are also free to choose any link in a wide link = as they=20 route connection requests through the domain.

 A=20 wide initiator port should not ask for more than one connection at = a time to=20 the same target port unless it knows the target port is also wide=20 (otherwise, all but one of its requests will lose arbitration to = the=20 winner).
 -- 
Rob Elliott, = elliott at hp.com=20 
Hewlett-Packard = Industry=20 Standard Server Storage Advanced Technology 
https://ecardfile.com/id/Ro= bElliott=20 

-----Original Message-----
From: = Kevin_Marks at Dell.com [mailto:Kevin_Marks at Dell.com] = 
Sent:=20 Tuesday, December 10, 2002 1:20 PM
To:=20 t10 at t10.org
Subject: SAS Wide link = question

 With an initiator = and target=20 in a SAS domain having a wide link configuration, on a reconnect = does the=20 target for example have to maintain allegiance to the original = path (phy)=20 when reconnecting or can arbitrate for any path back to the=20 initiator. Kevin Marks ( = Kevin_Marks at Dell.com=20 )

------_=_NextPart_001_01C2A6B8.FF83CA44--