SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


    [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

    RE: iSCSI: draft 7: Abort Task and RefCmdSN



    Correct - expect CmdSN should take a value of one more the last CmdSN (the
    value that would have beein give to it would it not have been immediate).
    And tere is bound bound to be a slight correction som simplify abbort task
    in most of the cases comming in several days (nothing major only a
    clarification and a recommended behaviour).
    
    Julo
    
    "Nicolson, Alex" <Alex.Nicolson@emulex.com> on 07-08-2001 16:28:42
    
    Please respond to <Alex.Nicolson@emulex.com>
    
    To:   Julian Satran/Haifa/IBM@IBMIL
    cc:
    Subject:  RE: iSCSI: draft 7: Abort Task and RefCmdSN
    
    
    
    Julo,
    
    Thanks for having patience with me.
    
    The way I understand the CmdSN for Task Management in the Rev 7 spec, it
    looked that for all practical purposes, the CmdSn was going to have to be
    set to the last value sent for it to properly handle all outstanding
    commands, otherwise there would be lost Task Tags. I read it as there would
    be no responses for any tasks greater than CmdSN, therefore CmdSN seems to
    requires some not so arbitrary value. Is there something I should read more
    carefully?
    
    Alex
    
    
    
    > -----Original Message-----
    > From: Julian Satran [mailto:Julian_Satran@il.ibm.com]
    > Sent: Tuesday, August 07, 2001 1:58 AM
    > To: Alex.Nicolson@emulex.com
    > Subject: RE: iSCSI: draft 7: Abort Task and RefCmdSN
    >
    >
    > Alex
    >
    > CmdSN can take any value. CmdSN is meant mainly as a generalized
    reference
    > for where in the stream an immediate command was issue - or as a
    > CmdSN when
    > not immediate (even Abort Task does not have to be immediate).
    >
    > Julo
    >
    > "Nicolson, Alex" <Alex.Nicolson@emulex.com> on 07-08-2001 00:21:01
    >
    > Please respond to <Alex.Nicolson@emulex.com>
    >
    > To:   Julian Satran/Haifa/IBM@IBMIL
    > cc:
    > Subject:  RE: iSCSI: draft 7: Abort Task and RefCmdSN
    >
    >
    >
    > Julo,
    >
    > I'm sorry I am being so thick, I am quite confused. My understanding was
    > that to Abort Task aborted the task based on the Referenced Task
    > Tag. Is it
    > now required that to abort a task both the CmdSn and Task Tag must be
    used
    > ID a particular Task? I'm not sure why both are required.
    >
    > I can understand that a Abort Task might arrive before the iSCSI SCSI PDU
    > for that task to be aborted arrived, but can't the problem be detected
    and
    > manged by use of the CmdSN.
    >
    > (I guess I liked it better when the only purpose of CmdSn was for command
    > ordering between the iSCSI and SCSI layer. :))
    >
    > Alex.NicolsoN@emulex.com
    >
    >
    > > -----Original Message-----
    > > From: Julian Satran [mailto:Julian_Satran@il.ibm.com]
    > > Sent: Monday, August 06, 2001 4:21 PM
    > > To: Alex.Nicolson@emulex.com
    > > Cc: ips@ece.cmu.edu
    > > Subject: RE: iSCSI: draft 7: Abort Task and RefCmdSN
    > >
    > >
    > >
    > > Alex,
    > >
    > > That is correct but has no relevance if you want to abbort one specific
    > > task.
    > >
    > > Julo
    > >
    > > "Nicolson, Alex" <Alex.Nicolson@emulex.com> on 06-08-2001 21:16:28
    > >
    > > Please respond to <Alex.Nicolson@emulex.com>
    > >
    > > To:   Julian Satran/Haifa/IBM@IBMIL, <ips@ece.cmu.edu>
    > > cc:
    > > Subject:  RE: iSCSI: draft 7: Abort Task and RefCmdSN
    > >
    > >
    > >
    > > Julo,
    > >
    > > I must be misunderstanding something. I thought that Task Management
    > > commands were to be executed as if they had arrived in sequence based
    on
    > > their CmdSN.
    > >
    > >         "Task management commands must be executed
    > >         as if all the commands having a CmdSN lower or equal to the
