RE: iSCSI-countdown to new version

["BURBRIDGE,MATTHEW (HP-UnitedKingdom,ex2)" <matthew_burbridge@hp.com>]

Julian,

The text describes that the first login PDU of a connection and not the
session is set as non-immediate.  However, I think all login PDUs shouls be
immediate:

Is the only reason for making the initial login PDU of a session
non-immediate is to set the CmdSN?  However, if all login PDUs are immediate
then the target can set the ExpCmdSN from the CmdSN of the login PDU.  This
has the advantage that processing all login PDUs is handled the same and
that the protocol is simplified.

Cheers

Matthew

-----Original Message-----
From: Julian Satran [mailto:Julian_Satran@il.ibm.com]
Sent: Sunday, September 16, 2001 10:50 AM
To: BURBRIDGE,MATTHEW (HP-UnitedKingdom,ex2)
Cc: 'eddy_quicksall@ivivity.com'; ips@ece.cmu.edu
Subject: RE: iSCSI-countdown to new version



Not for the session establishing login request.  Julo

"BURBRIDGE,MATTHEW (HP-UnitedKingdom,ex2)" <matthew_burbridge@hp.com>
@ece.cmu.edu on 13-09-2001 20:33:30

Please respond to "BURBRIDGE,MATTHEW (HP-UnitedKingdom,ex2)"
      <matthew_burbridge@hp.com>

Sent by:  owner-ips@ece.cmu.edu


To:   "'eddy_quicksall@ivivity.com'" <eddy_quicksall@ivivity.com>,
      ips@ece.cmu.edu
cc:
Subject:  RE: iSCSI-countdown to new version



Hi Eddy,

The last part of the sentence was added just to highlight that if the first
full feature phase commands were immediate then it does not increment the
CmdSN.  I would therefore proprose:

  Command numbering starts with the login request on the first
  connection of a session (the leading login) and includes every
  non-immediate command issued afterwards including into the full
  feature phase.

The first login must be immediate otherwise if it is queued and there is a
missing PDU earlier in the command queue then the login will not get
processed.  If the login was for recovery we have stalemate!

Cheers

Matthew


-----Original Message-----
From: Eddy Quicksall [mailto:eddy_quicksall@ivivity.com]
Sent: Thursday, September 13, 2001 6:25 PM
To: 'BURBRIDGE,MATTHEW (HP-UnitedKingdom,ex2)'
Subject: RE: iSCSI-countdown to new version


I'm not a good writer but I would write it as:

  Command numbering starts with the login request on the first
  connection of a session (the leading login) and includes every
  non-immediate command issued afterwards.

The reason it works better is because all login commands have to be
immediate (with the correction you have stated).

BTW, I thought the reason for the very first login to have the I = 0 was so
hardware does not have to worry about the leading login and will just
always
advance CmdSN for the next Immediate command.

Does Julian know the leading login is supposed to also be Immediate?

Eddy


-----Original Message-----
From: BURBRIDGE,MATTHEW (HP-UnitedKingdom,ex2)
[mailto:matthew_burbridge@hp.com]
Sent: Thursday, September 13, 2001 1:21 PM
To: 'eddy_quicksall@ivivity.com'
Subject: RE: iSCSI-countdown to new version


Eddy,

I think I may be missing something here: what would you suggest the wording
to be?

Matthew

-----Original Message-----
From: Eddy Quicksall [mailto:eddy_quicksall@ivivity.com]
Sent: Thursday, September 13, 2001 6:17 PM
To: 'BURBRIDGE,MATTHEW (HP-UnitedKingdom,ex2)'
Subject: RE: iSCSI-countdown to new version


Because it says "afterwards whether during login".

Eddy

-----Original Message-----
From: BURBRIDGE,MATTHEW (HP-UnitedKingdom,ex2)
[mailto:matthew_burbridge@hp.com]
Sent: Thursday, September 13, 2001 1:16 PM
To: 'eddy_quicksall@ivivity.com'
Subject: RE: iSCSI-countdown to new version


Hi Eddy,

Why does 1.2.2.1 need modifying?

Matthew

-----Original Message-----
From: Eddy Quicksall [mailto:eddy_quicksall@ivivity.com]
Sent: Thursday, September 13, 2001 6:13 PM
To: 'BURBRIDGE,MATTHEW (HP-UnitedKingdom,ex2)'; julian_satran@il.ibm.com
Cc: Sanjay Goyal; ips@ece.cmu.edu
Subject: RE: iSCSI-countdown to new version


So a correction is needed in 2.12.2.

Also, if all other login phase PDUs MUST be immediate, Section 1.2.2.1
needs
change because it clearly implies the opposite.

Eddy

-----Original Message-----
From: BURBRIDGE,MATTHEW (HP-UnitedKingdom,ex2)
[mailto:matthew_burbridge@hp.com]
Sent: Thursday, September 13, 2001 12:41 PM
To: 'eddy_quicksall@ivivity.com'
Subject: RE: iSCSI-countdown to new version


Hi Eddy,

IMO 2.12.2 is incorrect and the initial login command MUST be immediate as
are all other login phase PDUs.  Therefore 1.2.2.1 is correct.

Cheers

Matthew

-----Original Message-----
From: Eddy Quicksall [mailto:eddy_quicksall@ivivity.com]
Sent: Thursday, September 13, 2001 5:15 PM
To: julian_satran@il.ibm.com; ips@ece.cmu.edu
Subject: RE: iSCSI-countdown to new version


Section 1.2.2.1 says:

  Command numbering starts with the login request on the first
  connection of a session (the leading login) and includes every
non-immediate
  command issued afterwards whether during login or in full-feature
  phase.

But 2.12.2 says:

  Login MUST be issued as an immediate request (I=1) except for the
  first Login request (C=0) of the leading connection that MUST have
  I=0.

Since all commands during login must have the I bit set, I believe 1.2.2.1
3rd line should read

  command issued afterwards in full-feature

Am I correct?

Eddy