Number of pad bytes with DT transfers

[Jim McGrath]

* From the T10 Reflector (t10 at t10.org), posted by:
* Jim McGrath <Jim.McGrath at quantum.com>
*

One extra point - the intention of the DT pad mechanism was to allow the pad
word (2 bytes) to be discarded at the lowest layer of the initiator
architecture (i.e. your firmware might never even see it, depending on your
initiator design).  In that case, all you end up seeing is a possible 1 byte
pad case, which is what you should be use to today.  So your actions should
be the same as it is today (if the initiator architecture adequately hid the
pad word from you).

Jim


> -----Original Message-----
> From:	Richard Moore [SMTP:r_moore at qlc.com]
> Sent:	Wednesday, March 08, 2000 2:44 PM
> To:	'Wolfgang Michel'; t10 at t10.org
> Subject:	RE: Number of pad bytes with DT transfers
> 
> * From the T10 Reflector (t10 at t10.org), posted by:
> * Richard Moore <r_moore at qlc.com>
> *
> Wolfgang,
> 
> DT transfers include a pad mechanism which is controlled through the
> P_CRCA
> signal. P_CRCA is switched low for two REQ edges (an assertion and a
> negation
> of REQ) to indicated a CRC transfer with no pad bytes. P_CRCA is switched
> low
> for three REQ edges (a negation, an assertion, and another negation) to
> indicate
> two pad bytes followed by CRC. This is sufficient for transfers of length
> mod 4
> equal to 0 or 2.
> 
> For the 1 and 3 cases, a single residue byte, or a single residue byte
> plus
> two
> pad bytes using P_CRCA, is used. The residue byte is indicated by IGNORE
> WIDE
> RESIDUE, and can only be placed at the end of the transfer, not on
> intermediate
> CRCs. On the other hand pad bytes, which always come two bytes at a time,
> can
> happen immediately prior to any CRC.
> 
> The above description is for the non-packetized case. Packetized will have
> the
> transfer length explicitly declared in the L_Q packet that precedes the
> data
> packet.
> 
>  -- Richard Moore
>     QLogic Corp.
> 
> 
> >-----Original Message-----
> >From: Wolfgang Michel [mailto:wolfgang_michel at vortex.de]
> >Sent: Tuesday, March 07, 2000 2:26 AM
> >To: t10 at t10.org
> >Subject: Number of pad bytes with DT transfers
> >
> >
> >* From the T10 Reflector (t10 at t10.org), posted by:
> >* "Wolfgang Michel" <wolfgang_michel at vortex.de>
> >*
> >
> >Given a sequential device (like a DAT, etc.) is transfering data to 
> >the initiator with DT data group transfers. And further given the 
> >device has to disconnect from the initiator due to an internal 
> >resource condition on a non module 4 boundary. The device than has to 
> >fill up the last data field up to the next 4 byte boundary with 
> >padding bytes before sending the CRC. How is the initiator able to 
> >identify the number of pad bytes (1,2 or 3)? Up to my understanding 
> >the IGNORE WIDE RESIDUE message "invalidates" only 1 byte.
> >
> >Your assistance/help is really appreciated.
> >
> >Thanks
> >Wolfgang
> >
> >
> >
> >***************************************************
> >Wolfgang Michel, Director R&D
> >ICP vortex Computersysteme GmbH

> >Konrad Zuse Str. 9
> >74172 Neckarsulm, Germany
> >Tel.: +49-7132-9620-535
> >Fax : +49-7132-9620-200
> >email: w_michel at vortex.de
> >
> >http://www.icp-vortex.com
> >
> >http://www.vortex.de
> >*************************************************** 
> >
> >*
> >* For T10 Reflector information, send a message with
> >* 'info t10' (no quotes) in the message body to majordomo at t10.org
> >
> *
> * For T10 Reflector information, send a message with
> * 'info t10' (no quotes) in the message body to majordomo at t10.org

*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org