    task
    > >         management CmdSN have been received by the target (i.e.,
    > > have to be
    > >         executed as if received for ordered delivery even when
    > marked for
    > >         immediate delivery)."
    > >
    > > Why wouldn't that be the case with Abort Task?
    > >
    > > Alex.Nicolson@emulex.com
    > >
    > >
    > > > -----Original Message-----
    > > > From: owner-ips@ece.cmu.edu [mailto:owner-ips@ece.cmu.edu]On Behalf
    Of
    > > > Julian Satran
    > > > Sent: Sunday, August 05, 2001 2:37 PM
    > > > To: ips@ece.cmu.edu
    > > > Subject: Re: iSCSI: draft 7: Abort Task and RefCmdSN
    > > >
    > > >
    > > > Robert,
    > > >
    > > > This is related to the fact that the abort command can arrive before
    > the
    > > > command to be aborted if sent on a different connection (or the
    > > > command has
    > > > been dropped due to some error). As such RefCmdSN indicates where the
    > > > command would have been in the command queue.  If they don't agree
    the
    > > > target should not abbort. And the main criteria for finding a
    > > task is the
    > > > Initiator Task Tag.
    > > > I've changed the wording to:
    > > >
    > > > 1.1.1     RefCmdSN
    > > >
    > > >    For abort-task the task CmdSN to enable task removal. If
    > > RefCmdSN does
    > > >    not match the CmdSN of the command to be aborted at the target
    > > > The abort action MUST not be performed and the response MUST be
    > function
    > > > rejected.
    > > >
    > > > Julo
    > > >
    > > >
    > > >
    > > > "Elliott, Robert" <Robert.Elliott@compaq.com>@ece.cmu.edu on
    > 05-08-2001
    > > > 20:36:10
    > > >
    > > > Please respond to "Elliott, Robert" <Robert.Elliott@compaq.com>
    > > >
    > > > Sent by:  owner-ips@ece.cmu.edu
    > > >
    > > >
    > > > To:   "'ips@ece.cmu.edu'" <ips@ece.cmu.edu>
    > > > cc:
    > > > Subject:  iSCSI: draft 7: Abort Task and RefCmdSN
    > > >
    > > >
    > > >
    > > > The Task Management [function] Command PDU includes two fields
    > > > currently only used by the ABORT TASK function:
    > > >
    > > > 2.5.2 Referenced Task Tag
    > > >   Initiator Task Tag of the task to be aborted - for abort task
    > > >
    > > > 2.5.3 RefCmdSN [Referenced command sequence number]
    > > >   For abort-task the task CmdSN to enable task removal. If RefCmdSN
    is
    > > >   is lower that ExpCmdSN or higher than MaxCmdSN the target
    > will ignore
    > > >   RefCmdSN.
    > > >
    > > > Both fields identify the task to be aborted.  The Referenced Task
    > > > Tag field sits at the SCSI level and matches the SAM-2 function
    > > > call description (SAM-2 revision 18):
    > > > 6.2 ABORT TASK
    > > >   Function call:
    > > >     Service Response = ABORT TASK (IN (I_T_L_Q Nexus) )
    > > >
    > > > The RefCmdSN field sits at the iSCSI level.
    > > >
    > > > I suggest removing one of these fields.  Having two ways to
    > > > specify the same thing just raises the question of what to do
    > > > when the values don't agree.  If that happens, should the
    > > > target:
    > > > send back a Reject PDU
    > > > abort both tasks
    > > > abort the task indicated by the Referenced Task Tag
    > > > abort the task indicated by the RefCmdSN
    > > > abort one of the tasks but also report an error
    > > > do any of the above
    > > >
    > > > Since task management functions in general may rely on the
    > > > SCSI tag (although Abort Task is the only current user of it),
    > > > I suggest keeping that flag and dropping the iSCSI field.
    > > >
    > > > --
    > > > Robert.Elliott@compaq.com
    > > > Compaq Computer Server Storage
    > > >
    > > >
    > > >
    > >
    > >
    > >
    > >
    >
    >
    >
    >
    
    
    
    
    


Home

Last updated: Tue Sep 04 01:04:04 2001
6315 messages in chronological order