Re: iSCSI: F bits in Command and DataIn PDUs, too?

[julian_satran@il.ibm.com]

Stephen,

I've corrected the statement and I was already considering moving the X bit
to the command and have the P/F be always a Poll/Final bit (or 0)(not
exactly what you wanted but close enough).  I have however to check all the
side effects...

Thanks,
Julo

"Ferrari, Stephen" <smf@pirus.com> on 13/01/2001 02:27:30

Please respond to "Ferrari, Stephen" <smf@pirus.com>

To:   ips@ece.cmu.edu
cc:
Subject:  iSCSI: F bits in Command and DataIn PDUs, too?




According to the spec, the F bit in DataOut PDUs indicates "the last PDU of
immediate data or the last PDU of a sequence answering a R2T".  I believe
this is a typo and should be "the last PDU of _unsolicited_ data or the
last
PDU of a sequence answering an R2T".

For consistency, it would seem that Command PDUs should also have an F bit
defined to indicate the last PDU of unsolicited data if all of the
unsolicited data is sent as immediate data, and DataIn PDUs should have an
F
bit to indicate the last PDU of data for the command.  Granted, these two
cases are easier to detect than the case already covered, since there can
be
at most one PDU of immediate data per command (so there's no overlap
possible), and the Response PDU implies that there's no more DataIn coming.
But having a single bit, preferably in the same position in all three PDU
types, to indicate that "this data transfer is done" would simplify data
transfer management.  (Note that for DataOut and bi-directional operations
there can be multiple data transfers per command.)