SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


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

    Re: Question On R2T Behaviour?



    Suresh,
    
    The only statement that can be made is that the target is the master of 
    the data transfers.
    For the case you describe any of the behaviors you describe is legal.
    The behavior is defined by SSCSI (SAM).
    The results are implementation dependent.
    The only think that we might say is that the results "should match 
    expectation" and "the residual counts should match transfers".
    In practical terms you will find many targets that will terminate the 
    transfer with an error status if an R2T  request is not properly 
    fulfilled.
    
    Julo
    
    
    
    
    
    "Suresh Tanjore" <sureshtk@aarohicommunications.com>
    Sent by: owner-ips@ece.cmu.edu
    16-11-01 19:39
    Please respond to "Suresh Tanjore"
    
     
            To:     <ips@ece.cmu.edu>
            cc: 
            Subject:        Question On R2T Behaviour?
    
     
    
    Hi all,
    
    On page 75 of iSCSi 0.8 revision
    
                     " The data length of the DATA PDU MUST *not exceed* the 
    desired
                     data transer length specified in the R2T. If the R2T is 
    answered
                     with a sequence of data PDUs the buffer offset and Length
                     must be within the range of those specfied by R2T, the 
    last PDU
                     SHOULD have the F bit set to 1. The data-out PDU ordering 
    is goverened
                     by  DataPDUInOrder."
    
    
    My question is if the data length of the DATA PDU is less than
    the desired data transfer length and the F bit is set, what should be the
    behaviour of the target. For the sake of this discussion let us
    assume DataPDUInOrder to be set to Yes.
    
                     1. It looks intutive that target may send another R2T 
    with different
                        sequence number and collect the data initially that 
    the host
                        was not able to send for the previous R2T request.
    
                     Or
    
                     2. It is an error that need to be detected by target that 
    the host was not
                        able to fully satisfy the R2T request sent by the 
    target in the first
    place.
    
    Please let me know what is the right behaviour and should this be 
    documented
    in the
    specification if it is not already documented. If it is documented may be 
    i
    missed
    reading this, please point to me where it is defined.
    
    Thanks
    sureshtk
    
    
    
    
    
    
    
    


Home

Last updated: Sun Nov 18 10:17:57 2001
7845 messages in chronological order