Description of rounding rules for PPR transfer agreements

George Penokie gop at us.ibm.com
Thu Dec 20 06:08:48 PST 2001


* From the T10 Reflector (t10 at t10.org), posted by:
* "George Penokie" <gop at us.ibm.com>
*

I believe the wording Gerry is referring to was in SPI-4 r6 (see below). It
is located just above table 63.

>From SPI-4 r6 section 16.3.12.1:
'The initiator sets its values according to the rules above to permit it to
receive data successfully. If the target is able to receive data
successfully with these values (or smaller periods or larger REQ/ACK
offsets or both), it returns the same values in its PARALLEL PROTOCOL
REQUEST message, except for the PCOMP_EN value. If it  requires a larger
period, a smaller REQ/ACK offset, or a smaller transfer width in order to
receive data successfully, it substitutes values in its PARALLEL PROTOCOL
REQUEST message as required, returning unchanged any value not required to
be changed. If the TRANSFER PERIOD FACTOR contains a value greater than
08h, each SCSI device when transmitting data shall respect the negotiated
limits set by the other's PARALLEL PROTOCOL REQUEST message, but it is
permitted to transfer data with larger periods, smaller synchronous REQ/ACK
offsets, or both. If the TRANSFER PERIOD FACTOR contains a value equal to
08h, each SCSI device when transmitting data shall transmit data at the
negotiated transfer period but smaller synchronous REQ/ACK offsets are
allowed. The completion of an exchange of PARALLEL PROTOCOL REQUEST
messages implies an agreement as shown in table 63. If the target does not
support the selected protocol option it shall clear as many bits as
required to set the protocol option field to a legal value that it does
support.'

I have looked over r8 and can only find the statement below which does not
contain several of the allowable characteristics of the transmitting port
defined in r6. This information is important and should not have been
dropped. If I missed it then someone needs to point to the location in
SPI-4 where it is stated.

>From SPI-4 rev 8 section 4.12.2:
'Ports shall not set message fields to values they do not support. The
originating port should set the fields in the originating negotiation
message to the maximum values (e.g., fastest transfer period, largest
REQ/ACK offset) it supports. If the responding port is able to support the
requested values, it shall return the same values in the responding
negotiation message. If the responding port requires different values
(i.e., a subset of the originating port's request), it shall return those
values in the responding negotiation message (e.g., if the originating port
asks for a REQ/ACK offset of 32 and the responding port only supports a
REQ/ACK offset of 16, then the responding port replies with an offset of
16).'

Bye for now,
George Penokie

Dept 2C6  114-2 N212
E-Mail:    gop at us.ibm.com
Internal:  553-5208
External: 507-253-5208   FAX: 507-253-2880



Gerry.Houlder at seagate.com@t10.org on 12/17/2001 05:52:44 PM

Sent by:    owner-t10 at t10.org


To:    t10 at t10.org
cc:
Subject:    Description of rounding rules for PPR transfer agreements



* From the T10 Reflector (t10 at t10.org), posted by:
* Gerry.Houlder at seagate.com
*
I recall that SPI-4 used to contain rules for "rounding down" a proposed
transfer agreement for PPR message. There were several cases (e.g., period
factor is too small to be legal for ST, or IU was requested but not
supported) that were mandated to return the default transfer agreement
(aynchronous narrow) instead of a slightly slower "rounded down" speed. I
can't seem to find these rules in SPI-4 rev. 8. Were these rules removed as
part of Rob Elliott's remaking of the transfer agreement rules, or perhaps
removed earlier, or perhaps I just am not looking in the right places?

*
* 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




More information about the T10 mailing list